Research ArticleHost-Pathogen Interactions

Shigella flexneri Infection Generates the Lipid PI5P to Alter Endocytosis and Prevent Termination of EGFR Signaling

See allHide authors and affiliations

Science Signaling  20 Sep 2011:
Vol. 4, Issue 191, pp. ra61
DOI: 10.1126/scisignal.2001619


The phosphoinositide metabolic pathway, which regulates cellular processes implicated in survival, motility, and trafficking, is often subverted by bacterial pathogens. Shigella flexneri, a bacterium that causes dysentery, injects IpgD, a phosphoinositide phosphatase that generates the lipid phosphatidylinositol 5-phosphate (PI5P), into host cells, thereby activating the phosphoinositide 3-kinase–Akt survival pathway. We show that epidermal growth factor receptor (EGFR) is required for PI5P-dependent activation of Akt in infected HeLa cells or cells ectopically expressing IpgD. Cells treated with PI5P had increased numbers of early endosomes with activated EGFR, no detectable EGFR in the late endosomal or lysosomal compartments, and prolonged EGFR signaling. Endosomal recycling and retrograde pathways were spared, indicating that the effect of PI5P on the degradative route to the late endocytic compartments was specific. Thus, we identified PI5P, which was enriched in endosomes, as a regulator of vesicular trafficking that alters growth factor receptor signaling by impairing lysosomal degradation, a property used by S. flexneri to favor survival of host cells.


Bacterial pathogens have evolved various strategies to subvert phosphoinositide metabolism in host cells to efficiently cope with a hostile environment (1, 2). Shigella flexneri, the causative agent of bacillary dysentery in humans, injects effector proteins into host cells through a type III secretion system to hijack the intracellular machinery and promote uptake and colonization (3). The S. flexneri effector IpgD is a phosphoinositide phosphatase that transforms a fraction of phosphatidylinositol 4,5-bisphosphate [PI(4,5)P2] in host cells into phosphatidylinositol 5-phosphate (PI5P) (4). PIs are considered to be critical players in the spatial and temporal organization of signaling pathways, actin cytoskeleton rearrangements, and vesicle trafficking (5, 6). Their metabolism is precisely controlled by kinases and phosphatases that target their 3-, 4-, or 5-hydroxyl groups (7), and mutations in several of these enzymes contribute to the development of human diseases such as myotubular myopathies, diabetes, and cancer (8, 9). Because of their ability to bind specific protein domains such as pleckstrin homology (PH), Fab1p/YOTB/vac1p/early endosome antigen 1 (EEA1) (FYVE), or Phox homology (PX) domains, PIs are recognized as essential integrators and regulators of various signals originating from membranes (5, 7, 10). Among the eight phosphoinositides, PI5P is the least characterized and its functions remain elusive (11, 12). Development of a specific mass assay (13) and imaging using the plant homeodomain (PHD) from inhibitor of growth 2 (ING2) (14) led several groups to suggest that PI5P may act as a second messenger (11, 12). The basal amount of PI5P is low in cells but can be increased by different stimuli such as thrombin in platelets (13), insulin in 3T3-L1 cells (15), T cell receptor engagement (16), increased tyrosine kinase activity (17, 18), and stress signals (19). In addition, an increase in PI5P was reported during the G1 phase of the cell cycle (20). Infection by the intracellular pathogens S. flexneri or Salmonella Typhimurium results in increased amounts of PI5P in host cells (4, 21). Several enzymes have been proposed to regulate the amount of PI5P, including the 5-kinase PIKfyve, which can phosphorylate phosphatidylinositol in vitro (22); the 3-phosphatases of the myotubularin family, which can hydrolyze phosphatidylinositol 3,5-bisphosphate [PI(3,5)P2] (23); type I and II PI(4,5)P2 4-phosphatases (24, 25); and type II PI5P 4-kinases, which specifically use PI5P as a substrate (13, 18, 19, 26).

A nuclear pool of PI5P interacts with ING2 and contributes to stress-induced acetylation of p53 (14, 19), but this lipid has also been detected in the plasma membrane as well as in internal membranes or vesicles (27). The greatest production of PI5P at the plasma membrane reported so far is induced by the S. flexneri effector IpgD, which transforms a considerable portion of the plasma membrane PI(4,5)P2 into PI5P (4, 28). IpgD expression (26, 28) or inactivation of type II PI5P 4-kinases (26, 29) leads to an increase in Akt activation. Stimulation of the phosphoinositide 3-kinase (PI3K) and Akt survival pathway by IpgD delays the onset of host cell apoptosis to promote better bacterial colonization (28). S. Typhimurium injects SigD and Salmonella Dublin injects SopB, which are potent phosphoinositide phosphatases that, like IpgD, produce PI5P (21), suggesting that this lipid may facilitate invasion of various intracellular pathogens. The model of S. flexneri infection with the effector IpgD provides an opportunity to decipher how this lipid can alter intracellular mechanisms.

Here, we demonstrate that the epidermal growth factor receptor (EGFR) is recruited to S. flexneri entry sites in an IpgD-dependent manner and is required for PI5P-mediated activation of Akt. Increased amounts of PI5P resulted in an accumulation of active EGFR on early endosomes, protecting it from degradation. Indeed, this lipid blocked the maturation process or the transport from early endosome to late endosome and lysosome compartments but spared the recycling and the retrograde pathways. Thus, PI5P is a key molecule in a strategy evolved by bacterial pathogens to subvert and exploit the endocytic machinery and growth factor receptor activation to increase the activity of the Akt survival pathway.


EGFR is activated in an IpgD-dependent manner and is required for PI5P-induced Akt activation

To identify the molecular mechanisms involved in IpgD- and PI5P-mediated Akt activation (28), we examined EGFR activation in HeLa cells infected with wild-type S. flexneri (M90T). Confocal analysis demonstrated that, after 10 min of infection, PI5P colocalized with activated EGFR (as indicated by phosphorylation of Tyr1173) at bacterial entry foci, characterized by actin rearrangement (Fig. 1A). EGFR was not activated by infection with an IpgD-deleted strain (ΔIpgD), which can still rearrange actin at entry sites, or a noninvasive mutant (BS176, which lacks the type III secretion system), suggesting that recruitment and activation of the receptor were dependent on IpgD and therefore PI5P production (Fig. 1A). Knockdown of EGFR by small interfering RNA (siRNA) abolished EGFR activation (Fig. 1B and fig. S2A). Western blot analysis confirmed that phosphorylation of EGFR at Tyr1173 occurred only upon infection with wild-type M90T strain (Fig. 1C and fig. S2B). Accordingly, ectopic expression of IpgD, but not that of the inactive IpgD-C438S mutant, induced phosphorylation of EGFR (Fig. 1D and fig. S2C). Transformation of similar amounts of PI(4,5)P2 into phosphatidylinositol 4-monophosphate (PI4P) by the yeast 4-phosphatase Inp54p (fig. S3) did not induce EGFR phosphorylation or Akt activation (Fig. 1D and fig. S2C). EGFR phosphorylation, induced either by S. flexneri infection (Fig. 1E) or by IpgD expression (Fig. 1F), was partially attenuated by transfection of the PI5P 4-kinase II, which converts PI5P back to PI(4,5)P2 (fig. S4B). Coexpression of the PI5P biosensor green fluorescent protein (GFP)–PHD3x, which sequesters PI5P and impairs downstream signaling (14), also reduced EGFR phosphorylation (Fig. 1F and fig. S2D). Moreover, we observed a dose-dependent activation of EGFR upon treatment with exogenous short-chain PI5P (Fig. 1G and fig. S1A). The effect of C4-PI5P on EGFR activation was reduced by expression of PI5P 4-kinase IIα (fig. S1, B and C). Under similar conditions, application of exogenous short-chain PI4P had no effect on EGFR activation (Fig. 1G and fig. S1A). Collectively, these data indicate that IpgD can induce EGFR activation and that PI5P is essential in this process.

Fig. 1

Activation of EGFR by IpgD and PI5P production. (A) Serum-starved HeLa cells were infected for 10 min with wild-type (WT) S. flexneri (M90T), the IpgD-deficient mutant (ΔIpgD), or a noninvasive strain (BS176) and then fixed. Actin was visualized with Alexa Fluor 488–phalloidin, PI5P with biotinylated GST-PHD2x, and Alexa Fluor 594–labeled streptavidin and phosphorylated EGFR at Tyr1173 (pEGFR Tyr1173) with anti-pEGFR Tyr1173 antibody. White arrows show entry foci sites. Right: line scan analysis of fluorescence intensity at entry sites (n = 3 experiments; 10 cells per experiment). Scale bars, 10 μm. (B) siRNA-mediated knockdown of EGFR abolishes pEGFR Tyr1173 signal at the entry foci sites of WT S. flexneri (M90T)–infected serum-starved cells (five sites per 10 cells on average in untreated or siRNA-treated cells; n = 3 experiments, 20 cells per experiment). (C) Lysates from serum-starved cells infected for the indicated times were immunoblotted for pEGFR Tyr1173 and total EGFR (n = 3 experiments). (D) Lysates from serum-starved HeLa cells transfected as indicated were probed with antibodies to pEGFR Tyr1173 or Tyr1068, total EGFR, Akt phosphorylated at Ser473, or total Akt (n = 3 experiments). (E) Cells were transfected with pEGFP-PI5P4KIIβ, infected with the WT (M90T) strain for 15 min, and serum-starved. Lysates were immunoblotted for pEGFR Tyr1173, EGFR, and GFP (n = 3 experiments). (F) Serum-starved HeLa Tet-Off IpgD–expressing cells were transfected as indicated, and lysates were immunoblotted for pEGFR Tyr1173, EGFR, mCherry, GFP, and myc (n = 3 experiments). (G) HeLa cells serum-starved overnight were incubated with C4-PI4P or C4-PI5P for the indicated times. Lysates were immunoblotted for pEGFR Tyr1173 and EGFR. Data are calculated as normalized integrated densities of the bands of pEGFR Tyr1173 compared to total EGFR and as means ± SD (n = 3 experiments; *P < 0.05).

In addition to the conventional ligand-induced activation, EGFR can also be activated by transactivation, a process that involves maturation of heparin-binding EGF-like growth factor (HB-EGF) and EGF secretion after activation of matrix metalloproteinases by various signaling pathways, such as heterotrimeric guanosine 5′-triphosphate–binding protein (G protein)–coupled receptors (30). EGFR and other tyrosine kinase and cytokine receptors can be activated, in the absence of ligand, by ultraviolet (UV) irradiation, osmotic stress, or fatty acid production (31, 32). We therefore determined whether ectopic expression of IpgD could activate EGFR through autocrine production of EGF. Conditioned medium experiments were conducted by incubating naïve HeLa cells with medium from IpgD-expressing cells. Under these conditions, no activation of EGFR or Akt could be detected in naïve cells (fig. S5A). Moreover, EGF was not detected in the media of IpgD-expressing cells with an enzyme-linked immunosorbent assay (ELISA) (fig. S5B). Finally, the matrix metalloproteinase inhibitors GM6001 and marimastat had no effect on IpgD-mediated phosphorylation of EGFR (fig. S5C). These data suggest that EGFR activation was not due to secreted ligand and indicate that PI5P production was likely to activate EGFR in the absence of its bona fide ligands in infected HeLa cells.

To demonstrate that EGFR was a critical intermediate between PI5P and Akt, we treated cells infected by S. flexneri with AG1478, an inhibitor of EGFR kinase activity. This led to a decrease of 60.6 ± 19% of activated Akt at the entry foci (Fig. 2, A and B). In addition, in cells expressing IpgD, EGFR knockdown by siRNA abolished Akt activation (Fig. 2, C and D). Because PI5P production was not affected by EGFR knockdown or AG1478 treatment (Fig. 2E), and because PI3K inhibition by LY294002 decreased Akt phosphorylation but spared EGFR activation (Fig. 2F and fig. S6), the succession of events was likely to be (i) production of PI5P, (ii) phosphorylation and activation of EGFR, and (iii) subsequent PI3K and Akt activation. Moreover, destabilization of the plasma membrane by addition of the cholesterol-removing agent methyl-β-cyclodextrin impaired IpgD-mediated activation of Akt (fig. S7). This suggested that Akt activation occurs at the plasma membrane, where cholesterol- and sphingolipid-rich nanodomains are important for its recruitment and activation upon phosphatidylinositol 3,4,5-trisphosphate [PI(3,4,5)P3] production (33).

Fig. 2

EGFR is a key intermediate between PI5P and Akt activation. (A) Serum-starved cells were treated with the EGFR inhibitor AG1478 for 2 hours and infected for 10 min with the WT S. flexneri (M90T). Actin was visualized with Alexa Fluor 488–phalloidin and Akt phosphorylation with anti-pAkt Ser473. White arrows show entry foci sites. Scale bars, 10 μm (n = 3 experiments). (B) The average intensity of pAkt Ser473 signal at foci entry sites 10 min after infection was measured (n = 3 experiments, 10 entry foci and 10 cells analyzed per experiment; *P < 0.05). (C) Lysates from serum-starved cells transfected with control or EGFR siRNA and plasmids encoding the indicated proteins were probed with antibodies to pAkt Ser473, Akt, and EGFR. (D) Densitometry analysis of the immunoreactive bands corresponding to pAkt Ser473 shown in (C) (*P < 0.05). (E) PI5P amounts under the conditions described in the above experiments were quantified by mass assay. Data are presented as fold increase of control and as means ± SD (n = 3 experiments; *P < 0.05). (F) Serum-starved cells expressing the indicated proteins were treated with vehicle [dimethyl sulfoxide (DMSO)] or 25 μM LY294002 (LY) for 2 hours. Lysates were probed with antibodies to pEGFR Tyr1173, pAkt Ser473, and EGFR and Akt. Data are presented as means ± SD (n = 3 experiments).

PI5P enriches early endosomes in active EGFR and prevents its lysosomal degradation

Upon ligand binding, the signaling capacity of EGFR decreases within minutes after endocytosis and degradation (34). Dimerized activated EGFR is ubiquitinated, internalized in early endosomes, and sorted to be either recycled back to the plasma membrane or sent to late endosomes and lysosomes where degradation occurs (34). In our system, EGFR phosphorylation was still detected after 48 hours of IpgD expression (Figs. 1, E and F, and 2F). In addition, the activation of Akt was strong and sustained, suggesting attenuation of mechanisms that decrease activity. Moreover, the abundance of EGFR remained stable after 24 hours of IpgD expression (Figs. 1, E and F, and 2F). These observations suggest that PI5P could maintain extended Akt activation by two nonexclusive ways: (i) by inducing a reduction in the dephosphorylation of Akt by phosphatases such as protein phosphatase 2A (PP2A) and (ii) by affecting EGFR processing and degradation. We previously showed that an increase in PI5P amounts led to PP2A inhibition (35), possibly through phosphorylation by EGFR (36). Thus, we investigated whether EGFR processing could also be affected. Although EGF treatment of control cells induced complete degradation of EGFR within 60 min, the amount of EGFR in IpgD-expressing cells was stable and comparable to that in unstimulated control cells (Fig. 3A and fig. S8). When PI5P was transformed back to PI(4,5)P2 by expression of PI5P 4-kinase IIα, degradation of EGFR was restored (Fig. 3A and fig. S8). Accordingly, addition of C4-PI5P prevented EGFR degradation upon sustained EGF stimulation of HeLa cells (fig. S1, D and E), demonstrating a key role for PI5P in this process.

Fig. 3

PI5P induces internalization and localization of EGFR in EEA1-positive endosomes. (A) Serum-starved HeLa Tet-Off IpgD–expressing cells transfected with the indicated proteins were stimulated with EGF (200 ng/ml) for 0 or 60 min. Lysates were probed with antibodies to EGFR, mCherry, myc, or tubulin (n = 3 experiments). (B) Serum-starved HeLa Tet-Off IpgD–expressing cells transfected with the indicated proteins were assessed for cell surface expression of EGFR by surface biotinylation (n = 3 experiments). (C and E) Serum-starved cells expressing EGFP or EGFP-IpgD-WT were fixed and labeled with antibodies against EGFR, EEA1, or LAMP1. Insets correspond to magnifications of the regions shown in boxes. Scale bars, 10 μm. (D and F) Individual endosomes containing EGFR and EEA1 or LAMP1 were counted in cells expressing EGFP, EGFP-IpgD-WT, or EGFP-IpgD-C438S (49.5 ± 8.4% of EGFR and EEA1 colocalization compared to 21.4 ± 5.0% in control cells). Values are percentages of the total number of EGFR-containing endosomes. Data are presented as means ± SD (n = 5 experiments, 10 cells per experiment; *P < 0.05).

Biotinylation of surface proteins showed that IpgD expression induced a decrease in cell surface EGFR, an effect that was rescued by expression of PI5P 4-kinase IIα (Fig. 3B). Confocal analysis provided similar observations and indicated that transformation of comparable amounts of PI(4,5)P2 into PI4P by Inp54p had no effect on EGFR plasma membrane localization (fig. S9). To determine at which step PI5P affected EGFR degradation, we monitored EGFR trafficking by confocal microscopy. In IpgD-expressing cells, we observed a significant increase in EGFR in EEA1-labeled early endosomes (Fig. 3, C and D), whereas in cells expressing the inactive IpgD-C438S mutant, EGFR remained localized at the plasma membrane (Fig. 3D). Moreover, in IpgD-transfected cells, the distribution of EGFR largely overlapped with that of sorting nexin-1 (SNX1), another marker of the early endosome compartment (fig. S10, A and B). Under these conditions, the amount of EGFR did not increase in LAMP1 (lysosomal-associated membrane protein 1)–positive late endosomes (Fig. 3, E and F). Accordingly, IpgD expression did not increase the amount of EGFR in lysosomes because no change was observed in the colocalization of EGFR with LysoTracker, a probe that labels this acidic organelle (fig. S10, C and D). Together, these data demonstrate that PI5P promotes endocytosis of EGFR and blocks its routing toward the lysosomal compartment.

EGFR and platelet-derived growth factor receptor (PDGFR) are active in the early endosome compartment where they can generate efficient signals through Akt or mitogen-activated protein kinases (MAPKs) (37, 38). Therefore, we next tested whether EGFR was active when located in early endosomes in our model. IpgD expression induced colocalization of phosphorylated EGFR and EEA1 (Fig. 4, A and B), indicating that active EGFR preferentially localized in EEA1-positive compartments. Moreover, a significant portion of phosphorylated Akt colocalized with EGFR (Fig. 4, C and D), suggesting that PI5P production promoted relocation of signaling EGFR in early endosomes.

Fig. 4

The active form of EGFR localizes to EEA1-positive endosomes. (A and C) Serum-starved HeLa cells expressing EGFP or EGFP-IpgD-WT were fixed and labeled with antibodies to pEGFR Tyr1173, EGFR, pAkt Ser473, or EEA1. Insets are magnifications of the regions shown in boxes. Scale bars, 10 μm. (B and D) Individual endosomes containing pEGFR Tyr1173 and EEA1 or EGFR and pAkt Ser473 were counted. Values are percentages of the total number of EGFR-containing endosomes. Data are presented as means ± SD (n = 3 experiments, 10 cells per experiment; *P < 0.05).

Increased PI5P amounts modify routing of EGFR from EEA1 and SNX1 endosomes to lysosomes but spare recycling and retrograde transport

We next determined whether PI5P affected EGFR degradation or whether this phosphoinositide could modulate other vesicular pathways. First, the bulk transport from early to late endosomes was followed by pulse-chase experiments of endocytosis with the fluid phase marker tetramethylrhodamine dextran. The tracer accumulated in EEA1-positive compartments upon IpgD expression, whereas it reached the LAMP1 compartment in control cells (Fig. 5, A to D), showing that PI5P blocked general vesicular transport from early to late endosomes. EEA1 endosomes function as a vesicle sorting compartment, where cargo sorting and maturation or transport occur to form either multivesicular bodies and lysosomes, recycling vesicles, or vesicles directed to the retrograde pathway (39). The effect of PI5P on recycling was then assessed by monitoring the trafficking of the transferrin receptor, which, upon activation and endocytosis, is recycled to the plasma membrane (40). IpgD expression did not affect recycling because Texas Red–labeled transferrin bound to its receptor was internalized and reached the cell surface to be released back into the medium (Fig. 5E).

Fig. 5

PI5P increase alters the maturation of early to late endosomes but spares recycling. (A) Tetramethylrhodamine dextran was pulsed for 10 min at 37°C in serum-starved HeLa cells expressing EGFP or EGFP-IpgD-WT and then chased for an additional 45 min. (C) Cells were fixed and labeled with antibodies to EEA1 (A) or LAMP1 (C). Insets are magnifications of the regions shown in boxes. Scale bars, 10 μm. (B and D) Individual endosomes containing both dextran and EEA1 (B) or dextran and LAMP1 (D) were counted. Values are the percentage of the total number of dextran-containing endosomes and are presented as means ± SD (n = 3 experiments, 10 cells per experiment; *P < 0.05). (E) Serum-starved HeLa cells expressing EGFP or EGFP-IpgD-WT were incubated with Texas Red–labeled transferrin (20 μg/ml) for the indicated times at 37°C and then fixed and observed by confocal microscopy (n = 3 experiments, 10 cells per experiment). Scale bars, 10 μm.

Finally, we tested whether IpgD affected retrograde transport from endosomes to the trans-Golgi network (TGN), which is regulated by the retromer machinery (41). Shiga toxin uses this pathway to bypass the recycling route to the plasma membrane and the late endocytic pathway (41). Using Cy3-labeled Shiga toxin B subunit, we demonstrated that the retrograde pathway was not affected by increased amounts of PI5P (fig. S11). Together, these observations demonstrate that IpgD-generated PI5P specifically alters the degradation route.

PI5P is enriched in early endosomes

Having established that PI5P selectively modifies the routing from EEA1 and SNX1 endosomes to lysosomes, we next isolated the whole endosomal compartments of IpgD-expressing HeLa cells for biochemical analysis. We observed that PI5P was enriched in endosomes [553.3 pmol per nanogram of protein in the endosomal fraction compared to 98.8 pmol per nanogram of protein in the post-nuclear supernatant (PNS)]. Early and late endosomal fractions from baby hamster kidney (BHK) cells were separated (42), and in BHK cells, PI5P produced by IpgD also activated Akt in an EGFR-dependent manner (Fig. 6A). Again, the relative amount of PI5P increased in early endosomes and, to a lesser extent, in the late endosomal fraction with IpgD expression (Fig. 6B). Moreover, phosphorylated EGFR and phosphorylated Akt were enriched in the early endosome fraction of IpgD-expressing BHK cells (Fig. 6C).

Fig. 6

Enrichment of PI5P in endosomes. (A) Serum-starved BHK cells expressing Myc-IpgD-WT or not were treated with AG1478 for 2 hours. Lysates were immunoblotted with antibodies to pAkt473 and Akt (n = 3 experiments). (B) Serum-starved BHK cells expressing Myc-IpgD-WT or not were lysed, and the post-nuclear supernatant (PNS) was further fractionated on a sucrose gradient into heavy membrane (HM), early endosome (EE), and late endosome (LE). The amount of PI5P was quantified in each fraction by mass assay and calculated as picomoles per nanogram of protein. Data are presented as the fold increase for each fraction compared to PNS of control cells; 2.5 μg of proteins from each fraction was immunoblotted for the presence of EEA1. Data are presented as mean ± SD (n = 3 experiments; *P < 0.05). (C) The PNS and EE from serum-starved BHK cells expressing Myc-IpgD-WT or not were probed with antibodies to pAkt Ser473, pEGFR Tyr1173, and Rab5 (n = 3 experiments). (D) Schematic representation of the role of PI5P in the control of vesicular trafficking and growth factor signaling duration. This effect may be important for survival of infected cells to benefit the pathogen. RE, recycling endosomes; Ly, lysosomes.


Here, we provide new insights into the role of PI5P that appears to be an important regulator of the endocytic pathway. Specifically, the production of this lipid by the bacterial phosphatase IpgD triggers EGFR activation, which is essential for the sustained activation of the Akt pathway. How EGFR is activated remains to be established, but we provide evidence that this activation is ligand-independent, as previously shown for EGFR activation induced by UV or osmotic stress (31). We show that a bacterial pathogen has evolved a strategy to produce PI5P through IpgD and that this lipid then induces the accumulation of active EGFR in EEA1- and SNX1-positive early endosomes, thereby blocking its degradation and leading to prolonged survival signals (Fig. 6D). We show that conversion of PI(4,5)P2 into PI5P, but not into PI4P, prevents EGFR degradation induced by addition of high concentrations of EGF, whereas transformation of PI5P back into PI(4,5)P2 restores EGFR degradation. Moreover, short-chain PI5P reproduces the effect of IpgD expression on EGFR activation and on prevention of its degradation upon EGF stimulation. These results highlight the effect of phosphoinositide metabolism on the dynamics of endosomes, which is emerging as an important compartment for signal transduction (34, 37, 43, 44). Accumulation of EGFR and its downstream effectors in endosomes has been previously described, and endosome-specific signaling of activated EGFR has been proposed to be sufficient to promote cell survival through the PI3K and Akt pathway (38). In line with these observations, our results show that active EGFR and Akt accumulate in early endosomes upon increase of PI5P by IpgD expression.

In our system, IpgD produces PI5P at the plasma membrane (28), which then rapidly accumulates in early endosomes, leading to modification in trafficking: Lysosomal degradation is blocked, whereas the recycling and retrograde pathways remain effective. We propose that this subversion of the trafficking machinery by PI5P could result from a mislocalization of regulatory proteins of the endosomal compartment, which impairs maturation and sorting. A phylogenetic study (45) suggested that mammalian enzymes involved in PI5P metabolism, including PIKfyve (22) and myotubularin phosphatases (23), have been evolutionarily conserved as active regulators of endosome functions, thus predicting a role for PI5P in endosomal compartments. Both enzymes regulate the amounts of two endosomal lipids, phosphatidylinositol 3-phosphate (PI3P) (in early endosomes) and PI(3,5)P2 (in late endosomes), that play important roles in the homeostasis of these compartments through their capacity to recruit FYVE, PX, or β-propeller–containing proteins (46). Myotubularins can produce PI5P from PI(3,5)P2 (23), thus modulating the amount of these lipids in endosomes. Consistent with our data, overexpression of myotubularin can protect EGFR from degradation by blockade of the late endosome-to-lysosome pathway (47). Furthermore, forced myotubularin targeting to Rab5-positive endosomes, leading to depletion in PI3P, can alter this endosomal fraction to become an early endocytic intermediate with signaling properties, resulting in enhanced growth factor signaling (43). This is in agreement with data showing that inhibition of PI3P signals does not affect bulk transport from early to late endosomes but retains EGFR in early endosomes (48). Detectable amounts of PI5P are present not only in the plasma membrane but also in internal vesicles and, like PI3P, in the endoplasmic reticulum and TGN (27). Overall, these data indicate that PI3P, PI(3,5)P2, and PI5P are critical regulators of the plasticity of the trafficking machinery and act at different points in the endocytic pathway (5, 45, 49). Changes in their concentration, potentially by interconversion, can alter the fate of endocytic vesicles and in turn modulate signaling mechanisms. Endocytosis is indeed emerging as a spatiotemporal organizer of signaling circuits (37, 39, 44). Furthermore, loss-of-function mutations in enzymes controlling the amounts of PI3P, PI(3,5)P2, and PI5P can induce pathologies such as myopathies or neuropathies (9). Bacterial pathogens have evolved strategies to subvert and exploit the vesicular trafficking machinery by specifically manipulating phosphoinositide amounts (50) and activation of growth factor receptors, particularly the EGFR, as previously reported for S. Typhimurium (51). We show how S. flexneri combines these two aspects through PI5P production to increase survival of host cells (Fig. 6D).

In conclusion, our data underline the role of the relatively uncharacterized lipid PI5P as a key regulator of intracellular trafficking that impairs the maturation or the transport of endosomes to lysosomes, which in turn extends the duration of survival signals. Because PI5P is present in various cell types in basal amounts that increase upon cell stimulation or stress, this lipid appears to be an important regulator of traffic and may thus affect physiology and pathologies, from infection to genetic diseases and cancer (52).

Materials and Methods

Plasmids, reagents, and antibodies

The following plasmids were used: pRK5-Myc-IpgD-WT and pRK5-Myc-IpgD-C438S (27); pEGFP-IpgD-WT and pEGFP-IpgD-C438S; pEGFP-PHD3x and pGEX-PHD2x (14); pcDNA3-EGFP-Lyn-Inp54p (gift from T. Meyer, Stanford University School of Medicine, Stanford, CA); pEGFP-PI5P4KIIβ-WT (gift from L. Rameh, Boston Biomedical Research Institute, Watertown, MA). The antibodies used were anti-pAkt (Ser473 and Thr308), anti-EGFR and anti-pEGFR (Tyr1068) (Cell Signaling Technology); anti-pEGFR (Tyr1173), anti-Akt1/2/3, anti-GFP, anti-Rab6 (Santa Cruz Biotechnology); anti-Rab5 (BD Pharmingen); anti-EEA1, anti-SNX1 (Transduction Laboratories); anti-human LAMP1 (BD Pharmingen); anti-myc (Sigma-Aldrich); anti-mCherry (Clontech); and anti-transferrin receptor (Invitrogen). LY294002, AG1478, and methyl-β-cyclodextrin were from Sigma-Aldrich, and GM6001 and marimastat were from Calbiochem. Texas Red transferrin complex, LysoTracker, and dextran tetramethylrhodamine lysine fixable (molecular weight, 10,000) were from Molecular Probes. Cy3-labeled Shiga toxin B subunit was a gift of L. Johannes (Institut Curie, Paris, France). Enhanced green fluorescent protein (EGFP)–PI5P4KIIα WT and D273K (gift from K. A. Hinchliffe, University of Manchester, Manchester, UK) were subcloned in the BsrG I and Xho I sites of pmCherry-C1 vector (Clontech).

Cell lines and transfections

HeLa cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS). Plasmid, SMARTpool siRNA against EGFR (L-003114-00-0005), and Non-Targeting Pool (D-001810-10-05; Dharmacon SiGenome) siRNA transfections were performed with Effectene according to the manufacturer’s instructions (Qiagen). When required, siRNA transfections were performed 24 hours before plasmid transfections. Transfected cells were incubated in 10% FBS for 12 hours, and then serum-starved for 12 hours before processing, at which point RNA interference had been occurring for 48 hours total. BHK cells were grown in Glasgow minimum essential medium (GMEM) supplemented with 5% FBS (Invitrogen) and tryptone (Sigma-Aldrich), transfected with Effectene (Qiagen), and serum-starved for 12 hours before processing.

To generate stable cell lines, we subcloned Myc-EGFP-IpgD-WT into the pTRE2-hyg-myc expression vectors (Clontech) by standard techniques, and plated 1 × 106 HeLa Tet-Off cells (Clontech) expressing the tetracycline-controlled transactivator onto 100-mm dishes and transfected them with 2 μg of pTRE2-hyg-myc-EGFP-IpgD-WT using Effectene (Qiagen). Forty-eight hours after transfections, hygromycin (400 μg/ml; Sigma-Aldrich) was added to cells with doxycycline (2 μg/ml; Sigma-Aldrich) to suppress protein expression during the selection process. Transformants were selected and single cell–cloned with the limiting dilution method. To test for induced expression, we removed doxycycline from the medium. EGFP-positive clones were selected and analyzed by Western blot for expression of EGFP-IpgD-WT.


Phospholipids were extracted and analyzed by high-performance liquid chromatography (HPLC) after metabolic labeling with [32P]orthophosphate (200 μCi/ml; Perkin Elmer), except for PI5P that was quantified by mass assay as previously described (4, 13, 23, 28).

For experiments with short-chain lipids, C4-PI4P and C4-PI5P (Echelon Biosciences Inc.) were resuspended in water and used at 15 μM for the indicated times on HeLa cells that had been serum-starved overnight.

Serum-starved HeLa cells were incubated with the indicated concentration of methyl-β-cyclodextrin for 1 hour before lysis and separation on SDS–polyacrylamide gel electrophoresis (SDS-PAGE) for immunoblotting analysis.

S. flexneri strains and cell infection

Semiconfluent HeLa cells were infected with exponentially growing wild-type S. flexneri serotype 5 (M90T), invasive IpgD-deficient (ΔIpgD), and noninvasive BS176 strains as previously described (4, 28). At the indicated times, cells were fixed with paraformaldehyde for immunofluorescence or scraped in Laemmli sample buffer for Western blotting.

Western blotting and immunofluorescence

Cells were scraped in Laemmli sample buffer, and proteins were separated by SDS-PAGE before Western blotting on Immobilon-P membranes (Millipore). Immunoreactive bands were detected by chemiluminescence with the SuperSignal detection system (Pierce Chemical Co.). For immunofluorescence, HeLa cells were fixed for 10 min in 4% paraformaldehyde before permeabilization for 2 min with 0.1% Triton X-100 and saturation in 3% bovine serum albumin (BSA)/phosphate-buffered saline (PBS). Cells were incubated with the appropriate antibodies or the glutathione S-transferase (GST)–PHDx2 biotinylated probe (40 μg/ml in 1% BSA/PBS) for 1 hour (28). After incubation with fluorescent secondary antibodies or Alexa Fluor 488–phalloidin (Molecular Probes), cells were washed and mounted with Mowiol. Confocal imaging was captured with an LSM 510 or LSM 710 microscope equipped with a 63× Plan-Neofluar objective (Carl Zeiss).

Cell surface EGFR measurements

Cell surface EGFR was assessed by cell surface biotinylation. Briefly, cells were incubated at 4°C with sulfo-SS-biotin (0.5 mg/ml; Pierce) for 30 min, washed in PBS, quenched for 20 min with 20 mM NH4Cl, and washed in PBS. Cells were then lysed in TNE buffer [20 mM tris (pH 7.4), 250 mM NaCl, and 1 mM EDTA] with 1% NP-40 and antiproteases for 45 min, with gentle shaking at 4°C. Cells were scraped and lysates were centrifuged for 10 min at 14,000g. The supernatant was rotated overnight with streptavidin-coated agarose bead resin (Pierce). The beads were washed three times with TNE buffer with 0.1% NP-40, boiled in Laemmli without dithiothreitol, and run on SDS-PAGE before immunoblotting.

Enzyme-linked immunosorbent assay

EGF amounts in culture media were determined with the Quantikine Human EGF ELISA (R&D Systems) according to the manufacturer’s instructions.

Endosome purification

Subcellular fractionation was performed as described (42). Briefly, cells were homogenized in homogenization buffer (HB) containing 250 mM sucrose and 3 mM imidazole (pH 7.4). The PNS was adjusted to 40.6% sucrose and loaded at the bottom of an SW41 tube. For BHK cells, the PNS was overlaid with 35 and 25% sucrose, 3 mM imidazole (pH 7.4), and HB and, for HeLa cells, with 35% sucrose, 3 mM imidazole (pH 7.4), and HB. After centrifugation at 35,000 rpm for 90 min in an SW41 rotor, the endosomal fraction was collected at the interface of the 35% sucrose-HB layers. For BHK cells, early endosomes were collected at the interface of the 35 to 25% sucrose layers and the late endosomes at the interface of the 25% sucrose-HB layers.

Transferrin recycling and Shiga toxin endocytosis assays

HeLa cells, grown on coverslips and transfected with GFP or GFP-IpgD-WT, were serum-starved overnight. For the transferrin recycling assay, cells were washed with DMEM and 0.1% BSA and incubated on ice in the same medium in the presence of Texas Red transferrin (20 μg/ml) for 1 hour. After a PBS wash, cells were chased for 30 or 120 min at 37°C in DMEM with 0.1% BSA. The incubation was stopped on ice; cells were washed once with DMEM, once with 0.2 M acetic acid and 0.5 M NaCl, and twice with PBS before fixation in 3.7% formaldehyde. For the Shiga toxin assay, cells were washed twice in PBS and incubated for 1 hour on ice in DMEM plus Shiga toxin B–Cy3 (0.5 μg/ml). After three washes in PBS supplemented with BSA (5 g/liter), cells were incubated at 37°C for 0, 10, or 50 min. The incubation was stopped on ice, and after three washes in cold PBS, cells were fixed in 3% paraformaldehyde for 20 min. After permeabilization, cells were incubated with either anti-transferrin antibody followed by anti-mouse Alexa 647 (for the 10-min chase) or anti-Rab6 antibody followed by anti-rabbit Alexa 647 (for the 50-min chase).

Statistical analysis

Statistical analysis was performed by Student’s t test or one-way analysis of variance (ANOVA) followed by Bonferroni posttest with GraphPad Prism 4.0 software (GraphPad Software Inc.). Statistical analysis of colocalization was performed by Fisher’s exact test. Differences were considered significant when the P value was less than 0.05 (as indicated by an asterisk).

Supplementary Materials

Fig. S1. Exogenous short-chain PI5P induces activation of EGFR and protects EGFR from degradation.

Fig. S2. Quantification of immunofluorescence and Western blots corresponding to Fig. 1.

Fig. S3. Phosphoinositide concentrations in HeLa cells expressing EGFP, EGFP-IpgD-WT, or EGFP-Inp54p.

Fig. S4. Characterization of the Tet-Off IpgD-WT HeLa stable cell line.

Fig. S5. PI5P production activates EGFR in the absence of detectable ligand.

Fig. S6. Quantification of Western blots corresponding to Fig. 2F.

Fig. S7. Cholesterol depletion attenuates IpgD-induced Akt activation.

Fig. S8. Quantification of Western blots corresponding to Fig. 3A.

Fig. S9. Transformation of PI(4,5)P2 into PI5P, but not into PI4P, changes EGFR localization.

Fig. S10. PI5P production localizes EGFR to SNX1-positive early endosomes and prevents its routing to lysosomes.

Fig. S11. PI5P production does not affect retrograde transport.

References and Notes

  1. Acknowledgments: We thank E. Saland for excellent technical assistance and S. Manenti, C. Racaud-Sultan, and C. Lamaze for stimulating discussions. We thank S. Allart and D. Sapede (Cellular Imaging Platform, IFR150) for their help with confocal imaging and M. Tremblay-Franco (Plateforme AXIOM, Genotoul) for her help in statistical analysis. Funding: This work was supported by the Agence Nationale de la Recherche (ANR; Programme Blanc), ANR E-rare, and Association pour la Recherche sur le Cancer (ARC). D.R. and F.L. were supported by grants from the Ministere de la Recherche et de la Technologie and ARC. P.J.S. is a Howard Hughes Medical Institute foreign scholar. Author contributions: D.R. and F.L. conducted biochemical and cell biology experiments with the assistance of H.T.; S.D.-C. and V.P. conducted endosome isolation; J.M. performed S. flexneri infection experiments; G.C. performed HPLC phosphoinositide analysis; P.J.S. directed and interpreted S. flexneri infection experiments; and F.G.-I., H.T., and B.P. directed the study, analyzed the data, and wrote the paper. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article