Expression of Functionally Distinct Variants of the β4A Integrin Subunit in Relation to the Differentiation State in Human Intestinal Cells*

Integrins are important mediators of cell-laminin interactions. In the small intestinal epithelium, which consists of spatially separated proliferative and differentiated cell populations located, respectively, in the crypt and on the villus, laminins and laminin-binding integrins are differentially expressed along the crypt-villus axis. One exception to this is the integrin α6β4, which is thought to be ubiquitously expressed by intestinal cells. However, in this study, a re-evaluation of the β4 subunit expression with different antibodies revealed that two forms of β4 exist in the human intestinal epithelium. Furthermore, we show that differentiated enterocytes express a full-length 205-kDa β4A subunit, whereas undifferentiated crypt cells express a novel β4A subunit that does not contain the COOH-terminal segment of the cytoplasmic domain (β4Actd−). This new form was not found to arise from alternative β4 mRNA splicing. Moreover, we found that these two β4A forms can associate into α6β4A complexes; however, the β4Actd− integrin expressed by the undifferentiated crypt cells is not functional for adhesion to laminin-5. Hence, these studies identify a novel α6β4Actd− integrin expressed in undifferentiated intestinal crypt cells that is functionally distinct.

Epithelial cells are characterized by special structural features such as polarized morphology, specialized cell-cell contacts, and their attachment to an underlying basement membrane (1). This attachment mediates various crucial cell functions including adhesion, migration, proliferation, differentiation, and cell survival. The biological effects of basement membranes are largely mediated by laminins, a growing family of structurally related molecules (2,3), which are expressed in a tissue-specific manner (4). Epithelia bind to laminins via various cell membrane receptors, many of which are members of the integrin superfamily (5)(6)(7)(8)(9) which can subsequently ini-tiate intracellular signaling cascades upon ligation to their ligand (10,11). Some integrins, such as ␣ 2 ␤ 1 and ␣ 3 ␤ 1 , are considered quite promiscuous (12) and bind various molecules including laminins, while others such as ␣ 6 ␤ 1 , ␣ 7 ␤ 1 , and ␣ 6 ␤ 4 are specific for laminins (8,13).
The integrin ␣ 6 ␤ 4 has a number of features which distinguish it structurally and functionally from the other laminin receptors of the ␤ 1 family (14 -16). The ␤ 4 subunit has an unusually long cytoplasmic domain having no homology with other integrin ␤ subunits. Upon binding to laminin-5, ␣ 6 ␤ 4 becomes concentrated in hemidesmosomes and is associated with intermediate filaments, making it essential for the organization and maintenance of the epithelial structure (17)(18)(19). This role was confirmed using transgenic mice in which the ␤ 4 subunit had been knocked out (20,21), causing the loss of hemidesmosomal structures. Furthermore, mice carrying a targeted deletion of the integrin ␤ 4 cytoplasmic domain showed defects in both cell adhesion and proliferation (22), suggesting that ␤ 4 coordinates hemidesmosome assembly but may be responsible for additional signals, some of which may be essential for cell survival and cell cycle regulation (22,23). Indeed, upon ligation to laminin-5, ␣ 6 ␤ 4 has been shown to become phosphorylated and to cause recruitment of Shc and Grb2-mSOS, which go on to activate Ras/MAP kinase pathways (17,24).
The intestinal epithelium, which is in constant and rapid renewal, represents an interesting model for the study of mechanisms involved in the determination of the cell state (25). Within its functional unit, the crypt-villus axis, are two main distinct cell populations: the proliferative and poorly differentiated crypt cells and the mature enterocytes of the villus (26 -28). Processes of epithelial cell growth and functional differentiation need to be tightly regulated along the crypt-villus axis (29,30). Analysis of integrins and basement membrane molecules in the human intestine has revealed particular patterns of expression for many of these molecules, namely those involved in laminin-cell interactions (13,25). Of interest is the reciprocal expression of laminin-1 and laminin-2 that has been found along the crypt-villus axis, with laminin-1 occurring as a villus form and laminin-2 as a crypt form, suggesting a relation between laminin form expression and cell differentiation (31,32). The functional relevance of these observations was provided by the demonstration that laminin-1, but not laminin-2, can precociously induce differentiation in intestinal cells (33), suggesting that laminin-1 is critical in triggering terminal enterocytic differentiation (33,34). A differential crypt-villus pattern of expression for laminin-binding integrins was also reported in the intestinal epithelium. The distribution of ␣ 2 ␤ 1 was mainly restricted to the crypt while ␣ 3 ␤ 1 was found predominantly in the villus (35,36). More recently, ␣ 7 B␤ 1 , a spe-cific laminin receptor largely involved in muscle development in response to laminin (37) has been identified in the intestinal epithelium. It was found to be located at the crypt-villus junction in the intact intestine while its expression was transiently up-regulated in differentiating enterocytes (38). In contrast, the integrin ␣ 6 ␤ 4 has been reported to be present at the base of intestinal epithelial cells all along the crypt-villus axis suggesting that this laminin receptor is ubiquitously expressed by intestinal cells (31,32,36,39).
In this study, the expression of the ␤ 4 subunit in intestinal cells was re-evaluated on the basis of the observation that antibodies directed to the COOH-terminal segment of the cytoplasmic domain of the molecule (40) detected the ␤ 4 subunit only in the villus cells of the intact small intestine, suggesting that two distinct forms of ␤ 4 may exist in the intestinal epithelium. To further investigate this, we used two well characterized intestinal cell models which allow to some extent the recapitulation of the crypt-villus axis in vitro: the crypt-like HIEC cells, which are proliferative and undifferentiated (41), and the Caco-2/15 cells which have the ability to differentiate into fully functional villus-like enterocytes (42)(43)(44). Our data show that differentiated enterocytes express a full-length 205-kDa ␤ 4 A subunit while undifferentiated crypt cells express a novel ␤ 4 A subunit which does not contain the COOH-terminal segment of the cytoplasmic domain (␤ 4 A ctdϪ ). Moreover, we found that these two ␤ 4 A variants form ␣ 6 ␤ 4 A complexes that are functionally distinct with regard to adhesion activity, the ␣ 6 ␤4A ctdϪ receptor expressed on undifferentiated crypt cells being inactive as opposed to the fully functional ␣ 6 ␤ 4 A integrin expressed on differentiated villus cells.

EXPERIMENTAL PROCEDURES
Tissues-Specimens of adult small intestine (jejunum) were obtained from non-diseased parts of resected segments. Only specimens obtained rapidly were used; the overall period required before freezing the tissue after surgery never exceeded 60 min. The project was in accordance with the protocol approved by the Institutional Human Research Review Committee for the use of human material.
The HIEC-6 cells have been generated from the normal fetal human small intestine. These cells express a number of crypt cell markers but no villus cell markers and are thus considered as poorly differentiated crypt cells (41,46). They were grown in Dulbecco's modified Eagle's medium supplemented with 4 mM glutamine, 20 mM HEPES, 5 ng/ml recombinant epidermal growth factor (Life Technologies, Inc.), 0.2 IU/ml insulin (Connaught Novo Laboratories, Willowdale, Ontario), and 5% fetal bovine serum in 100 mM plastic culture dishes (41). Cells were used between passages 5 and 15.
Primary Antibodies-Antibodies used in this study were the mono- Indirect Immunofluorescence-The preparation and Optimum Cutting Temperature embedding compound (Tissue Tek, Miles laboratories, Elkhart, IN) embedding of tissue samples for cryosections was performed as described previously (35). Frozen sections 3-m thick were cut on a Jung Frigocut 2800N cryostat (Leica Canada, St. Laurent, Québec), spread on silane-coated glass slides, and air-dried 1 h at room temperature. Tissue sections were fixed in methanol or ethanol (10 min, Ϫ20°C) before immunostaining, as described elsewhere (31,35). Primary antibodies were diluted 1/100 (anti-␤ 4 cyto and AB1922) or used at 5 g/ml (3E1 and 439-9B). The secondary antibody was an fluorescein isothiocyanate-conjugated goat anti-rabbit or anti-mouse IgG (Roche Molecular Biochemicals Canada, Laval, Québec) or anti-rat IgG (Caltag, Cedarlane Laboratories, Hornby, Ontario) used 1/25 in 2% bovine serum albumin in PBS. 1 Sections were stained with 0.01% Evan's blue in PBS, mounted in glycerol-PBS (9:1) containing 0.1% paraphenylenediamine, and viewed with a Reichart Polyvar 2 microscope (Leica Canada) equipped for epifluorescence. In all cases no specific immunofluorescent staining was observed when primary antibodies were omitted or replaced by the corresponding non-immune serum.
Metabolic Cell Labeling and Immunoprecipitation-Cells were metabolically labeled using Promix [ 35 S]methionine and cystine (Amersham Pharmacia Biotech), 100 Ci/ml for HIEC and 200 Ci/ml for Caco-2/15 for the indicated times as described previously (50). For certain points, cells were labeled as described above and after the labeling period the radioactive medium was removed, cells were washed once with complete Dulbecco's modified Eagle's medium, and chased with 10 ml of complete Dulbecco's modified Eagle's medium containing 10 ϫ methionine and 10 ϫ cystine for the indicated times. Cells were solubilized in ice-cold lysis buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 5 mM EDTA, 0.5% Triton X-100, 2 M phenylmethylsulfonyl fluoride, 50 g/ml leupeptin, 50 g/ml pepstatin, 100 g/ml aprotinin) for 20 min on ice, then centrifuged for 15 min at 13,000 ϫ g. Samples were pre-cleared using 100 l of heparin-agarose (Bio-Rad) followed by 50 l of protein G-Sepharose (Life Technologies, Inc.) each for 1 h at 4°C. Antibodies were added to the samples and incubated for 15 h at 4°C followed by the addition of protein G-Sepharose for 1 h at 4°C. Radioactive samples were analyzed under nonreduced conditions by SDS-PAGE, the gels were fixed (10% acetic acid, 45% methanol, 45% deionized water) for 1 h and soaked for 15 min in Amplify (Amersham Pharmacia Biotech), dried and exposed using Hyperfilm ECL (Amersham Pharmacia Biotech).
RT-PCR-Total RNA was isolated from cell lines or tissue homogenates using TriZOL (Life Technologies, Inc.). The integrity of the RNA was verified by ethidium bromide staining and the quantities were determined spectrophotometrically. Reverse transcriptase Superscript (Life Technologies, Inc.) and 0.5 g of oligo(dT) [12][13][14][15][16][17][18] primer (Amersham Pharmacia Biotech) were added to 5 g of total RNA, as described elsewhere (51). The integrin ␤ 4 subunit was amplified using five different sets of primers. . Conditions for S14 amplification and sequence analysis of the ␤ 4 B3/B6 PCR products were described elsewhere (38). The ␤ 4 B9/B10 1202-bp PCR product was digested with SmaI and SacII prior to cloning in pBluescript for sequence analysis.
Cell Adhesion Assays-The adhesion of HIEC and Caco-2/15 cells on laminin-5 was carried out in 96-well plates based on a procedure described previously (53). Bovine serum albumin, 1%, was used to set background binding. Purified laminin-5 was used at 10 g/ml. The wells were coated for 1 h at 37°C and subsequently blocked with 0.25% bovine serum albumin for 1 h at 37°C. Cells harvested using 10 mM EDTA, washed twice, and 5 ϫ 10 4 cells were plated per well for 30 min. Unbound cells were removed by gently washing twice with PBS and bound cells were fixed in 1% formaldehyde, colored with a 1% crystal violet solution, and solubilized using 2% SDS. The plates were read at 595 nm with a Microplate Reader (model 3550, Bio-Rad). For inhibition studies, neutralizing antibodies were added to cells before plating and incubated for 30 min, each at a final concentration of 10 g/ml.

RESULTS
As shown previously with the 3E1 antibody (31,36) and confirmed herein by using 439-9B, another antibody directed to the extracellular domain of the molecule, the ␤ 4 subunit was found to be expressed in significant amounts at the base of all epithelial cells in both crypt and villus (Fig. 1A). The use of another antibody, the anti-␤ 4 c, raised against the last 31 amino acids of the COOH-terminal of the ␤ 4 subunit, produced a distinct pattern of expression by strongly labeling the basal surface of villus cells but not the crypts (Fig. 1B). The specificity of the staining pattern of the anti-␤ 4 c antibody was confirmed with AB1922, another antibody directed against the last 22 amino acids of the COOH-terminal of the ␤ 4 subunit (Fig.  1C). This observation suggests that at least two immunologically distinct forms of ␤ 4 exist in the intestinal epithelium and that they are associated with different cell populations along the crypt-villus axis.
Expression of ␤ 4 in Intestinal Epithelial Cells-This hypothesis was tested by analyzing the expression of the ␤ 4 subunit in undifferentiated (HIEC) and differentiated (Caco-2/15) cells. Western blot analysis of cell lysates using anti-␤ 4 c ( Fig. 2A) detected the ␤ 4 subunit in Caco-2/15 (lane 2) cells but not in HIEC cells (lane 1). However, RT-PCR amplification using two different sets of primers, located in the region encoding for the cytoplasmic domain of ␤ 4 , revealed that both cell models express the ␤ 4 transcript (Fig. 2B, lanes 2 and 4). The B1/B2 primer set spans the region which is alternatively spliced to produce isoforms ␤ 4 A, ␤ 4 B, or ␤ 4 C (15,16,40). This set of primers amplifies a band of 369 bp which corresponds to ␤ 4 A, indicating that this is the major ␤ 4 isoform in both cell lines. The primer set B3/B4 has been used previously to identify the ␤ 4 D isoform which contains a 21-bp deletion (52). The amplification of a single band of 83 bp with this set of primers in HIEC and Caco-2/15 cells indicated that neither cell line expresses this isoform.
Undifferentiated Crypt Cells Contain a Novel Form of the ␤ 4 A Subunit-The differential expression of ␤ 4 subunits between differentiated and undifferentiated intestinal cells was further investigated by immunoprecipitation using the 3E1 antibody, followed by Western blot detection using the anti-␤ 4 c antibody (Fig. 3A). As expected from Western blot analyses (previous section), the ␤ 4 subunit was only detected in Caco-2/15 cells (lane 2) with this procedure, suggesting that the ␤ 4 c epitope is absent in HIEC. To verify this, HIEC and Caco-2/15 cells were metabolically labeled using [ 35 S]methionine and cystine for 6 h, then ␤ 4 was immunoprecipitated using either 3E1 or the anti-␤ 4 c antibody. As shown in Fig. 3B, both antibodies immunoprecipitated comparable amounts of the ␤ 4 subunit in Caco-2/15 cells, demonstrating that both 3E1 and anti-␤ 4 c epitopes are expressed in these cells (Fig. 3B, lanes 3 and 4). In HIEC, however, the ␤ 4 subunit was detected as a 205-kDa band using 3E1 (lane 2) while the anti-␤ 4 c failed to immunoprecipitate the protein (lane 1), indicating that HIEC express the ␤ 4 subunit, but under a form which lacks the ␤ 4 c epitope, thus distinct from the ␤ 4 subunit found in Caco-2/15 cells.
It has been previously reported that the ␤ 4 subunit can undergo proteolytic cleavage in its cytoplasmic tail (40,54,55). The possibility that this occurred in undifferentiated crypt cells was investigated by pulse-chase experiments and sequential immunoprecipitation. HIEC and Caco-2/15 cells were metabolically labeled with [ 35 S]methionine and cystine for 2 h and chased for 0 -8 h before lysis (Fig. 4A). For each sample, the ␤ 4 subunit was first immunoprecipitated using anti-␤ 4 c antibody. Then, after two additional rounds of immunoprecipitation with anti-␤ 4 c to ensure complete depletion of the immunoreactive ␤ 4 c form in the samples, the remaining ␤ 4 was isolated with a last immunoprecipitation using the 3E1 antibody. This sequen- tial immunoprecipitation procedure showed that in Caco-2/15 cells, the anti-␤ 4 c identified a major band at 205 kDa at all time points corresponding to the ␤ 4 subunit (Fig. 4A, upper left-hand  panel). Subsequent immunoprecipitation of the ␤ 4 c-depleted lysates using 3E1 indicated that the majority of the ␤ 4 subunit contains the anti-␤ 4 c epitope since very little material was recovered with 3E1 (Fig. 4A, upper right-hand panel). In contrast, in undifferentiated HIEC cells, ␤ 4 was barely detected after immunoprecipitation using the anti-␤ 4 c (Fig. 4A, lower  left-hand panel), even with a chase time of 0 min, the almost complete majority of the ␤ 4 subunit being recognized only by 3E1 (Fig. 4A, lower right-hand panel). Shortening the labeling period down to 30 min (Fig. 4B) did not allow the detection of the ␤ 4 protein containing the anti-␤ 4 c epitope (Fig. 4B, lefthand panel), again the majority of the ␤ 4 subunit present being recognized only by 3E1 (Fig. 4B, right-hand panel). This indicates that in HIEC cells, the loss of the ␤ 4 c epitope in the ␤ 4 precursor occurs extremely rapidly.
The possibility that the ␤ 4 transcript may not code for the COOH-terminal domain in undifferentiated crypt cells was verified by designing primers (B5/B6; see "Experimental Procedures") for RT-PCR that would amplify the region overlapping the nucleotides encoding the last 30 amino acids against which the anti-␤ 4 c antibody was raised. As shown in Fig. 5, both cell lines expressed a ␤ 4 A transcript that contains this sequence. To rule out the possibility that the lack of the ␤ 4 c epitope in undifferentiated crypt cells arises from alternative splicing, the sequence of a large section of the ␤ 4 mRNA expressed by these cells was determined. Sequencing of the HIEC B3/B6 and B9/B10 PCR products (Fig. 5), which correspond to the last 1490 bp of the ␤ 4 mRNA encoding the carboxyl-terminal end of the protein, did not reveal any difference with ␤ 4 A (14). Hence, altogether these results indicate that undifferentiated crypt cells express a novel form of the ␤ 4 A integrin subunit which is distinct in its COOH-terminal domain from that found in differentiated intestinal cells.
The ␣ 6 ␤ 4 A ctdϪ Complex in Intestinal Crypt Cells Is Not Functional-As shown in Fig. 4A, the co-immunoprecipitation of ␣ 6 with its ␤ 4 A partner in both HIEC (with 3E1) and Caco-2/15 cells (with the anti-␤ 4 c) indicates that ␤ 4 A is expressed in association with ␣ 6 in both cell lines. The ␣ 6 ␤ 4 integrin serves as a specific receptor for laminins, namely laminin-5 (17, 18, 53, 56 -58). Adhesion of HIEC and Caco-2/15 cells to laminin-5 was thus analyzed (Fig. 6). Adhesion of Caco-2/15 cells was partially inhibited by the GoH3 (anti-␣ 6 ) or 3E1 (anti-␤ 4 ) antibodies. When both of these antibodies were used in combination, they significantly inhibited cell adhesion, indicating that the ␣ 6 ␤ 4 A complex modulates substantially Caco-2/15 cell adhesion to laminin-5 in these cells. The mAb13 (anti-␤ 1 ) antibody also significantly impaired laminin-5 binding, suggesting that Caco-2/15 cells may use a ␤ 1 integrin in cooperation with ␣ 6 ␤ 4 A, but one other than ␣ 2 ␤ 1 or ␣ 3 ␤ 1 since the neutralizing antibodies P1E6 (anti-␣ 2 ) and P1B5 (anti-␣ 3 ), alone or in combination with GoH3, failed to affect significantly Caco-2/15 cell adhesion to laminin-5. In contrast, HIEC cell binding to laminin-5 was unaffected in the presence of GoH3 or 3E1, either alone or in combination, suggesting that ␣ 6 ␤ 4 A ctdϪ is not functional. However, HIEC cells appear to use ␣ 3 ␤ 1 instead, an-  1 and 3) or 3E1 (lanes 2 and 4) from HIEC (lanes 1 and 2) and Caco-2/15 (lanes 3 and 4) cells metabolically labeled using [ 35 S]methionine and cystine for 6 h. Samples were separated on a 10% SDS-PAGE and the gel was dried and exposed for 48 h.  1-4). After two successive rounds of depletion with the anti-␤ 4 c, the remaining ␤ 4 was then immunoprecipitated from the same lysates using 3E1 (lanes 5-8). Samples were migrated on 10% SDS-PAGE under nonreducing conditions. The ␤ 4 subunit migrates at 205 kDa and the ␣ 6 subunit migrates at 150 kDa. B, HIEC were labeled using [ 35 S]methionine and cystine for 30 min with chase times of 0 min ( lanes  1 and 4), 30 min (lanes 2 and 5), or 1 h (lanes 3 and 6). Cells were lysed in a nondenaturing buffer and immunoprecipitated using anti-␤ 4 c (lanes 1-3). After immunodepletion with the anti-␤ 4 c, the remaining lysates were then immunoprecipitated for a last round using 3E1 (lanes 4 -6). All samples were separated on a 10% SDS-PAGE under nonreducing conditions and the gels were dried and exposed for 72 h. other laminin-5 receptor, since adhesion is markedly inhibited with P1B5 or mAb13. In combination, P1B5 and GoH3 did not increase inhibition, supporting the interpretation that ␣ 6 ␤ 4 A ctdϪ is not involved in the binding of undifferentiated intestinal cells to laminin-5.

DISCUSSION
Epithelial cell proliferation, migration, and differentiation are tightly regulated along the crypt-villus axis of the small intestine and evidence that cell-laminin interactions play a role in their regulation is strengthening (33,34). As in other systems, cell-laminin interactions are highly dynamic in the renewing intestinal epithelium, a differential expression of both receptors and ligands having been observed along the cryptvillus axis (13,25). In the present study, we have examined the function of the ␣ 6 ␤ 4 integrin in the mediation of intestinal cell-laminin interactions. We have identified two distinct forms of the ␤ 4 subunit expressed according to the cell state in the intestinal epithelium: a previously undescribed variant of ␤ 4 A that lacks a small terminal fragment of the cytoplasmic domain (␤ 4 A ctdϪ ), which is expressed by crypt cells and their normal in vitro counterpart, the HIEC cells, and the full-length ␤ 4 A subunit, which is expressed by Caco-2/15 cells and by villus cells in the intact intestine. Our results provide clear evidence that both variants of ␤ 4 A can associate with the ␣6 subunit to form a stable ␣ 6 ␤ 4 A complex. However, in contrast to the ␣ 6 ␤ 4 A present in differentiated intestinal cells, the ␣ 6 ␤ 4 A ctdϪ expressed by undifferentiated crypt cells was found to be nonfunctional for cell adhesion to laminin-5.
The ␤ 4 subunit can exhibit a high degree of structural complexity resulting from alternative splicing of its mRNA as well as proteolytic cleavage. Up to now, five distinct mRNA variants of the human ␤ 4 integrin have been identified, each altering the cytoplasmic domain, and designated (59) ␤ 4 A, which is the most common form (14), ␤ 4 B (15), ␤ 4 C (16), ␤ 4 D (52), and the newly identified ␤ 4 E (60). Information about the distribution of these variants is still limited (16,52) and the implication of their expression has only begun to be investigated at the functional level (59,61). Analysis of these variants in intestinal epithelial cells by RT-PCR using specific primer sets revealed that both undifferentiated and differentiated cells express mainly, if not exclusively, the ␤ 4 A form. Furthermore, RT-PCR with primers designed to amplify the 3Ј end of the ␤ 4 A transcript encoding the anti-␤ 4 c epitope ruled out the possibility that the ␤ 4 A ctdϪ form observed in undifferentiated cells arises from an alternative splicing mechanism.
The expression of a ␤ 4 A subunit lacking the immunoreactive terminal portion of the cytoplasmic domain may thus result from alteration(s) of the protein itself. Conformation changes are unlikely since the anti-␤ 4 c antibody used in this study is a polyclonal serum raised against a 31-amino acid stretch (40), and lack of the anti-␤ 4 c epitope was observed under both denaturing and nondenaturing conditions (see Figs. 2 and 3 for comparison). However, post-translational proteolytic cleavage in the cytoplasmic domain of the ␤ 4 subunit has been previously reported yielding a characteristic pattern of additional bands migrating at 165 and 125/130 kDa (40,54,55). These proteolytic fragments of ␤ 4 were not detected in either HIEC or Caco-2/15 cell extracts prepared under conditions in which proteolysis was inhibited (Refs. 40 and 55; see "Experimental Procedures"), indicating that such proteolysis does not occur in these cells. Furthermore, immunofluorescent cytoplasmic staining of the ␤ 4 subunit, used as an indicator of endogenous proteolysis in the intact tissue (40), was not observed in the intestinal epithelium. These observations suggest that such post-translational proteolytic cleavage of ␤ 4 does not occur significantly in the human intestinal epithelium. Since the lack of the anti-␤ 4 c epitope seems to be the result of proteolysis, it is likely to take place by a distinct mechanism. First, the actual portion of the cytoplasmic tail that appears to be deleted in ␤ 4 A ctdϪ would be very short since no significant difference was observed in the apparent size of ␤ 4 A ctdϪ from HIEC cells as compared with the full-length ␤ 4 A isolated from Caco-2/15 cells. Second, its proteolysis should occur very rapidly in the biosynthetic pathway, i.e. co-translationally, since a ␤ 4 A bearing the anti-␤ 4 Ac epitope was not detected in HIEC even after only 30 min of labeling. Taken together, these data indicate that undifferentiated intestinal cells constitutively express ␤ 4 A ctdϪ , a form of ␤ 4 A that lacks a short portion of the terminal cytoplasmic domain. The exact mechanism involved in the generation of the ␤ 4 A ctdϪ form remains to be elucidated but our observations suggest that it arises from a co-translational processing of the integrin subunit.
A second interesting observation is that even though the ␤ 4 A subunit was able to associate with its ␣ 6 partner in the two cell lines studied, the ␣ 6 ␤ 4 A ctdϪ complex expressed by undifferentiated HIEC cells was found to be inactive in terms of laminin-5 adhesion. The HIEC do bind to laminin-5 but use the ␣ 3 ␤ 1 integrin, which has also been reported to have a high affinity for this ligand (8,62,63). Interestingly in these cells, ␣ 3 ␤ 1 appears to function without the cooperation of ␣ 6 ␤ 4 A ctdϪ since the combination of neutralizing antibodies to ␣ 3 and ␣ 6 did not inhibit adhesion of laminin-5 significantly more than anti-␣ 3 alone. In a recent study, it has been suggested that ␣ 3 ␤ 1 and ␣ 6 ␤ 4 can function cooperatively to mediate adhesion (64). This was not observed in HIEC adhesion to laminin-5 even when incubation times were extended to over 1 h (not shown), which is consistent with the apparent inability of ␣ 6 ␤ 4 A ctdϪ to bind to laminin-5. This lack of activity was surprising in view of the fact that tail-less ␤ 4 mutants (65,66) can still bind to laminin-5, albeit at a lower affinity. However, this is not without precedent since a role for the COOH-terminal end of various other ␤ subunit in integrin-mediated adhesion is well documented (67)(68)(69)(70). In this context, the relation between an apparently minor alteration in the COOH-terminal domain of the ␤ 4 A ctdϪ subunit in undifferentiated intestinal crypt cells and the lack of activity of its ␣ 6 ␤ 4 complex is of particular interest. Indeed, the cytoplasmic domain of ␤ 4 is composed of distinct FIG. 6. The ␣ 6 ␤ 4 A ctd؊ integrin is not a functional receptor for laminin-5 in undifferentiated HIEC cells. Adhesion of HIEC and Caco-2/15 cells on purified laminin-5 (10 g/ml) in the presence of neutralizing antibodies to ␣ 2 (P1E6), ␣ 5 (mAb16), ␣ 3 (P1B5), ␣ 6 (GoH3), ␤ 1 (mAb13), and ␤ 4 (3E1) integrin subunits, either alone or in the following combinations: ␣ 2 ϩ ␣ 6 , ␣ 3 ϩ ␣ 6 , and ␣ 6 ϩ ␤ 4 . All antibodies were added to the cells at 10 g/ml and incubated for 30 min before plating. Results are from four separate experiments and are expressed as the percentage of cells bound after 30 min, with 100% being adhesion on laminin-5 in the absence of antibodies (n ϭ 4). structural regions, namely two pairs of type III fibronectin-like modules separated by a region referred to as the connecting segment and the COOH-terminal segment. Both type III fibronectin-like modules and the connecting segment were shown to be important for hemidesmosome assembly including recruitment of HD-1/plectin, BP180, and the association with the keratin cytoskeleton (17,59,61,66,(71)(72)(73). However, recent evidence indicates that the COOH-terminal segment of ␤ 4 also plays a key role in these interactions, being involved in the ␤ 4 interaction with HD1/plectin (72), BP180 (73), and itself (72,73). This latter observation that the COOH-terminal end of ␤ 4 can bind directly to a more NH 2 terminally located region, which encompasses the sequences essential for the interaction with other hemidesmosomal components, is particularly interesting as it suggests that the cytoplasmic domain of ␤ 4 can fold back upon itself and thus be part of a mechanism regulating ␤ 4 interactions with HD1/plectin and PB180 (73). Further characterization of the COOH-terminal end may reveal an additional role(s) for this ␤ 4 cytoplasmic segment. For instance, the functional significance of the strong phosphorylation of the COOH-terminal region of ␤ 4 reported in various epithelial cell lines (74,75) remains to be determined while a number of interesting features pertaining to the cytoplasmic portion of the ␤ 4 subunit such as the recruitment of Shc/Grb2-mSOS and subsequently activating the Ras/MAP kinase pathways (17,24), the ability to modulate the cyclin-dependent kinase inhibitors p21 WAF/Cip1 and p27 Kip (22,65), as well as the activation of the phosphatidylinositol 3-OH kinase (76), have not yet been mapped.
What would be the biological relevance for undifferentiated crypt cells to express a ␣ 6 ␤ 4 A ctdϪ complex unable to mediate adhesion to laminin? While further investigations will be required to clearly answer this question, it is interesting to note that the epithelial basement membrane in the crypt region is negative for laminin-5 in the human small intestine, while it is expressed in the villus basement membrane (39, 77), 2 and therefore coincides almost perfectly with the pattern of ␤ 4 A ctd؉ expression. Furthermore, the expression of HD1/plectin has also been reported to be exclusively expressed by villus intestinal cells (77). The repression of laminin-5 and HD1/plectin expression, coupled to the production of an inactive ␣ 6 ␤ 4 A ctdϪ integrin, may thus represent two distinct control mechanisms ensuring that type II hemidesmosomes will not form in undifferentiated intestinal cells. It is noteworthy that the expression of an inactive form of a molecule is not without precedent in human intestinal crypt cells. For instance, sucrase-isomaltase, a brush-border hydrolytic enzyme, is the subject of a posttranslational regulatory mechanism that depends on the state of differentiation for the acquisition of its mature fully active form (50). On the other hand, the lack of laminin binding activity of the ␣ 6 ␤ 4 A ctdϪ complex in the intestinal crypt may be of some importance, because of the presence of laminin-2 which is distributed according to an increasing gradient from the upper half to the bottom of the gland (31). Indeed, in addition to laminin-1 and laminin-5, which are restricted to the villus, laminin-2 can also serve as a ligand for ␣ 6 ␤ 4 (8,23,66). Therefore, it could be speculated that the inactivity of this receptor in terms of laminin binding is required to allow enterocytes to migrate upward to the villus. Finally, in light of the recent evidence that ␣ 6 ␤ 4 may exert biological activities in conditions where the interaction with its ligand does not occur (65,78), the ␣ 6 ␤ 4 A ctdϪ complex present in crypt cells may nevertheless exert ligand independent activities of functional relevance for undifferentiated epithelial cells.