Biochimica et Biophysica Acta 1459 (2000) 266^273

www.elsevier.com/locate/bba

Nitric oxide reductases in bacteria

Janneke Hendriks 1 , Arthur Oubrie 1 , Jose Castresana, Andrea Urbani,
Sabine Gemeinhardt, Matti Saraste *
European Molecular Biology Laboratory, Meyerhofstrasse 1, Postfach 102209, D-69012 Heidelberg, Germany

Received 12 April 2000; received in revised form 29 May 2000; accepted 29 May 2000

Abstract

Nitric oxide reductases (NORs) that are found in bacteria belong to the large enzyme family which includes cytochrome
oxidases. Two types of bacterial NORs have been characterised. One is a cytochrome bc-type complex (cNOR) that receives
electrons from soluble redox protein donors, whereas the other type (qNOR) lacks the cytochrome c component and uses
quinol as the electron donor. The latter enzyme is present in several pathogens that are not denitrifiers. We summarise the
current knowledge on bacterial NORs, and discuss the evolutionary relationship between them and cytochrome oxidases in
this review. ß 2000 Elsevier Science B.V. All rights reserved.

Keywords: Denitri¢cation; Bacterium; Nitric oxide reductase; Evolution

1. Regeneration of the N-N bond as part of the
nitrogen cycle
Nitrogen is an essential element for all living or-
ganisms due to its occurrence in nucleic and amino
acids. It is highly abundant in the earth's atmosphere
as elementary dinitrogen gas (N2 ). Many species of
bacteria are capable of reducing nitrogen gas by a
process called dinitrogen ¢xation. The produced am-
monium ion (NH‡ 4 ) can be used by biological sys-
tems as a nitrogen donor for amino acid and nucleic
acid biosynthesis. Alternatively, ammonium can be
converted via nitrite (NO3 3
2 ) into nitrate (NO3 ) by
soil bacteria of the genera Nitrosomonas and Nitro-
bacter in a process termed nitri¢cation. Nitrate can
be taken up by higher plants, algae or fungi, which
* Corresponding author. Fax: +49 (6221) 387306;
E-mail: saraste@embl-heidelberg.de
1
Both authors contributed equally to this work.
convert it back into ammonium for synthesis of or-
ganic compounds in a process that is called assimi-
latory nitrate reduction. However, nitrate can also be
used as terminal electron acceptor by several micro-
organisms, as an alternative to oxygen. This process
is called nitrate ammoni¢cation when its product is
ammonium, or denitri¢cation when nitrate is sequen-
tially reduced to dinitrogen via nitrite, nitric oxide
(NO) and nitrous oxide (N2 O). In addition, a sepa-
rate pathway that oxidises ammonium back to dini-
trogen is called anaerobic ammonium oxidation ^ it
presumably consists of direct reduction of ammo-
nium and nitrite to dinitrogen gas and water. The
bacterium responsible for this process is a lithotro-
phic planctomycete [1].
The N-N bond formation in nature mainly occurs
by nitric oxide reduction during denitri¢cation. NO
is produced from nitrite (NO3 2 ) in a reductive reac-
tion catalysed by nitrite reductases (see [2]). Because
of its toxicity to cells, NO is reduced to nitrous oxide
(N2 O) immediately after it has been generated. Until

0005-2728 / 00 / $ ^ see front matter ß 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 0 0 5 - 2 7 2 8 ( 0 0 ) 0 0 1 6 1 - 4

BBABIO 44916 4-8-00 Cyaan Magenta Geel Zwart

 J. Hendriks et al. / Biochimica et Biophysica Acta 1459 (2000) 266^273
recently, denitri¢cation has been regarded as an ex-
clusively prokaryotic activity. However, the identi¢-
cation of denitri¢cation in several fungi [3,4] and
yeasts [5] has revised this thinking. In denitrifying
fungi, NO reduction is catalysed by a soluble enzyme
(P-450nor), whereas a membrane protein complex
catalyses this reaction in bacteria.
2. Fungal P-450nor
The reduction of NO to nitrous oxide in denitrify-
ing fungi is catalysed by a soluble, single-subunit
protein, called P-450nor. This enzyme is a member
of the cytochrome P-450 superfamily [4^7], but it
lacks the N-terminal membrane anchor that is
present in all other eukaryotic cytochrome P-450 en-
zymes.
Cytochrome P-450s are ubiquitous enzymes that
normally catalyse mono-oxygenation or hydroxyla-
tion of a range of hydrocarbon substrates. In this
reaction, one atom of dioxygen is inserted into the
substrate, whereas the other oxygen atom is reduced
to water using electrons that are taken from assisting
redox proteins, or from the substrate. In contrast to
other cytochrome P-450s, P-450nor does not have
mono-oxygenation activity. Instead, it catalyses the
reduction of NO according to the equation [6]:
2 NO ‡ NAD P†H ‡ H‡ ! N2 O ‡ H2 O ‡ NAD P†‡
P-450nor is also unique in that it takes the electrons
directly from NAD(P)H without intervention of re-
dox proteins, such as iron-sulphur or £avoproteins
[6].
The structure of P-450nor is similar to that of
other P-450 enzymes, consisting of a L-sheet domain,
an K-helical domain and a coil named `meander'. A
large opening to the distal side of the haem is present
in this structure [7]. The BP helix, which aligns this
opening, contains a NAD-like binding motif that is
highly conserved in P-450nor [8]. Studies on chimeric
proteins support the idea that NAD(P)H interacts
with P-450nor at this site [8].
P-450nor is not e¤ciently reduced by NAD(P)H in
the absence of NO [9]. Only after binding of NO to
the ferric enzyme, a rapid reduction by NAD(P)H
takes place [6]. The two-electron reduction of the
BBABIO 44916 4-8-00 Cyaan Magenta Geel Zwart
267
NO-bound ferric haem by NAD(P)H results in the
formation of an intermediate that is probably a fer-
rous nitroxyl species (Fe2‡ NO3 ) [10]. A second mol-
ecule of NO is thought to react with this intermedi-
ate, thereby causing the formation of N2 O, H2 O and
regeneration of ferric haem [11]. This reaction may
occur via the formation of bound hyponitrite
(3 ONNO3 ) [12], which is known to rapidly decom-
pose into N2 O and OH3 after reaction with a pro-
ton. Water could be generated by reaction of this
OH3 with a second proton.
Such a mechanistic model suggests that at least
one of the two protons needed in catalysis has to
be transferred into the active site. In the structure
of P-450nor in the ferrous CO-bound state [7] and
in the ferric NO-bound state [13], a hydrogen-bond-
ing network including residues S286 and D393 is
observed between the ligand on the haem and the
solvent. Both residues are speci¢c to P-450nor. Mu-
tagenesis studies have con¢rmed the importance of
this hydrogen-bonding network for catalysis [7,13].
3. Bacterial nitric oxide reductase
In bacteria, an integral membrane protein complex
is responsible for the reduction of NO to N2 O. The
bacterial nitric oxide reductase (NOR) has been orig-
inally isolated as a two-subunit complex [14]. After
sequencing of the ¢rst nor genes from Pseudomonas
stutzeri [15], it was noticed [16,17] that the large sub-
unit, NorB, displays the characteristic features of the
main catalytic subunit in the superfamily of haem/
copper cytochrome oxidases. These comprise the ex-
istence of 12 transmembrane helices that contain six
invariant histidines at de¢ned positions. These histi-
dines are involved in co-ordination of metal centres
and are all conserved in NorB in topologically cor-
rect positions.
The haem/copper cytochrome oxidases form a
large family with members in all three domains of
life. The best known is the mitochondrial cytochrome
c oxidase. Haem/copper oxidases normally catalyse
the reduction of molecular oxygen to water as part
of the respiratory chain. In all members of the super-
family, the catalytic subunit contains two redox-ac-
tive metal centres that are ligated by the conserved
histidines. These are a low-spin haem centre, and the
268 J. Hendriks et al. / Biochimica et Biophysica Acta 1459 (2000) 266^273
Fig. 1. Schematic overview of the predicted topology of the
multi-subunit cNOR and the single-subunit qNOR. The genes
encoding the polypeptides are represented by a bar underneath
the drawing. P and C indicate the periplasmic and cytosolic
sides of the membrane, respectively.
bimetallic active site that is made by a high-spin
haem and a copper ion, called CuB . However, in
the nitric oxide reductases the position of the copper
atom is occupied by a non-haem iron (see [18], and
references therein). A low-resolution structure, that
has been obtained through the analysis of two-di-
mensional crystals using electron microscopy, con-
¢rms the structural similarity between bacterial
NOR and cytochrome oxidases (Gohlke, Warne
and Saraste, unpublished results).
4. Two types of bacterial nitric oxide reductases
The bacterial nitric oxidase reductases have been
puri¢ed from Ps. stutzeri and other proteobacteria
[14,18^22] as the complex of NorB and NorC pro-
teins. NorC is a membrane-anchored c-type cyto-
BBABIO 44916 4-8-00 Cyaan Magenta Geel Zwart
chrome (NorC), and both haems of NorB are b-
type. For reasons that will become obvious, we shall
call the NorBC complex cNOR. In the bacterial
chromosome, the norC gene is always located directly
upstream from the norB gene and both are part of a
large cluster (operon) that contains additional nor
genes. Mutational analysis has also suggested that
the active enzyme may contain additional subunits
in situ [23]. The enzyme catalyses the reduction of
two molecules of NO to N2 O and water, using elec-
trons derived from soluble electron donors such as
c-type cytochromes (hence the name, cNOR), and
possibly cupredoxins such as pseudoazurin (see
[24,25] for reviews).
In the denitrifying Ralstonia eutropha, there is no
norC gene upstream from norB [26]. Instead, the
norB gene encodes a protein that has an N-terminal
extension as compared to the homologous subunit in
the NorBC complex. This extension includes two ad-
ditional transmembrane spans and a (putative) peri-
plasmic domain that is sandwiched between them
(Fig. 1). The Ralstonia NOR has been isolated as a
single subunit enzyme [27]. The activity measured
using NADH and diaphorase is virtually absent in
the presence of cytochrome c, but it is stimulated by
menadione. Therefore, it appears to be a quinol-ox-
idising enzyme [27].
Thus, we can distinguish two types of bacterial
nitric oxide reductases. One is a cytochrome bc com-
plex (cNOR) that can use a c-type cytochrome as an
electron donor, whereas the other lacks a cytochrome
c component and accepts electrons from quinols. We
call the latter qNOR.
Until now, qNOR genes have been identi¢ed in the
genomes of eight eubacterial species. A phylogenetic
analysis of the currently available qNOR and NorB
sequences is shown in Fig. 2. As the outgroup, we
have used the sequences of the major subunit from
the FixN-type cytochrome oxidases (so-called cyto-
chrome cbb3 ), because this enzyme is a close relative
of NOR in the haem/copper oxidase family [17]. The
cNOR and qNOR sequences separate into distinct
groups. In both groups, the relationships are in
good agreement with the known phylogeny. The
main exception is the paraphyly of qNOR in Firmi-
cutes, but this is likely due to the low resolution of
the tree in the short internode branches. The deep
and well-supported branching of the two groups in-
 J. Hendriks et al. / Biochimica et Biophysica Acta 1459 (2000) 266^273 269

Fig. 2. Neighbour-joining tree with Dayho¡ distances of an alignment of NorB, the homologous part of qNOR and FixN sequences
calculated with PHYLIP [41]. The alignment was made using ClustalW [42] and 230 positions suitable for phylogenetic analysis were
selected by Gblocks [43]. Bootstrap values are from 100 replications (only values above 50% are shown). The scale bar represents a
distance of 0.2 substitutions per site. The sequences have been obtained from the SPTREMBL or SWISSPROT databases (see acces-
sion numbers in parentheses after the species name) and the genome sequencing projects that are in progress at the Institute of Ge-
nome Research (TIGR), the University of Oklahoma and the Sanger Centre.

dicates that the two enzyme subfamilies are the prod-
uct of an early gene duplication at the base of the
eubacterial tree. This is di¡erent from the situation in
the case of cytochrome oxidase in which the gene
duplication that gave rise to the eubacterial quinol
oxidase, occurred within the Firmicutes and was then
followed by a gene transfer to some Proteobacteria
[28]. One of the two copies of the qNOR genes of
Ra. eutropha is in the chromosome, whereas the oth-
er is present in a megaplasmid. They form a single
subgroup, which indicates that they have been gen-
erated by a recent gene duplication followed by an
internal gene transfer between the chromosome and
the megaplasmid.
It has been previously noted that outside the haem
C-binding motif, NorC has no signi¢cant similarity
to other soluble or membrane-anchored c-type cyto-
chromes (see [25]). However, we have now noticed
that the N-terminal extension of qNOR shows se-
quence similarity to NorC (Fig. 3). The similarity is
low but detectable by a Smith-Waterman search in
which the N-terminal extension in qNOR (until the
end of the second transmembrane span, see Fig. 1) is
compared against the non-redundant sequence data-

BBABIO 44916 4-8-00 Cyaan Magenta Geel Zwart

270 J. Hendriks et al. / Biochimica et Biophysica Acta 1459 (2000) 266^273

BBABIO 44916 4-8-00 Cyaan Magenta Geel Zwart

 J. Hendriks et al. / Biochimica et Biophysica Acta 1459 (2000) 266^273 271

base. This homology suggests that the evolutionary
relationship between qNOR and cNOR is not re-
stricted to the NorB part of the enzyme but also
involves the NorC portion of the enzyme. The pres-
ence of the transmembrane helix between the peri-
plasmic domain and the NorB-homologous part per-
mits the overall topology of qNOR to be similar to
that of cNOR (Fig. 1).
It has been suggested that the ancestor of the
haem/copper cytochrome oxidase family has been a
NO-reducing enzyme [29]. It is obviously not clear
whether such an ancestor would have been similar to
qNOR or cNOR. It has been argued that the en-
zymes that use quinols as a substrate, such as the
cytochrome bc1 complex, are likely to have evolved
very early (see [30]). However, the homology between
the haem-binding domain in NorC and the N-termi-
nal extension in qNOR would suggest a scenario in
which the latter must have evolved from the former.
The easiest explanation for the structural similarity
between a haem-binding domain and an `empty' do-
main is that the haem has been lost during evolution.
Such a `loss-of-function' argument is similar to the
one that we have presented before for the relation-
ship between the CuA -binding domain in subunit II
of cytochrome c oxidase and the `empty' domain in
the homologous subunit of the quinol oxidase
[31,32].
5. Active sites are conserved in cNOR and qNOR
In cNOR, the electron entry into the enzyme oc-
curs via haem c in NorC. The homologous N-termi-
nal extension in qNOR is thought to play a role in
quinol binding, and in its oxidation. Deletion studies
have shown that this part of the enzyme is essential
(see [27]). Until now, the Ralstonia qNOR is the only
quinol-oxidising NOR that has been biochemically
characterised. The data indicate one non-haem iron
to be present in this enzyme, and only low amounts
of copper can been detected (0.2 mol Cu/mol en-
zyme). Thus, the enzyme is thought to contain an
active site formed by a high-spin haem and a non-
haem iron similarly to cNOR. In the visible spectrum
of qNOR, only haem b is detected [27]. The absence
of c-type haem is in line with the absence of a haem
c-binding motif in the sequence of qNOR. Addition-
ally, a weak signal absorbing at 580 nm in the visible
spectrum decreases upon reduction, similarly to the
behaviour of the 595 nm band in the corresponding
spectrum of cNOR. The 604 nm band in the latter
has been attributed to a mixed-valence binuclear
centre, in which the non-haem iron is reduced and
the high-spin haem is oxidised [33]. Further reduc-
tion of qNOR results in an absorbance band at 599
nm [27]. This signal may be related to the 604 nm
band in the spectrum of cNOR upon reduction of
the active site [33].
The EPR spectrum of qNOR contains a set of
signals that are similar to those arising from the
low-spin b haems in cNOR [18,22]. There are also
small signals at g = 6 and g = 2.01. The former indi-
cates the presence of a high-spin haem and the latter
is assigned to either a non-metallic radical or a non-
haem iron [27]. The spectroscopic features of qNOR
compare to the results obtained with cNOR, and
although a proper experimental comparison needs
yet to be carried out, the cross-spectral features sup-
port cross-similarity of the haem/Fe active sites in the
two enzymes.
The mechanism of NO reduction catalysed by the
bacterial reductases is not known at present. The
similarity of the catalytic sites of cNOR and qNOR
suggests that the mechanism is very similar in these
enzymes. The cytochrome oxidases of the haem/cop-
per family cannot e¤ciently reduce NO under anaer-
obic conditions [34,35] although some turn it over
very slowly [36]. The replacement of copper in the
active site of the oxidases with iron in NOR is
thought to be an important factor for the e¤cient
reduction of NO. Since the mechanism in P-450nor
involves only a haem, the bacterial reductases are
expected to reduce NO via a di¡erent mechanism

6
Fig. 3. Alignment of the predicted protein sequences of NorC and the N-terminal extension of qNOR (helix-domain-helix). Conserved
residues are highlighted on a grey background. The haem c binding motif in NorC is indicated by triangles above the sequence. The
alignment has been made using ClustalW [42]. The transmembrane helices predicted for the sequence of Neisseria meningitidis and
Paracoccus denitri¢cans by the program PHD [44] are represented as black and grey bars respectively underneath the sequence. The
¢gure has been prepared using Alscript [45].

BBABIO 44916 4-8-00 Cyaan Magenta Geel Zwart

272 J. Hendriks et al. / Biochimica et Biophysica Acta 1459 (2000) 266^273
involving the non-haem iron. Two di¡erent mecha-
nisms have been suggested. One involves the binding
of both substrate molecules to the non-haem iron in
the active site [37], whereas the other comprises the
binding of one NO molecule to each metal in the
binuclear site [22,38]. The involvement of nitroxyls
(NO3 ) in the reduction mechanism has been inferred
from trapping studies using dithiothreitol [39], which
is compatible with either mechanism. The observa-
tion that nitrosyl complexes form with both the
haem and non-haem iron during turnover [18] does
not discriminate between the two models either. In
the former model, the haem-nitrosyl species can be
interpreted as an inhibited state of the enzyme that is
induced by the high concentration of NO after mix-
ing with the sample [22].
6. Physiological roles of cNOR and qNOR
All bacteria that contain a cNOR are capable of
denitri¢cation. Upstream from norC, a conserved se-
quence motif for an FNR-type regulator (TTGAT-
N4 -ATCAA) is usually present in all loci that encode
cNOR. A gene encoding such a regulator has been
identi¢ed near the nor locus in several cNOR-con-
taining species. The same regulator also in£uences
the expression of the NO-producing nitrite reductase,
and is, therefore, called NNR (nitrite and nitric oxide
regulator). This regulator guarantees that the enzy-
matic production and reduction of NO are coupled
at the gene expression level, and this regulation is
thought to be important for preventing the accumu-
lation of toxic NO during denitri¢cation.
Ra. eutropha is a denitrifying organism. In con-
trast, the other organisms that contain a qNOR-
type enzyme (Fig. 2) are classi¢ed as non-denitrifying
[40], and many are intracellular pathogens that in-
vade mammalian cells. However, the Neisseria spe-
cies, Corynebacterium diphtheriae and Bacillus stear-
othermophilus all contain in their genome a gene that
encodes a protein with homology to NirK, the cop-
per-containing dissimilatory nitrite reductase. In the
complete Synechocystis genome, a homologous gene
is missing and there is no gene encoding a cyto-
chrome cd1 -type nitrite reductase either. In Ra. eu-
tropha, the upstream region of the promoter that
controls the expression of qNOR (see [26]) does
BBABIO 44916 4-8-00 Cyaan Magenta Geel Zwart
not contain an FNR box. Thus, nor expression ap-
parently is not coupled to nir expression in Ra. eu-
tropha.
When qNOR is expressed in a background that
lacks the other functions of denitri¢cation, its phys-
iological activity may relate to the detoxi¢cation of
NO that is not produced by the bacterium itself, but
is present in its environment. In the mammalian
pathogens, the host's macrophages are a likely
source for this NO. qNOR expressed by the patho-
gen provides protection against the host defence
mechanism. Additionally, in combination with a ni-
trite and a nitrate reductase, the ability to denitrify
might aid the pathogen's growth, allowing survival
under anaerobic conditions when nitrate is avail-
able.
7. Conclusion
The N-N bond forming NO reductases in fungi as
well as in bacteria are members of two distinct en-
zyme families that normally use oxygen as substrate.
The fungal P-450nor belongs to the cytochrome
P-450 family, whereas the bacterial NORs are related
to the haem/copper cytochrome oxidases. The bacte-
rial NORs fall into two subclasses. These use di¡er-
ent electron donors, i.e. either quinols (qNOR) or
soluble redox proteins such as cytochrome c
(cNOR). The presence of cNOR is characteristic of
the denitrifying activity. In many (pathogenic) organ-
isms, qNOR may be responsible for detoxi¢cation of
NO that is produced in their environment.
The existence of cNORs and qNORs parallels the
existence of cytochrome c- and quinol-oxidising cy-
tochrome oxidases. Thus, the occurrence of catalysts
that swap the electron donor during evolution may
be a characteristic of membrane-bound enzymes that
have access to di¡erent pools of redox compounds.
This parallel evolutionary process has apparently
taken place within the NO-reducing as well as the
oxygen-reducing branches of the large haem/copper
cytochrome oxidase family.
Acknowledgements
This work has been supported by the EC grant
 J. Hendriks et al. / Biochimica et Biophysica Acta 1459 (2000) 266^273
BIO 4-CT98-0507 (SENORA Project). We are grate-
ful to Dr. Rainer Cramm for discussion.
References
[1] M. Strous, J.A. Fuerst, E.H.M. Kramer, S. Logemann, G.
Muyzer, K.T. van de Pas-Schoonen, R. Webb, J.G. Kuenen,
M.S.M. Jetten, Nature 400 (1999) 446^449.
[2] F. Cutruzzola©, Biochim. Biophys. Acta 1411 (1999) 231^249.
[3] H. Shoun, T. Tanimoto, J. Biol. Chem. 266 (1991) 11078^
11082.
[4] H. Shoun, D.H. Kim, H. Uchiyama, J. Sugiyama, FEMS
Microbiol. Lett. 94 (1992) 277^282.
[5] S. Tsuruta, N. Takaya, L. Zhang, H. Shoun, K. Kimura, M.
Hamamoto, T. Nakase, FEMS Microbiol. Lett. 168 (1998)
105^110.
[6] K. Nakahara, T. Tanimoto, K. Hatano, K. Usuda, H.
Shoun, J. Biol. Chem. 268 (1993) 8350^8355.
[7] S.-Y. Park, H. Shimizu, S. Adachi, A. Nakagawa, I. Tanaka,
K. Nakahara, H. Shoun, E. Obayashi, H. Nakamura, T.
Iizuka, Y. Shiro, Nat. Struct. Biol. 4 (1997) 827^832.
[8] T. Kudo, D. Tomaru, D.L. Liu, X.Q. Dai, H. Shoun, Bio-
chimie 78 (1996) 792^799.
[9] Y. Shiro, M. Fujii, Y. Isogai, S.-I. Adachi, T. Iizuki, Bio-
chemistry 34 (1995) 9052^9058.
[10] E. Obayashi, K. Tsukamoto, S.-I. Adachi, S. Takahashi, M.
Nomura, T. Iizuka, H. Shoun, Y. Shiro, J. Am. Chem. Soc.
119 (1997) 7807^7816.
[11] Y. Shiro, M. Fujii, T. Iizuka, S. Adachi, K. Tsukamoto, K.
Nakahara, H. Shoun, J. Biol. Chem. 270 (1995) 1617^1623.
[12] N. Toritsuka, H. Shoun, U.P. Singh, S.-Y. Park, T. Iizuka,
Y. Shiro, Biochim. Biophys. Acta 1338 (1997) 93^99.
[13] H. Shimizu, E. Obayashi, Y. Gomi, H. Arakawa, S.-Y. Park,
H. Nakamura, S. Adachi, H. Shoun, Y. Shiro, J. Biol.
Chem. 275 (2000) 4816^4826.
[14] B. Heiss, K. Frunzke, W.G. Zumft, J. Bacteriol. 171 (1989)
3288^3297.
[15] W.G. Zumft, C. Braun, H. Cuypers, Eur. J. Biochem. 219
(1994) 481^490.
[16] J. Van der Oost, A.P.N. de Boer, J.-W. de Gier, W.G.
Zumft, A.H. Stouthamer, R.J.M. van Spanning, FEMS Mi-
crobiol. Lett. 121 (1994) 1^10.
[17] M. Saraste, J. Castresana, FEBS Lett. 341 (1994) 1^4.
[18] J. Hendriks, A. Warne, U. gohlke, T. Haltia, C. Ludovici,
M. Lu«bben, M. Saraste, Biochemistry 37 (1998) 13102^
13109.
[19] G.J. Carr, S.J. Ferguson, Biochim. Biophys. Acta 1017
(1990) 57^62.
[20] M. Dermastia, T. Turk, T.C. Hollocher, J. Biol. Chem. 266
(1991) 10899^10905.
[21] T. Fujiwara, Y. Fukumori, J. Bacteriol. 178 (1996) 1866^
1871.
BBABIO 44916 4-8-00 Cyaan Magenta Geel Zwart
273
[22] P. Girsch, S. de Vries, Biochim. Biophys. Acta 1318 (1997)
202^216.
[23] A.P.N. De Boer, J. van der Oost, W.N.M. Reijnders, H.V.
Westerho¡, A.H. Stouthamer, J.M. van Spanning, Eur. J.
Biochem. 242 (1996) 592^600.
[24] B.C. Berks, S.J. Ferguson, J.W. Moir, D.J. Richardson, Bio-
chim. Biophys. Acta 1232 (1995) 97^173.
[25] J. Hendriks, U. Gohlke, M. Saraste, J. Bioenerg. Biomembr.
30 (1998) 15^24.
[26] R. Cramm, R.A. Siddiqui, B. Friedrich, J. Bacteriol. 179
(1997) 6769^6777.
[27] R. Cramm, A. Pohlmann, B. Friedrich, FEBS Lett. 460
(1999) 6^10.
[28] J. Castresana, M. Lu«bben, M. Saraste, D.G. Higgins,
EMBO J. 13 (1994) 2516^2525.
[29] J. Castresana, M. Saraste, Trends Biochem. Sci. 20 (1995)
443^448.
[30] S.M. Musser, S.I. Chan, J. Mol. Evol. 46 (1998) 508^520.
[31] J. Van der Oost, P. Lappalainen, A. Musacchio, T. Warne,
L. Lemieux, J. Rumbley, R.B. Gennis, R. Aasa, T. Pascher,
B.G. Malmstro«m, M. Saraste, EMBO J. 11 (1992) 3209^
3217.
[32] M. Wilmanns, P. Lappalainen, M. Kelly, E. Sauer-Eriksson,
M. Saraste, Proc. Natl. Acad. Sci. USA 92 (1995) 11955^
11959.
[33] K.L.C. Gro«nberg, M.D. Roldän, L. Prior, G. Butland, M.R.
Cheesman, D.J. Richardson, S. Spiro, A.J. Thomson, N.J.
Watmough, Biochemistry 38 (1999) 13780^13786.
[34] G.W. Brudvig, T.H. Stevens, S.I. Chan, Biochemistry 19
(1980) 5275^5285.
[35] G. Stubauer, A. Giu¡re©, M. Brunori, P. Sarti, Biochem.
Biophys. Res. Commun. 245 (1998) 459^465.
[36] A. Giu¡re©, G. Stubauer, P. Sarti, M. Brunori, W.G. Zumft,
G. Buse, T. Soulimane, Proc. Natl. Acad. Sci. USA 96
(1999) 14718^14723.
[37] C.S. Butler, H.E. Seward, C. Greenwood, A.J. Thomson,
Biochemistry 36 (1997) 16259^16266.
[38] P. Moe«nne-Loccoz, S. de Vries, J. Am. Chem. Soc. 120
(1998) 5147^5152.
[39] T. Turk, T.C. Hollocher, Biochem. Biophys. Res. Commun.
183 (1992) 983^989.
[40] N.E. Krieg, Bergey's Manual of Systematic Bacteriology,
Williams and Wilkins, Baltimore, MD, 1984.
[41] J. Felsenstein, PHYLIP (phylogeny inference package). Ver-
sion 3.5c, Distributed by the author, Department of Genet-
ics, University of Washington, Seattle, WA, 1993.
[42] J.D. Thompson, D.G. Higgins, T.J. Gibson, Nucleic Acids
Res. 22 (1994) 4673^4680.
[43] J. Castresana, Mol. Biol. Evol. 17 (2000) 540^552.
[44] B. Rost, R. Casadio, P. Fariselli, C. Sander, Protein Sci. 4
(1995) 521^533.
[45] G.J. Barton, Protein Eng. 6 (1993) 37^40.