Listeria monocytogenes Exploits Host Caveolin for Cell-to-Cell Spreading.

Listeria monocytogenes moves from one cell to another as it disseminates within tissues. This bacterial transfer process depends on the host actin cytoskeleton as the bacterium forms motile actin-rich membranous protrusions that propel the bacteria into neighboring cells, thus forming corresponding membrane invaginations. Here, we examine these membrane invaginations and demonstrate that caveolin-1–based endocytosis is crucial for efficient bacterial cell-to-cell spreading. We show that only a subset of caveolin-associated proteins (cavin-2 and EHD2) are involved in this process. Despite the absence of clathrin at the invaginations, the classical clathrin-associated protein epsin-1 is also required for efficient bacterial spreading. Using isolated L. monocytogenes protrusions added onto naive host cells, we demonstrate that actin-based propulsion is dispensable for caveolin-1 endocytosis as the presence of the protrusion/invagination interaction alone triggers caveolin-1 recruitment in the recipient cells. Finally, we provide a model of how this caveolin-1–based internalization event can exceed the theoretical size limit for this endocytic pathway.

L isteria monocytogenes kills ϳ25% of the people that it infects (1). This microbe uses clathrin-mediated endocytosis to initially invade a target host cell (2), resulting in vacuole-residing bacteria. Secretion of pore-forming toxins from L. monocytogenes bacteria that include listeriolysin O (LLO) enables the microbe to escape the vacuole and reside in the host cell cytoplasm (3,4). Once free in the host cell cytosol, L. monocytogenes initiates several processes to spread into neighboring cells, thus propagating infection. To accomplish this, the bacterium generates actin-rich structures (comet/rocket tails) at one of its poles, which enables its intracellular motility (3; see also reference 5 for a review). When in close opposition to the host cell plasma membrane, these actin-rich tails provide the propulsive force necessary to distend the host plasma membrane, forming bacterium-led actin-rich protrusions that can extend up to 100 m (3, 6; see also references 7 and 8 for reviews). The force generated by these structures is thought to drive the bacterium into neighboring cells. This idea of an indispensable role of actin polymerization at the bacterial cell surface is supported by several observations of abolished intracellular and intercellular movement of L. monocytogenes bacteria within and among host cells treated with filamentous actin inhibitors (3,9,10). The subsequent uptake of L. monocytogenes membrane protrusions by neighboring host cells ultimately allows the bacteria to continue the disease process.
Classical endocytic mechanisms often exploit clathrin or caveolin as the key protein for the internalization of various particles (11)(12)(13). Although clathrin commonly internalizes material into 30-nm to 150-nm vesicles, bacteria (including L. monocytogenes) have devised strategies to control the clathrin-mediated endocytic machinery for their own large-scale endocytic events (2,14,15). Alternatively, caveolin-based endocytosis remains a process that internalizes extracellular material into bulb-shaped caveolae of a maximum size of ϳ100 nm (16)(17)(18).
Both clathrin endocytosis and caveolin endocytosis require an assortment of accessory proteins to form invaginations and ultimately cut the membranous neck at the terminal stages of the formation of the structures prior to vesicle release (13). While some of these accessory proteins (dynamin-2 [19,20] and actin [21,22]) are shared by both endocytic mechanisms, most remain segregated to their respective internalization pathways.
Fundamental mechanistic questions concerning how L. monocytogenes membrane protrusions are internalized into neighboring cells remain unanswered. In this study, we explored the involvement of the host endocytic machinery during L. monocytogenes infections and found that L. monocytogenes intercellular spreading relies primarily on caveolar elements (caveolin-1, cavin-2, EHD2) but also usurps phosphatidylserine, dynamin-2 and epsin-1.

RESULTS
To begin to unravel the endocytic mechanism used by L. monocytogenes bacteria as they move from cell to cell, we initially immunolocalized clathrin and caveolin-1 to invaginations generated by L. monocytogenes protrusions during MDCK cell infections. We found that caveolin-1, but not clathrin, accumulated at those sites (Fig. 1A). Morphologically, caveolin-1 delineated the entire invagination and accumulated as bright puncta along the structures (Fig. 1A). To confirm that the cell forming the bacterial invagination (receiving the L. monocytogenes membrane protrusion) was the source of the observed increase in caveolin-1 levels, we utilized a mixed-cell assay whereby infected HeLa cells were overlaid on uninfected cells that had also been transfected previously with fluorescently tagged caveolin-1. We found that caveolin-1-mCherry was recruited along the entire length of the invaginations when expressed in the invagination-forming cells (Fig. 1B). Conversely, there was no obvious enrichment of clathrin-green fluorescent protein (GFP) at the invaginations (Fig. 1B). To characterize caveolin-1 accumulation at these sites in more detail, we plotted the pixel intensity profiles of F-actin and caveolin-1 (or clathrin) based on a 1.5-m line drawn perpendicularly across the membrane protrusion/invagination. As seen in our analyses of Color intensities are enhanced in insets to clearly visualize the spreading events in MDCK and HeLa cells, endogenous caveolin-1 as well as the caveolin-1-mCherry signal at the structures generally presented as 2 peaks that were peripheral to a strong single F-actin peak originating from the actin-rich core of the membrane protrusions ( Fig. 1C and D; see also Fig. S1A to FЉ in the supplemental material). Neither endogenous clathrin nor clathrin-GFP generated the 2 characteristic peaks such as were observed with caveolin-1. Instead, clathrin generated a low-level signal across the entire structure ( Fig. 1E and F; see also Fig. S2A to FЉ). The increased signal of caveolin-1 at L. monocytogenes membrane invaginations did not coincide with alterations to endogenous caveolin-1 protein levels during 8-h or 24-h infections of MDCK or HeLa cells compared to the results seen with uninfected samples (Fig. 1G).
We recently identified CD147 as a marker of L. monocytogenes membrane invaginations (23). Thus, to quantify the frequency with which caveolin-1 delineates L. monocytogenes membrane invaginations, we coexpressed CD147-GFP with caveolin-1-mCherry in invagination-forming host cells. We found that ϳ95% of CD147-GFPpositive membrane invaginations were also enriched with caveolin-1-mCherry during infections of either HeLa or MDCK cells (Fig. S3A to D). We also examined CD147positive membrane invaginations using CD147-targeting antibodies and found that ϳ92% of (endogenous) CD147-positive invaginations were delineated with caveolion-1-mCherry in both HeLa and MDCK cell lines ( Fig. S3E to H). The empty mCherry vector did not localize to the CD147-positive membrane invaginations (Fig. S3). Taken together, these data point to caveolin-1 as another component of L. monocytogenes membrane invaginations.
Using the same mixed-cell assay, we continued to catalogue proteins used by the caveolin pathway that might also be hijacked at L. monocytogenes membrane invaginations and found only cavin-2 and EHD2 present at the structures ( Fig. 2A). Caveolar components such as cavin-1, cavin-3, and pacsin2 as well as the clathrin-associated proteins AP2 and Eps15 were absent from the L. monocytogenes membrane invaginations ( Fig. S4 and S5). Line scan analysis of cavin-2 and EHD2 at the invaginations resembled those performed on L. monocytogenes membrane invaginations labeled by caveolin-1, with 2 peaks present at the periphery of a single actin peak ( To further characterize L. monocytogenes membrane invaginations, we examined them at various stages of their progression using caveolin-1-mCherry localization. Caveolin-1-mCherry was present at the initial contact sites when the membrane protrusions interacted with the receiving cell membrane (Fig. 2C). A trailing streak of caveolin-1-mCherry was routinely observed when the invaginations elongated within the receiving cells. This presumably was labeling the plasma membrane prior to vacuole formation (Fig. 2C). Caveolin-1-mCherry maintained its presence following apparent scission events as it continued to surround the microbes contained within vacuole-like structures (Fig. 2C). Multiple bacteria were often found contained within those structures (Fig. 2C). Vacuolar localization of caveolar proteins also held true for cavin-2-GFP and EHD2-GFP ( Fig. 2D and E). Scale bars are 5 m or 2 m (inset). (B) Mixed HeLa cell assay demonstrating caveolin-1-mCherry (pseudocolored green) but not clathrin-GFP (green) concentrated at invaginations when expressed in L. monocytogenes protrusion-receiving cells. Samples were fixed and stained with Alexa 488-phalloidin (red) to visualize actin and with DAPI (blue) to visualize host DNA and bacteria within the invaginations. The white star indicates the location of the untransfected protrusion-sending cell. Insets are enlargements of the boxed regions. Color intensities are enhanced in insets to more clearly visualize the labeled proteins. Solid arrowheads indicate the invaginations, and open arrowheads indicate spreading bacteria. Scale bars are 5 m or 1 m (inset). (C to F) Line scan analysis of the L. monocytogenes membrane protrusion/invagination from samples stained with fluorescent phalloidin (red) to visualize F-actin and with DAPI (blue) to confirm the presence of the bacteria at the structures. The caveolin-1 (C and D) or clathrin (E and F) signal is representative of exogenous protein (pseudocolored green) from the mixed HeLa cell assays (C and F) or endogenous protein (green) from infections of MDCK cells (D and F). A 1.5-m line (white line) was drawn through the protrusion/invagination, and F-actin intensity and the corresponding caveolin-1 or clathrin intensity were plotted. Scale bar is 1 m. (G) Caveolin-1 protein levels are unaltered during L. monocytogenes infections. Whole MDCK or HeLa cell lysates from uninfected (Uninf) cells versus cells from 8 h (top blots) or 24 h (bottom blots) L. monocytogenes infections (Inf) were probed for endogenous caveolin-1 using rabbit polyclonal anti-caveolin-1 (Cav-1) antibodies. ␣-Tubulin is shown as a loading control. A.U., arbitrary units. Our observations of linear and punctate caveolin-1 localization at L. monocytogenes membrane invaginations suggested that at least a subset of these structures could have caveolae associated with them. To examine this, we used electron microscopy and found structures resembling caveolae at the tips of some of the invaginations (Fig. 2F). However, in most instances the invaginations lacked any noticeable caveolar structures (Fig. 2F=).
It is well known that caveolin and other caveolar components interact with or regulate phospholipid species at the plasma membrane (24)(25)(26)(27). Phosphatidylserine plays a key role in caveolae biogenesis and stability as well as in proper caveolin-1 clustering at the structures (28). We examined whether this lipid was also enriched at L. monocytogenes membrane invaginations. To visualize the distribution of phosphatidylserine in the invagination-forming cells, we utilized a previously generated phosphatidylserine sensor that makes use of the discoidin C2 domain of lactadherin fused to GFP (Lact-C2-GFP) (29). Using our mixed-cell assays, we saw an enrichment of the phosphatidylserine probe along the entire length of L. monocytogenes invaginations (Fig. 2G). Line scan analysis of Lact-C2-GFP at the structures also generated the characteristic dual peak surrounding actin ( Fig. S7A to AЉ). Although few studies have examined the precise lipid composition of caveolae, recent work from Román-Fernández and colleagues (2018) showed that PIP2 [phosphatidylinositol (3,4)bisphosphate] is a crucial determinant of caveolin-1-positive apical endocytic bodies (30). Thus, to further our examination of lipids at L. monocytogenes invaginations, we utilized the pleckstrin homology (PH) domain of Akt fused to mCherry (Akt-PH-mCherry) (31), which targets both PIP2 [phosphatidylinositol (3,4)-bisphosphate] and PIP3 [phosphatidylinositol (3,4,5)-trisphosphate]. Examination of L. monocytogenes invaginations revealed this probe along the entire structure (Fig. S8A). To confirm that the localization of these caveola-associated lipids at L. monocytogenes membrane invaginations was specific, we looked for lactosylceramide (LacCer), a lipid species known to be absent at caveolae (32). Using the mixed-cell assay, we saw no obvious enrichment of LacCer at the invaginations in cells that were prelabeled with the fluorescent LacCer probe (BODIPY-LacCer) (Fig. S8B).
Induction of plasma membrane curvature is crucial during endocytosis (33,34). This is often accomplished by the activity of the BAR domain-containing family of proteins such as amphiphysin 1 as well as the clathrin-associated protein epsin-1 (35)(36)(37)(38)(39). To determine whether these types of proteins are involved in the internalization of L.

Dhanda et al.
® monocytogenes during cell-to-cell spreading, we utilized the mixed-cell assay to examine epsin-1 and the BAR domain-containing proteins amphiphysin 1 and FCHO1 as well as other membrane curving proteins such as intersectin-1 and NECAP at L. monocytogenes invaginations. We found that only epsin-1 was present at these sites ( Fig. 3A and B). This was surprising as epsin-1 is generally recognized as a clathrin-dependent protein but was present despite the absence of clathrin at the invaginations. Finally, we used the mixed-cell assay to examine dynamin-2 and filamentous actin in the invagination-forming cells as both proteins are used during terminal endocytic processes (40,41). Interestingly, dynamin-2 did not concentrate at the scission point but instead localized predominantly around the bacterial region of the invagination (Fig. 3C). To examine the actin filaments in only the invagination-forming cells, we utilized a fluorescently tagged version of the small F-actin-binding peptide LifeAct (42). The LifeAct-associated actin filaments from those cells clearly accumulated around the entire length of the invagination (Fig. 3C). Line scan analyses of both epsin-1 and F-actin (LifeAct) at invaginations again produced 2 peaks, characteristic of those previously seen at the sites (Fig. 3D). Although several studies previously reported on the extent to which L. monocytogenes membrane protrusions can grow in size (6,43,44), there is a dearth of analysis on the size of the corresponding invaginations that are generated. To address this issue, we measured the lengths of 35 caveolin-1-mCherrypositive membrane invaginations where fusion of the membrane at their distal ends was morphologically identifiable. The average length of these structures was found to be 9.5 m (Fig. 3E).
We next set out to ascertain the functional importance of caveolin-1 during the cell-to-cell spreading of L. monocytogenes. Researchers have routinely used cholesteroldepleting agents such as methyl-␤-cyclodextrin and filipin to study caveolin/caveolamediated endocytosis (45)(46)(47)(48). However, it is well known that these agents can severely inhibit an assortment of endocytic processes, including clathrin-mediated endocytosis and lipid raft homeostasis (49)(50)(51)(52). To get around the broad action of these drugs, we generated a stably transfected short hairpin RNA (shRNA) caveolin-1 knockdown (KD) cell line (Fig. 4A). Western blot analysis of these cells indicated an ϳ95% reduction in caveolin-1 protein levels compared to control shRNA cells (Fig. 4B). We first examined the general appearance of the caveolin-1 KD cells by phase imaging and phalloidin staining and saw that single isolated cells as well as cell monolayers appeared morphologically indistinguishable from control shRNA cells (data not shown). Furthermore, caveolin-1 depletion did not visibly affect the endogenous protein levels of clathrin or epsin-1 ( Fig. 4C and D). We used these cells for bacterial intercellular spreading assays and found that cell-to-cell spreading of L. monocytogenes was significantly (ϳ70%) impaired relative to the control cell results ( Fig. 4E and F; see also Fig. S11A at https://figshare.com/s/8f0fd1b824a579d16cfa), with the majority of bacteria occasionally even restricted to a single host cell (Fig. 4G). To ensure that the observed defects in spreading from the caveolin-1 KD cells were not caused by bacterial replication or invasion deficiencies, we performed gentamicin protection assays. Examining bacterial loads from 3-h infections, we found that the caveolin-1 KD cells showed a marginal (ϳ1.13ϫ) and yet significant increase in bacterial loads compared to those from control shRNA cells (Fig. 4H). After 8-h infections, we found no significant differences between the bacterial loads obtained from caveolin-1 KD and control shRNA cells (see also Fig. S11B at https://figshare.com/s/8f0fd1b824a579d16cfa). Another potential factor which could affect bacterial spreading in the caveolin-1 KD cells is the proper formation of actin comet/rocket tails, as those structures are used to propel the bacteria toward the host cell periphery prior to membrane protrusion generation. We measured the length and linearity (or tortuosity [10]) of 60 comet/rocket tails generated in the caveolin-1 KD and control shRNA cell lines and found that the morphology of comet/rocket tails remained unchanged with caveolin-1 depletion ( Fig. 4I and J). Similarly, we analyzed the morphology of at least 40 L. monocytogenes membrane protrusions, as the proper formation of these structures is also required for efficient bacterial cell-to-cell dissemination (23,44,53,54), and found no significant differences in the length or tortuosity of the structures (Fig. 4K and L). Finally, we quantified the frequency of membrane protrusion formation and found that there was no significant difference in the number of protrusions generated in the caveolin-1 KD and control shRNA cell lines (see also Fig. S11C at https://figshare.com/s/8f0fd1b824a579d16cfa). These findings suggest that the spreading defects likely arise due to an irregularity at the invaginations. One potential explanation for this might be that the invaginations generated in caveolin-1 KD cells are shorter than those formed in control shRNA cells. To test this, we measured the length of invaginations generated in the caveolin-1 KD and control shRNA cell lines and found that the invaginations were ϳ47% shorter in the caveolin-1 KD cells than their counterparts generated in control cells (see also Fig. S11D at https://figshare.com/s/ 8f0fd1b824a579d16cfa).
The concentration of epsin-1 along the entire L. monocytogenes invagination phenocopied the caveolin-1 localization, and, similarly to caveolin-1, epsin-1 expression levels were unaltered compared to uninfected cells (Fig. 5A). To examine the importance of epsin-1 during L. monocytogenes cell-to-cell spreading, we utilized stably transfected small interfering RNA (siRNA) epsin-1 KD cells (Fig. 5B). Western blot analysis of these cells showed an ϳ92.5% reduction in epsin-1 protein levels compared to stably transfected control siRNA cells (Fig. 5C). L. monocytogenes intercellular spreading assays performed in these cells revealed a significant (ϳ25%) reduction in bacterial dissemination ( Fig. 5D and E; see also Fig. S11E to F at https://figshare.com/s/ 8f0fd1b824a579d16cfa). During these assays, we saw on occasion bacteria that were also restricted to a single host cell (Fig. 5F). Similarly to our previous finding with respect to caveolin-1 depletion, bacterial loads were augmented marginally (ϳ1.15fold), but statistically significantly, in the epsin-1 KD cells compared to control siRNA cells (Fig. 5G). To examine in more mechanistic detail the role of epsin-1 at invaginations, we used domain deletion mutant constructs to determine which parts of epsin-1 were involved in its localization to the structures (Fig. 5H). All but the ΔENTH mutant of epsin-1 maintained localization to invaginations (Fig. 5I). Line scan analyses of all epsin-1 constructs (but not the ΔENTH mutant of epsin-1) depicted the characteristic 2 peaks peripheral to a strong single F-actin peak coming from the membrane protrusions (see Fig. S12A at https://figshare.com/s/cf00a3963800acb09230). Epsin-1 ENTH interacts with plasma membrane phosphatidylinositol-4,5-bisphosphate [PtdIns(4,5)P2], and it is this interaction which permits epsin-1 to directly modify membrane curvature during endocytosis (35,39,55). When expressed alone, the ENTH domain of epsin-1   localized to L. monocytogenes invaginations, suggesting that the ENTH domain likely plays a crucial role during L. monocytogenes invagination formation (Fig. 5J). Line scan analysis of ENTH-GFP at the invagination also generated the characteristic dual peak surrounding actin (see Fig. S12B at https://figshare.com/s/cf00a3963800acb09230).
The prevailing hypothesis regarding the mechanism behind L. monocytogenes cellto-cell transfer largely hinges on the actin-based motility of the microbes and the resulting propulsive forces generated at their cell surface (3,43,56). While it is possible that our observed recruitment of caveolin-1 to invaginations arose solely due to the actin-based motility of the bacteria, we do not think this is the principal mechanism. In fact, potential mechanisms underlying the requirement of caveolar components in L. monocytogenes spreading could involve global effects on the plasma membrane, such as lipid/protein composition and/or membrane tension. To begin to resolve these possibilities, we set out to demonstrate that the physical membrane protrusions alone can trigger caveolin-1 recruitment upon their contact with the host cell surface. To do this, we adapted a previously developed technique to isolate L. monocytogenes membrane protrusions from infected epithelial cells (6) and added these preparations onto naive epithelial cells that had previously been transfected with fluorescently tagged caveolin-1. To first confirm the presence of intact isolated membrane protrusions, we isolated the structures from cells expressing the plasma membrane marker pmKate2f-mem and saw that pmKate2-f-mem clearly delineated the structures (Fig. 6A). We also stained samples of isolated membrane protrusions for the host protein ezrin, a wellknown marker of L. monocytogenes membrane protrusions (53), and saw that endogenous ezrin levels were enriched within the actin-rich core of the structures, confirming that the isolated structures were in fact bacterially derived membrane protrusions (Fig. 6B). In both preparations, DAPI (4=,6-diamidino-2-phenylindole) staining as well as phase-contrast images indicated the presence of bacterial cells at one pole of the structures (Fig. 6A and B). When we overlaid isolated membrane protrusions onto naive cells, we saw that the membrane protrusions, some as large as ϳ8 m, had concentrated caveolin-1 around the entire structure (Fig. 6C). Caveolin-1 enrichment around the isolated membrane protrusion was further reflected in our line scan analysis of the structure (Fig. 6D). When we performed similar experiments examining other caveolar proteins, fluorescently tagged cavin-2 and EHD2, but not dynamin-2, was found concentrated around the overlaid isolated membrane protrusions ( Fig. 6E; see also Fig. S13 at https://figshare.com/s/b7f90aaaf70fbb1e0dfc).

DISCUSSION
Examinations of cell-to-cell spreading events during L. monocytogenes infections have historically concentrated on the formation of the protrusions, leaving the invaginations poorly studied. Here, we report that L. monocytogenes hijacks the caveolinendocytic machinery to move from one epithelial cell to another. Our data point to a subset of classical caveolar proteins (caveolin-1, cavin-2, and EHD2) as key components for the internalization of these large formations. We also document the presence of known caveolin-associated lipids at the structures.
The involvement of the clathrin-mediated endocytic protein epsin-1 at the membrane of these caveolin-based invaginations is peculiar, as other examples of clathrinassociated machinery such as clathrin itself, eps15, and AP2 are excluded from the sites. The reasons for the presence of epsin-1 may lie in its ability to curve the plasma membrane during endocytosis (35). Epsin-1 is classically known to induce the curvature of membrane through the direct insertion of its N-terminal alpha-helix, helix0, into the  cytosolic leaflet of the plasma membrane (35,57,58). Given that a spheroid shape would be generated at the region where the L. monocytogenes protrusion makes initial contact with the plasma membrane of the neighboring cell, which is a curved shape similar to that of an initially forming endocytic vesicle, epsin-1 could function at these sites as a membrane curving protein, inserting itself into the membrane to help promote the initial formation of the invagination. Although epsin-1 is normally used for clathrin-coated endocytic pit curve formation (35)(36)(37)(38), work in the McMahon laboratory clearly demonstrated that under conditions of coincubation with liposomes (35; see also references 39 and 57), epsin-1 (or even just the ENTH domain of epsin-1) can generate elongated tubules that morphologically resemble the tubular component of L. monocytogenes membrane invaginations. Thus, we postulate that epsin-1 could also interact with the highly curved membrane along the tubular aspect of the L. monocytogenes membrane invagination (oriented perpendicularly to the long axis of the structures). Further support for this possibility comes from the work of Capraro and colleagues, who demonstrated that the epsin-1 ENTH domain binds preferentially to highly curved membrane tethers (59). In the future, it will be interesting to see whether epsin-1 is also used for the caveolin-dependent internalization of other large particles.
How might caveolae and their associated proteins manage to physically distort the plasma membrane and engulf the entirety of L. monocytogenes protrusions? Clues to this may lie in evidence from Sinha and coworkers, who demonstrated that caveolae flatten with a concomitant integration of caveolin-1 into the plasma membrane as the cells experience mechanical stress (60). Our images show caveolin-1 delineating the invaginating membrane in a manner similar to the model proposed by Sinha and colleagues, and our observation that caveolae are present at invaginations supports this model. Additionally, while we have electron microscopic evidence documenting structures resembling caveolae at the tip of the membrane invagination (adjacent to the bacteria), it is likely that these structures also populate the stalk and distal regions of invaginations, as fluorescent images of endogenous caveolin-1 show a punctate staining pattern reminiscent of caveolae staining. Thus, given the presence of the caveola-like structures in electron micrographs together with our observation of linear and punctum-like caveolar protein localization along the entirety of invaginations, we propose a model whereby the L. monocytogenes protrusion initially contacts the recipient cell and uses the membrane from already present caveolae to begin to Consequently, the resulting elongated invagination contains caveolar proteins but no caveolae (Fig. 7). Our findings showing that invaginations are shorter in cells depleted of caveolin-1 gives credence to this possibility. Whether or not caveolae preferentially flatten at the earlier stages of the invagination process rather than once they have elongated will require future study.
Cargo uptake via caveolin-1-mediated endocytosis is normally triggered through ligand-receptor interactions (45,(61)(62)(63)(64)(65)(66)(67)(68)(69); see also reference 70 for a review). However, support for ligand-independent caveolin-1 function at plasma membrane caveolae also exists (71). Could ligand receptor-mediated signaling also play a role in the recruitment of caveolin-1 and other caveolar proteins to invaginations? From our experiments involving isolated L. monocytogenes membrane protrusions, the results implied that contact of the structures with the host cell surface alone is sufficient to trigger recruitment of caveolin-1, cavin-2, and EHD2. Coupled with our recent finding of the host plasma membrane receptor CD147 at L. monocytogenes membrane invaginations (23), the potential for the presence of a ligand(s) on the surface of protrusions, interacting with a corresponding receptor(s) on the cell forming the invaginations, presents a compelling area of further investigation. Furthermore, the visually apparent elevated levels of these caveolar proteins surrounding the structures also point to the possibility that preassembled caveolar vesicles could incorporate their proteins into the plasma membrane as an endocytic unit at sites of membrane invaginations (72,73) rather than being recruited separately. Unlike those aforementioned proteins, dynamin-2 remained absent at contact sites of the isolated membrane protrusions and the underlying cell. During cell-to-cell-spreading experiments, dynamin-2 showed an unusual localization at the invagination, being restricted to the region of the bacterium. This recruitment contrasts with its classical function at the neck of endocytic pits, where assemblies of dynamin-2 collars catalyze, via GTP, membrane scission of the budding vesicle (19,(74)(75)(76). One potential explanation may pertain to the L. monocytogenes internalin family of proteins which are involved in several stages of its infection cycle. Cell surface internalin A (InlA) and InlB control initial bacterial invasion into cells (77,78; see also reference 79 for a review), whereas secreted internalin C (InlC) and InlP have been shown to promote bacterial cell-to-cell and cell-to-basement membrane transfer, respectively (80,81). Consequently, future investigations into the L. monocytogenes infection cycle should focus on evaluating whether or not these bacterial components are also involved in the caveolin-mediated engulfment of membrane protrusions.
Despite the sparsity of studies examining the internalization of L. monocytogenes membrane protrusions, one study has provided insight into a process that is at play during the infections. The process of efferocytosis involves the removal of dead (or dying) material primarily by phagocytic cells. This process causes the weakening of the phagocytic cell plasma membrane prior to the removal of the dying material, and L. monocytogenes takes advantage of this process (82). Phosphatidylserine is commonly found within host cells; however, Czuczman and coworkers found that annexin V (a phosphatidylserine marker) was present on the surface of membrane protrusions generated by L. monocytogenes during HeLa cell infections, suggesting a certain degree of damage to the host cell plasma membrane (82). Through the use of TIM-4 (an efferocytosis component that recognizes phosphatidylserine) knockout (KO) mice and macrophages, they also found that this protein was involved in the dissemination of L. monocytogenes. We also examined phosphatidylserine but looked specifically for its presence at invaginations, as phosphatidylserine is also involved in organizing caveolar biogenesis (28). How the efferocytosis event at the protrusion may be linked to the formation of the corresponding invagination in epithelial cells remains to be determined.
Microbes often find strategies to exceed classical endocytic size limits. L. monocytogenes, Staphylococcus aureus, and Candida albicans use clathrin-mediated endocytosis, a process once thought to be limited to ϳ150-nm vesicles, for their initial entry into epithelial cells (14,83), whereas Shigella flexneri bacteria depend on clathrin and its associated proteins for cell-to-cell spreading (84). Caveolae, strictly defined, are too small (ϳ100 nm) (16)(17)(18) to engulf L. monocytogenes protrusions; however, our evidence suggests that caveolae have a direct local effect on the membrane that permits engulfment of large structures into the cell. The idea that caveolae can be locally permissive for endocytosis by potentially providing membrane to release into large invaginations, without themselves budding from the structures, opens the door for researchers to consider this process during other microbial infections and during the general internalization of particles whose size is beyond the current theoretical restriction. Our work also provides a way to reinterpret some of the literature on caveolar endocytosis and to resolve some of the many controversies in the field.

MATERIALS AND METHODS
Cell culture. Human cervical (HeLa) and Canis familiaris kidney (MDCK) epithelial cells were purchased from American Type Culture Collection (ATCC) (catalog no. CCL-2 and CCL-34, respectively). Stable HeLa cell lines transfected with control (nontargeting) or epsin-1-targeting sequences were generated from a wild-type HeLa cell line and cultured as described previously (85). All cells were cultured using Dulbecco's modified Eagle's medium (DMEM) containing high levels of glucose (HyClone; GE Healthcare) and supplemented with 10% fetal bovine serum (FBS) (Gibco, Thermo Fisher Scientific). All cell lines were maintained in a cell culture incubator at 37°C and 5% CO 2 . To seed cells for experiments, flasks containing cells were washed three times with Dulbecco's phosphate-buffered saline without Ca 2ϩ and Mg 2ϩ (PBS [Ϫ/Ϫ]) (Gibco, Thermo Fisher Scientific), trypsinized with 0.05% trypsin-EDTA (Gibco, Thermo Fisher Scientific), counted, and seeded onto clear polystyrene 6-well or 24-well plates (Corning) containing glass coverslips. For electron microscopy, flexible silicone elastomer membranes (Flexcell International) were used in place of glass coverslips.
Caveolin-1 shRNA. HeLa cells with stable knockdown of caveolin-1 (CAV1) as well as control (nontargeting) cells were generated via lentivirus-mediated transduction. Lentivirus particles containing pLKO.1-based plasmids that included gene-specific targeting sequences were produced. This was followed by puromycin selection (2 g/ml). The ShCav1 sequence used was TRCN0000008002 (targeting open reading frame).
Bacterial strains and growth conditions. L. monocytogenes strain EGD BUG 600 (gifted by Pascale Cossart) was used throughout this study and was grown at 37°C using either brain heart infusion (BHI) agar or BHI broth (BD Biosciences).
Listeria monocytogenes infections. To infect cells, broth cultures of L. monocytogenes (shaken overnight) were diluted 10-fold in fresh BHI broth (final volume of 10 ml) and then incubated at 37°C in a shaking incubator (on an angle) until A 600 ϭ 1.00. At A 600 ϭ 1.00, 1 ml of bacteria were centrifuged for 5 min at 10,000 rpm (25°C) and washed twice with prewarmed (37°C) PBS [-/-]. Pelleted bacteria were resuspended with 1 ml of prewarmed serum-free media (DMEM; 37°C) and then diluted 100ϫ. Diluted bacteria were added onto culture plates containing host cells and incubated for at least 6 to 8 h to study actin comet/rocket tail and membrane protrusion/invagination formation.
Mixed-cell infections. One batch of HeLa cells was seeded at a density of 2 ϫ 10 6 per well in 6-well-format plates without coverslips. On the same day, in separate 6-well plates (containing coverslips), a second batch of HeLa cells seeded at 2.25 ϫ 10 6 was added per well. The following day, the HeLa cells seeded previously at a density of 2.25 ϫ 10 6 were transfected (as described below) with DNA plasmids encoding the fluorescently tagged protein of interest to be examined at L. monocytogenes invaginations. On day 3, the untransfected HeLa cells (seeded previously at a density of 2 ϫ 10 6 per well) were infected (as described above) with wild-type L. monocytogenes at a multiplicity of infection (MOI) of ϳ40. At 2-h postinfection, the infected cells were washed three times with PBS [Ϫ/Ϫ] and then 1 ml of prewarmed DMEM containing 10% FBS and gentamicin (50 g/ml) was added to each well to kill any remaining extracellular bacteria. After 3 h of infection, the infected cells were detached and enumerated. Approximately 1 ϫ 10 6 infected cells were then overlaid on well plates containing the previously uninfected/transfected cells. Gentamicin was added to reach a final concentration of 50 g/ml. Samples were fixed 5 h following the overlaying procedure and stained with fluorescent phalloidin and DAPI as described below. The examination of fluorescently tagged proteins of interest at L. monocytogenes spreading events was performed by microscopic analysis whereby untransfected but infected cells (the cells appeared black to the eye of the microscopist) were sending L. monocytogenes membrane protrusions directly into and generating invaginations in the adjacent transfected cells (as visually determined by the microscopist through a combination of fluorescent and phase microscopy). Care was taken to ensure that each membrane protrusion could be visually traced back to the original sending location (untransfected cell).
Immunolocalization. Cells on glass coverslips were fixed at room temperature (rt) (in the dark) for 15 min using prewarmed (37°C) 3% paraformaldehyde (prepared in 150 mM NaCl, 4 mM Na/K PO 4 , 5.0 mM KCl, pH 7.3) and then washed three times using PBS [Ϫ/Ϫ]. Cells were permeabilized using room temperate 0.2% Triton X-100 (prepared in PBS [Ϫ/Ϫ]) for 5 min or Ϫ20°C acetone for 10 min. Following Triton permeabilization, coverslips were rinsed three times with PBS [Ϫ/Ϫ], whereas acetone treated coverslips were dried at room temperature for 30 min. All samples were blocked with 5% normal goat serum (in PBS [Ϫ/Ϫ]) for 25 min and then incubated overnight at 4°C with primary antibodies prepared in Tris PBS (TPBS)/BSA (PBS [Ϫ/Ϫ], 0.5% Tween 20, 0.1% bovine serum albumin [BSA]). The next day, the samples were washed three times with TPBS/BSA for 10 min and then treated with secondary antibodies (Alexa Fluor 594-or 488-conjugated goat anti-rabbit or goat anti-mouse) at room temperature in the dark for 2 h. To visualize F-actin, samples were treated with Alexa Fluor 594-or 488-conjugated phalloidin (prepared in PBS [Ϫ/Ϫ]) for 20 min. Samples were washed three times with PBS [Ϫ/Ϫ] and mounted onto glass microscope slides using Prolong Diamond antifade mounting medium (Invitrogen, Thermo Fisher Scientific) (with or without DAPI).
Lysate preparation and Western blotting. Cells were washed three times with prewarmed PBS [ϩ/ϩ] and then treated with 4°C radioimmunoprecipitation assay (RIPA) lysis buffer (150 mM NaCl, 50 mM Tris [pH 7.4], 5 mM EDTA, 1% Nonidet P-40, 1% deoxycholic acid, 10% SDS) containing cOmplete Mini EDTA-free protease inhibitor cocktail (Roche) on ice for 5 min. Cell scrapers were used to disrupt the cells, and lysates were then collected into microcentrifuge tubes. Lysates were spun at 4°C and 10,000 ϫ g for 10 min to pellet cellular debris and DNA; supernatants were then collected into fresh 4°C microcentrifuge tubes and immediately stored at Ϫ80°C. Protein concentrations were ascertained using a bicinchoninic acid (BCA) assay kit (Pierce). For Western blotting, lysates samples were prepared in 6ϫ Laemmli buffer and then boiled (100°C) for 10 min. Equal amounts of protein were loaded onto 10% SDS-polyacrylamide gels and resolved by electrophoresis. Gels were rinsed in distilled water for 5 min and then transferred onto nitrocellulose membranes using a Trans-Blot SD semidry transfer cell (Bio-Rad). Membranes were washed for 5 min in TBST (Tris-buffered saline, 0.05% Tween 20) with shaking, blocked with 4% Blotto (Santa Cruz Biotechnology) prepared in TBST (1 h shaking), and then treated with primary antibodies (diluted in TBST-1% BSA) overnight at 4°C. The next day, membranes were rinsed three times with TBST for 10 min prior to incubation with secondary antibodies (HRP-conjugated goat anti-rabbit or goat anti-mouse) for 1 h at room temperature. To visualize protein bands, membranes were treated with Western Lightning Plus-ECL (PerkinElmer) following the manufacturer's instructions and imaged using a Fujifilm LAS-4000 imager (Fujifilm). To confirm equivalent levels of loading, membranes were stripped using mild stripping buffer (1.5% glycine, 0.1% SDS, 1% Tween 20, pH 2.2) and reprobed using mouse anti-␣-tubulin targeting antibody.
Cell culture transfections. DNA transfections were performed using jetPEI or jetPRIME transfection reagents (Polyplus Transfection) and carried out according to the manufacturer's instructions. Briefly, cells were transfected (3 l reagent and 1.5 g plasmid DNA per well [6-well plate]) and then incubated at 37°C for 4 h. After 4 h, the medium in the wells was replaced and the cells were incubated for at least 24 h at 37°C to allow expression of the respective gene product.
Gentamicin protection assay. Cultured cells in 24-well cell-binding plates were infected with wild-type L. monocytogenes at an MOI of ϳ15 to 20 for 2 h to allow bacterial invasion. After 2 h of infection, the well plates were rinsed three times with Dulbecco's phosphate-buffered saline with Ca2 ϩ and Mg2 ϩ (PBS [ϩ/ϩ]) (Gibco, Thermo Fisher Scientific). The infections were then allowed to proceed for an additional 1 h in media containing 50 g/ml gentamicin (to kill extracellular bacteria). Bacterial loads were also examined after 8 h following the intercellular spreading assay infection protocol. At the end of the infections, cells were rinsed five times with PBS [ϩ/ϩ] and intracellular bacteria were released by treatment of cells with a 1% solution of Triton X-100 (made in PBS [ϩ/ϩ]) for 5 min. After the 5-min incubation, serial dilutions of the wells were performed in 96-well-format assay blocks. Dilutions were selected (so as to give a count of between 30 and 300 bacterial colonies), and cells were spread onto BHI agar plates and incubated for 24 h at 37°C prior to their enumeration.