|Home | About | Journals | Submit | Contact Us | Français|
Recent work has shown that in addition to cholera toxin (CT) and the toxin-coregulated pilus (TCP), other cytotoxic proteins in Vibrio cholerae also cause disease symptoms, and this is particularly evident in strains lacking CT. One such protein is the hemolysin encoded by hlyA. Here we show that, like CT and TCP, HlyA is repressed by the quorum-sensing-regulated transcription factor HapR. This repression occurs on two levels: one at the transcriptional level that is independent of the metalloprotease HapA and one at the posttranslational level that is mediated by HapA. The transcriptional regulation is significantly more apparent on solid media than in liquid cultures. This is the first time that hemolysis has been shown to be directly regulated by quorum sensing in V. cholerae, and it is interesting that, like other virulence factors, HlyA is also repressed by HapR, which is expressed late in infection.
Vibrio cholerae is the etiologic agent of the severely dehydrating diarrheal disease cholera. Epidemic cholera is caused predominantly by the O1 serogroup, although a new serogroup, O139, has recently emerged as a cause of cholera epidemics. The O1 serogroup is divided into two biotypes known as classical and El Tor (19). Isolates from the sixth pandemic were almost all of the classical biotype, while isolates from the seventh and current pandemic are mostly of the El Tor biotype (5). Both biotypes produce cholera toxin (CT), an ADP-ribosylating protein that induces the secretion of massive amounts of fluid from intestinal epithelial cells (8). This virulence factor is responsible for the profuse watery diarrhea that is characteristic of the disease. Both biotypes also produce the toxin-coregulated pilus (TCP), which is necessary for efficient colonization (15, 40). However, naturally occurring V. cholerae strains without CT are still sometimes associated with enterocolitis and extraintestinal symptoms (2, 30). Furthermore, CT-negative El Tor strains developed for use as vaccines still cause mild to severe diarrhea as well as inflammation in humans (37, 39). In contrast, a CT-negative classical strain-based vaccine is much less reactogenic (21). A study with the suckling mouse model also showed that both TCP-negative classical and El Tor V. cholerae strains are cleared from the small intestine, but the TCP-negative El Tor strain is able to persist at significantly higher levels than the TCP-negative classical strain in the cecum and large bowel (1). Thus, it is highly likely that there are toxins specific to the El Tor biotype that are important for pathogenesis. There are at least three potential toxins that are expressed at higher levels in El Tor than in classical strains and therefore may be responsible for the difference in phenotypes seen in vivo: a hemolysin encoded by hlyA, a hemagglutinin (HA)/protease encoded by hapA, and a multifunctional autoprocessing RTX toxin (4, 22, 32, 38). In one study using a streptomycin-fed murine model of disease, it was shown that in the absence of CT, HlyA is the predominant cause of death among these three toxins, with RTX playing a minor role and HapA having little to no effect on lethality (31). However, HlyA does not significantly affect virulence in a murine pulmonary cholera model (9).
Hemolysins, by definition, have lytic activity against red blood cells in vitro, and they are often implicated in virulence. However, hemolysins are often cytotoxic to other mammalian cells types as well, and sometimes it is their action against other cell types that is more important for their function as virulence factors. V. cholerae HlyA has been shown to have both vacuolating and cytocidal activity against a number of cell lines, including human intestinal Caco-2 cells (26, 33, 48). HlyA has also been shown to induce apoptosis in Caco-2 cells (33). Animal studies have shown that purified HlyA is capable of inducing intestinal fluid accumulation in the adult rabbit ligated intestinal loop model, infant rabbit model, and suckling mouse model. However, unlike the diarrhea seen with CT, HlyA-induced diarrhea usually contains both mucus and blood (16). Thus, it is likely that HlyA is a virulence factor important for V. cholerae symptoms other than the classical profuse watery diarrhea.
In V. cholerae, the expression of the primary virulence factors, CT and TCP, is controlled by changes in cell density via a signaling cascade known as the quorum-sensing pathway (47). Each bacterium produces small molecules known as autoinducers that increase in concentration with increasing bacterial numbers. At low cell density, the components of the quorum-sensing pathway act as kinases, and phosphorylated LuxO activates the transcription of small RNAs that destabilize hapR mRNA (20). At high cell density, autoinducers bind to cognate sensors on the bacterial surface and induce conformational changes in the sensors (29). The net result is dephosphorylation of LuxO. The small RNAs are no longer transcribed, the hapR transcript is stabilized, and HapR protein is expressed in high quantities so that it is able to mediate its downstream effects. HapR is a TetR family transcriptional regulator that controls many phenotypes, including repression of virulence factor production (CT and TCP) and biofilm formation and activation of protease production (HapA) (12, 17, 25, 46, 47). It has been shown that HapR levels increase during the course of infection in a murine model, suggesting a corresponding decrease in virulence factor production (23). It is thought that this repression of virulence factors coincides with clearance of the bacteria from the host, when virulence factors are no longer needed. Recently, our lab published a bioinformatics-based approach for identifying novel targets of HapR (41). Many new members of the HapR regulon were identified, indicating that quorum sensing controls many more functions than was previously recognized. In this study, we show for the first time that quorum sensing directly regulates V. cholerae hemolytic activity at both the transcriptional and posttranslational levels, which may represent yet another mechanism by which quorum sensing regulates virulence in this organism.
All V. cholerae strains used in this study were derived from El Tor C6706 (6). For the hemolysis, azocasein, and luminescence time courses, single colonies were inoculated from plates into liquid Luria broth (LB) with the appropriate antibiotics and grown at 37°C under shaking conditions. The plasmids used to measure transcription of hlyA (VCA0219) and the divergently transcribed gene VCA0218 were constructed by PCR amplifying the intergenic region lying between VCA0218 and VCA0219 and cloning the product into pBBR-lux in both orientations (13). The resulting plasmids were then introduced into the various V. cholerae mutants by conjugal transfer. In-frame luxO, hapR, hapA, prtV, and hlyA deletion mutants were constructed as previously described (9, 10, 42, 47). The VCA0880 mutant was constructed as described previously (46). The VCA0883 mutant was constructed by cloning an internal fragment of VCA0883 into pVIK112 (18), and the resulting plasmid was integrated into the VCA0883 locus by homologous recombination. The His6-HlyA overexpression plasmid was constructed by PCR amplifying the hlyA coding sequence and cloning into pET32a (Novagen). Ptac-hapA was constructed by PCR amplifying the hapA coding sequence and cloning into pMal-c2x (New England Biolabs). The strain containing a chromosomal copy of Ptac-hapA was constructed by inserting Ptac-hapA into the ~2-kb intergenic region between VCA0104 and VCA0105 through double crossover.
Hemolysis assays were performed as previously described (7) with modifications. Culture supernatants were collected by centrifugation at the desired time points. Rabbit blood (HemoStat Laboratories, Becton-Dickinson BBL) was prepared by washing three times with phosphate-buffered saline (PBS). Culture supernatants (245 μl) were combined with 5 μl of washed rabbit blood and incubated for 1 h at 37°C under shaking conditions. LB was substituted for culture supernatant as a negative control, and 1% Triton X-100 was used as a positive control. After incubation, samples were centrifuged, and hemoglobin release was measured by determining the optical density at 540 nm (OD540). The reading for the LB negative control was subtracted from each sample reading. Percent hemolysis was calculated by taking the Triton X-100 control as 100% hemolysis. The percentages were then normalized by the OD600 of the cultures at the time supernatants were collected.
Culture samples were collected, and the OD600 was measured at the indicated time points. Culture supernatants were then harvested by centrifugation. One hundred microliters of culture supernatant was combined with 100 μl of azocasein (Sigma-Aldrich) (5 mg/ml in 100 mM Tris, pH 8) and incubated at 37°C for 1 h. LB was used in place of the culture supernatant as a negative control. After incubation, 400 μl of 10% trichloroacetic acid was added to each sample, and the resulting mixtures were centrifuged for 15 min. Three hundred microliters of each of the resulting supernatants was transferred to 350 μl of 525 mM NaOH, and the OD442 was determined. The reading for the LB negative control was subtracted from each sample reading. Azocasein units were calculated by multiplying by 100 and normalizing to the OD600 of the culture at the time of sample collection.
His-tagged HlyA was purified over an Ni column according to the manufacturer's instructions (Novagen). Clear cell lysate from Escherichia coli either containing a vector or expressing HapA was prepared by growing 100 ml of isopropyl β-d-1-thiogalactopyranoside (IPTG)-induced cultures to an OD600 of ~1. Cell pellets were then resuspended in 10 ml of PBS, lysed with lysozyme, and cleared by centrifugation. The indicated combinations of 12 μl of purified HlyA, 4 μl of E. coli cell lysate, and/or 1 μl of 0.5 M EDTA were incubated with an appropriate amount of PBS for a total reaction volume of 24 μl. Reaction mixtures were incubated at 37°C for 1 h and then separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The gel was then stained with Coomassie blue.
Various V. cholerae strains containing either the hlyA-luxCDABE or the VCA0218-luxCDABE reporter fusion were grown at 37°C under shaking conditions. Periodically, luminescence was read using a Bio-Tek Synergy HT spectrophotometer. Results are reported as light units normalized by OD600.
Maltose-binding protein fused to the N terminus of HapR (MBP-HapR) was expressed and purified as described previously (41). PCR products containing the hlyA (VCA0219) promoter region as well as the region further upstream (the VCA0218 promoter region) were digested with EcoRI and end labeled using [α-32P]dATP and the Klenow fragment of DNA polymerase I. Binding reactions were performed as described previously (41), except that when indicated, 1 μg of unlabeled hlyA promoter DNA was added in the reaction mixture. Radioactivities of free DNA and HapR-DNA complexes were visualized by using a Typhoon 9410 variable model imager (Molecular Dynamics).
In a recent study published by our lab, we identified a new HapR target, VCA0880, which resides in an operon with VCA0881, VCA0882, and VCA0883 (41). All of the genes in this operon are annotated as encoding hypothetical proteins. However, a BLAST search using their amino acid sequences revealed homology between VCA0883 and both the L1 component of hemolysin BL and the nonhemolytic enterotoxin NheB, both of which are found in Bacillus cereus (11). To investigate the possibility that HapR controls hemolytic activity via regulation of this operon, we tested a VCA0880 mutant and a VCA0883 mutant for their abilities to lyse red blood cells relative to that of a wild-type strain. We also included a hapR mutant in the study since we previously showed that HapR is a strong activator of transcription from the VCA0880 promoter. While neither the VCA0880 nor the VCA0883 mutant had hemolytic phenotypes different from that of the wild type, the hapR mutant consistently exhibited higher levels of hemolysis than the wild-type strain (data not shown). This indicates that hemolytic activity in V. cholerae is in fact regulated by hapR, but this regulation is independent of VCA0880 and VCA0883. Additionally, the hemolytic activity of the wild-type strain varied dramatically on different days, which we hypothesized may have been due to the fact that we tested cultures at different cell densities on different days. This is consistent with a role for HapR in the regulation of hemolysis, since the expression of HapR in wild-type V. cholerae is known to vary with cell density due to quorum sensing. The V. cholerae quorum-sensing pathway has never previously been shown to control hemolysis, so we decided to further investigate the mechanism of this regulation.
Since HapR protein levels increase with increasing cell density due to quorum sensing, detailed time course experiments were conducted to examine hemolytic activity throughout the growth curves of a wild-type strain and a hapR deletion mutant. We also tested a luxO deletion mutant, which has no LuxO to transcribe the small RNAs that destabilize hapR mRNA and thus constitutively overexpresses HapR. Therefore, each strain has a different pattern of HapR expression, with the luxO mutant expressing HapR at high levels even at low ODs, the wild-type strain expressing HapR at a later time point after autoinducer concentrations have reached a certain threshold, and the hapR mutant not expressing HapR at all. We found that despite some HapR-independent regulation of hemolysis (all strains showed an initial increase in hemolytic activity), hemolytic activity is inversely correlated with HapR levels, which is consistent with HapR repression of hemolytic activity. A luxO hapR double mutant mimicked the hapR mutant in terms of its hemolytic phenotype, indicating that the decreased hemolysis seen in the luxO mutant is in fact due to increased HapR levels as opposed to some other downstream target of LuxO (Fig. (Fig.11).
The data in Fig. Fig.11 suggest that HapR represses hemolytic activity, but this repression may be either direct or indirect. Previous studies of a related species, V. vulnificus, have indicated that a metalloprotease, VvpE, may inactivate a hemolysin, VvhA, but there have been conflicting reports (34, 35). Studies of V. mimicus and V. tubiashii have also shown a similar relationship between a protease and a hemolysin (14, 27). In V. cholerae, HapR increases the expression levels of two genes encoding metalloproteases, hapA and prtV (17, 46). Thus, we tested deletion mutants of both of these metalloproteases for changes in hemolytic activity relative to wild type. The prtV mutant behaved essentially the same as the wild-type strain, while the hapA mutant exhibited elevated hemolytic activity, much like the hapR mutant (Fig. (Fig.2A).2A). Additionally, constitutive expression of HapA from the chromosome can complement a hapR deletion strain in terms of hemolytic activity (Fig. (Fig.2B).2B). Taken together, these results strongly suggest that HapR's regulation of hemolysis is mediated through its regulation of HapA expression. Since HlyA is the predominant hemolysin in V. cholerae, we thought it was likely that HapR regulates hemolysis by controlling HlyA activity or expression. Indeed, the hapA hlyA double mutant showed a complete loss of hemolytic activity (Fig. (Fig.2A),2A), which is consistent with HlyA being the hemolysin controlled by the quorum-sensing network.
While hapA transcription is activated by HapR, it is also dependent on other factors, including the cyclic AMP receptor protein and RpoS (36, 45). Thus, hapA expression does not always directly correlate with HapR levels. HapA is the predominant protease found in the supernatant of wild-type V. cholerae cultures, and cleavage of azocasein is often used as a quantitative measure of HapA activity. Previous studies have shown that even though HapR is constitutively expressed in luxO mutants, HapA activity is not detected in luxO mutant supernatants at low cell density (47). However, HapA activity is detected at lower cell densities and at higher levels in luxO mutants than in wild-type strains. This would help explain the existence of a small peak in hemolytic activity at early time points for the luxO mutant (Fig. (Fig.1),1), since HapA activity may not be induced until slightly after the initial rise in hemolytic activity. To better correlate proteolytic activity with hemolytic activity, we measured the hemolytic and proteolytic activities of the various mutants at time points where there were the greatest differences between strains. Protease activity was undetectable for all strains at an OD600 of ~0.7 (data not shown), a time point when all strains, including the luxO mutant, had approximately equal levels of hemolytic activity (Fig. (Fig.11 and and2A).2A). At an OD600 of ~0.9, high protease activity was observed in luxO culture supernatants but not in those of any of the other strains (Fig. (Fig.2D),2D), and this was correlated with very low hemolytic activity in the luxO mutant and high hemolytic activity in all the other strains (Fig. (Fig.2C).2C). Finally, at an OD600 of ~1.2, protease activity was high in both the wild-type strain and the luxO mutant (Fig. (Fig.2D),2D), and this was correlated with repressed hemolysis in both strains (Fig. (Fig.2C).2C). At the same time, hapR, luxO hapR, and hapA strains exhibited high hemolytic activity (Fig. (Fig.2C)2C) but no protease activity (Fig. (Fig.2D).2D). Thus, proteolytic activity and hemolytic activity are inversely related. Since protease activity is almost completely abolished when hapA is deleted and hemolytic activity is almost completely abolished when hlyA is deleted, the protein products of these two genes are the main factors responsible for the proteolytic and hemolytic phenotypes exhibited by the various strains. Together, these data strongly support the hypothesis that quorum sensing controls hemolytic activity by activating the expression of the metalloprotease HapA, which represses HlyA.
Since HapA is a metalloprotease, one way that it could be repressing HlyA activity is by degrading the protein. To test this hypothesis, in vitro degradation assays were performed using purified HlyA. Cell lysate from E. coli containing a vector had no effect on HlyA when coincubated with the protein, whereas cell lysate from E. coli expressing HapA resulted in the degradation of HlyA after coincubation. EDTA, a chelating agent which renders HapA inactive (3), inhibits the degradation of HlyA during incubation with HapA (Fig. (Fig.3).3). Taken together, these results indicate that HapR represses hemolytic activity by activating the transcription of the metalloprotease HapA, which degrades the hemolysin HlyA.
In the previous section, we provided evidence for posttranslational regulation of HlyA indirectly by HapR. To determine if transcriptional regulation is involved, we constructed a transcriptional reporter by fusing the hlyA promoter upstream of luxCDABE. Expression from the hlyA promoter was then determined for each of the quorum-sensing mutants. Interestingly, hlyA expression appeared to be regulated by the quorum-sensing network at the transcriptional level. The luxO mutant showed repressed transcription from the hlyA promoter compared to the wild type, while the luxO hapR double mutant mimicked the wild type (Fig. (Fig.4A).4A). This result indicates that the repression of hlyA transcription seen in the luxO mutant is mediated by increased HapR levels. Even though the luxO and luxO hapR mutant phenotypes indicate a HapR-dependent transcriptional regulation of hlyA, there did not appear to be a difference in hlyA transcription between the wild-type strain and the hapR mutant in liquid culture (Fig. (Fig.4A).4A). This was inconsistent with the clear difference in hemolytic activity between the wild type and the hapR mutant seen previously (Fig. (Fig.1).1). To confirm the inconsistency between hlyA transcription and hemolytic activity, both luminescence and hemolytic activity were measured simultaneously for wild-type, hapR, luxO hapR, and hapA strains containing the hlyA-luxCDABE reporter at an OD600 of ~1.2. Indeed, the previously observed high hemolytic activities in hapR, luxO hapR, and hapA mutants compared to the wild type were still seen in these strains even though there was no clear difference in hlyA transcription (data not shown). This further supports the hypothesis that there are two distinct levels of regulation of hemolytic activity by HapR, and these two modes of regulation differ temporally.
At the end of the hlyA-luxCDABE time course experiment, each culture was spotted onto an LB agar plate, and light production was examined after overnight growth at 37°C. While the luxO mutant was almost completely dark, both the hapR and luxO hapR mutants were clearly brighter than the wild-type strain (Fig. (Fig.4B).4B). These are the expected phenotypes for these strains if HapR transcriptionally represses hlyA expression and are a bit different from what was seen in liquid culture. Additionally, the hapA mutant was dim like the wild-type strain, indicating that HapA's effect on hemolytic activity is not mediated transcriptionally.
It is interesting to note that VCA0218, which is transcribed divergently from hlyA (VCA0219), is also annotated as a hemolysin. Since their promoter regions both lie in the intergenic region separating them, we thought it was possible that HapR binds to this intergenic region and regulates the transcription of both hemolysins. However, unlike transcription from the hlyA promoter, transcription from the VCA0218 promoter did not seem to be dependent upon HapR (data not shown).
To determine if HapR is able to directly bind to the hlyA promoter, a gel retardation assay was performed using purified HapR protein and a radiolabeled hlyA promoter fragment. HapR was able to bind to the hlyA promoter fragment, and this binding was specific, since nonradiolabeled hlyA promoter DNA was able to compete for binding to HapR (Fig. (Fig.4C).4C). Consistent with the promoter-luxCDABE data, HapR did not bind to the DNA fragment directly upstream of VCA0218.
In this work, we show for the first time that hemolysis is a phenotype controlled by quorum sensing in V. cholerae. Specifically, HapR, a transcriptional regulator expressed at high cell density, represses HlyA at both the transcriptional and posttranslational levels. HapR's posttranslational repression of hemolysis is mediated via the metalloprotease HapA. HlyA is produced as a preprotoxin that is subsequently activated by two cleavage steps to produce the fully active form of the hemolysin (43). One study showed that HapA, as well as all the other tested proteases, including alpha-chymotrypsin and papain, was able to cleave and activate pro-HlyA. However, the N-terminal amino acid of mature HlyA resulting from cleavage by all of the tested proteases, including HapA, differed from the N-terminal amino acid of mature HlyA purified from culture supernatants, indicating that the endogenous protease that processes HlyA to its active form has yet to be identified (28). Also, longer periods of incubation of HapA with HlyA led to decreased hemolytic activity. Our data clearly show that HapA represses hemolytic activity in vivo and that HapA degrades HlyA in vitro. Since HapA does not affect hlyA transcription, our current model for HapA's regulation of hemolytic activity is that HapA decreases the amount of active HlyA posttranslationally by cleaving the protein. We believe that the differences seen between our in vitro experiment and the one showing activation of HlyA by HapA may be due to differences in the relative concentrations of the two proteins (perhaps a lower concentration of HapA properly processes HlyA, while a higher concentration results in degradation). Given the phenotype of the hapA mutant, it is clear that physiological concentrations of HapA decrease HlyA activity rather than increase it. A study of V. mimicus revealed a very similar situation, in which an endogenous metalloprotease cleaves and activates a hemolysin at early time points but inactivates the hemolysin upon further incubation. In that case, it was also true that deletion of the protease gene resulted in higher rather than lower levels of hemolytic activity, so it was concluded that that metalloprotease is not responsible for proper processing of the hemolysin and in fact leads to degradation of the hemolysin (27).
In addition to repressing HlyA activity through HapA, we show that HapR also represses transcription from the hlyA promoter directly. A review of previously published data showed that hlyA had in fact been identified in microarrays comparing the gene expression profiles of various quorum-sensing mutants (44, 47). We go one step further and show that this regulation is transcriptional (as opposed to purely posttranscriptional) and also show that HapR directly binds to the hlyA promoter. The difference in hlyA-luxCDABE phenotypes seen in liquid culture versus on solid media was intriguing, and we hypothesize that HapR is able to repress hlyA transcription only at high levels and that in the luxO mutant that threshold is reached while in the wild type it is not, at least not in liquid culture. On solid media, it appears that the wild type does accumulate high enough levels of HapR to repress hlyA relative to the hapR deletion mutant. This may be due to an alteration in signaling when bacteria are attached to a solid surface, or it may be due to autoinducers being trapped by the agar, which would increase regional autoinducer concentrations. Increased autoinducer levels would trigger increased HapR expression. A similar phenomenon of autoinducer trapping and elevated levels of HapR is seen in biofilms (24). Regardless of the mechanism, it is clear that assaying for phenotypes in liquid versus solid media is not always the same, and it would be worthwhile to take this into consideration during experimental design and interpretation.
It is interesting to note that the hlyA promoter was not identified as a direct target of HapR in our previous bioinformatics study (35). This was because although the hlyA promoter contained a region similar to one of the HapR binding motifs, it was different at one base pair that was highly conserved across all of the other confirmed targets and which was therefore regarded as an important nucleotide for HapR binding. Thus, the hlyA promoter was not selected for experimental verification. Since we have now shown that the hlyA promoter can in fact be bound and regulated by HapR, we can modify our HapR consensus binding motif to reflect the fact that it is not essential for that particular nucleotide to be conserved.
In summary, we have shown that HapR represses HlyA hemolytic activity at both a transcriptional level and a posttranslational level. Thus, it seems that there must be some advantage to low levels of hemolytic activity in V. cholerae, either during the infectious cycle or while it is in its environmental reservoir. Previous studies with both mouse and rabbit models have indicated that HlyA is a virulence factor of significant physiological importance during infection, and now we show that it is regulated similarly to the well-studied V. cholerae virulence factors CT and TCP, in that all three are repressed by HapR. Since HapR levels increase during infection, expression of these virulence factors presumably decreases. It is possible that this transcriptional repression by HapR is a mechanism to conserve energy, since these virulence factors will no longer be needed at the end of the infectious cycle when the bacteria are exiting the intestines to enter the environment. However, once a protein is already made, posttranslational degradation does not have any benefits in regard to energy, and this may indicate that there is a reason for HlyA repression beyond simply conservation of energy. Interestingly, HapA expression late in infection is thought to enhance bacterial release from the intestines since HapA is thought to have “detachase” activity, and this study suggests that HapA has an additional role in degrading a hemolysin that is likely not necessary for environmental survival. At the conclusion of our previous study identifying novel components of the HapR regulon (41), we suggested that our consensus sequences could be generalized to reveal even more HapR targets, and indeed, we have now identified yet a new role for quorum-sensing regulation.
We are grateful to Rahul Kulkarni for insightful discussions and to Tao Cai for help with constructing strains. We also thank J. J. Mekalanos and K. J. Fullner for providing the hapA and hlyA deletion constructs used in this study.
This study was supported by NIH/NIAID R01 (AI072479).
Editor: V. J. DiRita
Published ahead of print on 26 October 2009.