Recruitment of the Major Vault Protein by InlK: A Listeria monocytogenes Strategy to Avoid Autophagy

L. monocytogenes is a facultative intracellular bacterium responsible for listeriosis. It is able to invade, survive and replicate in phagocytic and non-phagocytic cells. The infectious process at the cellular level has been extensively studied and many virulence factors have been identified. Yet, the role of InlK, a member of the internalin family specific to L. monocytogenes, remains unknown. Here, we first show using deletion analysis and in vivo infection, that InlK is a bona fide virulence factor, poorly expressed in vitro and well expressed in vivo, and that it is anchored to the bacterial surface by sortase A. We then demonstrate by a yeast two hybrid screen using InlK as a bait, validated by pulldown experiments and immunofluorescence analysis that intracytosolic bacteria via an interaction with the protein InlK interact with the Major Vault Protein (MVP), the main component of cytoplasmic ribonucleoproteic particules named vaults. Although vaults have been implicated in several cellular processes, their role has remained elusive. Our analysis demonstrates that MVP recruitment disguises intracytosolic bacteria from autophagic recognition, leading to an increased survival rate of InlK over-expressing bacteria compared to InlK− bacteria. Together these results reveal that MVP is hijacked by L. monocytogenes in order to counteract the autophagy process, a finding that could have major implications in deciphering the cellular role of vault particles.


Introduction
Listeria monocytogenes is a Gram-positive bacterium responsible for listeriosis, a severe food-borne human infection with an overall mortality rate of 30% [1]. L. monocytogenes has evolved efficient strategies to survive in the intestine and cross the intestinal, bloodbrain and placental barriers [2,3] leading to clinical features of the disease that include gastroenteritis, septicemia, central nervous system infections, and mother-to-child infections [4]. Inside the host, this facultative intracellular bacterium is able to invade phagocytic and non-phagocytic cells, replicate intracellularly, and spread directly from cell-to-cell, thereby escaping the immune response [3]. L. monocytogenes has thus emerged as a paradigm to study host-pathogen interactions and fundamental processes in cell biology [5]. For instance, the study of actin rearrangements upon entry and intracellular movements [6][7][8][9] is an example of how understanding a bacterial-induced process can yield insight into basic cellular processes. Namely, the listerial virulence factor ActA triggers the recruitment of Arp2/3 complex and Ena/VASP to mediate actin polymerization and propel the bacterium from one infected cell to another without exposure to the extra-cellular milieu [8,10]. Interestingly, as shown recently ActA also disguises the bacteria from autophagic recognition within the cytosol as ActA-bacteria becomes rapidly ubiquitinated and targeted to autophagy [9,11]. It is currently viewed that ubiquitin-associated bacteria recognized by the autophagy machinery are trapped by autophagosomal membrane for delivery into the lytic compartment where they undergo degradation by autolysosomes [11,12]. Interestingly, a variety of studies had noticed that autophagic markers can accumulate around intracytosolic L. monocytogenes, unless bacteria were forming actin tails [13,14]. Consequently, it has been hypothesized and shown that L. monocytogenes avoids ubiquitination and autophagic recognition by expressing ActA, and ActA mutants are efficiently targeted by autophagy [11]. While the role of ActA in autophagy is now established, the role that many other surface proteins play during Listeria infection remains fragmentary [15].
The vault particle is the largest cytoplasmic ribonucleoprotein complex known to date [16]. Originally identified as contaminants of clathrin-coated vesicles preparation, these complexes were named vault particles because of their barrel shaped morphology resembling the ceiling of cathedrals [17]. Mammalian vaults are composed of the highly conserved major vault protein (MVP) constituting more than 70% of the mass of the particle [16,18,19] which spontaneously forms vault particles without the need of other vault components [20]. The two other vault components are the telomerase associated protein (TEP-1) [21] and the vault poly(ADP)ribose polymerase (vPARP) [22][23][24]. Vault preparations have additionally been shown to contain several small untranslated RNAs [25,26]. Vaults exist in thousands of copies per cell and are widely expressed in all eukaryotic organisms, from Dictyostelium discoideum to mammals, except plants, Saccharomyces cerevisiae, Caenorhabditis elegans and Drosophila melanogaster [27]. Diverse roles have been proposed for MVP and/or vaults [27], including roles in drug resistance [28], cellular differentiation [29], innate immunity [30], virus infections [31], signaling cascades [28,[32][33][34][35] and cell survival [33,36]. However, the precise cellular function(s) of MVP and vaults remains poorly understood. In addition, the MVP 2/2 mice are viable, healthy and show no obvious abnormalities [37,38].
The genome sequence of L. monocytogenes EGD-e has revealed the presence of 25 genes encoding proteins of the internalin family [39]. Proteins of this family, which are characterized by the presence of leucine-rich-repeats (LRRs), are mostly surface proteins [40]. Their binding to the bacterial surface is mediated by different anchoring domains, in particular the LPXTG motif which allows a sortase A mediated covalent attachment to the peptidoglycan [41]. The invasion protein, Internalin, is one such protein [42]. Comparative post-genomic studies have established that several members of the L. monocytogenes internalin family are absent in L. innocua, a closely related non-pathogenic species [40]. Lmo1290 is an internalin gene absent in L. innocua, herein referred to as inlK, which is expressed at very low levels in brain-heartinfusion medium [43,44] and induced during infection [43].
In this study we investigated the role of InlK in the infectious process. We first explored the expression of InlK and the virulence phenotype of the inlK deletion mutant. We then searched for potential host partners of InlK and identified MVP. We demonstrated that the InlK/MVP interaction occurs in the cytosol of infected cells at the bacterial surface. Moreover, our results reveal that MVP recruitment protects L. monocytogenes from autophagic recognition, leading to an increase in bacterial survival in infected cells.

L. monocytogenes inlK encodes a virulence factor
The gene lmo1290 ( = inlK) is 1797 bp long. It is located 331 bp downstream from gene lmo1289 which is followed by a transcriptional terminator. Lmo1290 is also followed by a transcriptional terminator upstream from the divergently transcribed oatA gene which encodes a peptidoglycan O-acetyltransferase ( Figure 1A) [45]. The inlK gene is present in all 22 L. monocytogenes genomes sequenced to date and absent from the genomes of L. ivanovii and all non-pathogenic Listeria strains including L. innocua ( Figure 1A), L. seeligeri, L. welschimeri and L. grayi, suggesting that InlK could be involved in Listeria virulence.
To assess the role of InlK in virulence, we generated an inlKdeletion mutant (DinlK) in the strain EGD-e. The DinlK mutant grew as rapidly as the wild-type (WT) in broth medium and infected cells (macrophages and epithelial HeLa cells) (data not shown). The LD50 of the DinlK mutant after intravenous (i.v.) injection in BALB/c mice was 2.2610 4 CFU, compared with 1.7610 3 CFU for the WT strain. Inactivation of inlK resulted in complete survival of animals infected intravenously with 10 4 bacteria ( Figure 1B). In contrast, infection with the same number of WT bacteria led to 100% mortality. Moreover, the number of CFU recovered from spleens and livers of i.v. infected BALB/c mice after 24 h, 48 h, 72 h and 96 h of infection was significantly lower (,1 Log 10 ) for the mutant compared to the WT ( Figure 1C), and virulence of the mutant was fully restored by complementation ( Figure 1C). Together, these results establish a role for InlK in the virulence of L. monocytogenes.

InlK is expressed in vivo
InlK is a 598 amino acid LPXTG surface protein predicted to be anchored to the peptidoglycan by sortase A (Figure 2A). To address whether L. monocytogenes produces InlK in vitro, we first generated an antibody against a purified recombinant InlK protein ( Figure S1A) and used it to detect the protein at the bacterial surface by immunofluorescence. In agreement with previous whole genome transcriptomic results that demonstrated a low expression level of inlK in vitro [44], bacteria grown in brain-heart infusion (BHI) medium were not stained by the InlK antibody ( Figure 2B), suggesting that InlK protein was poorly expressed on the surface or not produced. We then showed that InlK was not detected in bacterial total extracts ( Figure 2C), also in agreement with previous data indicating that InlK is not present in the cell wall proteome of L. monocytogenes EGD-e grown in BHI medium [46]. Moreover, consistent with the fact that the two major regulators of virulence genes, PrfA and sigmaB, were not required for basal inlK transcription [44,47], the InlK protein level was also not detectable when bacteria were grown in charcoal supplemented medium or at low pH (data not shown).
To verify that the inlK open reading frame encoded a surface protein, inlK was expressed under the control of two constitutive promoters active in Listeria. We used either the promoter of the protease gene from Lactococcus lactis subsp. cremoris on the multicopy plasmid pPRT-inlK or the promoter P Hyper after integration on the chromosome of the plasmid pADc-inlK [48,49]. InlK antibodies efficiently labeled InlK on the surface of bacteria that constitutively expressed inlK ( Figure 2B) and also detected the protein in bacterial total extracts ( Figure 2C). This labeling was specific, as the InlK antibody did not label WT or inlK mutant bacteria grown in same conditions. Interestingly, when InlK was over-expressed by Listeria under the control of constitutive promoters, a polypeptide with a lower mass than expected was also detected by Western-blot ( Figure 2C) indicating that the protein may be processed. Moreover, InlK was not detected by immunofluorescence at the surface of a DsrtA sortase mutant over-expressing inlK ( Figure 2B), but was then detected in the supernatant of the culture medium ( Figure 2D). Taken together these results established that, when inlK is expressed, the protein is anchored at the bacterial surface in a sortase A-dependent manner.

Author Summary
L. monocytogenes is a food-born pathogen responsible for listeriosis, a severe illness with a high mortality rate, which mainly affects immunocompromised patients and pregnant women. The bacterium is a facultative intracellular pathogen able to invade, survive and multiply in large variety of cells. Although the infectious process at the cellular level has been extensively studied, the role of InlK, a surface protein specific to L. monocytogenes, remains unknown. Here we established that L. monocytogenes use InlK to interact with a mammalian cytoplasmic protein named Major Vault Protein (MVP). Although MVP has been implicated in diverse cellular processes, its role remains elusive. Here we demonstrate that, inside the cell, L. monocytogenes is able, via InlK, to decorate its surface with MVP to escape autophagy, an innate immune defense system that protects the cell from invading pathogens. L. monocytogenes uses this camouflage strategy to efficiently survive inside cells.
Recently, a whole genome transcriptomic analysis of L. monocytogenes during infection revealed that the gene inlK was better expressed in vivo compared to growth in BHI [43]. We thus investigated whether the InlK protein was indeed produced in vivo by testing for the presence of anti-InlK antibodies. Purified InlK was submitted to migration on polyacrylamide gel ( Figure S1A) and blotted with two different rabbit anti-Listeria sera. As shown in Figure 2E, a rabbit anti-L. monocytogenes serum obtained after immunization with killed bacteria was not able to detect the purified InlK whereas the serum directed against live bacteria detected InlK. This signal was specific to InlK as the antibodies did not label bovine serum albumine (BSA) used at the same concentration.
To confirm in vivo inlK expression, we constructed an expression reporter vector in which the expression of the bioluminescent operon lux ABCDE was under the control of inlK promoter (pPL2-P inlK -lux ABCDE ). This construct was integrated in the chromosome of WT L. monocytogenes EGD-e, and the resulting strain was used to infect cell lines or BALB/c mice [50]. As shown in Figure S1B, inlK was neither expressed in BHI growth medium (right panel) nor in cells infected with bacteria previously grown in BHI (left panel). Conversely, inlK was expressed in vivo in i.v. infected mice, 24 h post-infection ( Figure S1C). This signal was specific to inlK expression as it did not superimpose on those obtained with the control strain of L. monocytogenes that contains a bioluminescent reporter of LLO promoter (pPL2-P hly -lux ABCDE ). Together, these results confirm that InlK is expressed in vivo.

InlK interacts with the Major Vault Protein
To identify InlK interaction partners in the eukaryotic cell, we used InlK as a bait in a large-scale yeast two-hybrid screen and identified the Major Vault Protein (MVP) as a prey with a very high interaction score (Table S1). To confirm this interaction we performed a bacterial pull down assay and showed that GST-MVP purified protein bound to InlK over-expressing bacteria, but not to WT bacteria ( Figure 3A). This interaction was specific as (i) the WT strain (which expresses InlA, InlB and InlH) did not bind MVP (data not shown), (ii) the overexpression of InlJ (i.e. another internalin not expressed in BHI [49]) was not able to mediate bacterial binding to MVP, and (iii) InlK over-expressing bacteria were not able to bind another GST fusion protein, GST-ScarA. Finally, bacterial incubation with MVP-GFP transfected cells lysates confirmed the interaction between InlK and MVP ( Figure 3B). This interaction occurred when InlK was either expressed on a multicopy plasmid, or integrated in the chromosome ( Figure S2A).
This interaction between purified InlK and endogenous MVP was confirmed by co-immunoprecipitation assays ( Figure 3C). Indeed, when purified InlK was incubated with HT29 cell lysate, it interacted with endogenous MVP and the two partners coimmunoprecipitated, as shown using an anti-MVP antibody ( Figure 3C). Similar results were obtained with stable HEK293-HTP-InlK cells that were engineered to express InlK in their cytosol, under the control of a tetracyclin inducible promoter ( Figure S2B).
The InlK/MVP interaction occurs in the cytosol at the bacterial surface and does not depend on actin polymerization In agreement with a specific interaction between InlK and MVP, we observed that InlK over-expressing bacteria co-localized with MVP in MVP-GFP transfected HeLa cells whereas the inlK mutant or wild type bacteria that do not express InlK in vitro did not co-localize with MVP ( Figure 3D). As MVP has been mainly described as a cytoplasmic protein [51,52] and InlK is targeted and anchored to the bacterial surface ( Figure 2D), we hypothesized that the InlK/MVP interaction should occur in the cytosol of infected cells after lysis of the internalization vacuole. To test this hypothesis, we analyzed the localization of MVP recruiting bacteria. A differential immuno-staining protocol allowing extraand intracellular Listeria to be distinguished showed that MVP was recruited to intracellular bacteria ( Figure 4A To determine whether MVP was recruited to intracellular bacteria before or after lysis of the internalization vacuole, we used a marker of early times points after vacuole escape, YFP-CBD, a YFP fused phage protein known to bind L. monocytogenes peptidoglycan as soon as the vacuole membrane begins to lyse ( Figure S3A) [53]. Cells were co-transfected with MVP-Tomato and YFP-CBD, fixed 4 h post-infection and immuno-stained for actin. As expected, bacteria that polymerized actin were efficiently labeled with YFP-CBD ( Figure 4B, Figure S3B inset 2), confirming that YFP-CBD efficiently labels intracytosolic Listeria. Moreover, all MVP-positive bacteria were also labeled with YFP-CBD ( Figure 4B, Figure S3B) revealing that MVP was recruited by intracytosolic bacteria after lysis of the internalization vacuole.
Interestingly, we did not observe the co-recruitment of MVP-GFP and endogenous actin to intracellular bacteria ( Figure 4B, Figure S3B). Co-recruitment was also not observed in infected cells previously co-transfected with MVP-GFP and actin-CFP ( Figure  S3C). We therefore analysed the kinetics of MVP and actin recruitment by performing live-cell imaging. Cells were cotransfected with MVP-Tomato and actin-GFP, and infected with InlK expressing L. monocytogenes. Strikingly, MVP was recruited rapidly by InlK over-expressing bacteria and could then be replaced by actin ( Figure 4C and Video S1), showing that MVP recruitment occurs before actin polymerization. We then verified that MVP recruitment occurred independently of actin polymerization using a actA mutant. Intracytosolic DactA over-expressing InlK were efficiently labeled with MVP ( Figure S3D). This MVP recruitment was more efficient than for wild type bacteria. Indeed, the percentage of intracytosolic DactA over-expressing InlK having recruited MVP at 4 h post-infection was 88.3612.7% (mean 6 SEM from n = 3 experiments), compared to the 16.862.5% (mean 6 SEM from n = 3 experiments) observed when using the InlK over-expressing strains that are able to polymerize actin via ActA. Together, these results suggested that ActA at least partially impairs MVP recruitment. As ActA protects bacteria from autophagy [11], these data also suggested that both InlK and ActA may protect bacteria from autophagy.

MVP recruitment protects Listeria from autophagy
To test if MVP recruitment could lead to autophagic escape, we used two well-established markers of autophagy, p62 (SQSTM1) and LC3 (Atg8) [54]. p62 has emerged as the prototypic adaptor involved in directing cytoplasmic substrates towards autophagic degradation [55]. p62 interacts with ubiquitinated subtrates via its ubiquitin-binding domain, and links them to the autophagosomal structural protein LC3. We infected MVP-transfected HeLa cells with InlK over-expressing bacteria for 4 h and after fixation, immuno-stained for endogenous p62 and actin. No co-localization could be observed between MVP and p62 ( Figure 5A, Figure S4A) or MVP and LC3 ( Figure 5B, Figure S4B). Interestingly, the vast majority of MVP-positive bacteria were completely devoid of anti-p62 labeling (95.162.0%; mean 6 SEM from n = 3 experiments) but 4.962.0% (mean 6 SEM from n = 3 experiments) were stained at one pole with MVP and at the other pole with p62. Similar results were obtained using GFP-LC3 ( Figure 5, Figure  S4B). As previously demonstrated [11], bacteria that had started to recruit actin were not labeled by p62 ( Figure 5A) or GFP-LC3 ( Figure 5B). Strikingly, when the MVP-positive bacteria that exhibited a recruitment of LC3 at one pole were examined by livecell-imaging ( Figure 5C), the membrane elongation leading to the autophagosome formation failed to occur (Video S2 and S3). Together these results indicate that bacteria which either recruit MVP or have started to polymerize actin evade autophagic recognition.
We thus studied autophagy marker recruitment by the actA mutant over-expressing InlK. In agreement with our previous observations ( Figure S3D), InlK over-expressing DactA bacteria efficiently recruited MVP [88.6612.8% (mean 6 SEM from n = 3 experiments)] ( Figure 5D and S4C). These MVP positive bacteria were neither surrounded by ubiquitinated proteins nor recognized by LC3 ( Figure 5D and S4C). Furthermore, the level of LC3-II, the active form of LC3 that correlates with active autophagy [54], was significantly lower in cells infected with DactA+InlK as compared with DactA (1.8260.14 fold) ( Figure 5E). Together, these results show that in the absence of ActA, Listeria is able to evade autophagic recognition via MVP recruitment.

MVP-dependent escape from autophagy leads to increased Listeria survival
Autophagy is recognized as a cell-autonomous innate defense mechanism that may control growth of intracellular microbes [56]. We thus tested if MVP-mediated autophagy escape leads to increased bacterial survival. As macrophages are among the cells which express the highest levels of MVP [29,57,58], the intracellular survival of WT, WT+InlK, DactA and DactA+InlK was analysed in RAW 264.7 macrophages ( Figure 6A). These four strains (WT, WT+InlK, DactA and DactA+InlK) were first verified to grow identically in culture medium (data not shown). As previously described by Yoshikawa et al. [11], the intracellular survival rate of DactA bacteria at 4 h post-infection was significantly lower than that of WT bacteria ( Figure 6B). Strikingly, the expression of InlK by the DactA strain restored the intracellular survival rate to the level of WT bacteria ( Figure 6B), indicating that InlK could functionally replace ActA in its role in autophagy escape. Infection of MVP-transfected epithelial cells with DactA and DactA+inlK led to similar results ( Figure 6C and 6D).
The intracellular survival of DactA and DactA+InlK was then analysed in RAW 264.7 macrophages treated with control or MVP siRNA ( Figure 6E). As previously observed ( Figure 6B), the DactA+InlK strain replicated better than the DactA strain in control cells ( Figure 6E). Strikingly, in MVP-depleted cells, the Dac-tA+InlK strain did not replicate faster than the DactA strain ( Figure 6F), confirming the role of InlK/MVP interaction in survival rate. Taken together, these data show that the specific recruitment of MVP to the bacterial surface via InlK leads to a better survival of L. monocytogenes.

Discussion
L. monocytogenes has emerged as a paradigm to study hostpathogen interactions and fundamental processes in cell biology [5,59]. However the role of the many proteins expressed on the bacterial surface during Listeria infection remains fragmentary [15]. In this study we report that InlK, a L. monocytogenes surface protein of the internalin family, plays a critical role in Listeria virulence. We show that InlK is anchored to the listerial surface through its LPXTG peptidoglycan anchoring signal by sortase A and is produced during in vivo infection, whereas it cannot be detected on bacteria grown in BHI medium [44] or within the cytosol of tissue-cultured cells. This in vivo specific expression profile was previously described for other virulence factors of L. monocytogenes, e.g. the internalin InlJ, that behaves as an adhesin [49] and recently LntA, a secreted bacterial protein involved in chromatin remodeling and type III interferon response [60]. Furthermore, our results confirm and extend our recently published transcriptomic analysis of L. monocytogenes [43] which identified inlK as a gene highly activated during in vivo infection and that may play a role in the infectious process. Together, our results demonstrate that InlK is a so far undescribed virulence factor of L. monocytogenes.
To enter, survive and spread from cell-to-cell, L. monocytogenes has been shown to interact with several host partners. We revealed here that MVP is a specific cellular interactor of InlK. The highly conserved MVP protein constitutes more than 70% of the mass of Figure 2. InlK is expressed in vivo. A. InlK amino acid sequence. The signal sequence is underlined and the different regions of leucine rich repeats (LRRs) are outlined. The consensus pentapeptide LPXTG at the C-terminal end is boxed. B. Detection by immunofluorescence microscopy of InlK overexpressing in L. monocytogenes EGD-e (WT), DinlK, WT+pADc-inlK, DinlK+pPRT-inlK and the DsrtA mutant over-expressing inlK (DsrtA+pPRT-inlK) grown in BHI medium using the rabbit polyclonal anti-InlK antibody. InlK was detected at the surface of InlK over-expressing bacteria (WT+pADc-inlK and DinlK+pPRT-inlK), whereas it was undetectable at the surface WT bacteria or at the surface of the DstrA mutant over-expressing inlK. C. Detection of InlK by Western blot on total lysates of L. monocytogenes EGD-e (WT), DinlK and DinlK+pPRT-inlK grown in BHI using the rabbit polyclonal anti-InlK antibody. Decreased concentrations of recombinant purified InlK were used as a positive control. D. Detection of secreted InlK in the supernatant of DsrtA mutants over-expressing InlK. Western blotting was carried out on trichloroacetic acid precipitates of DsrtA and DsrtA+pPRT-inlK culture (OD 600 = 1) supernatants using the rabbit polyclonal anti-InlK antibody. E. Detection of purified recombinant InlK protein with rabbit polyclonal antilive Listeria antibody, rabbit polyclonal anti-killed Listeria antibody, rabbit polyclonal anti-InlK and a rabbit pre-immune serum. InlK was detected only with the rabbit polyclonal anti-live Listeria antibody indicating that it is expressed during the in vivo infectious process. BSA was used as control protein. Two different amounts of proteins were tested (500 ng and 200 ng) to access signal specificity. doi:10.1371/journal.ppat.1002168.g002 the largest cytoplasmic ribonucleoprotein (RNP) complex known, i.e. vault particles [16,18,19]. Since its first description in 1986 [61], several putative functions have been attributed to this RNP complex. Data that link the vault complex to various functions have suggested roles in multidrug resistance [28,62], transport [63], signaling [28,[32][33][34][35], apoptosis resistance [33,36] or innate immunity [30]. However, no compelling evidence for a cellular role was reported unequivocally and MVP was mainly considered as a scaffold protein. Nevertheless, vaults were previously found to be implicated in g-herpesvirus (Epstein-Barr and Kaposi's sarcoma virus) [31,64] and Pseudomonas aeruginosa infectious processes [30]. During Epstein-Barr or Kaposi's sarcoma virus infection, the expression of vault RNAs (vRNAs) was shown to be specifically upregulated in human lymphocytes [31,64]. However, the function of this overexpression was not assessed. In addition, not only vRNA but also MVP was reported to be upregulated during viral infection by human T-cell lymphotropic virus type I (HTLV-I) infection [65]. In the case of bacteria, MVP was implicated in host resistance to P. aeruginosa lung infection [30]. Indeed, a rapid recruitment of MVP to lipid rafts was observed when human lung epithelial cells were infected with P. aeruginosa. This recruitment was dependent on bacterial binding to the cystic fibrosis transmembrane conductance regulator CFTR. However, no evidence of direct binding between MVP and bacteria was observed. Our results provide the first report of a direct interaction between a microbial protein and a component of the vault particles. Indeed, we demonstrated that InlK over-expressing L. monocytogenes were able to directly bind MVP. In agreement with previous observations that MVP/vaults are predominantly (.90%) localized in the cytoplasm [28,51,52], we established that the InlK/MVP interaction occurs in the cytosol of infected cells, after the disruption of the internalization vacuole, and independently of actin polymerization.
As with a variety of intracellular microbes, intracytosolic L. monocytogenes are recognized by autophagy, a cell-autonomous effector mechanism of innate immunity that protects the cytosol against bacterial invasion [66]. Perrin et al. first demonstrated that cytosolic L. monocytogenes occasionally colocalized with ubiquitin in infected cells, and this association was more frequent in case of the DactA strain [13,14]. More recently, Yoshikawa et al. demonstrated that the recruitment of VASP, Arp2/3 complex and actin via ActA protect bacteria from ubiquitination and autophagic recognition [11]. Here we reveal that L. monocytogenes has a second strategy to escape autophagy in the absence of ActA (Figure 7). Indeed, no significant difference could be observed between the intracellular survival rate of WT and WT+InlK bacteria in infected RAW 267.4 macrophages ( Figure 6B), suggesting that when ActA is expressed it is sufficient for Listeria to escape from autophagy. In contrast, in absence of ActA, InlK protects against autophagy. Together, our results show that the bacteria are able, via InlK, to decorate their surface with MVP in order to escape from autophagy ( Figure 7). It will be thus of the highest importance to decipher in which cells InlK is expressed in vivo and when the InlK/MVP interaction takes place during infection. These data will be critical to unravel the role of InlK in the pathophysiology of Listeria infection. It will also be of great interest to further study the link between actin polymerization, MVP, autophagy, and pathogen dissemination.

Bacterial strains, growth conditions and reagents
Listeria strains (Table S2) were grown in brain-heart infusion (BHI) medium (Difco; BD) and Escherichia coli were grown in Luria-Bertani Medium (LB) medium (Difco; BD). When required, chloramphenicol and erythromycin were used at final concentrations of 7 mg/ml and 5 mg/ml respectively for L. monocytogenes and kanamycin, erythromycin and chloramphenicol were used at final concentration of 50 mg/ml, 150 mg/ml and 35 mg/ml, respectively for E. coli.  [67] digested by BamHI and NcoI, yielding the pMAD-DinlK plasmid. The sequence was verified by sequencing. This plasmid was electroporated into L. monocytogenes EGD-e. Independent colonies were used for allelic exchange in L. monocytogenes wild-type EGD-e, which was performed as previously described [49], generating a DinlK isogenic deletion mutant (Table S2). Deletion of the entire inlK gene was confirmed by PCR amplification and sequencing.

Generation of EGD-e DinlK mutant strain and inlK over-expressing strains
Generation of InlK over-expressing strains. To express InlK in L. monocytogenes the pPRT-and pADc-derivative plasmids were constructed as described below. In the pPRT-inlK plasmid, inlK was expressed under the control of the promoter region of the protease gene from Lactococcus lactis subsp. cremoris, which is active in Listeria [49]. This is a multicopy plasmid which expresses an erythromycin resistance gene used for cloning selection.
The pADc-inlK plasmid generated as previously described by Balestrino et al [48] was derived from the integrative pPL2 plasmid, which inserts in the Listeria chromosome at the tRNAArg-attBB site, thereby avoiding the requirement for antibiotic pressure to maintain the plasmid and preventing heterogeneity of InlK expression due to variation in the plasmid copy number.  The following day, the medium was changed and the cells were incubated in complete medium for another 24 hours. Infections were performed as above-mentioned and the efficiency of siRNA knock-down was assessed by performing Western-blot on total cell lysates in each experiment ( Figure 6E).

InlK purification
The inlK coding sequence (aa 27-568) was amplified using primers lmo1290-Fw: 59-GAG TCG GAT CCG GTA TTT GCT GCA GAG CAAC C-39 and lmo1290-Rev: 59-GAG TCG TCG ACA GCC TCT TTA CTT GGT TCT G-39. The PCR product was purified and ligated with pET28b (Novagen) plasmid after double digestion with BamHI and SalI enzymes. The ligation product was electroporated in E.coli XL-1 Blue and positive clones were selected on 50 mg/ml supplemented BHI and sequenced (BUG 2812). For purification, E.coli BL21(DE3) (Invitrogen) were chemically transformed with the purified His 6 -InlK-His 6 expressing pET28b plasmid. Bacteria were grown in 50 mg/ml supplemented LB until OD 600 0.6 and IPTG was added at the final concentration 1 mM for 4 additional hours. Bacteria were lysed using a French press and the supernatant was recovered. His 6 -InlK-His 6 purification was performed using TALON His-Tag Purification Resins (Clontech). Increased concentration of imidazole (0-200 mM) in Tris-HCl 20 mM, NaCl 0.5 M (pH = 8) were used for purification and elution of InlK. The purified protein was dialysed against Tris-HCl 20 mM, NaCl 0.5 M (pH = 8) buffer and concentrated using AmiconUltra centrifugal filter (Millipore).

Immunofluorescence microscopy
Cells were fixed in 4% paraformaldehyde (PFA) in 1X PBS for 20 minutes at room temperature. Cells were then rinsed in 1X PBS before incubation in blocking solution (0.5% BSA, 50 mM NH4Cl in PBS, pH 7.4) containing 0.05% saponin. Cells were then incubated with the primary antibodies diluted in the blocking solution for 30 min at room temperature, rinsed 5 times in 1X PBS and further incubated for 30 minutes with the secondary antibodies diluted in blocking solution. Where needed, fluorescent phalloidin was added with the secondary antibodies to label actin. Cells were then rinsed 5 times in 1X PBS and mounted on glass coverslip using Fluoromount mounting medium (EMS, PA). The differential immuno-staining between extra-and intracellular Listeria was previously described [69]. Samples were analysed either with a Zeiss Axiovert 135 epifluorescence microscope connected to a CCD camera or with a Zeiss LSM510 confocal microscope (Carl Zeiss, Germany). Images were acquired with a 100X oil immersion objective and images were processed using MetaMorph (Universal Imaging Corp.).

Plasmids
The MVP-GFP plasmid that encodes EGFP fused to MVP Cterminus has been previously described [70]. To construct MVP-CFP (BUG 2908) and MVP-Tomato (BUG 2909), the MVP coding sequence was isolated from the MVP-GFP (BUG 2907) plasmid by double enzyme digestion (HindIII and BamHI) and found between the 4 time points. B. Detection of MVP recruitment at the surface of intracytosolic InlK over-expressing bacteria. HeLa cells were transfected with MVP-tomato (red) and YFP-CBD (green), infected with InlK over-expressing Listeria (DinlK+pPRT-inlK) for 4 h, fixed for fluorescence light microscopy and stained with phalloidin (blue). MVP positive bacteria were also labeled with YFP-CBD revealing that MVP was recruited by intracytosolic bacteria after the lysis of the internalization vacuole. Inset regions are magnified. The scale bar represents 1 mm. C. Kinetics of MVP and actin recruitment at the surface of InlK over-expressing bacteria. HeLa cells were transfected with MVP-tomato (red) and actin-GFP (green), infected with InlK overexpressing Listeria (DinlK+pPRT-inlK) for 4 h, and prepared for real-time video microscopy. Image series were collected every 15 min for 2 h. The left part shows an MVP positive bacterium that never recruits actin. The right part shows MVP replacement by actin around the bacterium. No colocalization of MVP-Tomato and actin-GFP was detected. Time is indicated along the Y axis. The entire image sequence can be viewed as Video S1. doi:10.1371/journal.ppat.1002168.g004 ligated in pECFP-N1 and ptdTomato-N1. Briefly, ptdTomato-N1 was constructed by replacing EGFP, from pEGFP-N1 vector (Invitrogen), by tdTomato, from ptdTomato-LCa vector (BUG 2420) [69]. Plasmid encoding CBD-YFP (BUG 2305) [53], actin-GFP (BUG 2421), actin-CFP (BUG 2155) and GFP-LC3 (BUG 3046) [71] were described elsewhere. Cells transfections were performed with FuGENE HD (Roche) as recommended by the manufacturer.

Bacterial pull down assay
To test the binding of bacteria to GST-MVP, L. monocytogenes strains were grown in BHI to an OD 600 of 1.0, and 1 ml of each culture was taken for each reaction. Bacterial cells were washed twice in buffer with 20 mM HEPES pH 7.5, 150 mM NaCl, resuspended in 1 ml of binding buffer (20 mM HEPES pH 7.5, 150 mM NaCl, 1 mM CaCl2, 1 mM MgCl2, and 0.2% BSA), and incubated at room temperature on a rotating wheel for 30 min. GST-MVP recombinant protein was added to a final concentration of 1 mg/ml and samples were incubated with rotation for an additional hour. Samples were centrifuged and pellets were washed three times in 20 mM HEPES pH 7.5, 300 mM NaCl, 0.05% Tween 20 and three times in buffer lacking Tween 20. The final bacterial pellets were resuspended in 20 microliters of 2X sample buffer, boiled for 10 min, and stored at 220uC before migration on 8% SDS/PAGE gels and Western blotting.
To analyse the binding of bacteria to transfected MVP-GFP HeLa cell lysates, HeLa cells were grown on 75 cm 2 flask, then transfected with MVP-GFP plasmid 24 h prior to the experiment. Cells were lysed at 4uC for 10 min in a 1 ml of lysis buffer (Tris-HCl 50 mM, pH 7.5, NaCl 150 mM, EDTA 2 mM, NP40 1%, AEBSF 1 mM, Na 3 VO 4 3 mM). Cells were scraped and the lysates were incubated with rotation for additional 15 min. Lysates were cleared by 15 min centrifugation at 13 000 g at 4uC. L. monocytogenes strains were grown in BHI to an OD 600 of 1.0, and 1 ml of each culture was taken for each reaction. Bacterial cells were washed twice in lysis buffer and were resuspended in cell lysate supernatants, mainly corresponding to cytoplasmic fraction, and incubated at room temperature on a rotating wheel for 1 hour. Samples were centrifuged to pellet bacteria, and washed five times in washing buffer (Tris-HCl 50 mM, pH 7.5, NaCl 150 mM, EDTA 2 mM, NP40 0.2%, AEBSF 1 mM, Na 3 VO 4 3 mM). The final bacterial pellets were treated as described above.
In each experiments bacterial inoculi were counted to ensure that an equal number of bacteria were used for pull-down assay.

Co-immunoprecipitation
HT29 (colon epithelial cells, ATCC HTB-38) cells were cultured to confluence in 75 cm 3 flask. Cells were lysed for 10 min at 4uC in 1 ml of lysis buffer (Tris-HCl 50 mM, pH 7.5, NaCl 150 mM, EDTA 2 mM, NP40 1%, AEBSF 1 mM, Na 3 VO 4 3 mM). Cells were scraped and the lysates were incubated with rotation for additional 15 min. Lysates were cleared by 15 min centrifugation at 13 000 g at 4uC and were incubated overnight at 4uC with 20 mg of purified InlK. Then, coimmunoprecipitation was performed using 1 mg of anti-MVP antibody. After five washes in lysis buffer, samples were resuspended in 20 microliters of 2X sample buffer, boiled for 10 min, and stored at 220uC before migration on 8% SDS/ PAGE gels and Western blotting.
Stable HEK293-HTP Blue and HEK293-HTP InlK were constructed as previously described [60]. When necessary cells were treated with doxycyclin 24 h prior the assay. Co-immunoprecipitations were performed using anti-MVP antibody as described above.

Yeast two-hybrid screening
The InlK encoding sequence (aa 27-568) was amplified by PCR from EGD-e and cloned into pB29 (N-bait-LexA-C fusion) plasmid. Randomly primed cDNA from human placenta poly(A) were constructed into a prey plasmid derived from pBTM116. The two-hybrid screen was performed by Hybrigenics (www. hybrigenics.com). The DNA inserts of the positive clones were sequenced to identify the corresponding gene in GenBank database using a fully automated procedure. Results of the yeast two-hybrid screening are recapitulated in Table S1.

Immunoblotting
Cells were seeded into 6 well plates and incubated 24 h before infection. Infections were done as described above. At the indicated times cells were lysed and lysates were analysed by Western blot. The immunoblots shown are representative of three independent experiments. Analysis of immunoblots was performed using G:Box Gel documentation system (Syngene).

Ethics statement
All animals were handled in strict accordance with good animal practice as defined by the relevant national and local animal welfare bodies, and all animal work was approved by the Institut  . Model for escape of autophagic recognition for L. monocytogenes expressing InlK. During intracellular growth, cytoplasmic bacteria are able to escape from autophagy process using two independent virulence factors, ActA and InlK. On the one hand, the recruitment of VASP, Arp2/3 complex and actin via ActA masks the bacteria from ubiquitination and autophagic recognition. On the other hand, MVP recruitment via InlK is also able to protect bacteria from ubiquitination and autophagic recognition. In that way, depending on ActA and InlK expression, four possibilities could be distinguished: (1) When neither ActA nor InlK are expressed, the bacterial ubiquitination is followed by p62 and LC3 recruitment, leading to autophagosome formation around the bacterium. (2) When ActA is expressed (e.g. wild-type bacterium (WT) grown in BHI before cell infection) it is sufficient for Listeria to escape from autophagy. (3) In contrast, in the absence of ActA, InlK efficiently protects bacterium against autophagy recognition via MVP recruitment. (4) Finally, when ActA and InlK are co-expressed by the bacterium, InlK rapidly recruits MVP at the surface of the bacterium. Then, in some instance, ActA replaces InlK leading to a switch of the bacteria disguised from MVP to actin. The model is partially based on the results of Yoshikawa et al [11]. doi:10.1371/journal.ppat.1002168.g007 Pasteur animal experimentation committee which comply with European regulations (directive 2010/63/EU of the European parliament and of the council of 22 September 2010 on the protection of animals used for scientific purposes). Figure S1 In vitro, in cellulo and in vivo expression of inlK. A. Coomassie staining of purified InlK recombinant protein and bovine serum albumine (BSA). B. In cellulo and in vitro expression of inlK revealed by bioluminescence. Left panel: RAW 267.4 macrophages and HeLa epithelial cells were infected for 4 h with wild-type L. monocytogenes EGD-e that contain a bioluminescent reporter of either inlK promoter [EGD-e-(pPL2-P inlKlux ABCDE )] or hly promoter [EGD-e-(pPL2-P inlK -lux ABCDE )], and submitted to photon detection with IVIS 100 (Xenogen/Caliper) system. Right panel: EGD-e-(pPL2-P inlK -lux ABCDE ) and EGD-e-(pPL2-P inlK -lux ABCDE ) were grown in BHI to OD 600 1.0 and submitted to photon detection with IVIS 100 (Xenogen/Caliper) system. C. In vivo expression of inlK revealed by bioluminescence. Left panel: Five BALB/c mice were i.v. infected with either EGDe-(pPL2-P inlK -lux ABCDE ) or EGD-e-(pPL2-P inlK -lux ABCDE ). Each 24 h mice were anesthetized and submitted to photon detection with IVIS 100 (Xenogen/  [44]. B. Detection of MVP recruitment at the surface of InlK overexpressing bacteria that do no recruit actin. HeLa cells were transfected with MVP-GFP (green) and actin-CFP (yellow), infected with InlK over-expressing Listeria (DinlK+pPRT inlK) for 4 h, fixed for fluorescence light microscopy, and stained with anti-InlK (blue) and anti-ActA (red) antibodies. MVP-GFP and actin-CFP and their respective bacterial interactors, InlK and ActA, are never co-recruited Inset regions are magnified. Inset region 1 represents an MVP-GFP positive bacterium which is also labeled for InlK but not for actin-CFP and ActA. Opposingly, the inset 2 represents a bacterium that recuits actin-CFP which is also labeled for ActA, but not for MVP-GFP and InlK. The scale bar represents 1 mm. C. Detection of MVP recruitment at the surface of intracytosolic InlK over-expressing bacteria. HeLa cells were transfected with MVP-CFP (red) and YFP-CBD (green), infected with InlK over-expressing Listeria (DinlK+pPRT inlK) for 4 h, fixed for fluorescence light microscopy and stained with phalloidin (blue). MVP positive bacteria were also labeled with YFP-CBD revealing that MVP was recruited by intracytosolic bacteria after the lysis of the internalization vacuole. Inset regions are magnified. The scale bar represents  Video S1 Kinetics of MVP and actin recruitment at the surface of InlK over-expressing bacteria. HeLa cells were transfected with MVP-tomato (red) and actin-GFP (green), infected with InlK over-expressing Listeria (DinlK+pPRT-inlK) for 4 h, and prepared for real-time video microscopy. Image series were collected every 15 min for 2 h. (AVI) Video S2 Kinetics of autophagy escape for MVP positive Listeria. Jeg3 cells were transfected with MVP-tomato (red) and GFP-LC3 (green), infected with InlK over-expressing Listeria (DinlK+pPRT-inlK) for 4 h, and prepared for real-time video microscopy. Image series were collected every 5 min for 2 h. The video shows that the GFP-LC3 membranous aggregate detaches from the MVP positive Listeria.

(AVI)
Video S3 Kinetics of autophagy escape for MVP positive Listeria. Jeg3 cells were transfected with MVP-tomato (red) and GFP-LC3 (green), infected with InlK over-expressing Listeria (DinlK+pPRT-inlK) for 4 h, and prepared for real-time video microscopy. Image series were collected every 5 min for 2 h. The video shows that the GFP-LC3 membranous aggregate on MVP positive bacteria does not lead to an autophagosome formation, whereas those bacteria efficiently divided. (AVI)