You are on page 1of 12

KINETIC ALTERATION OF HUMAN ALDOSEREDUCTASE

BY MUTAGENESIS OF CYSTEINE RESIDUES

J. Mark Petrash, Theresa Harter, Ivan Tarle and David Borhani*

Departments of Ophthalmology and Visual Sciences and of Genetics,


Washington University School ofMedicine, St. Louis, MO 63110 and
*BioCryst Pharmaceuticals, Inc., Birmingham, AL 35244

INTRODUCTION

Aldose reductase (ALR2 1 : alditol:NADPH oxidoreductase: E.C. 1.1.1.21) catalyzes the


NADPH-linked reduction of aldoses to their corresponding alcohols or polyols, the first step
of the polyol pathway. Enhanced flux of glucose through the polyol pathway and consequent
biochemical imbalances are thought to be crucial to the onset and progression of many com-
plications of diabetes mellitus including cataract, retinopathy, neuropathy and nephropathy
(Kinoshita and Nishimura, 1988). In light of its rate-limiting position in the polyol pathway
as well as its apparent metabolic dispensability (Yancey et al., 1990), strategies to control or
prevent the onset of diabetic complications through inhibition of aldose reductase are being
aggressively pursued. While a structurally-diverse array of aldose reductase inhibitors (ARI)
have yielded impressive results in animal studies, their effectiveness in arresting or preventing
diabetic neuropathy (Boulton et al., 1990) and retinopathy (Sorbinil Retinopathy Trial
Research Group, 1990) in human trials has been less encouraging (Frank, 1990).
Altered (so-called "activated") forms of aldose reductase, characterized by changes in
kinetic parameters for aldose and aldehyde substrates and reduced susceptibility to inhibition
by a variety of ARis, have been described recently (Das and Srivastava, 1985; Del Corso et
al., 1989a,b; Vander Jagt, 1990). Such isoforms have been extracted from tissues obtained
frolll diabetic patients (Das and Srivastava, 1985), from cells cultured under hyperglycemic
conditions (Srivastava et al., 1986), or produced by treatment of purified aldose reductase
under conditions thought to induce protein oxidation (Del Corso et al., 1987; Vander Jagt et
al., 1990). Conversion from native to the modified form is prevented or reversed by thiol
reducing agents and occurs with loss of two titratable protein thiol groups, lending support to
the idea that formation of an intramolecular disulfide bridge is the structural basis for the
altered kinetic behavior (Vander Jagt et al., 1990). The rationale for studying this process
takes on added importance if one considers that inhibition of altered forms of aldose
reductase in vivo would be expected to be very poor considering the marked loss of
sensitivity of these isoforms to ARis.
Human aldose reductase contains 7 cysteine residues, 2 or 3 of which appear to be
relatively solvent-accessible on the basis of their reaction kinetics with 5,5'-dithiobis-(2-nitro-
benzoic acid) (Liu et al., 1989; Vander Jagt et al., 1990). The recently-solved structure of

Enzymology and Molecular Biology of Carbonyl Metabolism 4


Edited by H. Weiner, Plenum Press, New York, 1993 289
porcine lens aldose reductase revealed the presence of 3 solvent exposed cysteine residues
located in a region of the protein relatively near the active site (Rondeau et al., 1992). These
cysteine residues correspond to cysteines -80, -298 and -303 in human aldose reductase . In
order to investigate the potential involvement of these cysteines in catalysis, we constructed
mutants with serine substitutions at each individual cysteine site. Mutant ALR2:C298S
displayed kinetic properties characteristic of the "activated" aldose reductase such as elevated
apparent Km and Vmax for aldose and aldehyde substrates and reduced sensitivity to certain
aldose reductase inhibitors. We have examined the structure of this enzyme by x-ray
crystallography to evaluate the spatial distribution and potential interaction of the remaining
cysteine residues.

EXPERIMENTAL METHODS

Complementary DNA sequences encoding aldose reductase were isolated from a human
placenta cDNA library by screening with a bovine lens aldose reductase cDNA (Petrash et al.,
1989). Plasmids containing cDNA sequences were recovered from bacteriophage vectors by
excision with helper phage (Figure 1). Aldose reductase coding sequences were then trans-
ferred into the expression plasmid pMON5842 (see Olins and Rangwala, 1990) giving rise to
pMON5997 . For mutagenesis, single stranded wild type coding sequences were produced
from phagemid stocks and reacted with synthetic oligonucleotides encoding one or more
nucleotide substitutions using the site directed mutagenesis procedure developed by Kunkel
and coworkers (1987). The mutated coding sequences were then exchanged for wild type
sequences in pMON5997 for over-expression in Escherichia coli strain JM101. Nucleotide
sequencing of mutant plasmids confirmed the mutations and assured that no unintended
nucleotide sequence changes had been incorporated .
Expression of human aldose reductase in shaker flask cultures of Escherichia coli was
conducted as described previously (Petrash et al., 1992). Recombinant aldose reductase was
extracted from host cells by osmotic shock, concentrated by precipitation with 50-80% am-
monium sulfate, and purified by chromatofocusing and hydroxylapatite column chromatogra-
phy. All purification steps were conducted at 4°C in buffers containing 1 mM DTT .
Crystals of the ALR2:C298S-NADPH complex were grown by vapor diffusion in
hanging drops containing 15 mg/ml ALR2:C298S, 9% polyethylene glycol (PEG) 6000, 25
mM MES, 1 mM DTT, 0.1 mM EDTA, 1 mM NADPH, pH 7. Drops were equilibrated at
4°C against 1 MNaCl; box-like crystals grew from precipitated protein in a few days. These
crystals, which belong to space group 1222, a=74.71 A, b=85.32 A, c=106.84 A, diffract X-
rays to at least 2.2 A resolution. The crystal structure was solved by the method of
• molecular replacement, using porcine aldose reductase apoenzyme as the search model
(Rondeau et al., 1992; J.-M Rondeau, F. Tete-Favier, A. Podjarny, D. Moras, personal com-
munication). Rigicl body refinement (CORELS; Sussman et al., 1977) of the initial model (R-
factor 0.456, 20-3 A, F>crF), deletion of three large loops, and further rigid body refinement
reduced the R-factor to 0.305 (6-3 A) . Multiple cycles of least squares refinement of the
model (TNT ; Tro"~'1.!det al., 1987) and manual rebuilding using computer graphics (FRODO;
Jones, 1978) allowed the re-introduction of most of the deleted residues, and gradual exten-
sion to higher resolution. Since clear difference density was present for the coenzyme,
NADPH was added to the model at this stage. The current model contains 308 amino acid
residues and NADPH (2366 atoms); the R-factor for all data from 6-2.4 A is 0.286, with
RMS deviations in bond lengths of0 .022 A and in bond angles of3 .85°. Completion of the
refinement at the diffraction -limit of these crystals (2.2 A) is in progress .

290
S~--E-•~_A_L_:_K_?_D~N-A_~_~½A~R~

/n v,·vo excision with

l
Helper phage

Eco RI NcoI Eco RI


H,ndllI
ALR2cDNA
Hu PAR 53- SK-
Amp'

Nco l
!
EcoR[ Htndill

½
~ ;lR2cDNA]
+ H,nd m
(
Transcr iptionol
Ptoc terminat ors

pMON 5842

ALR2cDNA
!
pMON 5997

Figure 1. Construction ofaldose reductase expression plasmid pMON5997. cDNA sequences contained in
bacteriophage isolate l PAR 53 were recovered by in vivo excision of its phagemid vector (HuPAR53SK-).
cDNA sequences were excised from the phagemid by digestion with Hind III and Ncol ; the resulting cDNA
fragment was transferred to expression vector sequences derived from pMON5842 (a derivative of the vector
described by Olins and Rangwala , 1990) giving rise to pMON5997. Functional elements in the expression
plasmids include a tac promoter (Ptac), a ribosome binding domain (glO-L rbs), transcriptional terminators
derived from phage P22 ant gene, and a spectinomycin resistance gene (Spr).

291
RESULTS AND DISCUSSION

Cloning and Expression of Human Aldose Reductase

Marked variability observed in the properties of human aldose reductase extracted from
different tissues or studied at various levels of purification (Poulsom, 1987) provide the
rationale for establishing a stable and renewable source of aldose reductase for detailed
chemical study. In light of the significant occupational hazards faced when using human
donor tissue, animal tissues would seem to be an alternative source of starting material.
However, species differences in the kinetic properties of aldose reductases also indicate that
evaluation of the enzyme with regard to potential human therapy would best be carried out
with enzyme derived from human sources (Kador et al., 1980). To satisfy these needs, we
focused our initial efforts on establishing a recombinant source of the enzyme by cloning and
expressing cDNA sequences encoding human aldose reductase.
Human aldose reductase cDNA clones were isolated from a placenta library by screening
with a probe derived from bovine lens aldose reductase (Petrash and Favello, 1989). The
coding regions of 6 out of 36 aldose reductase cDNA clones isolated from the library were
substantially or completely analyzed by nucleotide sequencing and were found to contain
identical sequences in their overlapping regions. The consensus sequence was also perfectly
co-linear with human aldose reductase cDNA sequences published previously (Graham et al.,
1989; Grundmann et al., 1990). These results indicate that a single aldose reductase gene is
expressed in human placenta and that the coding sequence is identical to that expressed in
human muscle and retina (Nishimura et al., 1990).
Steps involved in the transfer of aldose reductase coding sequences from the original
bacteriophage cloning vector to the final expression vector construct are outlined in Figure 1.
Abundant quantities of aldose reductase activity were obtained from Escherichia coli cul-
tures transformed with the aldose reductase expression plasmid pMON5997. As noted
previously (Grundmann et al., 1990), the enzyme could be readily extracted from host cells
by osmotic shock, presumably reflecting its localization in the periplasmic space. Rapid
purification by chromatofocusing (Figure 2) and hydroxylapatite chromatography provided an
enzyme preparation virtually homogeneous judging by the appearance of a single protein

75 I
3.0
Polyb uffe,
6.5 i°a.
Ie 25
Appl;ed
5.5 2 .5

~ 2 .0 2.0
IE
>-" C:
>- 1.5 0
1.5
~ 00
N
>-
LJ.J
~ 1.0 1.0 u
z
LJ.J <l:
~ 0 .5 0 .5 co
N
z
""
0
V>
LJ.J 0 0 co
·0 .25 <!
0 . 100 200 300 400 500 600
ELUTION VOLUME (ml)

Figure 2. Purification of recombinant human aldose reductase by column chromatofocusing. This


representative chromatogram was obtained by loading approximately 200 mg protein extracted from E. coli
cells transformed with pMON5997. Each fraction was measured for absorbance at 280 nm (•-•), aldose
reductase activity (o-o) and pH( •-•) .

292
band following SDS-polyacrylamide gel electrophoresis (SDS-P AGE). Recombinant and
native human aldose reductases co-migrated at a position corresponding to M,-36,000 on
SDS-PAGE and cross-reacted with antibodies directed against bovine lens aldose reductase .
Ten cycles ofEdman degradation of the recombinant aldose reductase revealed the following
sequence: NHrAla-Ser-Arg-Leu-Leu-Leu-Asn-Asn-Gly-Ala -COO- . This structure agrees
with the sequence predicted from the cDNA, and indicates that the amino terminal alanine
residue is not covalently blocked (acetylated) as is the case for aldose reductase synthesized
in eucaryotic cells (Schade et al., 1990).
Although other investigators have reported over-expression of recombinant human
aldose reductase in procaryotic (Grundmann et al., 1990; Carper et al., 1990; Bohren et al.,
1992) and eucaryotic (Nishimura et al, 1990) host cells, concerns about the usefulness and
reliability of recombinant-derived aldose reductase as a substitute for enzyme purified from
human tissues have been raised (Carper et al., 1991). Therefore, we devoted considerable
effort in our initial studies to evaluate the relatedness of aldose reductases from recombinant
and native sources. Michaelis constants (KnJand turnover numbers (kc aUmeasured with
recombinant human aldose reductase using a variety of structurally-diverse substrates were
strikingly similar to that reported previously for native human aldose reductase (Table I).
The sensitivity of recombinant aldose reductase to inhibition by a variety of ARls was also
similar to that reported for aldose reductase extracted from native human tissues. Tolrestat
(N-[[ 5-(trifluoromethyl)-6-methoxy- l -naphthalenyl]thioxomethyl]-N-methylglycine) and
zopolrestat (3, 4-dihydro-4-oxo-3-[[ 5-(trifluoromethyl)-2-benzothiazolyl]methyl]- l -phthal-
azineacetic acid); Mylari et al, 1991) were potent inhibitors, with Ki values -6 to 16-fold
lower than that for sorbinil ((S)-6-fluoro-spiro-[chroman-4,4'-imidazolidine]-2',5'-dione).

0 ,,;,:
j/
/~

-2 .:\ ,/
\i
~
it
0
E -4

\ j
-0 .

E
u
bn -6
<I)

::::.
M

6 'III /
\,\\ ;~---,J,j
\ /

1r
X -8
~
I ,- ,/
' ' --
-10
I\\ tl
' , -.1

- 12
200 2 10 220 230 240 250
Wavelength (nm)
Figure 3. Circular dichroism spectra of human aldose reductase . Spectra of the enzyme (5 µM in 50 mM
potassium phosph ate, pH 7.0) and coenzyme or coenzyme analogs (~200 µM) were recorded at 23°C using a
0.1 cm path length quartz cell under constant nitrogen flush . Values are corrected for buffer and coenzyme
(or analog). ALR2 (--) ; ALR2 plus NADPH (- -) ; ALR2 plus NADP+ (- - -). Spectra collected with
ALR2 in presence of 2',5'-adenosine diphosphate were identical to that collected with the enzyme alone.

293
We also examined the interaction of the recombinant ALR2 with NADPH, NADP+ and
the coenzyme analog adenosine 2',5'-diphosphate (2',5'-ADP). As shown in Figure 3, binding
of either NADPH or its oxidized form, NADP+, induced a major conformational change as
measured by circular dichroism spectroscopy. In contrast, binding of 2',5'-ADP failed to
induce a measurable conformational change although this analog was a potent inhibitor
(Ki=S µM; competitive versus NADPH) . Crystallographic studies have now provided the
structural basis for this conformational change (vide infra).

Table 1. Comparison ofRecombinant and Native Human Aldose Reductases .

Recombinant ALR2 Native ALR2a

Specific Activity
(U/mg) 3.5 2.4

DL-glyceraldehyde
K... (µM) 54 72
kcat (sec· 1) 1.0 1.2
kcat'K... (sec· 1M· 1) 1.9 X 104 2 X 104

D-glucose
K... (mM) 212 651
kcat (sec·') 0.09 0.61
kca1'K...(sec•IM-1) 0.41 0.94

Benzaldehyde
K... (µM) 73 224
kca1(sec- 1) 0.44 0.86
kca/Km (sec· 1M· 1) 6.lxl0 3
4 X 103

p-nitrobenzaldehyde
K...(µM) 16 31
kcal (sec-I) 0.47 1.3
kca/Km (sec· 1M· 1) 3 X 104 4 X 104

Effect oflnhibitors (Ki, µM)


Sorbinil 0.098 0.22b
Tolrestat 0.015 0.017b
Zopolrestat 0.006 0.002c

Specific activity was determined at 23°C using 5 mM DL-glyceraldehyde, 150 µM NADPH, 0.4 M lithium
sulfate, 10 mM sodium phosphate, pH 6.2, and 5 mM ~-mercaptoethanol. Substrate kinetic parameters were
determined under these conditions but with the indicated substrates. Inhibition constants (Ki) were
determined at pH 7.1 in 50 mM potassium phosphate, 0.4 M ammonium sulfate, 150 µM NADPH with DL-
glyceraldehyde as variable substrate. Comparative data are from: a (Morjana and Flynn, 1989); b (Vander
Jagt et al., 1990); c (Mylari et al., 1991).

294
Mutagenesis of Cys-80, -298, and -303

Recent structure determination of porcine and human aldose reductases revealed that the
enzyme consists of a j3/a.-barrel structural motif, with the coenzyme-binding domain located
at the carboxy-terminal ends of the 13-strands (Rondeau et al., 1992; Wilson et al., 1992;
Borhani et al., 1992). Three cysteine residues, namely those corresponding to cysteine resi-
dues 80, 298 and 303 in human aldose reductase, are located in this region. Previous studies
have implicated the involvement of cysteine residues in inhibitor and coenzyme binding and in
post-translational modifications leading to enzyme isoforms with altered kinetic properties.
As an initial attempt to evaluate the potential role of these cysteine residues in catalysis and
inhibitor binding, we examined the kinetic properties of mutants containing serine substitu-
tions at cysteine residues 80 (ALR2:C80S), 298 (ALR2 :C298S) and 303 (ALR2 :C303S) .

Kinetic Effects of Cysteine Mutations. In comparison with ALR2:WT, all three


cysteine • serine mutants were expressed with equal if not greater efficiency in E. coli and
could be routinely isolated to apparent homogeneity without modification of the purification
procedure. All mutants exhibited robust reductase activity with a variety of aldose and aro-
matic aldehyde substrates. Although detailed kinetic properties of these mutants will be pub-
lished elsewhere (Petrash et al., 1992), three model substrates are presented here . These
compounds were chosen as they represent three structural classes of potential aldose
reductase substrates: glyceraldehyde as a model for short-chain 11aldoses 11 or trioses (Vander
Jagt, 1992), galactose as representative of potential hexose substrates utilized in the polyol
pathway and linked to ocular disease (Kinoshita and Nishimura, 1988), and benzaldehyde as
an aromatic aldehyde similar in structure and hydrophobicity to other substituted aromatic
aldehydes that could be potential physiological substrates of the aldo-keto reductases (Flynn,
1982).
When quantified under standard assay conditions (5 mM DL-glyceraldehyde as substrate
and 150 µMNADPH as coenzyme), the specific activity (units/mg) of the C80S and C303S
mutants was approximately 50% that of wild type (Table 2) . However, the specific activity
of the C298S mutant was approximately 2.5-fold that of wild type, due in part to an almost
6-fold increase (relative to wild type) in kcat· The apparent Km (DL-glyceraldehyde)for all
mutants was elevated, particularly so for ALR2:C298S which was increased
approximately 57-fold compared to wild type. Despite differences in their respective Km and
kcat values, the catalytic efficiency (kcat1Km) measured with DL-glyceraldehyde was uniformly
decreased among all the mutants . Similar changes in kinetic parameters were also apparent
when reactivity with other substrates was examined. As with DL-glyceraldehyde, the Km and
kcat (measured with galactose and benzaldehyde) observed with ALR2:C298S were markedly
elevated relative to wild type . Mutants C80S and C303S showed higher (approximately 6 to
IO-fold) apparent Km (galactose); utilization of this substrate by the C303S mutant was
particularly poor (kcat!Km reduced >50-fold). In contrast, the C80S was notable in its
enhanced reactivity with benzaldehyde ( decreased Km and increased kcat; kcat1Km increased
to a level 4.5-fold that of wild type).
The kinetic behavior of mutant C298S is strikingly similar to that attributed to
11
activated II forms of aldose reductase . This is exemplified by its elevated Km and V max for
aldehyde substrates and its reduced sensitivity to inhibition by ARis such as sorbinil. As
conversion of native aldose reductase to the modified form is thought to occur concomitantly
with formation of an intramolecular disulfide bridge, a structural model of C298S was exam-
ined to evaluate the spatial distribution and chemical environment of the cysteine residues.
STRUCTURE OF ALR2:C298S COMPLEXED TO NADPH

Recombinant human aldose reductase (C298S mutant) was crystallized in the presence of
NADPH. These crystals, which diffract X-rays to at least 2.2 A resolution, belong to a

295
different space group (/222) than previously reported crystals of porcine aldose reductase
apoenzyme (Pl; Rondeau et al., 1992) or wild type recombinant human aldose reductase
complexed with NADPH (P2 12 12 1; Wilson et al., 1992). The crystal structure of
ALR2:C298S·NADPH was solved by molecular replacement; the search model was one
monomer of porcine aldose reductase apoenzyme. The structure has been refined at 2.4 A
resolution to an R-factor of 0.286.
The structure of ALR2:C298S is for the most part identical to the previously described
structures of wild type human (Wilson et al., 1992) and porcine (Rondeau et al., 1992) aldose
reductases . As shown in Figure 4, ALR2:C298S consists predominantly of a (i3/a) 8-barrel;
this regular fold is preceded by a ~-hairpin, and is interrupted by three large C-terminal loops
and two extra a-helices . Comparison of our model with that of the porcine apoenzyme
reveals the nature of the conformational changes which occur upon coenzyme binding. Upon
binding NADPH, loop 7 of ALR2:C298S rotates by 51° around Gly 213 and Ser 214. This
movement brings residues 213-216 into contact with the pyrophosphate and ribose-2'-
phosphate regions of NADPH, and locks the coenzyme into its binding cleft. The detailed
interactions between aldose reductase and NADPH which we observe appear to be identical
to those found for the wild type recombinant human enzyme (Wilson et al., 1992). When the
localized changes which occur upon coenzyme binding are excluded from the analysis, our

Table 2. Kinetic constants for wild type and mutant aldose reductases

Wild Type C80S C298S C303S

Specific Activity 3.5 1.7 9.0 1.7

DL-glyceraldehyde
¾(µM) 54 195 3100 405
kcat (sec·1) 1.02 0.38 5.82 0.82
kcat~ (M· 1sec·1) 18907 1931 1878 2035

D-Galactose
¾ (mM) 94.5 578 2620 889
kcal (sec•!) 0.32 0.83 1.93 0.05
kcal~ (M•lsec•l) 3.4 1.4 .74 .06

Benzaldehyde
¾(µM) 73 28 1061 68
kcal (sec•l) 0.44 0.76 4.62 0.30
kca/Km (M•lsec•l) 6093 27473 4404 4420

K; sorbinil (µM)* 0.37 0.84 3.81 0.55

Kinetics measurements were .carried out with recombinant enzymes purified to apparent homogeneity.
Specific activities were determined as described in Table 1. Kinetics constants were determined at 23°C in 50
mM potassium phosphate buffer, pH 7.0, containing 0.4 M ammonium sulfate and 150 µM NADPH. Rate
measurements were analyzed using a software program originally described by Cleland (1979). Protein
concentration was determined by the dye-binding method of Bradford (1976). *Prior to kinetics analysis to
determine Ki values, the enzymes were treated with 0. 1 M OTT, followed by rapid filtration through a
desalting column (Petrash et al., 1992).

296
Figure 4. Stereoview of the a-carbon backbone of ALR2:C298S. The view is down the mouth of the barrel;
the N-terminal 13-hairpin is at the bottom center of the picture, and the C-terminus is at the upper left.
NADPH is shown, as are the six cysteines and Ser 298, which are labelled at they-sulfur (oxygen) atoms.
The dashed lines represent residues 217-229 of loop 7, which are not yet included in our model.

Figure S. Stereoview of the positions of Ser 298, Cys 80, and Cys 303 relative to the active site of
ALR2:C298S. The direction of view is similar to Figure 4. Ser 298, Cys 80, Cys 303, and Phe 115 are
labelled. The nicotinamide ribose portion of NADPH is visible at the lower right.

297
model of ALR2 :C298S superimposes on the porcine apoenzyme model with an RMS
deviation of0 .6 A (293 a-carbon atoms) .
The proximity of cysteine residues 80, 298 and 303 to the active site of aldose reductase
is clear from our structure . As shown in Figure 5, the nicotinamide ring ofNADPH is tucked
deeply in the center of the 13-barrel,the walls of which are lined with hydrophobic residues.
The active site of aldose reductase, as identified by the pro-R C4 hydrogen atom ofNADPH,
thus has the shape of an inverted cone, with the nicotinamide ring exposed to solvent at the
point of the cone. The locations of cysteines 80 and 303, as well as the mutated residue,
serine 298, are essentially identical in our structure and the porcine apoenzyme. These
residues form the apices of an isosceles triangle; the sulfur atoms of Cys 80 and Cys 303 are
separated by 6.6 A, and both of these sulfur atoms are about 10.5 A from the hydroxyl
oxygen atom of Ser 298 . Cysteines 80 and 303 are quite distant from the active site. In
contrast, Ser 298, which forms part of the upper wall of the inside of the 13-barrel,is in van
der Waals contact with the nicotinamide ring ofNADPH and thus is clearly "in" the active
site.
In our structure, we find no evidence for the formation of any disulfide bridges . It is
clear that Cys 80, Cys 298, and Cys 303 are exposed to solvent; these residues are suscepti-
ble to chemical modification (Rondeau et al., 1992). Formation of a disulfide bridge between
the two most closely apposed of these three residues (Cys 80 and Cys 303) would require the
movement of Phe 115, which physically separates Cys 80 and Cys 303. Movement of Phe
115 would be expected to be thermodynamically unfavorable, as it would require the disrup-
tion of the hydrophobic packing of numerous residues in this region of the enzyme (see
Figure 5). As can be seen in Figure 4, the remaining cysteines, Cys 44, Cys 92, Cys 186, and
Cys 199, are all distant from the active site(> 12.5 A), distant from each other(> 11.0 A), and
relatively sequestered in the interior of the enzyme. Disulfide bridges among these residues
would appear to be impossible to construct. On the basis of our model, one may conclude
that the modified kinetic behavior of ALR2:C298S does not arise from the formation of a
disulfide bridge between any two of the remaining 6 cysteine residues . Indeed, our structural
model would predict that none would be capable of forming without dramatic reorganization
of the hydrophobic core of the !3/cx. barrel. Therefore, we question the hypothesis that
oxidative modification of aldose reductase involves formation of . a disulfide bridge
(Bhatnagar et al., 1989; Vander Jagt et al., 1990).
It seems plausible to conclude that cysteine residues do not participate directly in cataly-
sis, as the cysteine mutants containing serine substitutions nearest the active site displayed
near-wild type activity levels under standard assay conditions . However, demonstration that
any given residue is "unessential" to the catalytic reaction will require the construction of
additional mutants with other structurally conservative substitutions at each given site.
Further study with additional substrates will also be required to sort out the apparent struc-
tural and/or functional roles each of these residues may contribute to the enzyme. As pointed
out by Plapp and coworkers in their study of alcohol dehydrogenases (1991), single muta-
tions can potentially affect many steps in the catalytic mechanism and that is probably what
we are measuring with these cysteine mutations . It is difficult to ascribe a definitive role to
Cys 298 in the catalytic mechanism, and thereby understand why mutation of this residue to
serine affects the enzyme's kinetic parameters. One possibility consistent with the kinetic and
structural data we have presented is that Cys 298 helps to bind and position the nicotinamide
ring of NADPH and/or the substrate, and that mutation of this residue to serine appears to
modify these interactions. Alternatively, mutation of this residue may affect the rate at which
enzyme isomerization occurs with the binding and release of NADP(H) (Borhani et al.,
1992), a step thought to be rate-limiting for aldehyde reduction catalyzed by aldose reductase
(Kubiseski et al., 1992).

298
ACKNOWLEDGMENTS

This work was supported by research grants from the National Eye Institute (EY05856
and 5T32 EY07108), a Diabetes Research and Training Center grant (P60 DK20579) and in
part by grants to the Department of Ophthalmology and Visual Sciences from Research to
Prevent Blindness, Inc. We would like to thank Dr . Jean-Michel Rondeau (Biostructure,
S.A.), and Dr. Alberto Podjarny and Professor Dino Moras (University of Strasbourg) for
providing the porcine aldose reductase apoenzyme and 2'-monophosphoadenosine-5'-diphos-
phoribose (ADPRP)-complex coordinates to BioCryst prior to publication.

REFERENCES

Bhatnagar, A., Liu, S., Das, B., Srivastava , S.K., 1989, Involvement ofsulfhydryl residues in aldose
reductase-inhibitor interaction . Mo/. Pharmacol. 36,825-830 .
Bohren , K.M ., Page, J.L., Shankar, R., Henry, S.P., and Gabbay, K.H., 1992, Expression of human aldose
and aldehyde reductases: Site-directed mutagenesis of a critical lysine 262. J. Biol. Chem. 266,24031-
24037.
Borhani, D.W ., Harter, T.M., and Petrash, J.M., 1992, The crystal structure of the aldose reductase ·NADPH
binary complex. J. Biol. Chem. (submitted).
Boulton , A.J.M., Levin, S., and Comstock, J., 1990, A multicentre trial of the aldose-reductase inhibitor,
tolrestat, in patients with symptomatic diabetic neuropathy . Diabeto/ogia , 33 :431-437 .
Bradford , M .M. , 1976, A rapid and sensitive method for the quantitation of microgram quantities of protein
using the principle of dye-binding. Anal. Biochem. 72,248-254.
Carper, D., Sato, S., Old, S., Chung, S., and Kador , P.F ., 1990, In vitro expression of human placental
aldose reductase in Escherichia coli. In, : "Enzymology and Molecular Biology of Carbonyl Metabolism
3," H. Weiner et al., eds., Plenum Press , New York, NY.
Carper , D.A., Old, S.E., Sato, S. and Kador, P.F ., 1991, Characterization of recombinant human placenta and
rat lens aldose reductase expressed in Escherichia coli . ARVO abstracts . Supplement to Invest.
Ophthal. Vis. Sci., 1991, p.975 , J.B. Lippincott, Philadelphia.
Cleland, W.W., 1979, Stastical analysis of enzyme kinetic data. Methods Enzymol. 63,103-138.
Das, B. and Srivastava, S.K., 1985, Activation ofaldose reductase from human tissue. Diabetes 34,1145-
1151.
Del Corso, A., Camici, M. and Mura, U., 1987, In vitro modification of bovine lens aldose reductase activity.
Biochem. Biophys . Res. Commun. 148:369-375 .
Del Corso, A., Barsacchi , D. , Camici , M ., Garland , D., and Mura , U. 1989a, Bovine lens aldose reductase :
Identification of two enzyme forms. Arch. Biochem . Biophys . 270:604-610 .
Del Corso, A., Barsacchi , D., Giannessi , M. , Tozzi, M.G., Camici , M., and Mura , U., 1989b, Change in
stereospecificity of bovine lens aldose reductase modified by oxidative stress . J. Biol. Chem. 264,17653-
17655.
Del Corso, A., Barsacchi , D ., Giannessi, M ., Tozzi, M.G., Camici, M ., Houben, J.L., Zandomeneghi, M., and
Mura, U., 1990, Bovine lens aldose reductase: Tight binding of the pyridine coenzyme. Arch.
Biochem. Biophys. 283:512-518 .
Flynn, T.G., 1982, Aldehyde reductases : Monomeric NADPH-depencent oxidoreductases with
multifunctional potential. Biochem . Pharmacol. 31,2705-2712 .
Frank, R.N ., 1990, Aldose reductase inhibition : The chemical key to the control of diabetic retinopathy?
(editorial)Arch. Ophthal. 108, 1229-1231.
Graham, A., Hedge, P.J., Powell, S.J., Riley, J., Brown, L., Gammack , A., Carey, F ., and Markham,
A.F., 1989, Nucleotide sequence of cDNA for human aldose reductase . Nucl. A cids Res . 17, 8368.
Grunrlmann, Ulrich , Bohn, H., Obermeier , R., and Amann, E., 1990, Cloning and prokaryotic expression ofa
biologically active human placental aldose reductase . DNA and Cell Biol . 9, 149-157.
Jones, T.A., 1978, A graphics model building and refinement system for macromolecules. J . Appl . Cryst.
11,268-272.
Kador, P .F., Kinoshita, J.H., Tung, W.H., and Chylack, L.T .,Jr., 1980, Differences in the susceptibility of
various aldose reductases to inhibition . Invest . Ophtha/. Vis. Sci. 19:980-982
Kinoshita , J.H. and Nishimura, C., 1988, The involvement of aldose reductase in diabetic complications.
Diabetes/Metabolism Reviews , 4:323-337.

299
Kubiseski, T.J., Hyndman, D.J., Morjana, N.A., Flynn, T.G., 1992, Studies on pig muscle aldose reductase:
Kinetic mechanism and evidence for a slow conformational change upon coenzyme binding. J. Biol.
Chem. 267,6510-6517.
Kunkel, T.A., Roberts, J.D. and Zakour, R.A., 1987, Rapid and efficient site-specific mutagenesis without
phenotypic selection. Methods Enzymol. 154, 367-382.
Liu, S., Bhatnagar, A., Das, B., and Srivastava, S.K., 1989, Functional cysteinyl residues in human placental
aldose reductase, Arch. Biochem . Biophys. 275:112-121.
Morjana, N.A. and Flynn, T.G., 1989, Aldose reductase from psoas ·muscle: Purification, substrate
specificity, immunological characterization, and effect of drugs and inhibitors, J. Biol . Chem. 264,2906-
2911.
Mylari, B.L., Larson, E.R., Beyer, T.A., Zembrowski, W.J., Aldinger, C.E., Dee, M.F., Siegel, T.W ., and
Singleton, D.H., 1991, Novel, potent aldose reductase inhibitors: 3,4-dihydro-4-oxo-3-[[5-
(trifluoromethyl)-2-benzothiazolyl]methyl]-l-phthalazine-acetic acid (Zopolrestat) and congeners, J.
Med. Chem. 34,108-122.
Nishimura, C., Matsuura, Y., Kokai, Y. Akera, T., Carper, D., Morjana, N., Lyons, C., and Flynn, T.G.,1990,
Cloning and expression of human aldose reductase. J. Biol. Chem. 265, 9788-9792.
Olins, P.O. and Rangwala, S.H., 1990, Vector for enhanced translation of foreign genes in Escherichia coli.
Methods Enzymol. 185,115-119.
Petrash, J.M., Harter, T.M., Devine, C., Olins, P., Bhatnagar, A., Liu, S., and Srivastava, S.K., 1992,
Involvement of cysteine residues in catalysis and inhibition of human aldose reductase: Site-directed
mutagenesis of cys-80, -298 and -303. J. Biol. Chem. (submitted) .
Petrash, J.M. and Favello, A.D. , 1989, Isolation and characterization of cDNA clones encoding al dose
reductase. Curr. Eye Res. 8, 1021-1027.
Plapp, B.V., Ganzhom, A.I., Gould, R.M., Green, D.W., Warth, J.T., and Kratzer, D.A., 1991, Catalysis by
yeast alcohol dehydrogenase . Adv. Exp. Med. Biol. 284,241-251.
Poulsom, R., 1987, Comparison of aldose reductase inhibitors in vitro: Effects of enzyme purification and
substrate type. Biochem. Pharm. 36,1577-1581.
Rondeau, J.-M., Tete-Favier, F., Podjamy, A., Reymann, J.-M., Barth, P., Biellmann, J.-F., and Moras, D.,
1992, Novel NADPH-binding domain revealed by the crystal struture of aldose reductase. Nature
355:469-472.
Schade, S.Z., Early, S.L., Williams, T.R., Kezdy, F.J., Heinrikson, R.L., Grimshaw, C.E. and Doughty, C.C.,
1990, Sequence analysis of bovine lens aldose reductase. J. Biol. Chem. 265,3628-3635.
Sorbinil Retinopathy Trial Research Group, 1990, A randomized trial of sorbinil, an aldose reductase
inhibitor, in diabetic retinopathy. Arch. Ophthalmol. 108, 1234-1244.
Srivastava, S.K., Ansari, N.H., Hair, G.A., Jaspan, J., Rao, M.B., and Das, B., 1986, Hyperglycemia-induced
activation of human erythrocyte aldose reductase and alteration in kinetic properties. Biochim. Biophys.
Acta 870:302-311.
Sussman, J.L., Holbrook, S.R., Church, G.M., Kim, S.-H., 1977, A structure-factor least-squares refinement
procedure for macromolecular structures using constrained and restrained parameters. Acta Cryst.
A23,800-804.
Tronrud, D.E., Ten Eyck, L.F., Matthews, B.W., 1987, An efficient general-purpose least-squares refinement
program for macromolecular structures. Acta Cryst. A43,489-501.
Vander Jagt, D.L., Robinson, B., Taylor, K.K., and Hunsaker, L.A., 1990, Aldose reductase from human
skeletal and heart muscle: Interconvertible forms related by thiol-disulfide exchange. J. Biol. Chem.
265: 20982-20987.
Vander Jagt, D.L., Robinson, B., Taylor, K.K., and L. Hunsaker, 1992, Reduction oftrioses by NADPH-
dependent aldo-keto reductases: Aldose reductase, methylglyoxal, and diabetic complications. J. Biol.
Chem. 267,4364-4369.
Wilson, D.K., Bohren, K.M., Gabbay, K.H., and Quiocho, F.A., 1992, An unlikely sugar substrate site in the
1.65A structure of the human aldose reductase holoenzyme implicated in diabetic complications.
Science 257:81-84.
Yancey, P.H., Haner, R.G. , and Freudenberger, T.H., 1990, Effects of an aldose reductase inhibitor on
organic osmotic effectors in rat renal medulla. Am. J. Physiol. 259,F733-F738 .

300

You might also like