Divergent Control of Two Type VI Secretion Systems by RpoN in Pseudomonas aeruginosa

Three Type VI Secretion System (T6SS) loci called H1- to H3-T6SS coexist in Pseudomonas aeruginosa. H1-T6SS targets prokaryotic cells whereas H2-T6SS mediates interactions with both eukaryotic and prokaryotic host cells. Little is known about the third system, except that it may be connected to H2-T6SS during the host infection. Here we show that H3-T6SS is required for P. aeruginosa PAO1 virulence in the worm model. We demonstrate that the two putative H3-T6SS operons, called “left” and “right”, are coregulated with H2-T6SS by the Las and Rhl Quorum Sensing systems. Interestingly, the RpoN σ54 factor has divergent effects on the three operons. As for many T6SSs, RpoN activates the expression of H3-T6SS left. However, RpoN unexpectedly represses the expression of H3-T6SS right and also H2-T6SS. Sfa2 and Sfa3 are putative enhancer binding proteins encoded on H2-T6SS and H3-T6SS left. In other T6SSs EBPs can act as σ54 activators to promote T6SS transcription. Strikingly, we found that the RpoN effects of H3-T6SS are Sfa-independent while the RpoN mediated repression of H2-T6SS is Sfa2-dependent. This is the first example of RpoN repression of a T6SS being mediated by a T6SS-encoded EBP.


Introduction
Protein secretion is an essential for host colonization by pathogenic bacteria. Multiple systems have evolved in order to secrete proteins into the extracellular medium or directly into target cells [1]. The most recently described system, the Type Six Secretion System (T6SS), was first discovered in Vibrio cholerae [2] and in Pseudomonas aeruginosa [3]. A unique feature of T6SSs is their capacity to deliver toxic proteins into eukaryotic host cells as well as into bacteria [4,5]. These systems were originally thought of as virulence determinants towards eukaryotic host cells [6,7], although now they have also been shown to be unambiguously involved in interbacterial interactions and competition [8]. Nevertheless, the anti-prokaryotic T6SSs may also facilitate the colonization of specific niches where pathogens can then express virulence towards eukaryotic cells. Another striking feature of T6SSs is that multiple distinct T6SS loci are often present in a single genome. For example, the genomes of Burkholderia pseudomallei [9,10] and Yersinia pseudotuberculosis [11] harbor six and four T6SSs respectively. The various T6SS within a single strain may serve different functions and/or be differentially regulated. In Burkholderia thailandensis, the T6SS-5 was shown to be required for virulence in a murine melioidosis model, while inactivation of T6SS-1 rendered B. thailandensis more susceptible to contact with other bacteria [12].
The genome of the P. aeruginosa contains three T6SS loci, called H1-to H3-T6SS [3,7]. H1-T6SS has been widely studied and was shown to deliver three bacteriolytic toxins to the periplasm of target bacteria [8,13]. It may give P. aeruginosa a survival benefit in a multi-bacterial environment. However, less is known about the two other T6SSs. The H2-T6SS locus of the PAO1 strain of P. aeruginosa promotes bacterial internalization into epithelial cells, and indeed plays a role in virulence in the worm model [14]. H2-T6SS also mediates interbacterial competition through Tle5/PldA [15], a phospholipase D which was previously shown to contribute to P. aeruginosa persistence in a chronic pulmonary infection model [16]. As with many P. aeruginosa virulence factors, the H2-T6SS machinery is controlled by Quorum-Sensing (QS) and iron availability in the environment [14]. H2-T6SS and H3-T6SS loci in the PA14 strain of P. aeruginosa differ from their counterparts in the PAO1 strain by the absence of three and one putative effector genes [17]. PA14 H2-T6SS and H3-T6SS are both required for virulence in the plant model Arabidopsis thaliana. In a mouse model of acute infection, while a H2-T6SS mutant was affected, a H3-T6SS mutant was as virulent as the Wild-Type (WT) strain. Interestingly, the double H2-and H3-T6SS mutant exhibited dramatically reduced virulence, this suggesting compensation in vivo between the two systems [17].
In this study, we aimed at determining whether the H2-T6SS and H3-T6SS loci of PAO1 are related at a transcriptional level. We observed that, like H2-T6SS, the two H3-T6SS gene clusters are activated by Quorum Sensing, but are under divergent control by RpoN.

Bacterial Strains, Plasmids, and Growth Conditions
The bacterial strains and plasmids used in this study are described in Table 1. LB and TSB broths and agar were used for the growth of P. aeruginosa and Escherichia coli strains at 37uC. Cultures were inoculated at an optical density at 600 nm (OD 600 ) of 0.1 with overnight cultures, and strains were grown at 30, 37 or 42uC with aeration in TSB. Recombinant plasmids were introduced into P. aeruginosa using the conjugative properties of pRK2013 (Table 1) or by electroporation. Pseudomonas transconjugants were selected on Pseudomonas isolation agar (PIA, Difco Laboratories) supplemented with appropriate antibiotics. The antibiotic concentrations were as follows: for E. coli, ampicillin (50 mg ml -1 ), kanamycin (25 mg ml -1 ), tetracycline (15 mg ml -1 ), gentamicin (10 mg ml -1 ); for P. aeruginosa, tetracycline (200 mg ml -1 for plates or 50 mg ml -1 for liquid growth), gentamicin (50 mg ml -1 ), carbenicillin (500 mg ml -1 ).

lacZ Reporter Fusion and b-galactosidase Assay
The H3-T6SS left-lacZ and H3-T6SS right-lacZ transcriptional fusions were constructed by PCR amplification of respectively 486 and 494 bp upstream DNA region from the lip3 or hsiB3 gene by using TSO15/TSO16 and TSO17/TSO18 primers (Table 1). PCR amplification products were directly cloned into the pMini-CTX::lacZ vector [18], yielding pTS12 and pTS13, in pCR2.1, and pTS19 and pTS20 in MiniCTX-'lacZ. Nucleotide sequences were verified by sequencing (GATC). The promoter fragment was integrated at the CTX phage attachment site in PAO1 and isogenic mutants using established protocols [18].
Overnight culture, grown in TSB, was diluted in TSB to OD 600 = 0.1. Growth and ß-galactosidase activity were monitored by harvesting samples at different time intervals. ß-galactosidase activity was measured according Miller [19], based on onitrophenyl-b-D-galactopyranoside hydrolysis. ß-galactosidase activities were expressed in Miller units.

Construction of the DclpV2DcplV3 Mutant
To generate the DclpV2cplV3 mutant, the pTS27 mutator plasmid [14] was mobilized in the P. aeruginosa strain PAO1DclpV3 [20]. Mutants which had undergone a double recombination event, resulting in the non-polar deletion of the clpV2 gene, were verified by PCR with the primers OA14 and OA17 that flank clpV2.

Construction of sfa2 and sfa3 Mutants
To generate sfa2 and sfa3 mutants, internal fragments of 490 and 511 bp were respectively amplified with TSO118-TSO119 and TSO120-TSO121 and cloned in the pCR2.1, resulting in pSBC56 and pSBC57. The mutator plasmids were electroporated into P. aeruginosa PAO1 and the mutant bacteria selected on PIA medium containing Carbenicillin. The insertions were verified by PCR with the primer pairs TSO39-TSO40-TSO45 and TSO41-TSO42-TSO46 that hybridize outside and inside of the sfa2 and sfa3 genes.

Caenorhabditis elegans Killing Assay
The slow killing assay was performed as described previously [14]. Each independent assay consisted of three replicates. E. coli OP50 was used as a control. L4 to adult stage C. elegans were removed from food and placed on unseeded NGM plates for 24 hours at 25uC. 50 worms were then picked onto plates containing overnight grown bacteria. Worms were evaluated for viability on a daily basis. Animal survival was plotted using the PRISM 5.0 program. Survival curves are considered significantly different from the control when P-values are ,0.05. Prism calculates survival fractions using the product limit (Kaplan-Meier) method. Prism compares survival curves by two methods: the log-rank test (also called the Mantel-Cox test) and the Gehan-Breslow-Wilcoxon test.

Results
H2 and H3-T6SS are Both Involved in Virulence Towards C. elegans Functional compensation between PA14 H2-and H3-T6SS has been observed in a lung infection model [17]. We previously showed that the H2-T6SS of PAO1 mediates virulence in a worm model [14]. Therefore, we wondered whether the H3-T6SS of PAO1 is also required for virulence. We indeed found decreased virulence in a PAO1 H3-T6SS mutant, which harbors a deletion of the clpV3 gene ( Fig. 1). clpV3 encodes the ClpV AAA + ATPase [20], a core component of the secretion machinery (Fig. 1). The worms infected with the mutant appeared to die with a 2 day delay when compared to the WT strain. H3-T6SS is thus necessary for killing C. elegans, however to a lesser extent than H2-T6SS [14]. In contrast with the findings for PA14 [17], the virulence of the double mutant DclpV2DclpV3 was not lower than in the single mutants. Both T6SS are thus required for virulence in PAO1.   [14]). B. H3-T6SS is organized in two putative operons. The genes are labeled hsiA3 to hsiJ3 for the left operon and hsiB3 to hsiH3 for the right operon and, where applicable, with the given name, i.e., clpV3 or sfa3. Gene annotation numbers are also indicated (e.g. PA2359). The promoter region of each operon is also shown. C. The intergenic sequence between lip3 and hsiB3 genes is represented. The -35 box and the -10 box of the s 70 promoters predicted by Bprom are highlighted in green and red respectively. The translation initiation codons of lip3 and hsiB3 genes are underlined. 486 bp of the left operon upstream region and 494 bp of the right operon upstream region were used for the transcriptional fusions, which are encoded by pTS19 and pTS20 respectively. doi:10.1371/journal.pone.0076030.g002 H2-T6SS and H3-T6SS are Coregulated by Quorum-Sensing While the H1-T6SS is not expressed in a PAO1 WT background [3,21], H2-T6SS expression is controlled by QS in a cell-density dependent manner [14] (Fig. 2A). We therefore investigated the transcriptional regulation of the H3-T6SS gene cluster in PAO1. The H3-T6SS locus is organized into two divergent gene clusters (Fig. 2B), and we analyzed the intergenic DNA region between lip3 (PA2364) and hsiB3 (PA2365) for potential regulatory elements (Fig. 2C). The BProm algorithm identified one s 70 dependent promoter upstream of the lip3 gene and another, in the opposite direction, upstream of the hsiB3 gene (http://linux1.softberry.com/berry.phtml?topic = bprom&group = programs&subgroup = gfindb) (Fig. 2C). To construct chromosomal transcriptional lacZ fusions, the regions upstream of the ATG of lip3 and hsiB3, the first gene of each of the two H3-T6SS clusters respectively (hereinafter called ''left'' and ''right operons'' for simplicity) were fused to lacZ and then integrated at the CTX phage attB site as a single copy on the chromosome, yielding strains PAO1TS19 and PAO1TS20 respectively ( Table 1). Strain PAO1Z was similarly constructed by integrating a promoter-less lacZ gene to serve as a negative control. The b-galactosidase activity profiles associated with the two H3-T6SS transcriptional fusions were similar. Expression was induced at the transition from log-tostationary phase and a maximal level was reached late in stationary phase, after 9 hours of growth ( Fig. 3A and Fig. S1). The expression then stayed and maximal for at least 4 more hours. No promoter activity was detectable in the control strain PAO1Z. Moreover, H3-T6SS was clearly more expressed at 37uC than at the other tested temperatures, 30uC and 42uC (Fig. S1). As previously observed [14], H2-T6SS expression was also induced at the transition phase, but a maximal and steady level of expression was reached earlier in stationary phase (Fig. 3A and Fig. S1).
The cell density dependent expression profile of the H3-T6SS reporter fusions suggested that the transcription of the H3-T6SS might be regulated by QS. This would be in line with the LasRmediated regulation of hcp3 in the H3-T6SS right operon of PA14 [17], and with two transcriptomic studies indicating de-regulated expression of genes in the right H3-T6SS operon of PAO1 in QS mutants [22,23]. We therefore examined the expression of the two H3-T6SS fusions in P. aeruginosa QS mutants, a lasR mutant (PAOR) [24] and a rhlI mutant (PDO100) [25]. Compared to the WT PAO1 strain (Fig. 3B & 3C), the expression of the left and right H3-T6SS operons was significantly decreased in the lasR mutant (2.7-fold and 3.3-fold respectively) and in the rhlI mutant (2.4-fold and 2.7-fold respectively). In conclusion, H3-T6SS is coregulated with H2-T6SS by the Las and Rhl QS systems.

Divergent Effects of RpoN on H2-and H3-T6SS Gene Expression
Several reports in the literature have shown that T6SS transcriptional activity requires the sigma factor RpoN (s54) and cognate activators encoded within the T6SS operon. In the initial genetic screen that led to the discovery of T6SS genes in V. cholerae [2], one of the attenuated mutants in virulence towards Dictyostelium was in the vasH gene which encodes a s54-activator. Similarly, in Aeromonas hydrophila, a VasH homologue was shown to be required for cytotoxicity towards macrophages and epithelial cells since it was found to be essential for the expression of the genes encoding the T6SS machinery [26]. In agreement, Bernard and coworkers demonstrated, in a reconstituted heterologous system, that s54-activators from various T6SSs together with the E. coli s54-RNAP (RNA polymerase) complex allowed expression of T6SS genes [27].
We thus monitored expression of the H2-T6SS and H3-T6SS lacZ fusions in the rpoN mutant (Table 1) at the entry to stationary phase (Fig. 4). The rpoN mutation is complemented (PAO6360) or not (PAO6358) on the chromosome [28]. Expression of the left H3-T6SS operon decreased 2.0-fold in the rpoN mutant, and complementation of the mutation restored WT expression levels (Fig. 4B). This means that expression of the left H3-T6SS operon depends on RpoN, and may require a s54-activator as observed in for other T6SSs. Unexpectedly, H2-T6SS and the right H3-T6SS operons were overexpressed in the rpoN mutant (4.5-fold and 4.7-fold increase respectively). Expression was restored to WT levels upon complementation (Fig. 4A & 4C). This suggests that RpoN mediates the transcriptional repression of these two operons. In conclusion, RpoN has divergent roles in regulating P. aeruginosa T6SS gene expression.

Sfa2 Decreases H2-T6SS Expression whereas H3-T6SS Expression is Sfa-independent
In P. aeruginosa, the sfa2 and sfa3 genes (sigma factor activator) from the H2-and H3-T6SS loci ( Fig. 2A & 2B) encode putative RpoN activators also called EBPs (enhancer binding protein) [7]. As for other EBPs, Sfa2 and Sfa3 contain two Walker A and B motifs that have roles in nucleotide binding and hydrolysis, and the highly conserved ''GAFTGA'' domain that is indispensable for the nucleotide-dependent interactions with s54-RNAP that drives open complex formation and transcription (Fig. S2) [29,30]. We thus wondered if Sfa2 and Sfa3 proteins could be involved in H2and H3-T6SS regulation. To test this hypothesis, sfa2 and sfa3 mutants were constructed and expression of the three lacZ fusions were assayed in the mutant backgrounds. While mutations in sfa2 and sfa3 had no effect on H3-T6SS expression levels (Fig. 5A &  5B), H2-T6SS expression was increased in the sfa2 mutant (3.7fold). Expression could be restored to WT levels upon complementation in trans with a WT copy of sfa2 (Fig. 5C). Moreover Sfa3 has no effect on H2-T6SS expression (Fig. 5C). Taken together, these data suggest that H3-T6SS expression is Sfa-independent and that the H2-T6SS operon may be indirectly repressed by RpoN in a Sfa2-dependent manner.

Sfa2 Contributes to Repression of H2-T6SS by RpoN
To demonstrate that the RpoN repression of H2-T6SS is mediated by Sfa2, we hypothesized that overproduction of Sfa2 should have no effect in a rpoN mutant while it should decrease expression in a WT background. We thus monitored H2-T6SS expression upon Sfa2 overproduction in exponential phase in these two backgrounds. We chose to probe expression at this particular moment of growth to focus on the effect of the overproduced Sfa2 and not of the chromosomal sfa2 gene, which not yet fully expressed at this time. In support of our hypothesis we indeed found that Sfa2 overproduction led to the decreased expression of H2-T6SS in the WT, and had no effect in the rpoN background (Fig. 6). Hence Sfa2 mediates to the indirect repression of H2-T6SS by RpoN.

Discussion
The P. aeruginosa genome harbors three T6SS clusters. The H1-T6SS of the PAO1 strain targets toxins into host bacteria [8,13] and H2-T6SS is involved in interactions with eukaryotic and prokaryotic hosts [14,15]. The role of H3-T6SS has been mainly studied in the PA14 strain of P. aeruginosa, which appears to differ from PAO1 at the H2-and H3-T6SS gene level. The PA14 H3-T6SS is required for virulence in the plant model A. thaliana and may compensate for the loss of H2-T6SS in mouse virulence because only the double H2-T6SS H3-T6SS mutant is less virulent than the WT strain [17]. Here, we have shown that H3-T6SS is required for P. aeruginosa PAO1 virulence towards worms. However, we did not observe any compensation between the two T6SSs, at least in the worm model. But the clpV3 mutation phenotype is dominant on clpV2 since the double mutant has the same virulence defect as the clpV3 mutant this suggesting a genetic interaction. Moreover, we found that RpoN divergently regulates expression of PAO1 T6SSs, by repressing H2-T6SS and the right H3-T6SS and activating the left H3-T6SS. Interestingly solely Sfa2, the s54 activator encoded within the H2-T6SS cluster, participates in RpoN-control. This is the first example of a RpoNrepression mechanism mediated by a T6SS-encoded activator.
We observed that like H2-T6SS [14], the two H3-T6SS operons of the PAO1 strain are induced at the growth phase transition by LasR and RhlR QS regulators (Fig. 3). This is in line with data establishing a differential regulation of the three PA14 T6SS loci by LasR [17,31]. We also noticed that the two H3-T6SS operons are fully expressed late in stationary phase (Fig. 3A), suggesting induction by the RpoS sigma factor. In agreement, a transcriptomic study showed that the right H3-T6SS operon is strongly dependent on RpoS [32], however the authors did not identify the heptameric consensus sequence CTATACT defined as the -10 of RpoS-controlled promoters [33] upstream of the right H3-T6SS operon. Interestingly, the right H3-T6SS operon also belongs to a group of genes that are induced by contact with eukaryotic cells [34]. This is in agreement with our data on the involvement of H3-T6SS in P. aeruginosa virulence in a eukaryotic model (Fig. 1). Furthermore, we also observed that H3-T6SS is more highly expressed at 37uC, the temperature of certain eukaryotic hosts (Fig. S1). Finally, the H3-T6SS from PAO1 was recently shown to be dispensable for bacterial competition (Russell 2013). All together, and in contrast to H2-T6SS, the H3-T6SS machinery in PAO1 appears to be exclusively dedicated to interactions with eukaryotic cells.
We also showed a complex RpoN-control of T6SS expression. Indeed H2-T6SS and the right H3-T6SS operons are unexpectedly under a negative control by RpoN, to the contrary of other T6SSs (for a review see [5]), (Fig. 4A & 4C). Moreover, the RpoNmediated repression of H2-T6SS requires Sfa2, the EBP encoded within H2-T6SS (Fig. 5C & 6). H3-T6SS right is Sfa-independent, although H3-T6SS left encodes an EBP, Sfa3 (Fig. 5B). We also showed that the H3-T6SS left operon is activated by RpoN, and independently of any Sfa (Fig. 5A). Hence the two H3-T6SS operons are divergently regulated by RpoN. This could fit with the recent observations of Dong and Mekalanos [35]. In V. cholerae, they observed that RpoN positively regulates the expression of the hcp operons and vgrG3 that encode Type VI secreted proteins, but has no effect on the expression of the main T6SS cluster encoding sheath and other structural components of the phage tail-like machinery [5,36]. This is presumably because the latter are recycled. The role of Sfa3 is also intriguing. It may be required for coregulation of substrate genes that are not part of the H3-T6SS locus. In P. aeruginosa, RpoN plays important roles in mobility, in the transport of nutrients, in the formation of pili, in mucoidy and in cell-to-cell signaling (for a review see [37]). As a sigma factor, RpoN positively regulates its target genes (e.g. type IV pili and flagellum genes), but has also been shown to negatively regulate QS [28], the expression of sadB, coding an important protein during biofilm formation [38], and of aceA, coding an isocitrate lyase, an enzyme required for the metabolic pathway utilized by P. aeruginosa during chronic pulmonary infections [39]. Unlike RpoNmediated activation, RpoN repression is indirect. In agreement we have been unable to identify RpoN consensus elements in the promoter regions of H2-and of the right H3-T6SS operons with Virtual footprint (http://www.prodoric.de/vfp/vfp_promoter. php), while a RpoN binding site was previously proposed for the H3 left operon that we found RpoN-activated [27] (Bernard 2011).
Taken together, our data allow us to propose a model in which the expression of H2-T6SS and H3-T6SS is induced by QS [14] at the transition between the exponential and stationary phase, with H3-T6SS reaching full expression later than H2-T6SS in the stationary phase. The H2-T6SS operon codes Sfa2, that once produced activates RpoN to repress its own expression. A first explanation in line with Heulier et al. [28] could have been that Sfa2 is the EBP which activates RpoN in order to repress QS, thus arresting H2-T6SS induction. However, we found that lasR and rhlR transcriptional lacZ fusions were not impacted upon Sfa2 overproduction (Fig. S3). We would thus like to propose that RpoN together with Sfa2 activates an unknown repressor of H2-T6SS. Repression of the H3-T6SS right operon might be explained by the RpoN/GacA/RsmA pathway [40]. Indeed, the H3-T6SS right operon is strongly activated by GacA and RsmYZ [41], while RpoN has a negative effect on gacA expression [28] and thus on expression of H3-T6SS right. Future studies will be required to decipher this network of regulation. Figure S1 H2-T6SS and H3-T6SS are differentially thermoregulated. The expression pattern of the H2-T6SS-lacZ (A), H3-T6SS left-lacZ (B), and H3-T6SS right-lacZ (C) transcriptional fusions in the WT PAO1 strain is given in Miller Units at different time points over the growth and at 3 different temperatures: 30uC (blue), 37uC (green) and 42uC (red). The OD 600 is also presented (diamonds). A control strain (PAO1Z) (grey squares) is included for each graph. Each experiment was done in triplicate and independently repeated three times; error bars indicate the standard deviation.