Activation of BAD by Therapeutic Inhibition of Epidermal Growth Factor Receptor and Transactivation by Insulin-like Growth Factor Receptor*

Novel cancer chemotherapeutics are required to induce apoptosis by activating pro-apoptotic proteins. Both epidermal growth factor (EGF) and insulin-like growth factor (IGF) provide potent survival stimuli in many epithelia, and activation of their receptors is commonly observed in solid human tumors. Here we demonstrate that blockade of the EGF receptor by a new drug in phase III clinical trails for cancer, ZD1839, potently induces apoptosis in mammary epithelial cell lines and primary cultures, as well as in a primary pleural effusion from a breast cancer patient. We identified the mechanism of apoptosis induction by ZD1839. We showed that it prevents cell survival by activating the pro-apoptotic protein BAD. Moreover, we demonstrate that IGF transactivates the EGF receptor and that ZD1839 blocks IGF-mediated phosphorylation of MAPK and BAD. Many cancer therapies kill tumor cells by inducing apoptosis as a consequence of targeting DNA; however, the threshold at which apoptosis can be triggered through DNA damage is often different from that in normal cells. Our results indicate that by targeting a growth factor-mediated survival signaling pathway, BAD phosphorylation can be manipulated therapeutically to induce apoptosis.

Deregulated growth in cancer is not solely due to an increase in proliferation but is also due to an alteration in the balance between proliferation and apoptosis (1). EGFR 1 and IGF-IR can both provide powerful mitogenic and anti-apoptotic signals in a wide variety of cell types but activate distinct downstream signaling pathways. The EGFR family consists of four members, EGFR (ErbB1, HER1), ErbB2 (Neu, HER2), ErbB3 (HER3), and ErbB4 (HER4) (2). Activation of the receptor by ligand binding results in phosphorylation of tyrosine residues within the cytoplasmic domain, providing docking sites for a number of SH2 domain containing signaling proteins, such as the Grb2/mSOS complex, and phosphatidylinositol-3Ј kinase (PI3K) (2). Ligand-bound IGF-IR interacts with specific adaptor molecules, insulin receptor substrates (IRS) 1 and 2, which in turn become phosphorylated on tyrosine residues and recruit SH2 domain-containing signaling molecules (3,4).
A number of growth factor and cytokine receptors have been shown to suppress the pro-apoptotic function of BAD, a BH3 only member of the Bcl-2 family of apoptosis regulators (5)(6)(7)(8)(9)(10). Bcl-2 family proteins share conserved BH (for Bcl-2 homology) domains (11). Multi-BH domain proteins (e.g. Bax and Bak) have the capacity to regulate pore formation in the outer membrane of the mitochondrion and thereby regulate the release of factors such as cytochrome c (12). The BH3 only proteins (e.g. Noxa, Bid, and BAD) appear to act upstream of multi-domain proteins and are regulated by a variety of signals including p53 (Noxa), Fas ligand (Bid), and withdrawal of survival factors (BAD) (13)(14)(15)(16). Signaling pathways initiated by a variety of soluble growth factors, and cytokines inhibit apoptosis by phosphorylating BAD on multiple serine residues (Ser 112 , Ser 136 , and Ser 155 ) (10,16,17). Activation of receptors that regulate PI3K result in phosphorylation of BAD on serine 136 (18), whereas cytokines that activate the MAPK pathway phosphorylate BAD on serines 112 and 155 (8). Phosphorylation of these sites inactivates BAD by sequestering it in the cytosol and inhibiting its interaction with Bcl-X L (10,16,19). The central position of BAD between multiple growth factor signaling pathways and apoptosis raises the possibility of its potential as a therapeutic target.
Most therapies for solid tumors utilize DNA-damaging agents, inducing apoptosis of cancer cells and thereby reducing tumor mass (20,21). Tumor suppressor genes, such as p53 and ATM, encode molecules that integrate DNA damage with apoptosis (22,23). Inactivation of these genes are frequently found in cancer, thus decreasing the sensitivity of tumor cells to chemotherapy and radiotherapy and limiting the usefulness of these therapeutic approaches (24). Despite this, cancer cells still maintain the potential to undergo apoptosis, and scope therefore exists for therapies that induce cell death via different pathways (25)(26)(27). In this context, growth factor receptors, such EGFR, are potential drug targets (28). The EGFR family, and EGFR and Erb2 in particular, have been strongly implicated in tumor progression, including tumors of breast, lung, head, neck, bladder, and ovary, where receptor overexpression is associated with advanced disease and poor prognosis (29,30).
Herceptin, a recently developed humanized anti-ErbB2 antibody inhibits ErbB2 activity and cell proliferation and is effective in treating a subset of breast cancers that overexpress ErbB2 (31). Other growth factor receptors may prove to be effective therapeutic targets, and a number of drugs are now undergoing development (32,33).
In this paper we have examined whether inhibitors of growth factor receptors can be used to induce activation of pro-apoptotic Bcl-2 family proteins. ZD1839 is a novel, low molecular mass, 4-anilinoquinazoline derivative that is a powerful and selective inhibitor of EGFR tyrosine kinase activity (IC 50 ϭ 0.08 M in vitro) which blocks cell proliferation in response to EGF (34). It is currently undergoing phase III clinical trials for non-small cell lung cancer (35). In vitro and in vivo studies have shown ZD1839 to inhibit the mitogenic effect of EGFR activation or overexpression (36). In this paper we demonstrate that ZD1839 potently induces apoptosis in normal and malignant breast epithelial cells. We show that ZD1839 inhibits mammary epithelial cell survival through its effects on EGFRdependent MAPK signaling and BAD phosphorylation. Furthermore, we find that IGF-1 induces phosphorylation of BAD and epithelial cell survival and that this occurs through both PI3K-and MAPK-dependent pathways. By using ZD1839, we demonstrate that in mammary epithelial cells, IGF-IR-mediated activation of MAPK occurs through transactivation of the EGFR.
Our results suggest that regulation of BAD and apoptosis in response to diverse growth factors is dependent upon EGFR activation and that this can be manipulated therapeutically. Moreover, our results suggest that by targeting pathways independent of DNA damage, ZD1839 may have significant potential as an anticancer agent in tumors that are resistant to conventional therapies.

EXPERIMENTAL PROCEDURES
Cell Culture and Inhibitors-Primary mouse mammary epithelial cells, isolated from pregnant ICR mice, BAD Ϫ/Ϫ mice, or wild type littermates (37), and FSK-7 cells (38), were grown in Dulbecco's modified Eagle's medium/Ham's F-12 supplemented with 2% fetal calf serum, 5 ng/ml EGF, and 880 nM insulin. Freshly isolated pleural effusion cells were obtained with consent from a 62 year-old female patient with an estrogen receptor-positive primary invasive ductal carcinoma. The cells were kindly donated by Prof. Charles Coombes (Department of Cancer Medicine, Hammersmith Hospital, London, UK). The patient was lymph node-positive at primary surgery and had relapsed with metastatic disease at multiple sites within 4 years, after successive rounds of anti-estrogen therapy (tamoxifen and aromatase inhibitors), radiotherapy, and chemotherapy, including taxane and vinca alkaloid treatment. The malignant cells were separated from the reactive mesothelial cells in the effusion by a 2-h attachment in tissue culture medium, which removed the latter as a result of their rapid adhesion to the substratum. The purified malignant breast carcinoma cells were then assayed for ZD1839 sensitivity. For all of the growth factor experiments, the cells were starved of serum and growth factors for 4 h before stimulating with either EGF (5 ng/ml) or IGF (10 m) for 15 min. PP2, AG1478, LY-294002, PD-98059 (all from Calbiochem), and ZD1839 were added to cells 30 min prior to growth factor stimulation. To quantify apoptosis, both the detached cells and the remaining adherent cells were harvested by trypsinization and pooled. The cells were cytospun onto polysine slides (Merck) and fixed in 2% paraformaldehyde. The nuclear morphology was examined after staining cells with 4 g/ml Hoescht 33258 (Molecular Probes).
Protein Extraction and Immunoblotting-The cells were lysed in radioimmune precipitation buffer (50 mM Tris⅐Cl, pH 7.6, 150 mM NaCl, 1% Nonidet P-40, 1% sodium deoxycholate, 0.1% SDS, 2 mM EDTA, 10 mM NaF, 1 mM Na 3 VO 4 , protease inhibitors), and the insoluble material was cleared by centrifugation. The samples were normalized for protein content and either immunoprecipitated using protein A-Sepharose or separated by SDS-PAGE. The samples were transferred to nitrocellulose for immunoblotting with antibodies from the following sources: caspase 3 and MAPK (Santa Cruz); EGFR and phosphotyrosine (PY-20, BD Transduction Laboratories); phospho-MAPK and Ser(P) 112 and Ser(P) 136 BAD (New England Biolabs); BAD (R and D systems); and IRS-1 and p85 PI3K (Upstate Biotechnology, Inc.).
Transfections-The cells were transfected with up to 1 g of plasmid DNA using LipofectAMINE Plus reagent according to the manufacturers instructions (Invitrogen). pIRES-hrGFP and pEBG-BAD were from Stratagene and New England Biolabs, respectively. Other plasmids were generously donated by the following people: pEH-Hm-Raf and pCDNA3-p110CAAX were from Julian Downward (ICRF, London, UK), pEGFP-Bax was from Richard Youle (National Institutes of Health, Bethesda, MD), and pMT2-myrRsk-1 was from John Blenis (Harvard Medical School, Boston, MA).
Determination of BAD Phosphorylation State-FSK-7 cells were transfected with 1 g of pEBG.BAD and grown for 18 h post-transfection. The cells were starved for 4 h before stimulating with either EGF or IGF for 15 min. Inhibitors were added to cells 30 min prior to stimulation with growth factor. The cells were lysed by scraping into radioimmune precipitation buffer, followed by centrifugation. GST-BAD was precipitated on glutathione-agarose (Sigma) for 1 h at 4°C and separated by SDS-PAGE before immunoblotting using antibodies specific for Ser(P) 112 and Ser(P) 136 BAD.

Blockade of EGF Receptor Activation Induces Apoptosis of Breast Epithelial Cells and Primary Cultures of Breast
Carcinoma Cells-IGFs have previously been shown to be survival factors for mammary epithelial cells (39). We tested whether EGF also regulates survival in primary cultures of mammary epithelial cells. High levels of spontaneous apoptosis occur in primary cells plated on tissue culture plastic or collagen I (40), but EGF, as well as fetal calf serum and IGF, suppresses apoptosis (Fig. 1a). To confirm a survival role for EGF, we treated primary mammary cells with ZD1839 and observed a time-dependent suppression of EGF-mediated survival (Fig.  1a). The data were similar in a mammary epithelial cell line, FSK-7, isolated from luminal epithelial cells (38), although the cell line exhibits lower spontaneous apoptosis than primary cells both in the absence of growth factors or with fetal calf serum, EGF, or IGF (Fig. 1b). FSK-7 cells also underwent apoptosis in a dose-dependent manner when treated with varying concentrations of ZD1839 (Fig. 1c). Together, these data indicate that ZD1839 blocks EGF-and IGF-mediated survival of mammary epithelial cells.
ZD1839 is in phase III clinical trials for lung cancer (35), but it is not yet known whether it will be effective for other types of human cancer. Our results with primary cultures of mouse mammary cells and with mammary epithelial cell lines suggest that ZD1839 might also be a valuable therapy for breast cancer. We therefore assessed whether the drug induced apoptosis in a primary culture isolate of a pleural effusion from a malignant breast tumor. These cells were from a patient whose tumor had failed radiotherapy and chemotherapy, and they show low spontaneous apoptosis when cultured without growth factors (not shown) or with EGF (Fig. 1d). ZD1839 induces a dramatic apoptotic response in these cells. Thus, our results demonstrate that ZD1839 is successful in inducing apoptosis of primary breast carcinoma cells.
EGF Promotes Survival by Phosphorylating BAD-EGF induces cytoplasmic signal transduction through the MAPK pathway (2). One target of the MAPK pathway is the proapoptotic protein, BAD. Phosphorylation of BAD inhibits BADinduced apoptosis (7). In mammary epithelial cells, EGF potently activates MAPK phosphorylation, whereas ZD1839 inhibits the phosphorylation of both MAPK and EGFR (Fig.  2a). In the same cultures EGF modulates the phosphorylation status of endogenous BAD. BAD is visible by SDS-PAGE as a discrete 23-kDa protein, but additional more slowly migrating isoforms can be detected that represent the phosphorylated form of the protein (16). These slower migrating phosphorylated isoforms of BAD can be detected in the presence of EGF, both in FSK-7 cells (Fig. 2a) and in primary cultures of mam-mary cells (Fig. 2b). ZD1839 inhibits the appearance of the additional forms of BAD in response to EGF (Fig. 2, a and b).
To test for the specificity of ZD1839 inhibition of EGFR signaling, FSK-7 cells were treated with increasing concentrations of ZD1839 before stimulation with EGF. Inhibition of EGF-dependent activation of MAPK is consistent with the IC 50 of ZD1839 for EGFR (0.08 M). 1 M ZD1839 completely inhibits MAPK and BAD phosphorylation (Fig. 2c). FSK-7 cells were treated with AG1478, another well characterized EGFR antagonist (41). AG1478 also inhibits EGF-dependent MAPK and BAD phosphorylation in a dose-dependent manner (Fig. 2d). In contrast to these results, PP2, a tyrosine kinase inhibitor that does not reportedly affect EGFR, does not block EGF-mediated phosphorylation of MAPK or BAD (Fig. 2e). Thus, ZD1839 prevents EGF-dependent MAPK activation and BAD phosphorylation, correlating with its inhibition of mammary cell survival (Fig. 1).
One of the EGFR-regulated forms of BAD is phosphorylated on Ser 112 (BAD-S112), and this can be detected using a phospho-specific antibody after transfecting cells with a cytomegalovirus promoter-driven vector expressing GST-BAD (Fig. 2f). BAD-S112 is known to be phosphorylated downstream of MAPK (7,42). In mammary cells, the phosphorylation of this residue is inhibited by either ZD1839 or the MEK inhibitor PD98059 (Fig. 2f). This suggests that BAD phosphorylation by EGF occurs through the MAPK pathway. To confirm this, the cells were transfected with vectors encoding constitutively active Raf or p90 Rsk-1 together with GST-BAD. Raf is upstream of MAPK, whereas p90 Rsk-1 is downstream and has been shown to phosphorylate BAD (8,43). Both of these proteins bypass the inhibition of EGF-mediated BAD-S112 phosphorylation that occurs with ZD1839 (Fig. 2g). These results demonstrate that EGF induces BAD phosphorylation through a MAPK pathway involving p90 Rsk-1 .
To determine whether BAD is directly involved in the apoptotic response to ZD1839, we overexpressed BAD in mammary epithelial cells and asked whether this sensitized cells to apoptosis induced by ZD1839. In cells overexpressing BAD, the endogenous factors that suppress its pro-apoptotic function may be sequestered, and apoptosis ensues even in the presence of survival factors such as EGF (Fig. 3a). These cells are therefore sensitized for apoptosis, and significantly more death occurs after blocking EGFR signaling by ZD1839 (p ϭ 0.05). In contrast, ZD1839 does not potentiate apoptosis in cells overexpressing another pro-apoptotic protein, Bax. Furthermore in cells sensitized by BAD overexpression, apoptosis (in the absence of ZD1839) is enhanced by the removal of EGF (Fig. 3b). Finally, we assessed the sensitivity of BAD Ϫ/Ϫ mammary epithelial cells to ZD1839. Mammary epithelial cells isolated from BAD Ϫ/Ϫ or wild type mice were cultured in the presence of EGF with or without ZD1839. Whereas apoptosis increased in wild type cells following treatment, the BAD Ϫ/Ϫ cells were less sensitive, showing no increase in apoptosis over controls without drug at 5 h (Fig. 3c). Interestingly, we also observed that the rate of spontaneous apoptosis was higher in the wild type cells compared with the BAD Ϫ/Ϫ cells (data not shown). Thus, BAD is a mediator of apoptosis that becomes activated after blocking EGF signaling, either by removing ligand or by inhibiting EGFR with ZD1839. Together these data show that ZD1839 is a potent inhibitor of mammary epithelial cell survival through its effects on MAPK signaling and BAD activation.
IGF-I Signaling Promotes Survival Partly through Transactivating EGF Receptor-IGF-I activates a separate pathway leading to BAD phosphorylation. This is mediated by PI3K and results in phosphorylation of BAD on a different serine residue, Ser 136 (44,45). In mammary epithelial cells, IGF activates the IGF-IR, thus phosphorylating IRS-1 and causing it to associate with the regulatory p85 subunit of PI3K (46). ZD1839 does not inhibit the kinase activity of IGF-IR receptor in vitro (47) or block IRS-1 tyrosine phosphorylation and PI3K association in mammary cells (Fig. 4a). Furthermore, ZD1839 does not prevent BAD-S136 phosphorylation in response to either IGF-1 (Fig. 4b) or the constitutively active p110 subunit of PI3K (Fig.  4c). In contrast, EGF fails to activate the IRS-1 pathway and BAD phosphorylation on residue Ser 136 (Fig. 4, a and b). These results point to the specificity of ZD1839 in inhibiting EGFR but not IGF-IR activation.
We noted above that ZD1839 does partially reverse the IGF suppression of apoptosis in primary mammary epithelial cultures (Fig. 1a). This suggests that IGFs regulate mammary cell survival through an additional, IRS-1-independent pathway. We therefore examined the possibility that IGF-1 might contribute to survival through a MAPK pathway and phosphorylation of BAD-S112 (Fig. 4, b, d, and e). IGF-I has been shown to activate EGFR signaling indirectly through heparin-binding EGF (48), suggesting that the MAPK pathway might also be activated through cross-talk with EGFR in mammary cells. Indeed, IGF-1 induces phosphorylation of EGFR (Fig. 4e), and ZD1839 inhibits both this as well as the phosphorylation of MAPK and BAD-S112 (Fig. 4, b,  d, and e). We confirmed that IGF mediates BAD-S112 phosphorylation independently of PI3K, because expression of the constitutively active p110 subunit of PI3K does not prevent inhibition of BAD-S112 phosphorylation by ZD1839 (Fig. 4d, left panel). Furthermore, inhibition of PI3K with LY294002 fails to block IGF-induced phosphorylation of BAD-S112, whereas PD98059 does (Fig. 4d, right panel). Similarly, IGF activation of MAPK FIG. 2. ZD1839 prevents EGF mediated BAD phosphorylation through a MAPK pathway involving p90 Rsk-1 . a and b FSK-7 cells (a) or primary mammary epithelial cells (b) were cultured in the absence of growth factors for 4 h. The cells were then either left untreated or treated with 1 M ZD1839 for 30 min before stimulating with EGF for 15 min. In a, FSK-7 cell lysates were analyzed by immunoblotting for phosphorylation of EGFR, MAPK, and BAD. Note the higher molecular mass bands of phosphorylated endogenous BAD in the presence of EGF but not ZD1839 (arrows). In b, primary mouse mammary epithelial cell lysates were analyzed for BAD phosphorylation. c, serum-starved FSK-7 cells were treated for 30 min with the indicated concentrations of ZD1839, before stimulating with EGF for 15 min. Cell lysates were analyzed by immunoblotting for phospho-MAPK, total MAPK, and BAD. Note that both MAPK and BAD phosphorylation were abrogated by 1 M ZD1839 and partially reduced by 0.5 M. d, serum-starved FSK-7 cells were treated with either 1 M ZD1839 or AG1478 at the indicated concentrations before stimulating with EGF. The lysates were immunoblotted as in c. e, serum-starved FSK-7 cells were treated with either 1 M ZD1839 or PP2 at the indicated concentrations before stimulating with EGF. The lysates were immunoblotted as in c. f, FSK-7 cells transfected with pEBG-BAD were treated with either no growth factors (Ϫ) or EGF with or without preincubation with 1 M ZD1839 or 10 M MEK inhibitor PD98059. The cells were lysed, and GST-BAD was precipitated on glutathione-agarose, separated by SDS-PAGE, and analyzed by immunoblotting with antibodies to Ser(P) 112 and total BAD. g, cells were transfected with pEBG-BAD, together with pCDNA3, pEF-Hm-Raf, or pMT2-myrRsk-1. Phosphorylated GST-BAD was analyzed as in f. was inhibited by ZD1839 and PD98059 but not LY29402 (Fig. 4f). These data indicate that IGF signaling occurs partially through EGFR and that this is independent of IGF-IR-mediated activation of the PI3K pathway.
Our results demonstrate that both the MAPK and PI3K signaling pathways lead to phosphorylation of BAD in mammary epithelial cells. To determine which of these pathways is critical for EGF-mediated survival, we transfected mammary epithelial cells with vectors encoding activated signaling enzymes and then assessed cell survival in the presence or absence of ZD1839. In cells transfected with only the control vector, pLacZ, ZD1839 induces significant apoptosis (Fig. 5). Activation of both Raf and p90 Rsk-1 bypasses ZD1839-mediated apoptosis (Fig. 5) and BAD-S112 phosphorylation (Fig 2g), confirming the role of MAPK signaling in EGF-dependent survival. PI3K also has a role in survival that is independent of EGFR signaling, because the constitutively active p110 subunit of PI3K circumvents the apoptotic effect of ZD1839 (Fig. 5) and maintains phosphorylation of BAD-S136 in the presence of ZD1839 (Fig. 4c).
Thus, EGF signaling results in BAD-S112 phosphorylation through MAPK, whereas IGF activates PI3K leading to BAD-S136 phosphorylation, and both of these pathways contribute to mammary cell survival (Fig. 6). Furthermore, IGF can additionally transactivate EGFR resulting in phosphorylation of BAD-S112, and this can be inhibited by ZD1839. Through this mechanism ZD1839 may have the potential to target survival mediated by a wide range of ligands whose signaling converges on the EGFR (48 -52). DISCUSSION Activation of the EGFR provides a potent survival signal in many cell types. In this paper we have shown that ZD1839, a small molecular mass inhibitor of the EGFR that is currently in clinical trials for lung cancer, potently induces apoptosis in both normal mammary epithelial cells and primary cultures of mammary carcinoma cells (Fig. 1). ZD1839 induced apoptosis by inhibiting MAPK signaling downstream of the EGFR, resulting in dephosphorylation of BAD on serine 112 (Fig. 2). In parallel experiments we found that ZD1839 also caused dephosphorylation of BAD on serine 155. 2 Inhibition of BAD phosphorylation and subsequent apoptosis could be circumvented by the expression of constitutively active Rsk-1, a kinase downstream of MAPK (Figs. 2g and 5). Other studies have shown BAD to be downstream of growth factor receptors that activate the MAPK pathway and have implicated Rsk-1 as a BAD kinase (8,53). Our work now suggests that phosphorylation of BAD on serine 112 is downstream of EGF signaling in mammary epithelial cells (Fig. 2), and our data indicate the involvement of Raf, MEK, and p90Rsk-1 as signaling intermediates (Figs. 2, f and g, and 5), as shown in Fig. 6a.
IGF-IR activation also regulates apoptosis, and this has been proposed to occur primarily through PI-3 kinase and protein kinase B (3,54,55). We found that IGF-1 promotes survival of mammary epithelial cells (Fig. 1, a and b), and phosphorylation of BAD on serine 136 (Fig. 4, b and c), the residue indirectly targeted by PI3K (Fig. 6b). However, IGF-I was not sufficient to maintain survival in the presence of ZD1839, even though BAD-S136 phosphorylation was independent of EGFR ( Fig. 4b) This suggests that a significant component of IGF-IR-mediated survival signaling in epithelial cells occurs through the transactivation of the EGFR (Fig. 4e), resulting in the phosphorylation of MAPK (Fig. 4, e and f) and BAD on serine 112 (Fig. 4d). This transactivation of EGFR by IGF-1 is shown in Fig. 6.
Survival Signaling by IGF Can Be Mediated through Crosstalk with EGFR-The IGF-IR has been shown to protect a wide variety of cells from apoptosis (18,56,57). A number of pathways have been described by which IGF-IR activates survival pathways. The main pathway proposed to protect cells is through the interaction with IRS-1, one of its major substrates (54). IRS-1, bound to and phosphorylated by the receptor, interacts with the p85 regulatory subunit of PI3K (4). PI3K subsequently activates protein kinase B, which can phosphorylate BAD-S136. The IGF-IR is also known to activate the MAPK pathway through an interaction with Shc, one of its substrates (18,58,59). This can also contribute to cell survival through BAD phosphorylation (18). A third pathway has been FIG. 3. ZD1839 induces apoptosis through BAD. a, FSK-7 cells transfected with pEBG-BAD, pEGFP-Bax, or control vector, pCDNA3, were grown in the presence of EGF for 24 h, with or without 1 M ZD1839, and the percentage of apoptotic transfected cells was determined. p values represent the significance between pairs with or without ZD1839. ZD1839 increased apoptosis in BAD-transfected cells (p ϭ 0.05, Mann-Witney U test) but not following transfection with Bax (p ϭ 0.27). b, transfected FSK-7 cells were maintained with or without EGF for 24 h, and the percentage of transfected apoptotic cells was determined. Transfection of pEBG-BAD sensitized the cells to EGF withdrawal (p ϭ 0.05). c, primary cultures of mammary epithelial cells isolated from BAD Ϫ/Ϫ mice or wild type (WT) litter mates were cultured in the presence of EGF and 1 M ZD1839 for 5 h, after which apoptosis was quantified and compared with parallel control cultures grown in the presence of EGF alone. The data are expressed as the levels of apoptosis relative to EGF alone controls.
proposed to involve the translocation of Raf to mitochondria, involving Nedd-4 (56).
We found that IGF-1 resulted in phosphorylation of BAD on Ser 112 and Ser 136 , with activation of both the PI3K and MAPK pathways. The phosphorylation of BAD-S136 was dependent upon PI3K, which became associated with IRS-1 upon IGF-1 stimulation, because BAD-S136 phosphorylation was inhibited by LY294002. 2 Neither the binding of PI3K to IRS-1 (Fig. 4a) nor the phosphorylation of BAD-S136 (Fig. 4, b and c) was affected by ZD1839, showing it to have no direct effect on the activity of the IGF-IR. This is consistent with data on the specificity of ZD1839 for EGFR (34). In contrast to these results, the phosphorylation of BAD-S112 in response to IGF-I occurred in a MAPK-dependent manner, and both MAPK phosphorylation (Fig. 4f) and BAD-S112 phosphorylation (Fig. 4d) were inhibited by the MEK inhibitor PD98059. However, the phosphorylation of BAD-S112 was also inhibited by ZD1839 (Fig. 4, b and d), indicating dependence on EGFR activation. Therefore, although IGF-1 mediates survival signaling through activation of PI3K and phosphorylation of BAD-S136, some of the anti-apoptotic effects of IGF-IR occur through the transactivation of the EGFR (Fig. 6c). Indeed, as ZD1839 exhibited potent apoptotic effects in the presence of IGF-1 (Fig. 1a), this transactivation of the EGFR must provide a major mechanism for IGF-1-mediated survival in mammary cells.
It is important to note that BAD-S136 remained phosphorylated downstream of IGF-1 in the presence of ZD1839 (Fig. 4b), yet the cells underwent apoptosis. It may be that the apoptotic activity of BAD is only fully suppressed when it is phosphorylated on more than one site. Serines 112 and 136 both interact with 14-3-3 when phosphorylated, sequestering BAD in the cytosol (16). Serine 155 is within the binding site for Bcl-X L , and its phosphorylation inhibits this interaction (6,10). Binding to Bcl-X L is critical for the pro-apoptotic function of BAD. Evidence has indeed been presented that cooperative phosphorylation at multiple sites may be required to fully suppress the apoptotic function of BAD (19). This may provide an explanation for why BAD phosphorylated on just serine 136 (i.e. in the presence of IGF as well as ZD1839) is unable to completely suppress apoptosis in mammary cells (Fig. 1a). Although IGF transactivates EGFR, we found no evidence that the reverse occurs, because EGF does not induce phosphorylation of either FIG. 4. IGF signaling is a target for ZD1839 through cross-talk with EGFR. a, ZD1839 does not inhibit IGF-IR activation of IRS-1 and PI3K. Serum-starved FSK-7 cells were stimulated with either no growth factors (Ϫ), EGF, or IGF, with or without preincubation with 1 M ZD1839. IRS-1 was immunoprecipitated from lysates, separated by SDS-PAGE, and immunoblotted with antibodies to IRS-1, phosphotyrosine (PY-20), or the p85 subunit of PI3K. b, ZD1839 inhibits phosphorylation of BAD on Ser 112 but not Ser 136 . FSK-7 cells transfected with pEBG-BAD were stimulated with EGF or IGF or left without growth factor, and half of the cultures were treated with 1 M ZD1839. Phosphorylated GST-BAD was analyzed as in Fig. 2c. c, IGF-I-mediated phosphorylation of BAD-S136 is via PI3K. FSK-7 cells were co-transfected with pEBG-BAD and either pCDNA3 or pCDNA-p110CAAX. 18 h post-transfection, the cells were starved and treated with IGF and 1 M ZD1839, and phosphorylated GST-BAD was analyzed as in Fig. 2c. d, IGF-I-mediated phosphorylation of BAD-S112 does not require activation of PI3K. FSK-7 cells were transfected with either pEBG-BAD alone (right panel) or co-transfected with pEBG-BAD and pCDNA-p110CAAX (left panel). The cells were starved, treated with the indicated inhibitors, and stimulated with IGF-I, before being lysed for analysis of GST-BAD phosphorylation. e, ZD1839 inhibits IGF-mediated activation of MAPK. Serum-starved FSK-7 cells were stimulated with EGF or IGF and treated with or without 1 M ZD1839. EGFR was immunoprecipitated from cell lysates and immunoblotted with PY-20 anti-phosphotyrosine. Total cell lysates were separated by SDS-PAGE and immunoblotted for phospho-MAPK and total MAPK. f, serumstarved FSK-7 cells were pretreated with either ZD1839, LY294002, or PD98059 as indicated. The cells were then treated with EGF or IGF for 15 min. The cell lysates were immunoblotted with antibodies against total MAPK or phospho-MAPK. IRS-1 (Fig. 4a) or BAD-S136 (Fig. 4b). We therefore speculate that for EGF-mediated survival, phosphorylation of BAD on serine 112 and serine 155 may be sufficient without the additional need for serine 136 phosphorylation.
Transactivation of EGFR has been seen in response to a variety of receptors, including G protein-coupled receptors, Ecadherin, and integrins (48 -50, 52, 60). The mechanism of transactivation varies, although in all cases it is rapid. In the case of IGF in COS-7 cells and G protein-coupled receptor ligands in Rat-1 cells, metalloproteinase-dependent release of heparin-binding EGF is required (48,50). In mammary epithelial cells, transactivation of endogenous EGFR may occur through a different mechanism, because it is not blocked with the broad-spectrum metalloproteinase inhibitor, Illomastat. 2 In the case of E-cadherin-mediated activation, EGFR is recruited to sites of cell adhesion where it becomes activated (60). Similarly, EGFR is recruited into a large multireceptor complex in response to ␤ 2 -adrenergic receptor activation (52). As of yet, we have not determined the mechanism for transactivation in mammary epithelial cells.
Therapeutic Induction of Apoptosis by Inhibition of EGFR Signaling-Several types of regulatory pathway can contribute to the cellular decision to enter apoptosis. These include DNA damage sensing signals, dominant death signals from "death receptors" such as Fas or tumor necrosis factor receptor, and survival signaling pathways from growth factor receptor-activated kinase cascades that block apoptosis and are intrinsically required for normal cell survival. Most cancer therapeutics induce immediate or delayed genomic injury, which can be countered in the tumor cell by acquired resistance to damage-induced apoptosis triggers (20,24). Mutations in p53, ATM, and BRCA1 are examples where DNA damage is not translated into a subsequent apoptotic response (23,61,62). The treatment of metastatic breast carcinomas that do not respond to conventional therapy is a major problem. Thus, novel strategies for eliminating cancer cells are required to target apoptosis pathways not related to the damage sensor, such as receptor tyrosine kinase-mediated survival signaling, which regulates Bcl-2 family proteins. What sets ZD1839 apart from most cancer therapeutics is that it mediates apoptosis by targeting a kinase-driven cellular survival pathway rather than DNA damage. It therefore offers the potential for treating a range of cancers that have become resistant to radiotherapy and other forms of chemotherapy. In this study we examined a pleural effusion culture from a patient whose tumor had failed in sequence to radiotherapy, tamoxifen, anastrozole, standard chemotherapy, epirubicin, taxane, and lastly vinorelbine. These cells were sensitive to treatment with ZD1839 in culture, showing dramatic induction of apoptosis (Fig. 1d).
EGFR and ErbB signaling is frequently activated in cancer (30,32,63). Moreover, other EGF-related ligands including transforming growth factor-␣ and amphiregulin are present at significantly higher levels in cancer than in normal breast (64 -66). Thus, molecules such as ZD1839 that inhibit the signaling pathway responsive to EGFR stimulation may selectively affect breast cancer cells overexpressing either EGFR or its ligands and be of significant clinical value. It is not clear yet whether such an approach would also have an effect on the surrounding normal cells, even though this is not likely to represent a problem in post-menopausal breast cancer patients. EGFR is expressed on normal mammary epithelial cells, which respond to EGFR ligands with an anti-apoptotic and proliferative response (67). Indeed, ZD1839 can inhibit proliferation of xenografts of ductal carcinoma human breast tissue implanted into nude mice (68). However, the precise role of EGFR in normal mammary gland development and function in vivo is unclear. Tissue recombination studies have demonstrated that EGFR is not required in the epithelium for normal mammary ductal growth and morphogenesis or for lobo-alveolar development (69). In mice with triple null mutations in EGF, amphiregulin, and transforming growth factor-␣, cellular proliferation and apoptosis within ductal terminal end buds is not affected, and furthermore, EGF and transforming growth factor-␣ are dispensable for full ductal morphogenesis (70). Because EGFR is expressed in 90% of ductal carcinoma in situ breast tumors, the effect of ZD1839 may be more significant on tumor tissue than on normal breast epithelium. Higher levels of IGF-1 are also associated with increased risk in a variety of cancers, including prostate, breast, colorectal, and lung cancers (71). Similarly, IGF signaling has strong mitogenic and antiapoptotic effects in a wide range of cancer cell lines (72). Our results suggest that ZD1839 might also be effective in reducing some of the effects of IGF signaling amplification.
Some tumors contain mutations in oncogenes downstream of EGFR, such as activated Ras (73,74), or have inactivated apoptosis components downstream of mitochondria, such as altered Apaf-1 expression (75), and ZD1839 may not directly affect such cancers. However, by lowering the level of survival signaling, ZD1839 may increase the sensitivity of these cancers to other drugs in combination therapy. Indeed, some studies have already indicated that ZD1839 increases sensitivity of ovarian, breast, colon, lung, and prostate tumors to taxanes and platinum compounds (63,76). These results, together with favorable pharmacokinetic and pharmacodynamic properties and low toxicity of ZD1839 (34,35), suggest that it may have significant potential as an anticancer agent. Our study is one of the first examples where a potential cancer therapeutic in advanced clinical trials has been demonstrated to induce apoptosis through a growth factor survival signaling pathway.