Direct association of Csk homologous kinase (CHK) with the diphosphorylated site Tyr568/570 of the activated c-KIT in megakaryocytes.

The Csk homologous kinase (CHK), formerly MATK, has previously been shown to bind to activated c-KIT. In this report, we characterize the binding of SH2CHK to specific phosphotyrosine sites on the c-KIT protein sequence. Phosphopeptide inhibition of the in vitro interaction of SH2CHK-glutathione S-transferase fusion protein/c-KIT from SCF/KL-treated Mo7e megakaryocytic cells indicated that two sites on c-KIT were able to bind SH2CHK. These sites were the Tyr568/570 diphosphorylated sequence and the monophosphorylated Tyr721 sequence. To confirm this, we precipitated native CHK from cellular extracts using phosphorylated peptides linked to Affi-Gel 15. In addition, purified SH2CHK-glutathione S-transferase fusion protein was precipitated with the same peptide beads. All of the peptide bead-binding studies were consistent with the direct binding of SH2CHK to phosphorylated Tyr568/570 and Tyr721 sites. Binding of FYN and SHC to the diphosphorylated Tyr568/570 site was observed, while binding of Csk to this site was not observed. The SH2CHK binding to the two sites is direct and not through phosphorylated intermediates such as FYN or SHC. Site-directed mutagenesis of the full-length c-KIT cDNA followed by transient transfection indicated that only the Tyr568/570, and not the Tyr721, is able to bind SH2CHK. This indicates that CHK binds to the same site on c-KIT to which FYN binds, possibly bringing the two into proximity on associated c-KIT subunits and leading to the down-regulation of FYN by CHK.


EXPERIMENTAL PROCEDURES
Materials-Tyrosine-phosphorylated and nonphosphorylated synthetic peptides were obtained from the Dana Farber Cancer Institute Molecular Biology Core Facility (Boston, MA). Peptides were analyzed for purity by high pressure liquid chromatography, mass spectroscopy, and amino acid analysis. SH2 CHK -GST fusion protein was prepared as described previously (12). PI3-kinase p85␣ SH2-GST fusion protein was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-GST monoclonal antibodies were produced against Schistosoma japonicum glutathione S-transferase according to the procedure of Harlow and Lane (18). Anti-CHK polyclonal antibody (anti-LSK), anti-SHC polyclonal antibody (C- 20), and anti-FYN polyclonal antibody were purchased from Santa Cruz Biotechnology. Anti-SHC monoclonal antibody S14620 and anti-p85 polyclonal antibody were purchased from Transduction Laboratories (Lexington, KY). Anti-c-KIT polyclonal and recombinant SCF/KL were generously provided by Dr. B. Bennett, Amgen Inc. (Thousand Oaks, CA). Interleukin-3 and GM-CSF were purchased from R & D Systems (Minneapolis, MN). All other reagents were from Sigma unless otherwise specified. Factor-dependent megakaryocytic Mo7e cells were grown in RPMI 1640 medium (Mediatech, Washington, D. C.) including 20% fetal bovine serum, 10 ng/ml interleukin-3, 10 ng/ml GM-CSF, 100 IU/ml penicillin, and 10 g/ml streptomycin. Prior to stimulation with SCF/KL, cells were starved for 15 h in RPMI 1640 medium with 1% fetal bovine serum, 100 IU/ml penicillin, and 10 g/ml streptomycin.
Peptide Inhibition of SH2 CHK -GST Fusion Protein/c-KIT Interaction-Serum-starved Mo7e cells (2 ϫ 10 6 /ml) were stimulated with 500 ng/ml SCF/KL at 37°C for 2 min and centrifuged for 2 min at 2200 rpm. Cell pellets were placed on ice and solubilized in 20 mM Tris-HCl, pH 7.4, 137 mM NaCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 10% glycerol, 1.0% Nonidet P-40, 1 mM Na 3 VO 4 , 2 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, and 10 g/ml aprotinin (lysis buffer). Cell lysates (40 ϫ 10 6 cells/ml) were mixed briefly and centrifuged for 15 min at 4800 rpm. During centrifugation, SH2-GST fusion proteins (10 g/incubation) were preincubated with 0.5 ml of lysis buffer along with synthetic peptides at twice the specified concentration. Cell supernatants (0.5 ml, equivalent to 20 ϫ 10 6 cells) were added to SH2 CHK -GST fusion protein plus peptide. Lysates were rotated during incubation for 1 h at 5°C and then precipitated by the addition of Glutathione-Sepharose 4B (Pharmacia Biotech Inc.) for 0.5 h at 5°C. Precipitates were washed by centrifugation and resuspension in 1 ml of lysis buffer (repeated three times). SDS sample buffer was added to washed precipitates, and the samples were run on 7.5% polyacrylamide SDS-PAGE. Proteins were transferred onto nitrocellulose filters (Bio-Rad). Transfers were blocked by 4% bovine serum albumin in 0.1% Tween 20, phosphate-buffered saline (PBST). Primary antibody incubation for 1.5 h was followed by washing in PBST. Horseradish peroxidase-linked secondary antibody incubation was in PBST for 45 min. After washing in PBST, transfers were treated with ECL chemiluminescent reagent (Amersham Corp.) and exposed to x-ray film.
Immunoprecipitation-Mo7e cell extract supernatants prepared as described above were also immunoprecipitated by specified antibodies. Incubations of 20 ϫ 10 6 cell equivalents with 1 g of affinity-purified antibody were for 1.5 h at 5°C with rotation. Protein G-Sepharose (Pierce) was added, and incubation was continued for 0.5 h. Precipitates were centrifuged and resuspended in 1 ml of lysis buffer (repeated three times). SDS sample buffer was added to precipitates, and samples were run on 10% or 7.5% polyacrylamide SDS-PAGE. Gels were transferred onto nitrocellulose and immunoblotting was conducted as in the previous section.
Association of SH2 CHK -GST or Native CHK with Peptide Beads-Phosphorylated and nonphosphorylated peptides were linked to Affi-Gel 15 by adding 1 mol of peptide in 40 mM MOPS buffer, pH 7.4, to a 0.5-ml bed volume of washed beads. After reaction for 1 h at room temperature, Tris-HCl, pH 8.0, was added to a final 50 mM concentration to block unreacted sites. After incubation overnight at 5°C, the resin was washed with 0.5 M NaCl, 50 mM MOPS, pH 7.4, followed by 50 mM MOPS, pH 7.4, and stored at 5°C with 1 mM sodium azide as a preservative. In order to test for peptide linkage, we measured the A 280 of the initial supernatant of the peptide-bead reaction and compared this to the A 280 prior to linkage, correcting for dilution. In all cases, 80 -90% of the peptide was linked to the beads. Peptide beads (15-l bead volume) were incubated with 10 g of SH2 CHK in 20 mM Tris-HCl, pH 7.4, 137 mM NaCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 10% glycerol, 1.0% Nonidet P-40, 1 mM Na 3 VO 4 , and 2 mM phenylmethylsulfonyl fluoride (association buffer). Incubation was for 1.5 h at 5°C. The beads were centrifuged and resuspended in 1 ml of association buffer (repeated three times). SDS sample buffer was added to the washed precipitates, and the samples were subjected to SDS-PAGE followed by transfer to nitrocellulose and immunoblotting as described above with anti-GST as the primary antibody. In order to test binding of native CHK to the peptide beads, 15 l of beads were incubated with 1 ml of Mo7e extract (20 ϫ 10 6 cells equivalent) prepared as described in the previous section. Extracts were incubated for 1.5 h at 5°C, with gentle mixing. Precipitates were treated with SDS sample buffer and subjected to SDS-PAGE/Western blotting with anti-CHK or other antibodies, followed by the appropriate secondary antibody and ECL development. Nitrocellulose filters were stripped for reblotting by incubating in 0.2 M glycine, 0.5 M NaCl, pH 2.8. After neutralization with Tris base, the blots were extensively washed with deionized water.

Transient Transfection of Wild Type and Mutant c-KIT Expression
Vectors-Wild type and mutant c-KIT pcDNA3 vectors were co-transfected into COS 7 cells along with a ␤-galactosidase pSV reporter plasmid (Promega, Madison, WI). For each 10-cm plate of COS 7 cells, 10 g of c-KIT pcDNA3 plus 5 g of ␤-galactosidase pSV DNA were co-transfected using LipofectAMINE reagent (Life Technologies, Inc.). After 5 h of transfection, full medium was added, and cells were continued in culture for another 48 -72 h. At this time, cells were serumstarved, stimulated with SCF/KL (10 min, 350 ng/ml), and harvested into lysis buffer (as described in the previous section). Supernatants were then subjected to precipitation by SH2 CHK -GST fusion protein as described in the previous section. Glutathione-Sepharose precipitates were resolved by SDS-PAGE and transferred to nitrocellulose followed by Western blotting with c-KIT polyclonal antibody. In each case, amounts of extract used in precipitation were normalized for the expression of ␤-galactosidase as measured by an O-nitrophenyl ␤-D-galactopyranoside reporter assay (Promega).

Inhibition of SH2 CHK /c-KIT Interaction by
Tyrosine-phosphorylated Peptides-In order to identify the tyrosine phosphorylation on c-KIT which is critical for SH2 CHK association, we synthesized a series of tyrosine-phosphorylated heptapeptides derived from the cytoplasmic portion of human c-KIT. Since SH2 domains usually associate with residues on the C-terminal side of phosphotyrosine, we chose peptide sequences with tyrosine as the N-terminal residue. We did not include phosphotyrosine sites within the two catalytic domains of c-KIT. These peptides were used to inhibit the interaction of c-KIT from SCF/KL-activated Mo7e cells and the SH2 CHK -GST fusion protein as described under "Experimental Procedures." Complexes of c-KIT and SH2 CHK -GST were shown by the presence of c-KIT in washed SH2 CHK -GST precipitates (Fig. 1, A and B). Of the peptides that inhibited complex formation, the most significant was the Y*MDMKPG peptide (Fig. 1A, lane 10) with approximately 80% inhibition of association at 100 M peptide. The nonphosphorylated YMDMKPG peptide failed to inhibit complex formation in this assay (Fig. 2, lane 5). We also found a slight inhibition by the peptide Y*VYIDPT (Fig. 1, lane 9).
Recently, it has been reported that the analogous peptide sequence in the PDGF receptor is a site for binding of FYN (15) and of SHC (16,17). In the case of FYN, binding was significantly greater when both tyrosines of the sequence were phosphorylated. We tested whether this would be the case with SH2 CHK binding to c-KIT. As shown in Fig. 2, the diphosphorylated peptide Y*VY*IDPT (lane 1) gave significantly greater inhibition than did the monophosphorylated Y*VYIDPT (lane 2). Furthermore, the monophosphorylated Y*IDPTQL failed to give significant inhibition (Fig. 1, lane 3). Thus, both phosphates are required for inhibition at this site. In order to test the relative abilities of Y*MDMKPG and Y*VY*IDPT to inhibit SH2 CHK /c-KIT interaction, various concentrations of peptide from 5 M to 100 M were tested (Fig. 3). Densitometry of c-KIT bands (not shown) indicated an IC 50 of ϳ50 M for both peptides. In a similar experiment substituting SH3-SH2 CHK for SH2 CHK , both peptides inhibited to the same extent but with a slightly lower IC 50 of 20 M. These results indicated that binding of CHK is primarily at the Tyr 721 and Tyr 568/570 sites on c-KIT.
Binding of SH2 CHK and Native CHK to Immobilized Peptides-In order to further test the binding of CHK to the phosphorylated Tyr 721 and Tyr 768/770 sites, we linked these peptides to Affi-Gel 15 beads, and the association of either SH2 CHK -GST fusion protein or native CHK from crude extracts of Mo7e cells was analyzed. As is shown in Fig. 4A, SH2 CHK -GST associated in a phosphotyrosine-dependent manner to Y*MDMKPG and Y*VY*IDPT linked to beads as shown by an anti-GST immunoblot of the precipitates. In a control precipitation, GST alone did not associate with the beads (Fig. 4B). A non-receptor peptide derived from the Tyr 416 phosphorylation site of FYN (IEDNEY*TARQGA) showed very little phosphate-dependent association to SH2 CHK . We also observed association of SH2 CHK to the phosphorylated C-terminal tyrosine peptide of SRC kinases (EPQY*QPGENL) (Fig. 4A, lane 8). When we looked at the association of native CHK to peptide beads as shown by anti-CHK immunoblotting, we observed a similar specificity (Fig. 5A). This specificity was in agreement with the peptide inhibition experiments and also indicated that the association was by direct binding of the SH2 CHK domain to the tyrosine-phosphorylated site(s) on c-KIT.
Possible Association of SH2 CHK to c-KIT through Other Com-ponents-We have shown by peptide inhibition that the PI3kinase SH2 p85␣ binds specifically to the Y*MDMKPG peptide as does SH2 CHK (data not shown). To confirm that CHK binding to c-KIT is not indirectly via PI3-kinase p85, we stripped and reblotted the transfers from Fig. 1 with anti-p85 subunit antibody (Fig. 6). The distribution of p85 in this blot was similar to the distribution of c-KIT seen in Fig. 1. Thus the lack of p85 in the Y*MDMKPG-inhibited precipitate indicates that both p85 and CHK are competing for the same site. If CHK was binding through p85, inhibition with Y*MDMKPG would dissociate only c-KIT from the complex and would retain p85 in the SH2 CHK -GST precipitate. In the case of the phosphorylated Tyr 568/570 site, we have demonstrated by immunoblotting that association of crude extracts to peptide beads resulted in not only CHK, but also FYN and SHC binding at this di-phosphorylated site (Fig. 5, B and C). Csk did not associate with the diphosphorylated site and only slightly associated with the monophosphorylated site (Fig. 5D, left panel). This result was observed in spite of having a large amount of Csk in the crude extract (Fig. 5D, right panel). In order to confirm that the CHK Washed precipitates were run on SDS-PAGE, and transfers were blotted with anti-GST antibody. Panel B, GST protein alone was incubated with the same phosphorylated and nonphosphorylated peptides linked to beads. As in panel A, washed precipitates were run on SDS-PAGE, and transfers were blotted with anti-GST antibody. association at this site was direct, and not through either FYN or SHC, we studied the association of Mo7e extracts to peptide beads with SCF/KL-activated and nonactivated extracts. As can be seen in Fig. 7, there was no observable difference in the association of CHK to the beads when comparing SCF/KLactivated and nonactivated extracts. Thus, it is unlikely that CHK associated with c-KIT through another component which itself was tyrosine-phosphorylated in the course of activation. Experiments attempting to co-precipitate SHC and CHK were negative (data not shown). Experiments attempting to inhibit the SH2 CHK /c-KIT receptor complex with phosphopeptides present in SRC kinases were also negative (data not shown), again emphasizing that FYN and SHC were not intermediaries in CHK/c-KIT binding.
Effect of Site-directed Mutagenesis of c-KIT on Association with SH2 CHK -In order to confirm the significance of the binding of SH2 CHK to the activated c-KIT at the Tyr 721 and Tyr 568/570 sites, we constructed mutant c-KIT cDNAs for either Tyr 721 3 Phe (designated 2162 c-KIT) or Tyr 568 3 Phe (designated 1702 c-KIT) and attempted to demonstrate whether these mutant proteins were still able to associate with the SH2 CHK . As can be seen in Fig. 8B, the wild type, Tyr 568 3 Phe (1702), and Tyr 721 3 Phe (2162) c-KIT constructs expressed comparable amounts of c-KIT protein as observed in c-KIT Western blots of crude extracts from the transfectants, when corrected for the extent of transfection as measured by the ␤-galactosidase reporter assay. When such extracts were precipitated by anti-c-KIT antibody and blotted with anti-phos-photyrosine antibody, comparable SCF/KL-dependent phosphorylation was observed for wild type and both mutant c-KIT proteins (data not shown). This indicated that the expressed protein is capable of ligand stimulated tyrosine autophosphorylation even with the replacement of the specific tyrosines by phenylalanines. When wild type c-KIT-expressing cell extracts were incubated with SH2 CHK -GST, c-KIT associated in an SCF/ KL-dependent manner as was observed with endogenous c-KIT (Fig. 8A, lanes 1 an 2). As expected, no c-KIT association was observed in the GST control precipitation (Fig. 8A, lane 8). In these experiments, when SH2 CHK was incubated with either of the two mutant c-KIT-expressing extracts, the SCF/KL-activated Tyr 568 3 Phe c-KIT was completely unable to associate (lanes 3 and 4), while the Tyr 721 3 Phe c-KIT still retained the ability to associate with SH2 CHK -GST in an SCF/KL-dependent manner (lanes 5 and 6). DISCUSSION In this report, we have identified the sites of binding of the SH2 CHK domain in the activated human c-KIT. Initial studies involving phosphopeptide inhibition indicated that it was  of panels A and B) were associated with the phosphorylated or nonphosphorylated peptides linked to beads as indicated above each lane. Phosphorylated peptide associations were conducted in duplicate, while nonphosphorylated peptide associations were done singly. Western immunoblotting was performed with anti-CHK antibody; the arrow indicates the position of the CHK protein.
the monophosphorylated Tyr 721 and the diphosphorylated Tyr 568/570 sites on c-KIT which were important in binding SH2 CHK . To test the relative affinities of the two sites, the concentration dependence of the two phosphopeptides for inhibition of the c-KIT/SH2 CHK interaction was assessed. Both peptides appeared to inhibit c-KIT/SH2 CHK equally well. As has been shown with the PDGF receptor in the binding of SH2 FYN to the diphosphorylated site, the second phosphotyrosine of Y*VY*IDPT resembles the glutamic acid which is present in the high affinity Y*EEI peptide as predicted (15,19). The PI3-kinase site, on the other hand, has a glutamate residue at the ϩ2 position. However, like the diphosphorylated site, the PI3-kinase site has hydrophobic amino acids at the ϩ1 and ϩ3 positions. Thus, it is not surprising that the SH2 CHK was capable of being competed by the Y*VY*IDPT and the Y*MDMKPG peptides. The SH2 FYN and SH2 SHC both favored the binding to the Tyr 568/570 site over the Tyr 721 site. Studies of SH2 SHC binding to activated PDGF receptor indicated binding to multiple sites (17). However, in these studies, binding of the diphosphorylated Tyr 579/581 peptide was not analyzed in the peptide association experiments. Thus, it is likely in the case of the PDGF receptor that binding of SHC is favored at the Tyr 579/581 site in light of our results. Other studies of FYN binding to the PDGF receptor (15) showed a clear preference for the diphosphorylated Tyr 579/581 site over other sites. The binding affinity of FYN to the PDGF receptor (IC 50 ϭ 2 M) was higher than was the binding of c-KIT to CHK (IC 50 ϭ 50 M for SH2 CHK /c-KIT and 20 M for SH3-SH2 CHK /c-KIT). This difference may be due to a less native structure in the isolated SH2 compared to the intact protein. The lowering of the IC 50 in the case of SH3-SH2 CHK /c-KIT is consistent with this. Thus, had we utilized the native CHK, we might have seen an even lower IC 50 . However, both Tyr 568/570 and Tyr 721 peptides showed a reduction in IC 50 when SH3-SH2 CHK was bound to c-KIT, indicating that the relative bindings of native CHK to the two sites are likely to be similar. To demonstrate that the interaction of SH2 CHK at the two c-KIT sites is direct and not mediated by other proteins in the system, we showed that native CHK from cell extracts as well as a bacterial SH2 CHK -GST fusion protein can bind to phosphorylated peptides linked to beads. This was an unequivocal indication that the CHK/c-KIT interaction was direct and not a result of binding via intermediates in the crude extracts. The possibility of intermediates in the binding was a concern because the sites identified are known to bind other SH2-containing proteins. The Tyr 721 site is known to bind the p85 regulatory subunit of the PI3-kinase (20). In the PDGF receptor, the analogous Tyr 579/581 juxtamembrane site is known to bind SH2 domains of SHC and SRC family kinases (15)(16)(17). The demonstration of direct binding could not completely rule out an additional association via FYN or SHC. Our results of blotting the peptide bead/lysate precipitates with anti-SHC and anti-FYN confirmed that indeed, SHC and FYN were binding to the juxtamembrane phosphorylated Tyr 568/570 sequence of c-KIT. If these or other components were to serve as intermediates or adapters in the binding of SH2 CHK to c-KIT, they themselves would have to be phosphorylated. However, nearly all tyrosine phosphorylation occurs after growth factor stimulation. Thus, the fact that there was no difference in association of peptide beads to CHK from SCF/KL-activated or nonactivated cells indicated that, again, an intermediate was unlikely. Therefore, it was reasonable to expect that we would not be able to co-precipitate SHC and SH2 CHK , or that SRC phosphorylation site peptides would not interfere with SH2 CHK /c-KIT interaction.
It is also notable that blotting the precipitates of extracts and peptide beads with anti-Csk did not show significant binding to the Tyr 568/570 site, but did show some weak binding to the Tyr 721 site. This observation is in spite of the fact that a significant amount of Csk was present in the crude lysates of these cells. Specificity of SH2 Csk as determined by peptide libraries (19) shows a preference for Y*TKM. The similarity to the site Y*MDMKPG particularly at the ϩ1 and ϩ3 positions may explain the slight binding to this peptide. Csk has been shown to bind to tyrosine-phosphorylated paxillin, tensin, and focal adhesion kinase (FAK) (21,22). This emphasizes the role of Csk in the regulation of SRC kinases located in focal adhesions. There have been no reports in the literature of Csk having the ability to bind to growth factor activated receptor tyrosine kinases. Thus, it is more likely that SRC family kinases bound to growth factor receptors such as c-KIT are phosphorylated and down-regulated by CHK and not Csk.
The implication of the c-KIT site directed mutagenesis is that in the association of SH2 CHK to intact c-KIT, only the Tyr 568/570 site is effectively able to bind while the Tyr 721 site is not able to bind this SH2. This is in apparent contradiction to the data involving phosphopeptide inhibition and direct binding of CHK to phosphopeptide beads. A possible explanation for this is that the heptapeptides used in the experiments do not accurately reflect the 3-dimensional structure of the native c-KIT. Thus, the Y*MDMKPG peptide may resemble the Y*VY*IDPT peptide sufficiently such that it could compete for the Tyr 568/570 site. It may also be that this Y*MDMKPG peptide linked to beads, is able to bind to the SH2 CHK even though the native c-KIT is unable to bind to SH2 CHK at the Tyr 721 site. The mutagenesis experiment on the other hand does not rely on peptides to distinguish the binding and more accurately reflects the specificities of the protein-protein interactions. Furthermore, the mutation experiment indicates that there are no other sites on c-KIT that are able to bind the SH2 CHK since binding was completely abolished by the mutation at the Tyr 568 site. Thus, the data taken together indicate that it is only the Tyr 568/570 site that is able to bind CHK through its SH2 domain.
A potential model for the regulation of SRC kinase by CHK involves the SCF/KL stimulated dimerization of c-KIT, one bearing an activated SRC kinase and the other bearing the CHK kinase. This dimerization would bring activated SRC kinase into proximity to CHK, leading to the down-regulation of SRC kinase activity. Further studies will be needed to establish whether SRC kinase bound to c-KIT is the primary target for CHK, and to elucidate the downstream effects resulting from these interactions.