Professional Documents
Culture Documents
15:28
Annual Reviews
AR232-FM
CONTENTS
FRONTISPIECEMinor J. Coon
CYTOCHROME P450: NATURES MOST VERSATILE BIOLOGICAL
CATALYST, Minor J. Coon
CYTOCHROME P450 ACTIVATION OF ARYLAMINES AND
HETEROCYCLIC AMINES, Donghak Kim and F. Peter Guengerich
GLUTATHIONE TRANSFERASES, John D. Hayes, Jack U. Flanagan,
and Ian R. Jowsey
xii
1
27
51
89
119
147
177
203
227
247
269
291
311
P1: JRX
vi
15:28
Annual Reviews
AR232-FM
CONTENTS
335
357
385
413
439
465
477
495
529
565
587
605
629
657
P1: JRX
15:28
Annual Reviews
AR232-FM
CONTENTS
vii
689
725
INDEXES
Subject Index
Cumulative Index of Contributing Authors, Volumes 4145
Cumulative Index of Chapter Titles, Volumes 4145
ERRATA
An online log of corrections to Annual Review of Pharmacology and
Toxicology chapters may be found at
http://pharmtox.annualreviews.org/errata.shtml
751
773
776
P1: JRX
3 Dec 2004
20:34
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
10.1146/annurev.pharmtox.45.120403.100030
Minor J. Coon
Victor C. Vaughan Distinguished University Professor of Biological Chemistry, Emeritus,
Department of Biological Chemistry, Medical School, The University of Michigan,
Ann Arbor, Michigan 48109; email: mjcoon@umich.edu
BACKGROUND
To my knowledge, my forebears, who migrated to the state of Colorado in the
United States from Germany (via Russia) and from the Netherlands, had no scientific credentials. However, my father and my paternal grandmother were highly
interested in reading about scientific advances despite having had no formal training, and it was my good fortune that my parents were fully supportive, even pleasantly surprised, at my own scientific bent. I also had the benefit of exposure to
rigorous courses in the Denver public schools. Our teachers frequently told us that
the schools in our city were ranked among the ten best in the country. We did not
ask for documentation of that fact, but the science courses in my high school were
challenging and so interesting that I considered a future career in several such
fields. Geology particularly intrigued me, possibly because from our classroom
windows we could see in the foothills of the Rocky Mountains the formations we
were studying. The chemistry courses, however, opened a new world to me, and I
knew I would continue to pursue some branch of this subject. That turned out to be
the relatively new field of biochemistry, which I first learned about as an undergraduate at the University of Colorado, Boulder. Professor Reuben Gustavson, whose
0362-1642/05/0210-0001$14.00
3 Dec 2004
20:34
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
COON
training had been in steroid biochemistry at the University of Chicago, taught the
freshman chemistry course, in which he included examples of problems in biology
waiting to be solved by the application of chemistry. His enthusiasm and engrossing stories about the early history of science and the personalities involved made
the subject come alive. I readily accepted his invitation to join his small research
group working on estrogen metabolism, and, at his recommendation, spent the
summer following my junior year at the University of Chicago in the laboratory of
Professor F.C. Koch. From my experience, I am convinced that research in some
field of scholarly endeavor is as crucial to undergraduate education as the usual
didactic studies.
GRADUATE STUDY
Present-day academic institutions advertise widely to attract the best qualified
faculty members, postdoctoral associates, and even medical and graduate students,
and they spend much effort in interviewing and impressing applicants. It was much
simpler in 1943, when I had a brief discussion with Dr. Gustavson about my desire
to enter graduate school. He suggested a few top institutions and we decided on
the University of Illinois. As a result of his correspondence with Professor William
C. Rose, Head of the Biochemistry Division in the Chemistry Department there, I
moved to Urbana in September as a Graduate Research Assistant, and I undertook
a laboratory project a few weeks later.
There were no laboratory rotations in the 1940s, but after considering several
attractive possibilities, I chose Professor Will Rose as my mentor. I paraphrase here
what I have written at greater length elsewhere about his personality and accomplishments (1). Roses research interests included the intermediary metabolism of
amino acids, creatine, uric acid, and related compounds, and he was renowned
for the discovery, isolation, and identification of a new amino acid, -amino-hydroxy-n-butyric acid, which he named threonine. This was the culmination of
experiments in which rats failed to grow on diets containing the 19 previously
known amino acids. When I arrived in Urbana, the identity of the 10 amino acids
essential for growth in rats and the 8 essential for the maintenance of nitrogen
equilibrium in the human (that is, male graduate students) was already known. It
fell to my lot to isolate or synthesize, purify, and analyze amino acids and then
feed them to my fellow students enlisted as human guinea pigs to establish the
quantitative requirements for the essential amino acids and the availability of their
D-isomers or N-acetyl derivatives. In those days, the recruits were grateful for the
free synthetic diets, the dollar a day they were paid, and the prospect of seeing
their initials in Roses widely read publications, but they required my constant encouragement because the rations were unpalatable. I have often remarked that any
skills I developed to persuade my recalcitrant fellow students to continue in these
difficult experiments (and therefore lead to completion of my Ph.D. thesis) became
useful many years later when, as chair of a biochemistry department, I had to deal
3 Dec 2004
20:34
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
with faculty colleagues having contrary views. Roses students were somewhat in
awe of the professor, probably wondering whether they could meet his exacting
standards or hope to emulate the seeming ease with which he succeeded in all his
professional endeavors. I learned in time that behind his somewhat reserved and
formal manner was genuine warmth and an understanding that young scientists
develop their full potential only by profiting from their mistakes; we became close
friends until his death at age 98.
3 Dec 2004
20:34
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
COON
3 Dec 2004
20:34
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
3 Dec 2004
20:34
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
COON
3 Dec 2004
20:34
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
3 Dec 2004
20:34
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
COON
the activities and spectra of liver microsomes isolated from the treated animals
for two types of CO-binding pigments, or possibly two interconvertible forms of
a single cytochrome. With respect to the active oxidant produced by P450, oxene,
analogous to compound I of peroxidases, was proposed. All in attendance agreed
that the very intriguing field of drug metabolism was on the threshold of major
progress.
This prefatory chapter is concerned with my research interests that led our laboratory to study the biochemical aspects of drug metabolism, and no attempt is
made to provide a general review of what has become a huge field of endeavor.
However, mention should be made of the outstanding contributions of the Gunsalus laboratory with bacterial P450cam, a nonmembranous cytochrome that is
specific for camphor oxidation (35, 36) and has served as a model for the versatile
but less tractable mammalian P450s. Readers interested in developments in this
field over the years are referred to the proceedings of several series of international
meetings, all with an emphasis on basic science: Symposia on Microsomes and
Drug Oxidations, as already mentioned; Conferences on the Biochemistry and
Biophysics of Cytochrome P450 (37), originated in 1976 by Klaus Ruckpaul, who
was working at Berlin-Buch to overcome the barriers that had divided eastern and
western Europe since the end of World War II, and whose valiant efforts in this
endeavor attracted worldwide support as acknowledged by Sinisi Maricic, the organizer of the first conference (38); meetings on Cytochrome P450 Diversity (37),
with an emphasis on microbial and plant systems, initiated by Hans-Georg Mueller,
a colleague of Ruckpauls at Berlin-Buch; and meetings of the International Society for the Study of Xenobiotics, started by Bruce Migdaloff in discussions with
Fred DiCarlo, John Baer, and Ina Snow at the 1980 Gordon Conference on Drug
Metabolism and launched the following year. Perhaps surprisingly, sufficient new
results are obtained from laboratories around the world to justify all of these and
other related meetings on a regular basis. I had the pleasure of chairing the committees that provided oversight for the Microsomes and Drug Oxidations symposia
and P450 conferences for many years. Without a doubt, the collaborations and
friendships that grew out of such international meetings were a major stimulus to
the rapid development of this broad field, including its application to drug design
and development.
3 Dec 2004
20:34
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
3 Dec 2004
20:34
10
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
COON
the question of how many of these unique catalysts occur in different species. The
reductase, which fortunately occurs in only a single form, was purified from both
rat (57) and rabbit liver microsomes (58), the structural features were established
(59), and the crystal structure was subsequently reported (60). The interactions
among the various purified components of the enzyme system have been investigated by several techniques (61, 62), including the puzzling effects of cytochrome
b5, which may be stimulatory or inhibitory (6366).
Among the other rabbit cytochromes identified and then isolated in our laboratory were P450s 2C3 (form 3b) (67), 3A6 (form 3c) (68), and 1A1 (form 6)
(69). In addition, the P450 that catalyzes the 12-hydroxylation of 7-hydroxy4-cholesten-3-one, an intermediate in the conversion of cholesterol to cholic
acid, was purified in collaboration with Kyu-Ichiro Okuda and his associates in
Hiroshima (70). Thanks to the painstaking work of Dennis Koop, Edward Morgan,
and George Tarr (71), ethanol-inducible cytochrome P450 was discovered in rabbit
liver microsomes, purified, and characterized as a unique isozyme with unusually
interesting properties. This cytochrome, designated 2E1 (form 3a), displayed the
highest activity of the rabbit isozymes in the oxidation of ethanol to acetaldehyde and was also found to oxidize other alcohols, aniline, and several drugs (72).
The existence of such a microsomal ethanol-oxidizing system, first proposed by
Lieber & DeCarli (73), had previously been the subject of much debate. Although
this may not be a major pathway for alcohol oxidation under most circumstances,
the increased levels of 2E1 resulting from the diabetic state, fasting, and exposure
to ethanol and several other diverse agents, including acetone, imidazole, benzene,
and isoniazid, is a cause for concern because of resulting toxicities (74). In particular, acetaminophen, a widely used antipyretic and analgesic drug, is normally
nontoxic, but in large doses it produces acute hepatic necrosis when converted to a
reactive metabolite. Of a series of P450 isoenzymes examined, 2E1 was one of the
most active in producing this metabolite (75). As summarized elsewhere (76), the
predominant role of alcohol-inducible P450 in oxidative damage involves activation of carcinogenic nitrosamines (77) and the leakiness of this cytochrome in
generating hydrogen peroxide and oxygen radicals (78), as well as alkoxy radicals
in the cleavage of lipid hydroperoxides (79). The rabbit is apparently unique in
having two genes in the alcohol-inducible P450 subfamily, the exon-intron organization of which was determined by restriction mapping and sequence analysis
(80). The genes are not coordinately expressed or regulated, and chemical inducers
act through changes in the rate of synthesis or degradation of the enzyme, rather
than through increased gene transcription (81). The corresponding enzymes are
97% identical in amino acid sequence and have similar substrate selectivity, with
2E2 always somewhat less active. The regulation of 2E1 is particularly complex
and includes effects of insulin and thyroid hormone on mRNA turnover (82) as
well as of cytokines on the transcriptional regulation of both 2E isoforms (83).
Todd Porter, already an expert on NADPH-cytochrome P450 reductase when
he joined the Biological Chemistry Department as a faculty member, contributed
greatly to our research progress with his experience in molecular biology and
3 Dec 2004
20:34
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
11
3 Dec 2004
20:34
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
COON
12
AR
Figure 2 Joint function of P450 and reductase in drug metabolism. The schemes
account for the oxygenase, oxidase, and peroxygenase reactions of cytochrome P450
with electron transfer from NADPH via the reductase. (A) The reductase cycle is
modified from that in Reference 94 with the model for rapid interflavin electron transfer
in Reference 95. (B) The P450 cycle is based on that in Reference 96. Fe represents the
heme iron atom, RH a drug or other substrate, and ROH the corresponding product.
3 Dec 2004
20:34
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
13
3 Dec 2004
20:34
14
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
COON
3 Dec 2004
20:34
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
15
3 Dec 2004
20:34
16
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
COON
More recently, we have gained further insight into the details of oxygen activation by site-directed mutagenesis of mammalian 2B4 and 2E1 in which the highly
conserved I helix threonine residue was replaced by alanine (115, 116). The impetus for our studies was the finding by the Ishimura (117) and Sligar (118) laboratories that the analogous mutation in P450cam apparently caused disruption of proton
delivery, thereby interfering with the conversion of dioxygen to the oxenoid species
and, therefore, the oxidation of the substrate, camphor. Replacement of threonine302 by alanine in 2B4 virtually obliterated benzphetamine demethylation and also
caused decreases in cyclohexane hydroxylation and phenylethanol oxidation. In
sharp contrast, the deformylation of cyclohexane carboxaldehyde was increased
approximately tenfold (115, 119). On the basis of these findings and our previous evidence that P450-dependent aldehyde deformylation is supported by added
H2O2, we concluded that the iron-peroxo species, not oxenoid-iron, is the direct
oxygen donor (115). Furthermore, in a study of olefin epoxidation (with cyclohexene, styrene, and the cis- and trans-isomers of 2-butene as substrates) by the T302A
and T303A mutants of P450s 2B4 and 2E1, respectively, we obtained evidence for
hydroperoxo-iron (as well as oxenoid-iron) as an electrophilic oxidant (116). Thus,
our results support the involvement of three functional species produced during the
reduction of molecular oxygen: peroxo-iron, hydroperoxo-iron (or its protonated
version, iron-complexed hydrogen peroxide), and oxo-iron, as shown in Figure 3.
In the past few years, in collaboration with Martin Newcomb and Paul Hollenberg, we have examined hydroxylations of unactivated C-H bonds in hydrocarbons
3 Dec 2004
20:34
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
17
and related compounds by use of highly reactive radical clocks. These mechanistic probes, including trans-2-methoxy trans-3-phenylmethylcyclopropane and
methylcubane, were chosen to differentiate between cationic and radical species
because for these two kinds of intermediates different structural rearrangements
occur. When these probes were used with several P450s, cationic products were
found, and the small amounts of radical-derived products indicated that radicallike species were very short lived (subpicosecond) (122). A recent review presents
our knowledge of the very complex P450-catalyzed hydroxylation reaction (121).
In summary, in addition to the commonly accepted iron-oxo species, a second
electrophilic oxidant is believed to exist. This scheme also takes into account
computational work by Shaik et al. (123) and Yoshizawa et al. (124) that suggests the iron-oxo species may function in multiple spin states, possibly one that
would involve oxygen insertion as envisioned in the early days of P450 mechanistic research (125, 126), and the other that would, by hydrogen abstraction,
give a radical intermediate and thus resemble the oxygen-rebound pathway (99).
A related long-standing question is whether the thiolate provided by a cysteine
residue as the proximal heme ligand contributes to the chemical reactivity of these
catalysts. Replacement of the active site cysteine-436 by serine has recently been
shown to convert P450 2B4 into an NADPH oxidase with negligible monooxygenase activity (119). Remaining problems of oxygen activation will continue to
be solved, but it is now clearly evident that the occurrence of multiple oxidizing
species contributes to the remarkable versatility of the P450 family of isozymes
in the modification of drugs and other substrates.
Of major importance, some crystal structures of mammalian P450s are now
known. The first, reported in 2000 by Cosme & Johnson (127) and Williams
et al. (128), was that of isoform 2C5 that had been engineered to delete the single
N-terminal transmembrane domain and to mutate a peripheral membrane-binding
site. More recently, Scott et al. (129) have reported the structure of 2B4, which
revealed a large open cleft that extends from the protein surface directly to the
heme iron. Differences between the two structures suggest that defined regions of
these xenobiotic-metabolizing cytochromes may assume a substantial range of energetically accessible conformations. This flexibility is likely to facilitate substrate
access, metabolic versatility, and product egress. The structural and functional data
available suggest that conformational flexibility may be central to the ability of
family 2 cytochromes to bind such a diverse array of xenobiotics. Thus, both the
structural features of the cytochromes and the generation of multiple oxidants with
different properties (120) may contribute to their exceptional diversity in catalysis.
Sixty years have passed since I decided on the branch of science I would pursue. During that time, technological achievements and increasing overlap among
disciplines have made advances possible in fields that were poorly understood, and
I have been fortunate to share in such progress. In addition to the thrill of research
discoveries, I have enjoyed friendships and interactions with students, postdoctoral
associates, and other collaborators. These have included biochemists, pharmacologists, toxicologists, chemists, biophysicists, molecular biologists, and occasionally
3 Dec 2004
20:34
18
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
COON
LITERATURE CITED
1. Coon MJ. 2002. Enzyme ingenuity in
biological oxidations: a trail leading to
cytochrome P450. J. Biol. Chem. 277:
2835163
2. Coon MJ, Gurin S. 1949. Studies on the
conversion of radioactive leucine to acetoacetate. J. Biol. Chem. 180:115967
3. Coon MJ. 1950. The metabolic fate of the
isopropyl group of leucine. J. Biol. Chem.
187:7182
4. Bachhawat BK, Robinson WG, Coon MJ.
1954. Carbon dioxide fixation in heart extracts by -hydroxyisovaleryl coenzyme
A. J. Am. Chem. Soc. 76:3098
5. del Campillo-Campbell A, Dekker EE,
Coon MJ. 1959. Carboxylation of methylcrotonyl coenzyme A by a purified
enzyme from chicken liver. Biochim. Biophys. Acta 31:29092
6. Coon MJ, Kupiecki FP, Dekker EE,
Schlesinger MJ, del Campillo A. 1959.
The enzymic synthesis of branched-chain
acids. In Ciba Foundation Symposium on
the Biosynthesis of Terpenes and Sterols,
ed. GEW Wolstenholme, M OConnor,
pp. 6272. London: J & A Churchill
7. Dutler H, Coon MJ, Kull A, Vogel H,
Waldvogel G, Prelog V. 1971. Fatty acid
synthetase from pig liver. 1. Isolation of
the enzyme complex and characterization of the component with oxidoreductase activity for alicyclic ketones. Eur. J.
Biochem. 22:20312
8. Ochoa S. 1980. The pursuit of a hobby.
Annu. Rev. Biochem. 49:130
9. Verkade PE, Elzas M, van der Lee J, de
Wolff HH, Verkade-Sandbergen A, van
10.
11.
12.
13.
14.
15.
16.
17.
3 Dec 2004
20:34
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
19
Properties of a yeast cytochrome P-450containing enzyme system which catalyzes the hydroxylation of fatty acids,
alkanes, and drugs. Eur. J. Biochem. 36:
58392
Ryan KJ, Engel LL. 1957. Hydroxylation
of steroids at carbon 21. J. Biol. Chem.
225:10314
Klingenberg M. 1958. Pigments of rat
liver microsomes. Arch. Biochem. Biophys. 75:37686
Garfinkel D. 1958. Studies on pig liver microsomes. I. Enzymic and pigment composition of different microsomal fractions. Arch. Biochem. Biophys. 77:493
509
Omura T, Sato R. 1962. A new cytochrome in liver microsomes. J. Biol.
Chem. 237:PC137576
Omura T, Sato R, Cooper DY, Rosenthal
O, Estabrook RW. 1965. Function of cytochrome P-450 of microsomes. Federation Proc. 24:118189
Lu AYH. 1998. The Bernard B. Brodie
lecture. A journey in cytochrome P450
and drug metabolism research. Drug
Metab. Dispos. 26:116873
Gillette JR, Conney AH, Cosmides GJ,
Estabrook RW, Fouts JR, Mannering GJ,
eds. 1969. Microsomes and Drug Oxidations. New York: Academic. 547 pp.
Coon MJ, Lu AYH. 1969. Fatty acid oxidation in a soluble microsomal enzyme
system containing P-450. See Ref. 33,
pp. 15166
Gunsalus IC, Sligar SG. 1978. Oxygen
reduction by the P450 monooxygenase
systems. Adv. Enzyme Regul. 47:144
Coon MJ, Sligar SG. 2003. Irwin C.
Gunsalus, talented and creative scientist.
Biochem. Biophys. Res. Commun. 312:1
6
Ruckpaul K. 2003. Early years of
cytochrome P450 research in BerlinBuch: its present state and origin of
the biochemical and biophysical conferences. Biochem. Biophys. Res. Commun.
312:6574
3 Dec 2004
20:34
20
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
COON
38. Maricic S. 1977. Foreward to proceedings of the Scientific Conference on Cytochrome P-450structural aspects (Primosten, Yugoslavia), ed. MJ Coon, IC
Gunsalus, S Maricic. Croat. Chem. Acta
49:163388
39. Conney AH. 2003. Induction of drugmetabolizing enzymes: a path to the discovery of multiple cytochromes P450.
Annu. Rev. Pharmacol. Toxicol. 43:1
30
40. Nebert DW, Heidema JK, Strobel HW,
Coon MJ. 1973. Genetic expression of
aryl hydrocarbon hydroxylase induction.
Genetic specificity resides in the fraction
containing cytochrome P448 and P450. J.
Biol. Chem. 248:763136
41. Lu AYH, Kuntzman R, West S, Conney AH. 1972. Reconstituted liver microsomal enzyme system that hydroxylates
drugs, other foreign compounds, and endogenous substrates. I. Determination of
substrate specificity by the cytochrome P450 and P-448 fractions. Biochem. Biophys. Res. Commun. 42:12006
42. Autor AP, Kaschnitz RM, Heidema JK,
Coon MJ. 1973. Sedimentation and
other properties of the reconstituted
liver microsomal mixed-function oxidase system containing cytochrome P450,
reduced triphosphopyridine nucleotidecytochrome P450 reductase, and phosphatidylcholine. Mol. Pharmacol. 9:93
104
43. Coon MJ, van der Hoeven TA, Kaschnitz
RM, Strobel HW. 1973. Biochemical
studies on cytochrome P-450 solubilized
from liver microsomes: partial purification and mechanism of catalysis. Ann. NY
Acad. Sci. 212:44957
44. van der Hoeven TA, Coon MJ. 1974.
Preparation and properties of partially
purified cytochrome P-450 and reduced nicotinamide adenine dinucleotide
phosphate-cytochrome P-450 reductase
from rabbit liver microsomes. J. Biol.
Chem. 249:630210
45. Haugen DA, van der Hoeven TA, Coon
46.
47.
48.
49.
50.
51.
52.
53.
54.
MJ. 1975. Purified liver microsomal cytochrome P-450. Separation and characterization of multiple forms. J. Biol.
Chem. 250:356770
Haugen DA, Coon MJ. 1976. Properties of electrophoretically homogeneous phenobarbital-inducible and naphthoflavone-inducible forms of liver
microsomal cytochrome P-450. J. Biol.
Chem. 251:792939
Nebert DW, Adesnik M, Coon MJ, Estabrook RW, Gonzalez FJ, et al. 1987. The
P450 gene superfamily: recommended
nomenclature. DNA 6:111
Dean WL, Coon MJ. 1977. Immunochemical studies on two electrophoretically homogeneous forms of rabbit liver microsomal cytochrome P-450: P-450LM2 and
P-450LM4. J. Biol. Chem. 252:325561
Park SS, Persson AV, Coon MJ, Gelboin
HV. 1980. Monoclonal antibodies to rabbit liver cytochrome P-450LM. FEBS Lett.
116:23135
Chiang YL, Coon MJ. 1979. Comparative study of two highly purified forms
of liver microsomal cytochrome P-450:
circular dichroism and other properties.
Arch. Biochem. Biophys. 195:17887
Inouye K, Coon MJ. 1985. Properties of
the tryptophan residue in rabbit liver microsomal cytochrome P-450 isozyme 2
as determined by fluorescence. Biochem.
Biophys. Res. Commun. 128:67682
Haugen DA, Coon MJ, Nebert DW. 1976.
Induction of multiple forms of mouse
liver cytochrome P-450. Evidence for genetically controlled de novo protein synthesis in response to treatment with naphthoflavone or phenobarbital. J. Biol.
Chem. 251:181727
Zaphiropoulos PG, Folk WR, Coon MJ.
1986. Isolation and characterization of
a novel cytochrome P-450-like pseudogene. Biochem. Biophys. Res. Commun.
134:499505
Haugen DA, Armes LG, Yasunobu
KT, Coon MJ. 1977. Amino-terminal
sequence of phenobarbital-inducible
3 Dec 2004
20:34
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
21
cytochrome P-450-containing monooxygenase system by second derivative spectroscopy. Biochim. Biophys. Acta 626:41
56
Vatsis KP, Theoharides AD, Kupfer D,
Coon MJ. 1982. Hydroxylation of prostaglandins by inducible isozymes of rabbit liver microsomal cytochrome P-450:
participation of cytochrome b5. J. Biol.
Chem. 257:1122129
Morgan ET, Coon MJ. 1984. Effects
of cytochrome b5 on cytochrome P450-catalyzed reactions. Studies with
manganese-substituted cytochrome b5.
Drug Metab. Dispos. 12:35864
Pompon D, Coon MJ. 1984. On the mechanism of action of cytochrome P-450.
Oxidation and reduction of the ferrous
dioxygen complex of liver microsomal cytochrome P-450 by cytochrome b5. J. Biol.
Chem. 259:1537785
Gorsky LD, Coon MJ. 1986. Effects
of conditions for reconstitution with cytochrome b5 on the formation of products
in cytochrome P-450-catalyzed reactions.
Drug Metab. Dispos. 14:8996
Koop DR, Coon MJ. 1979. Purification
and properties of P-450LM3b, a constitutive form of cytochrome P-450, from rabbit liver microsomes. Biochem. Biophys.
Res. Commun. 91:107581
Koop DR, Persson AV, Coon MJ. 1981.
Properties of electrophoretically homogeneous, constitutive forms of liver microsomal cytochrome P-450. J. Biol. Chem.
256:1070411
Koop DR, Coon MJ. 1984. Purification of liver microsomal cytochrome P450 isozymes 3a and 6 from imidazoletreated rabbits. Evidence for the identity
of isozyme 3a with the form obtained
by ethanol treatment. Mol. Pharmacol.
25:494501
Ishida H, Noshiro M, Okuda K, Coon MJ.
1992. Purification and characterization
of 7-hydroxy-4-cholesten-3-one 12hydroxylase. J. Biol. Chem. 267:21319
23
3 Dec 2004
20:34
22
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
COON
81.
82.
83.
84.
85.
86.
87.
88.
89.
3 Dec 2004
20:34
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
101.
102.
103.
104.
105.
106.
107.
108.
23
3 Dec 2004
20:34
24
109.
110.
111.
112.
113.
114.
115.
116.
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
COON
P-450. Arch. Biochem. Biophys. 175:524
33
Ellin A, Orrenius S. 1975. Hydroperoxide-supported cytochrome P-450linked fatty acid hydroxylation in liver
microsomes. FEBS Lett. 50:37881
Blake RC II, Coon MJ. 1981. On the
mechanism of action of cytochrome P450. Evaluation of homolytic and heterolytic mechanisms of oxygen-oxygen
bond cleavage during substrate hydroxylation by peroxides. J. Biol. Chem.
256:1212733
Blake RC II, Coon MJ. 1989. On the
mechanism of action of cytochrome P450. Spectral intermediates in the reaction
with iodosobenzene and its derivatives. J.
Biol. Chem. 264:3694701
Osawa Y, Coon MJ. 1989. Selective
mechanism-based inactivation of the major phenobarbital-inducible cytochrome
P-450 from rabbit liver by phencyclidine
and its oxidation product, the iminium
compound. Drug Metab. Dispos. 17:713
Vaz ADN, Kessell KJ, Coon MJ. 1994.
Aromatization of a bicyclic steroid
analog, 3-oxodecalin-4-ene-10-carboxaldehyde, by liver microsomal cytochrome P450 2B4. Biochemistry 33:
1365161
Vaz ADN, Coon MJ, Peegel H, Menon
KMJ. 1992. Substituted pyridines: nonsteroidal inhibitors of human placental aromatase cytochrome P450. Drug
Metab. Dispos. 20:10812
Vaz ADN, Pernecky SJ, Raner GM, Coon
MJ. 1996. Peroxo-iron and oxenoid-iron
species as alternative oxygenating agents
in cytochrome P450-catalyzed reactions:
switching by T302A mutagenesis of cytochrome P450 2B4. Proc. Natl. Acad.
Sci. USA 93:464448
Vaz ADN, McGinnity DF, Coon MJ.
1998. Epoxidation of olefins by cytochrome P450: evidence from sitespecific mutagenesis for hydroperoxoiron as an electrophilic oxidant. Proc.
Natl. Acad. Sci. USA 95:355560
3 Dec 2004
20:34
AR
AR232-PA45-01.tex
AR232-PA45-01.sgm
LaTeX2e(2002/01/18)
P1: GCE
25
3 Dec 2004
20:35
AR
AR232-PA45-02.tex
AR232-PA45-02.sgm
P1: GCE
LaTeX2e(2002/01/18)
10.1146/annurev.pharmtox.45.120403.100010
INTRODUCTION
Many arylamines, e.g., 2-naphthylamine (2-NA), benzidine, and 4-aminobiphenyl
(4-ABP), are of industrial importance because of their use as intermediates in
the synthesis of azo dyes, antioxidants in rubber products, and other commercial
materials (1, 2). Epidemiological observations of the toxicity of arylamines were
first reported in aniline dye factories by Rehn in 1895, with the report that German
and Swiss workers suffered urinary bladder tumors (2, 3). A major toxicological
issue is reaction with DNA and induction of carcinomas, primarily in the urinary
bladder, liver, or other tissues in humans and experimental animals (1, 2, 46).
In 1939, the Swedish chemist Widmark demonstrated that extracts of fried horse
meat induced cancer when applied to mouse skin (7, 8). Sugimura and his associates
investigated the smoke produced by broiling fish and meat; they demonstrated that
the smoke condensate and charred surfaces of broiled fish and meat were highly
mutagenic in Salmonella typhimurium test systems (Table 1) (911). Subsequently,
the heterocyclic arylamine (HAA) products formed as a consequence of pyrolysis
of amino acids or protein-containing foods were isolated, their structures were
0362-1642/05/0210-0027$14.00
27
3 Dec 2004
20:35
28
AR
KIM
TABLE 1
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
GUENGERICH
HAA
TA98
2-amino-3-methylimidazo[4,5-f]quinoline (IQ)
433,000
7000
2-amino-3,5-dimethylimidazo[4,5-f]quinoline (MeIQ)
661,000
30,000
75,000
1500
2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline (MeIQx)
145,000
14,000
2-amino-3,4,8-trimethylimidazo[4,5-f]quinoxaline (4,8-DiMeIQx)
183,000
8000
2-amino-3,7,8-trimethylimidazo[4,5-f]quinoxaline (7,8-DiMeIQx)
163,000
9900
2-amino-3-methylimidazo[4,5-f]quinoline (IQx)
TA100
2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP)
1800
120
3-amino-1,4-dimethyl-5H-pyrido[4,3,-b]-indole (Trp-P-1)
39,000
1700
104,000
1800
49,000
3200
1900
1200
41
23
56,800
3-amino-1-methyl-5H-pyrido[4,3,-b]-indole (Trp-P-2)
2-amino-6-methyldipyrido[1,2-a:3 ,2 -d]imidazole (Glu-P-1)
Reference 11.
CHEMISTRY OF BIOACTIVATION
Arylamines and HAAs require metabolic activation to be mutagenic or carcinogenic (31). The major metabolic process is N-oxidation, which is mediated primarily by cytochrome P450 (P450) enzymes but also by flavin-containing monooxygenases (FMOs) and peroxidases (3139). The resulting N-hydroxylamine
products can be further activated to produce highly reactive ester derivatives that
20:35
AR
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
3 Dec 2004
P1: GCE
29
3 Dec 2004
20:35
30
AR
KIM
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
GUENGERICH
bind covalently to DNA. At least four enzyme systems are known to be involved in
this secondary activation step in mammals: N-acetyltransferase (NAT), sulfotransferase, prolyl tRNA synthetase, and kinases, yielding reactive N-acetoxy, N-sulfonyloxy, N-prolyloxy, and N-phosphatyl esters, respectively (4043). The N-hydroxy HAAs can react directly with DNA (32), but the reaction is facilitated when
reactive ester derivatives undergo heterocyclic cleavage to yield reactive aryl nitrenium ion species, which preferentially react to form DNA adducts (Figure 2).
NAT-catalyzed acetylation of N-hydroxy HAAs and arylamines enhances genotoxic activity and DNA adduct levels through formation of reactive N-acetoxyl
esters (4448). In a similar way, the sulfur esters formed by the action of sulfotransferases are unstable and react (42, 49). The role of the sulfotransferases
has been given less attention than NAT in human epidemiology studies. Even less
information is available about the in vivo roles of the prolyloxy and phosphatyl
esters.
Arylamines and HAAs yield adducts primarily with guanine (50, 51), reacting
at the N2 and C8 atoms. Wild et al. (52) showed that the photoactivated azide
derivatives of IQ, MeIQx, and PhIP bind to DNA to form the same adducts as
the N-acetoxy species, indicating that the nitrenium ion may be a common intermediate for both reactive intermediates. Mechanisms for the reaction at the C8
atom have been less clear (53). A direct reaction is possible (54), and a stepwise
mechanism via an N7-guanyl intermediate has also been proposed (55, 56). The
latter has an advantage of also explaining several accompanying products (e.g.,
8-oxo-7,8-dihydroguanine, depurination, imidazole ring opening) (56).
Several approaches to syntheses of DNA adducts of these amines have been
reported. HAA-DNA adducts (including IQ, MeIQ, MeIQx, 4,8-DiMeIQx, PhIP,
Glu-P-1, and Trp-P-2) have been synthesized and characterized spectroscopically
(5763). Oligonucleotides containing guanyl-C8 2-aminofluorene (2-AF) (and Nacetyl 2-AF) derivatives have been synthesized, and structures have been determined using NMR spectroscopy and mutagenic properties have been examined
in cell-based systems (Figure 3). PhIP has been site-specifically incorporated into
oligonucleotides by a biomimetic approach in which N-acetoxy-PhIP was reacted
with oligonucleotides containing a single guanine (64). In a nonbiomimetic approach, the Rizzo group has reported the synthesis of C8-guanyl adduct of IQ
through palladium-catalyzed N-arylation of a protected 8-bromo-2 -dG derivative
with IQ as the key reaction (65).
3 Dec 2004
20:35
AR
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
Figure 2 General pathway for activation of arylamines and HAAs, as shown for IQ.
31
3 Dec 2004
20:35
32
AR
KIM
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
GUENGERICH
mutation response is also found in other bacterial genes, such as lacZ, lacZ, and
lacI of Escherichia coli (71, 72). An E. coli lacZ reversion mutation assay has
also been applied to study HAA genotoxicity (7376); a tester stain carrying a
(-GC) copy of lacZ gene can regain functional lacZ activity following induction
of frameshifts by HAAs (76, 77). Systems have also been developed that incorporate the heterologous expression of P450s and NADPH-P450 reductase (74).
This system allows the detection of HAA mutagenicity by recombinant human
P450 without a need for rat liver fractions. These bacteria have also been genetically engineered to express S. typhimurium NAT, and the DNA nucleotide excision
repair system has been inactivated (UvrABC) to improve the sensitivity (74, 77,
78). S. typhimurium tester strains have also been used to express human P450s
that activate arylamines and HAAs (73, 79, 80). The E. coli strains overexpressing P450s and NAT have been used to characterize P450 1A2 allelic and random
variants (8184). Another use of this genotoxicity system has been the screening
and characterization of P450 inhibitors. For instance, a P450 1B1based system
was used to characterize the potent inhibition of the enzyme by tetramethoxystilbene (85) and a P450 1A2based system was sensitive to the drug oltipraz (86).
Other bacterial systems have utilized the SOS response in S. typhimurium NM2009
as a measure of DNA damage (47). This strain contains a plasmid-based umuDC
gene linked to a lacZ reporter gene and is activated by induction of the SOS pathway (47). Sensitivity to arylamines and HAAs was also improved by incorporating
plasmids coding for P450 enzymes, NADPH-P450 reductase, and NAT (48).
DNA adducts are considered biomarkers of potential mutagenesis and carcinogenesis (87, 88), and HAAs are thought to induce mutagenesis by producing
mutations in oncogenes and tumor suppressor genes in experimental animals (23,
89). Although base pair mutations are not a major feature of HAAs in bacterial test
systems, one was dominantly induced by IQ in the p53 gene in rat Zymbals gland
tumors and monkey hepatocellular carcinoma (89, 90). Also, a C T transversion
3 Dec 2004
20:35
AR
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
33
in rasH was predominantly observed in mouse forestomach and rat Zymbals gland
tumors induced by MeIQ. However, PhIP was reported to induce a number of characteristic one-base deletions in the lacI gene of the colon mucosa of transgenic
mice (91).
The major analytical methods used for the detection and quantification of arylamine and HAA adducts are the 32P-postlabeling assay and mass spectrometry.
32
P-postlabeling assays use polynucleotide kinase to label the adducted or nonadducted 3 -nucleotides with [ -32P]ATP after digestion of DNA to mononucleotides
(92). The radiolabeled 3 , 5 -bisphosphonucleotide adducts can be separated by
two-dimensional thin-layer chromatography (TLC) or high-pressure liquid chromatography (HPLC) (93). A combination of HPLC and electrospray ionizationtandem mass spectrometry can provide structural information on adducts, and the
incorporation of stable isotopically labeled internal standards in assays provides
precise and accurate quantification of the DNA adducts (93).
N-Hydroxylamines and other metabolites of arylamines and HAAs are mainly
identified and measured with combinations of HPLC, using [3H]- or [14C]radiolabeled substrates, and mass spectrometry (36, 37, 94, 95). Care is necessary because of the instability of these oxidized amine species. The in vitro
N-hydroxylation of arylamines and HAAs can be measured colorimetrically by a
modification of a Fe3+-reduction method using 4,7-diphenyl-1,10-phenanthroline
(83, 84, 96). Owing to the inherent instability of aryl N-hydroxylamines and the
large numbers of assays required in enzyme kinetic studies (83), this method can
be used quite sensitively and routinely.
Many cancer studies on arylamines have been done following the initial epidemiological findings in workers (2, 3), beginning with the induction of urothelial
tumors in dogs with 2-NA by Hueper (97).
The carcinogenicity of HAAs has been intensively studied in rodent models.
HAAs induce tumors at multiple organs in rats and mice, including liver, lung,
colon, small and large intestine, forestomach, the hematopoietic system, prostate,
mammary gland, Zymbals gland, clitoral gland, oral cavity, and urinary bladder
(20, 98100).
ACTIVATION BY P450
Studies of arylamine oxidation go back to the 1940s, with early work on azo dyes by
the Millers (101). Several lines of evidence implicated the N-hydroxyl derivatives
of the arylamines as being responsible for carcinogenic activity, as well as for the
methemoglobenemia induced by some drugs (2, 31). N-Hydroxylation was first
demonstrated with the acetamide derivative of 2-AF (102). Subsequently, P450 was
demonstrated to be involved in this reaction (103), and further studies extended
the work to unsubstituted arylamines (31, 96).
In early work on the multiplicity of P450s, several lines of investigation had suggested the role of Ah locus-linked P450 enzymes (i.e., now recognized as Family 1)
in the N-hydroxylation of 2-acetylaminofluorene (AAF) (104, 105). Subsequently,
3 Dec 2004
20:35
34
AR
KIM
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
GUENGERICH
these P450s were recognized as being involved in the N-hydroxylation of this substrate and several arylamines in various species (32, 33, 106). This is generally
the case, with many arylamines preferentially N-hydroxylated by rat P450 1A2
and, to a lesser extent, by P450 1A1 (36). However, 4-ABP is N-hydroxylated by
rat P450 proteins in the order 1A2 > 1A1 > 2A6 > 2C11 > 2B1 (36). In the
same study, the order for N-hydroxylation of 4,4 -methylene-bis(2-chloroaniline)
(MOCA) was 2B1 > 2B2 > 1A2 > 2A6 > 1A1.
The human P450 enzymes involved in the metabolism of arylamines, HAAs,
and other chemical carcinogens have long been a subject of interest. Some efforts
had been made at analysis with early preparations of human P450s (107). A key
development was the purification of the human P450 involved in phenacetin Odeethylation, now recognized as P450 1A2 (108). Analysis of the animal models
and subsequent correlations of hepatic expression levels with the N-hydroxylation
of 4-ABP (36) led to the view that this enzyme, then known as P450PA, has a major role in the N-hydroxylation of many arylamines and HAAs. Further evidence
followed, with the demonstration that the same enzyme is involved in caffeine
N3-demethylation (37) and that many HAA activations can be attributed to this enzyme (109). Phenacetin metabolism had been studied in humans in vivo (110), and
the characterization of human P450 1A2 (108) led to insight into the inducibility
of P450 1A2 in humans. However, phenacetin can no longer be used as a human
drug because of concerns about its carcinogenicity, and the demonstration of caffeine N3-demethylation led to the use of caffeine as a noninvasive in vivo probe
(37, 111).
The roles of human P450s in the bioactivation of arylamines and HAAs have
been considerably documented (24, 37, 109, 112116). P450s 1A1 and 1A2 have
been generally recognized to be the major forms involved in the bioactivation of
arylamines and HAAs in human liver and lung microsomes (109, 113, 116). A representative study with HAAs is presented in Table 2. The findings with P450 1A2
have been confirmed in vivo in human studies, at least with PhIP and MeIQx. The
P450 1A2selective inhibitor furafylline blocked most of the in vivo elimination
in studies in which the human volunteers consumed burned meat (117). Another
P450 Family 1 member, P450 1B1, has been also shown to be an important enzyme
involved in the activation of HAAs and has been considered regarding mechanisms
of development of human cancers (Table 2) (118, 119). It should be emphasized
that some of the P450s (other than Family 1) do have measurable activity with some
of the arylamine and HAA substrates. MOCA N-hydroxylation, in contrast with
other arylamines, was shown to be preferentially catalyzed by P450 3A4 in human
liver (114). Kitada & Kamataki (120) have shown that P450 3A7 can activate some
HAAs to mutagens in human fetal liver, where P450 1A2 is not expressed.
Although the early epidemiology linked P450 2D6 with lung cancer incidence
(121), subsequent efforts to provide a basis were not fruitful. We have been unable
to find any carcinogens that are preferentially activated by P450 2D6, including
arylamines, HAAs, and crude cigarette smoke fractions (112, 122).
We have treated P450 1A2 (and other P450s) only in terms of the wild-type
(or more correctly, the predominant) allele thus far. The possibility exists that
3 Dec 2004
20:35
AR
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
35
HAA
Glu-P-1
PhIP
1.5
220
1A2
1B1
2C9
2D6
2E1
3A4
27
91
10
13
24
MeIQx
0.3
16
442
MeIQ
0.05
24
179
IQ
0.5
10
214
Trp-P-1
536
321
232
179
Trp-P-2
578
39
138
555
2-AA
0.1
90
374
50
23
2-AF
12.5
46
676
22
13
25
Reference 48.
Lowest concentration used in assay. See original reference for more results (48).
some individuals will have variants that provide unusual catalytic properties. For
instance, we have utilized screens involving the activation of MeIQ to a genotoxic
N-hydroxylamine to identify P450 1A2 mutants with high activity in laboratorygenerated random libraries (81, 83). Some of these variants have activities (Nhydroxylation) 12-fold higher than the wild type. To our knowledge, none of these
has been reported in the population. We have expressed the known allelic variants
and found catalytic efficiencies (kcat/Km) for N-hydroxylation of several HAAs
within a threefold range, although one P450 1A2 variant failed to incorporate
heme and was inactive (84).
Although the discussion is mainly about the activation of arylamines and HAAs,
we should also emphasize that P450s are involved in detoxication of the same
compounds by other routes (36, 123). This aspect can be important, as shown
in the classic Richardson experiment (124) in which enzyme induction by 3methylcholanthrene lowered the tumorigenicity of an aminoazobenzene compound
to rats.
A study with P450 1A2/ mice indicated that P450 1A2 plays an important
role in DNA adduct formation with PhIP and IQ in vivo (125). Differences owing
to the absence/presence of P450 1A2 were seen in liver, kidney, and colon but
not in mammary glands. However, a neonatal bioassay study with P450 1A2/
mice suggests that an unknown pathway, unrelated to P450 1A2, appears to be
responsible for the carcinogenesis of PhIP (126).
Interspecies differences in metabolism of HAAs by rat and human P450 1A2
were found in the metabolism of MeIQx and PhIP (127, 128). Although rat and
human P450 1A2 have 75% amino acid sequence identity (129), relatively high
levels of P450 1A2 expression in human liver and catalytic activities for HAA
3 Dec 2004
20:35
36
AR
KIM
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
GUENGERICH
N-hydroxylation compared to the rat P450 1A2 were observed (127, 128). Important differences between human and rat P450s 1A2 were also found in the C8- and
N-oxidation of MeIQx (130, 131). These suggest that the interspecies differences
in P450 enzyme expression and catalytic activities may be significant and must be
carefully considered when assessing human health risk.
The carcinogenicity of IQ, MeIQ, and PhIP was examined in cynomologous
monkeys, with up to seven years of administration (132134). IQ and PhIP were
potent liver carcinogens. IQ and PhIP formed high levels of DNA adducts in a
number of organs, particularly the liver, kidney, and heart. However, low mutagenic and carcinogenic activation of MeIQx was observed in this species. The
poor activation of MeIQx was explained by the lack of constitutive expression
of P450 1A2 and an inability of other P450s to hydroxylate this quinoxaline
(133).
Arylamines can also be N-hydroxylated by FMO and peroxidases (32, 33, 135,
136). These reports, along with those demonstrating DNA adduct formation by
these routes, suggested that the formation of a common DNA-reactive species
could be generated by alternate pathways and these pathways could be considered
to contribute to the burden of DNA adducts found in extrahepatic tissues. This
appears to be the case because P450 1A2 is not expressed in extrahepatic tissues,
and some N-hydroxy HAAs are too unstable to be transported from the liver to
distant sites (3335, 136, 137), although contributions of P450s such as 1A1 and
1B1 may also be an issue.
3 Dec 2004
20:35
AR
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
37
Peroxidases generate nitro compounds from HAAs as well as N-hydroxy products (141, 142). Nitroaromatic hydrocarbons, which are abundant in the environment and can cause cancer (143), undergo nitroreduction to produce N-hydroxy
intermediates in bacteria or in mammalian cells and then follow the same process
for mutagenesis as described for amines (77, 112, 136).
3 Dec 2004
20:35
38
AR
KIM
AR232-PA45-02.tex
Figure 5
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
GUENGERICH
Metabolism of SMX.
3 Dec 2004
20:35
AR
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
39
LITERATURE CITED
1. Radomski JL. 1979. The primary aromatic amines: their biological properties
and structure-activity relationships. Annu.
Rev. Pharmacol. 19:12957
2. Garner RC, Martin CN, Clayson DB.
1984. Carcinogenic aromatic amines and
related compounds. In Chemical Carcinogens, ed. CE Searle, 1:175276. Washington, DC: Amer. Chem. Soc. 2nd ed.
6.
7.
8.
9.
3 Dec 2004
20:35
40
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
AR
KIM
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
GUENGERICH
20.
21.
22.
23.
24.
25.
26.
27.
28.
DiMeIQx from the model system fructose, alanine and creatinine. Comparison
with the isomeric 5,8-DiMeIQx. Mutat.
Res. 174:59
Eisenbrand G, Tang W. 1993. Food-borne
heterocyclic amines. Chemistry, formation, occurrence and biological activities.
A literature review. Toxicology 84:182
Sugimura T. 1986. Past, present, and future of mutagens in cooked foods. Environ. Health Perspect. 67:510
Felton JS, Jagerstad M, Knize MG, Skog
K, Wakabayashi K. 2000. Contents in
foods, beverages and tobacco. In Food
Borne Carcinogens Heterocyclic Amines,
ed. M Nagao, T Sugimura, pp. 3171.
West Sussex, UK: Wiley
Turesky RJ. 2002. Heterocyclic aromatic
amine metabolism, DNA adduct formation, mutagenesis, and carcinogenesis.
Drug Metab. Rev. 34:62550
Shimada T, Guengerich FP. 1991. Activation of amino--carboline, 2-amino-1methyl -6 -phenylimidazo[4,5-b]pyridine,
and a copper phthalocyanine cellulose
extract of cigarette smoke condensate by
cytochrome P-450 enzymes in rat and
human liver microsomes. Cancer Res.
51:528491
Manabe S, Tohyama K, Wada O, Aramaki T. 1991. Detection of a carcinogen,
2-amino-1-methyl-6-phenylimidazo[4,5b]pyridine (PhIP), in cigarette smoke
condensate. Carcinogenesis 12:194547
Svensson CK. 2003. Do arylhydroxylamine metabolites mediate idiosyncratic
reactions associated with sulfonamides?
Chem. Res. Toxicol. 16:103543
Naisbitt DJ. 2004. Drug hypersensitivity
reactions in skin: understanding mechanisms and the development of diagnostic
and predictive tests. Toxicology 194:179
96
Gorlewska-Roberts K, Green B, Fares
M, Ambrosone CB, Kadlubar FF. 2002.
Carcinogen-DNA adducts in human
breast epithelial cells. Environ. Mol. Mutagen. 39:18492
3 Dec 2004
20:35
AR
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
37.
38.
39.
40.
41.
42.
43.
44.
41
3 Dec 2004
20:35
42
45.
46.
47.
48.
49.
50.
AR
KIM
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
GUENGERICH
3 Dec 2004
20:35
AR
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
43
3 Dec 2004
20:35
44
75.
76.
77.
78.
79.
80.
81.
82.
83.
AR
KIM
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
GUENGERICH
84.
85.
86.
87.
88.
89.
90.
91.
92.
3 Dec 2004
20:35
AR
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
93.
94.
95.
96.
97.
98.
99.
100.
101.
45
3 Dec 2004
20:35
46
110.
111.
112.
113.
114.
115.
116.
117.
118.
AR
KIM
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
GUENGERICH
119.
120.
121.
122.
123.
124.
125.
126.
3 Dec 2004
20:35
AR
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
127.
128.
129.
130.
131.
132.
133.
47
3 Dec 2004
20:35
48
144.
145.
146.
147.
148.
149.
150.
151.
AR
KIM
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
GUENGERICH
carcinogens that may be involved in human breast cancer etiology. Chem. Res.
Toxicol. 5:58590
Klehr H, Eyer P, Schafer W. 1987. Formation of 4-ethoxy-4 -nitrosodiphenylamine
in the reaction of the phenacetin metabolite 4-nitrosophenetol with glutathione.
Biol. Chem. Hoppe-Seyler 368:895902
Eyer P. 1988. Detoxication of Noxygenated arylamines in erythrocytes.
An overview. Xenobiotica 18:132733
Mandell GL, Sande MA. 1985. Antimicrobial agents: sulfonamides, trimethoprim-sulfamethoxazole and agents for
urinary tract infections. In The Pharmacological Basis of Therapeutics, ed.
AG Gillam, LS Goodman, TW Rall,
R Marud, pp. 1095114. New York:
Macmillan
Cribb AE, Spielberg SP, Griffin GP.
1995. N4-hydroxylation of sulfamethoxazole by cytochrome P450 of the cytochrome P4502C subfamily and reduction of sulfamethoxazole hydroxylamine
in human and rat hepatic microsomes.
Drug Metab. Dispos. 23:40614
Naisbitt DJ, Hough SJ, Gill HJ, Pirmohamed M, Kitteringham NR, et al. 1999.
Cellular disposition of sulphamethoxazole and its metabolites: implications
for hypersensitivity. Br. J. Pharmacol.
126:1393407
Cribb AE, Nuss CE, Alberts DW, Lamphere DB, Grant DM, et al. 1996. Covalent binding of sulfamethoxazole reactive
metabolites to human and rat liver subcellular fractions assessed by immunochemical detection. Chem. Res. Toxicol. 9:500
7
Carr A, Tindall B, Penny R, Cooper DA.
1993. In vitro cytotoxicity as a marker of
hypersensitivity to sulphamethoxazole in
patients with HIV. Clin. Exp. Immunol.
94:2125
Meekins CV, Sullivan TJ, Gruchalla
RS. 1994. Immunochemical analysis of
sulfonamide drug allergy: identification
of sulfamethoxazole-substituted human
152.
153.
154.
155.
156.
157.
158.
159.
160.
3 Dec 2004
20:35
AR
AR232-PA45-02.tex
AR232-PA45-02.sgm
LaTeX2e(2002/01/18)
P1: GCE
161.
162.
163.
164.
49
7 Jan 2005
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
P1: GCE
LaTeX2e(2002/01/18)
10.1146/annurev.pharmtox.45.120403.095857
GLUTATHIONE TRANSFERASES
0362-1642/05/0210-0051$14.00
51
7 Jan 2005
13:55
52
AR
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
LaTeX2e(2002/01/18)
P1: GCE
INTRODUCTION
The glutathione transferases (EC 2.5.1.18) have historically also been called glutathione S-transferases, and it is this latter name that gives rise to the widely
used abbreviation, GST. These enzymes catalyze nucleophilic attack by reduced
glutathione (GSH) on nonpolar compounds that contain an electrophilic carbon,
nitrogen, or sulphur atom. Their substrates include halogenonitrobenzenes, arene
oxides, quinones, and ,-unsaturated carbonyls (15). Three major families of
proteins that are widely distributed in nature exhibit glutathione transferase activity.
Two of these, the cytosolic and mitochondrial GST, comprise soluble enzymes that
are only distantly related (6, 7). The third family comprises microsomal GST and
is now referred to as membrane-associated proteins in eicosanoid and glutathione
(MAPEG) metabolism (8). A further distinct family of transferases exists, represented by the bacterial fosfomycin resistance proteins FosA and FosB (9); this
family is not discussed further.
Cytosolic and mitochondrial GST share some similarities in their three-dimensional fold (6) but bear no structural resemblance to the MAPEG enzymes (10).
However, all three families contain members that catalyze the conjugation of
GSH with 1-chloro-2,4-dinitrobenzene (CDNB) and exhibit glutathione peroxidase activity toward cumene hydroperoxide (CuOOH); these reactions are shown in
Figure 1. The cytosolic GST and MAPEG enzymes catalyze isomerization of various unsaturated compounds (8, 11, 12) and are intimately involved in the synthesis
of prostaglandins and leukotrienes (4, 8).
Cytosolic GSTs represent the largest family of such transferases and have activities that are unique to this group of enzymes. They catalyze thiolysis of 4nitrophenyl acetate; display thiol transferase activity; reduce trinitroglycerin, dehydroascorbic acid, and monomethylarsonic acid; and catalyze the isomerization
of maleylacetoacetate and 5-3-ketosteroids (Figure 1) (1, 1317).
Glutathione transferases are of interest to pharmacologists and toxicologists
because they provide targets for antiasthmatic and antitumor drug therapies (18
21), and they metabolize cancer chemotherapeutic agents, insecticides, herbicides, carcinogens, and by-products of oxidative stress. Overexpression of GST
in mammalian tumor cells has been implicated with resistance to various anticancer agents and chemical carcinogens (2). Furthermore, elevated levels of GST
have been associated with tolerance of insecticides and with herbicide selectivity
(22, 23).
In microbes, plants, flies, fish, and mammals, expression of GSTs are upregulated by exposure to prooxidants (2430). Increase in transferase activity is also
observed in animals that undergo prolonged torpor or hibernation when comparisons are made between their estivated state and their wakeful condition (31). It
is similarly observed during transition of the common toad Bufo bufo from an
aquatic environment to the land (32). Collectively, these findings indicate that
induction of GST is an evolutionarily conserved response of cells to oxidative
stress.
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
LaTeX2e(2002/01/18)
Figure 1 Reactions catalyzed by GST. Example of conjugation, reduction, thiolysis, and isomerization reactions catalyzed by GST. The
following substrates are shown: (a) CDNB, (b) sulforaphane, (c) CuOOH, (d) 4-nitrophenyl acetate, (e) trinitroglycerin, ( f ) maleylacetoacetate, and (g) PGH2 (conversion to PGD2 is depicted).
7 Jan 2005
P1: GCE
GLUTATHIONE TRANSFERASES
53
7 Jan 2005
13:55
54
AR
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
LaTeX2e(2002/01/18)
P1: GCE
7 Jan 2005
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
LaTeX2e(2002/01/18)
P1: GCE
GLUTATHIONE TRANSFERASES
55
7 Jan 2005
13:55
56
AR
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
LaTeX2e(2002/01/18)
P1: GCE
catalase, and glutathione peroxidase and nonenzymatically by -tocopherol, ascorbic acid, GSH, and bilirubin. Despite these antioxidant defenses, reactive oxygen
species inflict damage on membrane lipid, DNA, protein, and carbohydrate. Oxidation of these macromolecules gives rise to cytotoxic and mutagenic degradation
products (53). Thus, although O
2 can damage DNA directly, it can also damage
DNA indirectly through the production of these reactive secondary metabolites.
Aldehyde dehydrogenase, alcohol dehydrogenase, aldo-keto reductase, GST, and
Se-dependent glutathione peroxidase (GPx) are some of the enzyme systems that
protect against the by-products of oxidative stress.
Free radical-initiated peroxidation of polyunsaturated fatty acids in membranes
is a particular problem as it results in chain reactions that serve to amplify damage to lipids. The process produces short-lived lipid hydroperoxides that breakdown to yield secondary electrophiles, including epoxyaldehydes, 2-alkenals,
4-hydroxy-2-alkenals, and ketoaldehydes, some of which are genotoxic (53). GST
isoenzymes exhibit modest Se-independent glutathione peroxidase activity toward
1-palmitoyl-2 - (13 - hydroperoxy-cis-9,trans-11-octadecadienoyl)-L-3-phosphatidylcholine and phosphatidylcholine hydroperoxide, indicating they may reduce
lipid hydroperoxides within membranes (5456). The transferases can also reduce
cholesteryl hydroperoxides (57) and fatty acid hydroperoxides, including (S)-9hydroperoxy-10,12-octadecadieonic acid and (S)-13-hydroperoxy-9,11-octadecadieonic acid (56). Presumably, reduction of phospholipid, fatty acid, and cholesteryl
hydroperoxides curtails formation of downstream epoxides and reactive carbonyls
arising from oxidation of membranes. Among the end-products of lipid peroxidation, GSTs conjugate GSH with the 2-alkenals acrolein and crotonaldehyde (2, 4),
as well as 4-hydroxy-2-alkenals of between 6 and 15 carbon atoms in length (58)
(Figure 2); conjugation of GSH with the (S) enantiomer of 4-hydroxynonenal
is favored over the (R) enantiomer (59). Further, GSTs catalyze the conjugation of cholesterol-5,6-oxide, epoxyeicosatrienoic acid, and 9,10-epoxystearic
acid with GSH (2). These findings indicate that GST, along with other antioxidant enzymes, such as Se-dependent GPx1, provide the cell with protection
against a range of harmful electrophiles produced during oxidative damage to
membranes (4).
The 1-cys peroxiredoxin, Prx VI, defends against cellular membrane damage
by reducing phospholipid hydroperoxides to their respective alcohols. Reduction
of these substrates results in oxidation of Cys-47 in Prx VI to sulfenic acid. It
has been proposed that GST reactivates oxidized Prx VI through glutathionylation
followed by reduction of the mixed disulfide (60). Through this process, GST may
indirectly combat oxidative stress by restoring the activity of oxidized Prx VI.
Oxidation of nucleotides yields base propenals, such as adenine propenal, and
hydroperoxides that are detoxified by GST (Figure 2). Oxidation of catecholamines
yields aminochrome, dopachrome, noradrenochrome, and adrenochrome that are
harmful because they can produce O
2 by redox cycling. These quinone-containing
compounds can be conjugated with GSH through the actions of GST, a
reaction that prevents redox-cycling (61). O-quinones formed from dopamine can
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
Figure 2 GST-catalyzed conjugation of ,-unsaturated carbonyls and o-quinones with GSH. Reactions catalyzed by GST on the
following substrates are shown: (a) acrolein, (b) crotonaldehyde, (c) 4-hydroxynonenal, (d) adenine propenal, (e) dopa-o-quinone, and ( f )
aminochrome.
7 Jan 2005
LaTeX2e(2002/01/18)
P1: GCE
GLUTATHIONE TRANSFERASES
57
7 Jan 2005
13:55
58
AR
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
LaTeX2e(2002/01/18)
P1: GCE
also be conjugated with GSH by GST, and this reaction is similarly thought
to combat degenerative processes in the dopaminergic system in human brain
(Figure 2).
7 Jan 2005
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
LaTeX2e(2002/01/18)
P1: GCE
GLUTATHIONE TRANSFERASES
59
facilitates the elimination of these eicosanoids from the cell via MRP1 and MRP3
transporters (68).
Leukotrienes (LTs) are another group of eicosanoids formed from arachidonic
acid. MAPEGs are critically involved in their synthesis because one member
uniquely activates 5-lipoxygenase, whereas several others catalyze the formation
of LTC4.
7 Jan 2005
13:55
60
AR
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
LaTeX2e(2002/01/18)
P1: GCE
transduction pathways, such as JNK, p38, and protein kinase C, as well as increasing p53 protein and promoting apoptosis (77). It is anticipated that conjugation of
4-HNE with GSH will influence many signal transduction pathways and modulate
the activity of transcription factors, including NF-B, c-Jun, and Nrf2.
GST FAMILIES
Cytosolic Enzymes
Mammalian cytosolic GSTs are all dimeric with subunits of 199244 amino acids
in length. Based on amino acid sequence similarities, seven classes of cytosolic GST are recognized in mammalian species, designated Alpha, Mu, Pi, Sigma,
7 Jan 2005
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
LaTeX2e(2002/01/18)
P1: GCE
GLUTATHIONE TRANSFERASES
61
Theta, Omega, and Zeta (25). Other classes of cytosolic GST, namely Beta, Delta,
Epsilon, Lambda, Phi, Tau, and the U class, have been identified in nonmammalian species (5, 23, 81). In rodents and humans, cytosolic GST isoenzymes
within a class typically share >40% identity, and those between classes share
<25% identity.
At least 16 cytosolic GST subunits exist in the human. As those in the Alpha and
Mu classes can form heterodimers (2), a significantly larger number of isoenzymes
can be generated from these subunits. The total of 16 homodimers listed in Table 1
includes the relatively poorly characterized GSTM4-4 (82) and GSTM5-5 (83), as
well as a transferase, GSTA5-5, that has been identified by genomic cloning but
has not been characterized at the protein level (84). An additional human enzyme,
hGST5.8, with high activity toward 4-HNE, has been reported and is presumed to
be a class Alpha transferase (85). This enzyme seems to be distinct from GSTA1-1,
GSTA2-2, GSTA3-3, and GSTA4-4 but it is not included in Table 1 as its primary
structure has not been described. The transferases display overlapping substrate
specificities, a feature that makes it difficult to identify isoenzymes solely on their
catalytic properties. Substrates identified for each of the human cytosolic GST are
listed in Table 1 (some examples are illustrated in Figures 1 and 2).
Besides catalyzing conjugation, reduction, and isomerization reactions, cytosolic GST also bind, covalently and noncovalently, hydrophobic nonsubstrate ligands (2). This type of activity contributes to intracellular transport, sequestration,
and disposition of xenobiotics and hormones. Such compounds include azo-dyes,
bilirubin, heme, PAHs, steroids, and thyroid hormones; it is the nonsubstrate binding activity that led originally to class Alpha GST being called Ligandin (2).
Affinity labeling of rat class Alpha GST has revealed a high-affinity nonsubstrate binding site within the cleft between the two subunits (86), indicating that
there are two distinct xenobiotic-binding sites in certain isoenzymes. The second
nonsubstrate binding site formed in heterodimers will be distinct from those in
homodimers, and it may provide an evolutionary reason why it is beneficial for
members within the Alpha and Mu classes to heterodimerize.
Class Mu and Pi GST have been reported to inhibit Ask1 and JNK during
nonstressed conditions through physical interactions with the kinases (8789). It
has been shown that GSTM1 dissociates from Ask1 by heat shock (88), whereas
GSTP1 dissociates from JNK in response to oxidative stress (89). As described
above, GSTP1 also physically interacts with Prx VI, a process that leads to recovery of peroxiredoxin enzyme activity through glutathionylation of the oxidized
protein (60).
The majority of cytosolic GST isoenzymes are found in the cytoplasm of the
cell. However, mouse and human Alpha-class GSTA4-4 can associate with mitochondria and membranes (9092), as can mouse GSTM1-1 (91). In the case
of GSTA4-4, this entails phosphorylation of the transferase, and targeting is dependent on the Hsp70 chaperone (92). Using monkey COS cells, treatment with
4-HNE increases the amount of GSTA4-4 associated with the mitochondria (92).
A human transferase that is closely related to GSTA1-1 has been purified from
liver microsomes (56), and it appears that certain class Alpha enzymes have a
7 Jan 2005
13:55
62
AR
HAYES
TABLE 1
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
P1: GCE
Family
Class, enzyme
Substrates or reaction
Cytosolic
Alpha, A1-1
Alpha, A2-2
Alpha, A3-3
Alpha, A4-4
Alpha, A5-5
Annu. Rev. Pharmacol. Toxicol. 2005.45:51-88. Downloaded from arjournals.annualreviews.org
by Universitaet Heidelberg on 10/01/05. For personal use only.
LaTeX2e(2002/01/18)
Cytosolic
Mu, M1-1
Mu, M2-2
Mu, M3-3
Mu, M4-4
Mu, M5-5
Cytosolic
Pi, P1-1
Cytosolic
Sigma, S1-1
PGH2 PGD2
Cytosolic
Theta, T1-1
Theta, T2-2
Cytosolic
Zeta, Z1-1
Cytosolic
Omega, O1-1
Omega, O2-2
Mitochondrial
Kappa, K1-1
MAPEG
gp I, MGST2
gp I, FLAP
gp I, LTC4S
MAPEG
gp II, MGST3
MAPEG
gp IV, MGST1
gp IV, PGES1
A systematic study of all these enzymes toward substrates has not been undertaken, and therefore it is not possible to
define relative activities toward the compounds listed. These data are taken from papers cited in the text.
Abbreviations: 5-ADD, 5-androstene-3,17-dione; BCDE, benzo[g]chrysene diol epoxide; BCNU, 1,3-bis(2-chloroethyl)-1-nitrosourea; BPDE, benzo[a]pyrene diol epoxide; BPhDE, benzo[c]phenanthrene diol epoxide; CDE, chrysene1,2-diol 3,4-epoxide; COMC-6, crotonyloxymethyl-2-cyclohexenone; DBADE, dibenz[a,h]anthracene diol epoxide;
DBPDE, dibenzo[a,l]pyrene diol epoxide; EA, ethacrynic acid; EPNP, 1,2-epoxy-3-(p-nitrophenoxy)propane; N-a-PhIP,
N-acetoxy-2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine.
7 Jan 2005
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
LaTeX2e(2002/01/18)
P1: GCE
GLUTATHIONE TRANSFERASES
63
Mitochondrial GST
The mammalian mitochondrial class Kappa GST isoenzymes are dimeric and
comprise subunits of 226 amino acids. Mouse, rat, and human possess only a
single Kappa GST (6, 7, 98, 99). Molecular cloning and crystallography of the
mitochondrial GST have provided definitive evidence that it represents a distinct
type of transferase (6, 7). The three-dimensional fold of Kappa is more similar
to bacterial 2-hydroxychromene-2-carboxylate isomerase, a GSH-dependent oxidoreductase that catalyzes conversion of 2-hydroxy-chromene-2-carboxylate to
trans-O-hydroxy-benzylidenepyruvate, and to prokaryotic disulfide-bond-forming
DsbA and TcpG oxidoreductases, than to any of the cytosolic GST isoenzymes.
As such, it has provided a new insight into the evolution of GST.
GST class Kappa has high activity for aryl halides, such as CDNB, and can
reduce CuOOH and (S)-15-hydroperoxy-5,8,11,13-eicosatetraenoic acid (99). In
view of its homology with 2-hydroxychromene-2-carboxylate isomerase, it will
be interesting to establish whether GST Kappa can metabolize aromatic hydrocarbons, such as naphthalene.
In the mouse, GST Kappa is present in large amounts in liver, kidney, stomach,
and heart, and electron microscopy has confirmed that it is associated with liver
and kidney mitochondria (100). Its tissue distribution in the rat seems similar to
that in the mouse (98). By contrast, GST Kappa appears to be more widely and
uniformly expressed in human tissues (99).
Although this transferase was originally isolated from mitochondria and is not
present in cytoplasm (98), it has also been shown to be located in peroxisomes
(99). The presence of GSTK1-1 in both organelles suggests it may be specifically
involved in -oxidation of fatty acids, either through its catalytic activity, some
transport function, or interaction with a membrane pore. The process of targeting
GST Kappa to mitochondria is unclear. It has been reported to associate with the
Hsp60 chaperone (7), and a possible cleavage site for a mitochondrial presequence
exists at the N-terminus (99). A peroxisomal targeting sequence (tripeptide ARL)
has been identified in the C-terminus of the human GSTK1 subunit (99).
7 Jan 2005
13:55
64
AR
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
LaTeX2e(2002/01/18)
P1: GCE
Figure 4 Schematic diagram showing evolution of the GST fold. The secondary structure arrangements of thioredoxin, the canonical cytosolic GST fold, and that of mitochondrial GSTK1 [the latter is predicted to be closely similar to the secondary structure
of bacterial DsbA (6, 7)] are illustrated. Arrows represent -sheets; rectangles represent
-helices. The regions corresponding to the core thioredoxin structure are shown for
the cytosolic GST fold and GSTK1. The positions of helical domain insertion that result in either fold are also shown, and they clearly illustrate two sites in the thioredoxin
fold that appear to have less evolutionary constraint. The differences in architecture
also provide substantial evidence that soluble GSTs have evolved through two differing
pathways.
7 Jan 2005
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
LaTeX2e(2002/01/18)
P1: GCE
GLUTATHIONE TRANSFERASES
65
indicate that mammalian GSTM1 and GSTP1 can probably exist as monomers
through interactions with other proteins, such as Ask1, JNK, and Prx IV (60, 87
89). It is interesting to note that Pettigrew & Colman (101) have reported that
heterodimers can be formed between class Mu and class Pi polypeptides in vitro
without the need for denaturants, an observation that might reflect some promiscuity in the subunit dimerization in these two classes of GST. Monomeric forms
of cytosolic GST have been demonstrated convincingly in nonmammalian species
(102). The recent identification of a structural relationship between the cytosolic
GSTs and isoforms of the CLICs (94, 95, 102, 103) has revealed the potential for
proteins possessing the canonical GST fold to exist as soluble monomers when
purified in a functionally active state, in this case forming chloride ion channels.
It has also been shown that these monomers can undergo structural rearrangement
under oxidizing conditions to form dimers (103). Whether CLIC adopts this form
in the membrane is at this point unknown, but it has been proposed that a large
conformational rearrangement occurs, facilitating membrane insertion (102).
Identification of the canonical cytosolic GST fold in proteins involved in nondetoxication processes illustrates that this structure is amenable to many different
functions, yet it is not clear whether these proteins represent pathways of convergent evolution or the continued evolution of the cytosolic GST.
MAPEG Enzymes
These members of the GST superfamily constitute a unique branch where most of
the proteins are involved in the production of eicosanoids. Throughout nature, a
total of four MAPEG subgroups (IIV) have been described, with proteins within
a subgroup sharing >20% sequence identity. Six human MAPEGs have been
identified, and these fall within subgroups I, II, and IV (8).
The founding member of the MAPEG family, MGST1, was initially identified
as a microsomal CDNB-metabolizing enzyme that, in contrast to most cytosolic
GST, can be activated by treatment with N-ethylmaleimide (2, 8). Three further
MAPEG members with roles in eicosanoid synthetic pathways were subsequently
identified as leukotriene C4 synthase (LTC4S), a microsomal transferase that conjugates leukotriene A4 with GSH; 5-lipoxygenase-activating protein (FLAP), an
arachidonic acid-binding protein required for 5-lipoxygenase to exhibit full activity; and prostaglandin E2 synthase 1 (PGES1), which catalyses GSH-dependent
isomerization of PGH2 to PGE2 (8). Following the discovery of MGST1, FLAP,
and LTC4S, bioinformatic approaches were used to isolate cDNAs for MGST2
and MGST3, encoding enzymes that reduce (S)-5-hydroperoxy-8,11,14-cis-6trans-eicosatetraenoic acid (104). According to sequence-based subdivision of
the MAPEG family, subgroup I consists of FLAP, LTC4S, and MGST2; the only
member of subgroup II is MGST3; and MGST1 and PGES1 make up subgroup IV.
Subgroup III contains microsomal GST-like proteins from Escherichia coli and
Vibrio cholera.
Evidence suggests MGST1 functions solely as a detoxication enzyme. By
contrast, human MGST2 and MGST3 are capable of both detoxifying foreign
7 Jan 2005
13:55
66
AR
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
LaTeX2e(2002/01/18)
P1: GCE
compounds and synthesizing LTC4 (104); in the rat, MGST3 is apparently unable
to synthesize LTC4 (105). FLAP does not have catalytic activity but binds arachidonic acid and appears to be essential for the synthesis of all leukotrienes formed
downstream of 5-lipoxygenase. LTC4S and PGES1 seem to make no contribution
to detoxification, their catalytic actions being restricted to synthesis of LTC4 and
PGE2, respectively (see Table 1).
crystal structure for MGST1 has illustrated the hoDetermination of a 6 A
motrimeric quaternary structure of the enzyme (10), a quaternary structure also
observed for the other subgroup IV enzyme PGES1 (106). By contrast with the
trimeric structure of these enzymes, subgroup I contains members that either
form monomers or more complex aggregates. For example, FLAP can exist in
monomeric, dimeric and trimeric forms, and LTC4S can similarly form multimeric
complexes (107). FLAP and LTC4S can also form heterodimers and heterotrimers
with each other (107). More research is required to understand the stoichiometry
and membrane topology of these proteins.
7 Jan 2005
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
LaTeX2e(2002/01/18)
P1: GCE
GLUTATHIONE TRANSFERASES
TABLE 2
Class
Allele
Nucleotide(s) in gene at
variable position(s)
Protein affected
Alpha
GSTA1 A
GSTA1 B
GSTA2 A
GSTA2 B
GSTA2 C
GSTA2 D
GSTA2 E
Mu
GSTM1 A
GSTM1 B
GSTM1 0
GSTM1 1x2
GSTM3 A
GSTM3 B
GSTM4 A
GSTM4 B
519G
519C
gene deletion
gene duplication
wild-type
3 bp deletion in intron 6
wild-type
T2517C change in intron
Lys173
Asn173
No protein
Overexpression of M1 protein
Reference protein levels
Protein unchanged
Reference protein levels
Protein unchanged
Pi
GSTP1 A
GSTP1 B
GSTP1 C
GSTP1 D
Sigma
GSTS1 A
GSTS1 B
IVS2 + 11 A
IVS2 + 11 C
Theta
GSTT1 A
GSTT1 0
GSTT2 A
GSTT2 B
wild-type gene
gene deletion
415A
415G
Zeta
GSTZ1 A
GSTZ1 B
GSTZ1 C
GSTZ1 D
Omega
GSTO1 A
GSTO1 B
GSTO1 C
GSTO1 D
GSTO2 A
GSTO2 B
Ala140, Glu155
Ala140, Glu155 deleted
Asp140, Glu155
Asp140, Glu155 deleted
Asn142
Asp142
Numbering of amino acids includes initiator methionine. Adapted from Reference 108.
67
7 Jan 2005
13:55
68
AR
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
LaTeX2e(2002/01/18)
P1: GCE
homozygous +/+ individuals (113). An assay has been developed that can identify heterozygotes at the GSTT1 locus (113a) though useful medical applications
remain to be established.
Besides influencing susceptibility to carcinogenesis, GSTP1 polymorphisms
are modifiers of response to chemotherapy in patients with metastatic colorectal
cancer (114) and those with multiple myeloma (115). It also influences risk of
therapy-related acute myeloid leukemia in patients successfully treated for breast
cancer, non-Hodgkins lymphoma, ovarian cancer, and Hodgkins disease (116).
By contrast with the weak effect that class Mu, Pi, and Theta GST polymorphisms have on tumorigenesis, a number of studies indicate that loss of these
genes increase susceptibility to inflammatory diseases, such as asthma and allergies, atherosclerosis, rheumatoid arthritis, and systemic sclerosis (117119).
In addition to allelic variants in class Mu, Pi, and Theta GST, polymorphisms
have also been identified in all the other classes of cytosolic GST (120122). Class
Alpha represents quantitatively a major group of transferases in the liver and these
enzymes presumably influence substantially detoxification processes. It has been
shown that both GSTA1 and GSTA2 are polymorphic, and the various alleles either
influence the amount of protein synthesized or the activity of the encoded proteins
(84, 123, 124). Further, GSTM4 and GSTT2 exhibit promoter polymorphisms that
are of functional significance (125). It will be interesting to know whether polymorphisms in these genes influence not only susceptibility to degenerative disease
but also efficacy of therapeutic drugs or adverse drug reactions.
7 Jan 2005
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
LaTeX2e(2002/01/18)
P1: GCE
GLUTATHIONE TRANSFERASES
TABLE 3
Previous designations
for subunits
Accession
number
Chromosomal
location
GSTA1
GSTA2
GSTA3
GSTA4
GSTA5
Ya
Ya2
GT10.6, Ya3, Yc
Yk, GST5.7
5
9
9
1
9
GSTM1
GSTM2
GSTM3
GSTM4
GSTM5
GSTM6
GSTM7
GT8.7, Yb1
Yb2
GT9.3, 4
Yb5, 7
Fsc2, mGSTM5
(also called mGSTM5)
3
GSTP1
GSTP2
Yf, piB
Yf, piA
Sigma
Ptgds2
Theta
GSTT1
GSTT2
GSTT3
5
Yrs
NP 032211
n
NM 010361
n
NM 133994
10
10
10
Zeta
GSTZ1
MAAI
12
GSTO1
GSTO2
p28
p
p
19
19
GSTK1
MGST2
FLAP
LTC4S
n
n
3
5
11
MAPEG, subgroup II
MGST3
MAPEG, subgroup IV
MGST1
Ptges1
6
2
Class or family
Alpha
69
Mu
Pi
Omega
Kappa
MAPEG, subgroup I
NP 032207
NP 032208
p
CAA46155
p
NP 034487
NP 034488
AF319526
p
P19639
p
NP 081040
p
NP 034490
n
AJ000413
n
AK002213
n
NP 038569
NP 861461
NP 062328
NP 034493
NP 034492
NP 080895
AAP20655
BC028535
BC026209
n
NM 008521
NM 025569
NM 019946
n
NM 022415
3
3
3
3
3
3
3
19
19
Superscript prefix n = accession number for nucleotide sequence, superscript prefix p = accession number for protein
sequence.
The genes encoding the cytosolic class Sigma GSTS1 and the MAPEG PGES1 are called Ptgds2 and Ptges1, respectively.
This is adapted from the Web site established by Dr. William Pearson on mouse GST (132). The nomenclature for Mu-class
GST differs from that of Andorfer et al. (162): The subunit they called 3 is GSTM7, the subunit they called 4 is GSTM3,
and the subunit they call 7 is GSTM4.
7 Jan 2005
13:55
70
AR
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
LaTeX2e(2002/01/18)
P1: GCE
reduced to between 23% and 64% of wild-type levels in the tissues examined,
but it was particularly marked in brain, heart, kidney, and lung. Substantial increases in 4-HNE and malondialdehyde were found in the livers of KO animals
(133). The livers and brain of GSTA4/ mice contained increases in mRNA for
GSTA1/2, GSTA3, GSTM1, catalase, superoxide dismutases 1 and 2, and GPx1.
Activation of ARE-driven gene expression (78) appears to be one of the mechanisms by which these genes are upregulated in GSTA4 KO mice. Certainly, 4-HNE
is a Michael reaction acceptor (75) and many cancer chemopreventive blocking
agents that induce GST can be included in this category of compound (134). It is
therefore presumed that induction of transferases and antioxidant proteins in the
mutant mice represents a compensatory response to increases in the intracellular
levels of reactive aldehydes resulting from loss of GSTA4-4.
The GSTA4 subunit is induced in mice fed on diets containing the cancer
chemopreventive agents -angelicalactone, butylated hydroxyanisole, ethoxyquin,
indole-3-carbinol, limettin, oltipraz, or sulforaphane (135). These data suggest the
mouse GSTA4 gene contains an ARE. Consistent with this hypothesis, we have
found, using a bioinformatic search, that the 5 -upstream region of mouse GSTA4
contains the sequence 5 -TGAGTCAGC-3 . This sequence closely resembles the
5 -TGAGTCGGC-3 ARE in mouse NAD(P)H:quinone oxidoreductase 1 (136);
both differ from the prototypic core ARE, 5 -TGACnnnGC-3 (137), in having
a G rather than a C at nucleotide position 4 (shown underlined). Assuming this
putative ARE in GSTA4 is functional, induction of the gene by 4-HNE is likely
to be mediated by Nrf2. It is envisaged that increased concentrations of 4-HNE
lead to modification of cysteine residues in Keap1, stabilization and nuclear accumulation of Nrf2, and increased GSTA4-4 and glutathione levels, resulting in
increased capacity to metabolize 4-HNE (see Figure 5, color insert). According
to these predictions, mouse GSTA4-4 appears to comprise part of an autoregulatory homeostatic defense mechanism against lipid peroxidation products. Another
characteristic of the putative ARE in GSTA4 is that it contains an embedded 12-Otetradecanoylphorbol-13-acetate (TPA) response element and may therefore also
be regulated by the c-Jun transcription factor; for a review of transcriptional regulation of genes through the ARE and related enhancers, see References 138 and 139.
A mouse line lacking GSTM5, which encodes the brain/testisspecific transferase, has been generated, but no clear phenotype has been reported
to date (140).
CLASS Mu GST
Mice lacking both GSTP1 and GSTP2 have been generated (141).
Under normal conditions, the double gene knockout on 129MF1 or C57/BL6
backgrounds had no obvious phenotype. At a biochemical level, the mutant mice
demonstrated a complete lack of transferase activity toward ethacrynic acid in the
liver (141). Although GSTP1-1 is quantitatively the principal transferase in male
mouse liver, Western blotting failed to demonstrate compensatory increases in
expression of hepatic GSTA1/2, GSTA3, and GSTM1 subunits in the double gene
CLASS Pi GST
7 Jan 2005
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
LaTeX2e(2002/01/18)
P1: GCE
GLUTATHIONE TRANSFERASES
71
KO animals (141). However, livers from GSTP1/P2/ mice have been reported
to contain a higher activator protein-1 activity than livers from GSTP1/P2+/+ mice
(142), a finding that is consistent with the hypothesis that class Pi GST inhibits
JNK (21, 89).
In a skin tumorigenesis regimen, GSTP1/P2/ mice yielded approximately
threefold more papillomas using 7,12-dimethylbenzanthracene as initiator and
TPA as promoter (141), demonstrating a role for GSTP1-1 in xenobiotic defense.
Surprisingly, GSTP1/P2/ mice are more resistant than wild-type mice to liver
toxicity caused by the analgesic acetaminophen, and this is attributed to faster
regeneration of hepatic GSH in the double gene KO animals (143). It was proposed that while Pi-class GST does not catalyze the conjugation of acetaminophen
with GSH, it contributes to oxidative stress by facilitating redox-cycling of the
acetaminophen metabolite NAPQI, possibly through formation of labile ipso
adducts with intracellular thiol groups (143). It is postulated that the absence of Pi
class GST lessens the ability of NAPQI to redox-cycle and thus deplete GSH.
This class of GST encodes the hematopoietic, or GSHdependent, prostaglandin D2 synthase. Knockout of the gene for this enzyme results in generation of mice with an allergic reaction that is weaker than wild-type
mice (144).
7 Jan 2005
13:55
72
AR
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
LaTeX2e(2002/01/18)
P1: GCE
7 Jan 2005
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
LaTeX2e(2002/01/18)
P1: GCE
GLUTATHIONE TRANSFERASES
73
MAPEG SUBGROUP IV
7 Jan 2005
13:55
AR
74
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
LaTeX2e(2002/01/18)
P1: GCE
induction of GST genes in the KO animals described above. According to this proposal, increased levels of 4-HNE, either MAA or its metabolites, and peroxides
in the GSTA4/, GSTZ1/, and Trsp/ mice modify Keap1, causing accumulation of Nrf2 and its translocation to the nucleus. Thereafter, Nrf2 is recruited
to ARE enhancers in the promoters of inducible genes. A substantial number of
GST genes have been found to contain an ARE or related sequences. Table 4
provides a compilation of those GST, NQO1, and SOD1 genes that contain such
elements (136, 137, 139, 155162) and could therefore be regulated by this mechanism; it also contains inducible GST genes that have ARE-like sequences that have
yet to be shown to be functional enhancers (these uncharacterized enhancers are
Gene
Enhancer
5 -USR
Rat
GSTA2
ARE
Enhancer
687
Rat
GSTA5
ARE
470
Rat
GSTP1
GPEI
2430
Mouse
GSTA1
EpRE
719
Mouse
GSTA3
ARE
138
Mouse
GSTA4
n.c.
147
Mouse
GSTM1
n.c.
1643
Mouse
GSTM2
n.c.
202
Mouse
GSTM3
n.c.
2315
Mouse
GSTP1
ARE
50
Mouse
GSTP2
n.c.
61
Human
MGST1
EpRE
490
Rat
NQO1
ARE
412
Mouse
Nqo1
ARE
426
Human
NQO1
ARE
460
Human
SOD1
ARE
323
ARE core
T-MARE
TGACnnnGC
TGC TGACTCAGCA
The mouse GSTM3 gene was called GSTM4 and 4 in Reference 162.
The core ARE required for gene induction is usually regarded as 5 -TGACnnnGC-3 , based on mutational analysis
of the promoter of rat GSTA2 (137). The nucleotides located in the 5 -upstream region (5 -USR) of the GSTA2-ARE
have been found to influence basal expression without altering the relative magnitude of induction, and therefore this
region is included in the line-up. Nucleotides in capital bold print are those that share identity with the Maf recognition
element (MARE); this contains an embedded TPA-response element, denoted by the abbreviation T-MARE (138).
The numbering in the right-hand column is the position of the 3 A nucleotide with respect to the transcriptional start
site; in the cases of rat GSTA5 and mouse GSTA4 this nucleotide is a G, and in the cases of GSTM2 and SOD1 this
nucleotide is a T. Data are taken from References 136, 137, 155162. The abbreviation n.c. stands for not characterized.
7 Jan 2005
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
LaTeX2e(2002/01/18)
P1: GCE
GLUTATHIONE TRANSFERASES
75
indicated by the abbreviation n.c.). The observation that disruption of GSTA4 and
GSTZ1 genes upregulates the ARE-gene battery supports the hypothesis that the
transferases encoded by these genes not only make a major functional contribution to an antioxidant and electrophile defense network but that their substrates are
endogenous activators of Nrf2.
The notion that Nrf2 mediates basal expression of GST by endogenous thiolactive endobiotics is supported by the fact that in mice nulled for this transcription
factor the normal homeostatic levels of many class Alpha, Mu, and Pi transferases
are reduced (163). For example, the levels of mRNA encoding GSTA1, GSTA2,
GSTM1, and GSTM3 in the livers of Nrf2/ mice fed on a normal diet have
been reported to be less than 20% of the levels observed in Nrf2+/+ mice (131).
In addition to changes in expression of cytosolic GST, microarray analyses have
shown that expression of MAPEG genes is also affected in Nrf2 KO mice (164,
165). Further work is required to establish how important Nrf2 is in regulating
GST in species other than the mouse.
It should be appreciated that Nrf2 is not the only transcription factor involved
in regulating GST through the ARE. The 5 -upstream region immediately adjacent
to the core ARE in genes such as rat GSTA2, mouse GSTA1, mouse GSTM2,
mouse GSTP1, and mouse GSTP2 conforms more closely to a TRE-containing
Maf recognition element (i.e., T-MARE) than does the same region in rat GSTP1,
mouse GSTA3, or any of the NQO1 genes; for a review of transcriptional regulation
of AREs and MAREs, see Reference 138. It appears that some of these GST genes
may be regulated entirely by Nrf2-small Maf heterodimers, whereas others may
be regulated not only by Nrf2-small Maf heterodimers but also by small and large
Maf homodimers. The positive and negative regulation of ARE-driven genes is an
area that needs further study.
7 Jan 2005
13:55
76
AR
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
LaTeX2e(2002/01/18)
P1: GCE
CONCLUDING REMARKS
This review describes recent advances in knowledge about the transferases. The
availability of gene KO models has given unprecedented insights into the in vivo
functions of GST and MAPEG proteins. These studies have demonstrated that
cytosolic GST are an integral part of a dynamic and interactive defense mechanism
that protects against cytotoxic electrophilic chemicals and allows adaptation to
exposure to oxidative stress. They have antioxidant and antiinflammatory activities.
Similar investigations have shown MAPEG members contribute to inflammatory
responses, although it is likely that some are also involved in antioxidant defenses.
Further work is required to elucidate the biological functions of the mitochondrial
class Kappa GST.
Evidence suggests cytosolic GST metabolize many endogenous and foreign
compounds that stimulate expression of the ARE-gene battery. Their catalytic actions therefore negatively regulate Nrf2 by protecting Keap1 from modification of
those cysteines (Cys-273 and Cys-288) that are required to capture and destabilize
the transcription factor. A most important consequence of this conclusion is that
GST indirectly control the levels of other antioxidant and drug-metabolizing enzymes that are regulated through the Keap1/Nrf2 pathway. In addition to phase I,
phase II, and phase III detoxication proteins, GST will negatively regulate chaperones, ubiquitin-proteasome components, inflammation-associated proteins, and
apoptosis-associated proteins (165, 166).
The gene KO mouse models have revealed the importance of GST in detoxifying 4-HNE and tyrosine catabolites. It is predicted that glutathione transferases
similarly contribute to the elimination of 15d-PGJ2 in vivo. Thus, knockout of
certain GST genes will cause relative accumulation of 15d-PGJ2 and constitutive
upregulation of PPAR -driven gene expression and a decrease in expression of
NF-B-driven genes. A possible candidate for this function is GSTA3-3 because its
levels increase markedly in mouse 3T3-L1 cells during adipogenesis (70). It can be
hypothesized that induction of GSTA3 reflects a cellular response to accumulation
of 15d-PGJ2 designed to metabolize and eliminate the prostanoid.
A possibility that remains to be explored is whether polymorphisms in human
GST genes influence the activity of Nrf2, PPAR or NF-B.
ACKNOWLEDGMENTS
We are enormously grateful to the many colleagues in the GST field who have
generously given advice and details of their ongoing work. We can only apologize
to this community that space constraints have prevented us from citing many
excellent papers from our fellow researchers. We particularly thank Drs. Philip
Board, Irving Listowsky, and Bill Pearson for critical comments about the mouse
GST nomenclature. The work from the Hayes laboratory is funded by the Medical
Research Council (G0000281), the Association for International Cancer Research
(02049, 03074), and the World Cancer Research Fund (2000/11).
7 Jan 2005
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
LaTeX2e(2002/01/18)
P1: GCE
GLUTATHIONE TRANSFERASES
77
LITERATURE CITED
1. Keen JH, Jakoby WB. 1978. Glutathione
transferases. Catalysis of nucleophilic
reactions of glutathione. J. Biol. Chem.
253:565457
2. Hayes JD, Pulford DJ. 1995. The glutathione S-transferase supergene family:
regulation of GST and the contribution
of the isoenzymes to cancer chemoprotection and drug resistance. Crit. Rev.
Biochem. Mol. Biol. 30:445600
3. Armstrong RN. 1997. Structure, catalytic mechanism, and evolution of
the glutathione transferases. Chem. Res.
Toxicol. 10:218
4. Hayes JD, McLellan LI. 1999. Glutathione and glutathione-dependent enzymes represent a co-ordinately regulated defence against oxidative stress.
Free Radic. Res. 31:273300
5. Sheehan D, Meade G, Foley VM, Dowd
CA. 2001. Structure, function and evolution of glutathione transferases: implications for classification of nonmammalian members of an ancient
enzyme superfamily. Biochem. J. 360:1
16
6. Ladner JE, Parsons JF, Rife CL, Gilliland
GL, Armstrong RN. 2004. Parallel evolutionary pathways for glutathione transferases: structure and mechanism of
the mitochondrial class Kappa enzyme
rGSTK1-1. Biochemistry 43:35261
7. Robinson A, Huttley GA, Booth
HS, Board PG. 2004. Modelling and
bioinformatics studies of the human
Kappa class glutathione transferase
predict a novel third transferase family with homology to prokaryotic
2-hydroxychromene-2-carboxylate isomerases. Biochem. J. 379:54152
8. Jakobsson P-J, Morgenstern R, Mancini
J, Ford-Hutchinson A, Persson B.
9.
10.
11.
12.
13.
14.
15.
7 Jan 2005
13:55
78
16.
17.
18.
19.
20.
21.
22.
23.
AR
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
24.
25.
26.
27.
28.
29.
30.
31.
32.
LaTeX2e(2002/01/18)
P1: GCE
7 Jan 2005
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
LaTeX2e(2002/01/18)
P1: GCE
GLUTATHIONE TRANSFERASES
33.
34.
35.
36.
37.
38.
39.
Amphibian transition to the oxidant terrestrial environment affects the expression of glutathione S-transferases isoenzymatic pattern. Biochim. Biophys. Acta
1691(23):18192
Haimeur A, Conseil G, Deeley RG,
Cole SP. 2004. The MRP-related and
BCRP/ABCG2 multidrug resistance
proteins: biology, substrate specificity
and regulation. Curr. Drug Metab. 5:
2153
Paumi CM, Ledford BG, Smitherman
PK, Townsend AJ, Morrow CS. 2001.
Role of multidrug resistance protein 1
(MRP1) and glutathione S-transferase
A1-1 in alkylating agent resistance.
Kinetics of glutathione conjugate formation and efflux govern differential cellular sensitivity to chlorambucil
versus melphalan toxicity. J. Biol. Chem.
276:795256
Morrow CS, Smitherman PK, Townsend
AJ. 2000. Role of multidrug-resistance
protein 2 in glutathione S-transferase
P1-1-mediated resistance to 4-nitroquinoline 1-oxide toxicities in HepG2
cells. Mol. Carcinog. 29:17078
Awasthi S, Sharma R, Singhal SS,
Zimniak P, Awasthi YC. 2002. RLIP76,
a novel transporter catalysing ATPdependent efflux of xenobiotics. Drug
Metab. Dispos. 30:130010
Hamilton DS, Zhang X, Ding Z, Hubatsch I, Mannervik B, et al. 2003. Mechanism of the glutathione transferasecatalyzed conversion of antitumor
2-crotonyloxymethyl-2-cycloalkenones
to GSH adducts. J. Am. Chem. Soc. 125:
1504958
Lien S, Larsson A-K, Mannervik B.
2002. The polymorphic human glutathione transferase T1-1, the most efficient glutathione transferase in the
denitrosation and inactivation of the
anticancer drug 1,3-bis(2-chloroethyl)1-nitrosourea. Biochem. Pharmacol. 63:
19197
Abel EL, Bammler TK, Eaton DL. 2004.
40.
41.
42.
43.
44.
45.
46.
79
7 Jan 2005
13:55
80
47.
48.
49.
50.
51.
51a.
52.
53.
AR
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
LaTeX2e(2002/01/18)
P1: GCE
7 Jan 2005
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
LaTeX2e(2002/01/18)
P1: GCE
GLUTATHIONE TRANSFERASES
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
81
7 Jan 2005
13:55
82
AR
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
76. Echtay KS, Esteves TC, Pakay JL, Jekabsons MB, Lambert AJ, et al. 2003. A signalling role for 4-hydroxy-2-nonenal in
regulation of mitochondrial uncoupling.
EMBO J. 22:410310
77. Awasthi YC, Sharma R, Cheng JZ,
Yang Y, Sharma A, et al. 2003. Role
of 4-hydroxynonenal in stress-mediated
apoptosis signalling. Mol. Aspects Med.
24:21930
78. Tjalkens RB, Luckey SW, Kroll DJ, Petersen DR. 1999. ,-Unsaturated aldehydes mediate inducible expression of
glutathione S-transferase in hepatoma
cells through activation of the antioxidant response element (ARE). Adv. Exp.
Med. Biol. 463:12331
79. Ishii T, Itoh K, Ruiz E, Leake DS, Unoki
H, et al. 2004. Role of Nrf2 in the regulation of CD36 and stress protein expression in murine macrophages. Activation by oxidatively modified LDL and
4-hydroxynonenal. Circ. Res. 94:60916
80. Levonen A-L, Landar A, Ramachandran
A, Ceaser EK, Dickinson DA, et al. 2004.
Cellular mechanisms of redox cell signalling: the role of cysteine modification in controlling antioxidant defences
in response to electrophilic lipid oxidation products. Biochem. J. 378:373
82
81. Ding Y, Ortelli F, Rossiter LC, Hemingway J, Ranson H. 2003. The Anopheles
gambiae glutathione transferase supergene family: annotation, phylogeny and
expression profiles. BMC Genomics 4:35
82. Patskovsky YV, Patskovska LN, Listowsky I. 2000. The enhanced affinity for
thiolate anion and activation of enzymebound glutathione is governed by an
arginine residue of human Mu class glutathione S-transferase. J. Biol. Chem.
275:3296304
83. Patskovsky YV, Patskovska LN, Listowsky I. 1999. An asparagine-phenylalanine substitution accounts for catalytic differences between hGSTM3-3
and other human class Mu glutathione
84.
85.
86.
87.
88.
89.
90.
91.
LaTeX2e(2002/01/18)
P1: GCE
7 Jan 2005
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
LaTeX2e(2002/01/18)
P1: GCE
GLUTATHIONE TRANSFERASES
92. Robin M-A, Prabu SK, Raza H, Anandatheerthavarada HK, Avadhani NG.
2003. Phosphorylation enhances mitochondrial targeting of GSTA4-4 through
increased affinity for binding to cytoplasmic Hsp70. J. Biol. Chem. 278:1896070
93. Kodym R, Calkins P, Story M. 1999.
The cloning and characterization of a
new stress response protein. A mammalian member of a family of class
glutathione S-transferase-like proteins.
J. Biol. Chem. 274:513137
94. Harrop SJ, DeMaere MZ, Fairlie WD,
Reztsova T, Valenzuela SM, et al. 2001.
Crystal structure of a soluble form of the
intracellular chloride ion channel CLIC1
(NCC27) at 1.4-A resolution. J. Biol.
Chem. 276:449935000
95. Dulhunty A, Gage P, Curtis S, Chelvanayagam G, Board P. 2001 The glutathione transferase structural family includes a nuclear chloride channel and a
ryanodine receptor calcium release channel modulator. J. Biol. Chem. 276:3319
23
96. Jeppesen MG, Ortiz P, Shepard W, Kinzy
TG, Nyborg J, Andersen GR. 2003.
The crystal structure of the glutathione
S-transferase-like domain of elongation factor 1B from Saccharomyces
cerevisiae. J. Biol. Chem. 278:47190
98
97. Marco A, Cuesta A, Pedrola L, Palau
F, Marn I. 2004. Evolutionary and
structural analyses of GDAP1, involved
in Charcot-Marie-Tooth disease, characterize a novel class of glutathione
transferase-related genes. Mol. Biol.
Evol. 21:17687
98. Jowsey IR, Thomson RE, Orton TC,
Elcombe CR, Hayes JD. 2003. Biochemical and genetic characterization of
a murine class Kappa glutathione Stransferase. Biochem. J. 373:55969
99. Morel F, Rauch C, Petit E, Piton A,
Theret N, et al. 2004. The human glutathione transferase Kappa: gene and
protein characterization and evidence
100.
101.
102.
103.
104.
105.
106.
107.
83
7 Jan 2005
13:55
84
AR
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
115.
116.
117.
118.
119.
120.
121.
LaTeX2e(2002/01/18)
P1: GCE
Association between glutathione Stransferase P1, T1, and M1 genetic polymorphism and survival of patients with
metastatic colorectal cancer. J. Natl.
Cancer Inst. 94:93642
Dasgupta RK, Adamson PJ, Davies FE,
Rollinson S, Roddam PL, et al. 2003.
Polymorphic variation in GSTP1 modulates outcome following therapy for multiple myeloma. Blood 102:234550
Allan JM, Wild CP, Rollinson S, Willett EV, Moorman AV, et al. 2001. Polymorphism in glutathione S-transferase
P1 is associated with susceptibility to
chemotherapy-induced leukaemia. Proc.
Natl. Acad. Sci. USA 98:1159297
Palmer CAN, Young V, Ho M, Doney A,
Belch JJF. 2003. Association of common
variation in glutathione S-transferase
genes with premature development of
cardiovascular disease in patients with
systemic sclerosis. Arthritis Rheumatism
48:85455
Gilliland FD, Li Y-F, Saxon A, DiazSanchez D. 2004. Effect of glutathione
S-transferase M1 and P1 genotypes
on xenobiotic enhancement of allergic responses: randomised, placebocontrolled crossover study. Lancet 363:
11925
Romieu I, Sienra-Monge JJ, RamrezAguilar M, Moreno-Macas H, ReyesRuiz NI, et al. 2004. Genetic polymorphism of GSTM1 and antioxidant supplementation influence lung function in
relation to ozone exposure in asthmatic
children in Mexico City. Thorax 59:810
Noguchi E, Shibasaki M, Kamioka M,
Yokouchi Y, Yamakawa-Kobayashi K,
et al. 2002. New polymorphisms of
haematopoietic prostaglandin D synthase and human prostanoid DP receptor
genes. Clin. Exp. Allergy 32:9396
Board PG, Anders MW, Blackburn AC.
2005. Catalytic function and expression
of glutathione zeta. In Drug Metabolism
and Transport: Molecular Methods and
Mechanisms, ed. LH Lash, Methods
7 Jan 2005
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
LaTeX2e(2002/01/18)
P1: GCE
GLUTATHIONE TRANSFERASES
122.
123.
124.
125.
126.
127.
128.
129.
130.
131.
132.
133.
134.
135.
136.
85
7 Jan 2005
13:55
86
137.
138.
139.
140.
141.
142.
143.
144.
145.
AR
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
146.
147.
148.
149.
150.
151.
152.
153.
LaTeX2e(2002/01/18)
P1: GCE
7 Jan 2005
13:55
AR
AR232-PA45-03.tex
AR232-PA45-03.sgm
LaTeX2e(2002/01/18)
P1: GCE
GLUTATHIONE TRANSFERASES
154.
155.
156.
157.
158.
159.
160.
161.
162.
163.
164.
165.
165a.
165b.
87
by 2,3,7,8-tetrachlorodibenzo-p-dioxin
through two different regulator sites,
the antioxidant responsive element and
xenobiotic response element. Mol. Cell.
Biochem. 24:4755
Jowsey IR, Jiang Q, Itoh K, Yamamoto M, Hayes JD. 2003. Expression
of the aflatoxin B1-8,9-epoxide-metabolizing murine glutathione S-transferase A3 subunit is regulated by the Nrf2
transcription factor through an antioxidant response element. Mol. Pharmacol.
64:101828
Andorfer JH, Tchaikovskaya T, Listowsky I. 2004. Selective expression of
glutathione S-transferase genes in the
murine gastrointestinal tract in response
to dietary organosulfur compounds. Carcinogenesis 25:35967
Hayes JD, Chanas SA, Henderson CJ,
McMahon M, Sun C, et al. 2000. The
Nrf2 transcription factor contributes both
to the basal expression of glutathione Stransferases in mouse liver and to their
induction by the chemopreventive synthetic antioxidants, butylated hydroxyanisole and ethoxyquin. Biochem. Soc.
Trans. 28:3341
Thimmulappa RK, Mai KH, Srisuma S,
Kensler TW, Yamamoto M, Biswal S.
2002. Identification of Nrf2-regulated
genes induced by the chemopreventive
agent sulforaphane by oligonucleotide
microarray. Cancer Res. 62:5196
203
Kwak M-K, Wakabayashi N, Itoh K, Motohashi H, Yamamoto M, Kensler TW.
2003. Modulation of gene expression by
cancer chemopreventive dithiolethines
through the Keap1-Nrf2 pathway. Identification of novel gene clusters for
cell survival. J. Biol. Chem. 278:8135
45
Solt DB, Farber E. 1976. New principle
for the analysis of chemical carcinogenesis. Nature 263:7013
Ikeda H, Nishi S, Sakai M. 2004. Transcription factor Nrf2/MafK regulates rat
7 Jan 2005
13:55
88
AR
HAYES
AR232-PA45-03.tex
AR232-PA45-03.sgm
JOWSEY
FLANAGAN
LaTeX2e(2002/01/18)
P1: GCE
HI-RES-PA45-03-Hayes.qxd
1/7/05
2:21 PM
Page 1
GLUTATHIONE TRANSFERASES
C-1
HI-RES-PA45-03-Hayes.qxd
HAYES
FLANAGAN
2:21 PM
Page 2
JOWSEY
C-2
1/7/05
7 Jan 2005
13:56
AR
AR232-PA45-04.tex
AR232-PA45-04.sgm
P1: GCE
LaTeX2e(2002/01/18)
10.1146/annurev.pharmtox.45.120403.095748
James K. Liao
Vascular Medicine Research, Brigham & Womens Hospital, Cambridge,
Massachusetts 02139; email: jliao@rics.bwh.harvard.edu
Ulrich Laufs
Klinik Innere Medizin III, Universitat des Saarlandes, 66421 Homburg, Germany;
email: ulrich@laufs.com
INTRODUCTION
Cardiovascular disease, in particular coronary heart disease (CHD), is the principal cause of mortality in developed countries. Among the causes of cardiovascular
disease, atherosclerosis is the underlying disorder in the majority of patients. Although the development of atherosclerosis is dependent on a complex interplay
between many factors and processes (1), a clear association has been established
between elevated levels of plasma cholesterol and increased atherosclerotic disease (2, 3). Indeed, several landmark clinical trials, such as the Scandinavian
Simvastatin Survival Study (4S) (4), Cholesterol and Recurrent Events (CARE)
(5), Long-term Intervention with Pravastatin in Ischemic Disease (LIPID) (6),
West of Scotland Coronary Prevention Study (WOSCOPS) (7), Air Force/Texas
0362-1642/05/0210-0089$14.00
89
7 Jan 2005
13:56
90
AR
LIAO
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAUFS
Coronary Atherosclerosis Prevention Study (AFCAPS/TexCAPS) (8), Heart Protection Study (HPS) (9), and the Anglo-Scandinavian Cardiac Outcome Trial
Lipid-lowering Arm (ASCOT-LLA) (10), have demonstrated the benefit of lipid
lowering with 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase
inhibitors or statins for the primary and secondary prevention of CHD.
7 Jan 2005
13:56
AR
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
91
the reductase by similar mechanism (Figure 1). Recently, the structure of the catalytic portion of human HMG-CoA reductase complexed with different statins was
determined (22). The bulky, hydrophobic compounds of statins occupy the HMGbinding pocket and block access of the substrate HMG. The tight binding of statins
is due to the large number of van der Waals interactions between statins and HMGCoA reductase. The structurally diverse, rigid, hydrophobic groups of the different
statins are accommodated in a shallow nonpolar groove that is present only when
COOH-terminal residues of HMG-CoA reductase are disordered. There are subtle
differences in the modes of binding between the various statins, with the synthetic
compounds atorvastatin and rosuvastatin having the greatest number of bonding
interactions with HMG-CoA reductase (22). Statins bind to mammalian HMGCoA reductase at nanomolar concentrations, leading to effective displacement
of the natural substrate, HMG-CoA, which binds at micromolar concentrations
(23).
Oral administration of statins to rodents and dogs showed that these drugs
are predominantly extracted by the liver and resulted in >30%50% reduction in
plasma total cholesterol levels and substantial decrease in urinary and plasma levels
of mevalonic acid, the end product of the HMG-CoA reductase reaction. Similar
reduction in cholesterol synthesis and decrease in circulating total and low-density
lipoprotein (LDL)-containing cholesterol (LDL-C) by these agents have been subsequently confirmed in humans. Because hepatic LDL-C receptors are the major
mechanism of LDL-C clearance from the circulation, the substantial declines in
serum cholesterol levels are accompanied by an increase in hepatic LDL-C receptor activity. Statins, therefore, effectively reduce serum cholesterol levels by two
separate mechanisms. They not only inhibit endogenous cholesterol biosynthesis
via HMG-CoA reductase inhibition but also increase cholesterol clearance from
the bloodstream via increases in LDL-C receptor.
The rank order of potency for HMG-CoA reductase inhibition among the
second-generation statins is simvastatin > pravastatin > lovastatin
= mevastatin,
with tissue IC50 values of simvastatin and mevastatin being approximately 4 nM
and 20 nM, respectively (24). The IC50 values for these statins correspond to their
relative potency for lowering serum cholesterol levels in vivo (i.e., simvastatin >
lovastatin) (25). The newer third-generation synthetic statins, which include fluvastatin, cerivastatin, the penta-substituted pyrrole atorvastatin, pitavastatin (NK104), and rosuvastatin, are much more potent than the mevastatin derivatives.
These newer statins are active compounds, which share some physico-chemical
properties with pravastatin, but have greater lipophilicity and half-life (26). Consequently, these statins, especially atorvastatin, pitavastatin, and rosuvastatin, appear
to be quite effective in lowering serum cholesterol levels, perhaps, in part, owing
to their ability to bind hepatic HMG-CoA reductase at higher affinity and inhibit
the enzyme for a longer duration.
Because statins differ in their tissue permeability and metabolism, they possess
different potencies for extrahepatic HMG-CoA reductase inhibition. These differences in tissue permeability and metabolism may account for some of the observed
7 Jan 2005
13:56
92
AR
LIAO
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAUFS
differences in their peripheral side effects (27). Lipophilic statins, such as simvastatin, are considered more likely to enter endothelial cells by passive diffusion
than hydrophilic statins, such as pravastatin and rosuvastatin, which are primarily
targeted to the liver. However, lipophilicity does not entirely predict the ability
of statins to exert extrahepatic effects in animal and human studies, and so other
unidentified factors may play a role. It may be that there are specific mechanisms
for hydrophilic statins to enter extrahapetic cells, such as endothelial cells. Such a
mechanism is present in the liver, where the organic anion transporter (OATP-C)
enables hydrophilic statins to enter hepatocytes (28).
Until recently, all cholesterol-independent or pleiotropic effects of statins
were believed to be mediated by inhibition of mevalonate synthesis. However,
statins can reportedly bind to a novel allosteric site within the 2 integrin functionassociated antigen-1 (LFA-1), independent of mevalonate production (29). LFA-1
belongs to the integrin family and plays an important role in leukocyte trafficking
and in T cell activation. Random screening of chemical libraries identified the
HMG-CoA reductase inhibitor, lovastatin, as an inhibitor of the LFA-1/intercellular
adhesion molecule (ICAM)-1 interaction.
7 Jan 2005
13:56
AR
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
93
systems. The activation of RhoA in Swiss 3T3 fibroblasts by extracellular ligands, such as platelet-derived lysophosphatidic acid, leads to myosin light chain
phosphorylation and formation of focal adhesion complexes (30, 31, 36). Indeed,
Rho-associated protein kinase increases the sensitivity of vascular smooth muscle
to calcium in hypertension (37) and coronary spasm (38). In contrast, activation
of Rac1 leads to the formation of lamellipodia and membrane ruffles, whereas
activation of Cdc42 induces actin-rich surface protrusions called filopodia.
These distinct but complementary functions of Rho family members also extend to their effects on cell signaling. When cells undergo reorganization of their
actin cytoskeleton in response to extracellular signals, such as growth factors, or
7 Jan 2005
13:56
94
AR
LIAO
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAUFS
during cell movement and mitosis, they alter the three-dimensional colocalization
of intracellular proteins (30, 31). Thus, changes in Rho-induced actin cytoskeleton can affect intracellular transport, membrane trafficking, mRNA stability, and
gene transcription. It is therefore not surprising to find that Rho-induced changes
in the actin cytoskeleton and gene expression are related. Indeed, experimental
evidence suggests that inhibition of Rho isoprenylation mediates many of the
7 Jan 2005
13:56
AR
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
95
cholesterol-independent effects of statins not only in vascular wall cells (32, 39),
but also in leukocytes (40) and bone (41).
7 Jan 2005
13:56
96
AR
LIAO
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAUFS
Additional important effects of statin treatment on eNOS function include inhibition of caveolin (60, 61). Statins also increase eNOS activity via posttranslational
activation of the phosphatidylinositol 3-kinase/protein kinase Akt (PI3K/Akt) pathway (55). Phosphorylation of Akt is an important event in several cellular activities.
Indeed, production of NO by the endothelium can be regulated by phosphorylation
and activation of eNOS by Akt, which is promoted in the presence of statins (62,
63). Caveolin-1 binds to eNOS in caveolae, thereby negatively regulating the enzyme (64). Exposure of cultured endothelial cells to hypercholesterolemic serum
upregulates caveolin-1 abundance and promotes association of caveolin-1 and
eNOS into inhibitory complexes, thereby decreasing NO production (65). Statins
have been shown to reduce caveolin-1 abundance and decrease its inhibitory action
on both basal and agonist-stimulated eNOS activity.
Another potential mechanism by which statins may improve endothelial function is through their antioxidant effects. For example, statins enhance endotheliumdependent relaxation by inhibiting the production of reactive oxygen species
(ROS), such as superoxide and hydroxy radicals, from aortas of cholesterol-fed rabbits (66). Although lipid lowering by itself can lower vascular oxidative stress (67),
some of these antioxidant effects of statins appear to be cholesterol-independent.
For example, statins attenuate angiotensin IIinduced free radical production in
vascular smooth muscle cells (SMCs) by inhibiting Rac1-mediated NADH oxidase
activity and downregulating angiotensin AT1 receptor expression (68) (Figure 4).
Because NO is scavenged by ROS, these findings indicate that the antioxidant
properties of statins may also contribute to their ability to improve endothelial
function (49, 50).
7 Jan 2005
13:56
AR
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
97
Figure 4 Antioxidative mechanisms of statins. The core NAD(P)H oxidase comprises five
components: p40phox (PHOX for phagocyte oxidase), p47phox, p67phox, p22phox, and gp91phox.
In the resting cell (left), three of these five components, p40phox, p47phox, and p67phox, exist in
the cytosol as a complex. The other two components, p22phox and gp91phox, are located in the
membranes. When it is stimulated by angiotensin, the cytosolic component becomes heavily
phosphorylated and the entire cytosolic complex migrates to the membrane. Activation requires the participation not only of the core subunits but also of two low-molecular-weight
guanine nucleotide-binding proteins, Rac and Rap. During activation, Rac binds GTP and
migrates to the membrane along with the core cytosolic complex. Treatment with statin downregulates AT1-receptor expression and inhibits Rac1 GTPase, a necessary component of the
NAD(P)H oxidase complex.
7 Jan 2005
13:56
98
AR
LIAO
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAUFS
indicate that statins inhibit vascular SMC proliferation by arresting cell cycle
between the G1/S phase transition. It remains to be determined whether the upregulation of p27Kip1 is responsible for the cell cycle arrest and whether there are
differences between different statins in terms of p27Kip1.
Because the small GTP-binding proteins, Ras and Rho, require posttranslational modification for membrane localization and activity and are implicated in
cell cycle regulation, they are likely targets for the direct antiproliferative vascular effects of statins. Ras can promote cell cycle progression via activation
of the MAP kinase pathway (84), whereas RhoA causes cellular proliferation
through destabilizing p27Kip1 protein (85). Interestingly, inhibition of vascular
SMC proliferation by statins was reversed by GGPP, but not FPP or LDL-C (39).
Indeed, direct inhibition of RhoA by Clostridium botulinum C3 transferase, which
ADP-ribosylates and inactivates RhoA, or by a dominant-negative RhoA mutant increased p27Kip1 and inhibited Rb hyperphosphorylation and SMC proliferation following PDGF stimulation. Taken together, these findings indicate that
RhoA mediates PDGF-induced SMC proliferation and that inhibition of RhoA
by statins is the predominant mechanism by which statins inhibit vascular SMC
proliferation.
7 Jan 2005
13:56
AR
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
99
7 Jan 2005
13:56
100
AR
LIAO
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAUFS
molecules, such as intercellular adhesion molecule-1 (ICAM-1), which are involved in the recruitment of inflammatory cells (116). Furthermore, statins attenuate P-selectin expression and leukocyte adhesion in normocholesterolemic animals
by increasing endothelial NO production (117, 118). This cholesterol-independent
effect of statins was absent in eNOS-deficient mice, suggesting that eNOS mediated the vascular protective effects of statins (119).
The activation of T-lymphocytes and the control of the immune response are
mediated by the major histocompatibility complex class II (MHC-II) and CD40/
CD40L. Under physiological conditions, antigen-presenting cells express MHC-II
constitutively, whereas the induction of interferon gamma (INF- ) leads to an increase of MHC-II expression in numerous cells, including human endothelial cells
and monocytes. An important regulator in this pathway is the transactivator CIITA.
Statins inhibit MHC-II expression on endothelial cells and monocyte-macrophages
via inhibition of the promotor IV of the transactivator CIITA and thereby repress
MHC-II-mediated T cell activation (120). In addition, statins have been shown to
decrease CD40 expression and CD40-related activation of vascular cells (121).
A clinical marker of inflammation is high-sensitivity C-reactive protein (hsCRP) (122). hs-CRP is an acute phase reactant that is produced by the liver in
response to proinflammatory cytokines, such as interleukin-6 (IL-6), and reflects
low-grade systemic inflammation (123). Elevated levels of hs-CRP have been
shown to be predictive of increased risk for coronary artery disease (CAD) in apparently healthy men and women (124, 125). hs-CRP is elevated in patients with
CAD, coronary ischemia and myocardial infarction compared with normal subjects
(126). It has been suggested that CRP could also contribute to the development
of atherosclerosis by binding to modified LDL-C within atherosclerotic plaques
(127, 128). Once CRP becomes bound, it activates complement, which has been
shown to play a role in promoting atherosclerotic lesion progression (129). Furthermore, CRP has been shown to induce plasminogen activator inhibitor (PAI)-1
expression and complement activation, increase the expression of cellular adhesion
molecules, and decrease eNOS expression, leading to propensity for thrombosis,
inflammation, and endothelial dysfunction. Indeed, transgenic overexpression of
human CRP in transgenic mice leads to increased thrombosis and vascular inflammation following arterial injury (130). However, further studies are needed to fully
elucidate the role CRP plays in atherosclerosis and cardiovascular risk.
Statin therapy lowers hs-CRP levels in hypercholesterolemic patients (122, 131,
132). In the CARE trial, statins significantly decreased plasma hs-CRP levels over
a five-year period in patients who did not experience recurrent coronary events
(133, 134). Similarly, an analysis of baseline and one-year follow-up from the AFCAPS/TexCAPS demonstrated that hs-CRP levels were reduced in statin-treated
patients who were free of acute major coronary events (122). Furthermore, preliminary data from the Pravastatin Inflammation/CRP Evaluation (PRINCE) study
confirm that statin therapy can significantly reduce serum hs-CRP levels in primary and secondary prevention populations (135). Following 24 weeks of therapy
with a statin, the hs-CRP level was reduced by approximately 13% in primary and
7 Jan 2005
13:56
AR
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
101
7 Jan 2005
13:56
102
AR
LIAO
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAUFS
7 Jan 2005
13:56
AR
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
103
20,000 people with cerebrovascular disease or other high-risk conditions (156). The
proportional reductions in stroke were approximately one quarter in all subcategories studied, including those aged over 70 years at entry and those presenting
with different levels of blood pressure or lipids, even when the pretreatment LDLC was below 3.0 mmol/L (116 mg/dl). Thus, the findings of these large statin trials
raise the interesting question of how a class of cholesterol-lowering agents can
reduce ischemic stroke when ischemic stroke is not related to cholesterol levels. It
appears likely that there are cholesterol-independent effects of statins, which are
beneficial for ischemic stroke. Some of these beneficial effects may relate to the
effects of statins on endothelial and platelet function.
Cerebrovascular tone and blood flow are regulated by endothelium-derived
NO (157). Mutant mice lacking eNOS (eNOS/) are relatively hypertensive and
develop greater proliferative and inflammatory response to vascular injury (158).
Indeed, eNOS/ mice develop larger cerebral infarcts following cerebrovascular
occlusion (159). Thus, the beneficial effects of statins in ischemic stroke may, in
part, be due to their ability to upregulate eNOS expression and activity (32, 55). For
example, mice that were prophylactically treated with statins for up to two weeks,
have 25%30% higher cerebral blood flow and 50% smaller cerebral infarct sizes
following cerebrovascular occlusion (160). No increase in cerebral blood flow or
neuroprotection was observed in eNOS/ mice treated with statins, indicating
that the upregulation of eNOS accounts for most, if not all, of the neuroprotective
effects of these agents. Interestingly, treatment with statins did not affect blood
pressure or heart rate before, during, or after cerebrovascular ischemia and did not
alter serum cholesterol levels in mice, consistent with the cholesterol-independent,
neuroprotective effects of statins.
In addition to increases in cerebral blood flow, other beneficial effects of statins
are likely to occur that can impact on the severity of ischemic stroke. For example, statins attenuate P-selectin expression and leukocyte adhesion via increases
in NO production in a model of cardiac ischemia and reperfusion (161, 162). Others have reported that statins upregulate tissue-type plasminogen activator (t-PA)
and downregulate plasminogen activator inhibitor (PAI)-1 expression through a
similar mechanism involving inhibition of Rho geranylgeranylation (57). Thus,
the absence of neuroprotection in eNOS-deficient mice emphasizes the importance of endothelium-derived NO in not only augmenting cerebral blood flow
but also, potentially, in limiting the impact of platelet and white blood cell accumulation on tissue viability following ischemia. In humans, atherosclerosis of
precerebral arteries causes stroke through plaque disruption and artery-to-artery
thromboembolism, and, in contrast to the mouse models, statins exert additional
stroke-protective effects in humans through their anti-atherosclerotic and plaquestabilizing effects. Furthermore, the antiinflammatory actions and mobilization of
endothelial progenitor cells of statins may also contribute to neuroprotection. It
is therefore possible that statins have contributed to the decrease in the incidence
of ischemic strokes in clinical trials, in part, by reducing cerebral infarct size to
levels that were clinically unappreciated.
7 Jan 2005
13:56
104
AR
LIAO
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAUFS
7 Jan 2005
13:56
AR
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
105
7 Jan 2005
13:56
106
AR
LIAO
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAUFS
Figure 5 Relationship between LDL-C reduction and risk of cardiovascular events. (Left
panel) Decrease in LDL-C (% reduction) is correlated with reduction in risk of nonfatal
myocardial infarctions (MI) or coronary heart disease (CHD) among statin (WOSCOPS,
CARE, and 4S) and nonstatin (LRC-CPPT and POSCH) trials. Note that the relationship
(slope) holds between statin and nonstatin trials, suggesting that the beneficial effects of
statins are likely due to only cholesterol lowering. (Right panel) Decrease in LDL-C (%
reduction) is correlated with reduction in risk of nonfatal myocardial infarctions (MI) or
coronary heart disease (CHD) among statin (WOSCOPS, CARE, and 4S) and nonstatin
(LRC-CPPT and POSCH) trials after 4.5 years of treatment. Note that the nonstatin trials
(LRC-CPPT and POSCH; dashed lines) show less cardiovascular benefits than statin trials
(WOSCOPS, CARE, and 4S) and they no longer fall on the same slope (solid line).
7 Jan 2005
13:56
AR
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
107
TABLE 1
Statin pleiotropy
Effect
Benefit
7 Jan 2005
13:56
108
AR
AR232-PA45-04.tex
LIAO
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAUFS
SUMMARY
Statins exert many pleiotropic effects in addition to the lowering of serum cholesterol levels. These additional properties include beneficial effects on endothelial
function and blood flow, decreasing LDL-C oxidation, enhancing the stability
of atherosclerotic plaques, inhibiting vascular smooth muscle proliferation and
platelet aggregation, and reducing vascular inflammation (Table 1). Recent evidence suggest that most of these effects are mediated by statins inhibitory effect
on isoprenoid synthesis. In particular, inhibition of Rho GTPases in vascular wall
cells by statins leads to increased expression of atheroprotective genes and inhibition of vascular SMC proliferation. It remains to be determined which of and
to what extent these pleiotropic effects account for the clinical benefits of statin
therapy beyond cholesterol lowering.
ACKNOWLEDGMENTS
The work described in this paper was supported in part by the National Institutes of
Health (HL-52233, HL-48743, and NS-10828), the American Heart Association
Bugher Foundation Award, and the Deutsche Forschungsgesellschaft. Dr. Liao is
an Established Investigator of the American Heart Association.
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Libby P, Sukhova G, Lee RT, Liao JK.
1997. Molecular biology of atherosclerosis. Int. J. Cardiol. 62 (Suppl 2):S2329
2. Sytkowski PA, Kannel WB, DAgostino
7 Jan 2005
13:56
AR
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
11.
12.
13.
14.
15.
16.
17.
18.
19.
109
7 Jan 2005
13:56
110
AR
LIAO
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAUFS
7 Jan 2005
13:56
AR
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
111
7 Jan 2005
13:56
112
59.
60.
61.
62.
63.
64.
65.
66.
AR
LIAO
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAUFS
67.
68.
69.
70.
71.
72.
73.
74.
7 Jan 2005
13:56
AR
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
113
7 Jan 2005
13:56
114
93.
94.
95.
96.
97.
98.
99.
100.
AR
LIAO
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAUFS
101.
102.
103.
104.
105.
106.
107.
108.
7 Jan 2005
13:56
AR
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
119.
120.
121.
122.
123.
124.
125.
126.
127.
115
7 Jan 2005
13:56
116
128.
129.
130.
131.
132.
133.
134.
135.
136.
AR
LIAO
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAUFS
atherogenesis: binding of CRP to degraded, nonoxidized LDL enhances complement activation. Arterioscler. Thromb.
Vasc. Biol. 19:234854
Zhang YX, Cliff WJ, Schoefl GI, Higgins
G. 1999. Coronary C-reactive protein distribution: its relation to development of
atherosclerosis. Atherosclerosis 145:375
79
Torzewski J, Bowyer DE, Waltenberger
J, Fitzsimmons C. 1997. Processes in
atherogenesis: complement activation.
Atherosclerosis 132:13138
Danenberg HD, Szalai AJ, Swaminathan
RV, Peng L, Chen Z, et al. 2003. Increased
thrombosis after arterial injury in human
C-reactive protein-transgenic mice. Circulation 108:51215
Musial J, Undas A, Gajewski P, Jankowski
M, Sydor W, Szczeklik A. 2001. Antiinflammatory effects of simvastatin in
subjects with hypercholesterolemia. Int. J.
Cardiol. 77:24753
Ridker PM, Rifai N, Lowenthal SP. 2001.
Rapid reduction in C-reactive protein with
cerivastatin among 785 patients with primary hypercholesterolemia. Circulation
103:119193
Ridker PM, Rifai N, Pfeffer MA, Sacks
F, Braunwald E. 1999. Long-term effects
of pravastatin on plasma concentration of
C-reactive protein. The Cholesterol and
Recurrent Events (CARE) Investigators.
Circulation 100:23035
Ridker PM, Rifai N, Pfeffer MA, Sacks
FM, Moye LA, et al. 1998. Inflammation, pravastatin, and the risk of
coronary events after myocardial infarction in patients with average cholesterol levels. Cholesterol and Recurrent
Events (CARE) Investigators. Circulation
98:83944
Albert MA, Staggers J, Chew P, Ridker
PM. 2001. The pravastatin inflammation
CRP evaluation (PRINCE): rationale and
design. Am. Heart J. 141:89398
Wang SM, Tsai YJ, Jiang MJ, Tseng YZ.
1997. Studies on the function of rho A pro-
137.
138.
139.
140.
141.
142.
143.
144.
145.
7 Jan 2005
13:56
AR
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
146.
147.
148.
149.
150.
151.
152.
153.
117
7 Jan 2005
13:56
118
164.
165.
166.
167.
168.
169.
170.
171.
172.
AR
LIAO
AR232-PA45-04.tex
AR232-PA45-04.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAUFS
173.
174.
175.
176.
177.
178.
179.
180.
HI-RES-PA45-04-Liao.qxd
1/7/05
2:22 PM
Page 1
HI-RES-PA45-04-Liao.qxd
C-2
LIAO
1/7/05
2:22 PM
Page 2
LAUFS
7 Jan 2005
13:57
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
P1: GCE
LaTeX2e(2002/01/18)
10.1146/annurev.pharmtox.45.120403.095843
INTRODUCTION
Obesity can be viewed as an energy storage disorder where weight gain results
from an energy imbalance (i.e., energy input exceeding output), with most of the
excess calories stored as triglyceride in adipose tissue (AT). A strong correlation
exists between the prevalence of obesity and the prevalence of type 2 diabetes.
Excessive AT accumulation is a key pathological contributor to the metabolic
syndrome characterized by insulin resistance and dyslipidemia that leads to type
2 diabetes and an increased risk for cardiovascular diseases (1). What causes this
association between obesity, insulin resistance, and the development of type 2
diabetes? Although skeletal muscle, liver, and pancreas dysfunctions have been
implicated as the major sites for development of insulin resistance, a number of
recent results focus attention on AT as being a primary site (2, 3).
ATs represent complex organs, and a preliminary definition is useful to delineate
the topic covered in the present review. The major distinctive characteristics that
distinguish white AT (WAT) and brown AT (BAT) are detailed in a number of
0362-1642/05/0210-0119$14.00
119
7 Jan 2005
13:57
120
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAFONTAN
7 Jan 2005
13:57
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
121
in the regulation of appetite, food intake, glucose disposal, and energy expenditure.
They are able to protect against the establishment of insulin resistance by actions on
liver, skeletal muscle, and pancreatic function. They also contribute to the prevention or worsening of atherogenic processes. In addition, some of the factors secreted
by the adipocyte exert local autocrine and paracrine actions mainly affecting AT
remodeling, adipogenesis, and angiogenesis and are not found in the circulation.
The topography of fat distribution plays an important role in the appearance
of health risks. Abdominal visceral fat extent is an important link between the
many facets of the metabolic syndrome: glucose intolerance, hyperinsulinemia,
hypertriglyceridemia, and other features such as hypertension and altered HDL
and VLDL levels (1618). Although it is quite well accepted that upper body
obesity, with increased visceral fat, should be considered as a factor that initiates
or exacerbates an individuals susceptibility to the components of the metabolic
syndrome, the causality is poorly understood. What causes this consistent association between obesity and the development of type 2 diabetes? What are the factors
interfering with the adipocytes from various depots that could explain the appearance of metabolic disorders? What is the contribution of adipocyte secretions to
the generation of diseases related to the development of obesity?
7 Jan 2005
13:57
122
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAFONTAN
pathways are depicted in Figure 1. Rodents possess abundant 3-adrenergic receptors ( 3-ARs) in the white fat cells, whereas in human fat cells, the role of the
3-AR remains quite puzzling and controversial (24). In human fat cells, both 1and 2-ARs initiate the activation of the lipolytic cascade by stimulation of cyclic
AMP (cAMP) production, activation of cAMP-dependent protein kinase A (PKA)
leading to phosphorylation of perilipin and hormone-sensitive lipase (HSL), and
promotion of lipolysis in vitro. The originality of the human fat cell is related to the
presence of abundant 2-adrenergic receptors ( 2-ARs): their stimulation inhibits
cAMP production and lipolysis (24, 25).
Differences exist in the adrenergic regulation of lipolysis in AT from different
sites in normal-weight subjects and in obese subjects (18, 26, 27). The lipolytic
response of isolated fat cells to catecholamines is weaker in the subcutaneous
gluteal/femoral and abdominal AT than in visceral AT. Lipolytic defects are explainable by the reduced expression or function of HSL and/or of proteins that interact with HSL [e.g., adipocyte lipid binding protein (ALBP)] or the lipid droplet,
such as perilipin. Alterations of the signaling pathways, such as reduced 12-AR
responsiveness or increased 2-AR responsiveness (and a possible association or
combination of defects), are also important. These site-related differences are more
noticeable in women than in men (24, 2729). An enhanced 2-AR responsiveness
associated with a concomitant decrease in -AR responsiveness explains the lower
lipolytic effect of catecholamines in gluteal/femoral fat cells of normal and obese
women and abdominal fat cells of obese men. Reduced lipid mobilization occurs
during exercise in subcutaneous AT of obese subjects (30). Functional changes in
12/ 2-AR balance appear with the extent of the fat mass and are related to fat cell
hypertrophy. Hypertrophic subcutaneous (abdominal, femoral) fat cells are known
to be the least responsive to the lipolytic action of catecholamines; they exhibit
the highest amount of 2-ARs and the lowest amount of 12-ARs. Increased expression of the 2-AR (and the concomitant decrease of -responsiveness) with
fat cell hypertrophy could be a physiological adaptation leading to a reduction of
the lipolytic responsiveness of the hypertrophied adipocytes of some fat deposits.
The mechanisms leading to the opposite regulation of the expression of 12- and
2-ARs as cells become hypertrophied are unknown.
Whatever the mechanism controlling AR expression, limitation of basal and
SNS-dependent lipolysis avoids excessive NEFA release from some fat deposits.
A recent study aimed at the direct assessment of fasting AT metabolism using
arterio-venous differences in defined depots has shown that the buttock is metabolically silent in terms of fatty acid release compared with the abdomen (31). The
buffering action of NEFAs by AT is an important phenomenon (32). When the
NEFA buffering system is inadequate, other tissues are exposed to elevated NEFA
concentrations. A role for the 2-AR gene in determining the propensity to store
fat in the abdominal area, independent of total body fatness, has recently been
reported (33).
Profound unresponsiveness of the subcutaneous AT to neurally stimulated lipolysis has been described in obese subjects (34). Reduced 2-adrenergic lipolytic
responsiveness has been reported in fat cells from obese subjects or subjects with
7 Jan 2005
13:57
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
123
a reduced isoproterenol sensitivity (35). In addition, an increased antilipolytic responsiveness linked to 2-AR stimulation has also been found in subcutaneous
adipocytes of obese individuals of both sexes. The lipolytic defects revealed in
fat cells have been confirmed during in vivo studies (36, 37). Using in situ microdialysis, a specific impairment in the capacity of 2-AR agonists to promote
lipolysis has been reported in the subcutaneous abdominal AT of obese adolescent girls (38). Moreover, when selective 1- and 2-AR-agonists are administered
intravenously, the increase in lipolysis and thermogenesis promoted by selective
2-adrenergic stimulation (salbutamol) was reduced in obese subjects. Conversely,
1-AR-mediated (dobutamine) metabolic processes (i.e., lipolysis, thermogenesis, and lipid oxidation) were similar in obese and lean men. In conclusion, 2adrenergic-mediated increases in thermogenesis and lipid oxidation are impaired
in the obese (39).
The putative role of 2- and 3-AR polymorphisms in the etiology of lipolysis disturbances and obesity has recently been reviewed (35, 40). To summarize, polymorphisms in the coding and noncoding sequences in the human 2-AR
gene could be of major importance for obesity, energy expenditure, and 2-ARdependent lipolytic function. Full -adrenergic activation of the human fat cell
usually requires synergistic activation of 1- and 2-ARs. A 2-adrenergic defect
could be sufficient to alter normal -adrenergic responsiveness. In addition, in human fat cells, any reduction of 2-AR-mediated lipolytic response will disturb the
normal functional balance existing between 2- and -AR-mediated effects and
amplify the reduction of the lipolytic responsiveness initiated by the physiological
amines in stressful situations. All the discussions related to the adrenergic regulation of lipolysis must be expanded in terms of regulation to all cAMP-related
events existing in fat cells.
7 Jan 2005
13:57
124
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAFONTAN
GROWTH HORMONE
7 Jan 2005
13:57
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
125
commonly acting in rodent fat cells have no effect on human fat cells. Glucagon
and glucagon-like peptide-1 (GLP-1) do not stimulate in vitro lipolysis. Moreover, no significant effect of either GLP-1 or glucagon on either lipolysis rate or
blood flow was detected in muscle or AT during local or experimental i.v. hyperglucagonemia (57, 58). Parathyroid hormone (59) stimulates lipolysis in human fat
cells at rather high extraphysiological concentrations. Human adipocytes express
both IL-6 and its receptor system consisting of the IL-6 receptor and the signal
transducing protein gp130. IL-6 administered in the normal physiological concentration range elicits lipolytic effects in human subcutaneous AT in vivo (60, 61).
IL-6 stimulates lipolysis in human adipocytes and exerts anti-insulin actions (62).
IL-6 also induces the expression of SOCS-3, a potential inhibitor of insulin signaling (63, 64). Whatever the results, it could be premature to include IL-6 inside
a physiological loop of lipid mobilization regulation. Leptin induces a novel form
of lipolysis in which glycerol is released without proportional release of NEFA
and increase in peroxisome proliferator-activated receptor- (PPAR-) and NEFA
oxidation in rat fat cells (65). It is unknown if the same leptin-dependent action
operates in human fat cells.
Cortisol is less potent than catecholamines in the stimulation of lipolysis; the
lipolytic effect is delayed and in vivo action is counteracted by corticoid-promoted
insulin release (66). Cortisol-induced lipid mobilization is observed when cortisolinduced insulin increment is prevented (67). Short-term treatment with a standard
dose of corticosteroids (i.e., prednisolone) induces increased abdominal adipose
tissue lipolysis, hyperglucagonemia and insulin resistance, whereas GH levels are
unaffected (68).
Stimulation of lipolysis by tumor necrosis factor-
(TNF-) is not direct because it becomes apparent only after long-lasting exposure
of human and rodent adipocytes to the cytokine (69). TNF--induced lipolysis,
as well as inhibition of insulin-stimulated glucose transport, is predominantly
mediated by the receptor TNFR1 (70, 71). TNF- could regulate lipolysis, in part,
by decreasing perilipin protein levels at the lipid droplet surface and activating the
extracellular signal-related kinase (ERK) pathway (72). Blunting the endogenous
inhibition of lipolysis through Gi protein downregulation is also another possible
mechanism (73). In human fat cells, TNF- activates the three mammalian mitogen
activated protein kinases (MAPK) in a distinct time- and concentration-dependent
manner. TNF--induced lipolysis is mediated by only p44/42 and Jun kinase but
not by p38 kinase (74).
MISCELLANEOUS AGENTS
7 Jan 2005
13:57
126
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAFONTAN
tumor-related LMF, was detected in the major fat deposits of mice and in 3T3-L1
cells. ZAG expression and protein was also found in human fat cells (visceral and
subcutaneous AT). ZAG is a new adipose tissue protein factor that may be involved
in the modulation of lipolysis in adipocytes (78).
Various hormones and
autacoid agents are known to negatively control adenylyl cyclase activity and
inhibit cAMP production and lipolysis in fat cells. The effects are mediated by
plasma membrane receptors, the stimulation of which inhibits adenylyl cyclase
and cAMP production (Figure 1). In addition, the stimulation of leptin secretion
was also observed with various agonists (A1-adenosine, 2-AR, and NPY-Y1 receptor agonists) (79, 80). The receptor of nicotinic acid (niacin), a well-known
lipid-lowering drug, has recently been discovered. The orphan G proteincoupled
receptor protein up-regulated in macrophages by interferon (mouse PUMAG, human HM74), which is highly expressed in adipose tissue, is a nicotinic acid
receptor that mediates the antilipolytic and lipid-lowering effect of nicotinic acid
in vivo (81). Agonists leading to activation of Gi protein-coupled receptors of the
adipocytes limit NEFA release and represent putative antihyperlipidemic drugs.
All these antilipolytic agents will also exert leptin-secreting effects. Antagonists
of such receptors, relieving inhibition of cAMP production promoted by the endogenous ligands, enhance the lipolytic activity of the fat cell. The physiological
relevance of all these in vitro investigations is yet to be established.
Leptin
It is clear that leptin is an important part of the lipostatic system because it signals
the size of the energy reserves existing in the body and controls fuel mobilization
and utilization (82). Leptin crosses the blood-brain barrier, enters the central nervous system (CNS), and stimulates the long form of its receptor (Ob-Rb) located
in the arcuate nucleus of the hypothalamus. A complex set of leptin receptor isoforms exist (83). Results in rodents fit with the lipostatic paradigm, postulating
that the adipocyte, through leptin, is able to provide a peripheral message of fat
7 Jan 2005
13:57
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
127
mass repletion to the CNS areas involved in the regulation of energy balance to
regulate energy intake and energy expenditure and limit fat deposition (82, 84).
Leptin is a pleiotropic molecule that may regulate a number of other biological
processes, as recently reviewed (15, 85, 86). Understanding of metabolic effects
of leptin expanded after the discovery of its mechanism of action in liver and muscle. Leptin directly stimulates 5 -AMP-activated protein kinase (AMPK), which
increases ATP-producing catabolic pathways, such as beta-oxidation, glycolysis,
and mitochondrial biogenesis, and concomitantly decreases ATP-consuming anabolic pathways (87).
Under experimental adenovirus-induced hyperleptinemia in rats, the fatty acids
inside the white adipocytes appear to be oxidized (88, 89). Leptin confines storage
of excess calories to adipocytes and spares the appearance of chronic steatosis in
nonadipocyte cells. It was proposed that in humans the metabolic syndrome might
be the equivalent of the lipotoxic syndrome described in rodents (90). It will be
essential to evaluate whether these novel effects appear at leptin concentrations
that are found in a physiological context in humans. Some actions of leptin at high
concentrations could be associated to effects independent of Ob-Rb and recruit
additional transducing pathways. Cross talk of leptin pathways with other cytokinerelated pathways cannot be excluded because leptin has similarities to the family of
long-chain helical cytokines that includes IL-6, IL-11, ciliary neurotrophic factor,
and leukemia inhibitory factor.
The subcutaneous fat depot is the major source of leptin, owing to the combination of a mass effect (subcutaneous fat being the major depot in men and women)
and the higher secretion rate in the subcutaneous than in the visceral depots (91).
Adipocyte size and anatomical location appear to be the major determinants of leptin mRNA expression. In vivo, overfeeding and obesity, glucocorticoid treatments,
glucose, and insulin administration increase circulating leptin levels, whereas fasting, sustained exercise, cold exposure, and SNS activation reduce leptin levels.
In vitro, positive effectors of leptin production include glucose, insulin, glucocorticoids, TNF-, estrogens, IL-1, agents acting through Gi protein-dependent
pathways, and melanin concentrating hormone. Conversely, catecholamines and
cAMP agonists, -adrenergic agonists, androgens, polyunsaturated fatty acids,
peroxisome proliferator-activated receptor agonists, and phorbol esters negatively regulate leptin production. Insulin and glucocorticoids affect transcriptional
mechanisms that increase leptin mRNA levels but also increase the traffic out of
the adipocyte, which involves constitutive and regulated secretory pathways (14,
86, 92).
Interleukin-6
Plasma IL-6 levels are increased in obese subjects and correlate with fat mass and
body mass index (BMI). High levels of plasma IL-6 are found in type 2 diabetes
and also correlate with fasting insulin levels. The capacity of human adipocytes
to release IL-6 was observed in AT explants and freshly isolated fat cells (93) and
7 Jan 2005
13:57
128
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAFONTAN
human adipocytes differentiated in vitro (62). In vivo studies have shown that the
cytokine IL-6 is secreted by subcutaneous fat in humans (94). In subcutaneous
AT, IL-6 secretion increases tenfold during the postexercise rest period following
a one-hour endurance exercise, and a concomitant increment of NEFA output was
observed; this suggests a postexercise lipid mobilizing contribution of the cytokine
(95). The IL-6 receptor and the gp-130 protein of cytokine pathways are expressed
in human fat cells, suggesting that a direct paracrine action of IL-6 on the human
fat cell is possible (96). IL-6 secretion is strongly stimulated by -AR activation
and mildly suppressed by glucocorticoids. To conclude, a hormonally regulated
IL-6 secretion occurs in mature human fat cells and it is probable that a local
paracrine action of IL-6 on adipocytes exists.
7 Jan 2005
13:57
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
129
Resistin
Resistin, for resistance to insulin, is 10-kDa adipocyte-secreted protein that possesses hormonal properties and has been claimed to represent an important link
between obesity and insulin resistance. In the original paper, resistin administration
was reported to cause glucose intolerance and insulin resistance in mice, whereas
resistin antibody administration improved glucose intolerance. Moreover, serum
levels of resistin were higher in mouse models of obesity and decreased after peroxisome proliferator-activated receptor (PPAR ) agonist (e.g., thiazolidinedione)
treatment. WAT resistin mRNA and serum protein levels dropped during fasting
and increased during refeeding (118). Other groups using different methods have
confirmed the existence of resistin (119121). Resistin has a rapid effect on hepatic, but not peripheral, insulin sensitivity (119). Mice lacking resistin (rstn/)
exhibit low blood glucose levels after fasting owing to reduced hepatic glucose
production. This is partly mediated by AMPK activation and decreased expression
of gluconeogenic enzymes in the liver (122). The original concept is still open to
discussion after major controversial results concerning resistin expression in obese
rodents and thiazolidinedione effects. The role of resistin in human insulin resistance remains quite controversial. Very low levels of resistin mRNA were found in
7 Jan 2005
13:57
130
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAFONTAN
7 Jan 2005
13:57
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
131
7 Jan 2005
13:57
132
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAFONTAN
7 Jan 2005
13:57
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
133
7 Jan 2005
13:57
134
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAFONTAN
NEFA release by fat cells is also under the control of all the antilipolytic agents,
including insulin, but also hormones and agents acting via Gi protein-coupled receptors. Enhancement of antilipolytic effects will reduce plasma NEFA levels in
subjects with increased NEFA (e.g., insulin-resistant subjects and patients with
type 2 diabetes). Antilipolytic strategies based on nicotinic acid derivatives and
adenosine A1-receptor agonists have been proposed to reduce plasma NEFA and
TAG. Insulin sensitization of fat cells contributes to reducing the release of NEFAs
by fat cells. Insulin sensitizers that enhance insulin action by modulating the events
following the binding of insulin to its receptor and/or by activating transcription
factors affecting the expression of the genes involved in the action of insulin in
insulin-sensitive tissues are of interest to improve insulin action in the obese. Inhibitors of the protein tyrosine phosphatase 1B activity as well as of other putative
negative regulators of insulin signaling are candidates to reduce insulin resistance
and improve insulin action and may have potential in the future as antiobesity
agents (168). Several agonists active at both PPAR and PPAR represent promising tools with potential antidiabetic and lipid-lowering properties (3, 169). PPARs,
widely distributed in tissues and cell types, constitute multiple therapeutic targets
(168). Ideally, drugs possessing both PPAR/ agonist potencies are expected to
provide the best means to decrease multiple risk factors for morbidity and mortality existing in diabetic patients by acting on fat cells and liver (170). PPAR is
predominantly expressed in adipocytes, and the various beneficial metabolic effects reported for PPAR agonists are thought to result from direct actions on AT
along with secondary impact in skeletal muscle and liver. The beneficial actions
of PPAR agonists on muscle, liver, and vessels (i.e., atherosclerosis risk) are mediated by their ability (a) to improve insulin-mediated uptake and metabolism of
glucose and NEFA in the adipocyte; (b) to induce the production of adiponectin by
adipocytes (e.g., adiponectin is a relatively early and specific response to activation of PPAR ); and (c) to reduce production of adipocyte-derived factors leading
to insulin resistance, such as resistin, inflammatory molecules, and TNF-. The
insulin-sensitizing potency of PPAR agonists could be related to their antiinflammatory action because they inhibit TNF- action on adipocytes and limit production of inflammatory molecules by fat cells and monocytes/macrophages, which
have been found to be abundant in obese AT. In view of the multiple metabolic
and vascular actions of adiponectin, it is possible that a number of ameliorations
of metabolic disturbances related to the metabolic syndrome attributable to the effects of PPAR agonists could be related to their action on adiponectin production
and release by fat cells. Investigations in adiponectin-deficient mice will facilitate
the answer to the question.
7 Jan 2005
13:57
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
135
7 Jan 2005
13:57
136
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAFONTAN
Although they are on the market and of interest for the treatment of type 2 diabetes,
PPAR agonists, such as thiazolidinediones, are not effective against obesity because they are known to promote recruitment/differentiation of new fat cells while
improving insulin sensitivity. Moreover, adipogenesis could be induced in bone
marrow stromal cells. It is a major negative side effect; PPAR agonists might
promote weight gain in patients still having serious metabolic disorders, and their
short-term benefits could be reduced in the long term. Important efforts are being made to identify new PPAR modulators having antidiabetic action without
promoting fat cell differentiation and weight increase. In addition to adipogenesis, angiogenic processes play an important role in the development of AT mass
(176). Adipocyte productions, such as leptin and vascular endothelial growth factor (VEGF), are known to exert proangiogenic effects and contribute to vascular
development in AT. The extent of AT mass is sensitive to angiogenesis inhibitors
(177); strategies aimed at the limitation of vascular supply in fat are opening new
perspectives that merit future attention.
In humans, in whom there are no BAT depots in adults, conversion of white
adipocytes into brown-like fat cells could be a great challenge. Appearance of
brown adipocytes is possible in certain conditions in adults with pheochromocytoma. In a recent study using adenovirus expressing human PGC-1, a PPAR
coactivator has demonstrated that it is possible to promote a metabolic shift in human white fat cells from lipid storage to fatty acid utilization with a concomitant
induction of UCP-1, mitochondria respiratory chain proteins, and fatty acid oxidation enzymes. Palmitate oxidation was indeed elevated in such modified adipocytes
(178). Such a study suggests that future strategies aimed at altering the phenotype
of human white adipocytes could be envisaged for the treatment of obesity.
7 Jan 2005
13:57
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
137
LITERATURE CITED
1. Reaven GM, Abbasi F, McLaughlin T.
2004. Obesity, insulin resistance and cardiovascular disease. Rec. Prog. Horm.
Res. 59:20723
2. Hotasmiligil GS. 2000. Molecular mechanisms of insulin resistance and the role of
the adipocyte. Int. J. Obes. Relat. Metab.
Disord. 24:S2327
3. Arner P. 2003. The adipocyte in insulin
resistance: key molecules and the impact of the thiazolidinediones. Trends Endocrinol. Metab. 14:13745
4. Lafontan M, Berlan M. 1993. Fat cell
adrenergic receptors and the control of
white and brown fat cell function. J. Lipid
Res. 34:105791
5. Robidoux J, Martin TL, Collins S. 2004.
b-Adrenergic receptors and regulation of
energy expenditure: a family affair. Annu.
Rev. Pharmacol. Toxicol. 44:297323
6. Cannon B, Nedergaard J. 2004. Brown
adipose tissue: function and physiological
significance. Physiol. Rev. 84:277359
7. Rousset S, Alves-Guerra M, Mozo J,
Miroux B, Cassard-Doulcier A-M, et al.
2004. The biology of mitochondrial uncoupling proteins. Diabetes 53(Suppl. 1):
S13035
8. McGarry JD. 2002. Dysregulation of fatty
9.
10.
11.
12.
13.
14.
15.
16.
7 Jan 2005
13:57
138
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAFONTAN
perils of portliness. Causes and consequences of visceral adiposity. Diabetes
49:88388
Wajchenberg BL. 2000. Subcutaneous
and visceral adipose tissue: their relation
to the metabolic syndrome. Endocr. Rev.
21:697738
Lafontan M, Berlan M. 2003. Do regional
differences in adipocyte biology provide
new pathophysiological insights? Trends
Pharmacol. Sci. 24:27683
Frayn KN, Karpe F, Fielding BA, McDonald IA, Coppack SW. 2003. Integrative
physiology of human adipose tissue. Int.
J. Obesity 27:87588
Bartness TJ. 2002. Dual innervation of
white adipose tissue: some evidence for
parasympathetic nervous system involvement. J. Clin. Invest. 10:123537
Dodt C, Lonnroth P, Fehm HL, Elam
M. 1999. Intraneural stimulation elicits
an increase in subcutaneous interstitial
glycerol levels in humans. J. Physiol.
521:54552
Kreier F, Fliers E, Voshol PJ, Eden
CGV, Havekes LM, et al. 2002. Selective
parasympathetic innervation of subcutaneous and intra-abdominal fat: functional
implications. J. Clin. Invest. 110:1243
50
Fliers E, Kreier F, Voshol J, Havekes LM,
Sauerwein HP, et al. 2003. White adipose tissue: getting nervous. J. Neuroendocrinol. 15:100510
Lafontan M, Berlan M. 1995. Fat cell
2-adrenoceptors: the regulation of fat
cell function and lipolysis. Endocr. Rev.
16:71638
Lafontan M, Berlan M. 1980. Evidence for the alpha2-nature of the alphaadrenergic receptor inhibiting lipolysis in
human fat cells. Eur. J. Pharmacol. 66:87
93
Arner P. 1988. Control of lipolysis and
its relevance to development of obesity in
man. Diabetes Metab. Rev. 4:50715
Langin D, Lucas S, Lafontan M. 2000.
Millenium fat-cell lipolysis reveals unsus-
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
7 Jan 2005
13:57
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
39.
40.
41.
42.
43.
44.
45.
46.
47.
139
7 Jan 2005
13:57
140
57.
58.
59.
60.
61.
62.
63.
64.
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAFONTAN
growth hormone on lipid metabolism. Am.
J. Physiol. 279:E5017
Bertin E, Arner P, Bolinder J, HagstromToft E. 2001. Action of glucagon and
glucagon-like peptide-1-(7-36) amide on
lipolysis in human subcutaneous adipose
tissue and skelatal muscle in vivo. J. Clin.
Endocrinol. Metab. 86:122934
Gravholt C, Moller N, Jensen M, Christiansen J, Schmitz. 2001. Physiological
levels of glucagon do not influence lipolysis in abdominal adipose tissue as assessed
by microdialysis. J. Clin. Endocrinol.
Metab. 86:208589
Taniguchi A, Kataoka K, Kono T, Oseko F, Okuda H, et al. 1987. Parathyroid hormone-induced lipolysis in human adipose tissue. J. Lipid Res. 28:490
94
Lyngso D, Simonsen L, Bulow J. 2002.
Metabolic effects of interleukin-6 in human splanchnic and adipose tissue. J.
Physiol. 543:37986
van Hall G, Steenberg A, Sacchetti M,
Fisher C, Keller C, et al. 2003. Interleukin6 stimulates lipolysis and fat oxidation
in humans. J. Clin. Endocrinol. Metab.
88:300510
Path G, Bornstein SR, Gurniak M,
Chrousos GP, Scherbaum WA, Hauner
H. 2001. Human breast adipocytes express interleukin-6 (IL-6) and its receptor system: increased IL-6 production by
beta-adrenergic activation and effects of
IL-6 on adipocyte function. J. Clin. Endocrinol. Metab. 86:228188
Lagathu C, Bastard J-P, Auclair M,
Maachi M, Capeau J, Caron M. 2003.
Chronic interleukin-6 (IL-6) treatment increased IL-6 secretion and induced insulin resistance in adipocyte: prevention
by rosiglitazone. Biochem. Biophys. Res.
Commun. 311:37279
Rotter V, Nagaev I, Smith U. 2003.
Interleukin-6 (IL-6) induces insulin resistance in 3T3-L1 adipocytes and is, like
IL-8 and tumor necrosis factor-, overexpressed in human fat cells from in-
65.
66.
67.
68.
69.
70.
71.
72.
73.
7 Jan 2005
13:57
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
141
7 Jan 2005
13:57
142
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAFONTAN
105.
106.
107.
108.
109.
110.
111.
112.
113.
114.
7 Jan 2005
13:57
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
115.
116.
117.
118.
119.
120.
121.
122.
123.
124.
125.
126.
127.
128.
129.
130.
131.
132.
133.
143
7 Jan 2005
13:57
144
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAFONTAN
134. Moller DE. 2000. Potential role of TNF in the pathogenesis of insulin resistance
and type 2 diabetes. Trends Endocrinol.
Metab. 11:21217
135. Lofgren P, van Harmelen V, Reynisdottir
S, Naslund E, Ryden M, et al. 2000. Secretion of tumor necrosis factor- shows
a strong relationship to insulin-stimulated
glucose transport in human adipose tissue.
Diabetes 49:68892
136. Ruan H, Lodish HF. 2003. Insulin resistance in adipose tissue: direct and indirect effects on tumor necrosis factor-.
Cytokine Growth Factor Rev. 14:44755
137. Fasshauer M, Paschke R. 2003. Regulation of adipokines and insulin resistance.
Diabetologia 46:1594603
138. Uysal KT, Wiesbrock SM, Marino MW,
Hotamisligil GS. 1997. Protection from
obesity-induced insulin resistance in mice
lacking TNF-a function. Nature 389:610
14
139. Weisberg SP, McCann D, Desai M,
Rosenbaum M, Leibel RL, Ferrante AW
Jr. 2003. Obesity is associated with
macrophage accumulation in adipose tissue. J. Clin. Invest. 112:1796608
140. Xu H, Barnes GT, Yang Q, Tan G, Yang
D, et al. 2003. Chronic inflammation in fat
plays a crucial role in the development of
obesity-related insulin resistance. J. Clin.
Invest. 112:182130
141. Takahashi K, Mizuarai S, Araki H,
Mashiko S, Ishihara A, et al. 2003.
Adiposity elevates plasma MCP-1 levels leading to the increased CD11bpositive monocytes in mice. J. Biol. Chem.
278:4665460
142. Sartipy P, Loskutoff DJ. 2003. Monocyte
chemoattractant protein-1 in obesity and
insulin resistance. Proc. Natl. Acad. Sci.
USA 100:726570
143. Bouloumie A, Marumo T, Lafontan M,
Busse R. 1999. Leptin induces oxidative
stress in human endothelial cells. FASEB
J. 13:123138
144. Yamagishi S-I, Edelstein D, Du XL, Kaneda Y, Guzman M, Brownlee
145.
146.
147.
148.
149.
150.
151.
152.
7 Jan 2005
13:57
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
163.
164.
165.
166.
167.
168.
169.
170.
171.
145
7 Jan 2005
13:57
146
172.
173.
174.
175.
AR
AR232-PA45-05.tex
AR232-PA45-05.sgm
LaTeX2e(2002/01/18)
P1: GCE
LAFONTAN
novel role of leptin. Proc. Natl. Acad. Sci.
USA 96:232732
Shimomura I, Hammer RE, Ikemoto S,
Brown MS, Goldstein JL. 1999. Leptin reverses insulin resistance and diabetes mellitus in mice with congenital lipodystrophy. Nature 401:7376
Petersen KF, Oral EA, Dufour S, Befroy
D, Ariyan C, et al. 2002. Leptin reverses
insulin resistance and hepatic steatosis
in patients with severe lipodystrophy. J.
Clin. Invest. 109:134550
Combs T, Wagner J, Berger J, Doebber T, Zhang B, et al. 2002. Induction
of Acrp30 levels by PPAR agonists: a
mechanism of insulin sensitization. Endocrinology 143:9981007
Moller DE, Berger JP. 2003. Role of
HI-RES-PA45-05-Lafontan.qxd
1/7/05
2:23 PM
Page 1
C-1
Figure 1 Control of human fat cell lipolysis: signal transduction pathways for catecholamines via - and 2-adrenergic receptors, atrial natriuretic peptide via type A receptor
(NPR-A), and insulin. Catecholamines, insulin, and various inhibitory receptors negatively coupled to adenylyl cyclase control cAMP production, whereas atrial and brain natriuretic peptides (ANP and BNP, respectively) control cGMP production. cAMP and cGMP
both contribute to the protein-kinase [PKA and PKG (cGK-I)]-dependent phosphorylation
of HSL and perilipin. Perilipin phosphorylation induces an important physical alteration
of the droplet surface that facilitates the action of HSL on triglyceride lipolysis. HSL phosphorylation promotes its translocation from the cytosol to the surface of the lipid droplet.
Docking of ALBP to HSL favors the efflux of NEFA released by the hydrolysis of triglycerides. PKA and PKG (cGK-I) phosphorylate a number of other substrates that are not
shown in the diagram and can influence the secretion of various adipocyte products such
as leptin, adiponectin, and interleukin-6. Question marks show pathways that are still
hypothetical or the relevance of which has not been fully demonstrated. AC, adenylyl
cyclase; ALBP, adipocyte lipid binding protein; AR, adrenergic receptor; EP3-R, EP3prostaglandin receptor; adenosine-A1-R, type A1 adenosine receptor; NPY-Y1-R, type Y1
neuropeptide Y receptor; GC, guanylyl cyclase; Gi, inhibitory GTP-binding protein; Gs,
stimulatory GTP-binding protein; HSL, hormone-sensitive lipase; IRS, insulin receptor
substrate; PDE-3B, phosphodiesterase 3B; PI3-K, phosphatidylinositol-3-phosphate
kinase; PKA, protein kinase A; PKB, protein kinase B/Akt; PKG (cGK-I), protein kinase
G; NEFA, nonesterified fatty acid; ALBP, adipocyte lipid binding protein; (), inhibition;
(), stimulation.
HI-RES-PA45-05-Lafontan.qxd
C-2
1/7/05
2:23 PM
Page 2
LAFONTAN
Figure 2 Overview of the major functions under the control of products secreted by the
adipocyte. All the hormones and secretions (and abbreviations) are defined in the text.
3 Dec 2004
20:38
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
P1: GCE
LaTeX2e(2002/01/18)
10.1146/annurev.pharmtox.45.120403.095847
INTRODUCTION
The hazards associated with human exposure to degradation products of volatile
anesthetics have long been recognized. The effects of light, air, and heat on chloroform to produce phosgene, along with other degradation products, are well known.
The interaction of volatile anesthetic agents within the anesthetic circuit itself
was also a concern. The most notable interaction was that of trichloroethylene
with soda lime to produce dichloroacetylene. This highly toxic compound produced considerable morbidity and, perhaps, mortality, and it is discussed in this
review because of its historical importance. As these problems were identified and
corrected, concern shifted to the toxicity associated with the metabolism of inhaled anesthetics. Fluoride-induced nephropathy associated with methoxyflurane
and the halothane-associated hepatoxicity were of concern to anesthesiologists.
Recently, however, there has been renewed interest in the formation and toxicity of degradation products of volatile anesthetics, particularly the formation of
0362-1642/05/0210-0147$14.00
147
3 Dec 2004
20:38
148
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
ANDERS
Compound A from sevoflurane and the formation of carbon monoxide from anesthetics with -CHF2 groups, i.e., desflurane, enflurane, and isoflurane.
This review addresses the formation, fate, and animal and human toxicity of
dichloroacetylene (from trichloroethylene); 2-bromo-2-chloro-1,1-difluoroethylene (from halothane); 2-(fluoromethoxy)-1,1,3,3,3-pentafluoro-1-propene or
Compound A (from sevoflurane); and carbon monoxide (from desflurane, isoflurane, and enflurane). All of these degradation products have known or suspected
toxic potential for humans.
3 Dec 2004
20:38
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
149
stabilized by high concentrations of trichloroethylene, which may limit its decomposition during anesthesia.
The metabolic fate of dichloroacetylene has been investigated in experimental
animals. In rats exposed by inhalation to [14C]dichloroacetylene (40 ppm, 1 h), S(1,2-dichlorovinyl)-N-acetyl-L-cysteine 7 (Figure 2) (61.8%), 2,2-dichloroethanol
(12.2%), 2,2-dichloroethyl glucuronide (4.5%), dichloroacetic acid (8.9%), chloroacetic acid (4.7%), and oxalic acid (8.3%) were excreted in the urine over 96 h (8).
The finding that mercapturate 7 is the major metabolite of dichloroacetylene indicates that glutathione-dependent metabolism is the major pathway of metabolism.
The biotransformation and bioactivation of a range of nephrotoxic and cytotoxic
haloalkenes is dependent on glutathione S-conjugate formation and activation of
cysteine S-conjugates by cysteine conjugate -lyase. This pathway includes glutathione transferase-catalyzed glutathione S-conjugate formation, hydrolysis of the
conjugates by -glutamyltransferase and dipeptidases to give the corresponding
cysteine S-conjugates, active uptake of the cysteine S-conjugates by the kidney, and
bioactivation by cytosolic and mitochondrial -lyases. Reviews about the -lyase
pathway have appeared (9, 10).
Dichloroacetylene 2 undergoes bioactivation by the -lyase pathway (Figure 2).
The reaction of dichloroacetylene 2 with glutathione is catalyzed by rat hepatic and renal glutathione S-transferases to give S-(1,2-dichlorovinyl)glutathione 3
(11). The bioactivation mechanism of S-(1,2-dichlorovinyl)glutathione 3 has been
elucidated (12): S-(1,2-dichlorovinyl)glutathione 3 is hydrolyzed by -glutamyltransferase and dipeptidases to give S-(1,2-dichlorovinyl)-L-cysteine 4, which
undergoes bioactivation by renal cysteine conjugate -lyase or detoxication by
N-acetylation to give S-(1,2-dichlorovinyl)-N-acetyl-L-cysteine 7, which is the
major urinary metabolite. S-(1,2-Dichlorovinyl)-L-cysteine 4 undergoes a -lyasecatalyzed -elimination reaction to give 1,2-dichloroethenethiolate 5, pyruvate,
and ammonia. Thiolate 5 may lose chloride to give chlorothioketene 6 or may tautomerize to give chlorothionoacetyl chloride 8. Both thioketene 6 and thionoacyl
chloride 8 may contribute to the toxicity of S-(1,2-dichlorovinyl)-L-cysteine 4, but
the finding that thioketene 6 is highly unstable in aqueous environments favors a
role for thionoacyl chloride 8 (13). The formation of 1,2-dichloroethenethiolate 5
and chlorothioketene 6 has been demonstrated by Fourier-transform ion cyclotron
resonance mass spectrometry (14).
Toxicity
The toxicity of dichloroacetylene and its glutathione and cysteines S-conjugates has
been investigated in experimental animals and in a range of in vitro systems. The
high reactivity of dichloroacetylene has prevented investigation of its cytotoxicity.
Dichloroacetylene is nephrotoxic,
nephrocarcinogenic, neurotoxic, and hepatotoxic in laboratory animals, but nephrotoxicity is the prominent feature of dichloroacetylene-induced toxicity (1518).
3 Dec 2004
20:38
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
ANDERS
150
AR
3 Dec 2004
20:38
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
151
Rabbits exposed to dichloroacetylene show extensive tubular and focal necrosis in the collecting tubules and accompanying clinical-chemical indices of renal
damage, including marked increases in blood urea nitrogen concentrations. As indicated above, the nephrotoxicity of dichloroacetylene is associated with -lyasedependent bioactivation (11). Dichloroacetylene is a potent nephrocarcinogen in
rats and mice: Cystadenomas and adenocarcinomas of the proximal tubules were
observed in all animals exposed to dichloroacetylene (18). The hepatotoxicity of
dichloroacetylene is characterized by fatty degeneration of parenchymal cells, but
only transient elevations in serum transaminases are observed (17).
The neurotoxicity of dichloroacetylene in rabbits is characterized by morphological changes in the sensory and motor trigeminal nuclei and in the facial and
oculomotor nerves and by functional neurological deficits that are manifested
as decreased thermal sensitivity (16). In mice, damage to the Purkinje layer of
the cerebellum is also observed (15). The mechanism of the neurotoxicity of
dichloroacetylene has not been elucidated. Both S-(1,2-dichlorovinyl)glutathione
3 and S-(1,2-dichlorovinyl)-L-cysteine 4 are efficiently taken up by the brain, and
-lyase activity is present in the brain, indicating a possible role for the -lyase
pathway in dichloroacetylene-induced neurotoxicity in rodents (19, 20).
The cytotoxicity of dichloroacetylene itself has apparently not been reported.
The dichloroacetylene-derived conjugates S-(1,2-dichlorovinyl)glutathione and S(1,2-dichlorovinyl)-L-cysteine are, however, cytotoxic in isolated rat renal proximal tubular cells (21). The cytotoxicity of S-(1,2-dichlorovinyl)glutathione is
blocked by the -glutamyltransferase inhibitor acivicin and by the dipeptidase inhibitors 1,10-phenanthroline and phenylalanylglycine, indicating that hydrolysis
of the glutathione S-conjugate to the cysteine S-conjugate is required for toxicity.
The -lyase inhibitor (aminooxy)acetic acid blocks the cytotoxicity of both the
glutathione and cysteine S-conjugates. S-Conjugate-induced mitochondrial dysfunction plays an important role in S-(1,2-dichlorovinyl)-L-cysteine-induced cytotoxicity (22). Similarly, S-(1,2-dichlorovinyl)glutathione and S-(1,2-dichlorovinyl)-L-cysteine are cytotoxic in pig kidney-derived cultured LLC-PK1 cells, and
their cytotoxicity is blocked by (aminooxy)acetic acid (23).
Pure dichloroacetylene is mutagenic in Salmonella typhimurium strain TA100
but not in strain TA98 (24). The glutathione and cysteine S-conjugates of dichloroacetylene are also mutagenic in the Ames test with S. typhimurium strain TA2638
(25). The -lyase inhibitor (aminooxy)acetic acid blocks the mutagenicity of both
S-(1,2-dichlorovinyl)-L-cysteine and S-(1,2-dichlorovinyl)glutathione, indicating
a role for -lyase in S-conjugate-induced mutagenicity. -Glutamyltransferase,
which catalyzes the hydrolysis of glutathione S-conjugates to cysteine S-conjugates,
is present in extracts of S. typhimurium and converts the glutathione S-conjugates
to cysteine S-conjugates, which undergo -lyase-dependent bioactivation. Finally,
S-(1,2-dichlorovinyl)--methyl-DL-cysteine, which cannot undergo bioactivation
by the pyridoxal phosphate-dependent -lyase, is not mutagenic. S-(1,2-Dichlorovinyl)-L-cysteine induces unscheduled DNA synthesis and micronucleus formation in Syrian hamster embryo fibroblasts and expression of c-fos and c-myc in
LLC-PK1 cells (26, 27).
3 Dec 2004
20:38
152
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
ANDERS
Damage to cranial nerves is a distinctive feature of dichloroacetylene poisoning in man and is often associated with symptoms such as skin
irritation, headache, nausea, dizziness, and confusion (28). This unusual pattern of symptoms was described in one of the early reports of toxicity after
trichloroethylene anesthesia (4, 29). Patients given trichloroethylene through sodalime-containing circuits showed neurological symptoms that ranged from mild
trigeminal anesthesia to general encephalitis and death. The most striking feature
in all patients was trigeminal neuropathy, but many patients also showed involvement of other cranial nerves. Although the investigators did not establish the exact
cause of the toxicity, they believed that the most likely cause was dichloroacetylene
formed by a chemical reaction of trichloroethylene with soda lime. Accordingly,
they recommended that soda lime not be used during anesthesia with trichloroethylene. Detailed studies of the conditions required to produce dichloroacetylene from
trichloroethylene in soda lime revealed that not only were the temperature and base
content of the soda lime important, but also its degree of hydration: Only dry soda
lime produced significant amounts of dichloroacetylene (5).
Interestingly, nephrotoxicity, which is a prominent feature of dichloroacetyleneinduced toxicity in rodents, was apparently not observed in human subjects anesthetized with trichloroethylene. Although the evidence strongly indicates that
dichloroacetylene undergoes -lyase-dependent bioactivation in rodents, the failure to observe nephrotoxicity in human subjects may be attributed to the low
-lyase activities present in human kidney tissue (3032). The possible role of lyase-dependent bioactivation in the observed neurotoxicity of dichloroacetylene
merits further investigation.
Serious toxicity after trichloroethylene anesthesia ceased to be a problem once
the cause was identified. Anesthesia circuits that lacked carbon dioxide absorbents
were used to deliver trichloroethylene. Additionally, the base content of absorbents
was reduced and their formulations were changed to minimize the temperature
increase during use so that if they were used in error during trichloroethylene
anesthesia, risks would be minimized. Despite the apparent safety of modern absorbents, they have been implicated in the production of all of the other degradation
products of currently used anesthetics described in this review.
HUMAN TOXICITY
2-BROMO-2-CHLORO-1,1-DIFLUOROETHYLENE
FROM HALOTHANE
Introduction
Halothane was introduced into clinical anesthesia practice in 1956 (33) and soon
became the most commonly used volatile anesthetic because of its lack of flammability and its desirable anesthetic properties. By the early 1960s, however, reports
appeared that its use was occasionally associated with a type of fulminant hepatitis.
Although rare, approximately 1 case in 35,000 administrations, concern about this
3 Dec 2004
20:38
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
153
3 Dec 2004
20:38
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
ANDERS
154
AR
3 Dec 2004
20:38
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
155
anesthetized with halothane (40, 42). Its formation can be rationalized by the
addition of glutathione to 2-bromo-2-chloro-1,1-difluoroethylene 10 to give S-(2bromo-2-chloro-1,1-difluoroethyl)glutathione 11, which may undergo -glutamyltransferase- and dipeptidase-catalyzed hydrolysis to give (2-bromo-2-chloro-1,
1-difluoroethyl)-L-cysteine 12 (Figure 4). N-Acetylation would give the observed
mecapturate S-(2-bromo-2-chloro-1,1-difluoroethyl)-N-acetyl-L-cysteine 17.
Cysteine S-conjugate 12 undergoes -lyase-dependent bioactivation: Glyoxylic
acid 16 was identified as a terminal metabolite of S-(2-bromo-2-chloro-1,1-difluoroethyl)-L-cysteine 12 (43). This was an unexpected finding, because other
bromine-lacking cysteine S-conjugates afford dihaloacetic acids as terminal products. Detailed mechanistic studies showed that 2-chloro--thiolactone 14 may
be an intermediate in the bioactivation of S-(2-bromo-2-chloro-1,1-difluoroethyl)L-cysteine (Figure 4). Subsequent experiments also showed, however, that
3-chloro-2,2-difluorothiirane 15 is also formed as an intermediate in the -lyasecatalyzed bioactivation of S-(2-bromo-2-chloro-1,1-difluoroethyl)-L-cysteine 12
(44) (Figure 4). Hydrolysis of thiirane 15 would give glyoxylic acid 16. Subsequent
computational chemistry studies indicated that a role for 2-chloro--thiolactone 14
is unlikely and that 3-bromo-2,2-difluorothiirane 14 may be more important (45).
The finding that 3-chloro-2,2-difluorothiirane 15 is formed in the biotransformation of cysteine S-conjugate 12 marked the first demonstration of 2,2,
3-trihalothiirane formation, although 2,2,3-trihalothiiranes had earlier been suggested, but not identified, as possible intermediates in the bioactivation of cysteine
S-conjugates (46, 47). 2,2,3-Trihalothiirane formation from cysteine S-conjugates
was later confirmed by Commandeur et al. (48).
Toxicity
The toxicity of 2-bromo-2-chloro-1,1-difluoroethylene and its glutathione and cysteines S-conjugates has been investigated in experimental animals and in vitro
systems.
2-Bromo-2-chloro-1,1-difluoroethylene is nephrotoxic in mice: Its LC50 is approximately 0.025% (v/v) (36). Animals
exposed to 2-bromo-2-chloro-1,1-difluoroethylene show kidney damage characterized by intense renal tubular degeneration. To determine whether the formation
of 2-bromo-2-chloro-1,1-difluoroethylene in the anesthetic circuit might lead to
kidney damage, monkeys were anesthetized with halothane, but no abnormalities
were found on postmortem examination. In dogs anesthetized with halothane, concentrations of 0.00005 to 0.001% (v/v) 2-bromo-2-chloro-1,1-difluoroethylene are
found in the reservoir bag, but no macroscopic or microscopic changes are observed
on postmortem examination.
Detailed studies on the mechanism of 2-bromo-2-chloro-1,1-difluoroethyleneinduced kidney damage have been conducted and were designed to test the hypothesis that 2-bromo-2-chloro-1,1-difluoroethylene 10 undergoes glutathione
3 Dec 2004
20:38
156
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
ANDERS
3 Dec 2004
20:38
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
157
2-(FLUOROMETHOXY)-1,1,3,3,3-PENTAFLUORO-1-PROPENE
(COMPOUND A) FORMATION FROM SEVOFLURANE
[FLUOROMETHYL 1-(TRIFLUOROMETHYL)2,2,2-TRIFLUOROETHYL ETHER]
Introduction
Sevoflurane is a fluorinated volatile anesthetic agent that is approved for use
in over 40 countries, including the United States. Its low blood-gas partition
coefficient allows rapid induction and awakening (56). In addition, sevoflurane
is nonirritating to the airways and is, therefore, useful for inhaled induction.
Although sevoflurane underwent clinical trials in the United States in the 1970s and
3 Dec 2004
20:38
158
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
ANDERS
was considered an excellent anesthetic, two concerns about its potential toxicity
have been raised: First, sevoflurane undergoes metabolism to inorganic fluoride,
which has the potential to induce nephrotoxicity. Second, sevoflurane undergoes
Baralyme- and soda lime-dependent degradation to the fluoroalkene Compound
A, which is nephrotoxic in rats.
3 Dec 2004
20:38
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
159
3 Dec 2004
20:38
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
ANDERS
160
AR
Figure 6 Glutathione S-transferase- and cysteine conjugate -lyase-dependent bioactivation of 2-(fluoromethoxy)-1,1,3,3,3-pentafluoro-1-propene (Compound A) 19.
20, S-[2-(fluoromethoxy)-1,1,3,3,3-pentafluoropropyl]glutathione; 21, S-[2-(fluoromethoxy)-1,3,3,3-tetrafluoro-1-propenyl]glutathione; 22, S-[2-(fluoromethoxy)-1,1,
3,3,3-pentafluoropropyl]-L-cysteine; 23, S-[2-(fluoromethoxy)-1,3,3,3-tetrafluoro-1propenyl]-L-cysteine; 24, 2-(fluoromethoxy)-3,3,3-trifluoropropanoic acid; 25,
trifluorolactic acid. GST, glutathione S-transferase; GSH, glutathione; -GT, glutamyltransferase; -lyase, cysteine conjugate -lyase.
3 Dec 2004
20:38
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
161
Figure 7 N-Acetylation and sulfoxidation of S-[2-(fluoromethoxy)-1,1,3,3,3-pentafluoropropyl]-L-cysteine 22 and S-[2-(fluoromethoxy)-1,3,3,3-tetrafluoro-1-propenyl]-L-cysteine 23. 26, S-[2-(fluoromethoxy)-1,1,3,3,3-pentafluoropropyl]-N-acetylL-cysteine; 27, S-[2-(fluoromethoxy)-1,3,3,3-tetrafluoro-1-propenyl]-N-acetyl-L-cysteine; 28, S-[2-(fluoromethoxy)-1,1,3,3,3-pentafluoropropyl]-N-acetyl-L-cysteine
sulfoxide; 29, S-[2-(fluoromethoxy)-1,3,3,3-tetrafluoro-1-propenyl]-N-acetyl-Lcysteine. NAT, N-acetyltransferase; P450, cytochrome P450.
the presence of the Compound A metabolites S-[2-(fluoromethoxy)-1,1,3,3,3pentafluoropropyl]-N-acetyl-L-cysteine 26, (E)- and (Z)-S-[2-(fluoromethoxy)-1,
3,3,3-tetrafluoro-1-propenyl]-N-acetyl-L-cysteine 27, 2-(fluoromethoxy)-3,3,3trifluoropropanoic acid 24, 3,3,3-trifluorolactic acid 25, and inorganic fluoride,
indicating metabolism of Compound A by the -lyase pathway. Similar results
were found in human subjects anesthetized with sevoflurane under conditions
designed to maximize Compound A formation (75). The inspired Compound A
concentrations were 29 14 ppm (range 1067 ppm). Mercapturates 26 and
27 were identified along with 2-(fluoromethoxy)-3,3,3-trifluoropropanoic acid
24.
In vitro studies on the N-acetylation, N-deacetylation, and -lyase-catalyzed biotransformation of Compound Aderived cysteine S-conjugates and
3 Dec 2004
20:38
162
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
ANDERS
mercapturates by human kidney microsomes and cytosol showed significant interindividual variability (78). In human kidney cytosol, the rates of N-acetylation
of cysteine S-conjugates 22 and 23 were 0.024 0.01 (range 0.0080.045) nmol
mercapturate mg1 min1 and 0.024 0.02 (range 0.0010.07) nmol mercapturate mg1 min1, respectively. Similar results were obtained in human kidney
microsomes: The rates of N-acetylation of cysteine S-conjugates 22 and 23 were
0.025 0.02 (range 0.0050.055) nmol mercapturate mg1 min1 and 0.030
0.02 (range 0.0010.06) nmol mercapturate mg1 min1, respectively. The lyase-catalyzed biotransformation of cysteine S-conjugates 22 and 23 amounted
to 0.051 0.04 (range 0.0040.14) nmol pyruvate mg1 min1 and 0.26 0.08
(range 0.100.40) nmol pyruvate mg1 min1, respectively. The rates of hydrolysis of the mercapturates 26 and 27 were 1.25 0.57 (range 0.82.5) nmol mg1
min1 and 0.17 0.10 (range 0.050.37) nmol mg1 min1, respectively. These
data show that rates of -lyase-catalyzed bioactivation of Compound Aderived
cysteine S-conjugates in human kidney tissue were greater than the rates of Nacetylation of the cysteine S-conjugates and that the rates of N-deacetylation of
Compound Aderived mercapturates were greater than the rates of N-acetylation
of Compound Aderived cysteine S-conjugates. Hence, rates of bioactivation (lyase and N-deacetylation) of cysteine S-conjugates of Compound A exceed rates
of detoxication (N-acetylation) in human kidney tissue.
Recent studies also show that Compound Aderived cysteine S-conjugates and
mercapturates undergo biotransformation to the corresponding sulfoxides (79).
The sulfoxidation of (Z)-S-[2-(fluoromethoxy)-1,3,3,3-tetrafluoro-1-propenyl]-Nacetyl-L-cysteine 26 and S-[2-(fluoromethoxy)-1,1,3,3,3-pentafluoropropyl]-Nacetyl-L-cysteine 27 to give sulfoxides 28 and 29, respectively (Figure 7), is
catalyzed by rat liver microsomal fractions; little sulfoxidation is observed in
renal microsomal fractions. In contrast to the mercapturates, S-[2-(fluoromethoxy)1,1,3,3,3-pentafluoropropyl]-L-cysteine 22 underwent nonenzymatic sulfoxidation. Although both cytochromes P450 and flavin-containing monoxygenases
catalyze sulfoxidation reactions, P450 3A isoforms are the major enzymes responsible for the sulfoxidation of Compound Aderived mercapturates. Finally,
the sulfoxidation of mercapturates 26 and 27 is significantly greater in rat than in
human liver microsomes.
Toxicity
The toxicity of Compound A and its glutathione and cysteine S-conjugates has
been investigated in experimental animals and in vitro systems.
Compound A is nephrotoxic in Wistar
rats exposed by inhalation for 1 h to 700 to 1400 ppm Compound A or for 3 h to 110
to 460 ppm Compound A (60). The LC50s for 1-h exposures of male and female
rats are 1090 ppm and 1050, respectively, and, for 3-h exposures, 420 ppm and 400
ppm, respectively. The toxicity of Compound A is characterized by renal tubular
3 Dec 2004
20:38
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
163
necrosis, increased urine glucose and protein concentrations, and increased blood
urea nitrogen concentrations. Lung congestion and hyperemia are also observed,
but hepatotoxicity was not reported.
A more detailed investigation of the toxicity of Compound A reported a LC50 of
331 ppm for a 3-h exposure in Wistar rats (80). Compound Ainduced nephrotoxicity is characterized by corticomedullary necrosis; significant evidence of corticomedullary necrosis was not seen in 10 rats exposed to 50 ppm Compound A, but
was seen in rats exposed to concentrations of Compound A greater than 100 ppm.
Liver and brain injury, but not lung injury, were observed in some animals. The
effect of exposure time on Compound Ainduced toxicity has also been studied
(81). The LC50s in Wistar rats exposed to Compound A for 6 or 12 h were 203 or
127 ppm, respectively. As in other studies, the nephrotoxicity is characterized by
corticomedullary necrosis. Proliferating cell nuclear antigen, which is an indicator
of cell proliferation, increased with increasing exposure concentration. Lung injury
was seen only at near-lethal concentrations of Compound A. Additional studies on
Compound Ainduced toxicity showed a threshold for nephrotoxicity, as measured
by histopathological examination, of 150 to 200 ppm for a 1-h exposure (82).
The toxicity of Compound A was studied in Sprague-Dawley rats exposed by
nose-only inhalation to 0, 30, 61, 114, or 202 ppm Compound A (83). Increases in
blood urea nitrogen and creatinine concentrations are observed in male and female
rats exposed to 202 ppm Compound A, and renal tubular necrosis is observed in
rats exposed to 114 or 202 ppm Compound A.
The mechanism of Compound Ainduced nephrotoxicity has not been fully resolved, but most evidence implicates the -lyase pathway. A range of 1,
1-difluoroalkenes undergo -lyase-dependent bioactivation, including 2-bromo2-chloro-1,1-difluoroethylene (49, 54), bromotrifluoroethylene (54), chlorotrifluoroethylene (84), 1,1-dichloro-2,2-difluoroethylene (54), hexafluoropropene (85),
and tetrafluoroethylene (86).
Evidence for a role for the -lyase pathway in Compound Ainduced nephrotoxicity has been presented: (Aminooxy)acetic acid, a -lyase inhibitor (12), partially blocks Compound Ainduced nephrotoxicity and reduces the excretion of
2-fluoromethoxy-3,3,3-trifluoropropanoic acid 24 in rats given Compound A (71,
73). Also, the Compound Aderived cysteine S-conjugates S-[2-(fluoromethoxy)1,1,3,3,3-pentafluoropropyl]-L-cysteine 22 and S-[2-(fluoromethoxy)-1,3,3,3tetrafluoro-1-propenyl]-L-cysteine 23 are substrates for rat, human, and nonhuman primate renal -lyase (32). Finally, the Compound Aderived glutathione
S-conjugates S-[2-(fluoromethoxy)-1,1,3,3,3-pentafluoropropyl]glutathione 20
and S-[2-(fluoromethoxy)-1,3,3,3-tetrafluoro-1-propenyl]glutathione 21 and cysteine S-conjugate S-[2-(fluoromethoxy)-1,1,3,3,3-pentafluoropropyl]-L-cysteine
22 and S-[2-(fluoromethoxy)-1,3,3,3-tetrafluoro-1-propenyl]-L-cysteine 23 are
nephrotoxic in rats in vivo, and their nephrotoxicity is partially blocked by (aminooxy) acetic acid (87).
Martin et al. purported to show that the -lyase pathway is not involved in
the mechanism of Compound Ainduced nephrotoxicity (88, 89). These workers
3 Dec 2004
20:38
164
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
ANDERS
reported that the -lyase inhibitor (aminooxy)acetic acid and the -glutamyltransferase inhibitor acivicin either failed to block or increased Compound Ainduced
nephrotoxicity. The interpretation of these studies is limited, however, by the experimental design: In one study (88), only one exposure concentration (150 ppm)
and one exposure time (3 h) were used. In a second study (89), the concentrations
of Compound A used (600 and 800 ppm for 1 h) are greater than one half of
the observed LC50 of Compound A in Wistar rats (60). Moreover, in both studies
nephrotoxicity was assessed only by histopathological studies; no clinical chemical parameters were reported. Subsequent studies showed that (aminooxy)acetic
acid and acivicin also potentiate the nephrotoxicity of Compound Aderived glutathione and cysteine S-conjugates (90). The observation that acivicin potentiates
the toxicity of Compound A has been confirmed by others (91, 92). The mechanism
by which acivicin may increase the nephrotoxicity of Compound A is not apparent.
Acivicin also increases the nephrotoxicity of hexachlorobutadiene in rats (93), but
blocks hexachlorobutadiene-induced nephrotoxicity in mice (94); hexachlorobutadiene undergoes -lyase-dependent bioactivation in rats (95). Lantum et al. (96)
tested the hypothesis that acivicin may reduce renal glutathione concentrations and,
thereby, render the kidney susceptible to injury. It was found, however, that acivicin
significantly increases renal glutathione concentrations. Finally, probenecid blocks
the nephrotoxicity of Compound A, perhaps by preventing the renal uptake of glutathione and cysteine S-conjugates of Compound A (91, 92). Additional studies
are needed to clarify fully the mechanism of Compound Ainduced nephrotoxicity
in rats and, particularly, the effects of acivicin. Nevertheless the weight of the evidence supports a role for the -lyase pathway in the nephrotoxicity of Compound
A in rats.
The cytotoxicity of Compound A and several of its metabolites has been studied
in human-kidney-derived HD-2 cells (97). Compound A was cytotoxic only at concentrations greater than 0.9 mM, which is higher than would be achieved during the
clinical use of sevoflurane. Glutathione S-conjugates 20 and 21 of Compound A
were not cytotoxic in HK-2 cells. The Compound Aderived cysteine S-conjugates
22 and 23 were cytotoxic in HK-2 cells, but were much less cytotoxic than
S-(1,2-dichlorovinyl)-L-cysteine 4. The mercapturate (Z)-S-[2-(fluoromethoxy)1,3,3,3-tetrafluoro-1-propenyl]-N-acetyl-L-cysteine 27 was cytotoxic at the highest
concentration tested (2.7 mM), but S-[2-(fluoromethoxy)-1,1,3,3,3-pentafluoropropyl]-N-acetyl-L-cysteine 26 was not cytotoxic.
The Compound Aderived sulfoxides 28 and 29 (Figure 7) are also cytotoxic in HK-2 cells (97); indeed, these sulfoxides are more cytotoxic than the
corresponding cysteine S-conjugates and mercapturic acids. Several sulfoxides
of haloalkene-derived cysteine S-conjugates or mercapturates are nephrotoxic in
vivo or cytotoxic in vitro, or both, including S-(1,2-dichlorovinyl)-L-cysteine sulfoxide (98), S-(1,2-dichlorovinyl)-N-acetyl-L-cysteine sulfoxide (99), S-(1,2,3,4,
4-pentachlorobutadienyl)-N-acetyl-L-cysteine sulfoxide (100), S-(trichlorovinyl)N-acetyl-L-cysteine sulfoxide (99), and (cis-3-chloro-2-propenyl)-N-acetyl-L-cysteine and (trans-3-chloro-2-propenyl)-N-acetyl-L-cysteine (101). These findings
3 Dec 2004
20:38
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
165
raise the question of whether sulfoxidation of cysteine S-conjugates or mercapturates is also a bioactivation pathway in addition to the -lyase pathway. As
expected, the nephrotoxicity of S-(1,2-dichlorovinyl)-L-cysteine sulfoxide, S-(1,
2-dichlorovinyl)-N-acetyl-L-cysteine sulfoxide, and S-(trichlorovinyl)-N-acetyl-Lcysteine sulfoxide is not blocked by the -lyase inhibitor (aminooxy)acetic acid
(98). Moreover, S-(1,2-dichlorovinyl)-L-cysteine sulfoxide is more cytotoxic in rat
renal distal tubular cells than in proximal tubular cells, but S-(1,2-dichlorovinyl)L-cysteine-induced nephrotoxicity is characterized by necrosis of renal proximal tubular cells rather than distal tubular cells (102). Finally, the -methyl
analogs of S-(1,2-dichlorovinyl)-L-cysteine and S-[2-(fluoromethoxy)-1,1,3,3,3pentafluoropropyl]-L-cysteine are not nephrotoxic in rats (12, 87). These compounds cannot undergo -lyase-catalyzed bioactivation but could presumably
undergo sulfoxidation. Hence, studies on the sulfoxidation of -methyl analogs of
cysteine S-conjugates and the toxicity of the sulfoxides are needed to resolve this
point.
Mutagenicity studies showed that Compound A does not induce reverse mutations, either in the absence or presence of a S-9 activating system, with S. typhimurium strains TA98, TA100, TA1535, and TA1537 or with Escherichia coli
strain WP2uvrA (60). Also, Compound A did not induce chromosome aberrations
or increase the number of micronuclei in bone-marrow cells (60).
The human toxicity of Compound A has been reviewed (103,
104). As discussed above, Compound A is produced when sevoflurane is delivered through absorbent-containing circuits. Concern has been expressed that
sevoflurane-derived Compound A may place patients at increased risk for
Compound Ainduced kidney damage. First, the relatively low safety factor of
Compound A concentrations produced in anesthetic circuits compared with concentrations that produce toxicity in rats is a potential concern. The safety factor is
hard to assess but is estimated to be in the range of 2 to 8, which is far less than
the safety factor of 10 to 100 that is commonly accepted by many toxicologists
as providing an adequate margin of safety (105). Second, evidence of renal injury
in human volunteers anesthetized with sevoflurane has been reported (106, 107).
Other workers have, however, failed to find evidence of renal injury or have observed mild and transient evidence of renal injury in human subjects anesthetized
with sevoflurane compared with reference anesthetics (see, for example, 62, 63,
108113). Also, a retrospective evaluation of the effect of sevoflurane on renal
function in adult surgical patients (1941 patients received sevoflurane and 1495
received control agents) found no evidence that sevoflurane administration is associated with renal injury (114).
There are well-documented scientific reasons why the nephrotoxicity of Compound A in man would be expected to be less than that in rats: If -lyase-dependent
bioactivation is the basis of the observed nephrotoxicity of Compound A, human
kidney tissue has much lower -lyase activities than does rat kidney (3032, 115).
Allometric scaling, which does not account for metabolic differences, indicates
HUMAN TOXICITY
3 Dec 2004
20:38
166
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
ANDERS
that the human threshold for Compound Ainduced nephrotoxicity would be approximately 9000 ppm h1, which is approximately 20 times the highest reported
human exposure (113). To date, it is estimated that more than 120 million human
subjects worldwide have been anesthetized with sevoflurane and no established
cases of Compound Aassociated renal damage have been reported (personal communication, R.D. Ostroff, Abbott Laboratories, Abbott Park, IL).
3 Dec 2004
20:38
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
167
strong bases, such as Amsorb, prevents the formation of carbon monoxide (and
the formation of Compound A from sevoflurane) (123).
The mechanisms by which desflurane and isoflurane are converted to carbon
monoxide have been elucidated (124). Their studies, based on well-established carbene and elimination chemistry, describe a mechanism for the formation of carbon
monoxide from anesthetics bearing a -CHF2 moiety. Base-catalyzed abstraction of
a proton from desflurane 30a or isoflurane 30b (Figure 8) affords the intermediate
carbanions 31a,b. Elimination of halide from the intermediate carbanions gives
difluorocarbene 32 and trifluoroacetaldehyde 34. Hydrolysis of carbene 32 affords
carbon monoxide 33. The hydrolysis of methylene carbenes to carbon monoxide is
known (125). Difluorocarbene 32 was trapped by reaction with -methylstyrene to
give 1,1-difluoro2-methyl-2-phenylcyclopropane, thereby establishing its formation as an intermediate. Moreover, the oxygen atom in carbon monoxide is derived
from water: When [18O]H2O was incorporated into barium hydroxide, [18O]carbon
monoxide was formed. Deuterium substitution in the -CHF2 group of desflurane
and isoflurane resulted in decreased carbon monoxide formation. Although mechanistic studies on the formation of carbon monoxide from enflurane have apparently
not been reported, a pathway similar to that described for desflurane and isoflurane
can be envisioned.
The biological fate of carbon monoxide is largely determined by its elimination
via the respiratory tract, although a small fraction of the body burden of carbon
monoxide is oxidized to carbon dioxide (126, 127).
Toxicity
The toxicity of carbon monoxide has been thoroughly investigated in animal studies and in cases of accidental or intentional human exposure; hence, only brief
comments about human carbon monoxide toxicity are warranted.
Moon et al. reported 28 cases of unexplained elevations of
COHb saturations during anesthesia (116, 117). Eight cases had COHb saturations
greater than 27%, and three cases had saturations of 30% or greater. To determine
the source of the carbon monoxide, gas samples were collected from inside the
soda lime canisters in idle anesthesia machines on 320 occasions. Carbon monoxide
concentrations were less than 20 ppm in 271 of the samples, between 100 and 1000
ppm in 10, and greater than 1000 ppm in 6.
The clinical significance of exposure to carbon monoxide concentrations in this
range and the physical properties and toxicity of carbon monoxide are well known.
The Haldane equation describes quantitatively the competition between oxygen
and carbon monoxide for the same ferrous heme binding sites on hemoglobin:
HUMAN TOXICITY
[Hb(CO)4 ]
[Pco ]
= 245
.
[Hb(O2 )4 ]
[Po2 ]
The constant, 245 at pH 7.4, indicates that if Pco = 1/245PO2 , then blood
will be half-saturated with oxygen and half with carbon monoxide at equilibrium.
3 Dec 2004
20:38
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
ANDERS
168
AR
For a human breathing room air containing only approximately 0.085% (850 ppm)
carbon monoxide at sea level, the COHb saturation will be 50% at equilibrium. This
relationship accounts for the dangerous toxicity of low concentrations of carbon
monoxide. The toxicological effects of carbon monoxide depend on a range of
factors, including preexisting anemia and cardiovascular disease. In general, COHb
3 Dec 2004
20:38
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
169
saturations between 30% and 50% result in tachycardia, hypoxic ECG changes,
headache, weakness, nausea, dizziness, and failing vision. Saturations between
50% and 80% result in coma, convulsions, and death.
CONCLUSIONS
The degradation of anesthetic agents to toxic products has been associated with
both old and modern agents. The common element that leads to the formation of
degradation products is the bases in carbon dioxide absorbents in the anesthetic
circuit. With some agents, e.g., sevoflurane, the degradation of an anesthetic to a
toxic product was discovered during manufacture or early in its clinical use. With
other agents, e.g., desflurane, isoflurane, and enflurane, however, the degradation
products were discovered after many years of clinical use. The elucidation of the
mechanisms of anesthetic degradation also provides a means for minimizing the
formation of toxic degradation products by use of absorbents that lack strong bases,
e.g., Amsorb, rather than Baralyme or soda lime.
ACKNOWLEDGMENTS
Research in the authors laboratory was supported by Abbott Laboratories and
by the National Institute of Environmental Health Sciences grant ES03127. The
author has served as a consultant to Abbott Laboratories.
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Striker C, Goldblatt S, Warm IS, Jackson
DE. 1935. Clinical experiences with the
use of trichloroethylene in the production
of over 300 analgesias and anesthesias.
Curr. Res. Anesth. 14:6871
2. Hewer CL. 1943. Further observations
on trichloroethylene. Proc. R. Soc. Med.
36:46365
3. McAuley J. 1943. Trichloroethylene and
trigeminal anaesthesia. Br. Med. J. 2:713
14
4. Carden S. 1944. Hazards in the use of the
closed-circuit technique for trilene anaesthesia. Br. Med. J. 1:31920
5. Firth JB, Stuckey RE. 1945. Decomposition of trilene in closed circuit anaesthesia.
Lancet 1:81416
6. Ott E, Ottemeyer W, Packendorff K. 1930.
Uber
das Dichlor-acetylen. Chem. Ber.
63:194144
7. Reichert D, Metzler M, Henschler D.
1980. Decomposition of the neuro- and
nephrotoxic compound dichloroacetylene
in the presence of oxygen: separation and identification of novel products. J. Environ. Pathol. Toxicol. 3:525
32
8. Kanhai W, Koob M, Dekant W, Henschler D. 1991. Metabolism of 14C-dichloroethyne in rats. Xenobiotica 21:905
16
9. Dekant W, Vamvakas S, Anders MW.
1994. Formation and fate of nephrotoxic
and cytotoxic glutathione S-conjugates:
cysteine conjugate -lyase pathway. Adv.
Pharmacol. 27:11562
3 Dec 2004
20:38
170
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
ANDERS
10. Anders MW, Dekant W. 1998. Glutathione-dependent bioactivation of haloalkenes. Annu. Rev. Pharmacol. Toxicol.
38:50137
11. Kanhai W, Dekant W, Henschler D.
1989. Metabolism of the nephrotoxin
dichloroacetylene by glutathione conjugation. Chem. Res. Toxicol. 2:5156
12. Elfarra AA, Jakobson I, Anders MW.
1986. Mechanism of S-(1,2-dichlorovinyl)glutathione-induced nephrotoxicity. Biochem. Pharmacol. 35:28388
13. Volkel W, Dekant W. 1998. Chlorothioketene, the ultimate reactive intermediate formed by cysteine conjugate
-lyase-mediated cleavage of the trichloroethene metabolite S-(1,2-Dichlorovinyl)-L-cysteine, forms cytosine adducts
in organic solvents, but not in aqueous solution. Chem. Res. Toxicol. 11:108288
14. Zhang T-L, Wang L, Hashmi M, Anders MW, Thorpe C, Ridge DP. 1995.
Fourier-transform ion cyclotron resonance mass spectrometric evidence for the
formation of -chloroethenethiolates and
thioketenes from chloroalkene-derived,
cytotoxic 4-thiaalkanoates. Chem. Res.
Toxicol. 8:90710
15. Reichert D, Ewald D, Henschler D. 1975.
Generation and inhalation toxicity of
dichloroacetylene. Food Cosmet. Toxicol.
13:51115
16. Reichert D, Liebaldt G, Henschler D.
1976. Neurotoxic effects of dichloroacetylene. Arch. Toxicol. 37:2338
17. Reichert D, Henschler D. 1978. Nephrotoxic and hepatotoxic effects of dichloroacetylene. Food Cosmet. Toxicol. 16:
22735
18. Reichert D, Spengler U, Romen W,
Henschler D. 1984. Carcinogenicity of
dichloroacetylene: an inhalation study.
Carcinogenesis 5:141120
19. Patel NJ, Fullone JS, Anders MW.
1993. Brain uptake of S-(1,2-dichlorovinyl)glutathione and S-(1,2-dichlorovinyl)-L-cysteine, the glutathione and
cysteine S-conjugates of the neuro-
20.
21.
22.
23.
24.
25.
26.
27.
3 Dec 2004
20:38
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
171
3 Dec 2004
20:38
172
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
ANDERS
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
3 Dec 2004
20:38
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
66.
67.
68.
69.
70.
71.
72.
173
73. Spracklin DK, Kharasch ED. 1996. Evidence for metabolism of fluoromethyl 2,
2-difluoro-1-(trifluoromethyl)vinyl ether
(Compound A), a sevoflurane degradation
product, by cysteine conjugate -lyase.
Chem. Res. Toxicol. 9:696702
74. Iyer RA, Anders MW. 1997. Cysteine
conjugate -lyase-dependent biotransformation of the cysteine S-conjugates
of the sevoflurane degradation product
2-(fluoromethoxy)-1,1,3,3,3-pentafluoro1-propene (Compound A). Chem. Res.
Toxicol. 10:81119
75. Kharasch ED, Jubert C. 1999. Compound A uptake and metabolism to
mercapturic acids and 3,3,3-trifluoro-2fluoromethoxypropanoic acid during lowflow sevoflurane anesthesia: biomarkers for exposure, risk assessment, and
interspecies comparison. Anesthesiology
91:126778
76. Uttamsingh V, Iyer RA, Baggs RB,
Anders MW. 1998. Fate and toxicity of
2-(fluoromethoxy)-1,1,3,3,3-pentafluoro1-propene (Compound A)-derived mercapturates in male, Fischer 344 rats.
Anesthesiology 89:117483
77. Iyer RA, Frink EJ Jr, Ebert TJ, Anders MW. 1998. Cysteine conjugate
-lyase-dependent metabolism of Compound A (2-(fluoromethoxy)-1,1,3,3,3pentafluoro-1-propene) in human subjects anesthetized with sevoflurane and in
rats given Compound A. Anesthesiology
88:61118
78. Altuntas TG, Kharasch ED. 2002. Biotransformation of L-cysteine S-conjugates
and N-acetyl-L-cysteine S-conjugates
of the sevoflurane degradation product
fluoromethyl-2,2-difluoro-1-(trifluoromethyl)vinyl ether (Compound A) in
human kidney in vitro: interindividual variability in N-acetylation, N-deacetylation, and -lyase-catalyzed metabolism. Drug Metab. Dispos. 30:148
54
79. Altuntas TG, Park SB, Kharasch ED.
2004. Sulfoxidation of cysteine and
3 Dec 2004
20:38
174
80.
81.
82.
83.
84.
85.
86.
87.
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
ANDERS
mercapturic acid conjugates of the
sevoflurane degradation product fluoromethyl-2,2-difluoro-1-(trifluoromethyl)vinyl ether (compound A). Chem. Res.
Toxicol. 17:43545
Gonsowski CT, Laster MJ, Eger EI II,
Ferrell LD, Kerschmann RL. 1994. Toxicity of compound A in rats. Effect of
a 3-hour administration. Anesthesiology
80:55665
Gonsowski CT, Laster MJ, Eger EI II,
Ferrell LD, Kerschmann RL. 1994. Toxicity of compound A in rats. Effect of increasing duration of administration. Anesthesiology 80:56673
Kandel L, Laster MJ, Eger EI II, Kerschmann RL, Martin J. 1995. Nephrotoxicity in rats undergoing a one-hour exposure to Compound A. Anesth. Analg.
81:55963
Keller KA, Callan C, Prokocimer P,
Delgado-Herrera L, Friedman MB, et al.
1995. Inhalation toxicity study of a
haloalkene degradant of sevoflurane,
Compound A (PIFE), in Sprague-Dawley
rats. Anesthesiology 83:122032
Dohn DR, Leininger JR, Lash LH,
Quebbemann AJ, Anders MW. 1985.
Nephrotoxicity of S-(2-chloro-1,1,2-trifluoroethyl)glutathione and S-(2-chloro1,1,2-trifluoroethyl)-L-cysteine, the glutathione and cysteine conjugates of
chlorotrifluoroethene. J. Pharmacol. Exp.
Ther. 235:85157
Koob M, Dekant W. 1990. Metabolism of
hexafluoropropene. Evidence for bioactivation by glutathione conjugate formation in the kidney. Drug Metab. Dispos.
18:91116
Odum J, Green T. 1984. The metabolism
and nephrotoxicity of tetrafluoroethylene
in the rat. Toxicol. Appl. Pharmacol.
76:30618
Iyer RA, Baggs RB, Anders MW. 1997.
Nephrotoxicity of the glutathione and
cysteine conjugates of the sevoflurane
degradation product 2-(fluoromethoxy)1,1,3,3,3-pentafluoro-1-propene (Com-
88.
89.
90.
91.
92.
93.
94.
95.
96.
3 Dec 2004
20:38
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
107.
108.
109.
110.
111.
112.
113.
114.
175
3 Dec 2004
20:38
176
115.
116.
117.
118.
119.
120.
121.
AR
AR232-PA45-06.tex
AR232-PA45-06.sgm
LaTeX2e(2002/01/18)
P1: GCE
ANDERS
in adult surgical patients. Anesth. Analg.
90:68388
Green T, Odum J, Nash JA, Foster JR.
1990. Perchloroethylene-induced rat kidney tumors: an investigation of the mechanisms involved and their relevance to humans. Toxicol. Appl. Pharmacol. 103:77
89
Moon RE, Ingram C, Brunner EA, Meyer
AF. 1991. Spontaneous generation of carbon monoxide within anesthetic circuits.
Anesthesiology 75:A873
Moon RE, Meyer AF, Scott DL, Fox
E, Millington DS, Norwood DL. 1990.
Intraoperative carbon monoxide toxicity.
Anesthesiology 73:A1049
Fang ZX, Eger EI II. 1994. Source of toxic
CO explained: -CHF2 anesthetic + dry
absorbent. Anesthesia Patient Saf. Found.
Newsl. 9:2530
National Research Council CoMaBEoEP.
1977. Carbon Monoxide. Washington,
DC: Natl. Acad. Sci.
Pankow D. 1996. Carbon monoxide formation due to metabolism of xenobiotics.
In Carbon Monoxide, ed. DG Penney, pp.
2543. Boca Raton: CRC Press
Fang ZX, Eger EI II, Laster MJ, Chortkoff
BS, Kandel L, Ionescu P. 1995. Car-
122.
123.
124.
125.
126.
127.
3 Dec 2004
20:38
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
P1: GCE
LaTeX2e(2002/01/18)
10.1146/annurev.pharmtox.45.120403.100058
INTRODUCTION
Adverse drug reactions (ADRs) are significant health problems that contribute to
patient morbidity and mortality. There are many different types of ADRs, affecting
every organ system in the body. However, drug-induced liver injury is the most
frequent reason for the withdrawal of an approved drug from the market, and
it also accounts for more than 50% of cases of acute liver failure in the United
States today (1). More than 600 drugs have been associated with hepatotoxicity.
The clinical picture is diverse, even for the same drug when given to different
0362-1642/05/0210-0177$14.00
177
3 Dec 2004
20:38
178
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
PARK ET AL.
3 Dec 2004
20:38
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
BIOACTIVATION-INDUCED HEPATOTOXICITY
179
Figure 1 Relationship between drug metabolism and toxicity. Toxicity may accrue
through accumulation of parent drug or, via metabolic activation, through formation
of a chemically reactive metabolite, which, if not detoxified, can effect covalent modification of biological macromolecules. The identity of the target macromolecule and
the functional consequence of its modification will dictate the resulting toxicological
response.
detectable in plasma. Their intracellular formation can be inferred from endogenous trapping reactions or physico-chemical techniques. Their formation may be
modulated by enzyme induction, enzyme inhibition, and gene deletion in animals.
However, none of these experimental procedures is directly applicable to man.
Hence, human exposure to chemically reactive metabolites in the liver is almost
impossible to quantify.
The concept that small organic molecules can undergo bioactivation to electrophiles and free radicals and elicit toxicity by chemical modification of cellular
macromolecules has its basis in chemical carcinogenicity and the pioneering work
of the Millers (2, 3). The application of such concepts to human drug-induced hepatotoxicity was established through the studies of Brodie, Gillette, and Mitchell
(4, 5) on the covalent binding to hepatic proteins of toxic (over) doses of the widely
used analgesic acetaminophen.
However, the relationship between bioactivation and the occurrence of hepatic
injury is not simple. For example, many chemicals undergo bioactivation in the
liver but are not hepatotoxic. The best example is the lack of hepatotoxicity with
3 Dec 2004
20:38
180
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
PARK ET AL.
therapeutic doses of acetaminophen. Tight coupling of bioactivation with bioinactivation may be one reason for this. Many enzymic and nonenzymic pathways
of bioinactivation are present in the liver, which is perhaps the best equipped
of all the organs in the body to deal with toxins. Typical examples of bioinactivation pathways include glutathione conjugation of quinones by glutathione
S-transferases (GSTs) and hydration of arene oxides to dihydrodiols by epoxide
hydrolases. It is only when a reactive metabolite is a poor substrate for such enzymes that it can escape bioinactivation and thereby damage proteins and nucleic
acids.
Moreover, covalent binding per se does not necessarily lead to drug hepatotoxicity. The regioisomer of acetaminophen, 3-hydroxyacetanilide, becomes covalently bound to hepatic proteins in rodents without inducing hepatotoxicity (6).
It is therefore necessary to identify the subset of targets, i.e., covalently modified
macromolecules, that is critical to the toxicological process. Hard electrophiles
generally react with hard nucleophiles, such as functional groups in DNA and
lysine residues in proteins. Soft electrophiles react with soft nucleophiles, which
include cysteine residues in proteins and in glutathione, which has a concentration
of approximately 10 mM in the liver. Free radicals can also react with lipids and initiate lipid peroxidative chain reactions. Unfortunately, there are no simple rules to
predict the target macromolecule(s) for a particular chemically reactive metabolite
or the biological consequences of a particular modification. Furthermore, noncovalent interactions also play a role because covalent binding of hepatotoxins is not
indiscriminate with respect to proteins. Even within a single protein there can be
selective modification of an amino acid side-chain found repeatedly in the primary
structure. Thus, the microenvironment (pKa, hydrophobicity, etc.) of the amino
acid in the tertiary structure appears to be the crucial determinant of selective
binding, and therefore the impact of covalent binding on protein function. The
extent of binding and the biochemical role of the protein will in turn determine
the toxicological insult of drug bioactivation. The resulting pathological consequences will be a balance between the rates of protein damage and the rates of
protein replacement and cellular repair.
It is therefore not surprising that irreversible chemical modification of a protein, which has a profound effect on function, is a mechanism of drug-induced
hepatotoxicity. However, it is also important to note that a number of drugs (e.g.,
penicillins, aspirin, omeprazole) rely on covalent binding to proteins for their efficacy, and thus prevention of covalent binding through chemical modification of
the compound may also inadvertently lead to loss of efficacy. Similarly, endogenous compounds, such as cyclopentenone prostaglandins, are Michael acceptors,
which react with specific cysteine residues in transcription factors to elicit their
physiological effects in cell signaling (7).
The considerable task therefore facing the molecular toxicologist and drug
metabolist is to differentiate between those protein modifications that are critical
for a particular type of drug toxicity (and drug efficacy) and the white noise of
noncritical, background covalent binding.
3 Dec 2004
20:38
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
BIOACTIVATION-INDUCED HEPATOTOXICITY
181
3 Dec 2004
20:38
182
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
PARK ET AL.
certainty. This information is essential if one is to relate global changes in gene expression, proteomics, and metabonomics in a way that can be used by the medicinal
chemist in drug design.
Acetaminophen
Acetaminophen is a major cause of drug-related morbidity and mortality in humans, producing massive hepatic necrosis after a single toxic dose. A similar
3 Dec 2004
20:38
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
BIOACTIVATION-INDUCED HEPATOTOXICITY
183
3 Dec 2004
20:38
184
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
PARK ET AL.
depletion occurs, which is an obligatory step for covalent binding and toxicity
(19). The standard treatment for acetaminophen intoxication is N-acetylcysteine,
which replaces hepatic glutathione and prevents toxicity. N-acetylcysteine is most
beneficial if given within 16 h of the overdose. The early signs of cellular disruption
in isolated hepatocytes can be reversed by a disulphide reductant, dithiothreitol
(20, 21).
The massive chemical stress mediated by an acetaminophen overdose leads to
an immediate adaptive defense response in the hepatocyte. This involves various
mechanisms, including the nuclear translocation of redox-sensitive transcription
factors such as Nrf-2, which sense chemical danger and orchestrate cell defense
(Figure 4). Thus, with respect to acetaminophen, Nrf-2 genes of immediate significance are those involved in glutathione synthesis such as -glutamylcysteine
3 Dec 2004
20:38
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
BIOACTIVATION-INDUCED HEPATOTOXICITY
185
synthetase ( -GCS), GSTs, glucuronyltransferases, and heme oxygenase (22). Importantly, it has been observed that nuclear translocation occurs at nontoxic doses
of acetaminophen and at time-points before overt toxicity is observed. However,
with increasing doses of acetaminophen, there is progressive dislocation of nuclear
translocation, transcription, translation, and protein activity (23) as the rate of drug
bioactivation overwhelms cell defense through the destruction of critical proteins.
Since the initial discovery that covalent
binding of acetaminophen to hepatic proteins was associated with hepatotoxicity, there has been a progression of techniques that have been used to identify the
protein targets. Thus, radiolabeled drugs and Western blotting enable the detection
and quantification of adduct formation, whereas more recently proteomics has
allowed the simultaneous identification of several adducted proteins. The latter
technique offers the possibility of determining the amino acids modified and the
nature of that modification. This in turn allows a molecular rationale for the change
in activity of that protein.
At least 17 liver enzymes that show a loss of activity ex vivo after administration
of a toxic dose of acetaminophen to a rodent species have now been investigated;
these are listed in Table 1. An additional 14 liver enzymes are known to be adducted
by paracetamol in vivo and in vitro but have yet to be shown to be inhibited.
It is notable that modification of proteins can occur in most intracellular compartments of the hepatocyte, e.g., endoplasmic reticulum (ER), cytosol, mitochondria, and plasma membrane, which is an indication of the intracellular mobility of
the reactive metabolite once glutathione is depleted (Figure 5). The loss of hepatocyte viability is likely to be a function of the summation and extent of inhibition
of protein activity. Thus, inhibition of -GCS, glyceraldehydes-3-phosphate dehydrogenase (GAPDH), and Ca2+/Mg2+ ATPase will severely impair hepatocyte
function by uncoupling mitochondria, depleting glutathione and ATP, and disturbing Ca2+ homeostasis, which could lead to the expression of TNF and Fas receptors
on cell membranes. -GCS catalyses the rate-limiting step in glutathione synthesis,
the primary biochemical defense of the hepatocyte against NAPQI. GAPDH,
which, as a component of the glycolytic pathway, contributes to ATP production,
is more than 80% inhibited at 2 h after a toxic dose of acetaminophen. On the basis
of reaction with NAPQI in vitro, inhibition is thought to be due to modification of
a critical cysteine (cys-149) within the active site of the enzyme (24). The loss of
calcium homeostasis is one of the first pathological features of acetaminophen
toxicity. It is clear that as NAPQI diffuses from its site of formation, a number
of enzymes are chemically modifiedusually at cysteine or lysine residuesbut
there is a degree of protein selectivity and variation in amino acid modification:
Acetaminophen appears to react with lysine residues of three intraluminal ER
proteins (25). Presumably, noncovalent interactions and the microenvironment of
amino acid residues determine the precise structure of modified proteins.
The rapid inactivation of several proteins suggests that cellular failure is a consequence of multiple parallel events rather than a simple cascade or signaling
Mouse
Mouse
Mouse
Rat
Rat
Rat
Cytosol
Mitochondria
Cell membrane
32%, 2.5 h
52%, 3 h
Covalent?
?
Covalent
Covalent?
Covalent
Covalent
Noncovalent
?
Covalent
Covalent
Covalent
Covalent
Covalent
Noncovalent?
Covalent
?
?
?
?
Cysteine?g
?
?
?
?
0.71%
4.3%
1.4%
2%
?
?
2.05%
1.53.3
0.2%
?
0.55%
?
?
?
?
Cysteine?f
?
?
Proline
?
?
?
?
Cysteine?e
Abundanced
Modified
amino acid
Measured ex vivo.
The inhibited enzymes protein band was less extensively labeled by a thiol-modifying fluorescent reagent (monobromobimane).
LaTeX2e(2002/01/18)
Taken to be covalent modification (arylation) if radiolabel or immunologically reactive drug moiety coincided with protein on gel electrophoretogram except for the MIF tautomerase
adduct, which was characterized by MALDI-TOF analysis of the adducted protein isolated from mouse liver.
AR232-PA45-07.sgm
a
These enzymes have been reported to be inhibited in animals or hepatocytes exposed to acetaminophen. A number of hepatic enzymes known to be covalently modified by acetaminophen
in vitro or in vivo have yet to be reported to be inhibited by the drug: calreticulin and two protein disulfide isomerases (25); aryl sulfotransferase, carbonic anhydrase III, 2,4-dienoyl-CoA
reductase, glutathione peroxidase, glycine N-methyltransferase, 3-hydroxyanthranilate 3,4-dioxygenase, inorganic pyrophosphatase, protein synthesis initiation factor 4A, sorbitol
dehydrogenase, thioether S-methyltransferase, urate oxidase (not found in primates) (91).
2.5 g/kg
850 mg/kg
40%, 4 h
65%, 6 h
35%, 1 h
35%, 24 h
ca. 50%, ? h
31%, 1 h
83%, 2 h
46%, 6 h
45%, 6 h
72%, 8 h
25%, 2 h
18%, 4 h
50%, 2 h
44%, 3 h
Covalent
Modificationc
AR232-PA45-07.tex
600 mg/kg
400 mg/kg
600 mg/kg
650 mg/kg
800 mg/kg
530 mg/kg
500 mg/kg
500 mg/kg
400 mg/kg
200 mg/kg
400 mg/kg
10 mM
600 mg/kg
3 100 mg/kg
65%, 6 h
Inhibitionb
AR
Rat
Mouse
Mouse
Mouse
Mouse
Mouse
Mouse
Mouse
Mouse
Rat
400 mg/kg
Dose/
concentration
20:38
Mouse
Microsomes
Species
Enzyme
186
Fraction
TABLE 1
3 Dec 2004
P1: GCE
PARK ET AL.
3 Dec 2004
20:38
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
BIOACTIVATION-INDUCED HEPATOTOXICITY
187
Figure 5 Levels of interaction of the chemically reactive metabolite of acetaminophen, NAPQI, with cellular function. The cellular locations of specific proteins
involved in cell defense and cell damage, whose functions are known to be modified
by NAPQI following exposure to acetaminophen, are indicated.
mechanism. It is well established that one of the main events in isolated hepatocytes is overall energy failure (26, 27), which is accompanied by the generation of
megamitochondria that are apparently ATP-depleted and nonfunctional (28). The
execution of hepatocytes involves interplay between hepatocyte damage mediated
by chemical stress and the activation of nonparenchymal cells and the subsequent
release of various mediators. The role of Kupffer cells has been demonstrated by the
fact that mice treated with dichloromethylene diphosphonate (DMDP), which depletes 99% of macrophages from the liver, were protected against acetaminophen
toxicity (29). Indeed, acetaminophen-treated rats have four- to sevenfold more
infiltrating macrophages than resident Kupffer cells (30). Furthermore, neutralization of Fas ligand (31) and TNF (32) affords a degree of protection against the early
3 Dec 2004
20:38
188
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
PARK ET AL.
apoptotic processes and the final overwhelming necrosis, which is the overriding
feature of acetaminophens hepatotoxicity.
Nitric oxide has a dual role in the hepatic response to acetaminophen. Nitric
oxide derived from iNOS contributes to acetaminophen-induced parenchymal cell
injury and to microvascular disturbances, whereas nitric oxide derived from constitutive NOS exerts a protective role in liver microcirculation and thereby minimizes
liver injury. In this context, it is of interest to note that glutathione depletion can
lead to oxidative deactivation of nitric oxide and thus produce hypertension (33).
It has also been suggested that liver blood flow is an important determinant
of toxicity. Consistent with this, it has been demonstrated that alpha-blockers,
which mediate vasodilatation, protect against acetaminophen toxicity even when
given after bioactivation and covalent binding of the drug has occurred (L. Randle,
unpublished data).
Halothane
Halothane is the best-studied drug with respect to immunoallergic hepatitis. A
significant proportion of patients exposed to this inhalation anesthetic develop
asymptomatic rises in transaminases. Fulminant irreversible hepatitis is a rare but
life-threatening phenomenon. Most of the patients recorded in the literature with
immunoallergic hepatitis had more than one exposure (34). Antibodies have been
detected in such patients that recognize autoantigens and neoantigens created by
3 Dec 2004
20:38
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
BIOACTIVATION-INDUCED HEPATOTOXICITY
TABLE 2
189
Drug
Bioactivation
Immune response
Reference
Halothane
Oxidative dehalogenation
(39, 45)
Tienilic acid
Thiophene sulphoxidation
Anti-CYP2C9 IgG
(42)
Dihydralazine
Anti-CYP1A2 IgG
(44)
Sulphamethoxazole
N-hydroxylation
IgG antibodies
Drug and metabolite
T cells
(93)
(94)
Carbamazepine
Arene oxidation
Drug T cell
(46, 95)
Nevirapine
Arene oxidation
Drug T cell
Unpublished data
trifluoroacetylation of hepatic proteins (Figure 6). Preincubation of halothanepretreated, but not of control, rabbit hepatocytes with sera from patients with
halothane-induced fulminant hepatic failure rendered the hepatocytes susceptible
to the cytotoxic effects of normal lymphocytes in vitro (35). It is thus likely that
drug-specific T cells may play a role in the pathogenesis of hepatocyte injury, but
direct evidence for this is lacking.
It is likely that the common chemical trigger for both the mild and severe
forms of hepatocyte injury is drug bioactivation to an acyl halide. Bioactivation
of halothane is substantial and is a consequence of the presence of a vulnerable
proton alpha to halide groups, which are effective leaving groups. In this sense, the
only metabolic route available to the molecule is bioactivation. There is direct and
indirect evidence for this concept. First, the detection of drug metabolite-specific
antibodies in affected patients. Second, a global evaluation of the relationship between the metabolism and toxicity of inhalation anesthetics reveals that the newer,
metabolically inert anesthetics such as enflurane and isoflurane are rarely associated with hepatotoxicity in man. Pohl and colleagues (3639) have identified
a number of target proteins modified by halothane; trifluoroacetylation of lysine
residues is believed to be the principal chemical modification. Precisely how such
chemical modifications trigger an immune response and what is the immunological
mechanism of cell killing is still very much a matter of debate. Animal models of
experimental autoimmune hepatitis indicate that T cells rather than immunoglobulins provide the immunological trigger for cell death (40, 41). A feature of such
animal models is the minimal level of tissue injury, with protection partly afforded
by the presence of T suppressor cells. In man, therefore, the balance between the
different T cell subsets with different functions may be crucial in determining not
only individual susceptibility but also the severity of the injury.
A further clue to the mechanisms involved in such reactions was the discovery
of antibodies directed against the P450 enzymes responsible for the bioactivation
of tienilic acid, dihydralazine, and halothane (4244). In the case of halothane,
3 Dec 2004
20:38
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
PARK ET AL.
190
AR
3 Dec 2004
20:38
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
BIOACTIVATION-INDUCED HEPATOTOXICITY
191
autoantibody generation was extensive (45). Collectively, these data provide evidence for a loss of tolerance to autologous proteins chemically modified as a
consequence of drug bioactivation. Further studies are required to examine patients with immunoallergic hepatitis for drug-specific T cells (46) and for genetic
variants in drug metabolism and immune responsiveness, which might provide the
key to understanding the idiosyncratic nature of such reactions.
Studies from our laboratories have already shown that patients with deranged
liver function as a consequence of taking carbamazepine or nevirapine have circulating T cells that recognize the drug (D.J. Naisbitt, unpublished data). Thus
most of the available information is compatible with the hapten mechanism of
drug-induced immunoallergic toxicity outlined in Figure 7.
3 Dec 2004
20:38
192
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
PARK ET AL.
Isoniazid
Isoniazid (INH) is still the most widely used drug in the treatment of tuberculosis (TB) and has high activity against Mycobacterium tuberculosis, although
resistant strains have emerged. INH is used in combination with drugs such as
rifampicin and pyrazinamide to reduce the chance of inducing resistant strains of
the mycobacterium.
INH causes two major adverse reactions: hepatitis and peripheral neuropathy.
The incidence and severity of the adverse drug reactions are related to dose and
duration of therapy. Toxicity may be delayed by several weeks. A minor asymptomatic increase in liver aminotransferases (less than threefold) is seen in 10%
20% of patients within the first two months of therapy, whereas fatal hepatitis is
seen in less than 1% of patients. Mortality is greater than 10% in patients with
jaundice (47, 48). INH typically produces diffuse massive necrosis or chronic hepatitis. Clinical features resemble acute viral-induced hepatitis. Anorexia, fatigue,
nausea, and vomiting are the usual prodromal features, but jaundice and dark urine
may be the first evidence of injury (49). Combination therapy is a risk factor for
hepatitis, although formal studies evaluating the mechanisms of this have not been
undertaken. Interestingly, of the three anti-TB compounds, it has been suggested
that pyrazinamide is the most hepatotoxic, with a rate of hepatitis three and five
times higher than that of rifampicin and INH, respectively (5052).
Studies in the rat (53) and rabbit (54), along with in vitro studies, indicate that
INH undergoes acetylation to give N-acetylisoniazid. This is then hydrolyzed to
acetylhydrazine, which undergoes bioactivation by P450 enzymes to give an acetyl
radical, a reactive species identified by trapping as a glutathione conjugate (53).
Precisely how such a reactive intermediate induces hepatocyte damage remains
to be elucidated, as do the reasons for the increased incidence of hepatotoxicity
when combination therapy is used. Target proteins have not been identified for the
reactive metabolite formed from INH. To date, there is no convincing clinical or
laboratory evidence to suggest an immunological mechanism.
Interestingly, bioactivation plays a role in the pharmacology of INH, with elimination of nitrogen being the driving force for the formation of an isonicotinoyl
radical intermediate (Figure 8). INH can thus be considered a prodrug, which is
activated by the mycobacterial catalase-peroxidase enzyme KatG. The product
of bioactivation forms a covalent adduct with NADH, which is an inhibitor of
InhA, an enoyl-acyl carrier protein reductase that is involved in the biosynthesis
of mycolic acids present in the mycobacterium cell wall (55, 56).
Diclofenac
The nonsteroidal antiinflammatory drugs (NSAIDs) as a class have a strong
association with hepatotoxicity. Several NSAIDs have been withdrawn after obtaining approval for a license, the most recent being bromfenac (57). The mechanism of hepatotoxicity appears to be complex and multifactorial, involving both
pharmacological and metabolic mechanisms. For example, inhibition of the COX
3 Dec 2004
20:38
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
BIOACTIVATION-INDUCED HEPATOTOXICITY
193
enzymes may lead to a reduction in cytoprotective prostaglandins, whereas bioactivation may occur by both oxidation and conjugation. This metabolic complexity
is illustrated with reference to diclofenac, which undergoes acyl glucuronylation
(58), acyl thiolation (59), and multiple P450-catalyzed oxidations producing two
p-benzoquinoneimines via phenols and an as yet uncharacterized intermediate
possibly an epoxideof mechanism-based inhibition (6062). The relative contributions of these metabolites to protein adduction, cytotoxicity, and hepatotoxicity
in vivo remains to be determined. In isolated rat hepatocytes, although the binding
of diclofenac to protein appears to derive principally from reactions of the acyl
glucuronide, the cytotoxicity has been attributed to products of oxidative pathways (63). However, diclofenac-protein adduct formationand especially on the
cell surfacemight be causally relevant to the expression of immune-mediated
hepatitis.
3 Dec 2004
20:38
194
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
PARK ET AL.
Thiazolidinedione Antidiabetics
Chemically reactive metabolites have also been described for a number of drugs
that cause idiosyncratic hepatotoxicity, but for which no mechanistic studies are
available. An important example is troglitazone, a 2,4-thiazolidinedione, which
was the first of a new class of drugs for type 2 diabetes. Troglitazone was associated with a significant frequency of reversible increases in serum transaminases. Reports of severe and fatal liver injury finally led to the withdrawal of this
3 Dec 2004
20:38
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
BIOACTIVATION-INDUCED HEPATOTOXICITY
195
important new drug (66). Fortunately, it could be replaced by newer and safer
2,4-thiazolidinediones (glitazones): pioglitazone and rosiglitazone (Figure 10).
The mechanism of troglitazone-induced hepatic injury is not known. The drug
undergoes oxidative bioactivation at both the chroman ringwhich is unique
to troglitazoneand the thiazolidinedione ring in rats, forming several reactive
metabolites that are eliminated as thioether and thioester conjugates of glutathione
(67, 68). It is also bioactivated in human hepatocytes (69, 70) and is cytotoxic (71).
An association of troglitazone hepatotoxicity in diabetic patients with a glutathione
S-transferase double null genotype provides indirect evidence for the importance
of bioactivation and bioinactivation in its pathogenesis (72). However, the less
hepatotoxic and cytotoxic glitazonespioglitazone and rosiglitazoneas well as
troglitazone undergo NADPH-dependent covalent binding to human microsomal
protein (73). At present, the toxicological significance of troglitazones metabolic
activation remains an open question; even the relative extents of the glitazones
bioactivation in vitro is unquantified. Finally, it is important to note that the heterogeneous clinical picture of troglitazone hepatotoxicity has prompted the suggestion
that this may be a reflection of interindividual variation in the balance of different
mechanisms of drug toxicity as well as varying patient characteristics (74).
3 Dec 2004
20:38
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
PARK ET AL.
196
AR
3 Dec 2004
20:38
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
BIOACTIVATION-INDUCED HEPATOTOXICITY
197
the ultimate hepatotoxin interferes with signaling, and the sequence of molecular
events that impair cell defense, which ultimately lead to hepatocyte destruction. It
is only when such a mechanistic framework is established that we will be in position
to understand the time-course of the toxicity, the nature of the toxicity, and the
direction that the toxicity takes in a particular patient. It is therefore imperative that
such studies begin at the clinical level, but are then translated into molecular studies
in the laboratory with the design of appropriate in vitro and in vivo model systems
to fully exploit the molecular technology now available in the post genomic era.
ACKNOWLEDGMENT
The authors wish to acknowledge the generous and continuing support of the
Wellcome Trust.
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Lee WM. 2003. Drug-induced hepatotoxicity. New Engl. J. Med. 349:47485
2. Miller E, Miller J. 1947. The presence
and significance of bound aminoazo
dyes in the livers of rats fed p-dimethylaminobenzene. Cancer Res. 7:468
80
3. Miller E, Miller J. 1952. In vivo combinations between carcinogens and tissue constituents and their possible role in carcinogenesis. Cancer Res. 12:54756
4. Brodie BB, Reid WD, Cho AK, Sipes
G, Krishna G, Gillette JR. 1971. Possible mechanism of liver necrosis caused
by aromatic organic compounds. Proc.
Natl. Acad. Sci. USA 68:16064
5. Gillette JR, Mitchell JR, Brodie BB. 1974.
Biochemical mechanisms of drug toxicity.
Annu. Rev. Pharmacol. 14:27188
6. Tirmenstein MA, Nelson SD. 1989. Subcellular binding and effects on calcium
homeostasis produced by acetaminophen
and a nonhepatotoxic regioisomer, 3 hydroxyacetanilide, in mouse liver. J.
Biol. Chem. 264:981419
7. Kawamoto Y, Nakamura Y, Naito Y, Torii
Y, Kumagai T, et al. 2000. Cyclopen-
8.
9.
10.
11.
3 Dec 2004
20:38
198
12.
13.
14.
15.
16.
17.
18.
19.
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
PARK ET AL.
glutathione S-transferases. Proc. Natl.
Acad. Sci. USA 95:527580
Carthew P, Rich KJ, Martin EA, De Matteis F, Lim CK, et al. 1995. DNA damage
as assessed by 32P-postlabelling in three
rat strains exposed to dietary tamoxifen:
the relationship between cell proliferation
and liver tumour formation. Carcinogenesis 16:1299304
Group EBCTC. 1998. Tamoxifen for
early breast cancer: an overview of
the randomised trials. Lancet 351:1451
67
Martin EA, Rich KJ, White IN, Woods
KL, Powles TJ, Smith LL. 1995. 32Ppostlabelled DNA adducts in liver obtained from women treated with tamoxifen. Carcinogenesis 16:165154
Boocock DJ, Maggs JL, White IN, Park
BK. 1999. Alpha-hydroxytamoxifen, a
genotoxic metabolite of tamoxifen in
the rat: identification and quantification
in vivo and in vitro. Carcinogenesis 20:
15360
Boocock DJ, Maggs JL, Brown K,
White INH, Park BK. 2000. Major
inter-species differences in the rates of
O-sulphonation and O-glucuronylation
of alpha-hydroxytamoxifen in vitro: a
metabolic disparity protecting human
liver from the formation of tamoxifenDNA adducts. Carcinogenesis 21:1851
58
Raucy JL, Lasker JM, Lieber CS,
Black M. 1989. Acetaminophen activation by human liver cytochromes P450IIE1 and P450IA2. Arch. Biochem. Biophys. 271:27083
Thummel KE, Lee CA, Kunze KL, Nelson SD, Slattery JT. 1993. Oxidation
of acetaminophen to N-acetyl-p-aminobenzoquinone imine by human CYP3A4.
Biochem. Pharmacol. 45:156369
Davis DC, Potter WZ, Jollow DJ, Mitchell
JR. 1974. Species differences in hepatic
glutathione depletion, covalent binding
and hepatic necrosis after acetaminophen.
Life Sci. 14:2099109
20. Albano E, Rundgren M, Harvison PJ, Nelson SD, Moldeus P. 1985. Mechanisms of
N-acetyl-p-benzoquinone imine cytotoxicity. Mol. Pharmacol. 28:30611
21. Rafeiro E, Barr SG, Harrison JJ, Racz
WJ. 1994. Effects of N-acetylcysteine
and dithiothreitol on glutathione and protein thiol replenishment during acetaminophen-induced toxicity in isolated
mouse hepatocytes. Toxicology 93:209
24
22. Goldring C, Kitteringham N, Elsby R,
Randle L, Clement Y, et al. 2004. Activation of hepatic Nrf2 in vivo by acetaminophen in CD-1 mice. Hepatology
39:126776
23. Kitteringham NR, Powell H, Clement
YN, Dodd CC, Tettey JN, et al. 2000. Hepatocellular response to chemical stress
in CD-1 mice: induction of early genes
and gamma-glutamylcysteine synthetase.
Hepatology 32:32133
24. Dietze EC, Schafer A, Omichinski JG,
Nelson SD. 1997. Inactivation of glyceraldehyde-3-phosphate dehydrogenase by a
reactive metabolite of acetaminophen and
mass spectral characterization of an arylated active site peptide. Chem. Res. Toxicol. 10:1097103
25. Zhou L, McKenzie BA, Eccleston ED Jr,
Srivastava SP, Chen N, et al. 1996. The
covalent binding of [14C]acetaminophen
to mouse hepatic microsomal proteins:
the specific binding to calreticulin and
the two forms of the thiol:protein disulfide oxidoreductases. Chem. Res. Toxicol.
9:117682
26. Andersson BS, Rundgren M, Nelson SD,
Harder S. 1990. N-acetyl-p-benzoquinone
imine-induced changes in the energy
metabolism in hepatocytes. Chem. Biol.
Interact. 75:20111
27. Burcham PC, Harman AW. 1991. Acetaminophen toxicity results in sitespecific mitochondrial damage in isolated
mouse hepatocytes. J. Biol. Chem. 266:
504954
28. Karbowski M, Kurono C, Wozniak M,
3 Dec 2004
20:38
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
BIOACTIVATION-INDUCED HEPATOTOXICITY
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
199
3 Dec 2004
20:38
200
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
PARK ET AL.
46. Naisbitt DJ, Britschgi M, Wong G, Farrell J, Depta JP, et al. 2003. Hypersensitivity reactions to carbamazepine: characterization of the specificity, phenotype,
and cytokine profile of drug-specific T cell
clones. Mol. Pharmacol. 63:73241
47. Mitchell JR, Zimmerman HJ, Ishak KG,
Thorgeirsson UP, Timbrell JA, et al. 1976.
Isoniazid liver injury: clinical spectrum,
pathology, and probable pathogenesis.
Ann. Intern. Med. 84:18192
48. Zimmerman HJ. 1990. Update of hepatotoxicity due to classes of drugs in common
clinical use: non-steroidal drugs, antiinflammatory drugs, antibiotics, antihypertensives, and cardiac and psychotropic
agents. Semin. Liver Dis. 10:32238
49. Zimmerman HJ, Ishak KG. 1995. Morphologic spectrum of drug-induced hepatic disease. Gastroenterol. Clin. North
Am. 24:75986
50. Durand F, Bernuau J, Pessayre D, Samuel
D, Belaiche J, et al. 1995. Deleterious influence of pyrazinamide on the outcome
of patients with fulminant or subfulminant liver failure during antituberculous
treatment including isoniazid. Hepatology 21:92932
51. Ormerod LP, Skinner C, Wales J. 1996.
Hepatotoxicity of antituberculosis drugs.
Thorax 51:11113
52. Schaberg T, Rebhan K, Lode H. 1996.
Risk factors for side-effects of isoniazid,
rifampin and pyrazinamide in patients
hospitalized for pulmonary tuberculosis.
Eur. Respir. J. 9:202630
53. Nelson SD, Mitchell JR, Timbrell JA,
Snodgrass WR, Corcoran GB 3rd. 1976.
Isoniazid and iproniazid: activation of
metabolites to toxic intermediates in man
and rat. Science 193:9013
54. Sarich TC, Zhou T, Adams SP, Bain AI,
Wall RA, Wright JM. 1995. A model
of isoniazid-induced hepatotoxicity in
rabbits. J. Pharmacol. Toxicol. Methods
34:10916
55. Rozwarski DA, Grant GA, Barton DH,
Jacobs WR Jr, Sacchettini JC. 1998. Mod-
56.
57.
58.
59.
60.
61.
62.
63.
3 Dec 2004
20:38
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
BIOACTIVATION-INDUCED HEPATOTOXICITY
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
201
3 Dec 2004
20:38
202
81.
82.
83.
84.
85.
86.
87.
88.
89.
AR
AR232-PA45-07.tex
AR232-PA45-07.sgm
LaTeX2e(2002/01/18)
P1: GCE
PARK ET AL.
prodrugs of L-cysteine. Toxicol. Lett. 132:
18
Senter PD, Al-Abed Y, Metz CN, Benigni F, Mitchell RA, et al. 2002. Inhibition of macrophage migration inhibitory
factor (MIF) tautomerase and biological activities by acetaminophen metabolites. Proc. Natl. Acad. Sci. USA 99:144
49
Pumford NR, Halmes NC, Martin BM,
Cook RJ, Wagner C, Hinson JA. 1997. Covalent binding of acetaminophen to N-10formyltetrahydrofolate dehydrogenase in
mice. J. Pharmacol. Exp. Ther. 280:5015
Bruno MK, Khairallah EA, Cohen SD.
1998. Inhibition of protein phosphatase
activity and changes in protein phosphorylation following acetaminophen exposure in cultured mouse hepatocytes. Toxicol. Appl. Pharmacol. 153:11932
Pierce RH, Franklin CC, Campbell JS,
Tonge RP, Chen W, et al. 2002. Cell
culture model for acetaminophen-induced
hepatocyte death in vivo. Biochem. Pharmacol. 64:41324
Daya S, Anoopkumar-Dukie S. 2000. Acetaminophen inhibits liver trytophan-2,3dioxygenase activity with a concomitant
rise in brain serotonin levels and a reduction in urinary 5-hydroxyindole acetic
acid. Life Sci. 67:23540
Landin JS, Cohen SD, Khairallah EA.
1996. Identification of a 54-kDa mitochondrial acetaminophen-binding protein as aldehyde dehydrogenase. Toxicol.
Appl. Pharmacol. 141:299307
Halmes NC, Hinson JA, Martin BM, Pumford NR. 1996. Glutamate dehydrogenase
covalently binds to a reactive metabolite
of acetaminophen. Chem. Res. Toxicol.
9:54146
Parmar DV, Ahmed G, Khandkar MA,
Katyare SS. 1995. Mitochondrial ATPase:
a target for paracetamol-induced hepatotoxicity. Eur. J. Pharmacol. 293:22529
Tsokos-Kuhn JO, Hughes H, Smith CV,
Mitchell JR. 1988. Alkylation of the
liver plasma membrane and inhibition
90.
91.
92.
93.
94.
95.
96.
97.
98.
3 Dec 2004
20:40
AR
AR232-PA45-08.tex
AR232-PA45-08.sgm
P1: GCE
LaTeX2e(2002/01/18)
10.1146/annurev.pharmtox.45.120403.095950
OVERVIEW
Plant products contain bioactive phytochemicals that are finding increasing importance in foods as NCs and in HPs as medicinal principles. HPs are a very
diverse category of plant products and extracts; for example, they are known as
Abbreviations used in text: ADE, adverse drug event; AUC, area-under-the-plasmaconcentration-time curve; CAM, complementary and alternative medicine; Cmax, maximum concentration; Cmin, minimum concentration; CYP, cytochrome P450; EROD,
7-ethoxyresorufin O-deethylation; GST, glutathione S-transferase; HPs, herbal products;
MIC, minimal inhibitory concentration; NCs, nutraceuticals; NHPs, natural health products; Pgp, P-glycoprotein; PK, pharmacokinetic; P450, cytochrome P450; TM, traditional
medicine; SAI, single-active ingredient; SJW, St. Johns wort; UGT, uridine diphosphoglucuronosyl transferase.
203
3 Dec 2004
20:40
204
AR
FOSTER
AR232-PA45-08.tex
ARNASON
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
BRIGGS
3 Dec 2004
20:40
AR
AR232-PA45-08.tex
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
205
3 Dec 2004
20:40
206
AR
FOSTER
TABLE 1
AR232-PA45-08.tex
ARNASON
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
BRIGGS
Chemotaxonomic
category
Typical interacting
compounds
Monoterpenes
Organosulfur
compounds
Diallyl sulfides
Isothiocyanates
Cruciferous vegetables,
horseradish, radishes
Flavonoids (genistein,
naringenin)
Theaflavins
Catechins
Curcuminoids (curcumin)
Gingerols and diarylheptanoids
Hydroxycoumarins
Phenolics
polyphenols
Furanocoumarins
Acids (cinnamic, ellagic, sinapic)
Indoles
Lignans (sesamol, sesaminol)
Chlorophyll derivatives
(chlorophyllin)
Hypericin, hypericum
Botanical source
Tocopherols
tocotrienols
Triterpenes
Saponins/sapogenins
Glycyrrhetic acid and derivatives
Citrus fruits
Green and yellow vegetables,
fruits, grains and cereals,
soybeans
Ginseng (leaves, roots), soybeans
Licorice root
Gluten, fiber
Protease inhibitors, phytic acid
Phthalides
Whole grains
Soybeans
Celery, parsley, carrots
Others
3 Dec 2004
20:40
AR
AR232-PA45-08.tex
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
207
Labeling Information
Information printed on some labels for products such as Echinacea, SJW, and
valerian root state that the products were standardized. SJW may be standardized to 0.3% hypericin or 4% hyperforin. Some labels mention hypericins. Wide
variation in hyperforin (0.006%2.64%), hypericin (0.008%0.08%), and pseudohypericin (0.014%0.19%) content was observed in tested products (52a). In
these products, the 0.3% hypericin standard was only approached (0.26%) when
total hypericin content was determined. Valerian root labels mention valerenic
acid, valerenic acids, and valeric acid (52b). This is indicative of the confusion created by the industry, as valeric acid is a five-carbon molecule that is
not related to the much larger valerenic acids. Silymarin is considered the active constituent of the milk thistle seed. It is a mixture of several flavonolignans, including silibinin (silybin A and B), isosilybinin, silichristin (silychristin),
and silidianin. Constituent analysis of five milk thistle extract products identified six major constituents: 3.3% taxifolin, 23.6% silychristin, 5.3% silidianin,
20% silybin A, 30.7% silybin B, and 17.3% isosilybin (B.C. Foster, C.E. Drouin,
J.F. Livesey, J.T. Arnason & E. Mills, unpublished findings). The total amounts
of silybin A and B ranged from 45.7 to 61%. The biological effect of each constituent is not known. Spectrophotometric analysis as used by many producers and
investigators for quality control would not provide sufficient information for a
critical comparison of these products.
3 Dec 2004
20:40
208
AR
FOSTER
AR232-PA45-08.tex
ARNASON
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
BRIGGS
Tea
Tablet
Tablet
Tincture
Softgel
Softgel
62
73a
74
29
69
71a
Not detected.
2299
1920
793
ND
ND
ND
NDb
ND
32,423
258
ND
1852
8916
5895
953
5.9 10.19d
54.9 4.54
49.4 6.88
71.9 7.07d
ND
59.0 3.45
52.5 7.73
ND
16.2 2.03
61.2 6.13
85.3 4.93d
65.0 2.01
ND
65.5 1.67
77.1 7.35
52.4 6.42
68.5 0.75
91.4 2.22
49.8 0.75
48.1 1.49
9.3 3.52
41.5 4.03
66.1 0.73
79.9 0.49
64.5 0.38c
3A4
98.2 0.55
60.7 1.23
66.8 1.46
80.0 0.61
91.4 1.17
73.4 3.13
72.6 5.86
2D6
2C19
2C9
AR232-PA45-08.sgm
Standardized to 4% echinacosides.
Tea
27
2572
Echinacoside
AR232-PA45-08.tex
Tea
10B
Cichoric
acid
AR
Form
20:40
TABLE 2 Biomarker analysis and percentage inhibition of human cytochrome P450-mediated metabolism by
aliquots of 1.25 mg/ml extracts from products containing Echinacea (n 6 SD)
3 Dec 2004
LaTeX2e(2002/01/18)
P1: GCE
209
3 Dec 2004
20:40
210
AR
FOSTER
AR232-PA45-08.tex
ARNASON
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
BRIGGS
Stability
Bilia et al. (5960) investigated the stability of 40% and 60% v/v tinctures of
artichoke, SJW, Calendula flower, Milk thistle fruit, and Passionflower. Stability
was related both to the class of flavonoids and water content of the tinctures. Shelf
life at 25 C of the most stable tincture (Passionflower 60% v/v) was approximately
six months, whereas that of the Milk thistle tinctures was approximately three
months. Budzinski et al. (61) examined 21 tinctures and noted that there was
marked variation in the ability of these products to inhibit CYP3A4-mediated
metabolism. Several clinical trials have subsequently been conducted on some
of the products examined in this study. Based on the stability concerns (5960)
3 Dec 2004
20:40
AR
AR232-PA45-08.tex
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
211
SELECTED EXAMPLES
Citrus
The potential for some members of the Rutaceae family, such as grapefruit, lime,
and Seville orange, to interact with 3A4 and Pgp and affect PK has been extensively
examined (2, 48, 10, 6267). The mechanisms involved are understood in part.
Activity was initially attributed to bioflavonoids and then to furanocoumarins. Furanocoumarin derivatives can be characterized into two main groups: angular with
a furan ring attached to 7,8-position or linear with the furan ring at 6,7-position with
methoxy, prenyloxy, and geranlyoxy substitution at 5- and/or 8-position. Some, but
not all are inhibitory (10). In grapefruit, geranyloxy derivatives of furanocoumarins
(psoralens) are thought to be involved in competitive or mechanism-based inhibition (10).
What is generally poorly understood is that furanocoumarins, natural lightactivated toxins that protect plants from herbivores such as insects and microbes,
are also present in plants belonging to the Fabaceae (legumes), Moraceae (fig and
mulberry), and Apiaceae (carrot, celery) families (10). Hot water decoctions or
40% ethanol infusions from Apiaceae: Baizhi (Angelica dahurica and varieties),
Qianghuo (Notopterygium incisum or N. forbesii), Duhuo (Angelica biserrata),
Fangfeng (Saposhnikovia divaricata), Danggui (Angelica sinensis), and Rutaceae:
Zhishi or Zhiqiao (Citrus aurantium) resulted in various degrees of human CYP3A
inhibition as determined by microsomal testosterone 6-hydroxylation (67). The
inhibitory potency was consistent with the abundance of the hydrophobic components for each sample. Some products showed increased inhibition after preincubation, suggesting mechanism-based inhibition. Some formulated prescriptions,
however, showed intense inhibition with their hydrophilic fractions rather than
with their hydrophobic fractions, suggesting that components other than furanocoumarins in herbal prescriptions may also cause CYP3A inhibition. Studies
suggest that furanocoumarins can inhibit or induce a wide range of P450s in
addition to CYP3A4, such as CYP1a1, CYP1A2, Cyp1b1, Cyp2a5, CYP2A6,
CYP2B1, CYP6B1/3, CYP6B4, and CYP6D1 (10).
3 Dec 2004
20:40
212
AR
FOSTER
AR232-PA45-08.tex
ARNASON
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
BRIGGS
Cranberry Juice
Cranberry juice (Vaccinium macrocarpon) is a popular drink that has been used
to reduce or prevent urinary infections. Five reports suggesting changes in the
International Normalized Ratio (INR) values for the Prothrombin Time have been
received by the Committee on Safety of Medicines, indicating an interaction between cranberry juice and warfarin, including one fatal case (68).
Garlic
Garlic (Allium sativum L.) and garlic products generally have been regarded as
safe, but conflicting reports in the literature make it difficult to unequivocally
establish the clinical efficacy and safety of these products either alone or in the
presence of therapeutic products. One case report identified two HIV-infected persons taking garlic or garlic supplements for more than two weeks who developed
severe gastrointestinal toxicity after beginning ritonavir-containing antiretroviral
therapy (400 or 600 mg twice daily) (69). The symptoms, including nausea, vomiting, and diarrhea, resolved with discontinuation of garlic or ritonavir. A recent
study analyzed 24 representative garlic products, including three fresh garlic bulbs
(P.S. Ruddock, M. Liao, B.C. Foster, L. Lawson, J.T. Arnason & J.R. Dillon, unpublished data). Interestingly, within the odorless garlic entities, the range for the
allicin/alliin ratio varied from 0 to 4.7. The major biomarker varied with manufactured product, and the constituent content did not correlate with the in vitro
inhibition of CYP-mediated metabolism. The results were consistent with earlier findings on their inhibitory effect on CYP 2C9 1, 2C9 2, 2C19, 2D6, and
3A-mediated metabolism (45). Extracts from garlic exhibited a similar inhibitory
effect on all 3A isoforms. Chinese and elephant garlic (Allium ampeloprasum) had
a lesser inhibitory effect on 3A7; Chinese garlic extracts also had a lesser effect
on the 3A5 isoform studied. All fresh varieties had a slight inhibitory effect on
2C9 1-mediated metabolism but highly stimulated metabolism of the marker substrate with the 2C9 2 isoform. The extracts had negligible to no effect on 2C19- and
2D6-mediated metabolism. However, all extracts strongly inhibited 3A4-mediated
metabolism. The effects of aqueous extracts from aged garlic capsules and the three
fresh varieties were examined for their ability to interact with human Pgp. Relative to 20 M verapamil as the positive control, the phosphate buffer extracts
of aged, common, and Chinese garlic had moderate levels of product-stimulated
vanadate-sensitive ATPase activity. Elephant garlic was inactive.
The effect of odorless garlic on single-dose pharmacokinetics of ritonavir was
examined in ten healthy volunteers (five male, five female) who received 400 mg
of a single dose of ritonavir either alone or with 10 mg of odorless garlic in a
randomized crossover design (70, 71). Coadministration of garlic decreased the
AUC by 17% (90% CI, 31% to 0%; range 46% to 68%) and decreased peak
plasma concentration of ritonavir by 1% (90% CI, 25% to 31%; range 51% to
136%). Although the trend was toward lower levels of ritonavir, the effect was not
significant. In a longer study, 10 healthy volunteers received 10 doses of saquinavir
at a dosage of 1200 mg 3 times daily with meals for 4 days on study days 14,
3 Dec 2004
20:40
AR
AR232-PA45-08.tex
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
213
2225, and 3639, and they received a total of 41 doses of garlic caplets taken 2
times daily on study days 525 (72). In the presence of garlic, the mean saquinavir
AUC during the 8-h dosing interval decreased by 51%, trough levels at 8 h after dosing decreased by 49%, and the Cmax decreased by 54%. After the 10-day washout
period, the AUC, trough, and Cmax values returned to 60%70% of their values
at baseline. Markowitz et al. (73) reported contradictory findings with no effect on
2D6-mediated metabolism of dextromethorphan and 3A4-mediated metabolism
of alprazolam with no significant differences in pharmacokinetic parameters at
baseline and after garlic extract treatment.
Foster et al. (74) demonstrated that garlic and SJW could have an antagonistic
or synergistic effect on antibiotics, indicating that herbal effects on host drug
disposition mechanisms may also affect response to antibiotics. Ward et al. (75),
using Staphylococcus aureus ATCC 29,213 or Escherichia coli ATCC 25,922 as
the indicator organisms, showed a general increase in the MIC of ampicillin by the
products they studied. There were 13 product-related increases in the MIC and 2
decreases. All garlic products increased the MIC of norfloxacin-sensitive organism
to greater than fourfold above baseline. With Escherichia coli ATCC 25922, the
greatest product-antibiotic interaction was with the ampicillin-sensitive organism.
Garlic, Echinacea, and zinc products all caused large increases in the MIC to
ampicillin over baseline values.
Ginkgo biloba
The effects of Ginkgo biloba leaf extract on the pharmacokinetics of diltiazem
were examined in rats (76). The simultaneous addition of extract to small intestine
and liver microsomes inhibited the formation of the active N-demethyl metabolite by CYP3A in a concentration-dependent manner with an IC (50) of approximately 50 and 182 g/ml, respectively. After a single oral pretreatment with extract
(20 mg/kg), both the rate of formation of the metabolite and total amount of CYP in
intestinal or hepatic microsomes decreased transiently. Pretreatment significantly
decreased the terminal elimination rate constant and increased the mean residence
time after intravenous administration of diltiazem (3 mg/kg). Furthermore, it significantly increased the AUC and absolute bioavailability after oral administration
of 30 mg/kg. These results indicated that the concomitant use of Ginkgo extract
in rats increased the bioavailability of diltiazem by inhibiting both intestinal and
hepatic metabolism, at least in part, via a mechanism-based inhibition for CYP3A.
A study in healthy volunteers phenotyped as CYP2D6 extensive metabolizers
with Ginkgo biloba (77) concluded that the products used in these studies at the
recommended dose was unlikely to significantly alter the disposition of coadministered medications primarily dependent on the CYP2D6 or CYP3A4 pathways
for elimination.
Goldenseal
Goldenseal (Hydrastis canadensis Ranunculaceae), a popular herbal supplement
for gastrointestinal ailments, has been shown to affect CYP3A4-mediated
3 Dec 2004
20:40
214
AR
AR232-PA45-08.tex
FOSTER
ARNASON
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
BRIGGS
metabolism and ATPase activity (27, 61). It contains the alkaloids berberine
and hydrastine, hydrastinine, and canadine. Extracts of goldenseal containing
approximately equal concentrations (approximately 17 mM) of two methylenedioxyphenyl alkaloids, berberine and hydrastine, inhibited with increasing potency
(CYP2C9) diclofenac 4 -hydroxylation, (CYP2D6) bufuralol 1 -hydroxylation,
and (CYP3A4) testosterone 6-hydroxylation activities in human hepatic microsomes (17). The inhibition of testosterone 6-hydroxylation activity was noncompetitive, with an apparent Ki of 0.11% extract. Of the methylenedioxyphenyl
alkaloids, berberine (IC50 = 45 M) was a better inhibitor toward bufuralol
1 -hydroxylation and hydrastine (IC50 approximately 350 M for both isomers),
and was an inhibitor toward diclofenac 4 -hydroxylation. For testosterone 6hydroxylation, berberine was the least inhibitory component (IC50 approximately
400 M). Hydrastine inhibited testosterone 6-hydroxylation with IC50 values
for the (+)- and ()-isomers of 25 and 30 M, respectively. For ()-hydrastine,
an apparent Ki value of 18 M without preincubation and an NADPH-dependent
mechanism-based inhibition with a kinactivation of 0.23 min1 and a KI of approximately 110 M were determined. CYP metabolic-intermediate complex formation
could be demonstrated for both hydrastine isomers. Hydrastine formed a CYP complex with CYP2C9, CYP2D6, and CYP3A4. Coexpression of cytochrome b5 with
the CYP isoforms enhanced the rate but not the extent of complex formation.
The pharmacokinetics of indinavir in 10 healthy volunteers before and after
14 days of treatment with goldenseal root (1140 mg twice daily) were not altered
significantly (78).
Three other herbals, barberry (Berberris vulgaris), Oregon grape (Mahonia
aquifolium), and Goldenthread (Coptis groenlandica), also contain measurable
amounts of these compounds (Table 3) and have a long ethnobotanical record in
TABLE 3 The chemical characterization and percentage inhibition of four berberinecontaining botanicals on cytochrome P450-mediated metabolism of three isozymes
Product
BERa
g/ml
HSb
g/ml
HSNc
g/ml
CDNd
g/ml
CYP
3A4
CYP
2C19
CYP
19
Mahonia aquifolium
124
NDe
ND
27.0
13.6
31.7
ND
135
ND
ND
46.2
24.7
41.3
Hydrastis canadensis
9000
4801
209
46.3
54.0
49.4
60.1
Berberris vulgaris
790
ND
ND
ND
ketoconazole
a
Berberine.
b
c
d
e
Hydrastine.
Hydrastinine.
Canadine.
Not detected.
47.5
25.4
45.5
84.2
40.6
33.9
3 Dec 2004
20:40
AR
AR232-PA45-08.tex
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
215
North America (R. Leduc, I. Scott, R. Marles, J. Dillon, J.T. Arnason & B.C.
Foster, unpublished findings). The relative amounts of alkaloids varied greatly in
the four species tested, with H. canadensis having considerably higher concentrations of the marker compounds. Extracts of H. canadensis were inhibitory to CYP
3A4, 2C19 and 2C19. C. groenlandica and B. vulgaris were less active, whereas
M. aquifolium extracts were the least inhibitory. These botanicals have the potential
to affect human drug and intermediary metabolism independent of the concentration of the major biomarkers for these botanicals.
Herbal Teas
Herbal and black teas were analyzed for their capacity to inhibit in vitro metabolism
of drug marker substrates by human CYP isoforms (20). Aliquots and infusions of
all products inhibited 3A4 metabolism. Of the aliquots from teas tested with 2C9,
2C19, and 2D6, many demonstrated inhibitory activity. Black teas and herbal tea
mixtures were generally more inhibitory than single-entity herbal teas. Maliakal &
Wanwimolruk (41) investigated the effect of herbal teas (peppermint, chamomile,
and dandelion) on the activity of hepatic Phase I and II metabolizing enzymes using
female rat liver microsomes. After four weeks of pretreatment, CYP isoforms and
Phase II enzyme activities were determined by incubation of liver microsomes or
cytosol with appropriate substrates. Activity of CYP1A2 in the liver microsomes of
rats receiving dandelion, peppermint, or chamomile tea was significantly decreased
(P < 0.05) to 15%, 24%, and 39% of the control value, respectively. CYP1A2
activity was significantly increased by pretreatment with caffeine solution. No
alterations were observed in the activities of CYP2D and CYP3A in any group of
pretreated rats. Activity of CYP2E in rats receiving dandelion or peppermint tea
was significantly lower than in the control group, 48% and 60% of the control,
respectively. There was a dramatic increase (244% of control) in the activity of
UGT in the dandelion teapretreated group. There was no change in the activity of
GST. The results suggested that certain herbal teas can cause modulation of Phase
I and II drug metabolizing enzymes.
Kava
Inhibition of CYP enzymes by kava extract was investigated (15, 79). Whole
kava extract (normalized to 100 M total kavalactones) caused concentrationdependent decreases in P450 activities, with significant inhibition of the activities
of CYP1A2 (56% inhibition), 2C9 (92%), 2C19 (86%), 2D6 (73%), 3A4 (78%),
and 4A9/11 (65%) following preincubation for 15 min; CYP2A6, 2C8, and 2E1
activities were unaffected. These data indicate that kava has a high potential for
causing drug interactions through inhibition of P450 enzymes.
Milk Thistle
Venkataramanan et al. (40) evaluated the effect of silymarin on the activity of
hepatic drug-metabolizing enzymes in human hepatocyte cultures. Treatment with
3 Dec 2004
20:40
216
AR
AR232-PA45-08.tex
FOSTER
ARNASON
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
BRIGGS
silymarin significantly reduced the activity of CYP3A4 enzyme (by 50% and
100%, respectively) as determined by the formation of 6--hydroxy testosterone
and the activity of UGT1A6/9 by 65% and 100%, respectively, as measured by
the formation of 4-methylumbelliferone glucuronide. Silymarin also significantly
decreased mitochondrial respiration in human hepatocytes.
At least three studies have been conducted with products containing milk thistle
(Silybum marianum) extract to characterize the pharmacokinetics of indinavir in
healthy subjects. Piscitelli et al. (80) and DiCenzo et al. (81) concluded that these
silymarin products had no apparent effect on indinavir plasma concentrations. In
a third study with 16 subjects lasting 28 days by Mills et al. (83), the AUC0-8
indinavir was reduced by a mean 4.4% (90% CI, 26% to 27.5%, P = 0.6) from
Phase I to Phase II in the active group, rebounding to a Phase III reduction of 17.3%
(90% CI, 9% to 37.3%, P = 0.6) of baseline. Control AUC0-8 reduced by 21.5%
(90% CI, 8 to 43%, P = 0.2) from Phase I to Phase II and rebounded to a further
reduction at Phase III of 38.5% (90% CI, 15.3% to 55.3%, P < 0.01) of baseline.
This study has important implications for the conduct and design of herb-drug
interaction trials. The significant decline of AUC0-8 in the control group indicates
that factors other than the exposure of interest may affect drug metabolism.
3 Dec 2004
20:40
AR
AR232-PA45-08.tex
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
217
the single-dose treatment phase, SJW caused a significant 35% decrease (P < 0.05)
in maximum plasma concentration and a significant 47% increase (P < 0.05) in
fexofenadine oral clearance.
The effect of SJW on CYP activity was examined with a probe drug cocktail (91). Twelve healthy subjects (five female, seven male) completed this threeperiod, open-label, fixed-schedule study. Tolbutamide (CYP2C9), caffeine
(CYP1A2), dextromethorphan (CYP2D6), oral midazolam (intestinal wall and
hepatic CYP3A), and intravenous midazolam (hepatic CYP3A) were administered
before short-term SJW dosing (900 mg), and after two weeks of intake (300 mg
tid) to determine CYP activities. Short-term administration of SJW had no effect on CYP activities. Fourteen-day administration caused a significant (P < .05)
increase in oral clearance of midazolam from 121.8 70.7 to 254.5 127.8 and a
corresponding significant decline in oral bioavailability from 0.28 0.15 to 0.17
0.06. In contrast to the >50% decrease in the AUC when midazolam was administered orally, 14-day administration caused a 20% decrease in AUC when
midazolam was given intravenously. Fourteen-day SJW administration resulted
in a significant and selective induction of CYP3A activity in the intestinal wall.
SJW did not alter the CYP2C9, CYP1A2, or CYP2D6 activities in these healthy
subjects.
Spices
Ground fancy clove (Sri Lanka), ground ginger (China or India), oregano leaf
(Turkey), ground sage (Turkey), thyme leaf (Spain), and ground turmeric (India) extracts were found to inhibit CYP2C9, CYP2C19, CYP2D6, and CYP3A4mediated metabolism (20).
3 Dec 2004
20:40
218
AR
FOSTER
AR232-PA45-08.tex
ARNASON
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
BRIGGS
Ueng et al. (12) examined the effects of methanol and aqueous extracts of
Evodia rutaecarpa on CYP, UGT, and GST in C57BL/6J mice. Methanol extract
caused a dose-dependent increase of liver microsomal EROD activity. In liver,
methanol extract increased benzo(a)pyrene hydroxylation, 7-methoxyresorufin Odemethylation (MROD), 7-ethoxycoumarin O-deethylation (ECOD), benzphetamine N-demethylation, and N-nitrosodimethylamine N-demethylation activities.
Aqueous extract increased EROD, O-demethylation, and O-deethylation activities
68%, twofold, and 83%, respectively. For conjugation activities, methanol extract
elevated UGT and GST activities. Aqueous extract elevated UGT activity without
affecting GST activity. Immunoblot analyses showed that methanol extract increased the levels of CYP1A1, CYP1A2, CYP2B-, and GSTYb-immunoreactive
proteins. Aqueous extract increased CYP1A2 protein level. In kidney, neither extract had any effect on most activities. Rutaecarpine, evodiamine, and dehydroevodiamine contributed, at least in part, to the increase of hepatic EROD activity.
Ge-gen, the root of a wild leguminous creeper, Pueraria lobata (Willd.) Ohwi
(13), possesses a high content of flavonoid derivatives, the most abundant of which
is puerarin. Puerarin and Ge-gen crude extracts inhibited the steady-state chemiluminescent reaction in a dose-dependent fashion. Although both CYP content and
NADPH-(CYP)-c-reductase activity were significantly increased in all situations,
a complex pattern of CYP modulation was observed, including both induction
(puerarin: CYP2A1, 1A1/2, 3A1, 2C11; Ge-gen: CYP1A2, 3A1, 2B1) and inactivation (Ge-gen and puerarin: CYP3A, 2E1, 2B1). The latter are due to either
parental agents or metabolites, as demonstrated by in vitro studies. Ge-gen contains compounds with potent antioxidant activity, which impair CYP-catalyzed
drug metabolism.
Ohnishi et al. (96) examined the possibility of pharmacokinetic interactions
between Sho-saiko-to extract powder, a widely used traditional Japanese herbal
(Kampo) medicine and carbamazepine in rats. Sho-saiko-to inhibited hepatic microsome 10,11-epoxylase activity in a concentration-dependent manner. Liver
weight, amounts of CYP and cytochrome b(5) in hepatic microsomes, and the
formation of the 10,11-epoxide by microsomes were not influenced by two-week
repeated oral pretreatment, although pretreatment with phenobarbital (80 mg/kg/d,
i.p.) significantly increased these parameters. Simultaneous oral administration of
Sho-saiko-to significantly decreased Cmax of carbamazepine and the AUC of the
epoxide and lengthened the time to reach Cmax. Two-week repeated oral pretreatment with Sho-saiko-to, however, did not affect the plasma concentration-time
profile or any pharmacokinetic parameter of carbamazepine. A single oral administration of Sho-saiko-to (1 g/kg) significantly delayed gastric emptying and simultaneous oral administration of TJ-9 with carbamazepine CBZ to rats decreased the
gastrointestinal absorption of carbamazepine, at least in part, by delaying gastric
emptying without affecting the metabolism.
Ueng et al. (97) examined the effects of Wu-chu-yu-tang on hepatic and renal
CYP, UGT, and GST in C57BL/6J mice. Treatment of mice with 5 g/kg per day Wuchu-yu-tang for 3 days caused 2.5-fold and 2.9-fold increases of liver microsomal
3 Dec 2004
20:40
AR
AR232-PA45-08.tex
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
219
EROD and 7-methoxyresorufin O-demethylation activities, respectively. CYP activities toward 7-ethoxycoumarin, benzphetamine, N-nitrosodimethylamine, erythromycin, and nifedipine, and conjugation activities of UGT and GST were not
affected. In kidney, Wu-chu-yu-tang treatment had no effects on CYP, UGT, and
GST activities. Among the four component herbs of Wu-chu-yu-tang, only Evodiae fructus (Wu-chu-yu) extract increased EROD activity and CYP1A2 protein
level. The main active alkaloids in E. fructus are rutaecarpine, evodiamine, and
dehydroevodiamine. At doses corresponding to their contents in Wu-chu-yu-tang,
rutaecarpine treatment increased hepatic EROD activity, whereas evodiamine and
dehydroevodiamine had no effects. These results demonstrate that ingestion of
Wu-chu-yu-tang increases mouse hepatic Cyp1a2 activity and protein level.
3 Dec 2004
20:40
220
AR
FOSTER
AR232-PA45-08.tex
ARNASON
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
BRIGGS
3 Dec 2004
20:40
AR
AR232-PA45-08.tex
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
221
benefits and the limitations of in vitro and clinical PK studies to determine what
risk, if any, may be associated with their combined drug and HP exposure.
ACKNOWLEDGMENTS
The Ontario HIV Treatment Network and Health Canada supported our work cited
in this article.
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Chan K. 2003. Some aspects of toxic contaminants in herbal medicines. Chemosphere 52:136171
2. Ameer B, Weintraub RA. 1997. Drug
interactions with grapefruit juice. Clin.
Pharmacokinet. 33:10321
3. Barnes J, Anderson LA, Phillipson JD.
2001. St. Johns wort (Hypericum perforatum L.): a review of its chemistry,
pharmacology and clinical properties. J.
Pharm. Pharmacol. 53:583600
4. Harris R, Jang G, Tsunoda S. 2003. Dietary effects on drug metabolism and
transport. Clin. Pharmacokinet. 42:1071
88
5. Huang SM, Hall SD, Watkins P, Love LA,
Serabjit-Singh C, et al. 2004. Drug interactions with herbal products and grapefruit juice: a conference report. Clin.
Pharmacol. Ther. 75:112
6. Ioannides C. 2002. Pharmacokinetic interactions between herbal remedies and
medicinal drugs. Xenobiotica 32:45178
7. Izzo AA. 2004. Drug interactions with St.
Johns Wort (Hypericum perforatum): a
review of the clinical evidence. Int. J. Clin.
Pharmacol. Ther. 42:13948
8. Zhou S, Gao Y, Jiang W, Huang M, Xu
A, Paxton JW. 2003. Interactions of herbs
with cytochrome P450. Drug Metab. Rev.
35:3598
9. Sun M, Sakakibara H, Ashida H, Danno
G, Kanazawa K. 2000. Cytochrome
P4501A1-inhibitory action of antimuta-
10.
11.
12.
13.
14.
15.
16.
3 Dec 2004
20:40
222
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
AR
FOSTER
AR232-PA45-08.tex
ARNASON
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
BRIGGS
27.
28.
29.
30.
31.
32.
33.
34.
35.
3 Dec 2004
20:40
AR
AR232-PA45-08.tex
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
36.
37.
38.
39.
40.
41.
42.
43.
44.
Vandenhoek S, Krantis A, et al. 2001b. Effect of St. Johns wort on cytochrome P450 and flavin monooxygenase enzymes
and on P-glycoprotein. Can. J. Infect. Dis.
12(Suppl. B):132P
van Zanden JJ, Ben Hamman O, van
Iersel ML, Boeren S, Cnubben NH, et al.
2003. Inhibition of human glutathione
S-transferase P1-1 by the flavonoid
quercetin. Chem. Biol. Interact. 145:139
48
Ferreira T, Drouin CE, Krantis A, Arnason JT, Foster BC. 2003. Effect of a natural health product containing echinacea
on human N-acetyltransferase expression.
Can. J. Infect. Dis. 14(Suppl. A):20A
Lin RD, Hou WC, Yen KY, Lee MH.
2003. Inhibition of monoamine oxidase B
(MAO-B) by Chinese herbal medicines.
Phytomedicine 10:65056
Wentworth JM, Agostini M, Love J,
Schwabe JW, Chatterjee VK. 2000. St
Johns wort, a herbal antidepressant, activates the steroid X receptor. J. Endocrinol.
166:R1116
Venkataramanan R, Ramachandran V,
Komoroski BJ, Zhang S, Schiff PL, Strom
SC. 2000. Milk thistle, a herbal supplement, decreases the activity of CYP3A4
and uridine diphosphoglucuronosyl transferase in human hepatocyte cultures. Drug
Metab. Dispos. 28:127073
Koh HL, Woo SO. 2000. Chinese proprietary medicine in Singapore: regulatory
control of toxic heavy metals and undeclared drugs. Drug Saf. 23:35162
Awang DV. 1991. Maternal use of ginseng and neonatal androgenization. J. Am.
Med. Assoc. 266:363
Binns SE, Livesey JF, Arnason JT, Baum
BR. 2002. Phytochemical variation in
echinacea from roots and flowerheads of
wild and cultivated populations. J. Agric.
Food Chem. 50(13):367387
Saez F. 2001. Volatile oil variability
in Thymus serpylloides ssp. gadorensis growing wild in Southeastern Spain.
Biochem. Syst. Ecol. 29:18998
223
3 Dec 2004
20:40
224
53.
54.
55.
56.
57.
58.
59.
60.
61.
AR
FOSTER
AR232-PA45-08.tex
ARNASON
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
BRIGGS
3 Dec 2004
20:40
AR
AR232-PA45-08.tex
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
72.
73.
74.
75.
76.
77.
78.
79.
225
3 Dec 2004
20:40
226
90.
91.
92.
93.
94.
95.
AR
FOSTER
AR232-PA45-08.tex
ARNASON
AR232-PA45-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
BRIGGS
96.
97.
98.
99.
100.
3 Dec 2004
20:41
AR
AR232-PA45-09.tex
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
10.1146/annurev.pharmtox.45.120403.095758
Multiple Domains
Peter R. Bieck1 and William Z. Potter
1
Eli Lilly & Company, Neuroscience Therapeutic Area, Lilly Corporate Center,
Indianapolis, Indiana 46285; email: bieck@lilly.com
INTRODUCTION
The drug development process has multiple phases and decision points that require
development of stage-specific biomarkers (1, 2). Biomarker assays have to be
validated for criteria such as reference range, accuracy (sensitivity, selectivity,
specificity), and stability (3, 4).
There is general agreement that the main utilities of biomarkers in drug development are the following:
0362-1642/05/0210-0227$14.00
227
3 Dec 2004
20:41
228
AR
BIECK
AR232-PA45-09.tex
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
POTTER
3 Dec 2004
20:41
AR
AR232-PA45-09.tex
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
229
Biomarker procedures
Brain imaging technique
Cell-based imaging
Electrophysiological marker
Electroencephalogram (EEG),
pupillometry, saccadic eye movements
Laboratory-based markerc
Concentrations of catecholamines,
hormones, enzymes, proteins, drugs, and
drug metabolites
Psycho-immunological marker
Neuroendocrine marker
Genetic markers
Proteomic identification
Reference 5.
b
c
Reference 6.
Matrices mostly from plasma, urine, CSF, tissue, saliva, and hair.
Altman has reviewed the challenges regarding the integration and analysis of
genomic, molecular, cellular, and clinical data and has merged at Stanford University what was called Clinical Informatics and Bioinformatics to Biomedical
Informatics (9).
The publicly available internet research tool Pharmacogenetics and Pharmacogenomics Knowledge Base (PharmGKB at http://www.pharmgkb.org/)
3 Dec 2004
20:41
230
AR
BIECK
AR232-PA45-09.tex
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
POTTER
demonstrates how a database can help researchers in understanding the contribution of genetic variation among individuals to differences in reactions to drugs. It is
an integrated resource and can be researched for drugs, diseases, clinical outcome,
pharmacodynamics, pharmacokinetics, molecular/cellular functional assays, and
for genotype (10). As more and more measures of CNS during drug action become
available, integration of genetic, proteomic, metabolic, and pharmacokinetic data
can utilize these tools that are being developed.
Type of study
Example
Phenomenology
Physiology/Pharmacology
Biochemical Pharmacology,
Anatomy, Histology
Molecular biology
Analogy
3 Dec 2004
20:41
AR
AR232-PA45-09.tex
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
231
date, the underlying physiological processes regulating a drugs access to brain are
generally ignored in a still empirical approach with focus on the phenomenological
level.
Cerebrospinal fluid (CSF) sampling is a tool that allows access to the central
compartment, and ideally it would provide a better matrix than plasma to assess
drug concentration close to the site of action. Early studies showed that sporadic
simultaneous measurement of the CSF to plasma concentration ratio usually is
inadequate to describe CSF penetration. CSF concentration-time curves lag behind
those in plasma. The areas under the concentration-time curves in CSF and plasma
at steady state or after a short-term infusion are accepted as measures of CSF
passage (13). Continuous CSF collections for 12 h or longer provide concentrationtime curves for drug and or biomarkers and serve as dynabridge study (14).
This term was introduced to describe a type of exploratory study in which drug
concentrations and activities in the central compartment of patients are measured.
The techniques have been described as safe and well tolerated (15). The question
remains concerning the extent to which CSF concentration is representative of that
at the site of action.
One approach to systematically categorize intercellular communication in the
brain is based on the concepts of wiring transmission (WT) and volume transmission (VT) (16). Morphological and functional observations suggest that CSF might
represent an important vector for convection of VT signals, especially to peri- and
paraventricular areas. Any brain cell can participate in VT, and any kind of substance, such as ions, drugs, classical transmitters, peptides, and neurosteroids, can
be a signal (17, 18). MRI studies have indicated the existence of fluid movements
from the CSF via the paravascular space and the extracellular space into the brain
capillaries (19). Despite these supportive factors, it still remains to be shown when
and if CSF concentrations reflect target drug concentrations.
Direct measures of drug concentrations in human brain tissue can be obtained
in vivo using imaging techniques (20) or measured in postmortem brain (21, 22).
Neurosurgeons apply microdialysis and voltammetry/spectrophotometry for continuous monitoring of substrates, metabolites, or neurotransmitters in the human
brain with the disadvantage that these probing methods are invasive and focal (23).
Ahmed et al. have previously summarized the advantages and limitations of
selected biomarker technologies for assessing CNS access (Table 3) (24). The
main difference between the use of CSF and imaging consists in the ability for
continuous monitoring versus the intermittent snap-shot imaging. We later discuss
examples comparing studies utilizing these different methods.
3 Dec 2004
20:41
AR
232
BIECK
AR232-PA45-09.tex
TABLE 3
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
POTTER
Advantages and limitations of selected biomarker technologiesa
Technology
Advantages
Limitations
CSFb,c
Measurement of drug PK in
central compartment
Several surrogate markers
available
Possible to combine PK and
PD measures in one
protocol
Invasive
Sensitive assay required
EEGc
MRI
b,c
MRS
c
fMRI
c
sMRI
Pharmacological doses of
certain drugs can be
accurately measured
Structural and functional
modalities can be combined
to enhance overall signal
detection
PET/SPECT
b
Tracer techniques
c
Functional methods
c
Occupancy studies
Exposure to ionizing
radiation
Tracer not available for
every application
Expensive
Reproduced with permission from the American Journal of Geriatric Psychiatry (Copyright 2002). American
Psychiatric Publishing, Inc. Reference 24.
b
c
Method can show central PD effect of a drug and/or serve as a surrogate marker in selected situations.
PK: pharmacokinetic(s); PD: pharmacodynamic(s); CSF: cerebrospinal fluid; EEG: electroencephalography; MRI: magnetic resonance imaging; MRS: magnetic resonance spectroscopy; fMRI: functional MRI;
sMRI: structural MRI; PET: positron emission tomography; SPECT: single photon emission computed tomography.
Because this technology is generally limited to robust blockade and displacement of antagonist binding, other measures are required to establish the interaction of a drug with a target. These fall into the general class of functional
measures, the most generalizable and promising of which may prove to be
proteomics.
Proteomics is a research field aiming to characterize molecular and cellular
dynamics in protein expression and function on a global level (25, 26). Clinical
proteomics is a new subdiscipline that involves the application of these technologies at the bedside. Most advanced is the analysis of serum proteomic patterns
3 Dec 2004
20:41
AR
AR232-PA45-09.tex
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
233
to provide diagnostic end points for cancer detection (27). Proteomic approaches
to CNS disorders are progressing with the hope of establishing a CNS proteome
database derived from primary human tissues (28). Areas of research are encompassing proteomes of nerve cells (29, 30), proteomic profiling of autopsy brain
tissue from patients with Alzheimers and Parkinsons disease (compared with
control specimen from healthy subjects) (31, 32), and proteomic analysis of the
CSF of patients with schizophrenia (33) or Alzheimers disease (34). Multinational
pharmaceutical and smaller private biotechnology-based companies show a huge
interest in using proteomic techniques for new discoveries in psychotropic drug
development (antipsychotics, anxiolytics, depression, schizophrenia) (25, 35).
Application of proteomics as a read-out of target-drug interactions is just
beginning to be explored. Its success depends on whether there really are discretely
identifiable patterns of changes in proteins associated with specific drug-target
molecular interactions.
From a practical experimental viewpoint, the combination of CSF sampling
with proteomics enables access to a body fluid in close contact with brain cells.
CSF has only minimal protein content, which makes analyses less complicated
compared with serum proteomic patterns. We are currently exploring the best
means of achieving standardization of CSF collection, which is mandatory for
its use in proteomic studies. In a parallel effort, because blood contamination
cannot be entirely avoided during lumbar puncture, methods to correct for the
variable contamination-associated changes in the CSF proteomic profile are being
developed (36). A large survey in Europe confirmed that CSF/serum quotients of
proteins represent method-independent values approaching the quality of reference
values (37, 38).
3 Dec 2004
20:41
234
AR
BIECK
AR232-PA45-09.tex
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
POTTER
not be surrogates for effects in the brain. Nonetheless, because there is only one
type of NET in the nervous system (peripheral and central), and metabolites of
NE formed in the brain are excreted, some of the peripheral biomarkers might
reflect changes in the CNS. Past experience has shown that peripheral biomarkers
obtained early during clinical pharmacological evaluation of MAO inhibitors (40)
can be validated at a later stage with PET imaging in peripheral organs (41) and
in brain (42).
Interestingly, investigations of clinical syndromes can also provide support for
an array of measures as indices of drug action. Recently, a functional polymorphism in the human NET has been discovered (43) in patients with orthostatic
intolerance taking the form of a NET deficiency that can be assessed simultaneously with a multitude of measures (44). The same type of measures should also be
applicable for testing drug-induced NET deficiency. In a study with duloxetine, a
5-HT and NE reuptake inhibitor, the following measures were applied: vital signs,
tyramine pressor test, posture test, NE and its metabolite DHPG in plasma and
urine, plasma melatonin, plasma inhibition of [3H]-nisoxetine binding (ex vivo
ligand), and plasma duloxetine. Selected results, consistent with a dose response
on measures related to NET inhibition, are shown in Figure 1ad. The findings of
this study suggest that a portfolio of biomarkers is useful for the assessment of
NET inhibition because there were substantial differences in the sensitivity with
which the different downstream biomarkers were affected: ex vivo binding =
DHPG: NE ratio > tyramine pressor test > heart rate (45). Assessment of such
biomarkers in CSF, plasma, and urine during treatment with the potent NET reuptake inhibitor atomoxetine (46) is presently ongoing. In the future, it will be
possible to validate such peripheral NET reuptake biomarkers with the highly sensitive (but expensive!) PET imaging as well as by comparing them with proteomic
patterns in the CSF.
CLINICAL RESPONSE
Evidence of drug effect in a model experimental paradigm can also be considered
as a type of biomarker. Historically, provocative anxiety tests have been given to
diagnose patients with panic disorders, but they also have been used in studies
on healthy subjects for psychotropic drug development. The scope in drug development has been to delineate the mechanism of action of potential antianxiety
agents and to determine a minimum effective dose and the duration of effect. However, these tests have not proved predictive of efficacy and are therefore limited
to providing evidence of some drug effect (47). Pharmacological challenges with
lactate, carbon dioxide, or cholecystokinin (CCK) can produce anxiety in healthy
human subjects (4852). They are usually regarded as models of unconditioned
anxiety, comparable to panic disorder. Their use requires careful consideration of
the experimental setting and the physiological changes. For example, Na-lactate
requires a 20-min infusion and CO2 needs several inhalations for 20 min each (53).
CCK can be given as i.v. bolus.
3 Dec 2004
20:41
AR
AR232-PA45-09.tex
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
235
Figure 1 (a) Quantal dose-response curves of DHPG/NE ratio in plasma before and
during duloxetine treatment. (b) Quantal dose-response curves of DHPG/NE ratio in
urine before and during duloxetine treatment. (c) Effect of duloxetine on ex vivo [3H]
nisoxetine binding to NE transporters. (d) Effect of duloxetine on tyramine pressor
dose to raise systolic blood pressure by 30 mm Hg (PD30). From Reference 45.
3 Dec 2004
20:41
236
AR
BIECK
AR232-PA45-09.tex
POTTER
Figure 1 (Continued)
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
3 Dec 2004
20:41
AR
AR232-PA45-09.tex
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
237
There are major differences between challenges with the endogenous occurring
CCK, Na-lactate, and CO2. Patients with panic disorders show a left-shifted dose
response (i.e., are more sensitive) to CCK in comparison to healthy subjects (54). It
has been argued that CCK fulfills the criteria of an ideal anxiety-challenging model
(55). CCK not only elicits symptoms of anxiety but also produces physiological
and hormonal changes through activation of the HPA axis (48, 51, 52). This is not
the case during lactate or CO2 challenge. Thus, there is a wider range of acute CCK
effects that can be affected by drugs of multiple classes, such as imipramine (56),
benzodiazepines (57), and vigabatrine (58), all of which antagonize CCK effects
on panic symptoms.
3 Dec 2004
20:41
238
AR
BIECK
AR232-PA45-09.tex
TABLE 4
LaTeX2e(2002/01/18)
P1: GCE
POTTER
Time course of biomarker collection
Drug
Fluoxetine SSRI
AR232-PA45-09.sgm
Publication
Biomarker
mGlu2/3 receptor Ki
PK: plasma and CSF
PK: plasma and CSF
CSF proteomics
Anxiety model: CO2
Anxiety model: CCK
Clinical studies
Reference
(62)
(64)
(65)
(66)
(63, 67)
(68)
(69)
(70)
(71)
(72)
(73)
& Asberg:
Brit. J. Psychiat. (1979), 134, 382-9). 5-HIAA: 5-hydroxy indole acetic acid; PD: pharmacodynamic(s); PK: pharmacokinetic(s); 19F MRS: Fluorine 19 magnetic resonance spectroscopy; PET: positron
emission tomography; 5-HT: 5-hydroxytryptamine; Ki: inhibitory constant; mGlu: metabotropic glutamate;
CCK: cholecystokinin.
CNS receptor occupancy was defined in healthy subjects to predict the occupancy
of central NK-1 receptors. The effective doses of aprepitant in patients with CINV
were 125 mg and 375 mg (78), doses that lead to a receptor occupancy of >90%
in healthy subjects. In a Phase II trial, therapy with aprepitant was associated with
improvements in depression and anxiety symptoms that were quantitatively comparable with those seen with selective serotonin reuptake inhibitors (SSRIs) and
significantly greater than those seen with placebo (79). But, in 2003 the Phase III
clinical program was halted because the compound failed to demonstrate efficacy
for the treatment of depression despite being used at doses producing >90% NK-1
receptor occupancy in the brain. Thus, the hypothesis that antagonism of NK-1
receptors produces antidepressant effects was properly tested and not supported.
If trials with other NK-1 antagonists also fail to show sustained antidepressant
effects, this will stand as the first example in antidepressant research of using a
biomarker to show that a drug really did engage its target and thereby reject a
hypothesis, not just a compound.
3 Dec 2004
20:41
AR
AR232-PA45-09.tex
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
239
3 Dec 2004
20:41
240
AR
BIECK
TABLE 5
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
POTTER
Drug
AR232-PA45-09.tex
CSF
PKa
Proteomics
CSF
Brain
access
Receptor
[Ki]b
PDc
effects
Rxd
effects
Atomoxetine
Strattera
ADHDe
Fluoxetine
Prozac
Depression
Duloxetine
Cymbalta
Depression
LY354740
Anxiety
(+)h
Aprepitant
Emend
CINVf
PD: Pharmacodynamic(s).
d
e
+ (I)g
PK: Pharmacokinetic(s).
b
c
+ (I)g
Rx: Treatment.
CONCLUSION
Many extensive reviews on biomarkers in drug development have been published
(24, 82, 83). This review focuses on the current status of biomarkers and/or approaches critical to assessing novel neuroscience targets with an emphasis on new
paradigms and challenges in this field of research. Nevertheless, some old questions
remain, such as how to verify access to the brain. Early imaging studies including
micro-PET (84, 85) can help to overcome this, at least for those compounds that
can be labeled and/or shown to affect another ligand. However, in case of delayed
tracer development or because of no feasible application of brain imaging effects
of the molecule, using CSF as a matrix could fill this gap. It is hoped that proteomic
research using CSF will have a major impact on the development of treatments for
psychiatric disorders.
In this review, the importance of biomarker data integration for psychotropic
drug development has been illustrated with examples for both clinically used medications and investigational drugs. The combination of biomarker development with
3 Dec 2004
20:41
AR
AR232-PA45-09.tex
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
241
current biomedical technologies applied to drug discovery can improve the level of
innovation and efficiency of drug discovery and developmental programs because
whether or not a drug engages its prestated target can be formally tested.
ACKNOWLEDGMENTS
The authors would like to acknowledge Robert A. Padich, Ph.D. of Lilly Global
Scientific Information and Communications for assisting with the preparation of
this manuscript. The authors are full-time employees of Eli Lilly and Company,
but the views expressed in this review are those of the authors.
APPENDIX
Definitions
Generally accepted definitions defined by working groups (86):
Pharmacologic marker (effect or response): A change representing a molecular interaction between drug and body constituent or the observable output.
Proteomics: Proteomics represents the effort to establish the identities, quantities, structures, and biochemical and cellular functions of all proteins in an
organism, organ, or organelle, and how these properties vary in space, time,
or physiological state (87).
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Bailey WJ, Ulrich R. 2004. Molecular profiling approaches for identifying
novel biomarkers. Expert Opin. Drug Saf.
3:13751
2. Colburn WA. 2003. Biomarkers in drug
discovery and development: from target
3 Dec 2004
20:41
242
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
AR
BIECK
AR232-PA45-09.tex
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
POTTER
16.
17.
18.
19.
20.
21.
22.
23.
24.
24a.
3 Dec 2004
20:41
AR
AR232-PA45-09.tex
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
243
3 Dec 2004
20:41
244
44.
45.
46.
47.
48.
49.
50.
51.
52.
AR
BIECK
AR232-PA45-09.tex
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
POTTER
53.
54.
55.
56.
57.
58.
59.
60.
3 Dec 2004
20:41
AR
AR232-PA45-09.tex
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
69.
70.
71.
72.
73.
74.
75.
76.
245
3 Dec 2004
20:41
246
77.
78.
79.
80.
81.
AR
BIECK
AR232-PA45-09.tex
AR232-PA45-09.sgm
LaTeX2e(2002/01/18)
P1: GCE
POTTER
82.
83.
84.
85.
86.
87.
7 Jan 2005
13:58
AR
AR232-PA45-10.tex
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
10.1146/annurev.pharmtox.45.120403.095930
INTRODUCTION
Pest insect control, an essential component of crop protection and public health,
has evolved over a recorded history of three millennia (1, 2). Sulfur was first referred to by Homer in 1000 BC as a fumigant for pest control, and, in California,
it is still used in larger amounts than any other pesticide. Nicotine in the form of
tobacco extracts was reported in 1690 as the first plant-derived insecticide, followed by the pyrethrins from pyrethrum flowers and rotenone from derris roots
in the early 1800s. Synthetic organics in the 1940s to the 1970s largely replaced
inorganics and botanicals with the introduction of organophosphates, methylcarbamates, organochlorines, and pyrethroids. With each new chemical class, resistant
strains were soon selected to limit their effectiveness. Genetically modified crops
expressing Bacillus thuringiensis (Bt) -endotoxin were introduced for pest insect
0362-1642/05/0210-0247$14.00
247
7 Jan 2005
13:58
248
AR
AR232-PA45-10.tex
TOMIZAWA
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
CASIDA
control in 1995. Many of the remaining gaps in pest control capabilities were filled
recently by the neonicotinoids (Figure 1), which combine outstanding effectiveness
with relatively low toxicity to vertebrates (37).
NEONICOTINOIDS
The current synthetic organic insecticides were discovered by modifying natural products, e.g., using the pyrethrins as a prototype for synthetic pyrethroids,
or by screening hundreds of thousands of structurally diverse compounds for
novel leads. Nicotine is still used as a minor insecticide, particularly in China,
but attempts to improve its insecticidal activity, e.g., 3 ,4 -dehydronicotine and
3-(alkylaminomethyl)-pyridines (8), were not successful. The lead for the neonicotinoids, 2-(dibromonitromethyl)-3-methylpyridine, was discovered in 1970 by
Shell Development Company in California to have modest activity against house
flies and pea aphids (911). Molecular modifications to achieve optimal potency
on corn earworm larvae (a major lepidopterous pest) culminated with nithiazine,
but unfortunately it could not be commercialized for crop protection due to photoinstability (10, 12). This shortcoming relegated nithiazine to a niche market for
fly abatement in poultry and animal husbandry (11). A major improvement of its
structure was made by Nihon Tokushu Noyaku Seizo in Japan (presently Bayer
Crop Science Japan) by introducing a chloropyridinylmethyl group, leading to
a nitromethylene prototype of outstanding potency on green rice leafhopper (a
major pest of rice and vegetables). However, photoinstability again prevented its
use for crop protection. Further structure-activity studies established that good
activity was retained on replacement of the imidazolidine by thiazolidine or oxadiazinane or acylic counterpart, and the chloropyridinylmethyl by chlorothiazolylmethyl or tetrahydrofuranmethyl. Changing the nitromethylene to nitroguanidine
or cyanoamidine afforded photostability and produced highly effective compounds
in field conditions (3, 13, 14). The current neonicotinoids and their year of patent are
the heterocyclics nithiazine (1977), imidacloprid (IMI) (1985), thiacloprid (1985),
and thiamethoxam (1992); and the acyclics nitenpyram (1988), acetamiprid (1989),
clothianidin (1989), and dinotefuran (1994). The physical properties of the neonicotinoids and nicotine are compared in Table 1. Molecular weights range from 160
to 292 and log P values from 0.66 to 1.26. The compounds vary in water solubility
from 0.1850.61 g/l for clothianidin, IMI, and thiacloprid, to infinite for nicotine.
The neonicotinoids are the only major new class of insecticides developed in the
past three decades. Worldwide annual sales of neonicotinoids are approximately
one billion dollars, accounting for 11%15% of the total insecticide market. They
are readily absorbed by plants and act quickly, at low doses, on piercing-sucking insect pests (aphids, leafhoppers, and whiteflies) of major crops. The neonicotinoids
are poorly effective as contact insecticides and for control of lepidopterous larvae.
They are used primarily as plant systemics; when applied to seeds, soil, or foliage
they move to the growing tip and afford long-term protection from piercing-sucking
7 Jan 2005
13:58
AR
AR232-PA45-10.tex
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
249
NEONICOTINOID TOXICOLOGY
7 Jan 2005
13:58
250
AR
AR232-PA45-10.tex
TOMIZAWA
TABLE 1
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
CASIDA
Compound
Molecular weight
Log Pa
222.7
249.7
202.2
255.7
270.7
160.1
252.7
291.7
4.25
0.300.34
54.3
0.61
>590
200
0.185
4.1
0.80
0.7
0.64
0.57
0.66
0.60
1.26
0.13
162.2
Neonicotinoids
Acetamiprid
Clothianidin
()-Dinotefuran
Imidacloprid
Nitenpyram
Nithiazine
Thiacloprid
Thiamethoxam
Nicotinoid
()-Nicotine
P = 1-octanol/water partition.
TOXICOLOGY
The neonicotinoids have unique physical and toxicological properties as compared
with earlier classes of organic insecticides (Table 2). They generally have the lowest log P values, which is consistent with their outstanding plant systemic activity
shared by some organophosphates and methylcarbamates but not by the more
lipophilic organochlorines and pyrethroids. The neonicotinoids and pyrethroids
7 Jan 2005
13:58
AR
AR232-PA45-10.tex
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
251
NEONICOTINOID TOXICOLOGY
TABLE 2
Potency
(LD50 , mg/kg)c
Systemic
action
Nerve targetb Insects Rats
Class
Log P
Neonicotinoids
0.7 to 1.3 +
Selectivity
factor
nAChR
2.0
912
456
Organophosphates 1 to 5.5
AChE
2.0
67
33
Methylcarbamates 1 to 3
AChE
2.8
45
16
Organochlorines
5.5 to 7.5
Na+ or Cl
channels
2.6
230
91
Pyrethroids
4 to 9
Na+ channel
0.45
2000
4500
Insecticide examples are the organophosphates parathion and malathion (as their oxon metabolites) and
methylcarbamates carbaryl and aldicarb inhibiting AChE, the organochlorine DDT and the pyrethroid deltamethrin
acting on the voltage-sensitive sodium channel, and the organochlorines endosulfan and lindane blocking the
-aminobutyric acid (GABA)-gated chloride channel.
Geometric means of large data sets (11 to 83 items each) for rat acute oral and insect topical (principally four
species) LD50 values for all classes of compounds except neonicotinoids (24). Values for the neonicotinoids are
geometric means for rat oral LD50 data in Table 3 and arbitrary for insects to reflect similar potency of neonicotinoids
and organophosphates on the same target insects.
have higher selectivity factors for insects versus mammals than the organophosphates, methylcarbamates, and organochlorines. This is attributable to both target
site specificity and detoxification, which are considered later. The neonicotinoids
act as agonists at the nicotinic acetylcholine receptors (nAChRs) of insects and
mammals (particularly the 42 subtype) (7).
The toxicological profiles of the individual neonicotinoids and nicotine are compared in Table 3. The acute oral LD50 values (mg/kg) for rats range from 5060 for
nicotine to >5000 for clothianidin. When ranked on the basis of chronic toxicity
to rats, reported as no-observed-adverse-effect-level (NOAEL; the principal toxicological parameter used in risk assessment), thiacloprid and thiamethoxam have
the lowest values (0.61.2 mg/kg/day) and are rated as likely human carcinogens.
Intermediate values (5.79.8 mg/kg/day) are observed for acetamiprid, clothianidin, and IMI, whereas dinotefuran has the highest value. The EPA has not followed
a cumulative risk approach in determining pesticide tolerances for neonicotinoids
and has not assumed that each neonicotinoid has a common mechanism of toxicity
with other substances (2530). Of the commercial neonicotinoids, acetamiprid,
IMI, and thiacloprid are the most toxic to birds, and thiacloprid to fish. Several
neonicotinoids are harmful to honeybees, either by direct contact or ingestion, but
potential problems can be minimized or avoided by treating seeds and not spraying
flowering crops (15).
The mammalian toxicity of neonicotinoids is considered to be centrally mediated because the symptoms of poisoning are similar to those of nicotine. Toxicity
correlates with agonist action and binding affinity at the vertebrate 42 nAChR,
7 Jan 2005
13:58
252
AR
AR232-PA45-10.tex
TOMIZAWA
TABLE 3
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
CASIDA
P1: GCE
Birdf
Fishg
Compound
Acute oralc
LD50 (mg/kg)
NOAELd
(mg/kg/day)
Carcinogene
Acute oral
LD50 (mg/kg)
LC50
(ppm)
Neonicotinoids
Acetamiprid
Clothianidin
()-Dinotefuran
Imidacloprid
Nitenpyram
Nithiazine
Thiacloprid
Thiamethoxam
182
>5000
2400
450
1628
300
640
1563
7.1
9.8
127
5.7
1.2
0.6
No
No
No
No
Yes
Yes
180
>2000
>2000
31
>2250
49
1552
>100
>100
>40
211
>1000
150
31
>100
Nicotinoid
()-Nicotine
5060
Toxic
b
c
Dermal LD50 values of neonicotinoids are >2000 to >5000 mg/kg (rat) except for ()-nicotine 50 mg/kg (rabbit).
Average data for male and female rats with sex difference less than twofold.
No-observed-adverse-effect-level (NOAEL) for chronic toxicity studies in rats. This value also applies to all adverse effects
in chronic toxicity studies with mice and dogs.
Thiacloprid gives thyroid and uterine tumors in rats and ovary tumors in mice. Thiamethoxam gives hepatocellular adenomas
and carcinomas in male and female mice. They are considered to be likely human carcinogens.
7 Jan 2005
13:58
AR
AR232-PA45-10.tex
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
NEONICOTINOID TOXICOLOGY
253
BIOTRANSFORMATIONS
Metabolism of the commercial neonicotinoids has been extensively studied in
crops, rats, lactating goats, and laying hens (15, 40). These studies are part of
the EPA registration requirements for approved uses. Extensive data has been
published for IMI (28, 41), thiacloprid (29, 42), clothianidin (26, 43, 44), and to
a lesser extent, nithiazine (45), nitenpyram (40), acetamiprid (40), thiamethoxam
(30), and dinotefuran (27). Owing to their relatively high water solubility and slow
metabolism in mammals, some (IMI and thiacloprid) to almost all (clothianidin,
dinotefuran, and nitenpyram) of an oral neonicotinoid dose is excreted unchanged
in urine. The chemical fate of neonicotinoids in and on crops is governed both by
metabolic and photochemical reactions. These processes may produce identical or
different products depending on the mechanisms involved.
Analysis of neonicotinoid residues to enforce crop tolerances and registered
uses involves the parent compound and toxic metabolites. IMI residues are determined as the parent compound plus metabolites with the chloropyridinyl moiety.
Thiacloprid is combined with an amide and a hydroxy derivative in evaluating
residues. Clothianidin and acetamiprid residues are regulated as the parent compounds. Thiamethoxam residues are considered along with those of its principal
metabolite clothianidin. Dinotefuran residues are combined with those of its guanidine and urea metabolites. The analyses are achieved by various combinations of
high-performance liquid chromatography or gas chromatography with UV, electron capture, or mass spectrometry for detection and characterization.
Most neonicotinoids undergo metabolic alterations at multiple sites (Figure 2).
For convenience, different parts of the molecules are considered separately, indicating the known or presumed effects of the reactions on bioactivity. In Figure 2A,
oxidation of the nitromethylene carbon of nithiazine is likely a detoxification
mechanism (45). IMI is hydroxylated in the imidazolidine moiety at either one of
the two methylene substituents, which is followed by conjugation or dehydration
to form the olefin, apparently with little or no ring opening; these unconjugated
metabolites retain insecticidal activity or insect nAChR potency (4648). Some
N-demethylation is observed in each case with compound-dependent effects on
product potency. It greatly increases insecticidal and/or receptor activity for Nmethyl-IMI (a model compound) (16), nitenpyram (49), and thiamethoxam (50,
51). Thiamethoxam is readily converted to clothianidin by ring methylene hydroxylation in insects and plants (52), whereas clothianidin undergoes N-demethylation
(43, 44). The thiazolidine ring of thiacloprid is opened and the sulfur oxidized
7 Jan 2005
13:58
254
AR
AR232-PA45-10.tex
TOMIZAWA
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
CASIDA
7 Jan 2005
13:58
AR
AR232-PA45-10.tex
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
NEONICOTINOID TOXICOLOGY
255
ring opening and liberation of an aldehyde that forms cyclic derivatives (27). The
=N NO2 (nitroguanidine) moiety (Figure 2C) of IMI is reduced to C=
=N NO
C=
=N NH2 (aminoguanidine), and cleaved to the C=
=NH
(nitrosoguanidine) and C=
=O (urea) derivatives. The aminoguanidine of clothianidin is
(guanidine) and C=
also conjugated with pyruvic or acetic acid, followed by cyclization (26). The
nitrosoguanidine metabolite of IMI has moderate to high insecticidal activity and
insect nAChR potency (53), whereas the guanidine metabolite is highly activated
against mammalian but deactivated against insect nAChRs (23, 38, 54).
Cytochrome P450 (CYP450) isozymes are involved in oxidative IMI metabolism. Based on studies with individual recombinant enzymes, human CYP3A4
is the predominant IMI N-methylene hydroxylase (55). The insecticidal activity
of many neonicotinoids is strongly synergized by CYP450 inhibitors, such as
piperonyl butoxide and O-propyl O-propynyl phenylphosphonate, suggesting that
these enzymes limit their efficacy (16, 56). Fruit flies (Drosophila melanogaster)
overexpressing CYP6G1 are resistant to IMI, probably owing to enhanced detoxification (57, 58). Several human P450s also reduce IMI to the nitroso derivative in
an oxygen-sensitive manner (55). A relatively oxygen-insensitive neonicotinoid
nitro reductase of rabbit liver cytosol (59) was recently identified as aldehyde
oxidase and readily converts IMI to nitrosoguanidine and aminoguanidine (60).
The aminoguanidine of IMI (a hydrazone) is further derivatized to an acetaldehyde
imine upon incubation with mammalian liver cytosol (60) and to a triazinone and
=N CN moiety of thiaother conjugates in vitro and in vivo (59, 60a). The C=
=NC(O)NH2] and also undergoes N CN
cloprid is hydrolyzed to the amide [C=
cleavage. Descyanothiacloprid (a plant metabolite) (42) is a particularly potent
mammalian nAChR agonist (23).
Toxicokinetic studies in mammals, so important in pharmaceutical research, are
often given lower priority in pesticide investigations. As an example, it is not clear
if the toxicity of IMI in mammals is due to the parent compound or the desnitro
metabolite (which enters the brain following direct intraperitoneal administration
in mice) (54). IMI is highly absorbed in a human intestinal cell model, suggesting
potential effects in mammals following ingestion (61).
NICOTINIC RECEPTORS
The vertebrate nAChR is an agonist-gated ion channel responsible for rapid excitatory neurotransmission. It is a pentameric transmembrane complex in the superfamily of neurotransmitter-gated ion channels, including -aminobutyric acid
(GABAA and GABAC), glycine, and 5-HT3 serotonin receptors. The nAChR consists of diverse subtypes assembled in combinations from ten , four , , , and
subunits. The skeletal muscle or electric ray (Torpedo) subtype is made up of
two 1 subunits and one each of 1, , and (or in adult muscle) subunits.
Neuronal nAChR subtypes expressed in vertebrate brain and ganglia are assembled in combinations of 210 and 24, and are pharmacologically classified
into two groups based on sensitivity to -bungarotoxin (-BGT). The 26 and
7 Jan 2005
13:58
256
AR
AR232-PA45-10.tex
TOMIZAWA
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
CASIDA
7 Jan 2005
13:58
AR
AR232-PA45-10.tex
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
NEONICOTINOID TOXICOLOGY
257
Compound
Insect
Vertebrate 42b,c
Selectivity ratio
Neonicotinoids
Acetamiprid
Clothianidin
()-Dinotefuran
Imidacloprid
Nitenpyram
Nithiazine
Prototype
Thiacloprid
Thiamethoxam
8.3
2.2
900
4.6
14
4800
0.24
2.7
5000
700
3500
>100,000
2600
49,000
26,000
210
860
>100,000
84
1591
>111
565
3500
5.4
875
319
>20
Nicotinoids
Desnitroimidacloprid
Descyanothiacloprid
()-Nicotine
()-Epibatidine
1530
200
4000
430
8.2
4.4
7.0
0.04
0.005
0.022
0.002
0.0001
IC50 values for displacing [3H]imidacloprid binding to the house fly (Musca domestica) (acetamiprid),
aphid (Myzus persicae) (thiamethoxam), and fruit fly (the other neonicotinoids) receptor, and [3H]nicotine
binding to the vertebrate 42 nAChR.
IC50 values (M) for the vertebrate 7 nAChR subtype (assayed by [125I]-BGT binding) are acetamiprid,
290; clothianidin, 190; ()-dinotefuran, >1000; imidacloprid, 270; nitenpyram, >300; nithiazine, >300;
prototype neonicotinoid, 6.1; thiacloprid, 100; thiamethoxam, >300; desnitroimidacloprid, 9.9;
descyanothiacloprid, 6.0; ()-nicotine, 21; and ()-epibatidine, 0.031.
7 Jan 2005
13:58
258
AR
AR232-PA45-10.tex
TOMIZAWA
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
CASIDA
coincide with that of ()-nicotine (4.8 A) (104, 105). This is rationalized by placing
an additional directional requirement on the ammonium nitrogen atom, which, in
modeling studies, confers reasonable overlap of ()-nicotine and ()-epibatidine
by orienting the pyridine nitrogen atom proximal to the sp3 nitrogen atom (N N
(68, 106). The iminium (+C NH2 C=
=+NH2) metabolites
distance of 4.79 A)
of IMI and thiacloprid (desnitro and descyano, respectively) are mostly protonated
at physiological pH (23).
Neonicotinoids have, instead of an easily protonated nitrogen, the nitro or cyano
or equivalent electronegative pharmacophore, and they demonstrate a coplanarity
7 Jan 2005
13:58
AR
AR232-PA45-10.tex
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
NEONICOTINOID TOXICOLOGY
259
between this tip and the substituted guanidine/amidine moiety (Figure 3) (23, 53,
107, 108). The guanidine moiety of IMI has pKa values of 1.56 and 11.12 for
protonation and deprotonation, respectively (109), indicating less than 0.0002%
protonation at physiological pH. The coplanarity allows a conjugated electronic
system that facilitates negative charge flow toward the electronegative tip to consolidate the binding. Interestingly, high affinity and insect receptor selectivity are
=N NO) analogs of neonicotinoids, suggesting that only
retained in nitroso (C=
one (O1) of the electronegative oxygen atoms is essential (53). The role of N1
(Figure 3) in neonicotinoid action is of particular interest. The distance between
the van der Waals surfaces of the two nitrogen atoms in nicotine (102) is the same
as that between the pyridinyl and N1 nitrogen atoms in the neonicotinoids (5.45
Additionally, a partial positive charge ( +) for the N1 nitrogen atom is
6.06 A).
conferred by the electron-withdrawing nitro or cyano substituent (74, 83, 107)
as supported by comparative molecular field analysis (110) and semiempirical
molecular orbital theory (PM3) calculation (111). The same relationship in the
above distance is also observed between N1 and O1 of the neonicotinoids (107).
However, the N1 atom does not have a significant positive charge in PM3 and
high-level ab initio calculations (17, 53, 112), and it can be replaced by a carbon
atom with retention of significant binding activity to the Drosophila receptor and
toxicity to insects (108, 113). The pyridin-3-ylmethyl substituent or equivalent
moiety greatly enhances the binding affinity (82, 108). Thus, in terms of binding,
there is a crucial role for the nitro or cyano group, an important contribution from
the pyridine nitrogen, and a complementary role for N1 of the neonicotinoids (23,
53). Other molecular features also affect the selectivity. The N-methyl group of
thiamethoxam favors a specific receptor interaction particularly at low temperature
with Myzus compared with Drosophila or other insects (114, 114a); the preference
of insect nAChRs for chloropyridinyl versus chlorothiazolyl moieties depends on
the rest of the molecule (Figure 1) (51); introducing azido or amino at the 5-position
of the 6-chloropyridin-3-yl moiety of neonicotinoids and epibatidine reduces the
potency for Drosophila but not for 42 and 7 nAChRs (115, 116).
7 Jan 2005
13:58
260
AR
AR232-PA45-10.tex
TOMIZAWA
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
CASIDA
protein as a complex with nicotine reveals the following molecular features: the
carbonyl oxygen of a tryptophan in loop B contacts through a hydrogen bond with
the ammonium nitrogen of nicotine; the carbonyl oxygen of a leucine and amide
nitrogen of a methionine (both from complementary loop E) make hydrogen bonds
with the pyridine nitrogen of nicotine through a bridging water molecule (123a).
The neonicotinoids are an anomaly for the nicotinoid cation- interaction model
(53). The electronegative pharmacophore, crucial for optimum potency of the
neonicotinoids, is proposed to associate with a cationic subsite (possibly lysine,
arginine, or histidine) in the insect nAChR (Figure 4) (23, 53, 108). Lysine and arginine are prominent (and histidine minor) in the extracellular domain of D2, the
main neonicotinoid-binding subunit (94, 98100). Although no direct information
is available on the actual location of the relevant residue(s), photoaffinity labeling
with a suitable neonicotinoid ligand (115) coupled with computer-assisted docking
simulation (118) may help define the orientation of the neonicotinoid electronegative tip in the binding domain. A point mutation (glutamine to arginine or lysine)
on the avian 7 subunit confers enhanced electrophysiological response for IMI
at 3 mM compared to that in the wild type (although the affinity of IMI on this
Figure 4 Binding subsite specificity shown as hypothetical schematic models for neonicotinoid imidacloprid binding in the insect nAChR and nicotinoid desnitroimidacloprid binding
in the mammalian nAChR, each at the ACh agonist site. The positioning of desnitroimidacloprid and the interacting amino acids in the mammalian site is based on earlier modeling with
ACh and nicotinoids (23, 53, 62, 65, 117, 118). In the mammalian binding site, loops A-C are
from an subunit and loop D from a complementary subunit. The insect (Drosophila) nAChR
subunits conserve the aromatic and vicinal cysteine residues suitably positioned from homology modeling. Imidacloprid is arbitrarily placed in the same way as desnitroimidacloprid and
a lysine (or alternatively arginine or histidine) cationic residue is introduced to interact with
the negatively charged ( ) tip important in selectivity for insect versus mammalian nicotinic
receptors.
7 Jan 2005
13:58
AR
AR232-PA45-10.tex
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
NEONICOTINOID TOXICOLOGY
261
mutated receptor remains unchanged) (124), providing possible support for a binding model featuring the role of the neonicotinoid electronegative pharmacophore
(23, 53). These relationships provide a testable model for the hypothesis that
specific subsite differences between insect and vertebrate receptors confer neonicotinoid selectivity.
LITERATURE CITED
1. Ware GW. 2000. The Pesticide Book.
Fresno, CA: Thomson Publ. 418 pp.
2. Casida JE, Quistad GB. 1998. Golden
age of insecticide research: past, present,
7 Jan 2005
13:58
262
4.
5.
6.
7.
8.
9.
10.
11.
AR
AR232-PA45-10.tex
TOMIZAWA
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
CASIDA
7 Jan 2005
13:58
AR
AR232-PA45-10.tex
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
NEONICOTINOID TOXICOLOGY
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
263
7 Jan 2005
13:58
264
45.
46.
47.
48.
49.
50.
51.
52.
AR
AR232-PA45-10.tex
TOMIZAWA
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
CASIDA
7 Jan 2005
13:58
AR
AR232-PA45-10.tex
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
NEONICOTINOID TOXICOLOGY
60a.
61.
62.
63.
64.
65.
66.
67.
68.
69.
265
7 Jan 2005
13:58
266
79.
80.
81.
82.
83.
84.
85.
86.
87.
AR
AR232-PA45-10.tex
TOMIZAWA
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
CASIDA
88.
89.
90.
91.
92.
93.
94.
95.
7 Jan 2005
13:58
AR
AR232-PA45-10.tex
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
NEONICOTINOID TOXICOLOGY
96.
97.
98.
99.
100.
101.
102.
103.
104.
105.
106.
107.
108.
109.
110.
111.
112.
267
7 Jan 2005
13:58
268
113.
114.
114a.
115.
116.
117.
118.
119.
AR
AR232-PA45-10.tex
TOMIZAWA
AR232-PA45-10.sgm
LaTeX2e(2002/01/18)
P1: GCE
CASIDA
120.
121.
122.
123.
123a.
124.
125.
HI-RES-PA45-10-Tomizawa.qxd
1/7/05
2:24 PM
Page 1
NEONICOTINOID TOXICOLOGY
C-1
7 Jan 2005
13:59
AR
AR232-PA45-11.tex
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
10.1146/annurev.pharmtox.45.120403.095902
GLYCERALDEHYDE-3-PHOSPHATE
DEHYDROGENASE, APOPTOSIS, AND
NEURODEGENERATIVE DISEASES
De-Maw Chuang,1 Christopher Hough,2
and Vladimir V. Senatorov3
1
INTRODUCTION
Recent research has revealed a small class of proteins whose members are endowed with multiple functions (for review, see 1). Glyceraldehyde-3-phosphate
dehydrogenase (GAPDH) is a typical example. Historically, GAPDH has been
The U.S. Government has the right to retain a nonexclusive, royalty-free license in and to
any copyright covering this paper.
269
7 Jan 2005
13:59
270
AR
AR232-PA45-11.tex
CHUANG
HOUGH
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
SENATOROV
considered a glycolytic enzyme with a key role in energy production. It has also
been regarded as a product of a housekeeping gene whose transcript level remains
constant under most experimental conditions, and it has been frequently used as an
internal control in studying the regulation of gene expression. Mounting evidence,
however, supports the view that the expression of GAPDH is regulated and that
GAPDH is a protein with multiple intracellular localizations and diverse activities independent of its traditional role in glycolysis (for review, see 2, 3). These
new activities include regulation of the cytoskeleton (4, 5), membrane fusion and
transport (68), glutamate accumulation into presynaptic vesicles (9), and binding
to low-molecular-weight G proteins (10). A role of GAPDH in nuclear function
is also suggested by its ability to activate transcription in neurons (11), to export
nuclear RNA (12), and to effect DNA repair (13). Particularly intriguing are the
increasing reports that GAPDH is an integral part of various forms of apoptosis and
may participate in neuronal death in some neurodegenerative diseases. The aim of
this review is to provide evidence for the involvement of GAPDH in the apoptotic
cascade, to discuss potential underlying mechanisms, and to evaluate the roles of
GAPDH in preclinical models of neurodegeneration and human disease and in
the pharmacological treatments of neurodegenerative diseases. Several reviews in
these areas have appeared (2, 1420).
7 Jan 2005
13:59
AR
AR232-PA45-11.tex
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
271
role in cell death came from studies using an antisense oligonucleotide knockdown
strategy. Antisense oligonucleotides directed against either the translation initiation
site or a segment of the coding region of GAPDH mRNA delay cell death; in
contrast, the corresponding sense and scrambled oligonucleotides are ineffective
(28, 29). Additionally, antisense, but not sense, oligonucleotides block the increase
in GAPDH mRNA and protein preceding the apoptotic death with little or no effect
on basal levels. Using similar approaches, GAPDH was found to be associated with
apoptosis, but not necrosis, of cerebellar granule cells grown in culture medium
with reduced concentrations of KCl (30).
The paradigm of cytosine arabinoside (AraC)-induced cell death has been used
to further investigate the generality of the role of GAPDH in neuronal apoptosis and the molecular events associated with this process. AraC is a pyrimidine
antimetabolite that has been used clinically for the treatment of acute leukemia.
Freshly plated cerebellar granule cells exposed to AraC undergo rapid apoptotic
neuronal death that is preceded by an upregulation of the tumor suppressor protein
p53 followed by an increase in levels of GAPDH and Bax, another proapoptotic
protein (31). Again, the role of GAPDH in AraC-induced apoptosis was confirmed
by antisense experiments (31, 32). Interestingly, a p53 antisense oligonucleotide
not only suppresses apoptosis and decreases p53 and Bax mRNA induced by AraC,
but also reduces the levels of upregulated GAPDH mRNA and protein. In the same
study, it was shown that neurons prepared from p53-deficient mice are resistant
to AraC neurotoxicity, and that p53 gene knockout also suppresses AraC-induced
GAPDH expression. Moreover, infection of PC12 pheochromocytoma cells with
an adenoviral vector containing the wild-type p53 gene dramatically increases
GAPDH expression and triggers cell death (31). Taken together, these results lend
support for the view that GAPDH is a novel target of p53, directly or indirectly
regulated by this proapoptotic transcription factor, and could be one of the downstream apoptotic mediators. A scan of the rat GAPDH promoter reveals three
potential p53 consensus binding sequences at positions 2008 to 1989, 1184 to
1165, and 1087 to 1068 from the ATG sequence and start site of translation,
and suggests that p53 induces GAPDH expression directly.
7 Jan 2005
13:59
272
AR
AR232-PA45-11.tex
CHUANG
HOUGH
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
SENATOROV
7 Jan 2005
13:59
AR
AR232-PA45-11.tex
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
273
translocation, and subsequent apoptosis; all three events are blocked by Bcl-2
overexpression (46). In human neuroblastoma cells, transfection-mediated Bcl-2
overexpression completely suppresses nuclear accumulation of GAPDH induced
by N-methyl-(R)-salsolinol (41). In R6 cells overexpressing Bcl-2, the nuclear
levels of endogenous GAPDH are markedly reduced compared with the wildtype control (42). These observations suggest that Bcl-2 may participate in the
regulation of GAPDH nuclear translocation and this effect may be part of the
protective mechanisms of Bcl-2 against apoptosis. However, Bcl-2 has no effect
on transfection-induced nuclear accumulation of GAPDH-GFP fusion protein in
R6 cells, indicating the complexity of the translocation process. The observation
that GAPDH first localizes to the Golgi before proceeding to the nucleus following
6-hydroxydopamine treatment in neuroblastoma cells may be an indication of how
GAPDH is transported to the nucleus (47).
It has been reported that GAPDH binds to a nuclear localization signal (NLS)containing protein, Siah, to initiate the GAPDH translocation to the nucleus (48).
GAPDH appears to stabilize Siah following their association, and thereby enhances
Siah-mediated proteolytic cleavage of its nuclear substrates, such as N-CoR, to
trigger apoptosis. Siah is also a target of p53 induction (49, 50) and can trigger
apoptosis either alone (49) or in conjunction with Pw1/Peg3 (51). In addition,
Siah acts as an E3 ubiquitin-ligase in the ubiquitination/proteasome degradation
pathway for Numb and DCC (52). Although GAPDH is known to interact with
microtubules and microfilaments (53) and binds to a variety of proteins linked to
neurodegenerative diseases (see below), there is no evidence yet supporting the
notion that these proteins mediate the transport of GAPDH to the nucleus. In NIH
3T3 fibroblast cells, serum depletion-induced GAPDH import to the nucleus is a
reversible process; readdition of serum or stimulation with growth factors causes
GAPDH export from the nucleus (54). The GAPDH nuclear export requires the
activity of phosphatidylinositol 3-kinase but is not mediated by exportin 1 (54, 55).
However, a novel exportin1-dependent nuclear export signal (NES) comprising 13
amino acids of the C-terminal domain of GAPDH has been identified (56). Truncation or mutation of this NES abrogates exportin1 binding and causes nuclear
accumulation of GAPDH-GFP fusion protein expressed in colorectal adenocarcinoma cells. This suggests that GAPDH would be a nuclear protein were it not for
its NES, and that the cytoplasmic localization of GAPDH is an active rather than
a passive process. Reversible GAPDH nuclear translocation following its overexpression has also been found in human monocytes infected with vaccinia virus
(57).
It appears that there is a change in the structure of the GAPDH protein following its nuclear translocation. For example, sodium nitropruside (an NO donor)
-induced NAD labeling of nuclear GAPDH is decreased by 60% in cerebellar
granule cells after AraC treatment (37), suggesting that the active site of GAPDH
may be covalently modified, denatured, or improperly folded. GAPDH is present
as a tetramer in the cytoplasm where the enzyme catalyzes its glycolytic activity. However, the uracil glycosylase activity of GAPDH is associated with the
7 Jan 2005
13:59
274
AR
AR232-PA45-11.tex
CHUANG
HOUGH
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
SENATOROV
7 Jan 2005
13:59
AR
AR232-PA45-11.tex
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
275
7 Jan 2005
13:59
276
AR
AR232-PA45-11.tex
CHUANG
HOUGH
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
SENATOROV
active site oxidation appear to be required for at least three nuclear functions of
GAPDH. This does not, however, exclude all mechanisms of nuclear translocation
involving active site oxidation of GAPDH, and an explanation and function for
GAPDH translocation to the nucleus during oxidative stress remains to be found.
The function of a sensor is to detect and signal changes in intracellular conditions. In addition to nuclear translocation, increased GAPDH gene expression
with oxidation of the GAPDH active site cysteine has been observed (80). Could
the oxidation of GAPDH Cys149 signal transcriptional activation of its own gene?
Increased NADPH levels resulting from inhibition of GAPDH enable thioredoxin
to become reduced, which in turn, through Ref-1, reduces a putative Cys275Cys135 bridge within p53, enhances DNA binding, and activates the transcription
of several proteins, including, presumably, GAPDH (31, 81). Reduced thioredoxin,
through Ref-1, also enhances the function of NF-B, AP-1, and HIF-1. It has long
been thought that p53 mediates oxidative stress-induced apoptosis (82). Recently,
however, it has become apparent that p53 acts through another protein, p66Shc,
to mediate apoptosis (83). Overexpression of p66Shc potentiates overexpressed
p53-mediated apoptosis in DLD-1 colorectal cancer cells, whereas overexpressed
p53 induces no apoptosis in p66Shc/ mouse embryonic fibroblast cells. Overexpressed p66Shc alone has little effect and hence acts downstream of p53. Under
high oxidative stress, p66Shc plays a permissive role in cytochrome c release and
collapse of the mitochondrial membrane potential (84). Exactly how p53 and/or
GAPDH interact with p66Shc is still unclear.
The one property of GAPDH that ties the multiple functions, locations, and
ligands of this protein with its putative role as a sensor of oxidative stress is
its capacity to bind NAD+. Direct induction of GAPDH expression by p53 suggests that GAPDH plays a role in the function of p53, but the details of this role
are unknown. Perhaps GAPDH provides NAD+ to the NAD+-dependent protein
deacetylase, Sirt1. The human homologue of yeast Sir2a, Sirt1, inhibits p53 by
removing the acetyl group from Lys382 in the C-terminal tail of p53 (85). This action results in resistance to stress and enhanced survival (86). Because irreversible
oxidation of GAPDH Cys149 dramatically reduces its affinity for NAD+, it could
prevent delivery of the cofactor to Sirt1 and result in unchecked p53 activity toward
apoptosis. This function could explain why GAPDH is translocated to the nucleus
under conditions of cellular stress.
7 Jan 2005
13:59
AR
AR232-PA45-11.tex
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
277
(coding for glutamine) in the gene of huntingtin, (reviewed in 87). The hallmarks
of apoptosis, such as DNA fragmentation, caspase activation, and induction of
apoptosis-related genes, have been found in brain neurons associated with the
deposits of -amyloid peptide in Alzheimers disease (AD) (for review, see 27).
Growing evidence also links Parkinsons disease (PD) with apoptotic death of
dopaminergic neurons in the substantia nigra (reviewed in 20, 27, 87). In brain
ischemia, the activation of caspases 1, 3, 8, 9, and 11 and apoptosis have been
reported in the ischemic penumbra, where hypoxia and energy depletion are not
as severe (reviewed in 8789). The participation of GAPDH in apoptosis in vitro
suggests significant roles in human pathophysiology. This section reviews literature reporting abnormal expression and intracellular localization of GAPDH and
implicating the involvement of GAPDH in the apoptosis and neurodegeneration
observed in HD, AD, PD, and ischemia.
7 Jan 2005
13:59
278
AR
AR232-PA45-11.tex
CHUANG
HOUGH
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
SENATOROV
7 Jan 2005
13:59
AR
AR232-PA45-11.tex
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
279
Alzheimers Disease
AD is the most common cause of progressive neurodegeneration and is characterized by dementia and memory loss. The deposition of -amyloid protein in
the cerebral cortex and hippocampus is one of the most distinct morphological
features in AD (reviewed in 106). -Amyloid peptide derives from irregular proteolytic processing of -amyloid precursor protein (-APP). The deposition of
-amyloid peptide is associated with neuronal loss in these brain regions and is,
at least in part, due to apoptosis. Initial in vitro studies found GAPDH binding to
the cytoplasmic carboxyl terminal of -amyloid precursor protein (-APP) (107).
Such binding could alter the normal processing of -APP to produce -amyloid
protein. A significant nuclear role for the C terminus of -APP has also been suggested (108, 109). The recognition of GAPDH by a monoclonal antibody raised
against amyloid plaques from the brains of patients with AD indicates the presence of GAPDH in these plaques and suggests an interaction between GAPDH and
-APP (110, 111). A more recent preliminary report suggests that cotransfection
of COS-7 cells with GAPDH and wild-type -APP cDNAs induces synergistic
cytotoxicity (112).
The initial postmortem study showed a significant (19%) reduction in GAPDH
glycolytic activity in the homogenates of the temporal cortex of AD patients (94).
However, a subsequent study failed to detect a change in GAPDH activity in homogenates of the frontal, temporal, parietal, and occipital lobes of AD brains,
although there was a significant increase in activity in the same brain regions in
Downs syndrome patients (113). Such a discrepancy could be related to the fact
that whole-cell homogenates, rather than subcellular fractions, were used in the
activity measurements. The presence of multiple cell types in a given brain region might have also masked potential changes in a specific cell population. Using
fibroblasts from AD patients, it has been reported that GAPDH glycolytic activity is decreased by approximately 27% in both postnuclear and nuclear fractions
compared with age-matched controls (58, 96). A high-molecular-weight species
of GAPDH-immunoreactivity was detected exclusively in the postnuclear fraction
of AD fibroblasts (58). The latter displayed reduced GAPDH activity and was
not present in postnuclear fractions from control subjects. Whether the shift in
molecular weight reflects GAPDH binding to -APP is unknown. Interestingly,
the association of GAPDH with a high-molecular-weight species was not detected
in sonicated AD whole-cell extracts, which exhibited normal levels of GAPDH
activity. This suggests that GAPDH is weakly bound to the high-molecular-weight
protein complex and that dissociation of GAPDH from the high-molecular-weight
complex restores its glycolytic activity. In the postmortem AD brain, nuclear aggregated GAPDH in neurons of affected areas has been found (114). It is also
noteworthy that potential drugs for treating AD, such as tetrahydroaminoacridine
(THA) and ONO-1603, effectively suppress GAPDH overexpression and nuclear
translocation in rat brain neurons undergoing apoptosis in cultures (115, 116).
Moreover, THA and another antidementia drug, donepezil, inhibit AraC-induced
increase in GAPDH promoter activity (114).
7 Jan 2005
13:59
280
AR
AR232-PA45-11.tex
CHUANG
HOUGH
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
SENATOROV
Parkinsons Disease
PD is characterized by the loss of catecholaminergic neurons, including those in
the substantia nigra compacta and noradrenergic neurons in the locus coeruleus.
The first implication that GAPDH is involved in the pathogenesis of PD is from the
observation that CGP 3466 (also known as TCH 346), a structurally related analog
of R-(-)-deprenyl, protects human neuroblastoma PAJU cells from apoptosis induced by rotenone, a toxin producing PD-like neuropathology (44). The MAO-B
inhibitor, (-)-deprenyl, slows neurodegeneration and reduces the clinical deficits
of PD, HD, and AD. Unlike (-)-deprenyl, CGP 3466 does not inhibit MAO-B activity, but like (-)-deprenyl, binds to GAPDH. These results are compatible with
the view that GAPDH, rather than MAO-B, is the target of deprenyl-like compounds effective against PD neuropathology. An independent study suggested that
deprenyl-like compounds inhibit apoptosis by inducing GAPDH to dissociate from
its usual tetrameric form to a dimer, and thereby interfere with GAPDH nuclear
translocation (117). Nuclear localization of GAPDH monomers and dimmers were
readily detected in HeLa cells, however, following GAPDH active site oxidation
(76). R-2HMP (R-2-heptylmethyl-pargylamine) and other aliphatic pargylamines
bind GAPDH, prevent GAPDH overexpression, and block p53-dependent apoptosis (16, 17, 118). Other studies show that apoptosis of neurons or related cell-lines
induced by dopaminergic toxins such as MPP+, 6-hydroxydopamine, or N-methyl(R)-salsolinol involves GAPDH overexpression and nuclear translocation, and
these effects are prevented by GAPDH antisense oligonucleotides and anti-PD
drugs (3941, 47). In total, these observations suggest that GAPDH is a potential
molecular target of drugs used to treat PD and other neurodegenerative diseases.
Despite the number of cell culture studies of the role of GAPDH in apoptosis,
knowledge concerning GAPDH changes in the PD brain is limited. This may
be because there is still no clear consensus that apoptosis contributes to the loss
of dopaminergic neurons in PD. Interestingly, an accumulation of GAPDH was
found in the nuclei of melanized neurons of the nigra in postmortem brain sections
from PD patients, whereas GAPDH was found only in the cytoplasm of melanized
cells of age-matched control sections (119). Nuclear inclusion bodies, known as
Marinescos bodies, are immunoreactive for GAPDH in numerous nigral neurons
from PD, but not control brains. Many cytoplasmic inclusion bodies, known as
Lewy bodies, in the melanized neurons of PD brain are also immunopositive for
GAPDH, although it is unknown whether all Lewy bodies are immunopositive.
Moreover, GAPDH appears to be colocalized with Bax and caspase3 in melanized
neurons of the PD nigra. Although a direct link between PD and GAPDH-mediated
apoptosis is still undetermined, these results suggest a potential role of GAPDH
nuclear accumulation in dopaminergic cell death in the PD brain.
7 Jan 2005
13:59
AR
AR232-PA45-11.tex
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
281
increase in extracellular glutamate in the brain following ischemia and the brain
damage that occurs as a result are well recognized (for review, see 87, 89, 120).
Few studies address the involvement of GAPDH in the ischemic brain. In a rat focal ischemia stroke model, nuclear accumulation of GAPDH has been found in the
ischemic core of the parietal cortex of rats subjected to 2 h of middle cerebral artery
occlusion without reperfusion (121). During subsequent reperfusion, GAPDH immunostaining in the ischemic core decreases but cytoplasmic and nuclear staining
in the penumbra becomes detectable. The increase in nuclear GAPDH immunoreactivity persists up to 48 h with a concomitant decrease in cytoplasmic reactivity.
Double labeling of GAPDH-positive cells with TUNEL suggests an association of
GAPDH overexpression/nuclear accumulation with excitotoxicity-induced apoptotic death. Cell culture studies have provided insights into mechanisms by which
GAPDH is induced by hypoxia. A pioneering study has shown that hypoxia stimulates GAPDH overexpression in the cytoplasm and nucleus of endothelial cells
at both transcriptional and posttranscriptional levels (122). Subsequent studies
identified at least two hypoxia-responsive elements in the GAPDH gene promoter
(123, 124). Hypoxia induces an elevation of GAPDH protein in the cytosolic, nuclear, and particulate fractions by 4.0-, 2.3-, and 4.2-fold, respectively, in a mouse
brain capillary endothelial cell line (125). Little or no increase in GAPDH glycolytic activity was found in these fractions, however, suggesting a dynamic steady
state or an inactivation of newly induced GAPDH. The same study also shows that
GAPDH expression is suppressed by inhibiting the activation of nonselective Ca2+
channels, Na+/Ca2+ exchanger, Ca2+/calmodulin-dependent kinase, and c-Jundependent AP-1 binding. GAPDH has been shown to interact directly with heat
shock proteins (HSP), and in particular, with HSP70 (61, 126). HSP70 has a major
role in protection against ischemia-induced brain damage (88, 127, 128). Therefore, if GAPDH is involved in the apoptotic death induced by ischemia/hypoxia,
then GAPDH binding-induced inactivation of cytoprotective HSP70 and nuclear
translocation may be involved as well.
7 Jan 2005
13:59
282
AR
AR232-PA45-11.tex
CHUANG
HOUGH
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
SENATOROV
Androgen deprivation
AD
PD
7 Jan 2005
13:59
AR
AR232-PA45-11.tex
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
283
LITERATURE CITED
1. Jeffery CJ. 1999. Moonlighting proteins.
Trends Biochemical Sci. 24:811
2. Sirover MA. 1999. New insights into an
old protein: the functional diversity of
mammalian glyceraldehyde-3-phosphate
dehydrogenase. Biochim. Biophys. Acta
1432:15984
3. Alderete JF, Millsap KW, Lehker MW,
Benchimol M. 2001. Enzymes on microbial pathogens and Trichomonas vaginalis: molecular mimicry and functional
diversity. Cell. Microbiol. 3:35970
4. Huitorel P, Pantaloni D. 1985. Bundling
of microtubules by glyceraldehyde-3phosphate dehydrogenase and its modulation by ATP. Eur. J. Biochem. 150:26569
5. Fuchtbauer A, Jockusch BM, Leberer
7 Jan 2005
13:59
284
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
AR
AR232-PA45-11.tex
CHUANG
HOUGH
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
SENATOROV
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
7 Jan 2005
13:59
AR
AR232-PA45-11.tex
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
30.
31.
32.
33.
34.
35.
36.
37.
An antisense oligodeoxynucleotide to
glyceraldehyde-3-phosphate dehydrogenase blocks age-induced apoptosis of mature cerebrocortical neurons in culture. J.
Pharmacol. Exp. Ther. 278:44754
Ishitani R, Sunaga K, Tanaka M, Aishita
H. Chuang D-M. 1997. Over-expression
of glyceraldehyde-3-phosphate dehydrogenase is involved in low K+-induced
apoptosis, but not necrosis, of cultured
cerebellar granule cells. Mol. Pharmacol.
51:54250
Chen R-W, Saunders PA, Wei H, Li Z,
Seth P, Chuang D-M. 1999. Involvement
of GAPDH and p53 in neuronal apoptosis: evidence that GAPDH is upregulated
by p53. J. Neurosci. 19:965462
Ishitani R, Chuang D-M. 1996. Glyceraldehyde-3-phosphate dehydrogenase
antisense oligodeoxynucleotides protect
against cytosine arabinoside-induced
apoptosis in cultured cerebellar neurons.
Proc. Natl. Acad. Sci. USA 93:993741
Saunders PA, Chalecka-Franaszek E,
Chuang D-M. 1997. Subcellular distribution of GAPDH in cerebellar granule cells
undergoing cytosine arabinoside-induced
apoptosis. J. Neurochem. 69:182028
Sawa A, Khan AA, Hester LD, Snyder
SH. 1997. Glyceraldehyde-3-phosphate
dehydrogenase: nuclear translocation participates in neuronal and nonneuronal cell
death. Proc. Natl. Acad. Sci. USA 94:
1166974
Ishitani R, Tanaka M, Sunaga K, Katsube
N, Chuang D-M. 1998. Nuclear localization of over-expressed glyceraldehyde3-phosphate dehydrogenase in cultured
cerebellar neurons undergoing apoptosis.
Mol. Pharmacol. 53:7017
Shashidharan P, Chalmers-Redman RME,
Carlile GW, Rodic V, Gurvich N, et al.
1999. Nuclear translocation of GAPDHGFP fusion protein during apoptosis. NeuroReport 10:114953
Saunders PA, Chen R-W, Chuang D-M.
1999. Nuclear translocation of glyceraldehyde-3-phosphate dehydrogenase iso-
38.
39.
40.
41.
42.
43.
44.
45.
285
7 Jan 2005
13:59
286
46.
47.
48.
49.
50.
51.
52.
AR
AR232-PA45-11.tex
CHUANG
HOUGH
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
SENATOROV
7 Jan 2005
13:59
AR
AR232-PA45-11.tex
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
287
7 Jan 2005
13:59
288
79.
80.
81.
82.
83.
84.
85.
86.
87.
AR
AR232-PA45-11.tex
CHUANG
HOUGH
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
SENATOROV
88.
89.
90.
91.
92.
93.
94.
95.
96.
7 Jan 2005
13:59
AR
AR232-PA45-11.tex
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
97.
98.
99.
100.
101.
102.
103.
104.
105.
106.
107.
108.
109.
110.
111.
112.
289
7 Jan 2005
13:59
290
113.
114.
115.
116.
117.
118.
119.
120.
121.
AR
AR232-PA45-11.tex
CHUANG
HOUGH
AR232-PA45-11.sgm
LaTeX2e(2002/01/18)
P1: GCE
SENATOROV
122.
123.
124.
125.
126.
127.
128.
HI-RES-PA45-11-Chuang.qxd
1/7/05
2:25 PM
Page 1
C-1
4 Dec 2004
13:28
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
P1: GCE
LaTeX2e(2002/01/18)
10.1146/annurev.pharmtox.45.120403.100004
NON-MICHAELIS-MENTEN KINETICS IN
CYTOCHROME P450-CATALYZED REACTIONS
William M. Atkins
Department of Medicinal Chemistry, University of Washington, Seattle,
Washington 98195-7610; email: winky@u.washington.edu
OVERVIEW
The cytochrome P450 monooxygenases (CYPs) are ubiquitous heme-containing
enzymes that catalyze an immense range of chemical reactions in prokaryotes,
plants, and animals (1, 2). CYPs participate in the biosynthesis of hormones,
second messengers, and other natural products. CYPs also dominate xenobiotic
detoxification and human drug metabolism. As a result, CYPs are of primary
importance in the pharmaceutical industry (36). In fact, characterization of the
interactions between new drugs and human CYPs is now a routine component of
early drug development. An enigmatic behavioral characteristic of CYPs, which
has only recently been appreciated fully, is their tendency to exhibit atypical
steady-state kinetic patterns in vitro, and possibly in vivo. In fact, several excellent
recent reviews have focused on this atypical behavior, also referred to as allosterism, thus highlighting its perceived importance (711). This review explores some
recent observations, while minimizing duplication with the previous reviews, and
it considers mechanistic aspects of the atypical kinetics in the context of recently
determined X-ray structures.
From a historical perspective, it is interesting that nonhyperbolic CYP kinetics
were documented as early as the 1980s (1214), but this received little attention.
0362-1642/05/0210-0291$14.00
291
4 Dec 2004
13:28
292
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
ATKINS
Subsequently, Korzekwa et al. (15, 16) provided thoughtful accounts of the relationship between atypical kinetics observed with CYP3A4 and the possibility that
multiple ligands could occupy the active site simultaneously. Today, it is widely
accepted that several CYP isoforms, including 3A4, 1A2, 2E1, 2D6, and 2C9, can
exhibit nonartefactual atypical kinetics in vitro. Furthermore, it is highly likely that
the kinetic behavior is related in some cases to simultaneous binding of multiple
ligands to a single active site, as elaborated here. Also, it is notable that experimental evidence for multiple ligand binding to CYP101 (P450cam) was provided
by Sligar and coworkers as early as 1994, based on NMR approaches (17).
In contrast to the widespread acceptance of this behavior in vitro, examples
of in vivo kinetics that deviate from Michaelis-Menten kinetics are sparse. Examples include interactions between diclofenac and quinidine in monkeys (18),
carbamezepine and felbamate in humans (19), and a marginal effect between flurbiprofen and dapsone in humans (20). Although examples of in vivo allosteric CYP
interactions are limited, they are likely to become more widespread as awareness of
their possibility increases and with improved analytical methods. Regardless, the
apparent universality of allosteric effects across several CYP isoforms and many
drugs in vitro (2126) demands a mechanistic understanding that could dramatically enhance in vitro predictability of drug-drug interactions. Presumably, this
understanding would translate directly into increased predictive power in vivo.
The behavior of CYPs also is extremely important from an academic perspective
because it demands significant revision of the paradigms of traditional allosteric
enzymes. Nearly all allosteric proteins are multisubunit oligomers (2729). Moreover, allosteric behavior of normal enzymes can be rationalized within their biological niche as a mechanism for achieving metabolic control through highly
specific molecular recognition. In contrast, although CYPs may sample several
aggregation states (30, 31), they can exhibit non-Michaelis-Menten kinetics under conditions in which they are predominantly monomeric. Although CYP-CYP,
CYP-reductase, and CYP-Cyt b5 interactions may provide an additional mechanism by which allosteric effects occur, they are considered only briefly here.
Also, according to traditional paradigms, allostery requires specificity. However, as detoxification enzymes, CYPs do not utilize specific molecular recognition. Rather, they are extraordinarily substrate diverse. The resulting nonspecific
allosterism is also of academic interest because it deviates from well-understood
allosterism of substrate-specific enzymes. It is not clear what biological advantage,
if any, is gained from the allosterism of CYPs, wherein some toxic substrates are
metabolized more efficiently and others less efficiently in the presence of allosteric
effectors. Both the mechanism and the biological purpose of CYP allosterism are
challenging (32).
4 Dec 2004
13:28
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
293
types of CYP allosterism are summarized below. Throughout this review, the term
allosterism is used interchangeably with atypical kinetics because both require
formation of a ternary complex, [CYPSS] or [CYPSE], where S and E are
substrate and effector, respectively, and these complexes have kinetic properties
that differ from [CYPS].
The implication of nonhyperbolic kinetics is that the Michaelis-Menten steady
state model is insufficient to describe drug clearancc, CLint. The Michaelis-Menten
model describes the velocity of product formation, v, as
v=
Vmax K M
,
[S] + K M
where Vmax and KM have their usual meanings. When the Michaelis-Menten relationship does apply, the clearance of a drug may accurately be estimated, in
principle, from the Vmax/KM. This parameter, approaches the intrinsic drug clearance (Clint = v/[S]) or the slope of a hyperbolic Michaelis-Menten plot at low
[S]. Furthermore, the in vitro clearance is frequently used to estimate in vivo clearance, after appropriate scaling for the CYP capacity of the liver or other tissue.
Obviously, the accuracy of the in vivo prediction is limited by the accuracy of the
model used to extract metabolic velocities from the in vitro data (6, 9).
Heterotropic Effects
Alternatively, heterotropic effects occur when one drug alters the CYP interactions
with a second drug, either activating or inhibiting the rate of product formation
(33, 34). Here, the drug acting as substrate may yield classic hyperbolic velocity
versus [substrate] curves, but the second drug changes the parameters Vmax or
4 Dec 2004
13:28
294
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
ATKINS
Figure 1 Velocity versus [substrate] plots depicting possible kinetic profiles with
homotropic effects. Top left: hyperbolic kinetics with no allosterism. Bottom left:
sigmoidal kinetics resulting from homotropic activation. Top right: biphasic kinetics
resulting from a low-affinity second ligand site. Bottom right: substrate inhibition,
wherein binding of the second substrate decreases Vmax. In each case, the inset depicts
an Eadie-Hoffstee plot (V versus V/[S]) corresponding to the velocity curves.
4 Dec 2004
13:28
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
295
second drug cleared by the same CYP isoform. Equally important, a single effector molecule may change from an activator at low concentration to an inhibitor
at higher concentration. Thus, the behavior of any effector compound depends
on the substrate that is being metabolized, as well as on the concentrations of
effector and substrate (35, 36). For example, testosterone inhibits with different
apparent potencies the metabolism of terfenadine and midazolam by CYP3A4.
In contrast, testosterone does not inhibit metabolism of nifedipine, but terfandine
does. Moreover, testosterone itself is a substrate for CYP3A4, and its metabolism is
partially inhibited by nifedipine (3537). Similar nonreciprocal effects have been
observed with CYP3A4-dependent interactions between -napthoflavone (-NF)
and aflatoxin B1. The -NF activates metabolism of the aflatoxin, but the latter
has no effect on the metabolism of aflatoxin B1 (38). Houston and coworkers have
initiated the categorization of various CYP3A4 substrates into subgroups based
on kinetic traits and heterotropic effects in which they participate (37). Clearly,
the behavior of any substrate or inhibitor depends on what other compounds are
present, and this is a major challenge for describing CYP allosterism. Moreover,
the heterotropic effects of any ligand pair are CYP isoform dependent. For example, the highly homologous CYPs 3A4 and 3A5 exhibit different heterotropic
interactions for several ligand pairs (39).
At least two molecular mechanisms may contribute to context-dependent ligand
effects. The first is ligand-dependent conformational change, wherein the enzyme
is sufficiently flexible that each combination of ligands induces a different enzyme conformation with different kinetic properties. This contrasts the case with
substrate-specific enzymes in which only a few specific conformations are coupled
to a few specific ligands (27, 28). If a wide range of ligand-dependent conformational space is available to the enzyme, this will promote context-dependent ligand
effects (40).
Based on flash photolysis and CO recombination experiments, it was proposed
that slowly equilibrating conformations of a single CYP isoform could differentially interact with ligands (4143). This possibility has been reconsidered based on
studies using hydrostatic pressure (44). Such persistent conformations could cause
allosteric kinetics, even in the absence of multiple ligand binding, just as mixtures
of isoforms can yield non-Michaeles-Menten kinetics. In contrast, ligand-induced
conformational changes, in the absence of nonequilibrating conformational states,
cannot cause allosteric kinetics. In the absence of persistent conformations with different properties, ligand-dependent conformational change is neither a necessary
nor sufficient condition for allosteric kinetics. Multiple ligand binding, however, is
a necessary but not a sufficient condition for allosterism. Conformational change
provides one mechanism by which multiple ligand binding can yield complex
kinetics (40).
Conformational changes induced by nonactive site ligands may also contribute.
For example, Schrag & Wienkers (45) found that addition of Mg2+ to CYP3A4
incubations with the substrate pyrene resulted in the conversion from positive
homotropic kinetics to hyperbolic patterns, and this correlated with a change in
4 Dec 2004
13:28
296
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
ATKINS
4 Dec 2004
13:28
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
297
Scheme 1 depicts the simplest generic model for homotropic effects, which has
been used by numerous investigators. Here S is substrate; P is product; Ks is the
affinity of the CYP for substrate; kcat is the rate of formation of product from the
[CYPS] complex; and Vmax is defined as 2kcat/[E]t, where [E]t is the total enzyme
concentration. The equation describing the fraction of maximal velocity at any
[S] is
Scheme 1
In this model, the substrate can bind in either of two sites, as indicated by
[SCYP] versus [CYPS], and these complexes have identical dissociation constants for substrate, KS, and identical kcat values for product formation. The widespread use of the two-site model in Scheme 1 in the CYP literature, or variations
of it, reflects the popular belief that at low occupancy, the bound ligand is localized in a discrete binding site, rather than sampling all parts of the active site,
i.e., that [CYPS] and [SCYP] are two different molecular species that can only
interconvert via substrate dissociation and rebinding, but neither is preferred thermodynamically. Regardless, binding of the second substrate leads to a complex
[CYPSS] with different kinetic properties, KS and kcat. Here, is the effect
that the first substrate has on the KS for the second substrate, and is the effect
that the presence of the first substrate has on the kcat for the second. Thus when
either < 1 or > 1, positive homotropic cooperativity may be apparent and
velocity curves will be sigmoidal. Alternatively, if > 1 the curves may appear
biphasic, and if < 1 substrate inhibition will be evident. The detailed shape of
the corresponding velocity versus [S] plot will be determined by KS, kcat, , and
. Although this model has been extremely useful for conceptualizing homotropic
allosteric kinetics for CYPs, it is inherently oversimplified because of the kinetic
equivalence of [SCYP] and [CYPS], which form with equal apparent affinities
4 Dec 2004
13:28
298
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
ATKINS
and generate product at equivalent velocities. Such kinetic symmetry is useful for
reducing the complexity of the system, but in the absence of any structural symmetry of the CYP enzymes, it is likely to be an inaccurate depiction of what really
occurs at the molecular level. This model is more suitable for normal allosteric
enzymes with multiple copies of identical active sites.
A more likely scenario, possibly, is that multiple [CYPS] complexes are
formed, with multiple orientations of S in rapid equilibrium, [SCYP] and [CYPS],
which form with different affinities and different kcat values associated with them,
as in Scheme 2. In this case, the system behaves like a mixture of enzyme-substrate
complexes, with the fractional contribution of [SCYP] versus [CYPS] determined by Ks1/Ks2, and with the reaction velocity equation shown. Here Vmax1 =
kcat[ET], Vmax2 = kcat[ET], and [ET] is the total enzyme concentration, and , ,
and are scaling factors that modulate the KS1, KS2, and kcat , respectively. The
parameters Vmax1 and Vmax2 are virtual parameters that represent the rate of product
formation if all of the enzyme could be forced into the [CYPS] or the [SCYP]
states; however, this cannot actually occur. Note that the number of fitting parameters has increased to eight (Ks1, Ks2, kcat, kcat , , , , ). This model was used
recently to explore the metabolism of verapamil by CYP3A4 (61). It was found
that formation of several metabolites could be described by Scheme 2, wherein
negative cooperativity was associated with and values less than 1 and and
values greater than 1.
Scheme 2
Comparison of Schemes 1 and 2 reveals the compromise that must accompany a choice between models. The model in Scheme 1 suffers from potentially
unrealistic features, such as the existence on a single unsymmetrical CYP enzyme
4 Dec 2004
13:28
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
299
Scheme 3
The model in Scheme 3 accounts for multiple substrate binding with homotropic
effects, heterotropic binding, and multiple effector (E) binding with homotropic
4 Dec 2004
13:28
300
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
ATKINS
effects. Differential affinity of both substrate and effector is caused by the presence of another effector or inhibitor, and differential effects on kcat are allowed.
This model, and variations, have been applied to examine context-dependent heterotropic effects (7, 36, 37, 64, 65).
The steady-state kinetics provide a powerful tool for conceptualizing the possible mechanisms responsible for the variety of atypical kinetics observed, and
they accurately predict metabolism rates. The examples described above demonstrate the necessary compromise between a level of complexity adequate to describe the experimental data and the need to use many data points to avoid
overparameterization.
4 Dec 2004
13:28
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
301
cooperativity or negative cooperativity at the level of ligand binding, per se, and
positive cooperitivity on the ligand-dependent spin state shift (69, 70). It is unclear
how these results relate mechanistically to changes in the Ks parameter used in
kinetic models, as described above.
Several conceptual models that are based on multiple ligand binding to a single
active site remain viable. It is possible that there are discrete and static subsites
within the large active site, and each subsite has its own personality. Each subsite may have a characteristic affinity for each ligand and hold it in a preferred
orientation, which is static on the timescale of oxidative turnover (66, 68, 71,
72). In this extreme case, multiple ligands bind sequentially to the highest affinity available subsite and then to the lowest affinity site. From their respective
binding sites, ligands may alter the metabolism of other ligands by inducing conformational changes, causing minor shifts in the distances between oxidizable
sites on the drug and the heme iron-oxo species, or by altering relative uncoupling rates to nonproductive formation of superoxide. Support for discrete static
subsites has come, partly, from mutagenesis studies as championed by Halpert
and coworkers (6972). For example, midazolam appears to be an example of
this type of ligand wherein distinct subsites within the active site are responsible for the formation of the 1 -hydroxy- versus 4-hydroxy-midazolam products.
The results with midazolam support the unlikely suitability of Scheme 1 as a
model; the different binding subsites for midazolam, if they exist, are proposed
to have different Vmaxs and substrate affinities, and to even generate different
products.
Alternatively, the large active site may be fluid and multiple bound ligands may
sample several subsites within the large active site, either dynamically or through
a static heterogeneity. Evidence for a fluid active site has come mainly from kinetic deuterium isotope effects. Trager and coworkers have provided numerous
examples, and appropriate theory, to understand CYP substrates as moving within
the active site and presenting several points of oxidation on a single molecule to
the [FeO]3+ intermediate (7375). In effect, they have varied distances between
deuterium- and hydrogen-bearing benzylic methyl groups on ring systems of increasing size. The substrate size could be correlated to the extent of masking
of the isotope effect (kH/kD), as expected if smaller substrates rapidly reorient
within the active site. In principle, this approach could be used to determine the
effect of multiple ligand binding on substrate dynamics. Both homotropic and
heterotropic effects should modulate the magnitude of the observed isotope effects if they change the effective size of the active site. In fact, deuterium isotope
effects have already been used to observe multiple ligand binding to CYP BM3
(CYP102), wherein deuteration of laurate caused a change in the regioselectivity
of hydroxylation of palmitate (76). This could only occur if both ligands bound
simultaneously. As with the kinetic modeling, the mechanism of multiple ligand
binding within CYP active sites may be context-dependent. Some ligands may
dynamically or statically sample several parts of the active site, whereas others
may occupy well-defined subsites.
4 Dec 2004
13:28
302
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
ATKINS
An additional mechanistic complexity arises from the possibility that uncoupling may occur. Many [CYPsubstrate] complexes generate superoxide anion,
hydrogen peroxide, or water as the reduction products of O2 at the expense of
substrate oxidation (7779). As competing processes, these pathways decrease
the apparent kcat for product (oxidized substrate) formation. Hutzler and coworkers recently demonstrated that the branching ratios between substrate oxidation
and uncoupling could be altered by the addition of an effector (80). Specifically,
dapsone caused a decrease in the uncoupling of [CYP2C9flurbiporfen], thus explaining the activation by dapsone of flurbiprofen metabolism. Allosteric effects
on coupling are likely to be common.
Recent elegant strategies, and nearly heroic efforts, have led to the successful
solubilization of several mammalian CYP isoforms by engineering the membrane
binding regions (8185). Specifically, the N-terminal membrane anchor has been
partially truncated, and the F-G region, thought to be a peripheral membranebinding patch, has been mutated or chimerized in several ways. The resulting soluble proteins have been crystallized and they have afforded X-ray structural models.
The structures provide an obvious tool to look for mechanistic clues concerning
the atypical kinetic behavior described above, so they are briefly discussed here.
First, it is useful to highlight an important relevant aspect of the crystal structure of the bacterial CYPeryF as it relates to atypical kinetics. Cupp-Vickery
and coworkers provided a crystal structure of CYPeryF complexed with either
androstenedione or 9-amino-phenanthrene (57). For both, clear electron density
revealed the simultaneous presence of two molecules in the active site cavity proximal to the heme. Interestingly, for both complexes, direct ligand-ligand interactions
were observed, suggesting a possible contribution to positive homotropic cooperativity as noted above. Also, for both cases, only one of the bound ligands appears to
be in a location that would allow metabolism (of course, the 9-aminophenanthrene
forms a 6-coordiante nitrogen-liganded complex that is not expected to be metabolized, but if the exocyclic amine were not present, only one of the phenanthrene
rings would be sterically accessible to the heme iron-oxo complex). In short, the
structures demonstrated for the first time the possibility that two ligands could
simultaneously occupy a CYP active site, although only one could be a target for
oxidation with reorientation on the timescale of turnover.
The first X-ray structure of a mammalian CYP was that of rabbit CYP2C5 (86).
Perhaps the single most important conclusion resulting from this structure was
that mammalian CYPs are structural homologs of the bacterial CYPs, for which a
wealth of structural data already exists (8791). This was not a surprising conclusion, but its experimental validation was comforting and important. Subsequent
structures of the engineered rabbit CYP2C5 have allowed for a comparison of
ligand-free enzyme with complexes of diclofenac (92) or a benzenesulfonamide
(DPZ) derivative (93). This comparison reveals the likely ligand-dependent conformations of the protein in the B -C loop and the F-G loop, and it has prompted
the use of induced fit models for discussing CYP dynamics. These results extend
to the mammalian CYPs the notion that ligand binding alters the conformational
dynamics, particularly in these regions, as expected from the bacterial structures.
4 Dec 2004
13:28
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
303
In addition, the complex with DPZ suggests the possibility that the substrate binds
in two different orientations, with each providing electron density in partially occupied complexes. Interestingly, neither orientation, including the one with substrate
several angstroms from the heme iron and expected to yield less product, revealed
water bound to the iron. Thus, the crystallographic models are at odds with solution
spectroscopy, wherein addition of diclofenac or DPZ to 2C5 does not induce a high
spin transition (92, 93). Crystallographic evidence for multiple binding orientations
of a single substrate has also been provided with a [CYP101 nicotine complex]
(94). Here the major substrate orientation is nonproductive, with a substrate-heme
coordinate bond. Upon reduction and stabilization with CO, the nicotine orientation changes to a productive one, consistent with the metabolism of nicotine. This
clearly demonstrates the complexity of simple ligand binding with CYPs.
A structure of human CYP2C9 was recently solved, and reveals a striking behavior that is particularly relevant to CYP allosterism (95). The [CYP2C9warfarin]
complex positions warfarin in a corner of the active site, far from the heme iron, and
in an orientation inconsistent with the experimentally established regiospecificity
of warfarin hydroxylation (96). Based on this complex, the authors performed
docking experiments to demonstrate that there is ample room within the active
site for two ligands, suggesting that productive and nonproductive binding modes
may be available for any substrate, and that the relative population of these modes
will change with single occupancy versus multiple occupancy, i.e., [CYPS] versus [CYPSS]. For example, it is tempting to speculate that the first ligand can
merely take up space, without being a good target for oxidation, as suggested by
the crystal structure. Williams et al. note (see 95 for supporting information) that
in their attempt to obtain a ligand-free structure they observed undefined electron
density in the active site directly adjacent to the heme iron. Although they were
unable to identify the species yielding this density, it demonstrates that this part
of the active site is accessible to ligands, as required for metabolism. This amplifies
the possible preference of warfarin to not bind near the heme iron. In this case,
the bound warfarin may only occasionally sample portions of the active site closer
to the heme iron. However, at higher occupancy [CYPSS], the second ligand is
forced into more productive binding modes, thus providing a structural model for
positive homotropic or positive heterotropic effects. However, it should be emphasized that warfarin does not demonstrate atypical kinetics when metabolized by
CYP2C9, so this intuitive model is either incorrect or oversimplified.
A structure of CYP2B4 provides evidence for the possible contribution of conformational dynamics in atypical kinetics (97). Owing to structural rearrangements
in the B region and the F-G loops, an open conformation is captured in the crystal
state and stabilized by dimerization with a second CYP2B4. In fact, the cleft is
sufficiently pronounced to allow heme ligation by His-226 of the other CYP2B4
molecule of the dimer. Importantly, evidence for this dimer existing in solution
as well is presented. Comparison with the CYP2C5 structures suggests a range
of conformations in the B -C and F-G regions, including a significantly altered
conformation with a very large solvent-exposed crevice above the heme. Speculatively, the two structures, CYP2B4 and CYP2C5, may provide benchmarks for
4 Dec 2004
13:28
304
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
ATKINS
the range of conformations accessible to any single isoform and underscore the
extensive flexibility of the protein in these regions.
No structure of mammalian CYPs has revealed two ligands simultaneously
bound at the active site, so these structures have not provided any direct clues
about the mechanisms of atypical kinetics. In fact, a recent structure of human
2C8 reveals a nonsubstrate palmitic acid binding site that is peripheral to the active site, which could modulate catalytic properties (98). The apparent fatty acid
binding site includes determinants within the F , G , and G helices, which are
contiguous with ligand-sensitive regions of other isoforms, and this site communicates with substrate binding regions. At one level, this may be taken as evidence
for a true allosteric binding site remote from the heme iron and the active site
per se. However, without significant rearrangement, it is not obvious that this
site could accommodate hydrophobic drugs, and it is unlikely to be responsible
for the heterotropic effects discussed above. Rather, it supports the importance
of nonsubstrate ligand-dependent conformational effects (4754). Most recently,
crystal structures of CYP3A4 have been solved, and they further complicate the
existing paradigms (99, 100). Specifically, a structure of CYP3A4 complexed with
progesterone indicates that this ligand also binds at a site remote from the active
site, which suggests a separate allosteric site (99). Interestingly, with either progesterone at this remote site or with metyrapone coordinated to the heme iron,
no dramatic ligand-induced conformational changes are evident, compared to the
ligand-free CYP3A4 (99). However, the dimensions of the active site cavity are
significantly greater for CYP3A4 than the 2C isoforms in the immediate vicinity of
the heme (100). Thus, the possibility of mulitple ligand binding within the active
site remains. The available structures have not proven the central assumption of
current models for mammalian CYP allosterism: multiple ligand binding within a
single active site. However, collectively the available structures provide fundamentally important insights. For example, the presence of warfarin in a nonproductive
binding mode that limits the space available for a second ligand on CYP2C9, if
it binds, clearly demonstrates that each ligand can present new surfaces and handles to a second ligand, and direct ligand-ligand interactions can contribute, as
suggested for pyrene binding to CYP3A4 (59).
CONCLUSIONS
Atypical steady-state kinetics are now commonly observed among CYPs directly
involved in xenobiotic and drug metabolism for a wide range of drug structures.
In the past few years, the notion that multiple ligands bind within a single active
site of mammalian CYPs has evolved from an interesting speculation to a likely
possibility for many CYP-drug combinations. Of the experimental approaches used
to understand complex CYP kinetics, including kinetic modeling, crystallography,
and spectroscopic approaches, none alone are likely to reveal the mechanism of
CYP allosterism. Rather, there are likely to be multiple mechanisms spanning
different combinations of CYP isoform, substrate, and effector. An understanding
4 Dec 2004
13:28
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
305
of both ligand and protein dynamics will be necessary to fully understand CYP
kinetics. The combination of these approaches may be required to learn any general
rules of CYP allosterism, if they exist.
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Guengerich FP. 2002. Cytochrome P450
enzymes in the generation of commercial
products. Nat. Rev. Drug Discov. 1:359
66
2. Coon MJ, Ding XX, Pernecky SJ, Vaz
AD. 1992. Cytochrome P450: progress
and predictions. FASEB J. 6:66973
3. Wienkers LC. 2001. Problems associated
with in vitro assessment of drug inhibition
of CYP3A4 and other P-450 enzymes and
its impact on drug discovery. J. Pharmacol. Toxicol. Methods 45(1):7984
4. Szklarz GD, Halpert JR. 1998. Molecular basis of P450 inhibition and activation: implications for drug development
and drug therapy. Drug Metab. Dispos.
26(12):117984
5. MacCoss M, Baillie TA. 2004. Organic
chemistry in drug discovery. Science
303(5665):181013
6. Venkatakrishnan K, Von Moltke LL,
Greenblatt DJ. 2001. Human drug
metabolism and the cytochromes P450:
application and relevance of in vitro models. J. Clin. Pharmacol. 41(11):114979
7. Shou M. 2002. Kinetic analysis for multiple substrate interaction at the active site
of cytochrome P450. Methods Enzymol.
357:26176
8. Shou M, Lin Y, Lu P, Tang C, Mei
Q, et al. 2001 Enzyme kinetics of cytochrome P450-mediated reactions. Curr.
Drug Metab. 2:1736
9. Houston JB, Kenworthy KE. 2000.
In vitro-in vivo scaling of CYP kinetic
data not consistent with the classical
Michaelis-Menten model. Drug Metab.
Dispos. 28:24654
4 Dec 2004
13:28
306
18.
19.
20.
21.
22.
23.
24.
25.
26.
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
ATKINS
aromatic bases: a 1H NMR study. J. Am.
Chem. Soc. 116(11):486673
Tang W, Stearns RA, Kwei GY, Iliff SA,
Miller RR, et al. 1999. Interaction of diclofenac and quinidine in monkeys: stimulation of diclofenac metabolism. J. Pharmacol. Exp. Ther. 291:106874
Egnell AC, Houston B, Boyer S. 2003.
In vivo CYP3A4 heteroactivation is a possible mechanism for the drug interaction
between felbamate and carbamazepine. J.
Pharmacol. Exp. Ther. 305:125162
Hutzler JM, Frye RF, Korzekwa KR,
Branch RA, Huang SM, Tracy TS. 2001.
Minimal in vivo activation of CYP2C9mediated flurbiprofen metabolism by dapsone. Eur. J. Pharm. Sci. 14:4757
Ekins S, Ring BJ, Binkley SN, Hall SD,
Wrighton SA. 1998. Autoactivation and
activation of the cytochrome P450s. Int.
J. Clin. Pharmacol. Ther. 36:64251
Tracy TS, Hutzler JN, Haining RL, Rettie AE, Hummel MA, Dickmann LJ.
2002. Polymorphic variants (CYP2C9 3
and CYP2C9 5) and the F114L active site
mutation of CYP2C9: effect on atypical
kinetic metabolism profiles. Drug Metab.
Dispos. 30:38590
He YA, Roussel F, Halpert JR. 2003.
Analysis of homotropic and heterotropic
cooperativity of diazepam oxidation by
CYP3A4 using site-directed mutagenesis and kinetic modeling. Arch. Biochem.
Biophys. 409:92101
Hutzler JM, Hauer MJ, Tracy TA. 2001.
Dapsone activation of CYP2C9-mediated
metabolism: evidence for activation of
multiple substrates and a two-site model.
Drug Metab. Dispos. 29:102934
Khan KK, He YA, Domanski TL, Halpert
JR. 2002. Midazolam oxidation by cytochrome P450 3A4 and active-site mutants: an evaluation of multiple binding
sites and of the metabolic pathway that
leads to enzyme inactivation. Mol. Pharmacol. 61:495506
Ngui JS, Chen Q, Shou M, Wang RW,
Stearns RA, et al. 2001. In vitro stimu-
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
4 Dec 2004
13:28
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
37.
38.
39.
40.
41.
42.
43.
44.
307
45. Schrag ML, Wienkers LC. 2000. Topological alteration of the CYP3A4 active
site by the divalent cation Mg(2+). Drug
Metab. Dispos. 28(10):1198201
46. Hutzler JM, Powers FJ, Wynalda MA,
Wienkers LC. 2003. Effect of carbonate
anion on cytochrome P450 2D6-mediated
metabolism in vitro: the potential role
of multiple oxygenating species. Arch.
Biochem. Biophys. 417(2):16575
47. Maenpaa JVA, Hall SD, Ring BJ, Strom
SC, Wrighton SA. 1998. Human cytochrome P450 3A (CYP3A) mediated
midazolam metabolism: the effect of assay conditions and regioselective stimulation by -naphthoflavone, terfenadine and
testosterone. Pharmacogenetics 8:137
55
48. Guengerich FP. 1999. Cytochrome P450 3A4: regulation and role in drug
metabolism. Annu. Rev. Pharmacol. Toxicol. 39:117
49. Schenkman JB, Jansson I. 2003. The
many roles of cytochrome b5. Pharmacol.
Ther. 97(2):13952
50. Yamazaki H, Johnson WW, Ueng YF,
Shimada T, Guengerich FP. 1996. Lack
of electron transfer from cytochrome
b5 in stimulation of catalytic activities
of cytochrome P450 3A4. Characterization of a reconstituted cytochrome P450
3A4/NADPH cytochrome P450 reductase
system and studies with apo-cytochrome
b5. J. Biol. Chem. 271(44):2743844
51. Gilep AA, Guryev OL, Usanov SA, Estabrook RW. 2001 Apo-cytochrome b5
as an indicator of changes in heme accessability: preliminary studies with cytochrome P450 3A4. J. Inorg. Biochem.
87(4):23744
52. Yamazaki H, Nakamura M, Komatsu T,
Ohyama K, Hatanaka N, et al. 2002. Roles
of NADPH-P450 reductase and apo- and
holo-cytochrome b5 on xenobiotic oxidations catalyzed by 12 recombinant human
cytochrome P450s expressed in membranes of Escherichia coli. Protein Expr.
Purif. 24(3):32937
4 Dec 2004
13:28
308
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
ATKINS
53. Guryev OL, Gilep AA, Usanov SA, Estabrook RW. 2001. Interaction of apocytochrome b5 with cytochromes P4503A4 and P45017A: relevance of heme
transfer reactions. Biochemistry 40(16):
501831
54. Modi S, Gilham DE, Sutcliffe MJ, Lian
LY, Primrose WU, et al. 1997. 1-methyl4-phenyl-1,2,3,6-tetrahydropyridine as a
substrate of cytochrome P450 2D6: allosteric effects of NADPH-cytochrome
P450 reductase. Biochemistry 36(15):
446170
55. Hanna IH, Krauser JA, Cai H, Kim MS,
Guengerich FP. 2001. Diversity in mechanisms of substrate oxidation by cytochrome P450 2D6. Lack of an allosteric
role of NADPH-cytochrome P450 reductase in catalytic regioselectivity. J. Biol.
Chem. 276(43):3955361
56. Loughran PA, Roman LJ, Miller RT, Masters BS. 2001. The kinetic and spectral
characterization of the E. coli-expressed
mammalian CYP4A7: cytochrome b5 effects vary with substrate. Arch. Biochem.
Biophys. 385(2):31122
57. Cupp-Vickery J, Andersen R, Hatziris Z.
2000. Crystal structures of ligand complexes of P450eryF exhibiting homotropic
cooperativity. Proc. Natl. Acad. Sci. USA
97(7):305055
58. Torimoto N, Ishii I, Hata M, Nakamura H,
Imada H, et al. 2003. Direct interaction between substrates and endogenous steroids
in the active site may change the activity
of cytochrome P450 3A4. Biochemistry
42(51):1506877
59. Dabrowski MJ, Schrag ML, Wienkers LC,
Atkins WM. 2000. Pyrene-pyrene complexes at the active site of cytochrome
P4503A4: evidence for a multiple substrate binding site. J. Am. Chem. Soc.
124(40):1186667
60. Segel IH. 1993. Multisite and allosteric
enzymes. In Enzyme Kinetics. pp. 346
404. New York: Wiley
61. Shen L, Fitzloff JF, Cook CS. 2004. Differential enantioselectivity and product-
62.
63.
64.
65.
66.
67.
68.
69.
4 Dec 2004
13:28
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
309
4 Dec 2004
13:28
310
87.
88.
89.
90.
91.
92.
93.
AR
AR232-PA45-12.tex
AR232-PA45-12.sgm
LaTeX2e(2002/01/18)
P1: GCE
ATKINS
first structure of a microsomal P450
implications for drug discovery. Curr.
Opin. Drug Discov. Dev. 4(5):67177
Lepesheva GI, Waterman MR. 2004.
CYP51the omnipotent P450. Mol. Cell
Endocrinol. 215(12):16570
Li H, Poulos TL. 1996. Conformational
dynamics in cytochrome P450-substrate
interactions. Biochimie 78(89):69599
Poulos TL. 1988. Cytochrome P450:
molecular architecture, mechanism, and
prospects for rational inhibitor design.
Pharm. Res. 5(2):6775
Schlichting I, Berendzen J, Chu K,
Stock AM, Maves SA, et al. 2000.
The catalytic pathway of cytochrome
p450cam at atomic resolution. Science
287(5458):161522
Ravichandran KG, Boddupalli SS, Hasermann CA, Peterson JA, Deisenhofer J.
1993. Crystal structure of hemoprotein
domain of P450BM-3, a prototype for
microsomal P450s. Science 261(5122):
73136
Wester MR, Johnson EF, Marques-Soares
C, Dijols S, Dansette PM, et al. 2003.
Structure of mammalian cytochrome
P450 2C5 complexed with diclofenac at
2.1 A resolution: evidence for an induced
fit model of substrate binding. Biochemistry 42(31):933545
Wester MR, Johnson EF, Marques-Soares
C, Dansette PM, Mansuy D, Stout CD.
2003. Structure of a substrate complex
of mammalian cytochrome P450 2C5 at
2.3 A resolution: evidence for multiple
substrate binding modes. Biochemistry
42(21):637079
7 Jan 2005
14:1
AR
AR232-PA45-13.tex
AR232-PA45-13.sgm
P1: GCE
LaTeX2e(2002/01/18)
10.1146/annurev.pharmtox.45.120403.095920
INTRODUCTION
Epoxide-containing compounds are ubiquitously found in the environment from
both natural and man-made sources, and a large variety of aromatic and alkenic
compounds are also metabolized to epoxides endogenously (1, 2). An epoxide (or
oxirane) is a three-membered cyclic ether that has specific reactivity patterns owing
to the highly polarized oxygen-carbon bonds in addition to a highly strained ring
(3). Some reactive epoxides are responsible for electrophilic reactions with critical
biological targets such as DNA and proteins, leading to mutagenic, toxic, and
carcinogenic effects (4, 5). Although most epoxides are of intermediate reactivity,
relatively stable at physiological pHs, and do not present acute dangers to cells,
they still need to be transformed in a controlled manner (6). The catalytic addition
of water to epoxides or arene oxides by epoxide hydrolases (EHs, E.C.3.3.2.3) to
yield the corresponding 1,2-diols, or glycols (7), is only one of several ways that
cells transform oxiranes. However, EHs are ubiquitous and hydration seems to be a
common route of epoxide transformation. The reaction is energetically favorable,
with water as the only cosubstrate.
0362-1642/05/0210-0311$14.00
311
7 Jan 2005
14:1
312
AR
AR232-PA45-13.tex
MORISSEAU
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
HAMMOCK
MECHANISM
Formation of a Hydroxyl Alkyl-Enzyme Intermediate
The observation that both the mammalian mEH and sEH sequences are similar to a
bacterial haloalkane dehalogenase and other related proteins was key in suggesting
7 Jan 2005
14:1
AR
AR232-PA45-13.tex
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
313
that both EHs have a similar mechanism to the bacterial enzyme (53) and that they
are members of the /-hydrolase fold family of proteins (54). All the enzymes
in this family are characterized by a nucleophile-histidine-acid catalytic triad and
have a two-step mechanism involving the formation of a covalent intermediate
(55, 56). This suggested that these EHs hydrolyze epoxides through the formation
of a hydroxyl alkyl-enzyme intermediate as described in Figure 1A. Before this
time, the generally accepted mechanism of EHs involved a general-based-catalyzed
direct attack of water on the epoxide ring (Figure 1B; 5759). Around the same
time that the sequence analysis was done (54), Lacourciere & Armstrong (60)
demonstrated the formation of a covalent intermediate for the mEH with a single
turnover experiment (excess of enzyme) in H218O showing that the 18O was not
incorporated in the formed glycol but rather in the protein. A second step was
shown to incorporate the 18O in the product, even in H216O. Further evidence was
gained through the isolation of the covalent intermediates for the sEH and mEH
(61, 62). Chemical characterization of the enzyme-product intermediate indicated
a structure consistent with an -hydroxyl alkyl-enzyme (61).
7 Jan 2005
14:1
314
AR
AR232-PA45-13.tex
MORISSEAU
TABLE 1
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
HAMMOCK
Basic
histidine
Orienting
acid
Polarizing
tyrosines
Human mEH
Asp226
His431
Glu404
Human sEH
Asp334
His523
Asp495
2
3
E+S
E-I E + P
ES
k
2
1.
Furthermore, it was found that the formation of the ester intermediate was reversible (k2 = 0), indicating that the enzyme could stabilize the oxyanion in the
alkylation reaction (first step). Other /-hydrolase fold enzymes have an oxyanion hole that stabilizes tetrahedral intermediates for the formation and hydrolysis
of the covalent bond between the enzyme and the substrate (55, 56). The presence
of such an oxyanion hole in sEH was proposed with a push-pull mechanism that
7 Jan 2005
14:1
AR
AR232-PA45-13.tex
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
315
7 Jan 2005
14:1
316
AR
AR232-PA45-13.tex
MORISSEAU
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
HAMMOCK
Figure 3 Catalytic mechanism of EH. The amino acid residue numbers correspond to the human sEH.
residues (Asp333 and His523). This suggests that Van der Walls interactions make
a significant contribution to substrate binding (77). However, such an analysis
also indicates a number of potential hydrogen bonding sites primarily located on
the surface opposite of Asp333, which could be important in the formation of the
intermediate (top right of Figure 3).
The substrate epoxide is polarized by two tyrosine residues (382 and 465),
which hydrogen bond with the epoxide oxygen. At the same time, the nucleophilic
carboxylic acid of Asp334, present on the side of the catalytic cavity opposite to the
tyrosines, makes a backside attack on the epoxide, usually at the least sterically
hindered and most reactive carbon. The nucleophilic acid is oriented and activated
by His523, a second carboxylic amino acid (Asp495) and possibly other amino acids
in the catalytic site for this attack. A recent study based on molecular dynamics
simulations (80) suggests that the protonation of His523 is essential for the right
orientation of Asp334; however, no experimental proofs exist yet. Because the
mEH has a higher optimal pH for activity (pH 8.09.0) than the sEH (pH 7.07.5)
7 Jan 2005
14:1
AR
AR232-PA45-13.tex
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
317
(32), the corresponding His in mEH is probably not protonated (80). The opening
of the epoxide results in an ester bond between the enzyme carboxylic acid and
one alcohol functionality of the diol. This is termed the hydroxyl alkyl-enzyme
intermediate (bottom right of Figure 3). An important unanswered question is
where the oxygen of the epoxide catches the hydrogen to form a hydroxyl because it
was determined that an oxyanion is not formed (73). It was proposed that the proton
came from one of the tyrosine side chains (77), as it is shown in Figure 3, and that the
formed tyrosinate ion is stabilized by edge-to-face -interactions with surrounding
aromatic residues (79). However, the presence of a tyrosinate ion has yet to be
proven. Furthermore, it could be argued that the high pKa (10) of the tyrosine
side chain makes it difficult for the phenolate to form at the pH (7.4 for sEH and
8.09.0 for mEH) at which the reaction is catalyzed (51). It would be intellectually
satisfying if the hydrogen come from the protonated His523, especially because
this histidine should be a base (not protonated) for the second half of the catalytic
reaction (80). However, crystal structures show that this histidine residue is on the
wrong side of the catalytic site to directly donate its proton to the epoxide (77).
Therefore, a proton shuttle mechanism was proposed to transfer the proton from the
histidine to the tyrosine (80), but this proton transfer pathway has yet to be shown.
Once the covalent hydroxyl alkyl-enzyme is formed, the histidine moves far
enough from the nucleophilic acid (now ester) to allow a water molecule to be
activated by the acid-histidine pair (bottom left of Figure 3). This movement may
account for the fluorescence shift during the enzyme reaction (60). The activation
of the water can occur only if the histidine is not protonated (80). This very
basic water attacks the carbonyl of the ester, releasing the diol product and the
original enzyme. As we described above, a variety of lines of evidence support
this mechanism for both the mammalian microsomal and soluble EH and EHs from
numerous other organisms. Interestingly, a few reports suggest that the cholesterol
epoxide hydrolase has a different mechanism (see below).
7 Jan 2005
14:1
318
AR
AR232-PA45-13.tex
MORISSEAU
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
HAMMOCK
mechanism similar to the one described in Figure 1B. This hypothesis is supported by the recent report of the structure for the limonene-1,2-epoxide hydrolase
from Rhodococcus erythropolis that has a one-step general-based-catalyzed direct
hydration of the epoxide (86).
Quaternary Structures
Analysis of the primary structure suggests that the sEH gene (EPXH2) was produced by the fusion of two primordial dehalogenase genes; the C-terminal sEH
domain has high homology to haloalkane dehalogenase, whereas the N-terminal
domain is similar to haloacid dehalogenase (HAD) (54, 69). This gene fusion hypothesis was recently supported by an X-ray crystal structure of the mouse and
human sEH that exhibit a domain-swapped architecture (Figure 5) where both
domains of each monomer are separated by a proline-rich linker (76, 87). Furthermore, the three-dimensional (3-D) structures confirmed that unlike the mEH, the
sEH is a homodimer with with a monomeric unit of 62.5 kDa as determined from
biochemical analysis (6, 32). The C-terminal domain of one subunit interacts with
both the C- and N-terminal domains of the other monomer. Beside the physical
interaction between the two C-terminal domains, which contain the EH activity,
no cooperative allosteric effects have been reported for the sEH activity (6). The
fact that the C-terminal catalytic cavities are not close to any interdomain or interprotein interface may explain the lack of cooperativity in epoxide hydrolysis
(76). Alternatively, it was found that in solution the monomer and dimer sEH are
active (88, 89), suggesting a natural equilibrium between the two forms of sEH. Although a dissociation constant for the dimer formation has yet to be measured, it is
possible that, under the conditions where sEH activity is generally measured (low
nanomolar), the enzyme could be mainly in its monomeric form, thus preventing
the detection of any possible allosteric effects.
Analysis of the sEH crystal structure reveals that while the C-terminal domain containing the EH activity adopts an /-hydrolase fold as expected, the
7 Jan 2005
14:1
AR
AR232-PA45-13.tex
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
319
epoxides by the sEH (93). Although sEH inhibitors do not influence the phosphatase activity, kinetic analysis revealed a positive cooperative Hill coefficient of
2 for the hydrolysis of the monophosphate of dihydroxy stearic acid, suggesting
an allosteric interaction between the two monomers (93). The recent 3-D structure
INHIBITOR DESIGNS
Specific enzyme inhibitors are important research tools to help understand the
catalytic mechanism of an enzyme and the pathologies that may be associated with
dysfunctions of this enzyme. This statement has been particularly true over the past
few years for sEH, and the recent design of potent inhibitors for sEH has opened
the door to new therapies for hypertension and inflammation (3440). To start
with a historic point of view, the first inhibitors discovered for the sEH and mEH
were epoxide-containing compounds (Figure 6) (see 6 and references therein).
However, most of these compounds are in fact substrates of the corresponding EH
with a relatively low turnover that gives only a transient inhibition in vitro and
are inefficient in cell cultures and in vivo (6, 73, 106, 107). A widely used mEH
inhibitor, trichloropropene oxide, not only reacts with many proteins directly but
7 Jan 2005
14:1
320
AR
AR232-PA45-13.tex
MORISSEAU
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
HAMMOCK
Figure 6 Epoxide-containing inhibitors of (A) the mEH and (B) the sEH.
it is rapidly turned over by the mEH (108). The mEH and sEH activities are found
to be inhibited by Hg2+ and Zn2+, and the sEH is also inhibited by Cu2+ and Cd2+
(109). The inhibition of the human mEH by Zn2+ is found to be competitive with a
KI of 60 M, whereas the inhibition of the human sEH is noncompetitive with a
KI of 20 M (109). This divalent cation is found to also inhibit the phosphatase
activity of the sEH (93). One could hypothesize that the binding of Zn2+ at the Mg2+
site in the N-terminal domain resulted in loss of both activities through some as yet
unknown allosteric mechanism that is suggested by the sEH quaternary structure
(see above). During inflammation, the concentrations of divalent cation metals,
especially zinc, increase in the liver (110), suggesting that the binding of Zn2+
could be a simple way for the organism to naturally reduce the sEH activity that
was shown to be proinflammatory (34, 38).
Approximately a decade ago, valpromide treatment was reported to affect the
normal metabolism of carbamazepine by inhibiting the mEH in vivo (111, 112).
The anticonvulsant properties of this compound could cause undesirable secondary
effects in experimental systems if used as mEH inhibitor, but other unsubstituted
amides could be used (113). Recently, primary ureas, amides, and amines were
described as mEH inhibitors (Figure 7A; 114). The most potent inhibitor obtained,
elaidamide, has a mix of competitive and noncompetitive inhibition kinetics with
a KI of 70 nM. It is efficient in vitro (114); however, its fast turn over by amidases limits its use in vivo, underlying the need of new potent and stable mEH
inhibitors.
1,3-Disubstituted ureas, carbamates, and amides (Figure 7B) were recently
described as new potent and stable inhibitors of sEH (115). These compounds
are competitive tight-binding inhibitors with nanomolar Ki that interact stoichiometrically with purified recombinant sEH (115, 116). Crystal structures show
that the urea inhibitors establish hydrogen bonds and salt bridges between the
7 Jan 2005
14:1
AR
AR232-PA45-13.tex
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
321
urea functionality of the inhibitor and residues of the sEH active site, mimicking
the intermediate formed during the enzymatic epoxide ring opening, as described
in Figure 3 (76, 77, 87). Furthermore, the side chains of the inhibitors (R and
R ) need to be hydrophobic to bind tightly in the hydrophobic catalytic site,
as shown in Figure 4 (76, 77). Interestingly, because of the presence of a methionine residue (Met337) pointing into the catalytic cavity, the orientation of
the urea inhibitors is reversed in the human sEH compared with the mouse enzyme (87). Using classical quantitative structure activity relationship (QSAR),
3-D-QSAR, and medicinal chemistry approaches, the structure of these inhibitors
were improved to yield compounds that have an order of magnitude better inhibition potency (116119). This new generation of sEH inhibitors display on
one side of the urea functionality secondary and tertiary pharmacophores at 5
and 11 atoms away from the urea carbonyl group, respectively (119). These inhibitors efficiently inhibit epoxide hydrolysis in several in vitro and in vivo models (3840, 115, 120). The beneficial biological effects observed are discussed
below.
BIOLOGICAL ROLES
The biological role of any enzyme is intimately linked to the substrates the enzyme
transforms. Substrate specificity is probably the best way to distinguish between
the mEH and sEH. These two enzymes have been found to hydrolyze a broad
and complementary range of substrates (6, 32). In general, mEH seems to prefer
mono- and cis-disubstituted epoxides, whereas the sEH prefers gem-di-, trans-di-,
7 Jan 2005
14:1
322
AR
AR232-PA45-13.tex
MORISSEAU
Figure 8
the sEH.
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
HAMMOCK
cis-di-, tri-, and tetra substituted epoxides (32). The latter enzyme hydrates epoxide
on cyclic system very poorly (107). A few examples of mEH and sEH substrates
are shown in Figure 8.
mEH is a key hepatic enzyme involved in the metabolism of numerous xenobiotics, such as 1,3-butadiene oxide 1, styrene oxide 2, and benzo()pyrene 4,5-oxide
3 (6, 33, 52). In addition, mEH is likely involved in the extrahepatic metabolism
of these agents (33, 121, 122). Styrene 2 and cis-stilbene 4 oxides are widely used
as surrogate substrates for mEH (32). The mEH action is part of a detoxification
process for most of the substrates (6, 33). This detoxification action is illustrated
by the example of a man who had a defect in mEH expression and suffered from
acute and severe phenytoin toxicity (123). However, in some cases, such as for
7 Jan 2005
14:1
AR
AR232-PA45-13.tex
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
323
benzo()pyrene 4,5-oxide 5, diol formation can lead to the stabilization of a secondary epoxide, increasing the mutagenic and carcinogenic potential of the product
(124). The procarcinogenic role of mEH was illustrated in mEH knockout mice that
were less sensitive to the carcinogenic activity of 7,12-dimethylbenz[]anthracene
than control mice (125). Furthermore, in human populations, polymorphism in the
mEH gene is associated with the onset of numerous cancers (4244, 126, 127).
The role of mEH in xenobiotic metabolism is probably also linked to the observed
relationship between mEH polymorphism and emphysema (41) or Crohns disease
(45).
Despite the fact that mEH knockout mice do not present an obvious phenotype
(125), there are several new lines of evidence suggesting an endogenous role for
this enzyme. A potential role of mEH in sexual development is supported by the
fact that androstene oxide 5 is a very good mEH substrate (128), and that mEH
is an apparent subunit of the antiestrogen-binding-site (105). Such a role could
be related to the observed relation between mEH polymorphism and spontaneous
abortion (129) or preeclampsia (130). Furthermore, mEH is well expressed in
ovaries (131), especially in follicle cells (132). During the past decade, mEH was
also described as mediating the transport of bile acid in the liver (133, 134). The
mechanism by which mEH participates in this transport is not known. Potent mEH
inhibitors could provide new tools to better understand the multiple roles of this
enzyme.
sEH was thought to participate in the metabolism of xenobiotics, like the mEH;
however, there is no evidence supporting this hypothesis in vivo in mammals (6,
52). Radioactive aromatic epoxides, such as trans-diphenyl-propene oxide 6 and
trans-stilbene oxide 7, are classically used as surrogate substrates for this enzyme
in vitro (32, 135). On the other hand, the sEH is clearly involved in the metabolism
of arachidonic epoxides (8, also called epoxyeicosatrienoic acids or EETs) and
linoleic acid epoxides (9, also called leukotoxins) both in vitro and in vivo (34, 35,
136, 137). The sEH hydrates all of these epoxy-fatty acids with high VM and low
KM. The sEH-dependent transformation of EETs decline as the epoxide approaches
the carboxyl terminal (i.e., 14,15-EET is hydrolyzed 20-fold faster than 8,9-EET
and 5,6-EET is hydrolyzed very slowly), whereas both mono-epoxides of linoleic
acid are hydrolyzed at similar rates (138, 139). Although epoxy-fatty acids are
relatively poor substrates for mEH compared to sEH (138), the former enzyme
hydrolyzes them with a high enantioselectivity, whereas the latter shows little or
no enantiomeric preference (140, 141).
The EETs, which are endogenous chemical mediators (142), act at the vascular, renal, and cardiac levels to regulate blood pressure (143, 144). The vasodilatory properties of EETs are associated with an increased open-state probability
of calcium-activated potassium channels, which lead to hyperpolarization of the
vascular smooth muscle (145). Hydrolysis of the epoxides by sEH diminishes
this activity (146). The sEH-dependent hydrolysis of EETs also regulates their
incorporation into coronary endothelial phospholipids, suggesting a regulation of
endothelial function by sEH (147). Recently, blood pressure reduction was achieved
7 Jan 2005
14:1
324
AR
AR232-PA45-13.tex
MORISSEAU
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
HAMMOCK
in the spontaneous hypertensive rat (SHR) and in angiotensin IIinduced hypertension rat models with pharmacological sEH inhibition (35, 39). Additionally,
male knockout sEH mice have significantly lower blood pressure than wild-type
mice (36), further supporting the role of sEH in blood pressure regulation and sEH
inhibition as a potential new therapeutic treatment for hypertension.
The EETs also display antiinflammatory properties in endothelial cells (37,
148). In contrast, diols derived from epoxy-linoleate (leukotoxin) perturb membrane permeability and calcium homeostasis (34), which results in inflammation
that is modulated by nitric oxide synthase and endothelin-1 (149, 150). Micromolar concentrations of leukotoxin reported in association with inflammation and hypoxia (151) depress mitochondrial respiration in vitro (152) and cause mammalian
cardio-pulmonary toxicity in vivo (150, 153, 154). Leukotoxin toxicity presents
symptoms suggestive of multiple organ failure and acute respiratory distress syndrome (ARDS) (151). In both cellular and whole organism models, leukotoxinmediated toxicity is dependent on epoxide hydrolysis (34, 115), suggesting a role
for sEH in the regulation of inflammation. Treatment with sEH inhibitors increases
EET levels in cell cultures and reduces indicators of vascular inflammation (38,
155), suggesting that sEH is a potential therapeutic target for the treatment of
several vascular inflammatory diseases, including atherosclerosis and kidney failure (38, 46). Inhibition of the sEH seems to result in general antiinflammatory
properties in many systems.
ACKNOWLEDGMENTS
The authors want to particularly thank Dr. John Newman for his precious help
in the preparation of the figures and review of this manuscript. Figures 2, 4, and
5 were prepared using the Cn3D program version 4.1 produced by the National
Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov). This work
was supported in part by NIEHS Grant R37 ES02710, NIEHS Superfund Basic
Research Program Grant P42 ES04699, NIEHS Center for Environmental Health
Sciences Grant P30 ES05707, and NIH/NHLBI R01 HL59699-06A1.
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Jerina DM. 1974. Biological formation
and disposition of arene oxides. Lloydia
37:21218
2. Oliw EH. 1994. Oxygenation of polyunsaturated fatty acids by cytochrome
P450 monooxygenases. Prog. Lipid Res.
33:32954
3. Parker RE, Isaacs NS. 1959. Mechanism of epoxide reactions. Chem. Rev.
264:931013
4. Szeliga J, Dipple A. 1998. DNA adducts
formation by polycyclic aromatic hydrocarbon dihydrodiol epoxides. Chem. Res.
Toxol. 11:111
7 Jan 2005
14:1
AR
AR232-PA45-13.tex
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
325
7 Jan 2005
14:1
326
26.
27.
28.
29.
30.
31.
32.
33.
34.
AR
AR232-PA45-13.tex
MORISSEAU
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
HAMMOCK
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
7 Jan 2005
14:1
AR
AR232-PA45-13.tex
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
327
7 Jan 2005
14:1
328
64.
65.
66.
67.
68.
69.
70.
71.
72.
AR
AR232-PA45-13.tex
MORISSEAU
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
HAMMOCK
73.
74.
75.
76.
77.
78.
79.
80.
involves reversible formation and rate limiting hydrolysis of the alkyl-enzyme intermediate. J. Am. Chem. Soc. 118:943637
Morisseau C, Du G, Newman JW, Hammock BD. 1998. Mechanism of mammalian soluble epoxide hydrolase inhibition by chalcone oxides derivatives. Arch.
Biochem. Biophys. 356:21428
Satoh T, Hosokawa M. 1998. The mammalian carboxylesterases: from molecules
to functions. Annu. Rev. Pharmacol. Toxicol. 38:25788
Nardini M, Ridder IS, Rozeboom HJ,
Kalk KH, Rink R, et al. 1999. The Xray structure of epoxide hydrolase from
Agrobacterium radiobacter AD1. An enzyme to detoxify harmful epoxides. J.
Biol. Chem. 274:1457986
Argiriadi MA, Morisseau C, Hammock
BD, Christianson DW. 1999. Detoxification of environmental mutagens and carcinogens: structure, mechanism, and evolution of liver epoxide hydrolase. Proc.
Natl. Acad. Sci. USA 96:1063742
Argiriadi MA, Morisseau C, Goodrow
MH, Dowdy DL, Hammock BD, et al.
2000. Binding of alkylurea inhibitors to
epoxide hydrolase implicates active site
tyrosines in substrate activation. J. Biol.
Chem. 275:1526570
Zou J, Hallberg BM, Bergfors T, Oesch
F, Arand M, et al. 2000. Structure of Aspergillus niger epoxide hydrolase at 1.8 A
resolution: implications for the structure
and function of the mammalian microsomal class of epoxide hydrolases. Struct.
Fold Des. 8:11122
Yamada T, Morisseau C, Maxwell JE, Argiriadi MA, Christianson DW, et al. 2000.
Biochemical evidence for the involvement
of tyrosine in epoxide activation during
the catalytic cycle of epoxide hydrolase.
J. Biol. Chem. 275:2308288
Schitt B, Bruice TC. 2002. Reaction
mechanism of soluble epoxide hydrolase:
insights from molecular dynamics simulations. J. Am. Chem. Soc. 124:14558
70
7 Jan 2005
14:1
AR
AR232-PA45-13.tex
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
91.
92.
93.
94.
95.
96.
97.
98.
99.
329
1999. Mechanistic studies of phosphoserine phosphatase, an enzyme related to Ptype ATPases. J. Biol. Chem. 274:33985
90
Ndubuisil MI, Kwok BH, Vervoort J, Koh
BD, Elofsson M, et al. 2002. Characterization of a novel mammalian phosphatase having sequence similarity to
Schizosaccharomyces pombe PHO2 and
Saccharomyces cerevisiae PHO13. Biochemistry 41:784148
Cronin A, Mowbray S, Durk H, Homburg S, Fleming I, et al. 2003. The Nterminal domain of mammalian soluble
epoxide hydrolase is a phosphatase. Proc.
Natl. Acad. Sci. USA 100:155257
Newman JW, Morisseau C, Harris TR,
Hammock BD. 2003. The soluble epoxide
hydrolase encoded by EPXH2 is a bifunctional enzyme with novel lipid phosphate
phosphatase activity. Proc. Natl. Acad.
Sci. USA. 100:155863
Craft JA, Baird S, Lamont M, Burchell B.
1990. Membrane topology of epoxide hydrolase. Biochim. Biophys. Acta 1046:32
39
Friedberg T, Lollmann B, Becker R,
Holler R, Oesch F. 1994. The microsomal
epoxide hydrolase has a single membrane
signal anchor sequence which is dispensable for the catalytic activity of this protein. Biochem. J. 303:96772
Bulleid NJ, Graham AB, Craft JA. 1986.
Microsomal epoxide hydrolase of rat
liver. Purification and characterization of
enzyme fractions with different chromatographic characteristics. J. Biochem.
233:60711
Griffin M. 1999. Regulation of rat liver
epoxide hydrolase by tightly bound phosphoinositides. Proc. Okla. Acad. Sci. 79:
16
Galteau MM, Antoine B, Reggio H. 1985.
Epoxide hydrolase is a marker for the
smooth endoplasmic reticulum in rat liver.
EMBO J. 4:2793800
Von Dippe P, Amoui M, Alves C, Levy
D. 1993. Na(+)-dependent bile acid
7 Jan 2005
14:1
330
100.
101.
102.
103.
104.
105.
106.
107.
AR
AR232-PA45-13.tex
MORISSEAU
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
HAMMOCK
108.
109.
110.
111.
112.
113.
114.
115.
116.
117.
7 Jan 2005
14:1
AR
AR232-PA45-13.tex
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
118.
119.
120.
121.
122.
123.
124.
125.
2000. 3-D QSAR analysis of inhibition of murine soluble epoxide hydrolase (MsEH) by benzoylureas, arylureas,
and their analogues. Bioorg. Med. Chem.
8:266373
McElroy N, Jurs PC, Morisseau C, Hammock BD. 2003. QSAR and classification of murine and human soluble
epoxide hydrolase inhibition by urea-like
compounds. J. Med. Chem. 46:106680
Kim I-H, Morisseau C, Watanabe T,
Hammock BD. 2004. Design, biosynthesis, and biological activity of 1,3disubstituted ureas as potent inhibitors of
the soluble epoxide hydrolase of increased
water solubility. J. Med. Chem. 47:2110
22
Newman JW, Denton DL, Morisseau C,
Koger CS, Wheelock CE, et al. 2001.
Evaluation of fish models of soluble epoxide hydrolase inhibition. Environ. Health.
Perspect. 109:6166
Zheng J, Cho M, Brennan P, Chichester
C, Buckpitt AR, et al. 1997. Evidence for
quinone metabolites of naphthalene covalently bound to sulfur nucleophiles of
proteins of mouse Clara cell after exposure to naphthalene. Chem. Res. Toxicol.
10:100814
Teissier E, Fennrich S, Strazielle N, Daval
J-L, Ray D, et al. 1998. Drug metabolism
in in vitro organotypic and cellular models
of mammalian central nervous system: activities of membrane-bound epoxide hydrolase and NADPH-cytochrome P-450
(c) reductase. Neurotoxicology 19:34756
Yoo JH, Kang DS, Chun WH, Lee WJ, Lee
KH. 1999. Anticonvulsant hypersensitivity syndrome with an epoxide hydrolase
defect. Br. J. Dermatol. 140:18183
Szeliga J, Dipple A. 1998. DNA adducts
formation by polycyclic aromatic hydrocarbon dihydrodiol epoxides. Chem. Res.
Toxicol. 11:111
Miyata M, Kudo G, Lee Y-H, Yang TJ,
Gelboin HV, et al. 1999. Targeted disruption of the microsomal epoxide hydrolase
gene: microsomal epoxide hydrolase is
126.
127.
128.
129.
130.
131.
132.
133.
331
7 Jan 2005
14:1
332
134.
135.
136.
137.
138.
139.
140.
141.
AR
AR232-PA45-13.tex
MORISSEAU
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
HAMMOCK
7 Jan 2005
14:1
AR
AR232-PA45-13.tex
AR232-PA45-13.sgm
LaTeX2e(2002/01/18)
P1: GCE
333
HI-RES-PA45-13-Morisseau.qxd
1/7/05
2:26 PM
Page 1
C-1
Figure 2 Structure of the active site of the mouse sEH showing the presence of two
tyrosine residues, 381 and 465 (gray), positioned opposite of the catalytic triad
(Asp333 in red, His523 in blue, and Asp495 in red). Structure obtained from Reference
76.
HI-RES-PA45-13-Morisseau.qxd
C-2
MORISSEAU
1/7/05
2:26 PM
Page 2
HAMMOCK
Figure 4 Hydrophobicity map of the mouse sEH substrate binding pocket cocrystalyzed with the inhibitor 1-cyclohexyl-3-dodecyl urea (77). Amino acid side chains
within 6 of the inhibitor are displayed as space-filling models. The residues shown
in bright red and blue are the urea oxygen and nitrogens, respectively. A color gradient of brown to blue indicates degrees of hydrophobicity. Panel A shows a view of
the catalytic pocket from the inside of the enzyme toward the outside, and panel B
shows the opposite view. A series of hydrophilic residues are observed on the top
side of the channel (Phe265, Pro266, Trp334, Val337, Pro363, Pro369, Ile373, Phe385,
Phe406, Ile427, Thr468, Trp472), whereas the bottom of the channel is very
hydrophobic (Thr359, Met361, Pro363, Val372, Phe379, Ile416,Val418, Val497, Lys498,
Trp524), with the exception of the catalytic aspartate and histidine (Asp333 and
His523). This structural analysis indicates that a number of potential hydrogen bonding sites (Tyr381, Gln382, Tyr465) are observed in the substrate binding pocket of the
soluble epoxide hydrolase, primarily located on the surface opposite Asp333.
HI-RES-PA45-13-Morisseau.qxd
1/7/05
2:26 PM
Page 3
C-3
Figure 5 Structure of the mouse sEH dimer (76). The N-terminal domains (residues
Arg4-Gly218) are in yellow-orange, the C-terminal domains (residues Val235-Ala544)
are in blue-green, and the proline-rich linker (Thr219-Asp234) is in magenta. Catalytic
residues for both the C- and N-terminal domains are displayed as space-filling
residues with blue for positive charge, red for negative charge, and gray for neutral.
The alternating helices and the beta sheet floor typical of the /-hydrolase fold
enzymes is clearly shown in the C-terminal domain.
4 Dec 2004
13:24
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
P1: GCE
LaTeX2e(2002/01/18)
10.1146/annurev.pharmtox.45.120403.095959
and Pharmacology
Jon M. Fukuto,1 Christopher H. Switzer,2
Katrina M. Miranda,3 and David A. Wink4
1
INTRODUCTION
The biological activity and biological chemistry of nitrogen oxide species in mammalian systems have received considerable attention over the past 15 years. Interest
in this area is primarily the result of the discovery of endogenous generation of
nitric oxide (NO) by mammalian cells. Although the focus of much of the past
research has been NO, it is becoming increasingly clear that other nitrogen oxides
derived in vivo from NO may have significant physiological and/or pathophysiolgical functions. Although significant advances have been made in our understanding
of the chemical biology of NO and related/derived nitrogen oxides, such as nitrogen dioxide (NO2), dinitrogen trioxide (N2O3), and peroxynitrite (ONOO),
nitroxyl (HNO) remains the least studied and least understood of all the biologically relevant nitrogen oxides. Despite the original description of HNO more
0362-1642/05/0210-0335$14.00
335
4 Dec 2004
13:24
336
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
FUKUTO ET AL.
than 100 years ago, understanding of the chemistry and biochemistry of HNO
has seriously lagged behind other redox nitrogen oxide congeners, even after the
discovery of endogenous mammalian nitrogen oxide generation. However, recent
reports have indicated that HNO has novel and potentially important biological activity (see below), which prompted numerous labs to investigate the physiological
and chemical properties and reactivity of HNO. Much of this recent work has led
to redefinition of the fundamental chemistry of this enigmatic species, which aided
in partial understanding of the chemistry responsible for the newly discovered biological properties of HNO. In this chapter, we review some of the physiologically
relevant chemical properties of HNO and discuss some of its recently discovered
biological/pharmacological properties.
1.
Since then, others have proposed HNO as an intermediate in a variety of chemical and biological processes. For example, HNO has been proposed to be generated
during bacterial denitrification (6), released from acid-catalyzed solvolysis of acinitroalkanes (Nef reaction) (7), the product of the reaction of NO with hydrogen
(8), and a product of the reaction of NO with hydroxylamine (9, 10). Direct observation of HNO was accomplished by Brown & Pimentel (11) when they trapped
it in an argon matrix during the photolysis of methyl nitrite. HNO has also been
generated by pulse radiolysis (12, 13), a technique that has led to the elucidation
of some of the fundamental chemical properties of HNO (although some of these
properties have been reevaluated and revised recently, see below).
One of the most unique and fascinating aspects of HNO chemistry involves its
simple deprotonation. The electronic ground state of HNO is the singlet where all
4 Dec 2004
13:24
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
NITROXYL (HNO)
337
HNO
3 NO + H+
2.
3.
Thus, equilibrium between HNO and NO cannot be achieved at suprananomolar concentrations because HNO can be siphoned off to N2O and H2O via
Reaction 3. Regardless, in a 1970 study where NO was generated using pulse
radiolysis, a pKa for HNO of 4.7 was reported (12). This report did not specify
the spin states of the relevant equilibrium species, and until recently this was the
exclusively quoted value for the pKa of HNO. Theoretical reevaluation of the HNO
pKa led to a revision of 7.2 (19). Of particular note, this work specifically indicated
that the relevant equilibrium in solution, between HNO and 3NO, was the same as
that proposed by Janaway & Brauman for the gas phase (22). Further experimental
4 Dec 2004
13:24
338
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
FUKUTO ET AL.
and theoretical work by Shafirovich & Lymar (24) and Bartberger and coworkers
(23) provided a consensus agreement that the pKa of HNO is 11.4.
Bartberger and coworkers (19) noted that the revision of the HNO pKa required reevaluation of an aspect of NO chemistry as well. An often-quoted reduction potential for the NO/3NO couple was calculated to be 0.39 V (versus
NHE throughout) using the assumptions, among others, that the HNO pKa was 4.7
and the relevant acid-base equilibrium was between HNO and 1NO (20). Considering the dramatic revision in the pKa and the establishment that the relevant
equilibrium is between HNO and 3NO, recalculation of the NO/3NO couple
gives a value of 0.8 V (23). This recalculated reduction potential is consistent
with experimentally derived values (27, 28), and this previously irreconcilable difference between experiment and calculation can now be explained. Protonation of
3
NO to HNO will be highly favorable at physiological pH and therefore results
in a positive shift in the potential to approximately 0.5 to 0.6V as the pH is
lowered (23, 24). These negative values for the one-electron reduction potentials
for both the NO/3NO and NO,H+/HNO couples indicate that direct one-electron
reduction of NO to reduced species by an outer sphere electron transfer process
is thermodynamically unfavorable and not likely to occur under biological (mammalian) conditions. This idea has, however, been challenged on the basis that if
the intracellular concentrations of the reductants and oxidants are considered, a
much less negative potential will be realized (29). Moreover, it has been pointed
out that the reduction potentials in prokaryotic cells may be capable of reducing
NO (29) and may be part of an antipathogenic response of NO.
The discussion of nitroxyl chemistry thus far has focused on HNO and 3NO.
However, a triplet protonated species and a singlet anionic species have been
examined in previous studies. Protonation of 3NO has been proposed to occur
on the more electronegative oxygen atom, generating 3NOH (30, 31). This triplet
protonated species has been calculated to be approximately 2023 kcal/mol less
stable than 1HNO (32, 33). Thus, interconverison between 1HNO and 3NOH is
biologically inaccessible. As noted earlier, the singlet anionic nitroxyl, 1NO,
has been determined to be approximately 1720 kcal/mol above the ground state
triplet species, which agrees reasonably well with theoretical studies (19). From
a biological perspective, the only accessible nitroxyl species are HNO and 3NO
(which is the reason the chemistry of these has been the focus of discussion).
Figure 1 depicts the energy relationships between all of these protonation- and
spin-related species.
As is evident from the above discussion, fundamental nitroxyl chemistry is
conceptually distinct and, at times, requires one to suspend commonly held chemical dogma/beliefs when thinking about it. When addressing the chemistry of
nitroxyl in biological/physiological systems, it will be important to remember the
following:
1. The pKa of HNO is approximately 11 and therefore, if formed initially, HNO
will be the predominant species present at physiological pH.
T o x i c o l .
2 0 0 5 . 4 5 : 3 3 5 - 3 5 5 .
D o w n l o a d e d
f r
H e i d e l b e r g
o n
1 0 / 0 1 / 0 5 .
F o r
p e r s o n a l
u s e
A n n u .
R e v .
P h a r m a c o l .
b y
U n i v e r s i t a e t
4 Dec 2004
13:24
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
NITROXYL (HNO)
339
2. The relevant acid-base equilibrium for nitroxyl is between the singlet protonated, 1HNO, and the triplet anion, 3NO.
3. The requirement for a spin-state change for nitroxyl protonation-deprotonation means these reactions are slow relative to all other protonation-deprotonation events, which are extremely fast.
4. If circumstances exist whereby 3NO is generated/formed, it will have a
significant lifetime (milliseconds) because protonation to HNO is slow.
5. The other protonation/spin-state congeners of nitroxyl, namely 1NOH and
1
NO, are biologically inaccessible and irrelevant to most all discussions of
biological nitroxyl activity.
6. Generation of HNO or 3NO via single-electron reduction of NO by an outer
sphere process is not favorable, although it may be possible.
REACTIVITY OF NITROXYL
As already mentioned, an important (and bothersome) reaction of HNO is dimerization with itself followed by dehydration to give N2O and H2O (Reaction 3).
This propensity to dimerize necessitates the use of donor molecules for most studies of HNO. The ground state triplet anion, 3NO, reacts with O2 to generate
peroxynitrite, OONO (34). This reaction (Reaction 4) is isoelectronic with the
well-studied reaction of NO with O
2 (Reaction 5), and both reactions occur at
near diffusion-controlled rates (24, 35, 36):
3
NO + 3 O2
OONO
(2.7 109 M1 s1 )
4.
4 Dec 2004
13:24
340
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
FUKUTO ET AL.
NO + O
2
OONO
(47 109 M1 s1 )
5.
decomposing to N2O and NO2 (12, 13, 37) (Reactions 6, 7, and 8):
NO + NO N2 O
2
Annu. Rev. Pharmacol. Toxicol. 2005.45:335-355. Downloaded from arjournals.annualreviews.org
by Universitaet Heidelberg on 10/01/05. For personal use only.
N2 O
2 + NO N3 O3
(k = 1.73.3 109 M1 s1 )
6.
(k = 34.9 106 M1 s1 )
7.
N3 O
3 N2 O + NO2
(k = 87330 s1 )
8.
(5.8 106 M1 s1 )
9.
(8 106 M1 s1 )
10.
1 1
11.
(1.6 10 M
4
s )
The catenation reactions of HNO/NO with NO may be of some biological/pharmacological interest because both species may be present simultaneously
under certain circumstances. Indeed, nitroxyl may be capable of attenuating the
actions of NO (and vice versa).
Formation of N2O
2 /HN2O2 (Reactions 6 and 9) has been hypothesized to lead
to the generation of the potent oxidant hydroxyl radical (HO) (13) (Reaction 12):
N2 O
2 N2 O + O ( HO)
(3.5 102 s1 )
12.
Although this reaction has been proposed to account for some of the oxidative
chemistry and/or toxicity associated with nitroxyl (3840), unequivocal demonstration of this reaction is lacking.
One of the most important and biologically relevant aspects of HNO chemistry
is its ability to react as an electrophile with thiols. In fact, this reaction has been
used to distinguish between the biology of HNO versus NO because HNO is
much more reactive toward thiols compared with NO (41). The electrophilicity
of HNO appears to be particularly high for thiols and much less so with oxygenbased nucleophiles (19). The reactivity of HNO with amine nucleophiles has been
calculated to be intermediate between thiols and oxygen-based nucleophiles and
can be favorable. The initial product of the reaction of HNO with a thiol is an
4 Dec 2004
13:24
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
NITROXYL (HNO)
341
13.
14.
15.
16.
Alkyl and aryl diazenes are known to be unstable with respect to dinitrogen (N2)
loss, and oxidative decomposition leads to the formation of alkyl radicals (50, 51).
The reaction of HNO with amines has not been extensively examined beyond
the recently published theoretical treatment (19). However, one study in 1965
reported that reaction of secondary amines with the HNO-donor Angelis salt led
to the generation of dinitrogen and products consistent with radical intermediates
(52). Whether HNO release from Angelis salt was required for this chemistry was
not determined, however.
The reaction of nitroxyl with dioxygen has become a topic of considerable
interest. As discussed above, there appears to be little doubt that the reaction
4 Dec 2004
13:24
342
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
FUKUTO ET AL.
of 3NO with O2 to give ONOO is a fast reaction. Of, course, the biological
relevance of this reaction is largely dependent on the existence/levels of 3NO.
The pKa of HNO is approximately 11, making equilibrium concentrations of
3
NO extremely small under physiological conditions. However, as noted above,
biological circumstances whereby 3NO is made directly (if they exist) could lead
to reaction with O2 because protonation is very slow and, therefore, 3NO can
have a significant (millisecond) lifetime (25). A more intriguing, provocative, and
biologically relevant process is the reaction of HNO with O2. Although this reaction
has been analyzed experimentally and theoretically, it remains a controversial and
contentious topic. Several studies by Miranda and coworkers indicate that the
reaction of HNO with O2 results in the generation of a potent two-electron oxidant
whose reaction profile is distinct from that of OONO and/or N2O3 (53, 54).
Interestingly, aerobic decomposition of the HNO-donor molecule Angelis salt
primary nitroxyl-O2 product (owing to the fact that OONO decomposes to give
predominantly NO
3 ) (2). Moreover, a direct and spin-forbidden reaction of HNO
with O2 to generate HOONO/OONO would be very slow and highly unlikely
(25). Thus, the chemistry and biology of the HNO/O2 reaction remains one of the
most significant and important enigmas in the field of HNO chemistry and biology.
As mentioned above, NO is very difficult to reduce to 3NO (indicated by the
very low reduction potential for the NO/3NO couple). This means that 3NO, if
formed, can be a one-electron reducing agent. An example of this is the reduction
of the cupric form of the enzyme superoxide dismutase (SOD) to the cuprous form
by NO (49, 5557) (Reaction 17):
NO + SODCuII NO + SODCuI
17.
Most of the studies examining the interaction of nitroxyl with SOD were performed prior to the understanding that the primary nitroxyl species present in
solution at physiological pH is HNO rather than 3NO. Thus, all reactions were
written as occurring through the deprotonated anionic species. Although this is possible, previously mentioned difficulties in generating significant concentrations of
NO under biological conditions indicate that coordination of HNO to the metals
followed by deprotonation may be an equally likely mechanism for these reactions. HNO can also react with oxidized hemoproteins, such as methemoglobin, to
generate the ferrous nitrosyl adducts via reductive nitrosylation (26, 42) (Reaction
18):
HNO + HbFeIII HbFeII -NO + H+
Hb = hemoglobin
18.
4 Dec 2004
13:24
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
NITROXYL (HNO)
343
conditions. The H-NO bond strength is only 4850 kcal/mol (see, for example,
19). This relatively low bond strength indicates that HNO will be a good hydrogen
atom donor, and, therefore, can be a good reducing agent for radical species. The
product of hydrogen atom abstraction from HNO is NO, which can also react
rapidly with other radical species. Thus, it may be expected that HNO can be
an efficient radical scavenger via H-atom donation with subsequent generation of
another radical scavenger, NO.
To accurately predict which nitrogen oxide species are relevant following biological HNO exposure and which biological targets are affected, it is imperative
that the kinetics of the reactions of HNO are known. To this end, Miranda and
coworkers (60) used competition kinetics to determine the relative rates of reaction of HNO with a variety of possible chemical/biological reactants and derive
approximate rate constants. This study showed that relative reactivity toward HNO
is oxymyoglobin (1 107 M1 s1) > glutathione (GSH), horseradish peroxidase
(2 106 M1 s1) > N-acetyl cysteine, CuZnSOD, MnSOD, metmyoglobin, catalase (310 105 M1 s1) > Tempol, ferricytochrome c (48 104 M1 s1) >
O2 (3 103 M1s1). Considering the high concentrations of GSH in cells, these
results indicate that reaction of HNO with GSH may be a primary fate for cytosolic HNO. However, HNO partitioned into membrane compartments may have a
considerably longer lifetime.
NITROXYL DONORS
Dimerization of HNO (Reaction 3) precludes convenient and direct accessibility
to HNO for chemical or biological studies. Therefore, most of the studies of HNO
chemistry and biology utilize HNO-donor molecules. The best studied, most established and most utilized HNO-donor is sodium trioxodinitrate (Na2N2O3), or
Angelis salt (1 and references therein) (Reaction 1). This inorganic salt is fairly
stable in base but will spontaneously release HNO between pH 48 with a firstorder rate constant of 4.6 104 s1 (61). Thermal degradation of Angelis salt
can never be used as a source of 3NO because conditions for significant HNO
deprotonation (strong base) inhibits the release of HNO. At low pH, Angelis salt
becomes an NO-donor, possibly owing to protonation of a relatively nonbasic site,
resulting in a different mechanism of decomposition (62). Owing to its ability
of release HNO at physiological pH with predictable kinetics, most of the novel
biological effects of HNO have been discovered using Na2N2O3. However, this
compound is limited in that its half-life is only 2.1 min at 37 C (61), making
prolonged HNO delivery difficult. Moreover, NO2 is released as a coproduct that
exhibits its own array of chemistry/biology (see, for example, 63).
Another possible source of nitroxyl is via the decomposition of compounds
with the N-hydroxysulfonamide functional group. The best known of these is
N-hydroxybenzenesulfonamide (Pilotys acid), which, under basic conditions, decomposes to nitroxyl and benzenesulfinate (Reaction 19):
4 Dec 2004
13:24
344
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
FUKUTO ET AL.
19.
The basic conditions required for HNO release allow HNO deprotonation to
occur, making these compounds possible sources of 3NO (unlike Angelis salt,
which cannot be used as a ready source of 3NO). Like Angelis salt, the spin
state of HNO initially generated from Pilotys acid is singlet (64). Biological
studies using N-hydroxysulfonamides can be troublesome because strongly basic
conditions are required for HNO release. N-Hydroxysulfonamides are also readily
oxidized by one electron to give the corresponding nitroxide intermediate, which
releases NO, not HNO (65). For these reasons, Angelis salt has been used more
extensively for biological studies.
Derivatives of N-hydroxysulfonamides have also been developed as HNOdonors. For example, the Nagasawa lab has synthesized a series of N- and/or
O-substituted N-hydroxysulfonamides, which could be activated in biological systems to release HNO (6671). Similarly, N-hydroxybenzenecarboximidic acid
derivatives have also been developed as HNO-donors (72).
HNO can also be generated via a retro-Diels-Alder reaction from acyl- or
phosphinoyl-nitroso-diene cycloadducts (7377). Water-soluble analogs of acylnitroso-anthracene adducts, which are amenable to biological studies, have been
shown to release the acylnitroso moiety, followed by hydrolysis to give HNO
(78).
NITROXYL TRAPS/DETECTION
Studies on the biological actions and biochemistry of HNO are severely hindered
by the lack of specific and convenient traps and/or detectors for HNO. Many previous studies relied on the detection of N2O, which is the ultimate product of HNO
dimerization (Reaction 3), as an indication of the intermediacy or existence of
HNO. This is, however, equivocal because mechanisms exist whereby N2O can be
generated without the intermediacy of free HNO (see, for example, 43). Moreover, HNO dimerization is a second-order process requiring high concentrations
of HNO to get significant reaction in the thiol and metalloprotein-rich environment
of a cell. Considering the existence of other more likely fates for HNO in biological systems (i.e., reaction with thiols), it is not likely that low levels of HNO in
biological systems can be detected via N2O.
Other traps/detectors for HNO exist. For example, one of the first described
traps for nitroxyl is via reaction with tetracyanonickelate [Ni(CN4)2)] to give the
nickel-nitrosyl species (79) (Reaction 20):
2
Ni(CN)2
4 + HNO/NO NiNO(CN)3 + HCN/CN
20.
However, this is a pH-dependent (only occurs at high pH) and inefficient process
and unlikely to be useful in biological systems.
4 Dec 2004
13:24
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
NITROXYL (HNO)
345
21.
HNO can also be trapped and detected via reaction with nitrosobenzene (44).
This reaction generates cupferron (N-nitroso-N-phenylhydroxylamine) that can
chelate the cupric ion to form a colored complex (Reaction 22):
Phenyl-NO + HNO Phenyl-NH(OH)NO chelates Cu2+
22.
23.
Although this represents an efficient trap for HNO, the products of these reactions can also be generated by reaction with NO via a more involved series of
reactions (see, for example, 83) (Reactions 24 and 25):
+
Porphyrin-Fe3+ + NO + H2 O Porphyrin-Fe2+ + NO
2 + 2H
24.
25.
4 Dec 2004
13:24
346
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
FUKUTO ET AL.
react with other thiols to generate HNO according to Reaction 26 (43, 84):
26.
NITROXYL PHARMACOLOGY/TOXICOLOGY
It remains uncertain whether HNO is endogenously generated in mammalian cells.
Therefore, the question of whether HNO is an endogenous signaling/effector
species or simply a metabolite of NO remains open. However, numerous studies indicate that exogenous HNO administration results in interesting, novel, and
potentially important pharmacology and toxicology. Some of the earliest studies
of the biological activity of HNO reported that nitroxyl can be a potent vasorelaxant (see, for example, 101). Because NO is known to elicit vasodilation via
activation of the enzyme soluble guanylate cyclase (sGC) (see, for example, 102),
it is possible that HNO was being converted to NO in these experiments. This
seems especially likely because HNO itself has been reported to be incapable of
activating sGC (103). It should be noted that this study was performed using an
in vitro preparation of the enzyme in the presence of the thiol dithiothreitol. As
discussed above, thiols can react rapidly with HNO precluding interaction with
the enzyme. Thus, possible HNO-mediated sGC activation needs reinvestigation.
The ability of HNO to react with thiols predicts that it will be capable of
disrupting thiol proteins. The Nagasawa group was one of the first to demonstrate this as they observed inhibition of the enzymes aldehyde dehydrogenase and
glyceraldehyde-3-phosphate dehydrogenase when exposed to HNO donors (67,
104, 105). The finding that aldehyde dehydrogenase was inhibited by HNO was
the result of the elucidation of the mechanism of action of the alcohol deterrent
drug cyanamide by the Nagasawa lab. They found that cyanamide could be hydroxylated by the enzyme catalase, which resulted in the formation of an unstable
4 Dec 2004
13:24
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
NITROXYL (HNO)
347
4 Dec 2004
13:24
348
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
FUKUTO ET AL.
factor in the effect of HNO. That is, under hypoxic conditions, HNO appears able
to decrease steady-state ion currents, whereas under normoxic conditions blockade
of glycine-independent desensitization occurs, leading to ion current enhancement
(114). Thus, the ultimate effect of HNO on the NMDA receptor can be a function
of cellular/tissue O2 levels.
One of the most recent and provocative pharmacological actions of nitroxyl
appears to be as a unique cardiovascular agent. Paolocci and coworkers (115)
have reported that HNO increases left ventricular contractility while lowering
cardiac preload and diastolic pressure without an increase in arterial resistance.
This novel activity is likely responsible for much of the recent attention given
to HNO (see below). The actions of HNO on the vascular system have been
found to be mediated by calcitonin gene-related peptide (CGRP). Significantly,
the effects of HNO were unaffected by -receptor blockade and additive to those
of the 1-selective agonist dobutamine, indicating that the effects of HNO are
independent of -adrenergic signaling (60, 116). Also, HNO administration to
animals does not result in an increase in cGMP, indicating that the vascular effects
were not due to enhancement of levels of this second messenger. CGRP is a broadly
distributed neuropeptide found in many cardiovascular tissues and is the most
potent vasorelaxant currently known, with established positive inotropic effects
on the human heart (see, for example, 117, 119). The actions of CGRP occur
through activation of the calcitonin-receptor-like receptor (CRLP), which leads
to activation of adenylate cyclase and elevation of intracellular cAMP (see, for
example, 118, 119). Increases in cAMP results in PKA activation followed by
phosphorylation of L-type Ca2+ channels and, eventually, vasodilation. Thus, the
actions of HNO, at least with regard to the cardiovascular system, appear to occur
primarily through a cAMP-mediated pathway. This is in contrast to NO, whose
actions in the cardiovascular system are mediated by cGMP. This fundamental
difference in signaling indicates that HNO is not merely converted to NO and that
the two species have orthogonal signaling pathways (60).
The ability of nitroxyl to elicit CGRP-mediated responses in vivo makes it a
candidate for the treatment of heart failure because it will increase heart contractility while decreasing vascular resistance. As pointed out by Feelisch (119), with the
current lack of selective CGRP-mimetics and the increasing interest in inodilators,
the potential for HNO-donors to be developed as drugs to be used in heart failure
is significant. Of course, it needs to be realized that these ideas are in the early
stages of development; much more work needs to be done before the therapeutic
utility of HNO and HNO-donor drugs can be properly evaluated.
SUMMARY
Recent reevaluation of some of the fundamental chemical properties and reactivity of HNO provides a basis to begin to understand the chemistry responsible for some of its novel and potentially important biology. However, a clear
4 Dec 2004
13:24
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
NITROXYL (HNO)
349
understanding of the chemistry and specific biological targets associated with the
physiological/pharmacological actions of HNO remains to be established and is an
area of extreme interest/need. It is clear that nitroxyl possesses biological properties
unique from those of other nitrogen oxides and that may be of significant physiological/pharmacological importance. The idea that the actions of HNO are, in part,
mediated through cAMP, whereas NO regulates through cGMP is intriguing and
represents an interesting physiological paradigm whereby redox nitrogen oxide
congeners have orthogonal signaling properties. Whether this redox nitrogen oxide system is physiologically important is a question that remains and is contingent
upon, among other things, determining whether HNO is generated endogenously,
and, if so, under what conditions.
The utility of HNO as a possible therapeutic agent, for example, in the treatment of heart failure or as a preconditioning agent to prevent ischemia-reperfusion
injury needs to be reconciled with its possible toxicity. Of course, this is not an
uncommon situation, as the same issues are important for the development of NO
as a therapeutic agent. However, it is worth noting that although HNO donors at
millimolar levels have significant toxicity (46), animal studies have shown that
long-term administration of the HNO donor Angelis salt is very well tolerated,
with an LD50 well above 130 mg/kg with no observable carcinogenesis (120). This
concentration is greater than 10,000 times that which has been shown to demonstrate beneficial cardiovascular effects. Regardless, HNO can now take its place
among other nitrogen oxides and oxygen-derived species as an important signaling/effector species possessing novel and possibly useful pharmacology as well
as toxicological properties. Considering that many of the major discoveries of the
physiological chemistry and biology of nitroxyl are relatively recent, it may be
expected that many more interesting and important discoveries await.
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Bonner FT, Hughes MN. 1988. The aqueous solution chemistry of nitrogen in low
positive oxidation states. Comments Inorg. Chem. 7:21534
2. Hughes MN. 1999. Relationships between nitric oxide, nitroxyl anion, nitrosonium cation and peroxynitrite. Biochim.
Biophys. Acta 1411:26372
3. Miranda KM. 2004. The chemistry of nitroxyl (HNO) and implications in biology.
Coord. Chem. Rev. In press
4. Koppenol WH, Traynham JG. 1996. Say
NO to nitric oxide: nomenclature for
5.
6.
7.
8.
4 Dec 2004
13:24
350
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
FUKUTO ET AL.
deuterium and hydrogen atoms with nitric
oxide. J. Chem. Phys. 46:407584
Bonner FT, Dzelzkalns LS, Bonucci JA.
1978. Properties of nitroxyl as intermediate in the nitric oxide-hydroxylamine reaction and in trioxodinitrate decomposition. Inorg. Chem. 17:248794
Cooper JN, Chilton JE Jr, Powell RE.
1970. Reaction of nitric oxide with alkaline hydroxylamine. Inorg. Chem. 9:
23034
Brown HW, Pimentel GC. 1958. Photolysis of nitromethane and of methyl nitrite
in an argon matrix; infrared detection of
nitroxyl, HNO. J. Chem. Phys. 29:88388
Gratzel M, Taniguchi S, Henglein A.
1970. A pulse radiolytic study of shortlived byproducts on nitric oxide reduction in aqueous solution. Ber. Bunsenges.
Phys. Chem. 74:100310
Seddon WA, Fletcher JW, Sopchyshyn
FC. 1973. Pulse radiolysis of nitric oxide in aqueous solution. Can. J. Chem.
51:112330
Gorman AA, Rogers MA. 1989. CRC
Handbook of Organic Photochemistry,
ed. JC Scaiano, vol. II, pp. 22947. Boca
Raton: CRC Press
Tennyson J, Noble CJ. 1986. Low-energy
electron scattering by the NO molecule.
J. Phys. B. 19:402533
Tronc M, Huetz A, Landau M, Pichou F,
Reinhardt J. 1975. Resonant vibrational
excitation of the NO ground state by electron impact in the 0.10.3 eV energy
range. J. Phys. B. 8:116069
Burrow PD. 1974. Temporary negative
ion formation in NO and O2. Chem. Phys.
Lett. 26:26568
Szmytkowski C, Maciag K. 1991. Total
cross section for electron impact on nitrogen monoxide. J. Phys. B. 24:427379
Bartberger MD, Fukuto JM, Houk KN.
2001. On the acidity and reactivity of
HNO in aqueous solution and biological systems. Proc. Natl. Acad. Sci. USA
98:219498
Stanbury DM. 1989. Reduction potentials
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
involving inorganic free radicals in aqueous solution. Adv. Inorg. Chem. 33:69
138
Janaway GA, Zhong M, Gatev GG,
Chabinyc ML, Brauman JI. 1997.
[FHNO]: An intermediate in a spin
forbidden proton transfer reaction. J. Am.
Chem. Soc. 119:1169798
Janaway GA, Brauman JI. 2000. Direct observation of spin forbidden proton
transfer reactions: 3NO + HA 1HNO
+ A. J. Phys. Chem. A 104:179598
Bartberger MD, Liu W, Ford E, Miranda
KM, Switzer C, et al. 2002. The reduction
potential of nitric oxide (NO) and its importance to NO biochemistry. Proc. Natl.
Acad. Sci. USA 99:1095863
Shafirovich V, Lymar SV. 2002. Nitroxyl
and its anion in aqueous solutions: spin
states, protic equilibria, and reactivities
toward oxygen and nitric oxide. Proc.
Natl. Acad. Sci. USA 99:734045
Shafirovich V, Lymar SV. 2003. Spinforbidden deprotonation of aqueous nitroxyl (HNO). J. Am. Chem. Soc. 125:
654752
Bazylinski DA, Hollocher TC. 1985. Evidence from the reaction between trioxodinitrate(II) and 15NO that trioxodinitrate(II) decomposes into nitrosyl hydride
and nitrite in neutral aqueous solution. Inorg. Chem. 24:428588
Ehman DL, Sawyer DT. 1968. Electrochemistry of nitric oxide and of nitrous
acid at a mercury electrode. J. Electroanal. Chem. 16:54149
Benderskii VA, Krienko AG, Ponomarev
EA. 1989. Homogenous and electrode reactions of NO2
2 ions. Sov. Electrochem.
25:15461
Liochev SI, Fridovich I. 2003. The
mode of decomposition of Angelis salt
(Na2N2O3) and the effects thereon of oxygen, nitrite, superoxide dismutase and glutathione. Free Radic. Biol. Med. 34:1399
404
Peslak J Jr, Klett DS, David CW. 1971.
An ab initio molecular orbital study of the
4 Dec 2004
13:24
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
NITROXYL (HNO)
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
351
ide (NO) and nitroxyl (NO) using cysteine. Biochem. Biophys. Res. Commun.
201:5462
Doyle MP, Mahapatro SN, Broene RD,
Guy JK. 1988. Oxidation and reduction of
hemoproteins by trioxodinitrate(II). The
role of nitrosyl hydride and nitrite. J. Am.
Chem. Soc. 110:59399
Wong PSY, Hyun J, Fukuto JM, Shiroda FN, DeMaster EG, Nagasawa HT.
1998. The reaction between nitrosothiols
and thiols: generation of nitroxyl (HNO)
and subsequent chemistry. Biochemistry
37:536271
Shoeman DW, Nagasawa HT. 1998. The
reaction of nitroxyl (HNO) with nitrosobenzene gives cupferron (N-nitrosophenylhydroxylamine). Nitric Oxide Biol.
Chem. 2:6672
Woo HA, Chae HZ, Hwang SC, Yang KS,
Kim K, Rhee SG. 2003. Reversing the
inactivation of peroxiredoxins caused by
cysteine sulfinic acid formation. Science
300:59294
Wink DA, Feelisch M, Fukuto J,
Christodoulou D, Jourdheuil D, et al.
1998. The cytotoxicity of nitroxyl: possible implications for the pathophysiological role of NO. Arch. Biochem. Biophys.
351:6674
Reif A, Zecca L, Riederer P, Feelisch M,
Schmidt HHHW. 2001. Nitroxyl oxidizes
NADPH in a superoxide dismutase inhibitable manner. Free Radic. Biol. Med.
30:8038
Liochev SI, Fridovich I. 2001. Copper,
zinc superoxide dismutase as a univalent NO-oxidoreductase and as a dichlorofluorescin peroxidase. J. Biol. Chem.
38:3525357
Liochev SI, Fridovich I. 2002. Nitroxyl
(NO-): a substrate for superoxide dismutase. Arch. Biochem. Biophys. 402:166
71
Huang PC, Kosower EM. 1968. Diazenes.
III. Properties of phenyldiazene. J. Am.
Chem. Soc. 90:236776
Kosower EM. 1971. Monosubstituted
4 Dec 2004
13:24
352
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
FUKUTO ET AL.
diazenes (diimides). Surprising intermediates. Acc. Chem. Res. 4:19398
Lemal DM, Rave TW. 1965. Diazenes
from Angelis salt. J. Am. Chem. Soc. 87:
39394
Miranda KM, Espey MG, Yamada K, Krishna M, Ludwick N, et al. 2001. Unique
oxidative mechanisms for the reactive nitrogen oxide species, nitroxyl anion. J.
Biol. Chem. 276:172027
Miranda KM, Yamada K, Espey MG,
Thomas DD, DeGraff W, et al. 2002. Further evidence for distinct reactive intermediates from nitroxyl and peroxynitrite:
effects of buffer composition on the chemistry of Angelis salt and synthetic peroxynitrite. Arch. Biochem. Biophys. 401:134
44
Murphy ME, Seis H. 1991. Reversible
conversion of nitroxyl anion to nitric oxide by superoxide dismutase. Proc. Natl.
Acad. Sci. USA 88:1086064
Fukuto JM, Hobbs AJ, Ignarro LJ. 1993.
Conversion of nitroxyl (HNO) to nitric
oxide in biological systems: the role of
physiological oxidants and relevance to
the biological activity of HNO. Biochem.
Biophys. Res. Commun. 196:70713
Hobbs AJ, Fukuto JM, Ignarro LJ. 1994.
Formation of free nitric oxide from Larginine by nitric oxide synthase: direct
enhancement of generation by superoxide dismutase. Proc. Natl. Acad. Sci. USA
91:1099296
Lin R, Farmer PJ. 2000. The HNO adduct
of myoglobin: synthesis and characterization. J. Am. Chem. Soc. 122:239394
Sulc F, Immoos CE, Pervitsky D, Farmer
PJ. 2004. Efficient trapping of HNO by
deoxyhemoglobin. J. Am. Chem. Soc.
126:1096101
Miranda KM, Paolocci N, Katori T,
Thomas DD, Ford E, et al. 2003. A biochemical rationale for the discrete behavior of nitroxyl and nitric oxide in the cardiovascular system. Proc. Natl. Acad. Sci.
USA 100:9197201
Bonner FT, Ravid B. 1975. Thermal de-
62.
63.
64.
65.
66.
67.
68.
69.
70.
4 Dec 2004
13:24
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
NITROXYL (HNO)
71.
72.
73.
74.
75.
76.
77.
78.
79.
353
4 Dec 2004
13:24
354
90.
91.
92.
93.
94.
95.
96.
97.
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
FUKUTO ET AL.
of N-hydroxyguanidine compounds: a
chemical model for the oxidation of Nhydroxy-L-arginine by nitric oxide synthase. J. Med. Chem. 36:266670
Cho JY, Dutton A, Miller T, Houk
KN, Fukuto JM. 2003. Oxidation of Nhydroxyguanidines by copper(II): model
systems for elucidating the physiological chemistry of the nitric oxide (NO)
biosynthetic intermediate N-hydroxyl-Larginine. Arch. Biochem. Biophys. 417:
6576
Hobbs AJ, Fukuto JM, Ignarro LJ. 1994.
Formation of free nitric oxide from Larginine by nitric oxide synthase and enhancement of generation by superoxide
dismutase. Proc. Natl. Acad. Sci. USA
91:1099296
Schmidt HHHW, Hofman H, Schindler U,
Shutenko ZS, Cunningham DD, Feelisch
M. 1996. No NO from NO synthase. Proc.
Natl. Acad. Sci. USA 93:1449297
Rusche KM, Spiering MM, Marletta
MA. 1998. Reactions catalyzed by
tetrahydrobiopterin-free nitric oxide synthase. Biochemistry 37:1550312
Adak S, Wang Q, Stuehr DJ. 2000. Arginine conversion to nitroxide by tetrahydrobiopterin-free neuronal nitric oxide
synthase. J. Biol. Chem. 275:3355461
Clarkson RB, Norby SW, Smirnov A,
Boyer S, Vahidi N, et al. 1995. Direct
measurement of the accumulation and
mitochondrial conversion of nitric oxide
within Chinese hamster ovary cells using
an intracellular electron paramagnetic resonance technique. Biochim. Biophys. Acta
1243:496502
Zhao X-J, Sampath V, Caughey WS. 1995.
Cytochrome c oxidase catalysis of reduction of nitric oxide to nitrous oxide.
Biochem. Biophys. Res. Commun. 212:
105460
Poderoso JJ, Carreras MC, Schopfer F,
Lisdero CL, Riobo NA, et al. 1999. The
reaction of nitric oxide with ubiquinol:
kinetic properties and biological significance. Free Radic. Biol. Med. 26:92535
4 Dec 2004
13:24
AR
AR232-PA45-14.tex
AR232-PA45-14.sgm
LaTeX2e(2002/01/18)
P1: GCE
NITROXYL (HNO)
107. Sidorkina O, Espey MG, Miranda KM,
Wink DA, Laval J. 2003. Inhibition of
poly(ADP-ribose) polymerase (PARP) by
nitric oxide and reactive nitrogen species.
Free Radic. Biol. Med. 35:143138
108. Shiva S, Crawford JH, Ramachandran
A, Ceaser EK, Hillson T, et al. 2004.
Mechanisms of the interaction of nitroxyl with mitochondria. Biochem. J.
doi:10.1042/BJ20031758
109. Espey MG, Miranda KM, Thomas DD,
Wink DA. 2002. Ingress and reactive
chemistry of nitroxyl-derived species
within human cells. Free Radic. Biol.
Med. 33:82734
110. Ma XL, Cao F, Liu GL, Lopez BL,
Christopher TA, et al. 1999. Opposite effects of nitric oxide and nitroxyl on postischemic myocardial injury. Proc. Natl.
Acad. Sci. USA 96:1461722
111. Pagliaro P, Mancardi D, Rastaldo R,
Penna C, Gattullo D, et al. 2003. Nitroxyl
affords thiol-sensitive myocardial protective effects akin to early preconditioning.
Free Radic. Biol. Med. 34:3343
112. Takahira R, Yonemura K, Fujise Y,
Hishida A. 2001. Dexamethazone attenuates neutrophil infiltration in the rat kidney in ischemia/reperfusion injury: the
possible role of nitroxyl. Free Radic. Biol.
Med. 31:80915
113. Kim W-K, Choi Y-B, Rayudu PV, Das
P, Asaad W, et al. 1999. Attenuation of
NMDA receptor activity and neurotoxicity by nitroxyl anion. Neuron 24:46169
355
7 Jan 2005
14:2
AR
AR232-PA45-15.tex
AR232-PA45-15.sgm
P1: GCE
LaTeX2e(2002/01/18)
10.1146/annurev.pharmtox.45.120403.100124
INTRODUCTION
Cancer is the second most common cause of death in developed countries and
is a rising health problem in less developed parts of the world. The diagnosis
of cancer carries great physical and mental suffering for affected individuals and
poses a significant burden on the health care system. For many tumors, conventional
management strategies (surgery, radiation, chemotherapy) have high toxicity with
0362-1642/05/0210-0357$14.00
357
7 Jan 2005
14:2
358
AR
AR232-PA45-15.tex
TIBES
TRENT
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
KURZROCK
marginal efficacy. The consensus that has emerged among investigators is that
surmounting the cancer therapeutic problem will be greatly facilitated by an indepth understanding of the molecular genetics underlying individual malignancies.
Autonomous cell growth resulting in tissue invasion and metastases is the defining feature of all malignant neoplasms (1). Cancers do not necessarily arise solely
as a result of an accelerated rate of cell proliferation. Rather, they are the consequence of an imbalance between the rate of cell-cycle progression (cell division)
and cell growth (cell mass) on one hand and programmed cell death (apoptosis)
on the other. Researchers now recognize that aberrant cellular signal transduction pathways play a vital role in driving this imbalance and hence in malignant
transformation (1).
Perhaps one of the most critical groups of signaling molecules involved in normal and abnormal cellular regulation are the tyrosine kinases (2). These proteins
constitute a family of enzymes that catalyze the phosphorylation of the tyrosine
residues of various target molecules. This process controls fundamental cellular
processes including cell cycle, migration, metabolism, proliferation, differentiation, and survival (2). Importantly, several tyrosine kinases are aberrantly expressed
in malignancies. The underlying defects may include, but are not limited to, mutation, hybrid gene formation, amplification, and perturbation of transcriptional
machinery (3).
In this review, we will highlight the role of select tyrosine kinasesBcr-Abl,
KIT, and platelet-derived growth factor receptor (PDGFR)in the clinical setting.
A specific inhibitor (imatinib mesylate) has been developed against these kinases,
and this compound demonstrates definitive therapeutic activity. More recently,
other kinases, including epidermal growth factor receptor (EGFR) and the vascular
endothelial growth factor (VEGF) system, have also been targeted successfully.
On the basis of the knowledge gained in the emerging field of molecular cancer
therapeutics, scientists are now developing a wealth of new compounds.
7 Jan 2005
14:2
AR
AR232-PA45-15.tex
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
359
7 Jan 2005
14:2
360
AR
AR232-PA45-15.tex
TIBES
TRENT
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
KURZROCK
pockets, has been the focus of inhibitor design that exploits differences in kinase
structure and pliability in order to achieve selectivity. Imatinib mesylate [also
known as Gleevec (USA), Glivec (Europe), STI 571, or CGP57148] has shown
remarkable clinical activity in Philadelphia chromosomepositive leukemias, in
gastrointestinal stromal tumors (GIST), and in several unusual tumors with alterations in the PDGF system.
Imatinib mesylate was developed from a lead compound identified in a highthroughput compound screening program searching for protein kinase C and PDGF
receptor inhibitors (13). The initial compound was a phenylaminopyrimidine that
was modified to increase cellular activity, solubility, and oral bioavailability (16).
Imatinib mesylate occupies the nucleotide-binding cleft of the Bcr-Abl protein
tyrosine kinase, preventing access of ATP to the substrate and thus competitively
inhibiting phosphorylation of downstream effector molecules (14).
In pioneering work, Druker and coworkers demonstrated that imatinib mesylate
suppressed proliferation of BCR-ABLpositive chronic myelogenous leukemia
(CML) cells in vitro (15). Normal hematopoietic progenitors were mostly unaffected (15). Imatinib mesylate also showed activity against Philadelphia chromosomepositive acute lymphoblastic leukemia (ALL) cells and in in vivo models
(17). This compound is also an effective inhibitor of the PDGF receptor tyrosine
kinase and kit (CD 117) (stem cell factor receptor) tyrosine kinases (18). Imatinib
mesylate is very specific, with 50% inhibiting concentrations (IC50s) of 188 nM
for c-Abl, 413 nM for c-Kit, and 386 nM for PDGFR-, as opposed to IC50s of
>10,000 nM for most of the other cellular tyrosine kinases (13). These observations laid the groundwork for the use of imatinib mesylate in the clinical setting,
with potential for killing tumor cells harboring the target kinases without harm to
normal host tissue. Imatinib mesylate shows striking antitumor effects in Bcr-Abl
positive (Philadelphia chromosomepositive) leukemias, GISTs, with activating
KITmutations, and in a variety of cancers with alterations in the PDGF system
(1942) (Tables 1 and 2).
Bcr-Abl
Bcr-Abl
KIT
KIT or
PDGFR-
PDGF
PDGFR-
PDGFR-
Gastrointestinal stromal
tumor
Dermatofibrosarcoma
protuberans
Hypereosinophilic
syndrome
Chronic myelomonocytic
leukemia
Anecdotal cases
up to 90%
(35)
(3134)
(29, 30)
(2528)
LaTeX2e(2002/01/18)
TEL-PDGFR
Anecdotal cases
50%
(2224)
(1921)
(36, 39)
Reference
AR232-PA45-15.sgm
FIP1L1PDGFR
Col1/PDGF
FIP1L1-PDGFR
KIT [Phe522sys]
Responses short-lived
CHR 520%
p210Bcr-Abl
p190Bcr-Abl
4090%, depending
on criteria
Responses durable
Refer to hematologic response
p210Bcr-Abl
KIT mutation
(exons 9, 11)
Comment
Response rate
Molecular
aberration
AR232-PA45-15.tex
Target
AR
Tumor type
14:2
TABLE 1
7 Jan 2005
P1: GCE
361
1015
90100
40% at
6 months
>20
(41, 42)
(19, 20)
(19, 21)
(40)
Number of patients in most studies ranges from about 100 to more than 1000.
Abbreviations: ALL = acute lymphoblastic leukemia; CML = chronic myelogenous leukemia; CCR = complete cytogenetic response; CHR = complete hematologic remission.
2040
Not stated
Median response
duration 3 months
median response
duration 6 months
LaTeX2e(2002/01/18)
Blast crisis
40
10
520
4060
510
520
75
AR232-PA45-15.sgm
100
1520
3040
Accelerated
Phase CML
TRENT
(38, 39)
(36, 37)
Reference
TIBES
>95
Markedly superior to
standard interferon-
and cytarabine
Comment
AR232-PA45-15.tex
90
40
95
>95
Survival at
18 mos (%)
AR
60
>95
75
>95
Chronic Phase
CML (previously untreated)
Progression free
survival at 12 mos (%)
CCR (%)
CHR (%)
Stage/Status of disease
362
14:2
TABLE 2
7 Jan 2005
P1: GCE
KURZROCK
7 Jan 2005
14:2
AR
AR232-PA45-15.tex
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
363
7 Jan 2005
14:2
364
AR
AR232-PA45-15.tex
TIBES
TRENT
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
KURZROCK
assessed by the sensitive polymerase chain reaction test that can detect BCR-ABL
transcripts from one leukemic cell among 105 normal cells. Thirty-nine percent
of imatinib-treated patients achieved more than a 3 log reduction in BCR-ABL
transcripts. For these patients, the probability of progression-free survival at 24
months was 100% (36).
Results of imatinib mesylate in patients with chronic phase CML who had failed
interferon- were also impressive. Complete hematological responses were reported in 95% of patients. Major and complete cytogenetic response rates were approximately 60% and 40%, respectively. In general, hematologic responses were
observed within days to weeks. Progression-free survival after 18 months followup was 89% (39).
Imatinib mesylate is now used as front-line therapy for chronic-phase CML.
Complete hematologic remission is expected by three months with major (<35%
Philadelphia chromosomepositive metaphases) or complete cytogenetic response
by 612 months. In patients who do not achieve these milestones, the imatinib
mesylate dose can be increased or a different treatment strategy may be considered
(58).
7 Jan 2005
14:2
AR
AR232-PA45-15.tex
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
365
(20, 21). This is likely a consequence of decreased marrow reserve and progression
of underlying disease. Combination therapies of imatinib mesylate with more
conventional chemotherapy or other investigational agents are being studied.
7 Jan 2005
14:2
366
AR
AR232-PA45-15.tex
TIBES
TRENT
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
KURZROCK
(g) using dual src-abl inhibitors (such as BMS-354825), which impose less stringent conformational requirements on ABL for kinase inhibition. Indeed, BMS354825 has proved effective in animal models and clinical trials are underway (72).
The efficacy of these strategies may depend on the mechanism of resistance,
which could vary among patients. Although a large body of data implicates BCRABL mutations in the emergence of imatinib resistance, cells with mutated BCRABL often do not make up the predominant population in resistant disease. Hence,
other mechanisms of resistance must be operative in some individuals. Loss of
the Bcr-Abl kinase target or activation of pathways that supplant the role of BcrAbl may play a role (71). Approaches that target Bcr-Abl function or levels may
be moot for persons in whom molecular pathways other than Bcr-Abl mediate
resistance to imatinib mesylate.
7 Jan 2005
14:2
AR
AR232-PA45-15.tex
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
367
four weeks. As in CML, new mutations in this fusion gene may lead to resistance.
Rarely, patients with systemic mast cell disease also carry a FIP1L1-PDFGR fusion gene; patients harboring this mutation show response to imatinib mesylate
treatment (26).
Atypical CML
In several cases of atypical CML, the BCR region is fused to PDGFR because of a
translocation between chromosomes 4 and 22 instead of the usual t(9:22). Rapid
responses to imatinib mesylate have been reported in patients with this variant,
demonstrating the activity against the PDGFR tyrosine kinase in vivo in these
individuals (74).
Dermatofibrosarcoma Protuberans
Dermatofibrosarcoma protuberans is an uncommon, low-grade, fibrohistiocytic
tumor of intermediate malignant potential. This neoplasm represents a unique
molecular situation where the PDGF ligand, rather than the PDGF receptor itself, is altered. Patients with dermatofibrosarcoma protuberans harbor a t(17, 22)
translocation that generates a Col1-PDGF fusion gene (75). Fusion to Col1 enhances PDGF action by allowing constitutive expression of Col1-PDGF ligand
and constitutive PDGFR kinase activation through autocrine stimulation. Exposure of primary cultures of dermatofibrosarcoma protuberans to imatinib mesylate
in vitro has been shown to inhibit cell growth (76). Further, patients with this tumor
respond to imatinib mesylate even in the case of inoperable, metastatic disease (29,
30).
7 Jan 2005
14:2
368
AR
AR232-PA45-15.tex
TIBES
TRENT
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
KURZROCK
145-kD receptor tyrosine kinase (79, 80). KIT is similar in structure and function to
the PDFGR or Flt3-receptor. It is expressed by many cells including hematopoietic
progenitor cells, germ cells, and mast cells (81). Physiologic functions of KIT
include cell survival, proliferation, differentiation, adhesion, and apoptosis (82,
83).
Several activating mutations of KIT lead to constitutive, ligand-independent
activation of the receptor tyrosine kinase and intracellular downstream pathways
including STAT, PI3K and MAPK (82, 84). Therefore, KIT expression is thought to
represent a crucial step in tumorigenesis. Some GIST tumors lack a mutation in the
KIT gene but still possess an activated, phosphorylated KIT protein (85). Mechanisms that might account for this finding include other undetected KIT mutations,
KIT ligand up-regulation, KIT heterodimerization, or alteration of phosphatases
that inhibit KIT (85). The natural ligand of KIT is stem cell factor. Interestingly,
6070% of GIST also express the CD34 antigen, a hematopoietic stem cell marker
of unknown function (86).
Prior to the availability of imatinib mesylate, prognosis of GIST was grave.
Conventional chemotherapy had response rates close to zero. This changed dramatically with the use of imatinib mesylate. The first patient reported had rapidly
progressive GIST, resistant to chemotherapy but responded to imatinib mesylate
with a complete metabolic response and tumor shrinkage (87). In subsequent
large studies for recurrent or advanced GIST, imatinib mesylate exhibited overall response rates (stable disease or partial response) anywhere from >40% (by
conventional response criteria) to 85% meaningful clinical responses (2224, 88).
The value of conventional response criteria [assessed by imaging studies, usually
in the form of computerized tomography (CT) scans] is in doubt because >90% of
treated patients showed clinical benefit, as manifest by long-term relief of cancerrelated symptoms. Patients should, therefore, be kept on the medication if they do
well symptomatically. Clinical improvement often occurs within one to two days.
Remarkably, positron emission tomography (PET) scans demonstrate metabolic
turn off of the tumor within several days and are highly predictive of anatomic
response (Figure 2). Conventional CT scans, on the other hand, may be misleading. They may demonstrate stability of disease or even disease growth for months,
despite clear-cut PET responses. On this basis, Choi and colleagues (89) proposed
new CT response criteria, including a greater than 10% decrease in tumor size
(rather than the usual greater than 50% decrease) or a greater than 15% decrease
in Hounsfield units, a measurement of tumor density on CT scan. Of particular
importance is that responses continue to increase with duration of treatment (23).
This is in contrast to chemotherapy, in which lack of early response predicts the
futility of further treatment. In the large phase III trials, progression-free survival
at 12 months was close to 70%, and overall survival was about 85%. Until now,
there has been no established dose response difference between 400, 600, and 800
mg imatinib mesylate (22, 90). However in patients failing treatment with 400 mg
of imatinib mesylate per day, increase to 800 mg per day of imatinib mesylate can
still be effective in up to 7% of patients (J. Trent, unpublished data). Response rate
7 Jan 2005
14:2
AR
AR232-PA45-15.tex
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
369
Figure 2 PET scan of a patient with GIST before and after treatment with imatinib mesylate.
Positron emission tomography (PET) scan uses a small amount of radioactive glucose [(18F)
fluorodeoxyglucose (FDG, FDG-PET)] injected intravenously. This material enriches in areas
of increased metabolic activity, such as in tumors. Emitted gamma ray photons are detected
with a scanner reflecting cell/tumor metabolism and showing tumor distribution in vivo. PET
scans can be used for diagnosis, staging, and monitoring treatment of cancers. This scan
shows a GIST patient before and after treatment with imatinib mesylate. Dark areas represent
tumor (pretreatment). These areas disappear posttreatment.
appears to correlate with site of mutation. Mutations in exon 11 of KIT are more
favorable than mutations in exon 9. The least favorable prognostic group in GIST
lacks the KIT mutation and has no other identifiable mutations (91).
Taken together, these observations suggest that the molecular defect in GIST
is inextricably related to tumor response, and that evaluation of response with
new targeted therapies may require clinical and imaging endpoints that differ from
those established over the years for chemotherapy.
7 Jan 2005
14:2
370
AR
AR232-PA45-15.tex
TIBES
TRENT
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
KURZROCK
involving the tyrosine kinase domain of KIT (92). Pardanani et al. (28) prospectively treated 10 adults suffering from symptomatic systemic mast cell disease
with imatinib mesylate at a dose of either 100 mg or 400 mg per day. Five of
the patients had a measurable response to the drug, four of whom had important
mast cell cytoreduction and two of whom had complete clinical and histological
remission. Three of the five patients with eosinophilia had major responses. The
other two, who did not respond to treatment, were the only patients with the KIT
Asp816Val mutation. It appears that these KIT mutations confer resistance to imatinib mesylate by interfering with the binding of the drug to the enzymatic site of
the KIT molecule (27).
To date, two imatinib mesylatesensitive molecular genetic defects have been
identified in mast cell disease. Akin et al. (25) reported a point mutation within the
transmembrane segment of KIT that resulted in a substitution of a phenylalanine
residue by a cysteine at codon 522 in a patient who was amenable to treatment
with imatinib mesylate. Pardanani et al. (26) demonstrated that FIP1L1-PDGFRA
is the therapeutic target of imatinib mesylate in the specific subset of patients with
mast cell disease and associated eosinophilia and that virtually all of these patients
respond to imatinib.
7 Jan 2005
14:2
AR
AR232-PA45-15.tex
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
Agent
371
Target
Reference
Imatinib
mesylate
Bcr-Abl
PDGFR
Kit
Gefitinib
EGFR
(127)
Cetuximab
EGFR
Colorectal cancer
Head and neck cancer
(97, 104)
Colorectal cancer
Renal cell carcinoma
(99, 100,
133)
Bevacizumab VEGF
Trastuzumab
(28, 35,
37, 73,
93, 94,
96)
(102, 103)
Abbreviations: EGFR = epidermal growth factor receptor; FDA = Federal Drug Administration; PDGFR = platelet-derived
growth factor receptor: VEGF = vascular endothelial growth factor.
400600 mg/day, but 800 mg/day are easily tolerated and may yield better results.
Pediatric doses are calculated by body surface area (93, 94).
7 Jan 2005
14:2
372
AR
AR232-PA45-15.tex
TIBES
TRENT
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
KURZROCK
to the intracellular domain with tyrosine kinase activity (107, 108). Endogenous
activating ligands are EGF, TGF-alpha, heparin-binding EGF, amphiregulin, betacellulin, epiregulin, and many others (105, 109). Transactivation through G-protein
coupled receptors and cytokines plays a role (109). ErbB family receptors are
widely expressed in all tissues where they regulate diverse functions, including
mitogenesis, differentiation, and cell survival (105). Downstream target activation includes PLC , Ras-Raf-MEK-MAPK (gene transcription and proliferation),
phosphatidylinositol-3 kinase (PI3K)/Akt (cell survival), the tyrosine kinase Scr,
the stress-activated protein kinases, PAK-JNKK, JNK, and the signal transducers
and activators of transcription (STAT) (110, 111). Of particular importance for the
ErbB receptor family is the ability of Erb2/Her2Neu to form heterodimers with the
other receptor subunits. ErbB2 does not have an extracellular ligand but is the most
oncogenic ErbB receptor-dimer known (109). The EGFR degradation constitutes
an important regulatory mechanism. After ligand binding, the receptor is internalized, with signal termination often within seconds, and either further endocytotic
degradation or recycling of receptor components to the cell surface for repeated
signaling (112).
Mechanisms mediating transformation include receptor overexpression, gene
amplification, activating mutations, alterations in the dimerization process, and
structural rearrangements (reviewed and referenced in 105). Furthermore activation of autocrine growth factor loops (113) and deficiency of specific phosphatases
may be of importance, as well. Receptor and ligand overexpression and gene amplification are the common causes of oncogenic transformation (114). Indeed, the
rationale for targeting the EGF receptor tyrosine kinases is based on the receptor
overexpression discerned in many human malignancies including colorectal, head
and neck, esophageal, ovarian, cervical, breast, endometrial, and nonsmall cell
lung cancer (105, 115, 116).
Several strategies to interfere with aberrant EGFR signaling have emerged. The
most successful in the clinic are antibodies that block the extracellular ligandbinding site (preventing ligand activation of the receptor) and small molecule
tyrosine kinase inhibitors that suppress the intracellular tyrosine kinase. To date,
the small molecule tyrosine kinase inhibitors Gefitinib (ZD1839, Iressa) (117)
and the antibody Cetuximab (118) have been approved by the FDA for use in lung
and colorectal cancer, respectively. Trastuzumab, which targets Erb-2 (Her2/neu),
is approved for treatment of Her2/neu-positive breast cancer. Numerous other
inhibitors of the EGFR machinery are in development (reviewed in 119).
Initially, the modest response rates to some of these compounds were considered
disappointing. However, recent data demonstrate that it is possible to identify
subsets of patients whose tumors have mutations in the targeted kinase that confer
profound susceptibility to the inhibitor.
Trastuzumab (Herceptin)
The approval of Trastuzumab in 1998 for the treatment of breast cancer is a milestone in the field of EGFR-directed therapy. It is now used worldwide for breast
cancer management.
7 Jan 2005
14:2
AR
AR232-PA45-15.tex
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
373
Her2/neu acts as a major signaling partner for other EGFR family members
by forming heterodimers with potent signaling activity leading to proliferative
and antiapoptotic effects. Her2/neu is amplified/overexpressed in about 30% of
breast cancers, and correlates with a poor outcome (101, 120). Trastuzumab is an
anti-ErbB-2 receptor (Her2/neu) humanized monoclonal antibody with benefit in
Her2/neu-positive metastatic breast cancer as a single agent (102). Compared to
chemotherapy alone, it demonstrates statistically significant increases in response
rate, time to progression, and survival time when combined with chemotherapy
(103). Benefit from Trastuzunab is only derived in Her2/neu positive patients, however, so selection of the study population for new targeted agents is very important.
Gefitinib (Iressa)
Gefitinib is an oral, highly bioavailable, EGFR-specific, anilinoquinazoline, smallmolecule inhibitor. It binds to the ATP site of the EGFR tyrosine kinase domain with
approximately 100-fold increased affinity compared to other kinases and reversibly
inhibits autophosphorylation of the receptor by competitively blocking access of
ATP to the EGFR kinase domain (121, 122), which prevents downstream kinase
activation. The IC50 for the inhibition of autophosphorylation of the EGFR/Her1
receptor in intact cells is 0.033 M (123). In preclinical models, gefitinib inhibited
the growth of multiple cell lines and mouse tumor xenografts in a dose-dependent
manner and was synergistic and additive with chemotherapeutic agents (platinum,
taxanes, etoposide), radiation therapy, as well as with the monoclonal antibody
Trastuzumab (124126).
In patients with nonsmall cell lung cancer, there was no synergism when gefitinib was administered with cytotoxic agents, according to two large randomized
trials (INTACT-1 and INTACT-2) (134, 135). The combination of gefitinib with radiation therapy might be more promising (124). Even so, gefitinib is now approved
for single-agent, third-line therapy of patients with nonsmall cell lung cancer who
have failed chemotherapy, on the basis of a response rate of about 10% (127).
This modest response rate was initially considered disappointing, and the lack of
a correlation between response and EGFR overexpression was frustrating (128).
However, a recent discovery indicates that activating mutations in the ATP-binding
pocket of the EGFR kinase domain confers susceptibility to gefitinib in nonsmall
cell lung cancer, hence allowing identification of subgroups of responsive patients
(137).
Gefitinib is well tolerated. The most common side effects are skin rash and
gastrointestinal complaints. An oral dose of 250 mg of gefitinib per day is recommended. Steady-state concentrations are achieved in most patients at seven
days. Gefitinib can be used in extensively pretreated patients or those with poor
performance status, for whom a therapy with a low toxicity profile is needed.
Erlotinib (TarcevaTM)
A second small-molecule EGFR inhibitorerlotinib, TarcevaTM, OSI-774demonstrates response rates of 10% to 19% as a single agent in patients with
7 Jan 2005
14:2
374
AR
AR232-PA45-15.tex
TIBES
TRENT
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
KURZROCK
nonsmall cell lung cancer who failed chemotherapy (129). However, large randomized trials testing erlotinib in combination with chemotherapy in this disease
did not show any benefit derived from the addition of erlotinib (R. Herbst, personal
communication). Single agent TarcevaTM in patients with advanced NSCLC failing standard therapy has recently shown to have a survival advantage in a study
with 751 patients from Canada (F. Shepherd, Proc. ASW, June 2004).
Bevacizumab (Avastin)
Bevacizumab is a humanized monoclonal antibody. It binds and inhibits the VEGF
growth factor ligand. A phase III trial involving 815 patients established Bevacizumab as the first angiogenesis-targeted agent to improve overall survival in
patients with metastatic colorectal cancer and lead to FDA approval. The addition of Bevacizumab in the frontline setting to a regimen containing multiagent
chemotherapy (irinotecan, 5-fluoruracil, and leucovorin) improved overall survival
from 15.6 to 20.3 months and progression-free survival from 6.4 to 10.6 months
when compared to the chemotherapy arm (99). These results, although modest,
were statistically significant. Bevacizumab has also been tested alone and/or in
combination with different conventional cytotoxic agents in several malignancies
(i.e., breast cancer, renal cell cancer, and nonsmall cell lung cancer). Overall results were disappointing and few partial tumor regressions were observed (132).
Promising results have been seen in renal cell cancer, however (133).
7 Jan 2005
14:2
AR
AR232-PA45-15.tex
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
375
7 Jan 2005
14:2
376
AR
AR232-PA45-15.tex
TIBES
TRENT
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
KURZROCK
patients with lung cancer in whom EGFR inhibitors can elicit dramatic responses
(137).
A synthesis of the knowledge gleaned from the clinical application of tyrosine
kinase inhibitors indicates that a new paradigm for cancer therapy is emerging.
Optimal exploitation of targeted designer drugs as part of the oncology arsenal
mandates that treatment should be based on the molecular fingerprint of the tumor
rather than its anatomic locale.
7 Jan 2005
14:2
AR
AR232-PA45-15.tex
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
377
kinase inhibitors may have serious limitations in their salutary activity, our current
successes almost certainly represent the tip of the therapeutic iceberg for targeted
treatments.
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Hanahan D, Weinberg RA. 2000. The
hallmarks of cancer. Cell 100:5770
2. Schlessinger J. 2000. Cell signaling by receptor tyrosine kinases. Cell 103:21125
3. Blume-Jensen P, Hunter T. 2001. Oncogenic kinase signalling. Nature 411:355
65
4. Van Etten RA, Jackson P, Baltimore D.
1989. The mouse type IV c-abl gene product is a nuclear protein, and activation of
transforming ability is associated with cytoplasmic localization. Cell 58:66978
5. Wetzler M, Talpaz M, Van Etten RA,
Hirsh-Ginsberg C, Beran M, Kurzrock
R. 1993. Subcellular localization of Bcr,
Abl, and Bcr-Abl proteins in normal and
leukemic cells and correlation of expression with myeloid differentiation. J. Clin.
Invest. 92:192539
6. Wetzler M, Talpaz M, Yee G, Stass SA,
Van Etten RA, et al. 1995. Cell cyclerelated shifts in subcellular localization of
BCR: association with mitotic chromosomes and with heterochromatin. Proc.
Natl. Acad. Sci. USA 92:348892
7. Hubbard SR, Till JH. 2000. Protein tyrosine kinase structure and function. Annu.
Rev. Biochem. 69:37398
8. Ullrich A, Schlessinger J. 1990. Signal
transduction by receptors with tyrosine kinase activity. Cell 61:20312
9. Heldin CH. 1995. Dimerization of cell
surface receptors in signal transduction.
Cell 80:21323
10. Heldin CH, Westermark B. 1999. Mechanism of action and in vivo role of
platelet-derived growth factor. Physiol.
Rev. 79:1283316
7 Jan 2005
14:2
378
20.
21.
22.
23.
24.
25.
26.
AR
AR232-PA45-15.tex
TIBES
TRENT
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
KURZROCK
7 Jan 2005
14:2
AR
AR232-PA45-15.tex
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
36.
37.
38.
39.
40.
41.
42.
Response to imatinib mesylate in patients with chronic myeloproliferative diseases with rearrangements of the plateletderived growth factor receptor beta. N.
Engl. J. Med. 347:48187
Hughes TP, Kaeda J, Branford S, Rudzki
Z, Hochhaus A, et al. 2003. Frequency of
major molecular responses to imatinib or
interferon alfa plus cytarabine in newly
diagnosed chronic myeloid leukemia. N.
Engl. J. Med. 349:142332
OBrien SG, Guilhot F, Larson RA, Gathmann I, Baccarani M, et al. 2003. Imatinib
compared with interferon and low-dose
cytarabine for newly diagnosed chronicphase chronic myeloid leukemia. N. Engl.
J. Med. 348:9941004
Cortes J, Giles F, OBrien S, Thomas D,
Garcia-Manero G, et al. 2003. Result of
high-dose imatinib mesylate in patients
with Philadelphia chromosome-positive
chronic myeloid leukemia after failure of
interferon-alpha. Blood 102:8386
Kantarjian H, Sawyers C, Hochhaus A,
Guilhot F, Schiffer C, et al. 2002. Hematologic and cytogenetic responses to imatinib mesylate in chronic myelogenous
leukemia. N. Engl. J. Med. 346:645
52
Talpaz M, Silver RT, Druker BJ, Goldman JM, Gambacorti-Passerini C, et al.
2002. Imatinib induces durable hematologic and cytogenetic responses in patients with accelerated phase chronic
myeloid leukemia: results of a phase 2
study. Blood 99:192837
Kantarjian HM, OBrien S, Cortes JE,
Giralt SA, Rios MB, et al. 2002. Imatinib mesylate therapy for relapse after allogeneic stem cell transplantation
for chronic myelogenous leukemia. Blood
100:159095
Olavarria E, Ottmann OG, Deininger M,
Clark RE, Bandini G, et al. 2003. Response to imatinib in patients who relapse
after allogeneic stem cell transplantation
for chronic myeloid leukemia. Leukemia
17:170712
379
7 Jan 2005
14:2
380
54.
55.
56.
57.
58.
59.
60.
61.
62.
AR
AR232-PA45-15.tex
TIBES
TRENT
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
KURZROCK
63.
64.
65.
66.
67.
68.
69.
70.
71.
7 Jan 2005
14:2
AR
AR232-PA45-15.tex
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
381
7 Jan 2005
14:2
382
AR
AR232-PA45-15.tex
TIBES
TRENT
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
KURZROCK
97.
98.
99.
100.
101.
102.
7 Jan 2005
14:2
AR
AR232-PA45-15.tex
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
115.
116.
117.
118.
119.
120.
121.
122.
123.
124.
383
7 Jan 2005
14:2
384
AR
AR232-PA45-15.tex
TIBES
TRENT
AR232-PA45-15.sgm
LaTeX2e(2002/01/18)
P1: GCE
KURZROCK
125. Ciardiello F, Caputo R, Bianco R, Damiano V, Pomatico G, et al. 2000. Antitumor effect and potentiation of cytotoxic
drugs activity in human cancer cells by
ZD-1839 (Iressa), an epidermal growth
factor receptor-selective tyrosine kinase
inhibitor. Clin. Cancer Res. 6:2053
63
126. Normanno N, Campiglio M, De LA,
Somenzi G, Maiello M, et al. 2002. Cooperative inhibitory effect of ZD1839
(Iressa) in combination with trastuzumab
(Herceptin) on human breast cancer cell
growth. Ann. Oncol. 13:6572
127. Fukuoka M, Yano S, Giaccone G, Tamura
T, Nakagawa K, et al. 2003. Multiinstitutional randomized phase II trial of
gefitinib for previously treated patients
with advanced non-small-cell lung cancer.
J. Clin. Oncol. 21:223746
128. Sirotnak FM. 2003. Studies with ZD1839
in preclinical models. Semin. Oncol. 30:
1220
129. Perez-Soler R CA, Huberman M. 2004.
Final Results from a phase II study of erlotinib (Tarceva) monotherapy in patients
with advanced non-small cell lung cancer following failure of platinum-based
chemotherapy. Lung Cancer 42 (Suppl.
2):246 (Abstr.)
130. Baselga J. 2001. The EGFR as a target for
anticancer therapyfocus on cetuximab.
Eur. J. Cancer 37 (Suppl. 4):S1622
131. Bergers G, Benjamin LE. 2003. Tumorigenesis and the angiogenic switch. Nat.
Rev. Cancer 3:40110
132. Miller KD, Rugo HS, Cobleigh MA. MPKCLI. 2002. Phase III trial of capecitabine
(Xeloda) plus bevacizumab (Avastin)
versus capecitabine alone in women
133.
134.
135.
136.
137.
138.
HI-RES-PA45-15-Tibes.qxd
1/7/05
2:26 PM
Page 1
C-1
4 Dec 2004
13:21
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
P1: GCE
LaTeX2e(2002/01/18)
10.1146/annurev.pharmtox.45.120403.095731
385
No
129/Sv C57BL/6
Adenosine
deaminase
No
129/JEms C57BL/6
Adenosine kinase
No
No
129 C57BL/6
129 B6D2
A3 receptor
CD1, N12
129/Sv, N = 1
C57BL/6 N = 6
129/Sv CD1
129/Sv C57BL/6
No
129/SvJ C57BL/6
(42)
(39)
(34)
(31)
(32, 104)
(28)
(27)
References
AR232-PA45-16.sgm
Perinatal death,
die at 3 weeks
after trophoblast
rescue
Die at P4
No
No
No
No
No
No
No
Lethality
AR232-PA45-16.tex
A2A receptor
No
129/OlaHsd C57BL/6
A1 receptor
Congenic
Parent strains
Target
386
AR
Target disruption of adenosine receptors and adenosine kinase and deaminase in mice
13:21
TABLE 1
4 Dec 2004
LaTeX2e(2002/01/18)
FREDHOLM ET AL.
P1: GCE
4 Dec 2004
13:21
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
387
Metabolic Pathways
In addition to mice lacking specific adenosine receptor subtypes, there are mouse
strains in which the metabolic pathways controlling the levels of adenosine have
been genetically modified (Table 1). It is well known that adenosine levels are
regulated by adenosine kinase (phosphorylates adenosine to AMP) and adenosine
deaminase (converts adenosine to inosine).
A mouse strain with a targeted disruption of the adenosine kinase gene was
recently reported (39). These mice developed normally until birth, but they died
soon after birth. Low levels of adenine nucleotides and high levels of S-adenosyl
homocysteine are signature features of this genetic manipulation (39). It has been
shown in studies on yeast KOs that adenosine kinase plays an important role in
methyl transfer reactions (40). The fatal outcome in adenosine kinase KO mice
may be due, in particular, to the high levels of S-adenosyl homocysteine and
the consequent depression of several transmethylation reactions. For this reason,
we have to await the generation of region- and time-dependent KOs for adenosine
kinase before we can get clear information about its roles in the CNS. It is also worth
noting that adenosine may play a role in the association between cardiovascular
morbidity and hyperhomocysteinemia (41).
An adenosine deaminase KO mouse has been generated and provides a model
for increased adenosine levels (23, 42). Lack of adenosine deaminase is classically
associated with immune deficiency, but this is probably due to an accumulation of
2-deoxy-adenosine and subsequent accumulation of dATP, and not to adenosine
accumulation (43). Indeed, blockade of adenosine kinase in adenosine deaminase
4 Dec 2004
13:21
388
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
FREDHOLM ET AL.
KO mice, which is expected to massively increase adenosine accumulation, decreased thymocyte death in parallel with decreased dATP accumulation (44).
Deamination of adenosine to inosine largely, but not completely eliminates the
actions of adenosine on adenosine receptors: The A1 and particularly the A3R
can also respond to inosine, although inosine is not a full agonist (4547). Indeed,
studies on KO mice have shown that the A3 receptor can mediate some of the effects
of inosine in the immune system, but for other effects of administered inosine only
mice lacking both the A2A and the A3 receptor were unresponsive (48), suggesting
that A2A receptors are also involved in mediating the effects of inosine. This does
not necessarily mean that inosine acts on A2A receptors, however. It could mean
that inosine increases levels of adenosine, which in turn acts on A2A receptors. In
addition, inosine may influence energy levels and polyADP-ribosylation (49).
Use of Heterozygotes
Attention is most often paid to the phenotype of the homozygous KO. However,
detailed examination of heterozygotes (HZ) can also be very revealing.
a) How well adjusted is the receptor level? If receptor number is directly proportional to gene dosageas is the case in A1R and A2AR HZthis argues
against strong autoregulation of transcription. Therefore, it seems likely that
neither A1 nor A2AR levels are regulated to a major extent by the ongoing
signaling via these receptors.
b) Heterozygotes often provide a better model for the effects likely to be seen
with antagonists because it is only rarely the case that antagonists can be
given at a dose that will inhibit all the receptors all the time. An especially
relevant aspect is that caffeine in doses commonly consumed by humans
gives plasma concentrations very close to the KD for caffeine at human A1
and A2ARs (3). Because responses to adenosine are shifted to the right, and
because there are only half the normal number of receptors in heterozygous
mice, it seems possible that heterozygous mice can be used as a genetic
model for caffeine use.
c) Heterozygous mice have also been used to circumvent the problems associated with the developmental effects that can potentially confound studies
on homozygous KO mice (50). In this approach, pharmacological agents
are given to heterozygous mice at doses that are subthreshold in WT mice.
There is no biological effect in WT mice treated with a subthreshold dose of
the drug or in HZ mice treated with vehicle, but a subthreshold dose elicits
a biological effect when combined with heterozygous genetic inactivation
of the target molecule (51). This approach may be particularly useful in
examining some adenosine receptor functions where discrepancies between
the pharmacological and genetic approaches have been reported (such as the
psychostimulant effect of A2AR KO and A2AR antagonists; see below under
the section on striatum and dopamine receptors).
4 Dec 2004
13:21
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
389
4 Dec 2004
13:21
390
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
FREDHOLM ET AL.
twice that of the WT, whereas EMAX is unaffected (see Figure 1). Despite the
loss of tonic inhibition, no compensatory responses were found in other receptor
subtypes mediating similar G proteincoupled presynaptic inhibition of synaptic
transmission (27), but a wide range of potential compensatory mechanisms remain
to be explored.
Slices from homozygous A1R KO show no evidence of any remaining endogenous inhibitory influence of adenosine in the Schaffer collateral pathway in the
CA1 region of the hippocampus or at the mossy fiber synapses in the CA3 region (61). Furthermore, there is no inhibition of synaptic transmission when large
concentrations of adenosine (100 M) are applied exogenously (27). Using the
Cre-loxP system and an adeno-associated viral vector, a targeted deletion of the
A1R was induced separately in the CA1 and in the CA3 region of the hippocampus
(38). This approach holds promise for dissecting out specific pre- and postsynaptic actions of adenosine and its synaptic interactions with other molecules and
neurotransmitters, but so far no major results have been reported. Similar to the
constitutive KO model, there was no response to adenosine in the targeted inducible
knockout. Application of adenine nucleotides such as ATP was also ineffective in
hippocampal slices from KO mice (62) (see Figure 1). Together, these data suggest that the inhibitory effects of both adenosine and adenine nucleotides in the
hippocampus are mediated ultimately through the adenosine A1 receptor (27, 38),
4 Dec 2004
13:21
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
391
or alternatively, that effects of the other receptors require the presence of A1Rs, as
suggested for A2ARs (63). It will be important to determine the potency of adenosine in regulating neurotransmission via A1R in mice lacking the other receptors
to determine if there is such an interaction.
Dunwiddie and coworkers, using pharmacological tools (64), reported some
role of A3Rs in modulating the responses to A1 stimulation. This question was
re-examined recently and no significant interactions between A1 and A3Rs were
discovered using a host of different methods, including binding studies and electrophysiological studies (65). Thus, if A3Rs do play a role it is likely to be small
and indirect.
ISCHEMIA
It is generally believed that adenosine can protect tissues against the negative consequences of hypoxia or ischemia (1, 66), and that A1Rs play a particularly important
role. Hence, survival after a hypoxic challenge may be reduced if A1Rs are absent
or blocked (27). One consequence is that use of caffeine or other methylxanthines
in doses that would completely block A1Rs may be hazardous in hypoxic human
newborns. In keeping with the proposed role for adenosine acting at A1Rs, hippocampal slices taken acutely from adult mice do show greater functional recovery
from both hypoxic and ischemic insults when A1Rs are intact (27, 67). Moreover,
acute administration of an A1R antagonist did enhance ischemic damage in vivo,
giving further evidence that compensatory mechanisms may be providing protection in the knockout (68).
However, the severity of ischemic damage either in vivo or in organotypic
hippocampal slice cultures is not increased in the A1R KO model (68). The lack
of any obvious difference between the WT and the KO after pathophysiological
insult where A1Rs are considered neuroprotective is surprising. In immature brain,
blockade of A1Rs in fact attenuated ischemic injury. For example, the loss of white
matter that is a typical consequence of hypoxia in the newborn actually appears
to be mediated by adenosine acting on A1Rs (69). Thus, blockade of adenosine
receptorseven incomplete blockade like that achieved by caffeinereduces such
white matter loss (69). In addition, the consequences of prenatal hypoxic ischemia
in rats are reduced if the dams have been given caffeine (70).
Brain damage after focal ischemia has been reported to be attenuated in adult
A2AR KO mice compared with WT mice (32). On the other hand, aggravated brain
damage is observed after hypoxic ischemia in immature seven-day-old A2AR KO
mice (71). These results suggest that, in contrast to the situation in adult animals,
A2ARs play an important protective role against hypoxic ischemic brain injury in
neonates. Interestingly, a recent study using a novel approach where A2AR KO is
combined with bone marrow transplantation demonstrated that selective reconstitution of the A2AR in bone marrowderived cells of A2AR KO mice abolished the
neuroprotection against ischemic brain injury afforded by global depletion of A2AR
4 Dec 2004
13:21
392
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
FREDHOLM ET AL.
CAFFEINE
One reason why studies of adenosine and its receptors attract interest is that adenosine receptors (A1, A2A, and A2B) are the targets for the most widely used of all
psychoactive drugs, caffeine. Studies on KO animals have provided compelling
evidence that the psychostimulant effects of caffeine require blockade of A2ARs.
Caffeine has a mild stimulant effect in A2AR WT mice, but becomes a depressant of
locomotor activity in A2AR KO mice (31). Thus, A2ARs appear to be required for the
stimulant effect of caffeine (see Figure 2). In fact, caffeine dependently decreases
locomotion in A2AR KO mice over a wide range of doses (76). This effect probably
results from the other biological effects of caffeine, the blockade of A1Rs being a
candidate. Examining immediate early gene expression in WT and A2AR KO mice,
the Schiffmann group also concluded that A1R blockade was important for some
of the high-dose effects of caffeine (77). However, the role of A1R in the effects of
caffeine on motor activity is less clear. Recently, Halldner et al. (78) showed that
the A1R is not crucial for the stimulatory effect of caffeine, although the effect is
facilitated in the A1R KO mice. The results also suggest that the inhibitory effects
4 Dec 2004
13:21
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
393
Modifications
Aggressiveness
A1
A2A
A3
Increased
Increased
Not determined
Anxiety
A1
A2A
A3
Increased/no change
Increased
No change
Despair-like
A1
A2A
A3
Not determined
Decreased
Increased
Memory
A1
A2A
A3
No change
Not determined
Not determined
Motor activity
A1
A2A
A3
No change/decreased
Decreased/slightly increased
Slightly increased
Neuroprotection
A1
A2A
A3
Sensorimotor gating
A1
A2A
A3
Not determined
Reduced startle inhibition and prepulse inhibition
Not determined
Thermal nociception
A1
A2A
A3
Hyperalgesia
Hypoalgesia
Hyperalgesia/no change
Function
of higher doses of caffeine are not due to blockade of the A1R. Rather, this effect
is likely to be independent of adenosine receptors. Clearly, many more studies of
the actions of caffeine in single and double KOs are necessary to delineate which
effects are entirely due to adenosine receptor blockade and which are not.
SLEEP
One of the best-known effects of caffeine is its effect on sleep (3). There is also
considerable evidence that adenosine is an endogenous promoter of sleep (for
references see 79). Adenosine levels, particularly in the basal forebrain, increase
4 Dec 2004
13:21
394
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
FREDHOLM ET AL.
4 Dec 2004
13:21
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
395
shown if some level of adaptation also occurs after long-term, high-dose exposure
to caffeine.
Despite the fact that most attention has been focused on the role of A1 receptors,
there is increasing evidence that A2A receptors also play a role. For example, in
fetal sheep there is evidence for a tonic role of A2ARs in regulating REM sleep
state (91), and administration of A2AR agonists into the subarachnoid space close
to the preoptic area increases sleep (92). The possibility exists that the abundant
A2ARs present in the nucleus accumbens (92) or tuberculum olfactorium (93)
play a role. Thus, there are changes in A2ARs and the corresponding mRNA in
tuberculum olfactorium following sleep deprivation (93). Furthermore, the sleepinducing effect of the A2AR agonist CGS 21680 was eliminated in A2AR KO mice
(94). It will be of considerable interest to examine sleep in mice that lack both A1
and A2ARs and to determine whether caffeine has any effect in such mice.
Locomotion
Basal locomotion is marginally affected in A1R KO mice (30, 78). Thus, no differences in overall spontaneous motor activity were detected over a 23-h monitoring
period, but activity was reduced in some parts of the light-dark cycle.
4 Dec 2004
13:21
396
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
FREDHOLM ET AL.
A2ARs are highly expressed in the dorsal and ventral striatum, where they
could be involved in the physiological control of motor activity, and a major role
of A2AR stimulation is to modulate locomotor activity. In most studies performed
so far (31, 32, 101, 106), the exploratory behavior of A2AR KO mice was reduced
as compared to A2AR WT mice. As expected, treatment with the A2A agonist CGS
21680 strongly reduced locomotor activity in A2AR WT mice and had no significant
effect on A2AR KO mice (31, 32). However, this reduction of locomotor activity
en et al. (71)
is not an invariable characteristic of A2AR KO mice because as Ad
observed there is a small increase in basal locomotor activity at four weeks of
age in A2AR KO mice compared with A2AR WT mice. Locomotor behavior was
reported to be increased in A3R KO mice (75).
Parkinsons Disease
Among new therapeutic approaches for Parkinsons disease, one possibility being
investigated is modulation of dopamine-mediated striatal functions through the
blockade of A2ARs. The past ten years have witnessed significant progress in the
development and characterization of a new generation of A2AR antagonists for
use in Parkinsons disease. The motor enhancement afforded by A2A antagonists
was well documented in early pharmacological studies (19), and activity at A2ARs
reduces motor responses in both normal and dopamine-depleted animals. The
multiple benefits of A2AR inactivation (seen either after genetic deletion, as in
A2AR KO mice, or after treatment with A2A antagonists) advance the prospects of
A2AR antagonists as a novel treatment strategy for Parkinsons disease (18, 107)
(see Table 3).
Proof of principle comes from large epidemiological studies that firmly establish
an inverse relationship between caffeine consumption and the risk of developing
Parkinsons disease (108110). Studies carried out in mice support these epidemiological findings, providing evidence for neuroprotective effects of caffeine and
specific A2AR antagonists, as well as genetic deletion of the A2AR (111). Several
other pharmacological studies employing various A2AR antagonists also support
this neuroprotective effect (112, 113).
A2AR antagonism also provides symptomatic relief of surgical lesion or druginduced motor dysfunction. Catalepsy, a state where the animals remain immobile
for long periods, even if they are placed in awkward postures, can be induced by
dopamine D1 or D2 receptor antagonists or a muscarinic acetylcholine receptor agonist. In A2AR KO mice, such catalepsy was reduced as compared with A2AR WT
mice (104, 114). These results suggest that A2ARs influence not only dopamine D1
and D2 receptormediated neurotransmission but also that mediated via muscarinic
acetylcholine receptors. Interestingly, caffeine and muscarinic antagonists act in
synergy to inhibit haloperidol-induced catalepsy (115). The results on catalepsy
show that deletion of the A2A receptor alleviates dysfunction of basal ganglia motor
circuitry caused by drugs acting at dopamine and acetylcholine receptors. These
preclinical studies led to the clinical trial of the A2AR antagonist KW6002 in patients with Parkinsons disease, and the initial results were encouraging (116, 117).
4 Dec 2004
13:21
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
397
Animal models
Effect
References
Parkinsons
disease
MPTP
MPTP
6-OHDA
Haloperidol-catalepsy
D2R KO-induced
hypolocomotion
MPTP-bradykinesia
in monkey
(111)
(111, 153)
(112)
(111, 114)
(154, 155)
3-NP
3-NP
3-NP
(111)
(111)
(120)
Huntingtons
disease
3-NP
(156)
Multiple
sclerosis
Experimental
autoimmune
encephalomyelitis
(EAE)
(132)
Stroke (ischemic
brain injury)
Hypoxia
(27)
(32)
Hypoxic-ischemia
Prenatal hypoxic
ischemia
MCAO
MCAO
MCAO
Carbon monoxide
Alzheimers
disease
Beta-amyloid
aggregation
(69)
(71)
(68)
(72)
(75)
(157)
Finally, recent studies with A2AR KO mice and pharmacological agents suggest the possibility of another potentially beneficial effect of A2AR antagonists,
namely prevention of the development of dyskinesia after repeated treatment with
L-DOPA (118, 119). Debilitating motor complications such as dyskinesia are the
major limiting factors of management in the later stages of Parkinsons disease.
Thus the finding that repeated administration of L-DOPA did not lead to behavioral
sensitization in A2AR KO mice indicates that the A2AR may be required for the development of maladaptive changes after long-term treatment with L-DOPA (113).
This notion is further supported by recent studies in MPTP-treated nonhuman
primates, showing that coadministration of KW6002 with the dopamine agonist
apomorphine completely abolished apomorphine-induced dyskinesia (119).
4 Dec 2004
13:21
398
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
FREDHOLM ET AL.
Huntingtons Disease
The cellular localization of the A2AR in striatopallidal neurons suggests that the
A2AR may contribute to selective vulnerability to neurotoxins in Huntingtons
disease. Indeed, there is also evidence that A2ARs may play a role in Huntingtons
disease (16, 120) (Table 3). In a neurochemical model of Huntingtons disease,
pharmacological and genetic inactivation of the A2AR have been shown to attenuate
striatal damage induced by the mitochondrial toxin 3-nitropropionic acid (121) or
the excitotoxin quinolinic acid (122). However, the role of the receptors is complex,
and it is difficult at present to envision A2A antagonism as a therapy in this disorder.
The complex actions were interpreted as resulting from a balance between negative
effects owing to blockade of presynaptic A2A receptors regulating glutamate release
and positive effects owing to blockade of postsynaptic receptors (120). However,
glutamate levels are also regulated by glutamate transporters on glial cells, which
express A2AR. Therefore, the interpretation of glutamate changes is not yet clear.
Schizophrenia
Another pathological condition involving both adenosine and dopamine in the
striatumand where A2A agonists might be beneficialis schizophrenia. Patients
with schizophrenia show impaired sensorimotor gating. Normally, this gating prevents excessive irrelevant sensory stimuli from disturbing integrative mental processes in the brain. In schizophrenic patients, the impairment in sensorimotor
gating results in reduced prepulse inhibition (PPI) and reduced startle habituation.
In experimental animals, both parameters are modulated by dopaminergic and
adenosine receptor agonists and antagonists. Wang et al. (123) recently found that
startle amplitude, startle habituation, and PPI were significantly reduced in A2AR
KO mice, which provides evidence that this receptor may be involved in the regulation of these phenomena. In addition, responses to an NMDA antagonist and amphetamine were altered (123). These data suggest substances with A2A receptor agonist properties may be of interest in the development of antipsychotic drugs (124).
4 Dec 2004
13:21
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
399
A2AR WT mice but not in A2AR KO mice. The selective A2AR antagonist SCH
58261, but not the selective A1 receptor antagonist DPCPX, shortened the duration
of the loss of righting reflex induced by ethanol, thus mimicking the lack of the
receptor in A2AR-deficient mice. Caffeine (25 mg/kg) also reduced ethanol-induced
hypnotic effects. These results indicate that the activation of A2A receptors plays
a role in the hypnotic effect of ethanol.
The cessation of chronic ethanol intake or ethanol withdrawal is an experimental procedure recognized to produce seizures in mice. This convulsant activity is associated with an increase in excitatory neurotransmission in the brain.
Whereas A2AR KO mice and controls ingested similar amounts of ethanol during
forced ethanol consumption, the severity of handling-induced convulsions during
withdrawal was significantly lower in the A2AR KO mice than in A2AR WT mice.
Because the selective A2AR antagonist ZM 241385 also attenuated the intensity
of withdrawal-induced seizures, it was suggested that selective A2AR antagonists
may be useful in the treatment of alcohol withdrawal (126). The role of A2ARs
in ethanol consumption and neurobiological responses to this drug of abuse was
further characterized by Naassila et al. (127). Male and female A2AR KO mice consumed more ethanol than WT mice. This slightly higher ethanol consumption was
also related to ethanol preference. Relative to A2AR WT mice, A2AR KO mice were
found to be less sensitive to the sedative and hypothermic effects of ethanol. No
major difference in the development of tolerance to ethanol-induced hypothermia
was found between the two phenotypes, although female A2AR KO mice showed
a lower tolerance-acquisition rate. These results suggest that activating the A2ARs
may play a role in suppressing alcohol-drinking behavior and be associated with
sensitivity to the intoxicating effects of acute ethanol administration.
There is also evidence that morphine dependence is modified by A2ARs. Opiate
withdrawal was enhanced in mice lacking A2A receptors, and this enhancement
was abolished when both the cannabinoid CB1 receptor and A2AR were eliminated
(106).
Because there is considerable evidence for interactions between adenosine receptors and central stimulants (see above), for a role of adenosine in some actions
of morphine (17), for various interactions between adenosine and ethanol (128),
and because adenosine receptors are very important in regulating dopaminergic
transmission in the reward pathways in nucleus accumbens, it is important to
further examine the effects of addictive drugs in AR KO mice.
Seizures
It has long been known that adenosine can suppress repetitive neuronal firing, and a
role of adenosine as an endogenous modifier of seizures has been suspected. This
notion recently received support (129) when it was found that seizure-inducing
lesions can increase the level of adenosine kinase in astrocytes, and that this,
by reducing adenosine levels, contributes to increased seizure susceptibility. This
raises the possibility that modifying the extracellular adenosine level in brain may
be of therapeutic value against seizures. Indeed, cells that generate adenosine have
4 Dec 2004
13:21
400
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
FREDHOLM ET AL.
been transplanted into rat brain and this has led to decreased seizure susceptibility
(130, 131). In particular, activation of A1Rs appears to be an interesting target
for therapy in drug-resistant epilepsy (131). Unless there are major compensatory
mechanisms in effect, seizure thresholds would be expected to be lower in A1R
KO animals, but this awaits further investigation.
Multiple Sclerosis
There is some evidence that adenosine may play a role in multiple sclerosisat
least there are effects in an experimental model (132). Thus, in A1R KO mice the
demyelination and axonal degeneration was much more pronounced than in WT
littermates. There was also a stronger activation of microglia/macrophages. Furthermore, macrophages from A1R KO animals exhibited increased expression of
the proinflammatory genes IL-1 and matrix metalloproteinase-12 on immune activation compared to control cells from A1R WT animals (132). This would imply
first that A1Rs are very important in regulating macrophages/microglial cells. However, this is not immediately obvious from other data where these cells have been
examined (e.g., 133). Furthermore, the role of A1Rs in regulating oligodendrocyte
function and survival appears to differ between the adult spinal cord (132) and the
immature brain (69). This again emphasizes that the roles of adenosine receptors
may be complex, and that they could differ with age, location, and pathology.
Memory
Despite some hints from experiments with drugs that affect adenosine receptors,
the evidence from KO animals does not reveal any clear effect of the A1R KO
genotype on memory (30, 134). Minor effects in the water maze were suggested
(134) to be due to the altered emotional stability reported for these mice (27,
30). Long-term potentiation (LTP), an in vivo model of memory formation, has
generally been observed to be inhibited by A1R activation (135) and enhanced by
A2AR activation (136, 137). Deletion of adenosine A2ARs did not alter ongoing
synaptic transmission in either striatum (138) or nucleus accumbens (137), but
accumbens neurons showed significantly reduced LTP when the effects of the
A2AR were removed (137). LTP was reduced greatly in the mossy fiber pathway in
hippocampal slices from A1R KO mice as well as rat hippocampal slices pretreated
with an A1R antagonist (61), providing strong evidence that adenosine acting at
the A1R augments LTP in this pathway.
Anxiety
The neurobiology of anxiety, including the role of adenosine, was recently comprehensively reviewed (139). Interestingly, anxiety-related behavior in the classical
light/dark box test was increased in the A1R KO mice, as shown by a reduction
in the number of entries into as well as the total time spent in the lit compartment
compared with A1R WT mice (27, 30). The A1R KO mice also showed a decrease
in exploratory behavior in the open-field and in the hole-board, results that could
reflect an anxiogenic state in A1R KO mice. However, another strain of A1R KO
4 Dec 2004
13:21
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
401
mice with a similar genetic background displayed a normal overall level of motor
activity, with very modest behavioral changes in the direction of increased anxiety
(134). It is likely that different environmental conditions have contributed substantially to the behavioral discrepancies between the two lines. This might prompt
us to ask whether the increased sensitivity to caffeine reported in patients with
panic disorders (140) is indeed linked to a disorder of adenosine neuromodulation
at A1Rs in the brain.
A2AR KO mice showed higher rates of spontaneous anxiety-like responses
in two different anxiety-like behavioral tests, the elevated plus-maze and the
light/dark box (31, 106, 141). Thus, A2AR KO mice and at least one strain of
A1R KO mice exhibit increased anxiety, consistent with the well-known, pronounced, anxiogenic effects of high doses of caffeine. High doses of caffeine will
presumably block most of these adenosine receptor subtypes, but low doses will
not. Despite several studies using pharmacological tools and performed in rodent
models (141144) there is no clear consensus concerning the role of A1 and A2ARs
in anxiety. However, on the basis of screening tests, it has been proposed that A1R
agonists exert anxiolytic effects, whereas A1R antagonists in some cases, but not
consistently, exert anxiogenic effects. On the other hand, it is still unclear whether
the A2AR also plays a major role in anxiety states. Selective A2AR antagonists
seem to be devoid of effects in tests on rodents (141). However, recent data from
humans shed fresh light on the potential role of A2ARs in the anxiogenic effects
of caffeine. In a study conducted by Alsene et al. (145), the association between
variations in anxiogenic responses to caffeine and polymorphisms in the adenosine A1 and A2AR genes has been examined. They found a significant association
between self-reported anxiety after oral administration of 150 mg of caffeine and
two linked polymorphisms on the A2AR gene. Individuals with the 1976T/T and
the 2592Tins/Tins genotypes reported greater increases in anxiety after caffeine
administration than the other genotypic groups. Moreover, in patients with panic
disorder, a psychiatric condition characterized by recurrent panic attacks and anticipatory anxiety, a single-nucleotide polymorphism haplotype in the A2AR gene
was found to be associated with the disease (146). Alpha-melanocyte-stimulating
hormone (alpha-MSH) influences anxiety, aggressiveness, and motor activity, all
of which are also influenced by A2AR gene disruption. In A2AR KO mice, significantly increased alpha-MSH content was observed in the amygdala and cerebral
cortex. Plasma corticosterone concentration was significantly higher in A2AR KO
mice, revealing hyperactivity of their pituitary-adrenocortical axis. Results suggest
that A2ARs are involved in the control of POMC gene expression and biosynthesis
of POMC-derived peptides in pituitary melanotrophs and corticotrophs (147).
Aggression
Several studies have suggested that adenosine receptors are involved in the modulation of aggressive behavior. In agreement with the decrease of offensive behavior
induced by a selective stimulation of A1Rs (148), A1R KO mice isolated for the
resident-intruder aggression test showed enhanced aggressive behavior (27). A
4 Dec 2004
13:21
402
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
FREDHOLM ET AL.
similar enhancement was also observed in isolated male A2AR KO mice in the
resident-intruder test (31). The increased aggressiveness observed in both A1R
KO mice and A2AR KO mice is in agreement with the increase of offensive behavior induced by selective blockade of either A1 or A2ARs (M. El Yacoubi &
J.M.Vaugeois, unpublished observations). These results suggest that both adenosine receptor subtypes are involved in the effect of adenosine on aggressiveness.
The link between these effects and the increase in nervousness and irritability reported in humans (3) after chronic administration of high doses of caffeine remains
a matter of speculation.
Depression
In behavioral procedures used to screen potential antidepressants, such as tail
suspension and forced swim tests, A2AR KO mice were found to be less sensitive
to depressant challenges than their WT littermates, which were less immobile
than A2AR WT mice in both tests (149). Consistently, A2AR blockers reduced
the immobility times in tail suspension and forced swim tests. Taken together,
the results support the hypothesis that blockade of the A2AR might be an interesting
target for the development of effective antidepressant agents. Although their mode
of action in potentially alleviating mood disorders is unknown, modulation of
dopamine transmission might play a role (149). Future clinical trials with selective
A2AR antagonists as potential therapeutic agents for major depressive episodes will
help to delineate the role of adenosine in the pathophysiology of mood disorders.
Whereas A2AR antagonists have been proposed as antidepressants (149), A3R KO
mice showed an increase in the amount of time spent immobile in the two tests of
behavioral depression, the forced swim test and the tail suspension test (75).
Pain
The role of adenosine as an endogenous analgesic substance has also been evaluated
(27). A1Rs are abundant in mouse spinal cord, with the highest levels in the outer
lamina of the dorsal horns, where the density of receptors was close to that observed
in the hippocampus. A1Rs are responsible for the analgesic effects of intrathecally
administered A1 agonists. A1R KO mice react faster to thermal pain than A1R
WT mice. However, this increase is not matched by an increased sensitivity to
mechanical stimulation. The authors suggested that endogenous adenosine acting
at A1Rs decreases nociception, mediated via C fibers. These results also suggest
that the A1R may be a target for the development of antinociceptive drugs.
The response of A2AR KO mice to acute pain stimuli is slower in the hot
plate and tail-flick tests compared to A2AR WT mice (31). Similar reduced pain
responses were also found when a tail-immersion test was used (106). This higher
nociceptive threshold suggests that the peripheral lack of A2ARs predominates over
the spinal defect. Thus, depending on the site of action and the receptor activated
(A1 or A2A), adenosine may exert very different effects on pain. This variety of
effects may explain why caffeine has analgesic effects against some, but not all,
types of pain (3).
4 Dec 2004
13:21
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
403
CONCLUSIONS
Whereas deletion of genes for enzymes critically involved in adenosine metabolism
leads to lethal phenotypes, deletion of A1, A2A, and A3 receptors has rather subtle
effects and the mice are remarkably normal. This agrees well with the conclusion
drawn before, i.e., that adenosine receptors are involved in modulating physiological responses and that they are particularly important under pathophysiological
conditions. Thus, to determine the roles of the adenosine receptors, the genetically
modified mice must be subjected to various types of challenges.
The results obtained so far have both confirmed previous data and yielded some
surprises. The important role of the A1R in modulating excitatory transmission and
its role in pain transmission was expected, as was the critically important role of
A2ARs in striatal function. Among the major surprises were the noncritical role of
A1Rs in brain ischemia and in sleep and the finding that A2ARs mediate aggravated
brain damage mainly via peripheral receptors. Our examination of the literature has
also indicated studies that can and should be performed to further define the roles
of adenosine receptors in the nervous system. Because the goal of these studies is to
examine the possibility for novel drug therapies, the use of KO mice to determine
that the drugs are indeed selective is very important. Indeed, data obtained already
suggest that some of the drugs used to delineate adenosine receptor effects are not
as selective as previously hoped (36, 151, 152).
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Fredholm BB. 1996. Adenosine and
neuroprotection. In Neuroprotective
Agents and Cerebral Ischemia, ed. AR
Green, AJ Cross, pp. 25980. London:
Academic
4 Dec 2004
13:21
404
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
FREDHOLM ET AL.
13. Thong FS, Graham TE. 2002. The putative roles of adenosine in insulin- and
exercise-mediated regulation of glucose
transport and glycogen metabolism in
skeletal muscle. Can. J. Appl. Physiol.
27:15278
14. Klinger M, Freissmuth M, Nanoff C.
2002. Adenosine receptors: G proteinmediated signalling and the role of accessory proteins. Cell. Signal. 14:99
108
15. Schwarzschild MA, Chen JF, Ascherio A.
2002. Caffeinated clues and the promise
of adenosine A(2A) antagonists in PD.
Neurology 58:115460
16. Blum D, Hourez R, Galas MC, Popoli
P, Schiffmann SN. 2003. Adenosine receptors and Huntingtons disease: implications for pathogenesis and therapeutics.
Lancet Neurol. 2:36674
17. Sawynok J, Liu XJ. 2003. Adenosine in
the spinal cord and periphery: release
and regulation of pain. Prog. Neurobiol.
69:31340
18. Chen JF. 2003. The adenosine A(2A) receptor as an attractive target for Parkinsons disease treatment. Drug News Perspect. 16:597604
19. Fredholm BB, Svenningsson P. 2003.
Adenosine-dopamine interactions: development of a concept and some comments on therapeutic possibilities. Neurology 61:S59
20. Dhalla AK, Shryock JC, Shreeniwas R,
Belardinelli L. 2003. Pharmacology and
therapeutic applications of A1 adenosine
receptor ligands. Curr. Top. Med. Chem.
3:36985
21. Headrick JP, Hack B, Ashton KJ. 2003.
Acute adenosinergic cardioprotection in
ischemic-reperfused hearts. Am. J. Physiol. Heart Circ. Physiol. 285:H1797
818
22. Schulte G, Fredholm BB. 2003. Signalling from adenosine receptors to
mitogen-activated protein kinases. Cell.
Signal. 15:81327
23. Blackburn MR. 2003. Too much of a good
4 Dec 2004
13:21
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
24.
25.
26.
27.
28.
29.
30.
31.
32.
405
4 Dec 2004
13:21
406
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
FREDHOLM ET AL.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
4 Dec 2004
13:21
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
71.
72.
73.
74.
75.
76.
77.
78.
407
4 Dec 2004
13:21
408
79.
80.
81.
82.
83.
84.
85.
86.
87.
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
FREDHOLM ET AL.
lacking adenosine A1 and/or A2A receptors. Neuropharmacology 46:100817
Porkka-Heiskanen T, Alanko L, Kalinchuk A, Stenberg D. 2002. Adenosine and
sleep. Sleep Med. Rev. 6:32132
Porkka-Heiskanen T, Strecker RE, Thakkar M, Bjorkum AA, Greene RW, McCarley RW. 1997. Adenosine: a mediator
of the sleep-inducing effects of prolonged
wakefulness. Science 276:126568
Alanko L, Heiskanen S, Stenberg D,
Porkka-Heiskanen T. 2003. Adenosine kinase and 5 -nucleotidase activity after prolonged wakefulness in the cortex and the
basal forebrain of rat. Neurochem. Int.
42:44954
Mackiewicz M, Nikonova EV, Zimmerman JE, Galante RJ, Zhang L, et al. 2003.
Enzymes of adenosine metabolism in the
brain: diurnal rhythm and the effect of
sleep deprivation. J. Neurochem. 85:348
57
Ticho SR, Radulovacki M. 1991. Role
of adenosine in sleep and temperature regulation in the preoptic area of
rats. Pharmacol. Biochem. Behav. 40:33
40
Benington JH, Kodali SK, Heller HC.
1995. Stimulation of A1 adenosine receptors mimics the electroencephalographic
effects of sleep deprivation. Brain Res.
692:7985
Virus RM, Ticho S, Pilditch M, Radulovacki M. 1990. A comparison of the
effects of caffeine, 8-cyclopentyltheophylline, and alloxazine on sleep in rats.
Possible roles of central nervous system
adenosine receptors. Neuropsychopharmacology 3:24349
Rainnie DG, Grunze HCR, McCarley
RW, Greene RW. 1994. Adenosine inhibition of mesopontine cholinergic neurons: implications for EEG arousal. Science 263:68992
Chamberlin NL, Arrigoni E, Chou TC,
Scammell TE, Greene RW, Saper CB.
2003. Effects of adenosine on gabaergic
synaptic inputs to identified ventrolateral
88.
89.
90.
91.
92.
93.
94.
95.
96.
97.
4 Dec 2004
13:21
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
98.
99.
100.
101.
102.
103.
104.
105.
106.
107.
108.
109.
110.
111.
112.
113.
409
4 Dec 2004
13:21
410
114.
115.
116.
117.
118.
119.
120.
121.
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
FREDHOLM ET AL.
and striatum, respectively. J. Neurosci.
23:1098287
El Yacoubi M, Ledent C, Parmentier
M, Costentin J, Vaugeois JM. 2001.
Adenosine A2A receptor knockout mice
are partially protected against druginduced catalepsy. NeuroReport 12:983
86
Moo-Puc RE, Gongora-Alfaro JL, Alvarez-Cervera FJ, Pineda JC, ArankowskySandoval G, Heredia-Lopez F. 2003. Caffeine and muscarinic antagonists act in
synergy to inhibit haloperidol-induced
catalepsy. Neuropharmacology 45:493
503
Hauser RA, Hubble JP, Truong DD. 2003.
Randomized trial of the adenosine A(2A)
receptor antagonist istradefylline in advanced PD. Neurology 61:297303
Bara-Jimenez W, Sherzai A, Dimitrova T,
Favit A, Bibbiani F, et al. 2003. Adenosine
A(2A) receptor antagonist treatment of
Parkinsons disease. Neurology 61:293
96
Fredduzzi S, Moratalla R, Monopoli A,
Cuellar B, Xu K, et al. 2002. Persistent behavioral sensitization to chronic L-DOPA
requires A2A adenosine receptors. J. Neurosci. 22:105462
Bibbiani F, Oh JD, Petzer JP, Castagnoli N, Chen JF, et al. 2003. A2A
antagonist prevents dopamine agonistinduced motor complications in animal
models of Parkinsons disease. Exp. Neurol. 184:28594
Blum D, Galas MC, Pintor A, Brouillet E,
Ledent C, et al. 2003. A dual role of adenosine A2A receptors in 3-nitropropionic
acid-induced striatal lesions: implications for the neuroprotective potential of
A2A antagonists. J. Neurosci. 23:5361
69
Fink JS, Kalda A, Ryu H, Stack EC,
Schwarzschild MA, et al. 2004. Genetic
and pharmacological inactivation of the
adenosine A2A receptor attenuates 3nitropropionic acid-induced striatal damage. J. Neurochem. 88:53844
4 Dec 2004
13:21
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
132.
133.
134.
135.
136.
137.
138.
139.
140.
141.
142.
143.
144.
145.
146.
147.
148.
149.
411
4 Dec 2004
13:21
412
AR
AR232-PA45-16.tex
AR232-PA45-16.sgm
LaTeX2e(2002/01/18)
P1: GCE
FREDHOLM ET AL.
4 Dec 2004
13:20
AR
AR232-PA45-17.tex
AR232-PA45-17.sgm
P1: GCE
LaTeX2e(2002/01/18)
10.1146/annurev.pharmtox.45.120403.100045
413
4 Dec 2004
13:20
414
AR
AR232-PA45-17.tex
KROETZ
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
XU
Figure 1 Major pathways of arachidonic acid metabolism catalyzed by cytochrome P450. Arachidonic acid is metabolized into epoxyeicosatrienoic acids
(EETs) by CYP2C and CYP2J epoxygenases and into 20-hydroxyeicosatetraenoic acid
(20-HETE) by CYP4A and CYP4F -hydroxylases. EETs can also be further metabolized by soluble epoxide hydrolase (sEH) into their corresponding dihydroxyeicosatrienoic acids (DHETs).
limited, although recent efforts in this area hold the promise that new drug targets
will also emerge from this pathway.
CYP enzymes can metabolize arachidonic acid into numerous eicosanoids with
the relative abundance dependent on the tissue and species (Figure 1). The major
products in most tissues are the -hydroxylated metabolite 20-hydroxyeicosatetraenoic acid (20-HETE) and regio- and stereospecific epoxyeicosatrienoic acids
(EETs). Hydroxylation of arachidonic acid at the -1 and midchain (carbons 16
18) positions is less common. CYP4A and CYP4F enzymes catalyze the - and
-1 hydroxylation reactions, whereas members of the CYP2C and CYP2J families are responsible for epoxidation (35). A novel CYP isoform, CYP2U1, has
recently been identified as a human arachidonic acid -hydroxylase (6). EETs are
efficiently hydrated by soluble epoxide hydrolase (sEH) into the corresponding dihydroxyeicosatrienoic acids (DHETs) (7, 8). CYP eicosanoids can also be further
metabolized by CYPs, dehydrogenases, or cyclooxygenases (4, 911); -oxidized
(4, 10); or incorporated into membrane phospholipid pools (10, 11). The focus of
this review is on pathologic regulation and inhibition of the CYP -hydroxylase
pathway.
4 Dec 2004
13:20
AR
AR232-PA45-17.tex
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
415
4 Dec 2004
13:20
416
AR
AR232-PA45-17.tex
KROETZ
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
XU
4 Dec 2004
13:20
AR
AR232-PA45-17.tex
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
417
4 Dec 2004
13:20
418
AR
AR232-PA45-17.tex
KROETZ
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
XU
4 Dec 2004
13:20
AR
AR232-PA45-17.tex
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
419
(90, 91). Peroxisome proliferators typically exert their effects through activation of
the isoform of the peroxisome proliferator-activated receptor (PPAR). Unfortunately, the use of these inducers to characterize 20-HETE function is complicated
by the fact that the CYP4A isoforms, CYP2C23, and soluble epoxide hydrolase
are induced (35, 37, 92), whereas CYP4F and CYP2C11 are downregulated (67,
93) following exposure to peroxisome proliferators. With effects on both the CYP
-hydroxylase and epoxygenase pathways, often in opposite directions, the interpretation of results using these chemicals should be made with caution. The
abundant cellular signaling molecule nitric oxide inhibits CYP4A -hydroxylase
expression and function (94) and a novel mechanism of regulation involving covalent attachment of the prosthetic heme group through an ester link at a glutamic
acid residue conserved in the I-helix of the active site of most CYP4 members
has recently been described (9598). Interestingly, in some cases covalent modification of the enzymes results in increased activity. Heme levels are highest in
the liver and vary throughout the body, suggesting that covalent heme binding
may influence tissue-specific regulation of CYP4 arachidonic acid -hydroxylase
activity, a novel mechanism of CYP regulation. Although each of these mechanisms is interesting and of value in the field of eicosanoid biology, the focus of the
sections below is on the regulation of arachidonic acid -hydroxylation in various
disease states and the use of chemical inhibitors to study 20-HETE biology.
HYPERTENSION
4 Dec 2004
13:20
420
AR
AR232-PA45-17.tex
KROETZ
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
XU
4 Dec 2004
13:20
AR
AR232-PA45-17.tex
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
421
PREGNANCY
INFLAMMATION
4 Dec 2004
13:20
422
AR
AR232-PA45-17.tex
KROETZ
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
XU
stimuli, as barium sulfate increased hepatic CYP4F4 mRNA and protein levels as
well as corresponding enzyme activity. By contrast, hepatic CYP4F1 and CYP4F6
were unaffected by these inflammatory stimuli. In the kidney, LPS had no effect
on CYP4F mRNA levels, but barium sulfate induced CYP4F1 and CYP4F6 up to
threefold (121). PPAR plays some role in mediating the inductive effects of LPS
on Cyp4f15 in the kidney and the downregulation of Cyp4f15 and Cyp4f16 in the
liver (67). Interestingly, a traumatic brain injury model is associated with isoformdependent changes in both hepatic and renal CYP activity, including a twofold
increase in renal CYP4F expression and activity that is sustained for at least two
weeks (122). It is tempting to speculate that changes in renal 20-HETE levels resulting from CYP4F activity might contribute to the renal effects associated with
head trauma.
Alterations in CYP4A expression and function have been noted in
animal models of diabetes. Following the induction of diabetes with streptozotocin,
CYP4A expression and function are increased in liver and kidney microsomes
(113, 123126). The effect of diabetes can be reversed with insulin treatment
(113, 123, 125, 126) or correction of the hyperketonic state (127). An elevation of
intracellular fatty acids during diabetes contributes to the effects on CYP4A (128).
Activation of PPAR is a necessary step in mediating the effects of diabetes on
CYP4A transcription, as streptozotocin treatment has no effect in PPAR/ mice
(126). It is postulated that levels of an endogenous fatty acid activator of PPAR are
increased in the diabetic state, resulting in PPAR activation and CYP4A induction.
The effects of streptozotocin-induced diabetes on CYP4A expression appear to be
a direct result of the disease state, as similar findings are reported for the fa/fa
Zucker rat and the ob/ob mice (129). The recent report of a negative correlation
between insulin levels and urinary 20-HETE excretion in humans (112) suggests
that renal CYP4A expression and function is likely altered in human diabetics.
Lower renal 20-HETE production in diabetics would provide a protective effect
from the vasoconstrictive properties of this eicosanoid but may alter the pressurenatriuresis profile and contribute to the hypertensive complications of diabetes.
DIABETES
MECHANISM-BASED INHIBITORS
4 Dec 2004
13:20
AR
AR232-PA45-17.tex
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
423
4 Dec 2004
13:20
424
AR
AR232-PA45-17.tex
KROETZ
TABLE 1
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
XU
Inhibitor
Structure
ABT
10-UDYA
11-DDYA
17-ODYA
DMDYA
10-SUYS
DBDD
DDMS
HET0016
6(Z),15(Z)-20-HEDE
4 Dec 2004
13:20
AR
AR232-PA45-17.tex
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
425
(17-ODYA)] (Table 1) carbons (138, 139) have also been synthesized and characterized as arachidonic acid -hydroxylase inhibitors. These compounds are
oxidized to ketenes that inactivate the CYP protein instead of alkylating the prosthetic heme group (140). They are highly selective inhibitors of rat liver CYP
isoforms that are active toward fatty acid substrates without affecting total P-450
levels or other P-450-dependent activities. 10-UDYA and 11-dodecynoic acid (11DDYA) (Table 1) specifically inhibit hepatic CYP enzymes that catalyze lauric acid
- and -1 hydroxylation (138). 11-DDYA and 17-ODYA inhibit the lauric acid
- and -1 hydroxylation by microsomes prepared from the lungs of pregnant
rabbits and reconstituted P-450 (141). 17-ODYA is a very potent inhibitor toward
arachidonic acid metabolism; however, the inhibition is nonspecific (142, 143).
It irreversibly inhibits both -hydroxylation and epoxidation of arachidonic acid
with IC50 values of 7 and 5 M, respectively (142). It also potently inhibits and -1 hydroxylation of arachidonic acid catalyzed by recombinant rat CYP4A1,
CYP4A2, CYP4A3, CYP4F1, and CYP4F4 with a similar IC50 for all isoforms
(51, 74).
Despite its lack of selectivity, 17-ODYA has been widely used in in vitro and
in situ studies to characterize the biological function of 20-HETE. For example,
17-ODYA has been used to establish the role of 20-HETE in the regulation of
renal blood flow and tubuloglomerular feedback and as a K+ channel inhibitor
in rat renal arterioles (20, 21, 143). It also has been used to demonstrate that
20-HETE mediates the vasoconstrictor response to angiotensin II in isolated renal
arterioles and the myogenic response of renal, cerebral, and skeletal muscle arteries
(19, 137, 144). Intrathecal administration of 17-ODYA prevents the acute fall in
cerebral blood flow after subarachnoid hemorrhage in the rat (145).
Unfortunately, the terminal acetylenic fatty acid inhibitors are of little value
for in vivo inactivation of fatty acid hydroxylases because of their rapid metabolic
degradation by -oxidation, their esterification and storage in the liver, and extensive protein binding (139). Introduction of two methyl groups vicinal to the
carboxylic acid group in 10-UDYA yields 2,2-dimethyl-11-dodecynoic acid
(DMDYA) (Table 1), and the replacement of the carboxyl group in 10-UDYA
with a sulfate yields sodium 10-undecynyl sulfate (10-SUYS) (Table 1). Both of
these compounds have in vitro activities similar to that of 10-UDYA and are resistant to -oxidation and storage and exhibit substantial in vivo activity (146).
10-SUYS selectively inhibits arachidonic acid -hydroxylation in rat cortical microsomes (74). The IC50 of 10-SUYS for inhibition of 20-HETE formation is 10
M, whereas epoxygenase activity was not affected at a concentration up to 50
M. 10-SUYS also shows isoform-specific inhibition of rat recombinant CYP4F1and CYP4F4-catalyzed 20-HETE formation (74). The IC50 of 10-SUYS for inhibition of CYP4F4-catalyzed 20-HETE formation is 25 M, whereas 10-SUYS
has only minimal inhibition toward CYP4F1-catalyzed 20-HETE formation.
Administration of 150 mg/kg of 10-SUYS intraperitoneally to SHRs results
in a dose-dependent and selective inhibition of renal cortical arachidonic acid hydroxylase activity (147). A single dose of 10-SUYS (5 mg/kg) causes an acute
4 Dec 2004
13:20
426
AR
AR232-PA45-17.tex
KROETZ
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
XU
COMPETITIVE INHIBITORS
4 Dec 2004
13:20
AR
AR232-PA45-17.tex
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
427
ANTISENSE OLIGONUCLEOTIDES
4 Dec 2004
13:20
428
AR
AR232-PA45-17.tex
KROETZ
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
XU
20-HETE ANTAGONISTS
4 Dec 2004
13:20
AR
AR232-PA45-17.tex
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
429
20-HETE effects. To further delineate the role of individual -hydroxylases, additional isoform-specific inhibitors will need to be characterized. The plausibility of
such an approach is supported by our recent studies that indicate isoform-specific
inhibition of the CYP4F enzymes by 10-SUYS and DDMS (74). Another challenge will be the continued improvement of the biopharmaceutical properties of
these inhibitors. Stability and appropriate pharmacokinetic properties are essential
for the application of inhibitors in vivo.
Future studies will begin to focus more on the application of our current knowledge of 20-HETE effects and CYP arachidonic acid -hydroxylase regulation in
human biology and disease. Sensitive LC/MS/MS and GC/MS assays are becoming more widely available for the quantitation of 20-HETE in urine, plasma, and
other biological samples. This technology will no doubt prove useful in exploring
the hypothesis that arachidonic acid -hydroxylation is altered in human disease,
e.g., diabetes, as well as hypertension, pregnancy, and inflammation. The recent
reports of 20-HETE urinary excretion patterns correlating with insulin levels and
natriuresis (110112) provide promise that therapeutic modulation of CYP arachidonic acid -hydroxylase may prove useful in the management of human disease.
Another exciting area that should be explored is genetic variability in 20-HETE
synthesis and the importance of this variation in eicosanoid function and disease
susceptibility. The clinical significance of drug-induced alterations in CYP arachidonic acid -hydroxylase activity should also be studied. Although much remains
unknown about the role of 20-HETE in human disease, the wealth of information
in animal models will no doubt stimulate increasing interest in this field of study,
with the hope that new drug targets might be identified for the management of
vascular reactivity, renal function, and likely additional clinical conditions that
remain to be discovered.
ACKNOWLEDGMENTS
Work from the authors laboratory cited in this article was supported by a grant
from the National Institutes of Health (HL53994) and the UCSF Liver Core Center
Facility supported by National Institutes of Health Grant P30 DK26743.
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Mukherjee AB, Miele L, Pattabiraman N.
1994. Phospholipase A2 enzymes: regulation and physiological role. Biochem.
Pharmacol. 48:110
2. Funk CD. 2001. Prostaglandins and
leukotrienes: advances in eicosanoid biology. Science 294:187175
4 Dec 2004
13:20
430
5.
6.
7.
8.
9.
10.
11.
12.
13.
AR
AR232-PA45-17.tex
KROETZ
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
XU
14.
15.
16.
17.
18.
19.
20.
21.
4 Dec 2004
13:20
AR
AR232-PA45-17.tex
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
431
31. Amlal H, Legoff C, Vernimmen C, Paillard M, Bichara M. 1996. Na+-K+(NH4+)2Cl cotransport in medullary thick ascending limb: control by PKA, PKC, and
20-HETE. Am. J. Physiol. Cell Physiol.
271:C45563
32. Birks EK, Bousamra M, Presberg K,
Marsh JA, Effros RM, et al. 1997. Human
pulmonary arteries dilate to 20-HETE,
an endogenous eicosanoid of lung tissue.
Am. J. Physiol. Lung Cell Mol. Physiol.
272:L82389
33. Zhu D, Birks EK, Dawson CA, Patel
M, Falck JR, et al. 2000. Hypoxic pulmonary vasoconstriction is modified by
P-450 metabolites. Am. J. Physiol. Heart
Circ. Physiol. 279:H152633
34. Jacobs ER, Zeldin DC. 2001. The lung
HETEs (and EETs) up. Am. J. Physiol.
Heart Circ. Physiol. 280:H110
35. Hardwick JP, Song BJ, Huberman E, Gonzalez FJ. 1987. Isolation, complementary DNA sequence, and regulation of
rat hepatic lauric acid -hydroxylase (cytochrome P-450LA). Identification of a
new cytochrome P-450 gene family. J.
Biol. Chem. 262:80110
36. Kimura S, Hanioka N, Matsunaga E,
Gonzalez FJ. 1989. The rat clofibrateinducible CYP4A gene subfamily. I.
Complete intron and exon sequence of
the CYP4A1 and CYP4A2 genes, unique
exon organization, and identification of a
conserved 19-bp upstream element. DNA
8:50316
37. Kimura S, Hardwick JP, Kozak CA,
Gonzalez FJ. 1989. The rat clofibrateinducible CYP4A subfamily. II. cDNA sequence of IVA3, mapping of the Cyp4a
locus to mouse chromosome 4, and coordinate and tissue-specific regulation of the
CYP4A genes. DNA 8:51725
38. Stromstedt M, Hayashi S, Zaphiropoulos
PG, Gustafsson JA. 1990. Cloning and
characterization of a novel member of the
cytochrome P450 subfamily IVA in rat
prostate. DNA Cell Biol. 9:56977
39. Bellamine A, Wang Y, Waterman MR,
4 Dec 2004
13:20
432
40.
41.
42.
43.
44.
45.
46.
47.
AR
AR232-PA45-17.tex
KROETZ
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
XU
48.
49.
50.
51.
52.
53.
54.
55.
4 Dec 2004
13:20
AR
AR232-PA45-17.tex
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
65.
66.
67.
68.
69.
70.
71.
72.
433
A novel human cytochrome P450 4F isoform (CYP4F11): cDNA cloning, expression, and genomic structural characterization. Genomics 68:16166
Bylund J, Bylund M, Oliw EH. 2001.
cDNA cloning and expression of
CYP4F12, a novel human cytochrome
P450. Biochem. Biophys. Res. Commun.
280:89297
Hashizume T, Imaoka S, Hiroi T, Terauchi Y, Fujii T, et al. 2001. cDNA cloning
and expression of a novel cytochrome
P450 (CYP4F12) from human small intestine. Biochem. Biophys. Res. Commun.
280:113541
Cui X, Kawashima H, Barclay TB, Peters
JM, Gonzalez FJ, et al. 2001. Molecular
cloning and regulation of expression of
two novel mouse CYP4F genes: expression in peroxisome proliferator-activated
receptor -deficient mice upon lipopolysaccharide and clofibrate challenges. J.
Pharmacol. Exp. Ther. 296:54250
Kikuta Y, Kusunose E, Ito M, Kusunose
M. 1999. Purification and characterization
of recombinant rat hepatic CYP4F1. Arch.
Biochem. Biophys. 369:19396
Kikuta Y, Kusunose E, Kusunose M.
2000. Characterization of human liver
leukotriene B4 -hydroxylase P450
(CYP4F2). J. Biochem. (Tokyo) 127:
104752
Kikuta Y, Kusunose E, Sumimoto H,
Mizukami Y, Takeshige K, et al. 1998. Purification and characterization of recombinant human neutrophil leukotriene B4
-hydroxylase (cytochrome P450 4F3).
Arch. Biochem. Biophys. 355:2015
Bylund J, Harder AG, Maier KG, Roman RJ, Harder DR. 2003. Leukotriene B4
-side chain hydroxylation by CYP4F5
and CYP4F6. Arch. Biochem. Biophys.
412:3441
Jin R, Koop DR, Raucy JL, Lasker JM.
1998. Role of human CYP4F2 in hepatic
catabolism of the proinflammatory agent
leukotriene B4. Arch. Biochem. Biophys.
359:8998
4 Dec 2004
13:20
434
AR
AR232-PA45-17.tex
KROETZ
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
XU
83.
84.
85.
86.
87.
88.
89.
90.
91.
4 Dec 2004
13:20
AR
AR232-PA45-17.tex
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
435
4 Dec 2004
13:20
436
111.
112.
113.
114.
115.
116.
117.
118.
AR
AR232-PA45-17.tex
KROETZ
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
XU
4 Dec 2004
13:20
AR
AR232-PA45-17.tex
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
128.
129.
130.
131.
132.
133.
134.
135.
136.
137.
138.
139.
140.
141.
142.
143.
144.
437
4 Dec 2004
13:20
438
145.
146.
147.
148.
149.
150.
151.
152.
AR
AR232-PA45-17.tex
KROETZ
AR232-PA45-17.sgm
LaTeX2e(2002/01/18)
P1: GCE
XU
7 Jan 2005
14:4
AR
AR232-PA45-18.tex
AR232-PA45-18.sgm
P1: GCE
LaTeX2e(2002/01/18)
10.1146/annurev.pharmtox.45.120403.100127
INTRODUCTION
Ubiquitination (or ubiquitylation as it is now often called) is a process in which cellular proteins are covalently modified, posttranslationally, with a single molecule
(monoubiquitination) or chains (polyubiquitination) of ubiquitin (Ub) (1 and references therein). Ub is an evolutionarily highly conserved 76-residue polypeptide
(8565 Da) that, as implied by its name, is ubiquitously present in all eukaryotic cells either as free species (monomers or in preformed chains) or covalently
0362-1642/05/0210-0439$14.00
439
7 Jan 2005
14:4
440
AR
AR232-PA45-18.tex
CORREIA
SADEGHI
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
MUNDO-PAREDES
bound to proteins. Such covalent protein decoration with Ub serves many important physiological functions. Monoubiquitination can serve as a sorting signal
in endocytic vesicular transport as well as a critical regulator of transcription,
replication, and DNA repair (2 and references therein). In contrast, polyubiquitination largely targets proteins for degradation via the 26S proteasome, thereby
critically regulating various essential cellular processes such as cell cycle progression, antigen presentation, apoptosis and stress response, in addition to the vital
function of quality control by cellular disposal of aberrant, misfolded, damaged,
and/or abnormal proteins (1, 310). Indeed, in the latter context, high molecular mass (HMM) ubiquitinated species of certain (but not all) cytochromes P450
(P450s) have been detected in liver cells after structural damage and/or blockade
of their normal physiological turnover by proteasomal inhibitors (11, 12). The
HMM profile of this P450 ubiquitination and the striking temporal relationship
between its detection and P450 proteasomal degradation are consistent with a role
for polyubiquitination as a targeting signal in this process. Although a biological
role for monoubiquitination in P450 regulation may exist, it remains obscure. Furthermore, although monoubiquitination of the various multiple P450 surface Lys
residues could yield a HMM profile similar to that generated by polyubiquitination, its plausibility has been excluded by our in vitro studies with methylated Ub
(MeUb) (13), the Ub analog incapable of polyubiquitination because of chemical
methylation of its Lys residues. For these combined reasons, this review will focus
on what is currently known about P450 polyubiquitination and its association with
the proteasomal destruction of these enzymes.
7 Jan 2005
14:4
AR
AR232-PA45-18.tex
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
441
ligase (E3). Such sequential shuttling of the Ub-thioester from E1 to the target
protein entails an initial transthiolation of an E2-Cys residue, with or without a
subsequent similar transthiolation of an E3-Cys residue. This Ub relay via E2 and
E3 in this process apparently insures substrate specificity by preventing the attack
of random proteins by the E1 charged molecule.
E2s differ in size, usually with low molecular weights ranging between 14
and 36 kDa1, and containing a 1416 kDa core that is 35% conserved among
family members (69). The remainder of the protein may contain N- and/or Cterminal extensions that confer substrate and/or E3 specificity or promote physical
interactions between the three entities, often by serving as membrane anchors. Both
ER-bound and soluble E2s exist for the ubiquitination of luminal and integral
endoplasmic reticulum (ER) proteins such as the P450s (9, 15, 16; see below).
This E2 multiplicity apparently insures functional redundancy on one hand, and
substrate specificity on the other (9).
The E3 Ub-ligases, considerably more numerous than E2s, often exist as monomeric proteins or heteromeric multisubunit protein complexes. The multiplicity
and structural diversity of E3s contribute to their remarkable substrate diversity
and/or specificity in the recognition of proteolytic targets (9, 1721). Three general
classes of E3s are known: the HECT-E3s (Homologous to E6-AP C-terminus, the
first HECT-E3 identified), CHIP-E3s (C-terminus of Hsc70 interacting protein),
and the RING-finger E3s. The same E2 can apparently interact with either the
HECT or the RING-finger domain of an E3 (9). HECT-E3s contain a conserved
Cys-SH in their C-terminal domain for Ub-thioester relay from its cognate E2 to
the target protein (9). The N terminus of some but not all HECT-E3s contains a
WW domain (with 2 Trps, 2022 residues apart and an invariant Pro within a 40residue region) that interacts with Pro-rich sequences including those containing
phosphorylated Ser/Thr residues (9, 22).
The first CHIP-E3 prototype was identified as a cochaperone of Hsc70. A typical
CHIP-E3 contains a tetratricopeptide repeat (TPR) motif that interacts with both
Hsc70 and Hsp90; its U-box domain exhibits E3 Ub-ligase activity. CHIP-E3s
play an active role in quality control through recognition of chaperone-associated
aberrant proteins that are ubiquitinated before their removal by the proteasome
(17, 18). These E3s may play a similar role in the Ub-dependent proteasomal
degradation of unfolded and/or misfolded P450 proteins.
The increasingly numerous RING-finger E3s, on the other hand, exhibit an
interleaved or cross-braced ring pattern with eight conserved metal-binding Cys
and His residues that coordinate two Zn atoms (9, 1921). The three RING-finger
motifs (RING-CH, RING-HC, and RING-H2) are distinguished by whether one
or two His are the middle two conserved residues. These E3s may exist as single
subunits with both substrate recognition and RING-finger E2 docking domains
on the same polypeptide or as multisubunit protein scaffolds that include a small
1
Although Ubcs are known to possess low molecular weights, a larger (582 kDa) polytopic
membrane-anchored Ubc (BRUCE) has been reported (9).
7 Jan 2005
14:4
AR
442
AR232-PA45-18.tex
CORREIA
SADEGHI
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
MUNDO-PAREDES
Frequently, this is detected as a smear rather than a distinct step-ladder. However, in the
case of P450s it is important to distinguish between the essential features of ubiquitination
and protein aggregation.
7 Jan 2005
14:4
AR
AR232-PA45-18.tex
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
443
sites with a total of six proteolytic sites/20S species (27, 3033). Each of these
three (1, 2, and 5) sites differs in its preferential cleavage after acidic, basic,
or hydrophobic residues, respectively, giving rise to the typical chymotrypsinlike, trypsin-like, peptidoglutamyl hydrolase, and caseinolytic activities of the
proteasome (34). These catalytic -subunits are unusual in exhibiting a unique
catalytically active N-terminal Thr residue, a critical nucleophile in peptide bond
hydrolysis, strategically lining the barrel cavity. This unusual characteristic qualifies the 20S proteasome as an amino-terminal nucleophile (NTN) hydrolase. The
active site Thr residue is the target of selectively designed proteasomal inhibitors
such as MG-132, lactacystin -lactone, epoxomycin, and others (3537).
In the eukaryotic 26S complex, either (or both) of the 20S -ring structures
is capped by a 900 kDa multisubunit complex variably termed PA700 (proteasome activator) or the 19S regulatory complex, an assembly that is ATP-dependent
(27, 3841). Thus the 26S proteasome may exist as heterooligomeric 20S barrels
capped at one or both ends with this 19S complex. The 19S cap complex is composed of 17 or more subunits arranged in two structurally and functionally distinct
assemblies: The base situated directly above the 20S -ring contains six functionally distinct AAA-family ATPase or Rpt (regulatory particle ATPase) subunits and
two non-ATPase Rpn (regulatory particle non-ATPase) subunits. The 19S base is
topped by a lid containing eight non-ATPase Rpn subunits. In the mammalian 26S
proteasome, these two structures are further linked together by the seventeenth subunit, Rpn10. High salt can dissociate the lid from the base-bound 20S proteasome,
leaving a catalytic species that is capable of ATP-dependent degradation of an
unfolded protein but incapable of degrading ubiquitinated substrates. This finding
implies that the lid contains polyUb-recognition elements and unfoldases in addition to deubiquitinating enzymes. More recently, additional proteasome-associated
proteins (including E3s) have been identified with the 19S complex, although it
is unclear whether they are adventitiously bound or bona fide lid components.
Together the multiple subunits of the 19S regulatory complex are responsible for
the initial acceptance of the polyUb-tagged target substrates as well as the coordination of the subsequent release of this polyUb-tag with their high energy-coupled
unfolding and translocation through the 20S proteolytic core to be digested (27
34, 3841). Accordingly, the polyUb tag is initially recognized by the Rpt5 and/or
Rpn10 (or any other) of the 19S base subunits, thereby tethering the substrate to
the 19S subcomplex and bringing its termini or any other loosely folded domain in
close proximity to one or more of the 19S base Rpt ATPase subunits for unfolding
and translocation of the unfolded substrate through the 19S base pore. This event,
requiring ATP hydrolysis, denatures the substrate to enable its onward movement
through the juxtaposed 20S -subunit pore into the adjacent 20S catalytic chamber where it is processively digested into short peptides that exit through the distal
20S axial pore. As the entire polyubiquitinated substrate is unfolded and strung
through the 20S catalytic chamber to be proteolyzed, its polyUb tag is released
intact by hydrolyses of the UbGly76-NH2 substrate isopeptide bond by the 19S lid
subunits Rpn11, a deubiquitinase with a Zn-metalloprotease-like domain (27, 42,
7 Jan 2005
14:4
444
AR
AR232-PA45-18.tex
CORREIA
SADEGHI
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
MUNDO-PAREDES
43), and Ubp6, another 19S deubiquitinase. Additional 19S subunits containing
Ub-hydrolases, the cysteine proteases that disassemble the Lys48-linked polyUb
chain, may be subsequently recruited to regenerate Ub for fresh proteolytic cycles
(2730).
The enmeshed 20S -subunit N termini normally clog the two gates to the
adjacent 20S proteolytic -subunit chamber, keeping its pores closed and inaccessible to peptides and unfolded proteins and therefore catalytically inactive. Thus
an additional critical function of the 19S base ATPase subunits is to activate the
20S proteolytic chamber by physically lifting these -subunit tails to open its gate
pore diameter that is accessible to unfolded proteins and polypepto a 20 A
tides but not to normally folded proteins (27, 44, 45). Such restrictive gating thus
protects native cellular proteins from promiscuous/indiscriminate proteasomal attack. Higher eukaryotes also contain other hybrid proteasome species, including
the immunoproteasomes, that are specifically engaged in antigen processing for
presentation to the immune system on major histocompatibility complex (MHC)
class I molecules (35, 4649). Because antigenic peptides from several P450 enzymes have been reported (5053), these proteasome species are relevant to the
current discussion. Unlike the above described constitutive 20S species of the 26S
proteasome, that of the immunoproteasome contains interferon- inducible 1, 2
and 5 proteolytic subunits responsible for generating antigenic peptides (27, 49).
This 20S immunoproteasome species may be capped on either or both ends by the
11S activator complex (also known as the PA28 activator, which consists of two
alternating non-ATPase subunits, PA28 and PA28, in a concentric heptameric
complex of 200 kDa) (27, 49), which modulates the proteasome-catalyzed generation of antigenic peptides. The PA28 activator apparently activates the 20S
proteolytic species by regulating the gating into the 20S chamber in a mechanism
analogous to that of the 19S complex (27, 49).
7 Jan 2005
14:4
AR
AR232-PA45-18.tex
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
445
proteasomal degradation (Figure 2). Any specific substrate sorting for proteolysis
thus has to occur before the protein is polyubiquitinated.
Indeed, such specific substrate sorting does occur and E3s apparently play a
critical role in the selection of a substrate for Ub-dependent 26S proteasomal
degradation through the recognition of specific substrate-based structural determinants that partly or fully constitute destruction signals called degrons (9, 27). Such
critical structural determinants of substrate recognition by E3s may be intrinsic to
the proteins primary sequence or acquired through posttranslational processing.
Degrons encoded in a short discrete sequence include the Deg1 sequence of the
yeast MAT2 transcriptional regulator, the destruction box of mitotic cyclins, the
degradation motif of IB proteins, the stability regulating region of cMOS, specific
phosphorylatable Ser/Thr residues, PEST sequences, Pro-rich domains, and some
N-terminal residues of target proteins (9, 27; reviewed in Reference 57). On the
other hand, instead of a discrete modular degron, the structural information for
Ub-dependent 26S proteasomal degradation may be distributed over a considerably large protein domain, as in the case of the entire N-terminal 523-residue-long
transmembrane domain of HMGR (58, 59) as well as the IB N-terminal phosphorylatable (Ser32/Ser36) and ubiquitinatable (Lys21/Lys22) and C-terminal PEST
domains (6062). Although certain P450s are indeed phosphorylated (13, 6369)
and/or ubiquitinated (13, 69) before their proteolytic degradation, it remains to
be determined whether they similarly harbor any intrinsic modular or distributed
degrons that are either normally accessible or unmasked for this event.
7 Jan 2005
14:4
446
AR
AR232-PA45-18.tex
CORREIA
SADEGHI
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
MUNDO-PAREDES
yielded no comparable proteolytic loss of the enzyme nor a similar HMM ubiquitination profile (70). ABT inactivates most P450s via heme-N-arylation rather
than heme-modification of the protein (72). However, although ABT rapidly and
effectively abolished CYP2E1 function, minimal CYP2E1 proteolysis was observed along with relatively minor ubiquitination of the microsomal protein detectable only after 9 h of ABT treatment (70). Thus, despite marked P450 functional loss, the protein moiety apparently escapes unscathed and remains relatively
stable.
The second report documented the time-dependent ubiquitination and proteolytic loss of rat liver microsomal CYPs 3A after their mechanism-based inactivation by 3,5-dicarbethoxy-2,6-dimethyl-4-ethyl-1,4-dihydropyridine (DDEP), a
4-ethyl analog of the calcium channel antagonist nitrendipine (11, 73, 74). DDEPinduced CYP3A inactivation also results in heme-modification of the protein that
is accompanied within 30 min by ubiquitination of the microsomal protein and
immunochemically detectable loss of these microsomal enzymes (11, 73, 74).
This ubiquitination is, as expected, enhanced by the coadministration of hemin,
a known 26S proteasome inhibitor (7577; see below). More importantly, when
the microsomal CYPs 3A were first immunoprecipitated with goat anti-CYP3A23
IgGs and then immunoblotted with rabbit anti-Ub IgGs, they exhibited the characteristic step-ladder ubiquitination profile discussed earlier (11; Figure 3a). This
finding unequivocally established that the DDEP-inactivated CYPs 3A were indeed ubiquitinated. This DDEP-induced ubiquitination of liver microsomal CYPs
3A in intact rats could be reproduced in DDEP-incubations of freshly isolated hepatocytes obtained from dexamethasone (DEX)-pretreated rats (12). This system
provided a convenient experimental model for the definitive mechanistic characterization of the proteolytic process. Accordingly, CYP3A immunoprecipitation
analyses revealed that incubation of these freshly isolated hepatocytes with DDEP
resulted in the ubiquitination of CYPs 3A within 15 min of their inactivation, an
event that preceded the onset of their proteolytic degradation detectable at 30 min
(12). Inclusion of the proteasomal inhibitors aclarubicin or MG-132 in these incubations, while blocking the DDEP-induced immunochemically detectable loss
of microsomal CYPs 3A, also intensified the CYP3A ubiquitination profile (12).
These findings in freshly isolated rat hepatocytes thus unequivocally established
that DDEP-inactivated, heme-modified CYPs 3A undergo Ub-dependent 26S proteasomal degradation (12).
Both ubiquitination and 26S proteasomal degradation of heme-modified CYPs
3A can also be documented in in vitro reconstituted systems (13) containing purified recombinant CYP3A4, the major human liver CYP3A ortholog. For this
purpose, 35S-labeled CYP3A4 either native or heme-modified by inactivation with
cumene hydroperoxide (CuOOH) and Fraction II (a rat liver cytosolic subfraction
containing the requisite soluble E1, E2 and E3 enzymes and the 26S proteasome)
were incubated in the presence of Ub, an ATP-generating system, protease inhibitors (to block the ubiquitous lysosomal protease contaminants), MgCl2, and
Ub-aldehyde (Ubal). The inclusion of Ubal, an inhibitor of Ub-hydrolases and
7 Jan 2005
14:4
AR
AR232-PA45-18.tex
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
447
isopeptidases, in these incubations is essential for blocking protein deubiquitination so that Ub-conjugation of a substrate can be detected. In the absence of other
proteins, thermally sensitive CYP3A4 tends to aggregate on incubation at 37 C,
thereby generating undesirable and confounding cross-linking artifacts (see below). To preclude such incubation-induced aggregation artifacts, liver microsomal
membranes from female rats that are devoid of appreciable CYP3A content and
exhaustively washed free of luminally trapped cytosolic ubiquitinating and proteolytic enzyme contaminants were included as a membrane platform for CYP3A4.3
Parallel immunoblotting analyses of these incubates with either anti-CYP3A or
anti-Ub IgGs revealed that only the heme-modified but not the native CYP3A4
exhibited a time-dependent ubiquitination profile which was enhanced by the inclusion of the proteasome inhibitor, Z-IE(OtBu)ALCHO (PSI) (78). No CYP3A4
ubiquitination profile was detected if Fraction II was omitted from the incubations (13). Moreover, no similar profile was observed when Ub was replaced by
MeUb (the methylated Ub-analog incapable of any Lys48-Gly76 polyubiquitination
linkages; see above) in the incubation (13). If appreciable CYP3A4 aggregation
were to have occurred, it should have been detected in the incubations with MeUb.
The absolute dependence of the observed CYP3A4 ubiquitination profile on both
Fraction II and Ub convincingly attests to its authenticity, while excluding the recently raised possibility that this finding represents a CYP3A4 aggregation artifact
(79, 80).
Two native rat liver P450s, CYP2B1 and CYP2C11, that have relatively long
half-lives and reportedly are degraded by the lysosomal pathway in vivo, are also
subject to Ub-dependent 26S proteasomal degradation when suicidally inactivated
(13, 57, 81). Accordingly, in an in vitro reconstituted system similar to the one
described above, CuOOH-inactivated heme-modified CYP2B1 was shown to be
ubiquitinated and degraded by the 26S proteasome-species (13). On the other
hand, DDEP incubation of freshly isolated hepatocytes from untreated male rats
also caused a time-dependent structural and functional inactivation of CYP2C11
that is associated with CYP2C11 protein ubiquitination and proteolytic loss (Z.J. Song & M.A. Correia, unpublished observations; 81). Although such DDEPinduced CYP2C11 inactivation is largely due to its prosthetic heme destruction
to an N-ethylporphyrin, with little or no heme modification of the protein, no
structural or functional restoration of the enzyme was observed when hemin was
included in the incubations (81). Instead, as revealed by CYP2C11 immunoprecipitation analyses, inclusion of hemin in the DDEP-incubations resulted in an
accumulation of ubiquitinated CYP2C11 species, consistent with hemin blockade of proteasomal function at a step beyond protein ubiquitination (Figure 3e).
Unlike the recent findings in primary hepatocytes in culture (79), incubation of
3
In retrospect, this strategy for preventing CYP3A4 aggregation was fortuitous, given that
ERAD substrates such as CYP3A4 apparently require integral and ER-associated enzyme
components for their ubiquitination. The inclusion of ER membranes inadvertently provided
the ER ubiquitination components later found to be essential (16, 113).
7 Jan 2005
14:4
448
AR
AR232-PA45-18.tex
CORREIA
SADEGHI
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
MUNDO-PAREDES
7 Jan 2005
14:4
AR
AR232-PA45-18.tex
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
449
detected in vivo, even under conditions optimized for its maximal detection
(Figure 3a,c; see below). This ubiquitination is considerably augmented after
the protein is structurally unraveled by catalytic insults such as futile oxidative
cycling and/or chemically induced suicide inactivation. Such structural damage
may expose normally concealed Lys residues and/or other degron components
such as phosphorylatable residue PEST sequences and other targetable domains
to the ubiquitinating enzymes. Together, the above findings attest to the indisputable fact that when inactivated, misfolded, or otherwise structurally deformed,
certain hepatic P450s incur ubiquitination and/or 26S proteasomal degradation.
But are the native CYPs 3A also ubiquitinated in the course of their physiological
turnover?
7 Jan 2005
14:4
450
AR
AR232-PA45-18.tex
CORREIA
SADEGHI
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
MUNDO-PAREDES
Ubc7p/Cue1p but not Ubc6p (93, 97, 98). The HRD/DER machinery also includes
(b) Hrd2p, a 19S protein subunit that is a functionally indispensable component of
the 26S proteasome (98), and (c) Hrd1p/Hrd3p complex, the ER-associated Ubligase (E3) composed of Hrd1p/Der3p and its partner, Hrd3p. Hrd1p is an integral
ER-membrane protein with two distinct domains: a multitransmembrane-spanning
N-terminal hydrophobic region and a cytosolic C-terminal hydrophilic RING-H2
motif (24, 25, 99, 100) that binds Ubc1p or Ubc7p. Hrd3p is an ER glycoprotein with a single C-terminal membrane-anchor and a large N-terminal domain
in the ER-lumen. Hrd3p is found to stabilize Hrd1p in the ER membrane (24).
The Hrd1p/Hrd3p complex catalyzes the Ubc7p-dependent ubiquitination of target substrates such as Hmg2p and CPY (24, 99101). The HRD/DER machinery
also includes Cdc48p (p97, an AAA ATPase required for cellular processes such
as cell division, protein degradation, and ER membrane fusion), Npl4/Hrd4p, an
ER-specific adapter of undefined function, and Ufd1p, a Cdc48p protein adapter
for polyUb chain recognition and/or Cdc48p-association with Hrd4p (102104).
The Cdc48p-Ufd1p-Hrd4p complex is apparently involved in the recognition of
polyubiquitinated luminal and integral ER proteins, their dislocation from the ER,
and their subsequent delivery to the 26S proteasome. Mammalian homologs of
the yeast HRD/DER machinery such as Ubc6p, Ubc7p, Cue1p, Hrd2p, Hrd1p
Ufd1p, Hrd4p and Cdc48p have been recently documented (102112), indicating
that ERAD is evolutionarily a highly conserved process.
Because of this high homology between yeast and mammalian ubiquitination
enzymes and the availability of validated genetic S. cerevisiae strains with defined
defects in the ubiquitination and proteasomal degradation of several integral ER
proteins including Hmg2p, the yeast model was used to identify and characterize
the enzymes participating in the ubiquitination of CYP3A4, the dominant human
liver P450 (113, 114). To identify the Ubc involved in CYP3A4 ubiquitination,
isogenic wild-type (wt) yeast strains and strains deficient in Ubc6p, Ubc7p, or
in both Ubc6p and Ubc7p were transformed with the CYP3A4 expression vector
pAAH5/NF25 or the control vector (113). At the early stages of logarithmic cell
growth, CYP3A4 was equivalently expressed in all four strains, indicating their
comparable transcriptional and translational efficiencies (113). At the later stages
of culture, CYP3A4 was greatly stabilized only in mutants deficient in Ubc7p and
Ubc6p/Ubc7p but not in Ubc6p alone, thereby revealing the relative importance
of Ubc7p-dependent ubiquitination in the ERAD of this native integral protein.
Thus the monotopic CYP3A4 and the polytopic Hmg2p are alike in that they
require Ubc7p-dependent ubiquitination for their ERAD. Presumably, such Ubc7pmediated CYP3A4 ubiquitination also requires the ER-protein adapter Cue1p.
To determine whether the RING-H2 Hrd1p/Hrd3p Ub-ligase complex and
Hrd2p involved in Hmg2p Ub-dependent proteasomal degradation were similarly
involved in that of CYP3A4, isogenic wt and mutant hrd1, hrd2-1, and hrd3 S.
cerevisiae strains were transformed as discussed above with a CYP3A4 expression
plasmid and corresponding vector control. CYP3A4 protein was equivalently expressed in all four yeast strains during the early logarithmic growth phase, but at the
7 Jan 2005
14:4
AR
AR232-PA45-18.tex
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
451
later stationary growth stage of culture, CYP3A4 protein was comparably reduced
in wt and hrd1-deficient yeast, consistent with its degradation predominating over
its de novo synthesis after transient expression of both strains (113). In contrast,
consistent with the role of 26S proteasome in CYP3A4 degradation, the microsomal CYP3A4 protein content was significantly stabilized in hrd2-deficient yeast.
These findings thus reveal that in yeast, the RING-H2 finger Ub-ligase Hrd1p
is not required for CYP3A4 Ub-dependent proteasomal degradation in contrast
to that of Hmg2p (113). It is unclear whether the lesser, albeit statistically significant, CYP3A4 protein stabilization observed in hrd3-deficient yeast reflects
the interaction of Hrd3p with an Ub-ligase partner other than Hrd1p. The yeast
Ub-ligase required for CYP3A4 ubiquitination currently remains to be identified. Any of the several ERAD-associated E3 ligases such as the ER-localized
yeast RING-CH Doa-10 or HECT-like Rsp5p (115, 116) remain plausible E3
candidates in CYP3A4 ubiquitination. Similarly, the recently identified human
Hrd1p homolog HRD1 (110) or its related mammalian E3 homolog gp78/AMFR
(autocrine motility factor receptor; 111) could be involved in CYP3A4 ubiquitination in the human liver. Because all these E3 ligases duly engage Ubc7p or its human
counterpart UBC7 in their CYP3A4 ubiquitination reactions, their E3 candidacy is
plausible.
It is noteworthy, however, that the rapid disposal of human liver CYP3A4 via
the Ub-dependent 26S proteasomal degradation in S. cerevisiae is not because
it is an alien protein. Corresponding expression of other mammalian P450s (rat
liver CYPs 2B1 and 2C11) in these same yeast strains (114, 117) results in their
vacuolar lysosomal degradation rather than proteasomal degradation. The latter
findings are not only consistent with similar observations in intact rats (57 and
references therein), but also confirm the validity of the yeast model for mammalian
P450 turnover analyses.
7 Jan 2005
14:4
452
AR
AR232-PA45-18.tex
CORREIA
SADEGHI
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
MUNDO-PAREDES
7 Jan 2005
14:4
AR
AR232-PA45-18.tex
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
453
7 Jan 2005
14:4
AR
454
AR232-PA45-18.tex
CORREIA
SADEGHI
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
MUNDO-PAREDES
NEM would inactivate the sulfhydryl groups of several proteins including the ubiquitinating
enzymes, and should therefore be used only at the termination of the ubiquitination reactions.
5
The upward migration of CYP3A due to the formation of dimers, trimers, and/or oligomers
effectively reduces the levels of the parent proteins detected at 55 kDa, thereby leading
to the erroneous conclusion that P450 is lost because of protein degradation.
7 Jan 2005
14:4
AR
AR232-PA45-18.tex
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
455
from immunized goat serum and thus were not fully purified, the corresponding
immunoblot background reflecting the presence of serum contaminants is also
shown with a mock immunoprecipitate from mixtures without liver microsomes
(Figure 3c; last lane). Densitometric assessment of CYP3A4 ubiquitination requires that this background be subtracted from the corresponding profiles in the
other four lanes. Figure 3d illustrates the absolute need for autoclaving in the
immunoblotting analyses of P450 ubiquitination, particularly if neither the Ub
nor P450 molecules are radiolabeled and thus undetectable by autoradiography or
PhosphoImager analyses. Autoclaving of the electroblotted membranes is found to
improve the immunodetection of polyUb epitopes by rehydration of the denatured
proteins (127).
The source of the confounding artifacts often encountered during in vitro P450
ubiquitination assays are illustrated in Figure 4. Freshly prepared liver microsomes
from DEX-pretreated rats (stored as pellets at 80 C for 1 wk) were used for
CYP3A immunoblotting analyses before or after incubation at 37 C for 2 h, under
conditions identical to those previously detailed (129), except that CaCl2, ZnCl2
or MgCl2 were individually included instead of the previously used Ca+2/Zn+2
combination. Note that in the nonincubated (0 h) microsomes, no CYP3A aggregation was detected even after sample overload (Figure 4a). Incubation of these
microsomes at 37 C for 2 h in the presence of CaCl2, ZnCl2, or MgCl2 only minimally increased the detection of CYP3A dimers and trimers, even after sample
overload (Figure 4a). Figure 4b depicts the corresponding protein ubiquitination
pattern of the nonincubated (0 h) microsomes. This profile was clearly enhanced
by the inclusion of NEM during homogenization of the liver (Figure 4b).
On the other hand, immunoblotting analyses of liver microsomes stored as
pellets at 80 C for longer periods (1 year) clearly documented the formation of CYP3A dimers and trimers even without incubation (Figure 4b). Furthermore, immunoblotting analyses of these stored microsomes after incubation at
37 C for 2 h as described (Figure 4a) clearly revealed the presence of CYP3A
oligomers/aggregates at the stacking/running gel interphase (Figure 4c; marked
with an asterisk) in addition to CYP3A dimers and trimers and irrespective of
the cation present. Corresponding immunoblotting analyses with anti-Ub IgGs
of incubations depicted in Figure 4c are shown in Figure 4d. Note the salient
differences in this profile and that seen in Figures 3a and 3c, particularly at the
stacking/running gel interphase (marked with an asterisk). Additional storage of
the nonincubated microsomal suspensions for just a week at 20 C resulted in
the dramatic CYP3A aggregation, which was further intensified by incubation at
37 C for 2 h (Figure 4e). Only slightly lesser aggregation was detected if instead
these suspensions were stored at 80 C for a week (Figure 4e).
Collectively, these findings clearly indicate that maximal detection of liver
P450 ubiquitination requires (a) addition of NEM during liver homogenization;
(b) minimal (<2 wk) storage of microsomal pellets at 80 C; (c) immunoprecipitation of the P450 under scrutiny; and (d) autoclaving of the electroblotted membranes for maximal immunochemical detection. Artifacts such as those depicted in
7 Jan 2005
14:4
456
AR
AR232-PA45-18.tex
CORREIA
SADEGHI
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
MUNDO-PAREDES
7 Jan 2005
14:4
AR
AR232-PA45-18.tex
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
457
Figure 4d,e are observed if microsomes are stored for prolonged periods as pellets
or even as suspensions for just a week. More importantly, comparative inspection
of the immunoblots in Figures 3 and 4 clearly distinguishes the authentic CYP3A
ubiquitination profiles (Figure 3a and c) from the CYP3A aggregation artifacts illustrated in Figure 4d by the absence of any immunochemically detectable CYP3A
at the stacking/running gel interphase and/or bottoms of the gel wells (marked by
an asterisk). A distinguishing feature of truly ubiquitinated CYP3A species is that
they predominantly migrate to a region well above the 55 kDa parent protein but
distinctly below the stacking/running gel interphase where CYP3A aggregates are
usually found.
ACKNOWLEDGMENTS
The authors acknowledge Ms. Zhi-Juan Song for the CYP2C11 ubiquitination
studies, as well as Ms. Suzanne Davoll and Drs. Katy Korsmeyer, Huifen (Faye
Wang) and Bernard Murray for their invaluable contributions to the CYP3A ubiquitination studies. We gratefully acknowledge the financial support of NIH grants
GM44037 and DK26506 that made these studies possible.
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Peters JM, Harris JR, Finley D, eds . 1998.
Ubiquitin and the Biology of the Cell. New
York: Plenum
2. Hicke L. 2001. Protein regulation by
monoubiquitin. Nat. Rev. Mol. Cell Biol.
2:195201
3. Pickart CM. 1997. Targeting of substrates
to the 26S proteasome. FASEB J. 11:
105566
4. Haas AL, Siepmann TJ. 1997. Pathways of ubiquitin conjugation. FASEB J.
11:125768
5. Hershko A, Ciechanover A. 1998. The
ubiquitin system. Annu. Rev. Biochem.
67:42579
6. Scheffner M, Smith S, Jentsch S. 1998.
The ubiquitin conjugation system. See
Ref. 1, pp. 6598
7. Hochstrasser M. 2000. Evolution and
function of ubiquitin-like protein-conjugation systems. Nat. Cell Biol. 2:E153
57
7 Jan 2005
14:4
458
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
AR
AR232-PA45-18.tex
CORREIA
SADEGHI
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
MUNDO-PAREDES
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
7 Jan 2005
14:4
AR
AR232-PA45-18.tex
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
459
7 Jan 2005
14:4
460
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
AR
AR232-PA45-18.tex
CORREIA
SADEGHI
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
MUNDO-PAREDES
dihydralazine-induced hepatitis: specificity of autoantibodies and inductive capacity of the drug. Mol. Pharmacol. 42:
28085
Zhang M, Pickart CM, Coffino P.
2003. Determinants of proteasome recognition of ornithine decarboxylase, a
ubiquitin-independent substrate. EMBO
J. 22:148896
Ghoda L, Sidney D, Macrae M, Coffino
P. 1992. Structural elements of ornithine
decarboxylase required for intracellular degradation and polyamine-dependent
regulation. Mol. Cell. Biol. 12:217885
Sheaff RJ, Singer JD, Swanger J, Smitherman M, Roberts JM, et al. 2000. Proteasomal turnover of p21Cip1 does not require
p21Cip1 ubiquitination. Mol. Cell 5:403
10
Correia MA. 2003. Hepatic cytochrome
P450 degradation: mechanistic diversity
of the cellular sanitation brigade. Drug
Metab. Rev. 35:10743
Gardner R, Cronin S, Leader B, Rine J,
Hampton R, et al. 1998. Sequence determinants for regulated degradation of yeast
3-hydroxy-3-methylglutaryl-CoA reductase, an integral endoplasmic reticulum membrane protein. Mol. Biol. Cell
9:261126
Gardner RG, Hampton RY. 1999. A distributed degron allows regulated entry
into the ER degradation pathway. EMBO
J. 18:59946004
Gilon T, Chomsky O, Kulka R. 1998.
Degradation signals for ubiquitin system
proteolysis in Saccharomyces cerevisiae.
EMBO J. 17:275966
Chen ZJ, Hagler J, Palombella VJ, Melandri F, Scherer D, et al. 1995. Signalinduced site-specific phosphorylation targets IB to the ubiquitin-proteasome
pathway. Genes. Dev. 9:158697
Chen ZJ, Parent L, Maniatis T. 1996.
Site-specific phosphorylation of IB by
a novel ubiquitination-dependent protein
kinase activity. Cell 84:85362
Pyerin W, Taniguchi H, Horn F, Oesch
64.
65.
66.
67.
68.
69.
70.
7 Jan 2005
14:4
AR
AR232-PA45-18.tex
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
71.
72.
73.
74.
75.
76.
77.
78.
Selective loss after labilization of the enzyme. Arch. Biochem. Biophys. 293:9
16
Davies HS, Britt SG, Pohl LR. 1986.
Carbon tetrachloride and 2-isopropyl-4pentenamide-induced inactivation of cytochrome P-450 leads to heme-derived
protein adducts. Arch. Biochem. Biophys.
244:35287
Ortiz de Montellano PR, Mathews JM.
1981. Autocatalytic alkylation of the cytochrome P-450 prosthetic haem group
by 1-aminobenzotriazole. Isolation of an
NN-bridged benzyne-protoporphyrin IX
adduct. Biochem. J. 195:76164
Correia MA, Decker C, Sugiyama K,
Caldera P, Bornheim L, et al. 1987.
Degradation of rat hepatic cytochrome
P-450 heme by 3,5dicarbethoxy-2,6dimethyl-4-ethyl-1,4-dihydropyridine to
irreversibly bound protein adducts. Arch.
Biochem. Biophys. 258:43651
Correia MA, Rettie A, Wrighton SA,
Waxman DJ. 1992. Differential apoprotein loss of rat liver cytochromes P-450
after their inactivation by 3,5-dicarbethoxy-2,6-dimethyl-4-ethyl-1,4-dihydropyridine: A case for distinct proteolytic
mechanisms? Arch. Biochem. Biophys.
294:493503
Etlinger JD, Goldberg AL. 1980. Control of protein degradation in reticulocytes
and reticulocyte extracts by hemin. J. Biol.
Chem. 255:456368
Haas AL, Rose IA. 1981. Hemin inhibits
ATP-dependent ubiquitin-dependent proteolysis: role of hemin in regulating ubiquitin conjugate degradation. Proc. Natl.
Acad. Sci. USA 78:684548
Vierstra RD, Sullivan ML. 1988. Hemin
inhibits ubiquitin-dependent proteolysis
in both a higher plant and yeast. Biochemistry 27:329095
Figueiredo-Pereira ME, Berg KA, Wilk
S. 1994. A new inhibitor of the chymotrypsin-like activity of the multicatalytic
proteinase complex (20S proteasome) induces accumulation of ubiquitin-protein
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
461
7 Jan 2005
14:4
462
89.
90.
91.
92.
93.
94.
95.
96.
97.
98.
99.
AR
AR232-PA45-18.tex
CORREIA
SADEGHI
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
MUNDO-PAREDES
100.
101.
102.
103.
104.
105.
106.
107.
108.
dation of misfolded lumenal and integral membrane proteins. Mol. Biol. Cell
9:20922
Gardner RG, Swarbrick GM, Bays NW,
Cronin SR, Wilhovsky S, et al. 2000. Endoplasmic reticulum degradation requires
lumen to cytosol signaling. Transmembrane control of hrd1p by hrd3p. J. Cell
Biol. 151:6982
Plemper RK, Bordallo J, Deak PM, Taxis
C, Hitt R, Wolf DH. 1999. Genetic interactions of Hrd3p and Der3p/Hrd1p with
Sec61p suggest a retro-translocation complex mediating protein transport for ER
degradation. J. Cell Sci. 112:412334
Bays N, Wilhovsky S, Goradia A,
Hodgkiss-Harlow K, Hampton R. 2001.
HRD4/NPL4 is required for the proteasomal processing of ubiquitinated ER proteins. Mol. Biol. Cell 12:411428
Braun S, Matuschewski K, Rape M,
Thoms S, Jentsch S. 2002. Role of the
ubiquitin-selective CDC48(UFD1/NPL4)
chaperone (segregase) in ERAD of OLE1
and other substrates. EMBO J. 21:61521
Jarosch E, Taxis C, Volkwein C, Bordallo
J, Finley D, et al. 2002. Protein dislocation from the ER requires polyubiquitination and the AAA-ATPase Cdc48. Nat.
Cell Biol. 4:13439
Meyer H, Shorter J, Seemann J, Pappin
D, Warren G. 2000. A complex of mammalian ufd1 and npl4 links the AAAATPase, p97, to ubiquitin and nuclear
transport pathways. EMBO J. 19:2181
92
Ye Y, Meyer H, Rapoport T. 2001. The
AAA ATPase Cdc48/p97 and its partners
transport proteins from the ER into the cytosol. Nature 414:65256
Rabinovich E, Kerem A, Frohlich K,
Diamant N, Bar-Nun S. 2002. AAAATPase p97/Cdc48p, a cytosolic chaperone required for endoplasmic reticulumassociated protein degradation. Mol. Cell
Biol. 22:62634
Tsai B, Ye Y, Rapoport TA. 2002.
Retro-translocation of proteins from the
7 Jan 2005
14:4
AR
AR232-PA45-18.tex
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
109.
110.
111.
112.
113.
114.
115.
endoplasmic reticulum into the cytosol. Nat. Rev. Mol. Cell Biol. 3:246
55
Flierman D, Ye Y, Dai M, Chau V,
Rapoport TA. 2003. Polyubiquitin serves
as a recognition signal, rather than a
ratcheting molecule, during retrotranslocation of proteins across the endoplasmic reticulum membrane. J. Biol. Chem.
278:3477482
Kikkert M, Doolman R, Dai M, Avner R,
Hassink G, et al. 2004. Human HRD1 is
an E3 ubiquitin ligase involved in degradation of proteins from the endoplasmic reticulum. J. Biol. Chem. 279:3525
34
Fang S, Ferrone M, Yang C, Jensen JP,
Tiwari S, Weissman AM. 2001. The tumor autocrine motility factor receptor,
gp78, is a ubiquitin protein ligase implicated in degradation from the endoplasmic reticulum. Proc. Natl. Acad. Sci. USA
98:1442227
Elkabetz Y, Shapira I, Rabinovich E, BarNun S. 2004. Distinct steps in dislocation of luminal endoplasmic reticulumassociated degradation substrates: roles
of endoplamic reticulum-bound p97/Cdc
48p and proteasome. J. Biol. Chem. 279:
398099
Murray BP, Correia MA. 2001. Ubiquitindependent 26 S proteasomal pathway: A
role in the degradation of the native human liver CYP3A4 expressed in Saccharomyces cerevisiae? Arch. Biochem. Biophys. 393:10616
Murray BM, Zgoda VG, Correia MA.
2002 Native CYP2C11: Heterologous expression in Saccharomyces cerevisiae reveals a role for vacuolar proteases rather
than the ubiquitin-26S proteasome system in the degradation of this endoplasmic reticulum enzyme. Mol. Pharmacol.
61:114653
Swanson R, Locher M, Hochstrasser M.
2001. A conserved ubiquitin ligase of the
nuclear envelope/endoplasmic reticulum
that functions in both ER-associated and
116.
117.
118.
119.
120.
121.
122.
123.
124.
463
7 Jan 2005
14:4
464
AR
AR232-PA45-18.tex
CORREIA
SADEGHI
AR232-PA45-18.sgm
LaTeX2e(2002/01/18)
P1: GCE
MUNDO-PAREDES
125. Laney JD, Hochstrasser M. 2002. Assaying protein ubiquitination in Saccharomyces cerevisiae. Methods Enzymol. 351:
24857
126. Tsuji H, Akasaki K. 1994. Identification and characterization of lysosomal enzymes involved in the proteolysis
of phenobarbital-inducible cytochrome
P450. Biol. Pharm. Bull. 17:56871
127. Swerdlow PS, Finley D, Varshavsky A.
1986. Enhancement of immunoblot sensitivity by heating of hydrated filters. Anal.
Biochem. 156:14753
128. Haas AL, Warms JVB, Hershko A, Rose
IA. 1982. Ubiquitin-activating enzyme:
Mechanism and role in protein-ubiquitin
conjugation. J. Biol. Chem. 257:2543
48
129. Zangar RC, Kimzey AL, Okita JR, Wunschel DS, Edwards RJ, et al. 2002. Cytochrome P450 3A conjugation to ubiquitin in a process distinct from classical
1/7/05
2:27 PM
Figure 1 The cellular enzymatic machinery available for P450 ubiquitination. The roles of E1, E2, and E3 enzymes are discussed in the
text. E2-substrate Ub-thioester relay via an E3-Cys residue (a) or directly (b) is shown.
HI-RES-PA45-18-Correia.qxd
Page 1
C-1
C-2
CORREIA
1/7/05
SADEGHI
2:27 PM
Figure 2 Hepatic P450 ERAD: Putative cellular participants. The plausible events in the Ub-dependent 26S proteasomal degradation of an ER-bound DDEP-inactivated CYP3A are illustrated. See the text for specific details.
HI-RES-PA45-18-Correia.qxd
Page 2
MUNDO-PAREDES
HI-RES-PA45-18-Correia.qxd
1/7/05
2:28 PM
Page 3
HI-RES-PA45-18-Correia.qxd
C-4
CORREIA
1/7/05
SADEGHI
2:28 PM
Page 4
MUNDO-PAREDES
7 Jan 2005
14:5
AR
AR232-PA45-19.tex
AR232-PA45-19.sgm
P1: GCE
LaTeX2e(2002/01/18)
10.1146/annurev.pharmtox.45.120403.100037
INTRODUCTION
Multiple myeloma (MM) remains fatal despite all available therapies (1, 2), and
novel approaches that target mechanisms regulating MM cell growth, survival,
and apoptosis are urgently needed. Apoptosis is the primary means by which most
radio- and chemotherapy modalities kill cancer cells (3); conversely, resistance to
apoptosis is one potential mechanism whereby tumor cells evade cytotoxic druginduced and immune-mediated cell death (4). Our studies to date have delineated
apoptotic signaling triggered by various conventional and novel anti-MM agents
(5). Importantly, recent studies show remarkable anti-MM activity of the proteasome inhibitor PS-341/bortezomib (VelcadeTM) even in MM cells refractory to
multiple prior therapies, including dexamethasone (Dex), melphalan, and thalidomide (6, 7). In addition to directly inducing apoptosis of MM cells, multiple other
lines of evidence provided rationale for the use of proteasome inhibitors (PIs) to
0362-1642/05/0210-0465$14.00
465
7 Jan 2005
14:5
466
AR
AR232-PA45-19.tex
CHAUHAN
AR232-PA45-19.sgm
HIDESHIMA
LaTeX2e(2002/01/18)
P1: GCE
ANDERSON
treat MM. First, adhesion of MM cells to bone marrow stromal cells (BMSCs)
triggers transcription and secretion of MM growth factors, such as interleukin-6
(IL-6) or IGF-1, which stimulate the growth of MM cells and also block the cytotoxic effects of chemotherapy (8, 9). Inhibition of proteasomes downregulates
adhesion molecules and secretion of cytokines, thereby abrogating bone marrow
(BM)-dependent growth of MM cells (1012). Second, angiogenesis plays a role
in MM pathogenesis (13, 14), and bortezomib is an antiangiogenic agent (1517).
Finally, in vitro studies showed that bortezomib adds to the cytotoxicity of conventional anti-MM agents including Dex- and DNA-damaging agents (6, 18). Indeed,
based on our preclinical (6) and phase II clinical studies (19), the FDA recently
approved bortezomib for the treatment of relapsed/refractory MM. This successful
development of bortezomib therapy for MM has established proteasome inhibition
as an effective therapeutic strategy for the treatment of cancer.
7 Jan 2005
14:5
AR
AR232-PA45-19.tex
AR232-PA45-19.sgm
LaTeX2e(2002/01/18)
P1: GCE
467
abnormal protein catabolism, neural and muscular degeneration, cellular differentiation, antigen processing, and DNA repair (31). The 26S proteasome complex,
which constitutes up to 2%3% of the total protein in cells, has two 19S units flanking a barrel-shaped 20S proteasome core (24, 29, 32) (Figure 1A). Four stacked
rings comprise the 20S structure: two central rings are surrounded by two
rings, each composed of seven proteins. Most action occurs at six sites located
in the rings: Two sites act like chymotrypsin, which cleaves after hydrophobic
residues; two trypsin-like sites cleave after basic residues; and two are like caspase,
cleaving after acidic residues (33, 34) (Figure 1B).
The 19S units regulate entry into the 20S core chamber of only those proteins
marked for degradation (29, 35). Each 19S unit contains binding sites for ubiquitinated protein, enzymes to depolymerize the ubiquitin chain, and six ATPases
that unfold the proteins, thereby preparing them for entry into the proteasome
(Figure 1C). Attachment of ubiquitin to a target protein is the principal mechanism whereby proteins are marked for degradation by the proteasome. Importantly,
blocking proteasome activity leads to stabilization of inhibitory proteins, thereby
abrogating growth, survival, and triggering apoptosis (Figure 1D).
Most proteasome inhibitors fall in three categories: peptide aldehydes, peptide
boronates, and nonpeptide inhibitors such as lactacystin. Peptide aldehydes (MG132, MG-115, ALLN, or PSI) potently, but reversibly, block the chymotrypsinlike activity; however, they also inhibit lysosmal cysteine and serine proteases and
calpains. The peptide boronates, such as bortezomib/PS-341, are reversible and
more potent and selective than peptide aldehydes. Finally, lactacystin is a natural,
irreversible, nonpeptide inhibitor that is more selective than peptide aldehydes but
less selective than peptide boronates.
7 Jan 2005
14:5
468
AR
AR232-PA45-19.tex
CHAUHAN
AR232-PA45-19.sgm
HIDESHIMA
LaTeX2e(2002/01/18)
P1: GCE
ANDERSON
malignant cells have altered or defective cell cycle proteins, which leads to an
increased proliferation rate. These cells therefore accumulate damaged proteins at
a much higher rate than do normal cells, which in turn increases dependency on the
proteasomal degradation. In contrast, another study showed that quiescent cancer
cells are more susceptible to proteasome inhibition than are normal counterparts
(40). NF-B is linked to proliferation and drug-resistance in cancer cells (41, 42),
and PIs downregulate NF-B activation, thereby enhancing the cytotoxic effects
of chemotherapy (43). Together, these findings suggest that the proteasome is a
valid target for chemotherapy with a tolerable therapeutic index.
7 Jan 2005
14:5
AR
AR232-PA45-19.tex
AR232-PA45-19.sgm
LaTeX2e(2002/01/18)
P1: GCE
469
adhesion molecules [intracellular adhesion molecule (ICAM), vascular cell adhesion molecule (VCAM) and E-selectin] (45).
In the context of MM, NF-B mediates key cellular functions, including immune responses, as well as growth, survival, and apoptosis (46, 47). Intrinsic
activation of NF-B is associated with growth and survival of MM cells. Specifically, adhesion of MM cells to BMSCs triggers NF-B-mediated transcription and
secretion of IL-6 and insulin-like growth factor-I (4648); both IL-6 and IGF-1
promote the survival of MM cells in the BM by blocking apoptosis triggered by
conventional agents such as Dex (49) (Figure 2). Patient MM-derived primary cells
and BMSCs have upregulated NF-B activity relative to normal cells (50). Furthermore, drug-sensitive MM cells show lower NF-B activity than drug-resistant
MM cells, suggesting that NF-B confers chemoresistance (50). Elevated NF-B
levels have also been reported in MM cells derived from patients relapsing after chemotherapy (47). These findings indicate that NF-B is a key regulator of
growth and survival of MM cells in the BM milieu. Importantly, treatment of MM
with bortezomib prevents degradation of IB, thereby blocking not only NF-B
activation but also related cytokine production and the survival advantage for MM
cells conferred by BMSCs (Figure 2).
Bortezomib downregulates NF-B; however, our recent work shows that NF-B
inhibition alone is unlikely to account for the total anti-MM activity of bortezomib
(51, 52). The evidence for this finding is derived from the experiments using a
specific inhibitor of IB, PS-1145. Both PS-1145 and bortezomib blocked TNF-induced NF-B activation by inhibiting phosphorylation and degradation of
IB-. Dex, a conventional anti-MM agent, increases IB- protein and thereby
enhances blockade of NF-B activation by PS-1145. Importantly, both bortezomib
and PS-1145 block NF-B activation; however, in contrast to bortezomib, PS-1145
only partially inhibits MM cell growth (20%40% inhibition by PS-1145 versus
80%90% inhibition by bortezomib) (51), suggesting that NF-B inhibition cannot
account for the overall anti-MM activity of bortezomib.
Multiple genomics and proteomic studies have now established that besides
modulating NF-B, bortezomib indeed affects various other signaling pathways.
For example bortezomib-induced apoptosis is associated with initiation of the
following additional events: (a) activation of classical stress response proteins
such as heat shock proteins, Hsp27, Hsp70, and Hsp90 (17, 53); (b) upregulation
of c-Jun-NH2-terminal kinase (JNK) (54) (Figure 3); (c) alteration of mitochondrial membrane potential and generation of reactive oxygen species (ROS) (55)
(Figure 3); (d) induction of intrinsic cell death pathway, i.e., the release of mitochondrial proteins cytochrome-c/Smac into cytosol and activation of caspase-9
> caspase-3 cascade (Figure 3) (49); (e) activation of extrinsic apoptotic signaling through Bid and caspase-8 cleavage (17) (Figure 3); ( f ) inactivation of
DNA-dependent protein kinase (DNA-PK) (18) (Figure 2), which is essential
for the repair of DNA double-strand breaks; (g) inhibition of MM to BMSCshost interaction, thereby blocking of associated MM growth factor transcription and secretion from BMSCs (56) (Figure 2); and (h) inhibition of MM cell
7 Jan 2005
14:5
470
AR
AR232-PA45-19.tex
CHAUHAN
AR232-PA45-19.sgm
HIDESHIMA
LaTeX2e(2002/01/18)
P1: GCE
ANDERSON
7 Jan 2005
14:5
AR
AR232-PA45-19.tex
AR232-PA45-19.sgm
LaTeX2e(2002/01/18)
P1: GCE
471
Bortezomib in Clinic
It is known that bortezomib mediates its effects by inhibiting cellular proteasomes;
however, whether proteasome inhibition is universally required for bortezomibtriggered apoptosis is unclear. Our findings showed that treatment with bortezomib
led to 82% and 88% inhibition of proteasome activity in bortezomib-resistant
SUDHL4 and bortezomib-sensitive SUDHL6 lymphoma cells, respectively (53).
Together, these data confirm that (a) the proteasome inhibition pathway is not
defective in bortezomib-resistant DHL4 cells, and (b) proteasome inhibition is not
correlated with apoptosis.
Direct determination of proteasome inhibition in patient blood and tissue samples was examined in phase I studies. Bortezomib was well tolerated at doses,
resulting in up to 80% proteasome inhibition (65). Furthermore, extended dosing
did not further reduce sensitivity to proteasome inhibition. These data suggest
that proteasome inhibition is the main function of the proteasome inhibitor, but
that proteasome blockade may not correlate with degree of cytotoxicity in cancer
cells.
7 Jan 2005
14:5
472
AR
AR232-PA45-19.tex
CHAUHAN
AR232-PA45-19.sgm
HIDESHIMA
LaTeX2e(2002/01/18)
P1: GCE
ANDERSON
PHASE II STUDIES IN MM
A phase II bortezomib study was conducted in MM patients who had relapsed
and were refractory to multiple prior therapies. Each cycle of therapy included
bortezomib (1.3 mg/m2) administered twice weekly for 2 weeks, with 1 week off
(19). Oral dexamethasone was given to patients with a suboptimal response after
two cycles, and eight cycles of therapy were given to responders. Two hundred and
two patients were enrolled, all of whom received corticosteroids, 92% alkylating
agents, 81% anthracyclines, 83% thalidomide, and 64% stem-cell transplant; the
median number of prior therapies was six. Poor prognostic factors at enrollment included elevated beta-2-microglobulin and abnormal cytogenetics. Of 193 patients,
4% achieved a CR, evidenced by MM protein undetectable by both electrophoresis
and immunofixation; 6% achieved showed a near CR, evidenced by MM protein
detectable only by immunofixation. Partial response (PR) was noted in 18% and
minimal response (MR) in 7% patients. Overall response rate (CR + PR + MR)
was 35%.
Response to bortezomib was examined relative to prognostic factors, including
the patients age; gender; type of MM; beta-microgloblin levels; extent of disease,
i.e., plasma cells in BM; chromosomal abnormalities, i.e., deletion of chromosome
13; and intensity of prior therapies. Statistical analysis (univariate) indicated that
>50% plasma cells in BM significantly predicted a lower response rate. Multivariate analysis suggested that older age (65 years or older) and >50% plasma cells
in BM significantly predicted for lower response rate. Major responses (CR and
PR) were associated with improved hemoglobin and nonmyeloma immunoglobulin levels, decreased transfusion requirements, increased platelet counts, and improvements in global quality of life and disease symptoms. Median time to disease
progression (TTP) for all patients was 7 months, compared with TTP of 3 months
for the last treatment before enrollment, and 13 months of TTP for patients achieving a CR or PR. Responses were durable: Median response duration was 12 months
among patients achieving an objective response (CR + PR + MR) and 15 months
among those achieving a CR or near CR. The median survival for the entire population (n = 202) was 16 months and patients achieving a major response (CR + PR)
7 Jan 2005
14:5
AR
AR232-PA45-19.tex
AR232-PA45-19.sgm
LaTeX2e(2002/01/18)
P1: GCE
473
survived significantly longer than those who did not (P = 0.007). Of 74 patients
who did not achieve at least a MR and therefore received dexamethasone in combination with bortezomib, 18% improved; these included 6 patients with steroidrefractory disease, indicating that bortezomib can overcome resistance to steroids.
Most commonly reported adverse events were nausea, vomiting, diarrhea, fatigue,
loss of appetite including anorexia, constipation, anemia, thrombocytopenia, peripheral neuropathy, and pyrexia.
CONCLUSIONS
The proteasome is a promising target in the treatment of cancer. Ongoing research
in this field will unveil the complex mechanisms whereby proteasome inhibitors
impact a wide array of cellular functions with a differential sensitivity of normal
versus cancer cells. More specific therapeutic targets may be the E2 and E3 ubiquitin enzymes, which target unique proteins. Proteasomes have six active sites, and
blocking the chymotrypsin-like site decreases protein degradation significantly,
whereas inhibition of trypsin- or caspase-like sites is less effective. Whether simultaneous inhibition of all three activities is more cytotoxic to cancer versus
normal cells remains to be examined. The proteasome inhibitor bortezomib/PS341 has shown potent preclinical activity in vitro as well as therapeutic activity
in hematologic malignancies, especially MM. Importantly, bortezomib is the first
treatment in more than a decade to be FDA approved for patients with MM, and
the European Commission also granted marketing authorization for bortezomib
for the treatment of patients with MM who have received at least two prior therapies and have demonstrated disease progression on their last therapy. Ongoing
clinical trials are examining its efficacy in other hematologic malignancies and
solid tumors.
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Anderson KC. 2004. Bortezomib therapy
for myeloma. Curr. Hematol. Rep. 3:65
2. Anderson KC. 2004. Plasma cell tumors. In
Cancer Medicine, ed. JF Holland, E Frei III,
RC Bast, DW Kufe, DL Norton, RR Weichselbaum. Baltimore: Williams and Wilkins.
In press
3. Wyllie A. 1980. Glucocorticoid-induced
thymocyte apoptosis is associated with endogenous endonuclease activation. Nature
284:55556
7 Jan 2005
14:5
474
7.
8.
9.
10.
11.
12.
13.
14.
15.
AR
AR232-PA45-19.tex
CHAUHAN
AR232-PA45-19.sgm
HIDESHIMA
LaTeX2e(2002/01/18)
P1: GCE
ANDERSON
7 Jan 2005
14:5
AR
AR232-PA45-19.tex
AR232-PA45-19.sgm
LaTeX2e(2002/01/18)
P1: GCE
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
475
7 Jan 2005
14:5
476
49.
50.
51.
52.
53.
54.
55.
56.
57.
AR
AR232-PA45-19.tex
CHAUHAN
AR232-PA45-19.sgm
HIDESHIMA
LaTeX2e(2002/01/18)
P1: GCE
ANDERSON
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
1/7/05
2:28 PM
Figure 1 (A) Structure and function of proteasomes; (B) cross-sectional view of 26S proteasome complex; (C) process
of degradation of ubiquitinated proteins by proteasome complex; and (D) bortezomib/velcade blocks the proteasomal
protein degradation resulting in accumulation of cytotoxic proteins.
HI-RES-PA45-19-Anderson.qxd
Page 1
C-1
HI-RES-PA45-19-Anderson.qxd
C-2
CHAUHAN
1/7/05
HIDESHIMA
2:28 PM
Page 2
ANDERSON
4 Dec 2004
13:15
AR
AR232-PA45-20.tex
AR232-PA45-20.sgm
P1: GCE
LaTeX2e(2002/01/18)
10.1146/annurev.pharmtox.45.120403.095821
Pharmacogenetics
Allan E. Rettie1 and Jeffrey P. Jones2
1
OVERVIEW
Cytochrome P450s are a superfamily of oxygen-activating enzymes that carry out
an enormous range of metabolic reactions on endogenous and exogenous substrates
in processes both beneficial and deleterious to the organism (1, 2). Therapeutically
administered drugs, endogenous eicosanoids, steroids, and bile salts, as well as
carcinogens and environmental pollutants, are but a few important targets for these
versatile catalysts (3). Annotation of the human genome has revealed the presence
of some 57 human P450 genes (4), but less than a dozen of these play important
roles in the hepatic clearance of drugs (5).
CYP2C9 is a major human liver form of P450 that has drawn considerable
attention from researchers owing, in large part, to its role in causing adverse drug
reactions (ADRs). ADRs, which are projected to cause hundreds of millions of
0362-1642/05/0210-0477$14.00
477
4 Dec 2004
13:15
478
AR
RETTIE
AR232-PA45-20.tex
AR232-PA45-20.sgm
LaTeX2e(2002/01/18)
P1: GCE
JONES
dollars in health care costs per year in the United States alone, often result from
unanticipated changes in P450 enzyme activity secondary to genetic polymorphisms or metabolically based drug-drug interactions (6, 7). Both mechanisms
are highly pertinent to ADRs involving CYP2C9. These can be especially serious
because several of the enzymes substrates exhibit a narrow therapeutic index,
therefore the resulting clinical consequences can be severe.
This review summarizes our current knowledge of the enzymes structurefunction relationships, drug-drug interactions, pharmacogenetics, and gene regulation, and it attempts to relate these base-line parameters to key clinical and
toxicological features of this important enzyme. Several earlier reviews of CYP2C
enzyme properties, function, and genetics may prove to be useful adjuncts to this
material (811).
INTRODUCTION
CYP2C9 is one of four functional human CYP2C genes located on the long arm
of chromosome 10. Within the CYP2C subfamily, which also comprises CYP2C8,
CYP2C18, and CYP2C19, CYP2C9 is arguably the most important member for
several reasons. First, it is the largest contributor among these four isoforms to
total human liver microsomal P450 content, with estimates of mean microsomal levels ranging from 40 10 pmol/mg (12) to as high as 89 9 pmol/mg
(13). Only CYP3A4 is a more quantitatively significant P450 enzyme in human
liver. Second, like CYP3A4, CYP2C9 metabolizes a host of clinically important
drugs (Table 1). Indeed, it has been estimated that CYP2C9 is responsible for the
metabolic clearance of up to 15% of all drugs that undergo Phase I metabolism (5).
Third, drug-drug interactions pose therapeutic management problems for several
CYP2C9 substrates, especially those involving low therapeutic index substrates,
such as (S)-warfarin, tolbutamide, and phenytoin. Fourth, CYP2C9 is subject to
significant genetic polymorphism, such that up to 40% of Caucasian populations
are carriers of alleles that encode partially defective functional forms of the enzyme. Such gene-drug interactions can exacerbate adverse drug reactions with the
same battery of CYP2C9 substrates that display an intrinsically low margin of
safety.
4 Dec 2004
13:15
AR
AR232-PA45-20.tex
AR232-PA45-20.sgm
LaTeX2e(2002/01/18)
P1: GCE
Oral
hypoglycemics
Oral
anticoagulants
Diuretics
and uricosurics
Angiotensin II
blockers
Flurbiprofen
Diclofenac
Naproxen
Piroxicam
Suprofen
Ibuprofen
Mefenamic acid
Celecoxib
Tolbutamide
Glyburide
Glipizide
Glimepiride
(S)-Warfarin
(S)-Acenocoumarol
(Phenprocoumon)
Torsemide
Ticrynafen
Sulfinpyrazone
sulfide
Losartan
Irbesartan
Candesartan
Class
(cont.)
Antiasthmatics
Anticonvulsants
Anticancer agents
Substrates
(cont.)
Zafirlukast
(Zileuton)
Phenytoin
(Phenobarbital)
(Trimethadione)
Cyclophosphamide
(Tamoxifen)
Class
Substrates
479
Endogenous
compounds
Arachidonic acid
5-Hydroxytryptamine
Linoleic acid
Miscellaneous
Mestranol
Fluvastatin
9-Tetrahydrocannabinol
(Benzopyrene)
(Pyrene)
(Fluoxetine)
(Sildenafil)
(Rosiglitazone)
Parentheses indicates that (where known) other P450s or metabolic pathways play a major role in clearance.
4 Dec 2004
13:15
480
AR
RETTIE
AR232-PA45-20.tex
AR232-PA45-20.sgm
LaTeX2e(2002/01/18)
P1: GCE
JONES
4 Dec 2004
13:15
AR
AR232-PA45-20.tex
AR232-PA45-20.sgm
LaTeX2e(2002/01/18)
P1: GCE
481
4 Dec 2004
13:15
482
AR
RETTIE
AR232-PA45-20.tex
AR232-PA45-20.sgm
LaTeX2e(2002/01/18)
P1: GCE
JONES
aromatic residues in binding of this ligand (44). In contrast, Arg97 plays a more
important role in binding heme, whereas Asp293 appears to have dual functions
in controlling product regioselectivity as well as maintaining holo-enzyme stability (42, 4446). Ser286 and Asn289 are I-helix residues important in conferring
substrate specificity toward the NSAIDs, ibuprofen, and diclofenac (38), whereas
Ser365 appears to be the target nucleophile adducted by activation of tienilic acid
(43). Arg144 and Ile359 together largely determine the genetic background that
confers a wild-type phenotype (see following section).
Although significant progress was made prior to the initial crystallization of
CYP2C9 in mapping hydrophobic active site residues of CYP2C9 by mutagenesis,
the nature of the anionic binding site in CYP2C9 remained elusive. Early homology
modeling efforts gave rise to several predictions for charged amino acids that could
4 Dec 2004
13:15
AR
AR232-PA45-20.tex
AR232-PA45-20.sgm
LaTeX2e(2002/01/18)
P1: GCE
483
4 Dec 2004
13:15
484
AR
RETTIE
AR232-PA45-20.tex
AR232-PA45-20.sgm
LaTeX2e(2002/01/18)
P1: GCE
JONES
crystal structures of CYP2C9 are indicative of a large active site that might readily
accommodate more than one ligand (44, 48).
Although CYP2C9 activation is now a well-documented phenomenon in vitro,
it remains to be seen whether it has clinical relevance. Few studies have been
performed as yet, but Tracy and coworkers did report a modest, yet statistically
significant, increase (11%) in flurbiprofen clearance following cotreatment with
dapsone in vivo (61). It is worth noting that an in vivo significance for such allosteric
phenomena with P450s is not established even for the much more intensively
studied CYP3A4 enzyme. Therefore, for the foreseeable future CYP2C9 activation
may remain an in vitro curiosity, albeit one that promotes new ideas about the
elasticity of the P450 active site.
CYP2C9 Pharmacogenetics
The first indications of polymorphism in the CYP2C9 gene arose when multiple
cDNAs were cloned in the late 1980s and early 1990s. Subsequently, a systematic
investigation of possible sites of allelic variation confirmed the existence of the
CYP2C9 2 and CYP2C9 3 variants at significant frequencies (close to 10%) in
a Northern European population (62). Population studies by several other groups
extended these findings and it is now clear that up to 40% of Caucasians possess one
or more variant CYP2C9 allele (63). This high frequency has prompted numerous
studies aimed at determining the functional effects of these common CYP2C9
variants.
The CYP2C9 2 allele reflects a missense mutation in exon 3 that causes a
nonconservative ArgCys substitution at amino acid 144. The consensus view
from in vitro studies conducted with the recombinantly expressed CYP2C9.2 is
that this mutation causes a small decrement in Vmax (0%35%) and little or no
change in the Km for substrate catalysis (64). In vivo studies have generally been
difficult to interpret owing to the paucity of CYP2C9 2/ 2 homozygotes available
for study, but recent clinical investigations that did include this test group also
suggest modest decreases in drug clearance attributable to this mutation (15, 65).
Arg144 maps to helix C, which is located on the exterior of the protein and forms
part of the putative P450 reductase binding site (66). Loss of activity may reflect
altered affinity for the coenzyme P450 reductase, which appears to bind reversibly
to positively charged surface amino acids on P450s (67).
The CYP2C9 3 allele arises from a missense mutation in exon 7 that causes
an IleLeu substitution at amino acid 359. In vitro and in vivo experiments consistently demonstrate substantial loss of enzyme activity owing to this mutation
(64). In fact, we recently identified five 3/ 3 homozygotes in a Caucasian anticoagulation clinic population and were able to demonstrate that this mutation
is associated with low warfarin dose and increased risk of bleeding during the
warfarin stabilization phase (68). Loss of activity for CYP2C9.3 reflects a combination of decreased Vmax and increased Km for CYP2C9 substrates (40). Recent
(unpublished) studies from our group confirm that the spectral binding constant,
4 Dec 2004
13:15
AR
AR232-PA45-20.tex
AR232-PA45-20.sgm
LaTeX2e(2002/01/18)
P1: GCE
485
Ks, is increased substantially for CYP2C9.3 relative to the wild-type enzyme. The
structural basis for diminished P450 activity as a consequence of the 3 allele is
not yet clear, as new 3D information for CYP2C9 places this residue outside the
active site of the enzyme, some distance from the heme (44, 48). Ile359 is situated
below the SRS-6 region (70), which contains the important orienting amino acid
F476. The physical proximity of the 3 locus and this active-site region could conceivably result in global conformational changes that secondarily diminish binding
affinity. However, more studies are needed to better explain the functional deficit
attributable to this important CYP2C9 polymorphism.
Extensive resequencing of CYP2C9 in ethnically diverse populations demonstrates that this gene is highly polymorphic (71, 105) (http://egp.gs.washington.
edu). At the time of writing (April, 2004), a total of 12 CYP2C9 coding-region alleles were listed on the P450 Allele Web Site (http://www.imm.ki.se/CYPalleles),
and to our knowledge, all but the 4 and 7 alleles have been independently verified
by multiple research groups. In our own laboratory, resequencing across 60 kb of
CYP2C9 in 192 warfarin patients of Caucasian origin revealed a total of 129 single
nucleotide polymorphism (SNP) sites (105). The prevalence of coding-region mutations in this study population (allele frequency in parentheses) decreased in the
following order: 2 (11%), 3 (6%), 11 (1%), and 12 and 9 (0.5%). Consideration of sequence variation in the CYP2C9 gene allowed us to infer 23 haplotypes,
10 of which are represented at a frequency of >3% (105). In another study, resequencing of DNA from 92 individuals across three different racial groups predicted
at least 21 haplotypes (71). The 2 and 3 alleles are each isolated on one major
haplotype background, and both appear to be more significant contributors to variability in warfarin maintenance dose than any of the other eight major haplotypes
(105). A qualitatively similar situation has been reported for the warfarin analog
acenocoumarol (72).
CYP2C9 polymorphisms vary dramatically between different ethnic populations. An early genotyping study by Goldsteins group demonstrated that the
CYP2C9 2 and CYP2C9 3 variants that are common in Caucasians were represented in African-Americans but at much lower allele frequencies (1%2%) (73).
CYP2C9 6 was detected by the same group in one African-American patient who
had an adverse reaction to phenytoin. This rare, null allele is devoid of activity
owing to a splicing mutation that results in a truncated protein (74). CYP2C9 5,
D360E, is selectively expressed in African-Americans at an allele frequency of
1% (75, 76). Recombinant CYP2C9.5 exhibited a large increase in Km for several
substrates, but little change in Vmax, and we have suggested that this was predictive of a decrease in the in vivo catalytic efficiency of the enzyme. However, the
infrequence of this variant complicates further in vivo studies. Genotyping studies
in Korean and East Asian populations have not detected the CYP2C9 2 polymorphism, although the CYP2C9 3 allele is present at low frequencies (1%2%)
(77). Similar to the situation with CYP2C9 6, a rare polymorphism CYP2C9 4
(I359T), has been reported in one Japanese patient who had an adverse reaction
to phenytoin (78). Not surprisingly, recombinant CYP2C9.4 exhibited defective
4 Dec 2004
13:15
486
AR
RETTIE
AR232-PA45-20.tex
AR232-PA45-20.sgm
LaTeX2e(2002/01/18)
P1: GCE
JONES
metabolism of several substrates in vitro (79). Recently, four new CYP2C9 SNPs
were reported in a Hong Kong Chinese anticoagulation clinic population, I181L,
H184P, Q192P, and L208V, but on closer examination these appear to be
spurious and likely a consequence of improper primer design (80). Further studies
are required to determine the frequency of occurrence of the newer ethnic-specific
CYP2C9 alleles, establish haplotypes in these populations, and delineate their
functional consequences.
4 Dec 2004
13:15
AR
AR232-PA45-20.tex
AR232-PA45-20.sgm
LaTeX2e(2002/01/18)
P1: GCE
487
values. As for constitutive expression of the gene, further studies are needed to
evaluate upstream regulatory sequences and establish basic mechanisms of ontogenic regulation.
4 Dec 2004
13:15
488
AR
AR232-PA45-20.tex
RETTIE
AR232-PA45-20.sgm
LaTeX2e(2002/01/18)
P1: GCE
JONES
In conclusion, in the 20 plus years since CYP2C9 was first identified in human
liver (103), this isoform has become one of the most studied human P450s owing largely to its quantitative significance in oxidative drug metabolism, role in
adverse drug reactions, and pharmacogenetic variability. CYP2C9 is also the first
human P450 enzyme crystallized, and this pivotal event can be expected to propel
future structural, biochemical, biophysical, and clinical studies aimed at a fuller
understanding of this enzymes role in xenobiotic and endobiotic disposition.
ACKNOWLEDGMENTS
This work was supported in part by NIH grants GM32165, GM68797, and ES009
122. The authors would like to thank Dr. T.S. Tracy for helpful comments and
criticism.
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Guengerich FP. 2001. Uncommon P450catalyzed reactions. Curr. Drug Metab.
2:93115
2. Guengerich FP. 2003. Cytochrome P450
oxidations in the generation of reactive
electrophiles: epoxidation and related reactions. Arch. Biochem. Biophys. 409:59
71
3. Gonzalez FJ, Kimura S. 2003. Study of
P450 function using gene knockout and
transgenic mice. Arch. Biochem. Biophys.
409:15358
4. Nelson DR. 2002. Introductory remarks
on human CYPs. Drug Metab. Rev. 34:
15
5. Evans WE, Relling MV. 1999. Pharmacogenomics: translating functional genomics into rational therapeutics. Science
286:48791
6. Pirmohamed M, Park BK. 2003.
Cytochrome P450 enzyme polymorphisms and adverse drug reactions.
Toxicology 192:2332
7. Thummel KE, Kunze KL, Shen DS. 2000.
Metabolically-based drug-drug interactions: principles and mechanisms. See
Ref. 104, pp. 319
4 Dec 2004
13:15
AR
AR232-PA45-20.tex
AR232-PA45-20.sgm
LaTeX2e(2002/01/18)
P1: GCE
14.
15.
16.
17.
18.
19.
20.
21.
489
4 Dec 2004
13:15
490
AR
RETTIE
AR232-PA45-20.tex
AR232-PA45-20.sgm
LaTeX2e(2002/01/18)
P1: GCE
JONES
39.
40.
41.
42.
43.
44.
45.
46.
4 Dec 2004
13:15
AR
AR232-PA45-20.tex
AR232-PA45-20.sgm
LaTeX2e(2002/01/18)
P1: GCE
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
491
4 Dec 2004
13:15
492
65.
66.
67.
68.
69.
70.
71.
72.
73.
AR
RETTIE
AR232-PA45-20.tex
AR232-PA45-20.sgm
LaTeX2e(2002/01/18)
P1: GCE
JONES
4 Dec 2004
13:15
AR
AR232-PA45-20.tex
AR232-PA45-20.sgm
LaTeX2e(2002/01/18)
P1: GCE
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.
96.
97.
98.
99.
100.
493
4 Dec 2004
13:15
494
AR
RETTIE
AR232-PA45-20.tex
AR232-PA45-20.sgm
LaTeX2e(2002/01/18)
P1: GCE
JONES
7 Jan 2005
14:6
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
P1: GCE
LaTeX2e(2002/01/18)
10.1146/annurev.pharmtox.45.120403.095825
Nonstandard abbreviations: AOE, 2-amino-8-oxo-9,10-epoxy-decanoic acid; ATRA, alltrans-retinoic acid; CBHA, m-carboxylcinnamic acid bis-hydroxamide; CDK, cyclindependent kinase; DAC, 5-aza-2 -deoxycytidine; HAT, histone acetyltransferase; HDAC,
histone deacetylase; HMT, histone methyltransferase; IMID-1, immunomodulatory thalidomide derivative 1; PB, phenyl butyrate; SAHA, suberoylanilide hydroxamic acid; SB,
sodium butyrate; TRAIL, tumor necrosis factor-related apoptosis-inducing ligand; TSA,
trichostatin A.
0362-1642/05/0210-0495$14.00
495
7 Jan 2005
14:6
496
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
DRUMMOND ET AL.
INTRODUCTION
The packaging of DNA into the higher order and dynamic structure of chromatin
provides a pivotal point of control for gene expression by regulating access of
transcription factors to DNA. Chromatin is composed of multiple repeating units
termed nucleosomes, which are comprised of 146 base pairs of DNA wrapped
around a core of eight histone proteins composed of two copies each of H2A,
H2B, H3, and H4. Posttranslational modifications play a prominent role in the
regulation of gene expression and signal transduction pathways. Phosphorylation,
methylation, acetylation, ubquitination, and sumoylation are the known modifications thought to influence chromatin architecture and regulate gene transcription. The composition and consequences of these various histone modifications
are often referred to as the histone code, orchestrating an intricate regulation of
nucleosomal structure, DNA accessibility, and gene transcription (Figure 1). To
date, acetylation is the most thoroughly studied of these modifications; and while
the acetylation state of chromatin proteins is unquestionably very dynamic, it
seems to depend on the net local balance between histone acetyltransferase (HAT)
and histone deacetylase (HDAC) activities. Empirically observed is the fact that
HDAC activity is invariably increased in cancer cells, resulting in altered gene
transcription, impaired differentiation, increased cell survival, and dysregulated
proliferation (1).
7 Jan 2005
14:6
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
497
transcriptional control (4, 5). The code is set by a variety of histone tailderivatizing
enzymes, including HATs for the acetylation of lysine residues (6), histone methyltransferases (HMTs) for methylation of histone lysine and arginine residues (7,
8), serine kinases for the phosphorylation of specific histone serine residues (9),
ubiquitin ligase for the addition of the 76-amino-acid 9-kDa protein ubiquitin
to specific lysine residues (10), and the sumoylation of lysine by the 11-kDa
small ubiquitin-related modifier (SUMO) (11). In addition, the activities of modification destabilizing enzymes such as HDACs, methylases, phosphatases, and
ubiquitin and ULP-related proteases help shape the status of the code. The complexity of transcriptional regulation by histone modifications is further enhanced
by the interaction of HATs and HDACs with other proteins involved in chromatin modification, including methyl CpG-binding proteins and ATP-dependent
chromatin-remodeling complexes, which can lead to replication propagated and
more enduring epigentic modifications of DNA, such as the gene silencing cytosine
methylation of specific CpG dinucleotides (1214).
The setting of the histone code involves establishing defined patterns of histone
tail modifications, whereupon a particular modification in turn affects subsequent
modifications. For example, histone deacetylation has been shown to activate lysine
(K) methylation, resulting in relatively stable transcriptional silencing (15). In an
eloquent experiment demonstrating sequential histone modifications, Kouzarides
and colleagues showed that upon estrogen stimulation, H3 is acetylated initially
at K18, then at K23, and finally methylated at R17 (16). A specific set of histone modifications was proposed to direct DNA methylation (17). The reading of
the code can be accomplished through recognition of particular modifications or
groups of modifications (18, 19). The bromodomain of proteins such as BRG1 and
TAFII250 and the chromodomain of HP1 recognize acetylated lysines and methylated lysines, respectively (2022). Certain combinations of modifications can also
dictate the recruitment of various cis- or trans-acting regulatory proteins. The role
of the particular modification in transcriptional signaling may also be influenced
by the degree and stability of the modification. Lysine residues may be modified
with one, two, or three methyl groups, and the degree of methylation determines
if transcription of certain genes is activated or repressed (23, 24). The methylated
lysines, and more so the methylated cytosines in DNA, are more stable modifications than the relatively dynamic modifications of histone tail acetylation and
phosphorylation. Thus, with the lack of any known histone or DNA demethylases,
methylation may be more important in epigenetic memory, whereas the acetylation status of histones may be more of a switch that can be rapidly reset and allow
transcription to respond more rapidly to changes in the cells environment.
The histone code is just beginning to be deciphered and thus its complexity
and its role in carcinogenesis are far from understood. Although it is obvious that
a wide variety of posttranslational protein modifications are responsible for regulating transcription of any given gene and as such can play important roles in
human cancer cell behavior, the remainder of this review focuses specifically on
the preclinical and clinical development of HDAC inhibitors as potential anticancer
7 Jan 2005
14:6
498
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
DRUMMOND ET AL.
agents. In this regard it is important to note that the activity of a wide variety of
nonhistone transcription factors and co-regulators of transcription are known to
be modified by acetylation, and both are structurally and functionally affected by
HDAC inhibitors. Acetylation may enhance or inhibit the function of transcriptional activators as well as transcriptional repressors; therefore, enhancing their degree of acetylation by cell treatment with an HDAC inhibitor can either increase or
repress the transcription of genes regulated by such nonhistone proteins (Table 1).
TFIIE (25), TFIIF (25), p53 (26), androgen receptor (27), estrogen receptor- (28),
and GATA-1 (29, 30) are promoter-binding and transcription-regulating proteins
shown to be acetylated in response to HDAC inhibition. In addition, other DNA
binding nonhistone proteins are functionally affected by acetylation. For example,
HMG-17 is a nucleosomal binding protein responsible for unfolding the higher order structure of chromatin and thus exerts indirect control over gene transcription;
and acetylation of HMG-17 has been shown to reduce its binding to chromatin
(31).
Classification of HDACs
There are three major groups or classes of mammalian HDACs based on their
structural homologies to the three distinct yeast HDACs: Rpd3 (class I), Hda1
(class II), and Sir2/Hst (class III). Class III HDACs consist of the large family of
sirtuins (SIRs) that are evolutionarily distinct, with a unique enzymatic mechanism dependent on the cofactor NAD+, and are virtually unaffected by all HDAC
inhibitors currently under development (32, 33). This review focuses on the NAD+independent class I and II HDACs (Figure 3), as they are evolutionarily similar,
contain an active site zinc as a critical component of their enzymatic pocket, have
been more thoroughly described in association with cancer, and are thought to
be comparably inhibited by most currently available HDAC inhibitors. The Rpd3
homologous class I HDACs include HDAC1, HDAC2, HDAC3, and HDAC8.
They are widely expressed in a variety of tissues and are primarily localized in
the nucleus. The Hda1 homologous class II HDACs include HDAC4, HDAC5,
HDAC6, HDAC7, HDAC9 (a and b isoforms), and HDAC10, and are structurally
much larger in size. Class II HDACs can shuttle between the nucleus and cytoplasm (3437), suggesting different functions and cellular substrates from Class I
HDACs. HDAC6 in particular is predominantly localized in the cytoplasm (38).
Class II HDACs also display a more limited tissue distribution (3941). HDACs
4, 8, and 9 are expressed to a greater extent in tumor tissues than in normal tissues,
with HDAC 4 demonstrating the greatest difference in this regard (39). Class II
enzymes have also been shown to be specifically involved in differentiation (40).
Finally, HDAC6 and HDAC10 are unique among class II HDACs in having two
catalytic domains (37, 40). Although there is some evidence that certain HDAC inhibitors display different degrees of HDAC specificity, considerable research must
still be performed to delineate differences in HDAC function, their roles in cancer,
and their sensitivities to drugs. Some of these differentiating features are reviewed
7 Jan 2005
14:6
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
499
Intracellular function
Reference(s)
p53
Tumor suppressor
c-Myb
(147)
GATA-1
(29, 30)
Estrogen
receptor-
(28)
TFIIE
(25)
TFIIF
(25)
Androgen
receptor
(27)
hsp90
(82)
-tubulin
Microtubule component
(61, 148)
HMG-17
(31)
HMGI
(149)
TCF
Transcriptional regulator
(150)
PCNA
(151)
EKLF
(152)
ACTR
(153)
HNF-4
Transcriptional activation
(154)
Importin-
(155)
NF-B
(156)
ER81
(157)
SF-1
(158)
Ku70
Suppresses apoptosis
(159)
UBF
(160)
Sp3
(161)
TAL1
(162)
YY1
(163)
E2F1
(164)
MyoD
(165)
PCNA, proliferating cell nuclear antigen; SF-1, steroidogenic factor-1; UBF, architectural upstream binding factor.
7 Jan 2005
14:6
500
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
DRUMMOND ET AL.
in detail elsewhere (39). Unraveling specific roles by these HDAC isozymes during
human tumorigenesis will further incentivize development of more specific HDAC
inhibitors (42), potentially enhancing their clinical activity as well as decreasing
their nonspecific toxicities, while also optimizing potential interactions with other
rationally designed and integrated therapeutic agents.
7 Jan 2005
14:6
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
501
Figure 2 Structural classes of HDAC inhibitors. Six basic classes of HDAC inhibitors
are shown: (a) small-molecular-weight carboxylates, including sodium butyrate, valproic
acid, and sodium phenylbutyrate; (b) hydroxamic acids, including CBHA, TSA, SAHA,
and LAQ824; (c) benzamides, including MS-275 and CI-994; (d) epoxyketones, including
AOE and trapoxin B; (e) cyclic peptides, including depsipeptide and apicidin; and ( f ) hybrid
molecules, such as CHAP31 and CHAP50.
7 Jan 2005
14:6
502
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
DRUMMOND ET AL.
Figure 3 Structural basis for hydroxamic acid inhibition of HDACs. (A) Structural homology of class I and II HDACs showing hydroxamate-inhibiting catalytic domains.
(B) Functional components of hydroxamic acidtype HDAC inhibitors (LAQ824). Hydroxamic acidbased HDAC inhibitors are composed of four primary functional components: (a) a
zinc-chelating hydroxamic acid, (b) a linker region, (c) a polar site, and (d) a hydrophobic cap
that blocks the active site. (C) Predicted transition state inhibition of HDAC1 by LAQ824.
hydrogen bond contacts with the ketone. Elimination of the ketone, or reduction of
the ketone to an alcohol, abolishes the activity of these molecules (50). Trapoxin
B and HC-toxin also contain a five-carbon linker for transversing the cavity and a
cyclic tetrapeptide capable of acting as a hydrophobic cap for the cavity. Trapoxin
B is a hybrid molecule and can also be listed with the cyclic peptide HDAC inhibitors. The combination of cyclic peptide and epoxyketone resulted in nanamolar
HDAC inhibitory activity.
The carboxylates or short-chain fatty acids, including sodium butyrate, valproic
acid, and sodium phenylbutyrate, have much weaker HDAC inhibition constants
(Kis), commonly in the millimolar range. In spite of their weak activity, several of
these agents have been studied clinically (51) owing in part to their clinical use for
alternative medical indications. The most commonly studied members of this class
are simple molecules with alkyl or phenylalkyl carboxylates. The carboxylate is
thought to coordinate with the zinc ion in the active site, albeit more poorly than
in the case of hydroxamates.
Cyclic peptide HDAC inhibitors have been discovered or developed that either
contain an epoxyketone group (HC-toxin, trapoxin B) or are devoid of it (Apicidin,
7 Jan 2005
14:6
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
503
Depsipeptide). In general, these inhibitors have nanamolar HDAC inhibitory activity and can have either irreversible (epoxyketone-based) or reversible mechanisms
of action. The macrocyclic peptide portion of the inhibitor binds tightly to the rim
or opening of the channel to the active site, whereas an aliphatic linker navigates the
channel to the active site (44). Depsipeptide, also known as FK228, is a prodrug
that requires intracellular reduction to liberate a sulfhydryl-containing aliphatic
group that enters the active site and binds the active site zinc and water molecule
(44, 52). Hybrid molecules, including CHAP31 and CHAP50, that possess both
a cyclic peptide and an aliphatic hydroxamate have been prepared and shown to
have a reversible mechanism of action and remarkable inhibitory activity when
optimized in the range of 15 nanamolar (53, 54). The optimal linker in these
studies was found to have five methylenes, similar to that described previously for
other hydroxamates (47).
Inhibitors of the benzamide class, such as CI-994 (55) and MS-275 (56), are
in general less active than members of the hydroxamate or cyclic peptide classes,
with Kis in the micromolar range (44, 56). The mechanism of HDAC inhibition
for benzamides remains uncertain at present. In addition to the structural classes
of HDAC inhibitors described thus far, a variety of inhibitors have been prepared
that are not readily classified into one of the above mentioned five classes. Brosch
and colleagues have recently described 3-(4-Aroyl-1-methyl-1H-2-pyrrolyl)-Nhydroxy-2-alkylamides containing a range of different metal chelating groups with
IC50s in the micromolar range (57, 58). Another series of Psammaplin derivatives
containing novel metal chelating groups have demonstrated considerably greater
inhibitory activity, with the most active of these compounds having nanomolar Kis
(59).
Little is presently known about the potential selectivity of various HDAC class I
or II isoforms for structurally different inhibitors. HDAC6 and HDAC10 both possess two catalytic domains that appear to be differentially inhibited by drugs that
preferentially bind near the entrance of the catalytic site (37, 54, 55). These class
II HDAC isoforms appear relatively resistant to trapoxin when compared to class
I HDACs. Despeptide, MS-275, and several of the hybrid CHAP derivatives also
appear considerably more selective for HDAC1 over HDAC6 (54). TSA is generally considered a nonspecific HDAC inhibitor, as it has a similar Ki for all isoforms
examined. Recently, Schrieber and colleagues described an HDAC6-specific inhibitor, tubacin (Figure 2), responsible for the deacetylation of tubulin, as well as
another histacin, which appears to be a histone-selective deacetylase (60, 61).
The continued development of isoform-specific inhibitors will undoubtedly remain
a major emphasis of HDAC inhibitor development.
7 Jan 2005
14:6
504
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
DRUMMOND ET AL.
(e.g., translocation, amplification, mutation, overexpression) in a subset of hematological and epithelial malignancies, as well as the aberrant genomic recruitment
of otherwise normal HDACs in conjunction with oncogenic transcription factors.
Such observations have led to the conclusion that defects and/or imbalances in the
genomes acetylation machinery accompany changes in local chromatin structure
and oncogenic dysregulation of genes controlling cell cycle progression, differentiation, and apoptosis. Despite these observations and conclusions, there are as yet
no specific HAT or HDAC measurements devised that can predict the sensitivity
of any given tumor to any class of HDAC inhibitor.
It has also been generally accepted that more actively transcribed chromatin
regions are associated with histone hyperacetylation and recruitment of HATs
(although HDACs are also known to be recruited), and histone deacetylation associated with recruitment of HDACs often restores these genomically active regions
to a more repressed and condensed chromatin state. Thus, an attractive paradigm
for the antitumor action of HDAC inhibitors has been the induction of histone
acetylation producing transcriptional activation of critical genes needed for tumor growth arrest (1, 43, 44, 6467). Unquestionably, HDAC inhibitors produce
a global increase in histone acetylation within hours of treatment of many different malignant and nonmalignant tissue types, including those showing little if
any biological consequences upon treatment with HDAC inhibitors. Thus, while a
global increase in the level of histone acetylation by itself cannot explain selective
changes in gene expression or specific patterns of antitumor activity following
HDAC inhibition, assaying for enhanced histone acetylation in readily sampled
cells or tissues (e.g., peripheral white blood cells) is being routinely employed
to demonstrate HDAC inhibitor bioavailability and activity. Greater attention is
currently being given to the expanding list of nonhistone proteins acetylated in direct response to HDAC inhibition (Table 1), especially because many of these are
tissue/development-specific (EKLF, GATA-1, ER, MyoD), oncogenic (c-Myb),
tumor-suppressing (p53), or even rather ubiquitous (TFIIE, TFIIF, TCF, HNF-4)
transcription factors.
Virtually all HDAC inhibitors currently in clinical development show some degree of preclinical activity against malignant cells proliferating in culture and also
tumors growing in animal models; this antitumor activity may be characterized as
either inducing cytostasis (cell cycle arrest), differentiation, or apoptosis. However, the HDAC-dependent mechanisms accounting for the observed and rather
selective modulation of gene expression, as well as specific patterns of antitumor
activity, remain poorly understood. Several studies have now revealed that fewer
than 10% of expressed genes in a given malignant cell population are affected
by an antitumor dose of an HDAC inhibitor, with a near equal number of transcriptionally active genes being repressed as those being stimulated; structurally
different HDAC inhibitors can similarly modulate expression of a relatively limited set of core genes (2, 3, 68). As shown in Table 2, among the commonly
up- and down-modulated gene transcripts identified in these expression microarray studies, as well as in numerous single-gene expression studies (6678), are
7 Jan 2005
14:6
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
505
Reference(s)
(81)
TGF-
(166, 167)
Thioredoxin
(168)
Telomerase
(97)
RECK
(86)
VEGF
Angiogenic factor
(87, 169)
bFGF
Angiogenic factor
(87)
Myb/c-MyBL2
(68)
raf-1
Effector of Ras
(68)
cyclin A
(111)
cyclin B
(111)
DAF
(170)
abl
(68)
DEK
(68)
Proteasome
(68)
Proapoptotic
(76)
Bcl2
Proapoptotic
(78)
p53
Proapoptotic
(169)
Proapoptotic
(171)
c-myc
Inhibitor of differentiation
(100)
Caspase 3
(125, 172)
Carboxypeptidase
A3 (CPA3)
(173)
RECK
(86)
p21WAF1/Cip1
Gelsolin
(66, 70)
(70)
(Continued)
7 Jan 2005
14:6
AR
506
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
DRUMMOND ET AL.
TABLE 2
AR232-PA45-21.tex
(Continued)
Regulated
protein
ER
(174)
TSSC3
(68)
IGFBP-3
(175)
TBP-2
(168)
Reference(s)
Bak, Bcl2 antagonist killer; Bax, Bcl2-associated X protein; DAF, decay-accelerating factor; TBP-2, thioredoxin binding
protein; TSSC3, tumor supressing subtransferable candidate.
7 Jan 2005
14:6
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
507
other HDAC inhibitors for their ability to inhibit ErbB2 promoter function. Of
interest, the rank order of potency for the HDAC inhibitors shown in this screening assay (LAQ824 > TSA > A1616906 > SAHA CI994) is comparable to
their relative antitumor activity against several ErbB2 overexpressing breast cancer cell lines. When evaluated further against SkBr3 and other ErbB2-dependent
breast cancer cell lines (e.g., BT-474, MDA-453), HDAC inhibitors were shown to
inhibit the synthesis and elongation of nascent ErbB2 transcripts as well as destabilize and accelerate the decay of mature cytoplasmic ErbB2 transcripts (Figure 4,
panels B and C). Although ongoing preclinical studies are confirming that ErbB2dependent cancers appear somewhat more sensitive to HDAC inhibitors than
ErbB2-independent cancers, molecular studies are attempting to define the drugsensitive HDAC-dependent nuclear and cytoplasmic mechanisms that differentially regulate ErbB2 transcription and ErbB2 transcript stability, respectively. The
presence of multiple distinct HDAC-dependent mechanisms capable of controlling ErbB2 transcript levels suggests that even among ErbB2-dependent cancers,
there will be differential sensitivity to structurally different classes of HDAC inhibitors. Other investigators have identified HDAC-dependent posttranslational
mechanisms that can also downregulate the expression of oncoprotein kinases like
ErbB2 and bcr/abl (80, 82). Acetylation of the chaperone protein, Hsp90, induced
by HDAC inhibition, results in the enhanced proteasomal degradation of ErbB2
and bcr/abl kinases. These examples of multiple mechanisms by which HDAC
inhibitors potentially downregulate critical oncogenic pathways also suggest new
combinatorial strategies for possible clinical evaluation, including HDAC inhibitor
treatment in conjunction with tyrosine kinase inhibitors (80, 8284) or Hsp90 antagonists (85).
Apart from directly affecting transformed cells, HDAC inhibitors have also been
shown to inhibit tumor angiogenesis, suggesting additional therapeutic mechanisms for the observed in vivo activity of these antitumor drugs (76, 8688).
Depsipeptide was shown to suppress the expression of pro-angiogenic factors, including vascular endothelial growth factor (VEGF) and basic fibroblast growth
factor (bFGF) (87). VEGF and bFGF mRNA levels were significantly reduced
in prostate tumor xenografts sensitive to this cyclic peptide HDAC inhibitor. The
hydroxamic acid HDAC inhibitor TSA was shown to upregulate the RECK protein responsible in part for inhibiting tumor metastasis and angiogenesis through
its action on matrix metalloproteases (86). The carboxylate and short-chain fatty
acid HDAC inhibitor, valproic acid, was also shown to inhibit angiogensis both
in vitro and in vivo via a mechanism involving diminished expression of endothelial
nitric oxide synthase (88). Additional miscellaneous or less well-studied tumorassociated mechanisms may prove to be important in determining the ultimate
clinical utility of some HDAC inhibitors.
Last, various drug-resistance phenotypes have been shown to be modulated
by HDAC inhibitors (8995). Treatment of different multidrug-resistant cell lines
with TSA or SAHA was shown to downregulate P-glycoprotein (93), helping
7 Jan 2005
14:6
508
AR
AR232-PA45-21.tex
DRUMMOND ET AL.
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
7 Jan 2005
14:6
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
509
Figure 4 Transcriptional repression of ErbB2 induced by hydroxamic HDAC inhibitors is caused by a combination of both ErbB2 promoter repression and transcript
destabilization. (A) Employing our previously described high-throughput screening assay (81), an ErbB2-independent subline of MCF-7 breast cancer cells (MCF/R06pGL4) bearing a chromatin-integrated ErbB2 promoter-driven luciferase construct was
used to compare the ErbB2 promoter repressing potency of four structurally different hydroxamic acidtype HDAC inhibitors (SAHA, A1616906, TSA, LAQ824) and
a benzamide-type HDAC inhibitor (CI994). After 24 h culture exposure to the indicated drug doses, cell viability as measured by MTT assay (squares) shows little
change, whereas specific repression of ErbB2 promoter activity is detected by luciferase expression (diamonds). The benzamide inhibitor (CI994) shows slight ErbB2
promoter stimulation with no evidence of promoter repression; in contrast, the hydroxamic acid inhibitors show ErbB2 promoter repression at different potencies as
indicated by the 25% luciferase inhibitory concentration (M IC25) values. (B) When
ErbB2-dependent SkBr3 breast cancer cells in culture are treated for 5 h with an
ErbB2 promoterrepressing dose of TSA, nascent ErbB2 transcript synthesis and
elongation, as measured by nuclear run-off assays (81), appears completely inhibited, whereas nascent transcript synthesis of the Ets transcription factor ESX appears
marginally increased. (C) Total RNA extracted and Northern blotted after 5-h treatment of cultured SkBr3 breast cancer cells shows treatment effects on mature longlived (8 h half-life) ErbB2 transcripts (4.8 kb) in comparison to short-lived (<2 h
half-life) ESX transcripts (2.2 kb). Treatment for 5 h with an RNA polymerase inhibiting dose of Actinomycin D (10 g/ml) demonstrates the expected absence of
ESX transcripts and partial decline in total ErbB2 transcripts. In contrast, and after 5-h treatment with comparable doses of the HDAC inhibitors TSA, CI994, and
LAQ824, ESX levels appear marginally increased, whereas ErbB2 transcript levels
are reduced below levels caused by Act D treatment, demonstrating the independent
ability of HDAC inhibitors to destabilize and accelerate the decay of mature ErbB2
transcripts.
7 Jan 2005
14:6
510
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
DRUMMOND ET AL.
7 Jan 2005
14:6
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
511
7 Jan 2005
14:6
512
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
DRUMMOND ET AL.
in several murine tumor and human tumor xenograft models (108), and has shown
some encouraging results in early clinical trials (109).
The benzamide HDAC inhibitors, MS-275 (56, 71, 110, 111) and CI-994 (55),
have also shown in vivo activity against various tumor models. MS-275 was shown
to inhibit the growth of three different orthotopic pediatric tumor xenografts (110).
In another study, MS-275 administered orally was shown to have potent antitumor
activity against a series of seven different human tumors xenografts (71). Unfortunately, antitumor doses of MS-275 in mice were also myelosuppressive, causing
decreases in red and white blood cells as well as platelets (111). A similar pattern
of toxicity was observed with CI-994 (112); and thrombocytopenia was a major
dose-limiting toxicity seen in Phase I and II clinical testing with CI-994 as well as
with the cyclic peptide depsipeptide (113, 114).
7 Jan 2005
14:6
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
513
Pharmacokinetic Considerations
Although receiving less reported attention than studies elucidating the pharmacodynamics of various HDAC inhibitors, the pharmacokinetic characteristics and
limitations of different HDAC inhibitors are of critical interest and will likely
prove to be an important determinant of their ultimate clinical utility. Inhibition of
intracellular HDAC activity commonly requires continuous systemic circulation
and drug exposure to achieve maximal tumor cytostasis or apoptosis and clinical
response. Rapid clearance, a high degree of protein binding, rapid metabolism,
or rapid inactivation of reactive functional groups (i.e., epoxy groups) are factors
that can adversely affect HDAC inhibitor bioavailability and antitumor activity.
Although occasionally used in the clinic, prolonged or daily infusions of any
drug are generally undesirable. The requirement of constant systemic exposure
by parenteral administration to achieve an active antitumor drug concentration
will most likely limit the clinical development of any HDAC inhibitor that is not
orally bioavailable. For this reason, the clinical development of SAHA shifted
from Phase I evaluation of daily intravenous infusions to a more recently designed
oral formulation (115).
Novel drug delivery systems that allow for controlled drug release may help circumvent the clinical inconvenience of daily infusions as well as generally enhance
the therapeutic index of HDAC inhibitors. We have recently evaluated the potential
of liposomes for delivering the HDAC inhibitor LAQ824. When formulated properly, liposomes can entrap and concentrate amphipathic drugs (achieving >10,000
drug molecules per liposomal nanoparticle), releasing them slowly over time in
the plasma or delivering them specifically to solid tumors where they deposit their
drug in close proximity to the tumor, allowing for increased tumor accumulation
and drug exposure (118). In a pilot study administering liposomal LAQ824 on a
once-weekly schedule for three weeks, we observed significant growth arrest of
rapidly growing human breast tumor xenografts (Figure 5B). As shown in other
studies involving various tumor model systems, free LAQ824 requires daily injections of generally higher doses to slow tumor growth (Figure 5A). More recent
studies with optimized formulations and targeted liposomal constructs have shown
even greater efficacy (119).
7 Jan 2005
14:6
514
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
DRUMMOND ET AL.
more extensively in a separate review of the field (65), and are summarized in
Table 3. The pretreatment or coadministration of HDAC inhibitors with a wide
range of agents has repeatedly been shown to additively or synergistically enhance
apoptosis of cancer cells in culture (68, 82, 85, 120125) as well as antitumor
efficacy in vivo (126128). Notably, enhancements in activity have been observed
when HDAC inhibitors are combined with a number of different commonly used
chemotherapeutics (82, 120, 121). Nuclear receptor ligands (123, 127, 129, 130),
Hsp90 antagonists (85), proteasome inhibitors (68, 84, 131), signal transduction
inhibitors (80, 82, 124, 125, 132136), and DNA demethylating agents (72, 73, 122,
128, 137) represent some of the more promising classes of agents. Demethylating
agents such as 5-aza-2 -deoxycytidine (DAC) are particularly interesting owing to
the interaction of DNA methylation with histone deacetylation in gene silencing of
tumor suppressor genes, as mentioned above. Combinations of DAC with TSA or
depsipeptide were shown to reactivate silenced tumor suppressor genes including
MLH1, TIMP3, CDKN2B, CDKN2A, ARHI, gelsolin, and maspin (72, 73, 137),
synergistically increasing the level of tumor cell apoptosis (122). Combinations
of nuclear receptor ligands, such as all-trans retinoic acid (ATRA), or vitamin
D analogs, such as 1,25-dihydroxyvitamin D, with HDAC inhibitors have been
shown to increase differentiation and apoptosis in cancer cells (123, 127, 130) and
also inhibit tumor growth in vivo (127, 130, 138). Small-molecule kinase inhibitors
may also be rationally combined with HDAC inhibitors. Imatinib (Gleevec) is
a specific inhibitor of Bcr/Abl with impressive clinical activity in the treatment
of chronic myeloid leukemia and selected other malignancies. Because LAQ824
has been shown to downregulate the expression of Bcr/Abl and also promote its
degradation through acetylation of Hsp90 (80), combinations of imatinib with
LAQ824 as well as other HDAC inhibitors, such as SAHA and apicidin, have
been tested and shown to dramatically increase the apoptosis of Bcr/Abl positive
leukemic cells (80, 83, 132, 133). A similar effect is seen when malignant cells
known to be transformed by oncogenic tyrosine kinases (ErbB2/HER2, Src/Abl,
PI3 kinase) are treated with HDAC inhibitors in combination with appropriate
kinase inhibitors like Herceptin, PD180970, or LY294002 (80, 82, 124).
As noted above, expression of the CDK inhibitor p21WAF1/CIP1 is regulated by
HDACs and plays a critical role in determining whether cells undergo differentiation or apoptosis in response to treatment with HDAC inhibitors (66, 70, 139).
Flavopiridol is a CDK inhibitor that results in a disruption of p21WAF1/CIP1 induction and induces apoptosis. Its combination with HDAC inhibitors (SAHA,
depsipetpide, sodium butyrate) has been shown to result in a disruption of p21
induction and an additive or synergistic increase in tumor cell apoptosis (125, 135,
139, 140).
Proteasome inhibitors and Hsp90 antagonists represent two other groups of
interesting agents that may be rationally combined with HDAC inhibitors. Hsp90
is a molecular chaperone that stabilizes and controls the intracellular trafficking of
important client proteins, including ErbB2/HER2, Bcr/Abl, EGF, cyclin D1, c-Raf,
and steroid receptors. The inhibition of Hsp90 with amsacrine antagonists, such as
7 Jan 2005
14:6
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
515
TABLE 3 Therapeutic agents used in combination with HDAC inhibitors in preclinical and
clinical studies
Combined
therapeutic agent
HDAC inhibitor
Standard chemotherapy
Gemcitabine
CI-994
VP-16 (etoposide)
cytarabine, etoposide, and
topotecan
Doxorubicin, melphalan,
chloroambucil, cisplatin,
carboplatin, fludarabine
Taxotere, gemcitabine,
epothilone B
Etoposide
Fludarabine
IMID-1, dexamethasone
Demethylating agents
DAC
TRAIL
Combined
antitumor effects
Reference(s)
(176)
TSA, SAHA
PB
Phase II trial
increased toxicity
Synergistic
Synergistic
PB
Additive
(121)
LAQ824
Additive
(82)
TSA
MS-275
SAHA
Antagonistic
Synergistic
Synergistic
(177)
(178)
(68)
TSA, depsipeptide
Depsipeptide
PB
PB
Enhanced
Synergistic
Synergistic
Enhanced
(122)
(73)
(128)
(179)
CBHA
TSA
SB
Synergistic
Synergistic
(127)
(123)
Apicidin
SAHA
LAQ824
Synergistic
Synergistic
Synergistic
(83)
(132)
(80)
LAQ824
SAHA, SB, MS-275
SAHA
SB
Depsipeptide
SB
SB, SAHA
LAQ824
Synergistic
Synergistic
Synergistic
Synergistic
Synergistic
Enhanced
Synergistic
Enhanced
(82)
(124)
(125)
(135)
(136, 140)
(180)
(181)
(182)
Synergistic
Synergistic
Synergistic
Synergistic
Synergistic
(85)
(68)
(84)
(141)
(131)
(120)
(121)
ATRA, all-trans retinoic acid; CBHA, m-carboxylcinnamic acid bis-hydroxamide; DAC; 5-aza-2 -deoxycytidine; IMID-1,
immunomodulatory thalidomide derivative 1; TRAIL, tumor necrosis factor-related apoptosis-inducing ligand.
7 Jan 2005
14:6
516
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
DRUMMOND ET AL.
17-AAG, results in proteasomal degradation of client proteins. Owing to the regulation of Hsp90 function by acetylation, the combination of HDAC inhibitors and
Hsp90 antagonists is a reasonable therapeutic strategy. Experimentally, 17-AAG
in combination with SAHA or sodium butyrate inhibited induction of p21WAF1/CIP1,
inducing Bcl-2 cleavage, and synergistically enhanced tumor cell apoptosis (85).
Proteasome inhibitors slow the degradation of many important and diverse cellular
proteins, and their combination with HDAC inhibitors results in more complete
inhibition of proteasome activity, which may synergistically enhance tumor cell
apoptosis (68, 84, 131, 141).
There is additional evidence that HDAC inhibitors may improve the efficacy
of radiation therapy (142). In this study, pretreatment with depsipeptide greatly
increased radiation-induced apoptosis. In another study, the HDAC inhibitors
phenylbutyrate, TSA, and valproic acid were able to reduce cutaneous radiation
toxicity following radiotherapy (143). This poorly understood interaction whereby
HDAC inhibitors potentially increase radiation-induced tumor cell death while decreasing normal host cell toxicity deserves further study, as it may lead to more
novel clinical indications for HDAC inhibitors.
Although these provocative combination regimens based on cell culture studies
have rational appeal, they must be explored more fully in vivo to assure that
they do not also lead to enhanced host toxicity. One recent Phase II study of the
combination of gemcitabine and CI-994 in patients with NSCLC demonstrated no
improvement in efficacy over gemcitabine alone, primarily because of increased
toxicity that limited dose intensity and reduced the net therapeutic index of the
two-drug combination (176).
CONCLUSIONS
The complexities of the histone code and the various other nuclear as well as
cytoplasmic nonhistone proteins whose functions are modulated by acetylation
underscore why HDAC inhibition was an empirically discovered, as well as novel,
form of cancer therapy. The biology of the various HDAC isoforms and their relationship to tumorigenesis is just beginning to be elucidated and is largely driven by
the perceived clinical potential of HDAC inhibitors. It remains to be seen if a more
detailed understanding of the specific roles played by various HDAC isoforms
during human tumorigenesis leads not only to development of isoform-specific
inhibitors but also to more effective or less toxic antitumor therapeutics, as compared to the multiclass HDAC inhibitors that are currently undergoing clinical
evaluation. Rationally designed combinations of HDAC inhibitors with various
other types of approved or investigational anticancer agents are showing promise
in tumor cell culture systems but must yet be proven in clinical trials. Of great interest to many cancer investigators is the potential ability to derepress the expression
of epigenetically silenced tumor suppressor genes by administering HDAC inhibitors in combination with inhibitors of DNA methyltransferases. There is a
7 Jan 2005
14:6
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
517
LITERATURE CITED
1. Cress WD, Seto E. 2000. Histone deacetylases, transcriptional control, and cancer.
J. Cell Physiol. 184:116
2. Glaser KB, Staver MJ, Waring JF, Stender
J, Ulrich RG, Davidsen SK. 2003. Gene
expression profiling of multiple histone
deacetylase (HDAC) inhibitors: defining
a common gene set produced by HDAC
inhibition in T24 and MDA carcinoma cell
lines. Mol. Cancer Ther. 2:15163
7 Jan 2005
14:6
518
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
DRUMMOND ET AL.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
7 Jan 2005
14:6
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
519
7 Jan 2005
14:6
520
51.
52.
53.
54.
55.
56.
57.
58.
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
DRUMMOND ET AL.
ketone functionalized cyclic tetrapeptides. J. Med. Chem. 30:7178
Gore SD, Carducci MA. 2000. Modifying histones to tame cancer: clinical
development of sodium phenylbutyrate
and other histone deacetylase inhibitors.
Expert Opin. Invest. Drugs 9:292334
Furumai R, Matsuyama A, Kobashi N,
Lee KH, Nishiyama M, et al. 2002. FK228
(depsipeptide) as a natural prodrug that inhibits class I histone deacetylases. Cancer
Res. 62:491621
Komatsu Y, Tomizaki KY, Tsukamoto M,
Kato T, Nishino N, et al. 2001. Cyclic
hydroxamic-acid-containing peptide 31, a
potent synthetic histone deacetylase inhibitor with antitumor activity. Cancer
Res. 61:445966
Furumai R, Komatsu Y, Nishino N,
Khochbin S, Yoshida M, Horinouchi
S. 2001. Potent histone deacetylase inhibitors built from trichostatin A and
cyclic tetrapeptide antibiotics including
trapoxin. Proc. Natl. Acad. Sci. USA 98:
8792
el-Beltagi HM, Martens AC, Lelieveld P,
Haroun EA, Hagenbeek A. 1993. Acetyldinaline: a new oral cytostatic drug with
impressive differential activity against
leukemic cells and normal stem cells
preclinical studies in a relevant rat model
for human acute myelocytic leukemia.
Cancer Res. 53:300814
Suzuki T, Ando T, Tsuchiya K, Fukazawa
N, Saito A, et al. 1999. Synthesis and histone deacetylase inhibitory activity of new
benzamide derivatives. J. Med. Chem. 42:
30013
Mai A, Massa S, Ragno R, Esposito M,
Sbardella G, et al. 2002. Binding mode
analysis of 3-(4-benzoyl-1-methyl-1H2-pyrrolyl)-N-hydroxy-2-propenamide: a
new synthetic histone deacetylase inhibitor inducing histone hyperacetylation,
growth inhibition, and terminal cell differentiation. J. Med. Chem. 45:177884
Ragno R, Mai A, Massa S, Cerbara I, Valente S, et al. 2004. 3-(4-Aroyl-1-methyl-
59.
60.
61.
62.
63.
64.
65.
66.
67.
7 Jan 2005
14:6
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
77.
78.
79.
80.
81.
82.
83.
521
tosis and Fas/Fas ligand expression in human acute promyelocytic leukemia cells.
J. Biol. Chem. 277:207380
Cao XX, Mohuiddin I, Ece F, McConkey
DJ, Smythe WR. 2001. Histone deacetylase inhibitor downregulation of bcl-xl
gene expression leads to apoptotic cell
death in mesothelioma. Am. J. Respir. Cell
Mol. Biol. 25:56268
Sawa H, Murakami H, Ohshima Y, Sugino T, Nakajyo T, et al. 2001. Histone
deacetylase inhibitors such as sodium butyrate and trichostatin A induce apoptosis through an increase of the bcl-2related protein Bad. Brain Tumor. Pathol.
18:10914
Slack JL. 1999. The biology and treatment
of acute progranulocytic leukemia. Curr.
Opin. Oncol. 11:913
Nimmanapalli R, Fuino L, Bali P, Gasparetto M, Glozak M, et al. 2003. Histone deacetylase inhibitor LAQ824 both
lowers expression and promotes proteasomal degradation of Bcr-Abl and
induces apoptosis of imatinib mesylatesensitive or -refractory chronic myelogenous leukemia-blast crisis cells. Cancer
Res. 63:512635
Scott GK, Marden C, Xu F, Kirk L,
Benz CC. 2002. Transcriptional repression of ErbB2 by histone deacetylase inhibitors detected by a genomically integrated ErbB2 promoter-reporting cell
screen. Mol. Cancer Ther. 1:38592
Fuino L, Bali P, Wittmann S, Donapaty
S, Guo F, et al. 2003. Histone deacetylase
inhibitor LAQ824 down-regulates Her-2
and sensitizes human breast cancer cells to
trastuzumab, taxotere, gemcitabine, and
epothilone B. Mol. Cancer Ther. 2:971
84
Kim JS, Jeung HK, Cheong JW, Maeng H,
Lee ST, et al. 2004. Apicidin potentiates
the imatinib-induced apoptosis of BcrAbl-positive human leukaemia cells by
enhancing the activation of mitochondriadependent caspase cascades. Br. J.
Haematol. 124:16678
7 Jan 2005
14:6
522
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
DRUMMOND ET AL.
92.
93.
94.
95.
96.
97.
98.
99.
7 Jan 2005
14:6
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
100.
101.
102.
103.
104.
105.
106.
107.
108.
109.
110.
111.
112.
113.
114.
523
7 Jan 2005
14:6
524
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
DRUMMOND ET AL.
115. Kelly WK, Richon VM, O Connor O, Curley T, MacGregor-Curtelli B, et al. 2003.
Phase I clinical trial of histone deacetylase inhibitor: suberoylanilide hydroxamic acid administered intravenously.
Clin. Cancer Res. 9:357888
116. Piekarz RL, Robey R, Sandor V, Bakke
S, Wilson WH, et al. 2001. Inhibitor
of histone deacetylation, depsipeptide
(FR901228), in the treatment of peripheral and cutaneous T-cell lymphoma: a
case report. Blood 98:286568
117. Atadja P, Gao L, Kwon P, Trogani
N, Walker H, et al. 2004. Selective
growth inhibition of tumor cells by a
novel histone deacetylase inhibitor, NVPLAQ824. Cancer Res. 64:68995
118. Drummond DC, Meyer OM, Hong K,
Kirpotin DB, Papahadjopoulos D. 1999.
Optimizing liposomes for delivery of
chemotherapeutic agents to solid tumors.
Pharmacol. Rev. 51:691743
119. Benz C, Scott G, Berger C, Amend C,
Guo Z, et al. 2003. Liposome encapsulation of histone deacetylase inhibitor enhances in vivo activity against ErbB2positive breast cancers. Clin. Cancer Res.
9:6259s
120. Kim MS, Blake M, Baek JH, Kohlhagen
G, Pommier Y, Carrier F. 2003. Inhibition
of histone deacetylase increases cytotoxicity to anticancer drugs targeting DNA.
Cancer Res. 63:7291300
121. Witzig TE, Timm M, Stenson M, Svingen PA, Kaufmann SH. 2000. Induction of
apoptosis in malignant B cells by phenylbutyrate or phenylacetate in combination
with chemotherapeutic agents. Clin. Cancer Res. 6:68192
122. Zhu WG, Lakshmanan RR, Beal MD,
Otterson GA. 2001. DNA methyltransferase inhibition enhances apoptosis induced by histone deacetylase inhibitors.
Cancer Res. 61:132733
123. Rashid SF, Moore JS, Walker E, Driver
PM, Engel J, et al. 2001. Synergistic
growth inhibition of prostate cancer cells
by 1 alpha,25 dihydroxyvitamin D(3) and
124.
125.
126.
127.
128.
129.
130.
its 19-nor-hexafluoride analogs in combination with either sodium butyrate or trichostatin A. Oncogene 20:186072
Rahmani M, Yu C, Reese E, Ahmed
W, Hirsch K, et al. 2003. Inhibition of
PI-3 kinase sensitizes human leukemic
cells to histone deacetylase inhibitormediated apoptosis through p44/42 MAP
kinase inactivation and abrogation of
p21(CIP1/WAF1) induction rather than
AKT inhibition. Oncogene 22:623142
Almenara J, Rosato R, Grant S. 2002. Synergistic induction of mitochondrial damage and apoptosis in human leukemia cells
by flavopiridol and the histone deacetylase inhibitor suberoylanilide hydroxamic
acid (SAHA). Leukemia 16:133143
Keen JC, Yan L, Mack KM, Pettit C,
Smith D, et al. 2003. A novel histone
deacetylase inhibitor, scriptaid, enhances
expression of functional estrogen receptor
alpha (ER) in ER negative human breast
cancer cells in combination with 5-aza 2 deoxycytidine. Breast Cancer Res. Treat.
81:17786
Coffey DC, Kutko MC, Glick RD, Butler LM, Heller G, et al. 2001. The histone deacetylase inhibitor, CBHA, inhibits growth of human neuroblastoma
xenografts in vivo, alone and synergistically with all-trans retinoic acid. Cancer
Res. 61:359194
Belinsky SA, Klinge DM, Stidley CA,
Issa JP, Herman JG, et al. 2003. Inhibition
of DNA methylation and histone deacetylation prevents murine lung cancer. Cancer Res. 63:708993
Kosugi H, Towatari M, Hatano S, Kitamura K, Kiyoi H, et al. 1999. Histone
deacetylase inhibitors are the potent inducer/enhancer of differentiation in acute
myeloid leukemia: a new approach to antileukemia therapy. Leukemia 13:131624
Zhou DC, Kim SH, Ding W, Schultz C,
Warrell RP Jr, Gallagher RE. 2002. Frequent mutations in the ligand-binding domain of PML-RARalpha after multiple relapses of acute promyelocytic leukemia:
7 Jan 2005
14:6
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
131.
132.
133.
134.
135.
136.
137.
analysis for functional relationship to response to all-trans retinoic acid and histone deacetylase inhibitors in vitro and
in vivo. Blood 99:135663
Giuliano M, Lauricella M, Calvaruso G,
Carabillo M, Emanuele S, et al. 1999. The
apoptotic effects and synergistic interaction of sodium butyrate and MG132 in
human retinoblastoma Y79 cells. Cancer
Res. 59:558695
Yu C, Rahmani M, Almenara J, Subler M, Krystal G, et al. 2003. Histone
deacetylase inhibitors promote STI571mediated apoptosis in STI571-sensitive
and -resistant Bcr/Abl+ human myeloid
leukemia cells. Cancer Res. 63:211826
Nimmanapalli R, Fuino L, Stobaugh C,
Richon V, Bhalla K. 2003. Cotreatment
with the histone deacetylase inhibitor
suberoylanilide hydroxamic acid (SAHA)
enhances imatinib-induced apoptosis of
Bcr-Abl-positive human acute leukemia
cells. Blood 101:323639
Rosato RR, Almenara JA, Yu C, Grant
S. 2004. Evidence of a functional role
for p21WAF1/CIP1 down-regulation in
synergistic antileukemic interactions between the histone deacetylase inhibitor
sodium butyrate and flavopiridol. Mol.
Pharmacol. 65:57181
Rosato RR, Almenara JA, Cartee L,
Betts V, Chellappan SP, Grant S. 2002.
The cyclin-dependent kinase inhibitor
flavopiridol disrupts sodium butyrateinduced p21WAF1/CIP1 expression and
maturation while reciprocally potentiating apoptosis in human leukemia cells.
Mol. Cancer Ther. 1:25366
Nguyen DM, Schrump WD, Tsai WS,
Chen A, Stewart JHt, et al. 2003.
Enhancement of depsipeptide-mediated
apoptosis of lung or esophageal cancer
cells by flavopiridol: activation of the
mitochondria-dependent death-signaling
pathway. J. Thorac. Cardiovasc. Surg.
125:113242
Fujii S, Luo RZ, Yuan J, Kadota M, Oshimura M, et al. 2003. Reactivation of the
138.
139.
140.
141.
142.
143.
144.
525
7 Jan 2005
14:6
526
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
DRUMMOND ET AL.
7 Jan 2005
14:6
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
LaTeX2e(2002/01/18)
P1: GCE
166.
167.
168.
169.
170.
171.
172.
527
7 Jan 2005
14:6
528
AR
AR232-PA45-21.tex
AR232-PA45-21.sgm
P1: GCE
DRUMMOND ET AL.
LaTeX2e(2002/01/18)
1/7/05
2:29 PM
HI-RES-PA45-21-Drummond.qxd
Page 1
C-1
HI-RES-PA45-21-Drummond.qxd
C-2
1/7/05
2:29 PM
Page 2
DRUMMOND ET AL.
Figure 1 Histones can be modified in a variety of ways, primarily in the tails of core histones in a process that is referred to as the histone code. Acetylated lysine residues, methylated arginines, methylated lysines, phosphorylated serines, sumoylated lysines, and ubquitinated lysine residues all contribute to the histone code. The relative positions for each
modification on the various histone tails are depicted by symbols that are defined in the key.
The actual chemical modification of the various amino acids in the histone tails is shown with
the colored bonds indicating the modification and the amino acid residue shown in black.
This figure was adapted from Turner (4) and Spotswood & Turner (144) with permission.
4 Dec 2004
18:50
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
10.1146/annurev.pharmtox.45.120403.100120
INTRODUCTION
Humankinds battle with tuberculosis (TB) dates back to antiquity. TB, which
is caused by Mycobacterium tuberculosis, was a much more prevalent disease
in the past than it is today, and it was responsible for the deaths of about one
billion people during the last two centuries (1). Improved sanitation and living
conditions significantly reduced the incidence of the disease even before the advent
of chemotherapy. The introduction of TB chemotherapy in the 1950s, along with
the widespread use of BCG vaccine, had a great impact on further reduction in TB
incidence. However, despite these advances, TB still remains a leading infectious
disease worldwide, especially in the third world countries.
M. tuberculosis is a particularly successful pathogen that latently infects about
2 billion people, about one third of world population (2). Each year, there are about
8 million new TB cases and 2 million deaths worldwide. TB is on the increase
in recent years, largely owing to HIV infection, immigration, increased trade,
and globalization (2). The increasing emergence of drug-resistant TB, especially
0362-1642/05/0210-0529$14.00
529
4 Dec 2004
18:50
530
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
ZHANG
4 Dec 2004
18:50
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
Figure 1
531
drugs for the treatment of TB and subsequently led to synthesis of thiosemicarbazones such as Conteben (also called amithiazone), which were more active than
sulfanilamide and had definite clinical value but were not as effective as streptomycin (19). In 1946, two years after the discovery of streptomycin, Lehmann
from Sweden discovered para-aminosalicylic acid (PAS) (Figure 1) as an effective
TB drug (20), a discovery based on a curious observation made by Bernheim that
salicylate and benzoate stimulated the oxygen consumption of tubercle bacillus
(21). This was quickly followed in 1952 by the sensational discovery of the highly
active TB drug isoniazid (INH) (Figure 1) simultaneously by three drug companies: Hoffman LaRoche, E.R. Squibb & Sons, and Bayer. The discovery of INH
was based on the nicotinamide activity against tubercle bacilli in the animal model
observed by Chorine in 1945 (22) and the reshuffling of chemical groups in the
thiosemicarbazone (2325). INH represented a major milestone in the chemotherapy of TB because it is highly active, inexpensive, and without significant side
effects (26). Remarkably, the nicotinamide lead also led to the discovery of pyrazinamide (PZA) (Figure 1) in 1952 by the Lederle Research Laboratories (27)
and ethionamide (ETH)/Prothionamide (PTH) (Figure 1) in 1956 (28). Ethambutol (EMB) was discovered in 1961 at Lederle on the basis of the observation that
polyamines and diamines had activity against tubercle bacilli; subsequent synthesis
of diamine analogs led to the identification of EMB (29). Further screening for antibiotics from soil microbes led to discovery of many other antituberculosis drugs:
cycloserine (30); kanamycin (31) and its derivative amikacin; viomycin (32); capreomycin (33); and rifamycins (34) and its derivative rifampin (RIF), developed at
4 Dec 2004
18:50
532
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
ZHANG
Dow-Lepetit Research Laboratories, Italy (35), which has been the drug of choice
for treatment of TB since the 1970s. The 1950s and 60s represent a golden era
of TB drug discovery. Most of the TB drugs in use today were discovered during
this period, except the broad-spectrum quinolone drugs, which were developed in
1980s on the basis of the antibacterial activity of nalidixic acid discovered in the
1960s (36). Although quinolone drugs were not initially used for TB treatment,
they were subsequently shown to have high activity against tubercle bacillus and
were used as second-line drugs for the treatment of drug-resistant TB since the
late 1980s (37, 38).
4 Dec 2004
18:50
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
533
(c) dormant bacteria. Although all three types of phenotypic resistance may share
some common mechanism, the mechanism of phenotypic resistance in M. tuberculosis is unknown. There is currently considerable interest in the study of
mycobacterial persistence and dormancy (4143), with the aim to better understand the basis of this phenomenon and devise therapeutic strategies that target
the persistent or dormant organisms for improved treatment of TB. Current TB
drugs are mainly active against growing bacilli, and except for RIF and PZA, they
are not good at killing persisters. Although RIF and PZA are important sterilizing
drugs that significantly reduce the number of bacilli in lesions and play an important role in shortening the therapy from 1218 months to 6 months, there are still
other persister populations that are not killed by RIF or PZA. TB is like a little
universe or a Russian Doll, consisting of layer after layer of different bacterial
populations within a large bacterial population. At this time, we have little knowledge about the biology of these persisters, despite significant interest in this area
(4143). The intracellular location of the bacilli could render some drugs such as
streptomycin inactive. However, most drugs do penetrate the necrotic tissues (40),
although they cannot effectively kill nonreplicating bacilli in the lesions. Third,
host immune system may not effectively eliminate tubercle bacilli in the lesions.
In many bacterial infections, small numbers of residual bacteria that remain after
antibiotic therapy can be effectively mopped up by the immune system. However,
in TB, it appears that the host immune system is not very effective in controlling
the residual bacteria not killed by TB chemotherapy. Thus, although achieving
a clinical cure, the current chemotherapy cannot achieve a bacteriological cure,
i.e., the therapy cannot completely eradicate all bacilli in the lesion (40). This
depressing fact underscores the need for developing better sterilizing drugs and
other interventions, such as improving host immune status, as adjunct treatment
for more effective therapy.
The varying types of lesions determine different metabolic status of tubercle
bacilli in vivo and are the basis for diverse bacterial populations. According to
Mitchison (44), tubercle bacilli in lesions consist of at least four different subpopulations: (a) those that are actively growing, which are killed primarily by INH
[but in case of INH resistance, are killed by RIF, SM (streptomycin), or inhibited by EMB]; (b) those that have spurts of metabolism, which are killed by RIF;
(c) those that are of low metabolic activity and reside in acid pH environment,
which are killed by PZA; and (d) those that are dormant, which are not killed
by any current TB drug. A modified version of the Mitchison hypothesis is shown
in Figure 2, where the speed of growth in the original Mitchison hypothesis is
replaced with metabolic status.
4 Dec 2004
18:50
534
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
ZHANG
Figure 2
Special bacterial populations and TB
chemotherapy.
between static and cidal drugs is only relative, because some static drugs can
be cidal under some conditions (such as with higher drug concentrations, smaller
inoculum, or change in bacterial physiological status). For example, PZA can show
cidal activity against small numbers of nongrowing bacilli at acid pH but primarily
shows static activity for growing bacilli with active metabolism (45). Cidal drugs
exhibit higher activity over static drugs in reducing the number of bacilli in the
lesions. The current TB drugs can also be categorized as either first-line drugs or
second-line drugs. The first-line drugs include INH, RIF, PZA, EMB, and SM; the
second-line drugs include kanamycin, amikacin, capreomycin, cycloserine (CS),
PAS, ETH/PTH, thiacetazone, and FQ. According to their specificity, TB drugs can
also be grouped as TB or mycobacteria-specific drugs such as INH, PZA, EMB,
PAS, ETH, and thiacetazone, and the broad-spectrum antibiotics such as RIF, SM,
kanamycin, amikacin, capreomycin, CS, and FQ. The mechanisms of action and
resistance to TB-specific drugs are specific to M. tuberculosis, whereas mechanisms of action and resistance of the broad-spectrum drugs in M. tuberculosis are
the same as in other bacterial species. The chemical structures and the targets of inhibition for the first-line and second-line TB drugs are shown in Figure 1 and Table
1, respectively. The mechanisms of action and resistance of TB drugs have been
reviewed recently (46, 47). For the purpose of comparison with new drug targets,
the mechanisms of action and resistance of the current TB drugs will be briefly reviewed here. These drugs can be grouped as cell wall synthesis inhibitors, nucleic
acid synthesis inhibitors, protein synthesis inhibitors, and energy inhibitors.
INH
0.010.2
0.050.5
20100
pH 5.5 or 6.0
15
28
18
0.24
0.62.5
18
520
Isoniazid (1952)
Rifampin (1966)
Pyrazinamide (1952)
Ethambutol (1961)
Streptomycin (1944)
Kanamycin (1957)
Quinolones (1963)
Ethionamide (1956)
PAS (1946)
Cycloserine (1952)
Inhibition of peptidoglycan
synthesis
D-alanine
racemasec
alrA, Ddlc
Unknown
inhA
etaA/ethAb
gyrA
gyrB
rrs
rpsL, rrs
DNA gyrase
16S rRNA
embCAB
TUBERCULOSIS DRUG TARGETS
Bacteriostatic
Arabinosyl transferase
pncAb
rpoB
katGb
inhA
ndh
XMLPublishSM (2004/02/24)
Bacteriostatic
Bacteriostatic
Bactericidal
Bactericidal
Bactericidal
Targets
AR232-PA45-22.tex
Bacteriostatic
Bacteriostatic/
bactericidal
Bactericidal
Bactericidal
Mechanisms of action
Genes
involved in
resistance
AR
MICa (g/ml)
Effect on
bacterial cell
18:50
Drug (year of
discovery)
4 Dec 2004
P1: JRX
535
4 Dec 2004
18:50
536
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
ZHANG
organic radicals, which then attack multiple targets in the tubercle bacillus. The
primary target of inhibition is the cell wall mycolic acid synthesis pathway (49),
where enoyl ACP reductase (InhA) was identified as the target of INH inhibition
(50). The active species for InhA inhibition has been found to be isonicotinic
acyl radical, which reacts with NAD to form INH-NAD adduct and then inhibits
the InhA enzyme (51, 52). The reactive species produced during INH activation
could also cause damage to DNA, carbohydrates, and lipids (53) and inhibit NAD
metabolism (54, 55). Changes in the NADH/NAD ratios caused by mutations in
NAD dehydrogenase II (ndh) could cause resistance to INH (56, 57). The cidal
activity of INH is very likely to be due to its effect on multiple targets in tubercle
bacillus (47). Mutations in KatG involved in INH activation (48), in the INH
target InhA (50), and Ndh II (NADH dehydrogenase II) (57) could all cause INH
resistance. KatG mutation is the major mechanism of INH resistance (46, 47).
ETH, structurally related to INH (Figure 1), is also a prodrug that is
activated by the enzyme EtaA (a monooxygenase, also called EthA) (58, 59) and
inhibits the same target InhA as INH (50) of the mycolic acid synthesis pathway.
PTH (prothionamide) shares almost identical structure and activity as ETH, where
the R group in ETH is C2H5 and the R group in PTH is C3H7 (Figure 1). EtaA is an
FAD-containing enzyme that oxidizes ETH to the corresponding S-oxide, which is
further oxidized to 2-ethyl-4-amidopyridine, presumably via the unstable oxidized
sulfinic acid intermediate (60). EtaA also activates thiacetazone, thiobenzamide,
and perhaps other thioamide drugs (60). Mutations in the drug-activating enzyme
EtaA/EthA and the target InhA cause resistance to ETA (61).
ETH/PTH
EMB
CS
4 Dec 2004
18:50
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
537
RIF
The first quinolone drug, nalidixic acid, was obtained as an impurity during
the manufacture of quinine in the early 1960s (36, 78). Since then, many FQ derivatives have been synthesized and evaluated for antibacterial activity. Ciprofloxacin
(Figure 1), ofloxacin, levofloxacin, and sparfloxacin are the best studied of these
agents and are highly active against M. tuberculosis (79). FQ inhibits DNA synthesis by targeting the DNA gyrase A and B subunits. FQ drugs are now used to
treat MDR-TB as second-line drugs but MDR-TB strains are becoming resistant
to FQ (80). An Indian study showed some promise of oxifloxacin in combination
with first-line drugs in ultra-short course of TB treatment in three months (81).
Strains of M. tuberculosis can develop resistance to FQ by mutations in GyrA or
GyrB subunit (82, 83).
FQ
4 Dec 2004
18:50
538
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
ZHANG
at 16S rRNA position 1400 are associated with high-level resistance to kanamycin
and amikacin (9092). Cross-resistance may be observed between kanamycin and
capreomycin or viomycin (9092), but a recent study found little cross-resistance
between kanamycin and amikacin (93).
New Fluoroquinolones
The new C-8-methoxy-FQ moxifloxacin (MXF) (Figure 3) and gatifloxacin with
longer half-lives are more active against M. tuberculosis, with MIC of 0.125 and
0.06 g/ml, than are ofloxacin and ciprofloxacin, with MIC of 2 and 4 g/ml,
respectively (103, 104). MXF was active against M. tuberculosis comparable to
INH in a mouse model (105, 106). MXF appeared to kill a subpopulation of
tubercle bacilli not killed by RIF, i.e., RIF tolerant persisters in vitro (107). A
4 Dec 2004
18:50
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
Figure 3
539
recent study showed that MXF in combination with RIF and PZA killed the bacilli
more effectively than the INH +RIF +PZA in mice (108). This higher activity of
MXF-RIF-PZA regimen than INH-RIF-PZA combination could be due to MXF
killing a subpopulation of bacilli not killed by INH and RIF (107), or it could
be due to the absence of the curious antagonism between INH and PZA (109)
such that replacing INH with MXF relieved such antagonism and thus showed
better sterilizing activity of MXF and PZA. The higher activity of MXF-RIF-PZA
than INH-RIF-PZA has generated considerable excitement and raises the hope that
MXF may replace INH in combination with RIF and PZA to shorten the TB therapy
in humans. However, scientists are also concerned about the potential toxicity of
MXF-RIF-PZA combination in the absence of INH as seen in the treatment of
latent TB infections with RIF-PZA (110). MXF has early bactericidal activity
against tubercle bacilli comparable to INH in a preliminary human study (111)
and was well tolerated. Combination therapy with MXF seems to be as effective
as current standard drug combinations (112). MXF and gatifloxacin are currently
being evaluated in clinical treatment of TB in combination with RIF and PZA
(R. Chaisson, D. Mitchison, personal communication). The highly active MXF
or gatifloxacin may have the potential to be used as first-line drugs for improved
treatment of TB and MDR-TB.
4 Dec 2004
18:50
540
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
ZHANG
flu-like symptoms and transient dose-dependent decrease in white blood cell and
platelet counts and did not show any better efficacy than RIF (117). Further studies are needed to more definitively assess RLZ for treatment of TB in human
trials.
Oxazolidinones (Linezolid)
Oxazolidinones are a new class of antibiotics developed by Pharmacia which were
approved by the FDA for the treatment of drug-resistant gram-positive bacterial
infections (118). Oxazolidinones inhibit an early step of protein synthesis by binding to ribosomal 50S subunits, most likely within domain V of the 23S rRNA
peptidyl transferase and forming a secondary interaction with the 30S subunit
(118, 119). Oxazolidinones had significant activity against M. tuberculosis with
an MIC of 24 g/ml and were also active against tubercle bacilli in mice (120,
121). One derivative, PNU100480 (Figure 3) had activity against M. tuberculosis
comparable to that of INH and RIF in a murine model (122). Recently, a series
of 3-(1H-pyrrol-1-yl)-2-oxazolidinone analogues of PNU-10,0480 were synthesized and some of them were found to have significant activity against M. avium
in vitro (123). Oxazolidinones may have promising potential for the treatment of
mycobacterial infections. However, treatment of human TB with oxazolidinones
has not yet been reported.
Azole Drugs
The azole drugs that are used to treat fungal infections have been shown to have
activity against M. tuberculosis (124). The azole drugs miconazole (Figure 3) and
clotrimizole were quite active against growing M. tuberculosis with an MIC of
25 g/ml, and they were also active against stationary phase bacilli (124). The
subsequent identification of cytochrome P450 homologs, a target for azole drugs, in
the M. tuberculosis genome (125) provides an explanation for the activity of azole
drugs against M. tuberculosis and led to studies to examine the correlation between
the presence of P450 and susceptibility to azole drugs in M. tuberculosis (126
129). The M. tuberculosis cytochrome P450 enzyme has recently been crystallized
and is being pursued as a target for TB drug development (130). Further in vivo
studies are needed to assess whether azole drugs can be used for the treatment of
TB.
Nitro-Containing Drugs
M. tuberculosis is quite susceptible to nitro-containing compounds. For example, niclosamide, furazolidone, 2-nitroimidazole, and 4-nitroimidazole are active
against tubercle bacilli (124). The nitro-containing compounds are likely to be
prodrugs that require activation by nitroreductases in M. tuberculosis to produce
reactive species that can damage DNA. Nitrofuran was active against nonreplicating bacilli in the Wayne dormancy model (131). It is interesting to note that
4 Dec 2004
18:50
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
541
Riminophenazine Derivatives
Clofazimine (Figure 3) is a riminophenazine derivative originally developed in the
1950s from components in lichens active against M. tuberculosis (133). Clofazimine is commonly used to treat leprosy in combination with dapsone and RIF,
and it is also used to treat M. avium intracellulare infections (134). The emergence
of drug-resistant TB has stimulated renewed interest in developing phenazines as
TB drugs. The MIC of clofazimine and its derivative B669 for M. tuberculosis is
0.152.5 g/ml (134). The mode of action of riminophenazines is not clear, but
was proposed to induce mycobacterial phospholipase A2 activity, causing interference with bacterial potassium transport (135). However, a recent study failed
to confirm this proposition (136). Clofazimine at the maximum tolerated dose of
5 mg/kg had no effect on tubercle bacilli in mice (137), but the liposomal form of
clofazimine at 50 mg/kg reduced the bacterial numbers in infected organs by 23
logs (137). Novel tetramethylpiperidine (TMP)-substituted phenazines were found
to be more active than clofazimine against M. tuberculosis and MDR-TB strains in
vitro and also had higher activity against intracellular bacilli than clofazimine and
RIF in macrophages (138). No animal studies with TMP-substituted phenazines
are available.
Phenothiazines
Phenothiazines such as chlorpromazine (CPZ) (Figure 3), thioridazine, and trifluroperazine are antipsychotic drugs with antituberculosis activity (139). Phenothiazines are calmodulin antagonists and their antituberculous activity appears
to correlate with the presence of a calmodulin-like protein in mycobacteria (140).
Phenothiazines are also active against MDR-TB (141, 142), suggesting that they inhibit a novel target in M. tuberculosis. The MIC of trifluoperazine was 832 g/ml
in vitro (143). CPZ inhibited intracellular mycobacteria at lower concentrations
0.233.6 g/ml because of its accumulation inside macrophages (144). CPZ may
also enhance the effectiveness of TB drugs against intracellular mycobacteria
(144). However, because of significant side effects, CPZ is not recommended for
treating human TB but may be used along with other TB drugs to treat TB in psychiatric patients (139). Thioridazine, which has identical anti-TB activity as CPZ
but fewer side effects, has been proposed as a candidate for human testing (139,
141).
4 Dec 2004
18:50
542
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
ZHANG
Nitroimidazopyran (PA-824)
PA-824 (Figure 3) is a new nitroimidazole derivative developed by PathoGenesisChiron (145) on the basis of an earlier observation by Indian researchers that 5nitroimidazole had good in vitro and in vivo activity against M. tuberculosis (146,
147). PA-824 was highly active against M. tuberculosis with an MIC of 0.015
0.25 g/ml (145). PA-824 was also active against nonreplicating tubercle bacilli.
PA-824 is a prodrug that is activated by F420-dependent glucose-6-phosphate
dehydrogenase and a nitroreductase activity in the bacilli (145). The resulting active
metabolites interfere with cell wall lipid biosynthesis by inhibiting an enzyme
responsible for the oxidation of hydroxymycolic acid to ketomycolate (145). PA824 was also active against MDR-TB strains, suggesting that it inhibits a new
target in tubercle bacilli. PA-824 was as active as INH in animal models of TB
infection (145). A preliminary toxicity study indicated that mice tolerated a single
dose of PA-824 at 1000 mg/kg or 500 mg/kg daily for 28 days (145). However, no
safety and efficacy data in humans are available. PA-824 is being jointly developed
by the Global Alliance for TB Drug Development and Chiron.
4 Dec 2004
18:50
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
543
inhibit novel targets are needed. In choosing targets for drug development, it is important that they be involved in vital aspects of bacterial growth, metabolism, and
viability. These targets could include cell wall synthesis, nucleic acid biosynthesis,
protein biosynthesis, and energy metabolism, resulting in either growth inhibition
or death of the bacteria. Recent developments in mycobacterial molecular genetic
tools such as transposon mutagenesis, signature-tagged mutagenesis, gene knockout, and gene transfer will facilitate the identification and validation of new drug
targets essential for the survival and persistance of tubercle bacilli not only in vitro
but also in vivo. Below is a list of potential targets whereby new drugs may be
developed for improved treatment of TB.
DosR-Rv3133/DevR-DevS
RelA
4 Dec 2004
18:50
544
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
ZHANG
Using a transposon
mutagenesis approach based on changes in colony morphology, a gene called
pcaA encoding a novel methyl transferase involved in the modification of mycolic
acids in mycobacterial cell wall was identified (168). Although the PcaA knockout
mutant grew normally in vitro and replicated in mice initially like the parent strain,
the mutant was defective in persisting in mice (168) and could be a target for drug
design against persistent bacilli.
4 Dec 2004
18:50
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
545
4 Dec 2004
18:50
546
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
ZHANG
4 Dec 2004
18:50
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
547
These inhibitors of fatty acid and mycolic acid synthesis could be good candidates
for further development. However, drugs that target cell wall synthesis are likely
to be active mainly against growing bacilli but not against persisters, and they may
not be able to shorten the lengthy therapy (212).
4 Dec 2004
18:50
548
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
ZHANG
to see if compounds that generate reactive oxygen or nitrogen species inside bacilli
could be designed as TB drugs.
4 Dec 2004
18:50
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
549
DRUG SCREENS
Because of the problem of drug-resistant TB and the need to shorten the lengthy TB
chemotherapy, there is currently a great deal of interest in TB drug development (7,
8, 11). NIH supports some antimycobacterial drug discovery research through the
NIAID Division of AIDS Opportunistic Infections Branch. The NIH-sponsored
consortium consists of in vitro screening facilities at the Tuberculosis Antimicrobial Acquisition & Coordinating Facility (TAACF) at Southern Research Institute
and at Hansen Disease Center (Baton Rogue), and at an animal testing facility at
Colorado State University. GlaxoSmithKline also has a program called Action TB
for TB drug discovery research. A private organization, the Global Alliance for
TB Drug Development, was recently established to facilitate TB drug development
(http://www.tballiance.org) and aims to have at least one TB drug registered by
2010 (229).
Both whole cell screens and cell-free target-based screens are used for antimicrobial drug discovery. The target-based screen is a relatively recent invention and
has so far been generally disappointing (233), except the recent development of
peptide deformylase inhibitors which represents the first success of the target-based
approach (148, 151, 152). However, all current TB drugs, with the exception of
PZA, were identified by in vitro whole cell screens. The current NIAID-sponsored
TB drug development effort is primarily based on screening of compounds active against growing bacilli using AlarMar Blue redox dye in a 96-well microtiter
plate format. About 70,000 compounds have been screened so far (R. Reynolds,
personal communication), and data for about 50,000 compounds were recently
published (230), where 11% (5251) had high activity against M. tuberculosis
in vitro. Of these, 53 were tested in vivo, and 9 were found to significantly reduce
bacterial numbers in the lungs of infected mice. A luciferase-reporter mycobacterial strain has also been used for screening more than 62,000 EMB analogs
generated by combinatorial chemistry for more active compounds (231). Twentysix compounds were identified; N-Geranyl-N -(2-adamantyl)ethane-1,2-diamine
(Compound 109), the most active of these diamines, was 14- to 35-fold more
active than EMB (231). Further development is required to assess its in vivo
activity. A green fluorescent protein based screening system utilizing acetamidase gene promoter was recently established for high throughput antimycobacterial compound screen (232). The combinatorial chemistry can be applied to
generate diverse compounds for screens in both whole cell and target-based screens.
4 Dec 2004
18:50
550
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
ZHANG
Although the whole cell screens are useful for TB drug development, we must
recognize the potential problem of developing drugs active against growing tubercle bacilli: drugs only active against growing bacilli are not going to be very
useful for killing nonreplicating persisters, which are the biggest stumbling block
for a more effective therapy. Although sterilizing drugs that can kill persisters and
shorten the TB therapy are desperately needed, it is not clear how this objective
can be effectively achieved. There is no good in vitro correlate of high sterilizing
activity against persisters in vivo. That is, we cannot infer from the MIC whether
the drug is going to be active against persistent bacilli or have high sterilizing activity. Low MIC does not mean the drug will have good sterilizing activity against
persistent bacilli in vivo. INH is a wonderful drug that is highly active against
growing tubercle bacilli with a very low MIC of 0.020.06 g/ml, but has no
activity against nonreplicating bacilli and therefore cannot effectively sterilize the
lesions (235). In contrast to INH, PZA is a paradoxical drug that has poor in vitro
activity against growing tubercle bacilli with a high MIC of 50100 g/ml at
pH 5.56.0 and is completely inactive against tubercle bacilli at normal culture
conditions near neutral pH, which is commonly used for whole cell MIC-based
screens. Unlike common antibiotics which are active against growing bacteria with
no activity against nonreplicating bacteria, PZA is exactly the opposite and is more
active against nonreplicating old bacilli (98) and under hypoxic conditions (99). It
is these properties that are responsible for its high sterilizing activity in vivo and
its ability to shorten the therapy from 912 months to 6 months. PZA was discovered by a serendipitous observation in 1940s that nicotinamide had activity against
mycobacteria in animal models; subsequent synthesis of nicotinamide analogs and
direct screen in mice without MIC testing identified PZA as the most active agent in
vivo (45). In a sense, we should feel fortunate that we have the wonderful sterilizing
drug PZA, which would have been missed altogether had the conventional MICbased screens been used. As we can see, the MIC-based approach does not work
here! If there is any lesson to be learned from the PZA story, it is that we cannot use
the MIC-based screens to identify drugs that have high sterilizing activity against
persisters.
To identify drugs that effectively kill nongrowing persisters and shorten the
therapy, we must design new and unconventional screens that mimic the persisters in vivo, such as using nonreplicating bacilli at low oxygen and acid pH
in the screen, a process that is more challenging. There are different persistence models that can potentially be used for screening for sterilizing drugs (41,
212). Stationary phase bacilli, old and starved bacilli, and persisting bacilli after drug treatment can all be used in such screens. Synergy screens with different agents should also be considered. Because of our limited understanding of
mycobacterial persisters, it is difficult to judge if one model is better than another. However, testing in animals will show which in vitro persistence model
is more relevant to the goal of shortening the therapy. In addition, potential targets involved in persistence (see above) could also be selected for target-based
screens.
4 Dec 2004
18:50
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
551
CONCLUDING REMARKS
The development of modern TB chemotherapy is indeed a remarkable achievement
of modern medicine and represents a major milestone in humankinds fight against
TB. Yet despite the availability of TB chemotherapy and the BCG vaccine, TB is
still a leading infectious disease worldwide. Along with the socio-economic and
host factors that underlie this problem, a fundamental problem that hinders more
effective TB control is the tenacious ability of M. tuberculosis to persist in the
host and to develop drug resistance, often as a consequence of poor compliance to
lengthy therapy. Novel screens targeting persisters are needed but such screens are
challenging. PZA represents a prototype model drug that can shorten TB therapy,
and improved understanding of PZA should help us to design drugs that are more
active against persisters. Although having another new drug like INH that only
kills growing bacilli may be useful for treating drug-resistant TB, it is unlikely
to improve the current TB therapy. The development of new sterilizing drugs that
target persisters and shorten the TB therapy must be a top priority. This represents
a paradigm shift from previous approaches, which focused on just finding another
drug, to beating mycobacterial persistence. In the big picture, we must recognize
that better control of TB extends beyond better chemotherapy; it requires a multifaceted approach, including improved socio-economic conditions and nutrition,
better management of adverse psychological factors, and improved host immunity
as adjunct treatment (41). The recent developments in mycobacterial genetic tools
and TB genomics, new technology of combinatorial chemistry and high throughput screening, structure-based drug design, and improved understanding of the
unique biology of tubercle bacillus provide an exciting opportunity to discover
new Magic Bullets that kill persisters and shorten the current TB treatment from
six months to a few weeks. A new era of TB chemotherapy will arrive when these
new Magic Bullets are identified.
ACKNOWLEDGMENTS
The support from NIH (AI44063 and AI/HL49485) is gratefully acknowledged.
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Ryan F. 1993. The Forgotten Plague. How
the Battle against Tuberculosis Was Won
and Lost, p. 3. Boston: Little, Brown. 460
pp.
2. World Health Organization. 2003. The
World Health Organization Global
Tuberculosis Program. http://www.who.
int/gtb/
4 Dec 2004
18:50
552
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
ZHANG
prisons: treatment failure and MDR-TB.
Int. J. Tuberc. Lung Dis. 3:45153
De Cock KM, Chaisson RE. 1999. Will
DOTS do it? A reappraisal of tuberculosis control in countries with high rates of
HIV infection. Int. J. Tuberc. Lung Dis.
3:45765
Tuberculosis: a global emergency. 1993.
World Health Forum 14:438
Global Alliance for Tuberculosis Drug
Development. 2001. Scientific blueprint
for TB drug development. Tuberculosis
81(Suppl.):152
Smith CV, Sharma V, Sacchettini JC.
2004. TB drug discovery: addressing issues of persistence and resistance. Tuberculosis 84:4555
Duncan K. 2003. Progress in TB drug development and what is still needed. Tuberculosis 83:2017
Kremer LS, Besra GS. 2002. Current status and future development of antitubercular chemotherapy. Expert. Opin. Investig. Drugs 11:103349
Zhang Y, Amzel LM. Tuberculosis drug
targets. Curr. Drug Targets 3:13154
Barry CE 3rd. 2001. Preclinical candidates and targets for tuberculosis therapy.
Curr. Opin. Investig. Drugs 2:198201
Inderlied CB, Salfinger M. 1999. Antimycobacterial agents and susceptibility
tests. In Manual of Clinical Microbiology, ed. PR Murray, EJ Baron, MA Pfaller,
FC Tenover, RH Yolken, pp. 16011623.
Washington DC: ASM Press. 7th ed.
Domagk G. 1935. Ein Beitrag zur
Chemotherapie der bakteriellen Infektionen. Deutschemedizinische Wochenschrift 61:25053
Hatfull G, Jacobs WR Jr, eds. 2000.
Molecular Genetics of Mycobacteria.
Washington DC: ASM Press
Deleted in proof
Schatz AB, Bugie E, Waksman S. 1944.
Streptomycin, a substance exhibiting antibiotic activity against gram-positive and
gram-negative bacteria. Proc. Soc. Exp.
Biol. Med. 55:6669
4 Dec 2004
18:50
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
553
4 Dec 2004
18:50
554
53.
54.
55.
56.
57.
58.
59.
60.
61.
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
ZHANG
13C NMR characterization of hemiamidal isoniazid-NAD(H) adducts as possible
inhibitors of InhA reductase of Mycobacterium tuberculosis. Chemistry 9:2034
38
Winder F. 1982. Mode of action of the
antimycobacterial agents and associated
aspects of the molecular biology of mycobacteria. In The Biology of Mycobacteria, ed. C Ratledge, J Stanford, pp. 354
438. New York: Academic
Winder F, Collins P. 1968. The effect of
isoniazid on nicotinamide nucleotide levels in Mycobacterium bovis strain BCG.
Am. Rev. Respir. Dis. 97:71920
Sriprakash KS, Ramakrishnan T. 1969.
Isoniazid and nicotinamide adenine dinucleotide synthesis in M. tuberculosis. Indian J. Biochem. 6:4950
Miesel L, Weisbrod T, Marcinkeviciene
JA, Bittman R, Doshi P, et al. 1998.
NADH dehydrogenase defects confer resistance to isoniazid and conditional
lethality in Mycobacterium smegmatis. J.
Bacteriol. 180:245967
Lee AS, Teo AS, Wong SY. 2001. Novel
mutations in ndh in isoniazid-resistant
Mycobacterium tuberculosis isolates. Antimicrob. Agents Chemother. 45:215759
DeBarber AE, Mdluli K, Bosman M,
Bekker LG, Barry CE 3rd. 2000.
Ethionamide activation and sensitivity
in multidrug-resistant Mycobacterium tuberculosis. Proc. Natl. Acad. Sci. USA
97:967782
Baulard AR, Betts JC, Engohang-Ndong
J, Quan S, McAdam RA, et al. 2000. Activation of the pro-drug ethionamide is
regulated in mycobacteria. J. Biol. Chem.
275:2832631
Vannelli TA, Dykman A, Ortiz de Montellano PR. 2002. The antituberculosis
drug ethionamide is activated by a flavoprotein monooxygenase. J. Biol. Chem.
277:1282429
Morlock GP, Metchock B, Sikes D, Crawford JT, Cooksey RC. 2003. ethA, inhA,
and katG loci of ethionamide-resistant
62.
63.
64.
65.
66.
67.
68.
69.
4 Dec 2004
18:50
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
555
4 Dec 2004
18:50
556
89.
90.
91.
92.
93.
94.
95.
96.
97.
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
ZHANG
RNA pseudoknot. Mol. Microbiol. 9:
123946
Zhang Y, Telenti A. 2000. Genetics of
drug resistance in Mycobacterium tuberculosis. See Ref. 15, pp. 23554
Alangaden GJ, Kreiswirth BN, Aouad
A, Khetarpal M, Igno FR, et al. 1998.
Mechanism of resistance to amikacin
and kanamycin in Mycobacterium tuberculosis. Antimicrob. Agents Chemother.
42:129597
Suzuki Y, Katsukawa C, Tamaru A, Abe
C, Makino M, et al. 1998. Detection of
kanamycin-resistant Mycobacterium tuberculosis by identifying mutations in
the 16S rRNA gene. J. Clin. Microbiol.
36:122025
Taniguchi H, Chang B, Abe C, Nikaido
Y, Mizuguchi Y, et al. 1997. Molecular
analysis of kanamycin and viomycin resistance in Mycobacterium smegmatis by
use of the conjugation system. J. Bacteriol. 179:4795801
Kruuner A, Jureen P, Levina K, Ghebremichael S, Hoffner S. 2003. Discordant resistance to kanamycin and
amikacin in drug-resistant Mycobacterium tuberculosis. Antimicrob. Agents
Chemother. 47:297173
Scorpio A, Zhang Y. 1996. Mutations
in pncA, a gene encoding pyrazinamidase/nicotinamidase, cause resistance to
the antituberculous drug pyrazinamide
in tubercle bacillus. Nat. Med. 2:662
67
Scorpio A, Lindholm-Levy P, Heifets
L, Gilman R, Siddiqi S, et al. 1997.
Characterization of pncA mutations in
pyrazinamide-resistant Mycobacterium
tuberculosis. Antimicrob. Agents Chemother. 41:54043
Tarshis MS, Weed WA. 1953. Lack of significant in vitro sensitivity of Mycobacterium tuberculosis to pyrazinamide on
three different solid media. Am. Rev. Tuberc. 67:39195
McDermott W, Tompsett R. 1954. Activation of pyrazinamide and nicotinamide
98.
99.
100.
101.
102.
103.
104.
105.
106.
4 Dec 2004
18:50
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
115.
116.
117.
118.
119.
120.
121.
122.
557
4 Dec 2004
18:50
558
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
ZHANG
132.
133.
134.
135.
136.
137.
138.
139.
4 Dec 2004
18:50
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
140.
141.
142.
143.
144.
145.
146.
147.
148.
559
4 Dec 2004
18:50
560
159.
160.
161.
162.
163.
164.
165.
166.
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
ZHANG
for hypoxic dormancy. J. Bacteriol.
184:676067
Voskuil MI, Schnappinger D, Visconti
KC, Harrell MI, Dolganov GM, Sherman
DR, Schoolnik GK. 2003. Inhibition of
respiration by nitric oxide induces a Mycobacterium tuberculosis dormancy program. J. Exp. Med. 198:70513
Cashel M, Gentry DR, Hernandez VJ,
Vinella D. 1996. The stringent response.
In Escherichia coli and Salmonella: Cellular and Molecular Biology, ed. FC Neidhardt, R Curtis, JL Ingraham, EC Lin,
KB Low, et al., pp. 148896. Washington, DC: ASM Press
Primm TP, Andersen SJ, Mizrahi V, Avarbock D, Rubin H, et al. 2000. The stringent
response of Mycobacterium tuberculosis
is required for long-term survival. J. Bacteriol. 182:488998
Dahl JL, Kraus CN, Boshoff HI, Doan B,
Foley K, et al. 2003. The role of RelMtbmediated adaptation to stationary phase in
long-term persistence of Mycobacterium
tuberculosis in mice. Proc. Natl. Acad.
Sci. USA 100:1002631
McKinney JD, Honer Zu, Bentrup K,
Munoz-Elias EJ, Miczak A, Chen B, et al.
2000. Persistence of Mycobacterium tuberculosis in macrophages and mice requires the glyoxylate shunt enzyme isocitrate lyase. Nature 406:73538
Wayne LG, Lin KY. 1982. Glyoxylate metabolism and adaptation of Mycobacterium tuberculosis to survival under anaerobic conditions. Infect. Immun.
37:104249
Graham JE, Clark-Curtiss JE. 1999. Identification of Mycobacterium tuberculosis
RNAs synthesized in response to phagocytosis by human macrophages by selective capture of transcribed sequences
(SCOTS). Proc. Natl. Acad. Sci. USA
96:1155459
Fenhalls G, Stevens L, Moses L,
Bezuidenhout J, Betts JC, et al. 2002. In
situ detection of Mycobacterium tuberculosis transcripts in human lung granulo-
167.
168.
169.
170.
171.
172.
173.
174.
175.
4 Dec 2004
18:50
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
561
4 Dec 2004
18:50
562
196.
197.
198.
199.
200.
201.
202.
203.
204.
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
ZHANG
of diaryltriazoles against bacterial twocomponent regulatory systems and their
antibacterial activities. Bioorg. Med.
Chem. Lett. 8:92934
Deleted in proof
Via LE, Curcic R, Mudd MH, Dhandayuthapani S, Ulmer RJ, et al. 1996.
Elements of signal transduction in Mycobacterium tuberculosis: in vitro phosphorylation and in vivo expression of the
response regulator MtrA. J. Bacteriol.
178:331421
Supply P, Magdalena J, Himpens S, Locht
C. 1997. Identification of novel intergenic
repetitive units in a mycobacterial twocomponent system operon. Mol. Microbiol. 26:9911003
Ewann F, Locht C, Supply P. 2004.
Intracellular autoregulation of the Mycobacterium tuberculosis PrrA response
regulator. Microbiology 150:24146
Zahrt TC, Deretic V. 2001. Mycobacterium tuberculosis signal transduction
system required for persistent infections.
Proc. Natl. Acad. Sci. USA 98:12706
11
Perez E, Samper S, Bordas Y, Guilhot C,
Gicquel B, et al. 2001. An essential role
for phoP in Mycobacterium tuberculosis
virulence. Mol. Microbiol. 41:17987
Parish T, Smith DA, Roberts G, Betts
J, Stoker NG. 2003. The senX3-regX3
two-component regulatory system of
Mycobacterium tuberculosis is required
for virulence. Microbiology 149:1423
35
Rickman L, Saldanha JW, Hunt DM,
Hoar DN, Colston MJ, et al. 2004. A
two-component signal transduction system with a PAS domain-containing sensor is required for virulence of Mycobacterium tuberculosis in mice. Biochem.
Biophys. Res. Commun. 314:25967
Dasgupta N, Kapur V, Singh KK, Das
TK, Sachdeva S, et al. 2000. Characterization of a two-component system, devRdevS, of Mycobacterium tuberculosis. Tuber. Lung Dis. 80:14159
4 Dec 2004
18:50
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
P1: JRX
214.
215.
216.
217.
218.
219.
220.
221.
222.
563
4 Dec 2004
18:50
564
AR
AR232-PA45-22.tex
XMLPublishSM (2004/02/24)
ZHANG
232. Changsen C, Franzblau SG, Palittapongarnpim P. 2003. Improved green fluorescent protein reporter gene-based
microplate screening for antituberculosis
compounds by utilizing an acetamidase
promoter. Antimicrob. Agents Chemother.
47:368287
233. Projan S. 2002. New (and not so new) anAnnu. Rev. Pharmacol. Toxicol. 2005.45:529-564. Downloaded from arjournals.annualreviews.org
by Universitaet Heidelberg on 10/01/05. For personal use only.
P1: JRX
4 Dec 2004
18:51
AR
AR232-PA45-23.tex
XMLPublishSM (2004/02/24)
P1: JRX
10.1146/annurev.pharmtox.45.120403.095946
INTRODUCTION
Malaria is one of the greatest of all infectious diseases, afflicting more than 500
million people and causing approximately 2 million deaths each year. Estimates
of the economic burden of malaria in terms of lost productivity are staggering
(1). Malaria is transmitted by mosquitoes and caused by intracellular protozoan
parasites from the genus Plasmodium. By far the most significant species is P. falciparum, which causes severe infections and death, enjoys widespread geographic
distribution, and is most likely to be drug resistant. In the years after World War II,
public health workers had ambitious plans to eradicate malaria by various means,
including DDT against mosquitoes and chloroquine against the parasite. These
0362-1642/05/0210-0565$14.00
565
4 Dec 2004
18:51
566
AR
AR232-PA45-23.tex
ARAV-BOGER
XMLPublishSM (2004/02/24)
P1: JRX
SHAPIRO
efforts unfortunately failed; among the reasons for failure was the appearance and
spread of chloroquine-resistant malaria, an event that is aptly considered a public
health crisis. Malaria now features prominently among the reemerging infectious
diseases (2).
Although much work is being done to develop malaria vaccines, estimates are
that it will be many years before these are suitable for use in humans, and drugs
are therefore required not only for treatment of established infections but also
for prevention of malaria in healthy travelers, tens of millions of whom go to
malarious countries every year (3). Malaria therapy is complicated by a number of
factors, including the considerable requirement for safety (huge numbers afflicted,
disproportionate severity in children and pregnant women, prophylaxis of healthy
travelers), the fact that selective toxicity may be more difficult to attain against these
eukaryotic pathogens, and by the inherent complexity of the parasites lifecycle
within the human host. Each lifecycle stage varies in its drug-sensitivity profile;
hence, for a given patient multiple drugs may be needed to eradicate the infection.
Infection begins with the bite of an infected Anopheline mosquito. Parasites first
invade hepatocytes and replicate there before bursting the cell. The released forms
then infect, replicate within, rupture, and reinfect red cells in a cycle that repeats
every 23 days. This asexual replication leads to tremendous amplification, with
parasite burdens that may reach 1012 organisms per patient. Drug-resistance genes
that arise and are selected in this setting are further spread through the gene pool
by the meiotic exchange that occurs during the sexual reproduction of Plasmodium
within the mosquito.
The recently available genome for P. falciparum provides powerful information
for understanding resistance mechanisms and opens exciting new avenues for
drug development (4). P. falciparum contains 14 chromosomes and approximately
5300 protein-encoding genes, almost two-thirds of which seem to be unique to
this organism. Newly recognized cellular pathways and organelles, such as the
apicoplast (a chloroplast-like structure with unique metabolism), provide novel
targets for the development of selectively toxic new therapies. Information on P.
falciparum genes and their expression is available on the PlasmoDB Web site
(http://www.plasmoDB.org).
In this review, we provide an overview of the problem of antimalarial drug
resistance, consider potential solutions, and refer interested readers to the many
excellent and detailed reviews that have appeared in recent years (512). Given its
clinical and public health importance, and because it is by far the most likely to
be drug resistant, the discussion focuses almost entirely on P. falciparum.
4 Dec 2004
18:51
AR
AR232-PA45-23.tex
XMLPublishSM (2004/02/24)
P1: JRX
567
and most recently to assays for known gene mutations. Each of these approaches
has its merits, but for many reasons they may not be concordant. The assessment of
antimalarial drug resistance, and the correlation of clinical and laboratory findings,
is confounded by many variables. These include the obvious generic issues: distinguishing genuine resistance from suboptimal therapy, immunity and nutritional
factors, and culturing parasites in conditions where key nutrients far exceed those
in blood. There are also confounding variables more particular to malaria. In the
field, resistant parasites may take weeks to recrudesce, at which point it becomes
difficult to distinguish drug failure from reinfection. Furthermore, patients may
harbor many clones of P. falciparum, each with a distinct set of mutations that
impart resistance. Thus, if two mutations in a single gene are detected in a patients blood sample, unless clonal parasites are isolated and assayed, it is difficult
to know whether both mutations are in one cell line or whether two cell lines each
have one mutation. Fluorogenic assays that distinguish between these possibilities
may provide a solution to this problem (15).
A rich variety of genetic mechanisms are exploited for drug resistance in bacteria and tumor cells (16, 17). These range from discrete point mutations to the
rearrangement of large blocks of DNA (e.g., inversion, duplication, insertion, deletion, transposition), and even to the acquisition of foreign DNA. Alterations in gene
transcription, in the posttranscriptional control of RNA, and in the posttranslational
modification of proteins, play important roles in drug resistance. By comparison,
relatively few mechanisms are recognized in malaria and, as described below, the
best understood of these are confined to point mutations and changes in steady-state
transcript levels. Point mutations provide a satisfying and consistent explanation
for many cases of antimalarial drug resistance. Almost certainly, however, there
are mechanisms at work in these parasites that remain to be found. The availability of the fully sequenced genome and proteome that follows will be key in this
discovery process.
DRUG-SPECIFIC RESISTANCE
4-Substituted Quinolines
The members of this largest class of antimalarial agents share obvious structural
analogy, which reflects their derivation from the natural product quinine (Figure 1).
As described below, they also have a common molecular mechanism of antimalarial activity. The preeminent agent in this class has been chloroquine, which in
retrospect has aptly been termed a wonder drug (18). The focus of some intrigue during the years of World War II (19), this fully synthetic antimalarial is
inexpensive, safe, and orally bioavailable. For decades, chloroquine provided reliable prophylaxis for travelers, therapy for those with established infection, and a
powerful tool for public health workers in their efforts to control malaria. The emergence in the early 1960s and subsequent spread of chloroquine-resistant parasites
created a tremendous therapeutic void, which has not yet been filled satisfactorily.
4 Dec 2004
18:51
568
AR
AR232-PA45-23.tex
ARAV-BOGER
XMLPublishSM (2004/02/24)
P1: JRX
SHAPIRO
4 Dec 2004
18:51
AR
AR232-PA45-23.tex
XMLPublishSM (2004/02/24)
P1: JRX
569
selectively disrupts the aggregation of malarial pigment within the food vacuole
(24), and more recent experiments have refined this picture to show that chloroquine
effectively blocks the sequestration of toxic heme into hemozoin (25). Chloroquine
accumulates in parasitized red cells, particularly in the acidic digestive vacuole,
to reach levels hundreds of times those in plasma, and the accumulation is reduced substantially in chloroquine-resistant cells (26). Subsequent studies with
chloroquine-resistant P. falciparum confirmed these findings; noted the lack of
cross-reactivity with quinine, mefloquine, or chloroquine analogs (27); described
a paradoxically increased sensitivity to some antimalarials (28); found that reduced steady-state levels were attributable to enhanced efflux, not reduced uptake
(29); and revealed that verapamil could partially restore the accumulation of, and
sensitivity to, chloroquine (30). Although these phenotypic characteristics have
been invaluable in suggesting and corroborating the molecular mechanisms of resistance, the definitive studies have been genetic. Despite heavy drug pressure, it
took many years for chloroquine-resistant P. falciparum to emerge in the field.
This observation, together with the fact that chloroquine resistance in the laboratory could only be generated in the presence of mutagens, led to the suspicion that
chloroquine-resistance might well be multigenic.
As described in detail below, two entirely distinct experimental approaches
have yielded two independent genetic sources of chloroquine resistance in P. falciparum. One of these, pfcrt (P. falciparum chloroquine-resistance transporter)
is now recognized to be both necessary and sufficient to impart chloroquine resistance. The other, pfmdr1 (P. falciparum multidrug-resistance 1), may further
modulate the degree of resistance.
One experimental approach was an undirected search for a gene(s) that would
sort with chloroquine resistance when sensitive and resistant parent lines were
crossed during sexual reproduction in the mosquito (31). The resulting progeny
were fully sensitive or resistant, consistent with changes at a single genetic locus in
these haploid forms. Some ten additional years of work were required to identify
the rather cryptic 13-exon pfcrt on chromosome 7 (32). This gene encodes a novel
45 kDa protein with ten predicted transmembrane domains that immunolocalizes
to the membrane of the digestive vacuole. It has no obvious homology to the
large family of ABC (ATP-binding cassette) transporters that pump drugs against
a concentration gradient at the expense of ATP (33, 34). The predicted protein
is thought to be a transporter or channel that reduces chloroquine levels in the
digestive vacuole, which in turn reduces the accumulation of free heme and relieves
cytotoxicity. The mechanism by which pfcrt affects chloroquine levels is not yet
clear but it may involve altered ion fluxes that change the acidity of the vacuole, or
alternatively, pfcrt may interact directly with chloroquine itself (35). Studies of this
process have been hampered by the difficulty in expressing this transmembrane
protein in heterologous systems.
Analysis of pfcrt in cell lines obtained from many geographic locations revealed
a consistent wild-type sequence in the sensitive lines, and a remarkable array of
mutations in chloroquine-resistant lines (32). Genes from resistant cells have at
4 Dec 2004
18:51
570
AR
AR232-PA45-23.tex
ARAV-BOGER
XMLPublishSM (2004/02/24)
P1: JRX
SHAPIRO
least five and up to eight mutations, all confined to ten positions that are clustered
within or near transmembrane domains. Common to all resistant lines are a K76
mutation, which now provides a valuable molecular marker in surveillance studies
and a predictor of chloroquine efficacy (36). The limited patterns of mutations
suggest that resistant lines originated in just a few discrete geographic locations
from which they then spread. This notion is strengthened by a more recent genomewide satellite marker analysis in dozens of strains of P. falciparum, which reveals
a striking lack of polymorphism surrounding pfcrt in chromosome 7, relative to all
other portions of the genome (37). This prominent aberration reflects the powerful
selective pressure that extensive chloroquine use has exerted on this parasites
evolution. The essential role of pfcrt was firmly established by allelic exchange of
the endogenous pfcrt in chloroquine-sensitive cells for mutant alleles from resistant
lines, which effectively conferred a chloroquine-resistant phenotype (38).
Mutations in pfcrt have now been shown to account for the recognized characteristics of chloroquine-resistant cells described above: reduced accumulation of
chloroquine (38, 39); lack of cross-resistance with quinine and mefloquine (35),
indeed a paradoxically increased sensitivity to some antimalarials (38); and an
acceptable fulfillment of the expectation for a multigenetic mechanism (e.g., multiple mutations required, although all in a single gene). Finally, although perhaps
not a consequence that should be expected, the chloroquine resistance imparted
by mutant PfCRT is partially reversible by verapamil (38, 39). The latter is a
well-recognized antagonist of drug efflux pumps (33, 34); however, its action is
confined to just one of the seven classes of ABC transporters, and PfCRT is not
even a member of the ABC transporter family.
A second independent experimental approach to understanding the genetic basis for chloroquine resistance actually preceded that described above, and was a
directed search for ABC transporter genes whose sequence or expression might be
altered in drug-resistant cells. This logical search was prompted by accumulating
evidence that upregulation of ABC transporter gene expression is associated with
multidrug resistance in tumor cells, and by the finding that chloroquine resistance
is partially reversed by verapamil (30). With the completion of the human genome,
the family of ABC transporters is now divided into seven different classes on the
basis of sequence homology (33, 34). All are membrane-spanning proteins and
have highly characteristic nucleotide-binding domains. Best studied of the ABC
transporters is ABCB1 (also termed Pgy1, MDR1, Pgp, or GP170), whose preferred substrates (hydrophobic, planar aromatic rings, with the presence of tertiary
amino groupscriteria all fulfilled by chloroquine; Figure 1) are pumped against
a concentration gradient at the expense of ATP hydrolysis and whose action is antagonized by verapamil. Notably, mutations in human ABCB1 are not associated
with recognizable disease or with altered drug transport; the latter is mediated by
upregulated expression.
Using phylogenetically conserved ABCB1 sequences, two laboratories identified mdr genes (termed pfmdr1 and pfmdr2) in P. falciparum (40, 41). Subsequent
studies implicated only pfmdr1 in drug resistance, although the association was
4 Dec 2004
18:51
AR
AR232-PA45-23.tex
XMLPublishSM (2004/02/24)
P1: JRX
571
imperfect and variably involved either gene amplification or point mutations. The
clearest data bearing on the role of pfmdrs in drug-resistant P. falciparum indicate that these genes do not sort with chloroquine resistance in genetic crosses of
sensitive and resistant cells (31), and that the introduction of mutant pfmdr1 into
cells with wild-type pfcrt has no effect on chloroquine sensitivity (42). Importantly, however, the addition of mutant pfmdr1 to cells already harboring mutant
pfcrt does enhance chloroquine resistance, indicating that mutations in this gene
may modulate the overall response to chloroquine (38, 42). Of considerable interest and distinct from pfcrt, mutations in pfmdr1 are associated with resistance to
mefloquine, quinine, and halofantrine (42).
Although the exposure of P. vivax and P. falciparum to chloroquine has been
similar, the appearance of chloroquine-resistant P. vivax took nearly 30 additional
years to appear. First reported from Papua New Guinea in 1989 (43), chloroquineresistant P. vivax has now spread through Southeast Asia and into South America.
Unexpected and intriguing is the finding that chloroquine resistance in P. vivax is
apparently not mediated by mutations in the vivax homolog of pfcrt (44). Despite
the interest and importance of this problem, the technical difficulties in studying
P. vivax seriously hamper definitive studies.
A number of new therapeutic approaches have been taken on the basis of lessons
learned from chloroquine and the mechanisms of chloroquine resistance. These
include the use of analogs that differ only in the length of the 4-aminoalkyl side
chain, which retain antimalarial activity but are not cross-resistant with chloroquine (45); coadministration of chloroquine with various chemosensitizers in an
effort to reverse the efflux mechanism (46), although for antitumor agents this
approach has met with very limited success (33); and use of chloroquine in combination with other antimalarials, most notably an artemisinin (47). The documented
reemergence of chloroquine-sensitive parasites when drug pressure is removed is
fascinating and may afford an opportunity to reintroduce chloroquine after years
of nonuse (48).
4 Dec 2004
18:51
572
AR
AR232-PA45-23.tex
ARAV-BOGER
XMLPublishSM (2004/02/24)
P1: JRX
SHAPIRO
(53); second, the structural analogy of proguanil to a series of antibacterial 2,4diaminopyrimidines was recognized by Hitchings, who then demonstrated potent
antimalarial activity in this new chemical class of antifolates (54); and third, it
provided an effective means to forestall the emergence of resistance, which even
in earliest experiments was recognized as a serious problem. The eventual consequence was an antifolate/sulfonamide combination of pyrimethamine/sulfadoxine
(Figure 1) that was carefully selected for matching pharmacokinetics, formulated
in fixed ratios to maximize synergy, and marketed as Fansidar . (The antibacterial trimethoprim/sulfamethoxazole was similarly developed.) The extraordinary
degree of synergism in these combinations, which allows some 20-fold reduction
in the dose of each component, is still attributed to multiple blockades in a single metabolic pathway, although evidence to support this widely cited mechanism
remains circumstantial and other mechanisms may contribute (5558).
Tetrahydrofolate is an essential cofactor in the methyl transfer reactions that
generate monomers for protein and nucleic acid synthesis (59). In several important respects, folate biosynthesis in malaria parasites is distinctly different from
that in other systems (pathways and key points of drug inhibition in Figure 2, see
color insert). First, from biochemical studies and the annotated genome it is now
clear that P. falciparum is unique in that both para-aminobenzoic acid (6062) and
dihydrofolic acid (58, 63, 64) can be synthesized de novo as well as salvaged from
the environment. The availability of these salvage pathways has severely complicated in vitro inhibition studies, and they clearly modulate antifolate efficacy in
patients, whose blood levels of para-aminobenzoic acid and dihydrofolate may
vary widely (65). A second dissimilar feature in Plasmodium folate metabolism is
that sequential reactions may be catalyzed by a single bifunctional protein. Thus,
dihydro-6-hydroxymethylpterin pyrophosphokinase and dihydropteroate synthase
are encoded by the same gene and contained within the same protein (66, 67). Dihydrofolate reductase and thymidylate synthase activities are similarly linked (68,
69). This structural organization may improve catalytic efficiency by channeling
substrates in a processive fashion through two sequential transformations; it may
also offer novel strategies for drug-mediated disruption. Finally, malaria parasites
are especially susceptible to inhibition of dihydrofolate reductase because (unlike
mammalian cells) transcriptional inhibition, mediated by the protein binding to its
own message, is not relieved by the accumulation of substrate that occurs in the
presence of inhibitor (70). This precludes the upregulation of protein synthesis as
a means to counter antifolate inhibitors and it contributes to the selective toxicity
of antifolates against the parasite.
Chloroquines efficacy, safety, and low cost made it the clear drug of choice
for many decades, but the advent of chloroquine-resistant parasites established
pyrimethamine/sulfadoxine as the next best option, despite the recognized propensity for resistance and the concern about antifolate teratogenicity (71). Malaria
parasite resistance to sulfonamides and antifolates has been known for more
than 50 years (7274). Although available mechanisms reportedly include gene
amplification, which is the only recognized mechanism associated with clinical
4 Dec 2004
18:51
AR
AR232-PA45-23.tex
XMLPublishSM (2004/02/24)
P1: JRX
573
resistance to antifolate therapy in cancer (17), a large body of evidence now indicates that in Plasmodium the major effector of resistance is point mutations in the
key target enzymes: dihydropteroate synthase and dihydrofolate reductase. Unlike
the transmembrane proteins that mediate chloroquine resistance, native and recombinant forms of the synthase and reductase are soluble and assayable; hence,
the findings in genetic studies have been bolstered by biochemical and structural
experiments.
Molecular epidemiology studies from South America and Africa provide multiple lines of evidence that application of pyrimethamine/sulfadoxine therapy leads
to the progressive and orderly accumulation of point mutations, first in dihydrofolate reductase and then in dihydropteroate synthase. The sequential addition of
new mutations is evident in field isolates collected over years of time (75, 76), in
pre- versus posttreated patients (77), and in correlation with the degree of clinical resistance for a given patient or geographic region (78). Evaluation of these
mutations in the context of surrounding polymorphisms in noncoding sequences
is consistent with focal origin of mutant strains followed by spread through the
population via gene flow (75, 76). Highest levels of clinical resistance result from
parasites with four mutations in dihydrofolate reductase and two in dihydropteroate
synthase, which may represent the maximum number of mutations that can be tolerated in competition with less-affected strains. The utility of these mutations as
predictors for therapeutic response is modulated by host immunity, as evidenced
by the persistent efficacy of pyrimethamine/sulfadoxine in holoendemic Malawi,
despite ongoing use of these agents in a population that has harbored highly mutant
parasites for at least five years (79).
Laboratory findings that corroborate these field data and underscore the central
importance of point mutations include the appearance of the appropriate drugresistant phenotype in genetic crosses or when mutant genes are introduced into
wild-type cells (8082) and analysis of the inhibition kinetics of recombinant wildtype versus mutant enzymes (83, 84). The recently available crystal structure for
dihydrofolate reductase-thymidylate synthase provides satisfying evidence that
the critical mutations mediating clinical drug resistance map to the dihydrofolate
reductase active site (85).
The well-studied and proven value of the folate synthetic machinery as an antimalarial target has prompted several ingenious research efforts to devise new
interventions against tetrahydrofolate production and use. These include inhibition of the shikimate pathway, which provides an intracellular source of paraaminobenzoic acid (Figure 2), alone or in combination with downstream inhibitors (61); dihydrofolate reductase inhibitors rationally designed and selected
for activity against the clinically important quadruple mutant malaria enzyme
but not the human reductase (86); identification of novel chemical classes by in
silico docking of large chemical libraries into the known dihydrofolate reductase three-dimensional (3-D) structure (87); and deployment of folate analogs
against thymidylate synthase (88). More immediate clinical efforts have focused
on using sulfonamide/antifolate combinations that are less cross-resistant and/or
4 Dec 2004
18:51
574
AR
AR232-PA45-23.tex
ARAV-BOGER
XMLPublishSM (2004/02/24)
P1: JRX
SHAPIRO
have a shorter plasma half-life (89, 90) and adding a third antimalarial to the
pyrimethamine/sulfadoxine dosing regimen (47, 91).
4 Dec 2004
18:51
AR
AR232-PA45-23.tex
XMLPublishSM (2004/02/24)
P1: JRX
575
MULTIDRUG-RESISTANT PARASITES
In cancer chemotherapy, resistance to structurally and mechanistically diverse
agents can be mediated by alterations in expression of a single ABC transporter
gene (17, 105). As we now understand it, multidrug resistance for P. falciparum
is different: It involves genetic alterations in at least two, and often more, proteins [the difficult problem of multidrug-resistant P. falciparum has been thoughtfully defined and reviewed recently (8)]. Typically, this means resistance to both
chloroquine and pyrimethamine/sulfadoxine, mediated by mutations in pfcrt, dihydropteroate synthase and dihydrofolate reductase, as described above. However,
strains resistant to chloroquine, sulfadoxine/pyrimethamine, mefloquine, and partially resistant to quinine and quinidine have been described (106). Malaria in
Southeast Asia is notorious for its propensity to develop early and multidrug resistance. This prompted an interesting experiment comparing the emergence of
resistance in a parasite clone from Africa (which was fully susceptible to conventional antimalarials) to that of a multidrug-resistant clone from Indochina (107).
Two compounds were selected that had novel killing mechanisms and had never
before been applied to these parasites. The Indochina clone acquired resistance
some 1000 times more frequently, suggesting these parasites may have an underlying accelerated mutator or hyperrecombination phenotype.
4 Dec 2004
18:51
576
AR
AR232-PA45-23.tex
ARAV-BOGER
XMLPublishSM (2004/02/24)
P1: JRX
SHAPIRO
may inhibit ATPase and alter intracellular calcium stores (110). As a class the
artemisinins are potent, fast-acting, and remarkably impervious to resistance, although recrudescence of fully sensitive parasites is common. Human safety for this
class is regularly claimed despite the unfortunate rarity of systematic safety evaluations available in the literature. The current recommended use for artemisinins
is in combination therapy, where they effect a rapid and massive decrease in parasite burden and their gametocytocidal activity may lessen transmission of resistant
parasites to the mosquito. As noted above, several large clinical trials have already demonstrated their meaningful contribution to efficacy (47, 111), and even
larger studies are underway as a likely prelude to national health policy recommendations (6).
Combination Therapy
For both antitumor and antiinfective therapies, abundant laboratory and clinical
evidence attests to the fact that coadministration of drugs reduces the emergence
of resistance. As detailed above, this strategy has provided a useful antimalarial
therapeutic life span for pyrimethamine/sulfadoxine and atovaquone/proguanil,
agents that readily provoke resistance when used alone. To stem the further development and spread of antimalarial drug resistance, the combined use of three or
more drugs is under extensive study, and will likely succeed in reducing resistance
(119). Less easy to predict is how multiple agents will interact in terms of antimalarial potency and host toxicity, where the net effects may be additive, synergistic, or
even antagonistic. Distinguishing among these important outcomes requires careful attention to study design. Investigational combinations include coartemether,
a fixed dose of artemether and lumefantrine; the latter has structural similarities
to mefloquine and halofantrine. This combination originated in China and is in
advanced clinical development (120). The combination of dihydroartemisinin and
piperaquine has been evaluated in patients from Cambodia with uncomplicated falciparum malaria (121). Amodiaquine combined with sulfadoxine/pyrimethamine
4 Dec 2004
18:51
AR
AR232-PA45-23.tex
XMLPublishSM (2004/02/24)
P1: JRX
577
had substantial antimalarial activity in spite of preexisting resistance to each component drug (122).
CONCLUDING REMARKS
In recent years the severe problem of drug-resistant malaria has been featured
extensively in the scientific and lay literature, leading to increased public awareness, new and better funding opportunities for research, and a growing sense that
the situation requires thoughtful public health policies to preserve the utility of
current therapies. Spurred by powerful genetic tools and availability of the fully
sequenced genome, effective new drugs will almost certainly be discovered. Less
certain is whether these agents will be inexpensive enough for widespread use in
developing countries. Also of obvious concern is the propensity for resistance,
which atovaquone has taught can appear immediately and at high levels. It is interesting to speculate that would-be new antimalarial drugs might better have a
nonprotein target (e.g., chloroquine against the growing hemozoin crystal) or an
irrational molecular mechanism (e.g., the artemisinins whose activated free radicals may pose a nonspecific oxidative stress). Although the pathway to design
such agents prospectively is less obvious than, for example, that for an enzyme
inhibitor, they may be inherently less affected by point mutations, which are the
4 Dec 2004
18:51
578
AR
AR232-PA45-23.tex
ARAV-BOGER
XMLPublishSM (2004/02/24)
P1: JRX
SHAPIRO
ACKNOWLEDGMENTS
We apologize to our many colleagues whose work was not directly cited because
of strict space limitations, and thank Tom Kulikowicz and Rahul Bakshi for their
generous help with preparing the figures. Our work has been supported by the Johns
Hopkins Malaria Research Institute (TS), the Johns Hopkins Clinician Scientist
Award (RB), and the National Institutes of Health (RR-00052).
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Krogstad DJ. 2000. Plasmodium species
(Malaria). See Ref. 133, pp. 281832
2. Nchinda TC. 1998. Malaria: a reemerging disease in Africa. Emerg. Infect. Dis.
4:398403
3. Wellems TE, Miller LH. 2003. Two
worlds of malaria. N. Engl. J. Med. 349:
149698
4. Gardner MJ, Hall N, Fung E, White
O, Berriman M, et al. 2002. Genome
sequence of the human malaria parasite Plasmodium falciparum. Nature
419:498511
5. White NJ. 2004. Antimalarial drug resistance. J. Clin. Invest. 113:108492
6. Talisuna AO, Bloland P, dAlessandro
U. 2004. History, dynamics, and public
health importance of malaria parasite resistance. Clin. Microbiol. Rev. 17:235
54
7. Rosenthal PJ. 2003. Antimalarial drug
discovery: old and new approaches. J.
Exp. Biol. 206:373544
8. Wongsrichanalai C, Pickard AL, Wernsdorfer WH, Meshnick SR. 2002. Epidemiology of drug-resistant malaria. Lancet Infect. Dis. 2:20918
9. Hyde JE. 2002. Mechanisms of resistance
10.
11.
12.
13.
14.
15.
16.
4 Dec 2004
18:51
AR
AR232-PA45-23.tex
XMLPublishSM (2004/02/24)
P1: JRX
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
579
4 Dec 2004
18:51
580
39.
40.
41.
42.
43.
44.
45.
46.
47.
AR
AR232-PA45-23.tex
ARAV-BOGER
XMLPublishSM (2004/02/24)
P1: JRX
SHAPIRO
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
4 Dec 2004
18:51
AR
AR232-PA45-23.tex
XMLPublishSM (2004/02/24)
P1: JRX
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
581
4 Dec 2004
18:51
582
78.
79.
80.
81.
82.
83.
84.
AR
AR232-PA45-23.tex
ARAV-BOGER
XMLPublishSM (2004/02/24)
P1: JRX
SHAPIRO
4 Dec 2004
18:51
AR
AR232-PA45-23.tex
XMLPublishSM (2004/02/24)
P1: JRX
102.
103.
104.
105.
106.
107.
108.
109.
110.
111.
112.
583
www.cdc.gov/travel/diseases/malaria/ind
ex.htm
Drugs for parasitic infections. 2004. Med.
Lett. Drugs Ther. August:112
Fivelman QL, Butcher GA, Adagu IS,
Warhurst DC, Pasvol G. 2002. Malarone
treatment failure and in vitro confirmation
of resistance of Plasmodium falciparum
isolate from Lagos, Nigeria. Malar. J. 1:1
Schwartz E, Bujanover S, Kain KC. 2003.
Genetic confirmation of atovaquoneproguanil-resistant Plasmodium falciparum malaria acquired by a nonimmune
traveler to East Africa. Clin. Infect. Dis.
37:45051
Lage H. 2003. ABC-transporters: implications on drug resistance from microorganisms to human cancers. Int. J. Antimicrob. Agents 22:18899
Syafruddin D, Asih PB, Aggarwal SL,
Shankar AH. 2003. Frequency distribution of antimalarial drug-resistant alleles among isolates of Plasmodium falciparum in Purworejo district, Central Java
Province, Indonesia. Am. J. Trop. Med.
Hyg. 69:61420
Rathod PK, McErlean T, Lee PC. 1997.
Variations in frequencies of drug resistance in Plasmodium falciparum. Proc.
Natl. Acad. Sci. USA 94:938993
Meshnick SR. 2001. Artemisinins and its
derivatives. See Ref. 134, pp. 191202
Borstnik K, Paik IH, Shapiro TA, Posner
GH. 2002. Antimalarial chemotherapeutic peroxides: artemisinin, yingzhaosu A
and related compounds. Int. J. Parasitol.
32:166167
Eckstein-Ludwig U, Webb RJ, Van
Goethem ID, East JM, Lee AG, et al. 2003.
Artemisinins target the SERCA of Plasmodium falciparum. Nature 424:95761
Nosten F, Luxemburger C, ter Kuile FO,
Woodrow C, Eh JP, et al. 1994. Treatment of multidrug-resistant Plasmodium
falciparum malaria with 3-day artesunatemefloquine combination. J. Infect. Dis.
170:97177
Geary TG, Jensen JB. 1983. Effects of
4 Dec 2004
18:51
584
113.
114.
115.
116.
117.
118.
119.
120.
121.
AR
AR232-PA45-23.tex
ARAV-BOGER
XMLPublishSM (2004/02/24)
P1: JRX
SHAPIRO
122.
123.
124.
125.
126.
127.
128.
129.
4 Dec 2004
18:51
AR
AR232-PA45-23.tex
XMLPublishSM (2004/02/24)
P1: JRX
585
HI-RES-PA45-23-Shapiro.qxd
1/7/05
2:30 PM
Page 1
C-1
4 Dec 2004
18:52
AR
AR232-PA45-24.tex
XMLPublishSM (2004/02/24)
P1: JRX
10.1146/annurev.pharmtox.45.120403.095807
INTRODUCTION
Enormous progress has been made during the past decade identifying sets of molecular interactions that transmit information between different parts of the cell. The
increasing number of databases containing lists of interacting biomolecules has
sparked the development of the burgeoning field of network biology and with it
the realization that, to a first approximation, biological networks can be described
mathematically by scale-free power functions (1). These functions predict the hierarchical interconnectedness of molecules through their participation in different
classes of interacting units such as nodes, hubs, modules, and motifs. Power functions mathematically describe the organization of thermodynamically far from
equilibrium systems like living organisms. The network structure derived from
this type of abstract analysis is very useful for understanding the patterns created
by interconnecting the molecular constituents of different signaling pathways. Understanding exactly how these molecular interactions become determinants of cell
structure and function, however, remains a significant challenge. The molecular
interactions underlying biological networks take place in living cells, and network
analysis inherently is unable to consider the contribution of different intracellular
environments to signal transduction. Therefore, an important next step is to develop a high resolution map of signaling networks in living cells and the location
0362-1642/05/0210-0587$14.00
587
4 Dec 2004
18:52
588
AR
WHITE
AR232-PA45-24.tex
XMLPublishSM (2004/02/24)
P1: JRX
ANDERSON
of interacting signaling units (i.e., hubs, motifs, modules) relative to cell structures
like the plasma membrane, mitochondria, the nucleus, etc.
Cell biologists use many techniques to map the distribution of molecules in cells.
Cell fractionation as well as light and EM immunocytochemistry are the principal
methods that have been used to demonstrate that cell signaling molecules tend to
be concentrated in different cellular compartments. The compartmentalization of
interacting sets of signaling molecules has several implications for understanding
signaling networks in situ. First, these compartments often can be isolated in a
way that preserves the functionality of the resident signaling units. They contain
dynamic information about the behavior of molecules that make up specific signaling networks, and embedded in the pattern of molecular interactions are the codes
that govern cell behavior. Compartments also contain the molecular signature of
unknown signaling pathways that cannot be detected using ex vivo techniques.
For example, current estimates indicate that the human genome contains vastly
more signaling molecules than have been classified and assigned to pathways.
Determining the compartment where these molecules reside is a valuable first
step in identifying, mapping, and characterizing their function. Another important consideration is that similar sets of signaling units can be found in different
compartments, although the same class of compartments at other times contains
different sets of signaling units. Nothing is known about the rules that control the
compartmentalization of signaling units, nor how the spatial distribution of these
units and the environment created by the host compartment influences signal transduction. Deciphering the rules of compartmentalization can only be achieved by
studying the function of signaling units when they are in different host compartments. Each type of compartment is spatially restricted, so compartmentalization
also contributes to the spatial organization of signal transduction in the cell. A
final consideration is that the mechanism of action of a signaling molecule in a
compartment cannot be predicted simply by knowing all its interacting partners.
Compartmentalized sets of signaling molecules display emergent behavior that
can only be understood by studying the entire ensemble of molecules interacting
in their native environment. The fidelity of signal transduction depends as much
on the molecular ecology of the compartment as it does on the interaction between
individual signaling molecules.
4 Dec 2004
18:52
AR
AR232-PA45-24.tex
XMLPublishSM (2004/02/24)
P1: JRX
589
pathways, creating a situation in which diverse inputs (S1, S2, S312, Figure 1A,
see color insert) can transmit through a limited number of core transducers (E1,
E2, E3, Figure 1A) to multiple outputs (R1, R2, R3, Figure 1A). The apparent
complexity of this organization raises questions about how these signaling units
are able to establish high fidelity coupling between diverse stimuli and discrete
physiological effects in the living cell. Moreover, the identity of interacting signaling units in a network does not reveal how these units are spatially and temporally
organized in the cell. The ERK1/2 MAP kinase cascade is a well-studied signaling
pathway that illustrates the dilemma.
The ERK1/2 protein kinases are a core signaling element implicated in the
regulation of assorted biological processes, ranging from cellular proliferation and
tumorigenesis to differentiation and cell specialization (2, 3). ERK1/2 apparently
function as the terminal kinase in a three-kinase cascade that includes the Mek MAP
kinase kinase (MAP2K) and the Raf MAP kinase kinase kinase (MAP3K). This
linked set of protein kinases is a signal propagation cassette (E1, E2, E3, Figure 1A)
typical of many signaling kinase families. ERK1/2 activation can be stimulated by
growth factor receptors, heterotrimeric G-protein coupled receptors, and integrins.
Many ERK1/2 effector proteins (R1, R2, R3, Figure 1A) have been identified,
including nuclear transcription factors like c-Fos and Elk-1 (4, 5), cytoplasmic
protein kinases like p90RSK (6, 7) and myosin light chain kinase (8), and lipases
like phospholipase A2 (9). Moreover, recent proteomic studies identified 20 new
targets for this kinase that are involved in such diverse activities as nuclear import,
nucleotide excision repair, membrane traffic, and cytoskeleton assembly (10). The
pleiotropic consequences of ERK1/2 activation imply that activated ERK1/2 is
directly connected to many different targets.
The ERK1/2 protein kinase cascade is functionally coupled to stimuli in part
by Ras family small GTPases (3). Ras proteins are tethered to membranes rich
in sensory receptors through carboxy-terminal lipidation (11). The bulk of the
kinase elements in the ERK1/2 kinase cascade, by contrast, are in the cytosol of
unstimulated cells. In response to stimulus, Ras proteins are activated through
GDP/GTP exchange as a consequence of receptor-driven association with guanyl
nucleotide exchange factors (GEFs). Ras-GTP adopts a conformation that favors
the direct interaction with downstream effector proteins, including Raf kinases
(11). Recent structural and biochemical analysis of Ras GEFs (12), together with
single-molecule imaging of activated Ras (13), suggests that these enzymes function processively by generating interactive surfaces in the activated state; these
surfaces form signaling platforms for combinatorial sets of protein-protein interactions. Normally, activation of Ras recruits Raf-1 to the plasma membrane, but Raf-1
artificially targeted to membranes causes activation of ERK1/2 independently of
Ras. Therefore, the current paradigm for Raf-Mek-Erk pathway activation suggests
that Ras-GTP acts as molecular flypaper that snares Raf kinases at the plasma
membrane where they are subsequently activated by other membrane-associated
components that have not yet been defined [reviewed in Kolch (14)]. Raf in turn
mediates activation of MEK and ERK through a linear cascade of kinase/substrate
4 Dec 2004
18:52
590
AR
WHITE
AR232-PA45-24.tex
XMLPublishSM (2004/02/24)
P1: JRX
ANDERSON
interactions (Figure 1A). Although this model partially explains how various
signals are linked to ERK1/2 activation, it does not clarify how activated ERK1/2
is tuned to the diverse signaling targets it controls.
A growing number of observations suggest that scaffolding proteins (M1, M2,
M3, Figure 1B) can selectively couple ERK1/2 activation to distinct regulatory
programs. Genetic screens for modifiers of the Ras-Raf-Mek-ERK cascade, together with protein/protein interaction studies, have identified proteins like KSR,
CNK, MP-1, and Sur-8 that possess no obvious intrinsic enzymatic activities but
physically interact with multiple core components of the ERK1/2 cascade (15).
These scaffolds appear to be obligate components of ERK1/2 signaling modules
(1618), are required for ERK1/2 activation in cells (1921), and contribute to
the functional coupling of the ERK1/2 cascade to selective stimuli (20, 22). For
example, CNK can interact directly with Raf kinases (17), is required for Raf activation in response to insulin in insect cells (21), and functions at least in part to
partition Raf into membrane compartments (21). In mammalian cells, the CNK
family member CNK2 may help neuronal precursor cells distinguish between neurotrophic and proliferative signals. For example, CNK2 is required for activation
of ERK1/2 by TrkA receptor signaling but not EGF receptor signaling (20). The
ERK1/2 scaffold Sur-8, by contrast, appears to be more important for EGF receptor signaling (22). These observations suggest that scaffold proteins mediate the
assembly of signaling modules that are selectively coupled to discrete receptor
inputs (Figure 1B).
An additional mechanism for establishing specific input-output connections for
these signaling modules may be to spatially segregate each module into a different
cell compartment. Thus, the higher-ordered molecular organization generated by
scaffold proteins may also function to target the kinases (E1, E2, E3) to a specific
location in the cell (Figure 1C). This hypothesis would require address information
on the scaffolding protein, or on some component of the module, that would target
it to a compartment, thereby creating spatially restricted signaling activity. There
is considerable evidence that the Raf-Mek-Erk1/2 modules can signal from multiple cellular compartments (Figure 1C), including the late endosomes (M1) (23),
caveolae (M2) (24), and the Golgi apparatus (M3) (25). Ras family GTPases may
control, in part, the localization of each module. These GTPases carry autonomous
address information at the carboxy-terminus specified by a pattern of methylation,
prenylation, and palmitoylation that controls the targeting of the protein to a distinct membrane domain (11). The scaffold associated with the kinase cascade may
also control compartmentalization. CNK contains a pleckstrin homology domain
that mediates association with membrane phosphoinositides, which may mediate
the association of insect Raf with membrane fractions (17, 26). KSR carries a
cysteine-rich motif that can mediate membrane association, perhaps through interactions with phosphatidylserine (27). Finally, MP-1, a potential scaffold for
MEK1 and MEK2, is required for the localization of ERK1/2 to late endosomes
via an interaction with the resident endosomal protein p14 (23).
From this brief analysis of signaling through the ERK1/2 kinase cascade, we
conclude that compartmentalization is a fundamental component of signaling
4 Dec 2004
18:52
AR
AR232-PA45-24.tex
XMLPublishSM (2004/02/24)
P1: JRX
591
network design. In these compartments, signaling is organized and local environmental factors exert control. An important strategy for understanding signal
transduction in cells, therefore, will be to devise methods to probe the functionality of signaling networks in the compartment where they normally reside.
Caveolae as a Compartment
Cell compartments have a distinctive morphology and dynamics that are important for understanding the functionality of the signaling units they contain. Like
all compartments, caveolae are constructed and maintained by specialized cellular
machinery and exhibit characteristic behaviors that define their lifetime functions.
They typically are recognized as flask-shaped membrane invaginations (31) decorated with a coat protein called caveolin-1 (32) and are best known as endocytic
organelles that internalize specific classes of molecules. There appear to be two
distinctive modes of internalization (Figure 2, see color insert). Some caveolae
(Type 1) invaginate much the same as clathrin-coated pits do and pinch off from
the membrane using dynamin to complete the fission step (33, 34). These caveolae
are able to travel to the interior of the cell. Other caveolae (Type 2) become deeply
invaginated to the point where they are functionally sealed off from the extracellular space but remain associated with the plasma membrane. These caveolae open
and close without ever leaving the vicinity of the cell surface. Type 1 and Type 2
caveolae can be distinguished by their ligand internalization patterns. For example, uptake of folate by the GPI-anchored folate receptor involves caveolae that
open and close in a 1 hr cycle, without ever leaving the vicinity of the plasma
membrane, by a process called potocytosis (35). Internalization of SV40 virus,
by contrast, depends on caveolae that pinch off from the plasma membrane and
travel to the cell interior (36). Inhibiting either PKC (36, 37) or tyrosine kinase
activity (36, 38) blocks both types of internalization, although uptake by coated
pits is unaffected.
The vesicles produced by caveolae (39) during endocytosis (cavicles) are impossible to identify by EM unless they are loaded with recognizable cargo. The
introduction of caveolin-GFP (green fluorescent protein) has made it possible for
the first time to study caveolae membrane traffic in detail. Unexpectedly, three
4 Dec 2004
18:52
592
AR
WHITE
AR232-PA45-24.tex
XMLPublishSM (2004/02/24)
P1: JRX
ANDERSON
different kinds of traffic can be detected in tissue culture cells. The most common
pattern (up to 75%) is a sessile behavior where the caveolin-GFP positive membrane appears to be firmly anchored at the cell surface (40). Sessile caveolin-GFP is
also concentrated at the cleavage furrow of dividing cells (41). Relatively immotile
caveolin-GFP is the expected behavior of a Type 2 caveola (Figure 2). A second
behavior (Figure 2) is a rapid bidirectional, microtubule-dependent movement of
caveolin-GFP positive vesicles (cavicles) between the center of the cell and the
cell surface (39). This movement most likely corresponds to Type 1 caveolae that
have budded from the membrane. In the case of CHO cells, cavicles appear to be
traveling to the recycling endosome (42), although it is not clear if they fuse with
this compartment. By contrast, cavicles carrying SV40 virus travel to a special
endocytic compartment called the caveosome (36). The third type of movement
detected with caveolin-GFP is the projection and retraction of fine tubular elements that can extend from the plasma membrane to the center of the cell (Figure
2). Recently these tubules have been captured in EM images of cells internalizing
protein A-gold bound to prions (43). Because prions are also concentrated in flaskshaped caveolae (44), the tubular elements, designated Type 3 caveolae (Figure 2),
may be derived from either Type 1 or 2 caveolae. To the extent that caveolin-GFP
marks caveolae and cavicles, there appears to be a high degree of plasticity to the
movement of caveolae-derived membranes. There are even instances in which entire sheets of caveolae membrane appear to internalize en masse to form internal,
endosome-like structures (39), which may be how certain bacterial pathogens are
internalized (45).
4 Dec 2004
18:52
AR
AR232-PA45-24.tex
XMLPublishSM (2004/02/24)
P1: JRX
593
4 Dec 2004
18:52
594
AR
WHITE
AR232-PA45-24.tex
XMLPublishSM (2004/02/24)
P1: JRX
ANDERSON
between neuronal and non-neuronal cells (62). Here we focus the discussion on
three of these activities.
Receptor tyrosine kinases such as the PDGF receptor (24), the EGF receptor
(63), and insulin receptor (64) have been localized to caveolae using cell fractionation, immunocytochemistry, or caveolin-1 interaction. Investigators have used
several different experimental protocols to show that these tyrosine kinases are
linked to signaling units in caveolae. One successful approach demonstrates that
these receptors are coupled to other signaling molecules in caveolae in vivo. For
example, binding of PDGF to PDGFR in caveolae stimulates the phosphorylation of multiple caveolar substrates (65) and silences EGFR phosphorylation in
response to EGF (66). By contrast, EGF causes the recruitment of Raf-1 kinase to
caveolae where it is activated (67), and it stimulates the local generation of inositol trisphosphate from a pool of caveolar phosphatidylinositol 4,5-bisphosphate
(PtdIns 4,5-P2 ) (68). The general protocol for these experiments is to expose the
cells to the growth factor, prepare caveolae fractions, and assay for changes that
occur in this membrane fraction but not in noncaveolae fractions.
An extension of this method is to assay for the presence of functional signaling
units in isolated caveolae. The first test of this technique found that PDGFR was
functionally linked to the activation of the MAP kinase ERK1/2 in isolated caveolae (24). The interaction of as many as 11 different molecules (PDGF, PDGFR,
SOS, Ras, Raf-1, Grb2, SHC, 143-3, MEK-1, and ERK1/2) can be involved in
activating ERK, so all the members of this signaling unit must be preorganized in
the caveolae membrane, because nothing else was added to the preparation except
PDGF. Indeed, immunoblotting has documented that many of these molecules are
enriched in caveolae fractions (65).
Finding functional signaling units in caveolae fractions indicates that the isolated compartment can retain the cellular complexity necessary to study the natural
switching and branching that occurs between signaling units in the living cell. Recent studies on activation of eNOS support this reasoning (69). eNOS is targeted
to caveolae by an N-terminal acylation motif (59) and in this location can be activated by a number of different humoral and mechanical stimuli including estradiol,
bradykinin, VEGF, HDL, isometric vessel contraction, and shear stress. The linkage between the stimulus and the activation of eNOS depends on the interaction
of many cofactors and connectors, including nonreceptor tyrosine kinases, calcium, heterotrimeric G proteins, PI3 kinase, Akt kinase, ERK1/2, PKC, PKA,
calmodulin, and HSP90. Many of these molecules and ions have been localized to
caveolae. More important, the connectivity between these molecules is preserved
in isolated caveolae. Incubation of isolated endothelial cell caveolae in the presence
of estradiol (70), bradykinin (70), acetylcholine (70), or HDL (71) all stimulate
eNOS enzymatic activity (Figure 3A), although these ligands have no effect on
isolated noncaveolae membrane (Figure 3B). Thus, the natural organization of this
signaling pathway is preserved so well that caveolae eNOS can be activated by
four different ligands, each binding a receptor that is wired to eNOS through a
distinct set of connecting molecules (72).
4 Dec 2004
18:52
AR
AR232-PA45-24.tex
XMLPublishSM (2004/02/24)
P1: JRX
595
Figure 3
The eNOS signaling unit is located in caveolae. L-[3 H]-arginine conversion
to L-[3 H]-citrulline was measured in caveolae (A), and noncaveolae (B) membrane
fractions isolated from endothelial cells by the method of Smart et al. (48) and incubated
in the absence (B) or presence of 108 M estradiol (E2 ), 106 M acetylcholine (Ach),
or 106 M bradykinin (BK). Values are mean +/ SEM (n = 46), p < 0.05 relative
to basal. [Modified from (70).]
Probing the functionality of signaling units in isolated compartments will uncover unexpected molecular connections that will need to be verified in the living
cell. A major technical advance that makes this possible is the development of designer fluorescence resonance energy transfer (FRET) probes capable of detecting
specific signaling pathways in live cells (73). The typical probe is a chimeric protein consisting of a donor and an acceptor GFP, which have matched overlapping
excitation and emission spectra, connected together by a sensor that is designed to
bind a specific ionic or molecular intermediate in a signaling pathway (74). When
the sensor binds the molecule or ion of interest it undergoes a conformational
change that adjusts the distance between the two GFPs (e.g., cyan fluorescent
protein and yellow fluorescent proteins). When the two GFPs are close together,
intermolecular FRET occurs. Therefore, the emission ratio before and after cell
stimulation is a relative measure of how much of the signaling molecule or ion is
in the vicinity of the probe. If the probe is targeted to a cell compartment, then the
probe will record signal transduction at that location.
The calcium sensor yellow cameleon is an example of a FRET probe that has
been successfully used to monitor signal transduction from caveolae in living cells
(75). There is considerable evidence that caveolae contain the molecular machinery
for sensing [Ca2+ ] (76, 77). Yellow cameleon was used to monitor the dynamics
of [Ca2+ ] in endothelial cells. To do this, the cameleon was targeted either to the
cytoplasm, the plasma membrane, or caveolae. The internal ER Ca2+ stores were
4 Dec 2004
18:52
596
AR
WHITE
AR232-PA45-24.tex
XMLPublishSM (2004/02/24)
P1: JRX
ANDERSON
then depleted and the relative [Ca2+ ] detected by each cameleon from the respective
sites was recorded as the concentration of extracellular Ca2+ was increased. The
comparative response of the probe to being in the three different locations (i.e.,
compartments) indicates that caveolae are preferred sites of Ca2+ entry and that
the entering Ca2+ is linked to the activation of eNOS. FRET-based probes promise
to be extremely useful for mapping signaling networks in live cells.
4 Dec 2004
18:52
AR
AR232-PA45-24.tex
XMLPublishSM (2004/02/24)
P1: JRX
597
4 Dec 2004
18:52
598
AR
WHITE
AR232-PA45-24.tex
XMLPublishSM (2004/02/24)
P1: JRX
ANDERSON
CONCLUSION
This analysis of caveolae illustrates how signal transduction is organized in a
compartment that controls important spatial and temporal parameters necessary
for fidelity in intracellular signaling. There are strong indications that other cellular
compartments attract specific sets of signaling molecules and modules, so most
likely, compartmentalization is a general way cells organize signaling networks.
Some of the questions that emerge from the current analysis include (a) How can
the same compartment contain operationally different sets of signaling molecules?
(b) How do the different molecular ecologies of compartments affect the input and
output signals of the same signaling unit? (c) What are the rules for targeting
signaling units to specific compartments? (d) Are interacting sets of signaling
units always in the same compartment of each cell type, or do they move around?
(e) What are the thermodynamic rules for how signaling networks are superimposed on to the architecture of the cell? Clearly, a systems biology approach to
understanding cell structure and function will require answers to these and many
similar questions.
ACKNOWLEDGMENTS
We would like to thank Brenda Pallares for administrative assistance. Some of
the work cited in this report was supported by grants from the National Institutes of Health, HL 20948, GM 52016 (RGWA), and CA71443 (MAW); Robert
Welch Foundation I-1414; the Perot Family Foundation; and the Cecil H. Green
Distinguished Chair in Cellular and Molecular Biology.
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Barabasi AL, Oltvai ZN. 2004. Network biology: understanding the cells functional
organization. Nat. Rev. Genet. 5:10113
2. Zhu JJ, Qin Y, Zhao M, Van Aelst L, Malinow R. 2002. Ras and Rap control AMPA
receptor trafficking during synaptic plasticity. Cell 110:44355
3. Pearson G, Robinson F, Beers Gibson
T, Xu BE, Karandikar M, et al. 2001.
Mitogen-activated protein (MAP) kinase
pathways: regulation and physiological
functions. Endocr. Rev. 22:15383
4. Gille H, Kortenjann M, Thomae O,
Moomaw C, Slaughter C, et al. 1995.
4 Dec 2004
18:52
AR
AR232-PA45-24.tex
XMLPublishSM (2004/02/24)
P1: JRX
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
599
4 Dec 2004
18:52
600
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
AR
WHITE
AR232-PA45-24.tex
XMLPublishSM (2004/02/24)
P1: JRX
ANDERSON
40.
41.
42.
43.
44.
45.
46.
47.
48.
4 Dec 2004
18:52
AR
AR232-PA45-24.tex
XMLPublishSM (2004/02/24)
P1: JRX
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
601
man Z, Yuhanna IS, et al. 1996. Acylation targets endothelial nitric-oxide synthase to plasmalemmal caveolae. J. Biol.
Chem. 271:651822
Rechardt L, Hervonen H. 1985. Cytochemical demonstration of adenylate cyclase activity with cerium. Histochemistry 82:501
5
Sugi H, Suzuki S, Daimon T. 1982. Intracellular calcium translocation during
contraction in vertebrate and invertebrate
smooth muscles as studied by the pyroantimonate method. Can. J. Physiol. Pharmacol. 60:57687
Anderson RG. 1993. Caveolae: where incoming and outgoing messengers meet.
Proc. Natl. Acad. Sci. USA 90:1090913
Mineo C, Gill GN, Anderson RG. 1999.
Regulated migration of epidermal growth
factor receptor from caveolae. J. Biol.
Chem. 274:3063643
Saltiel AR, Pessin JE. 2003. Insulin signaling in microdomains of the plasma membrane. Traffic 4:71116
Liu P, Ying Y, Ko YG, Anderson RG. 1996.
Localization of platelet-derived growth
factorstimulated phosphorylation cascade
to caveolae. J. Biol. Chem. 271:10299303
Liu P, Anderson RG. 1999. Spatial organization of EGF receptor transmodulation by
PDGF. Biochem. Biophys. Res. Commun.
261:695700
Mineo C, James GL, Smart EJ, Anderson RG. 1996. Localization of epidermal growth factorstimulated Ras/Raf-1
interaction to caveolae membrane. J. Biol.
Chem. 271:1193035
Pike LJ, Casey L. 1996. Localization
and turnover of phosphatidylinositol 4,5bisphosphate in caveolin-enriched membrane domains. J. Biol. Chem. 271:26453
56
Shaul PW. 2002. Regulation of endothelial
nitric oxide synthase: location, location, location. Annu. Rev. Physiol. 64:74974
Chambliss KL, Yuhanna IS, Mineo C, Liu
P, German Z, et al. 2000. Estrogen receptor alpha and endothelial nitric oxide
4 Dec 2004
18:52
602
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
AR
WHITE
AR232-PA45-24.tex
XMLPublishSM (2004/02/24)
P1: JRX
ANDERSON
4 Dec 2004
18:52
AR
AR232-PA45-24.tex
XMLPublishSM (2004/02/24)
P1: JRX
603
1/7/05
2:31 PM
Figure 1 Schematic representation of how high fidelity signal transduction machines can be generated through modular organization of commonly engaged signaling proteins. (A) Representation of the biochemical relationships in a regulatory network
with multiple distinct inputs (stimuli S1, S2, and S3) and outputs (responses R1, R2, and R3) that propagate through a common
core enzymatic cascade (enzymes E1, E2, and E3). (B) Non-enzymatic accessory proteins (M1, M2, and M3) functionally segregate the core enzymatic cascade into separate modules with discrete input/output relationships. (C) Selective compartmentalization may restrict/facilitate coupling of spatially discrete stimuli to distinct signaling modules thereby generating fidelity
among stimulus/response pathways.
HI-RES-PA45-24-White.qxd
Page 1
HI-RES-PA45-24-White.qxd
C-2
WHITE
1/7/05
2:31 PM
Page 2
ANDERSON
Figure 2 Multiple pathways of caveolae traffic. Type 1 caveolae are able to invaginate and bud from the membrane, probably in a dynamin-dependent process (33, 34).
The vesicles that form, called cavicles, are able to travel on microtubules to various
endosomal compartments. Cavicles also can travel from endosomes to other places
in the cell. Type 2 caveolae invaginate and seal off from the plasma membrane but
are retained at the surface by the actin cytoskeleton. We imagine that type 3 caveolae begin as membrane invaginations similar to the other types but then get caught on
microtubules and become stretched by microtubule motor activity into tubules.
(Diagram adapted from 39.)
HI-RES-PA45-24-White.qxd
1/7/05
2:31 PM
Page 3
HI-RES-PA45-24-White.qxd
C-4
WHITE
1/7/05
2:31 PM
Page 4
ANDERSON
4 Dec 2004
19:4
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
10.1146/annurev.pharmtox.45.120403.095906
Drug Targets
Sophie Lotersztajn,1 Boris Julien,1 Fatima Teixeira-Clerc,1
Pascale Grenard,1 and Ariane Mallat1,2
1
INTRODUCTION
Chronic liver injury produces liver fibrosis, and its endstage, cirrhosis, is a major
public health problem worldwide owing to life-threatening complications of portal
hypertension and liver failure and to the risk of incident hepatocellular carcinoma.
A variety of adverse stimuli may trigger fibrogenesis, including viruses, toxins
such as alcohol, autoimmune diseases, chronic biliary stasis, metabolic disorders,
genetic defects, or hypoxia. In western countries, the prevailing causes of cirrhosis include chronic alcohol consumption, hepatitis C virus, and nonalcoholic
steatohepatitis. Current treatment of hepatic fibrosis is limited to withdrawal of
the noxious agent, which not only prevents fibrosis progression but may also induce its regression, as discussed below. Major advances have been made in this
respect during the past decade, with the advent of efficient antiviral treatments for
hepatitis B and C. Nevertheless, suppression of the cause of hepatic injury is not
0362-1642/05/0210-0605$14.00
605
4 Dec 2004
19:4
606
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
LOTERSZTAJN ET AL.
always feasible, and, therefore, numerous efforts are directed at the development
of liver-specific antifibrotic therapies. Although effective antifibrotic treatments
are not available as yet, several ongoing clinical trials are evaluating molecules
identified from the joint efforts of many researchers. In addition, recent advances
in the physiopathology of liver fibrosis are paving the way for the design of new
molecules interfering with regulatory pathways in fibrogenic cells. This review
highlights recent advances in the molecular mechanisms of liver fibrosis and discusses mechanistically based strategies that have emerged recently.
4 Dec 2004
19:4
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
607
cell populations, known as activated hepatic stellate cells and hepatic myofibroblasts (10, 11).
In the normal liver, hepatic stellate cells compose 5% to 10% of cells and
are located in the subendothelial space between hepatocytes and sinusoidal endothelial cells. Following acute or chronic liver diseases, they undergo phenotypic
changes, switching from a quiescent vitamin A-rich phenotype to a myofibroblastic phenotype (referred as to activated HSC) (12). Activated hepatic stellate cells
show de novo fibrogenic properties, including proliferation and accumulation in
areas of parenchymal cell necrosis, secretion of proinflammatory cytokines and
chemokines, and synthesis of a large panel of matrix proteins and of inhibitors of
matrix degradation, leading to progressive scar formation (Figure 1).
Hepatic myofibroblasts are another source of fibrogenic cells that derive from
fibroblasts of the portal connective tissue, perivascular fibroblasts of portal and
central veins, and periductular fibroblasts in close contact with bile duct epithelial
cells. Contribution of these cells to fibrogenesis was initially demonstrated in
experimental biliary cirrhosis by showing that myofibroblastic transformation of
Figure 1
4 Dec 2004
19:4
608
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
LOTERSZTAJN ET AL.
4 Dec 2004
19:4
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
609
hepatic stellate cells cultured in a three-dimensional matrix of collagen I or matrigel retain a quiescent vitamin-A rich phenotype (22). In contrast, induction of
matrix degradation is rapidly associated with acquisition of the myofibroblastic
phenotype. Several lines of evidences also indicate that adhesion molecules are
important mediators of matrix-induced activation of hepatic stellate cells (23).
Chemotaxis
Migration of fibrogenic cells toward injured areas may contribute to their accumulation at sites of injury. Migration is promoted by growth factors (e.g., PDGF,
FGF-2) or chemokines (MCP-1, CCl21) produced by inflammatory cells and involving the PI3 kinase pathway (23, 34).
4 Dec 2004
19:4
610
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
LOTERSZTAJN ET AL.
Fibrogenesis
The profibrogenic potential of activated hepatic stellate cells and hepatic myofibroblasts is due to their capacity to synthesize fibrotic matrix proteins and components that inhibit fibrosis degradation. Among the large number of factors identified
as activators of matrix production, TGF-, CTGF (35), and leptin (36) play a major
role.
Hepatic stellate cells express a wide range of metalloproteinases (MMPs) as well
as MMP activators that cleave pro-MMP into their active form. In addition, they
also produce specific tissue inhibitors of the metalloproteinase family (TIMPs).
Production of MMPs and TIMPs is tightly regulated according to the activation
state of hepatic stellate cells, and it reflects extracellular matrix remodeling during
chronic liver injury. At early stages, hepatic stellate cells express MMP-1, MMP-2,
MMP-3, and MMP-9 and their activators, but do not produce TIMPs; this allows degradation of normal matrix in the subendothelial space and its substitution
by fibrillar collagens. In contrast, fully activated hepatic stellate cells shut down
expression of MMPs and turn on expression of TIMPs, resulting in a dramatic
reduction of collagenolytic activity within the liver (37).
Strikingly, a number of cytokines simultaneously govern several functions of
fibrogenic cells. Thus, TGF-, interleukin-1, and leptin promote stellate cell activation, enhance collagen synthesis, and markedly induce TIMP-1. In addition,
TGF- also promotes cell survival (38).
4 Dec 2004
19:4
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
611
cells that spontaneously undergo apoptosis (10). However, this hypothesis needs to
be explored by expression profiling of passaged cells. Therefore, in the following
sections, we refer to activated hepatic stellate cells or hepatic myofibroblasts, as
stated in the publications. Other culture models include rodent or human hepatic
stellate cell lines with myofibroblastic features obtained either spontaneously or
by transfection of the coding region of SV-40 (12). However, the relevance of these
models to the in vivo situation is questionable.
Animal Models
Rodent fibrosis models are widely used because of their convenient time frame.
Features of the fibrogenic process depend on the nature of liver injury. Compounds
such as carbon tetrachloride, dimethylnitrosamine, or galactosamine generate significant hepatocyte necrosis, associated with marked inflammation. In these models, antifibrotic effects of tested drugs may therefore result either from a direct
effect on fibrogenic cells or from nonspecific antiinflammatory effects. Therefore,
additional models with low degrees of cell damage and inflammation, such as bile
duct ligation or thioacetamide administration, should be used in parallel to validate efficiency of an expected antifibrotic molecule. It should also be stressed that
models of fibrosis recovery after cessation of chronic tetrachloride intoxication (2)
or following biliodigestive anastomosis in bile duct ligated rats (25) have proved
useful recently for the study of curative antifibrotic effects.
4 Dec 2004
19:4
612
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
LOTERSZTAJN ET AL.
ANTIFIBROTIC STRATEGIES
4 Dec 2004
19:4
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
613
4 Dec 2004
19:4
614
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
LOTERSZTAJN ET AL.
Compound
density of
fibrogenic
cells in vitro
fibrogenesis
and/or
fibrolysis
in vitro
Antifibrotic
effects in
animals
Reference(s)
Adiponectin
ND
(116)
Amiloride
(125)
Antiangiotensin
Antioxidants (tocopherol,
resveratrol, sylimarin,
S-adenosylmethionine,
Sho-saiko-to. . .)
(6770)
Anti-TGF-
(5861)
Cannabinoid receptor 1
antagonism
ND
ND
(121)
Cannabinoid receptor 2
agonism
ND
(74)
Endothelin A receptor
antagonists
ND
ND
(84)
Endothelin B receptor
agonists
ND
ND
Gliotoxin
ND
(102, 103)
Halofuginone
(126)
Integrin antagonists
(26, 127)
Interleukin-10
ND
(53, 54)
Interferon-
(62)
Interferon-
(62)
Noradrenergic
antagonists
(128, 129)
Pentoxifylline
(130, 131)
15-D-prostaglandin J2
(73,
104107)
Prostaglandin E2
ND
Sphingosine-1 phosphate
ND
(17, 30)
Thiazolininediones
(104106,
108)
4 Dec 2004
19:4
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
615
signaling pathway may also reduce liver fibrogenesis, as shown by the beneficial
effects of an adenovirus carrying Smad 7 cDNA in the bile duct ligation model
(61).
Although attractive given their efficiency, systemic anti-TGF- strategies may
be limited by adverse effects, such as the risk of autoimmune disease secondary
to its prominent immunoregulatory properties.
Among antifibrogenic Th1 cytokines, interferons have been
the subject of extensive studies. Interferon- and interferon- inhibit activation,
proliferation, and collagen synthesis in cultures of activated hepatic stellate cells
and hepatic myofibroblasts (62); in addition, both cytokines directly inhibit collagen gene transcription in vivo and reduce progression of fibrosis, as shown in a
model of transgenic mice harboring the 2(I) collagen gene (63). In keeping with
these experimental findings, studies in patients with chronic hepatitis C suggest
that IFN- may improve the stage of fibrosis irrespective of virological response,
suggesting a direct inhibitory effect of the cytokine on fibrosis progression (5,
64). This hypothesis is being further evaluated in several ongoing clinical trials.
Beneficial effects of hepatocyte growth factor (HGF) delivered as a recombinant
protein or by gene therapy have also been reported following dimethylnitrosamine
administration (65). However, HGF being a promitogenic factor for parenchymal cells, long-term administration raises concern as to the risk of epithelial
tumors.
OTHER CYTOKINES
4 Dec 2004
19:4
616
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
LOTERSZTAJN ET AL.
A number of vasoregulatory peptides are overproduced during liver fibrogenesis and show pro- or antifibrogenic properties. These observations have stimulated assessment of pharmacological activator or inhibitors of these compounds.
Endothelin-1, the angiotensin system, and prostaglandins have provided the most
convincing data.
Endothelin-1 is a potent vasoconstrictor that binds at least two G
proteincoupled receptors, ETA and ETB (7577). Investigation of the role of endothelins in liver fibrogenesis was stimulated by the finding that both endothelin-1
and its receptors are markedly induced in fibrogenic cells during chronic liver
diseases (78, 79) and by the previous demonstration of a profibrogenic role of the
peptide in kidney fibrogenesis (80). Culture studies have shown that endothelin1 displays dual pro- and antifibrogenic effects in the liver according to receptor
subtype: thus, binding of ETA receptors stimulates activation of hepatic stellate
cells and induces a weak mitogenic effect. In contrast, binding of ETB receptors
promotes marked growth inhibition (28, 81) by a mechanism involving the sequential generation of sphingosine-1-phosphate (S1P), cyclooxygenase-2 (COX2)-derived prostaglandins, and elevation of cAMP (28, 82, 83). Therefore, these
results suggested that antifibrotic effects may be achieved by selectively inhibiting
ETA receptors, whereas beneficial antifibrogenic effects of ETB receptors should
be protected, or even better enhanced. In keeping with these in vitro studies, administration of a selective ETA receptor antagonist prevents the development of
liver fibrosis in bile ductligated rats (84), whereas treatment with a nonselective
ETA/ETB receptor antagonist accelerates liver fibrosis in carbon tetrachloridetreated rats (85).
ENDOTHELIN-1
Angiotensin II is involved in cardiac and kidney fibrogenesis, and several recent studies support a significant role in liver fibrosis.
AT1 receptors are upregulated in fibrotic areas during experimental liver fibrosis
(86). Accordingly, cultured activated stellate cells express AT1 receptors and produce angiotensin II in response to growth factors via the renin angiotensin system
(87). Furthermore, activation of AT1 receptors stimulates secretion of TGF- and
proliferation of cultured activated stellate cells (88, 89). Finally, the relationship
between angiotensin II and liver fibrogenesis is supported by experimental and
clinical studies. Thus, mice invalidated for AT1 receptors show reduced liver fibrosis following administration of carbon tetrachloride (90). These observations
are corroborated by the beneficial effect of angiotensin antagonism in experimental
models of liver fibrosis, whether using angiotensin inhibitors or antagonists of AT1
receptors (88, 91, 92). In patients with chronic hepatitis C, there is a statistically
significant relationship between inheritance of a high angiotensinogen-producing
genotype and progression of hepatic fibrosis (93). Finally, a controlled pilot study
in hepatitis C recently showed that losartan reduces liver fibrosis as compared to
4 Dec 2004
19:4
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
617
untreated controls (94). Multicenter prospective trials assessing angiotensin antagonism in liver fibrosis are currently under way.
A number of studies have demonstrated antifibrogenic potential of prostaglandins. Thus, PGE2 reduces fibrosis progression in bile ductligated
rats (95). Beneficial effects are related to inhibition of proliferation and collagen
synthesis in hepatic myofibroblasts and activated hepatic stellate cells, as shown
in culture studies (95, 96). Interestingly, we have shown that growth inhibitory
effects of several factors, such as endothelin-1, TNF-, and S1P, involve induction of COX-2 and subsequent generation of PGE2 (17, 83, 96). Finally, we also
demonstrated that the mitogenic effects of PDGF-BB and thrombin result from a
balance between a promitogenic pathway and a parallel COX-2-dependent growth
inhibitory pathway (24). Together, these data point to COX-2 as a source of antifibrogenic prostaglandins in the liver.
PROSTAGLANDINS
Enhancement of Apoptosis
It has been demonstrated conclusively in experimental models that apoptosis of
hepatic fibrogenic cells is a key mandatory step in the recovery process following
fibrosis induction. Thus, available data indicate that during liver fibrogenesis, proliferation of fibrogenic cells predominates over spontaneous apoptosis, whereas
cessation of liver injury is associated with a reduction of proliferation and a marked
increase in apoptosis. Importantly, apoptosis of fibrogenic cells is accompanied
by a restoration of the collagenolytic capacities of MMP-1 and MMP-2 in the
liver, subsequent to a decrease in TIMP-1 and TIMP-2 expression, which allows
progressive matrix degradation (2, 97).
These observations have been strong incentives to characterize pathways regulating apoptosis and survival of fibrogenic cells. Available studies have been
performed mainly in cultures and have identified a number of apoptotic stimuli.
Classical apoptotic factors such as Fas-L, TRAIL 2, and TRAIL 5, and their receptors Fas and TRAIL, are upregulated during transition of hepatic stellate cells to
their activated myofibroblastic phenotype (98100). Other receptor-mediated stimuli include nerve growth factor and benzodiazepines (25, 101); however, expression
of the benzodiazepine receptor is transient and declines in activated hepatic stellate cells. Nonreceptor-mediated apoptosis of hepatic myofibroblasts also occurs
in response to a COX-2-derived prostaglandin, 15-deoxy 12,14 prostaglandin J2
(15-D-PGJ2) (73). Furthermore, we have also recently shown that hepatic myofibroblasts undergo apoptosis following exposure to sphingomyelinase metabolites,
including ceramide, sphingosine, and sphingosine-1-phosphate (S1P) (30). Investigation of the role of S1P arose from the findings that hepatic myofibroblasts express Edg receptors for the molecule (17, 30) and that sphingosine kinase activity
is increased in carbon tetrachloridetreated rats (P. Grenard, T. Levade, A. Mallat
& S. Lotersztajn, unpublished results). We found that S1P stimulates two parallel pro- and antiapoptotic pathways in human hepatic myofibroblasts, probably
4 Dec 2004
19:4
618
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
LOTERSZTAJN ET AL.
via distinct receptors. The apoptotic signal is mediated by caspase-3, whereas the
survival signal is conveyed by activation of ERK and PI3K (30).
Two experimental studies using the fungal toxin gliotoxin have documented
the potential efficiency of a proapoptotic strategy in vivo. It was shown that the
compound kills activated hepatic stellate cells in culture (102), and that in both carbon tetrachloride- and thioacetamide-treated rats, treatment with gliotoxin reduces
the number of fibrogenic cells and decreases fibrosis (102, 103). A major issue of
a proapoptotic strategy is that of cell specificity because nonselective effects may
result in life-threatening side effects, such as severe or fulminant hepatitis.
4 Dec 2004
19:4
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
619
4 Dec 2004
19:4
620
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
LOTERSZTAJN ET AL.
DRUG TARGETING
CONCLUSION
During the past decade, characterization of molecular mechanisms of liver fibrogenesis and resolution has revealed novel approaches for therapeutic intervention
based on interference with major pro- or antifibrogenic pathways in liver fibrogenic
cells. A large number of approaches have been validated in culture studies and in
animal models. Clinical trials are underway or anticipated for a growing number
of molecules, and will obviously be facilitated by the availability of noninvasive
methods for staging fibrosis. However, proof of effectiveness is still lacking in
humans. Combination of drugs with distinct antifibrogenic actions may result in
therapeutic benefits at low dosages and reduce the risk of unwanted side effects.
ACKNOWLEDGMENTS
P. Grenard was supported by INSERM and B. Julien by a fellowship from the
Minist`ere de la Recherche et de la Technologie. This work was supported by
the INSERM, the Universite Paris-Val-de-Marne, and by grants (to S.L.) of the
Association pour la Recherche sur le Cancer and the Ligue departementale du Val
de Marne de la Recherche contre le Cancer.
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Benyon RC, Arthur MJ. 2001. Extracellular matrix degradation and the role of
hepatic stellate cells. Semin. Liver Dis.
21:37384
4 Dec 2004
19:4
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
621
4 Dec 2004
19:4
622
20.
21.
22.
23.
24.
25.
26.
27.
28.
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
LOTERSZTAJN ET AL.
bone marrow cells to hepatic stellate cells
in mouse. J. Hepatol. 40:25560
Forbes SJ, Russo FP, Rey V, Burra P,
Rugge M, et al. 2004. A significant proportion of myofibroblasts are of bone marrow origin in human liver fibrosis. Gastroenterology 126:95564
Friedman SL. 2000. Molecular regulation
of hepatic fibrosis, an integrated cellular
response to tissue injury. J. Biol. Chem.
275:224750
Han YP, Zhou L, Wang J, Xiong S, Garner WL, et al. 2004. Essential role of
matrix metalloproteinases in interleukin1-induced myofibroblastic activation of
hepatic stellate cell in collagen. J. Biol.
Chem. 279:482028
Pinzani M, Marra F. 2001. Cytokine receptors and signaling in hepatic stellate
cells. Semin. Liver Dis. 21:397416
Mallat A, Gallois C, Tao J, Habib A,
Maclouf J, et al. 1998. Platelet-derived
growth factor-BB and thrombin generate
positive and negative signals for human
hepatic stellate cell proliferation. Role of
a prostaglandin/cyclic AMP pathway and
cross-talk with endothelin receptors. J.
Biol. Chem. 273:273005
Iredale JP. 2001. Hepatic stellate cell behavior during resolution of liver injury.
Semin. Liver Dis. 21:42736
Zhou X, Murphy FR, Gehdu N, Zhang
J, Iredale JP, Benyon RC. 2004. Engagement of alpha Vbeta 3 integrin regulates
proliferation and apoptosis of hepatic stellate cells. J. Biol. Chem. 24:24
Tao J, Mallat A, Gallois C, Belmadani S,
Mery PF, et al. 1999. Biological effects of
C-type natriuretic peptide in human myofibroblastic hepatic stellate cells. J. Biol.
Chem. 274:2376169
Mallat A, Preaux AM, Serradeil-Le Gal
C, Raufaste D, Gallois C, et al. 1996.
Growth inhibitory properties of endothelin-1 in activated human hepatic stellate cells: a cyclic adenosine monophosphate-mediated pathway. Inhibition
of both extracellular signal-regulated ki-
29.
30.
31.
32.
33.
34.
35.
36.
4 Dec 2004
19:4
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
623
cirrhosis using DNA sequencer-based total serum protein glycomics. Nat. Med.
10:42934
Ziol M, Barget N, Sandrin L, Fournier C,
Mal F, et al. 2004. Correlation between
liver elasticiy measured by transient
elastography and liver fibrosis assessed
by morphometry in patients wih HCV
chronic hepatitis. J. Hepatol. 40(Suppl.
1):136
Canbay A, Higuchi H, Bronk SF, Taniai
M, Sebo TJ, Gores GJ. 2002. Fas enhances fibrogenesis in the bile duct ligated mouse: a link between apoptosis and
fibrosis. Gastroenterology 123:132330
Song E, Lee SK, Wang J, Ince N, Ouyang
N, et al. 2003. RNA interference targeting
Fas protects mice from fulminant hepatitis. Nat. Med. 9:34751
Schiff ER, Pockros P, Schiffman ML,
McHutchinson J, Gish R, et al. 2004. Oral
IDN-6556, an anti apoptotic caspase inhibitor lowers aminotransferases in HCV
patients. J. Hepatol. 40(Suppl. 1):24
Canbay A, Feldstein A, Baskin-Bey E,
Bronk SF, Gores GJ. 2004. The caspase
inhibitor IDN-6556 attenuates hepatic injury and fibrosis in the bile duct ligated mouse. J. Pharmacol. Exp. Ther.
308:119196
Titos E, Claria J, Planaguma A, LopezParra M, Villamor N, et al. 2003. Inhibition of 5-lipoxygenase induces cell
growth arrest and apoptosis in rat Kupffer cells: implications for liver fibrosis.
FASEB J. 17:174547
Louis H, Van Laethem JL, Wu W,
Quertinmont E, Degraef C, et al. 1998.
Interleukin-10 controls neutrophilic infiltration, hepatocyte proliferation, and liver
fibrosis induced by carbon tetrachloride in
mice. Hepatology 28:160715
Thompson K, Maltby J, Fallowfield J,
McAulay M, Millward-Sadler H, Sheron
N. 1998. Interleukin-10 expression and
function in experimental murine liver inflammation and fibrosis. Hepatology 28:
1597606
4 Dec 2004
19:4
624
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
LOTERSZTAJN ET AL.
64.
65.
66.
67.
68.
69.
70.
71.
72.
4 Dec 2004
19:4
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
82.
83.
84.
85.
86.
87.
88.
89.
625
4 Dec 2004
19:4
626
90.
91.
92.
93.
94.
95.
96.
97.
98.
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
LOTERSZTAJN ET AL.
proliferation of human hepatic stellate
cells. Gastroenterology 118:114956
Kanno K, Tazuma S, Chayama K. 2003.
AT1A-deficient mice show less severe
progression of liver fibrosis induced by
CCl(4). Biochem. Biophys. Res. Commun.
308:17783
Croquet V, Moal F, Veal N, Wang J,
Oberti F, et al. 2002. Hemodynamic and
antifibrotic effects of losartan in rats with
liver fibrosis and/or portal hypertension.
J. Hepatol. 37:77380
Jonsson JR, Clouston AD, Ando Y,
Kelemen LI, Horn MJ, et al. 2001.
Angiotensin-converting enzyme inhibition attenuates the progression of rat hepatic fibrosis. Gastroenterology 121:148
55
Powell EE, Edwards-Smith CJ, Hay JL,
Clouston AD, Crawford DH, et al. 2000.
Host genetic factors influence disease progression in chronic hepatitis C. Hepatology 31:82833
Terui Y, Saito T, Watanabe H, Togashi
H, Kawata S, et al. 2002. Effect of angiotensin receptor antagonist on liver fibrosis in early stages of chronic hepatitis
C. Hepatology 36(4 Pt. 1):1022
Beno DW, Espinal R, Edelstein BM,
Davis BH. 1993. Administration of
prostaglandin E1 analog reduces rat hepatic and Ito cell collagen gene expression
and collagen accumulation after bile duct
ligation injury. Hepatology 17:70714
Mallat A, Fouassier L, Preaux AM,
Mavier P, Lotersztajn S. 1995. Antiproliferative effects of ET-1 in human liver
Ito cells: an ETB- and a cyclic AMPmediated pathway. J. Cardiovasc. Pharmacol. 26(Suppl. 3):S13234
Preaux AM, DOrtho MP, Bralet MP,
Laperche Y, Mavier P. 2002. Apoptosis of
human hepatic myofibroblasts promotes
activation of matrix metalloproteinase-2.
Hepatology 36:61522
Gong W, Pecci A, Roth S, Lahme B, Beato
M, Gressner AM. 1998. Transformationdependent susceptibility of rat hepatic
99.
100.
101.
102.
103.
104.
105.
106.
4 Dec 2004
19:4
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
107.
108.
109.
110.
111.
112.
113.
114.
115.
116.
117.
118.
119.
120.
121.
122.
123.
124.
125.
627
4 Dec 2004
19:4
628
AR
AR232-PA45-25.tex
XMLPublishSM (2004/02/24)
P1: JRX
LOTERSZTAJN ET AL.
4 Dec 2004
19:7
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
10.1146/annurev.pharmtox.45.120403.095832
HISTORICAL INTRODUCTION
The field of DNA methylation is attracting the interest of many researchers and
clinicians around the world. Some of the best laboratories are gradually changing their old interests and are moving into the emerging fields of epigenetics
and, particularly, DNA methylation. Biotechnological and pharmaceutical companies are developing research programs specifically designed to develop new DNA
0362-1642/05/0210-0629$14.00
629
4 Dec 2004
19:7
630
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
ESTELLER
4 Dec 2004
19:7
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
631
Figure 1 Protein occupancy and methylation status of a promoter CpG island of a tumor
suppressor gene in normal and cancer cells. Gray boxes, exons; white circles, unmethylated
CpGs; black circles, methylated CpGs.
4 Dec 2004
19:7
632
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
ESTELLER
Figure 2 Techniques available for the study of DNA methylation according to the researcher
interests.
4 Dec 2004
19:7
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
633
The remaining DNA methyltransferases were identified by searches of expressed sequence tag (EST) databases. The first of these was DNMT2 (17). This
lacks the large N-terminal regulatory domain common to other eukaryotic methyltransferases and does not exhibit comparable DNA methyltransferase activity,
although it does seem to have some residual activity in vitro (18). DNMT3a and
DNMT3b were soon identified by searching EST databases (19) and were proposed to be the enzymes responsible for de novo methylation (20). Mutations in
the human DNMT3B gene are responsible for ICF syndrome. Figure 1A shows a
schematic representation of the DNMT family.
Although DNMTs were originally classified as maintenance or de novo DNA
methyltransferases, there are several strands of evidence that indicate that all three
DNMTs not only cooperate but also may possess both de novo and maintenance
functions in vivo (2125).
The information stored by methylation of CpGs has functional significance only
in the context of chromatin. Since its discovery, DNA methylation has been associated with a transcriptionally inactive state of chromatin; however, the mechanisms
by which DNA methylation is translated into transcriptionally silent chromatin
have only recently started to be unveiled.
Historically, several hypotheses have been proposed to explain the way by which
DNA methylation is interpreted by nuclear factors. The first possibility is that DNA
methylation inhibits the binding of sequence-specific transcription factors to their
binding sites that contain CpG (26). In this context, a protein with an affinity
for unmethylated CpGs has also been identified that is associated with actively
transcribed regions of the genome (27). In this case, methylation of CpGs would
result in release of this protein. An alternate model proposed that methylation may
have direct consequences for nucleosome positioning, for instance, by leading
to the assembly of specialized nucleosomal structures on methylated DNA that
silence transcription more effectively than conventional chromatin (28). The third
possibility is that methylation leads to the recruitment of specialized factors that
selectively recognize methylated DNA and either impede binding of other nuclear
factors or have a direct effect on repressing transcription (29).
Although there are examples that support all three possibilities, the active recruitment of methyl-CpG binding activities appears to be the most widespread
mechanism of methylation-dependent repression.
MeCP1 and MeCP2 were the first two methyl-CpG binding activities described
(29). Although MeCP1 was originally identified as a large multiprotein complex,
MeCP2 is a single polypeptide with an affinity for a single methylated CpG.
Characterization of MeCP2 in subsequent years led to the identification of the
minimum portion with affinity for methylated DNA, i.e., its methyl-CpG binding
domain (MBD) (30) and its transcriptional repression domain (TRD).
Database searches led to the identification of additional proteins harboring the
MBD, namely MBD1, MBD2, MBD3, and MBD4 (31). Whereas mammalian
MBD1 and MBD2 are bona fide methylated DNA binding proteins, MBD3 is able
to bind methylated DNA only in certain species (31, 32). In the case of MBD4, this
protein binds preferentially to m5CpG x TpG mismatches. The primary product
4 Dec 2004
19:7
634
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
ESTELLER
4 Dec 2004
19:7
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
635
methylation of the cytosine located within the dinucleotide CpG. 5mC in normal human tissue DNAs constitutes 0.75%1% of all nucleotide bases, and about
4%6% of all cytosines are methylated in normal human DNA (2, 3, 47).
CpG dinucleotides are not randomly distributed throughout the vast human
genome. CpG-rich regions, known as CpG islands (48), are usually unmethylated
in all normal tissues and frequently span the 5 -end region (promoter, untranslated region, and exon 1) of a number of genes; they are excellent markers of the
beginning of a gene. If the corresponding transcription factors are available, the
histones are in an acetylated and unmethylated state, and if the CpG island remains
in an unmethylated state, then that particular gene will be transcribed (Figure 3,
see color insert).
Of course, there are exceptions to the general rule. We can find certain normally
methylated CpG islands in at least four cases: imprinted genes, X-chromosome
genes in women, germline-specific genes, and tissue-specific genes (49). Genomic
or parental imprinting is a process involving acquisition of DNA hypermethylation in one allele of a gene early in the male and female germline that leads
to monoallelic expression (50). A similar phenomenon of gene-dosage reduction can also be invoked with regard to the methylation of CpG islands in one
X-chromosome in women, which renders these genes inactive to avoid redundancy. Finally, although DNA methylation is not a widely occurring system for
regulating normal gene expression, sometimes it does indeed accomplish this
purpose, as with the genes whose expression is restricted to the male or female
germline and not expressed later in any adult tissue, such as the MAGE gene
family. Finally, methylation has been postulated as a mechanism for silencing
tissue-specific genes in cell types in which they should not be expressed. However, it is still not clear whether this type of methylation is secondary to a lack of
gene expression owing to the absence of the particular cell typespecific transcription factor or whether it is the main force behind transcriptional tissue-specific
silencing.
What is the significance of the presence of DNA methylation outside the CpG
islands? One of the most exciting possibilities for the normal function of DNA
methylation is its role in repressing parasitic DNA sequences (51, 52). Our genome
is plagued with transposons and endogenous retroviruses acquired throughout the
history of the human species. We can control these imported sequences with direct
transcriptional repression mediated by several host proteins, but our main line of
defense against the large burden of parasitic sequence elements (more than 35%
of our genome) may be DNA methylation. Methylation of the promoters of our
intragenomic parasites inactivates these sequences and, over time, will destroy
many transposons.
The perfect epigenetic equilibrium of the previously described normal cell is
dramatically transformed in the cancer cell. The epigenetic aberrations observed
can be summarized as falling into one of two categories: transcriptional silencing
of tumor suppressor genes by CpG island promoter hypermethylation and global
genomic hypomethylation.
4 Dec 2004
19:7
636
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
ESTELLER
GENOMIC HYPOMETHYLATION
OF TRANSFORMED CELLS
At the same time that certain CpG islands become hypermethylated, as discussed
below, the genome of the cancer cell undergoes dramatic global hypomethylation.
The malignant cell can have 20%60% less genomic 5mC than its normal counterpart (2, 3). The loss of methyl groups is accomplished mainly by hypomethylation
of the body (coding region and introns) of genes and through demethylation of
repetitive DNA sequences , which accounts for 20%30% of the human genome.
The degree of genomic DNA hypomethylation increases through all the tumorogenic steps, from the benign proliferations to the invasive cancers (14) (Figure 4,
see color insert).
How does global DNA hypomethylation contribute to carcinogenesis? Three
mechanisms can be invoked: chromosomal instability, reactivation of transposable
elements, and loss of imprinting. Undermethylation of DNA might favor mitotic recombination, leading to loss of heterozygosity as well as promoting karyotypically
detectable rearrangements. Additionally, extensive demethylation in centromeric
sequences is common in human tumors and may play a role in aneuploidy. It has
been reported that patients with germline mutations in DNA methyltransferase
3b (DNMT3b) have numerous chromosome aberrations (53). Hypomethylation of
malignant cell DNA can also reactivate intragenomic parasitic DNA, such as L1
(long interspersed nuclear elements, LINES) and Alu (recombinogenic sequence)
repeats (51, 52). These, and other previously silent transposons, may now be transcribed and even jump to other genomic regions where they can disrupt normal
cellular genes. Finally, the loss of methyl groups can affect imprinted genes. The
best-studied case concerns the effects of the H19/IGF-2 locus on chromosome
11p15 in certain childhood tumors (54, 55).
However, we still know very little about the real role of DNA hypomethylation
in the development of cancer cells. Is it really a causative factor? Or just a
modulator of cancer risk? Or only a bystander passenger? The studies in mouse
models are extremely interesting but puzzling: When the mouse deficient in DNA
methylation owing to a defect in DNMT1 is crossed with the colon adenomaprone Min mouse (with a genetic defect in APC), the resulting mouse has fewer
tumors (56); but another DNMT1 defective mouse may have an increased risk of
lymphomas (57). This paradox is an important question that needs to be addressed
in the near future.
4 Dec 2004
19:7
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
637
methylated in every tumor type, but strong specificity is apparent with respect to the
tissue of origin (11, 58). We have recently described the exquisite profile of hypermethylation that occurs in primary human tumors (11). Furthermore, the number
of hypermethylated genes increases with the malignant potential (14) (Figure 4).
We do not currently know why some genes became hypermethylated in certain
tumors, whereas others with similar properties (a typical CpG island, a history
of loss of expression in certain tumors, and the absence of mutations) remain
methylation-free. We can hypothesize, as researchers have done before with genetic
mutations, that a particular gene is preferentially methylated with respect to others
in certain tumor types because inactivation confers a selective advantage, in the
Darwinian sense, on the former. Another option is that aberrant DNA methylation is
directly targeted. It has been proposed that fusion proteins, such as PML-RAR, can
contribute to aberrant CpG-island methylation by recruiting DNMTs and HDACs
to aberrant sites (59). This latter activity is somewhat controversial but, in any
case, does not seem to be a general mechanism, at least in leukemia patients
(60). Selection and targeting are not exclusive events, and they are most probably
happening together in the generation and maintenance of hypermethylated CpG
islands of tumor suppressor genes.
The tumor suppressor genes, bona fide and look-alike, that undergo aberrant
CpG island methylation in human cancer affect all the cellular pathways and have
relevant consequences (49). A brief list of the most significant genes inactivated
by DNA hypermethylation is represented in Table 1 and includes the following:
(a) Cell cycle. The cell cycle inhibitor p16INK4a is hypermethylated in a wide
variety of human primary tumors and cell lines (68), allowing the cancer
cell to escape senescence and start proliferating. The Rb gene and the cell
cycle inhibitor p15INK4b can also suffer occasionally aberrant methylation
(4, 61).
(b) p53 network. p53 is the most frequently mutated tumor suppressor gene in
human cancer; nevertheless, half of human primary tumors are wild-type
p53. Another way to inactivate p53 is through the methylation-mediated silencing of the tumor suppressor gene p14ARF (6264) because in this way the
MDM2 oncogenic protein is not inhibited by p14ARF and is free to induce p53
degradation (64). p73, a gene that is a p53-homolog, is also hypermethylated
in leukemias (65).
(c) APC/-catenin/E-cadherin pathways. APC is commonly mutated in sporadic
colon tumors, but little was known about the relevance of this particular pathway in noncolorectal tumorogenesis until recently. Now it is recognized that
aberrant methylation of APC is a common lesion in other neoplasms of the
aerodigestive tract (66). E-cadherin, H-cadherin, and FAT tumor-suppressor
cadherin promoter hypermethylation is also important in the cancer biology
of breast, colon, and other tumor types (25, 67, 68). Finally, methylationassociated silencing of the genes encoding secreted frizzled-related proteins
(SFRPs), which possess a domain similar to one in the WNT-receptor frizzled
4 Dec 2004
19:7
638
AR
XMLPublishSM (2004/02/24)
P1: JRX
ESTELLER
Table 1
AR232-PA45-26.tex
A selected list of genes that undergo CpG island hypermethylation in human cancer
Gene
Function
Consequences
p16INK4a
Cyclin-dependent kinase
inhibitor
9p21
Multiple types
Entrance in cell
cycle
p14ARF
MDM2 inhibitor
9p21
Colon,
stomach,
kidney
Degradation of p53
p15INK4b
Cyclin-dependent kinase
inhibitor
9p21
Leukemia
Entrance in cell
cycle
hMLH1
3p21.3
Colon,
endometrium,
stomach
Frameshift
mutations
MGMT
DNA repair of
06-alkyl-guanine
10q26
Multiple types
Mutations,
chemosentivity
GSTP1
Conjugation to
glutathione
11q13
Prostate, breast,
kidney
Adduct
accumulation?
BRCA1
DNA repair,
transcription
17q21
Breast, ovary
Double-strand
breaks?
p73
p53 homolog
1p36
Lymphoma
Unknown
19p13.3
Colon, breast,
lung
Unknown
ER
Estrogen receptor
6q25.1
Breast
Hormone
insensitivity
PR
Progesterone receptor
11q22
Breast
Hormone
insensitivity
AR
Androgen receptor
Xq11
Prostate
Hormone
insensitivity
PRLR
Prolactin receptor
5p13p12
Breast
Hormone
insensitivity
RAR2
3p24
Colon, lung,
head, and neck
Vitamin
insensitivity?
RASSF1A
3p21.3
Multiple types
Unknown
NORE1A
1q32
Lung
Unknown
VHL
Ubiquitin ligase
component
3p25
Kidney, hemangioblastoma
Loss of hypoxic
response?
Rb
13q14
Retinoblastoma
Entrance in cell
cycle
THBS-1
Thrombospondin-1,
antiangiogenic
15q15
Glioma
Neovascularization
(Continued)
4 Dec 2004
19:7
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
639
Table 1 (Continued)
Gene
Function
Consequences
CDH1
E-cadherin, cell
adhesion
16q22.1
Breast,
stomach,
leukemia
Dissemination
CDH13
H-cadherin, cell
adhesion
16q24
Breast, lung
Dissemination?
FAT
Cadherin, tumor
suppressor
4q34-35
Colon
Dissemination?
HIC-1
Transcription factor
17p13.3
Multiple types
Unknown
APC
Inhibitor of -catenin
5q21
Aerodigestive
tract
Activation
-catenin route
SFRP1
Secreted Frizzled-related
Protein 1
8p12p11
Colon
Activation WNT
signaling
COX-2
Cyclooxygenase-2
1q25
Colon, stomach
Antiinflammatory
resistance?
SOCS-1
Inhibitor of JAK/STAT
pathway
JAK2 activation
SOCS-3
Inhibitor of JAK/STAT
pathway
17q25
Lung
JAK2 activation
GATA-4
Transcription factor
8p23p22
Colon, stomach
Silencing of target
genes
GATA-5
Transcription factor
20q13
Colon, stomach
Silencing of target
genes
SRBC
BRCA1-binding protein
1p15
Breast, lung
Unknown
SYK
Tyrosine kinase
9q22
Breast
Unknown
RIZ1
Histone/protein
methyltransferase
1p36
Breast, liver
Aberrant gene
expression?
DAPK
Pro-apoptotic
9q34.1
Lymphoma,
lung, colon
Resistance to
apoptosis
TMS1
Pro-apoptotic
16p11
Breast
Resistance to
apoptosis
2q33
Colon, bladder
Unknown
proteins and can inhibit WNT receptor binding to downregulate pathway signaling during development, has also been found in colorectal cancer (69).
(d) DNA repair. DNA methylation is one of the major players at this crossroads
of all cell pathways. Selected examples are the methylation-mediated silencing of the mismatch DNA repair gene hMLH1 in sporadic cases of colorectal
(70, 71), endometrial (72, 73), and gastric tumors (74) that cause the unusual
4 Dec 2004
19:7
640
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
ESTELLER
phenotype known as microsatellite instability; the promoter hypermethylation of MGMT (75) that prevents the removal of groups at the O6 position of
the guanine and leads to the appearance of K-ras and p53 mutations (7678);
the hypermethylation of the mitotic checkpoint gene CHFR (79); and the somatic inactivation of BRCA1 by aberrant methylation in breast and ovarian
tumors (80), which alters its role in the repair of double-strand breaks in
the DNA and leads to the same global expression changes that occur in the
carriers of BRCA1 germ line mutations (81).
(e) Hormonal response. Aberrant methylation of the estrogen, progesterone,
androgen, and prolactin receptors occurs in breast and uterine tumors and
may render these cancer cells unresponsive to steroid hormones (45, 8284).
The differentiating action of the retinoids may also be abolished in tumors
that show promoter hypermethylation of the retinoic acid receptor-2 (60,
8587) and the cellular retinol-binding protein I (87).
(f) Cytokine signaling. The suppressor of cytokine signaling (SOCS) family of
proteins has been implicated in the negative regulation of several cytokine
pathways, particularly the receptor-associated tyrosine kinase/signal transducer and activator of transcription (Jak/STAT) pathways of transcriptional
activation. SOCS-1 and SOCS-3 undergo methylation-associated silencing
in human cancer (8890).
(g) The remaining pathways. This is not an exhaustive list, but I would like to
emphasize that every imaginable molecular route can be affected by a candidate gene: the proapoptotic death-associated protein kinase (DAPK) (91) and
TMS1 (92); the kidney tumor and hemangioblastoma-related VHL gene (5);
the serine-threonine kinase LKB1/STK11 in hamartomatous neoplasms (93);
the ras-effector genes RASSF1A (94, 95) and NORE1A (96); the antiangiogenic factor thrombospondin-1 (THBS-1) (97); the prostaglandin generator
cyclooxygenase 2 (98); the TPEF gene that contains epidermal growth factor domains (99); the electrophilic detoxifier glutathione S-transferase P1
(GSTP1) in prostate, breast, and kidney tumors (100, 101); the transcription
factors GATA-4 and GATA-5 (102); and many more.
Finally, it is important to mention that as a consequence of the increasing
number of hypermethylated genes in human cancer, we need to demonstrate a
role for the methylation-associated silencing of the studied gene in tumor biology.
For example, we can check if the reintroduction of the gene in a deficient cancer
cell line reduces colony formation (25, 45, 103) or inhibits xenograft growth in
nude mice (95); if the hypermethylation of that gene correlates with a particular
molecular or clinical phenotype, as is the case with the MGMT methylation that
is associated with the appearance of transition mutations and chemosensitivity to
alkylating agents (78); if the methylation-mediated silencing has the same effects
as a frameshift mutation, as it has been shown for BRCA1 (81); or if mutations
for that gene are not described, generating a knockout mouse, as has been done
for HIC-1 (104).
4 Dec 2004
19:7
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
641
4 Dec 2004
19:7
642
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
ESTELLER
4 Dec 2004
19:7
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
643
4 Dec 2004
19:7
644
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
ESTELLER
4 Dec 2004
19:7
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
645
chemoresponse of human tumors to alkylating agents is not limited to BCNUlike alkylating agents; it also extends to other drugs such as cyclophosphamide
(122). This has been demonstrated in diffuse large cell lymphomas treated with
cyclophosphamide, where MGMT hypermethylation was the strongest predictor
of overall survival and time to progression and was far superior to classical clinical
factors such as the international prognostic index (122). Finally, we have extended
in gliomas the use of MGMT methylation as a marker of good clinical response
for another drug recently introduced, temozolomide (123).
Similar cases to that described for MGMT can be cited for other DNA repair and
detoxifier genes that also undergo aberrant DNA methylation. For example, the
response to cisplatin and derivatives may be a direct function of the methylation
state of the CpG island of hMLH1 (124), the response to adriamicine may be
related to the methylation status of GSTP1 (101), and the response to certain DNAdamaging drugs could be a function of the state of BRCA1 hypermethylation (80,
125).
Finally, gene inactivation by promoter hypermethylation may be the key to
understanding the loss of hormone response in many tumors. The inefficacy of
the antisteroids, estrogen-progesterone-androgen-related compounds such as tamoxifen, raloxifene, or flutemide, in certain breast, endometrial, and prostate cancer cases may be a direct consequence of the methylation-mediated silencing
of their respective cellular receptors (ER, PR, and AR genes). A similar picture
can be painted for the retinoids: Why has chemoprevention with these agents
not produced the results that we so desire and expect? A highly convincing explanation is that the tumors and the premalignant lesions become insensitive to
these compounds owing to epigenetic silencing of genes that are crucial in the
retinoid response, especially the retinoic acid receptor 2 (RAR2) (60, 8587)
and the cellular retinol binding protein I (CRBPI) (87). This is a dynamic process, and we have demonstrated that a suitable supply of dietary retinoids prevents the aberrant methylation of RAR2 and CRBPI in colorectal tumorigenesis
(87).
4 Dec 2004
19:7
646
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
ESTELLER
(130), which give us more hope; however, the nonspecificity problem persists.
If only tumor suppressor genes were hypermethylated, this would not be a great
problem. However, we do not know if we have disrupted some essential methylation
at certain sites, and global hypomethylation may be associated with even greater
chromosomal instability. Another drawback is the toxicity to normal cells. Indeed,
this phenomenon was observed when higher doses were first used.
The first analog tested as a possible inhibitor of DNA methylation was 5azacytidine (131). This substance causes covalent arresting of DNMTs, resulting
in cytotoxicity, and tumors with increased levels of these enzymes are expected
to present higher sensitivity toward the drug (131). 5-azacytidine was tested as
an antileukemic drug before its demethylating activity was known (132, 133). It
is reported that 5-azacytidine has interference at very low concentrations (below
0.1 M) in RNA processing, tRNA methylation, and protein synthesis owing to its
incorporation preferential into RNA in vivo and in cultured cells. Treatment with
equimolar amounts of both cytidine and 5-azacytidine inhibits the incorporation
of the latter one in nucleic acids, resulting in no alteration of the cell cycle either in
vivo or ex vivo. 5-azacytidine is degradated by a nucleoside deaminase, so cells that
highly express this enzyme are not sensitive to this compound (132). Therefore, 5azacytidine is much less employed in studies related to methylation. However, more
recently, it has been concluded that irreversible cell cycle arresting at phases G1/G0,
G2, and S caused by this compound when used at micromolar concentrations is
due to its effects on DNA methylation and not on RNA metabolism (134). It is still
used in clinical trials (135, 136).
The analog 5-aza-2 -deoxycitidine (Decitabine) is one of the most used demethylating drugs for assays with cultured cells. It overcomes the major incorporation
of 5-azacytidine into RNA and reduces its side effects. Indeed, Decitabine is only
incorporated into DNA. However, it has been shown that cytidine deaminase can
degradate 5-aza-2 -deoxycytidine to 5-aza-2 -deoxyuridine (137), resulting in the
complete loss of DNMTs inhibition. The high level of cytidine deaminase in liver
and spleen may reduce the half-life of this compound to 1520 min when tested
in vivo (138). A Phase I clinical trial has suggested that deamination is the major
pathway for this compound (133).
In reference to new DNA demethylating agents, in addition to the previously
mentioned procaine and procainamide (129, 130), we should discuss zebularine [1(beta-D-ribofuranosyl)-1,2-dihydropyrimidin-2-one]. Zebularine is the first DNA
demethylating agent that can be administered orally and exhibits chemical stability
and minimal cytotoxicity both in vitro and in vivo (139, 140).
These compounds and their derivatives have been used in the clinic with some
therapeutic benefit, especially in hematopoietic malignancies such as myelodysplastic syndrome and acute myeloid leukemia (141143). Lower doses of 5azacytidine associated with inhibitors of histone deacetylases (such as trichostatin,
depsipeptide, SAHA, or sodium butyrate) may also reactivate tumor suppressor
genes (126). This was an encouraging discovery with respect to avoiding toxicity.
Hypermethylation of the CpG island is not a solitary phenomenon, but it occurs
4 Dec 2004
19:7
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
647
FINAL THOUGHTS
Cancer is a poligenetic disease, but it is also a poliepigenetic disease. We cannot understand the dynamics and plasticity of cancer cells if we do not invoke
epigenetic changes. This review has focused on DNA methylation alteration,
but the whole epigenetic setting of the transformed cell, including histone
4 Dec 2004
19:7
648
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
ESTELLER
7.
8.
9.
10.
methylation is associated with transcriptional silencing of the tumour suppressor p16/CDKN2/MTS1 in human cancers.
Nat. Med. 1:68692
Gonzalez-Zulueta M, Bender CM, Yang
AS, Nguyen T, Beart RW, et al. 1995.
Methylation of the 5 CpG island of the
p16/CDKN2 tumor suppressor gene in
normal and transformed human tissues
correlates with gene silencing. Cancer
Res. 55:453135
Herman JG, Merlo A, Mao L, Lapidus
RG, Issa JP, et al. 1995. Inactivation of
the CDKN2/p16/MTS1 gene is frequently
associated with aberrant DNA methylation in all common human cancers. Cancer Res. 55:452530
Clark SJ, Harrison J, Paul CL, Frommer M. 1994. High sensitivity mapping of
methylated cytosines. Nucleic Acids Res.
22:299097
Herman JG, Graff JR, Myohanen S,
Nelkin BD, Baylin SB. 1996. Methylation-specific PCR: a novel PCR assay for
methylation status of CpG islands. Proc.
Natl. Acad. Sci. USA 93:982126
4 Dec 2004
19:7
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
21.
22.
23.
24.
25.
26.
27.
28.
29.
649
4 Dec 2004
19:7
650
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
ESTELLER
41.
42.
43.
44.
45.
46.
47.
48.
4 Dec 2004
19:7
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
61.
62.
63.
64.
65.
66.
67.
68.
69.
651
4 Dec 2004
19:7
652
70.
71.
72.
73.
74.
75.
76.
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
ESTELLER
Schuebel KE, Markowitz SD, et al. 2004.
Epigenetic inactivation of SFRP genes allows constitutive WNT signaling in colorectal cancer. Nat. Genet. 36:41722
Kane MF, Loda M, Gaida GM, Lipman
J, Mishra R, et al. 1997. Methylation of
the hMLH1 promoter correlates with lack
of expression of hMLH1 in sporadic colon
tumors and mismatch repair-defective human tumor cell lines. Cancer Res. 57:808
11
Herman JG, Umar A, Polyak K, Graff
JR, Ahuja N, et al. 1998. Incidence
and functional consequences of hMLH1
promoter hypermethylation in colorectal
carcinoma. Proc. Natl. Acad. Sci. USA 95:
687075
Esteller M, Levine R, Baylin SB, Ellenson LH, Herman JG. 1998. MLH1
promoter hypermethylation is associated
with the microsatellite instability phenotype in sporadic endometrial carcinomas.
Oncogene 17:241317
Esteller M, Catasus L, Matias-Guiu X,
Mutter GL, Prat J, et al. 1999. hMLH1
promoter hypermethylation is an early
event in human endometrial tumorigenesis. Am. J. Pathol. 155:176772
Fleisher AS, Esteller M, Wang S, Tamura
G, Suzuki H, et al. 1999. Hypermethylation of the hMLH1 gene promoter in
human gastric cancers with microsatellite
instability. Cancer Res. 59:109095
Esteller M, Hamilton SR, Burger PC,
Baylin SB, Herman JG. 1999. Inactivation of the DNA repair gene O6methylguanine-DNA methyltransferase
by promoter hypermethylation is a common event in primary human neoplasia.
Cancer Res. 59:79397
Esteller M, Toyota M, Sanchez-Cespedes
M, Capella G, Peinado MA, et al.
2000. Inactivation of the DNA repair
gene O6-methylguanine-DNA methyltransferase by promoter hypermethylation
is associated with G to A mutations in
K-ras in colorectal tumorigenesis. Cancer
Res. 60:236871
4 Dec 2004
19:7
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
94.
95.
96.
97.
98.
99.
100.
101.
653
4 Dec 2004
19:7
654
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
ESTELLER
110.
111.
112.
113.
114.
115.
116.
117.
118.
119.
4 Dec 2004
19:7
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
120.
121.
122.
123.
124.
125.
126.
127.
128.
129.
130.
131.
132.
133.
134.
135.
136.
655
4 Dec 2004
19:7
656
137.
138.
139.
140.
141.
142.
AR
AR232-PA45-26.tex
XMLPublishSM (2004/02/24)
P1: JRX
ESTELLER
Randomized controlled trial of azacitidine
in patients with the myelodysplastic syndrome: a study of the cancer and leukemia
group B. J. Clin. Oncol. 20:242940
Laliberte J, Marquez VE, Momparler RL.
1992. Potent inhibitors for the deamination of cytosine arabinoside and 5aza-2 -deoxycytidine by human cytidine
deaminase. Cancer Chemother. Pharmacol. 30:711
Ho DH. 1973. Distribution of kinase and deaminase of 1-beta-D-arabinofuranosylcytosine in tissues of man and
mouse. Cancer Res. 33:281620
Cheng JC, Matsen CB, Gonzales FA, Ye
W, Greer S, et al. 2003. Inhibition of DNA
methylation and reactivation of silenced
genes by zebularine. J. Natl. Cancer Inst.
95:399409
Cheng JC, Weisenberger DJ, Gonzales
FA, Liang G, Xu GL, et al. Continuous
zebularine treatment effectively sustains
demethylation in human bladder cancer
cells. Mol. Cell Biol. 24:127078
Wijermans PW, Krulder JW, Huijgens
PC, Neve P. 1997. Continuous infusion
of low-dose 5-aza-2 -deoxycytidine in elderly patients with high-risk myelodysplastic syndrome. Leukemia 11:15
Schwartsmann G, Fernandes MS,
Schaan MD, Moschen M, Gerhardt
LM, et al. 1997. Decitabine (5-aza-2 -
143.
144.
145.
146.
147.
1/7/05
2:31 PM
Figure 3 Illustrative examples of DNA methylation analysis. (A) Chromatograms of the bisulfite genomic sequencing of a small fragment of a CpG island: left, unmethylated sequence (cytosines changed to thymines); right, methylated sequence (cytosines remined as
cytosines). (B) Methylation-specific PCR in primary tumors. The presence of a band under the M lanes represents hypermethylated neoplasms. (C) Staining for the 5-methylcytosine antibody in a cancer cell line.
HI-RES-PA45-26-Esteller.qxd
Page 1
C-2
1/7/05
2:31 PM
Figure 4 Cancer as a poligenetic and poliepigenetic disease: the model of mouse multistage skin carcinogenesis. Through all the tumoral
different stages (from benign lesions to invasive carcinomas) there is an accumulation of gene mutations, but also a double epigenetic
lesion: an increase in the number of genes undergoing methylation-associated silencing and in the degree of genomic hypomethylation
(14).
HI-RES-PA45-26-Esteller.qxd
Page 2
ESTELLER
10 Dec 2004
17:42
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
10.1146/annurev.pharmtox.45.120403.095802
Key Words
INTRODUCTION
As global human populations develop economically and technologically, human
physiological function is faced with challenges outside evolutionary experience.
The consequence is a paradigm shift in the causes of human mortality away from
extrinsic factors, such as infectious disease or nutritional insufficiency, and toward
failures of intrinsic physical function owing to longer life span or inherent genetic
abnormalities. These circumstances have resulted in a pandemic of heart disease.
In the United States alone, heart failure accounts for 400,000700,000 deaths
per year, $20$40 billion in yearly healthcare costs, and is the leading hospital
discharge diagnosis (1). These considerations provide the impetus for an ongoing
search for novel approaches to therapy.
Heart failure reflects the end result of a variety of primary or secondary causes,
including the hereditary and idiopathic cardiomyopathies, and the sequelae of
0362-1642/05/0210-0657$14.00
657
10 Dec 2004
17:42
658
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
BROWN ET AL.
10 Dec 2004
17:42
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
659
Figure 1
Structural organization of cardiac ECM. Schematic arrangement of fibrillar collagen in relation to cardiac myocytes and the coronary vasculature is shown.
Collagen weave surrounding individual myocytes and collagen struts tethering adjacent myocytes comprise the endomysium. Groups of myocytes are bundled within the
perimysium. The epimysium encloses groups of perimysial bundles. Capillaries and
coronary microvessels have free diffusion access to cardiac myocytes throughout the
ECM. Adapted from References 3 and 49.
are sequestered in inactive forms, serve important roles in regulating cell function
upon disruption of the ECM (6).
Extracellular matrix homeostasis involves ongoing cycles of synthesis and
degradation. Both synthetic and degradative aspects of collagen metabolism are
tightly regulated (5, 7). Fibrillar collagen is synthesized as a precursor polypeptide, exported from the cell, and proteolytically processed by removal of aminoand carboxy-terminal propeptides before insertion into nascent fibrils. Collagen
monomers are then cross-linked through hydroxyproline and hydroxylysine
residues to produce the mature structure. Mature fibrillar collagen is highly stable, with a turnover half-life of around 100 days in normal myocardium. Collagen biosynthesis is regulated transcriptionally by fibrogenic growth factors,
particularly TGF, and posttranscriptionally by the rate-limiting enzyme prolyl4-hydroxylase (5, 7). Collagen degradation is accomplished in stepwise fashion by members of the matrix metalloproteinase (MMP) superfamily, which are
10 Dec 2004
17:42
660
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
BROWN ET AL.
themselves under multiple levels of regulatory control. MMP production is regulated by three mechanisms: transcriptionally; posttranscriptionally by the activation of the proenzyme to the active form; and posttranslationally by endogenous
pseudosubstrate antagonists, the tissue inhibitors of metalloproteinases (TIMPs).
Of diagnostic significance, collagen metabolism has been monitored in vivo by
immunochemical measurement of key metabolic products and enzymes in serum
(Figure 2). Synthesis of collagens I or III produce stoichiometric equivalents of
the procollagen amino-terminal peptides (PINP, PIIINP) and procollagen carboxyterminal peptides (PICP and PIIICP), respectively, which are cleared from the circulation by the liver. Degradation of collagens I and III releases the corresponding
carboxy telopeptides, ICTP or IIICTP, which are excreted by the kidney (5, 7,
8). Serum concentrations of specific MMPs and TIMPs reflect the release of these
Figure 2
Extracellular metabolism of fibrillar collagen. Monomers of collagens I
and III are exported from the cell as propeptides and assembled. Amino- and carboxyterminal propeptides (P[I/III]NP and P[I/III]CP, respectively) are released into the
interstitial space during assembly. Mature collagen fibrils are further processed by
cross-linking through hydroxylated lysine and proline residues. Degradation of collagen fibrils by collagenase (MMP-1) releases stable carboxy-terminal telopeptides
([I/III]CTP).
10 Dec 2004
17:42
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
661
molecules into the interstitial space and are measured by ELISA techniques. These
methods have been validated to monitor altered ECM metabolism in the context
of heart disease (9).
Cardiac fibroblasts are activated in response to myocardial infarction (MI) or
injury and participate as key cells in the wound healing response. In common
with injury responses in other tissues, myocardial injury sets in motion a complex sequence of events involving coordinated interactions among multiple cell
types in the wound environment (10). The sequential steps in response to injury
include hemostasis, infiltration of immune and inflammatory cells, degradation
and phagocytosis of necrotic myocytes and cellular debris, repopulation of cardiac
fibroblasts within the zone of injury by chemotaxis and increased proliferation,
reconstruction of a granulomatous scar, and subsequent ECM remodeling to produce a mature scar. Thus, net ECM degradation, resulting from increased MMP
expression, dominates the initial phase of the injury response, whereas net ECM
deposition, arising from enhanced collagen synthesis, dominates the later phase
of healing.
Transitions of fibroblast phenotype and functional capabilities are regulated by
cytokines and growth factors released from other cell types and by the fibroblasts
themselves. Cardiac fibroblasts serve important roles as intermediate sensors and
amplifiers of signals from immune cells and myocytes, through production of autocrine and paracrine mediators such as cytokines, growth factors, prostaglandins,
and nitric oxide (NO) (reviewed in 10, 17). These agents are presumed to regulate the functional responses of cardiac fibroblasts through intracellular signaling networks, which converge at the level of transcription of coordinated gene
programs.
In the heart, as in other systems, termination of injury responses appears to
occur by apoptosis of activated cells (11). Chronic or repeated injury in the heart
ultimately overcomes the compensatory reactions of the myocardium. The ensuing progression to heart failure is accompanied by persistent inflammation and
fibrosis. The mechanisms that govern the resolution of acute injury responses,
versus the transition to chronic activation of cardiac fibroblasts, are not well
understood.
Myofibroblasts have been described as a specialized phenotype of activated
fibroblasts (12). These cells express contractile proteins, including smooth muscle -actin, vimentin, and desmin; effectively contract collagen gels in vitro;
and are postulated to be important for wound closure and structural integrity of
healing scars. In addition to normal wound healing, myofibroblasts are associated with hypertrophic fibrotic scars in injury models from multiple organ systems, and differentiation to the myofibroblast phenotype is strongly promoted by
the reference fibrogenic growth factor TGF. Myofibroblast apoptosis has been
associated with progression of granulomatous tissue to a mature scar, whereas
failure of myocyte apoptosis has been suggested to drive the progression to fibrosis (13). Cardiac myofibroblasts were shown to persist in mature infarct scars
(14).
10 Dec 2004
17:42
662
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
BROWN ET AL.
10 Dec 2004
17:42
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
663
to injurious stimuli. However, emerging data from other systems suggest three
possible sources: (a) epithelial-to-mesenchymal transition from epithelial cells
(26, 27); (b) recruitment of circulating, collagen-secreting, bone marrowderived
hematopoietic precursor cells described as fibrocytes (28, 29); and (c) activation of
resident fibroblasts (28, 30). More than one route of recruitment may occur within
a single tissue (26, 28). In irradiation-induced pulmonary fibrosis, Hashimoto et al.
observed that myofibroblasts arose from resident fibroblasts, whereas the principal collagen-producing fibroblasts actually derived from bone marrow recruitment
(28). By contrast, peritoneal myofibroblasts were reported to derive from circulating precursor cells (31). These observations provide exciting new insights into the
biology of wound healing.
10 Dec 2004
17:42
664
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
BROWN ET AL.
Hypertrophic Cardiomyopathy
Hypertrophic cardiomyopathy (HCM; defined as increased septum and LV wall
thickness inappropriate for hemodynamic load) in affected individuals may be
asymptomatic or lead to alternate outcomes, including atrial fibrillation, symptomatic cardiac dysfunction progressing to heart failure, or sudden cardiac death
(SCD) from ventricular tachycardia (34). The latter result is encountered in young
competition athletes, where sudden death may be the first symptom of HCM. At
the cellular level, HCM is most commonly associated with myocyte hypertrophy
and myocyte disarray, accompanied by replacement fibrosis at foci of myocyte
death.
HCM in most cases has been shown to arise from hereditary or spontaneous
mutations in myocyte sarcomeric proteins responsible for force generation (35).
This observation provides proof of the principle that a single gene defect in the
cardiac myocyte is sufficient to drive the full syndrome of cardiac hypertrophy
and failure. However, the relationship between mutational genotype and disease
phenotype is highly variable and not well understood (36). These findings clearly
suggest the presence of additional modifiers of disease.
The association of cardiac fibrosis with diastolic dysfunction and SCD has been
studied in HCM. Both myocyte disarray and fibrosis are associated with diastolic
dysfunction and electrical instability in HCM patients (37, 38). Varnava et al.
(39), in examination of autopsy specimens from HCM-induced sudden death,
concluded that myocyte disarray correlated most strongly with SCD in young
patients, whereas fibrosis was associated with SCD in older patients. By contrast,
a second autopsy study by Shirani et al. (40) found morphologic abnormalities
and increased amounts of ECM in children and young-adult SCD victims, arguing
that expanded ECM is involved early in the disease process. These data clearly
suggest an association between cardiac fibrosis and HCM severity, although the
link to SCD needs clarification.
10 Dec 2004
17:42
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
665
10 Dec 2004
17:42
666
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
BROWN ET AL.
10 Dec 2004
17:42
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
667
Agent class
Mechanism of action,
if known
Renin-angiotensin-aldosterone inhibitors
ACE
Enzymatic inhibitors of
inhibitors
angiotensin converting
enzyme, reduce Ang II
production
Angiotensin
AT1 receptor antagonists
receptor
blockers
Aldosterone
Mineralocorticoid receptor
antagonists
antagonists
References
(60, 61,
6365)
(6669)
(7173)
Endothelin receptor
antagonists
(77)
Statins
HMG-CoA reductase
inhibitors
(8183)
(84)
Cytokine therapies
Anti-TNF
IFN
Inhibit myofibroblasts,
collagen synthesis
Anti-TGF
(85)
(89)
MMP inhibitors
(103, 123,
124)
Pirfenidone
(109, 110)
Tranilast
(111)
Nuclear receptor
agonists
PPAR agonists
(114, 115)
PPAR agonists
(118)
10 Dec 2004
17:42
668
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
BROWN ET AL.
10 Dec 2004
17:42
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
669
fibrotic patients than in patients with nonsevere fibrosis. These results suggest
heterogeneity of fibrosis responses to therapy (69).
In summary, inhibitors of the RAS clearly appear to derive a significant portion
of their therapeutic benefit from actions on cardiac fibroblasts and fibrotic remodeling of the heart. Despite a tremendous amount of research, however, the specific
mechanisms of these actions, and the underlying role of the RAS in myocardial remodeling and homeostasis remain enigmatic. Identifying the primary mechanisms
should provide further opportunities for therapeutic development.
10 Dec 2004
17:42
670
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
BROWN ET AL.
therapy that demonstrates the value of targeting cardiac fibrosis to improve cardiac
performance and prognosis in appropriately selected patients.
Statins
The proven clinical benefit of statins as cholesterol-lowering drugs in atherosclerotic disease was more recently complemented by the unexpected finding that these
agents exert pleiotropic effects on a variety of cell signaling pathways through inhibition of protein prenylation (78, 78a). These observations have fueled interest
in the potential utility of statins in heart failure. Studies in experimental animal
models provide support for beneficial actions of statins directed toward cardiac
fibroblasts. Treatment with statins reduces myocardial remodeling, fibrosis, and
collagen synthesis in models of myocardial injury including surgical infarction
(79), transgenic models of hypertrophic cardiomyopathy (79a) or NaCl-induced
pressure overload (80). In many of these studies, statins exert concordant antifibrotic and antiinflammatory actions. These results underscore that inflammation
and fibrosis represent aspects of a continuum of responses of cardiac fibroblasts
to myocardial injury.
In theory, statin therapy might confer either positive or negative impacts in
the setting of congestive heart failure (discussed in 81). Statins may act on cardiac fibroblasts to attenuate inflammatory signaling through reduced prenylation
of small GTPases. In this regard, elevated serum concentrations of the inflammatory marker C-reactive protein were shown to predict the likelihood of nonfatal
MI or fatal coronary events following an initial infarct. Treatment with pravastatin normalized serum concentrations of CRP and reduced the risks of cardiac
events equivalent to normal subjects (82). Moreover, amelioration of coronary
artery disease and ischemic events would relieve proapoptotic stress on cardiac
myocytes and indirectly diminish replacement fibrosis. The large-scale prospective CORONA trial is underway to explicitly test therapy with rosuvastatin in heart
failure (83). This trial does not include measurement of fibrotic or inflammatory
markers as a primary endpoint, but such a substudy would be of considerable
interest.
10 Dec 2004
17:42
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
671
10 Dec 2004
17:42
672
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
BROWN ET AL.
10 Dec 2004
17:42
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
673
The effects of MMP inhibitors in the progression to heart failure have also been
examined. Pacing-induced supraventricular tachycardia of three-weeks duration
in pigs produces congestive heart failure characterized by LV dilation; increased
activities of MMP-1, -2, and -3; and decreased collagen content. Pathological remodeling was attenuated and cardiac function was preserved by treatment with
the nonselective MMPI, PD-166 793 (100). Similar results were obtained more
recently with a newer generation MMPI, PGE 7113313, designed to spare inhibition of MMP-1, which is downregulated in chronic human heart failure (101).
It should be noted that inhibitors were administered prior to and throughout the
duration of pathological stimulus rather than after the establishment of congestive
heart failure.
Preliminary studies in humans lend further support for MMPs as targets in heart
disease (reviewed in 94). Gene polymorphisms have been identified in the promoters of MMP-1, -3, -9, and -12, which influence MMP expression. Polymorphisms
in the promoters for MMP-3, -9, and -12 were shown to confer susceptibility to
coronary artery disease and abdominal aortic aneurysm. Concentration ratios of
serum MMP/TIMP correlate with functional values of LV volume and ejection
fraction and predict clinical outcome of myocardial infarction. A Phase II trial
was recently completed to test the effect of MMP inhibitor PG 116800 to prevent adverse cardiac remodeling following a first myocardial infarction (102). The
tetracycline derivative Periostat (doxycycline) is the only MMP inhibitor currently
approved for clinical use, but its application is limited to periodontal disease. Treatment of coronary heart disease patients with Periostat reduced serum inflammatory
markers (C-Reactive Protein, IL-1, and IL-6) as well as circulating concentrations
of MMP-9 (103). Considering its safety, efficacy, and cost, clinical trials of doxycycline in myocardial remodeling also appear worth pursuing.
10 Dec 2004
17:42
674
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
BROWN ET AL.
10 Dec 2004
17:42
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
675
type in myocardial remodeling. Questions remain that bear on the therapeutic issues
discussed above and that may provide novel opportunities for drug development.
First, what are the fibroblast phenotypes of normal, injured, and failing myocardium? It will be important to resolve whether a single activated fibroblast
phenotype is capable of multiple functional responses, such as proliferation, migration, ECM metabolism, and production of autocrine/paracrine mediators, or
whether different subsets of fibroblasts subserve distinct endpoint responses. To
address this question, it will be necessary to define the cellular precursors of
activated fibroblasts; for example, whether activated fibroblasts derive from homogeneous or polyclonal populations of quiescent resident fibroblasts, from epithelialmesenchymal transitions of undifferentiated resident precursor cells, or from recruitment of circulating precursor cells. These same issues need to be addressed
for cardiac myofibroblasts.
These ideas lead to related thoughts about the mechanisms that underlie the termination of the myocardial injury response in normal wound healing compared to
the transition to maladaptive responses in fibrotic heart failure. One may speculate
that heart failure progression results from unresolved cardiac myocyte dysfunction.
Within this environment of chronic injury, cardiac fibrosis could reflect either a failure to terminate a normal injury phenotype, or alternatively, the de novo appearance
of a novel failure phenotype of cardiac fibroblasts. The role of fibroblast apoptosis
as a termination mechanism in adaptive myocardial healing versus the role of hyperplasia as a mechanism for fibroblast recruitment in fibrotic progression should
be a specific focus for investigation. Identification of populations of (myo)fibroblasts that are involved in distinct functional aspects or disease stages of myocardial
remodeling would offer obvious opportunities for therapeutic intervention.
Continued research on the signaling mechanisms that regulate cardiac fibroblast
phenotype in response to inflammatory and fibrotic stimuli represents a second major area of emphasis. The unique properties of the cardiac fibroblast relative to other
fibroblastic cells, the ability of cardiac fibroblasts to integrate many stimuli through
receptor-specific signaling pathways, and the diversity of endpoint responses all
point to a more complex regulatory organization than has been appreciated (Figure 3). The functional response of regulated gene expression reflects combinatorial
inputs in the evolving wound environment. These regulatory interactions underlie
the spatiotemporal sequencing of events that occurs in wound healing and provide
the basis for determining the intervals of opportunity for defined therapeutic targets. This model further suggests that genomic or proteomic pattern recognition
analyses will help to identify groups of genes that are expressed in concert, in
turn generating hypotheses for conserved regulatory motifs within the gene promoters and associated signaling pathways. Additional levels of complexity come
from the discovery that signaling molecules as well as transcriptional regulators
are spatially organized within the cell (119, 119a). In theory, these elements offer
a wealth of targets for therapy to intersect the inflammatory-fibrotic cascade, but
the dissection of key mechanisms will require acquisition of detailed molecular
information. Genomic analyses of myocardial injury models, and particularly of
10 Dec 2004
17:42
676
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
BROWN ET AL.
Figure 3
Signaling and transcriptional regulation of fibroblast function. Cardiac
fibroblasts respond to diverse humoral and mechanical inputs through cell surface
receptors. Environmental stimuli are integrated through receptor-mediated signaling
pathways leading to activation of nuclear transcription factors and altered gene expression. Integration of nuclear and cytoplasmic signaling networks controls fibroblast
phenotype. Abbreviations: ALDO, aldosterone; R, receptor; GPCR, G proteincoupled
receptor; FAK, focal adhesion kinase; IL-1 RA, IL-1 receptor antagonist; MAPK, mitogen activated protein kinase.
cardiac fibroblasts, are at an early stage, but a torrent of data is surely coming.
We are unaware of clinical trials utilizing agents that focus on cardiac fibroblast
intracellular signaling pathways. However, a number of agents are in preclinical or
early phase clinical trials for other applications, and it is likely that coming years
will see their evaluation in myocardial remodeling (120).
A third area for research will be to explore further the mechanisms of intercellular communication in the normal and failing heart. Homeostatic maintenance
and remodeling of the heart requires communication among cardiac myocytes, fibroblasts, and immune cells, as well as interactions with the coronary vasculature
(Figure 4). Mechanotransduction via integrins and the extracellular matrix, direct
cell-cell communication via gap junctions, or humoral transmission by diffusible
chemokines and cytokines all may contribute to this process. Therapies aimed at
intramyocardial signaling by AngII and aldosterone demonstrate the value of this
approach.
Conversely, cardiovascular exercise training has been shown to promote physiologically adaptive cardiac hypertrophy and nonfibrotic ECM remodeling, which
are associated with improved cardiac performance, and protect against the risk
of adverse cardiac events (121, 121a). These results raise questions of how the
10 Dec 2004
17:42
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
677
myocardium distinguishes the healthy stress of exercise from the pathologic stress
of injury. It is intriguing to speculate that exercise activates survival pathways in
cardiac myocytes, resulting in intercellular signals to cardiac fibroblasts, which are
distinct from signals of injury (121b). Knowledge of cardiac fibroblast responses
to exercise training may provide additional insight into approaches to oppose the
progression of fibrosis. An observational study is ongoing to evaluate cardiac fibrosis by serum collagen markers and MRI in relation to exercise tolerance in
hypertrophic cardiomyopathy (122).
10 Dec 2004
17:42
678
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
BROWN ET AL.
basic and pharmaceutic research sectors so far have yielded a single approved agent,
doxycycline (Periostat), for periodontal disease. Lessons from these experiences
have been thoughtfully reviewed and are relevant to cardiac fibroblasts (123, 124).
Cardiac fibrosis is not a primary cause, but rather a disease modifier, that affects
the progress, severity, and outcome of heart disease. Furthermore, fibrotic remodeling of the heart represents a spectrum of responses that may vary depending on
the specific etiology of myocardial injury; the stage of disease progression; and
differences in age, gender, ethnicity, and genetic polymorphisms between individual patients. It therefore is essential to develop algorithms for stratification of risk
in large patient populations. Preliminary results discussed above suggest that more
extensive fibrosis may confer measurable risk and that individuals differ in their
responses to therapy. However, these studies are in their initial stages compared to
the quantitative databases that have accumulated for prognostic indicators, such
as LV hypertrophy, serum cholesterol, or inflammatory status. Analysis of serum
collagen metabolites in archival samples from large-scale clinical trials in relation
to clinical outcome could offer a cost-effective starting point to obtain this type of
information.
Development and standardization of surrogate markers of fibroblast activation
or fibrosis are prerequisite to risk stratification and to evaluation of the efficacy
of pharmaceutical agents. The utility and economy of serum collagen metabolites
have been validated in this regard (5, 79). However, these measurements are
limited because of their lack of specificity for cardiac versus extracardiac ECM
remodeling, and the lack of sensitivity to detect key mechanistic steps in the
remodeling process. Analysis of coronary sinus blood may provide a means to
identify specific markers of cardiac ECM metabolism compared to general ECM
markers in the systemic circulation.
Selection of appropriate clinical endpoints is critical to assess therapeutic benefit
and to evaluate the relationship between target efficacy and clinical outcome.
Reduction in mortality may be the preferred endpoint in younger patients, whereas
reversal of disease progression or adverse events may assume greater importance
in the elderly.
Antifibrotic agents will likely be useful adjuncts for combination therapy in
selected patient populations, as is currently recommended for aldosterone antagonists. This approach offers the appeal of targeting complementary therapies for
cardiac myocytes and cardiac fibroblasts. The development of agents that combine
antifibrotic and antiinflammatory actions offers promise. However, issues of cost
and safety of polypharmacy, especially in older patients, must be considered.
Identification of the windows of opportunity for antifibrotic therapy is likely
to be crucial for effective intervention. Myocardial remodeling is progressive and
cumulative as the heart passes from the initial response to injury through the
transition to fibrosis and failure. It will be necessary to apply therapy at intervals
that are appropriate to the molecular target. Furthermore, it may be more feasible
to prevent fibrotic remodeling than to reverse it once it occurs. This is most likely
to be the case for replacement fibrosis following myocyte loss. On the other hand,
10 Dec 2004
17:42
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
679
reversal of reactive fibrosis may be more feasible as has been seen with regression
of LVH and fibrosis in hypertension.
Finally, choice of physiologically appropriate experimental models is important to extrapolate preclinical findings to positive results in human trials. Studies
with cultured fibroblasts in vitro typically examine responses to acute stimuli, and
animal studies examine the initial onset of myocardial remodeling and failure in
younger animals. By contrast, human heart failure is more commonly a progressive disease of the elderly, and clinical trials are aimed at therapy of established
disease. Transgenic mouse technology has provided powerful insights into the
roles of specific gene products in cardiovascular physiology, but the consequences
of transgene expression throughout the life span of the animal may not accurately reflect the sequential and coordinated activation of myocardial remodeling
in response to a pathological insult. Newer methodologies allowing conditional
transgenic expression will help address this disparity.
In conclusion, the available evidence provides a strong rationale for therapies
directed toward cardiac fibroblasts to improve outcomes in heart disease. Continued basic and translational research, and perseverance through the inevitable
setbacks, will be needed to bring this promise to fruition.
ACKNOWLEDGMENT
Work in the authors laboratories was supported by NIH (HL59428).
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Graziano JM. 2001. Global burden of
cardiovascular disease. In Heart Disease: A Textbook of Cardiovascular
Medicine, ed. E Braunwald, DP Zipes,
P Libby, pp. 118. Philadelphia: WH
Saunders Co. 6th ed.
2. Richardson P, McKenna W, Bristow M,
Maisch B, Mautner B, et al. 1996. Report of the 1995 World Health Organization/International Society and Federation of Cardiology Task Force on the
definition and classification of cardiomyopathies. Circulation 93:84142
3. Eghbali M, Weber KT. 1990. Collagen
and the myocardium: fibrillar structure,
biosynthesis and degradation in relation
to hypertrophy and its regression. Mol.
Cell Biochem. 96:114
10 Dec 2004
17:42
680
8.
9.
10.
11.
12.
13.
14.
15.
16.
16a.
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
BROWN ET AL.
system in extracellular matrix remodeling. Methods Find. Exp. Clin. Pharmacol. 25:54164
Lopez B, Gonzalez A, Varo N, Laviades
C, Querejeta R, et al. 2001. Biochemical assessment of myocardial fibrosis in
hypertensive heart disease. Hypertension
38:122226
Querejeta R, Varo N, Lopez B, Larman M, Artinano E, et al. 2000. Serum
carboxy-terminal propeptide of procollagen type I is a marker of myocardial fibrosis in hypertensive heart disease. Circulation 101:172935
Davidson JM. 1992. Wound repair. In
Inflammation: Basic Principles and
Clinical Correlates, ed. JI Gallin, RH
Goldstein, R Snyderman, pp. 809
19. New York: Raven Press. 2nd
ed.
Takemura G, Ohno M, Hayakawa Y,
Misao J, Kanoh M, et al. 1998. Role of
apoptosis in the disappearance of infiltrated and proliferated interstitial cells
after myocardial infarction. Circ. Res.
82:113038
Powell DW, Mifflin RC, Valentich JD,
Crowe SE, Saada JI, et al. 1999. Myofibroblasts. II. Intestinal subepithelial
myofibroblasts. Am. J. Physiol. Cell
Physiol. 277:C183201
Zhang H-Y, Phan SH. 1999. Inhibition
of myofibroblast apoptosis by transforming growth factor 1. Am. J. Respir. Cell
Mol. Biol. 21:65865
Sun Y, Weber KT. 2000. Infarct scar:
a dynamic tissue. Cardiovasc. Res. 46:
25056
Gittenberger-de Groot AC. 2003. Mannheimer lecture. The quintessence of the
making of the heart. Cardiol. Young
13:17583
Wessels A, Perez-Pomares JM. 2004.
The epicardium and epicardially derived
cells (EPDCs) as cardiac stem cells.
Anat. Rec. 276A:4357
Kalluri R, Neilson EG. 2003. Epithelialmesenchymal transition and its implica-
17.
18.
19.
20.
21.
22.
23.
24.
10 Dec 2004
17:42
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
35.
36.
37.
38.
39.
40.
41.
42.
43.
681
10 Dec 2004
17:42
682
44.
45.
46.
47.
48.
49.
50.
51.
52.
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
BROWN ET AL.
the dilated non-ischaemic myocardium.
Heart 89:125556
Marijianowski MM, Teeling P, Mann J,
Becker AE. 1995. Dilated cardiomyopathy is associated with an increase in the
type I/type III collagen ratio: a quantitative assessment. J. Am. Coll. Cardiol.
25:126372
Yokoseki O, Yazaki Y, Suzuki J, Imamura H, Takenaka H, et al. 2000. Association of matrix metalloproteinase expression and left ventricular function in
idiopathic dilated cardiomyopathy. Jpn.
Circ. J. 64:35257
Klappacher G, Franzen P, Haab D,
Mehrabi M, Binder M, et al. 1995. Measuring extracellular matrix turnover in
the serum of patients with idiopathic
or ischemic dilated cardiomyopathy and
impact on diagnosis and prognosis. Am.
J. Cardiol. 75:91318
Lip GY, Felmeden DC, Li-Saw-Hee FL,
Beevers DG. 2000. Hypertensive heart
disease. A complex syndrome or a hypertensive cardiomyopathy? Eur. Heart J.
21:165365
Kenchaiah S, Pfeffer MA. 2004. Cardiac remodeling in systemic hypertension. Med. Clin. North Am. 88:11530
Rossi MA. 1998. Pathologic fibrosis and
connective tissue matrix in left ventricular hypertrophy due to chronic arterial
hypertension in humans. J. Hypertens.
16:103141
Ohsato K, Shimizu M, Sugihara N, Konishi K, Takeda R. 1992. Histopathological factors related to diastolic function
in myocardial hypertrophy. Jpn. Circ. J.
56:32533
Eckardt L, Haverkamp W, Johna R,
Bocker D, Deng MC, et al. 2000.
Arrhythmias in heart failure: current
concepts of mechanisms and therapy.
J. Cardiovasc. Electrophysiol. 11:106
17
Pogwizd SM, Hoyt RH, Saffitz JE, Corr
PB, Cox JL, et al. 1992. Reentrant and
focal mechanisms underlying ventricu-
53.
54.
55.
56.
57.
58.
59.
60.
61.
10 Dec 2004
17:42
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
70.
71.
72.
73.
74.
75.
76.
77.
683
10 Dec 2004
17:42
684
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
BROWN ET AL.
85.
86.
87.
88.
89.
90.
91.
10 Dec 2004
17:42
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
100.
101.
102.
103.
104.
105.
106.
107.
108.
685
10 Dec 2004
17:42
686
109.
110.
111.
112.
113.
114.
115.
116.
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
BROWN ET AL.
proliferation and collagen production.
J. Clin. Endocrinol. Metab. 83:21923
Raghu G, Johnson WC, Lockhart D,
Mageto Y. 1999. Treatment of idiopathic
pulmonary fibrosis with a new antifibrotic agent, pirfenidone: results of a
prospective, open-label phase II study.
Am. J. Respir. Crit. Care Med. 159:1061
69
Fukagawa M, Noda M, Shimizu T,
Kurokawa K. 1999. Chronic progressive interstitial fibrosis in renal disease
are there novel pharmacological approaches? Nephrol. Dial. Transplant.
14:279395
Holmes DR Jr, Savage M, LaBlanche
JM, Grip L, Serruys PW, et al. 2002. Results of prevention of REStenosis with
tranilast and its outcomes (PRESTO)
trial. Circulation 106:124350
Kagitani S, Ueno H, Hirade S, Takahashi T, Takata M, et al. 2004. Tranilast attenuates myocardial fibrosis in association with suppression of monocyte/
macrophage infiltration in DOCA/salt
hypertensive rats. J. Hypertens. 22:
100715
Pinto YM, Pinto-Sietsma SJ, Philipp T,
Engler S, Kossamehl P, et al. 2000.
Reduction in left ventricular messenger RNA for transforming growth factor 1 attenuates left ventricular fibrosis and improves survival without
lowering blood pressure in the hypertensive TGR(mRen2)27 rat. Hypertension
36:74754
Francis GA, Annicotte JS, Auwerx J.
2003. PPAR- effects on the heart and
other vascular tissues. Am. J. Physiol.
Heart Circ. Physiol. 285:H19
Schiffrin EL, Amiri F, Benkirane K,
Iglarz M, Diep QN. 2003. Peroxisome
proliferator-activated receptors: vascular
and cardiac effects in hypertension. Hypertension 42:66468
Wayman NS, Hattori Y, McDonald MC,
Mota-Filipe H, Cuzzocrea S, et al. 2002.
Ligands of the peroxisome proliferator-
117.
118.
119.
119a.
120.
121.
121a.
121b.
10 Dec 2004
17:42
AR
AR232-PA45-27.tex
XMLPublishSM (2004/02/24)
P1: JRX
687
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
10.1146/annurev.pharmtox.44.101802.121444
Key Words
interaction
Abstract Recent studies have revealed the import role played by transporters in
the renal and hepatobiliary excretion of many drugs. These transporters exhibit a broad
substrate specificity with a degree of overlap, suggesting the possibility of transportermediated drug-drug interactions with other substrates. This review is an overview of
the roles of transporters and the possibility of transporter-mediated drug-drug interactions. Among the large number of transporters, we compare the Ki values of inhibitors
for organic anion transporting polypeptides (OATPs) and organic anion transporters
(OATs) and their therapeutic unbound concentrations. Among them, cephalosporins
and probenecid have the potential to produce clinically relevant OAT-mediated drugdrug interactions, whereas cyclosporin A and rifampicin may trigger OATP-mediated
ones. These drugs have been reported to cause drug-drug interactions in vivo with OATs
or OATP substrates, suggesting the possibility of transporter-mediated drug-drug interactions. To avoid adverse consequences of such transporter-mediated drug-drug
interactions, we need to be more aware of the role played by drug transporters as well
as those caused by drug metabolizing enzymes.
INTRODUCTION
The kidney and the liver play important roles in the elimination of drugs and xenobiotics from the body (15). Cumulative in vivo and in vitro studies have revealed
the importance of transporters in the renal and hepatobiliary excretion of many
drugs and other xenobiotics (15). Recent studies to investigate the molecular
mechanism of renal and hepatobiliary excretion have revealed that multiple transporters are expressed in the kidney and liver in animals and humans, as well as
revealing their function, tissue distribution, and intracellular localization (615).
These transporters exhibit broad substrate specificity with a degree of overlap.
0362-1642/05/0210-0689$14.00
689
4 Dec 2004
19:14
690
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
SHITARA▪SATO▪SUGIYAMA
Figure 1
Mechanism of drug elimination in the kidney. Drug elimination in the
kidney takes place by glomerular filtration and secretion at the proximal tubules. However, they may return to the systemic circulation via a process of drug reabsorption.
Transporters are involved in drug secretion and reabsorption.
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
691
1.
where fu , GFR, FR, CLsec , QR , and CLR,int represent protein unbound fraction
in the blood, glomerular filtration rate [ml min1 ], the fraction reabsorbed, renal
secretion clearance, renal blood flow rate, and intrinsic clearance of tubular secretion, respectively (4). FR and CLR,int are partly saturable and can be inhibited,
suggesting the possibility of drug-drug interactions.
QH fu CLH,int,all
,
QH + fu CLH,int,all
2.
where CLH , QH , and CLH,int,all represent the hepatic clearance; hepatic blood flow;
and overall intrinsic clearance of biliary excretion, including uptake, metabolism,
and biliary excretion, respectively. CLH,int,all can be described by the following
equation (3, 19):
CLH,int,all = PSinflux
CLH,int
,
PSefflux + CLH,int
3.
where PSinflux and PSefflux are the membrane permeability across the sinusoidal
membrane from the outside to the inside and from the inside to the outside of cells,
respectively, and CLH,int represents the exact intrinsic clearance for the metabolism
and/or biliary excretion of the unbound drugs. When CLH,int is negligibly low
4 Dec 2004
19:14
692
AR
AR232-PA45-28.tex
SHITARA
SATO
XMLPublishSM (2004/02/24)
P1: JRX
SUGIYAMA
Figure 2 Mechanism of drug elimination in the liver. Drugs are taken up into hepatocytes via transporters and/or passive diffusion, followed by metabolism and/or
biliary excretion. Drugs are possibly effluxed into the circulation via sinusoidal
membrane.
CLH,int
.
PSefflux
4.
5.
It should be noted that the uptake of drugs via the sinusoidal membrane (PSinflux ),
which is partly mediated by transporters, is a determinant of the net hepatic clearance regardless of the other processes, i.e., CLH,int and PSefflux . Therefore, hepatic clearance may be affected when the uptake clearance of drugs is altered,
even if the drug finally undergoes metabolism. On the other hand, the excretion of drugs via the bile canalicular membrane, which is partly mediated by
transporters, is a determinant of the net hepatic clearance, unless PSefflux is negligibly low compared with CLH,int . Therefore, except in this case, the change
in the biliary excretion may affect the net hepatic clearance. If PSefflux is much
lower than CLH,int , only a drastic reduction in the biliary excretion will affect
the net hepatic clearance, possibly leading to a transporter-mediated drug-drug
interaction.
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
693
Figure 3
Transporters in the kidney. Transporters expressed in human and rodent kidney
are summarized in this figure. Some of the transporters in rodents are expressed only in
either rats or mice.
4 Dec 2004
19:14
694
AR
AR232-PA45-28.tex
SHITARA
SATO
XMLPublishSM (2004/02/24)
P1: JRX
SUGIYAMA
The transport of organic anions is mainly mediated by organic anion transporters (OATs). Rat Oat1 has been isolated as a
renal transporter that is involved in the renal uptake of organic anions like paminohippuric acid (PAH) in an exchange of dicarboxylates (20). OAT15 have
been identified as human OAT family transporters (2124). Among them, OAT14
are expressed in the human kidney and OAT2 and 5 are expressed in the liver (21
24). In the kidney, OAT13 are localized on the basolateral membrane, whereas
OAT4 is localized on the brush border membrane (24). Each of these transporters
in the OAT family has a similar substrate specificity. These transporters accept organic anions with a relatively small molecular weight with some exceptions. They
accept PAH, methotrexate (MTX), nonsteroidal antiinflammatory drugs, and antiviral nucleoside analogues as substrates (2527). They also accept more lipophilic
organic anions, such as estrone 3-sulfate and ochratoxin A, and even an organic
cation, cimetidine (24, 25, 28).
P-GLYCOPROTEIN
PEPTIDE TRANSPORTERS
OCT1 and OCT2 are expressed in the kidney, whereas only OCT1 is expressed in the liver (14, 42). OCT2 is highly expressed in the kidney (14, 42). In human kidney, these transporters are localized in
the basolateral membrane and are important organic cation transporters for renal
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
695
tubular secretion (14, 42). These are pH-independent, electrogenic, and polyspecific transporters (14, 42). These transporters accept organic cations with relatively
low molecular weight (type I cations), such as tetraethylammonium (TEA), as substrates (42, 43). Cimetidine, choline, dopamine, acyclovir, and zidovudine are also
reported to be substrates (4446).
OCTN1 is strongly expressed in the kidney but not in the adult liver (47).
In OCTN1-expressing HEK293 cells, pH-sensitive uptake of TEA has been observed (47). An inward proton concentration gradient stimulated the efflux of TEA
in OCTN1-expressing Xenopus leavis oocytes, indicating that OCTN1-mediated
transport couples with proton antiport (48). OCTN1 is considered to be localized on the brush border membrane of the kidney. Substrates include quinidine
and adriamycin as well as TEA (47, 48). OCTN2, an isoform of OCTN1, was
isolated from human placenta and it was also found to be expressed in the kidney
(49, 50). Although OCTN2 accepts TEA as its substrate, the transporter activity
is not as high as that of OCTN1. OCTN2 can also accept carnitine, a zwitterion
that is a cofactor essential for -oxidation of fatty acids, and several mutations in
mRNA encoding OCTN2 result in systemic carnitine deficiency owing to the poor
renal reabsorption of carnitine (51). This fact suggests that OCTN2 plays a role
in the renal reabsorption of carnitine. This transporter also accepts cephaloridine
and other cationic compounds, such as verapamil, quinidine, and phyrilamine, in
addition to TEA and carnitine, although it is not yet known whether this transporter
takes part in the renal reabsorption and/or excretion of these compounds together
with OCTN1 (52, 53).
OCTN
4 Dec 2004
19:14
696
AR
AR232-PA45-28.tex
SHITARA
SATO
XMLPublishSM (2004/02/24)
P1: JRX
SUGIYAMA
Figure 4 Transporters in the liver. Transporters expressed in human and rodent liver are
summarized in this figure. Some of the transporters in rodents are expressed only in either
rats or mice.
suggesting there may be other drugs that are taken up into the liver via OATP
family transporters.
In TR and Eisai hyperbilirubinemic rats (EHBRs), which exhibit hyperbilirubinemia owing to a deficiency in
the biliary excretion of bilirubin glucuronide, mutations in the transporter, multidrug resistance-associated protein 2 (Mrp2), have been found (66, 67). This
finding and the comparison of the biliary excretion of several compounds between normal rats and Mrp2-deficient rats suggests that it plays an important
role in the biliary excretion of multispecific organic anions, including glucuronide
conjugates [bilirubin glucuronide, E3040 glucuronide, estradiol 17-D-glucuronide (E2 17G), grepafloxacin glucuronide, SN-38 glucuronide, glycyrrhizin, etc.],
glutathione conjugate [glutathione bimane (GSB), dinitrophenyl glutathione (DNPSG), leucotrienes (LTC4 , D4 and E4 ), etc.], grepafloxacine, MTX, pravastatin, SN38, and temocaprilat (8, 6872). This transporter is localized in the bile canalicular membrane of the liver (7375). Its human counterpart (MRP2) has been
isolated from the cisplatin-resistant tumor cells, KCP4 (76). This transporter is
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
697
also localized on the bile canalicular membrane of the liver and accepts multiple organic anions, including glucuronides (bilirubin monoglucuronide, bilirubin
bisglucuronide, E3040 glucuronide, E2 17G, grepafloxacin glucuronide, LTC4 ,
etc.), glutathione conjugates (DNP-SG, GSB, glutathione-methylfluorescein, etc.),
pravastatin, MTX, vinblastine, vincristine, etoposide, etc. (33, 7788).
Human MRP3 is also a MRP family transporter, which is expressed in the liver.
However, this transporter is localized on the sinusoidal membrane and considered
to be involved in secretion into the blood. It also accepts many glucuronides,
glutathione conjugates, MTX, etc. (89, 90).
Breast cancer resistant protein (BCRP) has
been cloned from human MCF-7 breast cancer cells as a multidrug resistance
transporter (9193). This transporter also belongs to the ABC transporter family
(9193). However, its structure differs from that of other ABC transporters, such as
MDR1 and MRP, which contain two tandem repeats of transmembrane and ABC
domains. BCRP consists of only one ABC and one transmembrane domain, and,
therefore, it is referred to as a half-sized ABC transporter (9193). This transporter
is also expressed in normal tissues including the liver (94). In the liver, it is located
on the bile canalicular membrane (94). Many sulfated conjugates, such as estrone
3-sulfate (E1 S), dehydroepiandrosterone sulfate, 4-methylumbelliferone sulfate,
etc., are transported by BCRP (95). MTX, estradiol 17-D-glucuronide, and 2,4dinitrophenyl-S-glutathione are also transported but to a lesser extent compared
with E1 S (95). BCRP preferentially transports sulfate conjugates (95).
KIDNEY SLICES
4 Dec 2004
19:14
698
AR
AR232-PA45-28.tex
SHITARA
SATO
XMLPublishSM (2004/02/24)
P1: JRX
SUGIYAMA
uptake of compounds via Oct2, Oat1, Oat3, and a novel peptide transporter has been
observed (9699). The uptake of compounds into kidney slices is much lower than
the renal intrinsic uptake clearance in vivo, which may be partly due to diffusion
from the surface of the slices. Although extrapolation of the renal clearance using
kidney slices has not been reported, it may be used as a tool for the prediction of
transporter-mediated drug-drug interactions in the kidney. At the time of writing,
there have been no reports using human kidney slices; however, the use of this tool
will be useful for the prediction of transporter-mediated drug-drug interactions in
the kidney.
Isolated and cultured hepatocytes have
been used as an in vitro model of the liver (100, 101). The hepatic uptake of
peptidic endothelin antagonists using isolated rat hepatocytes showed that their in
vitro uptake clearance could be extrapolated to give their in vivo uptake clearance,
assuming a well-stirred model (100). Thus, isolated hepatocytes are a good tool
for the evaluation of drug uptake in the liver and transporter-mediated drug-drug
interactions in the liver. Recently, because of the progress in the techniques of
cryopreservation, it seems possible to preserve frozen human hepatocytes in such
a way that most of their enzymatic activity is retained (102). They have been used
to examine drug metabolism interactions, including induction of metabolic enzymes (103105). Recently, we have examined the uptake of taurocholate (TC)
and estradiol-17-D-glucuronide in freshly isolated and cryopreserved human hepatocytes (106). This study suggested that their active transports were retained even
in cryopreserved human hepatocytes, although the activity was decreased after
cryopreservation in some lots of hepatocytes (106). Therefore, cryopreserved human hepatocytes, at least, retain transporter function and they can be used as a
useful experimental system for examining the mechanism of the hepatic uptake of
drugs and interactions with other drugs (106).
Liver slices are also used for the study of drug uptake in the liver
(107, 108). Olinga et al. examined the uptake of digoxin, a substrate of human
OATP8 [OATP1B3] and rat Oatp2 [Oatp1a4], and temperature-dependent uptake
was observed (107). Liver slices are supplemented with nonparenchymal cells,
and, therefore, the interaction between hepatocytes and other cells and the effect
of other cells on the function of hepatocytes can also be examined.
LIVER SLICES
MEMBRANE VESICLES
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
699
requires a driving force for transport, so it is impossible to use this system without
prior characterization.
Using transporter expression systems,
the kinetic parameters for the target transporter can be obtained. Once the responsible transporter for the drugs in question has been identified, the possibility
of drug-drug interactions can be examined using the gene expression system,
i.e., without hepatocytes, membrane vesicles, and tissue slices. As human tissue
samples are scarcely distributed, transporter-expressing systems greatly help drug
transport studies. With the information of contributions of specific transporter(s)
to the total uptake of drugs in human liver or kidney, quantitative prediction of
drug uptake in human tissues is possible. The method to estimate the contributions of specific transporters is described below. cDNA-transfected cells and
cRNA-injected oocytes can be used as gene expression systems. More recently,
cultured cells stably transfected with both uptake and efflux transporters have
become available (85, 115). OATP-C/OATP2 [OATP1B1] and MRP2 transfected
cells and OATP8 [OATP1B3] and MRP2 transfected cells have been reported (Figure 5) (85, 115). Using them, hepatobiliary transport can be measured as vectorial
transcellular transport when these cells are cultured on a porous membrane. This
will make it easy to predict transporter-mediated drug-drug interactions in the
liver.
The estimation of the contribution of specific transporter(s) is important for the quantitative prediction of the uptake in human tissues, including liver and kidney from the
in vitro data using transporter-expressing systems, and even for the quantitative
4 Dec 2004
19:14
700
AR
AR232-PA45-28.tex
SHITARA
SATO
XMLPublishSM (2004/02/24)
P1: JRX
SUGIYAMA
Contribution =
CLhep,ref /CLCOS,ref
,
CLhep,test /CLCOS,test
6.
where CLhep and CLCOS represent the uptake clearance of compounds into hepatocytes and transporter cDNA transfected cells, respectively. CLhep,ref and CLhep,test
represent the uptake clearance of the reference and test compounds, respectively.
The reference compounds should be specific substrates, otherwise the contribution
will be overestimated (119, 120). More recently, Hirano et al. proposed a method
to estimate the contributions of human transporters (OATP-C/OATP2 [OATP1B1]
and OATP8 [OATP1B3]) to the total hepatic uptake using estrone 3-sulfate and
cholecystokinine octapeptide (CCK8) as specific substrates, respectively, and actually estimated their contributions to the hepatic uptake of pitavastatin (121). They
also estimated their contributions by uptake in human hepatocytes and transporter
expression systems normalized by their transporter expression levels measured by
Western blot analysis (121). The contributions of OATP-C/OATP2 [OATP1B1] and
OATP8 [OATP1B3] estimated by these two different methods were comparable,
suggesting the validity of this method (121).
A specific inhibitor of a transporter also helps to estimate its contribution
to the total uptake. To identify a specific inhibitor, we examined the comparative inhibitory effects of many compounds on rat Oatp1 [Oatp1a1] and Oatp2
[Oatp1a4] (122). Among them, we found that digoxin specifically inhibited Oatp2
[Oatp1a4] with no effect on Oatp1 [Oatp1a1] (122). We also found several compounds which preferentially inhibited one of these transporters (122). These inhibitors may be used to estimate the contributions of Oatp1 [Oatp1a1] and Oatp2
[Oatp1a4] at appropriate concentrations (122). However, the selectivity of most
of the preferential inhibitors in this report was not very high, and inhibitors
that act as selective inhibitors over a wider range of concentrations are needed
(122).
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
701
EVALUATION OF TRANSPORTER-MEDIATED
DRUG-DRUG INTERACTIONS
In this section, the inhibitory effects of therapeutic drugs and the possibility of
clinically relevant drug-drug interactions based on transporter-mediated processes
are described.
Vmax
+ Pdif ,
Km + S u
7.
where CL is the influx or efflux clearance; Vmax , Km , and Pdif are the maximum
transport rate, Michaelis constant, and nonsaturable transport clearance, respectively; and Su is the protein-unbound substrate concentration. In the presence of
competitive inhibitors, it can be described as follows:
CL (+inhibitor) =
Vmax
+ Pdif ,
Km (1 + Iu /Ki ) + Su
8.
9.
Vmax
+ Pdif .
Km (1 + Iu /Ki )
10.
4 Dec 2004
19:14
702
AR
AR232-PA45-28.tex
SHITARA
SATO
XMLPublishSM (2004/02/24)
P1: JRX
SUGIYAMA
CLtransporter (+inhibitor)
1
= R,
=
CLtransporter (control)
1 + Iu /Ki
11.
Figure 6
A model for estimating the inhibitor concentration at the inlet to the liver
after oral administration. Iinlet is the inhibitor concentration at the inlet to the liver. It
can be estimated from the inhibitor concentration in the hepatic artery (Ia ) plus that in
the portal vein (Ipv ).
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
703
where Isys and Ipv are the inhibitor concentration in the circulating blood and portal
vein, respectively; fu is the blood protein unbound fraction; vabs is the absorption
rate from the intestine to the portal vein; and QH is the hepatic blood flow. When
the intestinal absorption is described by a first-order rate constant, this equation
becomes (123, 124)
F D ka ekat
F D ka
Iu = fu Isys +
fu Isys +
,
13.
QH
QH
where F is the fraction absorbed from the gastrointestinal tract, D is the dose, and
ka is the absorption rate constant. To avoid a false negative prediction, the unbound
a
inhibitor concentration should be estimated by fu (Isys + FDk
) for a drug-drug
QH
interaction based on a hepatic transporter-mediated process.
To date, there are many published inhibition studies of renal and hepatic uptake
transporters: OATs and OATPs. In this section, the inhibitory effects of therapeutic
drugs on these transporters are evaluated using Ki values, comparing them with
the therapeutic concentrations.
4 Dec 2004
19:14
704
AR
AR232-PA45-28.tex
SHITARA
SATO
XMLPublishSM (2004/02/24)
P1: JRX
SUGIYAMA
Oat-k1 [Oatp1a3], and k2, are reported to be expressed in the kidney (126130),
their human counterparts have not been characterized. As shown in Supplemental
Table 2 (Follow the Supplemental Material link from the Annual Reviews home
page at http://www.annualreviews.org), several therapeutic drugs are reported to
inhibit OATP family transporters. Because they are hepatic uptake transporters, R
values were calculated based on not only the maximum inhibitor unbound therapeutic concentration in the circulating blood but also that in the inlet to the
liver, calculated by Equation 13 (123, 124). Values calculated based on the unbound concentration in the inlet to the liver are given as R. Inhibitors of OATP
family transporters consist of bulky compounds, including anions, neutral compounds, and even cations (Supplemental Table 2). In Supplemental Table 2, only
cyclosporin A and rifampicin exhibited relatively low R and R values and may
lead to clinically relevant drug-drug interactions. On the other hand, pravastatin,
an HMG-CoA reductase inhibitor, is not a cause of a severe drug-drug interaction
based on OATP-mediated hepatic uptake because of its low plasma unbound concentration. As pravastatin is a potent HMG-CoA reductase inhibitor and is highly
distributed to the liver, its target organ, a low plasma concentration is sufficient
for its pharmacological effect, leading to a low risk of inhibition of transporter
function (132). A small number of inhibitors with relatively low R values may be
due to a lack of inhibition studies involving human OATP family transporters, and
further studies may provide other inhibitors that cause clinically relevant drug-drug
interactions. More inhibition studies on human OATP transporters are needed to
allow the quantitative prediction of transporter-mediated drug-drug interactions.
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
705
0.8, ka = 0.1 min1 , and fu = 0.13. Using this and assuming a cell-to-medium
1
concentration ratio of 10, the calculated R value is 1+104.6/5
= 0.098, suggesting
that hepatobiliary excretion will be reduced to at most 9.8% of the control. Actually, both the hepatobiliary and renal clearances of digoxin have been reported to
be reduced when concomitantly administered with quinidine (133).
4 Dec 2004
19:14
706
AR
AR232-PA45-28.tex
SHITARA
SATO
XMLPublishSM (2004/02/24)
P1: JRX
SUGIYAMA
Figure 7
Transcellular transport of cerivastatin (CER) mediated by OATP-C/OATP2
[OATP1B1] and MRP2 and the inhibitory effect of cyclosporin A. (a) Transcellular transport of [14 C]CER in OATP-C/OATP2 [OATP1B1] and MRP2 double-transfected MDCK
cells (closed squares) and in vector-transfected cells (closed circles) was examined. Addition of cyclosporin A (10 M) inhibited OATP-C/OATP2 [OATP1B1]- and MRP2-mediated
transport of CER (open squares), whereas it did not change the transcellular transport in
vector transfected cells (open circles). (b) Cyclosporin A inhibited the transcellular transport
(PSB>A ) in a concentration-dependent manner. The IC50 value obtained in this experimental
system was 0.084 0.015 M. p < 0.01, p < 0.001.
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
707
HMGCoAreductase Cmax
inhibitors
[ng/mL]
Ratio
AUC
[ng h/mL]
Ratio
7.97
2.55
Simvastatin
18.9/2.5
20.6/9.9
7.56
2.08
78.1/9.8
101/39.6
Pravastatin
223/28.0
7.95
1300/
57.1
Fluvastatin
155/119
1.30
373/192
Cerivastatin
7.82/1.56
5.01
36.2/9.53
Atorvastatin
58.0/8.8#
6.59
Pitavastatin
179/27.6
6.49
Major
clearance
mechanism
Reference
CYP3A4
193
194
OATP-C
143
1.94
CYP2C9
195
3.80
CYP2C8/
3A4OATPC
142
595/79.9#
7.45
CYP3A4OATP-C
145
347/76.9
4.51
OATP-C
144
4 Dec 2004
19:14
708
AR
AR232-PA45-28.tex
SHITARA
SATO
XMLPublishSM (2004/02/24)
P1: JRX
SUGIYAMA
HMG-CoA
reductase
inhibitors
Cmax
[ng/mL]
Lovastatin
Simvastatin
Pravastatin
Fluvastatin
Cerivastatin
2.38/2.69
6.15/6.87
120/66.3
54.3/48.4
8.0/3.2
Pitavastatin
Rosuvastatin
2.93/1.61 1.82
no data
1.30
109/49.5 2.20
Ratio
AUC
[ng h/mL]
Ratio
0.885
0.895
1.81
1.12
2.5
33.1/34.4
36.2/25.2
281/139
213/227
91.1/20.9
0.962
1.44
2.02
0.938
4.36
41.9/9.92
no data
771/410
4.22
1.45
1.88
Major
clearance
mechanism
CYP3A4
CYP3A4
OATP-C
CYP2C9
CYP2C8/3A4
OATP-C
OATP-C
CYP2C9
OATP-C
Reference
196
197
198
199
148
149
200
151
Xenopus laevis oocytes (150, 151). We also analyzed the inhibitory effects of
gemfibrozil and its metabolites on the P450-mediated metabolism of cerivastatin
and found that gemfibrozil and its glucuronide inhibited the CYP2C8-mediated
metabolism with IC50 values of 28 and 4 M, respectively (150). They are still
higher than the therapeutic unbound concentrations in the circulating blood. However, there are reports that, in rat perfusion studies, gemfibrozil-1-O-glucuronide
is actively taken up into the liver and accumulates there (152154). If this also
took place in human liver, the concentrated gemfibrozil-1-O-glucuronide might
act as an inhibitor of CYP2C8-mediated metabolism, leading to a drug-drug interaction. In this case, a transporter plays an important role, i.e., an inhibitor of the
metabolism leading to accumulation in the liver via transporter-mediated uptake.
Our hypothesis that interaction with gemfibrozil is not a transporter-mediated
one, but a metabolism-mediated one, is supported by the fact that gemfibrozil
does not cause a severe interaction with pravastatin and pitavastatin, which are
mainly cleared by the OATP-C/OATP2 [OATP1B1]-mediated hepatic uptake (Table 2). Therefore, we should also be more cautious about drug-drug interactions
when inhibitors of the metabolism are substrates of hepatic uptake transporters
(Figure 8).
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
709
described in MDR-Mediated Drug-Drug Interactions (above), the Ki value of quinidine for the MDR1-mediated efflux can be assumed to be 5 M. On the other hand,
the steady-state plasma concentration of quinidine in this study was 4.5 M, with
a protein unbound fraction of 0.13. Therefore, the protein unbound concentration
in the circulating blood is estimated to be 0.59 M. The unbound concentration of quinidine at the inlet to the liver estimated by Equation 13 is 4.6 M
using QH = 1.6 liters min1 , Fa Fg = 0.8, and ka = 0.1 min1 . With a safety
margin of 1 10 as a cell-to-medium concentration ratio, the estimated reduction
in the renal excretion of digoxin is 46% to 89% of the control, and the estimated
reduction in the hepatobiliary excretion of digoxin is 9.8% to 52% of the control. In
clinical situations, the hepatobiliary excretion was reduced to 42% of the control,
whereas the renal excretion was reduced to 60% of the control, which was within
the predicted range (133).
In rat hepatocytes, the inhibitory effect on the uptake of digoxin was more
potent for quinine than for quinidine, and the same tendency was observed using
the rat Oatp2 [Oatp1a4] expression system (122, 155). Therefore, the mechanism
of the drug-drug interaction between digoxin and quinine may be caused by the
inhibition of the transporter-mediated uptake. However, there is a study that shows
that both quinine and quinine had no inhibitory effects on the uptake of digoxin
into isolated human hepatocytes, although both of them inhibited the uptake of
digoxin into rat hepatocytes (156).
4 Dec 2004
19:14
710
AR
AR232-PA45-28.tex
SHITARA
SATO
XMLPublishSM (2004/02/24)
P1: JRX
SUGIYAMA
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
711
MTX is expected to be nonextensive and partial. Many NSAIDs also inhibit human
OAT3-mediated uptake of MTX with therapeutic relevant plasma concentrations
of unbound drugs (26). However, also in humans, the contribution of OAT3 to
the total renal uptake of MTX needs to be clarified for the identification of the
mechanism of the clinically relevant DDI.
CONCLUSION
In addition to phase I and phase II enzymes, transporters also play an important
role in drug elimination and distribution. Therefore, it is possible that transportermediated drug-drug interactions alter pharmacokinetics, and could result in severe
side effects.
A large number of transporters have been characterized in rodents and humans,
and the mechanism of the membrane transport of several compounds including
endogenous compounds and therapeutic drugs has been clarified. However, the
transport mechanism of most therapeutic drugs remains unknown. To predict a
transporter-mediated drug-drug interaction, the transporters involved in the membrane transport of the drug need to be characterized. As multiple transporters have
been characterized in the kidney and liver and their expression systems are available, it should be possible to predict a transporter-mediated drug-drug interaction
by using these systems with the information of the contribution made by each
transporter to the net transport in the kidney and liver.
We have estimated the possibility of a transporter-mediated drug-drug interaction from the R value, calculated using the maximum unbound concentration of
inhibitors. This method may avoid false negative predictions of drug-drug interactions. In conclusion, greater awareness of the possibility of transporter-mediated
drug-drug interactions is necessary.
The Annual Review of Pharmacology and Toxicology is online at
http://pharmatox.annualreviews.org
LITERATURE CITED
1. Petzinger E. 1994. Transport of organic
anions in the liver. An update on bile
acid, fatty acid, monocarboxylate, anionic
amino acid, cholephilic organic anion,
and anionic drug transport. Rev. Physiol.
Biochem. Pharmacol. 123:47211
2. Oude Elferink RPJ, Meijer DKF, Kuipers
F, Jansen PLM, Groen AK, Groothis
GMM. 1995. Hepatobiliary secretion of
organic compounds; molecular mechanism of membrane transport. Biochem.
Biophys. Acta 1241:21568
4 Dec 2004
19:14
712
AR
AR232-PA45-28.tex
SHITARA
SATO
XMLPublishSM (2004/02/24)
P1: JRX
SUGIYAMA
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
33.
34.
35.
36.
37.
38.
39.
40.
713
4 Dec 2004
19:14
714
AR
AR232-PA45-28.tex
SHITARA
SATO
XMLPublishSM (2004/02/24)
P1: JRX
SUGIYAMA
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
715
4 Dec 2004
19:14
716
AR
AR232-PA45-28.tex
SHITARA
SATO
XMLPublishSM (2004/02/24)
P1: JRX
SUGIYAMA
83.
84.
85.
86.
87.
88.
89.
90.
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
100.
101.
102.
103.
104.
105.
106.
107.
717
4 Dec 2004
19:14
718
108.
109.
110.
111.
112.
113.
114.
115.
116.
AR
AR232-PA45-28.tex
SHITARA
SATO
XMLPublishSM (2004/02/24)
P1: JRX
SUGIYAMA
117.
118.
119.
120.
121.
122.
123.
124.
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
125.
126.
127.
128.
129.
130.
131.
132.
133.
134.
135.
136.
137.
138.
139.
140.
719
of digoxin: stereoselective difference between quinine and quinidine. Clin. Pharmacol. Ther. 47:2026
Bohme M, Jedlitschky G, Leier I, Buchler M, Keppler D. 1994. ATP-dependent
export pumps and their inhibition by cyclosporins. Adv. Enzyme Regul. 34:371
80
Versantvoort CH, Broxterman HJ,
Lankelma J, Feller N, Pinedo HM. 1994.
Competitive inhibition by genistein
and ATP dependence of daunorubicin
transport in intact MRP overexpressing
human small cell lung cancer cells.
Biochem. Pharmacol. 48:112936
Chu XY, Kato Y, Niinuma K, Sudo KI,
Hakusui H, et al. 1997. Multispecific organic anion transporter is responsible for
the biliary excretion of the camptothecin
derivative irinotecan and its metabolites in
rats. J. Pharmacol. Exp. Ther. 281:30414
Araki E, Ishikawa M, Iigo M, Koide T,
Itabashi M, et al. 1993. Relationship between development of diarrhea and the
concentration of SN-38, an active metabolite of CPT-11, in the intestine and the
blood plasma of athymic mice following
intraperitoneal administration of CPT-11.
Jpn. J. Cancer Res. 84:697702
Gupta E, Lestingi TM, Mick R, Ramirez
J, Vokes EE, et al. 1994. Metabolic fate of
irinotecan in humans: correlation of glucuronidation with diarrhea. Cancer Res.
54:372325
Horikawa M, Kato Y, Tyson CA,
Sugiyama Y. 2002. The potential for an
interaction between MRP2 (ABCC2) and
various therapeutic agents: probenecid
as a candidate inhibitor of the bibiary excretion of irinotecan metabolites. Drug Metab. Pharmacokin. 17:23
33
Horikawa M, Kato Y, Sugiyama Y.
2002. Reduced gastrointestinal toxicity
following inhibition of the biliary excretion of irinotecan and its metabolites by
probenecid in rats. Pharm. Res. 19:1345
53
4 Dec 2004
19:14
720
AR
AR232-PA45-28.tex
SHITARA
SATO
XMLPublishSM (2004/02/24)
P1: JRX
SUGIYAMA
151.
152.
153.
154.
155.
156.
157.
158.
159.
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
160.
161.
162.
163.
164.
165.
166.
167.
721
4 Dec 2004
19:14
722
177.
178.
179.
180.
181.
182.
183.
184.
185.
AR
AR232-PA45-28.tex
SHITARA
SATO
XMLPublishSM (2004/02/24)
P1: JRX
SUGIYAMA
cephalosporins, but not of biliary excreted cefoperazone. Drug Metab. Pharmacokin. 17:12529
Babu E, Takeda M, Narikawa S,
Kobayashi Y, Enomoto A, et al. 2002.
Role of human organic anion transporter 4
in the transport of ochratoxin A. Biochim.
Biophys. Acta 1590:6475
Babu E, Takeda M, Narikawa S,
Kobayashi Y, Yamamoto T, et al. 2002.
Human organic anion transporters mediate the transport of tetracycline. Jpn. J.
Pharmacol. 88:6976
Uwai Y, Saito H, Hashimoto Y, Inui K.
2000. Interaction and transport of thiazide diuretics, loop diuretics, and acetazolamide via rat renal organic anion transporter rOAT1. J. Pharmacol. Exp. Ther.
295:26165
Uwai Y, Saito H, Hashimoto Y, Inui K.
2000. Inhibitory effect of anti-diabetic
agents on rat organic anion transporter
rOAT1. Eur. J. Pharmacol. 398:19397
Nagata Y, Kusuhara H, Endou H,
Sugiyama Y. 2002. Expression and functional characterization of rat organic anion transporter 3 (rOat3) in the choroid
plexus. Mol. Pharmacol. 61:98288
Takeda M, Narikawa S, Hosoyamada M,
Cha SH, Sekine T, et al. 2001. Characterization of organic anion transport inhibitors using cells stably expressing human organic anion transporters. Eur. J.
Pharmacol. 419:11320
Enomoto A, Takeda M, Shimoda M,
Narikawa S, Kobayashi Y, et al. 2002. Interaction of human organic anion transporters 2 and 4 with organic anion transport inhibitors. J. Pharmacol. Exp. Ther.
301:797802
Cvetkovic M, Leake B, Fromm MF,
Wilkinson GR, Kim RB. 1999. OATP and
P-glycoprotein transporters mediate the
cellular uptake and excretion of fexofenadine. Drug Metab. Dispos. 27:86671
Tokui T, Nakai D, Nakagomi R, Yawo H,
Abe T, et al. 1999. Pravastatin, an HMGCoA reductase inhibitor, is transported by
186.
187.
188.
189.
190.
191.
192.
193.
4 Dec 2004
19:14
AR
AR232-PA45-28.tex
XMLPublishSM (2004/02/24)
P1: JRX
723
4 Dec 2004
19:16
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
10.1146/annurev.pharmtox.45.120403.100040
Department of Pharmacology, the Combinatorial Chemistry Center and the Fiske Drug
Discovery Laboratory, University of Pittsburgh Cancer Institute, University of Pittsburgh,
Pittsburgh, Pennsylvania 15261; email: lazo@pitt.edu
2
Department of Chemistry, the Combinatorial Chemistry Center and the Fiske Drug
Discovery Laboratory, University of Pittsburgh Cancer Institute, University of Pittsburgh,
Pittsburgh, Pennsylvania 15261
Key Words
INTRODUCTION
Cellular signaling networks are controlled by reversible covalent phosphorylation,
which depends on a precise balance between protein kinase and phosphatase activities (1). These signaling networks govern processes such as cell growth, cell
division, and cell death; perturbation of these pathways, whether by environmental
stresses or genetic defects, underlies the pathophysiology of many diseased states.
The sequencing of the human genome predicts approximately 428 protein kinases,
the majority of which catalyze serine and threonine phosphorylation (Figure 1) (2).
Although protein kinases were originally considered the prime regulators of signal
transduction-mediated events, it is now recognized that protein dephosphorylation
is an equally important component, playing a central role in cell cycle transitions
and other signal transduction mechanisms (3). Furthermore, protein phosphatase
activity critically regulates fundamental cellular processes that are perturbed in diseased states. The human genome is estimated to encode 99 protein phosphatases,
0362-1642/05/0210-0725$14.00
725
4 Dec 2004
19:16
726
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
DUCRUET ET AL.
Figure 1
The balance of protein kinases and phosphatases in the human genome.
This figure is based on DNA sequence and protein structural analyses described by
others (2, 3, 6a). The total predicted number of human protein tyrosine (Tyr), serine
(Ser), threonine (Thr), and dual-specificity kinases and phosphatases are indicated.
Catalytically inactive phosphatases and kinases and the phosphatases with lipid or
nucleic acid substrates are not included. See text for details.
approximately one quarter the number of protein kinases, suggesting functional redundancy and/or substrate promiscuity (Figure 1) (2, 3). Protein phosphatases are
classified according to their substrate specificity, either serine/threonine-specific
protein phosphatases (PS/TPases) or tyrosine-specific protein phosphatases (PTPases) (4), although there have been recent efforts to exploit structural information (3), which may result in some reassignments. Dual-specificity phosphatases
(DSPases) represent a subclass of the protein tyrosine phosphatase superfamily by
virtue of their highly conserved PTPase active site motif and because they employ
the PTPase catalytic mechanism, which proceeds via the formation of a transient
enzyme-phosphosubstrate intermediate [4; reviewed in Zhang (5)]. DSPases, however, are unique in their ability to dephosphorylate protein substrates containing
both phosphotyrosine and phosphoserine or phosphothreonine, either immediately adjacent or separated by one amino acid; such substrates are exemplified
by the cyclin-dependent kinases (Cdks) and the mitogen-activated protein kinases
(MAPKs), which play essential roles in the signaling pathways that regulate cell
division and cell growth (Figures 2 and 3). Recent structural analyses suggest the
human genome encodes 38 DSPases, including 11 MAPK phosphatases (MKPs),
17 atypical DSPases, 4 PRL phosphatases, 3 Cdc14 phosphatases, and 3 Cdc25
phosphatases (3) (Figure 1). The DSPases share the conserved PTPase active site
4 Dec 2004
19:16
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
727
and catalytic mechanism but they have a shallower active site cleft than PTPases,
presumably to accommodate the sterically less accessible phosphoserine and phosphothreonine residues (4, 6). The most widely studied DSPases are the Cdc25
phosphatases and the MKPs, two protein families that play central roles in the
biology of the cell.
CDC25 DSPASES
The first DSPases to be discovered were the Cdc25 phosphatases, which were
functionally defined as promoters of the cell division cycle in yeast (7). More
specifically, Cdc25 phosphatases dephosphorylate and activate the Cdks (Figures
2 and 4), which are key participants in the cellular division program induced
in response to extracellular signals including growth factors. Cdks coupled to
their cyclin partner are maintained in an inactive state by dual phosphorylation at
adjacent threonine and tyrosine (-T-Y-) residues near their amino terminus; these
inactivating phosphorylations are mediated by Wee1 and Myt1 protein kinases
(8, 9). Cdc25s activate Cdks by dephosphorylating both phosphothreonine and
phosphotyrosine residues (Figure 2); regulation of Cdk kinase activity remains an
Figure 2
Cdc25 phosphatases dephosphorylate and activate the cyclin-dependent
kinases. Mitogenic signal transduction cascades induce cell division. Progression
through cell cycle transitions is achieved by dephosphorylation and activation of the
cyclin-dependent kinases by Cdc25 phosphatases. In contrast to the MKPs, the Cdc25
phosphatases activate Cdks by dephosphorylating both residues in the Cdk -T-Y- motif
(see text for details).
4 Dec 2004
19:16
728
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
DUCRUET ET AL.
Figure 3
Mitogen-activated protein kinase phosphatases dephosphorylate and inactivate the mitogen-activated protein kinases. Growth factor receptor signal transduction
cascades, cellular stresses, and chemotherapeutics can activate mitogenic signaling
pathways, culminating in the activation of upstream mitogen-activated protein kinase
kinase kinases (MAPKKKs), which phosphorylate and activate mitogen-activated protein kinase kinases (MAPKKs), which phosphorylate and activate mitogen-activated
protein kinases (MAPKs) in the -T-x-Y- motif. Downregulation of mitogenic signaling through MAPKs is achieved by dephosphorylation of both residues in the -T-x-Ymotif, a process regulated by the dual-specificity MAPK phosphatases (DS-MKPs).
4 Dec 2004
19:16
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
729
Figure 4
Cdc25 phosphatases promote mammalian cell cycle progression. Cdc25
phosphatases drive the cell division cycle by dephosphorylating and activating the Cdks.
The three human Cdc25 isoforms, Cdc25A, Cdc25B, and Cdc25C, have overlapping
roles in the cell cycle. Cdc25A exclusively promotes the G1/S transition and S phase
progression and contributes to the Cdc25 activity necessary for G2 phase progression,
the G2/M transition, and mitosis. Cdc25B contributes to G2 progression and is believed
to be the trigger for initiating the G2/M transition. Cdc25C activity is restricted to
mitosis. The Cdc25 phosphatases are targeted by the G1/S, intra-S, and G2/M cell cycle
checkpoints to inhibit their activity in response to genotoxic stress. Cdc25 activity is
influenced by Cdk activity in regulatory feedback loops: solid arrows indicate activation
by Cdc25 and dotted arrows represent known positive (+) or negative () feedback
loops. Cdk2 has both positive and negative effects on Cdc25A (+/). It is unclear
whether Cdk4/cyclin D is a bona fide substrate of Cdc25A in cells.
below). Cdc25C, the first human Cdc25 isoform identified, functions primarily in
mitosis and catalyzes mitotic progression by activating Cdk1/cyclin B; microinjection of anti-Cdc25C antibodies into HeLa cells prevented mitotic entry (1113).
Cdc25B also activates Cdk1/cyclin B, and microinjection of anti-Cdc25B antibodies inhibits mitotic entry, leading many to speculate that Cdc25B is functionally
4 Dec 2004
19:16
730
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
DUCRUET ET AL.
redundant to Cdc25C (14, 15). Nonetheless, Cdc25B and Cdc25C activities are
temporally distinct, with Cdc25B activity peaking prior to that of Cdc25C. More
recently, Cdc25B has been described as the trigger for the G2/M transition; Cdc25B
appears to initiate the mitotic transition by activating a particular pool of Cdk1/cyclin
B (1518). Cdc25B also contributes to the Cdk phosphatase activity necessary to
activate Cdk2/cyclin A in S phase and Cdk1/cyclin A in G2 (17, 19, 20). Cdc25A
promotes the G1/S cell cycle transition and S phase progression by activating
Cdk2/cyclin E (21, 22). Microinjection of anti-Cdc25A antibodies prevented S
phase entry in cells following serum induction, and overexpression of Cdc25A accelerated S phase entry with premature Cdk2 activation (2123). Cdc25A activity
is rate limiting for the G2/M transition and mitotic progression by contributing to
Cdk1/cyclin B activation (24, 25).
Although the emerging model for temporal and combinatorial contributions of
Cdc25A, Cdc25B, and Cdc25C activities to achieve precise control over cell cycle
progression is appealing, Cdc25B/ mice and Cdc25C/ mice are viable and
cells isolated from these mice undergo normal mitotic cell division, implying that
Cdc25A has the potential to drive the entire mitotic cell division cycle (26, 27).
The preeminence of Cdc25A is further illustrated by the prompt inactivation of
Cdk activity and cell cycle arrest seen with rapid Cdc25A degradation (24, 2830).
Cdc25A has, thus, been dubbed the master Cdk phosphatase (31), as it appears
to be responsible for Cdk activation to promote the G1/S cell cycle transition,
for maintaining Cdk activity throughout S phase and G2 progression, and for
contributing to the Cdk phosphatase activity necessary for the G2/M transition
and mitotic progression (Figure 4) (21, 22, 24, 25). It remains unclear why cells
have multiple Cdc25s to regulate mitotic cell division, although it is possible that
their combined activities ensure optimal Cdk activation to promote the irreversible
process of mitotic division. In such a model, the multiple Cdc25s would impose
a switch-like regulatory mechanism, consisting of a biological threshold of Cdk
activation, to achieve strict unidirectional control of cell division (32).
Cdc25 Regulation
As key controllers of cell division, Cdc25 DSPases are subject to precise regulation, including enzyme-substrate feedback loops involving specific Cdk/cyclin
complexes and their activating Cdc25. For example, Cdc25A activity is upregulated by Cdk2/cyclin E following its activation, and Cdc25A protein stability is
increased by Cdk1/cyclin B phosphorylation (21, 24); Cdk2 activity also appears
to negatively regulate Cdc25A protein stability (33). Cdc25B protein stability is
negatively regulated by Cdk1/cyclin A (16) and Cdc25C catalytic activity is upregulated by Cdk1/cyclin B (Figure 4) (34). In addition, the Cdc25 DSPases are
regulated by alternative gene splicing, which results in the expression of 12 splice
variants. The precise role of alternative splicing in Cdc25 biology remains unclear,
although the splice variants could have altered tissue or cell cycle phase activity
profiles, or they may have different specific catalytic activities as a result of loss
of consensus regulatory phosphorylation sites (35, 36).
4 Dec 2004
19:16
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
731
Throughout the cell cycle, Cdc25C protein expression does not appreciably fluctuate; however, Cdc25C remains inactive during interphase by 14-3-3-mediated
sequestration in the cytoplasm (37). Cdc25A and Cdc25B, on the other hand, are
labile proteins, most likely owing to their role as the major catalysts of the cell
cycle transitions (25, 38). Cdc25B protein levels accumulate throughout late S
and early G2, peaking at the G2/M transition (17, 39). Although a detailed understanding of Cdc25B protein turnover is lacking, its proteolysis requires prior
phosphorylation by Cdk1/cyclin A (16). Like Cdc25C, Cdc25B activity is also
regulated by its subcellular localization, which is facilitated by interactions with
14-3-3 (40). Cdc25A protein levels and activity remain elevated past S phase and
increase as cells enter mitosis. Cdc25A activity is primarily regulated by protein
turnover, although 14-3-3 can prevent the phosphatase from interacting with its
mitotic substrate, Cdk1/cyclin B. Furthermore, Cdc25A activity has been reported
to be upregulated by phosphorylation in response to mitogenesis (41, 42). Cdc25A
protein turnover is catalyzed by the ubiquitin-proteasome pathway; Cdc25A ubiquitination is catalyzed by the APC/CCdh1 ubiquitin ligase during mitotic exit and
early G1 and by the SCFTrCP ubiquitin ligase during interphase [reviewed in
Busino et al. (43)]. The subcellular localization of Cdc25A remains a matter of
some debate, as Cdc25A has been reported to localize in the nucleus, the cytoplasm, and the plasma membrane and to interact with proteins that reside in each
of these cellular compartments (21, 41, 42, 4446).
4 Dec 2004
19:16
732
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
DUCRUET ET AL.
Figure 5
Cdc25C inactivation by the G2/M cell cycle checkpoint. In response to
genotoxic stress, checkpoint kinases Chk1 and Chk2 phosphorylate Cdc25C, promoting its cytoplasmic sequestration by 14-3-3 binding.
MKP DSPASES
MKPs dephosphorylate and inactivate MAPKs on threonine and tyrosine residues
(Figure 3). MAPKs are widely studied protein kinases that play pivotal roles in
4 Dec 2004
19:16
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
733
4 Dec 2004
19:16
734
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
DUCRUET ET AL.
Figure 7
Cdc25A inactivation by cell cycle checkpoints. Cdc25A is rapidly and
irreversibly inactivated by the G1/S, intra-S phase, and G2/M cell cycle checkpoints.
In response to genotoxic stress or interruptions to DNA synthesis, stress-responsive
p38 MAPK and checkpoint kinases Chk1 and Chk2 phosphorylate Cdc25A (at multiple
sites), promoting its association with ubiquitin ligases. Following polyubiquitination
(Ub), Cdc25A is degraded by the 26S proteasome; dashed outlined Cdc25A indicates
degraded protein.
To date, 12 bona fide human DS-MKPs have been cloned and characterized
(Table 1). Table 1 also contains two putative DS-MKPs, namely hVYH1, whose
substrate has not been identified, and JSP-1, which fails to dephosphorylate MAPK
in cells but nonetheless specifically activates the JNK pathway by an as of yet undetermined mechanism (59). The first MKP discovered was 3CH134/MKP-1 (60),
which was later found to have PTPase activity (61) and DSPase activity (62). The
human homolog of 3CH134/MKP-1, CL100 or DUSP1, was independently cloned
(63). Other DS-MKPs were subsequently discovered in a variety of organisms [a
comprehensive listing of DS-MKPs from various species was compiled by Farooq
& Zhou (64)].
The DS-MKPs have unique but overlapping MAPK substrate specificities, as
recently reviewed by Farooq & Zhou (64). For example, the Erk isoforms are
4 Dec 2004
19:16
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
735
Figure 8
Cdc25 phosphatases regulate multiple signaling pathways. In addition to
driving cell cycle transitions, Cdc25 phosphatases promote hormone-responsive gene
expression by affecting steroid receptor activity, downregulate apoptotic responses to
genotoxic stresses by blocking Ask1 homo-dimerization (which is necessary for Ask1
activation), and downregulate mitogenic signaling by dephosphorylating the epidermal
growth factor receptor (EGFR) and Raf-1, which can also have a cytoprotective effect.
selectively dephosphorylated by MKP-3, whereas M3/6 selectively dephosphorylates JNK. MKP-1 recognizes JNK, ERK, and p38, and MKP-2 recognizes Erk
and JNK. PAC-1, a DSPase from human T cells that is similar to MKP-3, is specific
for Erk. MKP-5 appears to be somewhat selective for p38. The prototype DSPase
VHR dephosphorylates Erk and JNK. There is also evidence for cross-talk between
the MAPK pathways. For example, MKP-7 interacts with Erk, JNK, and p38, but
shows substrate specificity for JNK and is phosphorylated in an Erk-dependent
manner (65).
4 Dec 2004
19:16
736
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
DUCRUET ET AL.
Table 1 Human DS-MKPs identified by DUSP nomenclature based
on analysis described in Reference 64
DUSP
Synonyms
GenBank
Accession Number
DUSP1
NM 004417
DUSP2
PAC1, PAC-1
NM 004418
DUSP3
VHR
NM 004090
DUSP4
NM 001394
DUSP5
HVH3
NM 004419
DUSP6
MKP-3, PYST1
NM 001946
DUSP7
MKP-X, PYST2
NM 001947
DUSP8
NM 004420
DUSP9
MKP-4
NM 001395
DUSP10 MKP-5
NM 007207
NM 007026
DUSP16 MKP-7
NM 030640
DUSP12 YVH1
NM 007240
NM 020185
DS-MKP Regulation
The DS-MKPs are regulated on multiple levels. The majority of DS-MKPs are
inducible genes, and basal levels of DS-MKPs are low in nonstressed or unstimulated cells [reviewed in Keyse (58)]. Some DS-MKPs are immediate early genes.
For example, MKP-1, MKP-2, MKP-X (PYST2), and PAC-1 are rapidly induced
in response to serum stimulation (6668). In contrast, MKP-3 (PYST1), MKP4, MKP-5, MKP-X, and M3/6 are not encoded by immediate early genes (58).
MKP-3 and VHR are constitutively expressed (67), and while MKP-3 is moderately inducible after several hours of stimulation (67, 69), VHR is not known to be
inducible. Different DS-MKPs respond to different stimuli: MKP-1 is inducible by
mitogens, oxidative stress, heat shock (63, 69), and hypoxia (7072). In contrast,
MKP-X is only moderately induced by serum but not by cellular stress (67).
Inducible expression of DS-MKPs is thought to be a mechanism for attenuation of mitogenic signaling. Induction of MKP-1 in NIH3T3 cells (62) and
CCL39 hamster lung fibroblasts temporally correlates with Erk inactivation and is
dependent on Erk activity (66). An additional mechanism by which Erk induces
MKP-1 is through stabilization of MKP-1 protein levels. This is achieved by direct
phosphorylation of MKP-1 by Erk, leading to reduced MKP-1 ubiquitination and
proteasomal degradation (73). Furthermore, some DS-MKPs are activated by activated forms of their respective substrates. MKP-3 experiences a 25-fold increase
in catalytic activity when complexed to its phosphorylated substrate, Erk2 (74).
4 Dec 2004
19:16
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
737
This activation is specific, as neither p38 nor JNK activated MKP-3, but they did
activate a nonspecific DS-MKP (MKP-4) (74). Taken together, the data indicate
that inactivation of the Erk cascade is regulated through induction and stabilization
of DS-MKPs in an inhibitory feedback loop.
4 Dec 2004
19:16
738
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
DUCRUET ET AL.
transformed by upstream components of the Erk MAPK signaling cascade. Therefore, overexpression of Cdc25A and Cdc25B may contribute to the transformed
phenotype by endowing cells with a proliferative advantage or by generating resistance to genotoxic stress-induced cell cycle arrest and apoptosis.
The role of the Cdc25s in neurodegeneration remains unclear. Cdc25A and
Cdc25B are expressed and active in the brains of Alzheimers disease patients (77,
78), and there is increasing evidence that expression and activation of the cell cycle
machinery is associated with neurodegeneration in postmitotic neurons (8689).
Cell cycle activation appears to be a critical element of the apoptotic response
to DNA damage in postmitotic neurons, and Cdk activation is a precursor to the
neurodegeneration characteristic of Alzheimers disease (78, 90, 91); moreover,
inhibition of Cdk activity provides a neuroprotective effect, substantiating a role
for the cell cycle machinery in the pathophysiology of neurodegeneration (92).
The Cdc25 DSPases, therefore, constitute attractive potential targets for cancer
and neurodegenerative disease drug discovery.
4 Dec 2004
19:16
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
739
4 Dec 2004
19:16
740
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
DUCRUET ET AL.
other kinases, such as the Cdks, mitogen-activated protein kinase kinase (MEK),
Raf, and mTOR (124). Based on their roles in multiple signaling pathways and
altered expression in diseased states, there has been increasing interest in identifying DSPase inhibitors that are more potent and selective than the general tyrosine
phosphatase inhibitor sodium orthovanadate (55). Such targeted agents may provide value as therapeutics for cancer and Alzheimers disease. The shallow nature
of the DSPase active site, combined with the conserved nature of the PTPase active
site cleft, has lead some investigators to believe that DSPase-selective inhibitors
may be difficult to identify. Although the three Cdc25 isoforms possess the common, highly conserved PTPase active site motif, the architecture of their active site
appears to be different. Thus, the Cdc25 phosphatases have a shallow catalytic do cleft (125,126). Indeed, several groups
main, whereas the PTPases have a deep 9 A
have identified lead compounds with favorable selectivity profiles, suggesting that
phosphatase-selective inhibition is plausible (55).
Structure-activity relationships of natural and synthetic inhibitors of DSPases
have been partially reviewed (55, 127). Representative members in this group include the natural products dnacin B1 (1), dysidiolide (2), menadione (3), and coscinosulfate (4), which inhibited the Cdc25 family with IC50 values in the 110 M
range (Figure 9) (128132). The biological activities of these natural products inspired total syntheses as well as the preparations of synthetic analogs and chemical
libraries (133138). Structurally most conspicuous among the small-molecule inhibitors discovered through combinatorial library and random screening are highly
lipophilic acids [e.g., 5 (139), 6 (140), 7 (141)] as well as annulated para-quinones
[e.g., 8 (142), 9 (143), 10 (144), 11 (145)] (Figure 9). In addition, moderately
potent heterocyclic [e.g., 12 (146), 13 (147)] and phenolic derivatives [14 (148)]
have also been identified (Figure 9).
To date, compounds with quinone moieties have demonstrated the highest potency as well as considerable specificity in DSPase screens. Specifically, 10 was
found to inhibit Cdc25B and VHR with IC50 values of 206 nM and 4.0 M, respectively. Compound 11 had IC50 values of 22, 125, and 57 nM for Cdc25A, B, and
C, respectively, and showed partial mixed-inhibitory kinetics. It is also interesting
to note that indolyldihydroxyquinone 9 was found to bind competitively with the
substrate at the active site of Cdc25 and yield a Ki of 470 nM (143). Molecular
modeling of enzyme-inhibitor complexes is possible (145, 149) because several
crystal structures of Cdc25 isoforms are available, but rational design has so far
met with limited success for the improvement of the binding characteristics of lead
structures. A recent report presents the homology-modeling of a Cdc25B-inhibitor
complex, which might provide a more suitable starting point for rational inhibitor
design (150).
Whether Cdc25 isoform specificity can be achieved is a challenging issue.
All three Cdc25 isoforms possess identical amino acids in their highly conserved
PTPase active site motif (HCEFSSER), and they share a high degree of sequence
homology outside of this catalytic loop, posing a high hurdle for selective targeting
of the individual Cdc25 isoforms. Nonetheless, selective protein kinase inhibition
4 Dec 2004
19:16
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
741
Figure 9
Natural product and synthetic inhibitors of the Cdc25 family of DSPases.
Representative examples for small-molecule inhibitors of this enzyme class are natural
products (14), lipophilic acids (57), quinones (811), heterocycles (12, 13), and
phenols (13, 14).
has been achieved with ATP competitive inhibitors, which target similar active
site structures and catalytic mechanisms (124), lending credence to the hypothesis
that selective Cdc25 inhibition may yet be achieved. It is worth mentioning that
Cdc25-specific inhibitors lacking isoform selectivity may have some theoretical
therapeutic appeal, but additional small-molecule inhibitors will be required to
fully test this hypothesis.
Although screening strategies for Cdc25 inhibitors have focused on identifying
active site inhibitors, an alternate mechanism for targeted inhibition of Cdc25
4 Dec 2004
19:16
742
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
DUCRUET ET AL.
CONCLUSIONS
DSPases have critical roles in regulating cellular phosphorylation signaling networks and are deregulated in human cancer and Alzheimers disease. The uniqueness of their biochemical mechanism and the central role of their substrates make
DSPases an attractive target for further pharmacological studies. In recent years,
several natural products and novel small organic molecules have been identified
that can block phosphatase activity. Nonetheless, there continues to be a need
for more potent and selective inhibitors of DSPases to permit a further dissection
of their roles in biological systems and to clinically validate their potential as
anticancer targets.
The Annual Review of Pharmacology and Toxicology is online at
http://pharmtox.annualreviews.org
LITERATURE CITED
1. Hunter T. 1995. Protein kinases and phosphatases: the yin and yang of protein phosphorylation and signaling. Cell 80:225
36
4 Dec 2004
19:16
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
13.
14.
15.
16.
17.
18.
19.
20.
21.
743
4 Dec 2004
19:16
744
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
DUCRUET ET AL.
dependent phosphorylation at the G1/S
transition. EMBO J. 13:430210
Jinno S, Suto K, Nagata A, Igarashi M,
Kanaoka Y, et al. 1994. Cdc25A is a novel
phosphatase functioning early in the cell
cycle. EMBO J. 13:154956
Blomberg I, Hoffmann I. 1999. Ectopic
expression of Cdc25A accelerates the
G(1)/S transition and leads to premature activation of cyclin E- and cyclin
A-dependent kinases. Mol. Cell Biol.
19:618394
Mailand N, Podtelejnikov AV, Groth A,
Mann M, Bartek J, Lukas J. 2002. Regulation of G(2)/M events by Cdc25A
through phosphorylation-dependent modulation of its stability. EMBO J. 21:5911
20
Donzelli M, Squatrito M, Ganoth D, Hershko A, Pagano M, Draetta GF. 2002.
Dual mode of degradation of Cdc25 A
phosphatase. EMBO J. 21:487584
Chen MS, Hurov J, White LS, WoodfordThomas T, Piwnica-Worms H. 2001. Absence of apparent phenotype in mice lacking Cdc25C protein phosphatase. Mol.
Cell Biol. 21:385361
Lincoln AJ, Wickramasinghe D, Stein
P, Schultz RM, Palko ME, et al. 2002.
Cdc25b phosphatase is required for resumption of meiosis during oocyte maturation. Nat. Genet. 30:44649
Falck J, Mailand N, Syljuasen RG,
Bartek J, Lukas J. 2001. The ATM-Chk2Cdc25A checkpoint pathway guards
against radioresistant DNA synthesis. Nature 410:84247
Mailand N, Falck J, Lukas C, Syljuasen
RG, Welcker M, et al. 2000. Rapid destruction of human Cdc25A in response
to DNA damage. Science 288:142529
Zhao H, Watkins JL, Piwnica-Worms H.
2002. Disruption of the checkpoint kinase
1/cell division cycle 25A pathway abrogates ionizing radiation-induced S and G2
checkpoints. Proc. Natl. Acad. Sci. USA
99:14795800
Donzelli M, Draetta GF. 2003. Regulating
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
4 Dec 2004
19:16
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
745
4 Dec 2004
19:16
746
62.
63.
64.
65.
66.
67.
68.
69.
70.
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
DUCRUET ET AL.
protein-tyrosine-phosphatase. Proc. Natl.
Acad. Sci. USA 90:529296
Sun H, Charles CH, Lau LF, Tonks NK.
1993. MKP-1 (3CH134), an immediate
early gene product, is a dual specificity
phosphatase that dephosphorylates MAP
kinase in vivo. Cell 75:48793
Keyse SM, Emslie EA. 1992. Oxidative
stress and heat shock induce a human gene
encoding a protein-tyrosine phosphatase.
Nature 359:64447
Farooq A, Zhou MM. 2004. Structure and
regulation of MAPK phosphatases. Cell.
Signal. 16:76979
Masuda K, Shima H, Katagiri C, Kikuchi
K. 2003. Activation of ERK induces phosphorylation of MAPK phosphatase-7, a
JNK specific phosphatase, at Ser-446. J.
Biol. Chem. 278:3244856
Brondello JM, Brunet A, Pouyssegur J,
McKenzie FR. 1997. The dual specificity mitogen-activated protein kinase
phosphatase-1 and -2 are induced by the
p42/p44MAPK cascade. J. Biol. Chem.
272:136876
Dowd S, Sneddon AA, Keyse SM. 1998.
Isolation of the human genes encoding
the pyst1 and Pyst2 phosphatases: characterisation of Pyst2 as a cytosolic dualspecificity MAP kinase phosphatase and
its catalytic activation by both MAP and
SAP kinases. J. Cell Sci. 111:338999
Grumont RJ, Rasko JE, Strasser A, Gerondakis S. 1996. Activation of the mitogenactivated protein kinase pathway induces
transcription of the PAC-1 phosphatase
gene. Mol. Cell. Biol. 16:291321
Groom LA, Sneddon AA, Alessi DR,
Dowd S, Keyse SM. 1996. Differential regulation of the MAP, SAP and
RK/p38 kinases by Pyst1, a novel cytosolic dual-specificity phosphatase. EMBO J.
15:362132
Bernaudin M, Tang Y, Reilly M, Petit E,
Sharp FR. 2002. Brain genomic response
following hypoxia and re-oxygenation
in the neonatal rat. Identification of
genes that might contribute to hypoxia-
71.
72.
73.
74.
75.
76.
77.
78.
4 Dec 2004
19:16
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.
96.
97.
98.
747
4 Dec 2004
19:16
748
99.
100.
101.
102.
103.
104.
105.
106.
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
DUCRUET ET AL.
carcinomas reveals a high frequency of allelic loss on chromosome 8p1221. Cancer Res. 56:241116
Yokota J, Tsukada Y, Nakajima T, Gotoh
M, Shimosato Y, et al. 1989. Loss of heterozygosity on the short arm of chromosome 3 in carcinoma of the uterine cervix.
Cancer Res. 49:3598601
Theodosiou AM, Rodrigues NR, Nesbit MA, Ambrose HJ, Paterson H, et al.
1996. A member of the MAP kinase phosphatase gene family in mouse containing a complex trinucleotide repeat in the
coding region. Hum. Mol. Genet. 5:675
84
Bepler G, Garcia-Blanco MA. 1994.
Three tumor-suppressor regions on chromosome 11p identified by high-resolution
deletion mapping in human non-smallcell lung cancer. Proc. Natl. Acad. Sci.
USA 91:551317
Hoornaert I, Marynen P, Goris J, Sciot
R, Baens M. 2003. MAPK phosphatase
DUSP16/MKP-7, a candidate tumor suppressor for chromosome region 12p12-13,
reduces BCR-ABL-induced transformation. Oncogene 22:772836
Montpetit A, Boily G, Sinnett D. 2002. A
detailed transcriptional map of the chromosome 12p12 tumour suppressor locus.
Eur. J. Hum. Genet. 10:6271
Martell KJ, Kwak S, Hakes DJ, Dixon
JE, Trent JM. 1994. Chromosomal localization of four human VH1-like proteintyrosine phosphatases. Genomics 22:462
64
Weber RG, Rieger J, Naumann U, Lichter
P, Weller M. 2001. Chromosomal imbalances associated with response to
chemotherapy and cytotoxic cytokines in
human malignant glioma cell lines. Int. J.
Cancer 91:21318
Ojopi EP, Rogatto SR, Caldeira JR,
Barbieri-Neto J, Squire JA. 2001. Comparative genomic hybridization detects
novel amplifications in fibroadenomas of
the breast. Genes Chromosomes Cancer
30:2531
107. Murty VV, Reuter VE, Bosl GJ, Chaganti RS. 1996. Deletion mapping identifies loss of heterozygosity at 5p15.1-15.2,
5q11 and 5q34-35 in human male germ
cell tumors. Oncogene 12:271923
108. Peng HQ, Liu L, Goss PE, Bailey D, Hogg
D. 1999. Chromosomal deletions occur in
restricted regions of 5q in testicular germ
cell cancer. Oncogene 18:327783
109. Faulkner SW, Friedlander ML. 2000.
Molecular genetic analysis of malignant
ovarian germ cell tumors. Gynecol. Oncol. 77:28388
110. Emslie EA, Jones TA, Sheer D, Keyse
SM. 1994. The CL100 gene, which encodes a dual specificity (Tyr/Thr) MAP
kinase phosphatase, is highly conserved
and maps to human chromosome 5q34.
Hum. Genet. 93:51316
111. Dorfman K, Carrasco D, Gruda M, Ryan
C, Lira SA, Bravo R. 1996. Disruption of
the erp/mkp-1 gene does not affect mouse
development: normal MAP kinase activity in ERP/MKP-1-deficient fibroblasts.
Oncogene 13:92531
112. Khosravi-Far R, White MA, Westwick
JK, Solski PA, Chrzanowska-Wodnicka
M, et al. 1996. Oncogenic Ras activation of Raf/mitogen-activated protein
kinase-independent pathways is sufficient
to cause tumorigenic transformation. Mol.
Cell Biol. 16:392333
113. Magi-Galluzzi C, Mishra R, Fiorentino
M, Montironi R, Yao H, et al. 1997.
Mitogen-activated protein kinase phosphatase 1 is overexpressed in prostate cancers and is inversely related to apoptosis.
Lab. Invest. 76:3751
114. Bang YJ, Kwon JH, Kang SH, Kim JW,
Yang YC. 1998. Increased MAPK activity
and MKP-1 overexpression in human gastric adenocarcinoma. Biochem. Biophys.
Res. Commun. 250:4347
115. Wang HY, Cheng Z, Malbon CC. 2003.
Overexpression of mitogen-activated protein kinase phosphatases MKP1, MKP2
in human breast cancer. Cancer Lett.
191:22937
4 Dec 2004
19:16
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
125.
126.
127.
128.
129.
130.
131.
132.
133.
134.
749
4 Dec 2004
19:16
750
135.
136.
137.
138.
139.
140.
141.
142.
AR
AR232-PA45-29.tex
XMLPublishSM (2004/02/24)
P1: JRX
DUCRUET ET AL.
products are biologically validated starting points in structural space for compound library development: Solid-phase
synthesis of dysidiolide-derived phosphatase inhibitors. Angew. Chem.-Int. Ed.
41:30711
Sodeoka M, Sampe R, Kojima S, Baba
Y, Usui T, et al. 2001. Synthesis of a
tetronic acid library focused on inhibitors
of tyrosine and dual-specificity protein
phosphatases and its evaluation regarding VHR and cdc25B inhibition. J. Med.
Chem. 44:321622
Kar S, Lefterov IM, Wang M, Lazo JS,
Scott CN, et al. 2003. Binding and inhibition of Cdc25 phosphatases by vitamin K
analogues. Biochemistry 42:1049097
Poigny S, Nouri S, Chiaroni A, Guyot
M, Samadi M. 2001. Total synthesis and
determination of the absolute configuration of coscinosulfate. A new selective inhibitor of Cdc25 protein phosphatase. J.
Org. Chem. 66:726369
Wipf P, Aslan DC, Luci DK, Southwick
EC, Lazo JS. 2000. Synthesis and biological evaluation of a targeted library of
protein phosphatase inhibitors. Biotechnol. Bioeng. 71:5870
Kitaide M, Nagai K, Terada T, Asao T,
Sugimoto Y, Yamada Y. 2001. JP Patent
No. 2003104964
Peng H, Xie W, Otterness DM, Cogswell
JP, McConnell RT, et al. 2001. Syntheses and biological activities of a novel
group of steroidal derived inhibitors for
human Cdc25A protein phosphatase. J.
Med. Chem. 44:83448
Fritzen EL, Brightwell AS, Erickson LA,
Romero DL. 2000. The solid phase synthesis of tetrahydroisoquinolines having
cdc25B inhibitory activity. Bioorg. Med.
Chem. Lett. 10:64952
Galcera Contour MO, Lavergne O, Brezak
143.
144.
145.
146.
147.
148.
149.
150.
151.