The CRKL adaptor protein transforms fibroblasts and functions in transformation by the BCR-ABL oncogene.

The CRKL adaptor protein was recently identified as a substrate for the BCR-ABL tyrosine kinase in patients with chronic myelogenous leukemia, but its function is unknown. Here we report that CRKL is phosphorylated when overexpressed, activates RAS and JUN kinase signaling pathways, and transforms fibroblasts in a RAS-dependent fashion. We examined the potential role of CRKL in BCR-ABL function by deleting the CRKL binding site in BCR-ABL. This mutant BCR-ABL protein shows a 50% reduction in fibroblast transforming activity. The GRB2 adaptor protein has previously been implicated in this pathway, presumably linking BCR-ABL to RAS. To address the relative roles of CRKL and GRB2 in this system, we compared BCR-ABL mutants with defects in binding to one or both adaptors. Whereas each single mutant showed a 2-3-fold loss in transforming activity, the double mutant showed a 15-fold reduction, suggesting that GRB2 and CRKL both contribute to BCR-ABL transformation. These results demonstrate the oncogenic potential of CRKL and provide functional evidence that CRKL plays a role in fibroblast transformation by BCR-ABL in conjunction with other adaptor proteins.

Receptor tyrosine kinases (RTKs) 1 function in part by activation of Ras. This activation is mediated by adaptor molecules such as GRB2 and SHC, which physically link the RTK to guanine nucleotide exchange factors that activate Ras (1,2). Adaptor molecules have no catalytic activity but contain SH2 and SH3 (Src homology) domains, that mediate protein-protein interactions with phosphotyrosine (3) and proline (4) residues, respectively. Genetic studies have confirmed a critical role for these adaptors in RTK signal transduction (5,6). Loss of function mutations in the Drosophila and Caenorhabditis elegans homologues of GRB2 disrupt signaling by the Sevenless and Let-23 RTKs, respectively, and the mutant phenotypes are specifically rescued by expression of wild-type GRB2 (6 -9).
The CRK genes, originally identified as cellular homologues of the avian retroviral oncogene for v-CRK (10,11), also encode adaptor molecules. CRK-I and CRK-II are alternative RNA splicing products of the c-CRK gene and contain one SH2 domain and either one (CRK-I) or two (CRK-II) SH3 domains. v-CRK contains retroviral GAG sequences fused to the SH2 and SH3 domains of CRK-I. Oncogenic transformation by v-CRK is associated with elevated levels of phosphotyrosine in cells, presumably due to assembly of activated signaling complexes containing a cellular tyrosine kinase (10,12). The normal function of the CRK proteins is unclear. Binding studies have shown that the N-terminal SH3 domain of c-CRK binds to the guanine nucleotide exchange factors SOS and C3G (13), providing a potential link to RAS activation. The same domain also binds the c-ABL tyrosine kinase, and c-CRK is phosphorylated by c-ABL (14,15). Structural studies suggest that c-CRK is regulated by an intramolecular interaction between its SH2 domain and tyrosine 221, which is specifically phosphorylated by c-ABL (16). The SH2 domain of v-CRK binds to specific tyrosine-phosphorylated proteins such as p130 CAS, CBL, and paxillin (17)(18)(19). The functions of CAS and CBL are unknown, but paxillin is localized to focal adhesion plaques (20,21), perhaps linking CRK to adhesion molecules or the cytoskeleton (22). A new CRK family member, CRKL, was recently isolated as a phosphotyrosine substrate in leukemia cells from patients expressing the BCR-ABL fusion protein (23)(24)(25). Similar to CRK-II, CRKL contains one SH2 domain and two SH3 domains and appears to share the same range of binding properties, including interaction with SOS, C3G, c-ABL (13), and BCR-ABL (25).
BCR-ABL encodes a constitutively active cytoplasmic tyrosine kinase created by the Philadelphia chromosome translocation in patients with chronic myelogenous leukemia (CML) (26). BCR-ABL transforms fibroblasts (27) and hematopoietic cells (28) in culture and causes leukemias in mice (29 -32). Transformation is kinase-dependent (33) and results from the activation of multiple signaling pathways (34), including those involving RAS (35)(36)(37), MYC (38), and JUN (39). As with RTKs, it is believed that adaptor proteins mediate the connection between BCR-ABL and RAS. Evidence for this model includes observations that (i) GRB2 and SHC are associated with BCR-ABL in transformed cells (35,40,41), (ii) the GRB2 binding site in BCR-ABL is required for full activity in fibroblast transformation assays (35), and (iii) overexpression of SHC restores transformation to this GRB2 binding mutant (42).
Similar to GRB2 and SHC, CRKL binds to BCR-ABL (23,25). CRKL is a substrate for BCR-ABL in CML cells, and BCR-ABL is the only stimulus known to activate CRKL phosphorylation (24). However, the function of CRKL and its role in BCR-ABL transformation are unknown. In this report we use a well characterized fibroblast model to examine the biological activity of CRKL and its role in BCR-ABL transformation. We show that CRKL becomes phosphorylated when overexpressed, activates Ras-dependent and JNK pathways, and transforms fibroblasts, thereby demonstrating its oncogenic potential. We also show that CRKL plays a critical role as a BCR-ABL substrate because a BCR-ABL mutant that no longer binds CRKL has attenuated transforming activity in fibroblasts. CRKL and GRB2 both function in BCR-ABL transformation in a nonoverlapping manner because a double mutant BCR-ABL incapable of binding either adaptor protein shows a greater deficit in transformation assays than either single mutant. These findings show that CRKL is a functionally relevant tyrosine kinase substrate and suggest that multiple adaptors contribute to transformation in a combinatorial manner.

MATERIALS AND METHODS
Plasmids-A full-length cDNA of the coding region of CRKL was isolated from poly(A) RNA from K562 cells by PCR using primers derived from the published sequence (43). The resulting cDNA was confirmed by sequencing and was subcloned into the pSR␣MSVtkNeo retrovirus vector (44). pSR␣MSVtkNeo p185BCR-ABLwt (44) and pSR␣MSVtkNeo Asn-17 c-Ha-RAS (37) have been described previously. To construct a p185BCR-ABL mutant that lacks the ability to bind CRKL, an in-frame deletion of amino acids 925-975, spanning the previously identified c-CRK binding site in c-ABL (15), was made by PCR. p185BCR-ABLY177F was kindly provided by A. M. Pendergast (Duke University). The double mutant p185BCR-ABLY177F⌬925-975 was made by swapping a BsrGI to HindIII restriction fragment from BCR-ABL⌬925-975 with BCR-ABLY177F. The pB4XCAT plasmid containing the RAS-responsive ETS/AP-1 promoter element (45) (provided by B. Wasylyk, INSERM) was used for transcription activation studies as described previously (35).
Protein Analysis-For labeling studies, cells were incubated in medium containing 1 mCi/ml orthophosphate and lysed in 10 mM sodium phosphate, pH 7.0, 1% Triton X-100, 5 mM EDTA, 150 mM NaCl, 100 g/ml phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 2 mM Na 3 VO 4 , 50 mM NaF. CRKL protein was immunoprecipitated with 10 l of CRKL antiserum (Santa Cruz Biotechnology) and separated by SDS-PAGE. SDS-PAGE percentages were 7% and 12.5% for visualization of ABL and CRKL proteins, respectively. Proteins were transferred to nitrocellulose (Schleicher & Schuell), filters were probed with CRKL antiserum and visualized by ECL (Amersham) or subjected to autoradiography. For non-radioactive studies cells were lysed in the buffer described above and protein concentration was measured by the Bio-Rad DC protein assay. Equal amounts of protein were subjected to immunoprecipitation with CRKL antiserum or 5 l of Abl antiserum (42). Proteins were transferred to nitrocellulose and immunoblotted with antiserum to CRKL or ABL. For JNK assays cells were lysed in 25 mM HEPES, pH 7.5, 10% glycerol, 1% Triton X-100, 150 mM NaCl, 2 mM EDTA, 25 mM sodium glycerophosphate, 1 mM sodium vanadate, 10 g/ml leupeptin, 100 g/ml phenylmethylsulfonyl fluoride. Endogenous JNK was immunoprecipitated with 1 l of JNK antiserum (Santa Cruz Biotechnology) per sample, and immune complex kinase assays were performed with 10 Ci of [␥-32 P]ATP/reaction and using 3 g of GST-JUN as substrate (39) in the following buffer: 20 mM HEPES, pH 7.5, 20 mM MgCl 2 , 25 mM sodium glycerophosphate, 100 M sodium orthovanadate, 2 mM dithiothreitol, 20 M ATP. Chloramphenicol acetyltransferase and ␤-galactosidase activities were measured as described previously (35,46).
Transformation Assays-Retrovirus stocks were prepared by transient transfection of 293T cells using CaCl 2 as described previously (44). Cells were infected with retrovirus stocks and plated into soft agar after 48 h as described previously (38). Colonies were counted approximately 2 weeks after plating.

CRKL Becomes Phosphorylated when Overexpressed-Bio-
chemical studies have shown that CRKL forms a complex with BCR-ABL and is phosphorylated on tyrosine in cells expressing BCR-ABL (23)(24)(25), thereby implicating CRKL as a critical in vivo BCR-ABL substrate. However, no studies have examined the biological consequences of CRKL phosphorylation. We addressed this question by examining the cellular effects of overexpression of CRKL. A cDNA for the coding region of CRKL was isolated using PCR as described under "Materials and Methods" and subcloned into the pSR␣MSVtkNeo retrovirus vector for expression in mammalian cells. Previous studies in CML cells have shown that tyrosine-phosphorylated CRKL migrates more slowly than unphosphorylated CRKL in SDS-PAGE assays and can be distinguished by anti-CRKL immunoblot analysis due to a band shift (24,25). We obtained similar results in 293T cells transfected with a BCR-ABL expression plasmid (Fig. 1A, lane 3) compared to a NEO control (Fig. 1A, lane 1). When cells were transfected with the CRKL expression vector, we consistently observed three bands immunoreactive with antisera specific for CRKL (Fig. 1A, lane 2). These included (i) one band (lowest) which comigrates with the most prominent endogenous CRKL band observed in NEO-transfected cells (Fig. 1A, lane 1), (ii) a second, more prominent band (middle), which comigrates with the upper band in cells transfected with BCR-ABL, and (iii) a third band (top) of even slower mobility. These same three bands were also seen in cells transfected with CRKL and BCR-ABL; however, the intensity of the upper band was greatly enhanced (Fig. 1A, lane 4).
Next we asked if these slower mobility bands represent phosphorylated forms of CRKL by performing orthophosphate labeling experiments. As expected, CRKL immunoprecipitated from cells transfected with BCR-ABL showed a large increase in phosphate content (Fig. 1B, lane 3) when compared to the NEO-transfected control (Fig. 1B, lane 1). Anti-CRKL immunoblot analysis performed using the same filter confirmed that the prominent, phosphate-labeled band seen in cells transfected with BCR-ABL is the slower mobility CRKL band, as previously reported (24,25). In cells overexpressing CRKL, two phosphate-labeled bands were seen (Fig. 1B, lane 2), one of similar mobility to that observed in cells transfected with BCR-ABL and a second slower mobility band. This upper band was labeled even more prominently in cells co-transfected with both CRKL and BCR-ABL (Fig. 1B, lane 4). Phosphotyrosine immunoblot analysis confirmed that the slower migrating bands contained phosphotyrosine (data not shown). Based on these results, we conclude that CRKL becomes phosphorylated when overexpressed, similar to the effects induced by BCR-ABL. Differences in the mobility of phosphorylated forms of CRKL induced by BCR-ABL versus CRKL overexpression may result from differences in the stoichiometry and sites of phosphorylation and will require further study.
CRKL Activates Intracellular Signaling Pathways Involving RAS and JUN Kinase-Because overexpression of CRKL was 293T cells were transfected with the plasmids listed above the figure. A, cells were lysed and proteins separated by SDS-PAGE (12.5%) followed by immunoblot with CRKL antisera. B, after labeling with orthophosphate, lysates were subjected to immunoprecipitation with CRKL antisera. Immunoprecipitates were resolved by SDS-PAGE (12.5%) and exposed to film for autoradiography. The same filter was also analyzed by CRKL immunoblot to identify the position of unphosphorylated CRKL (data not shown). sufficient to activate its phosphorylation, we next asked if CRKL overexpression mimics the effects of BCR-ABL on intracellular signaling pathways. BCR-ABL is known to activate pathways involving RAS (34,36,47) and JNK (39). We tested the ability of CRKL to activate these pathways in Rat-1 fibroblasts, which have been extensively utilized for studies of BCR-ABL signal transduction (27,34). Stable cell lines expressing either CRKL (Rat-1/CRKL) or the G418 (NEO) drug resistance marker (Rat-1/NEO) were generated by retrovirus infection and drug selection for 2 weeks. Overexpression of CRKL was confirmed by immunoblot (Fig. 2B, bottom). Previous studies have shown that the plasmid pB4XCAT, which contains a Ras-responsive enhancer element adjacent to a minimal promoter driving expression of the CAT reporter gene (45), is activated by BCR-ABL in a RAS-dependent manner (35). Therefore, we measured the effect of CRKL on the activity of this RAS-dependent reporter gene. After transfection with pB4XCAT, cells expressing CRKL consistently showed 10-fold higher levels of CAT activity when compared to the Rat-1/NEO control ( Fig. 2A). The magnitude of this response is comparable to that previously reported for BCR-ABL (35). Next we examined the effect of CRKL on JNK. We have previously shown that JNK is activated 4 -5-fold by BCR-ABL (39). Endogenous JNK was immunoprecipitated from Rat-1/NEO and Rat-1/ CRKL cells, and its kinase activity was measured in vitro using GST-JUN as a substrate. JNK was activated in Rat-1/CRKL cells when compared to the Rat-1/NEO control (Fig. 2B, top). Similar activation was seen with transient retroviral infection, followed by JNK kinase assay (data not shown). Thus, overexpression of CRKL is sufficient to activate intracellular signal-ing pathways similar to those activated by BCR-ABL.
CRKL Transforms Fibroblasts in a RAS-dependent Manner-Next we asked if CRKL overexpression leads to cellular transformation using a single-step soft agar colony assay, which measures anchorage-independent growth. Rat-1 fibroblasts were infected with retrovirus expressing CRKL or the NEO marker gene and plated in duplicate into soft agar after 48 h. Immunoblot analysis with CRKL antisera showed overexpression of CRKL with multiple bands indicative of phosphorylation (Fig. 3A, lane 2). In two independent experiments plated in duplicate, cells infected with CRKL retrovirus formed numerous colonies after 2 weeks (Fig. 3, B and C). No colonies were observed in cells infected with the Neo control retrovirus (Fig. 3B). In addition, CRKL infected cells caused acidification of the soft agar media, a frequent finding with transformed cells. Similar results were obtained with the stable R1/CRKL cell line (data not shown).
Because CRKL activates the RAS-dependent pB4XCAT construct, we asked if RAS plays a role in transformation by CRKL. For these experiments Rat-1 fibroblasts were simultaneously infected with retroviruses expressing either CRKL and NEO or CRKL and Asn-17 RAS, a previously described dominant negative RAS mutant (48). Asn-17 RAS cannot be converted to an active GTP-bound state and is believed to function as a dominant negative mutation by titration of guanine nucleotide exchange factors (48). In two independent experiments, Asn-17 RAS completely blocked transformation by CRKL (Fig. 4A). This result is comparable to our previous studies of Asn-17 RAS and BCR-ABL (37). Immunoblot analysis confirmed expression of CRKL and Asn-17 RAS in the appropriate cell populations (Fig. 4B). These findings show that overexpression of CRKL transforms fibroblasts cells by activation of a Ras-dependent pathway.
CRKL Binds to a Proline-rich Region of BCR-ABL-The fact that overexpression of CRKL is oncogenic provides functional evidence that it may be a physiologically relevant substrate. To directly examine the role of CRKL as an adaptor protein in signal transduction by a tyrosine kinase, we designed a strategy to disrupt the signal from BCR-ABL to CRKL. Previous studies of the GRB2 adaptor molecule showed that mutation of the GRB2 binding site in BCR-ABL disrupts complex formation between BCR-ABL and GRB2. This mutant has impaired transforming activity in fibroblasts (35), thereby implicating GRB2 in the BCR-ABL transformation pathway. We adapted this approach to address the role of CRKL in BCR-ABL transformation. First, we had to define the CRKL binding site in BCR-ABL. A prior observation from studies of c-ABL and c-CRK showed that the N-terminal SH3 domain of c-CRK can bind two adjacent P-X-X-P motifs in c-ABL (15). We created a deletion in BCR-ABL that encompassed both of these P-X-X-P motifs (BCR-ABL⌬925-975) (Fig. 5A) and tested the effect of this deletion on binding to CRKL in two cell types commonly used for studies of BCR-ABL function: 293T cells and Rat-1 fibroblasts. Cells were transfected (293T) or infected (Rat-1/ CRKL) with retrovirus constructs expressing NEO, BCR-ABL, or BCR-ABL⌬925-975. Immunoprecipitations with CRKL antisera were performed on lysates from each cell type, then analyzed for co-precipitation of BCR-ABL by immunoblot using ABL antisera. As expected, wild-type BCR-ABL was present in anti-CRKL immunoprecipitates from both cell types (Fig. 5B,  top panel, lanes 2 and 5), indicating these two proteins form a complex. However, the BCR-ABL⌬925-975 mutant was not present in anti-CRKL immunoprecipitates from 293T cells or Rat-1/CRKL cells (Fig. 5B, top panel, lanes 3 and 6), despite high levels of expression of BCR-ABL⌬925-975 (Fig. 5B, bottom panel, lanes 3 and 6). This result defines the CRKL binding site in BCR-ABL and shows that deletion of this site disrupts complex formation between BCR-ABL and CRKL in these two cell types.
Deletion of the CRKL Binding Site in BCR-ABL Impairs Fibroblast Transformation-Next we examined the consequences of lack of CRKL binding on the biological activity of BCR-ABL. Because the CRKL binding site in BCR-ABL is adjacent to the tyrosine kinase domain, we first compared the kinase activity of wild-type BCR-ABL to the BCR-ABL⌬925-975 mutant using autophosphorylation on tyrosine in vivo as an end point. Whole cell lysates from 293T cells transfected with neo, wild-type BCR-ABL, or BCR-ABL⌬925-975 were analyzed by immunoblot for expression of BCR-ABL and autophosphorylation activity. Wild-type BCR-ABL and BCR-ABL⌬925-975 were expressed at similar levels and were phosphorylated at comparable levels on tyrosine (Fig. 6A). This result suggests that deleting the CRKL binding site does not alter the kinase activity of BCR-ABL. Next we measured the effect of this deletion on the transforming activity of BCR-ABL in Rat-1 fibroblasts. After infection with retrovirus stocks expressing NEO, wild-type BCR-ABL, or BCR-ABL⌬925-975, Rat-1 cells were plated into soft agar to assess anchorageindependent growth as a measure of transformation. In two independent experiments plated in duplicate, cells infected with BCR-ABL⌬925-975 retrovirus consistently showed a 50% reduction in colony formation when compared to cells infected with wild-type BCR-ABL retrovirus (Fig. 6B). Immunoblot analysis of whole cell lysates using ABL antisera confirmed that equivalent levels of expression of the wild-type and BCR-ABL⌬925-975 proteins were seen following retrovirus infec-

FIG. 4. Dominant negative RAS blocks CRKL transformation.
A, Rat-1 fibroblasts infected with the retrovirus stocks listed on the left of the figure were plated into soft agar. Colonies were counted after 2 weeks. Error bars represent the average of two experiments (Ϯ standard deviation). B, a portion of the infected cells was lysed before plating and protein expression levels were analyzed by SDS-PAGE (12.5%) and immunoblot with antisera to CRKL and RAS.

FIG. 5. Deletion of a proline-rich region in BCR-ABL abrogates binding to CRKL.
A, the schematic shows wild-type BCR-ABL protein and the BCR-ABL⌬925-975 mutant protein, which contains a 50-amino acid deletion spanning two P-X-X-P motifs. B, 293T cells were transfected with plasmids expressing NEO, wild-type BCR-ABL, or BCR-ABL⌬925-975. Rat-1/CRKL cells were infected with retrovirus stocks expressing the same proteins. Cell lysates were subjected to immunoprecipitation with CRKL antisera and immunoprecipitates were analyzed by SDS-PAGE (7%) followed by ABL immunoblot (top half, lanes [1][2][3][4][5][6]. The different mobilities of wild-type BCR-ABL and BCR-ABL⌬925-975 are indicated by solid black arrows and open arrows, respectively. Expression of the appropriate BCR-ABL protein in each cell type was confirmed by anti-ABL immunoblot analysis of whole cell lysates (bottom half). tion (Fig. 6C). These results show that CRKL binding is necessary for full transforming activity of BCR-ABL in fibroblasts.
GRB2 and CRKL Independently Contribute to Fibroblast Transformation by BCR-ABL-The ability of BCR-ABL to bind to both GRB2 and CRKL provides a possible explanation for why the BCR-ABL⌬925-975 mutant retains partial transforming activity. In the absence of binding to CRKL, BCR-ABL may utilize other adaptor proteins to function as an oncogene. To address this possibility, we created a BCR-ABL double mutant (BCR-ABLY177F⌬925-975) with a point mutation in the GRB2 binding site (Tyr-177 3 Phe) and a deletion of the CRKL binding site (⌬925-975). The relative transforming activities of single and double adaptor binding mutants of BCR-ABL were compared to wild-type BCR-ABL using the Rat-1 fibroblast soft agar assay. Colony number, colony size, and acidification of the media were all used as measures of transformation in three independent experiments (Fig. 7). Consistent with previous reports (35), a single mutation in the Grb2 binding site (BCR-ABLY177F) impaired transforming activity. However, it is important to note that this mutant does retain 30% activity under the conditions of this assay, suggesting a potential role for other adaptor proteins. A single mutation of the CRKL binding mutant (BCR-ABL⌬925-975) showed a 50% decrease in colony formation, consistent with the experiments reported in Fig. 6. When combined within the same protein, the effect of each single mutation was enhanced 5-10-fold. The double mutant incapable of binding either GRB2 or CRKL (BCR-ABLY177F-⌬925-975) showed a 15-fold reduction in transforming activity compared to wild-type BCR-ABL. This loss of function was not due to a change in kinase activity because the BCR-ABLY177F⌬925-975 double mutant showed levels of autophosphorylation equivalent to the wild-type protein as measured by phosphotyrosine immunoblot (data not shown). These results show that fibroblast transformation by BCR-ABL requires interaction with both GRB2 and CRKL, suggesting that multiple adaptors mediate the biological activity of this tyrosine kinase. DISCUSSION One approach toward defining mechanisms of transformation by tyrosine kinases is to identify relevant cellular substrates. CRKL was identified (49) and isolated (23,24) from CML cell lines and patient samples in a search for substrates for the BCR-ABL tyrosine kinase. CRKL is of particular interest because it is an adaptor protein that can bind a number of molecules which might affect signal transduction pathways. These include GDP/GTP exchange factors for RAS such as SOS and C3G and the tyrosine-phosphorylated proteins paxillin, CBL, and CAS (13,25). Despite these compelling biochemical associations, no functional studies of the biological activity of CRKL have been reported. In this work we have examined the biological effects of CRKL by overexpression in fibroblasts. We find that overexpression of CRKL activates cellular signaling pathways and leads to fibroblast transformation, thereby implicating CRKL as an oncogene. We have also examined the role of CRKL as an adaptor protein in RTK signaling using fibroblast transformation by BCR-ABL as a model system. Deletion of the CRKL binding site in BCR-ABL disrupts complex formation between the two proteins and causes a partial reduction in transforming activity. Because BCR-ABL also binds the GRB2 adaptor protein, we addressed the relative roles of CRKL and GRB2 in BCR-ABL function by preparing mutants that fail to bind to either one or both proteins. The results suggest that both GRB2 and CRKL contribute to transformation. Taken together, these findings provide evidence that CRKL is a physiologically relevant BCR-ABL substrate and functions in linking BCR-ABL to transformation pathways.
To date, BCR-ABL is the only stimulus known to activate human CRKL phosphorylation, despite an exhaustive look at cytokines and other mitogenic stimuli (24). Recent studies of murine CRKL have shown ubiquitous expression of the protein with highest levels in hematopoietic cells. Tyrosine-phosphorylated CRKL was found only during early embryogenesis and in lung tissue (50). For these reasons, we were surprised by our initial studies showing that overexpression of CRKL in fibroblasts was sufficient to lead to its phosphorylation. This result implies that a balance between cellular tyrosine kinases and phosphatases maintains CRKL in a mostly unphosphorylated state. Overexpression of CRKL might perturb this balance in favor of its tyrosine phosphorylation, perhaps by overcoming a regulatory phosphatase. Alternatively, high levels of CRKL might allow unhindered access to a cellular tyrosine kinase. The consequence is activation of cellular signal transduction pathways and transformation. Although similar findings have been reported for other adaptor proteins such as v-CRK (10, 11), CRK-I (51), NCK (52), and SHC (53), the closest homologue of CRKL, CRK-II, does not transform fibroblasts (51). Our FIG. 6. BCR-ABL⌬925-975 retains kinase activity but transformation is impaired. A, 293T cells were transfected with retrovirus constructs expressing NEO, wild-type BCR-ABL, or BCR-ABL⌬925-975. After 48 h whole cell lysates were analyzed by SDS-PAGE (7%) for BCR-ABL protein expression (upper panel) and phosphotyrosine content (lower panel) by immunoblot using antisera directed against ABL or phosphotyrosine. B, Rat-1 fibroblasts were infected with retrovirus stocks expressing NEO, wild-type BCR-ABL, or BCR-ABL⌬925-975. Cells were plated into soft agar and colonies counted after 2 weeks. Results show the average of two independent experiments (Ϯ standard error) and are expressed relative to wild-type BCR-ABL. C, a portion of the cells was lysed before plating and analyzed by SDS-PAGE (7%) and anti-ABL immunoblot to show expression levels of BCR-ABL and BCR-ABL⌬925-975. results suggest that the C-terminal SH3 domain of CRKL, which is postulated to negatively affect transforming activity in CRK-II (51), may be regulated differently in the context of CRKL. The mechanism of transformation by CRK proteins is unknown. Recent studies of v-CRK implicate the Src kinases because v-CRK can interfere with the normal regulation of Src by CSK (54,55). RAS pathway activation also appears to be of functional importance in mediating the effects of v-CRK in PC12 cells (56 -58), which is in agreement with our finding of RAS-dependent transformation by CRKL.
What role does CRKL play in the transforming activity of BCR-ABL? In addition to interacting with BCR-ABL, CRKL binds two guanine nucleotide exchange factors, SOS and C3G, which activate RAS (13). Our finding that CRKL transformation is RAS-dependent suggests that one function of CRKL might be to connect BCR-ABL to the RAS pathway. Consistent with this hypothesis, the CRKL binding mutant of BCR-ABL shows reduced activation of the RAS-responsive pB4XCAT reporter, 2 similar to results reported for the GRB2 binding mutant (35). Two other adaptor molecules, GRB2 (35) and SHC (42), have also been implicated in BCR-ABL transformation, presumably by activating RAS through the formation of signaling complexes. GRB2 links BCR-ABL to SOS by binding directly to both proteins through its SH2 and SH3 domains, respectively (35,40). SHC may function as an intermediate between BCR-ABL and GRB2, since both proteins are found in SHC immunoprecipitates (41), but the biochemical details of the SHC/BCR-ABL interaction are not yet clear. Because CRKL also binds BCR-ABL and SOS, it is possible that it functions analogous to GRB2. This interpretation would argue for multiple pathways for RAS activation by BCR-ABL, which may be redundant or may have distinct roles. The fact that BCR-ABL mutants defective in binding to either GRB2 or CRKL each show some loss of activity favors distinct roles for each adaptor. Furthermore, loss of binding to both GRB2 and CRKL has a greater effect on transformation than loss of binding to either one alone, suggesting that each adaptor contributes independently to transformation. This conclusion is also supported by studies using dominant negative adaptor proteins, which show that dominant negative GRB2 and dominant negative c-CRK each have distinct inhibitory effects on ERK activation by an oncogenic c-ABL mutant (59).
The concept that GRB2 and CRKL may have non-overlapping functions is not surprising if one considers certain differences between GRB2 and CRKL. GRB2 is postulated to form a ternary complex containing both BCR-ABL and SOS, whereas CRKL binds both BCR-ABL and SOS through the same Nterminal SH3 domain (22). Since a single SH3 domain would not be expected to bind two proteins simultaneously, it is unlikely that CRKL functions simply as a bridge between BCR-ABL and SOS. Alternatively, one consequence of direct binding between BCR-ABL and CRKL might be the efficient phosphorylation of CRKL by BCR-ABL, as has recently been reported for c-ABL and c-CRK (60). Once phosphorylated, CRKL might bind a novel set of proteins that play a role in signal transduction by BCR-ABL. In addition to RAS pathway activation through SOS and C3G, it is likely that phosphorylated CRKL affects other signaling pathways through interaction with phosphotyrosine containing proteins through its SH2 domain. CBL (18,61) and paxillin (22) are two such CRKL-binding proteins. Paxillin is of particular interest, since it forms a complex with CRKL and BCR-ABL and becomes phosphorylated in CML cells (22,62). Paxillin is also a substrate for focal adhesion kinase and is localized to focal adhesion plaques (21,63,64). It is likely that interaction with these proteins may contribute to the transformation phenotype. These observations may also provide insight into the observation that CML cells have defects in cellular adhesion (65).