Role of Host Type IA Phosphoinositide 3-Kinase Pathway Components in Invasin-Mediated Internalization of Yersinia enterocolitica

Abstract

Many bacterial pathogens subvert mammalian type IA phosphoinositide 3-kinase (PI3K) in order to induce their internalization into host cells. How PI3K promotes internalization is not well understood. Also unclear is whether type IA PI3K affects different pathogens through similar or distinct mechanisms. Here, we performed an RNA interference (RNAi)-based screen to identify components of the type IA PI3K pathway involved in invasin-mediated entry of Yersinia enterocolitica, an enteropathogen that causes enteritis and lymphadenitis. The 69 genes targeted encode known upstream regulators or downstream effectors of PI3K. A similar RNAi screen was previously performed with the food-borne bacterium Listeria monocytogenes. The results of the screen with Y. enterocolitica indicate that at least nine members of the PI3K pathway are needed for invasin-mediated entry. Several of these proteins, including centaurin-α1, Dock180, focal adhesion kinase (FAK), Grp1, LL5α, LL5β, and PLD2 (phospholipase D2), were recruited to sites of entry. In addition, centaurin-α1, FAK, PLD2, and mTOR were required for remodeling of the actin cytoskeleton during entry. Six of the human proteins affecting invasin-dependent internalization also promote InlB-mediated entry of L. monocytogenes. Our results identify several host proteins that mediate invasin-induced effects on the actin cytoskeleton and indicate that a subset of PI3K pathway components promote internalization of both Y. enterocolitica and L. monocytogenes.

INTRODUCTION

Many microbial pathogens exploit host signal transduction pathways in order to cause disease (1–4). One key pathway subverted by bacterial, viral, and protozoan pathogens involves a mammalian lipid kinase called type IA phosphoinositide 3-kinase (PI3K) (3). Type IA PI3K plays a critical role in internalization of several pathogens into host cells. These microbes include bacteria that cause anthrax (Bacillus anthracis) (5), respiratory infections (Pseudomonas aeruginosa and Chlamydia pneumoniae) (6, 7), and food-borne disease (Campylobacter jejuni and Listeria monocytogenes) (3, 4, 8). Type IA PI3K also promotes entry of Ebola virus (9), influenza A virus (10), and parasites causing Chagas' disease (Trypanosoma cruzi) (11) or toxoplasmosis (Toxoplasma gondii) (12). Overall, the mechanism by which this mammalian lipid kinase controls infection by these diverse pathogens is not well understood.

Mammalian type IA PI3K is an heterodimeric enzyme composed of a 110-kDa catalytic subunit and a 85-kDa regulatory subunit (13). This PI3K is coupled to growth factor, cytokine, or cell adhesion receptors and controls a variety of processes, including cell growth, survival, and motility (13, 14). Type IA PI3K promotes its biological effects through at least two mechanisms, the best understood of which involves lipid kinase activity. This PI3K produces phosphatidylinositol 3,4,5-trisphosphate [PI(3,4,5)P₃], a lipid second messenger. PI(3,4,5)P₃ is converted by phosphatases to phosphatidylinositol 3,4-bisphosphate [PI(3,4)P₂]. Together, PI(3,4,5)P₃ and PI(3,4)P₂ bind approximately 40 downstream “target” proteins, recruiting these proteins to the plasma membrane, where they exert their biological activities (13, 15). Apart from its lipid kinase activity, type IA PI3K also regulates signaling through protein-protein interactions (16–19).

In order to understand how class IA PI3K promotes entry of microbial pathogens, it is essential to identify human proteins that act upstream and downstream of this kinase to control microbial internalization. Many bacteria induce their own uptake into human cells by stimulating remodeling of the host actin cytoskeleton and plasma membrane (2, 3, 20). Since several PI(3,4,5)P₃ and/or PI(3,4)P₂ binding proteins control actin polymerization and membrane trafficking (14, 15, 21, 22), identifying PI3K pathway components important for infection may further understanding of how bacteria stimulate cytoskeletal and membrane dynamics. In addition, given the diverse cellular functions of type IA PI3K (13–15), a comprehensive study of the roles of individual components of the PI3K pathway in bacterial entry may reveal host physiological processes that were not previously appreciated to control infection.

To date, the only microbial pathogen that has been subjected to a systematic and comprehensive analysis of the role of the type IA PI3K pathway in infection is the food-borne bacterium Listeria monocytogenes. One of the major pathways of internalization of L. monocytogenes into human cells is mediated by interaction of the bacterial surface protein InlB with its host receptor, the Met receptor tyrosine kinase (23). Type IA PI3K is activated by L. monocytogenes and required for InlB-mediated entry (24–26). One of the major roles of PI3K in L. monocytogenes internalization is to stimulate the localized assembly of host actin filaments, an event thought to remodel the plasma membrane during bacterial uptake (26). In order to investigate how the type IA PI3K pathway controls InlB-mediated entry, an RNA interference (RNAi)-based genetic screen was performed (27). This screen targeted human genes encoding proteins known to bind PI(3,4,5)P₃ and/or PI(3,4)P₂, proteins that interact with the p85 regulatory or p110 catalytic subunit of PI3K, or proteins indirectly regulated by PI3K. The study led to the identification of nine components of the human type IA PI3K pathway required for InlB-mediated entry of L. monocytogenes (27). One of these proteins, ARAP2, was demonstrated to have a crucial role in polymerization of host F-actin during bacterial entry through antagonism of the human GTPase Arf6 (28).

An important unresolved question is whether type IA PI3K promotes internalization of different microbial pathogens through similar or distinct mechanisms. One way of addressing this question is to perform RNAi-based screens similar to the screen done with L. monocytogenes. If several different microbial pathogens require the same type IA PI3K-regulated target proteins for entry, then uptake may involve a common mechanism. On the other hand, if a completely different profile of PI3K-regulated proteins participates in uptake of different pathogens, then entry mechanisms are likely distinct.

Yersinia enterocolitica is a Gram-negative, food-borne pathogen that causes acute gastrointestinal illnesses, including enteritis and lymphadenitis (29). Critical for disease is the ability of Y. enterocolitica to induce its internalization into antigen-sampling M cells in the intestine (30, 31). This internalization process has been extensively studied in cultured epithelial cell lines, including HEp-2 (32–34), COS-1 (35, 36), and HeLa (37–39). Y. enterocolitica and the related pathogen Yersinia pseudotuberculosis infect host cells through the binding of an outer membrane protein, termed invasin, to host cell β1 chain integrin receptors (31, 32, 40–44). Treatment of human cells with wortmannin, an inhibitor of PI3K activity (45), impairs invasin-mediated internalization of Y. enterocolitica (39) or Y. pseudotuberculosis (46). In addition, infection of host cells with an Escherichia coli strain expressing Y. pseudotuberculosis invasin stimulates phosphorylation of the host kinase Akt, a known downstream target of type IA PI3K (34). Finally, infection of human cells with Y. pseudotuberculosis induces recruitment of a fluorescently labeled probe for PI(3,4,5)P₃ (35). Taken together, these findings suggest an important role for host type IA PI3K in invasin-mediated entry. However, the mechanism by which this lipid kinase promotes infection is not well understood.

In this work, we performed an RNAi screen to identify components of the type IA PI3K pathway involved in invasin-mediated entry of Y. enterocolitica. A nearly identical profile of human proteins was targeted as in the screen previously performed with L. monocytogenes (27). Nine host proteins in the type IA PI3K pathway were found to be required for efficient invasin-mediated internalization. Several of these proteins were recruited to sites of entry and/or needed for changes in the F-actin cytoskeleton that accompany uptake. Interestingly, a comparison of the results from the screens with Y. enterocolitica and L. monocytogenes (27) identified six PI3K pathway components required for both invasin-mediated uptake of Y. enterocolitica and InlB-mediated entry of Listeria. Three PI3K pathway proteins were involved only in invasin-dependent internalization. Taken together, the findings identify several host proteins that participate in uptake of Y. enterocolitica and indicate that a subset of PI3K pathway components promote entry of different bacterial pathogens.

MATERIALS AND METHODS

Bacterial strains, mammalian cell lines, and media.

Escherichia coli strain HB101 and strain HB101 harboring a plasmid expressing the invasin gene from Y. enterocolitica (inv⁺ HB101) were gifts of B. Brett Finlay (University of British Columbia) and have been previously described (37, 40). Y. enterocolitica strain 8081c, generously provided by Virginia Miller (University of North Carolina at Chapel Hill), was generated by curing strain 8081v of its virulence plasmid (47). Strain 8081c contains the chromosomally encoded invasin gene (47). L. monocytogenes strain BUG 947 is isogenic with wild-type strain EGD, has an in-frame deletion in the inlA gene, and exhibits normal expression of InlB (48). Consequently, entry of BUG 947 into HeLa or other host cells is mediated by the Listeria surface protein InlB and is independent of the protein InlA (23, 26). For infection studies, inv⁺ HB101 was grown overnight at 37°C in Luria-Bertani (LB) broth containing 50 μg/ml ampicillin. Y. enterocolitica strain 8081c was cultured in LB broth at 26°C. The L. monocytogenes strain BUG 947 was grown in brain heart infusion (BHI; Difco) broth and prepared for infection as described previously (25).

The human epithelial cell line HeLa (ATTC CCL-2) was grown in Dulbecco's modified Eagle's medium (DMEM) with 4.5 g of glucose per liter and 2 mM glutamine (11995-065; Invitrogen), supplemented with 10% fetal bovine serum (FBS). Cell growth, cell stimulation, and bacterial infections were performed at 37°C in 5% CO2.

siRNAs.

Sequences of small interfering RNAs (siRNAs) used to target human genes comprising the host type IA PI3K pathway are listed in Table S3 in the supplemental material. These siRNAs were designed by and purchased from Sigma-Aldrich or Ambion. As a negative control, a nontargeting siRNA (catalog no. D-001210-01; Dharmacon) that contains two or more mismatches with all sequences in the human genome was used.

Antibodies, inhibitors, and other reagents.

Polyclonal antibodies used were anti-Akt (9272; Cell Signaling), anti-Cbl (sc-170; Santa Cruz Biotechnology), anti-focal adhesion kinase ([FAK] PTK2) (sc-558; Santa Cruz Biotechnology), anti-glutathione S-transferase (GST) (G7781; Sigma-Aldrich), anti-InlB (23), anti-LL5α (PHLDB1) (HPA038448; Sigma-Aldrich), anti-mTOR (2972; Cell Signaling), anti-phospho-Akt (Ser473) (9271; Cell Signaling), anti-phospholipase Cγ1 (PLC-γ1) (sc-81; Santa Cruz Biotechnology), anti-Rab5c (HPA003426; Sigma-Aldrich), and anti-Yersinia (01-90-04; KPL). The monoclonal antibodies used were anti-Dock180 (sc-13163; Santa Cruz Biotechnology), anti-glutathione S-transferase (GST) (G1160; Sigma-Aldrich), anti-hemagglutinin (HA) (clone 16B12, MMS-101P; Covance) anti-myc (clone 9E10, MMS-150P; Covance), and anti-α-tubulin (T5168; Sigma-Aldrich). Secondary antibody horseradish peroxidase (HRPO) conjugates were from Jackson Immunolabs. Secondary antibodies or phalloidin coupled to Alexa Fluor 488, Alexa Fluor 555, or Alexa Fluor 647 were purchased from Life Technologies. The PI3K inhibitor LY294002 was from Sigma-Aldrich.

Plasmids.

Mammalian expression vectors used were EGFPC1 (Clontech) or plasmids expressing enhanced green fluorescent protein (EGFP)-tagged phospholipase D2 (PLD2) (S. Bourgoin, University of Laval), EGFP-PSCD3 (Grp1) (S. Bourgouin), ECFP-Rab5c (J. Brumell, University of Toronto), EGFP-Dock180 (S. Grinstein, University of Toronto), EGFP-PTK2 (FAK) (50515; Addgene), EGFP-LL5β (PHLDB2) (A. Akhmanova, Utrecht University), myc-mTOR (1861; Addgene), and HA-CENTA1 (ADAP1) (V. Koronakis, University of Cambridge). A plasmid expressing EGFP-LL5α (PHLDB1) was constructed through PCR using Pfu DNA polymerase and the human cDNA KIAA0638 (Kazusa DNA Research Institute) as a template. Primers 5′-GCAGTCGACATGGACGCTCTCAATAGGAACCAA-3′ and 5′-TTCGGATCCCCAGGAGGCCCACGACAGTTA-3′ were used for amplification, and the resulting product was subcloned into the SalI and KpnI sites of EGFPC1. The final construct was verified by DNA sequencing. The plasmid pET28a-InlB used for expression of 6×His-tagged InlB protein was previously described (25). A plasmid used for expression of a glutathione S-transferase (GST) fusion protein containing the carboxyl-terminal 397 amino acids of invasin from Y. enterocolitica (GST.Inv397) was constructed by Gene Art AG. The Y. enterocolitica gene fragment was assembled from synthetic oligonucleotides and PCR products and subcloned into pGex4T-1 using EcoRI and XhoI sites. The final construct was verified by DNA sequencing.

Transfection of HeLa cells with siRNA.

HeLa cells were cultured, and transfections were performed in 24-well plates using 100 nM siRNA and the lipid reagent LF2000 (Invitrogen) exactly as described previously (27). Control conditions for the experiments shown in Fig. 2 to 6 and 8 and in Fig. S1 to S6 and S8 in the supplemental material involved mock transfection in the absence of siRNA or transfection with 100 nM nontargeting control siRNA.

FIG 2.

Effects of siRNAs against category I host genes on target gene expression and invasin-mediated entry. (A) Inhibition of host gene expression by siRNAs. HeLa cells were transfected with siRNAs targeting the indicated category I host genes. As controls, cells were transfected with a nontargeting control siRNA or were mock transfected in the absence of siRNAs (none). Approximately 48 h after addition of the siRNA, gene expression was analyzed by qPCR (i). Since the cbl qPCR probe did not detect expression, Cbl protein levels were assessed by Western blotting (ii). Data are means ± SEM. Statistical analysis by ANOVA gave P values between 0.019 and 0.0001. *, P < 0.05, relative to results with the no-siRNA (none) control (by the Tukey-Kramer posttest). (B) Impact of siRNAs on internalization of E. coli expressing invasin (inv⁺ HB101) into host cells. HeLa cells were subjected to the same transfection conditions as described for the experiment shown in panel A. Bacterial entry was assessed using gentamicin protection assays. Relative entry values are means ± SEM. Statistical analysis by ANOVA gave a P value of <0.0001. *, P < 0.05, relative to results with the no-siRNA (none) control (by the Tukey-Kramer posttest).

FIG 6.

Role of host type IA PI3K components in entry of Y. enterocolitica. HeLa cells were either mock transfected in the absence of siRNA, transfected with control nontargeting siRNA, or transfected with siRNAs targeting the indicated human genes. Cells then were infected with Y. enterocolitica strain 8081c, and entry was assessed. The targeting siRNAs used in these experiments correspond to the following siRNAs shown in Fig. 5: CENTA1 (siRNA 2), DOCK1 (1), FRAP1 (1), PHLDB1 (1), PHLDB2 (1), PLD2 (2), PSCD3 (3), PTK2 (1), and RAB5C (1). Results are means ± SEM. Statistical analysis by ANOVA gave a P value of <0.0001. *, P < 0.05, relative to results under the no-siRNA (none) or control siRNA condition (by the Tukey-Kramer posttest).

FIG 8.

Role of type IA PI3K components in remodeling of the actin cytoskeleton during invasin-mediated internalization. HeLa cells were either mock transfected in the absence of siRNA, transfected with control siRNA, or transfected with siRNAs targeting the indicated human genes. Transfected cells were incubated for 5 min with beads coupled to GST.Inv397. As a control, cells were incubated with beads coupled to GST alone. (A) Representative images of HeLa cells subjected to various siRNA conditions and incubated with beads coated with GST-Inv397. Areas of cells with beads indicated by arrows in the left panels are shown in an expanded view in the right panels. Note that siRNAs against CENTA1, FRAP1, PLD2, or PTK2 reduced recruitment of F-actin to beads coupled to GST.Inv397. (B). Representative image of cells treated with control siRNA and incubated with control beads coupled to GST. (C) Quantification of F-actin recruitment as fold enrichment (FE) values. Data are means ± SEM. Statistical analysis by ANOVA gave a P value of <0.0001. *, P < 0.05, relative to results under the no-siRNA (none) or control siRNA condition (by the Tukey-Kramer posttest).

qPCR analysis.

HeLa cells in 24-well plates were used for analysis of gene expression about 48 h after transfection with siRNA. Preparation of samples for quantitative PCR (qPCR) was performed as described previously (27), except that RNA was isolated using a PureLink RNA Micro Scale kit (Ambion), and cDNA synthesis was carried out with a SuperScript Vilo kit (Invitrogen). Real-time PCR was performed in triplicate on each cDNA sample using an ABI7500 or ABI7900 instrument (Applied Biosystems). TaqMan gene expression assays (probes) used for each of the 69 target genes comprising the human type IA PI3K pathway are listed in Table S3 in the supplemental material. Each of the probes spans exon-exon junctions and should not detect genomic DNA. The glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene (gene expression assay Hs99999905_m1; Applied Biosystems) was used as an endogenous control. Threshold cycle (C_T) values for the 69 target genes ranged from 25 to 33 in the various experiments. C_T values for the GAPDH endogenous controls were typically between 20 and 22. Data were analyzed by the comparative C_T method, normalizing C_T values for target gene expression to those for GAPDH. Relative quantity (RQ) values were calculated by the formula RQ = 2^{−ΔΔ CT}. To obtain the relative expression values shown in Fig. 2A, 3A, 4A, and 5 and in Fig. S1C and S6E in the supplemental material, RQ values in a given experiment were normalized to the values in cells mock transfected in the absence of siRNA (no-siRNA condition). The data shown in Fig. 2 to 5 and in Fig. S1C and S6E in the supplemental material are means ± standard errors of the means (SEM) from three to nine independent experiments, depending on the gene and siRNA condition.

FIG 3.

Effects of siRNAs directed against category II host genes on target gene expression and internalization of invasin-mediated entry. (A) Inhibition of host gene expression. HeLa cells were transfected with siRNAs targeting the indicated category II host genes. Control transfection conditions and analysis of gene expression were performed as described in the legend for Fig. 2. Gene expression was analyzed by qPCR for all but one of the genes (i). In the case of PLCG1 (encoding PLC-γ1 protein), gene expression was assessed by Western blotting (ii). Data are means ± SEM. Statistical analysis by ANOVA gave P values between 0.0253 and 0.0001. *, P < 0.05, relative to results with the no-siRNA (none) control (by the Tukey-Kramer posttest). (B) Effects of siRNAs on entry of E. coli expressing invasin (inv⁺ HB101). HeLa cells were subjected to the same transfection conditions as described for panel A, followed by measurement of bacterial entry. Data are means ± SEM. Statistical analysis by ANOVA produced a P value of <0.0001. *, P < 0.05, relative to results with the no-siRNA (none) control (by the Tukey-Kramer posttest).

FIG 4.

Effects of siRNAs against category III host genes on target gene expression and invasin-mediated entry. (A) Inhibition of host gene expression. HeLa cells were transfected with siRNAs targeting the indicated category III host genes. Control transfection conditions were the same as those described in the legend for Fig. 2. Gene expression was measured using qPCR. Data are means ± SEM. Statistical analysis by ANOVA indicated P values between 0.012 and 0.0001. *, P < 0.05, relative to results with the no-siRNA (none) control (by the Tukey-Kramer posttest). (B) Impact of siRNAs on entry of E. coli expressing invasin (inv⁺ HB101). HeLa cells were subjected to the same transfection conditions described for panel A. Data are means ± SEM. Statistical analysis by ANOVA gave a P value of <0.0001. *, P < 0.05 relative to results for the no-siRNA (none) control (by the Tukey-Kramer posttest).

FIG 5.

Multiple siRNAs affecting target gene expression inhibit invasin-mediated internalization. HeLa cells were transfected with three different individual siRNAs targeting the human gene indicated in each panel (black bars). As controls, cells were either were mock transfected in the absence of siRNA (none) or transfected with a control nontargeting siRNA. Effects of siRNAs on host gene expression were assessed by qPCR (i). Data are means ± SEM. Statistical analysis by ANOVA gave P values between 0.0007 and 0.0001. *, P < 0.05, relative to results under the no-siRNA (none) or control siRNA condition (by the Tukey-Kramer posttest). Effects of targeting siRNAs on invasin-mediated entry were determined (ii). Data are means ± SEM. Statistical analysis by ANOVA gave a P value of <0.0001. *, P < 0.05, relative to results under the no-siRNA (none) or control siRNA condition (by the Tukey-Kramer posttest).

In the case of 15 out of the 69 human genes analyzed, expression could not be detected using qPCR and the available probe (see Tables S2 and S3 in the supplemental material). We considered that these genes were either not expressed in HeLa cells, expressed at levels below the limit of detection of real-time PCR, or possibly incapable of being detected because of a flaw in probe design.

Western blotting.

Approximately 48 h after transfection with siRNAs, HeLa cells were solubilized in radioimmunoprecipitation assay (RIPA) buffer (1% Triton X-100, 0.25% sodium deoxycholate, 0.05% SDS, 50 mM Tris-HCl [pH 7.5], 2 mM EDTA, 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride [PMSF], and 10 mg/liter each of aprotinin and leupeptin). Western blotting and detection using enhanced chemiluminescence (ECL) or ECL Plus reagents (GE Health Care) were performed as described previously (23). Film was used for detection of anti-phospho-Akt blots, whereas an Odyssey imaging system (Li-Cor Biosciences) was used for detection with all other antibodies.

Bacterial entry assays.

HeLa cells were used for bacterial infections approximately 48 h after transfection with siRNAs. Gentamicin protection assays to measure entry of inv⁺ HB101, Y. enterocolitica, or L. monocytogenes (see Fig. 2 to 6; see also Fig. S1 and S3 in the supplemental material) were performed by infecting cells for 1 h in the absence of gentamicin and then incubating the cells in DMEM with 100 μg/ml gentamicin (HB101 or Y. enterocolitica) or 20 μg/ml gentamicin (L. monocytogenes) for 2 h as described previously (23, 25). Multiplicities of infection (MOIs) were approximately 100:1 for inv⁺ E. coli, 50:1 for Y. enterocolitica, and 30:1 for L. monocytogenes. Entry experiments shown in Fig. 2 to 6 and in Fig. S1 and S3 were performed 3 to 10 times, depending on the particular experiment and conditions. Entry efficiencies were first expressed as the percentage of the bacterial inoculum that survived gentamicin treatment. To obtain relative entry values, absolute percent entry values in a given experiment were normalized to the value in cells subjected to control conditions. For the experiment shown in Fig. S1B in the supplemental material, the control condition was dimethyl sulfoxide (DMSO) treatment. In the case of all of other experiments, the control was mock transfection in the absence of siRNA.

Protein purification.

Recombinant 6×His-tagged InlB protein was expressed in E. coli strain BL21(λDE3) and purified as described previously (25). GST alone or GST fused to the carboxyl-terminal 397 amino acids of invasin from Y. enterocolitica was expressed in BL21(λDE3) at 24°C after induction with 0.1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) for 2 h. Purification using glutathione-Sepharose 4B beads (GE Healthcare) was performed essentially as described previously (49).

Measurement of bacterial lengths.

ImageJ (version 1.4g) software was used to measure the lengths of Y. enterocolitica strain 8081 or L. monocytogenes strain BUG 947 bacteria. Fixed samples used for the analysis consisted of HeLa cells infected with bacteria for 5 min. Bacteria were labeled with anti-Yersinia or -L. monocytogenes primary antibodies and secondary antibodies coupled to Alexa Fluor 647. The lengths of 50 bacteria were measured for each strain. The mean lengths ± standard deviations were 2.66 ± 0.43 μm for Y. enterocolitica and 2.73 ± 0.45 μm for L. monocytogenes.

Coupling of proteins to latex beads.

Carboxylate-modified latex beads 3 μm in diameter (catalog no. 09850; PolySciences) were coupled to InlB, GST.Inv397, or GST protein as described previously (26). Prior to coupling, GST proteins were eluted from glutathione-Sepharose 4B beads using reduced glutathione. The glutathione was then removed using Micro Bio-Spin 6 columns (Bio-Rad). Approximately 1 × 10⁹ beads were incubated with 0.50 mg of InlB, GST.Inv397, or GST protein, with rotation at room temperature for 3 h. Beads were washed four times in phosphate-buffered saline (PBS) with 1% bovine serum albumin (BSA) and stored at 4°C until use.

Measurement of internalization of beads coated with invasin or InlB.

HeLa cells grown on sterile 22- by 22-mm coverslips were transfected with siRNAs using Lipofectamine 2000 (Life Technologies) as described previously (26). About 48 h posttransfection, cells were incubated with GST.Inv397-coated beads using a 3:1 bead-to-cell ratio or with or InlB-coated beads using a 5:1 ratio. Incubation was for 30 min at 37°C in 5% CO₂, and cells were then fixed in PBS with 3% paraformaldehyde. Samples were labeled using a previously described approach that distinguishes extracellular or intracellular particles (26, 49). Labeling was performed with polyclonal anti-InlB antibodies or monoclonal or polyclonal anti-GST antibodies for particles coated with GST-invasin. Samples were analyzed using a Zeiss LSM 710 confocal microscope equipped with diode (405 nm), multiline argon (458/488/514 nm), helium-neon 1 (543 nm), and helium-neon 2 (633 nm) lasers. The data shown in Fig. S6B to E in the supplemental material are from three experiments in each case. In each experiment, at least 100 intracellular beads were scored for the no-siRNA (none) control condition. A similar number of total (intracellular plus extracellular) beads were analyzed for all other conditions. The data were initially expressed as the percentage of total cell-associated beads that were internalized. Results shown in Fig. S6C and E are relative internalization values obtained by normalizing percent internalization values to those of the no-siRNA (none) control.

Confocal microscopy analysis of recruitment of PI3K pathway components or F-actin.

For recruitment studies of type IA PI3K pathway components (see Fig. 7; see also Fig. S7 in the supplemental material), approximately 2 × 10⁵ HeLa cells were seeded on sterile 22- by 22-mm coverslips in six-well plates. About 24 h after seeding, cells were transfected with the plasmids indicated in Fig. 7 or Fig. S7 in the supplemental material, using Lipofectamine 2000 as described previously (26). Approximately 24 h after addition of plasmid DNA, cells were incubated with beads coated with GST.Inv397, InlB, or GST. The ratios of particles to human cells were approximately 3:1 for GST.Inv397-coated beads or 5:1 for InlB-coated beads. Cells were centrifuged at 1,000 rpm for 2 min at room temperature to enhance contact between particles and HeLa cells and then incubated at 37°C in 5% CO₂ for the times indicated in the legends to Fig. 7 and S7. Samples were fixed in PBS containing 3% paraformaldehyde. For data in shown Fig. 7A and in Fig. S7A in the supplemental material, extracellular beads were labeled before cell permeabilization using anti-GST or anti-InlB antibodies. In the case of data with GFP-Rab5c shown in Fig. 7B and S7B, both intracellular and extracellular beads were labeled as described previously (26). Anti-rabbit or anti-mouse Alexa Fluor 647 was used as a secondary antibody. In the case of cells transfected with EGFP-tagged proteins, samples were immediately mounted in Mowiol. For cells transfected with HA-tagged centaurin-α1 or myc-tagged mTOR, samples were permeabilized in PBS with 0.40% Triton X-100, followed by incubation with anti-HA or anti-myc antibodies and Alexa Fluor 488-conjugated secondary antibodies.

FIG 7.

Recruitment of host type IA PI3K components during invasin-mediated entry. HeLa cells were transfected with plasmids expressing the indicated HA- or GFP-tagged mammalian proteins. Names of genes encoding these proteins are shown in parentheses. As controls, cells were transfected with plasmids expressing HA-tagged luciferase or GFP alone. After transfection, cells were incubated with latex beads coated with a GST fusion protein containing the last 397 amino acids of invasin (GST.Inv397). Incubation was for 10 min for analysis of FAK or mTOR recruitment or for 15 min for all other host proteins. Fixed samples were imaged by confocal microscopy. (A) Recruitment of host proteins to extracellular particles. Representative images of beads associating with HeLa cells are shown (i). Arrows indicate regions of interest that are expanded in the panels on the right. These expanded views contain merged images of beads and the relevant tagged protein or the tagged protein alone. Scale bar, 5 μm. Recruitment was quantified as fold enrichment (FE) values (ii). GFP was used as a control for myc-mTOR because a myc-tagged luciferase construct was expressed at levels that were too low. (B) Rab5c localization near extracellular or intracellular beads. Representative images of localization of GFP-Rab5c or GFP alone are shown (i). Intracellular beads or extracellular beads were differentially labeled. Arrows and the white asterisk indicate regions of interest with intracellular beads, whereas arrowheads indicate regions with extracellular beads. Quantification of recruitment of GFP-Rab5c or GFP near intracellular or extracellular beads is shown (ii). Data in panels A and B are means ± SEM. Statistical analysis by ANOVA gave a P value of <0.0001. *, P < 0.05, for results compared to those with HA-luciferase or a GFP control (by the Tukey-Kramer posttest).

For studies on recruitment of F-actin (see Fig. 8; see also Fig. S8 in the supplemental material), 8 × 10⁴ HeLa cells were seeded on sterile 22- by 22-mm coverslips in six-well plates. About 24 h after seeding, cells were transfected with siRNAs, using Lipofectamine 2000 as described previously (26). Approximately 48 h after transfection, cells were incubated with beads coated with GST.Inv397, InlB, or GST for 5 min, followed by fixation and labeling for confocal microscopy as described previously (26). Extracellular beads were labeled using anti-GST or anti-InlB antibodies before cell permeabilization. After permeabilization with PBS containing 0.40% Triton X-100, F-actin was labeled using phalloidin conjugated to Alexa Fluor 555.

The degree of recruitment of type IA PI3K pathway components or F-actin was measured using confocal microscopy. Images from serial sections spaced 1.0 μm apart were used to ensure that all cell-associated beads were detected. ImageJ (version 1.43g) software was employed to determine fold enrichment (FE) values for each cell-associated bead. FE is defined as the mean pixel intensity in a ring-like structure around the bead, normalized to the mean pixel intensity throughout the human cell. The thresholding function of ImageJ was used to select ring-like structures of PI3K pathway proteins or F-actin around beads. This function was also used in measuring mean pixel intensity throughout the cell. In each experiment, approximately 20 extracellular, cell-associated beads were analyzed for each condition. The data shown in Fig. 7 and 8 and in Fig. S7 and S8 in the supplemental material are means ± SEM from three experiments.

Statistical analysis.

Statistical analysis was performed using Prism (version 6.0c; GraphPad Software). In comparisons of data from three or more conditions, analysis of variance (ANOVA) was performed. The Tukey-Kramer test was used as a posttest. A P value of 0.05 or lower was considered significant.

RESULTS

Role of host type IA PI3K in invasin-mediated entry.

The primary goal of this study was to perform an RNAi-based screen to identify human type IA phosphoinositide 3-kinase (PI3K) pathway components involved in invasin-mediated entry of Yersinia enterocolitica. Before carrying out the screen, we used E. coli strain HB101 expressing the Y. enterocolitica invasin gene (inv) to confirm the previously demonstrated role for PI3K activity in invasin-mediated uptake (39, 46, 50). Consistent with previous reports (37), inv⁺ HB101 was internalized by the human epithelial cell line HeLa, whereas HB101 lacking inv was not (see Fig. S1A in the supplemental material). Importantly, the PI3K inhibitor LY249002 caused a dose-dependent inhibition in uptake of inv⁺ HB101 (see Fig. S1B). This inhibitor also impaired entry of Y. enterocolitica strain 8081 cured of its virulence plasmid. These findings suggest a role for one or more PI3K enzymes in invasin-mediated bacterial uptake.

Although LY294002 has been extensively used as general inhibitor of PI3K enzymes, this compound is not specific for type IA PI3K (51). We therefore used RNAi to determine the roles of the three type IA PI3K catalytic subunits, p110α, p110β, and p110δ (13), in invasin-mediated entry. We initially found that an siRNA targeting PIK3CA, the gene encoding p110α, impaired internalization of inv⁺ HB101 and plasmid-cured Y. enterocolitica (see Fig. S1C in the supplemental material). In contrast, siRNAs against PIK3CB or PIK3CD, encoding p110β or p110δ, respectively, did not affect entry of these bacterial strains. In order to exclude off-target effects of the PIK3CA siRNA, two additional siRNAs against PIK3CA were tested and determined to inhibit bacterial entry (see Fig. S1C). Interaction of p110α with the p85α or p85β regulatory subunit controls PI3K activity (13). We found that siRNAs targeting PIK3R2, the gene encoding p85β, reduced uptake of inv⁺ HB101 or Y. enterocolitica. In contrast, an siRNA against PIK3R1, encoding p85α, failed to affect entry in a statistically significant fashion. Taken together, the results shown in Fig. S1C indicate an important role for the p110α catalytic subunit and p85β regulatory subunit of type IA PI3K in invasin-mediated entry.

We also examined if invasin-mediated entry stimulates type IA PI3K activity. Importantly, infection of host cells with inv⁺ HB101 or Y. enterocolitica resulted in activation of type IA PI3K, as indicated by increased phosphorylation of the downstream PI3K effector Akt (13, 15) (see Fig. S1Di in the supplemental material). As a positive control, HeLa cells were treated with the L. monocytogenes protein InlB, a known agonist of type IA PI3K (25) (see Fig. S1Dii). Bacterium- or InlB-induced phosphorylation of Akt was impaired by treatment of cells with LY294002 (see Fig. S1D). Collectively, the results shown in Fig. S1 demonstrate an important role for type IA PI3K in invasin-mediated bacterial entry.

siRNA library targeting the type IA PI3K pathway.

In previous work, an siRNA library was used to identify components of the host type IA PI3K pathway that promote InlB-mediated entry of L. monocytogenes (27). This library targeted human genes encoding known or putative upstream regulators or downstream effectors of type IA PI3K. In the present study, we expanded the original siRNA library to include nine additional type IA PI3K pathway genes. The new library was directed against a total of 69 human genes which fell into three categories depending on their relationship to type IA PI3K and its lipid products (Fig. 1; see also Table S1 in the supplemental material). Category I was comprised of 15 genes encoding proteins known to physically interact with the PI3K p85 regulatory and/or p110 catalytic subunits. Some of these proteins, including Ras GTPases and PTK2/FAK, act upstream to control p110 catalytic activity (52, 53). Other proteins, such as Rab4 and Rab5 GTPases, have biochemical activities that are regulated by type IA PI3K (18). These proteins might act downstream of the kinase. Category II contained 40 human genes whose products bind directly to type I PI3K lipid products PI(3,4,5)P₃ and PI(3,4)P₂. Most of these proteins interact with phosphoinositides through one or more pleckstrin homology (PH), Phox homology (PX), or DHR1 domains (54) (see Table S1). Lipid binding is thought to recruit these proteins to the plasma membrane and/or modulate their biochemical activities (13, 15). Category III has 14 genes encoding proteins that are indirectly regulated by type IA PI3K (see Table S1). These proteins are substrates of Akt or PDK1, kinases directly regulated by PI(3,4)P₂ and PI(3,4,5)P₃ (55).

FIG 1.

Host type IA PI3K pathway components targeted in the RNAi screen. Invasin-mediated entry of Y. enterocolitica or InlB-mediated uptake of L. monocytogenes occurs through interaction with host β1 integrin or Met receptors, respectively (23, 31, 32, 44). In both of these entry pathways, type IA PI3K is activated downstream of host receptors and plays a critical role in bacterial internalization (23, 24, 26, 34, 39). Type IA PI3K uses PI(4,5)P₂ as a substrate and produces the lipid second messenger PI(3,4,5)P₃ (13, 15). PI(3,4)P₂ is another second messenger, which is generated from PI(3,4,5)P3 by phosphatases. The RNAi-based screen performed in this study targeted three categories of host genes encoding proteins of the type IA PI3K signaling pathway. Category I genes encode 15 proteins that interact with the 85-kDa regulatory and/or 110-kDa catalytic subunit of PI3K. Category II genes code for 40 proteins that bind to the PI3K lipid product PI(3,4,5)P₃ and/or PI(3,4)P₂. Category III genes encode 14 products that are indirectly controlled by type IA PI3K.

RNAi-based screen of invasin-mediated entry.

Before beginning the genetic screen, qPCR was used to confirm expression of the 69 PI3K pathway genes in HeLa cells. In the case of 15 of the 69 genes, expression under control conditions was not detected by qPCR using the available probes (see Table S2 in the supplemental material). However, expression of the products of two of these 15 genes, Cbl and PLC-γ1, was confirmed by Western blotting. Effective antibodies against the products of the remaining 13 genes were not commercially available. In total, 56 of the 69 human genes were verified to be expressed in HeLa cells.

These 56 human genes were then tested for inhibition in expression by RNAi. Single siRNAs against each gene (see Table S3 in the supplemental material) were examined for effects on target gene expression at 48 h posttransfection. Control conditions were mock transfection in the absence of siRNA or transfection with a control nontargeting siRNA that has two or more mismatches with all known human mRNA transcripts. Target gene expression was measured using qPCR or, in the case of Cbl or PLC-γ1, Western blotting. Of the 56 genes tested, 53 exhibited statistically significant reductions in expression due to targeting siRNAs (Fig. 2A, 3A, and 4A). In the case of three genes, expression was detected by qPCR, but the siRNAs employed did not reduce gene expression (see Table S2).

siRNAs that caused statistically significant reductions in gene expression were next assessed for effects on internalization of inv⁺ HB101. Growth and transfection of HeLa cells were performed exactly as in the gene expression studies. Gentamicin protection assays were used to measure bacterial internalization efficiencies (23) (see Materials and Methods). siRNAs against 11 of the targeted host genes caused statistically significant changes in invasin-mediated entry compared to results under the control conditions (Fig. 2B, 3B, and 4B). For nine of these genes, bacterial entry was inhibited, suggesting positive roles for the corresponding human proteins (see Table S4 in the supplemental material). In the case of the remaining two genes, targeting siRNAs increased bacterial internalization, implying negative roles for the corresponding gene product. siRNAs targeting these 11 genes did not affect bacterial adhesion to human cells (see Fig. S2A), indicating that the gene products were involved in postbinding steps. Taken together, the results shown in Fig. 2 to 4 suggest that several components of the human type IA PI3K pathway control invasin-mediated internalization.

RNAi study of InlB-mediated entry of L. monocytogenes.

The nine human type IA PI3K components that were not previously tested for roles in uptake of L. monocytogenes (27) are indicated with asterisks in Table S1 in the supplemental material. siRNAs targeting these nine genes were examined for effects on L. monocytogenes entry into HeLa cells (see Fig. S3A). siRNAs against two of these genes, PHLDB1 and PHLDB2, inhibited internalization of L. monocytogenes in a statistically significant fashion. In contrast, siRNAs targeting each of the three genes AKAP13, PHLPP1, or RAPH1 augmented bacterial entry. siRNAs found to impact entry were also examined for effects on adhesion of L. monocytogenes to HeLa cells (see Fig. S2B). Only an siRNA targeting RAPH1 increased bacterial adhesion. Collectively, the data shown in Fig. S3A and S2B in the supplemental material suggest that host PHLDB1 and PHLDB2 positively control uptake of L. monocytogenes, whereas AKAP13, PHLPP1, and RAPH1 may limit bacterial entry.

Addressing potential off-target effects of siRNAs.

For the type IA PI3K components that appeared to have positive roles in entry of inv⁺ HB101 or of L. monocytogenes, we performed additional RNAi experiments to address potential off-target effects of siRNAs. One common method of minimizing off-target effects is to verify that several different siRNA molecules recognizing distinct sequences in a given mRNA cause the same biological phenotype (56). Importantly, three different siRNAs directed against each of the nine human genes implicated in internalization of inv⁺ HB101 (Fig. 2 to 4) caused statistically significant reductions in target mRNA expression and bacterial entry (Fig. 5). In the case of five of the nine human genes, effective antibodies were available that allowed confirmation of depletion at the protein level (see Fig. S4 in the supplemental material). Taken together, the data shown in Fig. 5 and S4 indicate that off-target effects for the nine human genes are unlikely. The results are consistent with important roles for host CENTA1, DOCK1, FRAP1, PHLDB1, PHLDB2, PLD2, PSCD3, PTK2, and RAB5C in invasin-mediated entry (Fig. 5). The findings also indicate a critical function for PHLDB1 and PHLDB2 in uptake of L. monocytogenes (see Fig. S3Bi).

An important question is whether some type IA PI3K pathway components control both invasin- and InlB-mediated entry. In a previous RNAi-based screen with L. monocytogenes, experiments with multiple siRNAs demonstrated requirements for FRAP1 and RAB5C in InlB-dependent uptake (27). In that same study, single siRNAs against CENTA1 and PSCD3 impaired internalization of L. monocytogenes, suggesting that these host proteins might affect entry. However, roles in entry of these two genes were not verified by use of multiple siRNAs. Since CENTA1 and PSCD3 promoted invasin-mediated internalization (Fig. 5), we tested the effect of multiple siRNAs targeting these host genes on uptake of L. monocytogenes. The results confirm important functions for CENTA1 and PSCD3 in InlB-dependent entry (see Fig. S3Bii in the supplemental material). Based on results shown in Fig. 5 and S3Bii and on our previous work (27), we conclude that host CENTA1, FRAP1, PHLDB1, PHLDB2, PSCD3, and RAB5C are needed for both invasin- and InlB-mediated internalization.

Role of host PI3K pathway components in internalization of Y. enterocolitica.

We determined if host genes required for entry of inv⁺ HB101 (Fig. 5) also affect uptake of Y. enterocolitica. siRNAs targeting CENTA1, DOCK1, FRAP1, PHLDB1, PHLDB2, PLD2, PSCD3, PTK2, and RAB5C each caused statistically significant reductions in internalization of Y. enterocolitica strain 8081 (41) (Fig. 6). The siRNAs against these genes did not reduce cell growth or viability, as measured by MTT [3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide] assays (57) (see Fig. S5 in the supplemental material). We conclude that at least nine host PI3K pathway components are needed for efficient uptake of Y. enterocolitica.

Recruitment of host PI3K pathway components during invasin- or InlB-mediated entry.

We performed experiments to determine if any of the PI3K pathway components shown in Fig. 5 (or in Fig. S3B in the supplemental material) accumulate at sites of invasin- or InlB-dependent uptake. The protein names of these components, which sometimes differ from their cognate gene names, are listed in Table S4 in the supplemental material. Any observed accumulation or recruitment of these host proteins would imply active manipulation by invasin- or InlB-mediated pathways to allow internalization.

For these recruitment studies, we used inert particles coated with invasin or InlB. Latex beads coupled to truncated forms of invasin or full-length InlB have been previously employed as a model for entry of Y. enterocolitica, Y. pseudotuberculosis, or L. monocytogenes (26, 28, 33, 58–63). The efficient internalization of invasin- or InlB-coated beads makes these particles well suited for analysis of recruitment of host proteins (26, 28, 59, 60, 63). We selected beads with a diameter of 3.0 μm since the mean length of Y. enterocolitica or L. monocytogenes bacteria was each estimated as ∼2.7 μm (see Materials and Methods). Latex beads were coated with a GST fusion protein containing the carboxyl-terminal 397 amino acids of Y. enterocolitica invasin (GST.Inv397) or with full-length InlB (see Fig. S6A in the supplemental material). As expected, particles coupled to GST.Inv397 or InlB were efficiently internalized into HeLa cells, whereas beads coated with GST alone did not enter into these cells (see Fig. S6B). Importantly, incubation of HeLa cells with GST.Inv397-coated beads resulted in recruitment of centaurin-α1, Dock180, FAK, Grp1, LL5α, LL5β, mTOR, and PLD2 to extracellular, cell-associated particles (Fig. 7Ai). The degree of accumulation of these human proteins near beads was quantified as fold enrichment (FE) values. FE was defined as the mean fluorescence intensity in a ring-like structure around beads divided by the mean fluorescence intensity throughout the cell (28). An FE value greater than 1.0 indicates enrichment, implying recruitment. An FE value less than or equal to 1.0 is defined as lack of enrichment. FE numbers for HA- or GFP-tagged forms of centaurin-α1, Dock180, FAK, Grp1, LL5α, LL5β, mTOR, and PLD2 ranged between 1.4 and 2.6 (Fig. 7Aii). FE values for control proteins consisting of HA-tagged luciferase or GFP alone were 1.0 or less. Of the nine human proteins investigated, only Rab5c failed to be recruited to extracellular or intracellular GST.Inv397-coated beads (Fig. 7B).

In the case of InlB-dependent entry, extracellular beads associated with HeLa cells induced recruitment of centaurin-α1, Grp1, LL5α, LL5β, and mTOR (see Fig. S7A in the supplemental material). Interestingly, Rab5c did not accumulate around extracellular InlB-coated beads but, rather, localized exclusively to intracellular particles (see Fig. S7B). Collectively, the results shown in Fig. 7 and S7 demonstrate that several host proteins on the type IA PI3K pathway are recruited to sites of invasin- or InlB-mediated uptake.

Role of host PI3K pathway components in F-actin cytoskeletal changes during entry.

Many of the human genes identified in the RNAi screen (Fig. 5; see also Fig. S3 in the supplemental material) encode proteins known to regulate the actin cytoskeleton (see Table S5) (64–67). Since host actin polymerization is critical for invasin- and InlB-mediated internalization (24, 26, 68, 69), we examined if any of the genes shown in Fig. 5 are needed for F-actin cytoskeletal changes that accompany entry. Beads coated with GST.Inv397 or InlB have been effectively used as tools for examining F-actin recruitment (26, 28, 63). Before investigating actin accumulation near beads, we used RNAi to confirm that the various host genes needed for GST.Inv397- or InlB-mediated entry of bacteria (Fig. 5; see also Fig. S3) are also required for internalization of beads (see Fig. S6C). We also confirmed that the host actin cytoskeleton is required for entry of invasin- or InlB-coated particles. As expected, treatment of HeLa cells with the actin polymerization inhibitor cytochalasin D resulted in a marked decrease in uptake of beads coated with GST.Inv397 or InlB (see Fig. S6D). Actin polymerization during invasin- or InlB-mediated entry is mediated, at least in part, by the host Arp2/3 complex (58, 59, 63, 70). Importantly, internalization of GST.Inv397- or InlB-coated particles was also dependent on this host complex since siRNA-mediated depletion of Arp3 impaired uptake (see Fig. S6E in the supplemental material).

Having verified roles for actin and type IA PI3K pathway components in bead entry, we examined the requirements for these host components in invasin- or InlB-mediated actin cytoskeleton remodeling. As previously reported (26, 28, 63), F-actin accumulated in cup-like structures around GST.Inv397- or InlB-coated beads in HeLa cells subjected to mock transfection or treated with a control siRNA (Fig. 8; see also Fig. S8 in the supplemental material). Mean FE values for control conditions were 1.8 or 2.8 for GST.Inv397- or InlB-coated particles, respectively (Fig. 8C and S8C). siRNAs targeting CENTA1, FRAP1, PLD2, or PTK2 reduced FE values for F-actin near GST.Inv397-coated particles (Fig. 8C). These results demonstrate important roles for human centaurin-α1 (CENTA1), mTOR (FRAP1), PLD2, and FAK (PTK2) proteins in actin remodeling during invasin-mediated entry. In contrast, siRNAs against DOCK1, PHLDB1, PHLDB2, PSCD3, and RAB5C did not affect FE values, suggesting that these genes are dispensable for cytoskeletal changes induced by invasin. In the case of InlB-coated beads, RNAi results indicated that F-actin remodeling requires CENTA1, PHLDB1, and PSCD3 but not FRAP1, PHLDB2, or RAB5C (see Fig. S8). In summary, the results shown in Fig. 8 and S8 identify several human proteins required for actin accumulation induced by invasin or InlB. Interestingly, only one host protein, centaurin-α1, had a shared role in both invasin- and InlB-dependent actin cytoskeletal changes.

DISCUSSION

In this work, we used RNAi to identify nine members of the human type IA PI3K pathway that are needed for efficient entry mediated by the invasin protein of Y. enterocolitica (see Table S4 in the supplemental material). Of these nine host proteins, eight were recruited to invasin-coated particles, implying that their functions are actively exploited to allow internalization. Four of the nine proteins, centaurin-α1, FAK, mTOR, and PLD2, were required for F-actin remodeling, a process essential for invasin-dependent uptake (68). In contrast, depletion of Dock180, Grp1, LL5α, or Rab5c failed to affect F-actin recruitment by invasin-coated beads. This lack of effect suggests that these proteins may control entry through processes apart from actin assembly.

Based on results in this study and previous work by Jiwani and coworkers (27), six of the nine human proteins involved in invasin-mediated entry of Y. enterocolitica are also needed for InlB-dependent internalization of L. monocytogenes (see Table S5 in the supplemental material). An interesting question is whether these six proteins promote invasin- and InlB-dependent entry through the same or distinct mechanisms. Importantly, only one of the six proteins, centaurin-α1, was required for actin assembly during both invasin- and InlB-mediated uptake (Fig. 8; see also Fig. S8). Conversely, two of the six proteins, LL5β (encoded by PHLDB2) and Rab5c, were dispensable for F-actin remodeling in both entry pathways. The host proteins LL5α (encoded by PHLDB1), mTOR (FRAP1), and Grp1 (PSCD3) exhibited different roles in actin rearrangements, depending on the entry route. Taken together, these results suggest that some of the host proteins involved in both invasin- and InlB-mediated uptake probably use distinct mechanisms to promote infection by Y. enterocolitica or L. monocytogenes. On the other hand, centaurin-α1, LL5β, and Rab5c might have similar modes of action. Clearly, more work is required to definitively address this issue. Compelling evidence for similar or distinct mechanisms will require investigation of biochemical activities, substrates, and protein-protein interactions used by centaurin-α1, mTOR, or Rab5c to facilitate the invasin or InlB entry pathways.

Some type IA PI3K components are required specifically for invasin- or InlB-mediated internalization (see Table S5 in the supplemental material). These findings are additional indications that mechanistic differences exist between two entry pathways. In some cases, the absence of a role could be due to a failure of the host protein to respond to engagement of the invasin or InlB receptor, β1 integrin or c-Met, respectively. However, this scenario is probably not true for FAK, ARAP2, or ARNO, as the literature indicates tyrosine phosphorylation of FAK downstream of c-Met (71) and effects of ARAP2 or ARNO on trafficking of β1 integrins in HeLa cells (72, 73). Below, we discuss what is known about the biological functions of the host proteins identified from our RNAi screen and describe possible ways in which these proteins might control invasin- or InlB-mediated entry.

Membrane remodeling during bacterial uptake.

Internalization of Yersinia spp. or L. monocytogenes requires host activities that remodel the plasma membrane, resulting in extension of pseudopods around adherent bacteria (20, 74, 75). To date, the main mechanism known to stimulate membrane remodeling during invasin- or InlB-mediated entry involves localized polymerization of the host actin cytoskeleton. In both entry pathways, actin polymerization is mediated, at least in part, by the human Arp2/3 complex and upstream regulators of this complex, including nucleation promoting factors (NPFs) and the GTPase Rac1 (58, 59, 61, 63). An important question is whether PI3K pathway components needed for F-actin recruitment during invasin- or InlB-dependent entry affect the Arp2/3 pathway or, instead, regulate the actin cytoskeleton through other mechanisms. Also of interest is how some PI3K pathway components might control bacterial entry independently of effects on the actin cytoskeleton. Below, we discuss the known biological functions of host proteins identified in the RNAi screen and describe possible ways that these proteins might control bacterial uptake.

Host proteins that participate only in invasin-mediated entry.

Three host proteins Dock180, FAK, and PLD2 were needed for invasin-dependent internalization but not for InlB-mediated uptake (see Table S4 in the supplemental material). Dock180 is a guanine nucleotide exchange factor (GEF) for Rac1 GTPase (67). Despite the fact that Rac1 activity is required for F-actin recruitment to invasin-coated beads (63), our results indicate that Dock180 is dispensable for actin accumulation. It is therefore possible that Dock180 cooperates with other GEFs to activate Rac1 during invasin-mediated entry.

FAK is a cytoplasmic tyrosine kinase found in focal adhesions, integrin-containing structures that link the actin cytoskeleton to the extracellular matrix (76). Our finding that FAK is needed for entry of Y. enterocolitica is consistent with previous results demonstrating a role for this kinase in internalization of Y. pseudotuberculosis (77). Importantly, we found that FAK is needed for F-actin accumulation near particles coated with invasin. FAK is known to enhance actin polymerization by binding directly to the Arp2/3 complex (78), and it is possible that this process contributes to invasion-dependent uptake.

PLD2 (phospholipase D2) is one of two mammalian PLD enzymes that convert phosphatidylcholine to phosphatidic acid (PA) (79, 80). PA stimulates membrane remodeling through several effector proteins, including phosphatidylinositol 4-phosphate 5 kinase (PI4P5K). This enzyme produces PI(4,5)P₂, a phosphoinositide that promotes membrane transport and F-actin assembly (14). In future work, it will be important to determine the role of PI4P5K in entry of Y. enterocolitica and also whether PLD2 stimulates accumulation of PI(4,5)P₂ during bacterial uptake.

Human proteins that promote only InlB-mediated internalization.

Based on this study and a previous RNAi-based screen (27), seven type IA PI3K components needed for InlB-mediated entry of L. monocytogenes are dispensable for invasin-dependent entry of Y. enterocolitica (see Table S5 in the supplemental material). Potential mechanisms by which these host proteins might control InlB-mediated uptake were previously discussed (27, 28).

Host proteins involved in both invasin- and InlB-mediated entry.

Uptake of both Y. enterocolitica and L. monocytogenes requires the host proteins Rab5c, centaurin-α1, Grp1, mTOR, LL5α, and LL5β (see Table S5 in the supplemental material). Rab5c is one of three mammalian Rab5 GTPases that promote the maturation of early endosomes or phagosomes (81, 82). We found that Rab5c localized exclusively to internalized InlB-coated particles. This localization differed from localizations of other type IA PI3K components, which were typically recruited to cup-like structures around the extracellular portion of beads. In addition, Rab5c was dispensable for invasin- or InlB-mediated accumulation of F-actin. Actin polymerization is thought to act early during bacterial entry to promote pseudopod extension (20, 74, 75). Collectively, our results suggest that Rab5c might localize to the nascent phagosome and control late events in InlB-mediated entry. In the case of invasin-dependent internalization, we failed to detect Rab5c around intracellular or extracellular particles. This lack of recruitment suggests that the role of Rab5c in invasin-mediated entry of Y. enterocolitica could be indirect. Specifically, invasin might exploit existing Rab5c activity without actively manipulating the localization of this GTPase. Alternatively, it is possible that Rab5c recruitment is too transient to detect in studies with fixed cells. It is worth mentioning that a recent study describes an important role for Rab5 GTPases in uptake of Y. pseudotuberculosis (83). Rab5 GTPases were found to be required for Y. pseudotuberculosis entry and to accumulate on nascent phagosomes. However, since the study employed antibodies that likely detect all Rab5 GTPases, the specific role of the Rab5c isoform in bacterial uptake was not addressed.

Centaurin-α1, also known as ADAP1, contains two PH domains and a GTPase activation protein (GAP) domain capable of acting on the GTPase Arf6 (64, 84). The first PH domain of centaurin-α1 binds PI(3,4,5)P₃, whereas the second PH domain interacts with both PI(3,4,5)P₃ and PI(3,4)P₂ (84, 85). These PH domains mediate translocation of centaurin-α1 from the cytoskeleton to the plasma membrane upon stimulation of cells with growth factors (64, 86). Our finding that centaurin-α1 is needed for accumulation of F-actin during invasin- or InlB-mediated uptake (see Table S5 in the supplemental material) is in general agreement with results indicating a role for this human protein's GAP activity in growth factor-induced remodeling of the actin cytoskeleton (64). Whether centaurin-α1 promotes entry of Y. enterocolitica or L. monocytogenes through regulation of Arf6 or other Arf family GTPases is currently unknown.

Grp1, also referred to as cytohesin-3, is a GEF that catalyzes nucleotide exchange on the Arf family GTPases Arf1, Arf5, and Arf6 (87, 88). Grp1 has a carboxyl-terminal PH domain that interacts specifically with PI(3,4,5P₃) and mediates recruitment to the plasma membrane in response to growth factor stimulation (89–91). Binding of PI(3,4,5)P₃ to the PH domain stimulates the GEF activity of Grp1, indicating that type IA PI3K can control Grp1 function (89). It is noteworthy that Grp1 is needed for actin cytoskeletal remodeling induced by InlB, but not by invasin (Fig. 8; see also Fig. S8 in the supplemental material). These results indicate that Grp1 likely uses distinct mechanisms to promote entry of L. monocytogenes or Y. enterocolitica, perhaps by activating different Arf family GTPases.

mTOR, encoded by FRAP1, is a serine/threonine kinase that acts downstream of type IA PI3K to control a plethora of biological processes, including translation, lipid synthesis, autophagy, and cytoskeletal organization (66). mTOR is present in two different complexes, mTORC1 or mTORC2. mTORC1 regulates synthesis of proteins or lipids, autophagy, lysosome biogenesis, and energy production. In contrast, mTORC2 controls cell survival and the actin cytoskeleton. The fact that mTOR has distinct roles in F-actin remodeling induced by invasin or InlB (Fig. 8; see also Fig. S8) raises the possibility that different mTOR complexes might control internalization of Y. enterocolitica or L. monocytogenes.

LL5α and the related protein LL5β are large scaffolding proteins with coiled-coil regions and a PH domain that interacts with PI(3,4,5)P₃ (92–96). Interestingly, our results indicate that LL5α is needed for actin filament remodeling during entry mediated by InlB, but not by invasin (Fig. 8; see also Fig. S8 in the supplemental material). In contrast, LL5β was dispensable for actin rearrangements induced by invasin or InlB. One possible explanation for actin-independent roles of LL5 proteins in bacterial entry could be related to the recognized functions of these proteins in organization of microtubules. Coiled-coil domains in LL5α or LL5β interact with CLASPs, proteins that bind the plus end of microtubules (94, 95). By simultaneously engaging PI(3,4,5)P₃ at the plasma membrane and CLASPs, LL5α and LL5β anchor microtubule plus ends to the cell cortex (94, 95). Although the purpose of such anchoring is not well understood, there is recent evidence to suggest that it may direct exocytosis to specific sites in the cell (93, 96). While a role for host exocytosis in uptake of Y. enterocolitica or L. monocytogenes has yet to be reported, this process could potentially contribute to remodeling of the plasma membrane during bacterial engulfment.

In summary, by screening components of the type IA PI3K pathway, we have identified several host proteins that are required for internalization of Y. enterocolitica or L. monocytogenes into human cells. Six host proteins have important roles in infection by both bacterial pathogens. However, when the effects of host proteins on the actin cytoskeleton are taken into account, our findings argue for some differences in the mechanisms of invasin- and InlB-mediated uptake. The Arf GAP centaurin-α1 might affect both entry pathways through similar means although future experiments are needed to definitively address this idea. It is noteworthy that several human proteins identified in the screen appear to control internalization independently of F-actin remodeling. Future work on these proteins could possibly identify host physiological processes that were not previously appreciated to control internalization of Y. enterocolitica, L. monocytogenes, or other bacterial pathogens.