Competence for genetic transformation in Vibrio cholerae is triggered by chitin-induced transcription factor TfoX and quorum sensing (QS) regulator HapR. Transformation requires expression of ComEA, described as a DNA receptor in other competent bacteria. A screen for mutants that poorly expressed a comEA–luciferase fusion identified cytR, encoding the nucleoside scavenging cytidine repressor, previously shown in V. cholerae to be a biofilm repressor and positively regulated by TfoX, but not linked to transformation. A ΔcytR mutant was non-transformable and defective in expression of comEA and additional TfoX-induced genes, including pilA (transformation pseudopilus) and chiA-1 (chitinase). In Escherichia coli, ‘anti-activation’ of nucleoside metabolism genes, via protein–protein interactions between critical residues in CytR and CRP (cAMP receptor protein), is disrupted by exogenous cytidine. Amino acid substitutions of the corresponding V. cholerae CytR residues impaired expression of comEA,pilA and chiA-1, and halted DNA uptake; while exogenous cytidine hampered comEA expression levels and prevented transformation. Our results support a speculative model that when V. cholerae reaches high density on chitin, CytR–CRP interactions ‘anti-activate’ multiple genes, including a possible factor that negatively controls DNA uptake. Thus, a nucleoside scavenging mechanism couples nutrient stress and cell–cell signalling to natural transformation in V. cholerae as described in other bacterial pathogens.
Vibrio cholerae is the bacterium responsible for the fatal diarrhoeal disease cholera, but it is also a natural inhabitant of marine and estuarine environments where it commonly forms biofilms on abiotic and biotic surfaces, such as chitinous chironomids (non-biting flies) and zooplankton moults (Tamplin et al., 1990; Halpern et al., 2004). Horizontal gene transfer (HGT) of traits among Vibrios is thought to promote rapid genetic exchange that is responsible for the mosaic genome structure of Vibrios revealed by recent genomic sequencing efforts (Chun et al., 2009). Competence for genetic transformation in V. cholerae was recently reported and represents a newly appreciated mode of HGT for this aquatic bacterium (Meibom et al., 2005). V. cholerae natural transformation is induced by two environmental signalling pathways: a quorum sensing system and a chitin utilization system. The network connecting these two systems to natural competence remains poorly understood.
Quorum sensing (QS) enables bacterial populations to collectively control gene expression and thus co-ordinate behaviours presumably unproductive for individuals (Ng and Bassler, 2009). Like many other Vibrio species, V. cholerae populations accomplish QS by producing and responding to autoinducer (AI) signal molecules, specifically two AIs, CAI-I and AI-2 (for review see Hammer and Bassler, 2008). At low cell density (LCD) the unbound receptors of CAI-1 and AI-2 (CqsS and LuxP/Q respectively) behave as kinases and phosphorylate response regulator LuxO, via LuxU. Phosphorylated LuxO activates transcription of four small RNAs, termed Qrr1–4 (Quorum Regulatory RNAs). In association with the RNA chaperone Hfq, the Qrrs base-pair with the mRNAs of several target genes including hapR, which encodes the master regulator of QS, HapR (Lenz et al., 2004; Hammer and Bassler, 2007; Svenningsen et al., 2009; Bardill et al., 2011; Rutherford et al., 2011; Bardill and Hammer, 2012). At high cell density (HCD), binding of AIs to their cognate receptors switches them to phosphatases, reversing the phosphorylation cascade and inactivating LuxO. Thus, HapR is produced and activates expression of numerous genes at HCD including hapA, which encodes a protease that plays a role in interactions of V. cholerae with aquatic hosts (Halpern et al., 2003), and the comEA gene, which encodes a periplasmic DNA-binding protein shown to be the DNA receptor for transformation in Bacillus subtilis (Provvedi and Dubnau, 1999; Meibom et al., 2005) (Fig. 1). Therefore, wild-type (WT) V. cholerae is naturally competent at HCD, a ΔhapR mutant is non-transformable, and a ΔluxO mutant that constitutively produces HapR takes up DNA independently of cell density (Meibom et al., 2005; Blokesch and Schoolnik, 2008). So too, a V. cholerae strain unable to produce either AI only expresses comEA and takes up DNA when provided exogenous AIs (Antonova and Hammer, 2011; Suckow et al., 2011).
Activation of comEA expression by QS AIs requires an additional extracellular signal, namely chitin, the most abundant carbon source in the ocean. Genetic studies (Li and Roseman, 2004; Meibom et al., 2004; Yamamoto et al., 2010; 2011) support that when chitin is present, (GlcNAc)2 binds to the CBP (chitin-binding protein) activating the ChiS sensor kinase, which in turn leads to the production of the Hfq-dependent TfoR sRNA. TfoR promotes translation of the mRNA encoding an orthologue of Haemophilus influenzae regulator, Sxy (called TfoX in Vibrio species), which appears to promote transcription of competence genes in other bacteria by direct interactions with the cAMP receptor protein, CRP (Meibom et al., 2004; Redfield et al., 2005) (Fig. 1). In V. cholerae, experimental induction of tfoX, such as from the tac promoter, is sufficient to promote transformation and activate expression of the comEA and multiple chitinase genes (including chiA-1) independent of chitin (Meibom et al., 2005; Yamamoto et al., 2010; Antonova and Hammer, 2011). In addition to comEA and chiA-1, competence pseudopilus genes (pilA, pilB, pilQ) are also under TfoX control (Meibom et al., 2005) (Fig. 1). Thus, for V. cholerae both the chitin responsive pathway (i.e. TfoX) and QS (i.e. HapR) are required for sufficient comEA expression to promote uptake of DNA, as Δtfox and ΔcomEA mutants, like a ΔhapR mutant, are severely impaired for transformation (Meibom et al., 2005).
DNA uptake by V. cholerae was demonstrated in 2005, and more recently in other marine Vibrios, such as V. parahemolyticus, V. fischeri and V. vulnificus (Srivastava et al., 2009; Chen and Ge, 2010; Pollack-Berti et al., 2010). Recent genome sequencing efforts, however, predict that not only the Vibrionaceae, but also the Enterobacteriaceae, encode orthologues of many DNA uptake genes, including tfoX and comEA (Cameron and Redfield, 2006). Despite the fact that many members of the Enterobacteriaceae are not naturally competent, it has been proposed that a common regulatory mechanism for natural competence involving both TfoX and CRP may be shared among these two families of the γ-proteobacteria (Redfield et al., 2005; Cameron and Redfield, 2006). Studies that bolster this hypothesis include demonstration that expression of natural transformation genes in Escherichia coli and V. cholerae is subjected to carbon catabolite repression (CCR) (Meibom et al., 2005; Cameron and Redfield, 2006; Blokesch, 2012; Zheng et al., 2012). CRP, the global regulator of CCR, in Gram-negative bacteria, together with its allosteric effector cAMP, controls the expression of multiple genes involved in the utilization pathways of alternative carbon sources when glucose levels in the cell are low (Bruckner and Titgemeyer, 2002; Deutscher, 2008). In Gram-positive bacteria, CCR also co-ordinates a response to low glucose levels but instead utilizes the CcpA transcription factor (Warner and Lolkema, 2003). Interestingly, a recent study demonstrating that CcpA in Streptococcus may induce natural competence as a consequence of CCR supports perhaps an even broader role for nutrient starvation in inducing DNA uptake (Zheng et al., 2012).
The complex CRP-mediated response to nutrient stress in Gram-negative bacteria may also include participation by additional regulatory factors that enable the bacteria to utilize not only varied carbohydrates but also nucleic acids as well. In E. coli, for example, many proteins for scavenging of extracellular free nucleosides are encoded by genes in a regulon that is negatively controlled by the CytR regulator (Valentin-Hansen et al., 1996). The presence of the nucleoside cytidine in the growth medium alleviates repression by CytR, which works in conjunction with CRP in E. coli as an ‘anti-activator’ of a subset of CRP-activated genes that are involved in nucleoside transport and metabolism. In V. cholerae, CytR has been shown to repress expression of the udp gene for nucleoside catabolism in nucleoside-poor environments, in addition to impairing biofilm development by unknown mechanisms (Haugo and Watnick, 2002). Moreover, several studies in V. cholerae demonstrated that CRP is involved in QS, biofilm formation, motility and cholera toxin production, as well as natural competence (Callahan and Richardson, 1973; Liang et al., 2007a,b; Fong and Yildiz, 2008; Blokesch, 2012). While CRP may well participate in many steps of a regulatory network for DNA uptake, a molecular mechanism linking CRP to the control of natural competence genes has not been demonstrated in V. cholerae. Here we provide evidence that the CytR anti-activator, which requires CRP to function, co-ordinates a nutrient scavenging response in V. cholerae that controls natural competence.
Identification of a competence-deficient V. cholerae mutant
Transcription of comEA depends on induction of TfoX in response to chitin, and on HapR, which is produced as a result of accumulated quorum sensing (QS) autoinducers (AIs) at HCD. V. cholerae strains that do not produce TfoX, or are unable to synthesize QS AIs, and thus produce no HapR, are severely impaired for comEA expression (Meibom et al., 2004; 2005; Antonova and Hammer, 2011; Suckow et al., 2011). To identify one or more positive regulators of the competence gene, comEA, in V. cholerae, we performed a transposon mutagenesis of El Tor V. cholerae strain C6706 using a well-described Tn5 system (Larsen et al., 2002). We eliminated the requirement of chitin by first constructing a derivative of C6706 that expresses the tfoX gene from the chromosome under control of a constitutive tac promoter (referred to here as a tfoX* allele). C6706 has a frameshift mutation in lacI and does not encode a functional LacI protein, thus tfoX is not LacI-repressed in this strain. Maximal expression of a comEA–lux reporter that we constructed previously (Antonova and Hammer, 2011) was observed in the tfoX* strain at HCD, when the QS response results in HapR production (Fig. 2A, bar 1). A deletion of hapR in this background results in a reduction in comEA expression by ∼ 10 000-fold, as expected (Fig. 2A, bar 2). In contrast, a ΔluxO, tfoX* strain expresses comEA to levels similar to the tfoX* strain at HCD, but does so independent of chitin and AI accumulation (data not shown). We further introduced a second copy of hapR at the lacZ locus into the ΔluxO, tfoX* strain to minimize screening of mutants with Tn5 insertion in hapR, or spontaneous hapR mutations that can occur in a ΔluxO strain (Hammer and Bassler, 2003). As expected, the merodiploid behaved in a manner indistinguishable from the isogenic ΔluxO, tfoX* strain, expressed comEA maximally (Fig. 2A, compare bar 1 and 3), and was used as the recipient for Tn5 transposon mutagenesis.
Three independent pools of ∼ 20 000 Tn5 kanamycinR (KanR) mutants were screened for defective comEA–lux expression, and one isolate was identified that expressed comEA as poorly as a ΔhapR mutant (Fig. 2A, bar 4). The location of the Tn5 insertion mapped to vc2677, which is annotated in the database as a transcriptional repressor of the LacI family. In 2002, vc2677 was shown to encode CytR, which represses biofilm formation in V. cholerae strain MO10 (Haugo and Watnick, 2002). Importantly, it was later identified by Meibom et al. in a microarray study as one of ∼ 100 TfoX-induced genes (Meibom et al., 2005). Because it was predicted to participate in nucleoside metabolism and DNA uptake based on genomic comparison with other naturally competent bacteria (Cameron and Redfield, 2006), we sought to determine a role for CytR in natural competence of V. cholerae.
CytR positively regulates comEA expression and DNA uptake in V. cholerae
To confirm that the defect in comEA expression in the transposon-insertion mutant was indeed the result of cytR inactivation, we introduced an in-frame cytR gene deletion (ΔcytR) into various genetic backgrounds of V. cholerae using standard methods (Skorupski and Taylor, 1996), and measured expression of comEA–lux (Fig. 2A) and transformation frequency (Fig. 2B) of each strain. A ΔcytR, tfoX* mutant was 10 000-fold reduced for comEA expression compared with the WT strain carrying tfoX* (Fig. 2A, compare bar 5 with bar 1), and similar to the comEA defect of the ΔhapR, tfoX* mutant (Fig. 2A, compare bar 5 with bar 2). Similarly, a ΔluxO, tfoX*, ΔcytR mutant was also deficient in comEA expression compared with the isogenic ΔluxO, tfoX* mutant and to the ΔluxO, lacZ::hapR, tfoX* mutant (data not shown, and Fig. 2A, compare bar 6 with bar 3). The ΔcytR, tfoX* strain was fully complemented either by inserting a copy of the cytR gene under control of its native promoter into the chromosome at the lacZ site (Fig. 2A, compare bar 7 with bar 1), or by introducing the same cytR gene on a plasmid (data not shown), as comEA–lux expression was restored to levels observed with the tfoX* strain.
Transformation frequencies of corresponding V. cholerae strains that do not carry a tfoX* allele, but rather require chitin for tfoX expression, were measured by a crab-shell microcosm system described previously (Meibom et al., 2005). The results were consistent with comEA–lux expression (Fig. 2A and B). Specifically, no transformants were detected for the Tn5 mutant, ΔcytR mutant, and the ΔluxO, ΔcytR double mutant, similar to the ΔhapR strain. The DNA uptake level of the complemented strain was similar to WT, the ΔluxO mutant, and the ΔluxO, lacZ::hapR mutant of V. cholerae (Fig. 2B; Antonova and Hammer, 2011). These results confirm the role of CytR as a positive regulator of natural competence of V. cholerae.
CytR and QS regulate expression of multiple TfoX-induced genes
Based on prior microarray studies, in addition to the cytR gene itself, three groups of genes have been shown to be under TfoX control in V. cholerae: DNA binding and processing genes (such as comEA), chitin degradation and utilization genes (such as chiA-1), and competence pseudopilin genes (such as pilA) (Meibom et al., 2005). To test whether genes in each of these categories are also regulated by CytR and/or QS, we constructed luciferase-based transcriptional fusions of representative genes in each group and quantified expression in Luria broth (LB medium) without chitin for V. cholerae WT (QS+, CytR+), ΔhapR (QS−, CytR+) and ΔcytR (QS+, CytR−) strains that carry the constitutive tfoX* allele (TfoX+) or the native tfoX allele (TfoX−) that requires chitin for induction. Consistent with microarray studies, expression of all three genes, comEA, chiA-1 and pilA, increased in WT V. cholerae in the presence of the tfoX* allele (∼ 115-, 55-, 15-fold respectively), which confirms that they are activated by TfoX (Fig. 3A–C, compare first two bars). However, expression of all three genes in a ΔhapR and a ΔcytR mutant was approximately as low as in the strains lacking constitutive TfoX, which suggested that not only comEA, but also chiA-1, and pilA are positively regulated by both HapR and CytR (Fig. 3A–C).
As a control, we also constructed a lux-based transcriptional fusion to the V. cholerae udp gene, encoding uridine phosphorylase, which is repressed by CytR (Haugo and Watnick, 2002). As expected, the transcription pattern of udp–lux was reciprocal of the comEA–lux results (Fig. 3D). Namely, expression decreased in the WT strain when tfoX was expressed (Fig. 3D, compare the first two bars). In a ΔhapR mutant, a similar pattern was observed with slightly greater udp expression than the WT strain in the absence of TfoX. Levels of udp in the ΔhapR strain without tfoX* were slightly higher than in WT (Fig. 3D, compare the second white bar with the first white bar), as described prior (Yildiz et al., 2004; Liang et al., 2007a). These results suggest that HapR not only positively regulates comEA, pilA and chiA-1, but also negatively regulates udp in a manner independent of TfoX induction (and CytR) (Fig. 1). Moreover, in a ΔcytR mutant, repression of udp–lux was eliminated (Fig. 3D), consistent with studies that showed repression of the udp gene by CytR in V. cholerae (Haugo and Watnick, 2002; Zolotukhina et al., 2003).
As an additional control, we measured the expression pattern of the hapA gene, which encodes a haemagglutinin protease that is positively regulated by HapR (Hammer and Bassler, 2003; Bardill et al., 2011), but not predicted to be under control of either TfoX or CytR. The HapR-dependent expression pattern of hapA–lux was as described prior (Hammer and Bassler, 2003). The maximal transcription of hapA was observed in both WT and ΔcytR mutants, and minimal in a ΔhapR mutant and, as predicted, independent of TfoX and CytR (Fig. 3E). These results confirm that regulation of hapA is QS-dependent and CytR-independent. Taken together, these data indicate that CytR positively regulates genes controlling the natural competence and chitin utilization network in V. cholerae, and as it was shown previously, negatively regulates the udp gene involved in nucleoside metabolism.
V. cholerae CytR behaves like a CRP-dependent anti-activator
Extensive studies in E. coli have demonstrated that CytR associates with global cAMP receptor protein, CRP, to inhibit transcription of a subset of promoters activated by CRP (for review see Valentin-Hansen et al., 1996). For example, the CRP-dependent promoter of udp in E. coli contains both a distal (CRP-2) and a proximal (CRP-1) binding site allowing activation of udp transcription by CRP in the absence of glucose when intracellular cAMP levels are high (Brikun et al., 1996; Zolotukhina et al., 2003). Specifically, binding of a CRP dimer at the twofold symmetric CRP-2 site and at the CRP-1 site, which overlaps the −35 region, positions each CRP dimer to recruit RNA polymerase (RNAP) (Fig. 4A, top). The CRP consensus site in E. coli is depicted in Fig. 4B. Optimal spacing of 52–53 bp between the CRP-2 and CRP-1 sites of several CRP-dependent promoters, including udp, ensures that a CytR dimer inhibits initiation of transcription of such genes by protein–protein interactions with each CRP dimer. Interaction of CytR with a highly degenerate operator site between the two CRP sites plays a minor role (Busby and Ebright, 1999), and indeed overexpression of E. coli CytR that lacks a DNA-binding domain can still repress the deoP promoter, which is also under CytR–CRP control (Sogaard-Andersen and Valentin-Hansen, 1993; Pedersen and Valentin-Hansen, 1997). Thus, CytR–CRP interactions are necessary for ‘anti-activation’ of genes in E. coli, including udp, which are involved in nucleoside scavenging in the absence of preferred carbon sources (Fig. 4A, bottom) (Valentin-Hansen et al., 1996). As a result, we sought to test whether a CRP-dependent CytR anti-activation mechanism may control natural competence in V. cholerae, and if such a mechanism was consistent with prior observations that a Δcrp mutant is defective in transformation and that the presence of glucose inhibits chitin-induced DNA uptake (Meibom et al., 2005; Blokesch, 2012).
To determine whether the CytR–CRP protein–protein interactions important for the E. coli nucleoside scavenging response play a role in V. cholerae natural competence, we first examined the degree of similarity between the regulatory proteins involved. It has been shown previously that V. cholerae cytR complements an E. coli cytR mutant, confirming that it is a functional homologue (Haugo and Watnick, 2002). Protein alignments indicated that V. cholerae CytR and CRP are 81% and 98% similar (65% and 95% identical), respectively, to their E. coli homologues (Fig. 4C). Importantly, alignment revealed that specific residues of E. coli CytR, notably residues L169 and F173 (corresponding to L161 and F165 in V. cholerae) that form a patch on the surface of CytR crucial for the CytR–CRP interactions in E. coli (Kallipolitis et al., 1997), are conserved in V. cholerae CytR (Fig. 4C). As expected, residues that are components of a corresponding patch on the surface of CRP are also conserved between these nearly identical proteins (not shown). Thus, to determine whether a CytR–CRP anti-activation mechanism is conserved and responsible for natural competence in V. cholerae, we measured expression of our lux-based gene fusions in a WT V. cholerae control strain expressing tfoX* and compared each to expression levels in an isogenic strain with an in-frame cytR gene deletion (ΔcytR), or with an L161A amino acid substitution in CytR (cytRL161A), which corresponds to the E. coli L169A CytR mutation that abolishes CytR-dependent anti-activation (Kallipolitis et al., 1997). Based on the results of the genetic screen with comEA (Fig. 2A), the expression pattern of QS- and CytR-dependent reporters, comEA, chiA-1 and pilA in the control strains was as expected. Namely, the maximal transcription of each gene was observed in the WT strain carrying the tfoX* allele, with minimal expression of each gene fusion in an isogenic ΔcytR, tfoX* strain (Figs 3 and 4D–F; compare bar 1 with bar 2). We reasoned that the V. cholerae tfoX* strain expressing cytRL161A would behave like a ΔcytR mutant, because a L161A amino acid substitution in CytR prevents CytR–CRP protein–protein interaction. As predicted, a tfoX*, cytRL161A mutant was severely impaired for expression of comEA and chiA-1, like the isogenic ΔcytR strain (Fig. 4D and E). Expression of pilA was slightly higher (threefold) in the tfoX*, cytRL161A double mutant compared with the ΔcytR strain, but was still over 10-fold lower than the tfoX* mutant that expresses pilA maximally (Fig. 4F; compare bar 2 with bars 3 and 1). We also measured expression levels of comEA, chiA-1 and pilA in a cytRF165A, tfoX* strain and observed similar alterations in expression (data not shown), consistent with E. coli studies (Kallipolitis et al., 1997).
Previous studies predicted and also documented a role for carbon catabolite repression (CCR) in regulating transformation by V. cholerae, and a requirement of CRP for this process (Meibom et al., 2005; Cameron and Redfield, 2006; Blokesch, 2012). We identified CytR as a critical regulator for transformation and expression of multiple competence genes in V. cholerae (Figs 2 and 3). The impact of the loss of CytR depends on CRP in E. coli, thus, to test whether CRP is also required for the observed effects of CytR in V. cholerae, we measured comEA, chiA-1 and pilA expression in a V. cholerae tfoX*, Δcrp mutant that constitutively expressed tfoX, but carried an in-frame deletion in crp. It is important to note that the V. cholerae strain used does not encode a functional LacI, and the tfox* allele is controlled here by the tac promoter, which is also insensitive to catabolite repression. The Δcrp, tfoX* double mutant showed a level of expression intermediate between the WT (tfoX*) and ΔcytR, tfoX* strains for all three gene fusions tested (Fig. 4D–F, bar 4).
To confirm that CytR function in V. cholerae requires CRP, we designed an epistasis experiment test to provide additional evidence that competence gene expression and DNA uptake require CytR–CRP interaction. Specifically, we predicted that the Δcrp mutation, which resulted in intermediate levels of competence gene expression would be epistatic to the ΔcytR mutation which produced minimal levels of expression for comEA, pilA and chiA-1. We constructed a ΔcytR, Δcrp, tfoX* mutant and measured expression of each of these gene fusions. Indeed, the ΔcytR, Δcrp, tfoX* mutant displayed intermediate levels of expression for each of the three gene fusions; like the Δcrp, tfoX* mutant (Fig. 4D–F, compare bar 5 with bar 4), and unlike the minimum values observed in the ΔcytR, tfoX* mutant (Fig. 4D–F, compare bar 5 with bar 3).
While the genetic analysis supported a potential model that three TfoX-activated genes previously described (Meibom et al., 2005) are positively controlled by CytR–CRP regulation, this regulation is likely indirect since CytR–CRP anti-activates, or effectively represses, promoters under their direct control (see Discussion). In contrast, the udp gene in V. cholerae is directly controlled by CytR, like in E. coli (Haugo and Watnick, 2002; Zolotukhina et al., 2003). Indeed, upstream of the V. cholerae udp gene is a well-conserved distal CRP-2 site centred at −154 separated by 52 bp from a proximal CRP-1 binding site centred at −102, relative to the ATG start codon (Fig. 4B) (Zolotukhina et al., 2003). Thus, we predicted that the expression pattern of a udp–lux gene fusion would be the reciprocal to that observed for the competence genes in V. cholerae; namely, udp–lux expression would be higher in a cytRL161A mutant, as it is in a ΔcytR mutant. As already shown in Fig. 3, udp expression was lowest in the tfoX* strain, and maximal (although only approximately fivefold higher) in a tfoX*, ΔcytR strain. As predicted, the tfoX*, cytRL161A mutant had a udp expression level similar to the ΔcytR mutant and still higher than the tfoX* strain (Fig. 4G). Surprisingly, the Δcrp mutants did not appear to have an intermediate level of expression, perhaps due to the difficulty of resolving difference of less than threefold with lux-based gene fusions as described prior (Hammer and Bassler, 2009). While the fold differences in udp–lux expression observed here in V. cholerae were not identical to observations by Zolotukhina et al. with a lacZ transcriptional fusion of V. cholerae udp measured in E. coli (Zolotukhina et al., 2003), the pattern in each case is consistent with negative regulation of V. cholerae udp by CytR.
Productive CytR–CRP interactions yield maximal expression of the three competence and chitin utilization genes tested (Fig. 4D–F), thus we predicted that DNA uptake would likely follow a similar pattern in transformation assays with corresponding V. cholerae strains that expressed tfoX in the presence of chitin, rather than in response to a constitutive tfoX* allele. Indeed, transformation frequencies were consistent to the expression of QS- and CytR-dependent reporters (Fig. 4H). Compared with the WT strain that has a transformation frequency of ∼ 10−5, a ΔcytR and cytRL169A mutants were non-transformable; as were the Δcrp and ΔcytR, Δcrp mutants. These observations are consistent with previous studies (Meibom et al., 2005; Blokesch, 2012), and confirm the essential role of CytR–CRP interactions in natural competence of V. cholerae. Specifically, they suggest that productive protein–protein interaction between these two regulators ensures the expression of genes for DNA uptake in nutrient poor settings where glucose is absent, but the alterative carbon source, chitin, is abundant.
CytR overexpression is not sufficient for maximal comEA expression
Previously, Meibom et al. demonstrated that not only were comEA, pilA and chiA-1 positively regulated by TfoX, but also cytR as well (Meibom et al., 2005). Because we had identified cytR in a screen with a strain that expressed TfoX and HapR constitutively (Fig. 2), we sought to determine whether CytR may fit in a regulatory pathway downstream of TfoX, or possibly HapR. To test this we first constructed a cytR–lux fusion; however, we observed no change in expression when we compared levels in a WT strain to strains carrying a tfoX* allele, or ΔtfoX and ΔhapR strains (data not shown). As an additional test to determine whether CytR may be under control of TfoX, we constructed a plasmid (p-tac-cytR) to control transcription of the cytR gene by the IPTG-inducible, tac promoter. We reasoned that if the only role of TfoX, when induced by chitin (mimicked by the tfoX* allele), was to activate cytR transcription, then a strain carrying p-tac-cytR (+ IPTG) would express comEA to maximal levels sufficient for DNA uptake, even in a ΔtfoX strain. Alternatively, if CytR acts on competence gene expression in a manner that is independent of TfoX, then tfoX* would still be required for maximal comEA expression in strains overexpressing cytR from the p-tac-cytR plasmid. As Fig. 5 shows, in the absence of chitin, WT V. cholerae (with the native tfoX allele) carrying the p-tac-cytR plasmid expresses comEA minimally in the absence of IPTG, and reaches an intermediate level of comEA upon IPTG induction of cytR (Fig. 5, first set of bars). In a ΔcytR mutant, and a ΔtfoX mutant, a similar modest increase in comEA expression (10-fold) was also observed upon cytR expression by IPTG from p-tac-cytR (Fig. 5, second and third set of bars). However, the 100-fold increase in comEA expression to the maximal level of > 107 was only achieved in strains carrying tfoX* strain (Fig. 5, fourth and fifth sets of bars). Consistent with the epistasis test shown in Fig. 4, the ΔcytR, Δcrp and ΔcytR, Δcrp, tfoX* strains express intermediate levels of comEA that are not altered by CytR overexpression (Fig. 5, sixth and seven sets of bars). These results suggest that cytR has a positive impact on competence gene expression in the absence of TfoX; however, an independent contribution by TfoX is required to achieve the levels of competence gene expression necessary for DNA uptake.
Cytidine is a repressor of natural competence
In E. coli, protein–protein interactions between CytR and CRP result in anti-activation of numerous CRP-activated metabolism genes, including udp and cytR itself (Valentin-Hansen et al., 1996). CytR is termed the Cytidine Repressor because the accumulation of cytidine is thought to induce conformational changes in CytR that weaken contact with CRP. Prior studies in E. coli have shown that cytidine relieves CytR-dependent anti-activation (repression) and permits CRP-dependent activation of nucleoside scavenging genes including udp in minimal media lacking glucose (Barbier and Short, 1992). Thus, we next tested the role of cytidine scavenging in CytR anti-activation of genes controlling natural transformation in V. cholerae. We measured levels of comEA–lux expression in V. cholerae strains incubated in AB minimal media that was supplemented with chitin to activate tfoX expression and to provide the bacteria with a carbon source. Under these conditions where induction of tfoX by chitin promotes competence, the presence of 100 mM cytidine reduced transcription of comEA–lux in WT V. cholerae > 100-fold, to levels comparable to a ΔcytR mutant, which was unresponsive to cytidine addition (Fig. 6A).
Escherichia coli CytR with amino acid substitution D281N (corresponding to D273N in V. cholerae) (Fig. 4C) binds cytidine with three orders of magnitude lower affinity than native CytR and this mutation severely curtails cytidine-mediated disruption of CytR–CRP interactions (Barbier and Short, 1992; Barbier et al., 1997). We constructed a V. cholerae strain that carries the corresponding cytRD273N allele and compared comEA expression in this mutant to the isogenic WT control strain carrying the native cytR allele in the presence or absence of 100 mM cytidine. The expression levels of comEA–lux in the cytRD273N mutant were similar to WT when incubated in AB minimal medium with chitin but lacking exogenous cytidine. However, unlike the WT strain that experienced a > 100-fold reduction in comEA in response to cytidine, the cytRD273N mutant showed only a slight (12-fold) reduction in comEA–lux expression (Fig. 6A), indicating that CytR mediates the cytidine response observed in this assay. Expression of the comEA reporter in a Δcrp mutant, which was already ∼ 10-fold decreased relative to WT strain incubated without cytidine (as in Fig. 4D), remained unresponsive to cytidine (Fig. 6A), because the absence of CRP is epistatic to both CytR- and cytidine-mediated effects.
To uncouple the role of chitin as a nutrient source and inducer of tfoX in our assay, we performed a complementary experiment with strains that expressed the constitutive tfoX* allele. The expression patterns in the presence of cytidine were similar, even under conditions where tfoX was no longer under control of its native promoter (Fig. 6B). The effect of cytidine on comEA expression in the WT strain carrying tfoX* was enhanced, resulting in an ∼ 800-fold reduction in the presence of cytidine. The remaining tfoX* strains displayed a similar pattern as in Fig. 5A, with no effect of cytidine on the ΔcytR and Δcrp mutant, and a minor (10-fold) reduction in the cytRD273N mutant.
Finally, to test whether the presence of exogenous cytidine not only halted comEA expression in a CytR-dependent manner, but also prevented DNA uptake, we measured transformation efficiency of corresponding strains carrying the native tfoX allele but incubated on chitin. Unlike the WT strain that is highly transformable, the WT strain incubated on chitin with 100 mM cytidine showed an ∼ 10 000-fold defect in transformation, with no DNA uptake recorded, similar to the ΔcytR mutant (Fig. 6C), which was not transformable without and with 100 mM cytidine. In contrast, transformation frequency of the cytRD273N mutant was similar to WT grown without cytidine, and only slightly decreased (10-fold) in the presence of cytidine (Fig. 6C). The Δcrp mutant that expressed intermediate comEA levels that were unresponsive to cytidine addition, was not transformable as expected (Fig. 6C). Deoxycytidine also severely impaired CytR-dependent transformation in a manner similar to cytidine (Fig. S1), suggesting that DNA uptake in V. cholerae is responsive to the presence of extracellular nucleosides and deoxynucleosides. Thus, a cytidine-responsive CytR-dependent nucleoside scavenging mechanism, described in E. coli (Barbier et al., 1997), appears to be a critical component of a regulatory network controlling natural competence in V. cholerae.
Why bacteria become naturally competent to take up DNA is a matter of controversy that has persisted since the pioneering studies of the ‘transforming principle’ by Griffith and later Avery, McLeod and McCarthy who established Streptococcus pneumoniae as a model organism for studying natural competence for DNA uptake (Griffith, 1928; Avery et al., 1944). Numerous studies support the hypotheses that natural competence evolved in bacteria to aid in three major processes: DNA repair, HGT and nutrition (for reviews see Redfield, 1993; Solomon and Grossman, 1996; Dubnau, 1999). However, it is acknowledged that DNA taken up by bacteria may not be used exclusively for one function or another, since extracellular DNA scavenged as nutrient may also be available for recombination onto the chromosome when of sufficient sequence identity (Dubnau, 1999). Indeed, the nutrition hypothesis has been viewed with particular scepticism as the sole explanation for maintenance of competence systems in bacteria because Neisseria gonorrhoeae and H. influenzae exclude DNA that is not species-specific, and many bacteria including B. subtilis can secrete exoenzymes to degrade extracellular DNA and then utilize nucleoside scavenging transporters for acquisition of the extracellular bases (Dubnau, 1999; Claverys et al., 2006; 2009; Mortier-Barriere et al., 2007). Our results presented here begin to define components of a nucleoside scavenging system in V. cholerae that, along with chitin and quorum sensing signalling, alters expression of transformation gene expression. Indeed, features of this emerging network suggest that regulation of V. cholerae natural competence has characteristics of pathways described below for both Gram-negative and Gram-positive bacteria.
Despite the unique features and regulatory components of the chitin- and QS-induced natural competence system described in Vibrios, the general architecture of this regulatory network shares some features with other naturally competent bacteria. In particular, Gram-positive S. pneumoniae and B. subtilis require a peptide-based QS system to regulate a phosphorylation cascade, which induces a regulator (sigma factor) that controls genes for the uptake of DNA without sequence preference (Claverys et al., 2006). In contrast, Gram-negative N. gonorrhoeae and H. influenzae only take up DNA carrying species-specific uptake sequences; yet do not appear to use QS to mediate this process (Chen and Dubnau, 2004). Interestingly, it has been proposed (Dubnau, 1999; Tortosa and Dubnau, 1999) that S. pneumoniae and B. subtilis regulate competence in response to species-specific QS AIs to limit competence induction to HCD conditions that may favour acquisition of ‘self’ and not ‘foreign’ DNA. It is believed that N. gonorrhoeae and H. influenzae, in contrast, utilize a sequence-based mechanism rather than QS to ensure ‘sexual isolation’. However, such a model is insufficient to explain the V. cholerae competence network elucidated here. First, unlike the other Gram-negative bacteria that regulate competence without QS system input, Vibrios appear to be notable exceptions to this generalization. Second, V. cholerae lacks uptake sequences and instead appears to take up DNA broadly (Suckow et al., 2011), in contrast to the archetypal Gram-negative N. gonorrhoeae and H. influenzae. Finally, the competence network in Vibrios shares additional features with Gram-positive systems by not only utilizing QS signalling but also a regulatory circuit for monitoring nutrient stress. A complex regulatory circuitry in B. subtilis co-ordinates competence and sporulation in response to nutrient cues (Grossman, 1995). So too, the CcpA regulator in Streptococcus gordonii, which orchestrates a CCR similar to that in Gram-negative bacteria, controls both biofilm formation and natural competence as a response to nutrient depravation (Zheng et al., 2012).
Vibrios commonly form biofilms on biotic chitinous surfaces, such as chironomids and zooplankton moults (Tamplin et al., 1990; Halpern et al., 2004) producing chitinases that allow exploitation of this abundant GlcNAc polymer in an otherwise nutrient-poor aquatic biosphere (Li and Roseman, 2004). Initial studies demonstrating that chitin also induced natural competence noted that the presence of glucose suppressed DNA uptake, which prompted the suggestion that competence in V. cholerae was under control of carbon catabolite repression (CCR) (Meibom et al., 2005). Indeed, Blokesch recently confirmed a role for CRP in DNA uptake by V. cholerae, although a specific mechanism for the role of CRP in competence was not validated (Blokesch, 2012).
Cameron and Redfield have proposed a model for γ-proteobacteria (including E. coli, V. cholerae and H. influenzae) that transcription from the promoters of comEA and pilABCD may be under Sxy (TfoX) control (Cameron and Redfield, 2006). Specifically, H. influenzae Sxy (TfoX) is proposed to direct CRP to interact with a competence regulatory element (CRE) sequence (TGCGA-N6-TCGCA) in the comE1 (comEA) and pilA promoters, although a precise mechanism has not been revealed to fully explain how CRP and/or Sxy engage at the CRE site (which is remarkably similar to the CRP binding site: TGTGA-N6-TCACA) (Redfield et al., 2005). Inspection of the promoter region of V. cholerae comEA that is included in our reporter fusion indicates one potential CRE site (TGCGA-N6-AAGCA); and the pilA promoter contains a potential CRP binding site or a CRE (TGAGA-N6-TCAAA), but is not in our reporter fusion (data not shown). Thus it is possible that in V. cholerae, as in H. influenzae, CRP (via TfoX) directly promotes transcription of competence genes like comEA. In addition, our results here also support a role for CRP (via CytR) in indirectly regulating a similar class of genes as described below. The independent contribution of TfoX and CytR could explain why tfoX induction (tfoX*) was required for maximal comEA expression when cytR was overexpressed (Fig. 5). Although naturally competent H. influenzae does not encode a CytR homologue, our data linking CytR to competence in V. cholerae may be useful in discovering why Enterobacteriacaea such as E. coli, Shigella and Yersinia species that encode homologues of both TfoX and CytR (Redfield et al., 2005) are not naturally transformable.
In E. coli, which is not naturally competent, the CytR regulator has been extensively studied for its role in anti-activation of a set of CRP-dependent nucleoside scavenging genes (including cytR, udp, deoP, nupG, cdd, tsx, cytX-rot) that are anti-activated by CytR (for review see Valentin-Hansen et al., 1996). However, blastp analysis suggests that CytR–CRP-regulated promoters in V. cholerae and E. coli may be different (data not shown). In E. coli, CytR-controlled promoters contain two CRP binding sites, with the exception of cytR itself, which is autorepressed and yet contains a single CRP site (Pedersen et al., 1992). This is likely the case for V. cholerae cytR as well, which appears to contain a single CRP binding site in its promoter region (data not shown). So, while we predict that the V. cholerae CytR regulon includes cytR and udp (Figs 3 and 4; Haugo and Watnick, 2002; Zolotukhina et al., 2003), the apparent V. cholerae deoP homologue, deoC, lacks two CRP sites in its promoter, and V. cholerae does not encode an obvious nupG homologue (data not shown). As a result, we are currently defining the CytR regulon by experimental and computational methods to further define whether CytR has direct or indirect effects on competence gene expression.
Our results presented here are consistent with a model that CytR–CRP interactions have a positive effect on competence gene expression and DNA uptake. However, it is likely that one or more intermediate steps exist between CytR–CRP and the competence genes described here. Only the udp promoter contains two CRP binding sites for direct anti-activation, while the promoter regions of comEA, chiA-1 and pilA do not have two CRP binding sites (data not shown). While it remains possible that V. cholerae CytR may not act identically to its E. coli counterpart as an anti-activator, we favour a speculative model supported by our data that CytR and CRP, via protein–protein interactions, interact with the promoter of a putative factor X, which in turn represses comEA, chiA-1 and pilA (Fig. 7). Given such a mechanism, WT V. cholerae would result in maximal competence gene expression, a ΔcytR mutant would maximally repress competence, and a ΔcytR, Δcrp double mutant unable to activate X could results in intermediate comEA levels (Fig. 7). However, since CRP is a pleiotropic regulator, and a V. cholerae Δcrp mutant displays a growth defect (data not shown; Skorupski and Taylor, 1996; Blokesch, 2012), it is also possible that the intermediate expression observed here for Δcrp mutants may be complex and result from changes in a CytR response, as well as consequence due to growth alterations or additional role(s) that CRP may play in directly regulating one or more competence genes. Nonetheless, our epistasis results support that the CytR-mediated effects observed here do not occur in the absence of CRP, consistent with our model that CytR function in V. cholerae requires CRP.
Haugo and Watnick demonstrated that CytR represses biofilm formation in V. cholerae strain MO10, although a direct mechanism linking CytR to biofilm genes was not revealed (Haugo and Watnick, 2002). Recently, Garavaglia et al. have also demonstrated CytR repression of E. coli biofilms, as a ΔcytR mutant displayed reduced expression of the csgDEFG operon, which controls assembly and transport of curli fibres that promote aggregation (Garavaglia et al., 2012). Modulation of intracellular pyrimidine concentrations appears responsible for the changes in curli expression leading the authors to propose that biofilm gene expression is an indirect consequence of CytR control of nucleoside pools in the cell. It remains possible that CytR functions in a similar indirect manner to control expression of the competence genes in V. cholerae and experiments are underway to test this.
In V. cholerae A1552, cytR was identified one of ∼ 100 genes positively regulated at least 2.5-fold by tfoX induction (Meibom et al., 2005); however, we did not observe an increase in cytR–lux expression in V. cholerae strain C6706 under similar conditions (data not shown). So too, we showed that maximal comEA expression required tfoX activation (tfoX*), as cytR overexpression was not sufficient to bypass the need for Tfox (Fig. 5). It is possible that TfoX controls transcription of comEA, and other competent genes via direct interactions (at CRE sites as indicated above), while CytR plays an additional independent role. Determining the identity of a putative factor X that may be anti-activated by CytR in V. cholerae could indeed provide insight into how CytR–CRP mediates its effects on competence. Complementary biochemical, bioinformatics and genomic methods are being developed to identify the set of CytR-regulated targets.
The manner in which HapR influences V. cholerae competence genes is likely to be complex. Earlier models predicting a more direct role for HapR exclusively on comEA (Antonova and Hammer, 2011; Suckow et al., 2011) appear insufficient to explain the observations here that pilA and chiA-1 are also regulated like comEA in a manner that depends on TfoX, CytR and HapR (Fig. 3). Rather than direct interaction of HapR with each promoter (or a unique upstream regulator for each gene), it is likely that HapR may also impinge on this network by directly controlling a single factor that in turn regulates multiple competence genes. Induction of cytR by HapR was not observed in prior studies or in this current study (Meibom et al., 2005, and data not shown). In accordance with this, a strain carrying a constitutive tfoX* allele still requires hapR for comEA expression and DNA uptake (Fig. 2A and B); thus like CytR, HapR likely regulates competence in a manner that is TfoX-independent. HapR also regulate udp transcription, as a ΔhapR mutant relative to an isogenic HapR+ strain is increased for udp transcription (Fig. 3D), and also slightly increased for udp transcript abundance in V. cholerae C7258 strain (Liang et al., 2007a). Lo Scrudato and Blokesch recently concluded that HapR does not regulate pilA expression, but acknowledge that variations in regulatory circuits of diverse V. cholerae strains may result in different interpretations regarding the role of quorum sensing and also CRP (Lo Scrudato and Blokesch, 2012). QS in V. cholerae C6706 controls many genes including several that alter levels of the intracellular second messenger molecule, cyclic dimeric GMP (c-di-GMP), which acts on many targets in the cell (Waters et al., 2008; Hammer and Bassler, 2009). HapR-mediated effects on competence may also be an indirect consequence of alterations in this pool of di-nucleotides in response to HCD conditions. Interplay between c-di-GMP and levels of intracellular and extracellular nucleic acids would suggest complex metabolic changes while V. cholerae is in the naturally competent state.
A similar QS system to that described in V. cholerae C6706 is used by many members of the Vibrio genus (Ng and Bassler, 2009); and chitin-induced DNA uptake has been demonstrated for several Vibrios including V. parahemolyticus, V. fischeri and V. vulnificus (Gulig et al., 2009; Chen and Ge, 2010; Pollack-Berti et al., 2010). Counterparts to the competence genes and cytR are also present in other Vibrios (Bartlett and Azam, 2005, and data not shown); however, a more complete understanding of the network connections between these systems in Vibrios is obviously required. Curiously, Vibrio harveyi does not appear to be naturally competent under the assay conditions described here for V. cholerae (Antonova and Hammer, 2011), but has been proposed to use CytR to control a regulon important for pathogenesis of marine hosts (Rattanama et al., 2012). Naturally competent V. fischeri encodes a CytR protein with an amino acid substitution at a position corresponding to F165 in V. cholerae (data not shown), suggesting that additional contacts may mediate CytR–CRP interaction in some Vibrios.
As we have proposed previously, the reliance of V. cholerae on a genus-wide QS system to control DNA uptake may be a contributing factor sculpting the genome of V. cholerae, which has undergone rampant HGT (Chun et al., 2009; Antonova and Hammer, 2011). As nucleotide scavenging, via CytR–CRP, appears to halt DNA uptake, as shown here, it may be that the evolutionary role of competence in the Vibrios includes HGT as well as nutrient acquisition. Determining whether extracellular DNA and nucleosides can support growth of V. cholerae, as shown for E. coli (Finkel and Kolter, 2001), is an important next step. Uncovering the complex network connections linking TfoX, HapR and CytR to competence will likely contribute to our knowledge of signalling in other naturally competent Vibrios, as well as elucidate an expanding role for CytR-based regulation. Defining of the CytR role in Vibrios may also be applied to understanding Enterobacteria, like E. coli, that encode CytR and many other competence genes and appear to have genetic differences that limit DNA uptake (Redfield et al., 2005; Sinha and Redfield, 2012).
Bacterial strains, plasmids and culture conditions
The relevant genotypes of the V. cholerae strains and plasmids used in the study are listed in Table 1. V. cholerae strains were incubated at 37°C on Luria–Bertani (LB) agar, and in LB broth with shaking. AB minimal medium (Clark and Maaloe, 1967) modified to include 0.7 mM Na2SO4 (C.M. Waters, unpublished) was used for bioluminescence assays where noted. Artificial sea water (ASW; Instant Ocean) was used for chitin-induced natural transformation assays as described previously (Antonova and Hammer, 2011). Antibiotics (Fisher BioReagents) chloramphenicol (Cm), kanamycin (Kan) and streptomycin (Str) were used at concentrations of 10, 100, 5000 μg ml−1 respectively. Expression of the cytR gene encoded on p-tac-cytR was induced with 0.5 mM isopropyl-b-d-thiogalactopyranoside (IPTG; Fisher BioReagents). Where noted, V. cholerae cultures were supplemented with 100 mM cytidine or deoxycytidine (Sigma).
Table 1. V. cholerae strains and plasmids used in this study
Standard protocols were used for all DNA manipulations (Sambrook, 2001). Restriction enzymes, T4 DNA ligase (New England Biolabs) and Phusion DNA polymerase (Finnzymes) were used for cloning and PCR reactions. In-frame deletions, amino acid substitutions and insertion mutants in V. cholerae were constructed by allelic exchange pKAS32-based plasmids (Skorupski and Taylor, 1996). Genomic DNA marked with KanR gene at the lacZ locus (Antonova and Hammer, 2011) was extracted using a ZR Fungal/Bacterial DNA kitTM (Zymo Research) for chitin-induced natural transformation assays. Plasmids carrying the luciferase-based transcriptional reporters (comEA–lux, pilA–lux, chiA-1–lux and udp–lux) were constructed as previously described (Antonova and Hammer, 2011). Briefly, the promoter and a portion of the coding region of corresponding gene from WT V. cholerae was PCR-amplified with an upstream primer containing a SpeI site and a downstream primer containing a BamHI site, and then cloned into the SpeI/BamHI-digested pBBRlux vector (Lenz et al., 2004). The IPTG-inducible p-tac-cytR plasmid was constructed by amplifying the entire coding region of vc2677 and cloning it into the pEVS143 vector by insertion into the EcoRI and BamHI restriction sites. The primer sequences used for plasmid construction and additional cloning details are described in Table S1.
Transposon mutagenesis of V. cholerae
The suicide delivery plasmid pRL27 encoding the Tn5 transposon conferring resistance to kanamycin (KanR) (Larsen et al., 2002) was transferred by conjugation from E. coli S17λpir to the V. cholerae ΔluxO, tfoX* recipient strain merodiploid for hapR (Table 1, EA349). Three independent pools of ∼ 105 Tn5 KanRV. cholerae mutants were conjugated with an E. coli S17λpir donor carrying the comEA–lux reporter plasmid to create a library of transposon mutants. KanR transconjugant mutant colonies were arrayed to microtitre plates with a Genetix QPix2XT colony picker followed by screening for candidates with defective comEA–lux expression using a BioTek multimode plate reader. The identity of candidate target genes found in the screen was determined by blast analysis to the V. cholerae C6706 genome of the DNA sequences adjacent to the Tn5 insertion, by standard methods described previously (Hammer et al., 2002; Larsen et al., 2002).
Vibrio cholerae bioluminescence expression was assayed as described previously (Miller et al., 2002; Zhu et al., 2002). V. cholerae strains carrying a lux-based reporter plasmid were incubated in LB at 37°C overnight with appropriate antibiotics and IPTG where noted, and bioluminescence and absorbance were quantified thereafter. For bioluminescence measurements in the presence of cytidine, V. cholerae strains carrying a reporter plasmid were incubated overnight for 9–11 h at 37°C in AB minimal media containing appropriate antibiotics and supplemented with 100 mM cytidine. A chitin tile was added to each tube to provide a carbon source and for induction of tfoX. Bioluminescence was measured using a Wallac model 1409 liquid scintillation counter as described previously (Hammer and Bassler, 2007). The optical density of each culture was measured with a spectrophotometer. Relative Light Units (RLU) are defined as counts min−1 ml−1/OD600. Single-time-point experiments were performed with triplicate samples.
Chitin-induced natural transformation assay
Chitin-induced transformation experiments were performed as described previously (Meibom et al., 2005). Two micrograms of V. cholerae genomic DNA marked with a KanR gene at the lacZ locus was used as an extracellular DNA source. For enumeration of transformants, cultures were plated onto LB medium containing Kan. Transformation frequency was defined as KanR cfu ml−1/total cfu ml−1. To measure transformation frequencies of V. cholerae strains in the presence of cytidine, ASW medium was supplemented with 100 mM cytidine (or 100 mM deoxycytidine in Fig. S1). Independent experiments were performed in triplicate.
We thank the Hammer lab for discussions and critical manuscript review, and B. Duke and M. Nellesson for assistance with luciferase assays. V. Rajan and M. Borodovsky provided support for genome analysis. This study was funded by National Science Foundation grants (MCB-0919821 and MCB-1149925) to B.K.H.