YopH is a protein tyrosine phosphatase (PTP) that is delivered into host mammalian cells via a type III secretion pathway in pathogenic Yersinia species. Although YopH is a highly active PTP, it preferentially targets a subset of tyrosine-phosphorylated proteins in host cells, including p130Cas. Previous in vitro studies have indicated that the carboxy-terminal PTP domain contributes specificity to the interaction of YopH with substrates. However, it is not known if the PTP domain is sufficient for substrate recognition by YopH. Here, we have identified paxillin as an additional substrate of YopH in HeLa cells. In addition, we have identified a domain in the amino-terminal region of YopH that binds to both p130Cas and paxillin and is required for the efficient recognition of substrates by the wild-type enzyme. This ‘substrate-binding’ domain exhibits a ligand specificity that is similar to that of the Crk Src homology 2 (SH2) domain, and it binds substrates directly in a phosphotyrosine-dependent manner. The substrate-binding domain of YopH may represent a novel type of protein–protein interaction module, as it lacks significant sequence similarity with any known SH2 or phosphotyrosine-binding (PTB) domain.
The 468-amino-acid YopH protein appears to be composed of several modular domains. By fusing fragments of YopH to a CyaA reporter protein, Sory et al. (1995) have mapped the minimal regions required for secretion or translocation of this protein to the amino-terminal 17 or 71 residues respectively. In addition, the binding site for the YopH-specific chaperone SycH has been mapped between residues 20 and 69 (Wattiau et al., 1994; Woestyn et al., 1996). Residues 205–468 of YopH correspond to the PTP domain (Guan and Dixon, 1990). A catalytic Cys residue that is conserved in all PTPs is located within this domain at residue 403 (Guan and Dixon, 1990). Conversion of this Cys residue to either Ser (C403S) or Ala (C403A) abolishes the catalytic activity of YopH (Guan and Dixon, 1990; Bliska et al., 1993). However, these catalytically inactive proteins retain the ability to bind to substrates and are referred to as ‘substrate-trapping’ mutants (Bliska et al., 1992; Black and Bliska, 1997).
Although YopH is an extremely active PTP, several studies have demonstrated that this enzyme acts selectively on a subset of tyrosine-phosphorylated proteins in host cells at early infection times (Bliska et al., 1992; Andersson et al., 1996). In addition, Zhang et al. (1993) have shown that the PTP domain of YopH exhibits substrate specificity in vitro. Compared with the mammalian PTP1 enzyme, the YopH PTP domain displayed a unique preference for phosphopeptide substrates that contain a proline at the +3 position relative to the phosphotyrosine (Zhang et al., 1993). One of the major targets of YopH in focal adhesions in HeLa cells was recently identified as the Crk-associated substrate p130Cas (Cas) (Black and Bliska, 1997; Persson et al., 1997). Interestingly, Cas contains 15 copies of the motif YxxP (single letter code, where × represents any amino acid and P is proline at the +3 position; Sakai et al., 1994). YxxP corresponds to the optimal phosphotyrosine-based recognition sequence for the SH2 domain of the Crk protein (Birge et al., 1993). Taken together, these results suggest that the PTP domain of YopH acts preferentially on substrates that are tyrosine phosphorylated at YxxP motifs (Black and Bliska, 1997). However, it is not known if the PTP domain is sufficient for the recognition of substrates by YopH in vivo. Most PTPs also contain an accessory domain, for example an SH2 domain, that influences substrate recognition (reviewed by Mauro and Dixon, 1994). No such sequence has yet been identified in YopH.
Here, we report the identification of paxillin as an additional substrate of YopH in HeLa cells. In addition, we have identified a domain within the amino-terminal 130 amino acids of YopH that binds to the same proteins that are dephosphorylated by the PTP domain and is required for the efficient recognition of these substrates by the wild-type enzyme. This ‘substrate-binding domain’ displays a ligand-binding specificity that is similar to that of the Crk SH2 domain, and it binds substrates directly in a phosphotyrosine-dependent manner. The substrate-binding domain of YopH is potentially a novel type of protein–protein interaction module, as it lacks significant sequence similarity with any known SH2 or PTB domain.
YopH selectively dephosphorylates Cas and a 68 kDa protein in HeLa cells infected with Y. pseudotuberculosis
The goal of this study was to examine the mechanism of substrate recognition by YopH. Initially, we wanted to determine if there are other focal adhesion proteins, in addition to Cas, that are preferential substrates of YopH in HeLa cells. In a previous study, we detected a major 68 kDa tyrosine-phosphorylated protein (p68) in HeLa cells that was dephosphorylated in a YopH-specific fashion (Black and Bliska, 1997). As the phosphorylation of p68 was constitutive in adherent HeLa cells (Black and Bliska, 1997), we hypothesized that p68 was tyrosine phosphorylated in response to integrin-mediated cell adhesion. To test this idea, we plated HeLa cells in dishes coated with the extracellular matrix protein fibronectin. Adhesion of cells to fibronectin has been shown to induce sustained tyrosine phosphorylation of focal adhesion proteins over a several hour time period (Burridge et al., 1992). Sixty minutes after plating, the plated cells, or control cells held in suspension for the same time period, were lysed in detergent, and samples of the lysates were analysed by immunoblotting with the anti-phosphotyrosine antibody 4G10. 1Figure 1A shows that p68 and a higher molecular weight protein band of approximately 130 kDa underwent tyrosine phosphorylation in response to cell plating on fibronectin (compare lanes 1 and 2). Immunodepletion experiments indicated that the 130 kDa band was composed primarily of two proteins, focal adhesion kinase (Fak) and Cas, with Fak representing the major tyrosine-phosphorylated component of this complex (data not shown). In order to monitor changes in tyrosine phosphorylation of Cas and Fak accurately, these proteins were immunoprecipitated from cell lysates and analysed by 4G10 immunoblotting. 1Figure 1B and C show that both Cas and Fak were tyrosine phosphorylated in response to cell plating on fibronectin (compare lanes 1 and 2). These results indicated that p68, like Cas and Fak, is tyrosine phosphorylated in response to integrin-mediated cell adhesion.
We compared the selectivity of YopH for these different tyrosine-phosphorylated proteins in a time course infection assay. HeLa cells plated on fibronectin were infected for different lengths of time with IP17/yopH+, a Y. pseudotuberculosis yopE yopH mutant expressing wild-type yopH+in trans from a multicopy plasmid. For these experiments, the bacteria were pregrown at 37°C in the presence of 2.5 mM Ca2+, conditions that prevent activation of the type III pathway. Therefore, we anticipated a lag of several minutes after bacterial–host cell contact before the type III pathway would be activated to deliver the Yops into the HeLa cells. After 15 min of infection with IP17/yopH+, there was an apparent further stimulation in phosphorylation of multiple proteins including p68 (Fig. 1A, lane 3), Cas (Fig. 1B, lane 3) and Fak (Fig. 1C, lane 3). This enhanced protein phosphorylation may have resulted from attachment of the bacteria to β1 integrin receptors mediated by the invasin or YadA proteins (Bliska et al., 1993; Isberg and Tran Van Nhieu, 1994). However, this enhanced protein phosphorylation during infection was not observed reproducibly (for example, see Fig. 2). After 30 min of infection with IP17/yopH+, p68 and Cas were significantly dephosphorylated (Fig. 1A and B, lane 4), while Fak was not dephosphorylated to completion even after 60 min (Fig. 1C, lane 5). The resistance of Fak to dephosphorylation by YopH was also evident in the relatively small decrease in tyrosine phosphorylation of the 130 kDa band, which is composed largely of Fak, after 60 min of infection by IP17/yopH+ (Fig. 1A, lane 5). p68 and Cas were not dephosphorylated when HeLa cells were infected with a control strain producing the catalytically inactive form of YopH (data not shown; Black and Bliska, 1997), indicating that the dephosphorylation events we observed were specific to the PTP activity of YopH. These results indicated that, in HeLa cells, the PTP domain of YopH selectively targets two proteins that are tyrosine phosphorylated in response to integrin-mediated cell adhesion, Cas and p68.
Identification of p68 as paxillin
Paxillin is a 68 kDa focal adhesion protein that is tyrosine phosphorylated in response to integrin-mediated cell adhesion (Burridge et al., 1992). To determine if p68 corresponds to paxillin, lysates from HeLa cells plated on fibronectin-coated dishes were subjected to immunoprecipitation with an anti-paxillin antibody. The lysates and the immune complexes were then analysed by immunoblotting with 4G10 or anti-paxillin antibody. As shown in 2Fig. 2A and B, both p68 and paxillin were depleted from the lysate by the anti-paxillin antibody (compare lanes 2 and 3). In addition, paxillin was tyrosine phosphorylated in response to cell plating on fibronectin (Fig. 2A, compare lanes 4 and 5). When HeLa cells plated on fibronectin were infected for different lengths of time with IP17/yopH+, paxillin was dephosphorylated to near completion after 30 min of infection (Fig. 2A, lanes 6–8). These results demonstrate that p68 is paxillin.
Identification of a substrate-binding domain in the amino-terminal region of YopH
An in vitro substrate binding assay was established in order to identify structural features in YopH that mediate specific interactions with Cas and paxillin. Initially, a protein consisting of glutathione S-transferase (GST) fused to full-length, catalytically inactive YopHC403S (GST1-468) was purified (Fig. 3A and B) and tested for its ability to bind Cas or paxillin in detergent lysates of fibronectin-plated HeLa cells. Lysates that had been precleared by incubation with GST were subsequently incubated with GST1-468 immobilized on beads and, after extensive washing, bound proteins were analysed by 4G10 immunoblotting. For comparison, similar binding reactions were performed with a GST fusion protein containing the SH2 domain of v-Crk (GST-CrkSH2) (Fig. 3B), which has been shown to bind tyrosine-phosphorylated Cas and paxillin selectively in cell lysates (Birge et al., 1993). As shown in 4Fig. 4A, GST1-468 and GST-CrkSH2 bound to similar, but non-identical, arrays of tyrosine-phosphorylated proteins in HeLa cell lysates (compare lanes 3 and 9). A control binding reaction performed with immobilized GST demonstrated that these tyrosine-phosphorylated proteins were not binding to the GST domain of either fusion protein (data not shown). The tyrosine-phosphorylated proteins of 130 and 68 kDa that bound most efficiently to GST1-468 and GST-CrkSH2 were identified as Cas and paxillin, respectively, by reprobing the blot with anti-Cas or anti-paxillin antibodies (data not shown, see Fig. 4B). These results indicated that Cas and paxillin interact specifically with full-length YopHC403S in cell lysates.
To determine if there are structural elements in YopH outside of the PTP domain that are capable of specific interaction with Cas and paxillin, a GST fusion protein consisting of the first 203 amino acids of YopH (GST1-203) was purified (Fig. 3A and B) and tested for substrate-binding activity. Surprisingly, GST1-203 bound a similar array of tyrosine-phosphorylated proteins as did GST1-468 and GST-CrkSH2 (Fig. 4A, compare lanes 3, 6 and 9). The blot shown in 4Fig. 4A was stripped and reprobed with an anti-paxillin antibody to confirm that the 68 kDa protein that bound to GST1-203 and GST-CrkSH2 was paxillin (Fig. 4B, lanes 6 and 9). These binding assays were repeated using lysates of rat embryo fibroblast (REF-52) cells, which contain significantly higher levels of tyrosine-phosphorylated paxillin than HeLa cells. Again, GST1-203 bound significant amounts of Cas and paxillin (Fig. 5, lane 9). In addition, this binding was selective, as an equally abundant phosphoprotein of 60 kDa was not recognized by the GST fusion proteins (Fig. 5, compare lanes 7 and 9). YopH contains a proline-rich region located between residues 154 and 171 (Bliska, 1996). To determine if this proline-rich region was required for substrate-binding activity, a GST fusion protein containing only the first 130 amino acids of YopH (GST1-130) was purified (Fig. 3A and B) and tested. GST1-130 displayed the same high level of Cas- and paxillin-binding activity as did GST1-203 (Fig. 5, compare lanes 6 and 9). Therefore, the proline-rich region of YopH was dispensable for this activity. These results indicated that, in addition to the PTP domain, YopH contains an additional domain that is capable of specific recognition of Cas and paxillin. This ‘substrate-binding domain’, which is located within the amino-terminal 130 amino acids of YopH, appears to bind to the same proteins that are dephosphorylated by the PTP domain. Furthermore, the substrate-binding domain displays a ligand preference that is similar to that of the Crk SH2 domain (Fig. 5, compare lanes 6 and 12).
Deletions falling within the first 130 amino acids of YopH result in global defects in protein structure
Additional GST fusion proteins encoding residues 1–75, 1–91, or 1–104 of YopH were produced in an attempt to define the carboxy-terminal boundary of the substrate-binding domain further. However, unlike GST1-130, these fusion proteins (GST1-75, GST1-91 and GST1-104) were insoluble when overproduced in E. coli (data not shown). We also introduced a number of small in frame deletions into the amino-terminal region of the full-length YopHC403S protein in an attempt to inactivate the substrate-binding domain. A derivative of YopHC403S with residues 130–153 deleted was found to be secreted and translocated at wild-type levels (see below). Using this construct as a starting point, a nested series of in frame deletions extending into the amino-terminal region was generated by removing residues 105–153, 92–153 or 76–153. As the SycH-binding domain has been mapped between residues 20 and 69 (Wattiau et al., 1994; Woestyn et al., 1996), we expected that all of these proteins would be secreted and translocated at near wild-type levels. Plasmids expressing these various proteins were introduced into IP17, and IPTG was used to induce high-level protein expression under growth conditions that promote Yop secretion. After centrifugation of the cultures, bacterial-associated proteins or proteins secreted into the media were analysed by SDS–PAGE. As shown in Fig. 6, full-length YopHC403S (1–468) and YopHΔ130–153 were produced efficiently in the bacteria (lanes 8 and 9) and secreted at high levels (lanes 2 and 3). In contrast, the YopHΔ76–153, Δ92–153 and Δ105–153 proteins were produced at lower levels in the bacteria (lanes 10–12) and secreted at lower levels into the media (lanes 4–6). The strains expressing the YopHΔ76–153, Δ92–153 and Δ105–153 proteins also secreted significantly lower levels of other Yops (lanes 4–6), indicating that the mutant YopH proteins were somehow interfering with either the production or the secretion of other Yops. The YopHΔ76–153, Δ92–153 and Δ105–153 proteins were also produced at very low levels when the bacteria were used to infect HeLa cell cultures (data not shown). The defect in expression of these mutant proteins did not appear to be related to the disruption of SycH-binding activity, as they were also produced at very low levels in E. coli (data not shown). We interpret these data to indicate that small deletions extending into the amino-terminal 130 residues of YopH result in a global disruption of protein structure. As a consequence, the proteins containing these deletions are either rapidly degraded or aggregated into insoluble complexes.
The substrate-binding domain of YopH binds substrates directly in a phosphotyrosine-dependent manner
An overlay assay was performed to determine if the substrate-binding domain of YopH binds directly to substrates in a phosphotyrosine-dependent manner. For this purpose, tyrosine-phosphorylated Cas was co-precipitated with an epitope-tagged form of YopHC403S (see Experimental procedures). The immune complex was divided into equal portions, boiled to dissociate YopHC403S and Cas, and one sample was treated with purified YopH to dephosphorylate Cas. The treated and untreated samples were then subjected to SDS–PAGE in parallel and transferred to a membrane. The membrane was incubated with GST1-130 and, after washing, binding of the fusion protein was detected by immunoblotting with anti-GST antibody. Subsequently, the membrane was stripped and reprobed with either anti-Cas or 4G10 antibodies as a control. As shown at the top of 7Fig. 7A, GST1-130 bound to the region of the membrane containing tyrosine-phosphorylated Cas (lane 1), but did not bind to the region containing dephosphorylated Cas (lane 2). Cas was completely dephosphorylated by YopH, as shown by its increased gel mobility and decreased staining with 4G10 in the control blots (lanes 1 and 2, middle and bottom of Fig. 7A). Therefore, the substrate-binding domain of YopH binds directly to Cas in a phosphotyrosine-dependent manner.
To investigate the mechanism of interaction between the substrate-binding domain of YopH and tyrosine-phosphorylated proteins further, we assayed binding of Cas and paxillin to GST1-130 in the presence of phosphotyrosine. Phosphotyrosine has been shown to act as a competitive inhibitor of the binding of tyrosine-phosphorylated proteins to either SH2 or PTB domains (for example, see Kavanaugh and Williams, 1994). The addition of 10 mM phosphotyrosine to HeLa cell lysates completely inhibited binding of both Cas and paxillin to GST1-130 (Fig. 7B, lane 5). The band migrating below the position of paxillin that was not competed appears to be non-specific. Decreased binding of paxillin, but not of Cas, to GST1-130 was also observed at 1 mM phosphotyrosine (lane 4), suggesting that these proteins interact with the substrate-binding domain with different affinities. No inhibition of binding of either substrate was observed in the presence of 10 mM phosphoserine (lane 9). These results provided additional evidence that, like the SH2 and PTB domains, the substrate-binding domain of YopH recognizes targets in a phosphotyrosine-dependent manner.
The substrate-binding domain increases the efficiency of substrate dephosphorylation by YopH
The functional role of the substrate-binding domain was addressed by examining the effect of its removal on the ability of YopH to dephosphorylate a physiological substrate. As we were unable to generate a form of YopH lacking the substrate-binding domain that could be efficiently produced and secreted by the bacteria (see above), we turned to an in vitro dephosphorylation assay (Garton et al., 1996). Lysates of REF-52 cells were used as the source of tyrosine-phosphorylated substrates. These lysates were incubated for different lengths of time with one of two purified proteins, either GST fused to wild-type YopH (GST–YopH) or GST fused to a truncated form of YopH lacking the entire amino-terminal region (GSTΔ1–170). Like full-length YopH, this latter protein can be expressed at high levels in E. coli in a soluble form. After incubation with the enzymes, samples of the lysates were analysed by immunoblotting with 4G10 to detect the dephosphorylation of substrates. As shown in Fig. 8, GST–YopH selectively dephosphorylated paxillin within 0.5 min of its addition to the lysate (lane 2). In contrast, approximately 10 min was required for the complete dephosphorylation of paxillin by GSTΔ1–170 (lane 12). Although the rate of paxillin dephosphorylation by GSTΔ1–170 was significantly decreased, its substrate selectivity was not altered, as evidenced by the fact that GSTΔ1–170 displayed a dramatic preference for paxillin compared with other equally abundant tyrosine-phosphorylated proteins in the lysate (lanes 8–14). Dephosphorylation of Cas was not apparent in this experiment because of the fact that it co-migrates with the more abundant tyrosine-phosphorylated Fak protein. Zhang et al. (1992) have demonstrated that full-length YopH and a mutant protein lacking the first 162 amino acids have the same Km and kcat values using p-nitrophenyl phosphate (p-NPP) as a substrate. In addition, we confirmed that GST–YopH and GSTΔ1–170 have similar activities in assays of p-NPP hydrolysis (data not shown). Therefore, the decreased efficiency of paxillin dephosphorylation by GSTΔ1–170 was not caused by a general reduction in PTP activity because of truncation of the amino-terminal region. These results confirmed that the PTP domain of YopH displays intrinsic substrate specificity, and further demonstrated that the substrate-binding domain dramatically increases the efficiency of substrate dephosphorylation by this enzyme.
In this study, we examined the mechanism of substrate recognition by the Yersinia tyrosine phosphatase YopH. We found that YopH preferentially dephosphorylated two focal adhesion proteins in HeLa cells plated on fibronectin. One of these proteins is the previously identified substrate Cas (Black and Bliska, 1997; Persson et al., 1997). The other substrate of 68 kDa was identified as paxillin. Interestingly, Cas and paxillin share several features in common. Both proteins are tyrosine phosphorylated in response to integrin-mediated cell adhesion, and both function as adaptors to which downstream signalling components, such as Crk, bind via their SH2 domains (reviewed by Turner, 1994; Burridge and Chrzanowska-Wodnicka, 1996). Cas contains 15 repetitions of the amino acid sequence YxxP, which corresponds to the optimal binding site for the Crk SH2 domain (Sakai et al., 1994). As Cas forms a complex with v-Crk (Sakai et al., 1994), at least some of these sites must be tyrosine phosphorylated in vivo. Paxillin contains three YxxP motifs, corresponding to tyrosines 31, 118 and 181 (Turner and Miller, 1994; Salgia et al., 1995). Phosphopeptide analysis of paxillin has confirmed that residues 31 and 118 of paxillin are tyrosine phosphorylated in vivo, and that both sites are recognized by the Crk SH2 domain in vitro (Birge et al., 1993; Schaller and Parsons, 1995). The other focal adhesion protein we analysed was Fak. Fak has been shown to be tyrosine phosphorylated at six sites, corresponding to tyrosines 397, 407, 576, 577, 861 and 925 (Calalb et al., 1995; Ilic et al., 1997). Only one of these sites, tyrosine 407, is contained within a YxxP motif (YTMP) (Calalb et al., 1995; Ilic et al., 1997). The biological role of this phosphorylation site is not known, although it has been proposed to act as a binding site for an SH2 domain-containing protein (Calalb et al., 1995; Ilic et al., 1997). Based on anti-phosphotyrosine immunoblotting, Fak was relatively resistant to complete dephosphorylation by YopH compared with Cas and paxillin. Although we were unable to determine if tyrosine 407 of Fak was efficiently dephosphorylated by YopH, it is clear that other major sites of tyrosine phosphorylation in Fak were not. These results provided support for the concept that the PTP domain of YopH selectively targets tyrosine-phosphorylated sites that are contained within YxxP motifs.
Next, we examined the binding of Cas and paxillin to a truncated form of YopH lacking the PTP domain in order to determine if there are structural elements in the amino-terminal region of YopH that contribute to its specific interaction with these substrates. We identified a domain contained within the first 130 amino acids of YopH that binds selectively to Cas and paxillin in cell lysates. This finding was somewhat unexpected, as the amino-terminal region of YopH was previously thought only to contain signals important for the secretion and translocation of this protein (Sory et al., 1995). The minimal sequences required for the secretion or translocation of Yop–CyaA reporter proteins have been mapped to the amino-terminal 17 and 71 residues of YopH, respectively (Sory et al., 1995), and the binding site for the YopH-specific chaperone SycH has been mapped between residues 20 and 69 (Wattiau et al., 1994; Woestyn et al., 1996). Our attempts to refine the boundaries of the substrate-binding domain and to define its relationship to sequences important for secretion/translocation by deletion analysis were unsuccessful because of apparent protein misfolding problems. Preliminary structural analysis has revealed that the amino-terminal 130 amino acids of YopH are folded into a single, highly ordered domain (C. Smith and M. Saper, personal communication). Taken together with the results of our deletion analysis, it appears that the amino-terminal 130 amino acids of YopH comprise a single, multifunctional domain that is highly sensitive to misfolding upon perturbation of its primary sequence. Our findings appear to be at odds with the demonstration that residues 1–71 of YopH were able to mediate the translocation of a Yop–CyaA fusion protein into macrophages efficiently (Sory et al., 1995). One possibility is that the CyaA domain has special properties that allow it to stabilize small fragments of the YopH amino-terminal domain.
To investigate the mechanism of substrate recognition by the substrate-binding domain, we examined its interaction with Cas in an overlay assay. The results of this assay showed that the substrate-binding domain binds Cas directly in a phosphotyrosine-dependent manner. In addition, we were able to inhibit the binding of Cas and paxillin competitively to the substrate-binding domain using phosphotyrosine, but not phosphoserine. Thus, the substrate-binding domain functions in the same way as two previously identified domains that mediate phosphotyrosine-dependent protein–protein interactions, the SH2 domain and the PTB domain (reviewed by Cohen et al., 1995). However, comparison of the sequence of the amino-terminal region of YopH with sequences in computer databases revealed no significant matches with any known SH2 domain or PTB domain. SH2 domains share significant similarity at the level of protein sequence and structure, while PTB domains share similarity only at the structural level (Cohen et al., 1995). Therefore, it is possible that the amino-terminal region of YopH contains a PTB domain, although the three-dimensional structure of this region will need to be determined to reconcile this issue. The only protein in the databases that shares significant similarity with this region of YopH is YscM (LcrQ), a 115-amino-acid secreted Yersinia protein that appears to function as a negative regulator of the type III pathway (Michiels et al., 1991; Rimpilainen et al., 1992). As reported previously, residues 51–129 of YopH share 41% identity with residues 42–115 of YscM (Michiels et al., 1991; Rimpilainen et al., 1992). The functional significance of this similarity is not known.
In addition to one or more catalytic domains, PTPs often contain some type of accessory domain that controls their subcellular localization and/or their ability to interact with substrates (reviewed by Mauro and Dixon, 1994). It is well documented that SH2 domains can influence substrate recognition by PTPs. For example, the SH2 domains in SHP1 and SHP2 appear to localize these enzymes to signalling complexes by binding to specific phosphotyrosine residues (Mauro and Dixon, 1994). Binding of the SH2 domains in SHP1 or SHP2 to a tyrosine-phosphorylated protein leads to activation of the PTP domain, which then dephosphorylates the site bound to the SH2 domain or other sites in its vicinity (Mauro and Dixon, 1994). The functional role of the substrate-binding domain in YopH was assessed by examining the effect of its removal on the ability of YopH to dephosphorylate paxillin in cell lysates. Our results show that deletion of the substrate-binding domain significantly reduces the rate of substrate dephosphorylation, without diminishing the intrinsic specificity of the PTP domain. These findings are similar to those reported by Garton et al. (1997), who showed that binding of a proline-rich region in PTP-PEST to the SH3 domain in Cas enhances the ability of PTP-PEST to recognize Cas as a substrate. However, mutation of the proline-rich region in PTP-PEST did not qualitatively alter the intrinsic specificity of its PTP domain for Cas (Garton et al., 1997). The interaction of the Cas SH3 domain with the proline-rich region in PTP-PEST is likely to prolong the association of these two proteins, thereby facilitating complete dephosphorylation of Cas (Garton et al., 1997). The substrate-binding domain in YopH is also likely to increase the efficiency of substrate recognition. It is less obvious how the substrate-binding domain in YopH enhances the efficiency of substrate dephosphorylation by the PTP domain, as both domains potentially compete for the same tyrosine-phosphorylated sites. One possibility is that the substrate-binding domain allows the PTP domain to act processively in cases in which the substrate contains multiple sites of tyrosine phosphorylation. This situation would be analogous to the role of SH2 domains in promoting processive phosphorylation events by cytoplasmic tyrosine kinases (Mayer et al., 1995). In both cases, the processivity of the reaction is driven by the presence, within a single protein, of a substrate-binding domain and a catalytic domain with overlapping ligand/substrate recognition sequences (Mayer et al., 1995). The fact that both Cas and paxillin are tyrosine phosphorylated at multiple YxxP motifs in vivo (Sakai et al., 1994; Schaller and Parsons, 1995) is consistent with this model.
In summary, we have shown that Cas and paxillin are preferential targets of YopH in HeLa cells and that the amino-terminal substrate-binding domain dramatically enhances the efficiency of substrate dephosphorylation by this PTP. The substrate-binding domain exhibits a ligand preference that is similar to that of the Crk SH2 domain. From these data, we conclude that YopH preferentially targets proteins that are tyrosine phosphorylated on YxxP motifs. Recent studies have indicated that tyrosine phosphorylation of Cas and paxillin is important for integrin-mediated cell spreading and cell migration (Richardson et al., 1997; Cary et al., 1998; Klemke et al., 1998). Furthermore, current evidence suggests that Cas promotes integrin-mediated cell migration through interaction with Crk and that Crk is coupled via a downstream signalling pathway to the Rho family GTPase Rac (Klemke et al., 1998). YopH is likely to interfere with this type of signal transduction pathway through its ability to dephosphorylate Cas and paxillin. This activity may provide the mechanism by which YopH disrupts focal adhesion complexes in HeLa cells (Black and Bliska, 1997; Persson et al., 1997). A similar mechanism may also underlie the ability of YopH to inhibit phagocytosis and the production of superoxide in response to phagocytosis, as these processes have also been linked to tyrosine phosphorylation of proteins, such as paxillin, and activation of Rho family GTPases (reviewed in Bokoch, 1995; Greenberg, 1995).
Bgl II linkers were purchased from New England Biolabs. The oligonucleotide PTP5 has been described previously (Bliska and Black, 1995). The following oligonucleotides were synthesized: PTP6 5′-AGAATTCTGTGGTGAGGCGGCTG-3′; PTP7 5′-CGGATCCTGAATTCTTAGCTATTTAATAATGGTCGCCCTT-3′; PTP17 5′-CGAATTCAATATTACGTCTCTGTCTCAAAAGGAGGTAGGCGATCCCGGGAATTTAATA- ATGGTCGCCCTTGTCC-3′; and PTP18 CGGATCCCATATGAACTTATCATTAAGC-3′. The monoclonal antibody M45, which recognizes the 12-amino-acid epitope SRDRLPPFETET, was provided by Dr P. Hearing (SUNY, Stony Brook, USA). Monoclonal anti-phosphotyrosine antibody 4G10 was purchased from Upstate Biotechnology. Monoclonal antibodies directed against paxillin (03-6100) or p130Cas (P27820) were purchased from Zymed and Transduction Laboratories respectively. Rabbit anti-Fak (C-20), rabbit anti-Cas (C-20) and monoclonal anti-GST (12) antibodies were purchased from Santa Cruz Biotechnology. Anti-mouse and anti-rabbit IgG conjugated to horseradish peroxidase (HRP) were purchased from Sigma.
Bacterial strains and growth conditions
The Y. pseudotuberculosis yopH yopE mutant IP17 containing the IPTG-inducible expression plasmids pYOPH or pYOPHC403S has been described previously (Black and Bliska, 1997). pYOPHC403SM45 was constructed in the same fashion as pYOPHC403S (Bliska and Black, 1995), except that PTP17, which encodes the M45 epitope, was used as the reverse primer in a polymerase chain reaction (PCR). pYOPHC403SM45 was introduced into IP17 by conjugation (Black and Bliska, 1997), yielding IP17/yopHC403SM45. Before infection, bacteria were grown overnight at 26°C with shaking in Luria broth containing 100 μg ml−1 ampicillin (LB-Amp). The next morning, bacteria were subcultured into LB-Amp supplemented with 2.5 mM CaCl2 to an OD600 = 0.1. Cultures were shaken at 37°C for 2 h. Bacteria were pelleted by centrifugation and resuspended in warm Hanks' balanced salt solution (HBSS) to an OD600 = 1.0 (≈1 × 109 cfu ml−1). Growth of bacteria under low-calcium conditions and analysis of Yop secretion was performed as described previously (Palmer et al., 1998).
HeLa and rat embryo fibroblast (REF-52) cells were cultured in DMEM (Gibco BRL) supplemented with 10% heat-inactivated fetal calf serum (Gibco BRL) and 1 mM sodium pyruvate in a 5% CO2 humidified incubator at 37°C. Unless otherwise stated, HeLa and REF-52 cells were prepared for assays by plating 2 × 106 cells in 10 ml of media into 100-mm tissue culture dishes. The cells were cultured overnight in a 5% CO2 humidified incubator at 37°C. Alternatively, 2 × 106 suspended HeLa cells in 10 ml of serum-free media were seeded into fibronectin-coated 100 mm tissue culture dishes as described previously (Black and Bliska, 1997).
HeLa cells were seeded into fibronectin-coated 100 mm tissue culture dishes or held in suspension for 60 min (Black and Bliska, 1997). The suspended cells and one dish of plated cells were then lysed in RIPA as described previously (Black and Bliska, 1997). The remaining dishes of plated cells were infected with IP17/yopH+ at a multiplicity of infection (MOI) of 50:1. At 15 min, 30 min or 60 min after infection, the cells were lysed in RIPA. The lysates were subjected to immunoprecipitation with anti-paxillin (2 μg), anti-Cas (0.4 μg) or anti-Fak (0.4 μg) antibodies as described previously (Black and Bliska, 1997), except that GammaBind Plus Sepharose (Pharmacia) was used for the paxillin immunoprecipitation instead of protein A-Sepharose.
GST fusion vectors
pGEX-CrkSH2 (Birge et al., 1993) was provided by H. Hanafusa. pGEX2T-YOPHC403S was constructed using PCR and PTP18 and PTP7 as 5′ and 3′ primers, respectively. The PCR product was amplified using the yopHC403S gene (Bliska and Black, 1995) as a template. The resulting product was digested with BamHI and EcoRI and inserted between the BamHI and EcoRI sites in the multicloning region of pGEX-2T (Pharmacia). The inserted DNA was sequenced to confirm its structure. pGEXKG-YOPH1–203 was constructed similarly using primers PTP5 and PTP6, except that the product was digested with NdeI, then treated with Klenow to create a blunt 5′ end and, lastly, digested with EcoRI. The resulting fragment was inserted between the SmaI and EcoRI sites in pGEX-KG (Guan and Dixon, 1991). pGEXKG-YOPH1–130 was constructed using primers PTP5 and PTP6 to amplify a DNA fragment. This fragment was digested with BamHI and NlaIII and then inserted between the BamHI and SphI sites of pUC19 (Yanisch-Perron et al., 1985), yielding pUC191–130. pUC191–130 was digested with NdeI and treated with Klenow to create blunt ends. Subsequently, pUC191–130 was digested with HindIII, and the DNA fragment encoding the first 130 amino acids of YopH was isolated and inserted in frame between the SmaI and HindIII sites of pGEX-KG.
Introduction of deletions into the amino-terminal domain
pGEXKG-YOPH1–130 was modified by the insertion of Bgl II linkers at one of three unique restriction sites (BlpI, Afl II or SnaBI) within the coding region of the amino-terminal domain. Before insertion of the linkers, the BlpI and Afl II sites were made blunt using Klenow fragment. The change in reading frame at the site of linker insertion resulted in the premature termination of the fusion protein (a small number of amino acids were added to the carboxy-termini of each protein before a stop codon was encountered). pYOPHC403S was modified likewise by the insertion of Bgl II linkers at the BlpI, Afl II or SnaBI restriction sites. To create internal deletions, the resulting plasmids were digested with Bgl II and EcoRI, and the coding region downstream of the Bgl II site was replaced with a BamHI–EcoRI fragment specifying residues 154–468 of YopHC403S.
GST fusion protein purification and binding assays
GST fusion proteins were purified as described previously (Guan and Dixon, 1991). REF-52 cells cultured under standard conditions or HeLa cells plated on fibronectin for 2 h were used as sources of lysates. Dishes were placed on ice and washed twice with 10 ml of ice-cold PBS containing 10 mM NaF and 1 mM Na3VO4. Cells were lysed with 0.5 ml of ice-cold modified RIPA buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1 mM EDTA, 1% Nonidet P40, 0.1% SDS, 0.5% deoxycholic acid, 1 mM Na3VO4, 10 mM NaF, 200 μM AEBSF, 20 μM leupeptin and 1 μM pepstatin) for 15 min on ice with occasional rocking. Cells were scraped into Eppendorf tubes and centrifuged for 10 min at 12 000 r.p.m. at 4°C. The cleared cell lysates were transferred to new tubes, and protein concentrations were determined using the Bio-Rad protein assay. For standard binding experiments, lysate corresponding to 750–1000 μg of total protein in approximately 1 ml was added to 10 μg of GST immobilized on gluthathione agarose beads (Pharmacia). After incubation for 1 h with rocking at 4°C, the beads were removed by centrifugation, and 10 μg of GST fusion protein immobilized on beads was added to the lysates. After incubation for 1 h with rocking at 4°C, the beads were pelleted by centrifugation and washed three times with 1 ml of ice-cold modified RIPA buffer. The beads were resuspended in 2 × Laemmli sample buffer, boiled, and eluted proteins were analysed by SDS–PAGE and immunoblotting. Similar procedures were used for phosphotyrosine competition assays, except that 15 μg of precleared lysate protein in 0.5 ml was incubated with 1 μg of GST fusion protein immobilized on beads. In some cases, before the addition of the GST fusion protein, phosphotyrosine or phosphoserine was added to the lysate to a final concentration of 10, 1 or 0.1 mM. After incubation for 1 h with rocking at 4°C, the beads were washed and analysed as above.
In vitro dephosphorylation assays
Dephosphorylation reactions using p-nitrophenyl phosphate as a substrate were performed as described previously (Guan et al., 1991). Dephosphorylation reactions using cell lysates were performed essentially as described by Garton et al. (1996), except that REF-52 cells were used as the source of tyrosine-phosphorylated proteins, and pervanadate treatment was omitted. An aliquot of 280 μl of lysate (0.5 mg protein ml−1) was mixed with 70 μl of diluted enzyme (≈1.0 pmol of GST-YOPH or GST-YOPHΔ170) and incubated on ice. At several time points ranging from 0.5 min to 30 min, 50 μl aliquots were removed, and the reaction was terminated by the addition of an equal volume of 2 × Laemmli sample buffer. In one reaction, enzyme dilution buffer (50 mM imidazole, pH 7.5) was added to the lysate instead of enzyme and incubated for 30 min as a control for endogenous phosphatase activity. Dephosphorylation of proteins in the lysates was measured by immunoblotting with the 4G10 antibody.
Samples were separated by SDS–PAGE under reducing conditions and transferred electrophoretically to nitrocellulose. The nitrocellulose filters were blocked in TBST (50 mM Tris, pH 8.0, 137 mM NaCl, 2.7 mM KCl, 0.05% Tween 20) containing 1% BSA for 1 h. Filters were then incubated with the specified primary antibody, either anti-paxillin (1:1000), 4G10 (1:1000), anti-p130Cas (1:1000) or anti-Fak (1:100), for 1 h. Filters were rinsed and then incubated with the appropriate secondary antibody for 1 h. The filters were washed and developed using the Renaissance (DuPont NEN) chemiluminescence system. In some cases, the blots were stripped of bound antibodies by incubation in 62.5 mM Tris-HCl, pH 6.8, 2% SDS, 100 mM 2-mercaptoethanol at 50–55°C for 30 min.
Hela cells plated in 100 mM dishes were infected with IP17/yopHC403SM45 for 120 min at an MOI of 50:1. Immunoprecipitations were performed as described above using 20 μl of M45 antibody. The immune complex was washed once in 1 ml of 50 mM imidazole (pH 7.3), divided into two tubes, and each tube was boiled for 2 min. After the tubes had cooled to room temperature, purified Yop51 (400 U; New England BioLabs) was added to one of the tubes, and both tubes were then incubated for 10 min at room temperature. After the addition of Laemmli sample buffer, the tubes were boiled for 5 min. The immune complexes were subjected to SDS–PAGE and transferred to Immobilon-P transfer membrane (Millipore). The membrane was blocked with PBS–1% BSA and then probed with 5 μg ml−1 purified GST–YOPH1–130 for 1 h at room temperature. The membrane was washed with PBS–1% BSA–0.1% Tween 20 (PBSTB) and then incubated in TBST containing anti-GST antibody (1 μg ml−1) for 1 h. After washing in PBSTB, the membrane was developed with HRP-conjugated anti-mouse IgG as described above.
We thank L. Palmer and T. Miller for helpful discussions and comments on the manuscript. We are indebted to P. Hearing for providing the M45 antibody. This work was supported by grants from the National Institutes of Health (AI35175), the Sinsheimer Foundation and the Catacasinos Young Investigator Award. J.B.B. is a PEW scholar in the biomedical sciences.