Heparin interferes with translocation of Yop proteins into HeLa cells and binds to LcrG, a regulatory component of the Yersinia Yop apparatus

Authors

  • Aoife P. Boyd,

    1. Microbial Pathogenesis Unit, International Institute of Cellular and Molecular 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
  • Marie-Paule Sory,

    1. Microbial Pathogenesis Unit, International Institute of Cellular and Molecular 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
  • Maite Iriarte,

    1. Microbial Pathogenesis Unit, International Institute of Cellular and Molecular 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, International Institute of Cellular and Molecular 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, E-mail cornelis@mipa.ucl.ac.be; Tel. (2) 764 7449; Fax (2) 764 7498.

Abstract

Yersiniae are equipped with the Yop virulon, an apparatus that allows extracellular bacteria to deliver toxic Yop proteins inside the host cell cytosol in order to sabotage the communication networks of the host cell or even to cause cell death. LcrG is a component of the Yop virulon involved in the regulation of secretion of the Yops. In this paper, we show that LcrG can bind HeLa cells, and we analyse the role of proteoglycans in this phenomenon. Treatment of the HeLa cells with heparinase I, but not chondroitinase ABC, led to inhibition of binding. Competition assays indicated that heparin and dextran sulphate strongly inhibited binding, but that other glycosaminoglycans did not. This demonstrated that binding of HeLa cells to purified LcrG is caused by heparan sulphate proteoglycans. LcrG could bind directly to heparin-agarose beads and, in agreement with these results, analysis of the protein sequence of Yersinia enterocolitica LcrG revealed heparin-binding motifs. In vitro production and secretion by Y. enterocolitica of the Yops was unaffected by the addition of heparin. However, the addition of exogenous heparin decreased the level of YopE–Cya translocation into HeLa cells. A similar decrease was seen with dextran sulphate, whereas the other glycosaminoglycans tested had no significant effect. Translocation was also decreased by treatment of HeLa cells with heparinitase, but not with chondroitinase. Thus, heparan sulphate proteoglycans have an important role to play in translocation. The interaction between LcrG and heparan sulphate anchored at the surface of HeLa cells could be a signal triggering deployment of the Yop translocation machinery. This is the first report of a eukaryotic receptor interacting with the type III secretion and associated translocation machinery of Yersinia or of other bacteria.

Introduction

The genus Yersinia includes three species that are pathogenic for rodents and humans: Y. pestis, the agent of the plague, Y. pseudotuberculosis and Y. enterocolitica. Y. enterocolitica is a common human pathogen, which causes gastrointestinal diseases of various severity, ranging from diarrhoea to mesenteric adenitis. To survive in the host tissues, Yersinia are endowed with a sophisticated, plasmid-encoded, virulence apparatus called the Yop virulon. This apparatus allows extracellular bacteria to deliver toxic Yop proteins inside the host cell cytosol in order to sabotage their communication networks or to kill them (for review, see Cornelis and Wolf-Watz, 1997). It is composed of four elements: (i) a contact or type III secretion system called Ysc, devoted to the secretion of Yop proteins outside the bacterium (for review, see Cornelis, 1994); (ii) a system consisting of YopB and YopD, which is designed to deliver the toxic Yop proteins inside eukaryotic target cells (Rosqvist et al., 1994; Sory and Cornelis, 1994; Persson et al., 1995; Sory et al., 1995; Boland et al., 1996; Håkansson et al., 1996a); (iii) a recognition and control complex consisting of YopN and LcrG (Forsberg et al., 1991; Skrzypek and Straley, 1993; Rosqvist et al., 1994; Boland et al., 1996); and (iv) a set of toxic effector Yop proteins, namely YopE, YopH, YopM, YpkA/YopO and YopP (Rosqvist et al., 1994; Sory and Cornelis, 1994; Persson et al., 1995; Sory et al., 1995; Boland et al., 1996; Håkansson et al., 1996b; M.-P. Sory, unpublished). YopH is a 51 kDa broad-spectrum protein tyrosine phosphatase (Guan and Dixon, 1990), and YopO/YpkA is an 81 kDa serine/threonine kinase (Galyov et al., 1993; Håkansson et al., 1996b). The enzymatic properties of YopE, YopM and YopP are not known.

Delivery of Yop effectors by infecting extracellular Yersinia was first demonstrated with human cervical epithelial cell lines (HeLa) (Rosqvist et al., 1994; Sory and Cornelis, 1994). After delivery, YopE causes disruption of the actin microfilament structure of HeLa cells (Rosqvist et al., 1991), while YopH dephosphorylates the focal adhesion kinase and p130Cas (Black and Bliska, 1997; Persson et al., 1997). On human intestinal epithelial cell lines (T84), the Yop virulon reduces the release of interleukin 8 (IL-8) induced by the Yersinia infection (Schulte et al., 1996). Yop effectors are also delivered into macrophage cell lines (Sory et al., 1995), where they inhibit phagocytosis (via the action of YopH; Rosqvist et al., 1988), the oxidative burst (Hartland et al., 1994; Bliska and Black, 1995), the release of tumour necrosis factor alpha (Nakajima and Brubaker, 1993; Ruckdeschel et al., 1997) and even induce apoptosis, a phenomenon that requires YopP (Mills et al., 1997). Yop effectors can also be delivered inside other cell types, such as fibroblasts (M.-P. Sory, unpublished). During the animal infection, macrophages seem to be the most relevant target for the delivery of Yops, but other cell types could also represent important targets.

In vitro, Yop translocators and effectors are specifically released, either by chelating Ca2+ ions or upon contact between bacteria and eukaryotic cells (Rosqvist et al., 1994; Sory and Cornelis, 1994). In addition, in the presence of eukaryotic cells, the effector Yop proteins are preferentially delivered inside the eukaryotic cells rather than released into the surrounding media. Therefore, the bacteria must be able to sense the eukaryotic cell and to react to its presence. This recognition presumably involves both a bacterial ligand and a eukaryotic receptor. The interactive partner on the bacterium could be the 35 kDa YopN or the 11 kDa LcrG, based on the fact that yopN and lcrG mutants are deregulated for Yop secretion in the presence of Ca2+ as well as in the absence of eukaryotic cells (Forsberg et al., 1991; Skrzypek and Straley, 1993; Rosqvist et al., 1994; Persson et al., 1995; Boland et al., 1996). One could hypothesize that LcrG and YopN and perhaps other Yop virulon proteins could form a complex stop-valve system controlling the release and translocation of the effectors. We set out in this work to study the interaction of these proteins with eukaryotic cells.

Results

Binding of LcrG to HeLa cells

We first tested the possibility that LcrG and/or YopN could bind to eukaryotic cells. To do this, LcrG and YopN were purified as glutathione S-transferase fusion proteins (GST–LcrG and GST–YopN respectively). The GST–YopN preparation was very unstable and so unfortunately could not be tested. The GST–LcrG preparation was used to coat microtitre plates. Increasing amounts of GST–LcrG were then incubated with HeLa cells, and the presence of bound cells was determined by assaying hexosaminidase, a constitutive lysosomal protein. HeLa cells were able to bind to wells coated with GST–LcrG but not with GST alone. Binding to GST–LcrG was concentration dependent and saturable (Fig. 1A), indicating that the LcrG is specifically able to interact with eukaryotic cells.

Figure 1.

. Binding of HeLa cells to LcrG. GST–LcrG, GST or LcrG were bound to 96-well plates at the concentrations shown. HeLa cells were incubated with immobilized protein for 1 h at 37°C, and the hexosaminidase assay was performed to determine the level of HeLa cell binding (expressed as OD405). A. GST and GST–LcrG were bound on the plates and 7 × 104 HeLa cells were added to each well. B. LcrG was bound on the plates and 10 × 104 HeLa cells were added to each well. These are the results of triplicate experiments ± SD.

To investigate the nature of the interaction between LcrG and HeLa cells further, we purified the GST–LcrG fusion protein on glutathione beads and then cleaved this with thrombin protease to release the LcrG protein with an additional 13 amino acids at the N-terminal (GSPGISGGGGGIH). This protein was more than 95% pure as judged by SDS–PAGE (see Fig. 6). This LcrG was coated on plastic, incubated with HeLa cells, and binding was quantified by the hexosaminidase assay. As shown in 1Fig. 1B, purified LcrG bound HeLa cells in a saturable manner, thus confirming the results obtained with GST–LcrG. Hence, we used this protein preparation for the further experiments described in this study.

Figure 6.

. Dimerization of LcrG. LcrG was incubated for 90 min in Tris buffer and 1 mM PMSF. Samples were electrophoresed on a SDS–PAGE tricine gel in reducing or non-reducing conditions. Molecular weight standard sizes in kDa are indicated on the lefthand side. Lane 1, LcrG-reducing conditions; lane 2, LcrG in non-reducing conditions.

Inhibition of binding by modification of HeLa cells

We then studied the nature of the receptor for LcrG on HeLa cells, using the same binding assay and hexosaminidase detection. First, we pretreated the HeLa cells with proteases to see if a protein on the HeLa cells was responsible for the interaction with LcrG. As can be seen in Fig. 2, treatment of HeLa cells with proteinase K, trypsin or chymotrypsin all decreased the binding of HeLa cells to LcrG. Thus, a cell surface eukaryotic protein is involved in binding. As Y. enterocolitica is able to translocate YopE efficiently into a variety of mammalian cell lines, we hypothesized that the eukaryotic receptor responsible for interaction with the translocation machinery is present ubiquitously on mammalian cells. Proteoglycans, surface proteins to which glycosaminoglycans (GAGs) are attached, are found on practically all types of eukaryotic cells, and they have been shown to be receptors on eukaryotic cells for a variety of microorganisms and cellular factors via their GAGs. We thus tested the possibility that proteoglycans were responsible for the binding of HeLa cells to LcrG. We first pretreated the HeLa cells with xyloside, which inhibits the attachment of the GAGs to the core proteins. Binding of HeLa cells to LcrG was inhibited by more than 90% by this treatment, indicating that LcrG indeed binds to the proteoglycans on the surface of HeLa cells (Fig. 2). To confirm this, HeLa cells were then treated with two GAG lyases that remove the GAGs from the surface of cells. Treatment with chondroitinase ABC, which removes chondroitin sulphate (CS) A, B and C, did not affect the binding of HeLa cells to LcrG (Fig. 2). However, treatment with heparinase I did inhibit binding. Thus, we concluded that heparan sulphate proteoglycan acts as a receptor for LcrG on HeLa cells.

Figure 2.

. Effect of HeLa cell modification on the binding to LcrG. Purified LcrG was bound to 96-well plates at a concentration of 12 μg ml−1. HeLa cells were treated as described, incubated with LcrG for 1 h at 37°C, and the hexosaminidase assay was performed to determine the degree of binding. Percentage binding is relative to control conditions, which were carried out without modification of the HeLa cells. These are the results of triplicate experiments ± SD.

Competitive inhibition of binding by GAGs

To confirm the binding of proteoglycans, we tested the ability of various exogenously added GAGs to inhibit the binding of HeLa cells to LcrG (Fig. 3). Heparin (a GAG with the same backbone structure as heparan sulphate but with a higher degree of sulphation) virtually eliminated the binding of HeLa cells to LcrG, while CSA and CSC did not significantly affect binding. CSB (dermatan sulphate) did decrease binding, but just to 40% of the binding observed in control conditions lacking GAG. The presence of iduronic acid in CSB and its absence in CSA and CSC may explain the differences observed with the CS species. Two other GAGs — keratan sulphate and hyaluronic acid — had no effect on binding. Dextran sulphate, an artificial substrate consisting of a high-molecular-weight glucose polymer with a large percentage of sulphate substitutions, strongly inhibited binding. Dextran, which lacks sulphate groups, did not inhibit binding, showing the importance of the sulphate groups on the GAGs for recognition by LcrG. It has been shown that the sulphate groups on the GAGs are responsible for binding with the ligands through electrostatic interactions, and that the pattern of sulphate modification is important for the specificity of binding too (Lindahl et al., 1984). This latter point was confirmed by using de-N-sulphated heparin, which did not inhibit the binding of HeLa cells to LcrG, in contrast to heparin which did (Fig. 3). These results showed that heparin can specifically inhibit the interaction of HeLa cells with LcrG. Dextran sulphate could also inhibit, presumably because of its very large sulphate charge. The N-sulphate group on heparin is particularly important for the inhibition effect, as presumably are iduronic acid residues. So, heparan sulphate with iduronic acid and N-sulphate is the preferred GAG substrate for LcrG.

Figure 3.

. Effect of exogenously added GAGs on binding of HeLa cells to LcrG. Purified LcrG was bound to 96-well plates at a concentration of 12 μg ml−1. GAGs were added at 30 μg ml−1 to the binding buffer and allowed to incubate for 1 h before addition of HeLa cells. After incubation of the HeLa cells with LcrG for 1 h at 37°C, the hexosaminidase assay was performed to determine the degree of HeLa cell binding. Percentage binding is relative to control conditions, which were carried out in the absence of GAGs. These are the results of triplicate experiments ± SD.

To look further at the relative abilities of these GAGs to affect the binding of HeLa cells to LcrG, inhibition curves were carried out with heparin, dextran sulphate, CSB and CSC. As shown in Fig. 4, heparin was the most effective inhibitor with a 50% inhibition concentration (IC50) of 0.3 μg ml−1. Dextran sulphate differed only slightly with an IC50 of 0.5 μg ml−1. The IC50 for CSB was 20 μg ml−1, showing that it has a much weaker affinity for LcrG than heparin or dextran sulphate and presumably plays a minor role in binding LcrG on the cell surface. CSC did not inhibit at any concentration. Taken together, the competition and HeLa cell modification experiments suggest that heparan sulphate on HeLa cells acts as a receptor for LcrG.

Figure 4.

. Competitive inhibition of HeLa cell binding to LcrG by heparin, dextran sulphate and CSB. The experiment was carried out as described in Fig. 3 using decreasing concentrations of GAGs. Percentage binding is relative to control conditions, which were carried out in the absence of GAGs. These are the results of triplicate experiments ± SD.

Direct binding of LcrG to heparin. Dimerization of LcrG

We examined whether LcrG could bind heparin directly by testing the ability of purified LcrG to bind to heparin conjugated to agarose beads. The purified LcrG was incubated with heparin beads and then eluted in batches with increasing concentrations of NaCl. A control experiment using LcrV in place of LcrG was also performed. LcrV is a Yop virulon protein produced by a gene in the same operon as lcrG. The results are shown in Fig. 5. The control protein, LcrV, was recovered in the wash fraction only. In contrast, only a small fraction of LcrG was found in the unbound fraction. LcrG remained on the beads and was eluted preferentially at between 250 and 1000 mM NaCl. Thus, LcrG can bind directly to heparin. Polymerization of heparin ligands by heparin has been shown previously and can lead to changes in the functional activities of these ligands (Sakata et al., 1997; Seiffert, 1997). So, we tested if LcrG could polymerize and whether heparin could induce such LcrG polymerization. When the LcrG protein samples were electrophoresed in reducing conditions, a single band of approximately 12–14 kDa was seen, corresponding to an LcrG monomer (Fig. 6). In non-reducing conditions, an additional protein band could be seen at 25–30 kDa, corresponding in size to an LcrG dimer. As this dimer could be found only in non-reducing conditions, disulphide bond interactions must be responsible for the formation of the dimer. Only one cysteine residue is present in LcrG at amino acid 34 (Fig. 7), and this must be the residue responsible for the disulphide bond interactions. No major differences were seen in the presence or absence of heparin. So, even though LcrG can form dimers, the formation of these dimers was not augmented by incubation with heparin.

Figure 5.

. Direct binding of LcrG to heparin. LcrG and LcrV were incubated with heparin-agarose beads. After washing, batch elutions with increasing concentrations of NaCl were performed. The eluted proteins were electrophoresed on a SDS–PAGE tricine gel. A. LcrG. B. LcrV. Molecular weight standard sizes in kDa are shown on the lefthand side. Lane 1, unbound fraction; lane 2, 62 mM NaCl elution; lane 3, 125 mM NaCl elution; lane 4, 250 mM NaCl elution; lane 5, 500 mM NaCl elution; lane 6, 1 M NaCl elution; lane 7, 2 M NaCl elution; lane 8, fraction that remains on beads after 2 M NaCl elution. The LcrG and LcrV bands are indicated with arrows.

Figure 7.

. Protein sequence of LcrG from Y. enterocolitica. The comparison of the Y. enterocolitica and the Y. pseudotuberculosis /Y. pestis LcrG sequences is shown. The residues of the LcrG sequence of Y. pseudotuberculosis /Y. pestis that are identical to those of the Y. enterocolitica protein are linked by a line. The asterisk indicates the sole cysteine residue in LcrG. The crosses above the sequence show the basic residues of LcrG. In bold are the basic amino acids with similarity to the heparin-binding motifs that are shown below the LcrG sequences.

Protein sequence analysis of LcrG

We analysed the sequence of LcrG to see if it showed similarity to amino acid motifs that have been identified as binding heparin. The sequences for lcrG from Y. pseudotuberculosis and Y. pestis were already available (Price et al., 1989; Bergman et al., 1991) and are identical at the amino acid level. Sequencing of the lcrG gene from Y. enterocolitica gave rise to the derived amino acid sequence shown in Fig. 7. The LcrG sequence from Y. enterocolitica differed from that of Y. pseudotuberculosis and Y. pestis at three positions. These are L18→M, Y79→N and R85→K. Analysis of the Y. enterocolitica sequence shows that the C-terminal part of the protein is rich in basic residues (B). Basic residues are implicated in heparin binding in a number of proteins, including fibronectin and the circumsporozoite protein of malaria (Barkalow and Schwarzbauer, 1991; Sinnis et al., 1994). Closer inspection shows that LcrG contains two sequences with similarity to heparin-binding motifs (Cardin and Weintraub, 1989; Margalit et al., 1993). Both these motifs are characteristic of protein segments with an alpha-helical structure. The first motif is XBBXBX (where B is a basic residue), which is present in the sequence IKRQRE (amino acids 67–72) and in GKRPKK (amino acids 81–86) (GKRPRK in Y. pseudotuberculosis/Y. pestis). The second motif is BX12B, which is seen in the sequence RQRERQPQHPNDGK (amino acids 69–82) (RQRERQPQHPYDGK in Y. pseudotuberculosis/Y. pestis) that links the first two motifs. Thus, the LcrG protein shares the structural characteristics of heparin-binding proteins.

Effect of heparin on Yop secretion in brain–heart infusion (BHI)

In vitro, lcrG mutant bacteria have a deregulated Yop secretion phenotype, in that they secrete Yops in the presence as well as in the absence of Ca2+ (Skrzypek and Straley, 1993). We tested the possibility that the LcrG–heparin interaction could have the same effect as Ca2+ chelation and allow Yop secretion even in the presence of Ca2+. Addition of heparin did not modify the SDS–PAGE profile of Yops secreted by Yersinia grown at 37°C: all Yops were secreted in BHI-OX and none in BHI-Ca2+ (data not shown). So, the interaction of LcrG and heparin does not interfere with the Ca2+ control of Yop secretion. As the relative quantities of secreted Yops were similar in the presence and absence of heparin, we can also say that production of the Yops is unaffected by heparin.

Inhibition of translocation of YopE–Cya into HeLa cells by GAGs

The interaction of LcrG with heparin could be a necessary step in the control of Yop translocation upon contact with eukaryotic cells. To test this, we used the Yop–adenylate cyclase methodology (Sory and Cornelis, 1994). We infected HeLa cells with Y. enterocolitica E40(pMS111) producing YopE130–Cya, and we monitored the cAMP accumulation in infected HeLa cells. As expected from our previous studies, YopE–Cya was delivered efficiently into HeLa cells in the absence of exogenously added heparin (Fig. 8). The addition of heparin to this assay significantly decreased the translocation of YopE–Cya. Dextran sulphate also decreased translocation but dextran did not, in agreement with the LcrG–HeLa cell binding data. None of the other glycosaminoglycans tested significantly inhibited YopE–Cya translocation. We also performed the translocation assays in the presence of various concentrations of heparin. As can be seen in Fig. 9, maximal inhibition was reached at 1 mg ml−1. At this concentration, translocation was inhibited by 75%. Addition of exogenous heparin did not influence Yersinia–host cell binding, presumably because of the high efficiency of the Yersinia adhesins, YadA and Inv (data not shown). These results indicate that heparin interferes with the translocation process. We conclude that heparan sulphate proteoglycans present on the surface of HeLa cells trigger Yop translocation, presumably because of their interaction with LcrG.

Figure 8.

. Effect of exogenously added GAGs on the translocation of YopE–Cya by Y. enterocolitica. HeLa cells were preincubated with GAGs at 100 μg ml−1 for 30 min in RPMI-1640, 2 mM L-glutamine. E40(pMS111) was preincubated with GAGs at 100 μg ml−1 for 15 min at 37°C in RPMI-1640, 2 mM L-glutamine. Bacteria were inoculated onto HeLa cell monolayers and the translocation assay performed. Control conditions were carried out in the absence of GAGs. These are the results of triplicate experiments ± SD.

Figure 9.

. Competitive inhibition of YopE–Cya translocation by heparin. The experiment was carried out as described in the legend to Fig. 8 using the concentrations of heparin indicated. These are the results of triplicate experiments ± SD.

Effect of GAG removal on the translocation of YopE–Cya

To confirm the implication of heparan sulphate in the process of Yop translocation, we treated the HeLa cells with GAG lyases before performing the translocation assay. Treatment with chondroitinase ABC did not affect translocation of YopE–Cya (Fig. 10). On the other hand, treatment with heparinase III did decrease translocation. Adherence of Y. enterocolitica strain E40(pMS111) to HeLa cells was unaffected by heparinitase treatment of the HeLa cells (data not shown). This confirms the importance of heparan sulphate specifically for translocation into HeLa cells.

Figure 10.

. Inhibition of translocation of YopE–Cya by heparinitase. HeLa cells were pretreated with heparinitase or chondroitinase ABC as indicated before the translocation assay. Percentage translocation is relative to control conditions, which were carried out without the addition of GAG lyases. These are the results of triplicate experiments ± SD.

Discussion

We have shown here that LcrG binds HeLa cells by interacting with heparan sulphate proteoglycans. Proteoglycans consist of a core protein to which are covalently attached glycosaminoglycans (GAGs) — sulphated sugar chains (for reviews, see Hardingham and Fosang, 1992; David et al., 1995). Heparan sulphate is structurally heterogeneous depending on the degree of sulphation, epimerization and acetylation. A total of 24 different disaccharide combinations are possible, which result in GAGs of different properties and with different functions. The varying structures allow binding of ligands to specific GAGs on particular cells. We did not identify a particular heparin structure that can bind LcrG, but we showed that N-sulphate groups, and perhaps iduronic acid residues, are important. Proteoglycans can be found on the cell surface of virtually all eukaryotic cells and also in the extracellular matrix. They are concentrated in focal adhesions and participate in cellular adhesion and intracellular signalling. As they are highly charged, the GAGs interact electrostatically with soluble ligands and matrix proteins, including enzymes, cell adhesion molecules, growth factors, proteinase inhibitors and extracellular matrix components. Proteoglycans have been found to be receptors for a number of different pathogens, including viruses, parasites and bacteria. Heparan sulphate-binding proteins include circumsporozoite protein (CS) of Plasmodium falciparum (Pancake et al., 1992; Frevert et al., 1993), the trypanosome adhesin, penetrin (Ortega-Barria and Pereira, 1991), gp120 of human immunodeficiency virus (HIV; Roderiquez et al., 1995) and gB and gC of herpes simplex virus (HSV; Herold et al., 1991). Heparan sulphate is also known to bind to a number of proteins exposed at the surface of bacterial pathogens, such as the filamentous haemagglutinin (FHA; Menozzi et al., 1991) and the fimbrial protein, Fim2 (Geuijen et al., 1996), of Bordetella pertussis, OpaA of Neisseria gonorrhoeae (Chen et al., 1995; van Putten and Paul, 1995), the major outer membrane protein (MOMP) of Chlamydia trachomatis (Su et al., 1996), ActA of Listeria monocytogenes (Alvarez-Domínguez et al., 1997) and the heparin binding haemagglutinin (HBHA) of Mycobacteria (Menozzi et al., 1996). Such interactions between proteoglycans and bacterial ligands have been shown to occur on various cell types, such as epithelial cells, human umbilical vein epithelial cells (HUVECs), hepatocytes and macrophages (for review, see Rostand and Esko, 1997).

In all the examples listed above, the interaction between GAGs and the ligand leads to adherence and, in some instances, ultimately to invasion. However, in the present case, binding triggers a bacterial action, namely translocation of Yop effectors inside the target HeLa cells. This is the first example of a functionally relevant interaction between a eukaryotic cell and the Yop virulon. This suggests that binding of Yersinia to cells via the adhesins YadA, Inv, Myf and Ail (Isberg and Falkow, 1985; Miller and Falkow, 1988; Iriarte et al., 1993; Yang and Isberg, 1993) might not be sufficient to trigger translocation.

The observation that binding of LcrG to heparan sulphate proteoglycans is important for translocation would suggest that LcrG has a critical role to play in translocation. We are at present investigating the importance of LcrG in the translocation of each of the translocated effector Yop proteins. It has been proposed that the role of LcrG in the blockage of Yop secretion is carried out in the cytoplasm and that interaction with LcrV removes this blockage (Nilles et al., 1997). The results presented in this paper would suggest that LcrG is surface located, but we have no direct evidence that LcrG is located on the outer surface of Yersinia. However, previous studies have reported that, although LcrG is mainly cytoplasmically located, some LcrG is present extracellularly in a manner dependent on the Ysc export machinery (Skrzypek and Straley, 1993). Thus, LcrG can transverse the bacterial membrane, and it is possible that a small fraction of the LcrG population could remain on the cell surface. It is at present not easy to predict how the LcrG–heparan sulphate interaction could regulate the translocation process. However, in other systems, binding to GAGs has been shown to induce modifications of the ligands. This interaction may induce polymerization of the ligand or stablilize the protein and protect it from proteolytic cleavage (Lookene et al., 1996; Sakata et al., 1997; Seiffert, 1997; Soncin et al., 1997). It can also induce profound effects on the ligand's reactivity and biological activities. For example, binding of heparin to the proteinase inhibitor antithrombin induces a conformational change that renders the active bond of antithrombin more accessible to the proteinase (Nordenman and Björk, 1978; Nordenman et al., 1978). In addition, heparin enhances the reaction by bridging antithrombin and the protease on the same polysaccharide strand (Laurent et al., 1978). Alternatively, the binding of LcrG to heparan sulphate may influence the function of heparan sulphate proteoglycans, rather than the other way round. For instance, the binding could induce proteoglycans to initiate signalling effects in the eukaryotic cell that allow translocation to occur (Rosenshine et al., 1996; Oh et al., 1997).

Although heparin and heparinitase treatment significantly reduced translocation, we were unable to inhibit translocation by 100% in either case. This suggests that perhaps the requirement for heparan sulphate–LcrG interaction in translocation can be partially compensated for by another bacterial–eukaryotic cell interaction. In this way, the heparan sulphate–LcrG interaction can be viewed as maximizing the efficiency of the translocation process but may not be absolutely essential. Some heparin-binding proteins are multifunctional, e.g. FHA binds β2 integrin CR3 on macrophages as well as binding heparan sulphate (Relman et al., 1990). LcrG may also be multifunctional, in which case heparan sulphate may act as a co-receptor. We can also not rule out the possibility that there is another receptor on the cell surface, which could use another Yop virulon protein, such as YopN, as a ligand.

Like viruses, bacterial pathogens must be able to cope with the different types of eukaryotic cells that they encounter. Translocation of Yop proteins has been demonstrated in vitro with epithelial cells, such as HeLa cells (Rosqvist et al., 1994; Sory and Cornelis, 1994), macrophages (Sory et al., 1995) and fibroblasts (M.-P. Sory, unpublished). It will thus be necessary to assess the importance of proteoglycans in the translocation of Yops into each of these cell types and probably also into other types of cells. In particular, the interaction between LcrG and heparan sulphate anchored at the surface of HeLa cells should also be studied further to investigate the manner in which this interaction induces translocation. In addition, the verification and the localization of the heparan sulphate-binding site on LcrG by site-directed mutagenesis of the putative binding residues could produce some interesting information.

Experimental procedures

Cell lines, bacterial strains, plasmids and growth conditions

HeLa human cervical epithelial carcinoma cells were grown routinely at 37°C in an 8% CO2 atmosphere in RPMI-1640 medium (Seromed) supplemented with 2 mM L-glutamine (Seromed), 10% (v/v) fetal bovine serum (Gibco) and 100 μg ml−1 streptomycin (Sory and Cornelis, 1994). E. coli LK111 received from M. Zabeau (Ghent, Belgium) was used for standard genetic manipulations. E. coli strain LK111(pMRS49) produces glutathione S-transferase (GST)–LcrG upon IPTG induction (M. R. Sarker, unpublished). Similarly, LK111(pMRS47) produces GST–LcrV (M. R. Sarker, unpublished), and LK111(pIML216) produces GST–YopN (M. Iriarte, unpublished). E40 (pYV40) is a 0:9 Y. enterocolitica isolate (Sory et al., 1995). E40(pMS111) carries a plasmid that encodes a YopE–Cya fusion protein under yopE promoter control (Sory and Cornelis, 1994). The bacterial strains were grown routinely in tryptic soy broth (TSB; Oxoid) and plated on tryptic soy agar (Oxoid). All media were supplemented with the relevant selective agents. Unless otherwise specified, the concentrations were as follows: 200 μg ml−1 ampicillin; 10 μg ml−1 chloramphenicol; 35 μg ml−1 nalidixic acid and 0.4 mM arsenite (Neyt et al., 1997). All chemicals were from Sigma unless otherwise stated. All molecular procedures were performed according to standard methods (Sambrook et al., 1989) and manufacturers' directions.

Protein production

The production of GST fusion proteins was performed basically as described by Smith and Johnson (1988). Overnight cultures of strains LK111(pMRS49), LK111(pMRS47) and LK111(pGEX) were inoculated into TSB containing 200 μg ml−1 ampicillin and vigorously shaken at 37°C. IPTG was added to a final concentration of 1 mM when the OD600 reached 0.6. The cells were collected after three more hours of growth, washed with phosphate-buffered saline (PBS) and stored at −20°C until needed. The cell pellet was resuspended in PBS, 0.5 mM dithiothreitol (DTT), 0.1 mM ethylenediamine tetraacetic acid (EDTA), 1 mM phenylmethylsulphonyl fluoride (PMSF) and disrupted with a Branson sonicator. A cell-free extract was obtained by centrifugation for 20 min at 10 000 r.p.m. Glutathione-sepharose beads (Pharmacia Biotech) were added to this extract and mixed with gentle agitation at 4°C for 1–2 h. The beads were washed three times with PBS. For some experiments, GST and GST fusion proteins were eluted from the beads with 50 mM glutathione. In other experiments, cleavage of LcrG or LcrV from the fusion protein was performed by incubation of thrombin (50 U ml−1 in PBS) with the beads at 20°C for 2–6 h. The beads were separated by centrifugation, and the supernatant containing LcrG or LcrV cleaved from the GST fusion protein was collected. For each protein preparation, 1 mM PMSF was added, and the protein was stored on ice until required. The Bradford method was used for protein quantification (Bradford, 1976) (Bio-Rad protein assay).

Binding experiments monitored with the hexosaminidase assay

HeLa cells were grown to 80% confluency and detached from the culture flask surface with PBS and 2 mM EDTA (PBS-EDTA). After washing twice in binding buffer (RPMI-1640, 10 mM L-glutamine, 50 mM HEPES, pH 7.0, 0.5% BSA), the cells were counted and adjusted to the required concentration. Purified protein (GST, GST–LcrG or LcrG) was bound in 40 μl of PBS at the required protein concentration in 96-well plates at 4°C for 16 h. The wells were washed three times with PBS, and the unoccupied sites were blocked with PBS, 0.5% BSA for 1 h at 37°C. After washing, 5–10 × 104 HeLa cells were added to each well in 100 μl of binding buffer and incubated at 37°C for 1 h. The wells were washed three times and then the hexosaminidase assay was performed (Landegren, 1984). Briefly, 60 μl of substrate solution (3.25 mM p-nitrophenyl-N-acetyl-β-D-glucosaminide, 50 mM citrate buffer, pH 5.0, 0.25% Triton X-100) was added. Incubation was carried out at 37°C for 1 h. Stop solution (90 μl) (5 mM EDTA, 50 mM glycine buffer, pH 10.4) was added and the absorbance was read at 405 nm. All experiments were performed several times in triplicate.

Modification of HeLa cells for hexosaminidase binding assay

For protease treatment, HeLa cells were first detached with PBS-EDTA and washed. Proteases (proteinase K, trypsin or chymotrypsin) were added to a concentration of 100 μg ml−1 in RPMI-1640 supplemented with L-glutamine and incubated for 5 min at 20°C. The cells were washed three times and added to the hexosaminidase assay. For xyloside treatment, p-nitrophenyl-β-D-xylopyranoside was added to the HeLa cells in growth media 48 h before detachment at a concentration of 40 mM. Cells were detached and collected as described above. For treatment with GAG lyases, HeLa cells were washed with RPMI and treated with chondroitinase ABC or heparinase I at a concentration of 50 mU ml−1 in RPMI-1640, 0.1% BSA for 3 h at 37°C. The GAG lyases were used in terms of conventional units (one conventional unit equals 1.7 mIU). The cells were processed as described above. HeLa cell controls for each modification were treated in the same way but without the addition of the active ingredient. For all these assays, purified LcrG was bound to 96-well plates at a concentration of 12 μg ml−1.

Competition assays for hexosaminidase-binding assay

LcrG was bound to the 96-well plates at a concentration of 12 μg ml−1, and the wells were blocked with PBS, 0.5% BSA. After washing, 50 μl of binding buffer was added containing the GAG at the desired concentration. After 1 h, 1 × 105 HeLa cells were added in 50 μl of reaction buffer supplemented with the GAG to maintain the correct concentration. The hexosaminidase assay was performed as described above.

Sequencing

The Sal I–SmaI fragment encoding the lcrG gene from pMRS49 was cloned into the Sal I–SmaI sites of pBluescript-KS− (to give pIM252) and sequenced with the forward and backward primers. pMRS49 was constructed with a polymerase chain reaction (PCR) product obtained using pYV22703 as a template (M. R. Sarker, unpublished). The lcrG gene present in plasmid pMRS64 was also sequenced. This clone was constructed with a PCR product obtained using pYV40 as a template (M. R. Sarker, unpublished). The two sequences were identical and were deposited in GenBank under accession number AF022645.

Binding of LcrG to heparin beads

Purified LcrG and LcrV were incubated with heparin-agarose beads (Sigma) for 16 h at 20°C with gentle rotation in Tris buffer [50 mM Tris-(hydroxymethyl)-aminoethane, pH 8.0] and 1 mM PMSF. The beads were then washed three times with Tris buffer. Tris buffer (30 μl) supplemented with 62 mM NaCl was added to the beads, allowed to incubate for 15 min at 20°C, and the supernatant was recovered from the beads. The incubation step was repeated with Tris buffer supplemented with increasing concentrations of NaCl (125, 250, 500, 1000 and 2000 mM), and the supernatants were collected after each incubation. These supernatants, along with the beads remaining after the last incubation step, were then electrophoresed on SDS–PAGE tricine gels as described by Allaoui et al. (1995).

Dimerization of LcrG

Purified LcrG was incubated in Tris buffer, 1 mM PMSF in the presence or absence of 10 μg ml−1 heparin for 1 h at 20°C. Non-reducing [no β-mercaptoethanol (BME)] or reducing (+ BME) SDS–PAGE loading dye was then added to the protein samples. These samples were electrophoresed on a tricine SDS–PAGE gel.

Yop induction

For the induction of the yop regulon, Y. enterocolitica E40 was grown in BHI broth supplemented with 4 mg ml−1 glucose, 20 mM MgCl2 and 20 mM sodium oxalate (BHI-OX). For non-permissive conditions, BHI was supplemented with 4 mg ml−1 glucose and 5 mM CaCl2 (BHI-Ca). In some cases, heparin was added to the growth media at a concentration of 50 μg ml−1. Yop preparation from culture supernatants and analysis by SDS–PAGE were performed as described by Cornelis et al. (1987).

Yop translocation assay

Translocation assays were carried out essentially as described by Sory and Cornelis (1994). HeLa cells were seeded into 24-well tissue culture plates at a density of 1–2 × 105 cells per 1 ml medium per well and allowed to adhere for 20 h. Before infection with Y. enterocolitica, cells were washed and covered with RPMI-1640 supplemented only with 2 mM L-glutamine. Cytochalasin D was added 30 min before infection at a final concentration of 5 μg ml−1 (stock solution 2 mg ml−1 in dimethyl sulphoxide). A freshly isolated transconjugant colony of Yersinia E40 (pMS111) was cultured overnight in BHI at 20°C and diluted the next day to an OD600 of 0.2 in 5 ml of medium. After growth with shaking at 20°C for 2 h and 37°C for 1 h, bacteria were washed and suspended in saline. Samples of 100 μl, containing about 107 bacteria, were added to the monolayer, and the infected cultures were incubated at 37°C for 2 h in an 8% CO2 atmosphere. Cells were washed and then lysed in denaturing conditions (100°C for 5 min in 50 mM HCl, 0.1% Triton X-100) as described previously (Sory and Cornelis, 1994). The lysate was neutralized by NaOH, and cAMP was extracted with ethanol. After centrifugation, the supernatant was dried, and cAMP was assayed by an enzyme immunoassay (Biotrak; Amersham). All experiments were performed several times in triplicate.

Competition of translocation assay with GAGs

GAGs at the desired concentration were added to the HeLa cells at the time of cytochalasin D addition. Y. enterocolitica E40(pMS111) was grown in BHI at 20°C for 2 h and then shifted to growth at 37°C for 1 h. The cells were harvested and washed in saline. GAGs were added to 1 × 108 bacteria ml−1 in RPMI, 2 mM L-glutamine, and incubated at 37°C for 15 min. Aliquots (100 μl) were added to the HeLa cell monolayer, and the translocation assay was performed.

Lyase treatment of HeLa cells for the translocation assay

For treatment with GAG lyases, HeLa cells were washed with RPMI and treated with chondroitinase ABC or heparinase III (heparinitase) at a desired concentration in RPMI-1640, 0.1% BSA for 3 h at 37°C. The cells were washed, and fresh RPMI-1640, 2 mM L-glutamine supplemented with 5 μg ml−1 cytochalasin D and the appropriate GAG lyase was added. The translocation assay was performed as described above.

Yersinia–HeLa cell adherence assay

To monitor the adherence of bacteria to eukaryotic cells, HeLa cells were grown on coverslips in 24-well plates, and infection with E40(pMS111) was carried out as described above for the translocation assays. The infected cells were fixed with 2% glutaraldehyde in PBS, stained with Giemsa and examined under the microscope to count the attached bacteria.

Acknowledgements

We thank Isabelle Lambermont for technical assistance. We thank Cecile Geuijen for critical reading of the manuscript. A.P.B. was a recipient of a fellowship attributed by ICP. 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).

Ancillary