Activation of p53 Function in Carcinoma Cells by the α6β4 Integrin*

The interaction of integrins with extracellular matrix is known to promote cell survival by inhibiting apoptotic signaling. In contrast, we demonstrate here that the α6β4 integrin induces apoptosis in carcinoma cells by stimulating p53 function. Specifically, we show that expression of α6β4 in carcinoma cells that lack this integrin stimulates an increase in the transactivating function of p53 as demonstrated by the ability of this integrin to up-regulate the expression of a p53-sensitive reporter gene as well as the endogenous p53 response gene, bax. In addition, we report that α6β4 triggers apoptosis in carcinoma cells that express wild-type but not mutant p53 and that these α6β4 functions are inhibited by a dominant negative p53 construct. Importantly, we provide a link between integrin signaling and p53 activation by demonstrating that the clustering of α6β4 with a β4integrin-specific antibody promotes p53-dependent apoptosis in cells that express both α6β4 and wild-type p53. These studies are the first to demonstrate that a specific integrin can promote apoptosis by activating p53. Moreover, given the ability of α6β4 to stimulate invasion (Shaw, L. M., Rabinovitz, I., Wang, H. F., Toker, A., and Mercurio, A. M. (1997) Cell 91, 949–960), these studies suggest that the ability of α6β4 to promote carcinoma progression will be enhanced in tumor cells that express mutant, inactive forms of p53.

The interaction of integrins with extracellular matrix is known to promote cell survival by inhibiting apoptotic signaling. In contrast, we demonstrate here that the ␣ 6 ␤ 4 integrin induces apoptosis in carcinoma cells by stimulating p53 function. Specifically, we show that expression of ␣ 6 ␤ 4 in carcinoma cells that lack this integrin stimulates an increase in the transactivating function of p53 as demonstrated by the ability of this integrin to up-regulate the expression of a p53-sensitive reporter gene as well as the endogenous p53 response gene, bax. In addition, we report that ␣ 6 ␤ 4 triggers apoptosis in carcinoma cells that express wild-type but not mutant p53 and that these ␣ 6 ␤ 4 functions are inhibited by a dominant negative p53 construct. Importantly, we provide a link between integrin signaling and p53 activation by demonstrating that the clustering of ␣ 6 ␤ 4 with a ␤ 4 integrin-specific antibody promotes p53-dependent apoptosis in cells that express both ␣ 6 ␤ 4 and wild-type p53. These studies are the first to demonstrate that a specific integrin can promote apoptosis by activating p53. Moreover, given the ability of ␣ 6 ␤ 4 to stimulate invasion ( Integrins are the primary receptors used by cells to interact with extracellular matrices. Although initial studies had emphasized the functional contribution of integrins to cell adhesion and migration, a significant finding was the observation that integrins are essential for cell survival (2,3). Specifically, epithelial cells, endothelial cells, and fibroblasts are prone to growth arrest and apoptosis when deprived of integrin-mediated contact with the extracellular matrix (4,5). To date, several integrins including ␣ 5 ␤ 1 (6 -7), ␣ v ␤ 3 (8 -10), and ␣ 6 ␤ 1 (11) have been implicated in the promotion of cell growth and survival.
Arguably, one of the most complex integrins in terms of both structure and function is ␣ 6 ␤ 4 . This integrin is distinguished structurally from other integrins on the basis of the unusually large cytoplasmic domain of its ␤ 4 subunit (12)(13)(14). Aside from its involvement in cell adhesion and migration (1,(15)(16)(17)(18)(19)(20), the ␣ 6 ␤ 4 integrin can promote growth arrest and apoptosis in some carcinoma cells. Specifically, we reported that ␣ 6 ␤ 4 expression induces the growth arrest and apoptosis of the RKO colon carcinoma cell line (21), a finding that has been substantiated in other carcinoma cell lines (22)(23), as well as in endothelial cells (24). These findings, however, conflict with considerable evidence that supports a role for ␣ 6 ␤ 4 in promoting carcinoma invasion and progression (1,19). In order to understand how ␣ 6 ␤ 4 can deliver these apparently conflicting signals, we analyzed the mechanism by which ␣ 6 ␤ 4 promotes apoptosis in more detail. Specifically, we examined the hypothesis that ␣ 6 ␤ 4 activates the p53 tumor suppressor.
To obtain expression of the dnp53 construct, RKO/mock and RKO/␤ 4 subclones were co-transfected using calcium phosphate with plasmids expressing the puromycin resistance gene (25) and a dominant negative p53 (dnp53) construct (provided by M. Oren, Weizmann Institute for Science, Israel) encoding for a carboxyl-terminal domain of p53 that can heterodimerize with endogenous p53 and inhibit its transcriptional activity. The transfected cells were subcloned, and those subclones that expressed high levels of dnp53 were selected by FACS 1 using the Pab122 mAb (Roche Molecular Biochemicals), which recognizes a conserved, denaturation stable epitope in dnp53. Mock-transfected RKO/␤ 4 subclone D4.3 and dnp53-expressing RKO/␤ 4 subclone D4.DD1.4c were selected for analysis. All assays were performed on early passage cells (less than 8 passages).
For transient expression of ␣ 6 ␤ 4 in bulk populations of RKO and dnp53-expressing RKO cells, the LipofectAMINE reagent was used to transfect these cells according to the manufacturer's instructions with the pRC-CMV full-length ␤ 4 cDNA or a control vector. After 48 h, the cells were double-stained with the ␤ 4 -specific mAb 439 -9B followed by a secondary (R)-phycoerythrin (PE)-conjugated goat anti-rat serum and annexin V-FITC. The stained cells were analyzed by FACS, and the ␤ 4 -positive population was gated and analyzed for annexin V reactivity. For transient expression of E1B in ␣ 6 ␤ 4 -expressing RKO subclones, RKO/␤ 4 cells were transfected using the calcium phosphate-DNA precipitation technique with 1 g of plasmid expressing the puromycin resistance gene (25) in the presence or absence of 20 g of the pCMV-55K plasmid (kindly provided by A. Zantema, Leiden University, The Netherlands), encoding the adenovirus E1B 55-kDa protein of type 5 adenovirus under the control of the CMV promoter. After 24 h, transfected cells were selected with 2 g/ml puromycin for 48 h and then analyzed for E1B 55-kDa expression by indirect immunofluorescence and for annexin V FITC and PI reactivity by FACS. Integrin Clustering Experiments-Cells were harvested and incubated in suspension with either the 439-9B (␤ 4 -specific) or SAM1 (␣5specific; Immunotech) monoclonal antibody at a concentration of 10 g/ml for 1 h on ice. These cells were washed with phosphate-buffered saline before plating them in wells (2 ϫ 10 5 cells per well) of a 12-well tissue culture plate (Costar) that had been coated overnight at 4°C with either goat anti-rat IgG (Jackson ImmunoResearch) or goat anti-mouse IgG (Jackson ImmunoResearch) and blocked for 1 h at 37°C with 1% bovine serum albumin.
Analysis of Apoptosis-To assess annexin V reactivity, cells were stained with annexin V FITC (Bender MedSystems) or annexin V PE (PharMingen) using the manufacturers' recommended protocols. For in situ analysis of apoptosis in cells transfected transiently with the green fluorescence protein (GFP)-expressing vector pEGFP-1 (CLONTECH) and dnp53, adherent and supernatant cells were combined, stained with annexin V-PE (PharMingen) according to the manufacturer's directions, and plated on coverslips. The percentage of GFP-positive cells that were annexin V PE positive was determined by fluorescence microscopy. To assess ApopTag reactivity, cells were fixed and subjected to ApopTag reactions (Oncor) following the methods recommended by the manufacturer. Samples (10 4 cells) were analyzed by flow cytometry.
Analysis of p53 Activity-Cells were co-transfected using the Lipofectin reagent (Life Technologies, Inc.) with 1 g of CMV-␤-galactosidase and 5 g of either PG 13 CAT, a reporter gene plasmid consisting of a polyoma early promoter and the chloramphenicol acetyltransferase (CAT) gene downstream of a p53-binding motif (PG 13 CAT), or MG 15 CAT, the control plasmid that contains mutant p53-binding sites. Both plasmids were provided by Dr. Bert Vogelstein. The amount of CAT enzyme in equivalent amounts of total protein from whole cell extracts was determined using an enzyme-linked immunosorbent assay (Roche Molecular Biochemicals). The data obtained were normalized for transfection efficiency by assaying ␤-galactosidase.

RESULTS
Ectopic Expression of the ␣ 6 ␤ 4 Integrin Activates p53-dependent Apoptosis-We and others (21)(22)(23) have shown that the ␣ 6 ␤ 4 integrin can induce apoptosis in carcinoma cells. A possible involvement of p53 in this ␣ 6 ␤ 4 function was explored by comparing the effects of ␣ 6 ␤ 4 expression on the apoptosis of ␤ 4 -deficient carcinoma cells that differed in their p53 status. Stable transfectants of RKO colon carcinoma and MDA-MB-435 breast carcinoma cells were generated that expressed either the ␣ 6 ␤ 4 integrin (␤ 4 ) or a cytoplasmic domain deletion mutant of ␤ 4 (␤ 4 -⌬cyt). RKO colon carcinoma cells express wild-type p53, and MDA-MB-435 breast carcinoma cells contain a homozygous point mutation in the p53 gene that inhibits the apoptotic functions of this tumor suppressor (26). ␣ 6 ␤ 4 expression resulted in a significant induction of apoptosis in RKO but not in MDA-MB-435 cells, as determined by annexin V FITC staining (Table I), as well as ApopTag staining (Table  I). The difference in the magnitude of ␣ 6 ␤ 4 -induced apoptosis of the two RKO/␤ 4 subclones analyzed correlates with their level of ␣ 6 ␤ 4 surface expression (Table I). Of note, we were only able to select RKO/␤ 4 clones with relatively low levels of surface ␣ 6 ␤ 4 because of the extensive cell death observed in cells expressing higher levels of this integrin, as we reported previously (21).
The importance of p53 in ␣ 6 ␤ 4 -mediated apoptosis was examined further by inhibiting p53 function in the RKO/␤ 4 cells. Expression of a dnp53 construct blocked their apoptosis, as determined by annexin V FITC staining (Fig. 1A) and ApopTag reactivity (Fig. 1B). These data were corroborated by the finding that expression of the adenovirus E1B 55-kDa protein, another inhibitor of p53 activity (27), blocked ␣ 6 ␤ 4 -induced apoptosis in these cells (Fig. 1C). To demonstrate that the results described were not limited to one subclone of RKO/␤ 4 transfectants, mixed populations of mock-transfected RKO cells (RKO/mock) or dnp53-expressing RKO cells (RKO/dnp53) were transfected transiently with a full-length ␤ 4 cDNA (␤ 4 ) or a control vector (Mock) and subsequently stained with annexin V FITC. ␣ 6 ␤ 4 expression in RKO/mock, but not in RKO/dnp53 cells, induced a substantial increase in the percentage of annexin V FITC-positive cells (Fig. 1D), demonstrating that the ability of ␣ 6 ␤ 4 to induce p53-dependent apoptosis is not an artifact of clonal selection. We observed a higher level of apoptosis in RKO cells transfected transiently with ␤ 4 as compared with that observed in RKO/␤ 4 subclones, most likely because higher surface ␣ 6 ␤ 4 expression was achieved in the transient transfectants (data not shown).
Ectopic Expression of the ␣ 6 ␤ 4 Integrin Stimulates p53 Activity-Based on the above findings that ␣ 6 ␤ 4 expression in p53 wild-type but not in p53-mutant carcinoma cell lines can induce p53-dependent growth arrest and apoptosis, we investigated whether ␣ 6 ␤ 4 could stimulate p53 function using a p53-sensitive reporter gene plasmid (PG 13 CAT) (28). RKO/␤ 4 /PG 13 CAT cells displayed a significant increase in CAT activity in comparison to both RKO/mock/PG 13 CAT (Table II) and RKO/␤ 4 -⌬cyt/PG 13 CAT cells (Table II). This increase in CAT activity was seen in both of the RKO/␤ 4 subclones analyzed (Table II). The relative level of p53 activity correlated with the magnitude of apoptosis observed in these clones, with a higher level of ␣ 6 ␤ 4 expression, p53 activity, and apoptosis observed in RKO/␤ 4 clone 1 relative to RKO/␤ 4 clone 2 (Tables I and II). As a negative control for these experiments, these cells were transfected with the same reporter plasmid mutated in the p53binding sites (MG 15 CAT). No increase in CAT activity was observed in the RKO/␤ 4 /MG 15 CAT transfectants (Table II). We did not detect CAT activity in either MDA-MB-435 cells or MDA/␤ 4 subclones transfected with PG 13 CAT (Table II), as expected for cells expressing mutant forms of p53 that are functionally inactive. To corroborate the CAT activity data, we analyzed the relative expression of bax, a p53 target gene TABLE I ␣ 6 ␤ 4 induces apoptosis in p53 wild-type, but not in p53 mutant carcinoma cells Mock-transfected (mock), ␣ 6 ␤ 4 -⌬cyt (␤ 4 -⌬cyt), and ␣ 6 ␤ 4 (␤ 4 )-expressing RKO and MDA-MB-435 subclones were plated on poly-L-lysine for 48 h. To assess the level of apoptosis in these cells, they were either stained with annexin V-FITC (Bender MedSystems) and propidium iodide (PI), or subjected to ApopTag reactions (Oncor). These cells were analyzed on a Becton Dickinson flow cytometer using CellQuest software. Similar results were observed in three independent experiments for the annexin V stains. The ApopTag results are expressed as the percent ApopTag-positive cells (ϮS.E.) from three trials. To assess the relative surface expression of ␣ 6 ␤ 4 , these clones were incubated with either normal rat IgG or the ␤ 4 -specific antibody, 439 -9B, followed by FITC-conjugated anti-rat IgG and analyzed by flow cytometry.  2). In contrast, equivalent STAT-1 protein levels were detected in ␣ 6 ␤ 4 -expressing and mock-transfected RKO cells (data not shown). Importantly, the expression of ␣ 6 ␤ 4 did not increase Bax expression in either MDA-MB-435 cells (Fig. 2) or in dnp53-expressing RKO cells (data not shown), confirming that the ability of the ␣ 6 ␤ 4 integrin to up-regulate Bax expression is dependent on p53 activity. The decreased level of Bax in ␣ 6 ␤ 4expressing as compared with mock-transfected MDA-MB-435 cells (Fig. 2) may be attributable to the ability of ␣ 6 ␤ 4 to activate a transcriptional repressor of Bax, as was recently identified in T cells (32). Also, we are currently investigating whether the ␣ 6 ␤ 4 -associated decrease in Bax expression in MDA-MB-435 cells is related to our preliminary observations that ␣ 6 ␤ 4 can promote the survival of p53-deficient cells. 2 These studies demonstrate that the expression of ␣ 6 ␤ 4 in RKO cells is sufficient to induce p53 activity in the apparent absence of ␣ 6 ␤ 4 ligation or clustering, an observation that is consistent with other ligand-independent functions of this integrin (1,20,(22)(23)(24). Ligation of the ␣ 6 ␤ 4 Integrin Stimulates p53 Activity and Apoptosis in HCT116 Carcinoma Cells-An important question that arose from the above findings is whether ␣ 6 ␤ 4 activates p53 and promotes p53-dependent apoptosis in carcinoma cells that naturally express this integrin. We investigated whether clustering of ␣ 6 ␤ 4 with a ␤ 4 -specific antibody could stimulate the apoptosis of HCT116 cells, a colon carcinoma cell line that expresses wild-type p53 (33) and ␣ 6 ␤ 4 (data not shown). To demonstrate the specificity of the effects observed with ␣ 6 ␤ 4 clustering, we clustered the ␣ 5 ␤ 1 integrin. Importantly, the surface expression level of the ␣ 5 and ␤ 4 subunits is similar in HCT116 cells (data not shown). The clustering of ␣ 6 ␤ 4 resulted in a significant increase in the percentage of annexin V FITC ϩ , PI Ϫ HCT116 cells in comparison to the clustering of ␣ 5 ␤ 1 (Table  III). The apoptotic potential of HCT116 cells did not correlate with their degree of adherence or cell spreading following integrin clustering (Table III) (34). As evidence that ␣ 5 ␤ 1 clustering did not augment the basal level of apoptosis in these 2 R. E. Bachelder, unpublished data.  cells, we observed that HCT116 cells bound the same amount of annexin V FITC following the clustering of ␣ 5 ␤ 1 (Table III) or HLA antigens (data not shown), which have not been implicated in the promotion of apoptosis. The possibility that the increased level of apoptosis observed in HCT116 cells following ␣ 6 ␤ 4 clustering could be attributed to the inability of ␣ 5 ␤ 1 to deliver survival signals to these cells was discounted by the finding that either the simultaneous clustering of ␣ 5 ␤ 1 and ␣ 6 ␤ 4 integrins or the clustering of ␣ 6 ␤ 4 alone resulted in the same increase in the percentage of annexin V FITC ϩ , PI Ϫ HCT116 cells (data not shown). Based on our observation that ␣ 6 ␤ 4 stimulated p53 activity in the RKO cells, we postulated that ␣ 6 ␤ 4 clustering should also increase p53 activity in HCT116 cells. In fact, using the PG 13 CAT reporter gene construct, we observed that the clustering of ␣ 6 ␤ 4 , as compared with the clustering of ␣ 5 ␤ 1 , induced a 2.4-fold increase in p53 activity in HCT116 cells (␤ 4 clustering: 47.8 Ϯ 0.08 pg of CAT enzyme/mg of total protein versus ␣5 clustering: 19.6 Ϯ 0.01 pg of CAT enzyme/mg of total protein). This increase in p53 activity correlated with the 3.2-fold increase in the level of apoptosis observed in HCT116 cells following ␣ 6 ␤ 4 as compared with ␣ 5 ␤ 1 clustering (Table III). The CAT activity we detected in ␣ 5 ␤ 1 cross-linked, PG 13 CAT-transfected HCT116 cells represents the basal p53 activity in these cells based on our detection of similar levels of CAT activity in HLA cross-linked, PG 13 CAT-transfected HCT116 cells.
Finally, to establish definitively the importance of p53 in ␣ 6 ␤ 4 -mediated HCT116 cell apoptosis, we investigated the effect of dnp53 expression in the cells. As shown in Table III, dnp53 inhibited the apoptosis observed in HCT116 cells upon antibody-mediated ␤ 4 integrin clustering. Our demonstration that the antibody-mediated clustering of ␣ 6 ␤ 4 in HCT116 cells stimulates p53 activity, as well as p53-dependent apoptosis, provides direct evidence that p53 activity can be regulated by the signaling functions of this integrin. DISCUSSION Although p53 can be activated by a number of "stress" signals (35), few cell-surface receptors have been described that can activate p53 and trigger a p53-dependent apoptotic response (36 -38). We demonstrate here that the ␣ 6 ␤ 4 integrin, a cell adhesion receptor, can stimulate p53-transactivating function and promote p53-dependent apoptosis. This finding is in contrast with most studies on integrins that have focused on their ability to promote cell survival by inhibiting apoptotic signaling pathways (6,7,10,11,39), including the inhibition of p53 activity (10,40). Our finding that ␣ 6 ␤ 4 -dependent apoptosis is p53-dependent explains why the expression of the ␣ 6 ␤ 4 integrin has been reported to inhibit the growth of some carci-noma cell lines (21)(22)(23) but has no effect on the growth of other carcinoma cells (1). Relating to our previous demonstration that ␣ 6 ␤ 4 stimulates chemotactic migration and invasion (1,19,20), a major ramification of the current study is that carcinoma cells that express ␣ 6 ␤ 4 and inactive forms of p53 will have the propensity to progress more readily than those that express wild-type p53.
Our demonstration that antibody-mediated clustering of ␣ 6 ␤ 4 stimulates p53 activity and apoptosis in HCT116 cells provides direct evidence that p53 activity can be regulated by the signaling functions of this integrin. We also observed that the expression of ␣ 6 ␤ 4 in ␤ 4 -deficient carcinoma cells that express wild-type p53 is sufficient to induce p53 activity and p53-dependent apoptosis in the apparent absence of ␣ 6 ␤ 4 ligation or clustering. The ability of ␣ 6 ␤ 4 expression to induce apoptosis in a ligand-independent manner has also been observed in other cell types (22)(23)(24). This finding is consistent with recent reports that some functions of ␣ 6 ␤ 4 can be mediated entirely by the ␤ 4 cytoplasmic domain and are independent of the extracellular domains of this integrin (41). The ability of the ␤ 4 cytoplasmic domain to self-associate may account for this "ligand-independent" behavior (42). In contrast, cells that express this integrin endogenously may regulate its signaling as a means of stimulating the apoptotic function of this integrin only at the appropriate anatomical sites where ligand is available.
An important issue that arises from these data is the mechanism by which ␣ 6 ␤ 4 stimulates p53 activity. p53 can be modified by acetylation (43) and phosphorylation (44 -46), events that alter the stability and DNA binding activities of this tumor suppressor. It will be informative in future studies to determine whether the apoptotic function of ␣ 6 ␤ 4 can be attributed to its induction of such posttranslational modifications in p53, as well as to identify the domains of the ␤ 4 subunit important for this activity. Moreover, it will be important to determine whether the growth inhibition that has been reported to be mediated by ␤ 2 , ␤ 1 C, and ␤ 1 D integrins is also dependent on tumor suppressor activity (47)(48)(49)(50).
Based on our demonstration that ␣ 6 ␤ 4 delivers p53-dependent apoptotic signals to transformed cells, it will be informative in future studies to determine whether this integrin can also deliver these signals in non-transformed cells. Although ␣ 6 ␤ 4 has been reported to promote T cell receptor-driven thymocyte proliferation (51), as well as primary keratinocyte proliferation  MB-435 cells). Proteins from mocktransfected and ␤ 4 -transfected RKO and MDA-MB-435 (MDA) cells that had been serum-starved for 48 h were extracted in a modified RIPA buffer, normalized for protein content, and resolved by 12% SDS-polyacrylamide gel electrophoresis. These proteins were transferred to nitrocellulose and probed with a rabbit anti-human Bax polyclonal serum. As a control for protein loading, we demonstrated that equivalent amounts of STAT-1 protein were expressed in RKO/Mock, RKO/␤ 4 , MDA/Mock, and MDA/␤ 4 cells (data not shown). Similar results were observed in four additional trials.

TABLE III
Inhibition of ␣ 6 ␤ 4 -dependent HCT116 cell apoptosis by dnp53 HCT116 cells were transfected with pGFP in addition to either the dnp53-expressing or appropriate control vector. After 24 h, these cells were harvested and stimulated by antibody-mediated ␤ 4 or ␣ 5 integrin clustering, as described under "Experimental Procedures." These stimulated cells were stained with phycoerythrin-conjugated annexin V and examined by fluorescence microscopy. The percentage of GFP-positive cells that were also stained by annexin V-PE was determined by collecting data for at least 70 GFP-positive cells. Similar results were observed in three separate trials. To quantitate cell spreading, HCT116 cells were incubated with either an ␣ 5 -or ␤ 4 -specific monoclonal antibody and then plated on secondary antibody-coated wells, as described under "Experimental Procedures." After 1 h, the spread surface area of these cells was quantitated by digital image analysis. The cell spreading data are presented as the mean area (m 2 Ϯ S.D.) of 30 cells.  (52,53) and survival (54), this integrin can also induce the apoptosis of primary endothelial cells (24). The fact that ␣ 6 ␤ 4 is expressed in most epithelia, which by definition are growtharrested and differentiated, suggests that this integrin may contribute to epithelial renewal and differentiation. Interestingly, cells not only arrest in the G 0 /G 1 phase of the cell cycle during differentiation, but they also undergo events normally associated with apoptosis including DNA cleavage and chromosome condensation. In fact, p53 has been suggested to play an important role in cell differentiation in addition to its role in mediating growth arrest and apoptosis (55,56). The potential involvement of ␣ 6 ␤ 4 in differentiation is also supported by a number of studies that have concluded that the laminins, which are the predominant ligands for ␣ 6 ␤ 4 , are potent inducers of epithelial differentiation (57,58). Given these observations, further studies investigating the role of ␣ 6 ␤ 4 in epithelial differentiation are warranted.
In summary, these studies are the first to implicate an integrin in apoptosis by the activation of a tumor suppressor. The ability of the ␣ 6 ␤ 4 integrin to activate this growth inhibitory pathway likely plays an important role in the selection of mutations in the apoptotic pathway of ␣ 6 ␤ 4 -expressing carcinomas.