The regulatory network of Vibrio parahaemolyticus type VI secretion system 1.

Type VI secretion systems (T6SSs) are widespread, tightly regulated, protein delivery apparatuses used by Gram-negative bacteria to outcompete their neighbours. The pathogen, Vibrio parahaemolyticus, encodes two T6SSs. These T6SSs are differentially regulated by external conditions. T6SS1, an antibacterial system predominantly found in pathogenic isolates, requires warm marine-like conditions and surface sensing for activation. The regulatory network that governs this activation is not well understood. In this work, we devised a screening methodology that allows us to easily monitor the outcome of bacterial competitions and thus to identify mutants that are defective in T6SS1-mediated bacterial killing. The methodology, termed Bacterial Competition Fluorescence (BaCoF), relies on detection of a fluorescent signal as an indicator of the survival and growth of a T6SS-sensitive, GFP-expressing prey that has been co-cultured with mutants derived from a T6SS+ attacker of interest. Using BaCoF, we screened a random transposon insertion mutant library and identified genes required for V. parahaemolyticus T6SS1 activation, among them TfoY and Tmk. We used epistasis experiments to determine the relationships between the newly identified components and other regulators that were previously described. Thus, we present here a detailed biological understanding of the T6SS1 regulatory network.

Introduction

Bacteria use various strategies to carve a niche by outcompeting other bacteria. One of these strategies includes producing antimicrobial proteins that are secreted by toxin‐delivery apparatuses, such as the type VI secretion system (T6SS) (Mougous et al., 2006; Pukatzki et al., 2006; Russell et al., 2011). T6SS is a contractile phage tail‐like, multi‐protein apparatus that is widespread in Gram‐negative bacteria (Bingle et al., 2008; Boyer et al., 2009; Basler et al., 2012). It is used to deliver toxins, termed effectors, into neighbouring cells in a contact‐dependent manner. Effectors can be delivered into either bacterial or eukaryotic neighbours, thus implicating T6SS in both antibacterial competition and virulence (Pukatzki et al., 2007; Russell et al., 2011). Notably, bacteria encode cognate immunity proteins that protect them against self‐intoxication by their own antibacterial T6SS effectors (Russell et al., 2011; 2012).

The production of antibacterial arsenals, such as the T6SS, is energy consuming. Therefore, bacteria evolved to regulate these weapons and activate them under specific conditions in which they are required to enhance fitness. Indeed, T6SSs were shown to be tightly regulated in many bacteria, and they may be induced by external conditions and cues such as quorum sensing (Zheng et al., 2010), salinity (Salomon et al., 2013), temperature (Salomon et al., 2013), mucin (Bachmann et al., 2015), chitin (Borgeaud et al., 2015), surface sensing (Salomon et al., 2013) and membrane damage (Basler et al., 2013; Ho et al., 2013).

Vibrio parahaemolyticus, a Gram‐negative halophylic bacterium, which is a major cause of seafood‐borne gastroenteritis worldwide (Newton et al., 2012; Zhang and Orth, 2013) and of acute hepatopancreatic necrosis disease (AHPND) in shrimp (Tran et al., 2013; Lai et al., 2015), encodes two functional T6SSs (T6SS1 and T6SS2) (Yu et al., 2012; Salomon et al., 2013). These T6SSs were shown to be differentially regulated by various external cues in the clinical isolate RIMD 2210633. T6SS2 is found in all V. parahaemolyticus isolates; however, its role remains unknown. It is active in low salinity (1% NaCl), at low temperatures (23°C), and is repressed upon surface sensing (which can be mimicked in suspension by the polar flagella inhibitor phenamil) (Salomon et al., 2013). In contrast, T6SS1, an antibacterial system predominantly found in clinical and AHPND‐causing isolates (Yu et al., 2012; Li et al., 2017), is active under warm marine‐like conditions and requires high salinity (3% NaCl), warm temperatures (30°C) and surface sensing for activation (Salomon et al., 2013). We previously identified three antibacterial effectors that are delivered by V. parahaemolyticus RIMD 2210633 T6SS1 to mediate this antibacterial activity: VP1388, VP1415 and VPA1263 (Salomon, Kinch, et al., 2014).

We previously described regulators that contribute to the activation of T6SS1 in V. parahaemolyticus. Two proteins encoded within the T6SS1 gene cluster (spanning vp1386‐vp1420), VP1391 and VP1407, act as positive regulators of T6SS1 and are required for its activation (Salomon et al., 2013; Salomon, Klimko, et al., 2014), whereas the histone‐like nucleoid structuring protein (H‐NS) acts as a negative regulator and represses the expression of T6SS1 when the external conditions are not ideal for inducing the system (Salomon, Klimko, et al., 2014). OpaR, the high cell density quorum‐sensing master regulator, also acts as a negative regulator of this system and its deletion de‐represses T6SS1 expression (Salomon et al., 2013). However, it remains largely unknown how this bacterium senses and translates surface, salinity and temperature conditions into a T6SS1‐activating signal. Moreover, we do not know whether additional components found outside of the T6SS1 gene cluster are required for its activity (e.g. domesticated core‐genome components).

In this study, we set out to identify components that govern the activity of V. parahaemolyticus T6SS1. To this end, we devised a fluorescence‐based bacterial competition assay that provides a simple, intermediate‐throughput method for screening transposon insertion libraries to identify mutants impaired in T6SS‐mediated antibacterial activity. Using this method, we identified new components required for T6SS1 activity, and we also determined their role within the T6SS1 regulatory network.

Results

Fluorescence‐based assay to identify mutants defective in T6SS‐mediated antibacterial activity

To identify components that are required for V. parahaemolyticus T6SS1 activity, we first sought to develop an assay that would allow us to screen a large number of V. parahaemolyticus mutants searching for those impaired in T6SS1‐mediated antibacterial activity. To this end, we devised a screening methodology termed BaCoF, which uses a green fluorescent protein (GFP) output to detect the outcome of bacterial competitions. GFP serves as an indicator of the survival and growth of a T6SS‐sensitive, GFP‐expressing prey that has been co‐cultured with mutants derived from a T6SS+ attacker. Competition with a T6SS+ attacker results in prey death and thus no GFP signal, whereas competition with a T6SS‐defective mutant allows prey growth and hence the subsequent detection of the GFP signal.

Briefly, the BaCoF methodology (Fig. 1A) comprises the following steps: (i) generating a random transposon insertion mutant attacker library; (ii) picking individual attacker mutants into 96‐well plates and allowing them to grow overnight; (iii) adding T6SS1‐sensitive prey that constitutively express GFP in the 96‐well plates; (iv) spotting the attacker: prey mixtures onto T6SS‐inducing, solid media plates; (v) following incubation, evaluating the fluorescence of spotted mixtures (detecting fluorescence indicates the lack of prey killing by the mutant attacker, which is thus considered a potential hit); (vi) streaking bacterial mixtures containing potential hits onto selective plates to isolate the attacker; and (vii) sequencing flanking regions of the transposon to identify the position of its insertion into the genome. The detailed methodology is found in the Methods section.

Figure 1

BaCoF screen reveals mutants defective in T6SS1‐mediated bacterial killing.

(A) Schematic representation of the BaCoF methodology. Erm, erythromycin; ON, overnight.

(B and C) Genomic map of BaCoF hits' transposon insertion sites (red triangles) within known T6SS1 gene clusters and modules (B) and previously non‐T6SS1 associated operons (C). Arrows indicate the direction of gene transcription. Locus numbers are listed above and gene names or annotations are listed below. In C, serial numbers of transposon (Tn5) mutant BaCoF hits are shown below in red.

Applying BaCoF to V. parahaemolyticus T6SS1

To apply the BaCoF screen on V. parahaemolyticus T6SS1, we used the V. parahaemolyticus RIMD 2210633 derivative, POR1, as the parental attacker strain (previously shown to use its T6SS1 to kill competing bacteria under warm marine‐like conditions), and the POR1 T6SS1‐sensitive derivative, Δvp1415‐6, constitutively expressing GFP from a plasmid, as prey. Δvp1415‐6 harbours a deletion in one of the T6SS1 effector/immunity pairs, vp1415/6, and is thus unable to protect itself against attackers that deliver the effector VP1415 via T6SS1 (Salomon, Kinch, et al., 2014).

After screening ~12 000 individual mini‐Tn5 transposon mutant attackers, we collected 95 potential hits that allowed the growth of the GFP‐expressing, T6SS1‐sensitive prey. Of these potential hits, 76 had a transposon inserted within the T6SS1 gene cluster (vp1386‐vp1420) or in a known T6SS1 immunity gene (vti2) (Fig. 1B). We expected that mutations within the T6SS1 cluster would affect the antibacterial activity because the cluster encodes core components essential for T6SS1. Therefore, these hits were not further investigated. The remaining 19 potential hits in which the transposon was inserted outside of the T6SS1 gene cluster were subjected to two additional rounds of BaCoF. Ten of the 19 hits passed this validation step and possessed transposon insertion in genes not previously associated with V. parahaemolyticus T6SS1: vp1028, vp2050, vp0013, vp2153 and vp2388 (Fig. 1C). These 10 potential hits were chosen for further investigation.

Identification of genes required for T6SS1 induction and activity

First, we determined the ability of the potential hits to mediate T6SS1‐dependent bacterial killing using a quantifiable bacterial competition assay. All 10 hits were impaired in their ability to intoxicate the Δvp1415‐6 prey, although vp2153::Tn and vp0013::Tn were only partially impaired (Fig. 2A). In addition, as the above self‐competition assay only monitors the delivery of one of the V. parahaemolyticus T6SS1 effectors (VP1415), we also measured the bacterial killing ability of these 10 hits by monitoring the survival of Escherichia coli prey. E. coli was chosen as prey because it was previously shown to be T6SS1 sensitive, and it is predicted to be targeted by all V. parahaemolyticus T6SS1 effectors as it does not possess immunity genes against them. As shown in Fig. 2B, potential hits with the transposon inserted in vp1028 and vp2050 lost their ability to kill E. coli prey completely and they were similar to a Δhcp1 strain used as a T6SS− control. However, only an intermediate effect on bacterial killing was observed for potential hits with the transposon inserted in vp0013, vp2153 and vp2388.

Figure 2

BaCoF hits present T6SS1‐mediated bacterial killing, secretion and growth defects.

(A and B) Viability counts of POR1/Δvp1415‐6 (A) and DH5α (B) prey before (0 h) and after (4 h) co‐culture with the indicated attackers on MLB media (containing 3% NaCl) at 30°C. Asterisks mark statistical significance between samples and POR1 parental attacker at 4 h timepoint by unpaired, two‐tailed Student's t‐test (*P < 0.05). DL, Detection limit.

(C) Expression (cells) and secretion (media) of VgrG1 were detected by immunoblotting using specific antibodies against VgrG1. Loading control (LC), visualized as trihalo compounds' fluorescence of the immunoblot membrane, is shown for total protein lysates.

(D) Growth of V. parahaemolyticus strains in MLB at 30°C is shown as OD600 measurements. Data are mean ± SD, n = 3.

Hampered killing activity could have resulted from inability to induce T6SS1 expression following the sensing of external cues (surface, temperature and salinity) or from inability to properly fire the effector‐decorated T6SS tail‐tube. To distinguish between these possibilities, we monitored the expression and secretion of the hallmark T6SS‐secreted tail‐tube component, VgrG1, under T6SS1‐inducing conditions. Hits with the transposon inserted in vp1028 and vp2050 were unable to express VgrG1 (and thus unable to secrete it). In contrast, hits with the transposon inserted in vp0013, vp2153 and vp2388 exhibited VgrG1 expression and secretion levels comparable to the parental POR1 strain (Fig. 2C).

As hits in vp0013, vp2153 and vp2388 exhibited parental‐level VgrG1 secretion (Fig. 2C) and were only partially defective in their E. coli killing ability (Fig. 2B), we suspected that the effect on bacterial killing could have indirectly resulted from impaired growth of some transposon insertion mutants. Of the mutants lacking bacterial killing activity against E. coli, transposon insertions in vp1028 caused no apparent growth defect. However, insertion in vp2050 had a dramatic growth defect. Of the mutants retaining intermediate killing activity, only transposon insertion in vp2153 had no apparent growth defect (Fig. 2D). These results suggest that, whereas the effect of transposon insertions in vp1028 on bacterial killing did not result from a growth defect, it is possible that the effect of the vp2050 insertion did. Moreover, it is also plausible that the intermediate killing ability of insertions in vp0013 and vp2388 resulted from reduced growth rather than impaired T6SS1 activity per se.

Next, we investigated whether the effects on T6SS1 activity can be separated from the effect that some of the hits had on growth. We reasoned that if the defect in T6SS1 activity was only the result of diminished growth or of a general defect in protein secretion across bacterial membranes, then similar effects should be observed in other protein secretion systems. Therefore, we tested how the transposon insertions affected the activity of another protein secretion system, T6SS2, by monitoring the expression and secretion of the hallmark secreted tail‐tube component, Hcp2. The hits that were dramatically affected in T6SS1 activity exhibited no defect in Hcp2 expression and secretion, whereas vp2388::Tn and vp0013::Tn did exhibit reduced Hcp2 secretion levels, compared with the parental strain (Supporting Information Fig. S1). Moreover, we were surprised to find reproducibly high levels of Hcp2 in the secreted fraction of the vp2050::Tn mutant, compared with the parental strain. Becuase T6SS1 induction was nearly completely abolished in this mutant, the finding that T6SS2 is active indicated that the growth defect observed in this mutant was not accompanied by a general secretion defect. Thus, these results suggest that the defect in T6SS1 induction in vp2050::Tn might not necessarily be linked to its growth defect.

Taking the above results together, we decided to focus our subsequent efforts on the hits in which T6SS1 activity was abolished (i.e. vp1028::Tn and vp2050::Tn). The other potential hits that resulted in only an intermediate effect on T6SS1‐mediated antibacterial activity were not included in subsequent analyses because we concluded that either (i) we cannot separate their growth defect from a general defect on protein secretion (vp0013::Tn and vp2388::Tn) or (ii) the effect on T6SS1 activity was minor (vp2153::Tn).

TfoY is a positive regulator required and sufficient for T6SS1 activation

Two of our BaCoF hits had a transposon inserted in vp1028, which encodes TfoY. Ectopic expression of TfoY from a plasmid restored the bacterial killing ability of the two transposon mutants, indicating that the mutation in tfoY caused the loss of T6SS1 activity (Supporting Information Fig. S2). Indeed, deletion of the tfoY open reading frame (ORF) recapitulated the phenotype observed for vp1028::Tn and abolished T6SS1 antibacterial activity (Fig. 3A and B).

Figure 3

TfoY is an essential positive regulator of V. parahaemolyticus T6SS1.

(A and B) Viability counts of POR1/Δvp1415‐6 (A) and DH5α (B) prey before (0 h) and after (4 h) co‐culture with the indicated attackers on MLB media (containing 3% NaCl) containing 0.1% L‐arabinose at 30°C. Asterisks mark statistical significance between samples and POR1 + pEmpty parental attacker at 4 h timepoint by unpaired, two‐tailed Student's t‐test (*P < 0.05). DL, Detection limit.

(C and D) Expression (cells) and secretion (media) of VgrG1 were detected by immunoblotting using specific antibodies against VgrG1. Loading control (LC), visualized as trihalo compounds' fluorescence of the immunoblot membrane, is shown for total protein lysates. As indicated, the strains contained arabinose‐inducible expression vectors of TfoY (pTfoY), TfoX (pTfoX), or an empty vector (pEmpty).

TfoY and its paralogue, TfoX (VP1241; 36% identity to TfoY), are considered regulatory proteins and are widespread in vibrios (Pollack‐Berti et al., 2010). The homologues of TfoY and TfoX in Vibrio cholerae (hereafter TfoYVc and TfoXVc respectively) were recently shown to orchestrate the activity of T6SS. TfoXVc (69% identity to its V. parahaemolyticus homologue) was found to be a master regulator of V. cholerae T6SS because it was required for chitin‐mediated induction of T6SS, and its overexpression resulted in T6SS induction and antibacterial activity even in the absence of chitin (Borgeaud et al., 2015). TfoYVc (67% identity to its V. parahaemolyticus homologue) was shown to be required for the constitutive activity of V. cholerae T6SS in the strain V52, and its deletion resulted in greatly reduced bacterial killing activity and Hcp expression. Moreover, the overexpression of TfoYVc bypassed the requirement for chitin induction of T6SS in strains in which the system is not constitutively active (Metzger et al., 2016).

As V. cholerae T6SS differs from V. parahaemolyticus T6SS1 in gene content, regulatory components and activating cues, we investigated whether the concerted regulation of T6SS in V. cholerae by both TfoY and TfoX also occurs in V. parahaemolyticus T6SS1. We found that whereas TfoY was necessary for surface‐induced activation of T6SS1 in V. parahaemolyticus [mimicked in suspension by the addition of the polar flagella inhibitor, phenamil (Salomon et al., 2013)], TfoX (encoded by vp1241) was not required, because T6SS1 activation in ΔtfoX resembled that of the parental strain (Fig. 3A–C). Moreover, the overexpression of TfoX from a plasmid could not compensate for the loss of T6SS1 activity in ΔtfoY, whereas TfoX overexpression had a negative effect on VgrG1 secretion in the presence of TfoY (Fig. 3C and D). We also found that the overexpression of TfoY, but not of TfoX, resulted in T6SS1 induction that bypassed the requirement for surface sensing activation (mimicked by the addition of phenamil) (Fig. 3D). TfoY overexpression also resulted in T6SS1 activation in non‐inducing salinity conditions (1% NaCl) and to a lesser extent also in non‐inducing temperatures (23°C) (Supporting Information Fig. S3). Notably, deletion of either tfoY or tfoX had no effect on V. parahaemolyticus growth (Supporting Information Fig. S4). These results suggest that whereas TfoY is a major positive regulator of surface‐induced T6SS1 in V. parahaemolyticus, TfoX has a negative effect on the system when it is overexpressed. Thus, the roles of TfoY and TfoX in the T6SS regulatory network differ between V. parahaemolyticus and V. cholerae.

Thymidylate kinase is required for T6SS1 activation

The mutant containing a transposon insertion in vp2050 was severely hampered in its ability to express and secrete VgrG1, as well as in its ability to kill competing bacteria and grow (Fig. 2A–C). Ectopic expression of VP2050 from a plasmid did not rescue these phenotypes, suggesting that the mutation in vp2050 was not the direct cause of the defect in T6SS1 activation (Fig. 4A and B). This result prompted us to investigate whether the inserted transposon caused a polar effect. Indeed, ectopic expression of the gene found directly downstream of vp2050, namely vp2049, rescued the defect in bacterial killing, as well as VgrG1 expression and secretion, and even the growth defect observed in this transposon mutant (Fig. 4). These results indicated that the insult to T6SS1 activation and to growth in this mutant resulted from the polar effect of the transposon insertion on vp2049. Ectopic expression of VP2049 even reversed the increased activity of T6SS2 observed in the vp2050::Tn mutant (Supporting Information Fig. S5). vp2049 is an essential gene encoding thymidylate kinase (Tmk), an enzyme that catalyses the phosphorylation of thymidine 5′‐monophosphate (dTMP) to form thymidine 5′‐diphosphate (dTDP) in both de novo and salvage pathways of dTTP synthesis (Reynes et al., 1996). In agreement with this prediction, we were unable to generate a Δvp2049 strain. Given the annotated activity of VP2049, the mechanism underlying its requirement for T6SS1 activation is not immediately apparent. Notably, it appears that Tmk does not actively play a positive regulatory role in T6SS1 activation because its overexpression did not bypass the need for surface sensing (Supporting Information Fig. S6). Taken together, these results suggest a role for Tmk, or one of the downstream products of its enzymatic activity, in activating T6SS1.

Figure 4

VP2049 (Tmk) is required for T6SS1 activity.

(A and B) Viability counts of POR1/Δvp1415‐6 (A) and DH5α (B) prey before (0 h) and after (4 h) co‐culture with the indicated attackers on MLB media (containing 3% NaCl) with 0.1% L‐arabinose at 30°C. Asterisks mark statistical significance between samples and POR1 + pEmpty parental attacker at 4 h timepoint by unpaired, two‐tailed Student's t‐test (*P < 0.05).

(C) Expression (cells) and secretion (media) of VgrG1 were detected by immunoblotting using specific antibodies against VgrG1. Loading control (LC), visualized as trihalo compounds' fluorescence of the immunoblot membrane, is shown for total protein lysates.

(D) Growth of V. parahaemolyticus strains in MLB containing 0.1% L‐arabinose and 250 μg ml−1 kanamycin at 30°C is shown as OD600 measurements. Data are mean ± SD, n = 3. As indicated, the strains contained arabinose‐inducible expression vectors of VP2050 (pVP2050), VP2049 (pVP2049), or an empty vector (pEmpty).

A complex regulatory network of negative and positive regulators controls T6SS1 activation

Identifying TfoY as a major positive regulator of V. parahaemolyticus T6SS1 prompted us to determine its role and position within the T6SS regulatory network. To this end, we performed epistasis experiments and determined the relationships between known T6SS1 regulators. We constructed mutant strains by combining deletions of T6SS1 negative (opaR or hns) and positive (tfoY, vp1391 or vp1407) regulators, as well as strains in which positive regulators are overexpressed in the background of a deletion in another positive regulator. We then monitored the effects of these genetic interactions on T6SS1 activity, as measured by antibacterial activity and by VgrG1 expression and secretion. De‐repression of T6SS1 by deletion of hns required the presence of all three positive regulators (tfoY, vp1391 and vp1407) (Fig. 5A and B). In contrast, whereas de‐repression by deletion of opaR required vp1391 and vp1407, a double deletion of both opaR and tfoY (ΔopaR/ΔtfoY) retained an intermediate activity of T6SS1 (Fig. 5A and C). These findings suggest that H‐NS and OpaR repress T6SS1 differently, whereby de‐repression by removal of OpaR but not by removal of H‐NS, can bypass the need for activation via TfoY. Moreover, TfoY‐mediated activation of T6SS1 was dependent on the presence of both vp1391 and vp1407 (Fig. 5D and E). Notably, neither the single nor the double gene deletion strains that were tested exhibited any growth defect, compared to the parental strain (Supporting Information Fig. S7). Taken together, these epistasis interactions reveal a complex regulatory mechanism governing the activation of V. parahaemolyticus T6SS1.

Figure 5

T6SS1 is regulated by a complex network of positive and negative regulators.

(A) Expression (cells) and secretion (media) of VgrG1 were detected by immunoblotting using specific antibodies against VgrG1. Loading control (LC), visualized as trihalo compounds' fluorescence of the immunoblot membrane, is shown for total protein lysates.

(B and C) Viability counts of POR1/Δvp1415‐6 prey before (0 h) and after (4 h) co‐culture with the indicated attackers on MLB media (containing 3% NaCl) at 30°C. DL, Detection limit.

(D) Expression (cells) and secretion (media) of VgrG1 were detected by immunoblotting using specific antibodies against VgrG1. Loading control (LC), visualized as trihalo compounds' fluorescence of the immunoblot membrane, is shown for total protein lysates.

(E) Viability counts of POR1/Δvp1415‐6 prey before (0 h) and after (4 h) co‐culture with the indicated attackers on MLB media containing 0.1% L‐arabinose at 30°C. As indicated, the strains contained the arabinose‐inducible expression vector of TfoY (pTfoY) or an empty vector (pEmpty).

Asterisks mark statistical significance between samples and POR1 or POR1 + pEmpty parental attacker at 4 h timepoint by unpaired, two‐tailed Student's t‐test (*P < 0.05).

Discussion

In this work, we devised a methodology, termed BaCoF, which can be used to screen large numbers of bacterial mutants in order to identify those in which antibacterial activity has been modified. BaCoF relies on fluorescence output as an indicator of bacterial competition outcomes. Using BaCoF, we screened V. parahaemolyticus transposon mutants and identified new players in the T6SS1 regulatory network.

Most of the V. parahaemolyticus transposon mutants that lost their ability to mediate T6SS1‐dependent bacterial killing had a transposon inserted within the T6SS1 gene cluster (vp1386‐vp1420). This result was to be expected, because the T6SS1 gene cluster encodes all the T6SS core components known to be required for normal activity (Boyer et al., 2009; Salomon et al., 2013). Although most genes in the T6SS1 cluster were represented at least once in the list of T6SS1‐defective mutants identified, there were a few that were not, such as vp1386‐vp1389 and vp1395‐vp1396‐7 [we recently showed that the region previously thought to encode two proteins, VP1396 and VP1397, actually contains only a single open reading frame termed vp1396‐7. The erroneous annotation resulted from a mistake in the original deposited sequence (Li et al., 2017)], and vp1418‐vp1420. This implies that these unrepresented genes, which are not known T6SS core components (Boyer et al., 2009), are not essential for T6SS1 functionality. Indeed, our previous work showed that vp1388 and vp1389 encode a T6SS1 effector/immunity pair, and that they are not required for T6SS1 activity (Salomon, Kinch, et al., 2014). Moreover, vp1417‐vp1420 appear to be diverged duplications of the immunity gene vp1416, which is required to provide immunity against the T6SS1 effector VP1415 (Salomon, Kinch, et al., 2014). Therefore, it is possible that vp1418‐vp1420 are not required to provide immunity against self‐intoxication by VP1415, but rather, are used to provide immunity against divergent effectors from other Vibrio strains, which would make them expendable under the conditions of our BaCoF assay.

Two transposon insertion hits in our BaCoF assay revealed that vp1028, known as tfoY, is a required positive regulator of T6SS1. Previous work identified the TfoY homologue in V. cholerae as a positive regulator of T6SS. However, because the cues that activate the T6SSs, as do the T6SSs themselves, differ between V. cholerae and V. parahaemolyticus, we continued to investigate the role of TfoY in the regulation of V. parahaemolyticus T6SS1. Indeed, we found that the effects of TfoY and of its paralogue, TfoX, are quite different in these two vibrios. Whereas TfoY played a positive role in both T6SSs, TfoX did not seem to play a positive role in the regulation of V. parahaemolyticus T6SS1. On the contrary, TfoX overexpression resulted in diminished levels of T6SS1 activity. This negative effect of TfoX on V. parahaemolyticus T6SS1 can be explained by a previous observation made by Borgeaud and colleagues, who found that the overexpression of TfoX in V. cholerae resulted in downregulation of TfoY (Borgeaud et al., 2015), which is evidently essential for proper activation of V. parahaemolyticus T6SS1.

Our findings that the overexpression of TfoY bypasses the need for surface sensing and salinity‐mediated induction of T6SS1 suggest that under natural conditions surface sensing and salinity activate TfoY, which in turn, activates V. parahaemolyticus T6SS1. Notably, because overexpression of TfoY under low temperature conditions (23°C) only partially activated T6SS1, it is possible that temperature regulates T6SS1 via additional components or regulators besides TfoY. Similar to TfoY‐mediated activation of T6SS1 by surface sensing, TfoXVc was shown to be activated and induce the V. cholerae T6SS in response to the external inducer chitin (Borgeaud et al., 2015). Thus, it seems that the major positive activator, as well as the extracellular cue, differ between V. cholerae T6SS and V. parahaemolyticus T6SS1. Nevertheless, it is interesting that the paralogues TfoY and TfoX play a major role in activating antibacterial T6SSs in these two Vibrio species. As both TfoX and TfoY are highly conserved in vibrios (Pollack‐Berti et al., 2010), it is plausible that they also control the activity of other Vibrio T6SSs. Interestingly, even though TfoY is a central positive regulator of V. parahaemolyticus T6SS1, which is induced by surface sensing, apparently it can be bypassed and the system can also be activated by a different pathway. A double deletion of tfoY and the negative regulator opaR (ΔtfoY/ΔopaR) exhibited intermediate T6SS1 activity. Thus, it is possible that under certain conditions, quorum‐sensing regulation can control T6SS1 activity independent of surface sensing and TfoY.

Using epistasis experiments, we further demonstrated the role of TfoY and of previously identified positive and negative regulators in the T6SS1 regulatory network. As summarized in Fig. 6, we propose a model in which, upon sensing the T6SS1‐inducing conditions and cues (surface sensing, 3% NaCl, and 30°C), TfoY is activated and consequently activates the two positive regulators encoded within the T6SS1 gene cluster, VP1391 and VP1407. These two positive regulators are essential for the activation of T6SS1, because they are required for TfoY‐mediated activation and also for de‐repression of the system, which occurs upon removal of the negative regulation of H‐NS or OpaR. The repression mediated by H‐NS and OpaR, however, differs, because de‐repression by removal of H‐NS requires TfoY, whereas that is not the case when OpaR is removed. Therefore, H‐NS probably exerts its repression upstream of TfoY, whereas OpaR acts downstream of TfoY to repress T6SS1 activity. This result also implies the existence of a quorum‐sensing dependent, surface sensing/TfoY‐independent pathway for T6SS1 activation. In support of this model, OpaR was previously suggested to directly bind to promoter regions of V. parahaemolyticus T6SS1 operons, including the promoter of the operon encoding the positive downstream regulator, VP1407 (Zhang et al., 2017).

Figure 6

Model of the V. parahaemolyticus T6SS1 regulatory network. T6SS1 is induced under warm marine‐like conditions and upon surface sensing (via inhibition of the polar flagella). This leads to activation of TfoY, which in turn, activates VP1391 and VP1407 (positive regulators encoded within the T6SS1 cluster) that turn on the antibacterial T6SS1. The quorum‐sensing master regulator, OpaR, represses T6SS1 downstream of TfoY, possibly by direct binding to promoters of T6SS1 cluster operons, whereas H‐NS represses the system upstream of TfoY. Dashed arrows denote links that may be indirect.

Note that a previous study also implicated ToxR and AphA in the regulation of V. parahaemolyticus T6SS1 (Zhang et al., 2017); however, we did not include them in our epistasis analyses because we could not detect any effect of these regulators on T6SS1 activity in our previous work (i.e. their deletion had no negative effect on T6SS1‐mediated secretion or antibacterial activity) (Salomon, Klimko, et al., 2014). The discrepancy between our previous results and those of Zhang et al. could stem from the fact that experiments performed by Zhang et al. were conducted under T6SS1 non‐inducing conditions (i.e. in the absence of surface sensing activation and at 37°C; conditions under which we previously demonstrated that T6SS1 is inactive (Salomon et al., 2013; Salomon, Klimko, et al., 2014). In contrast, our previous work, as well as the current one, examined the effect of regulators under T6SS1‐inducing conditions.

Another BaCoF hit that had a dramatic effect on T6SS1 activity was vp2049, which encodes Tmk. Although the transposon mutant also exhibited a growth defect, we were able to distinguish the effect on growth from the effect on protein secretion by showing that the negative effect on T6SS1 activation was not evident for T6SS2. Whereas Tmk was required for proper T6SS1 activation, it does not seem to play an active, positive role in activation, because its overexpression did not bypass the need for surface sensing‐mediated induction. The link between Tmk (or the products of its activity) and T6SS1 activation remains unknown, and we will investigate it further in future studies.

Finally, the BaCoF methodology, used here to identify positive regulators and components of V. parahaemolyticus T6SS1, can be easily adapted to other systems. To use BaCoF, one needs a system in which conditions that activate an antibacterial T6SS (or any other antibacterial system) are known, and where a suitable sensitive prey that can grow under these conditions is available and can be fluorescently tagged. BaCoF can also be used to identify negative regulators of an antibacterial system by searching for mutants in which the antibacterial system becomes active under otherwise non‐inducing conditions. We are currently applying BaCoF to map regulatory networks of other Vibrio T6SSs.

Experimental procedures

Bacterial strains and media

Vibrio parahaemolyticus RIMD 2210633 derivative POR1 (ΔtdhAS) (Park et al., 2004), used here as the parental strain, and its derivatives, were routinely grown in Marine Lysogeny broth (MLB; LB broth supplemented with NaCl to a final concentration of 3% w/v) or on Marine Minimal Media (MMM) agar (2% w/v NaCl, 0.4% w/v galactose, 5 mM MgSO4, 5 mM K2SO4, 77 mM K2HPO4, 35 mM KH2PO4, 20 mM NH4Cl, 1.5% w/v agar) at 30°C. To induce the expression of genes from a plasmid, 0.1% (w/v) L‐arabinose was included in the media. Escherichia coli strain DH5α was used for plasmid maintenance and amplification, and as prey in competition assays (see below). E. coli strain S17‐1 (λ pir) was used for maintenance of pDM4 plasmids, mating and maintenance of isolated transposon‐containing genomic regions (see below). E. coli was routinely grown in 2xYT broth (1.6% w/v tryptone, 1% w/v yeast extract and 0.5% w/v NaCl) at 37°C. When necessary, media were supplemented with 30 μg ml−1 (for E. coli) or 250 μg ml−1 (for V. parahaemolyticus) kanamycin, and 10 μg ml−1 chloramphenicol. For a list of bacterial strains used in this study, see Supporting Information Table S1.

Plasmids

For arabinose‐inducible expression in bacteria, the coding sequences of TfoY (vp1028) and TfoX (vp1241) were amplified from V. parahaemolyticus RIMD 2210633 genomic DNA and inserted into the multiple cloning site of the pBAD/Myc–His vector (Invitrogen) harbouring a kanamycin‐resistance cassette (Salomon et al., 2013) in‐frame with the C‐terminal Myc‐6xHis tag (producing pTfoY and pTfoX respectively). pEVS104 conjugative helper and pEVS170 mini‐Tn5 delivery vector were a generous gift from Prof. Eric V. Stabb (Lyell et al., 2008). For a list of plasmids used in this study, see Supporting Information Table S1. For a list of primers used in this study, see Supporting Information Table S2.

Construction of deletion strains

For in‐frame deletions of tfoY (vp1028) and tfoX (vp1241), 1 kb sequences directly upstream and downstream of each gene were cloned into pDM4, a CmROriR6K suicide plasmid (O'Toole et al., 1996). These pDM4 constructs were inserted into V. parahaemolyticus via conjugation by S17‐1(λ pir) E. coli. Transconjugants were selected on MMM agar plates containing chloramphenicol. The resulting transconjugants were plated onto MMM agar plates containing 15% (w/v) sucrose for counter selection and loss of the sacB‐containing pDM4. Deletions were confirmed by PCR. The generation of in‐frame deletions of hcp1 (vp1393), vp1391, vp1407, hns (vp1133) and opaR (vp2516) were described previously (Salomon et al., 2013; Salomon, Klimko, et al., 2014).

BaCoF screen

A mini‐Tn5 transposon containing OriR6K and erythromycin resistance (ErmR) found on the pEVS170 mini‐Tn5 delivery vector was conjugated into the V. parahaemolyticus POR1 parental attacker. Transconjugates containing a chromosomally integrated transposon were selected on MMM plates containing Erm (100 μg ml−1). Single insertion mutant colonies, as well as the parental POR1 (T6SS1+) and the POR1/Δhcp1 derivative (T6SS1−) (Salomon et al., 2013), were picked into 96‐well plates containing 100 μl MLB per well. Plates were incubated overnight at 30°C. The following day, 40 μl of overnight‐grown T6SS1‐sensitive prey V. parahaemolyticus POR1/Δvp1415‐6 (Salomon, Kinch, et al., 2014) containing pGFP [a high copy number plasmid that is stably maintained without selection and constitutively expresses GFP from the lac promoter (Ritchie et al., 2012)] were added to each well. Attacker:prey mixtures were spotted onto MLB agar plates using a stainless steel pin replicator, and plates were left open for 10 min to dry. The MLB agar plates were then incubated overnight at 30°C and visualized the following morning using a Fusion FX6 imaging system (Vilber) equipped with a RGB excitation module and a GFP emission filter to identify GFP‐positive spots. When spots containing GFP fluorescence were identified (indicating the survival and growth of the T6SS1‐sensitive prey), 5 μl of the attacker:prey mixtures from the wells in the 96‐well plate that corresponded to the GFP‐positive spots were spread on MMM plates containing Erm (100 μg ml−1) to isolate the insertion mutant attackers. The location at which the transposon was inserted into these hits was determined as previously described (Lyell et al., 2008). Briefly, chromosomal DNA was isolated from insertion mutant hits, digested with HhaI (which does not cut within the transposon), and then was self‐ligated. Ligations were transformed into E. coli S17‐1 (λ pir) and selected on LB agar plates containing Erm (100 μg ml−1). Circular transposons containing the flanking chromosomal DNA were isolated, and transposon chromosomal DNA junctions were sequenced using the M13 forward primer.

Bacterial competition assays

Quantitative bacterial competition assays were performed as previously described (Salomon et al., 2013). Prey strains, either E. coli DH5α or V. parahaemolyticus derivatives, harboured a pBAD33 plasmid to provide selectable resistance against chloramphenicol. Assays were repeated at least three times with similar results. Results from a representative experiment are shown.

Secretion assays

Vibrio parahaemolyticus secretion assays for T6SS1 and T6SS2 were performed as previously described (Salomon et al., 2013). Briefly, for T6SS1 secretion, 5 ml of bacterial culture at an initial OD600 = 0.18 in MLB media was incubated for 5 h at 30°C in the presence or absence of 20 μM phenamil. For T6SS2 secretion, 5 ml of bacterial culture at an initial OD600 = 0.9 in LB media was incubated for 5 h at 23°C. For maintaining plasmids, appropriate antibiotics were added to the media. To induce expression from pBAD vectors, 0.1% (w/v) L‐arabinose was added to the media.

For expression fractions (cells) 1.0 OD600 units of cells were collected and re‐suspended in 100 μl of Tris‐glycine SDS sample buffer x2 (Novex, Life Sciences). Supernatants of volumes equivalent to 10 OD600 units were filtered (0.22 μm) and precipitated with deoxycholate and trichloroacetic acid (Bensadoun and Weinstein, 1976). Precipitated proteins were washed twice with ice‐cold acetone prior to re‐suspension in 20 μl of 10 mM Tris–HCl pH = 8.0, followed by the addition of 20 μl of 2× protein sample buffer. Expression and secretion samples were resolved on TGX stain‐free gels (Bio‐Rad), transferred onto PVDF or nitrocellulose membranes, and immunoblotted with custom‐made α‐VgrG1 (for T6SS1) or α‐Hcp2 (for T6SS2) polyclonal antibodies (Li et al., 2017). Loading of total protein lysates was visualized by analysis of trihalo compounds' fluorescence of the immunoblot membrane. Experiments were performed at least three times with similar results.

Bacterial growth

Overnight‐grown cultures of V. parahaemolyticus were normalized to an OD600 = 0.01 in MLB media and transferred to 96‐well plates (200 μl per well). For each experiment, n = 3 or n = 6, as indicated. Cultures were grown at 30°C in a BioTek EPOCH2 or SYNERGY H1 microplate reader with continuous shaking at 205 cpm. OD600 readings were acquired every 10 min. When the cultures contained pBAD expression vectors, kanamycin (250 μg ml−1) and L‐arabinose (0.1% w/v) were added. Experiments were performed at least three times with similar results. Results from a representative experiment are shown.

Conflict of interest

The authors declare that they have no conflict of interest.

Supplementary Figure S1. BaCoF hits display varying degrees of T6SS2 activity. Expression (cells) and secretion (media) of Hcp2 were detected by immunoblotting using specific antibodies against Hcp2. Loading control (LC), visualized as trihalo compounds' fluorescence of the immunoblot membrane, is shown for total protein lysates.

Click here for additional data file.

Supplementary Figure S2. TfoY restores antibacterial activity to Viability counts of DH5α prey before (0 h) and after (4 h) co‐culture with the indicated attackers on MLB media (containing 3% NaCl) supplemented with 0.1% L‐arabinose at 30°C. The attackers harboured an arabinose‐inducible expression vector, either empty (pEmpty) or encoding TfoY (pTfoY). Asterisks mark statistical significance between samples and POR1 + pEmpty parental attacker at 4 h timepoint by unpaired, two‐tailed Student's t‐test (* P < 0.05).

Click here for additional data file.

Supplementary Figure S3. TfoY overexpression bypasses the required external conditions for T6SS1 activation. Expression (cells) and secretion (media) of VgrG1 were detected by immunoblotting using specific antibodies against VgrG1. Loading control (LC), visualized as trihalo compounds' fluorescence of the immunoblot membrane, is shown for total protein lysates. As indicated, the strains contained arabinose‐inducible expression vectors of TfoY (pTfoY) or an empty vector (pEmpty).

Click here for additional data file.

Supplementary Figure S4. Deletions of Growth of V. parahaemolyticus strains in MLB at 30°C is shown as OD600 measurements. Data are mean ± SD, n = 6.

Click here for additional data file.

Supplementary Figure S5. Overexpression of VP2049 restores T6SS2 secretion levels. Expression (cells) and secretion (media) of Hcp2 were detected by immunoblotting using specific antibodies against Hcp2. Loading control (LC), visualized as trihalo compounds' fluorescence of the immunoblot membrane, is shown for total protein lysates.

Click here for additional data file.

Supplementary Figure S6. Overexpression of VP2049 does not bypass the surface sensing requirement for T6SS1 activation. Expression (cells) and secretion (media) of VgrG1 were detected by immunoblotting using specific antibodies against VgrG1. Loading control (LC), visualized as trihalo compounds' fluorescence of the immunoblot membrane, is shown for total protein lysates. Strains contained an arabinose‐inducible expression vector of VP2049 (pVP2049) or an empty vector (pEmpty).

Click here for additional data file.

Supplementary Figure S7. Deletions of T6SS1 regulators do not affect growth. The growth of V. parahaemolyticus strains in MLB at 30°C are shown as OD600 measurements. Data are mean ± SD, n = 6.

Click here for additional data file.

Supplementary Table S1. List of bacterial strains and plasmids used in this study.

Supplementary Table S2. List of primers used in this study.

Click here for additional data file.