INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Disulfide bonds are important for the structure and stability of numerous proteins. For Escherichia coli it is now well established that the formation of disulfide bonds is an assisted process which occurs in the periplasm and is catalyzed by the thiol-disulfide oxidoreductase DsbA (7, 34). DsbA acts as a donor of disulfides to newly synthesized periplasmic proteins (23) and is reoxidized by DsbB, a second protein located in the inner membrane (5, 6). DsbA and DsbB are members of the Dsb system (the system for disulfide bond formation) which consists of at least six redox proteins belonging to the thioredoxin superfamily. These proteins contain a canonical C-X-X-C motif in the dithiol active site and seem to be conserved throughout the gram-negative bacteria (49). DsbC is another member which is suggested to act as a disulfide isomerase (56, 67). DsbD (39) keeps DsbC in a reduced state (57). More recently, DsbG was described as a novel member of the Dsb family in E. coli which oxidizes so far unknown substrates (4, 8).

Undoubtedly, the process of protein secretion in gram-negative bacteria is related to the function of the Dsb system. However, the results obtained so far are contradictory. Four major secretion pathways exist in gram-negative bacteria to direct proteins into the extracellular medium (37). In type I and type III pathways the secreted proteins directly pass both the inner and the outer membrane using a machinery formed by either three or more than 20 different proteins, respectively. In the type II pathway, which is also called the general secretory pathway, secretion occurs in two consecutive steps (48), with an intermediate state in the periplasm where the formation of disulfide bonds can take place. Since E. coli does not secrete exoproteins via the GSP under normal laboratory growth conditions (22), most studies focus on other gram-negative bacterial species. The DsbA-dependent disulfide bond formation has been described to be essential for secretion of pectate lyases and cellulase EGZ in Erwinia chrysanthemi (11, 54) and for cholera toxin and hemagglutinin-protease in Vibrio cholerae (45). However, a lipase acyltransferase from Aeromonas hydrophila is secreted without prior formation of a disulfide bond (15). The disulfide bond in Klebsiella oxytoca pullulanase is not necessary for secretion; nevertheless, secretion still depends on dsbA (52).

The opportunistic pathogen Pseudomonas aeruginosa secretes an array of hydrolytic enzymes and toxins. Many of these extracellular enzymes contain disulfide bonds, and they are secreted by a type II pathway involving at least 12 different Xcp proteins (21). Until now, only one DsbD-analogous protein, named DipZ, has been described (44) which could be a member of a hypothetical Dsb system operating in P. aeruginosa. This protein has the capacity to reduce disulfide bonds and is involved in the maturation of c-type cytochromes.

The present study describes the identification of dsbA- and dsbC-homologous genes in P. aeruginosa which encode two functional thiol-disulfide oxidoreductases. Three different extracellular enzymes were chosen to investigate the influence of Dsb proteins on secretion. dsb-negative knockout mutants were constructed, and the activities of alkaline phosphatases (APs) elastase and lipase were determined. At least two different APs exist in P. aeruginosa; the light AP (M_r, 39,000; GenBank accession no. AF047381) contains four cysteine residues which presumably form two disulfide bonds (60). Elastase (LasB) is a metalloprotease which contains two disulfide bonds (62) and is secreted in a tight noncovalent association with its propeptide, which functions as an intramolecular chaperone (13, 35). Lipase contains one disulfide bond (32) and also depends on the presence of a lipase-specific foldase (Lif) which assists in periplasmic folding and subsequent secretion (33). We report here that mutations in genes dsbA and dsbC lead to significantly altered levels of extracellular enzyme activities, thereby suggesting an important role for the Dsb system in protein secretion by P. aeruginosa.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, and growth conditions. The strains and plasmids used in this study are listed in Table 1. All medium components were from Difco (Detroit, Mich.) or Oxoid (Wesel, Germany). E. coli strains were grown at 37°C in Luria-Bertani (LB) broth medium and P. aeruginosa strains were grown at 37°C in LB or nutrient broth medium. For lipase activity assays P. aeruginosa strains were grown at 30°C in 0.5% peptone-0.3% yeast extract supplemented with 1% n-hexadecane (Sigma, Deisenhofen, Germany). A 20% n-hexadecane stock solution was solubilized by sonication in the presence of 10% gum arabic. Growth media were solidified by addition of 1.5% Bacto agar or 0.25% Bacto agar for motility assays. The following antimicrobial agents were used (concentrations [in micrograms per milliliter] are given in parentheses): ampicillin (100), carbenicillin (100), gentamycin (10), spectinomycin (50), and streptomycin (50) for E. coli and carbenicillin (300), gentamycin (2.5 to 20), tetracycline (100), spectinomycin (400), and streptomycin (400) for P. aeruginosa.

DNA manipulation. Plasmid DNA was isolated using the alkaline lysis method (9), followed by anion-exchange chromatography on Qiagen tips (Qiagen). Chromosomal DNA was prepared as described elsewhere (24). General DNA procedures were performed as described previously (50). Reaction endonucleases and bacteriophage T4 DNA ligase were purchased from MBI Fermentas and used according to manufacturer's instructions.

DNA sequencing and analysis. DNA was sequenced using the dideoxy-chain termination method (51) on an ALF-Express automated sequencer (Pharmacia) and was kindly performed by the Lehrstuhl für Biochemie, Ruhr-Universität Bochum, Bochum, Germany. Multiple alignments were performed with the program MegAlign of the program package DNAstar (Lasergene). Pairwise alignments were done with the BLASTP program (3). The PSORT program (http://psort.nibb.ac.jp/) was used for prediction of protein localization (41).

Southern blot analysis. Genomic DNA (5µg/lane) was digested to completion with restriction endonucleases, electrophoresed in a 0.6% agarose gel, and transferred to nylon membranes (Qiabrane; Qiagen, Duesseldorf, Germany) by capillary and gravity action (16). Specific DNA probes were labeled with digoxigenin using a labeling and detection kit (Roche Diagnostics). Hybridization under high-stringency conditions was done according to the manufacturer's protocols at 55°C for dsbA and at 65°C for dsbC in 50 mM Na phosphate buffer (pH 7.0) containing 2% blocking reagent, 7% sodium dodecyl sulfate (SDS), and 0.1% laurylsarcosin.

PCR amplifications. The following oligonucleotides were designed as degenerated primers to amplify fragments of genomic DNA: for dsbA, sense primer 5'-CTCSTTCTACTGCCCSCACTGCTA-3' and antisense primer 5'-GTTSACSACSACSGCCGGSACGC-3' and for dsbC, sense primer 5'-ACCGTSTTCACCGACATCACCTGC-3' and antisense primer 5'-GTTCTTGTCCTTSRCGCACCAGAT-3'. NdeI and XhoI sites were introduced at the 5' and 3' ends, respectively, of the dsb genes to allow for cloning in the expression vector pET21b (Novagen) by using the sense primer 5'-TTATTATCATATGGACGACTATACCGCCGGC-3' and the antisense primer 5'-TTTTTTTCTCGAGCTTCTTGGCCGCGCGCTC-3' for dsbA and the sense primer 5'-ATATCATATGGACAATGCCGATCAGAA-3' and the antisense primer 5'-ATATCTCGAGTTTGGCCTCCAGCGCCAG-3' for dsbC. Plasmid pDDA2C34S was constructed using PCR-mediated site-directed mutagenesis with the sense primer 5'-CGCCGCCTACTTCGCCAGCCAGAA-3' and the antisense (mutagenic) primer 5'-AACGCGTAGCAATGCGGGCTGCCATAC-3' (mutagenic bases are underlined and correspond to codon 34 with a substitution for Cys by a Ser residue). In plasmid pDDA2 a 562-bp AgeI/MluI fragment was replaced by the AgeI/MluI fragment of the PCR product to obtain pDDA2C34S. PCR was performed in 30 cycles with a Robo Cycler Gradient 40 (Stratagene) under standard conditions using an annealing temperature of 50 to 65°C.

Construction of dsb mutants. The dsbA mutant was constructed by cloning the dsbA gene on a 2.3-kb XhoI fragment from plasmid pMDS2 into the suicide plasmid pME3087 (63). The dsbA gene was inactivated by deletion of a StyI fragment and subsequent insertion of an Sm/Sp cassette from plasmid pHRP316 (47) into the same sites to obtain pMDS2. This plasmid was mobilized from E. coli S17.1 into P. aeruginosa PAO1. Transconjugants were selected on plates containing streptomycin-spectinomycin, tetracycline, and Irgasan (25 µg/ml) and enriched with carbenicillin (2,000 µg/µl) to select a derivative in which the vector DNA including the tetracycline resistance gene was deleted, yielding strain P. aeruginosa PAML1 dsbA::Sm/Sp. Plasmid pMDS2 was mobilized into P. aeruginosa PABS2 lipA::lacZ to obtain the double mutant P. aeruginosa PAAU3 dsbA::Sm/Sp lipA::lacZ. A nonpolar mutation in the dsbA gene was obtained by cloning an AgeI/MluI PCR fragment containing the dsbA(C34S) mutation into the same sites of pMDS2. This plasmid named pMDS2C34S was mobilized into P. aeruginosa PAML1. An Sm^s-Sp^s transconjugant named P. aeruginosa PAAU5 was selected, and the mutation was confirmed by DNA sequence analysis. The dsbC mutant was constructed by inactivating the dsbC gene from plasmid pMEDC6 which is a derivative of pME3087. A 240-bp internal StyI fragment was deleted, and the Sm/Sp cassette was inserted in its place. P. aeruginosa PATE1 dsbC::Sm/Sp was constructed as previously described for the dsbA mutant P. aeruginosa PAML1. A dsbA dsbC double mutant was constructed by cloning a Gm cassette from plasmid pBSL141 (2) into the StyI sites of pMDS2. This plasmid pMDS2G was mobilized into P. aeruginosa PATE1 dsbC::Sm/Sp. A transconjugant selected on plates containing gentamycin, streptomycin-spectinomycin, tetracycline, and Irgasan (25 µg/µl) was enriched with carbenicillin (2,000 µg/µl) to select a derivative in which the vector DNA was deleted, yielding P. aeruginosa PAAU5 dsbA::Gm dsbC::Sm/Sp. Mutations were confirmed by Southern blotting of chromosomal DNA from P. aeruginosa strains PAML1, PATE1, PAAU1, and PAAU3.

SDS-PAGE and immunoblotting. For immunodetection of lipase protein cells were grown for 20 h at 30°C as described above and pelleted by centrifugation for 10 min at 3,000 × g. Supernatants or cell extracts were precipitated with trichloroacetic acid as previously described (46). Samples were separated by SDS-polyacrylamide gel electrophoresis (PAGE) as described by Laemmli (36). A 5% polyacrylamide stacking gel and 12 or 15% polyacrylamide separating gels, respectively, were used for detection of lipase or DsbA. Molecular mass standard proteins of 66, 45, 36, 29, 24, 20, and 14 kDa were from Sigma (Deisenhofen, Germany). For immunodetection proteins were blotted onto a polyvinylidene difluoride membrane (Bio-Rad) using carbonate buffer (pH 9.3) containing 20% methanol (20). A polyclonal antiserum against the DsbA protein of P. aeruginosa was obtained by using the purified His-tagged DsbA as the antigen and a standard immunization protocol for rabbits (Eurogentec, Seraing, Belgium). The antiserum was diluted 1:100,000 in TBST (50 mM Tris-HCl [pH 6.8], 150 mM NaCl, 1 mM MgCl₂, 2% Tween 20), and horseradish peroxidase-labeled goat anti-rabbit antibody (Bio-Rad) diluted 1:5,000 in TBST was used as the second antibody. Detection was done with the ECL system (Amersham) according to the manufacturer's protocol. Lipase was detected with a polyclonal antiserum (1:80,000 dilution) using the same protocol.

Protein purification. DsbA and DsbC proteins of P. aeruginosa were purified under native conditions using immobilized-metal affinity chromatography (IMAC) with a Ni-nitrilotriacetic acid-agarose column (Qiagen). Carboxy-terminally His₆-tagged proteins were produced by cloning into vector pET21b NdeI/XhoI PCR fragments encoding dsbA and dsbC, though lacking both their putative N-terminal signal sequences and their stop codons, resulting in hybrid plasmids pE21DAT3 (dsbA) and pEDCChis (dsbC), respectively. These plasmids were transferred into E. coli BL21(DE3) and the resulting strains were grown in 500 ml of LB-M9 medium to an optical density at 580 nm of 0.6. After the addition of 0.5 mM isopropyl--D-thiogalactopyranoside (IPTG) and an additional 3 h of growth, cells were disrupted by sonication and the cell extract was centrifuged, filtered (Schleicher & Schuell membranes; pore diameter, 0.2 µm), and loaded on an IMAC column (volume, 2 ml) previously equilibrated with 50 mM phosphate buffer (pH 8.0) containing 0.3 M NaCl and 40 mM imidazole. For the DsbC purification an addition of 10 mM -mercaptoethanol was necessary. Chromatography was performed with the GradiFrac System from Pharmacia Biotech (Freiburg, Germany) and an increasing imidazole gradient (40 to 400 mM) at a flow rate of 0.1 ml/min for elution. Protein concentration was determined by the method of Bradford (12) with bovine serum albumin as the standard.

Enzyme activity assays. Oxidoreductase activities was determined as described elsewhere (29). DsbA or DsbC protein (5 µM) was incubated with 167 µM insulin in the presence of 0.83 µM dithiothreitol (DTT) in 0.1 M potassium phosphate buffer (pH 7.0) containing 2 mM EDTA. DsbA protein from E. coli (Roche Diagnostics) was used as a control. Substrate-containing indicator plates were used for detecting alkaline phosphatase with X-phosphate (14) and elastase with elastin (43). Elastase activities are indicated by the formation of halos around the colonies, whereas alkaline phosphatase activity results in blue-colored colonies. Lipase activity in P. aeruginosa culture supernatants was determined with p-nitrophenyl-palmitate (Sigma) as the substrate, as described elsewhere (65).

Nucleotide sequence accession numbers. The nucleotide sequences reported here are accessible from GenBank under accession no. U84726 for dsbA and AF057031 for dsbC.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Identification, cloning, and characterization of dsbA and dsbC. A comparison of DsbA protein sequences obtained from E. coli (7, 34), Haemophilus influenzae (GenBank accession no. M94205), V. cholerae (66), Legionella pneumophila (GenBank accession no. U15278), and E. chrysanthemi (54) allowed us to identify conserved regions of amino acid homology. Two degenerated primers were designed and used to amplify a 376-bp PCR fragment of P. aeruginosa DNA which served as a homologous probe to identify by Southern hybridization analysis of P. aeruginosa genomic DNA the dsbA gene which was located on a 4.0-kb PstI fragment (data not shown). A gene library containing 4.0-kb PstI fragments was cloned into pBluescript II SK. A positive E. coli clone containing the hybrid plasmid pDS83 was identified by colony-filter hybridization. The dsbA gene on a 2.3-kb XhoI fragment was subcloned to obtain pDS2, and the DNA sequence was determined revealing an open reading frame of 633 bp. The corresponding protein exhibited a significant homology to other known DsbA proteins (Fig. 1A). The highest homology score (68% identity, 82% similarity) was obtained with Azotobacter vinelandii DsbA (42). A putative signal sequence consisting of 22 amino acids was identified at the amino terminus of DsbA, suggesting a periplasmic location of the mature protein, which has calculated molecular mass of 21 kDa.

View larger version (43K): [in this window] [in a new window]   FIG. 1.   Alignments of the active site amino acid residues of DsbA (A) and DsbC (B) proteins from P. aeruginosa and other bacterial species. The F-(X)4-C-X-X-C consensus motive characteristic of the thioredoxin superfamily is boxed. Percentages of sequence identity (id) and similarity (si) were obtained by pairwise alignments of whole sequences.

Purification and enzyme activity of DsbA and DsbC. Plasmids pE21DAT3 and pEDCCHis containing genes dsbA and dsbC, respectively, but lacking their signal sequences and carrying 3' fusions encoding His₆ tags were expressed in E. coli BL21(DE3). Upon addition of IPTG to induce T7 pol expression, two additional proteins of 22 and 25 kDa, respectively, were observed in whole-cell extracts from E. coli BL21(DE3). They were purified to electrophoretic homogeneity by IMAC under nondenaturating conditions (Fig. 2A). His₆-tagged DsbA and DsbC were eluted at 150 mM imidazole. Yields of purified proteins from a 1-liter culture were 140 mg for DsbA and 60 mg for DsbC.

View larger version (17K): [in this window] [in a new window]   FIG. 2.   Purification and thiol-disulfide oxidoreductase activities of His-tagged DsbA and DsbC. (A) Cell extracts of E. coli BL21(DE3) carrying plasmids pE21DAT (dsbA) or pEDCChis (dsbC) were loaded onto a Ni2+-nitrilotriacetic acid affinity column and eluted with 150 mM imidazole. Fractions were analyzed by SDS-PAGE and staining with Coommassie brilliant blue G (B) The reduction of insulin in the presence of DTT was determined by the increase in turbidity measured at 600 nm. DsbA from E. coli was used as a positive control. As a negative control a sample without protein was used.

DsbA and DsbC restore motility of E. coli and P. aeruginosa dsbA mutants. E. coli JCB571 is defective in disulfide bond formation due to a Km cassette insertion in the dsbA gene (7). Whereas E. coli JCB571 was nonmotile on soft agar plates due to its inability to form functional flagella (18), transformation with pDS2 carrying the P. aeruginosa dsbA gene restored motility to a level equal to that observed for wild-type E. coli JCB570 containing a functional dsbA gene (Table 2). Introduction of plasmid pBKDC1 carrying the P. aeruginosa dsbC gene only partially restored the motility of E. coli JCB571.

Dsb mutations affect extracellular enzyme activities. Extracellular elastase activity was determined using indicator plates containing the substrate elastin. Elastolytic activity of wild-type P. aeruginosa PAO1 is indicated by a halo around the colony. In the mutant strain P. aeruginosa PAML1 dsbA elastolytic activity was drastically reduced, as indicated by the lack of a halo around the colony (Fig. 3A). As expected, the type II secretion-deficient mutant P. aeruginosa 2B18 pilD xcpA (58) also lacked extracellular elastase activity, as did the double mutant P. aeruginosa PAAU1 dsbA dsbC. A mutation in dsbC did not result in a reduction of elastase activity. Alkaline phosphatase activities of wild-type P. aeruginosa PAO1 and dsb mutants were determined using indicator plates with the substrate X-phosphate. Figure 3B shows that P. aeruginosa PAO1 produced AP, as indicated by blue-colored colonies (darkly stained in Fig. 3B). In mutants P. aeruginosa PAML1 dsbA and PAAU1 dsbA dsbC AP activity was not detectable. A mutation in dsbC alone did not affect AP activity.

View larger version (71K): [in this window] [in a new window]   FIG. 3.   AP and elastase production by P. aeruginosa wild-type (wt) and dsb mutants. Bacteria were grown on indicator plates containing the substrates elastin (A) and X-phosphate (B).

View larger version (14K): [in this window] [in a new window]   FIG. 4.   Extracellular lipase activity of P. aeruginosa dsb mutants. (A and C) Lipase assays of supernatants from the wild type (wt), a dsbA mutant, a dsbA mutant with plasmid harboring dsbA (pDDA2), or a plasmid control (pUCPKS) (A) and a dsbA dsbC mutant and a dsbC mutant (C). The results are presented as percentages of the maximum lipase activity of the wild type. As controls we use a lipA mutant and a xcpQ mutant from P. aeruginosa. (B and D) Detection of lipase in culture supernatants assayed for parels A and C by Western blotting. An amount of each supernatant corresponding to an optical density (at 580 nm) of 0.6 was subjected to SDS-PAGE. rel., relative.

View larger version (26K): [in this window] [in a new window]   FIG. 5.   Effect of catalytically inactive DsbA on the formation of extracellular lipase. (A) Li-pase activities were determined with supernatants from a P. aeruginosa dsbA-negative strain carrying plasmid pDDA2C34S encoding DsbA(C34S) and a P. aeruginosa strain PAAU5 carry-ing the nonpolar mutation dsbA(C34S) which was grown for 20 h at 30°C. (B and C) Extracellular lipase (B) and DsbA in whole-cell extracts (C) were detected by Western blotting. rel., relative.

DsbA affects lipase stability and secretion. We were unable to detect any lipase in whole-cell extracts of P. aeruginosa PAO1 and P. aeruginosa PAML1 dsbA. Lipase protein became detectable by immunoblotting only upon increasing the lipA copy number. Constitutive expression of lipA from the lac promotor on plasmid pBBL7 in P. aeruginosa PAAU3 lipA dsbA resulted in a decrease by 75% of lipase protein and activity in both culture supernatants and whole-cell extracts indicating that lipase was at least partly degraded in a dsbA-negative background (Fig. 6A, C, and D). The role of the disulfide bond for lipase protein stability and secretion efficiency was analyzed using the lipase mutant C183G expressed from plasmid pBBRC183G (H. Duefel and K.-E. Jaeger, unpublished data). Overexpression of lipA(C183G) in the lipase-negative strain P. aeruginosa PABS1 did not result in any detectable lipase protein, either in the supernatant or in whole-cell extracts [Fig. 6B and C]). The stability of secreted lipase was investigated in cell culture supernatants obtained from P. aeruginosa wild-type and the dsbA mutant (Fig. 7A). Both lipases remained stable for 16 h of incubation. The presence of DTT in the wild-type culture supernatant did not affect extracellular lipase activity. However, when DTT was added to a growing culture of P. aeruginosa PAO1, no extracellular lipase was detectable (Fig. 7B).

View larger version (27K): [in this window] [in a new window]   FIG. 6.   Expression of genes lipA and lipA(C183G) in trans. (A) Lipase assays of supernatants and whole cell extracts from P. aeruginosa strains lipA, and lipA dsbA each containing plas-mid pBBL7 (lipA) or pBBRC183G (lipA[C183G]). (B and C) Strains were grown for 4 h in LB medium. Immunodetection of lipase in supernatants (B) and whole-cell extracts (C) rel., relative.

View larger version (23K): [in this window] [in a new window]   FIG. 7.   Effect of DsbA and DTT on lipase stability. (A) Lipase activities of cell culture supernatants obtained after 12 h of growth and further incubated for 16 h at 30°C in the presence or absence of 10 mM DDT, (B) detection of extracellular lipase by immunoblotting of P. aeruginosa PAO1 supernantant obtained from a culture grown in the presence of 10 mM DTT. The amount of culture supernatant loaded into each lane corresponds to a culture optical density at 580 nm of 0.6.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Disulfide bonds must be formed in an oxidative environment which is provided by the periplasm of E. coli and other gram-negative bacteria. Proteins secreted via the type II pathway can form disulfide bonds during their passage through the periplasm (48). Here, the DsbA-DsbB system seems to represent the main route for disulfide bond formation (54).

The opportunistic pathogen P. aeruginosa secretes a large number of hydrolytic enzymes and toxins. Many of them, including lipase, elastase, and AP, use the type II secretion pathway (21) and also contain disulfide bonds. The question of how these disulfide bonds are formed in P. aeruginosa has not yet been adressed. So far, only the presence of a DsbD (DipZ)-homologous protein which has the capacity to reduce disulfide bonds has been described (44). More recently, DsbA was identified as a virulence factor in a Caenorhabditis elegans killing assay (61).

We have identified, cloned, and sequenced both dsbA and dsbC of P. aeruginosa (Fig. 1). Analysis of the deduced amino acid sequences revealed a striking difference from other redox proteins belonging to the thioredoxin superfamily, such as PDI, glutaredoxin, thioredoxin, or other Dsb proteins: the phenylalanine residue present in the conserved dithiol-active site consensus motif F-(X)₄-C-X-X-C was not found in P. aeruginosa DsbA but was present in DsbC. This residue is also absent in DsbA from A. vinelandii, which shares the highest homology with P. aeruginosa DsbA (Fig. 1A).

The thiol-disulfide oxidoreductase activity of DsbA was unequivocally demonstrated by the following: (i) P. aeruginosa dsbA was capable of complementing the motility-negative phenotype of an E. coli dsbA mutant, (ii) the phenotype of the P. aeruginosa dsbA mutant was characterized by significantly reduced motility and AP activity as previously described for dsbA mutants of other gram-negative bacteria (7, 34, 54), and (iii) DsbA efficiently catalyzed a disulfide exchange reaction in vitro (Fig. 2B).

With the exception of H. influenzae the dithiol-active sites of all known DsbC proteins contain the conserved motif C-G-Y-C (Fig. 1B). In P. aeruginosa DsbC, we have found this consensus motif changed to C-P-Y-C, a motif which is also present in DsbG of E. coli (4). The X-X residues located in between the cysteine residues in the conserved C-X-X-C consensus motif are different within the thioredoxin-like redox protein family, yet they are highly conserved within each subfamily. Any alteration of these residues perturbs the reduction potential of the disulfide bond of each redox protein (25, 30). Three independent lines of evidence point to the fact that we have isolated the P. aeruginosa dsbC rather than the dsbG gene, as follows. (i) The homology of P. aeruginosa DsbC, particularly around the C-X-X-C consensus motif, to known DsbC proteins is much higher (up to 40% identity, 56% similarity) than that to the E. coli DsbG protein (22% identity, 38% similarity) (Fig. 1B). (ii) We have identified in the P. aeruginosa genome an open reading frame of 768 bp whose putative gene product revealed a higher homology (46% identity, 70% similarity) to DsbG from E. coli than to DsbC from P. aeruginosa. (iii) DsbG from E. coli was unable to catalyze the reduction of insulin (8). In contrast, we have shown here (Fig. 2B) that P. aeruginosa DsbC was even more effective than DsbA in the DTT-dependent reduction of insulin.

DsbA and DsbC play an important role for secretion of enzymes via the sec-dependent type II pathway. In a dsbA mutant extracellular elastase and lipase activities were significantly reduced; in a dsbA dsbC double mutant lipase activity was hardly detectable (Fig. 4C and D), indicating that DsbC alone can also catalyze disulfide bond formation in vivo to an extent detectable only in a DsbA-negative background. A comparable result was found for E. coli where AP activity was even more reduced in a dsbA dsbC double mutant than in a dsbA mutant (38).

The physiological role of DsbC in E. coli seems to be the shuffling of misoxidized disulfide bonds (56, 67). However, it is still unknown how the disulfide bond isomerase DsbC can distinguish nonnative from native disulfide bonds (19). Interestingly, a dsbC mutation in P. aeruginosa resulted in the doubling of extracellular lipase activity (Fig. 4C). We assume that part of the lipase molecules which possess one surface-located disulfide bond (32, 40) may first be correctly oxidized by DsbA but subsequently be reduced by DsbC. A dsbC knockout mutant would then be expected to produce a higher amount of active extracellular lipase as we have observed experimentally (Fig. 4C and D).

The reduced lipase activity in the P. aeruginosa dsbA mutant strictly correlated with a reduced amount of extracellular lipase protein (Fig. 4B), suggesting that a correctly formed disulfide bond is a prerequiste to ensure efficient secretion. If its formation is hindered, e.g., by disruption of the dsbA gene, by introduction of the mutation C183G (Fig. 6), or by addition of DTT to the growth medium (Fig. 7B), secretion of lipase into the culture supernatant cannot occur. Instead, lipase may quickly and efficiently be degraded in the periplasm of P. aeruginosa. Indeed, we have observed in a cell extract obtained from the dsbA mutant strain a decrease of lipase activity by 75%. Also, we were unable to detect by Western blotting lipase variant C183G in whole-cell extracts (Fig. 6C). These results were supported by the findings of Liebeton et al. who have studied the role of the disulfide bond in P. aeruginosa lipase (36a). Wild-type lipase and variants C183S, C235S, and C183S-C235S were constructed, expressed in E. coli, and subsequently refolded to enzymatic activity. Lipolytic activity was detected with wild-type and all variant lipases indicating that an intact disulfide bond was required neither for refolding of lipase by its cognate foldase (Lif) nor for reaching and maintaining an enzymatically active conformation.

How does the Dsb-mediated formation of disulfide bonds affect secretion of extracellular enzymes? Studies with a number of different extracellular proteins have revealed confusing results. Cellulase EGZ from E. chrysanthemi (11, 54) and cholera toxin from V. cholerae (45) both required correct disulfide bond formation for enzyme stability and secretion, whereas aerolysin (26) and the lipase/acyltransferase of A. hydrophila were secreted irrespective of the presence of disulfide bonds (15). In a Burkholderia cepacia mutant defective in the DsbA-DsbB-dependent disulfide bond formation both variations of the theme were found: though extracellular protease activity was missing, lipase activity remained unaffected. Interestingly, both enzymes contain disulfide bonds and are secreted via a type II pathway (1). K. oxytoca pullulanase was secreted in E. coli without intramolecular disulfide bonds; however, DsbA was still necessary for efficient secretion. A general chaperone activity of DsbA (31, 68) as well as an indirect effect of DsbA via disulfide bond formation in proteins forming the secretion machinery has also been discussed (52). P. aeruginosa producing the enzymatically inactive DsbA variant C34S showed significantly reduced extracellular lipase activity, suggesting that the oxidoreductase activity of DsbA is necessary to allow the formation of a lipase which is stable enough to be secreted via the type II pathway in P. aeruginosa (Fig. 5).

At present, we can draw the following general conclusion regarding the role of the Dsb system for folding and/or secretion of extracellular and disulfide bond-containing enzymes: Those enzymes which do not need their disulfide bonds to achieve a stable conformation in the periplasm also do not rely on a functional DsbA-DsbB-system for efficient secretion. Examples of this type include the aerolysin and the lipase/acyltransferase from A. hydrophila. However, extracellular enzymes cellulase EGZ and pectate lyase from E. chrysanthemi, as well as elastase, AP and lipase from P. aeruginosa, require intact disulfide bonds for maintaining a stable and secretion-competent conformation in the periplasm. In a dsbA mutant background these enzymes will form unstable periplasmic intermediates which will rapidly be degraded.