INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Vibrio cholerae O1, the causative agent of Asiatic cholera, has been isolated from coastal and estuarine environmental samples, both as free-living bacteria and in association with phytoplankton, zooplankton, crustaceans, and mollusks (26). These observations and the capacity of the organism to adaptively respond to changes in salinity, temperature, and the availability of nutrients (48, 49) have led to the idea that this species can successfully occupy one or more ecological niches in a variety of aquatic habitats. If this is so, then it seems likely that the physiology and structure of V. cholerae O1 in environmental reservoirs might differ in fundamental ways from these features of the organism in the intestine.

Recent studies support this view. Upon nutrient deprivation, comma-shaped V. cholerae cells become coccoid, significantly decrease their volume (4), and reduce their lipid, carbohydrate, RNA, and protein contents (22). Changes of this kind to nutrient deprivation have been observed in other bacterial species as well, most notably Escherichia coli, where, in addition to metabolic and structural changes, nutrient starvation also induces resistance to heat shock, oxidative stress, osmotic shock, and acid (27, 28, 31, 33, 35, 36).

The global response by E. coli to starvation and stress requires rpoS, which is induced to express an alternative sigma factor, ^S, during nutrient deprivation or the stationary phase of growth (31, 36). In turn, ^S governs the expression of multiple genes that mediate the observed changes in physiology and structure. These properties of the E. coli rpoS system have led us, in the work reported here, to examine the possibility that a V. cholerae homolog of rpoS exists and is required for survival of the organism under conditions of growth that it might encounter in environmental aquatic habitats. Here we report the cloning and characterization of V. cholerae rpoS (rpoS_vc), the phenotype of a rpoS_vc knockout mutant, and the unexpected finding that this mutant is markedly deficient in its capacity to produce and/or secrete an extracellular protease.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Media and growth conditions. All strains were maintained at 80°C in Luria-Bertani (LB) broth supplemented with glycerol (15%, vol/vol). For the experiments described below, the cells were grown aerobically at 37°C in LB broth, on MacConkey plates (Difco, Detroit, Mich.), in M9 minimal medium (37) with either no glucose or 0.02% (wt/vol) glucose, in artificial sea water (ASW; Sigma Chemical Co., St. Louis, Mo.), and on thiosulfate-citrate-bile salts-sucrose (TCBS; Difco) agar. The following supplements were added as appropriate: ampicillin (100 µg/ml), kanamycin (35 µg/ml), chloramphenicol (5 µg/ml), rifampin (100 µg/ml), and X-Gal (5-bromo-4-chloro-3-indolyl--D-galactopyranoside; 40 µg/ml).

Bacterial strains and plasmids. Bacterial strains and plasmids used in this study are described in Table 1. Plasmids were introduced into E. coli strains by transformation and into V. cholerae strains through electroporation or through mating with E. coli S17-1.

DNA manipulations and analysis. Plasmid DNA and chromosomal DNA preparation, DNA ligation, bacterial transformation, agarose gel electrophoresis, and Southern blotting were carried out by standard techniques described by Sambrook et al. (46). Restriction enzymes and DNA modification enzymes were purchased from New England Biolabs, Inc. (Beverly, Mass.). [-³²P]dCTP was obtained from Amersham Life Science Inc. (Arlington Heights, Ill.). Oligonucleotides were purchased from Operon Technologies, Inc.

Genomic library construction. Wild-type V. cholerae 92A1552 DNA was partially digested with Sau3A and size selected by agarose gel electrophoresis using SeaKem (FMC Bioproducts, Rockland, Maine). DNA fragments of 3 to 6 kb were pooled and ligated into a multicopy vector (pBluescriptKS⁺) that had been digested with BamHI and treated with bacterial alkaline phosphatase. The ligation mixture was transformed into E. coli DH5. The colonies were pooled, and the plasmid DNA from these pools was isolated. A V. cholerae genomic library was also constructed in  DashII (Stratagene, La Jolla, Calif.), using wild-type 92A1552 genomic DNA which was partially digested with Sau3A and treated with bacterial alkaline phosphatase before being ligated into BamHI-digested  DashII. The ligation mixture was packaged by using Gigapack II (Stratagene). The library was then screened with the rpoS gene to obtain clones that carry rpoS and nlpD and the upstream promoter region of nlpD. The positive clones were mapped by standard molecular techniques, and a 4-kb EcoRI fragment that contains whole rpoS and nlpD genes as well as its upstream region was identified. This fragment was cloned into pBluescriptKS⁺ and pWKS130 which had been digested with EcoRI, yielding pFY8 and pFY9, respectively.

Isolation of V. cholerae rpoS by functional complementation. The wild-type V. cholerae Sau3A recombinant library in pBluescriptKS⁺ was introduced into E. coli ZK918 by transformation. Transformants were selected on MacConkey-lactose plates containing ampicillin (100 µg/ml) and kanamycin (35 µg/ml), on which colonies of the parent strain are white. Red colony transformants were tested for catalase activity, using H₂O₂ on LB plates. Plasmid DNA was isolated from red, catalase-positive transformants and analyzed by restriction enzyme digestions. Plasmids with overlapping fragments that were able to complement E. coli ZK918 were identified. A 1.698-kb XbaI fragment from pFY1 that complemented both lactose fermentation and catalase production in E. coli ZK918 was selected for sequence analysis.

DNA sequence analysis. The 1.698-kb XbaI fragment was cloned into the XbaI site of pGEM-7 ZF⁺ (Promega, Madison, Wis.), generating plasmid pFY3. A series of plasmids containing nested deletions of the insert DNA, generated by using an Erase-a-Base kit (Promega), were used as a template for sequence analysis. Double-stranded DNA sequencing was performed with an Applied Biosystems 373A automated DNA sequencer. The remaining gaps in the sequence were filled, using specifically designed oligonucleotide primers. Sequence analysis and database searches were performed with IntelliGenetics (Menlo Park, Calif.) and Genetics Computer Group (Madison, Wis.) software packages and the NCBI BLAST server.

Construction of insertion mutants in rpoS. A mutant strain, derived from V. cholerae 92A1552, which carries a chloramphenicol resistance (cm) cassette in the rpoS coding sequence was constructed as follows. A 1.022-bp PCR fragment containing the rpoS coding sequence was generated by using pFY2 as the template and the primers 5'-CGCGTCGACATGAGTGTCAGCAATACC-3' and 5'-CGCGTCGACGTTGTCGTATTCGACGTT-3'. This fragment contained a 1-kb region from the rpoS start codon to the rpoS stop codon, which is flanked by SalI restriction sites which were used to clone into pUC19/18 that had been digested with SalI. The resulting plasmid, pFY4, was digested with BamHI and ligated to a cm cassette contained within a BamHI restriction fragment of pHP45W-Cm (3.6 kb). The resulting plasmid, pFY5, was digested with SalI, and the resulting 4.6-kb fragment was cloned into pKNG101, yielding pFY6. FY6 was mobilized into the wild-type V. cholerae strain from E. coli S17-1  pir by conjugation.

SDS-polyacrylamide gel electrophoresis (PAGE) and immunoblot analysis. Total cell lysates were prepared from overnight grown cultures by sonication. Protein concentration was determined with a bicinchoninic acid protein assay kit (Pierce, Rockford, Ill.). Lysates containing 25 µg of protein were separated on sodium dodecyl sulfate (SDS)-12% polyacrylamide gels and directly electroblotted onto Immobilon-P membranes (Millipore, Bedford, Mass.). Blots were blocked and then incubated with an antiserum to E. coli RpoS followed by incubation with an anti-rabbit horseradish peroxide-conjugated secondary antibody. The blots were developed with the Amersham enhanced chemiluminescence system.

Survival assays. Long-term starvation survival was analyzed in M9 medium supplemented with 0.02% glucose, in M9 without any carbon source, and in nutrient-free ASW basal salts medium. For these experiments, cultures were grown overnight in LB, centrifuged, washed, resuspended in 25 ml of the appropriate medium at a density of approximately 10⁶ to 10⁷ CFU ml¹, and incubated at 37°C aerobically (200 rpm). Immediately after inoculation and for up to 5 and 30 days, viable cell numbers were determined on LB plates after appropriate dilution. For hydrogen peroxide (15 mM) and osmolarity (NaCl, 2.4 M) survival studies, the overnight grown cultures in LB were collected by centrifugation, resuspended, and incubated under the indicated conditions. At the indicated time points, samples were taken and plated onto LB agar plates for determination of viable cell numbers.

2D gel electrophoresis. Two-dimensional (2D) gel electrophoresis was performed by the method of O'Farrell (42). Cells were grown in M9 minimal medium supplemented with 0.2% glucose to mid-logarithmic phase and to stationary phase. For logarithmically grown cultures, 50 µCi of L-[³⁵S]methionine (Du Pont NEN, Wilmington, Del.) was used to label a 1-ml sample for 2 min followed by a 1-min chase with 0.2 mM unlabeled methionine. Labeling time was increased to 30 min in the case of stationary-phase cultures, followed by a 15-min chase with unlabeled methionine. The samples were washed twice with phosphate-buffered saline. The amount of L-[³⁵S]methionine incorporated was determined by trichloroacetic acid (10%) precipitation and liquid scintillation counting of a small portion of the cell lysate. Equal amounts of counts (10⁶ cpm) were loaded onto the tube gels. 2D gel electrophoresis was carried out with a Protean II (Bio-Rad, Hercules, Calif.) device as directed by the manufacturer. Labeled proteins were visualized by autoradiography.

Protease assay and hemagglutinin/protease (HA/protease) Western analysis. The initial screen for protease production was done on LB agar plates supplemented with 1% skim milk, where protease clearance zones of strains were analyzed. Following the initial screen, a fluorescence-based EnzCheck assay (Molecular Probes, Eugene, Oreg.) was used to determine protease activity in the samples. The BODIPY-casein substrate was incubated in a 96-well microplate with culture supernatants or with cell sonicates for 1 h at 37°C; fluorescence was measured (485 and 530 nm for excitation and emission, respectively) with a fluorescence microplate reader.

In vivo assay for V. cholerae intraintestinal growth. A competition assay for intraintestinal growth in 4- to 5-day-old CFW suckling mice was performed by administering mixtures containing equal numbers of the rpoS::cm mutant and wild-type V. cholerae parent, by gavage, into 4- to 5-day-old CFW mice. After 24 h, the small intestine was removed and intestinal homogenates were serially diluted and plated onto TCBS agar. Enumeration of mutant strain viable counts was performed after replica plating of recovered cells onto TCBS-chloramphenicol agar plates. Then the ratio of values for the mutant and wild-type strains was determined and used to derive the competitive index (15).

Nucleotide sequence accession number. The GenBank accession number for the sequence reported in this study is AF000945.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Cloning and characterization of V. cholerae rpoS. The V. cholerae O1 functional and structural homolog of E. coli rpoS was cloned by complementation using E. coli ZK918, a strain with an insertionally inactivated rpoS gene, a deletion in the lac operon, and a chromosomal bolAp1-lacZ fusion (6). bolAp1 requires rpoS for expression; it encodes a morphogene which is necessary for changes in bacterial shape that occur during starvation (6, 30). Lactose fermentation by this strain, as determined by the production of red colonies on MacConkey-lactose agar, does not normally occur but is regained when a functional copy of rpoS is provided. To clone rpoS_vc, a genomic library was prepared in pBluescriptKS⁺ with DNA from V. cholerae 92A1552, a wild-type strain (El Tor biotype, Inaba serotype) isolated in 1992 from a traveler from Latin America (Table 1), and this library was introduced into E. coli ZK918. Red colonies were picked and tested for catalase production, since expression of katE, the HPII catalase structural gene that confers resistance to H₂O₂, is also controlled by rpoS. This complementation strategy is shown in Fig. 1A. When a plasmid containing either the E. coli rpoS gene (pDEB2) or the V. cholerae rpoS gene (pFY1) is present in E. coli ZK918, the recombinant ferments lactose, imparting a red color to the corresponding colony on MacConkey indicator plates (Fig. 1B). In contrast, the same host carrying the cloning vector (pBluescriptKS⁺) does not ferment lactose. The same recombinants shown in Fig. 1B were also tested for catalase production on LB agar; all but the strain containing the cloning vector tested positive.

View larger version (39K): [in this window] [in a new window]   FIG. 1.   (A) Schematic diagram of the complementation strategy. E. coli ZK918 carries an rpoS::cm insertion mutation and a bolAp1::lacZ fusion and has a deletion of the lac operon. This strain was transformed with a V. cholerae genomic library, and transformants that were lac+ were selected on MacConkey agar and then assayed for catalase activity. (B) Complementation by V. cholerae rpoS. E. coli ZK918 was transformed with a clone harboring V. cholerae rpoS (pFY1), vector alone (pBluescriptKS+), or E. coli rpoS clone (pDEB2). Ampicillin- and kanamycin-resistant transformants were streaked onto MacConkey-lactose plates, yielding lactose-positive (red) colonies (which appear darker).

View larger version (50K): [in this window] [in a new window]   FIG. 2.   (A) Genomic organization of rpoS and its flanking genes on the V. cholerae O1 El Tor chromosome. rpoS is surrounded by nlpD and mutS, based on homologous sequences that flank E. coli rpoS. The transcriptional orientation of each gene is indicated by an arrow. (B) Alignment of amino acid sequences of RpoS from S. typhimurium (S47534), E. coli (Z14969), S. flexneri (U00119), E. carotovora (U66542), Y. enterocolitica (P47765), V. cholerae (this study), and P. aeruginosa (P45684). The amino acid sequences were retrieved from the databases via the accession numbers indicated in parentheses, and alignment was determined by the PileUp program of the Genetics Computer Group. Amino acids which are identical in all of these sequences are shown in the consensus. The V. cholerae RpoS translated sequence overall identity ranges from 60 to 71% for the other RpoS sequences shown.

Insertional inactivation of rpoS. To identify rpoS-associated phenotypes, we inactivated the rpoS gene of the wild-type V. cholerae strain 92A1552 by inserting the cm cassette at the BamHI site of a PCR-generated 1.002-kb fragment corresponding to the rpoS coding region. The resulting construct, rpoS::cm, interrupts translation at amino acid 149 of ^S and prevents transcription downstream of the insertion site. rpoS::cm was then cloned into suicide vector pKNG101, yielding plasmid pFY6. This plasmid was introduced into V. cholerae 92A1552 by conjugation, and Rif^r, Sm^r, and Cm^r transconjugates were selected. Two cointegrate transconjugants containing no plasmid and representing a single homologous recombination event were identified and used for sucrose selection to induce second-site recombination.

View larger version (24K): [in this window] [in a new window]   FIG. 3.   Confirmation of the structure of the rpoS insertion mutation. (A) Physical maps of the rpoS regions of the chromosomes from the rpoS+ wild-type parent strain and the rpoS::cm mutant derivative. The two probes that were used for the hybridization studies shown in panels B and C are indicated under the restriction maps as rpoS and cm, respectively. Restriction enzyme sites: H, HindIII; S, SalI; B, BamHI. (B and C) Southern analysis of the parent and rpoS::cm insertion mutant. Chromosomal DNA was digested with HindIII, separated on a 0.8% agarose gel, blotted, and hybridized with the rpoS (B) or cm (C) probe. (D) Immunoblot analysis to detect of RpoS production by the parent strain (rpoS+), mutant (rpoS::cm), and complemented mutant, using pFY7 (rpoS::cm/pFY7). Total-cell extracts from the indicated strains were separated by SDS-PAGE (12% gel), blotted, and then probed with an antiserum to the RpoS protein of E. coli.

View larger version (18K): [in this window] [in a new window]   FIG. 4.   Diagram of the inserts used to construct the plasmids used for complementation analysis. pFY8 and pFY9 carry complete nlpD and rpoS genes and thus, by analogy to E. coli promoter analysis (29, 32), are predicted to carry both the logarithmic- and stationary-phase-driven promoters of rpoS, indicated by dashed and straight arrows, respectively. Both pFY1 and pFY7 carry only 464 bp of the nlpD gene and thus are predicted to encompass only the stationary-phase-driven promoter, which is indicated by the arrow. pFY7 and pFY9 are low-copy-number plasmids, whereas both pFY1 and pFY8 are high-copy-number plasmids. pFY10 harbors the entire coding sequence of nlpD, its upstream region, and the first 500 bp of rpoS. Construction of these plasmids is described in Materials and Methods.

Survival of the rpoS::cm mutant during environmental stress. rpoS mutants of E. coli, S. typhimurium, Y. enterocolitica, and S. flexneri have been studied by other investigators and found to be more susceptible than the corresponding wild-type strain to a variety of physicochemical conditions that are thought to simulate different host or natural environmental conditions of growth (3, 10, 25, 31, 35, 47, 50, 53). Among the tested conditions were carbon starvation, hyperosmolarity, heat shock, low pH, and oxidative stress. From these results, we reasoned that the V. cholerae rpoS::cm mutant might survive less well than the parent strain under conditions of nutrient deprivation (in particular, carbon starvation), hyperosmolarity, and oxidative stress. To exclude the possibility that the lack of ^S-mediated functions might affect the general fitness of the mutant during nonstressed cultivation, we compared its growth to that of the parent strain in LB broth at 37 and 30°C. No differences in viable plate counts were observed during the exponential or stationary phase of growth (data not shown).

View larger version (15K): [in this window] [in a new window]   FIG. 5.   Effect of rpoS::cm on resistance to carbon starvation, hyperosmolarity, and H2O2. Stress survival of stationary-phase V. cholerae wild-type (open squares), rpoS::cm (filled squares), and rpoS::cm/pFY7 (open circles) bacteria was tested by incubating the cultures in carbon-free ASW (A), in M9 medium without added glucose (B), at high osmolarity (2.4 M NaCl, in LB) (C), and in the presence of H2O2 (15 mM) (D). At indicated times, the viable cell count was determined by plating dilutions onto LB; 100% survival corresponds to the viable cell count determined just prior to exposure to the indicated stress.

rpoS-dependent protein expression. To identify changes in the protein synthesis pattern caused by the rpoS::cm mutation, cultures were intrinsically labeled with [³⁵S]methionine during both the exponential and stationary phases of growth, and expression of labeled proteins was determined by 2D gel electrophoresis.

View larger version (70K): [in this window] [in a new window]   FIG. 6.   2D gel electrophoresis pattern of [35S]methionine-labeled proteins in lysates of wild-type V. cholerae and rpoS::cm strains. Shown are protein expression patterns of the wild type during logarithmic growth (A), the rpoS::cm mutant during logarithmic growth (B), the wild type during stationary phase (C), and the rpoS::cm mutant during stationary phase (D). Bacteria were labeled with [35S]methionine either during logarithmic growth phase or during late stationary phase. Processing of cell extracts and 2D gel electrophoresis were performed as described in Materials and Methods. Proteins expressed only in rpoS+ strains during logarithmic growth or in stationary phase are shown by arrows in panels A and C, respectively. Proteins expressed only by the rpoS::cm mutant during logarithmic (B) or during stationary phase (D) are also indicated by arrows. Proteins whose expression is increased compared to that of the other strain during the specific growth condition are circled. IEF, isoelectric focusing. Sizes are given in kilodaltons on the left of each panel.

Effect of rpoS on the production of the HA/protease. During the course of these experiments, we observed that SDS-PAGE analysis of V. cholerae outer membrane proteins revealed marked degradation of many proteins in lysates of the wild-type strain but not in lysates prepared with the rpoS::cm mutant. We reasoned that the rpoS mutation may have affected either the secretion or production of one or more of the previously described V. cholerae O1 proteases (55). This possibility was examined by measuring the protease activity secreted into culture supernatants by the wild-type strain, the rpoS::cm mutant, and the mutant with complementing plasmids. The results, shown in Table 2, demonstrate that the rpoS::cm mutant produces and/or secretes ~108-fold less protease activity than the wild-type parent strain. The absence of protease activity in the mutant was partially, but not completely restored, by providing either pFY1, pFY8, or pFY9; the resulting complemented strains then exhibited 37.8-, 5.9-, and 9.8-fold, respectively, less protease activity than the parent strain. In contrast, pFY7, which had complemented the stress-related phenotypes shown in Fig. 5, did not complement the protease-deficient phenotype of the mutant. This result points to a possible difference between the transcriptional control of rpoS for genes conferring resistance to stress versus genes required for protease production or secretion. Alternatively, the difference between the protease complementation capacities of pFY1 and pFY7 may have been due to differences in copy number between the two plasmids: pFY1 is present in high copy number, whereas pFY7 is present in low copy number. As expected, when only the cloning vectors pBluescript KS⁺ and pWKS130 were provided to the mutant strain, no increase in protease activity was detected.

View larger version (18K): [in this window] [in a new window]   FIG. 7.   HA/protease production and/or secretion. Immunoblot analysis using an antiserum to the HA/protease was performed on concentrated culture supernatants of the wild-type (wt+) strain, the rpoS::cm mutant, and the mutant strain carrying rpoS-complementing plasmids. HA/protease-negative mutant (HAP-1) was included as a negative control together with its wild-type parent (3083), which served as a positive control.

Intraintestinal survival of the V. cholerae rpoS mutant. To analyze the importance of RpoS for intraintestinal survival, we tested the intestinal colonizing capacity of the mutant and the wild-type parent strain. Equal numbers (3.7 × 10⁵ CFU) of the wild-type parent and the rpoS::cm mutant were combined and administered by gavage to 4- to 5-day-old CFW mice. Twenty hours later, the animals were sacrificed, the small intestine was removed and homogenized, and dilutions of the homogenate were plated onto TCBS agar plates. Individual colonies were then replicate plated onto LB agar containing chloramphenicol (to select for the rpoS::cm mutant) and LB agar without antibiotic (for growth of both the antibiotic-sensitive parent strain and the rpoS mutant). The output ratio of the wild type to the rpoS mutant strain surviving in the murine intestine was then calculated from these plate counts. Using an initial inoculum containing a ratio of wild type to rpoS mutant of 1.0, the output ratio was found to be 0.97 ± 0.49 (average ± standard deviation, n = 12). This result indicates that rpoS is not required for intraintestinal survival under the conditions tested.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

We have shown that an rpoS mutant of a recently isolated V. cholerae O1 El Tor strain is less resistant than its wild-type parent to carbon starvation, hyperosmolarity, and oxidative stress. Further, based on the 2D gel electrophoresis results shown in Fig. 6, we have demonstrated that rpoS_vc seemingly regulates the expression of 25 or more genes during stationary-phase growth. In these respects, the function of rpoS_vc appears to be similar to that of the well-studied rpoS homolog of E. coli, which also regulates the expression of multiple genes, some of which participate in the general stress response (katE, xth, appA, and mcc) (33), are required for starvation-induced changes in cell shape (bolA) (30), or confer an adaptive response to hyperosmolarity (osmB, osmE, otsAB, and treA) (20, 21).

In other respects, the role of rpoS in V. cholerae seems to differ significantly from its apparent role in two enteric pathogenic genera, Salmonella and Shigella. This is exemplified by our finding that the V. cholerae rpoS mutant and the parent strain survive equally well in the murine small intestine. In contrast, an S. typhimurium rpoS mutant was found to be attenuated for virulence in mice (11, 54), probably because it controls the expression of spv plasmid virulence genes that are induced during stationary-phase growth and in the intracellular environment (11). However, further analysis of these contrasting in vivo results reveals that they may not be due to essential differences in the physiological role of rpoS in the two species but rather may have resulted from differences in how each species interacts with the murine intestinal tract. V. cholerae does not normally enter host cells during an infectious episode. Consequently, our use of the murine model was intended to measure rpoS-dependent adaptation to an extracellular, intraintestinal environment that is rich in nutrients, is iso-osmolar, and probably lacks oxidative stress. In contrast, S. typhimurium causes murine enteric fever, which requires survival within host macrophages, an environment that entails oxidative stress through the production of oxygen free radicals. Thus, these seemingly different results may in fact point to a common function for rpoS because resistance to oxidative stress, whether generated by UV radiation in an aquatic environment (2), as may be the case for V. cholerae, or by the oxidative burst of a phagocyte, in both species may require rpoS.

The Shigella rpoS homolog has also been assigned a role in pathogenesis that does not seem to be shared by the V. cholerae homolog. rpoS is required for the resistance of Shigella to acid and thus is plausibly responsible for the small inoculum required to produce shigellosis in humans by stimulating a protective response by organisms transiting the acidic compartment of the stomach (50, 53). We could identify no such function for rpoS_vc (data not shown), perhaps in part because V. cholerae is intrinsically only very weakly resistant to acid.

The most striking functional difference between the rpoS homologs of V. cholerae and most other enteric pathogens is the requirement of rpoS_vc for the production and/or exportation of the HA/protease. We could find only one other example of this function for rpoS in another species, namely, the requirement of rpoS for the expression by Y. enterocolitica of yst, which encodes a secreted heat-stable enterotoxin (25). In the case of V. cholerae, we have not elucidated the mechanism by which rpoS_vc subserves this function. Possibilities that are now under study include a direct effect by rpoS on the transcription of the HA/protease coding sequence or through an indirect effect on the transcription of the hap gene by affecting the levels of its transcriptional activator(s).

We have no explanation for the capacity of pFY1, pFY7, pFY8, and pFY9 to complement the rpoS::cm mutant for resistance to H₂O₂, carbon starvation, and hyperosmolarity but to compensate rather poorly for the effect of the rpoS mutation on the production and/or secretion of the HA/protease (Table 2). In trying to understand this difference, we reasoned that the promoter of rpoS that activates its transcription during the exponential phase (by analogy to the E. coli rpoS promoter) also might be required to complement the HA/protease phenotype. In contrast, the clone carrying less of the 5' upstream promoter region, which by analogy to E. coli would encompass the stationary-phase promoter (carried by pFY7), might suffice to complement those functions most directly related to survival during stress and carbon starvation. Although the longer clone (carried by pFY8 and pFY9), which we predicted would contain the exponential-phase promoter of rpoS_vc, provided better complementation of the HA/protease phenotype, the levels of complementation attained with these plasmids were still 6- to 10-fold less than that for the parent strain. We further ruled out plasmid copy number as a probable cause for partial complementation, since pFY9, which is a low-copy-number plasmid, also provided only incomplete complementation of the protease activity. Explanations for these discrepant results that are now under investigation include the possible differences in the expression of chromosomal versus episomal copies of the rpoS gene and a possible growth phase control over either the secretion or production of HA/protease.

The effect of rpoS on the secretion and/or production of the HA/protease suggests that this sigma factor could be important for the pathogenesis of cholera. Cholera toxin is the quintessential virulence determinant of V. cholerae O1, and its A subunit may need to be activated by the HA/protease before it can function as a secretogogue (14). Therefore, in this manner, its biological activity, if not its production and secretion, may require HA/protease. HA/protease was once considered to be a virulence determinant in its own right because of its capacity to degrade mucin, fibronectin, and lactoferrin (14, 17). However, because an HA/protease-negative mutant was found to be as virulent as the parent strain when tested in an infant rabbit model, it is now believed to function in vivo to detach V. cholerae from epithelial cell surfaces by degrading the receptors for bacterial adhesins (13). This "detachase" role would allow the bacteria to exit an infected host and reenter environmental reservoirs. Alternatively, and in accord with our starvation survival results which suggest a role for rpoS in the survival of V. cholerae in aquatic habitats, another potential function of the HA/protease is the digestion of proteinaceous dissolved organic matter in water, leading to the liberation of amino acid nutritional substrates. Indeed, studies of a marine vibrio have shown that it secretes proteases under conditions of carbon starvation (1), but the possible rpoS dependency of this phenomenon has not been reported. It will be interesting to explore this possibility by comparing the survival of V. cholerae strains mutated in rpoS or the HA/protease gene with the wild-type parent strains in laboratory microcosms that contain complex macromolecules as dissolved organic matter, in concentrations found in natural aquatic habitats, as the only extrinsic source of carbon and nitrogen.

We undertook this study with the preconception that the life style of V. cholerae takes place in one of two loci: the nutrient-rich intestinal lumen or in nutrient-poor aquatic habitats. We further hypothesized that many of the organism's most highly regulated genes would belong either to the pathogenic set of genes (expressed mainly in the intestine) or the environmental set of genes (expressed mainly in contaminated water). Those genes regulated by ToxR, including the tcp operon and ctxAB, would surely belong to the former set (7-9, 16, 52). On balance, rpoS belongs more in the latter set, an assignment supported by its effect on carbon starvation survival, resistance to hyperosmolarity, and H₂O₂, and its apparent lack of effect on intraintestinal survival, and its far more potent effect on gene expression in stationary-phase growth than in exponential-phase growth.