Listeria InlB Expedites Vacuole Escape and Intracellular Proliferation by Promoting Rab7 Recruitment via Vps34

ABSTRACT Rapid phagosomal escape mediated by listeriolysin O (LLO) is a prerequisite for Listeria monocytogenes intracellular replication and pathogenesis. Escape takes place within minutes after internalization from vacuoles that are negative to the early endosomal Rab5 GTPase and positive to the late endosomal Rab7. Using mutant analysis, we found that the listerial invasin InlB was required for optimal intracellular proliferation of L. monocytogenes. Starting from this observation, we determined in HeLa cells that InlB promotes early phagosomal escape and efficient Rab7 acquisition by the Listeria-containing vacuole (LCV). Recruitment of the class III phosphoinositide 3-kinase (PI3K) Vps34 to the LCV and accumulation of its lipid product, phosphatidylinositol 3-phosphate (PI3P), two key endosomal maturation mediators, were also dependent on InlB. Small interfering RNA (siRNA) knockdown experiments showed that Vps34 was required for Rab7 recruitment and early (LLO-mediated) escape and supported InlB-dependent intracellular proliferation. Together, our data indicate that InlB accelerates LCV conversion into an escape-favorable Rab7 late phagosome via subversion of class III PI3K/Vps34 signaling. Our findings uncover a new function for the InlB invasin in Listeria pathogenesis as an intracellular proliferation-promoting virulence factor.


RESULTS
InlB is required for efficient intracellular proliferation. Isogenic L. monocytogenes in-frame inlA and inlB deletion mutants were tested in human epithelial HeLa cells using gentamicin protection assays. Since the intracellular population at a given time point depends directly on the initial number of internalized bacteria, the multiplicity of infection (MOI) was adjusted such that each strain had comparable intracellular numbers at t = 0 (average of ;5 Â 10 4 colony forming units [CFU]/well). In addition, the intracellular proliferation data were normalized to the bacterial counts at t = 0 using an intracellular growth coefficient (IGC) (27) (see Materials and Methods).
As shown in Fig. 1A, while the DinlA mutant behaved as the wild type (WT), a distinct intracellular growth defect was observed for the DinlB mutant. This was reproduced in other cell lines (human intestinal Caco-2 cells, murine J774 macrophages) (see Fig. S1 in the supplemental material) and corroborated early observations in mouse hepatocytes (31).
The effect was not attributable to impaired bacterial fitness, since the DinlB mutant showed no growth defect in broth (Fig. S1B). Complementation of DinlB with the inlB gene on a plasmid, but not empty vector, rescued InlB production (Fig. S1C), normal cell entry (Fig. S1D), and WT intracellular proliferation (Fig. 1A), linking the observed phenotype to InlB.
L. monocytogenes proliferation in a cell monolayer is dependent on both successful escape to the cytosol and actin-based spreading to adjacent cells. The contribution of cell-to-cell spread to the DinlB proliferation defect can be tested using a strain lacking ActA (8). While, as expected, overall proliferation was generally reduced in the absence of ActA due to the spread deficiency, intracellular growth was significantly diminished in a DactA DinlB double mutant compared to the DactA single mutant. Again, complementation with inlB, but not empty vector, reverted the proliferation defect (Fig. 1B). Together, these data indicated that the absence of InlB impairs L. monocytogenes intracellular proliferation due to effects upstream of cell-to-cell spread.
InlB promotes early vacuole escape. The DinlB phenotype could be due either to a reduced ability to replicate within the cytoplasm or to an earlier defect in vacuole escape. To test this, the proportion of vacuole-associated and cytosolic bacteria was monitored over a 90-min infection time course in HeLa cells using two separate microscopy-based assays. Vacuole-associated bacteria were identified by colocalization with FM 1-43FX, a fluorescent membrane probe that constitutively labels endosomes and phagosomes (32,33). The vacuole marker was associated as early as 10 min after infection with WT L. monocytogenes, with a peak at 20 to 30 min before rapidly dropping to ;20% of the bacterial population by 90 min ( Fig. 2A and B). Although entering in fewer numbers (Fig. S2B), the DinlB bacteria that successfully invaded the monolayer showed similar FM 1-43FX colocalization dynamics during the first 30 to 45 min of infection. Subsequently, however, the DinlB mutant remained membrane associated in high numbers, the percentage only dropping by 20% over the rest of the time course ( Fig. 2A and Fig. S2A). To quantify the phenotype while controlling for variations in cell entry rate, vacuolar persistence (VP) was calculated based on the length of time between 50% initial bacterial association and 50% dissociation with the membrane probe relative to peak values. The VP values were 34.2 6 8.0 and 69.6 6 5.0 min for the WT and DinlB mutant, respectively ( Fig. 2A, right), suggesting the latter had a vacuole escape defect. In parallel, cytosolic bacteria were identified both through ActA-mediated F-actin accumulation (8) and decoration with a transfected fluorescent probe that binds to Listeria after escape to the cytosol (CBD, a listeriophage endolysin cell wall-binding domain fused to yellow fluorescent protein [YFP]) (5). Consistent with the vacuole association data, significant decoration with the CBD probe was observed later for the DinlB mutant than the WT. The dynamics of actin accumulation around bacteria mirrored the CBD probe data ( Fig. 2C and D). With both approaches, the mutant phenotype reverted to WT in the inlB-complemented DinlB mutant ( Fig. 2A and C and Fig. S2A and B). These data indicated that InlB is required for efficient vacuole escape in HeLa cells.
InlB promotes Rab7 recruitment to the Listeria-containing vacuole (LCV). To explore the mechanism of DinlB's delayed escape, different endosomal trafficking markers were probed over the infection time course in HeLa cells. Consistent with previous findings in macrophages (5), the key regulator of early to late endosome maturation, Rab5 (34,35), was not found to localize to WT LCVs, and the same was observed for DinlB (Fig. S3A). A positive signal was clearly visible on control bacteria-containing vacuoles known to associate with Rab5 (Rhodococcus equi and Staphylococcus aureus) (Fig. S4A), excluding a technical issue with the antibody. Early endosomal antigen-1 (EEA1) (36) could be seen for both WT and DinlB bacteria, but only at 10 min after infection and for a small proportion of LCVs (;10%) ( Fig. 3A, left, and Fig. S3B). Thus, EEA1 might have a role during the initial stages of Listeria endocytosis, as previously observed (25), but apparently only transient and independent of InlB.
Differences between the WT and DinlB mutant were evident, however, for the Rab7 late endosomal marker. Rab7 associated with WT LCVs soon after internalization, as previously seen in macrophages (5). The Rab7 GTPase was detected on a large proportion of phagosomes 10 min after infection, with a peak at 20 min, while it was absent 25 min later, coinciding with vacuole escape. In contrast, DinlB LCVs showed late acquisition of Rab7, with a flatter peak at 30 min followed by prolonged association up to 90 min. The phenotype could be rescued using the inlBcomplemented mutant, indicating that it was due to InlB expression (Fig. 3A, right, and Fig. S4B). This echoed the vacuole escape experiments showing that the DinlB mutant reached the cytoplasm much later than WT (Fig. 2). DinlB was not found to colocalize with the lysosomal-associated membrane protein 1 (LAMP1), implying that although delayed in escape, the mutant was still able to avoid lysosomal fusion (Fig. S3C). Collectively, these data indicated that InlB promotes Rab7 recruitment to the LCV, and this correlated with InlB-dependent early vacuole escape.
InlB-dependent PI3P accumulation and class III PI3K (Vps34) recruitment. The phosphoinositide PI3P plays a key role in the conversion of early endosomes/phagosomes into Rab7-positive vesicles (37,38) and was previously detected in the LCV of macrophages (5). To determine whether InlB had an effect on PI3P levels in HeLa LCVs, infection time course experiments were conducted in cells expressing a fluorescent PI3P probe (GFP-PX, consisting of the PI3P-binding PX domain from the vacuolar NADPH oxidase subunit, p40 phox , tagged with green fluorescent protein) (39). Accumulation of PI3P was observed after only 10 min in WT LCVs, with a peak by 20 min, declining by 30 min, and almost complete disappearance by 45 min after infection. In marked contrast, PI3P remained undetectable in DinlB LCVs during the 90-min time course. This again reverted to WT upon inlB complementation, indicating that, as for Rab7, PI3P accumulation was linked to InlB (Fig. 4A, left, Fig. 4B, and Fig. S5A). PI3P is primarily generated in endosomes and phagosomes by vacuolar protein sorting-34 (Vps34), the sole mammalian class III PI3K isoform (40,41). Parallel experiments in which HeLa cells were stained with a Vps34 antibody showed Vps34 localizing to WT and inlB-complemented DinlB LCVs as early as 10 min after infection, while remaining absent from DinlB LCVs (Fig. 4A, right, Fig. 4C, ). This pattern mirrored the PI3P data, consistent with the absence of PI3P being due to a lack of Vps34 recruitment. Thus, the phagosomal escape deficiency caused by absence of InlB was associated with a lack of effective Vps34 recruitment and accumulation of its lipid product on the LCV. Vps34 supports InlB-dependent intracellular proliferation but not entry. If InlBdependent intracellular proliferation involves class III PI3K Vps34 activity, a pan-PI3K inhibitor such as wortmannin (42) should abolish it. Wortmannin treatment of preinfected HeLa cells inhibited L. monocytogenes intracellular proliferation to an extent similar to that seen in the absence of InlB. The effect was clearly dependent on InlB, as wortmannin did not affect the intracellular proliferation of the DinlB mutant, while significant inhibition was observed with the inlB-but not mock-complemented mutant (Fig. 5A, left). Wortmannin had a similar InlB-dependent inhibitory effect in Caco-2 cells and J774 macrophages (Fig. S6A).
To pinpoint the specific PI3K isoform involved, we first used isoform-specific drugs that inhibit class I PI3K. HeLa cells were treated with PIK75 (PI3Ka), TGX-221 (PI3Kb), AS252524 (PI3Kg ) and PI-103 (PI3Ka,b,d ) (42)(43)(44), and PI3K inhibition efficacy was confirmed by Western blotting for pAktS473 (45) (Fig. S6B). None of these inhibitors , indicating that the effect caused by wortmannin might be linked to non-class I PI3K. We next tested the involvement of class II and class III PI3K using small interfering RNA (siRNA) knockdown experiments. The siRNA treatments resulted in effective depletion of class II (PIKC2a and PIK3C2b; PIK3C2g did not appear to be expressed in HeLa) and class III (Vps34) PI3K. Similar targeting of the class I PI3K regulatory subunit PIKR1 as a control resulted in a partial knockdown of p85 and a complete loss of class I PI3K activity as determined by pAktS473 Western blotting (Fig. S6D). Transfection with siRNA to the Met receptor required for InlB-mediated entry (17) was also used as an additional control. Both Listeria invasion and intracellular proliferation were investigated in the siRNA-transfected HeLa cells. While as expected (17,24), both siPIKR1 and siMet significantly reduced cell invasion (60% and 80%, respectively), siRNAs to class II and class III PI3Ks had no effect, indicating they were not involved in the entry process (Fig. S6E). Conversely, in Vps34-, but not class I or class II PI3K-depleted cells, intracellular proliferation was significantly impaired for WT Listeria (Fig. 5A, right, and Fig. S6F), indicating that class III PI3K did indeed have a role in cell infection downstream entry. None of the siRNAs, including siVps34, had any effect on the already reduced intracellular proliferation of the DinlB mutant. As seen with wortmannin, complementation with inlB rescued the proliferation inhibition by siVps34 (Fig. 5A, right). Overall, these data indicated that Vps34 required InlB Endosomal Vps34 Subversion by Listeria InlB mBio to exert its effect, suggesting that InlB and the class III PI3K act together to promote Listeria intracellular proliferation. Vps34 is required for Rab7 recruitment and efficient vacuole escape. To establish whether the (InlB-dependent) effect of Vps34 on intracellular proliferation was linked to vacuole escape efficiency, siVps34-or control siRNA-treated HeLa cells were infected with WT L. monocytogenes and LCVs were monitored over 90 min using either FM 1-43FX or Rab7 antibody. Whereas bacteria appeared to rapidly disengage with LCVs in control cells (as determined by dissociation with the membrane probe and association with F-actin), they remained membrane associated for up to 90 min after infection in Vps34-depleted cells ( Fig. 5B and Fig. S7A). Additionally, in contrast to control cells, no Rab7 could be seen around LCVs at any time point in siVps34-treated cells ( Fig. 5C and Fig. S7B). These effects were essentially similar to, albeit more severe than, those observed with the L. monocytogenes DinlB mutant (Fig. 3A, right, and Fig. 3B), supporting the notion that InlB promotes vacuole escape by subverting LCV membrane trafficking via Vps34. These data also demonstrated the importance of Vps34 in Rab7 recruitment to the Listeria phagosome.
Rapid escape requires Rab7 and is not hindered by Rab5 inhibition. Listeriae appear to escape from Rab7-positive phagosomes (see reference 5 and data herein), and InlB/Vps34-promoted vacuole escape correlated with Rab7 recruitment to the LCV (this study). To test the role of Rab7 in InlB-facilitated escape, HeLa cells were transfected with control or Rab7-specific siRNAs and infected with WT or DinlB L. monocytogenes ( Fig. 6A to E). Since the acidic pH of the maturing Rab7-positive phagosome is thought to be critical for efficient LCV disruption by LLO (5,6), an isogenic Dhly mutant was also included to determine whether Rab7 recruitment is linked to LLO-mediated escape. Although inefficiently, escape may occur in human epithelial cells in the absence of LLO, resulting in LLO-independent intracellular growth (46). We therefore verified that LLO was required for rapid vacuole escape under our conditions during the infection time course experiments in HeLa cells (Fig. 6A to D and Fig. S8A and B). Rab7 depletion correlated with significantly reduced intracellular growth of WT L. monocytogenes (Fig. 6A), while it did not affect internalization (not shown). Similarly reduced intracellular growth was observed for both DinlB and Dhly mutants in control cells, with no significant further change in Rab7-depleted cells (Fig. 6A). Proliferation defects correlated with significantly impaired vacuole escape, as determined by fluorescence microscopy using FM 1-43FX (Fig. 6B) and F-actin staining (Fig. 6C). As expected, Rab7 was not detected in siRab7-treated cells, while in control cells, the Rab7 acquisition pattern of WT and DinlB LCVs reproduced our previous results ( Fig. 6D; see also Fig. 3A,  right). No defect in Rab7 acquisition, but rather a general increase, was observed for Dhly LCVs, followed by a protracted association consistent with a vacuole escape deficiency (Fig. 6D). Together, these data suggest that the effects of InlB on intracellular proliferation depend on Rab7 recruitment, in turn promoting rapid, LLO-mediated vacuole escape.
Although acquisition of the early endosomal Rab GTPase Rab5 normally precedes that of Rab7 in the endosome/phagosome maturation sequence (34,37), Rab5 could not be recognized on HeLa (our data) or mouse macrophage (5) LCVs even immediately after infection. While this may reflect only poor or transient recruitment, evidence in macrophages suggested that Rab5 might be excluded from the LCV (5). We therefore explored in a final set of experiments the role of Rab5 in Listeria intracellular infection. Rab5 was inhibited in HeLa cells prior to infection either by transfection with Rab5-S34N, a dominant-negative mutant in which the GTPase is locked in GDP-bound inactive form (47), or by using an siRNA pool targeting Rab5 isoforms A, B, and C ( Fig. 6J and Fig. S9A). Proliferation of WT L. monocytogenes was not impaired, suggesting that entry into the cytosol was not significantly affected when Rab5 function was perturbed. The defective proliferation phenotype of the DinlB mutant also remained unaffected by Rab5 inhibition (Fig. 6F and Fig. S9B). Microscopic analysis of siRNAtreated cells confirmed that vacuole escape was not delayed in Rab5-depleted cells ( Fig. 6G and H). Furthermore, no interference with Rab7 recruitment was observed in siRab5-treated cells (Fig. 6I), indicating that Rab5 does not seem to play a significant role in LCV's maturation to the Rab7-positive stage.

DISCUSSION
The InlB protein is a well-known Listeria invasin that promotes entry into nonphagocytic cells via hijacking of the ligand-dependent Met receptor endocytic pathway and downstream activation of type I PI3K (18,24,25,48). Here, we provide evidence in HeLa cells that InlB has an additional key function at later stages of intracellular infection, accelerating vacuole escape, presumably through subversion of class III PI3K signaling. InlB is sufficient to mediate this process, as shown using gene complementation analysis and pharmacological and siRNA inhibition studies. The role of InlB in LCV maturation and vacuole escape is not unique to HeLa cells as similar effects were observed in Caco-2 human enterocytes (see Fig. S10). InlB has been reported to support Listeria infection in epithelial cells by modulating transcriptional host cell responses via SIRT2-mediated histone H3K18 deacetylation (49). Recently, InlB-induced c-Met/class I PI3K (PI3Ka) signaling has been also found to facilitate L. monocytogenes in vivo persistence and invasive (e.g., brain) infection by inhibiting Fas-mediated killing of infected macrophages (50). Together with these studies, our findings highlight the growing evidence that bacterial virulence factors are multifunctional, encapsulating multiple key host-cell modulatory activities to support infection.
Cell invasion permits Listeria to evade extracellular host defenses while vacuolar escape is essential for intracellular survival (3,6). Since cytosolic replication relies on successful vacuole escape, optimization of the latter by InlB maximizes listerial intracellular proliferation. In vivo survival also crucially depends on the ability of L. monocytogenes to avoid autophagy and the cytotoxic immune response directed against infected cells via a "runaway" strategy involving ActA-mediated actin-based cell-to-cell spread after cytosolic replication (3, 10). The rapid amplification of the bacterial load is clearly an essential component of this strategy because it increases the chances of Endosomal Vps34 Subversion by Listeria InlB mBio spreading before the infected cell is targeted by host defenses. Indeed, L. monocytogenes possesses a dedicated virulence factor to promote rapid intracellular replication, the hexose phosphate transporter Hpt (7). The vacuolar escape-accelerating activity of InlB here identified (together with its recently reported role in protecting infected monocytes from CD8 1 -mediated killing) (50) can therefore be seen as contributing, alongside Hpt, to provide L. monocytogenes a competitive edge against host immunity. Strikingly, InlB targets both extracellularly and intracellularly the endocytic pathway through interference with PI3K signaling. It seems unlikely that InlB-dependent class III PI3K (Vps34)-promoted escape is directly linked to the stimulatory effect of InlB on class I PI3K during entry because Vps34 is not known to be downstream of the Met receptor. Indeed, our data show that Vps34 supports InlB-dependent intracellular proliferation but not entry. Moreover, InlB/Vps34-promoted vacuole escape was resistant to class I PI3K isoform inhibition. Rather, these two distinct InlB functions appear to be spatiotemporally coordinated, since the hijacking of the endocytic pathway occurs at two sequential levels involving stage-specific PI3K isoforms (51) coincident with consecutive steps in Listeria intracellular pathogenesis. After Met/class I PI3K-mediated internalization and phagocytic cup closure, a receptor-free fraction of InlB, or InlB in complex with Met before sorting for lysosomal degradation (21,52), might direct Vps34 recruitment to the early LCV. Cotransfection experiments demonstrated colocalization of Vps34 and PI3P in InlB-positive vesicles, but we could not detect binding between InlB and Vps34 by immunoprecipitation (data not shown). This may reflect either an interaction too weak to be captured or that InlB acts upstream of Vps34.
The catalytic subunit of class III PI3K Vps34 is an effector of the small GTPase Rab5 in the endocytic pathway and localizes to nascent endosomes in complex with several regulatory subunits through interaction with membrane-bound GTP-Rab5 (41,51). Via Vps34, Rab5 controls recruitment of PI3P-binding proteins necessary for early endosome sorting and progression to the Rab7 late endosome stage (35,41,51,53,54). Intriguingly, however, although Vsp34 was obviously present, we could not detect Rab5 in the HeLa LCVs, recapitulating previous observations in macrophages (5). The FYVE-domain (PI3P-binding) Rab5 effector EEA1 requires not only PI3P but also corecognition by GTP-Rab5 for targeting to early endosomes (35,51,55). EEA1 was barely detectable in the Listeria phagosomes regardless of significant PI3P accumulation, consistent with Rab5 being only transiently or poorly recruited to the LCV. Indeed, in contrast to Vps34 knockdown, inhibition of Rab5 did not impede Rab7 acquisition by the LCV (or Listeria escape/intracellular proliferation), suggesting that Rab5 has a marginal or accessory role in the process.
Our findings do not seem to support the previous proposition that L. monocytogenes actively recruits and inhibits Rab5 to block LCV maturation and delay phagolysosome fusion (56,57). Indeed, the late endosomal marker Rab7 is rapidly recruited to the LCV after Listeria entry in human epithelial cells (our data) and murine macrophages (5), inconsistent with phagocytic vacuole maturation being arrested. On the contrary, L. monocytogenes appears to expedite the process via InlB-promoted recruitment of Vps34, driving LCVs to a Rab7-positive, prelysosome (LAMP1-negative) stage. This might be facilitated by diminished Rab5 levels/EEA1-mediated fusion with other (bacterium-free, Rab5-loaded) early endosomes (34,54,58), as this would contribute to isolating LCVs from normal endosomal traffic. The experiments using inlB and hly mutants in Rab7-depleted cells indicated that rapid (InlB-promoted) escape takes place from Rab7-positive vacuoles and that Rab7 facilitates LLO-mediated escape, likely because of the acidified intraluminal environment of the maturing endosome (6). Consequently, InlB could be seen as acting intracellularly in concert with the low-pH-activated LLO toxin to achieve efficient vacuole escape and avoid lysosomal killing.
The apparent lack of Rab5 recruitment to the LCV is puzzling, as the coordinated replacement of Rab5 by Rab7 appears to be key in the early-to-late endosome conversion process (34,54,59,60). According to the current model, upon accumulation of PI3P, the Mon1/Sand1-Ccz1 protein complex displaces the Rab5-activating guanine nucleotide exchange factor (GEF) Rabex5, causing membrane dissociation of Rab5, and recruits/activates Rab7 through interaction with the HOPS (homotypic fusion and vacuole protein sorting) complex in association with the Rab7 GEF, Ccz1 (38,54,59,61). Although GTP-Rab5 interacts with Mon1 and HOPS components (34,38,62), it is unclear whether Rab5 is strictly needed in the process once PI3P is generated by its downstream effector Vps34, because the Mon1-Ccz1 complex binds to PI3P (63,64). Conceivably, therefore, L. monocytogenes might directly promote Rab7 accumulation without a sustained Rab5 input via InlB-modulated Vps34 recruitment and concomitant PI3P generation (Fig. 7). It is worth noting, however, that even in the absence of InlB, and thus of detectable Vps34 and PI3P accumulation, Rab7 acquisition still occurred in the LCV, albeit more slowly. This observation is consistent with previous data showing significant Rab7 recruitment to phagosomes in wortmannin-treated cells (37), implying that the phagosome maturation process can be, at least partially, independent of Vps34 activity (65). Nevertheless, our siRNA experiments support a crucial role for Vps34 because its depletion appeared to prevent Rab7 recruitment to the LCV. Since via a regulatory loop Vps34 negatively regulates Rab5 during endosome maturation (66), InlB-promoted Vps34 recruitment itself might have a role in the lack  (41,51,54,59,60), and further research is needed to clarify the mechanism underlying the apparent Rab5 exclusion and how InlB promotes Vps34 and Rab7 recruitment in the LCV. Avoidance of lysosomal killing by vacuolar pathogens often involves interference with class III PI3K/PI3P signaling, but the specific underlying mechanism differs for each pathogen (67,68). M. tuberculosis, for example, causes early phagosome maturation arrest through depletion of PI3P by the secreted SapM phosphatase as well as inhibition of Vps34 by mycobacterial cell envelope components. Rab5 is retained, but EEA1 and Rab7 are excluded, blocking acquisition of the acidifying vacuolar proton pump ATPase and lysosomal fusion (69)(70)(71)(72). Salmonella, in contrast, promotes homotypic fusion of PI3P-enriched, Rab5-and EEA1-positive vesicles to form a large vacuole in which the bacteria proliferate. This is mediated by the type III secreted effector SopB, which promotes PI3P generation by dephosphorylating phosphatidylinositol diand triphosphates and through recruitment of Rab5 and Vps34 activation (73,74). Here, we show that L. monocytogenes, via InlB, promotes the recruitment of Vps34 and induces PI3P generation on the LCV, boosting the formation of an escape-favorable Rab7 late phagosome. Our data offer a novel example of bacterial modulation of endosomal trafficking, unexpectedly in a cytosolically replicating pathogen that only very transiently interacts with the phagocytic vacuole after internalization.

MATERIALS AND METHODS
Bacteria, plasmids, antibodies, and reagents. Isogenic derivatives of L. monocytogenes serovar 4b clinical isolate P14 with constitutive in vivo-like virulence gene expression (75) were used in this study. The construction of DinlA and DinlB in-frame deletion mutants is described below, the DactA and Dhly mutants have been reported elsewhere (27,76). The R. equi 103S and S. aureus USA300 strains used as controls for Rab5 detection in bacterial phagosomes were from our bacterial isolate collection and R. Fitzgerald's laboratory (Roslin Institute, University of Edinburgh), respectively. Mammalian expression plasmids pEYFP-CBD, encoding the Listeria cell wall-binding cytosolic probe, and pGFP-iPX, encoding the PI3P-binding probe, were kindly provided by J. Swanson's lab (University of Michigan Medical School, Ann Arbor, MI, USA) and the L. Stephen/P. Hawkins's lab (Babraham Institute, Cambridge, United Kingdom), respectively (5,39). The plasmid pHinlB used for complementation of the DinlB mutant is a derivative of the E. coli Gram-positive shuttle vector pHPS9 (77) containing the inlB gene. The insert was prepared by PCR amplification of inlB and upstream inlAB operon regulatory regions from L. monocytogenes PAM14 DinlA genomic DNA using oligonucleotides mainl13SalI13 (ACGCGTCGACGAACATAAAGGGTAGAGG) and inlBBamHI (CGGGATCCCGATTCTTGCTAGACCACC), which carry SalI and BamHI restriction sites (underlined), respectively. After cloning into the pTOPO cloning T-vector (Invitrogen), the SalI/BamHI insert was transferred to pHPS9.
Reagents, antibodies, and oligonucleotides. Chemicals were from Sigma-Aldrich unless stated otherwise. Class I PI3K inhibitors PIK-75, TGX-221, AS-604850, and PI-103 were from Cayman Chemicals. FM 1-43FX membrane probe was from Life Technologies. Antibodies to Akt, pAktS473, Rab7, Rab5, EEA1, and Vps34 were from Cell Signaling Technology, GAPDH (glyceraldehyde-3-phosphate dehydrogenase) from Chemicon, p85 from Millipore, PI3KC2a and PI3KC2b from BD Biosciences, PI3KC2g from Abgent, LAMP1 (H4A3) from Santa Cruz, and horseradish peroxidase (HRP)-conjugated-mouse and rabbit antibodies from Amersham. Specific antibodies to InlA and InlB were kindly provided by P. General DNA techniques. Restriction enzymes were used according to the manufacturer's instructions (Promega and NEB). Chromosomal DNA from Listeria was extracted and purified as previously described (75). Plasmid DNA was extracted from E. coli using the Qiagen plasmid purification kit and was introduced into E. coli or L. monocytogenes by electroporation using a GenePulserXcell apparatus (Bio-Rad). PCR was carried out with Taq DNA polymerase (Biotools, Madrid, Spain) for detection/mapping purposes or high-fidelity ProofStart DNA polymerase (Qiagen) for plasmid or mutant constructions. The reaction mixtures contained 100 ng of DNA template, 200 mM deoxynucleoside triphosphates (dNTPs), 0.25 mg oligonucleotide primers, 2.5 mM MgCl 2 , a suitable amount of polymerase buffer, and 1 U of polymerase per kb in a 25-mL volume. The standard amplification program was 3 min at 94°C, followed by 30 cycles of 15 s at 95°C, 30 to 60 s at 48 to 58°C, and 1 to 3 min at 72°C, and a final 3 min at Endosomal Vps34 Subversion by Listeria InlB mBio 72°C. PCR products were purified with the Qiagen PCR purification kit. DNA sequences were determined on both strands. Construction of L. monocytogenes inlA and inlB in-frame deletion mutants. Chromosomal gene deletions were generated by allelic exchange via homologous recombination as previously described (76). For the DinlA mutant, the recombinogenic plasmid was constructed by in-frame ligation of two PCR fragments containing small portions of the 59-and 39-terminal regions of the inlA gene plus part of the upstream and downstream flanking regions, respectively. Oligonucleotides inlA1 (ATTTGGATCCTAAAG GGTAGAGG) and inlA2 (CATACCCCGGGGCCAAATACT) carrying BamHI and XmaI/SmaI sites, respectively (underlined), were used to PCR amplify a 643-bp fragment from the 59 region of inlA, including the first 72 bp of the gene. Similarly, oligonucleotides inlA3 (TCACCCGGGAATTCAGCTAGC) and inlA4 (TGAACGG ATCCAATATCACTATTAT) carrying XmaI/SmaI and BamHI sites, respectively, were used to amplify a 709-bp fragment from the 39 region of inlA, including the last 174 nucleotides of the gene. The two PCR products were digested with XmaI and ligated, and the ligation product was used as a template for PCR amplification with oligonucleotides inlA1 and inlA4. The resulting amplicon was inserted into the pTOPO T-vector (Invitrogen) and then the BamHI fragment containing the DinlA construct was transferred to the thermosensitive shuttle vector pMAD (78) to give the recombinogenic plasmid pMDinlA. The in-frame-deleted allele for the DinlB mutant was generated by "recombinant PCR" (79). Oligonucleotide primers inlB1 (TCCTGTGGATCCACCAACAACT) and inlB2R (TACCGGAACTTTTGTCACTAGATCCGTCACAC) amplified a 573bp fragment from the 59 region of inlB, including the first 213 bp of the gene, and primers inlB3R (GATCTAGTGACAAAGTTCCGGTAGTAGATAGC) and inlB4 (GTGATGGATCCCACATTTTGGC) were used to amplify the 39 region of inlB. Primers inlB1 and inlB4 carried BamHI sites, and primers inlB2R and inlB3R had complementary sequences (underlined). The two PCR products were fused by splicing overlap extension using oligonucleotides inlB1 and inlB4, and the recombinant PCR product was processed as described above to produce the pMAD-based recombinogenic plasmid pMDinlB.
Mammalian cell cultures and transfections. HeLa cells (and Caco-2 and J774 cells for some control experiments) were sourced from ATCC, passaged at no more than 70% confluence, and used between passages 3 and 15. Cells were cultured in 24-well plates or on 13-mm glass coverslips at 37°C under 5% CO 2 in DMEM (Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum [FCS] and L-glutamine [Invitrogen]) without antibiotics, as previously described (45). HeLa cells were seeded at a density of 3.5 Â 10 4 or 5 Â 10 4 cells per well (24-well plates) or coverslip 24 h prior to transfection with siRNA or plasmid, respectively. Plasmid DNA or siRNAs diluted in optiMEM without FCS were premixed with optiMEM-diluted Lipofectamine 2000 reagent (Invitrogen) as per the manufacturer's instructions. Cells were transfected for 12 h in 1 mL DMEM, giving a final concentration of 80 to 200 nM oligonucleotide or 1 to 5 mg/mL plasmid DNA. Transfected cells were incubated for 24 h prior to infection, fixation for immunofluorescence microscopy, or production of cell lysates. Plasmid transfection efficiency was typically >85%.
Cell infections. Culture plates containing %1 Â 10 5 cells per well were inoculated with bacterial suspensions in DMEM and immediately centrifuged (180 Â g, 3 min) to synchronize infection. Internalization and intracellular proliferation assays were carried out using a gentamicin protection assay as previously described (27). After centrifugation, inoculated plates were incubated for 30 min (15 min for J774 macrophages) and washed several times with Dulbecco's phosphate-buffered saline (DPBS) (Invitrogen) followed by a further incubation in DMEM containing 100 mg/mL gentamicin for 30 min to kill the remaining extracellular bacteria. Samples were taken at this point to determine invasion (t = 0), and medium was replaced with fresh medium containing 10 mg/mL gentamicin to prevent extracellular bacterial growth. Where applicable, PI3K inhibitors were added at this stage diluted in 0.1% dimethyl sulfoxide (DMSO). At the sampling time points, cells were washed in DPBS, lysed in 10 mM Tris-HCl (pH 7.4), 1% Triton X-100, and viable bacteria plate counted. Intracellular proliferation data were normalized using an intracellular growth coefficient to correct for variation in cell entry using the formula IGC = (CFU tn 2 CFU t0 )/CFU t0 (27).
Western blotting. HeLa cells were processed for immunoblotting as detailed by Cain et al. (45), and bacterial proteins were prepared as previously described (76). Immunoreactive proteins were visualized using the ECL enhanced chemiluminescence detection system (Amersham).
Fluorescence microscopy. Coverslips for time course quantifications were collected 10, 20, 30, 45, and 90 min after plate centrifugation, thoroughly washed four times in warm PBS to remove extracellular bacteria, fixed with 3.7% (wt/vol) paraformaldehyde, and processed as described by Cain et al. (45). For time points 45 and 90 min, 100 mg/mL gentamicin was added to the medium 30 min after infection to prevent extracellular growth. Alexa Fluor 488-or 568-conjugated anti-rabbit or anti-mouse IgG secondary antibodies (1 h) were used for immunofluorescence staining, followed by Alexa Fluor 568-or 488-conjugated phalloidin (Invitrogen) and DAPI (49,6-diamidino-2-phenylindole) (Sigma) (20 min) to visualize F-actin and bacterial DNA/cell nuclei, respectively. To determine vacuole-associated bacteria, cells were incubated with 2 mM of the fixable fluorescent membrane probe FM 1-43FX (Life Technologies) in DMEM for 45 min, washed twice, and incubated with fresh DMEM for 30 min before infection. Bacteria that escaped from the vacuole were identified based on their association with either F-actin (46) or a transfected listerial cell wall-binding fluorescent cytosolic probe (yellow fluorescent protein [YFP]-CBD) (5). Images were acquired at room temperature using a Leica CTR-6000 fluorescence microscope with a 63Â oil immersion objective, collecting data as a z-stack, which were later flattened by deconvolution. Images were processed using Leica LAF software (Leica) and ImageJ (NIH; http://rsbweb.nih.gov/ij/), and figures were assembled using Adobe Photoshop.
Statistics. In all cases, data are shown as the mean 6 standard error of the mean (SEM) of values from a minimum of three independent experiments performed in triplicate. Data for image-based quantifications were collected from 60 to 70 bacteria/vacuoles across five randomly selected microscopic fields per time point per condition in each of three independent experiments. The statistical significance of data was determined by Student t tests using Prism software (http://www.graphpad.com/scientific -software/prism/).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.