YopT, a new Yersinia Yop effector protein, affects the cytoskeleton of host cells

Authors

  • Maite Iriarte,

    1. Microbial Pathogenesis Unit, Christian de Duve Institute of Cellular Pathology, and Faculté de Médecine, Université Catholique de Louvain, Avenue Hippocrate, 74, UCL 74.49, B-1200 Brussels, Belgium.
    Search for more papers by this author
  • Guy R. Cornelis

    1. Microbial Pathogenesis Unit, Christian de Duve Institute of Cellular Pathology, and Faculté de Médecine, Université Catholique de Louvain, Avenue Hippocrate, 74, UCL 74.49, B-1200 Brussels, Belgium.
    Search for more papers by this author

Guy R. Cornelis at the Microbial Pathogenesis Unit. E-mail cornelis@mipa.ucl.ac.be; Tel. (2) 764 74 49; Fax (2) 764 74 98.

Abstract

Extracellular Yersinia disarm the immune system of their host by injecting effector Yop proteins into the cytosol of target cells. Five effectors have been described: YopE, YopH, YpkA/YopO, YopP and YopM. Delivery of these effectors by Yersinia adhering at the cell surface requires other Yops (translocators) including YopB. Effector and translocator Yops are secreted by the type III Ysc secretion apparatus, and some Yops also need a specific cytosolic chaperone, called Syc. In this paper, we describe a new Yop, which we have called YopT (35.5 kDa). Its secretion required an intact Ysc apparatus and SycT (15.0 kDa, pI 4.4), a new chaperone resembling SycE. Infection of macrophages with a Yersinia, producing a hybrid YopT–adenylate cyclase, led to the accumulation of intracellular cAMP, indicating that YopT is delivered into the cytosol of eukaryotic cells. Infection of HeLa cells with a mutant strain devoid of the five known Yop effectors (ΔHOPEM strain) but producing YopT resulted in the alteration of the cell cytoskeleton and the disruption of the actin filament structure. This cytotoxic effect was caused by YopT and dependent on YopB. YopT is thus a new effector Yop and a new bacterial toxin affecting the cytoskeleton of eukaryotic cells.

Introduction

Yersinia spp. are able to resist the non-specific immune response of their host and to multiply in the lymphoid tissues. This capacity depends on a sophisticated apparatus called the Yop virulon, which is encoded by a 70 kb plasmid called pYV (for review, see Cornelis and Wolf-Watz, 1997). This apparatus consists of a complex type III secretion machinery (Ysc) and Yop proteins that are secreted by this system (Michiels et al., 1991; Plano et al., 1991; Allaoui et al., 1994; 1995; Bergman et al., 1994; Fields et al., 1994; Woestyn et al., 1994). The Yop proteins themselves fall into two groups. The translocator Yops presumably form a delivery apparatus that assembles at the interface between the bacterium and the eukaryotic cell and translocates the effector Yops across the plasma membrane of cells involved in the immune response (for review, see Cornelis and Wolf-Watz, 1997). The Ysc secretion apparatus is thought to be installed as soon as the bacterium enters its host and the temperature reaches 37°C. However, Yops are not secreted until there is contact between the bacterium and a eukaryotic cell. This contact control may involve YopN, TyeA and LcrG that are thought to gate the secretion channel (Persson et al., 1995; Boland et al., 1996; Håkansson et al., 1996; Boyd et al., 1998; Iriarte et al., 1998, Sarker et al., 1998a). The translocation apparatus consists of at least YopB and YopD, which require LcrV for their secretion, and LcrG (Sarker et al., 1998a,b; see other references in Cornelis and Wolf-Watz, 1997). The YopN-associated TyeA is needed for translocation of some Yop effectors but not for others (Iriarte et al., 1998), while YopN itself is dispensable for translocation (Forsberg et al., 1994; Rosqvist et al., 1994; Boland et al., 1996). Secretion and translocation of the YopE and YopH effectors requires the presence of individual cytosolic chaperones called SycE and SycH (for review, see Wattiau et al., 1996), which bind to a specific domain of their Yop partner (Woestyn et al., 1996). The SycE chaperone protects YopE from intrabacterial degradation and acts as a pilot to drive YopE to the Ysc secretion apparatus (Wattiau and Cornelis, 1993; Frithz-Lindsten et al., 1995; Cheng et al., 1997). The Yop effectors are YopE, YopH, YopM, YpkA/YopO and YopP. YopE exerts a cytotoxic effect on eukaryotic cells, resulting in depolymerization of the actin filaments, and has an anti-phagocytic effect on macrophages (Rosqvist et al., 1990; 1991). A yopE mutant of Y. pseudotuberculosis is no longer cytotoxic (Rosqvist et al., 1990; 1991), but a yopE mutant of Y. enterocolitica induces the same cytotoxic effect as the wild-type strain (Hartland et al., 1994). YopH is a protein tyrosine phosphatase (PTPase) that also has an anti-phagocytic effect on macrophages (Fällman et al., 1995; Andersson et al., 1996; Ruckdeschel et al., 1996). It causes dephosphorylation of p130Cas and FAK and disruption of peripheral focal adhesion complexes, which results in inhibition of bacterial uptake (Black and Bliska, 1997; Persson et al., 1997). YpkA/YopO is a serine–threonine protein kinase (Galyov et al., 1993). Finally, YopP (YopJ in Y. pseudotuberculosis) triggers apoptosis of macrophages (Mills et al., 1997; Monack et al., 1997) and prevents the release of tumour necrosis factor alpha (TNFα) (Boland and Cornelis, 1998; Palmer et al., 1998).

In this work, we describe YopT, a new Y. enterocolitica Yop protein. Secretion of YopT requires SycT, a protein of the SycE chaperone family. We show that YopT is an effector protein, delivered into eukaryotic cells, where it induces a cytotoxic effect and disruption of the actin filaments. Translocation of YopT across the eukaryotic cell membrane requires YopB but not TyeA.

Results

Two new ORFs between yopQ and yopM

The region between yopQ and yopM in the pYV plasmid from Y. enterocolitica W227 contains two adjacent open reading frames (ORF1 and ORF2), oriented clockwise (Fig. 1). ORF1 encodes a protein of 322 amino acids with a predicted molecular mass of 35.5 kDa and no classical N-terminal signal sequence.

Figure 1.

. Genetic map of the pYV plasmid of Y. enterocolitica. The region encoding YopT and SycT and the overlap between the genes encoding these two proteins are detailed.

The start codon of ORF2 overlaps one nucleotide of the stop codon of ORF1 and is preceded by a fairly good ribosome binding site (AAGGAGN5ATG). Translation of ORF2 gives a 130-residue protein (15.1 kDa) with an acidic pI (4.4). The putative protein is 20.9% identical and 69.7% similar to the SycE chaperone (Fig. 2A) and 23.0% similar to SycH (data not shown; Wattiau and Cornelis, 1993; Wattiau et al., 1994). The hydrophobic moment plot (Eisenberg et al., 1984) of the product of ORF2 resembles that of SycE with a major peak at the C-terminal end (data not shown). The amino acids of the last peak closely match the Syc chaperone consensus sequence described by Wattiau et al. (1996) (Fig. 2B).

Figure 2.

. Similarity between SycT and SycE. A. Comparison between the amino acid sequences of SycT and SycE. B. Alignment of amino acids 81–112 of SycT with the corresponding region of other chaperones of the SycE family (Wattiau et al., 1996), Orf1 (Yahr et al., 1995) OrfU (Jerse et al., 1990) and SycH (Wattiau et al., 1994). Vertical lines correspond to pairwise sequence alignments that were significant. A consensus sequence derived from the aligned residues that completes the consensus of Wattiau et al. (1996) is shown below the sequences.

In conclusion, ORF1 and ORF2 form an operon encoding two proteins, one of which seems to be a member of the Syc chaperone family. By analogy with yopE–sycE (Wattiau and Cornelis, 1993) and yopH–sycH (Wattiau et al., 1994), this organization suggests that ORF1 encodes a new Yop protein.

ORF1 encodes a new Yop: YopT

To determine whether the product of ORF1 is a new Yop, we first tried to identify it among the Yops secreted by Y. enterocolitica E40 (pYV40), but we could not clearly identify a 35.5 kDa protein. As secretion of less abundant Yops can be facilitated when the other Yops are missing, we analysed the proteins secreted by a polymutant (yopH, yopO,yopP,yopE,yopM) derivative of MRS40 called ΔHOPEM (Boland and Cornelis, 1998), and we observed a protein of about 35.5 kDa. To determine whether this protein was the product of ORF1, we constructed an ORF1 mutant by creating an out of frame deletion starting at codon 135, and we replaced the wild-type gene with the mutated allele in the pYV plasmid of strain ΔHOPEM. The new mutant strain was called MRS40 (pIML421) or ΔHOPEM,ORF1. We also raised an antiserum against two peptides derived from the sequence of ORF1. As shown in 3Fig. 3A (lanes 4 and 5) and B (lanes 1 and 2), the 35.5 kDa band reacted with the antiserum and disappeared from the proteins secreted by the ΔHOPEM, ORF1 strain.

Figure 3.

. Secretion of YopT. SDS–PAGE and Coomassie blue staining of culture supernatant proteins of the following strains of Y. enterocolitica. A. Lane 1, E40 (pYV40); lane 2, E40 (pIM409); lane 3, E40 (pIM402); lane 4, MRS40 (pABL403); lane 5, MRS40 (pIML421); lane 6, MRS40 (pIML422); lane 7, MRS40 (pMSK50). B. Western blot analysis with a polyclonal anti-YopT serum. Lane 1, MRS40 (pABL403); lane 2, MRS40 (pIML421); lane 3, MRS40 (pIML422); lane 4, MRS40 (pIML422) (pIM279); lane 5, KNG22703 (pPW2269) (pIM279); lane 6, KNG22703 (pSW2276) (pIM279). C. Secretion of YopT124–Cya by Y. enterocolitica E40. Plasmid pAB29 encoding YopT124–Cya was introduced in Y. enterocolitica E40 (pYV40), synthesis of the Yop virulon was induced for 4 h at 37°C and the secreted proteins were analysed by SDS–PAGE. The letters identify the various Yop proteins. Molecular weight markers are shown on the left.

To analyse whether secretion of this protein is dependent on the Ysc secretion machinery, we analysed the proteins secreted by a ΔHOPEM strain with an additional mutation in yscN. As shown in 3Fig. 3A, lane 7, the 35.5 kDa protein disappeared like all the Yops, suggesting that this protein is secreted by the same Ysc apparatus. To confirm that the secretion of YopT was indeed dependent of the Ysc secretion machinery and that the absence of the protein in the supernatant was not caused by a downregulation of yopT expression, we cloned yopT,sycT under the control of the Plac promoter (pIM279). This plasmid was then introduced in the sycD mutant and in the yscN mutant. As shown in Fig. 3, YopT was secreted by the sycD mutant (Fig. 3B, lane 5) but not by the yscN mutant (Fig. 3B, lane 6). We did not observe an accumulation of YopT in the total cell proteins of the yscN mutant, which can result from a degradation of the protein. As is the case for the other Yops, the 35.5 kDa protein was not secreted when bacteria were grown at room temperature or in the presence of Ca2+ (data not shown). We conclude from these experiments that ORF1 encodes a new Yop protein secreted by the Ysc machinery, and we have called it YopT.

ORF2 encodes SycT, the chaperone of YopT

As the chaperones are needed to allow secretion of their cognate Yop, we wondered whether ORF2 was necessary for secretion of YopT. We created an in frame deletion of codons 16–94 in ORF2, and we replaced the wild-type gene with the mutated allele in the pYV plasmid of the polymutant ΔHOPEM strain to obtain strain MRS40 (pIML422) or ΔHOPEM,ORF2. We then compared the proteins secreted by the ΔHOPEM, the ΔHOPEM,yopT and the ΔHOPEM,ORF2 strains. As shown in Fig. 3A (lane 6) and B (lane 3), YopT disappeared from the culture supernatant of the ΔHOPEM,ORF2 mutant, but reappeared when plasmid pIML266, encoding SycT under the control of the plac promoter, was introduced (Fig. 3B, lane 4). As expected from the sequence analysis, the product of ORF2 thus behaved like a chaperone for YopT, and we called it SycT. To analyse whether SycT is required only for the secretion of YopT, we introduced the sycTΔ16–94 allele in the wild-type strain E40 (pYV40) to obtain E40 (pIM402), and we analysed the pattern of secreted proteins. As shown in Fig. 3A (lane 3), the sycT mutation did not affect the secretion of any other Yop.

To confirm that SycT acts as a chaperone for YopT, we tried to visualize the binding of SycT to YopT by an overlay method (Wattiau and Cornelis, 1993). As shown in Fig. 4, radiolabelled SycT reacted with immobilized YopT and with a YopT–Cya hybrid (see below) containing 124 N-terminal residues of YopT. We conclude from these results that SycT binds YopT and that at least one binding domain for the chaperone is located in the first 124 amino acids of YopT.

Figure 4.

. A. SDS–PAGE and autoradiography of total extracts of E. coli LK111 (pGP1-2) (pBluescript KS−) (left) and LK111 (pGP1-2) (pIM154) (right) after induction of the T7 expression system in the presence of labelled methionine and cysteine. The arrows points to SycT. B. Binding of SycT to YopT. Left, Coomassie blue staining of secreted proteins. Right, overlay assay. Proteins were transferred to nitrocellulose and incubated with the extract shown in (A). The membrane was then washed and autoradiographed. Lane 1, E40 (pYV40) (pAB29); lane 2, KNG22703 (pPW2269) (pIM163); lane 3, KNG22703 (pPW2269) (pBluescript KS−).

Transcription analysis of the yopT,sycT operon

We analysed the influence of the growth conditions and of the transcriptional activator VirF (Cornelis et al., 1989; Lambert de Rouvroit et al., 1992) on the transcription of yopT,sycT by Northern blotting. For comparison, we analysed the transcription of yopE in parallel. As can be seen in Fig. 5, maximal expression of yopT,sycT was observed when bacteria were grown at 37°C in the absence of Ca2+. Transcription of yopT,sycT was dependent on VirF, like all the other yop genes.

Figure 5.

. Northern blot analysis showing the influence of Ca2+, temperature and VirF on transcription of the yopT,sycT operon (top) and of yopE (bottom). Bacteria were grown at 22°C or 37°C in BHI supplemented with oxalate (20 mM) and with 2.5 mM Ca2+. Total RNA was extracted and probed with an internal fragment of sycT or a PCR-amplified yopE.

YopT is not a component of the delivery apparatus

To determine whether YopT is an element of the delivery apparatus like YopB and YopD, we studied the delivery of the YopE130–Cya, a calmodulin-dependent hybrid adenylate cyclase used as reporter (Sory and Cornelis, 1994), by isogenic YopT+ and YopT strains. We infected PU5-1.8 macrophages and monitored the accumulation of cAMP inside the eukaryotic cells. Infection with both strains led to cAMP increase in the macrophages (4.6 and 7.4 nmol cAMP mg−1 protein, respectively, versus 0.6 nmol detected after infection with a translocator yopB mutant). These results indicate that YopT is not required for the delivery of YopE, and presumably also the other effector Yops, into eukaryotic cells, and they suggest that YopT could be a new effector Yop.

YopT is translocated inside eukaryotic cells

To determine whether YopT is delivered inside macrophages, we also used the adenylate cyclase reporter strategy (Sory and Cornelis, 1994). We engineered a hybrid gene consisting of the first 124 codons of yopT fused to cyaA′. The hybrid yopT124–cyaA′ gene was cloned downstream from the yopE promoter in a pACYC184 derivative, giving plasmid pAB29.

To test the construct, we introduced plasmid pAB29 in Y. enterocolitica E40 (pYV40), and we induced Yop synthesis and secretion. As shown in 3Fig. 3C, the transconjugant secreted a hybrid protein of the expected molecular mass (57.4 kDa). We then tested the adenylate cyclase (AC) activity of the culture supernatant. As a control, we introduced pAB29 in the yscN secretion mutant E40 (pMSL41). The supernatant of E40 (pAB29) had a significant AC activity in the presence of calmodulin (11.6 ± 3.8 units ml−1) but not in the absence of calmodulin (0.04 ± 0.02 units ml−1). No AC activity was detected in the supernatant of the yscN mutant carrying pAB29. These results confirm that YopT is a secreted protein, and they show that the secretion domain is located within the first 124 amino acids or codons of YopT. They also confirm that the YopT–Cya fusion is active.

The injection of poorly produced Yops into eukaryotic cells can be improved by using a polymutant Yersinia strain unable to produce the major effector Yops (Håkansson et al., 1996; Iriarte et al., 1998). To study the internalization of YopT124–Cya, we therefore introduced plasmid pAB29 and, as a control, plasmid pMS111 encoding YopE130–Cya in the polymutant ΔHOPEM strain. As a second control, we also introduced pAB29 in the isogenic strain carrying an additional mutation in yopB, ΔHOPEMB. We then infected PU5-1.8 macrophages and measured the accumulation of cAMP inside the cells. As shown in Table 1, cAMP was detected in macrophages infected with strain ΔHOPEM (pAB29) or ΔHOPEM (pMS111) but not with the isogenic yopB mutants. These results indicate that, like YopE130–Cya, YopT124–Cya is internalized in macrophages by a mechanism that requires YopB. To test whether TyeA is required for the delivery of YopT, we introduced plasmids pAB29 and pMS111 in the ΔHOPEM strain carrying a mutation in tyeA (but not in yopB). Infection of macrophages with the former strain but not with the latter one led to a significant increase in cAMP, indicating that TyeA is not required for the delivery of YopT as it is for YopE (Table 1).

Table 1. . Delivery of the YopE130–Cya and YopT124–Cya hybrid proteins into PU5-1.8 macrophages by Y. enterocolitica E40. a. Data (nmol cAMP mg−1 protein) are means ± SD of at least three experiments, each one done in duplicate. Experiments were carried out in the presence of cytochalasin D.Thumbnail image of

YopT mediates a cytotoxic effect on eukaryotic cells

To discover the role of YopT, we infected HeLa cells with Y. enterocoliticaΔHOPEM, ΔHOPEM,ORF1 (now called ΔHOPEMT) and, as a control, ΔHOPEMB. Cells infected with the ΔHOPEM strain detached more easily from the plastic plates than those infected with ΔHOPEMT or ΔHOPEMB strains. The ΔHOPEM strain induced a clear cytotoxic effect on all the HeLa cells characterized by a rounding up of the cells (Fig. 6). This cytotoxic effect was not observed when HeLa cells were infected with the ΔHOPEMT or the ΔHOPEMB strains, but it could be completely restored after the introduction in the ΔHOPEMT strain of plasmid pIM279, expressing YopT and SycT under the control of the Plac promoter (Fig. 6). The same cytotoxic response was also observed when J774A.1 or PU5-1.8 macrophages were infected instead of HeLa cells (data not shown).

Figure 6.

. Effect of YopT on cultured HeLa cells. Bacteria of the indicated strains were pregrown for 2 h at room temperature, washed with saline and used to infect HeLa cells with a multiplicity of infection of 100. The morphology of the cells was recorded after 2 h of infection.

These results indicate that YopT induces a cytotoxic response in eukaryotic cells in the absence of any other known effector Yop. The induction of this cytotoxic effect requires an intact delivery apparatus, indicating that YopT needs to reach the cytosol of the target cell to exert its action.

In order to make a better comparison of the cytotoxic effect induced by YopT and YopE, we infected HeLa cells with the wild-type strain MRS40 (pYV40), the yopE mutant MRS40 (pAB4052), the yopT mutant MRS40 (pIM409) and a double mutant yopE,yopT MRS40 (pIM425). As shown in Fig. 6, cells infected with the wild-type strain, the yopE mutant or the yopT mutant exhibited a cytotoxic effect after 2 h of infection, while cells infected with the double mutant yopE,yopT did not. These results suggest that both YopE and YopT are able to induce a cytotoxic effect in eukaryotic cells. Taking into account that the amount of YopE secreted by Y. enterocolitica is much higher than the amount of YopT, the latter seems to be more cytotoxic than YopE.

We studied the influence of YopT on cell viability using the trypan blue exclusion assay. Less than 1% of the cells infected with the wild-type strain, the yopE mutant or the yopT mutant stained with trypan blue, indicating that neither YopT nor YopE alone have an influence on the membrane integrity of the target cells.

YopT leads to disruption of the cytoskeleton structure

We stained the actin filaments of HeLa cells infected with the ΔHOPEM, ΔHOPEMB or ΔHOPEMT strains with fluorescein isothiocyanate (FITC)-conjugated phalloidin. As shown in Fig. 7, HeLa cells infected with the ΔHOPEM strain were smaller than uninfected cells, rounded up and their cytoskeleton structure was disrupted. The actin filaments broke up and actin appeared as dispersed patches in the cytosol. In contrast, HeLa cells infected with the ΔHOPEMT strain, unable to produce YopT, or the ΔHOPEMB strain, unable to translocate YopT, had an intact cytoskeleton, a flat irregular shape and clear stress fibres of actin filaments. The effect of YopT is thus very reminiscent of that of YopE.

Figure 7.

. YopT-mediated disruption of the actin filaments. HeLa cells were infected with the ΔHOPEM, ΔHOPEMB and ΔHOPEMT strains. After 2.5 h of infection, actin filaments were stained with FITC-conjugated phalloidin. Arrows indicate stress fibres. Asterisks indicate patches of actin.

Presence of YopT in the other pathogenic Yersinia spp

To investigate whether the other pathogenic Yersinia species also produce YopT, we analysed the proteins secreted by Y. pseudotuberculosis pYPIII and Y. pestis EV76P by Western blotting. As a control, we included Y. enterocolitica KNG22703, MRS40 and its ΔHOPEM derivative. YopT was detected among the proteins secreted by Y. pseudotuberculosis, but it was not detected among the proteins secreted by Y. pestis (Fig. 8). However, while YopT was detected among the proteins secreted by Y. enterocolitica KNG22703 and MRS40ΔHOPEM, it was not detected among the proteins secreted by the wild-type MRS40. From these results, we can conclude that YopT is secreted by Y. pseudotuberculosis, but we cannot exclude that YopT is also secreted in lower amounts by Y. pestis.

Figure 8.

. Secretion of YopT by different Yersinia spp. Lane 1, Y. enterocolitica KNG22703; lane 2, Y. pseudotuberculosis, pYPIII; lane 3, Y. pestis EV76; lane 4, Y. enterocoliticaΔHOPEM (polymutant derivative of MRS40); lane 5, Y. enterocolitica MRS40.

Role of YopT in colonization of the mouse Peyer's patches

To have a rough insight into the role of YopT in pathogenicity, we infected four groups of five mice each intragastrically with 1010 bacteria of the wild-type strain MRS40 (pYV40), the yopE mutant and the yopT mutant and determined the number of bacteria in Peyer's patches after 2 days of infection. The capacity of the yopE mutant to colonize Peyer's patches was 100 times (2 ± 2 × 104 bacteria/Peyer's patches) lower than that of the wild-type strain (7 ± 4 × 106 bacteria/Peyer's patches). However, the yopT mutant was not affected in its capacity to multiply in Peyer's patches (3 ± 1 × 106 bacteria/Peyer's patches).

Discussion

In this work, we have characterized a bicistronic operon situated between yopQ and yopM (co-ordinates 8.4–9.8) in the pYV227 plasmid of Y. enterocolitica. The first gene encodes a new Yop of 35.5 kDa that we called YopT. Like the other Yops, YopT was secreted by the Ysc apparatus, and its secretion domain is located within the first 124 amino acids or codons. The product of the second gene, SycT, closely resembles the SycE chaperone (Wattiau and Cornelis, 1993), and it behaved as the specific chaperone of YopT. As for SycE and SycH, SycT was found to bind specifically to its cognate, YopT, and was only required for the secretion of YopT but not for the secretion of any of the other Yops. The binding domain of SycT is localized within the 124 N-terminal residues. The proximity between yopT and sycT is reminiscent of that of yopE–sycE (Wattiau and Cornelis, 1993) and yopH–sycH (Wattiau et al., 1994).

Construction and analysis of a YopT–Cya hybrid protein showed that YopT is translocated into target cells by a mechanism that requires YopB. YopT is thus a new effector Yop protein, and its translocation domain, like its chaperone-binding domain, is located within the first 124 amino acids. At variance with YopE and YopH, YopT did not require TyeA for its translocation into eukaryotic cells (Iriarte et al., 1998).

YopT induced a cytotoxic effect on HeLa cells and J774A.1 or PU5-1.8 macrophages in the absence of any other Yop effector. There was a perfect correlation between the translocation of YopT124–Cya and the induction of a cytotoxic response in cells infected with a YopT-producing strain. The induced cytotoxic response required an intact delivery apparatus, reinforcing the idea that YopT must reach the cytosol of the target cell to exert its action.

The main cytotoxin of Y. pseudotuberculosis seems to be YopE (Rosqvist et al., 1991). However, in Y. enterocolitica, a yopE mutant is as cytotoxic as the wild-type strain (Hartland et al., 1994; this work), while a double yopE,yopT mutant was not cytotoxic after 2 h of infection. YopT- and YopE-induced cytotoxicity was observed after the same length of infection. Taking into account that the amount of YopT secreted is lower than that of YopE, our results suggest that YopT is more cytotoxic than YopE.

One may wonder why Y. enterocolitica needs two cytotoxins, YopE and YopT, whose action results in the disruption of actin filament structure. Several possibilities could be foreseen. First, the two cytotoxins could reinforce each other by acting through different pathways. YopE and YopT could, for instance, act at different levels in the signalling cascade involving the small Rho GTP-binding proteins (Ridley and Hall, 1992; for a review, see Aktories, 1997). A second hypothesis to explain the multiplicity of effectors would be that all the effectors are not delivered to the same target cell in vivo. In support of this hypothesis, we have shown recently that Yops could be divided into two subsets for their delivery into eukaryotic cells. YopE and YopH require TyeA for delivery, whereas YopO, YopM and YopP do not (Iriarte et al., 1998). While the significance of this difference remains unknown, it suggests that different Yops could be delivered to different cell types depending on TyeA. In contrast to YopE and YopH, YopT does not require TyeA to be delivered into the target cell, and this suggests that YopT would be delivered to a different cell type than YopE and YopH. Yersinia would thus use two different cytotoxins, YopE and YopT, to disorganize the actin filaments of different target cells. This hypothesis, which remains very speculative, awaits further investigation. Thirdly, the two cytotoxins could have their optimal effect on the same cells but at different stages of their development.

We did not observe any significant difference between the invasion capacity of the wild-type strain and the yopT mutant. However, we believe that these results do not rule out a role for YopT in virulence. It is likely that the mouse model and the infection procedure used bypass a number of steps that could be critical in the complexity of a human infection and its transmission chain.

Other effector Yops, namely YopE, YopH and YpkA/YopO of Y. pseudotuberculosis, have also been shown to induce a cytotoxic effect on eukaryotic cells. The cytotoxic effect induced by YopT and YopE in Y. enterocolitica closely resembles that induced by YopE in Y. pseudotuberculosis (Rosqvist et al., 1991). The cells detach from the plastic support, round up and become smaller than the uninfected cells. The actin filaments break up, the stress fibres disappear and the actin appears as dispersed patches in the cytosol.

YopH induces a weaker cytotoxic effect on host cells that can only be detected in the absence of YopE (Rosqvist et al., 1990; Bliska et al., 1993; Andersson et al., 1996). The infected cells become round in shape and detach from the substrate. YopH induces dephosphorylation of p130Cas and FAK, two proteins present in the focal adhesion (FAs) sites, which results in the disruption of the F-actin stress fibres, the peripheral focal complexes and, thereby, inhibition of bacterial uptake (Black and Bliska, 1997; Persson et al., 1997).

According to Håkansson et al. (1996), infection of HeLa cells with a Y. pseudotuberculosis polymutant strain yopH,yopM,yopE,yopK overexpressing YpkA/YopO also results in a cytotoxic response. The morphology of the infected cells is nevertheless different from that of cells infected with the YopE-producing strain. YpkA/YopO-induced cytotoxicity is characterized by a rounding up of the cells without detachment from the extracellular matrix, resulting in very pronounced and branched retraction fibres (Håkansson et al., 1996). YpkA is mainly found associated with the plasma membrane of the target cell (Håkansson et al., 1996).

Other cytotoxins have also been identified in pathogens endowed with a type III secretion system. ExoU/PepA, a protein secreted by the type III secretion pathway of Pseudomonas aeruginosa, is also a cytotoxin involved in bacterial-mediated epithelial cell damage (Finck-Barbançon et al., 1997; Hauser et al., 1998). The exoenzyme S (ExoS) of P. aeruginosa is a bifunctional enzyme endowed with cytotoxic and an ADP-ribosylating activity. The cytotoxicity depends on a domain that is similar to YopE (Frithz-Lindsten et al., 1997). The Salmonella typhimurium tyrosine phosphatase SptP and the Salmonella typhi StpA also share some similarity to YopE in their N-terminal part and also lead to disruption of the cell actin cytoskeleton (Arricau et al., 1997; Fu and Galan, 1998).

Finally, YopT shows some similarity to the C-terminal end of p76, an immunoglobulin binding protein encoded by a 13.4 kb segment present in the chromosome of the serum-resistant strains of Haemophilus somnus (Cole et al., 1992; 1993; Corbeil et al., 1997). This 13.4 kb segment contains two 1.2 kb direct repeats (DRs), and the coding region for p76 starts in DR1 and extends beyond DR2, which has flanking inverted repeats similar to insertion elements (Cole et al., 1993). The similarity between YopT (322 amino acids) and p76 (874 amino acids) is limited to the C-terminal end of p76 (30.5% similarity in 308 amino acids overlap). This similarity is thus not necessarily significant with regard to protein function. However, it raises the question whether H. somnus also has a type III secretion system. In agreement with this hypothesis, Cole et al. (1993) and Corbeil et al. (1997) have shown that, even though p76 does not contain any obvious signal sequence, it can easily be extracted from the bacterial surface with sarkosyl, indicating that it is loosely associated with the outer membrane. This is in agreement with the fact that p76 is highly hydrophilic and does not have hydrophobic regions characteristic of membrane-spanning proteins (Cole et al., 1993). However, to our knowledge, no type III secretion–translocation system has been identified up to now in H. somnus.

Experimental procedures

Bacterial strains, plasmids and growth conditions

This work was carried out with Y. enterocolitica serotype O:9 strains W22703 (pYV227) (Cornelis and Colson, 1975) and E40 (pYV40) (Sory et al., 1995). Y. enterocolitica KNG22703 (pYV227) (Kaniga et al., 1991) and MRS40 (pYV40) (Sarker et al., 1998a) are mutant strains in which the blaA gene encoding β-lactamase A (Cornelis and Abraham, 1975) was replaced by the luxAB gene (Kaniga et al., 1991). E. coli LK111 was used for standard manipulations. E. coli SM10 lambda pir+ constructed by Miller and Mekalanos (1988) was used to deliver mobilizable plasmids in Y. enterocolitica. The plasmids used in this study are listed in Table 2.

The strains were grown routinely in tryptic soy broth (TSB) and plated on tryptic soy agar. For in vitro induction of the yop genes, Y. enterocolitica was grown in brain–heart infusion (BHI), supplemented with 20 mM sodium oxalate, 20 mM MgCl2 and 0.4% glucose. Selective agents were used at the following concentrations: ampicillin 200 μg ml−1, chloramphenicol 10 μg ml−1, nalidixic acid 35 μg ml−1, streptomycin 100 μg ml−1, sucrose 5% and 0.4 mM arsenite (Neyt et al., 1997).

Eukaryotic cell growth conditions

HeLa cells and the monocyte–macrophage cell lines PU5-1.8 (ATCC TIB61) and J774A.1 (ATCC TIB67) were grown as described by Sory and Cornelis (1994), Sory et al. (1995) and Mills et al. (1997).

DNA sequencing and computer analysis

The sequence of 2314 bp of pYV227 encoding YopT and SycT was performed on both strands using a LICOR automatic sequencer. The nucleotide sequence of yopT and sycT has been submitted to GenBank under accession number AF054981.

DNA and protein sequences were analysed using the BLAST (Altschul et al., 1990; Gish and States, 1993) and FASTA programs (Pearson and Lipman, 1988). The isoelectric point and the antigenic domains of YopT were determined using the Genetics Computer Group (University of Wisconsin, Madison, WI, USA) sequence analysis software package computer program. The signal sequence was analysed with the SIGSEQ program (Popowicz and Dash, 1989).

Contruction of a yopT mutant and a yopE, yopT double mutant

To construct a yopT mutant, we created an internal out of frame deletion of 428 bp (after codon 135) by digestion of pIM157 with NheI and EcoRI, followed by Klenow treatment and religation. The deletion was checked by sequence analysis, and the resulting plasmid was called pIM166. In a second step, a XbaI–EcoRI fragment of pIM166 encoding YopT135 (and 94 codons of sycT ) was subcloned in the corresponding sites of pBluescript KS– giving pIMB199. A SalI–XbaI fragment of pIMB199 encoding YopT135 and SycT94 was subcloned into the corresponding sites of the suicide vector pKNG101 (Kaniga et al., 1991). The resulting mutator plasmid, pIML200, was mobilized in Y. enterocolitica E40, Y. enterocolitica MRS40 and Y. enterocolitica MRS40 (pABL403) by conjugation, and the wild-type gene was replaced by the yopT135 allele by double recombination, giving strains E40 (pIM409), MRS40 (pIM409) and MRS40 (pIML421), respectively, as described by Kaniga et al. (1991). Allelic exchange was checked by polymerase chain reaction (PCR) using oligonucleotides MIPA 370 (5′-TGCTCTAGACATGGACAGTATTCACGGACA-3′) and MIPA 399 (5′-ACGCGTCGACAGCT-

GTTGCATAAGTTCTG-3′).

The double mutant yopE,yopT was constructed by introducing the yopT135 allele in the yopE mutant MRS40 (pAB4052) (Mills et al., 1997).

Construction of a sycT mutant

To construct a sycT mutant, we created an inframe deletion of 234 bp in sycT carried in plasmid pIM154, by digestion with ClaI and HindIII, followed by Klenow treatment and religation. The resulting plasmid called pIM161 encodes SycTΔ16–94. In a second step, a SalI–XbaI fragment of pIM161 containing the mutated sycT allele was subcloned into the corresponding sites of the suicide vector pKNG101, giving the mutator plasmid pIM164. The sycTΔ16–94 allele was then introduced into the pYV plasmid of Y. enterocolitica E40 (pYV40) and Y. enterocolitica MRS40 (pABL403) by double recombination, giving E40(pIM402) and MRS40 (pIML422). Allelic exchange was checked by PCR using oligonucleotides MIPA 275 (5′-CCGGAATTCAGACAACCTTCACAGAACTT-3′) and MIPA 276 (5′-ACGCGTCGACTTGGAGGCATCACTGAAATAC-3′).

Radiolabelling of SycT and study of the binding of SycT to YopT

We cloned sycT downstream from the T7 promoter in pBluescript KS−, giving plasmid pIM154. Exclusive labelling of SycT was obtained in E. coli as described by Wattiau and Cornelis (1993) using the specific T7 expression system (Tabor and Richardson, 1985) and a mixture of [35S]-methionine and [35S]-cysteine (Promix; Amersham). Binding of the radioactive SycT protein on membrane-immobilized YopT was performed as described by Wattiau and Cornelis (1993).

Preparation of a serum against YopT

Two synthetic peptides corresponding to amino acid residues 89–108 and 135–150 of YopT (prepared at the Ludwig Institute for Cancer Research, Brussels, Belgium) were linked to keyhole limpet haemacyanin and bovine serum albumin and used to raise rabbit antibodies (Laboratoire d'Hormonologie, Marloie, Belgium). The antiserum was used at a dilution of 1:100 for immunoblot analysis.

RNA analysis

DNA and RNA analysis were performed as described by Michiels et al. (1991). The probe used to detect the yopT,sycT transcript consists of an internal EcoRV–HindIII fragment of sycT. To detect the yopE transcript, we used a PCR-amplified yopE.

Cytotoxic assay

The eukaryotic cells were infected with the different Yersinia strains pregrown at room temperature for 2 h in BHI, washed and resuspended in saline. The morphology of the cells was observed after 1, 2, 2.5 or 3 h of infection by phase contrast microscopy. Cytotoxicity was manifested as a rounding up of the cells.

Immunofluorescence staining and confocal microscopy analysis

Hela cells were grown on coverslips and infected with different strains. After 2.5 h of infection, the first sign of cytotoxicity appeared, and the cells were fixed in 2% paraformaldehyde for 20 min. After washing with PBS, membranes were permeabilized with PBS–0.5% Triton X-100 for 10 min. Actin microfilaments were labelled after incubation at 37°C for 30 min with FITC-conjugated phalloidin. Samples were mounted in Mowiol, and cells were examined on a confocal laser scanning microscopy equipped with dual detectors and an argon–krypton laser (MRC 1024; Bio-Rad Laboratories).

Determination of Hela cell viability using trypan blue staining

HeLa cells infected with the different strains for 2–4 h were washed once with prewarmed PBS containing 0.01% CaCl2, 0.01% MgCl2 and stained with 0.4% trypan blue (Sigma) for 5 min. The cells were washed once with the same PBS and examined by phase-contrast microscopy.

Virulence assay

Virulence of the yopT mutant was studied as described by Iriarte et al. (1995).

Acknowledgements

We thank Isabelle Lambermont and Corinne Kerbourch for excellent technical assistance. We are grateful to Anne Boland for constructing pAB29 encoding YopT124–Cya, and we also thank Aoife Boyd and Cecile Geuijen for suggestions and for critical reading of the manuscript. This work was supported by the Belgian ‘Fonds National de la Recherche Scientifique Médicale’ (convention 3.4595.97), the ‘Direction générale de la Recherche Scientifique-Communauté Française de Belgique’ (Action de Recherche Concertée' 94/99-172) and by the ‘Interuniversity Poles of Attraction Program — Belgian State, Prime Minister's Office, Federal Office for Scientific, Technical and Cultural Affairs’ (PAI 4/03). Confocal microscopy was funded by grant 9.4531.94F from FRSM (Loterie Nationale).

Ancillary