Involvement of CD44v6 in InlB-dependent Listeria invasion


*E-mail; Tel. (+49) 7247 826523; Fax (+49) 7247 823354.


Listeria monocytogenes, a Gram-positive bacterium, is the causative agent for the disease called listeriosis. This pathogen utilizes host cell surface proteins such as E-cadherin or c-Met in order to invade eukaryotic cells. The invasion via c-Met depends on the bacterial protein InlB that activates c-Met phosphorylation and internalization mimicking in many regards HGF, the authentic c-Met ligand. In this paper, we demonstrate that the activation of c-Met induced by InlB is dependent on CD44v6, a member of the CD44 family of transmembrane glycoproteins. Inhibiting CD44v6 by means of a blocking peptide, a CD44v6 antibody or CD44v6-specific siRNA prevents the activation of c-Met induced by InlB. Subsequently, signalling, scattering and the entry of InlB-coated beads into host cells are also impaired by CD44v6 blocking reagents. For the entry process, ezrin, a protein that links the CD44v6 cytoplasmic domain to the cytoskeleton, is required as well. Most importantly, this collaboration between c-Met and CD44v6 contributes to the invasion of L. monocytogenes into target cells as demonstrated by a drastic decrease in bacterial invasion in the presence of blocking agents such as the CD44v6 peptide or antibody.


Bacteria have developed complex strategies to invade mammalian cells and to spread from one cell to another. They express proteins that mimic functions of cellular proteins in order to target host cells and promote their entry. One such bacterium is Listeria monocytogenes, a food-borne pathogen that can cause listeriosis (for review see Hamon et al., 2006). This disease manifests as meningitis, encephalitis, gastroenteritis and mother-to-fetus infections that lead to abortion in pregnant women. L. monocytogenes induces its uptake into non-phagocytic host cells using surface proteins of the internalin family. Although L. monocytogenes expresses several internalins (Inl), only two members, InlA and InlB, are well characterized (reviewed in Bierne et al., 2007). InlA is covalently linked to the bacterial cell wall (Gaillard et al., 1991; Dhar et al., 2000) and binds to E-cadherin (Mengaud et al., 1996). InlB is non-covalently bound to bacterial lipoteichoic acids (Jonquieres et al., 1999) and stimulates c-Met phosphorylation (Shen et al., 2000). Both InlA and InlB can independently mediate L. monocytogenes invasion. After the initial binding to the cell surface, the bacterium is engulfed by a ‘zipper’ mechanism. Activation of cell surface proteins leads to actin remodelling and membrane extensions, so that the bacteria can be driven into the cells (reviewed in Veiga and Cossart, 2006). The vacuoles containing the bacteria are then lysed by listeriolysin O, a molecule that can lyse phagosomal membranes (reviewed in Schnupf and Portnoy, 2007). The bacteria are released into the cytoplasm and move from one cell to another using the actin cytoskeleton.

InlB, like the physiological c-Met ligand HGF, activates c-Met (Shen et al., 2000) and induces downstream signalling pathways, such as the MAPK (Tang et al., 1998; Copp et al., 2003) or the PI3-kinase pathway (Ireton et al., 1996; Ireton et al., 1999), that lead to changes in the actin cytoskeleton organization. Inhibition of the MAPK (Tang et al., 1998) and the PI3K pathways by specific inhibitors or transfection of a dominant negative form of p85α that prevents PI3-kinase activation (Ireton et al., 1996) led to reduced bacterial invasion.

Activated c-Met is internalized, a process that normally regulates receptor tyrosine kinase activation and turnover (Petrelli et al., 2002). Indeed, shortly after activation by their ligands, RTKs like c-Met are endocytosed, in most cases through a clathrin-dependent mechanism (reviewed in Teis and Huber, 2003). This internalization process is subverted by L. monocytogenes to invade eukaryotic cells (Li et al., 2005; Veiga and Cossart, 2005).

Despite their functional resemblance, InlB and HGF share no sequence homology and are structurally unrelated, and the known binding sites for InlB and HGF on c-Met are distinct (Stamos et al., 2004; Niemann et al., 2007). InlB consists of several domains including an N-terminal internalin domain (also called InlB321) followed by a so-called B-repeat and three C-terminal glycine/tryptophan-rich (GW) domains (for review see Bierne et al., 2007). The internalin domain is critical for binding to c-Met and essential for the invasion of bacteria (Braun et al., 1999; Shen et al., 2000). The concave face of the internalin domain forms a functionally critical high affinity interface with c-Met (Machner et al., 2003; Niemann et al., 2007). The GW domains interact with lipoteichoic acid in the bacterial cell wall (Jonquieres et al., 1999) and they bind to host–cell heparan sulphate proteoglycans (HSPG) (Jonquieres et al., 2001; Banerjee et al., 2004). The binding to HSPGs enhances bacterial entry and is critical for full stimulation of c-Met (Jonquieres et al., 2001; Banerjee et al., 2004). HSPGs most likely induce clustering of c-Met bound to InlB, because the highly sulphated, soluble glycosaminoglycan heparin, induces higher order oligomerization of InlB/c-Met complexes in vitro (Niemann et al., 2007). The soluble internalin domain on its own induces phosphorylation of c-Met, but cannot elicit phenotypic cellular responses like cell scattering or proliferation. Interestingly, in the case of HGF, heparan sulphation is not required for activation of the c-Met receptor. Indeed, a mutant of HGF deficient in heparan-sulphate binding is even more potent for c-Met activation than wild-type HGF (Hartmann et al., 1998). Furthermore, removal of heparan sulphate moieties from cell surface molecules does not interfere with the activation of c-Met (Orian-Rousseau et al., 2002).

In many different cancer cells as well as primary cells, HGF and c-Met require specific CD44 isoforms, collectively named CD44v6 (see below), for their functions (Orian-Rousseau et al., 2002). CD44 forms a family of transmembrane glycoproteins that play important roles in many cellular processes, among which are the regulation of growth, survival, differentiation and migration (Ponta et al., 2003). The smallest isoform is called CD44s or CD44 standard. Larger isoforms differ thereof mainly in their extracellular domain in which inclusion of 10 so-called ‘variant exons’ in various combinations can take place (Ponta et al., 2003). In this paper, the term CD44v6 designates isoforms that contain the v6 variant exon either alone or in combination with other variant exons. All CD44v6 isoforms are able to act as a co-receptor for c-Met (Orian-Rousseau et al., 2002). In pathological situations, these isoforms confer metastatic propensity to several tumor cells (reviewed in Naor et al., 2002).

The role of CD44v6 for c-Met activation is twofold: the extracellular part of CD44 is needed for activation of the receptor itself whereas the cytoplasmic tail is instrumental for signalling (Orian-Rousseau et al., 2002; Orian-Rousseau et al., 2007). In the extracellular part, a mutation of three amino acids in the exon v6 leads to loss of c-Met activation (Matzke et al., 2005). The tripeptide sequence corresponds to EWQ in rat, RWH in human and GWQ in mouse. Small peptides (the minimal size being 5 amino acids) covering this region can block the activation of c-Met in vitro similarly to CD44v6-specific antibodies (Matzke et al., 2005). Furthermore, the metastatic spreading of carcinoma cells expressing CD44v6 was completely blocked by intravenous or intratumoral injection of CD44v6 peptides (A. Matzke, unpubl. results). CD44v6 peptides show species specificity, so that, e.g. rat peptides exert their blocking function on rat, but not on human or mouse cells. This contrasts with the effect of the complete, soluble CD44v6 ectodomain. Transfection of, e.g. the rat CD44v6 ectodomain abrogated activation of c-Met in human cells. However, this inhibition was released by incubation with the rat peptides. These data give hints on the mode of action and suggest that the peptides address CD44v6 itself, probably interfering with its correct folding.

The cytoplasmic domain of CD44v6 recruits the actin cytoskeleton via binding to ERM (Ezrin-Radixin-Moesin) proteins (Orian-Rousseau et al., 2007). CD44v6, HGF, c-Met and ERM proteins, together with the cytoskeleton, form a signalosome complex that promotes signalling. For instance, Ras activation by its guanine–nucleotide exchange factor SOS did not occur in cells transfected with a mutant form of CD44v6 where the cytoplasmic domain had been removed. Furthermore, siRNA repressing ezrin abrogated HGF induced signalling to Erk. Finally, an ezrin protein that lacked the actin-binding domain (ezΔABD) repressed the activation of Erk induced by HGF (Orian-Rousseau et al., 2007).

The co-receptor function of CD44v6 for c-Met is not unique. For instance, VEGFR-2 that plays a pivotal role in angiogenesis also requires CD44v6 for activation and for signalling. Interestingly, the same CD44v6 peptides that block c-Met activation also inhibited the activation of VEGFR-2 induced with VEGFA-165 and reveal a role of CD44 in angiogenesis (M. Tremmel, unpubl. results). More and more evidence shows that several RTK-ligand units recruit other players such as cell adhesion molecules most likely to fine-tune the signalling events. Examples are FGFRs or EGFRs that recruit members of the syndecan and cadherin family as co-receptors (reviewed in Orian-Rousseau and Ponta, 2008).

Here, we show that CD44v6 is required for InlB-induced c-Met activation. A CD44v6 peptide, a CD44v6 antibody or downregulation of CD44v6 by means of siRNA blocked the activation of c-Met and downstream signalling after InlB treatment. Furthermore, a CD44v6-negative cell line was rendered InlB-responsive by transfection of a CD44v6 isoform and InlB induces formation of a complex between CD44v6, c-Met and ezrin. CD44v6 blocking agents impaired scattering induced by InlB, and most strikingly, the entry of InlB-coated beads into cells. The entry of beads is also tightly dependent on ezrin binding to the cytoskeleton. Finally, bacterial infection was strongly decreased by CD44v6 blocking agents.


InlB requires CD44v6 for c-Met activation

CD44v6 is necessary for HGF-induced c-Met activation (Orian-Rousseau et al., 2002; Matzke et al., 2005; Orian-Rousseau et al., 2007). This was demonstrated by means of CD44v6-specific antibodies, CD44v6 peptides and CD44v6-specific siRNA that all blocked the activation of c-Met. We used these tools to investigate whether CD44v6 also plays a role in InlB-induced activation of c-Met and signalling. In HT29 and in HeLa cells, where a contribution of CD44v6 to HGF-induced c-Met activation was already observed (for HT29, Orian-Rousseau et al., 2002; Matzke et al., 2005; for HeLa, our own unpublished experiments), a human CD44v6 14-mer peptide completely abrogated InlB-induced activation of c-Met (Fig. 1A and B). The optimal concentration of InlB used for c-Met activation throughout the paper is 1 nM. This contrasts to the optimal HGF concentration, which is 0.15 nM (Orian-Rousseau et al., 2007). Also downstream signalling monitored by Erk and Akt activation was inhibited (Fig. 1A and B), indicating that both the MAPK and PI3K pathways were blocked. A control peptide had no effect. By means of siRNA against CD44v6, we further confirmed the requirement of CD44v6. Two different siRNAs downregulated CD44v6 expression and thereby inhibited InlB-induced c-Met and Erk activation (Fig. 1C). The isolated internalin domain (InlB321) was also able to activate c-Met and downstream signalling but only at high concentrations (100 nM instead of 1 nM for InlB), confirming previously published results (Banerjee et al., 2004; Niemann et al., 2007). This activation was dependent on CD44v6 as well (Fig. 1D).

Figure 1.

InlB requires CD44v6 for c-Met activation.
A. HeLa cells were induced with InlB at a concentration of 1 nM for 5 min at 37°C. Pretreatment with a control peptide or a CD44v6 14-mer peptide at a concentration of 100 ng ml−1 for 30 min was performed where indicated. Phosphorylation of c-Met, of Erk and Akt were measured as described in Experimental procedures.
B. The same experiments were performed with HT29 cells.
C. HeLa cells were transfected with two different siRNA (v6-1 and v6-2, described in Experimental procedures) to downregulate CD44v6 and with a control siRNA. Forty-eight hours after transfection, cells were starved for additional 24 h. InlB-induced c-Met and Erk phosphorylation were measured. The amount of CD44v6 protein was detected by Western blot analysis.
D. HeLa cells were induced with InlB321 at a concentration of 100 nM for 5 min at 37°C. Pretreatment with a control peptide or a CD44v6 14-mer peptide was performed as described in (A). Phosphorylation of c-Met and of Erk was measured as described in Experimental procedures.
The numbers reflect –fold induction as determined by densitometric scanning (Image J program).

Thus, a CD44v6-specific peptide, antibody and siRNA against CD44v6 are able to block InlB-induced activation of c-Met and subsequent signalling. The HT29 and HeLa cells used in these experiments express CD44 variant isoforms containing exon v6 together with other variant exons (for HT29 see Orian-Rousseau et al., 2002 and for HeLa, our unpublished data). In order to confirm the dependency of c-Met activation on CD44v6 and to test whether an isoform containing exclusively exon v6 was sufficient for the activation of c-Met, we used rat pancreatic carcinoma cells (BSp73AS abbreviated AS, Fig. 2) or these cells transfected with such a CD44v6 isoform (BSp73ASs6 abbreviated ASs6). These cells express this CD44v6 isoform as the only variant isoform in addition to CD44s (Orian-Rousseau et al., 2002). In the AS cells, no activation of Erk was observed after addition of InlB, whereas activation was detected in cells expressing CD44v6 (Fig. 2A). Therefore, this experiment not only confirmed that a CD44v6 isoform is needed for InlB-induced activation of c-Met, but it also demonstrated that exon v6 inclusion alone was sufficient for this function. As expected, InlB-induced c-Met and Erk phosphorylation were abrogated in these ASs6 cells upon treatment with the CD44v6 peptide (Fig. 2B).

Figure 2.

A CD44v6 isoform is sufficient for InlB-induced c-Met activation.
A. AS cells or ASs6 cells were induced with InlB as indicated and Erk phosphorylation was determined.
B. ASs6 cells were stimulated with InlB in the absence and presence of a CD44v6 14mer peptide or an unrelated (control) peptide. Erk phosphorylation and c-Met phosphorylation were determined.
C. HeLa cells were induced either with HGF or InlB for 5 min at 37°C. CD44 was immunoprecipitated from the lysates and a Western blot was performed using CD44v6, c-Met and ezrin antibodies as indicated. For control, an IgG antibody was used.
The numbers reflect –fold induction as determined by densitometric scanning (Image J program).

CD44v6, c-Met and InlB form a complex

Our results so far suggest that CD44v6, c-Met and InlB are in close vicinity and form a complex. To identify such a complex we immunoprecipitated CD44 isoforms and checked the presence of c-Met in the complex. Indeed, upon induction with InlB, endogenous c-Met and CD44 were co-immunoprecipitated (Fig. 2C), whereas no complex was observed in the absence of InlB. In the control IgG immunoprecipitation, no association between c-Met and CD44 was observed.

CD44v6 mediates InlB-induced scattering and entry of InlB-coated beads

InlB, similarly to HGF, can induce complex biological responses such as cell scattering upon activation of c-Met (Shen et al., 2000). We tested whether scattering was also dependent on CD44v6. As expected, both a CD44v6 antibody and a CD44v6-specific peptide blocked this response (Fig. 3).

Figure 3.

CD44v6 plays a role in InlB-induced scattering. Scattering of HT29 cells was determined after treatment with InlB, alone or together with a control peptide, or a CD44v6 peptide or antibody as indicated. The magnification used was ×20.

InlB-activated c-Met is internalized, leading to the entry of the bacteria into the host cells (Li et al., 2005; Veiga and Cossart, 2005). This process can be simulated using InlB-coated latex beads that are also internalized into mammalian cells (Braun et al., 1998; 1999). The latex beads invasion assay was used to test the effect of a CD44v6 peptide on InlB-induced cellular entry. HeLa cells were incubated with InlB-coated beads or control beads. In phase-contrast microscopy all beads, whether inside or outside the cells, appear black. A Cy3-labelled antibody detects the beads that remain outside, as the cells are not permeabilized. An overlay between the two pictures allows distinguishing between beads inside (black) and outside (red) the cells (Fig. 4A). Control beads without InlB are all excluded from the cells whereas InlB-coated beads can enter (Fig. 4A). A drastic decrease of the entry was observed when cells were pre-incubated with the CD44v6 peptide. The inhibition of the entry was close to 100%. The control peptide had only a minor effect on the uptake (Fig. 4B). These data demonstrate that CD44v6 is instrumental for the entry of InlB-coated beads into the cells.

Figure 4.

Uptake of InlB-coated latex beads is dependent on CD44v6.
A. HeLa cells were incubated with control beads or beads coated with InlB (see Experimental procedures). Cells were pretreated with a CD44v6 peptide or a control peptide for 1 h as indicated. Extracellular beads bind to a Cy3-labelled antibody and are stained red in immunofluorescence microscopy. In phase-contrast microscopy both extracellular and intracellular beads are detected. An overlay of both pictures is shown. To better visualize the internalized beads, the picture was enlarged (inset).
B. Counts of internalized beads are represented. On the y-axis is the relative uptake (±SD) from four experiments.

ERM proteins are essential for entry of InlB-coated beads in cells

The recruitment of ERM proteins, together with the cytoskeleton to CD44v6, is a decisive step in HGF-dependent signal transduction (Orian-Rousseau et al., 2007). Also in the case of InlB, ezrin can be found in a complex with CD44 and c-Met (Fig. 2C). In order to investigate if a link of ezrin to the cytoskeleton is necessary for internalization of InlB-coated beads, we used an ezrin protein that lacks the actin-binding site (ezΔABD) (Algrain et al., 1993). This protein prevents signal transduction (Orian-Rousseau et al., 2007). In the case of HGF, it competes with endogenous ezrin and has a dramatic effect on Erk phosphorylation whereas c-Met activation itself is not affected. The same is true for InlB-induced c-Met and Erk activation (Fig. 5A). Transfection of HeLa cells with this ezΔABD construct completely abrogated the entry of InlB-coated beads, whereas transfection with a control vector had no effect (Fig. 5B and C). Thus, the link of ezrin to the cytoskeleton is also instrumental for entry of InlB-coated beads.

Figure 5.

Ezrin-dependent uptake of InlB-coated beads.
A. InlB-dependent c-Met and Erk phosphorylation were measured in HeLa cells transiently transfected with a control vector or a vector coding for a truncated version of ezrin lacking the actin binding domain (ezΔABD).
B. In these cells also the uptake of latex beads was determined as described in Fig. 4. HeLa cells were transfected with a CD44v6 protein devoid of the cytoplasmic domain (CD44v6Δcyt) in parallel with the ezΔABD transfection. The uptake of InlB-coated beads was measured as compared with untransfected cells and control beads.
C. Uptake of beads has been quantified. On the y-axis is the relative uptake (±SD) from four experiments.

Is the binding of ezrin to the membrane via CD44 also necessary for bacterial uptake? To test this assumption, we transfected cells with a CD44v6 construct devoid of the cytoplasmic domain (CD44v6Δcyt) (Fig. 5C). This mutant protein completely blocked the entry of InlB-coated beads, suggesting that a link between the cytoplasmic domain of CD44, ezrin and the cytoskeleton drives signalling from c-Met induced by InlB.

InlB-dependent invasion of Listeria relies on CD44v6

The data accumulated till now strongly point towards a requirement of CD44v6 for bacterial invasion. To prove this assumption, we performed bacterial invasion experiments in HeLa cells using the gentamicin protection assay. In order to exclusively address c-Met-dependent infection, we used a mutant strain with a chromosomal in-frame deletion in inlA (ΔinlA2) (Lingnau et al., 1995). Invasion of HeLa cells by this strain was only about 60% as compared with wild-type bacteria. Invasion of an isogenic mutant with a chromosomal in-frame deletion in inlB (ΔinlB2) (Lingnau et al., 1995) was reduced to 20%, whereas a double mutant ΔinlAB2 (Parida et al., 1998) hardly invaded at all (data not shown), indicating that only these two internalins are crucial for the invasion in the HeLa cells. In the presence of the CD44v6 peptide and the CD44v6 antibody, there was an inhibition of ∼60% and 70%, respectively, of the invasion of HeLa cells by the ΔinlA2 mutant strain (Fig. 6). The control peptide and the control antibody have only a marginal effect. Invasion by the ΔinlB2 mutant strain is not affected at all by the CD44v6 blocking reagents (data not shown). These reagents had no effect on adhesion of bacteria to HeLa cells (Fig. 6 inset), indicating that it is the invasion process itself that is CD44v6 dependent.

Figure 6.

Requirement of CD44v6 for bacterial uptake. Inhibition of L. monocytogenes DinlA2 mutant entry by incubation of HeLa cells with a CD44v6 peptide and a CD44v6 antibody. HeLa cells were incubated with DMEM (1% FCS) alone (−) or DMEM (1% FCS) containing a CD44v6 peptide, a control peptide, a CD44v6 antibody or a control antibody as indicated. Cells were then infected with L. monocytogenes DinlA2 and a gentamicin protection assay was performed as described in Experimental procedures. On the y-axis the relative invasion or adhesion (inset) (±SD) from four independent experiments is represented. Data were expressed as relative adhesion and relative invasion compared with the L. monocytogenes DinlA2 mutant without the peptide or the antibody.

In conclusion, the co-receptor function of CD44v6 for c-Met is not only required for InlB-dependent c-Met activation and signalling but appears essential for InlB-dependent entry of L. monocytogenes into host cells.


In this paper, we show that InlB-mediated activation of c-Met depends on CD44v6. Indeed, InlB-induced c-Met activation as well as downstream Erk phosphorylation can be blocked by means of a CD44v6 peptide, antibody and upon downregulation of CD44v6 by siRNA. Consequently, InlB-induced cell scattering is also inhibited using CD44v6 blocking reagents. The induction of c-Met via InlB in cells that lack CD44v6 is restored upon transfection of CD44v6. InlB promotes complex formation between c-Met, CD44v6 and ezrin as shown by co-immunoprecipitation. Uptake of InlB-coated beads that mimics the entry of L. monocytogenes into cells depends on full-length CD44v6 and on ezrin binding to the cytoskeleton. Finally, a proof of the role of CD44v6 in L. monocytogenes invasion comes from the observation that treatment of cells with the CD44v6 peptide or the CD44v6 antibody leads to a drastic decrease of bacterial uptake.

We have previously shown that CD44v6 acts as a co-receptor for c-Met when induced by its authentic ligand, HGF (Orian-Rousseau et al., 2002; Matzke et al., 2005; Orian-Rousseau et al., 2007). According to the results presented here, c-Met requires the same co-receptor for its activation by two different ligands. In the case of EGFR, we observed a different situation. EGFR binds several growth factors, among which are EGF and transforming growth factor-α (TGFα) (reviewed in Hynes and Lane, 2005). Activation of EGFR via EGF is dependent on CD44v6, whereas activation of EGFR via TGFα is independent of this isoform (our own unpublished data).

InlB recruits HSPGs to activate c-Met (Jonquieres et al., 2001; Banerjee et al., 2004). Does CD44v6 itself act as an HSPG in the case of InlB induction of c-Met? This seems not to be the case as only the CD44v3-containing isoforms can be heparan sulphated and HT29 cells do not express the v3 and the v6 exons in the same isoform (Orian-Rousseau et al., 2002). Furthermore, we have shown that an isoform of CD44 containing only the v6 exon was sufficient to support InlB-induced Erk phosphorylation. This isoform can certainly not be heparan sulphated. These results suggest that the heparan sulphate component is provided by another HSPG present on the cell surface.

The recruitment of HSPGs by InlB leads to the clustering of c-Met, a process required for full activation (Niemann et al., 2007). As CD44v6 is not heparan sulphated, we can conclude that the CD44v6 co-receptor function is not directly involved in the clustering of the c-Met receptor. This is in agreement with the observation that InlB321 that cannot bind to HSPGs is also dependent on CD44v6 to induce phosphorylation of c-Met. Interestingly, in the case of HGF-dependent activation of c-Met, heparan sulphate seems not to play a role (Orian-Rousseau et al., 2002) and is not involved in the oligomerization of c-Met (Gherardi et al., 2003).

How does the CD44v6 peptide interfere with c-Met activation? Binding studies using purified InlB proteins revealed that these proteins are unable to bind to cells in the presence of the CD44v6 peptide (our own unpublished results). This observation suggests that CD44v6 enables the binding of InlB to cells and that the interference of the peptide with c-Met activation may occur on the level of binding of InlB. We hypothesize that the ectodomain of CD44v6 can bind to several growth factors including VEGF, HGF, EGF and to InlB and that the CD44v6 peptide interferes with the conformation of the ectodomain, thereby abrogating the binding of the different ligands. This would explain how CD44v6 acts as a co-receptor for different RTKs (Trk, c-Met, Ron, VEGFR-2, EGFR) and why these functions are all inhibited by the same peptide.

The activation of the c-Met receptor via InlB is the first step necessary for InlB-mediated uptake of L. monocytogenes into host cells. We have simulated bacterial uptake by means of InlB-coated beads. Their uptake can be completely inhibited by CD44v6-specific peptides suggesting that CD44v6 is instrumental for infection of cells by L. monocytogenes. This was directly shown for invasion of HeLa cells by L. monocytogenes as it is blocked by both the CD44v6 peptide and antibody. The adhesion of bacteria to cells was not significantly changed. This is in agreement with the observation that in a ΔinlAB mutant other proteins, like the autolysin Ami, significantly contribute to adhesion of L. monocytogenes to eukaryotic cells (Milohanic et al., 2001).

InlB-dependent signalling requires binding of the actin cytoskeleton to ezrin and CD44. Transfection of an ezrin protein lacking the actin-binding domain leads to a drastic reduction of Erk phosphorylation but not of c-Met activation. More strikingly, also the uptake of InlB-coated beads is dependent on ezrin binding to the cytoskeleton and on the CD44 cytoplasmic domain that has been shown to bind ERM proteins (Tsukita et al., 1994). Ezrin might play a role on two levels as far as bacterial invasion is concerned. The link between the actin cytoskeleton to the membrane via CD44 and ezrin might be necessary for bacterial invasion in that it facilitates the clathrin-dependent internalization. Interestingly, a key role for actin during clathrin-mediated endocytosis in mammalian cells was demonstrated (Yarar et al., 2005). Furthermore, ezrin binding to the cytoskeleton and to CD44 is necessary for signal transduction via c-Met (Orian-Rousseau et al., 2007) that in turn is necessary for actin remodelling (for review see Ivetic and Ridley, 2004). In addition to several host molecular effectors such as the Arp2/3 complex (for review see Seveau et al., 2007), ezrin might be required to orchestrate actin remodelling and bacterial invasion.

Listeria monocytogenes represents a class of bacteria that enters via a zipper mechanism, whereas Shigella is the prototype of a bacterium that invades cells through a trigger mechanism (for review see Veiga and Cossart, 2006). In both cases, a CD44 isoform is required for invasion. Indeed, Shigella, which is responsible for bacillary dysentery in humans, also makes use of CD44 for its entry into mammalian cells (Skoudy et al., 2000). It secretes several proteins that are essential for the infection process. One of them, IpaB, interacts with CD44s. This binding is essential for bacterial invasion as blocking of CD44 by means of antibodies inhibits bacterial entry. The common feature of all CD44 isoforms is the binding of the ERM proteins through the cytoplasmic domain. This is a property that might be generally required for invasion. Interestingly, Shigella invasion via CD44 requires ezrin for the entry process (Skoudy et al., 1999).

c-Met is also involved in infectious diseases other than listeriosis. Both Plasmodium falciparum and Helicobacter pylori depend on or subvert c-Met signalling (Carrolo et al., 2003; Churin et al., 2003; Oliveira et al., 2006). The requirement for CD44v6 should also be addressed in these different cases where the CD44v6 blocking peptide might be useful as tool to fight infection.

Taken together, our data indicate that a link from c-Met to the cytoskeleton via the CD44v6 cytoplasmic domain and ERM proteins is required for InlB-dependent uptake of beads into cells. Furthermore, the InlB-dependent invasion of L. monocytogenes also relies on the collaboration between c-Met and CD44v6.

Experimental procedures

All the experiments throughout the paper were performed at least three times except for the latex bead uptake assay (Figs 4 and 5), and the bacterial invasion and adhesion assays (Fig. 6) in which at least four independent experiments have been combined.

Cells and bacteria

The human colon adenocarcinoma cell line HT29, a gift from A. Zweibaum (INSERM; France) and the human cervix carcinoma cell line HeLa (American tissue culture collection, ATCC; Wesel, Germany. Accession No: CCL-2) were grown in Dulbecco's modified Eagle's medium (DMEM; Invitrogen, Karlsruhe, Germany) supplemented with 10% fetal calf serum (FCS; PAA Cölbe, Germany). The rat pancreatic carcinoma cell line BSp73AS and its transfectant BSp73ASs6 have previously been described (Orian-Rousseau et al., 2002) and were grown in RPMI (Invitrogen, Karlsruhe, Germany) plus 10% FCS.

Listeria monocytogenes EGD (Glaser et al., 2001) was grown in BHI (brain–heart infusion) broth. The isogenic deletion mutants ΔinlA2, ΔinlB2 and ΔinlAB2 have been described previously (Lingnau et al., 1995; Parida et al., 1998).

Antibodies and other reagents

The human monoclonal antibody against CD44v6 (Biwa) was obtained from Bender (Vienna, Austria). The pan-CD44 antibody IM7 was from Pharmingen and the pan-CD44 antibody Hermes 3 was a kind gift from S. Jalkanen (Turku, Finland). The antibody against Erk 1 (K-23) and Akt (H-136) were from Santa Cruz (CA, USA) and the ezrin antibody (3C12) from Neomarker (Fremont, CA, USA). The phospho-Erk (Phospho-p44/42 Map Kinase), the phospho-Akt (Ser437), phospho-Met (Tyr 1234/1235) and Met (25H2) antibodies were purchased from Cell Signaling Technology (Beverly, England). Secondary antibodies labelled with HRP were purchased from Dako (Hamburg, Germany). Mouse IgG was obtained from Santa Cruz (CA, USA). The hybridoma cell supernatant of anti-InlB monoclonal antibody [IC100 (Lingnau et al., 1995)] has been described previously. HGF was a generous gift of George Vande Woude (Van Andel Institute, USA). InlB was prepared as described (Niemann et al., 2007). The CD44 v6 peptide (14mer: KEQWFGNRWHEGYR) and the control peptide (described in Matzke et al., 2005) were synthesized by NMI Technology Transfer (Reutlingen, Germany). An ezrin construct in which the last 29 amino acids encoding the actin-binding domain have been deleted was kindly provided by Monique Arpin (Institute Curie, Paris, France). The CD44Δv6 construct was described previously (Orian-Rousseau et al., 2002). The Cy-3 labelled secondary antibody was bought from Dianova (Hamburg, Germany).

Detection of c-Met, Erk and Akt phosphorylation

In all experiments cells were induced with 1 nM of InlB or 100 nM of InlB321 after 24 h of starvation where indicated. Induction was performed for 5 min at 37°C. Blocking experiments were performed by incubating the CD44v6 14-mer peptide (100 ng ml−1) or the control peptide (100 ng ml−1) for 30 min at 37°C prior to induction with InlB. Cells were lysed with sample buffer + DTT and subjected to SDS-PAGE. The corresponding blots were treated with the respective antibodies according to the manufacturer's instructions.


HeLa cells were induced with InlB (1 nM) or with HGF (0.15 nM) for 5 min at 37°C after 24 h of starvation. Cells were lysed using a buffer containing 25 mM HEPES, pH 7.5; 100 mM NaCl; 10 mM MgCl2; 1 mM EDTA; 10% Glycerol; 1% NP40. Lysates were cleared by centrifugation at 15000 r.p.m. for 30 min. Cleared lysates were incubated with a pan-CD44 antibody (IM7) over night followed by an incubation with a mixture of Protein A and Protein G Agarose beads (Pierce, Rockford, USA) for 2 h. The beads were washed three times with the lysis buffer and solubilized in sample buffer + DTT. The samples were loaded on an SDS-PAGE gel and blotted with the phospho-Met, ezrin or CD44v6 (Biwa) antibody as indicated in the figure.

Scattering assay

The scattering assay has been described previously (Orian-Rousseau et al., 2002). Briefly, HT29 cells were seeded at a density of 3 × 105 cells per well in a 6-well plate and starved for 24 h. They were then incubated with a InlB (1 nM) for 48 h at 37°C. For the blocking experiments, CD44 anti-v6 antibody (Biwa; 100 μg ml−1) or a CD44v6 peptide (100 ng ml−1) were added before addition of InlB. Pictures were taken using a phase-contrast microscope 48 h after induction.

siRNA inhibition

Cells were transfected with CD44v6-specific siRNAs (eurofins MWG GmbH, Ebersberg, Germany): v6-1: siRNA 25nt 5′-AGU AGU ACA ACG GAA GAA ATT-3′; v6-2: siRNA 25nt 5′-GGA UAU CGC CAA ACA CCC ATT-3′; and control siRNA (AATTCTCCGAACGTGTCACGT) (Qiagen, Hilden, Germany, Cat. No. 1022076) using lipofectamine 2000 (Invitrogen, Karlsruhe, Germany) according to the manufacturer's protocol. Forty-eight hours after transfection the cells were starved for 24 h and then treated with InlB as described above.

Latex bead invasion assay

Coating of the beads with InlB was performed according to Machner et al. (2003). Briefly, 25 μl of latex beads (4 × 108 beads ml−1, Dynabeads M-450, Dynal, Invitrogen, Karlsruhe, Germany) were coated with goat anti-mouse IgG and then incubated with 0.5 ml of hybridoma cell supernatant of monoclonal anti-InlB (IC100) antibody for 2 h at 4°C. After washing with PBS, the beads were incubated with InlB (3 μM). Latex beads coated with both the goat anti-mouse IgG and the anti-InlB mAb were used as control beads.

For the invasion assay, 2 × 104 cells were seeded in chamber slides (LabTek Chamber slides, Nunc, Langenselbold, Germany). Two types of experiments were performed. In the case of the blocking with the CD44v6 peptide, the cells were starved for 24 h and then treated with the CD44v6 peptide or control peptide at a concentration of 100 ng ml−1 for 30 min.

In the case of the transient transfection with the ezrinΔABD construct, the CD44v6Δcyt construct or the control vectors, the cells were transfected using lipofectamine 2000 according to the manufacturer's protocol.

In both cases, the cells were then incubated for 2 h with the InlB-coated beads or control beads in 0.2 ml of complete DMEM at 37°C. The cells were washed with 0.5 ml of medium and incubated in medium for 1 h at 37°C. After this incubation time, the cells were washed three times with 1 ml of CB buffer (10 mM Pipes, 150 mM NaCl, 5 mM EGTA, 5 mM glucose, 5 mM MgCl2, 100 μg ml−1 streptomycin), fixed with 4% paraformaldehyde in CB buffer for 30 min and washed with PBS. The non-permeabilized cells were incubated with 0.2 ml of BSA (1% in CB buffer) and incubated with a Cy3-labelled goat anti-mouse antibody (1:200 dilution) to detect the extracellular beads. Intracellular and extracellular beads were detected by phase-contrast microscopy. An overlay of fluorescence scans and bright field scans show the internalized beads in black and the extracellular beads in red. Cells were mounted in PVA and viewed using microscope [Axioskop 200 M (fluo), Zeiss, Jena, Germany]. Intracellular beads were counted. Each experiment was repeated at least four times and results were analysed statistically.

Adherance and invasion assay (gentamicin protection assays)

Adherence of L. monocytogenes (EGD or ΔInlA2, ΔInlB2 and ΔinlAB2) to and invasion into HeLa cells were measured with a gentamicin protection assay. In brief, for adherence assays, the HeLa cells were infected with log phase bacteria and a multiplicity of infection (moi) of 5 (∼5 × 106 bacteria/1 × 106 cells per well) in DMEM tissue culture medium supplemented with 1% heat-inactivated FCS for 1 h at 37°C in 5% CO2. For the inhibition studies HeLa cells were pre-incubated with the CD44v6 peptide and control peptide at a concentration of 100 ng ml−1, Biwa (anti-CD44v6 antibody) and control mouse IgG at a concentration of 100 μg ml−1 for 30 min. After 1 h of infection the HeLa cells were washed three times with DMEM and were subsequently lysed by adding 500 μl of 1% Triton X-100. The number of cell-adherent bacteria was determined by plating appropriate dilutions of the lysate onto agar plates. Due to the lysis of the eukaryotic cells in this process, the calculation of cell-adherent bacteria also included bacteria that had invaded into HeLa cells. Therefore, the number of invaded bacteria was subtracted from the numbers of cell-adherent bacteria to calculate the actual number of adherent bacteria. For invasion assays, the HeLa cells were also infected with log phase L. monocytogenes and an moi of 5. Bacteria were incubated for 1 h at 37°C in 5% CO2, and washed three times with DMEM. Subsequently, the infected cells were incubated for 2 h in tissue culture medium supplemented with gentamicin (50 μg ml−1) to kill extracellular bacteria. After three washes with DMEM, the HeLa cells again were lysed by adding 500 μl of 1% Triton X-100. The number of invasive bacteria was quantified by plating serial dilutions of the lysate onto agar plates.


We are grateful to Helmut Ponta (Forschungszentrum Karlsruhe, Institute for Toxicology and Genetics, Germany) for critical reading of the manuscript. We are thankful to Professor Horst Schroten (University Children Hospital, Mannheim, Germany) for helpful discussions. We thank Christina Geerds (Bielefeld University) for expert technical assistance.