Distinct roles of σ54 during both free-swimming and colonizing phases of the V. cholerae pathogenic cycle
In the present study we have identified the presence of the alternative sigma factor σ54 in the human pathogen V. cholerae. Bacterial alternative sigma factors typically co-ordinate the transcription of genes that contribute to a common physiological response; two examples of this would be σ28-dependent motility genes (Blair, 1995) and σ38 (RpoS)-dependent stationary-phase genes (Hengge-Aronis, 1993). σ54 is unique in transcribing genes with diverse roles in different organisms. Moreover, there are a number of examples in which the σ54-holoenzyme transcribes genes with diverse physiological roles within the same bacterium. One particularly relevant example is the pathogen Ps. aeruginosa, in which σ54 is involved in the expression of genes for flagellar synthesis, pilus synthesis, glutamine prototrophy, non-pilus adhesin synthesis and osmoprotection (Ishimoto and Lory, 1989; Totten et al., 1990; Ramphal et al., 1991; Sage et al., 1997). As in Ps. aeruginosa, σ54 of V. cholerae appears to be involved in the expression of multiple sets of genes with different roles. We have identified at least three distinct roles of σ54 in V. cholerae. First, it is involved in the expression of glutamine synthetase (glnA). Second, it is required for polar flagellar synthesis. Third, it is involved in the expression of genes that enhance colonization, and these genes are not related to glnA transcription or motility per se.
Other alternative holoenzyme forms of RNA polymerase, such as σ38-holoenzyme (Nguyen et al., 1993), activate transcription in the absence of any accessory activating proteins (at least in vitro ), thus limiting their specificity to a single physiological response. The absolute requirement for an accessory-activating protein by σ54-holoenzyme allows this form of RNA polymerase to transcribe genes in response to different environmental cues; a single σ54-activator directs transcription of the appropriate set of genes for a single physiological response, but the presence of multiple activators responding to different environmental cues allows for multiple responses by σ54-holoenzyme. We have shown the involvement of two different σ54-activators, FlrA and FlrC, in flagellar synthesis, and have additionally implicated these activators in the expression of genes that enhance colonization. However, neither FlrA nor FlrC appear to play a role in the expression of glnA, indicating the presence of at least one additional Gln-specific σ54-activator in V. cholerae. In fact, we have identified six additional V. choleraeσ54-activators using PCR with degenerate oligonucleotides designed to recognize the conserved σ54-activation domain (K. Klose and J. Mekalanos, unpublished data); we do not yet know the function of all these activators, but one appears to be an NtrC homologue involved in glnA transcription.
The presence of multiple σ54-activators in V. cholerae provides a mechanism to allow σ54-holoenzyme to express different sets of genes in response to the changing environmental conditions experienced during the pathogenic cycle, both in free-swimming and colonizing phases. In the closely related fish pathogen V. anguillarum, σ54 similarly demonstrates distinct roles during the two phases of its pathogenic cycle: during the free-swimming phase it is required for flagellar synthesis that is necessary for infection of the host, and during the colonizing phase within the fish it is required for transcription of unidentified gene(s) that are not related to motility (O'Toole et al., 1996). O'Toole et al. speculate that the defect of the V. anguillarum rpoN mutant during colonization may be due to this strain's inability to grow in the presence of low amounts of iron. The V. cholerae rpoN mutant does not demonstrate a similar growth defect in limiting iron medium (data not shown), thus the σ54-dependent genes transcribed within the host by these similar organisms may be different and host adapted, whereas the free-swimming σ54-dependent genes may be very similar, given that both bacteria inhabit similar aquatic environments. In the developmental cycle of Myxococcus xanthus fruiting body formation, multiple σ54-activators are likewise involved in interacting with σ54-holoenzyme to express various sets of genes at different stages of development (Kaufman and Nixon, 1996; Keseler and Kaiser, 1997).
The role of σ54 in expression of glutamine synthetase
It is clear from the significant colonization defect of a V. cholerae glnA mutant that glutamine prototrophy is important during colonization. Apparently, glutamine is in limited supply in the infant mouse intestine. Yet the Gln phenotype of an rpoN mutant is not responsible for its colonization defect because the glnA gene that complements the Gln phenotype failed to complement its ability to colonize. In addition, a strain containing a mutation in the NtrC homologue (discussed above; presumably this is the activator of σ54-dependent glnA expression), which has a similar Gln phenotype, exhibits no defect for colonization (data not shown).
Although we have not yet cloned and sequenced the V. cholerae glnA promoter, we can already draw some conclusions about the arrangement of its regulatory elements. Like the glnA gene of S. typhimurium (Porter et al., 1995), the V. cholerae glnA gene is probably transcribed by σ54-holoenzyme and the NtrC homologue mentioned above, as well as from a separate σ54-independent promoter. A S. typhimurium ntrA (rpoN ) mutant, unlike the V. cholerae rpoN mutant, is a glutamine auxotroph due to repression from NtrC binding preventing transcription initiation at the σ54-independent promoter (McCarter et al., 1984; North et al., 1996); thus, the V. choleraeσ54-independent glnA promoter must not be similarly repressed by binding of the NtrC homologue, which otherwise would result in glutamine auxotrophy.
The flagellar transcription hierarchy of V. cholerae
Based on the results presented here, we can begin to define the transcription cascade that leads to flagellar synthesis in V. cholerae (Fig. 9). We have identified the flagellar regulatory locus flrABC of V. cholerae that encodes two σ54-activators, FlrA and FlrC. Two lines of evidence demonstrate that FlrA and σ54-holoenzyme transcribe flrBC : (i) transcription of the flrB promoter in V. cholerae is dependent on intact flrA and rpoN genes; and (ii) expression of FlrA in S. typhimurium activates transcription of the flrB promoter only in strains containing σ54. Moreover, a putative σ54-binding site was identified in the flrB promoter; deletion of this site reduced transcription to background levels in all V. cholerae strains. We were unable to identify any transcription components that modulate flrA expression and it is synthesized in an active form in S. typhimurium; thus, the synthesis and activity of FlrA may be modulated in V. cholerae but we currently lack evidence to support this hypothesis.
FlrB and FlrC have domains that classify them as members of the two-component family of bacterial signal transducers. By analogy with other two-component systems (see, for example, Stock et al., 1995), we infer that under inducing conditions the autokinase FlrB transfers a phosphate to the conserved aspartate residue at position 54 in the amino terminus of FlrC, thereby activating its ability to catalyse σ54-dependent transcription initiation. We have identified a mutation in the amino terminus of FlrC (M114→I) that appears to bypass the need for phosphorylation for transcriptional activation; this mutation possibly changes the conformation of the regulatory domain to more closely resemble the phosphorylated form. Interestingly, FlrB lacks a transmembrane motif common to many bacterial sensor kinases and is predicted to be cytoplasmic, indicating that it probably responds to an intracellular signal that leads to the formation of phosphorylated FlrC.
FlrA function may be required for the phosphorylation of FlrC, as well as its synthesis, based on the phenotypes of flrA strains that express the flrBC operon from a heterologous promoter. Only an flrA strain expressing FlrCM114→I demonstrates increased motility in a swarm assay and the cells appear somewhat more flagellated than the parental flrA strain. However, this strain is not flagellated to the uniform extent of a wild-type strain, demonstrating that FlrA must be required for transcription of other flagellar genes in addition to flrBC.
The polar flagellum of V. cholerae, similar to those of the closely related pathogens V. anguillarum and V. parahaemolyticus (McCarter, 1995; McGee et al., 1996), is composed of multiple flagellin subunits that are differentially regulated (Klose and Mekalanos, 1998). In other bacteria, the flagellin gene(s) lie at the bottom of the flagellar transcription hierarchy because of their need to be transcribed last, given their structural role in forming the flagellar filament. We have previously demonstrated that only one of the V. cholerae flagellins, FlaA, is essential for motility, and the gene encoding it is the only flagellin gene transcribed by σ54-holoenzyme (Klose and Mekalanos, 1998). The results presented here are consistent with the active (phosphorylated) form of FlrC activating σ54-dependent transcription of flaA : (i) high level transcription of the flaA promoter requires intact rpoN, flrA, flrB and flrC genes (because of the FlrA and FlrB requirements for FlrC expression and activity, respectively, we infer that the lack of flaA transcription in flrA and flrB strains is due to lack of synthesis or activity of FlrC); and (ii) introduction of the flrCM114→I mutation in an flrB strain results in a dramatic increase in flaA transcription, indicating that when FlrC is synthesized in an active form, it can activate flaA transcription.
The promoters for the non-essential flagellin genes flaE, flaD, and flaB are transcribed by σ28-holoenzyme (Klose and Mekalanos, 1998), yet there is essentially no transcription of these genes in the absence of σ54 or FlrA, demonstrating a hierarchy in which some aspect of FlrA function regulates σ28-dependent expression. In other bacteria, the activity of σ28 is negatively regulated by an anti-sigma factor; upon synthesis of a correctly formed flagellar hook–basal body complex, the anti-sigma factor is exported from the cell, allowing σ28 to associate with RNA polymerase and initiate transcription (Hughes et al., 1993). We speculate that FlrA and σ54-holoenzyme affect the activity of σ28 and possibly FlrC by transcribing flagellar genes necessary for σ28 anti-sigma factor export. Thus FlrA would appear to belong to a higher level in the flagellar hierarchy than either σ28 or FlrC, and correct expression of its target genes is required to proceed to expression of the subsequent level.
Expression of FlrCM114→I during in vitro conditions that induce virulence gene expression leads to a distinct difference in cell morphology, implicating a role for this σ54-activator in cell division. Inducing conditions involve the growth of the organisms under specific conditions of pH, temperature and osmolarity into stationary phase. It is still not clear how these in vitro conditions relate to the environmental signals present within the human intestine, but electron microscopy of adherent V. cholerae in rabbit intestines (Teppema et al., 1987) reveals elongated cells similar to those seen in stationary phase. In contrast, V. cholerae cells growing logarithmically exhibit a rounded, shortened appearance. Noticeably, when FlrCM114→I is expressed from the ctx promoter, these cells have the shortened appearance of logarithmically growing cells, even in stationary phase.
It is perhaps not surprising that a polar flagellar regulatory protein would also affect cell division because a single flagellum is assembled only at one pole, tying cell division to flagellar expression prevents assembly of multiple flagella. In the developmental bacterium Caulobacter crescentus, σ54-dependent gene expression is similarly tied to cell division (Brun and Shapiro, 1992). C. crescentusσ54 is required for synthesis of a polar flagellum, and is also required for normal cell division. Not only do our results indicate that active (phosphorylated) FlrC might stimulate cell division, but also that FlrC becomes less active during stationary phase or under colonizing conditions, to allow for cell elongation. Although adherent V. cholerae cells also appear flagellated (Teppema et al., 1987), these results suggest flagellar transcription is downregulated during colonization. This prediction is also consistent with the reciprocal nature of virulence gene expression and the motility phenotype previously observed (Gardel and Mekalanos, 1996).
The role of σ54 during colonization
The colonization defect of the rpoN mutant, as stated above, is not related to its Gln phenotype, and is apparently not due to non-motility per se because a variety of non-motile V. cholerae mutants do not exhibit the same reduced ability to colonize in the infant mouse cholera model (Richardson, 1991; Gardel and Mekalanos, 1996). Unlike Ps. aeruginosa, in which σ54 is required for transcription of the pilus that is the major colonization factor (Ishimoto and Lory, 1989), the V. cholerae rpoN mutant produces normal levels of TCP (and CT) under in vitro inducing conditions (data not shown). Unless σ54 contributes to the expression of these two major virulence factors only in vivo, the σ54-dependent colonization gene(s) is likely to not be a gene(s) related directly to TCP. Although TCP is the major colonization factor of V. cholerae, there is little evidence that suggests these type IV pili themselves bear adhesins for intestinal receptors. In P. aeruginosa, the expression of the adhesin associated with this organism's type IV pili is σ54 dependent (Totten et al., 1990). Thus, the colonization defect associated with the V. choleraeσ54 mutant may involve regulation of a TCP associated adhesin that is regulated differently than the rest of the TCP operon.
Mutations in the two flagellar σ54-activators FlrA and FlrC not only result in a non-flagellate phenotype but also result in significant colonization defects; an flrC mutant exhibits a similar colonization defect to that of an rpoN strain. The FlrC homologue FleR of Ps. aeruginosa has been shown to be involved not only in flagellar synthesis but also in the synthesis of non-pilus adhesin(s) to mucin (Ritchings et al., 1995). We speculate that FlrC of V. cholerae is similarly required for expression of some accessory colonization factor(s) in addition to flagellar genes. Given that in the natural host the bacterium must swim through the mucus lining of the intestinal epithelia to arrive at a permissive colonization site, adhesins to mucin would probably facilitate colonization and need to be expressed simultaneously with flagellar genes. Alternatively, some of the flagellar gene(s) themselves may serve as adhesins; a fucose-sensitive haemagglutinating activity has been described in V. cholerae that appears to be associated with motility (Gardel and Mekalanos, 1996).
The cell morphology phenotypes suggest that FlrC-mediated gene expression may be downregulated during colonization, and yet FlrC appears to be required for some aspect of gene expression during colonization. This temporal disparity could be explained in several ways. First, although non-motility does not cause a significant colonization defect in the infant mouse model, some other gene(s) transcribed by FlrC simultaneously with flagellar genes may be necessary to reach a permissive site for colonization, after which FlrC-dependent transcription is downregulated; one possibility would be transcription of an adhesin, as discussed above. Second, FlrC may play a negative role as a repressor of gene expression by virtue of DNA binding during colonization when its positive role as activator is downregulated. Other σ54-activators play important roles as repressors of gene expression; one such example already discussed is repression of the σ54-independent glnA promoter by NtrC in S. typhimurium (McCarter et al., 1984; North et al., 1996). Third, FlrC may be required for transcription of another regulatory protein that is active once the organism colonizes and FlrC activity is downregulated. There are several examples of regulatory proteins activating expression of other regulatory proteins that respond to different environmental conditions. One example is the cyclic AMP (cAMP) receptor protein (CRP) that activates many genes in response to growth on poor carbon sources, including the σ54-independent (σ70 dependent) glnA promoter already mentioned that also drives expression of ntrC, whose product activates transcription in response to limiting nitrogen availability.