EPEC effector EspF promotes Crumbs3 endocytosis and disrupts epithelial cell polarity

Summary

Enteropathogenic Escherichia coli (EPEC) uses a type three secretion system to inject effector proteins into host intestinal epithelial cells causing diarrhea. EPEC infection redistributes basolateral proteins β1-integrin and Na⁺/K⁺ ATPase to the apical membrane of host cells. The Crumbs (Crb) polarity complex (Crb3/Pals1/Patj) is essential for epithelial cell polarization and tight junction (TJ) assembly. Here we demonstrate that EPEC displaces Crb3 and Pals1 from the apical membrane to the cytoplasm of cultured intestinal epithelial cells and colonocytes of infected mice. In vitro studies show that EspF, but not Map alters Crb3, while both effectors modulate Pals1. EspF perturbs polarity formation in cyst morphogenesis assays and induces endocytosis and apical redistribution of Na⁺/K⁺ ATPase. EspF binds to sorting nexin 9 (SNX9) causing membrane remodeling in host cells. Infection with ΔespF/pespFD3, a mutant strain that ablates EspF binding to SNX9, or inhibition of dynamin attenuates Crb3 endocytosis caused by EPEC. In addition, infection with ΔespF/pespFD3 has no impact on Na⁺/K⁺ ATPase endocytosis. These data support the hypothesis that EPEC perturbs apical-basal polarity in an EspF-dependent manner, which would contribute to EPEC-associated diarrhea by disruption of TJ and altering the crucial positioning of membrane transporters involved in the absorption of ions and solutes.

Introduction

Enteropathogenic Escherichia coli (EPEC) causes diarrhea in infants in developing countries. EPEC infection occurs in several steps: i) attachment to host cells, ii) delivery of bacterial effector proteins through a type III secretion system into host cells and iii) formation of actin pedestals and attaching/effacing lesions (A/E) with loss of microvilli (Lai, Rosenshine, Leong, & Frankel, 2013). The dramatic loss of the absorptive microvilli contributes to loss of water and altered intestinal transport of electrolytes and solutes leading to diarrhea, dehydration and sometimes death (Nataro & Kaper, 1998). Although the mechanisms by which EPEC induces diarrhea are not completely understood, it is known that EPEC effectors EspF, Map, Tir, and Intimin cooperate to induce a loss of the activity of the sodium-D-glucose co-transporter (SGLT-1), a major water pump in the small intestine (Dean et al., 2006). Map and NleH1 bind to Na⁺/H⁺ exchanger regulatory factors I and II (NHERF1/2), and Map regulates NHERF1 activity (Simpson et al., 2006; Martinez et al., 2010). EPEC also inhibits the activity of the Cl⁻/HCO₃⁻ exchanger, SLC26A3 (Down Regulated in Adenoma, DRA), and this process is mediated by EspG1/G2 effectors (Gill et al., 2007; Gujral et al., 2015).

In addition to altered ion transport, EPEC also disrupts the architecture and barrier function of intestinal epithelial tight junctions (TJ) (McNamara et al., 2001; Muza-Moons, Schneeberger, & Hecht, 2004) although the mechanisms have yet to be clearly elucidated. TJ form the intercellular barrier between epithelial cells and control paracellular permeability by regulating the flux of water and solutes across the epithelium. Select EPEC effectors EspF, Map, NleA, and EspG have been studied with respect to their specific effects on TJ, regulation of the size-selective paracellular permeability and TJ restoration in epithelial cells (McNamara et al., 2001; Dean & Kenny, 2004; Muza-Moons, Schneeberger, & Hecht, 2004; Singh & Aijaz, 2015; Matsuzawa, Kuwae, & Abe, 2005; Tomson et al., 2005; Thanabalasuriar et al., 2010; Glotfelty et al., 2014). EspF is a multifunctional molecule that interacts with several host proteins. The N-terminal of EspF contains the mitochondrial targeting signal (MTS) and nucleolar targeting domain (NTD) (Holmes, Muhlen, Roe, & Dean, 2010). EspF contains three proline-rich repeats (PRR), which bind to the SH3 domain of SNX9 (Marches et al., 2006; Alto et al., 2007). The interaction of EspF with SNX9 promotes the formation of elongated plasma membrane tubules (Weflen, Alto, Viswanathan, & Hecht, 2010; Alto et al., 2007) as well as the invasion of intestinal epithelial cells by EPEC (Weflen, Alto, Viswanathan, & Hecht, 2010). SNX9 localizes and binds to dynamin at the plasma membrane mediating endocytosis and cytoskeletal organization (Lundmark & Carlsson, 2003); this complex is recruited to clathrin-enriched areas before scission of the clathrin-coated vesicles (Soulet, Yarar, Leonard, & Schmid, 2005).

TJ contribute to the maintenance of apical-basal polarity by restricting the intermixing of apical and lateral plasma membrane components. Apical-basal polarity is crucial for cell morphology, directional vesicle transportation, ion and solute transport, and specific localization of proteins and lipids to specific membrane domains (Tepass 2012; Rodriguez-Boulan & Macara, 2014). Three polarity complexes are involved in the establishment and maintenance of epithelial cell apical-basal polarity. The Crumbs complex consists of Crumbs (Crb), protein associated with Lin-7 (Pals1), and Pals1-associated tight junction protein (Patj). Par3/Par6/aPKC/cdc42 form the Par complex and the third complex is comprised of Scribble (Scrib)/Lethal giant larvae (Lgl)/Disc large (Dlg) (Rodriguez-Boulan & Macara, 2014). Polarity proteins localize to the apical membrane and have been implicated in TJ formation and polarization. The Crb family of proteins possesses a single spanning transmembrane with a highly conserved cytoplasmic tail (Crb-CT) which contains a protein 4.1/Ezrin/Radixin/Moesin (FERM)-binding motif (FBM), a PDZ-binding motif (PBM), and a potential aPKC phosphorylation site (Makarova et al., 2003; Klebes & Knust, 2000; Klose, Flores-Benitez, Riedel, & Knust, 2013; Sotillos et al., 2004; Laprise et al., 2006). Overexpression of Crb3 disrupts polarity in MDCKII cysts and delays TJ assembly (Roh, Fan, Liu, & Margolis, 2003). MCF10A cells, which possess limited endogenous expression of Crb3, fail to assemble TJ; however, exogenous expression of Crb3 is sufficient for the development of functional TJ in these cells (Fogg, Liu, & Margolis, 2005; Tilston-Lunel et al., 2016). Mutation in either the PDZ or FERM sequence of Crb3 or expression of a dominant-negative chimeric protein impairs the development of apical polarity and TJ (Roh, Fan, Liu, & Margolis, 2003; Fogg, Liu, & Margolis, 2005).

Despite the large amount of data regarding the effects of EPEC on TJ, the impact of these effectors on epithelial apical-basal polarity has not been investigated. The basolateral proteins β1-integrin and Na⁺/K⁺ ATPase are also found on the apical membrane of EPEC-infected cells (Muza-Moons, Koutsouris, & Hecht, 2003; Muza-Moons, Schneeberger, & Hecht, 2004). In addition, aberrant TJ strands are present throughout the lateral membrane of EPEC-infected cells (Muza-Moons, Schneeberger, & Hecht, 2004). Similar to EPEC infection, Helicobacter pylori disrupts the organization and assembly of apical junctions in polarized epithelial cells. Cag-A, an effector protein of H. pylori, localizes to sites of TJ formation and associates with ZO-1 (Amieva et al., 2003; Bagnoli et al., 2005). It also perturbs apical-basal polarity and cell adhesion by interacting with and inhibiting Par1/MARK and PRK2 kinases activity, essential enzymes for apical-basal polarity and cell adhesion, respectively (Saadat et al., 2007; Zeaiter et al., 2008; Mishra et al., 2015) The hypothesis of this study is that EPEC modulates the integrity of the Crb complex thus disrupting epithelial polarity.

Results

EPEC infection promotes cytoplasmic accumulation of Crb polarity complex proteins

To determine the effect of EPEC on the polarity complexes of intestinal cells, SKCO-15 monolayers were plated on Transwell inserts and infected on the apical side for 2h. EPEC induced the movement of Crb3, and to a lesser extent Pals1, from the membrane to the cytoplasm (Figure 1A, arrows). In contrast, Patj and E-cadherin remain unchanged (Figure 1A). Fluorescence intensity indicates that EPEC significantly reduced the level of membrane-associated Crb3 and Pals1 (−60.0±5.0% and −43.0±8.0%, respectively) and increased their cytoplasmic accumulation (+100.0±16.0% and +131.0±17.0%, respectively) compared to uninfected monolayers (Figure 1B). Similar experiments were conducted using MDCKII cells, which were infected with EPEC for 1–4h and immunolocalization and western blot analysis were performed. Both Crb3 and Pals1 were redistributed following EPEC infection as was seen in SKCO-15 monolayers although the internalization of Pals1 was more pronounced in MDCKII cells (Supplemental Figure S1A). The level of protein expression was unchanged (Supplemental Figure S1B). These results indicate that EPEC promotes the redistribution of Crb3 and Pals1 to the cytoplasm without altering the level of protein expression.

Figure 1.

EPEC induces internalization of Crb polarity proteins Crb3 and Pals1, but not Patj in intestinal epithelial cells. SKCO-15 cells were plated on Transwells and infected on the apical side, or not (UI), with EPEC for 2h. Immunolocalization of endogenous Crb3/Pals1/Patj and E-cad was performed. (A and B) Representative confocal images and quantification of the fluorescence intensity are shown. Arrows indicate the loss of membrane-associated Crb3 and Pals1 induced by EPEC. Hoechst was used to mark the nuclei (blue) in all the images. Scale bar, 10 μm. Data represent the mean ± SEM (n=3); ***P < 0.001 values were calculated using ANOVA Tukey’s Multiple Comparison Test.

EPEC infection alters TJ structure, attenuates barrier function and increases paracellular permeability in vivo. The translocation of occludin and ZO-1 from cell-cell contacts to the cytoplasm of intestinal epithelial cells correlates with this functional phenotype (Shifflett et al., 2005; Zhang et al., 2010; Zhang et al., 2012). We questioned whether EPEC affects the localization of polarity proteins in vivo. Uninfected control mice show a well-delineated pattern of Crb3 at the apical membrane of intestinal epithelial cells (Figure 2A). In contrast, Crb3 in EPEC infected mice is more diffuse even when present in the membrane (Figure 2A). Similarly, Pals1 shifted from the apical membrane to the cytoplasm in EPEC-infected mice (Figure 2A). In concordance with our in vitro data from SKCO-15 and MDCKII cells (Figure 1A and Supplemental Figure S1A), Patj localization remained unchanged during EPEC infection in mice (Figure 2A). Fluorescence intensity of murine colonic tissues confirms that EPEC significantly reduced the level of membrane-associated Crb3 and Pals1 (−36.0±4.0% and −38.0±3.0%, respectively) and increased cytoplasmic accumulation (+82.0±5.0% and +106.0±6.0%, respectively) compared to uninfected mice (Figure 2B). In order to study the integrity of Crb complex using a murine pathogen similar to EPEC, mice infected with Citrobacter rodentium were also examined. Our data show that C. rodentium significantly reduced the amount of Crb3 in the apical membrane and increased its cytoplasmic accumulation (−36.0±4.0% and +186.0±3.0%, respectively) in a manner similar to that seen with EPEC infection (Figure 2C and D). Interestingly, the amount of Patj in the membrane slightly diminished during C. rodentium infection (−19.0±6.0%) (Figure 2C and D). These results indicate that EPEC and C. rodentium infection redistribute Crb polarity members from cell-cell contacts to the cytoplasm of murine colonic epithelial cells.

Figure 2.

EPEC and C. rodentium redistribute Crb complex proteins from the plasma membrane to the cytoplasm of murine colonocytes. Mice were infected with EPEC or C. rodentium strain by oral gavage, sacrificed on day 3 or day 10 post-infection respectively; intestinal tissues were processed for immunofluorescence. (A–D) Representative confocal images of Crb3/Pals1/Patj and quantification of the fluorescence intensity of the colonic tissues are shown. Scale bar, 40 μm. Data represent the mean ± SEM (n=3); ***P < 0.001 values were calculated using ANOVA Tukey’s Multiple Comparison Test.

Mislocalization of Crb3 from the membrane is EspF-dependent

To identify the EPEC effectors responsible for disruption of the Crb complex, in vitro models were used. SKCO-15 monolayers were infected with EPEC strains harboring mutations in effectors (Δmap and ΔespF) known to contribute to TJ disruption (McNamara et al., 2001; Dean & Kenny, 2004; Viswanathan et al., 2004a; Shifflett et al., 2005). Deletion of espF, but not map, prevented the mislocalization of Crb3 (Figure 3A). Complementation of espF (ΔespF/pespF) restored the wild-type EPEC phenotype of Crb3 internalization (Figure 3A). Deletion of either espF or map attenuated the redistribution of Pals1 (Figure 3B), while complemented strains (ΔespF/pespF or Δmap/pmap) reverted this phenotype to that of wild-type EPEC (Figure 3B). Patj and E-cadherin, as expected, were not affected by either mutant strain (Figure 3C). Total protein expression of Crb complex proteins remained unaltered following infection regardless of the infecting strain (Supplemental Figure 2A). However, fluorescence intensity demonstrated that EPEC, Δmap and the complemented strains (ΔespF/pespF or Δmap/pmap) significantly increased the cytoplasmic accumulation of Crb3 and reduced its membrane localization; both phenotypes are protected by ΔespF (Figure 3D). Although ΔespF and Δmap attenuated the membrane disruption of Pals1 caused by EPEC, these mutant strains do not significantly alter cytoplasmic accumulation (Figure 3E). The fluorescence intensity of Patj remained unchanged following infection by wild-type EPEC or mutant strains (Figure 3F).

Figure 3.

Deletion of espF protects against Crb3 mislocalization while deletion of either map or espF attenuates the internalization of Pals1. SKCO-15 cells were plated on Transwells and infected apically with wild-type EPEC, ΔespF, Δmap, or complemented strains (ΔespF/pespF and Δmap/pmap) for 2h. Cells were immunostained for Crb3/Pals1/Patj and E-cad was used to label the lateral membrane. (A–F) Representative confocal images and quantification of the fluorescence intensity of membrane, cytoplasm, and total protein are shown. Arrows show the absence of membrane-associated Crb3 and Pals1 in infected monolayers. Scale bar, 10 μm. Data represent the mean ± SEM (n=3); ***P < 0.001 values were calculated using ANOVA Tukey’s Multiple Comparison Test.

To determine if EspF has a direct role in Crb3 localization, pRetroX-Tight-Pur-EspF-HA or pRetroX-Tight-Pur-Map-HA was cloned into a Tet-inducible plasmid (Retro-X Tet-On Advanced) allowing for controlled expression of EspF-HA or Map-HA. Crb3 was localized as expected at the cell membrane of control cells (−dox), whereas induction of EspF (+dox), but not Map, disrupted Crb3 membrane localization and increased cytoplasmic accumulation (Figure 4A). In addition, transient expression of GFP-EspF displaced Crb3 from the membrane (Figure 4B). Consistent with our previous observations with EPEC infection, Patj remains at cell-cell contacts in monolayers expressing EspF (Figure 4B). Expression of HA or EspF was determined by immunodetection (Supplemental Figure S2B and C, respectively). These data suggest that EspF specifically promotes the membrane disruption of Crb3.

Figure 4.

Ectopic expression of EspF, but not Map, displaces Crb3 from the membrane. A) espF-HA or map-HA was cloned into the doxycycline-inducible pRetroX-Tight-Pur vector and transfected into SKCO-15 cells. Tet-On SKCO-15 cells were plated in absence of doxycycline (−dox) and transgene expression was induced with doxycycline (+dox) for 3 days. B) SKCO-15 cells were transiently transfected or not (control), with GFP-vector or GFP-EspF. Ectopic expression of EspF or Map was evaluated by immunodetection of HA (Green) or GFP (A and B, respectively). The impact of EspF or Map on localization of Crb3 and Patj was evaluated. Scale bar, 10 μm.

EPEC drives endocytosis of Crb3

The endocytosis of Crb3 is regulated by Rab5 (Lu & Bilder, 2005), an early endosome marker. To analyze whether EPEC promotes the endocytosis of Crb3 via the Rab5 pathway, SKCO-15 cells were infected with wild-type EPEC, ΔespF, or complemented espF (ΔespF/pespF) (Figure 5A). EPEC infection significantly increased the association of Crb3 with Rab5 in both intracellular and cell membrane-associated vesicles (Figure 5A and B). Deletion of espF reduced the association of Crb3 and Rab5 in both cytoplasmic and membrane-associated vesicles similar to uninfected cells, while complementation of espF (ΔespF/pespF) restored the wild-type EPEC phenotype (Figure 5A and B). To provide insight into the mechanism by which EspF disrupts Crb3 membrane localization, SKCO-15 monolayers were treated with inhibitors that block different endocytic pathways, then infected with EPEC for 2h. Inhibition of lipid rafts with MβCD or autophagic vacuoles with MDC failed to prevent Crb3 endocytosis induced by EPEC and showed a pattern similar to cells treated with DMSO alone (Supplemental Figure S3). Treatment of cells with dynasore, an inhibitor of dynamin GTPase activity, ablated the membrane disruption and attenuated the cytoplasmic accumulation of Crb3 caused by EPEC (Figure 5C), suggesting that EspF-induced clathrin-mediated endocytosis is involved. These data support the conclusion that EPEC EspF increases internalization of Crb3 via the clathrin pathway.

Figure 5.

Depletion of espF or treatment with dynasore inhibits EPEC-induced cytoplasmic internalization of Crb3. (A–B) SKCO-15 cells were infected with wild-type EPEC, ΔespF or complemented strain (ΔespF/pespF) for 4h and processed for immunodetection of Crb3 and Rab5. (A and B) Representative confocal images and quantification of the area of the co-localization vesicles Crb3/Rab5 are shown. Data represent the mean ± SEM (n=3); ***P < 0.001 values were calculated using ANOVA Tukey’s Multiple Comparison Test. C) SKCO-15 cells were plated in Transwells and treated with DMSO or Dyn, then cells were infected or not (UI) with EPEC for 2h. Samples were processed for immunodetection of Crb3. Scale bar, 10 μm.

The interaction of EspF with SNX9 is critical for Crb3 endocytosis

EspF has several functions in host cells including formation of a complex with 14-3-3ζ and cytokeratin 18 (CK-18) and interaction with the SH3 domain of SNX9 through its RxAPxxP motif (Viswanathan et al., 2004b; Marches et al., 2006; Alto et al., 2007). To determine the contribution of these interactions to Crb3 endocytosis, SKCO-15 cells were incubated for 2h with wild-type EPEC, ΔespF, or ΔespF complemented to express specific site-directed EspF mutations (ΔespF/pSer47A, ΔespF/pSer47/50A and ΔespF/pespFD3). Infection with ΔespF/pSer47A and ΔespF/pSer47/50A, which prevent 14-3-3ζ binding, induced the redistribution of Crb3 from the membrane to the cytoplasm in a similar manner to that seen in cells infected with wild-type EPEC (Figure 6A), indicating that EspF/14-3-3ζ interactions are not involved in Crb3 endocytosis. In contrast, cells infected with ΔespF/pespFD3, a mutant that prevents the interaction of EspF with SNX9, blocked the mislocalization of Crb3 (Figure 6A). Fluorescence intensity shows that EPEC, ΔespF/pSer47A and ΔespF/pSer47/50A strains, diminished the level of membrane-associated Crb3 and increased its cytoplasmic accumulation to the same degree as wild-type EPEC (Figure 6B). In contrast, infection with ΔespF/pespFD3 afforded the same level of protection against Crb3 redistribution as the ΔespF mutant strain (Figure 6B). These data suggest that binding of EspF to SNX9 is crucial for promoting the endocytosis of Crb3.

Figure 6.

The interaction of EspF with SNX9 is crucial for Crb3 endocytosis. SKCO-15 cells were infected with wild-type EPEC, ΔespF, or ΔespF complemented to express specific site-directed EspF mutations (ΔespF/pSer47A, ΔespF/pSer47/50A and ΔespF/pespFD3) for 2h. (A and B) Representative images of Crb3 localization and quantification of the fluorescence intensity are shown. Only infection with ΔespF and ΔespF/pespF-D3, which cannot bind SNX9, preserved Crb3 localization. Scale bar, 10 μm. Data represent the mean ± SEM (n=3); ***P < 0.001 values were calculated using ANOVA Tukey’s Multiple Comparison Test.

EspF alters apical-basal polarity

Polarity proteins are the main regulators of cyst formation. Since the data presented above demonstrate that EspF perturbs the membrane localization of Crb3, the impact of EspF on cyst formation in 3D cultures of MDCKII cells was investigated. Cells were transfected to express various GFP-EspF constructs and cyst formation was assessed. As expected, cells transfected with GFP-vector alone developed cysts containing a single lumen exhibiting cortical actin, the TJ markers ZO-1 and occludin at apical cell-cell contacts, as well as E-cadherin basolaterally (Figure 7A and Supplemental Figure S4, A and B), indicating no defect in cell polarity. In contrast, cells expressing GFP-EspF showed evidence of perturbed polarity (Figure 7A and Supplemental Figure S4, A and B) with the presence of ZO-1 and occludin along the lateral membrane (Figure 7A and Supplemental Figure S4A, arrows). In addition, cells transfected with GFP-EspF formed a significantly higher number of multi-lumen cysts and fewer single lumen cysts compared to cells expressing GFP-vector alone (Figure 7B). Expression of EspF in 2D MDCKII cultures induced a redistribution of Crb3 without changing its expression levels (Supplemental Figure S4, C and D).

Figure 7.

EspF prevents cyst formation. MDCKII cells were transfected with GFP-vector, GFP-EspF, GFP-EspF-L16E or GFP-EspF-D3, and after 24h placed as a single-cell suspension in Matrigel; cyst formation occurred after 2–3 days. Cysts were fixed and phalloidin (red) and ZO-1 (magenta) were used to label the lumen and the TJ region, respectively. Visualization of GFP was used to determine cysts containing transfected cells. A) Representative confocal images of cysts. EspF and EspF-L16E displaced ZO-1 from tight junctions to the lateral membrane (A, arrows). In contrast, cyst morphogenesis and ZO-1 localization in the presence of GFP-EspF-D3 expression was not different from that seen with GFP-vector. Scale bar, 10 μm. B) Twenty cysts from each group harboring transfected cells were analyzed for the number of lumens formed: single-lumen, 2–3 lumens and multi-lumen. Expression of GFP-EspF and EspF-L16E increased the number of multi-lumen (>3) cysts compared to GFP-vector. Data represent the mean ± SEM (n=3); *P < 0.05 values were calculated using Student t-test analysis.

The N-terminal sequence of EspF targets mitochondria and promotes apoptosis (Nougayrede & Donnenberg, 2004). Substitution of leucine at position 16 to glutamic acid (L16E) within the mitochondria targeting sequence prevents EspF localization to the mitochondria and cell death (Nougayrede & Donnenberg, 2004; Nagai, Abe, & Sasakawa, 2005). To investigate whether the effect of EspF on polarity disruption is due to increased cell death, GFP-EspF-L16E was expressed in MDCKII cells and cyst formation was analyzed. Similar to cells expressing wild-type EspF, cells expressing GFP-EspF-L16E formed significantly more multi-lumen cysts compared to those transfected with GFP-vector and displayed a redistribution of ZO-1 and occludin (Figure 7, A and B and Supplemental Figure S4A, arrows). In addition, the formation of single lumen cysts was reduced by GFP-EspF-L16E expression compared to GFP-vector (Figure 7B). Expression of GFP-EspF-L16E also mislocalized Crb3 from the cell-cell contacts without altering its protein expression in a similar manner to that of wild-type EspF (Supplemental Figure S4, C and D). To verify that EspF-L16E expression reduced apoptosis caused by wild-type EspF, cleaved caspase-3 was examined. GFP-EspF-L16E diminished the number of cells positive for cleaved caspase-3 compared to GFP-EspF expression (Supplemental Figure S5). These results indicate that the interference of EspF with the formation of apical-basal polarity is independent of its apoptotic function.

Since infection with ΔespF/pespFD3 mutant protects the mislocalization of Crb3 (Figure 6A), we questioned the impact of EspF/SNX9 interaction on cyst formation and polarity. Interestingly, cells expressing GFP-EspF-D3 formed normal single lumen cysts with the apical marker actin outlining a single central lumen, ZO-1 and occludin restricted to the TJ, and E-cadherin distributed to the basolateral membrane in a similar manner to cells transfected with GFP-vector (Figure 7A and Supplemental Figure S4, A and B). No significant differences were observed in cyst formation between cells transfected with GFP-vector and those expressing GFP-EspF-D3 (Figure 7B). Expression of GFP-EspF-D3 did not alter the redistribution of Crb3 from the plasma membrane or the level of protein expression (Supplemental Figure S4, C and D). These results indicate that EspF via its interaction with SNX9 impairs the development of apical-basal polarity in epithelial cells.

EPEC infection induces the redistribution of the basolateral proteins β1-integrin and Na⁺/K⁺ ATPase to the apical membrane of T-84 intestinal cells (Muza-Moons, Koutsouris, & Hecht, 2003; Muza-Moons, Schneeberger, & Hecht, 2004). To determine the impact of EPEC on the apical-basal polarity of intestinal epithelial cells, SKCO-15 were plated on Transwells, infected with wild-type EPEC or EspF mutant strains (ΔespF, ΔespF/pespFD3, ΔespF/pespF), and the localization of the basolateral protein Na⁺/K⁺ ATPase was determined. EPEC induced the endocytosis of Na⁺/K⁺ ATPase after 2h infection and re-distributed it from the basolateral membrane to the apical domain by 3h post-infection (Figure 8). Infection of monolayers with ΔespF or ΔespF/pespFD3 ablated the apical redistribution of Na⁺/K⁺ ATPase (Figure 8). In contrast, complementation of ΔespF with wild-type espF (ΔespF/pespF) restored the wild-type phenotype of Na⁺/K⁺ ATPase mislocalization to the apical domain (Figure 8). Together these results indicate that EPEC EspF plays a major role in the perturbation of intestinal epithelial apical-basal polarity.

Figure 8.

EPEC induces the internalization of the basolateral protein Na⁺/K⁺ ATPase and its redistribution to the apical membrane in an EspF-dependent manner. SKCO-15 cells were infected wild-type EPEC, ΔespF, ΔespF/pespFD3 or complemented strain (ΔespF/pespF) for 2–3h, and monolayers immunostained for Na⁺/K⁺ ATPase. Figure shows representative confocal images of three independent experiments. Scale bar, 10 μm.

Discussion

EPEC infection disrupts intestinal epithelial cell architecture and TJ complexes leading to altered barrier and fence function (Simonovic, Rosenberg, Koutsouris, & Hecht, 2000; McNamara et al., 2001; Dean & Kenny, 2004; Shifflett et al., 2005; Guttman et al., 2006; Tomson et al., 2005; Matsuzawa, Kuwae, & Abe, 2005; Thanabalasuriar et al., 2010; Singh & Aijaz, 2015). EPEC also modulates intestinal epithelial cell electrolyte transport, reducing the expression of sodium hydrogen exchanger 3 (NHE3), the major intestinal transporter for Na⁺ absorption (Gawenis et al., 2002; Hecht et al., 2004; Hodges et al., 2008). Furthermore, EPEC alters the activity and surface expression of the intestinal Cl⁻/HCO₃⁻ exchanger, SLC26A3/DRA, resulting in a reduced Cl⁻ uptake and its accumulation in the lumen, driving water loss (Gill et al., 2007; Gujral et al., 2015). EPEC effectors EspF, Map, Tir, and Intimin, act together to inactivate SGLT-1, a cotransporter responsible for 70% of the total fluid uptake by the small intestine (Dean et al., 2006; Meinild et al., 1998). Although the effect of EPEC on TJ barrier function has been widely investigated, few studies have focused on the impact on TJ fence function and maintenance of apical-basal polarity. EPEC infection redistributes the basolateral proteins β1-integrin and Na⁺/K⁺ ATPase to the apical surface (Muza-Moons, Koutsouris, & Hecht, 2003) and induces aberrant TJ strands throughout the lateral membrane (Muza-Moons, Schneeberger, & Hecht, 2004), indicating a loss in apical-basal polarity. Other pathogens have been reported to induce the loss of polarity. For example, the H. pylori effector, CagA, interacts with and inhibits Par1 kinase activity causing junctional and polarity defects (Saadat et al., 2007). Neisseria meningitidis recruits the Par3/Par6/aPKC complex beneath colonies, thus disorganizing cell-cell junctions and perturbing the endothelial barrier (Coureuil et al., 2009). This is the first study to demonstrate that EPEC perturbs apical-basal polarity by targeting Crb polarity complex.

In polarized epithelial cells, the Crb complex located at the apical membrane is important for the development and maintenance of apical-basal cell polarity (Roman-Fernandez & Bryant, 2016). Crb3 is regulated by multiple mechanisms, including endocytosis, recycling, degradation, and interaction with cytoplasmic partners. For example, expression of Snail, down-regulation of either Patj/Cdc42/Lgl/Adaptor Protein 2 (AP-2), or mutant aPKC with reduced kinase activity, all destabilize Crb inducing its internalization from the membrane into sub-apical vesicles causing polarity defects (Harder et al., 2012; Michel et al., 2005; Harris & Tepass, 2008; Kim et al., 2009; Fletcher et al., 2012; de Vreede et al., 2014; Whiteman, Liu, Fearon, & Margolis, 2008; Lin et al., 2015). Our findings show that internalized Crb3 induced by EPEC infection is recruited to the endosomal pathway and that this phenotype is dependent on the interaction of EspF and SNX9. EspF regulates actin cytoskeletal dynamics through its binding and activation of neuronal Wiskott-Aldrich syndrome protein (N-WASP) and dynamically colocalizes to clathrin-coated pits (CCPs) (Alto et al., 2007). SNX9 associates with N-WASP and dynamin, and together with Arp2/3 and associated proteins, these complexes are recruited to CCPs promoting the clathrin-mediated endocytosis (CME) of certain plasma membrane receptors (Badour et al., 2007; Merrifield, Qualmann, Kessels, & Almers, 2004; Benesch et al., 2005; Yarar, Waterman-Storer, & Schmid, 2007; Lundmark & Carlsson, 2003; Soulet, Yarar, Leonard, & Schmid, 2005; Shin et al., 2007). AP-2, also involved in CME, binds to and controls Crb3 levels at the plasma membrane by regulating its internalization (Lin et al., 2015). Our data demonstrate that inhibition of either dynamin GTPase activity or EspF/SNX9 interaction (EspF-D3 mutant) blocks EPEC-induced Crb3 endocytosis suggesting that the EspF/SNX9 complex and possibly other binding partners, as NWASP/Arp2/3/dynamin/clathrin, promote Crb3 endocytosis.

Little is known about the endocytosis of the other Crb polarity complex proteins, Pals1 and Patj. Amot is an adaptor protein that binds both Patj and Pals1 (Sugihara-Mizuno et al., 2007; Wells et al., 2006) and promotes their internalization to endosomes and induces the loss of TJ integrity (Wells et al., 2006; Heller et al., 2010; Campbell et al., 2016). Amot is found in Rab11 positive endosomes but not with Rab5 or Rab7 in MDCK cells, indicating that Amot has an important role in protein recycling (Heller et al., 2010). Our data show that Pals1, but not Patj, moves to the cytoplasm following EPEC infection and that this phenotype is diminished by deletion of espF or map. The finding that EspF, but not Map, alters Crb3 while both effectors modulate Pals1 suggests that the internalization of Crb3 and Pals1 is differentially regulated. The maintenance and regulation of apical-basal polarity depends largely on the interaction of Crb3 with Pals1 and ezrin (Makarova et al., 2003; Roh, Fan, Liu, & Margolis, 2003; Li et al., 2014; Whiteman et al., 2014). Pals1 binds to and regulates ezrin localization at the apical membrane of gastric parietal cells (Cao et al., 2005). EPEC activates ezrin, thus contributing to disruption of intestinal barrier function (Simonovic et al., 2001). In addition, reduced expression of ezrin decreases transepithelial electrical resistance (TER) and augments permeability of MCF10A cells overexpressing Crb3b (Tilston-Lunel et al., 2016). EPEC induces microvillus effacement of the intestinal epithelium of infected human and mice (Savkovic et al., 2005). Our in vivo data correlate with our in vitro findings, demonstrating that EPEC redistributes Crb3 and Pals1 from cell-cell contacts to the cytoplasm of colonocytes. Similar to EPEC, infection with C. rodentium also induces the redistribution of Crb3. Interestingly, membrane-associated Patj is slightly diminished in murine colonocytes infected with C. rodentium indicating that despite the similarities between EPEC and C. rodentium strains, there are some pathophysiological differences between these two organisms. The difference in this phenotype could simply be the increase in attached C. rodentium as compared to EPEC. Interestingly, the intestines of Crb3- or ezrin-null mice show similar defects to those caused by EPEC, intestinal villus fusion and disrupted microvilli (Whiteman et al., 2014; Charrier, Loie, & Laprise, 2015; Saotome, Curto, & McClatchey, 2004; Casaletto, Saotome, Curto, & McClatchey, 2011). Although the mechanism by which EPEC displaces Crb3 and Pals1 from plasma membrane is unknown, it is possible that EPEC destabilizes the cytoplasmic interaction of Crb3 with Pals1/ezrin leading to a loss of cell polarity and barrier function.

The establishment of apical-basal polarity allows the asymmetric segregation of proteins and lipids into apical and basolateral domains and contributes to the development of lumens allowing vectorial transport, an essential process for the balance of ions, solutes and water along epithelia (Shivas, Morrison, Bilder, & Skop, 2010; Rodriguez-Boulan & Macara, 2014). EPEC infection is characterized by an increased loss of ions and water, and profound diarrhea. In the intestine, the assimilation of electrolytes, nutrients, and fluid occurs via Na-dependent cotransport processes (Rose & Valdes, 1994). The basolateral protein Na⁺/K⁺ ATPase provides for the export of Na⁺ across the basolateral membrane, thus controlling the osmotic balance and volume regulation of cells (Robinson & Flashner, 1979; Pavlov & Sokolov, 2000). The proper function of Na⁺/K⁺ ATPase is essential for efficient nutrient and ion absorption. Decreased Na⁺/K⁺ ATPase activity during acute and transient enteritis diminishes the reabsorption of sodium and water (Allgayer et al., 1988; Ejderhamn, Finkel, & Strandvik, 1989; Rachmilewitz, Karmeli, & Sharon, 1984). In an animal model of chronic ileal inflammation, decreased Cl⁻/HCO⁻₃ and stimulated Na⁺/K⁺ ATPase exchange in villus and crypts cells, respectively, has been shown, suggesting that chronic inflammation alters electrolyte transport (Sundaram & West, 1997). Na-glucose uptake, Na-alanine, Na-glutamine and Na-taurocholate cotransport are also down regulated during inflammation due to diminished expression and activity of Na⁺/K⁺ ATPase (Sundaram, Wisel, Rajendren, & West, 1997; Sundaram, Wisel, & Fromkes, 1998; Saha, Arthur, Kekuda, & Sundaram, 2012; Coon, Kekuda, Saha, & Sundaram, 2010). Campylobacter jejuni, an intestinal pathogen, decreases Na⁺/K⁺ ATPase activity in the rat ileum, causing impaired electrolyte absorption (Kanwar et al., 1994). This underscores the importance of apical-basal polarity maintenance and supports the conclusion that the perturbation of Na⁺/K⁺ ATPase activity contributes to EPEC pathophysiology.

Although the mechanisms are not well defined, several studies have focused on the importance of Na⁺/K⁺ ATPase endocytosis. Na⁺/K⁺ ATPase interacts with Src leading to the subsequent activation of PI3K, Ras/Raf/ERKs, PLC/PKC and ERK1/2 pathways (Liang et al., 2006; Tian et al., 2006). PI3K activation, as well as treatment with ouabain, a cardiac glycoside that inhibits Na⁺/K⁺ ATPase, facilitates the binding of Na⁺/K⁺ ATPase to AP-2, recruiting Na⁺/K⁺ ATPase into CCPs and triggering its localization to both early and late endosomes (Yudowski et al., 2000; Liu et al., 2004). Our findings show that EPEC induces Na⁺/K⁺ ATPase endocytosis in an EspF-dependent manner. Interestingly, the ablation of EspF/SNX9 interaction inhibits Na⁺/K⁺ ATPase redistribution to the apical membrane indicating that EspF/SNX9 interaction plays an important role in maintaining apical-basal polarity.

In summary, this study shows that EPEC induces the redistribution of Crb3 and Pals1 in an EspF-dependent manner (Figure 9). The data presented herein support the conclusion that binding of EspF to SNX9 is crucial for both preventing the formation of polarity (cyst multi-lumen structures) and disrupting established epithelial polarity (movement of basolateral Na⁺/K⁺ ATPase to the apical domain). We therefore speculate that EspF targeting and disruption of the Crb polarity complex may initiate crucial signaling processes that ultimately impact host epithelial physiology and integrity contributing to EPEC pathogenesis.

Figure 9.

Model depicting the effect of EPEC on apical-basal polarity. Polarized epithelial cells consist of the apical membrane facing the lumen and the basolateral domain contacting the underlying basement membrane. The apical polarity complex Crb (Crb3/Pals1/Patj) localizes at the TJ (Control). i) During EPEC infection, the injected EPEC effectors, including EspF and Map, induce the endocytosis of polarity proteins Crb3 and Pals1 and the basolateral protein Na⁺/K⁺ ATPase. ii) Increased endocytosis of Crb3, Pals1 and likely other polarity proteins leads to a loss in the apical-basal polarity, as demonstrated by the redistribution of Na⁺/K⁺ ATPase to the apical membrane.

Experimental procedures

Tissue culture

SKCO-15 and MDCKII epithelial cells were grown in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum, 100 U/I penicillin and 100 U/I streptomycin at 5% CO₂.

Antibodies and Reagents

Antibodies used include (species-antigen): rat-Crb3 (ab180835, Abcam), rabbit-Patj (ab102113, Abcam), rabbit-Pals1 (07-708, Millipore), rabbit-ZO-1 (61-7300, Zymed), rabbit-EspF (Hecht Lab), rabbit-cleaved caspase-3 (9661S, Cell Signaling), rabbit-actin (A2066, Sigma), rabbit-HA (C29F4, Cell Signaling), rabbit-GFP (A11122, Life Technologies), mouse-Rab5 (sc-46692, Santa Cruz), mouse-E-cad (610181, BD), mouse-occludin (33-1500, ThermoFisher Scientific), mouse-Na⁺/K⁺ ATPase (610992, BD), Phalloidin Alexa Fluor 568 (Life Technologies). Secondary antibodies used for immunofluorescence were Alexa fluor (Life Technologies). Drugs used were from Sigma, 10μM methyl-beta-cyclodextrin (MβCD), 400 μM monodansyl cadaverin (MDC) and 80μM dynasore (Dyn).

Bacteria culture

The following EPEC strains were used: wild-type EPEC 0127:H6 E2348/69, ΔespF (McNamara & Donnenberg, 1998), ΔespF/pespF, Δmap (Kenny & Jepson, 2000), Δmap/pmap, ΔespF/pespFD3 (Alto et al., 2007), ΔespF/pSer47A and ΔespF/pSer47/50A (Hecht unpublished). Bacterial strains were grown at 37°C in Luria broth overnight and with addition of appropriate selective antibiotics. For infections, overnight bacterial cultures were diluted in serum- and antibiotic-free T84 1:1 (vol/vol) mixture of low glucose Dulbecco’s modified Eagle’s medium (DMEM) and Ham’s F-12 (Gibco) medium containing 0.5% mannose and grown to mid-log growth phase. Monolayers were infected with EPEC at a MOI of 50 for the indicated times.

Murine infection

For murine infection, six to eight-week-old male C57BL/6J mice were used (Jackson Laboratory Bar Harbor, ME, USA) and housed in a specific pathogen-free facility at Loyola University Chicago, Maywood, IL for 7–14 days with free access to food and water. Mice were infected with EPEC by oral gavage as previously described (Rhee et al., 2011) and sacrificed on day 3 post-infection. All animal protocols were approved by LU Animal Care and Use Committee. Mouse intestines were removed, cleaned, processed as “Swiss rolls”, and fixed in 10% phosphate-buffered formalin for 24h. Fixed tissues were embedded in paraffin, cut into 5 μm sections, and processed for immunofluorescence. Paraffin embedded sections of intestine from C. rodentium infected mice were obtained from the laboratory of Dr. Katherine Knight (Department of Microbiology; Loyola University School of Medicine). Infection and tissue preparation were performed as previously described (Jones & Knight, 2012).

DNA plasmids

For transient transfection, the following plasmids were used: pGFP (vector alone), pEGFP-EspF and pEGFP-EspF-D3 (Alto et al., 2007). The pEGFP-C1-EspF-L16E plasmid was constructed using QuikChange site-directed mutagenesis kit according to manufacturer’s protocol (Agilent Technologies) using pEGFP-EspF as template and the primer pair (5′-cgacttgcgatacctacttcctgccgccctagtgtag-3′ and 5′-ctacactagggcggcaggaagtaggtatcgcaagtcg-3′). For inducible cell lines, EspF-HA and Map-HA were generated by amplifying EPEC genomic DNA using the following primers: EspF-FOR (5′-cttgaagcggccgcatgcttaatggaattagtaacgc-3′ and EspF-REV-HA (5′-aatcaggaattcttaagcgtaatctggaacatcgtatgggtaccctttcttcgattgctc-3′); Map-FOR (5′-cttgaagcggccgcatgtttagtccaacggcaatgg-3′) and Map-REV-HA (5′-aatcaggaattcctaagcgtaatctggaacatcgtatgggtacagccgagtatcctgcacatt-3′) and separately cloned into the NotI/EcoRI sites of the retroviral plasmid pRetroX-Tight-Pur (Clontech), then co-transfected with pRetroX-Tet- On-Advance Inducible Expression System (632104, Clontech) in SKCO-15 cells.

Western blotting

For protein isolation, cells were washed twice with ice-cold phosphate buffered saline (PBS), and lysed in 0.25 ml of radioimmunoprecipitation assay (RIPA) buffer (150 mM NaCl, 40 mM Tris-HCl, pH 7.5, 2 mM EDTA, 10% glycerol, 1% Triton X-100, 0.5% sodium deoxycholate and 0.2% SDS) and protease inhibitor cocktail (P8340, Sigma). Proteins were boiled in SDS sample buffer, and analyzed by SDS-PAGE. For immunoblotting, nitrocellulose or Immobilon^R-FL (IPFL00010, Millipore) membranes were incubated in blocking solution (Invitrogen) for 1h, and probed with primary antibodies overnight at 4°C. Membranes were washed three times for 10 min with TBS + 0.05% Tween and incubated for 1h with secondary antibodies horseradish peroxidase-conjugated (Sigma) or IRDye^R 800CW (915-32213, LI-COR). Chemiluminescent exposure was performed with ECL (GE Healthcare) or using LI-COR Odyssey Classic (Image Studio software).

Immunofluorescence

Cells were plated on Transwell inserts or glass coverslips and fixed with cold methanol at −20°C. Fixation of cysts in Matrigel (Corning) was performed using 3% PFA and permeabilized with 0.1% Triton X-100 in PBS. For mouse tissue samples, paraffin embedded tissues were deparaffinized in xylene, rehydrated with ethanol and H₂O washes, and antigen retrieval performed in Tris-EDTA plus Tween20 buffer (pH 9) (Crb3 and Pals1) and in sodium citrate plus Tween20 buffer (Patj). Samples were blocked and incubated with primary antibodies overnight at 4°C or 30°C. Tissue samples were blocked and incubated with primary and secondary antibodies in 5% normal goat serum in 1x PBS plus 0.1% Tween20 and 0.1% saponin. Samples were washed with PBS, and then processed for immunofluorescence using Alexa fluor conjugated goat secondary antibodies and counterstained with Hoechst to visualize nuclei. Slides were analyzed using a Leica DM 4000B microscope (MetaMorph software) or confocal Leica TCS SPE DMI 4000B (LAS X software) microscope. Images were processed using Adobe Photoshop and ImageJ software.

Tet-On system and transient transfections

To generate SKCO-15 Tet-On system, cells were transfected with p-RetroX-Tet-On Advanced and pRetroX-Tight-Pur-EspF-HA or pRetroX-Tight-Pur-Map-HA using the Xfect nanoparticles polymer kit (ST0153, Clontech) following manufacturer’s instructions. After transfection, cells were selected with G418 and puromycin for 2–3 weeks. Expression of pRetroX-Tight-Pur-EspF-HA or pRetroX-Tight-Pur-Map-HA was induced in presence of Doxycycline (500 ng/ml) for three days. For transient transfection, MDCKII and SKCO-15 cells were transfected with Lipofectamine 2000 (Invitrogen) or nanoparticles according to manufacturer instructions. Cells were processed for immunofluorescence 72–96h post-transfection.

Cyst formation assay

MDCKII cells were plated and transfected with GFP-vector, GFP-EspF, GFP-EspF-L16E or GFP-EspF-D3 DNA constructs using the Xfect nanoparticles polymer kit. The following day, cells were trypsinized and plated in Matrigel to allow cyst formation according to (Debnath, Muthuswamy, & Brugge, 2003). Immunostaining was performed 3–4 days post-transfection.

Abbreviations used

EPEC

Enteropathogenic Escherichia coli

TJ

tight junctions

Crb3

Crumbs3