See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/12887217

TorC apcytochrome negatively autoregulates the trimethylamine N-oxide

(TMAO) reductase operon in Escherichia coli

Article  in  Molecular Microbiology · August 1999

DOI: 10.1046/j.1365-2958.1999.01468.x · Source: PubMed

CITATIONS READS

28 157

4 authors, including:

Mireille Ansaldi
French National Centre for Scientific Research
96 PUBLICATIONS   1,070 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Lysogeny maintenance View project

Phage-based biosensors View project

All content following this page was uploaded by Mireille Ansaldi on 25 October 2017.

The user has requested enhancement of the downloaded file.

Molecular Microbiology (1999) 33(2), 284±295
TorC apocytochrome negatively autoregulates the
trimethylamine N-oxide (TMAO) reductase operon
in Escherichia coli
Mireille Ansaldi, Christophe Bordi, MicheÁle
Lepelletier and Vincent MeÂjean*
Laboratoire de Chimie BacteÂrienne, Institut de Biologie
Structurale et Microbiologie, Centre National de la
Recherche Scienti®que, 31, chemin Joseph Aiguier,
BP 71, 13402 Marseille Cedex 20, France.
Summary
The trimethylamine N-oxide (TMAO) anaerobic
respiratory system of Escherichia coli comprises a
periplasmic terminal TMAO reductase (TorA) and a pen-
tahaem c-type cytochrome (TorC), which is involved
in electron transfer to TorA. The structural proteins
are encoded by the torCAD operon whose expression
is induced in the presence of TMAO through the TorS/
TorR two-component system. By using a genomic
library cloned into a multicopy plasmid, we identi®ed
TorC as a possible negative regulator of the tor operon.
Interestingly, in trans overexpression of torC not only
decreased the activity of a torA8± 8lacZ fusion, but also
dramatically reduced the amount of mature TorC cyto-
chrome. This led us to propose that, after trans-
location, TorC apocytochrome downregulates the tor
operon unless it is properly matured. In agreement
with this hypothesis, we have shown that mini-Tn10
insertions within genes involved in the c-type cyto-
chrome maturation pathway or haem biosynthesis
decreased tor operon expression. Dithiothreitol (DTT),
which reduces disulphide bonds and thus prevents
the ®rst step in c-type cytochrome formation, also
strongly decreases the tor promoter activity. The DTT
effect is TorC dependent, as it is abolished when
torC is disrupted. In contrast, overexpression of the
c-type cytochrome maturation (ccm ) genes relieved
the tor operon of the negative control and allowed
the bacteria to produce a higher amount of TorC holo-
cytochrome. Therefore, the TorC negative autoregula-
tion probably means that maturation of the c-type
cytochrome is a limiting step for Tor system biogen-
esis. Genetic experiments have provided evidence
that TorC control is mediated by the TorS/TorR
Received 4 March, 1999; revised 14 April, 1999; accepted 19 April,
1999. *For correspondence. E-mail mejean@ibsm.cnrs-mrs.fr; Tel.
(‡33) 4 91 16 40 32; Fax (‡33) 4 91 71 89 14.
Q 1999 Blackwell Science Ltd
two-component system and different from the tor
anaerobic control. In our working model, TMAO and
apoTorC bind to the periplasmic side of TorS, but
TMAO activates TorS autophosphorylation, whereas
apoTorC inhibits the TorS kinase activity.
Introduction
Several bacteria, including Escherichia coli, are classi-
®ed as facultatively anaerobic bacteria, because they
can use alternative electron acceptors when oxygen is
depleted. These electron acceptors are usually low-mole-
cular-weight oxygen-containing molecules abundant in the
environment. For example, E. coli can produce energy
by reducing nitrate, nitrite, dimethyl sulphoxide (DMSO)
or trimethylamine N-oxide (TMAO; Gennis and Stewart,
1996). TMAO is widespread in the marine environment
and acts as a powerful osmoprotector for the tissues of
marine ®sh (Wang and Bolen, 1997). Obviously, not only
enterobacteria but also marine bacteria (Photobacterium,
Shewanella and Vibrio species) and photosynthetic bac-
teria found in ponds (Rhodobacter species) are capable
of TMAO respiration (Barrett and Kwan, 1985). Although
these strains are quite distant from each other, it is note-
worthy that the TMAO reductase systems are clearly
related (Dos Santos et al., 1998). First, the terminal
enzymes are homologous molybdenum cofactor-containing
enzymes located in the periplasm (Czjzek et al., 1998). The
TMAO reductase of E. coli and Shewanella massilia has
been called TorA and that of Rhodobacter has been
named DorA because the former is highly TMAO speci®c,
whereas the latter can also reduce DMSO ef®ciently
(Iobbi-Nivol et al., 1996; Dos Santos et al., 1998). Sec-
ondly, the electron transfer to the terminal enzyme involves
a conserved pentahaemic c-type cytochrome (TorC or
DorC) anchored to the inner membrane. Thirdly, the
TMAO reductase respiratory genes are always organized
in an operon of at least three genes (Ujiiye et al., 1996;
Mouncey et al., 1997).
In E. coli, the TMAO reductase system is encoded by
the torCAD operon (MeÂjean et al., 1994). It has been
shown recently that the torD gene encodes a TorA private
chaperone (Pommier et al., 1998). The TMAO reductase
systems of Shewanella and Rhodobacter species also
contain a TorD-like protein whose role is probably similar
 TorC autoregulation 285
to that of the E. coli TorD protein. Interestingly, TorD is
located in the cytoplasm of E. coli. This emphasizes the
fact that TorA is translocated to the periplasm only if it is
previously folded with the molybdenum cofactor attached
to it (Santini et al., 1998). Extensive studies have recently
established that TorA belongs to a large family of metallo-
enzymes that are exported by a sec-independent pathway
after being folded (Weiner et al., 1998; Dalbey and Robin-
son, 1999). In E. coli, this pathway involves at least four
proteins that are encoded by the tat genes, also called
mtt genes (Chanal et al., 1998; Sargent et al., 1998).
Furthermore, the signal peptide of these metalloenzymes
is longer than classical signal peptides and contains a
highly conserved sequence R-R-X-F-L-K in its ®rst half.
Accordingly, these signal peptides have been called the
`twin-arginine signal peptides'.
In contrast, c-type cytochromes are metalloenzymes
translocated via the classical Sec pathway as unfolded
apoproteins (ThoÈny-Meyer and Kunzler, 1997). Indeed,
c-type cytochrome maturation is a complex process occur-
ring exclusively on the periplasmic side of the inner mem-
brane (Page and Ferguson, 1989; 1990; Metheringham
et al., 1995; for reviews ThoÈny-Meyer, 1997; Kranz et
al., 1998). After translocation of the apocytochrome, the
cysteines of the haem binding site C-X-X-C-H are rapidly
oxidized by the DsbA/DsbB system (Metheringham et al.,
1996). Although the cysteines must be reduced again
before covalent haem attachment, disulphide bond forma-
tion is an essential step during cytochrome c maturation
that probably converts the unfolded apocytochrome to a
prefolded form required for the following maturation
steps (Raina and Missiakas, 1997; ThoÈny-Meyer, 1997).
The ®nal steps of the c-type cytochrome formation involve Fig. 1. Multicopy effect of the torC gene.
the products of the ccmABCDEFGH genes and the DipZ A. Physical map of the tor locus and plasmid inserts of pUW13 and
pUW13D1.
(DsbD) protein (Crooke and Cole, 1995; ThoÈny-Meyer et
B. b-Galactosidase activities of torA8± 8lacZ fusion strain (LCB620)
al., 1995; Grove et al., 1996; Tanapongpipat et al., containing either pUC18, pUW13 or pUW13D1, grown anaerobically
1998). Haem, which is synthesized in the cytoplasm by with TMAO. Activities are expressed in Miller units.
C. Soluble fractions of strain MC4100, containing either no plasmid
the hem gene products is probably transported across
(10 mg), plasmid pUW13 (27 mg) or pUW13D1 (10 mg), were
the cytoplasmic membrane by the CcmABC exporter submitted to rocket immunoelectrophoresis on an agarose plate
complex (ThoÈny-Meyer, 1997; Goldman et al., 1998). into which 100 ml of TorA antiserum had previously been added.
The prosthetic group is then captured in the periplasm
by the CcmE protein (Schulz et al., 1998; Schulz et al., has shown that these regulatory genes encode positive
1998). After the cysteines of the cytochrome haem regulators essential for the torCAD operon expression.
binding site have been rereduced by the CcmG/DipZ TorS and TorR form a two-component regulatory system
thioredoxin, haem is transferred from the CcmE haemo- in which TorS is the transmembranous sensor that detects
protein to the apocytochrome (Schulz et al., 1998). This the presence of TMAO in the medium. Once activated,
leads to the mature form of the c-type cytochrome. TorS transphosphorylates its TorR partner, which, in turn,
TorC, which is one of the ®ve c-type cytochromes triggers the transcription of the tor operon by binding to
synthesized anaerobically by E. coli, follows this complex four decameric boxes located in the torCAD promoter
maturation pathway (Iobbi-Nivol et al., 1994; Tanapongpipat (Simon et al., 1995; Jourlin et al., 1997). Interestingly,
et al., 1998). TorS belongs to the restricted family of sensors that
In addition to the torCAD structural operon, the E. coli tor contains three sites of phosphorylation (Mizuno, 1997;
locus (Fig. 1A) is composed of the three regulatory genes Goudreau and Stock, 1998). As a result, the phosphoryla-
torR, torS and torT (Jourlin et al., 1996a). Genetic analysis tion of TorR requires a four-step phosphorelay (Jourlin et
Q 1999 Blackwell Science Ltd, Molecular Microbiology, 33, 284±295
286 M. Ansaldi, C. Bordi, M. Lepelletier and V. MeÂjean
al., 1997). The third regulator, TorT, is a periplasmic protein the torA and torS genes (Fig. 1A). As the tor operon pro-
similar to the periplasmic ribose-binding protein. Therefore, moter was present on pUW13, we thought at ®rst that
TorT probably binds a still unknown inducer and then inter- the observed negative effect was caused by titration of
acts with the periplasmic region of TorS to stimulate tor the TorR activator by the high-copy-number plasmid. To
induction (Jourlin et al., 1996b). test this hypothesis, we deleted a large DNA fragment of
In order to have a better understanding of the Tor sys- the downstream region of the insert. This deletion included
tem complex regulation, we used a genetic approach to most of the torC coding region but did not affect the tor
identify the proteins that downregulate the tor operon. Sur- operon promoter (Fig. 1A). When introduced into LCB620,
prisingly, we found that the TorC protein negatively con- the resulting plasmid (pUW13D1) did not decrease torA8±
trols its own synthesis. Additional results indicated that 8lacZ expression in inducing conditions (Fig. 1B). The
this control involves the TorC apoprotein and is mediated deleted plasmid, which still contained the tor promoter,
by the TorS/TorR two-component system. This negative thus behaved as the control plasmid pUC18. Moreover,
autoregulation prevents the maturation apparatus from the amount of TMAO reductase was similar in strain
overloading and, consequently, allows optimal biogenesis MC4100 either with or without pUW13D1 (Fig. 1C). These
of the Tor respiratory system. To our knowledge, this type results invalidate our previous proposal and suggest that
of regulation by the maturation state of a c-type cyto- the torC coding sequence is involved in the downregula-
chrome is unprecedented. tion of the torCAD operon.
To con®rm that TorC alone negatively autoregulates the
tor operon, we cloned its coding sequence into plasmid
Results pBAD24, downstream from the arabinose-dependent
PBAD promoter. This construct (pBC5H) allowed us to
The torC gene product negatively autoregulates the
torCAD operon
To identify proteins that negatively regulate the tor operon,
we looked for genes, which, when present on a multicopy
plasmid, affect the expression of a torA8± 8lacZ fusion.
An E. coli genomic library cloned into a pUC18 vector
was introduced by transformation into strain LCB620
(MC4100 but torA8± 8lacZ ). The transformed bacteria
were then plated onto MacConkey lactose-TMAO medium.
As TMAO induces the torA8± 8lacZ fusion, most colonies
were red, but some white ones appeared. To characterize
these clones more thoroughly, we measured their b-galac-
tosidase activities in the presence or in the absence of
TMAO. As shown in Fig. 1B, torA8± 8lacZ expression of
one of these clones is dramatically affected. In inducing
conditions, the b-galactosidase activity was decreased
about 10-fold when compared with the strain containing
pUC18. As a control, the plasmid (pUW13) was extracted
from this strain and then introduced back into strain
LCB620 and strain MC4100. b-Galactosidase measure-
ments con®rmed that the plasmid affects the expression
of the chromosomal fusion of LCB620. Moreover, as
revealed by rocket immunoelectrophoresis, the quantity of
TorA protein was greatly decreased in strain MC4100
when pUW13 was present (Fig. 1C). Therefore, we con-
cluded that plasmid pUW13 is responsible for the down- Fig. 2. Effect of torC overexpression.
regulation of the chromosomal tor operon. A. b-Galactosidase activities of strain LCB620 (torA8± 8lacZ )
The insert carried by plasmid pUW13 was polymerase carrying plasmid pBC5H grown anaerobically with TMAO.
Overnight induction with 0.05½ arabinose was performed when
chain reaction (PCR) ampli®ed by using appropriate vec- indicated.
tor-based primers, and the junctions between the vector B. Membranous fractions (10 mg), from cells grown as in (A), were
and the insert were determined by direct DNA sequencing. loaded on a SDS±10% polyacrylamide gel. After electrophoresis,
the gel was ®rst stained for haem with TMBZ (top); then, the same
The cloned genomic fragment is 4.7 kb long and contains gel was stained with Coomassie blue (bottom). The position of
the entire torC, torR and torT genes and the beginning of TorC and apoTorC is indicated by arrows.

Q 1999 Blackwell Science Ltd, Molecular Microbiology, 33, 284±295


Fig. 3. Effect of overproduction of TorC and Ccm proteins.
A. b-Galactosidase activities of strain LCB620 overproducing either
TorC alone, or TorC together with Ccm proteins. LCB620 carrying
pBC5H (open squares) and LCB620 carrying pBC5H and pEC86
(closed squares) were grown anaerobically with TMAO (10 mM).
Overnight induction was performed with arabinose at the indicated
concentrations.
B: Maturation of TorC in the presence of plasmid pEC86.
Membranous fractions (10 mg) of LCB620 containing plasmids
pEC86 and pBC5H grown as in (A) were loaded on a SDS±10%
polyacrylamide gel. After electrophoresis, the gel was stained for
haem with TMBZ. As high concentrations of arabinose lowered
growth rate, samples were normalized according to membranous
protein amounts.
regulate the in trans expression of torC tightly. Figure 2A
shows that expression of the torA8± 8lacZ fusion is almost
unaffected in the absence of arabinose (compare Fig. 2A
and Fig. 1B), whereas it is repressed when arabinose was
added to the medium. Moreover, increasing the arabinose
concentration led to a decrease in b-galactosidase activities
(Fig. 3A). Therefore, the synthesis of TorC clearly affects
the expression of the tor operon.
Modi®cation of the TorC maturation pattern affects
tor operon expression
When we looked for the presence of TorC holocytochrome
in membrane extracts, we noted that overexpression of torC
did not lead to increasing amounts of mature TorC protein
(Fig. 2B). Rather, the overall TorC maturation decreased
as the haem-staining band that corresponds to the TorC
c-type cytochrome faded away when TorC synthesis from
pBC5H was induced by arabinose. In contrast, the apo-
form of TorC could be detected by Coomassie blue stain-
ing in inducing conditions (Fig. 2B). These results not only
con®rm that E. coli has a limited capacity for maturation
Q 1999 Blackwell Science Ltd, Molecular Microbiology, 33, 284±295
TorC autoregulation 287
of c-type cytochromes (Arslan et al., 1998; Roldan et al.,
1998), but also reveal that overproduction of the TorC apo-
protein somehow jams the maturation process. As in trans
overproduction of TorC also represses the tor operon
expression, we propose that, if TorC could not be properly
matured, it then acts as a downregulator of the tor operon.
Accordingly, in our model, the TorC holoprotein plays no
role in the tor regulation, whereas the TorC apoprotein
decreases tor expression by an unknown mechanism.
The fact that almost no TorC holocytochrome is present
when torC is overexpressed, whereas expression of the
tor fusion is strongly repressed agrees with this model
(Fig. 2).
To con®rm the model, we decided to improve the matura-
tion of TorC in order to reduce the TorC apoprotein content
in the bacteria. This was achieved by introducing plasmid
pEC86, which allowed overexpression of the cytochrome c
maturation genes ccmABCDEFGH and, thus, increased c-
type cytochrome yields in E. coli (Arslan et al., 1998). The
presence of pEC86 led to a signi®cant increase in both
torA8± 8lacZ expression and TorC holocytochrome synth-
esis (Fig. 3). When torC was overexpressed in trans from
plasmid pBC5H by increasing the concentration of arabi-
nose, both tor operon expression and the amount of mature
TorC decreased. Interestingly, the decrease in b-galacto-
sidase activities parallels that observed in the strain
devoid of plasmid pEC86. However, the levels of b-galac-
tosidase activities are always about three times higher
when pEC86 is present, regardless of the arabinose con-
centration. A similar result is observed in strain LCB620
carrying pEC86 alone (924 Miller units compared with
304 in the absence of pEC86). These results strongly sug-
gest that increasing TorC maturation removes a tor-nega-
tive regulator. Alternatively, although improbable, one of
the ccm gene products might be a tor-positive regulator.
As overproduction of TorC from pBC5H again affects tor
operon expression and TorC maturation, the negative
regulator is likely to be the TorC apocytochrome.
Disruption of genes involved in c-type cytochrome
maturation or haem biosynthesis decreases tor
operon expression
From the model, one could predict that any defect in the
machinery for c-type cytochrome formation would lead
to a decrease in tor operon expression because TorC apo-
cytochrome, which could no longer be matured, would
become available to act as a negative autoregulator. To
answer this question, we isolated random mini-Tn10 inser-
tions that affect the expression of the chromosomal torA8±
8lacZ fusion. Thousands of mutated clones were plated onto
MacConkey lactose-TMAO, and white or pink colonies
were isolated and checked for their reduced b-galactosi-
dase activities in comparison with the control strain
288 M. Ansaldi, C. Bordi, M. Lepelletier and V. MeÂjean
LCB620. Nine mutants unable to express the fusion and
17 others with b-galactosidase activities two- to ®vefold
lower than that of LCB620 were analysed further. Mapping
of the insertions was done by inverse PCR with mini-Tn10-
based primers, followed by sequencing of one DNA
junction located between the transposon and the mutated
gene. The insertions that abolish tor operon expression
were located within the torS or torR regulatory genes
(seven and two insertions respectively). This was expected
as TorR and TorS form a two-component system essential
for the positive regulation of the tor operon (Jourlin et al.,
1997). Of greater interest, nine insertions among the 17
that lower tor operon expression were mapped within the
dsbA, dsbB, ccmC or hemN genes (®ve, one, one and
two insertions respectively). The precise location of these
insertions is summarized in Table 3, and the b-galactosi-
dase activities measured for each of the four mutant
classes in inducing conditions are shown in Fig. 4A.
An obvious feature shared by the four classes of mutants
is that they are all unable to produce TorC holocytochrome
(Fig. 4B) and, more generally, c-type cytochromes (ThoÈny-
Meyer, 1997). Indeed, DsbA is a periplasmic protein
required for the ef®cient formation of disulphide bonds in
translocated proteins and, in particular, c-type cytochromes
(Metheringham et al., 1995; Sambongi and Ferguson,
1996). DsbB, the DsbA partner, is located in the inner
membrane and maintains DsbA in the active oxidized
state (Missiakas et al., 1993; Raina and Missiakas, 1997).
CcmC is one of the subunits of an ABC transporter that is
essential for c-type cytochrome maturation (Tanapongpi-
pat et al., 1998). Finally, HemN is an enzyme involved
in haem biosynthesis (Troup et al., 1995). Interestingly,
both HemN and HemF catalyse the oxidative decarboxy-
lation of coproporphyrinogen III haem precursor, but
HemF is an oxygen-dependent oxidase, whereas HemN
is an oxygen-independent dehydrogenase. As E. coli
c-type cytochrome synthesis is induced solely in anaero-
bic growth conditions, HemN proves to be essential for
c-type cytochrome haem supply (Tyson et al., 1997).
Although the four classes of mutants are undoubtedly
affected in the tor operon expression, it is puzzling that
the tor expression levels of dsbA and dsbB mutants are
lower than that of ccmC or hemN mutants (Fig. 4A). Pos-
sible explanations will be proposed in the Discussion. In
any case, the fact that preventing TorC c-type cytochrome
maturation increases the tor negative regulation agrees
with our model and suggests strongly that neither TorC
holoprotein nor proteins involved in the c-type cytochrome
maturation pathway contribute to the tor downregulation.
As a last control, we stopped the cytochrome maturation
process arti®cially by adding a non-lethal concentration of
dithiothreitol (DTT; 5 mM) to the growth medium. Indeed,
DTT is a reducing agent that has been shown to reduce
in vivo disulphide bonds of periplasmic proteins (Raina
Fig. 4. Tn10 insertions that affect TorC maturation also affect tor
expression.
A. b-Galactosidase activities of strains LCB620 (wild type) with or
without DTT (5 mM), LCB3070 (dsbA), LCB3094 (dsbB ), LCB3001
(ccmC ) and LCB3201 (hemN ), grown anaerobically with TMAO.
B. Membranous fractions (10 mg) of the same strains were loaded
on a SDS±10% polyacrylamide gel. After electrophoresis, the gel
was stained for haem with TMBZ.
and Missiakas, 1997). Therefore, DTT antagonizes DsbA/
DsbB action and prevents c-type cytochrome formation.
As expected no TorC haemoprotein was detected after
haem staining when strain LCB620 was grown in the pre-
sence of DTT (Fig. 4B). Moreover, the b-galactosidase
activities of the fusion dropped at least fourfold when
DTT was added and became close to that measured
under normal inducing conditions for the dsbA or dsbB
strains (Fig. 4A). In addition, the effect of DTT in a ccmC
or hemN mutant strain was not greater than that observed
previously in strain LCB620 (data not shown). We con-
cluded that the DTT effect on the tor regulation supports
the role of TorC apocytochrome as a downregulator.
Effect of torC gene disruptions on tor regulation
To analyse the effect of torC gene disruption on tor pro-
moter activity, we constructed a hybrid plasmid in which
the torC promoter region was fused to the lacZ coding
sequence of the operon fusion vector pGE593. In this plas-
mid (pPTor43) the tor DNA fragment is devoid of any TorC
coding sequence and corresponds to the minimal tor pro-
moter from the torC transcription start site (‡1) to position
86 (Simon et al., 1995). When pPTor43 was introduced
into strain MC4100, the b-galactosidase activity decreased
three- to fourfold upon DTT addition in anaerobiosis (Table
1). We concluded from this experiment that the negative
regulation can be monitored from the plasmid construct.
As we only had a mutant (LCB628) in which the V-inter-
poson was inserted in the middle of the torC gene, we
decided to construct a true null torC mutant (LCB502) by
inserting the V-interposon at the very beginning of the
Q 1999 Blackwell Science Ltd, Molecular Microbiology, 33, 284±295

Table 1. Effect of DTT and TorC on tor promoter activity.
b-Galactosidase activitya
O2 ‡O2
Strain DTT ‡DTT DTT
MC4100/pPTor43 599 173 96
LCB502/pPTor43 784 740 111
LCB628/pPTor43 685 718 115
a. Activities are expressed in Miller units.
Values were obtained from cells grown anaerobically ( O2 ) or aero-
bically (‡O2 ) in LB medium containing ampicillin and TMAO in the
presence (‡DTT) or in the absence ( DTT) of DTT (5 mM).
chromosomal torC coding sequence. Plasmid pPTor43
was then introduced into the two torC mutant strains. As
shown in Table 1, activities from the plasmid-encoded tor
promoter are quite similar in both torC strains. Compared
with strain MC4100, b-galactosidase activities were slightly
higher in the absence of DTT and, much more important,
DTT no longer caused a negative effect in the torC strains.
To con®rm that the DTT effect was entirely TorC depen-
dent, we also veri®ed that this negative effect was not
modi®ed by a torA gene insertion (data not shown).
These results con®rm that the tor-negative regulation
revealed by DTT addition depends on the torC gene pro-
duct. Moreover, this negative effect seems to require the
entire torC gene, as even V insertion in the middle of
torC in strain LCB628 abolishes such an effect. As synth-
esis and maturation of a stable TorC8 c-type cytochrome
can occur in strain LCB628 (MeÂjean et al., 1994), we pro-
pose that the carboxy-terminal half of TorC is somehow
involved in the tor negative control.
TorC autoregulation is different from tor anaerobic
control
The torCAD operon is fully induced in anaerobiosis but, in
contrast to all other known anaerobic respiratory systems,
this control is mediated neither by FNR, a pleiotropic anae-
robic regulator, nor by ArcA, a response regulator involved
in redox control of numerous genes (Simon et al., 1994).
To test whether TorC plays any role in the tor anaerobic
regulation, we monitored the tor promoter expression
from pPTor43 in wild-type and torC strains grown in the
presence or absence of oxygen. Although the levels of
b-galactosidase activities were always slightly higher for
the torC strains, as mentioned previously, full induction
of the tor promoter was observed only in anaerobic condi-
tions for all three strains (Table 1). Moreover, the ratio
between activities measured in anaerobic and aerobic
growth conditions is almost the same (about 6) for the
wild-type and the torC strains. This indicates clearly that
the tor anaerobic control occurs normally even in the
Q 1999 Blackwell Science Ltd, Molecular Microbiology, 33, 284±295
TorC autoregulation 289
absence of a functional torC gene. In aerobiosis, the addi-
tion of DTT lowered the activity of the tor promoter in strain
MC4100, whereas tor transcription is unaffected in torC
strains. Consequently, the tor regulation mediated by
TorC is distinct from the tor anaerobic control whose reg-
‡DTT
ulators still remain to be discovered.
64
123
111 The TorC signalling pathway involves the TorS/TorR
two-component system
TorC is anchored to the inner membrane by an N-terminal
hydrophobic segment, but its main body is located in the
periplasm where c-type cytochrome maturation occurs
(MeÂjean et al., 1994; ThoÈny-Meyer, 1997). By substituting
the TorC N-terminal hydrophobic sequence with a classi-
cal signal peptide, we have shown that the periplasmic
part of TorC alone can still downregulate the tor operon
(data not shown). This result rules out a possible involve-
ment of the TorC hydrophobic segment in the autoregu-
lation process and also reinforces the idea that TorC
apocytochrome is ®rst translocated to the periplasm
where it downregulates its own synthesis unless it is prop-
erly matured. Obviously, owing to its location, TorC could
not be a repressor acting directly on the tor transcription,
for example by binding to the tor promoter. Therefore,
additional regulatory proteins are necessary for this signal
transduction pathway, and these proteins must receive the
TorC signal from the periplasm and transduce it to the
cytoplasm. The TorS/TorR two-component system, which
was initially shown to induce the tor operon in response
to TMAO availability (Jourlin et al., 1996a), is a good
candidate to belong to the TorC regulatory pathway.
Our approach to test this hypothesis was based on
epistatic experiments. We used either a `gain-of-function'
TorS allele (TorS726), which renders expression of
the tor operon constitutive, or a plasmid (pGS1) that
allows TorR overproduction and thus constitutive tor
transcription (Simon et al., 1994; Jourlin et al., 1996a). If
the TorC negative control requires TorS and TorR,
then the tor constitutive phenotypes should be unaffected
by the TorC control, as TorC must act upstream from TorS
and TorR. In contrast, if the TorC negative pathway
involves proteins different from TorS and TorR, no bypass
would be expected and, therefore, the TorC-dependent
repression should still occur.
The results of this experiment are summarized in
Table 2. As published previously, the torA8± 8lacZ fusion
is expressed at a high level in strain LCB620 containing
pGS1 or in strain LCB726 (LCB620 but torS726 ) in the
presence as well as in the absence of TMAO (Simon et
al., 1994; Jourlin et al., 1996a). More importantly, the addi-
tion of DTT to the growth medium, in order to enhance
TorC negative regulation, only slightly decreases the b-
galactosidase activities of the two latter strains, whereas
290 M. Ansaldi, C. Bordi, M. Lepelletier and V. MeÂjean
Table 2. TorS/TorR bypass experiments.
b-Galactosidase activitya
TMAO ‡TMAO
Strain DTT ‡DTT DTT ‡DTT
LCB620 <5 <5 300 72
LCB620/pGS1 784 654 808 770
LCB726 821 768 947 800
a. Activities are expressed in Miller units.
Values were obtained from cells grown anaerobically in LB medium
containing TMAO (10 mM) and DTT (5 mM) as indicated.
the activity of strain LCB620 drops about fourfold. This
clearly shows that overproduction of TorR or the presence
of a TorS constitutive mutation relieves the strain from the
TorC negative control. Strikingly, the level of constitutive
tor expression is close to that reached by strain LCB620
carrying plasmid pEC86, which allows overproduction of
mature TorC protein (compare Table 2 and Fig. 3A).
This is consistent with the idea that, in terms of tor regu-
lation, reducing the concentration of TorC apocytochrome
is equivalent to the genetic bypass of the TorC negative
control. Together, these results provide genetic evidence
for an involvement of the TorS/TorR two-component sys-
tem in the TorC autoregulatory pathway.
Discussion
Before this study was undertaken, no negative regulator
was known for the torCAD operon. The screening of a
multicopy library to identify a downregulator revealed
that the torC structural gene, which encodes the pentahae-
mic c-type cytochrome involved in electron transfer to the
terminal TMAO reductase TorA, negatively regulates the
tor operon. TorC maturation can be prevented by TorC
overproduction, DTT addition or disruption of genes invol-
ved in holocytochrome formation. This decreases tor
operon transcription. We therefore propose that the TorC
apocytochrome, rather than mature TorC protein, is a tor
downregulator (Fig. 5). In agreement with this hypothesis,
we have shown that improving TorC maturation by over-
expression of the c-type cytochrome maturation (ccm )
genes led to increased tor operon expression (Fig. 3).
This effect was probably caused by the reduction in the
TorC apocytochrome concentration in the cell. This result
also indicates that the chromosomal ccm genes do not
allow E. coli to mature large amounts of TorC protein.
This limited capacity for c-type cytochrome maturation
has been noticed previously for other endogenous or exo-
genous c-type cytochromes (Arslan et al., 1998; Roldan et
al., 1998). These observations give physiological rele-
vance for the TorC negative autoregulation that can now
be viewed as a regulatory process that prevents the cyto-
chrome maturation machinery from saturation and thus
allows optimal biogenesis of TorC. The fact that the tor pro-
moter activity was higher in a torC strain, together with the
tor-positive effect of ccm overexpression, strongly suggests
that the TorC negative control occurs even in usual growth
conditions. Finally, it is striking that TorC overproduction
strongly decreased the overall amount of TorC holo-
enzyme in the bacteria, as if the presence of an excess
of apocytochrome jams the cytochrome maturation system.
Therefore, too much transcription of the torCAD operon
might lead to lower amounts of mature TorC protein and,
consequently, reduced TMAO respiratory ability. This
again highlights the important role played by the TorC
negative autoregulation.
Although TorC apocytochrome acting as a negative
regulator is an attractive model, it does not explain the
mechanism by which the tor operon is downregulated.
As the TorC body is located in the periplasm, additional
proteins are necessary to mediate the TorC negative con-
trol up to the tor promoter. Using a genetic approach, we
tested whether the TorS/TorR two-component system
responsible for TMAO induction was also involved in TorC
control. The epistatic experiments showed that either over-
production of the TorR response regulator or a constitutive
mutation in the TorS sensor protein allowed the bacteria
to bypass the TorC regulation (Table 2). The simplest
Fig. 5. Model of tor regulation. The apoTorC is
located on the periplasmic side of the inner
membrane. In the early steps of TorC maturation,
the conserved cysteines of the haem binding
sites (SH) are oxidized by the DsbA/B system
(S±S). Transport of haem (X) across the inner
membrane and haem attachment to apoTorC
require the Ccm proteins. When this maturation
process is affected, the apoform of TorC has a
negative effect ( ) on the tor expression via the
TorS/TorR system. In our model, apoTorC
interacts negatively with the periplasmic detector
region of TorS, whereas TorT and TMAO activate
(‡) the TorS kinase.
Q 1999 Blackwell Science Ltd, Molecular Microbiology, 33, 284±295

explanation deduced from these experiments is that TorC,
TorS and TorR belong to the same pathway, with TorC act-
ing upstream from TorS and TorR. Using the same genetic
approach, we have shown previously that the anaerobic
control of the tor operon is TorS/TorR independent
(Simon et al., 1994; Jourlin et al., 1996a). Accordingly,
the tor anaerobic regulatory pathway is also distinct from
the TorS/TorR-dependent TorC regulatory system
(Table 1).
It is quite intriguing that the TorS/TorR two-component
system mediates the TorC negative control in addition to
TMAO induction and TorT signalling (Fig. 5). Based on
genetic data and, in particular, on the location of the con-
stitutive TorS726 mutation, which corresponds to a deletion
of three residues in the large periplasmic amino-terminal
region of TorS, we proposed previously that TMAO bind-
ing occurs on the periplasmic-exposed `detector' region
(Jourlin et al., 1996a). The TorT regulatory protein could
also interact with this region, as TorT is located in the peri-
plasm and belongs to the TorS/TorR pathway (Jourlin et al.,
1996b). Similarly, the TorC apocytochrome could contact
the TorS periplasmic region. However, we cannot exclude
that lack of TorC maturation somehow impairs TorC trans-
location and, as a consequence, the TorC polypeptide
would interact with a cytoplasmic region of TorS. In any
case and in contrast to TMAO and TorT, this interaction
might reduce the TorS kinase activity. Although the situa-
tion in which a sensor protein detects more than one signal
is not unusual, the fact that, for TorS, these signals appear
as kinase inducer or inhibitor is much more uncommon
(Hoch and Silhavy, 1995). Consequently, the TorS detec-
tion mechanism deserves further study.
There are at least two possible mechanisms for TorS-
mediated TorC regulation. The apoform of TorC could inter-
act with a speci®c domain of the TorS detector region dif-
ferent from those of TMAO and TorT, and this interaction
might induce a TorS conformational change that leads to
the inhibition of TorS activity. Alternatively, the TorC apo-
protein might prevent the binding of TMAO and/or TorT to
TorS, for example by hiding one of the TorS inducer bind-
ing sites. However, before embarking on experiments to
distinguish between these two possibilities, we ®rst have
to demonstrate a direct interaction between the TorS peri-
plasmic region and apoTorC, TorT or even TMAO.
In general, mature c-type cytochromes are stable,
whereas apocytochromes produced, for example, as a
consequence of mutations in the maturation apparatus,
are less stable (ThoÈny-Meyer, 1997). It has been shown
recently that DegP, which is a stress-induced serine pro-
tease active in the periplasm, can degrade certain apo-
cytochromes (Arslan et al., 1998). This is not surprising,
as the main physiological role of DegP is to degrade mis-
folded proteins in the periplasm (Pallen and Wren, 1997).
Although entire TorC apoprotein can be detected when
Q 1999 Blackwell Science Ltd, Molecular Microbiology, 33, 284±295
TorC autoregulation 291
torC is overexpressed (Fig. 2), its half-life remains unknown.
Consequently, one could imagine that TorC negative con-
trol implies a TorC degradation product rather than the
complete TorC apoprotein. This also raises an interesting
question concerning the TorC apocytochrome region
essential for the autoregulatory process. Although we
have no de®nite answer, it is worth mentioning that the
autoregulation is lost when torC is mutated, even if the V
insertion is located in the middle of the gene (Table 1).
Remarkably, in such a strain (LCB628), a mature c-type
cytochrome corresponding to the TorC8 tetrahaemic pro-
tein can be synthesized from the truncated torC gene
(MeÂjean et al., 1994). This leads us to suppose either
that the entire TorC apoprotein is required for autoregula-
tion or that the carboxy-terminal domain of TorC, which is
a monohaemic domain found exclusively in TMAO/DMSO
respiratory systems (Dos Santos et al., 1998), is suf®cient
for the negative control. Unfortunately, attempts to over-
produce the TorC carboxy-terminal domain alone were
unsuccessful. The effect of directed mutagenesis on the
haem-binding sites would help to de®ne the TorC region
involved in negative autoregulation.
Whatever the TorC region involved in the tor negative
control, it is quite surprising that different mutations pre-
venting c-type cytochrome maturation have different effects
on tor operon expression. Thus, the tor promoter activity is
lower in dsbA or dsbB mutants than in ccmC or hemN
mutants (Fig. 4). One possible explanation is that the fold-
ing state of the TorC apocytochrome is different in dsb
mutants than in ccmC or hemN mutants. Indeed, apoTorC
is probably completely unfolded in dsb mutants, as no dis-
ulphide bond could be formed, whereas TorC is likely to be
prefolded by the action of DsbA in ccmC or hemN mutants
(Fig. 5). If true, the unfolded apoform of TorC is a stronger
downregulator than the prefolded TorC apocytochrome
that contains disulphide bonds. The fact that DTT addition
decreased tor activity at least as much as a dsb mutation
con®rms that the absence of disulphide bonds converts
TorC into an ef®cient tor downregulator. Together, the
results reported in this study agree with the idea that
TorC autoregulation is a sophisticated regulatory process,
which allows ®ne-tuned expression of the tor operon in
response to the maturation state of TorC.
In bacteria, TorC is the ®rst c-type cytochrome that has
been found to have dual functions, one as a structural pro-
tein involved in electron transfer and the other as a down-
regulator. Usually, c-type cytochromes are only considered
as electron transfer proteins. However, a c-type haem-bind-
ing protein (DcrA), found in the sulphate-reducing bacteria
Desulfovibrio vulgaris Hildenborough, seems rather to be
a chemoreceptor sensing the oxygen concentration of
the environment (Fu et al., 1994). In eukaryotes, mito-
chondrial c-type cytochrome (Cyt c) controls both cellular
energetic metabolism and apoptosis (Cai et al., 1998). In
292 M. Ansaldi, C. Bordi, M. Lepelletier and V. MeÂjean
the mitochondria, Cyt c behaves as a classical electron
transfer protein but, once released into the cytosol, Cyt c
can initiate the activation cascade of caspases through
interaction with apoptotic protease-activating factors.
However, Cyt c apocytochrome alone does not activate
apoptosis.
c-type cytochromes are haemoproteins widespread in
nature. TorC, for example, belongs to a large bacterial
family that contains both tetrahaemic and pentahaemic
c-type cytochromes (Roldan et al., 1998). Given the essen-
tial respiratory role played by c-type cytochromes and the
complex maturation process required for their biogenesis,
it would not be surprising to discover other autoregulatory
pathways involving c-type cytochromes.
Experimental procedures
Strains, growth conditions and general methods
E. coli strains and plasmids used in this study are listed in
Table 3. Bacteria were grown on L broth medium (Miller,
1972). When necessary, TMAO (10 mM), DTT (5 mM) and/or
arabinose (0.001±0.1½) were added. Antibiotics were used
at the following ®nal concentrations: ampicillin, 50 mg ml 1;
chloramphenicol, 25 mg ml 1; and spectinomycin, 25 mg ml 1.
Anaerobic cultures were performed in full cap tubes and
grown overnight; in contrast, aerobic cultures were grown in
agitated ¯asks until the OD 600 reached 0.4. DNA preparations
were carried out with high pure DNA isolation kits from Boeh-
ringer Mannheim. Transformations were performed accord-
ing to the method of Chung and Miller (1988).
Genetic procedures
The genomic library, a kind gift from C. Bohn and P. Bouloc,
was obtained by partial digestion of chromosomal DNA from
strain C600 with Sau 3A to produce DNA fragments 4±8 kb
in length. These DNA fragments were puri®ed and ligated
into the BamHI site of pUC18. A library of at least 20 000 inde-
pendent recombinant clones was made, and DNA from such a
pool was used to transform a torA8± 8lacZ fusion-containing
strain (LCB620). Transformants were plated onto MacConkey
lactose (1%) TMAO (10 mM) medium, and white colonies were
picked. Twenty-three plasmid clones, whose presence affected
the expression of the torA8± 8lacZ fusion, were thus selected
and sequenced using primers derived from the pUC18
sequence (lacZ874: 58-GGTCGACTCTAGAGGA and pUCR:
58-CGAGGGAGCTTCCAGGGGGAAAC). Eight of these car-
ried nearly the same fragment, containing the torT, torR and
torC genes, in both orientations with respect to the lac promo-
ter of the pUC18 vector. Plasmid clone pUW13 was chosen
for further experiments.
Random mini-Tn10 mutagenesis was performed according
to Kleckner et al. (1991) with the lNK1324 (Cam R ). Location
of the insertions was performed using a rapid inverse PCR
method as follows: mutated bacteria were grown in LB medium

Table 3. Bacterial strains and plasmids used in this study.

Relevant characteristics Source/reference

Strains
MC4100 araD139 D(lacIPOZYA-argF ) U169 rpsL thi M. J. Casadaban
LCB502 MC4100 torC 2::V (Mun I site, position 66)a This work
LCB620 MC4100 torA8 ::MudII 1734 (torA 8±8lacZ ) MeÂjean et al. (1994)
LCB628 MC4100 torC 1::V (Bss HII site, position 586)a MeÂjean et al. (1994)
LCB726 LCB620 torS726 Jourlin et al. (1996a)
LCB3001 LCB620 ccmC ::Tn10 position 655 b This work
LCB3003 LCB620 dsbA::Tn10 position 580 b This work
LCB3026 LCB620 dsbA::Tn10 position 534 b This work
LCB3035 LCB620 dsbA::Tn10 position 509 b This work
LCB3058 LCB620 dsbA::Tn10 position 550 b This work
LCB3070 LCB620 dsbA::Tn10 position 486 b This work
LCB3094 LCB620 dsbB::Tn10 position 205 b This work
LCB3201 LCB620 hemN::Tn10 position 332 b This work
LCB3213 LCB620 hemN::Tn10 position 991 b This work
KM29 AB1157 D(recC ptr recB recD )::Plac -bet exo kan recJ Murphy (1998)
Plasmids
pUW13 Sau 3A E. coli genomic fragment (1055 kb to 1059.7 kbc ) cloned into pUC18 This work
pUW13D1 MunI±Eco RI deletion of pUW13 This work
pUW13V V sequence inserted at the Mun I site of pUW13 This work
pPTor43 torC promoter ( 86 to ‡1)a cloned into pGE593 This work
pBAD24 vector containing the arabinose PBAD promoter Guzman et al. (1995)
pBC5H torC coding sequence inserted into pBAD24 This work
pGS1 torR coding sequence inserted into pJF119EH Simon et al. (1994)
pEC86 ccmABCDEFGH inserted into pACYC184 Arslan et al. (1998)
pHP45 plasmid containing the V insertion sequence Prentki and Krisch (1984)

a. Nucleotide position relative to the transcription start site of torC .

b. Nucleotide position relative to the first nucleotide of the initiation codon.
c. Chromosomal region relative to the replication origin of E. coli .

Q 1999 Blackwell Science Ltd, Molecular Microbiology, 33, 284±295


with chloramphenicol; then chromosomal DNA was extracted
and digested by the restriction endonuclease HpaII in the pre-
sence of RNase I. HpaII was then heat inactivated (658C,
20 min), and the mixture was diluted until a DNA concentration
of 1 mg ml 1 was obtained. Intramolecular ligation was carried
out by adding T4 DNA ligase (Biolabs). After a dialysis step
against water, inverse PCR was performed using divergent
primers designed from the Tn10 sequence: Cam3, 58-
CTGCCTCCCAGAGCCTGATAAAAACGG; and Cam4, 58-
TCATATCGTCAATTATTACCTCCACG. The PCR product
was then puri®ed (Gene Clean kit from Bio101) and sequenced
using an ABI apparatus with primer Cam3.
Construction of the torC null mutant (LCB502)
R coli genomic library, L. ThoÈny-Meyer for the gift of plasmid
The 2 kb EcoRI fragment containing the V-interposon (Spc )
from pHP45 (Prentki and Krisch, 1984) was cloned into plas-
mid pUW13, previously linearized at the MunI site within torC,
to create plasmid pUW13V. Disruption of the chromosomal
torC gene was performed according to the method of Murphy
(1998). Plasmid pUW13V was linearized with PvuII and intro-
duced into strain KM29 by electrotransformation. Electrocom-
petent cells were prepared as described previously (Murphy,
1998). IPTG (1 mM) was added during growth to induce the
recombination functions. Linear DNA (100 ng) was mixed with
40 ml of electrocompetent cells into a precooled 0.1 cm cuvette
and submitted to a discharge of 25 kV cm 1 with an RC time
constant of 4.2 ms. Spectinomycin-resistant colonies were
then analysed by PCR using primers V 58-GCTTGCTCAAT-
CAATCACCG (derived from the V sequence) and ML13 58-
GCCACAAAGCGACCAT (derived from the torA sequence).
A deletion of about 0.7 kb of the torC sequence occurred dur-
ing the V insertion, and PCR analysis revealed that the torA
gene and the torC promoter remained unaffected.
Construction of plasmids
To create plasmid pBC5H, we used PCR to generate a pro-
moterless torC DNA fragment with an upstream EcoRI site
and a downstream SmaI site. This PCR product was then
cloned into the same sites on pBAD24 (Guzman et al.,
1995), thus placing torC under the control of the arabinose
promoter. To create plasmid pPTor43, we ampli®ed the torC
promoter sequence by PCR (from position 86 to the tran-
scription start site). The DNA fragment was blunted using
the blunting kit from Takara and then introduced into plasmid
pGE593 previously linearized with SmaI, thus placing the lacZ
gene under the control of the torC promoter. A EcoRI±MunI
deletion of pUW13 led to plasmid pUW13D1.
b-Galactosidase assays
b-Galactosidase activities were measured on whole cells
according to the method of Miller (1972); values represent
the average of at least three determinations with a variation
of no more than 10% from the mean.
Analytical procedures
Preparation of soluble and membranous fractions was per-
formed as described by Silvestro et al. (1988). Protein
Q 1999 Blackwell Science Ltd, Molecular Microbiology, 33, 284±295
TorC autoregulation 293
separation was carried out using SDS±10% polyacrylamide
gels. After the run, the presence of haems in TorC was
revealed by staining the gel for peroxidase activity using
3,38,5,58 tetramethylbenzidine (TMBZ; Thomas et al., 1976);
then Coomassie blue staining of total proteins was performed.
The amount of TMAO reductase antigen present in the
extracts was determined by rocket immunoelectrophoresis
using a polyclonal antiserum speci®c for TMAO reductase
(Pommier et al., 1998). Protein concentrations were estimated
using the technique of Lowry.
Acknowledgements
We thank C. Bohn and P. Bouloc for providing us with the E.
pEC86 and for helpful suggestions, and K. C. Murphy for the
kind gift of strain KM29. We also thank C. Iobbi-Nivol, C. Jourlin
and S. Wells for critical reading of the manuscript. We are
grateful to S. Solia for technical assistance, and J. Pommier
for helpful advice. This work was supported by grants from
the Centre National de la Recherche Scienti®que, the Univer-
sity of Aix-Marseille II and a MENRT fellowship to M.A.
References
Arslan, E., Schulz, H., Zufferey, R., Kunzler, P., and ThoÈny-
Meyer, L. (1998) Overproduction of the Bradyrhizobium
japonicum c-type cytochrome subunits of the cbb3 oxidase
in Escherichia coli. Biochem Biophys Res Commun 251:
744±747.
Barrett, E.L., and Kwan, H.S. (1985) Bacterial reduction of
trimethylamine oxide. Annu Rev Microbiol 39: 131±149.
Cai, J., Yang, J., and Jones, D.P. (1998) Mitochondrial con-
trol of apoptosis: the role of cytochrome c. Biochim Biophys
Acta 1366: 139±149.
Chanal, A., Santini, C., and Wu, L. (1998) Potential receptor
function of three homologous components, TatA, TatB and
TatE, of the twin-arginine signal sequence-dependent
metalloenzyme translocation pathway in Escherichia coli.
Mol Microbiol 30: 674±676.
Chung, C.T., and Miller, R.H. (1988) A rapid and convenient
method for the preparation and storage of competent bac-
terial cells. Nucleic Acids Res 16: 3580.
Crooke, H., and Cole, J. (1995) The biogenesis of c-type
cytochromes in Escherichia coli requires a membrane-
bound protein, DipZ, with a protein disulphide isomerase-
like domain. Mol Microbiol 15: 1139±1150.
Czjzek, M., Dos Santos, J.P., Pommier, J., Giordano, G.,
MeÂjean, V., and Haser, R. (1998) Crystal structure of oxi-
dized trimethylamine N-oxide reductase from Shewanella
massilia at 2.5 AÊ resolution. J Mol Biol 284: 435±447.
Dalbey, R.E., and Robinson, C. (1999) Protein translocation
into and across the bacterial plasma membrane and the
plant thylakoid membrane. Trends Biochem Sci 24: 17±
22.
Dos Santos, J.P., Iobbi-Nivol, C., Couillault, C., Giordano, G.,
and MeÂjean, V. (1998) Molecular analysis of the trimethyl-
amine N-oxide (TMAO) reductase respiratory system from
a Shewanella species. J Mol Biol 284: 421±433.
Fu, R., Wall, J.D., and Voordouw, G. (1994) DcrA, a c-type
294 M. Ansaldi, C. Bordi, M. Lepelletier and V. MeÂjean
heme-containing methyl-accepting protein from Desulfo-
vibrio vulgaris Hildenborough, senses the oxygen concen-
tration or redox potential of the environment. J Bacteriol
176: 344±350.
Gennis and Stewart, V. (1996) Respiration. In Escherichia
coli and Salmonella. Neidhardt, F.C. (ed.). Washington,
DC: American Society for Microbiology Press, pp. 217±261.
Goldman, B.S., Beck, D.L., Monika, E.M., and Kranz, R.G.
(1998) Transmembrane heme delivery systems. Proc Natl
Acad Sci USA 95: 5003±5008.
Goudreau, P.N., and Stock, A.M. (1998) Signal transduction
in bacteria: molecular mechanisms of stimulus-response
coupling. Curr Opin Microbiol 1: 160±169.
Grove, J., Tanapongpipat, S., Thomas, G., Grif®ths, L.,
Crooke, H., and Cole, J. (1996) Escherichia coli K-12
genes essential for the synthesis of c-type cytochromes
and a third nitrate reductase located in the periplasm.
Mol Microbiol 19: 467±481.
Guzman, L.M., Belin, D., Carson, M.J., and Beckwith, J.
(1995) Tight regulation, modulation, and high-level expres-
sion by vectors containing the arabinose PBAD promoter.
J Bacteriol 177: 4121±4130.
Hoch, J.A., and Silhavy, T.J. (eds) (1995) Two-Component
Signal Transduction. Washington, DC: American Society
for Microbiology Press.
Iobbi-Nivol, C., Crooke, H., Grif®ths, L., Grove, J., Hussain, H.,
Pommier, J., et al. (1994) A reassessment of the range of
c-type cytochromes synthesized by Escherichia coli K-12.
FEMS Microbiol Lett 119: 89±94.
Iobbi-Nivol, C., Pommier, J., Simala-Grant, J., MeÂjean, V., and
Giordano, G. (1996) High substrate speci®city and induction
characteristics of trimethylamine N-oxide reductase of
Escherichia coli. Biochim Biophys Acta 1294: 77±82.
Jourlin, C., Bengrine, A., Chippaux, M., and MeÂjean, V.
(1996a) An unorthodox sensor protein (TorS) mediates
the induction of the tor structural genes in response to tri-
methylamine N-oxide in Escherichia coli. Mol Microbiol 20:
1297±1306.
Jourlin, C., Simon, G., Pommier, J., Chippaux, M., and
MeÂjean, V. (1996b) The periplasmic TorT protein is required
for trimethylamine N-oxide reductase gene induction in
Escherichia coli. J Bacteriol 178: 1219±1223.
Jourlin, C., Ansaldi, M., and MeÂjean, V. (1997) Transphos-
phorylation of the TorR response regulator requires the
three phosphorylation sites of the TorS unorthodox sensor
in Escherichia coli. J Mol Biol 267: 770±777.
Kleckner, N., Bender, J., and Gottesman, S. (1991) Uses of
transposons with emphasis on Tn10. Methods Enzymol
204: 139±180.
Kranz, R., Lill, R., Goldman, B., Bonnard, G., and Merchant,
S. (1998) Molecular mechanisms of cytochrome c bio-
genesis: three distinct systems. Mol Microbiol 29: 383±396.
MeÂjean, V., Iobbi-Nivol, C., Lepelletier, M., Giordano, G.,
Chippaux, M., and Pascal, M.C. (1994) TMAO anaerobic
respiration in Escherichia coli : involvement of the tor
operon. Mol Microbiol 11: 1169±1179.
Metheringham, R., Grif®ths, L., Crooke, H., Forsythe, S., and
Cole, J. (1995) An essential role for DsbA in cytochrome c
synthesis and formate-dependent nitrite reduction by
Escherichia coli K12. Arch Microbiol 164: 301±307.
Metheringham, R., Tyson, K.L., Crooke, H., Missiakas, D.,
Raina, S., and Cole, J.A. (1996) Effects of mutations in
genes for proteins involved in disulphide bond formation
in the periplasm on the activities of anaerobically induced
electron transfer chains in Escherichia coli K12. Mol Gen
Genet 253: 95±102.
Miller, J.H. (1972) Experiments in Molecular Genetics. Cold
Spring Harbor, NY: Cold Spring Harbor Laboratory Press.
Missiakas, D., Georgopoulos, C., and Raina, S. (1993) Iden-
ti®cation and characterization of the Escherichia coli gene
dsbB, whose product is involved in the formation of disul-
®de bonds in vivo. Proc Natl Acad Sci USA 90: 7084±
7088.
Mizuno, T. (1997) Compilation of all genes encoding two-
component phosphotransfer signal transducers in the
genome of Escherichia coli. DNA Res 4: 161±168.
Mouncey, N.J., Choudhary, M., and Kaplan, S. (1997) Char-
acterization of genes encoding dimethyl sulfoxide reductase
of Rhodobacter sphaeroides 2.4.1T: an essential metabolic
gene function encoded on chromosome II. J Bacteriol 179:
7617±7624.
Murphy, K.C. (1998) Use of bacteriophage lambda recombi-
nation functions to promote gene replacement in Escheri-
chia coli. J Bacteriol 180: 2063±2071.
Page, M.D., and Ferguson, S.J. (1989) A bacterial c-type
cytochrome can be translocated to the periplasm as an
apo form; the biosynthesis of cytochrome cd1 (nitrite
reductase) from Paracoccus denitri®cans. Mol Microbiol
5: 653±661.
Page, M.D., and Ferguson, S.J. (1990) Apo forms of cyto-
chrome c550 and cytochrome cd1 are translocated to the
periplasm of Paracoccus denitri®cans in the absence of
haem incorporation caused either mutation or inhibition
of haem synthesis. Mol Microbiol 7: 1181±1192.
Pallen, M.J., and Wren, B.W. (1997) The HtrA family of serine
proteases. Mol Microbiol 26: 209±221.
Pommier, J., MeÂjean, V., Giordano, G., and Iobbi-Nivol, C.
(1998) TorD, a cytoplasmic chaperone that interacts with
the unfolded trimethylamine N-oxide reductase enzyme
(TorA) in Escherichia coli. J Biol Chem 273: 16615±16620.
Prentki, P., and Krisch, H.M. (1984) In vitro insertional muta-
genesis with a selectable DNA fragment. Gene 29: 303±
313.
Raina, S., and Missiakas, D. (1997) Making and breaking
disul®de bonds. Annu Rev Microbiol 51: 179±202.
Reid, E., Eaves, D.J., and Cole, J.A. (1998) The CcmE pro-
tein from Escherichia coli is a haem-binding protein. FEMS
Microbiol Lett 166: 369±375.
Roldan, M.D., Sears, H.J., Cheesman, M.R., Ferguson, S.J.,
Thomson, A.J., Berks, B.C., et al. (1998) Spectroscopic
characterization of a novel multiheme c-type cytochrome
widely implicated in bacterial electron transport. J Biol
Chem 273: 28785±28790.
Sambongi, Y., and Ferguson, S.J. (1996) Mutants of Escheri-
chia coli lacking disulphide oxidoreductases DsbA and
DsbB cannot synthesise an exogenous monohaem c-type
cytochrome except in the presence of disulphide com-
pounds. FEBS Lett 398: 265±268.
Santini, C.L., Ize, B., Chanal, A., Muller, M., Giordano, G.,
and Wu, L.F. (1998) A novel sec-independent periplasmic
protein translocation pathway in Escherichia coli. EMBO J
17: 101±112.
Q 1999 Blackwell Science Ltd, Molecular Microbiology, 33, 284±295

Sargent, F., Bogsch, E.G., Stanley, N.R., Wexler, M., Robin-
son, C., Berks, B.C., et al. (1998) Overlapping functions of
components of a bacterial Sec-independent protein export
pathway. EMBO J 17: 3640±3650.
Schulz, H., Hennecke, H., and ThoÈny-Meyer, L. (1998) Proto-
type of a heme chaperone essential for cytochrome c
maturation. Science 281: 1197±1200.
Silvestro, A., Pommier, J., and Giordano, G. (1988) The indu-
cible trimethylamine N-oxide reductase of Escherichia coli
K12: biochemical and immunological studies. Biochim Bio-
phys Acta 954: 1±13.
Simon, G., MeÂjean, V., Jourlin, C., Chippaux, M., and Pascal,
M.C. (1994) The torR gene of Escherichia coli encodes
a response regulator protein involved in the expression of
the trimethylamine N-oxide reductase genes. J Bacteriol
176: 5601±5606.
Simon, G., Jourlin, C., Ansaldi, M., Pascal, M.C., Chippaux,
M., and MeÂjean, V. (1995) Binding of the TorR regulator
to cis-acting direct repeats activates tor operon expression.
Mol Microbiol 17: 971±980.
Tanapongpipat, S., Reid, E., Cole, J.A., and Crooke, H.
(1998) Transcriptional control and essential roles of the
Escherichia coli ccm gene products in formate-dependent
nitrite reduction and cytochrome c synthesis. Biochem J
334: 355±365.
Thomas, P.E., Ryan, D., and Levin, W. (1976) An improved
staining procedure for the detection of the peroxidase
activity of cytochrome P-450 on sodium dodecyl sulfate
polyacrylamide gels. Anal Biochem 75: 168±176.
ThoÈny-Meyer, L. (1997) Biogenesis of respiratory cyto-
chromes in bacteria. Microbiol Mol Biol Rev 61: 337±376.
Q 1999 Blackwell Science Ltd, Molecular Microbiology, 33, 284±295
View publication stats
TorC autoregulation 295
ThoÈny-Meyer, L., and Kunzler, P. (1997) Translocation to the
periplasm and signal sequence cleavage of preapocyto-
chrome c depend on sec and lep, but not on the ccm
gene products. Eur J Biochem 246: 794±799.
ThoÈny-Meyer, L., Fischer, F., Kunzler, P., Ritz, D., and Hen-
necke, H. (1995) Escherichia coli genes required for cyto-
chrome c maturation. J Bacteriol 177: 4321±4326.
Troup, B., Hungerer, C., and Jahn, D. (1995) Cloning and
characterization of the Escherichia coli hemN gene encod-
ing the oxygen-independent coproporphyrinogen III oxidase.
J Bacteriol 177: 3326±3331.
Tyson, K., Metheringham, R., Grif®ths, L., and Cole, J. (1997)
Characterisation of Escherichia coli K12 mutants defective
in formate-dependent nitrite reduction: essential roles for
hemN and the menFDBCE operon. Arch Microbiol 168:
403±411.
Ujiiye, T., Yamamoto, I., Nakama, H., Okubo, A., Yamazaki,
S., and Satoh, T. (1996) Nucleotide sequence of the genes,
encoding the pentaheme cytochrome (dmsC ) and the trans-
membrane protein (dmsB ), involved in dimethyl sulfoxide
respiration from Rhodobacter sphaeroides f. sp. denitri®-
cans. Biochim Biophys Acta 1277: 1±5.
Wang, A., and Bolen, D.W. (1997) A naturally occurring pro-
tective system in urea-rich cells: mechanism of osmolyte
protection of proteins against urea denaturation. Biochem-
istry 36: 9101±9108.
Weiner, J.H., Bilous, P.T., Shaw, G.M., Lubitz, S.P., Frost, L.,
Thomas, G.H., et al. (1998) A novel and ubiquitous system
for membrane targeting and secretion of cofactor-contain-
ing proteins. Cell 93: 93±101.