Presently, studies from C. jejuni and the closely related gastric pathogen, H. pylori, report mostly the phenotypic effects of csrA mutation
[13, 23]. Furthermore, in C. jejuni as well as H. pylori the small RNA molecules (e.g. csrB, csrC) and the other proteins (e.g. CsrD) known to be involved in the Csr pathway in E. coli are either unidentified or absent
[7, 27–30, 39]. This suggests that the molecular mechanisms may be somewhat different than those in E. coli. In this study, we sought to determine the capability of the C. jejuni CsrA ortholog to complement the phenotypes of an E. coli csrA mutant to gain insight into the mechanisms of C. jejuni CsrA function.
The E. coli csrA mutation has several phenotypes that can be used as tools for determining the capability of CsrA orthologs from other bacteria to complement the well-characterized E. coli strain. For instance, mutation of csrA in E. coli alters glycogen biosynthesis, biofilm accumulation, motility, and cellular morphology, as well as several other cellular processes. Mercante and colleagues
 used the glycogen, biofilm, and motility phenotypes as a means to analyze the effects of comprehensive alanine-scanning mutagenesis of E. coli CsrA. In that study, the authors were able to identify which amino acids were most important for regulating glycogen biosynthesis, biofilm production, and motility, while also defining two regions of CsrA that are responsible for RNA binding.
When we compared representative CsrA orthologs from other bacteria, we found that C. jejuni CsrA is considerably divergent, as it clustered distantly from the E. coli ortholog. In part this is due to the significantly larger size of CsrA orthologs in the C. jejuni cluster (75–76 amino acids) as compared to the E. coli cluster (61–67 amino acids, Figure
1A). Considering the phylogenetic divergence of C. jejuni CsrA, we also examined the amino acid sequences of several CsrA orthologs of the pathogenic bacteria represented in Figure
1A to investigate the conservation of individual residues known to be important for the function of E. coli CsrA
, and found that C. jejuni CsrA is considerably divergent in several key amino acid residues. Variability is found in both RNA binding domains, region 1 and region 2, although greater variation is found in region 2. The first region, residues 2–8, contains only two conservative substitutions (T5S and R7K) while the other four residues are identical. RNA binding region 2 is highly variable consisting of two residues that are identical to E. coli (R44 and E46), three similar amino acids (V40L, V42I, and I47L), and three non-conservative substitutions (S41M, H43L, and E45K). Between the defined binding regions, there were two non-conservative substitutions (T19E and N35E) we found to be particularly interesting because of their reported ability to improve the regulatory functions of CsrA in E. coli. Presently, we are not able to draw any specific conclusions as to the significance of the individual amino acid substitutions in C. jejuni as compared to E. coli; however, it is likely that this divergence from E. coli plays a role in the ability of the C. jejuni ortholog to bind to E. coli targets appropriately.
In several studies, researchers characterizing the CsrA orthologues of different bacteria have used the glycogen biosynthesis phenotype of the E. coli csrA mutant to determine the functional similarities. One such study, of particular interest to our laboratory, reported that the H. pylori ortholog of CsrA would not functionally complement the E. coli mutant as it failed to repress glycogen biosynthesis
. It is likely that the H. pylori CsrA complementation failure was due to differences in the functional mechanism of ε-proteobacterial CsrA, however, this may have been specific to the two CsrA-binding sites of the glgCAP mRNA but not to other CsrA targets. To test this for C. jejuni CsrA, we examined the ability of CsrACJ to complement multiple E. coli csrA mutant phenotypes. We first expressed the C. jejuni ortholog in the E. coli csrA mutant and assessed its ability to repress glycogen biosynthesis under gluconeogenic conditions. Similar to H. pylori CsrA, the C. jejuni CsrA ortholog was incapable of repressing glycogen accumulation in the E. coli csrA mutant.
We next examined the ability of the C. jejuni protein to complement the motility, biofilm accumulation, and cellular morphology phenotypes of the E. coli mutant as well. As with glycogen biosynthesis, CsrA-mediated regulation of biofilm formation in E. coli is based on repression of a synthetic pathway, in this case the pgaABCD operon
. However, CsrA mediated expression of PgaABCD appears to be more complicated than that of glycogen biosynthesis, as it was reported that the mRNA leader sequence of the operon contains as many as six CsrA binding sites compared to the two binding sites observed on the glg leader sequence. Regardless of the complexity of the molecular mechanism of CsrA regulation of PGA we found that, when expressed in the E. coli csrA mutant, C. jejuni CsrA successfully complemented the biofilm formation phenotype (p<0.001).
Considering that the regulation of the glg and pga operons are both examples of CsrA-mediated repression of a biosynthetic pathway, we wanted to determine the ability of C. jejuni CsrA to substitute for its E. coli ortholog when the activation of gene expression is required. Wei and colleagues demonstrated that CsrA is a potent activator of flhDC expression and is therefore required for synthesis of the E. coli flagellum
. When we expressed C. jejuni CsrA within the non-motile E. coli csrA mutant the phenotype was completely rescued (p<0.001) suggesting that the C. jejuni ortholog is capable of promoting FlhDC expression.
Finally, we assessed the ability of C. jejuni CsrA to rescue an uncharacterized phenotype such as the altered cellular morphology of the E. coli csrA mutant. When CsrA was discovered, Romeo and colleagues reported that the csrA mutant displayed a greater cellular size as compared to the wild type, which was most obvious in early stationary phase
. This phenotype was explained as a possible indirect effect of endogenous glycogen accumulation. When we grew the wild type, csrA mutant, and complemented E. coli strains in LB media and observed their morphology via scanning electron microscopy, we observed a modest but significant decrease in the size of the mutant as compared to the wild-type as reported by Romeo
 in the absence of arabinose; however, in the presence of arabinose we observed a marked decrease in the size of the mutant as compared to the wild type (p<0.001). This decrease in size was complemented by both orthologs of CsrA (p<0.001). As Romeo suggested that the size differences between the mutant and wild type may be due to the role of endogenous glycogen cellular morphology, it is possible that the presence of arabinose used for protein expression may play a separate metabolic role within the cell leading to the observed phenotype.
A number of studies have shown that regulation of mRNA targets by E. coli CsrA is complex
[12, 15, 35, 41]. Mercante et al.
 showed that proper regulation depends on simultaneous binding of E. coli CsrA to multiple sites on target mRNAs, involving both of the RNA-binding surfaces of CsrA, using a multi-site bridging mechanism, and also the formation of higher order ribonucleoprotein complexes. Therefore, it is possible that the lack of regulation of the E. coli glg genes by C. jejuni CsrA is not due just to simple binding of one glg site vs. another, but rather due to changes in the dynamics (i.e. not ‘all or nothing’) of one or more of these bridging or ribonucleoprotein formation processes. For example, even moderately decreased affinity of C. jejuni CsrA for one of the glg sites may inhibit the formation of multi-site bridges and ribonucleoprotein complexes and therefore not result in productive regulation.
Finally, the binding of some but not all E. coli CsrA binding sites by C. jejuni CsrA infers that ε-proteobacterial CsrA binding sites are likely to show at least subtle differences from such sites in E. coli. It further underscores that predictive algorithms based solely or primarily on E. coli CsrA binding sites may be problematic for identifying CsrA binding sites in ε-proteobacteria and other divergent bacteria (Figure
, and that experimental approaches are preferable (such studies are ongoing in our lab).