Identification of a Sequence within the Integrin β6 Subunit Cytoplasmic Domain That Is Required to Support the Specific Effect of αvβ6 on Proliferation in Three-dimensional Culture

The integrin αvβ6 augments the proliferation of epithelial cells in collagen gels and in vivo. This effect depends on the presence of a unique carboxyl-terminal region of the β6 subunit cytoplasmic domain. In the present study, we have utilized deletional and alanine substitution mutagenesis within this region to map the amino acids responsible for αvβ6-mediated proliferation in more detail. Replacement or deletion of any of 6 amino acids (glutamic acid 778, lysine 779, lysine 781, valine 782, aspartic acid 783, and leucine 784) largely abolished the proliferative effects of αvβ6, but none of the mutants examined interfered with αvβ6-mediated cell adhesion or with localization of αvβ6 to focal adhesions. These findings suggest that residues contained within the sequence EKXKVDL are critical for the effects of αvβ6 on proliferation in collagen gels and that pathways initiated by interaction with this sequence are distinct from those required for integrin-mediated cell attachment or focal adhesion formation.

The integrin ␣v␤6 augments the proliferation of epithelial cells in collagen gels and in vivo. This effect depends on the presence of a unique carboxyl-terminal region of the ␤6 subunit cytoplasmic domain. In the present study, we have utilized deletional and alanine substitution mutagenesis within this region to map the amino acids responsible for ␣v␤6-mediated proliferation in more detail. Replacement or deletion of any of 6 amino acids (glutamic acid 778, lysine 779, lysine 781, valine 782, aspartic acid 783, and leucine 784) largely abolished the proliferative effects of ␣v␤6, but none of the mutants examined interfered with ␣v␤6-mediated cell adhesion or with localization of ␣v␤6 to focal adhesions. These findings suggest that residues contained within the sequence EKXKVDL are critical for the effects of ␣v␤6 on proliferation in collagen gels and that pathways initiated by interaction with this sequence are distinct from those required for integrin-mediated cell attachment or focal adhesion formation.
The integrin ␣v␤6 is a receptor for the extracellular matrix proteins fibronectin and tenascin (1)(2)(3). This integrin is restricted in its distribution to epithelial cells and is principally expressed during development, in response to injury and/or inflammation, and in epithelial neoplasms (4,5). In an effort to identify the functional significance of ␣v␤6, we have previously examined the effects of heterologous expression of this integrin on cell behavior. When ␣v␤6 was expressed in the human embryonic kidney cell line, 293, or the colon carcinoma cell line, SW480, the receptor was shown to mediate cell attachment to appropriate ligands and to localize to focal adhesions, properties shared by other members of the integrin family (6). In addition, ␣v␤6 was uniquely capable of enhancing the ability of transfected cells to proliferate in a three-dimensional culture system and in vivo in nude mice (6).
Most integrin ␤ subunit cytoplasmic domains contain a highly conserved region of 48 amino acids containing subdomains that have been shown to be critical for localization to focal adhesions and for interaction with cytoplasmic signaling proteins such as the focal adhesion kinase and paxillin (7)(8)(9)(10). In addition to containing this highly conserved region, the ␤6 cytoplasmic domain also contains a completely unique 11-amino acid carboxyl-terminal extension. In a previous study, we found that this unique carboxyl-terminal extension was not required for localization of ␣v␤6 to focal adhesions nor for ␣v␤6-mediated cell adhesion to fibronectin (6). However, deletion of this region completely abrogated ␣v␤6-mediated proliferation, both in three-dimensional culture and in nude mice (6). In the present study, we have utilized smaller deletions and saturation alanine substitution mutagenesis within this region to map the amino acids responsible for this specific effect of ␣v␤6 in more detail.
Preparation of Expression Plasmids-A cDNA containing the entire coding region of human ␤6 in the mammalian expression vector pcDNAIneo (Invitrogen, San Diego, CA) was constructed as described previously (3). To construct truncated versions of pcDNAIneo␤6 that lacked selected portions of the cytoplasmic domain, cDNA fragments were amplified by polymerase chain reaction using a 5Ј upstream primer corresponding to nucleotides 2050 -2070 and 3Ј downstream primers at the following positions: 1) nucleotides 2568 -2587 to prepare pcDNAIneo␤6T-1; 2) nucleotides 2559 -2578 to prepare pcDNAIneo-␤6T-4; and 3) nucleotides 2550 -2569 to prepare pcDNAIneo␤6T-7. The full-length plasmid, pcDNAIneo␤6, was used as a template. The 3Ј primers introduced stop codons to replace amino acids Cys 788 , Ser 785 , and Val 782 , respectively, and each primer included an XbaI recognition sequence to facilitate ligation into the unique XbaI site in the pcDNAIneo polylinker. Each polymerase chain reaction fragment included a unique BstEII site in ␤6 (nucleotide 2087). The polymerase chain reaction products were digested with XbaI and BstEII and ligated into pcDNAIneo␤6 that had been digested with the same enzymes. The authenticity of each mutant clone was confirmed by dideoxy sequencing using Sequenase 2.0 (U. S. Biochemical Corp.).
For site-directed mutagenesis, full-length ␤6 cDNA was ligated into the mutagenesis plasmid pAlter-1 (Promega) between SmaI and XbaI sites in the polylinker. Nine alanine substitution mutations were generated with the Altered Sites In Vitro Mutagenesis System utilizing nine distinct mutant oligonucleotides designed to individually replace each of amino acids 778 -786 in the ␤6 cytoplasmic domain with alanine. pAlter-1 contains two antibiotic resistance genes encoding resist-* This work was supported in part by National Institutes of Health Grants HL/A 133259, HL47412, and HL53949. 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  ance to ampicillin and tetracylcine, but the ampicillin resistance gene is rendered inactive by a single point mutation. Single-stranded DNA of the vector was purified and allowed to anneal simultaneously with a mutant oligonucleotide and an oligonucleotide designed to repair the ampicillin resistance gene. The mutant strand was synthesized with T4 DNA polymerase and ligated with T4 DNA ligase. BMH71-18 mutS Escherichia coli was transformed with each mutant plasmid, plasmids were purified and used to transform JM109 E. coli, and mutants were selected on ampicillin-containing plates. The correct mutations were confirmed by direct sequencing, as above. Mutant plasmids containing the correct mutations were cloned into pcDNAIneo between unique XhoI and XbaI sites.
Transfection of SW480 Colon Carcinoma Cells-SW480 cells were transfected with pcDNA1neo constructs containing wild type or mutant forms of ␤6 or the expression plasmid only (pcDNA1neo) by a modification of the calcium phosphate precipitation method (12). DNA/calcium phosphate precipitates were added to 10 6 cells in suspension. 48 h after transfection, cells were plated and selected in DMEM supplemented with 1 mg/ml of the neomycin analogue, G418. G418-resistant, individual colonies were harvested and further subcloned by limiting dilution. Stable transfectants were screened for ␤6 expression by flow cytometry.
Cell Proliferation in Collagen Gels-Proliferation assays were performed as described previously (6). Bilayer collagen gels comprising of a 500-l cell-free underlayer and a 500-l cell-containing upper layer were prepared in 24-well tissue culture plates. Type 1 collagen (3.9 mg/ml) was dissolved in 40 mM acetic acid for 48 h. To this solution an equal volume of a 2 ϫ concentrate of DMEM supplemented with 10% fetal bovine serum, penicillin, streptomycin, and 1 mg/ml G418 was added. The pH was adjusted to 7.4 with 1 M NaOH. 500 l of this was plated and allowed to gel at 37°C. 10 4 cells were suspended in 500 l of the DMEM/collagen mixture and overlaid on this layer. After gelation, 500 l of culture medium containing 5% fetal bovine serum, 1 mg/ml G418 was added to each well. In all experiments, duplicate gels were prepared and were incubated at 37°C in 5% CO 2 for 7 days, at the end of which cells were harvested by dissolving the collagen matrix with collagenase (15 mg/ml). Total number of cells harvested was determined by counting cells in a hemocytometer.
Cell Adhesion Assays-Cell adhesion assays were performed as described previously (1). Briefly, wells of nontissue culture treated polystyrene 96-well flat bottom microtiter plates (Linbro/Titertek, Flow Laboratories, McLean, VA) were coated with 100 l of 10 g/ml fibronectin or 1% bovine serum albumin for 1 h at 37°C, washed with PBS, and blocked with 1% bovine serum albumin in serum-free DMEM. Cells were incubated in the presence or the absence of the ␤1-blocking antibody P5D2 for 15 min at 4°C, and 50,000 cells/well were then plated onto the matrix-coated wells. The plates were centrifuged (top side up) at 10 ϫ g for 5 min and incubated for 1 h at 37°C in humidified 5% C0 2 . Nonadherent cells were removed by centrifuging the plates top side down at 10 ϫ g for 5 min. The attached cells were fixed and stained with 1% formaldehyde and 0.5% crystal violet in 20% methanol, and the excess dye was washed off with PBS. The cells were solubilized in 50 l of 2% Triton X-100 and quantified by measuring the absorbance at 595 nm in a Microplate Reader (Bio-Rad). The data were expressed as the means of the absorbance of triplicate wells. Measurements of absorbance were determined to be linearly related to input cell number over the range of 10,000 -70,000 cells/well.
Immunofluorescence-Cells were harvested by EDTA treatment before reaching 75% confluence. 3-5 ϫ 10 3 cells were plated on 6-mm diameter wells of 10-well slides (Structure Probe, West Chester, PA) that were previously coated with 10 g/ml fibronectin and blocked with 2% bovine serum albumin in PBS. The slides were incubated for 4 h at 37°C in 5% C0 2 and washed twice with PBS, and the cells were fixed with 2% paraformaldehyde (Fisher) in PBS for 10 min and permeablized for 20 min in 0.1% Triton X-100 in PBS. The cell monolayers were blocked with 2% bovine serum albumin and processed for immu-nofluorescence microscopy. The slides were incubated with anti-human ␣v␤6 antibody (E7P6, 1:10), with biotinylated sheep anti-mouse IgG (1:50), and with streptavidin-fluorescein (1:100) for 60, 45, and 30 min, respectively. For vinculin staining, the fixed cells were incubated for 1 h with primary mouse anti-vinculin and then for 30 min with fluorescein isothiocynate-coupled goat anti-mouse IgG (1:50). Stained cells were washed with PBS between incubations. Slides were briefly rinsed in distilled water and mounted with coverslips using Fluoromount G (Fisher).

Deletions within the Carboxyl-terminal 11 Amino Acids Are Well Expressed on the Cell Surface and Do Not Impair ␣v␤6mediated Cell Adhesion or Localization of the Receptor to Focal
Adhesions-We have previously reported that heterologous expression of the integrin ␣v␤6 augments proliferation of SW480 cells in three-dimensional collagen gels and that a unique region of the ␤6 cytoplasmic domain is required for this effect (6). This unique region corresponding to the carboxyl-terminal 11 amino acids ( Fig. 1) was not required for cell attachment or for localization of ␣v␤6 to focal adhesions. To determine the amino acids within this unique region that are responsible for this proliferative effect, we first generated stably transfected SW480 cells expressing truncated ␤6 lacking the carboxylterminal one, four, or seven amino acids, respectively (Fig. 1). As shown in Table I, each of these truncation mutants was well expressed on the cell surface, as detected by flow cytometry using the ␣v␤6 complex-specific monoclonal antibody, E7P6 (3).
To determine the ability of each truncation mutant to support ␣v␤6-mediated cell attachment, adhesion assays were performed for each transfectant on fibronectin-coated 96-well plates in the presence of the ␤1-blocking antibody, P5D2 (Fig.  2). As expected, mock transfectants demonstrated only minimal ␤1-independent adhesion to fibronectin, whereas each of the deletion mutants adhered as well as the wild type ␤6 transfectants. Each of the deletion mutants also localized to focal adhesions as determined by fluorescence microscopy (data not shown). These findings are consistent with our previously reported observations that the carboxyl-terminal 11 amino acids were not required for ␣v␤6-mediated cell adhesion or localization of ␣v␤6 to focal adhesions (6,13) and suggest that the truncation mutations we generated did not grossly alter the conformation of the conserved membrane-proximal 48 amino acids of the ␤6 cytoplasmic domain.
To assess the ability of each of the cytoplasmic domain trun- cations to support ␣v␤-mediated proliferation in three-dimensional culture, mock, ␤6, ␤6T-1, ␤6T-4, and ␤6T-7 transfectants were plated in collagen gels for 7 days (Fig. 3). Deletion of the carboxyl-terminal cysteine (T-1) had no effect on proliferation. SW480 cells stably transfected with this construct proliferated as well as cells transfected with full-length ␤6, whereas mocktransfected cells did not proliferate. Although deletion of the carboxyl-terminal four amino acids (T-4) reduced proliferation by approximately 50%, SW480 cells transfected with this construct retained considerable capacity to proliferate in this culture system. These data suggest that the carboxyl-terminal four amino acids are not absolutely required to produce this effect. In contrast, deletion of the carboxyl-terminal seven amino acids essentially abrogated ␣v␤6-mediated proliferation, suggesting that one or more critical amino acids is included in the region including valine 782, aspartic acid 783, and leucine 784.
We analyzed at least two clones for each mutation and obtained similar results with each clone, excluding clonal variation as an explanation for any of the observations made. Furthermore, each clone was analyzed for their expression of other integrins on the cell surface (Table I). There was considerable variation among clones in the expression of ␣2, ␣3, ␣5, ␣6, ␤1, and ␤5. However, there was no consistent relationship between expression of other integrins and proliferation in three-dimensional culture.
Effects of Alanine Substitution Mutations-To further delineate the critical amino acids required for this unique effect of the ␤6 cytoplasmic domain, SW480 cells were stably transfected with expression plasmids encoding mutant forms of ␤6 in which each of amino acids Glu 778 through Thr 786 in the ␤6 cytoplasmic domain were replaced by alanine (Fig. 1). As shown in Table II, all of the alanine substitution mutations were expressed on the cell surface as determined by flow cy-tometry, but there was considerable variation in the level of ␣v␤6 expressed. Despite this range of expression levels, each of the alanine substitution mutants could mediate ␤1-independent cell adhesion to fibronectin (Fig. 4) and localized to focal adhesions in cells plated on fibronectin (data not shown). These data suggest that the alanine substitutions tested, like the deletion mutations described above, did not produce gross changes in the conformation of the conserved membrane-proximal 48 amino acids of the ␤6 cytoplasmic domain.
The ability of each of these substitution mutants to mediate proliferation in collagen gels is shown in Fig. 5. Alanine substitution for either glutamine 780 or threonine 786 had little effect on ␣v␤6-mediated cell proliferation, suggesting that neither of these residues is critical. The ability of T786A to proliferate is consistent with the results obtained for deletion of this amino acid (in ␤6T-4) and is also informative because this mutant was consistently expressed at the lowest level on the cell surface of stable transfectants. Apparently, even this level of expression (a level similar to the native expression seen in a number of epithelial tumor cell lines) is sufficient to mediate maximal proliferation in collagen gels. In contrast, alanine substitutions for lysine 781, valine 782, aspartic acid 783, leucine 784, or serine 785 completely abolished this proliferative response. The effect of substitution for serine 785 is most likely the result of a change in the conformation of an adjacent motif, because deletion of this residue in T-4 only reduced proliferation by approximately 50%. However, the results of both deletional and substitution mutagenesis suggest that at least a portion of the sequence Lys 781 -Val 782 -Asp 783 -Leu 784 (KVDL) is critical for ␣v␤6-mediated proliferation in collagen gels. Alanine substitution for glutamic acid 778 or lysine 779 also markedly inhibited ␣v␤6-mediated proliferation in this culture system, suggesting either that these residues are also part of a critical motif or that alanine substitution at these positions critically alters the conformation of the adjacent KVDL motif. We analyzed at least two clones for each mutation to exclude unrelated clonal variation as an explanation for any  Mock  312  878  166  128  148  1383  838  6  ␤6  287  837  382  281  201  1745  683  404  T-1  218  837  239  134  459  1423  764  615  T-4  522  1143  496  183  521  2224  623  508  T-7  112  558  88  73  807  1111  242  of the observations made. In every case, multiple clones proliferated in similar fashion in collagen gels (data not shown). To determine whether heterologous expression of substitution mutants affected the expression of other integrin subunits, we analyzed surface expression of ␣2, ␣3, ␣5, ␣6, ␤1, and ␤5 in all the transfectants by flow cytometry (Table II). As with the deletion mutants, there was variation in the expression of other integrins among the different transfectants; however, we were unable to detect any correlation between the level of expression of any of the integrin subunits and the magnitude of the proliferative responses.

DISCUSSION
The results of the present study confirm our previous finding that distinct regions of the ␤6 cytoplasmic domain are required for localization to focal adhesions and mediating stable cell adhesion and for enhancing cell proliferation in three-dimensional culture (6). For focal adhesion localization and cell adhesion, regions within the highly conserved, membrane-proximal 48 amino acids are sufficient, as confirmed by the lack of effects of any of the deletion or substitution mutants examined in the present study on either of these end points. In contrast, proliferation in three-dimensional culture is absolutely dependent on the presence of a region within the unique 11-amino acid carboxyl-terminal extension of ␤6, which has the minimal essential sequence EKXKVDL. Either deletion or alanine substitution for any of these amino acids markedly attenuated ␣v␤6-mediated proliferation in this culture system. Although alanine substitution for serine at position 785 also abolished proliferation, this amino acid does not appear to be essential, because the mutant (T-4) from which it was deleted was able to support substantially enhanced proliferation (albeit 50% of that induced by full-length ␤6). The carboxyl-terminal 11 amino acids of ␤6 and the critical amino acid sequence we have identified within this region are not homologous with any known signaling motifs. Although the carboxyl-terminal 11 amino acids includes two potential phosphorylation sites (serine 785 and threonine 786), neither of these residues appears to be critical.
The cytoplasmic domain of the integrin ␤1 subunit has been extensively studied and has regions within it that have been shown to be important for association with the linker proteins ␣-actinin and talin that connect integrins to the actin cytoskeleton and lead to localization of integrins to the focal adhesions that form at the ends of actin stress fibers in spread cells (14 -17). Each of these regions is highly homologous to regions within the the membrane-proximal 48 amino acids of the ␤6 subunit cytoplasmic domain (6,13). It is therefore not surprising that none of the mutations analyzed in the present study, all of which involved amino acids outside this conserved region, affected localization to focal adhesions. Similarly, because we have previously reported that deletion of the carboxyl-terminal 11 amino acids of the ␤6 cytoplasmic domain does not interfere with ␣v␤6-mediated adhesion to fibronectin (6), it is not surprising that none of the smaller deletions or point mutations examined in the present study inhibited ␣v␤6-mediated cell adhesion.  5. Proliferation of mock-, wild type ␤6-, and ␤6 alanine substitution mutant-transfected SW480 cells in three-dimensional collagen gels. Gels were seeded with 10 4 cells for each of the transfectants, gels were dissolved 7 days later with collagenase, and cell number was determined in a hemocytometer. The bars represent mean values of 10 data points from duplicate wells in five separate experiments, and the lines above each bar represent standard errors.  Mock  312  878  166  128  148  1383  838  6  ␤6  287  837  382  281  201  1745  683  404  E778A  280  877  159  186  162  1332  593  198  K779A  326  1124  119  63  404  1408  894  161  Q780A  256  955  108  64  287  1310  766  244  K781A  316  1244  393  103  413  1540  782  271  V782A  287  1451  158  120  396  1658  629  254  D783A  429  1688  128  88  427  2137  1086  134  L784A  378  1138  127  75  323  1644  913  164  S785A  667  806  383  230  294  2160  582  170  T786A  81  1193  170  169  343  1516  897  57 The results of mutagenesis experiments should always be interpreted with caution, because deletions or substitutions in one portion of a polypeptide could produce effects by changing the conformation of other regions. For example, in the present study, as noted above, substitution of alanine for serine 785 abolished ␣v␤6-mediated enhancement of proliferation, but deletion of this same amino acid did not, a finding that suggests an effect of the alanine at this position on the conformation of another region of the cytoplasmic domain. Nonetheless, several lines of evidence suggest that at least amino acids with the linear sequence KVDL are directly involved in the proliferative effects of ␣v␤6. Deletion or alanine substitution of each of these amino acids consistently eliminated ␣v␤6-mediated enhancement of proliferation. Furthermore, none of the deletion or substitution mutations within this region affected ␣v␤6-mediated cell adhesion or localization of the receptor to focal adhesions. We have previously reported that each of five different deletion mutations within the conserved membrane-proximal 48 amino acids of the ␤6 cytoplasmic domain completely abolished both cell adhesion and focal adhesion localization (6,13). Together, these findings suggest that the mutations we examined in the present study did not dramatically alter the functionally significant regions of the membrane-proximal 48 amino acids.
Over the past few years, studies from several different laboratories have described a number of intracellular signaling pathways that can be activated by ligation of several members of the integrin family. A number of these pathways, including activation of Src, Ras, and phosphatidylinositol 3,4,5-trisphosphate kinase could be initiated as downstream consequences of phosphorylation of the focal adhesion kinase or the adapter protein, paxillin (18,19). However, little is known about how ligation of specific integrins leads to signaling specificity. Both focal adhesion kinase and paxillin have been shown to bind to an 11-amino acid, membrane-proximal region of the ␤1 cytoplasmic that is highly conserved among integrin ␤ subunits (20), including ␤6 (21). It is therefore not surprising that phosphorylation of focal adhesion kinase and paxillin has been shown to occur after ligation of several members of the integrin family (18), including ␣v␤6. 2 However, cellular responses, such as the one described in this paper, that are specific for specific family members are likely to involve distinct pathways. The stable transfectants we have describe in the present study should be useful tools for identifying the signaling pathways involved in the unique cellular responses to ␣v␤6.