Identification of Residues in the N-terminal Domain of theYersinia Tyrosine Phosphatase That Are Critical for Substrate Recognition*

YopH is a 468-amino acid protein-tyrosine phosphatase that is produced by pathogenic Yersiniaspecies. YopH is translocated into host mammalian cells via a type III protein secretion system. Translocation of YopH into human epithelial cells results in dephosphorylation of p130Cas and paxillin, disruption of focal adhesions, and inhibition of integrin-mediated bacterial phagocytosis. Previous studies have shown that the N-terminal 129 amino acids of YopH comprise a bifunctional domain. This domain binds to the SycH chaperone in Yersinia to orchestrate translocation and to tyrosine-phosphorylated target proteins in host cells to mediate substrate recognition. We used random mutagenesis in combination with the yeast two-hybrid system to identify residues in the YopH N-terminal domain that are involved in substrate-binding activity. Four single codon changes (Q11R, V31G, A33D, and N34D) were identified that interfered with binding of the YopH N-terminal domain to tyrosine-phosphorylated p130Cas but not to SycH. These mutations did not impair YopH translocation into HeLa cells infected with Yersinia pseudotuberculosis. Introduction of the V31G substitution into catalytically inactive (substrate-trapping) forms of YopH interfered with the ability of these proteins to bind to p130Cas and to localize to focal adhesions in HeLa cells. In addition, the V31G substitution reduced the ability of catalytically active YopH to dephosphorylate target proteins in HeLa cells. These data indicate that the substrate- and SycH-binding activities of the YopH N-terminal domain can be separated and that the former activity is important for recognition and dephosphorylation of substrates by YopHin vivo.

Three Yersinia species (Yersinia pestis, Yersinia pseudotuberculosis, and Yersinia enterocolitica) are highly pathogenic for humans. All three harbor a related 70-kilobase pair plasmid that is essential for virulence (1). Encoded on this plasmid is a type III protein secretion system that functions to translocate a set of protein toxins termed Yops into infected eukaryotic cells. Yops function to prevent phagocytosis, superoxide production, and cytokine synthesis by professional phagocytes and other types of host cells (2)(3)(4)(5). YopT, YopE, and YopH target key proteins that regulate the host cell actin cytoskeleton. YopT modifies and inactivates the small GTP-binding protein RhoA (6). YopE is a GTPase-activating protein for RhoA, Rac1, and Cdc42 (7,8). YopH is a protein-tyrosine phosphatase (PTP) 1 (9) that dephosphorylates multiple focal adhesion proteins (10 - 14).
The 468-amino acid YopH protein appears to be composed of two distinct modular domains. Residues 206 -468 comprise the C-terminal PTP catalytic domain (9). The PTP activity of this domain is essential for the antiphagocytic function of YopH and Yersinia virulence (15,16). Residues 403-410 form a phosphate-binding loop (P-loop) within the active site (17). Substitution of the nucleophilic Cys at position 403 with either Ser (C403S) or Ala (C403A) has been shown to inactivate the enzyme (9). Catalytically inactive forms of YopH can form stable complexes with substrates in vivo ("substrate trapping") (18) and localize to focal adhesion complexes in infected cultured cells (10,12). The focal adhesion proteins p130 Cas (Cas) (12) (10), focal adhesion kinase (12), and paxillin (11) have been identified as substrates of YopH in cultured human epithelial cells. A region within the PTP domain (residues 223-226) has been shown to be important for targeting of YopH to focal adhesion complexes (19).
The N-terminal 129 residues of YopH comprise a second modular domain that is bifunctional. This domain binds to the SycH chaperone in Yersinia to orchestrate type III-mediated translocation of YopH into host cells (20). A binding site for SycH has been localized between residues 20 and 69 (20). The N-terminal domain also binds to Cas and paxillin in vitro in a phosphotyrosine-dependent manner (11). The efficiency of substrate dephosphorylation by YopH in vitro is diminished by removal of the N-terminal domain, suggesting that it is important for substrate recognition (11). As a first step toward elucidating the mechanism of substrate recognition mediated by the YopH N-terminal domain, we have identified several residues that are critical for this activity.

EXPERIMENTAL PROCEDURES
Construction of Plasmids-The plasmid pLP15 (21) contains a DNA fragment coding for YopH fused to a C-terminal M45 epitope tag inserted between the BamHI and EcoRI sites of pGEX-2T. Here, pLP15 is designated pGEX2T-YopHM45. A NdeI site overlaps the initiation codon of the YopH reading frame, and an XbaI site is present at the point of fusion between the YopH and M45 sequences in pGEX2T-YopHM45. The plasmid pGEX-YopH1-129M45 was constructed as follows. A DNA fragment encoding the first 129 residues of YopH flanked by 5Ј and 3Ј restriction sites was synthesized by PCR. The PCR was performed using pYopH (22) as template and PTP18 (11) and PTP23 * This research was supported by National Institutes of Health Grants AI35175 and AI3389 (to J. B. B.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ Present address: Nathan S. Kline Inst. for Psychiatric Research, Orangeburg, NY 10962.
(5Ј-GATCCCGGGACCCCCTCGCTCCCGACTCTTG-3Ј) as forward and reverse primers, respectively. The primers were designed to incorporate BamHI and NdeI restriction sites into the 5Ј end of the product and an XbaI site into the 3Ј end. The PCR product was digested with BamHI and XbaI. The pGEX2T-YopHM45 vector was digested with BamHI and XbaI, and the region coding for full-length YopH was removed and replaced with the PCR product encoding YopH1-129. This resulted in a translational fusion between the codons specifying residues 1-129 of YopH and the 12-amino acid C-terminal M45 epitope tag. The structure of pGEX-YopH1-129M45 was verified by sequencing.
The plasmids pBDMI and pADMI were derived from the yeast twohybrid vectors pGBT9 and pGAD424 (27), respectively, by the insertion of new polylinker regions. The polylinker regions of pGBT9 or pGAD424 were removed by digestion with EcoRI and SalI. Two complementary oligonucleotides, MIsiteF (5Ј-AATTGGGATCCCCGGGAATTCGCGGC-CGCG-3Ј) and MIsiteR (5Ј-CCCTAGGGGCCCTTAAGCGCCGGCG-CAGCT-3Ј), were annealed and inserted between the EcoRI and SalI sites in pGBT9 and pGAD424. The polylinker regions of pBDMI and pADMI were confirmed by sequencing. A segment of DNA coding for YopH1-129M45 was removed from pGEX-YopH1-129M45 by digestion with BamHI and EcoRI. This DNA was inserted between the BamHI and EcoRI sites of pBDMI, yielding pBD-YopH1-129M45. The plasmid pAD-Src was constructed from pADMI by the insertion of a DNA fragment that encodes a mutant form of c-Src (Y416F/Y527F) (23). The DNA fragment encoding c-Src was removed from pBTM116 (23) by digestion with BamHI and inserted into the unique BglII site in pADMI. A DNA fragment encoding Cas was removed from pEBG-p130Cas (24) by digestion with BamHI and NotI and inserted between the BamHI and NotI sites of pADMI or pAD-Src, yielding pAD-Cas and pAD-CasϩSrc, respectively.
Codon substitutions isolated in the two-hybrid assay (see below) were introduced into the full-length YopHM45 reading frame by restriction fragment subcloning. DNA fragments containing the various codon substitutions were removed from pBD-YopH1-129M45 by digestion with NdeI and SnaBI and substituted for the corresponding NdeI-SnaBI fragment of pGEX2T-YopHM45. A standard restriction fragment subcloning procedure was used to combine the V31G mutation with either the C403S or the R409A substitutions. DNA fragments coding for full-length YopH proteins with the various codon substitutions were removed from the pGEX-2T backbone by digestion with NdeI and EcoRI and inserted between the NdeI and EcoRI sites in the plasmid pPROH for expression in Y. pseudotuberculosis. The plasmid pPROH was generated by the insertion of the yopH promoter region into the polylinker region of pMMB67HE (25). The yopH promoter region was amplified by PCR using virulence plasmid pIB1 as template and the primers PTP26 (5Ј-CGGATCCGCTGCGCGATGTACTGAC-CCG-3Ј) and PTP27 (5Ј-TTACATTAGGAATTCATATGTCCCTCCT-TAATTAAATACACGCC-3Ј). The primers were designed to incorporate a BamHI site into the 5Ј end of the product and NdeI and EcoRI sites into the 3Ј end. The PCR product was digested with BamHI and EcoRI, inserted between the BamHI and EcoRI sites of pMMB67HE, and verified by sequencing.
Identification of Codon Changes in the N-terminal Domain of YopH That Interfere with Binding to Cas in the Two-hybrid System-The DNA sequence coding for the first 129 residues of YopH was subjected to random mutagenesis using a PCR under suboptimal conditions (26). The PCR was performed using pBD-YopH1-129M45 as template and the oligonucleotides 5ЈPBD2H (5Ј-CCGTCACAGATAGATTGGCT-TCAGTGG-3Ј) and 3ЈPBD2H (5Ј-CCTGAGAAAGCAACCTGACCTA-CAGGA-3Ј) as forward and reverse primers, respectively. The resultant PCR product contained the YopH1-129M45 coding region flanked on either side by ϳ100 base pairs of sequence derived from pBDMI. The PCR product was introduced into pBDMI via homologous recombination in yeast cells. For this purpose, pBDMI was digested with BamHI, and the linear plasmid was mixed with the PCR product and supercoiled pAD-CasϩSrc. The mixture was used to transform the yeast two-hybrid reporter strain Y153 to Trp and Leu prototrophy on synthetic dextrose media lacking Trp and Leu (SD-Trp-Leu) (27).
Mutations that interfered with binding of the YopH N-terminal domain to tyrosine-phosphorylated Cas were identified by screening ϳ1500 Trpϩ Leuϩ colonies of Y153 on filters for reduced expression of ␤-galactosidase activity (27). Colonies that displayed reduced ␤-galactosidase activity were tested for the production of BD-YopH1-129M45 fusion protein by immunoblotting. Yeast cell lysates were prepared from 25 ml of culture grown to an A 600 value of ϳ0.68 (equivalent to 2.5 ϫ 10 8 cells) in SD-Trp-Leu media. After centrifugation the pellet was resuspended in 50 l of ice-cold ddH 2 O, 250 l of glass beads, and 200 l of 1ϫ Laemmli sample buffer supplemented with 10% ␤-mer-captoethanol. Samples were vortexed at high speed for 1 min and then chilled on ice for 30 s. This procedure was repeated twice. Samples were boiled for 3 min and briefly centrifuged prior to loading. Samples of 10 l were separated on 12% SDS polyacrylamide gels and analyzed by immunoblotting using anti-M45 (see below). The two-hybrid assay was repeated with reconstructed binding domain vectors to verify that reduced expression of ␤-galactosidase was specific to mutations in the YopH1-129M45 coding region. Plasmid DNA was isolated from 53 Y153 colonies that produced full-length BD-YopH1-129M45 protein. The plasmid DNA was used to transform Escherichia coli strain MH4 to Leu prototrophy to select for the LEU2 gene carried by the pBD-YopH1-129M45 plasmid (27). Plasmid DNA was isolated from Leuϩ colonies of MH4 and digested with BamHI and EcoRI to remove the YopH1-129M45 coding regions. The regions coding for mutant YopH1-129M45 proteins were inserted between the BamHI and EcoRI sites of pBDMI, and the resulting plasmids were introduced into Y153 along with the pAD-CasϩSrc plasmid. Colonies of Leuϩ Trpϩ transformants on filters were tested for expression of ␤-galactosidase as described above. This procedure was repeated using pAD-SycH in place of pAD-CasϩSrc to identify mutations that selectively interfered with binding of the YopH N-terminal domain to Cas. The regions coding for YopH1-129M45 in purified pBD-YopH1-129M45 plasmids were sequenced to identify mutations that resulted in reduced binding to Cas. The G41S and D106G codon substitutions were separated by a standard restriction fragment subcloning procedure.
␤-galactosidase activity in Y153 strains containing mutant pBD-YopH1-129M45 plasmids was quantitated using a colorimetric liquid assay as described previously (27). Cultures of Y153 strains containing mutant pBDMI-YopH1-129M45 plasmids and pAD-CasϩSrc were grown in selective media (SD-Trp-Leu) to an A 600 value of ϳ1.0. Yeast cells in 5 ml of culture were assayed for ␤-galactosidase activity using chlorophenol red-␤-D-galactopyranoside as described (27). Reactions were incubated for 35 min. Measurement of ␤-galactosidase activity in yeast cells containing mutant pBDMI-YopH1-129M45 plasmids and pAD-SycH was performed using cells derived from 0.1 ml of culture and the substrate O-nitrophenyl-␤-D-galactopyranoside (27). Reactions were incubated for 15 min. Units of ␤-galactosidase activity were calculated as described (27).
Antibodies-Hybridoma supernatant containing monoclonal anti-M45 antibody was provided by Dr. P. Hearing (State University of New York at Stony Brook). Anti-M45 recognizes the 12-amino acid epitope SRDRLPPFETET. Anti-M45 was purified from the supernatant using protein A-Sepharose as described (28). Anti-M45 was used at a dilution of 1:1000 for immunoblotting and at a final concentration of 0.5 g/ml for immunofluorescence labeling. Immunoprecipitations were performed with 1.0 g of M45 per sample. Monoclonal anti-phosphotyrosine (pTyr) antibody 4G10 was purchased from Upstate Biotechnology. The anti-pTyr antibody was used at a dilution of 1:1000 for immunoblotting and at final concentration of 5 g/ml for immunofluorescence labeling. Monoclonal anti-Cas (P27820) was purchased from Transduction Laboratories and used at a dilution of 1:1000 for immunoblotting. Immunoprecipitations were performed with 1.0 g of anti-Cas per sample. The rabbit anti-YopH antibody SB360 was prepared in a commercial facility (Calico Biologicals, Inc.) using a purified glutathione S-transferase-YopH fusion protein as antigen. SB360 was used at a dilution of 1:1000 for immunoblotting. Anti-mouse and anti-rabbit IgG conjugated to horseradish peroxidase were purchased from Sigma and used at dilutions of 1:1000 or 1:15,000, respectively. Tetramethyl rhodamine isothiocyanate (TRITC)-conjugated F(abЈ) 2 goat anti-mouse secondary antibody was purchased from Jackson ImmunoResearch Laboratories, Inc. and used at a dilution of 1:250.
Bacterial and HeLa Cell Cultures and Infection Conditions-The Y. pseudotuberculosis serogroup III strains YP17 (yopHyopE) and YP19 (yopHyopEyopB) and their growth conditions have been described previously (21). Both strains carry a naturally occurring deletion in the plasmid-borne yopT gene and are devoid of YopT activity. 2 Expression plasmids derived from pPROH that produce wild-type or mutant YopH proteins were introduced into these strains from E. coli by conjugation (10). For infection assays, bacteria were grown overnight at 26°C with shaking in Luria broth containing 100 g/ml ampicillin. Bacteria were subcultured into Luria broth containing 100 g/ml ampicillin and 2.5 mM CaCl 2 to an A 600 value of 0.1. Cultures were shaken at 37°C for 2 h. Bacteria were pelleted by centrifugation and resuspended in warm (37°C) Hanks' balanced salt solution to an A 600 value of 1.0 (ϳ1 ϫ 10 9 colony-forming units/ml).
HeLa cells were cultured in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% heat-inactivated fetal calf serum (Life Technologies, Inc.) and 1 mM sodium pyruvate in a 5% CO 2 humidified incubator at 37°C. For the immunofluorescence experiments, 1 ϫ 10 5 HeLa cells were seeded in 1 ml of media onto sterile glass coverslips placed in a 24-well tissue culture plate and cultured overnight. For the translocation, immunoprecipitation, and in vivo dephosphorylation assays, 2 ϫ 10 6 HeLa cells in 10 ml of media were seeded into 100-mm tissue culture dishes and cultured overnight. HeLa cells were overlaid with fresh media 30 min prior to bacterial infection. Cells were left uninfected or infected with bacteria grown as described above at a multiplicity of infection of 50:1 at 37°C in a 5% CO 2 incubator.
Immunofluorescence Assays-All steps following a 2-h infection (see above) were performed at room temperature. Coverslips were washed twice with phosphate-buffered saline (PBS) containing 1 mM Na 3 VO 4 , fixed with 4% paraformaldehyde for 10 min, and then permeabilized with 0.2% Triton X-100 for 10 min. Coverslips were washed twice with PBS containing 1% bovine serum albumin (BSA) and then incubated for 1 h with primary antibody (anti-pTyr or anti-M45) diluted in PBS containing 3% BSA. Coverslips were washed with PBS and then incubated for 1 h with TRITC-conjugated secondary antibody diluted in PBS containing 3% BSA. Coverslips were washed well with PBS before mounting on slides in 10% Airvol (Air Products, Inc.), 100 mM Tris, pH 8.5, 25% glycerol, and 2.5% DABCO anti-fade (Sigma). The stained cell samples were analyzed by epifluorescence microscopy using a Zeiss AxioPlan 2 microscope equipped with a 100ϫ Plan-NeoFluar (numerical aperture 1.4) oil immersion objective. Epifluorescent images were captured with a Spot CCD camera and Adobe Photoshop, version 5.5 software running on a Macintosh G4 computer.
Translocation Assay-After a 2-h infection, the dishes were placed on ice and washed twice with 10 ml of ice-cold PBS. Cells were lysed in 0.5 ml of Nonidet P-40 lysis buffer (150 mM NaCl, 50 mM Tris, pH 8.0, 1% Nonidet P-40, 1 mM Na 3 VO 4 , and 10 mM NaF) for 15 min on ice with occasional rocking. Cells were scraped into microcentrifuge tubes and centrifuged for 10 min at 12,000 ϫ g at 4°C. The supernatants were transferred to new tubes, and protein concentrations were determined using the Bio-Rad protein assay. Samples of 20 l containing 5 g of total cell protein each were separated on 10% SDS polyacrylamide gels and analyzed by immunoblotting using anti-M45.
Immunoprecipitation Assay-After 2 h of infection the dishes were placed on ice and washed twice with 10 ml of ice-cold PBS containing 1 mM Na 3 VO 4 . Cells were lysed in 0.5 ml of ice-cold Triton X-100 lysis buffer (10 mM Tris, pH 7.6, 150 mM NaCl, 5 mM EDTA, 10% glycerol, 1% Triton X-100, 1 mM Na 3 VO 4 , 10 mM NaF, 200 M 4-(2-aminoethyl)benzenesulfonyl fluoride hydrochloride, 20 M leupeptin, and 1 M pepstatin) for 15 min on ice with occasional rocking. Cells were scraped into microcentrifuge tubes and centrifuged for 10 min at 12,000 ϫ g at 4°C. The supernatants were transferred to new tubes, and protein concentrations were determined using a Bio-Rad protein assay. Cell lysates were adjusted to contain 450 g of protein in a volume of 1 ml. 50 l of a 50% suspension of protein A-Sepharose beads (Amersham Pharmacia Biotech) was added to each lysate sample, and the tubes were incubated for 30 min at 4°C with rotation as a pre-clearing step. After the protein A-Sepharose beads were removed by centrifugation, the supernatants were transferred to new tubes and mixed with anti-M45. The tubes were incubated for 3 h at 4°C with rotation. Immune complexes were recovered by addition of 50 l of 50% protein A-Sepharose, followed by incubation for 3 h at 4°C with rotation. The beads were pelleted by centrifugation, washed three times with 1 ml of 4°C lysis buffer, resuspended in 60 l of 1ϫ Laemmli sample buffer, and boiled for 5 min. Immunoprecipitated proteins were separated on 7.5% SDS polyacrylamide gels under reducing conditions and analyzed by immunoblotting with anti-Cas, anti-pTyr, or anti-YopH antibodies as described below.
In Vivo Dephosphorylation Assay-HeLa cells were left uninfected or infected for 15, 30, 60, or 120 min. Dishes were placed on ice and washed twice with 10 ml of ice-cold PBS containing 1 mM Na 3 VO 4 . Cells were lysed in 0.5 ml of ice-cold Triton X-100 lysis buffer for 15 min on ice with occasional rocking. Cells were scraped into microcentrifuge tubes and centrifuged for 10 min at 12,000 ϫ g at 4°C. The supernatants were transferred to new tubes, and protein concentrations were determined using the Bio-Rad protein assay. Supernatant samples containing 50 g of protein in a volume of 20 l were mixed with an equal volume of 2ϫ Laemmli sample buffer and boiled for 3 min. The resulting samples were separated on 7.5% SDS polyacrylamide gels under reducing conditions and analyzed by immunoblotting with anti-pTyr or anti-YopH as described below.
Immunoblotting-Proteins separated in SDS polyacrylamide gels were electrophoretically transferred to nitrocellulose filters (Schleicher & Schü ll). Unless indicated all subsequent steps were performed at room temperature. The nitrocellulose filters were blocked in Tris-buffered saline containing 0.05% Tween 20 (TBST) and 1% BSA for 1 h. Filters were incubated with primary antibody (see above) diluted in TBST for 1 h. Filters were washed four times in TBST and then incubated for 1 h with the appropriate secondary antibody diluted (1:1000 for anti-mouse and 1:15,000 for anti-rabbit) in TBST. The filters were washed four times in TBST and developed using the Renaissance (PerkinElmer Life Sciences) 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. After the filter was extensively washed in TBST, the immunoblotting procedure was then repeated starting with the blocking step.

Interaction of the YopH N-terminal Domain with Tyrosinephosphorylated Cas in the Yeast Two-hybrid
System-A modified yeast two-hybrid system (23) was used to detect association of the YopH N-terminal domain with tyrosine-phosphorylated Cas. A yeast strain carrying a lacZ reporter gene under control of the Gal4p transcription factor was transformed with two expression vectors. One vector produced the DNA binding domain of Gal4p fused to residues 1 to 129 of YopH adjoined to a C-terminal M45 epitope tag (BD-YopH1-129M45). The second vector produced a modified form of the Src tyrosine kinase and the activation domain of Gal4p fused to Cas (AD-Cas). Based on previous studies (23), ectopic expression of Src was expected to result in tyrosine phosphorylation of the AD-Cas fusion protein in yeast cells. ␤-galactosidase activity was detected when both fusion proteins and Src were produced in the yeast reporter strain (Fig. 1A). In contrast, ␤-galactosidase activity was not detected when both fusion proteins were expressed in yeast cells in the absence of Src (data not shown). Thus, a specific two-hybrid interaction between the N-terminal domain of YopH and tyrosine-phosphorylated Cas was detected.

Identification of Amino Acid Substitutions in the YopH Nterminal Domain That
Interfere with Binding to Tyrosine-phosphorylated Cas-Random mutagenesis was used to identify amino acid substitutions in the YopH N-terminal domain that interfere with binding to tyrosine-phosphorylated Cas in the two-hybrid system. The DNA sequence encoding the first 129 residues of YopH was amplified using a PCR and Taq DNA polymerase under suboptimal conditions (26). The resulting DNA product was inserted into a Gal4p binding domain fusion vector in a yeast reporter strain using in vivo recombination. Approximately 1500 yeast colonies were screened on filters for reduced ␤-galactosidase activity. 43 colonies that showed reduced ␤-galactosidase activity were chosen for further characterization (see "Experimental Procedures"). Five of these colonies produced intact BD-YopH1-129M45 protein that interacted weakly with tyrosine-phosphorylated Cas in the two-hybrid system. The binding domain vectors were isolated from these colonies, and the regions coding for residues 1-129 of YopH were sequenced. Four plasmids contained single codon changes in the YopH coding region (corresponding to Q11R, V31G, A33D, and N34D). One plasmid contained two codon changes (G41S and D106G) that were subsequently separated by restriction fragment subcloning. As shown in Fig. 1A, five of the six single amino acid substitutions (the exception being D106G) resulted in reduced levels of ␤-galactosidase activity in the two-hybrid assay.
We tested each of the mutant BD-YopH1-129 proteins by two-hybrid assay for interaction with an AD-SycH fusion protein to determine whether any of the amino acid substitutions interfered with binding to the chaperone. With the exception of the Q11R mutation, all of the substitutions resulted in higher levels of ␤-galactosidase activity as compared with the wild-type (Fig. 1B). Lysates of the yeast strains producing the mutant BD-YopH1-129M45 proteins were analyzed by immunoblotting with the M45 antibody to examine levels of protein expression. As shown in Fig. 2, all of the mutant fusion proteins were produced at approximately the same level as the wildtype fusion protein. These results suggested that the V31G, A33D, N34D, G41S, and D106G substitutions actually resulted in tighter binding of the YopH N-terminal domain to SycH.

Amino Acid Substitutions in the YopH N-terminal Domain That Reduce Binding to Cas Do Not Interfere with YopH Translocation into HeLa Cells-We next examined the effect of the substitutions in the N-terminal domain on translocation of
YopH into HeLa cells. The Q11R, V31G, A33D, and N34D codon changes were inserted into the full-length yopH gene carried on a bacterial expression vector (pPYopHM45), and the resulting plasmids were introduced into a yopEyopHyopT mutant Y. pseudotuberculosis strain (YP17). HeLa cells were infected with the bacterial strains for 2 h and lysed in 1% Nonidet P-40. Soluble fractions of the lysates were prepared by centrifugation. Samples of the soluble fractions were analyzed by immunoblotting with the M45 antibody to measure the amount of translocated YopH protein in the soluble fractions. As seen in Fig. 3, YopH proteins containing the Q11R, V31G, A33D, and N34D substitutions were detected in the soluble fractions at levels comparable with wild-type YopH. YopH was not detected in the soluble fraction when HeLa cells were infected with a yopB mutant of YP17 (YP19/pPYopHM45) that is defective for Yop translocation (lane 1). These results indicated that YopH proteins containing the Q11R, V31G, A33D, and N34D substi-tutions are not defective for translocation into HeLa cells. Similar results were obtained when translocation of mutant YopH proteins was analyzed by immunofluorescence microscopy (data not shown; see Fig. 4, A and B). Thus, amino acid substitutions in the YopH N-terminal domain that interfere with binding to Cas do not interfere with YopH translocation.
The N-terminal of YopH Domain Is Required for Efficient Substrate Trapping in Vivo-The V31G substitution was engineered into a catalytically inactive form of YopHM45 (YopHC403SM45) to determine whether the N-terminal domain is important for substrate trapping in vivo. The V31G mutation was selected for this purpose, because it strongly interfered with substrate-binding activity in the two-hybrid system (Fig. 1A). A plasmid encoding the mutant protein was introduced into Y. pseudotuberculosis YP17, and the resulting strain was used to infect HeLa cells. Initially, the HeLa cells were processed for immunofluorescence microscopy using the M45 antibody to examine the effect of the V31G substitution on localization of inactive YopH to focal adhesions. Analysis by epifluorescence microscopy showed that the V31G substitution strongly reduced, but did not eliminate, localization of inactive YopH to focal adhesions (Fig. 4, compare panels C and E). Infected HeLa cells were also labeled with an anti-pTyr antibody to demonstrate that focal adhesions were not disrupted under the conditions of the infection (Fig. 4G).
Detergent lysates of the infected HeLa cells were prepared and subjected to immunoprecipitation with M45 antibody to determine the amount of Cas that was directly bound to the mutant YopH proteins. The immune complexes were analyzed by immunoblotting with either anti-Cas antibody (Fig. 5A) or anti-pTyr antibody (Fig. 5B). The results obtained with the anti-pTyr antibody were more informative because of higher sensitivity. The V31G substitution had a dramatic inhibitory effect on binding of Cas to inactive YopH (Fig. 5B, compare  lanes 2 and 3). However, the V31G substitution did not abolish coprecipitation of Cas with inactive YopH (Fig. 5B, lane 3), suggesting that the PTP domain contributes, as well, to substrate-trapping activity. The filter analyzed in Fig. 5B was stripped of the anti-pTyr antibody and analyzed by immunoblotting with anti-YopH as a control for recovery of inactive YopH protein in the immunoprecipitation (Fig. 5C).
To determine whether Cas was binding to the P-loop in the PTP domain, the V31G substitution was introduced into a catalytically inactive form of YopH in which Arg-409 in the P-loop motif was changed to Ala (R409A). Arg-409 appears to stabilize the transition state of the enzyme and is thus critical for YopH catalytic activity (29). In addition, Arg-409 plays an important role in substrate binding (29), and therefore the R409A substitution was predicted to interfere with binding of tyrosine-phosphorylated Cas to the P-loop. The R409A substitution by itself did not interfere with localization of YopH to focal adhesions (Fig. 4D) or with binding of YopH to Cas (Fig.  5B, lane 4), indicating that the N-terminal domain is the major determinant of substrate binding in vivo. When the V31G and the R409A substitutions were combined, localization of YopH to focal adhesions was abolished (Fig. 4F) and Cas-binding activity was further decreased but not eliminated (Fig. 5B, lane  5). These results indicated that the P-loop containing the C403S substitution does bind tyrosine-phosphorylated Cas in vivo but that this substrate-binding activity is substantially weaker than that of the N-terminal domain.
The Substrate-binding Activity of the YopH N-terminal Domain Is Required for Efficient Substrate Dephosphorylation in Vivo-We next examined the effect of the V31G substitution on dephosphorylation of substrates by YopH in infected HeLa cells. Lysates prepared from HeLa cells infected for 15, 30, 60, or 120 min with YP17 strains producing either YopH or YopHV31G were analyzed by immunoblotting with anti-pTyr antibody. Two heavily phosphorylated protein bands, of 130 and 68 kDa, were detected in lysates of uninfected HeLa cells (Fig. 6A, lanes 1 and 6). The 68-kDa band has previously been shown to correspond to paxillin (11). The efficiency of paxillin dephosphorylation over time by wild-type YopHM45 was greater than that observed for YopHV31GM45 (Fig. 6A, compare lanes 2 and 7). The filter analyzed in Fig. 6A was stripped of the anti-pTyr antibody and analyzed by immunoblotting with anti-YopH. The results indicated that differences in levels of YopH protein production could not account for differences in rates of substrate dephosphorylation (Fig. 6B). Because Cas comigrates with focal adhesion kinase, Cas was immunoprecipitated from HeLa cell lysates and analyzed by anti-pTyr immunoblotting to directly monitor its rate of dephosphorylation. As shown in Fig. 7, the efficiency of Cas dephosphorylation was greater for wild-type YopH than for YopHV31G (com- pare lanes 2 and 7). Analysis of the same filter with anti-Cas antibodies confirmed that equivalent amounts of Cas were recovered from each of the lysate samples. These data indicate that the substrate-binding activity of the YopH N-terminal domain is required for efficient dephosphorylation of substrates in vivo.

DISCUSSION
The goal of this study was to identify residues in the YopH N-terminal domain that are important for substrate-binding activity but not SycH-binding activity. We used random mutagenesis to isolate single amino acid changes in the N-terminal domain that interfere with binding to tyrosine-phosphorylated Cas in a yeast two-hybrid system. Four single codon changes (Q11R, V31G, A33D, and N34D) and one double codon change (G41S and D106G) were identified in the screen. Separation of the G41S and D106G mutations indicated that the G41S change was largely responsible for the mutant phenotype.
It has been determined by truncation experiments that a region between residues 20 and 69 of YopH is critical for binding to SycH (30,31). All of the codon changes that resulted in reduced binding of the YopH N-terminal domain to Cas, except for Q11R, fell within this putative binding site for SycH. This may indicate that tyrosine-phosphorylated proteins and SycH bind to an overlapping region of the YopH N-terminal domain. Surprisingly, all of the single codon changes, except for Q11R, appeared to result in tighter binding of SycH to the YopH N-terminal domain in the two-hybrid system. Binding experiments carried out in vitro with purified components will be required to confirm that SycH binds tighter to the mutant N-terminal domain, because other variables inherent to the two-hybrid system could also result in increased expression of ␤-galactosidase activity. YopH proteins containing amino acid substitutions in the N-terminal domain were translocated into HeLa cells as efficiently as the wild-type protein, as determined by cellular fractionation experiments and immunofluorescence microscopy. Because binding of SycH to YopH is required for translocation (32), we conclude from these experiments that it is possible to separate the substrate-binding activity of the N-terminal domain from its SycH-binding activity.
It has recently been reported that the Yersinia yop virulon regulators LcrQ/YscM1 and YscM2, which share significant sequence similarity with the YopH N-terminal domain, also bind to SycH (33). By sequence alignment of YopH1-129 with LcrQ and YscM1, it is possible to predict which residues are involved in SycH-binding activity. None of the residues identified in our screen as being critical for substrate-binding activity in the YopH N-terminal domain are conserved among the three proteins. This observation further suggests that distinct residues in the YopH N-terminal domain mediate interaction with tyrosine-phosphorylated substrates or SycH.
The three-dimensional structure of the YopH N-terminal domain has recently been determined to a resolution of 2.2 Å. 3 It is a single, highly compact domain composed of 4 ␣-helices sandwiched between one two-stranded ␤-sheet and one threestranded ␤-sheet. The fold of the YopH N-terminal domain is unlike that of other known phosphotyrosine binding domains such as the SH2 domain of Src (34) or the phosphotyrosine binding domain of insulin receptor substrate-1 (35). Q11, V31, A33, N34, and G41 are located on the surface along a loop connecting ␤-strand 1 and ␤-strand 2. This region is rich in positively charged residues (e.g. K26, R28, K35, and R49) and is flanked on either side by pockets, which may together form a shallow phosphotyrosine-binding cleft.
The role of the N-terminal domain in substrate recognition in vivo was addressed by the introduction of the V31G substitution into the full-length YopH reading frame. Initially, the V31G substitution was introduced into two different catalytically inactive forms of YopH, one in which the P-loop in the PTP domain retained the capacity to bind substrate (C403S) and the other in which binding of substrate to the P-loop is blocked (R409A). This strategy allowed us to independently assess the contributions of the N-terminal domain and the P-loop to the formation of a stable enzyme-substrate complex. Substratebinding activity was scored by the ability of the mutant proteins to localize to focal adhesions in infected HeLa cells and to coprecipitate with Cas in detergent lysates of infected HeLa cells. These results of both assays indicated that the N-terminal domain is the major determinant of substrate-binding activity in YopH but that the P-loop also contributes, albeit weakly, to substrate trapping. The ability of Cas to coprecipitate with YopH was not completely eliminated by the presence of the V31G and R409A substitutions, suggesting that an additional substrate-binding interface may be present elsewhere in the protein. Persson et al. (19) have reported that part of a surface-exposed loop in the PTP domain (residues 223-226) is 3 C. Smith and M. Saper, personal communication. involved in targeting YopH to focal complexes. However, their data argue that this region functions as a localization sequence and is not involved in substrate recognition (19).
The V31G substitution was introduced into catalytically active YopH to determine whether the N-terminal domain is important for efficient dephosphorylation of substrates by YopH in vivo. The presence of the V31G substitution resulted in a dramatic decrease in the efficiency of Cas and paxillin dephosphorylation by YopH in infected HeLa cells. These results support our original proposal that the N-terminal domain functions to increase the efficiency of substrate recognition by YopH in vivo (11). It is also possible that the N-terminal domain allows YopH to act processively during the dephosphorylation of multiply phosphorylated substrates such as Cas and paxillin.
It is becoming apparent that translocated proteins of other type III secretion systems may be arranged in a modular fashion with at least two, and possible more, distinct effector domains (36 -38). In addition, a single modular domain may in fact be multifunctional, as is the case for the YopH N-terminal domain. This raises the possibility that the N-terminal domains of other translocated type III proteins will perform multiple functions.