A Novel Non-catalytic Mechanism Employed by the C-terminal Src-homologous Kinase to Inhibit Src-family Kinase Activity* □ S

Although C-terminal Src kinase (CSK)-homologous kinase (CHK) is generally believed to inactivate Src-family tyrosine kinases (SFKs) by phosphorylating their consensus C-terminal regulatory tyrosine (Tyr T ), exactly how CHK inactivates SFKs is not fully understood. Herein, we report that in addition to phosphorylating Tyr T , CHK can inhibit SFKs by a novel non-catalytic mechanism. First, CHK directly binds to the SFK mem-bers Hck, Lyn, and Src to form stable protein complexes. The complex formation is mediated by a non-catalytic Tyr T -independent mechanism because it occurs even in the absence of ATP or when Tyr T of Hck is replaced by phenylalanine. Second, the non-catalytic CHK-SFK interaction alone is sufficient to inactivate SFKs by inhibiting the catalytic activity of autophosphorylated SFKs. Third, CHK and Src co-localize to specific plasma membrane mixture containing autophosphorylated Hck (3 pmol at 60 n M final concentration) were assessed for (i) the effect of CHK on the activity of autophosphorylated Hck and (ii) the formation of a stable complex between CHK and autophosphorylated Hck. Briefly, autophosphory- lated Hck was preincubated at 30 °C for 30 min in varying CHK concentrations (0–10 (cid:2) M ). Hck activity was assayed using cdc2-(6–20)K19 peptide. The kinase assay was allowed to proceed for only 5 min. The choice of using a lower concentration of Hck (60 n M ) and a shorter reaction time (5 min) assisted in preventing further Tyr A and Tyr T phosphorylation during the course of assaying Hck activity. Anti-CHK immunoprecipitation was conducted to detect Hck bound to CHK following the procedures as detailed for Fig. 3. is sufficient

Src-family kinases (SFKs) 1 are non-receptor protein-tyrosine kinases that participate in many cellular functions ranging from cell growth and proliferation to memory and learning (1). The kinase activity of SFKs is regulated by phosphorylation, as well as by their interaction with other cellular proteins. Among the various regulatory mechanisms, the most important are autophosphorylation of a consensus tyrosine (Tyr A ) 2 in the kinase domain and phosphorylation of a consensus regulatory tyrosine near the C terminus (Tyr T ) 2 (2,3). Autophosphorylation of Tyr A leads to activation of SFKs (1,4,5). The crystal structure of the autophosphorylated kinase domain of the Srcfamily kinase Lck reveals that the phosphorylated Tyr A (Tyr(P) A ) stabilizes the active kinase domain configuration by forming ionic interactions with the conserved Arg in the catalytic loop (6). We previously reported that the Src-family member Hck could undergo autophosphorylation at a novel site (Tyr-29) 2 in the Unique domain and that autophosphorylation of Hck at Tyr-29 contributed to Hck activation (4). However, the structural basis of activation by Tyr-29 autophosphorylation is not yet known. In contrast to the activating effect of Tyr A and Hck Tyr-29 autophosphorylation, Tyr T phosphorylation results in inactivation (7). Crystal structures of Tyr T -phosphorylated c-Src (8) and Hck (9) reveal that the inactive configuration of the kinase domain is stabilized by intramolecular interactions involving binding of (i) the Tyr(P) T to the SH2 domain and (ii) the SH2 kinase linker to the SH3 domain.
The consensus Tyr T of SFKs is thought to be phosphorylated exclusively by two upstream regulatory tyrosine kinases, Cterminal Src kinase (CSK) and CSK-homologous kinase (CHK) 2 (2, 10 -12). CSK is ubiquitously expressed in all mammalian tissues, whereas the expression of CHK is much more restricted; it is expressed predominantly in neurons and hemopoietic cells. Extensive biochemical evidence indicates that CSK inactivates SFKs primarily by phosphorylating their Tyr T . Although there are several pieces of preliminary evidence suggesting that CHK can also phosphorylate Tyr T of several SFK members, including c-Src, Lyn, Lck, and Fyn (10,11), CHK-mediated SFK inactivation remains largely uncharacterized. Using Hck, Lyn, and Src as model targets, we investigated the mechanism by which CHK inactivates SFKs. We discovered that in addition to inactivation of SFKs by specifically phosphorylating their Tyr T , CHK could inactivate SFKs by a novel inhibitory mechanism. In this mechanism CHK directly binds to SFKs to form stable complexes, and this binding alone is sufficient to inactivate SFKs. To further confirm the validity our in vitro observations, we attempted to determine if CHK could bind and inactivate Hck in transfected HEK293T cells. Results of our experiments using the transfected HEK293T cells reveal that CHK⅐Hck complex formation and inactivation of Hck occurred even when Tyr T of Hck was replaced by phenylalanine. The results indicate that CHK⅐Hck complex formation and Hck inactivation by this novel inhibitory mechanism are not mediated by binding of Tyr T of Hck to the active site of CHK. Hence, this novel inhibition occurs by a non-catalytic mechanism. To support the physiological relevance of our findings, we reveal that CHK and Src co-localize to specific microdomains of rat brain plasma membrane and that CHK and Src form stable protein complex(es) in rat brain. Furthermore, we also demonstrate that the interaction can suppress SFK activity regardless of the level of autophosphorylation of Tyr A . Thus, by binding to and suppressing the activity of both the less active unphosphorylated form and the fully active autophosphorylated form of SFKs, CHK is capable of providing a fail-safe mechanism to down-regulate SFK signaling under all circumstances.
Chemical Cross-linking of Antibodies to Protein A-Sepharose-Data generated by the use of these reagents are shown in Fig. 10C. Rabbit IgG from preimmune sera, anti-CHK, and anti-CSK antibodies were chemically cross-linked to protein A-Sepharose with dimethyl pimelimidate (Sigma-Aldrich) following the procedures described by Harlow and Lane (16).
Mutagenesis and Construction of Expression Vectors-The cDNA fragment encoding full-length CHK was retrieved from the pRK7-BatK-flagC plasmid by digestion with EcoR1 and BamH1. It was then used as template for a PCR reaction that introduced EcoR1 and BglII restriction sites, and the stop codon. The oligonucleotides used were 5Ј-CCG GAA TTC GCT AAG ATG CCA ACG CAA CGC-3Ј (Primer 1 sense) and 5Ј-G ACT AGA TCT TCA GGG GTC CTG GCT CCG-3Ј (Primer 2 antisense); the restriction sites are underlined, and the start and stop codons are in bold italics. The PCR product generated was cloned into the pBAK-PAK9 vector (Clontech) using procedures described previously (4). For the K221M mutation pBacPAK9-CHK plasmid was used as template for PCR-based mutagenesis. The oligonucleotides used were Primer 1 Sense, Primer 2 Antisense, and Primer 3 Antisense (5Ј-G TAC TTG GGA CAG AAG GTG GCC GTG ATG AAT ATA-3Ј); the A to T codon change is underlined in the sequence. The resultant plasmids, pBacPAK9-CHK or pBacPAK9-CHK(K221M), were sequenced to confirm that only the required mutation was introduced during the PCR procedure. Both plasmids were used to co-transfect with BAC-PAK6 baculoviral DNA into Sf9 cells to generate the recombinant CHK and CHK(K221M) baculoviruses according to manufacturer's instructions (Clontech).

Purification of Recombinant CHK and Its Mutant Expressed in Sf9
Insect Cells-All purification procedures were carried out at 4°C. One liter of CHK baculovirus-infected Sf9 cells (0.7 ϫ 10 6 cells/ml) were harvested by centrifugation at 1000 ϫ g for 5 min. The cell pellet was homogenized in lysis buffer (50 mM Tris, pH 7.0, 1% Nonidet P-40, 1 mM EDTA, 0.2 mg/ml benzamidine, 50 mM ␤-glycerophosphate, 0.1 mg/ml phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 5 mM sodium potassium tartrate, 1 mM p-nitrophenylphosphate, 10% glycerol, 0.1 mg/ml trypsin inhibitor, and 50 mM NaF). The supernatant was applied to a DEAE column (80-ml bed volume) pre-equilibrated with Buffer A (25 mM Hepes, pH 7.0, 0.1% Nonidet P-40, 10% glycerol, 1 mM EDTA, 0.2 mg/ml benzamidine, 10 mM ␤-glycerophosphate, 0.1 mg/ml phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 0.5 mM Na 3 VO 4 , and 50 mM NaF). The unbound fraction, which contained the majority of the CHK protein was then loaded onto a carboxymethyl cation exchange column (100 ϫ 22 mm). Proteins bound to the carboxymethyl column were eluted with a 0 -1 M NaCl gradient in Buffer A. Fractions containing CHK were further purified on a Superose-12 gel-filtration column equilibrated in Buffer A with 0.2 M NaCl. The CHK-containing column fractions were pooled and applied to a phosphotyrosine-agarose column (5-ml bed volume, 5-15 mol of immobilized phosphotyrosine per ml). The column was washed with Buffer A containing 0.2 M NaCl. The bound proteins were eluted from the column with 100 mM p-nitrophenylphosphate in Buffer A. Fractions containing CHK were pooled and purified by Mono S cation exchange column chromatography. As shown in Fig. 1C, the final CHK preparation was Ͼ90% pure. The recombinant CHK(K221M) mutant was also purified using an identical procedure.
Preparation of Crude Rat Brain Lysate for Detection of CHK⅐Src Protein Complexes-All procedures described were performed at 4°C. Brain tissues from 10 adult rats were homogenized in 140 ml of lysis buffer. The homogenate was centrifuged twice at 40,000 ϫ g for 40 min. The supernatant was separated into aliquots and frozen at Ϫ70°C before immunoprecipitation analysis to generate the data shown in Fig. 10C.
Fractionation of Rat Brain Membrane-This section describes Fig.  10B. All procedures were performed at 4°C. Brain tissues from 10 adult rats were homogenized in Buffer B (0.32 M sucrose, 25 mM Tris-HCl, pH 7.0, 1 mM MgCl 2 , 0.2 mg/ml benzamidine, 0.1 mg/ml phenylmethylsulfonyl fluoride). The homogenate was centrifuged at 600 ϫ g for 10 min to remove the nuclei and large debris. The supernatant was centrifuged at 100,000 ϫ g for 30 min to separate the plasma membrane and cytoplasm. The crude membrane was further separated into different membrane microdomains by sucrose density gradient centrifugation and treatment with Triton X-100 using a procedure adapted from Prior et al. (17) and Waheed et al. (18). Briefly, the crude membrane was resuspended in 2 ml of 45% (w/v) sucrose in MBS solution (25 mM MES, pH 6.5, 150 mM NaCl, 0.2 mg/ml benzamidine, and 0.1 mg/ml phenylmethylsulfonyl fluoride). The suspension was loaded into a centrifuge tube and then overlaid with a MBS-sucrose gradient of 35% sucrose (2.8 ml), 30% sucrose (2.4 ml), 25% sucrose (2.4 ml), and 5% sucrose (2.4 ml). After centrifugation at 300,000 ϫ g in the Beckman SW41 rotor for 16 h, 12 1-ml fractions were harvested from the top of the tube. Fractions 2-5 and 9 -12 correspond to the low and high density membrane microdomains, respectively. To pellet the membrane in each fraction, MBS (9 ml) was added to the individual fractions, and the mixture was centrifuged at 300,000 ϫ g for 30 min. The pellet from each fraction was treated with 0.5 ml of 1% Triton X-100 in Buffer B. The samples were again subjected to centrifugation at 300,000 ϫ g for 15 min. The supernatant corresponds to the Triton X-100-soluble portion, whereas the pellet corresponds to the Triton X-100-insoluble portion. Aliquots of both portions were analyzed for the presence of CHK and Src by Western blotting.
Assay of CHK Protein Kinase Activity-Three different substrates, poly(Glu/Tyr), Hck(K267M), and SFK C-terminal peptide were used to assay CHK activity. For poly(Glu/Tyr) phosphorylation the kinase assay mixture (25 l) consisted of 8 -12 g of CHK, kinase assay buffer (20 mM Tris-HCl, pH 7.0, 10 mM MgCl 2 , 1 mM MnCl 2 , 50 M Na 3 VO 4 ), 2.5 mg/ml poly(Glu, Tyr), and 100 M [␥-32 P]ATP. An assay mixture without CHK was used as a blank to quantify the background radioactivity. Phosphorylation was allowed to proceed at 30°C for 30 min. After termination of the reaction, CHK activity was determined following procedures previously described (3). The results are shown in Fig. 1B.
Hck(K267M) was used as a bona fide substrate of CHK to investigate the efficiency of CHK in catalyzing Tyr T phosphorylation. The reaction mixture contained kinase assay buffer, 100 M [␥-32 P]ATP, CHK (0.33 M), and varying concentrations of Hck(K267M) (0 -5.8 M). The reaction mixture was incubated at 30°C. After termination of the reaction, Hck(K267M) was separated by SDS-PAGE, and the gel was subjected to autoradiography. The individual protein bands corresponding to Hck(K267M) were excised, and the associated radioactivity was quantitated ( Fig. 2C).
An SFK C-terminal peptide, modeled after the C terminus of the Src-family kinase Lyn (KEKAEERPTFDYLQSVLDDFYTATEGQY T Q-QQP, Y T is Tyr T ) , was synthesized (3). Because the sequence of this peptide exhibits Ͼ90% sequence identity with the sequence of Hck C-terminal regulatory domain, we compared the kinetics of CHK phosphorylation on the peptide and Hck(K267M) under identical conditions. The K m and V max values were determined by double-reciprocal transformation of the Michaelis-Menten curve (Fig. 2B).
Demonstration of Stable Enzyme-Substrate Complex Formed by CHK and Hck(K267M)-This section describes Fig. 2D. Two identical sets of reaction mixtures containing CHK (0.33 M) and varying amounts of Hck(K267M) (0 -5.8 M) were prepared. The reaction mixtures were incubated at 30°C for 30 min in the presence of either (i) 100 M ATP and kinase assay buffer or (ii) Buffer A. After 30 min the reaction mixtures were diluted 4-fold with an EDTA stock solution that gave a final concentration of 5 mM EDTA; this amount of EDTA was sufficient to chelate Mg 2ϩ and Mn 2ϩ supplied to the reaction mixtures. Immunoprecipitation was performed by adding 4.3 g of the anti-CHK antibody to each preincubated mixture. CHK was allowed to react with the anti-CHK antibody at 4°C for an hour before being transferred onto Protein A-Sepharose beads (10-l bed volume). The antibody-containing mixtures were further incubated at 4°C for another hour. The immunoprecipitates were then washed extensively with 6 ϫ 1.0 ml of 0.5 M NaCl in Buffer A, and the bound proteins were eluted with SDS-PAGE sample buffer. The proteins were resolved on 10% SDS-PAGE and then blotted onto PVDF membranes. The resultant blots were probed with the anti-Hck antibody followed by horseradish peroxidase-conjugated protein A (protein A-HRP conjugate). Because the IgG heavy chain and CHK (molecular mass, ϳ50 kDa each) show similar mobility on SDS-PAGE, the use of Protein A-HRP conjugate helped to maximize the detection of CHK immuno-signals due to its insensitivity to cross-react with the denatured IgG.
Demonstration of Stable CHK⅐SFK Complex Formation by Co-immunoprecipitation-This procedure describes Figs. 3 and 5-10. For immunoprecipitation with the anti-CHK antibody, CHK (0 -10 M) was incubated at 30°C for 30 min with Hck, Lyn, or Src (88 nM to 0.48 M). As a control, SFKs were omitted from the incubation mixture. The binding assay mixture was applied to a tube containing anti-CHK antibody (1.7 g) immobilized to protein A-Sepharose (bed volume, 10 -15 l). The mixture was further incubated for 2 h at 4°C, and the immunoprecipitates were washed and processed for subsequent Western blot analysis. For anti-Hck immunoprecipitation, the binding assay mixture was transferred to protein G-Sepharose (10 l bed volume) precoated with a rat monoclonal anti-Hck antibody. As for the in vivo experiment, given that CHK, CSK, and Src are less abundant in the crude brain lysate, immunoprecipitation was performed with antibodies covalently crosslinked to protein A-Sepharose in an effort to minimize interference from the IgG heavy chain (Fig. 10C).
Assay for the Ability of CHK to Inactivate SFKs (Inactivation Assay)-This section describes Figs. 1B, 4A, 6B, 8A, and 9A. The two-step assay method described previously (3) was used to monitor the ability of CHK to inactivate Hck, Lyn, and Src. In Step 1, each SFK (0.11-0.16 M Hck, 0.23 M Lyn, and 0.48 M Src) was preincubated at 30°C for 30 min in a reaction volume of 25 l containing varying amounts of CHK, kinase assay buffer, and 100 M [␥-32 P]ATP. In this step, SFKs were capable of undergoing autophosphorylation at Tyr A , whereas Tyr T was capable of phosphorylation by CHK. As a control, SFKs were preincubated in the absence of CHK under identical conditions. In Step 2, an activity assay was conducted to specifically monitor the kinase activity of the preincubated SFK. Because cdc2-(6 -20)K19 peptide was previously reported as a specific and efficient substrate for many SFKs (19,20) but was a poor substrate for CSK (3) and CHK (data not shown), it was used as the substrate for this assay. The preincubation mixtures (10 l per sample) were separated into aliquots, and phosphorylation was initiated upon the addition of an assay mix containing 5 l each 100 M [␥-32 P]ATP, kinase assay buffer, and 1.5 mM cdc2-(6 -20)K19 peptide. The reaction was allowed to proceed for 20 -30 min. SFK kinase activity was determined from the radioactivity associated with the peptide substrate as previously described (4). The ratio of SFK kinase activity in the preincubation mixtures containing CHK (inactivation reaction) versus that without CHK (control reaction) was determined. From the ratio, the % inactivation of the SFK was calculated as previously described (3). The formula involved is % Inactivation ϭ ͫ 1 Ϫ SFK activity in inactivation reaction SFK activity in control reaction ͬ ϫ 100 (Eq. 1) Co-elution of CHK in gel-filtration chromatography with protein-tyrosine kinase activity and the activity that inhibits Hck. CHK partially purified by DEAE/carboxymethyl ion-exchange column chromatography was applied to the Superose 12 gelfiltration column. The column fractions were monitored for anti-CHK immunoreactivity, poly(Glu/Tyr) tyrosine kinase activity, and activity that inhibits Hck. The ability of CHK to inhibit Hck was assayed by the inactivation assay procedure. A, elution profiles of CHK from the gelfiltration column. WB, Western blot. B, profiles of the protein-tyrosine kinase activity and the activity that inhibits Hck from the gel-filtration column. C, analysis of the final purified CHK preparation by SDS-PAGE and Western blotting to show purity and authenticity.

Determination of the Stoichiometry of SFK Phosphorylation-This
was incubated at 30°C for 30 min. After termination of the reaction, the phosphoproteins were separated by SDS-PAGE. The bands corresponding to the phosphorylated SFKs were excised, and the associated radioactivity was determined.
Phosphopeptide Mapping-This procedure was used to generate Figs. 2A, 4D, 8D, and 9D. Hck, Lyn, and Src were incubated with [␥-32 P]ATP at varying CHK concentrations under conditions detailed in the previous section. The phosphoproteins were separated by SDS-PAGE and transferred to a nitrocellulose filter, and the bands corresponding to the radioactively phosphorylated SFKs were excised and processed for tryptic digestion. The tryptic fragments were separated and identified by the two-dimensional thin layer electrophoresis/TLC procedures described previously (13). The spots corresponding to Tyr(P) T , Tyr(P) A (Src and Hck), and Tyr(P)-29 (Hck only) on the phosphopeptide maps were identified as described in our previous reports (4,13). Similar conditions and procedures were used for phosphopeptide mapping of Hck(K267M) phosphorylated by CHK.

FIG. 2. CHK specifically phosphorylates Tyr T of Hck, and the phosphorylation reaction deviates from Michaelis-Menten kinetics.
A, two-dimensional tryptic phosphopeptide map of Hck(K267M) radioactively phosphorylated by CHK in vitro. The four-pronged arrow denotes phosphopeptide fragments containing Tyr(P) T (pY T ) of Hck(K267M). As confirmed by our previous report (4), four phosphopeptide fragments were generated as a result of incomplete tryptic digestion. Because the Lys-to-Arg mutation rendered Hck catalytically inactive, the mutant could not undergo autophosphorylation, and therefore, the phosphopeptide map revealed no autophosphorylation at Tyr A and Tyr-29. B, kinetic analysis of phosphorylation of the SFK C-terminal peptide by CHK. CHK (1. The autoradiogram in the inset shows the levels of phosphorylation of Hck(K267M) at varying concentrations. D, demonstration of stable complex formation between CHK and Hck(K267M) by co-immunoprecipitation. CHK (0.33 M) and Hck(K267M) were incubated with assay buffer in the presence (ϩ) and absence (Ϫ) of Mg 2ϩ /ATP at 30°C for 30 min. CHK was immunoprecipitated (IP) using protein A-Sepharose precoated with the anti-CHK antibody. Hck(K267M) bound to CHK in the immunoprecipitate was detected by anti-Hck Western blotting (WB). TLE, thin layer electrophoresis; TLC, thin layer chromatography. cells using FuGENE 6 TM transfection reagent (Roche Applied Science). Twenty hours post-transfection, the cells were lysed with lysis buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, 1% Nonidet P-40, 10% glycerol, 1 mM Na 3 VO 4 , 0.1 mM sodium molybdate, and 1 mM Pefabloc). Insoluble material was removed by centrifugation (10 min, 10,000 ϫ g). The total amount of cellular proteins present in the crude cell lysates was determined by the Bradford assay (42).
Assay for Hck and Hck(Y499F) Activity in the Crude Lysates of the Transfected HEK293T Cells-This section describes Fig. 5, B-D. The catalytic activity of Hck and Hck(Y499F) in HEK293T cell lysates (ϳ62.5 g of total proteins) was measured by kinase assay. Src-optimal peptide (AEEEIYGEFEAKKKK) was employed as the exogenous substrate because it has been documented as a specific and efficient substrate for Hck (13,21). The phosphorylated peptide was separated from free [␥-32 P]ATP by a p81 filter paper assay procedure as described previously (13).
Demonstration of CHK Inactivation and Binding to Autophosphorylated Hck-This procedure describes Fig. 7. Hck (68 pmol at 1.4 M final concentration) was autophosphorylated upon incubation with 100 M ATP and kinase assay buffer in 50 l at 30°C for 1 h. Autophosphorylation was stopped by diluting the reaction mixture in 175 l of Buffer A containing 5 mM EDTA. Aliquots (10 l per sample) of the diluted mixture containing autophosphorylated Hck (3 pmol at 60 nM final concentration) were assessed for (i) the effect of CHK on the activity of autophosphorylated Hck and (ii) the formation of a stable complex between CHK and autophosphorylated Hck. Briefly, autophosphorylated Hck was preincubated at 30°C for 30 min in varying CHK concentrations (0 -10 M). Hck activity was assayed using cdc2-(6 -20)K19 peptide. The kinase assay was allowed to proceed for only 5 min. The choice of using a lower concentration of Hck (60 nM) and a shorter reaction time (5 min) assisted in preventing further Tyr A and Tyr T phosphorylation during the course of assaying Hck activity. Anti-CHK immunoprecipitation was conducted to detect Hck bound to CHK following the procedures as detailed for Fig. 3.

Purified Recombinant CHK Displays Both Protein-tyrosine
Kinase Activity and the Activity to Inhibit the Src-family Kinase Hck in Vitro-Recombinant CHK was expressed in Sf9 cells and purified by sequential chromatography on ion-exchange, gel-filtration, and phosphotyrosine-agarose affinity columns. Throughout the course of purification, recombinant CHK coeluted with protein-tyrosine kinase activity as well as the "Hck inhibition" activity (measured by the inactivation assay). Fig.  1, A and B, illustrate co-elution of CHK with these activities in the gel-filtration chromatography. The co-elution indicates that the recombinant CHK was catalytically active and confirms that Hck was inactivated by CHK and not by any low molecular mass contaminants in the CHK preparation. Furthermore, no protein-tyrosine phosphatase activity and negligible ATPase activity (less than 1% of the total ATP present in the Hck inactivation assay was hydrolyzed in 30 min) were detected in the CHK-containing column fractions, indicating that the inhibition of Hck activity was not a result of dephosphorylation of Hck by contaminating phosphatases nor was it due to depletion of ATP. The purity and authenticity of the purified CHK are depicted in Fig. 1C.
CHK Can Specifically Phosphorylate Tyr T of Hck by Following a Non-Michaelis-Menten Kinetics Mechanism-To confirm that CHK could specifically phosphorylate the consensus Tyr T near the C termini of SFKs, we generated the kinase-dead Hck(K267M) mutant as an in vitro substrate of CHK. Because Hck(K267M) lacks kinase activity, its use as the CHK substrate eliminates the complication of additional tyrosine phosphorylation due to Hck autophosphorylation. As shown in Fig.  2A, despite the existence of multiple tyrosine residues in Hck(K267M), CHK phosphorylates this Hck mutant exclusively at Tyr T (Tyr-499), suggesting that CHK can inactivate Hck and possibly other SFKs by phosphorylating the consensus Tyr T . To characterize the kinetics of Hck phosphorylation by CHK, we compared the initial velocities of CHK phosphorylation of Hck(K267M) and those of CHK phosphorylation of the SFK C-terminal peptide (Fig. 2, B and C). As shown in Fig. 2B, phosphorylation of the SFK C-terminal peptide conforms to Michaelis-Menten kinetics. A double reciprocal transformation of the data (1/V versus 1/[S]) yields the V max and K m values of 16.5 Ϯ 1.4 pmol/min/g of CHK and 1.1 Ϯ 0.1 mM, respectively (inset). In contrast to the phosphorylation of the SFK C-terminal peptide, increasing the Hck(K267M) concentration gives rise to a higher than expected increase in the rate of Tyr T phosphorylation by CHK. For example, doubling the Hck(K267M) concentration from 2.9 to 5.8 M resulted in an ϳ2.5-fold increase in initial velocity (inset, Fig. 2C). A doublereciprocal transformation of the data yields a non-linear plot (data not shown), indicating that CHK phosphorylation of Hck Tyr T does not follow the Michaelis-Menten kinetic model.
CHK and Hck Form a Tight Complex-The non-Michaelis-Menten kinetics of Tyr T phosphorylation by CHK indicates that, in addition to the conventional substrate-enzyme interactions (i.e. binding of Tyr T of Hck(K267M) to the CHK active site), Tyr T phosphorylation by CHK is also governed by noncatalytic interactions between the two proteins. To examine the validity of this hypothesis, we used co-immunoprecipitation to assess whether the two proteins could bind together to form a stable complex. Fig. 2D shows that CHK bound Hck(K267M) in the presence and absence of ATP and that the complex was stable to extensive washing with a buffer containing 0.5 M NaCl. As the concentration of Hck(K267M) was increased in the incubation mixture, the amount of Hck(K267M) co-immunoprecipitated with CHK also increased. The ability of CHK to bind Hck(K267M) even in the absence of Mg 2ϩ -ATP further suggests that the CHK⅐Hck(K267M) complex formation was mediated by non-catalytic interactions between the two proteins, i.e. it was not mediated by the binding of Tyr T of Hck(K267M) to the active site of CHK. Conceivably, the interactions increased the effective concentrations of CHK and Hck(K267M) and in turn led to a higher than expected increase in the rate of Tyr T phosphorylation by CHK, accounting for the deviation of the phosphorylation kinetics from the conventional Michaelis-Menten model. Similar experiments were conducted to ascertain if CHK could bind wild type Hck. As shown in Fig.  3, CHK was present in the anti-Hck immunoprecipitate. In the reciprocal co-immunoprecipitation experiment, Hck was detected in the anti-CHK immunoprecipitate. All these data verify direct binding of CHK to Hck and its K267M mutant in vitro.
CHK Can Inactivate Hck in Vitro, and the Inactivation Correlates with the Suppression of Autophosphorylation at Tyr A and Tyr-29 but Not with Extensive Phosphorylation at Tyr T -Autophosphorylation of Tyr A (Tyr-388) and Tyr-29 is the major mechanism of stimulating Hck activity (4). Therefore, we examined how CHK affected the degree of Hck phosphorylation at these sites. As shown in Fig. 4A, the presence of increasing concentrations of CHK correlates with the extent of Hck inactivation. At the CHK concentration of 7.2 M, the catalytic activity of Hck was suppressed by more than 95%. Accompanying the Hck inactivation, the increase in CHK concentration resulted in a significant drop in the total phosphorylation level of Hck (Fig. 4B). Because the total phosphorylation level of Hck is contributed by Tyr A and Tyr-29 autophosphorylation as well as Tyr T phosphorylation by CHK, we used phospho-specific antibodies to examine the extents of autophosphorylation at the two sites at varying CHK concentrations. We also employed phosphopeptide mapping to monitor phosphorylation of Hck at all three sites (i.e. Tyr A , Tyr-29, and Tyr T ) in the presence of 0 and 7.2 M CHK.
Western blotting of anti-Tyr(P) A and anti-Tyr(P)-29 antibodies shows that the inactivation of Hck correlated with the decrease in the levels of phosphorylation of Tyr A and Tyr-29 ( Fig. 4C). At the CHK concentration of 7.2 M, at which the Hck activity was less than 5% of the control Hck activity (i.e. Ͼ95% inactivation), autophosphorylation at Tyr A and Tyr-29 was suppressed to less than 5% that of the level of Hck autophosphorylation observed in the absence of CHK. To assess the effect of the CHK-mediated Tyr T phosphorylation on Hck autophosphorylation levels, we compared the degree of [␥-32 P]ATP incorporation at the three phosphorylation sites by two-dimensional phosphopeptide mapping. In the absence of CHK, Hck readily underwent autoactivation by autophosphorylation at Tyr A and Tyr-29, reaching a stoichiometry of 0.42 mol of phosphate/mol of Hck in 30 min (Fig. 4B). The presence of increasing concentrations of CHK led to a significant decrease in both the catalytic activity and the total phosphorylation level of Hck (Fig. 4, A and B). Stoichiometry measurement (Fig. 4B) and phosphopeptide mapping (Fig. 4D) of the phosphorylated Hck in the presence of 7.2 M CHK revealed that the level of phosphorylation of Hck at all sites in total was only 0.03 mol of phosphate per mol of Hck, indicating that the stoichiometry of Tyr T phosphorylation must be very low (Ͻ0.03 mol of phosphate/mol of Hck). Despite such a low level of Tyr T phosphorylation, CHK at this concentration induced 96% inactivation of Hck (Fig. 4A). This clearly demonstrates that Tyr T phosphorylation by CHK contributes very little to the Hck inactivation we observed.

CHK Forms a Stable Complex with Hck and Inactivates Hck in HEK293T Cells Even When Tyr T of Hck Is Replaced by
Phe-The in vitro data presented so far unequivocally demonstrate that Hck was inactivated by CHK by a novel non-catalytic mechanism involving the formation of a stable CHK⅐Hck complex that occurred without significant Tyr T phosphorylation. To examine if Hck and CHK could form a stable complex in mammalian cells, the two proteins were co-expressed in was monitored by Western blotting. As shown in Fig. 5A, CHK specifically co-immunoprecipitated with Hck, demonstrating that CHK and Hck formed a stable protein complex in mammalian cells.
Because CHK could specifically phosphorylate Tyr T of Hck, it is logical to postulate that binding of the Tyr T of Hck to the active site of CHK contributes to the CHK⅐Hck complex formation. To examine the validity of this hypothesis, we co-expressed CHK and the constitutively active Hck(Y499F) mutant (of which Tyr T Tyr-499 was replaced by phenylalanine) in HEK293T cells and determined if CHK and the Hck mutant could form a stable complex. As shown in Fig. 5A, despite the replacement of Tyr T by phenylalanine, the Hck mutant still retained the ability to form a stable complex with CHK in HEK293T cells, indicating that the CHK⅐Hck complex formation did not require a tyrosine at this site. The result further confirms that the CHK⅐Hck complex formation was mediated by Tyr T -independent non-catalytic interactions between the two proteins.

Overexpression of CHK Suppresses Autophosphorylation and Autoactivation of Wild Type Hck and Hck(Y499F) in HEK293T
Cells-As shown in Fig. 5B, overexpression of CHK induced a small decrease (17%) in Hck activity. We previously demonstrated that most of the recombinant Hck expressed in HEK293T cells was already in the inactive configuration as it was phosphorylated at Tyr T by the endogenous CSK (4). Consequently, the degree of reduction in total Hck activity caused by CHK overexpression is small because the available pool of active Hck was low. Nevertheless, Fig. 5B shows that CHK co-expression suppressed Hck autophosphorylation at Tyr A and Tyr-29, indicating that CHK could suppress Hck activation and prevent it from undergoing autophosphorylation. Because Hck(Y499F) is not subject to inhibition by Tyr T phosphorylation by the endogenous CSK, it was significantly more active than the wild type Hck (Fig. 5, B-D). Co-expression of CHK, however, resulted in a dramatic reduction in Hck(Y499F) autophosphorylation at Tyr A and Tyr-29 and a decrease (Ͼ87% reduction) in the catalytic activity of the Hck mutant. These results confirm the hypothesis that CHK can inactivate Hck by a mechanism other than Tyr T phosphorylation. Because Hck(Y499F) cannot be phosphorylated by CHK but could form a stable complex with CHK in HEK293T cells (Fig. 5A), we postulate that the inactivation of Hck(Y499F) is solely a consequence of the binding of CHK to the Hck mutant.
The K221M Mutation of CHK Abolishes Its Ability to Bind and Inactivate Hck-Because the binding and inactivation of Hck by CHK is independent of Tyr T phosphorylation, it is logical to expect that the same effect can be observed when a CHK mutant lacking tyrosine kinase activity is used in the experiment. To examine the validity of this notion, the kinasedead CHK(K221M) mutant was generated. Lys-221 in CHK is homologous to the conserved lysine residue critical for ATP binding as well as maintenance of the active configuration of many protein kinases. As expected, the replacement of Lys-221 by methionine rendered the recombinant CHK mutant inactive (data not shown).
In contrast to wild type CHK, CHK(K221M) is unable to form a stable complex with the activated Hck(Y499F) mutant in HEK293T cells (Fig. 5A). Furthermore, overexpression of CHK(K221M) has no significant negative impact on the protein kinase activity and the Tyr A and Tyr-29 phosphorylation levels of Hck(Y499F) co-expressed in HEK293T cells (Fig. 5D). The results strongly suggest that the K221M mutation completely abolishes CHK ability to bind and inactivate SFKs. However, it is possible that the mutation causes gross unfolding of the CHK mutant, and as such the recombinant mutant fails to inactivate Hck(Y499F) in HEK293T cells. To address this possibility, we expressed CHK(K221M) in Sf9 cells and purified the recombinant CHK mutant for biochemical and biophysical studies. To ascertain if CHK(K221M) folds correctly, we examined the secondary structure of CHK and CHK(K221M) by circular dichroism. The circular dichroism spectra of CHK and CHK(K221M) indicate similar secondary structure content (data not shown). Furthermore, the CHK mutant remained soluble throughout the course of its purification, and it displayed chromatographic properties very similar to those of wild type CHK. All these results indicate that the K221M mutation does not cause gross unfolding of CHK. Similar to what was observed in HEK293T cells co-expressing Hck(Y499F) and the CHK mutant, the K221M mutation abolishes the ability of CHK to both bind and inactive Hck in vitro (Fig. 6). Taken together, the results shown in Figs. 5 and 6 suggest that the K221M mutation structurally perturbs or reduces the accessibility of the determinants mediating CHK interaction with and inactivation of Hck.
CHK Can Bind and Inactivate Autophosphorylated Hck-Although Fig. 4 shows that CHK can inactivate Hck by suppressing its autophosphorylation, it remains unclear whether CHK can inactivate the fully active autophosphorylated form of Hck. This issue is relevant to many physiological and pathological situations in which stimulation of growth factor receptors leads to autophosphorylation and activation of the Srcfamily kinases. The fully active autophosphorylated form of the Src kinases mediates the cell growth signals originating from the stimulated receptors. Fig. 7 shows the result of our investigation into this question. In this experiment, Hck was allowed to undergo autophosphorylation by incubation with [␥-32 P]ATP and assay buffer for 1 h. The autophosphorylation was stopped by diluting the reaction mixture with the stopping buffer containing EDTA to chelate the Mg 2ϩ and Mn 2ϩ essential for the autophosphorylation. Aliquots of the diluted autophosphorylated Hck preparation were then assayed for peptide kinase activity in the absence and presence of increasing concentrations of CHK (Fig. 7A). In addition, the assay mixture was used for the co-immunoprecipitation experiment to ascertain if CHK could bind the autophosphorylated Hck. As shown in Fig. 7B, autophosphorylated Hck co-immunoprecipitated with CHK, indicating that CHK can form a tight complex with the autophosphorylated Hck. More importantly, the binding effectively inactivates the autophosphorylated Hck, with near complete inactivation of the autophosphorylated Hck occurring when incubated with 10 M CHK (Fig. 7A). The results indicate that CHK can bind and inactivate autophosphorylated Hck.

CHK Inactivation of Lyn and Src Is Associated with Tyr T Phosphorylation and the Formation of Stable CHK⅐Lyn and
CHK⅐Src Complexes-To ascertain if the mechanism employed by CHK to inactivate Hck is also applicable to CHK inactivation of other SFKs, we studied in vitro whether CHK can specifically phosphorylate Tyr T of Lyn and Src and form stable complexes with them.
As shown in Fig. 8A, CHK effectively inactivates Lyn protein kinase activity. At CHK concentrations (Ն0.37 M) that are higher than that of Lyn (0.23 M), CHK is capable of suppressing more than 90% of Lyn kinase activity. In contrast to the effect of CHK on Hck (Fig. 4B), CHK causes a significant increase in the stoichiometry of Lyn phosphorylation (Fig. 8B). Because CHK induces a reduction in Lyn Tyr A phosphorylation level (Fig. 8C), the increase in stoichiometry of Lyn phosphorylation in the presence of CHK is very likely attributed to a significant increase in phosphorylation at Tyr T . Phosphopeptide mapping of the phosphorylated Lyn digested sequentially by trypsin and chymotrypsin (Fig. 8D) was performed to fur- Lysates of the transfected cells were analyzed for Hck and Hck(Y499F) kinase activity using Src-optimal peptide. The specific activities of Hck and its mutant were expressed as pmol of phosphate incorporated/ min/densitometry unit of anti-Hck immunoreactivity. The autophosphorylation status of Hck and its mutant was monitored by anti-Tyr(P) A (Anti-pY A ) and anti-Tyr(P)-29 immunoblotting. Anti-Hck immunoblotting was also performed to confirm that equal amounts of Hck and its mutant were present. The multiple immunoreactive bands are consistent with the different degrees of mobility, attributed to various extents of Hck phosphorylation at Tyr-29, Tyr A , and Tyr T (4). Densitometric analysis of the immunoreactive bands reveals that co-expression of CHK with wild type Hck led to a 26 and 61% reduction in its Tyr(P) A and Tyr(P)-29 levels, respectively, whereas co-expression of Hck(Y499F) with CHK resulted in 67 and 55% reduction in its Tyr(P) A and Tyr(P)-29 levels, respectively. Wt, wild type. ther examine the effect of CHK on Lyn autophosphorylation and phosphorylation at Tyr T . Phosphopeptide maps of the phosphorylated Lyn show that CHK induces a decrease in Lyn autophosphorylation at Tyr A , whereas it induces a significant increase in Tyr T (Tyr-508) phosphorylation (Fig. 8D).
We recently discovered that Lyn can undergo autophosphorylation in vitro at Tyr A Tyr-397 in the kinase domain as well as at a novel site Y N (Fig. 8). 3 In addition to suppressing autophosphorylation at Tyr A , CHK also induces a reduction in Lyn autophosphorylation at Y N (Fig. 8).
Our data, therefore, indicate that both phosphorylation at Tyr T and the suppression of autophosphorylation are mechanisms employed by CHK to inactivate Lyn. It is noteworthy that at 0.37 M, CHK suppresses more than 90% of Lyn kinase activity (Fig. 8A), and densitometric analysis of the corresponding phosphopeptide map (Fig. 8D) reveals that at this CHK concentration only 19% of total Lyn phosphorylation (0.68 mol of phosphate incorporated/mol of Lyn (Fig. 8B)) is contributed by Tyr T phosphorylation. From these values, the stoichiometry of Lyn phosphorylation at Tyr T is estimated to be 0.13 mol of phosphate incorporated/mol of Lyn. It is obvious from this low level of Tyr T phosphorylation by 0.37 M CHK that Tyr T phosphorylation alone cannot fully account for the CHK ability to almost completely suppress Lyn activity. Furthermore, the re-quirement of CHK at concentrations higher than that of Lyn for effective suppression of Lyn activity implies that the inactivation process involves binding of CHK to Lyn to form the CHK⅐Lyn complex. Indeed, Lyn is co-immunoprecipitated with CHK as a stable CHK⅐Lyn complex (Fig. 8E). Analysis of Lyn in the CHK⅐Lyn complex by anti-Tyr(P) A Western blotting reveals that CHK can bind to Lyn even when Lyn is autophosphorylated at Tyr A (Fig. 8E), indicating that CHK can bind Lyn regardless of its Tyr A phosphorylation status. In summary, our data indicate that CHK can inactivate Lyn in vitro both by phosphorylating its Tyr T and by the non-catalytic Tyr T -independent mechanism that involves formation of the stable CHK⅐Lyn complex.
Similar to Hck and Lyn, Src is efficiently inactivated by CHK (Fig. 9A). Intriguingly, incubation of Src with increasing concentrations of CHK leads to a significant increase in the level of autophosphorylation (Fig. 9, B and C). As shown in the phos- , and the residual Hck catalytic activity was determined by the inactivation assay procedures. B, demonstration of stable complex formation between CHK and autophosphorylated Hck. CHK was immunoprecipitated from the preincubated reaction mixture, and the presence of Hck in the immunoprecipitate (IP) was assessed by anti-Tyr(P) A (Anti-pY A ), anti-Tyr(P)-29, and anti-Hck Western blotting (WB). The amount of CHK in each immunoprecipitate was also monitored. It is noteworthy that the amount of anti-CHK antibody used in this study is 3.8-fold higher that of the previous experiment (Fig. 3). The higher amount of antibody used led to the appearance of IgG heavy chain in all blots.
phopeptide maps (Fig. 9D), Src undergoes autophosphorylation mainly at Tyr A . Reminiscent of a previous observation by MacAuley et al. (22), the phosphopeptide map also reveals Src autophosphorylation at Tyr T , albeit only to a very low level. However, the presence of CHK only causes a small increase in Tyr T phosphorylation (Fig. 9D). Thus, despite the increased level of Tyr A autophosphorylation, CHK effectively inactivates Src without significant Tyr T phosphorylation. The result strongly suggests that CHK also employs the non-catalytic mechanism to inactivate Src. Indeed, results of the co-immu- A, the residual Lyn kinase activity was monitored by the inactivation assay procedure. B, the stoichiometry of Lyn phosphorylation at all sites (Y N , Tyr A , and Tyr T ) was determined. C, the relative level of Lyn autophosphorylation at Tyr A was monitored by anti-Tyr(P) A (Anti-pY A ) Western blotting. D, tryptic/chymotryptic maps of Lyn phospho rylation with varying concentrations of CHK. The origin of each map is marked by an asterisk. The key represents the migration patterns of tryptic/chymotryptic phosphopeptides derived from Tyr(P) N , Tyr(P) A , and Tyr(P) T . TLE, thin layer electrophoresis; TLC, thin layer chromatography. E, demonstration of stable complex formation between CHK and Lyn. The reaction mixture containing Lyn phosphorylated in the absence and presence of CHK was subjected to anti-CHK immunoprecipitation (IP). Lyn and its associated Tyr A phosphorylation were verified by Western blotting (WB). The amounts of CHK in the immunoprecipitates were also revealed by immunoblotting.
FIG. 9. CHK inactivates Src by phosphorylating its Tyr T and forming a stable CHK⅐Src complex. Src (0.48 M) was incubated at 30°C for 30 min with varying concentrations of CHK (0 -10 M) in a final volume of 25 l. A, the residual Src kinase activity was monitored by the inactivation assay procedure. B, the stoichiometry of Src phosphorylation at all sites (Tyr A and Tyr T ) was determined. C, the relative level of Src autophosphorylation at Tyr A was monitored by anti-Tyr(P) A (Anti-pY A ) Western blotting. D, tryptic phosphopeptide maps of Src phosphorylated in the absence and presence of varying CHK concentrations. The key shows the migration patterns of Tyr(P) A -and Tyr(P) T -derived tryptic fragments, which have been previously identified (13). The origin of each map is marked by an asterisk. TLE, thin layer electrophoresis; TLC, thin layer chromatography. E, demonstration of stable complex formation between CHK and Src. The reaction mixture containing Src phosphorylated in the absence and presence of CHK was subjected to anti-CHK immunoprecipitation. The presence of Src in the immune complex together with its degree of autophosphorylation were monitored by Western blotting (WB). Likewise, the amounts of CHK in the immunoprecipitates (IP) were also monitored. noprecipitation experiment (Fig. 9E) reveal that CHK forms a stable complex with Src. Western blot analysis of the CHK⅐Src complex with the anti-Tyr(P) A antibody reveals that CHK is capable of binding to Src even when Src is autophosphorylated. Thus, similar to CHK inactivation of Hck and Lyn, CHK is capable of inactivating Src both by phosphorylating its Tyr T and by the non-catalytic mechanism that involves binding of CHK to Src to form the stable CHK⅐Src complex.
CHK and Src Co-localize to Specific Microdomains of the Plasma Membrane, and They Form Tight CHK⅐Src Complexes in Rat Brain Cells-An important issue associated with inactivation of SFKs by CHK in vivo relates to their subcellular localizations; SFKs, containing fatty acid moieties at the N terminus, are bound to the plasma membrane, whereas CHK, lacking obvious structural features for its targeting to the plasma membrane, is expected to reside in the cytosol. Further-more, micromolar concentrations of CHK, Hck, Lyn, and Src are needed for CHK to bind and inactivate the SFKs (Figs. 4, 6 , 7, 8, and 9) in vitro. This suggests that efficient inactivation of SFKs by CHK entails their co-localization such that their effective concentrations are significantly increased. Thus, it is important to ask the question, do CHK and SFKs co-localize to specific microdomains of the plasma membrane? Because CHK and Src are co-expressed in brain cells (14), we therefore studied their subcellular localization and whether the CHK⅐Src protein complexes exist in rat brain. As shown in Fig. 10A, both Src and CHK are present in the plasma membrane fraction of rat brain cells. Further purification by sucrose density gradient ultracentrifugation separates the membrane preparation into the low density and high density plasma membrane microdomains. Western blot analysis of the fractionated membrane reveals that both CHK and Src co-localize to both microdo- FIG. 10. Co-localization of CHK and Src to the plasma membrane and the existence of a stable CHK⅐Src complex in rat brain cells. A, distributions of Src and CHK in the plasma membrane and cytoplasm of rat brain. The plasma membrane and cytoplasm were separated from the crude brain lysate by centrifugation. Immunoblotting (WB) was subsequently performed to analyze the distributions of Src and CHK. Because flotillin has been reported to reside primarily in the plasma membrane (40), it was employed as a marker to identify the isolated membrane fraction. B, distributions of Src and CHK in the specific membrane microdomains with various lipid contents. After sucrose gradient ultracentrifugation of crude rat brain membrane preparation, 12 gradient fractions were harvested from the top of the centrifuge tube, with fractions 2-5 and 9 -12 corresponding to the low and high density membrane microdomains, respectively. All gradient-fractions were again centrifuged. The pellets of the individual fractions were treated with Triton X-100 before an additional round of centrifugation, which gave rise to the Triton X-100-soluble and -insoluble portions. The distribution patterns of Src and CHK in the specific membrane microdomains were analyzed by immunoblotting. C, demonstration of the existence of a stable CHK⅐Src complex by co-immunoprecipitation. 1% of the total crude brain lysate was applied to protein A-Sepharose beads cross-linked to the anti-CHK and anti-CSK antibodies. The beads were also cross-linked to IgG from preimmune serum and were used as a control for the immunoprecipitation analysis. The mixtures were rocked gently at 4°C for 2 h, and the immunoprecipitates (IP) were processed for Western blotting. MAb, monoclonal antibody. mains (Fig. 10B). Conceivably, co-localization of CHK and Src in these membrane microdomains significantly increases their effective concentrations and in turn facilitates CHK inactivation of Src. Our finding is in agreement with the observation made by Chow et al. that CHK localizes to the detergentresistant membrane microdomain (23).
The results presented in Fig. 5 show that CHK can bind and inactivate Hck when they are co-expressed at high levels in HEK293T cells. However, it is still possible that the CHK⅐Hck complex formation is an artifact resulting from their overexpression. To address this possibility, we need to demonstrate that native stable CHK⅐SFK complexes exist in cells or tissues where these kinases are co-expressed. As shown in Fig. 10C, Src is present in the anti-CHK immunoprecipitate but not in the preimmune serum immunoprecipitate, suggesting that endogenous CHK and Src form stable protein complexes in brain cells. Because CSK is also expressed in rat brain cells and recently Lee et al. (24) demonstrated that CSK could bind Src in vitro, we therefore examined if CSK could bind Src or to the CHK⅐Src complexes in vivo. As shown in Fig. 10C, Src is not present in the anti-CSK immunoprecipitate. Furthermore, we failed to detect CSK in the anti-CHK immunoprecipitate of brain lysate (data not shown).

DISCUSSION
It is well documented that CHK can down-regulate the activity of SFKs in vivo (25)(26)(27)(28)(29). For example, CHK inactivation of Src and Lyn was associated with suppression of breast carcinoma cell growth (29). Because of the significant sequence similarity displayed by CHK and CSK, CHK was postulated to follow the same mechanism as CSK in inactivating SFKs, i.e. CHK inactivates SFKs exclusively by phosphorylating their consensus C-terminal regulatory tyrosine Tyr T . In this manuscript we report that CHK can specifically phosphorylate Tyr T of three SFK members, Hck, Lyn, and Src, and that CHK can suppress their catalytic activity. Intriguingly, we demonstrated both in vitro and in transfected HEK293T cells that CHK can suppress SFK activity by a novel mechanism that does not require Tyr T phosphorylation. This novel mechanism involves direct binding of CHK to SFKs and suppression of their kinase activity toward exogenous peptide substrates. Further analysis reveals that binding and suppression of SFK activity are independent of Tyr T , as CHK could still bind and effectively inactivate the Hck(Y499F) mutant, where the Tyr T was substituted with phenylalanine. Moreover, we found that CHK and Src co-localize to specific microdomains in rat brain plasma membrane and that native CHK⅐Src protein complexes exist in rat brain cells. These findings, therefore, support the physiological relevance of our in vitro observations as co-localization of CHK and Src would increase their effective concentrations and in turn facilitate CHK phosphorylation of Src and binding to Src by the Tyr T -independent mechanism.
What is the physiological significance of CHK ability to inactivate SFKs by this novel non-catalytic mechanism? SFKs are known to be constitutively active and autophosphorylated in cancer cells (for review, see Ref. 30). Furthermore, autophosphorylation permits SFKs to remain active even when their Tyr T is phosphorylated (31). Recently, Lerner and Smithgall (32) reported that human immunodeficiency virus Nef binding or mutations of the SH2 kinase linker caused Hck activation by autophosphorylation even when Tyr T remains phosphorylated and bound to the SH2 domain. All these data indicate that Tyr T phosphorylation and the intramolecular Tyr(P) T -SH2 interaction may not be sufficient to completely inhibit SFK activity in vivo. Thus, this novel non-catalytic mechanism, which both inactivates the autophosphorylated Hck and prevents Hck from undergoing autophosphorylation, represents a fail-safe cellular mechanism to suppress SFK signaling under all circumstances.
Despite the high degree of specificity displayed by CHK in phosphorylating Tyr T of SFKs (Figs. 2, 4, 8, and 9), its efficiency in phosphorylating SFKs in vitro is very low. For example, CHK phosphorylated Hck to a stoichiometry of less than 0.03 mol of phosphate/mol of Hck even when CHK was present in a 3-4-fold excess. Similarly, Src is a poor in vitro substrate of CHK (Fig. 9D). Lyn appears to be a better substrate (Fig.  8D), but even so, Lyn Tyr T was phosphorylated only to a stoichiometry of 0.13 mol of phosphate/mol of Lyn when CHK was present in a 1.6-fold excess. Our results suggest that CHK either needs to be activated or requires the assistance of other cellular proteins to enhance its efficiency in phosphorylating Tyr T of SFKs.
It is intriguing that CHK binds Src and suppresses its phosphorylation of exogenous peptide substrates and yet it enhances Src autophosphorylation (Fig. 9C). Nevertheless, despite the elevated Src autophosphorylation, CHK can still bind and suppress Src phosphorylation of exogenous peptide substrates (Fig. 9, A and E). How might CHK binding enhance Src autophosphorylation? Relevant to this question, Wang et al. (33) report that Src can undergo dimerization. We postulate that CHK binding also facilitates Src dimerization and this in turn enhances Src trans-autophosphorylation. But how CHK binding suppresses Src kinase activity toward the exogenous peptide substrates while it permits Src trans-autophosphorylation, is unclear.
Although stable CHK⅐Src protein complexes exist in rat brain, we failed to detect stable CSK-Src complexes, suggesting either that the CSK-Src complexes are much less abundant than the CHK⅐Src complexes in rat brain or that the CSK-Src complexes are of transient existence in brain cells. Recently, Lee et al. (24) were able to demonstrate the formation of a stable CSK-Src complex in vitro when both kinases were at high concentrations. Given the high degree of sequence identity of CHK and CSK (Ͼ50% in their overall sequences and Ͼ70% in their kinase domains), it is possible that both CHK and CSK use similar motifs to bind SFKs.
In an attempt to elucidate the structural basis of the stable association between CHK and SFKs, we revealed that direct binding of CHK to Hck was abolished when Lys-221 in the active site of CHK was mutated to methionine. Based upon the crystal structures of other protein kinases (6,8,9), Lys-221 in ␤-strand 3 of the kinase domain is predicted to fulfill two structural requirements; (i) it binds to the ␣ and ␤ phosphates of ATP, and (ii) it forms a salt bridge with the conserved Glu-235 on ␣-helix C in the minor lobe of the kinase domain. Our results show that CHK⅐SFK complexes were maintained even in the absence of ATP (Fig. 2D), indicating that interactions of Lys-221 with ATP play no role in directing CHK to bind SFKs. This implies that formation or accessibility of the motifs critical for SFK binding and inhibition are dependent upon interactions of Lys-221 with other residues of the kinase domain, e.g. formation of the Lys-221-Glu-235 salt bridge.
Because CHK and CSK display significant sequence identity, it is logical to postulate that the three-dimensional structures of the two enzymes also share a high degree of similarity. For this reason we turned to the recently published x-ray crystal structure of CSK for clues to explain how the K221M mutation completely abolishes the CHK ability to bind and inactivate Hck. The unit cell of the CSK crystal structure contains six copies of CSK; four adopt a configuration in which the corresponding Lys-Glu salt bridge is formed, and two adopt configurations in which the salt bridge is broken (Fig. 11). There are significant conformational differences between the two forms.
In the minor lobe of the kinase domain, ␤-strand 1, ␤-strand 2, ␤-strand 3, the ␤3-␣C loop, and ␣-helix C are displaced. Also, the two lobes of the kinase domain change their relative disposition, and the SH2 domain and its flanking linkers move with respect to the kinase domain (Fig. 11). We suggest that such conformational differences could also exist for CHK. Thus, the disruption of the Lys-221-Glu-235 salt bridge by the K221M mutation could have far-reaching effects on the conformation of CHK that limit the accessibility or prevent the formation of the motifs critical for SFK binding and inhibition. Recently, Lee et al. (24) identified a number of residues in and around ␣-helix D in the major lobe of the kinase domain of CSK form. CHK inhibits SFK activity by binding to the latent form and in turn preventing it from undergoing autophosphorylation. Moreover, CHK can bind to the fully active autophosphorylated state Tyr(P) A [Tyr T ], and the binding alone is sufficient to inhibit SFK activity. In addition to inhibiting SFK activity upon direct binding, CHK can also inhibit its activity by phosphorylating its Tyr T . The ability of CHK to bind and inhibit both autophosphorylated and latent forms of SFKs suggests that CHK is capable of providing a fail-safe mechanism to shut down SFK signaling regardless of the phosphorylation status of SFKs. The question mark indicates that the mechanisms of efficient Tyr T phosphorylation and dissociation of the CHK⅐Hck complex remain unknown.

FIG. 11. Conformational changes in CSK.
A, schematic representation of the conformation adopted by CSK (PDB 1K9A chain A) when the salt bridge is formed between the residues corresponding to Lys-221 and Glu-235 in CHK. The side chains involved are shown in stick mode and are labeled Lys and Glu, respectively. The relative position of selected secondary structure elements in the kinase domain (␤1, ␤1-␤2 loop, ␤2, ␤3, ␤3-␣C loop, ␣C in the minor lobe and ␣D in the major lobe) are shown within an outline of the whole molecule. B, conformation adopted by CSK when the salt bridge is broken (PDB 1K9A chain C). The two images were generated with the major lobe of the kinase domain in the same orientation. as parts of a binding motif responsible for recognition of SFK substrates. It is possible that this motif is also used by CHK to bind and inactivate SFKs by the non-catalytic mechanism. Indeed, this site is relatively close to residues in ␤1 (ϳ7-8 Å) that undergo a large conformational change between the two aforementioned forms, which could explain the loss of binding after the K221M mutation in CHK.
Based upon the data presented in this report, we suggest the following model for the mechanism of CHK inactivation of SFKs (See Fig. 12). Inactive SFKs exist as a compact intramolecular complex, where the interaction of the Tyr T -phosphorylated C-terminal tail with the SH2 domain and the interaction of the SH2 kinase linker with the SH3 domain stabilizes the inactive configuration of the kinase domain (the [Y A ][pY T ] form in Fig. 12). Receptor stimulation results in phosphorylation events that recruit the SFK to the signaling complex where it binds via its SH2 and/or SH3 domains. Ligand binding to the SH2 and/or SH3 domain releases the kinase domain from the inactive configuration. The next steps in the course of activation of SFKs are (i) removal of the phosphate from Tyr T by a co-localized phosphatase to yield the non-phosphorylated, latent form of the kinase (the [Y A ][Y T ] form), and (ii) autophosphorylation at Tyr A to convert the kinase to its fully active, autophosphorylated form (the [pY A ][Y T ] form).
CHK in the cytoplasm is also recruited to the phosphorylated signaling complex via its SH2 domain (34). This brings CHK in close proximity to the active SFK. CHK then binds the active, autophosphorylated SFK and inhibits its catalytic activity. Furthermore, CHK can also bind the non-phosphorylated, latent form of the SFK, preventing the SFK from undergoing autoactivation by autophosphorylation (Fig. 12). Presumably, the binding is followed by an unknown activation event that stimulates CHK phosphorylation of Tyr T . The Tyr T phosphorylation would reconstitute the intramolecular SH2-binding site, and this would allow the SH2 and SH3 domains of the SFK to re-bind to their intramolecular ligands, displacing CHK and releasing the inactive SFK from the complex. We believe that, in addition to inactivation of SFKs by the intramolecular Tyr(P) T -SH2 domain interaction, CHK-SFK binding represents a new and universal mechanism of SFK inactivation.
Activation and elevated expression of SFKs have been associated with the development of many types of cancer including breast cancer and colon cancer (30,35). In this context our finding that CHK can inactivate SFKs by a novel non-catalytic mechanism that involves the formation of a tight CHK⅐SFK complex is particularly important because compounds that mimic the CHK ability to bind and suppress SFK activity are potential therapeutics for the treatment of breast cancer. Further investigations to define the structural determinants in both CHK and SFKs that mediate their stable interactions will facilitate the discovery of such compounds.