Identification of Grb4/Nckβ, a Src Homology 2 and 3 Domain-containing Adapter Protein Having Similar Binding and Biological Properties to Nck*

Adapter proteins made up of Src homology (SH) domains mediate multiple cellular signaling events initiated by receptor protein tyrosine kinases. Here we report that Grb4 is an adapter protein closely related to but distinct from Nck that is made up of three SH3 domains and one SH2 domain. Northern analysis indicated that both genes are expressed in multiple tissues. Both Nck and Grb4 proteins could associate with receptor tyrosine kinases and the SH3-binding proteins PAK, Sos1, and PRK2, and they synergized with v-Abl and Sos to induce gene expression via the transcription factor Elk-1. Although neither protein was transforming on its own, both Nck and Grb4 cooperated with v-Abl to transform NIH 3T3 cells and influenced the morphology and anchorage-dependent growth of wild type Ras-transformed cells. Nck and Grb4 therefore appear to be functionally redundant.

Growth factor binding to receptor protein tyrosine kinases (R-PTKs) 1 induces their dimerization and trans-phosphorylation, creating docking sites for proteins containing SH2 and PTB protein interaction domains (1). Many of these phosphotyrosine-binding proteins are effector enzymes, e.g. phospholipase C␥, the protein phosphatases SHP-1 and SHP-2, and p120 RasGAP (1). However, R-PTKs also bind to a number of adapter proteins that lack enzymatic activity and contain SH3 in addition to SH2 domains. Through SH3 domains, adapter proteins can bind to proline-rich motifs in downstream effectors, often recruiting them into multiprotein complexes at the plasma membrane. The adapter protein Grb2, for example, has the domain structure SH3-SH2-SH3 and can bind to prolinerich sequences in the Ras guanine nucleotide exchange factor, Sos, recruiting it to the plasma membrane, where it triggers the Ras-ERK pathway (2,3).
Another abundantly expressed adapter protein, Nck, consists of three juxtaposed SH3 domains and a C-terminal SH2 domain (4). The SH2 domain of Nck has been reported to bind a variety of growth factor receptors, including those for EGF (5), PDGF (6), vascular endothelial growth factor (7), and hepatocyte growth factor (8), as well as the insulin receptor substrate, IRS-1 (9), Eph (10,11), p62 dok (12), and focal adhesion kinase (13). The SH3 domains of Nck can interact with prolinerich motifs in multiple binding partners, including the Ser/Thr protein kinases PAK (14 -17), Prk2 (18), casein kinase I ␥ (19), and Nck interacting kinase (20). Nck has also been found to interact with the protein tyrosine kinase Abl (21), Sos (22), the Wiskott-Aldrich syndrome protein (18,23), a Drosophila protein tyrosine phosphatase dPTP61F (24), c-Cbl (25), and SAM68 (26). Although the physiological significance of most of these interactions is undetermined, several of the above SH3 binding partners (PAK, PRK2, and the Wiskott-Aldrich syndrome protein) appear to be downstream effectors of Rho family GTPases (18,27,28). Because Rho proteins have been implicated in cytoskeletal reorganization (29), Nck may be responsible for coupling R-PTK activation to cytoskeletal regulation. In support of this notion, the Drosophila homolog of Nck, DOCK, is located in photoreceptor growth cones and is involved in axonal guidance (30). Use of dominant inhibitory SH domain mutants has also implicated Nck in dorso-ventral axis development during Xenopus embryogenesis (31). The nck gene is located at a chromosomal break point associated with a variety of cancers (32), and Nck has been reported to weakly transform rodent fibroblast cell lines (5,33), also implicating it in oncogenesis.
In 1992, a partial mouse cDNA that encoded for an SH2 domain and part of an SH3 domain was isolated from a bacterial expression library and designated Grb4 (34). The predicted amino acid sequence shared 74% identity with human Nck. It was not clear from this original study whether the partial grb4 cDNA encoded the mouse ortholog of Nck or a separate protein.
Here we demonstrate that grb4 is, indeed, a unique human gene. Isolation of the full-length human grb4 cDNA showed that Grb4 shares the same SH domain composition and is 69% identical to Nck. Like Nck, Grb4 bound to Sos, PAK, and Prk2 and to activated growth factor receptors. Both Nck and Grb4 cooperated with v-Abl, Ha-Ras(WT), and Sos1 to modulate cell morphology and transformation and to induce gene expression via the Elk-1 transcription factor. These findings suggest that Nck and Grb4 may be functionally redundant.

MATERIALS AND METHODS
Library Screen-A human brain Marathon-ready cDNA library (CLONTECH) was used to isolate full-length grb4. The 5Ј-end of the gene was amplified using polymerase chain reaction with anchor primers provided with the library and nested grb4 primers located within the middle SH3 domain. Due to difficulty obtaining the complete 3Ј-end of the clone by rapid amplification of cDNA ends, we used further nested grb4 primers to isolate a central portion and specific primers to polymerase chain reaction amplify the 3Ј coding region. The grb4 fragments were subcloned into pCR2.1 (Invitrogen) and sequenced. Fulllength grb4 was generated by ligation using existing restriction sites within the three overlapping fragments.
Northern Analysis-Full-length nck cDNA (4) or a 266-base pair fragment of grb4 spanning base pairs Ϫ44 to ϩ222 were radiolabeled with [␣-32 P]dCTP using a random-priming DNA labeling kit (Boehringer Mannheim) as outlined by the manufacturer. Two human multitissue Northern blots (CLONTECH) were probed overnight with each individual fragment in hybridization solution (5ϫ SSPE, 10ϫ Denhardt's solution, 100 mg/ml denatured sheared salmon sperm DNA, 2.0% SDS plus 50% formamide) at 42°C. Blots were then washed extensively in 2ϫ SSC, 0.05% SDS at room temperature followed by two 20-min washes in 0.1ϫ SSC, 0.1% SDS at 50°C. The blots were exposed at Ϫ80°C with intensifying screens.
Mammalian Cell Culture-Cos, NIH 3T3, and SV40-immortalized mouse embryo fibroblast cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Hyclone) or, for NIH 3T3 cells, in 10% calf serum (Colorado Serum Company or Life Technologies, Inc.). Transforming focus and soft agar assays were performed as described (35).
Mammalian Protein Expression and GST Fusion Protein Interaction-For transient overexpression of Prk2 and PAK proteins, 100-mm plates of 50% confluent Cos cells were transfected with 6 g of either pFLAG-CMV2(prk2) (18) or pCMV6(pak(WT)) (16) using Lipo-fectAMINE (Life Technologies, Inc.) and harvested after 48 h. Sos1 was stably overexpressed from pRC-bac(sosF*) (39) in NIH 3T3 cells. Confluent plates of cells were lysed in 1 ml lysis buffer consisting of 20 mM Tris, pH 7.4, 50 mM NaCl, 0.5% IGEPAL CA-630, 5 mM EDTA, 10% glycerol, 19 g/ml aprotinin, 1 mM Na 3 VO 4 , and 1 mM phenylmethylsulfonyl fluoride. Cell lysates were cleared by microcentrifugation for 10 min at 4°C. GST fusion proteins of Nck, Grb4, or a control GST (containing the 40 C-terminal amino acids of Prk2) were expressed in BL21-DE3(lysE) bacteria as described (18). 2-10 g of glutathioneagarose bead-bound protein was tumbled with 0.5 ml of each individual cell lysate for 3 h at 4°C. Beads were then washed twice with the above lysis buffer and once with phosphate-buffered saline. Protein was solubilized by boiling in 2ϫ Laemmli buffer, and 50% of the resultant sample from Prk2-or Sos-incubated GST fusion proteins was separated by SDS-PAGE (7% gel). Because PAK co-migrated with the GST-Nck and GST-Grb4 fusion proteins during SDS-PAGE separation, beadbound fusion proteins incubated with lysate from cells overexpressing PAK were further incubated for 30 min at room temperature with a peptide representing residues 6 -21 of PAK1 to specifically release bound PAK into the solution. 50% of the supernatant from this competition was separated by SDS-PAGE (8% gel).
GST-fused SH2 Domain Interaction with Tyrosine-phosphorylated Proteins-Using human cDNA as a template, individual SH2 domains of Grb2 (codons 57-154), Nck (codons 275-377), and Grb4 (codons 278 -388) were generated by polymerase chain reaction and subcloned into pGEX-2T. Glutathione-agarose bead-bound GST fusion proteins of the SH2 domains were prepared as described above. Mouse embryo fibroblasts (40) were serum-starved for 48 h prior to stimulation with EGF (100 ng/ml, Upstate Biologicals Inc.) or PDGF-AA (50 ng/ml, Collaborative Biomedical Products) for the indicated times and lysed in 50 mM HEPES, 150 mM NaCl, 10% glycerol, 1% Triton X-100, 1.5 mM MgCl 2 , 1 mM EGTA, 100 mM NaF, 10 mM sodium pyrophosphate. 25% of the lysate from a confluent 100-mm plate was then incubated with 5 g of GST-fused SH2 domain proteins or full-length GST adapter proteins, as indicated, for 2 h at 4°C. The bead-bound proteins were washed as described above. The proteins were separated by SDS-PAGE (10% gel) and detected using the anti-phosphotyrosine antibody, PY99.
Co-immunoprecipitation of Sos1-Cos cells were transfected with pFLAG-CMV2(nck) or pFLAG-CMV2(grb4) together with pRC-bac-(sosF*) using LipofectAMINE. After 48 h, cells were lysed as above, and tagged Nck and Grb4 were immunoprecipitated with 3 g of M2 anti-FLAG antibody. Proteins were separated by SDS-PAGE and co-precipitated Sos1 detected by Western blotting using anti-Sos antibody.
Western Blotting-Proteins separated by SDS-PAGE were transferred onto Immobilon-P polyvinylidene fluoride membrane (Millipore) and blocked for 1 h at room temperature with 5% milk powder in TBS-Tween 20 (Tris-buffered saline, 0.05% polyoxyethylene-sorbitan monolaurate) or, in the case of phosphorylated tyrosine detection, with 3% bovine serum albumin in TBS-Tween 20. All antibody dilutions and wash steps were performed in TBS-Tween 20. Primary antibody incubations were for 1 h at room temperature with indicated antibodies. Secondary incubations were done for 30 min at room temperature with antibody conjugated to horseradish peroxidase (Amersham Pharmacia Biotech) and detected by enhanced chemiluminescence (Amersham Pharmacia Biotech). Anti-Nck antibody was from Pharmingen, anti-HA tag antibody was from Berkeley Antibody Co., anti-FLAG (M2) was from Sigma, anti-Sos was from Transduction Laboratories, and PY99 was from Santa Cruz Biotech. Other anti-Nck sources included Transduction Laboratories and Upstate Biotechnology.
[ 35 S]Methionine Labeling-Ha-Ras(Q61L)-transformed NIH 3T3 cells were metabolically labeled overnight with 250 Ci of [ 35 S]cysteine/ methionine (Amersham Pharmacia Biotech) per 60-mm plate in cysteine/methionine-free Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum that had been dialyzed against 20 mM HEPES, pH 7.4. Cell monolayers were washed with phosphatebuffered saline and lysed in 150 mM NaCl, 1% IGEPAL CA-630, 0.25% deoxycholate, 20 mM Tris, pH 7.4, 5 mM EDTA, 10% glycerol, 19 g/ml aprotinin, 1 mM Na 3 VO 4 , and 1 mM phenylmethylsulfonyl fluoride. The lysate was cleared by microcentrifugation and then tumbled with 20 g of the indicated GST fusion proteins on glutathione-agarose beads for 3 h at 4°C. Beads were then washed three times with lysis buffer and once with phosphate-buffered saline. 0.5 ml of each protein sample was separated by SDS-PAGE (7% gel) and then fixed and stained with 0.25% Coomassie Blue to confirm equal loading of GST fusion proteins. The gel was soaked in Amplify (Amersham Pharmacia Biotech), dried, and exposed to film for 48 h at Ϫ80°C.

RESULTS
Isolation of Full-length Grb4 cDNA-Sequence homology between human nck and a partial mouse grb4 cDNA isolate indicated that grb4 could either be the mouse ortholog of nck or a separate, related gene (34). Human expressed sequence tags, as well as an additional partial cDNA clone, 2 supported the notion that grb4 was a separate gene. Therefore, to obtain the fulllength Grb4 cDNA, we screened a human brain library using polymerase chain reaction/rapid amplification of cDNA ends (see under "Materials and Methods"). Like Nck, Grb4 consisted of three N-terminal SH3 domains and a C-terminal SH2 domain with short linker regions. The two proteins share 69% identity, with most divergence occurring in the regions between the SH domains ( Fig. 1). An in-frame stop codon was located just 5Ј of the presented sequence, indicating that the Grb4 N terminus does not extend beyond that of Nck. Indeed, significant homology between the nck and grb4 5Ј untranslated regions supports a common ancestry. Overall Nck has slightly more identity with the Xenopus Nck homolog than does Grb4, whereas the Drosophila protein is somewhat more divergent (Fig. 1).
Nck and Grb4 mRNAs Are Ubiquitously Expressed-Because Nck and Grb4 shared such a high degree of homology within their functional SH domains, we anticipated that they might serve similar functions in different tissues. We looked for differential expression of nck/grb4 mRNAs in a panel of 16 human tissues using commercially available Northern blots. Surprisingly, both nck and grb4 messages were almost ubiquitously expressed (Fig. 2). This apparent ubiquitous expression pattern was not due to cross-hybridization of the probes because their messages were of distinct sizes (the nck mRNA was slightly smaller). Furthermore, there were slight variations in abundance of the major transcripts in different tissues, most notably pancreas and thymus. The broad tissue distribution of nck was consistent with previously findings (5). Nck and Grb4 Bind to Common Target Proteins-We next tested the two adapter proteins to see whether the minor deviations in SH domain sequence might impart differential binding properties. Incubation of GST fusions of SH domain adapter proteins with lysates from [ 35 S]methionine-labeled NIH 3T3 cells revealed that, although distinct from that of Grb2, the binding profiles of Nck and Grb4 were similar to each other (Fig. 3A).
The SH3 domains of Nck have been shown to bind to a number of target proteins, such as Sos, PAK, and PRK2, all of which contain proline-rich peptide sequences (14 -18, 22, 41). We therefore addressed whether GST-Grb4 would also interact with these molecules. Recombinant Grb4 was found to precipitate FLAG-tagged PRK2 and Myc-tagged PAK1 from cell lysates with similar efficacy to Nck (Fig. 3B). Sos1 could also be co-immunoprecipitated by FLAG-tagged Grb4 or Nck from Cos cell lysates. The SH2 domain of Nck binds to a number of activated growth factor receptors and their substrates (5)(6)(7)(8)(9)(10)(11)(12)(13). Grb4 and Nck interacted with a similar pattern of phosphotyrosine-containing proteins. 3 To look more specifically at inter-action with these proteins, serum-starved mouse embryo fibroblasts were challenged with EGF or PDGF prior to incubation of lysates with glutathione bead-immobilized full-length adapter proteins or isolated SH2 domains. As shown in Fig. 3C, the SH2 domains of Grb2, Nck, and Grb4 all precipitated acti- Nck and Grb4 Serve as Substrates for Protein Kinases-Nck has been reported to be phosphorylated on multiple sites (primarily Ser/Thr residues) following agonist stimulation (5,16,33,(42)(43)(44). Examination of the Nck sequence revealed that most consensus phosphorylation sites are located between the SH domains, where Grb4 and Nck are most divergent. Following 32 PO 4 labeling of Cos cells transiently overexpressing FLAG-tagged Nck or Grb4, both molecules were heavily phosphorylated, even in the absence of challenge with forskolin/ IBMX (Fig. 4), serum, or phorbol myristate acetate. 3 Therefore, although the regions between the SH3 domains share less homology, both proteins are still susceptible to phosphorylation. Indeed, three consensus PKA phosphorylation sites are conserved in the region between the first and second SH3 domains of Nck and Grb4 (see Fig. 1). No such sites were present in the epitope tag.
Adapter Proteins Cooperate with v-Abl and Sos to Induce Transcriptional Activation-Nck has been reported to activate Ras and or Ras-induced phenomena by binding the Ras guanine nucleotide exchange factor, Sos, and replacing the role of Grb2 in mediating growth factor-induced recruitment of Sos to the site of Ras activation (22). Because Sos could co-precipitate with Nck and Grb4, we looked at the ability of these adapters to stimulate transcriptional activation via Elk-1 that is located downstream of the Ras/ERK cascade. In these assays, we cotransfected cells with a plasmid encoding the tyrosine kinase v-Abl to provide an upstream initiating signal (45). As shown in Fig. 5, co-transfection of NIH 3T3 cells with Grb2 augmented the transcriptional activity induced by v-Abl. Nck and Grb4 also enhanced the ability of Abl to induce luciferase expression, and although the effect of these adapter proteins was less than that seen with Grb2, they had similar potency to each other, suggesting a common mode of action. None of the adapter plasmids enhanced luciferase activity when transfected in the absence of v-Abl.
Because PAK has been implicated in Ras-independent ERK activation (14, 46), we next tested whether Nck and Grb4 might signal through PAK to induce gene expression. In our assay system, PAK did not induce transcriptional activation via Elk-1, either alone or in combination with Nck or Grb4. 3 As shown for v-Abl, however, we did find that both Nck and Grb4 could weakly synergize with Sos to induce transcriptional activation (Fig. 5).
Nck and Grb4 Cooperate with Ras and v-Abl to Induce Morphologic Transformation of NIH 3T3 Cells-Although it has previously been reported that Nck can induce transformation of rodent (NIH 3T3 and 3Y1) fibroblast cell lines (5, 33), in our hands Nck did not affect the morphology, growth rate, or sat-

FIG. 3. Nck and Grb4 interact with a common set of proteins.
A, lysates from [ 35 S]methionine/cysteine-labeled NIH 3T3 cells were incubated with GST fusion proteins containing the PRK2 C terminus (negative control), Grb2, Nck, or Grb4. Co-precipitated proteins were separated by SDS-PAGE and visualized by fluorography. B, lysates from Cos cells transiently expressing FLAG-tagged PRK2 or Myc-tagged PAK1 were incubated with 2-10 g of the indicated GST fusion proteins. Co-precipitation of PRK2 and PAK1 with GST fusion proteins was determined by immunoblotting. GST fusion proteins were detected by Ponceau S staining of polyvinylidene fluoride membranes prior to blocking. Lower panel, 48 h after co-transfection of Cos cells with plasmids encoding Sos1 and FLAG-Nck or FLAG-Grb4, cells were lysed and subjected to immunoprecipitation with anti-FLAG antibody. Sos co-precipitation was detected by Western blotting. C, phosphotyrosinecontaining proteins precipitated from EGF-stimulated (upper panel) or PDGF-stimulated (lower panel) mouse embryo fibroblasts by the indicated GST fusion proteins were detected using anti-Tyr(P) antibody, PY99. PDGF stimulation was for 5 min. Similar results were obtained using full-length adapter proteins or isolated SH2 domains (see Footnote 3). All data representative of at least three experiments. uration density of either cell type. 3 When NIH 3T3 cells were cotransfected with pZIP Ha-ras(WT) and pZIP(nck) or pZ-IP(grb4), however, we found that a number (1-10%) of the Ras-induced foci took on a unique morphology of highly rounded cells (Fig. 6). In contrast, Grb2 had no effect on Rasinduced focus forming activity (47). 3 When cells coexpressing Ras/Nck or Ras/Grb4 were isolated from the round transforming foci and plated in semi-solid medium, they developed anchorage-independent colonies that were distinct from those induced by oncogenic Ras(Q61L). Ras(Q61L) expression caused the formation of spherical colonies in approximately 50% of the transfected cells, whereas nearly 100% of Nck/Ras(WT)-and Grb4/Ras(WT)-expressing cells grew into diffuse colonies with poorly defined edges (Fig. 6C).
Because Nck and Grb4 cooperated with v-Abl to induce transcriptional activation, we also compared the ability of Nck and Grb4 to cooperate with v-Abl to induce transformation of NIH 3T3 cells. Co-transfection of v-Abl with plasmids encoding either adapter protein significantly enhanced focus formation. Whereas v-Abl alone induced the formation of only 1-5 small foci/g of transfected DNA, inclusion of the adapter proteins Grb2, Nck, or Grb4 resulted in the formation of 10 -20 larger foci/g that contained highly rounded cells (Fig. 6D). None of the v-Abl-induced foci could be established into transformed cell lines. Foci co-expressing the adapter proteins could, however, be maintained. Furthermore, the cells overexpressing Nck or Grb4 were much rounder than those expressing Grb2 (Fig. 6D). DISCUSSION A number of SH2/SH3 adapter proteins, including Grb2, Crk, and Nck, have been identified that couple R-PTKs to downstream effectors. A partial mouse grb4 cDNA sequence was described in 1992 having 66 and 74% identity to human nck at the DNA and predicted protein levels, respectively (34). Because a number of spliced variants and paralogs of the adapter proteins Grb2 (Grb3.3 and Grap) and Crk (CrkI, II, and CrkL) exist (48 -50), it was not clear whether Grb4 represented mouse Nck or a novel protein. We have now isolated a full-length human cDNA encoding the Grb4 protein, demonstrating that it is a distinct gene product rather than an ortholog of Nck.
In contrast to Grap, which shared 59% amino acid identity to Grb2 but had a very limited tissue distribution (49), the expression of the grb4 transcript appears to be almost ubiquitous. Furthermore, with a few exceptions, e.g. pancreas and thymus, the intensity of grb4 mRNA expression closely mirrored that of nck. Because all commercial anti-Nck antibodies tested crossreacted with Grb4 3 and we failed to generate Grb4-specific anti-peptide antisera, we were unable to confirm the tissue distribution of Nck and Grb4 at the protein level. However, while this study was under review, Chen et al. (51) reported the independent isolation of Grb4 as Nck␤ and confirmed that it was ubiquitously expressed at the protein level.
Comparison of the amino acid sequences of Grb4 and Nck indicate that they are most highly conserved within their SH domains. The key SH domain residues involved in ligand binding have been predicted by x-ray crystallographic and NMR structures. Because these residues are mostly conserved between Nck and Grb4, it was not surprising that the pull-down profile from [ 35 S]Met-labeled cells, phosphotyrosine-containing proteins, and specific SH3-binding proteins was similar for the two adapters. In contrast, it was reported that Nck binds to the PDGF receptor and to the EGF-receptor-associated adapter protein p62 dok less effectively than Nck␤ (51). This may have been due to differences between the Grb4 and Nck␤ sequences. A valine at codon 26 of Nck␤ (51) was not present in the Grb4 sequence or that of the human, Drosophila, or Xenopus Nck (Fig. 1). Nck␤ also has a unique E199Q substitution (51). Another possible explanation for the different SH2 binding affinities may be the instability of recombinant GST-Nck or Nck-SH2 domain proteins. We did observe a significant difference in the ability of Nck or Grb4 SH2 domains to precipitate R-PTKs when not using freshly prepared fusion proteins. 3 Both Nck and Grb4 were expressed at similar levels in mammalian cells, suggesting that this differential stability may not be significant in vivo. However, Grb4 but not Nck was reported to inhibit growth factor-induced DNA synthesis, suggesting that the adapter proteins play unique roles (51). Because we did not find any difference between the ability of Nck and Grb4 to induce Elk activation, the effect on DNA synthesis appears not to be mediated by the Ras/ERK pathway. An explanation for the apparent redundancy in the function of Nck and Grb4 could be that they are differentially expressed or regulated during development. For example, differential regulation has been reported for the three Ras proteins, in which Ki-but not Ha-or N-Ras was required for embryonic development (52,53). It will be interesting to determine whether Grb4 and Nck are functionally redundant following gene-targeted knockout. Grb4 has also recently been described as Nck2 (59).
Phosphorylation of Nck has been reported to be induced by growth factor stimulation (5,(42)(43)(44) or association with the downstream target kinase, PAK1 (16). In our hands, both Nck and Grb4 were constitutively phosphorylated and not sensitive to stimuli. This discrepancy may be due to the use of Cos cells or overexpression of the proteins in this study. Regardless, both Nck and Grb4 were phosphorylated to similar levels. Most consensus phosphorylation sites in Nck are located within the poorly conserved inter-SH domain regions, providing the potential for differential regulation of adapter protein binding by phosphorylation. However, a potential protein tyrosine kinase substrate site and three consensus cyclic AMP-dependent protein kinase phosphorylation sites (located between the first and FIG. 5. Nck and Grb4 cooperate with v-Abl and Sos1 to induce transcriptional activation via Elk-1. NIH 3T3 cells were cotransfected with either pSR␣MSVtk neo (v-abl), pZIP(sos1) or appropriate empty vector, pZIP(nck), or pZIP(grb4) as indicated, plus Gal4-Elk-1 and Gal4-luc reporter plasmids. Luciferase activity was detected as described (38). Data are representative of five or more experiments performed in triplicate. The columns and error bars show mean Ϯ S.D. second SH3 domains) are conserved between Nck and Grb4. Which sites are phosphorylated and whether there is any differential regulation of Nck versus Grb4 by phosphorylation will require further study. Indeed, no physiological role has so far been ascribed to Nck phosphorylation.
Nck has been shown to bind to the Ras guanine nucleotide exchange factor, Sos, in vivo and to activate the c-Fos promoter in a Ras/ERK-dependent fashion (22). In another study, however, Nck dominant inhibitory mutants, unlike Grb2 mutants, did not block the activation of ERKs induced by oncogenic (⌬SH3) c-Abl when transiently overexpressed in 293-T cells (54). Furthermore, overexpression of Nck did not rescue NIH 3T3 cells from the inhibitory effects of the Sos C terminus, implying that Nck did not couple Sos to MAP kinase activation FIG. 6. Nck and Grb4 can transform NIH 3T3 cells by cooperating with Ha-Ras(WT) and v-Abl. A, NIH 3T3 cells were transfected with pZIPH(ras) along with pZIP, pZIP(nck), or pZ-IP(grb4), as indicated. Transforming foci were photographed (magnification, ϫ 10) after culture in regular growth medium for 14 days. B, individual transforming foci from A were isolated and selected on G418. Subconfluent cells were photographed (magnification, ϫ 20). C, cells from B were grown in 0.6% agar for 14 days. Pooled populations of cells stably expressing Nck and Grb4 in the absence of Ras did not produce colonies (magnification, ϫ 4). D, NIH 3T3 cells were transfected with pSR␣MSVtk neo (v-abl) along with pZIP, pZIP(nck), or pZIP(grb4), as indicated. Transforming foci were photographed (magnification, ϫ 4) after culture in regular growth medium for 14 -20 days. Individual transforming foci from D were isolated and selected on G418. Subconfluent cells were photographed (magnification, ϫ 10). Focus assays are representative of three or more experiments done in quadruplicate. Soft agar data are representative of two experiments performed in triplicate. (41). One possible explanation for this discrepancy is that the binding efficiency of Nck and Grb4 to Sos is weaker than that of Grb2, such that competition assays are less effective. Although we found no effect of Nck, Grb4, or Grb2 on basal transcriptional activity, each of the adapters could co-immunoprecipitate Sos and weakly cooperated with v-Abl and Sos1 to induce luciferase expression via Elk-1. The differences in cell types or culture conditions employed in these studies likely influenced the biological observations. Indeed, we have observed that growth of NIH 3T3 cells in bovine calf serum from various sources can result in significant differences in the magnitude of adapter protein-induced transcriptional responses. 3 Typically, we have seen less luciferase activity induced by Nck/Grb4 than by Grb2, suggesting that Grb2 is more effective at recruiting Sos to the site of Ras activation. Transcriptional activation differences among the adapters could be due to the presence of R-PTKs that preferentially interact with Grb2. Alternatively, the receptors that associate with Grb2 may be more effectively localized to the vicinity of Ras. It has been reported that activation of PAK can result in phosphorylation of ERKs independently of Ras (14,46). Because Nck can bind to PAK and influence both its location and activation (14,16,17), it was possible that Nck and Grb4 were circumventing Ras and activating Elk via PAK. However, in our hands, transfection of pCMV6(pak1) had no effect on luciferase activity in the presence or absence of Nck.
Two previous studies reported weak transformation of rodent fibroblasts by Nck (5,33). Although we saw no effect of Nck or Grb4 on their own, both proteins could cause a rounding up of Ras(WT)-transformed cells. A similar phenotype was observed by co-expression of Nck with R-Ras(G38V). 4 The rounding of the cells suggests a loss of stress fibers and/or cell contact with its substratum. A similar morphological change was reported following membrane targeting of PAK in 293T kidney cells using a myristoylated Nck-SH3 domain (14), implicating PAK in Nck-induced cytoskeletal events. The morphology of cells derived from the Ras plus Nck/Grb4 foci, as well as the diffuse/migratory nature of the soft agar colonies compared with those induced by Ras(Q61L), is similar to that induced by co-expression of Ras with activated Rac1 or RhoA (55). This supports the notion that Nck is responsible for coupling R-PTK to Rho family effectors, such as PAK or PRK2.
The observation that Nck and Grb4 only affected the morphology of a subset of NIH 3T3 cells was surprising. Heterogeneity in the NIH 3T3 cell population has, however, been reported. Although v-Abl induces cell cycle arrest at G 1 in the majority of NIH 3T3 cells, it transforms a subset representing 5-8% (56). The reason for this heterogeneity is not clear but may involve the absence of a proposed inhibitory feedback loop in some 3T3 cells or lack of a v-Abl effector (56). It would be interesting to determine whether the same population of cells respond to both Nck-and v-Abl-induced transformation. Furthermore, this heterogeneity may, at least in part, explain the differences in biological activity of the Nck adapter protein observed in different studies on NIH 3T3 cells (Refs. 5, 22, 33, and 41 and this report).
Although activating mutations in ras are found in only ϳ30% of human malignancies, Ras activation by upstream oncoproteins, e.g. BCR/Abl in leukemias or ErbB2/Neu in breast cancer, is also believed to contribute to transformation. Adapter proteins, such as Grb2, are required to transduce these deregulated signals from PTKs to Ras (57), and their amplification could contribute to the transformed phenotype. Indeed, the adapter protein Grb7 has also been shown to be up-regulated along with ErbB2 in breast cancer cell lines and tumor samples (58). Here, we have described the characterization of Grb4, an adapter protein that shares many structural and functional properties with Nck. Both proteins share the ability to couple R-PTKs to Ras and to Rho protein effectors and to cooperate with signaling molecules to promote cellular transformation. It will be interesting to determine whether Nck or Grb4 overexpression contributes to human malignancies and in what tissues or at what point in development the body relies on Grb4 expression.