A Vibrio Effector Protein Is an Inositol Phosphatase and Disrupts Host Cell Membrane Integrity

See allHide authors and affiliations

Science  24 Sep 2010:
Vol. 329, Issue 5999, pp. 1660-1662
DOI: 10.1126/science.1192850

This article has a correction. Please see:


The marine bacterium Vibrio parahaemolyticus causes gastroenteritis in humans and encodes the type III effector protein VPA0450, which contributes to host cell death caused by autophagy, cell rounding, and cell lysis. We found that VPA0450 is an inositol polyphosphate 5-phosphatase that hydrolyzed the D5 phosphate from the plasma membrane phospholipid phosphatidylinositol 4,5-bisphosphate. VPA0450 disrupted cytoskeletal binding sites on the inner surface of membranes of human cells and caused plasma membrane blebbing, which compromised membrane integrity and probably contributed to cell death by facilitating lysis. Thus, bacterial pathogens can disrupt adaptor protein–binding sites required for proper membrane and cytoskeleton dynamics by altering the homeostasis of membrane-bound inositol-signaling molecules.

The Gram-negative marine bacterium Vibrio parahaemolyticus is a leading cause of gastroenteritis from the consumption of contaminated seafood (1). Many virulent strains of V. parahaemolyticus encode thermostable direct hemolysins and two type III secretion systems (T3SS1 and T3SS2) (2, 3). The T3SS is a syringe-like mechanism often used by Gram-negative bacteria to introduce effector proteins into eukaryotic target cells during infection (4).

The V. parahaemolyticus T3SS1 injects effectors that cause a rapid, orchestrated cell death mediated by autophagy, cell rounding, and then cell lysis (5). One effector, VopQ, is both necessary and sufficient to induce autophagy (6), whereas another effector, VopS, is an AMPylator that contributes to cell rounding by modifying a conserved threonine residue on the Rho family of guanosine triphosphatases (GTPases) with adenosine 5'-monophosphate (AMP), preventing their interaction with downstream-signaling molecules (7). Upon analysis of the contribution of other T3SS1 effectors (8) involved in this paradigm of cell death, we found that a strain with only a functional T3SS1 (POR3) that was deleted for the effector VPA0450 (POR3Δvpa0450) (Fig. 1B and fig. S1A) caused cell rounding faster than the parental POR3 strain or the complemented strain (POR3Δvpa0450 + VPA0450) during infection of HeLa cells (Fig. 1, A and C, respectively) (9). Additionally, both the POR3 strain (Fig. 1, E and I) and the complemented POR3Δvpa0450 + VPA0450 strain (Fig. 1, G and J) induced a transient blebbing of the host cell membrane before cell rounding, whereas POR3Δvpa0450 initiated rounding without any blebbing (Fig. 1, B and F). Further analysis revealed that POR3Δvpa0450 delayed lysis of the infected cell by approximately 1 hour as compared with the POR3 strain (Fig. 1K). Complementation of the deleted strain with VPA0450 restored the lysis activity of this strain back to wild-type levels (Fig. 1K). Thus, the presence of VPA0450 contributed to a blebbing phenomenon during the early stages of infection and to efficient lysis during the later stages of infection.

Fig. 1

Expression of VPA0450 leads to rapid host cell lysis. HeLa cells were infected with POR3, POR3Δvpa0450, POR3Δvpa0450 + VPA0450, or POR3Δvpa0450 + VPA0450-H356A and visualized with confocal microscopy at, respectively, (A to D) 1 hour and (E to H) 1.5 hours. Scale bar, 10 μm. Blebbing is shown in detail from (I) POR3 and (J) POR3Δvpa0450 + VPA0450 infection at 1.5 hours. Scale bar, 5 μm. Actin cytoskeleton was stained with rhodamine-phalloidin (red), and nuclei were stained with Hoechst (blue). (K) Hela cells were infected with POR3 (circles), POR3Δvpa0450 (squares), or POR3Δvpa0450 + VPA0450 (triangles), and lactate dehydrogenase (LDH) release was evaluated as a measure of cytotoxicity and host cell lysis. Data are means ± SD (n = 3 samples) from a representative experiment repeated in triplicate.

Having correlated the presence of VPA0450 with efficient lysis of infected cells, we next investigated whether VPA0450 alone was sufficient to induce host cell lysis. Although HeLa cells transfected with empty vector exhibited the predicted morphology (Fig. 2A), cells transfected with VPA0450 did not lyse but remained attached to the plate and were covered by small membrane blebs that were consistent with those seen during infection (Fig. 2B). Blebs appeared not only on the surface of the cell but also on the surface of other blebs (movie S1). Thus, on its own VPA0450 did not directly lyse cells but did dramatically change the surface of transfected cells.

Fig. 2

VPA0450 mimics the synaptojanin inositol polyphosphate 5-phosphatase catalytic domain. HeLa cells were transfected with pEGFP-N1 and (A) pSFFV empty vector (B) pSFFV-VPA0450-flag, or (C) pSFFV-VPA0450-H356A-flag and visualized with confocal microscopy. Expression of GFP (green) denotes transfected cells, actin cytoskeleton stained with rhodamine-phalloidin (red), and nuclei stained with Hoechst (blue). Scale bar, 10 μm. (D) Quantitation of blebbing in Hela cells transfected with empty vector, SPsynj IPP5C, VPA0450, or VPA0450 point mutants. Data are means ± SD from three independent experiments. Asterisks refer to statistically significant differences between VPA0450-transfected cells and the other constructs (*P < 0.001, **P < 0.01, ***P < 0.05, n = 3 independent experiments) using a pairwise, two-tailed t-test.

To investigate the activity that VPA0450 might contain to cause blebbing and facilitate lysis during infection, we performed bioinformatic analysis and discovered that VPA0450 has homology with effectors from other pathogenic bacteria, including Aeromonas salmonicida (Ati2, 87% identity), which is the causative agent of furunculosis in salmonid fish and of considerable importance to the fisheries industry (10), and the entemopathogenic nematode commensal Photorhabdus luminescens TTO1 (Plu4615, 93% identity). These effectors are most similar to proteins in the endonuclease/exonuclease/phosphatase family of eukaryotic proteins (fig. S2). VPA0450 contains five of the six catalytic motifs found in the active site for the family of inositol polyphosphate 5-phosphatases (IPP5Cs), enzymes that specifically dephosphorylate the D5 phosphate on an inositol ring (fig. S3, A and B) (11, 12).

Disrupting the homeostasis of phosphatidylinositols in a cell can result in a wide range of consequences, including the promotion of tumor progression and disruption of normal neuronal development (1113). Synaptojanin from the fission yeast Schizosaccharomyces pombe (SPsynj) is a multi-domain protein containing an IPP5C domain that is associated with the regulation of intracellular calcium signaling, actin cytoskeleton, and cell growth and motility (11, 14, 15). This domain is different from the synaptojanin Sac1 domain that is homologous to Salmonella effectors SigD and SopB, which are necessary for phagocytosis as well as sustaining the Salmonella-containing vacuole during intracellular infection by that pathogen (fig. S3B) (16). Overexpression of membrane-associated Inp54p, an SPsynj IPP5C domain homolog, induces cellular blebbing (14). Expression of the SPsynj IPP5C domain in HeLa cells also resulted, albeit to a lesser extent, in cellular blebbing (fig. S3C) similar to that seen for VPA0450 (Fig. 2B). The observation that VPA0450 causes a more dramatic phenotype as compared with that of an endogenous eukaryotic homolog is characteristic of T3SS effectors. Thus, VPA0450 might contain an inositol polyphosphate 5-phosphatase activity that can alter membrane dynamics in infected and transfected cells.

Structural analysis of this family of D5-phosphatases revealed a catalytic pocket similar to that of apurinic/apyrimidinic endonucleases that uses a histidine as a base for catalysis. To test whether the conserved histidine in the putative catalytic site of VPA0450 was important for activity, we transfected the mutant VPA0450-H356A into HeLa cells. No substantial morphological changes were observed in cells as compared with those from vector control transfected cells (Fig. 2C). Complementation of the POR3Δvpa0450 strain with VPA0450-H356A did not rescue the defect in blebs or the delayed cytotoxicity (Fig. 1, D and H, and fig. S1). On the basis of this finding, we went further to ask whether other residues in the putative catalytic pocket might contribute to the activity of VPA0450. Mutation of amino acids that are predicted to coordinate substrate binding and facilitate catalysis resulted in enzymes that were less effective at inducing cellular blebbing (Fig. 2D and fig. S3, D to H). Thus, VPA0450 might be disrupting homeostasis in the infected cell by disrupting phosphatidylinositol signaling.

To directly test for putative phosphatidylinositol phosphatase activity, we performed a malachite green assay (17) with recombinant proteins, including VPA0450, VPA0450-H356A, and SPSynj IPP5C, so as to measure phosphate release. rVPA0450 and rSPSynj IPP5C, but not mutant rVPA0450-H356A, hydrolyzed phosphate off phosphatidylinositol 4,5-bisphosphate [PtdIns(4,5)P2] (Fig. 3A). None of the recombinant enzymes hydrolyzed phosphate off of phosphatidylinositol 3,4-bisphosphate (Fig. 3A). Both rVPA0450 and rSPSynj IPP5C, but not mutant rVPA0450-H356A, cleaved phosphate off phosphatidylinositol 3,4,5-triphosphate (Fig. 3A). Thus, rVPA0450 contained phosphatidylinositol phosphatase activity that specifically hydrolyzed the D5 inositol phosphate.

Fig. 3

VPA0450 has inositol polyphosphate 5-phosphatase activity. (A) Phosphatase activity of rSPsynj IPP5C, rVPA0450, and rVPA0450-H356A on phosphatidylinositol (4,5) bisphosphate di-C8 [PI(4,5)P2], phosphatidylinositol (3,4) bisphosphate di-C8 [PI(3,4)P2], and phosphatidylinositol (3,4,5) triphosphate di-C8 [PI(3,4,5)P3], measured by means of malachite green assay of inorganic phosphate release. Asterisks refer to statistically significant differences between PtdIns control and each recombinant protein for a given substrate (*P < 0.001, **P < 0.01, n = 3 independent experiments) using a pairwise, two-tailed t-test. HeLa cells transfected with (B) PH(PLCδ1)-GFP and empty vector, (C) pSFFV-VPA0450-flag, or (D) pSFFV-VPA0450-H356A-flag and visualized by means of confocal microscopy. Green denotes expression of PH(PLCδ1)-GFP, orange denotes expression of VPA0450-flag, and grayscale denotes all cells. Scale bar, 10 μm. Quantitation of (E) blebbing or (F) GFP at the membrane in Hela cells transfected with 2xFYVE domain and empty vector, VPA0450, or VPA0450-H356A. Data are means ± SD from three independent experiments. Asterisks refer to statistically significant differences between VPA0450 transfected cells and the other constructs (*P < 0.001; **P < 0.01; ***P < 0.05, n = 3 independent experiments) using a pairwise, two-tailed t-test.

The plasma membrane PtdIns(4,5)P2 mediates signaling between the plasma membrane and the actin cytoskeleton by acting as a docking site for adaptor molecules. The pleckstrin homology (PH) domain from phospholipase-Cδ1 [PH(PLCδ1)] binds PtdIns(4,5)P2 to tether the actin cytoskeletal machinery to the membrane. Reversible inositol-mediated tethering and changes in association with the actin cortex permits membranes to swell, promoting cell movement (14, 18). We predicted that the constitutive hydrolysis of PtdIns(4,5)P2 by VPA0450 would result in unregulated membrane blebbing because PH(PLCδ1) domain–like proteins could no longer dock efficiently at the membrane. To test this hypothesis, we transfected Hela cells with a PH(PLCδ1)–green fluorescent protein (GFP) domain in the presence of empty vector, VPA0450, or VPA0450-H356A (Fig. 3, B to F, and fig. S4A). In cells cotransfected with either empty vector or VPA0450-H356A, we observed GFP-mediated fluorescence around the edge of the cell, whereas in the presence of wild-type VPA0450 the GFP fluorescence appeared cytoplasmic, indicating that VPA0450 and its IPP5C activity disrupted the proper association of the PH(PLCδ1) domain with the membrane. HeLa cells infected with either POR3 or POR3Δvpa0450 + VPA0450, but not POR3Δvpa0450, did not show localization of PtdIns(4,5)P2 at plasma membrane when cells began to bleb and round (fig. S4, B to E). 2xFYVE-GFP binds specifically to phosphatidylinositol 3-phosphate enriched on endosomes (19). Cells cotransfected with 2xFYVE-GFP and empty vector, VPA0450, or VPA0450-H356A all resulted in cytoplasmic GFP punctae that is consistent with endosomal staining, indicating that VPA0450 does not affect the inositol D3 phosphate (fig. S5, A to F). Thus, VPA0450 can act to hydrolyze PtdIns(4,5)P2 specifically and destroy docking sites for PtdIns(4,5)P2–specific PH domains, disrupting the interaction between the actin cytoskeleton and the membrane.

Bacterial pathogens use many tools to disrupt the actin cytoskeleton in an infected cell, including GTP-activating proteins and exchange factors, kinases, AMPylators, and phosphatases (7, 20, 21). V. parahaemolyticus uses a distinct repertoire of effectors during infection to unregulate autophagy, then cause cell rounding, followed by cell lysis. In an uninfected cell, integrity of the plasma membrane is maintained through the dynamic interaction of the cytoskeleton with various actin-binding and -nucleating proteins (fig. S6A). We have shown that transfection of VPA0450 into HeLa cells is sufficient to form membrane blebs similar to those seen during infection by V. parahaemolyticus carrying the vpa0450 gene but is not able to directly cause cell lysis (fig. S6B). We propose that VPA0450 facilitates cell lysis by using its IPP5C domain to hydrolyze PtdIns(4,5)P2, compromising the integrity of the cell membrane during late stages of infection (fig. S6C). The target of VPA0450, the D5 phosphate on the inositol ring, is critical to mediate a conversation between the actin cytoskeleton and the cell membrane. Disruption of this dynamic process by VPA0450, coupled with the activities of other effectors, including VopQ and VopS, leads to destabilization of the actin cytoskeleton and rapid cell lysis during infection. The discovery of the activity for VPA0450 highlights a lipid target that is crucial for cellular movement that bacterial pathogens can exploit to disrupt the integrity of the cell membrane and its dynamic interaction with the actin cytoskeleton.

Supporting Online Material

Materials and Methods

Figs. S1 to S6


Movie S1

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank N. Alto, H. Yin, J. Hurley, F. Robinson, J. Albanesi, B. Barylko, K. Luby-Phelps, Chris Smith, T. Iida, T. Honda, L. McCarter, and the Orth lab for insightful discussions, critical reading, and generous supply of reagents. K.O., L.Z., H.G., M.L.-A., and C.A.B. are supported by grants from NIH–Allergy and Infectious Disease (R01-AI056404 and R01-AI087808) and the Welch Foundation (I-1561). C.A.B is supported by National Institute of General Medical Sciences training grant 5T32GM008203 in cellular and molecular biology. K.O. is a Burroughs Wellcome Investigator in Pathogenesis of Infectious Disease and a W.W. Caruth Jr. Biomedical Scholar.

Stay Connected to Science

Navigate This Article