RelA and SpoT of Gram-negative organisms critically regulate cellular levels of (p)ppGpp. Here, we have dissected the spoT gene function of the cholera pathogen Vibrio cholerae by extensive genetic analysis. Unlike Escherichia coli, V. choleraeΔrelAΔspoT cells accumulated (p)ppGpp upon fatty acid or glucose starvation. The result strongly suggests RelA-SpoT-independent (p)ppGpp synthesis in V. cholerae. By repeated subculturing of a V. choleraeΔrelAΔspoT mutant, a suppressor strain with (p)ppGpp0 phenotype was isolated. Bioinformatics analysis of V. cholerae whole genome sequence allowed identification of a hypothetical gene (VC1224), which codes for a small protein (∼29 kDa) with a (p)ppGpp synthetase domain and the gene is highly conserved in vibrios; hence it has been named relV. Using E. coliΔrelA or ΔrelAΔspoT mutant we showed that relV indeed codes for a novel (p)ppGpp synthetase. Further analysis indicated that relV gene of the suppressor strain carries a point mutation at nucleotide position 676 of its coding region (ΔrelAΔspoT relV676), which seems to be responsible for the (p)ppGpp0 phenotype. Analysis of a V. choleraeΔrelAΔspoTΔrelV triple mutant confirmed that apart from canonical relA and spoT genes, relV is a novel gene in V. cholerae responsible for (p)ppGpp synthesis.
In natural environment bacteria are constantly challenged by a variety of stressful conditions among which nutrient limitation is a critical determinant for their survival and growth. Thus, bacteria have evolved complex strategies to adapt quickly by modulating various cellular functions when faced with diminishing nutrient supplies. The adaptive response to nutritional stress in bacterial cells leads to rapid and complex metabolic adjustments through negative and positive regulations of gene expression, which is widely known as the stringent response. The abrupt global changes associated with the stringent response are mainly triggered by the intracellular accumulation of two small molecules called guanosine 3′-diphosphate 5′-triphosphate and guanosine 3′,5′-bis(diphosphate), collectively termed (p)ppGpp. Elevated intracellular (p)ppGpp level quickly inhibit the transcription of stable RNA (rRNAs and tRNAs) (Cashel et al., 1996), protein synthesis (Svitil et al., 1993; Milon et al., 2006), DNA replication (Wang et al., 2007), and induces transcription of genes involved in amino acid biosynthesis (Stephens et al., 1975; Choy, 2000), and stationary phase genes needed for survival (Cashel et al., 1996; Jishage et al., 2002).
In the model organism Escherichia coli and other Gram-negative bacteria of γ-subdivision, cellular accumulation of (p)ppGpp is governed by two homologous enzymes, RelA and SpoT. It is interesting to note that while the RelA enzyme is able to synthesize (p)ppGpp only upon amino acid starvation, the SpoT protein, on the other hand, is a bifunctional enzyme having mainly (p)ppGpp hydrolase activity along with a weak (p)ppGpp synthesizing ability. In contrast to Gram-negatives, Gram-positive bacteria usually contain only one gene rel, which codes for the Rel protein having both (p)ppGpp synthesis and hydrolase activities (Hogg et al., 2004; Mittenhuber, 2001). Recent studies also confirmed that certain plant species may carry single or multiple copy of relA and spoT like gene homologues, called rsh, the product of which, i.e. Rsh, is responsible for metabolism of (p)ppGpp (van der Biezen et al., 2000; Kasai et al., 2002; Takahashi et al., 2004; Potrykus and Cashel, 2008). Although the rsh genes were first named in plants, now they have become a generic term for full-length rel-spo homologues from any source. As (p)ppGpp synthesis and degradation are carried out by similar enzymes, several studies including bioinformatics analyses have revealed interesting domain architecture of these group of proteins. In general RelA, SpoT or Rel contains four domains from N-terminus to C-terminus, which are widely known as HD or (p)ppGpp hydrolase (p)ppGpp synthetase, TGS (named after threonyl-tRNA synthetase, GTPases and SpoT proteins where the domain is conserved) and ACT (named after acetolactate synthase, chorismate mutase and TyrR proteins where this domain is conserved) (Aravind and Koonin, 1998; Potrykus and Cashel, 2008). The function of hydrolase domain (it contains two highly conserved amino acid residues, histidine and aspartatic acid) is to hydrolyse (p)ppGpp (Aravind and Koonin, 1998). As RelA is unable to hydrolyse (p)ppGpp, as expected, these conserved amino acids of HD domain are replaced with other amino acids leading to loss of its hydrolysis function (Aravind and Koonin, 1998). On the other hand (p)ppGpp synthetase domain in RelA, SpoT and Rel is highly conserved and this domain catalyses transfer of pyrophosphate group from ATP to the 3′-OH group of GTP or GDP to synthesize (p)ppGpp. While the N-terminal part containing (p)ppGpp hydrolase and synthetase domains is responsible for enzymatic activities, the C-terminally located TGS and ACT domains are involved in regulation of these enzymatic activities (Schreiber et al., 1991; Gropp et al., 2001; Mechold et al., 2002; Hogg et al., 2004; Battesti and Bouveret, 2006; Potrykus and Cashel, 2008).
The activation signal for (p)ppGpp synthesis by SpoT enzyme is glucose starvation (Xiao et al., 1991) or inhibition of fatty acid synthesis (Seyfzadeh et al., 1993; Battesti and Bouveret, 2006) or iron depletion (Vinella et al., 2005). While (p)ppGpp synthetase activity of RelA is stimulated by binding of uncharged tRNAs to the ribosomal A-site during amino-acid starvation (Haseltine and Block, 1973; Wendrich et al., 2002), at present comparatively less is known about the mechanism of activation of the (p)ppGpp synthetase function of SpoT, although both are ribosome-associated proteins. Mechold et al. (2002) showed that under in vitro conditions removal of the C-terminal half of RelSeq protein, a single Rsh protein of Streptococcus equisimilis, activated its (p)ppGpp synthetase activity and simultaneously strongly inhibited (p)ppGpp degrading activity. The fact that (p)ppGpp is detectable in a relA–spoT+ strain but not in a relA–spoT– strain (Xiao et al., 1991) suggests that SpoT is indeed needed for (p)ppGpp synthesis under in vivo condition. As both RelA and SpoT have extensive amino acid sequence homology (Metzger et al., 1989), it certainly suggests that SpoT most likely has synthetase activity. Recently, a physical interaction between TGS domain of SpoT and the acyl carrier protein (ACP), a central cofactor in fatty acid biosynthesis, has been demonstrated and a model has been proposed in which ACP carries information about the cellular status of fatty acid metabolism, which in turn can trigger the conformational switch in SpoT to (p)ppGpp synthesis mode followed by (p)ppGpp accumulation (Battesti and Bouveret, 2006).
Although our knowledge about stringent response is mostly based on E. coli, there are relevant studies on different Gram-positive organisms including Bacillus subtilis. However, recently it has been reported that apart from rel, B. subtilis genome also carries two genes, yjbM and ywaC, that encode two novel small (p)ppGpp synthetases, YjbM and YwaC respectively (Nanamiya et al., 2008). Similarly, the Gram-positive organism Streptococcus mutans genome has also been reported recently to code for two novel proteins, called RelP and RelQ, with (p)ppGpp-synthesizing activities (Lemos et al., 2007). Similar sequences coexist generally in the genomes of bacteria belonging to the class Firmicutes, e.g. bacilli, streptococci, staphylococci, Listeria, clostridia along with canonical full-length rel gene in each case. Moreover, recently it has been reported that the plant Arabidopsis thaliana also carries four functional rsh genes (Mizusawa et al., 2008).
The Gram-negative organism Vibrio cholerae, an aetiological agent of the severe diarrhoeal disease cholera, resides in aquatic environment during interepidemic periods. As availability of nutrient in natural aquatic habitats is generally low, adaptation in such environmental conditions contributes significantly to the fitness of any organism including pathogenic V. cholerae. Thus, it appears that stringent response may play a significant role in growth and survival of V. cholerae both during intestinal phase and when they were released from a human body into the natural aquatic environment. Like E. coli, V. cholerae genome also contains the homologues of relA (TIGR annotation no. VC2451) and spoT (TIGR annotation no. VC2710) (Heidelberg et al., 2000). Similar to E. coli, RelA and SpoT proteins of V. cholerae also contain four conserved domains, i.e. hydrolase, synthetase, TGS and ACT. Recently, we have constructed and characterized different ΔrelA (Haralalka et al., 2003) and ΔrelAΔspoT mutants of V. cholerae (Das and Bhadra, 2008). The results of these studies indicate that RelA of V. cholerae has a similar function like that of E. coli, i.e. synthesis of (p)ppGpp upon amino acid starvation (Haralalka et al., 2003). However, unlike E. coli, V. choleraeΔrelAΔspoT mutants are not (p)ppGpp0 because they are able to produce (p)ppGpp upon glucose (Das and Bhadra, 2008) or fatty acid starvation (this study) and the results strongly suggest presence of RelA and SpoT-independent (p)ppGpp synthesis pathway in V. cholerae. Thus, the exact function of V. cholerae SpoT (hereafter simply referred to as SpoTVc) is not clear and the present study aimed to understand the spoT gene function in more detail. Apart from analysis of the spoT gene function, this study also reports identification of a novel gene (TIGR annotation no. VC1224, a hypothetical gene), the product of which appears to be responsible for RelA and SpoT-independent (p)ppGpp synthesis in V. cholerae. As the gene VC1224 is highly conserved in other Vibrio genomes like V. parahaemolyticus, V. vulnificus, V. fischeri, V. angustum, etc., it has been named relV[relA-like (p)ppGpp synthetase domain coding gene in vibrios]. To our knowledge this is the first report which demonstrates that apart from canonical relA and spoT genes, a third gene is involved in (p)ppGpp synthesis in a Gram-negative organism. The relV gene is 780 bp long and should code for the RelV protein of molecular mass of ∼29 kDa. Bioinformatics analysis indicates that although RelV contains a (p)ppGpp synthetase domain (KEGG ID vch:VC1224, amino acid position 106–196) like Rsh proteins, but it completely lacks HD, TGS and ACT domains (http://ssdb.genome.jp/ssdb-bin/ssdb_motif?kid=vch:VC1224). As V. choleraeΔrelAΔspoT mutant is not a (p)ppGpp0 strain, by repeated subculturing we isolated a suppressor strain BS1.1S which showed (p)ppGpp0 phenotype. We have also constructed a V. choleraeΔrelAΔspoTΔrelV triple mutant, called BRV1. Experimental evidences obtained from extensive genetic, physiological and complementation studies using a well-characterized (p)ppGpp0 strain of E. coli and the suppressor strain of V. cholerae BS1.1S or BRV1 with (p)ppGpp0 phenotype allowed us to conclude that the V. cholerae relV gene product is a novel (p)ppGpp synthetase. Thus, it appears that intracellular homeostasis of (p)ppGpp in V. cholerae is regulated by relA, spoT and relV gene products and it is indeed quite complex.
Vibrio choleraeΔrelAΔspoT strain can accumulate (p)ppGpp during fatty acid starvation
Recently we have constructed and characterized different V. choleraeΔrelAΔspoT double mutants and found that these strains can accumulate (p)ppGpp during glucose starvation suggesting RelA-SpoT-independent (p)ppGpp synthesis in V. cholerae (Das and Bhadra, 2008). It has been shown earlier that (p)ppGpp synthetase activity of SpoT of E. coli (henceforth abbreviated as SpoTEc) could also be induced by inhibiting fatty acid biosynthesis using the inhibitor cerulenin (Seyfzadeh et al., 1993). In E. coli, two enzymes, 3-ketoacyl–ACP synthetase I and II (products of fabB and fabF genes respectively) involved in fatty acid biosynthesis have been shown to be inhibited by cerulenin (Omura, 1981; Moche et al., 1999). Accumulation of (p)ppGpp in cerulenin-treated E. coliΔrelA mutant but not in ΔrelAΔspoT strain indicated that the phenomenon is SpoT dependent (Seyfzadeh et al., 1993; Battesti and Bouveret, 2006). As V. choleraeΔrelAΔspoT strain is not (p)ppGpp0 upon glucose starvation (Das and Bhadra, 2008), we wished to understand whether inhibition of fatty acid biosynthesis of these V. cholerae cells still can lead to cellular accumulation of (p)ppGpp. When the wild-type V. cholerae strain N16961 (henceforth designated vc wt, Table 1) or its ΔrelA or ΔrelAΔspoT derivative (Table 1) was starved for glucose or treated with cerulenin (see Experimental procedures), cellular accumulation of (p)ppGpp was noted in all the cases (Fig. 1A and B respectively). However, cerulenin-treated E. coli wild-type strain CF1648 and its ΔrelA (CF1652), but not its ΔrelAΔspoT (CF1693) derivative (Table 1), when used as controls, showed (p)ppGpp accumulation (Fig. 1C) as reported previously (Seyfzadeh et al., 1993). The result suggests that like E. coli, V. cholerae cells also behave similarly upon fatty acid starvation and accumulation of (p)ppGpp in vc ΔrelA cells indicated possible role of the SpoTVc enzyme. However, this was not the case because previously constructed vc ΔrelAΔspoT strain BS1.1 (Table 1), when treated similarly with cerulenin, like glucose starvation (Fig. 1A), accumulated significant amount of (p)ppGpp (data not shown). To rule out any residual activity of the truncated RelA protein in vc ΔrelAΔspoT strain BS1.1 (Das and Bhadra, 2008), we constructed another strain N16961-RN1 (Table 1) with the deletion of entire relA open reading frame (ORF), i.e. a ΔrelAorf strain (see Experimental procedures) followed by deletion of entire spoT gene in the genome of ΔrelAorf strain and thus generating the vc strain BSN1 (Table 1). Like BS1.1 (ΔrelAΔspoT), BSN1 (ΔrelAorfΔspoT) exhibited identical phenotypes (data not shown) including accumulation of (p)ppGpp during glucose (Fig. 1A) or fatty acid starvation (Fig. 1B). Thus, it appears that fatty acid biosynthesis inhibition-mediated (p)ppGpp accumulation in V. cholerae could be either SpoTVc dependent (in the case of ΔrelA cells) or independent (in the case of ΔrelAΔspoT cells).
Table 1. Bacterial strains and plasmids used in this study.
2.2 kb PCR amplified relV region was cloned into SacI-XbaI-digested pUC18; Apr
0.6 kb internal relV sequence of pBS17 replaced with 1.3 kb aadA1 gene; Apr Spr
2.9 kb XbaI-SacI fragment containing relV::aadA1 allele of pBS18 cloned into similarly digested pKAS32, Apr Spr
In vivo (p)ppGpp synthetase activity of SpoTVc
An E. coliΔrelAΔspoT strain is devoid of (p)ppGpp and therefore unable to grow on M9 minimal agar (MMA) plate without amino acids, while a ΔrelA mutant is able to grow on it due to (p)ppGpp synthesized by SpoTEc (Xiao et al., 1991; Gentry and Cashel, 1996). In fact, based on these explanations in vivo (p)ppGpp synthetase activity of SpoT and Rsh like enzymes of different organisms were established (Kasai et al., 2002; Tozawa et al., 2007). To assess the capacity of the V. cholerae spoT gene product to synthesize (p)ppGpp, if any, similar complementation experiments were conducted with the E. coliΔrelAΔspoT strain CF1693 (Table 1). To do this, the plasmid DNA pBSvc1 (Table 1), which carries entire spoT ORF of V. cholerae under an arabinose-inducible promoter PBAD, was used. As a control CF1693 carrying the plasmid pBSec1 (Table 1) containing entire spoT ORF of E. coli under PBAD promoter or the empty vector pBAD24 (Table 1) was used. Interestingly, while CF1693(pBSec1) strain restored growth on MMA containing proper antibiotics and 0.01% l-arabinose, CF1693(pBSvc1) failed to grow on this medium like the control strain CF1693 carrying the empty vector pBAD24 (Fig. 2A). The result suggests that while SpoTEcin trans could complement CF1693 for growth on MMA, SpoTVc probably has no synthetase activity. For further confirmation, accumulation of (p)ppGpp in glucose-starved CF1693(pBSvc1) and CF1693(pBSec1) cells was estimated. As controls E. coli wild-type (CF1648), ΔrelA (CF1652) and ΔrelAΔspoT (CF1693) strains (Table 1) carrying with or without empty vector pBAD24 were used. It was found that only CF1693(pBSec1) strain could accumulate substantial amount of (p)ppGpp during glucose starvation like the control strains CF1648 and CF1652 and no cellular accumulation of (p)ppGpp in CF1693(pBSvc1) (Fig. 2B). Furthermore, it has been reported that the morphology of E. coli (p)ppGpp0 (ΔrelAΔspoT) cells are of elongated type (Magnusson et al., 2007; Shah et al., 2008), which could be overcome when SpoTEc function is provided in trans (Xiao et al., 1991). However, molecular basis of elongated cell morphology phenotype of E. coli (p)ppGpp0 strain is currently not known. Using scanning electron microscopy (SEM) we did similar experiments with the CF1693(pBSec1) cells and as reported earlier such complementation in fact rescued the cell elongation defect of CF1693 (Fig. 3). However, similar experiments performed with CF1693(pBSvc1) cells failed to rectify the elongated cell morphology (Fig. 3). The results suggest either that SpoTVc has no (p)ppGpp synthetase activity or that its synthetase activity is not functional in heterologous E. coli genetic background due to some unknown reason or requirement of some factor(s) which are missing in E. coli. Interestingly, unlike E. coliΔrelAΔspoT, vc ΔrelAΔspoT strain BS1.1 had no such morphological defect (Fig. S1). However, BS1.1 is not a (p)ppGpp0 strain (Das and Bhadra, 2008). We argued that as vc ΔrelAΔspoT strain BS1.1 (Table 1) produces substantial amount of (p)ppGpp in the absence of a functional (p)ppGpp-degrading SpoTVc enzyme, sustained concentration of (p)ppGpp above optimal level may be harmful for bacterial cells, which may lead to development of suppressors with ppGpp0 phenotype for its survival. Based upon this rationale, the vc ΔrelAΔspoT strain BS1.1 was subcultured repeatedly on a rich medium like Luria agar (LA) for a few days and tested the growth of each individual colony on MMA plates. As expected, screening of several colonies allowed us to isolate suppressor strains with (p)ppGpp0 phenotype, which were unable to grow on MMA plates like the E. coli (p)ppGpp0 strain CF1693 (Fig. 4A). We have reported recently that initial growth of a vc ΔrelAΔspoT strain in a nutrient-rich medium is defective, which may be due to high basal level of (p)ppGpp produced from RelA-SpoT-independent source in V. cholerae (Das and Bhadra, 2008). Interestingly, V. cholerae suppressor strains showed no such initial growth defect like vc ΔrelAΔspoT mutant (data not shown) supporting further that they are most possibly devoid of (p)ppGpp. For detailed studies we selected one such vc ΔrelAΔspoT (BS1.1) derived suppressor strain, called BS1.1S (Table 1). BS1.1S was indeed a (p)ppGpp0 strain because it failed to accumulate (p)ppGpp upon glucose or fatty acid starvation like the E. coli (p)ppGpp0 strain CF1693 (Fig. 4B and C respectively). Like vc wt, cell morphology of BS1.1 [ΔrelAΔspoT (p)ppGpp+] or BS1.1S [ΔrelAΔspoT (p)ppGpp0] was found to be normal (Fig. S1). When BS1.1S cells carrying the plasmid pBSvc1 or pBSec1 were tested for growth on MMA plates with 0.01% arabinose, as expected, the growth defect of BS1.1S was readily rescued by the presence of plasmid pBSvc1 or pBSec1, but not by the control strain BS1.1S(pBAD24) (Fig. 4A). The results also support that the plasmid pBSvc1 indeed expressed SpoTVc and the enzyme has synthetase activity. For further confirmation of SpoTVc-mediated synthetase activity in homologous genetic background, the wild-type spoT gene was inserted in the ΔspoT locus of BS1.1S using the recombinant suicide vector pKST2.7 (Table 1), which contained 2.7 kb genomic region of the vc wt (N16961) carrying the entire spoT gene with its natural promoter and the constructed strain was designated BS1.1SR (BS1.1S:spoT::pKST2.7; Table 1). Insertion of the suicide plasmid pKST2.7 in the correct locus of BS1.1SR was confirmed by PCR as well as by DNA sequencing (data not shown). Unlike the parent BS1.1S strain, BS1.1SR (SpoTVc+) cells were able to grow on MMA medium with arabinose but the growth was not optimal as it was observed in the case of BS1.1S(pBSec1) or BS1.1S(pBSvc1) (Fig. 4A). The results indicate that either BS1.1SR expressed very low level of SpoTVc due to insertion of about 7.1 kb DNA containing entire recombinant suicide vector pKST2.7 in the spoT locus or it may be possible that expression level of SpoTVc under this merodiploid state was similar to that of monocopy containing wild-type cells but the enzyme has a very weak (p)ppGpp synthetase activity. The latter explanation appears to be true because it is obvious that overexpression of SpoT through an arabinose-inducible promoter PBAD showed slightly better growth on MMA plates (Fig. 4A). However, further studies are needed on (p)ppGpp synthetase activity of SpoTVc, which is currently undergoing in our laboratory.
In vivo (p)ppGpp hydrolase activity of SpoTVc
The hydrolase activity of SpoTVc was examined in E. coli genetic background based on the information that the (p)ppGpp hydrolase activity of SpoT is essential for viability under relA+ background (Xiao et al., 1991). We hypothesized that if functional V. cholerae relA and spoT genes carrying plasmids, pBD2A (Fig. 5) and pBSvc1 (Table 1) respectively, are co-transformed in E. coli relA–spoT– strain CF1693 (Table 1), the transformed cells should be viable because it will become relA+spoT+ and no transformant will be obtained with pBD2A alone because the cells will become relA+spoT–, which leads to accumulation of substantial amount of (p)ppGpp that is toxic for bacterial cells (Xiao et al., 1991; Haralalka et al., 2003; Lemos et al., 2007). When such experiments were performed, viable transformants of CF1693 were obtained only with co-transformation with the plasmids pBD2A and pBSvc1. Furthermore, co-transformation of CF1693 cells with pBSec1 (Table 1) carrying the spoT gene of E. coli and the plasmid pBD2A also yielded similar results. The results were confirmed by preparing plasmid DNA from CF1693(pBSvc1+pBD2A) or CF1693(pBSec1+pBD2A) cells followed by PCR analysis using specific primers and the plasmid DNA as templates (data not shown). We also did restriction digestion analysis of the prepared plasmid DNA, which gave consistent results (data not shown). Altogether, the experimental results support that the SpoTVc has (p)ppGpp hydrolase activity.
ACT domain is essential for (p)ppGpp hydrolase activity of SpoTVc
As it has been shown that C-terminal part of Rsh plays a critical role in regulation of enzyme activities (Schreiber et al., 1991; Gropp et al., 2001; Mechold et al., 2002; Hogg et al., 2004; Battesti and Bouveret, 2006; Potrykus and Cashel, 2008), we tried to find out the effect of deletion of ACT domain on hydrolase activity of SpoTVc. To address this issue, ACT domain coding region of the spoT ORF was deleted from the genome of vc ΔrelA strain (N16961-R13) using the recombinant suicide vector pBS7.1 (Table 1) and thus creating the vc ΔrelAΔspoTACT strain BS2.4 (Table 1). This is well established that E. coli cells are able to grow on MMA plate containing either 3-amino-1,2,4-triazole (AT), or serine, methionine, glycine and leucine (SMGL), only if their basal (p)ppGpp levels are high enough (Uzan and Danchin, 1978; Rudd et al., 1985). Although a vc ΔrelA strain (SpoT+) showed severe growth defect on AT or SMGL plate (Das and Bhadra, 2008), but deletion of the ACT domain coding region of V. cholerae spoT in a relA– background (ΔrelAΔspoTACT) rescued the growth defect of vc ΔrelA strain (Fig. 5A and B, respectively). The phenotype of vc ΔrelAΔspoTACT strain (BS2.4) was thus similar to that of vc ΔrelAΔspoT strain (BS1.1). It was suggested that there is a RelA-SpoT-independent source of (p)ppGpp in V. cholerae, which in the absence of hydrolase activity of SpoTVc most likely helps the ΔrelAΔspoT mutant to overcome the inhibitory actions of AT or SMGL (Das and Bhadra, 2008). Thus, the result indicates that the deletion of ACT domain most probably inactivates the hydrolase activity of SpoTVc. This conclusion is further supported by the fact that vc ΔrelAΔspoT mutant can accumulate significant amount of (p)ppGpp upon fatty acid starvation as already mentioned (Fig. 1B). AT and SMGL assays suggest that similar mechanism is most probably operative in the ΔrelAΔspoTACT strain (BS2.4).
In silico identification of a novel (p)ppGpp synthetase domain coding gene of V. cholerae
The observation that vc ΔrelAΔspoT strains can accumulate (p)ppGpp under glucose or fatty acid starvation (Fig. 1A and B respectively) prompted us to search the whole genome sequence of the V. cholerae strain N16961 (Heidelberg et al., 2000) for genes with (p)ppGpp synthetase domain coding region. Such analysis identified one hypothetical gene VC1224 in the large chromosome of V. cholerae coding for a putative (p)ppGpp synthetase domain containing protein and the genetic organization of the region in the genome of the strain N16961 (Table 1) is shown in Fig. 6A. In silico study indicated that the ORF size of VC1224 is 780 bp and it codes for a hypothetical protein of ∼29 kDa in size. Thus, it is about one-third in size compared with canonical RelA or SpoT protein (Fig. 6B). Comparison of the deduced amino acid sequences of VC1224 with those of known (p)ppGpp synthetase domain containing proteins revealed very low sequence similarities to the (p)ppGpp synthetase domain, which is one of the catalytic domains of the Rel/Spo protein family (Fig. 6C). Analysis of whole genome sequence data of other species of vibrios (http://biocyc.org/VCHO/select-org-prefs?object=VC1224andrequest=INVOKE-BROWSE-ORTHOLOGS) indicated that the gene is highly conserved in the genus Vibrio. Furthermore, blastp (http://blast.ncbi.nlm.nih.gov/Blast.cgi version 2.2.19) analysis of VC1224 protein sequence also supported strongly that the protein is highly conserved in vibrios as shown in Table 2. Based on this information and involvement of the gene in (p)ppGpp metabolism as evident from experimental results described below, we designated this hypothetical gene relV. COG analyses (accession number COG2357) also revealed that 21 proteins from 13 different genomes contained the orthologues of RelV among which YjbM, YwaC, Spy0873 and Spy1125 have already been characterized and reported for their (p)ppGpp synthetase activities (Lemos et al., 2007; Nanamiya et al., 2008). Although KEGG (http://www.genome.jp/kegg/) analysis showed all the orthologues of relV (KEGG ID, Vch:VC1224) including homologues from other Vibrio spp., but for an unknown reason in COG analysis of RelV orthologues of other vibrios were not included.
Expression of relV is lethal in E. coli spoT– cells and identification of relV suppressor allele
To determine whether the product of relV has any (p)ppGpp synthetase activity, we transformed E. coliΔrelA and ΔrelAΔspoT strains CF1652 and CF1693 respectively, with the plasmid pDPS3 (Table 1) carrying the entire relV gene including its natural promoter with flanking regions as a 2.3 kb insert (Fig. 6A). The construct pDPS3 was first obtained in E. coli DH5α cells and purified plasmid was used to transform strains CF1652 and CF1693. While transformants of CF1652 using pDPS3 were easily obtainable in LA plates with appropriate antibiotics, in sharp contrast, no transformant was obtained in the case of CF1693. The experiments were repeated at least five times and in each time similar results were obtained indicating that the spoT gene activity is essential for survival of E. coli cells carrying the functional copy of relV. For further confirmation, CF1693 strain was transformed with a coexpression recombinant plasmid pBPS3Hyb80 (Table 1) carrying the spoT and relV genes of V. cholerae, where the spoT ORF was under the control PBAD promoter of the vector pBAD24 and the relV gene contained its natural promoter. This time we obtained numerous transformants in LA plates with appropriate antibiotics and 0.01% l-arabinose. The result strongly suggests that the function of (p)ppGpp hydrolase activity of SpoTVc is essential to overcome the lethal effect of the product of relV, which is most likely able to synthesize (p)ppGpp in E. coli background.
As function of relV in a SpoT-deficient E. coli was lethal, we hypothesized that probably large-scale screening of E. coliΔrelAΔspoT strain CF1693 transformed with the plasmid pDPS3 carrying functional relV gene might allow to isolate a viable CF1693(pDPS3) strain with a mutation in relV to suppress the lethal effect of its wild-type allele. To check such possibility, we transformed ∼109 cfu of CF1693 with 1 μg of pDPS3 plasmid DNA and the total transformed cell population was plated on LA plates containing ampicillin, chloramphenicol and kanamycin antibiotics as selective markers. Such transformation-based screening allowed us to isolate one viable transformant of CF1693 on selective agar plates. The plasmid DNA, called pDPS3cf (Table 1), was isolated from this viable transformant followed by restriction analysis, which revealed identical profile with that of pre-transformation pDPS3 DNA (data not shown). For further confirmation of presence of any mutation, the entire relV insert DNA carried by the plasmid pDPS3cf was subjected to sequencing. Surprisingly, sequence analysis revealed point mutations at two nucleotides of the relV ORF, one at nucleotide position 296 (C(r)T) and another one at 685 (G(r)A) (data not shown), which resulted in replacement of the amino acid residues threonine to isoleucine at the position 99 (T99I) and glutamic acid to lysine at the position 229 (E229K) respectively. To rule out that CF1693 carrying the plasmid pDPS3cf (Table 1) was viable due to unknown chromosomal mutation in the host cell, the CF1693 strain was taken out from −70°C and directly inoculated in Luria–Bertani (LB) medium with appropriate antibiotics followed by transformation with the plasmid DNA pDPS3cf containing the relV296,685 suppressor allele, which yielded several viable transformants suggesting strongly that the point mutations present in the relV allele are actually responsible for viability of transformants which is otherwise lethal for CF1693 cells.
It is to be noted that apart from relV, the insert DNA of pDPS3 (Table 1) also carried VC1223 gene, which seems to be arranged in an operon (Fig. 6A). However, VC1223 codes for a 78-amino-acid-long hypothetical protein of about 7.8 kDa in size without any functional domain as shown by KEGG analysis (http://www.genome.jp/dbget-bin/www_bget?vch:VC1223). Now, to rule out any role of VC1223 gene product in lethality of CF1693, if any, several pDPS3-derived deletion constructs in plasmids were made and the construct containing intact VC1223 gene failed to show lethality in CF1693 (data not shown). Furthermore, in-frame deletion of relV in vc ΔrelAΔspoT background also abolished (p)ppGpp accumulation under glucose or fatty acid starvation (see below). Although currently the function of VC1223 is unknown, its role in regulation of relV cannot be ruled out because both genes are present in an operon, which we are currently investigating.
V. cholerae suppressor mutant BS1.1S also carries a point mutation in the relV gene
As BS1.1S suppressor strain of vc ΔrelAΔspoT behaved like a (p)ppGpp0 strain, we hypothesized that there may be similar mutations in the relV coding region of this strain like viable CF1693(pDPS3cf) carrying the relV296,685 allele. To examine such possibility, the entire relV gene region including its promoter was PCR amplified from each of vc ΔrelA, vc ΔrelAΔspoT and BS1.1S [ΔrelAΔspoT (p)ppGpp0] strain and each amplified product was cloned followed by sequencing. Interestingly, only the nucleotide sequences of the insert DNA originating from the suppressor strain BS1.1S revealed a single point mutation at the nucleotide position 676 of relV (relV676 allele) coding region where the base A was replaced by the base C (A(r)C) (Fig. S2). The recombinant plasmid containing relV676 allele was designated pDPS3S (Table 1). The deduced amino acid sequence of the relV allele (relV676) revealed replacement of the amino acid threonine with proline at position 226 of the RelV protein (RelV-T226P). It is noteworthy that the point mutations detected in pDPS3cf (relV296,685) and pDPS3S (relV676) are beyond the putative (p)ppGpp synthetase domain coding region, which covers nucleotide positions 318–588 and deduced amino acid positions 106–196 (http://ssdb.genome.jp/ssdb-bin/ssdb_motif?kid=vch:VC1224). At present it is not clear how these mutations are affecting the activity of the relV gene product. Altogether, the results indicate that relV is most likely a candidate gene for point mutation in the development of the suppressor strain BS1.1S with (p)ppGpp0 phenotype.
Co-expression and complementation analysis of relV in E. coli
We have shown that the (p)ppGpp hydrolase but not the synthetase activity of SpoTVc is retained in E. coli genetic background and transformation of CF1693 with the coexpression vector pBPS3Hyb80 (RelV+ SpoTVc+Table 1) but not with pDPS3 (RelV+Table 1) could yield viable transformants on LA plates. When the strain CF1693(pBPS3Hyb80) was streaked on an MMA plate with arabinose, as expected, the strain showed growth almost similar to its isogenic parent E. coli strains CF1648 (relA+spoT+), CF1652 (relA–spoT+) or CF1693(pDPS3cf) as shown in Fig. 7A. This result suggests that although there is a point mutation in the coding region of relV, which allowed the strain CF1693(pDPS3cf) to grow normally on LA plates, residual (p)ppGpp synthetase activity of the product (RelV-T99I,E229K) of the allele (relV296,685) was still retained, which most likely rescued severe growth defect of CF1693 (Fig. 7A).
To establish further the putative (p)ppGpp synthetase activity of relV, we directly measured cellular (p)ppGpp in CF1693(pBPS3Hyb80). As controls E. coli strains CF1648 (relA+spoT+), CF1652 (relA–spoT+), CF1693(pBSvc1) and CF1693(pBAD24) (Table 1) were used. Such analysis revealed that CF1693(pBPS3Hyb80) cells indeed can synthesize (p)ppGpp as shown in Fig. 7B.
Functional characterizations of ΔrelV mutants
To assess further the contribution of RelV to (p)ppGpp synthesis in V. cholerae, about 77% of the relV ORF containing entire (p)ppGpp synthetase domain was replaced with a spectinomycin-resistance gene cassette in vc wt (N16961), ΔrelA (N16961-R13) and ΔrelAΔspoT (BS1.1) strains (see Experimental procedures) and thus generating the single, double and triple mutants, NRV1 (ΔrelV), RRV1 (ΔrelAΔrelV) and BRV1 (ΔrelAΔrelVΔspoT) respectively (Table 1). All mutations were verified by PCR followed by DNA sequencing of the products (data not shown). Microscopical examinations of V. choleraeΔrelV mutants grown in LB revealed almost normal cellular morphologies (data not shown) like the relA (N16961), relA spoT (BS1.1) and suppressor mutant relA spoT relV676 (BS1.1S) cells as shown in Fig. S1.
The growth phenotypes of vc wt (N16961), ΔrelV (NRV1), ΔrelA (N16961-R13), ΔrelAΔrelV (RRV1), ΔrelAΔspoTΔrelV (BRV1) and ΔrelAΔspoT relV676 (BS1.1S) on MMA (Fig. 8A), AT (Fig. 8B) and SMGL (Fig. 8C) plates along with E. coli control strains CF1648 (relA+spoT+), CF1652 (relA–spoT+) and CF1693 (relA–spoT–) were tested. While vc wt, ΔrelV, ΔrelA, ΔrelAΔrelV, ΔrelAΔspoT all showed growth on MMA plates suggesting synthesis of (p)ppGpp, the triple mutant ΔrelAΔspoTΔrelV and the (p)ppGpp0 suppressor strain ΔrelAΔspoT relV676 failed to grow on MMA due to lack of (p)ppGpp. Silva and Benitez (2006) as well as our results (Das and Bhadra, 2008) indicated that growth of a relA mutant is poor in M9 minimal salt solution. However, when the vc ΔrelA strain N16961-R13 was streaked on MMA solid medium it showed growth but not like vc wt and the result is consistent as reported by Silva and Benitez (2006). At present the basis of this differential growth phenomenon of vc ΔrelA strains in M9 solid and liquid media is not known. When vc ΔrelAΔspoTΔrelV (BRV1), ΔrelAΔspoT relV676 suppressor strain (BS1.1S) and the control of E. coliΔrelAΔspoT (CF1693) were streaked on an MMA plate with all amino acids (Xiao et al., 1991), the mutants showed optimal growth (Fig. S3) suggesting that due to lack of (p)ppGpp, upregulation of amino acid biosynthesis operons in mutants was not possible and hence the mutant cells were unable to grow on MMA medium without amino acids. On the other hand, only vc wt (N16961), ΔrelV (NRV1) and ΔrelAΔspoT (BS1.1) were able to overcome the growth inhibitory effects of AT and SMGL when present in MMA (Fig. 8B and C respectively). The ability of vc ΔrelAΔspoT strain (BS1.1) to grow on these media and inability of ΔrelAΔspoTΔrelV mutant (BRV1) to do so confirms that RelV synthesizes (p)ppGpp and its amount reached at least to a requisite level so that the cells can overcome the inhibitory effects of AT and SMGL. The result is also consistent with similar analysis carried out on the (p)ppGpp0 suppressor strain BS1.1S (ΔrelAΔspoTΔrelV676). When the BRV1 (ΔrelAΔspoTΔrelV) triple deletion mutant was complemented with the relV gene region carrying plasmid pDPS3 (Table 1), it showed growth on MMA without amino acids; however, its growth was slow, which was possibly due to the multicopy expression of RelV through the high copy number plasmid of pDrive origin (data not shown).
Finally, when (p)ppGpp accumulation in vc ΔrelV (NRV1), ΔrelAΔrelV (RRV1) and ΔrelAΔspoTΔrelV (BRV1) mutants under glucose starvation was estimated, as expected, only NRV1 and RRV1 strains were able to produce (p)ppGpp spot on a thin-layer chromatography (TLC) plate as shown in Fig. 8D. In this assay we used vc ΔrelAΔspoT strain (BS1.1) as a control. Similar experiment carried out with cerulenin-treated cells for fatty acid starvation gave similar results (data not shown). Similar results were also obtained using the (p)ppGpp0 suppressor strain BS1.1S (ΔrelAΔspoT relV676) under glucose or fatty acid starvation as discussed above. Low intensities of (p)ppGpp spots obtained with vc ΔrelV (NRV1) and ΔrelAΔrelV (RRV1) compared with vc ΔrelAΔspoT (BS1.1) suggest that between SpoTVc and RelV, the latter protein most likely had a major contribution in (p)ppGpp production under the starvation conditions used in this study. This is further supported by the fact that the intensity of (p)ppGpp spot in RRV1, a ΔrelAΔrelV but spoT+ strain, was the lowest of all the spots detected in this experiment. All these experimental evidences allowed us to conclude that the relV gene product is indeed responsible for (p)ppGpp synthesis in vc ΔrelAΔspoT mutant.
To respond to changing environmental conditions particularly availability of nutrients, bacteria have evolved a multitude of cellular regulatory mechanisms among which stringent response mediated by the cellular alarmone (p)ppGpp is most important. Recently, we have provided evidences that there are fundamental differences in the metabolism of (p)ppGpp in V. cholerae compared with other bacterial systems particularly the model Gram-negative organism E. coli (Das and Bhadra, 2008). Analysis of several ΔrelAΔspoT mutant strains of V. cholerae indicated presence of RelA-SpoT-independent (p)ppGpp synthesis in V. cholerae. In continuation of our previous report (Das and Bhadra, 2008), this study has identified a third source of (p)ppGpp synthesis in V. cholerae and the new gene (TIGR annotation no. VC1224) coding for this (p)ppGpp synthetase enzyme has been designated relV. Furthermore, bioinformatics analysis of relV indicated that it is highly conserved among different Vibrio species like V. fischeri, V. parahaemolyticus, V. vulnificus, etc., the whole genome sequences of which are available in public database as well as in other vibrios with incomplete genome sequences (Table 2). We have also previously shown that mutation in the relA gene leads to no accumulation of (p)ppGpp in V. cholerae cells starved for amino acids (Haralalka et al., 2003). Therefore, the role of spoT gene function particularly its (p)ppGpp hydrolase activity seems to be crucial in V. cholerae, which has been studied here by genetic approaches.
It is known that inhibition of fatty acid biosynthesis in E. coli by using the inhibitor cerulenin leads to accumulation of (p)ppGpp which is SpoT dependent (Seyfzadeh et al., 1993). Although the mechanism by which (p)ppGpp synthetase activity of SpoT is induced by cellular fatty acid starvation is currently not clear, Battesti and Bouveret (2006) recently provided evidences that it is an ACP-mediated induction of (p)ppGpp synthetase activity of SpoTEc in E. coli. According to them the ACP interacts with the TGS domain of SpoTEc and this interaction is influenced by the ratio of unacylated ACP to acylated ACP in cells. Fatty acid starvation causes change in the ratio of unacylated to acylated ACP, which in turn leads to increase in synthetase activity of SpoTEc followed by cellular accumulation of (p)ppGpp. As glucose starvation indirectly affects fatty acid biosynthesis leading to increase in unacylated ACP molecules, they proposed that sensing of increased population of cellular unacylated ACP by SpoTEc might be responsible for increase in (p)ppGpp synthetase activity of SpoTEc followed by accumulation of (p)ppGpp. Whether similar mechanism of (p)ppGpp synthesis is operative upon glucose or fatty acid starvation in V. cholerae is currently not known. As V. cholerae relA spoT null mutants can accumulate substantial amount of (p)ppGpp upon glucose starvation (Das and Bhadra, 2008), it appears that ACP/SpoT interaction-independent pathway related to (p)ppGpp synthesis exists in V. cholerae. This conclusion was further supported by the fact that vc ΔrelAΔspoT cells were also able to accumulate (p)ppGpp upon fatty acid starvation (Fig. 1B). Thus (p)ppGpp metabolism appears to be complex in V. cholerae. Although from bioinformatics analysis it seems that SpoTEc and SpoTVc proteins are homologous (see Fig. 6B), but RelA-SpoT-independent (p)ppGpp production in V. cholerae raised the question about the role of (p)ppGpp synthetase activity of SpoTVc.
Our attempt to study the (p)ppGpp synthetase activity of SpoTVc by using an arabinose-inducible expression plasmid pBSvc1 (Table 1) and the E. coliΔrelAΔspoT strain CF1693 indicated that synthetase activity of SpoTVc is most likely not functional in E. coli background, which is based on the following evidences: (i) when SpoTVc was expressed in trans through the plasmid pBSvc1 in CF1693, it failed to rescue severe growth defect on MMA (Fig. 2A), (ii) CF1693(pBSvc1) strain failed to accumulate (p)ppGpp upon glucose starvation (Fig. 2B) and (iii) elongated cell morphology of CF1693, a property of E. coli (p)ppGpp0 strain (Xiao et al., 1991; Magnusson et al., 2007), was not rescued by the plasmid pBSvc1 as revealed by SEM analysis (Fig. 3). In sharp contrast, CF1693 similarly complemented with SpoTEc[CF1693(pBSec1), Table 1] overcame easily the growth defect on MMA (Fig. 2A), accumulated (p)ppGpp upon glucose starvation (Fig. 2B) and rescued elongated cell morphology (Fig. 3). Interestingly, using the same plasmid pBSvc1 (p)ppGpp synthetase activity of SpoTVc could be demonstrated in V. cholerae genetic background because the suppressor strain BS1.1S (ΔrelAΔspoT relV676) or the triple mutant BRV1 (ΔrelAΔspoTΔrelV) with (p)ppGpp0 phenotype (see below) carrying this plasmid was able to grow on MMA. However, for unknown reasons we failed to detect clear accumulation of (p)ppGpp in strains BS1.1S(pBSvc1) upon glucose or fatty acid starvation (data not shown). This could be due to very weak synthetase activity of SpoTVc leading to very low quantity of (p)ppGpp production, which was not detectable by the TLC method. The synthetase activity of SpoTVc was finally tested and confirmed by constructing a more physiologically relevant ΔrelAΔrelV but spoT+ strain RRV1 (Table 1), which showed accumulation of (p)ppGpp upon glucose (see Fig. 8B) or fatty acid starvation (data not shown); however, (p)ppGpp produced by RRV1 was quite low, which further supports that SpoTVc has a weak synthetase activity. Interestingly, unlike an E. coli ppGpp° strain, cellular morphology of the suppressor strain BS1.1S (ΔrelAΔspoT relV676) or the triple mutant BRV1 (ΔrelAΔspoTΔrelV) with (p)ppGpp0 phenotype was similar to that of vc wt (Fig. S1) and thus, it is expected that the SpoTVc-complemented strain of BS1.1S(pBSvc1) or BRV1(pBSvc1) will show no change in cellular morphology (data not shown). It is noteworthy that the growth defect of BS1.1S or BRV1 on MMA was also rescued by expressing SpoTEc through the plasmid pBSec1 (Fig. 2A, data not shown). Thus, SpoTEc retains its (p)ppGpp synthetase activity in both homologous and heterologous genetic backgrounds. Recently, it has been shown by Sajish et al. (2007) that the synthetase catalytic sites of monofunctional RelA-like enzymes have a conserved acidic triad of amino acid residues (ExDD) that differs from the conserved basic triad of amino acid residues (RxKD) found in bifunctional SpoT-like enzymes. It is notable that SpoTVc carries the same RxKD motif of SpoTEc and other bifunctional Rsh proteins including RelSeq. Thus, it is not understood why synthetase activity of SpoTVc is inactive in E. coli. However, we are not ruling out the possibility of a strong imbalance between hydrolysis and synthesis, tilted towards its main function, i.e. hydrolysis when SpoTVc was expressed in E. coliΔrelAΔspoT strain CF1693. Furthermore, Mechold et al. (1996) have shown that highly expressed full-length RelSeq, a Rel/Spo homologue of Streptococcus equisimilis, in E. coli favours hydrolysis of (p)ppGpp rather than synthesis. It is also to be noted that no significant change in expression levels of SpoTEc and SpoTVc under 0.01% arabinose-inducible conditions were noticed by SDS-PAGE analysis (data not shown) indicating that the observed difference in activity of SpoTVc was not due to poor expression in E. coli. Thus, future studies are definitely needed to elucidate the basis of this host-specific behaviour of (p)ppGpp synthetase activity of SpoTVc.
We have previously shown that V. cholerae relA gene is functional in E. coli and its expression is lethal in CF1693 ppGpp0 strain (Haralalka et al., 2003). In this study we used a simple assay system developed by Xiao et al. (1991) where introduction of a functional relA gene, in single or multiple copy, in E. coli (p)ppGpp0 cells is lethal even in rich medium because of excessive levels of (p)ppGpp accumulation in absence of (p)ppGppase due to deletion in spoT. As co-transformation of CF1693 relA–spoT– cells with functional relA and spoT genes of V. cholerae yielded transformants, we concluded that SpoTVc has hydrolase activity. However, there are other assays for testing hydrolase activity that depend on reversing growth inhibition when known sources of (p)ppGpp synthesis are present, like in E. coli spoT203, which displays high basal (p)ppGpp levels due to defect in (p)ppGpp degradation and mutant cells grows slowly even on rich medium (Mechold et al., 1996; Battesti and Bouveret, 2006). As all these tests are based on E. coli, we believe that similar reporter strains of V. cholerae are to be developed for precise determination of the functions of SpoTVc under homologous genetic background, which is currently in progress in our laboratory.
The current concept that relA and spoT are the sole genetic determinants responsible for (p)ppGpp homeostasis in Gram-negative bacteria, although applicable in the case of V. cholerae, but at the same time we have provided experimental evidences that apart from relA and spoT genes, a new gene relV of V. cholerae also codes for a novel (p)ppGpp synthetase. Consistent to this report, recently, Lemos et al. (2007) have discovered two extra small proteins RelP (207 amino acids) and RelQ (221 amino acids) in Gram-positive S. mutans with (p)ppGpp synthetase activities, although the organism codes for a canonical Rel enzyme. Similarly, Nanamiya et al. (2008) have also identified two genes, yjbM and ywaC, in Gram-positive B. subtilis Rel+ strain that encode novel type of (p)ppGpp synthetases, which they called small alarmone synthetase or SAS. Like RelP and RelQ, the predicted products of these genes are also small, YjbM (211 amino acids) and YwaC (210 amino acids). Like these genes, the relV gene of V. cholerae also appears to code for a small protein (259 amino acids) having only the synthetase domain. However, alignment of RelV protein with YjbM and YwaC showed 15% and 17% identities respectively, while with RelP and RelQ it showed 15% and 12% identities indicating poor homology among these proteins. This is not unexpected, because all these bacteria itself are phylogenetically distantly located from V. cholerae and their niches are also different. Despite this difference, alignment of the synthetase domain of RelV with that of RelA, SpoT of V. cholerae and RelP, RelQ, YjbM or YwaC revealed high conservation of certain amino acid residues as shown in Fig. 6C. Further mutational studies on conserved amino acids of these novel proteins in relation to function may provide insight in their mechanism of action.
In this study we have provided several experimental evidences that in the relV gene product is responsible for RelA-SpoT-independent accumulation of (p)ppGpp under glucose or fatty acid starvation in V. cholerae. The most important observation was that introduction of cloned relV gene region present in the plasmid pDPS3 (Table 1) in E. coli RelA– SpoT– strain CF1693 is lethal even in a rich medium but not in CF1652 RelA– SpoT+ strain or in CF1693 provided with SpoTVcin trans (Fig. 7A). Thus, it is clearly understandable that the lethal effect in E. coli spoT– background was most likely due to (p)ppGpp synthesized by RelV. Detection of two point mutations in the coding region of the relV gene present in the plasmid pDPS3cf (Table 1), which allowed CF1693 to survive and grow in a rich medium and also on MMA, but slowly, supported further that the expression of RelV-T99I,E229K mutant protein with possibly lesser (p)ppGpp synthetase activity than the wild-type RelV, as a result optimal levels of (p)ppGpp concentration were attained in CF1693(pDPS3cf) hence growth was observed on MMA (Fig. 7). Isolation of BS1.1S suppressor strain of vc ΔrelAΔspoT relV676 with (p)ppGpp0 phenotype further allowed us to establish that relV indeed codes for a novel type of (p)ppGpp synthetase. Finally, we constructed a vc ΔrelAΔspoTΔrelV triple mutant strain BRV1 (Table 1), which failed to grow on MMA, AT and SMGL plates (Fig. 8A, B and C respectively) and did not accumulate (p)ppGpp under glucose (Fig. 8D) or fatty acid starvation (data not shown). It is to be noted that point mutations in the relV gene carried by pDPS3cf and a single point mutation detected in the relV gene of the suppressor strain BS1.1S were not localized with respect to the (p)ppGpp synthetase domain coding region of relV rather they occurred just before or after that region. Thus, at present it is not clear how these point mutations in relV are inactivating the synthetase function of its products. Further studies are required to get more insight about how RelV-like (p)ppGpp synthetases work and what are the conserved amino acids playing a role in its function.
It is interesting to note that deletion of spoT in V. cholerae under relA– background was not lethal due to cellular accumulation of (p)ppGpp through the activity of the relV gene product. This observation indicates that there may be a SpoT-independent alternative pathway of (p)ppGpp degradation in V. cholerae or it may be possible that the function of the relV gene product is regulated by some unknown factors which ultimately control the cellular level of (p)ppGpp. This might be possible because in E. coli spoT-negative background no such regulation exists as a result accumulation of high amount of (p)ppGpp occurs through the activity of the relV gene product leading to severe growth inhibition. In fact, the hypothesis that SpoT-independent degradation of (p)ppGpp is applicable to all those organisms where the third source of (p)ppGpp synthesis has been discovered in rel/spo-deleted strains like in B. subtilis (Nanamiya et al., 2008) and S. mutans (Lemos et al., 2007). Further investigations are warranted in this direction to understand comprehensively about how RelA, SpoT and RelV of V. cholerae intricately regulate basal level of (p)ppGpp under various conditions of growth.
In summary, this study has revealed that in contrast to E. coli, fatty acid starvation in V. cholerae leads to accumulation of (p)ppGpp irrespective of mutation in the relA or in relA spoT genes, which indicated probable presence of an alternative source of (p)ppGpp in this pathogen. Through genetic analysis we also showed that (p)ppGpp hydrolase activity of SpoTVc both in homologous and in E. coli backgrounds. Deletion analysis indicated that the C-terminal ACT domain of SpoTVc is essential for its hydrolase activity as reported earlier in other organisms. On the other hand (p)ppGpp synthetase activity could only be demonstrated in V. cholerae but not in E. coli cells, suggesting probable requirement of some cofactors for the synthetase activity in the latter strain which may be available in homologous genetic background. Most important observation in this study is the discovery of the relV gene in V. cholerae that codes for a new type of small Rsh protein containing only the (p)ppGpp synthetase domain, similar to rsh genes recently found in firmicutes (Lemos et al., 2007; Nanamiya et al., 2008). Thus, in V. cholerae the products of relA, spoT and relV genes have (p)ppGpp synthetase activities (Haralalka et al., 2003; Das and Bhadra, 2008; this study) and (p)ppGpp metabolism seems to be complex in this organism. To the best of our knowledge, this is the first report about the existence of a novel (p)ppGpp synthetase in Gram-negative bacteria and definitely future studies are needed to understand clearly about the function of the relV gene.
Bacterial strains, plasmids and growth conditions
The bacterial strains and plasmids used in this study are listed in Table 1. For cloning purpose we used E. coli DH5α strain. Both E. coli and V. cholerae cells were routinely grown in LB at 37°C with shaking as described previously (Haralalka et al., 2003). For plate culture LA was used containing 1.5% agar. MMA plates were prepared as described (Das and Bhadra, 2008). MMA plates with all amino acids (each 20 μg ml−1) were prepared according to Xiao et al. (1991). For induction of gene expression through arabinose-inducible promoter containing plasmids of pBAD origin, 0.01% l-arabinose (Sigma-Aldrich, USA) was used. Antibiotics were used at the following concentrations unless otherwise indicated: ampicillin, 100 μg ml−1; streptomycin, 100 μg ml−1; spectinomycin, 50 μg ml−1; kanamycin, 50 μg ml−1 for E. coli and 40 μg ml−1 for V. cholerae; chloramphenicol, 34 μg ml−1 for E. coli and 3 μg ml−1 for V. cholerae; tetracycline 10 μg ml−1 for E. coli and 1 μg ml−1 for V. cholerae. Bacterial cells including mutants were always maintained at −70°C in LB containing 20% glycerol. To avoid development of any suppressor mutant, all genetically engineered strains were minimally subcultured after their isolation and before any experiment the strains were directly inoculated from their −70°C stock. The growth of bacterial cultures was monitored spectrophotometrically by measuring the optical density of the culture at 600 nm (OD600).
Molecular biological methods
Standard molecular biological methods (Ausubel et al., 1989) were followed for chromosomal and plasmid DNA preparations, electroelution of DNA fragments, restriction enzyme digestion, DNA ligation, bacterial transformation, conjugation, agarose gel electrophoresis and Southern blotting unless stated otherwise. All restriction enzymes and nucleic acid-modifying enzymes were purchased from New England Biolabs, USA and were used essentially as directed by the manufacturer. Southern hybridization experiment and electroporation of V. cholerae cells were done as described previously (Das and Bhadra, 2008).
The recombinant suicide vector pBD2 (Table 1) was constructed by the following method: the plasmid DNA pRELVCH (Table 1) was double digested with EcoRV-HincII and the total relA ORF was replaced with a HincII digested out kanamycin resistance gene (kan) cassette from the plasmid pUC4K (Table 1) and the recombinant vector was designated as pBRK (Table 1). The relA::kan allele of pBRK was obtained by double digestion with KpnI-SacI followed by ligation of the fragment in similarly digested suicide vector pKAS32 (Skorupski and Taylor, 1996) followed by transformation in E. coli SM10λpir (Table 1) and thus generating the plasmid pBD2 (Table 1). Authenticity of the recombinant clone pBD2 was confirmed by DNA sequencing (data not shown). The recombinant vector pBD2A was constructed as follows: pRELVCH (Table 1) was double digested with the enzymes EcoRI-PstI to obtain 2.1 kb truncated relA gene of V. cholerae with a deletion of the C-terminal ACT domain coding region followed by gel purification of the fragment and cloning of the fragment in similarly digested 0.75 kb portion of bla gene deleted vector DNA pBR322 (Table 1). Recombinant vector pBS7.1 (Table 1) was constructed as described below: first by digesting pDS6.2 (Table 1) with HincII to remove the ACT domain coding region of spoT gene of V. cholerae followed by insertion of an EcoRV fragment containing chloramphenicol-resistant gene cassette (cam) from the plasmid ProTet.E (BD Biosciences, USA; Table 1) in the deleted site of the spoT ORF generating the recombinant plasmid pDS6.5 (Table 1), which was further double digested with KpnI-SacI to get the 2.7 kb ΔspoT::cam allele and it was cloned in similarly digested suicide vector pKAS32 to generate pBS7.1. The plasmids pBSvc1 and pBSec1 (Table 1) were constructed as follows: the spoT ORF of V. cholerae was PCR amplified using the primers STvcEx2F/STvcExR (Table S1) and genomic DNA of the V. cholerae strain N16961 (Table 1) as templates and the amplified 2.1 kb DNA fragment containing the entire spoT region was double digested with KpnI-XbaI followed by cloning of the fragment into similarly digested expression vector pBAD24 (Table 1). The desired clone was confirmed by DNA sequencing using the primers PBAD-F/PBAD-R (Table S1). For construction of pBSec1, the spoT ORF of E. coli was PCR amplified using the primers STecEx-F/STecEx-R (Table S1) and the wild-type CF1648 genomic DNA as template followed by cloning of the amplified E. coli spoT ORF pBAD24 as described in the case of pBSvc1. The clone pBSec1 was confirmed by DNA sequencing (data not shown). To construct the plasmid pDPS3 (Table 1) two primers 1224wF/1224wR (Table S1) were used to PCR amplify the VC1224 hypothetical gene region (Heidelberg et al., 2000) taking genomic DNA of V. cholerae strain N16961 as template. A desired single band of 2.3 kb in size of VC1224 gene region was obtained by PCR amplification, which was purified by electroelution and the fragment was cloned into a commercially available PCR cloning vector pDrive (Table 1) to generate the recombinant plasmid pDPS3 (Table 1). The recombinant plasmid pDPS3 was confirmed by DNA sequencing (data not shown). The coexpression vector pBPS3Hyb80 (Table 1), containing the spoT ORF and VC1224 gene of V. cholerae, was constructed as follows: VC1224 gene region was PCR amplified using the primers 1224Hy-R/M13R and pDPS3 DNA as a template followed by digestion of 1.1 kb PCR product with PstI. The standard M13 reverse primer (M13R) was purchased from NEB (USA). The plasmid pBSvc1 (Table 1) containing the spoT ORF of V. cholerae was digested with PstI and the PstI-digested VC1224 region was cloned into the same site, which is located just after the stop codon of the spoT ORF. The coexpression plasmid pBPS3Hyb80 (Table 1) thus constructed was confirmed by DNA sequencing (data not shown). Recombinant plasmid pKST2.7 was constructed by the following method: the plasmid pDSC2.7 (Table 1) was double digested with KpnI-SacI and the 2.7 kb fragment containing the spoT gene of V. cholerae along with its natural promoter was gel purified and cloned in a similarly digested suicide vector pKAS32 (Table 1). For the construction of the recombinant plasmid pBS20 (Table 1), about 2.2 kb relV gene containing region was PCR amplified using the primers Dw1224F/Dw1224R (Table S1) and wild-type V. cholerae N16961 genomic DNA as template. The amplified product was digested with XbaI-SacI and cloned into similarly digested pUC18 (Table 1) and designated as pBS17 (Table 1). To replace about 0.6 kb (∼77%) internal sequence of relV ORF with aadA1 gene the recombinant plasmids pBS17 and pFX389 were PCR amplified using primers Di1224F/Di1224R and SpecR-F/SpecR-R respectively, and amplified products were digested with KpnI-BamHI and ligated by using T4 DNA ligase and thus generating the plasmid pBS18 (Table 1). The ΔrelV::aadA1 allele was digested out from pBS18 by using restriction enzymes XbaI-SacI and cloned into similarly digested suicide vector pKAS32 and named pBS20 (Table 1). Authenticity of each clone was always verified by DNA sequencing.
Construction of strains
In the present study, we have used the positive selection vector pKAS32 (Table 1) for construction of the relA null mutant of V. cholerae. The plasmid pBD2 (Table 1) was utilized to disrupt entire relA ORF of V. cholerae N16961 essentially as described by Das and Bhadra (2008) and mutant was designated N16961-RN1 (Table 1). The presence of the ΔrelAorf::kan allele in the correct locus of the genome of V. cholerae strain was confirmed by PCR analysis using the primers PMBEX1/VCR1 (Table S1) and by DNA sequencing. The relA phenotypes of the strain N16961-RN1 was checked as described previously (Haralalka et al., 2003; Das and Bhadra, 2008). Deletion of the spoT gene from ΔrelAorf strain N16961-RN1 was done essentially as described by Das and Bhadra (2008) and thus generated the ΔrelAorfΔspoT strain BSN1 (Table 1). The presence of the ΔspoT::cm allele in the correct locus of the genome of BSN1 was confirmed by PCR, Southern hybridization, and reverse transcriptase (RT) PCR assay and by DNA sequencing (data not shown). For deletion analysis of ACT domain, vc ΔrelAΔspoTACT strain BS2.4 (Table 1) was constructed using the suicide pBS7.1 and ΔrelA strain N16961-R13 (Table 1) as described previously (Das and Bhadra, 2008). Southern hybridization, PCR analysis, RT-PCR assay and DNA sequencing (data not shown) confirmed the authenticity of mutation in the correct locus. Similarly, the suicide plasmid pBS20 (Table 1) was used to construct the V. choleraeΔrelV, ΔrelAΔrelV and ΔrelAΔspoTΔrelV mutants and they were designated NRV1, RRV1 and BRV1 (Table 1) respectively. All these mutants were confirmed by PCR and DNA sequencing analyses (data not shown).
DNA sequencing reactions were performed using the BigDye® Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems) essentially as recommended by the manufacturer. The samples were run on an ABI3130 genetic analyser using the POP-7 polymer (Applied Biosystems). Results were analysed using the software DNA Sequencing Analysis V5.1 (Applied Biosystems).
AT and SMGL tests
For functional assay of the relA or relA spoT or relA spoT relV mutant of V. cholerae, the AT test was performed using M9 minimal medium (Sigma, USA) supplemented with glucose (0.2%), all amino acids (4 μg ml−1) except histidine, adenine (1 mM), thiamine (1 mM), and AT (15 mM; Sigma, USA) as recommended previously (Rudd et al., 1985). Sensitivity of the E. coli relA or relA spoT double mutant to the amino acids SMGL in combination has been demonstrated earlier (Rudd et al., 1985). To detect the sensitivity of V. cholerae relA, relA spoT and relA spoT relV strains to SMGL, M9 minimal medium with glucose (0.2%), SMGL (100 μg ml−1 each), adenine (50 μg ml−1), thymine (50 μg ml−1) and calcium pantothenate (1 μg ml−1) were supplemented essentially as described earlier (Silva and Benitez, 2006; Das and Bhadra, 2008). V. cholerae mutants and their corresponding parent strains were streaked on AT or SMGL medium and the plates were incubated overnight at 37°C. E. coli relA strain CF1652, relA spoT strain CF1693 and their isogenic wild-type isolate CF1648 (Table 1) were always used as controls.
Determination of intracellular (p)ppGpp by TLC
V. cholerae strains were screened for patterns of (p)ppGpp accumulation after amino acid starvation as described previously (Haralalka et al., 2003). Level of (p)ppGpp in glucose starved bacterial cells was determined essentially as described by Das and Bhadra (2008). (p)ppGpp labelling during fatty acid starvation was done essentially as described by Battesti and Bouveret (2006) with following modifications. The bacteria were grown to exponential phase (OD600∼0.4) in LB at 37°C with continuous shaking. Then the exponential phase cells were washed and grown in low phosphate containing (about 0.2 mM phosphate) MOPS minimal medium with all supplements at 37°C with shaking for 10 min followed by labelling with [32P] H3PO4 (100 μCi ml−1) (BRIT, Mumbai, India) and fatty acid starvation was simulated by the addition of 200 μg ml−1 of cerulenin (Sigma, USA). Remaining steps including analysis of samples by TLC were essentially same as described above for amino acid starved cells.
Scanning electron microscopy
Scanning electron microscopy of bacterial samples was done as described previously (Shah et al., 2008) with minor modification. Briefly, LB-grown bacterial cells in log phase were harvested, washed in phosphate-buffered saline (PBS, pH 7.4), and resuspended in an equal volume of PBS. Bacterial cell suspension (20 μl) was diluted with 80 μl of PBS, mixed gently, and pipetted onto a poly-l-Lysine (1 mg ml−1, Sigma-Aldrich, USA) coated glass coverslips followed by fixing with 4% glutaraldehyde (Sigma-Aldrich) and post-fixed with 4% osmium tetroxide (OsO4; Sigma-Aldrich) for 1 h. Coverslips were then rinsed several times with PBS and cells were dehydrated with chilled ethanol series (10 ml each of 40%, 60%, 80%, 90%, and two changes with 100%) for a minimum of 10 min at each stage. Samples were critical point dried with an E3000 Critical Point Dryer (Quorum Technologies, UK). Glass coverslip containing treated cells was placed on an aluminium stub and sputter coated with gold (Model SC7620, Quorum Technologies, UK). Bacterial cells were examined using a scanning electron microscope (Model VEGA II LSU, Tescan, Czech Republic) at 10.0 kV. The images in the figures are representative of what was observed in 10 random fields in each of two independent experiments.
DNA sequence data were compiled and analysed by using the dnasis program (Hitachi Corporation, Yokohama, Japan). National Center for Biotechnology Information (NCBI) blastn program of version 2.2.15 was used to search for homologous sequences in the database (http://www.ncbi.nlm.nih.gov). The ORFs were subsequently subjected to database search using the blastp program of version 2.2.15 (http://www.ncbi.nlm.nih.gov). For designing PCR and other primers, Primer3 software was used (http://frodo.wi.mit.edu/). GeneDoc software (http://www.nrbsc.org/downloads/) was used for the alignment of protein sequences.
We thank Siddhartha Roy for his constant support in this study. We are grateful to M. Cashel, National Institute of Health, Bethesda, MD, for the generous gift of E. coli strains CF1648, CF1652 and CF1693. We thank F.X. Barre, Centre de Genetique Moleculaire, CNRS, France for providing the spectinomycin-resistance gene cassette (aadA1) containing plasmid pFX389 and help in constructing the relV::aadA1 allele. We also thank Kalpataru Halder for DNA sequencing and Sangita Shah and Pratap Koyal for their sincere help in this work. The work was supported by the research grants from the Council of Scientific and Industrial Research (CSIR), Government of India (GOI). B.D. and R.R.P. are grateful for research fellowships from CSIR and S.B. is grateful to Indian Council of Medical Research, GOI.