key: cord-0293841-sebvmx5g
authors: Radhakrishnan, Prathima; Sathe, Mugdha; Theriot, Julie A.
title: Listeria monocytogenes Co-opts Caveolin-Mediated E-cadherin Trafficking and Macropinocytosis for Epithelial Cell-to-Cell Spread
date: 2022-04-06
journal: bioRxiv
DOI: 10.1101/2022.04.06.487361
sha: f12cb681a76da1416d3a710e02f5c60eeca64ef8
doc_id: 293841
cord_uid: sebvmx5g

Listeria monocytogenes is an intracellular bacterial pathogen that spreads directly between adjacent host cells without exposure to the extracellular space. Recent studies have identified several host cell factors that promote L. monocytogenes cell-to-cell spread in epithelial monolayers, but details of the mechanism remain unclear. We find that the adherens junction protein, E-cadherin, promotes L. monocytogenes cell-to-cell spread at the recipient side of cell contacts. In particular, two point mutations in E-cadherin’s cytoplasmic domain that prevent its ubiquitination hinder bacterial cell-to-cell spread efficiency without reducing the extent of contact between neighboring cells. As ubiquitination induces E-cadherin endocytosis, we hypothesize that E-cadherin promotes protrusion engulfment, where donor cell protrusions containing L. monocytogenes are taken up by the recipient cell concurrently with E-cadherin internalization. In support of this hypothesis, we show that inhibiting caveolin-mediated membrane trafficking reduces L. monocytogenes cell-to-cell spread only under conditions where E-cadherin can be ubiquitinated. Additionally, we demonstrate that macropinocytosis also contributes to dissemination of L. monocytogenes through an epithelial monolayer.

Listeria monocytogenes is a facultative bacterial pathogen that causes spontaneous abortions and illness in immunocompromised individuals, ranking third among foodborne pathogens in number of deaths caused per year (Hamon et al., 2006) . When ingested orally, L. monocytogenes first infects epithelial cells of the human gut (Schlech and Acheson, 2000) . It adheres to these cells by binding receptors such as E-cadherin on the host cell surface (Mengaud et al., 1996 , Lecuit et al., 1999 , Ortega et al., 2017 and is then actively taken up by the host cells. Upon entry, the bacterium secretes phospholipases and the pore-forming toxin Listeriolysin O (LLO) to rupture its vacuole and escape into the host cell's cytoplasm (Portnoy et al., 1988 , Marquis et al., 1995 . It propels itself within the cytoplasm using actin-based motility (Tilney and Portnoy, 1989 , Dabiri et al., 1990 , Theriot et al., 1992 , eventually reaching the plasma membrane and forming a membrane-covered protrusion that can push into the neighboring cell.

Intercellular spread occurs when the recipient cell engulfs the L. monocytogenes-containing protrusion and the bacterium escapes the secondary vacuole into the cytoplasm of the recipient cell (Robbins et al., 1999) . This process allows the pathogen to propagate through monolayers without exposure to the extracellular space, thereby avoiding the host's humoral immune response (Zenewicz and Shen, 2007) .

Several molecular factors have been identified in both the pathogen and the host cell that promote the formation or elongation of protrusions at the donor side of cell contacts and enhance the efficiency of cell-to-cell spread. For example, the secreted bacterial virulence factor internalin C (InlC) supports protrusion formation and elongation by relieving cortical tension at the donor side of cell contacts (Rajabian et al., 2009) and recruiting exocyst machinery to deliver additional membrane to protrusions (Dowd et al., 2020) . By secreting cyclic-di-AMP, L. monocytogenes enhances nitric oxide production in host cells, which increases the speed of the pathogen's actin-based motility, to allow for the generation of protrusions that are more frequently taken up by a recipient cell (McFarland et al., 2018) .

Two recent studies have identified specific host cell factors that participate in cell-to-cell spread at the recipient side rather than the donor side of cell contacts (Sanderlin et al., 2019 , Dhanda et al., 2020 . Using an RNAi screen to identify host cell factors that contribute to L. monocytogenes cell-to-cell spread, Sanderlin et al. identified caveolin-1, a mediator of caveolar endocytosis, as contributor to bacterial spread in recipient cells (Sanderlin et al., 2019) . Using a more targeted candidate-based approach, Dhanda et al. also described a role for caveolar elements including caveolin-1, cavin-1, and EHD2 in promoting L. monocytogenes cell-to-cell spread, along with other endocytosis-related proteins including dynamin (Dhanda et al., 2020) .

In addition to caveolins, the RNAi screen conducted by Sanderlin et al. also found a role for the epithelial cell adherens junction protein, E-cadherin, in promoting cell-to-cell spread (Sanderlin et al., 2019) . E-cadherin is robustly detected all along protrusions containing L.

monocytogenes (Robbins et al., 1999) , and cadherin expression is required for the epithelial cellto-cell spread of Shigella flexneri, another intracellular pathogen that uses actin-based motility for spread in epithelial monolayers (Sansonetti et al., 1994) . Interestingly, adhesion proteins including VE-cadherin and nectins have been shown to promote trans-endocytosis, the process by which cytoplasmic material and transmembrane proteins are exchanged between neighboring cells through a receptor-mediated process (Cagan et al., 1992 , Sakurai et al., 2014 , Generous et al., 2019 . For these reasons, we began this study of L. monocytogenes cell-to-cell spread by investigating the mechanism by which the adhesion protein, E-cadherin, promotes bacterial intercellular spread. Our results are consistent with a simple model where caveolin-mediated trafficking of E-cadherin at the recipient side of cell-cell contacts promotes internalization of L. monocytogenes-containing protrusions, and thereby contributes to spread.

To quantify the extent to which E-cadherin influences L. monocytogenes cell-to-cell spread, we compared efficiency of cell-to-cell spread in epithelial cell monolayers grown in tissue culture. To this end, we used A431D epithelial cells, a cell line derived from a human primary epidermoid carcinoma that have lost expression of E-cadherin but retain expression of accessory proteins associated with adherens junctions (Lewis et al., 1997) . Because of this property, it is convenient to use A431D cells as the background for expression of E-cadherin constructs with specific molecular lesions, in order to evaluate the roles of the protein's various functional domains in complex biological processes (McEwen et al., 2014) . For our first measurement, we compared the efficiency of L. monocytogenes spread between A431D epithelial cells expressing full length (wild-type) E-cadherin (WT E-cad) and parental A431D cells expressing no E-cadherin (Null E-cad). As E-cadherin is required for the initial invasion step through which the bacterium enters the monolayer from the extracellular space (Mengaud et al., 1996 , Lecuit et al., 1999 , we compared focus size at 12 hours post-infection in a monolayer of WT E-cad cells with 1:100 WT E-cad: Null E-cad A431D cells ( Fig. 1A -B, Suppl. Movie 1), expecting that focus size at this late time point would be dominated by spread between the numerous Null E-cad cells rather than the single WT E-cad cell required for initial invasion.

Because L. monocytogenes foci in epithelial monolayers are highly irregular in shape (Ortega et al., 2019) , simple methods for determining the size of the focus using the smallest convex hull typically overestimate the true extent of cell-to-cell spread (Suppl. Fig. S1A ). As an alternative, we optimized a method for calculating a tightly wrapped "alpha-shape" (Edelsbrunner et al., 1983 ) that more accurately correlates with the total number of infected host cells in a focus (Suppl. Fig. S1B ). Using this metric ( Fig. 1C-D) , we found that the focus area was diminished by about 50% when L. monocytogenes spread between Null E-cad cells as compared to WT E-cad cells (Fig. 1E) , suggesting that E-cadherin facilitates but is not strictly required for L. monocytogenes cell-to-cell spread.

E-cadherin might promote spread by aiding in the formation of protrusions at the donor cell, uptake of protrusions at the recipient cell, or both. To distinguish between these possibilities, we infected WT E-cad cells or 1:100 WT E-cad: Null E-cad A431D cells for a short time period of five hours, such that only one round of cell-to-cell spread would have occurred.

Then, for each focus, we quantified the ratio between the total number of bacteria in neighboring recipient cells to the number of bacteria in the originally infected donor cell, thereby comparing the efficiency of transfer from WT E-cad to WT E-cad cells with the efficiency of transfer from WT E-cad to Null E-cad cells (Fig. 1F ). This ratio was significantly higher for spread into WT Ecad recipient cells (Fig. 1G ), although the total number of bacteria in each focus was comparable for both conditions (Suppl. Fig. S1D ). As the lack of E-cadherin in the recipient cell specifically reduced the efficiency of cell-to-cell spread, we conclude that E-cadherin is involved in the uptake of protrusions at the recipient side of cell contacts, although we cannot rule out the possibility that it might also participate in the formation of protrusions by the donor cell.

To further explore the mechanism by which E-cadherin promotes spread, we compared L. monocytogenes cell-to-cell spread through a monolayer of WT E-cad cells with spread through a monolayer of A431D cells expressing an E-cadherin truncation mutant that lacked its entire cytoplasmic domain, deleting amino acids 731-822 (Δcyto E-cad) (Ringwald et al., 1987) .

Although this truncated protein was enriched at cell-cell junctions, it was unable to recruit cytoplasmic binding partners such as β -catenin ( Fig. 2A) . Focus area was reduced in the Δ cyto E-cad cells as compared to the WT E-cad cells, nearly to the extent that focus area was reduced in the Null E-cad cells (Fig. 2B) . Using a strain of L. monocytogenes deficient for cell-to-cell spread, we confirmed that the initial replication rate over the first 3 to 5 hours post-infection for bacteria in WT E-cad cells and in Δ cyto E-cad cells was indistinguishable (Suppl. Fig. S2A ), consistent with the idea that these focus size defects are caused by a specific deficiency in cellto-cell spread rather than effects on earlier events in bacterial invasion or replication. Our results suggest that it is not E-cadherin's extracellular domain, which physically links neighboring cells (Shapiro et al., 1995) , that is primarily involved in augmenting L. monocytogenes cell-to-cell spread. Rather, E-cadherin's cytoplasmic domain, which reinforces cell-cell junctions due to its linkage to the underlying actin cytoskeleton (Buckley et al., 2014) and enables E-cadherin internalization and turnover (Delva and Kowalczyk, 2009) , contributes to L. monocytogenes cellto-cell spread.

In order to further narrow down the part of E-cadherin's cytoplasmic domain that contributes to bacterial spread, we used a line of A431D cells expressing an E-cadherin truncation mutant (missing amino acids 810-882) that cannot bind β -catenin (Δβ E-cad) and therefore does not associate with the actin cytoskeleton, although it retains other protein-protein interaction sites in the juxtamembrane domain (JMD) (Ferber et al., 2002 , Fujita et al., 2002 ( Fig. 2A) . Efficiency of bacterial cell-to-cell spread in this host cell line was intermediate between that in the WT E-cad cell line and the Δ cyto E-cad cell line (Fig. 2B) , suggesting that the lack of association between Δ cyto E-cad and the actin cytoskeleton may contribute to the defect in bacterial cell-to-cell spread, but cannot completely explain the magnitude of this effect.

As β -catenin links E-cadherin to cortical actin (Yamada et al., 2005) , cell junctions between neighboring Δ β E-cad or Δ cyto E-cad cells are likely to be under less tension than cell junctions between neighboring WT E-cad cells (Verma et al., 2012) . Moreover, Δ β E-cad or Δ cyto E-cad cells would be unable to extend points of adhesion between two cells into larger zones of contact as this junctional reinforcement is normally a consequence of actin nucleation at cell junctions (Verma et al., 2004) . The extent of cell-cell contact in an epithelial monolayer is expected to depend on the physical packing density of cells within the monolayer as well as on the ability of the cells to reinforce junctional integrity by attachment to the actin cytoskeleton.

We therefore carefully controlled cell plating density to ensure that the physical packing of cells in the monolayer was consistent across all experiments (Suppl. Fig. S2B ). In order to examine the extent of cell-cell contact more directly, we used structured illumination microscopy (SIM) to measure the height of junctions in the monolayers as compared to the maximum height of the cells (Suppl. Fig. S2C ). Using this assay, we found that the extent of cell-cell contact between neighboring cells in a monolayer of A431D cells expressing WT E-cad was substantially greater than the extent of contact for Null E-cad cells plated at the same density, as expected ( Fig. 2 

The E-cadherin JMD includes binding sites for several interaction partners that might contribute to L. monocytogenes cell-to-cell spread, including p120 catenin (Ferber et al., 2002) and the ubiquitin ligase hakai (Fujita et al., 2002) . We constructed a modified full-length Ecadherin including two point mutations, K738R and K816R, that block ubiquitination and normal turnover of the E-cadherin JMD (Hartsock and Nelson, 2012) , and expressed this construct in A431D cells. In contrast to Null E-cad cells, Using the quantitative assay for L. monocytogenes spread described above, we found a 20% reduction of L. monocytogenes focus size in the K738R, K816R E-cad background as compared to the WT E-cad background (Fig. 2B ). This observation is consistent with the hypothesis that the JMD of E-cadherin contributes to bacterial cell-to-cell spread in a way that is dependent on E-cadherin ubiquitination by hakai. Because ubiquitination at K738 and K816 is known to induce E-cadherin endocytosis (Fujita et al., 2002) , we further hypothesized that Ecadherin could contribute to L. monocytogenes cell-to-cell spread by promoting protrusion uptake, a role that is consistent with our demonstration that E-cadherin contributes to bacterial spread at the recipient side of cell-cell contacts. In this scenario, L. monocytogenes can be transferred from donor to recipient cell concurrently with E-cadherin endocytosis.

The best-characterized mechanism through which E-cadherin is internalized is clathrinmediated endocytosis (CME), where adaptor proteins such as Numb and AP-2 bind to Ecadherin to bring about its clathrin-dependent internalization (Le et al., 1999) . This pathway seemed a promising candidate, as binding of the protein p120-catenin at the cadherin JMD is known to inhibit endocytosis by obstructing endocytic motifs that permit CME (Miyashita and Ozawa, 2007) and preventing cadherin clustering at clathrin-coated pits (Chiasson et al., 2009 ).

However, clathrin heavy chain and AP-2 were not identified as genes affecting L.

monocytogenes cell-to-cell spread in an RNAi screen for genes involved in spread (Sanderlin et al., 2019) and clathrin was absent from invaginations made by donor cell protrusions containing L. monocytogenes (Dhanda et al., 2020) . In accordance with these prior results, we did not observe a reduction of focus size when L. monocytogenes propagated through A431D cells treated with pitstop 2, an inhibitor of CME (Dutta et al., 2012) . Instead, we observed a small, 10% increase in focus size in the WT E-cad cells and no change in the Δ cyto E-cad, Δ β E-cad, or K738R, K816R E-cad cells (Fig. 3A ). To confirm that pitstop 2 was inhibiting CME in A431D cells, we quantified uptake of transferrin, a standard target of CME (Harding et al., 1983) and

found that it was reduced upon addition of pitstop 2 (Suppl. Fig. S3A ). Next, we used an antibody-based surface accessibility assay to measure what proportion of total E-cadherin is available for antibody binding at the cell surface in the presence and absence of pitstop 2. We observed a modest but significant increase in surface E-cadherin in WT E-cad A431D cells upon treatment with pitstop 2, as would be expected if the drug causes a defect in E-cadherin internalization (Suppl. Fig. S3B ). These results confirm that CME contributes to E-cadherin internalization in A431D cells, and that pitstop 2 effectively disrupts CME under our assay conditions, but suggest that CME is not a major pathway contributing to L. monocytogenes cellto-cell spread.

An alternative pathway that has been shown to contribute to E-cadherin internalization and turnover in epithelial cells is caveolin-mediated uptake (Lu et al., 2003 , Orlichenko et al., 2009 , Akhtar and Hotchin, 2001 . To assess whether L. monocytogenes might use caveolinmediated E-cadherin internalization to spread from cell to cell, we quantified spread efficiency in A431D cells treated with filipin III, a cholesterol sequestering reagent (Behnke et al., 1984) As filipin III alters membrane fluidity (Zhang et al., 2019) , it can also disrupt endocytic pathways that do not involve caveolin (Hao et al., 2004) and so we sought to confirm our results using the more specific approach of treating A431D cells with the caveolin-scaffolding domain peptide (CSD peptide). This peptide competes with endogenous caveolin-1 to disrupt its interaction with binding partners (Bucci et al., 2000) . As the N-terminal cytoplasmic domain of caveolin-1 immunoprecipitates with E-cadherin (Galbiati et al., 2000) , the two proteins are likely to associate in vivo. In WT E-cad A431D cells treated with vehicle control, caveolin-1 colocalized with E-cadherin at cell boundaries ( Fig. 3C ). Upon addition of CSD peptide, Ecadherin remained at the cell junctions but caveolin-1 was mislocalized (Fig. 3C) , confirming that the peptide interfered with the association between the two proteins. Using the surface accessibility assay, we also demonstrated that CSD peptide reduced E-cadherin internalization in the WT E-cad and Δ β

Δ cyto E-cadherin or K738R, K816R E-cad (Fig. 3D ). In addition to suggesting that an interaction between caveolin-1's scaffolding domain and E-cadherin's JMD is required for caveolin-mediated E-cadherin internalization, this result indicates that internalization of E-cadherin after ubiquitination by hakai occurs through a caveolin-mediated process.

Using our quantitative assay to measure the efficiency of L. monocytogenes cell-to-cell spread, we found that focus size was diminished in WT E-cad and Δ β E-cad cells that were treated with CSD peptide, but remained unaltered in CSD-treated Δ cyto E-cad and K738R, K816R E-cad cells (Fig. 3E ). This result suggests that caveolin-mediated E-cadherin internalization contributes to L. monocytogenes cell-to-cell spread, in a manner that is dependent on ubiquitination of the E-cadherin JMD by hakai.

A third mechanism through which E-cadherin may be internalized is macropinocytosis (Bryant et al., 2007) . When we treated A431D cells with 5-[N-ethyl-N-isopropyl] amiloride (EIPA) and Rac1 Inhibitor (NSC23766), inhibitors of macropinocytosis (Koivusalo et al., 2010 , Bryant et al., 2007 , we observed a reduction of focus size, regardless of E-cadherin background ( Fig. 3F, 3G ). As an independent approach to measure the role of macropinocytosis in L.

monocytogenes intercellular spread, we used siRNA knockdown of CtBP1 (Suppl. Fig. S3D ), a protein that is required for macropinosome scission (Liberali et al., 2008) . Once again, we observed a decrease in efficiency of L. monocytogenes cell-to-cell spread after CtBP1 knockdown in A431D cells expressing any of our four E-cadherin constructs (Fig. 3H ). These results led us to conclude that macropinocytosis is an additional mechanism through which L.

monocytogenes spreads from cell to cell, but this mechanism is not dependent on any specific protein interactions with the cytoplasmic domain of E-cadherin.

To confirm that these drug treatments were blocking macropinocytosis under our experimental conditions, we monitored nonspecific macropinocytic uptake of fluorescent dextran from the cell culture medium, and found it to be substantially reduced upon treatment of A431D cells with EIPA and NSC23766 (Suppl. Fig. S3E ) as well as after siRNA knockdown of CtBP1 (Suppl. Fig. S3F ). As an additional control, we used the surface accessibility assay to demonstrate that EIPA and NSC23766 increased the proportion of E-cadherin remaining on the cell surface (Suppl. Fig. S3G , S3H), as expected if macropinocytosis normally contributes to Ecadherin internalization and turnover in these cells. However, none of these treatments affected the average number of L. monocytogenes present in A431D cells in the first 3 to 5 hours postinfection, as determined using a mutant strain of bacteria incapable of actin-based cell-to-cell spread (Suppl. Fig. S3I -M). These results support the conclusion that macropinocytosis contributes to L. monocytogenes cell-to-cell spread, but is not involved in earlier stages of bacterial invasion or replication.

So far we have shown that inhibiting ubiquitin-dependent caveolin-mediated internalization of E-cadherin and inhibiting macropinocytosis both independently decrease the efficiency of L. monocytogenes cell-to-cell spread. We were therefore curious whether stimulation (rather than inhibition) of these pathways might actually enhance the efficiency of bacterial spread. As the growth factor EGF can stimulate both macropinocytosis and caveolinmediated internalization of E-cadherin (Lu et al., 2003 , Bryant et al., 2007 , we sought to measure bacterial spread efficiency upon treatment of A431D cells with EGF.

EGF stimulation of macropinocytosis in A431 cells is usually observed in the absence of serum (Hamasaki et al., 2004) but as serum is present in our standard infection assay, it was first necessary for us to measure the effect of EGF on macropinocytosis when we replicated the conditions of our infection assay. Under these conditions, we found no increase in internalized dextran when EGF was added to A431D cells regardless of E-cadherin background (Fig. 4A 4B ). This EGF-dependent increase in focus size is abrogated by treatment with the EGFR inhibitor gefitinib (Lynch et al., 2004) , but gefitinib has no effect on focus size in the absence of exogenous EGF (Suppl. Fig. S4B ). Finally, we confirmed that EGF treatment does not alter the number of bacteria in foci at early stages of infection (Suppl. Fig. S4C ). Given the evidence described above that a major effect of EGF stimulation in these cells is to promote caveolin-mediated uptake of E-cadherin, these results further confirm our central conclusion that this particular pathway for E-cadherin trafficking contributes to the efficiency of L.

monocytogenes cell-to-cell spread.

Our results support the hypothesis that caveolin-mediated uptake of E-cadherin, but not its clathrin-mediated uptake, promotes the engulfment of L. monocytogenes-containing protrusions on the recipient side of host cell-cell contacts, thereby enhancing the efficiency of bacterial cell-to-cell spread in epithelial monolayers. Significantly, we made use of a subtle mutational alteration in the juxtamembrane domain of E-cadherin, converting two lysine residues to arginine residues in order to block ubiquitination of the E-cadherin cytoplasmic domain by the ubiquitin ligase hakai (Hartsock & Nelson, 2012) , without disrupting the ability of E-cadherin to mediate formation of normal cell-cell junctions or to interact with the actin cytoskeleton via binding to β -catenin. Using this mutant, we found that disruption of cell-to-cell spread by inhibiting caveolin-mediated trafficking can affect the efficiency of bacterial spread only when E-cadherin itself can be ubiquitinated. Prior work has provided evidence for the involvement of E-cadherin in L. monocytogenes cell-to-cell spread (Sanderlin et al., 2019) and suggested a role for caveolins and other caveolar components in this process (Sanderlin et al., 2019; Dhandha et al 2020) . We confirm and extend this work by suggesting that it is most likely a direct interaction between caveolin and E-cadherin that facilitates this function.

In addition, our results suggest that macropinocytosis is an additional pathway that contributes to the efficiency of L. monocytogenes cell-to-cell spread, although this mechanism is not dependent on any particular domain of E-cadherin. While macropinocytosis is known to be a mechanism through which many pathogens enter host cells (Bloomfield and Kay, 2016) and has also been implicated in the intercellular spread of prions (Zeineddine and Yerbury, 2015) and viruses such as HIV-1 (Wang et al., 2008) and coronaviruses (Freeman et al., 2014) , to our knowledge, L. monocytogenes is the only bacterial pathogen for which macropinocytosis has been shown to contribute to intercellular spread (Fukumatsu et al., 2012) .

Finally, it is illuminating to compare L. monocytogenes cell-to-cell spread with the mechanism used by Shigella flexneri, another intracellular bacterial pathogen that employs actinbased motility to disseminate through an epithelial monolayer (Bernardini et al., 1989) . Spread of both pathogens occurs at cell-cell junctions (Robbins et al., 1999 , Rajabian et al., 2009 , Duncan-Lowey et al., 2020 and loss of E-cadherin reduces efficiency of spread (Fukumatsu et al., 2012 , Sanderlin et al., 2019 . However, S. flexneri targets tricellular junctions, while L. monocytogenes does not exhibit a preference for tricellular over bicellular junctions. Moreover, knockdown of tricellulin, which is enriched at tricellular junctions, greatly diminishes S. flexneri cell-to-cell spread but has no effect on L. monocytogenes spread (Fukumatsu et al., 2012) .

Consistent with our results, Fukumatsu et al. also showed that inhibiting macropinocytosis with EIPA reduced efficiency of L. monocytogenes cell-to-cell spread, while suppressing clathrin-mediated uptake with phenylarsine oxide did not significantly impair spread in Caco-2 epithelial cells. Additionally, they showed that inhibiting caveolar uptake with the cholesterol depleting agent, methyl-β-cyclodextrin, limited L. monocytogenes cell-to-cell spread, consistent with our results using filipin III to deplete cholesterol and using the CSD peptide to interfere with caveolin protein-protein interactions more directly. Significantly, Fukumatsu et al. showed that the impact of these drugs on S. flexneri cell-to-cell spread was distinct from that for L. monocytogenes; specifically, macropinocytosis did not appear to contribute to spread at all, while inhibiting caveolin-mediated endocytosis had a modest effect and clathrin-mediated endocytosis was found to be the major pathway S. flexneri exploits to spread from cell-to-cell (Fukumatsu et al., 2012) . These results suggest that the bacteria are not merely carried into neighboring cells during trans-endocytosis events that are occurring constitutively at some basal level even in uninfected cells (Robbins et al., 1999; Generous et al., 2019 , Sakurai et al., 2014 .

Instead, there appears to be some specificity to the choice of trafficking pathway utilized by each pathogen.

We thank Cara Gottardi for the generous gift of the human full-length E-cadherin cDNA as well as the parental A431D cell line. We additionally thank Fabian Ortega for generating the WT, 

The authors declare no competing interests.

The 

As described previously (Ortega et al., 2017) , various E-cadherin constructs were derived using the full-length human E-cadherin sequence from the pcDNA3 human E-cadherin plasmid (a gift from Cara Gottardi's lab at Northwestern University). 

To conduct infection assays, A431D cells were seeded on cleaned 18 mm glass coverslips that were coated with 1μg/mL fibronectin at a density of 4×10 5 cells/mL and were allowed to Data shown in Figure 1G originates from two independent experiments.

In the case of spread between WT E-cad and Null E-cad cells, the WT E-cad cell is assumed to be the donor, as L. monocytogenes adheres to E-cadherin on the cell surface prior to invading epithelial cells (Ortega et al., 2017) . Time lapse experiments revealed that the cell at the center of the focus was most likely to be the donor in the case of spread between WT E-cad cells (Data not shown).

A431D cells were seeded on clean fibronectin-coated 25mm coverslips at a density of 7.5×10 5 cells/mL and incubated at 37 o C for 36 hours. They were subsequently fixed and Cell membranes of A431D cells were labeled using 5μg/mL rhodamine labeled wheat germ agglutinin (Vector Labs; RL-1022) in HBSS (Hank's balanced salt solution) for 10 minutes at room temperature. After washing with PBS and permeabilizing with 0.1% Triton X-100 as described above, cells were stained for E-cadherin using the protocol previously outlined and were also labeled with DAPI to illuminate the cell's nucleus as detailed in the infection assay section. Then, coverslips were mounted onto slides and superresolution images were collected using a Nikon Apo TIRF 100X 1.49 NA objective for STORM/SIM using a VisiTech iSIM coupled with an inverted Nikon Eclipse TI2 and a Hamamatsu ORCA/Fusion sCMOS Camera.

We acquired confocal images of mTagRFP expressing L. monocytogenes as well as DAPI-stained host cell nuclei with approximately 10 slices per infection focus and a z-spacing of 1μm. These images were taken using a Yokogawa W1 Spinning Disk Confocal with Borealis

Upgrade on a Nikon Eclipse Ti2 inverted microscope with a 50μm disk pattern (Andor).

Furthermore, a Plan Apo 20X 0.95 NA water immersion objective was used to acquire the images, a piezo z-stage (Ludl 96A600) was used to increase the rate of image acquisition between z-frames, and MicroManager v.1.4.23 was used to control all microscopy equipment.

Upto fifty bacterial foci were imaged on each coverslip with two replicate coverslips per condition and each experiment repeated on three separate days.

After performing a maximum intensity projection of the confocal images and cropping out the bacterial focus to eliminate extraneous bacteria in Fiji, Matlab_R2018B was used to threshold and binarize the image of the focus. We then generated an alpha shape around each binarized focus using the Matlab function alphaShape with an alpha radius of 30, suppressed all holes within the alpha shape and calculated its area. This was a deviation from the traditional use of the convex hull area as a measure of spread efficiency as the convex hull tends to highly weight outlying bacteria that occupy extreme positions within the focus (Ortega et al., 2019) . for a condition were similarly transferred onto ice but fixed first using the same method described above and then permeabilized with 0.1% Triton-X 100 at room temperature for 10 minutes. The protocol for staining cells was identical to the non-permeabilized sample except that all steps were conducted at room temperature. Epifluorescence images of 10 fields of view (FOV) per dish were obtained using an inverted Nikon Eclipse TI2 with an EMCCD Camera (Andor Technologies) and a 20X 0.75 NA Plan Apo Air Objective and Micromanager. The average intensity of each FOV in the non-permeabilized sample was divided by the average intensity of a FOV in the corresponding permeabilized sample to obtain a ratio of surface Ecadherin to total E-cadherin.

To conduct the dextran uptake assay, A431D cells were seeded on a 35 mm glass-bottom dish with a 20 mm well (Cellvis; D35-20-1.5N) that was coated with 1μg/mL fibronectin at a density of 6×10 

To conduct the transferrin internalization assay, A431D cells were seeded on a 35 mm glass-bottom dish with a 20 mm well (Cellvis; D35-20-1.5N) that was coated with 1μg/mL fibronectin at a density of 6×10 5 cells/mL and allowed to incubate at 37 o C for 36 hours. A431D cells were then pulsed with this solution for 10 minutes in a 37 o C water bath and then quickly transferred onto ice to inhibit any further endocytosis. After washing with ice-cold PBS, the cells were incubated in ascorbate buffer (160mM sodium ascorbate, 40mM ascorbic acid, 1mM MgCl 2 , 1mM CaCl 2 ) twice for five minutes each to strip transferrin from the surface of the cell. They were then incubated in ice-cold PBS for five minutes to return the sample to a neutral pH. Next, cells were fixed on ice with 4% methanol-free PFA for 30 minutes before transferrin receptor was labeled with CD71 Transferrin Receptor Antibody (Thermofisher; 14-0719-82) as a primary antibody and Alexa Fluor 488 Goat Anti-Rabbit IgG (H+L) antibody as the secondary.

A431D cells were incubated with both antibodies for 30 minutes each and three PBS washes were conducted post-fixation, after addition of the primary and secondary antibodies. Two independent experiments are reported in Figure S4A . Images were acquired and the ratio of internalized labeled transferrin to surface transferrin receptor was obtained by dividing the average fluorescence intensity of the former from the latter. comparison of cell-to-cell spread efficiency were calculated using the linear mixed-effects model (Gałecki and Burzykowski, 2013) , which takes into account the variability introduced by random effects such as the day of experimentation and coverslip on which cells were seeded. As at least three independent experiments were conducted to evaluate whether a perturbation affected the focus area, the threshold for significance was set at 0.01 to account for multiple hypothesis testing.

The results of the surface accessibility assays ( Figures 3D, S3B , S3E, S3F, S4A), transferrin internalization assay ( Figure S3A ) and macropinosome quantitation assay ( Figures   S3D, 4A) were represented as bar graphs with the midlines signifying the mean of the data points and the whiskers denoting the standard deviation. Ratio of junction length to cell height ( Figure   2C ) and bacterial count was represented in the same way in Suppl. Figs. S1D, S2A, S3G, S3H, S3I, S3J, S3K, and S4C. The ratio of bacteria in recipient cells to the donor cell in Figure 1G was represented as a beeswarm plot with midlines signifying the mean of the data points and the whiskers denoting the standard deviation. P-values were calculated using the Wilcoxon rank-sum test in GraphPad PRISM8 with a threshold of significance set at 0.05.

Requests for reagents and detailed protocols may be directed to Julie A. Theriot (jtheriot@uw.edu).

Materials developed in this study are available on request to the corresponding author.

Data collected and computer codes are available on request to the corresponding author. imaged for each condition from left to right is 437, 496, 183, 188, 194, 204, 224 and 190 standard deviation, dashed line shows the median, and dotted lines the first and third quartiles. P values were determined using the linear mixed-effects model.

RAC1 Regulates Adherens Junctions through Endocytosis of E-Cadherin

Filipin as a cholesterol probe

Filipin-cholesterol interaction in red blood cell membranes

Identification of icsA, a plasmid locus of Shigella flexneri that governs bacterial intra-and intercellular spread through interaction with F-actin

Uses and abuses of macropinocytosis

EGF induces macropinocytosis and SNX1-modulated recycling of E-cadherin

In vivo delivery of the caveolin-1 scaffolding domain inhibits nitric oxide synthesis and reduces inflammation

The minimal cadherin-catenin complex binds to actin filaments under force

The bride of sevenless and sevenless interaction: internalization of a transmembrane ligand

p120-catenin inhibits VE-cadherin internalization through a Rho-independent mechanism

Listeria monocytogenes moves rapidly through the host-cell cytoplasm by inducing directional actin assembly

Regulation of cadherin trafficking

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

Listeria monocytogenes exploits host exocytosis to promote cell-to-cell spread

Shigella flexneri Disruption of Cellular Tension Promotes Intercellular Spread

Pitstop 2 is a potent inhibitor of clathrin-independent endocytosis

On the shape of a set of points in the plane

An octapeptide in the juxtamembrane domain of VE-cadherin is important for p120ctn binding and cell proliferation

Hakai, a c-Cbl-like protein, ubiquitinates and induces endocytosis of the E-cadherin complex

Shigella targets epithelial tricellular junctions and uses a noncanonical clathrin-dependent endocytic pathway to spread between cells

Caveolin-1 Expression Inhibits Wnt/β-Catenin/Lef-1

Signaling by Recruiting β -Catenin to Caveolae Membrane Domains*

Linear Mixed-Effects Model

Linear Mixed-Effects Models Using R: A Step-by-Step Approach

Trans-endocytosis elicited by nectins transfers cytoplasmic cargo, including infectious material, between cells

Association of early endosomal autoantigen 1 with macropinocytosis in EGF-stimulated A431 cells

Listeria monocytogenes : a multifaceted model

Effects of cholesterol depletion and increased lipid unsaturation on the properties of endocytic membranes

Receptor-mediated endocytosis of transferrin and recycling of the transferrin receptor in rat reticulocytes

Competitive Regulation of E-Cadherin JuxtaMembrane Domain Degradation by p120-Catenin Binding and Hakai-Mediated Ubiquitination

Amiloride inhibits macropinocytosis by lowering submembranous pH and preventing Rac1 and Cdc42 signaling

E-cadherin phosphorylation occurs during its biosynthesis to promote its cell surface stability and adhesion

RECON-Dependent Inflammation in Hepatocytes Enhances Listeria monocytogenes Cell-to-Cell Spread

Ecadherin is the receptor for internalin, a surface protein required for entry of L. monocytogenes into epithelial cells

Increased internalization of p120-uncoupled E-cadherin and a requirement for a dileucine motif in the cytoplasmic domain for endocytosis of the protein

Caveolae mediate growth factor-induced disassembly of adherens junctions to support tumor cell dissociation

Listeria monocytogenes cell-tocell spread in epithelia is heterogeneous and dominated by rare pioneer bacteria. eLife

Adhesion to the host cell surface is sufficient to mediate Listeria monocytogenes entry into epithelial cells

Role of hemolysin for the intracellular growth of Listeria monocytogenes

The bacterial virulence factor InlC perturbs apical cell junctions and promotes cell-to-cell spread of Listeria

The structure of cell adhesion molecule uvomorulin. Insights into the molecular mechanism of Ca2+-dependent cell adhesion

Listeria monocytogenes Exploits Normal Host Cell Processes to Spread from Cell to Cell✪

Inter-Cellular Exchange of Cellular Components via VE-Cadherin-Dependent Trans-Endocytosis

RNAi screen reveals a role for PACSIN2 and caveolins during bacterial cell-to-cell spread

Cadherin expression is required for the spread of Shigella flexneri between epithelial cells

Foodborne Listeriosis

Structural basis of cell-cell adhesion by cadherins

The rate of actin-based motility of intracellular Listeria monocytogenes equals the rate of actin polymerization

Actin filaments and the growth, movement, and spread of the intracellular bacterial parasite, Listeria monocytogenes

A WAVE2-Arp2/3 actin nucleator apparatus supports junctional tension at the epithelial zonula adherens

Macropinocytosis and cytoskeleton contribute to dendritic cell-mediated HIV-1 transmission to CD4+ T cells

Deconstructing the cadherin-catenin-actin complex

The role of macropinocytosis in the propagation of protein aggregation associated with neurodegenerative diseases

Innate and adaptive immune responses to Listeria monocytogenes: a short overview

Focus area measured as in part A, with and without treatment with EIPA. Between 3 and 8 independent experiments were imaged for each condition and the number of individual foci imaged for each condition from left to right was 493, 460. 182, 194, 158, 113, 210 and 103. P values were determined using the linear mixed-effects model. G. Focus area measured as in part A, with and without treatment with NSC23766. Between 3 and 7 independent experiments were imaged for each condition and the number of individual foci imaged for each condition from left to right was 458, 491. 235, 217, 211, 159, 199 and 137. P values were determined using the linear mixed-effects model. H. Focus area measured as in part A, upon siRNA knockdown of CtBP1 as compared to a non-targeting control (NT). Between 3 and 6 independent experiments were imaged for each condition and the number of individual foci imaged for each condition from left to

Focus