Global Gene Expression and Phenotypic Analysis of a Vibrio cholerae rpoH Deletion Mutant

Bacterial strains and growth conditions.The Vibrio cholerae El Tor clinical isolate N16961 was used as the parental strain throughout this study. The annotated genome of this strain has been published (19). E. coli strain DH5α λpir was used as the host strain for plasmid construction. E. coli SM10 λpir was used as the donor in mating experiments. Cells were grown in Luria-Bertani (LB) medium and stored at −80°C in LB containing 20% glycerol.

The antibiotic concentrations used for bacterial selection were as follows for V. cholerae: 200 μg/ml streptomycin, 50 μg/ml kanamycin, 5 μg/ml chloramphenicol, and 80 μg/ml ampicillin. For E. coli, antibiotics were used at the following concentrations: 50 μg/ml kanamycin, 20 μg/ml chloramphenicol, and 100 μg/ml ampicillin. Arabinose and glucose were added at final concentrations of 0.1 and 0.2%, respectively. In the growth experiments described in Fig. 2, 10μl of serial 10-fold dilutions from freshly thawed frozen stocks were spotted on the indicated plates and incubated for 18 h (for 30°C, 37°C, and 42°C experiments) or for 48 h (for 15°C experiments). X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactoside) was used at a final concentration of 40 μg/ml.

DNA and RNA manipulations.Plasmid DNA was extracted from E. coli with QIAGEN kits. Chromosomal DNA was extracted from N16961 with Genome DNA kits from Q-Biogene. RNeasy kits and RNase-free DNase kits (QIAGEN) were used to purify RNA from exponential-phase cells. RNA was purified from stationary-phase cells with TRIzol reagent (Invitrogen). Contaminating DNA was removed from RNA preparations with the DNA-free DNase I kit from Ambion. Restriction enzymes and T4 DNA ligase were purchased from New England Biolabs; SuperScript II reverse transcriptase was purchased from Invitrogen. PCR was performed with a PTC-200 thermal cycler from MJ Research, with Pfx Platinum polymerase (Invitrogen) or the HotStart Taq Mastermix from QIAGEN. The oligonucleotides for PCR amplification, 5′ rapid amplification of cDNA ends (RACE), and real-time reverse transcription-PCR (RT-PCR) were purchased from the Tufts University Core facility or from Operon (Huntsville, AL). Nucleotide sequencing was performed at the Tufts University Core facility.

Plasmid and strain construction.pRpoH contains the promoterless rpoH gene (vc0150) under the control of the arabinose-inducible P_BAD promoter. It was constructed by subcloning rpoH into pBAD18-Cm (17). The rpoH gene was amplified by PCR from N16961 chromosomal DNA using primers RSP3 and rpoH8 (Table 1). DNA sequencing confirmed that the subcloned rpoH sequence was identical to that of vc0150.

The allele exchange vector pCVD442ΔrpoH::Kn was constructed in the following steps. First, the 5′ and 3′ regions of the rpoH gene were amplified by PCR, using primer pairs rpoH3/rpoH4 and rpoH5/rpoH6 (Table 1). The 5′ end was purified as an XbaI/EcoRI fragment, and the 3′ end was purified as a BamHI/SphI fragment. The kanamycin resistance (Kn^r) cassette was amplified from plasmid pKD4 (8) with primers kan3 and kan5 (Table 1) and digested with EcoRI and BamHI. The three fragments were ligated with XbaI- and SphI-digested pCVD442 (10), a pir-dependent sacB-containing allele exchange vector. The resulting plasmid, pCVD442ΔrpoH::Kn, was then introduced by electroporation into the Mob⁺ E. coli strain SM10 λpir. The rpoH gene of strain N16961 was deleted by allele exchange with pCVD442ΔrpoH::Kn essentially as follows. In brief, the plasmid was initially transferred from SM10 λpir by conjugation into N16961. One of the resulting Ampicillin-resistant (Ap^r) and Kn^r transconjugants was then electroporated with pRpoH. The transformants were grown at 15°C in the presence of kanamycin and chloramphenicol and then plated on sucrose-kanamycin-chloramphenicol plates (at 15°C) to select for cells that had lost the allele exchange vector after homologous recombination. The resulting strain, N16961 ΔrpoH::Kn (pRpoH), was designated LS1.

The structures of the strains and plasmids constructed for this study were all verified by PCR and DNA sequencing.

Plasmid loss assay.LS1 [N16961 ΔrpoH (pRpoH)] and LS2 [N16961 (pRpoH)] cells were grown overnight in LB supplemented with chloramphenicol and arabinose at 22°C. The cells were then diluted (1/1,000) at time zero into fresh medium supplemented with only arabinose and incubated either at 30, 37, or 42°C for 8 h or at 15°C for 32 h. Serial dilutions of the cultures were then plated on streptomycin- and arabinose-containing agar and incubated for 16 h at 30°C. Colonies were counted, and replicas were generated on plates supplemented with streptomycin, chloramphenicol, and arabinose. The plates were incubated at room temperature until colonies appeared, and they were subsequently compared to the master plates, allowing calculation of the ratio of streptomycin- and chloramphenicol-resistant (Sm^r Cm^r) cells to Sm^r cells. The same procedure was used to calculate this ratio for the cells at time zero.

Intestinal colonization assay.LS1 (Lac⁺) and LS2 (Lac⁻) cells were grown overnight at 22°C in LB in the presence of chloramphenicol and arabinose. The cultures were then diluted 1:1,000 into fresh LB supplemented with chloramphenicol and either arabinose or glucose and grown for an additional 1.5 h at 22°C. The experiment was carried out essentially as described previously (43). Briefly, one-to-one mixtures of LS1 and LS2 from the glucose and from the arabinose cultures were intragastrically inoculated into 5-day-old CD-1 suckling mice. In a parallel in vitro competition experiment, 30 μl of the same mixtures were diluted 100-fold in LB medium and grown for 20 h at 37°C. Mice were sacrificed 20 h after inoculation, and the small intestine was harvested and ground, and serial dilutions were plated on LB supplemented with X-gal, arabinose, and streptomycin.

Microarray analyses.To prepare RNA for microarray analyses, LS2 [N16961 (pRpoH)] and LS1 [N16961 ΔrpoH (pRpoH)] cells were grown as follows. First, the strains were grown at 15°C in LB supplemented with arabinose to an optical density at 600 nm (OD₆₀₀) of ∼0.5. Then, the cells were harvested and resuspended in LB supplemented with glucose and incubated for 20 additional minutes at 15°C. Finally, the two strains were subjected to a heat shock at 37°C for 20 min, and then the cells were harvested and RNA was extracted. The subsequent steps of cDNA preparation and labeling, with microarray hybridization and data analysis, were carried out as previously described by Ding et al. (9). The experiment was repeated three times on independently isolated RNA samples, and dye swaps were performed on each RNA sample. Thus, in one experiment Cy3 (Amersham Biosciences) was used to label cDNA from LS1 and Cy5 (Amersham Biosciences) was used to label cDNA from LS2 and vice-versa, generating replicas. The results from the three experiments were averaged, and genes were counted as downregulated in LS1 compared to LS2 if the ratio of LS1/LS2 transcripts was ≤0.5 (see Table 2).

5′ RACE.To map rpoH promoters, 5′ RACE was performed on total RNA extracted from stationary-phase N16961 and from the isogenic rpoE mutant BD1581 (9). The 5′ RACE system kit (Invitrogen) was used according to the manufacturer's recommendations. The specific primer RSP1 (Table 1) was used to generate cDNA, and the nested primers RSP2 and RSP (Table 1) were used for the subsequent PCR steps.

To map the transcription start sites of dnaK and vca0446, 5′ RACE was performed on two of the LS2 RNA samples that were used for the microarray experiments. The specific primers DSP1 (for dnaK) and HSP1 (for vca0446) were used to generate cDNA. The nested primers DSP2 and DSP (for dnaK) or HSP2 and HSP (for vca0446) (Table 1) were used for the subsequent PCR steps.

The final products were purified from an agarose gel and sequenced.

Real-time RT-PCR analyses.The RNA samples used for the microarray experiments were also used for real-time RT-PCR analyses, as previously described (35). The specific oligonucleotide primer pairs GroES2Fd/GroES2Rv, 2675Fd/2675Rv, 0706Fd/0706Rv, and 0177Fd/0177Rv (Table 1) were used for PCR. The primers were designed with Primer3 (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi ) and verified with Amplify software and by using the search for short, nearly exact match option of BLAST (http://www.ncbi.nlm.nih.gov/BLAST/ ). The real-time RT-PCR assays were performed on two different sets of RNA. The reactions were performed and analyzed with an ABIPrism 7000 sequence detection system.

Bioinformatics.The V. cholerae RpoH binding site consensus sequence was generated with BioProspector (27). Bioprospector is designed to identify motifs conserved among groups of sequences. To utilize BioProspector, the user must initially specify the general structure of the motif being sought. The E. coli σ³² binding site consensus sequence motif is comprised of two conserved blocks separated by a gap of variable length (52). The blocks and gap widths used to identify the consensus sequence for the V. cholerae σ³² binding site were based on the E. coli σ³² binding site consensus sequence. Bioprospector was used to search for a consensus motif corresponding to this structure in the 200-bp sequence upstream of each of the RpoH-regulated genes presented in Table 2. The search parameters were adjusted until the motif identified corresponded to the published sequence of the V. cholerae σ³²-regulated htpG promoter (34). The final search parameters were as follows: motif, 2 blocks; width of the first motif block, 6; width of the second motif block, 8; minimum gap between the blocks, 13; maximum gap between the blocks, 16.

Identification of putative σ³² binding sites throughout the V. cholerae genome was then accomplished in two steps. First, the program PATSER (46) was used to search the upstream regions of all V. cholerae genes for sequences corresponding to the consensus motif of each block. Second, PromoterFinder, a C++ program, was developed and used to find those blocks separated by gaps of 13 to 16 bp in length. Each putative promoter was assigned a score corresponding to the average score of its associated blocks as determined by PATSER and was also assigned a rank among the scores of all the promoters.

Sequence comparisons were done with BLAST on the NCBI website (http://www.ncbi.nih.gov/BLAST/ ) and with DnaStrider software. Promoter predictions were performed by using Bprom on the Softberry website (http://www.softberry.com/berry.phtml ).

Microarray accession number.The microarray data have been submitted to the Gene Expression Omnibus database (http://www.ncbi.nih.gov/geo/ ) and assigned the following accession number: GSE6097.

Absence of RpoH induction impairs V. cholerae growth.Generating an rpoH deletion mutant to explore the role of the alternative sigma factor RpoH in V. cholerae physiology proved difficult. Despite several attempts, we were unable to replace the rpoH gene in the sequenced El Tor V. cholerae clinical isolate N16961 with a cassette conferring resistance to kanamycin, via allelic exchange, at temperatures ranging from 15°C to 37°C. However, the same strategy was successful when RpoH, provided by a plasmid that carries rpoH under the control of the arabinose-inducible P_BAD promoter, was introduced into N16961 prior to the allelic exchange (see Materials and Methods). The resulting N16961 ΔrpoH (pRpoH) strain was designated LS1.

To explore the growth phenotype of the V. cholerae ΔrpoH mutant, we assessed LS1 and LS2 [N16961 (pRpoH)] growth in LB broth supplemented with chloramphenicol and either arabinose, to induce rpoH expression, or glucose, to repress rpoH expression, at 22°C and 42°C. Transcriptome analyses presented below confirmed that these conditions altered rpoH mRNA levels as expected. We measured changes in OD₆₀₀ and counted CFU on plates containing arabinose (Fig. 1). Growth rates were calculated as the change in OD₆₀₀/time. At both temperatures, the growth rates of wild-type strain LS2 were approximately the same in arabinose and glucose (Fig. 1A and B), suggesting that overexpression of rpoH does not alter V. cholerae growth. In contrast, the growth rate of the rpoH deletion mutant LS1 at 42°C was ∼1.7 times greater in arabinose than in glucose. At 42°C, the number of LS1 CFU was approximately five times greater in arabinose- than in glucose-containing cultures (Fig. 1C). At 22°C, although the growth rates of LS1 were similar in arabinose and glucose (Fig. 1B), there was a somewhat higher (∼2.0-fold) number of LS1 CFU in the arabinose-containing cultures (Fig. 1D). This discrepancy may, at least in part, result from the mild filamentation observed in LS1, but not LS2, at both 22° and 42°C in glucose-containing media (data not shown). Overall, these observations suggest that rpoH has a significant effect on V. cholerae growth at 42°C and a more subtle influence at 22°C, perhaps by influencing cell division.

• Open in new tab
• Download powerpoint

FIG. 1.

Comparison of the growth and colony-forming ability of wild-type (LS2) and ΔrpoH (LS1) V. cholerae at 22°C and 42°C. LS1 (black lines) and LS2 (gray lines) cells were grown in LB with chloramphenicol and arabinose at 22°C to an OD₆₀₀ of 0.03. The cells were then harvested, washed, and resuspended in fresh medium supplemented with chloramphenicol and either arabinose (solid lines) or glucose (dotted lines) and grown in a shaker incubator at 42°C (A and C) or 22°C (B and D). Each graph is representative of at least two independent experiments; the data in panels A and B and those in panels C and D come from two different experiments. Time zero indicates the time at which the cells were placed in the incubator at either 22°C or 42°C. Note that the y axis is a logarithmic scale.

V. cholerae σ³² may be essential at all temperatures.To determine whether V. cholerae cells are able to survive in the absence of RpoH, we compared the frequency of pRpoH loss from the rpoH deletion mutant LS1 and the wild-type strain LS2 at temperatures ranging from 15°C to 42°C. In these experiments, overnight cultures of LS1 and LS2 were grown at 22°C in LB supplemented with arabinose and chloramphenicol, the marker carried by pRpoH. Then, the cultures were diluted in fresh LB supplemented with only arabinose and incubated at 30°C, 37°C, or 42°C for 8 h or at 15°C for 32 h. Regardless of the temperature, all of the ΔrpoH cells present at the end of the experiment carried pRpoH (Fig. 2, black bars). In contrast, there was marked loss of pRpoH from the wild-type cells. After incubation at 30°C or 37°C, only 0.32% and 0.01% of LS2 cells, respectively, harbored pRpoH. The marked difference in the stability of pRpoH in LS1 and LS2 in the absence of selection indicates that regardless of temperature, provision of rpoH in trans is beneficial to the ΔrpoH mutant. Interestingly, a larger fraction of the wild-type cells still carried the plasmid after incubation at 42°C or 15°C than at 30°C or 37°C. Wild-type cells harboring the empty vector pBAD18-Cm showed a similar pattern of plasmid loss (Fig. 2). The similarity in the patterns of pRpoH and pBAD18-Cm loss from wild-type cells suggests that the excess of RpoH provided by pRpoH does not explain the differences in pRpoH loss observed at different temperatures; instead, temperature-dependent differences in plasmid segregation likely account for the marked difference in pRpoH loss from LS2 over the temperature range we tested.

• Open in new tab
• Download powerpoint

FIG. 2.

Maintenance of a plasmid-borne copy of rpoH in wild-type versus ΔrpoH V. cholerae at various temperatures. Retention of the Cm^r plasmid pRpoH in LS1 (Δ rpoH) and LS2 (wild type) after 32 h at 15°C, or after 8 h at 30°C, 37°C and 42°C, is shown as the normalized ratio of Sm^r Cm^r to Sm^r CFU. The black bars represent LS1 cells, the light gray bars represent LS2 cells, and the dark gray bars represent wild-type cells carrying the empty vector pBAD18-Cm. Normalization was done by dividing the ratios of Sm^r Cm^r/Sm^r cells after incubation at the various temperatures by the ratio of Sm^r Cm^r/Sm^r cells at the time of inoculation (t = 0). Note that the y axis is a logarithmic scale. The results are the mean values of two experiments, and the error bars represent standard deviations.

RpoH is required for V. cholerae resistance to kanamycin at high temperatures.During the course of our experiments investigating the growth of the V. cholerae rpoH deletion mutant, we made the unexpected observation that rpoH influences V. cholerae resistance to the aminoglycoside antibiotic kanamycin in a temperature-dependent fashion. LS1 contains a Kn^r gene in the place of rpoH. When we plated LS1 on plates containing either arabinose or glucose, the results were remarkably different depending on whether we used kanamycin to select for LS1 CFU (Fig. 3A; compare the top and bottom halves of each panel). We counted the number of LS1 [ΔrpoH (pRpoH)] CFU on LB plates containing either arabinose, to induce RpoH expression, or glucose, to repress RpoH expression, at four different temperatures, in the presence or absence of kanamycin. The results of a representative spot dilution experiment are shown in Fig. 3A. At 15°C, RpoH induction was dispensable for V. cholerae colony formation, independently of the presence of kanamycin. However, there was a dramatic reduction in the number of LS1 CFU on the medium containing glucose versus arabinose at 30°C and 37°C on plates containing the antibiotic. Finally, at 42°C, no LS1 CFU were detected on the plates supplemented with glucose and kanamycin (Fig. 3A, bottom half of the bottom panel). In contrast, LS1 colony formation was not impaired below 42°C in the presence of streptomycin (data not shown), another aminoglycoside antibiotic like kanamycin, or chloramphenicol, another antibiotic that, like kanamycin, inhibits protein synthesis (Fig. 3B). Thus, RpoH appears to be essential for V. cholerae resistance to kanamycin at high temperatures or for V. cholerae resistance to high temperatures in the presence of kanamycin.

• Open in new tab
• Download powerpoint

FIG. 3.

Temperature influences the requirement of an rpoH deletion mutant for RpoH in the presence of kanamycin. In panel A, 10 μl of serial 10-fold dilutions from a freshly thawed frozen stock of LS1 were spotted on LB and LB-kanamycin (LB Kn) plates. The numbers in the upper panel indicate the dilution factor. Plates on the left contain arabinose (0.1%), and plates on the right are supplemented with glucose (0.2%). The CFU/ml of the freshly thawed frozen stocks plated on LB supplemented with chloramphenicol are shown in the bar graph (B). Gray bars indicate plates supplemented with arabinose, and black bars indicate plates supplemented with glucose. Note that the y axis is a logarithmic scale. The results are the mean values of at least two experiments, and the error bars represent standard deviations.

RpoH is critical for V. cholerae growth in the intestine.We carried out competition experiments to compare the growth of LS1 and LS2 in the intestines of suckling mice to assess the importance of RpoH for V. cholerae growth in vivo. We designed these experiments to approximate the temperature shift that V. cholerae may undergo during infection of the human gastrointestinal tract. To prepare the inocula for these experiments, the two strains were initially grown overnight at 22°C in LB supplemented with chloramphenicol and arabinose, to ensure maintenance of pRpoH and expression of rpoH. Prior to mixing of the strains for in vivo inoculation, they were grown for an additional 1.5 h at 22°C in LB supplemented with chloramphenicol and either arabinose, to promote rpoH expression, or glucose, to repress rpoH. One-to-one mixtures of LS1 and LS2 grown in glucose or LS1 and LS2 grown in arabinose were intragastrically inoculated into 5-day-old suckling mice. After 20 h, the mice were sacrificed, their small intestine was harvested and ground, and serial dilutions of the mixture were plated on LB supplemented with X-gal and streptomycin. In a parallel in vitro competition experiment, the same inocula were diluted in LB medium and grown at 37°C for 20 h to compare the growth of these two strains in rich medium, at approximately the same temperature as the mice.

As shown in Fig. 4, LS1 was profoundly attenuated for growth in vivo. Induction or repression of rpoH expression prior to inoculation did not rescue LS1 colonization. In both cases, the average LS1/LS2 competitive index (the ratio of LS1 to LS2 CFU recovered from individual mice divided by the ratio of LS1 to LS2 CFU in the inocula) for growth in vivo was ∼0.001 (Fig. 4). These numbers represent an underestimate of the true degree of the in vivo growth impairment of the rpoH mutant, since several mice did not have any detectable LS1 in the small intestine. Thus, RpoH appears to be critical for V. cholerae growth in vivo. The maintenance of pRpoH in vivo and in vitro was similar (Fig. 1 and data not shown). All the LS1 CFU recovered from the intestinal homogenates harbored pRpoH whereas nearly all LS2 CFU lost the plasmid. The finding that induction or repression of RpoH production prior to inoculation did not appear to significantly influence LS1 colonization may suggest that RpoH is required throughout the duration of infection and not strictly for the shift in growth conditions that the organism undergoes at the beginning of the experiment. In LB, LS1 was also outcompeted by LS2 in the vitro competition experiment. Thus, provision of rpoH in trans without inducer does not appear to provide adequate RpoH to enable wild-type growth of the rpoH mutant in rich medium when competing with wild-type cells. However, the magnitude of the competitive defect of LS1 versus LS2 in vivo was more than 150 times lower (0.16 versus 0.001) than in vitro. The most likely explanation for this dramatic difference is that there is a greater necessity for RpoH and the genes that it controls for growth in the intestine than in rich medium.

• Open in new tab
• Download powerpoint

FIG. 4.

Comparison of the growth of LS1 and LS2 in LB at 37°C and in suckling mice. Prior to inoculation, the strains were grown at 22°C in either glucose or arabinose. Then, the two strains were mixed 1:1 and inoculated into suckling mice or LB broth. The ratio of LS1 to LS2 CFU in intestinal homogenates or in the overnight LB broth cultures was divided by the ratio of these strains in the inocula to yield the competitive indices. Open triangles and diamonds represent mice that had no detectable LS1 CFU.

The V. cholerae RpoH regulon.To begin to define the V. cholerae RpoH regulon, we compared the transcriptomes of LS1 [N16961 ΔrpoH (pRpoH)] and LS2 [N16961 (pRpoH)] following a temperature up-shift. The strains were initially grown at 15°C in medium containing arabinose, to induce RpoH expression, until the culture reached an OD₆₀₀ of ∼0.5. Then, the cells were shifted to medium containing glucose, to repress RpoH expression, and incubated at 15°C for an additional 20 min and subsequently shifted to 37°C for 20 min prior to extraction of RNA for cDNA synthesis. Thus, this comparison should yield knowledge of the RpoH-dependent genes that are induced by a heat shock similar to the change in temperature that occurs upon infection of the human gastrointestinal tract by V. cholerae from a natural water environment.

The transcripts of 49 genes were more abundant in wild-type cells than in ΔrpoH cells 20 min after the temperature up-shift (ratio of mutant/wild-type transcript, ≤0.5) (Table 2). As expected, the rpoH transcript was much less abundant (sevenfold) in the LS1 cells than in the LS2 cells. This confirmed that rpoH transcription was reduced by the shift to glucose in LS1 compared to LS2, which contains the native rpoH gene under the control of its own promoter. While 10 different categories of genes are found in the list of genes whose temperature up-shift induction depends on RpoH, more than 60% of the genes identified here are either involved in protein fate, like chaperones and proteases, or are of unknown function.

Most of the genes encoding products involved in protein fate have orthologs in the E. coli RpoH regulon, as recently defined by Zhao et al. (54), Nonaka et al. (33), and Wade et al. (47). Four other genes (vc0706, vca0752 [trxC], vc0977, and vc2735) also appear to be common between the two regulons. Although vc2735 was classified as encoding a “conserved hypothetical protein” by Heidelberg et al. (19), the product of this gene is homologous to E. coli Hsp15 (encoded by yrfH), a heat shock protein that possesses an S4-like RNA binding domain. vc2735 is likely to be in an operon with vc2736, a gene also annotated as encoding a conserved hypothetical protein; vc2736 bears similarity to the E. coli Hsp33 protein, a redox-regulated chaperone. Although vc2736 did not meet the criteria to be included in Table 2, it is probably RpoH regulated, as its transcript level was reduced more than 1.5-fold in LS1 relative to LS2.

There are also several differences between the V. cholerae and E. coli RpoH regulons. Not all genes in the V. cholerae RpoH regulon that have E. coli homologs were found to be RpoH regulated in E. coli (e.g., vc0466, vc0467, vc0468, and vc0469). Twelve genes in the V. cholerae RpoH regulon do not have E. coli homologs; the products of these genes mainly belong to the “conserved hypothetical protein” category. Some of the RpoH-regulated genes that are involved in protein fate in E. coli, such as ftsH, were absent from our list. Although most of the genes whose products are involved in regulating the levels and activity of RpoH are present in our list (such as DnaK, DnaJ, and GroESL), the absence of ftsH suggests that RpoH regulation may differ in V. cholerae and in E. coli, as FtsH degrades RpoH in E. coli (20, 45).

Some of the RpoH-regulated genes listed in Table 2 may contribute to V. cholerae pathogenicity. Indeed, two genes, vca0446 and vca0447 (a hemagglutinin and a hemagglutinin-associated protein, respectively), in Table 2, are found in the “pathogenesis” category. Although vca0065 is classified in the “conserved hypothetical protein” category, it appears to be a secreted Zn-metalloprotease and thus may play a role in V. cholerae virulence. Furthermore, RpoH-regulated genes that contribute to the cellular response to the various stresses encountered in vivo are also likely to contribute to pathogenicity. For example, the RpoH-regulated chaperone DnaK has been implicated in the virulence of V. cholerae (6) and other pathogens and is known to be a target of the human immune response to V. cholerae (38).

To verify the accuracy of the microarray data, we performed real time RT-PCR analyses on transcripts of four randomly chosen genes from Table 2. For these experiments we used two of the three RNA sample sets that were used for the microarray experiments. As seen in Fig. 5, the relative levels of vca0819, vc2675, vc0706, and vc0177 transcripts were all reduced in the ΔrpoH mutant relative to the wild-type strain. In fact there is a reasonably good concordance with the ratios of the transcripts ascertained with the RT-PCR and microarray techniques. This concordance lends considerable credence to the microarray data.

• Open in new tab
• Download powerpoint

FIG. 5.

Relative levels of transcripts of four RpoH-regulated genes in ΔrpoH versus wild-type V. cholerae. RNA was isolated from LS1 and LS2 20 min after a temperature up-shift from 15°C to 37°C. The names of the transcripts quantified by real-time RT-PCR are indicated below the x axis. The transcript levels were normalized to the amount of rpoB transcripts in the cells. The results presented are the mean values of two experiments. The error bars represent the standard deviations.

A putative RpoH binding site consensus sequence.To begin to assess which genes in Table 2 are under the direct control of σ³², we used a bioinformatics-based approach (Fig. 6A; also see Materials and Methods) to define a putative σ³² binding site consensus sequence, cTTGAA(N_13-16)(a/c)CCATat(a/t), where lowercase letters indicate less conservation among the sequences aligned (shown in Fig. 6B). Higher sequence conservation was observed for the −35 region than for the −10 region. This consensus sequence is similar to the three reported E. coli σ³² binding consensus sequences (33, 47, 54). Eighteen potential σ³² binding sites were found upstream of 16 of the 49 genes in Table 2 (highlighted in pink in Fig. 6A). Thus, approximately one-third of the V. cholerae σ³² regulon appears to be under the direct control of this sigma factor. More than half of the genes found to be associated with a putative σ³²-regulated promoter are involved in protein fate.

• Open in new tab
• Download powerpoint

FIG. 6.

Bioinformatics approach for determination of a putative V. cholerae σ³² binding site consensus sequence. (A) Protocol used to determine a consensus sequence for the V. cholerae RpoH binding site shown in panel B. Sequence databases are shown as blue ovals, computer programs are shown in black ovals, and outputs are shown as magenta ovals. The consensus sequence was used to search upstream of all V. cholerae ORFs to identify putative RpoH-dependent promoters. The graph at the bottom of panel A shows the score distribution of these putative RpoH-dependent promoters. The pink lines correspond to putative promoters identified upstream of RpoH-regulated genes in Table 2. (B) Consensus sequence of the RpoH binding site displayed with WebLogo (http://weblogo.berkeley.edu ) (7, 40). The height of each column indicates the sequence conservation at that position, while the height of symbols within the stack indicates the relative frequency of each nucleic acid at that position. The number of times each nucleotide is found at each position in the consensus is shown below the sequence. N_(13-16) indicates the gap between the two blocks.

To test the accuracy of our proposed σ³² promoter consensus sequence, we used 5′ RACE experiments to determine the transcription start sites of two genes predicted to be directly regulated by σ³², dnaK and vca0446. Two potential σ³² binding sites are predicted upstream from dnaK. We identified two apparent transcription start sites for this gene. These sites, P1 and P2, are located 108 and 39 bp upstream from the putative dnaK translation start codon, respectively (Fig. 7A). P1 is located 6 bp downstream from the distal end of a predicted −10 motif, and P2 is 7 bp downstream from the second predicted −10 motif. Similarly, the apparent transcription start site for vca0446, which is located 27 bp upstream from the translation start codon, is 9 bp downstream from the distal end of the predicted −10 motif (Fig. 7B). The relative positions of the predicted σ³² binding sites upstream of dnaK and vca0446, with respect to the experimentally defined transcription start sites for these genes, are similar to the distances found between the −10 motifs and transcription start sites in bacterial promoters recognized by sigma factors from the σ⁷⁰ family to which RpoH belongs (18). Furthermore, in the recent comprehensive analysis of the E. coli σ³² regulon, Nonaka et al. found that the distance between the −10 motif and the transcription start site for σ³²-regulated promoters varies from 6 to 9 bp (33). These observations provide experimental support for our proposed σ³² binding site consensus sequence.

• Open in new tab
• Download powerpoint

FIG. 7.

dnaK and vca0446 transcription start sites. (A) Two dnaK transcription start sites were identified by 5′ RACE. P1 and P2 are positioned 6 and 7 bp downstream from the −10 motif of the two predicted σ³² binding sites, respectively. (B) One transcription start site upstream from vca0446 was identified by 5′ RACE. P1 is located 9 bp downstream from the −10 motifs of the predicted σ³² binding site. The translation start codon for each gene is boxed, and the likely Shine-Dalgarno (SD) sequence is indicated by a dashed line. The beginning of each ORF is shown with uppercase letters.

A bioinformatics approach to identifying new RpoH-regulated genes.Using the consensus sequence we identified, we took a bioinformatics approach to complement our microarray data and potentially find additional genes that may be directly regulated by σ³². First, we searched for the σ³² binding site consensus motif in the regions upstream of all V. cholerae open reading frames (ORFs) as a means of identifying possible σ³²-regulated promoters; 665 potential σ³² binding sites that likely correspond to σ³²-regulated promoters were found. Each of these was given a score corresponding to the average of the scores assigned by PATSER to its associated block 1 and block 2 sequences (Fig. 6A). In the PATSER scoring system, the closer a sequence is to the consensus matrix, the higher the score. Each of the putative promoters was then ranked according to its score, and promoter scores were plotted versus promoter ranks to create the graph shown at the bottom of Fig. 6A. Almost all of the 18 putative promoters upstream of genes in Table 2 are clustered among the highest-scoring promoters in Fig. 6A. The fact that promoters predicted upstream of RpoH-regulated genes ranked among the highest-scoring promoters supports the idea that the consensus motif identified by Bioprospector is specific to genes regulated by RpoH and may correspond to a potential RpoH binding site. However, this finding could also reflect the inherent circularity in our approach, i.e., the high scores of putative promoters found in regions upstream of RpoH-regulated genes might simply reflect the fact that these regions were among the sequences used by Bioprospector to identify the consensus motif. To address this possibility, we searched for a consensus motif upstream of 49 randomly selected V. cholerae genes by the approach used to determine the RpoH binding site consensus motif. We then used PATSER and PromoterFinder to search for sequences corresponding to this consensus motif upstream of all V. cholerae genes. Of 720 sequences identified, those identified in these 49 regions had an average rank of 241, which corresponds to the 66th percentile. In contrast, putative σ³²-dependent promoters predicted upstream of RpoH-regulated genes had an average rank of 41 out of 665 predicted promoters, corresponding to the 94th percentile. These findings suggest that the high-scoring putative promoters predicted upstream of the 16 RpoH-regulated genes correspond to actual RpoH binding sites.

Many of the 665 predicted σ³² binding sites are likely to be false positives because there is significant divergence among sequences recognized by sigma factors. However, this list also probably includes several σ³²-dependent promoters that were not identified in our microarray experiments. We examined whether any of the genes associated with the top 20% of putative σ³² binding sites (and not listed in Table 2) have homologs in the E. coli σ³² regulons as defined in three recent studies (33, 47, 54). We found three genes, vc0636, vc0350, and vc0566, homologous to E. coli ftsJ, hflC, and b0161 (which encodes a putative protease), respectively, that met these criteria. On the other hand, several genes that were associated with high-scoring putative σ³² binding sites do not have homologs in E. coli but are also likely to be under the direct control of σ³² in V. cholerae. For example, the region upstream of vc1379 has one of the highest-scoring potential σ³² binding sites. The ratio of LS1 to LS2 for vc1379 transcripts was 0.6 and therefore did not meet our parameters to be included in the regulon; however it is probable that this gene is under the control of RpoH. Furthermore, BLAST analysis suggests that the product of this gene is distantly related to a DnaJ-like protein. Thus, even though this bioinformatic method yields many false predictions, our findings suggest that computational approaches can be a valuable complement to experimental studies.

rpoH transcription start site.We also performed 5′ RACE experiments to explore the V. cholerae rpoH promoter region. rpoH transcription in E. coli utilizes four transcription start sites and several regulatory proteins (12, 22, 31, 48). Three of the four promoters are recognized by σ⁷⁰, whereas the fourth is σ²⁴ dependent (49). Our previous work suggested that rpoH transcription in V. cholerae is also influenced by σ²⁴ (RpoE), as rpoH mRNA levels were elevated in a V. cholerae strain with a deletion of rseA (9), the gene encoding the RpoE anti-sigma factor. We carried out 5′ RACE experiments on RNA extracted from N16961 and from an isogenic rpoE mutant strain (9). These experiments suggest that there are three rpoH promoters (designated P1, P2, and P3 in Fig. 8). P2- and P3-derived transcripts were detected in both wild-type and rpoE mutant cells. P2 and P3, the two promoters most proximal to the rpoH translation start site, are probably σ⁷⁰ dependent, as there are reasonable σ⁷⁰ −10 and −35 recognition elements upstream from these transcription start sites. These two promoters are also predicted to contain transcription start sites (scores of 4.22 and 3.32, respectively) by the Bprom program (available at www.softberry.com ). On the other hand, P1 has features of a σ²⁴-dependent promoter, as described by Kovacikova and Skorupski (26), and, as expected, no putative σ²⁴-dependent rpoH transcript was amplified from the rpoE cells.

• Open in new tab
• Download powerpoint

FIG. 8.

Features of the rpoH promoter region. Three rpoH transcription start sites, P1, P2, and P3, were identified by 5′ RACE. The likely −10 and −35 elements for P2 and P3 suggest that these promoters depend on σ⁷⁰. Potential σ²⁴ −10 and −35 sequences were identified upstream of the P1 transcription start site. The RpoH translation start site is boxed, and the likely Shine-Dalgarno (SD) sequence is indicated by a dashed line. The beginning of the rpoH ORF is shown with uppercase letters. The asterisks denote the ftsX translation stop codon. The dashed box indicates a putative DnaA binding sequence.

The rpoH promoter region in E. coli and other bacteria contains binding sites for regulators such as DnaA, CRP, and CytR (36). A putative DnaA box overlaps the −10 region of P2, but no other candidate regulatory binding sites were identified in the V. cholerae rpoH promoter region.