The Integrin Cytoplasmic-tail Motif EKQKVDLSTDC Is Sufficient to Promote Tumor Cell Invasion Mediated by Matrix Metalloproteinase (MMP)-2 or MMP-9*

Integrins promote cellular invasion through a combination of activities, including adhesion to an extracellular matrix ligand, which result in the generation of intracellular signals that lead to changes in cell behavior. Until now, there have been no data that identify a particular region of the cytoplasmic tail of integrin subunits as being responsible specifically for promoting the invasive activity of tumor cells. In this report, we show that amino acids with the sequence EKQKVDLSTDC, which are the C-terminal residues of the integrin β6 subunit, promote αvβ6-dependent invasion in a matrix metalloproteinase (MMP)-9-dependent fashion. This same peptide sequence, when expressed at the cytoplasmic end of the β3 integrin subunit, was able to enhance αvβ3-mediated invasive and enzymatic activity of tumor cells in an MMP-2-dependent fashion. Our results show that these 11 amino acids, when expressed at the C terminus of the β subunit, are responsible for regulating the activity of invasion-promoting degradative enzymes, whereas the specific MMP involved in this cellular behavior is dependent on the context of the remainder of the β integrin subunit.

A major goal of cancer research is to understand the molecular basis of tumor cell invasion, because this process represents a critical step in tumor metastasis, the most devastating aspect of malignancy. Integrins are key components of the invasive cancer cell in which they serve as the principal mediators of adhesion to, migration across, and invasion through, the extracellular matrix. In this article, we report that a discrete region of the cytoplasmic tail of the ␤6 subunit is essential, and sufficient, to promote invasion by carcinoma cells that express this integrin subunit.
Expression of the integrin ␣v␤6 heterodimer is restricted to epithelial cells, where it serves as a receptor for fibronectin, tenascin, and latency-associated peptide, a protein involved in the maintenance of tumor growth factor-␤ in an inactive state. Integrin ␣v␤6 levels are up-regulated in wound healing and in various epithelial neoplasms (1)(2)(3)(4)(5)(6)(7). Expression of ␣v␤6 in oral SCC has been implicated as having a role in tumor cell invasion, both in vitro and in vivo, through experiments involving integrin-specific antibody blockade (8 -11). Moreover, ␣v␤6-me-diated production and regulation of various MMPs, 1 specifically MMP-9 and MMP-3, also have been reported in several tumor types and in untransformed keratinocytes (9,(12)(13)(14). Further still, ␣v␤6 has been implicated in the regulatory control of the urokinase-plasminogen activator proteolytic cascade (15). Thus, ␣v␤6 is associated intrinsically with invasion that is regulated, in part, by the activity of certain proteases of both the MMP and the serine protease families.
Integrins modulate cell behavior in response to a range of events largely by generating signals via their cytoplasmic tails. There is a large degree of protein homology between the different integrin ␤-subunit cytoplasmic domains, particularly between cytodomains-1, -2, and -3 ( Fig. 1), these being regions of the molecule that are involved in the recruitment of integrins to focal adhesions and in modulating integrin affinity (16,17). However, in contrast to this conservation of primary sequence, the ␤6 subunit contains an unique C-terminal 11 amino acid sequence EKQKVDLSTDC (Fig. 1). We have demonstrated recently that this motif is essential for oral SCC invasion. 2 This domain also has been implicated in the ligand-independent regulation of MMP-9 in a colon cancer model (18) and the density-dependent regulation of ␣v␤6 cell surface expression (19). In the present study, we have investigated the role of the unique ␤6 C-terminal 11 amino acids in the regulation of MMPs and tumor cell invasion. We demonstrate, for the first time, by generating cells expressing chimeric integrins, that this unique, 11-residue motif not only is essential for ␣v␤6mediated invasion but also is sufficient to induce an enhanced invasive phenotype, through up-regulation of gelatinases. It is notable that the precise gelatinase up-regulated is determined not by the 11 amino acid sequence but by the residual integrin cytoplasmic tail with which the 11 amino-acids are associated.

EXPERIMENTAL PROCEDURES
Cell Lines-The low ␣v␤6-expressing oral SCC cell line, V3, was generated by transfection of ␣v cDNA into the ␣v-negative oral SCC cell line H357 (20 -22). We have created the C1, VB6, and V3B6⌬11aa cell lines previously through retroviral transduction into V3 cells of pBabepuro retroviruses encoding either the puromycin resistance gene alone, wild-type human ␤6 cDNA, or a truncation mutant ␤6 lacking the C-terminal 11 amino acids, respectively (8,9). 2 In this study, a further panel of oral SCC cell lines was generated, using similar techniques, expressing full-length wild-type human ␤3 or a chimeric extension mutant of human ␤3, whereby the cDNA sequence encoding the Cterminal 11 residues of ␤6 was added to that encoding the carboxyl terminus of wild-type ␤3. These cell lines were designated V3B3 and * 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.
V3B3⌺11aa, respectively. C1, the vector control cells, were used as negative control cells. All SCC cells were maintained in keratinocyte growth medium as described elsewhere (8).
Generation of Oral SCC Cell Populations Expressing Wild-type or Extension Mutant ␤3-Using standard molecular techniques, the wildtype ␤3 sequence was excised from the pCDM8 vector (CD3A; a kind gift of Joe Loftus, Mayo Clinic) using restriction enzymes and ligated into the pBabepuro retroviral expression vector (␤3pBabepuro). Wild-type ␤3, in the pCR2.1 vector (Invitrogen), was used as a PCR template to generate cDNA encoding an extension mutant of the ␤3 subunit extended by the C-terminal 11 residues of ␤6 (␤3⌺11aa) (Fig. 1). PCR primers also included 5Ј-XhoI and 3Ј-XhoI restriction sites (bold and underlined) to aid cloning. The forward primer (Beta3F) used was 5Ј-CCCCCCCTCGAGCGGGAGGCGGACGAGATGCGAGC-3Ј and the reverse primer (Beta3/6R) used was 5Ј-CCCCCCCTCGAGCTAGCAAT-CTGTGGAAAGGTCTACCTTTTGTTTTTCAGTGCCCCGGTACGTGA-TATTGGTGAAG-3Ј. The ␤3⌺11aa PCR product was ligated into the pCDNA3.1 shuttle vector (Invitrogen). Restriction digest of the encoded XhoI sites was used to excise the ␤3⌺11aa cDNA, and this restriction fragment was blunt-end ligated into the SalI restriction site of the pBabepuro multiple cloning site, thus generating ␤3⌺11aapBabepuro.
Retroviruses encoding either ␤3 or ␤3⌺11aa were generated as described by Thomas et al. (8) and used to infect V3 cells to create V3B3 and V3B3⌺11aa, respectively. Puromycin-resistant cells were treated with mouse anti-␣v␤3 antibodies (LM609; Chemicon International, Harrow, UK) and sorted using magnetic beads (sheep anti-mouse IgG; Dynal A.S., Oslo, Norway) to enhance the levels of cell surface ␣v␤3 expressed by the cell populations. Expression levels of the ␤3 integrin heterodimers were determined by flow cytometry.
Flow Cytometry-Standard flow cytometric techniques were used to assess levels of cell surface integrins on the various derivative lines. Cells grown on tissue culture plastic were detached with trypsin/EDTA (0.25% (w/v), 5 mM) and washed twice with 0.1% bovine serum albumin and 0.1% sodium azide in phosphate-buffered saline (0.1/0.1 solution). Cells were incubated with primary antibody for 30 min at 4°C and washed three times with 0.1/0.1 solution. Alexa Fluor 488-conjugated rabbit anti-mouse secondary antibody (Molecular Probes, Eugene, Oregon) (1:200 final dilution) was added to the cells at 4°C for 30 min. Cells were washed in 0.1/0.1 solution three times, resuspended to 0.5 ml, and analyzed on a FACSCalibur cytometer (BD Biosciences, Oxford, UK), using Cellquest software, collecting 1 ϫ 10 4 events.
Indirect Immunofluorescence-In a 24-well plate, 13-mm diameter glass coverslips (Chance Propper Ltd., Smethwick, UK), previously coated with 5 g/ml vitronectin (Sigma, Poole, UK), in Dulbecco's modified Eagle's medium for 1 h at 37°C, were seeded with 2 ϫ 10 4 cells. After 24 h in keratinocyte growth medium, cells were rinsed in phosphate-buffered saline and fixed in 4% (w/v) formaldehyde for 10 min. Cells were rinsed twice in phosphate-buffered saline then treated with 0.1% (v/v) Triton X-100 in phosphate-buffered saline for 5 min at ambient temperature before incubation in 0.1/0.1 solution for 20 min. LM609 primary antibody (anti-␣v␤3, 5 g/ml; Chemicon International Ltd., Harrow, UK), was added to cells for 60 min at 4°C. Coverslips were washed three times in 0.1/0.1 solution and cells incubated with Alexa Fluor 488-conjugated goat anti-mouse IgG (Molecular Probes, Eugene, Oregon) (1:200 dilution) for 30 min at room temperature. Cells were rinsed once and actin was labeled using tetramethylrhodamine B isothiocyanate-conjugated phalloidin (100 ng/ml; Sigma) (10 min incubation at room temperature). Otherwise, cells were labeled with rabbit anti-human ␤3 (Chemicon International) (secondary antibody ϭ goat anti-rabbit Alexa 488 (Molecular Probes)) and counterstained for vinculin, using mouse anti-human vinculin (Sigma) and Alexa 568-conju-gated rabbit anti-mouse IgG secondary antibody (Molecular Probes). Coverslips were washed three times, mounted in Mowiol 4 -88 (Novabiochem, Nottingham, UK), using 100 mg/ml of Citifluor mountant (Citifluor Ltd., London), and visualized with a confocal laser scanning microscope (LSM510; Zeiss, Welwyn Garden City, UK). To allow for direct comparison of levels of expression between different cell lines, the analysis parameters on the microscope were kept constant for all cell lines.
Migration Assay-Haptotactic migration assays were performed using vitronectin-coated polycarbonate filters (8-m pore size, Transwell, Costar Corning Inc.). The undersides of the filters were coated in plasma vitronectin (Sigma: 10 g/ml) for 1 h at 37°C, rinsed with ␣-MEM, and blocked for 30 min in migration buffer (0.5% bovine serum albumin in ␣-MEM supplemented with 4 mM glutamine). For blocking experiments, cells were preincubated at 4°C for 10 min in the presence of LM609 (anti-␣v␤3; 5 g/ml). Control cells were incubated with the anti-MHC class I mouse monoclonal antibody W6/32 (5 g/ml; a gift from Sir Walter Bodmer, Institute of Molecular Medicine, Oxford, UK). Cells (1 ϫ 10 5 ) were added to the upper chamber of quadruplicate wells in 100 l of migration buffer and 500 l of migration buffer were added to the lower chamber. Plates were incubated at 37°C for 16 h. After incubation, all cells were trypsin-detached separately from both the upper and lower chambers of the Transwell plates and counted on a CASY-1 cell counter (Schä rfe System GmbH). Percentage migration was determined by the number of cells in the lower chamber relative to the total number of cells in the well. Random motility (toward bovine serum albumin) was subtracted from the total percentage migration to determine substrate-specific migration. Experiments were repeated three times, with quadruplicate wells for each treatment, with similar results.
Analysis of MMP Secretion-Cells were allowed to attach to fibronectin-or vitronectin-coated 24-well plates in 500 l of serum-free ␣-MEM. After 24 h, the supernatant was collected and analyzed by one of two methods. Levels of MMP-2 or MMP-9 secretion were measured by gelatin zymography as described previously (9,14). MMP-9 levels were also assessed using an enzyme-linked immunosorbent assay kit (Amersham Biosciences; used per manufacturer's instructions). Pro-MMP-9 was identified as a 92-kDa band on the gelatin-PAGE gel, whereas pro-MMP-2 was seen as a 72-kDa band. After collecting the conditioned medium, the adherent cells were trypsinized and counted to express protease activity relative to cell number and to achieve equal loading on each gel.
Invasion Assay-70 l of basement membrane-like matrix, Matrigel (BD Biosciences), diluted 1:2 with ␣-MEM, was allowed to set at 37°C on the upper surface of a Transwell chamber. To the lower chamber was added 500 l of keratinocyte growth medium, to act as a chemoattractant, and 1 ϫ 10 5 cells were added in 200 l of invasion buffer (␣-MEM supplemented with 4 mM glutamine) on top of the Matrigel. After 48 h, all cells in the lower chamber were collected by trypsin detachment and counted on a CASY-1 cell counter. In some experiments, cells were pre-incubated on ice with 10 g/ml function-blocking antibodies (LM609 (anti-␣v␤3) or CA-4001 (anti-MMP-2; Chemicon International). Invasion was expressed relative to V3B3 or C1 cells. Antibody inhibition was expressed relative to the activity of cells in the presence of W6/32 control antibody.

RESULTS
The C-terminal 11 Residues of the ␤6 Integrin Subunit Are Essential for ␣v␤6-Dependent MMP-9 Up-regulation-We have demonstrated previously that the amino acids EKQKVDL- Underlined are cytodomains 1, 2, and 3. The unique 11 amino acids of the ␤6 subunit are in bold. In the amino acid sequence of the cytoplasmic tail of mutant ␤6⌬11aa, note that the C-terminal 11 amino-acids of wild-type ␤6 have been replaced with a myc-his tag (italics). In the amino acid sequence of the cytoplasmic tail of the ␤3 chimera ␤3⌺11aa, note that the unique 11 amino acids of ␤6 have been added to the C terminus of ␤3.
STDC, which are located at the C terminus of the ␤6 subunit, are essential for ␣v␤6-dependent invasion by oral SCC cells. This was determined by comparing the invasive capacity of C1, VB6, and V3B6⌬11aa cell lines. 2 Niu et al. (18) demonstrated, in a colon cancer model, that this same region of the molecule is required for ligand-independent up-regulation of pro-MMP-9. However, the behavior of an integrin in one cell line is not necessarily indicative of its behavior in another cell line (23). Therefore, we compared the levels of pro-MMP-9 secreted by V3B6⌬11aa cells, when plated on fibronectin, versus those of VB6 or C1 cells, both by gelatin zymography (Fig. 2, A and B) and by enzyme-linked immunosorbent assay (Fig. 2C). Zymographic analysis demonstrated that levels of secreted pro-MMP-9 were up-regulated in the VB6 cells relative to the C1-null transfectants (pro-MMP-9 levels: VB6, 100%; C1, 53.5%). We have reported previously that this increase is caused by ligand-dependent binding of ␣v␤6 (9). However, this ␣v␤6-dependent up-regulation was not observed in V3B6⌬11aa cells (pro-MMP-9 levels: V3B6⌬11aa, 50.5%). Enzyme-linked immunosorbent assay on supernatants taken from these cells confirmed these zymography results ( Fig. 2C; VB6, 100%; C1, 68.5%; V3B6⌬11aa, 21.3%). Pro-MMP-2 was not detected by gelatin zymography of conditioned medium from C1, VB6, or V3B6⌬11aa cells (data not shown). These results seem to demonstrate that the C-terminal 11 amino acids of the ␤6 subunit are critical for ␣v␤6-dependent up-regulation of pro-MMP-9 in these oral SCC cells, an up-regulation that correlates with the invasive behavior of the different cell lines. 2 Creation of an ␣v␤3⌺11aa-Expressing Oral SCC Cell Line-To determine whether the ␤6 C-terminal 11 amino acids are sufficient, rather than just necessary, to promote oral SCC invasion, additional oral SCC cell lines were generated. Thus, ␣v-positive/␤3-negative, V3 oral SCC recipient cells (20 -22) were transduced retrovirally with cDNAs encoding either wildtype ␤3 or a mutant ␤3 (␤3⌺11aa) that was extended by the C-terminal 11 residues of the ␤6 subunit (Fig. 1), to create the V3B3 and V3B3⌺11aa cell lines, respectively. After several rounds of magnetic bead sorting, flow cytometry was performed to determine levels of cell surface ␣v␤3 expression. V3B3⌺11aa was seen to express levels of cell surface ␣v␤3⌺11aa mutant heterodimer similar to those at which the V3B3 cells express wild-type ␣v␤3 (Fig. 3A).
Immunolocalization of ␣v␤3 in the Panel of SCC Cell Lines-To examine distribution of wild-type and ␤3-chimeracontaining ␣v␤3, indirect immunohistochemistry was performed on V3B3 and V3B3⌺11aa cells using the LM609 monoclonal antibody. Both the V3B3 and V3B3⌺11aa cell lines were seen to express similar levels of the ␣v␤3 integrin heterodimer. The pattern of ␣v␤3 localization in V3B3 and V3B3⌺11aa cells was similar also, and the heterodimer appeared at the ends of actin filaments in focal adhesion-like structures (Fig. 3B). To confirm that these structures were focal adhesions, V3B3 and V3B3⌺11aa cells were double-labeled for ␤3 and the focal adhesion-associated protein vinculin. Fig. 3, C and D, shows exact correlation between ␤3 or ␤3⌺11aa and vinculin, confirming that the focal distribution of ␣v␤3 at the ends of actin filaments actually was in focal adhesions. This provided the first indication that the ␣v␤3 and ␣v␤3⌺11aa heterodimers, expressed by these cell lines, were functional integrins.
The C-terminal 11 Amino Acids of ␤6 Do Not Affect ␣v␤3-Dependent Adhesion or Migration-Adhesion to vitronectin was assessed in the C1, V3B3, and V3B3⌺11aa cell lines. The C1 cells achieved 6.52% adhesion, whereas the V3B3 cells demonstrated 38.60% adhesion and V3B3⌺11aa cells achieved 39.45% (data not shown). Thus, the wild-type and mutant ␣v␤3 heterodimers expressed in V3B3 and V3B3⌺11aa can mediate adhesion to vitronectin to a comparable extent.
To determine whether the addition of the EKQKVDLSTDC sequence to the C terminus of ␤3 would affect ␣v␤3-dependent migration, haptotactic migration toward vitronectin was measured in Transwell assays. Fig. 4 shows that migration toward vitronectin was similar and significantly enhanced for both the V3B3 and V3B3⌺11aa cell lines (20.29 and 21.79%, respectively) compared with C1 cells that displayed only basal levels of migration (2.01%). Moreover, in the presence of an ␣v␤3blocking antibody (LM609), V3B3 and V3B3⌺11aa migration was reduced to basal levels (3.92 and 0%, respectively). Therefore, the wild-type and mutant ␣v␤3 heterodimers, in V3B3 and V3B3⌺11aa cells, are functional as vitronectin receptors and able to transduce signals required for ligand-binding, cell adhesion, and migration. Furthermore, the addition of EKQKVDLSTDC to the C terminus of ␤3, as a part of ␣v␤3⌺11aa, did not modulate ␣v␤3-dependent adherence or ␣v␤3-dependent migration as determined in these assays.
Integrin ␤6 C-terminal 11 Amino Acids Are Sufficient to Promote Invasion by SCC Cells-To assess whether expression of the ␤6 C-terminal 11 amino acids as a chimera with another integrin was sufficient to increase the invasive potential of SCC cells, C1, V3B3, and V3B3⌺11aa cells were allowed to invade Matrigel-coated Transwells for 48 h. Fig. 5 shows that, relative to the C1 control cells (100%), invasion through Matrigel of V3B3⌺11aa cells (220.8%) was significantly higher than that of the V3B3 cell line (92.9%) (p ϭ 0.0004).
Because the V3B3⌺11aa cell line invades Matrigel to a significantly greater extent than V3B3 cells, these data indicate that the C-terminal 11 amino acids are sufficient, when teth-

FIG. 2. Examination of pro-MMP-9 levels of oral SCC cell lines.
A, zymogram showing triplicate samples demonstrating secreted pro-MMP-9 bands. Levels of pro-MMP-9 were increased in the VB6 cells relative to both the C1 and V3B6⌬11aa cell lines. B, densitometric analysis of pro-MMP-9 zymographic band intensity relative to the cell number in each well. Results are expressed relative to MMP-9 expression for VB6 cells. Hence, VB6 ϭ 100 Ϯ 1.9%, C1 ϭ 53.5 Ϯ 6.6%, and V3B6⌬11aa ϭ 50.5 Ϯ 2.6%. Error bars represent standard deviation. C, enzyme-linked immunosorbent assay showing up-regulation of secreted MMP-9 by VB6 cells relative to both C1 and V3B6⌬11aa cells. MMP-9 expression: VB6 ϭ 100%, C1 ϭ 68.5%, and V3B6⌬11aa ϭ 21.3%. Data are the combined results from three independent experiments. Error bars represent standard error. ered to the ␤3 integrin, to promote an increase in the invasive characteristics of this SCC cell line.
The Amino Acid Sequence EKQKVDLSTDC Induces an ␣v␤3-Dependent Increase in pro-MMP-2-We have shown previously that ␣v␤6 promotes invasion via a pro-MMP-9-dependent mechanism and that this phenomenon involves the termi-nal 11 amino acids of the ␤6 subunit. To determine whether a similar mechanism exists in V3B3⌺11aa cells, gelatin zymography was performed on supernatants obtained from C1, V3B3, and V3B3⌺11aa cells plated on vitronectin, an ␣v␤3 substrate. Fig. 6, A and B, shows a typical   pared with the C1-null transfectant cell line (100%), although this difference was not statistically significant (p ϭ 0.125). Likewise, pro-MMP-9 secretion from the V3B3⌺11aa cell line (53.0%) was also reduced compared with C1 cells (43.8%; p ϭ 0.204). Therefore, the ␤6 C terminus was unable to potentiate pro-MMP-9 secretion in these SCC cells, despite the fact that the ␤6 C-terminal residues, when associated with ␤3, were capable of promoting invasion.
In contrast, Fig. 6, C and D, shows that there was a dramatic increase in the levels of pro-MMP-2 secreted by the V3B3⌺11aa cells when plated on vitronectin compared with either C1 or V3B3. Therefore, the C-terminal 11 amino acids of ␤6-integrin are sufficient to regulate secretion of gelatinases, but it seems that the specific gelatinase undergoing regulation is dependent upon the particular integrin heterodimer context in which these amino acids are placed.
V3B3⌺11aa Invasion Is Mediated by MMP-2-The above data suggested that the amino acid sequence EKQKVDLSTDC, when expressed as a chimeric-␤3 subunit within the integrin ␣v␤3, is sufficient to promote both invasion and an enhanced secretion of pro-MMP-2. To determine whether the increased expression of pro-MMP-2 was required for invasion, Matrigel invasion assays were conducted in the presence of an MMP-2 function-blocking antibody (CA-4001; 10 g/ml) relative to the W6/32 control antibody (Fig. 7B). Invasion of both C1 (112.9%) and V3B3 cells (100%) was not affected significantly upon introduction of the anti-MMP-2 antibody (91.3 and 93.4%, respectively). However, invasion of V3B3⌺11aa cells (194.3% with W6/32) was reduced by CA-4001 (118.4%) to levels comparable with either C1 or V3B3. Thus, the EKQKVDLSTDCdependent promotion of invasion by V3B3⌺11aa cells is the result of an enhanced expression of pro-MMP-2; blocking this enzymatic activity leads to abrogation of this enhanced invasive activity. DISCUSSION The ␣v␤6 integrin heterodimer is expressed de novo by keratinocytes during wound healing and in various epithelial neoplasms (2-6, 24, 25). In oral SCC cells, overexpression of ␣v␤6 enhances cellular invasion through a ligand-dependent up-regulation of secreted pro-MMP-9 (9). We have shown elsewhere that the unique C-terminal 11 amino acids of the ␤6 integrin subunit are critical for ␣v␤6-mediated oral SCC invasion. 2 This evidence (and accumulating pathological ␣v␤6-ex- pression data) indicates that ␣v␤6, specifically the C-terminal residues of the ␤6 subunit, may play an important role in the invasion of many carcinoma types and that this regulatory role seems to be (at least in part) via the control of MMP levels.
In this investigation, by assessing protease secretion from oral SCC cells expressing a truncation mutant ␤6 (lacking the C-terminal 11 amino acids) relative to full-length ␤6, we have confirmed that the unique C-terminal 11 residues are critical for ␣v␤6-mediated up-regulation of pro-MMP-9. These observations are consistent with the data of Niu et al. (1998) who, in a colon cancer model, demonstrated that this 11-amino acid sequence was necessary for a ligand-independent, protein kinase C-dependent up-regulation of MMP-9 (18). However, we have investigated this observation more intensively by generating oral SCC cells that express, equally, either wild-type ␤3 or an extension mutant ␤3 (␤3 extended by the ␤6 C-terminal 11 amino acids), V3B3 and V3B3⌺11aa, respectively. Assessment of adhesion to vitronectin (data not shown) and haptotactic migration toward vitronectin of V3B3 and V3B3⌺11aa cells, and the null-infectant C1 control cells (Fig. 4), demonstrated that the transduced wild-type or mutant ␣v␤3 integrin heterodimers were functional and able to generate comparable activities in these lines in terms of adhesion and cell migration.
Cells expressing ␣v␤3⌺11aa were considerably more invasive through a matrix of basement membrane-like extracellular matrix molecules than cells expressing wild-type ␣v␤3. However, gelatin zymography demonstrated no significant difference in MMP-9 expression between V3B3 and V3B3⌺11aa (p ϭ 0.204) (Fig. 6, A and B), apparently indicating that the promotion of invasion, dependent on the ␤6 C terminus, was not mediated by an induction of MMP-9. It was somewhat surprising, however, that V3B3⌺11aa cells secreted substantial quantities of pro-MMP-2, whereas the V3B3 and C1 cells failed to produce a detectable amount (Fig. 6, C and D). This indicates that the C-terminal 11 amino acids of the ␤6 integrin subunit, depending on the heterodimer context, are sufficient to enhance SCC cellular invasion by promoting either gelatinase A (MMP-2) or B (MMP-9) production.
Antibody-blockade demonstrated that the enhanced levels of V3B3⌺11aa invasion, mediated by the ␤6 C-terminal 11 amino acids of the chimeric integrin, were dependent upon both ␣v␤3 expression and MMP-2 levels, an observation in agreement with the correlation drawn between up-regulated levels of MMP-2 secreted by these cells and increased invasive activity. The enhanced levels of invasion, because of and mediated by the up-regulation of MMP-2, are consistent with many reports concerning the role of MMP-2 in cancer progression (26 -35).
The ␣v␤3 heterodimer has been implicated in the malignant behavior of various tumor types (36 -42), and the regulation of tumor angiogenesis when expressed on endothelial cells (37,(43)(44)(45)(46)(47)(48). In certain cases, expression of ␣v␤3 has been consistent with an up-regulation of MMP-2 (28, 36, 49 -51), an MMP that has been reported to associate directly with ␣v␤3 (45,(52)(53)(54). However, in this oral SCC model (V3B3 and V3B3⌺11aa), expression of wild-type ␣v␤3 promoted neither MMP-2 production nor invasion. Instead, the 11 amino acids of ␤6 were required to mediate such an effect. One possibility for these different findings from various research groups is that the pro-invasive functions of both ␣v␤3 and ␣v␤6 may well be cell-type specific; these findings could relate to the epithelial origins of the SCC cells, a cell type not used much in previous studies by other investigators (49 -51).
Thomas et al. (9) reported a small, yet significant, increase in pro-MMP-2 in VB6 cells relative to C1. Therefore, although pro-MMP-2 ordinarily was undetectable under the experimental conditions used in this study (in both C1 and VB6 cells), wild-type ␣v␤6 also may have a weak, and possibly inconsistent, role in MMP-2 up-regulation. A possible explanation for the pro-invasive pro-MMP-2 regulation, observed with the ␣v␤3⌺11aa heterodimer, could be that tethering the ␤6 C terminus to ␤3, rather than ␤6, altered the hierarchy of proteases regulated by the 11 amino acids. It is also possible that the substrate (fibronectin was used for VB6/V3B6⌬11aa and vitronectin was used for V3B3/V3B3⌺11aa because these were the major ligands for each integrin/mutant-integrin pair) may also have influenced which MMP was up-regulated.
In this study we show, for the first time, that the C-terminal 11 amino acids of ␤6 are not only essential for but also sufficient to promote invasion. Depending on heterodimer context, this occurs through regulation of either MMP-9 or MMP-2; in each case, however, the enhanced elaboration of these enzymes is a necessary component of invasion, because selective blockade of the specific MMP inhibits the observed increase in invasive capacity. As a group, these data demonstrate that the C-terminal 11 amino acid sequence of the ␤6 integrin subunit has a pivotal role in the regulation of MMPs and, therefore, in the mediation of invasion of oral SCC. The implication from our analyses is that this region of the ␤6 subunit is a fundamental regulator of protease activity, and hence invasion, in carcinoma cells.