EspH Suppresses Erk by Spatial Segregation from CD81 Tetraspanin Microdomains

Enteropathogenic Escherichia coli (EPEC) belongs to a group of enteric human pathogens known as attaching-and-effacing (A/E) pathogens, which utilize a type III secretion system (T3SS) to translocate a battery of effector proteins from their own cytoplasm into host intestinal epithelial cells. Here we identified EspH to be an effector that prompts the recruitment of the tetraspanin CD81 to infection sites.

ABSTRACT

Enteropathogenic Escherichia coli (EPEC) belongs to a group of enteric human pathogens known as attaching-and-effacing (A/E) pathogens, which utilize a type III secretion system (T3SS) to translocate a battery of effector proteins from their own cytoplasm into host intestinal epithelial cells. Here we identified EspH to be an effector that prompts the recruitment of the tetraspanin CD81 to infection sites. EspH was also shown to be an effector that suppresses the mitogen-activated protein kinase (MAPK)/extracellular signal-regulated kinase (Erk) signaling pathway at longer infection times. The inhibitory effect was abrogated upon deletion of the last 38 amino acids located at the C terminus of the protein. The efficacy of EspH-dependent Erk suppression was higher in CD81-deficient cells, suggesting that CD81 may act as a positive regulator of Erk, counteracting Erk suppression by EspH. EspH was found within CD81 microdomains soon after infection but was largely excluded from these domains at a later time. Based on our results, we propose a mechanism whereby CD81 is initially recruited to infection sites in response to EspH translocation. At a later stage, EspH moves out of the CD81 clusters to facilitate effective Erk inhibition. Moreover, EspH selectively inhibits the tumor necrosis factor alpha (TNF-α)-induced Erk signaling pathway. Since Erk and TNF-α have been implicated in innate immunity and cell survival, our studies suggest a novel mechanism by which EPEC suppresses these processes to promote its own colonization and survival in the infected gut.

INTRODUCTION

Enteropathogenic Escherichia coli (EPEC) and enterohemorrhagic Escherichia coli (EHEC) are severe human diarrheagenic pathogens that colonize and infect the small and large intestines, respectively. A significant hallmark of the disease caused by these pathogens is their ability to induce an attaching-and-effacing (A/E) lesion phenotype in the intestinal mucosa (1–5). Another A/E-inducing pathogen is Citrobacter rodentium, which colonizes the intestines of laboratory mice. Upon attachment to the epithelial host cell surface, these pathogens rapidly activate a syringe-like machinery, termed the type III secretion system (T3SS), through which they translocate about 20 distinct EPEC effector proteins or 50 distinct EHEC effector proteins from their own cytoplasm into the host. These effectors, which are encoded by genes located in the locus of enterocyte effacement (LEE) (6) and non-LEE pathogenicity islands (7), are strikingly multifunctional (8). They have been shown to hijack a broad array of key host organelles and pathways (7, 9), including the cytoskeleton, membrane trafficking, cell death and survival, and manipulation of the mitogen-activated protein kinase (MAPK) signaling pathways (10). These activities might help the pathogen to successfully colonize its host.

Among their numerous effector proteins, EPEC and EHEC translocate into the host a LEE-encoded protein effector, EspH, which has the capacity to modulate the actin cytoskeleton (11, 12), possibly due to its ability to bind and inhibit Dbl homology-pleckstrin homology (DH-PH) domains of Rho guanine nucleotide exchange factors (RhoGEFs) (13). Additional effects attributed to EspH include focal adhesion disassembly to trigger host detachment and activation of caspase-3 to promote host cell cytotoxicity and death (14). These EspH-dependent activities could facilitate bacterial A/E of the intestinal mucosa and promote the induction of disease (15, 16). At present, the mechanisms by which EspH regulates host signaling have not been comprehensively elucidated.

Tetraspanins are 20- to 50-kDa evolutionarily conserved membrane proteins with four transmembrane domains. To date, the sequences of 33 distinct tetraspanins have been identified in the human genome (17). A widely accepted view suggests that while these membrane proteins lack a known enzymatic or catalytic activity, they can facilitate a myriad of different signaling events by interacting among themselves and with other proteins and lipids to form fairly large and dynamic membrane platforms called tetraspanin-enriched microdomains (TEMs), or the tetraspanin web (17–20). Interestingly, several canonical TEM-associated components have been reported to be hijacked by EPEC, such as β1-integrins (21, 22), phosphatidylinositol 4,5-bisphosphate (23), Rho GTPases (13, 24), and ezrin (25). Studies have shown that the tetraspanin CD81 activates extracellular signal-regulated kinase (Erk) signaling (26). It was also shown that EPEC infection of epithelial host cells can manipulate Erk activity in a T3SS-dependent manner (27–31). However, despite the potential importance of the process in the microbe's life cycle, its underlying mechanism has not been fully explored.

Here we identified a novel interplay between EspH and CD81 in the modulation of the Erk signaling cascade. Our findings suggest that EspH suppresses Erk phosphorylation (pErk) in infected epithelial cells and that CD81 antagonizes this effect. Surprisingly, we found that while EspH promotes CD81 recruitment at infection sites, the ability of the effector to suppress pErK may take place outside (or at the outer edges) of these domains. Our findings shed new light on the mode by which the spatial organization of bacterial effectors can control host signaling.

RESULTS

CD81 clusters at EPEC infection sites.

To gain initial insights into the role of CD81 in EPEC infection, we examined the effect that EPEC may have on the distribution of CD81 in HeLa and Caco-2 cells. Cells infected with wild-type (wt) EPEC or the T3SS-defective mutant (EPEC escV) for 30 min at 37°C were fixed, permeabilized, and immunostained with anti-CD81 antibodies. The cells were also stained with DAPI (4′,6-diamidino-2-phenylindole) and Texas Red (TR)-phalloidin to visualize bacteria and F-actin, respectively. Confocal imaging showed significant clustering of CD81 and F-actin in the EPEC wt but to a much lesser extent in EPEC escV adherence sites (Fig. 1A and B). These results suggest that T3SS-dependent and, to some degree, T3SS-independent mechanisms contribute to CD81 clustering at infection sites. x-z analysis of polarized Caco-2 cell monolayers infected with the EPEC wt showed a shift mainly in the CD81 distribution from the basolateral to the apical pole of the cells. This shift was not observed in EPEC escV-infected cells, where the CD81 signal was mainly distributed at the basal surface (Fig. 1C). These results suggest that T3SS-dependent mechanisms redistribute CD81 to the infected apical surface, where it is potentially more accessible to manipulations by EPEC effectors.

FIG 1.

CD81 clusters at EPEC infection sites. (A and B) CD81 clustering in HeLa and Caco-2 cells. Cells were infected with the EPEC wt or EPEC escV for 30 (HeLa cells) or 45 min (Caco-2 cells) at 37°C, fixed, and stained with anti-CD81 (5A6) antibodies, TR-phalloidin (F-actin), and DAPI (nuclei and bacteria). Cells were then visualized by confocal microscopy. (Left) Representative x-y confocal images. Arrows point to cell-associated EPEC microcolonies. Bar = 5 μm. (Right) Quantitative evaluation of CD81 and F-actin clustering at infection sites was performed as described in Materials and Methods. Results are the mean ± SE for at least 30 infection sites imaged in 3 independent experiments. P values refer to the comparison with EPEC escV-infected cells. ***, P < 0.0005. (C) Apical-basal distribution of CD81 in polarized Caco-2 cell monolayers. (Left) Representative x-z images. (Right) Mean fluorescence levels of CD81 and F-actin along the x-z axis of the monolayer (see Materials and Methods). Results are the mean ± SE for values obtained from 5 different images.

CD81 clustering does not play a role in F-actin recruitment.

EPEC has the capacity to hijack the actin cytoskeleton of its host. At an early infection phase, the bacterium induces the transient formation of filopodia, an activity mediated by the effector Map. At a later stage, interactions between the host-incorporated translocated intimin receptor (Tir) and its cognate outer membrane protein, intimin, induce actin polymerization and the formation of actin-rich protrusions, called pedestals, located beneath bacterial adherence sites (32). CD81 also exhibits a significant cross talk with the actin cytoskeleton (33–36). Thus, we hypothesized that CD81 could be involved in F-actin recruitment to infection sites at any of these particular infection phases. To address this hypothesis, we generated HeLa cell lines whose CD81-encoding gene has been knocked out (CD81KO) using the CRISPR-Cas9 gene editing technology. HeLa cells transfected with the parental CRISPR-Cas9 plasmid served as negative controls (and are referred to here as CRISPR control cells). Western blot analysis showed that while CRISPR control cells exhibited normal CD81 expression, the protein was not detected in the CD81KO cells (see Fig. S1 in the supplemental material). Next, CRISPR control and CD81KO cells were infected with the EPEC wt at different time points reflecting the different phases of actin rearrangement during EPEC infection (32). CD81 and F-actin clustering at infection sites was evaluated by microscopy. Results showed similar levels of clustering of F-actin in CRISPR control and CD81KO cells at all infection times (Fig. 2). These data confirm that CD81 is not involved in F-actin recruitment at EPEC infection sites.

FIG 2.

CD81 is not required for F-actin recruitment at infection sites. CRISPR-control (Control) and CRISPR CD81KO (CD81KO) HeLa cells were infected with the EPEC wt for the indicated times. The cells were then fixed, and the level of clustering of F-actin at infection sites was determined as indicated in Materials and Methods and in the legend to Fig. 1A. Representative images (left) and the results of quantitative analysis of protein clustering at infection sites (right) are presented. P values refer to the comparison with infected CRISPR control cells. ns, nonsignificant statistical difference (P > 0.05). DIC, differential interference contrast. Bars = 10 μm.

EspH stimulates CD81 clustering at infection sites.

The differences in the capacity of the EPEC wt versus EPEC escV to promote CD81 clustering suggested that T3SS components are involved in the clustering effect. To identify these components, we infected HeLa cells with a battery of EPEC strains mutated in their LEE- or non-LEE-encoding genes (Table S1). The cells were then fixed and processed for confocal imaging. The results showed that infection with EPEC ΔespH and EPEC ΔcesT resulted in reduced CD81 clustering at infection sites (Fig. 3A). The CesT protein is a chaperone required for the T3SS-dependent translocation of several protein effectors, including EspH (37). However, our screening did not reveal any effects of the known CesT-dependent effectors (e.g., Tir, Map, NleH) on CD81 clustering, except for EspH (Fig. 3A). To further corroborate the finding that EspH is involved in CD81 recruitment, HeLa or Caco-2 cells were infected with EPEC ΔespH or EPEC ΔespH complemented with an EspH-expressing plasmid (EPEC ΔespH+EspH; Table S3). The results again showed low levels of CD81 clustering at EPEC ΔespH infection sites, reaching levels exhibited by EPEC escV. However, infection with EPEC ΔespH+EspH completely rescued CD81 clustering levels (Fig. 3B). Notably, in these experiments EspH was efficiently translocated into the infected hosts (Fig. S2A). These results demonstrate that EspH is an important effector that contributes to CD81 clustering at infection sites.

FIG 3.

Screening for bacterial effectors that mediate CD81 clustering at infection sites. (A) Screening for LEE and non-LEE effectors involved in CD81 clustering. HeLa cells were infected with the indicated EPEC strains (see Table S1 in the supplemental material), and the levels of CD81 and F-actin clustering at infection sites were determined as described in the legend to Fig. 1A. (B) EspH is involved in CD81 clustering. HeLa or Caco-2 cells were infected with EPEC ΔespH or EPEC ΔespH+EspH strain (Table S1), and EspH expression was induced using IPTG. Cells were processed for confocal imaging as described in the legend to Fig. 1A. Representative images (left) and the results of clustering analysis of CD81 and F-actin (right) are presented. Results are the mean ± SE from 3 experiments. P values refer to the comparison with EPEC wt-infected cells. ***, P < 0.0005; **, P < 0.005; *, P < 0.05; ns, nonsignificant statistical difference (P > 0.05). Bars = 15 μm.

EspH inhibits pErk levels, while CD81 counteracts this effect.

Several studies have linked CD81 to Erk signaling (26, 38–40). We therefore hypothesized that EspH-mediated recruitment of CD81 is somehow linked to Erk signaling. Compared to uninfected or EPEc escV-infected cells, HeLa or Caco-2 cell infection with the EPEC wt for 90 min at 37°C caused a significant reduction in pErk levels (Fig. 4A). It should be noted that the inhibitory effect was not noticeable after 30 min of infection and not consistent following infection for 60 min (not shown). Screening for bacterial effectors identified EspH to be a potential effector involved in the process (Fig. S3). The involvement of EspH was confirmed using mutants that overexpressed EspH: unlike infection with EPEC ΔespH, infection with EPEC ΔespH+EspH resulted in a significant reduction in pErk levels (Fig. 4B). Next, we explored the role of CD81 in the process. CRISPR control cells, CD81KO cells, and CD81KO cells transfected with an mEmerald-CD81-expressing plasmid (Table S3) and thus ectopically overexpressing the fusion protein (CD81KO–mEmerald-CD81; Fig. S4A) were infected with EPEC ΔespH+EspH. Infection with EPEC ΔespH+EspH diminished the pErk levels in CRISPR CD81KO cells compared to those in CRISPR control cells. In contrast, pErk levels were fully restored to levels that were even somewhat higher than those observed in infected CRISPR control cells upon infection of cells overexpressing CD81 (Fig. 4C). Notably, the overexpressed mEmerald-CD81 behaved similarly to the native protein with respect to its cellular distribution and clustering at infection sites (Fig. S4B). Collectively, these data suggest that EspH causes a reduction in pErk levels and that CD81 counteracts this effect.

FIG 4.

EspH reduces pErk levels, and CD81 counteracts this effect. (A) pErk levels are suppressed in EPEC wt-infected but not EPEC escV-infected HeLa and Caco-2 cells. Cells were infected with EPEC for 90 min at 37°C or left untreated. Cells were lysed and processed for pErk and tErk detection by Western blotting, as described in Materials and Methods. Representative Western blots are shown. (B) EspH mediates pErk suppression. HeLa cells were infected with the indicated EPEC strains for 90 min at 37°C or left uninfected. The cells were then lysed and processed for quantitative Western blotting of pErk levels, as described in the legend to panel A and Materials and Methods. Results are the mean ± SE from 3 experiments. P values refer to the comparison with uninfected cells. (C) CD81 counteracts EspH-mediated pErk suppression. CD81KO or CD81KO–mEmerald-CD81 HeLa cells were infected with the indicated EPEC strains and subjected to quantitative Western blotting of pErk levels, as described in Materials and Methods. Results are the mean ± SE from 3 experiments. P values refer to the comparison with infected CRISPR control cells. ***, P < 0.0005; **, P < 0.005; *, P < 0.05.

Deletion of 38 C-terminal amino acids caused a loss in the ability of EspH to suppress pErk.

Our next aim was to identify the structural motifs in EspH that mediate pErk inhibition. Secondary structure prediction analysis using the JPred4 server showed that EspH contains several putative α-helices and β-strand secondary structures which are conserved among EPEC, EHEC, and C. rodentium (Fig. 5A). As protein-protein interactions can be mediated by α-helical structures, we postulated that α-helices might be involved in EspH interactions, with host proteins promoting pErk reduction. To initially address this hypothesis, we generated an EPEC strain expressing EspH whose C-terminal 38-amino-acid segment contains a putative α-helix (Fig. 5A, boxed area) that has been deleted (EPEC ΔespH+EspH_Δ130–168). We found that this mutation disabled the capacity of EspH to reduce pErk (Fig. 5B). A similar phenotype was observed upon ectopic expression of EspH_Δ130–168 labeled with enhanced green fluorescent protein (eGFP) (EspH_Δ130–168-eGFP) (Fig. 5C). In contrast, the phosphorylation level of another MAPK, p38 (Fig. 5D), was statistically indistinguishable in EPEC ΔespH+EspH-, EPEC ΔespH+EspH_Δ130–168-, and EPEC ΔespH-infected cells. These results suggest that translocated EspH selectively inhibits pErk (Fig. 5D). Notably, the translocation efficiency of the mutant EspH was comparable to that of wild-type EspH (Fig. S2B), suggesting that the inability of EPEC ΔespH+EspH_Δ130–168 to suppress pErk cannot be attributed to differences in effector translocation. Finally, as EspH has been suggested to inhibit RhoA GEFs (13), we examined the effects that the wild-type and mutant EspH may have on Rho activity. Interestingly, we found that, unlike EspH_Δ130–168-eGFP, expression of EspH-eGFP inhibited Rho (Fig. 5E). Taken together, these results suggest that the deleted C-terminal segment of EspH may harbor structural information that promotes pErk and Rho inhibition.

FIG 5.

The C-terminal 38-amino-acid segment of EspH is important for pErk increase. (A) Secondary structure prediction of EspH. Prediction of the secondary structure of EspH in EPEC (GenBank accession number SLM08785), EHEC (GenBank accession number ACG59626), and Citrobacter rodentium (GenBank accession number CBG89721) was performed using the JPred4 protein secondary structure prediction server (68). The results were viewed using the Jalview (version 2.10.3b1) program (69). The C-terminal segment (amino acids 114 to 168) is shown. jnetpred predicts secondary structures in EspH. Red tubes and green arrows represent helices and sheets, respectively; JnetCONF shows the confidence of prediction (from 0 to 9; a higher value means a higher-confidence prediction); JnetPSSM refers to Jnet PSIBLAST pssm-based profile prediction; red tubes and green arrows indicate helices and sheets, respectively. The C-terminal part of EspH containing the C-terminal 38-amino-acid segment and a highly conserved predicted α-helix, which was deleted to generate the EspH_Δ130–168 mutant, is indicated with a green box. (B) pErk levels are significantly decreased in host cells infected with EPEC ΔespH+EspH compared to cells infected with EPEC ΔespH or EPEC EspH_Δ130–168. HeLa cells were infected with the indicated EPEC strain, and pErk levels were determined as described in the legend to Fig. 4. Results are the mean ± SE from 3 independent experiments. P values refer to the comparison with EPEC ΔespH-infected cells. (C) pErk levels are decreased in EspH-eGFP-expressing cells compared to eGFP- or EspH EspH_Δ130–168-eGFP-expressing cells. HeLa cells were transfected with plasmids expressing the indicated eGFP. Cells were lysed and analyzed for pErk levels by Western blotting as described in the legend to Fig. 4. eGFP expression levels in the lysates were detected using anti-eGFP antibodies. The results are the mean ± SE from 3 experiments. P values refer to the comparison with eGFP-transfected cells. (D) Phospho-p38 levels were not significantly altered in cells infected with EPEC ΔespH, EPEC ΔespH+EspH, or EPEC EspH_Δ130–168. HeLa cells were infected with the indicated EPEC strains, and phospho-p38 levels normalized to total p38 levels were determined by Western blotting. Results are the mean ± SE from 3 independent experiments. P values refer to the comparison with EPEC ΔespH-infected cells. (E) Analysis of Rho GTPase activity. HeLa cells were transfected with plasmids expressing eGFP, EspH-eGFP, or EspH_Δ130–168-eGFP. The cells were then analyzed for Rho activity by the luciferase-based assay, as described in Materials and Methods. Data were normalized to those for eGFP-expressing cells. P values refer to the comparison with eGFP-transfected cells. In control experiments, Rho GTPases were inhibited by the Rho inhibitor I, and the results were compared to those for dimethyl sulfoxide-treated cells (inset). The results are the mean ± SE for 12 measurements. ***, P < 0.0005; **, P < 0.005; ns, nonsignificant statistical difference (P > 0.05).

Translocated EspH is segregated from CD81-enriched domains.

The results so far suggested that cross talk between translocated EspH and CD81 determines the capacity of EspH to diminish pErk levels. To gain a better understanding of this phenomenon, we took a closer look at the localization of the two proteins in EPEC-infected cells. Confocal imaging showed that early upon infection (at 30 min), wild-type and mutant EspH_Δ130–168 clusters were contained within the CD81 aggregates residing beneath the infecting microcolonies (Fig. 6A). At a later stage (90 min), the signal contributed by wild-type EspH appeared to be largely excluded from the CD81 aggregates, while the translocated mutant effector was still fully contained within them (Fig. 6B). These results suggest a mechanism whereby wild-type EspH is initially translocated into a CD81 domain but at a later stage the effector protein segregates from that domain. This activity might facilitate efficient suppression of pErk levels. In contrast, translocated EspH_Δ130–168 is unable to do this and thus fails to suppress pErk. Interestingly, ectopically expressed EspH-eGFP, but not EspH_Δ130–168-eGFP, also prompted pErk suppression (Fig. 5C). Confocal imaging of these cells showed that CD81 was distributed in the plasma membrane and a large aggregate was located at the cell center, while EspH-eGFP was located primarily in the plasma membrane. In contrast, EspH_Δ130–168-eGFP localized on both the plasma membrane and the central CD81 aggregate (Fig. 6C). These data reinforce the hypothesis that CD81 platforms counteract the ability of EspH to diminish Erk phosphorylation and that the effector has to be located at a distance from the CD81 domain to exert its inhibitory effect on Erk.

FIG 6.

EspH moves out of the CD81 microdomain. (A and B) Localization of translocated EspH with respect to mEmerald-CD81. mEmerald-CD81-expressing cells were infected for 30 (A) or 90 (B) min at 37°C. Cells were then fixed and stained with anti-SBP antibodies to label EspH. Cells were imaged by confocal microscopy, and an intensity profile was created along a drawn line over CD81-EspH clusters, as exemplified in the boxed areas. The plot represents the average of intensity profile values measured for ∼20 different such clusters. a.u., arbitrary units. Bars = 15 μm. (C) Distribution of ectopically expressed eGFP tagged EspH with respect to immunostained CD81. HeLa cells expressing EspH-eGFP or EspH_Δ130–168-eGFP were immunostained with the anti-CD81 5A6 antibody and imaged by confocal microscopy. The representative confocal image shows EspH and CD81 located at the plasma membrane (arrows) or at a central aggregate (arrowheads). The fluorescence level of eGFP and immunostained CD81 positioned in the central CD81 aggregate of EspH-eGFP- and EspH_Δ130–168-eGFP-expressing cells was quantified by a method similar to the quantification method applied for estimating the clustering of CD81 and F-actin at infection sites (Fig. 1). The results are the mean ± SE for 25 measurements. P values refer to the comparison with EspH-eGFP transfected cells. ***, P < 0.0005; ns, nonsignificant statistical difference (P > 0.05). Bar = 10 μm.

EspH expression selectively inhibits TNF-α-induced Erk phosphorylation.

Finally, we asked whether EspH targets Erk by inhibiting its upstream regulators, e.g., MEK and c-Raf. Data in Fig. S5 show that host infection with EPEC ΔespH+EspH diminished phosphorylated MEK (pMEK) and phosphorylated c-Raf (pc-Raf) levels, suggesting that EspH suppresses the entire MEK–c-Raf–Erk cascade. As epidermal growth factor (EGF), tumor necrosis factor alpha (TNF-α), and interleukin-1β (IL-1β) are well-known regulators of this pathway, we examined the capacity of EspH to suppress the pErk evoked by each of these ligands. The results showed that only TNF-α-mediated stimulation of pErk was inhibited by EspH expression (Fig. 7A), suggesting that the protein effector targets only a subset of the receptor-stimulated MAPK signaling pathways. Unlike EspH-eGFP, ectopically expressed EspH_Δ130–168-eGFP did not effectively inhibit TNF-α-induced pErk (Fig. 7B). Next, we repeated these experiments by infecting HeLa cells with bacteria expressing the wild type or the EspH_Δ130–168 mutant in the presence or absence of TNF-α. In these experiments, cells were infected for 3 h with EPEC which did not undergo preactivation. TNF-α treatment slightly stimulated pErk. However, translocated wild-type EspH had markedly decreased pErk levels in both TNF-α-treated and untreated cells, while translocated EspH_Δ130–168 had no effect (Fig. 7C). These results suggest that the C-terminal 38-amino-acid segment of EspH can inhibit pErk in TNF-α-stimulated and infected cells. Interestingly, in the absence of CD81, the efficiency with which EspH-eGFP reduced the TNF-α-induced pErk level was even greater (Fig. 7D). Taken together, these results highlight the capacity of EspH to selectively inhibit the TNF-α-dependent activation of the MAPK signaling pathways and the fact that CD81 regulates this pathway by counteracting this effect.

FIG 7.

EspH suppresses TNF-α-induced pErk levels, and CD81 counteracts this effect. (A) Ectopic expression of EspH-eGFP specifically suppresses the TNF-α-induced pErk. HeLa cells were either transfected with the indicated eGFP-encoding constructs or remained untransfected. Cells were then exposed (+) or not exposed (No Ligand) to EGF, TNF-α, or IL-1β ligands, as described in Materials and Methods. Cells were lysed, and pErk levels were detected by quantitative Western blotting, as described in the legend to Fig. 4 and Materials and Methods. The expressed eGFPs were detected using anti-GFP antibodies. The results are the mean ± SE from three experiments. P values refer to the comparison with untransfected cells. (B) Ectopically expressed mutant EspH with a 38-amino-acid C-terminal truncation fails to suppress the TNF-α-evoked pErk. HeLa cells were transfected with plasmids expressing eGFP, EspH-eGFP, or EspH-eGFP_Δ130–168. The cells were then challenged with TNF-α, lysed, and subjected to determination of changes in pErk levels by quantitative Western blotting. The results are the mean ± SE from three experiments. P values refer to the comparison with eGFP-transfected cells. (C) Translocated EspH with a 38-amino-acid C-terminal truncation fails to suppress TNF-α-evoked pErk in infected cells. HeLa cells were infected with EPEC ΔespH, EPEC ΔespH+EspH, or EPEC ΔespH+EspH_Δ130–168, which were not preactivated, for 3 h at 37°C. Thereafter, bacterium-containing medium was replaced with fresh TNF-α (25 ng/ml)-containing DMEM (+TNF-α) for one set of cells, while for other cells, the bacterium-containing medium was replaced with just plain DMEM (−TNF-α). The cells were then incubated for a further 30 min at 37°C and processed for Western blotting analyses. The results are the mean ± SE from 3 experiments. P values refer to the comparison with untreated (−TNF-α) EPEC-ΔespH-infected cells. (D) CD81 counteracts the ability of EspH to suppress TNF-α-stimulated pErk. CRISPR control, CD81KO, and CD81KO–mEmerald-CD81 cells were transfected with the indicated eGFP-encoding constructs, and the relative changes in pErk levels were determined as described in the legend to panel B. The results are the mean ± SE from 3 experiments. P values refer to the comparison with eGFP-transfected cells. ***, P < 0.0005; **, P < 0.005; *, P < 0.05; ns, nonsignificant statistical difference (P > 0.05).

DISCUSSION

Previous studies reported the ability of EPEC to manipulate the MAPK/Erk pathway. Some studies suggested that EPEC activates the pathway (27, 31), while others showed an inhibitory effect (28–30). We believe that these discrepancies can be attributed to differences in the infection conditions, as EPEC has been shown to induce a transient activation of Erk (e.g., see reference 31). Previous reports have highlighted that EspH is a significant modulator of the actin cytoskeleton, likely due to its ability to interact and inactivate Rho GTPases (11, 13, 14, 41). Here we report another EspH function, namely, suppression of the Erk signaling pathway (Fig. 4B; see also Fig. S5 in the supplemental material). The observation that ectopic expression of EspH reduces pErk signaling (Fig. 5C) suggests that the effector is sufficient and that no other bacterial cues are needed to perform this function. The inhibitory effect was also shown to depend on the presence of the C-terminal 38 amino acids of EspH (Fig. 5B and C), suggesting that the C terminus of EspH holds the capacity to inhibit pErk, possibly by interacting with proteins that inhibit the MAPK pathway. Interestingly, several studies have shown that Rho GTPases can function as Erk upstream modulators (42–45), suggesting that EspH may suppress Erk by inactivating these GTPases (13, 14). The observation that the C terminus of EspH plays a role in Rho inhibition (Fig. 5E) may concur with this view.

Little is known regarding the contribution of CD81 to bacterial pathogenesis. Studies have shown that CD81 is required for the invasion (46) and infection (47) of Listeria monocytogenes. In the case of EPEC, CD81 does not seem to be required for bacterial colonization of cultured epithelial cells (not shown) or for F-actin clustering at infection sites (Fig. 2). Our findings do suggest, however, that CD81 membrane domains are hijacked by EspH or are recruited to infection sites in response to translocated EspH (Fig. 3), to counteract the suppressive activity exerted by the effector on Erk signaling (Fig. 4C). Notably, previous studies have implicated CD81 as a positive regulator of Erk (26, 40, 48). Thus, CD81 may act here similarly by antagonizing the Erk suppression mediated by EspH.

The molecular mechanisms underlying the functional interplay between EspH and CD81 in controlling Erk remain unknown. However, our studies provide an initial insight into these mechanisms by showing that EspH was initially accommodated within CD81 clusters at infection sites, while at later infection times, EspH was located at the outer edges or outside the CD81 domains (Fig. 6). The redistribution of EspH correlated with the ability of the effector protein to exert Erk inhibition, as shown by the following: (i) after deletion of the C-terminal 38 amino acids in EspH (EspH_Δ130–168), EspH could neither escape the CD81 domain (Fig. 6) nor decrease pErk levels (Fig. 5B), and (ii) EspH-mediated inhibition of pErk was more potent in cells lacking CD81 (Fig. 4C). The latter observation clearly argues that CD81 is not needed for Erk inhibition by EspH but, rather, is needed for regulation of its activity at the different infection phases. For example, EspH mediates CD81 recruitment at an early infection time (Fig. 3B) to fine-tune and turn off its Erk inhibition activity. However, at the later infection phase, the effector protein manages to escape the CD81 domain and suppress Erk. As previously noted, EspH may inhibit Erk by inhibiting Rho GTPases acting upstream of Erk. CD81 membrane platforms have been suggested to interact with Rho GTPases and/or with their regulators (34, 49–51). An intriguing scenario would then be that at the initial stage, EspH interacts with Rho GTPases associated with CD81 microdomains. However, at a subsequent stage the protein effector has to leave these domains in order to exert its Rho/Erk inhibitory function. Interestingly, a similar principle of domain-dependent regulation of signaling has been reported for the small GTPase H-Ras. This protein is initially associated with lipid rafts, where it is uploaded with GTP. However, Raf activation and downstream signaling by Ras are achieved only after its redistribution from rafts to the bulk plasma membrane (52).

MAP Erk1/2 kinases are central in diverse host responses, including mitogenic responses to growth factors, cytokines, innate immunity (53–55), cell survival, and apoptosis (54, 56, 57). Regarding the responses to the last two processes, EspH has been reported to induce cell toxicity and death by a yet unknown mechanism (14). Thus, EspH may target the MAPK/Erk pathway to selectively modulate these processes. We so far found that EspH downregulates the pErk evoked by TNF-α but not that evoked by EGF or IL-1β (Fig. 7A), an activity that depends on the presence of the C-terminal 38 amino acids of EspH (Fig. 7B). Interestingly, previous studies suggested that TNF-α stimulated MAPK activation, a process that is mediated by the Rho family Rac/Cdc42 GTPases (58). EPEC has been shown to elicit the synthesis and release of TNF-α from infected epithelial cells (59). Here we show that translocated EspH inhibits pErk in TNF-α-treated cells (Fig. 7C), suggesting that EspH downregulates TNF-α-mediated activation of MAPK signaling by targeting Rac and CDC42, an hypothesis that merits further investigation. Nonetheless, CD81 expression appeared to counteract this effect (Fig. 7D), suggesting again that CD81 acts as a positive regulator of pErk, preventing EspH from suppressing MAPK. TNF-α mediates many important functions in the regulation of intestinal epithelial homeostasis and innate immunity (60). EspH possibly attenuates some of these processes through Erk inhibition, to increase bacterial survival in the infected gut. Although the role of CD81 and EspH in EPEC infection in vivo has not been thoroughly investigated, the observations implicating EspH in promoting EHEC colonization (15) and findings suggesting that CD81 is expressed in the human gut (Fig. S6) provide initial in vivo supporting evidence for the possible involvement of these proteins in EPEC and EHEC infection.

In summary, our studies suggest a novel working model for EspH whereby the effector associates with CD81 membrane platforms in an early phase of the infection process, from which it exits to suppress pErk in other domains of the plasma membrane. This activity could be essential for inhibiting innate immunity and/or promoting host cytotoxicity in order to facilitate the microbes' survival, colonization, and dissemination in the infected gut (Fig. 8). Further research is required to validate this hypothesis.

FIG 8.

Schematic of the working model for EspH-mediated modulation of pErk controlled by its spatial distribution with respect to CD81 microdomains. At an early infection phase, EspH facilitates CD81 clustering at the plasma membrane infection site. At this stage, CD81 may act as a positive Erk regulator which antagonizes the ability of translocated EspH to diminish pErk and, thereby, the MAPK-Erk signaling pathway. As Erk signaling has been implicated to regulate innate immunity and host cell survival, the inability of EspH to inhibit Erk may allow innate immunity and host survival processes aimed at exterminating the infecting pathogen. At a later stage, EspH is segregated to the edges and external regions of these CD81 clusters, where it prompts pErk inhibition. Suppression of this signaling pathway could be important for inhibiting innate immunity and other functions attributed to EspH, such as the induction of cell toxicity and death, to allow pathogen colonization and survival in its host.

MATERIALS AND METHODS

Bacterial strains, antibodies, plasmids and reagents.

The bacterial strains, antibodies, and plasmids used in this study are listed in Tables S1 to S3 in the supplemental material, respectively. Plasmids pAA6271, pAA6284, pAA6712, and pAA6290 (Table S3) were constructed using the Gibson assembly protocol with the primers indicated in Table S4. EPEC mutant strains were generated using the bacteriophage lambda Red recombinase system (61). Bacteria were grown in Luria-Bertani medium supplemented with appropriate antibiotics (ampicillin [100 μg/ml], kanamycin [50 μg/ml], streptomycin [50 μg/ml], nalidixic acid [30 μg/ml], chloramphenicol [20 μg/ml]).

Cells.

HeLa and Caco-2BBe Tet-off cells were cultured and maintained as described previously (62, 63). Notably, Caco-2 cells were seeded on collagen (rat tail collagen, 75 mg/ml in acetic acid)-coated 12-mm-diameter Transwell filter inserts with a 0.4-μm membrane pore size (Corning, Acton, MA) under conditions that allowed them to form a tight and polarized epithelial cell monolayer (63). This cell line, which was originally derived from a human colon carcinoma, exhibits well differentiated cell-cell junctions and brush border characteristics reminiscent of those found in enterocytes lining the small intestine (64, 65). We used CRISPR-Cas9-mediated gene editing to stably knock out the CD81 gene (CD81KO) in HeLa cells. To achieve this, HeLa cells were transfected with the pSpCas9 (BB)-2A-GFP (PX458) plasmid into which small guide RNA (GCTGGCTGGAGGCGTGATCC) was subcloned, using a TransIT-X2 dynamic delivery system (MIR 6004; Mirus, Madison, WI). Cells transfected with the backbone vector alone served as CRISPR controls. After 40 h, green fluorescent protein (GFP)-positive cells were isolated by flow cytometry and seeded in 12-well plates. When they reached full confluence, the cells were stained in suspension using allophycocyanin–anti-human CD81 antibody and sorted for CD81-negative cells by flow cytometry. Cells were expanded, and several clones were isolated by the end-dilution method. The lack of expression of CD81 in these cell clones was confirmed by Western blotting and is exemplified in two cell clones presented in Fig. S1. For rescuing CD81 expression, CD81KO cells were transfected with a plasmid expressing mEmerald-CD81, and CD81-expressing cells were isolated by flow cytometry and expanded under antibiotic selection. The protein expression level was confirmed by Western blotting (Fig. S4A). Like native CD81, mEmerald-CD81 localized mainly to the cell surface, and it also clustered at EPEC infection sites (Fig. S4B).

Bacterial growth, preactivation, and cell infection.

Bacterial growth and preactivation of their T3SS were performed as described previously (63). Unless otherwise mentioned, cells were infected with preactivated bacteria (multiplicity of infection, ∼100) at 37°C in 5% CO₂. In cells infected with EPEC ΔespH+EspH or EPEC ΔespH+EspH_Δ130–168, protein expression was induced by adding 0.05 mM isopropyl-β-d-thiogalactopyranoside (IPTG) (catalog number V395D; Promega, Madison, WI) for the last 30 min of activation, or the infection phase.

Fluorescence microscopy and clustering analysis.

Immunofluorescence labeling of fixed cells, F-actin labeling with Texas Red (TR)-phalloidin, and labeling of bacteria and cell nuclei with DAPI were performed as previously described (23, 63). The primary and fluorescently labeled secondary antibodies used are listed in Table S2. Cells were mounted and visualized using an Olympus FV-1200 laser scanning confocal microscope equipped with a 60× (numerical aperture, 1.42) oil immersion objective. Confocal sections were acquired at z-axis intervals of 0.5 μm. The clustering levels of fluorescently labeled CD81 and F-actin were analyzed using the Fiji distribution of ImageJ software (66). A maximal-intensity projection was generated for each stack. The regions of cell-associated EPEC microcolonies were manually defined using the polygon selections tool of ImageJ software and were termed “infection sites.” The average pixel intensity (PI) at the infection site (PI_i) was measured and expressed as the percent difference from the average PI for an identically defined uninfected area located near the infecting microcolony (PI_u), as follows: percent difference = [(PI_i − PI_u)/PI_u] × 100.

The line intensity profiles of the mEmerald-CD81 emission and the emission of the translocated EspH immunostained with anti-SBP antibodies were plotted as follows: a line was drawn across an adhered EPEC microcolony, and the intensity of the two different channels was measured using the plug-in Intensity Profile, as shown in Fig. 6. The data presented are the average for ∼20 such intensity profiles measured for 20 different CD81-EspH clusters. For analyzing the F-actin and CD81 distribution across the polarized Caco-2 cell monolayer, we used Imaris ×64 (version 7.6.5) software to generate an x-z image. Mean fluorescence intensity plots were created using the plug-in Plot Z-Axis Profile of Fiji distribution of ImageJ software.

SDS-PAGE and Western blotting.

For CD81 analysis, HeLa cells were cultured on a 6-well plate (∼70% confluence), washed 3 times with phosphate-buffered saline (PBS), and lysed in sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) sample buffer lacking reducing reagents (40% glycerol, 12% SDS, 0.2 M Tris-HCl, pH 6.8, supplemented with bromophenol blue). Lysates were heated (100°C, 10 min) and rigorously vortex shaken (1 min, 22°C). Proteins of each sample (∼20 μl) were analyzed by SDS-PAGE (40 mA, 30 min; Bio-Rad mini-Protean Tetra system) and transferred to nitrocellulose membranes (2.5 A, 10 min; Bio-Rad Trans-Blot Turbo). Membranes were then washed with Tris-buffered saline–Tween 20 (15 mM NaCl, 1 mM Tris-HCl. pH 8.0, 1% Tween 20), blocked with bovine serum albumin (2.5%, wt/vol) and skim milk (1%, wt/vol), and probed with the 5A6 primary antibody, followed by an appropriate peroxidase-conjugated secondary antibody. The analysis of phosphorylated MAPK (pMAPK) and total MAPK (tMAPK) was done similarly, except that cells were lysed in sample buffer containing dithiothreitol (DTT; 100 mM), washed, blocked, and probed with anti-pErk, pMEK, or phosphorylated c-Raf (pc-Raf) antibodies. Antibodies were then stripped off by submerging the membranes in stripping buffer (62.5 mM Tris-HCl, pH 6.8, 100 mM 2-mercaptoethanol, 2% SDS), followed by incubation for 20 min at 55°C with gentle agitation. The membranes were then washed and reprobed with anti-total Erk (anti-tErk), total MEK (tMEK), or total c-Raf (tc-Raf) antibodies. Imaging was performed using a Fusion FX spectra imager (Vilber Smart Imaging, Collègien, France), and the band intensity was determined with the Fiji distribution of ImageJ software (NIH). In all experiments, the pMAPK level was normalized to the tMAPK levels, and the ratio obtained was normalized again to a control whose pMAPK/tMAPK was defined as 1.

Effector translocation assay.

The effector translocation assay was performed as described previously (11). Briefly, HeLa cells cultured in a 6-well plate (∼70% confluence) were infected for 90 min at 37°C with EPEC ΔespH+EspH or EPEC escV+EspH. Cells were then lysed in ice-cold NP-40-containing buffer (100 mM NaCl, 1 mM EDTA, 10 mM Tris-HCl, pH 7.4, 0.5% [vol/vol] NP-40) supplemented with protease and phosphatase inhibitors. The lysates were centrifuged (16,000 × g, 4°C, 10 min), and an aliquot of the detergent-soluble supernatant fraction (containing host cytoplasm and membranes containing the translocated protein effector) and the insoluble fraction (containing mainly bacteria expressing the protein effector) were analyzed by SDS-PAGE followed by Western blotting. EspH was detected with anti-SBP antibodies.

Ligand stimulation assay.

HeLa cells seeded on a 10-cm plate were transfected with an eGFP-, eGFP-EspH-, or eGFP-EspH_Δ130–168-expressing plasmid using linear polyethylenimine (PEI; 1 mg/ml; catalog number 23966; Polysciences Inc., Warrington, PA) following the manufacturer's protocol. At 20 to 24 h after transfection, cells were incubated with human epidermal growth factor (EGF; 10 ng/ml; catalog number PHG0314; Gibco), human tumor necrosis factor alpha (TNF-α; 25 ng/ml; catalog number 210-TA; R&D Systems), or interleukin 1β (IL-1β/IL-1F2; catalog number 201-LB; R&D Systems) at 37°C for 30 min. The cells were then washed with PBS, and the lysates were prepared for Western blot analysis as described above.

Human intestinal biopsy specimens and their analysis for CD81 expression.

Four biopsy specimens from a healthy transverse colon and the terminal ileums were obtained from colonoscopy procedures performed on young individuals (13 to 14 years old). Each biopsy specimen (approximately 3 g each) typically contained the mucosa, submucosa, and muscularis mucosa layers (67). All samples were macro- and microscopically normal. Tissue samples were homogenized in ice-cold radioimmunoprecipitation assay buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM EDTA, 1% NP-40, 0.1% SDS, 0.5% sodium deoxycholate, protease and phosphatase inhibitors). The samples were then centrifuged (16,000 × g, 10 min, 4°C), and protein supernatants (1.5 μg) were analyzed by SDS-PAGE followed by Western blotting. CD81 was detected using the 5A6 monoclonal antibody.

Ethics statement.

The Shaare Zedek Medical Center Institutional Review Board approved this study (number 74/13).

Luciferase assay.

Rho activity was measured by a luciferase-based assay. HeLa cells were seeded in a 96-well plate and cotransfected with a luciferase reporter plasmid (pGL4.34[luc2P/SRF-RE/Hygro]; catalog number E1350; Promega, Madison, WI) and the appropriate eGFP-expressing plasmid (eGFP, eGFP-EspH, or eGFP-EspH_Δ130–168) using linear PEI as the transfection reagent following the manufacturer's protocol. After 3 h of transfection, cells were serum starved (0.5% serum) for 12 h and then induced with an induction medium (40% fetal calf serum in Dulbecco modified Eagle medium [DMEM]) or control medium (plain DMEM) for 6 h. Luciferase activity was determined using a One-Glo luciferase assay system (catalog number E6110; Promega, Madison, WI) under both conditions. Each experiment was performed in duplicate and repeated 6 times. The amount of cells before and after induction was similar (∼3 × 10⁵ cells/well). The Rho inhibitor I (C3 transferase from Clostridium botulinum; catalog number CT04; Cytoskeleton, Inc.) was used to confirm the validity of the assay in each experiment. Luciferase activity was calculated as follows: luciferase activity = average number of relative light units of induced cells/average number of relative light units of control cells.

Statistical analysis.

Results are presented as means ± standard errors (SE). Statistical significance was determined by two-tailed Student's t test. A P value of <0.05 indicates a statistically significant difference.