Normal murine NMuMG and metastatic 4T1 cells were obtained from ATCC (Manassas, VA) and cultured as described previously 18. 4T1 cells were engineered to express stably firefly luciferase by their transfection with pNifty-CMV-luciferase 20 and selection with Zeocin (500 μg/ml; Invitrogen, Carlsbad, CA). NMuMG cells expressing WT or the nonfunctional mutant D119A-β₃ integrin were constructed by retroviral transduction, as described previously 19. The MCF10A cell derivates T1k, Ca1h, and Ca1a were cultured in DMEM/F12 supplemented with 5% horse serum. The construction of NMuMG and 4T1 cells lacking FAK was accomplished by their lentiviral-mediated transduction with a scrambled (i.e., nonsilencing shRNA) or verified FAK-specific shRNA sequence encoded in pLentilox3.7-puro 15. In brief, human 293T cells were transiently transfected with lentiviral packaging vectors (i.e., pMD2.G, pRRE, pRSV, and pLentiLox 3.7) according to standard protocols 21, and 48 h after transfection, the resulting conditioned medium was collected, filtered, and mixed with polybrene (8 μg/ml). Target cells were incubated in viral-containing supernatants for 48 h, and cells expressing nonsilencing or FAK-specific shRNAs were isolated by puromycin selection (5 μg/ml) for 14 days. Afterward, puromycin-resistant NMuMG and 4T1 cells were assayed for FAK-deficiency by immunoblotting with anti-FAK antibodies, as described later.

For FAK immunoblots, β₃ integrin-expressing NMuMG cells were trypsinized, pelleted, and maintained in a nonadherent state for 4 h in serum-reduced media (0.5%). Afterward, the cells either were immediately harvested or were replated in the absence or presence of TGF-β₁ (5 ng/ml) for an additional 4 h, at which point they were harvested to detect differences in FAK phosphorylation and expression by immunoblotting (see later). Whole-cell extracts prepared from normal and malignant MECs were immunoprecipitated with antibodies against TβR-II, Grb2, and β₃ integrin (Santa Cruz Biotechnology, Santa Cruz, CA), and the resulting immunocomplexes were immunoblotted various antibodies listed later 19. Where specified, 4T1 cells were incubated for 18 h in the absence or presence of the FAK inhibitors, PF-562271 or PF-573228 (Pfizer, Inc., New York, NY) at the indicated concentrations before immunoprecipitation of β₃ integrin. NMuMG cells also were incubated in serum-reduced media (0.5% FBS) supplemented with TGF-β₁ (5 ng/ml) for 18 h in the absence or presence of the Src inhibitor, PP2 (10 μg/ml; Calbiochem, Temecula, CA). For all cell-signaling experiments, 4T1 or NMuMG cells were serum starved (0 FBS) or deprived (0.5% FBS), respectively, for 6 h before TGF-β₁ stimulation (5 ng/ml). Control and FAK-deficient 4T1 cells were incubated for up to 48 hours with TGF-β₁ (5 ng/ml) and detergent-solubilized whole-cell extracts were prepared and immunoblotted for E-cadherin (E-Cad; BD Biosciences, San Jose, CA). Last, 4T1 cells were incubated with TGF-β₁ (5 ng/ml) for 24 h in serum-free medium, and the resulting conditioned medium was precipitated with 0.01% sodium deoxycholate/6.25% trichloracetic acid and immunoblotted for plasminogen activator inhibitor-1 (PAI-1; Santa Cruz Biotechnology).

Cell extracts were prepared by harvesting NMuMG and 4T1 cells on ice in 3-D RIPA buffer (50 mM Tris, 150 mM NaCl, 6 mM sodium deoxycholate, 1.0% NP-40, 0.1% SDS, pH 7.4) supplemented with protease inhibitor cocktail (Sigma, St. Louis, MO) and phosphatase inhibitors (10 mM sodium orthovanadate, 40 mM β-glycerophosphate, 20 mM NaF), and subsequently were clarified by microcentrifugation before immunoblotting with the following primary antibodies (dilution): (a) anti-phospho-Y397-FAK (1:1000; Cell Signaling, Danvers, MA); (b) anti-phospho-Y577-FAK (1:1000; Cell Signaling); (c) anti-phospho-Y925-FAK (1:1,000; Cell Signaling); (d) anti-phospho-p38 MAPK (1:500; Cell Signaling); (e) anti-phospho-Smad2 (1:1,000; Cell Signaling); (f) anti-phospho-Smad3 (1:500; Cell Signaling); (g) anti-E-cadherin (1:2,500 BD Biosciences); (h) anti-PAI-1 (1:1,000; Santa Cruz Biotechnology); (i) anti-FAK (1:1,000; Santa Cruz Biotechnology); (j) anti-β-actin (1:1,000; Santa Cruz Biotechnology); (k) anti-p38 MAPK (1:1,000; Santa Cruz Biotechnology); (l) anti-Smad2/3 (1:1,000; Cell signaling); (m) anti-phospho-Y416-Src (1:1,000; Cell Signaling); and (n) anti-Src (1:1,000, Cell Signaling).

Confluent NMuMG cell cultures were wounded with a micropipette tip (200 μl) and immediately placed in 1% serum-containing medium supplemented with or without TGF-β₁ (5 ng/ml) or the TβR-I inhibitor, SB-431542 (10 μM; Sigma). Bright-field images of wounded monolayers were obtained immediately after wounding (T₀) and at various times thereafter as indicated. The extent of wound closure was quantified by obtaining three wound measurements for each of three random fields (x100) per wound, and all wound conditions were performed in triplicate. Measurements were taken by using the SlideBook Imaging Software (Intelligent Imaging Innovations, Inc., Denver, CO). The ability of TGF-β₁ (5 ng/ml) to alter the invasion of 4T1 cells (50,000 cells/well) was analyzed by using a modified Boyden Chamber assay, as described previously 19.

Alterations in gene expression regulated by TGF-β were assessed by using a reporter gene assay that monitored changes in luciferase expression driven by the synthetic SBE (Smad-binding element) promoter, as described previously 22. In brief, NMuMG cells (25,000 to 30,000 cells/well) were allowed to adhere overnight to 24-well plates. The following morning, the cells were transiently transfected by overnight exposure to LT1-liposomes (Mirus, Madison, WI) that contained 300 ng/well of pSBE-luciferase cDNA and 50 ng/well of CMV-β-gal cDNA, which was used to control for differences in transfection efficiency. Afterward, the cells were washed twice with PBS and stimulated overnight with TGF-β₁ (5 ng/ml) in serum-deprived (0.5% FBS) media. Upon completion of agonist stimulation, firefly luciferase and β-gal activities present in detergent-solubilized cell extracts were determined. In addition, 4T1-luciferase cells that stably expressed firefly luciferase under control of the CMV promoter were cultured into 96-well plates at a density of 10,000 cells/well and subsequently were transiently transfected with an SBE reporter plasmid that drove expression of renilla luciferase. The transfectants were stimulated with TGF-β₁ as described previously, and subsequently were processed for the determination of renilla and firefly luciferase by using the Dual-Glo Assay System (Promega, Madison, WI).

NMuMG and 4T1 cells were stimulated with TGF-β₁ (5 ng/ml) for 24 h, and total RNA was isolated by using the RNeasy Plus Kit (Qiagen, Valencia, CA). Afterward, total RNA was reverse transcribed by using the iScript cDNA Synthesis System (BioRad, Hercules, CA), and semiquantitative real-time PCR was conducted by using iQ SYBR Green (BioRad) according to the manufacture's recommendations and as described previously 23. In all cases, differences in RNA concentration were controlled by normalizing individual gene signals to their corresponding GAPDH RNA signals. The oligonucleotide primer pairs used are provided in Additional data file 1.

NMuMG cells (25,000 cells/well) were allowed to adhere overnight to glass coverslips in a 24-well plate. Afterward, the cells were washed extensively in PBS and immediately stimulated with TGF-β₁ (5 ng/ml) in serum-deprived (0.5% FBS) media for 18 h. Upon completion of agonist stimulation, the cells were (a) fixed in 4% paraformaldehyde; (b) permeablized in 0.1% Triton X-100; (c) stained with phospho-Y925 FAK antibodies according to the manufacture's instructions (Cell Signaling); and (d) visualized by using FITC-labeled donkey anti-rabbit secondary antibodies (Jackson ImmunoResearch, West Grove, PA). The actin cytoskeleton was visualized by using TRITC-conjugated phalloidin (0.25 μM; Invitrogen) as described previously 19.

Control or various 4T1 derivatives engineered to express firefly luciferase stably were resuspended in sterile PBS (50 μl) and injected orthotopically into the mammary fat-pads (0.5 to 1.0 × 10⁴ cells/injection) of 6-week-old female Balb/C mice (Jackson Labs, Bar Harbor, ME). Primary tumor growth and metastasis development was assessed by using digital calipers (Fisher Scientific, Waltham, MA), and by weekly bioluminescent imaging on a Xenogen IVIS-200 (Caliper Life Sciences, Hopkinton, MA). Tumor volumes were calculated by using the following equation:

where "x" is the tumor width and "y" is the tumor length. Finally, serial histologic sections of control and FAK-deficient 4T1 tumors removed after the study were stained with phospho-specific antibodies against p38 MAPK and Smad2 and counterstained with hematoxylin, as previously described 17. Where indicated, mice were treated daily with PF-562271 (50 mg/kg) or vehicle (5% gelucire 44/14; Gattefosse, Saint Priest Cedex, France) by oral gavage. Histologic sections from these studies were stained with antibodies for the F4/80 macrophage marker (IHC Tech, Aurora, CO), or with phospho-specific antibodies against Y397-FAK, as described previously 17. All animal studies were performed in accordance with the animal protocol procedures approved by the Institutional Animal Care and Use Committee of University of Colorado.

Statistical values were defined by using an unpaired Student's t-test; a P value <0.05 was considered significant. P values for all experiments analyzed are indicated.

NMuMG cells exhibit several distinct morphologic features in response to TGF-β, most notably a dramatic reorganization of the actin cytoskeleton 24. Figure 1a shows that phosphorylated FAK localized primarily to the periphery in quiescent NMuMG cells, producing a staining pattern very similar to that of F-actin, which was visualized with phalloidin staining (Figure 1a). However, upon TGF-β₁ stimulation, phosphorylated FAK underwent a dramatic reorganization and localized primarily at the end of actin stress fibers (Figure 1a; overlay). Accordingly, immunoblotting NMuMG cell extracts with a panel of phospho-specific FAK antibodies showed that TGF-β stimulation dramatically increased the phosphorylation of FAK (Figure 1b). Figure 1b also shows that TGF-β stimulation increased FAK protein levels in NMuMG cells and induced an impressive upregulation of β₃ integrin 19. Both the increase in FAK phosphorylation and expression, as well as the increase in β₃ integrin expression were wholly dependent on Src activity, because treating these same cells with the Src inhibitor, PP2, abrogated FAK phosphorylation at the Src-dependent sites (Y577 and Y925) and prevented TGF-β₁-induced expression of FAK and β₃ integrin (Figure 1b). Furthermore, in contrast to control (i.e., GFP) and WT-β₃ integrin-expressing NMuMG cells, those engineered to express a signaling-deficient mutant of β₃ integrin, D119A-β3, exhibited drastically reduced maintenance of FAK protein levels and phosphorylation in response to nonadherent conditions (Figure 1c). To more thoroughly investigate the role of β₃ integrin in TGF-β-mediated stabilization and phosphorylation of FAK, nonadherent NMuMG cells were replated in the absence or presence of TGF-β₁ before analyzing FAK expression and phosphorylation by immunoblotting. As shown in Figure 1d, treatment of control (i.e., GFP) and WT-β₃ integrin-expressing NMuMG cells with TGF-β₁ stimulated increased FAK phosphorylation (Figure 1d). In stark contrast, TGF-β treatment of D119A-β₃ integrin-expressing NMuMG cells actually decreased their expression and phosphorylation of FAK (Figure 1d). Finally, we conducted real-time PCR for FAK in control (i.e., GFP), WT-β₃, D119A-β₃ integrin-expressing NMuMG cells. As shown in Figure 1e, chronic TGF-β stimulation (i.e., 24 h) had no effect on FAK mRNA levels in control or WT-β₃ integrin-expressing NMuMG cells; however, these same experimental conditions did increase FAK mRNA expression in D119A-β3 NMuMG cells. These data strongly suggest that (a) upregulated β₃ integrin expression is required to stabilize FAK protein levels upon TGF-β stimulation, and (b) activated β₃ integrin signaling acts as a negative-feedback mechanism governing FAK transcription. Along these lines, our use of oligonucleotide sequences that specifically amplified murine β₃ integrin sequences, not that of recombinant human WT or D119A β₃ integrin sequences, showed that NMuMG cells engineered to overexpress WT-β₃ integrin failed to upregulate endogenous murine β₃ integrin transcripts in response to TGF-β stimulation (Figure 1f). Thus, these findings provide the first evidence that the stability and extent of FAK phosphorylation induced by TGF-β is critically dependent on its ability to upregulate functional β₃ integrin, and that both of these events require the activity of Src kinase. These data also suggest that FAK may play a critical function with β₃ integrin and Src 19 in facilitating TGF-β signaling and function in MECs.

To assess the role of FAK in mediating downstream TGF-β signaling events, we next used shRNAs to deplete stably the expression of FAK in NMuMG cells (Figure 2a). As shown in Figure 2a and 2b, FAK-deficiency had no effect on canonical Smad2/3 activity stimulated by TGF-β, but did markedly disrupt the coupling of TGF-β to the noncanonical p38 MAPK pathway. Furthermore, the quiescent architecture of the actin cytoskeleton, as well as TGF-β-induced actin stress fibers were severely disrupted upon FAK depletion (Figure 2c). We also examined the impact of FAK deficiency on the ability of TGF-β to stimulate MEC migration. To do so, confluent monolayers of control or FAK-deficient NMuMG cells were wounded with a micropipette tip, and the extent of MEC migration into the denuded area was measured at various times thereafter. Stimulating FAK-deficient NMuMG cells with TGF-β₁ enhanced their wound-healing response 25, although at a significantly reduced rate as compared with control (i.e., scrambled shRNA) NMuMG cells (Figure 2d), suggesting that FAK plays a critical role in TGF-β-induced MEC migration. Accordingly, administration of the TβR-I inhibitor, SB-431542, inhibited control NMuMG cell wound closure, thereby identifying a role for autocrine TGF-β signaling in mediating the closure of MEC wounds. Interestingly, FAK-deficient NMuMG cells were refractory to administration of the TβR-I inhibitor (Figure 2d), suggesting that these cells have adapted a less-efficient mechanism of migration that is no longer dependent on the activities of TGF-β and FAK. Finally, as wound closure is driven by both cell migration and proliferation, the decreased growth rate of FAK-deficient cells (see Additional data file 2a) may contribute to their reduced wound-healing response. However, this does not appear to be the case in NMuMG cells, as control and FAK-deficient cells exhibit similar cytostatic responses to high-dose TGF-β1 treatment (5 ng/ml; see Additional data file 2b), which indicates that the difference in wound healing between control and FAK-deficient cells reflects alterations in their ability to migrate, not to proliferate (Figure 2d). Taken together, these data strongly suggest that FAK is directly involved in mediating TGF-β-induced MEC migration.

Imbalances between canonical and noncanonical TGF-β signaling contribute to mammary tumorigenesis 4. We illustrated this shift in TGF-β signaling by using the human MCF10A breast cancer progression model that consists of indolent (T1K), malignant nonmetastatic (Ca1h), and malignant metastatic (Ca1a) cells 2627. With this model system, we observed an enhanced ability of TGF-β to activate specifically p38 MAPK, but not Smad2, in a manner correlating with increasing metastatic potential (Figure 3a). Importantly, the improved coupling of TGF-β to p38 MAPK activation in malignant metastatic Ca1a cells correlated with a marked upregulation of FAK expression as compared with their premetastatic counterparts (Figure 3a). These findings are consistent with the notion that the acquisition of metastasis by breast cancer cells coincides with their elevated expression of FAK, which enhances p38 MAPK activation by TGF-β and its pro-metastatic activities.

To investigate the merits of this supposition, we compared the ability of control and FAK-deficient metastatic breast cancer cells to activate Smad2/3 and p38 MAPK in response to TGF-β. We found that FAK deficiency dramatically not only decreased basal p38 MAPK phosphorylation, but also completely abrogated the ability of TGF-β to activate p38 MAPK (Figure 3b). Interestingly, in contrast to what we observed in NMuMG cells (Figure 2a and 2b), the coupling of TGF-β to both Smad2 and Smad3 were also decreased in FAK-deficient 4T1 cells (Figures 3b and 3c). These data stress the increased dependence of metastatic breast cancer cells on FAK to facilitate not only noncanonical (p38 MAPK), but also canonical (Smad2/3) TGF-β signaling (Figure 3).

Previously, we established Src as being essential for the ability of TGF-β to stimulate p38 MAPK in MECs 19. To investigate the role of FAK in this mechanism, we now analyzed the phosphorylation and activation status of Src (i.e., phosphorylation at Y416) in control and FAK-depleted cells, which showed that FAK deficiency rendered Src hypophosphorylated in 4T1 cells (Figure 4d). Collectively, these findings are the first to demonstrate the increasing dependence of metastatic breast cancer cells on the reciprocal activation between FAK and Src in mediating downstream TGF-β signaling 28.

We next sought to address the mechanism by which FAK facilitates oncogenic TGF-β signaling. Previously, we observed β₃ integrin to interact aberrantly with TβR-II 19, resulting in the promotion of MAPK signaling by TGF-β 1718. Indeed, robust quantities of FAK were detected in β₃ integrin:TβR-II complexes specifically in NMuMG cells engineered to overexpress β₃ integrin (Figure 4a; 19). Furthermore, the formation of β₃ integrin:TβR-II complexes was readily induced in NMuMG cells subsequent to their induction of EMT by TGF-β (Figure 4b; 19). However, this same TGF-β treatment protocol failed to induce β₃ integrin:TβR-II interaction in FAK-depleted NMuMG cells, as (a) β₃ integrin immunocomplexes isolated from FAK-deficient NMuMG cells no longer included TβR-II, and (b) TβR-II immunocomplexes no longer included β₃ integrin (Figure 4b). Moreover, whereas the formation of β₃ integrin:TβR-II complexes was dependent on EMT in NMuMG cells, these same complexes were observed to form constitutively in 4T1 cells (Figure 4c). Importantly, depleting FAK expression in 4T1 cells also reduced the interaction between β₃ integrin and TβR-II, as β₃ integrin immunocomplexes isolated from FAK-deficient 4T1 cells no longer included TβR-II (Figure 4c). The formation of β₃ integrin:TβR-II complexes leads to Src-mediated phosphorylation of TβR-II on Y284, which coordinates the recruitment and binding of Grb2 to TβR-II 18. We now show that FAK deficiency prevented the interaction between TβR-II and Grb2, as Grb2 immunocomplexes isolated from FAK-depleted 4T1 cells no longer included TβR-II (Figure 4d).

Finally, we found that the PTK activity of FAK was absolutely required for the formation of β₃ integrin:TβR-II complexes, as treatment of 4T1 cells with an effective concentration (see Additional data file 3d) of FAK inhibitors (PF-562271 and PF-573228) similarly abolished the interaction between β₃ integrin and TβR-II (Figure 4e). Taken together, these findings demonstrate that FAK expression and activity are required for the formation of β₃ integrin:TβR-II:Grb2 complexes (Figure 4) and, consequently, for the initiation of aberrant oncogenic TGF-β signaling (Figure 3).

To elucidate the significance of FAK in mediating oncogenic TGF-β signaling, we first compared the ability of TGF-β to induce EMT in control and FAK-deficient 4T1 cells. Figure 5a shows that FAK-depleted 4T1 cells failed to undergo the characteristic cell scattering that is normally associated with the induction of EMT (Figure 5a). In contrast to control cells, FAK-depleted 4T1 cells maintained cell-cell junctions that, in many respects, reflect the cortical actin patterns observed in unstimulated normal MECs (Figure 1a). We corroborated these morphologic findings by examining the differential expression of several genes associated with EMT in control and FAK-deficient 4T1 cells before and after their treatment with TGF-β. In doing so, we observed TGF-β administration to reduce dramatically the 4T1 cell expression of E-cadherin (E-cad) protein and to increase the amount of PAI-1 protein in the conditioned media (Figure 5b). Importantly, both TGF-β-dependent responses were lost in 4T1 cells depleted for FAK expression (Figure 5b).

More thoroughly to characterize the role of FAK in TGF-β1-induced EMT, we also examined a panel of EMT markers by real-time PCR. As shown in Figure 5c, the downregulation of the epithelial markers E-cad and cytokeratin-19, which are characteristic features of EMT induced by TGF-β, was effectively prevented by depletion of FAK. Most interestingly, with the exception of PAI-1, the upregulation of mesenchymal markers was not appreciably affected by FAK deficiency (Figure 5c). The maintenance of an epithelial gene-expression profile is consistent with the epithelial morphology of FAK-depleted 4T1 cells (Figure 5a), and is further supported by a recent study in hepatocytes showing that the expression of a dominant-negative FAK prevents the downregulation of epithelial gene expression without affecting the ability of TGF-β to induce mesenchymal gene expression 29.

Consistent with the switch of TGF-β from a tumor suppressor to a tumor promoter, we and others observed TGF-β to induce the invasion of breast cancer cells, a result that is not recapitulated by normal MECs 1830. We therefore monitored the ability of control and FAK-deficient 4T1 cells to invade synthetic basement membranes in response to TGF-β. Figure 5d shows that whereas FAK-deficiency failed to affect the invasion of 4T1 cells induced by a nonspecific serum stimulus, this same cellular condition abrogated the ability of 4T1 cells to undergo enhanced invasion in response to TGF-β (Figure 5d). Previous findings by our laboratory established a model whereby constitutive expression of TβR-II increases the invasion of 4T1 cells 18. Importantly, depletion of FAK in "hypermetastatic" 4T1-TβR-II cells actually reversed the affects of TβR-II expression, as TβR-II-shFAK cells completely failed to invade to a serum stimulus (Figure 5e). Taken together, these data identify FAK as an essential player in mediating carcinoma cell EMT and invasion induced specifically by TGF-β.

Recent data suggest that FAK is required for mammary tumor progression and metastasis 101115; however, the precise mechanisms whereby FAK promotes tumor progression remain to be elucidated. Although FAK depletion had no affect on primary tumor growth (Figure 6a), bioluminescent imaging of mice bearing 4T1 tumors did show that pulmonary metastasis was reduced significantly upon FAK depletion (Figure 6b). In accordance with our in vitro findings (Figure 3), immunohistochemistry of 4T1 tumors demonstrated a dramatic decrease in the phosphorylation of p38 MAPK and Smad2 with FAK depletion (Figure 6c). Thus, these findings suggest that FAK plays a critical role in regulating TGF-β signaling and the metastasis of mammary tumors in mice.

In contrast to tumor cell depletion of FAK, therapeutic administration of the FAK inhibitor, PF-562271, significantly decreased the growth of primary 4T1 tumors (Figure 6d). The reduction in 4T1 tumor growth likely reflects diminished PTK activity of FAK, as tumors from biopsies of mice treated with PF-562271 possessed significantly less phosphorylated FAK as compared with their vehicle-treated counterparts (see Additional data file 3e). Moreover, PF-562271 decreased pulmonary metastasis in a fashion highly reminiscent of that observed in tumors depleted in FAK expression (Figure 6e). The difference in primary tumor growth between FAK-depleted cells (Figure 6a) and systemic FAK inhibition by PF-562271 (Figure 6d) suggests that FAK plays an important role in governing the composition or activity or both of nontumor cells in the tumor microenvironment, including the potential recruitment of systemic cell populations required for optimal mammary tumor growth and progression. Accordingly, we observed 4T1 tumors to exhibit strong staining for the macrophage marker F4/80, a result that was not recapitulated with PF-562271 administration (Figure 6f). Thus, we show for the first time that, in addition to the critical roles FAK plays in directing carcinoma cell function and behavior, the PTK activity of FAK is also clearly required for regulating innate immunity within the microenvironments of developing and progressing mammary tumors.

We next used the 4T1-TβR-II model to access the specific role of FAK in TGF-β-driven breast cancer metastasis. As shown in Figure 7a, FAK depletion had no effect on primary tumor growth of 4T1-TβR-II cells. Furthermore, although FAK-deficient 4T1-TβR-II cells were still highly metastatic, FAK depletion did significantly decrease the immediate pulmonary dissemination of 4T1-TβR-II cells (Figure 7b). These data suggest that FAK selectively regulates the initial steps of tumor cell dissemination stimulated by TGF-β, a result that is consistent with our findings of the requirement of FAK in (a) mediating EMT stimulated by TGF-β (Figure 5) and (b) preventing primary colonization of breast cancer cells in the lung (Figure 6b, weeks 2 to 3), but not their secondary outgrowth (Figure 6b, weeks 3 to 4). In addition, we found no differences in the ability of control or FAK-deficient 4T1 cells to colonize the lungs after their injection into the tail vein of Balb/C mice (see Additional data file 4c). Taken together, these data suggest that the coupling of TGF-β to FAK promotes the initial invasion and exit of breast cancer cells from the primary tumor site. Moreover, and similar to control 4T1 cells, PF-562271 administration beginning 1 week after engraftment of 4T1-TβR-II cells significantly reduced their growth in mice (Figure 7c); however, this same treatment protocol had no effect on the subsequent metastasis of 4T1-TβR-II cells (Figure 7d).

Collectively, these findings provide the first evidence that FAK activity can be inhibited chemotherapeutically as an effective two-pronged approach to reduce the growth and metastasis of breast cancers. Moreover, these results also show that amplified TGF-β signaling in breast cancer cells is capable of driving early tumor cell dissemination from the primary mammary tumor.