HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 45, 43050-43057, November 8, 2002

This Article Abstract Full Text (PDF) Supplemental Data All Versions of this Article: 277/45/43050    most recent M207638200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Philipps, B. Articles by Glockshuber, R. Search for Related Content PubMed PubMed Citation Articles by Philipps, B. Articles by Glockshuber, R. Social Bookmarking                      What's this?

Randomization of the Entire Active-site Helix 1 of the Thiol-disulfide Oxidoreductase DsbA from Escherichia coli^*,

Björn Philipps and Rudi Glockshuber^§

From the Institut für Molekularbiologie und Biophysik, Eidgenössische Technische Hochschule Hönggerberg, CH-8093 Zürich, Switzerland

Received for publication, July 29, 2002, and in revised form, August 16, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

DsbA from Escherichia coli is the most oxidizing member of the thiol-disulfide oxidoreductase family (E_o' = 122 mV) and is required for efficient disulfide bond formation in the periplasm. The reactivity of the catalytic disulfide bond (Cys³⁰-Pro³¹-His³²-Cys³³) is primarily due to an extremely low pK_a value (3.4) of Cys³⁰, which is stabilized by the partial positive dipole charge of the active-site helix 1 (residues 30-37). We have randomized all non-cysteine residues of helix 1 (residues 31, 32, and 34-37) and found that two-thirds of the resulting variants complement DsbA deficiency in a dsbA deletion strain. Sequencing of 98 variants revealed a large number of non-conservative replacements in active variants, even at well conserved positions. This indicates that tertiary structure context strongly determines -helical secondary structure formation of the randomized sequence. A subset of active and inactive variants was further characterized. All these variants were more reducing than wild type DsbA, but the redox potentials of active variants did not drop below 210 mV. All inactive variants had redox potentials lower than 210 mV, although some of the inactive proteins were still re-oxidized by DsbB. This demonstrates that efficient oxidation of substrate polypeptides is the crucial property of DsbA in vivo.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The thiol-disulfide oxidoreductase DsbA from Escherichia coli is a monomeric protein of 189 amino acids, which is required for disulfide bond formation in the bacterial periplasm (1, 2). The enzyme contains a catalytic disulfide bridge, which is located at the N terminus of the active-site helix 1 with the sequence Cys³⁰-Pro³¹-His³²-Cys³³-Tyr³⁴-Gln³⁵-Phe³⁶-Glu³⁷. DsbA-mediated disulfide bond formation in the periplasm involves oxidation of reduced, newly translocated substrate polypeptides by DsbA, followed by reoxidation of DsbA by the inner membrane protein DsbB. DsbB is in turn reoxidized by molecular oxygen through ubiquinone and terminal cytochrome oxidases (3, 4).

The three-dimensional structure of both oxidized and reduced DsbA has been solved by x-ray crystallography and NMR spectroscopy (5-7). The enzyme possesses a thioredoxin fold, which is common to most thiol-disulfide oxidoreductases and contains the active-site disulfide (8). In addition to the thioredoxin domain, DsbA possesses a second, purely -helical domain, which is inserted into the thioredoxin domain (5). The structural differences between oxidized and reduced DsbA are locally restricted to the active-site region, and there is no major domain motion related to the redox state of DsbA (6, 7).

Biophysical studies on DsbA have shown that the protein is the strongest known oxidant within the disulfide oxidoreductase family (9, 10), and it oxidizes thiol compounds randomly and extremely rapidly (11-13). The pK_a of the nucleophilic, active-site thiol of Cys³⁰ has an extremely low value of 3.4 (compared with 9 to 10 for normal cysteine side chains) (14). The low pK_a can, at least qualitatively, explain the high redox potential of the enzyme, as well as the fact that the reduced state of DsbA is thermodynamically more stable than the oxidized state (10, 14, 15). The x-ray and NMR structures of reduced DsbA indicate that the partial positive charge from the dipole of the active-site helix 1 stabilizes the thiolate anion of Cys³⁰, which is the most N-terminal residue of -helix 1. In addition, His³² presumably contributes about one pK_a unit to the lowered pK_a of Cys³⁰, possibly by formation of a hydrogen bond between the Cys³⁰ thiolate and N1 (7, 16-18). In addition, possible hydrogen bonds with the thiol of Cys³³, the amide hydrogen of His³², and the amide hydrogen of Cys³³ further stabilize the Cys³⁰ thiolate (7).

The redox potentials of the different members of the thiol-disulfide oxidoreductase family vary over a wide range, from 122 mV for the most oxidizing member DsbA to 270 mV for thioredoxin, the most reducing member (8-10, 19, 20). Previous mutagenesis studies have shown that the Xaa-Xaa dipeptide sequence between the active-site cysteines is a critical determinant of the redox properties of thiol-disulfide oxidoreductases. The first experiment in this direction was the replacement of the Gly-Pro dipeptide in E. coli thioredoxin with that of eukaryotic protein disulfide isomerase (PDI)¹ (Gly-His). The mutation shifted the redox properties of thioredoxin by 35 mV toward the redox potential of PDI (20). Analogous results were reported for the catalytic a-domain of PDI, in which the reverse exchange of the dipeptide Gly-His against the thioredoxin dipeptide Gly-Pro leads to a decrease in redox potential by 66 mV (18). For thioredoxin variants carrying the dipeptide Gly-His (like PDI), Pro-His (like DsbA), and Pro-Tyr (like glutaredoxin), the redox potential increases by 49, 66, and 75 mV (21). Conversely, lowered redox potentials were obtained for analogous dipeptide variants of DsbA (22). Random mutagenesis of the dipeptide in DsbA, thioredoxin, or PDI, combined with functional screening or selection methods, has also clearly demonstrated the functional importance of the Xaa-Xaa dipeptide (23-25).

In the present study, we have extended previous randomization experiments on the active-site region of DsbA to gain insights into the sequence requirements for the entire active-site helix of the enzyme. To this end, we have simultaneously randomized all six non-cysteine residues of the active-site helix 1, i.e. residues 31, 32, and 34-37 (see Fig. 1), and screened overproducing bacterial clones for biologically active and inactive variants. From the pool of 98 different sequences (see Supplemental Material, Table B), nine different variants were chosen for further in vitro characterization (see Table I). The redox potentials, pK_a values of Cys³⁰, reactivity toward the reduced polypeptide substrate hirudin, and the reoxidation by DsbB were measured for these variants and compared with the properties of wild type DsbA.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Materials-- Dithiothreitol, GSH, GSSG, ampicillin, 5,5'-dithiobis(2-nitrobenzoic acid), 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-Gal), and polymyxin B sulfate were purchased from Axon Labs AG (Baden, Switzerland). DE52-cellulose were purchased from Whatman (Maidstone, United Kingdom), and the Phenyl Superose HR 10/10 column was obtained from Amersham Biosciences. All other chemicals were from Merck and were of the highest purity available. DNA oligonucleotides were from Microsynth (Balgach, Switzerland). Polyclonal rabbit anti-DsbA antibodies were obtained from Drs. Rosskopf and Fraefel (Zürich, Switzerland). E. coli THZ2 (dsbA::kan, recA::cam, malF-lacZ102) was kindly provided by Dr. James C. A. Bardwell (University of Michigan).

Random Mutagenesis and Screening-- The plasmid pDsbA3-CC30/33AA for periplasmic production of the inactive DsbA variant Cys³⁰  Ala-Cys³³  Ala was derived after amplification of the mutant gene from a previously described plasmid (26) and replacement of the dsbA wild type gene in pDsbA3 (27) via the restriction sites NdeI and BamHI. Plasmids pDsbA3 and pDsbA3-CC30/33AA thus both contain the trc promoter and lac operator sequence for inducible gene expression and an f1 replication origin. Random mutagenesis of the active-site helix was performed according to Kunkel et al. (28), using single-stranded, uridinylated plasmid DNA. For identification of active variants after random mutagenesis, pDsbA3-CC30/33AA was used as template vector to eliminate active wild type background. For analogous reasons, we used the wild type expression plasmid pDsbA3 as mutagenesis template for identification of biologically inactive DsbA variants. In both experiments, the mutagenesis primer 5'-AGA AAT ATG CAG AAC TTC NNN NNN NNN NNN GCA NNN NNN GCA GAA GAA AGA GAA AAA CTC-3' was used.

Screening for DsbA⁺ and DsbA phenotypes was performed in the absence of isopropyl-1-thio--D-galactopyranoside with (i) a blue-white screening assay based on oxidative inactivation of periplasmically orientated -galactosidase (1) and (ii) by motility assays based on DsbA-dependent assembly of intact flagellae (29) as described (27) using E. coli THZ2 (dsbA::kan, recA::cam, malF-lacZ102) as expression strain. Periplasmic production of DsbA variants was analyzed by SDS-PAGE of periplasmic extracts, followed by immunoblotting and specific staining with polyclonal rabbit anti-DsbA antibodies (27). The sequences of the active and inactive variants were determined by dideoxynucleotide sequencing after restriction analysis of the corresponding expression plasmids.

The mutagenesis yield for generation of active variants was quantified after growth of the entire pool of DsbA-producing E. coli cells in LB-amp medium, preparation of the plasmid library, and complete digestion of the library with NsiI. The mutagenesis primer eliminates a single NsiI site in pDsbA3-CC30/33AA. The yield of mutagenesis could thus be determined after separation of the restriction digest products on an agarose gel and densitometric analysis of ethidium bromide-stained bands.

Cloning of Cytoplasmic Expression Plasmids and Purification of Selected DsbA Variants-- As previous results have shown that cytoplasmic expression of DsbA results in much higher production yields compared with periplasmic expression (27), the genes encoding the nine different DsbA variants selected for further characterization were amplified by PCR with the oligonucleotides 5'-GCG ACT GGA ATT CCA T AT GGC GCA GTA TGA AGA TG-3' (NdeI site is underlined) and 5'-G GGC GCG TGG GGA TCC-3' (BamHI site is underlined) and cloned via NdeI and BamHI into the vector pDsbA_cyto (27). The cytoplasmic production of the DsbA variants was performed essentially as described previously (27). E. coli BL21(DE3) harboring the corresponding derivative of pDsbA_cyto were grown at 37 °C in rich medium (20 g/liter tryptone, 10 g/liter yeast extract, 5 g/liter NaCl, 20 ml/liter glycerol, 50 mM K₂HPO₄, 10 mM MgCl₂, 10 g/liter glucose, 100 µg/liter ampicillin) to an A₆₀₀ of 1.0 and induced with 1 mM isopropyl-1-thio--D-galactopyranoside. After further growth for 5 h at 37 °C the cells were harvested by centrifugation, suspended in 12 ml of extraction buffer (50 mM Tris/HCl, pH 8.0, 1 mM MgCl₂, 1 mg/liter lysozyme) per liter of bacterial culture at 0 °C and disrupted by sonification. The soluble fraction of the extract was dialyzed against 10 mM MOPS/NaOH, pH 7.3, applied to a DE52-cellulose anion exchange column, and bound proteins were eluted by a linear gradient from 0 to 400 mM NaCl. Fractions containing the DsbA variants were pooled, mixed with 4 M ammonium sulfate to a final concentration of 0.8 M ammonium sulfate, and applied to a phenyl-Sepharose column equilibrated with 0.8 M ammonium sulfate, 10 mM MOPS/NaOH, pH 7.3. The proteins were eluted with a linear gradient from 0.8 to 0 M ammonium sulfate. Fractions with pure DsbA variants were pooled, dialyzed against distilled water, and concentrated.

Protein concentrations were measured by the specific absorbance at 280 nm (30), and the correct mass of the variants was verified with MALDI-TOF mass spectrometry. All variants were obtained in the oxidized form after purification, as shown by the lack of free thiol groups (31). The only exception was variant 71, which has an additional free cysteine at position 32 and yielded one molar equivalent of thiol per polypeptide (see Table I). Far- and near-UV CD spectra were measured at 25 °C on a Jasco 715 CD spectrapolarimeter as described (32).

Redox Potentials of Active-site Helix Variants-- The redox potentials of the variants were determined by measuring their equilibrium constants with glutathione, using the redox state-dependent fluorescence of the DsbA variants at 325 nm and assuming no significant equilibrium concentrations of DsbA-glutathione mixed disulfides (9). Measurements were performed in 100 mM sodium phosphate, pH 7.0, 1 mM EDTA, containing 0.1 mM GSSG and 0 to 2 mM GSH at 25 °C. A value of 240 mV was used for the standard redox potential of glutathione (33) to deduce E_o' values.

pK_a Values of the Active-site Variants-- The pK_a value of Cys³⁰ in the different DsbA variants was measured by the decrease in absorbance at 240 nm because of protonation of the Cys³⁰ thiolate (14, 34). Titrations were performed at protein concentrations of 3.5 µM in 10 mM sodium phosphate, 10 mM sodium citrate, 10 mM boric acid containing 200 mM KCl and 3 mM dithiothreitol by stepwise addition of small portions of 0.1 M HCl. Absorbance values were corrected for the volume increase. Identical measurements were performed with the corresponding oxidized proteins in the absence of dithiothreitol and used for baseline correction.

Hirudin Refolding-- Refolding of reduced hirudin by stoichiometric amounts of the oxidized DsbA variants was performed at pH 7.0 and 25 °C. Reduced, unfolded hirudin (28 µM) was prepared and mixed with oxidized DsbA (84 µM) as described (11, 27). Aliquots of 120 µl were removed after different reaction times and quenched with formic acid (final concentration, 10% (v/v) and pH < 2). The reaction products were separated by reversed-phase HPLC on a 218TP54 C18 column (Vydac, Hesperia, CA) at 55 °C with a 20 to 24% (v/v) acetonitrile gradient in 0.1% (v/v) trifluoroacetic acid, and hirudin folding intermediates were detected by their absorbance at 230 nm.

Oxidation of the Reduced Variants by DsbB in Vitro-- Membranes containing DsbB, ubiquinone, and terminal oxidases were prepared from E. coli JCB819 cells overexpressing DsbB as described (3, 35). The DsbB-catalyzed oxidation of reduced DsbA was followed by the decrease in specific DsbA fluorescence at 330 nm (excitation at 280 nm). The reaction was performed at 25 °C in a volume of 1 ml in 50 mM Tris/HCl, pH 8.0, 300 mM NaCl containing different concentrations of the reduced DsbA variants (1-40 µM) as described (35). The reaction was started by addition of DsbB-containing membranes to reduced DsbA.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Active-site Helix Randomization and Screening for Active and Inactive Variants-- Fig. 1 shows the active-site helix 1 of DsbA comprising residues 30-37 in the context of the tertiary structure of the oxidized protein. In both oxidized and reduced DsbA, the first turn of the active-site helix is a 3₁₀ helix, which then continues as a regular -helix to residue 37 (5-7). The 3₁₀-helical segment contains Pro³¹, which is invariant in the DsbA protein family. Besides the conserved His³², the residues in the C-terminal position of the helix are also rather well conserved. Residue 34 is usually either Tyr or Ala, residues 35 and 37 are always polar or charged, and residue 36 is exclusively a hydrophobic or aromatic amino acid (cf. Supplemental Material, Table A).

View larger version (36K): [in this window] [in a new window]   Fig. 1.   Three-dimensional structure of DsbA and conservation of residues in the active-site helix 1. A ribbon representation of oxidized E. coli DsbA (5, 7). The thioredoxin-like domain is shown in blue, and the -helical domains are in red. The segment comprising the active-site helix 1 (residues 30-37) is depicted in green, and the side chains of Cys30 and Cys33 are shown in ball-and-stick representation. The figure was created with the program MOLMOL (47).

To explore the available sequence space for the active-site helix, we randomized all non-cysteine residues of segment 30-37 by Kunkel mutagenesis and analyzed the resulting variants for biological activity in an E. coli dsbA deletion mutant. For this purpose, we used two independent DsbA complementation assays based on (i) oxidative inactivation of periplasmically oriented -galactosidase through active DsbA (1, 24) and (ii) DsbA-dependent flagellar assembly, which results in immotile cells in the absence of active DsbA (29). As we intended to look for both active and inactive DsbA variants, randomization of segment 30-37 was performed both starting from the wild type background and from a catalytically inactive variant in which both active-site cysteines were replaced by alanine residues. This allowed quantification of the mutagenesis yield and determination of the fraction of active and inactive variants after randomization (see "Experimental Procedures"). Despite the well conserved sequence of the active-site helix 1, 70% of all mutant proteins with randomized residues 31, 32, and 34-37 proved to be biologically active in complementation assays after mutagenesis of the inactive dsbA gene. Overall, we characterized 98 mutant proteins from arbitrarily selected single colonies further with respect to primary structure and periplasmic expression levels (assayed by immunoblots after SDS-PAGE of periplasmic extracts). We arbitrarily selected 69 variants, which were active in both DsbA complementation assays (variants 1-69) and 25 variants that proved inactive in both assays (variants 74-98, obtained from mutagenesis of the wild type dsbA gene) (see Supplemental Material, Table B). Moreover, we identified a third class of DsbA variants (variants 70-73), which we termed "semi-active." These variants did not reconstitute motility of the dsbA deletion strain but were active according to the -galactosidase assay (see Supplemental Material, Table B).

Previous studies on circularly permuted DsbA variants had shown a reasonably good correlation between thermodynamic stability of DsbA variants and periplasmic expression levels. This is most likely caused by a higher sensitivity of less stable proteins toward periplasmic proteases (27). Analysis of the expression levels of the 69 active variants showed that 48 active variants were still produced at levels comparable with wild type DsbA (see Supplemental Material, Table B). In the case of the inactive variants, the majority of mutant proteins (variants 82-98) were obtained at lower levels, although eight inactive variants were still produced at wild type level. Overall, the periplasmic expression levels qualitatively indicate that many of the active variants have stabilities similar to that of the wild type protein, whereas most of the inactive variants are thermodynamically less stable (see Supplemental Material, Table B).

Sequence Analysis of Active and Inactive Variants-- Supplemental Material, Table B shows the active-site helix sequences of all 98 DsbA variants selected for this study. In the case of the 69 active variants, the most unexpected result is the fact that practically any residue except proline is tolerated at positions 34-37. The same holds true for position 32, where proline is additionally tolerated. This demonstrates that tertiary structural context strongly controls folding of helix 1, independent of helix propensity, the degree of solvent accessibility of individual residues, and the size and charge of the residues introduced by mutagenesis. Most amino acids also appear to be tolerated at position 31. However, we never found arginine or lysine residues at this position. In contrast, basic residues were specifically enriched at position 31 in the set of inactive variants. Interestingly, all sequences of semi-active variants also had a basic residue at position 31 (see Supplemental Material, Table B). The variability of the dipeptide at positions 31 and 32 in the active variants is in accordance with previous random mutagenesis studies on these two residues (24).

Eighteen of the 25 sequenced, inactive DsbA variants had one or two proline residues within segment 34-37 (see Supplemental Material, Table A). Here, incorporation of the -helix breaker proline is certainly the most prominent mutation leading to an inactive enzyme. Nevertheless, two variants (20 and 47) from the set of active DsbA proteins also contained prolines in segment 34-37 (see Supplemental Material, Table B). Consequently, even larger conformational changes caused by prolines in this segment can be accommodated by the DsbA protein and do not necessarily lead to loss of biological activity.

Spectroscopic Characterization of Selected DsbA Variants-- For further characterization of the mutant proteins, we arbitrarily selected three active variants (1-3) from the set of 48 well expressed active proteins (see Supplemental Material, Table B) and the two well expressed, semi-active variants (70 and 71). From the set of inactive variants, we arbitrarily selected four moderately expressed variants (82-85) (see Table I and Supplemental Material, Table B). The genes encoding these nine variants were then cloned into a cytoplasmic expression plasmid as cytoplasmic expression gives much higher yields of recombinant DsbA compared with periplasmic production (27). The nine variants were then purified to homogeneity from cytoplasmic fractions by conventional chromatography (see "Experimental Procedures").

View this table: [in this window] [in a new window]   Table I Primary structures of E. coli DsbA variants with randomized active-site helix 1 that were arbitrarily selected after expression analysis for biochemical characterization

The far-UV CD spectra of the oxidized and reduced variants proved to be extremely similar to the spectra of oxidized and reduced wild type DsbA (Fig. 2). Consequently, the overall fold of the variants, even of the inactive variants, is essentially unchanged. The overall wild type-like tertiary structure of the mutant proteins was also confirmed by fluorescence spectroscopy. Specifically, the strong tryptophan fluorescence increase of DsbA upon reduction of the catalytic disulfide bond (9, 10, 32) was retained in all variants. This is a good indication for the intactness of the tertiary structure, as fluorescence quenching by the catalytic disulfide does not occur via direct contact between a tryptophan and the disulfide but through a complex through-space energy transfer mechanism that requires the correct relative orientations and local environments of the two tryptophan residues in the protein (32, 36). Overall, both the CD and fluorescence data indicate that even the biologically inactive DsbA variants 82-85 have wild type-like tertiary structures.

View larger version (39K): [in this window] [in a new window]   Fig. 2.   Comparison of the near-UV (top) and far-UV (bottom) circular dichroism spectra of the oxidized (left) and reduced (right) variants of DsbA. Spectra were recorded at pH 7.4 and 25 °C. Spectra of active, semi-active, and inactive variants are depicted with blue, green, and red lines, respectively, and spectra of wild type DsbA are shown with black lines. In the case of the far-UV CD data (bottom panels), the spectra of all nine purified variants proved to be identical within experimental error to the spectra of oxidized and reduced wild type DsbA. For this reason, only one representative spectrum is shown for the active, semi-active, and inactive variants (variants 1, 70, and 82, respectively).

Redox Potentials of the Variants and pK_a Values of Cys³⁰-- There is a theoretical and experimentally established correlation among the redox potentials of disulfide oxidoreductases, the pK_a of their active-site cysteine, and the difference in thermodynamic stability between the oxidized and reduced state of the enzymes (14, 19, 21, 22, 24, 37, 38). Moreover, the redox potential of the catalyst of disulfide bond formation in the bacterial periplasm and the rate of substrate oxidation proved to be critical factors for the function of DsbA in vivo (35). We first determined the intrinsic redox potentials of the purified variants at pH 7.0 and 25 °C by measuring their equilibrium constants with glutathione (see Fig. 3A and Table II). As observed for practically all DsbA variants investigated so far (22, 24, 27, 34, 39), all nine active-site helix variants were more reducing than wild type DsbA, with redox potentials between 165 and 222 mV. For comparison, a value of 123 mV was determined for the wild type protein (Table II). In contrast, all biologically inactive variants had redox potentials below 210 mV. This is in good agreement with DsbA complementation studies with periplasmically produced thioredoxin variants, where only thioredoxins with redox potentials above 220 mV could complement the DsbA deficiency (35).

View larger version (20K): [in this window] [in a new window]   Fig. 3.   Redox properties of DsbA variants after active-site helix randomization. A, determination of the intrinsic redox potentials of purified variants through equilibration of the proteins at different [GSH]2/[GSSG] ratios (cf. Table II). The fractions of oxidized and reduced protein were measured through the redox state-dependent tryptophan fluorescence (cf. Table II), and the redox transitions were normalized. As an example, the redox equilibria with glutathione of the active variant 3, the semi-active variant 70, and the inactive variant 83 are shown. For numbering and primary structure of the variants, see Table I. B, plot of measured equilibrium constants with glutathione (Keq) against the pKa of Cys30 for the purified active-site variants of DsbA. The solid and dashed lines indicate theoretically expected correlations based on the Brønsted theory of disulfide exchange reactions (37). The solid line represents a model that assumes a constant pKa of 11.7 for the buried Cys33 thiol, whereas the dashed line is based on a model which assumes pKa (Cys30) + pKa (Cys33) = 16.9 (38). Variants 1 and 70 deviate most strongly from the theoretical models.

We also determined the pK_a values of the nucleophilic, active-site cysteine (Cys³⁰) thiols for all variants, using the loss of thiolate anion absorbance at 240 nm upon protonation (14). As expected from the lowered redox potentials of the mutant proteins compared with the wild type, all variants showed increased pK_a values for Cys³⁰ (see Fig. 3B and Table II), ranging between 4.9 and 6.5 (compared with 3.4 for wild type DsbA). The Cys³⁰ pK_a values are thus still at least 3 pK_a units below the pK_a of a normal cysteine (9-9.5), again indicating that all variants have an intact tertiary structure. However, the observed Cys³⁰ pK_a values of the variants correlate only weakly with the corresponding equilibrium constants with glutathione, with the variants 1, 70, and 82 deviating most strongly from theory (Fig. 3B). A possible reason for this weak correlation is electrostatic effects on the pK_a value of Cys³⁰ at pH 7.0 resulting from new, charged residues introduced into segment 30-37 (cf. Table I).

Stoichiometric Oxidation of the Substrate Polypeptide Hirudin-- To probe the oxidative potential of the purified active-site helix variants of DsbA toward polypeptide substrates, we investigated their ability to oxidize the reduced thrombin inhibitor hirudin (65 amino acids, 3 disulfide bonds in the native state), a well established DsbA substrate (11). Reduced hirudin was mixed with 3 molar equivalents of the oxidized DsbA variants at 25 °C at pH 7.0, reactions were acid-quenched after different incubation times, folding intermediates were separated by reversed-phase HPLC, and the half-lives for refolding were estimated from the HPLC profiles (11, 27) (see Fig. 4 and Table II). In the case of wild type DsbA, this reaction is characterized by extremely rapid, random oxidation of hirudin, followed by slow isomerization of non-native disulfide bonds catalyzed by reduced DsbA (11-13). All active and semi-active variants yielded refolding rates between 10 and 90% of wild type activity and led to quantitative refolding, and the active variant 1 and the semi-active variant 70 proved to be even more efficient than DsbA wild type (Fig. 4) (see "Discussion" for details). In contrast, the inactive variants 83-85 were not capable of oxidizing hirudin efficiently, and the inactive variant 82 showed only about 10% of wild type activity (see Table II and Fig. 4). These data suggest that at least one of the reasons for the lack of biological activity of variants 82-84 is inefficient oxidation of polypeptide substrates.

View larger version (33K): [in this window] [in a new window]   Fig. 4.   Kinetics of the stoichiometric oxidation of reduced hirudin (28 µM) with 3 molar equivalents of purified DsbA variants at pH 7.0 and 25 °C. Reactions were quenched with acid after different reaction times, and disulfide folding intermediates of hirudin were separated by reversed-phase HPLC using a water-acetonitrile gradient. R and N denote the HPLC retention times of completely reduced hirudin (six sulfhydryl groups) and native hirudin (three disulfide bonds), respectively.

Interactions between the Active-site Variants of DsbA and DsbB-- Finally, we investigated the ability of the purified active-site helix variants of DsbA of being reoxidized by DsbB, using DsbB-containing membrane fractions of E. coli for catalysis of air oxidation of the reduced variants. Reactions were followed by the decrease in DsbA fluorescence, as established previously by Bader et al. (3). All active and semi-active DsbA variants had DsbB substrate properties similar to wild type DsbA, with k_cat and K_m values within a factor of 2 of the wild type values (see Fig. 5 and Table II). The only exceptions were the active variant 1 and the semi-active variant 70, which showed 5.4- and 3.0-fold higher k_cat values than wild type DsbA (see Fig. 5 and Table II). From the group of inactive proteins, variants 84 and 85 were reoxidized very slowly by DsbB such that we could not determine reliable catalytic parameters under the applied conditions. In contrast, the inactive variants 82 and 83 were oxidized with wild type efficiency (see Table II and Fig. 5). Thus, insufficient substrate oxidation activity is the likely reason for the lack of biological activity of variants 82 and 83, whereas both low oxidative force and slow reoxidation by DsbB cause the lack of activity in the case of variants 84 and 85. 

View larger version (31K): [in this window] [in a new window]   Fig. 5.   Michaelis-Menten kinetics of the DsbB-catalyzed oxidation of reduced DsbA variants with molecular oxygen at pH 8.0 and 25 °C. DsbA concentration was varied between 0 and 40 µM, and the reactions were followed by the decrease in DsbA fluorescence at 330 nm. A membrane preparation from an E. coli strain overproducing DsbB was used to catalyze the reaction as described previously (4, 35).

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Numerous mutagenesis studies in the last few years have demonstrated that proteins can be very tolerant toward multiple amino acid replacements and that tertiary structure context plays a major role in defining the eventual fold of a polypeptide chain (40-42). For example, randomization of all amino acids from the hydrophobic core of barnase leads to a surprisingly high tolerance toward substitutions (43). Moreover, extensive site-directed substitutions of amino acids in T4 lysozyme suggest that about half of the natural primary structure would be sufficient for defining the three-dimensional structure of the protein (44). In the case of the small SH3 domain, it was even possible to reduce the amino acid alphabet to only five different amino acids (45). Nevertheless, the fact that about 70% of all variants of DsbA obtained after complete randomization of six residues from the active-site helix 1 retained native tertiary structure and biological activity is remarkable and unexpected. This is because helix 1 is a well conserved element of regular secondary structure in the DsbA family, and because the dipole of helix 1 and hydrogen bonds within the 3₁₀-helical, N-terminal part of 1 are assumed to cause the low pK_a value of Cys³⁰ (7). The low pK_a of Cys³⁰ is indeed considered the key factor underlying the enormous oxidative force of DsbA and the extreme reactivity of its active-site disulfide bond (14, 24). Although we cannot make statements about the exact local structures of the randomized segments in the mutant proteins from our present data, all spectroscopic techniques applied so far, as well as the abnormally low pK_a values of Cys³⁰ observed for all mutant proteins, indicate that even the biologically inactive DsbA variants obtained after randomization of 1 have essentially wild type-like tertiary structures. Altogether, these results demonstrate that tertiary structure context almost alone determines folding of segment 30-37 to a conformation guaranteeing a functional enzyme. This is in agreement with the view that tertiary structure context is more important for folding of polypeptide segments within a protein than, for example, secondary structure propensities or steric factors (40). Our data also provide further evidence for the extremely degenerate code for protein folding, i.e. that a large number of sequences fulfill the requirements for the same tertiary structure.

The high sequence variability of the active-site helix of DsbA is also in agreement with results from a previous circular permutation study on DsbA, in which a series of permuted DsbA variants with novel N and C termini at different positions in the polypeptide chain was investigated. This study demonstrated that the -helical domain of DsbA, but not its catalytic thioredoxin domain, is critical for folding and stability and is the most stable part of the enzyme (27). The -helical domain of DsbA, which is inserted into the thioredoxin motif, could thus have provided a scaffold for evolving the thioredoxin domain of DsbA toward an oxidizing redox potential by interdomain stabilization. Our present data are also in agreement with previous mutagenesis studies on the DsbA segment 38-40, which connects 1 to the second part of the active-site helix 1' (residues 41-50). Deletion of residues 38-40, as well as other mutations that eliminate all charged residues in the vicinity of the catalytic disulfide bond, had practically no effect on the functional properties of DsbA (34, 39).

The measured redox potentials of the active-site helix variants generated in this study also provide interesting information on the functional requirements for the catalyst of disulfide bond formation in the bacterial periplasm. Together with previous data obtained for periplasmically produced thioredoxin variants, our data suggest that the redox potential of a biologically active catalyst that can complement DsbA deficiency in vivo must be above 210 mV. In the case of DsbA, it appears that the general consequence of a lowered redox potential in DsbA variants is less efficient oxidation of polypeptide substrates. In this context, the active variant 1 is particularly interesting, as its redox potential is already at the threshold value of 210 mV. The low redox potential of variant 1 may be compensated by its fast reoxidation through DsbB, which occurs 5.4-fold faster at DsbB saturation than wild type DsbA (see Fig. 5 and Table II). Rationally engineered, more reducing analogs of variant 1 could therefore be used to search for biologically active DsbA variants with even more reducing redox potentials.

The fact that some of the DsbA variants (variants 1 and 70) appeared to be more efficient oxidants in the hirudin oxidation and refolding experiments than the wild type, despite their lower redox potentials (Fig. 4), has to be interpreted carefully. One of the main disadvantages of DsbA is its extremely rapid, random oxidation of substrate proteins with multiple cysteine residues. This initially leads to a high fraction of fully oxidized but scrambled molecules in the case of the substrate hirudin (11). As scrambled, fully oxidized molecules cannot isomerize spontaneously to the native conformation without reduction of at least one of the disulfide bonds by an external reductant, isomerization of non-native disulfide bonds is rate-limiting for hirudin folding after oxidation with wild type DsbA. A slightly slower oxidation of hirudin, allowing intramolecular disulfide rearrangements of partially oxidized species, would thus lead to higher apparent folding rates. This appears to be exactly the case for the DsbA variants 1 and 70 (Fig. 4), which cause a slower disappearance of the HPLC peak corresponding to fully reduced hirudin but faster formation of native hirudin than wild type DsbA (Fig. 4).

Another interesting aspect of the present study with regard to the in vivo function of DsbA is the identification of the semi-active variants 70-73 (see Tables I and II and Supplemental Material, Table B), which all successfully inactivated periplasmically oriented -galactosidase by random oxidation but failed to restore motility to the dsbA deletion strain. The two purified variants 70 and 71 had sufficiently high redox potentials for biological activity (166 and 185 mV, respectively), proved to be fully active in oxidizing hirudin, and were efficiently reoxidized by DsbB in vitro. Thus, there is no obvious reason why these variants should not be fully active. We believe that the simplest explanation for the failure to reconstitute motility in the dsbA deletion strain is a more restricted substrate specificity of the semi-active variants. As mentioned above, all sequenced semi-active variants have a basic residue at position 31, which is normally not found in active variants (see Supplemental Material, Table B; only exception, variant 47). A basic residue at position 31 may prevent efficient oxidation of subunits of the flagellar P-ring, whose functional assembly is probed by the motility reconstitution assay. This result suggests that broad substrate specificity is another critical, functional property of DsbA, the common oxidant of all 148 periplasmic E. coli proteins (46).

	ACKNOWLEDGEMENT

We thank Dr. René Brunisholz (Protein Service Laboratory, ETH Zurich) for N-terminal Edman sequencing and recording of MALDI-TOF spectra.

	FOOTNOTES

^* This work was supported by the Schweizerische Nationalfonds and Eidgenössische Technische Hochschule Zürich within the framework of the National Center of Competence in Research in Structural Biology.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The on-line version of this article (available at http://www.jbc.org) contains Tables A and B and Refs. 1-5.

Present address: Novartis Pharma AG, WSJ-360.432, CH-4002 Basel, Switzerland.

^§ To whom correspondence should be addressed: Institut für Molekularbiologie und Biophysik, Eidgenössische Technische Hochschule Hönggerberg, CH-8093 Zürich, Switzerland. Tel.: 41-1-633-6819; Fax: 41-1-633-1036; E-mail: rudi@mol.biol.ethz.ch.

Published, JBC Papers in Press, August 21, 2002, DOI 10.1074/jbc.M207638200

	ABBREVIATIONS

The abbreviations used are: PDI, protein disulfide isomerase; MOPS, 4-morpholinepropanesulfonic acid; MALDI-TOF, matrix-assisted laser desorption ionization time-of-flight.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				L. Segatori, L. Murphy, S. Arredondo, H. Kadokura, H. Gilbert, J. Beckwith, and G. Georgiou Conserved Role of the Linker {alpha}-Helix of the Bacterial Disulfide Isomerase DsbC in the Avoidance of Misoxidation by DsbB J. Biol. Chem., February 24, 2006; 281(8): 4911 - 4919. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Supplemental Data All Versions of this Article: 277/45/43050    most recent M207638200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend