Mitochondrial Targeting of the Enteropathogenic Escherichia coli Map Triggers Calcium Mobilization, ADAM10-MAP Kinase Signaling, and Host Cell Apoptosis

Enteropathogenic E. coli (EPEC) is an important human diarrhea-causing bacterium. The pathogenic effects of EPEC largely depend upon its ability to inject a series of proteins, termed effectors, into the host cells. One such effector is the mitochondrion-associated protein (Map). Map has been shown to induce actin-rich projections (i.e., filopodia) on the infected cell surface and activate a Rho GTPase enzyme termed Cdc42. Nonetheless, although most injected Map localizes to host mitochondria, its functions in the mitochondria remain unknown. Here, we show that Map targeting of mitochondria stimulates the disruption of mitochondrial membrane potential to induce Ca2+ efflux into the host cytoplasm. The efflux stimulates the activity of a protein termed ADAM10, which induces activation of a mitogen-activated protein kinase cascade leading to host cell apoptosis. As apoptosis plays a central role in host-pathogen interactions, our findings provide novel insights into the functions of mitochondrial Map in promoting the EPEC disease.

effacement (LEE)-located effector genes to upregulate pERK in HeLa cells. The results identified EPEC-map as the only mutant strain that did not demonstrate such capability ( Fig. 2A). Notably, pERK levels in EPEC-map-infected cells were even lower than those exhibited by EPEC-escV-infected cells. This finding suggests that in the absence of Map, other bacterial components suppress ERK. We suspected that the injected EspH, a negative regulator of ERK (13), could be such a component. To address this prediction, pERK levels in HeLa cells infected with an EPEC strain doubly mutated in the espH and map genes (EPEC-map,espH) were examined. Indeed, data presented in Fig. 2B shows that pErK levels in cells infected with the EPEC-map,espH mutant strain were higher than in the EPEC-map-infected cells, reaching levels comparable to those exhibited by EPEC-escV-infected or uninfected cells. These results point to EspH as the effector that suppresses pERK in the EPEC-map-infected cells. Complementing EPEC-map with a Map-encoding plasmid increased pERK levels, which reached the levels observed in EPEC-wt-infected cells (Fig. 2B). A similar Map-dependent stimulatory effect, albeit somewhat lower, was observed in infected Caco-2 cells (Fig. 2C). Notably, in these experiments, Map was effectively translocated into the cells (Fig. S1A). Since the host actin cytoskeleton can be affected by bacterial targeted MAPKs (22), and since Map itself can modulate filamentous actin (F-actin) (10,19,23), the F-actin levels in EPEC-mapϩMap infection sites were compared to those of EPEC-map-infected cells. The results showed that the F-actin clustering levels were comparable at the infection sites of the two EPEC strains (Fig. S1B). These results argue that translocated Map is the type III secreted effector that confers pERK stimulation at an early infection time and that the effect is not contributed by changes in F-actin clustering at infection sites.
Map triggers phosphorylation of MEK, p38, and the EGFR but not B-Raf or C-Raf. Next, we asked what the upstream components that activate ERK upon stimulation by Map are. We found that the immediate upstream MEK (Fig. 3A), but not C-Raf ( Fig. 3B and C) or B-Raf (Fig. 3D), is hyperphosphorylated in response to EPEC-mapϩMap wt infection. Interestingly, pEGFR (Y1068), positioned upstream of MEK, was also upregulated (Fig. 3E), suggesting that translocated Map activates the epidermal  (Table S1) for 30 min at 37°C or remained uninfected. Cells were then lysed, and pERK and general (g)ERK levels were determined by SDS-PAGE followed by Western blotting, as described in Materials and Methods. Representative Western blots are shown. Results are means and SE from 3 independent experiments. ***, P Ͻ 0.0005; **, P Ͻ 0.005; ns, nonsignificant (P Ͼ 0.05). growth factor receptor (EGFR)/MEK/ERK signaling pathway, via a Raf-independent mechanism.
Map activates the Rho GTPase Cdc42 (19,20). Cdc42 and Rac1 can act as upstream mediators of p38 and JNK MAPK activities in response to stress or inflammatory stimuli (24). Therefore, we hypothesized that Map may stimulate the phosphorylation and activation of p38. Data in Fig. 3F indeed show a Mapdependent increase in phosphorylated-p38 (p-p38) levels. pJNK was also stimulated in response to EPEC-wt, but the clear involvement of Map in the process could not be demonstrated (data not shown). In summary, our data link Map to the stimulation of pEGFR, pMEK, pERK, and p-p38 but not to pRaf signaling.
Map mediates ADAM10 but not ADAM17 activation. ADAM17 acts as a sheddase of epiregulin, transforming growth factor ␣ (TGF-␣), amphiregulin, and heparin-binding EGF-like growth factor (25). Since these ligands are known to potentiate the EGFR/ERK/ MAPK pathway (26), we reasoned that Map might activate this pathway by stimulating ADAM17. Data in Fig. 4A suggest that this is not the case. EPEC infection activates the  (Table S1), and pERK levels were analyzed, as for Fig. 1. (B and C) Map wt translocation induces pERK in HeLa and CaCo-2 cells, and EspH counteracts this effect. Cells were infected with the indicated EPEC strains (Table S1), and pERK levels in the infected cells were analyzed, as for Fig. 1. Results are means and SE from 3 independent experiments. ***, P Ͻ 0.0005; **, P Ͻ 0.005; *, P Ͻ 0.05; ns, nonsignificant (P Ͼ 0.05).
ADAM17-mediated TGF-␣ sheddase activity, but the process is T3SS and Map independent. EGF and betacellulin (BTC) share overlapping signaling properties concerning EGFR activation (27). Because the metalloprotease ADAM10 acts as the main sheddase of BTC and EGF (25), we hypothesized that Map might be capable of activating the ADAM10 sheddase activity. Results presented in Fig. 4B show that cell infection with EPEC-wt or EPEC-mapϩMap wt indeed resulted in increased BTC shedding compared to EPEC-map or EPEC-escV infection or no infection. These data suggest that Map stimulates the ADAM10 sheddase activity. To further confirm this conclusion, we examined the ability of EPEC to upregulate pERK in cells whose ADAM10 expression was silenced by small interfering RNA (siRNA) (Fig. S1C). Data in Fig. 4C show that cell infection with EPEC-mapϩMap wt failed to increase pERK levels in the ADAM10 siRNA-treated cells, while the stimulatory effect was apparent in the control siRNA-treated cells. Notably, Map translocation was not affected by ADAM10 siRNA treatment (Fig. S1D). These results suggest that the Map-mediated stimulation of pERK is due to its ability to activate ADAM10.
The GEF and PDZ binding motifs of Map are not involved in ADAM10-EGFR-MAPK activation. As previously noted, Map consists of an MTS, a WxxxE Rho GEF, and a C-terminal TRL class I PDZ binding motif. To investigate which of these motifs mediate ERK activation, we generated EPEC mutant strains bearing a mutation in the WxxxE or  (Table S2). The levels of phosphorylated and total proteins were determined as described for Fig. 1. Results are means and SE from 3 independent experiments. Notably, we used three different anti-phospho-Raf antibodies to strengthen the conclusion that Map stimulates the EGFR-MAPK pathway in a Raf-independent mechanism. The anti-pRaf antibodies used to detect Raf kinase stimulation linked to the EGFR signaling were the classical anti-c-Raf phosphorylated on Ser338 and the anti-B-Raf phosphorylated on Ser445, a site in B-Raf shown to be equivalent to that in c-Raf (Ser338) (68), and anti-pc-Raf phosphorylated on Ser289, Ser296, and Ser301. These serine residues of c-Raf have been reported to be EGF induced (69). ***, P Ͻ 0.0005; **, P Ͻ 0.005; ns, nonsignificant (P Ͼ 0.05). the TRL motif (Table S1 and Fig. 5A). The capacity of these strains to alter the pERK levels in infected HeLa cells was evaluated, as before. pERK levels in EPEC-map-and EPEC-mapϩMap wt -infected cells served as a reference in these experiments. Results show that infection with EPEC-mapϩMap WxxxA and EPEC-mapϩMap ΔTRL caused significant upregulation of pERK levels compared to those in EPEC-map-infected cells (Fig. 5B). A similar effect was observed on pMEK (data not shown), pEGFR (Fig. 5C), and ADAM10 BTC sheddase activity (Fig. 5D). Compared to Map wt , the capacity of the mutant Maps to undergo translocation into the host cells was somewhat reduced, especially in the case of Map ⌬TRL (Fig. 5E). This may explain the slightly reduced levels of pERK and pEGFR levels in cells infected with these EPEC strains compared to levels in cells infected with the MAP wt strain. However, the amounts of the translocated effectors seemed to be sufficient for inducing the indicated effects. Notably, previous studies suggested that F-actin-rich filopodia are not generated in response to translocated Map ⌬TRL (19,21). Map WxxxA also fails to evoke Cdc42 and filopodium formation ( Fig. S1E) (19). These data suggest that Map activates the ADAM10-MAPK and the Rho GTPase signaling pathways by two distinct and independent mechanisms.
Finally, in this context, ectopic expression of Map tagged in its C terminus with eGFP (Map-eGFP) (Fig. S1F) resulted in an increase in pERK levels compared to eGFPexpressing cells (Fig. S1G). The fusion protein colocalized with the mitochondrial marker Hsp60 (Fig. S1H). In contrast, ectopic expression of N-terminal mCherry-tagged Map did not colocalize with Hsp60 (Fig. S1H) and did not affect pERK levels (Fig. S1G) compared to mCherry expression. These results suggest that the expression of Map is sufficient for stimulating the ADAM10-MAPK signaling cascade and that targeting of mitochondria by the effector may play a role in the process.
ADAM10 and MAPK activation requires mitochondrial targeting of Map. Next, we examined the importance of Map's mitochondrial targeting in the upregulation of the ADAM10-pERK axis. The N-terminal ϳ25-amino-acid sequences of bacterial type III effectors typically contain a signal that mediates their specific secretion via the T3SS (28). Thus, the MTS of Map, which was localized to the N-terminal 44-amino-acid segment of the protein (18), likely contains the signal that mediates its translocation via the T3SS. Therefore, the mere deletion of the MTS (Map ΔMTS ) would not be a reasonable approach for examining its role in signaling, as such mutation may affect the translocation of the effector into the host. To address this potential problem, the N-terminal 25-amino-acid type III secretion signal of EspH, an effector whose localization upon translocation has been reported to be confined to the bacterial infection sites and not to mitochondria (9), was fused to the N terminus of Map ΔMTS (Map ΔMTS -EspH 1-25 ) (Tables S1 and S3; Fig. 6A). This modification was expected to facilitate the translocation of the MTS deleted Map mutant. Data presented in Fig. 6B indeed show that the translocation  [4]) in altering mitochondrial morphology (18). To explore the significance of this motif, an EPEC-mapϩMap Δ101-152 mutant strain was generated (Tables S1 and S3; Fig. 6A). Map Δ101-152 was effectively translocated into the infected cells (Fig. 6B) and localized to mitochondria (Fig. S1I). Cell infection with EPEC-mapϩMap Δ101-152 resulted in a significant reduction in pMEK (data not shown), pERK (Fig. 6D), pEGFR (Fig. 6E), and ADAM10 activity (Fig. 6F), reaching the minimal values exhibited upon infection with EPEC-map. These data suggest that a Map MTR motif may be involved in promoting the ADAM10-MAPK signaling pathway.
Map promotes Ca 2؉ mobilization from mitochondria. Studies have shown that Map can disrupt the mitochondrial membrane potential (18,29). We confirmed this in our experimental system by showing that cell infection with EPEC-mapϩMap wt resulted in a continuous reduction in the fluorescence of the mitochondrial membrane potential reporter tetramethylrhodamine ethyl ester (TMRE). Only a moderate reduction in the TMRE fluorescent signal was observed in response to EPEC-map infection ( Fig. 7A; Movie S1). The disruption of mitochondrial membrane potential can lead to Ca 2ϩ efflux from mitochondria (reviewed in reference 30). Using the mitochondrial Ca 2ϩ -sensitive fluorescent reporter Rhod-2, we showed that cell infection with EPEC-mapϩMap wt results in a sharp loss of mitochondrial Ca 2ϩ levels, while infection with EPEC-mapϩMap Δ101-152 led to a slight reduction in mitochondrial Ca 2ϩ compared to EPECmap. Cell infection with EPEC-mapϩMap ΔMTS -EspH 1-25 resulted in an intermediate effect ( Fig. 7B; Movie S2). The efflux of mitochondrial Ca 2ϩ may cause a parallel increase in cytoplasmic Ca 2ϩ . Indeed, cytosolic Ca 2ϩ imaging by the Fluo-8 probe showed a parallel increase in cytoplasmic Ca 2ϩ levels that was essentially spread throughout the cytoplasm in response to EPEC-mapϩMap wt but not to EPEC-map or EPEC-mapϩMap Δ101-152 infection. Infection with EPEC-mapϩMap ΔMTS -EspH 1-25 resulted in a slight increase in cytoplasmic Ca 2ϩ ( Fig. 7C; Movie S3). Taken together, these data suggest that mitochondrially targeted Map disrupts the mitochondrial membrane potential, leading to Ca 2ϩ efflux from the mitochondria and Ca 2ϩ influx into the infected cell cytoplasm. The MTR motif of MAP seems to play a major role in this effect.
Mitochondrial membrane potential disruption by CCCP induces Ca 2؉ extrusion and ADAM10/pERK activation. Previous studies suggest that Ca 2ϩ can trigger ADAM10 (31)(32)(33)(34)(35) as well as MEK, ERK, and p38 activity (36)(37)(38). Based on these studies, we hypothesized that enrichment of the host cytoplasm with Ca 2ϩ might prompt ADAM10 and MAPK activation. To tackle this hypothesis, HeLa cells were treated with the protonophore carbonyl cyanide m-chlorophenylhydrazone (CCCP), which results in the dissipation of the mitochondrial membrane potential, allowing the mitochondria to release accumulated Ca 2ϩ and preventing its further accumulation in this organelle (39). Our results show that CCCP treatment resulted in a significant loss of mitochondrial Ca 2ϩ ( Fig. 8A; Movie S4), and a parallel increase in cytosolic Ca 2ϩ ( Fig. 8B; Movie S5). The activity of ADAM10 (Fig. 8C) and the levels of pERK (Fig. 8D) significantly increased in response to HeLa cell treatment with CCCP. These data, combined with data presented in Fig. 5 and 6, support the hypothesis that ADAM10 and the downstream ERK cascade are activated by Map-mediated Ca 2ϩ extrusion from mitochondria and the consequent Ca 2ϩ increase in the host cell cytoplasm.
Map induces apoptosis. Activation of the ERK cascade is linked to cell survival and death (40,41). Mitochondrial dysfunction, membrane potential perturbation, fragmentation (fission), and redistribution have all been shown to trigger cell death (30). Data have also linked Ca 2ϩ signaling, mitochondria, and various forms of cell death (42,43). Translocated Map seems to impact Ca 2ϩ compartmentalization ( Fig. 7B and C), and active ERK (Fig. 6D), EGFR (Fig. 6E), and ADAM10 (Fig. 6F) levels in a mitochondrial targeting-dependent manner. We, therefore, postulated that these Map-mitochondria dependent processes could promote host cell death. Data in Fig. 9 unambiguously show that cell infection with EPEC-mapϩMap wt resulted in a significant increase in the appearance of apoptotic cells compared to cells infected with EPEC-map or uninfected cells. This phenomenon was dependent on the ability of Map to target mitochondria, as apoptotic-cell levels in EPEC-mapϩMap ΔMTS -EspH 1-25 -and EPEC-mapϩMap Δ101-152infected cells were significantly reduced compared to those in EPEC-Map wt -infected cells.

DISCUSSION
Map was shown to target host cell mitochondria, in vitro and in vivo (17,18,29). Map has also been suggested to play a significant role in bacterial colonization and the induction of the diarrheal effect (21,29,44,45). Despite its importance as a virulent factor and the fact that most Map localizes at steady state in mitochondria, the functional significance of Map-mitochondrion interactions is not known. Here, we demonstrate, for the first time, that early in infection, mitochondrial targeting of Map, particularly via its MTR motif, plays a role in activating the sheddase activity of ADAM10 ( Fig. 5F and 7E). This activity may, in turn, induce the pEGFR (Fig. 3E, 5C, and 6E) and possibly another EGFR family member, ErbB4, because BTC has been shown to activate both the EGFR and ErbB4 (46). The activation of the EGFR may then elicit the MEK/ ERK/MAPK signaling cascade (Fig. 3A, 5B, and 6D). Our data also suggest that Map can activate p38 (Fig. 3F). Previous studies suggested that EPEC and similarly, enterohemorrhagic E. coli (47) rapidly induce T3SS-dependent ERK, p38, and JNK in infected cells (15,16). Here, we propose that Map is the effector that activates ERK and p38. All in all, these effects of Map do not depend on the ability of the effector to activate Cdc42 and induce actin-rich filopodia, because the mutant Map WxxxA , which fails to induce Cdc42 (Fig. S1E), retains the capacity to induce the ADAM10-MAPK signaling (Fig. 5). Thus, early upon translocation, Map localizes to infection sites, but not to mitochondria, where it exerts Rho GEF signaling to remodel the host actin cytoskeleton. In parallel, the effector fraction targeting mitochondria evokes the ADAM10-MAPK signaling, independently.
We also show that Map targeting of mitochondria mediates the disruption of the mitochondrial membrane potential and Ca 2ϩ release from mitochondria into the host cytoplasm (Fig. 7). Evidence suggests that the activity of ADAM10, MEK, ERK, and p38 is regulated by Ca 2ϩ (31-33, 37, 48, 49). Thus, a Map-mediated increase in cytoplasmic Ca 2ϩ could have contributed to the activation of these enzymes. This hypothesis may also be supported by the observation that disruption of mitochondrial membrane potential by the protonophore CCCP increases cytoplasmic Ca 2ϩ levels and the ADAM10 and ERK activities (Fig. 8). A proposed model for the action of Map is presented in Fig. 10. How does the activation of ADAM10-ERK and p38 relate to EPEC pathogenesis? MAPKs can act as a double-edged sword in promoting cell death and survival (50). In most cases, ERK signaling promotes cell survival, e.g., by activating prosurvival BCL2 proteins and repressing prodeath proteins (BAD, BIM, BMF, and PUMA). However, active ERK, under certain conditions, such as nutrient deprivation, can also drive the expression of the prodeath proteins to control autophagy or apoptosis (reviewed in references 51 and 52). Similarly, previous studies suggested that activation of p38 is required for inducing the mitochondrial death pathway (53-55). Hence, it is possible that by targeting mitochondria, Map induces host cell deprivation, which, combined with ERK and p38 activation, triggers host cell apoptosis (Fig. 9).
Similar to Map, the type III secreted effector EspF targets host mitochondria via an N-terminal targeting signal to cause mitochondrial dysfunction and cell death (56)(57)(58). However, in our screen, EspF was not identified as an effector that stimulates pERK ( Fig. 2A). This concern was investigated by generating new EPEC strains in which EspF was expressed in the background of EPEC-map (EPEC-mapϩEspF) or EPEC-map,espF (EPEC-map,espFϩEspF) (Table S1). We used these bacterial strains to infect HeLa cells under the experimental conditions used to investigate the role of Map. The levels of translocated EspF upon cell infection were nearly identical (Fig. S1J). Infection with either EPEC strain resulted in a minor increase in host pERK levels compared to EPEC-mapϩMap wt (Fig. S1K and L). Infection with EPEC-mapϩEspF also caused a slight decrease in mitochondrial Ca 2ϩ levels ( Fig. 7B; Movie S2) and a corresponding increase in cytoplasmic Ca 2ϩ levels ( Fig. 7C; Movie S3). Confocal imaging revealed that the localization of translocated EspF was confined to host cell areas juxtaposed to the infection sites, where the effector protein only partially colocalized with the mitochondrial marker Hsp60 (Fig. S1M). These observations, which are consistent with our most recent discoveries (59), suggest that early upon infection with EPEC-espF, Map is FIG 9 Map induces host cell apoptosis. HeLa cells infected with the indicated EPEC strains were subjected to a flow cytometry-based apoptosis assay, as described in Materials and Methods. The fraction of EPECexposed apoptotic cells (i.e., annexin V-positive and PI-positive cells) was normalized to the fraction of apoptotic cells detected in uninfected cells. The fractions of these cells in uninfected cells and in EPEC-map-, EPEC-mapϩMap wt -, EPEC-mapϩMap ΔMTS -EspH 1-25 -, and EPEC-mapϩMap Δ101-152 -infected cells were 8%, 24%, 11%, 16%, and 10%, respectively. The fraction of apoptotic cells in EPEC-espZ-infected cells, which was greater than 50% and ϳ6-fold higher than the fraction of apoptotic cells in uninfected cells (inset), served as a positive control for these experiments, as previous studies have shown that an EPEC-espZ mutant increases dramatically the levels of apoptotic and necrotic cells (70). The fractions of cells that were dually labeled with annexin V and PI were negligible (Յ6%) in all experiments. Results are means and SE from 3 independent experiments. ***, P Ͻ 0.0005; *, P Ͻ 0.05; ns, nonsignificant (P Ͼ 0.05). translocated and efficiently targets mitochondria, where it initiates the cascade of events that leads to pERK activation. In contrast, infection with EPEC-map allows EspF translocation, which only partially targets mitochondria and thereby exhibits a limited capacity to induce pERK. Additionally, Map's MTR motif, which we show here to be critical for promoting ADAM10-MAPK signaling, does not exist in EspF (18). Nevertheless, at a later infection time, EspF may reach the host mitochondria more effectively to promote cell death that is superimposed on the process that has been started by Map.
By targeting mitochondria, Map induces an apoptotic effect early upon infection. However, EspH may counteract this effect at a later infection time (13). These opposing effects may delay the enforcement of the host cell death and therefore be required for establishing bacterial colonization and successful infection. The early Map-dependent effects could be physiologically important, because in vivo infection studies suggested that Map expression is suppressed at an early colonization phase, and when Map expression is reduced, EPEC colonization is also significantly reduced (44). Thus, the downregulation of Map could be another mechanism that terminates its effects. Many intestinal microbial pathogens modulate mitochondrial cell death to achieve successful infection (60). Hence, mitochondrial targeting, e.g., by Map, may represent a universal event that plays a role in establishing bacterial pathogens' colonization of the gut. The mechanisms underlying these effects are presently unknown. One intriguing hypothesis is that by causing mitochondrial dysfunction, pathogens build a niche where they succeed in competing with nearby commensal bacteria (microbiota) whose viability depends upon functional mitochondria of the host (61).
The precise mechanism by which Map evokes ADAM10 activation through mitochondrial targeting is still not well understood. Recent studies have shown that membrane externalized phosphatidylserine is required for activating the ADAM10 sheddase activity (62). Hence, an interesting hypothesis would be that the induction of mitochondrial cell death by Map is caused by phosphatidylserine externalization. This process is shown here to take place because the detection of cell apoptosis utilizes the annexin V reporter, which binds specifically to externalized phosphatidylserine (Fig. 9). , where it disrupts the mitochondrial membrane potential, resulting in the release of mitochondrial Ca 2ϩ into the host cytoplasm (c). Elevated cytoplasmic Ca 2ϩ levels then stimulate the ADAM10 sheddase activity (d), which results in the release of EGF/BTC (e). The released ligands (f) bind EGFR (g), which in turn activates (phosphorylates) the EGFR-MEK-ERK signaling cascade (h and i). This sequence of events can lead to host cell apoptosis (j). The rise in cytoplasmic Ca 2ϩ may also directly trigger the activity of the MAPK p38 (k), which can also evoke apoptosis (l).
Finally, most recent findings may signify the importance of ERK in EPEC infection by showing that the effector NleD selectively cleaves several MAPKs but not ERK (63), probably leaving it intact for activation by Map.

MATERIALS AND METHODS
Bacterial strains, antibodies, primers, plasmids, and site-directed mutagenesis. Bacterial strains, antibodies, plasmids, and primers used in this study are listed in Tables S1 to S4, respectively. For constructing the EPEC-espH,map double mutant strain (Table S1), Red recombinase system (64) was used to mutate map in EPEC-espH::Kan strains. The upstream and downstream recombination sequences were PCR amplified from the genomic DNA of EPEC-wt using primer pairs 1371-4197 and 1495-4198, respectively (Table S4). Primers 1354 and 1355 (Table S4) were used to amplify a chloramphenicol cassette from pKD3 (Table S3). DNA of the map::Cam allele was made by isothermal assembly of these three PCR fragments (65), followed by electroporation into the EPEC espH::Kan strain containing pKD46 (Table S3) containing Red genes (␥, ␤, and exo). The desired mutants were selected, and pKD46 was cured at 42°C. The mutation was verified using sequencing and PCR with flanking primers. The Map mutants Map ΔMTS and Map WxxxA , which were subcloned into the pSA10 plasmid, were constructed using the 5=-phosphorylated primer pairs MTS delta F-MTS delta R and WxxxA F-WxxxA R, respectively (Table S4). Mutations were introduced by a site-directed mutagenesis method using Platinum Superfi II DNA polymerase, based on protocol B for deletions in the manufacturer's protocol (https://assets.thermofisher.com/TFS-Assets/LSG/manuals/MAN0014883_Platinum_SuperFi_PCR _MM_UG.pdf) (Thermo Fisher). The Map ΔTRL , Map ΔMTS -EspH 1-25 , and Map Δ101-152 mutants were prepared by the standard Gibson assembly method (65), using the Gibson assembly master mix (number E2611; NEB) and the respective linearization primers and the gblock (Table S4). The wild-type and mutated Map constructs were electroporated into EPEC-map, and Amp r EPEC-mapϩMap wt strains were isolated (Table S1). The Map wt -eGFP plasmid (Table S3) was constructed by the Gibson assembly method using mCherry-Map wt (Table S3) as the template. The primers EGFP F, EGFP R, Map EGFP F, and Map EGFP R (Table S4) were used for generating the Map-eGFP plasmid. All mutations were confirmed by sequencing. All bacterial strains used in this study were grown in Luria-Bertani medium supplemented with appropriate antibiotics (Table S1).
Bacterial growth, activation, and infection. The T3SS of all bacterial strains used in the study was activated (3 h at 37°C and 5% CO 2 ) before infection, as described elsewhere (13). In EPEC-mapϩMapinfected cells, Map expression was induced by supplementing the activation medium with 0.2 mM isopropyl-␤-D-thiogalactopyranoside (IPTG; Promega, Madison, WI) during the last 30 min of activation. All infections were performed in a CO 2 incubator (37°C, 5% CO 2 , 90% humidity) for 30 min, as described elsewhere (13).
Confocal microscopy. (i) Immunofluorescence labeling of fixed cells and confocal imaging. The procedures employed for confocal microscopy were described previously (13). Briefly, after infection, cells were fixed with 4% paraformaldehyde, F-actin was labeled with Texas Red (TR)-phalloidin, and the bacterial and HeLa cell DNA was labeled with DAPI (4=,6-diamidino-2-phenylindole). Primary and fluorescence-labeled secondary antibodies used in this study are listed in Table S2. Cells were mounted and visualized using an Olympus FV-1200 laser scanning confocal microscope equipped with a 60ϫ, 1.42-numerical-aperture oil immersion objective. The excitation wavelengths were 405 nm, 488 nm, 561 nm, and 633 nm, and the emission filter passbands were 430 to 470 nm, 505 to 525 nm, 575 to 625 nm, and 650 to 720 nm for DAPI, green, red, and far-red fluorescence, respectively. Confocal sections were acquired at z-axis intervals of 0.5 or 1 m. The levels of F-actin at the infection sites were analyzed using Fiji (NIH) (66). A maximum-intensity projection was generated for each stack. The regions of cell-associated EPEC microcolonies were manually defined using the polygon selections tool of Fiji and were termed infection sites. The average pixel intensity (PI) at the infection site (PI i ) was measured and expressed as normalized fluorescence from the average PI of an identically defined uninfected area located near the infecting microcolony (PI u ), as follows: normalized fluorescence ϭ (PI i Ϫ PI u )/PI u . For colocalization analyses, line intensity profiles of the anti-hemagglutinin (HA) (Map) and anti-Hsp60 (mitochondria) fluorescence were generated, as previously described (13), and are depicted in Fig. S1I. Basically, a line was drawn across a randomly chosen confocal section showing Map and mitochondrial staining, and the fluorescence intensity along the line of the two different channels was generated using the intensity profile tool (plot-profile plug-in) of Fiji (NIH). Colocalized labeling was scored when the fluorescence intensity copeaked at a given place along the line (indicated by the green arrows in the intensity profile plots in Fig. S1I). Data are presented as percentages of colocalizing and noncolocalizing fluorescence intensity peaks determined in 10 such intensity profile determinations.
(ii) Live-cell imaging. Cells were seeded on 8-well ibiTreat microslides (catalog no. 80826; ibidi) for 2 days (ϳ70% confluence). Cells were washed twice with plain Dulbecco modified Eagle medium (DMEM) and then incubated with either the fluorescent mitochondrial membrane potential indicator tetrameth-ylrhodamine ethyl ester (TMRE; 20 nM) (T669; Thermo Fisher), the fluorescent mitochondrial Ca 2ϩ indicator Rhod-2 AM (1 M; ab142780; Abcam), or the fluorescent cytoplasmic Ca 2ϩ indicator Fluo-8 AM (1 M; ab142773; Abcam) in DMEM for 15 min at 37°C in the CO 2 incubator. Cells were then washed three times with Hanks' balanced salt solution (HBSS) (no. 02-016-1A; Biological Industries) and exposed to EPEC strains or the protonophore carbonyl cyanide m-chlorophenyl hydrazone (CCCP; 20 M) (ab141229; Abcam) for 30 min at 37°C in a CO 2 incubator. Cells were then immediately subjected to time-lapse confocal imaging using an Olympus FV-1200 laser scanning confocal microscope (Olympus, Japan) equipped with a temperature and CO 2 incubator, using a 40ϫ, 0.95-numerical-aperture air objective. The excitation wavelengths were 488 nm or 561 nm and the emission filter passbands were 505 to 525 nm and 575 to 625 nm for green and red fluorescence, respectively. Scans were taken every 30 s for 35 min under identical conditions. Fluorescence intensity at all time points was measured using the polygonal selection tool of Fiji (NIH) to select the desired region of interest (ROI). In the case of TMRE and Rhod-2 AM, the ROI was confined to the labeled mitochondria. In the case of Fluo-8 AM, the ROI included the entire cell. Data are presented as the average fluorescence intensity of 10 cells normalized to their fluorescence intensity at time zero. In the supplemental movies, time and titles were introduced using the time stamper and label tool of Fiji (NIH).
SDS-PAGE and Western blotting. Cells (ϳ70% confluence) were infected with the respective bacteria, as described above. Cells were then washed with phosphate-buffered saline (PBS) three times, and equal numbers of cells were lysed in sample buffer (40% glycerol, 12% SDS, 0.2 M Tris-HCl [pH 6.8], and 100 mM dithiothreitol) supplemented with bromophenol blue, followed by heating (95°C for 5 min) and vortex shaking (10 min (Table S2). For stripping, the membranes were incubated in stripping buffer (62.5 mM Tris-HCl [pH 6.8], 100 mM 2-mercaptoethanol, 2% SDS) for 20 min at 55°C. Membranes were then washed with TBST and reprobed with anti-general-EGFR (gEGFR) or MAPK antibodies (Table S2). A Fusion FX Spectra imager (Vilber Smart Imaging, Collègien, France) was used to image the membranes. Band intensity was measured by Fiji (NIH). The band intensity of the phosphorylated protein was normalized to the band intensity of the total protein. The value obtained was further normalized to the protein level of the control experiment.
Effector translocation assay. The effector translocation assay was performed as described elsewhere (13). Briefly, HeLa cells (ϳ70% confluence) grown in 6-well plates were infected with preactivated EPEC strains for 30 min at 37°C. Cells were lysed with ice-cold lysis 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. Detergent-soluble (host cytoplasm and membranes containing the translocated protein effector) and -insoluble (bacteria and associated protein effectors) fractions were separated by centrifugation (16,000 ϫ g, 4°C, 10 min) and analyzed for the presence of Map by Western blotting. The level of translocated effector was calculated by normalizing the levels of effector present in the NP-40-soluble fraction to the levels of the effector detected in the detergent-insoluble fraction.
Sheddase activity assay. The ectodomain shedding assay was performed essentially as described elsewhere (25). Briefly, HeLa cells grown on a 6-well plate (ϳ60% confluence) were transfected with betacellulin-alkaline phosphatase (BTC-AP) or with TGF-␣-AP. After 24 h, cells were infected with the appropriate EPEC strains for 30 min at 37°C. Medium containing bacteria was aspirated and replaced with Opti-MEM (reduced serum medium; no. 31985-047; Gibco) for 1 h at 37°C. The medium was then collected, and the cells were lysed in lysis buffer (2.5% [vol/vol] Triton X-100 in triple-distilled water [TDW]). Cell lysates were centrifuged (13,000 rpm, 5 min), and the supernatants were diluted 1:10 in a Triton-double-distilled-water (DDW) solution. Samples (100 l) derived from media and cell lysates were placed in a 96-well plate and incubated with 4-nitrophenyl phosphate (100 l) for 45 min at 37°C in the CO 2 incubator. The levels of TGF-␣-AP and BTC-AP released to the medium and in the cell lysates were determined by absorbance at 405 nm, using a Synergy H1 microplate reader (Biotek). The levels of sheddase activity were expressed as AP activity levels measured in the supernatant (i.e., AP released ) normalized to the total AP activity (i.e., AP released ϩ AP lysate ). In experiments measuring sheddase activity in CCCP-treated cells, cells were first transfected with BTC-AP and then treated with CCCP (20 M, 30 min, 37°C).
Apoptosis assay. Apoptosis assay was performed using the MEBCYTO apoptosis kit (annexin V-fluorescein isothiocyanate [FITC] kit; no. 4700; MBL) per user instructions. Briefly, HeLa cells were grown in a 6-well plate (ϳ70% confluence) and infected with the respective EPEC strains, as described in Map-Induced MAPK Activity and Mitochondrial Cell Death ® Materials and Methods. Cells were washed once with PBS, trypsinized, and washed once with serumcontaining DMEM and once with PBS. They were suspended in 85 l binding buffer, treated with annexin V-FITC and propidium iodide for 15 min at 22°C, and analyzed by flow cytometry (Cell Stream flow cytometer; Merck). The fractions of viable cells (annexin V negative, propidium iodide negative), early apoptotic cells (annexin V positive, propidium iodide negative), late apoptotic cells (annexin V positive, propidium iodide positive), and necrotic cells (annexin V negative, propidium iodide positive) were determined in each sample. Voltage settings for forward scatter (FSC) and side scatter (SSC) and fluorescence channels (488 and 561 nm) were kept constant for all samples. Data for 10,000 cells were recorded in each experiment.
Statistical analyses. Results are presented as means and standard errors (SE) of the means. A two-tailed Student's t test determined statistical significance. A P value of Ͻ0.05 indicates a statistically significant difference.