The quorum sensing transcription factor AphA directly regulates natural competence in Vibrio cholerae

Summary

Many bacteria use population density to control gene expression via quorum sensing. In Vibrio cholerae, quorum sensing coordinates virulence, biofilm formation, and DNA uptake by natural competence. The transcription factors AphA and HapR, expressed at low and high cell density respectively, play a key role. In particular, AphA triggers the entire virulence cascade upon host colonisation. In this work we have mapped genome-wide DNA binding by AphA. We show that AphA is versatile, exhibiting distinct modes of DNA binding and promoter regulation. Unexpectedly, whilst HapR is known to induce natural competence, we demonstrate that AphA also intervenes. Most notably, AphA is a direct repressor of tfoX, the master activator of competence. Hence, production of AphA markedly suppressed DNA uptake; an effect largely circumvented by ectopic expression of tfoX. Our observations suggest dual regulation of competence. At low cell density AphA is a master repressor whilst HapR activates the process at high cell density. Thus, we provide deep mechanistic insight into the role of AphA and highlight how V. cholerae utilises this regulator for diverse purposes.

Keywords:

Vibrio cholerae – Sequence motif analysis – DNA transcription – Population density – Genetic footprinting – Deoxyribonucleases – RNA polymerase – DNA footprinting

Introduction

Vibrio cholerae colonises two distinct habitats; the human intestine and aquatic ecosystems [1]. In the aquatic niche, the microbe forms biofilms on chitinous surfaces [2,3]. This induces expression of the gene regulatory protein TfoX [4–7]. Natural competence, the ability to acquire exogenous DNA from the environment, is triggered as a result [8,9]. Specifically, TfoX stimulates production of a type IV pilus that extends to bind, and retracts to internalise, exogenous DNA [10–13]. Genes encoding important cell envelope factors ComEA and ComEC, are also TfoX regulated [14]. Biofilms disperse upon entering the human gut [1,15]. This triggers the expression of virulence factors. Notably, the toxin co-regulated pilus (TCP) and cholera toxin (CT) are produced [16]. These factors are encoded by the tcpPH and ctxAB operons respectively.

Quorum sensing detects changes in bacterial population density reported by auto-inducer molecules [17]. This information is used to modify patterns of gene expression [18]. For example, CqsS is a membrane bound sensor kinase that detects cholera auto-inducer 1 (CAI-1) [19]. When kin are scarce, and CAI-1 levels low, CqsS triggers a regulatory cascade, which cumulates in expression of five quorum regulatory RNAs [20]. These RNA molecules activate translation of AphA, a PadR family transcription factor with an N-terminal winged helix-turn-helix DNA binding motif [20–22]. In turn, AphA activates expression of tcpPH [23]. This event ignites the entire virulence gene expression programme [16]. Surprisingly, given the central role of AphA, the regulator is poorly understood. For instance, transcriptome analysis found just 6 operons controlled by AphA in V. cholerae, 5 of these indirectly [24]. Conversely, in closely related Vibrio harveyi, perturbation of aphA impacted transcription of hundreds of genes [22,25]. The DNA binding properties of AphA are also incompletely defined. Three studies have proposed a DNA consensus motif that binds AphA [24,26,27]. Though there are similarities, the sequence differs in each report. Thus, whilst a key factor in V. cholerae, the extent and mechanistic basis of regulation by AphA is not understood.

In this study, we sought to better understand the DNA binding and gene regulatory properties of AphA in V. cholerae. Using chromatin immunoprecipitation (ChIP) and DNA sequencing (ChIP-seq) we mapped genome-wide DNA binding by AphA. This revealed a palindromic consensus for AphA recognition. Interactions can involve single targets or co-operative AphA binding to adjacent sites. Though discovered as an activator of pathogenicity, AphA mainly targets genes related to cell surface physiology, including a subset required for natural competence. These target genes encode components of the type IV pilus, the competence regulator TfoX, and the quorum sensing factor CqsS. The mechanistic details vary but, at each locus, AphA acts as a transcriptional repressor. We conclude that AphA plays a key role by coupling natural competence to population density in V. cholerae.

Results

Genome-wide distribution of AphA in Vibrio cholerae

To identify genes targeted by AphA we used ChIP-seq. The data are illustrated in Fig 1A. In each plot, genes are shown as blue lines (outer two tracks) and the binding profile of AphA is in teal (inner track). We identified 49 binding peaks for AphA. The peaks were not distributed evenly; 27% were located within a 0.5 Mbp section of chromosome II accounting for only 12% of the total genome (Fig 1A, right arm of chromosome II). To assess data validity, we determined the position of each peak with respect to the nearest gene start codon (Fig 1B). As expected, AphA peak positions cluster at the 5' ends of genes. Counterintuitively, this was true even though two thirds of peaks were located within coding sequences (Fig 1B, inset). This apparent contradiction arises because many peaks close to start codons are within adjacent genes. Examples of such peaks are shown in Fig 1C. Next, we extracted DNA motifs from the peak sequences using MEME. A single significant (E = 2.2⁻⁷) DNA motif was found (Fig 1D). This palindromic sequence (5'-ATGCAA-N₄-TTGCAT-3') likely represents the preferred AphA binding target. We determined the distance between each occurrence of the motif and the centre of the ChIP-seq peak. As expected, the motif is biased to the centre of the ChIP-seq signal (Fig 1E, teal bars). The MEME analysis also identified motifs that were not statistically significant. These sequences were randomly distributed with respect to AphA peak centres. An example is shown as a control in Fig 1E (grey bars). The position of each peak, adjacent genes, and AphA binding motif is listed in Table 1. The distribution of functions associated with AphA targeted genes is shown in Fig 1F. We found that AphA primarily targeted genes for other regulatory proteins and components of the cell envelope. Most interestingly, AphA bound at three loci adjacent to genes known to influence DNA uptake by natural competence (Table 1). Briefly, these loci are the intergenic region between VC0857 and VC0858, the VC1153 (tfoX) promoter region, and the regulatory DNA for VCA0522 (cqsS). Importantly, purified AphA also recognised these targets specifically in vitro (S1 Fig). Hence, we next sought to understand the role of AphA at each locus.

AphA binds a single site at the intergenic region between VC0857 and VC0858

The genes VC0857 and VC0858 encode homologues of the minor pilins pilE and fimT. Such proteins are hypothesised to initiate assembly of the DNA uptake pilus and reside at the tip of the pilus fibre [28,29]. Indeed, the VC0858 and VC0857 gene products contribute to pilus-DNA interactions, which occur specifically at the pilus tip [13]. The ChIP-seq data for AphA binding at the intergenic region between VC0857 and VC0858 is shown in Fig 2A. The sequence of the intergenic region is shown in Fig 2B. To precisely identify the AphA binding site we used DNAse I footprinting (Fig 2C). The data show that AphA binding protects a 25 bp region of DNA from digestion by DNAse I. The footprint coincides with a DNA sequence (ATGAAT-N₄-TTGCAT) that matches our motif for AphA binding at 10 of 12 positions (boxed in Fig 2B). The sequence also coincides precisely with the centre of the ChIP-seq peak (asterisk in Fig 2B).

AphA prevents binding of RNA polymerase to the VC0857/VC0858 intergenic region and inhibits transcription

The previously defined transcription start sites (+1) for VC0857 and VC0858 are denoted by bent arrows in Fig 2B [30]. The corresponding promoter DNA elements are underlined. The AphA binding site overlaps the transcription start site for VC0858 and the promoter -35 element for VC0857 (Fig 2B). We reasoned that binding of AphA would be incompatible with binding of RNA polymerase to the regulatory region. To test this, we used electrophoretic mobility shift assays (EMSAs). The data show that AphA retarded migration of the regulatory DNA fragment (Fig 2D, compare lanes 1 and 2). A substantial change in electrophoretic mobility was evident when RNA polymerase was added instead of AphA (Fig 2D, compare lanes 1 to 3). Inclusion of AphA in incubations with RNA polymerase resulted in DNA migration similar to that with AphA alone (compare lanes 2 and 4). Hence, AphA appears to interfere with the binding of RNA polymerase. To measure the impact of AphA on promoter activity we used in vitro transcription assays. Hence, the DNA sequence in Fig 2B was cloned, in either the forward or reverse orientation, upstream of the λoop terminator in plasmid pSR. Transcripts generated from the cloned DNA can be quantified after electrophoresis. The 108 nt RNAI transcript derives from the pSR replication origin and serves as an internal control. We were unable to detect transcripts from the VC0858 promoter. Conversely, a transcript of the expected size was generated by the VC0857 promoter (Fig 2E, lane 1). Addition of AphA abolished production of the transcript but not synthesis of the RNAI control (Fig 2E, lane 2). We also fused the regulatory fragment to lacZ in plasmid pRW50T. Next, V. cholerae strains with or without aphA were transformed using the plasmid derivatives. As with our in vitro analysis, we detected no transcription from the VC0858 promoter. Hence, we can infer little about potential regulation of VC0858 by AphA. However, lacZ expression was driven by the VC0857 promoter. Recall that AphA is active at low bacterial population densities. Hence, we measured β-galactosidase activity in different phases of growth. The results are shown in Fig 2F. In wild type cells, VC0857 promoter activity increased rapidly once the culture reached the optical density of ~0.8. (Fig 2F, solid line). In cells lacking AphA, the promoter was active at a markedly lower population density (Fig 2F, dashed line). We conclude that AphA is a repressor of VC0857 expression and exerts this effect by occluding the binding site for RNA polymerase (Fig 2G). Interestingly, whilst VC0857 is activated by quorum sensing auto-inducer molecules, which prevent AphA expression, VC0858 is unresponsive (Table 1) [31].

AphA co-operatively binds two adjacent sites at the tfoX promoter

We next turned our attention to the regulatory region upstream of the tfoX gene. Briefly, TfoX is a master activator of all genes required for natural competence [11]. Hence, the decision to express tfoX is a major checkpoint. The ChIP-seq data for AphA binding upstream of tfoX are shown in Fig 3A. The corresponding gene regulatory region is shown in Fig 3B. The centre of the AphA binding peak is again denoted by an asterisk. Note that expression of tfoX is almost completely dependent on CRP binding to a site (orange in Fig 3B) centred 83.5 bp upstream of the tfoX transcription start site (bent arrow in Fig 3B) [32]. We first used DNAse I footprinting to pinpoint AphA binding to the tfoX regulatory region (S2A Fig, lanes 2–4). The AphA footprint occurs between promoter positions -73 and -110. It is important to note that the upstream boundary of the footprint is demarked by a subtle change in the pattern of DNAse I digestion; protection of the base at position -109 and DNAse I hypersensitivity at position -110. This transition is marked by the teal arrow adjacent to lane 4 in S2A Fig. Since the AphA footprint extends over 37 bp it likely represents AphA binding two adjacent sites. The proposed sites are labelled AphA I and AphA II in Fig 3B. The sequences coincide precisely with the centre of the ChIP-seq peak for AphA binding. Sites AphA I (5'-CAACAA-N₄-TTGACG-3') and AphA II (5'-GTGATA-N₄-TCTCAT-3') match the consensus for AphA binding at 6/12 and 7/12 positions respectively. Hence, these seem comparatively poor binding targets. To understand how AphA recognises these sequences we mutated site AphA I or AphA II (red in Fig 3B). We then used DNAse I footprinting to investigate the consequences. Mutations in the upstream AphA I site changed only the upstream half of the large AphA footprint (S2A Fig, lanes 5–7). Hence, the hypersensitive band at position -110 did not appear. We also observed poor protection of a band at position -100 (see triangle to the right of lane 7 in S2A Fig). Mutations in the downstream AphA II site had pronounced consequences. First, the mutations changed the overall pattern of DNAse I sensitivity (compare lanes 2 and 8 in S2A Fig). Second, the mutations rendered AphA unable to bind to either site I or site II (S2A Fig, lanes 8–10). We conclude that AphA binds its two adjacent sites at the tfoX regulatory region co-operatively. Hence, mutations in site AphA II, a closer match to the consensus, abolish recognition of both targets.

AphA and CRP compete for overlapping binding sites upstream of the tfoX promoter

Strikingly, the region of the tfoX promoter bound by AphA overlaps the binding site for CRP (orange in Fig 3B). We reasoned that AphA and CRP may compete for binding. To test this, we further utilised DNAse I footprinting (Fig 3C). Fortuitously, although CRP and AphA bound similar locations, the footprints produced were easily distinguished. Thus, CRP binding resulted in DNAse I hypersensitivity at positions -83 and -92 upstream of the tfoX promoter (Fig 3C, lane 3). As already noted, AphA binding protected the DNA and induced DNAse I hypersensitivity at promoter position -110 (Fig 3C, lane 4). The sites of DNAse I hypersensitivity due to CRP and AphA binding respectively are shown by orange and teal triangles to the right of Fig 3C. Next, we added increasing concentrations of AphA to incubations containing the tfoX promoter DNA fragment and CRP (Fig 3C, lanes 5–9). As the concentration of AphA increased so did the occurrence of DNAse I hypersensitivity at position -110. Concomitantly, DNAse I hypersensitivity at positions -83 and -92, due to CRP binding, was reduced. In parallel experiments we measured binding of AphA and/or CRP to the tfoX promoter region using EMSAs (Fig 3D). Addition of AphA or CRP to incubations altered migration of the tfoX promoter DNA fragment during electrophoresis. Importantly, the degree to which migration altered was different for each protein (compare lanes 1–3 in Fig 3D). This is most likely because CRP bends the DNA by 90° whilst AphA has little effect [21,33,34]. Addition of AphA reduced the abundance of complexes due to CRP and increased the abundance of the complexes due to AphA (Fig 3D, lane 4). Thus, AphA and CRP compete for binding the same section of the tfoX gene regulatory region. Note that AphA could outcompete CRP even if the latter was present at higher concentrations (Fig 3D). We conclude that AphA is likely an anti-activator of tfoX expression (i.e. the repressor targets the activator rather than RNA polymerase directly).

AphA prevents activation of tfoX expression mediated by CRP

To investigate the effects of AphA and CRP on tfoX promoter activity we used in vitro transcription assays. As expected, activation of tfoX transcription by CRP was evident (Fig 3E, lanes 1–4). Addition of AphA abolished activation by CRP (lanes 5–8). To investigate repression in vivo we utilised the lacZ reporter plasmid described above. In wild type cells tfoX promoter activity increased in line with culture optical density (Fig 3F, solid line). The pattern of tfoX promoter activity was different in the absence of AphA (Fig 3F, dashed line). In particular, β-galactosidase activity increased to 5-fold higher levels in the early- to mid-exponential phase of growth. Our model for regulation of tfoX promoter activity by AphA and CRP is shown in Fig 3G.

AphA and CRP bind the cqsS regulatory region in unison

The final AphA target selected for characterisation was adjacent to cqsS. The CqsS protein is at the top of the regulatory cascade triggered by the quorum sensing auto-inducer molecule CAI-1 [20]. At high population densities the cascade prevents expression of downstream genes including tfoX and aphA. The ChIP-seq signal for AphA binding upstream of cqsS is shown in Fig 4A and the sequence of the regulatory region is shown in Fig 4B. We again used DNAse I fooprinting to dissect binding of AphA (S2B Fig). The footprint due to AphA was 37 bp in length (S2B Fig, lanes 2–4). This was indicative of two adjacent AphA sites (labelled I and II in Fig 4B). Mutations made in each site are shown in Fig 4B (red text) and footprints done using the mutated DNA fragments are in S2B Fig. Mutations in the upstream AphA I site completely abolished binding of AphA to both targets (S2B Fig, lanes 5–7) whilst mutations in the downstream AphA II site only prevented binding to the second occurrence of the motif (S2B Fig, lanes 8–10). Hence, we again conclude that AphA recognises two sites co-operatively at the cqsS promoter region. Importantly, the two sites overlap precisely with the DNAse I footprint and the centre of the ChIP-seq peak (asterisk in Fig 4B). The AphA sites also overlap the cqsS transcription start site (bent arrow in Fig 4B) [30]. We noticed that the cqsS regulatory region contained an 8/10 match to the consensus sequence for CRP binding (orange in Fig 4B) centred 41.5 bp upstream of the cqsS promoter. This is intriguing because CRP enhances expression of CqsA, which synthesises CAI-1 detected by CqsS [35]. Hence, we sought to understand if CRP bound this site. The results of DNAse I footprinting experiments are shown in Fig 4C. As expected, CRP (lane 3) and AphA (lane 4) produced footprints at their respective target sites. Addition of AphA to incubations containing CRP resulted in footprinting of both the AphA and CRP targets (lanes 5–7). Similar results were obtained in parallel EMSAs (Fig 4D). Hence, AphA (lane 2) and CRP (lane 3) both individually bound to the cqsS DNA fragment. A super-shifted complex was observed when the proteins were added in unison (lane 4).

AphA and CRP oppositely regulate the cqsS promoter

To understand the effects of AphA and CRP on cqsS promoter activity we first used in vitro transcription assays (Fig 4E). A transcript was only generated from the cqsS promoter in the presence of CRP, albeit at low levels (lanes 1 and 2). In the presence of CRP and AphA this transcript was undetectable (lanes 4–6). Because the cqsS promoter was poorly active in our in vitro transcription analysis we also used KMnO₄ footprinting. This detects DNA opening by RNA polymerase at promoter -10 elements during transcription initiation. The results are shown in Fig 4F; the appearance of bands is indicative of DNA melting. Such bands were only observed in the presence of both CRP and RNA polymerase (lane 6). Addition of AphA to incubations with CRP and RNA polymerase prevented promoter unwinding (lanes 7–9). The low level transcription detected in vitro (Fig 4E) was recapitulated using a cqsS::lacZ fusion in vivo (Fig 4G). Hence, only 120 Miller units were detected for experiments with the cqsS promoter; 20-fold lower than equivalent experiments with the tfoX promoter (compare Figs 3F and 4G). Surprisingly, β-galactosidase activity in lysates of wild type and ΔaphA cells were similar (compare solid and dashed lines in Fig 4G). We conclude that AphA is likely to be a repressor of the cqsS promoter. However, specific conditions may be required to detect such repression in vivo (Fig 4H). We note that addition of quorum sensing auto-inducer molecules, which block production of AphA, activate cqsS expression (Table 1) [31].

Deleting aphA enhances natural competence at low cell density

Whether V. cholerae cells act individually or as a group is coupled to population density by the regulator LuxO [20,36]. Consequently, the LuxO^D61E derivative (earlier misnamed LuxO^D47E) locks cells in a low density state. This mutation is frequently exploited to study behaviours specific to this mode of life [22,37]. Our data suggest that AphA represses the regulatory cascade triggering natural competence at low population density. To test this, we introduced luxO^D61E into V. cholerae E7946 and a derivative lacking aphA. We then measured the frequency of transformation by natural competence for each strain. The result of the experiment is shown in Fig 5A (horizontally lined bars). As expected, the E7946 strain encoding LuxO^D61E was poorly transformable. Consistent with our model, deletion of aphA triggered a >500-fold increase in transformation frequency. We also measured the effect of deleting dns; a gene encoding an endonuclease expressed at low cell density to degrade any DNA obtained by natural transformation (compare striped and speckled bars) [38]. There was no significant effect of deleting dns when aphA was present. A 6-fold increase in transformation frequency was apparent when dns was lost from aphA null cells.

Expression of aphA reduces natural competence at high cell density

We next examined transformation of E7946, and the derivative lacking aphA, in the context of the wild type luxO allele. These strains transition to a high cell density, and become naturally competent, upon colonisation of chitinous surfaces. In this scenario, wild type E7946 transformed efficiently. However, because aphA is not expressed at high cell density, deleting the gene had no effect (Fig 5A, open bars). We reasoned that differences in transformation frequency would be observed if aphA was expressed ectopically. To achieve this, V. cholerae strain E7946 was transformed with plasmid pAMCFaphA that encodes the C-terminally 3xFLAG tagged AphA used in our ChIP-seq experiments. Importantly, the level of AphA generated from this plasmid precisely matches that of chromosomally encoded AphA (S3 Fig). As a control, we utilised plasmid pAMNFaphA encoding N-terminally 3xFLAG tagged AphA, which cannot bind DNA (see control ChIP-seq data). Expression of the active C-terminally tagged AphA reduced transformation frequency by ~1,300-fold. Comparatively, N-terminally tagged AphA had little effect (compare open and black bars in Fig 5A).

Uncoupling TfoX expression from AphA regulation largely restores natural transformation

We next sought to understand which of the AphA regulatory events described above (Figs 2–4) resulted in the loss of competence phenotype due to constitutive AphA production. First, we focused our attention on tfoX repression by AphA. If responsible for reduced competence, natural transformation should be restored by uncoupling tfoX expression from AphA regulation. To do this we replaced the native tfoX promoter with the IPTG inducible tac promoter (P_tac-tfoX) in strains containing pAMCFaphA or pAMNFaphA. Expression of tfoX almost completely abolished the effect of constitutive AphA production on natural transformation (compare black and grey bars in Fig 5A). We conclude that AphA mediated tfoX repression is the primary cause of reduced natural transformation in our experiments.

Constitutive expression of AphA does not impact pilus production

Our focus turned to the residual 6-fold effect of AphA observed in the presence of ectopic tfoX expression (see the difference between grey bars in Fig 5A). We reasoned that attenuated pilus activity, due to repression of VC0857 by AphA, might be responsible. To monitor pilus production, we utilised V. cholerae encoding PilA^S67C [13]. The cysteine substitution facilitates in vivo labelling of PilA with the fluorescent dye AF488-mal. Since PilA is a major component of the DNA uptake apparatus, this allows visualisation of pili in live cells. Recall that TfoX is an activator of VC0857 [11]. To avoid indirect effects of AphA, mediated by repression of tfoX, we again utilised P_tac-tfoX. The supplementary S1 and S2 Movies, and the representative images in Fig 5B, show dynamic pilus events (white arrows). We compared cells expressing the non-functional N-terminally tagged AphA or the active C-terminally tagged derivative. A quantification of the data is shown in Fig 5C; there was no significant difference between strains (P = 0.06). We conclude that repression of VC0857 by AphA has little impact on DNA uptake in the conditions of our assay. This is consistent with previous reports of VC0857 having only minor effects on competence [10].

Deleting luxO does not bypass inhibition of natural competence by AphA

To determine if repression of cqsS contributed to the remaining 6-fold drop in natural transformation, we deleted luxO. Deletion of luxO constitutively activates a high cell density expression profile regardless of CqsS activity or regulation. The 6-fold drop in natural transformation, caused by AphA in the presence of ectopic tfoX expression, was not altered by deleting luxO (compare grey and striped bars in Fig 5A).

Constitutive expression of AphA reduces levels of ComEA

Repression of VC0857 or cqsS cannot explain the lingering impact of AphA when tfoX is ectopically expressed. We resolved to determine which aspect of natural transformation was still impaired. As a starting point we measured DNA uptake. Hence, we incubated P_tac-tfoX cells with MFP488-labeled DNA and counted cells with DNA in their periplasm. Bacteria that constitutively expressed the non-functional N-terminally tagged AphA acquired DNA efficiently. Conversely, cells expressing functional C-terminally tagged AphA showed a significant 6-fold reduction (Fig 5D and 5E). Hence, DNA uptake was reduced by AphA even when TfoX was produced. Acquisition of DNA results from the concerted activity of type IV pili and periplasmically localized ComEA, the latter of which acts as a molecular ratchet to facilitate DNA uptake [14,39,40]. As we had already determined that pilus production was not affected by AphA in the P_tac-tfoX background (Fig 5C), we hypothesized that AphA may impact ComEA production. Using a functional fluorescent fusion, we found that ComEA expression was significantly reduced by expression of C-terminally tagged AphA (Fig 5D and 5F). Since AphA does not bind upstream of comEA (S4 Fig) this repression must be indirect.

Discussion

In V. cholerae, signals for natural competence are integrated at the tfoX promoter [4,5,14,32,41]. Triggers include: i) carbon starvation, mediated through CRP and ii) chitin-breakdown, sensed by membrane proteins ChiS and TfoS then communicated via the small RNA TfoR [5–8,32,42]. In comparison, regulatory links between cell density and competence are poorly characterised. At high cell density, HapR stimulates competence by blocking expression of the nuclease Dns [11,38]. Indirectly, HapR also activates expression of comEC and comEA via the LuxR-family transcription factor QstR [11,43,44]. The regulatory mechanisms are unknown and additional factors must be involved; QstR does not bind the comEA regulatory region [43]. At low cell density, we propose that AphA plays an essential role and directly represses competence (Table 1, Fig 6). Like other key factors, AphA targets tfoX expression (Fig 3). Surprisingly, the ability of AphA to repress competence is apparent even when dns is present (Fig 5). Indeed, for bacteria locked at low cell density (i.e. when HapR is poorly expressed), deletion of dns had a comparatively small effect observed only when aphA was absent (Fig 5). This indicates that aphA-dependent repression of tfoX, rather than loss of dns repression by HapR, plays the dominant role in limiting natural transformation at low cell density. Near maximal competence can be induced at low population density if aphA is deleted (Fig 5). Additionally, artificial expression of AphA blocks competence at high-cell density, when HapR is abundant. Taken together, this suggests that AphA is a key link between competence and population density.

Mechanistically, AphA blocks competence induction by preventing the activator CRP binding to the tfoX promoter (Fig 3). The ability of AphA to displace CRP, even with the latter in excess, is of key importance. This ensures that tfoX expression cannot be switched on by carbon starvation and chitin metabolism alone; the bacterial population must also reach an appropriate level. Our observations concerning AphA and CRP at the tfoX promoter are reminiscent of the tcpPH regulatory region. This locus also has overlapping sites for the two regulators that act antagonistically [45]. Opposing regulation was also observed at the cqsS promoter, although the regulators targeted distinct sites on the DNA (Fig 4). We speculate that antagonistic control of promoters by AphA and CRP is a common regulatory strategy in V. cholerae.

Unexpectedly, we found that AphA could interact with single DNA sites or co-operatively bind pairs of targets. In the examples tested here, binding a single site required the sequence to closely match the consensus for AphA binding (Fig 2). More divergent AphA sites could function in unison by co-operatively binding the regulator (S2 Fig). These different configurations of DNA binding may explain why three previous studies each proposed a slightly different AphA binding consensus [24,26,27]. Unusually, none of these sequences were palindromic. We suggest that AphA preferentially binds to the inverted repeat sequence 5'-ATGCAA-N₄-TTGCAT-3' (Fig 1D). Consistent with this, structures of PadR family regulators demonstrate DNA binding as a dimer with two-fold symmetry [34]. Confusion likely arose previously because the sequence 5'-TGCA-3' is embedded as a direct repeat within the larger motif identified here (Fig 1D). Furthermore, a paucity of known AphA binding sites hindered prior studies.

Surprisingly, transcriptome analysis found only six differentially regulated operons in V. cholerae cells lacking aphA. Similar studies in V. harveyi identified hundreds of genes [22,25,46]. We speculate that these discrepancies result from the growth conditions used. This may also explain why control of the competence regulon was not identified. In particular, transcription of tfoX and cqsS requires CRP, but transcriptome analysis used rich media that triggers catabolite repression [46]. Indirectly, this would impact VC0857 that is induced by TfoX. Our data are consistent with transcriptome profiling of cells treated with quorum sensing molecules. Hence, VC0857, tfoX and cqsS are all activated by auto-inducers that block expression of AphA (Table 1) [31]. Similarly, like Rutherford and co-workers, we note that AphA frequently targeted genes involved in cell envelope physiology [22]. In hindsight, this is not surprising since the TCP is itself membrane associated [47]. Importantly, our ChIP-seq analysis did detect binding of AphA at the tcpPH locus, albeit at low levels (S5 Fig). We were also able to detect AphA binding at its own promoter, consistent with previous reports of auto-regulation (Table 1) [22]. However, we did not detect AphA binding upstream of pva or alsR, the only other known targets [46,48].

In summary, our work better defines DNA binding by AphA and expands the direct regulon by >10-fold. Of particular interest are genes repressed by AphA that play key roles in the control of natural competence. Hence, as well as inducing the pathogenicity cascade at low population densities, AphA plays a key role by repressing genes utilised in the aquatic environment. We caution that our definition of the AphA regulon is unlikely to be complete. However, our work provides a solid basis for understanding changes in gene expression caused by transition of V. cholerae between the environmental niche and human host. In particular, we explain how competence can be controlled in this regard.

Materials and methods

Strains, plasmids and oligonucleotides

Standard procedures for cell culture and storage were used throughout. Strains were constructed using the approach of Dalia et al. [12]. Full descriptions of materials used are in S1 Table. Derivatives of pRW50T were transferred from E. coli DH5α into V. cholerae by tripartite mating. Overnight cultures were washed twice using 0.9% (w/v) NaCl, resuspended in LB, mixed in a 1:1:2 ratio of donor:recipient:helper, then spotted on non-selective LB plates. After overnight incubation at 30 ^oC, cells were resuspended in 0.9% NaCl and plated on TCBS agar containing 100 μg/ml streptomycin and 5 μg/ml tetracycline. After overnight incubation at 37 ^oC, colonies were re-streaked on LB agar containing 100 μg/ml streptomycin and 5 μg/ml tetracycline. Conjugants were confirmed by PCR.

Chromatin immunoprecipitation

ChIP-seq experiments were done as described in Haycocks et al. [49]. Briefly, V. cholerae E7946 was transformed with plasmid pAMCFaphA or pAMNFaphA. These encode AphA with a C- or N-terminal 3xFLAG epitope respectively. The N-terminally tagged AphA was unable to bind DNA in ChIP-seq experiments and so served as a useful control. Note that levels of AphA produced from plasmid pAMCFaphA and the native chromosomal locus were indistinguishable (S3 Fig). Cultures were incubated aerobically to mid-log phase in LB media at 37 ^oC. Cells were cross-linked with 1% (v/v) formaldehyde, washed, treated with lysozyme, and sonicated. The AphA-DNA complexes were immunoprecipitated with an anti-FLAG antibody (Sigma) and Protein A sepharose beads. Immunoprecipitated DNA was blunt-ended, A-tailed, and ligated to barcoded adaptors before elution and de-crosslinking. ChIP-seq libraries were then amplified by PCR and purified. Library quality was assessed using an Agilent Tapestation 4200 instrument and quantity determined by qPCR using an NEBnext library quantification kit (NEB).

Illumina sequencing and data analysis

Libraries were sequenced as previously described [50]. Each library was diluted to a concentration of 2 nM, before pooling and denaturation. Sequencing was done using an Illumina MiSeq instrument. Fastq files were deposited in Array Express (accession number E-MTAB-7953). Individual sequence reads were mapped against the Vibrio cholerae N16961 genome (Genbank accession numbers CP024162.1 and CP024163.1) using BWA (Burroughs-Wheeler Aligner) [51]. This facilitated comparison with other studies. Resulting Sequence Alignment Map (SAM) files were converted to Binary Alignment Map (BAM) files using the SAM-to-BAM tool [52,53]. Coverage per base was calculated using multiBAMsummary [54], and R was used to normalise each data set to the same overall read depth for each chromosome. To visualise the AphA binding profile coverage depth was plotted as a graph against genome features in Artemis. The graph window size was set to 100 bp and peaks with a coverage score of ≥ 10 over 300 consecutive bases were selected. The centre of the region passing the cut off was set as the peak centre. Next, 250 bp DNA sequences from each peak centre were collated. To identify DNA motifs associated with peak sequences we used MEME [55]. We scanned for motifs between 12 and 26 bp in length that occurred once per given sequence on the given DNA strand.

Natural transformation

Chitin-induced transformation assays of V. cholerae were done as described by using shrimp chitin flakes to induce competence (Sigma) [12]. Briefly, cells were grown to an OD₆₀₀ of ~1 in LB. Cells in 1 ml of culture were recovered by centrifugation and washed twice with 1 ml 0.7% (w/v) Instant Ocean (Aquarium Systems). Cells were diluted 10-fold with 0.7% Instant Ocean and 1 ml added to 10 mg of sterile chitin in a 2 ml Eppendorf tube. After incubation at 30 ^oC for 24 hours 200 ng of transforming DNA was added. Following a further 5 hours incubation cells were recovered in LB for 1–2 hours at 37 ^oC with shaking. Cells were then plated for quantitative culture on selective media (to quantify the number of transformants) and non-selective media (to quantify the total viable counts). The transformation frequency is defined as the number of transformants divided by the total viable count in each reaction.

Protein purification and western blotting

V. cholerae CRP was purified using cAMP-agarose as previously described [56]. RNA polymerase was purified from V. cholerae N16961 using a protocol based on the method of Burgess and Jendrisak as previously described [56,57]. V. cholerae σ⁷⁰ was purified by affinity chromatography as previously described [56]. To facilitate overexpression aphA was cloned in pET21a. The resulting plasmid was used to transform E. coli T7 Express cells. All colonies resulting from a single transformation were pooled and used to inoculate 500 ml of LB supplemented with 100 μg/ml ampicillin. Overexpression of C-terminally His₆ tagged AphA was induced with 1 mM IPTG for 3 hours. Cells were recovered by sonication and resuspended in 20 mM Tris-HCl pH 7.5, 1 mM EDTA pH 8.0, 10 mM NaCl and 1 mM PMSF. After cell lysis by sonication cell debris was removed by centrifugation and the lysate was passed through a His-Trap column (GE Healthcare). Proteins were eluted with a gradient of buffer containing 20 mM Tris-HCl pH 7.5, 1 mM EDTA pH 8.0, 10 mM NaCl and 500 mM imidazole. Fractions with AphA were identified by SDS-PAGE, pooled and concentrated. For long-term storage at -20 ^oC, glycerol was added to a final concentration of 50% (v/v). Western blots were done as described by Lamberte et al. [58].

Electrophoretic mobility shift assays (EMSAs)

DNA fragments for EMSA experiments were generated by PCR as previously described [59]. PCR products were cut using EcoRI and HindIII (NEB). End-labelling was done using γ³²-ATP and T4 polynucleotide kinase (NEB). Radiolabelled fragments were incubated with purified proteins in buffer containing 40 mM Tris acetate pH 7.9, 50 mM KCl, 5 mM MgCl₂, 500 μM DTT and 12.5 μg/ml Herring Sperm DNA for 15 minutes at 37 ^oC. Protein-DNA complexes were separated by electrophoresis using a 7.5% non-denaturing polyacrylamide gel. Subsequently, dried gels were exposed to a Biorad phophorscreen that was scanned using a Biorad Personal Molecular Imager. Full gel images are shown in S6 Fig.

DNAse I and KMnO₄ footprinting

DNA fragments were excised from pSR using AatII-HindIII. After end-labelling using γ³²-ATP and T4 PNK (NEB), footprints were done as previously described in buffer containing 40 mM Tris acetate pH 7.9, 50 mM KCl, 5 mM MgCl₂, 500 μM DTT and 12.5 μg/ml Herring Sperm DNA [60,61]. Resulting DNA fragments were analysed on a 6% denaturing gel. Subsequently, dried gels were exposed to a Biorad phophorscreen that was scanned using a Biorad Personal Molecular Imager. Full gel images are shown in S6 Fig.

In vitro transcription assays

We used the protocol of Kolb et al. [62] as described by Savery et al. [63]. Reactions contained different combinations of 16 μg/ ml supercoiled pSR template, V. cholerae RNA polymerase σ⁷⁰ holoenzyme, AphA and CRP. In experiments where CRP was used, CRP was pre-incubated with cAMP 37 ^oC prior to addition. The reaction buffer was 40 mM Tris pH 7.9, 5 mM MgCl₂, 500 μM DTT, 50 mM KCl, 100 μg/ ml BSA, 200 μM ATP/GTP/CTP, 10 μM UTP and 5 μCi α-P³²-UTP. If required, AphA and CRP were added to reactions for 10 minutes at 37 ^oC before the addition of 0.4 μM RNA polymerase, for a further 10 minutes. Transcripts were analysed on a 6% denaturing polyacrylamide gel. The dried gel was exposed to a Biorad Phosphorscreen, which was scanned using a Biorad Personal Molecular Imager. Full gel images are shown in S6 Fig.

β-galactosidase assays

V. cholerae harbouring pRW50T were grown to mid-log phase (OD₆₅₀ of ~1) in LB or M9 minimal media supplemented with 1% fructose unless stated in figure legends. Cells were lysed using 1% sodium deoxycholate and toluene, and assays carried out as previously described using the Miller method [56,64]. For cells containing plasmid pMMB-tfoX 1 mM IPTG was present.

Microscopy

Strain P_tac-tfoX ΔluxO containing pAMCFaphA or pAMNFaphA was grown in continually rolling containers at 30 ºC to late-log phase in LB Miller broth supplemented with 50 μg/mL Kanamycin, 100 μM IPTG, 20 mM MgCl₂ and 10 mM CaCl₂. Around 10⁸ colony-forming units (c.f.u.) were collected by centrifugation (18,000 x g for 1 minute) and resuspended in 0.7% Instant Ocean. Cells were then labelled with 25 μg/ml AlexaFluor488 maleimide (AF488-mal) in the dark for 15 minutes at room temperature. Cells were then washed twice with 0.7% Instant Ocean by sequential centrifugation and resuspension. Cells were imaged by time-lapse microscopy every 2 seconds for 2 minutes to monitor pilus production. The number of cells that made at least one pilus within the 2-minute window and the total number of cells were manually counted. To examine DNA internalisation, approximately 10⁸ c.f.u. of late-log culture were diluted 4-fold with Instant Ocean. The cells were then incubated with or without 100 ng MFP488-labelled DNA at room temperature in the dark. After 30 minutes, 10 units of DNase I (NEB) was added to all reactions and incubated for 2 minutes to degrade any remaining extracellular DNA. Cells were then washed twice with 0.7% Instant Ocean by sequential centrifugation and resuspension. Static images of cells were taken and the number of cells with a DNA-uptake event, indicated by MFP-488 DNA foci, compared to the total number of cells in a field of view, was manually counted. For all microscopy experiments, samples were placed under 0.2% Gelzan (Sigma) pads made with Instant Ocean medium. We used a Nikon Ti-2 microscope with a Plan Apo x60 objective, GFP and dsRed filter cubes and a Hamamatsu ORCAFlash4.0 camera. Image collection and analysis used Nikon NIS Elements imaging software and Image J.

Supporting information

S1 Fig [pdf]
Binding of AphA to different DNA fragments .

S2 Fig [pdf]
Binding of AphA to the and regulatory regions.

S3 Fig [ns]
Levels of chromosomal and plasmid encoded AphA are indistinguishable.

S4 Fig [pdf]
AphA does not bind to the regulatory region .

S5 Fig [pdf]
AphA binds to the tcpPH regulatory region in vivo.

S6 Fig [pdf]
Raw gel images.

S1 Movie [avi]
Pilus dynamics in the presence of AphA with an N-terminal 3xFLAG fusion.

S2 Movie [avi]
Pilus dynamics in the presence of AphA with a C-terminal 3xFLAG fusion.

S1 Table [docx]
Strains, plasmids and oligonucleotides.