Control of Type II Transforming Growth Factor-β Receptor Expression by Integrin Ligation*

Ectopic expression of the α5 integrin subunit in cancer cells with little or no endogenous expression of this integrin often results in reduced proliferation as well as reduced malignancy. We now show that inhibition resulting from ectopic expression of α5integrin is due to induction of autocrine negative transforming growth factor-β (TGF-β) activity. MCF-7 breast cancer cells do not express either α5 integrin or type II TGF-β receptor and hence are unable to generate TGF-β signal transduction. Ectopic expression of α5integrin expression enhanced cell adhesion to fibronectin, reduced proliferation, and increased the expression of type II TGF-β receptor mRNA and cell surface protein. Receptor expression was increased to a higher level in α5transfectants by growth on fibronectin-coated plates. Induction of type II TGF-β receptor expression also resulted in the generation of autocrine negative TGF-β activity because colony formation was increased after TGF-β neutralizing antibody treatment. Transient transfection with a TGF-β promoter response element in tandem with a luciferase cDNA into cells stably transfected with α5integrin resulted in basal promoter activities 5–10-fold higher than those of control cells. Moreover, when α5 transfectants were treated with a neutralizing antibody to either TGF-β or integrin α5, this increased basal promoter activity was blocked. Autocrine TGF-β activity also induced 3-fold higher endogenous fibronectin expression in α5 transfectants relative to that of control cells. Re-expression of type II receptor by α5 transfection also restored the ability of the cells to respond to exogenous TGF-β and led to reduced tumor growth in athymic nude mice. Taken together, these results show for the first time that TGF-β type II receptor expression can be controlled by α5β1 ligation and integrin signal transduction. Moreover, TGF-β and integrin signal transduction appear to cooperate in their tumor-suppressive functions.

As mediators of cell matrix-cell interaction, integrins impart diverse biological properties to the cells that express them. A number of studies have demonstrated that integrin expression affects tumor cell proliferation and progression (1)(2)(3). Immunohistochemical analysis showed that the expression of integrins is altered in human tumors compared with corresponding normal cells. Many neoplastic cells show reduced expression of integrins (4 -9). Overexpression of certain integrins, such as ␣ 5 ␤ 1 , can reverse tumorigenicity and anchorage-independent growth in some transformed cells (10,11). Perturbation of integrin ␣ 5 ␤ 1 binding to its ligand stimulates the growth of a variant of the K562 cells (12). Recently, we showed expression of integrin ␣ 5 subunit selectively blocks DNA synthesis (13) and disruption of ␣ 5 ␤ 1 ligation enhanced DNA synthesis in a mitogen-activated protein kinase-dependent but epidermal growth factor receptor-independent manner (14). Extracellular matrix recognition by integrin ␣ 5 ␤ 1 may, therefore, play a role in the negative control of cell growth, which may be lost in some cancer cells.
Loss of negative growth factors can also lead to abnormal growth in transformed cells. Transforming growth factor-␤ (TGF-␤) 1 has been identified as a potent growth inhibitor in various normal as well as transformed cell types (15,16). Escape from the negative growth control by TGF-␤ is an important step during oncogenic transformation (15) because the growth of normal mammary epithelial cells is inhibited by TGF-␤, whereas their transformed counterparts are often resistant to its inhibitory effects (17)(18)(19). Loss of TGF-␤ sensitivity in estrogen receptor ϩ breast cancer cells has frequently been associated with loss of the TGF-␤ type II receptor (RII), which, along with the TGF-␤ type I receptor (RI), is necessary for TGF-␤ signal transduction (20,21). The importance of loss of TGF-␤ signaling in breast cancer cells was demonstrated by differences in tumorigenicity of MCF-7 breast cancer clones with and without RII expression (22) and the ability of reexpression of RII in MCF-7 cells to inhibit tumorigenicity in athymic mice (21). RII has been shown to be a tumor suppressor gene by the criteria that its mutational inactivation is associated with a hereditary form of colon cancer and that RII ectopic expression in cancer cells from individuals with this form of cancer reverses malignancy in athymic mice (24,25). Re-expression of RI in cancer cells that are deficient in this receptor also reverses tumorigenicity (26).
Two lines of evidence indicate that autocrine negative TGF-␤ may be more important than response to exogenous TGF-␤ in controlling tumor growth. The first of these was the demonstration that removal of autocrine TGF-␤ activity by stable transfection of a TGF-␤ antisense expression vector leads to malignant progression of cancer cells in athymic mice (27,28). This approach blocked autocrine TGF-␤ activity because endogenous TGF-␤ was removed from these cells but did not affect the expression of TGF-␤ receptors and, therefore, permitted the retention of response to exogenously produced TGF-␤ in the tumor environment from nonmalignant cells. The occurrence of tumor formation indicated that exogenous TGF-␤ produced by nonmalignant cells was insufficient to achieve tumor suppression. The second line of evidence involved the re-expression of RII in a cell line that was homozygous for mutational inactivation of the gene. Re-expression of RII regenerated autocrine negative TGF-␤ but did not regenerate an inhibitory response to exogenous TGF-␤ (25). However, reversion of tumorigenicity did occur, indicating that autocrine TGF-␤ was critical. Thus, the available evidence indicates that mechanisms for regeneration of RII expression and autocrine negative TGF-␤ will be of importance in our understanding of controls of TGF-␤ signal transduction in malignancy and may lead to novel treatment or prevention approaches for cancer.
Previously, we showed that vitamin D 3 was inhibitory to wild type MCF-7 clones that expressed RII and to RII-null clones expressing ectopic RII (29). In contrast, RII-null clones were refractory to inhibition (29). Response to vitamin D 3 was associated with induction of higher RII levels and enhanced TGF-␤ autocrine negative activity, suggesting that other growth modulators may cause inhibition by inducing TGF-␤ autocrine negative activity. Given the ability of ␣ 5 integrin to affect growth in a negative fashion (13,14), we hypothesized that re-expression of cell surface ␣ 5 ␤ 1 integrin in cancer cells deficient in the expression of the ␣ 5 subunit would lead to regeneration of RII expression and autocrine negative TGF-␤ activity. This hypothesis was tested by stable transfection of a MCF-7 breast cancer clone lacking both RII and ␣ 5 integrin expression with an ␣ 5 integrin cDNA. Transfection resulted in re-expression of RII and regeneration of autocrine TGF-␤ activity, as well as response to exogenous TGF-␤. RII expression and TGF-␤ responses were dependent upon ␣ 5 ␤ 1 ligation to endogenous MCF-7 fibronectin (FN) and were enhanced when exogenous FN was used to coat culture plates, indicating that growth inhibition by ␣ 5 ␤ 1 ligation involves induction of autocrine TGF-␤ activity.

MATERIALS AND METHODS
Cell Culture-MCF-7 cells were originally obtained from American Type Culture Collection and adapted to McCoy's 5A medium supplemented with 10% fetal bovine serum (FBS), pyruvate, vitamins, amino acids, and antibiotics (30). Working cultures were maintained at 37°C in a humidified atmosphere of 5% CO 2 and routinely checked for mycoplasma contamination as described previously (21). MCF-7 limiting dilution clones were obtained by diluting the parental cells in 96-well tissue culture plates at 0.5 cell/well as described previously (21). The strain of MCF-7 cells utilized in this study is insensitive to TGF-␤ because it lacks RII (21).
Integrin ␣ 5 Stable Transfection-An integrin ␣ 5 expression vector for mammalian cells was used for stable transfection as described previously (13). The plasmid was linearized and transfected into a typical MCF-7 limiting dilution clone (designated MCF-7 LDC4) by electroporation with a Bio-Rad Gene Pulser at 250 V and 960 microfarads. Control cells were transfected with a Neo-containing plasmid. The transfected cells were plated in 100-mm culture dishes in 10% FBS medium for 2 days. Selection of stable transfectants was carried out by adding Geneticin (600 g/ml; Life Technologies, Inc.) into the medium. After three weeks, Geneticin-resistant clones were ring-cloned and expanded for screening of ␣ 5 expression. The control clones were pooled and designated the MCF-7 LDC4 Neo pool.
RNA Analysis-Total RNA was isolated from cultured cells by the guanidine isothiocyanate method (31). For detection of FN mRNA lev-els, cells (10 6 ) were plated in 100-mm culture dishes coated with poly-L-lysine (10 g/ml; Sigma) or FN (10 g/ml; Collaborative Biomedical Products) for 4 days in 10% FBS medium, and total RNA was then isolated. For detection of RII mRNA levels, cells (10 6 ) were plated in 100-mm culture dishes coated with poly-L-lysine (10 g/ml) or FN (10 g/ml) for 1 and 2 h in McCoy's 5A medium supplemented with 2% bovine serum albumin (Sigma) and 24 and 96 h in McCoy's 5A medium supplemented with 10% FBS.
The construction of the integrin ␣ 5 subunit and RII antisense probes has been described (21,32). The FN riboprobe plasmid was constructed by subcloning a 232-base pair BamHI-PvuII fragment of the human FN cDNA into a pBSK(Ϫ) vector (Stratagene Cloning System). T 7 RNA polymerase was used to synthesize the FN antisense probe (13,32). The RNase protection assay was performed by hybridization of the radioactive riboprobes with total RNA (20 g) isolated from the control or ␣ 5 -transfected cells as described previously (32).
Immunoprecipitation-To determine cell surface integrin ␣ 5 ␤ 1 expression, cell surface proteins were labeled with biotin, immunoprecipitated with an anti-␣ 5 subunit monoclonal antibody (Life Technologies, Inc.), and analyzed by SDS-polyacrylamide gel electrophoresis as described previously (32).
The specificity of cell adhesion to FN was determined using a monoclonal anti-human integrin ␣5 subunit antibody (Life Technologies, Inc.). The cells were incubated in the absence or presence of the antibody (1:100 dilution) for 30 min at 4°C and then plated at 4 ϫ 10 4 cells/well in 96-well plates coated with FN (10 g/ml). Determination of cell adhesion and the 3-(4,5-dimethyl thiazol-2-yl)-2,5-diphenyl tetrazolium bromide assay were performed as described above.
Tumorigenicity-Exponentially growing cells of MCF-7 Neo pool and ␣ 5 sense clone 15 were inoculated subcutaneously as described previously (21). Growth curves for xenografts were determined by measuring the volume (V) of tumors. V is expressed as V ϭ (L ϫ W 2 ) ϫ 0.5, where L is the length and W is the width of the xenograft.
Mitogenesis Assay-Inhibition of cell proliferation by exogenous TGF-␤ 1 in Neo and ␣ 5 -transfected cells was determined by measuring [ 3 H]thymidine incorporation as described previously (21). Briefly, cells were plated in 24-well plates at 1.5 ϫ 10 4 cells/well in the presence of various concentrations of TGF-␤ 1 (0.2-10 ng/ml). After 4 days of incubation, cells received a 1-h pulse with [ 3 H]thymidine (25 Ci) and were washed with supplemental McCoy's 5A medium three times, and DNA was precipitated with 10% trichloroacetic acid and then solubilized with 0.2 M sodium hydroxide. The amount of [ 3 H]thymidine incorporated was analyzed by liquid scintillation counting in a Beckman LS7500 scintillation counter. Growth inhibition by TGF-␤ 1 is represented as the percentage of [ 3 H]thymidine incorporation of TGF-␤ 1 -treated cells relative to untreated cells.
Plating Efficiency Assay-The effect of TGF-␤ 1 neutralizing antibody on the clonogenic potential of control and ␣ 5 -transfected cells was compared to determine autocrine TGF-␤ activity as described previously (25,26). Cells were seeded at low density (400 cells/well) in 24-well plates in the presence of control IgG (10 g/ml; R & D Systems, Inc.) or TGF-␤ 1 neutralizing antibody (10 g/ml; R & D Systems, Inc.). After 8 days of incubation without medium change, cell colonies were fixed with 1% glutaraldehyde, stained with 0.1% crystal violet, and dissolved in 1% Triton X-100 as described by Westergren-Thorsson et al. (34).
Transient Transfection and Luciferase Assay-The TGF-␤-responsive luciferase expression plasmid (p3TP-Lux) was used for transient transfections, and luciferase assays were performed as described previously (35). MCF-7 Neo control and ␣ 5 sense-transfected cells were transfected with 30 g of p3TP-Lux and 10 g of ␤-galactosidase plasmid by electroporation with a Bio-Rad Gene Pulser at 250 V and 960 microfarads. The electroporated cells were plated onto 6-well tissue culture plates. After the cells attached, control IgG (10 g/ml) and TGF-␤ 1 neutralizing antibody (10 g/ml) were added. The cells were harvested with 200 l of lysis buffer (luciferase assay system; Promega). Luciferase activity was measured in the first 10 s after substrate addition using a luminometer (Berthold Lumat LB 9501) and expressed as arbitrary units after normalization with ␤-galactosidase activity.
Receptor Cross-linking-Human TGF-␤ 1 was purified and iodinated by the chloramine T method as described previously (36). Equal numbers of the cells (10 5 ) were plated in 6-well plates, and after 5 days, binding and cross-linking of 200 mM 125 I-TGF-␤ 1 to the cell monolayer were performed as described by Segarini et al. (37). Labeled cells were solubilized in 200 l of 1% Triton X-100. Equal amounts of cell lysate protein were electrophoresed by 4 -10% gradient SDS-polyacrylamide gel electrophoresis under reducing conditions and exposed for autoradiography.

RESULTS
Expression of ␣ 5 ␤ 1 Integrin in MCF-7 Cells-We transfected the ␣ 5 subunit into a typical limiting dilution clone, MCF-7 LDC4. The ␣ 5 -positive transfectants were initially screened for increased expression of ␣ 5 mRNA by RNase protection assays (Fig. 1). Two positive clones (designated cl.5 and cl.15) were isolated that expressed higher levels of ␣ 5 subunit mRNA than the untransfected MCF-7 LDC4 and Neo-transfected (Neo pool) control cells. Increased ␣ 5 mRNA expression was accompanied by increased cell surface expression (Fig. 2). The levels of cell surface integrin ␤ 1 subunit dimerized by ␣ 5 subunit were also increased, indicating that the expression of ␤ 1 subunit might be up-regulated by increased ␣ 5 subunit expression or that excess ␤ 1 subunit already present could complex with the increased levels of ␣ 5 subunit. In contrast to ␣ 5 ␤ 1 , the levels of ␣ 2 ␤ 1 and ␣ 3 ␤ 1 were unchanged in ␣ 5 transfectants (data not shown), ruling out the possibility that increased ␣ 5 expression might compete with other ␣ subunits for dimerizing with the ␤ 1 subunit. We next determined whether the increased ␣ 5 ␤ 1 en-hanced adhesion to FN (Fig. 3A). Both ␣ 5 transfectants showed 7-8-fold increased adhesion to FN-coated plates at FN concentrations ranging from 5 to 10 g/ml, whereas Neo control cells showed only 3-4-fold enhancement. In addition, both ␣ 5 transfectants showed ϳ5-fold increases of binding at FN concentrations ranging from 0 to 2.5 g/ml, relative to Neo control cells. The specificity of cell adhesion was shown by blocking the attachment to FN with an anti-␣ 5 subunit antibody (Fig. 3B), thus indicating that enhanced cell attachment to FN was due to increased cell surface ␣ 5 ␤ 1 expression.
Ectopic ␣ 5 expression leading to enhanced ␣ 5 ␤ 1 ligation blocked DNA synthesis, whereas disruption of ligation led to increased DNA synthesis in other model systems (13,14). If our hypothesis that enhanced ␣ 5 ␤ 1 function leads to induction of autocrine negative TGF-␤ activity was correct, ectopic expression of ␣ 5 should result in reduced cell proliferation. Growth curves of ␣ 5 -transfected clones showed more than 50% inhibition of proliferation relative to wild type cells (Fig. 3C).
Expression of RII in ␣ 5 Transfectants-Previously, we showed that MCF-7 cells were insensitive to TGF-␤ 1 because they expressed nearly undetectable levels of RII (21). Therefore, if ␣ 5 ␤ 1 -mediated growth inhibition was associated with autocrine negative TGF-␤ activity, RII expression must be restored. To test this hypothesis, we examined RII mRNA levels in ␣ 5 transfectants by RNase protection assay (Fig. 4). High steady state RII mRNA levels were induced in ␣ 5 -transfected cells compared with Neo control cells when the cells were cultured on poly-L-lysine. This was probably due to enhanced production of endogenous FN, as described below. Growth on exogenous FN further increased RII mRNA levels to 2.5-fold (as determined by densitometry) in ␣ 5 -transfected cells, whereas it had no effect in Neo control cells. RI mRNA levels remained the same in ␣ 5 -transfected and Neo-transfected cells (data not shown). TGF-␤ receptor cross-linking with 125 I-TGF-␤ 1 showed little binding to RII of control cells, whereas substantially higher binding of TGF-␤ 1 was observed in the ␣ 5 transfectants (Fig. 5). Interestingly, binding of TGF-␤ 1 to RI was fairly prominent in control cells, as was binding to RIII. This has been observed in previous studies of this cell line (21,22), as well as in other cell lines (38). Transfection of ␣ 5 subunit resulted in a substantial increase in RI binding. This is in accordance with the increased RII expression because this receptor is thought to be responsible for RI recruitment to the cell surface (35).
Induction of Autocrine Negative TGF-␤ Activity by ␣ 5 ␤ 1 Transfection-To determine whether autocrine negative TGF-␤ activity was induced as a result of ␣ 5 ␤ 1 expression, TGF-␤ 1 neutralizing antibody blockade of endogenously produced TGF-␤ 1 was employed, utilizing a previously described clonal assay (25,26). Cells expressing autocrine negative TGF-␤ activity will show enhanced colony formation and growth as a result of the antibody-mediated neutralization of TGF-␤ 1 , whereas those cells that do not express autocrine negative activity will be unaffected by antibody treatment. Cell growth and colony formation are determined by crystal violet staining. Standard curves were performed measuring crystal violet levels with known numbers of MCF-7 cells to ensure that the assay was performed over a linear range of MCF-7 cells. As shown in Fig. 6, TGF-␤ 1 neutralizing antibody stimulated colony formation in ␣ 5 transfectants, whereas it had no effect on Neo control cells (Fig. 6A). The percentage of stimulation by the antibody was calculated and plotted in Fig. 6B. TGF-␤ 1 neutralizing antibody treatment resulted in 25% stimulation for clone 5 and 55% for clone 15.
To confirm the enhancement of autocrine TGF-␤ activity after ␣ 5 expression, we compared the activity of a TGF-␤responsive promoter in control cells with that in ␣ 5 -transfected cells. The p3TP-Lux promoter contains a TGF-␤ response element from the plasminogen activator inhibitor gene inserted upstream of the luciferase reporter gene and has been extensively utilized as a marker for TGF-␤ responsiveness (23,35). Therefore, it would be expected that induction of autocrine TGF-␤ activity would result in enhanced expression of the p3TP-Lux construct in ␣ 5 transfectants relative to Neo control cells. Fig. 7 shows that both ␣ 5 transfectants expressed 5-10fold higher levels of luciferase activity than Neo control. If increased luciferase activity of p3TP-Lux construct was due to autocrine TGF-␤, neutralizing antibody treatment would reduce expression of the reporter construct. As shown in Fig. 7A, TGF-␤ 1 neutralizing antibody treatment resulted in a substan- tial decrease in luciferase reporter activity in both ␣ 5 -transfected clones, whereas it had no effect on Neo control cells. This experiment was repeated four times, and similar results were obtained. Similarly, an ␣ 5 neutralizing antibody was used to show that disruption of ␣ 5 ␤ 1 ligation to FN resulted in approximately a 60% reduction in the enhanced luciferase activity associated with the ␣ 5 clone 15 transfectant (Fig. 7B), thus confirming that the enhanced endogenous TGF-␤ activity was dependent on the ectopic ␣ 5 expression.
We previously showed that autocrine TGF-␤ controlled steady state levels of FN in both native and RII-transfected cells (25,32). Consequently, induction of autocrine TGF-␤ should be associated with increased endogenous FN expression by ␣ 5 transfectants. The ␣ 5 -transfected cells showed a 3-fold increase (as determined by densitometry) in FN mRNA levels compared with Neo control cells (Fig. 8A). FN mRNA levels were further increased in ␣ 5 -transfected cells when the cells were plated on FN, whereas the levels in Neo control cells still remained the same. The enhanced FN expression was due to autocrine TGF-␤ as shown by the ability of TGF-␤ neutralizing antibody treatment to repress FN expression in ␣ 5 transfectant cells to a level comparable to that of NEO controls (Fig. 8B).
Growth Inhibitory Effects of TGF-␤ 1 on ␣ 5 Transfectants-Induction of autocrine TGF-␤ activity suggested that response to exogenous TGF-␤ effects should also result from ␣ 5 transfection. The MCF-7 LDC4 Neo pool was insensitive to TGF-␤ 1 in the absence or presence of exogenous FN (Fig. 9). The ␣ 5 transfectants showed reduced basal proliferation relative to NEO controls as indicated above in Fig. 3C. DNA synthesis in the ␣ 5 transfectants was further inhibited by TGF-␤ 1 in a dose-dependent manner (Fig. 9). When the two ␣ 5 transfectants were plated in 24-well culture plates coated with exogenous FN (10 g/ml), increased sensitivity to TGF-␤ 1 was demonstrated. Increased sensitivity on FN was likely due to the increased RII expression when cells were grown on FN as demonstrated in Fig. 4, above.
Effect of Integrin Expression on Tumorigenicity-To assess the effect of ␣ 5 expression on the malignant properties of MCF-7 cells, we inoculated Neo control and ␣ 5 -transfected clone 15 into ovariectomized, estrogen-supplemented nude mice as described previously (21). The size of xenografts formed was monitored with time (Fig. 10). Initially, MCF-7 LD 4 Neo pool and ␣ 5 clone 15 formed similar sized xenografts (Ͻ 200 FIG. 6. Effect of TGF-␤ 1 neutralizing antibody on the plating efficiency of MCF-7 LDC4 integrin ␣ 5 sense transfectants. A, cells (400) were plated on FN-coated (10 g/ml) 24-well plates either in the presence or absence of neutralizing TGF-␤ antibody (1:100). Cell growth was quantitated by crystal violet staining 8 days later. B, after staining, cell colonies were dissolved in 1% Triton X-100, and absorbance measurements at 595 nm and cell number are expressed as the percentage of stimulation. Each value is the mean Ϯ S.E. of four replicates.

FIG. 7. Effect of TGF-␤ 1 neutralizing antibody on the transcription of a TGF-␤-responsive promoter.
A, 3TP-Lux and pSV␤galactosidase plasmids were transiently transfected into Neo, ␣ 5 S.5, and ␣ 5 S5.15 cells by electroporation, and luciferase activity was measured 48 h after transfection in cells treated with IgG or TGF-␤ neutralizing antibody (10g/ml). Values are means of duplicate samples. B, 3TP-Lux transfectants were treated with either ␣ 5 antibody or TGF-␤ neutralizing antibody (10 g/ml). mm 3 ) until day 8 after inoculation. After day 8, growth was delayed in ␣ 5 clone 15 compared with Neo control. At day 28, Neo controls formed ϳ2.2-fold larger xenografts than ␣ 5 clone 15. This result indicates that integrin ␣ 5 ␤ 1 expression in MCF-7 cells can partially reverse the malignant properties of the cell line. DISCUSSION A number of studies have indicated that loss or reduced expression of integrin receptors results in abnormal cell growth (10,11,12). To test the hypothesis that ␣ 5 ␤ 1 ligation has tumor suppressor effects mediated through autocrine TGF-␤, we restored integrin ␣ 5 expression in MCF-7 cells. The MCF-7 cell line provided a good model system for this study in that it expressed low amounts of ␣ 5 integrin and was insensitive to the growth inhibitory effects of TGF-␤ 1 due to repression of RII expression (21)(22)(23). Integrin ␣ 5 transfectants expressed similar levels of cell surface integrin as Hs578T cells, another breast cancer cell line that was sensitive to growth inhibitory effects of TGF-␤ 1 . 2 The ␣ 5 transfection resulted in an increase in expression of RII, which was accompanied by increased autocrine TGF-␤ activity as assessed by 1) enhanced clonality following TGF-␤ neutralizing antibody treatment; 2) decreased endogenous activity of a TGF-␤-sensitive reporter system in response to TGF-␤ neutralizing antibody treatment with either TGF-␤ 1 or ␣ 5 antibodies; and 3) stimulation of FN expression, which was reversed by TGF-␤ neutralizing antibody treatment. Upregulation of RII expression was also reflected by increased sensitivity to inhibition by treatment with exogenous TGF-␤.
Previously, we showed that blockade of FN/␣ 5 ␤ 1 ligation by antibodies against FN or integrin ␣ 5 subunit stimulated DNA synthesis in cancer cell lines with moderate to high ␣ 5 ␤ 1 cell surface expression (13,14). These results are consistent with a previously described model that suggests that moderate adhesion to a loosely organized extracellular matrix facilitated both migration and growth, but strong adhesion to a fully organized extracellular matrix suppressed proliferation and contributed to inhibition of growth (4). Thus, the low level of expression of integrin ␣ 5 ␤ 1 on the cell surface of wild type MCF-7 cells could contribute to weak adhesion and hence abnormal growth in MCF-7 cells. Ectopic ␣ 5 expression leads to results consistent with a model suggesting that higher ␣ 5 ␤ 1 surface expression allows for greater adhesion due not only to ␣ 5 , but to greater endogenous FN expression as well. Inhibition of proliferation is either due to or augmented by the generation of autocrine TGF-␤ activity. Exogenous FN coating allows for even stronger adhesion and further enhancement of autocrine TGF-␤ activity. Most importantly, the results indicate that ␣ 5 ␤ 1 ligation and autocrine TGF-␤ interact in a reciprocal manner that is self-sustaining for both autocrine negative activity and cellextracellular matrix interactions. Moreover, this interaction is tumor-suppressive. This model may well apply to other systems given that autocrine TGF-␤ signaling and ␣ 5 ␤ 1 ligation have both been individually associated with tumor suppression in various model systems. Autocrine TGF-␤ has been shown to control steady state ␣ 5 expression (32) in model systems that also show tumor-suppressive TGF-␤ function (28).
Our results suggest that ␣ 5 ␤ 1 may have a negative growth regulatory role in some cancer cells and normal cells through modulation of TGF-␤ sensitivity. Apparently, signal transduction mechanisms for activating the TGF-␤ pathway are essentially intact in MCF-7 cells because restoration of ␣ 5 ␤ 1 expression leads to autocrine as well as exogneous TGF-␤ inhibitory responses. However, other types of cancer cells may be resistant to this mode of regulation despite ␣ 5 ␤ 1 expression because of perturbations of the TGF-␤ pathway resulting from malignant transformation. For example, HCT116 colon carcinoma cells express high levels of integrins that mediate adhesion to FN, but this cell line still exhibits a highly malignant phenotype due to a mutated RII gene (25). It is also possible that downstream signaling messengers encoded by oncogenes or tumor suppressor genes that participate in either a primary or secondary manner in signal transduction are abnormally modulated in some cell types.
An interesting aspect of this study was the enhancement of RII expression and TGF-␤ function when ␣ 5 transfectants were plated on FN-coated plates. These results, along with the demonstration that ␣ 5 antibody treatment blocks autocrine TGF-␤ activity, show that ␣ 5 ␤ 1 ligation to FN is critical to the generation of RII expression and hence TGF-␤-mediated signal transduction. Enhanced basal expression of FN in MCF-7 ␣ 5 transfectants that were not grown on FN-coated plates was also observed. We postulate that increased steady state expression of FN after ␣ 5 transfection allowed for ␣ 5 ␤ 1 ligation, which was critical to the basal RII expression and TGF-␤ sensitivity associated with transfectant cells that were not exposed to FN-coated plates.
In a previous study, we reported that disruption of ␣ 5 ␤ 1 /FN ligation resulted in stimulation of DNA synthesis (14). DNA synthesis was associated with up-regulated CDK2 activity without alterations of CDK inhibitors. DNA synthesis was also found to be dependent upon extracellular receptor kinase 1 and 2 activation. Thus, this previous study indicated ␣ 5 ␤ 1 ligation to FN had a repressive effect on cell cycle progression through repression of cyclin A and CDK2 expression. Exogenous treatment with TGF-␤ has been shown to down-regulate cyclin A (39). Interestingly, TGF-␤ has also been reported to repress ERK1/2 activation in some types of cells (40). This suggests that DNA synthesis resulting from disruption of ␣ 5 ␤ 1 ligation may be a reflection of the disruption of integrin related autocrine TGF-␤ activity resulting in up-regulation of ERK activation and subsequent promotion of cell cycle transit.
There is evidence that integrins transduce signals cooperatively with other growth factor systems in the regulation of cell proliferation. The proliferative response of murine mammary carcinoma cells to platelet-derived growth factor-BB and basic fibroblast growth factor is dependent on the extracellular matrix environment, indicating that modification of extracellular matrix and/or surface integrin receptors may regulate responsiveness to these growth factors (41). Reciprocal enhancement of ␣ 5 ␤ 1 -mediated adhesion by insulin and of insulin-mediated signal transduction by ␣ 5 ␤ 1 have recently been reported (42). However, ␣ 5 ␤ 1 ligation does not appear to modulate expression of insulin receptor in this system. Our results indicate that the TGF-␤ signaling pathway can be rescued by re-expression of integrin in MCF-7 cells. Ligation of integrins with their extracellular matrix ligands has been shown to regulate gene expression in a number of studies (43). However, because studies on the regulation of RII mRNA transcription and stability are limited, it is not yet clear how ligation of ␣ 5 ␤ 1 integrin to FN increases RII mRNA level. Nevertheless, the induction of RII and autocrine TGF-␤ activity by ␣ 5 ␤ 1 ligation suggests that integrin-mediated signal transduction plays a cooperative role with TGF-␤ signal transduction in tumor suppression. Moreover, the results indicating reciprocal positive control of autocrine TGF-␤ and ␣ 5 ␤ 1 ligation suggest that TGF-␤ signal trans- Neo pool and ␣ 5 sense clone 15 (cl15) were subcutaneously inoculated into ovariectomized athymic nude mice supplemented with 17␤-estradiol. Tumors were measured externally on the indicated days in two dimensions using a caliper. Xenograft volume was determined from the equation V ϭ (L ϫ W 2 ) ϫ 0.5, where L is the length and W is the width of the tumor. Each point represents the mean Ϯ S.E. of 10 xenografts. duction and ␣ 5 ␤ 1 integrin signal transduction participate in a mutually self-sustaining tumor-suppressive autocrine loop.