The cDNAs encoding wild-type (WT) human β₃ integrin, as well as its inactive mutant D119A (D→A point mutation at position 119), were generously provided by Dr Mark H Ginsberg (University of California at San Diego, San Diego, CA, USA 34). Retroviral β₃ integrin vectors were synthesized by PCR amplification using oligonucleotides containing BglII (amino terminus) and XhoI (carboxyl terminus) restriction sites, and subsequently ligated into identical sites immediately upstream of the IRES in the bicistronic retroviral vector pMSCV-IRES-GFP (plasmid murine stem cell virus-internal ribosomal entry site-green fluorescent protein) or pMSCV-IRES-YFP (yellow fluorescent protein). 35 All β₃ integrin inserts were sequenced in their entirety on an Applied Biosystems 377A DNA sequencing machine (Applied Biosystems, Foster City, CA USA).

Full-length human c-Src cDNA was PCR amplified from IMAGE clone 4871614 (Invitrogen, Carlsbad, CA, USA) using oligonucleotides containing HindIII (amino terminus) and XbaI (carboxyl terminus) restriction sites, respectively. The resulting PCR product was ligated into corresponding sites in pcDNA3.1/Myc-His B vector (Invitrogen) to carboxyl-terminally tag c-Src with a Myc and (His)₆-tag. Kinase dead (Lys295Met) or constitutively active (Tyr530Phe) c-Src mutants were generated by site-directed mutagenesis using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA, USA). Afterward, Myc-His tagged WT and mutant c-Src cDNAs were amplified by PCR and ligated into EcoRI (amino terminus) and BglII (carboxyl terminus) restriction sites in pMSCV-IRES-GFP. All c-Src inserts sequenced in their entirety on an Applied Biosystems 377A DNA sequencing machine.

NMuMG cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS) and 10 μg/ml insulin. MCF10A and MCF10CA1a cells were cultured as previously described 9. Stable expression of individual β₃ integrin subunits or c-Src derivatives in NMuMG and MCF10A cells was accomplished by their overnight infection with control (i.e. pMSCV-IRES-GFP or pMSCV-IRES-YFP), WT or D119A β₃ integrin, or mutant c-Src retroviral supernatants produced by EcoPac2 retroviral packaging cells (Clontech, San Diego, CA, USA), as described previously 36. Cells expressing GFP, YFP, or both fluorescent proteins were isolated and collected 48 hours later on a MoFlo cell sorter (Cytomation, Fort Collins, CO, USA), and subsequently were expanded to yield stable polyclonal populations of control (i.e. GFP or YFP), β₃ integrin (i.e. WT or D119A), or mutant c-Src (i.e. kinase-dead or constitutively active) expressing cells. Expression of recombinant β₃ integrins in individual NMuMG cell lines was monitored by immunoblotting detergent solubilized whole cell extracts with antibodies against the extracellular domain of β₃ integrin (Santa Cruz Biotechnology, Santa Cruz, CA, USA), whereas expression of mutant c-Src protein kinases was detected by immunoblotting with either anti-Src (1:1000 dilution; Santa Cruz Biotechnology) or anti-Myc (1:1000 dilution; Covance, Princeton, New Jersey USA) antibodies.

Control (GFP), WT, or D119A β₃ integrin expressing NMuMG cells were cultured in the absence or presence of TGF-β₁ (5 ng/ml) for 36 hours to stimulate EMT. Afterward, 1 × 10⁶ cells were trypsinized, washed, and incubated in fluorescence-activated cell sorter (FACS) buffer (1% bovine serum albumen in phosphate-buffered saline [PBS]) supplemented with a 1:20 dilution of either PE-conjugated anti-mouse α_v integrin (BD Pharmingen, San Diego, CA, USA) or PE-conjugated anti-human β₃ integrin antibodies (BD Pharmingen). After a 30 min incubation on ice, the cells were washed twice in PBS and immediately fixed with 1% paraformaldehyde before fluorescence-activated cell sorting (FACS) analysis of α_v or β₃ expression in GFP-positive NMuMG cells.

The ability of TGF-β to alter actin cytoskeletal architecture was monitored essentially as described previously 3738. Briefly, control or β₃ integrin expressing NMuMG cells (30,000 cells/well) were plated onto gelatin coated (0.1% in PBS) glass coverslips in 24-well plates. The cells were stimulated with TGF-β₁ (5 ng/ml) for 0–36 hours at 37°C. In some experiments, control or β₃ integrin-expressing NMuMG cells were stimulated with TGF-β₁ in the absence or presence of the Src kinase inhibitor PP2 (10 μmol/l; Calbiochem, Temecula, CA, USA) or its inactive counterpart PP3 (10 μmol/l; Calbiochem). Upon completion of agonist stimulation, the cells were washed in PBS, fixed in 4% paraformaldehyde, and permeabilized by Triton X-100 (0.2% in PBS). The cells were then blocked in PBS supplemented with 1.5% FBS, followed by incubation with TRITC-phalloidin or FITC-phalloidin (0.25 μmol/l). For α_vβ₃ integrin staining, the cells were blocked in goat β-globulin (200 μg/ml; Jackson Immunoresearch, West Grove, PA, USA) before sequential incubations with anti-α_vβ₃ LM609 antibody (1:250 dilution; BD Biosciences, San Jose, CA USA), followed by biotinylated goat anti-mouse antibody (5 μg/ml; Jackson Immunoresearch) and finally by Alexa-streptavidin (1.2 μg/ml; Invitrogen). All images were captured on a Nikon Diaphot microscope.

NMuMG cells lacking either β₃ integrin or c-Src were generated using SMARTpool small interfering RNAs (siRNAs), in accordance with the manufacturer's recommendations (Dharmacon, Lafayette, CO, USA). Briefly, NMuMG cells (30,000 cells/well) were plated either onto plastic or gelatin coated (0.1% in PBS) glass coverslips in 24-well plates and cultured overnight in antibiotic-free media. Fresh media was added the following morning and the cells were transiently transfected with DharmaFECT One reagent (Dharmacon) supplemented with β₃ integrin or c-Src siRNAs (100 nmol/l). Thirty-six hours after transfection, the cells were treated with TGF-β₁ (5 ng/ml) for varying times at 37°C. Upon completion of agonist stimulation, the cells were harvested and prepared for immunoblotting and immunofluorescence analyses as above. NMuMG cell transfection efficiency was monitored by co-transfection with siGLO RISC-Free siRNA (Dharmacon), which was visualized under fluorescent microscope.

Control or β₃ integrin-expressing NMuMG cells were cultured onto 24-well plates (30,000 cells/well) and allowed to adhere overnight. The next morning, the cells were washed twice in PBS and placed in Dulbecco's modified Eagle's medium supplemented with 0.5% FBS for 4 hours before stimulation with TGF-β₁ (5 ng/ml) for 0–120 min. In some experiments, the cells were stimulated with Prolactin (100 ng/ml; generously provided by National Hormone and Peptide Program, US National Institutes of Health), 4α-phorbol 12-myristate 13-acetate (PMA; 10 ng/ml; Sigma, St. Louis, MO, USA), or epidermal growth factor (EGF; 100 ng/ml; Upstate, Charlottesville, VA). Afterward, the cells were washed in ice-cold PBS and lysed in 200 μl of buffer H/1% Triton X-100 36. Detergent solubilized whole cell extracts were prepared, clarified by microcentrifugation, and subsequently concentrated by acetone precipitation. Recovered proteins were fractionated through 10% SDS-PAGE gels, immobilized electrophoretically to nitrocellulose membranes, and subsequently probed with a 1:250 dilution of either anti-phospho-Smad2, -ERK1/2, or -p38 MAPK polyclonal antibodies (Cell Signaling Technology, Beverly, MA, USA). The resulting immunocomplexes were visualized by enhanced chemiluminescence. Differences in protein loading could not be readily monitored by b-actin immunoreactivity because stable β₃ integrin expression and TGF-β stimulation significantly elevated b-actin expression in MECs (data not shown 53239). Thus, differences in protein loading were instead monitored by reprobing stripped membranes with anti-ERK1/2 antibodies (1:2500 dilution; Upstate, Charlottesville, VA USA), whose expression in MECs was unaltered by TGF-β treatment 53239.

The effect of WT and D119A β₃ integrin expression on various TGF-β stimulated activities in MECs was determined as follows: cell proliferation using 10,000 cells/well in a [³H]thymidine incorporation assay, as described elsewhere 35; cell invasion induced by 10% serum using 350,000 cells/well in a modified Boyden-chamber coated with Matrigel matrices (diluted 1:25 in serum-free Dulbecco's modified Eagle's medium), as described elsewhere 3540; and gene expression using 30,000 cells/well in a synthetic pSBE-luciferase reporter gene assay, as described previously 35.

Control and β₃ integrin-expressing NMuMG cells were plated onto 10 cm plates and grown until they reached 90% confluency. The radioligand binding and cross-linking of [¹²⁵I]TGF-β₁ (200 pmol/l) to NMuMG cells was performed as described previously 41. Afterward, cytokine:receptor complexes contained in detergent-solubilized whole cell extracts were isolated by immunoprecipitation with anti-TβR-II antibodies, as described elsewhere 41. Immunocomplexes were subsequently fractionated through 7.5% SDS-PAGE and then immobilized electrophoretically to nitrocellulose and probed with anti-β₃ integrin antibodies. Iodinated TGF-β₁ bound to cell surface TβR-I and TβR-II was visualized by exposure of the dried nitrocellulose membranes to a phosphor screen, which was developed 1–3 days later on a Molecular Dynamics Typhoon Scanner (GE Healthcare Bio-Sciences Corp., Piscataway, NJ USA).

Control (2.5 × 10⁶) or β₃ integrin (7 × 10⁵) expressing NMuMG cells were cultured onto 10 cm plates and subsequently stimulated with TGF-β₁ (5 ng/ml) for varying times in the absence or presence of the Src inhibitor PP2 (10 μmol/l). In some cases, NMuMG cells were held in suspension and replated onto culture dishes previously coated with vitronectin (1 ng/ml; Chemicon, Temecula, CA, USA). Following agonist stimulation, the cells were washed twice in ice-cold PBS and disrupted in Nonidet P-40 lysis buffer 42 (Nonidet P-40; Sigma). The resulting detergent-solubilized whole cell extracts were clarified by microcentrifugation and subjected to the following immunoprecipitation conditions: anti-β₁ integrin antibodies (Santa Cruz Biotechnology) using 1 mg whole cell extract; anti-β₃ integrin antibodies using 1 mg whole cell extract; anti-phosphotyrosine 4G10 antibodies (Upstate) using 1 mg whole cell extract; or anti-TβR-I or -TβR-II antibodies using 2 mg whole cell extract as described previously 43. All immunoprecipitations were incubated for 16 hours at 4°C with slow rotation. The resulting immunocomplexes were collected by microcentrifugation, washed, fractionated through 10% SDS-PAGE gels, and transferred electrophoretically to nitrocellulose membranes, which subsequently were probed with anti-β₃ integrin (1:1000), anti-TβR-II (1:1500; Santa Cruz Biotechnology), or anti-phosphotyrosine 4G10 (1:1500; Upstate) antibodies.

Tyrosine phosphorylation of TβR-II was determined using an in vitro protein kinase assay that measured the ability of either active FAK or Src to phosphorylate recombinant glutathione S-transferase (GST)-TβR-II, which contained only the cytoplasmic domain (i.e. the carboxyl terminal 380 amino acids) of TβR-II fused to GST. Protein kinase reactions were performed in a final volume of 30 μl, consisting of 1 μg of GST-TβR-II with either 0.2 μg of FAK or 1 Unit of Src (Cell Signaling Technology). Reactions were initiated by addition of 10 μl of 4× assay buffer 44 and were incubated for 30 min at 30°C. Afterward, the reactions were stopped by addition of 4× sample buffer 44 and boiled for 5 min. Reactions were diluted by addition of 1 ml buffer H/Triton X-100 and immunoprecipitated overnight with anti-TβR-II antibodies. The resulting immunocomplexes were collected, washed, and resuspended in 1X sample buffer. GST fusion proteins were fractionated through 10% SDS-PAGE before their immobilization to nitrocellulose membranes. Tyrosine phosphorylation of GST-TβR-II was visualized by immunoblotting nitrocellulose membranes with anti-phosphotyrosine antibodies. Differences in immunoprecipitation efficiency and loading were monitored by reprobing stripped membranes with anti-TβR-II antibodies.

The phosphorylation of recombinant Smad3 by activated TβR complexes was performed essentially as described previously 45. Briefly, quiescent NMuMG cells were pretreated with 10 μmol/l of either PP2 or SU6656 (EMD Biosciences, La Jolla, CA, USA) for 1 hour at 37°C, and subsequently were stimulated with TGF-β₁ (5 ng/ml) for 30 min at 37°C. Cytokine stimulations were terminated by washing the cells twice in ice-cold PBS, which then were lysed and solubilized on ice in buffer H/1% Triton X-100. The resulting cell extracts (1 mg/tube) were clarified by microcentrifugation, and subsequently were immunoprecipitated with anti-TβR-II antibodies for 2 hours at 4°C. TβR immunocomplexes were recovered by brief centrifugation and subsequently were washed twice in PBS. TβR phosphotransferase activity against recombinant GST-Smad3 was measured for 30 min at 30°C in a final reaction volume of 40 μl consisting of 30 μl of TGF-β receptor complexes, 5 mg GST-Smad3, and 10 μl of 4X assay buffer 43. Phosphotransferase reactions were stopped by the addition of 4X sample buffer and were boiled for 5 min before their fractionation through 10% SDS-PAGE. Fractionated proteins were immobilized electrophoretically to nitrocellulose membranes, which subsequently were probed with anti-phospho-Smad2/3 antibodies to visualize phosphorylated GST-Smad3. Differences in immunoprecipitation efficiency and loading were monitored by reprobing stripped membranes with anti-TβR-II antibodies.

Alterations in cell microenvironments and stromal compartments play critical roles in regulating the progression of tumors from indolent to aggressive states 4647. Differential integrin expression 27 and dysregulated TGF-β signaling 48 both contribute to the remodeling of cancer cell microenvironments, particularly those of the breast. Because integrins regulate MEC response to TGF-β 32, we hypothesized that differential integrin expression induced by TGF-β would contribute to its regulation of MEC proliferation and EMT. To test this hypothesis, we stimulated normal mouse mammary epithelial (NMuMG) cells with TGF-β₁ and monitored changes in actin stress fiber formation by immunofluorescence. As we have shown previously 38, resting NMuMG cells exhibited typical cuboidal epithelial cell morphology characterized by strong cortical acting staining (Figure 1a). When stimulated with TGF-β₁, these cells underwent EMT and exhibited significant formation of actin stress fibers that emanated from focal adhesions (Figure 1a).

Because α_vβ₃ integrin expression is elevated in breast cancers and enhances cancer cell invasion 273031, we monitored the ability of TGF-β to alter α_vβ₃ integrin expression in NMuMG cells. Immunofluorescence analysis showed that TGF-β₁ treatment of NMuMG cells significantly induced their cell surface expression of the α_vβ₃ integrin (Figure 1a). Western blot analysis confirmed the upregulation of α_vβ₃ integrin expression upon TGF-β₁ mediated EMT (Figure 1b). The association of upregulated α_vβ₃ integrin expression with TGF-β stimulation of EMT suggested that α_vβ₃ integrins may play an essential role in facilitating EMT stimulated by TGF-β. We tested this hypothesis by transiently transfecting NMuMG cells with siRNAs directed against β₃ integrin and subsequently monitored their ability to undergo EMT in response to TGF-β. Figure 1c shows that β₃ integrin deficiency abolished the ability of TGF-β to stimulate EMT in NMuMG cells. More importantly, NMuMG cells transfected with β₃ integrin siRNAs only exhibited rudimentary EMT characteristics coordinate with the expression of α_vβ₃ integrins, which exhibit significantly delayed expression in response to TGF-β (Figure 1c). Moreover, β₃ integrin deficiency failed to alter NMuMG cell expression of β₁ integrins (Figure 1c), indicating that signals arising from β₁ integrin were insufficient in mediating EMT by TGF-β.

Collectively, these findings show that TGF-β induces EMT in NMuMG cells, and that upregulated expression of α_vβ₃ integrin is essential for this process to occur. The findings also raise the possibility that β₃ integrins may play a direct role in regulating various MEC responses to TGF-β₃.

Integrins regulate growth factor signaling in part via formation of integrin:RTK complexes 2324. Based on our findings presented in Figure 1, we hypothesized that β₃ integrins interact physically with cell surface TβRs. In testing this hypothesis, NMuMG cells were stimulated with TGF-β for various periods of time, whereupon whole cell extracts were prepared and immunoprecipitated with anti-β₃ integrin antibodies. Probing the resulting immunocomplexes with anti-TβR-II antibodies showed that β₃ integrin did indeed interact with TβR-II (Figure 2a), and that EMT was necessary to induce this event. In addition, detergent-solubilized whole cell extracts prepared from control and TGF-β stimulated NMuMG cells were immunoprecipitated with antibodies against either β₁ or β₃ integrins, and the resulting immunocomplexes were probed for TβR-II. Figure 2b shows that β₁ integrins interacted constitutively with TβR-II, whereas EMT was again required to promote the association of β₃ integrin with TβR-II. Collectively, these findings indicate that the formation of β₃ integrin:TβR-II complexes is an EMT-specific phenomena that may play an important role in regulating TGF-β stimulation of EMT in NMuMG cells.

To further explore the role of β₃ integrin in regulating MEC function and TGF-β biology, we used bi-cistronic retroviral transduction to stably express β₃ integrin or its inactive counterpart D119A-β₃ integrin 34 in NMuMG cells. The D119A-β₃ integrin contains a point mutation in the conserved RGD integrin binding motif and therefore fails to bind RGD-containing ligands, such as fibronectin, vitronectin, and osteopontin 4950. Recombinant β₃ integrin expression in NMuMG cells was confirmed by immunoblotting whole cell extracts from basal or TGF-β₁ treated cells (Figure 3a). As observed previously, TGF-β stimulation of EMT in NMuMG cells induced their expression of α_v integrin (Figures 1 and 3a). Interestingly, WT β₃ integrin expression induced a dramatic compensatory upregulation of α_v integrin expression in NMuMG cells (Figure 3a). This β₃ integrin response, which was independent of TGF-β administration, was mimicked moderately by expression of its inactive D119A counterpart (Figure 3a). Although D119A β₃ integrin is considered to be functionally inactive, its heterologous expression in cells can elicit partial biologic activity in part through its ability to bind non-RGD-containing ligands, such as plasminogen 51. Thus, these findings suggest that signals arising from activated β₃ integrins induce α_v integrin expression in MECs.

To confirm that recombinant β₃ integrins were expressed on the cell surface of NMuMG cells, control or TGF-β₁ treated cells were incubated with PE-conjugated anti-human β₃ integrin antibodies, and subsequently analyzed by FACS analysis. As expected, retrovirally infected NMuMG cells expressed WT and D119A β₃ integrins on their cell surfaces (Fig. 3b, left). In addition, β₃ integrin expressing cells were also incubated with PE-conjugated anti-mouse α_v integrin antibodies, which confirmed the compensatory upregulation of α_v integrins and showed their ability to form functional complexes with cell surface β₃ integrin (Figure 3b, right).

Finally, iodinated TGF-β₁ radioligand binding and cross-linking assays were performed to confirm that β₃ integrins did indeed interact with TβR-II at the cell surface. As shown in Figure 3c, WT and D119A β₃ integrins both interacted with TβR-II at the cell surface, indicating that formation of this complex is independent of β₃ integrin activation. Quite surprisingly, WT β₃ integrin expression, but not that of D119A, increased cell surface expression of TβR-II (Figure 3c). Indeed, the ratio of TβR-II:TβR-I detected at the cell surface in WT β₃ integrin expressing NMuMG cells (1.31 ± 0.14; n = 4) was significantly higher (P = 0.009, Student's t-test) than those observed in their GFP β₃ integrin (0.76 ± 0.05; n = 4) and D119A β₃ integrin (0.72 ± 0.31; n = 4) expressing counterparts. Interestingly, semiquantitative real-time PCR analyses failed to detect differences in TβR-II transcript expression between these individual NMuMG cell lines (data not shown), suggesting that signals arising from β₃ integrins alter TβR-II translocation to and/or internalization from the cell surface of MECs.

Collectively, these findings show that retrovirally expressed human β₃ integrins were translocated to the cell surface of NMuMG cells, where they formed active heterodimers with α_v integrin, and more importantly they interacted physically with TβR-II at the plasma membrane. In addition, these findings and those presented in Figures 1 and 2 illustrate a unique and essential role for β₃ integrin in promoting EMT, particularly that mediated by TGF-β, and suggest that the formation of β₃ integrin:TβR-II complexes may alter the response of MECs to TGF-β by regulating its intracellular signaling systems.

The coupling of integrins to RTKs often is associated with activation of MAPKs 2352. This is noteworthy because the activation of MAPKs is essential for the ability of TGF-β to promote EMT 21516. We therefore determined whether β₃ integrin expression altered the ability of TGF-β to activate MAPKs in NMuMG cells. Although expression of β₃ integrin in NMuMG cells had negligible effects on TGF-β stimulated Smad2 phosphorylation (Figure 4a) and nuclear translocation (data not shown), its expression significantly enhanced the activation of ERK1/2 and p38 MAPK by TGF-β (Figure 4a). Integrin expression can elicit dramatic changes in cell adhesion and morphology 23, potentially resulting in global alterations in intracellular signaling systems. To determine the specificity of β₃ integrin expression to TGF-β signaling, we monitored the activation status of p38 MAPK in control and β₃ integrin expressing NMuMG cells before and after their stimulation with TGF-β₁, prolactin, PMA, or EGF. Figure 4b shows that β₃ integrin expression selectively regulated the coupling of TGF-β, but not that of prolactin, PMA, or EGF, to p38 MAPK activation in NMuMG cells. Thus, these findings implicate β₃ integrins as unique mediators of TGF-β signaling in MECs.

Based upon the ability of β₃ integrins to enhance MAPK activation by TGF-β, we next sought to determine whether β₃ integrins could alter the ability of TGF-β to regulate MEC cell proliferation, invasion, and EMT. We first measured the effects of β₃ integrin expression on the rate of DNA synthesis in resting NMuMG cells. As compared with GFP expressing NMuMG cells, those expressing WT β₃ integrin exhibited significantly increased rates of DNA synthesis (by 78 ± 0.12%; n = 3; P = 0.002), whereas those expressing D119A β₃ integrin exhibited significantly retarded rates of DNA synthesis (by 68 ± 4.0%; n = 3; P = 0.00007). Despite the relative growth rate differences observed between these β₃ integrin expressing NMuMG cell derivatives, the response of GFP and D119A expressing NMuMG cells to TGF-β mediated cytostasis was surprisingly similar (Figure 5a). In contrast, WT β₃ integrin expression significantly reduced the ability of TGF-β to induce growth arrest in NMuMG cells (Figure 5a), indicating that β₃ integrin expression significantly reduces cytostasis mediated by TGF-β, thereby circumventing its tumor suppressor function in MECs.

We also determined whether β₃ integrins are capable of enhancing the tumor promoting functions of TGF-β. In doing so, we compared the ability of β₃ integrin expressing NMuMG cell lines to invade through synthetic basement membranes. As shown in Figure 5b, GFP expressing NMuMG cells exhibited minimal invasion through Matrigel matrices in response to serum stimulation, a response that was enhanced insignificantly by TGF-β treatment. Similar to its inability to alter growth regulation by TGF-β, expression of the D119A β₃ integrin in NMuMG cells failed to affect their TGF-β regulated invasiveness (Figure 5b). In contrast, NMuMG cells expressing β₃ integrin exhibited a trend toward increased invasiveness to serum as compared with control cells (Figure 5b); however, when stimulated with TGF-β₁ these cells exhibited significantly enhanced invasion through synthetic basement membranes (Figure 5b). Because increased invasiveness is often associated with EMT 11, we examined the morphology and actin cytoskeleton of NMuMG cells expressing β₃ integrins. Consistent with our previous results, only those NMuMG cells that expressed WT β₃ integrin acquired an elongated, fibroblast-like morphology (Figure 5c), complete with the formation of actin stress fibers (Figure 5c). Taken together, these findings suggest that although both WT and D119A β₃ integrins can form complexes with TβR-II at the cell surface, only functional β₃ integrins are capable of inducing EMT and invasion in NMuMG cells.

Lastly, we monitored changes in TGF-β stimulated gene expression by measuring differences in luciferase expression driven by the synthetic SBE (Smad binding element) promoter. Figure 5d shows that only WT β₃ integrin expression significantly enhanced TGF-β stimulated gene expression as compared with control cells or those expressing its inactive β₃ integrin counterpart.

Collectively, these findings indicate that, unlike its nonfunctional counterparts, expression of WT β₃ integrin in MECs diminished TGF-β mediated growth arrest, enhanced TGF-β mediated invasion and EMT, and augmented TGF-β mediated gene expression. Moreover, our results suggest that β₃ integrin expression translates TGF-β from an inhibitor of MEC cell growth to a stimulator of their invasion and EMT.

Our findings thus far show that β₃ integrin and TβR-II interact upon TGF-β stimulation, β₃ integrin expression is essential for EMT stimulated by TGF-β, and β₃ integrin expression influences the MEC response to TGF-β. To further explore the consequences of β₃ integrin:TβR-II complex formation, GFP expressing NMuMG cells were incubated in the absence or presence of TGF-β for 36 hours, and subsequently were held in suspension before re-plating onto plastic or vitronectin-coated wells in the absence or presence of TGF-β₁ for 5 min. Afterward, the phosphorylation status of TβR-II and its ability to form complexes with β₃ integrin were monitored by co-immunoprecipitation analyses. Figure 6a shows that following TGF-β induced EMT, TβR-II was constitutively tyrosine phosphorylated and associated with β₃ integrin. Because chronic TGF-β stimulation of NMuMG cells induces β₃ integrin expression (Figure 1) and complex formation with TβR-II (Figure 2), it was difficult to assess the role of TGF-β receptor and integrin stimulation in the formation of β₃ integrin:TβR-II complexes in NMuMG cells. To circumvent this difficulty, resting WT β₃ integrin expressing NMuMG cells were dissociated and subsequently replated and stimulated for 5 min, as above. Figure 6b shows that TβR-II only interacted with β₃ integrins upon TGF-β₁ stimulation and, more importantly, that this response was enhanced significantly upon β₃ integrin adhesion. In addition, only upon dual receptor activation (i.e. β₃ integrin and TβRs) was TβR-II phosphorylated on tyrosine residues (Figure 6b). Thus, β₃ integrin regulates TGF-β signaling by targeting TβR-II for tyrosine phosphorylation in a manner reminiscent of integrin actions on RTK signaling 23.

Although integrins lack intrinsic protein tyrosine kinase (PTK) activity, they facilitate tyrosine phosphorylation of target proteins via their recruitment of FAK and Src PTKs to focal adhesions 22. We therefore hypothesized that FAK or Src are potential PTKs responsible for phosphorylating TβR-II on tyrosine residues in NMuMG cells. We tested this hypothesis by utilizing an in vitro PTK assay that measured the ability of purified, active FAK or Src to phosphorylate GST fusion proteins containing catalytically active or inactive (i.e. K277R) versions of the cytoplasmic domain of TβR-II. As shown in Figure 6c, Src phosphorylated the cytoplasmic domain of TβR-II on tyrosine residues. Although active and capable of undergoing autophosphorylation on tyrosine residues (Figure 6d), FAK was unable to phosphorylate TβR-II in vitro (Figure 6c), suggesting that Src phosphorylates TβR-II in response to β₃ integrin expression.

To further examine the functional link between Src and TβR-II, we attempted to inhibit Src activity or function in NMuMG cells and subsequently monitor the consequences of these manipulations on TβR-II tyrosine phosphorylation. Recent work conducted by Maeda and coworkers 53 showed that the broad spectrum protein kinase inhibitor PP1 antagonizes TGF-β signaling by inhibiting TβR-I and TβR-II protein kinase activity. As such, we monitored the effects of two Src inhibitors, PP2 and SU6656, on the ability of immunoprecipitated TGF-β receptor complexes to phosphorylate recombinant GST-Smad3 in vitro. As expected, TGF-β stimulation greatly enhanced the phosphorylation of recombinant Smad3 (Figure 7a), a reaction that was reduced markedly by SU6656 administration (Figure 7a). In contrast, PP2 treatment failed to inhibit TβR activity and instead appeared to elicit agonist independent receptor activation (Figure 7a). Based on its inactivity against TβRs, PP2 appeared to be a suitable pharmacological agent to inhibit Src activity in MECs. Accordingly, β₃ integrin-expressing NMuMG cells treated with PP2 abrogated the tyrosine phosphorylation of TβR-II in TGF-β stimulated NMuMG cells (Figure 7b), indicating that Src activity does indeed mediate tyrosine phosphorylation of TβR-II in β₃ integrin expressing NMuMG cells. Along these lines, transient expression of constitutively active Src in β₃ integrin expressing NMuMG cells resulted in significantly elevated tyrosine phosphorylation of TβR-II, an event that was not recapitulated by expression of dominant negative Src (Figure 7c). Taken together, these findings indicate that β₃ integrin-mediated recruitment of Src facilitates the tyrosine phosphorylation of TβR-II in MECs, and that this phosphorylation event may act to enhance their proliferation, invasion, and EMT by augmenting the ability of TGF-β to activate MAPKs.

We next explored the functional importance of Src in regulating various MEC responses to TGF-β. To do so, NMuMG cells were pretreated with either PP2 or its inactive counterpart PP3 and subsequently stimulated with TGF-β to activate ERK1/2 and p38 MAPK. As expected, β₃ integrin expression significantly increased ERK1/2 and p38 MAPK activation following TGF-β₁ stimulation (Figure 8a). This response to TGF-β was unaffected by PP3 administration, but it was reduced significantly by PP2 treatment of NMuMG cells (Figure 8a). More importantly, PP2 administration completely reverted the EMT phenotype induced by expression of β₃ integrin in NMuMG cells, and nearly abolished the ability of TGF-β to stimulate EMT in NMuMG cells (Figure 8b). We also engineered NMuMG cells to stably express dominant negative Src and subsequently measured its effect on the ability of TGF-β to induce EMT and invasion in MECs. As expected, GFP expressing NMuMG cells readily underwent EMT when stimulated by TGF-β (Figure 8c). In stark contrast, NMuMG cells expressing dominant negative Src were stunted in their ability to undergo EMT in response to TGF-β (Figure 8c), and instead formed small, disorganized stress fibers that contrasted sharply with those produced in their GFP expressing counterparts.

Proper organization of cell polarity is essential for cancer cell invasion and metastasis 54. Based on the aberrant EMT response observed in dominant negative Src expressing NMuMG cells, we suspected that their invasion through synthetic basement membranes would also be impaired. Accordingly, dominant negative Src expression significantly inhibited the ability of TGF-β to induce NMuMG cell invasion (Figure 8d). Interestingly, singular expression of either constitutively active Src or β₃ integrin in NMuMG cells significantly enhanced tonic and TGF-β stimulated cell invasion (Figure 8d). More importantly, introducing dominant negative Src into β₃ integrin expressing NMuMG cells completely abolished the ability of TGF-β to stimulate cell invasion in MECs (Figure 8d). Thus, activity of both β₃ integrin and Src is necessary for MEC invasion stimulated by TGF-β.

Finally, the findings above suggest that Src may play an essential role in facilitating EMT stimulated by TGF-β. We tested this hypothesis by transiently transfecting NMuMG cells with siRNAs directed against Src and subsequently monitored their ability to undergo EMT in response to TGF-β. Figure 8e shows that Src deficiency did indeed abolish the ability of TGF-β to stimulate EMT in NMuMG cells, a response reminiscent of that observed in NMuMG cells deficient in β₃ integrin (Figure 1c). Collectively, these findings indicate that Src expression and activity are essential for the ability of TGF-β to stimulate MAPKs, and to induce EMT and cell invasion in MECs.

Our findings have thus far solely investigated the relationship between β₃ integrins and TGF-β in normal MECs. It should be noted that cells undergoing neoplastic transformation, including those of the breast 2627, exhibit altered integrin expression profiles during tumorigenesis, suggesting that integrins could play a key role in the acquisition of neoplastic phenotypes. The human MCF10A system consists of a series MEC cell lines that share a common ancestry and represent distinct stages of breast cancer development, ranging from normal to highly invasive and metastatic 55; they also represent a model system for studying the conversion of TGF-β from a tumor suppressor to a tumor promoter 9.

Interestingly, breast cancer progression was associated with elevated expression of β₃ integrin and its effector, FAK, which correlated with the conversion of TGF-β from a suppressor to a promoter of breast cancer progression (Figure 9a). Moreover, exogenous β₃ integrin expression in normal, poorly invasive MCF10A cells significantly enhanced their invasion through synthetic basement membranes in response to TGF-β (Figure 9b). In contrast, the malignant, metastatic MCF10CA1a cells readily invaded Matrigel matrices when stimulated with TGF-β, a response that was augmented significantly by WT β₃ integrin expression (Figure 9c). Importantly, D119A β₃ integrin expression completely abolished the ability of TGF-β to stimulate MCF10ACA1a cell invasion (Figure 9c). Collectively, our findings suggest that mammary tumorigenesis alters β₃ integrin expression in neoplastic MECs, and that β₃ integrins may play a crucial role in translating TGF-β from a suppressor to a promoter of breast cancer growth and invasion.