Role of the S. typhimurium Actin-Binding Protein SipA in Bacterial Internalization

See allHide authors and affiliations

Science  26 Mar 1999:
Vol. 283, Issue 5410, pp. 2092-2095
DOI: 10.1126/science.283.5410.2092


Entry of the bacterium Salmonella typhimurium into host cells requires membrane ruffling and rearrangement of the actin cytoskeleton. Here, it is shown that the bacterial protein SipA plays a critical role in this process. SipA binds directly to actin, decreases its critical concentration, and inhibits depolymerization of actin filaments. These activities result in the spatial localization and more pronounced outward extension of the Salmonella-induced membrane ruffles, thereby facilitating bacterial uptake.

Entry of Salmonella typhimurium into host cells requires a bacterially encoded type III protein secretion system (1). Upon its activation after bacterial contact with the host, this system delivers several bacterial proteins into the host cell cytosol (2). These effector proteins trigger host cell signaling pathways that lead to profuse actin cytoskeleton rearrangements and membrane ruffling, ultimately resulting in bacterial internalization (3). One such protein is SopE (4, 5), which activates small guanosine triphosphatases (GTPases) of the Rho subfamily such as CDC42 and Rac-1.

In a search for other bacterial effectors, we focused on SipA (6), a substrate of the type III secretion system that shares amino acid sequence similarity to IpaA, aShigella spp. protein that is necessary for efficient bacterial entry into epithelial cells (7). We examined the interaction of a S. typhimurium sipA null mutant strain with cultured HeLa epithelial cells using an assay that allows the examination of both actin cytoskeleton reorganization and bacterial internalization after short infection times (8). The mutant was less effective than the wild-type strain in inducing actin cytoskeletal reorganization, and the rearrangements were diffuse rather than localized (Fig. 1, A and B). In addition, the internalized mutant bacteria appeared to be evenly distributed within the infected cell, whereas the wild-type strain was localized in clusters. A plasmid encoding SipA rescued the mutant phenotype (Fig. 1C). The sipA mutant was also impaired in its ability to be internalized by cultured epithelial cells after short infection times (Table 1). Over time, the defect in internalization was reduced, and beyond 30 min of infection there was almost no measurable difference between the entry levels of the wild type and the sipA mutant (Table 1). These results indicate that SipA is involved in the induction of localized actin cytoskeleton rearrangements and is required for efficient S. typhimurium entry into cultured epithelial cells.

Figure 1

Requirement of SipA for S. typhimuriuminduction of localized actin cytoskeleton rearrangements in HeLa cells. Cells were infected with wild-type S. typhimurium(A), the sipA mutant (B), thesipA mutant complemented with psipA (a plasmid carrying sipA) (C), or the invasion-deficientinvA mutant (D). The actin cytoskeleton was visualized by phalloidin staining (green), which stains F-actin.Bacteria (red) were stained with a polyclonal antibody to theS. typhimurium O-antigen (Difco, Detroit). Images were captured with a Hamamatsu C2400 charge-coupled device camera and pseudocolored. Exposure times for all the images are the same. Scale bar, 20 μm.

Table 1

Requirement of SipA for efficient S. typhimurium entry into HeLa cells. Bacterial internalization was measured by fluorescence microscopy (7). Values are averages ± SE of three independent experiments and have been normalized to the internalization level of wild-type bacteria after 30 min of infection, which was considered to be 100% (in this case, 1597 ± 185 bacteria internalized per 100 cells). A minimum of 250 cells were counted for each strain per time point. The Pvalues for the difference between the entry levels of the wild type and the sipA strains are as follows: 5 min, P = 0.002; 10 min, P = 0.05; 20 min, P = 0.03. IT, infection time.

View this table:

To search for cellular proteins that interact with SipA, we used a glutathione S-transferase (GST) pull-down assay (9). HeLa cell lysates were incubated with GST-SipA or GST alone that had been immobilized on glutathione-agarose beads. The major GST-SipA–binding protein was actin (Fig. 2, A and B). The interaction of these proteins was shown to be direct (Fig. 2C) and was confirmed in a yeast two-hybrid assay (10). To examine whether the SipA-actin complex forms as a result of bacterial infection, we infected HeLa cells with wild-type S. typhimurium or an isogenic sipA mutant and looked for the complex by coimmunoprecipitation analysis (11). Actin was readily detected in SipA immunoprecipitates from cells infected with wild-type bacteria, but not in those from uninfected cells or cells infected with the sipA mutant (Fig. 2D).

Figure 2

Interaction of SipA with actin in vivo and in vitro. The interaction of SipA with F-actin in HeLa cell extracts was examined with a GST pull-down assay (8). Proteins bound to GST or GST-SipA beads were stained with Coomassie blue (A) or probed with an actin mAb (B, upper panel). As a control, the interaction of GST-SipA with vinculin was examined in a similar manner with an antibody to vinculin (B, lower panel). (C) The interaction of SipA with actin was also probed in extracts of E. coli expressing GST-SipA (8). The presence of F-actin bound to the GST-SipA (pellet) or the control GST beads, unbound in the cell extracts (post), or in the extracts before the reaction (pre) was detected on an immunoblot developed with an actin mAb. (D) The formation of the SipA-actin complex in infected HeLa cells was investigated by examining the proteins immunoprecipitated with an antibody to SipA (10). IP, immunoprecipitate. Left lane in (A) contains molecular size markers.

To study the actin-binding activity of SipA in more detail, we carried out a cosedimentation assay (12). Constant amounts of filamentous (F) actin, which is composed of polymerized actin monomers (G-actin), were mixed with variable amounts of purified SipA before centrifugation, and the amount of SipA cosedimented with F-actin in the pellet was quantitated by densitometric analysis of Coomassie blue–stained gels. Under these experimental conditions, the binding of SipA to F-actin was saturable and the stoichiometry of SipA to actin was approximately 1:1, suggesting that there is only one actin-binding site per SipA molecule (Fig. 3A). We also examined the actin-SipA complex by electron microscopy (13). The SipA-actin filaments were slightly larger in diameter than the actin filaments alone (actin, 6.5 ± 0.8 nm; actin plus SipA, 7.2 ± 0.7 nm). They were straighter, and they showed less obvious helical twist and subunit structure along the filaments (Fig. 3B).

Figure 3

Stoichiometry of the SipA-actin interaction. (A) The interaction of SipA and F-actin was examined by a cosedimentation assay (11). Coomassie blue–stained SDS gels of supernatants (S) and pellets (P) resulted from a 100,000gsedimentation of solutions containing varying amounts of SipA. The densitometric quantitation of SipA in the pellet is shown in the lower panel. (B) Electron micrographs of negatively stained F-actin (2 μM) incubated in the presence or absence of SipA (2 μM) (13). Scale bar, 0.1 μm.

We investigated the effect of SipA on actin dynamics using a pyrene-actin assay (14). The fluorescence of pyrene-actin increases when it is incorporated into F-actin and therefore can be taken as a direct measure of actin polymerization. We first measured the concentration of G-actin required for polymerization (critical concentration) in the presence or absence of SipA. Addition of SipA significantly reduced the critical concentration of G-actin from ∼0.25 μM to as low as 0.02 μM (Fig. 4A). Analogous results were obtained in experiments in which F-actin was diluted in the presence or absence of SipA (Fig. 4B). We also investigated the influence of SipA on the kinetics of actin polymerization. Pyrene–G-actin (1 μM) was incubated with SipA (1 μM) and the rate of polymerization was monitored over time by fluorescence intensity measurements. Addition of SipA did not alter the rate of actin polymerization (Fig. 4C), suggesting that this protein does not nucleate actin. There was a slight decrease in fluorescence intensity in the presence of SipA, but this was due to the quenching of pyrene-actin by SipA rather than an effect on actin polymerization. At saturating levels, SipA quenched 32 ± 0.2% of the pyrene–F-actin fluorescence (mean ± SD of three determinations).

Figure 4

Effect of SipA on actin dynamics. (Aand B) Effect of SipA on the critical concentration of actin. In (A), pyrene-labeled G-actin was diluted to various concentrations in the presence and absence of equimolar amounts of SipA, and fluorescence was measured 4 hours after the initiation of the polymerization reaction (14). Alternatively, in (B), pyrene-labeled G-actin (4 μM) was polymerized at room temperature for 30 min in actin polymerization buffer. Polymerized actin was then diluted to different concentrations in the presence or absence of equimolar concentrations of SipA. (C) Effect of SipA on the kinetics of actin polymerization. Pyrene–G-actin (1 μM) was incubated with SipA (1 μM) in polymerization-inducing buffer, and fluorescence intensity was measured (14). (D) Effect of SipA on actin depolymerization. Pyrene–F-actin (1 μM) in the presence or absence of SipA (1 μM) or phalloidin (1 μM) was diluted to 0.1 μM in actin-polymerizing buffer, and the decrease in fluorescence intensity was followed over time. Bars represent SEs of three independent experiments.

We next examined the effect of SipA on F-actin depolymerization. Pyrene–F-actin (1 μM) in the presence or absence of SipA (1 μM) was diluted to 0.1 μM, and the fluorescence intensity was monitored over time. SipA greatly increased the stability of F-actin (Fig. 4D). In the presence of SipA, F-actin remained stable even after 60 min of incubation under conditions that, in the absence of SipA, resulted in almost complete depolymerization. The stabilizing effect of SipA was comparable to that of the fungal toxin phalloidin (Fig. 4D). Similar results were obtained when F-actin was diluted without KCl and MgCl2, indicating that SipA also inhibits depolymerization induced by dilution into low ionic strength buffer (15). Incubation of F-actin with SopE, an unrelated Salmonellaprotein purified in an identical manner, did not increase F-actin stability (15). Electron microscopy confirmed these results: F-actin filaments were readily observable in the presence but not in the absence of SipA (15). The ability of SipA to inhibit actin depolymerization without affecting the rate of actin elongation excludes the possibility that the stabilizing effect is due to capping of the ends of the actin filaments.

Our results indicate that SipA is largely responsible for the spatial restriction of the cytoskeletal rearrangements that result from the activation of Rho GTPases during S. typhimurium infection. The related Shigella IpaA protein is involved in a similar phenotype, although it appears to exert its function by a completely different mechanism that involves binding vinculin (7). SipA does not bind vinculin (Fig. 2B), but rather it decreases the critical concentration of actin and it stabilizes actin filaments by inhibiting their depolymerization. This stabilizing function may promote outward extension of membrane ruffles and filopodia, thereby facilitating bacterial uptake. Thus, SipA may help to increase the net accumulation of actin filaments at the point of bacterial–host cell contact, or even to influence the position and polarity of these cross-linked filaments. Stabilizing the actin filaments at the bacterial-induced membrane ruffles may also facilitate the persistence of these filaments even when the concentration of G-actin has fallen below the critical concentration for polymerization.

Our results suggest a two-step mechanism for efficient entry ofS. typhimurium into host cells. First, the bacterial protein SopE activates the small GTP-binding proteins CDC42 and Rac, components of the host signaling pathways that initiate actin cytoskeletal rearrangements and membrane ruffling. In a second step, these rearrangements are localized and enhanced by the bacterial effector SipA. The ability of S. typhimurium to modulate the actin cytoskeleton by influencing different stages in the formation of membrane ruffles is a remarkable example of pathogen evolution to modulate host cellular functions.

  • * To whom correspondence should be addressed. E-mail: jorge.galan{at}


View Abstract

Navigate This Article