|Home | About | Journals | Submit | Contact Us | Français|
Salmonella attachment to the intestinal epithelium triggers delivery of bacterial effector proteins into the host cytosol through a Type III Secretion System (T3SS), leading to pronounced membrane ruffling and macropinocytic uptake of attached bacteria. The tip of the T3SS is comprised of two proteins, SipB and SipC, which insert into the host plasma membrane forming a translocation pore . Both the N- and C-termini of SipC are exposed in the host cytosol and have been shown to directly modulate actin cytoskeleton assembly . We have identified a direct interaction between SipC and Exo70, a component of the exocyst complex, which mediates docking and fusion of exocytic vesicles with the plasma membrane. Here we show that exocyst components co-precipitate with SipC and accumulate at sites of invasion by Salmonella typhimurium. Exocyst assembly requires activation of the small GTPase RalA, which we show is triggered during Salmonella infection by the translocated effector, SopE. Knockdown of RalA or Sec5 results in reduced membrane ruffling at sites of attachment and impairs bacterial entry into host cells. These findings suggest that S. typhimurium enhances invasion efficiency by promoting localized membrane expansion, directly through SipC-dependent recruitment of the exocyst and indirectly via SopE-dependent activation of RalA.
SipC is thought to adopt a hairpin topology, with both N- and C-termini exposed in the host cytosol (Figure 1A). In addition to its function in protein translocation, the SipC C-terminus has been shown to nucleate actin filament assembly, while the N-terminus has filament bundling activity . To identify eukaryotic proteins that interact with SipC, we performed a yeast two-hybrid screen using the cytoplasmic C-terminus (amino acids 200-410, SipC-C) of the protein as bait. This screen yielded multiple interactors encoding the N-terminus of Exo70, a component of the eukaryotic exocyst complex, which targets and tethers exocytic vesicles to the plasma membrane prior to SNARE-mediated fusion . The carboxy-terminus of Exo70 contains several highly conserved, positively charged amino acids that enable electrostatic interactions with phosphatidylinositol 4,5-bisphosphate and localization to the plasma membrane [4, 5]. The C-terminus is also capable of interacting with the Arp2/3 complex , which is important for actin filament nucleation and branching. As a result, overexpression of Exo70 has been shown to induce the formation of filopodia in transfected cells [6, 7].
To verify the interaction between SipC-C and Exo70, Hela cells transfected with a GFP-tagged Exo70 construct were lysed and incubated with purified recombinant 6xHis-tagged SipC-C. After precipitation with nickel agarose (Ni-NTA), Exo70-GFP, but not GFP alone, was found to interact with recombinant SipC-C (Figure 1B, upper panels). This interaction was also observed with endogenous Exo70 from cell lysates (Figure 1C). No Exo70 was detected in control affinity precipitations with the 6xHis-tagged SipC N-terminal domain (aa 1-120, SipC-N) or with Ni-NTA in the absence of SipC-C.
Exo70 is one of eight proteins that associate to form the exocyst complex. As shown in Figure 1B (lower panels), SipC-C precipitates also contained endogenous Sec5 and Sec8, indicating that SipC-C can also bind to Exo70 when it is assembled into the octameric exocyst complex.
Finally, to confirm that the interaction between SipC and Exo70 is direct, recombinant 6xHis-tagged SipC-C and SipC-N pre-bound to Ni-NTA were incubated in vitro with purified recombinant Exo70. In agreement with the pulldown assays described above, we found that recombinant Exo70 associates specifically with SipC-C (Figure 1D). Taken together, these results suggest that SipC-C interacts directly with Exo70, thereby coupling the bacterial invasion machinery to the mammalian exocyst complex.
Exocyst components exist as subcomplexes that are brought together to mediate vesicle docking at specific sites on the plasma membrane. The complex present on carrier vesicles is thought to include Sec5, Sec6, Sec8, Sec10, Sec15, and Exo84 while Sec3 and Exo70 are thought to associate with the plasma membrane, where they provide spatial landmarks for vesicle targeting [8–10]. Delivery of exocytic vesicles to the plasma membrane leads to assembly of these subcomplexes into an intact, octameric vesicle tethering complex. To determine if both subcomplexes are recruited to sites of Salmonella entry into host cells, Hela cells expressing GFP-tagged Sec5, Sec8, or Sec10 were infected with S. typhimurium for 30 minutes. Fixed cells were then immunolabeled to detect the actin-rich membrane ruffles of invasion foci (phalloidin), the invading bacteria (anti-LPS), and the indicated exocyst protein (Figure 2 and Figure S1).
In agreement with the biochemical data, endogenous Exo70 became highly enriched at sites of bacterial attachment where it could be seen in close association with invading bacteria (Figure 2A). Similar enrichment was observed for the vesicle-associated subunits Sec10-GFP (Figure 2B), Sec5-GFP and Sec8-GFP (Figure S1A). Deconvolution microscopy verified the close association of Sec5-GFP with S. typhimurium (Supplemental Movies and Figure S1B), and confocal microscopy of single optical sections confirmed that the apparent enrichment of exocyst components around attached bacteria was not a result of increased cell thickness at invasion foci (Figure S2A). These observations suggest that exocytic vesicles are targeted directly to sites of bacterial engagement with the plasma membrane. In rare instances, exocyst components could be observed on vacuoles containing Salmonella that also labeled with the early endosomal marker EEA1 indicating that the exocyst association may persist during the early stages of vacuole biogenesis (Figure S2B).
Assembly of the exocyst requires activation of the small GTPase, RalA, which interacts directly with both the Exo84 and the Sec5 subunits . Immunostaining of infected Hela cells revealed that, in addition to the various exocyst subunits, endogenous RalA also became concentrated at sites of Salmonella invasion (Figure 2C). To determine if Salmonella infection triggers the activation of RalA, we made use of a previously described pulldown assay that takes advantage of the high affinity of GTP-bound RalA for Sec5 [11, 12]. Infection with S. typhimurium led to a rapid and sustained increase in RalA activation. Treatment of cells with a known RalA agonist, epidermal growth factor, generated a quantitatively similar level of activation (Figure 3A).
It is well established that the Salmonella effector proteins SopE and SopE2 are able to activate the Rho family GTPases Rac and Cdc42 by acting directly as guanine nucleotide exchange factors. A third Salmonella effector, the inositol phosphatase SopB, activates Rac indirectly, through the activation of an upstream Rac regulator, RhoG . To determine if Salmonella-mediated activation of RalA similarly requires translocated effector proteins, cells were infected with either wild-type S. typhimurium SL1344 or a T3SS mutant (ΔinvG) which cannot translocate bacterial effector proteins into the host cytosol . While infection with the wild-type strain induced a robust activation of RalA, infection with the ΔinvG strain had virtually no effect (Figure 3B), indicating that activation required one or more translocated effector proteins.
To identify the effector(s) responsible for RalA activation, we tested a panel of isogenic strains lacking SopE, SopE2, or SopB. The strain lacking SopE (ΔsopE) induced no detectable activation of RalA, while the strain lacking SopB (ΔsopB) activated RalA to a level that was indistinguishable from that induced by the wild-type strain (Figure 3C). The strain lacking SopE2 was slightly attenuated in its ability to activate RalA, suggesting that while SopE is quantitatively more important, both SopE and SopE2 contribute to Ral activation during Salmonella infection.
To determine if recruitment of the exocyst complex is functionally important for bacterial entry, cells were depleted of Exo70, Sec5 or RalA using siRNA, and invasion efficiency was quantified using a flow cytometry-based assay (Figure 4A). All three knockdowns resulted in a significant inhibition of bacterial internalization by host cells (38+/− 10% for Sec5; 41.9+/− 4.5% for Exo70; 40+/− 4.5% for RalA). Moreover, the inhibitory effects were quantitatively similar, suggesting that all three proteins act in the same pathway. It should be noted that, although multiple siRNAs were tested for each target protein, we were never able to achieve complete knockdown of Sec5, Exo70 or RalA (Figure S4). It is therefore possible that the observed inhibitory effects underestimate the role of the exocyst in Salmonella entry. Importantly, the inhibition of internalization observed upon Sec5 depletion could be reversed by expression of an siRNA-resistant construct, demonstrating that this defect is not due to off-target effects (Figure 4C).
Given that RalA has other functions in addition to its role in exocyst assembly, we also tested the requirement for RalA interaction with Sec5 using a Sec5 mutant (T11A) that fails to bind Ral . As shown in Figure 4C, expression of Sec5 T11A in Sec5-depleted cells failed to restore invasion efficiency to wild-type levels, indicating that the primary role of RalA in Salmonella entry is in the regulation of exocyst assembly.
The exocyst has been implicated in the delivery of new membrane to sites of membrane expansion in a variety of contexts [6, 16–19]. Since Salmonella invasion triggers localized membrane ruffling and macropinocytosis, we hypothesized that the bacteria use the SipC-mediated interaction with Exo70 to direct membrane traffic to sites of bacterial attachment. One prediction of this hypothesis is that inhibition of exocyst function should result in reduced ruffling at invasion foci due to impaired membrane expansion. Using quantitative fluorescence microscopy, we measured the size of the actin-rich foci associated with invading bacteria (Figure S3C). Knockdown of either Sec5 or RalA with siRNA led to a quantitatively similar reduction (35–40%) in the size of Salmonella invasion foci relative to controls (Figure 4B). Simultaneous knockdown of both Sec5 and RalA did not have additive effects (Figure S3B), again indicating that these proteins act in the same pathway. Interestingly, impaired focus formation was readily observed in foci containing 1 to 4 bacteria but became less apparent as the number of bacteria increased, suggesting that cooperative interactions among attached bacteria are sufficient to overcome the requirement for exocyst function. In agreement with the invasion assay described above, the size of invasion foci was restored to wild-type levels by expression of siRNA-resistant wild-type Sec5, but not by Sec5 T11A (Figure 4D). Taken together, these observations suggest that interaction between RalA and the exocyst is necessary for efficient Salmonella internalization.
Many Gram-negative bacterial pathogens, including Salmonella, Shigella, Yersinia, Pseudomonas, and enteropathogenic Escherichia coli (EPEC) utilize Type III secretion systems to inject an array of virulence factors into host cells . It is becoming increasingly clear that translocon components such as SipC and the homologous Shigella protein, IpaC, have effector functions in addition to their role in protein translocation. Both SipC and IpaC have been shown to stimulate actin remodeling directly, by nucleating actin filament assembly [2, 21]. IpaC can also promote actin reorganization indirectly by recruiting c-Src to its exposed C-terminus where the activity of the protein is necessary for efficient entry of S. flexneri into host cells .
To identify host proteins that interact with SipC, we conducted a yeast two-hybrid screen using the 200 amino acid SipC C-terminus as bait. Here we show that SipC interacts directly with the exocyst component, Exo70. Along with Sec3, Exo70 is thought to mark specific sites on the plasma membrane for the delivery of exocytic vesicles. A role for the exocyst in membrane expansion has been reported in other systems including phagosome biogenesis , lamellipodia formation [6, 23], axonal growth cone expansion , yeast bud expansion , and cytokinesis . Our data suggest that insertion of the translocon into the host membrane recruits Exo70, thereby promoting a redirection of exocytic vesicle traffic to sites of bacterial attachment. The localized delivery of these vesicles would then provide the additional new membrane required for macropinocytic uptake of the bound bacteria. This hypothesis is supported by our finding that the size of invasion foci induced by S. typhimurium is reduced in cells depleted of endogenous Sec5 or RalA, and that this correlates with the observed impairment in bacterial uptake. However, Exo70 is unique among the eight exocyst subunits in its ability to activate the Arp2/3 complex and stimulate actin filament assembly. Therefore, we cannot exclude that Exo70 also acts in this capacity during Salmonella invasion.
The exocyst consists of two hemicomplexes; one that marks the vesicle docking site on the plasma membrane and another that assembles on the exocytic vesicle. Assembly of the two hemicomplexes, which is necessary for vesicle docking to the plasma membrane, requires local activation of RalA . Here we demonstrate that Salmonella infection stimulates the activation of RalA, predominantly through the translocated effector protein SopE, and to a lesser extent, the highly related SopE2. These proteins are known to mimic eukaryotic guanine exchange factors: SopE has been shown to facilitate activation of both Rac and Cdc42 directly , while SopE2 appears to act more specifically on Cdc42 . Because SopE exhibits fairly promiscuous substrate specificity in vitro , it is likely that SopE acts directly on RalA. However, our data do not exclude the possibility that SopE acts upstream of RalA to activate it indirectly.
Although the exocyst complex has been shown to be important for phagosome biogenesis , our findings provide the first evidence for pathogen-directed recruitment of the exocyst to facilitate bacterial invasion of host cells.
The details of experimental methods are provided in Supplemental Experimental Procedures. Briefly, the yeast two-hybrid screen was performed using residues 200-410 of SipC as bait, and 3 ×106 colonies of a 7-day mouse embryonic library were screened. Ral activation assays were performed using a GST fusion containing the Ral-binding domain of Sec5. Bacterial internalization was assayed by flow cytometry, using S. typhimurium expressing GFP. Morphometric analysis of invasion foci was conducted using ImageJ software from randomly selected fields of infected cells.
We thank David Castle, Martin Schwartz and Nagaraj Balasubramanian for stimulating discussions, and David Castle for careful reading of the manuscript. We also thank Beth McCormick, Stanley Falkow, Edouard Galyov and Bobby Cherayil for Salmonella strains, and Charles Yeaman, Wei Guo, and Michael White for reagents and constructs. This research was funded by a National Institutes of Health grant (DK58536) to J.E.C. and a predoctoral training grant (T32 GM008136) to C.D.N.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.