Role of c-Src in Human MCF7 Breast Cancer Cell Tumorigenesis*

To study the role of c-Src in breast cancer tumorigenesis, we generated a cell line derived from MCF7 carrying an inducible dominant negative c-Src (c-SrcDN: K295M/Y527F) under tetracycline control (Tet-On system). c-SrcDN expression caused phenotypic changes, relocation of c-Src, Fak, and paxillin, and loss of correct actin fiber assembly. These alterations were coupled to increased Fak-Tyr397 autophosphorylation and to inhibition of Fak-Tyr925, p130CAS, and paxillin phosphorylation. An increased association of total Src with Fak and a decreased interaction of p130CAS and p85-PI3K with Fak were also observed. SrcDN inhibited cell attachment, spreading, and migration. Serum and EGF-induced stimulation of cell proliferation and Akt phosphorylation were also significantly reduced by SrcDN, whereas p27Kip1 expression was increased. Consistently, silencing c-Src expression by siRNA in MCF7 cells significantly reduced cell migration, attachment, spreading and proliferation. Inoculation of MCF7 cells carrying inducible SrcDN to nude mice generated tumors. However, doxycycline administration to mice significantly reduced tumorigenesis, and when doxycycline treatment was installed after tumor development, a significant tumor regression was observed. In both situations, inhibition of tumorigenesis was associated with decreased Ki67 staining and increased apoptosis in tumors. These data undoubtedly demonstrate the relevance of the Src/Fak complex in breast cancer tumorigenesis.

The Src family of non-receptor protein tyrosine kinases plays critical roles in a variety of cellular signal transduction pathways that are involved in cell division, motility, adhesion, angiogenesis, and survival (1).
Studies, mostly performed in colon cancer cells and fibro-blasts, revealed that the oncogenic potential of increased Src activity in tumor cells is pleiotropic involving control of cell proliferation and the regulation of cytoskeletal-linked events, such as ECM-adhesion, migration, spreading, and invasion (2)(3)(4). These biological effects are associated to Src substrates, among them, Fak 6 p130 CAS and paxillin, which are involved in integrin signaling and in adhesion dynamics (1,5). However, in breast cancer cells, the relevance and dispensability of each member of this complex is not fully understood, and it is still intriguing the importance of their kinase activity and their role as adaptor proteins to control the signaling pathways they trigger. A recent work in mouse embryo fibroblasts showed that Fak is dispensable for v-Src-induced transformation but would exert either positive or negative effects on signaling or motility depending on the pathways activated (6). The Src family of tyrosine kinases has a modular prototypic structure with a membrane-targeting signal, which is always myristoylated and sometimes palmitoylated, and Src homology domains SH2 and SH3, which facilitate protein-protein interactions. The kinase domain contains the ATP binding site Lys 295 (chicken nomenclature) and an autophosphorylation tyrosine residue (Tyr 416 ). The carboxyl tail contains Tyr 527 , which regulates Src kinase activity. Phosphorylation of Tyr 527 by CSK, favors intramolecular interaction with the SH2 domain, inducing an inactive conformation of the Src kinases (1,7). There are data in vivo and in tissue culture supporting the fact that Src kinases exert their biological actions not only through their enzymatic activity but also through their capacity to interact with other proteins via the SH2 and SH3 domains, generating complexes of signaling molecules involved in diverse signal transduction pathways (4, 8 -10). The combined mutations in the ATP binding site (K295M) and the regulatory C-terminal tyrosine residue (Y527F) generate a c-SrcDN mutant (K295M/Y527F) with an open conformation in which SH2 and SH3 domains may interact with other proteins but devoid of enzymatic activity; therefore, SrcDN is considered a dominant negative mutant of c-Src. Its expression was more efficient than c-SrcK-(K295M) in inhibiting prolactin-medi-ated proliferation of W53 lymphoid cells, by competing with endogenous activated Fyn and Lyn Src kinases (8).
c-Src is frequently overexpressed and/or aberrantly activated in a variety of epithelial cancers, including breast cancer (11)(12)(13)(14)(15). In addition, c-Src is associated with and activated by a variety of growth factor receptors involved in the growth and survival of human breast cancer cells (16 -22). Also, c-Src tyrosine kinase activity appears to be required for PyV middle T antigeninduced mammary tumorigenesis (23). Although these data support the importance of c-Src in breast cancer, the direct effect of c-Src functional inactivation on the transforming phenotype and tumorigenic potential of breast cancer cells has not been yet directly examined. Therefore, we generated MCF7 human breast cancer cells expressing SrcDN under the control of tetracycline (Tet-On system) in order to discriminate the specific role of the Fak and c-Src kinases, and to overcoming the drawback of Src inhibitors that could affect other signaling molecules (24,25). In MCF7 breast cancer cells, we observed that induction of SrcDN expression altered cell morphology, and caused a significant impairment of cell migration, adhesion, and spreading. These effects were associated with the delocalization and reduced tyrosine phosphorylation of critical proteins involved in integrin signaling and adhesion dynamics, such as Fak (Tyr 925 ), p130 CAS and paxillin. Expression of SrcDN also, significantly reduced cell proliferation in vitro, and led to a clear decrease in Akt phosphorylation and to a higher expression of the cell cycle inhibitor p27 Kip1 . Depletion of endogenous c-Src levels by means of siRNA had similar effects, showing the relevance of c-Src in breast cancer cells migration, adhesion, spreading and proliferation. Finally, expression of c-SrcDN reduced tumor growth in nude mice and more importantly, it was able to induce tumor regression. The antitumorigenic ability of induced SrcDN was associated with decreased proliferation, induction of apoptosis and reduction of angiogenic potential. Together these results provide insight on the mechanisms of the Src⅐Fak complex in the control of breast cancer cells proliferation, migration, spreading and survival, and the relevance of c-Src for the correct performance of this complex, pointing to this signaling protein as a potential target for therapeutic agents.
Generation of MCF7 Human Breast Cancer Cells Bearing a Doxycycline-inducible SrcDN-Clone 89rtTA expressing the rtTA (the reverse tetracycline transactivator) was obtained by transfection of MCF7 cells with pUHD17-1neo as previously described (26). Clone 89rTA cultured in complete medium (DMEM supplemented with 5% Tet-free FCS, 2 mM glutamine, 100 international units/ml penicillin, and 100 g/ml streptomycin), was maintained with 0.2 mg/ml G418. Cells were then cotransfected with SrcDN (avian c-Src K295M/Y527F) (8) cloned into pTET-Splice (Invitrogen) and pBABE-Puromycin by calcium phosphate method. Control cells were cotransfected with pTET-Splice (Invitrogen) and pBABE-Puromycin. Transfected cells were then selected with 2 g/ml puromycin. The resistant clones were expanded and tested for the induction of SrcDN expression upon addition of 2 g/ml Doxy by Western blotting with mAb EC10 to detect avian SrcDN. Four clones with very low basal SrcDN expression in the absence of Doxy, and high induction levels with Doxy were selected (1, 2, 29, and 30). For further experiments these clones were maintained in complete medium containing 0.2 mg/ml of G418 and 0.5 g/ml puromycin.
Generation of MCF7 Cells Expressing c-Src-siRNA-Duplexes for hairpin siRNA expression contained the unique 19-nucleotide sequence (AAACTCCCCTTGCTCATGTAC) that specifically targets the kinase domain of human c-Src (BLAST, NCBI Genome), in both sense and antisense orientation, separated by a 9-nucleotide (TTCAAGAGA) spacer sequence. The 5Ј-end corresponds to the BglII site, while the 3Ј-end contains the T5 sequence and HindIII corresponding nucleotides. The bacterial EGFP sequence was used as control (27). Forward and reverse oligonucleotides were annealed to produce the dsRNA, digested with BglII and HindIII and cloned into pSUPERIOR.puro (OligoEngine Inc., Seattle, WA), and correct insertion and insert sequence checked by sequencing. Stable transfection of MCF7 cells with the plasmids containing the siRNA for human c-Src or EGFP was carried out using jet-PEY TM (Polyplus Transfection, Ilkirch, France), according to the manufacturer's instructions. Clones were selected with puromycin (1 g/ml) and tested for c-Src expression by Western blotting using 327 mAb.
Cell Proliferation Assay-Cell proliferation was determined by staining the cells with crystal violet. Cells were plated at a density of 10 5 cells per well (24-well-plates) with complete medium in the presence or absence of Doxy. The next day medium was replaced for phenol red-free and serum-free DMEM supplemented with 10% serum or with EGF (100 ng/ml), in the presence or absence of 2 g/ml Doxy; 72 h later the number of cells per well was determined by crystal violet assay. Briefly, following cell fixation with glutaraldehyde 1%, cells were stained with crystal violet 1% and adhesion was quantified by washing the cells, eluting the dye with 1 ml/well of 10% acetic acid and measuring the absorbance at 600 nm. Cell proliferation was also determined by counting viable cells with trypan blue exclusion dye, as follows: cells were seeded (5 ϫ 10 5 /60-mm plate) in complete medium in the presence or absence of Doxy. After 16 h (t ϭ 0), the culture medium was changed for phenol red-free DMEM supplemented with 10% serum or with EGF (100 ng/ml), in the absence or presence of 2 g/ml Doxy, and 72 h later adherent cells were detached mixed with a 0.4% trypan blue solution (1:1) and loaded on a hemocytometer. The percentage of viable cells was calculated to determine cell growth.
Bromodeoxyuridine (BrdU) Pulse-label Experiments and Flow Cytometry Analysis-Cell cycle was analyzed by the BrdU/ anti-BrdU method. Briefly, cells were plated (1.5 ϫ 10 6 /90-mm plate) in serum-supplemented medium in the presence or absence of Doxy (2 g/ml) and incubated for 48 h. Afterward, cells were pulse-labeled for 30 min with 10 M of BrdU, washed twice with culture medium and incubated in fresh serum supplemented medium in the presence or absence of Doxy for different time periods. At given times, cells were detached labeled with fluorescein isothiocyanate (FITC)-conjugated-anti-BrdU (Becton Dickinson, San Jose, CA) and propidium iodide, and analyzed as previously described (8).
Cell Migration Assay-Cells were seeded at 5 ϫ 10 5 /60 mm, and grown in complete medium for 16 h. Then cells were incubated in the presence or absence of Doxy (2 g/ml) for 24 h to allow expression of SrcDN mutant. Subsequently, the cell monolayers were scarred with a sterile micropipette tip and incubated for another 24 h with or without Doxy. For each sample three defined areas were monitored during this period. The photographs were taken at the beginning of the assay (t ϭ 0 h) and 24 h later (t ϭ 24 h). Magnification, ϫ100.
Cell Adhesion and Spreading Assays-Cell adhesion and spreading assays were essentially performed as described previously (28). Briefly, 2 ϫ 10 5 cells, pretreated with or without Doxy for 48 h, were seeded in 96-well plates precoated with fibronectin (1 mg/ml) in serum-free DMEM in presence or absence of Doxy. Control cells were seeded in bovine serum albumin (1 mg/ml) precoated wells. After 2 h of incubation at 37°C cells were photographed (magnification, ϫ400) to determine cell spreading. Cell adhesion was quantified by washing twice with PBS to remove non-adherent cells, fixing the adhered cells and staining with crystal violet 1% and quantifying, as described previously (see cell proliferation assays). Spreading and adhesion were also determined in plastic wells. A similar method was used but with 2 ϫ 10 5 cells, pretreated with or without Doxy for 48 h. Cell spreading was quantified by counting round and spread cells in at least six fields (ϳ200 cells per field) per sample.
Immunofluorescence Microscopy and Confocal Analysis-Cells were plated at 1 ϫ 10 4 per chamber slide well, fixed, and permeabilized in 100% methanol for 10 min at Ϫ20°C (anti-Src 327, anti-Fak, anti-paxillin), or fixed with 2% formaldehyde in PBS and permeabilized in PBS-T (PBS containing 0.5% Triton X-100) at room temperature (phalloidin). Subsequently, cells were washed three times with PBS-T and blocked with 0.5% bovine serum albumin in PBS for 30 min at room temperature. Cells were incubated for 1 h at 37°C with 1:50 anti-Src 327, 1:50 anti-Fak (A17), 1:100 anti-paxillin, and 1:50 phalloidin. Coverslips were washed twice for 5 min with PBS-T, and then incubated with the corresponding secondary antibody for 1 h at 37°C, washed again, and mounted in Mowiol medium. Cells were examined using an Axiovert 200 inverted microscope (Zeiss, Gottingen, Germany) equipped with a color digital camera (Spot RT Slider) (Diagnostic Instruments, Sterling Heights, MI) with a 63/1.40 oil Plan-Apochromat objective. Confocal analysis was performed with a Leica TCS SPII Spectral microscope and 63X/1.3 NA oil objective.
Immunocytochemistry-Exponentially growing cells in complete medium in the presence or absence of Doxy during 48 h detached with trypsin, washed twice with complete medium to neutralize the protease, then washed twice with PBS to remove serum, and the final pellet of cells was fixed in formol. The fixed pellet was included in paraffin and 5-m sections were prepared. The paraffin sections were treated accordingly (29) for immunostaining with mAb 327 to detect total c-Src or with mAb EC10 to detect chicken c-SrcDN.
Evaluation of the Anti-tumorigenic Activity of the Src Double Mutant-Female athymic nude-Foxn1 nu mice 6 -8 weeks old, purchased from Harlan (Indianapolis, ID), were used. Mice were housed under pathogen-free conditions in a temperaturecontrolled room on 12/12 h light/dark schedule with food and water ad libitum. All procedures involving animals and their care were in accordance with national and international regulations. Each nude mice was implanted each with a 1.7 mg of 17␤-estradiol pellet (90 days release, Innovative Research of America) to supplement the estrogens requirement for MCF-7 proliferation. A week after implantation of 17␤-estradiol pellet, 5 ϫ 10 6 viable cells were subcutaneously injected in both back flanks to obtain solid tumors. Mice were divided into three experimental groups of 8 -9 animals. Group 1 (control) never received Doxy with the drinking water, group 2 received Doxy with the drinking water (2 mg/ml) from the first day after injection of MCF-7 Tet-on SrcDN cells, and group 3 received Doxy with the drinking water (2 mg/ml) 6 weeks after the injection of tumor cells. The bottles of water were changed every 3 days. The tumors were measured by vernier caliper, weekly, and their volumes were calculated using the formula LxWxHx0.52.
Immunohistochemistry of Xenografts-Tumors from untreated and Doxy-treated mice were removed after cervical dislocation twelve weeks after tumor cell implantation. Tumor specimens were frozen in liquid nitrogen and/or fixed in 4% formaldehyde solution immediately after surgical resection. Fixed samples were paraffin embedded (29), and 5-m sections were analyzed by hematoxylin/eosin staining, as described (29). Immunohistochemical staining for Ki67, c-Src, and avian c-Src was also performed. The number of Ki67 and total c-Src positive cells was estimated in six high-power fields containing 60 -80 cells/field (ϫ200). Results were classified as follows: 4 (Ն75%); 3 (50 -75%); 2 (25-50%); 1 (Յ25%); and 0 (no or weak staining in a few cells). Frozen samples were processed for Western blot analysis of PARP, total c-Src and avian Src expression.
Statistical Analyses-Analyses were performed with the SPSS/PC version 12.0X statistical package (SPSS, 2005, Chi-cago, IL). Comparisons between groups were performed by the Mann-Whitney U, Wilcoxon, Anova and chi 2 tests. When the expected frequency of any group was lower than 5, the Fisher exact test was used. All the tests applied were two-tailed and p Ͻ 0.05 was considered significant.

Inducible Doxycycline Expression of SrcDN in MCF7 Cells Results in Phe-
notypic Changes-Clone 89rtTA of MCF7 cells expressing the reverse tetracycline transactivator (rtTA) maintained in the presence of G418 (0.2 mg/ml) (26) was cotransfected with pTET-Splice-SrcDN and pBABE-puromycin. Cells maintained in complete medium were selected in the presence of G418 (0.2 mg/ml) with 2 g/ml of puromycin.
Resistant clones were expanded and tested by Western blotting for inducible expression of avian c-Src dominant negative mutant (SrcDN) upon addition of Doxy (0.2 and 2 g/ml) to cultures for 48 h. Using the mAb EC10, which only recognizes the avian c-Src mutant, but not the endogenous human c-Src expressed by MCF7 cells, four MCF7-Tet-On-SrcDN clones were selected for the following studies ( Fig. 1A). Total Src expression was also analyzed by Western blot using mAb 327 (Fig. 1A). Cells maintained in complete medium were treated with increasing concentrations of Doxy for 48 h to determine the dosedependent induction of SrcDN. The concentration of Doxy chosen to perform the following assays was 2 g/ml as the results showed an appropriate induction of SrcDN in all selected clones; higher concentrations of Doxy did not increase SrcDN expression (data not shown). We also determined the kinetics of SrcDN expression by collecting cells after incubation with Doxy (2 g/ml) for different periods of time, and observed maximal SrcDN expression between 48 and 72 h (data not shown). Next, we analyzed the expression of SrcDN in cells by immunocytochemistry (see "Experimental Procedures"). Consistent with results obtained by Western blot (Fig. 1A), expression of SrcDN was induced only by the addition of Doxy (Fig. 1B,  SrcDN (EC10)). In the presence of Doxy, staining of Total Src increased as a consequence of additional expression of SrcDN (Fig. 1B, Total-Src (327)). It is worth mentioning that  (1, 2, 29, and 30) were treated with 0, 0.2, and 2 g/ml of Doxy for 48 h, and expression of avian SrcDN (anti-c-Src mAb EC10) and total Src (327 mAb) was detected by Western blot. Membranes were reprobed with anti-␣-tubulin as a loading control. B, MCF7-Tet-On-SrcDN cells (clone 2) untreated or treated with Doxy (2 g/ml for 48h) were analyzed by immunocytochemistry with anti-Src EC10 to detect SrcDN and with anti c-Src 327 for total c-Src, as described under "Experimental Procedures." C, MCF7-Tet-On-SrcDN cells (clones 1, 2, 29, and 30), cultured in complete medium in the absence or presence of Doxy (2 g/ml) for 48 h were analyzed by phase contrast microscopy (magnification ϫ200).
SrcDN was mainly found at the plasma membrane where it is biologically functional (Fig. 1B).
We observed that SrcDN expression caused cells to become refractive and rounded as compared with cells grown in the absence of Doxy (Fig. 1C), and control cells (MCF7-Tet-On: Clone 89rtTA of MCF7 cotransfected with empty pTET-Splice and pBABE-puromycin; supplemental Fig. S1A). These results suggest that expression of SrcDN interferes with cell adhesion and with the correct assembly and dynamics of the cytoskeleton. The following experiments were carried out with all four clones obtaining similar results. The data presented here are referred to clone 2.
SrcDN Alters Cell Migration, Adhesion, and Spreading-To further evaluate the biological effects of SrcDN expression on MCF7-Tet-On-SrcDN cells, we next examined whether the expression of the c-Src mutant impaired cellular functions requiring intact integrin signaling and cytoskeleton machinery. For this purpose, we determined whether SrcDN expression altered cell migration, adhesion, and spreading.
Wound repair assays were performed to evaluate migration, as described under "Experimental Procedures." Cultures were pretreated with or without Doxy (2 g/ml) for 24 h and after wounding, cells were incubated in complete medium in the presence or absence of Doxy for another 24 h. MCF7 cells expressing SrcDN ( Fig. 2A, ϩDoxy) clearly showed a reduced capacity to migrate into the wounded area when compared with cells in which the mutant was not induced ( Fig. 2A, ϪDoxy). The wound repair assay was also performed with control cells incubated in the presence or absence of Doxy but no differences in migration were observed (supplementary Fig. S1B).
Tumor cell adhesion to extracellular matrix components such as fibronectin and the basement membrane has been shown to be an important step in cell migration and cell survival (30,31). We next examined whether the expression of SrcDN altered cell adhesion to fibronectin coated wells, and cell spreading (see "Experimental Procedures"). The expression of SrcDN caused a significant reduction in cell adhesion to fibronectin (42%, p Ͻ 0.01) as compared with Doxy-untreated cells (Fig.  2B). Similar results were obtained for direct adhesion to plastic (data not shown). Additionally, the expression of SrcDN inhibited cell spreading ( Fig. 2C; 94.06%, p Ͻ 0.01). Cell adhesion and spreading was not affected when the control cells were incubated in presence of Doxy (supplementary Fig. S1, C and D). SrcDN Induces Relocation of Fak, Paxillin, and c-Src-As described above, SrcDN expression caused morphological changes as well as inhibition of migration, adhesion and spreading. Therefore, we next analyzed if it also altered the cellular distribution of focal adhesion molecules involved in the control of such processes (4,32). Fluorescence microscopy analysis showed that c-Src, Fak, and paxillin in the absence of Doxy were mainly located at focal contacts homogeneously distributed along the membrane of spread cells. Phalloidin staining revealed the normal assembly of cytoskeleton actin stress fibers (Fig. 3, ϪDoxy). Interestingly, expression of SrcDN by addition of Doxy to cell cultures caused drastic changes in the distribution of these proteins (Fig. 3,  ϩDoxy). c-Src appeared evenly expressed at the plasma membrane of the cell, and the normal actin fiber assembly was disrupted. Fak and paxillin accumulated in patches, indicating large peripheral adhesions. The pattern of distribution of these proteins was unaltered when control cells were treated with Doxy (supplemental Fig. S2).
To further analyze the proteins involved in the assembly of large focal adhesions found in cells expressing SrcDN, colocalization analyses of c-Src, Fak, and paxillin staining were performed by confocal microscopy. At the basal cell level SrcDN overexpression caused relocation of c-Src, Fak, and paxillin. However, it did not alter the colocalization of these proteins at focal contacts, which concentrated into large patches at the cell surface (Fig. 4).
SrcDN Impairs Activation and Downstream Effects of the Src⅐Fak Complex-The c-Src⅐Fak complex controls different signaling pathways triggered by growth factors or by integrins (32), which in turn regulate proliferation, survival, migration and cell adhesion. Because expression of c-Src mutant impaired cell migration, adhesion, and spreading, we further analyzed whether it also affected the activation of the Src⅐Fak complex and its downstream substrates.
Growth factor or integrin stimulation induces Fak autophosphorylation on Tyr 397 , generating a high affinity binding site for Src SH2 domain (33). Thus, we then determined the extent of Fak-Tyr 397 phosphorylation in exponentially growing cells in the presence or absence of Doxy (2 g/ml) for 48 h. Interestingly we observed that expression of SrcDN increased the degree of Fak-Tyr 397 phosphorylation (Fig. 5A). A possible explanation for this result could be that the overexpressed SrcDN competes and displaces endogenous c-Src from its interaction with P-Tyr 397 -Fak, thus, protecting it from dephosphorylation. Consistently, the amount of total Src bound to Fak-Tyr 397 increased in cells treated with Doxy (Fig. 5D, IP Fak/WB c-Src), suggesting that the observed increased phosphorylation of Fak-Tyr 397 in Doxytreated cells could be caused by SrcDN SH2 domain protection from dephosphorylation. Alternatively, the binding of SrcDN may induce a greater percentage of Fak molecules to autophosphorylate, thereby increasing the Fak-Tyr 397 signal.
Once the association of Src with Fak has occurred, Src suffers a conformational change leading to Src activation and further phosphorylation of Fak on Tyr residues, which is required for full activation of Fak. Among these residues is Tyr 925 , which is exclusively phosphorylated by Src (4). To determine if Fak was activated in SrcDN overexpressing cells, the phosphorylation of Fak on Tyr 925 was assessed. As expected, Fak-Tyr 925 phosphorylation was diminished in Doxy-treated cells (Fig. 5A). Because the activated Fak⅐Src complex recruits and phosphorylates proteins involved in cell adhesion, actin dynamics, proliferation, and survival, we determined the extent of paxillin and p130 CAS phosphorylation in exponentially growing cells in the absence or presence of Doxy as before. pY-paxillin and pY-p130 CAS were clearly diminished upon addition of Doxy to cultures (Fig. 5, B and C). Likewise, the amount of p130 CAS associated with Fak was drastically decreased (Fig. 5D, IP Fak/WB p130 CAS ). Activation of the Src⅐Fak complex also causes association and activation of PI3K (34,35). We then analyzed whether SrcDN expression altered the association of p85-PI3K with Fak. To this end, Fak was immunoprecipitated from exponentially growing cells in the absence or presence of Doxy (2 g/ml) for 48 h, and the presence of p85-PI3K in Fak immune complex was determined by Western blot. The results showed that in the absence of Doxy p85-PI3K was associated with Fak (Fig. 5D, IP Fak/WB p85). However, expression of SrcDN abolished this interaction (Fig. 5D, IP Fak/WB p85).
Together, these results indicate that overexpression of SrcDN seems to be enough to inactivate the Src⅐Fak complex, which in turn is involved in the control of cell migration, adhesion, and spreading of breast cancer cells.
Reduction of Cell Proliferation Rate by SrcDN-c-Src expression and function has been reported as an important mediator of proliferation for a variety of cancer cells (1,36); however, it is not involved in colon cancer cells proliferation (4), albeit c-Src overexpression in these tumors (11). We have analyzed whether overexpression of the dominant negative form of c-Src in MCF7 breast cancer cells could alter serum-and EGFstimulated proliferation in vitro. For this purpose serum-starved and exponentially growing cells in serum or EGF-supplemented medium were incubated in the absence or presence of Doxy (2 g/ml) for 72 h. Afterward, cell proliferation was assessed (Fig.  6A). The induction of SrcDN expression significantly reduced the proliferation rate of serum-starved and exponentially growing cells in serum as well as in the EGF supplemented medium (Fig. 6A, p Ͻ 0.05). No significant differences in proliferation rate were found when control cells were incubated in the presence or absence of Doxy (supplemental Fig. S1E).
To define the effects of SrcDN expression on cell cycle progression, BrdU pulse-label experiments were performed in untreated and Doxytreated cells. In the absence of Doxy, cells progressed through the cycle, and after 36 h, the BrdU-labeled cells showed tendency toward to homogeneous distribution in G 1 , S, and G 2 /M phases (Fig. 6B). In contrast, SrcDN expression caused cell accumulation in the G 1 phase and decreased BrdUlabeled cells in S and G 2 ϩM (Fig. 6B). The ratio between the percentages of untreated and Doxy-treated cells in each phase at each time was calculated (Fig. 6B, ratio Ϯ Doxy) and clearly showed that SrcDN caused cells to accumulate in the G 1 phase, whereas cells in S and G 2 /M diminished with time (Fig. 6B, ratio Ϯ Doxy). These results indicate that c-Src is involved in the control of G 1 /S transition in exponentially growing breast cancer cells.
SrcDN Reduces Serum and EGF-mediated Activation of Erk1/2 and Akt, and Up-regulates p27 Kip1 Expression-Once we demonstrated the relevance of active c-Src for breast cancer cell proliferation in vitro, we analyzed its role in signaling pathways leading to cell proliferation and survival. Serum-starved cells were incubated in the absence or presence of Doxy (2 g/ml) for 48 h. Cells were stimulated with serum (10%) or with EGF (100 ng/ml) for different time periods, and activation of Erk1/2 and Akt and p27 Kip1 expression were determined by Western blot. Both serum and EGF activated Erk1/2 reaching maximal phosphorylation after 15 min and declining toward basal levels after 1 h (Fig. 6C). Expression of SrcDN caused a slight reduction of serum and EGF-mediated Erk1/2 activation at 15 min but produced a marked decrease at later times (Fig.  6C), suggesting that SrcDN exerts its greatest effect at late stages of mitogen-activated protein kinase signaling. Additionally, stimulation of Akt by serum or by EGF was drastically diminished in Doxy-treated cells (Fig. 6C). This result is consistent with the reduced association of p85-PI3K to the Fak⅐Src complex previously observed (Fig. 5D). In agreement with the reduced rate of cell proliferation observed upon expression of SrcDN (Fig. 6A), the expression of cyclin-Cdk inhibitor p27 Kip1 was highly increased in both serum and EGF-stimulated cells (Fig. 6C).
Reversibility of the Effects of Inducible Expression of SrcDN in MCF7 Cells-To test whether the effects of SrcDN expression could be reverted, we first analyzed the expression of total Src and SrcDN by Western blot in exponentially growing cultures of MCF7-Tet-On-SrcDN in complete medium in the absence or presence of Doxy (2 g/ml) for 48 h versus 24 h depletion of Doxy from the 48 h Doxy-induced cells. As observed in Fig. 7A, SrcDN induction was observed as before (Fig. 1A) after 48 h of Doxy induction. However, when cultures were washed with prewarmed culture medium and then incubated 24 h in the absence of Doxy SrcDN expression was abrogated (Fig. 7A). Consistent with the reversible expression of SrcDN, the refractive and rounded phenotype of cells upon induction of SrcDN recovered their normal epithelial phenotype when SrcDN was no longer expressed (Fig. 7B). Moreover, the signaling events triggered by SrcDN expression also reverted 24 h after the removal of Doxy. Thus, Doxy depletion reduced Fak phosphorylation at Tyr 397 and increased Tyr 925 phosphorylation to levels of non-induced cells (Fig. 7C).
Effect of c-Src-siRNA in MCF7 Cell Migration, Attachment, Spreading, and Proliferation-MCF-7 cells constitutively expressing c-Src-siRNA were generated to complement the studies performed with the SrcDN inducible mutant. EGFP-siRNA was used as control because it does not affect biological responses of eukaryotic cells (27,37). First, we analyzed the effect of c-Src-siRNA on endogenous c-Src expression. As compared with control cells expressing EGFP-siRNA, c-Src-siRNA expression in MCF7 cells strongly reduced the endogenous c-Src levels (Fig. 8A). Then we analyzed the biological effects of c-Src-siRNA expression in serum-stimulated versus EGFP-siRNA expressing cells. MCF7 cells expressing c-Src-siRNA showed a significant reduction of migration, attachment, spreading, and proliferation as compared with control cells (Fig. 8, B-E, p Ͻ 0.05), supporting the important role of c-Src in these biological responses.
Role of c-Src in Tumorigenesis-Because the anti-proliferative action of SrcDN mutant in vitro has been clearly established (Fig. 6), we next examined its inhibitory action in vivo. For this purpose, we analyzed whether overexpression of SrcDN under Doxy induction could alter the tumorigenic potential of MCF7 carcinoma cells when inoculated in female athymic nude-Foxn1 nu mice. Because MCF7 cells are estrogen dependent for growth in vivo, mice were implanted with a 17␤-estradiol pellet (90 days release) 1 week before cell injection. Animals were separated into three different groups: group 1 (never received Doxy in the drinking water; control group), group 2 (receiving 2 mg/ml of Doxy in the drinking water from the first day after implantation of cells) and group 3 (maintained without Doxy for the first 6 weeks, and then treated with 2 mg/ml of Doxy in the drinking water for the following 6 weeks). The experiment was carried out for 12 weeks (the period of time that estrogen pellets are active), and tumor size was measured weekly. At the end, tumors were excised and processed for biochemical and histological analyses.
Although there was not a time delay in the appearance of tumors between groups 1 and 2, the number of tumors developed in the group of mice receiving Doxy (group 2) was significantly smaller than in group 1 (control group, Fig. 9, A and B, *, p Ͻ 0.05). We next analyzed whether SrcDN expression could cause regression of tumors once they were established. To this end, tumors from groups 1 and 3 were compared after 6 and 12 weeks. The number of tumors within group 1 at 12 weeks was slightly increased when compared with the values at 6 weeks, although there were not significant differences (Fig. 9, A and B). However, there was a significant reduction in tumor number when comparing values at 12 versus 6 weeks in group 3 (Fig. 9, A and B, #, p Ͻ 0.05). Consistent with the inhibition of tumorigenesis and tumor regression, the c-Src expression in tumors was determined by immunohistochemical analysis (Fig. 9C), and the quantifications showed a significant increase in groups 2 and 3 because of the SrcDN expression when compared with group 1 (Fig.  9B,  †, p Ͻ 0.05). The immunohistochemical analysis of Src revealed the expected pattern of staining, which was restricted to the plasma membrane of inoculated breast cancer cells (Fig. 9C). This distribution of c-Src was consistent with immunocytochemistry of in vitro growing cells in presence or absence of Doxy (Fig. 1B). The inhibition of tumorigenesis was associated with a significant decrease in the proliferation of tumors belonging to groups 2 and 3 versus group 1, as revealed by immunohistochemical detection of Ki67 in tumors (Fig. 9, B, †, p Ͻ 0.05, and C). Concomitantly, apoptosis was significantly increased in SrcDN-expressing tumors (groups 2 and 3), as determined by cleavage of native poly (ADP-ribose) polymerase (PARP) p116 into PARP p85 in tumor biopsies by Western blot (Fig. 9D). The p116/p85 ratio calculated from the Western blot showed a significant decrease in the tumors of mice treated with Doxy (Fig. 9B, *, p Ͻ 0.05), indicating a higher degree of apoptosis upon expression of SrcDN. p53 levels were also determined by Western blot in tumors extracts and found clearly increased in mice belonging to groups 2 and 3 as compared with group 1. Given that p130 CAS cleavage has been associated with anoikis and suppression of tumor growth (38), to further elucidate the apoptosis mechanism, p130 CAS degradation was determined by Western blot in tumor extracts. In groups 2 and 3, a clear increment of a lower molecular mass protein (ϳ83 kDa) immunoreactive to the p130 CAS antibody was observed when compared with FIGURE 6. Role of SrcDN in serum and EGF-mediated cell proliferation, Erk1/2 and Akt activation, and p27 Kip1 expression. A, serum-starved and exponentially growing cells in serum or EGF-supplemented medium were incubated in the absence or presence of Doxy (2 g/ml) for 72 h. Cell proliferation was assessed by the crystal violet method (see "Experimental Procedures"). The percentage of cell growth was calculated considering the number of cells growing in serum-supplemented medium in the absence of Doxy for 72 h as 100%. B, cells were incubated in serum-supplemented medium in the presence or absence of Doxy for 48 h and then pulse-labeled for 30 min with 10 M BrdU, washed, incubated in fresh medium and detached at the indicated times. Afterward, cells were labeled with FITC-anti-BrdU and PI and then analyzed by flow cytometry (see "Experimental Procedures"). The results are expressed as percentage of BrdU-labeled cells in the G 1 , S, and G 2 ϩM phases of the cell cycle, as defined by PI label. The ratio between percentages of untreated and Doxytreated cells in each phase at given times is depicted in the right panel. This is a representative experiment of four independent tests. *, p Ͻ 0.05 (ANOVA test). C, serum-starved cultures were maintained in the absence or presence of Doxy for 48 h before stimulation with 10% serum or 100 ng/ml EGF for the indicated times. Activation of Erk1/2 and Akt was determined by Western blot with 25 g of protein lysates using specific phosphoantibodies. Membranes were reblotted with anti-Erk2 or anti-Akt for loading control. p27 Kip1 expression was detected by Western blot, and the membrane was reblotted with anti-␣-tubulin for loading control. group 1 tumors, which showed almost undetectable levels of the 83 kDa band. Because Src and p53 have been involved in VEGF expression in human tumor cells (39,40) we determined its expression in tumors extracts by Western blot. Results showed that levels of VEGF diminished in groups 2 and 3 tumors expressing SrcDN in contrast to control tumors of group 1 (Fig. 9, B and D).
Together these results showed that expression of SrcDN is able to reduce tumor development and to induce tumor regression once they were established. These effects are associated with increased apoptosis in tumors expressing SrcDN and with a reduction in cell proliferation.

DISCUSSION
Because the inherited or acquired deregulation of protein kinase activity has been implicated in the pathogenesis of many human diseases, including cancer, the inhibition of these enzymes has been postulated as a promising strategy for anticancer treatment, among them is the non-receptor tyrosine kinase c-Src (36,41,42).
c-Src controls pathways involved in cellular functions as diverse as cell growth, migration, adhesion, and survival (1) and has been found to be overexpressed or constitutively active in a large percentage of tumors, including breast and colon cancers (12,13,43). In the present study we have analyzed the effects that the inducible expression of a c-SrcDN (K295M/Y527F) has on the MCF7 human breast cancer cell line. The inducible c-SrcDN has no enzymatic activity but has a constitutively open conformation where domains SH2 and SH3 are accessible to interact with other proteins, and we have previously shown that it competes and blocks activation of other endogenously expressed members of the Src family when expressed in lymphoid cells (8).
According with several studies mainly performed with colon cancer cells and fibroblasts, the interplay between c-Src and Fak is essential for the control of diverse cellular functions (3,4,32,33,44). The initial interaction between c-Src and Fak occurs after Fak autophosphorylation on Tyr 397 . Autophosphorylated Fak binds the c-Src SH2 domain by P-Tyr 397 and subsequently, the Src⅐Fak complex phosphorylates and activates a plethora of substrates (32,33,44). To address the relevance of c-Src activity on the correct performance of this complex in MCF7 breast cancer cells, we have used SrcDN, which is able to interact with P-Tyr 397 -Fak but unable to phosphorylate Fak or Src⅐Fak complex substrates.
The induction of SrcDN in MCF7 led to morphological changes suggesting an aberrant function of focal adhesion assembly and cytoskeleton dynamics. When we further analyzed the effects of SrcDN expression on cellular functions that depend on the integrin signaling pathways and cytoskeleton machinery, such as cell migration, adhesion and spreading, we found them clearly impaired upon induction of the mutant. These alterations were also observed in MCF7 cells when c-Src was knocked-down by siRNA.
The morphological changes induced by SrcDN expression were associated with relocation of c-Src, Fak, and paxillin as well as with disruption of normal actin fiber assembly. SrcDN expression did not alter c-Src, Fak and paxillin co-localization but caused their rearrangement into large focal adhesions. This assembly of large, peripherally localized adhesions has been previously observed in fibroblasts. Fibroblasts from Fak Ϫ/Ϫ mice show decreased rate of migration and spreading, and increased size of peripherally localized adhesions (45). Also, the expression of Src kinase defective mutants in normal fibroblasts results in large peripheral adhesions (46). It has been previously suggested that increased number and size of adhesions in Fak Ϫ/Ϫ and Src Ϫ/Ϫ cells could be caused by an inhibition of adhesion turnover (45). An important role of the Fak⅐Src complex in the regulation of the rate of focal adhesion assembly and disassembly has been postulated. In fact, a recent study in fibroblasts has shown that Fak, Src, and their downstream substrates p130 CAS , paxillin, Erk1/2, and MLCK are required for adhesion disassembly (32). When we analyzed if the aberrant assembly of focal adhesions in SrcDN expressing MCF7 cells was linked to effects on activation of Src⅐Fak complex and its downstream substrates, we found increased Fak autophosphorylation on Tyr 397 associated with augmented co-immunoprecipitation of Src. Increased phosphorylation of Fak on Tyr 397 has also been recently observed in KM12C colon carcinoma cells expressing Src inactive mutants (4). In agreement with these authors, we suggest that the expressed SrcDN, via its SH2 domain, may stabilize autophosphorylated Fak protecting it from dephosphorylation. On the other hand, when we analyzed Fak phosphorylation on residue Tyr 925 , which is exclusively catalyzed by Src and required for adhesion turnover (4), we found it diminished in SrcDN-expressing cells. Also, phosphorylation of p130 CAS and paxillin, essential for adhesion assembly/disassembly and actin dynamics (32), was diminished in SrcDN expressing MCF7 cells. From these results, we conclude that expression of the SrcDN mutant, which is able to interact with P-Tyr 397 -Fak but is catalytically inactive, alters the Fak⅐Src complex control of cell adhesion, spreading and migration. The inactivation of the Fak⅐Src complex in SrcDN-expressing cells could be explained by SrcDN kidnapping of autophosphorylated Fak and subsequent inability to further phosphorylate Fak and Src⅐Fak substrates.
Expression of SrcDN was able not only to impair oncogenic properties of breast cancer cells such as migration, adhesion, and spreading resembling the effects of this FIGURE 8. Effect of c-Src-siRNA in the migration, adhesion, spreading, and proliferation of MCF7 cells. A, expression of c-Src was determined by Western blot with 25 g of protein lysates from exponentially growing MCF7 cells expressing c-Src-siRNA or EGFP-siRNA using 327 mAb. Membrane was reblotted with anti-␣-tubulin for loading control. B, cultures of serum-starved MCF7 cells expressing c-Src-siRNA or EGFP-siRNA were stimulated with 10% FCS for 24, 48, and 72 h. Cell proliferation was assessed as described under "Experimental Procedures." The percentage cell growth was calculated considering the number of EGFP-siRNA-expressing cells at 72 h as 100%. C, cell migration was determined by wound repair assays. MCF7 cells expressing c-Src-siRNA or EGFP-siRNA were seeded in complete medium and 24 h later the cell monolayers were scarred, recording the wound size (0 h); subsequently, cells were incubated for 24 h (24 h). Cell migration into the wound was analyzed by phase-contrast microscopy (magnification ϫ200). D, cell spreading was analyzed seeding c-Src-siRNA or EGFP-siRNA expressing MCF7 cells in 96-well plates, incubated for 2 h of incubation at 37°C, and the percentage of cell spreading was calculated as described under "Experimental Procedures." E, cell adhesion was analyzed seeding c-Src-siRNA or EGFP-siRNA expressing MCF7 cells in 96-well plates precoated with fibronectin (1 mg/ml), incubated for 2 h incubation at 37°C, and the percentage of adherent cells was calculated as described under "Experimental Procedures." Four independent experiments were performed in triplicates. **, p Ͻ 0,01 as compared with the control (ANOVA test). mutant on colon cancer cells (4), but it was also efficient to reduce cell proliferation in vitro and in vivo, which was not observed in colon cancer cells (4). Our data show that induction of SrcDN expression significantly reduced the proliferation rate of serum-starved and exponentially growing MCF-7 cells in serum and in EGF-supplemented medium, causing accumulation of cells in G 1 phase of the cell cycle. The inhibition of proliferation was associated with reduced activation of Akt after serum or EGF treatment and a marked increase in p27 kip1 expression. These data are in agreement with our previously reported results on W53 lymphoid cells cultivated in the presence of Src inhibitor PP1 (8). The relevance of c-Src in proliferation of breast cancer cells was confirmed by depletion of this oncoprotein by means of siRNA expression.
The inhibition of Akt activation by SrcDN described here, is in agreement with the role of Src kinases in mediating Akt stimulation observed in several experimental models (8,16,(47)(48)(49). The impairment of the Fak⅐Src complex correct functioning could be involved in the mechanism by which SrcDN inhibited serum and EGF-stimulated Akt phosphorylation. The p85-PI3K subunit has been shown to bind to phosphorylated Tyr 397 -Fak (50) activating PI3K and subsequently Akt. However, we observed that Fak interaction with p85-PI3K was inhibited, most probably by competition with overexpressed SrcDN. Serum and EGF induced a sustained activation of Erk1/2. In contrast, expression of SrcDN caused a rapid decrease on Erk1/2 activation by these stimuli. The role of Src kinases on Erk1/2 activation by growth factors depends not only on the mitogen used but also on the cell type studied. In prolactin stimulated MCF7 or T47D cells, Erk1/2 activation requires Src kinase activity (16), while in lymphoid W53 cells occurs even in the presence of Src kinase inhibitor PP1 (8). Nevertheless, in both cases, Erk1/2 activation by prolactin is needed for cell proliferation. Similarly, c-Src mediates activation of Erk1/2 that is required for estrogen dependent MCF7 proliferation (51), while in EGF stimulated MCF10A cells Erk5 activation is essential for proliferation, but stimulation of Erk1/2 does not play a major role (52).
The increased expression of p27 kip1 upon induction of SrcDN is consistent with Akt inhibition and the reduction of cell proliferation. These results are in agreement with those obtained in TGF-␤1 treated hepatoma cells, where inhibition of c-Src by PP2 augmented levels of p27 kip1 (53). Also, in ErbB2-overexpressing breast cancer cells, Herceptin causes increased expression of p27 kip1 by inhibiting PI3K and Akt (54). These effects can be linked to the inhibition of association and activation of c-Src with ErbB2 (22).
Proliferation of cancer cells in vitro does not always correlate with their tumorigenic potential. We therefore tested the antitumorigenic capacity of SrcDN in MCF7 cells inoculated in estrogen-primed nude mice. Our results showed that SrcDN expression significantly diminished the number of tumors and, even more, induced their regression once they were established. Interestingly, whereas treatment of cells growing in vitro for FIGURE 9. Effect of conditional expression of SrcDN in tumorigenesis and tumor regression. A, the number of tumors per mice in each group was measured weekly. Inhibition of tumorigenesis was evaluated by comparison between groups 1 and 2 and tumor regression by comparing groups 1 and 3 (data quantification is shown in B). C, representative immunohistochemistry analysis of tumors from each group detecting total c-Src (mAb 327) and Ki67 expression at the end of the experiment and quantified (see B) as described under "Experimental Procedures." D, PARP and p130 CAS cleavage, and p53 and VEGF expression analyzed by Western blot in homogenized tumor tissue. PARP p116 and p85, and p130 CAS p130 and p83 expression were quantified by densitometry and expressed as PARP p116/p85 and p130 CAS p130/p83 ratios; p53 and VEGF protein was quantified and normalized with tubulin as loading control (data are shown in B). Statistical significances were calculated as described under "Experimental Procedures." #, p Ͻ 0.05 (Wilcoxon); *, p Ͻ 0.05 (U-Mann Whitney); and †, p Ͻ 0.05 (Pearson chi 2 ).
48 -72 h with Doxy caused a clear reduction of proliferation rate, apoptosis was not observed, although inhibition of the PI3K-Akt signaling pathway may precede apoptotic events that would take place later on. In contrast, the induction of SrcDN in growing breast cancer cells in vivo led not only to reduction of their proliferation, but also to increased apoptosis, as shown by increased p53 expression and PARP and p130 CAS degradation. p130 CAS cleavage has been recently related to anoikis and suppression of tumor growth upon induction of Fak-Y397F in Fak Ϫ/Ϫ fibroblasts transformed by v-Src (38). The involvement of p130 CAS cleavage in anoikis is also supported by other studies (55)(56)(57)(58)(59)(60).
In addition with PARP and p130 CAS cleavage, and increased p53, we also observed decreased VEGF levels in SrcDN-expressing tumors. These effects are associated with augmented apoptosis and diminished tumorigenesis. Consistently with these results it has been previously shown, in other human tumor cells, that v-Src and p53 exert opposed effects on VEGF expression, whereas v-Src increases VEGF, p53 decreases it (61).
In conclusion, the data presented in this manuscript highlight the relevance of c-Src in Src⅐Fak complex control, which in turn regulates essential cellular processes, like migration, adhesion, spreading, and cell growth. In fact, expression of a c-Src mutant (SrcDN), which favors the interaction with Fak but has no kinase activity, is enough to impair cellular functions required for breast cancer growth and even to cause significant tumor regression. Additionally, we have studied the role of c-Src by silencing its expression with siRNA and obtaining similar effects to those induced by SrcDN expression. These data demonstrate the key involvement of c-Src in the biology of breast cancer cells, suggesting that tumors may become addicted to Src function. Thus, we consider that c-Src is a potential target for breast cancer therapy.