(Received for publication, August 7, 1996, and in revised form, October 7, 1996)
From the Institut für Molekularbiologie und Biophysik, Eidgenössische Technische Hochschule Hönggerberg, CH-8093 Zürich, Switzerland
The catalytic disulfide bond
Cys30-Cys33 of the disulfide oxidoreductase
DsbA from Escherichia coli is located at the amino terminus of an -helix, which has a kink caused by insertion of a tripeptide (residues 38-40). The oxidative force of DsbA
(E
0 =
125 mV) mainly results from the low
pKa of 3.4 of its Cys30 thiol. To
investigate the role of the kink and the electrostatic contribution of
Glu37 and Glu38 to the redox properties of
DsbA, we have characterized a series of DsbA variants (
38-40,
38-40/H41P, E37Q, E38Q, and E37Q/E38Q). In contrast to theoretical
predictions, the redox potentials of the variants are almost unchanged,
and the pKa values of Cys30 do not
differ by more than 0.5 units from that of DsbA wild type. All variants
show the same in vivo activity and dependence of redox
potential on ionic strength as the wild type. The mutations have no
influence on the polypeptide specificity of the protein, which is
independent of the isoelectric point of the polypeptide substrate and
most pronounced at acidic pH. We conclude that neither the kink in the
active-site helix nor Glu37 and Glu38 are
critical for the physical properties of DsbA.
DsbA is a soluble, monomeric 21.1-kDa protein (189 amino acids) that is required for efficient disulfide bond formation in the periplasm of Escherichia coli (1, 2). The enzyme contains a single, catalytic disulfide with the active-site sequence Cys30-Pro31-His32-Cys33. DsbA-catalyzed disulfide bond formation in the periplasm of E. coli involves two steps. First, the disulfide form of DsbA oxidizes reduced, folding substrate proteins. In the second step, DsbA becomes reoxidized by a disulfide exchange reaction with DsbB, a protein located in the inner E. coli membrane (3, 4, 5, 6).
The three-dimensional x-ray structure of oxidized DsbA (7) has revealed
that the enzyme possesses the thioredoxin fold, which is common to all
known structures of disulfide oxidoreductases (8). In addition to its
thioredoxin domain, DsbA possesses a second domain of 76 residues,
which is inserted into the thioredoxin motif and exclusively consists
of -helices (7) (Fig. 1A).
The biophysical characterization of DsbA has shown that the enzyme is the strongest known oxidant within the disulfide oxidoreductase family (12, 13) and oxidizes thiol compounds extremely rapidly (13, 14, 15, 16). In addition, DsbA specifically reacts with reduced, unfolded polypeptides, which can be explained by the presence of a polypeptide binding site (14, 15, 16, 17). DsbA is thus ideally suited as final oxidant of protein disulfides in the periplasm. The physical properties of DsbA are linked to the pKa of the nucleophilic thiol of Cys30, which is extremely low and has a value of 3.5 compared with pKa values of about 9 for normal cysteine side chains (18). On the one hand, the low pKa of Cys30 can explain the fact that the oxidized form of DsbA is about 20 kJ/mol less stable than its reduced form at neutral pH (13, 19) when one assumes that the oxidation of DsbA is nothing but the removal of the negative charge on the Cys30 thiol (18). On the other hand, the low pKa explains the high rate constants of disulfide exchange reactions involving DsbA (18, 20). The interdependence of the redox potential, the stability difference, and the pKa value of Cys30 has recently also been demonstrated by Grauschopf et al. (21) for a series of DsbA variants with amino acid replacements in the dipeptide sequence between the active-site cysteines.
In the known structures of disulfide oxidoreductases, the active-site
disulfide is located at the amino terminus of an -helix (8).
Interestingly, no positively charged residue is located in the vicinity
of the Cys30 side chain in the structure of DsbA that could
explain its low pKa. It was therefore postulated
that the helix dipole as well as the amino acid composition of the
active-site helix may be responsible for stabilizing the negative
charge on Cys30 (18, 22). In addition, electrostatic
calculations suggested that the side chains of Glu37 and
Glu38 as well as the backbone conformation of residues
38-40 have an influence on the redox properties and the low
pKa value of the Cys30 thiol (23, 24).
Comparison of the amino acid sequences and three-dimensional structures
of the active-site helices of DsbA, thioredoxin, and glutaredoxin shows
that the active-site helix of glutaredoxin is straight, whereas the
corresponding helices in DsbA and thioredoxin are kinked (Fig.
1, B and C). In DsbA, the kink is
about 60° and appears to be caused by the insertion of the tripeptide
Glu38-Val39-Leu40. The active-site
helix of thioredoxin has a kink of about 30°, which is caused by a
conserved proline residue at the corresponding position.
In this study, we address the question of whether the residues that
cause the kink in the active-site helix of DsbA are responsible for the
unusual biophysical properties of the enzyme. For this purpose, we have
constructed a DsbA variant in which the tripeptide insertion was
deleted (EVL), as well as a double variant, where His41
was additionally replaced by proline (
EVL/H41P) to mimic the active-site helices of glutaredoxin and thioredoxin, respectively. Since the kink in the active-site helix may also determine the orientation of the catalytic disulfide relative to the peptide binding
site, we have investigated the influence of the mutations on the
polypeptide specificity of DsbA. Since Glu37 and
Glu38 may contribute electrostatically to the
pKa of Cys30 and are part of an unusual
acidic patch on the surface of DsbA (7), we replaced these residues by
glutamines in another set of variants.
1,4-Dithio-DL-threitol
(DTT),1 reduced glutathione (GSH), oxidized
glutathione (GSSG), sodium succinate, amino acids, ampicillin, dithionitrobenzoic acid, 5-bromo-4-chloro-3-indoyl
-D-galactoside, and polymyxin B sulfate were purchased
from Sigma (Deisenhofen, Germany). Guanidinium
chloride (GdmCl) and maltose were from ICN, isopropyl-
-D-thiogalactoside (IPTG) was from AGS GmbH
(Heidelberg, Germany), DE52- and CM52-cellulose were purchased from
Whatman (Maidstone, United Kingdom), and phenyl-Sepharose was obtained from Pharmacia (Uppsala, Sweden). All other chemicals were from Merck
(Darmstadt, Germany) and of highest purity available. The strain THZ2
was kindly provided by T. Zander (University of Regensburg, Germany).
For construction and expression
of the dsbA mutants, the phasmid pDsbA2 was used, where the
dsbA gene is under control of the trc promotor
and its periplasmic expression is inducible by IPTG. pDsbA2 was derived
from the previously described expression plasmid pJW1717 (21) after
deletion of the rop gene.2
Single-stranded DNA was obtained after transformation of E. coli CJ236 and infection with the helper phage M13KO7 (25).
Site-directed mutagenesis was performed according to Kunkel et
al. (26, 27) using the kit supplied by Bio-Rad. The following
oligonucleotides were used (S corresponds to C or G): E37Q and E38Q,
5-C ATT ATC AGA AAT ATG CAG GAC CTS TTS GAA TTG ATA GCA GTG CGG GCA
G-3
;
EVL, 5
-CAC ATT ATC AGA AAT GTG CTC AAA CTG ATA GCA G-3
;
EVL/H41P, 5
-CTT CAC ATT ATC AGA AAT AGG CTC AAA CTG ATA GCA
G-3
.
Mutants were identified by restriction analysis and verified by complete sequencing of the mutated genes using the T7 sequencing kit from Pharmacia.
Expression and Purification of DsbACells of E. coli THZ2 (dsbA::kan,
recA::cam, malF-lacZ102; with
respect to this genotype, THZ2 is isogenic to THZ7; Ref. 21) harboring
the respective expression plasmids were grown in 2 × YT medium
containing 100 mg/liter ampicillin at 37 °C. After induction with 1 mM IPTG at an optical density of 1.0 (546 nm), the cells
were grown for 12 h at room temperature. The cells were harvested
by centrifugation, suspended in ice-cold extraction buffer (10 mM MOPS/NaOH, pH 7.0, 5 mM EDTA, 150 mM sodium chloride, 1 mg/ml polymyxin-B-sulfate; 2 ml/g wet
cells), stirred for 1 h at 4 °C, and centrifuged at 35,000 × g for 30 min. The supernatant (periplasmic extract) was
dialyzed against buffer A (10 mM MOPS/NaOH, pH 7.0) and
applied to a DE52-cellulose column (15 ml). The protein was eluted with
a linear gradient (500 ml) from 0 to 500 mM sodium chloride
in buffer A. Fractions containing DsbA were pooled, adjusted to 1.2 M ammonium sulfate and pH 8.0 by the addition of 4 M ammonium sulfate and 1 M Tris, respectively,
and applied to a phenyl-Sepharose column (15 ml). The column was washed
stepwise with 1.2 and 0.8 M ammonium sulfate in buffer B
(20 mM Tris/HCl, pH 8.0, 100 mM sodium
chloride). The protein was eluted by a linear gradient (125 ml) from
0.8 M ammonium sulfate in buffer B to 20 mM
Tris/HCl, pH 8.0. Fractions containing DsbA were dialyzed against 10 mM acetic acid/NaOH, pH 4.5 (buffer C). The protein was
applied to a CM52 column (15 ml) and eluted by a linear gradient (500 ml) from 0 to 500 mM sodium chloride in buffer C. Fractions
containing pure DsbA were dialyzed against distilled water and stored
at
20 °C.
The molecular mass of each variant was confirmed by electrospray mass
spectroscopy (error 1 Da). The correct amino termini of the variants
E37Q, E38Q, and E37Q/E38Q were also confirmed by Edman sequencing of
residues 1-45.
Protein concentrations were determined by measuring the absorbance at
280 nm (native, oxidized DsbA: 280 = 23,300 M
1 cm
1) (28). All purified DsbA
variants were obtained in the oxidized form, as shown by the lack of
free thiol groups (29).
The dsbA
mutants were tested for their ability to complement the phenotypes of
E. coli dsbA strains, namely the lack of
motility (30) and the ability to express functional
-galactosidase
in the periplasm (1). For all assays, E. coli THZ2 (see
above; Ref. 21) was used. Motility assays were performed at 37 °C on
0.3% agar plates containing ampicillin (100 mg/liter) and either rich
medium (2 × YT) or minimal medium supplemented with all amino
acids except cysteine and cystine (31).
-Galactosidase activity was
tested at 37 °C on indicator plates containing one of the above
media, supplemented with 0.4% maltose, 0.004%
5-bromo-4-chloro-3-indoyl
-D-galactoside, 1.5% agar,
and 100 mg/liter ampicillin.
Far- and near-UV CD spectra were measured at 25 °C on a Jasco 710 CD spectropolarimeter in 0.02- and 1-cm quartz cuvettes, respectively. Protein concentrations of 35 µM in 1 mM sodium phosphate, 10 mM sodium sulfate, pH 7.0, were used. Samples with reduced DsbA contained 100 µM DTT.
GdmCl-induced Unfolding EquilibriaFor unfolding, proteins (50 µM in distilled water) were diluted 1:50 with degassed 100 mM sodium phosphate, pH 7.0, 1 mM EDTA containing different concentrations of GdmCl and incubated at 25 °C for 3 days. In the case of reduced DsbA, DTT (1 mM) was included. For refolding experiments, DsbA was denatured at concentrations of 50 µM in 4 M GdmCl, 100 mM sodium phosphate, pH 7.0, 1 mM EDTA (additionally containing 50 mM DTT in the case of reduced DsbA) for 1 h at room temperature prior to the 1:50 dilution. The transitions were followed by the fluorescence at 365 nm and at 327 nm for oxidized and reduced DsbA, respectively (excitation at 280 nm), using a Hitachi F-4000 spectrofluorimeter. Data were evaluated according to a two-state equilibrium with a six-parameter fit as described previously (19, 32).
Determination of the Redox PotentialThe redox potentials
were determined fluorimetrically by measuring the equilibrium constant
between DsbA and glutathione, assuming no significant equilibrium
concentrations of DsbA/glutathione mixed disulfides (12). DsbA (1 µM) was incubated under a nitrogen atmosphere with 0.1 mM GSSG and different concentrations of GSH (0.003-3
mM) in 100 mM sodium phosphate, pH 7, 1 mM EDTA for 20 h at 25 °C. The relative amount of
reduced DsbA at equilibrium (R) was measured using the
specific DsbA fluorescence at 332 nm (excitation at 280 nm). The
equilibrium constant (Keq) was determined by
fitting the data to the equation, R = ([GSH]2/[GSSG])/(Keq + ([GSH]2/[GSSG])). The redox potential was calculated
using the Nernst equation and a value of 240 mV as the standard redox
potential of glutathione (33).
The pKa values of the thiol group of Cys30 were measured by the increase in absorbance at 240 nm upon formation of the thiolate (18). Oxidized DsbA was used as a reference. For preparation of reduced DsbA, the protein (0.15 mM in distilled water) was reduced by incubation with DTT (0.5 mM) for several minutes. DTT was removed by gel filtration on a PD 10 column (Pharmacia) equilibrated with 10 mM K2HPO4, 10 mM boric acid, 10 mM sodium succinate, 1 mM EDTA, and 200 mM potassium chloride (pH 8). The reference (oxidized DsbA) was subjected to the same gel filtration procedure.
The pH titration was performed with an initial DsbA concentration of 20 µM and an initial volume of 2 ml. The pH was lowered by the stepwise addition of 20-50 µl of 0.1 M HCl, and A240 and A280 were recorded for 10 s, averaged, and corrected for the volume increase. Since A280 proved to be essentially independent of pH the value of (A240, red/A280, red)/(A240, ox/A280, ox) was used to measure the fraction of the thiolate anion. Data were fitted according to the Henderson-Hasselbalch equation.
Determination of Peptide SpecificityAs a measure of the peptide specificity of DsbA, we used the factor by which it oxidizes reduced hirudin and RNaseA faster than the small dithiol DTT. Reduced, unfolded RNaseA and hirudin were prepared essentially as described by Wunderlich et al. (14) and Otto and Seckler (34), except that gel filtration was performed in 1 mM HCl, 200 mM KCl. Oxidized DsbA (1 µM) was reacted with reduced, unfolded RNaseA, hirudin, or DTT (100 µM total concentration of free thiols, pseudo-first-order conditions) at 25 °C in 20 mM K2HPO4, 20 mM boric acid, 20 mM sodium succinate, 200 mM potassium chloride, 1 mM EDTA (adjusted with HCl to pH 4-10).
The reaction was followed by the increase in DsbA fluorescence
(ex = 295 nm;
em
305 nm) on a SX-17MV
stopped-flow reaction analyzer (Applied Photophysics, Leatherhead, UK).
In principle, the fluorescence increase of DsbA only detects the
formation of the DsbA/substrate mixed disulfide, since previous
investigations had shown that the mixed disulfide between glutathione
and DsbA (at Cys30) exhibits the same fluorescence
characteristics as reduced DsbA.3 However,
the intramolecular attack of the mixed disulfide is very rapid.
Therefore, the measured apparent rate constants should be identical
with the overall rate constant of DsbA reduction. As reported
previously (14), the fluorescence increase upon reduction with DTT was
fully consistent with a second-order reaction. This was verified by the
linear dependence of the apparent pseudo-first-order rate constants on
DTT concentration. In the case of reduced polypeptide substrates, we
assumed identical reactivities of all thiols in the beginning of the
reaction (0-20% of final signal change), since DsbA randomly oxidizes
polypeptides (14). To avoid errors due to decreased reactivities of
buried thiols in folding intermediates, the initial velocities were
used to calculate the apparent second-order rate constants for the
reaction with hirudin and RNaseA.
We used the E. coli strain THZ2 lacking the
chromosomal dsbA gene as host for expression and phenotypic
characterization of the dsbA mutants. The in vivo
activity of the dsbA mutants was tested in the absence of
IPTG by the ability of the corresponding expression plasmids to
complement the dsbA phenotypes of E. coli THZ2. The strain THZ2 is immotile due to its inability to
form disulfide bridges in the P-ring protein of the flagellar motor
(30). Cells of E. coli THZ2 transformed with any of the
plasmids containing dsbA mutants showed normal motility on
soft agar comparable with the dsbA wild type gene. THZ2 also
harbors a gene encoding the MalF-
-galactosidase 102 fusion protein
that only results in a Lac+ phenotype in the absence of
disulfide bond formation in the periplasm (1, 3, 21). All
dsbA mutants also restored disulfide bond formation in the
periplasm and caused inactivation of
-galactosidase. The in
vivo activities of the dsbA mutants were not quantified further. The results of the assays were identical on rich medium and on
minimal medium supplemented with all amino acids except cysteine and
cystine. Hence, disulfide compounds of the growth medium can almost
certainly be excluded as oxidants, and the DsbA variants are likely to
be reoxidized by DsbB with an efficiency similar to the wild type.
DsbA WT and its variants were purified from periplasmic extracts by anion exchange chromatography on DE52-cellulose, hydrophobic chromatography on phenyl-Sepharose, and cation exchange chromatography on CM52-cellulose. The yields of the purified, oxidized protein varied from 13 to 19 mg of DsbA/liter of bacterial culture.
UV Absorbance, Fluorescence, and CD Spectra of DsbA VariantsUV absorbance spectra of native oxidized DsbA WT and all variants were identical and showed the same absorbance decrease at 280 nm upon denaturation in GdmCl.
The characteristic fluorescence properties of DsbA were also unchanged
by any of the amino acid replacements. All variants still showed a more
than 3-fold fluorescence increase at 330 nm (ex = 295 nm) upon reduction of the active-site disulfide bridge and a shift in
the emission maximum from 330 to 352 nm upon denaturation with GdmCl
(12).
All DsbA variants were also indistinguishable from DsbA WT with respect to their far-UV and near-UV CD spectra. As observed for DsbA WT (19), the variants show the same characteristic change in their near-UV CD spectra upon reduction (data not shown).
Overall, the spectroscopic properties of all variants indicate that they are properly folded and that conformational changes due to the amino acid replacements do not influence the global structure.
Thermodynamic Stability of DsbA VariantsThe influence of the
amino acid replacements on protein stability
(Gstab) was determined by GdmCl-induced
unfolding/refolding experiments (Fig. 2, Table
I), which were followed by fluorescence measurements.
For DsbA WT and all variants, cooperative and fully reversible
transitions were observed, which were evaluated according to the
two-state model of folding (32). Compared with DsbA WT, which is
destabilized by its disulfide bond, replacement of Glu37
and/or Glu38 of the active-site helix by glutamine results
in further destabilization of the oxidized form and stabilization of
the reduced form. Hence, the energy difference between the oxidized and
reduced form (
Gstab) is increased in these
variants.
|
Interestingly, deletion of the tripeptide EVL does influence the
stability of oxidized and reduced DsbA, but the additional exchange
H41P also leads to an increase in Gstab
(Table I).
The thermodynamic cycle
involving native oxidized, denatured oxidized, denatured reduced, and
native reduced DsbA implies a correlation between the values of
Gstab and the redox potentials of the
variants (13, 19, 35). Hence, all variants should be more oxidizing
than DsbA WT by 5-52 mV.
The redox properties of the DsbA variants were determined by measuring
their redox equilibrium constants (Keq) with
glutathione (12). Interestingly, none of the variants showed the
expected decrease in Keq (Fig. 3,
Table II). The variant E37Q is slightly more reducing
than DsbA WT, whereas the redox potential of E38Q is increased by 11 mV. These effects are qualitatively additive in the double variant
E37Q/E38Q. The redox potentials of EVL and
EVL/H41P are decreased
by 10 and 7 mV compared with DsbA WT, respectively.
|
Since negatively charged residues had been replaced by neutral residues in all these variants, the mutations may have led to a different electrostatic surrounding of the active-site disulfide. Therefore, the differences in their redox potential compared with DsbA WT should decrease upon increasing the ionic strength. In all cases, an increase in the sodium chloride concentration led to more reducing protein (Table III). The salt dependence of Keq is surprisingly low for DsbA WT and not changed in the variants. Therefore, we can exclude the possibility that negatively charged residues in the kink (Glu37, Glu38) have a large electrostatic effect on the active site of DsbA.
|
To test
whether the amino acids in the kink of the active-site helix have any
influence on the pKa value of the Cys30
thiol, the pH dependence of the fraction of the Cys30
thiolate was determined by its absorbance at 240 nm (18). All variants
show slightly higher pKa values of the
Cys30 thiol compared with DsbA WT, with the largest change
of 0.5 pH units in the variants EVL and
EVL/H41P (Fig.
4, Table II).
In general, all mutations had a stronger influence on the stability of oxidized and reduced DsbA compared with changes in the pKa of Cys30. The interdependence of thermodynamic stability, redox potential, and the pKa value of Cys30 thiol therefore does not seem to be that simple, since we would have expected a lowered pKa value for the variants E38Q and E37Q/E38Q due to their increased redox potentials.
pH Dependence of Polypeptide SpecificityDsbA seems to be especially important for disulfide bond formation at acidic conditions, since it acts as oxidant even at pH 4 (14, 15). In addition, DsbA preferably reacts with reduced unfolded polypeptides compared with small organic thiols like the strong reductant dithiothreitol (14), which is presumably caused by a peptide binding site (14, 15, 16, 17). Using stopped-flow techniques we determined the pH dependence of DsbA's polypeptide specificity between pH 4 and 10. We used reduced unfolded RNase A and hirudin as polypeptide substrates. These proteins contain four and three disulfide bridges in their native forms, respectively. These polypeptide substrates were chosen because their isoelectric points (pIs) are strongly different (RNaseA pI = 9.6, (36); hirudin pI = 3.9, (37)). This allowed us to investigate whether DsbA's polypeptide specificity is influenced by the net charge of the substrate. In addition, both substrate proteins do not contain tryptophans. Therefore, oxidation of the substrates was followed by the increase in the tryptophan fluorescence of DsbA and evaluated according to apparent second-order reactions (see "Experimental Procedures" for details).
The rate constants of the reduction of DsbA WT and the variants by DTT
are strongly dependent on pH and increase by 4 orders of magnitude from
pH 4 to 10. In contrast, the rate constants of DsbA's reduction by
both RNaseA and Hirudin increase by about 2 orders of magnitude between
pH 4 and 6 but become nearly pH-independent between pH 6 and 10 (Fig.
5). Hence, the polypeptide specificity of DsbA and all
variants is generally high under acidic conditions, decreases with
increasing pH, and is independent of the pI of the polypeptide
substrate. The negative charge of Glu37 appears to impair
polypeptide binding slightly between pH 6 and 8, because the
polypeptide specificities are significantly increased in the variants
E37Q and E37Q/E38Q. However, none of the variants investigated showed a
significantly altered pH dependence of polypeptide specificity. In
addition, neither DsbA WT nor any of the variants exhibited a
preference for the net charge of the substrate. Overall, we conclude
that residues within or near the kink in the active-site helix of DsbA
are not critical for the peptide specificity of the enzyme.
Until now, biophysical studies on DsbA were restricted to the wild type protein (12, 13, 14, 18, 19) and variants where either the active-site cysteines (38, 39) or the XY dipeptide between them (21)3 had been exchanged.
The aim of this work was to investigate whether the unusual properties of DsbA's active-site disulfide are influenced by other regions of the protein, namely the kink in the active-site helix and residues of the acidic patch (7).
The in vivo activity of all variants reflects that they are catalytically active and can be reoxidized by the transmembrane protein DsbB, a reaction that involves formation of a DsbA/DsbB mixed disulfide (5, 6). The exact mode of interaction between DsbA and DsbB is still unclear, but if a specific recognition between the proteins existed that accelerates their disulfide exchange, Glu37 and residues 38-40 of DsbA would presumably not be involved in the critical contacts with DsbB.
A thermodynamic cycle has been used to explain the high redox potential
of DsbA, which links the difference in stability of oxidized and
reduced DsbA to the stability of the disulfide bond (redox potential)
in the native protein (13, 19). As we observed an increase in
Gstab for all variants investigated, we
expected an increase in their intrinsic redox potentials. On the other hand, the observed increase of the pKa value of the
Cys30 thiol in all variants should lead to a decrease in
their redox potentials (21). Since the variants
EVL,
EVL/H41P,
and E37Q have a lowered redox potential and the redox potential of the variants E38Q and E37Q/E38Q is increased, we could not confirm the
predicted interdependence of redox potential,
Gstab and pKa of the
Cys30 thiol. Obviously, the mechanism underlying the high
redox potential of DsbA is more complex.3 One additional
parameter may be the pKa of the Cys33
thiol, which has been assumed to be constant in previous model studies
(21) but which might also change as a result of the mutations.
Electrostatic calculations had suggested that removal of Glu37 or Glu38 might further increase the redox potential of DsbA by 50 mV (23). However, only a slight increase of 11 mV was observed for the E38Q variant, and no change in redox properties could be detected for E37Q. E38Q and E37Q/E38Q are the first variants of DsbA that are more oxidizing than the wild type protein. Obviously, the charged side chains of Glu37 and Glu38 are not critical determinants of the oxidative force of DsbA.
DsbA accelerates thiol-disulfide exchange reactions even at acidic pH (14, 15) and oxidizes polypeptide substrates faster than small dithiol compounds such as DTT (14). This property can be explained by an additional noncovalent interaction between DsbA and the polypeptide substrate and suggests the existence of a polypeptide binding site (7, 16, 17). We show here that this peptide specificity is independent of the isoelectric point of the polypeptide substrate but strongly dependent on pH, with a higher specificity at low pH. Since only the thiolate acts as a nucleophile, the rate constants of thiol-disulfide exchange reactions strongly increase with increasing pH below the pKa of the attacking thiol (20). A higher substrate specificity with decreasing pH thus means that the rate of oxidation of polypeptide substrates is less pH-dependent than the rate of oxidation of small thiol compounds. This fact might become important in vivo for disulfide bond formation at acidic growth conditions. The growth limit for E. coli under acidic conditions is pH 4.4 (40), where spontaneous disulfide exchange reactions no longer occur.
Since the polypeptide specificity is not strongly influenced in any of the variants tested, we can almost certainly exclude that the kink is part of the proposed peptide binding site in DsbA.
It is generally accepted that the interactions of -helix dipoles
with charged residues are of functional and structural importance (41),
and they have been shown to influence protein stability (42, 43, 44, 45). The
partial positive charge at the amino terminus of the active-site
helices of disulfide oxidoreductases has recently been claimed to
stabilize the negative charge of the nucleophilic cysteine thiolate and
thereby lower its pKa (22). Removal of the
tripeptide Glu38-Val39-Leu40 and
the additional replacement H41P was not expected to cause exactly the
conformational change of the active-site helix of DsbA to the
conformations found in glutaredoxin and thioredoxin, respectively.
However, we at least expected changes in the physicochemical properties
of DsbA in the direction of glutaredoxins and thioredoxins. Indeed, we
observed slightly increased pKa values of Cys30 thiols and slightly lowered redox potentials (Table
II) for both variants. These differences are small, however, compared
with the effects reported for mutations in the XY dipeptide
in DsbA (21). It thus appears that residues in the amino-terminal part of the active-site helices of disulfide oxidoreductases have a much
stronger influence on their redox properties than residues in the
carboxyl-terminal part. The almost unchanged properties of the
EVL
variant and the
EVL/H41P double variant are in accordance with the
fact that the tripeptide insertion is absent and that a proline
equivalent to residue 41 in DsbA is present in the DsbA-related protein
from Vibrio cholerae (46).
Since the -helix dipole causes a partial negative charge at the
carboxyl terminus of the helix, the charged residues Glu37
and Glu38 at the carboxyl terminus of the first part of the
active-site helix should be energetically unfavorable. In addition,
both residues are part of the acidic patch and are thus located in a
strongly, negatively charged environment (7). The increased stability of the reduced variants E37Q, E38Q, and E37Q/E38Q compared with DsbA WT
is in accordance with this consideration. However, the origin of the
decreased stability of their oxidized forms remains unclear. Overall,
the stability changes are caused by changes in cooperativity rather
than by changes in the midpoint of transition. In principle,
replacement of energetically unfavorable, charged residues that cause
repulsive interactions within the protein might lead to a more compact
structure and thus to a higher cooperativity of the unfolding
transitions compared with DsbA WT. However, this was only observed for
the reduced forms of the variants. In contrast, the cooperativities
were unexpectedly decreased for the oxidized forms. This could mean
that the assumed two-state model of folding is no longer strictly valid
for the oxidized variants (47). Until now, only the three-dimensional
structure of DsbA's oxidized form is known (7). Therefore, no
information on structural changes that occur upon reduction of DsbA is
available. We speculate that the opposite effects of the amino acid
replacements on the unfolding/refolding transitions of DsbA might
indeed reflect conformational differences in the oxidized and reduced
form. However, three-dimensional structures of the oxidized and reduced
forms of E. coli glutaredoxin (48), E. coli
thioredoxin (49), and human thioredoxin (50, 51) have revealed that
only minor structural differences exist between the redox states of
these enzymes.
In conclusion, we have shown that neither Glu37, Glu38, nor the kink itself has a large influence on the physicochemical properties of DsbA, such as the pKa value of the Cys30 thiol or the redox potential. It appears that the properties of DsbA are largely determined by the PH dipeptide in the active-site CPHC, which may strongly influence the helix-dipole. We could demonstrate that Glu37 and Glu38, which are part of an unusual acidic patch on the surface of DsbA, do not contribute to its peptide specificity and are not critical for reoxidation of DsbA by DsbB.
We thank M. Huber-Wunderlich and A. Jacobi for fruitful discussions, V. Eggli for help during construction of the DsbA mutants, G. Frank for Edman sequencing, and P. James for performing electrospray mass spectroscopy.