Csk homologous kinase (CHK) and ErbB-2 interactions are directly coupled with CHK negative growth regulatory function in breast cancer.

Our previous studies demonstrated that Csk homologous kinase (CHK) acts as a negative growth regulator of human breast cancer through inhibition of ErbB-2/neu-mediated Src family kinase activity (Bougeret, C., Jiang, S., Keydar, I., and Avraham, H. (2001) J. Biol. Chem. 276, 33711-33720. The interaction between the CHK SH2 domain and Tyr(P)(1248) of the ErbB-2 receptor has been shown to be specific and critical for CHK function. In this report, we investigated whether the interaction of the CHK SH2 domain and ErbB-2 is directly related to the inhibition of heregulin-stimulated Src kinase activity. We constructed three CHK SH2 domain binding mutants: G129R (enhanced binding), R147K (inhibited binding), and R147A (disrupted binding). NMR spectra for the domains of each construct were used to evaluate their interaction with a Tyr(P)(1248)-containing ErbB-2 peptide. G129R showed enhanced binding to ErbB-2, whereas binding was completely disrupted by R147A. The enhanced binding mutant showed chemical shift changes at the same residues as wild-type CHK, indicating that this mutant has the same binding characteristics as the wild-type protein. Furthermore, inhibition of heregulin-stimulated Src kinase activity was markedly diminished by R147A, whereas G129R-mediated inhibition was stronger as compared with wild-type CHK. These results indicate that the specific interaction of CHK and ErbB-2 via the SH2 domain of CHK is directly related to the growth inhibitory effects of CHK. These new CHK high affinity binding constructs may serve as good candidates for inhibition of the ErbB-2/Src transduction pathway in gene therapy studies in breast cancer.

The majority of breast carcinomas appear to be sporadic and have a complex accumulation of molecular and cellular abnor-malities that constitute the malignant phenotype (1,2). In many cases, the random onset of breast cancer correlates with overexpression of the ErbB-2/neu receptor and Src tyrosine kinase activity (3,4). Downstream activation by the ErbB-2/neu receptor involves intracellular pathways mediated by Ras/mitogenactivated protein kinase, phosphatidylinositol 3-kinase, and phospholipase C␥; however, the molecular mechanisms of these processes are poorly understood (5). Src tyrosine kinase has been suggested to be a main downstream activator of the ErbB-2/neu receptor because the increased Src kinase activity observed in ErbB-2/neu-induced tumors results from the ability of the Src SH2 1 domain to interact directly with ErbB-2/neu in a phosphotyrosinedependent manner (6,7). Once the ErbB-2/neu receptor is activated by heregulin, it undergoes autophosphorylation at five tyrosine residues located in its non-catalytic carboxyl terminus. The autophosphorylation of ErbB-2/neu can also be induced in the absence of any ligand by high level overexpression of ErbB-2/neu protein (8), as occurs in BT474 or MDA-MB-361 cells. 2 The autophosphorylated tyrosine residues provide docking sites for proteins to connect to intracellular pathways (9,10). The individual target and effect of each phosphotyrosine are not clear, but an add-back mutation study showed that autophosphorylation of tyrosine residues is involved in both the positive and negative effects on ErbB-2/neu-mediated transformation (11). Tyr 1248 of ErbB-2/neu, which is conserved between human and rodent ErbB-2/neu, has been suggested to be the most critical residue for the oncogenicity of the constitutively activated receptor (12). Thus, the study of proteins that bind to Tyr(P) 1248 of ErbB-2/neu is important in elucidating ErbB-2/neu-mediated signaling and function in cancer development.
The Csk homologous kinase (CHK) protein comprises SH3, SH2, and tyrosine kinase domains. Its SH2 domain interacts with Tyr(P) 1248 of the ErbB-2/neu receptor in a ligand-and receptor-specific manner (13). CHK, like Csk, down-regulates Src kinase activity by phosphorylation of the conserved tyrosine residue in the carboxyl terminus of Src-related enzymes in vitro. However, CHK has been suggested to play a specific role as a novel negative growth regulator of human breast cancer on the basis of the following observations. 1) Unlike Csk, which is ubiquitously expressed and cannot associate with ErbB-2, CHK is specifically expressed in primary breast cancer specimens, but not in normal breast tissues (13)(14)(15). CHK expression in normal tissues is restricted to hematopoietic cells and brain (16 -18). 2) CHK binds directly to Tyr(P) 1248 of the ErbB-2/neu receptor kinase upon heregulin stimulation and inhibits Src kinase activity (17). Substantial evidence supports a role for CHK as a negative growth regulator of human breast cancer through inhibition of ErbB-2/neu-mediated Src family kinase activity. Overexpression of CHK in MCF-7 breast cancer cells markedly inhibits the cell growth, transformation, and invasion induced by heregulin and also causes a significant delay of cell entry into mitosis. Furthermore, the tumor growth of wildtype CHK-transfected MCF-7 cells in nude mice is significantly inhibited compared with that of non-transfected MCF-7 cells or cells transfected with kinase-dead CHK (18). The specific expression of CHK in breast cancer tissues and its inhibitory effect on cancer development strongly suggest the potential of the CHK protein as an anticancer drug and a target of gene therapy.
Mechanism-based target identification and structure-based drug design are promising for the development of selective anticancer drugs that would replace conventional cancer chemotherapy and its associated cytotoxic side effects (19). Precise biochemical and structural information on the CHK SH2 domain and the Tyr(P) 1248 containing peptide is necessary to develop CHK as a potential target of breast cancer therapy. We compared the primary sequence of the CHK SH2 domain with those of other SH2 domains (Fig. 1). The structures of a number of other SH2 domains and their complexes with phosphopeptides derived from biological targets were studied by NMR and x-ray crystallography. All contain similar secondary structural elements ( Fig. 1) (20 -22). The structures reveal general Tyr(P)-binding sites as well as specificity-determining sites in the SH2 domains. Although the mode of recognition of the cognate phosphopeptides by two types of SH2 domains is different (see "Discussion"), the geometry of the Tyr(P)-binding pocket within each SH2 domain is conserved in the SH2 domain family (23). SH2 domains display positively charged pockets lined with consensus basic residues of Arg (␣A1), Arg (␤B5), His (␤D4), and Lys/Arg (␤D6) (Fig. 1). Extensive interactions have been identified between the sequence C-terminal to Tyr(P) of the peptide and several residues in the ␤D strand, the ␤D-␣B loop, and the ␣B-␤E loop, which play a regulatory role in the specific binding. We constructed three CHK SH2 domain binding mutants in ␣A and ␤B to test the biological significance of their binding in growth inhibitory function. We also per-formed NMR experiments to identify, in the CHK SH2 domain, the binding sites for a phosphotyrosine-containing peptide derived from ErbB-2.
CHK has been suggested to have a specific role in breast cancer and to be a potential target for breast cancer drug development. Mutation of residues to confer modified binding to Tyr(P) 1248 of ErbB-2/neu will elicit functional insights into the binding of CHK to this receptor.
Cell Lines-Three different breast cancer cell lines with various levels of ErbB-2/neu protein expression were used: MCF-7 (normal level expression), T47D (moderate level overexpression), and BT474 (high level overexpression). All three cell lines were obtained from American Type Culture Collection (Manassas, VA). Cells were grown in RPMI 1640 medium (Cellgro, Inc.) supplemented with 10% fetal bovine serum and 3.5 g/ml insulin (Sigma). Prior to stimulation with heregulin, cells were starved overnight in medium containing 1% fetal bovine serum and then incubated for 4 h in serum-free medium.
Peptide Synthesis and Purification-A peptide containing Tyr(P) 1248 of ErbB-2, ENPEpYLGLDV, was synthesized using solid-phase Fmoc (N-(9-fluorenyl)methoxycarbonyl)-based peptide synthesis with an acetylated N terminus and amidated C terminus (Tufts Core Facility, Boston, MA). All peptides were purified by C 18 reverse-phase high performance liquid chromatography, and identities were confirmed using matrix-assisted laser desorption ionization mass spectroscopy.
Construction and Purification of the CHK SH2 Domain-The CHK SH2 domain constructs (residues 116 -217: G129R, R147A, R147K, and wild-type) were subcloned into the pGEX2T vector using the restriction endonuclease sites BamHI and EcoRI. Point mutations were generated by PCR using the QuikChange site-directed mutagenesis system (Stratagene) according to the manufacturer's instructions. Mutants were verified by sequencing. At least three independently generated mutants were tested for each construct. The glutathione S-transferase (GST)fused CHK SH2 domains were expressed in bacteria (BL21(DE3) cells) and purified following published procedures (13,14). Isolated SH2 domains were generated by thrombin cleavage, followed by purification on a benzamidine-Sepharose 6B column (Amersham Biosciences) (13,14). Approximately 10 mg of protein from all constructs were purified from 1-liter cultures in rich medium (LB medium) and 5 mg from culture in minimal medium. Generation of CHK-encoding pIRES2-EGFP Vectors-To investigate the effects of the generated mutants in breast cancer cells, the same mutations were generated in the full-length form of the CHK gene originating from previously described pcDNA3-based constructs (19 -21). The generation and characterization of a CHK mutant lacking kinase activity were described in detail previously (15). All studied forms of the CHK gene were cloned into the pIRES2-EGFP mammalian expression vector (CLONTECH). Expression of wild-type as well as mutant CHK proteins was assessed by transient transfection of 293T cells and by Western blot analysis.
Binding of ErbB-2 to GST Fusion Proteins-T47D or BT474 cells (ϳ5 ϫ 10 6 cells/plate) were starved overnight in medium containing 1% fetal bovine serum, followed by additional starvation in serum-free medium for 4 h at 37°C. The starved T47D or BT474 cells were then stimulated with 20 nM heregulin for 8 min at room temperature. The stimulation was terminated by the addition of ice-cold lysis buffer (0.1% SDS and 1% Triton X-100 in Tris-buffered saline containing 10% glycerol, 1 mM EDTA, 0.5 mM Na 3 VO 4 , and protease inhibitor mixture (Roche Molecular Biochemicals)). Lysates were precleared by centrifugation (14,000 rpm, 15 min) and incubated for 90 min at 4°C with 10 g of GST fusion proteins coupled to glutathione-Sepharose beads. Next, the beads were washed three times with lysis buffer. SDS buffer was then added, and samples were analyzed on an SDS-7% polyacrylamide gel. Proteins were transferred onto Immobilon-PM membranes (Millipore Corp.), and bound proteins were immunoblotted with anti-phospho-HER2/ErbB-2 (Tyr 1248 ) antibody. The blots were developed using the ECL system. NMR Spectroscopy-15 N, 13 C-Double-labeled protein samples or 15 Nlabeled samples were obtained by growing the transformed bacteria in minimal medium containing 15 NH 4 Cl and 13 C-labeled glucose or 15 NH 4 Cl and unlabeled glucose as the sole sources of nitrogen and carbon, respectively (22)(23)(24). Protein was purified following the same procedures as described above, except that the purified proteins were concentrated using Centricon centrifugation filtration units (Millipore Corp.) with a M r 3000 cutoff. The purified proteins were then exchanged into the final NMR sample buffer containing 50 mM phosphate (pH 7.5), 50 mM NaCl, 1 mM EDTA, and 2 mM perdeuterated dithiothreitol (Cambridge Isotope Laboratories, Cambridge, MA). Optimal conditions were predetermined using microdialysis against a variety of buffers (25)(26)(27). NMR experiments were performed on Bruker AMX 500-MHz and Avance 600-MHz spectrometers. Titration of the protein with the dissolved peptide (in the same buffer) was monitored by changes in 15 N-1 H heteronuclear single quantum correlation (HSQC) spectra collected at peptide/protein molar ratios of 0, 0.25, 0.5, 0.75, 0.85, 1.0, 1.25, 1.5, and 2.0. Dissociation constants (K d ) of peptide binding were determined by analyzing the titration data assuming fast exchange and by using CRVFIT (a nonlinear least-squares fitting program obtained from R. Boyko and B. D. Sykes). For backbone assignment of the protein, triple resonance experiments, including HNCA, HN(CO)CA, HNCACB, and HN(CO)CACB (27,28), were performed and 15 N-separated threedimensional nuclear Overhauser effect correlation and total correlation spectra were recorded. The data were processed and analyzed using FELIX 98 (Accelyrs, Inc.).
In Vitro Src Tyrosine Kinase Assay-For this experiment, MCF-7 cells were chosen because of their overexpression of Src kinase. The cells were grown on 100-mm Petri dishes until 80% confluent and then transfected with 10 g of various CHK-pIRES2-EGFP constructs using LipofectAMINE 2000 (Invitrogen) as recommended by the manufacturer. Twenty-four hours after transfection, the cells were starved overnight in medium containing 1% fetal bovine serum, followed by additional starvation in serum-free medium for 4 h at 37°C. The starved MCF-7 cells were then stimulated with 20 nM heregulin for 8 min at room temperature, and total protein extracts were prepared as described above. One milligram of protein was immunoprecipitated using antibodies against Src (Santa Cruz Biotechnology). The immunoprecipitates were washed three times with lysis buffer and then resuspended in 30 l of kinase buffer (50 mM Tris-HCl (pH 7.4), 10 mM MnCl 2 , 10 mM MgCl 2 , 0.1% Triton X-100, and 1 mM dithiothreitol) containing phosphatase and protease inhibitors, 0.25 mg/ml poly(Glu/Tyr) (4:1; Sigma) as an exogenous kinase substrate, 10 M unlabeled ATP, and 10 Ci of [␥-32 P]ATP (6000 Ci/mmol; PerkinElmer Life Sciences). After 10 min at 30°C, the reaction was stopped by adding SDS sample buffer and boiling the samples for 10 min. Subsequently, the samples were resolved on SDS-12% polyacrylamide gels, and the gels were stained with Coomassie Blue. The labeled poly(Glu/Tyr) was excised from the gel, and radioactivity was counted.

Generation of the Binding Mutants
Two arginine residues are conserved in the Tyr(P)-binding pocket. Alignment of the CHK SH2 domain with other SH2 domain sequences shows that whereas the arginine in ␤B5 is present in these SH2 sequences, the ␣A2 CHK SH2 domain has a glycine. This arginine in the binding pocket of other SH2 domains provides a positive charge to coordinate the phosphate of Tyr(P), implying either a different mode of binding or weak binding for CHK. Arg 147 in ␤B is a critical residue for Tyr(P) binding and is strictly conserved in SH2 domains (23). Based on these observations, we constructed three CHK SH2 domain mutants: G129R, R147A, and R147K. The G129R mutant encodes a CHK SH2 domain protein in which the ␣A2 glycine at the phosphotyrosine-binding pocket has been replaced by arginine. For the R147A and R147K mutants, the ␤B5 arginine has been replaced by alanine and lysine, respectively. Our hypothesis is that G129R will have enhanced ErbB-2 phosphopeptide binding, R147A will have disrupted binding, and R147K will have reduced binding.

Binding Studies
GST Pull-down Experiment-We conducted binding studies with the three CHK SH2 domain mutants as well as with wild-type CHK (Fig. 2). Two different breast cancer cell lines, T47D (moderate level of ErbB-2/neu expression) and BT474 (high level of expression), were tested. The conserved tyrosine residues of overexpressed ErbB-2/neu in BT474 cells were found to be autophosphorylated in the absence of ligand stimulation. The cells were serum-starved as described under "Experimental Procedures" and then activated with heregulin (20 nM) for 8 min. Unstimulated and stimulated cells were lysed and precipitated with GST-CHK SH2 fusion proteins as well as with GST protein alone. The precipitates were analyzed by SDS-PAGE and immunoblotted with anti-phospho-HER2/ ErbB-2 (Tyr 1248 ) antibody. In comparison with wild-type CHK, the substitution of Gly 129 with Arg 129 resulted in dramatically increased binding in both cell lines. In T47D cells ( Fig. 2A), only heregulin stimulation induced the association of ErbB-2 with the purified wild-type SH2 domain and also G129R, indicating that G129R binding is ligand-stimulated to a similar level compared with wild-type binding. The substitution of Arg 147 with Lys 147 slightly decreased binding, whereas the substitution with Ala 147 completely disrupted binding. These results indicate that the positive charge of Arg 147 is critical for phosphopeptide binding and that the length of the side chain has a moderate effect on binding. Because the R147K mutant showed only a moderate effect on binding in comparison with the wild-type SH2 domain, the R147K mutant was not analyzed in further experiments. In BT474 cells, no ligand stimulation was necessary for the CHK SH2 domain interaction with the constitutively phosphorylated ErbB-2/neu protein (Fig.  2B). Again, almost no association of R147A with ErbB-2/neu was seen, whereas G129R pulled down markedly more ErbB-2 protein than did the wild-type CHK SH2 domain.
NMR Experiments-The backbone atoms of the CHK SH2 domain were assigned using triple resonance experiments (Fig.  3A). NMR, which is a method used to determine the high resolution structure of macromolecules, is particularly valuable in providing rapid identification of a ligand-binding site. The 15 N-1 H HSQC experiment yields a well resolved spectrum, with single peaks for the backbone amides of most residues in the protein (25). Changes in the positions of these peaks upon titration of the ligand can identify the residues in the binding site, and analysis of the titration data can be used to determine the kinetics of binding (24).
To analyze the interaction between CHK SH2 domain constructs and the phosphopeptide, we titrated the peptide into 0.6 mM 15 N-labeled SH2 domains. Progressive changes in 1 H and 15 N chemical shifts were monitored with a series of 15 N HSQC spectra. Significant chemical shift changes in the wild-type SH2 domain were observed for several residues in the ␣A, ␤B, and ␤D secondary structures as well as in the ␤D-␣B and ␣B-␤E loops (Fig. 3A), which have been implicated in the binding of other SH2 domains to phosphopeptides (26). In particular, several positive residues in ␤D (His 168 and Arg 170 ) undergo large chemical shift changes upon complex formation, consistent with the positive charged residues in these secondary structural elements forming contacts with Tyr(P) of the ligand. In the ␣A helix, several residues changed chemical shifts upon complex formation (Ile 127 , Gly 129 , Gln 134 , and Gln 135 ). The ␤B-␤C loop has also been implicated in the binding of the peptide, and significant chemical shift changes were observed for residues in this loop (Ser 149 , Arg 151 , and Gly 154 ). Interestingly, Arg 147 showed little change in chemical shift despite its presumed role in contacting the phosphate of the ligand. Significant chemical shift changes were also observed in residues in ␤D (Tyr 169 , Val 171 , and Leu 172 ), the ␤D-␣B loop (Ile 180 and Asp 181 ), and the ␣B-␤E loop (Ile 203 ), which presumably form the hydrophobic environment for the hydrophobic residues of the C terminus to the Tyr(P) region of the peptide (see Fig. 5).
G129R, which showed enhanced binding, was also titrated with the phosphopeptide (Fig. 3B). The Gly 129 resonance was replaced with a new resonance (Arg 129 ), consistent with the substitution of glycine with arginine and the expected changes in chemical shift (27). Several residues showed chemical shift changes upon mutation. In particular, residues in ␣A (Ile 127 , Glu 131 , Ala 132 , and Gln 134 ) experienced chemical shift changes (0.3 Ϯ 0.1 ppm) presumably due to introduction of the long charged side chain of Arg. Several residues in ␤B and ␤D also showed chemical shift changes, suggesting that they are in proximity to the G129R mutation site. Upon titration with peptide, residues similar to the wild-type residues underwent significant chemical shift changes, except for residues near the mutation site in ␣A and ␤B (Table I).
R147A did not show chemical shift changes with any residue upon peptide titration, consistent with the complete disruption of binding seen in the GST pull-down experiment. With this mutation, the peak for Arg 147 disappeared, but a new Ala 147 peak was difficult to identify, presumably due to resonance overlap (Fig. 3C). Compared with G129R, the R147A substitution showed different chemical shifts for residues in regions such as ␣A (Ile 127 , Ser 128 , and Gly 129 ), ␤B (Ser 149 and His 152 ), the ␤B-␤C loop (Gly 154 and Asp 155 ), ␤C (Cys 159 and Val 160 ), and ␤D (Asp 165 , Ile 167 , and Val 171 ). In addition, Gly 129 disappeared with the mutation, implying line broadening consistent with the introduction of a motion on the millisecond-to-microsecond time scale. Although both mutants showed perturbation at several residues, the overall structure seemed to be unaltered, as the majority of the residues showed little change (Fig. 3).
Equilibrium dissociation constants (K d ) were determined from a plot of chemical shift changes versus the ratio of peptide to protein. Chemical shifts of residues Ile 127 (␤A-␣A loop); FIG. 3. A, superimposed 15 N-1 H HSQC spectra of the wild-type CHK SH2 domain without (black) and with (red) 1 eq of the Tyr(P) peptide (Ac-ENPEpYLGLDV-NH 2 ). B, superimposed 15 N-1 H HSQC spectra of the wild-type CHK SH2 domain (black) and the G129R mutant (red). C, superimposed 15 N-1 H HSQC spectra of the wild-type CHK SH2 domain (black) and the R147A mutant (red). The 500-MHz spectra were collected at 25°C from samples containing ϳ0.6 mM SH2 domain. Peaks that significantly changed chemical shifts upon binding the peptide (A) or upon mutation (B and C) are labeled in black by residue number. Residues targeted for mutation are labeled in red.
Gly 129 (␣A); Arg 147 (␤B); Gly 154 (␤B-␤C loop); Tyr 156 (␤C); His 168 , Tyr 169 , and Arg 170 (␤D); and Ile 203 (␣B-␤E loop) were monitored because these residues were involved in the phosphopeptide interaction and remained well resolved throughout the titration. Consistent with the results from the GST pulldown binding assay, overall, wild-type CHK has a K d of ϳ0.5 mM, showing five times weaker affinity than G129R (Table I).
In wild-type CHK, residues in ␤D (His 168 , Tyr 169 , and Arg 170 ) showed tighter binding than Gly 154 , Tyr 156 , and Ile 203 , which are predicted to be the C-terminal residues of the bound ErbB-2 phosphopeptide. In G129R, the affinity of Gly 154 , Tyr 156 , and Ile 203 increased Ͼ10 times over that seen in wild-type CHK, whereas the residues in ␤D (His 168 , Tyr 169 , and Arg 170 ) remained the same as observed in wild-type CHK.

Src Kinase Assay
To investigate the functional role of the association of CHK and ErbB-2, we tested the effects of transient transfection with CHK-pIRES2-EGFP constructs on heregulin-stimulated Src kinase activity in MCF-7 cells. The expression levels of all studied forms of the CHK protein were similar as assessed by Western blotting (Fig. 4). As shown in Fig. 4, stimulation of MCF-7 cells with heregulin caused a dramatic increase in the activity of Src kinase. Transfection with the wild-type CHK gene strongly inhibited the heregulin-stimulated Src activity, whereas no significant difference was seen in cells transfected with the kinase-dead mutant of CHK or an empty vector control. Furthermore, the inhibition of heregulin-stimulated Src kinase activity was markedly diminished by R147A, whereas the G129R-mediated inhibition was somewhat stronger compared with wild-type CHK.

DISCUSSION
The interaction of SH2 domains and their phosphotyrosinecontaining binding partners has been extensively studied because of its critical role in signal transduction and cancer development (23,28). Studying the binding of the CHK SH2 domain to the ErbB-2 receptor is particularly important for understanding the development of breast cancer (17). Studies using phosphopeptide libraries to probe sequence specificity showed that there are two types of SH2 domains, types I and II (23). Type I SH2 domains involve non-receptor tyrosine kinases such as Src, Lck, Fyn, and Abl and exhibit preferences for the motif Tyr(P)-hydrophilic-hydrophilic-(Ile/Pro) (29). Type II SH2 domains include phospholipase C␥1, Syp tyrosine phosphatase, Shc, and Grb2 and prefer the motif Tyr(P)-hydrophobic-X-hydrophobic. Structural analysis of the type I high affinity binding (10 Ϫ8 M) SH2 domain-peptide complex showed that Tyr(P) (in pYEEI) is inserted into a large positively charged pocket and Ile into a smaller hydrophobic pocket. In contrast, the type II phospholipase C␥1-peptide complex has relatively lower affinity binding (K d ϭ 10 Ϫ6 M) and demonstrates more extensive peptide interactions, using a long hydrophobic groove to accommodate the two hydrophobic residues in the peptide. Residues involved in binding Tyr(P) (Gly 129 , Arg 147 , His 168 , and Arg 170 ) are indicated in red for CHK and in blue for the corresponding residues of Src. Residues composing the hydrophobic site that interact with residues composing the C terminus to the Tyr(P) region of the phosphopeptide are indicated in pink (CHK) and in green (Src). The nonconserved Arg 151 of CHK is indicated in pink. The CHK SH2 domain binds to the ErbB-2 phosphopeptide (pYLGLDV), which contains the consensus sequence for binding to a type II SH2 domain (30). HSQC experiments identified the residues involved in phosphopeptide binding. Residues of the positively charged Tyr(P)-binding site of CHK, as observed in other SH2 domain-peptide interactions, are predicted to be Gly 129 in ␣A, Arg 147 in ␤B, and His 168 and Arg 170 in ␤D. Residues of this binding site participate in electrostatic interactions with the Tyr(P) ring. The hydrophobic residues in the specificity-determining site (␤D, ␤D-␣B loop, and ␣B-␤E loop) provide extensive hydrophobic contacts between the C-terminal hydrophobic residues and Tyr(P) of the peptide. Recently, the crystal structure of the CHK SH2 domain has been solved by Murthy et al. (34) (Protein Data Bank code 1JWO). The overall folding of the CHK SH2 domain is similar to that of other SH2 domains. Superposition of backbone atoms of the secondary structure residues of the CHK and Src SH2 domains showed that the overall folding and binding pocket are conserved (Fig. 5). The side chain of the nonconserved Arg 151 of the CHK SH2 domain is positioned near the Tyr(P)-binding site. The positive charge of this residue may also contribute to binding. The chemical shift of Arg 151 was changed by peptide addition.
The dissociation constant of the wild-type CHK SH2 domain was found to be 0.5 mM, which is much weaker than that of the low affinity type II SH2 domain (23). The low affinity for the phosphopeptide does not necessarily mean a low affinity of ErbB-2/neu for CHK in vivo in cells. Because Tyr(P) 1248 is the most critical residue in ErbB-2/neu-mediated cancer development, dynamic regulation can be involved in its interaction. We suspect that the unusual presence of Gly 129 in the Tyr(P)binding site is one reason for this weak binding. The GST pull-down experiment showed a dramatic enhancement of the binding when Gly 129 was substituted with Arg 129 . The K d of G129R decreased by Ͼ5-fold, as determined from the NMR titration experiments. Both experiments indicated that G129R showed higher binding affinity for the ErbB-2 peptide. The similar chemical shift changes of the residues in the specificitydetermining site of G129R and the wild-type SH2 domain indicate no alteration in specificity. Interestingly, the increased affinity of G129R is caused by a dramatic increase in the binding affinity of residues that are likely to bind to the C-terminal residues of the ErbB-2 phosphopeptide (Table I).
The R147K mutation decreased binding, whereas the R147A mutation completely disrupted binding. This result implies that the positive charge of Arg 147 is critical to coordinate the negative charge of Tyr(P) and that the length of the side chain also affects the binding. The importance of Arg 147 in SH2 domains has been tested in different systems, and the results indicate that this residue is critical for binding (31,32). Alanine mutation of this strictly conserved Arg 147 (␤B) resulted in a large increase in ⌬G 0 (⌬⌬G 0 ϭ 3.2 kcal/mol), whereas mutations of other residues each resulted in a significantly smaller (⌬⌬G 0 Ͻ 1.4 kcal/mol) reduction in affinity. This indicates that Arg 147 (␤B) is an important determinant of the Tyr(P)-binding recognition site (33).
How can CHK be involved in negative growth regulation in human breast cancer? How can the specific binding of CHK to the most critical and strictly conserved autophosphorylation site (Tyr(P) 1248 ) of ErbB-2 be related to the down-regulation of ErbB-2-mediated Src family kinase activity? One model is that upon heregulin stimulation, CHK association with the ErbB-2 receptor locates the CHK near the substrate, Src, thereby causing growth inhibitory effects (13,14). The enhanced binding or disrupted binding mutants were used to test whether the specific association of CHK and ErbB-2 is related to Src kinase activity. The R147A mutant markedly diminished the inhibition of Src activity, whereas the G129R mutant inhibited Src activity more potently in comparison with the wild-type SH2 domain. These results prove that the specific interaction of CHK and ErbB-2 via the SH2 domain is directly related to the growth inhibitory effects of CHK.
Because CHK shows restricted expression and specifically inhibits breast cancer development, CHK can be a potential candidate for gene therapy. However, the expression levels of CHK are very low in breast tumors, and CHK barely shows kinase activity in breast cancer tissues as compared with Csk (13)(14)(15)(16). Because highly effective inhibition of Src activity is important for gene therapy, improved expression of the CHK gene, development of CHK with higher kinase activity, and construction of a higher affinity CHK SH2 domain would be very important for inhibiting breast cancer growth. Thus, the enhanced binding mutant (G129R) is promising as a potential candidate for gene therapy.