Intracellular Bacteria Encode Inhibitory SNARE-Like Proteins

Pathogens use diverse molecular machines to penetrate host cells and manipulate intracellular vesicular trafficking. Viruses employ glycoproteins, functionally and structurally similar to the SNARE proteins, to induce eukaryotic membrane fusion. Intracellular pathogens, on the other hand, need to block fusion of their infectious phagosomes with various endocytic compartments to escape from the degradative pathway. The molecular details concerning the mechanisms underlying this process are lacking. Using both an in vitro liposome fusion assay and a cellular assay, we showed that SNARE-like bacterial proteins block membrane fusion in eukaryotic cells by directly inhibiting SNARE-mediated membrane fusion. More specifically, we showed that IncA and IcmG/DotF, two SNARE-like proteins respectively expressed by Chlamydia and Legionella, inhibit the endocytic SNARE machinery. Furthermore, we identified that the SNARE-like motif present in these bacterial proteins encodes the inhibitory function. This finding suggests that SNARE-like motifs are capable of specifically manipulating membrane fusion in a wide variety of biological environments. Ultimately, this motif may have been selected during evolution because it is an efficient structural motif for modifying eukaryotic membrane fusion and thus contribute to pathogen survival.


Introduction
In eukaryotic cells, intracellular membrane fusion events are mediated by members of the SNARE protein family. SNAREs are conserved in all eukaryotes and are present on the surface of all secretory compartments [1,2,3]. During membrane fusion, t-SNAREs present on target organelles assemble into a four-helix bundle with the v-SNAREs present on vesicles. This event brings the membranes in which they are embedded into close apposition and drives bilayer fusion [4,5,6,7]. The SNARE residues indispensable for membrane fusion form the ''SNARE motif'' [6], a 60 amino-acid sequence composed of coiled-coil heptad repeats [ Table 1 and [8]]. Similar structural motifs are used for the same purpose by viruses, highlighting the general role of coiled coil sequences in manipulating membrane fusion [9,10]. Here we investigated whether this particular motif is also utilized by bacteria to influence eukaryotic membrane fusion.
Intracellular bacteria such as Salmonella, Mycobacterium, Legionella or Chlamydia must manipulate membrane fusion of the host cells they inhabit in order to escape lysosomal fusion [11,12]. While intracellular, these bacteria modify their infectious phagosomes, also called inclusions or vacuoles, by expressing their own proteins to the surface [13]. As a result, the infectious phagosomes become protected against fusion with endocytic compartments [14,15].
Although the precise mechanism is unclear, it is likely that the bacterial proteins expressed on the surface of these infectious phagosomes are responsible for blocking fusion with the endocytic compartments [16,17]. Interestingly, over the past few years a growing number of SNARE-like proteins have been identified notably in Chlamydia and Legionella [18,19,20], two intracellular bacteria responsible for human diseases. For instance, IncA, a protein expressed by Chlamydia on the surface of the infectious vacuole displays two SNARE-like motifs [19,21]. IncA interacts directly with mammalian SNAREs [21] and IncA expressed by Chlamydia trachomatis, CtrIncA, has been implicated in homotypic membrane fusion [22,23]. Formation of Chlamydia inclusions by homotypic fusion is an event specifically occurring during C. trachomatis infection. Interestingly, most Chlamydia strains express IncA, yet not all strains have the capacity to undergo homotypic fusion, suggesting that IncA likely plays additional roles.
Expressed by Legionella pneumophila, IcmG/DotF only displays one SNARE-like motif. The precise function of IcmG/DotF is still unclear, although mutants are rapidly trafficked to, and degraded within lysosomal compartments [24]. Using these bacterial SNARE-like proteins as our models, we tested their function on SNARE-mediated membrane fusion. We discovered a novel inhibitory function of these proteins and characterized the molecular mechanism they use to block host membrane fusion.

Results and Discussion
Intracellular bacteria primarily protect their vacuoles against endocytic fusion [12,25], which is mediated by the association of the v-SNARE VAMP8 with the endocytic t-SNARE composed of Syntaxin 7, Syntaxin 8 and Vti1b [26,27].
Two distinct complementary mechanisms have been suggested concerning chlamydial avoidance of lysosomal fusion: 1) during the first ,8 hours of infection, the protection of the vacuole appears to be independent of Chlamydia protein synthesis [28]. Rather, structural components of the Chlamydia cell wall seems to be involved in this activity [29]. 2) Later however, at a time that coincides with IncA expression [30], an active modification of the inclusion membrane takes place to sustain the protection of the inclusion. In light of these evidences, we started to investigate the role of IncA in the protection of the Chlamydia inclusion.
Previously, IncA has been shown to co-precipitate with the endocytic SNAREs when expressed in cells [21]. Using an in vitro liposome fusion assay [4,6], we now tested both CtrIncA and CcaIncA expressed respectively by C. trachomatis and C. caviae, for their functional effect on endocytic SNARE-mediated membrane fusion (see Table 2 for a description of all the SNARE proteins studied here). To do so, we reconstituted the t-SNARE [Syntaxin7/Syntaxin8/Vti1b] and the v-SNARE [VAMP8] with or without IncA into acceptor and donor liposomes, respectively. Donor liposomes contain the FRET pair Rhodamine-PE [N-(lissamine rhodamine B sulfonyl) phosphatidyl ethanolamine] and NBD-PE [N-(7-nitro-2,1,3-benzoxadiazole-4-yl) phosphatidyl ethanolamine]. Liposome fusion results in lipid mixing of donor and acceptor liposomes. As the distance between NBD and rhodamine increases, the resonance energy transfer and the quenching of NBD are reduced. Fusion becomes detectable as an increased NBD fluorescence at 538nm [4,6]. After mixing different combinations of t-and v-liposomes+/2IncA, liposome fusion was allowed to proceed at 37uC for two hours. As shown in Fig. 1, CtrIncA strongly inhibits endocytic SNARE-mediated fusion. CtrIncA blocks membrane fusion whether present in v-SNARE (,70% inhibition) or in t-SNARE (,37% inhibition) liposomes  ADES SLEST TRRM MLQLV VEES SKDAG GIRT TLVML LDEQ QGEQL LERI IEEGM MDQI INKDM MKEA AEKNL   hSNAP25-Cterm 140 DARE ENEM MDENL LEQV VSGII IGNL LRHMA ALDM MGNEI IDTQ QNRQI IDRI IMEKA ADSN NKTRI IDEA Table 1 and [18]], we also observed inhibition of SNARE-mediated membrane fusion ( Fig. 1). Interestingly, IcmG/DotF has no effect when present on the t-SNARE side (Fig. 1J), but only interferes with the v-SNARE (Fig. 1I). Although both IncA and IcmG/DotF have a common inhibitory function on membrane fusion, it appears that bacterial SNARE-like proteins display different levels of efficiency. Most likely, other SNARE-like proteins, such as LegC3, play a major role in protecting Legionella's vacuole [31]. Although CtrIncA has been previously implicated in homotypic membrane fusion [22,23], we did not observe any fusion events between CtrIncA-containing liposomes (data not shown). Perhaps CtrIncA requires post-translational modifications such as phosphorylation, to become fusogenic [32]. Consistent with this possibility, IncA has multiple phosphorylation sites that become phosphorylated by host cells during infection [19,32]. Alternatively, additional proteins from either Chlamydia or the host cell might be necessary in combination with CtrIncA to promote fusion. Interestingly, some non-fusogenic strains do express a normal IncA protein on the inclusion membrane, supporting the possibility that other elements of the fusion machinery are missing in these strains [33].
We propose that CtrIncA could function as a switch to regulate the maturation of the inclusion. During the infectious cycle of C. trachomatis, each newly synthesized CtrIncA would first bind every resident SNARE on the inclusion, until all are blocked. As a consequence, SNARE-mediated fusion of the inclusion would be totally inhibited. As CtrIncA continues to accumulate, excess CtrIncA would then be available for further modification by the host cell (phosphorylation) and/or for binding additional proteins. CtrIncA would become active for fusion and inclusions could then undergo homotypic fusion.
Next, we determined whether the inhibitory function was encoded into the SNARE-like motif. Since IcmG/DotF has a limited inhibitory effect, we concentrated our efforts on IncA. IncA possesses two SNARE-like motifs [21] (Table 1, Fig. 1A). We focused our attention on the N-terminal motif due to its presence next to the trans-membrane domain mimicking the eukaryotic SNARE configuration. This makes it ideally located to interact directly with eukaryotic SNARE motifs. Furthermore, this motif has previously been shown to be compatible with the formation of a stable complex with SNARE proteins [19]. To determine whether the N-terminal SNARE-like motif has an inhibitory activity, truncated forms of CtrIncA were generated and their effects on endocytic SNARE-mediated fusion were examined. CtrIncA mutant containing only the N-terminal SNARE-like motif (CtrIncA 1-141 ) inhibited endocytic SNARE-mediated fusion in a dose-dependent manner similar to the full-length protein ( Fig. 2A). Next, we delineated the minimal IncA sequence necessary to retain the inhibitory function. As shown on Fig. 2B and Fig. 2C, CtrIncA 1-130 still displays a significant inhibitory effect (,15%, p = 0.028) when present on either t-or v-SNARE membrane. On the contrary, CtrIncA 1-120 , which contains only half of the SNARE-like motif, completely lost its ability to inhibit endocytic SNARE-mediated membrane fusion (Fig. 2D, 2E, p = 0.42). The outcome was similar regardless of whether this truncated form of IncA was reconstituted into t-SNARE or v-SNARE liposomes. This suggests that the CtrIncA N-terminal SNARE-like motif requires a SNARE-like motif of at least ,23 amino acids in order to exert an effective inhibitory activity and confirm the role of this motif in blocking membrane fusion. Although the function of the C-terminal domain remains to be determined, we cannot exclude its role in reinforcing the inhibitory effect of IncA.
Since Chlamydia inclusion membrane is derived from the plasma membrane, we then decided to test the effect of both CtrIncA and CcaIncA on the plasma membrane resident exocytic t-SNAREs ( Table 2). As shown in Fig. 3A, CtrIncA has no inhibitory effect on any of the exocytic complexes tested, regardless of its concentration, suggesting that CtrIncA is specific for the endocytic SNAREs. CcaIncA, on the other hand, exerts a significant inhibitory effect on [Syn2/SNAP23], [Syn3/SNAP23] and [Syn4/SNAP23] fusion (Fig. 3B), demonstrating that CcaIncA has a broader inhibitory effect. These results further show that bacterial SNARElike proteins display different levels of specificity. One might imagine that the capacity to inhibit a large range of membrane fusion events could potentially increase the number of hosts that intracellular bacteria could infect. For example, C. caviae has been detected in a wide range of hosts [34]. Alternatively, blocking a multitude of vesicular trafficking in the cells could impact the longterm outcome of an infection. In particular it would be interesting to correlate the level of SNARE-like protein inhibition with the capacity of certain bacteria to induce chronic diseases.
To confirm SNARE-like proteins inhibitory function in a more physiological environment, we tested IncA's role in vivo in mammalian cells that can potentially host infection. In order to obtain quantitative results, we chose the RBL-2H3 mast cell line as our model. Mast cells display a large number of endocytic compartments, including their secretory granules, which are secretory lysosomes [35]. During stimulation, the endocytic v-SNAREs VAMP8 present on the secretory lysosomes bind the exocytic t-SNAREs [Syntaxin 4/SNAP23] present on the plasma membrane to mediate exocytosis [36,37]. If IncA interferes with SNAREs when present in RBL-2H3 mast cells, as it does in the liposomes, we should observe an inhibition of the secretory pathway. Because CcaIncA full-length protein was toxic for the cells, RBL-2H3 were transfected with myc-CcaIncA 1-220 , a truncated form of CcaIncA still containing its SNARE-like N-terminal domain [19], and therefore still inhibitory (see Fig. 2). Myc-CcaIncA 1-220 was cloned together with GFP into an IRES vector to simultaneously express a transfection marker (40% average transfection efficiency). Using immunofluorescence, we observed that Myc-CcaIncA 1-220 co-localized with lysotracker (Fig. 4A), a marker of the RBL-2H3 secretory lysosomes [38]. This suggests that Myc-CcaIncA 1-220 is located on the secretory lysosomes where it can potentially interact with the lysosomal v-SNAREs VAMP8. This is physiologically relevant since VAMP8 is involved in the phagosomal fusion with lysosomes [39]. Therefore, interfering with VAMP8 would protect the phagosomal compartment against degradation. After stimulating transfected mast cells with both 10 27 M Phorbol 12-Myristate 13-Acetate (PMA) and 10 26 M ionomycin, we analyzed the release of b-hexosaminidase, a lysosomal enzyme stored inside mast cell secretory lysosomes. Kinetic analyses showed that after 30 min of stimulation, cells transfected with myc-CcaIncA 1-220 /GFP secrete significantly less b-hexosaminidase than the GFP control. The level of inhibition at 30 min (23% inhibition) and at 60 min (32% inhibition) is significant (p,0.05 and p,0.02 respectively) compared to GFP transfected cells (Fig. 4B). These data confirm  (Figs 3 and 1 respectively). Incidentally, this also indicates that the in vitro liposome fusion assay is able to accurately predict cellular data. Therefore, this assay represents a unique system by which more bacterial proteins could be screened for their effect on host vesicular trafficking.

Conclusion
The key for survival of intracellular bacteria in host cells is their capacity to manipulate host cellular processes -in particular membrane fusion-to allow the establishment of an intracellular replicative niche. An obvious host machinery to target in order to block membrane fusion is the SNARE machinery. Using CtrIncA, CcaIncA and IcmG/DotF as our models, we demonstrated that SNARE-like bacterial proteins differentially block SNAREsmediated membrane fusion. Furthermore, we showed that this inhibitory function is encoded into their SNARE-like motifs, validating the general function of such a motif for manipulating membrane fusion.
Interestingly, clinical isolates lacking IncA present defects in their infectious cycle, and the number of inclusions per cells is significantly decreased [40,41]. This would suggests that the level of protection exerted by Chlamydia cell wall during the first 8 hrs [28] is sufficient for small inclusions to develop, but that IncA synthesis is necessary for the inclusions to maturate further. Alternatively, it could also suggest that additional protective systems, although not as efficient IncA, are in place to insure such an important function. Interestingly, Chlamydia was found to express additional SNARE-like bacterial proteins, including CT813, which also interacts with host SNARES [21]. Although their inhibitory function remains to be confirmed, the redundancy of the SNARE-like protein system would further support its importance. Overlapping layers of protection would insure the survival of Chlamydia in case one of the protective systems fails. The differential timing of expression for each of these proteins could also ensure the protection of the vacuole over time [30]. This redundancy would explain the presence of a limited number of Chlamydia inclusions during infections with strains naturally lacking IncA [41]. Each SNARE-like protein may also be specific for a   and is dependent upon its concentration, reaching 55% of inhibition after 2 hrs with an estimated CtrIncA 1-141 :SNARE ratio of 2:1. This experiment is representative of n = 3. B-E-Two different concentrations of truncated CtrIncA (see representative coomassie gels inserted in each graph) were reconstituted into t-SNARE (B,D), and v-SNARE liposomes (C,E). As shown on graphs B and C, CtrIncA 1-130 still displays a significant inhibitory effect on the endocytic SNARE-mediated membrane fusion (average of 15% inhibition after 2 hrs, p = 0.022). On the contrary, CtrIncA 1-120 (D,E) completely fails to inhibit endocytic fusion (p.0.05). The mean from n = 5 independent experiments was determined at 30 min, 60 min and 120 min. The standard deviation is shown. One asterisk denotes statistically significant differences (p,0.05). For the purpose of comparison, maximal values of fusion obtained for the SNARE complex without IncA at 120 min were arbitrarily defined as 100%. doi:10.1371/journal.pone.0007375.g002 It is intriguing to notice that the inhibitory mechanism displayed by bacterial SNARE-like proteins is very similar to the one employed by the eukaryotic inhibitory-SNAREs (i-SNAREs). SNARE-mediated fusion is triggered by four fusogenic subunits and is highly specific [42,43,44,45]. It was shown that the presence of a fifth SNARE on the same compartment could result in an inhibition of fusion (therefore, such a SNARE has been called inhibitory-SNARE). An i-SNARE can substitute for one of the subunits of the functional tetramer leading to the formation of a non-functional tetramer (acting as a pseudo t-SNARE) [46]. In the Golgi, it has been demonstrated that a gradient of i-SNAREs across cisternae blocks SNARE-mediated membrane fusion and is likely used to fine-tune the specificity of membrane fusion [46]. Here we showed that bacterial SNARE-like proteins appear to function in a similar fashion. Similar to i-SNAREs, these bacterial proteins are capable to bind fusogenic SNAREs and inhibit membrane fusion. Altogether, this suggests that coiled-coil SNARE-like motifs may constitute one of the most effective motifs to manipulate membrane fusion and has been incorporated into intracellular bacteria genome as an adaptation to the pressures of survival [47]. Ultimately, one could take advantage of such a recurrence to develop a common therapeutic strategy for targeting a wide array of bacterial SNARE-like proteins and revert the fusion blockage.

DNA manipulation and plasmid construction
Standard genetic manipulations were performed throughout. All polymerase chain reaction (PCR) procedures were done with pfu turbo polymerase (Stratagene). All other DNA modifying enzymes were from New England Biolabs. The E. coli strain DH5a (Invitrogen) was used for standard cloning. Plasmid encoding CcaIncA 122 was generated as described [19]. We added a myc tag and cloned CcaIncA 122 into the pIRES2-EGFP vector (Clontech) using the oligonucleotides FO134 GGGAATTCCATATGA-CAGTATCCACAGACAACAC and FO135 CGGGATCCTCA-CAGATCCTCTTCTGAGATGAGTTTTTGTTCCAAAGAC-TGAGCTAATTTCT.

Reconstitution into liposomes
SNARE proteins were reconstituted into proteoliposomes by detergent dilution and isolated on an Accudenz density gradient flotation as previously described [6,49]. To insert bacterial proteins into liposomes, v-SNARE protein and preformed t-SNARE complexes were respectively preincubated with the bacterial protein at different concentration for 4 hrs at 4uC, before being mixed with the lipids, and dialysed for 16 hrs at 4uC.

Liposome fusion assay
Fusion reactions and data analysis were performed as previously described [6,49]. For most fusion assays, the mean from at least 5 independent experiments was determined at 30 min, 60 min and 120 min. For the purpose of comparison, maximal values of fusion obtained for the SNARE complex without IncA at 120 min were arbitrarily defined as 100%. The Mann-Whitney U test was used to compare the mean values of maximal fusion at 120 min between SNARE-containing liposomes and SNARE/IncA-containing liposomes. Significance was assumed at p values,0.05.

Cell transfection
The rat mast cell line RBL-2H3 was cultured as described [37]. We used the AMAXA nucleofector technology (AMAXA, Germany) to transiently transfect the RBL-2H3 cells. Briefly, 2610 6 cells were nucleofected in 100 ml solution V (AMAXA) using 1 mg of pIRES2-EGFP-CcaIncA 122 vector or pIRES2-EGFP vector (control). The cells were nucleofected using the program T-030. Cells were then plated in complete medium in 96 well plates for subsequent secretory cell assays 12 hrs later. Using these conditions, the efficiency of transfection was routinely in the range of 30 to 40% as determined by immunofluorescence (GFP positive).

Confocal immunofluorescence microscopy
Lysotracker labeling was performed following the manufacturer's instruction. Briefly, cells grown on coverslips were incubated with lysotracker 1:20,000 for 20 min in complete medium and washed three times. The Myc tag labeling was performed as described [37]. We used the anti-myc antibody (9E10) from Santa Cruz Biotechnology. Cy5-conjugated anti-mouse antibody was from Jackson Laboratories. All data were analyzed using a Leica TCS SP confocal microscope, LEICA CONFOCAL 2.5 software, HCX PL APO 63X oil immersion objective.

Secretory cell assay
Transfectants were plated in 96 well plates in triplicates at ,5610 5 cells in 100 ml of complete DMEM medium and incubated overnight at 37uC. After 12 hrs, adherent RBL cells were washed twice in prewarmed phenol red free DMEM and stimulated by Phorbol Myristate Acetate (10 27 M)/ionomycin (10 26 M). At different time points (0, 15 min, 30 min 1 hr), 25 ml of supernatant was collected and the granule secretion marker b-hexosaminidase was analyzed using test supernatants within the linear range of the assay [50]. Total cellular content of b-hexosaminidase was determined by lysis of the adherent cells in 0.5% Triton X-100. The absorbance was determined at 410 nm in a micro-titer plate reader. Results were calculated as a percentage of total b-hexosaminidase in cells after correction for spontaneous release in unstimulated cultures. For the purpose of comparison, all data were normalized to the maximal value of b-hexosaminidase release obtained in pIRES2-EGFP transfectants and arbitrarily taken as 100%. The Mann-Whitney U test was used to compare the mean values of maximal release between GFP and Myc-CcaIncA 1-220 transfectants. Significance was assumed at p values,0.05.

SDS-PAGE and Western blot analysis
Western blots were performed as described [51]. The anti-myc antibody (9E10) was from Santa Cruz Biotechnology, the anti-SNAP23 antibody from Synaptic System and both were used at 1:500. The secondary antibodies were from Biorad and were used at 1:20,000.