Pseudomonas aeruginosa type III-secreted toxin ExoT inhibits host-cell division by targeting cytokinesis at multiple steps Sasha H. Shafikhani* and Joanne Engel*†‡ *Division of Infectious Diseases, Department of Medicine and †Department of Microbiology and Immunology, University of California, San Francisco, CA 94143 Edited by E. Peter Greenberg, University of Washington School of Medicine, Seattle, WA, and approved August 31, 2006 (received for review July 14, 2006)
type III secretion system 兩 virulence factor 兩 microbial pathogenesis 兩 wound healing
paxillin and p130Cas, resulting in inhibition of cell migration and cell rounding (15, 16). We have previously demonstrated that P. aeruginosa inhibits the repair of wounded epithelium in an ExoT-dependent manner (17, 18). In this article, we report that P. aeruginosa inhibition of wound repair is in part due to its inhibition of host-cell division by targeting cytokinesis at multiple steps. We demonstrate that ExoT is both necessary and sufficient for this activity and that its GAP and the ADPRT domains each inhibit cytokinesis via distinct mechanisms. The GAP domain inhibits early steps in cytokinesis by inhibiting RhoA, whereas inactivation of Crk by the ADPRT domain results in inhibition of the late steps of cytokinesis by blocking the recruitment of syntaxin-2 to the midbody. These findings reveal that P. aeruginosa possesses multiple strategies to inhibit wound healing and provide an example of a bacterial pathogen targeting cytokinesis. Results
P
seudomonas aeruginosa, an important human pathogen, is ubiquitous in the environment. Despite constant exposure to P. aeruginosa, healthy individuals do not get infected. In the settings of epithelial cell injury and兾or immunocompromised hosts, however, P. aeruginosa is one of the leading causes of nosocomial infections (1). In addition, P. aeruginosa can chronically colonize the lungs of patients with cystic fibrosis, leading to persistent infections and lung damage (2). P. aeruginosa boasts a large arsenal of virulence factors that contribute to colonization and subsequent injury (1). Prominent among these is the type III secretion system (TTSS), a highly conserved protein export system that contributes to the virulence of a large number of Gram-negative pathogens. This syringe-like apparatus allows bacteria to directly translocate a set of toxins, termed effector proteins, into the eukaryotic host cell, where they modify signal transduction pathways to subvert the host immune response (3, 4). Although the secretion machinery itself is highly conserved, the effector proteins differ between different species or even, in the case of P. aeruginosa, among different clinical isolates. Four type III-secreted effectors have been identified in P. aeruginosa: ExoU, a phospholipase; ExoY, an adenylate cyclase; and ExoT and ExoS, which are bifunctional proteins (5–7). No strain possesses all four effectors, but virtually all strains encode and express ExoT (8), suggesting a more conserved role for this protein in the context of P. aeruginosa pathogenesis. Consistent with this notion, P. aeruginosa strains in which the ExoT gene is deleted exhibit reduced virulence and seem to be particularly defective in dissemination in mice (9–12). ExoT possesses an N-terminal GTPase-activating protein (GAP) domain with activity toward Rho family GTPases (13, 14) and is primarily responsible for the ExoT-mediated cell rounding, inhibition of cell migration, and antiinternalization activities (9, 14). At its C terminus, ExoT contains an ADP-ribosyl transferase (ADPRT) domain, recently shown to target the Crk adaptor proteins and to interfere with their ability to interact with www.pnas.org兾cgi兾doi兾10.1073兾pnas.0605949103
P. aeruginosa Inhibits Cytokinesis in a TTSS- and ExoT-Dependent Manner. We have previously shown that P. aeruginosa inhibits the
repair of wounded epithelium through ExoU-dependent and ExoU-independent mechanisms (17, 18). During studies designed to examine whether ExoU-independent epithelial cell injury is reversible, we observed that a portion of A549 human lung epithelial cells failed to complete cytokinesis, the final stage of cell division in which the daughter cells separate (Fig. 1A and Movie 1, which is published as supporting information on the PNAS web site). In these studies, we used scrape-wounded lung epithelial A549 cells that were infected with PA103 ⌬exoU (⌬U), a human lung isolate that contains an in-frame deletion in ExoU (Table 1, which is published as supporting information on the PNAS web site). Bacteria were removed 5 h postinfection, fresh medium containing antibiotics was added to kill the remaining bacteria, and wound healing was assessed by timelapse videomicroscopy after bacterial removal. No cytokinesis block was observed when A549 cells were infected with the isogenic pscJ mutant strain (19), which is defective in the TTSS and fails to secrete or translocate the type III-secreted effectors (Fig. 1 A and Movie 2, which is published as supporting information on the PNAS web site). Given that ExoU and ExoT are the only known effector proteins expressed in PA103 (6, 20) and that the observed Author contributions: S.H.S. and J.E. designed research; S.H.S. performed research; S.H.S. and J.E. analyzed data; and S.H.S. and J.E. wrote the paper. The authors declare no conflict of interest. This article is a PNAS direct submission. Abbreviations: TTSS, type III secretion system; ⌬U, PA103 ⌬exoU; ⌬U⌬T, PA103 ⌬exoU ⌬exoT; GAP, GTPase-activating protein; ADPRT, ADP-ribosyl transferase; PI, propidium iodide; IF, immunofluorescent. ‡To
whom correspondence should be addressed. E-mail:
[email protected].
© 2006 by The National Academy of Sciences of the USA
PNAS 兩 October 17, 2006 兩 vol. 103 兩 no. 42 兩 15605–15610
MICROBIOLOGY
Pseudomonas aeruginosa is an opportunistic pathogen that requires preexisiting epithelial injury to cause acute infections. We report that P. aeruginosa inhibits mammalian cytokinesis in a type III secretion system and exotoxin T (ExoT)-dependent manner. ExoT is a bifunctional type III secretion system effector protein that contains an N-terminal GTPase-activating protein domain and a C-terminal ADP-ribosyl transferase domain. Each of its domains inhibits cytokinesis in a kinetically, morphologically, and mechanistically distinct manner. The GTPase-activating protein-mediated inhibition of cytokinesis occurs early, likely as a consequence of its inhibitory effect on RhoA. The ADP-ribosyl transferase domain inhibits late steps of cytokinesis by blocking syntaxin-2 localization to the midbody, an event essential for completion of cytokinesis. These findings provide an example of a bacterial pathogen targeting cytokinesis.
Fig. 1. P. aeruginosa inhibits cytokinesis of epithelial host cells in an ExoTdependent manner. (A) A549 cells were mechanically wounded as described (18) and infected with the indicated strains (multiplicity of infection ⬇10). Five hours postinfection, media containing bacteria were removed, and fresh medium containing amikacin was added to prevent further bacterial growth. Video images were captured every 5 min at ⫻50 magnification. (B) HeLa cells were infected with PA103⌬U⌬T expressing either the control vector or the ExoT expression vector in the presence of ZVAD-fmk (to block apoptotic cell death) and PI, a cell-impermeable nuclear dye (to identify dying cells). Images were captured every 5 min at ⫻200 magnification. (C) Primary human epithelial keratinocyte cells were infected with the indicated strains. Images were captured every 5 min at ⫻200 magnification.
cytokinesis block was independent of ExoU, we sought to examine the role of ExoT in this virulence function. We infected HeLa cells with the following isogenic strains (see Table 1): ⌬U; PA103 ⌬exoU ⌬exoT (⌬U⌬T), which contains in-frame deletions in both exoU and exoT; ⌬U⌬T complemented with an ExoT expression vector (⌬U⌬T⫹pExoT); or ⌬U⌬T containing pUCP20 empty vector (⌬U⌬T⫹pUCP20). The block in cytokinesis was observed only in the presence of ExoT-producing strains ⌬U or ⌬U⌬T⫹pExoT (Fig. 1B and Fig. 6, which is published as supporting information on the PNAS web site), indicating that ExoT was necessary for inhibition of cytokinesis. We performed these experiments in the presence of propidium iodide (PI), a membrane impermeant nuclear dye to identify dying cells, and ZVAD-fmk, a broad-spectrum inhibitor of apoptosis (21), to determine whether the P. aeruginosa-induced apoptotic cell death (22) was the cause of the observed cytokinesis failure under these conditions. The ExoT-dependent cytokinesis block occurred in the presence of ZVAD and in cells that did not stain with PI, ruling out cell death as the cause of the observed cytokinesis block (Fig. 1B). PA103⌬U was also able to inhibit cytokinesis in Madin–Darby canine kidney cells (data not shown) and in a primary human epithelial keratinocyte cell line (Fig. 1C), also in an ExoT-dependent manner. Thus, the ability of ExoT to inhibit cytokinesis was neither cell linedependent nor specific to transformed cells. 15606 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0605949103
Fig. 2. Each domain of ExoT inhibits cytokinesis in a distinct morphologic and kinetic manner. A549 cells were infected for 5 h with PA103⌬U expressing a functional GAP domain [⌬U兾ExoT(G⫹A⫺)], a functional ADPRT domain [⌬U兾T(G⫺A⫹)], or WT ExoT (⌬U) and processed as described in Fig. 1 A. Video images were captured every 5 min at ⫻50 magnification. Representative frames are shown, indicating that GAP and ADPRT inhibit cytokinesis in kinetically and phenotypically distinct manners.
Cytokinesis has been traditionally divided into four stages: cleavage furrow positioning during anaphase, actomyosin contractile ring assembly at the site of the cleavage furrow, ingression of the cleavage furrow, and abscission, in which the cytoplasmic bridge connecting the daughter cells is resolved (23, 24). Using time-lapse videomicroscopy, we carried out a careful kinetic and morphologic analysis to determine the contribution of each domain of ExoT to the observed block in cytokinesis. Remarkably, the GAP or the ADPRT activities inhibited cytokinesis with distinct morphologic and kinetic characteristics. P. aeruginosa expressing a functional GA P domain, ⌬U兾 T(G⫹A⫺), inhibited cytokinesis at two early but distinguishable stages of cytokinesis: either at 1–2 h postmetaphase, after the initiation of ingression, or immediately after furrow completion (4–6 h postmetaphase) (Fig. 2). In contrast, in cells infected with ⌬U兾T(G⫺A⫹), expressing a functional ADPRT domain, the block in cytokinesis occurred much later, at 10–20 h postmetaphase during the abscission phase. In these cells, cleavage furrowing was completed but the cytoplasmic bridge failed to undergo abscission, and the daughter cells sprang back together to form binculeated cells (Fig. 2). Cells that exhibited cytokinesis failure after infection with PA103⌬U (expressing WT ExoT) exhibited all three modes of inhibition, likely reflecting the phase in the cell cycle at which ExoT was translocated into the host cytoplasm (Fig. 2). It was difficult to quantify the ExoT-dependent inhibition of cytokinesis in a large population for the following reasons. First, P. aeruginosa does not uniformly infect epithelial cells of a monolayer in these in vivo settings (25, 26), and, second, the host cells were not synchronized. We used various strategies to synchronize cells in G1, S, or G2 (27) and to assess this virulence phenotype on a large scale by flow cytometry. We expected that the ExoT-mediated cytokinesis failure would result in an increase in the G2 subpopulation (cells with 4N nuclear content). Upon infection with P. aeruginosa, however, cells that were Shafikhani and Engel
synchronized in the G1 or the S phase became arrested in S phase and did not undergo mitosis, indicating that P. aeruginosa also causes a cell-cycle arrest (data not shown). Synchronization in G2兾M by nocodazole severely affected cell division and therefore could not be used to study cytokinesis (data not shown). Nevertheless, the fact that we observed cytokinesis failure only in the presence of the TTSS-proficient, ExoT-producing bacteria gave us the confidence to further investigate this phenomenon. mine whether ExoT, in the absence of other bacterial factors, was sufficient to inhibit cytokinesis, we transiently transfected HeLa cells with ExoT fused directly to, or coexpressed with, GFP (see Table 1 and Materials and Methods) and assessed its effect on cytokinesis by immunof luorescent (IF) time-lapse microscopy. We quantified cytokinesis block by tallying cytokinesis events in transfected cells collected from multiple movies. Similar results were obtained with either set of expression vectors, indicating that GFP fusion did not interfere with the activities of ExoT or its domains. When transfected with the pIRES-GFP control vector or with the ExoT double mutant, pExoT(G⫺A⫺)-GFP, only 2.7% (n ⫽ 100) or 4% (n ⫽ 72) of the transfected cells failed to complete cytokinesis, respectively (Fig. 3B and Movie 3, which is published as supporting information on the PNAS web site). In contrast, transfection with pExoT-GFP, pExoT(G⫺A⫹)-GFP, or pExoT(G⫹A⫺)GFP resulted in cytokinesis failure in 62% (n ⫽ 62), 51% (n ⫽ 43), or 41% (n ⫽ 34) of cells, respectively (Fig. 3B and Movies 4 – 6, which are published as supporting information on the PNAS web site) (P ⬍ 0.001, as compared with the control vector or to the ExoT double mutant, determined by the 2 test). Kinetic and morphologic analysis revealed results remarkably similar to what was observed with bacterially translocated ExoT protein. Transient expression of pExoT(G⫹A⫺)-GFP blocked cytokinesis at two discrete time points, either at 1–2 h (early) or at 4–6 h (mid) postmetaphase (Fig. 3A, Movie 4, and data not shown). In contrast, transfection with pExoT(G⫺A⫹)-GFP resulted in cytokinesis failure at later time points, 10–20 h postmetaphase (Fig. 3A and Movie 5). Cytokinesis failure was observed at all three time points in cells transfected with the WT ExoT (Fig. 3A, Movie 6, and data not shown). Together, these results indicated that either domain of ExoT was sufficient to cause cytokinesis failure in HeLa cells. Based on their different morphological and kinetic effects on cytokinesis, we conclude that the GAP and the ADPRT domains act at distinct steps in cytokinesis. RhoA activity is required for the early steps in cytokinesis, including the actomyosin ring assembly, and the ingression (28, 29). More recently, it has been shown that Citron K, a RhoAdependent kinase, is required immediately after the ingression (30–32). The GAP-mediated inhibition of cytokinesis closely mimics the well chronicled block in cytokinesis that occurs when either RhoA or Citron K is inhibited (30–32). This finding is consistent with, and would be predicted by, the known inhibitory effect of the GAP domain of ExoT on RhoA (13). However, the mechanism by which the ADPRT domain inhibited late steps of cytokinesis was not explained by the published data, and we thus investigated this further. ADPRT Activity Inhibits Cytokinesis by Blocking Syntaxin-2 Midbody Localization. Published studies have demonstrated that, at late
stages of cytokinesis, the v-SNARE Endobrevin兾VAMP8 and the t-SNARE Syntaxin-2 are localized to the midbody, a specialized structure in the cytoplasmic bridge connecting the daughter cells (33). Their function is required for the final step in cytokinesis, abscission, in which the membranes undergo Shafikhani and Engel
MICROBIOLOGY
Each Domain of ExoT Is Sufficient to Inhibit Cytokinesis. To deter-
Fig. 3. Each domain of ExoT is sufficient to inhibit cytokinesis. (A) HeLa cells were transiently transfected with the indicated mammalian expression vectors harboring ExoT variants fused to GFP at their C termini. ZVAD-fmk was included to inhibit cell death, and PI (red) was added to identify dying cells. Video images were captured every 15 min at ⫻200 magnification, except for pExoT(G⫺A⫺), which is shown at ⫻100 magnification so that multiple clusters of cells undergoing successful cytokinesis can be seen. The arrows indicate the dividing cells. (B) The tabulated results from multiple movies are shown. The numbers above each column indicate the total number of transfected dividing cells that were scored. *, P ⬍ 1 (2 test) when compared with the control vector, pEGFP; **, P ⬍ 0.001 (2 test) when compared with the control vector or to pExoT(G⫺A⫺).
separation. We have found that Crk adaptor proteins are essential in mammalian cytokinesis, as they localize to the midbody and function to recruit syntaxin-2 in this compartment late during cytokinesis (S.H.S., K. Mostov, and J.E., unpublished data). The fact that ExoT ADP-ribosylates both the CrkI and CrkII isoforms and interferes with their functions in focal adhesion assembly and cell migration (15, 34) prompted us to investigate whether the ADPRT-mediated inhibition of cytokinesis also occurred as a consequence of its activity on Crk. We first tested whether the ADPRT domain affected Crk localization in the midbody. HeLa cells were transiently transfected with informative ExoT-GFP derivatives, and Crk midbody localization during cytokinesis was analyzed by IF. For these experiments, we used an antibody against Citron K, another component of the midbody (28, 31), PNAS 兩 October 17, 2006 兩 vol. 103 兩 no. 42 兩 15607
Fig. 4. ExoT localizes to the midbody, and its ADPRT activity alters Crk localization at the midbody. HeLa cells were transiently transfected with the indicated mammalian expression vectors harboring ExoT-GFP variants in the presence of ZVAD-fmk. Fifteen hours posttransfection, cells were fixed and stained with Citron K (blue), Crk (red), and GFP (green). Note that the Crk staining in the midbody is substantially reduced and appears split in the presence of functional ADPRT, despite significantly higher exposure (compare the Crk staining, red panels, in the cell bodies).
to identify midbodies. When transfected with the vector alone or the pExoT(G⫺A⫺)-GFP double mutant, Crk completely colocalized with Citron K (Fig. 4). However, in the presence of WT ExoT (pExoT) or a functional ADPRT domain, pExoT(G⫺A⫹), the Crk staining at the midbody was split and substantially reduced, despite significantly longer exposure time [Fig. 4; see Crk (red) and compare the cytoplasmic Crk staining among the panels]. Interestingly, ExoT-GFP variants were also present at the midbody (Fig. 4, GFP). Because Crk enrichment at the midbody was visible only after the cleavage furrow completion and the GAP-mediated block of cytokinesis occurred before that time point (Fig. 2), we did not examine the effect of GAP domain on Crk localization. If the ADPRT-mediated inhibition of cytokinesis occurs as a consequence of its inhibitory effect on Crk, we expected syntaxin-2 midbody localization to be similarly inhibited by the ADPRT activity of ExoT. To examine this prediction further, we first assessed syntaxin-2 midbody localization after infection with informative isogenic mutants of PA103⌬U. As shown in Fig. 5A, we found syntaxin-2 midbody localization to be blocked only in the presence of the WT ExoT- or the ADPRT (ExoT(G⫺A⫹))producing bacteria. We further confirmed these results by transfecting HeLa cells with pExoT-GFP, pExoT(G⫺A⫹)-GFP, or pExoT(G⫺A⫺)-GFP. Transfected cells were fixed and stained with antibodies to GFP (Fig. 5B, green) and syntaxin-2 (Fig. 5B, red). We identified midbodies by staining with an antibody to RhoA (Fig. 5B, blue), whose subcellular localization pattern during cytokinesis closely resembles Citron K (data not shown). Indeed, expression of pExoT-GFP or pExoT(G⫺A⫹)GFP prevented syntaxin-2 localization to the midbody in 73% or 70% of the transfected cells, respectively (Fig. 5 B and C; P ⬍ 0.001, 2 test). In contrast, syntaxin-2 localized to the midbody in 100% of HeLa cells transfected with pExoT(G⫺A⫺)-GFP double mutant or the control vector. Collectively, these data suggest that ADP-ribosylation of Crk by ExoT interferes with Crk localization to the midbody and with its ability to recruit syntaxin-2 during cytokinesis, thus providing a mechanism by which ADPRT inhibits cytokinesis. Discussion In response to injury, epithelial cells that line the mucosal barrier initiate a programmed series of separate yet interdependent responses that include cell migration, cell division, and matrix 15608 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0605949103
Fig. 5. Syntaxin-2 midbody localization is inhibited by the ExoT ADPRT activity. (A) HeLa cells were infected with the indicated strains. Five hours postinfection, the bacteria were removed, and fresh medium containing amikacin was added to kill extracellular bacteria. Three hours after antibiotic addition, the attached cells were fixed and stained with nuclear die DAPI (blue), -tubulin (red), and syntaxin-2 (green). (B) HeLa cells were transiently transfected with the indicated mammalian expression vectors harboring ExoTGFP variants in the presence of ZVAD-fmk. Fifteen hours posttransfection, cells were fixed and stained with RhoA (blue) to identify dividing cells, syntaxin-2 (red), and GFP (green). Note that syntaxin-2 localization to the midbody is not observed in cells transfected with pExoT-GFP or pExoT(G⫺A⫹)-GFP, despite increased exposure times. (C) Tabulated results from multiple events described in B. The numbers above each column indicate the total number of dividing cells examined. *, P ⬍ 1 (2 test) when compared with the control vector, pEGFP; **, P ⬍ 0.001 (2 test) when compared with the control, pEGFP, or to the pExoT(G⫺A⫺) vectors.
assembly (35). Cell division is necessary to replace the cell loss after an injury and thus is an important aspect of epithelial wound repair (24). The requirement for preexisting injury and兾or immunocompromise makes inhibition of wound healing paramount to the success of P. aeruginosa as an opportunistic pathogen. In this article, we have demonstrated that P. aeruginosa uses ExoT to impair epithelial wound repair by targeting cytokinesis. We used time-lapse videomicroscopy to carry out a careful morphologic and kinetic analysis of the inhibition of cytokinesis by ExoT, under conditions in which either the bacteria translocated ExoT into the target cell or when ExoT was expressed in target cells by transient transfection. These complementary Shafikhani and Engel
Materials and Methods Bacterial Strains and Media. All strains and plasmids are shown in Table 1. All cloning steps were done in chemically competent DH5␣ (Stratagene, La Jolla, CA) in the presence of appropriate antibiotics (50 g兾ml kanamycin or 50 g兾ml carbenicillin). Shafikhani and Engel
Except where noted, P. aeruginosa strains expressing ExoT or its mutant versions were grown in LB without shaking at 37°C overnight. Bacterial counts were determined by optical density and confirmed by plating serial dilutions for colony counts. Antibodies and Reagents. Mouse monoclonal anti-Crk antibody was purchased from BD Transduction Laboratories (Lexington, KY) and mouse monoclonal anti--tubulin was from Sigma (St. Louis, MO). Mouse monoclonal anti-RhoA and goat polyclonal anti-Citron K antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit polyclonal anti syntaxin-2 antibody was from Synaptic Systems (Goettingen, Germany). Goat and rabbit polyclonal anti-GFP antibodies were purchased from Novus Biologicals (Littleton, CO). The secondary antibodies, used in IF microscopy, were the Alexa-conjugated IgG series, purchased from Molecular Probes (Carlsbad, CA). Growth and Infection of Epithelial Cell Lines. In all experiments, wells or coverslips were precoated with poly(L-lysine) (Sigma) for 5 min at room temperature, followed by human fibronectin (40 g兾ml) for 20 min before cell seeding. HeLa, A549, and Madin–Darby canine kidney epithelial cell lines were grown and maintained as described (18). HEK primary cells and HEK growth medium were purchased from Cell Applications, Inc. (San Diego, CA) and were used in described experiments after the second passage. For experiments involving videomicroscopy, cells were grown essentially in the same media, except phenol red was omitted. Cell Wounding Assays. Wounding experiments were done as de-
scribed (18). Construction of Expression Vectors for Transient Transfection. Restriction enzymes were purchased from New England Biolabs (Ipswich, MA). Primer sequences shown in lowercase below are not homologous to the chromosomal sequences and contain the engineered restriction sites. The pIRES2-EGFP mammalian vector (Clontech, Mountain View, CA) was modified as follows. Using long PCR (37) and primers EGFP-f3 (5⬘-aaaactagtATGGTGAGCA AGGGCGAGG-3⬘) and EGFP-r2 (5⬘aaaagctagcGGATCTGACGGTTCAC-3⬘), which contain SpeI and NheI, respectively, the intervening sequences between the promoter and the EGFP-coding sequence including the IRES were removed. To construct ExoT, ExoT(G⫹A⫺), ExoT(G⫺A⫹), and ExoT(G⫺A⫺) C-terminal fusion to GFP, primers SS25-2 (5⬘cccgctagcatgctaATTCAATCATCTCAGC-3⬘), with the engineered NheI site, and SS26-EGFP-1 (5⬘-caaaaaaactagtGGCCAGGTCGAGGC-3⬘), with SpeI engineered site, were used to amplify W T ExoT, ExoT(G⫺A⫹), ExoT(G⫹A⫺), and ExoT(G⫺A⫺), using a PCR program described previously (38), and the pUC plasmid vector series, containing these ExoT mutant forms as the template (18). These PCR products were then digested with NheI and SpeI and directionally cloned into the modified pIRES2 vector. Transient Transfection. For transient transfection studies, we used
Effectene (Qiagen, Valencia, CA), using the vendor’s protocol. IF Microscopy. Fifteen to twenty-four hours posttransfection, the
attached cells were fixed with 10% trichloroacetic acid (Fisher Scientific, Fair Lawn, NJ). The primary antibodies were added at 1:50 dilution, except the mouse monoclonal anti--tubulin (1:200) and goat and rabbit polyclonal anti-GFP (1:300) and incubated overnight at 4°C. The secondary antibodies (1:400 dilution) and Hoechst 33342 nuclear stain (5 g兾ml) were added in the same blocking solution and incubated for 1 h at 37°C. Images were captured at a magnification of ⫻1,000 under oil PNAS 兩 October 17, 2006 兩 vol. 103 兩 no. 42 兩 15609
MICROBIOLOGY
approaches revealed identical results. Regardless of the mode of delivery, no cytokinesis block was observed when both domains of ExoT were inactivated or when ExoT was not present, indicating that ExoT is both necessary and sufficient for this virulence function. Moreover, the presence of either domain activity in host cells, whether delivered by bacteria or supplied on expression vectors, was sufficient to inhibit cytokinesis, but with distinct morphological and kinetic features. The GAP activity inhibited early steps in cytokinesis, and the ADPRT activity blocked late steps. Animal cytokinesis initiates by microtubule-directed positioning and assembly of the actomyosin contractile ring at the equatorial cell cortex during anaphase. A number of proteins, including RhoA, have been shown to partake in this delicate step (29). During ingression, the next step in cytokinesis, the GTP-bound active RhoA recruits and activates a number of proteins, including Citron K, culminating in the formation of the cleavage furrow (36). Not surprisingly, we found that the GAP domain of ExoT, a potent inhibitor of RhoA (9, 13), inhibits cytokinesis in a manner that phenotypically and kinetically mimics the cytokinesis failures observed when RhoA or RhoA-dependent Citron K is inhibited (30 –32). The GAP-mediated inhibition of cytokinesis occurs early at the onset of the ingression, 1–2 h postmetaphase, or during the furrowing step, 4 – 6 h postmetaphase. The ADPRT activity of ExoT specifically targets postcleavage furrow steps in cytokinesis. Cells failing cytokinesis in the presence of the ADPRT activity complete furrowing but fail to resolve the midbody and eventually fuse to form binucleated cells. This mode of cytokinesis block is similar to the cytokinesis failure observed in cells expressing dominant negative mutants of syntaxin-2 or endobrevin (33). We have found that Crk adaptor proteins localize to the midbody after cleavage furrow formation and are required to recruit syntaxin-2 (S.H.S., K. Mostov, and J.E., unpublished results). As would be predicted from a toxin that inhibits Crk activity (16), we demonstrate that, in the presence of the ADPRT activity, Crk enrichment at the midbody is substantially reduced (Fig. 4) and that syntaxin-2 fails to properly localize to the midbody (Fig. 5). Interestingly, ExoT variants also colocolize with Crk at the midbody. These findings provide a mechanism for the ADPRT inhibition of cytokinesis. The success of P. aeruginosa as a pathogen stems from its multiple fail-safe strategies to subvert its host cellular processes. These strategies likely have evolved to permit its survival in diverse ecological niches where it constantly faces single and multicellular organisms. Crucial to its success as a human opportunistic pathogen is its ability to inhibit epithelial wound repair. Previously published studies have demonstrated that the GAP and ADPRT domains of ExoT work together to inhibit cell migration, a key step in epithelial wound repair (17, 18). We now show that these domain activities also combine to effectively block cell division, another important cellular process in wound repair, by targeting cytokinesis at multiple steps. The end result is that this organism is able to both exploit preexisting tissue injury and perpetuate it, so as to enhance its ability to colonize and facilitate its spread to local tissues and distant organs. The combination of the potent cytotoxic activities of ExoU and ExoS, the cytoskeleton and the cell-cycle-disrupting activities of ExoT, and the high rate of intrinsic and acquired antibiotic resistance contribute to make P. aeruginosa a formidable and ever challenging human pathogen in demand of new therapeutic strategies.
immersion with a Nikon (Tokyo, Japan) Eclipse TE 2000-E fluorescence microscope, using a CCD camera and processed with Simple PCI imaging software. All video time-lapse microscopy studies were performed as described (18). Video images were captured every 5 min for phase only, or 15 min for phase and IF, by using a Zeiss (Jena, Germany) Axiovert microscope connected to a Cohu (San Diego, CA) video camera and analyzed by using OpenLab software (Improvision, Lexington, MA). Movies were further compressed by using Adobe Premiere 6.5 software (Adobe Systems, San Jose, CA).
Statistical Analysis. Web Chi-Square Calculator (http:兾兾schnoodles. com兾cgi-bin兾web chi.cgi) was used to determine the statistical significance by 2 test. P ⱕ 0.05 was taken as significant.
1. Engel JN (2003) in Severe Infections Caused by Pseudomonas aeruginosa, eds Hauser A, Rello J (Kluwer兾Plenum, New York), pp 201–230. 2. Mandell GL, Bennett JE, Dolin R (2000) Principles and Practice of Infectious Diseases (Churchill Livingstone, New York). 3. Mota LJ, Sorg I, Cornelis GR (2005) FEMS Microbiol Lett 252:1–10. 4. Hueck C (1998) Microbiol Mol Biol Rev 62:379–433. 5. Sato H, Frank DW (2004) Mol Microbiol 53:1279–1290. 6. Yahr TL, Vallis AJ, Hancock MK, Barbieri JT, Frank DW (1998) Proc Natl Acad Sci USA 95:13899–13904. 7. Barbieri JT, Sun J (2004) Rev Physiol Biochem Pharmacol 152:79–92. 8. Feltman H, Schulert G, Khan S, Jain M, Peterson L, Hauser AR (2001) Microbiology 147:2659–2669. 9. Garrity-Ryan L, Kazmierczak B, Kowal R, Commolli J, Hauser A, Engel J (2000) Infect Immun 68:7100–7113. 10. Hauser AR, Kang PJ, Fleiszig SJM, Mostov K, Engel J (1998) Infect Immun 66:1413–1420. 11. Lee VT, Smith RS, Tummler B, Lory S (2005) Infect Immun 73:1695–1705. 12. Shaver CM, Hauser AR (2004) Infect Immun 72:6969–6977. 13. Kazmierczak B, Engel J (2002) Infect Immun 70:2198–2205. 14. Krall R, Schmidt G, Aktories K, Barbieri JT (2000) Infect Immun 68:6066– 6068. 15. Sun J, Barbieri JT (2003) J Biol Chem 278:32794–32800. 16. Deng Q, Sun J, Barbieri JT (2005) J Biol Chem. 280:35953–35960. 17. Geiser T, Kazmierczak B, Garrity-Ryan L, Matthay M, Engel J (2001) Cell Microbiol 3:223–236. 18. Garrity-Ryan L, Shafikhani S, Balachandran P, Nguyen L, Oza J, Jakobsen T, Sargent J, Fang X, Cordwell S, Matthay MA, Engel JN (2004) Infect Immun 72:546–558.
19. Kang PJ, Hauser AR, Apodaca G, Fleiszig S, Wiener-Kronish J, Mostov K, Engel JN (1997) Mol Microbiol 24:1249–1262. 20. Fleiszig SMJ, Wiener-Kronish JP, Miyazaki H, Vallas V, Mostov K, Kanada D, Sawa T, Yen TSB, Frank D (1997) Infect Immun 65:579–586. 21. Garcia-Calvo M, Peterson EP, Leiting B, Ruel R, Nicholson DW, Thornberry NA (1998) J Biol Chem 273:32608–32613. 22. Hauser A, Engel J (1999) Infect Immun 67:5530–5537. 23. Glotzer M (2003) Curr Opin Cell Biol 15:684–690. 24. Wolf WA, Chew TL, Chisholm RL (1999) Cell Mol Life Sci 55:108–120. 25. Dacheux D, Attree I, Toussaint B (2001) Infect Immun 69:538–542. 26. Apodaca G, Bomsel M, Lindstedt R, Engel J, Frank D, Mostov K, WienerKronish J (1995) Infect Immun 63:1541–1551. 27. Cooper S (2003) Cell Mol Life Sci 60:1099–1106. 28. Bement WM, Benink HA, von Dassow G (2005) J Cell Biol 170:91–101. 29. D’Avino PP, Savoian MS, Glover DM (2005) J Cell Sci 118:1549–1558. 30. Naim V, Imarisio S, Di Cunto F, Gatti M, Bonaccorsi S (2004) Mol Biol Cell 15:5053–5063. 31. Eda M, Yonemura S, Kato T, Watanabe N, Ishizaki T, Madaule P, Narumiya S (2001) J Cell Sci 114:3273–3284. 32. Echard A, Hickson GR, Foley E, O’Farrell PH (2004) Curr Biol 14:1685–1693. 33. Low SH, Li X, Miura M, Kudo N, Quinones B, Weimbs T (2003) Dev Cell 4:753–759. 34. Sundin C, Hallberg B, Forsberg A (2004) FEMS Microbiol Lett 234:87–91. 35. Fenteany G, Janmey PA, Stossel TP (2000) Curr Biol 10:831–838. 36. Burgess DR (2005) Curr Biol 15:R310–R311. 37. Shafikhani S, Siegel RA, Ferrari E, Schellenberger V (1997) BioTechniques 23:304–310. 38. Shafikhani S (2002) Environ Microbiol 4:482–486.
15610 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0605949103
We thank Andy Finch and the Mt. Zion Cancer Center for technical assistance on videomicroscopy. We thank Drs. Zena Werb, Pat O’Farrell, Eric Brown, Daniel Portnoy, Zhien Wang, and Jessica Ray and members of the Engel laboratory for advice and suggestions. This work was supported by National Institutes of Health Grants F32 AI054056 (to S.H.S.) and R01 AI053194 and R01 AI42806 (to J.E.).
Shafikhani and Engel
