Dual Inhibition of Focal Adhesion Kinase and Epidermal Growth Factor Receptor Pathways Cooperatively Induces Death Receptor-mediated Apoptosis in Human Breast Cancer Cells*

The focal adhesion kinase (FAK) and epidermal growth factor receptor (EGFR) are protein-tyrosine kinases that are overexpressed and activated in human breast cancer. To determine the role of EGFR and FAK survival signaling in breast cancer, EGFR was stably overexpressed in BT474 breast cancer cells, and each signaling pathway was specifically targeted for inhibition. FAK and EGFR constitutively co-immunoprecipi-tated in EGFR-overexpressing BT474 cells. In low EGFR-expressing BT474-pcDNA3 vector control cells, inhibition of FAK by the FAK C-terminal domain caused detachment and apoptosis via pathways involving activation of caspase-3 and -8, cleavage of poly(ADP-ribose) polymerase, and caspase-3-dependent degradation of AKT. This apoptosis could be rescued by the dominant-negative Fas-associated death domain, indicating in-volvement of the death receptor pathway. EGFR overexpression did not inhibit detachment induced by the FAK C-terminal domain, but did suppress apoptosis, activating AKT and ERK1/2 survival pathways and inhibiting cleavage of FAK, caspase-3 and -8, and poly(ADP-ribose) polymerase. Furthermore, this SDS, 5 m M EDTA, 50 m M NaF, 1 m M NaVO 3 , 10% glycerol, and protease inhibitors (10 (cid:3) g/ml leupeptin, 10 (cid:3) g/ml phenyl-methylsulfonyl fluoride, and 1 (cid:3) g/ml aprotinin). The lysates were cleared by centrifugation at 10,000 rpm for 30 min at 4 °C. Protein concentration was determined using a Bio-Rad kit. The cleared lysates with equivalent amounts of protein were incubated with 5 (cid:3) l of primary antibody for 1 h at 4 °C and with 25 (cid:3) l of protein A/G-agarose beads (Oncogene Research Products Inc.). The precipitates were washed three times with the lysis buffer and resuspended in 30 (cid:3) l of 2 (cid:3) Laemmli buffer. For Western blotting, boiled samples were loaded on Ready SDS-10% polyacrylamide gels (Bio-Rad). The phosphorylation status of the proteins examined was detected with horseradish peroxidase-con-jugated anti-phosphotyrosine antibody RC20 in Tris-buffered saline/ Tween buffer containing 1% bovine serum albumin. The blots were stripped in 62.5 m M Tris-HCl (pH 6.8), 2% SDS, and 100 m M (cid:2) -mercap- toethanol at 60 °C for 30 min and reprobed with primary antibody to protocol. Simultaneous staining and quantification of apoptotic cells by TUNEL assay and Hoechst methods produced very similar results. The percent of apoptotic cells was calculated as the ratio of apoptotic cells to total number of cells in three independent experiments in several fields with a fluorescent microscope.

The invasion and metastasis of cancer require a controlled process of basement membrane degradation, cell motility, and anchorage-independent cell survival. The process of metastasis requires a disseminating cancer cell to survive an environment that actively promotes apoptosis. Thus, for a cancer cell to effectively metastasize, it must possess survival signals that suppress apoptosis. One survival signal that has recently been shown to modulate apoptotic signaling is the focal adhesion kinase (FAK) 1 (1)(2)(3). This non-receptor protein-tyrosine kinase localizes to points of cell contact with the extracellular matrix, the focal adhesions (4,5).
FAK was originally isolated as a tyrosine-phosphorylated 125-kDa protein in v-Src-transformed chicken embryo fibroblasts (6,7). FAK includes an N-terminal domain with a primary autophosphorylation site (Tyr 397 ) that directly interacts with the Src homology-2 domain (8); a central catalytic domain with major sites of phosphorylation (Tyr 576 /Tyr 577 ); and a C-terminal domain with two proline-rich segments and a focal adhesion targeting subdomain, which binds paxillin, talin, and other proteins (4,9). FAK activity is regulated by extracellular matrix receptors and integrins and is involved in cellular processes such as spreading, motility, proliferation, and survival (4,10). A non-catalytic domain of FAK, FAK-related nonkinase (FRNK; p41/p43), is expressed in chicken embryo cells (11), initiated from an alternative promoter and start site residing within an intron (12). Ectopic expression of FRNK causes dephosphorylation of FAK at Tyr 397 (13) and blocks FAK-mediated fibroblast migration (14).
FAK was shown to be overexpressed compared with normal tissue counterparts in many human tumors, including breast, colon, and thyroid carcinomas (15)(16)(17)(18). In human tumor cells, inhibition of FAK expression with antisense oligonucleotides to FAK or overexpression of the focal adhesion targeting subdomain leads to cell rounding, detachment, reduction of invasion, and apoptosis (1, 19 -22). Furthermore, FAK has been shown to suppress both transformation-associated apoptosis (2) as well as anoikis (detachment-induced apoptosis) of epithelial cells (23), suggesting that one function of FAK is to promote survival in cells subjected to apoptotic signals. Consistent with this hypothesis, constitutively active forms of FAK prevent anoikis and stimulate transformation of epithelial cells, resulting in anchorage-independent growth and tumor formation in nude mice (23). Further evidence for the anti-apoptotic role of FAK was shown in the leukemic cell line HL-60, where FAK is associated with activation of NF-B and inhibition of caspase-3 (24). Conversely, caspase-3 and -6 may promote apoptosis in part by cleaving FAK and generating a C-terminal FRNK-like polypeptide (25).
Recently, FAK was shown to be associated with the epidermal growth factor receptor (EGFR), also known as ErbB-1 (26,27). When epidermal growth factor (EGF) binds to 170-kDa EGFR, receptor homo-and heterodimerization is promoted, activating receptor tyrosine kinase activity (28) and downstream signaling (reviewed in Refs. 29 -32). EGFR is overexpressed or activated by autocrine growth factors in many types of tumors, including breast (33,34), thyroid (35), ovarian (36), colon (37), head and neck (38), and brain (38,39). Furthermore, EGFR overexpression has been linked to a poor prognosis in breast cancer (32,40) and may promote proliferation, migration, invasion, and cell survival as well as inhibition of apoptosis (41)(42)(43). Recent reports have suggested that FAK serves to integrate EGFR signals upon EGF induction, promoting tumor cell motility and invasion (22,27). However, another report suggests that FAK and EGFR are constitutively associated (26). Thus, the relationship between EGFR signaling and FAK expression and activity during progression from a noninvasive to an invasive and metastatic tumor phenotype is unknown, and their cooperation in preventing apoptosis has not been mechanistically examined.
In this study, we examined the role of FAK and EGFR in survival signaling in a human breast cancer cell line model system of EGFR overexpression. We stably overexpressed EGFR in a cell line that endogenously overexpresses FAK (BT474-EGFR cells) to compare the effects of EGFR survival signaling with the parental cell line without EGFR (BT474-pcDNA3 cells). We have demonstrated that dual inhibition of FAK and EGFR cooperatively caused apoptosis in breast cancer cells. In breast cancer cells that stably overexpressed EGFR, there was a constitutive association between FAK and EGFR. Furthermore, EGFR signaling suppressed death receptor-mediated apoptosis induced by the FAK C-terminal domain (FAK-CD). The mechanism included activation of AKT and ERK signaling pathways as well as protection of FAK from caspase degradation that was reversed by an EGFR kinase inhibitor. Dual inhibition of FAK by FAK-CD and of EGFR by AG1478 cooperatively enhanced apoptosis in human breast cancer cell lines via inhibition of signaling that involved both tumor necrosis factor receptor (TNFR) family-dependent AKT and ERK1/2 pathways. This is the first report of the role of FAK and EGFR in apoptosis showing that simultaneous inhibition of FAK and EGFR can be critical in induction of apoptosis in breast cancer cell lines.

EXPERIMENTAL PROCEDURES
Cells and Cell Culture-BT474 breast carcinoma cells, described by Xu et al. (2), were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum, 5 g/ml insulin, and 1 g/ml penicillin/ streptomycin. The clone of the BT474 cell line used in this study had low expression of EGFR (Her-1) and Her-2. BT20 breast carcinoma cells, which overexpress EGFR (44), were maintained in Eagle's minimal essential medium containing 10% fetal bovine serum. Cell lines were incubated at 37°C in a 5% CO 2 humidified incubator. For EGF stimulation (Western blotting and immunoprecipitation), cells were serumstarved overnight in serum-free medium. In FAK-CD-induced apoptosis experiments, EGF was added three times at 0, 6, and 20 or 22 h of adenoviral infection. EGF (Calbiochem) was used at 10 ng/ml for 10 min for EGF-dependent signaling experiments. For inhibition of EGF stimulation, pretreatment with the EGFR kinase inhibitor AG1478 was done at 5 M for 15 min. In FAK-CD-induced apoptosis experiments, BT474 cells were incubated with 5 M AG1478 for 24 h, and BT20 cells were incubated with AG1478 for Ͼ72 h at 5 M (added fresh every 24 h).
Adenoviral Infection-Recombinant adenoviruses carrying the lacZ gene (Ad-lacZ) and the HA-tagged FAK-CD gene, coding amino acids 693-1052 of FAK (Ad-FAK-CD) and the dominant-negative FADD protein (Ad-⌬FADD), were propagated by Dr. J. Samulski and the Gene Therapy Center Virus Vector Core Facility of the University of North Carolina and are described in Ref. 2. Cells were plated at 1.5 ϫ 10 6 in 100-mm culture plates and, after 24 h of attachment, infected with adenoviruses at the optimal concentration in 7 ml of medium with 10% serum. The optimal viral concentration was determined by infection of cells with different viral concentrations, and viral titer that produced Ͼ95% cell infectivity was used. The optimal viral titer was 500 focusforming units/cell (obtained from the Gene Therapy Center Virus Vector Core Facility), which produced 99% cell infectivity and no toxic effects for Ad-lacZ transduction, checked by X-gal staining. The same viral titer was used for Ad-FAK-CD infection, causing Ͼ90% of cell infectivity, determined by HA immunostaining. For co-infection experiments, Ad-⌬FADD was used with FAK-CD at 333 viral particles/cell, which was shown to produce high levels of ⌬FADD protein inside BT474 cells (2). For BT474 cells, caspase inhibitors (Ac-DEVD-CHO, caspase-3 inhibitor; and Ac-YVAD-CHO, interleukin-converting enzyme inhibitor) were added at 50 M (2) for 1 h before adenoviral infection and were present during incubation with adenoviruses for 23-24 h. For BT20 cells, the cell-permeable caspase-3 inhibitor DEVD-CHO (Calbiochem) was added at 5 M for 1 h before infection and was present during adenoviral incubation, added fresh every 24 h. The ERK1/2 inhibitor PD98059 was added at 10 M for 15 min before adenoviral infection and was present during incubation with adenoviruses for 24 h. Under these conditions, PD98059 inhibited phosphorylation of ERK1/2 in Ad-FAK-CD-infected BT474-EGFR cells (data not shown).
Staining with X-Gal-Cells were infected with Ad-lacZ as described above. Briefly, 24 h after infection, cells were fixed for 10 min on ice with 2% formaldehyde and 0.2% glutaraldehyde in 1ϫ phosphate-buffered saline (PBS). After washing twice with 1ϫ PBS, cells were stained for 1-3 h with 1 g/ml X-gal in X-gal buffer (5 mM K 3 Fe(CN) 6 , 5 mM K 4 Fe(CN) 6 , and 2 mM MgCl 2 in 1ϫ PBS). X-gal-positive (blue) cells were counted for determining of infection efficiency.
Transfections-To make a stable BT474 cell line expressing EGFR, cells were transfected with 10 g of pcDNA3 plasmid alone to create BT474-pcDNA3 vector control cells or with 10 g of EGFR-pcDNA3 plasmid (kindly provided by Dr. David Lee) to create BT474-EGFR cells. Transfections were accomplished with 20 l of LipofectAMINE (Invitrogen) in a 100-mm dish following the manufacturer's protocol. Stable BT474-EGFR and BT474-pcDNA3 clones were obtained using RPMI 1640 medium with the selective antibiotic Geneticin (500 M; G418, Invitrogen). Expression of EGFR was checked by Western blotting with anti-EGFR antibody, and a clone with maximal EGFR expression was used for the study.
Immunoprecipitation and Western Blotting-Cells were washed twice with cold 1ϫ PBS and lysed on ice for 30 min in buffer containing 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 5 mM EDTA, 50 mM NaF, 1 mM NaVO 3 , 10% glycerol, and protease inhibitors (10 g/ml leupeptin, 10 g/ml phenylmethylsulfonyl fluoride, and 1 g/ml aprotinin). The lysates were cleared by centrifugation at 10,000 rpm for 30 min at 4°C. Protein concentration was determined using a Bio-Rad kit. The cleared lysates with equivalent amounts of protein were incubated with 5 l of primary antibody for 1 h at 4°C and with 25 l of protein A/G-agarose beads (Oncogene Research Products Inc.). The precipitates were washed three times with the lysis buffer and resuspended in 30 l of 2ϫ Laemmli buffer. For Western blotting, boiled samples were loaded on Ready SDS-10% polyacrylamide gels (Bio-Rad). The phosphorylation status of the proteins examined was detected with horseradish peroxidase-conjugated anti-phosphotyrosine antibody RC20 in Tris-buffered saline/ Tween buffer containing 1% bovine serum albumin. The blots were stripped in 62.5 mM Tris-HCl (pH 6.8), 2% SDS, and 100 mM ␤-mercaptoethanol at 60°C for 30 min and reprobed with primary antibody to check equal loading of proteins. Immunoblots were developed with Renaissance chemiluminescence reagent (PerkinElmer Life Sciences).
Immunostaining-Attached or detached suspended cells (collected by centrifugation and spread evenly on the slide) were fixed in 4% paraformaldehyde in 1ϫ PBS for 10 min and permeabilized with 0.2% Triton X-100 for 5 min on ice. Cells were blocked with 25% normal goat serum in 1ϫ PBS for 30 min, washed with 1ϫ PBS, and incubated with primary antibody diluted 1:200 in 25% goat serum in 1ϫ PBS. Cells were washed three times with 1ϫ PBS, and a TRITC-conjugated secondary antibody (1:400 dilution in 25% goat serum) was applied to the coverslip. After washing three times with 1ϫ PBS, cells were incubated with fluorescein isothiocyanate-Bodipy FL -phallacidin (1:25 dilution in 25% goat serum; Molecular Probes, Inc.) for actin staining. For coimmunostaining experiments, cells were incubated with another primary antibody diluted 1:100 in 25% goat serum in 1ϫ PBS for 1 h. After washing three times with 1ϫ PBS, a fluorescein isothiocyanate-conjugated secondary antibody (1:100 dilution) was applied to the coverslip.
Apoptosis Assay-Detached cells were collected by centrifugation, fixed in 3.7% formaldehyde in 1ϫ PBS for 10 min, and stained with Hoechst 33342 or spread evenly on a slide for TUNEL staining. In brief, Hoechst 33342 in 1ϫ PBS solution (1 g/ml) was added to the fixed cells for 10 min, and cells were washed twice with 1ϫ PBS and spread evenly on the slide. TUNEL assay was done with an ApopTag fluorescein in situ apoptosis detection kit (Intergen) according to the manufacturer's protocol. Simultaneous staining and quantification of apoptotic cells by TUNEL assay and Hoechst methods produced very similar results. The percent of apoptotic cells was calculated as the ratio of apoptotic cells to total number of cells in three independent experiments in several fields with a fluorescent microscope.

EGF-dependent Tyrosine Phosphorylation and Its Inhibition by AG1478 in BT474-EGFR Cells-
To test the relationship between FAK and EGFR in breast cancer cell survival, we created a model system of EGFR overexpression in a clone of the BT474 breast carcinoma cell line (2), a cell line that expresses high levels of p125 FAK , but minimal levels of EGFR. Parental BT474 cells were stably transfected with the EGFR-pcDNA3 plasmid. These cells (called BT474-EGFR cells) expressed high levels of EGFR compared with the pcDNA3 vector control cells (called BT474-pcDNA3 cells) (Fig. 1A). In the BT474-EGFR cells, treatment with EGF at 10 ng/ml rapidly increased EGFR tyrosine phosphorylation, and this effect was inhibited in the presence of the tyrphostin AG1478, an EGFR kinase inhibitor (Fig. 1B). The BT474-EGFR cells had a higher level of tyrosine-phosphorylated cellular proteins than the vector control cells (Fig. 1C), and both BT474-pcDNA3 and BT474-EGFR cells rapidly increased phosphorylation of ERK1/2 at 10 min upon EGF stimulation and reversed this effect upon AG1478 treatment (Fig. 1D). In addition, AKT (Ser 473 ) and FAK (phospho-Tyr 397 and phospho-Tyr 577 ) (data not shown) were highly phosphorylated in the BT474-EGFR cells.
FAK and EGFR Constitutively Associate in BT474-EGFR Cells-To test whether FAK and EGFR were associated in the BT474-EGFR cells, EGFR and FAK were immunoprecipitated using anti-FAK monoclonal antibody ( Fig. 2A, left panels). In these experiments, anti-FAK antibodies precipitated EGFR that was tyrosine-phosphorylated in the presence of EGF ( Fig.  2A, upper left panel). Immunoprecipitation with anti-EGFR antibodies also precipitated FAK ( Fig. 2A, right panels). The FAK and EGFR association was constitutive, as detected in the presence and absence of EGF ( Fig. 2A). Thus, FAK and EGFR are physically associated in these breast cancer cells. Tyrosine phosphorylation of total cellular FAK was not affected by the EGFR inhibitor AG1478 ( Fig. 2A, left panels), although a small portion of phosphorylated FAK that was associated with EGFR was inhibited by AG1478 (upper right panel). Dual immunofluorescence assays in individual cells demonstrated that FAK and EGFR were co-localized at focal adhesions in the BT474-EGFR cells (Fig. 2B, left panels) and also in the BT20 cells, which endogenously overexpress EGFR (right panels). These results show that FAK and EGFR associate and co-localize at the focal adhesions of these breast cancer cells.
Overexpression of EGFR Suppresses Apoptosis Induced by FAK Down-regulation-Next, we tested whether overexpression of EGFR protects breast cancer cells from detachment and apoptosis induced by FAK inhibition. In these experiments, we down-regulated FAK function using Ad-FAK-CD in the BT474-EGFR and BT474-pcDNA3 vector control cells. To inhibit FAK, we used an adenoviral construct of the C-terminal domain (Ad-FAK-CD) that has been shown to act as a dominant-negative for FAK function (2,45,46). In these experiments, we used Ad-lacZ as a control at an equal multiplicity of infection. These conditions resulted in Ͼ95% infectivity as assessed by X-gal staining for Ad-lacZ (Fig. 3A) and by HA immunostaining for HA epitope-tagged Ad-FAK-CD (Fig. 3B).
After 24 h of FAK-CD expression, there were similar levels of loss of adhesion in the vector control cells (93 Ϯ 6%) and in the EGFR-overexpressing cells (83 Ϯ 17%) (Fig. 4A) and only 0.5% detachment by control Ad-lacZ at the same dose (data not FIG. 1. EGF-dependent signaling and its inhibition by AG1478 in BT474-EGFR cells. A, Western blotting was performed with anti-EGFR monoclonal antibody in the BT474-pcDNA3 vector control and BT474-EGFR cells. Equal protein loading was analyzed by Western blotting with anti-␤-actin antibody. B, EGFR increases phosphorylation after EGF stimulation and is inhibited in the presence of AG1478. Immunoprecipitation (IP) was performed with anti-EGFR monoclonal antibody, followed by the Western blotting (WB) with anti-phosphotyrosine antibody RC20. Cells were treated with EGF (10 ng/ml) for 10 min after overnight serum starvation. AG1478 (EGFR kinase inhibitor) was used at 5 M for 15 min before EGF stimulation. C, shown is the total phosphorylation of proteins. BT474 cells were serum-starved overnight and induced by EGF (10 ng/ml) for 10 min with or without AG1478 (5 M) pretreatment for 15 min. The human EGFR-overexpressing epidermoid carcinoma cell line A431 was used as a positive control for induction of EGFR phosphorylation. Western blotting was performed on total cell lysates with anti-phosphotyrosine antibody RC20. Western blotting with anti-␤-actin was performed on BT474 lysates to confirm equal protein loading. D, shown is ERK1/2 activation in response to EGF treatment reversed by AG1478. ERK1/2 activation was analyzed with anti-phospho-ERK1/2 antibody after EGF stimulation for 10 min and with AG1478 pretreatment after overnight serumfree medium starvation. Total ERK1/2 levels were determined with anti-ERK1/2 polyclonal antibody. p-ERK1/2 indicates phosphorylated active ERK1/2. shown). Furthermore, inhibition of EGFR kinase activity with AG1478 did not enhance the loss of adhesion (Fig. 4A). These results show that overexpression of EGFR does not augment the ability of BT474 cells to resist loss of adhesion induced by inhibition of FAK function.
In contrast, EGFR suppressed apoptosis induced by FAK inhibition with Ad-FAK-CD. After 24 h of Ad-FAK-CD infection, detached BT474-EGFR cells had significantly reduced levels of apoptosis (66%) compared with the BT474-pcDNA3 cells (97%) (Fig. 4B). Treatment of the BT474-EGFR cells with EGF (10 ng/ml) did not enhance this resistance to apoptosis, indicating that the maximal effect was already obtained by the background autophosphorylation and signaling caused by EGFR overexpression. However, inhibition of EGFR signaling with AG1478 in the BT474-EGFR cells increased the level of apoptosis to that in the BT474-pcDNA3 control cells (Fig. 4B). In Hoechst-stained cells (Fig. 4C), apoptotic condensed nuclei with fragmented chromatin in the BT474-EGFR cells were maximal with both FAK and EGFR inhibition. In addition, when FAK was inhibited by FAK-CD and ERK1/2 was inhibited by PD98059 (ERK1/2 inhibitor) in the BT474-EGFR cells, the level of apoptosis was increased to that in the BT474-pcDNA3 vector controls (Fig. 4D), suggesting that ERK1/2 enhances survival signaling in these breast cancer cells. Taken together, these results demonstrate that EGFR overexpression confers additional survival signals in the breast cancer cells that suppress the apoptosis-inducing effect caused by loss of FAK function.

FIG. 2. FAK constitutively associates with EGFR in BT74-EGFR cells.
A, BT474 cells were stimulated with EGFR as described in the legend to Fig. 1. Left panels, immunoprecipitation (IP) was performed with anti-FAK monoclonal antibody 4.47, and Western blotting (WB) was done with anti-phosphotyrosine antibody. The membrane was stripped, and Western blotting was performed with anti-FAK antibody and then stripped again and reprobed with anti-EGFR antibody to show that FAK coprecipitated with EGFR. Right panels, the same experiment was performed with immunoprecipitation of EGFR, detecting FAK and EGFR association. The experiment was done three times with the same results. The images are composites from the same films/gels. B, FAK and EGFR association and co-localization at focal adhesions was detected by co-immunostaining. Immunostaining was performed with mouse anti-FAK monoclonal primary antibody 4.47 and probing with rhodamine-conjugated anti-mouse secondary antibody. After washing, dual immunostaining was performed with anti-EGFR monoclonal antibody and probing with fluorescein isothiocyanate (FITC)conjugated anti-mouse secondary antibody. The merged images were obtained with Adobe Photoshop Version 6.0. BT474-pcDNA3 and BT474-EGFR cells were infected with Ad-FAK-CD or Ad-lacZ (see "Experimental Procedures") at 500 focus-forming units/cell, resulting in 100% cell infectivity. After 24 h, Ad-FAK-CD-infected detached cells were counted using a hemocytometer. Ad-lacZ-infected cells were resistant to detachment and were not apoptotic (data not shown). Four independent experiments were done with the same results, and a representative experiment is shown. The mean percent Ϯ S.D. of detached cells is shown from three independent cell counts using a hemocytometer. B, apoptosis induced by Ad-FAK-CD is inhibited by EGFR overexpression and reversed by the EGFR kinase inhibitor AG1478. Detached BT474 cells that were infected with Ad-FAK-CD were fixed and analyzed for apoptosis. Apoptosis was determined by Hoechst staining. Ad-lacZ-infected cells were not apoptotic (data not shown). Bars represent mean values Ϯ S.D. More than 100 cells were counted in three independent fields for each experimental treatment in three independent experiments. C, Hoechst staining of BT474 cells treated with Ad-FAK-CD. Apoptotic Ad-FAK-CD-infected cells with fragmented nuclei had bright Hoechst staining of condensed nuclear chromatin. Ad-lacZ-infected cells did not undergo apoptosis and had unfragmented nuclei, as did control BT474 cells (data not shown). Left panels, cells viewed for phase contrast; right panels, identical cells viewed for Hoechst-stained nuclei. Normal cells had faintly stained nuclei, whereas apoptotic condensed nuclei stained brightly. D, the ERK1/2 inhibitor PD98059 increases Ad-FAK-CD-induced apoptosis in BT474-EGFR cells. BT474-EGFR cells were pretreated with 10 M PD98059 for 15 min before Ad-FAK-CD treatment. PD98059 was present during adenoviral incubation for 24 h. After 24 h of adenoviral infection, apoptosis was measured as described for B. The graph shows PD98059-increased apoptosis in BT474-EGFR cells. Bars represent means Ϯ S.D. More than 100 cells were counted in three independent fields for each experimental treatment in three independent experiments. Statistical significance was determined using Student's t test. *, a significant difference from the pcDNA3 control (p Ͻ 0.02); **, a significant difference from the EGFR sample (p Ͻ 0.040). Hoechst-stained nuclei are also shown.

Inhibition of FAK and EGFR in BT474-EGFR Cells Downregulates Both ERK1/2 and TNFR Family-dependent AKT
Survival Pathways-Because EGFR appeared to function as a survival signal to protect breast cancer cells from apoptosis induced by FAK down-regulation, we investigated the downstream biochemical pathways for EGFR and FAK, beginning with the MAPK pathways. After 24 h of infection with Ad-FAK-CD, the BT474-EGFR cells demonstrated up-regulation of phosphorylated ERK1/2 that could be inhibited by the EGFR kinase inhibitor AG1478 (Fig. 5). This high level of ERK1/2 phosphorylation was not seen in the BT474-EGFR cells treated with control Ad-lacZ or in the pcDNA3 vector control cells treated with Ad-FAK-CD, showing that Ad-FAK-CD initiates an EGFR-mediated stress response and survival signaling that includes ERK1/2 phosphorylation.
We also examined the AKT (protein kinase B) pathway (47) because FAK has been shown to act upstream of this serine/ threonine kinase, which has important survival signal func-tions in tumor cells (48). AKT was constitutively expressed and Ser 473 -phosphorylated in the BT474-EGFR cells as well as in the BT474-pcDNA3 control cells treated with control Ad-lacZ (Fig. 5). The BT474-pcDNA3 cells treated with Ad-FAK-CD down-regulated AKT, as total AKT and Ser 473 -phosphorylated AKT were not present in these cells (Fig. 5, third and fourth  panels, third lanes). In contrast to the BT474-pcDNA3 control cells, the BT474-EGFR cells (independent of EGF ligand) expressed AKT and the active Ser 473 -phosphorylated form of AKT in response to down-regulation of FAK (Fig. 5). However, when EGFR kinase activity was inhibited by AG1478 in the BT474-EGFR cells, there were undetectable levels of AKT protein upon Western blotting, and AKT was completely dephosphorylated (Fig. 5). Based on these data, EGFR not only signals through ERK1/2, but also has an effect on the ability of AKT to resist down-regulation in response to FAK inhibition.
To further assess the relationship of EGFR overexpression and FAK down-regulation to the ERK1/2 and AKT pathways in BT474 cells, we inhibited TNFR family signaling in the BT474-EGFR and BT474-pcDNA3 cells, based on our recent data that the TNFR family regulates FAK-CD-induced apoptosis (2). To analyze if these receptors are important in reduction of AKT protein levels in apoptotic BT474-pcDNA3 cells infected with Ad-FAK-CD, we blocked death receptor pathways by co-infection of cells with adenoviral dominant-negative FADD lacking amino acids 1-79 of the death effector domain (Ad-⌬FADD) (49). Under these conditions, where the death complex was inhibited, AKT protein levels were not reduced, and AKT was Ser 473 -phosphorylated in the BT474-pcDNA3 and BT474-EGFR cells (Fig. 5, last two lanes). The results demonstrate that AKT down-regulation (reduction of protein levels) is mediated through a TNFR family pathway in the BT474-pcDNA3 cells infected with Ad-FAK-CD. In contrast, dominant-negative FADD had no effect on ERK1/2 phosphorylation in the cells, indicating the independence of the ERK1/2 survival pathway from the TNFR family pathways (Fig. 5). These results show that TNFR family signaling, which is involved in FAK-CDinduced apoptosis, is also important in down-regulation/cleavage of AKT. Taken together, these biochemical results parallel the cell biological results above, whereby inhibition of both FAK and EGFR caused both the highest level of apoptosis as well as inhibition of both the MAPK (ERK1/2) and TNFR family-dependent AKT signaling pathways.

EGFR Overexpression Protects p125 FAK and Caspase-3 and -8 from Complete Degradation in Response to FAK Inhibition in BT474 Cells
Reversed by AG1478 -To further examine the effects of EGFR on resistance to apoptosis, we tested whether overexpression of EGFR would protect endogenous p125 FAK in the breast cancer cells from degradation in response to Ad-FAK-CD. Previous work from our group has shown that p125 FAK is degraded 24 h after Ad-FAK-CD infection in parental BT474 cells and that this effect is mediated through caspase-8 and -3 (2). Similarly, in these studies, the p125 FAK protein was degraded in the BT474-pcDNA3 control cells upon infection with Ad-FAK-CD (Fig. 5). However, in the BT474-EGFR cells, EGFR overexpression protected p125 FAK from complete degradation by Ad-FAK-CD, as shown by the 125-kDa FAK band that was present in the BT474-EGFR cells, but not in the BT474-pcDNA3 cells (Fig. 5, third through fifth lanes). Down-regulation of both FAK and EGFR did lead to complete degradation of p125 FAK in the BT474-EGFR cells, whereby probing the Western blots with an antibody to the N terminus of FAK detected only 85-90-kDa degradation products (Fig. 5,  sixth and seventh lanes). Co-infection of the BT474 cells with Ad-FAK-CD and Ad-⌬FADD blocked cleavage of FAK (Fig. 5).
In the BT474-pcDNA3 cells, treatment with Ad-FAK-CD

FIG. 5. EGFR-overexpressing BT474 cells infected with Ad-FAK-CD activate ERK1/2 and AKT signaling pathways and partially protect FAK from degradation. Ad-⌬FADD blocked FAK degradation and AKT down-regulation caused by Ad-FAK-CD.
BT474-pcDNA3 and BT474-EGFR cells were treated with either control Ad-lacZ or Ad-FAK-CD with or without EGFR kinase inhibition by AG1478 and analyzed for ERK1/2 and AKT activation and FAK degradation. ERK1/2 activation was analyzed by Western blotting (WB) with anti-phospho-ERK1/2 and anti-total ERK1/2 antibodies. ERK1/2 was activated in BT474-EGFR cells infected with Ad-FAK-CD in the presence or absence of EGF and inhibited by AG1478 treatment. AKT status was analyzed with anti-phospho-Ser 473 AKT and anti-total AKT antibodies. AKT was not detected in BT474-pcDNA3 cells infected with Ad-FAK-CD, but was detected in control cells infected with Ad-lacZ. BT474-EGFR cells expressed active serine-phosphorylated AKT and total AKT after Ad-FAK-CD infection. Treatment with AG1478 in the presence or absence of EGF caused down-regulation of AKT. Ad-⌬FADD protected AKT from down-regulation and FAK from complete degradation. Western blotting with anti-HA antibody controlled for HA-tagged FAK-CD protein levels in the samples. Equal protein loading was controlled with anti-␣-tubulin antibody. Each experiment was performed five times with different adenoviral preparations with the same results, and a representative experiment is shown. induced caspase-3 activation and PARP (caspase-3 substrate) cleavage (Fig. 6A). Pretreatment of these cells with the Ac-DEVD-CHO peptide (a caspase-3 family inhibitor) prior to infection with Ad-FAK-CD blocked activation of caspase-3 and cleavage of PARP and increased the levels of total AKT and FAK (Fig. 6A), indicating that reduction of AKT and FAK protein levels is the result of FAK and AKT cleavage by a caspase-3 pathway in BT474 cells.
In contrast, overexpression of EGFR protected caspase-3 and -8 from degradation caused by FAK down-regulation. In the BT474-EGFR cells treated with Ad-FAK-CD, there was significant protection of caspase-3 from cleavage, with incomplete PARP cleavage (Fig. 6B, fourth and fifth lanes). However, when EGFR was inhibited under these conditions, the cleavage of caspase-3 and PARP was restored to levels equivalent to those in the BT474-pcDNA3 cells (Fig. 6B, third, sixth, and seventh  lanes). The upstream caspase-8 showed a similar effect, whereby the BT474-EGFR cells did not activate caspase-8 in response to FAK-CD, but in combination with EGFR inhibition, cleaved the inactive proform and activated the enzyme (Fig.  6C). Co-infection of the BT474-pcDNA3 cells with Ad-FAK-CD and Ad-⌬FADD protected caspase-8 and -3 and PARP from cleavage (Fig. 6).
Dual Inhibition of FAK and EGFR in BT20 Breast Cancer Cells, Which Endogenously Overexpress EGFR, Enhances Apoptosis, Down-regulating AKT and ERK1/2 Survival Pathways-In a final series of experiments, we tested whether endogenous EGFR in a breast cancer cell line would have similar survival signal effects as our model system of EGFR overexpression in the BT474-EGFR cells. We used the BT20 cell line, which has been shown to express high levels of endogenous EGFR (44).
Similar to the BT474-EGFR cells, inhibition of FAK by Ad-FAK-CD in BT20 breast cancer cells induced loss of adhesion, although this effect was seen at later time points, 46 -71 h after adenoviral infection. The BT20 cells treated with Ad-FAK-CD started to detach at 46 h, and Ͼ60% of the cells treated with FAK-CD alone or with AG1478 in the absence or presence of EGF detached by 71 h (Fig. 7A).
However, inhibition of FAK and EGFR enhanced the levels of apoptosis in these cells (Fig. 7B) in a fashion similar to that seen in the BT474-EGFR cells (Fig. 4B). At 46 h after infection, the levels of apoptosis were slightly increased when cells were  (first and second lanes). B and C, activation of caspase-3 and -8 and PARP in response to Ad-FAK-CD is blocked in BT474-EGFR cells and reversed by inhibition of EGFR. B, Ad-FAK-CD-induced activation of caspase-3 and PARP cleavage is blocked in BT474-EGFR and Ad-⌬FADD cells. Caspase-3 activation was determined with anti-caspase-3 antibody. BT474-EGFR and Ad-⌬FADD cells suppressed caspase-3 activation. PARP cleavage analysis was performed by Western blotting with anti-PARP antibody, specific to uncleaved 116-kDa PARP, on the same lysates as in A. Loading was controlled with anti-␣-tubulin antibody. Each experiment was repeated three times with two independent Ad-FAK-CD adenoviruses with the same results, and a representative experiment is shown. C, activation of caspase-8 cleavage in BT474 cells infected with Ad-FAK-CD is blocked by EGFR and Ad-⌬FADD expression. Caspase-8 activation was determined with anti-caspase-8 antibody, specific to uncleaved inactive 55-kDa caspase-8. BT474-EGFR and Ad-⌬FADD cells suppressed caspase-8 activation. ␤-Actin protein was used for normalization of protein levels. treated with FAK-CD and AG1478 compared with cells that had been treated with FAK-CD alone (7% versus 1.6%). However, at 71 h, this effect was more apparent, where the apoptotic rate with FAK-CD alone was 10%, but addition of the EGFR kinase inhibitor increased the rate to 33 and 43% (without and with EGF, respectively) (Fig. 7B).
Next, we directly compared the biochemical effects of FAK and EGFR inhibition in the BT474-EGFR and BT20 cells. As shown by control Ad-lacZ infection, the levels of endogenous EGFR and p125 FAK expression as well as the levels of AKT phosphorylation were higher in the BT20 cells (Fig. 8A, sixth lane) than in the BT474-pcDNA3 cells (first lane) or the BT474-EGFR cells (second lane). After 24 h of infection with Ad-FAK-CD, the BT20 cells totally protected FAK from degradation detected in the BT474-pcDNA3 cells (Fig. 8A). The ERK1/2 survival pathway was activated in the BT20 cells (as in the BT474-EGFR cells), as ERK1/2 was highly phosphorylated in BT20 cells infected with Ad-FAK-CD and dephosphorylated upon treatment with AG1478. However, AKT remained highly active and Ser 473 -phosphorylated in the BT20 cells after 24 h of FAK-CD infection, as the BT20 cells had high levels of AKT (Fig. 8A). At 72 h after FAK-CD expression and in the presence of AG1478, the BT20 cells detached and had down-regulated both ERK1/2 and AKT phosphorylation and also down-regulated total p125 FAK , caspase-3, and AKT proteins (Fig. 8B).
Next, we tested whether the degradation of FAK and AKT in the BT20 cells is caspase-3-dependent, similar to the BT474 cells. To down-regulate FAK and AKT more efficiently in the BT20 cells, we used the same dose of Ad-FAK-CD per cell, but concentrated it 3.5-fold in the culture medium. Under these conditions, FAK and AKT were completely down-regulated at 55 h (Fig. 8C). However, pretreatment of cells with the caspase-3 subfamily inhibitor DEVD-CHO blocked degradation of FAK and AKT (Fig. 8C), demonstrating that FAK and AKT down-regulation in the AG1478-and Ad-FAK-CD-treated BT20 cells was caspase-3-dependent. These results mechanistically support the increased levels of apoptosis seen upon Hoechst staining in the detached cells treated with Ad-FAK-CD and AG1478 and show that dual inhibition of EGFR and FAK increased apoptosis in the BT20 cell line, inhibiting the same survival signaling pathways as the BT474-EGFR cells. DISCUSSION This study demonstrates the cooperativity of both FAK and EGFR signals in suppressing apoptosis in breast cancer cells. Although there appears to be a physical association of these tyrosine kinases, their individual survival signals appear also in part to be in parallel. Thus, inhibition of both FAK and EGFR signaling pathways led to significantly higher levels of apoptosis than inhibition of either one alone. These results were supported at a biochemical level, whereby the EGFRoverexpressing cells had increased levels of both ERK1/2 and AKT phosphorylation and did not demonstrate complete p125 FAK or caspase-3 or -8 degradation until both FAK and EGFR signaling had been interrupted. Thus, this study is the first to show the cooperative effect of FAK and EGFR inhibitors in induction of apoptosis in human breast cancer cells. We propose a model of survival signaling in breast cancer cells whereby FAK and EGFR overexpression can promote survival signals via an AKT-dependent mechanism as well as via an ERK1/2 pathway (Fig. 9). Dual inhibition of FAK and EGFR led to apoptosis via death receptor-mediated signaling.
Individually, both FAK and EGFR have been shown to be overexpressed in human breast cancer specimens (15,17,18). However, the relationship between these kinases and the subsequent cellular effects in breast cancer remain unclear. It has been shown that FAK and EGFR can associate when coex-pressed in FAK Ϫ/Ϫ fibroblasts in the presence of EGF (27), suggesting that FAK can mediate a linkage between growth factor receptors and integrins. Similarly, the association of FAK and EGFR has been shown in A431 epidermoid cancer cells (26) and A549 adenocarcinoma cells (22), both of which express extraordinarily high levels of EGFR (50,51). However, these studies differed in whether FAK and EGFR constitutively associate (26) or whether this association requires EGF ligand (22). Our results support a constitutive association between FAK and EGFR in BT474 breast cancer cells stably overexpressing EGFR, but it is unclear what effect this association has on downstream signaling pathways.
The studies of FAK and EGFR in cancer cells have largely focused on their effects on tumor cell motility. Inhibition of FAK in A549 cells was shown to inhibit EGF-stimulated motility (22), providing further evidence that FAK can integrate motility signals from EGF to EGFR. Other motility studies suggest that FAK is dephosphorylated in response to EGF, promoting tumor invasion and motility (26). In studies of human glioblastoma cells, inhibition of FAK function by exogenously expressing the focal adhesion targeting domain also diminished EGFR-directed cell motility (19). We have examined motility of breast cancer cells in our system and found that FIG. 7. BT20 breast cancer cells, which endogenously overexpress EGFR, show detachment and increased apoptosis with FAK and EGFR inhibition. A, detachment assay was done as described for BT474 cells in the legend to Fig. 4A. BT20 cells started to detach at 46 h after Ad (Adeno)-FAK-CD infection, with Ͼ60% cells detached at 71 h. At 71 h, cells were equally detached in all treated samples. Cells did not detach after Ad-lacZ infection (data not shown). B, at 71 h, apoptosis was increased in BT20 cells by FAK and EGFR inhibition, with and without EGF. Apoptosis was determined as described in the legend to Fig. 4B. Two independent experiments were done with the same results, and a representative experiment is shown.
down-regulation of FAK rapidly diminished both random and EGF-directed cell motility (data not shown), supporting the hypothesis that FAK is involved in EGF-directed motility pathways. Thus, the model proposed by Hauck et al. (22) is consistent with our findings in breast cancer cells. In their model, interactions between FAK and EGFR and between FAK and integrins concomitant with Src family activity via FAK phosphorylation at Tyr 397 activate the downstream pathway that promotes motility.
However, it appears that the effects of FAK overexpression in breast cancer cells are not simply limited to motility and invasion, but can cooperate with EGFR signaling to suppress apoptosis and to enhance survival of breast cancer cells. Studies of FAK in primary breast cancer specimens have shown that up-regulation of FAK expression is an early event in tumorigenesis, occurring in ductal carcinoma in situ, before the tumor has developed the capacity for invasion and metastasis (18). These observations support the hypothesis that FAK functions to promote survival during tumor cell proliferation before invasion and migration have occurred. Furthermore, other studies in breast cancer cell lines suggest that FAK has two separate functions in human tumor cells: one promoting adhesive interactions between tumors and the matrix and the other providing survival signals to resist apoptosis (2). Our results in this study also support this hypothesis, whereby down-regulation of FAK function had effects on apoptosis in EGFR-overexpressing breast cancer cells.
In this study, the biochemical mechanisms of apoptotic resistance appeared to involve both the ERK and AKT pathways. Overexpression of EGFR was associated with a robust phosphorylation of ERK1/2 in the BT474-EGFR cells, and this appeared to augment the resistance to apoptosis induced by FAK down-regulation. This is consistent with other studies implicating the ERK pathways in apoptotic resistance, including tumor necrosis factor-␣-induced apoptosis in fibrosarcoma cells (52) and stress-induced apoptosis in A431 cells (53). In the latter study, Src-dependent phosphorylation of EGFR led to ERK activation and the induction of survival signals in response to UV irradiation (53). Other investigators have implicated EGFR activation in keratinocyte survival by sustained MEK/MAPK signaling activation (54). Intriguingly, EGFR has been shown to transmit a survival signal to MAPK, even in the absence of EGFR kinase activity (55), implicating other kinases such as the Src family in this pathway. Nonetheless, the persistent phosphorylation of ERK1/2 in our studies was also associated with the inability of FAK inhibition to cause downregulation and dephosphorylation of AKT in EGFR-overexpressing cell lines. This suggests that EGFR also has a survival signal function through this serine/threonine kinase. Furthermore, this effect appears to be dependent on EGFR kinase activity, as AG1478 abrogated the protection of AKT. In fact, several studies have shown that EGFR can signal directly to AKT via phosphatidylinositol 3-kinase (42, 56 -58). In a model of oxidative stress-induced apoptosis, H 2 O 2 activates AKT through an EGFR/phosphatidylinositol 3-kinase-dependent pathway (42). Similarly, activation of EGFR signaling in T47D breast cancer cells and HEK293 cells protects these cells from Fas-induced apoptosis via an AKT-dependent mechanism (56).

FIG. 8. Endogenous overexpression of EGFR protects FAK from degradation and activates AKT and ERK1/2 survival pathways after FAK-CD expression reversed by the AG1478 inhibitor.
A, FAK and phospho-Ser 473 AKT levels 24 h after Ad-FAK-CD infection and AG1478 treatment. Western blotting with anti-EGFR, anti-AKT, and anti-FAK antibodies was performed as described in the legend to Fig. 5. The BT20 cell line did not degrade FAK at 24 h after Ad-FAK-CD infection. BT20 cells had highly active AKT and ERK1/2 after FAK-CD infection. The HA-tagged FAK-CD expression level was analyzed with anti-HA-antibody. B, FAK, phospho-Ser AKT, phospho-ERK1/2, and caspase-3 levels 72 h after Ad-FAK-CD infection and AG1478 treatment. BT20 cells had lower FAK, AKT (total and phospho-Ser 473 ), and phospho-ERK1/2 protein levels at 72 h of Ad-FAK-CD infection in AG1478-treated samples. Phospho-ERK1/2 and ERK1/2 samples were run on a parallel gel with FAK and AKT samples, and a composite image from the same film is shown for ERK samples. Protein levels were analyzed as described for A with the same antibodies. C, down-regulation of FAK and AKT in Ad-FAK-CD-and AG1478-treated BT20 cells is caspase-3 family-dependent. BT20 cells were pretreated first with the cell-permeable caspase-3 family inhibitor DEVD-CHO at 5 M for 1 h and then with AG1478 at 5 M and infected with Ad-FAK-CD. Ad-FAK-CD was added at the same viral dose as in Fig. 8 (A and B), but was 3.5-fold more concentrated (added in a 3.5-fold less volume of medium for better infectivity and faster apoptotic response). Caspase-3 and AG1478 inhibitors were present during adenoviral incubation, added fresh at the same doses every 24 h. Cells were collected at 55 h after adenoviral infection. Western blotting (WB) was performed with anti-caspase-3, anti-PARP, anti-FAK, and anti-AKT antibodies as described in the legend to Fig. 6A. Equivalent amounts of total protein from cell lysates were loaded on the gel. These observations have recently been extended to TRAILinduced apoptosis, whereby signaling from EGFR to AKT protects HEK293 and MDA-MB-231 cells from apoptosis by inhibiting mitochondrial cytochrome c release (57). In our system, the apoptosis induced by FAK down-regulation appeared to function through similar receptor-mediated apoptotic pathways, consistent with our previous studies (2). In addition, inhibition of the death complex with dominant-negative FADD inhibited degradation of AKT, independent of ERK activation, suggesting that the effects of FAK inhibition in these cells are mediated through a TNFR family-mediated, AKT-dependent pathway (Fig. 9). Furthermore, TNFR family FADD-dependent AKT cleavage was recently reported in Madin-Darby canine kidney epithelial cells (59). Ad-FAK-CD may induce death receptors by activating death domain-containing proteins such as Fas, TNFR, and DR/TRAIL or by affecting adapter proteins such as FADD, TRADD, and RIP. These events appeared to activate the downstream caspase-8 and -3 cascade with cleavage of important survival proteins such as AKT and FAK (Fig.  9). In EGFR-and FAK-positive cells, binding of EGFR to FAK may partially block FAK-CD-induced apoptosis, whereas EGFR itself can signal directly to AKT and ERK1/2. Furthermore, EGFR survival signaling was kinase-dependent, as inhibition of EGFR with AG1478 induced Ad-FAK-CD apoptosis (Fig. 9). In addition, the magnitude of these effects was cell type-specific, as breast cancer cells such as BT20 that express high levels of FAK, EGFR, and AKT were more resistant to inhibition of these pathways than BT474 cells, which express lower levels of these survival proteins.
Our model of survival signaling in the BT474-EGFR and BT20 breast cancer cells is that there are multiple cross-talking signaling pathways via ERK and AKT that augment the resistance of tumor cells to the apoptosis-promoting effects of tumor dissemination. The persistent signaling to AKT and MAPK pathways contributes to the resistance to apoptosis in these cells, which overexpress EGFR. From these studies, we conclude that FAK and EGFR cooperatively suppress apoptosis in breast cancer cells, suggesting that targeting both signaling pathways will have an enhanced apoptotic effect in breast cancers that overexpress these kinases.