One of the major challenges in understanding T3SS-mediated disease is effector protein identification. However, knowing the arsenal of proteins that bacteria have at their disposal is an essential prerequisite to begin studying how disease occurs at the molecular level. In general, non-O1/non-O139 serogroup-associated cholera presents similarly to infection by O1 and O139 serogroup strains, with minor exceptions (e.g., a slight inflammatory component). Yet, most T3SS-positive non-O1/non-O139 strains do not encode TCP or CT. It is therefore presumed that the coordinated action of multiple effector proteins works to promote bacterial colonization in the intestinal epithelium, followed by disruption of cellular homeostasis, resulting in choleralike diarrhea. We therefore set out to survey the proteins encoded within the V. cholerae AM-19226 T3SS island for potential effector protein activity, using yeast growth inhibition and T3SS-dependent translocation into human cells as assays. It is important to note that the inability to cause a yeast growth defect does not necessarily exclude a protein as an effector, since some effector proteins might act cooperatively or in a coordinated manner with other effectors in vivo. In such a case, we would expect that the expression of multiple effectors in yeast would be required to produce a detrimental phenotype. However, we chose to begin our analysis by evaluating the potential activities of individual V. cholerae proteins.
A total of 15 proteins demonstrated the ability to inhibit growth when expressed in yeast, including four that required yeast-sensitizing agents to uncover the phenotype. The strongest inhibition was due to the expression of VopF (a known effector protein) and the product of A33_1706. VopF has also been shown by others to inhibit yeast growth and may function early in infection to reorganize the host cell actin network (56
). The expression of A33_1699 and A33_1663 (whose product is herein named VopX) also resulted in strong growth inhibition when expressed in yeast. The remaining proteins demonstrated mild to moderate inhibition. Collectively, our results indicate that yeast can demonstrate a range of growth defects and serve as a suitable system for screening V. cholerae
effector protein activity. This is consistent with the use of yeast to screen effector proteins from other bacterial T3SSs.
We used a FRET-based in vitro coculture assay to identify bona fide effector proteins that were translocated into eukaryotic host cells in a T3SS-dependent manner. We tested a total of 17 AM-19226 proteins for translocation: 14 identified by the yeast screen and an additional three that we chose for reasons discussed above. All were expressed as β-lactamase fusion proteins in V. cholerae cells that were cocultured with HeLa cells in the translocation assay. Eleven were detected as translocated proteins (Fig. and Tables and ). Interestingly, we detected various degrees of translocation with different A33_ORF-β-lactamase fusion proteins, with VopX consistently showing translocation to >50% of the cell population and VopF being one of the proteins demonstrating limited translocation. Because the MOI for this assay was similar for all assays (between 4 and 40), we conclude that the variation in reporting might be due to different translocation efficiencies in vitro for the fusion proteins or abilities of the fusion proteins to function once inside the HeLa cell. Western blot analysis suggested that all fusion proteins were made to similar levels, although the stability of each once inside eukaryotic cells was not determined. Certainly, it might be expected that some effector proteins are prioritized in their translocation, and given that all other effectors were present in addition to the highly expressed β-lactamase fusion protein, it is reasonable to speculate that the sensitivity of the FRET system facilitates detection under nonideal conditions where protein expression and levels might be dysregulated. Nonetheless, the assay provides a high level of confidence that all 11 proteins are bona fide effectors that depend on the T3SS for their injection into host cells.
We were surprised to find that A33_1706, one of the proteins producing the strongest growth defect when expressed in yeast, was not translocated in our assay. The predicted protein is larger than 450 amino acids and encodes two conserved domains having sequence similarity to bacterial IgG-like domains found in adhesins such as intimin. We therefore predict that A33_1706 encodes a protein that, rather than functioning as an effector, is important for the bacteria to colonize the intestinal epithelium in a TCP-independent manner. It is unclear at this time why the expression of the protein in yeast results in a strong growth defect, but based on sequence analysis using the SignalP algorithm, A33_1706 is predicted to encode a signal peptide (data not shown). We speculate that A33_1706 encodes a surface-expressed protein that may interact with eukaryotic cells and might have the capacity to interact with eukaryotic signaling pathways to promote colonization.
Two proteins that displayed relatively weak growth defects when expressed in yeast did not show translocation: A33_1668 and A33_1695. Our annotation suggested that the A33_1668 protein might begin ~160 bp upstream of the start site annotated at NCBI, and so we included the additional sequences in our β-lactamase fusion protein. The additional sequences could account for the lack of translocation, due to protein misfolding. However, we favor the interpretation that because the expression of A33_1668 has a weak phenotype in yeast, it is not a translocated effector. Additional studies using a fusion protein constructed based on the NCBI sequences are expected to resolve any remaining questions. The sequence of the A33_1695-encoded protein in NCBI matches our own annotation very closely (we included 9 additional base pairs at the 5′ end of the ORF), and so we expect that the results of the translocation assay indicate that A33_1695 is not an effector protein. The A33_1665-encoded protein also showed mild inhibition of yeast growth. Translocation assay results with the β-lactamase fusion to this protein were inconclusive: the protein appeared to be translocated in only two of the four assays that we conducted. Again, technical parameters of the assay that depend on protein stability and translocation efficiency may explain the inability to demonstrate consistent translocation.
We also conducted bioinformatic analysis of small T3SS-encoded ORFs that did not have a phenotype when expressed in yeast but were candidates for encoding chaperones. While T3SS chaperones do not exhibit primary amino acid sequence similarity, they often are small, have an acidic pI, and may possess a C-terminal amphipathic helix. Class I chaperones associate with effector proteins, while class II chaperones associate with the translocator proteins (44
). Class I chaperones share similar structures and contain five beta strands and three alpha helices, while class II chaperones often possess tetratricopeptidelike repeats in an all-alpha-helical array (28
). CesA is a chaperone in EPEC that has a unique structure, consisting of helices shaped like a hairpin, and has a pI of 9.5 (61
). CesA is sometimes referred to as a class IV chaperone (39
). Class V chaperones interact with the needle subunit in Yersinia
and also with a translocator in Salmonella
). The class V chaperone SsaE in Salmonella
contains three alpha helices, one of which is predicted to be amphipathic (39
). We have identified putative chaperones based on secondary-structure prediction (PSIPRED) and determination of the theoretical pI (ExPASy) of small, hypothetical proteins in the AM-19226 T3SS island (data not shown). Secondary-structure prediction did not reveal any proteins with similarities to the class I chaperones. The majority of the proteins predicted to have an acidic pI are mostly alpha helical in nature. The A33_1668 and A33_1694 proteins are almost entirely helical, with a small coiled region between two helices, most closely resembling the class V chaperones. The A33_1671 protein possesses two alpha helices which are connected by ~15 residues that are predicted to be coiled and a small beta strand. The A33_1683 protein is slightly large for a chaperone, at ~200 amino acids, but is predicted to have a pI of 5.10 and an alpha helix of almost 150 residues, followed by two beta strands and a small alpha helix.
We focused additional experiments on VopX as an effector protein because it encoded novel sequences and produced a strong phenotype when expressed in yeast. In addition, deletion of the vopX
gene resulted in a strain that showed a diminished ability to colonize the infant mouse intestine when competed against the isogenic strain encoding a wild-type VopX (Fig. ). The colonization defect was relatively mild compared to that observed for epidemic strains with deletions of TCP or even compared to that of an AM-19226 strain carrying a deletion in vopF
), so we postulate that VopX plays an accessory role in colonization or a role in maintaining colonization during infection or interacts with other effectors to promote full colonization. Because the model is limited to reporting on the colonization ability of a strain, we cannot say at this time whether VopX may play a more significant role in disease progression or in eliciting a diarrheal response.
While none of the other 13 yeast deletion strain we tested could restore the ability of yeast to grow in the presence of VopX, the deletion of the RLM1 gene suppressed the growth defect induced by VopX expression. Complementation of the RLM1 deletion reversed this suppression, resulting in a yeast strain that was again sensitive to the expression of VopX. These results further support the finding that the VopX-induced growth defect is mediated, either directly or indirectly, through Rlm1. We observed that growth inhibition was demonstrably weaker in the rlm1-Δ-complemented strain than in the strain with an intact, chromosomal RLM1 gene. The relatively weaker inhibition could be due to the cloning of limited sequences flanking the RLM1 gene (~500 bp upstream and downstream), which may have inadvertently omitted elements present in the chromosomal sequence that are necessary for proper RLM1 expression. Also, RLM1 is complemented on a plasmid. Therefore, the expression level of the gene may vary in comparison to its expression in the wild-type strain.
Rlm1 is a MADS box transcriptional factor regulating genes involved in cell wall biogenesis and maintenance. Rlm1 activity is regulated by the protein kinase C-mediated cell wall integrity (CWI) MAPK pathway (59
). The finding that the deletion of RLM1
suppresses the toxicity of VopX suggests several possible interpretations. First, based on our analyses, VopX activity may be involved in only one of the four MAPK pathways tested, likely the CWI pathway. Second, the negative effect of VopX may require its interaction with a gene product(s) whose expression depends on activation by Rlm1. Third, VopX may interact directly with Rlm1, causing deregulation of CWI components downstream of Rlm1 and resulting in inhibition of cell growth. Finally, it is possible that the expression of VopX activates the CWI pathway upstream of Bck1 and Slt2, the MAP kinases tested for suppression in our assay (Fig. ). Previous results showed that overexpression or overactivation of the Rho1 or Pkc1 inputs into the CWI pathway results in constitutive activity that is lethal for bck1
-Δ or slt2
-Δ strains but not for rlm1
-Δ strains (29
). Accordingly, the bck1
-Δ and slt2
-Δ mutations would fail to suppress VopX upstream activation of the pathway, while the rlm1
-Δ mutation would succeed. Further experiments are required to distinguish between these alternatives.
Rlm1 is one of four MADS box transcription factors present in yeast and is thought to function most similarly to the mammalian Mef2 family of proteins. As transcriptional regulators, Mef2-like proteins can be activated during the inflammatory process by phosphorylation as a result of the activity of the p38 MAP kinase (46
). Mef2 proteins have been shown to mediate smooth muscle cell differentiation but have recently been recognized as having a role in cell survival in response to cellular stress. It is therefore interesting to speculate that VopX interaction with Rlm1 and the CWI pathway in yeast may affect cell survival by disabling mechanisms important for cellular structure and that similar mechanisms of interaction may interfere with mammalian cell integrity and survival during V. cholerae
infection in the human intestine.
VopX expression appears to be coregulated with the genes encoding the T3SS structural apparatus. Its expression was detectable when the strain carrying the vopX-lacZ
promoter fusion was grown in LB alone and was strongly induced when the strain was grown in the presence of bile. VttRA
deletions markedly reduced the expression levels in LB plus bile, similar to the effect observed for reporter fusions for T3SS structural genes. It is unclear at this time whether all genes within the T3SS pathogenicity island are regulated similarly. However, previous results from reporter assays looking at vopF
expression suggest that at least some effector proteins are controlled by a regulatory network or set of signals that may vary from those governing the expression of the genes encoding the structural apparatus (1
The yeast screen proved to be both sensitive and effective, since we were able to detect growth defects of various degrees induced by V. cholerae protein expression. We also tested several V. cholerae proteins for their ability to inhibit yeast growth in the presence of stressors and discovered an additional four proteins that produced a phenotype in yeast. Because we did not employ stressors in looking for a phenotype for every A33_ORF or express putative effectors in combination in yeast, it is possible that we have not identified the complete assembly or repertoire of effector proteins encoded by AM-19226. However, the identification of 11 proteins in addition to VopF that are part of the V. cholerae T3SS arsenal provides an important first step in beginning to understand the pathogenic capabilities of T3SS-positive non-O1/non-O139 V. cholerae strains that cause diarrheal disease. Considering that epidemic strains rely on the toxin-coregulated pilus and cholera toxin as the main components of effecting colonization and diarrhea, it will be interesting to unravel the molecular mechanisms of the multiple effector proteins that lead to a choleralike disease in the absence of TCP and CT.