LcrV is a channel size-determining component of the Yop effector translocon of Yersinia

Authors


Abstract

Delivery of Yop effector proteins by pathogenic Yersinia across the eukaryotic cell membrane requires LcrV, YopB and YopD. These proteins were also required for channel formation in infected erythrocytes and, using different osmolytes, the contact-dependent haemolysis assay was used to study channel size. Channels associated with LcrV were around 3 nm, whereas the homologous PcrV protein of Pseudomonas aeruginosa induced channels of around 2 nm in diameter. In lipid bilayer membranes, purified LcrV and PcrV induced a stepwise conductance increase of 3 nS and 1 nS, respectively, in 1 M KCl. The regions important for channel size were localized to amino acids 127–195 of LcrV and to amino acids 106–173 of PcrV. The size of the channel correlated with the ability to translocate Yop effectors into host cells. We suggest that LcrV is a size-determining structural component of the Yop translocon.

Introduction

An important virulence mechanism that is common among several Gram-negative bacterial pathogens is the type III secretion/translocation system, which mediates the secretion and injection of antihost factors into eukaryotic cells via a contact-dependent mechanism. Type III secretion/translocation systems have been identified in a number of animal and plant pathogens, and several components of these systems also show homology to components of the secretion/assembly apparatus of the flagellar system (reviewed by Hueck, 1998). Structurally, the type III apparatus resembles the basal body of flagella, with an ‘injection needle’ at the tip replacing the hook and the flagellar filament. Such a ‘microinjection syringe’ has also been isolated for the SPI-1 system (encoded by pathogenicity island 1) of Salmonella typhimurium (Kubori et al., 1998). A similar structure has been visualized in Shigella flexneri (Blocker et al., 1999). Recently, the needle part of these isolated structures from Shigella and Salmonella has been shown to contain a small protein, MxiH and PrgI respectively (Kubori et al., 2000; Tamano et al., 2000). Homologues of MxiH and PrgI appear to be present in type III systems from animal pathogens but not from plant pathogens. Even if the systems from different bacteria show high similarity, the biological activity of the injected effector proteins varies between different pathogens. Intracellular delivery of key virulence effectors such as YopE, YopH and YopJ of pathogenic Yersinia into eukaryotic cells results in a general reduction in phagocytic capability as well as suppression of induction of inflammatory cytokines (Rosqvist et al., 1988; 1990; Fällman et al., 1995; Palmer et al., 1998; Schesser et al., 1998). In contrast, translocation of the S. typhimurium effector proteins by the type III system (SPI-1) induces the uptake and induction of a massive inflammatory response (for reviews, see Gálan, 1996; 1998). There is evidence from several systems that the overall mechanism of secretion and translocation is functionally conserved. This is supported by studies showing that type III-targeted effector proteins from different pathogens can be secreted and translocated by heterologous type III secretion/translocation systems (reviewed by Hueck, 1998).

The ability of pathogenic Yersinia species to evade the primary host defence and cause systemic infections in various animal hosts relies strictly on the delivery of Yop (Yersiniaouter proteins) virulence effectors into the cytosol of host cells via the type III secretion/translocation system (for reviews, see Cornelis and Wolf-Watz, 1997; Cornelis et al., 1998; Schesser et al., 2000). The type III secretion/translocation system of Yersinia has been studied extensively and has become the paradigm for this virulence mechanism. The components of the Yersinia type III system together with the secreted Yop proteins are encoded by a common plasmid. Several of the secreted Yops are involved in the delivery of effector Yops into the cytosol of the target cell. Secretion and translocation of Yops only occur at the zone of contact between the bacterium and the host cell, i.e. the Yop effectors are delivered by a vectorial mechanism where no effector proteins are secreted to the surrounding medium (Rosqvist et al., 1994; Persson et al., 1995; Boland et al., 1996).

The predominant view is that both YopB and YopD are absolutely required for Yop effector translocation (Rosqvist et al., 1994; Sory and Cornelis, 1994; Håkansson et al., 1996; Francis and Wolf-Watz, 1998; Fields et al., 1999). However, Lee and Schneewind (1999) have reported conflicting results for Yersinia enterocolitica, which suggested that YopB is not needed for the translocation of Yop effectors in a HeLa cell infection model. Using the same strains as in this latter study, we have been unable to reproduce the findings by Schneewind's laboratory (R. Nordfelth and H. Wolf-Watz, unpublished results). Recently, Pettersson et al. (1999) and Fields et al. (1999) presented evidence that LcrV is surface located and essential for translocation. Interestingly, some sera directed against LcrV provide both passive protection against Yersinia pestis infections and also prevent the translocation of Yop effectors in cell infection experiments (Hill et al., 1997; Fields et al., 1999; Pettersson et al., 1999). From studies using monoclonal antibodies, it appears that only antibodies directed against a certain epitope(s) in the central part of LcrV can provide protection in animal infections as well as block translocation (Hill et al., 1997; Pettersson et al., 1999). Therefore, it is not surprising that some antibodies are not blocking Yop effector translocation (Fields et al., 1999). Thus, at least three of the proteins secreted by the type III system of Yersinia, YopB, YopD and LcrV, are required for Yop effector translocation across the eukaryotic cell membrane. Strains mutated in the yopB, yopD and lcrV genes are all deficient for Yop delivery (Rosqvist et al., 1991; Sory and Cornelis, 1994; Hartland et al., 1995; 1996; Boland et al., 1996; Fields et al., 1999; Pettersson et al., 1999). In these mutants Yop effectors accumulate at the zone of contact between the pathogen and the target cell (Rosqvist et al., 1994; Persson et al., 1995; Håkansson et al., 1996; Francis and Wolf-Watz, 1998; Pettersson et al., 1999). YopB contains two central hydrophobic domains, and YopD contains one hydrophobic domain (Håkansson et al., 1993); therefore, these proteins have been proposed to interact with the target cell membrane. Håkansson et al. (1996) have shown that YopB is essential for the ability of Yersinia to form pores/channels mediating lysis of sheep erythrocytes. More recently, Neyt and Cornelis (1999a) extended these findings and showed that, in addition to YopB, YopD was also required for lytic activity. Furthermore, Tardy et al. (1999) were able to recover YopB and YopD from liposomes present in Y. enterocolitica cultures induced for Yop secretion. When these proteoliposomes were fused to a planar lipid bilayer, membrane channel activity was observed. The channel formation was dependent on YopB and YopD (Tardy et al., 1999). Based on these results, YopB and YopD were proposed to form the translocation channel in the eukaryotic cell membrane. When overexpressed in the Escherichia coli cytosol, YopB, YopD and LcrV all interact (Sarker et al., 1998; Neyt and Cornelis, 1999b). Therefore, it is possible that these three proteins are part of a translocon, a channel-forming complex whose function is to translocate Yop effectors into target cells.

Here, we show that, together with YopB and YopD, LcrV is also required for channel formation in erythrocytes. In addition, we found that purified LcrV induced channel formation in planar lipid bilayers. Replacement of LcrV with the corresponding protein PcrV of P. aeruginosa resulted in the formation of smaller channels. A channel size-determining region was localized to a central domain of LcrV (amino acids 127–195), and a strict correlation was found between the pore/channel size and the efficiency of translocation. Our findings support the concept that translocation occurs via a channel structure in the host cell membrane, and we suggest that LcrV is a size-determining component of the Yop effector translocon.

Results

LcrV and PcrV induce channels in the membranes of sheep erythrocytes

In agreement with the observations that YopD of Y. enterocolitica and LcrV of Y. pestis are required for the contact-dependent haemolysis on infected erythrocytes (Nilles et al., 1998; Neyt and Cornelis, 1999a), we found that yopD and lcrV mutant strains of Yersinia pseudotuberculosis were unable to induce haemolysis (Fig. 1). However, lcrV mutant strains are known to be downregulated for Yop expression/secretion (Pettersson et al., 1999). To circumvent the effect of LcrV on Yop regulation, we constructed a double lcrQ, lcrV mutant strain that exhibited derepressed Yop expression (Fig. 2). In contrast to the lcrQ mutant strain, the lcrQ, lcrV double mutant was unable to induce lysis of the infected sheep red blood cells (SRBCs). This phenotype resulted from the lack of LcrV, as the lytic effect was restored when lcrV was supplied in trans (Fig. 1). Thus, LcrV is essential for the Yersinia-mediated lysis of SRBCs and, hence, all three proteins that are essential for the translocation of Yops (YopB, YopD and LcrV) are also required for lytic activity on infected erythrocytes.

Figure 1.

Lytic activity of different Y. pseudotuberculosis strains during infection with sheep red blood cells (SRBCs). The activity is measured as released haemoglobin by spectrophotometric quantification of absorbance at 545 nm. Experiments were performed in triplicate, and the bars represent mean values ± SD. The relevant genotype of the strains used is shown (for details see Table 2). plcrV denotes plasmid pTB7 with lcrV cloned under the control of the tac promoter (Table 2).

Figure 2.

Proteins secreted by different Y. pseudotuberculosis strains. Yop expression and secretion was induced by growth at 37°C in BHI medium depleted of calcium. Secreted proteins were precipitated, separated on SDS–PAGE and visualized by staining with Coomassie brilliant blue. The relevant genotype of the strains used is shown above the respective lane (for details, see Table 2).

We showed recently that two regions of LcrV, amino acids 160–171 and amino acids 316–326, are essential for Yop effector translocation (Pettersson et al., 1999). Strains expressing variants of LcrV lacking these regions showed no haemolytic effect on the infected SRBCs (Fig. 1). The LcrV homologue in P. aeruginosa, PcrV, can substitute for LcrV with respect to Yop regulation, secretion and translocation (Pettersson et al., 1999). When PcrV was expressed in the lcrQ, lcrV mutant strain, only a low level of lytic activity was observed after infection of SRBCs (data not shown). However, the low level of lysis did not allow analysis of the channel size using the carbohydrate exclusion assay. Therefore, a yopK, lcrV mutant background was used, as yopK mutants show increased haemolytic activity (Holmström et al., 1997). When PcrV was expressed by the haemolysis-negative yopK, lcrV mutant strain, lysis of infected SRBCs was observed (Fig. 3). The PcrV-associated lytic activity was about 10% of that of an isogenic strain expressing LcrV (Fig. 3). These results show that LcrV (or PcrV) is also required to induce the lysis of infected SRBCs. In addition, the domains of LcrV required for Yop translocation are also essential for the induction of lytic activity.

Figure 3.

Lytic activity on erythrocytes by Y. pseudotuberculosis strains expressing LcrV, PcrV and LcrV/PcrV hybrid proteins expressed in an lcrV/yopK (KV) mutant strain. SRBCs were infected with different strains of Y. pseudotuberculosis, and the amount of released haemoglobin was quantified spectrophotometrically. See Experimental procedures for details. The experiment was performed in triplicate, and the bars represent the mean values ± SD. The relevant genotype of the strains used is shown (for details, see Table 2). plcrV denotes plasmid pTB7 with lcrV cloned under the control of the tac promoter, and ppcrV denotes plasmid pLJ33 with pcrV cloned under the control of the tac promoter (Table 2).

A central domain of LcrV (amino acids 127–195) and PcrV (amino acids 106–173) determines channel size

The fact that LcrV was required for the lytic effect of Yersinia on SRBCs argues strongly for a role for this protein in the formation/induction of the translocation channel. To study LcrV/PcrV-dependent channel formation in greater detail, we investigated the properties of the induced channel in the erythrocyte membrane using a carbohydrate exclusion assay (Braun et al., 1987; Holmström et al., 1997). In this experiment, carbohydrates of different sizes [raffinose (diameter 1.2–1.4 nm), dextrin 15 (diameter 2.2 nm), dextran 4 (diameter 3–3.5 nm); Bhakdi et al., 1986; Schönherr et al., 1994] were present during the haemolysis assay. Only the largest sugar, dextran 4, protected the SRBCs from lysis (Fig. 4). In contrast, even raffinose, the smallest carbohydrate, reduced the lytic effect of PcrV, whereas dextrin 15 and dextran 4 had an even greater effect (Fig. 4), indicating that LcrV induces larger channels than PcrV. From the sizes of the carbohydrates and using 50% inhibition of the lytic activity as a measurement of channel size, we estimated the LcrV-induced channels to be about 3 nm in diameter, whereas those induced by PcrV were ≈ 2 nm in diameter. We conclude that the LcrV-dependent lytic activity results from formation of channels in the membrane of SRBCs and that the channels associated with LcrV and PcrV differ in size.

Figure 4.

Osmoprotection of infected SRBCs by different carbohydrates. Sheep erythrocytes were infected with different strains of Y. pseudotuberculosis in the presence of carbohydrates of different sizes, and the amount of released haemoglobin was measured. The lytic activity is presented as the percentage of the lysis in the absence of sugars. Each column represents the mean and SD of three independent experiments. See Experimental procedures for further details. KV denotes the yopK, lcrV genotype, and plcrV and ppcrV denotes plasmids pTB7 and pLJ33 respectively (Table 2).

As PcrV is able to substitute for LcrV (Pettersson et al., 1999), and the two proteins display only 41% identity at the amino acid level, we decided to generate LcrV/PcrV hybrid proteins (Fig. 5). All hybrid proteins were secreted and functionally similar to LcrV with respect to the secretion and expression of Yop proteins (Fig. 5). Hybrids 1, 2 and 3 containing the first 195 amino acids or more of LcrV were found to induce lysis similar to that induced by strains expressing LcrV on infected SRBCs (Fig. 3). These hybrids all formed channels of similar sizes to those induced by LcrV (diameter 3 nm; Fig. 4). Bacteria expressing hybrid 4, which contained the N-terminal 126 amino acids of LcrV, induced an intermediate level of lysis (Fig. 3). Interestingly, the apparent size of these channels was similar to that of the PcrV-associated channels (diameter 2 nm; Fig. 4). These results indicate that a central region of LcrV (amino acids 127–195) and PcrV (amino acids 106–173) is involved in determining the size of the channels formed in conjunction with YopB and YopD on infected erythrocytes.

Figure 5.

A. Schematic representation of LcrV, PcrV and LcrV/PcrV hybrid proteins. The hybrids 1–4 contain the sequence of LcrV and PcrV indicated to the right; black represents LcrV, and white represents PcrV sequence.

B. Secretion of Yops, LcrV, PcrV and LcrV/PcrV hybrid proteins. The lcrV mutant strain YPIII (pIB19) carrying plasmids encoding lcrV, pcrV and lcrV/pcrV hybrid genes was induced for the expression of LcrV, PcrV and LcrV/PcrV hybrid proteins by the addition of IPTG in calcium-depleted media at 37°C. The secreted proteins of the culture supernatant were precipitated and separated by SDS–PAGE followed by Coomassie brilliant blue staining or Western blotting using either anti-LcrV or anti-PcrV antiserum.

The lytic effect induced by the LcrV–PcrV hybrids reflects Yop translocation ability

To investigate whether PcrV and the different hybrid proteins could promote Yop effector translocation, HeLa cells were infected with strains carrying the different constructs, and translocation of YopE and YopH was analysed. YopE translocation was studied by monitoring the YopE-induced cytotoxic effect on infected HeLa cells, and the levels of YopH translocation were analysed by indirect immunofluorescence staining. When HeLa cells were infected with a strain expressing LcrV, a rapid cytotoxic response was observed, and high amounts of YopH was detected in the infected cells (Fig. 6). When PcrV was expressed in an isogenic lcrV mutant, a delayed YopE-mediated cytotoxic response was observed (Table 1). This translocation assay is very sensitive, and the observed difference in cytotoxic response between the two strains used indicates that the level of translocation associated with PcrV is much lower than that associated with LcrV. Using the less sensitive immunostaining method, only low levels of YopH translocation were detected when a PcrV-expressing strain was used. This suggests that the translocation of Yop effectors is less efficient in the strain expressing PcrV compared with the strain expressing LcrV. Expression of hybrids 1, 2 and 3 resulted in the rapid rounding up of the infected cells similar to that observed for the LcrV-expressing strain (Table 1). Furthermore, YopH was found to be translocated into HeLa cells when infected with strains expressing these hybrids (Fig. 6). Hybrid 4 exhibited an intermediate level of translocation of both YopE and YopH when compared with strains expressing LcrV or PcrV (Table 1; Fig. 6). Thus, these results show that there is a correlation between the ability to translocate Yops into HeLa cells and the size of the translocation channel.

Figure 6.

YopH translocation into HeLa cells by Y. pseudotuberculosis strains expressing LcrV, PcrV and LcrV/PcrV hybrids. HeLa cells were infected by a double yopK, lcrV mutant strain complemented with plasmids encoding LcrV, PcrV, hybrid 3 or hybrid 4. YopH was detected by indirect immunofluorescence labelling (see Experimental procedures for details).

Table 1. Cytotoxicity on HeLa cells by different strains of Y. pseudotuberculosis.
StrainRelevant genotypeCytotoxicity on HeLa cells
  1. a.++ cytotoxicity after 1 h of infection.

YPIII(pIB102)Wild type++
YPIII(pIB15519) yopK, lcrV
YPIII(pIB15519, pTB7) yopK, lcrV(plcrV+)++
YPIII(pIB15519, pLJ33) yopK, lcrV(ppcrV+)+ (delayed 2 h)
YPIII(pIB15519, pHyb 1) yopK, lcrV(pHyb 1)++
YPIII(pIB15519, pHyb 2) yopK, lcrV(pHyb 2)++
YPIII(pIB15519, pHyb 3) yopK, lcrV(pHyb 3)++
YPIII(pIB15519, pHyb 4) yopK, lcrV(pHyb 4)+ (delayed 1 h)

LcrV and PcrV induce differentially sized channels in lipid bilayers

A frequently used approach for studying channel-forming proteins is to study channel-forming ability in lipid bilayer model membranes (Benz and Chakraborty, 1992). Previous work using this approach showed YopB to have membrane-disrupting activity (Håkansson et al., 1996). When purified LcrV was added to the artificial membrane, stepwise increases in conductivity across the bilayer were observed when voltage was applied (Fig. 7A; for details, see Experimental procedures). The conductance increase was around 3 nS in 1 M KCl, and the channels were formed preferentially in membranes containing cholesterol, i.e. in the absence of cholesterol, significantly fewer but similar channels were formed (data not shown). Interestingly, two of the truncated forms of LcrV, LcrVΔ316−326 and LcrVΔ306−326, also formed channels with similar properties in the lipid bilayers. Further truncation of the C-terminal part or deletion of a central domain (Δ of amino acids 160–171) of LcrV resulted in loss of channel-forming activity (data not shown).

Figure 7.

Single-channel recordings of the conductance of synthetic DiphPC/PS membranes in the presence of purified proteins.

A. LcrV.

B. PcrV.

C. Hybrid 3.

D. Hybrid 4.

The stepwise increase in conductance reflects single channel-forming events. Hybrids 3 and 4 are shown as representatives of chimeric proteins forming LcrV- and PcrV-like channels respectively. Approximately 1 ng ml−1 protein in 1% Genapol and 1% cholesterol was added to an unbuffered 1 M KCl solution. Applied voltage 20 mV, T = 20°C.

Like LcrV, PcrV was found to induce channels in the lipid bilayers. However, in this case, the properties of the channels were different; the stepwise conductance increase for PcrV was only around 1 nS in 1 M KCl (Fig. 7B), which is in agreement with the smaller channels observed for strains expressing PcrV in the erythrocytes (Fig. 4). Finally, we also purified the different LcrV/PcrV hybrid proteins and added them to the lipid bilayers. In these cases, the hybrid proteins that showed LcrV-like lytic activity (hybrids 1–3) on erythrocytes also formed channels with properties similar to those of purified LcrV (Fig. 7C). The hybrid protein 4 with PcrV-like activity on erythrocytes formed channels with properties similar to those formed by purified PcrV (Fig. 7D). Thus, these results demonstrate that LcrV and PcrV form channels in lipid bilayers and that LcrV and PcrV form channels with different properties.

Discussion

Several animal and plant pathogenic Gram-negative bacteria deliver antihost factors into the eukaryotic host cell via a common contact-dependent secretion/translocation system. Pathogenic Yersinia species constitute the paradigm for this virulence mechanism. Translocation of Yop effectors by Yersinia across the eukaryotic cell membrane requires the secreted proteins LcrV, YopB and YopD (Rosqvist et al., 1994; Sory and Cornelis, 1994; Håkansson et al., 1996; Francis and Wolf-Watz, 1998; Nilles et al., 1998; Fields et al., 1999; Pettersson et al., 1999). Previous work in several laboratories has indicated that translocation is associated with channel formation in the host cell membrane. Therefore, it has been suggested that Yop effector translocation occurs through a channel structure (Håkansson et al., 1996; Neyt and Cornelis, 1999a; Tardy et al., 1999). Here, we present evidence that LcrV is an essential component of channel formation and that LcrV is a size-determining element.

The diameter of the channel was estimated to be about 3 nm by a carbohydrate exclusion assay (Fig. 4). The LcrV homologue PcrV of P. aeruginosa can functionally complement an lcrV mutant with respect to regulation, secretion and translocation of Yop proteins (Pettersson et al., 1999). PcrV was also essential for the lysis of SRBCs, albeit at a lower level compared with LcrV, when expressed in an lcrV mutant background. Interestingly, the PcrV-mediated lysis correlated with the formation of channels with a diameter of about 2 nm (Fig. 4). Thus, the PcrV-associated channel is smaller than that of LcrV. PopB and PopD of P. aeruginosa can complement yopB and yopD mutants of Y. pseudotuberculosis with respect to translocation as well as channel formation. However, in this case, in which PopB and PopD are expressed together with LcrV, the channel size was unaltered (Frithz-Lindsten et al., 1998). These results indicate that LcrV and PcrV are the major size-determining components of the channel complex. This conclusion was reinforced by the finding that purified LcrV and PcrV both induced a stepwise conductance increase in a lipid bilayer model system. Both proteins showed increased channel-forming activity in cholesterol-containing membranes, indicating that these proteins exert their function in host cell membranes. While LcrV induced an increase in conductivity of about 3 nS, PcrV generated channels with a conductivity increase of about 1 nS in the presence of 1 M KCl.

The fact that LcrV and PcrV were associated with channels of different sizes made it possible to identify the domains of LcrV and PcrV determining the size of the channel. A region of LcrV spanning amino acids 127–195 and a corresponding region of PcrV (amino acids 106–173) were identified to be responsible for the difference in size of the channels induced by the two proteins. Computer-aided analysis of these two regions revealed the presence of amphipathic β-sheet conformations showing similarities to those found in other channel-forming proteins such as porins and bacterial toxins (Jap and Walian, 1996). In addition, deletion of amino acids 160–171 resulted in loss of channel formation and Yop translocation, further supporting the importance of this region.

The very C-terminal end of LcrV (316–326) is essential for function, and lcrV mutants lacking this domain are defective in Yop translocation (Pettersson et al., 1999). These mutants failed to induce lysis of RBCs, but here we show that the corresponding purified protein induced channel formation in lipid bilayers. Thus, the C-terminal domain of LcrV is redundant for channel formation in lipid bilayers but is required for the lysis of SRBCs. Therefore, it is likely that this part of LcrV interacts with other components involved in the translocation process. Sarker et al. (1998) showed that LcrV has the potential to interact with both YopB and YopD. It is possible that translocation and RBC lysis involves complex formation between LcrV, YopB and YopD. This complex may be needed for the functional interaction with the cell surface and the formation of the translocation channel in the cell membrane. YopB and YopD possess hydrophobic regions that could interact with biological membranes. In agreement with this idea, Tardy et al. (1999) showed that YopB and YopD were specifically enriched in liposomes. Therefore, it is likely that LcrV, YopB and YopD are all inserted into the eukaryotic cell membrane to form a functional translocation channel. In addition, Tardy et al. (1999) also showed that YopB and YopD are both required for the formation of channels in planar lipid membranes giving a conductance of 0.105 nS in 0.2–1 M NaCl. These authors found no evidence supporting a role for LcrV in channel formation. Our studies revealed that LcrV induces a channel in lipid bilayers with a conductance of 3 nS in 1 M KCl, which is considerably larger compared with that formed by YopB and YopD. It is possible that the LcrV-associated channel was not discovered by Tardy et al. (1999) due to the fact that LcrV was not inserted into the artificial proteoliposomes, and thus LcrV was not included in their assay system.

The result of channel formation in lipid bilayer membranes by LcrV alone seems to represent a contradiction to the need for YopB/D for the transport of effectors in target cells and for haemolysis. However, it has to be kept in mind that many cytolytic bacterial toxins, such as the RTX toxins (Ropele and Menestrina, 1989; Maier et al., 1996), e.g. α-toxin from Staphylococcus aureus (Chakraborty et al., 1990) and aerolysin from Aeromonas sobria (Buckley, 1992; Cabiaux et al., 1997), form channels in lipid bilayer membranes without any need of receptors, whereas they all need a receptor for biological activity, which means that they cannot form channels in these membranes in the absence of a receptor. Similarly, B-domains of toxins also reconstitute in lipid bilayer membranes without receptors and form channels. Lipid bilayers have smooth surfaces without any surface structure including the surface-exposed carbohydrates of biological membranes, which form a thick layer. This means that LcrV can interact with the hydrocarbon core of the lipid bilayers and can insert without the help of receptors and facilitators. In this work, we also attempted to study the effects of YopB/D on the LcrV-induced channels. Owing to the membrane-disruptive activity of purified YopB and YopD, we were unable to achieve conclusive results. As shown earlier for YopB and recently also for the related protein SipB of S. typhimurium, these proteins interact with membranes but do not form channels in lipid bilayer membranes (Håkansson et al., 1996; Hayward et al., 2000). This suggests that YopB and YopD are needed for guiding LcrV into the eukaryotic cell membrane. Alternatively, YopB and YopD could anchor the LcrV to the target cell membrane via the hydrophobic domains of YopB/D.

Based on all these findings, it is likely that LcrV, YopB and YopD together constitute a functional channel complex. However, our results showing that LcrV alone can form a channel in lipid bilayer membranes suggest that LcrV constitutes the backbone of the translocation complex. LcrV is present on the bacterial cell surface before target cell contact has been established. In addition, antibodies directed towards LcrV block translocation of Yop effectors and also provide passive protection against experimental infections, whereas antibodies directed against YopB or YopD show no such effect (Leary et al., 1995; 1999; Pettersson et al., 1999). These results indicate that LcrV is involved in the immediate early steps of Yop effector translocation. It is possible that, once target cell contact has been established, LcrV inserts into the eukaryotic cell membrane to initiate channel formation and that YopB and YopD are then inserted into the membrane to stabilize the channel to form a functional translocation complex.

YopD apparently has multiple roles in Yop translocation. One established function is as a negative regulator of Yop expression, but YopD is also translocated into the eukaryotic cell (Francis and Wolf-Watz, 1998; Williams and Straley, 1998). In addition, we confirm here the findings made by Neyt and Cornelis (1999a) that YopD is required for channel formation. This could mean that YopD stabilizes the translocation complex at the inner surface of the plasma membrane or that YopD has a more direct role in Yop translocation. The latter hypothesis is supported by the observation that YopD can interact with YopE and possibly other Yop effectors (Hartland and Robins-Browne, 1998). We speculate that YopD could have two different roles in the Yop translocation process. One is to position the translocation complex correctly in the eukaryotic membrane, and the other is to act as a guide in the transport of Yop effectors across the membrane of the target cell.

In summary, we have established that there is an absolute correlation between the components involved in Yop effector translocation across the host cell membrane and channel formation in erythrocyte membranes. In addition, the size of the translocation channel correlates with the translocation efficiency, arguing strongly for the importance of channel formation in Yop effector translocation. LcrV has a central role in this process, and we suggest that LcrV forms the core of the membrane channel through which Yop effectors are translocated.

We also used the LcrV/PcrV hybrid proteins to investigate whether the size of the translocation channel influenced the efficiency of Yop effector delivery into the eukaryotic cell. There was a strict correlation between the channel size and the amount of YopE and YopH that was found to be translocated (Table 1; Fig. 6). Thus, we have established that the in vitro properties of the membrane channel also reflect the in vivo translocation competence. These results provide support for the concept that Yop effector delivery into host cells is dependent on channel formation in the eukaryotic cell membrane.

Experimental procedures

Bacterial strains, plasmids, growth conditions and DNA methods

The bacterial strains and plasmids used in this study are listed in Table 2. E. coli strains were grown in LB broth or on LB agar plates. Yersinia strains were grown in LB or brain–heart infusion (BHI; Oxoid). BHI was supplemented with 5 mM EGTA and 20 mM MgCl2 (BHI). For solid media, Yersinia selective agar base (YSA; Difco) and blood agar base (BAB; Oxoid) were used. Antibiotics were used at the following concentrations: kanamycin (50 µg ml−1), chloramphenicol (20 µg ml−1) and carbenicillin/ampicillin (100 µg ml−1). Preparation of plasmid DNA, restriction enzyme digests, ligations and transformations into E. coli were performed essentially as described by Sambrook et al. (1989). Plasmids were introduced into Yersinia by electroporation or conjugation as described by Holmström et al. (1995). DNA fragments were purified from agarose gels using Geneclean (Bio 101) or GenElute (Supelco) according to the manufacturer's instructions.

Table 2a. Bacterial strains and plasmids used in this study.
Bacterial strainRelevant genotypeReference or source
Y. pseudotuberculosis
 YPIII(pIB102)Wild type Bölin and Wolf-Watz. (1984)
 YPIII(pIB155) yopK Holmström et al. (1997)
 YPIII(pIB15519) yopK, lcrVThis study
 YPIII(pIB19) lcrV (Δaa10–313) Pettersson et al. (1999)
 YPIII(pIB1926) lcrV, lcrQThis study
 YPIII(pIB26) lcrQ Pettersson et al. (1996)
 YPIII(pIB604) yopB Håkansson et al. (1996)
 YPIII(pIB621) yopD Francis and Wolf-Watz (1998)
E. coli
 S17-1λpirRP4::2-Tc::Mu::Km Tn7, λpir phage lysogen de Lorenzo and Timmis (1994)

Construction of mutants and plasmids

The double yopK, lcrV in frame deletion mutant strain was constructed by the same method used to make the lcrV and lcrV, yopN mutants (Pettersson et al., 1999). In short, E. coli S17-1 λpir containing the lcrV mutator plasmid pAH70 (Pettersson et al., 1999) was conjugated with the Yersinia strain YPIII(pIB155), yopK (Holmström et al., 1997), and two successive recombination events were selected for. The yopK, lcrV mutant strain was denoted YPIII(pIB15519).

The lcrQ, lcrV mutant was constructed by introducing an lcrQ deletion construct (Pettersson et al., 1996) into the lcrQ locus of Yersinia strain YPIII(pIB19), lcrV. The yopQ, lcrV mutant strain was denoted pIB1926.

Construction of plasmids expressing lcrVpcrV hybrid proteins

Hybrid clones containing parts of lcrV and pcrV were constructed as follows. The lcrV insert of plasmid pTB7 (Bergman et al., 1991) was excised by digestion of this plasmid with XbaI and SacI. lcrV-containing fragments were amplified by polymerase chain reaction (PCR) with YPIII(pIB102) as the template DNA, and pcrV-containing fragments were amplified with pMS9 (Frithz-Lindsten et al., 1998) as the template DNA (Table 3). lcrVpcrV and pcrVlcrV hybrid genes were generated using an overlapping fragment PCR method analogous to that described in detail by Holmström et al. (1997). Fragments contained XbaI and SacI sites to allow ligation back into the vector. The plasmids (constructs shown schematically in Fig. 5) were then transformed into E. coli S17-1 λpir, from which they were introduced by conjugation into the recipient Y. pseudotuberculosis strains YPIII (pIB19), lcrV, and YPIII (pIB15519), yopK, lcrV.

Table 2b. 
PlasmidDescriptionReference or source
pAH70ΔlcrV amino acids 10–313 mutator plasmid containing a 582 bp PCR fragment cloned in pDM4 (Milton et al., 1996) Pettersson et al. (1999)
pHyb 1–4 lcrVpcrV hybrids cloned under the control of the tac promoter of pMMB66HEThis study (see Experimental
procedures
for details)
pLJ33 pcrV cloned under the control of the tac promoter of pMMB66EH Pettersson et al. (1999)
pMS9The pcrGVHpopBD operon cloned in pUC19 Frithz-Lindsten et al. (1998)
pRN53ΔlcrQ mutator plasmid containing a Sp/Sm resistance gene cloned in pDM4 (Milton et al., 1996) Pettersson et al. (1996)
pTB7 lcrV cloned under the control of the tac promoter of pMMB66HE Bergman et al. (1991)
pV316pTB7 derivative, lcrV truncated at codon 316 Pettersson et al. (1999)
pV306pTB7 derivative, lcrV truncated at codon 306 Pettersson et al. (1999)
pVΔ160–171pTB7 derivative, amino acids 160–171 of lcrV deleted Pettersson et al. (1999)

Yop expression and analysis

Overnight cultures of Yersinia strains grown at 26°C were diluted (1:10) in fresh media (BHI) containing 0.1% Triton X-100, grown for 1 h at 26°C, then shifted to 37°C and grown for an additional 3 h. After measuring OD600, the cells were harvested, and the culture supernatant was collected and filtered (0.45 µm; Sartorius). The proteins from the supernatant were precipitated with trichloroacetic acid as described earlier (Forsberg et al., 1987). The samples from the supernatants were dissolved in SDS sample buffer, and the volume of the samples was adjusted in accordance with the OD600 values of the bacterial cultures. The proteins were separated by SDS–PAGE and visualized by staining with Coomassie brilliant blue or Western blotting using a rabbit antiserum recognizing LcrV or PcrV respectively.

Infection of HeLa cells and assay for YopE-mediated cytotoxicity

The cultivation and infection of HeLa cells has been described in detail previously (Rosqvist et al., 1990). Briefly, HeLa cells were grown as monolayers to semi-confluence in six- or 24-well tissue culture plates in RPMI medium with 10% heat-inactivated fetal calf serum (FCS) and 3 µg ml−1 gentamicin at 37°C in a humidified atmosphere. Before infection, the HeLa cells were washed free of gentamicin, and fresh medium was added (supplemented with 1 mM IPTG when suitable). Yersinia strains were cultivated in LB at 26°C overnight. The next day, the bacteria were diluted (1:300) into fresh RPMI medium. After growth for 30 min at 26°C on a rotary shaker, the bacteria were incubated for 1 h at 37°C. After the addition of bacteria, the infected HeLa cells were centrifuged for 5 min at 400 g to facilitate contact between the bacteria and the HeLa cells. Thereafter, the tissue culture plates were incubated at 37°C in a humidified atmosphere for 5 h, and the cells were examined every hour to detect changes in morphology.

Cultivation and infection of HeLa cells

The Y. pseudotuberculosis strains used to infect HeLa cells were grown overnight at 26°C in LB. Samples (10 µl) of the overnight cultures were added to 3 ml of modified Eagle medium (MEM) with 10% heat-inactivated FCS without any antibiotics. The inoculated RPMI cultures were grown for 30 min at 26°C followed by incubation at 37°C for 1 h before infection. IPTG was added to a final concentration of 1 mM at the time of the temperature shift. The cultivation and infection of HeLa cells has been described in detail elsewhere (Rosqvist et al., 1990). The HeLa cells were seeded (1.0 × 105 well−1) in a 24-well tissue culture plate in RPMI with 10% heat-inactivated FCS and 100 IU ml−1 gentamicin at 37°C in a 5% CO2 humidified atmosphere. For immunofluorescence studies, the HeLa cells were grown on 12 mm coverslips placed in a 24-well culture plate. Before infection, the HeLa cells were washed extensively with RPMI and 10% heat-inactivated FCS without any antibiotics was added. After infection (100 µl of induced culture), the HeLa cells were centrifuged for 5 min at 400 g to facilitate contact between the bacteria and the HeLa cells, followed by continued incubation at 37°C.

Cytotoxicity and immunofluorescence staining

Translocation of YopE was monitored as cytotoxicity, in which the infected cells were observed every hour by phase-contrast microscopy. A changed morphology, as visualized by rounding up of the HeLa cells, indicated a cytotoxic response. For immunofluorescence staining, the cell monolayers were washed twice in PBS 3 h after infection and stained with wheat germ agglutinin (WGA) conjugated to Texas red (Molecular Probes) (10 µg ml−1 for 10 min at 24°C); the WGA-stained cells were fixed in 2% paraformaldehyde for 10 min. Unless indicated, the cells were then permeabilized with 0.5% Triton X-100 in a buffer containing 1 mM EGTA, 4% polyethylene glycol 6000 and 100 mM piperazine-N,N′-bis(2-ethanesulphonic acid), pH 6.9, and processed further for indirect immunofluorescence labelling (for details, see Rosqvist et al., 1991) using affinity-purified goat anti-YopH antibodies, followed by fluorescein isothiocyanate (FITC)-conjugated anti-goat antibodies. The specimens were finally mounted in a mounting medium containing Citifluor as an antifading agent. The specimens were analysed using fluorescence microscopy (Zeiss Axioscope) equipped with a CCD camera (Hamamatsu).

Assay for haemolysis

The contact haemolytic assay was performed as follows. Overnight bacterial cultures were diluted to an OD600 of 0.2 in fresh culture medium (BHI). Bacteria were grown for 30 min at 26°C followed by incubation at 37°C for an additional 2 h. To the appropriate strains, IPTG was added to a final concentration of 1 mM at the time of the temperature shift. Sheep erythrocytes (SVA, Sweden) were washed twice with PBS and adjusted to 4 × 109 cells ml−1 with BHI. In large Eppendorf tubes, 200 µl of the erythrocyte suspension was mixed with 200 µl of bacterial culture adjusted to an OD600 of 0.7. Close contact between erythrocytes and bacteria was achieved by centrifugation at 855 g for 5 min at room temperature. After 3 h of incubation at 37°C, the pellets were resuspended in 600 µl of PBS, and the samples were centrifuged at 13 000 g for 5 min. The supernatants (100 µl) were transferred to a microtitre plate (F96 Maxisorp Immunoplate; Nunc), with each sample in quadruplicate. The release of haemoglobin was measured as the OD545 using an iEMS reader MF (Labsystems) and expressed as lytic activity × 10−3. Contact haemolytic assays were performed in the presence of carbohydrates essentially as described by Håkansson et al. (1996). The washed erythrocytes were resuspended in culture medium containing 60 mM of the respective carbohydrate to yield a final concentration of 30 mM in the assay tube, and PBS added after the second centrifugation contained 30 mM of the carbohydrate of interest. The percentage lytic activity was expressed as the lytic activity in the presence of carbohydrate divided by the lytic activity when no carbohydrates were present.

Protein purification

Protein secretion in BHI medium was induced as described above. Secreted proteins were precipitated from the cleared supernatant by the addition of 50% (v/v) ice-cold acetone. The resulting pellet was dissolved in SDS sample buffer without reducing agent, and proteins were separated by SDS–PAGE. The gel was stained using 0.3 M CuCl2, and the band of interest was excised and destained in 0.25 M Tris, 0.25 M EDTA, pH 9. The protein was electroeluted, precipitated with acetone (50% volume) and resuspended in PBS. When needed, the protein was run using the same procedure for a second time. YopB and YopD were prepared from YPIII(pIB 8219) yopN, lcrV. LcrV, PcrV and the hybrids thereof were prepared from YPIII(pIB19) lcrV, transcomplemented with the different LcrV/PcrV species. Purity was assayed by Western blot with antibodies against YopB, YopD, YopN, LcrV and PcrV, and none of these was found to contaminate the others.

In addition, LcrV, PcrV, V316 and V306 were also purified using an immunoaffinity method. Polyclonal anti-LcrV and anti-PcrV sera from rabbit were coupled to CNBr-activated Sepharose 4B (Pharmacia) according to the manufacturer's instructions, and 5 ml of affinity gel was packed in columns. Overnight cultures of E. coli strains harbouring pTB7, pLJ33, pV316 and pV306 were diluted to an OD600 of about 0.2, and protein expression was induced by adding IPTG to a final concentration of 1 mM when the bacterial culture had reached an OD600 of about 0.5. After 5–6 h, the cultures (0.5 l) were centrifuged, and the pellet was dissolved in 20 ml of PBS and sonicated for 3–5 min. The lysate was centrifuged, and the supernatant was run through the antibody column. The column was washed in succession with 10 volumes of PBS, 10 volumes of PBS containing 1 M NaCl and 10 volumes of PBS. Proteins were eluted by the addition of 3 M KCSN in aliquots of 0.5 ml and analysed by SDS–PAGE and Coomassie brilliant blue staining. Fractions containing eluted protein were pooled and dialysed against 1:10 PBS at 4°C for about 8 h.

Lipid bilayer experiments

The method used for the lipid bilayer experiments has been described in detail previously (Benz and Chakraborty, 1992). Black lipid bilayer membranes were built up by a 1% (w/v) solution of diphytanoyl phosphatidylcholine (PC) and phosphatidylserine (PS) (Avanti Polar Lipids) in a 4:1 ratio dissolved in n-decane + n-butanol. Protein samples were supplied with Genapol and cholesterol (Sigma). The temperature was maintained at 20°C during the experiments. The aqueous salt solution consisted of 1 M KCl, unbuffered (Merck). The single channel records were performed using Ag/AgCl electrodes with salt bridges connected in series to a voltage source and a current amplifier. The amplified signal was monitored with a storage oscilloscope (Tektronix 7633) and recorded on a strip chart.

Acknowledgements

This work was supported by grants from the Swedish Medical Research Council (B94-16X-07490-09B, B95-16X-11221-01A and project number 6251), the Swedish Natural Science Research Council (B-BU 04426-305 and B-AA/BU-06353-301), Swedish Society of Medicine, Magnus Bergvall Foundation, King Gustaf V 80 Year Foundation, the J. C. Kempe Memorial Foundation, the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen Industrie. Jeanette Bröms, Kerstin Kuoppa, Marlene Lundström and Lenore Johansson are greatly acknowledged for skilful technical assistance.

Ancillary