The Yersinia Yops inhibit invasion of Listeria, Shigella and Edwardsiella but not Salmonella into epithelial cells


Joan Mecsas ,E-mail


Yersinia virulence is dependent on the expression of plasmid-encoded secreted proteins called Yops. After bacterial adherence to receptors on the mammalian cell membrane, several Yops are transported by a type III secretion pathway into the host cell cytoplasm. Two Yops, YopH and YopE, prevent macrophages from phagocytosing Yersinia by disrupting the host cell cytoskeleton and signal transduction pathways. In contrast to this active inhibition of phagocytosis by Yersinia, other pathogens such as Salmonella, Shigella, Listeria and Edwardsiella actively promote their entry into mammalian cells by binding to specific host surface receptors and exploiting existing cell cytoskeletal and signalling pathways. We have tested whether Yersinia Yops can prevent the uptake of these diverse invasive pathogens. We first infected epithelial cells with Yersinia to permit delivery of Yops and subsequently with an invasive pathogen. We then measured the level of bacterial invasion. Preinfection with Yersinia inhibited invasion of Edwardsiella, Shigella and Listeria, but not Salmonella. Furthermore, we found that either YopE or YopH prevented Listeria invasion, whereas only YopE prevented Edwardsiella and Shigella invasion. We correlated the inhibitory effect of the Yops with the inhibitory action of the cell-signalling inhibitors Wortmannin, LY294002 and NDGA, and concluded that the four invasive pathogenic species enter epithelial cells using at least three distinct host cell pathways. We also speculate that YopE affects the rho pathway.


Many pathogenic bacteria disrupt or exploit mammalian cell processes to survive within their host (reviewed in Finlay and Falkow, 1997). Some pathogens actively promote their internalization by mammalian cells, whereas other pathogens actively inhibit their uptake by mammalian cells (Finlay and Falkow, 1997). Enteropathogenic Yersinia causes gastroenteritis, lymphadenitis and septicaemia in animals and humans (Cornelis, 1992). After Yersinia gains access to the Peyer's patch and mesenteric lymph nodes, it is found extracellularly in various host tissues (Hanski et al., 1989; Simonet et al., 1990).

Yersinia pathogenicity is dependent on chromosomal factors as well as a factors encoded by a virulence plasmid, pYV (Cornelis, 1992; Straley and Perry, 1995; Cornelis and Wolf-Watz, 1997). Yersinia binds avidly to host cell surfaces containing β-1 integrins, using either invasin or YadA (Isberg and Falkow, 1985; Isberg and Leong, 1990; Bliska et al., 1993; Yang and Isberg, 1993). Adherence triggers the transport of several plasmid-encoded proteins, called Yops, across the host cell membrane into the cell cytoplasm (Rosqvist et al., 1994; Sory and Cornelis, 1994; Persson et al., 1995). Two Yops, YopE and YopH, act synergistically to neutralize the phagocytic properties of macrophages, and disrupt both the focal adhesions and the cytoskeleton of epithelial cells (Rosqvist et al., 1990; 1991; Bliska et al., 1993). Yersinia strains that do not produce or secrete the Yops (ΔYop Yersinia) efficiently enter cells expressing β-1 integrins, including macrophages and epithelial cells (Rosqvist et al., 1990; Bliska et al., 1993; Yang and Isberg, 1993; Persson et al., 1995; Monack et al., 1997). Invasive Yersinia strains are avirulent (Rosqvist et al., 1990), and thus the inhibition of phagocytosis appears to be an essential feature of Yersinia pathogenicity.

Although many of the cellular targets of the Yops are not known, some of their phenotypes within mammalian cells and homologies to other proteins suggest possible biochemical functions. YopE causes disruption of the host cell cytoskeleton, including the collapse of actin stress fibres (Bliska et al., 1993; Rosqvist et al., 1991). YopE shares sequence homology with the N-terminal region of the Pseudomonas aeruginosa exoenzyme S, which ADP-ribosylates several GTP-binding proteins, including Ras (Coburn et al., 1989; Coburn and Gill, 1991). The shared regions of homology, however, do not include the catalytic ADP-ribosylating domain of exoenzyme S (Kulich et al., 1994). YopH is a potent tyrosine phosphatase that dephosphorylates proteins with src homology two and/or three domains (SH2, SH3) in vitro (Zhang et al., 1992; Black and Bliska, 1997; Persson et al., 1997). Two of its targets in vivo are the 125–135 kDa protein, CAS, and the focal adhesion kinase, FAK, both of which accumulate at focal adhesions (Black and Bliska, 1997; Persson et al., 1995; 1997). Focal adhesions contain integrins and other molecules involved in communicating between the extracellular matrix and the cytoskeleton (Clark and Brugge, 1995). The dephosphorylation of CAS and FAK presumably disrupts signalling to the cytoskeleton, which consequently halts cytoskeletal rearrangements necessary for Yersinia internalization (Fallman et al., 1995; Andersson et al., 1996; Black and Bliska, 1997; Persson et al., 1997). YpkA (YopO) shares amino acid homology with mammalian serine/threonine protein kinases and disrupts the host cytoskeleton of epithelial cells (Galyov et al., 1993; Hakansson et al., 1996). This effect on epithelial cells is distinct from the effect of YopE and is visible only after several hours of exposure to Yersinia strains overproducing YpkA and lacking YopE (Hakansson et al., 1996). YopJ induces programmed cell death (apoptosis) in infected macrophages but not in epithelial cells (Monack et al., 1997). YopM exhibits homology to the α-chain of the platelet receptor Gplα and prevents blood platelet aggregation (Reisner and Straley, 1992). All of the Yops described above are essential for Yersinia to cause a fatal infection in mice. Yersinia secretes at least seven more Yops, but their effects on cells are less well understood (Cornelis and Wolf-Watz, 1997).

In contrast to Yersinia, which paralyses host cell pathways required for bacterial uptake, other pathogenic bacteria exploit mammalian signal transduction pathways to actively promote their uptake by host cells (Francis et al., 1993; Dramsi et al., 1996; Watarai et al., 1996). The invasion of host cells is an important facet of Salmonella, Shigella and Listeria pathogenesis (Bernardini et al., 1989; Jones et al., 1993), whereas the importance of invasion in Edwardsiella pathogenesis is not yet clear. The cell processes triggered by these pathogens to facilitate their entry into mammalian cells have been the subject of considerable research (Finlay and Falkow, 1997). Invasion by Salmonella, Shigella, Listeria and Edwardsiella requires actin polymerization, as cytochalasin B and D inhibit their entry into epithelial cells (Finlay and Falkow, 1988; Janda et al., 1991). Small GTP-binding proteins that regulate the actin cytoskeleton have been implicated in the entry of some of these pathogens. Three such proteins, Cdc42, Rac and Rho, regulate assembly of filopodia, lamellipodia and actin stress fibres, respectively, in response to extracellular signals (Ridley and Hall, 1992; 1994; Ridley et al., 1992; Peppelenbosch et al., 1995). In some cases, activation of Cdc42 can result in activation of Rac, which can in turn activate Rho (Kozma et al., 1995; Nobes and Hall, 1995). Salmonella requires a Cdc42-dependent pathway to induce large actin-containing ruffles that engulf the bacterium (Chen et al., 1996), whereas Shigella induces a similar host response through a Rho-dependent pathway (Adam et al., 1996; Watarai et al., 1997).

Additional signal transduction pathways that result in tyrosine phosphorylation and/or changes in the host cell cytoskeleton are also activated during invasion. Shigella, Listeria and invasin- or YadA-mediated Yersinia entry all require tyrosine phosphorylation of specific host cell protein(s), as their uptake is inhibited by the tyrosine kinase inhibitor, genistein (Rosenshine et al., 1992; Velge et al., 1994; Andersson et al., 1996; Watarai et al., 1996); in contrast, Salmonella invasion is unaffected by genistein (Rosenshine et al., 1992). Activation of PI-3 kinase, which leads to actin rearrangement, is essential for Listeria entry into CHO, Vero and HeLa cells (Ireton et al., 1996). During Shigella invasion, cortactin, a cytoskeleton-associated protein, is phosphorylated by a tyrosine kinase, pp60c-src (Dehio et al., 1995). Both proteins, along with F-actin, co-localize with the invading Shigella, suggesting that they may be essential for Shigella entry (Dehio et al., 1995). Both Listeria and Salmonella invasion induce tyrosine phosphorylation of MAPK, but it is unclear if this phosphorylation is required for entry (Tang et al., 1994).

In this paper, we address whether the Yersinia pseudotuberculosis Yops disrupt the same host cell signal transduction pathways to remain extracellular that invasive pathogens exploit to enter cells. We hoped to learn more about the pathogenic traits of Salmonella, Shigella, Listeria and Edwardsiella, and to discern more about the potential mechanism(s) of Yop activity on host cellular processes. Our results show that one or more Yops have a profound effect on the entry of Shigella, Listeria and Edwardsiella but not Salmonella. From these data and data obtained by using several specific signal transduction pathway inhibitors, we conclude that the invasive pathogens enter cells by no less than three distinct host cell pathways and propose that Yersinia YopE may manifest its actions through a Rho-dependent pathway.


YopE and YopH block internalization of Yersinia pseudotuberculosis into HEp-2 cells

Mutant Y. pseudotuberculosis that do not express the Yops (ΔYop Yersinia) are rapidly internalized by cultured epithelial cells, whereas wild-type Yersinia is almost totally confined to the surface of cells (Rosqvist et al., 1991; Bliska et al., 1993). The basis for this striking difference was investigated by infecting epithelial cells with strains isogenic for the expression of one or more Yops (Fig. 1). We obtained similar results using two quantitative assays: the gentamicin protection assay, which measures the number of internalized bacteria, and a differential fluorescent staining assay, which permits direct counting of extracellular and intracellular bacteria (see Experimental procedures). Yersinia defective for YopE was internalized by epithelial cells at a frequency approaching, but not equal to, that exhibited by a ΔYop Yersinia strain (Fig. 1A and B). In contrast, Yersinia defective for YopH or YpkA was not effectively internalized and thus behaved like wild-type bacteria. Strains defective for both YopE and YopH showed the same high degree of internalization as ΔYop Yersinia. Although the invasive ΔYop Yersinia strain expresses both invasin and YadA, previous work has shown that under the conditions used to grow the bacteria, invasin is much more effective at facilitating internalization than YadA in epithelial cells (Bliska et al., 1993; Yang and Isberg, 1993). We concluded that YopE and YopH inhibit invasin-mediated Yersinia entry into epithelial cells.

Figure 1.

. The effect of various Yops on inhibiting Yersinia invasion. A. HEp-2 cells were infected with Yersinia pseudotuberculosis strains expressing various Yops at a multiplicity of infection of 10:1. One hour after infection gentamicin was added to the medium to kill extracellular bacteria. Intracellular bacteria were recovered and counted as described in Experimental procedures. Per cent invasion was calculated by dividing the number of intracellular bacteria recovered by the number of bacteria added to the assay. Infection by each strain was tested in triplicate to derive standard deviations and each assay was repeated at least three times. B. HEp-2 cells were exposed to Yersinia for 1 h and the number of cell-associated bacteria was assessed by differential immunofluorescence assays as described in Experimental procedures. The percentage of cell-associated bacteria that are intracellular is plotted with ‘n’ indicating the total number of cell-associated bacteria counted.

Given that wild-type Yersinia prevented its own invasion into epithelial cells through the activity of YopE and YopH, we next determined whether the Yops could prevent entry of invasive Yersinia bound to the same cell. In other words, was the inhibition a local event adjacent to the site of Yersinia attachment or was the effect on the host cell more global? HEp-2 cells were infected with wild-type Yersinia for 30 min to permit bacterial attachment and the subsequent transport of Yops into the cell cytoplasm to interact with their cellular targets. The cells were then infected with a ΔYop Yersinia strain constitutively expressing the green fluorescent protein (GFP) of Aquorea victoria. Entry of the ΔYop, GFP-producing Yersinia strain decreased more than 30-fold when HEp-2 cells were preinfected with wild-type Yersinia, whereas entry decreased twofold when HEp-2 cells were preinfected with a ΔYop Yersinia strain (Fig. 2). The twofold decrease in invasion may be due to the depletion of β1-integrin receptors on the cell surface after internalization of the ΔYop Yersinia strain. We have observed a similar phenomenon when the multiplicity of infection (MOI) of ΔYop Yersinia is greater than 10; the relative amount of invasion decreases (data not shown). Alternatively, at a high MOI, another Yersinia factor may reduce the entry of ΔYop Yersinia.

Figure 2.

. The effect of preinfection of various Yersinia strains on entry of GFP-ΔYop Yersinia into HEp-2 cells. HEp-2 cells were untreated or infected with wild-type, ΔYopH, ΔYpkA, ΔYopE, ΔYopEΔYpkA, ΔYopEΔYopH or ΔYop Yersinia strains for 30 min and then infected with GFP-ΔYop Yersinia. Entry of GFP-ΔYop Yersinia was quantified using the gentamicin assay as described in Experimental procedures. Each assay was carried out in triplicate to derive standard deviations and was repeated twice.

To identify which Yops were essential for the inhibition of invasin-mediated bacterial entry into epithelial cells, we sequentially infected HEp-2 cells with wild-type or mutant Yersinia strains followed by infection with a ΔYop, GFP-expressing Yersinia strain (Fig. 2). Preinfection with either wild type, a YopE mutant, a YopH mutant, a YpkA mutant or a YopE–YpkA mutant strain all resulted in a 10- to 20-fold inhibition of invasin-mediated bacterial uptake. Only preinfection with a Yersinia strain mutant for both YopH and YopE allowed significant invasin-mediated bacterial uptake. The observation that strains producing either YopE or YopH were capable of inhibiting invasin-mediated bacterial entry in trans suggests that both YopE and YopH block invasin-mediated bacterial uptake as Yersinia invaded epithelial cells only when both were absent. (In the ΔYopE Yersinia strain, YopH is still secreted and active in HEp-2 cells even although ≈1% of the bacteria become internalized.)

The Yops do not block Salmonella invasion or membrane ruffles in epithelial cells

When Salmonella invades epithelial cells efficiently, it induces massive actin accumulation at the site of entry, accompanied by macropinocytosis and the activation of several host cell signal transduction pathways (Francis et al., 1993). We investigated whether the mammalian cell mechanisms exploited by Salmonella to invade epithelial cells might include pathways disrupted by YopH and YopE to prevent Yersinia entry into cells. HEp-2 cells were first infected with wild-type Yersinia for 30 min and then infected with Salmonella for 30 min; entry of Salmonella was quantified using the gentamicin protection assay. Salmonella invasion was unaffected by preinfection of HEp-2 cells with Yersinia (Fig. 3). As illustrated in Fig. 4, the invading Salmonella induced marked actin polymerization around the site of invasion in both untreated cells and cells preinfected with Yersinia. In preinfected cells, the stress fibres have collapsed and the actin appears disorganized, except around the Salmonella. Thus, the Yersinia-induced collapse of the actin stress fibres does not impede Salmonella entry into epithelial cells or Salmonella-induced cytoskeletal reorganization. During the course of these experiments, we noted that when an MOI of greater than 10 was used for Salmonella, the number of internalized Yersinia increased several fold (data not shown). This observation is consistent with previous observations showing that Salmonella-induced macropinocytosis causes internalization of neighbouring particles (Francis et al., 1993).

Figure 3.

. The effect of preinfection of HEp-2 cells with wild-type and ΔYop Yersinia on the entry of invasive bacteria. HEp-2 cells were either uninfected or infected with wild-type or ΔYop Yersinia and subsequently infected with the following invasive bacteria: Salmonella typhimurium, Shigella flexneri, Listeria monocytogenes and Edwardsiella tarda. Invasion was quantified using the gentamicin assay. For each pathogen analysed, invasion into untreated HEp-2 was set to 100%. Each assay was performed in triplicate and repeated at least twice. In the assay presented, the actual percentage of invasion in untreated HEp-2 cells was 5% for Salmonella, 0.4% for Shigella, 0.4% for Listeria and 91% for Edwardsiella.

Figure 4.

. Immunofluorescence of HEp-2 cells after infection by Yersinia, Salmonella or sequential infections by Yersinia and Salmonella. The actin filaments of HEp-2 cells were stained with rhodamine-phalloidin and visualized in (A, B, D, F). Salmonella was visualized using an rabbit α-SL1344 polyclonal antiserum (C, E). The panels show (A) uninfected HEp-2 cells; (B) HEp-2 cells infected with wild-type Yersinia for 30 min; (C, D) same field of HEp-2 cells infected with Salmonella for 5 min; (E, F) same field of HEp-2 cells infected with wild-type Yersinia for 30 min and then with Salmonella for 5 min. The location of the Salmonella is indicated by arrows.

The Yops block Shigella, Edwardsiella and Listeria invasion of epithelial cells

Could the cellular effects of YopE, YopH or other Yersinia virulence factors influence the invasive properties of other pathogenic bacteria? We examined this question by determining if Yersinia preinfection of HEp-2 cells had any effect on the entry of three other invasive bacteria: Shigella, Listeria and Edwardsiella. Preinfection of HEp-2 cells with wild-type Yersinia decreased invasion of all three of these pathogens 8- to 10-fold, whereas preinfection with a ΔYop Yersinia strain did not (Fig. 3). Prolonged exposure to wild-type Yersinia has been reported to be cytotoxic to epithelial cells (Rosqvist et al., 1991). Hence, the decrease in gentamicin-protected Shigella, Listeria and Edwardsiella could have been a result of the HEp-2 cells becoming permeable to gentamicin during the course of the experiment. To exclude this possibility, HEp-2 cells were infected with Shigella for 30 min, washed, then subsequently infected with Yersinia for 60 min, and finally incubated with gentamicin for an additional 60 min. Infection with Yersinia made no significant difference in the number of Shigella recovered compared with the number of Shigella recovered from cells not infected with Yersinia (data not shown). This indicates that bacteria inside HEp-2 cells were not sensitive to gentamicin after a 60 min exposure to Yersinia. Therefore, the decrease in the number of gentamicin-protected Shigella, Listeria and Edwardsiella recovered after preinfection with Yersinia appears to be due to the disruption of cellular processes by the Yops.

YopE blocks Shigella and Edwardsiella entry, whereas either YopE or YopH block Listeria entry

To determine which Yersinia factors inhibited epithelial cell invasion by Shigella, Edwardsiella and Listeria, HEp-2 cells were infected first with various Yop-defective Yersinia strains and then with an invasive bacterium. YopE disrupted a process required for Shigella and Edwardsiella entry (Table 1 and Fig. 5A and B[link]), as cells preinfected with YopE-defective strains permitted Shigella and Edwardsiella invasion, whereas cells preinfected with any strain expressing YopE were unable to internalize Shigella and Edwardsiella (Fig. 5A and B). In contrast, Listeria invasion was inhibited by either YopE or YopH (Table 1, Fig. 5C[link]). Listeria only invaded cells preinfected with a ΔYopEΔYopH strain or a ΔYop Yersinia strain.

Table 1. . What blocks entry of bacteria?Thumbnail image of
Figure 5.

. The effect of preinfection of various Yersinia strains on entry into and association with HEp-2 cells of Shigella flexneri, Edwardsiella tarda and Listeria monocytogenes. A–C. HEp-2 cells were untreated or infected as indicated with wild-type, ΔYopE, ΔYopH, ΔYopEΔYopH, ΔYpkA, ΔYopEΔYpkA or ΔYop Yersinia for 30 min and then infected with (A) Shigella (B) Edwardsiella and (C) Listeria. Entry of the second bacteria was quantified using the gentamicin assay. Invasion of Shigella, Edwardsiella and Listeria in untreated cells was set to 100%. The actual percentage invasion in these experiments was 1.5%, 4.2% and 0.4% for Shigella, Edwardsiella and Listeria respectively.

The inhibition of Shigella, Edwardsiella and Listeria entry into HEp-2 cells could occur at one of several distinct steps in the invasion process. To determine whether the decrease in entry resulted from a decrease in binding to HEp-2 cells, we tested whether pretreatment with Yersinia altered the number of cell-associated bacteria. The number of Shigella, Edwardsiella and Listeria associated with HEp-2 cells was unaffected by pretreatment of cells with Yersinia, demonstrating that the Yops act subsequent to cell association (data not shown).

Wortmannin and NDGA inhibit Listeria and Yersinia invasin-mediated entry into HEp-2 cells

To further characterize the signal transduction pathways used by invasive bacteria, the effects of several inhibitors were tested on bacterial entry into HEp-2 cells. Wortmannin inhibits soluble PtINs 4-kinase, PI-3 kinases and PLA2 activity (Vlahos et al., 1994; Ui et al., 1995; Ireton et al., 1996; Ward et al., 1996) at IC50 values of 50 nM, 10 nM and 2 nM respectively. Treatment with 100 nM Wortmannin prevented entry of Yersinia and Listeria, but had no effect on invasion of Salmonella, Shigella and Edwardsiella (Fig. 6A and Table 1[link]). To distinguish which of these enzymes were critical for entry of ΔYop Yersinia, we assayed Yersinia invasion in the presence of various concentrations of Wortmannin. Preincubation of HEp-2 cells for 20 min with 10 nM Wortmannin, reduced entry of ΔYop Yersinia by 40%, whereas preincubation with 25 nM or 50 nM Wortmannin decreased entry by 95%. Thus, loss of PI-3, but not loss of PLA2, in HEp-2 cells appears to hinder ΔYop Yersinia entry. To confirm that PI-3 kinase activity is important for Yersinia entry, the effect of a second inhibitor of PI-3 kinase, LY294002, was tested on Yersinia entry. At the IC50 of LY294002, 10 μM, ΔYop Yersinia entry was reduced by 40% and at higher concentrations of 25 μM and 50 μM entry was 90% inhibited.

Figure 6.

. The effect of treatment with inhibitors on entry of ΔYop Yersinia, Salmonella, Shigella, Listeria and Edwardsiella. A. HEp-2 cells were incubated with 100 μM Wortmannin 2 h before infection with ΔYop Yersinia, Salmonella, Shigella, Listeria and Edwardsiella. For each bacterium, invasion into untreated HEp-2 cells was set to 100%. The actual percentage of invasion in untreated cells was 100% for ΔYop Yersinia, 13% for Salmonella, 6.8% for Shigella, 3.1% for Listeria and 1.2% for Edwardsiella. B. HEp-2 cells were incubated with 15 μg ml−1 NDGA for 15 min before infection. For each bacteria, invasion into untreated HEp-2 cells was set to 100%. The actual percentage of invasion in untreated cells was 19% for ΔYop Yersinia, 20% for Salmonella, 1.9% for Shigella, 0.2% for Listeria and 0.8% for Edwardsiella.

Nordihydroguaretic acid (NDGA) inhibits 5-lipoxygenase activity (Peppelenbosch et al., 1995), which generates leukotrienes from arachidonic acid. Leukotrienes act in many cell signal transduction cascades including signalling Rho after Rac activation (Peppelenbosch et al., 1995). At a concentration of 15 μg ml−1, NDGA blocked Listeria and invasin-mediated Yersinia entry but had no effect on the other pathogens tested (Fig. 6B and Table 1[link]). At lower concentrations (5 μg ml−1), NDGA had no effect on Listeria or invasin-mediated Yersinia entry. Incubation of Salmonella, Listeria, Yersinia, Shigella and Edwardsiella with Wortmannin and NDGA had no effect on bacterial viability (data not shown). The toxicity of these inhibitors on epithelial cells was clearly evident by visual inspection of the HEp-2 cells using a light microscope (data not shown). For instance, accumulation of large vacuoles could be seen easily in the Wortmannin-treated cells. However, given the observation that certain bacteria still entered these treated epithelial cells, they were clearly not devoid of all cellular functions.


During infection of mammals, pathogenic Yersinia migrates to the lymphatic tissue, its preferred niche in the host. After entering the intestinal tract, Yersinia migrates to the Peyer's patches of the gut (Fujimara et al., 1992; Marra and Isberg, 1997), where a single-cell layer of epitheloid tissue exists between the mucosal surfaces and the underlying lymphoid tissue. To reach the lymphoid tissue, Yersinia targets the M cell, a terminally differentiated epithelial cell that shows marked pinocytotic activity (Fujimara et al., 1992; Marra and Isberg, 1997). Invasin facilitates the entry of Yersinia into the murine Peyer's patch, presumably by binding to β-1 integrins on the M-cells, and thereby ensuring that Yersinia is efficiently internalized by, and traverses through, the M cell to reach the lymphoid tissue (Pepe et al., 1995; Marra and Isberg, 1997). However, once the epithelial barrier is breached, Yersinia remains primarily extracellular even in the vicinity of phagocytes (Hanski et al., 1989; Simonet et al., 1990). Hence, early in infection Yersinia promotes its own entry into cells, whereas at later stages Yersinia actively inhibits its own entry into cells. Either of these two phenotypes can be observed in cell culture assays depending upon the conditions used to grow the bacteria (Simonet and Falkow, 1992; Yang and Isberg, 1993).

Our experiments and those of Persson et al. (1997) show that YopH and YopE block invasin-mediated Y. pseudotuberculosis entry into epithelial cells; we and others have also shown that the Yops can inhibit macrophage phagocytosis, so that the adherent bacterium remains extracellular (Rosqvist et al., 1990; Andersson et al., 1996; Monack et al., 1997). There is some conflicting evidence about the relative importance of YopE and YopH in preventing bacterial uptake. In our experiments YopE alone is sufficient to inhibit significant Yersinia entry into epithelial cells when YopE is delivered to cells either by the invading bacteria or in trans. The effect of YopH was observed only in the absence of YopE or when cells were exposed to YopH for 30 min. We have complemented a Yersinia YopE mutant with YopE expressed from an inducible promoter and restored the inhibitory effects on bacterial uptake (our unpublished data). In contrast, using differential fluorescence staining, Rosqvist et al. (1990) have found that YopH is the predominant Yop neutralizing the phagocytic activity of bone marrow-derived macrophages. One possible explanation for these results is that different Yersinia adhesins may alter Yop delivery and/or Yersinia internalization. Under the conditions we use to grow the Yersinia strains, invasin-mediated entry into epithelial cells is more efficient than YadA-mediated entry (Bliska et al., 1993), whereas YadA attachment is more efficient at delivering Yops to bone marrow-derived macrophages (A. Hromockyj and S. Falkow, personal communication).

Most of the work reported here is focused on whether the cellular effects of YopE, YopH and YpkA inhibited the ability of other bacterial pathogens to invade cells. Based on our results, we delineate at least three distinct entry mechanisms for invasive bacteria (Table 1). Salmonella enter cells via processes that are unaffected by any Yop. Shigella and Edwardsiella enter cells via processes that YopE disrupts, and Listeria and Yersinia enter cells via processes disrupted by both YopH and YopE. Our conclusions about which Yops affect Shigella, Listeria and Edwardsiella entry are based on analysis of Yersinia strains that have deletions in YopE, YopH and YpkA. These studies raise several questions. For instance, other Yops also may affect bacterial entry by assisting YopE and YopH, but their role(s) have not yet been detected. Furthermore, our results indicate that YpkA does not inhibit invasion by any pathogen in this assay, although it alters the cytoskeleton of epithelial cells (Hakansson et al., 1996). As YpkA is synthesized at a much lower level than other Yops and its effect on epithelial cells is detected only after several hours (Hakansson et al., 1996), the HEp-2 cells may not be sensitive to the effects of YpkA during the course of the assays. Finally, as wild-type Yersinia remains bound to the cell surface whereas ΔYop Yersinia is internalized, wild-type Yersinia may sterically hinder the binding of invasive bacteria to its own host cell receptors or the Yops may disrupt trafficking of receptors to the cell surface. Future studies expressing YopE, YopH and YpkA from within mammalian cells will demonstrate whether YopE and/or YopH are sufficient to block bacterial invasion, and whether YpkA activity affects the invasion of host cells by Salmonella, Shigella, Listeria and Edwardsiella.

Although our observation that different invasive bacteria enter cells via distinct mechanisms is not novel (Isberg, 1991; Francis et al., 1993; Adam et al., 1995; Mengaud et al., 1996), one of our objectives was to use the accumulated knowledge about the entry mechanisms of these pathogens to further characterize potential mammalian cell targets of the Yops. Figure 7 depicts some of the known key components required for the invasion of Salmonella, Shigella, Listeria, Edwardsiella and ΔYop Yersinia. It also outlines some of the mammalian cell regulatory cascades that are involved in controlling actin cytoskeleton dynamics in fibroblasts, the enzymes inhibited by Wortmannin and NDGA, and our view of the potential targets of YopE and YopH. We recognize that there are several important caveats to this interpretation our data. YopE and YopH are probably responsible for pleiotropic effects on the cell either by interacting with a single regulatory protein or because they have multiple host cell targets. Therefore, different pathogens inhibited by the same Yop may not use identical cellular processes for invasion. Also, we have generally assumed that Yop activity dominates signals generated by the invading bacteria. This may not always be true. For instance, as the Yops had no effect on Salmonella invasion, we conclude that Salmonella and Yersinia affect different cellular processes. On the other hand, this conclusion could be erroneous if signals generated by Salmonella during its invasion actually act on the same pathway as the Yops but dominate or negate the activity of YopE and YopH. If so, our conclusion that the mechanism of Salmonella entry differs from that of the other invasive pathogens still stands, but the difference lies in the ability of Salmonella to modulate host cell mechanisms. In fact, Salmonella encodes a protein of unknown function that has homology to both YopE and YopH (Kaniga et al., 1996). This protein could possibly antagonize YopE and/or YopH activity. Finally, our knowledge of the actin cytoskeleton is based primarily on studies carried out in fibroblasts, thus our speculations assume that results found in fibroblasts are generalizable to epithelial cells.

Figure 7.

. Model of mammalian cell proteins necessary for invasion of Salmonella, Listeria, Yersinia, Shigella and Edwardsiella, and potential mammalian cell targets of the Yops. Evidence for the bacterial proteins and host cell receptors required for invasion of these pathogens are referenced in the text and in Finlay and Falkow (1997). In fibroblasts, evidence for activation of Cdc-42, Rac, and Rho, the signal transduction cascade from Cdc-42 to Rac to Rho and the downstream effects of their activation can be found in references Guinebault et al. (1995); Kozma et al. (1995); Nobes and Hall (1995); Peppelenbosch et al. (1995); Ridley and Hall (1992; 1994); Ridley et al. (1992); Ui et al. (1995); Vlahos et al. (1994) and references therein.

Shigella and Edwardsiella were inhibited from entering cells by YopE but were not inhibited by Wortmannin or NDGA. Therefore, cell-signalling mechanisms required for entry do not involve communication through PI-3 tyrosine kinases or conversion of arachidonic acid to leukotrienes. Activity of the small GTPase Rho is required for Shigella invasion, and Rho has been implicated in stress fibre formation (Ridley and Hall, 1994; Adam et al., 1996; Watarai et al., 1997). As YopE inhibits Shigella invasion and causes stress fibres to collapse, we speculate that YopE disrupts cellular mechanisms dependent on Rho by acting either directly on Rho, Rho-modifying proteins or molecules downstream of Rho (Fig. 7). The homology of YopE to exoenzyme S, which binds the small GTPase Ras (Ras shares homology with Rho), lends further credence to the idea that YopE may interact with Rho (Kulich et al., 1994). Experiments addressing whether YopE alters Rho are in progress. Finally, although Shigella invasion induces host tyrosine phosphorylation pp60c-src and cortactin, our observation that YopH does not inhibit Shigella entry suggests that these proteins are not substrates for YopH.

Invasion by Listeria and ΔYop Yersinia is clearly distinct from that of Shigella and Edwardsiella (Table 1). As either YopE or YopH appears to inhibit Listeria and invasin-mediated Yersinia entry, both YopE and YopH must disrupt cellular mechanisms required by these microorganisms for invasion. Wortmannin and NDGA also inhibit these bacteria, indicating that signals generated through activation of PI-3 kinases and through generation of leukotrienes from arachidonic acid are required for invasion in HEp-2 cells (Fig. 7 and Table 1[link]). A dominant negative PI-3 kinase mutant blocks Listeria invasion (Ireton et al., 1996); as YopH dephosphorylates proteins with SH2 domains, YopH may act on PI-3 kinase to halt signal transduction. Although this is an attractively simple explanation of the experimental results, PI-3 kinase may not actually be the primary target of YopH in macrophages and during Yersinia infection. YopH is already known to dephosphorylate several tyrosine kinases in vivo and thus may interfere in a variety of ways with cellular signal transduction pathways and other cell processes (Black and Bliska, 1997; Persson et al., 1997). Alternatively, one of the known in vivo substrates for YopH, CAS or FAK, may mediate Listeria invasion. In fact, PI-3 kinase can interact with FAK (Guinebault et al., 1995; Carpenter and Cantley, 1996). Future studies assessing phosphorylation of CAS, FAK and PI-3 kinase during Yersinia invasion in the presence of YopH should distinguish between these possibilities.

We observed that YopE also blocks Listeria invasion. Although we propose that YopE may act on Rho-controlled processes, we know of no published evidence for Listeria entry via a Rho-mediated cellular pathway. Nevertheless, the observation that NDGA also blocks Listeria uptake is consistent with the idea that Listeria enters cells by activating a PI-3 kinase that activates Rac, which in turn activates Rho (Fig. 7). Rac activation of Rho requires conversion of arachidonic acid to leukotrienes, a step known to be inhibited by NDGA (Peppelenbosch et al., 1995). Alternatively, YopE may act on Rac or Rac-modifying enzymes in addition to Rho, and NDGA may act to block functions downstream of Rac. Precedence for bacterial toxins inhibiting more than one small G protein exists; for example, Clostridium difficile synthesizes toxins that inactivate both Rac and Rho (Finlay and Falkow, 1997).

Salmonella entry is unaffected by preinfection of cells with Yersinia. Although the genes controlling the entry mechanisms of Salmonella and Shigella are in many ways quite similar, our data and several other recent observations by other laboratories indicate that Salmonella and Shigella enter cells by distinctly different mechanisms (Rosenshine et al., 1992). The Rho inhibitor, C3, from Clostridium botulinum has no effect on Salmonella entry (Jones et al., 1993), whereas a dominant negative Cdc42 allele affects both Salmonella entry and ruffle formation (Chen et al., 1996). In contrast, C3, blocks Shigella entry (Adam et al., 1996; Watarai et al., 1997). The observation that a cell exposed to YopE can still generate large actin-filled ruffles, indicates that YopE does not block all actin assembly, and suggests that YopE acts on actin mechanisms that are distinct from Cdc42-controlled mechanisms. The prediction that YopE acts on Rho-mediated mechanisms is consistent with these observations and is the focus of further experiments in our laboratory. We found that NDGA, which inhibits 5-lipoxygenase activity, did not inhibit Salmonella invasion of HEp-2 cells; in contrast, Pace et al. (1993) found that two different inhibitors of 5-lipoxygenase activity, 5,8,11-eicosatriynoic and 8,11,14-eicosatrienoic acid, did inhibit invasion of Salmonella into Henle cells. The apparent discrepancy in these results could be due to differences in cultured cells, Henle versus HEp-2, and/or differences in the way the Salmonella were grown, which can alter the mode of Salmonella entry into cells (Monack et al., 1996).

The evolution of pathogenicity in enteric bacteria such as Salmonella, Shigella and Yersinia appears to involve the inheritance of large blocks of genes that include a type III secretion pathway (Lee, 1997). This secretion pathway, which is often activated when the pathogen comes in contact with host cells, transports bacterial proteins to the host cell surface and in some cases across the plasma membrane and into the host cell cytoplasm. We are beginning to appreciate that these transported proteins, including the Yop proteins of Yersinia, are not only essential agents of bacterial pathogenicity, but that they represent important experimental tools to examine cellular processes in normal as well as infected cells.

Experimental procedures

Growth and construction of bacterial strains

All strains are listed in Table 2. For invasion and cell association assays, all Yersinia strains were grown in 2× YT overnight at 26°C. On the day of the assay, the bacteria were diluted 1:50 into 2× YT plus 20 mM sodium oxalate and 20 mM MgCl2 and were grown with aeration for 2 h at 26°C, then they were shifted to 37°C and were grown with aeration for an additional 2 h. The bacteria were diluted to a concentration of 8 × 107 to 1 × 108 bacteria per ml in PBS. To induce maximal expression of the invasive functions of Salmonella, cultures were grown overnight in Luria broth containing 0.3 M NaCl without aeration (i.e. standing) to maintain low oxygen conditions; 4 h before the assay, the cultures were diluted 1:100 and allowed to grow without aeration to an OD600 of 0.1–0.3. Edwardsiella and Listeria were grown with aeration at 37°C in brain–heart infusion broth overnight, whereas Shigella was grown at 37°C in trypticase soy broth. These strains were diluted 1:100 3–4 h prior to the assay and grown with aeration at 37°C. Logarithmically growing cultures were used for infection assays.

Table 2. . Bacterial strains. (Portnoy et al. (1984); Dramsi et al. (1995); Strauss et al. (1997).)Thumbnail image of

The following strategy was used to create ΔyopH and ΔypkA strains. A 400 bp XbaI–KpnI internal fragment of yopH was cloned into the vector p704.3 (Miller and Mekalanos, 1988), which carries the β-lactamase gene and requires the Pir protein for replication. An 800 bp EcoRV internal fragment of ypkA was cloned into p704.3. These plasmids were grown in E. coli strain SM10λpir and were introduced into appropriate Yersinia recipient strains by conjugation. Yersinia, but not E. coli, is naturally resistant to the drug irgasan. Therefore, we selected ampicillin–irgasan-resistant clones that arose from homologous recombination between the cloned internal yopH or ypkA fragment on the p704.3 plasmid and the native gene on the virulence plasmid. These recombination events disrupt the yopH or ypkA gene. Disruptions in the yopH and ypkA genes were confirmed by PCR and by analysing the secreted Yop proteins using SDS–PAGE. In all strains containing insertions in the yopH gene, YopH was not detected by SDS–PAGE. Furthermore, no phosphatase activity was detected when assayed (Persson et al., 1995). In wild-type Yersinia, the amount of secreted YpkA was too little to be detected by SDS–PAGE; however, YopJ, which is the downstream component of the ypkAyopJ operon, is detectable. In all strains containing insertions in the ypkA gene, YopJ was not detected, indicating a disruption of the ypkAyopJ operon (Monack et al., 1997).

Gentamicin protection assay

All assays were performed in triplicate. HEp-2 cells were grown in RPMI (Gibco BRL) + 5% fetal calf serum (Gibco BRL) at 37°C in 5% CO2 and were seeded at a density of 0.8–1 × 105 in 24-well tissue culture plates (Falcon) 1 day before the assay. This seeding resulted in a confluent but uncrowded monolayer. Bacteria were added to the cells at an approximate multiplicity of infection of 10:1 for Yersinia, Listeria and Shigella, 5:1 for Salmonella and 0.5:1 for Edwardsiella. Bacteria were spun onto the cells for 5 min at 1000 r.p.m. (185 × g) in a table-top centrifuge. Invasion was allowed to occur for 1 h at 37°C and 5% CO2, and then gentamicin was added for 1 h to kill the external bacteria (Isberg and Falkow, 1985). In experiments in which cells were sequentially infected with different strains of bacteria, the first strain was spun onto the cells. Thirty minutes later the second strain was spun onto the cells, and invasion was allowed to occur for 30 min. Finally, gentamicin was added for 60 min. For assays quantifying Yersinia, Salmonella or Edwardsiella invasion, 100 μg ml−1 gentamicin was added to cells; for assays quantifying Listeria or Shigella invasion, 20 μg ml−1 gentamicin was added to cells. Cells were washed twice with PBS to remove gentamicin, and then lysed with 100 μl of 1% Triton for 5 min. To each well, 900 μl of 2× YT was added and dilutions of each sample were plated onto the appropriate agar plates (see below).

Several strategies were used to recover and count the internalized bacteria of interest in experiments with mixed infections. To distinguish Yersinia or Shigella from Salmonella, bacteria were grown overnight at 37°C on Luria agar or trypticase soy agar plates respectively. Within 16 h of plating, Salmonella and Shigella colonies are clearly visible, whereas Yersinia requires 48 h to form visible colonies under these conditions. To distinguish Listeria, Edwardsiella or ΔYop Yersinia from wild-type Yersinia, which is sensitive to streptomycin and kanamycin, Listeria and Edwardsiella were plated on brain–heart infusion agar plates containing 50 μg ml−1 streptomycin and grown at 37°C, and ΔYop Yersinia was plated on Luria agar plates containing 50 μg ml−1 kanamycin and grown at 26°C. In some experiments, a ΔYop Yersinia expressing GFP was used. This strain was distinguished from other Yersinia strains by plating on Luria agar that contained 100 μg ml−1 ampicillin and counting green fluorescent colonies.

Cell association assays

For cell association assays, HEp-2 seeding and infections were performed as described above. After the second 30-min infection, cells were washed five times with PBS and lysed with 1% Triton for 5 min. To each well, 900 μl of 2× YT was added and dilutions were plated as described above. Shigella, Edwardsiella and Listeria adhered to the plastic of the 24-well plates at least 10-fold less efficiently than they adhered to HEp-2 cells. ΔYop Yersinia, however, adhered to the plastic wells at frequencies similar to its adherence to HEp-2 cells; thus, this experiment was not feasible with Yersinia.

Immunofluorescence assays

In 24-well tissue culture plates, 4 × 104 HEp-2 cells per well were seeded on glass coverslips 24 h before infection. Cells were infected as described above. Five minutes after infection, cells were fixed in 3.7% formaldehyde in PBS for 10 min, washed five times with PBS, permeabilized with 0.2% Triton for 15 min, and washed five times with PBS. Salmonella was detected by staining with a 1:500 dilution of rabbit α-Salmonella antibody in PBS for 20 min, washing four times with PBS and then incubating with a 1:200 dilution of an FITC-conjugated goat α-rabbit antibody (Sigma) in PBS for 20 min. Actin filaments were visualized by incubating cells with a 1:250 dilution of rhodamine-phalloidin (Molecular Probes) in PBS for 20 min. Glass coverslips were placed over a drop of antiquench on slides. Cells were viewed using a Nikon Optiphot fluorescence microscope and photographed using a Nikon FX-35a camera.

Differential fluorescence staining

To distinguish internal versus external Yersinia using fluorescent microscopy, the following strategy was used. HEp-2 cells were infected as described above; however, instead of adding gentamicin, cells were incubated with RPMI containing a 1:50 dilution of rabbit α-Yersinia antibody for 20 min. Cells were then fixed with 3.7% formaldehyde, washed four times with PBS and incubated with a 1:200 dilution of FITC-conjugated goat α-rabbit antibody. Only extracellular Yersinia is accessible to the Yersinia antibody and should fluoresce green. HEp-2 cells were then permeabilized with 0.2% Triton, washed four times with PBS, incubated with a 1:50 dilution of rabbit α-Yersinia antibody, washed four times with PBS and then incubated with a 1:200 dilution of TRITC-conjugated goat α-rabbit antibody. Both external and internal Yersinia should fluoresce red. The percentage of Yersinia inside cells is calculated as:

inline image

Inhibitor assays

HEp-2 cells were seeded as described above. Cells were treated with 100 nM Wortmannin 2 h before infection, or with 3–30 μg ml−1 of NDGA in DMSO 15 min before infection. In the presence of inhibitor, cells were infected for 30 min, treated with gentamicin plus inhibitor for 1 h, washed and lysed as described above, and then dilutions were plated. Parallel sets of experiments were performed in the absence of serum. HEp-2 cells were seeded and allowed to grow in the presence of serum for 4–6 h, and then washed three times with RPMI without serum and allowed to grow overnight in RPMI without serum.


  1. Present address: Max-Plank Institut für Infektionsbiologie, Berlin, Germany

  2. These authors contributed equally to this work


We thank Brendan Cormack, David Gunn, Anthea Lee, Denise Monack, James Nelson, Lalita Ramakrishnan, Nina Salama, Ulrich Schaible and Evelyn Strauss for helpful discussions and/or critical reading of the manuscript. We greatly appreciate Sara Fisher, Anthea Lee, Denise Monack, and Emily Senecal for editorial assistance. J.M. was supported by Damon Runyon Walter Winchell Cancer Research Fund no. 1277. B.R. was supported by Grant RA587/2-1 from the Deutsche Forschungsgemeinschaft and the NIH Digestive Disease Program grant DK38707. SF was supported by PHS grant AI26195 and Lederle-Praxis Biologicals.