Abstract
Quorum sensing, a cell-to-cell communication process, has been found to regulate a plethora of cellular events in Vibrio cholerae, including pathogenesis (Zhu et al., 2002). Recent work has established the complex architecture of the quorum sensing network of this bacterium, where multiple circuits work in parallel in response to population density. Genetic analysis reveals that sensory information from two parallel quorum sensing circuits (AI-2/LuxPQ and CAI-1/CqsS) converges into a central molecule, LuxO, through a shared phosphorelay protein, LuxU (Miller et al., 2002). At low cell density, LuxU activates LuxO through phosphorylation. This phosphorylated LuxO in concert with σ54 dictates the expression of cascades of small RNAs, which further cause the destabilization of hapR mRNA transcript. HapR is a positive regulator of hapA, which encodes haemagglutinin (HA)/protease. At high cell density, dephosphorylation of LuxO impairs its ability to activate the expression of small RNAs, which in turn promotes HapR expression and causes protease production (Lenz et al., 2004). In other words, the appearance of HA/protease, one of the developmentally regulated virulence factors, in the culture supernatant indicates the functional impairment of LuxO at high cell density.
In a recent effort, Bassler and co-workers have identified two additional systems of the quorum sensing network of V. cholerae (Lenz et al., 2005; Lenz & Bassler 2007). One such system is known as VarS/A-CsrA/BCD, which is constituted of a sensor kinase (VarS), a response regulator (VarA), an RNA-binding global regulatory protein (CsrA) and three small RNAs (CsrBCD). These small RNAs bind and antagonize the activity of CsrA. The functioning of the VarS/A two-component system depends on the growth phase. At low cell density, the two-component sensor kinase VarS remains inactive and does not phosphorylate the response regulator VarA. Unphosphorylated VarA is unable to turn on the expression of genes encoding CsrBCD small RNAs. As a consequence, CsrA is free to act on its cellular targets. In the quorum sensing network of V. cholerae, CsrA delivers sensory information to LuxO without involving LuxU (Lenz et al., 2005). It has been proposed that CsrA maintains the active form of LuxO by a hitherto unknown mechanism. At high cell density, VarS becomes active and phophorylates VarA. Phosphorylated VarA stimulates the expression of the genes that encode CsrBCD small RNAs. Subsequently, these small RNAs interfere with the activity of CsrA by titrating it away from its targets. As a result, CsrA can no longer promote the functioning of LuxO, which favours HapR production and HapR-mediated gene expression.
The key element of the fourth system is Fis, a small nucleoid protein. Recent studies have evidenced its role as a global regulator of exponential and stationary phase gene expression in many Gram-negative bacteria (Ishihama 1999; Kelly et al., 2004). In Escherichia coli, Fis was found to modulate transcription of genes encoding stable RNAs. Fis controls gene expression by altering the conformation of DNA through bending, as well as making contact with the α-subunit of RNA polymerase (Dorman & Deighan, 2003). Upon binding to a degenerate consensus sequence, it introduces a bend of between 4 ° and 9 ° (Hengen et al., 1997). Recently, Fis has been shown to control the expression of virulence genes in pathogenic strains of E. coli, Shigella flexneri and Salmonella typhimurium. In addition to virulence factors, Fis also coordinates the expression of a number of housekeeping genes that are involved in metabolism, transport and flagellar biosynthesis in Salmonella typhimurium (Kelly et al., 2004). Importantly, a large member of the Fis regulon is required for survival in the gut, which further demonstrates the contribution of Fis in the host–pathogen interaction of Salmonella typhimurium. The expression pattern of Fis is contingent on the growth phase: the level of this regulatory protein reaches a peak in the exponential phase and becomes undetectable during stationary phase under aerated condition (Walker et al., 2004). It is interesting to note that expression of Fis under non-aerated conditions continues in the stationary phase in the case of some E. coli strains and Salmonella typhimurium (O'Cróinin & Dorman 2007). It has also been demonstrated that alternate stress response sigma factor RpoS represses Fis expression and a reduced level of RpoS facilitates Fis expression under non-aerated condition (O'Cróinin & Dorman 2007). This observation is further expounded by a natural rpoS mutant of Salmonella typhimurium strain LT2, where the level of Fis is significantly higher under non-aerated conditions than that observed in strain SL1344. In the case of V. cholerae, expression of fis is also regulated by growth phase, as described in other bacteria (Lenz & Bassler, 2007).
In the quorum sensing system of V. cholerae, Fis works in parallel with LuxO to stimulate the expression of qrr1–4 small RNAs, thus controlling the production of HapR in a growth-dependent manner (Lenz & Bassler, 2007). Interestingly, Fis exhibits a gradation of affinity to the promoter regions of qrr small RNAs and promotes maximum production of qrr4 small RNA over others. In essence, Fis is a critical component of the quorum sensing network of V. cholerae and maximal transcription of genes encoding qrr small RNAs requires the involvement of LuxO-phosphate, σ54 and Fis. At present, the quorum sensing network of V. cholerae reveals an intricate circuit that involves an interplay of two autoinducer systems, one growth-phase regulated VarS/A–CsrA/BCD unit and a small nucleoid protein Fis to control population-dependent gene expression. It is also interesting to note that neither of the two global regulators CsrA and Fis control the expression of LuxO (Lenz et al., 2005; Lenz & Bassler, 2007).
Being a central molecule of quorum sensing circuit, a great deal of work has been dedicated to understanding the function of LuxO. Sequence analysis indicates a conserved aspartate residue (Asp-47) that is critical for its signalling function (Freeman & Bassler, 1999). Signal transduction is proposed to occur via a series of phosphorylation and dephosphorylation reactions. Inactivation of LuxO function occurs by dephosphorylation at Asp-47. Thus far, a number of variant LuxO molecules have been characterized from various strains of V. cholerae. One category of LuxO molecules harbour a specific mutation at a particular residue(s) that lock the protein in a constitutively active form mimicking phosphorylated LuxO, while a diverse set of mutations in the same residue or different residues convert the protein into an inactivated form simulating dephosphorylated LuxO. For example, mutant proteins mimicking unphosphorylated LuxO (D4K, D47A, D47N and K97A) inactivated its activator function and caused constitutive luminescence in Vibrio harveyi. In contrast, a mutant protein mimicking phospho-LuxO (D47E, F94W) resulted in a constitutively active form, which repressed luminescence in V. harveyi. Because of the constitutive nature of its activity, the mutant form D47E is referred to as constitutively active LuxO (Con-LuxO) (Freeman & Bassler, 1999). Similarly, a LuxOL104Q variant that was isolated from V. cholerae El Tor strain E7946 also acts constitutively, and thus represses protease production and enhances biofilm formation by this strain (Vance et al., 2003). It is noteworthy to mention that these variant LuxO molecules are not functionally equivalent. For example, the repressor phenotype of LuxO D47E is more pronounced than that of LuxO F94W. Similarly, ΔRR LuxO protein has a stronger repressor phenotype than does the LuxO D47E protein (Freeman & Bassler, 1999).
Previously, we identified a variant LuxO molecule from a non-O1, non-O139 V. cholerae PL91 (henceforth designated LuxOPL91) whose function remains unaffected in the absence of LuxU, thus conferring the protease-negative phenotype exhibited by this strain (Raychaudhuri et al., 2006). In this work, we wanted to evaluate whether LuxO functioning at high cell density in PL91 is due to a gain of function mutation or whether there exists a continuous input from CsrA and Fis to quorum sensing system in PL91, which in turn maintain the activity of LuxOPL91 at high cell density and thus render this strain protease negative.
In pursuit of our interest, csrA and fis single mutants of V. cholerae strain PL91 were generated by disrupting the chromosomal copy with suicidal constructs pSM and pSRF, respectively. For the construction of plasmid pSM, a 147 bp internal fragment of the csrA gene was amplified with oligonucleotides designated CsrA XbaI and CsrA EcoRI (Table 1). The fragment was cloned into similarly digested suicide vector pGP704 at XbaI and EcoRI and verified by sequencing. pSM was maintained in E. coli SM10 : λ pir as a host. In a similar fashion to the pSM construction, a 174 bp internal fragment of fis gene was amplified with oligonucleotides Fis XbaI and Fis EcoRI (Table 1) and cloned into similarly digested suicide vector pGP704. The resulting recombinant plasmid was designated pSRF and maintained in E. coli SM10 : λ pir as a host. Subsequently, pSM and pSRF were transferred in V. cholerae PL91 by conjugation. csrA and fis mutants of PL91 were screened by streaking onto Luria–Bertani agar plates containing ampicillin (100 µg ml–1) and streptomycin (10 µg ml –1). The csrA and fis mutants of PL91 were designated PL91-SM and PL91-SRF, respectively (Table 1). To further examine the integration of pSM, a PCR was carried out using a long PCR kit (Fermentas) following the manufacturer's protocol with the primer pair 100 bp UP CsrA and 100 bp DN CsrA (Table 2). An expected 4.2 kb band was obtained from the disrupted strain, which was purified and sequenced using an ABI PRISM dye terminator cycle sequencing kit (Perkin Elmer). Further analysis confirmed the disruption of csrA, which occurred at amino acid position 59 (CsrA is constituted of 65 aa) in PL91-SM. A similar approach was used to confirm the integration of pSRF. Towards this end, the genomic DNA of PL91-SRF was amplified by using Phusion polymerase (Finnzyme) with the primer pair SmaI FIS and PstI FIS (Table 2). An expected 4.2 kb band was obtained and sequenced using an ABI PRISM dye terminator cycle sequencing kit. Further analysis confirmed the disruption of fis, which occurred at amino acid position 72 (Fis is constituted of 98 aa) in PL91-SRF. As protease production in the stationary growth phase is one of the parameters used to evaluate the functional perturbation of wild-type LuxO, we wanted to evaluate the performance of LuxO in PL91-SM (PL91ΔcsrA) and PL91-SRF (PL91Δfis) by measuring the protease activity in culture supernatants of PL91-SM and PL91-SRF, along with PL91-C (PL91 carrying vector pKK177-3RI) and PL91-SVM (PL91ΔluxO). All these strains were grown in tryptic soy broth without dextrose (glucose) (TSB-D) containing appropriate antibiotics, with agitation, at 37 °C to stationary phase. The cell-free supernatants harvested from stationary phase cultures were tested for protease production using an azocasein assay as described elsewhere (Benitez et al., 2001; Raychaudhuri et al., 2006). We found that PL91-C, PL91-SM and PL91-SRF remain protease negative in comparison to PL91-SVM (Fig. 1a). These data clearly indicate that the absence of CsrA and Fis could not incapacitate the performance of LuxOPL91, which suppresses protease production in the corresponding mutants of PL91 at high cell density. In essence, our data clearly suggested that LuxOPL91 is a gain of function variant that is capable of maintaining its activity at high cell density in csrA and fis mutants of PL91, thus conferring a protease-negative phenotype in wild-type PL91 and isogenic csrA and fis mutants of PL91. Next, we wanted to examine whether the function of LuxOPL91 is a strain-dependent phenomenon or not. To resolve this issue, overexpression constructs of luxOPL91 and luxON16961 were used to transform into non-O1, non-O139 protease-positive strain of V. cholerae SC134 (a kind gift of Dr Ranjan K. Nandy, National Institute of Cholera and Enteric Diseases, Beliaghata, Kolkata, India). The only difference between LuxOPL91 and LuxON16961 is the absence of 12 aa in LuxOPL91 as described previously (Raychaudhuri et al., 2006). The recombinant strains SC134-PL (carrying luxOPL91 overexpression plasmid pPL) and SC134-NL (carrying luxON16961 overexpression plasmid pNL) were tested for protease production at high cell density under aerated conditions. Our data showed that SC134-NL remains protease positive, whereas SC134-PL becomes protease negative, which further suggested that the function of LuxOPL91 is not restricted to PL91, and can maintain its constitutive activity in strains other than PL91 (Fig. 1b). Taken together, our data confirmed that LuxOPL91 is gain of function variant, and the absence of CsrA and Fis does not perturb its activity. It can also exert its constitutive activity in strains other than PL91.