Uropathogenic Escherichia coli invades bladder epithelial cells by activating kinase networks in host cells

Uropathogenic Escherichia coli (UPEC) is the causative bacterium in most urinary tract infections (UTIs). UPEC cells adhere to and invade bladder epithelial cells (BECs) and cause uropathogenicity. Invading UPEC cells may encounter one of several fates, including degradation in the lysosome, expulsion to the extracellular milieu for clearance, or survival as an intracellular bacterial community and quiescent intracellular reservoir that can cause later infections. Here we considered the possibility that UPEC cells secrete factors that activate specific host cell signaling networks to facilitate the UPEC invasion of BECs. Using GFP-based reporters and Western blot analysis, we found that the representative human cystitis isolate E. coli UTI89 and its derivative UTI89ΔFimH, which does not bind to BECs, equally activate phosphatidylinositol 4,5-bisphosphate 3-OH kinase (PI3K), Akt kinase, and mTOR complex (mTORC) 1 and 2 in BECs. We also found that conditioned medium taken from UTI89 and UTI89ΔFimH cultures similarly activates epidermal growth factor receptor (EGFR), PI3K, Akt, and mTORC and that inhibition of EGFR and mTORC2, but not mTORC1, abrogates UTI89 invasion in vitro and in animal models of UTI. Our results reveal a key molecular mechanism of UPEC invasion and the host cells it targets, insights that may have therapeutic utility for managing the ever-increasing number of persistent and chronic UTIs.

Urinary tract infections (UTIs) 4 are among the most common types of community-acquired and nosocomial infectious diseases (1,2) and occur most frequently in women. More than half of all women will experience a UTI in their lifetime, and up to 30% of these women will experience a recurrent infection that may progress to chronic disease (3). The causative agents in 70 -80% of all UTIs are uropathogenic Escherichia coli (UPEC), which is capable of colonizing the urethra, bladder, and kidney (2,4,5). Infections of the urethra are usually acute and can be successfully treated with antibiotics, whereas those of the bladder and kidney may progress to be persistent and chronic with serious sequelae (2,5). The persistent and chronic UTIs are difficult to treat, in part because of the quiescent nature of the intracellular UPEC that can be "reawakened" to initiate a second wave of infection of host epithelial cells (6).
UPEC encode virulence factors, including secreted mediators and adhesive organelles (e.g. adhesins), that interact with the host cell to promote infection. Type I fimbriae (or pili) are the most common adhesins, being expressed in the majority of UPEC (7,8). The ability of UPEC to adhere to host bladder epithelial cells (BECs) is considered the most critical factor in uropathogenicity (9). Adhered UPEC have the capacity to invade, or enter, the epithelial cells and replicate in the intracellular space. Cell-based imaging studies have revealed that internalized UPEC may be found free in the cytosol or within membrane-enclosed vesicles. Free cytosolic UPEC rapidly multiply and form biofilm-like assemblies called intracellular bacterial communities (10,11). Intracellular bacterial communities are short-lived and may eventually disperse along with the infected cells following antibiotic treatment for UTI (12). Invaded UPEC can also be encased in lipid-enclosed vesicles that serve as portals to traffic the extracellular bacteria to the cytosol (13)(14)(15)(16)(17). Here, UPEC enter a dormant state, and the quiescent nature of the internalized UPEC renders them protected from natural bacterial flushing through the bulk flow of urine, immune cell surveillance, and extracellularly acting antibiotics. Evidence suggests that quiescent intracellular UPEC can persist for extended periods of time in the absence of clinical symptoms, even when antibiotics are used (12). Indeed, a recent study showed that latent intracellular E. coli egress from (previously) infected mouse bladder to promote recurrent UTI (6).
UPEC can invade bladder cells through the endocytic machineries of phagocytosis and pinocytosis (9,18). In phagocytosis, UPEC uptake involves specific cell surface receptors and Rho family GTPases. Here, activated Cdc42 and Rac trigger actin assembly and the formation of cell surface extensions that zipper up around the invading bacterium to execute the uptake (9). UPEC invasion of epithelial cells by pinocytosis may occur via caveolae and clathrin-coated pit portals (13)(14)(15)(16)(17). Caveolae are highly ordered plasma membrane invaginations enriched in scaffolding and signal transduction proteins. Clathrin-coated pits are comprised primarily of clathrin and assembly proteins that create latticed pit invaginations on the plasma membrane and can serve as cargo portals that transport, among others, pathogens (e.g. viruses and bacteria) from the extracellular milieu to inside host cells (19,20). Notably, ubiquitous large GTPase dynamin2 executes the last fission step of budding caveolae and clathrin-coated pits from the plasma membrane (21), and we have demonstrated a role for dynamin in vesicle trafficking and pathogen invasion (16,17,20). Nonetheless, the regulatory machineries and exact host cell mediators involved in the UPEC-induced vesicle formation and trafficking from the plasma membrane remain incomplete. In this work, we show that conditioned medium isolated from UPEC activate host cell protein networks of EGFR, Akt, and mTORC2 that promote the invasion of bladder epithelial cells.

Activation of PI3K/Akt is required for UPEC invasion of BECs
Our work has implicated dynamin2 in vesicle trafficking and UPEC invasion (16,17), but the exact mechanisms involved remain incomplete. Phosphatidylinositol 4,5-bisphosphate 3-OH kinase (PI3K) and Akt regulate dynamin function and vesicle trafficking (22)(23)(24). PI3K phosphorylates inositol 4,5bisphosphate (PIP 2 ) to generate PIP 3 , which, in turn, recruits the partner pleckstrin homology (PH) domain-containing proteins Sin1 and PDK1 to the plasma membrane (25). Sin1 is a unique component of mTOR complex 2 (mTORC2), and binding of the Sin1 PH domain to PIP 3 releases Sin1 inhibition of the mTOR kinase domain, activating mTORC2 (24,25). Activated mTORC2 phosphorylates Akt on Ser-473, and PDK1 phosphorylates it on Thr-308, rendering Akt fully activated to phosphorylate mTORC1 (Fig. 1A). We examined effect of UPEC on PI3K activation using the biosensor GFP-Bruton tyrosine kinase (Btk)-PH, which binds PIP 3 (26). Exposure of bladder epithelial 5637 cells to the human cystitis isolate UTI89 induced recruitment of GFP-Btk-PH to the plasma membrane, confirming PI3K activation (Fig. 1B). Deletion of the adhesin FimH, which is expressed on the tip of type I pili (to generate UTI89⌬FimH) renders the bacteria deficient in direct binding to BECs (Fig.  S1A), consistent with published reports (9,17). Notably, UTI89⌬FimH promoted the recruitment of GFP-Btk-PH to the plasma membrane (Fig. 1B), implying that FimH-mediated activates PI3K. BECs on glass cover slides and expressing the biosensor GFP-Btk-PH were incubated with UTI89 or UTI89⌬FimH (labeled with Hc-Red), fixed, and analyzed by microscopy. Arrowheads denote GFP-Btk-PH translocation to the plasma membrane. NI, not infected. C, UTI89 and UTI89⌬FimH promote Akt phosphorylation. BECs were incubated for the indicated time with UTI89 or UTI89⌬FimH. Cell lysates were analyzed by Western blotting using the indicated antibodies. Total Akt and GAPDH were used as protein loading controls. D, wortmannin inhibits UTI89-and UTI89⌬FimH-mediated Akt phosphorylation. BECs were pretreated with the indicated concentration of wortmannin for 1 h before adding bacteria. Cell lysates were analyzed for Akt phosphorylation by Western blotting. E, wortmannin inhibits UTI89 invasion. BECs were pretreated with the indicated concentration of wortmannin for 1 h and then incubated with UTI89 or UTI89⌬FimH at an m.o.i. of 10 for an additional hour. Cell invasion was performed using the gentamicin protection assay. NT, infected but not treated with inhibitor. Statistical analysis was performed by one-way ANOVA with Tukey comparison of columns. Error bars represent S.E. ***, p Ͻ 0.001; n ϭ 3. F, validation of Bay 80-6946 and MK-2206 as inhibitors of Akt activation. BECs were pretreated with Bay 80-6946 (0.4 M) or MK-2206 (4 M) for 1 h and then incubated with UTI89 or UTI89⌬FimH at an m.o.i. of 10 for an additional 10 min. Cell lysates were analyzed by Western blotting for Akt phosphorylation. G, Bay 80-6946 and MK-2206 inhibit UTI89 invasion of BECs. Cells were treated with inhibitors as in E, and UTI89 invasion was carried out using the gentamicin protection assay. Statistical analysis was performed by Student's t test compared with the untreated control. ***, p Ͻ 0.001; n ϭ 3.

Multistep bacterial invasion
physical association of the bacteria with host cells is not a prerequisite for PI3K activation in host BECs.

Opposing roles of mTORC1 and mTORC2 in UPEC invasion
UTI89⌬FimH bacteria are deficient in binding to BECs ( Fig.  S1A) and, like UTI89, are fully capable of activating PI3K and Akt (Fig. 1), suggesting possible involvement of bacterial secreted factor(s). To test this idea, UTI89 bacteria were allowed to grow in Luria Bertani (LB) broth, and conditioned medium (CM) was isolated by stepwise centrifugation at 6,000, 12,000, and 300,000 rpm, followed by filtration through 0.22-m pores that do not allow the passage of intact bacteria (Fig. S2A). BECs were incubated with equal volumes of LB alone (negative control) or the isolated UTI89 CM, and whole-cell lysates were obtained and subjected to Western blotting. The results evidenced a UTI89 CM-induced increase in p-Akt and p-mTOR signals compared with LB alone (Fig. S2B). Moreover, the Western blotting results showed a time-dependent increase of p-Akt by UTI89 (and UTI89⌬FimH) and their respective CM (Fig. S2C), supporting the conclusion that UPEC CM-found factor(s) act upon host BECs to activate specific signaling networks involved in bacterial invasion.
Activated Akt phosphorylates and activates mTORC1 (24). We used complementary pharmacologic and biologic reagents to parse the potential selective roles of mTORC1 and mTORC2 in UTI89 invasion of BECs. AZD8055 inhibits mTORC1 and mTORC2, whereas rapamycin (at low concentrations) selectively inhibits mTORC1. Treatment of BECs with AZD8055, but not rapamycin, inhibited phosphorylation of Akt in response to treatment with CM from UTI89 and UTI89⌬FimH ( Fig. 2A). Also, treatment with rapamycin inhibited phosphorylation of the mTORC1 substrate p70S6K ( Fig. 2A). Remarkably, AZD8055 significantly inhibited UTI89 invasion (Ͼ75%) compared with nontreated samples, whereas rapamycin did not significantly increase invasion (Fig. 2B), suggesting that mTORC2 facilitates the bacterial invasion of BECs. The results may also imply that inactivation of mTORC1 promotes bacterial invasion by increasing PI3K flux to mTORC2 and Akt. Neither AZD8055 nor rapamycin affected UTI89 attachment to BECs, as measured by counting GFP-UTI89 associated with BECs (Fig. S3).
The mTOR protein is a unit of the multiprotein complexes of mTORC1 and mTORC2, and we knocked down expression of mTOR using shRNA targeting sequences (Fig. 2C). Depletion of mTOR expression using two shRNAs reduced UPEC invasion of BECs by 75% (Fig. 2D). Notably, knockdown of mTOR protein expression effectively inhibited UTI89 CM-induced phosphorylation of p70S6K (Fig. 2E). To add support to the conclusion that mTORC2, but not mTORC1, promotes UPEC invasion (Fig. 2B), we knocked down expression of the mTORC1-and mTORC2-specific components Raptor (Fig. 2F) and Rictor (Fig. 2G), respectively. Knockdown of Raptor decreased UTI89 CM-mediated phosphorylation of p70S6K, but not appreciably Akt (Fig. 2F), and increased bacterial invasion by 2-fold (Fig. 2H). On the other hand, knockdown of Rictor inhibited UTI89 CM-induced p70S6K and Akt phosphorylation ( Fig. 2G) and, consistent with the pharmacologic reagents results (Fig. 2B), decreased bacterial invasion by about 50% (Fig. 2H). Neither knockdown of Raptor nor Rictor had any effect on UTI89 attachment to BECs (Fig. 2I). Collectively, these results demonstrate a requirement for mTORC2, but not mTORC1, in UTI89 entry but not attachment to BECs.

Activated EGFR supports UTI89 invasion
Our results suggest that PI3K, Akt, and mTORC2 facilitate bacterial invasion, but how these intracellular enzymes become activated in response to extracellularly found UPEC (UTI89⌬FimH) and bacterial CM remains unclear. Lipopolysaccharide (LPS) is the most studied UPEC virulence factor and has been implicated in a variety of host cell responses (27). To test for a potential role in UPEC invasion, we quantified the effect of purified LPS on Akt phosphorylation and could show a (high) dose-dependent increase in p-Akt (Fig. S4A). LPS is shed from bacteria and impacts target cells by binding to Toll-like receptors (TLRs). Our results show that treatment of BECs with the TLR4 inhibitor CLI-095 obliterated LPS-induced p-Akt (Fig. S4B). However, treatment with CLI-095 ( Fig. S4C), like knockdown of TLR4 expression with siRNA ( Fig. S4D), failed to reduce UTI89 invasion. Admittedly, LPS may impact host cells by binding other TLRs. Nonetheless, our results imply that UPEC invade BECs in a TLR4-independent manner. In agreement, TLR4 was recently reported to rather promote the expulsion of intracellular UPEC from invaded BECs (28,29).
Available evidence shows that growth factors, acting upon cognate plasma membrane-anchored receptors, activate PI3K and mTORC2 in a variety of cell types (24). The most studied of the growth factor receptors are the tyrosine kinase family, which undergo autophosphorylation-dependent activation upon ligand binding (30,31). We examined effect of UPEC on the landscape of BEC protein-tyrosine phosphorylation and observed a substantial increase in a prominent band that migrated to an apparent molecular mass of 170 kDa on SDS-PAGE. Epidermal growth factor receptor (EGFR) migrates at 170 kDa and has been implicated in the invasion of bacteria and viruses (32)(33)(34)(35)(36). BECs were exposed to CM from UTI89 and UTI89⌬FimH, and cell lysates were blotted using an antibody that recognizes autophosphorylated EGFR on tyrosine residue 1068 (p-EGFR), which denotes activation (30,31). The results showed a rapid and sustained increase of p-EGFR (up to 3-fold) following exposure to CM from both UTI89 and UTI89⌬FimH bacteria (Fig. 3A). We used the specific EGFR inhibitors AG1478 and afatinib to directly implicate it in the bacterially regulated Akt activation that is required for invasion. Treatment of BECs with AG1478 or afatinib (but not with LY294002, wortmannin, Bay 80-6946, MK-2206, rapamycin, or AZD8055, which act on PI3K, Akt, or mTORC) inhibited the UTI89 CM-mediated p-EGFR (Fig. 3B) and p-Akt (Fig. 3C), placing EGFR upstream of Akt activation. To address whether interference of EGFR signaling affects UPEC invasion, BECs were treated with AG1478 or afatinib. The results showed a 65-75% decrease in UTI89 invasion following EGFR signal blockade (Fig. 3D). Further support for the conclusion that EGFR mediates UPEC-induced p-Akt and bacterial invasion was found using an EGFR knockdown strategy with two shRNA sequences ( Fig. 3E). Knockdown of EGFR inhibited UTI89 invasion by 75-85% ( Fig. 3F) but had no effect on the attachment to target BECs (Fig. 3G).
To provide direct evidence for UTI89 CM-contained factor(s) in bacterial invasion, we performed an add-mix experiment where equal numbers of UTI89 bacteria were pelleted by centrifugation and then resuspended in equal volumes of fresh LB or original CM. The results showed that UTI89 invaded 60 -70% more when resuspended in CM compared with fresh LB (Fig. 4A). UTI89 CM promotes EGFR phosphorylation (Fig.  3), and to support the idea that EGFR facilitates bacterial invasion, we repeated the add-mix experiment in the presence of EGF (in LB) or AG1478 (in CM and in LB plus EGF  , followed by exposure to CM from UTI89 or UTI89⌬FimH for 10 min. Cell lysates were analyzed by Western blotting for phosphorylation of the indicated proteins. NI, not infected; NT, infected but not treated with inhibitor. B, AZD8055, but not rapamycin, inhibits UTI89 invasion of BECs. Statistical analysis was performed by Student's t test compared with the nontreated control. ***, p Ͻ 0.001; n ϭ 6. C, knockdown of mTOR. Lentiviruses encoding two different shRNA sequences targeting mTOR were used. The LKO lentivirus (shCon) was used as a control. Stable gene knockdown was achieved with selection using puromycin. mTOR gene expression was measured with quantitative RT-PCR, and results are presented as -fold difference relative to the shCon sample. Statistical analysis was performed by Student's t test compared with the LKO lentivirus control. ***, p Ͻ 0.001; n ϭ 3. D, knockdown of mTOR inhibits UTI89 invasion. Statistical analysis was performed by Student's t test compared with the LKO lentivirus control. ***, p Ͻ 0.001; n ϭ 6. E, mTOR knockdown decreases p70S6K phosphorylation. HSP90␤ was used as a protein loading control. F and G, effect of Raptor (F) and Rictor (G) knockdown on Akt and p70S6K phosphorylation. Cells were treated or not with UTI89 CM for 10 min, and lysates were analyzed by Western blotting for Akt (Ser-473) and p70S6K (Thr-389) phosphorylation. Total Akt, p70S6K, and GAPDH were used as protein loading controls. H, knockdown of Rictor inhibits UTI89 invasion. Bacterial invasion was determined using the gentamicin protection assay, and results are shown as -fold change from control (Con) cells. Statistical analysis was performed with Student's t test compared with the LKO lentivirus control. ***, p Ͻ 0.001; **, p Ͻ 0.01; n ϭ 6. I, Raptor and Rictor do not impact UTI89 attachment to BECs. GFP-UTI89 attachment to host BECs was quantified under fluorescence microscopy. The results show the number of attached UTI89 per cell, and 50 randomly selected cells were counted from each group. Statistical analysis was performed by Student's t test compared with the LKO lentivirus control.

Multistep bacterial invasion
showed that pretreatment with EGF (in LB) increased bacterial invasion by 2-fold compared with control LB alone (Fig. 4A).
The results also showed that inhibition of EGFR signaling by AG1478 obliterated the EGF-and CM-induced increases in UTI89 invasion (Fig. 4A). These findings lead us to conclude that UTI89 CM contains invasion-stimulating factor(s) that (activate EGFR and) prime host cells for bacterial invasion. Inhibition of EGFR (Fig. 3) or mTORC2 (Fig. 2) dramatically reduces UTI89 invasion. We explored whether co-inhibition of these mediators can better attenuate UTI89 invasion. We observed that co-treatment with (low doses of) AG1478 and AZD8055 was significantly more effective at inhibiting (about 80%) bacterial invasion than either reagent alone (Fig. 4B). These observations provide support for placing EGFR and mTORC2 in the cascade of UPEC invasion.

EGFR and mTORC2 regulate UTI89 invasion of mouse urothelium
Next, we wished to validate EGFR and mTORC2 as mediators of UPEC invasion of actual bladder epithelium to assess their potential as drug targets to more effectively manage persistent and chronic UTIs. We isolated mouse bladder transitional epithelium that contains superficial umbrella cells and

Multistep bacterial invasion
underlying intermediate and basal cells, but not smooth muscle or connective tissue, for use in an ex vivo invasion assay (Fig.  S5). The results validated the biologic activity of these reagents in mouse bladder epithelium tissue (Fig. 5A); treatment with AG1478 inhibited UTI89-induced p-EGFR and p-Akt, LY294002 inhibited p-Akt but not p-EGFR, AZD8055 inhibited p-Akt, and rapamycin increased p-Akt (Thr-308). The results also validated the in vitro findings that interference of EGFR (AG1478), PI3K-Akt (LY294002), and mTORC2 (AZD8055) but not mTORC1 (rapamycin) signaling inhibited UTI89 invasion of mouse urothelium (Fig. 5B). Notably, these chemicals had no effect on the bacterial growth rate in culture (Fig. S6). Also, imaging using GFP-UTI89 evidenced a lack of effect of the drugs on the attachment of bacteria to mouse urothelium. To evaluate contribution of EGFR and mTORC2 to actual bladder infection, we used a mouse UTI model. Female C3H/ HeJ mice were pretreated for 1 h with afatinib (an EGFR inhibitor) or AZD8055 (an mTORC2 inhibitor) by transurethral injection, which was followed by injection of 10 8 cfu of UTI89 (together with inhibitors) into the bladder for a total of 24 h. Following sacrifice, the bladders were harvested and subjected to bacterial invasion assays. The results showed that treatment with afatinib and AZD8055 reduced UTI89 infection by 76% (2.9 ϫ 10 4 cfu) and 68% (3.9 ϫ 10 4 cfu), respectively, compared with vehicle alone-treated mice (1.2 ϫ 10 5 cfu) (Fig. 5C). These data confirmed a role for EGFR and mTORC2 in UPEC invasion and identified these host cell mediators as possible targets for inhibiting UPEC invasion of the bladder.

Discussion
It is well-accepted now that UPEC invade host bladder epithelial cells, which may be accomplished through phagocytosis, via induction of actin polymerization, and/or pinocytosis of caveolae and clathrin-coated pits assembled on the plasma membrane. A compendium of the bacterial and host cell mol-ecules involved in the bacterial invasion remains incomplete.
Here we present a previously unrecognized mechanism of UPEC invasion that is initiated by bacterially secreted factor(s) that activate host cell signaling mediators in a manner that is independent of the bacterial binding to host cells to positively regulate invasion.
Invasive bacterial pathogens employ a variety of mechanisms to mediate their effects on host cells, including a type III secretion system for delivery of bacterial factors to target cells (37) and secreted toxins that act as ligands of host cell receptors (27). In the case of UPEC, the type 1 pilus-associated adhesin FimH binds to host cell uroplakin III proteins (38), and FimH has been reported to be both necessary and sufficient to promote bacterial invasion (9). Binding of FimH to the host cell membrane induced actin polymerization and activation of focal adhesion kinase (FAK), which, in turn, formed a complex with PI3K and presumably activated it, and both FAK and PI3K have been reported to be necessary for UPEC invasion (9). Notably, isogenic UPEC with deleted FimH did not promote FAK activation or formation of a FAK-PI3K complex (9). In contrast, our results show that UTI89 and UTI89⌬FimH equally activate PI3K, Akt, and mTORC2 signals, which are critical for successful bacterial invasion. The results also show that CM from UTI89 and UTI89⌬FimH similarly activate PI3K, Akt, and mTORC2. These findings expand the one-step classical model of direct UPEC binding to host cells and consequent activation of signal mediators involved in invasion. Rather, we propose that activated host cell signaling networks involved in productive UPEC invasion are dependent on a two-step mechanism. First, bacterial CM-containing factor(s) activate the host cell endocytic machinery, and, second, UPEC attach to cells primed for endocytosis, leading to their transport across the plasma membrane to the cytosol.
Many bacterial types have evolved the ability to bind to and invade host cells. For example, Neisseria meningitidis binds

Multistep bacterial invasion
host epithelial cells, leading to activation of EGFR and consequent invasion (33). Our results show that UTI89 and UTI89⌬FimH equally activate host cell EGFR and that the bacteria do not directly bind to EGFR. Knockdown of endogenous EGFR had no measurable effect on the attachment of UPEC to host BECs. Moreover, CM isolated from UTI89 and UTI89⌬FimH activated EGFR, excluding FimH as a direct mediator of EGFR activation. Our findings also exclude LPS and TLR4 in UPEC invasion of BECs. Rather, the results suggest that CM contains (a) so far unknown EGFR-activating factor(s). Notably, the bacterially secreted factor(s) is/are not produced by the E. coli K-12 strain and is susceptible to heat and proteinase K. 5 Bioinformatics analysis of the UPEC genome did not identify any encoded protein with homology to human EGF, implying that UPEC do not produce EGF (or EGF-like) peptides to directly activate EGFR. How UTI89 and CM activate EGFR remains incompletely understood, and prior work has not demonstrated a role for EGFR in UPEC invasion or UTI. EGFR is activated in response to direct binding to EGF (or EGFlike), leading to dimerization and assembly of signaling networks on the autophosphorylated receptor. We and others (39,40) have reported that, in addition to being directly activated by EGF, EGFR may be transactivated in response to stimulation of other types of receptors, including the large family of G protein-coupled receptors (GPCRs). Here, stimulated GPCRs promote activation of matrix metalloproteinases that cleave inactive pro-EGF to produce active soluble EGFR ligands that, in turn, bind to and activate EGFR (39,40). Hence, it is reasonable to suggest that UPEC CM contains protease activity able to cleave pro-EGF on the extracellular side of the host BEC plasma membrane, leading to the release of active EGFR ligands. It is equally plausible that UPEC CM contains GPCR ligand agonists (41,42) that promote a GPCR 3 EGFR transactivation signal. Support for the existence of a CM-activatable host cell receptor comes from the kinetics profile for p-Akt; stimulation with CM promoted the time-dependent increase and then decrease to a basal level of p-Akt. The temporal p-Akt signal is suggestive of a receptor-regulated event where an activated receptor promotes p-Akt and then that signal is down-regulated rapidly by receptor desensitization (43). UTI89 CM promotes p-EGFR and p-Akt and increases bacterial invasion. We reasoned that mutation of UPEC genes whose protein products are secreted (or participate in the secretory cascade) provides an unbiased genetic approach to identify the bacterially secreted factor(s) responsible for p-EGFR and UPEC invasion. We constructed a mini-TnPhoA transposon and plasmid delivery system and obtained 300 clones with insertion mutations in UTI89. Our preliminary screens identified 35 bacterial strains that were at least 50% less invasive than the parental UTI89. Our ongoing work aims to identify these bacterial genes and directly implicate them in UPEC invasion.
Vesicles trafficking from the plasma membrane to the cytosol serve as portals to shuttle UPEC from the extracellular milieu to inside target cells, and the internalized bacteria have been reported to encounter one of several outcomes, including degradation in the lysosome, expulsion to the extracellular environment for clearance by the innate immune machinery, acute replication as an intracellular bacterial community, or establishment of persistent quiescent intracellular reservoirs (10,28,44). The determinants responsible for the selected fate of internalized UPEC are not clear, and several vesicle types have been implicated in the invasion of UPEC into BECs. For example, UPEC internalize through phagocytosis, and the newly formed phagosomes fuse with lysosomes that have antibacterial activities, leading to the degradation of invading bacteria. Emerging evidence shows that cargo determines the intracellular trafficking route of vesicles (45). UPEC transported in caveolae and clathrin-coated pits may sort into specific pathways that determine the UPEC fate of, for example, expulsion and clearance through the immune system or fusion with the plasma membrane for release to initiate a second wave of infection that could mediate the persistent and chronic UTI. Hence, a better understanding of the nature of the vesicles mediating UPEC internalization is expected to shed light on the UPEC fate of clearance or reinfection.
In summary, our results support a revised mechanism of UPEC invasion comprised of bacterially secreted factor(s) activating host cell signaling networks. These findings present new drug targets in the armamentarium to more effectively treat the ever-increasing cases of persistent and chronic UTIs. Targeting of host cell proteins involved in UPEC invasion is appealing, as it overcomes UPEC resistance because cellular genes have a much lower mutation rate than UPEC genes under antibiotic pressure.

Bacteria, cells, and reagents
Human clinical cystitis isolate bacterium UTI89 and its derivative UTI89⌬fimH were a kind gift from M. Mulvey (University of Utah, Salt Lake City, UT). Bacteria were routinely cultured in LB broth or on LB agar (Difco, Sparks, MD) at 37°C. Liquid cultures were grown at 37°C under static conditions overnight, which induces expression of type 1 pili (which express FimH). To isolate bacterial pellet, UPEC were centrifuged at 12,000 ϫ g for 20 min. To isolate CM, bacteria were centrifuged in a stepwise manner at 6,000, 12,000, and 300,000 rpm, followed by filtration through a 0.22-m filter (Fig. S2A). The human bladder epithelial cancer cell line 5637 (BEC) was obtained from the ATCC and was cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum at 37°C in a humidified atmosphere of 5% CO 2 . Reagents were obtained as follows: LPS from E. coli 055:B5 and complete protease and phosphatase inhibitors from Sigma Aldrich (St. Louis, MO); CLI095 from InvivoGen (San Diego, CA); wortmannin, LY294002, and rapamycin from Cell Signaling Technology (Boston, MA); AZD8055 from Santa Cruz Biotechnology (Santa Cruz, CA); BAY 80-6946, afatinib, and MK-2206 from Selleckchem (Houston, TX); AG1478 from Cayman Chemical (Ann Arbor, MI); and gentamicin from Gibco. Antibodies were obtained as follows: p-Akt (Ser-473), p-Akt (Thr-308), total Akt, p-mTOR (Ser-2448), total mTOR, p-p70S6K, GAPDH, 5 W.-J. Kim and Y. Daaka, unpublished data.

Multistep bacterial invasion
p-EGFR (Tyr-1068), and total EGFR from Cell Signaling Technology and Alexa Fluor 568 -conjugated phalloidin from Life Technologies.

Bacterial invasion assay
BECs were seeded into 24-well plates at 5.0 ϫ 10 4 cells. Attached confluent cells were cultured overnight in starvation medium (10 mM HEPES (pH 7.5) and 0.1% BSA in RPMI 1640). BECs were then treated with bacteria at a multiplicity of infection (m.o.i.) of 10, centrifuged to synchronize infection, and incubated at 37°C for 1 h. Extracellular bacteria were killed by gentamicin-containing medium (100 g/ml) for 1 h, and cells were washed four times with PBS. Intracellular bacteria were recovered after hypotonic cell lysis (with PBS containing 0.4% Triton X-100). Dilutions (1:100) of the lysates were plated on LB agar, and colonies were allowed to grow overnight at 37°C and counted the next day. To determine the role PI3K/Akt, mTORC, and EGFR, specific inhibitors were used to pretreat the cells for 1 h prior to UPEC inoculation. BECs were then infected with bacteria, followed by determination of the colony number.

Gene silencing and quantitation
ON-TARGETplus SMARTpool siRNA specific for TLR4 and nontargeting controls were obtained from Dharmacon RNA Technologies (Lafayette, CO). BECs were transfected at 70% confluence with siRNA oligonucleotides (100 nM final concentration) using Lipofectamine RNAiMAX reagent (Invitrogen) and Opti-MEM. Stable knockdown of mTOR (RHS3979-9606083 and RHS3979 -9607174), Raptor (RHS3979 -9607161), Rictor (RHS3979 -97062588), and EGFR (RHS3979 -9607025) was achieved with shRNA (Open Biosystems) cloned in the lentiviral pLKO vector. Equal concentrations of pMD2G and PAX2 vectors were packaged in HEK293FT cells to propagate the virus. Lentivirus-containing medium was harvested, passed through a 0.22-m filter, and mixed with Polybrene (8 g/ml) to transduce BECs. The infected polyclonal cells were selected with 2 g/ml puromycin for 3 weeks before confirming gene knockdown. Total RNA was isolated using the HighPure RNA Isolation Kit (Roche), and mRNA was reverse-transcribed to complementary DNA using iScript Reverse Transcription Supermix (Bio-Rad). Gene expression levels were measured by iQ SYBR Green Supermix (Bio-Rad) and the iQ5 Thermal iCycler Detection System (Bio-Rad).

Western blot analysis
BECs seeded in 6-well plates were cultured in starvation medium overnight prior to exposure to bacteria (10 min). For inhibitor treatment, BECs were pretreated for 1 h prior to addition of CM or bacteria. For LPS experiments, BECs were exposed to purified LPS for 10 min with the indicated concentration. At the end of treatment, cells were washed and lysed in a modified radioimmune precipitation assay buffer (50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 2 mM EDTA, 1% NP-40, and 0.1% SDS) and fresh protease and phosphatase inhibitor mixture. Equal amounts of cell lysate (20 g protein) were analyzed by Western blotting using the indicated antibody.

Imaging studies
UTI89 and UTI89⌬FimH were made electrocompetent and transformed with a pHcRed-or GFP-encoding plasmid to generate HcRed-UTI89 and GFP-UTI89 bacteria. BECs were seeded on coverslips at a density of 1-2 ϫ 10 3 cells/well and transfected with a plasmid encoding GFP-Btk-PH (500 ng). Control cells were transfected with a plasmid encoding GFP-Btk. After 6 h of transfection, BECs were cultured in starvation medium overnight. BECs were then infected with the UTI89-HcRed or UTI89⌬FimH-HcRed bacteria at m.o.i. 30. Next, cells were washed three times with PBS and fixed for 10 min at room temperature with 4% formaldehyde solution. Slides were examined using a Leica confocal microscope (TCS SP5) equipped with a ϫ63/1.4-0.6 oil immersion lens.

Ethics statement
All experiments involving animals were performed according to the National Institutes of Health guidelines for housing and treatment of laboratory animals. Experiments were conducted using protocols preapproved by the University of Florida Institutional Animal Care and Use Committee.

Mouse bladder infection
C3H/HeJ female mice were obtained from The Jackson Laboratory. Mice were anesthetized and inoculated via transurethral catheterization. For drug treatment, anesthetized mice were injected transurethrally with 50 l of a 10 M solution of an EGFR inhibitor (afatinib) or mTORC2 inhibitor (AZD8055) in PBS 1 h before inoculation with UTI89 (10 8 cfu in 50 l of PBS) containing fresh inhibitors for a total of 24 h. At experiment termination, mice were killed, and bladders were harvested and processed for bacterial invasion. All analyses were performed in 8-to 10-week-old female mice (n ϭ 7 mice/experimental group).

Statistical analysis
Quantitative data are presented as mean Ϯ S.E. of the number of replicates or animals from at least three independent experiments. Significance was calculated by a two-tailed Student's t test or one-way ANOVA as indicated. Unpaired Student's t tests were used to determine statistical significance for in vivo experiments. p Ͻ 0.05 was considered statistically significant, and p values below 0.05, 0.01, and 0.001 are indicated in figures as *, **, and ***, respectively.