Thiol-Disulfide Exchange in an Immunoglobulin-like Fold: Structure of the
N-Terminal Domain of DsbD†,‡
Celia W. Goulding,§Michael R. Sawaya,§Angineh Parseghian,§Vincent Lim,|David Eisenberg,§and
Howard Hughes Medical Institute and Laboratory of Structural Biology and Molecular Medicine, UCLA-DOE,
P.O. Box 951570, Los Angeles, California 90095-1570, and Department of Biochemistry and Molecular Biology,
The UniVersity of Chicago, 920 East 58th Street, Chicago, Illinois 60637
ReceiVed December 13, 2001; ReVised Manuscript ReceiVed March 22, 2002
ABSTRACT: Escherichia coli DsbD transports electrons across the plasma membrane, a pathway that leads
to the reduction of protein disulfide bonds. Three secreted thioredoxin-like factors, DsbC, DsbE, and
DsbG, reduce protein disulfide bonds whereby an active site C-X-X-C motif is oxidized to generate a
disulfide bond. DsbD catalyzes the reduction of the disulfide of DsbC, DsbE, and DsbG but not of the
thioredoxin-like oxidant DsbA. The reduction of DsbC, DsbE, and DsbG occurs by transport of electrons
from cytoplasmic thioredoxin to the C-terminal thioredoxin-like domain of DsbD (DsbDC). The N-terminal
domain of DsbD, DsbDN, acts as a versatile adaptor in electron transport and is capable of forming disulfides
with oxidized DsbC, DsbE, or DsbG as well as with reduced DsbDC. Isolated DsbDNis functional in
electron transport in vitro. Crystallized DsbDNassumes an immunoglobulin-like fold that encompasses
two active site cysteines, C103 and C109, forming a disulfide bond between ?-strands. The disulfide of
DsbDNis shielded from the environment and capped by a phenylalanine (F70). A model is discussed
whereby the immunoglobulin fold of DsbDNmay provide for the discriminating interaction with thioredoxin-
like factors, thereby triggering movement of the phenylalanine cap followed by disulfide rearrangement.
Folding of secreted proteins within the periplasm of
Escherichia coli requires the formation of disulfide bonds,
a process that is dependent on the Dsb1(disulfide bond)
proteins. Disulfide bond formation occurs via oxidation of
a pair of cysteine sulfhydryl groups, a reaction that is
catalyzed by DsbA. Electrons generated by oxidative folding
are transferred from DsbA to DsbB and channeled into the
electron transport chain (1, 2) (Figure 1a). DsbA cannot
provide a proofreading activity for improperly formed
disulfides which are repaired by DsbC, DsbD, DsbE, and
DsbG. For example, maturation of c-type cytochromes
requires first oxidation and then reduction of two cysteine
residues. The reduction of cystines within c-type cytochromes
is catalyzed by DsbE (3, 4). Such redox rearrangement of
the apocytochromes is needed to prevent random oxidation
of the cysteines that are involved in the covalent attachment
of heme. Reshuffling of incorrectly paired disulfide bonds
of periplasmic proteins is also initiated by a reduction
reaction that involves DsbC or DsbG (5, 6). In all of these
reactions, DsbC, DsbG, and DsbE act as electron donors.
As no source of reducing equivalent is generated in the
bacterial periplasm, Gram-negative cells have evolved a
pathway whereby DsbD transfers electrons from cytoplasmic
NADPH across the cytoplasmic membrane to DsbC, DsbE,
and DsbG (Figure 1a) (7, 8).
DsbC, DsbG, and DsbE exist in reduced [Dsb-(SH)2] and
oxidized [Dsb-S2] states. Only reduced Dsb-(SH)2 can
transfer electrons to target substrates. In these redox-active
proteins, two sulfhydryl groups, -(SH)2, are positioned in a
dithiol C-X-X-C motif. This motif was first characterized
for thioredoxin, a small ubiquitous cytoplasmic protein that
reduces disulfide bonds in the cytoplasm (9, 10). The crystal
structures of both DsbC (11) and DsbE (12) reveal the
presence of a thioredoxin-like domain bearing the redox
active C-X-X-C motif. DsbG shares sequence similarity with
DsbC and is therefore presumed to contain a thioredoxin fold.
Furthermore, the C-terminal domain of DsbD, DsbDC
(residues 450-546), resembles protein disulfide isomerase,
another member of the thioredoxin superfamily. Both the
N- and C-terminal domains of DsbD, DsbDN, and DsbDC,
respectively, are located in the periplasm (13, 14), and each
harbors a pair of cysteines, C103-C109 and C461-C464,
respectively (Figure 1b). DsbDNand DsbDCare connected
by eight transmembrane (TM) segments with three cysteine
residues. Five out of the seven cysteines of DsbD are
absolutely required for electron transport activity (C103, C109,
C163, C285, and C461) (13). C163and C285are located in TM1
and TM4, respectively (Figure 1b).
Previous work examined the mechanism whereby DsbD
transfers electrons from the cytoplasm to the periplasm.
Electron transfer mediated by DsbD was demonstrated by
†This work was supported by Department of Energy Grant DOE
ER 60615-1004936-0000034 to D.E. and by U.S. Public Health Service
Grants GM62410 and GM58266 to D.E. and D.M., respectively.
‡The coordinates for DsbDNhave been deposited in the Brookhaven
Protein Data Bank with the accession number 1L6P.
* To whom correspondence should be addressed: dmissiak@
|The University of Chicago.
1Abbreviations: Dsb, disulfide bond formation; TrxA, thioredoxin;
TrxB, thioredoxin reductase; DsbDN, the N-terminal domain of DsbD
(1-144 amino acids); DsbDC, the C-terminal thioredoxin-like domain
Biochemistry 2002, 41, 6920-6927
10.1021/bi016038l CCC: $22.00© 2002 American Chemical Society
Published on Web 05/09/2002
isolating reaction intermediates that are tethered by a disulfide
bond between redox-active proteins (15). Such extremely
short-lived intermediates can be captured using redox
proteins containing only half of the redox-active site.
Substitution of DsbD C285with alanine and TrxA C35with
serine (C32-G-P-S35) led to the production of a stable
complex. Characterization of this complex showed that the
two proteins are trapped in a mixed disulfide species with a
disulfide bond between C32 of TrxA and C163 of DsbD,
suggesting that TrxA resolves a disulfide between C163and
C285 of DsbD. Upon TrxA-mediated reduction, C285 must
transfer electrons to C461and C464of DsbD, as substitution
of C461in the C-terminal thioredoxin-like domain of DsbD
(C461-V-A-C) with alanine abolishes electron transfer. The
C-terminal domain is assumed to transfer electrons to the
N-terminal domain of DsbD, presumably by forming a
disulfide between C109 and C461. Electron transfer also
involves a reduction of the DsbC dithiol motif (C98-G-Y-C)
and occurs between DsbC and DsbDN. Coexpression of DsbD
C103A and DsbC C101A, but not of DsbC C98A or DsbD
C109A, leads to the formation of a mixed disulfide species
between the two Dsb proteins. Thus, the final step of DsbD-
mediated electron transfer likely occurs via thiol-disulfide
exchange between C109of DsbD and C98of DsbC (16).
All electron transfer steps examined for DsbD occur by a
specific mechanism, i.e., the intramolecular exchange of
thiol-disulfide bonds (15, 17). Thioredoxin (TrxA) transfers
electrons to the transmembrane domain of DsbD. Electrons
are transferred to DsbDCbut not to DsbA, DsbC, DsbE, or
DsbG, even though all of these proteins contain a thio-
redoxin-like fold. Specificity may be solely achieved by
proximity of the two domains and the intramolecular nature
of the reaction. Electrons are then transferred to DsbDN.
DsbDNselectively transfers electrons to DsbC, DsbE, and
DsbG but not to DsbA or DsbB (16). DsbDNfunctions as
an adaptor unit that permits electron transfer between
thioredoxin moieties but avoids futile electron cycling
between reductants such as DsbC or DsbG and the oxidant
DsbA. Nonetheless, DsbDNitself must possess redox quali-
ties of thioredoxins as its sulfhydryls must be able to receive
and donate electrons very rapidly. We have crystallized
DsbDNand describe structural features that likely contribute
to the remarkable redox properties of the active site cysteines.
MATERIALS AND METHODS
Protein Purification. The nucleotide sequence encoding
DsbDNwas amplified by the polymerase chain reaction using
the following 5′- and 3′-primers: gtttttgccggtccttcgacgcgccg
and aagaattcacaattgcgcggtgggctg, respectively. The fragment
was cloned in vector pGEX-2T (Pharmacia) using the
restriction sites BamHI and EcoRI. The protein was expressed
as a glutathione S-transferase N-terminal DsbD (GST-DsbDN)
fusion protein in E. coli using DH5R cells. Ten milligrams
of protein was obtained from an 8 Lculture induced at an
OD600nm of 0.2 with 1 mM IPTG for 3 h at 37 °C. Cells
were collected by centrifugation, suspended in 50 mM Tris-
HCl, pH 7.5, 150 mM NaCl, and 10% glycerol, and broken
in a French press. DsbDNwas recovered with the total soluble
proteins by ultracentrifugation for 1 h at 100000g. This
soluble fraction was loaded on a glutathione-Sepharose
column (Pharmacia), and washed with 10 volumes of buffer.
DsbDNwas cleaved from bound GST by thrombin digestion
as described by the manufacturer (Pharmacia). Cleaved
DsbDNwas further purified over mono-Q-Sepharose using
20 mM Tris-HCl, pH 6.4, and eluted with a gradient of NaCl
from 0 to 250 mM. DsbDNeluted at 30-50 mM NaCl. Mass
spectrometry analysis of DsbDNrevealed a mass of 16073,
which corresponds to the expected mass for the 144 amino
acid protein domain (16072.78). As predicted from the
cloning procedure, the DsbDNsequence initiates with the
expected Gly followed by Ser instead of Leu.
Crystallization and Structure Determination. Initial crys-
tallization of purified DsbDNwas carried out by concentrating
FIGURE 1: Model for disulfide bond formation and rearrangement
catalyzed by Dsb proteins of E. coli. (a) Known Dsb proteins in
the E. coli system. DsbA oxidizes disulfide bonds of newly
translocated proteins. The transmembrane protein DsbB accepts
electrons from DsbA and transfers them to quinone or menaquino-
nes embedded in the bilayer. DsbC and DsbG are disulfide
isomerase proteins that catalyze the re-formation of incorrectly
formed disulfide bonds. DsbE is thought to reduce the cysteines of
aopcytochrome c for heme attachment (anchoring of DsbE in the
plasma membrane is not depicted on the figure). These three
proteins accept electrons from the cytoplasm through the DsbD
transmembrane protein. (b) Model for DsbD-mediated transfer of
electrons across the cytoplasmic membrane. TrxB (NADPH)
donates electrons to the active site dithiol motif of TrxA, C32-X-
X-C35. C32of TrxA interacts with C163of DsbD. This presumably
resolves the disulfide between C163and C285of DsbD (represented
as a gray broken line). In the next step, C163and C285presumably
donate electrons to the dithiols at C461and C464. The reduced dithiol
of C461attacks the disulfide at C103and C109, which transfer electrons
to the oxidized dithiol motif of DsbC. The square-shaped box
indicates the presence of a thioredoxin-like fold. Dark arrows
indicate the steps supported by experimental data whereas gray
arrows indicate steps presumed to occur.
Thiol-Disulfide Exchange in an Ig-like Fold
Biochemistry, Vol. 41, No. 22, 2002 6921
the protein to 10 mg/mL using 0.1 M Tris-HCl, and 0.5 M
NaCl, pH 7.4. The protein crystallized in 25% poly(ethylene
glycol) 4000, 0.1 M sodium acetate, pH 4.6, and 0.2 M NH4-
SO4. The crystals were swiped through a solution of 20%
glycerol and crystallization buffer (1:1) before mounting and
collecting data under cryoconditions. Potassium iodide soaks
were performed with 0.5 M potassium iodide in the cryo-
protectant. The crystals were soaked for 1 min before
mounting under cryoconditions. An iodo derivative of the
DsbDNcrystal diffracted to 2.3 Å, and a native crystal
diffracted to 1.65 Å with cell dimensions of 53.21 × 55.63
× 102.05 Å with one monomer per asymmetric unit in space
group C2221. Data were processed using DENZO and
SCALEPACK (18). SIRAS phasing proceeded by the
usual methods of heavy atom location [SHELXD (http://
shelx.uni-ac.gwdg.de/SHELXD/)], maximum likelihood phase
refinement [ML-PHARE (19)], and density modification
[DM (20)]. Phase extension to 1.65 Å permitted automated
model building with ARP/wARP (21) of all of the residues
of DsbDNexcept the initial 3 residues (1-3) and the last 21
residues (124-144). Model building was done in O (22),
and the model was refined using SHELXL. Data and
refinement statistics are presented in Table 1.
Structure and Sequence Analysis. BLAST (23) and CLUST-
ALW (24) were used for database searches and multiple
sequence alignments, respectively. Pairwise alignments were
calculated using the Smith-Waterman algorithm. Similar
protein structures were searched using DALI (25) and aligned
with combinatorial extension (CE) (26). Figure 2b is il-
lustrated using SETOR (27).
Members of the thioredoxin superfamily display small
amounts of sequence similarity and are predicted to assume
similar three-dimensional folded structures. An active site
C-X-X-C motif undergoes reversible oxidation-reduction
and is present in all members of the thioredoxin superfamily.
The primary sequence of DsbDNdoes not display sequence
similarity with thioredoxin, and DsbDNis not predicted to
assume a thioredoxin-like fold. Further, DsbDNdoes not
harbor the characteristic C-X-X-C motif. Comparison of
several different DsbD-like proteins revealed the presence
of a Q-G-C103-X5-C109-Y sequence motif within DsbDN
(Figure 2a), where C109is implicated in the redox reaction
that transfers electrons from DsbD to DsbC (15). Purified,
recombinant DsbDNwas able to transfer electrons and hence
reduced oxidized DsbG in vitro (data not shown). When used
in substoichiometric amounts over DsbG, DsbDNcould still
reduce oxidized DsbG provided stoichiometric amounts of
the purified and reduced C-terminal domain of DsbD, DsbDC,
were added to the reaction mixture (data not shown).
Together, these data suggested that DsbDNis a functional
domain, as it is capable of accepting electrons from DsbDC
and transferring them to DsbG.
OVerall Structure of DsbDN. DsbDNis a single domain
that consists of four antiparallel sheets, two of which form
a ?-sandwich (Figure 2b). The initial 3 residues (1-3) and
the last 21 residues (124-144) are presumably disordered.
The ?-sandwich consists of two larger antiparallel ?-sheets,
each of which has three strands (?1, ?2, ?8 and ?4, ?10,
?12, respectively). On one edge of the ?-sandwich is an
elongated segment of seven residues (residues 57-63,
between ?5 and ?6) and two ?-sheets, one near the
N-terminus and the other near the C-terminus of DsbDN. The
residues of the elongated segment have the backbone torsion
angles consistent with a ?-strand. Nonetheless, this elongated
segment is not hydrogen bonded to the neighboring ?-strand
(?5); instead, the hydrophobic side chains point into the core
of the protein. At the N-terminal end of DsbDN, the five-
stranded antiparallel ?-sheet is reminiscent of a partial
?-barrel (?6, ?7, ?3, ?10, ?11). At the C-terminal end is
another short antiparallel ?-sheet which consists of three
?-strands (?5, ?9, ?13). The only helical secondary structure
is a four-residue R-helix (R1) located at the beginning of
?-strand 1, which lies against the five-stranded ?-sheet
(Figure 2b). The Ramachandran plot shows that all residues
except two, Q26and D79, are in the allowed region. The model
locates residues Q26 and D79 in a ?-turn, which fits the
electron density map extremely well.
Structural Features of the ActiVe Site. Noticeably, the
structure of DsbDNdisplays two striking features. First, the
two cysteines, C103and C109, which have been shown to be
essential for activity, form a disulfide bond between strand
?10 and strand ?11 (Figure 2b). The distance between the
two Sγ atoms of C103and C109is 2.05 ( 0.05 Å, which is
consistent with a disulfide bond linking the two residues
(normally 2.030 Å) (28). The closest side-chain atoms to
the disulfide bond are the aromatic ring carbons of F70(3.34
Å); OH of Y42(3.55 Å) and OH of Y40(4.37 Å); O of D68
(5.07 Å); NH2of Q101(5.12 Å); and Y71(6.19 Å) (Figure
2c). With the exceptions of F70and Y71, which appear to be
interchangeable in position, all of these residues are con-
served among homologues of DsbDNas shown by the
sequence alignment in Figure 2a. Another region of con-
served residues is found in the first R-helix (R1) near the
active site, where the aromatic ring of F18is pointing toward
the active site contributing to its surrounding hydrophobic
Table 1: X-ray Diffraction Data Collection and Atomic Refinement
for E. coli DsbDN a
resolution range (Å)
unique reflections (total)b
no. of I sites/monomer
phasing resolution range (Å)
figure of meritd
resolution range (Å)
no. of reflections (working/free)
no. of protein atoms
no. of water molecules
bond lengths (Å)
bond angles (deg)
aThe second column refers to the iodide derivative, and the third
column refers to the native protein.bStatistics for the highest resolution
shell are given in parentheses.cRmerge(I) ) ∑hkl[(∑i|Ihkl,j- 〈Ihld〉|)/∑iIhld,i].
dValues are given before density modification at 2.5 Å and after density
modification and phase extension to 1.65 Å resolution.eRcryst) ∑hld||Fo|
- |Fc||/∑hld|Fo|. Rfreewas computed identically, except that 5% of the
reflections were omitted as a test set.
6922 Biochemistry, Vol. 41, No. 22, 2002
Goulding et al.
environment. Notably, nearly all of the conserved residues
are located in or around the active site.
The second striking feature is the loop region between
strands ?6 and ?7, which lies over the disulfide bond between
C103and C109of the active site. Both ?6 and ?7 strands are
located on the five-stranded ?-sheet (appearing in red in
Figure 2b). These residues are also in a conserved region of
the protein (Figure 2a). In particular, the aromatic ring of
F70within this loop region is situated approximately 160°
across from the disulfide bond within this partial ?-barrel.
F70forms an aromatic-sulfur interaction with the Sγ atom
of C109with a 62° angle and may contribute to the stability
of the disulfide bond. This loop lies over the otherwise
exposed active site cysteine residues. As C109is “capped”
by F70, we think it is plausible that the loop protects the active
site from illegitimate redox reactions and refer to this region
as the “cap” of DsbDN. Our model implies that structural
displacement of the cap from the active site must occur prior
to reduction of the disulfide bond of DsbDNby DsbDC.
Further, electron transfer between DsbDNand its protein
substrates DsbC, DsbG, and DsbE may also involve a
mechanism that requires cap displacement. Overall, the
disulfide bond is stabilized by S-π interactions via four
aromatic rings from residues F70, Y40, Y42, and Y71, and C103
has a strong hydrophobic interaction with Y42(29).
The N- and C-termini of DsbDNare located at the opposite
ends of the elongated domain. This is not unexpected as the
N-terminal is the cleavage site for the signal peptide, and
the C-terminal disordered strand is anchored to the hydro-
phobic transmembrane core of DsbD. It is noteworthy that
the electrostatic molecular surface potential at the C-terminal
end of DsbDNis hydrophobic, whereas the surface around
the active site five-stranded partial ?-barrel is negatively
charged with arginine side chains protruding at the surface.
Structurally Related Proteins. A search for structurally
related proteins in the Brookhaven Protein Data Bank,
FIGURE 2: Structure of E. coli DsbDN. (a) Sequence comparison of DsbDNhomologues. The top diagram depicts the E. coli DsbDNprotein
in terms of secondary structure elements (residue numbering is for the E. coli mature sequence). The yellow boxes show conserved residues
among the sequences. The two active site cysteines are indicated with arrows. The sequence alignment was made with CLUSTALW and
BoxShade. (b) Ribbon diagram of DsbDN. DsbDNcontains four ?-sheets and one R-helix, A. The ?-sheets contributing to the immunoglobulin-
like ?-sandwich are colored in yellow and blue. The active site is located on the partial ?-barrel (red); the sulfur atoms of the disulfude
bond (C103and C109) are in yellow, the atoms of the aromatic ring of F70capping C109are shown in white, and all ?-carbon atoms are shown
in white. The ?-sheet located at the C-terminal end of the protein appears in green. This figure was generated using SETOR. (c) The active
site of DsbDNas a view of the final 2Fo- Fcelectron density map. The electron density (blue) is contoured at 1.5σ. The distance between
the two sulfur atoms (yellow) in the disulfide bond form is 2.05 ( 0.05 Å. The figure shows the residues surrounding the active site (Y40,
Y42, D68, F70, and Y71) with oxygen atoms shown in red, nitrogen atoms in blue, and carbon atoms in white. The figure was generated using
Thiol-Disulfide Exchange in an Ig-like Fold
Biochemistry, Vol. 41, No. 22, 2002 6923
performed with the DALI server, identified proteins that are
involved in cell membrane-adhesion and membrane-inter-
acting proteins. The closest structural homologue of DsbDN
was found to be the ap-2 clathrin adaptor R subunit with a
Z-score of 7.1 (30). Also, there are structural similarities to
invasin, immunoglobulins, and major histocompatibility
proteins. The fold of DsbDNis part of the immunoglobulin
superfamily. This family of fold is normally comprised of a
single ?-sandwich, whereas DsbDNcontains two additional
antiparallel ?-sheets. The structural homology search gave
no positive results for proteins that function as a thiol-
disulfide oxidoreductant, as does DsbDN.
Proteins with similar function often share evolutionary
relationships that can be revealed by similarities in their
tertiary structures. Redox reactions involving reversible
thiol-disulfide exchanges are often carried out by dithiols
arranged in a C-X-X-C motif within a thioredoxin-like fold.
Three of the Dsb proteins have been shown to have
thioredoxin folds, including E. coli DsbA (31), DsbC (11),
and the DsbE homologue from Mycobacterium tuberculosis
(12). Sequence homology (24% identity) suggests that the
structure of DsbG is similar to that of DsbC. Presumably,
the C-terminal domain of DsbD also adopts a thioredoxin-
like fold. Conversely, some thioredoxin-like domains do not
carry any redox activity. Protein disulfide isomerase contains
four thioredoxin folds, of which two are redox inactive with
no C-X-X-C motif (32). In E. coli and other organisms,
thioredoxin is maintained reduced by thioredoxin reductase
(TrxB). TrxB is not a member of the thioredoxin superfamily.
Electrons are transferred from the C135and C138of TrxB to
thioredoxin (33). Recycling (reduction) of C135and C138is
achieved intramolecularly by an accessory domain that uses
flavin adenine dinucleotide as a cofactor. Much like TrxB,
DsbDNdoes not belong to the thioredoxin superfamily.
DsbDNobtains reducing equivalent intramolecularly from a
thioredoxin-like domain (DsbDC) and in turn reduces three
thioredoxin-like proteins. To our knowledge, the redox
activity of DsbDNis the first of its kind for an immunoglo-
As for thioredoxin, DsbDN-mediated thiol-disulfide oxi-
doreduction is achieved by formation of mixed disulfide
intermediates between the first cysteine of thioredoxin-like
domains and C109of DsbDN. C109, the active thiolate and its
redox partner, C103, are located on two separate, short,
antiparallel ?-strands (Figure 3a). How does the active site
of thioredoxin-like proteins compare with that of DsbDN?
Active site thiolates of thioredoxin-like proteins are stabilized
by electrostatic interactions with local positively charged
residues and presumably by an R-helix “dipole” (34-36).
In the thioredoxin fold, the C-X-X-C motif is always located
at the N-terminus of the R-helix dipole, and the second
cysteine is the third amino acid of the R-helix (Figure 3b).
The first thiol has an unusually low pKavalue, and it has
been suggested that the electrostatic field associated with
the R-helix pointing with its N-terminus toward the cysteine
residue is in part responsible for lowering the thiol pKavalue
by up to 5 pH units in glutaredoxin and DsbA (35, 36). In
thioredoxin, the thiolate form of C32, which is essential for
the nucleophilic attack, is stabilized by an interaction with
the thiol hydrogen of C35. The preference for deprotonating
C32before C35arises in part because of the greater solvent
exposure of C32, due to its location before the R-helix (34).
Other reports have concluded that redox properties of the
thioredoxin-like protein, i.e., propensity to contain a thiolate
at the active site, may be influenced by the two amino acids
(X-X) surrounded by the two cysteines in the active site (C-
X-X-C) (37, 38). Obviously, the contribution of the helix
dipole does not account for the thiolate form of C109 of
DsbDN. However, a close look at the active site suggests
that, in DsbDN, C109 thiolate is stabilized by a hydrogen-
bonding network and electrostatic interactions strikingly
reminiscent of those of thioredoxins (Figures 2c and 3).
Numerous residues surrounding C103and C109in the structure
are conserved among DsbD homologues. These highly
conserved amino acids are boxed in yellow in Figure 2a.
Assuming that, as for thioredoxins, the structure of reduced
FIGURE 3: Structure of DsbC and DsbDNactive sites. (a) Structure of the DsbDNactive site in the partial ?-barrel (pink). The distance
between D68and C103is 5.1 Å; the hydroxyl group of Y42is positioned between these two residues. (b) Structure of the DsbC active site.
C98and C101are disulfide bonded, bridging the R-helix (cyan) and the loop shown in white. An aspartic residue, D95, is found 6.2 Å away
from C101. The sulfurs atoms are shown in yellow, oxygen atoms in red, nitrogen atoms in blue, and carbon atoms in white. Both active
sites are illustrated in SETOR.
6924 Biochemistry, Vol. 41, No. 22, 2002
Goulding et al.
DsbDNis similar to that of oxidized DsbDN, the closest side-
chain atoms to the cysteines will be the hydroxyl groups of
Y42(3.55 Å), Y40(4.37 Å), and Y71(5.32 Å) (Figure 2c).
The oxygen and hydroxyl group of D68lie at 5.42 Å from
the sulfhydryl of C109. D68is conserved in all but one amino
acid sequence of DsbD homologues (Figure 2a). Helix R1
near the active site is also composed of conserved amino
acids. The conserved sequence E69FYGK73, between strand
?4 and ?5, appears as a loop capping the active site cysteines
C103and C109, and the aromatic ring of F70lies directly over
C109of the disulfide bond (2.48 Å). This capping appears to
protect the disulfide bond, which would be otherwise solvent
exposed and susceptible to nucleophilic attack (Figure 2b).
DsbDNreceives electrons from DsbDCand reduces oxi-
dized DsbC, DsbE, and DsbG. DsbDC, DsbC, DsbE, and
DsbG must all displace the “E69FYGK73” cap, allowing for
preferential deprotonation of C109over C103which remains
buried. The thiolate form of C109 performs a nucleophilic
attack on oxidized DsbC, leading to an intermolecular
disulfide bridge (Figure 4, step 1) (15, 17). We propose that
this event is followed by proton transfer between C103and
the conserved D68and subsequent nucleophilic attack on the
mixed disulfide bridge (Figure 4, step 2). This is followed
by release of reduced DsbC and oxidized DsbDN(Figure 4,
step 3). This model is reinforced by studies using thioredoxin
showing that a conserved aspartic residue (D26) serves as a
proton transfer catalyst for thioredoxin’s C35thiol during the
formation and breakdown of mixed disulfide intermediates
(39). A water molecule may possibly mediate proton transfer
between D26and C35(40). The distance between these two
residues is 6 Å (41), which is very close to the 5.1 Å
measured between the carboxyl group of D68and the sulfur
atom of C103in DsbDN(Figure 3a). Further, we propose that
in DsbDNproton transfer between D68and C103is mediated
by the hydroxyl group of Y42, instead of a water molecule.
The distance between the sulfur atom of C103 and the
hydroxyl group of Y42is 3.51 Å, and it is 2.69 Å between
the hydroxyl group of Y42 and the carboxyl group of D68
(Figure 3a). We note that a conserved aspartic residue, D95,
can also be observed in the structure of DsbC (11), in the
vicinity of the buried C101(Figure 3b). The side chains of
these two residues are separated by 6.2 Å. In the reduced
form of DsbDN, C103 and Y42 must move slightly nearer,
allowing proton transfer to occur between the three residues
(Figure 4, step 2). Also, the hydrophobic environment of
the active site disulfide bond will enhance the transfer of
electrons and protons within the partial ?-barrel. This
suggested catalytic diad between D68and Y42is reminiscent
of the well-known catalytic triad of serine, histidine, and
aspartic acid found in serine proteases. ∆5-3-Ketosteroid
isomerase is another enzyme containing similar residues in
the catalytic site (three tyrosines, two aspartates) (42).
How does DsbDNselect it substrates as compared to DsbC
or DsbA? DsbC shares structural features with DsbA,
including a helical insert in the thioredoxin fold. This helical
insert is smaller in the case of DsbC, and it is not clear what
its role might be (11). Both DsbA and DsbC contain a cleft
to accommodate potential substrates. The architecture of
these clefts is very different. The DsbA cleft is thought to
bind substrate peptide (31) and is absent in DsbC. Unlike
DsbA, DsbC is a homodimer with a much larger, uncharged
broad cleft, lying at the interface of the homodimer (11). It
has been hypothesized that this uncharged cleft permits
binding of unfolded proteins. Loose binding of unfolded
polypeptides may allow the proteins to experience various
conformations for the correct disulfide bonds to form. It is
unclear from our data whether DsbDNinteracts with DsbC
in a similar manner. However, it is possible that the cap
region of DsbDNmay contact this region of DsbC during
catalysis, as would a nonfolded substrate.
In DsbDN, electron transfer between thioredoxin folds is
mediated by an unrelated fold closely related to the immu-
noglobulin fold. The cysteines in this immunoglobulin-like
fold are less promiscuous than those in the thioredoxin fold,
FIGURE 4: Reduction mediated by DsbDN: a possible mechanism. (1) The thiolate form of the solvent-accessible C109performs a nucleophilic
attack on the oxidized target protein, DsbC, leading to an intermolecular disulfide bond between C109of DsbDNand C98of DsbC. (2)
Proton transfer between C103and D68is mediated by the hydroxyl group of Y42and triggers the nucleophilic attack of C103on the mixed
disulfide bridge. (3) Release of reduced DsbC and oxidized DsbDN.
Thiol-Disulfide Exchange in an Ig-like Fold
Biochemistry, Vol. 41, No. 22, 2002 6925
as no mixed disulfides have been observed between DsbDN
and random proteins containing dithiol-disulfide groups
(15). Immunoglobulin-like folds are not uncommon in the
periplasm of E. coli. PapD, a chaperone involved in assembly
of PapK pilin subunits, contains an immunoglobulin fold
(43). The crystal structure of the PapD-PapK chaperone-
subunit complex revealed that the chaperone functions by
donating one ?-strand to complete the immunoglobulin-like
fold of the subunit, via a mechanism termed donor strand
exchange (44). This stable interaction prevents premature
oligomerization of a pilus subunit such as PapK, until export
at the PapC outer membrane protein export site. However,
since interactions between DsbDNand its redox partners are
transient, it is unlikely that strand exchange is required to
bring the two folds together.
In another immunoglobulin-like fold, that of the chaperone
Caf1M from Yersinia pestis, oxidation of two cysteine
residues is thought to activate the chaperone activity of
Caf1M (43, 45, 46). Oxidation is mediated by DsbA (46,
47) on two cysteines located on two successive ?-strands,
as predicted from the structure of PapD. Caf1M is a
periplasmic chaperone involved in the assembly of composite
adhesive pili. Thus, this is an example of a protein predicted
to have an immunglobulin-like fold (Caf1M) that interacts
with a thioredoxin-like protein (DsbA) to become active.
Unlike DsbDN, however, Caf1M is not itself redox active,
as it does not oxidize any substrates subsequently.
Clearly, the DsbDNimmunoglobulin-like fold is a variation
of a common fold that has developed a relatively new
electron-transporting function, perhaps uniquely tailored to
provide reducing equivalents for thiol-disulfide exchanges
coupled to protein folding in bacterial cells. In a pathway of
interacting thioredoxin-like partners, DsbDNis shown to
adopt a completely different three-dimensional structure;
nonetheless, it carries a very similar function and strikingly
an almost identical catalytic site. This is likely an instance
of convergent evolution where two proteins evolved inde-
pendently, quite like the bacterial protease subtilisin and
trypsin proteases which have a very similar geometry of
residues in their active site but no structural similarities (48,
49). The structure of DsbDNand its proposed mechanism
will allow for careful probing of the residues that catalyze
reduction and oxidation of the active site cysteines.
We thank K. Faull for mass spectrometry measurements,
D. Cascio for invaluable help with data collection and general
crystallography, M. Gingery, G. Kleiger, C. Mura, and M.
Apostol for discussion, and O. Schneewind for carefully
reading the manuscript.
1. Kobayashi, T., Kishigami, S., Sone, M., Inokuchi, H., Mogi,
T., and Ito, K. (1997) Proc. Natl. Acad. Sci. U.S.A. 94, 11857-
2. Bader, M., Muse, W., Ballou, D. P., Gassner, C., and Bardwell,
J. C. (1999) Cell 98, 217-227.
3. Thony-Meyer, L. (2000) Biochim. Biophys. Acta 1459, 316-
4. Raina, S., and Missiakas, D. (1997) Annu. ReV. Microbiol.
5. Zapun, A., Missiakas, D., Raina, S., and Creighton, T. E.
(1995) Biochemistry 34, 5075-5089.
6. Bessette, P. H., Cotto, J. J., Gilbert, H. F., and Georgiou, G.
(1999) J. Biol. Chem. 274, 7784-7792.
7. Missiakas, D., Schwager, F., and Raina, S. (1995) EMBO J.
8. Rietsch, A., Bessette, P., Georgiou, G., and Beckwith, J. (1997)
J. Bacteriol. 179, 6602-6608.
9. Russel, M. (1995) Methods Enzymol. 252, 264-274.
10. Holmgren, A. (1995) Structure 3, 239-243.
11. McCarthy, A. A., Haebel, P. W., Torronen, A., Rybin, V.,
Baker, E. N., and Metcalf, P. (2000) Nat. Struct. Biol. 7, 196-
12. Goulding, C. W., Parseghian, A., Gennaro, M., and Eisenberg,
D. (2002) (manuscript in preparation).
13. Chung, J., Chen, T., and Missiakas, D. (2000) Mol. Microbiol.
14. Gordon, E. H., Page, M. D., Willis, A. C., and Ferguson, S. J.
(2000) Mol. Microbiol. 35, 1360-1374.
15. Krupp, R., Chan, C., and Missiakas, D. (2001) J. Biol. Chem.
16. Bader, M. W., Hiniker, A., Regeimbal, J., Goldstone, D.,
Haebel, P. W., Riemer, J., Metcalf, P., and Bardwell, J. C.
(2001) EMBO J. 20, 1555-1562.
17. Katzen, F., and Beckwith, J. (2000) Cell 103, 769-779.
18. Otwinowski, Z. M. W. (1997) Methods Enzymol. 276A.
19. Project, C. C. (1994) Acta Crystallogr. D 50, 760-763.
20. Cowtan, K., and Main, P. (1998) Acta Crystallogr. D 54, 487-
21. Perrakis, A., Morris, R., and Lamzin, V. S. (1999) Nat. Struct.
Biol. 6, 458-463.
22. Jones, T. A., Zou, J. Y., Cowan, S. W., and Kjeldgaard, M.
(1991) Acta Crystallogr. A 47, 110-119.
23. Altschul, S. F., Boguski, M. S., Gish, W., and Wootton, J. C.
(1994) Nat. Genet. 6, 119-129.
24. Thompson, J. D., Higgins, D. G., and Gibson, T. J. (1994)
Nucleic Acids Res. 22, 4673-4680.
25. Holm, L., and Sander, C. (1993) J. Mol. Biol. 233, 123-138.
26. Shindyalov, I. N., and Bourne, P. E. (1998) Protein Eng. 11,
27. Evans, S. V. (1993) J. Mol. Graphics 11, 134-138.
28. Engh, R. A., and a. H., R. (1991) Acta Crystallogr. A 47, 392-
29. Reid, K. S. C., Lindley, P. F., and Thornton, J. M. (1985)
FEBS Lett. 190, 209-213.
30. Traub, L. M., Downs, M. A., Westrich, J. L., and Fremont,
D. H. (1999) Biochemistry 96, 8907-8912.
31. Martin, J. L., Bardwell, J. C., and Kuriyan, J. (1993) Nature
32. Kemmink, J., Darby, N. J., Dijkstra, K., Nilges, M., and
Creighton, T. E. (1997) Curr. Biol. 7, 239-245.
33. Lennon, B. W., Williams, C. H. J., and Ludwig, M. L. (2000)
Science 289, 1190-1194.
34. Dillet, V., Dyson, H. J., and Bashford, D. (1998) Biochemistry
35. Guddat, L. W., Bardwell, J. C., and Martin, J. L. (1998)
Structure 6, 757-767.
36. Schirra, H. J., Renner, C., Czisch, M.,., Huber-Wunderlich,
M., Holak, T. A., and Glockshuber, R. (1998) Biochemistry
37. Mossner, E., Huber-Wunderlich, M., Rietsch, A., Beckwith,
J., Glockshuber, R., and Aslund, F. (1999) J. Biol. Chem. 274,
38. Grauschopf, U., Winther, J. R., Korber, P., Zander, T.,
Dallinger, P., and Bardwell, J. C. (1995) Cell 83, 947-955.
39. LeMaster, D. M., Springer, P. A., and Unkefer, C. J. (1997)
J. Biol. Chem. 272, 29998-30001.
40. Menchise, V., Corbier, C., Didierjean, C., Saviano, M.,
Benedetti, E., Jacquot, J.-P., and Aubrey, A. (2001) Biochem.
J. 359, 65-75.
41. Dyson, H. J., Jeng, M. F., Tennant, L. L., Slaby, I., Lindell,
M., Cui, D. S., Kuprin, S., and Holmgren, A. (1997)
Biochemistry 36, 2622-2636.
42. Kim, S. W., Cha, S.-S., Cho, H.-S., Ha, N.-C., Cho, M.-J.,
Joo, S., Kim, K. K., Choi, K. Y., and Oh, B.-H. (1997)
Biochemistry 36, 14030-14036.
6926 Biochemistry, Vol. 41, No. 22, 2002
Goulding et al.
43. Holmgren, A., and Branden, C. I. (1989) Nature 342, 248-
44. Sauer, F. G., Futterer, K., Pinkner, J. S., Dodson, K. W.,
Hultgren, S. J., and Waksman, G. (1999) Science 285, 1058-
45. MacIntyre, S., Zyrianova, I. M., Chernovskaya, T. V., Leonard,
M., Rudenko, E. G., Zav’Yalov, V. P., and Chapman, D. A.
(2001) Mol. Microbiol. 39, 12-25.
46. Zav’yalov, V. P., Chernovskaya, T. V., Chapman, D. A.,
Karlyshev, A. V., MacIntyre, S., Zavialov, A. V., Vasiliev,
A. M., Denesyuk, A. I., Zav’yalova, G. A., Dudich, I. V.,
Korpela, T., and Abramov, V. M. (1997) Biochem. J. 324,
47. Jacob-Dubuisson, F., Pinkner, J., Xu, Z., Striker, R., Padman-
haban, A., and Hultgren, S. J. (1994) Proc. Natl. Acad. Sci.
U.S.A. 91, 11552-11556.
48. Phadtare, S., Rao, M., and Deshpande, V. (1996) Arch.
Microbiol. 166, 414-417.
49. Liao, D. I., and Remington, S. J. (1990) J. Biol. Chem. 265,
Thiol-Disulfide Exchange in an Ig-like Fold
Biochemistry, Vol. 41, No. 22, 2002 6927