An Immortalization-Dependent Switch in Integrin Function Up-regulates MMP-9 to Enhance Tumor Cell Invasion

The development of squamous cell carcinoma (SCC) of the epidermis occurs via a multistage, stepwise process whereby keratinocytes accumulate distinct genetic lesions in oncogenes and tumor suppressor genes that promote cell growth and cell survival (1). Clonal expansion of cells after each genetic lesion leads to the outgrowth of tumor cells that have acquired novel phenotypes that promote carcinoma progression. Two critical phenotypes acquired early in SCC progression are cellular immortalization and cellular transformation, and loss-of-function mutations of p53 and oncogenic activation of H-Ras are known to immortalize and transform epithelial cells, respectively (2–4). In addition, these genetic lesions are common events in human epidermal SCC (1–3) and play crucial roles in its development (1, 5). Furthermore, mutations in both Ras and p53 are correlated with increased invasion, suggesting that these mutations result in changes in cellular behavior that are important for the acquisition of an invasive phenotype (6, 7). Therefore, identification of gene regulatory pathways that are acquired during immortalization and transformation could reveal attractive targets for therapies designed to inhibit tumor progression.

Interactions between tumor cells and the extracellular matrix (ECM) can directly contribute to SCC progression. Integrins are the major receptors for cell adhesion to the ECM (8), and they can activate intracellular signals that regulate many processes related to tumor progression, including proliferation, survival, migration, invasion, and gene expression (9, 10). Some integrins, such as α6β4, undergo functional changes during malignant progression of epithelial cells (11). In addition, a recent study suggests that α6β4 can either promote or suppress tumor development depending on which mutations the carcinoma cells have acquired (12). However, it is still unclear whether distinct genetic lesions that are commonly accumulated during tumor development lead to altered integrin function, and if so, how this altered integrin function contributes to specific carcinoma cell functions, such as invasion and metastasis.

Members of the matrix metalloproteinase (MMP) family play important and diverse roles during tumor progression (13). MMP-9 is important for processes that occur at both early and late stages of tumor progression, including induction of angiogenesis and early tumor growth (14, 15), tumor cell invasion (15, 16), and metastasis (17). MMP-9 may contribute to tumor growth and progression by degrading the ECM, or by proteolysing other substrates that regulate tumor progression, such as growth factors, other MMPs, or proteinase inhibitors (13). Although MMP-9 is often produced by stromal or inflammatory cells in the tumor microenvironment (13, 15), MMP-9 production by carcinoma cells has been shown in many tumors, including epidermal SCC, and is likely to contribute to tumor progression (18, 19).

Integrin α3β1 is a receptor for laminin-332/laminin-5 (LN-332) that is expressed in many epithelial cells, including the basal keratinocytes of the epidermis (20, 21). α3β1 and LN-332 are also highly expressed in many invasive carcinoma cells, and they have been implicated in regulating MMP-9 expression, tumor cell invasion, and metastasis (22–27). However, it remains unclear whether functions of this integrin are altered by tumor-promoting mutations, or whether acquired functions of α3β1 in tumor cells contribute to carcinoma cell invasion. In the current study, we sequentially introduced a null mutation of p53 and an activating mutation of H-Ras into primary mouse keratinocytes to test whether either of these two genetic lesions alters α3β1-dependent gene expression during early stages of carcinoma development. We found that keratinocytes acquire α3β1-dependent MMP-9 gene expression during immortalization caused by loss of p53 function, a phenotype that is maintained after subsequent transformation by oncogenic Ras. α3β1 was also required for MMP-9 expression in immortalized and transformed human cells harboring p53-null mutations. Importantly, we show that this acquired regulation of MMP-9 expression promotes invasiveness of tumorigenic cells. These results identify a novel tumor cell–specific change in α3β1 function that contributes to carcinoma cell invasion, indicating that this integrin may provide an attractive target for anticancer therapies.

Generation of mutant mice. To generate mice heterozygous for both the α3-null and p53-null mutations (p53+/−:α3+/−), we crossed mice that were heterozygous for a null mutation in the α3 gene (28) with mice that were heterozygous for a null mutation in the p53 gene (29). Because the genes for α3 and p53 are closely linked on chromosome 11, we first generated p53+/−:α3+/− mice that carry linked p53-null and α3-null mutations on the same copy of chromosome 11. To this end, initial p53+/−:α3+/− mice with unlinked α3-null and p53-null mutations were crossed with wild-type mice, and offspring were screened for double heterozygosity, resulting from a crossover event between the α3 and p53 genes of a p53+/−:α3+/− parent. Once obtained, p53+/−:α3+/− mice with linked null mutations were interbred to obtain neonatal mice homozygous for both mutations (p53−/−:α3−/−) at a frequency of ∼25%.

PCR genotyping. PCR primers and conditions for α3 genotyping were described previously (27). PCR genotyping for p53 was carried out using oligonucleotide primers that detected either the p53 wild-type allele (5′-ATGGGAGGCTGCCAGTCCTAACCC-3′ and 5′-GTGTTTCATTAGTTCCCCACCTTGAC-3′) or the p53 null allele (5′-TTTACGGAGCCCTGGCGCTCGATGT-3′ and 5′-GTGGGAGGGACAAAAGTTCGAGGCC-3′). p53 PCR reaction conditions were as follows: denaturation at 92°C for 60 s, extension at 62°C for 45 s, annealing at 72°C for 60 s. The p53-null and wild-type alleles were amplified separately using 35 and 25 amplification cycles, respectively.

Derivation and culture of cells. The immortalized human keratinocyte cell line HaCat (a kind gift from Dr. Paul Higgins, Albany Medical College, Albany, NY) was cultured in DMEM (Biowhittaker) supplemented with 10% fetal bovine serum (Hyclone). The human carcinoma cell line, SCC-25, (a kind gift from Dr. Jim Rheinwald, Brigham and Women's Hospital, Boston, MA) was cultured as described previously (27). SV40 LTAg–immortalized keratinocytes and primary epidermal keratinocytes were prepared from wild-type or mutant neonatal mice and cultured as described (27). p53-null immortalized mouse keratinocytes (IMK) that express or lack α3β1 (IMK:α3+/+ cells and IMK:α3−/− cells, respectively) were established by continued passage of primary keratinocytes isolated from mutant neonatal mice. Transformed mouse keratinocytes (TMK) were generated by stable transduction of IMKs with oncogenic RasV12 using retrovirus, as described below. IMKs and TMKs were grown in keratinocyte growth medium, as described (27). Primary keratinocytes, IMKs, and TMKs were maintained at 33°C, 8% CO₂, on tissue culture plates coated with 30 μg/mL denatured collagen (Cohesion). For experiments, cells were subcultured on collagen (30 μg/mL), or laminin-332–rich ECM (LN-332 ECM) prepared from SCC-25 cells, as described (27). Phase-contrast micrographs of live cells were taken on an Olympus IX70 inverted microscope.

Western blotting. Cell lysates were prepared in Cell Lysis Buffer (Cell Signaling Technology), and 20 μg of protein was subject to 10% SDS-PAGE, transferred to nitrocellulose membranes, and assayed by Western blot. Primary antibodies were used at the following concentrations: rabbit anti-α3 integrin subunit, 1:5,000; mouse monoclonal anti-p53 (Cell Signaling Technology), 1:1,000; rabbit anti-keratin 14 (Covance, Inc.), 1:5,000; mouse anti-Ras (BD Biosciences), 1:1,000; mouse anti-actin (Sigma), 1:2,500; mouse anti-p21 (BD Pharmagen), 1:500; rabbit anti–phospho-MAPK [extracellular signal-regulated kinase (ERK)1/2; Cell Signaling Technology], 1:1,000; rabbit anti-MAPK (ERK1/2; Pierce), 1:5,000; and rabbit anti-involucrine (Covance, Inc.), 1:1,000. Horse radish peroxidase–conjugated secondary antibodies were used at the following concentrations: goat anti-rabbit IgG (Cell Signaling Technology), 1:2,000; goat anti-rabbit IgG (Pierce), 1:15,000; and goat anti-mouse IgG (Pierce), 1:15,000. Chemiluminescence was performed with the SuperSignal kit (Pierce).

Analysis of MMP-9 mRNA expression and protein secretion. Cells were plated on either collagen or LN-332 ECM in the appropriate medium for times indicated in the figure legends, then serum starved for 24 h. Culture supernatants were assayed for MMP-9 protein secretion by gelatin zymography as described (27). Total RNA was isolated and assayed for MMP-9 and β-actin mRNA expression by reverse transcription-PCR (RT-PCR) using primer sequences and reaction conditions described previously (30, 31). For certain experiments, an additional set of β-actin primers used was as follows: (forward, 5′-GCCAGGTCATCACTATTGG-3′; reverse, 5′-AGTAACAGTCCGCCTAGAAGC-3′). Conditions for these primers were as follows: 94°C for 30 s, 51°C for 30 s, and 72°C for 30 s, with 18 amplification cycles. PCR primers for amplification of human MMP-9 and human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were described previously (32). Primers for human integrin α3 (hα3) were as follows: forward, 5′-AAGCCAAGTCTGAGACT-3′; reverse, 5′-GTAGTATTGGTCCCGAGTCT-3′. Human MMP-9 was amplified using the following conditions: 94°C for 30 s, 60°C for 30 s, and 72°C for 1 min, with 28 amplification cycles. hα3 and human GAPDH were amplified using the following conditions: 94°C for 30 s, 55°C for 30 s, and 72°C for 1 min, with 20 amplification cycles.

Adenoviral infection of keratinocytes. The adenoviral dominant-negative p53 mutant, p53mt135, (Clonetech) was described previously (33). Primary keratinocytes on LN-332 ECM were infected for 24 h with either adenovirus expressing p53mt135, or empty vector. Dose-response experiments indicated an optimal multiplicity of infection of 100. Twenty-four hours after infection, green fluorescent protein fluorescence was used to assess infection efficiency, and cells were prepared for RT-PCR or Western blotting, as described above. cDNAs for hMMP-9 (a kind gift from Dr. Ruth Muschel, Children's Hospital of Philadelphia, PA), or lacZ were cloned into the pAD/CMV/V5-DEST vector (Invitrogen) using the gateway cloning system (Invitrogen), and then transfected into the packaging cell line, QBI (Q-Biogene), using lipofectamine plus reagent (Invitrogen). Viral supernatants were isolated and purified as described (33). Keratinocytes were infected for 24 h, then viral supernatant was removed, and the cells were assayed for MMP-9 protein secretion by gelatin zymography or tested for invasion as described below.

Retroviral and lentiviral transduction of keratinocytes. Oncogenic RasV12 was cloned into the pBAbe retroviral vector and introduced into the ecotrophic Phoenix packaging cell line (a kind gift from Dr. Garry Nolan, Stanford University, Stanford, CA) by transient transfection. Viral supernatant was added to IMKs for 48 h followed by culture in keratinocyte growth medium containing 10 μm puromycin (MP Biomedicals, Inc.) to select for stably transduced cells. Human MMP-9, hα3 (a kind gift from Dr. Martin Hemler, Dana-Farber Cancer Institute, Boston, MA) or lacZ, were cloned using the gateway cloning system (Invitrogen), into the plenti-4/TO/V5-DEST vector (Invitrogen), which was modified to express the hygromycin resistance gene. Lentivirus was packaged in 293FT cells using the distributor's protocol (Invitrogen), and viral supernatants were added to cells for 24 to 48 h, followed by culture in growth medium containing 25 μg/mL hygromycin (Sigma) to select for stably transduced cells. For shRNA experiments, we used MISSION lentiviral shRNA constructs (Sigma) that encoded either mouse MMP-9 shRNA, hα3 shRNA, or a nontargeting control shRNA. Lentivirus was prepared as described above, and cells were stably transduced and selected in growth medium containing 10 μm puromycin (MP Biomedicals, Inc.).

Luciferase reporter assays. Luciferase assays to test the activity of the transfected MMP-9 promoter were described previously (30). For p21 promoter assays, primary keratinocytes infected with adenovirus expressing p53mt135 or control adenovirus (see above) were incubated for 48 h, then cotransfected with a p21 promoter–driven luciferase reporter plasmid, and a TK promoter/Renilla luciferase internal control plasmid (pRLTK; Promega) at a 50:1 ratio. Twenty-four hours after transfection, whole cell lysates were collected and assayed using the Dual-Luciferase Reporter Assay kit (Promega) and the TD-20/20 luminometer (Turner Designs). Expression of each luciferase reporter plasmid was normalized to that from the control pRLTK plasmid for each sample.

S.c. injection of IMKs or TMKs. IMKs or TMKs (5 × 10⁶) were injected into the right flank of NCR nude mice (Taconic) in 200 μL of complete growth medium. Tumor length (l) and width (w) were measured using a Vernier caliper (Bel-Art Scienceware), and tumor volume was estimated using the following formula: tumor volume = (w² × l)/2. Mean tumor volume was calculated for each test group for each day of measurement.

Matrigel invasion assays. Cell invasion was assayed using a modified boyden chamber assay. TMKs (8 × 10⁴) were seeded into the top of Growth Factor–Reduced Matrigel Invasion Chambers (8-μm pore; BD Biosciences) in complete growth medium. Cells were allowed to invade for 24 h, and then fixed with 3.7% formaldehyde (Sigma). Cells on the top of the filter were removed, and cells that invaded to the bottom of the filter were permeabilized in 0.05% triton-X (Sigma), stained with 4′,6-diamidino-2-phenylindole (1 μg/mL), and quantified using a digital inverted fluorescent microscope and Image ProPlus software. Unless otherwise noted in the figure legends, n represents the number of separate experiments in which duplicate wells were averaged.

α3β1-dependent regulation of MMP-9 gene expression is acquired by immortalized and transformed keratinocytes. As a model system to study changes in integrin function that occur during early SCC development, we established p53-null immortalized and RasV12-transformed keratinocyte cultures through sequential introduction of a p53-null mutation and oncogenic RasV12 into primary mouse keratinocytes. First, p53-null-IMKs were established through isolation and extended culture (>50 population doublings) of primary mouse keratinocytes homozygous for a null mutation of the p53 gene, as described previously (2, 3). We observed that p53−/− primary keratinocyte cultures proliferated for several days after isolation, and then entered a crisis period during which many of the cells displayed a large and flattened morphology, and detached from the dish (Fig. 1A). Western blots showed increased expression of the terminal differentiation marker, involucrine, during weeks 2 through 4 of this crisis period (Fig. 1B). Within 6 weeks after isolation, p53−/− cultures exited crisis, were doubling readily, and contained cells with morphology similar to established immortalized p53−/− cultures (IMK; Fig. 1A). Involucrine staining in 6-week cultures was also reduced (Fig. 1B). In contrast to p53−/− cultures, p53+/+ cultures rarely exited crisis (Fig. 1A).

The IMKs formed small palpable nodules when injected s.c. into nude mice, but these nodules failed to develop into tumors and eventually disappeared (Supplementary Fig. S1). Because it is well-established that expression of oncogenic Ras renders keratinocytes tumorigenic (1, 2), we next established TMKs through stable expression of an oncogenic form of H-Ras, RasV12, in the IMKs. Expression of RasV12 did not dramatically alter phenotypic appearance (Fig. 1A) or involucrine expression (Fig. 1B) compared with IMKs. However, in contrast with the IMKs, the RasV12-transduced TMKs were able to form tumors that persisted in vivo (Supplementary Fig. S1) indicating that RasV12 expression conferred tumorigenicity.

To study the importance of integrin α3β1 during cellular immortalization and transformation, we similarly established immortalized and RasV12-transformed α3β1-deficient keratinocyte cultures (IMK:α3−/− cells and TMK:α3−/− cells, respectively) to compare with the α3β1-expressing keratinocyte cultures (IMK:α3+/+ cells and TMK:α3+/+ cells) described above. Absence of α3β1 did not interfere with cellular immortalization because p53−/−:α3−/− cultures were able to proliferate for >50 population doublings without senescing (data not shown). In addition, stable expression of RasV12 caused increased levels of phosphorylated ERK in both the TMK:α3+/+ cells and TMK:α3−/− cells (Fig. 1C), indicating that α3β1 is not required for ERK activation by RasV12. We next tested the effects of α3β1 deficiency on tumorigenicity. Like the IMK:α3+/+ cells, the IMK:α3−/− cells formed small palpable nodules that failed to develop into tumors and eventually disappeared (Supplementary Fig. S1). However, although the RasV12-transformed TMK:α3−/− cells were able to form small tumors that persisted in vivo, these tumors showed dramatically reduced growth compared with tumors derived from the TMK:α3+/+ cells (Fig. 1D; Supplementary Fig. S1B), indicating that α3β1 promotes more rapid tumor growth by transformed keratinocytes.

MMP-9 is a known regulator of carcinoma growth and progression (15–17). Because MMP-9 expression can be regulated by α3β1 (27, 30, 31), we tested whether this regulation changes during early carcinoma development in our progression model. We compared MMP-9 mRNA levels in primary, immortalized, and transformed keratinocytes that express or lack α3β1. p53−/− primary keratinocytes expressed abundant MMP-9 mRNA regardless of α3β1 expression (Fig. 2A), showing that α3β1 is not required for MMP-9 expression in primary keratinocytes that lack p53. In contrast, MMP-9 mRNA expression was dramatically reduced in α3β1-deficient IMK:α3−/− cells and TMK:α3−/− cells compared with α3β1-expressing IMK:α3+/+ cells and TMK:α3+/+ cells (Fig. 2B), suggesting that immortalization and transformation by two genetic lesions known to promote SCC development results in α3β1-dependent MMP-9 gene expression. MMP-9 mRNA expression was found to be dependent on α3β1 even when IMKs or TMKs were plated on either collagen or fibronectin (data not shown), neither of which are strong ligands for α3β1, consistent with previous findings that keratinocytes adhere and respond to endogenous LN-332 that they deposit into other ECMs (34).

Next, we tested whether α3β1 was similarly required for MMP-9 expression in immortalized and transformed human keratinocytes. For these experiments, the immortalized human keratinocyte cell line, HaCat, and the transformed human carcinoma cell line, SCC-25, were stably transduced with lentivirus encoding either an shRNA that targets the human α3 subunit, or a control shRNA. Suppression of α3 mRNA by stable expression of the α3 shRNA led to reduced MMP-9 mRNA levels in both the HaCat and SCC-25 cells (Fig. 2C), suggesting that immortalized and transformed human keratinocytes also require α3β1 for MMP-9 expression.

MMP-9 expression decreases during culture of primary keratinocytes regardless of α3β1, but reacquisition of MMP-9 expression in immortalized cells requires α3β1. Results in Fig. 2 suggest that cellular immortalization, caused by null-mutation of p53, leads to changes in α3β1-mediated regulation of MMP-9 gene expression. Because p53−/− keratinocytes exit crisis and first behave as immortalized cells within 6 weeks of isolation (Fig. 1), we predicted that MMP-9 expression would become α3β1-dependent within these first 6 weeks of culture. Indeed, RT-PCR analysis confirmed that by 6 weeks after isolation, p53−/−:α3−/− cultures showed dramatically reduced MMP-9 mRNA expression compared with p53−/−:α3+/+ cultures, when cultured on either collagen or LN-332 ECM (Fig. 3A). MMP-9 mRNA expression increased with continued passage of p53−/−:α3+/+ cultures but remained barely detectable in p53−/−:α3−/− cultures for >50 weeks (Fig. 3A). Similar results were obtained for eight independent p53−/−:α3+/+ cultures and three independent p53−/−:α3−/− cultures.

To determine the point at which p53−/−:α3−/− cultures lose MMP-9 mRNA expression, we performed time course experiments to monitor changes in MMP-9 expression during the first several days of keratinocyte culture. For these experiments, we seeded the cells on collagen because α3−/− keratinocytes are recovered from neonatal skin less efficiently than α3+/+ keratinocytes when seeded on LN-332 ECM (data not shown). Importantly, MMP-9 mRNA expression remains dependent on α3β1 when IMKs are seeded on collagen (Fig. 3A). Consistent with results from Fig. 2, MMP-9 mRNA expression was equivalently high in both p53−/−:α3+/+ and p53−/−:α3−/− primary keratinocytes on the second day of culture (Fig. 3B). To our surprise, MMP-9 mRNA expression decreased in both p53−/−:α3+/+ and p53−/−:α3−/− cultures during the next few days of culture, and was barely detectable by the 8th day after isolation (Fig. 3B). This initial decrease in MMP-9 mRNA expression was observed in 10 independent p53−/−:α3+/+ cultures, and 4 independent p53−/−:α3−/− cultures, and indicates that the acquisition of α3β1-dependent MMP-9 expression during immortalization is preceded by a loss of MMP-9 expression that is independent of α3β1.

Because MMP-9 mRNA expression can be regulated at the posttranscriptional level by changes in mRNA stability as well as at the transcriptional level (30, 35), we next determined if either the initial loss, or the reacquisition of MMP-9 mRNA expression that occurred during keratinocyte immortalization, was associated with changes in MMP-9 promoter activity. Using an MMP-9 promoter–driven luciferase reporter plasmid, we found that the initial decrease in MMP-9 mRNA levels observed during keratinocyte culture (Fig. 3B) was correlated with a decrease in MMP-9 promoter activity (Supplementary Fig. S2). Interestingly, IMKs also showed dramatically lower luciferase levels compared with two-day cultures (Supplementary Fig. S2), indicating that the subsequent reacquisition of MMP-9 mRNA expression observed during keratinocyte immortalization (see Fig. 3A) was not associated with increased MMP-9 promoter activity. Rather, this reacquisition of MMP-9 mRNA expression, under conditions of reduced promoter activity is consistent with our previous finding that α3β1 promotes MMP-9 mRNA stability in immortalized keratinocytes (30).

α3β1-dependent reacquisition of MMP-9 expression by IMKs requires loss of p53 function. Results in Fig. 3 indicate that α3β1-dependent regulation of MMP-9 expression is acquired through at least two separate events during immortalization, a process that in our progression model is mediated by null mutation of p53. Because p53 is a potent transcription factor that regulates the expression of many genes, we next tested whether p53 gene dosage significantly affects MMP-9 mRNA expression in freshly-isolated primary keratinocytes. MMP-9 mRNA expression was equivalently high in 2-day-old primary keratinocyte cultures that were p53+/+, p53+/−, or p53−/− (Fig. 4A). To account for possible compensatory changes in MMP-9 regulation that may have occurred during development of p53−/− mice, we also used a dominant-negative approach to disrupt p53 function in freshly isolated wild-type (p53+/+, α3+/+) keratinocytes. Despite significantly disrupting p53 function, this dominant-negative construct had no effect on MMP-9 mRNA expression (Supplementary Fig. S3), confirming that p53 does not directly regulate MMP-9 mRNA expression in nonimmortalized, primary keratinocytes.

Next, we performed time course experiments in α3β1-expressing primary keratinocytes that either expressed p53 (p53+/+) or lacked p53 (p53−/−) to determine if either the initial loss of MMP-9 during culture, or the reacquisition of MMP-9 during immortalization was dependent on absence of p53. MMP-9 mRNA expression dramatically decreased during the first week of culture in both p53+/+ and p53−/− keratinocytes, indicating that the initial decrease in MMP-9 mRNA expression occurs regardless of p53 (Fig. 4B). MMP-9 mRNA expression was restored in the p53−/−:α3+/+ keratinocytes as early as 3 weeks after isolation, and increased in the 4th and 6th weeks (Fig. 4B). In contrast, MMP-9 mRNA expression was not restored to p53+/+:α3+/+ cultures over the same time course (Fig. 4B). These results suggest that α3β1-dependent reacquisition of MMP-9 mRNA expression requires loss of p53 function, whereas the initial decrease in MMP-9 mRNA expression occurs independently of p53 function.

The above results suggest that p53 prevents the α3β1-dependent reacquisition of MMP-9 mRNA expression. However, because the majority of cells in the p53+/+:α3+/+ cultures appeared unhealthy by 6 weeks after isolation (Fig. 1), it was possible that failure to reacquire MMP-9 expression was the result of reduced viability rather than a direct effect of p53. Therefore, we tested whether restoring p53 function to immortalized α3+/+ keratinocytes similarly blocks MMP-9 expression. For these experiments, we used IMKs that express a temperature-sensitive variant of the SV40 large T antigen (LTAg), tsA58 (36), which immortalizes cells in part due to its ability to bind to and inactivate p53 (37). Importantly, we showed previously that similar to the p53−/− cells, these LTAg-immortalized cells display α3β1-dependent MMP-9 expression when cultured at 33°C, the permissive temperature for tsA58 (27). However, here we observed that MMP-9 mRNA expression was dramatically reduced in these cells when cultured at 39°C to inactivate tsA58 (Fig. 4C), indicating that restoration of p53 function prevents α3β1-dependent MMP-9 mRNA expression in immortalized keratinocytes.

α3β1-dependent MMP-9 expression promotes invasion of transformed keratinocytes. Although acquisition of an invasive phenotype is considered a late event in SCC progression, it is likely that changes in cellular signaling and gene expression that occur early in tumor progression can contribute to the acquisition of this phenotype. To directly test if α3β1-mediated regulation of MMP-9 expression that is acquired during SCC progression promotes invasion of transformed keratinocytes, we next tested whether α3β1 was required for TMK invasion using a modified boyden chamber Matrigel invasion assay. Importantly, although the particular laminin isoforms, or other proteins, that are present in Matrigel may not provide strong ligands for α3β1-mediated keratinocyte adhesion, endogenous LN-332 secreted by keratinocytes onto exogenous substrates is sufficient to promote α3β1-mediated adhesion and signaling (34). Absence of α3β1 from TMKs reduced invasion by almost 75% (Fig. 5A), indicating that α3β1 is required for maximal invasion of transformed keratinocytes. To confirm that restoration of α3β1 expression to TMK:α3−/− cells could promote invasion, we used a lentiviral approach to stably express hα3 in TMK:α3−/− cells (Fig. 5B). TMK:α3−/− cells stably transduced with hα3 showed an ∼3-fold increase in invasion compared with the same cells stably transduced with control lentivirus (Fig. 5B). Importantly, stable expression of hα3 also increased the amount of MMP-9 secreted by TMK:α3−/− cells (Fig. 5C, compare mMMP-9 levels in lanes 2 and 4), indicating that α3β1-dependent invasion is correlated with α3β1-dependent induction of MMP-9.

α3β1 may also be required for other processes, in addition to MMP-9 gene expression, that are important for invasion. Therefore, we next tested whether restoring MMP-9 expression to TMK:α3−/− cells promotes invasion in the absence of α3β1, by transducing TMK:α3−/− cells with lentivirus expressing human MMP-9. Gelatin zymography confirmed secretion of human MMP-9 protein by these cells (Fig. 5C,, lane 3). Although the expression of human MMP-9 was considerably lower than endogenous mouse MMP-9 levels expressed by TMK:α3+/+ cells (Fig. 5C,, lane 5), it was sufficient to increase invasion ∼3-fold compared with TMK:α3−/− cells stably transduced with the control lentivirus (Fig. 5D). These results indicate that increasing the expression of MMP-9 in the absence of α3β1 is sufficient to enhance invasion of transformed keratinocytes.

Next, to determine if α3β1-dependent invasion requires MMP-9, TMK:α3+/+ cells were stably transduced with a lentiviral construct encoding a short hairpin RNA that targets mouse MMP-9, which efficiently suppressed endogenous MMP-9 mRNA expression (Fig. 6A) and protein secretion (data not shown). Importantly, shRNA-mediated suppression of MMP-9 significantly reduced invasion of TMK:α3+/+ cells (Fig. 6B). Adenoviral expression of a nontargeted human MMP-9 in these same cells (Fig. 6C) was able to restore invasion (Fig. 6B), whereas a control adenovirus had no effect (Fig. 6B and C), showing that reduced invasive potential was not due to off-target effects of the shRNA. Together, these results show that the acquired ability of α3β1 to promote MMP-9 expression in tumorigenic keratinocytes is important in determining the invasive phenotype of these cells.

Previous studies have shown that regulatory pathways of MMP-9 gene expression, triggered in response to diverse extracellular cues, can be acquired or altered during cellular immortalization or transformation (27, 32, 35), and that altered MMP-9 expression can greatly effect tumor progression and tumor cell invasion (24, 38). Here, we developed a model system that allowed us to test if changes in integrin α3β1 function that result from specific genetic lesions known to promote tumorigenicity can influence the regulation of MMP-9 expression, and to determine the importance of this altered regulation in processes essential for tumor progression. Our findings reveal that immortalization of mouse keratinocytes, mediated by null mutation of p53, is a key event for the acquisition of α3β1-dependent MMP-9 gene expression, a phenotype that was maintained during subsequent transformation by oncogenic Ras. In addition, we found that an immortalized human keratinocyte cell line, HaCat, and a human carcinoma cell line, SCC-25, both of which harbor mutations in p53 (3, 39), also showed α3β1-dependent MMP-9 expression, suggesting that α3β1-mediated MMP-9 expression is also linked to p53 mutation in human cells.

Mutations in the p53 gene occur in a high percentage of human carcinomas (5, 40) and have been directly correlated with the malignant phenotype, tumor cell invasion, and MMP-9 expression (41–43). We observed two separate changes in MMP-9 mRNA expression during keratinocyte immortalization that was mediated by p53-null mutation. First, MMP-9 mRNA levels decreased during culture of freshly isolated primary cells, regardless of α3β1 expression, but this decrease was rapid and independent of p53, suggesting that it was not a direct result of immortalization. Second, MMP-9 expression was reacquired in an α3β1-dependent manner during outgrowth of immortalized cells, and this reacquisition was inhibited by the presence of functional p53. It is well-know that changes in MMP-9 gene expression can result from changes in posttranscriptional mRNA stability, as well as altered transcription (30, 35), and previous studies suggest that MMP-9 mRNA is unstable in the absence of signals that promote mRNA stability, due to the presence of AU–rich elements in its 3′-untranslated region (44). Interestingly, α3β1 promotes MMP-9 mRNA stability in immortalized keratinocytes, which we believe facilitates MMP-9 mRNA accumulation when MMP-9 gene transcription is reduced (30). Here, we provide evidence that MMP-9 promoter activity is reduced significantly in immortalized cells compared with primary cells (Supplementary Fig. S2), yet high MMP-9 mRNA expression is reacquired by α3β1-expressing cells during immortalization (Fig. 3), presumably due to α3β1-mediated mRNA stability (30).

Our current findings show that mutation of p53 permits a switch in α3β1 function to maintain high MMP-9 mRNA expression in tumor cells, under conditions where other extracellular cues that promote robust MMP-9 gene transcription are reduced. The mechanism by which p53 inhibits α3β1-mediated MMP-9 mRNA expression remains to be determined. However, based on our findings, we hypothesize that p53 may promote degradation of MMP-9 mRNA. As depicted in Fig. 6D, we propose that under conditions where MMP-9 promoter activity is reduced, p53 function prevents accumulation of MMP-9 mRNA even if α3β1 is expressed. Thus, loss of p53 during carcinoma progression permits α3β1-mediated accumulation of MMP-9 mRNA in tumor cells. In support of such a model, p53 has been shown to promote the degradation of mRNA (45). Furthermore, a recent study showed that p53 regulates the expression of several microRNAs (46), and some microRNAs are known to promote mRNA degradation (47). Our finding that MMP-9 mRNA expression was high in primary keratinocytes, regardless of p53 status, may reflect relatively higher MMP-9 promoter activity in these cells, so that enhanced mRNA stabilization is not required for substantial accumulation of MMP-9 mRNA.

Both α3β1 and MMP-9 are highly expressed in several human SCCs (18, 19, 23, 24), and MMP-9 is an important regulator of tumor cell invasion (15, 16). Importantly, we found that keratinocytes that lack α3β1, and therefore have reduced MMP-9 expression, showed dramatically reduced invasive potential, and that invasion could be restored by exogenous expression of MMP-9 (Fig. 5). In addition, shRNA-mediated knockdown of endogenous MMP-9 expression in α3β1-expressing cells reduced invasive potential (Fig. 6). Collectively, these results show that the regulation of MMP-9 by α3β1 in tumorigenic cells is important for cell invasion. MMP-9 has also emerged as a critical regulator of tumor angiogenesis and metastasis (14, 15, 17, 24, 42), suggesting that the regulation of MMP-9 by α3β1 is likely to influence processes important for several stages of carcinoma progression. Interestingly, absence of α3β1 from transformed keratinocytes was also correlated with reduced tumor growth in vivo. We are currently investigating whether reduced tumor growth by α3β1-deficient TMKs is due in part to loss of MMP-9 expression, which would be consistent with other studies showing that tumor cell-derived MMP-9 promotes tumor growth and angiogenesis (48, 49).

Although inhibitors that target MMP-9 and other MMPs have been the focus of recent preclinical and clinical studies, these inhibitors often cause side effects due to inhibition of MMPs in normal cells (50). Therefore, development of therapies that target specific MMP regulatory pathways that are uniquely acquired by cancer cells should circumvent such side effects. The fact that α3β1-dependent regulation of MMP-9 gene expression is acquired by immortalized/transformed keratinocytes identifies this pathway as a potential target for such therapies.

No potential conflicts of interest were disclosed.

Grant support: NIH R01CA84238 (C.M. DiPersio), and NIH R01CA81419 (K.M. Pumiglia). J.M. Lamar was supported by a predoctoral training grant from the National Heart Lung and Blood Institute (NIH-T32-HL-07194) and a predoctoral fellowship from the National Cancer Center (06118).

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.

We thank Lee Stirling and Vin Milano for excellent technical assistance; Drs. Jordon Kreidberg, Lawrence Donehower, Paul Higgins, Yasuyuki Sasaguri, Garry Nolan, and Bert Vogelstein for reagents and mutant mice; and Drs. Vandana Iyer, Peter Vincent, Dorina Avram, Andrew Aplin and Patrick Bryant, as well as Ethan Abel and Kara Mitchell, for helpful discussions.