INTRODUCTION

Steel factor (SF)¹ supports growth and survival of immature hematopoietic cells of multiple lineages. The cytokine is synthesized as a transmembrane protein, and a soluble form is generated by a proteolytic cleavage. In vivo, the membrane form is required for normal hematopoiesis, but in vitro, both the membrane bound as well as the soluble form are biologically active. One interesting feature of SF is its ability to act synergistically with various other hematopoietic growth factors such as interleukin-3 or granulocyte macrophage-colony stimulating factor (1-5). The receptor for SF, the proto-oncoprotein c-Kit, belongs to a family of growth factor receptors with intrinsic tyrosine kinase activity, which also includes receptors such as c-Ret, the receptor for glia cell-derived neurotrophic factor, or the platelet-derived growth factor receptor (6-8). Mutations in the murine locus for SF, steel, and its receptor c-Kit, white spotting, have suggested pleiotropic functions for the SF/c-Kit pathway and include defects in melanogenesis, gametogenesis, or hematopoiesis (9, 10). The severity of the phenotype in mice depends on the mutation, with the most severe effects associated with absent kinase activity. Mice heterozygotic for Kit receptor mutations may also have a phenotype, possibly related to the fact that certain mutant receptors behave as dominant negatives (11, 12). Most humans heterozygous for Kit mutations do not have a significant hematopoietic disorder but do have mild defects in hair or skin pigmentation (13, 14).

SF and c-Kit have also been linked to several hematologic and nonhematologic neoplastic disorders. It has been suggested that one mechanism could include autocrine production of SF, leading to a clonal expansion of c-Kit expressing cells. For example, coexpression of SF and c-Kit has been reported in breast tumor cells, small cell lung cancer cells, and malignant glioma cell lines (15-17). Also, it has been shown that the tyrosine phosphorylation pattern of cellular proteins in chronic myelogenous leukemia resembles the pattern observed after SF stimulation in normal cells (18).

With the diverse effects of SF on cells of different lineages, there has been considerable interest in identifying the critical signal transduction pathways activated by c-Kit. The ligand is believed to induce dimerization or oligomerization of the receptor, activating the tyrosine kinase activity and inducing rapid but transient tyrosine phosphorylation of several cellular proteins, including c-Kit itself. Some substrates of c-Kit have been described to associate with c-Kit after activation, including phospholipase C-, p85^PI3K, and SHP2 (19-21). These proteins are believed to link c-Kit directly to various enzymatic pathways. However, as is the case for most receptors, the intermediate signaling events are not well understood.

In this study we have investigated the specific role of CRKL, an adapter protein containing one SH2 domain and two SH3 domains, in c-Kit signaling. CRKL is widely expressed and belongs to the family of CRK adapter proteins that includes v-Crk and c-CRK (22). The CRKL SH2 domain can bind to the proto-oncoprotein p120^CBL, p130^CAS, and the focal adhesion protein paxillin (23-27). The CRKL SH3 domain, like the c-CRK SH3 domain, binds in vitro and associates in vivo with c-ABL, SOS, and C3G (24, 28-32). CRKL is therefore likely to be involved in the regulation of the enzymatic pathways associated with these proteins.

Using a human megakaryoblastic cell line, MO7e, SF stimulation was found to result in the rapid tyrosine phosphorylation of CRKL and coprecipitation of CRKL with c-Kit. However, although this interaction was inducible, it was mediated through the CRKL SH3 domain. This suggested the possibility that CRKL was not binding to c-Kit directly, but associated with c-Kit as part of a larger complex of proteins. We found that the coprecipitation of CRKL with c-Kit is, at least in part, likely to be mediated through binding of p85^PI3K to c-Kit as well as to CRKL. Other proteins are likely to be involved in this complex, with the most prominent identified as the proto-oncoprotein p120^CBL. These data demonstrate that CRKL is involved in the formation of a complex of signaling proteins that bind to c-Kit through PI3K and further suggest that CRKL, and its SH2-binding partners such as p120^CBL, may function downstream of PI3K.

MATERIALS AND METHODS

The human promegakaryoblastic cell line MO7e (obtained from Dr. Steve Clark, Genetics Institute, Cambridge, MA) was maintained in Dulbecco's modified Eagle's medium (Mediatech, Washington, DC), 10 ng/ml granulocyte-macrophage colony-stimulating factor (Genetics Institute), and 20% (v/v) fetal calf serum (PAA Laboratories Inc., Newport Beach, CA) at 37 °C with 10% CO₂. For stimulation studies, MO7e cells were deprived of growth factors for 18 h at 37 °C in serum-free medium containing 1% (w/v) bovine serum albumin (Sigma).

Starved MO7e cells were left untreated or stimulated with recombinant human SF (Amgen, Thousand Oaks, CA) in Dulbecco's modified Eagle's medium at a concentration of 5 × 10⁶ cells/ml. Cells were washed in cold Dulbecco's phosphate-buffered saline, and cell lysates were prepared as described (33).

Immunoprecipitation and immunoblotting using a chemiluminescence technique was performed as described (33). Immunochemical detection of tyrosine phosphorylated proteins in Western blots utilized monoclonal antibody 4G10 (kindly provided by Dr. B. Druker, Oregon Health Science University, Portland, OR). Polyclonal rabbit antisera were obtained against c-Kit (Santa Cruz Biotechnology, Santa Cruz, CA), p120^CBL (Santa Cruz Biotechnology), CRKL (Santa Cruz Biotechnology), and p85^PI3K (Upstate Biotechnology Inc., Lake Placid, NY). A murine monoclonal antibody was generated against the N-terminal SH3 domain of CRKL and used for immunoblotting in this study. The pGEX vectors containing the N-terminal or C-terminal SH2 domain and the SH3 domain of p85^PI3K were kindly provided by Dr. L. Cantley (Beth Israel Hospital, Harvard Medical School, Boston, MA). pGEX vectors containing the SH2 and the SH3-SH3 domains of CRKL were kindly provided by Dr. J. Groffen (Children's Hospital, UCLA, Los Angeles, CA). The GST fusion proteins were expressed in Escherichia coli (DH-5) by isopropyl-1-thio--D-galactopyranoside induction and isolated from sonicated bacterial lysates using glutathione-Sepharose beads (Pharmacia Biotech Inc.) according to the manufacturer's directions.

Using previously established techniques (34), far Western blotting was performed as described (24). In brief, immunoprecipitated proteins were transferred after SDS-polyacrylamide gel electrophoresis to Immobilon-P (polyvinylidene difluoride) membrane (Millipore Corp., Bedford, MA) and blocked with 5% nonfat dry milk in 0.1% Tween 20 in phosphate-buffered saline, pH 7.4). The specific direct in vitro binding was evaluated by probing the membrane with GST fusion proteins and visualized with a combination of anti-GST monoclonal antibody (Santa Cruz Biotechnology) and horseradish peroxidase-coupled anti-mouse IgG antibody by chemiluminescence.

The activity of PI3K was measured in anti-c-Kit, CRKL, p120^CBL, and preimmune control immunoprecipitates of unstimulated and SF-stimulated MO7e cell lysates. The kinase assay was performed as described (35, 36).

RESULTS

Stimulation of c-Kit induces tyrosine phosphorylation of cellular proteins and the receptor itself. For initial experiments we stimulated MO7e cells with SF for different time points between 0 and 60 min. Fig. 1 shows that SF induces rapid and transient tyrosine phosphorylation of cellular proteins with apparent molecular masses of 39, 55, 70, 80, 120, and 155 kDa. The tyrosine phosphorylation of these proteins followed different kinetics. Whereas proteins with apparent molecular masses of 55, 70, 80, and 120 kDa had maximum tyrosine phosphorylation at 1 min, the tyrosine phosphorylation of proteins with apparent molecular masses of 39 and 155 kDa peaked at 10 min and decreased thereafter.

To investigate the potential role of the 39-kDa adapter protein CRKL in SF signaling, we looked for tyrosine phosphorylated proteins that coprecipitated with CRKL. SF induced tyrosine phosphorylation of CRKL itself and association of CRKL with prominent phosphotyrosine proteins with apparent molecular masses of 120 and 155 kDa (Fig. 2A). In addition, a single 39-kDa tyrosine phosphorylated band was also detected in CRKL immunoprecipitates of SF-stimulated but not of unstimulated denatured lysates, suggesting that this band is indeed CRKL (data not shown). The tyrosine phosphorylation of CRKL and association with the 120-kDa phosphotyrosine protein was maximal within 1 min, and the association with the 155-kDa phosphotyrosine protein was maximal within 10 min of SF stimulation.

The 155-kDa tyrosine phosphorylated protein that coimmunoprecipitated with CRKL comigrated with c-Kit. This membrane was therefore reblotted, and the 155-kDa protein was identified by immunoblotting as c-Kit (Fig. 2A). The lowest panel in Fig. 2A demonstrates that equal amounts of CRKL were loaded. Similarly, we found that a 39-kDa tyrosine phosphorylated protein was detected in c-Kit immunoprecipitates. This 39-kDa protein was identified as CRKL by reblotting with anti-CRKL antibodies (Fig. 2B). We also noted inducible coprecipitation of tyrosine phosphorylated proteins in c-Kit immunoprecipitates with apparent molecular masses of 55, 70, and 120 kDa. Preimmune rabbit serum did not precipitate tyrosine phosphorylated proteins. The upper panel on the right side of Fig. 2B demonstrates that equal amounts of c-Kit were loaded. In BCR/ABL-transformed cells, we and others have previously demonstrated that CRKL can bind to tyrosine phosphorylated p120^CBL. We therefore also asked if CRKL coprecipitates with p120^CBL after SF stimulation. Fig. 2C shows immunoprecipitates of p120^CBL after 0-60 min SF stimulation, demonstrating rapid association with CRKL within 1 min of SF stimulation, which decreased thereafter. p120^CBL coimmunoprecipitated also with a major tyrosine phosphorylated protein of an apparent molecular mass of 155 kDa. Similarly, in both the CRKL and c-Kit immunoprecipitations, the 120-kDa tyrosine phosphorylated protein was identified as p120^CBL (data not shown). The middle panel in Fig. 2C demonstrates that equal amounts of p120^CBL were loaded. These data suggest that CRKL forms an inducible complex with c-Kit and p120^CBL after SF stimulation.

We have previously shown that the BCR/ABL oncoprotein induces formation of a complex containing p210^BCR/ABL, p120^CBL, CRKL, and p85^PI3K (24). Because p85^PI3K is known to bind to c-Kit after SF activation (37), we looked for association of PI3K with CRKL or p120^CBL after SF stimulation of MO7e cells. In lysates of SF-stimulated MO7e cells, p85^PI3K was associated with tyrosine phosphorylated proteins with apparent molecular masses of 155, 120, 70, 55, and 39 kDa (Fig. 3, left panel). The 155-kDa protein was identified as c-Kit, the 120-kDa protein was identified as p120^CBL, and the 39-kDa protein was identified as CRKL by specific immunoblotting (Fig. 3, right panels). Whereas the association of p85^PI3K with c-Kit and p120^CBL was dependent on SF stimulation, a small amount of CRKL was already preassociated with p85^PI3K. The association of p85^PI3K with CRKL increased considerably upon SF stimulation of MO7e cells. These data suggest the formation of a multimeric signaling complex after SF stimulation that includes CRKL, c-Kit, p120^CBL, and p85^PI3K.

The previous results suggested that c-Kit, CRKL, and p120^CBL form a complex with p85^PI3K, suggesting that the p110^PI3K enzymatic subunit is also present. To determine if enzymatically active PI3K is present in this complex, we immunoprecipitated c-Kit, CRKL, and p120^CBL from lysates of unstimulated and SF-stimulated MO7e cells. Fig. 4 demonstrates that significant PI3K activity was detected in c-Kit, CRKL, and p120^CBL immunoprecipitates of SF-stimulated MO7e cells and in CRKL immunoprecipitates of unstimulated cells. However, very little PI3K coimmunoprecipitated with anti-p120^CBL compared with anti-c-Kit and CRKL immunoprecipitates; preimmune serum did not immunoprecipitate PI3K activity. These results also demonstrate that CRKL is preassociated with active PI3K, and after SF stimulation the association of active PI3K with CRKL and c-Kit is markedly increased.

The above results suggest the potential induction of one or more multimeric c-Kit receptor complexes containing CRKL, p85^PI3K, and p120^CBL. The association of c-Kit with CRKL and also with p85^PI3K and p120^CBL appears to require tyrosine phosphorylation of some of these proteins. Because CRKL has one SH2 and two adjacent SH3 domains, we sought to determine the mechanism of binding of CRKL to the proteins in this complex using GST fusion proteins containing various segments of each protein. Surprisingly, the SH3 domain of CRKL precipitated c-Kit from lysates of stimulated but not unstimulated cells. Using denatured lysates of unstimulated and SF-stimulated MO7e cells, we did not observe precipitation of the CRKL SH2 or CRKL SH3 domains with c-Kit (data not shown). In contrast, the SH2 domain of CRKL precipitated p120^CBL from cell lysates of stimulated but not unstimulated MO7e cells. However, both the SH2 and the SH3 domains precipitated p120^CBL from SF-stimulated and unstimulated lysate with increased specific binding of CRKL fusion protein to p120^CBL following stimulation (Fig. 5A). Because p85^PI3K contains two proline-rich motifs, we tested whether the CRKL-SH3 interacts with p85^PI3K. Fig. 5B demonstrates that p85^PI3K binds constitutively to the CRKL SH3 domain but does not associate with the CRKL SH2 domain or the ABL-SH3 domain. We next tested whether GST fusion proteins containing the N-terminal SH2 domain, the C-terminal SH2 domain, or the SH3 domain of p85^PI3K could precipitate c-Kit or p120^CBL. Fig. 5C demonstrates that the C-terminal and the N-terminal SH2 domain of p85^PI3K can associate with the c-Kit receptor. This is consistent with earlier findings demonstrating binding of p85^PI3K to c-Kit (37). This membrane was reblotted with antibodies to p120^CBL demonstrating in vitro association of p120^CBL with the SH3 domain and both SH2 domains of PI3K.

The in vitro GST fusion protein precipitations with c-Kit, p85^PI3K, and p120^CBL do not indicate whether binding of the SH2 or SH3 domains is direct or indirect. We therefore used a far Western technique to identify direct in vitro interactions. Cellular lysates from unstimulated and SF-stimulated MO7e cells were used for immunoprecipitations with c-Kit, CRKL, and anti-p120^CBL antibodies. Fig. 5D shows that GST protein alone does not bind to proteins in immunoprecipitations. Direct binding of a single 120-kDa protein band in p120^CBL immunoprecipitations using the GST-CRKL-SH2 protein as a probe was found after SF stimulation. We also observed in vitro binding of the CRKL SH3 domain to p85^PI3K, this interaction was not changed upon SF stimulation. However, we also observed inducible binding of the CRKL SH3 and the CRKL SH2 domains to a 155-kDa protein in anti-c-Kit and p85^PI3K immunoprecipitations but not in anti-p120^CBL immunoprecipitations. The identity of this protein remains unknown but we do not believe that it is c-Kit because we could not precipitate c-Kit with the CRKL SH2 domain. The p85^PI3K SH3 domain was found to bind directly to p120^CBL and the C-terminal SH2 domain to c-Kit. We did not observe significant binding of the N-terminal SH2 domain of p85^PI3K to c-Kit, CRKL, or p120^CBL.

These results suggest that in MO7e cells, CRKL is linked through its SH3 domain to the PI3K pathway, whereas the SH2 domain of CRKL binds to p120^CBL after SF stimulation. In addition, the results also suggest direct binding of the C-terminal p85^PI3K SH2 domain to c-Kit and binding of the p85^PI3K SH3 domain to p120^CBL.

DISCUSSION

Although there are a number of signaling molecules known to be tyrosine phosphorylated by c-Kit, most downstream signaling pathways are not well understood. After stimulation of c-Kit with SF, previous studies have demonstrated rapid tyrosine phosphorylation of c-Kit itself and cellular proteins including SHP2, phospholipase C-, SHC, and p120^CBL (19-21, 38, 39). These molecules in turn contribute signals to various pathways that influence growth, viability, adhesion, migration, or differentiation. PI3K has been the focus of several studies related to c-Kit signaling. For example, mutation of tyrosine 709 to phenylalanine in the Tyr-Xaa-Xaa-Met motif of the murine c-Kit receptor has been shown to reduce binding of the p85 subunit of PI3K (37). Further, this mutation also caused defects in SF-mediated adhesion and early gene expression, presumably by interfering with activation of PI3K (40). However, despite the apparent importance of PI3K signaling in this and other receptors, downstream signaling events have been difficult to identify.

In the studies reported here, we found that the adapter protein, CRKL, was tyrosine phosphorylated after c-Kit activation and coprecipitated with c-Kit. However, there was no evidence that CRKL bound to c-Kit directly, and the data presented here suggest rather that CRKL binds directly to p85^PI3K and indirectly to c-Kit through p85^PI3K. The interaction of CRKL with p85^PI3K was found to utilize the CRKL SH3 domains and could be further increased in response to factor stimulation. Also, we found that after SF stimulation, p120^CBL is tyrosine phosphorylated and coprecipitates with CRKL, suggesting the formation of a signaling complex that contains c-Kit, PI3K, CRKL, and p120^CBL. Overall, the data suggest the possibility that CRKL and/or p120^CBL play a role in sending or modulating signals from c-Kit that require PI3K.

One of the most intriguing findings here is the specific, inducible association of CRKL with p85^PI3K through the CRKL SH3 domains. Previous studies suggested that CRK is constitutively associated with c-ABL, C3G, or SOS through the CRK SH3 domains. These proteins were first described to bind to the CRKII SH3 domain, but we and others have shown that they also bind to the CRKL SH3 domain (24, 28-32). SOS has known guanine exchange factor activity for p21Ras, whereas C3G appears to have specific guanine exchange activity for p21Rap1 (41). C3G has unique binding affinities to the CRK family proteins, because it preferentially binds to the N-terminal SH3 domain (29). The exact function of the tyrosine kinase c-ABL is unknown, although c-ABL has been shown to be involved in transcriptional activation and possibly is activated in response to certain types of DNA damage (42, 43). Recently a consensus sequence for binding to the CRK SH3 domain, Pro-Xaa-Leu-Pro-Xaa-Lys, has been described (44, 45). Consistent with our finding of CRKL SH3 binding to p85^PI3K, a proline-rich motif with this consensus sequence, Pro-Ala-Leu-Pro-Pro-Lys (amino acids 305-310, human sequence), is present in p85^PI3K (46). The SH3 domain interaction of CRKL with p85^PI3K may occur at this site, although this has not yet been directly tested. Inducible association of an adapter protein through SH3 domain interactions has been previously described for the binding of GRB2 to SOS in T lymphocytes, although most SH3 domain-mediated interactions are constitutive (47). One explanation for this phenomenon would be the presence of other proteins binding to CRKL, p85^PI3K, and c-Kit that are important for overall stability of the complex and that are brought into the complex in response to receptor activation. We believe that p120^CBL may be important in this regard, and this will be discussed in more detail below. Overall, our data strongly suggest that PI3K should be added to the list of signaling proteins known to interact with the CRKL SH3 domains but in the specific situation of c-Kit activation.

The fact that the CRKL SH3 domains can bind to several different, unrelated signaling proteins suggests that CRKL may play a role in several different signaling pathways. Thus, it is possible that the biological functions of CRKL may vary widely in different cells. Alternatively, it is possible that different stimuli may activate different signaling pathways involving CRKL in the same cell. CRKL has been shown to be tyrosine phosphorylated in cells transformed by onco-proteins including BCR/ABL, v-Abl, v-Src, or in normal signaling after EGF receptor stimulation (34, 48, 49). Binding of CRKL to p120^CBL appears to be independent from CRKL tyrosine phosphorylation, because in other systems, including T cell signaling, CRKL is not tyrosine phosphorylated, although it binds to p120^CBL (50). Surprisingly, we did not find significant association of CRK with c-Kit in SF signaling in MO7e cells, despite abundant expression of CRK in this cell line (data not shown) and the reported similarity of target proteins selected by CRKL and CRK during in vitro binding studies. It may be worthwhile to look specifically for situations in which CRK and not CRKL is preferentially selected for tyrosine phosphorylation and activation in various cell lineages.

Thus, the data presented here support the notion that CRKL is involved in a signaling pathway that also involves PI3K. Our data suggest that a substantial proportion of the PI3K activity associated with an activated c-Kit receptor is also associated with CRKL, because approximately the same amount of PI3K enzymatic activity is found in anti-CRKL as in anti-c-Kit immunoprecipitates. The biological effects of activated PI3K may vary widely from cell to cell, ranging from regulation of apoptosis, viability, or early gene expression to regulation of adhesion. CRKL may be involved in signaling to any or all of these different biological events. Specifically with regard to c-Kit, mutant c-Kit receptors that fail to bind PI3K, fail to induce c-fos or junB expression, and lack the ability to induce binding of cells to fibronectin and CRKL could be involved in one or all of these events (40).

As noted above, a protein known to bind to the CRKL SH2 domain in certain transformed cells, p120^CBL, was found to coprecipitate with CRKL after c-Kit stimulation. In vitro binding studies suggested that the SH2 domains of both CRKL and p85^PI3K can bind to p120^CBL, and the SH3 domain of PI3K also interacted directly with p120^CBL. Of course, other proteins that can interact with the CRKL SH2 domain may also be brought into any such complex, and in addition to p120^CBL, the known possibilities include p130^CAS, and paxillin (25-27). However, our data in MO7e cells suggest that p120^CBL is the most abundant tyrosine phosphoprotein coprecipitating with CRKL after SF stimulation in these cells. The proto-oncoprotein p120^CBL (for Casitas B-lineage lymphoma) is the cellular homolog of v-Cbl, the oncoprotein in the CAS NS-1 retrovirus (51, 52) that induces pre-B cell lymphomas and myelogenous leukemias in mice (53). p120^CBL is also known to be a substrate of tyrosine kinases in response to T cell (54) and B cell (55) activation, FC- receptor cross-linking (56, 57), and growth factors (33, 58-60). In mammalian cells, the function of p120^CBL is not known; however, it may act downstream of c-Src signaling for bone resorption by osteoclasts (61). Also, the p120^CBL homolog Sli-1 in Caenorhabditis elegans is a negative regulator of the epidermal growth factor receptor tyrosine kinase homologue Let-23 (62). The formation of a complex containing CRKL, p85^PI3K, and p120^CBL has been previously demonstrated by us in BCR/ABL-transformed cells (24). Although the function of the CRK family proteins is not known, a model by Feller et al. shows a hypothetical role for the tyrosine phosphorylation of CRK (31). In this model the N-terminal SH3 domain interacts with the proline-rich domain of c-ABL. Subsequent tyrosine phosphorylation of CRK leads to dissociation of c-ABL from CRK and binding of the CRK SH2 domain to the tyrosine phosphorylated site in CRK. However, the data presented here and our previous studies in BCR/ABL-transformed cells do not suggest that CRKL, unlike CRK, undergoes intramolecular binding to its own SH2 domain.

Overall, one model consistent with our data would be that a tyrosine phosphorylated c-Kit receptor first attracts the p85^PI3K subunit through its SH2 domain to Tyr⁷⁰⁹. Some CRKL and 120^CBL may already be bound to PI3K and may become targets for the c-Kit tyrosine kinase or another kinase. The tyrosine phosphorylation of p120^CBL is likely to provide binding sites for the SH2 domains of CRKL and p85^PI3K, further potentially stabilizing the complex. Thereafter the complex may leave the receptor intact or disassembled, and it is not clear at this time if p120^CBL and CRKL are regulators of PI3K enzymatic activity or downstream effectors that require PI3K activity for an as yet unknown signaling function. In any case, it is likely that further elucidation of the functions of this complex will be helpful in understanding the signaling of c-Kit and PI3K in particular.