Identification of a Family of Vibrio Type III Secretion System Effectors That Contain a Conserved Serine/Threonine Kinase Domain

ABSTRACT Vibrio parahaemolyticus is a marine Gram-negative bacterium that is a leading cause of seafood-borne gastroenteritis. Pandemic strains of V. parahaemolyticus rely on a specialized protein secretion machinery known as the type III secretion system 2 (T3SS2) to cause disease. The T3SS2 mediates the delivery of effector proteins into the cytosol of infected cells, where they subvert multiple cellular pathways. Here, we identify a new T3SS2 effector protein encoded by VPA1328 (VP_RS21530) in V. parahaemolyticus RIMD2210633. Bioinformatic analysis revealed that VPA1328 is part of a larger family of uncharacterized T3SS effector proteins with homology to the VopG effector protein in Vibrio cholerae AM-19226. These VopG-like proteins are found in many but not all T3SS2 gene clusters and are distributed among diverse Vibrio species, including V. parahaemolyticus, V. cholerae, V. mimicus, and V. diabolicus and also in Shewanella baltica. Structure-based prediction analyses uncovered the presence of a conserved C-terminal kinase domain in VopG orthologs, similar to the serine/threonine kinase domain found in the NleH family of T3SS effector proteins. However, in contrast to NleH effector proteins, in tissue culture-based infections, VopG did not impede host cell death or suppress interleukin 8 (IL-8) secretion, suggesting a yet undefined role for VopG during V. parahaemolyticus infection. Collectively, our work reveals that VopG effector proteins, a new family of likely serine/threonine kinases, is widely distributed in the T3SS2 effector armamentarium among marine bacteria. IMPORTANCEVibrio parahaemolyticus is the leading bacterial cause of seafood-borne gastroenteritis worldwide. The pathogen relies on a type III secretion system to deliver a variety of effector proteins into the cytosol of infected cells to subvert cellular function. In this study, we identified a novel Vibrio parahaemolyticus effector protein that is similar to the VopG effector of Vibrio cholerae. VopG-like effectors were found in diverse Vibrio species and contain a conserved serine/threonine kinase domain that bears similarity to the kinase domain in the enterohemorrhagic Escherichia coli (EHEC) and Shigella NleH effectors that manipulate host cell survival pathways and host immune responses. Together our findings identify a new family of Vibrio effector proteins and highlight the role of horizontal gene transfer events among marine bacteria in shaping T3SS gene clusters.

clone, emerged and has been responsible for major outbreaks of gastroenteritis in diverse locations around the globe (2).
In addition to the presence of the characterized virulence factors thermostable direct hemolysin (TDH) and the tdh-related hemolysin (TRH), genome sequencing revealed that all V. parahaemolyticus strains encode a type III secretion system on chromosome 1 (T3SS1) (3). Furthermore, strains related to the pandemic clone harbor an evolutionarily distinct T3SS known as T3SS2 (4-6) encoded within an 80-kb V. parahaemolyticus pathogenicity island 7 (VPaI-7) on chromosome 2 (3). T3SSs are multicomponent nanomachines that enable Gram-negative bacteria to deliver proteins known as effectors directly from the bacterial cytosol into the cytosol of eukaryotic cells. Translocation of effectors into host cells enables pathogens to hijack host cell signaling, thereby manipulating a variety of host cell functions (reviewed in reference 7). Indeed, the virulence of many human, animal, and plant pathogens depends on the activity of the T3SS injectisome and the repertoire of effector proteins delivered to their respective hosts' cells (8,9).
Notably, most V. parahaemolyticus strains isolated from human clinical samples harbor T3SS2, and studies of animal models have shown that T3SS2 is essential for V. parahaemolyticus to colonize the intestine and to cause enteritis and diarrhea (10)(11)(12). Therefore, T3SS2 is considered a key V. parahaemolyticus virulence factor. Several T3SS2related gene clusters have been identified in other Vibrio species and are referred to as T3SS2 phylotypes (T3SS2a, T3SS2b, and T3SS2g) (13). T3SS2a include T3SS2 gene clusters related to those found in the tdh-positive V. parahaemolyticus pandemic strain RIMD2210633 and in Vibrio cholerae strain AM-19226. T3SS2b include T3SS2 gene clusters related to those found in V. parahaemolyticus strain TH3996 and V. cholerae strain 1587 (14). Finally, T3SS2g include T3SS2 gene clusters related to those encoded in V. parahaemolyticus strain MAVP-Q, which has features found in the T3SS2a and T3SS2b gene clusters (15).
In this study, we found that VPA1328, an open reading frame (ORF) in the V. parahaemolyticus VPaI-7, encodes a novel T3SS2 effector protein. VPA1328, renamed here VopG, due to its similarity to the uncharacterized V. cholerae effector VopG, is secreted in a T3SS2-dependent fashion. Comparative genomic and phylogenetic analyses revealed that VPA1328 and VopG are members of a larger family of T3SS2 effector proteins encoded within the T3SS2 clusters of vibrios outside V. parahaemolyticus and V. cholerae including Vibrio mimicus and Vibrio diabolicus and the marine bacterium Shewanella baltica. The association of vopG genes with insertion sequence elements in several of these clusters suggests independent horizontal gene transfer or rearrangement events in these loci. Furthermore, VopG proteins have a conserved domain that exhibits sequence and predicted structural similarity to the serine/threonine kinase domain in the well-characterized NleH family of T3SS effector proteins. These effectors have been linked to subversion of host cell survival pathways and suppression of innate immunity in infected cells (26)(27)(28). However, VopG did not block host cell death or interleukin 8 (IL-8) secretion in tissue culture-based infections, suggesting a yet undefined role for VopG during infection or functional redundancy with other T3SS2 effector proteins.

RESULTS
VPA1328 is a VopG homolog that is secreted and translocated into host cells by the Vibrio parahaemolyticus T3SS2. We carried out BLASTp-based homology searches of the V. parahaemolyticus VPaI-7 genomic island as a way to identify candidate new T3SS2 effector proteins. This approach suggested that VPA1328 (VP_RS21530 in the latest genome annotation of strain RIMD2210633) is a putative T3SS2 effector protein (see Table S2 in the supplemental material). VPA1328 is predicted to encode a 260amino-acid protein that shares ;42% amino acid sequence identity with the T3SS effector protein VopG, encoded in the phylogenetically related T3SS2 in V. cholerae AM-19226 (29) (Fig. 1A and B). The function of VopG remains unknown, but it is secreted and translocated by the V. cholerae T3SS2 and contributes to host cell cytotoxicity and colonization in a mouse model of infection (29), suggesting an important role for this effector in virulence (29). Even though VPA1328 and VopG are located in different locations within their respective T3SS2 clusters, their sequence similarity and presence in phylogenetically related T3SSs suggest that VPA1328 is a VopG homolog that functions as a V. parahaemolyticus T3SS2 effector protein. Below we refer to VPA1328 as VopG.
A band corresponding to the predicted size of the VopG-CyaA fusion (;74 kDa, along with some lower-molecular-weight species likely corresponding to degradation products) was observed in cell lysates from the WT strain harboring pVopG-CyaA, but not a control strain harboring the empty vector pCyaA ( Fig. 2A). VopG was detected only in supernatants when the WT (pVopG-CyaA) strain was grown under T3SS2-inducing conditions (LB 0.04% bile) and not in culture supernatants in strains lacking a functional T3SS2 ( Fig. 2A), suggesting that its secretion requires T3SS2 activity. Interestingly, previous transcriptomic analysis showed that expression of VPA1328 was increased by the presence of bile and controlled by VtrB, the master regulator of T3SS2 expression, suggesting that it is part of the VtrB regulon (31).
Analyses of VopG secretion from Dvscn1 (T3SS1-deficient) and Dvscn2 (T3SS2-deficient) strains strongly support the idea that VopG secretion requires T3SS2 and not T3SS1. When secretion by T3SS1 or T3SS2 or both T3SS was disabled by deletion of their respective ATPases, there was similar expression of VopG in cell lysates ( Fig. 2A); however, VopG was detected only in supernatants from the strain where T3SS1 was inactivated but not when T3SS2 was inactive. An identical pattern was observed with VopV, a known T3SS2 substrate ( Fig. 2A). The cytosolic RNA polymerase beta subunit (RpoB) was not detected in any of the culture supernatant samples, indicating that detection of VPA1328 in culture supernatants was not a consequence of bacterial lysis.
We also tested whether VPA1328 (VopG) is translocated into infected host cells by V. parahaemolyticus T3SS2. Caco-2 cells were infected with V. parahaemolyticus strains harboring the effector-CyaA reporter fusions described above, and the amount of intracellular cyclic AMP (cAMP) generated by each translocated effector was measured using an enzyme-linked immunosorbent assay (ELISA). As shown in Fig. 2B, after 1 h of infection, cAMP levels generated by VopV-CyaA (positive control) and VopG-CyaA were similar and significantly higher in cells infected with V. parahaemolyticus strains harboring a functional T3SS2 (Dvscn1, T3SS21) than in cells infected with a T3SS-deficient strain (Dvscn1 Dvscn2, T3SS2), which exhibited background cAMP levels (Fig. 2B). Together, these observations demonstrate that VopG is secreted and translocated into host cells in a T3SS2-dependent fashion and given its similarity to the V. cholerae VopG effector, strongly support the notion that VopG is a novel V. parahaemolyticus T3SS2 effector protein.
VopG homologs are widely distributed in vibrios harboring T3SS2 clusters. The presence of a VopG homolog encoded within the T3SS2 gene cluster in V. parahaemolyticus RIMD2210633 prompted us to investigate whether additional VopG homologs are present among distinct T3SS2 phylotypes. The VPA1328 sequence was used as a query to identify potential VopG homologs by sequential BLASTn, BLASTp, and tBLASTx searches, using publicly available bacterial genome sequences. With cutoff values of 60% sequence coverage and 40% sequence identity, 2,044 candidate VopG homologs were identified, including 123 nonredundant protein sequences ( Fig. 3A and B and Table S3). The majority of the VopG homologs (86%; n = 1,764) were encoded in V. parahaemolyticus strains and in V. cholerae strains (12.5%; n = 256), but homologs were also identified in V. mimicus (0.2%; n = 5), V. diabolicus (0.04%; n = 1), and Vibrio sp. (0.5%; n = 11) and in Shewanella strains (0.3%; n = 7), i.e., in most species known to harbor T3SS2 gene clusters. Interestingly, a T3SS2 gene cluster was not previously identified in V. diabolicus, a marine organism. However, it is important to note that not all Vibrio species, e.g., Vibrio anguillarum (32) which harbors T3SS2 gene clusters, carry genes that encode VopG homologs. Thus, even though VopG is widely distributed, this putative effector protein is not a universal component of the T3SS2.
Next, we evaluated the sequence relatedness of VopG homologs using phylogenetic analysis of the 123 nonredundant VopG sequences. Three distinct clades (A, B, and C) of VopG proteins were identified (Fig. 3C), but no clear correlation was found between these clades and T3SS2 phylotypes. For example, the VopG homologs of V. cholerae strain AM-19226 and V. parahaemolyticus RIMD2210633 (VPA1328) clustered in different clades (B and A, respectively) despite the fact that both these T3SS2 belong to the T3SS2a phylotype. The lack of correlation between the VopG clades and T3SS2 phylotypes suggests that VopG effectors have to some extent evolved independently of the T3SS2 machinery that delivers them to host cells.
Comparative genomic analyses were carried out to gain insights into variation of the genomic contexts of vopG genes within different T3SS2 gene clusters. Genome sequences from representatives of each clade of the VopG phylogenetic tree, including at least one genome for each different Vibrio species were used for these comparisons. As shown in Fig. 4, the overall genetic structure of these T3SS2 gene clusters is highly conserved, particularly in the regions encoding structural components of the T3SS2 apparatus. In most T3SS2 gene clusters, the relative position of vopG was similar with the exception of V. parahaemolyticus RIMD2210633. However, the nucleotide sequences and ORFs that are adjacent to the vopG homologs differed in most of the seven clusters analyzed in Fig. 4. In several of these cases, vopG was found to be close to sequences related to insertion sequence (IS) elements. This association raises the possibility that IS elements can promote the mobility of vopG loci and potentially account for the variations in the genetic contexts of these loci within different T3SS2 gene clusters.
Consistent with this idea, we identified two vopG homologs (FORC14_RS05860 and FORC14_RS06170) encoded in the T3SS2 gene cluster of V. parahaemolyticus strain FORC014. Analysis of their respective genetic contexts revealed that one of these vopG genes (FORC14_RS05860) is located at the end of the T3SS gene cluster and is flanked by IS200-like mobile genetic elements (see Fig. S1A in the supplemental material). These elements have high sequence identity to the ISVpa3 insertion sequence. ISVpa3 is an insertion sequence located adjacent to each copy of the TDH gene in V. parahaemolyticus strain RIMD2210633 and linked in some strains to deletion of TDH (33). While the T3SS2 gene cluster of V. parahaemolyticus RIMD2210633 has three copies of these ISVpa3 elements, V. parahaemolyticus strain FORC014 has six of these elements, two of them flanking one of the vopG homologs at the end of the cluster (Fig. S1B). Sequence analysis showed that the two vopG homologs in strain FORC014 share 69% nucleotide identity (Fig. S2). Both the sequence divergence of these two vopG genes and the mobile genetic elements flanking FORC14_RS05860 suggest that this vopG homolog was independently acquired, potentially via a horizontal gene transfer event, and not a duplication of FORC14_RS06170.
While the presence of a T3SS2 gene cluster in Shewanella baltica species has been inferred due to the presence of the vscn2 gene in strain S. baltica BA175 and S. baltica OS183 (34), information regarding the distribution and genetic context of the T3SS2 gene cluster in this genus has not been reported. We found that the Shewanella T3SS2 is located within a genomic island inserted between the SBAL678_RS45345 and SBAL678_RS45350 ORFs of reference strain OS678 (Fig. S1B). This genomic island includes 45 ORFs. Most of these genes encode structural components of the T3SS2 apparatus. Interestingly, not every T3SS2 gene cluster identified in Shewanella harbors a VopG-encoding gene (Fig. S1B).
VopG proteins have sequence and predicted structural similarity to the NleH family of serine/threonine kinases. The amino acid sequence conservation of the 123 VopG homologs was analyzed and depicted using WebLogo. The analysis showed a particularly striking conservation in the C termini of these amino acid sequences (Fig. S3), suggesting that this region of VopG includes a functional domain. To gain clues regarding the function of VopG proteins, we used the structure-based homology tools HHpred (35) and pGenTHREADER (36). For these analyses, the amino acid sequence of V. parahaemolyticus VPA1328 was used as a representative of the VopG family of effectors. Both algorithms detected a region in the VopG C terminus with similarity to NleH effectors, e.g., HHpred analysis uncovered the presence of a region of 22 amino acids in VPA1328 (positions 190 to 212) with identity to the T3SS effector proteins NleH1 from Escherichia coli O157:H7 strain Sakai (PDB accession no. 4LRJ chain B) and OspG from Shigella flexneri 2a strain 301 (PDB accession no. 4Q5E chain A).
Multiple sequence alignment of representatives of the NleH and VopG family of T3SS effector proteins were carried out to gain further insight into their similarity. Representatives from each clade of VopG proteins were included in these analyses. The analysis showed that the greatest similarity between VopG and NleH proteins is found in their C termini in the region that includes the characterized NleH serine/threonine kinase domain; in contrast, their N termini differ in both length and sequence ( Fig. 5A and Fig. S4). VopG proteins contain all the critical amino acid residues and motifs important for kinase activity, including the conserved catalytic residues glycine of the G-rich loop, the aspartic acid (D) and the asparagine (N) of the catalytic loop (alignment positions 200 to 205 in VPA1328) and the PID motif of the activation loop (alignment positions 220 to 233 in VPA1328). In addition, VopG proteins also share the invariant lysine (alignment position 109) involved in the autophosphorylation of the NleH family, and which has been used as a proxy to measure kinase activity (27) (Fig. 5A and  B). Thus, VopG family effector proteins harbor a NleH-like C-terminal serine/threonine kinase domain. Phylogenetic analysis of bacterial serine/threonine kinase domains also revealed the similarity of the kinase domains of VopG and NleH proteins (Fig. 6A). The VopG proteins clustered closer to the NleH proteins on this tree than to non-NleH serine threonine kinases from Legionella pneumophila, Yersinia pestis, and Salmonella enterica.
We derived a three-dimensional (3D) structural model of the C-terminal domain of VPA1328 using comparative homology modeling with I-TASSER (44) to gain further insight into the serine/threonine kinase domain of VopG proteins. In accord with the HHPred and pGenTHREADER analyses, I-TASSER identified NleH1, NleH2, and OspG as suitable models for comparative homology models using the crystal structures available for these proteins. Five models were obtained, and I-TASSER model 1 was chosen based on its error estimation, template modeling score (TM-score) and root mean square deviation (RMSD) values (Fig. S5). As shown in Fig. 6B, this model revealed the remarkable similarity of the predicted structure of the VPA1328 kinase domain with the NleH kinase domain. The structure of the catalytic pocket, including the positions of the predicted catalytic amino acid side chains (K109, D201, and N205) in VPA1328 and OspG structure overlap (Fig. 6B), strongly supporting the notion that the VopG family of proteins encode a NleH-like serine/threonine kinase domain.
VopG does not modulate T3SS2-mediated cytotoxicity or inhibit IL-8 production. NleH family effector proteins inhibit IL-8 expression (26) and host cell death during infection (26)(27)(28). Since V. parahaemolyticus capacity to suppress IL-8 secretion and to induce host cell death is partially dependent on a functional T3SS2 (19,45), we evaluated whether VopG contributes to these processes. A V. parahaemolyticus mutant strain harboring a deletion of vopG in the Dvscn1 genetic background was constructed to assess whether vopG modulates T3SS2-dependent killing of intestinal Caco-2 cells. As expected, Caco-2 cells were killed (,50% survival within 3.5 h of infection) by a V. parahaemolyticus strain harboring a functional T3SS2 (Dvscn1, T3SS21), whereas cells infected with a V. parahaemolyticus strain lacking both T3SSs (Dvscn1 Dvscn2) were not (Fig. 7A). However, the absence of vopG (Dvscn1 DvopG, T3SS21) did not alter survival of host cells infected with a functional T3SS2 (overlap of red and orange survival curves in Fig. 7A), suggesting that vopG does not modulate T3SS2-dependent cytotoxicity. We then tested whether VopG contributes to T3SS2-dependent suppression of IL-8 production in infected cells. HeLa cells were infected with V. parahaemolyticus strains for 1.5 or 3.5 h and then IL-8 secretion was stimulated by incubating the cells with IL-1b for 90 min, as previously described (19). Infection by V. parahaemolyticus inhibited IL-8 secretion in a T3SS2-dependent fashion, but the absence of vopG did not influence this phenotype (Fig. 7B). Together, these data suggest that VopG does not modulate T3SS2-dependent host cell death or inhibit IL-8 production in infected cells.

DISCUSSION
While all Vibrio parahaemolyticus strains harbor T3SS1, a hallmark of the pandemic V. parahaemolyticus O3:K6 clone and most human clinical V. parahaemolyticus isolates, is the presence of a second and phylogenetically distinct T3SS2. The latter T3SS is essential for both intestinal colonization and virulence in some animal models of disease (10,12). Here, we found that a T3SS2 ORF (VPA1328) likely corresponds to a novel V. parahaemolyticus T3SS2 effector protein. This ORF, which is secreted in a T3SS2-dependent fashion, bears similarity to the VopG effector found in the V. cholerae AM-19226 T3SS2. The function of the latter VopG protein is unknown, but it has been shown to be translocated to host cells and linked to V. cholerae AM-19226's pathogenicity. Bioinformatic analyses uncovered 123 nonredundant VopG-like proteins encoded in all three phylotypes of T3SS2 clusters in diverse Vibrio species and in Shewanella baltica. Interestingly, the evolutionary history of the T3SS2 phylotypes does not appear to correspond with the evolution of the three clades of VopG proteins that were uncovered by phylogenetic analysis. We found that the highly conserved C-terminal domains of VopG proteins bear striking structural similarity to the serine/threonine kinase domain of the NleH family of effectors found in enteric pathogens such as EHEC and Shigella (OspG). Thus, our findings support the idea that VopG effectors function as serine/threonine kinases in host cells.
The V. cholerae AM-19226 effector VopG had been classified as a V. cholerae-specific T3SS effector (29,46), but our analyses showed that VopG homologs belong to a larger family of putative effector proteins that is widely distributed among Vibrio species, including V. parahaemolyticus, V. cholerae, V. mimicus, and V. diabolicus as well as in strains of S. baltica. Recently, Matsuda et al. (13) proposed classifying T3SS2 effectors proteins as "core" effectors if they are conserved in both V. parahaemolyticus and non-O1/non-O139 V. cholerae and as "accessory" effectors if they are not. According to this classification, our work suggests that VopG corresponds to a core effector protein due to its presence in multiple Vibrio species. However, VopG homologs are not present in the T3SS2 gene clusters identified in all Vibrio species; e.g., the T3SS2 cluster in Vibrio anguillarum (32) lacks a VopG homolog and not all clusters in V. mimicus (47) encode a recognizable VopG.
T3SS2 gene clusters are classified into three phylotypes (T3SS2a, T3SS2b, and T3SS2g) that are believed to have been acquired through horizontal gene transfer events (13,15,47). Even though VopG homologs are not universally found in all T3SS2 gene clusters, we identified VopG homologs in all three T3SS2 phylotypes. Phylogenetic analysis identified three distinct VopG clades (Fig. 3C). These VopG clades did not correlate with T3SS2 phylotypes, i.e., all three clades were found in each T3SS2 phylotype. The apparent independent evolution of T3SS2 phylotypes and VopG clades supports the possibility that vopG genes have been independently acquired by different T3SS2 lineages.
The absence of vopG genes from certain T3SS2 clusters could be explained either by loss of vopG loci due to deletion event(s) or independent acquisition of vopG in some T3SS2 clusters. The presence of a second vopG homolog flanked by IS elements in V. parahaemolyticus strain FORC014 suggests insertion sequences may play a role in mobilizing vopG genes. These sequences bear similarity to the ISVpa3 insertion sequence first reported in V. parahaemolyticus RIMD2210633 (33). Since insertion sequences have been shown to shape bacterial genomic islands through rearrangements, insertion, and deletion events, it is plausible that ISVpa3-like elements have shaped the evolution of T3SS2 gene clusters through similar mechanisms. The apparent mobility of vopG loci adds an additional layer of complexity to our understanding of T3SS2 clusters. That is, these clusters appear to have been spread via horizontal gene transfer events among marine bacteria, and their repertoire of effector proteins appears to be "tunable" through independent horizontal gene transfer or rearrangement events.
Analysis of the amino acid sequences of the 123 nonredundant VopG homologs identified here revealed a particularly high degree of conservation in their C termini. This region of VopG proteins was found to be very similar to the conserved serine/threonine kinase domain in the NleH family of T3SS effector proteins. Thus, the conservation of this part of VopG effectors is likely explained by the presence of a functional kinase domain. Structural predictions, which showed that VopG proteins contain all the residues that constitute that catalytic pocket of NleH proteins, strongly support this hypothesis.
The NleH family of effectors contain a eukaryotic-like serine/threonine kinase domain that independently evolved in bacteria (37,38). VopG homologs harbor each of the key residues described in the NleH family of protein kinases. The classification of the NleH proteins as a distinct bacterial kinase family was made through structurebased phylogenetic analysis (37). Phylogenetic analysis showed that the C termini of VopG proteins have more similarity to the kinase domain of NleH proteins than to other bacterial protein kinases (Fig. 6A), but further structural information is required to determine whether VopG proteins are novel members of the NleH family or represent a distinct family on their own.
Although the conservation of key catalytic residues in VopG and NleH proteins provides evidence that supports the notion that VopG proteins are functional serine/threonine kinases, it is more problematic to speculate that their biological function is conserved as well. To date, NleH proteins have been shown to perturb the NF-k B pathway and impact cell survival and innate immune responses during infection through different molecular mechanisms (26-28, 40, 48). Both NleH1 and NleH2 proteins bind the host protein RPS3 leading to inhibition or activation of the NF-k B pathway, respectively (26,49), but only NleH1 suppresses IL-8 expression during EHEC infection (26). Despite these differences in control of IL-8 expression, both NleH1 and NleH2 inhibit apoptosis through interaction with the Bax inhibitor 1 protein (27). The Shigella OspG protein inhibits the NF-k B pathway by inhibiting the proteasomal destruction of Ik Ba (42), and the SboH protein of Salmonella bongori blocks intrinsic apoptotic pathways (28). The V. parahaemolyticus T3SS2 causes cell death (18,50,51) and suppresses IL-8 secretion through perturbation of the NF-k B pathway (19,45). While the exact mechanism of T3SS2-mediated host cell death is unknown, the VopZ effector protein plays an important role in inhibiting IL-8 production (19). Our tissue culture-based infection experiments did not reveal that VopG modulates T3SS2-dependent host cell death or IL-8 suppression (Fig. 7). Two possible scenarios might explain these observations. (i) VopG's contribution to these phenotypes are masked by the redundant effects of additional effectors such as VopZ. (ii) VopG targets host pathways that do not influence cell death or IL-8 synthesis. The N-terminal region of NleH proteins has been linked to substrate recognition and the observed functional differences between the NleH1 and NleH2 proteins (26,37,40). In this context, it is tempting to speculate that the sequence divergence observed within the N-terminal region of VopG homologs (see Fig. S3 and S4 in the supplemental material) has functional implications.
In summary, our work identifies a new family of VopG proteins that are likely T3SS2 effectors. These proteins contain a distinctive NleH-like serine/threonine kinase domain. Future biochemical and structural studies are required to corroborate these predictions. Moreover, defining the role(s) of these effectors in the pathogenicity and/or environmental adaptation of the diverse Vibrio and Shewanella species that encode them will be fruitful.

MATERIALS AND METHODS
Bacterial strains and growth conditions. All bacterial strains and plasmids used in this study are listed in Table S1 in the supplemental material. V. parahaemolyticus RIMD2210633 (3) and its Dvscn1, Dvscn2, and Dvscn1 Dvscn2 derivatives (19) were used in this study. Bacterial strains were routinely cultured in LB medium or on LB agar plates at 37°C. Culture medium was supplemented with the following antibiotics and chemicals: 0.04% bovine and ovine bile (Sigma catalog no. B8381); 5 mg/ml and 20 mg/ml chloramphenicol for V. parahaemolyticus and E. coli strains, respectively; 1 mM isopropyl-b-D-thiogalactopyranoside (IPTG) to induce expression vector pCyaA in secretion and translocation assays.
T3SS2 secretion assays. A reporter fusion was constructed between VPA1328 and the CyaA reporter encoded in plasmid pCyaA (a pMMB207 derivative) (19), generating plasmid pVPA1328 (VopG-CyaA), to investigate VPA1328 (VopG) secretion. A VopV-CyaA fusion (pVopV-CyaA) (19) was used as a positive control for T3SS2-dependent secretion, and the empty plasmid pCyaA was used as a negative control. Each plasmid was transformed into V. parahaemolyticus by electroporation as previously described (19). Secretion assays were performed by growing each V. parahaemolyticus strain for 1.5 h in LB medium supplemented with 0.04% bile. When cultures reached an optical density at 600 nm (OD 600 ) of 0.5 to 0.6, 1 mM IPTG was added to induce expression of the CyaA reporter fusion protein. After 1.5 h of induction, culture supernatants were collected by two centrifugations, one at 5,000 rpm for 20 min and a final centrifugation at 13,000 rpm for 5 min. The supernatants were then filter sterilized through a 0.22-mm filter and concentrated 100-fold by repeated centrifugation at 5,000 rpm for 30 min with an Amicon Ultra-15 centrifugal filter unit (Millipore) with a 10-kDa molecular weight cutoff and with two washes with 15 ml of 1Â phosphate-buffered saline (PBS) as an exchange buffer. Prior to concentrating the culture supernatant, bovine serum albumin (BSA) (1 mg/ml) was added to serve as a concentration/ loading control. Whole-cell lysates were prepared by solubilizing the bacterial pellets in 1Â Laemmli buffer. Lysate and supernatant samples were processed for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis by mixing them with loading buffer and boiling for 5 min; they were then run on 4 to 20% Mini-PROTEAN TGX Precast Gels (Bio-Rad) per 30 min at 200 V. For immunoblot analysis, gels were transferred to iBlot2 transfer stacks of polyvinylidene difluoride (PVDF) membranes (Invitrogen) and blocked by EveryBlot blocking buffer (Bio-Rad catalog no. 12010020). The Pierce Coomassie Plus (Bradford) assay kit (Thermo Fisher catalog no. 23236) was used for determination of protein concentrations. Antibodies were used at the following dilutions: anti-CyaA (mouse monoclonal, 1:2,000; Santa Cruz Biotechnology catalog no. sc-13582), anti-RpoB (rabbit monoclonal, 1:2,000; Abcam catalog no. ab191598), anti-mouse IgG conjugated to horseradish peroxidase (HRP) (anti-mouse IgG-HRP) (goat polyclonal 1:10,000; Thermo catalog no. 62-6520), and anti-rabbit IgG (H1L) secondary antibody conjugated to HRP (goat polyclonal 1:10,000; Invitrogen catalog no. G21234). The blots were developed with SuperSignal West Pico Plus substrate (Thermo Fisher catalog no. 35060), and imaging was performed on a C-DiGit Blot Scanner (LI-COR Biosciences). All blots are representative of at least three biological replicates.
Translocation of effector-CyaA fusion proteins. CyaA reporter fusion-based protein translocation assays were performed as previously described (18). Briefly, Caco-2 cells were seeded at 1.5 Â 10 4 cells/ well and cultured in DMEM210% FBS for 2 days in a 96-well plate at 37°C in 5% CO 2 . V. parahaemolyticus RIMD2210633 Dvscn1 and Dvscn1 Dvscn2 strains containing pCyaA, pVopV-CyaA or pVopG-CyaA were grown for 1.5 to 2 h until they reached an OD 600 of 0.6 in LB medium supplemented with 0.04% bile. The infection assay in Caco-2 cells was performed for 1 h at 37°C and 5% CO 2 and at a multiplicity of infection (MOI) of 50. The intracellular cyclic AMP (cAMP) levels in Caco-2 cells were determined using cAMP Biotrak enzyme immunoassay (EIA) kit (Cytiva catalog no. RPN2251) as described previously (52). Statistical analysis was performed with GraphPad Prism version 9 (GraphPad Software, San Diego, California, USA).
T3SS2-dependent cell death and IL-8 secretion assays. For cell survival assays, Caco-2 cells were seeded at 8.0 Â 10 4 cells/well into six-well plates and grown for 2 days in complete media. V. parahaemolyticus strains were cultured overnight and the next day diluted 1:100 into LB liquid media containing 0.04% bile (to induce T3SS2 expression) and grown for 2 h until attaining an OD 600 of 0.6. Cells were infected at an MOI of 1 and incubated at 37°C with 5% CO 2 . At each time point assayed (0.5, 1.0, 1.5, 2.5, and 3.5 h.), the medium was replaced with fresh complete DMEM medium supplemented with 100 mg/ ml of gentamicin. Cells were incubated overnight, and surviving cells were quantified either by trypan blue exclusion (0.4% trypan blue) and counted on a hemocytometer (Neubauer cell chamber). For the detection of secreted IL-8, Caco-2 cells were seeded at 1.0 Â 10 5 cells/well into 12-well plates and cultured in DMEM210% FBS for 24 h at 37°C in 5% CO 2 . V. parahaemolyticus strains were cultured, and T3SS2 was induced as described above. Cells were infected at an MOI of 1 and incubated at 37°C with 5% CO 2 for 1.5 or 3.5 h, and then, the infection was terminated by addition of gentamicin (100 mg/ml). In parallel with gentamicin, the cells were treated with IL-1b (25 ng/ml) or left untreated for 90 min. IL-8 in culture supernatant was then measured using a human IL-8 ELISA kit (ab46032). Statistical analysis was performed with GraphPad Prism version 9 (GraphPad Software, San Diego, California, USA).
Sequence and phylogenetic analysis. Identification of VopG orthologs was carried out using the VPA1328 amino acid and nucleotide sequences as queries in BLASTp, BLASTn, BLASTx, tBLASTn, and tBLASTx analyses (64) using publicly available bacterial genome sequences of the NCBI database (December 2020). A 94% sequence length, 40% identity and 60% sequence coverage threshold were used to select positive matches. Sequence conservation was analyzed by multiple sequence alignments using MAFFT (53) and T-Coffee Expresso (54) and visualized by ESPript 3.0 (55). WebLogo analysis was performed using multiple sequence alignments (56). Comparative genomic analysis of the T3SS2 gene clusters was performed using the multiple aligner Mauve (57) and the IslandViewer 4 pipeline (58) and EasyFig v2.2.2 (65). Nucleotide sequences were analyzed by the sequence visualization and annotation tool Artemis version 18.1 (59). Multiple sequence alignments were used for phylogenetic analyses that were performed with the Molecular Evolutionary Genetics Analysis (MEGA) software version 7.0 (60) and visualized by iTOL (61). Phylogenetic trees were built from the alignments by the bootstrap test of phylogeny (2000 replications) using the maximum-likelihood (ML) method with a Jones-Taylor-Thornton (JTT) correction model.
Remote homology prediction and homology modeling. Remote homology prediction of VPA1328 was performed using HHpred (35) and pGENTHREADER (36) on the PSIPRED server (62). Protein structure models of the VPA1328 C-terminal domain were obtained using I-TASSER (44), a protein structure homology-modeling server. Protein structure visualization and template alignment and superposition were performed using MAESTRO (63).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.