Epidermal growth factor receptor (EGFR), a member of the ErbB family of receptor tyrosine kinases, is overexpressed in as many as 60% cases of breast and other cancers. EGFR overexpression is a characteristic of highly aggressive molecular subtypes of breast cancer with basal-like and BRCA1 mutant phenotypes distinct from ErbB2-overexpressing breast cancers. Yet, EGFR is substantially weaker compared with ErbB2 in promoting the oncogenic transformation of nontumorigenic human mammary epithelial cells (human MEC), suggesting a role for cooperating oncogenes. Here, we have modeled the co-overexpression of EGFR and a biologically and clinically relevant potential modifier c-Src in two distinct immortal but nontumorigenic human MECs. Using a combination of morphologic analysis and confocal imaging of polarity markers in three-dimensional Matrigel culture together with functional analyses of early oncogenic traits, we show for the first time that EGFR and c-Src co-overexpression but not EGFR or c-Src overexpression alone unleashes an oncogenic signaling program that leads to hyperproliferation and loss of polarity in three-dimensional acinar cultures, marked enhancement of migratory and invasive behavior, and anchorage-independent growth. Our results establish that EGFR overexpression in an appropriate context (modeled here using c-Src overexpression) can initiate oncogenic transformation of nontumorigenic human MECs and provide a suitable in vitro model to interrogate human breast cancer–relevant oncogenic signaling pathways initiated by overexpressed EGFR and to identify modifiers of EGFR-mediated breast oncogenesis. [Cancer Res 2007;67(9):4164–72]
- mammary epithelial cells
- breast cancer
- 3D culture
- cell polarity
Members of the ErbB family of growth factor receptor tyrosine kinases (RTK), in particular the epidermal growth factor receptor (EGFR or ErbB1) and ErbB2, have been implicated in the pathogenesis of human cancers and have emerged as targets for rational cancer therapy, as exemplified by the development of trastuzumab (Herceptin) for the treatment of ErbB2-overexpressing breast cancer patients ( 1).
EGFR is overexpressed in over a quarter of breast cancer patients and in as many as 60% cases in some studies ( 2). EGFR overexpression has also been reported in other cancers such as those of the colon, lung, and ovary ( 3). Infrequent co-overexpression of EGFR and ErbB2 suggests that EGFR overexpression defines a distinct subset of breast and other solid tumors ( 3, 4). Indeed, microarray-based molecular profiling showed the EGFR overexpression to be highly prevalent in breast cancer subtypes with the basal cell–like and BRCA1 mutant phenotypes with the worst clinical outcomes ( 5– 7). Despite this evidence validating EGFR overexpression as a biomarker of poor prognosis and a therapeutic target, success with EGFR-directed kinase inhibitors has been limited primarily to non–small-cell lung cancer with activating EGFR mutations ( 8). It is crucial to gain further insights into oncogenic signaling by EGFR and the role of potential modifiers that may need to be concurrently targeted for effective therapy.
Although ErbB2 and EGFR are both linked to breast cancer, in vitro cell biological models have shown more potent transforming activities of ErbB2 ( 9– 11). Notably, inducible dimers of ErbB2 (or ErbB2-EGFR heterodimers) but not of EGFR expressed in nontumorigenic MEC line MCF10A reinitiated the proliferation of quiescent acini grown in three-dimensional Matrigel cultures, and only ErbB2 activation obliterated the lumen formation ( 12). These findings suggest that additional alterations promote EGFR-dependent oncogenic transformation in MECs, and that the three-dimensional acinar culture could reveal the role of such collaborators. The non–receptor protein tyrosine kinase c-Src is likely to be one such collaborator.
Clinical-pathologic studies have shown elevated c-Src protein levels and/or activity in about 70% of primary human breast and other epithelial tumors, often with concomitant ErbB2 or EGFR overexpression (reviewed in refs. 3, 13, 14). The c-Src–overexpressing tumors include a subset of late stage breast cancers with more aggressive disease (reviewed in refs. 3, 13). Experimental studies support a positive role of c-Src overexpression in signaling through ErbB receptors and other RTKs ( 3). Studies in breast cancer cell lines and limited analysis of tumor tissues showed that c-Src associates with EGFR or ErbB2 ( 3, 14). However, analyses of Src-dependent oncogenesis in the past have used activated Src mutants rarely seen in human cancers and never observed in breast cancer (reviewed in refs. 13– 15). Parsons et al. first showed the cooperative tumorigenesis by wild-type c-Src and EGFR in C3H10T1/2 murine fibroblasts ( 3, 16) and showed that c-Src overexpression in EGFR-expressing metastatic breast and other cancer cell lines promoted cell migration ( 3). Thus, c-Src represents a clinically and biologically relevant modifier of EGFR-mediated oncogenesis, but such a role has not been investigated in nontransformed human epithelial cell models.
Here, we modeled the co-overexpression of EGFR and c-Src in well-defined human MECs to address the hypothesis that c-Src is a modifier of EGFR-mediated mammary epithelial oncogenesis. We engineered immortal but nontumorigenic human MECs to overexpress EGFR and c-Src, either individually or in combination, and examined the biological consequences of these manipulations in two-dimensional and three-dimensional culture systems. We show that c-Src overexpression, while of minor consequence by itself, dramatically promotes the ability of overexpressed EGFR to induce a phenotype of branched and irregular acini formation in three-dimensional Matrigel cultures and enhances the migration, invasion, and anchorage-independent growth in soft agar. Thus, our findings establish that c-Src can modify EGFR-mediated mammary oncogenesis and provide a simple and tractable system to identify other EGFR modifiers and to dissect c-Src–regulated EGFR oncogenesis pathways.
Materials and Methods
Retroviral constructs. The wild-type human EGFR and ErbB2 were PCR amplified from cDNA templates ( 17, 18) and directionally cloned into the XhoI and HpaI sites of the pMSCV-Puro vector (Clontech). Mouse c-Src sequence was PCR amplified from the pLNCX-Src plasmid (provided by Dr. Joan Brugge) and directionally cloned into the BglII and XhoI sites of the pMSCV-Hygro vector (Clontech). The kinase-inactive c-Src K297R mutant was generated using the QuikChange site-directed mutagenesis kit (Stratagene). All constructs were verified by DNA sequencing.
Antibodies and other reagents. The following antibodies were obtained from commercial sources: rabbit polyclonal (pAb) anti-EGFR, pAb anti-ErbB2, monoclonal (mAb) anti-c-Src, pAb anti-PLCγ1, pAb anti–phosphorylated AKT (pAKT), and mAb anti-AKT were from Santa Cruz Biotechnology; pAb anti–phosphorylated EGFR-pY1068, anti–phosphorylated EGFR-pY845, anti–phosphorylated EGFR-pY1173, anti–phosphorylated EGFR-pY992, anti–phosphorylated Src (416), anti–phosphorylated Src homology and collagen (Shc), anti-Shc, anti–phosphorylated PLCγ1, anti–phosphorylated signal transducer and activator of transcription 3 (anti-pSTAT3), anti-STAT3, and anti-STAT5 were from Cell Signaling Technology; mAb anti-EGFR Ab-12 and pAb anti-pSTAT5b were from Lab Vision Corp.; mAb anti-β actin was from Sigma-Aldrich; mAb anti-α6 integrin was from Chemicon. mAb anti-EGFR (clone 528; American Type Culture Collection) and mAb anti-E-cadherin (clone E 4.6; provided by Drs. Jonathan Higgins and Michael Brenner; ref. 19) were Protein G purified from hybridoma supernatants. Purified mAb 4G10 (anti-pY) was provided by Dr. Brian Druker. Purified mouse EGF was from Sigma-Aldrich. The Src inhibitor PP2 and EGFR inhibitor AG1478 were from Calbiochem. Matrigel was purchased from BD Biosciences.
MEC lines. 16A5 is an HPV16 E6/E7–immortalized derivative of reduction mammoplasty-derived 76N normal MEC line ( 20). Spontaneously immortalized MEC line MCF10A ( 21) was obtained from Dr. Senthil Muthuswamy ( 22). The MECs and their retroviral transductants (below) were cultured in DFCI-1 medium supplemented with 12.5 ng/mL EGF ( 20).
Retroviral infections. Retroviral supernatants were generated by calcium phosphate–mediated cotransfection of the expression plasmids and the packaging plasmid pIK into the packaging cell line TSA54 ( 23). The retroviral supernatants, collected 24 h after transfection, were used to infect subconfluent MECs in three sequential 4-h incubations in the presence of 4 μg/mL polybrene (Sigma-Aldrich). Transductants were selected in puromycin (16A5, 0.5 μg/mL; MCF10A, 1 μg/mL) or hygromycin (16A5, 5 μg/mL; MCF10A, 10 μg/mL), 48 h after infection. Double transductants were derived by serial infection.
EGF stimulation, immunoprecipitations, immunoblotting, and gel electrophoresis of proteins. Cells at 60% to 70% confluence in DFCI-1 medium were switched to growth factor–deficient D3 medium ( 24) for 48 h and stimulated with EGF (10 ng/mL unless otherwise indicated) for the indicated time periods. Cell lysates were prepared in cold lysis buffer containing 0.5% Triton X-100, 50 mmol/L Tris (pH 7.5), 150 mmol/L sodium chloride, 1 mmol/L EDTA, 1 mmol/L phenylmethylsulfonyl fluoride, 1 μg/mL each of leupeptin and pepstatin, 1 mmol/L sodium orthovanadate, and 10 mmol/L sodium fluoride. Immunoprecipitations and Western blotting were done as described ( 25). The films were scanned with an HP ScanJet 4 scanner (Hewlett-Packard), and the density of bands was quantified using the Labworks 4.5 Image Acquisition and Analysis software for Windows (UVP).
Transwell cell migration assay. The Transwell chambers with 8-μm pores (Corning) were coated overnight at 4°C with 2% laminin followed by blocking with 1% bovine serum albumin for 2 h. Cells (2.5 × 104) deprived of EGF for 48 h in D3 medium were plated in the top chamber in 0.1 mL D3 for 2 h to allow attachment, and D3, with or without EGF (12.5 ng/mL), was added in the bottom chamber. PP2 was added to cells in top chamber 2 h before migration assay and to D3 medium in the bottom chamber. After 5 h, cells on the top side were removed by scraping, and the migrated cells were fixed in methanol at −20°C and visualized using the Diff-Quik stain (Dade Behring). The number of cells in three high-power (×400) fields was enumerated for each filter. Experiments were repeated a minimum of three times. Data were analyzed using the Student's t test, and P < 0.05 was considered significant.
Matrigel invasion assay. Invasion assays were carried out using the 24-well BD Biocoat Matrigel Invasion Chambers (BD Biosciences) according to the manufacturer's recommendations. Briefly, 2 × 104 cells in 0.2 mL D3 were added to the top chambers of Matrigel-coated or control wells with D3 medium containing 12.5 ng/mL EGF in the bottom chamber. After 22 h at 37°C, the filters were processed as in the migration assay, and the cells were enumerated in three high-power fields (×400) per filter. The experiments were done at least three times.
Anchorage-independent growth in soft agar. 16A5 (105) or MCF10A (5 × 105) transductants were plated per 60-mm dish in 2 mL of 0.3% agarose in DFCI-1 medium ( 20) on top of a bottom layer of 0.5% agarose. The cells were fed every 3 days. Phase-contrast images were obtained under ×40 magnification, and colonies were counted. Each experiment was done in triplicates.
Three-dimensional Matrigel culture and confocal immunofluorescence microscopy. Cells (2.5 × 103) in 0.4 mL of 2% Matrigel in DFCI-1 medium were plated per well of an eight-well chamber slide on top of a polymerized layer of 100% Matrigel, as described ( 22). The cultures were fed every 3 days. Phase-contrast images were obtained under ×100 magnification. For immunofluorescence analyses, the cultures were fixed in 4% paraformaldehyde on day 20, permeabilized with 0.5% Triton X-100 for 5 min, and stained with anti-E-cadherin (1:1,000) or anti-α6 integrin (1:200) primary antibodies followed by Alexa Fluor 488– and Alexa Fluor 594–conjugated secondary antibodies, respectively (Molecular Probes). The slides were mounted with Vectashield mounting medium containing 4′,6-diamidino-2-phenylindole (Vector Laboratories). Images were acquired with a Zeiss LSM510 UV META confocal microscope under ×400 magnification and analyzed with LSM510 META Image Examiner software (version 3.1.0; Zeiss).
Tumorigenicity assays. Six-week-old female athymic nude (nu/nu) mice (Charles River Laboratories) were injected s.c. close to the fourth mammary gland with 107 cells in 0.2 mL of 1:1 Matrigel and PBS, and tumor growth was monitored. Animals were euthanized, and necropsies were done when tumors reached 1 to 1.5 cm diameter, or after 8 months if no tumors formed. Each cell line was tested in at least five animals. All animal-related procedures were in accordance with the Institutional Animal Care and Use Committee guidelines.
Establishment of human MECs co-overexpressing EGFR and c-Src and demonstration of their biochemical cooperation. To model the clinically observed EGFR and c-Src co-overexpression in immortal, nontumorigenic human MECs, we derived polyclonal retroviral transductants of HPV E6/E7–immortalized 16A5 and spontaneously immortal MCF10A cell lines, as such cells have been widely employed to investigate the cell biology of early mammary tumor progression ( 26– 33). Western blot analysis of cell lysates confirmed the expression of endogenous and overexpressed levels of c-Src, EGFR, or ErbB2 as well as comparable overexpression of EGFR and c-Src in single and double transductants ( Fig. 1A and B ). We used the 16A5 transductants for in-depth analyses and the MCF10A transductants to confirm key observations.
To assess if EGFR and c-Src co-overexpressed in nontransformed MECs exhibit biochemical synergy reported in rodent fibroblasts and breast cancer cell lines ( 3), we examined the status of EGF-induced phosphorylation of EGFR in parental 16A5 cells and its transductants ( Fig. 2 ). Although EGF-deprived cells showed expected lack of tyrosine phosphorylation in anti-phosphotyrosine (anti-pY) blots of whole-cell lysates ( Fig. 2B, lanes 1, 4, 7, and 10), EGF stimulation induced substantial phosphorylation of EGFR ( Fig. 2B, lanes 2, 5, 8, and 11). Blotting with phospho-specific antibodies detected tyrosine phosphorylation of Y845, the target of Src kinase activity, and Y1068 an EGFR autophosphorylation site, in parental cells and all transductants ( Fig. 2A and B). Compared with c-Src or EGFR single transductants, EGFR phosphorylation in EGFR plus c-Src double transductants was markedly increased. The EGF-induced Y845 phosphorylation was markedly reduced in the presence of a Src kinase inhibitor PP2 ( 34), used at a concentration (3 μmol/L) that fully inhibited c-Src Y416 autophosphorylation (Supplementary Fig. S1A and B), in parental 16A5 and its transductants, whereas Y1068 phosphorylation was minimally or not affected ( Fig. 2A, compare lanes 3 and 4, lanes 7 and 8, lanes 11 and 12, and lanes 15 and 16). Conversely, when the cells were treated with an EGFR inhibitor AG1478 at 1 μmol/L (see Supplementary Fig. S1C for dose-response), overall EGF-induced phosphorylation of EGFR as well as Y845 and Y1068 phosphorylation were markedly reduced ( Fig. 2B, lanes 3, 6, 9, and 12). The specificity of c-Src–dependent Y845 phosphorylation was further supported by the loss of pY845 signal, but not the pY1068 signal, in 16A5 cells overexpressing EGFR and kinase-inactive K297R c-Src mutant ( Fig. 2B, lanes 13–15) when compared with a co-overexpressed wild-type c-Src ( Fig. 2B, lanes 10–12; for expression levels of K297R c-Src mutant, see Supplementary Fig. S1D). Notably, the increase in EGFR Y1068 phosphorylation seen with c-Src overexpression was not reversed by the c-Src inhibitor or the kinase-inactive c-Src. Together, the analyses of EGFR Y845 phosphorylation using a Src inhibitor and kinase-inactive c-Src mutant clearly showed biochemical cooperation between overexpressed EGFR and c-Src in human MECs.
We observed that phosphorylation of EGFR Y992 and Y1173 was also elevated in EGFR plus c-Src co-overexpressing cells compared with vector-transduced parental cells ( Fig. 2C). Because Y992 and Y1173 are known to bind PLCγ1 ( 35) and Shc, respectively ( 36), we asked if these pathways were altered. Indeed, higher levels of phosphorylated Shc were observed in EGFR plus c-Src co-overexpressing cells ( Fig. 2C), and phosphorylated PLCγ1 levels remained elevated longer in EGFR plus c-Src double transductants ( Fig. 2C). EGFR stimulation is also known to activate the phosphatidylinositol 3-kinase (PI3K)-AKT and STAT pathways ( 37, 38). Immunoblotting revealed elevated and prolonged phosphorylation of AKT, STAT3, and STAT5b in EGFR plus c-Src co-overexpressing cells ( Fig. 2D). Collectively, these findings validated the use of model cell lines generated here to further assess the biological cooperation between c-Src and EGFR in MECs.
c-Src overexpression disrupts acinar structure and induces branching morphogenesis in EGFR-overexpressing MECs grown in three-dimensional Matrigel culture. Three-dimensional culture on matrices, such as Matrigel, can reveal subtle biological alterations associated with early oncogenic transformation of MECs ( 32, 39). Therefore, we used the three-dimensional Matrigel culture, with a base layer of 100% and cells plated on top in 2% Matrigel (overlay Matrigel culture method; ref. 22) to assess the biological synergism between co-overexpressed EGFR and c-Src in human MECs. Phase-contrast microscopy on day 12 showed that 16A5 cells ( Fig. 3A, a ), similar to MCF10A cells ( Fig. 3B, a), formed predominantly regular acinar structures. The 16A5 and MCF10A cells transduced with control vectors (Puro, Hygro, or Puro plus Hygro) were indistinguishable from their respective parental cells (Supplementary Fig. S2). Although c-Src–overexpressing cells showed no discernible difference from control cells ( Fig. 3A and B, b), EGFR overexpression increased the proportion of irregular acini ( Fig. 3A and B, c). In contrast to the relatively mild effect of singly overexpressed EGFR or c-Src, MECs co-overexpressing EGFR and c-Src showed a dramatic shift towards irregular and hypercellullar acini, with a marked increase in the proportion of acini with branching morphogenesis ( Fig. 3A and B, d and C). The glandular epithelial tissues, such as the breast, are comprised of cells that exhibit a characteristic polarity, with apical poles pointing towards the central lumen. This polarity, identifiable by apical and basal membrane markers, is often disrupted in carcinomas ( 27). Hence, to characterize the acinar architecture further and to assess the MEC polarity, we cultured MEC acini for 20 days in three-dimensional and analyzed them for α6 integrin (basal marker) and E-cadherin (basolateral marker; ref. 22) localization using confocal immunofluorescence imaging. The parental 16A5 as well as MCF10A acini were comprised of a layer of polarized epithelial cells surrounding a lumen ( Fig. 4A and B, a and f ), as reported by others for MCF10A and other immortal nontumorigenic MECs ( 32). Notably, the overexpression of c-Src alone did not alter the overall acinar structure or disrupt the polarity ( Fig. 4A and B, b and g). The overexpression of EGFR alone resulted in a higher frequency of larger acini with increased numbers of cells in acini and partially filled lumens; however, the polarity, as determined with E-cadherin and α6 integrin staining, was maintained ( Fig. 4A and B, c and h). Strikingly, EGFR plus c-Src co-overexpressing cells stained aberrantly for α6 integrin and E-cadherin, indicating a dramatic loss of polarity ( Fig. 4A and B, d and i). In comparison, MECs with ErbB2 overexpression showed a predominance of irregular and hyperproliferative acini, as anticipated from earlier studies ( 12), with a substantial proportion of branched acini ( Fig. 3A and B, e; Fig. 4A and B, e and j); notably, whereas ErbB2 overexpression alone induced substantially more branched acini and loss of polarity compared with EGFR overexpression, these changes were more pronounced in EGFR plus c-Src co-overexpressing MECs ( Fig. 4A and B, compare d and i with e and j). Overall, the co-overexpression of c-Src dramatically enhanced the hyperproliferation and promoted the formation of branched structures and loss of polarity in EGFR-overexpressing MECs grown in three-dimensional culture ( Fig. 3C; Fig. 4A and B, d and i). The phenotypes of c-Src, EGFR, or EGFR plus c-Src overexpression were comparable in 16A5 and MCF10A cells, indicating that these are not restricted to a particular cell line.
Culture of the MECs in the presence of PP2 led to a dramatic reversion of the phenotypic abnormalities induced by EGFR, EGFR plus c-Src, or ErbB2 overexpression; this was indicated by the reduced size of acini and a drastic reduction in the proportion of irregular, hyperproliferative, and branched acinar structures ( Fig. 3A and B, f–j and Fig. 3C and D) as well as by the restoration of cell polarity (Supplementary Fig. S3). Thus, Src activity is critical for abnormal acinar structure induced by EGFR, EGFR plus c-Src, or ErbB2 overexpression.
c-Src overexpression enhances the migration of MECs in response to EGF. Src-family kinases play a key role in cell migration ( 40). As abnormally polarized acini of c-Src and EGFR co-overexpressing MECs were often surrounded by apparently migratory cells, we quantified the influence of c-Src on migration specifically triggered upon EGFR activation. Untreated or PP2-treated 16A5 cells or various transductants were subjected to migration assay as described in Materials and Methods. As anticipated, substantial migration was observed in parental 16A5 cells only when EGF was added to the lower chamber ( Fig. 5A , D3 versus D3 + EGF), consistent with the role of EGFR activation to trigger epithelial cell migration ( 41). EGF-induced migration was markedly reduced by PP2, indicative of a critical role of endogenous c-Src in EGF-induced MEC migration. Furthermore, 16A5 cells overexpressing c-Src alone showed a substantial increase in EGF-induced cell migration, and cells co-overexpressing EGFR plus c-Src showed a significantly higher level of migration compared with the c-Src–overexpressing (P = 0.0214) and the EGFR-overexpressing cells (P = 0.0066). As with the parental cells, EGF-induced migration was dramatically reduced by PP2 treatment of c-Src, EGFR, or EGFR plus c-Src–overexpressing MECs ( Fig. 5A). Thus, c-Src and EGFR cooperatively enhance the MEC migration, and c-Src activity is essential for the enhanced cell migration.
EGFR and c-Src co-overexpression renders human MECs invasive. In view of the c-Src–enhanced MEC migration, and the reported induction of invasive characteristics by unregulated Src activity (e.g., by v-Src; ref. 15), we examined if c-Src and EGFR co-overexpression promoted the invasiveness of MECs by analyzing cell migration through Matrigel. As anticipated, parental 16A5 cells exhibited a low level of Matrigel invasion that was inhibited by PP2 ( Fig. 5B). Notably, there was little enhancement of basal invasion upon c-Src overexpression. EGFR-overexpressing cells showed a modest increase in invasion, whereas ErbB2-overexpressing cells showed a somewhat higher level of invasion. Remarkably, the EGFR and c-Src co-overexpressing cells exhibited a dramatic increase in invasion, substantially greater than that for ErbB2-overexpressing cells ( Fig. 5B). The invasive behavior of EGFR plus c-Src–overexpressing as well as the single transductants was inhibited by PP2, consistent with the critical role of Src activity in the enhanced invasion.
EGFR and c-Src co-overexpression promotes the anchorage- independent growth of MECs. As reported with normal MECs and immortal nontumorigenic lines, such as MCF10A or HPV16 E6/E7–immortalized MECs ( 20, 21), 16A5 cells formed few (usually <10 per 105 cells) colonies when grown on soft agar ( Fig. 6A, a ). 16A5 cells overexpressing c-Src were comparable ( Fig. 6A, b), whereas EGFR overexpression led to a small but detectable increase in colony formation ( Fig. 6A, c), and ErbB2 overexpression resulted in a more robust increase ( Fig. 6A, e). Notably, the co-overexpression of EGFR and c-Src substantially increased the soft agar colony-forming ability of 16A5 cells ( Fig. 6A, d) and also led to substantially larger colonies (see insets in Fig. 6A, compare c with d), with many colonies showing loosely attached cells near the periphery of the colony, suggestive of migrating/invasive cells (see higher magnification in Fig. 6A, d). This was in contrast to smaller and compact colonies observed formed by parental and EGFR-overexpressing 16A5 cells, and even ErbB2-overexpressing 16A5 cells formed more compact colonies.
Similar analyses using MCF10A transductants showed that the parental cells essentially lacked anchorage-independent growth; very low level of colony formation was observed in cells overexpressing EGFR or ErbB2 but more readily in cells co-overexpressing EGFR and c-Src ( Fig. 6B and C). Overall, fewer colonies were obtained in MCF10A transductants possibly because they lack a permissive milieu created by the HPV E6 and E7 ( 22). Importantly, colony formation in both cell panels was significantly inhibited by PP2 ( Fig. 6A and B, f–j), establishing its c-Src dependence.
EGFR and c-Src co-overexpression is insufficient to drive in vivo tumor formation. Because the in vitro phenotypic changes induced by the co-overexpression of EGFR and c-Src suggested that these cells were transformed, we injected these cells s.c. near the mammary fat pad of female nude mice to determine whether they could form tumors in immunocompromised mice. Although almost all animals inoculated with transformed MECs, HMLER ( 29) and Ma-6-Ras ( 42), used as positive controls, formed tumors within 8 weeks of inoculation (5 of 5 for HMLER, 4 of 5 for Ma-6-Ras), none of the animals implanted with parental 16A5 or its c-Src, EGFR, or EGFR plus c-Src–overexpressing derivatives developed tumors even after 8 months of observation. Thus, EGFR plus c-Src are insufficient to promote MEC growth as xenotransplant tumors in immunocompromised mice.
Overall, the analyses of three-dimensional acinar structure and polarity together with functional indicators of early oncogenesis show that co-overexpressed EGFR and c-Src biologically cooperate to promote early oncogenic phenotypes in human MECs.
Although ErbB receptors are now well accepted as key players during oncogenesis of mammary and other epithelial tissues, the signaling pathways relevant to epithelial cell oncogenesis remain unclear because nearly all studies have been done in non-epithelial cell models or in metastatic tumor cells. Here, we have modeled, for the first time, the co-overexpression of EGFR and c-Src, clinically observed in human breast cancers, in immortalized nontumorigenic human MECs. Analyses of oncogenic traits using biologically relevant and sensitive three-dimensional culture in Matrigel, together with functional assays of in vitro oncogenesis, showed that c-Src overexpression can markedly enhance the ability of EGFR to promote oncogenic traits in MECs. These traits resemble those observed with the overexpression of ErbB2, generally recognized as a more potent epithelial transforming oncogene. Thus, biological modifiers, such as c-Src, can dramatically exaggerate the output of oncogenic signals downstream of EGFR, and the model presented here should facilitate the identification and characterization of such modifiers.
Controlled dimerization of ectopic ErbB2 but not EGFR was previously shown to reinitiate proliferation of growth-arrested MCF10A acini, whereas comparable growth promoting effects of EGFR or ErbB2 dimerization were seen in two-dimensional culture ( 12). Compared with these findings, we observed a modest degree of hyperproliferation and irregular acini as well as partial luminal filling with constitutive EGFR overexpression ( Fig. 4), likely reflecting the constitutive overexpression of EGFR in our system. However, the effects of EGFR overexpression alone were far less drastic compared with those of ErbB2, which induced irregular and hyperproliferative acini and also induced marked loss of polarity with branching morphogenesis in a substantial proportion of acini. Importantly, the c-Src co-overexpression with EGFR not only increased the proportion of hyperproliferative and irregular acini but also increased the proportion of acini with branching morphogenesis ( Figs. 3 and 4) above the levels seen in ErbB2 transductants. The phenotypic alterations induced by ErbB2 overexpression are also exaggerated by c-Src (data not shown), suggesting a general positive modifier role c-Src in ErbB receptor–mediated mammary oncogenesis. This suggestion is consistent with the activation of Src in ErbB2-driven transgenic mouse mammary tumors as well as Src overexpression in ErbB2-overexpressing breast cancers ( 14, 43). c-Src–like modifiers could play a more critical role in the case of weaker oncogenes, such as EGFR, compared with potent oncogenes such as ErbB2 ( 10, 11).
Similar to ErbB2 overexpression, EGFR plus c-Src induced a loss of cell polarity as shown by mislocalization of basal integrin and basolateral E-cadherin. Importantly, analyses using a c-Src inhibitor support the conclusion that hyperproliferation as well as loss of polarity induced by EGFR plus c-Src overexpression required c-Src activity. E-cadherin is the major adherens junction adhesion protein expressed by epithelial cells and is known to play an important role in the maintenance of cell polarity ( 44). Disruption of E-cadherin–mediated adhesion is associated with progression of noninvasive tumors to an invasive stage ( 45). Although mechanisms by which c-Src might regulate E-cadherin in MECs in three-dimensional remain unclear, analyses in other epithelial cells show that activated Src can enhance the down-regulation of E-cadherin, often in conjunction with EGFR stimuli, through ubiquitin-dependent endocytosis and lysosomal degradation ( 46). However, E-cadherin protein levels were not altered by c-Src or EGFR plus c-Src overexpression in our studies (Supplementary Fig. S5). Consistent with a role of Src in E-cadherin–mediated cell-cell adhesion, Src inhibitor PP2 was shown to promote the aggregation of breast cancer cells in vitro ( 47), whereas c-Src activation in hepatocellular carcinoma cells deregulated E-cadherin function ( 48). Elevated c-Src activity can also cause several components of the adherens junctions to redistribute to integrin-based adhesion complexes ( 15). As endogenous (in ErbB2 or EGFR single transductants) as well as overexpressed c-Src seemed important for ErbB receptor–driven perturbations of MEC polarity, the model system described here should help dissect the cell biological and biochemical pathways linking c-Src to E-cadherin function in a tractable system relevant to human epithelial cancers.
Compared with relatively little phenotypic alteration in three-dimensional acinar structure, c-Src overexpression alone induced a substantial increase in Transwell migration response to EGF exceeding that induced by EGFR overexpression ( Fig. 5). The effects of EGFR and c-Src co-overexpression were not dramatic in this assay as we observed at best an additive effect. However, the more stringent Matrigel invasion assay showed that EGFR plus c-Src overexpression dramatically promoted the invasive behavior of MECs ( Fig. 5). Invasion through the basement membrane requires matrix degrading activities in addition to enhanced migration ( 15), suggesting that cooperative signaling by c-Src and EGFR may be critical to trigger the release of appropriate proteolytic enzymes at the cell-matrix interface. This idea is consistent with the induction of matrix-degrading metalloproteases upon overexpression of v-Src in many cell systems ( 15). The model described here should help explore the mechanistic basis for c-Src–dependent invasive traits of human MECs.
Biological cooperativity between co-overexpressed EGFR and c-Src (Supplementary Fig. S4) was further evident in the ability of doubly transduced MECs to undergo anchorage-independent growth, a trait commonly used to gauge the oncogenicity of cells in vitro. Clear increase in soft agar colony formation was seen with EGFR plus c-Src overexpression ( Fig. 6), and as with three-dimensional acinar abnormalities, colony formation was inhibited by the Src inhibitor PP2, indicating its Src dependence. Finally, in vivo xenotransplant studies indicate that EGFR plus c-Src co-overexpression does not result in full tumorigenic transformation of MECs, unlike the rodent fibroblast system ( 16). This result is reminiscent of the inability of other rodent fibroblast-transforming oncogenes, such as activated Ras, to fully transform immortalized human MECs ( 29). Future analyses using the present system should identify cancer-relevant genes whose perturbations together with EGFR overexpression promote full oncogenesis of human MECs.
Notably, similar to the rodent fibroblast system as well as c-Src–overexpressing epithelial tumor cell lines, c-Src overexpression in nontumorigenic human MECs enhanced the phosphorylation of EGFR on Y845, a highly conserved activation loop tyrosine that is uniquely phosphorylated by Src ( 3). The c-Src dependence of Y845 phosphorylation in MECs was supported by its abrogation upon c-Src inhibitor treatment and expression of a kinase-inactive c-Src ( Fig. 2A and B). In contrast, the enhanced phosphorylation on Y1068 of EGFR was not reversed by the Src inhibitor and was seen upon co-overexpression of wild-type as well as kinase-inactive c-Src ( Fig. 2B). This kinase-independent enhancement of EGFR Y1068 is likely to be due to general protection of phosphorylation sites on EGFR by the bound SH2 domain of c-Src or its mutant; consistent with this suggestion, the effect of c-Src overexpression was lost when EGFR kinase activity was inhibited ( Fig. 2A).
Recent studies have linked EGFR pY845 to STAT activation and cell proliferation as well as to potential regulation of apoptosis through pY845-mediated interaction with mitochondrial cytochrome oxidase 2 ( 49). Our initial analyses are consistent with enhanced STAT activation in EGFR plus c-Src co-overexpressing MECs, which also exhibit increased PI3K and Shc pathway activation ( Fig. 2C and D). Given the recently shown roles of STAT proteins and PI3K mutations in mammary development and breast cancer ( 37, 38, 50), further studies using the present model should provide a more relevant system to dissect these oncogenic signaling pathways compared with traditional fibroblast and other model cellular systems. The simple assay systems that revealed the cooperative oncogenic progression in human MECs co-overexpressing EGFR and c-Src should be adaptable for high-throughput screening to identify additional modifiers of EGFR-mediated oncogenesis, screen for chemical inhibitors, and dissect oncogenic ErbB receptor signaling using ectopic gene expression or knockdown strategies.
Grant support: NIH grants CA 87986, CA 76118, CA 99900, CA99163 (H. Band), CA94143, CA96844, and CA81076 (V. Band); Department of Defense Breast Cancer Research grants DAMD17-02-1-0303 (H. Band) and DAMD17-02-1-0508 (V. Band); National Cancer Institute Cancer Center of Nanotechnology Excellence grant NCI 1U54 CA119341-01 (H. Band and V. Band); Jean Ruggles Romoser Chair of Cancer Research (H. Band); Duckworth Family Chair of Breast Cancer Research (V. Band); ENH Research Career Development Award (M. Naramura); and a grant from the Elsa U. Pardee Foundation (M. Naramura).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Dr. Senthil Muthuswamy (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY) for the MCF10A cells, Drs. Jonathan Higgins and Michael Brenner (Brigham and Women's Hospital, Boston, MA) for the anti-E-cadherin hybridoma, Dr. Brian Druker (Oregon Health Sciences University, Portland, OR) for 4G10 antibody, Dr. Joan Brugge (Harvard Medical School, Boston, MA) for c-Src cDNA, Drs. Valerie Weaver and Senthil Muthuswamy for sharing three-dimensional Matrigel protocols, and members of the Band laboratories for helpful suggestions and discussion.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received July 12, 2006.
- Revision received February 11, 2007.
- Accepted February 26, 2007.
- ©2007 American Association for Cancer Research.