|Home | About | Journals | Submit | Contact Us | Français|
Phosphatidylinositol 4,5-bisphosphate [PI(4,5)P2] and phosphatidylinositol 3,4,5-trisphosphate [PI(3,4,5)P3] are phosphoinositides (PIs) present in small amounts in the inner leaflet of the plasma membrane (PM) lipid bilayer of host target cells. They are thought to modulate the activity of proteins involved in enteropathogenic Escherichia coli (EPEC) infection. However, the role of PI(4,5)P2 and PI(3,4,5)P3 in EPEC pathogenesis remains obscure. Here we show that EPEC induces a transient PI(4,5)P2 accumulation at bacterial infection sites. Simultaneous actin accumulation, likely involved in the construction of the actin-rich pedestal, is also observed at these sites. Acute PI(4,5)P2 depletion partially diminishes EPEC adherence to the cell surface and actin pedestal formation. These findings are consistent with a bimodal role, whereby PI(4,5)P2 contributes to EPEC association with the cell surface and to the maximal induction of actin pedestals. Finally, we show that EPEC induces PI(3,4,5)P3 clustering at bacterial infection sites, in a translocated intimin receptor (Tir)-dependent manner. Tir phosphorylated on tyrosine 454, but not on tyrosine 474, forms complexes with an active phosphatidylinositol 3-kinase (PI3K), suggesting that PI3K recruited by Tir prompts the production of PI(3,4,5)P3 beneath EPEC attachment sites. The functional significance of this event may be related to the ability of EPEC to modulate cell death and innate immunity.
Enteropathogenic Escherichia coli (EPEC) is a major cause of a severe infantile diarrhea in developing countries. Studies performed on infected humans and animal models have shown that after ingestion, EPEC intimately adheres to the mucosal surface of the intestinal epithelium. Bacterial adhesion elicits a localized collapse of microvilli and a dramatic reorganization of the actin cytoskeleton, eventually leading to the establishment of a pedestal-like actin structure located underneath the adhering bacteria. These histopathological alterations, also termed attaching and effacing (A/E) lesions, are essential to promote successful EPEC colonization, but they also induce tissue damage and fluid loss, which may eventually lead to diarrhea.
A/E lesion formation requires a type III secretion system (T3SS) of EPEC that mediates the delivery of bacterial effector proteins directly into the host cell cytoplasm. Upon contact with the host cell, the T3SS translocates the intimin receptor, Tir, which is inserted into the host cell plasma membrane (PM), and interacts with intimin, a bacterial surface protein. Tir–intimin interaction leads to intimate attachment of the bacterium to the host cell surface and triggers signaling cascades that lead to polymerization of F-actin and pedestal formation. Clustering of Tir by intimin enhances the activity of cellular tyrosine kinases that phosphorylate two C-terminal tyrosines on the Tir molecule: tyrosine 474 (Y474) and tyrosine 454 (Y454) (Kenny, 1999 ; Campellone and Leong, 2005 ). This results in direct recruitment of the adaptor protein Nck, which in turn, recruits and activates the neural Wiskott-Aldrich syndrome protein (N-WASP) and the downstream actin-related protein (Arp) 2/3 complex (Gruenheid et al., 2001 ). The latter two modules are required for actin nucleation, polymerization, and pedestal formation (Lommel et al., 2001 ). However, Tir can also mediate the recruitment of other cytoskeletal proteins independent of Y474 phosphorylation, such as α-actinin, talin, vinculin, and cytokeratin 18 (Goosney et al., 2000 , 2001 ; Cantarelli et al., 2001 ; Itoh and Takenawa, 2002 ; Batchelor et al., 2004 ). Additional proteins have been reported to be recruited and required for pedestal formation (for a recent review, see Bhavsar et al., 2007 ). Although many proteins involved in EPEC adherence and in pedestal formation have been characterized, their mode of action, particularly in relation to their lipid environment, is poorly defined.
Phosphatidylinositol 4,5-bisphosphate [PI(4,5)P2], phosphatidylinositol 3,4,5-trisphosphate [PI(3,4,5)P3], and other inositol lipids are believed to act as signaling modules by maintaining diverse protein–lipid and protein–protein interactions (reviewed in (Lemmon, 2003 ; Balla, 2005 ; van Rheenen et al., 2005 ; Rusten and Stenmark, 2006 ). Thus, phosphoinositides (PIs) are implicated in the regulation of diverse cellular processes, including 1) anchoring signaling molecules at the cell PM (Heo et al., 2006 ); 2) restructuring the actin cytoskeleton (Raucher et al., 2000 ); 3) vesicle dynamics (De Matteis and Godi, 2004 ); and 4) cell infection by pathogens (Brumell and Grinstein, 2003 ; Pizarro-Cerda and Cossart, 2004 ).
Previous studies provided initial evidence suggesting that EPEC can subvert certain PIs during infection. First, several studies have shown that EPEC infection stimulates the production of inositol phosphates (Foubister et al., 1994 ), including inositol 3-phosphate (IP3; Dytoc et al., 1994 ; Ismaili et al., 1995 ; Guan et al., 2000 ), probably via intimin-dependent activation of phospholipase C-gamma (Kenny and Finlay, 1997 ). Second, a fraction of PI(4,5)P2 may be linked to cholesterol-sensitive lipid rafts (Laux et al., 2000 ; Rozelle et al., 2000 ; Kwik et al., 2003 ; Aoyagi et al., 2005 ; Meiri, 2005 ; van Rheenen et al., 2005 ), and rafts have been recently implicated in EPEC effector translocation and pedestal formation (Zobiack et al., 2002 ; Hayward et al., 2005 ; Riff et al., 2005 ; Allen-Vercoe et al., 2006 ). Third, it has been proposed that PI(4,5)P2 probes accumulate underneath EPEC microcolonies (Celli et al., 2001 ; Zobiack et al., 2002 ; Rescher et al., 2004 ), though the functional significance of this phenomenon in epithelial cell infection was not investigated. Fourth, EPEC inhibits its own phagocytosis by inhibiting a phosphatidylinositol 3-kinase (PI3K)-mediated pathway (Celli et al., 2001 ). However, these studies have utilized professional phagocytes as host cell targets for EPEC infection. PI3K recruitment and activation by EPEC-infected epithelial cells could occur by a different mechanism. Finally, PI(4,5)P2 has been suggested to bind and regulate the activity of numerous actin regulatory proteins, some of which associate with EPEC pathogenesis: e.g., annexin2 (Rescher et al., 2004 ; Gokhale et al., 2005 ), α-actinin (Michailidis et al., 2007 ), vinculin (Gilmore and Burridge, 1996 ), dynamin (Lin and Gilman, 1996 ; Zheng et al., 1996 ; Sever et al., 2000 ; Roux et al., 2006 ;), clathrin and associated adaptor proteins (Ford et al., 2001 ; Rohde et al., 2002 ; Krauss et al., 2003 ; Sun et al., 2007 ), and the N-WASP-Arp2/3 complex (Sechi and Wehland, 2000 ; Prehoda and Lim, 2002 ; Insall and Machesky, 2004 ). The interaction of PI(4,5)P2 with a large number of host cell proteins that associate with EPEC infection suggests that the PI is immensely involved in various steps of EPEC pathogenesis. Here, we show for the first time that PI(4,5)P2 is indeed required for optimal EPEC adherence to the host cell surface and for actin pedestal development. The possibility that EPEC subverts PI(3,4,5)P3 and the major kinase that induces its formation, i.e., the PI3K, was also explored in this work.
The cDNA constructs GFP-PH-PLCδ1, GFP-PH-PLCδ1-R40L, GFP-PH-FAPP1, GFP-PX-P40, GFP-PH-Akt, GFP-PH-Akt-R25C, mRFP-FKBP-5-ptase-dom, and PM-FRB-CFP were obtained from Tamas Balla (National Institute of Health, Bethesda, MD) and have been described previously (Varnai and Balla, 1998 ; Servant et al., 2000 ; Balla et al., 2005 ; Varnai et al., 2006 ). The constructs encoding myc-5-ptase and myc-5-ptase Δ1 were described (Ono et al., 2004 ). For constructing the pKB2690 plasmid, the mCherry gene was amplified from pRSET-BmCherry using the primers mCherry-F-EcoRI (AGGAATTCATGGTGAGCAAGGGCGAGG) and mCherry-R-SalI (TATCGTCGACTTACTTGTACAGCTCGTCC). The resulting fragment was digested and cloned into the EcoRI and SalI sites of pSA10. The lacI gene was deleted during construction, thus allowing constitutive expression of mCherry. Constructs encoding GFP-tagged phosphatidylinositol 4-phosphate 5-kinase [PI(4)P5-kinase] and PM-monomeric red fluorescent protein (mRFP) were provided by Dr. S. Grinstein. The latter consists of the N-terminal 11 amino acids of the Src-family kinase Lyn that is both myristoylated and doubly palmitoylated. These lipid moieties target the construct to the inner leaflet of the PM, mainly to lipid rafts.
Alexa Fluor 633 phalloidin, red rhodamine phalloidin, and Alexa Fluor 594–conjugated donkey anti-rabbit IgG were from Molecular Probes (Eugene, OR). 4′,6′-Diamidino-2-phenylindole (DAPI) and fluorescein isothiocyanate-dextran (FITC-dextran) of 4-kDa averaged molecular weight were from Sigma (St. Louis, MO). Plasmids encoding for the EPEC protein Tir and its various mutants were previously described (Campellone et al., 2002 ).
Rapamycin (rapa) was from EMD Chemicals (San Diego, CA). The PI3K specific inhibitor, LY294002, was purchased from Calbiochem and stored as 10 mM stock in DMSO. Rabbit α-EPEC (O:127) polyclonal antibodies were from the Israeli Ministry of Health. The 9E10 anti-myc tag monoclonal antibodies were kindly provided by Keith Mostov (UCSF).
Madin-Darby canine kidney cells (MDCK) were routinely cultured in minimal essential medium (MEM, Biological Industries, Beit Haemek, Israel), supplemented with 5% (vol/vol) fetal calf serum (FCS) and 1% (vol/vol) antibiotics (Biological Industries). Human embryonic kidney (HEK293) and African green monkey kidney fibroblast (COS-7) cells were grown in Dulbecco's modified Eagle's medium (DMEM, supplemented with 10%, vol/vol, FCS and 1%, vol/vol, antibiotics). Cells were maintained in a humidified incubator at 37°C and a 5% CO2 atmosphere.
Bacterial strains used in this study are listed in Supplemental Table S1.
A single colony of EPEC was picked from a Luria-Bertani (LB) agar plate and placed into a bacterial tube containing 3 ml LB media supplemented with the appropriate antibiotic solution (100 μg/ml ampicillin or 50 μg/ml kanamycin). Bacteria were cultured overnight at 37°C without shaking. The T3SS of EPEC was preactivated by diluting the overnight bacterial culture (1:50) in MEM and incubating it for 3 h in a humidified atmosphere of 5% CO2 at 37°C without shaking (Rosenshine et al., 1996 ; Kenny et al., 1997 ). Cells were infected at a multiplicity of infection (MOI) of 10 with preactivated bacteria supplemented with 5% FCS for 1 h at 37°C and 5% CO2. In some experiments, MDCK cells were treated with LY294002 (100 μM in DMSO) in growth medium for 1 h before infection and for an additional 1 h throughout the EPEC infection time.
MDCK cells were cultured on 12-mm filters (0.4-μm pore size, Transwell, Costar, Acton, MA). Growth medium was replaced in the next day. The experiment was preformed 3 d after culturing. On that day, bacteria were preactivated as described above, except for using phenol red–free MEM-Eagle medium supplemented with Hanks' salts (Biological Industries). Cells were washed three times with prewarmed phosphate-buffered saline (PBS) and then infected by placing 0.5 ml of the preactivated bacteria, supplemented with 5% FCS and 2 mg/ml 4-kDa FITC-dextran, into the apical chamber. Prewarmed plain MEM-Eagle/FCS, 1.5 ml, was added to the basolateral side. Transepithelial electrical resistance (TER) was measured using the Millicell-ERS system (Millipore, Bedford, MA) equipped with a silver/silver-chloride electrode. The value from a blank filter (with no cells, but otherwise treated identically) was subtracted from the resistance values of filters on which cells were plated. The fluorescence intensity of a 100-μl aliquot taken from the basolateral medium was measured in a BMG Galaxy Fluostar microplate reader (BMG Lab Instruments, Offenburg, Germany) equipped with a 485/538-nm excitation/emission filter pair. These measurements were regarded as “time zero.” Infected cells were then placed at 37°C, and similar measurements were conducted every hour throughout the 6-h experimental period.
MDCK or HEK293 cells were transfected using the ExGen 500 reagent (Fermentas, Glen Burnie, MD), according to the manufacturer's protocol. Expression was allowed for up to 18 h. COS-7 cells were transfected using the calcium phosphate precipitation protocol (Sambrook et al., 1989 ).
Indirect immunofluorescence was performed on paraformaldehyde-fixed cells, as previously described (Leyt et al., 2007 ). For F-actin visualization, cells were washed with PBS and permeabilized for 5 min with 0.1% Triton X-100 in PBS. Fluorescently labeled phalloidin (200 U/ml in MetOH) diluted 1:300 in PBS, pH 8.0, supplemented with 0.025% (wt/vol) saponin (Sigma) was added to the cells for 20 min. The reagent was removed by extensive washes with PBS. DNA/bacteria labeling was achieved by incubating fixed and permeabilized cells with DAPI (100 ng/ml) for 1 min at 22°C.
Confocal microscopy was performed using an FV-1000 confocal system (Olympus, Tokyo, Japan), based on an IX81 inverted microscope. A 60×/1.35 NA oil immersion objective was used. This system is equipped with an on-scope incubator (Life Imaging Services, Basel, Switzerland), which controls temperature and humidity. In addition, the IX81 is equipped with the ZDC (Zero Drift Controller) option, to maintain the same focus plane throughout the entire period of imaging. Confocal images were taken using the specific laser lines for excitation and the various emission filters listed in Supplemental Table S2. In all cases, data are representative of at least three independent experiments. Bar equals 10 μm.
MDCK cells were cultured to ~50% confluence on glass-bottom culture dishes (35-mm dish, 14-mm Microwell; MatTek, Ashland, MA) and transfected the next day with ExGen 500 as described above. Expression was allowed for 18 h, and then the cells were washed three times in PBS, and growth medium was added. The culture dish was then placed into the incubation chamber on the microscope stage, and the fields to be imaged were selected. Note that the expression of lipid-binding-domain probes may interfere with the turnover of these lipids by other enzymes that utilize these same domains. For this reason, we have cautiously observed only cells displaying moderate expression levels of the GFP fusion domains. Preactivated bacteria were then added to the medium, and XYZ image stacks were acquired every 30 s. The 3D stacks consisted of 5–10 sections (depending on the cell height). The Z-drive was incremented by 1 μm between sections. The confocal aperture was automatically set to 1 airy unit; therefore, the axial section width is estimated to be 0.5–1 μm.
Image analysis was performed using ImageJ ver. 1.38 (http://rsb.info.nih.gov/ij/), as follows. A circular region of interest (ROI) on the cell membrane, where an EPEC colony landed was selected, and the average GFP intensity in that ROI was calculated for each time point. To estimate the effect of photobleaching, a second ROI was defined in an area that was not occupied by an EPEC colony, and the average fluorescence intensity was measured in that ROI. The photobleaching was estimated by normalizing the intensity of the fluorescence at each time point according to the initial intensity. Then, the fluorescence intensity at each time point in the ROI containing the EPEC colony was corrected using the photobleaching factor for that time point, as found in the control ROI. Finally, the corrected intensity at each time point was normalized to the initial corrected intensity of the ROI (see Figure 3). Data are representative of three independent measurements, each of which was performed on a single cell infected by the EPEC microcolony.
The protocol for rapa-induced PM translocation of 5-ptase has been recently described (Varnai et al., 2006 ). Principally, this method relies on PM localization of the FK506-binding protein (FKBP) conjugated to 5-ptase domain and mRFP, as well as the rapa-binding fragment of mTOR (FRB) protein fused to cyan fluorescent protein (CFP). The mRFP-FKBP-5-ptase domain is translocated to the PM and heterodimerizes chemically with FRB by rapa. PM recruitment of the enzyme upon rapa treatment induces rapid dephosphorylation of PI(4,5)P2. COS-7 cells were triple transfected with CFP-FRB, mRFP-FKBP-5-ptase domain, and GFP-PH-PLCδ1 constructs. Cells were allowed to express the proteins for 18 h. Cell treatment with rapa for 30–60 s sufficed to generate significant effects by 5-ptase (Varnai et al., 2006 ). Generation of sizeable EPEC microcolonies requires ~1 h of cell exposure to bacterial infection. To avoid possible recovery of PI(4,5)P2 PM levels during the infection time, transfected cells were first pretreated with 100 nM rapa (diluted from 100 μM stock in DMSO) for 5 min, and subsequently infected with preactivated EPEC for 1 h at 37°C and 5% CO2 in the presence of rapa. Notably, rapa at the indicated concentration had no effect on the kinetics of EPEC growth in LB, as judged by monitoring the optical density at 600 nm (not shown). It is also worth noting that analyses were performed only on cells showing normal morphology.
MDCK cells (n = 1 × 107) were infected with EPEC strains expressing various hemagglutinin (HA)-tagged Tir proteins (Supplemental Table S1). After infection, the cells were washed with cold PBS and lysed for 30 min at 4°C in lysis buffer (20 mM Tris, pH 7.2, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 10% glycerol, 1 mM Na3VO4, COMPLETE inhibitor, Roche, Indianapolis, IN). Lysates were precleared with protein G-Sepharose (Amersham Pharmacia Biotech, Piscataway, NJ) for 2 h at 4°C. Next, 2 μg of mouse monoclonal α-HA antibody (NEB Cell Signaling, , Hitchin, Herts, UK) was added to the supernatants, which were incubated overnight at 4°C. Immune complexes were precipitated by the addition of protein G-Sepharose for 2 h and washed once with lysis buffer and three times with 0.5 ml PBS. Phosphorylated and nonphosphorylated Tir proteins were detected by incubating the membranes with mouse monoclonal α-phosphotyrosine antibody PY99 (Santa Cruz Biotechnology, Santa Cruz, CA) and the α-Tir antibody (Kenny, 1999 ). PI3K (p85) was identified using a rabbit polyclonal antibody (Santa Cruz Biotechnology). As a secondary antibody, horseradish peroxidase–conjugated anti-mouse or anti-rabbit polyvalent sheep immunoglobulin was used (Dako, Carpinteria, CA). Antibodies were detected with the ECL Plus chemiluminescence Western blot kit system for immunostaining (Amersham Pharmacia Biotech). Tir-dependent activation of Akt was determined by Western blots using the polyclonal rabbit anti-Akt-P-Ser-473 antibody and the polyclonal rabbit anti-Akt antibody (NEB Cell Signaling). Akt kinase activity in wild-type and mutant-infected cells was quantified using the luminescence image analyzer (Lumi-Imager F1, Roche).
We used green fluorescent protein (GFP)-tagged pleckstrin homology (PH) domains derived from phospholipase Cδ1 (GFP-PH-PLCδ) and Akt (GFP-PH-Akt; Balla and Varnai, 2002 ; Cozier et al., 2004 ; Balla, 2007 ), to monitor changes in PI levels and distribution in living cells. Initially, MDCK and HEK293 cells transfected with a plasmid expressing GFP-PH-PLCδ (a PI(4,5)P2 probe) were infected with wild-type EPEC and subsequently analyzed by fluorescence microscopy. As a negative control, we used an identical PI(4,5)P2 probe, except that the corresponding PH domain carried a R40L point mutation that rendered it deficient in PI(4,5)P2 binding (GFP-PH-PLCδ- R40L). The signal of the fluorescent probe indicated that GFP-PH-PLCδ, but not the mutant, was located primarily at the PM, having increased intensity in regions associated with bacterial adherence sites (Figure 1, A and B). Similar results were obtained with COS or HeLa cells (data not shown). These findings confirm previously reported EPEC-induced accumulation of GFP-PH-PLCδ (Celli et al., 2001 ; Zobiack et al., 2002 ; Rescher et al., 2004 ). In contrast to these findings, EPEC attachment did not stimulate the accumulation of GFP fused to PH–FAPP1 or PX-P40 (PI(4)P, and PI(3)P probes, respectively; Supplemental Figure S1). Collectively, these results suggest that upon infection, EPEC induces specific accumulation of PI(4,5)P2 at sites of its attachment to the cell surface.
We next transfected COS cells with GFP-PH-PLCδ–expressing plasmid and infected the cells with an EPEC mutant lacking bundle-forming pili (BFP; bfpA::TnphoA), which is deficient in microcolony formation. This mutant adheres to the cell surface as individual bacteria, forming extended F-actin pedestals (Figure 1C). The pedestal region of the BFP bacteria exhibited augmented GFP-PH-PLCδ labeling (Figure 1C), delineating the entire elongated F-actin–rich structure (Figure 1C, see inset). This result better defines the correlation between accumulated PI(4,5)P2 staining and the F-actin pedestal, suggesting the existence of a possible link between PI(4,5)P2 clustering and pedestal biogenesis.
We next tested the possibility that effectors transmitted by the T3SS mediate PI(4,5)P2 accumulation. MDCK cells transfected with GFP-PH-PLCδ were infected with the T3SS-defective EPEC-escV mutant, expressing the mCherry fluorescent protein (EPEC-escV::kan-cherry). Fluorescence images revealed GFP accumulation beneath the cell surface-attached EPEC-escV microcolonies (a representative example is shown in Figure 1D). This suggests that some T3SS-independent factors can cluster the fluorescent probe. To further analyze these observations, GFP-PH-PLCδ fluorescence levels under EPEC-wt and EPEC-escV microcolonies were quantified. Results in Figure 1E show that fluorescence levels associated with EPEC-wt are significantly greater than those measured for EPEC-escV, suggesting that the T3SS is actively involved in elevating the PI(4,5)P2 levels beneath EPEC-wt microcolonies. Importantly, GFP-PH-PLCδ labeling was not accumulated at all in the vicinity of the adhered E. coli K12 control strain (HB101), which ectopically expresses BFP from the plasmid pMAR7::Tn3 (Figure 1F). Thus, the factor that mediated the T3SS-independent accumulation of PI(4,5)P2 in response to EPEC-escV infection is EPEC specific, but distinct from the BFP.
Rescher and coworkers (Rescher et al., 2004 ) have shown that the type 1- PI(4)P5 kinase accumulates below wild-type EPEC microcolonies. These observations raised the possibility that the kinase is recruited to stimulate PI(4,5)P2 synthesis in EPEC infection sites. Thus, we extended this analysis and tested whether the enzyme also accumulates under microcolonies of EPEC-escV or Δtir mutants. To this end, MDCK cells transiently transfected with a plasmid encoding for GFP-tagged PI(4)P5 kinase were subsequently infected with EPEC expressing mCherry. The results show that the enzyme accumulates not only below wild-type EPEC, but also underneath the EPEC-escV (Figure 2) and Δtir (not shown) mutants. These results suggest that T3SS-dependent and independent factors can mediate local accumulation of kinase involved in PI(4,5)P2 synthesis.
After 1 h of infection, a fraction of EPEC-wt microcolonies did not show a prominent accumulation of the GFP-PH-PLCδ probe. Three scenarios, which are not mutually exclusive, can explain this phenomenon: 1) EPEC microcolonies induces PI(4,5)P2 clustering at diverse efficiencies; 2) The dynamics of PI accumulation elicited by each EPEC microcolony is different; and 3) PI accumulation is transient. To address these hypotheses, the dynamics of PI(4,5)P2 clustering at EPEC-cell contact sites was quantitatively determined, using confocal imaging of live cells. In these experiments, cells were allowed to coexpress GFP-PH-PLCδ and PM-mRFP, which label PI(4,5)P2 and the PM in general, respectively. Cells were then infected with wild-type EPEC or EPEC-escV mutant under the microscope, and images in the red and green channels were acquired simultaneously every 30 s. Cells and bacteria were also imaged by differential interference contrast (DIC) microscopy. EPEC microcolony landing was identified by the DIC images (see Movies S1 and S2). The fluorescence intensity level on the cell surface, in regions confined to the attached microcolony area, was determined quantitatively, as specified in Materials and Methods.
The results of a representative experiment shown in Figure 3A demonstrate a sharp increase in the PM-mRFP fluorescence intensity level, which occurred instantly upon bacterial microcolony attachment to the cell surface. The fluorescence intensity of that marker remained fairly stable throughout the remaining measurement periods. The accumulation of mRFP fluorescence could be contributed by EPEC-induced changes in PM architecture (e.g., by membrane microfolding) and/or by probe accumulation within a lipid domain generated in response to EPEC association. The fluorescence levels of GFP-PH-PLCδ also rose persistently after cell association of a wild-type EPEC microcolony (Figure 3A). Then, those levels remained rather steady for ~60 s (Figure 3A) or for longer periods of time (Supplemental Figure S2). Thereafter, the fluorescent signal declined for the remaining measurement time (Figure 3A and Supplemental Figure S2). These results indicate that immediately upon cell association, wild-type EPEC can prompt transient eruption of PI(4,5)P2 levels on the host cell surface, beneath its attachment site.
In contrast to wild-type EPEC, surface association of EPEC-escV resulted in a concomitant rise in GFP-PH-PLCδ and PM-mRFP fluorescence levels for ~150 s, which then reached plateau levels for the remaining measurement time (Figure 3B). These parallel responses of the fluorescent probes indicate that their accumulation is T3SS-independent, possibly contributed by PM architectural changes induced by the mere attachment of the bacterial microcolony to the PM. The above differences between wild type and EPEC-escV further stress the contribution of the T3SS to the EPEC-induced transient burst in the GFP-PH-PLCδ signal.
We next tested whether the dynamics of GFP-PH-PLCδ accumulation parallels in space and time with those of EPEC mediated accumulation of actin. This was done by measuring the time-dependent changes in GFP-actin and mRFP-PM fluorescence changes under EPEC infection sites. Data presented in Figure 3C (Movies S3 and S4) revealed significant accumulation of GFP-actin fluorescence underneath the EPEC-wt microcolony. The kinetic profile discloses a bell-shaped curve, which likely reflects the existence of transient actin accumulation events. Notably, after ~540 s, the fluorescence declined to steady-state levels that are higher than those recorded before EPEC attachment. These results suggest that complex actin-remodeling processes are elicited immediately upon EPEC attachment to the cell surface. Importantly, these alterations paralleled spatially and temporally with the observed variations in PI(4,5)P2 levels, suggesting a possible link between the two processes. In contrast to EPEC-wt, EPEC-escV caused a relatively minor accumulation of the actin probe, reaching plateau levels at ~150 s after infection (Figure 3C). The kinetic profile of actin accumulation was very similar to that observed for the PM probe, suggesting again that these changes could be contributed by T3SS-independent processes.
To test the significance of the PI in EPEC infection, we examined the effects of PI(4,5)P2 depletion on bacterial adherence to the cell surface (mediated by T3SS-independent and -dependent processes) and on the efficacy of F-actin pedestal biogenesis (a process stimulated by the T3SS).
To this end, we used a specific strategy whereby PI(4,5)P2 was rapidly degraded by the rapamicin (rapa)-mediated heterodimerization, consequently affecting recruitment to the PM, of the PI(4,5)P2 5-phosphatase (5-ptase) catalytic domain (5-ptase dom) with the membrane-targeted rapa-binding domain of mTOR (FRB; Varnai et al., 2006 ). For this purpose, COS-7 cells triply transfected with mRFP-FKBP-5-ptase-dom, PM-FRB-CFP, and GFP-PH-PLCδ were subjected to EPEC infection in the presence (+) or absence (−) of rapa. After fixation and cell permeabilization, the actin cytoskeleton was labeled with Alexa Fluor 633–tagged phalloidin. The four fluorescently labeled components were coimaged along with EPEC (DIC microscopy). In the absence of rapa, GFP-PH-PLCδ fluorescence could be readily visualized at cell edges, colocalizing with cortical F-actin staining, indicating PM localization of the probe (Supplemental Figure S3, a–d, indicated with arrowheads). Addition of rapa resulted in relocation of the fusion protein from the PM to the cytosol and the nucleus, confirming PI(4,5)P2 hydrolysis (Supplemental Figure S3, e–h). The GFP-PH-PLCδ did not detach from the PM in rapa-treated cells expressing the mRFP-FKBP-5-ptase domain alone (not shown). Although EPEC microcolony adherence to the cell surface and the development of pedestals in FKBP-5-ptase-dom/PM-FRB expressing cells could be readily observed (Figure 4A and Supplemental Figure S3, e–h, indicated with arrows), a fraction of the settled bacteria did not develop noticeable pedestals (Figure 4A, indicated with an arrowhead). The effects were quantified, and the results are presented in Figure 4, B and C.
In the absence of rapa (−rapa), the entire FKBP-5-ptase-dom/PM-FRB–expressing cell population was colonized with at least a single EPEC microcolony, whereas in rapa-treated cells (+rapa), 25 ± 7% of the expressing cells did not exhibit even a single adhered EPEC microcolony (Figure 4B). However, all transfected cells that did not express the FKBP-5-ptase and FRB proteins, yet were treated with rapa (not expressing), exhibited at least one adhered EPEC microcolony on their cell surface (Figure 4B).
Compared with cells that were not exposed to rapa treatment, the addition of rapa increased the fraction of EPEC that did not exhibit F-actin–stained pedestals in the FKBP-5-ptase/FRB–expressing cells (Figure 4C). F-actin pedestal staining was visualized beneath nearly every EPEC microcolony adhering to rapa-treated cells that did not express the fusion proteins (Figure 4C, not expressing). Finally, we noted that cell colonization and pedestal formation by EPEC were significantly impaired by transient overexpression of myc-tagged 5-ptase, but not by overexpression of an enzyme whose PI(4,5)P2-binding domain was mutationally inactivated (see Supplemental Figure S4). Taken together, these data argue that PI(4,5)P2 is essential for bacterium attachment to the host cell surface and for construction of the actin pedestal.
The levels of PI(4,5)P2 declined after they had accumulated (Figure 3A), most likely because EPEC activated mechanisms that consume PI(4,5)P2. PI(4,5)P2 can be eliminated by three possible mechanisms: dephosphorylation, hydrolysis by phospholipases, and phosphorylation. We examined the third scenario, namely, the ability of the microbe to stimulate the activity of PI3K, which phosphorylates the inositol ring 3′-OH group of PI(4,5)P2, thus, producing PI(3,4,5)P3. Confocal data show that the PH domain of Akt fused to GFP, a marker for PI(3,4,5)P3, forms clusters underneath bacterial microcolonies in cells infected with wild-type EPEC (Figure 5, top panel, indicated by arrows), but not under escV (middle) or Δtir (lower) EPEC mutants. PI(3,4,5)P3 accumulation was restored in cells infected with Δtir mutant complemented with the Tir-expressing plasmid (Supplemental Figure S5). The GFP-PH-Akt containing the R25C mutation, which exhibits reduced binding affinity to PI(3,4,5)P3 (Servant et al., 2000 ), showed only negligible levels of accumulation underneath microcolonies of the three EPEC strains (Supplemental Figure S6). Together, these data suggest that Tir of EPEC promotes local accumulation of PI(3,4,5)P3, possibly via recruitment and stimulation of PI3K that converts PI(4,5)P2 into PI(3,4,5)P3.
Coimmunoprecipitation experiments were performed to explore the ability of Tir to interact with PI3K (the p85 regulatory subunit). Cells were first infected with wild-type EPEC or with a Δtir mutant, complemented with plasmids expressing wild-type Tir or with plasmids encoding the TirY474F and TirY454F phospho-site mutants. Next, Tir was immunoprecipitated as described in Materials and Methods. Western blotting analysis revealed that Tir was immunoprecipitated from cell lysates at comparable levels (Figure 6A, top panel). Consistent with Y474 being the major phosphorylation site of Tir (Campellone and Leong, 2005 ), Y474F mutation, but not Y454F mutation, resulted in a significant reduction in tyrosine-phosphorylation levels of Tir (Figure 6A, middle panel). Interestingly, the PI3K was noticeably coimmunoprecipitated with wild-type and the Y474F mutant forms of Tir, but negligibly with TirY454F (Figure 6A, bottom panel). These results led us to suggest that the PI3K interacts with injected Tir and that phosphorylation of Y454 is essential for these interactions. This finding also prompted us to investigate whether the binding of PI3K to Tir can promote PI3K kinase activity. Indeed, we observed Tir-dependent stimulation of Akt phosphorylation on serine 473 (using the activation-specific α-Akt-P-Ser-473 antibody), which is commonly used as readout for PI3K activation (Figure 6B). In line with observations reported in panel A, we found that PI3K activation requires Tir and its Y454 (but not Y474) phosphorylation site.
A plethora of evidence suggests that EPEC infection disrupts epithelial tight junction barrier functions in vitro and in vivo (Shifflett et al., 2005 ; Guttman et al., 2006 ). Recent studies suggest that PI3K plays an important role in modulating adherence junctions, tight junctions, and epithelial cell polarity (Sheth et al., 2003 ; Hollande et al., 2005 ). Thus, an intriguing hypothesis is that EPEC-mediated activation of PI3K stimulates the breakdown of epithelial barrier functions and cell polarity. To test this hypothesis, the next set of experiments examined the ability of Tir expression to affect tight junction barrier functions of polarized MDCK cells. With this in mind, we used two classically used assays, which combine the monitoring of cell monolayer TER changes (Figure 7, top) with monolayer leakiness to 4-kDa FITC-dextran over time (Figure 7, bottom). As expected, MDCK infection with EPEC-wt significantly reduced the electrical resistance and correspondingly increased the monolayer permeability to FITC-dextran. Cell infection with EPEC-Δtir had also caused a net reduction of TER values and a concomitant increase in cell permeability to FITC dextran, but with delayed kinetics than EPEC-wt. Here it should be noted that previous reports yielded contradictory results concerning the ability of EPEC-Δtir to perturb the tight junctions. Although disruption of TER was independent of Tir in Caco-2–infected cells (Dean and Kenny, 2004 ), Tir deletion had no impact on TER values of T84-infected cells (Muza-Moons et al., 2003 ). The basis for these differences has yet to be explored.
Notably, overexpression of Tir (EPEC Δtir+Tir) resulted in a steady increase in TER values, above the control levels. In agreement, cell monolayer leakiness upon infection with this EPEC strain was minimal. These data suggest that overexpressed Tir tightens the junction barrier, unlike native Tir levels that contribute to tight junction disruption. In contrast, overexpression of the Tir mutants Y454F or Y474F caused partial increase in tight junctions permeability, suggesting that phosphorylation of Y454 and Y474 play a role in tightening the junctions upon Tir overexpression. However, the observed similar effects suggest that the mere ability of Tir Y454 to recruit and activate PI3K does not play a major role in EPEC-mediated disruption of tight junctions.
Collectively, our results suggest that EPEC induces the generation of a PI(4,5)P2/PI(3,4,5)P3-enriched domain on the plasma membrane, immediately underneath its attachment site. At least part of the PI accumulation could be contributed by de novo PI production via EPEC-mediated recruitment and activation of the PI(4)P5 and PI3 kinases. Our data also suggest that although the T3SS is required to promote efficient accumulation of PI(4,5)P2 (Figure 1E), a T3SS-independent component may also be involved in the process (Figures 1D and and2).2). We reasoned that BFP alone is probably insufficient to mediate this effect, because PI(4,5)P2 accumulation was apparent beneath EPEC lacking BFP (Figure 1C), but not beneath an E. coli laboratory strain that ectopically expresses BFP (Figure 1F). In this context, it is worth mentioning that besides pili and fimbria, a plethora of different bacterial nonpolymeric adhesions exist and recognize many different elements on the host cell PM (Pizarro-Cerda and Cossart, 2006 ). These adhesion factors could act independently or in concert with BFP, to facilitate the clustering of various cellular factors, including PI(4,5)P2 and PI(4)P5 kinase. This phenomenon may be related to recently reported data suggesting T3SS-independent involvement of cholesterol-enriched lipid rafts in modulating bacterium–cell adhesion and the efficiency of type III effector delivery (Riff et al., 2005 ; Allen-Vercoe et al., 2006 ). Thus, it is possible, for instance, that superimposed on the T3SS-independent activities, T3SS components recruit PI(4)P5 kinase activator(s), such as Arf GTPases (Honda et al., 1999 ) that facilitate local synthesis of PI(4,5)P2. Moreover, certain T3SS components could specifically bind and facilitate the recruitment of PI(4,5)P2 into bacterial-PM contact sites.
Although infection with the wild-type EPEC strain resulted in a transient accumulation of PI(4,5)P2 (Figure 3A and Supplemental Figure S2), the PI accumulation always reached unaltered plateau levels in EPEC-escV–infected cells (Figure 3C). This suggests that the PI(4,5)P2 elimination phase in wild-type infected cells is contributed by a T3SS component. An interesting possibility is that Tir–intimin interactions prompt tyrosine phosphorylation (activation) of phospholipase C-gamma, which degrades PI(4,5)P2 to IP3 and phosphatidic acid (Kenny and Finlay, 1997 ). An additional pathway of PI(4,5)P2 removal could be achieved by Tir-dependent activation of PI3Ks (Figure 6).
PI(4,5)P2 has been suggested to serve as a membranous platform critical for PM targeting of specific proteins and lipids (Heo et al., 2006 ). Proteins suggested to be involved in EPEC pedestal biogenesis, such as Nck/WASP/Arp2/3 (Gruenheid et al., 2001 ; Campellone and Leong, 2005 ; Schuller et al., 2007 ), α-actinin (Goosney et al., 2000 ), annexin2 (Goosney et al., 2001 ; Zobiack et al., 2002 ), dynamin (Unsworth et al., 2007 ), and clathrin (Veiga et al., 2007 ) were also reported to associate, directly or indirectly, with PI(4,5)P2 (Fukami et al., 1992 , 1994 ; Miki et al., 1996 ; Lin et al., 1997 ; Ford et al., 2001 ; Hayes et al., 2004 ; Rescher et al., 2004 ; Ho et al., 2006 ; Zoncu et al., 2007 ). The similar bell-shaped profiles of actin and PI(4,5)P2 accumulations (Figure 3) suggested a possible link between the two processes. Indeed, disruption of PI(4,5)P2 platforms using two independent strategies of PI(4,5)P2 depletion resulted in abrogation of actin pedestal construction (Figure 4C and Supplemental Figure S4D).
How does PI(4,5)P2 exert pedestal formation? It is possible that EPEC generates a sort of PI(4,5)P2 raft domain confined to infection sites. The association and clustering of a variety of signaling molecules, including WASP (Papayannopoulos et al., 2005 ), Fyn, and Src tyrosine kinases (Filipp and Julius, 2004 ; Taylor and Hooper, 2006 ) within that domain brings these protein factors into proximity, and also with EPEC T3SS effectors, such as Tir. Through interactions with PI(4,5)P2 domains, EPEC may transduce signal transduction pathways required for actin-pedestal formation (Bhavsar et al., 2007 ). Mechanistically, this event could be reminiscent of the previously described PI(4,5)P2-induced actin-tail production required for the motility of raft-enriched vesicles (Rozelle et al., 2000 ).
Notably, the effects of PI(4,5)P2 dephosphorylation on EPEC adherence and pedestal formation were partial. This could be explained by the fact that PI(4)P, the product generated upon PI(4,5)P2 hydrolysis, is efficiently rephosphorylated to PI(4,5)P2 by locally concentrated PI(4)P5 kinases (Balla, 2005 ). Additionally, reducing cellular levels of PI(4,5)P2 has been shown to increase Abl kinase activity (Plattner et al., 2003 ; Plattner and Pendergast, 2003 ). Abl is a key component in pedestal biogenesis (Backert et al., 2008 ). Thus, on the one hand, PI(4,5)P2 depletion may attenuate certain aspects of actin pedestal formation by affecting the localization and activity of crucial proteins. On the other hand, PI(4,5)P2 reduction may counteract these effects by stimulating processes that promote actin-pedestal biogenesis, e.g., through Abl activation. Finally in this context it should be noted that acute PI(4,5)P2 depletion can be toxic to the cells. Our measurements were performed only on cells showing a healthy morphology. Thus, it is likely to assume that the PI levels in those cells were decreased moderately, resulting in only partial effects on pedestal formation.
Our results show that PI(3,4,5)P3 accumulates underneath EPEC-wt microcolonies in a Tir-dependent manner (Figure 5). Remarkably, tyrosine 454 on Tir was shown to interact with PI3K and to be essential for its activation (Figure 6). These effects suggest that EPEC is capable of inducing local de novo production of PI(3,4,5)P3. The mechanism underlying PI3K activation may involve the ability of EPEC to induce the activation of receptor tyrosine kinases (Roxas et al., 2007 ). Previous studies yielded contradictory observations regarding the ability of EPEC to modulate PI3K activity (Celli et al., 2001 ; Quitard et al., 2006 ; Roxas et al., 2007 ). Our results are consistent with studies supporting the ability of EPEC to activate this kinase, and they provide additional novel information suggesting that the binding of PI3K to phosphorylated Y454 of Tir mediates this event. We propose that binding occurs via the phospho-tyrosine/SH2 domain interaction of the PI3K regulatory p85 subunit, as our coimmunoprecipitation experiments suggest.
What role does PI3K activation play in EPEC pathogenesis? PI3K inhibition by cell treatment with LY294002 had no effect on bacterial adherence and actin recruitment to sites of attachment (Supplemental Figure S7). This result may agree with observations suggesting that tyrosine 454 of Tir plays a minor role in pedestal biogenesis (Campellone and Leong, 2005 ), irrespective of PI3K recruitment and activation (Figure 6). Specific signaling pathways, resulting from PI3K activation, are involved in regulating cell death (Cully et al., 2006 ). PI3K is a negative regulator of Toll-like receptor signaling (Hazeki et al., 2007 ). Thus, a plausible scenario is that EPEC activates PI3K to regulate apoptosis (Crane et al., 1999 ; Abul-Milh et al., 2001 ; Heczko et al., 2001 ; Figueiredo et al., 2007 ; Roxas et al., 2007 ), and/or inactivation of innate immunity (Ruchaud-Sparagano et al., 2007 ).
It has recently been suggested that PI(3,4,5)P3 serves as a crucial regulator of basolateral PM formation in polarized MDCK cells (Gassama-Diagne et al., 2006 ). A very interesting scenario would be that EPEC adhesion to the apical cell surface of polarized enterocytes triggers the local formation of a PI(3,4,5)P3-rich domain with basolateral PM characteristics. By modulating the basolateral localization of proteins involved in colonic innate immunity (Lee et al., 2006 ), EPEC may cause perturbation of normal immunological functions.
In summary, our data support a model whereby at the very initial stages of EPEC association with the cell surface, yet unidentified EPEC external factors prompt the generation of a PI(4,5)P2-enriched membrane platform immediately underneath the bacterium infection site. This event could be promoted, at least in part, by the ability of these factors to recruit the PI(4)P5 kinase, which converts PI(4)P on the PM to PI(4,5)P2. The construction of this domain could strengthen the initial attachment of the bacterium to the cell surface and concomitantly facilitate the assembly of the T3SS channel in the host cell PM and the subsequent protein translocation. Upon translocation of Tir, and its interaction with intimin, the PI(4,5)P2 platform may expand, promoting the recruitment and activation of an array of proteins that interact with Tir and are essential for constructing the actin pedestal. Once these functions are fulfilled, the PI(4,5)P2 domain is destroyed. This activity could be achieved by PLC-gamma activity, which breaks down PI(4,5)P2 to IP3 and phosphatidic acid, and by phosphorylation of the inositol ring in the 3′ position by PI3K, generating a PI(3,4,5)P3-enriched membrane domain in the infection site. Tir phosphorylated on tyrosine 454 mediates recruitment and activation of PI3K. Activation of the PI3 kinase could play an essential role in various signaling cascades elicited by EPEC to modulate cell death, innate immunity, and the breakdown of epithelial cell surface polarity. These processes now deserve deeper mechanistic investigation, which we believe will lead to new insights into the molecular mechanisms underlying the ability of this fascinating pathogen to subvert normal cellular functions.
We are especially indebted to Dr. Julieta Leyt and Shulamit Cohen for consultation and advise. We also thank Dr. Sharon Eden (Hebrew University, Medical School) for contributing the GFP-actin construct. This work was supported by grants from the Israel Science Foundation (I.R.), and by the Israel Science Foundation Grants 626/04 and 1167/08 and a grant from the Ministry of Health (B.A.); I.R. is the Etta Rosensohn Professor of Bacteriology.
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E08-05-0516) on November 5, 2008.