[Frontiers in Bioscience 2, d501-518, October 15, 1997]
Reprints
PubMed
CAVEAT LECTOR




Table of Conents
 Previous Section   Next Section

ROLE OF c-Src TYROSINE KINASE IN EGF-INDUCED MITOGENESIS

Allison P. Belsches, Michelle D. Haskell, and Sarah J. Parsons

Department of Microbiology and Cancer Center, Box 441, Health Sciences Center, University of Virginia, Charlottesville, VA 22908

Received 10/7/97 Accepted 10/14/97

2. INTERACTION OF c-Src WITH MITOGENIC RECEPTORS

2.1 Platelet-derived Growth Factor

The PDGF receptor (PDGFR) is a 185 kDa transmembrane tyrosine kinase that becomes enzymatically activated following ligand binding and autophosphorylates at multiple sites in the cytoplasmic domain of the molecule. These phosphorylated tyrosine residues serve as docking sites for a variety of signal transducers, including the adaptor proteins, GRB2, Nck, and SHC, Raf-1 (a serine/threonine kinase), p120RasGAP (a GTPase activating protein for Ras), PI3 kinase (phosphatidylinositol-3 kinase), SHP2 (a tyrosine phosphatase) and PLC-gamma (phospholipase C-gamma) (1-3). The first evidence that c-Src and its family members were involved in PDGF signaling was provided by Ralston and Bishop (4), who observed that c-Src becomes activated upon PDGF stimulation. Kypta et al. (5) later demonstrated that not only c-Src, but also c-Fyn and c-Yes, are activated in a PDGF-dependent manner and become transiently associated with the receptor following PDGF treatment. Association between Src family kinases and the receptor is believed to involve phosphotyrosine (pTyr)-SH2 interactions, since the SH2 domain of c-Fyn is required for binding to the receptor in vitro (6), and mutation of two tyrosine phosphorylation sites in the juxtamembrane region of the receptor, Y579 and Y581, results in a reduction in both PDGF-induced c-Src activation and binding to the receptor in vivo (7). These data and results from in vitro peptide binding studies (8) suggest that Y579 and Y581 directly mediate binding of c-Src to the PDGFR, although it is not clear whether c-Src or the receptor phosphorylates these sites. The integrity of sequences surrounding Y857 of the PDGFR (the major autophosphorylation site in the kinase domain that is homologous to Tyr 416 of Src) is also important for the binding of c-Src, since mutation of this residue significantly reduces association of the two kinases. However, Y857 is not believed to be the direct site of interaction (9).

Activation of c-Src by PDGF is transient and accompanied by the appearance on c-Src of two novel serine phosphorylations in the N-terminal 16 kDa fragment (S12 and an unidentified S residue; 10), and a novel tyrosine phosphorylation, Y138 (11). Y138 is located in the SH3 domain of c-Src, and in vitro peptide binding experiments (11) suggest that phosphorylation of Y138 may alter the ability of the SH3 domain to associate with other signaling molecules through SH3/polyproline interactions. Mutation of Y138 generates a c-Src molecule that is inhibitory for PDGF-induced DNA synthesis (12, 13), suggesting that the SH3 domain is required for mitogenesis. That Src family members as a group are required for PDGF-initiated transit of cells through G1 to S phase is supported by the inhibitory effects of the introduction of kinase inactive c-Src or an antibody specific for the C-terminal domain of Src family members on BrdU incorporation (14).

Interaction of c-Src with the PDGFR appears to have consequences not only for c-Src but also for the receptor. Hansen et al. (15) have shown that Tyr 934 in the kinase domain of the PDGFR is phosphorylated by c-Src both in vitro and in vivo, and mutation of this site to phenylalanine and expression in intact cells results in a decreased mitogenic signal and an increase in chemotaxis and motility. In addition, the PDGF-stimulated tyrosine phosphorylation of PLC-gamma is enhanced in mutant receptor cells compared with wild-type (wt) receptor cells. These data suggest that phosphorylation of Y934 by c-Src positively regulates mitogenesis, while negatively regulating cell movement, possibly via a pathway that involves PLC-gamma.

2.1 Colony Stimulating Factor

The receptor for CSF-1, c-Fms, is a transmembrane molecule with strong homology to the PDGFR (16). CSF-1 is released by osteoblasts in response to treatment with parathyroid hormone (PTH) (17) and stimulates the proliferation, differentiation and survival of cells of the mononuclear phagocytic lineage (18). Osteoclasts respond to CSF-1 by rearranging actin and undergoing cytoplasmic spreading (17), a phenomenon shared with a variety of cell types in response to polypeptide growth factors. Interactions of signaling proteins with c-Fms are much less well-characterized as compared to other receptor tyrosine kinases. PI3 kinase has been demonstrated to associate with the receptor, but this association is not required for CSF-1-stimulated cell growth (19). Upon CSF-1 stimulation of macrophages (which normally express c-Fms) or mouse fibroblasts (which are engineered to express c-Fms), Src family members associate with the receptor and become activated (20). That binding between Src and the receptor may be mediated by pTyr/SH2 interactions is suggested by the finding that a GST-fusion protein of the SH2 region of c-Fyn complexes with the receptor in vitro. In vivo association can be reduced by mutation of the receptor at Tyr 809, a major phosphorylation site in the kinase domain that is homologous to Tyr 416 in Src. Roche et al. (21) have demonstrated that CSF-1-induced DNA synthesis requires Src family members, by microinjecting a C-terminally directed antibody that inhibits Src family tyrosine kinase activity in vitro. Though not as thoroughly studied as the PDGFR, these data suggest that the interaction of c-Fms with c-Src is regulated in much the same fashion as c-Src with the PDGFR and that the interaction has similar effects on the binding partners.

2.3 Fibroblast Growth Factor

The association of c-Src with the basic FGF receptor (bFGFR) appears to be cell type-specific. For example, in NIH3T3 cells c-Src has been shown to co-precipitate with FGFR in an FGF-dependent fashion. This association may be mediated by the SH2 domain of

c-Src, since GST fusion proteins of the SH2 domain of v-Src precipitate FGFR (22). In lung capillary endothelial cells, FGF stimulation leads to an increase in c-Src autophosphorylation activity but no complex formation between c-Src and FGFR. In porcine aortic endothelial cells engineered to express the FGFR, no complex between c-Src and FGFR is seen, and FGF stimulation results in a down-modulation of c-Src kinase activity (23). These results suggest the involvement of cell-type-specific factors which mediate interactions between the receptor and c-Src and regulate c-Src activity. These results also raise the question of whether c-Src binding to other receptors may be influenced by third party proteins.

2.4 Epidermal Growth Factor Receptor and family members

The EGFR is a 170kDa transmembrane receptor tyrosine kinase that dimerizes, phosphorylates multiple residues in its cytoplasmic domain, and binds and/or phosphorylates a variety of signaling proteins in a ligand-dependent fashion, similar to the PDGF, CSF-1, and bFGF receptors. Many of the binding proteins and substrates of the EGFR are the same as for the PDGFR, including GRB2, shc, p120RasGAP, PLC-gamma, and PI3 kinase (for review 24-27). Tyrosine phosphorylation of PLC-gamma initiates actin rearrangement, Ca2+ mobilization from intracellular stores, and activation of protein kinase C (28, 29), while changes in inositol phosphate metabolism and increases in anti-apoptotic signals are brought about by tyrosine phosphorylation of PI3 kinase (2). Proliferative signals are transduced through SHC and/or GRB2/SOS (2) to Ras and the MAP kinase cascade, which regulates gene expression via phosphorylation of transcription factors. The treatment of cells with EGF, therefore, stimulates multiple, parallel signaling pathways that are thought to coalesce to initiate cell division. These pathways are depicted in figure 1.

Figure 1. Postulated effect of c-Src on EGF mitogenic pathways. Two signaling pathways of the EGF receptor are depicted, namely, those involving PLC-gamma and MAP kinase. Tyrosine residues 845 and 1101 of the EGF receptor are phosphorylated by c-Src, and phosphorylation of Y845 is believed to augment receptor tyrosine kinase activity, as manifested by increased phosphorylation of the receptor targets, PLC-gamma and SHC.

2.4.1 Requirement for c-Src in EGF-induced mitogenesis

A major goal of our laboratory has been to determine if c-Src is required for EGF-induced cell proliferation, and if so, to elucidate its role. Our studies were initiated by examining c-Src from a variety of rodent and avian cells for alterations in specific kinase activity following EGF stimulation. While EGF-dependent activations of c-Src were detected in several experiments, such changes were not consistently observed (30), prompting us to test directly in intact cells the notion that c-Src is involved in mitogenesis. Thus, c-Src was overexpressed in C3H10T murine fibroblasts (a cell line that exhibits tight contact inhibition, a very low frequency of spontaneous focus formation in monolayer, and normal levels of EGFR), and transfectants were tested for their responsiveness to EGF, using [3H]-thymidine incorporation as a measure of progression through the cell cycle. We found that clonal cell lines overexpressing varying levels of c-Src ranging from 2-30 times above endogenous exhibit EGF-induced DNA synthesis that is 2-5 fold higher than in normal cells (30). Such cell lines are indistinguishable from normal, parental 10T cells in morphology and saturation density, and unable to form colonies in soft agar. Receptors in c-Src overexpressors are present in nearly equal numbers and exhibit similar affinities for ligand as receptors on control cells. Cell lines overexpressing structurally-altered forms of c-Src (kinase inactive, SH2-domain defective, and myristylation-defective) fail to potentiate the EGF response as observed with wildtype c-Src. In fact, all the c-Src mutants act in a dominant negative fashion (31), indicating that c-Src is required for signaling through the EGFR. This finding was corroborated by Roche, et al. (21), using microinjection of antisera to Src family members or kinase inactive c-Src cDNA and measuring EGF-induced DNA synthesis by BrdU incorporation.

Chronic stimulation of cells overexpressing the EGFR results in colony formation in soft agar and tumor formation in nude mice (32). To determine if c-Src affects EGF-induced tumorigenesis as it does EGF-induced mitogenesis, cell lines which overexpress the human EGFR (HER1) alone or in the presence of overexpressed c-Src were generated from the 10T parent and examined for levels of EGF-dependent [3H]-thymidine incorporation, growth in soft agar and tumor formation in nude mice. In this model system, expression of both HER1 and c-Src results in a synergistic increase in growth and tumorigenicity, as compared to cell lines overexpressing c-Src or HER1 alone. The differences in the tumorigenicity of the various cell lines is striking, with the double c-Src/HER1 overexpressors forming tumors of greater than 1.0 cm in diameter at each injection site within one week, and the single c-Src or HER1 overexpressors forming only pinhead-sized tumors at only 50% of the injection sites (33).

The mechanism of the synergism between HER1 and c-Src is not fully understood at this time, but formation of a stable complex mediated by SH2/pTyr interactions may play a role. Using antibodies specific to c-Src, an EGF-dependent, in vivo c-Src/HER1 complex can be detected in double overexpressors but not in single c-Src or HER1 overexpressors (33), and only when mild detergent conditions are used for extraction. These results suggest that the association is weak or indirect. Co-immunoprecipitation of c-Src with HER1 has also been observed in human carcinoma cell lines (30, 34-36). That the interaction may be direct is suggested by the finding that GST fusion proteins containing the c-Src SH2 domain both precipitate the receptor from crude cellular lysates and bind the receptor directly in a Far Western assay (37, Biscardi et al., 1997, in preparation). Using a panel of truncated receptors (38) or a panel of peptides to compete the precipitation of the receptor by a GSTc-SrcSH2 fusion protein (37, 34) other investigators have identified Y992 of the receptor as a potential site of c-Src SH2 binding, while Stover et al., (34) have implicated Y891 and 920 as additional sites. These results suggest that c-Src might bind multiple sites on the receptor.

Are there immediate consequences on c-Src of this association? Several investigators (39, 40) have reported an EGF-dependent activation of c-Src kinase activity in cells overexpressing HER1. However, results from our laboratory in both the 10T mouse model and human carcinoma cell lines, as well as from others (41), have indicated no reproducible, significant activation of c-Src, in the presence or absence of EGF treatment. However, Biscardi et al. (1997, in preparation) have evidence in the 10T model that HER1, in complex with c-Src, gains two novel sites of tyrosine phosphorylation, which have been identified as Y845 and Y1101 (figure 1). Both sites are phosphorylated in vitro and in vivo and require the presence of c-Src (Biscardi et al., 1997, in preparation). Phosphorylation of Y845 requires a kinase active c-Src, suggesting that c-Src phosphorylates Y845 directly, while phosphorylation of Y1101 requires the presence but not the kinase activity of c-Src (Tice et al., 1997, in preparation). This finding suggests that c-Src either recruits another tyrosine kinase into the complex or induces HER1 to phosphorylate Y1101. Y845, located in the kinase domain of HER1, is the Src Y416 homologue. When Y416 of Src is phosphorylated, the enzyme becomes hyperactivated, which suggests that phosphorylation of Y845 by c-Src may result in a hyper-activation of receptor kinase activity. Indeed, the HER1 substrates, SHC and PLC-gamma; are more highly phosphorylated in the HER1/c-Src double overexpressors than in either of the single overexpressors, providing evidence for the receptor hyper-activation hypothesis (33). Sato et al.(42) have also observed phosphorylation of Y845 in HER1 from A431 cells, and Stover et al. (34) have identified Y891 and Y920 as additional phosphorylation sites of c-Src.

2.4.2 Involvement of HER family members and c-Src in the etiology of human breast cancer

A survey of the literature yields much evidence for potential functional interactions between HER1 or its family members (HER2, 3, and 4) and c-Src in human breast tumor progression. Overexpression and/or gene amplification of HER1 and HER2 is seen in approximately 20-30% of human breast tumors, and overexpression of HER2 has been correlated with poor patient prognosis (43, 44). Furthermore, c-Src has been found to be elevated in 70/70 breast tumors examined (45). The increases in protein level and/or activity of both HER family members and c-Src in a significant proportion of human breast tumors suggest that the two families of molecules might functionally interact in human tumors, as they do in 10T cells. To the extent feasible, this hypothesis has been tested by Biscardi et al. (36) in a panel of 14 human breast tumor cell lines and three tissue samples from patients. A direct correlation between levels of c-Src/HER family member overexpression, complex formation, phosphorylation of Y845 and Y1101, relative levels of SHC phosphorylation, and tumor formation in nude mice (with several exceptions) were observed.

Taken together, the in vivo and in vitro data from the 10T mouse fibroblasts and the breast cancer cell lines suggest a physical and functional association of HER1 and c-Src and a potential synergism between the two kinases in human breast tumor progression. However, clinical data suggest that overexpression of HER1 occurs as a late event in breast tumor progression, while overexpression of HER2 occurs as an intermediate event. These findings argue that HER1/c-Src interactions are more important in later stage tumor progression, whereas HER2/c-Src interactions (if they occur) are more important in mid-stage tumor formation and progression.

2.4.3 c-Src and HER2 (erbB2/neu) Receptor

HER2 (erbB2/neu) is a 185kDa transmembrane protein that is highly homologous to HER1 (80%) and the rat neu gene (46, 47). When mutated at a single residue in the transmembrane domain, the rat neu gene becomes oncogenic (48), but no such mutations have been found in human cancers, even when HER2 is overexpressed (44). Activation of the HER2 kinase domain appears to occur by homodimerization under conditions of overexpression (48, 49). At the present time, downstream targets of HER2 appear indistinguishable from those of HER1 and include SHC (50), PLC-gamma (51), and p120RasGAP (51). In addition, activation of the MAP kinase pathway via GRB2/SOS/Ras has been demonstrated in breast cancer cell lines that overexpress HER2 (52).

In spite of the identification of downstream targets, no specific ligand for HER2 has yet been found that could account for its frequent activation in human breast cancers. Heregulin (which was at first thought to be the ligand for HER2) was found to bind HER3 and HER4, inducing not only homodimerization but also heterodimerization with HER2 (HER3/HER2 or HER4/HER2) (53-56). Heterodimerization appears to be the predominant complex responsible for the trans-phosphorylation of the receptors and initiation of cellular signaling (49, 54, 57, 58). Stimulation of cells with EGF also causes heterodimerization between HER1 and HER2 (in addition to homodimerization of HER1) (49, 57, 59, unpublished observations, Belsches and S. Parsons), thus providing a potential mechanism by which the two receptors share downstream targets.

The question under investigation at the present time is whether c-Src is a co-transducer of signals through HER2 as it is through HER1. Does c-Src enhance the inherent oncogenic potential of HER2? Because HER2 is so abundantly and frequently overexpressed in human tumors, as is c-Src, these question carry greater significance. Luttrell et al. (37) showed that HER2 can be precipitated by the GSTc-SrcSH2 fusion protein from extracts of the human breast carcinoma cell line, SKBR3, suggesting that stable complexes may also form between c-Src and HER2 in vivo. Our laboratory has detected in vivo HER2/c-Src complexes in three of fourteen human breast tumor cell lines and in three of thirteen tumor tissues (Belsches et al., 1997, in preparation), and Muthuswamy and Muller (60) have shown association between c-Src and HER2 in mammary tumors from HER2 transgenic mice. In addition, we have found that while constitutive association between c-Src and HER2 can be detected in serum-starved breast cancer cells, the amount of HER2 which co-immunoprecipitates with c-Src is augmented when the cells are stimulated with EGF. In contrast, heregulin stimulation of the cells does not increase the amount of the c-Src/HER2 complex above constitutive levels. These data suggest that the association between HER2 and c-Src may be augmented by activation of HER1 and HER1/HER2 heterodimerization and not through HER3 or HER4. Further studies on HER2/c-Src interactions are currently underway.