Regulation of natural competence by the orphan two-component system sensor kinase ChiS involves a non-canonical transmembrane regulator in Vibrio cholerae



In Vibrio cholerae, 41 chitin-inducible genes, including the genes involved in natural competence for DNA uptake, are governed by the orphan two-component system (TCS) sensor kinase ChiS. However, the mechanism by which ChiS controls the expression of these genes is currently unknown. Here, we report the involvement of a novel transcription factor termed ‘TfoS’ in this process. TfoS is a transmembrane protein that contains a large periplasmic domain and a cytoplasmic AraC-type DNA-binding domain, but lacks TCS signature domains. Inactivation of tfoS abolished natural competence as well as transcription of the tfoR gene encoding a chitin-induced small RNA essential for competence gene expression. A TfoS fragment containing the DNA-binding domain specifically bound to and activated transcription from the tfoR promoter. Intracellular TfoS levels were unaffected by disruption of chiS and coexpression of TfoS and ChiS in Escherichia coli recovered transcription of the chromosomally integrated tfoR::lacZ gene, suggesting that TfoS is post-translationally modulated by ChiS during transcriptional activation; however, this regulation persisted when the canonical phosphorelay residues of ChiS were mutated. The results presented here suggest that ChiS operates a chitin-induced non-canonical signal transduction cascade through TfoS, leading to transcriptional activation of tfoR.


Cellular functions are regulated in response to environmental stimuli by signal transduction pathways. In bacteria, two-component systems (TCSs) play predominant roles in transducing extracellular signals into the cell (Parkinson, 1993; Hoch and Silhavy, 1995; Stock et al., 2000; Mascher et al., 2006). The classical TCS is composed of a transmembrane sensor histidine kinase (HK) that regulates the activity of a single cytoplasmic response regulator (RR). In response to specific signals, the HK autophosphorylates by transferring a phosphoryl group from ATP to a histidine residue in its conserved kinase domain; the phosphoryl group is then transferred to an aspartate residue in the conserved receiver domain of the cognate RR. Phosphorylation of RR changes the activity of its output domain, which, in many cases, can control transcription through DNA binding (Galperin, 2010). A variant of the classical TCS signal transduction process involves multistep phosphorelays that are mediated by a hybrid HK containing both HK and RR domains but no output domain. In this process, the phosphoryl group is transferred from the hybrid HK to a histidine-containing phosphotransferase (HPt), and then to a terminal RR that possesses an output domain (Appleby et al., 1996). In some cases, HPt is incorporated into a single hybrid HK molecule (Kato et al., 1997).

TCS proteins are characterized by the presence of several signature domains: the HK A domain (HisKA, NCBI Accession Number: 119399), the HK ATPase domain (HATPase_c, NCBI Accession Number: 238030), the RR receiver domain (REC, NCBI Accession Number: 238088), and the HPt domain (NCBI Accession Number: 238041). Many bacteria possess multiple copies of TCS genes, ranging from a few to 278 copies (Whitworth and Cock, 2008). Genes encoding classical TCSs usually comprise an operon that contains both the HK and RR genes. Some bacteria also encode ‘hybrid’ TCS (HTCS) proteins that include all domains found in HKs and RRs in a single polypeptide (Sonnenburg et al., 2006). It is thought that such gene or protein organization may be the result of adaptation to avoid unwanted cross-phosphorylation between HK and non-cognate RR (Laub and Goulian, 2007). However, recent advances in bacterial genome sequencing have uncovered that multiple HK and RR genes are located independently, which are referred to as ‘orphan’ TCS (OTCS) genes. The Prokaryotic 2-Component System database ( suggests that different bacterial species have distinct numbers of OTCS genes; 23% of TCS genes in Escherichia coli are orphan, compared with 68% in Caulobacter crescentus and 74% in Synechococcus elongates (Barakat et al., 2011). These genetic organizations further complicate the investigation of connectivity between TCS proteins.

Vibrio cholerae causes the fatal diarrhoeal disease cholera. The sequenced V. cholerae strain N16961 contains 92 TCS genes (43 HK and 49 RR genes), of which 44 (48%, 21 HK and 23 RR genes) are orphan (Barakat et al., 2011). V. cholerae lives in aquatic environments, including rivers, estuaries, and coastal regions, and is often attached to the chitinous exoskeleton of zooplankton (Pruzzo et al., 2008). Chitin, a polymer of N-acetylglucosamine (GlcNAc), is the most abundant polysaccharide in the aquatic biosphere and is an important source of carbon and nitrogen for marine microorganisms. Attachment of V. cholerae to the chitin surface of zooplankton not only provides a nutrient-rich environment for growth (Nalin et al., 1979; Heidelberg et al., 2002), but may also aid the transmission of this pathogen from aquatic reservoirs to susceptible human hosts (Colwell et al., 2003). Chitin has multiple effects on the physiology of V. cholerae and other Vibrio spp., including the development of colonization and biofilm, stimulation of chemotaxis, survival during temperature and acid stresses, and the induction of natural competence for genetic transformation (see reviews, Pruzzo et al., 2008; Seitz and Blokesch, 2013a; Sun et al., 2013).

Comparative genomic studies with different strains revealed the mosaic-structured V. cholerae genome generated by extensive recombination via horizontal gene transfer (HGT) (Chun et al., 2009; Dziejman et al., 2002; 2005). Classically, transduction (e.g. the ctx genes encoding cholera toxin) and conjugation (e.g. the SXT integrating conjugative element encoding resistance to multiple antibiotics) were presumed to be the main contributors of HGT in V. cholerae (Waldor and Mekalanos, 1996; Waldor et al., 1996); however, this bacterium has other horizontally acquired genes that are not associated with known mobile genetic elements such as phages and plasmids or other clear signatures of HGT (Faruque and Mekalanos, 2003). Natural transformation induced by chitin has been only recently recognized as additional mechanism of HGT employed by V. cholerae (Meibom et al., 2005). Natural competence is the ability of bacteria to actively taken up DNA from the environment. The imported DNA is recombined into the recipient chromosome, thereby undergoing transformation. This mode of gene transfer does not require mobile genetic elements and is solely dependent on the recipient bacterium. Blokesch and Schoolnik demonstrated that the whole O1-specific antigen gene cluster of V. cholerae El Tor can be exchanged by the O37- or O139-antigen gene cluster through chitin-induced natural competence (Blokesch and Schoolnik, 2007). In addition, other virulence-associated large genomic islands (Udden et al., 2008; Morita et al., 2013) or variable genomic segments (Miller et al., 2007) can be transferred in the same manner. Recently, various members of the Vibrionaceae family, including Vibrio vulnificus (Gulig et al., 2009; Neiman et al., 2011), Vibrio fischeri (Pollack-Berti et al., 2010), and Vibrio parahaemolyticus (Chen et al., 2010), were shown to be naturally transformable upon exposure to chitin. Thus, chitin-induced natural competence provides an important subject to study how environmental signals drive the evolution of these aquatic bacteria.

Pioneering studies to understand chitin-induced global changes in V. cholerae gene expression have been performed previously (Meibom et al., 2004; 2005). Using microarray gene expression profiling, Meibom et al. showed that growth of V. cholerae on chitin or its oligosaccharide derivatives ((GlcNAc)n≥2) induces the expression of 41 genes, all of which are governed by an orphan HK named ChiS. Most of these genes are involved in chitin utilization, including a (GlcNAc)2 catabolic operon (chb), two extracellular chitinase genes (chiA-1 and chiA-2), and an outer membrane chitoporin gene (chiP). The genes required for natural competence, including one that encodes a transcriptional regulator (tfoX) and 15 that encode a predicted type IV pilus (pilA, pilB, and pilQ, etc.) (Seitz and Blokesch, 2013b), are also chitin-inducible. It is thought that V. cholerae evolved to couple sensing of the availability of chitin with its utilization and DNA uptake. This strategy might promote a niche-specific acquisition of genetic information. Based on genetic analyses, Li and Roseman proposed a unique mechanism for chitin perception and signalling (Li and Roseman, 2004). In the absence of chitin, the activity of ChiS is inhibited by a periplasmic chitin oligosaccharide-binding protein (CBP). In the presence of chitin, its major degradation product (GlcNAc)2 competes with ChiS for CBP binding, and then free ChiS activates a cytoplasmic RR that interacts with target promoters. However, the cognate RR for ChiS has not yet been identified.

Members of the ChiS regulon are controlled by two distinct pathways, one of which is independent of the transcriptional regulator TfoX and one that is TfoX-dependent (Meibom et al., 2004; 2005). TfoX was originally identified in Haemophilus influenzae as the central regulator of the competence regulon (Karudapuram and Barcak, 1997), which activates the transcription of competence genes in collaboration with the cAMP-CRP complex (Macfadyen et al., 1996; Redfield et al., 2005). In V. cholerae, TfoX is also dependent on cAMP-CRP (Antonova et al., 2012; Blokesch, 2012) and it positively regulates both competence genes and the chitin utilization genes chiA-1, chiA-2, and chiP (Meibom et al., 2004; 2005; Antonova et al., 2012; Lo Scrudato and Blokesch, 2012; 2013). By contrast, transcription of the chb operon in V. cholerae is independent of TfoX (Meibom et al., 2004; 2005).

We recently described the mechanisms underlying chitin-induced activation of tfoX expression (Yamamoto et al., 2010; 2011): Upon exposure to chitin, or more specifically (GlcNAc)2, the expression of tfoX is activated at both the transcriptional and translational levels, and the translational effect is much stronger than the transcriptional effect (Yamamoto et al., 2010). A subsequent genetic study led to the identification of an Hfq-dependent small RNA (sRNA), TfoR, which is essential for translation of tfoX mRNA and thus governs the TfoX pathway (Yamamoto et al., 2011). Since its transcription is induced by (GlcNAc)2, it is expected that tfoR is under the control of ChiS (Yamamoto et al., 2011). Studying the regulatory mechanism of tfoR expression will help to understand how the signalling pathways initiated by ChiS are transmitted to activate the TfoX pathway.

In this study, we describe the identification of a novel transmembrane regulator (TfoS) responsible for transcriptional activation of tfoR. The protein has no TCS signature domains, but its activity is strictly dependent on ChiS. The results presented here provide new insights into the connectivity between TCS and non-TCS proteins and their mechanisms of action during bacterial signal transduction.


ChiS is involved in chitin-induced expression of tfoR

Induction of natural competence requires the chitin-induced small RNA TfoR that activates translation of tfoX mRNA (Yamamoto et al., 2011). Although tfoR is thought to be a member of the ChiS regulon, direct evidence for the involvement of this HK in chitin-induced expression of tfoR has not been reported to date. Disruption of chiSchiS) abolished (GlcNAc)2-induced expression of tfoR and the competence genes tfoX and pilA (Table 1). As reported previously (Meibom et al., 2004; 2005), expression of the chb operon, which encodes proteins involved in the uptake and catabolism of (GlcNAc)2, was abolished following disruption of chiS but was not affected by disruption of tfoX (Table 1). In fact, both tfoR and tfoX were dispensable for chb expression (Table 1), suggesting that transcription of chb is controlled by a TfoR/TfoX-independent pathway.

Table 1. The effects of chiS and tfoS on expression of various lacZ fusion genes in V. cholerae
lacZ fusionaMutationPlasmidβ-Galactosidase activityb (Miller units ± SD) in M9L medium plus:Fold activationc
2.5 mM GlcNAc2.5 mM (GlcNAc)2
  1. aThe strains used were as follows: tfoR::lacZ, SY0627S; tfoR::lacZ ΔchiS, SY0627S30K; tfoR::lacZ chiSH469A, SY0627S32; tfoR::lacZ chiSH772A, SY0627S33; tfoR::lacZ ΔtfoS, SY0627S34; tfoR::lacZ tfoS::3FLAG, SY0627S36K; tfoX::lacZ, SY0616S; tfoX::lacZ ΔchiS, SY0616S30; tfoX::lacZ ΔtfoS, SY0616S34; pilA::lacZ, SY0623; pilA::lacZ ΔchiS, SY062330K; pilA::lacZ ΔtfoS, SY062334K; chb::lacZ, SY0635S; chb::lacZ ΔchiS, SY0635S30K; chb::lacZ chiSH469A, SY0635SS32; chb::lacZ chiSH772A, SY0635S33; chb::lacZ ΔtfoS, SY0635S34; chb::lacZ ΔtfoR, SY0635S26; and chb::lacZ ΔtfoX, SY0635S08.
  2. bβ-Galactosidase activity units are presented as means ± standard deviations. Experiments were performed three times.
  3. cFold activation is indicated as the relative difference from the value for the culture containing only GlcNAc.
tfoR::lacZ10 ± 1125 ± 213
tfoR::lacZΔchiS10 ± 212 ± 11.2
tfoR::lacZchiSH469A24 ± 3116 ± 44.8
tfoR::lacZchiSD772A21 ± 1124 ± 85.9
tfoR::lacZΔtfoS11 ± 112 ± 11.1
tfoR::lacZΔtfoSpMW1188 ± 19 ± 21.1
tfoR::lacZΔtfoSpMW-SYtfoS122 ± 1236 ± 511
tfoR::lacZtfoS::3×FLAG8 ± 1105 ± 1013
tfoX::lacZ23 ± 2669 ± 729
tfoX::lacZΔchiS21 ± 139 ± 41.9
tfoX::lacZΔtfoS18 ± 131 ± 51.7
pilA::lacZ19 ± 1899 ± 547
pilA::lacZΔchiS23 ± 321 ± 50.9
pilA::lacZΔtfoS21 ± 312 ± 10.6
chb::lacZ14 ± 1311 ± 1922
chb::lacZΔchiS20 ± 337 ± 61.9
chb::lacZchiSH469A31 ± 1241 ± 217.8
chb::lacZchiSD772A21 ± 1321 ± 1615
chb::lacZΔtfoS18 ± 3292 ± 816
chb::lacZΔtfoR16 ± 1291 ± 218
chb::lacZΔtfoX18 ± 1278 ± 315

(GlcNAc)2-induced competence for exogenous DNA was not detected for the ΔchiS mutant (Table 2). Shrimp shell was then used as a chitin source to determine competence in a near-natural environment (Morita et al., 2013). This condition induced competence with a high frequency of 2.1 × 10−4 in the wild-type strain, which was more than 1000-fold higher transformation frequencies than that induced by (GlcNAc)2 (Table 2). However, even in the presence of shrimp shell, the ΔchiS mutant still displayed a non-competent phenotype (Table 2). These results indicate that ChiS is essential for chitin-induced expression of tfoR and natural competence.

Table 2. The effects of chiS and tfoS on natural transformation
Recipient strainaPlasmidTransformation efficiency (Cmr cfu/total cfu)b in:
M9L medium plus:DASWLc plus shrimp shell
2.5 mM GlcNAc2.5 mM (GlcNAc)2
  1. aThe strains used were as follows: wild type, V060002; ΔchiS, SY0630K; ΔtfoS, SY0634S; tfoS::3×FLAG, SY0636K; ΔtfoR, SY0626K; ΔtfoX, SY0605S; ΔpilA, SY0609S.
  2. bRepresentative data from three independent experiments are shown. The transformation efficiencies of strains carrying plasmids were defined as Cm- and Ap-resistant cfu divided by Ap-resistant cfu.
  3. cDistilled artificial seawater medium containing 0.5% lactate (pH 7.6).
  4. d< DL, below detection limit (∼ 3.0 × 108 and ∼ 9.0 × 109 in M9L and DASWL respectively).
Wild type< DLd1.1 × 10−72.1 × 10−4
ΔchiS< DL< DL< DL
ΔtfoS< DL< DL< DL
ΔtfoSpMW118< DL< DL< DL
ΔtfoSpMW-SYtfoS1< DL2.5 × 10−72.5 × 10−4
tfoS::3×FLAG< DL9.0 × 10−86.8 × 10−5
ΔtfoR< DL< DL< DL
ΔtfoX< DL< DL< DL
ΔpilA< DL< DL< DL

A gene encoding a unique transcription factor is located close to tfoR

ChiS is a TCS HK (Li and Roseman, 2004); therefore, it seems unlikely that it regulates the transcription of tfoR directly. We hypothesized that the AraC-type transcription factor encoded adjacent to tfoR in all Vibrio spp. sequenced to date (Yamamoto et al., 2011; Fig. 1 and Fig. S1) was an orphaned RR regulated by ChiS. The gene encoding the AraC homologue in V. cholerae is annotated as vc2080 (Fig. 1) and encodes a predicted polypeptide of 1121 amino acids. A search for conserved domains in VC2080 revealed the presence of a periplasmic sensor domain (NCBI Accession Number: 225829) in the N-terminal region (residues 9–870) and an AraC-type helix–turn–helix (HTH) DNA-binding domain (HTH_18, NCBI Accession Number: 221799) in the C-terminal region (residues 1041–1115) (Fig. 2A). VC2080 is structurally related to HTCS proteins that incorporate all domains found in classical HKs and RRs into a single polypeptide (Sonnenburg et al., 2006). Although TCS signature domains were not identified in VC2080, the sensor domain of this protein is partially homologous (19–25% amino acid identity) to fragments of the sensor domains of multiple other HTCS proteins (data not shown). The sensor domains of most HTCS proteins comprise 14 repetitive β-strand units that adopt a double β-propeller fold (Reg_prop, NCBI Accession Number: 219429) (Menke et al., 2010). The HHrep program (Soding et al., 2006) predicted that VC2080 contains 14 repeat sequences in the sensor domain, each consisting of four β-strands (Fig. 2A and Fig. S2). These repeat sequences were similar to those of BT_4663 (Fig. S3), a structurally resolved HTCS protein from Bacteroides thetaiotaomicron (Lowe et al., 2012). Moreover, membrane topology prediction using the TMHMM program (Krogh et al., 2001) suggested that VC2080 may be an inner membrane protein with two transmembrane domains (residues 7–29 and 736–758), comprising the sensor and HTH_18 domains oriented to the periplasm and cytoplasm respectively (Fig. 2B). Intriguingly, the results described below suggest that VC2080 directly controls the transcription of tfoR and that its activity is strictly dependent on ChiS. We therefore designated vc2080 as ‘tfoS’ (tfoR regulator dependent on ChiS).

Figure 1.

Genomic colocalization of the tfoR and tfoS genes in Vibrio spp.

Bacterial strains shown are as follows: VC, V. cholerae N16961; VP, V. parahaemolyticus 2210633; VV, V. vulnificus CMCP6; VF, V. fischeri ES114.

Figure 2.

Schematic representation of domain structures in TfoS.

A. Predicted domains in TfoS (VC2080, NCBI Accession Number: NP_231712.1). Searches using the PSI-BLAST (Altschul et al., 1997) and TMHMM (Krogh et al., 2001) programs indicated the presence of conserved periplasmic sensor and HTH_18 domains, as well as two transmembrane (TM) domains. Sequence similarity with the HTH domain in AraC is shown; asterisks indicate identical residues. The conserved HTH motif of the AraC family (Gallegos et al., 1997) is indicated below the sequence. An HHprep analysis (Soding et al., 2006) also predicted the presence of 14 repetitive β-strands (Rps).

B. Membrane topology analysis of TfoS in E. coli. The predicted inner membrane location and topology of TfoS (VC2080) showing the two transmembrane domains (residues 7–29 and 736–758), comprising the sensor and HTH_18 domains oriented to the periplasm and cytoplasm respectively. Enzymatic activities of PhoA and LacZ proteins fused to the indicated positions of TfoS and expressed in SYEC002K were measured. The AP (PhoA) and β-galactosidase (LacZ) activities are presented as the mean ± standard deviation. Experiments were performed three times. The following plasmids were used: 1121 amino acid PhoA, pMW-SYtfoS::phoA1; amino acid 800 PhoA fusion, pMW-SYtfoS::phoA2; amino acid 500 PhoA fusion, pMW-SYtfoS::phoA3; amino acid 50 PhoA fusion, pMW-SYtfoS::phoA4; amino acid 1121 LacZ fusion, pMW-SYtfoS::lacZ1; amino acid 800 LacZ fusion, pMW-SYtfoS::lacZ2; amino acid 500 LacZ fusion, pMW-SYtfoS::lacZ3; amino acid 50 LacZ fusion, pMW-SYtfoS::lacZ4. The PhoA and LacZ activities of control cells carrying the pMW-SYphoA and pMW-SYlacZ, were 3 ± 1 and 5 ± 1 units respectively. OM, Peri, IM, and Cyto indicate the outer membrane, periplasm, inner membrane, and cytoplasm respectively.

TfoS is required for transcriptional induction of tfoR by chitin

Next, a deletion mutant of tfoS was constructed and its phenotype was examined. The ΔtfoS mutant was unable to induce the expression of tfoR, tfoX, or pilA (Table 1), and thus was non-competent (Table 2). The defects of tfoR expression and natural competence were complemented by the introduction of a low copy plasmid carrying tfoS (Tables 1 and 2). However, unlike the ΔchiS mutant, the ΔtfoS mutant displayed normal (GlcNAc)2-induced expression of chb, which is regulated independently of TfoX (Table 1). These data suggest that TfoS selectively controls the TfoX pathway in the ChiS regulon via transcriptional induction of tfoR.

ChiS and TfoS are sufficient to activate the tfoR promoter in Escherichia coli

To determine whether ChiS and TfoS are required for transcription of tfoR, an experimental system was constructed in E. coli. In this system, the lacZ promoter was replaced with the tfoR promoter to generate a chromosomally engineered strain with a tfoR::lacZ fusion (SYECREP1S), and then the transcriptional activity of the fusion gene was measured when ChiS and TfoS were expressed under the control of the araBAD promoter of pBAD33 (Guzman et al., 1995) and the lac promoter of pMW118 respectively. In this system, ChiS is expected to exist in a constitutively active state, because E. coli does not express CBP, an inhibitor of ChiS (Li and Roseman, 2004). Whereas the induction of either ChiS or TfoS individually failed to activate transcription of the tfoR::lacZ fusion gene, the induction of both of these proteins successfully stimulated transcription (Table 3). These results suggest that ChiS and TfoS are sufficient to activate the transcription of tfoR in E. coli.

Table 3. The effect of ChiS and TfoS on expression of the tfoR::lacZ fusion gene in E. coli
Strain backgroundaPlasmid 1 (mutation)bPlasmid 2 (mutation)β-Galactosidase activityc (Miller units ± SD) in LB
  1. aThe strains used were as follows: wild type, SYECREP1S; DR1, SYECREP2S; DR2, SYECREP3S.
  2. bAlterd amino acid residues are indicated.
  3. cβ-Galactosidase activity units are presented as means ± standard deviations. Assays were done from cells cultured in LB containing Cm, Ap, IPTG, and arabinose. When cells carried only pMAL-c2-based plasmid, LB containing Ap and IPTG was used for culture. Experiments were performed three times.
Wild typepMW118pBAD3324 ± 1
 pMW-SYtfoS1 (wild type)pBAD3329 ± 1
 pMW118pBAD-SYchiS1 (wild type)26 ± 1
 pMW-SYtfoS1 (wild type)pBAD-SYchiS1 (wild type)112 ± 4
 pMW-SYtfoS1 (wild type)pBAD-SYchiS2 (H469A)136 ± 4
 pMW-SYtfoS1 (wild type)pBAD-SYchiS3 (D772A)113 ± 7
 pMW-SYtfoS1 (wild type)pBAD-SYchiS4 (H469A and D772A)98 ± 4
 pMW-SYtfoS2 (Δ871 to 1,121)pBAD3325 ± 3
 pMW-SYtfoS2 (Δ871 to 1,121)pBAD-SYchiS1 (wild type)27 ± 3
 pMW-SYtfoS3 (Δ9 to 870)pBAD33116 ± 5
 pMW-SYtfoS3 (Δ9 to 870)pBAD-SYchiS1 (wild type)104 ± 3
 pMAL-c244 ± 4
 pMAL-SYtfoS3 (Δ1 to 870)497 ± 12
DR1pMAL-c265 ± 2
 pMAL-SYtfoS3 (Δ1 to 870)188 ± 8
DR2pMAL-c259 ± 2
 pMAL-SYtfoS3 (Δ1 to 870)59 ± 5

TfoS is an inner membrane protein with at least two functional domains

Next, we examined the subcellular localization of TfoS in bacterial cells. To detect the TfoS protein expressed from V. cholerae cells, a 3×FLAG sequence was chromosomally fused to the 3′-end of the tfoS gene. The response of the FLAG-tagged strain to (GlcNAc)2 was similar to that of the wild-type strain (Tables 1 and 2) As a control, the pMW-SYmtlA::phoA1 plasmid, which expresses the inner membrane protein MtlA::PhoA (Sugiyama et al., 1991) and the periplasmic control protein Bla, was introduced into the tfoS::3×FLAG strain. The bacterial cells were then fractionated and the expression levels of TfoS::3×FLAG, RpoA, Bla, MltA::PhoA, and OmpA in each fraction were examined by immunoblotting. As expected, the RpoA, Bla, MltA::PhoA, and OmpA proteins were detected in the cytoplasmic, periplasmic, inner membrane and cytoplasmic, and outer membrane fractions respectively. TfoS::3×FLAG was predominantly detected in the inner membrane fraction and the localization of this fusion protein was not affected by exposure of the cells to (GlcNAc)2 (Fig. 3).

Figure 3.

Subcellular localization of TfoS in V. cholerae. Immunoblot detection of TfoS::3×FLAG protein and control proteins (OmpA, OM; Bla, Peri; MtlA::PhoA, IM; and RpoA, Cyto) in outer membrane (OM), periplasmic (Peri), inner membrane (IM), and cytoplasmic (Cyto) fractions of a tester V. cholerae strain (SY0636K carrying pMW-SYmtlA::phoA1) grown in M9L medium containing GlcNAc or (GlcNAc)2. The proteins were separated by SDS-PAGE and then transferred onto a PVDF membrane. The TfoS::3×FLAG protein in each fraction was detected using an anti-FLAG antibody.

To determine the validity of the predicted membrane topology of TfoS, the PhoA-LacZ fusion system (Manoil, 1991) was used. In this system, fusions yielding periplasmic PhoA are highly active, whereas those yielding cytoplasmic PhoA are inactive. Fusions to LacZ show a reciprocal behaviour to those of PhoA. This system could not be tested directly in V. cholerae because this bacterium has a strong endogenous PhoA activity (data not shown). Therefore, the activities of various TfoS::PhoA and TfoS::LacZ fusions were measured in an E. coli strain lacking the phoA and lacZ genes. The PhoA fusions that presumably oriented to the periplasmic space (fusion points: residues 50 and 500) had higher activities than those oriented to the cytoplasm (fusion points: residues 800 and 1121) (Fig. 2B). Conversely, the activities of the LacZ fusions oriented to the cytoplasm were higher than those oriented to the periplasm (Fig. 2B). Taken together, the results of the cell fractionation and membrane topology analyses strongly suggest that TfoS is an inner membrane protein with an N-terminal periplasmic sensor domain and a C-terminal cytoplasmic HTH_18 domain.

To characterize the functions of these domains, the transcriptional activity of tfoR was determined in the E. coli strain SYECREP1S coexpressing ChiS and deletion variants of TfoS. Following deletion of its HTH_18 domain (residues 871–1121), TfoS lost the ability to activate tfoR transcription. By contrast, a variant lacking the periplasimic sensor domain (residues 9–870) activated transcription in both the presence and absence of ChiS (Table 3). These results indicate that the DNA-binding domain of TfoS is essential for transcriptional activation, while the sensor domain has an inhibitory effect on transcription. Upon the activation of TfoS, ChiS might transduce signal to the periplasmic sensor domain.

ChiS does not regulate TfoS at the transcriptional or translational level

To further characterize the connection between ChiS and TfoS, we examined the effect of chiS on TfoS expression in V. cholerae. The expression of the TfoS::3×FLAG protein was unaffected by deletion of chiS or the presence of (GlcNAc)2 (Fig. 4). This result raises the possibility that ChiS regulates TfoS at the post-translational level. Notably, in addition to the expected TfoS::3×FLAG protein band (125 kDa), a 35 kDa band was also detected (Fig. 4). The intensity of this band was dependent on the presence of both (GlcNAc)2 and the chiS gene, but was not affected by the FLAG-tag, suggesting that this product is a chitin-inducible protein that non-specifically reacts with the anti-FLAG antibody.

Figure 4.

The effect of chiS on TfoS expression. Immunoblot of total proteins from cells grown in M9L medium containing 2.5 mM GlcNAc or (GlcNAc)2. The proteins were separated on a 5–20% SDS-polyacrylamide gel, transferred onto a PVDF membrane, and then subjected to immunoblotting with an anti-FLAG antibody (upper panel) or anti-RpoA antibody (lower panel). In addition to the 125 kDa TfoS::3×FLAG protein, a 35 kDa band that was dependent on both (GlcNAc)2 and the chiS gene, but independent of the FLAG-tag sequence, was also detected. This protein is likely to be a chitin-inducible protein that reacts non-specifically with the anti-FLAG antibody. The following strains were used: WT (wild type), VC060002; WT tfoS::3×FLAG, SY0636K; ΔchiS tfoS::3×FLAG, SY0636K30.

The canonical phosphorelay of ChiS is dispensable for TfoS activity

ChiS, which is predicted to comprise 1129 amino acids, is a hybrid-type HK that contains both a HisKA domain (residues 457–523) with a predicted phosphorylation site at H469, and a REC domain (residues 725–839) with a predicted phosphorylation site at D772 (Fig. 5), suggesting that it is capable of intramolecular phosphorelay. A previous study reported that ChiS possesses an HPt domain in its C-terminal region (Li and Roseman, 2004); however, after careful re-annotation, we were unable to detect this domain. To determine whether the phosphotransfer activity of ChiS is crucial for signal transduction, its conserved phosphorelay residues were mutated to non-phosphatable alanines. The H469A and D772A single and double mutations had no significant effect on expression of the tfoR::lacZ fusion gene when ChiS and TfoS were coexpressed (Table 3). These mutations were also introduced into the native chromosomal copy of the chiS gene in a parental V. cholerae strain expressing the tfoR::lacZ fusion. Unlike the chiS null mutant, the alanine-substituted ChiS mutants still displayed a tfoR transcriptional response to (GlcNAc)2 (Table 1). These results suggest that the classical phosphorelay residues in ChiS are not essential for its function in vivo. The alanine mutations also had little effects on the transcription of chb, which is regulated independently of TfoS (Table 1).

Figure 5.

Schematic representation of domain structures in ChiS. Predicted domains in ChiS (VC0622, NCBI Accession Number: NP_230271.2) are shown. The PSI-BLAST (Altschul et al., 1997) and TMHMM (Krogh et al., 2001) analyses predicted that ChiS is a hybrid-type HK containing multiple TCS signature domains (HisKA, HATPase_c, and REC). This protein also has two transmembrane (TM) domains, a HAMP domain (NCBI Accession Number: 100122), and a PAS domain (NCBI Accession Number: 214512). The HisKA and REC domains contain a highly conserved H469 and D772 residue respectively. Examples of known phosphorylation sites in other hybrid-type HKs are shown in the alignment; asterisks indicate identical residues.

Identification of the tfoR promoter

We previously reported that tfoR comprises 102 nucleotides (nt) (Yamamoto et al., 2011); however, gene expression analyses of lacZ fusions inserted into different positions of tfoR suggested that the promoter region might be located more than 30 nt upstream of the 5′-end of the 102 nt RNA, and that the mature RNA may be processed from an as yet unidentified primary transcript (Yamamoto et al., 2011). The sequence ‘TATAGT’ (Fig. 6A), which is conserved in Vibrio spp. and resembles the −10 sequence of the σ70–type promoter (Yamamoto et al., 2011; Fig. S4), was identified as a candidate tfoR promoter; however, a region corresponding to the −35 sequence was not identified. As expected, when the lacZ gene was inserted immediately upstream of the −10 sequence, its expression was no longer induced, even when cells were grown in the presence of (GlcNAc)2 (Table 4). If the TATAGT sequence acts as the tfoR promoter, an RNA of at least 130 nt should be produced. Northern blotting using large amounts of total RNA (five times greater than the amount used in previous experiments) (Yamamoto et al., 2011) and Probe N1, which covered a 30 nt sequence within the 102 nt tfoR RNA (Fig. 6A), revealed the presence of an additional RNA transcript of approximately 130 nt (Fig. 6B). Production of both this longer transcript and the expected 102 nt RNA was dependent on the presence of the chiS, tfoS, and tfoR genes as well as (GlcNAc)2 (Fig. 6B and C). Northern blotting using Probe N2, which covered the 30 nt sequence immediately upstream of the 5′-end of the 102 nt RNA (Fig. 6A), only detected the 130 nt transcript (Fig. 6B). These results indicate that the two tfoR RNA transcripts have different 5′-ends.

Figure 6.

Identification of the tfoR promoter.

A. Schematic illustration of the tfoR promoter region. Bold letters indicate the sequence transcribed from the promoter. The transcriptional start site and post-transcriptional processing site are indicated as +1 and +31 respectively. The promoter sequence at position −10 is also shown.

B. Detection of transcripts from the tfoR gene. Total RNA was isolated from cells grown in M9L medium containing 2.5 mM GlcNAc or (GlcNAc)2, separated on a polyacrylamide gel containing 6 M urea, and then transferred onto a nylon membrane. RNA products were detected by Northern blotting with Probes N1 or N2, which covered the positions indicated in panel A. The V060002 strain was used.

C. The effect of chiS and tfoS on tfoR expression. Northern blot of total RNAs from cells grown in M9L medium containing 2.5 mM GlcNAc or (GlcNAc)2. The RNAs were separated on a 6% polyacrylamide gel containing 6 M urea, transferred onto a nylon membrane, and then tfoR RNA was detected by Northern blotting with Probe N1. As a loading control, Probe N3 was used to detect the expression of 5S rRNA. The following strains were used: WT (wild type), VC060002; ΔchiS, SY0630K; ΔtfoS, SY0634; ΔtfoR, SY0626K.

D. The tfoR products amplified by 5′-RACE. The two major bands indicated by arrows were cloned into pGEM-T and transformed in E. coli. Ten clones obtained from independent transformations were sequenced.

E. Characterization of the 5′-phosphorylation state of the tfoR RNAs. Total RNA isolated from cultures treated with 2.5 mM (GlcNAc)2 (panel A) was incubated with (+) or without (−) TAP or AP, and then digested with 5′-P-exo. The tfoR RNAs were detected by Northern blotting with Probe N1.

Table 4. Analysis of expression of lacZ inserted into the upstream region of tfoR in V. cholerae
Inserted positionaβ-Galactosidase activityb (Miller units ± SD) in M9L minimal medium plus:Fold activationc
2.5 mM GlcNAc2.5 mM (GlcNAc)2
  1. aThe position of insertion of lacZ is indicated as distance from the previously identified 5′-end of tfoR (Yamamoto et al., 2011). The strains used were as follows: +78, SY0627; −30, SY0628; −41, SY0629.
  2. bData are presented as the mean ± standard deviation. Experiments were performed three times.
  3. cRelative to the culture containing only GlcNAc.
+788 ± 1114 ± 1014
−309 ± 2116 ± 313
−4110 ± 27 ± 10.7

The sequence of the 5′-end of the 130 nt RNA was determined by 5′-RACE. Primers covering both the 130 nt and 102 nt RNAs amplified two differently sized bands (Fig. 6D). Sequencing analysis showed that the upper and lower bands were amplified from the 130 nt and 102 nt RNAs respectively. Consistent with our previous result (Yamamoto et al., 2011), the 5′-terminal nt of the 102 nt RNA was ‘A’ (Fig. 6A). The 5′-terminal nt of the 130 nt RNA was ‘G’, which is located 6 nt downstream of the −10 sequence (Fig. 6A). These results indicate that the G nucleotide is the transcription start site (+1) of the tfoR promoter and that a 132 nt transcript is produced.

To determine whether the 102 nt RNA is processed from the full-length RNA or is a primary transcript initiated from an as yet unidentified promoter, we examined the 5′-phosphorylation state of the 102 nt RNA. In bacteria, cellular RNAs include primary transcripts with a 5′-triphosphate group, and processed or degraded RNAs that contain either a 5′-monophosphate (5′-P) or 5′-OH group. After total RNA was digested with a 5′-P-dependent exonuclease (5′-P-exo) that specifically degrades RNA containing a 5′-P group (Celesnik et al., 2008), the tfoR RNAs were detected by Northern blotting with Probe N1. 5′-P-exo selectively degraded the 102 nt RNA but not the 132 nt RNA (Fig. 6E), indicating that the former contains a 5′-P. As a positive control, total RNA was monophosphorylated using tobacco acid pyrophosphatase (TAP) to render it completely susceptible to degradation. When treated with TAP, the 132 nt RNA became sensitive to 5′-P-exo (Fig. 6E), suggesting that the 132 nt RNA is the primary transcript containing a 5′-triphosphate group. In addition, a negative control experiment in which total RNA was treated with alkaline phosphatase (AP) to convert 5′-P to 5′-OH was also performed. When treated with AP, the 102 nt RNA was protected from degradation by 5′-P-exo (Fig. 6E). These results indicate that the 102 nt RNA is generated by post-transcriptional processing of the 132 nt primary transcript.

A purified TfoS fragment containing the HTH_AraC domain binds to and activates transcription from the tfoR promoter

The full-length and truncated TfoS proteins were highly insoluble when overproduced in E. coli (data not shown); therefore, our initial attempt to purify these proteins was unsuccessful. As an alternative, we were able to successfully purify a C-terminal fragment containing the HTH_18 domain of TfoS (residues 871–1121) fused to maltose-binding protein (MBP) (MBP::TfoSC) (Fig. S5). Transcription of tfoR was fully activated by the MBP::TfoSC protein in vivo (Table 3); therefore, this fusion protein was used for in vitro analysis of transcription from the tfoR promoter. For this analysis, plasmid pRL-SYtfoR carrying the promoter and transcribed sequence of tfoR was used as the template. If transcription occurred from the tfoR promoter, a 132 nt RNA should have been produced from this plasmid (Fig. 7A). After in vitro transcription with E. coli σ70-RNA polymerase and MBP::TfoSc, the production of tfoR was detected by Northern blotting. RNA I, which was also transcribed from the vector sequence (100 nt RNA), was detected as a loading control. In the absence of MBP::TfoSc, RNA I was produced but tfoR was not (Fig. 7B). In the presence of MBP::TfoSc, the amount of tfoR RNA increased in a concentration-dependent manner, whereas that of RNA I was consistent at all concentrations of MBP::TfoSc tested (Fig. 7B). By contrast, purified MBP did not stimulate the production of tfoR RNA (Fig. 7B). These results indicate that MBP::TfoSc specifically activates transcription from the tfoR promoter in vitro.

Figure 7.

The effect of TfoS on transcription from the tfoR promoter in vitro.

A. The template plasmid pRL-SYtfoR1 used for the in vitro transcription analysis. If the tfoR promoter is active, a 132 nt RNA is expected. A 100 nt RNA I transcript should also be produced from its respective promoter.

B. Northern blot analysis of tfoR transcribed in vitro. The pRL-SYtfoR1 (10 nM) vector was transcribed using E. coli RNA polymerase holoenzyme and various concentrations of MBP::TfoSc (0–1000 nM) or MBP (0 or 1000 nM) in the presence of 4NTP. The synthesized RNAs were separated on a 6% polyacrylamide gel containing 6 M urea, transferred onto a nylon membrane, and detected by Northern blotting with Probe N1 (Fig. 6). As a control, RNA I (100 nt) was visualized by Northern blotting with Probe N4.

Next, a gel mobility shift assay was performed to determine the ability of the MBP::TfoSc protein to bind to a 400 bp DNA fragment containing the upstream region of tfoR (Probe G1; −333 to +67, relative to the start of transcription) (Fig. 8A). Probe G2, which contained a 400 bp region located immediately downstream of Probe G1 (+68 to +467), was also used as a control. MBP::TfoSc shifted the mobility of Probe G1 in a concentration-dependent manner (Fig. 8B). By contrast, Probe G2 was not shifted even at a 250-fold molar excess of MBP::TfoSc to probe (Fig 8B). Furthermore, a 250-fold molar excess of MBP was unable to shift either Probe G1 or G2 (Fig. 8B). These data suggest that MBP::TfoSc specifically binds to the upstream region of tfoR.

Figure 8.

Gel shift assay using tfoR DNA and TfoS protein.

A. The location and length of Probes G1 and G2 relative to the transcription start site (+1) of tfoR. Both probes were 3′-end-labelled with DIG.

B. Gel mobility shift analysis of the specific binding of MBP::TfoSc to the upstream region of tfoR. Each reaction mixture contained 1 nM labelled DNA. After incubation of the DNA with the indicated concentration of MBP::TfoSc or MBP, the DNA-protein mixtures were separated on a 5–20% native polyacrylamide gel and transferred onto a nylon membrane. The labelled DNA was detected with an anti-DIG antibody. The positions of free and bound DNAs are indicated by arrows.

Characterization of the TfoS binding sites in tfoR

DNase I footprinting was used to identify the TfoS binding site(s) in the upstream region of tfoR. Several regions of the non-template and template strands of Probe G1 were protected by MBP::TfoSc (Fig. 9A). Direct repeat units composed of the heptamer sequence TGTCGTT (DR1: positions −70 to −64 and DR2: positions −43 to −36) separated by a 20 bp region were identified (Fig. 9B). The sequences of DR1 and DR2 are highly conserved in Vibrio spp. (Fig. S4). Members of the AraC family of proteins usually bind to the site overlapping the −35 region and additional upstream sites that are often organized as direct repeats (Gallegos et al., 1997). To determine their involvement in TfoS binding, point mutations of DR1 and DR2 were introduced into Probe G1 (TGTCGTT→TTTTTTT) to yield Probes G3 and G4 respectively. In a gel mobility shift assay, the effect of MBP::TfoSc on the progression of these mutant probes through the gel was markedly lower than its effect on the wild type Probe G1 (Fig. 10). To analyse transcriptional activation by TfoS in vivo, tfoR::lacZ fusions containing mutations in these repeat sequences were constructed on the E. coli chromosome. When MBP::TfoSc was expressed, the DR1 mutation reduced transcription to less than 50% of that observed for wild-type cells, and the DR2 mutation resulted in a complete loss of transcription (Table 3). These results suggest that the direct repeat units in tfoR, termed the ‘TfoS box’, are important for TfoS binding and regulation.

Figure 9.

DNase I footprinting analysis of TfoS binding site(s) in the tfoR upstream region.

A. Results of the DNase I footprinting experiment. Probe G1 was 5′-end-labelled with 32P on either the non-template or the template strand. Each reaction mixture contained 1 nM labelled DNA and 0, 100, or 250 nM MBP::TfoSc. Sequencing ladders (G, A, T, and C) were generated by cycle sequencing using the labelled primers for probe preparation. The sequences protected from DNase I digestion are indicated on the right side of each gel image.

B. Summary of the DNase I footprinting analysis. Protected regions and nucleotide are indicated by bold letters and asterisks respectively. The direct repeat units DR1 and DR2 are highlighted in grey. The boxed regions indicate the tfoR promoter (−10) and the transcription start site (+1).

Figure 10.

Gel shift assay showing the importance of DR1 and DR2 for binding of TfoS to the tfoR upstream region. Probes G1 (wild type), G3 (DR1: TGTCGTT→TTTTTTT), and G4 (DR2: TGTCGTT→TTTTTTT) were 3′-end-labelled with DIG and 1 nM of each probe was incubated with or without 250 nM MBP::TfoSc. The labelled probes were detected as described in Fig. 8. The positions of free and bound DNAs are indicated by arrows.


TCSs are the primary mechanism employed by bacteria to reprogramme gene expression in response to environmental changes. In the classical TCS paradigm, an HK specifically phosphorylates a cognate RR, and the genes encoding these paired proteins are chromosomally linked. However, ongoing accumulation of bacterial genomic information is uncovering the potential for an abundance of ‘orphan’ TCS genes in a single bacterium (Barakat et al., 2011). Like classical TCS protein pairs, OTCS proteins regulate a wide variety of cellular functions, including motility, biofilm formation, transport and synthesis of small substances, antibiotic resistance, and virulence (Hutchings et al., 2004; Ulijasz et al., 2004; Gueriri et al., 2008; Bell et al., 2010; McLaughlin et al., 2012). Therefore, it is important to understand the mechanisms by which OTCS proteins participate in bacterial signal transduction pathways. Although a number of OTCS proteins pair with specific partners and act as phosphoryl donors or receivers to regulate gene expression, others lack prototypical partners and operate without the use of classical phosphorelay reactions (Raghavan and Groisman, 2010). Several orphan RRs do not rely on phosphorylation to be modulated (Schar et al., 2005; Ruiz et al., 2008; Wang et al., 2009). An unusual example of such RRs is the Streptomyces venezuelae protein JadR1, which is regulated by direct interaction with the antibiotic jadomycin B (Wang et al., 2009). In addition, a recent study in Pseudomonas aeruginosa unveiled a new signalling pathway, in which the orphan HK RetS directly regulates the activity of the HK GacS in a phosphorylation-independent manner (Goodman et al., 2009). Research into non-canonical mechanisms used by TCSs would expand our current level of understanding of bacterial signal transduction.

Natural competence in V. cholerae is induced by chitin, the major component of the exoskeleton of the aquatic reservoirs. Chitin-induced natural competence is governed by the orphan HK ChiS. In the presence of chitin, ChiS stimulates production of TfoR, the small RNA activator of the competence regulon. This process is the key to initiate the V. cholerae developmental programme driving HGT through transformation. The present study examined the mechanism of transcriptional control of tfoR and the induction of natural competence in V. cholerae by ChiS. The central finding presented here is the identification of a non-TCS transmembrane regulator termed TfoS, which directly activates transcription of tfoR in a ChiS-dependent manner. On the basis of the results presented here and those reported in previous studies (Li and Roseman, 2004), we propose the following revised model for the initial stage of the chitin-signalling cascade that leads to the development of competence. In the absence of (GlcNAc)2, the activity of ChiS is inhibited by the chitin oligosaccharide-binding protein CBP and, thus, TfoS is inactive. (GlcNAc)2 relieves the inhibition of ChiS by competitively binding to CBP, and then activated ChiS provides a signal that stimulates TfoS, resulting in the initiation of transcription of tfoR by RNA polymerase (Fig. 11). The many sequenced genomes of Vibrio spp. encode TfoS and TfoR (Yamamoto et al., 2011; Fig. 1 and Fig. S1), as well as CBP and ChiS (Li and Roseman, 2004). This suggests the potential for a conserved mechanism of chitin-evoked competence regulation across this genus.

Figure 11.

Model of the initial stage of chitin signal transduction in V. cholerae. In the absence of (GlcNAc)2, the activity of ChiS is inhibited by CBP and thus TfoS is inactive. (GlcNAc)2 counteracts the inhibition of ChiS by binding to CBP. In turn, activated ChiS provides an as yet unidentified signal that stimulates TfoS, which binds to the TfoS box in the tfoR promoter and activates transcription by RNA polymerase (RNAP). Conversely, transcription of chb is regulated by the hypothetical regulator ChiR. The conserved phosphorelay residues of ChiS are not essential for activation of the TfoS or ChiR pathways. Furthermore, the possibility that CBP interacts with TfoS cannot be ruled out.

TfoS is an unusual AraC family protein in terms of its size (1121 residues versus the typical 250–300 residues) (Gallegos et al., 1997), subcellular localization, and its probable mode of regulation. A large portion of the TfoS protein is oriented into the periplasm, while the remainder is exposed to the cytoplasm (Fig. 2B). A cytoplasmic fragment (residues 871–1121) that contains HTH_18 was essential and sufficient for activating transcription both in vivo and in vitro (Table 3 and Fig. 7). In addition, this region specifically bound to the direct repeat units termed the ‘TfoS box’ (TGTCGTT-N20-TGTCGTT), which overlaps the −35 position of the tfoR promoter (Figs 8, 9 and 10). Searches of the V. cholerae genome sequence failed to identify additional candidate TfoS boxes, suggesting that the tfoR gene might be the only target for TfoS regulation. The TfoS box is structurally analogous to the binding sites of many AraC homologues. Generally, members of the AraC family of proteins exist in the cytoplasm and form a dimer as their default state (Gallegos et al., 1997). Ligand binding induces a conformational change of the DNA-binding domain, and each monomer binds to a half site of a direct repeat sequence located at the −35 region of the promoter, where it finally contacts RNA polymerase (Gallegos et al., 1997). Similar mechanisms of promoter binding and activation could be utilized by TfoS.

The mechanism by which TfoS responds to the signal from ChiS and regulates transcription requires further study. Although deletion of the periplasmic sensor domain (residues 9–870) activated transcription independent of ChiS, the intact protein absolutely required ChiS for activation (Table 3). This suggests that the sensor domain plays a dual role in the inhibition of DNA binding and in the sensing of ChiS state. The inhibitory effect of the TfoS periplasmic sensor domain is reminiscent of the N-terminal arm of AraC, which intramolecularly binds to the C-terminal DNA-binding domain to render it inactive in the absence of the ligand arabinose (Saviola et al., 1998). If the transmembrane protein TfoS has a self-inhibitory activity similar to AraC, it would be located within the cytoplasmic region (residues 759–870). However, we cannot exclude the additional possibility that TfoS drives transmembrane signal transduction in a similar manner to that utilized by the HTCS protein BT_4663, which controls heparin and heparan sulphate acquisition and degradation in B. thetaiotaomicron (Lowe et al., 2012). In the absence of signal, the cytoplasmic output domain of BT_4663 is maintained in an inactive form, while in the presence of signal, a conformational change in the external domain is transmitted across the membrane and this change transforms the output domain into an active state. If this assumption is correct for TfoS, mutations responsible for defective transmembrane signalling, including constitutively active mutations, would be mapped to the extracytoplasmic region. More extensive studies using diverse protein variants are required to address these questions.

One of the important questions raised by this study is the mechanism by which ChiS regulates TfoS. Many TCS HKs show both kinase and phosphatase activities. In EnvZ, a classical HK, a conserved phosphorylation site (H243) in the HisKA domain is essential for both activities (Zhu et al., 2000). However, in some hybrid HKs, the conserved histidine residue in HisKA is required for kinase activity but not for phosphatase activity. By contrast, the conserved aspartate residue in the REC domain is required for both activities (Uhl and Miller, 1996; Georgellis et al., 1998; Freeman et al., 2000; Clarke et al., 2002). ChiS was unexpectedly functional in V. cholerae and E. coli after mutation of the highly conserved H469 and D772 residues in the HisKA and REC domains respectively (Tables 1 and 3). This result suggests that ChiS may regulate the activity of TfoS using mechanisms distinct from classical phosphorelay reactions. It is possible that ChiS-mediated regulation of TfoS is achieved by a direct protein-protein interaction; however, positive interactions between these proteins were not detected by our bacterial two-hybrid assays (data not shown). It is also possible that additional factor(s) conserved in V. cholerae and E. coli may mediate the interaction between ChiS and TfoS.

We initially wondered whether TfoS itself could act as a sensor of (GlcNAc)2; however, exposure of the E. coli strain SYECREP1S expressing TfoS to (GlcNAc)2 did not induce transcription of tfoR (data not shown), despite the fact that this sugar can be taken up and catabolized by E. coli (Keyhani and Roseman, 1997). TfoS has a periplasmic sensor domain that is structurally related to HTCS proteins (Figs S2 and S3), some of which bind to specific sugar ligands (Sonnenburg et al., 2006; 2010; Lowe et al., 2012). The HTCS family has been widely expanded in the phylum Bacteroidetes. For example, the prominent human gut symbiont B. thetaiotaomicron has 32 HTCS genes that each encode a protein with a large N-terminal periplasmic sensor domain (typically 750–800 residues) flanked by two transmembrane domains, and a C-terminal cytoplasmic hybrid domain consisting of TCS signature domains (HisKA, HATPase_c, and REC) and an output domain (HTH_18) (Sonnenburg et al., 2006). The sensor domains of most HTCS proteins contain 14 repetitive β-strand units that adopt a double β-propeller fold (Menke et al., 2010). In some cases, this structure is responsible for ligand binding (Sonnenburg et al., 2006; Lowe et al., 2012). Importantly, the B. thetaiotaomicron HTCS protein BT3172 is thought to contain periplasmically and cytoplasmically oriented sensor domains that interact with α-mannosides and glycolytic enzymes, respectively, and couple glycan sensing to carbohydrate metabolism (Sonnenburg et al., 2006). Therefore, it will be interesting to determine whether proteins or small molecules associated with chitin catabolism bind to TfoS, or whether these ligands modulate the communication between ChiS and TfoS.

Chitin-induced transcriptional activation of chb and tfoR was dependent on ChiS (Table 1). This indicates that ChiS co-regulates chitin catabolism and natural competence. To utilize chitin as nutrient source, V. cholerae extracellularly degrades it into oligosaccharides, including (GlcNAc)2, and then takes up them (Li and Roseman, 2004). (GlcNAc)2 is sensed by the CBP-ChiS system located in the periplasmic space to induce chb and tfoR gene expression (Fig. 11). Soon after, the catabolic functions encoded on chb reduce periplasmic (GlcNAc)2 level, while TfoR promotes not only DNA uptake but also probably extraclellular chitin degradation and uptake by activating the TfoX-regulated genes (competence genes, chiA-1, chiA-2, and chiP), potentially generating ‘combined’ negative-positive feedback loops. In Bacillus subtillis, such architecture of gene regulatory circuit is associated to generating heterogeneity of the competent state at the individual cell level (Dubnau and Losick, 2006; Süel et al., 2006). On chitin surfaces, such as crab shell fragments, V. cholerae cells also show a heterogeneous competence gene expression pattern (Lo Scrudato and Blokesch, 2012). Therefore, co-regulation of chb and tfoR by ChiS might have some effect on the competence development of the individual cells in the V. cholerae population. Our results indicate that unlike tfoR, the activation of chb does not require TfoS (Table 1). Transcription of chb is controlled by an as yet unidentified regulator called ‘ChiR’ (Li and Roseman, 2004). We propose that the signalling pathway initiated by active ChiS is branched at the level of TfoS and ChiR stimulation to induce transcription of tfoR and chb respectively (Fig. 11). This model and the results presented here suggest that DNA uptake and chitin catabolism may be differentially modulated by chitin availability. Similar to tfoR, activation of chb was independent of the conserved phosphorelay residues of ChiS (Table 1). At present, we do not know whether non-canonical signalling is the default mode of action of ChiS. It will be necessary to elucidate whether ChiS is a bona fide HK, or whether it uses its intrinsic enzymatic activities to regulate expression of the regulon genes.

A comprehensive analysis of 145 bacterial genome sequences revealed that more than 400 (3%) of approximately 14 300 non-TCS DNA-binding proteins are predicted to be localized to the inner membrane (Ulrich et al., 2005). Although only a limited number of transmembrane DNA-binding proteins have been studied to date, all of the examined proteins play key roles in regulating clinically relevant functions of pathogenic bacteria, including virulence, antibiotic resistance, and the acid stress response (DiRita and Mekalanos, 1991; Beck et al., 2004; Tetsch et al., 2008; Gebhard et al., 2009). There is currently no evidence that TfoS (Fig. 4) or other regulators are proteolytically cleaved to liberate an active DNA-binding fragment like other prokaryotic and eukaryotic transmembrane regulators (Brown et al., 2000). Such non-cleavable regulators would provide an excellent model system to study the physiological rationale for membrane-anchored DNA-binding and transcription in growing cells. To our knowledge, TfoS is the first report of a transmembrane regulator that directly controls transcription of an sRNA gene. Several leading studies have demonstrated that the RNA chaperone Hfq is concentrated in close proximity to the membrane, where it post-transcriptionally controls mRNA stability in collaboration with an sRNA (Kawamoto et al., 2005; Diestra et al., 2009). Therefore, it is possible that TfoS directs the transcription machinery to produce TfoR close to the membrane (Fig. 11), thereby promoting localized translation of tfoX mRNA together with the resident Hfq.

Previously characterized transmembrane regulators are controlled by interaction with either functionally related membrane components (DiRita and Mekalanos, 1991; Tetsch et al., 2008) or specific small molecules (Gebhard et al., 2009). By contrast, TfoS function is dependent on the TCS HK ChiS, although it is unclear if direct interactions between these proteins occur. Nevertheless, the non-canonical mode of action of ChiS and the HK RetS (Goodman et al., 2009) indicates the existence of a functional plasticity of bacterial TCS HKs that creates additional signal transduction networks with membrane proteins. Further studies will provide additional insights into the mechanism and significance of ChiS-mediated activation of TfoS.

Experimental procedures

Bacterial strains, plasmids, media, and DNA manipulation

All of the strains and plasmids used in this study are listed in Table S1. The E. coli JM109 strain (Yanisch-Perron et al., 1985) was used for DNA cloning. The bacteria were routinely grown in Luria–Bertani broth (LB) and agar (LA), and SOC medium (2% tryptone, 0.5% yeast extract, 10 mM NaCl, 2.5 mM KCl, 10 mM MgCl2, 10 mM MgSO4, and 20 mM glucose). The following antibiotics were used: 100 μg ml−1 ampicillin (Ap), 12.5 μg ml−1 (for E. coli) or 1 μg ml−1 (for V. cholerae) chloramphenicol (Cm), and 30 μg ml−1 kanamycin. Arabinose, X-gal and IPTG were used at concentrations of 5 mg ml−1, 100 μg ml−1, and 1 mM respectively. PCR amplification of DNA was performed using the GeneAmp® PCR System 9700 (Applied Biosystems) and Herculase® II Fusion Polymerase (Stratagene). DNA sequencing was performed using the BigDye® Terminator v1.1 Cycle Sequencing Kit (Applied Biosystems). PCR products were purified using the High Pure PCR Product Purification Kit (Roche). Customized oligonucleotide primers (Table S2) were purchased from Operon. Chromosomal and plasmid DNAs were extracted using the Wizard® Genomic DNA Purification Kit (Promega) and High pure plasmid isolation kit (Roche) respectively.

Protein analysis

SDS-PAGE and immunoblot analyses of proteins were performed as previously described (Yamamoto et al., 2009).

Chromosomal engineering

Genetically modified E. coli and V. cholerae strains were constructed using the λ Red-FLP recombination system (Cherepanov and Wackernagel, 1995; Datsenko and Wanner, 2000; Uzzau et al., 2001; Yamamoto et al., 2009) or natural transformation induced by shrimp shell (Morita et al., 2013). When point mutations were introduced into the V. cholerae chromosome, markerless gene replacement was achieved using the sacB/sucrose counter selection system (Penfold and Pemberton, 1992; Metcalf et al., 1996; Fullner and Mekalanos, 1999). Recombinants were selected on LA lacking NaCl but containing 20% sucrose. Detailed procedures used to prepare donor DNAs for recombination are described in the Supplementary information (Tables S3, S4).

Plasmid constructions

See the Supplementary information in detail (Table S5).

β-Galactosidase enzyme assay and bacterial culture

β-Galactosidase (LacZ) activity was measured as previously described (Yamamoto et al., 2010), using E. coli and V. cholerae cells assayed in triplicate. The cells were grown under the culture conditions described below. Unless otherwise stated, these conditions were also used for other experiments.

E. coli

Cells were grown in LB at 37°C until they reached an OD600 of approximately 0.8. Cm, Ap, arabinose, and/or IPTG were added to the medium as required.

V. cholerae

Cells were grown in M9 minimal medium containing minimal essential medium vitamin solution (Gibco) and 0.5% lactate (pH 7.6) (termed ‘M9L’ medium), supplemented with 2.5 mM GlcNAc or 2.5 mM (GlcNAc)2 (Seikagaku Biobusiness), until they reached an OD600 of approximately 0.4. Lactate was added to support the growth of the chiS null mutant, which cannot use (GlcNAc)≥2 as its sole carbon source (Li and Roseman, 2004). If necessary, Ap was also added to the medium.

Natural transformation

Natural transformation was performed using the (GlcNAc)2 and shrimp chitin methods described below.

(GlcNAc)2 method

The (GlcNAc)2 method was performed as previously described (Yamamoto et al., 2011), with slight modifications. V. cholerae cells were cultured as described above, and then 10 μg of chromosomal DNA from strain SY1003, which harbours a cat gene in the lacZ gene, was added. After an overnight incubation, the cells were spread onto LA with or without Cm. If necessary, Ap was added to the medium. Transformation efficiency was defined as the number of Cm-resistant colony-forming units (cfus) divided by the total number of cfus.

Shrimp shell method

The shrimp shell method was performed as previously described (Morita et al., 2013), with slight modifications. Cells were cultured in distilled artificial seawater medium containing 0.5% lactate (pH 7.6), termed ‘DASWL’ medium. Chromosomal DNA from SY1003 (10 μg) was used as a donor. The transformation efficiency was calculated as described above. If necessary, Ap was added to the medium.

Cell fractionation analysis

Fractionation of V. cholerae cells was performed as previously described (Sandkvist et al., 1999; Wai et al., 2003), with some modifications. SY0636K carrying pMW-SYmtlA::phoA1 was used as the tester strain. After centrifugation of the bacterial culture, the pelleted cells were resuspended in a solution containing 20% sucrose, 20 mM Tris-HCl (pH 8.0), and 1 mM EDTA, and then incubated with 5000 U ml−1 polymyxin B for 15 min on ice to release the periplasmic content. The sample was centrifuged at 15 300 g for 15 min at 4°C, and the supernatant containing periplasmic proteins was transferred to a new tube. The cytoplasmic, outer membrane, and inner membrane fractions were then separated from the pellet as previously described (Wai et al., 2003). Proteins in the fractioned samples were separated by SDS-PAGE and the TfoS::3×FLAG protein was detected by immunoblotting with a monoclonal anti-FLAG M2-horseradish peroxidase antibody (Sigma). The control proteins OmpA, Bla, MtlA::PhoA, and RpoA were also analysed by immunoblotting with polyclonal anti-OmpA (Song et al., 2008), anti-AmpC (Chemicon international), anti-AP E. coli (Acris GmbH), and anti-RpoA antisera (Santa Cruz Biotechnology) respectively.

Alkaline phosphatase assay

AP (PhoA) activity was determined as previously described (Manoil, 1991). Assays were performed in triplicate.

Analysis of intracellular levels of TfoS

Proteins extracted from SY0636K and its derivative strains that harboured a tfoS::3×FLAG gene were separated on a 5–20% SDS-polyacrylamide gel and then transferred onto a PVDF membrane. The TfoS::3×FLAG protein was detected by immunoblotting with a monoclonal anti-FLAG M2-horseradish peroxidase antibody. RpoA was also detected using anti-RpoA antiserum.

Preparation of digoxigenin-labelled DNA probe for Northern blotting

Oligonucleotide DNA probes complementary to the specific RNAs were 3′-end-labelled using a digoxigenin (DIG) oligonucleotide 3′-end-labelling kit (Roche), according to the manufacturer's instructions. The oligonucleotides used were as follows: tfoR northern probe-D for Probe N1; tfoR northern probe-U for Probe N2; N16961 5S rRNA northern probe for Probe N3; and RNAI northern probe for Probe N4.

Northern blotting

Northern blotting was performed by the previously described procedure (Yamamoto et al., 2011) with slight modifications. Total RNA (7.5 μg) were separated on a polyacrylamide gel containing 6 M urea, transferred onto a nylon membrane, and hybridized with 0.5 μg DIG-labelled probe. Hybridized probe was subjected to a chemiluminescent detection using anti-DIG-AP Fab fragment (Roche) and Disodium 2-chloro-5-(methoxyspiro {1,2-dioxetane-3,2′-(5′-chloro)tricyclo [,7]decan}-4-yl)phenyl phosphate (Roche).

Determination of the 5′-ends of tfoR RNAs

The 5′-ends of tfoR RNAs were determined using a 5′ RACE system for rapid amplification of cDNA ends (Invitrogen), according to the manufacturer's instructions. Total RNA was isolated from the V060002 strain and cDNA was reverse-transcribed from 6 μg of the total RNA using the tfoR-GSP1 primer. A homopolymeric tail was then added to the 3′-end of the cDNA using a terminal deoxynucleotidyl transferase and dCTP. Next, PCR amplification was performed using ExTaq DNA polymerase (Takara), a nested primer (tfoR-GSP2) that annealed to a site located within the cDNA molecule, and a novel deoxyinosine-containing anchor primer (AAP). The primary PCR product was re-amplified using the nested primers tfoR-GSP3 and AUAP. The two major products obtained (Fig. 6D) were gel-purified and cloned into pGEM-T. Ten clones from independent transformants were sequenced using the T7P and SP6P primers.

Analysis of the 5′-phosphorylation states of tfoR RNAs

The 5′-phosphorylation states of the tfoR RNAs were analysed as previously described (Celesnik et al., 2008). Total RNA (7.5 μg) used for the 5′ RACE analysis was treated with or without TAP (Epicentre) and APex heat-labile AP (Epicentre) for 3 h at 37°C, and then the reaction was terminated by the addition of phenol-chloroform. After centrifugation, the aqueous fraction containing RNA was precipitated with ethanol and dissolved in H2O. 5′-modified or unmodified RNA was digested with Terminator 5′-phosphate-dependent exonuclease (Epicentre) for 3 h at 30°C, and then detected by Northern blotting using Probe N1.

Overproduction and purification of MBP and MBP::TfoSc

The E. coli BL21(DE3) strain carrying pMAL-c2 or pMAL-SYtfoS3 was grown at 30°C with shaking in 100 ml of LB containing Ap. When the cell growth reached an OD600 of 0.4, IPTG was added to a final concentration of 0.5 mM. After further cultivation for 2 h, the cells were harvested by centrifugation and resuspended in 10 ml of wash buffer comprising 20 mM Tris-HCl (pH 7.4), 200 mM NaCl, and 1 mM EDTA. After disruption of the cells by sonication, the sample was centrifuged and the resulting supernatant was mixed with 1.5 ml of amylose resin (New England Biolabs). After the mixture was shaken gently at 4°C for 2 h, the amylose resin was collected by centrifugation, washed four times with 10 ml of wash buffer, and resuspended in 300 μl of elution buffer comprising 20 mM Tris-HCl (pH 7.4), 200 mM NaCl, 1 mM EDTA, and 10 mM maltose. After centrifugation, the supernatant fractions were analysed by SDS-PAGE and those containing MBP (44 kDa) and MBP::TfoSc (72.5 kDa) were pooled (Fig. S5). The pooled fractions were dialysed twice against 500 ml of 20 mM Tris-HCl (pH 7.5) prior to use.

In vitro transcription

The in vitro transcription reaction was performed with a 15 μl mixture containing 10 nM template DNA (pRL-SYtfoR1), 1 U of E. coli RNA polymerase holoenzyme (Epicentre), 40 mM Tris-HCl, 150 mM KCl, 10 mM MgCl2, 1 mM dithiothreitol, 50 U of RNase inhibitor (Takara), and various concentrations (0–1000 nM) of MBP::TfoSc or MBP. After pre-incubation for 30 min at 37°C, the transcription reaction was initiated by the addition of 4NTP at a final concentration of 0.1 mM. After incubation for 30 min at 37°C, the reaction was terminated by incubation for 3 min at 90°C. Synthesized transcripts were detected by Northern blotting with Probes N1 and N4 (Table S2).

Gel shift analysis

To obtain Probes G1 and G2, 400 bp fragments of the tfoR locus were PCR amplified from V060002 chromosomal DNA using the tfoR-GSf1/tfoR-GSr1 and tfoR-GSf2/tfoR-GSr2 primer pairs respectively. Probes G3 and G4 were amplified by two-step PCR. The first round of PCR was performed using the tfoR-GSf1/tfoR-GSmtr1 and tfoR-GSmtf1/tfoR-GSr1 primer pairs for Probe G3 and the tfoR-GSf1/tfoR-GSmtr2 and tfoR-GSmtf2/tfoR-GSr1 primer pairs for Probe G4. The second round of PCR was performed using the first round products as templates and the tfoR-GSf1 and tfoR-GSr1 primers. The amplified probes were 3′-end-labelled with DIG as previously described above. The binding reaction mixture (10 μl) contained 1 nM labelled DNA and various concentrations (0–250 nM) of MBP::TfoSc or MBP in 40 mM Tris-HCl, 150 mM KCl, 10 mM MgCl2, 1 mM dithiothreitol, and 1 μg ml−1 poly(dI-dC). The reaction mixture was incubated for 10 min at 25°C and then electrophoresed on a native 5–20% polyacrylamide gel in Tris-glycine buffer (50 mM Tris base, 192 mM glycine, 50 mM EDTA, pH 8.3). After transfer onto a nylon membrane, DIG-labelled DNA was visualized as described above.

DNase I footprinting

DNase I footprinting was performed using 32P-labelled strand-specific probes. Probe G1 that was 5′-end-labelled on the non-template strand was PCR amplified using 32P-labelled primer tfoR-GSf1 and non-labelled primer tfoR-GSr1. The template strand-labelled probe was generated using non-labelled primer tfoR-GSf1 and 32P-labelled primer tfoR-GSr1. The binding reaction mixture (100 μl) contained 1 nM labelled DNA and 0, 100, or 250 nM MBP::TfoSc in the buffer used for gel shift assays. After incubation for 10 min at 25°C, 100 μl of a solution comprising 5 mM MgCl2, 5 mM CaCl2, and 180 ng of DNase I (Sigma) was added, and the mixture was incubated for 2 min at 25°C. The reaction was terminated by the addition of 100 μl of phenol-chloroform. After centrifugation at 4°C, the aqueous fraction containing DNA was precipitated with ethanol and separated on a 5% polyacrylamide gel containing 6 M urea. Labelled DNA was detected by autoradiography. Sequencing ladders were generated using a T7 sequencing kit (USB Corporation), the 32P-labelled primers described above, and the template pRL-SYtfoR1.


PSI-BLAST (Altschul et al., 1997) available via NCBI ( was used to identify sequences homologous to TfoS.


Membrane topology prediction of TfoS was performed using TMHMM (Krogh et al., 2001) available via the Center for Biological Sequence Analysis at

Multiple alignment

The alignment of multiple TCS HKs was performed using GENETIX-MAC version 12.0.9 software (Genetyx Corporation).


The periplasmic sensor domain of TfoS (residues 9–870) was analysed using HHprep (Soding et al., 2006) available via the Max-Planck Institute for Developmental Biology at Further details are provided in the supplementary information.


This work was supported by the Japan Society for the Promotion of Science (JSPS KAKENHI, no. 34102400), the Ministry of Health, Labour and Welfare of Japan (H23-Shinko-Shitei-020), and the Ministry of the Environment of Japan (Global Environment Research Fund, S-8). We thank the National Bioresource Project-E. coli at the National Institute of Genetics (Japan) for providing the pJP5608 and pWM91 plasmids.