Activation of the p21 pathway of growth arrest and apoptosis by the beta 4 integrin cytoplasmic domain.

The integrin alpha 6 beta 4, a receptor for members of the laminin family of basement membrane components, contributes to the function of epithelial cells and their oncogenically transformed derivatives. In our efforts to study alpha 6 beta 4-mediated functions in more detail and to assess the contribution of the beta 4 cytoplasmic domain in such functions, we identified a rectal carcinoma cell line that lacks expression of the beta 4 integrin subunit. This cell line, termed RKO, expresses alpha 6 beta 1 but not alpha 6 beta 4, and it interacts with laminin-1 less avidly than similar cell lines that express alpha 6 beta 4. We expressed a full-length beta 4 cDNA, as well as a mutant cDNA that lacks the beta 4 cytoplasmic domain, in RKO cells and isolated stable subclones of these transfectants. In this study, we report that subclones that expressed the full-length beta 4 cDNA in association with endogenous alpha 6 exhibited partial G1 arrest and apoptosis, properties that were not evident in RKO cells transfected with either the cytoplasmic domain mutant or the expression vector alone. In an effort to define a mechanism for these observed changes in growth, we observed that expression of the alpha 6 beta 4 integrin induced expression of the p21 (WAF1; CiP1) protein, an inhibitor of cyclin-dependent kinases. These data suggest that the beta 4 integrin cytoplasmic domain is linked to a signaling pathway involved in cell cycle regulation in the beta 4 transfected RKO cells.

The integrin ␣ 6 ␤ 4 , a receptor for members of the laminin family of basement membrane components, contributes to the function of epithelial cells and their oncogenically transformed derivatives. In our efforts to study ␣ 6 ␤ 4 -mediated functions in more detail and to assess the contribution of the ␤ 4 cytoplasmic domain in such functions, we identified a rectal carcinoma cell line that lacks expression of the ␤ 4 integrin subunit. This cell line, termed RKO, expresses ␣ 6 ␤ 1 but not ␣ 6 ␤ 4 , and it interacts with laminin-1 less avidly than similar cell lines that express ␣ 6 ␤ 4 . We expressed a full-length ␤ 4 cDNA, as well as a mutant cDNA that lacks the ␤ 4 cytoplasmic domain, in RKO cells and isolated stable subclones of these transfectants. In this study, we report that subclones that expressed the full-length ␤ 4 cDNA in association with endogenous ␣6 exhibited partial G 1 arrest and apoptosis, properties that were not evident in RKO cells transfected with either the cytoplasmic domain mutant or the expression vector alone. In an effort to define a mechanism for these observed changes in growth, we observed that expression of the ␣ 6 ␤ 4 integrin induced expression of the p21 (WAF1; CiP1) protein, an inhibitor of cyclin-dependent kinases. These data suggest that the ␤ 4 integrin cytoplasmic domain is linked to a signaling pathway involved in cell cycle regulation in the ␤ 4 transfected RKO cells.
The integrin ␣ 6 ␤ 4 is a receptor for members of the laminin family of basement membrane components. Initial studies established that ␣ 6 ␤ 4 is a receptor for laminin-1 (1)(2)(3), and subsequent work has shown that it also functions as a receptor for other laminin isoforms (4,5). In its capacity as a laminin receptor, ␣ 6 ␤ 4 is involved in the formation and maintenance of hemidesmosomes (6 -8) and in the dynamic adhesion and migration of carcinoma cells (2,3). Most likely, other functions of epithelial and carcinoma cells are dependent upon ␣ 6 ␤ 4 because it plays such a pivotal role in mediating their interac-tions with laminin matrices. It is widely assumed that the unusually large and structurally unique cytoplasmic domain of the ␤ 4 integrin subunit associates with cytoskeletal and signaling molecules and that such associations provide the basis for the distinct functions associated with ␣ 6 ␤ 4 (5,8,9).
In our efforts to study ␣ 6 ␤ 4 -mediated functions in more detail and to assess the contribution of the ␤ 4 cytoplasmic domain in such functions, we identified a rectal carcinoma cell line that lacks expression of the ␤ 4 integrin subunit. This cell line, termed RKO, expresses ␣ 6 ␤ 1 but not ␣ 6 ␤ 4 (2,3). In this study, we report that RKO transfectants, which expressed the fulllength ␤ 4 cDNA in association with endogenous ␣ 6 exhibited G 1 arrest and a basal rate of apoptosis, properties that were not evident in RKO cells transfected with either a ␤ 4 cytoplasmic domain mutant or the expression vector alone. In an effort to define a mechanism for these observed changes in growth, we observed that expression of the ␣ 6 ␤ 4 integrin induced expression of the p21 (WAF1; Cip1) protein, an inhibitor of G 1 cyclindependent kinases (10 -12). These data suggest that the ␤ 4 integrin cytoplasmic domain is linked to a signaling pathway involved in cell cycle regulation in the ␤ 4 -transfected RKO cells.

MATERIALS AND METHODS
Cloning of the ␤ 4 Integrin Subunit-A full-length ␤ 4 cDNA clone was isolated from the Clone A colon carcinoma cDNA library (Clontech). This library was screened using a polymerase chain reaction product that was obtained from a partial sequence of the ␤ 4 subunit that we previously published (13). Multiple clones that encompassed the fulllength ␤ 4 cDNA sequence were isolated and subcloned into pBluescript (Stratagene). The full-length ␤ 4 cDNA was ligated into the pRc/CMV expression vector (Invitrogen) using the ␤ 4 restriction sites BglII and BssHII and the unique EcoRI vector site. This cDNA lacked the 70-and 7-amino acid splice variants and the 5Ј upstream 49-base pair region (13,14). In addition, a ␤ 4 cDNA with a truncated cytoplasmic tail termed ␤ 4 -⌬CYT was constructed by polymerase chain reaction of a 400-base pair region from base pairs 2010 -2398 (14), which introduced a stop codon and an XbaI site after the first 4 amino acids of the tail. This product was digested with SmaI and XbaI and ligated together with the 5Ј end of ␤ 4 (digested with EcoRI and SmaI) into the mammalian expression pcDNA3 (Invitrogen) via the unique EcoRI and XbaI sites in this vector.
Transfections and Flow Cytometry-The rectal carcinoma cell line RKO (15) was obtained from M. Brattain (Medical College of Ohio) and maintained in RPMI 1640 medium supplemented with 25 mM HEPES buffer, 10% FCS, 1 1% penicillin/streptomycin, and 1% L-glutamine (Life Technologies, Inc.). RKO cells were transfected with either the fulllength ␤ 4 construct, the ␤ 4 -⌬CYT construct, or vector alone using 20 l of Lipofectin reagent (Life Technologies, Inc.) and 10 -15 g of plasmid DNA. Geneticin (G418; Life Technologies, Inc.) was added to the growth medium at a concentration of 0.5 mg/ml for selection purposes. The transfected cells were sorted by FACS using the ␤ 4 -specific mAb UM-A9 (16) (provided by T. Carey, Michigan). FACS sorting was repeated twice to obtain a population of cells that exhibited significant surface expression of ␤ 4 . These populations were then subcloned by single cell sorting into 96-well plates (Costar).
Growth Assays-The transfected cells were trypsinized, washed in divalent cation-free PBS (CMF-PBS), and resuspended in 5% FCS growth medium containing G418. Cells (1 ϫ 10 4 ) were plated in triplicate 12-well plates (Costar) and allowed to grow for 3-6 days. Each well was harvested by trypsinization, and the number of cells was counted using a Coulter counter.
For propidium iodide staining, cells at approximately 50% confluency * This work was supported by National Institutes of Health Grant CA44704. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. were plated overnight on either tissue culture plastic, EHS-laminin (4.2 g/cm 2 ), or laminin-5 (0.2 g/cm 2 ; provided by R. Burgeson) and then harvested by trypsinization. The cells were stained with propidium iodide (Sigma; 2 mg/ml in 4 mM sodium citrate containing 3% (w/v) Triton X-100 and RNase A (0.1 mg/ml)). The stained cells were analyzed by FACS.
DNA Fragmentation-Samples were prepared as described in Green and co-workers (17) for analysis of DNA fragmentation and analyzed by agarose gel electrophoresis (1.2%).
Cells for immunohistochemistry were plated for 18 h in complete growth medium on coverslips precoated with either poly-L-lysine (10 g/cm 2 ) or EHS-laminin (5 g/cm 2 ). Cells were fixed for 8 min in 4% paraformaldehyde, permeabilized for 2 min in 0.2% Triton X-100, and stained with the p21 (10 g/ml) and a fluorescein-conjugated donkey anti-mouse IgG (Jackson Laboratories; 1:30). The cells were examined using a confocal microscope (Bio-Rad MRC 600, Bio-Rad Microsciences, Cambridge, MA) attached to a Zeiss Axiovert 35 equipped with a ϫ63 Plan-Neofluar objective.

RESULTS
Representative RKO subclones from the full-length ␤ 4 cDNA and the ␤ 4 -⌬CYT transfections that expressed varying levels of ␤ 4 surface expression were chosen for functional studies (Fig.  1A). No ␤ 4 expression was evident in subclones obtained from transfection of the expression vector alone (Neo). The association of the transfected ␤ 4 subunits with endogenous ␣ 6 was confirmed by immunoprecipitation of surface-biotinylated cells with the A9 mAb (data not shown). Also, the full-length ␤ 4 transfectants exhibited increased adhesion, spreading, and migration on laminin-1 providing evidence that the expressed ␣ 6 ␤ 4 integrin is functional (18).
Initially, we observed that bulk sorts of the full-length ␤ 4 transfectants did not maintain high levels of ␤ 4 surface expression for more than 3-4 days after sorting. To gain insight into the behavior of these transfectants, we assessed their DNA content using propidium iodide. A significant A o peak, characteristic of cells undergoing apoptosis (19), was evident in the bulk transfectants (Fig. 1B). Stable subclones that expressed full-length ␤ 4 cDNA maintained ␤ 4 surface expression, although the level of expression was less than that seen in the initial bulk sorts (Fig. 1A). These subclones grew noticeably more slowly than the mock transfectants. Propidium iodide staining revealed that the number of cells in G 1 was significantly greater in these transfectants than in the mock transfectants (Fig. 1C). The D4 subclone, for example, exhibited twice the percentage of total cells in G 1 compared with the mock transfectants. However, subclones that expressed the ␤ 4 -⌬CYT subunit on the surface at levels comparable with that of full-length ␤ 4 exhibited no increase in the number of cells in G 1 . To extend these observations, the doubling time for each of the subclones was determined ( Table I). The doubling time for RKO cells, as well as for the mock transfectants, is approxi-mately 21 h. In contrast, the doubling times for subclones that expressed full-length ␤ 4 ranged from 25 to 39 h, and these times correlated with the level of ␤ 4 surface expression (cf. Table I and Fig. 1). However, the ␤ 4 -⌬CYT subclones exhibited doubling times similar to the Neo subclones, i.e. 20 -21 h. Neither DNA content nor doubling times were affected significantly by growth of the subclones on EHS laminin-1 or laminin-5 compared with tissue culture plastic.
The results described above suggested that expression of ␣ 6 ␤ 4 could induce partial G 1 arrest and possibly apoptosis in RKO cells. To examine this possibility, we assayed ApopTag reactivity by FACS. ApopTag is a fluorescein-conjugated antibody that recognizes digoxigenin-tagged 3Ј-OH DNA ends generated by DNA fragmentation, and its use for the detection of apoptotic cells has been documented (20). As shown in Table II, approximately 8% of the D4 transfectants and 3% of the B8 transfectants were ApopTag positive. In comparison, fewer than 2% of the ␤ 4 -⌬CYT or mock transfectants were ApopTag positive. Attachment to either EHS laminin-1 (data not shown) or laminin-5 (Table II) did not alter the pattern of ApopTag staining. In addition, DNA fragmentation was evident in the B8 and D4 subclones but not in the ␤ 4 -⌬CYT transfectants (Fig.  2). These results indicate that a low, but significant, rate of apoptosis occurs in the full-length ␤ 4 subclones. A possible mechanism for the observed changes in growth that correlate with ␣ 6 ␤ 4 expression was suggested by the report that RKO cells express relatively low levels of wild-type p53 (21). Moreover, the growth-suppressive function of p53 can result from its ability to induce expression of p21 (WAF1; Cip1), an inhibitor of G 1 cyclin-dependent kinases (10 -12). Based on these observations, we hypothesized that expression of the ␣ 6 ␤ 4 integrin in RKO cells increases p21 expression. This hypothesis was assessed initially by immunostaining using a p21-specific mAb. The results obtained revealed little p21 expression in either the Neo or ␤ 4 -⌬CYT subclones (Fig. 3, A and B). In these subclones, fewer than 5% of the cells exhibited p21 staining. However, nuclear staining of p21 was much more evident in both the B8 and D4 subclones (Fig. 3, C and D). The frequency of nuclear staining was greater in the D4 subclone (30 -40% cells stained) than it was in the B8 subclone (15-20% cells stained), an observation that correlates with the relative level of ␣ 6 ␤ 4 expression in these two subclones. The expression of p21 in the subclones was also assessed by immunoblotting detergent extracts prepared from the ␤ 4 subclones. Relatively little p21 expression was evident in either the Neo or ␤ 4 -⌬CYT subclones based on these immunoblots (Fig. 3E). The level of p21 expression in these subclones is consistent with other reports of cells that express low levels of wild-type p53 (12). In contrast, a substantial increase in p21 expression was seen in both the B8 and D4 ␤ 4 subclones. Attachment to EHS laminin-1 did not alter this pattern of p21 expression (Fig. 3E). DISCUSSION We have shown that expression of the ␣ 6 ␤ 4 integrin in RKO cells, a ␤ 4 -deficient carcinoma cell line, results in partial G 1 arrest as well as the apoptotic death of some transfectants. A possible mechanism for these observed changes in growth is provided by our finding that ␣ 6 ␤ 4 also induces expression of the G 1 cyclin-dependent kinase inhibitor p21. The specificity of the observed induction of p21 expression is provided by the finding that the ␣ 6 ␤ 4 -⌬CYT integrin failed to affect p21 expression or growth even though it was expressed on the cell surface at levels comparable with the full-length integrin. In addition, we had previously observed that expression of either the E-cadherin (22) or galectin-3 cDNAs 2 in RKO cells had no effect on their growth.
Our finding that   26,28), ␤4-⌬CYT subclone (3C1, 5A4), and full-length ␤4 subclone (A7, B8, D4, F10) cells were plated in triplicate 12-well plates (Costar) and allowed to grow for 3-6 days as described under "Materials and Method." Each well was harvested by trypsinization, and the number of cells was counted using a Coulter counter. Doubling times were determined by the formula: ((t Ϫ t 0 )(log 2))/log(N Ϫ N 0 ), where t is end time (hours), t 0 is initial time (hours), N is final cell number, and N 0 is initial cell number. The results shown represent the average multiple experiments (ϮS.D.). The p values shown reflect the significance of the difference between a particular subclone and the mock transfectants.   1, 3, 5, 7, and 9) or EHS laminin-1 (lanes 2, 4, 6, 8, and 10) for 18 h. After detergent extraction, the samples were normalized for protein content, resolved by SDS-polyacrylamide gel electrophoresis (12%), transferred to nitrocellulose, and blotted with the p21-specific mAb. Bound protein was detected by enhanced chemiluminescence. Lanes 1 and 2, Neo-24, a mock transfectant subclone; lanes 3 and 4, Neo-26, a mock transfectant subclone; lanes 5 and 6, 3C1, a ␤ 4 -⌬Cyt subclone; lanes 7 and 8, B8, a full-length ␤ 4 subclone; lanes 9 and 10, D4, a full-length ␤ 4 subclone. transfectants is intriguing because our efforts to isolate transfectants with ␤ 4 surface expression greater than that observed in the D4 subclone, for example, were not successful. These transfectants did not survive for more than a few days after their selection by FACS. A reasonable interpretation for these data is that lower levels of ␣ 6 ␤ 4 surface expression induce expression of p21 sufficient to induce partial G 1 arrest and some apoptosis and that higher levels of ␣ 6 ␤ 4 expression result in more p21 expression and more widespread apoptotic death.
The ␤ 4 -dependent increase in growth arrest and p21 expression does not appear to depend on attachment to either a laminin-1 or laminin-5 matrix. For this reason, we suggest that the ␤ 4 cytoplasmic domain is linked constitutively to the p21 pathway of growth arrest and apoptosis in RKO cells. This possibility of constitutive activation of adhesion-dependent signaling pathways in transformed cells is supported, for example, by the recent report that the binding of focal adhesion kinase to SH2-containing proteins in v-src-transformed 3T3 cells is independent of integrin engagement and cell attachment (23).
The results of this study raise the issue of the role of ␣ 6 ␤ 4 in the regulation of normal cell growth and apoptosis. The intestinal epithelium may provide a superb model for examining such a relationship. In this epithelium, undifferentiated cells at the base of the crypts proliferate rapidly and give rise to differentiated enterocytes that migrate to the tips where they are sloughed off into the intestinal lumen. Apoptosis occurs at the tips and may provide the mechanism for cell loss in this structure (24). Interestingly, p21 is expressed in mature enterocytes but not in undifferentiated crypt cells (25), and there is evidence that ␣ 6 ␤ 4 is not expressed in the undifferentiated crypt cells. 3 For these reasons, the possibility of a functional relationship between ␣ 6 ␤ 4 expression and p21 induction merits investigation.
Several reports have implicated cell adhesion as well as specific adhesion receptors in growth control and apoptosis (26 -30). The significance of our findings is that they establish a link between the cytoplasmic domain of a specific integrin subunit and the expression of a molecule known to be critical for growth suppression, apoptosis, and tumorigenesis. These findings should facilitate the elucidation of the signaling pathway(s) that results in the induction of p21 expression by cell surface receptors.