|Home | About | Journals | Submit | Contact Us | Français|
Increased expression of membrane-type matrix metalloproteinases (MT-MMPs) has previously been reported to correlate with increasing grade of malignancy in gliomas, a relationship shared with alterations in epidermal growth factor receptor (EGFR) signaling. To investigate the possibility of a causative role for EGFR signaling in increasing MT-MMP expression and subsequent peritumoral proteolysis, we characterized glioma cell lines for expression of MT1-MMP, MT2-MMP, MT3-MMP, and MT5-MMP by Western blotting and by quantitative real-time polymerase chain reaction analysis, and for MMP-2 activity following epidermal growth factor (EGF) stimulation. EGF stimulation of glioma cell lines resulted in a 2- to 4-fold increase in MT1-MMP mRNA levels. Although there were slight differences in MT2-, MT3-, and MT5-MMP mRNA expression following EGF stimulation, none of these demonstrated an increase similar to that of MT1-MMP expression. Treatment of high-grade glioma cell lines U251MG and IPSB-18 with EGF for 24 h resulted in a several-fold increase in MT1-MMP protein (2.5- and 5.1-fold, respectively) and in cyclin D1 (2.9-fold), as compared to untreated controls. No significant increase was detected in other MT-MMPs at the protein level. Although there was no detectable increase in proMMP-2 protein, there was an increase in MMP-2 activity. Furthermore, the MT1-MMP induction by EGF was prevented by pretreatment with the EGFR-specific tyrphostin inhibitor AG1478. Similarly, treatment with the phosphatidylinositol 3-kinase inhibitor LY294002 prevented the induction of MT1-MMP protein by EGF stimulation. These compounds additionally inhibited EGF-stimulated invasion in Matrigel Transwell assays. Our results indicate that one mechanism of EGFR-mediated invasiveness in gliomas may involve the induction of MT1-MMP.
Tumors of glial origin represent the most common intrinsic CNS cancer, accounting for 44% of all primary CNS tumors, of which 23% are glioblastomas (GBMs)3 (CBTRUS, 2002–2003). Despite recent advances in surgical and adjuvant technologies, the prognosis remains poor for most glioma patients because of the aggressive nature of these tumors and their propensity for recurrence. Following the evidence that glioma recurrence results from a capacity to invade the contiguous brain and perivascular spaces, novel attempts to prevent tumor growth and recurrence have begun to target invasion-related proteins such as matrix metalloproteinases (MMPs).
Membrane-type-1 matrix metalloproteinase (MT1-MMP) overexpression has been shown to correlate with malignancy and tissue invasion in many cancers, including malignant gliomas (Yamamoto et al., 1996). In an attempt to identify molecular mediators of this overexpression, we have chosen to examine the epidermal growth factor receptor (EGFR) signaling system because of its common deregulation in glioma malignancy and its known ability to stimulate tumor cell motility in other experimental systems (Huang et al., 1997). Oncogenic signaling via the EGFR has been found to be a frequent alteration, with 30% to 50% of glioblastoma cases showing gene deletion, gene amplification, gene overexpression, or duplication of EGFR gene segments (Ekstrand et al., 1992, 1994). All of these changes can result in over-activation of EGFR signaling. Stimulation of the EGFR by its cognate ligands, such as epidermal growth factor (EGF) and transforming growth factor α (TGFα), is thought to play a significant role in the invasive growth of gliomas and could therefore play a role in enhancing proteolysis (Ekstrand et al., 1991). The potential for enhanced EGFR signaling to increase the expression of genes implicated in invasion and metastasis has been addressed in several cancer models, demonstrating that, apart from the mitogenic signaling effects, proteases and other genes may be upregulated by growth factor signaling as well (Lohi et al., 1996; Maity et al., 2000; Miyagi et al., 1998; Monaghan et al., 2000).
There is strong evidence that MMPs are an integral part of extracellular matrix (ECM) remodeling processes involved in tumor cell metastasis. Other cancers in which EGFR signaling plays a prominent role, including breast and ovarian carcinoma, have shown abnormally high levels of activated MMPs (Davidson et al., 1999). MMPs, including the secreted MMP-2 and the membrane-bound MT1-MMP, have been shown to be overexpressed with increasing malignancy in gliomas (Nakano et al., 1995; Sawaya et al., 1996; Yamamoto et al., 1996).
In 1994, Sato and colleagues reported the cloning of a membrane-type MMP, designated MT1-MMP, with a single pass, transmembrane-spanning domain, in addition to the other conserved MMP domains (Sato et al., 1994). Subsequently, five other MT-MMPs have been described (MT2- to MT6-MMP; Llano et al., 1999; Puente et al., 1996; Takino et al., 1995; Velasco et al., 2000; Will and Hinzmann, 1995). In gliomas, several of the MT-MMP subtypes have been associated with increasing grade of malignancy, notably MT1-, MT2-, and MT5-MMP (Lampert et al., 1998; Llano et al., 1999; Yamamoto et al., 1996). Further studies have implicated these membrane-type MMPs as membrane receptors, which bind to and activate pro-gelatinase A (pro-MMP-2) through a heteromolecular complex with tissue inhibitor of metalloproteinases-2 (Shofuda et al., 1998). Activation of pro-MMP-2 by MT-MMPs appears to be a critical step in invasion and metastasis, initiating further proteolytic cascades via cleavage of the pro-forms of collagenase-3 (MMP-13) and 92-kDa gelatinase B (MMP-9) (Knäuper et al., 1996). More pertinent to gliomas, human MT1-MMP cDNA has been previously transfected into U251MG cells, resulting in an increase in active MMP-2 and increased invasive behavior in vitro, further implicating MT1-MMP as a causative factor in this aspect of glioma malignancy (Nakada et al., 2001). In U251MG xenografts, MT1-MMP was shown to augment growth and vascularity of tumors in vivo, suggesting a potential role for MT1-MMP in angiogenesis (Deryugina et al., 2002).
In the present study we tested the hypothesis that stimulation of EGFR signaling in human glioma cell lines would result in an increase in expression levels of MT-MMPs. MT-MMPs may be critical to invasive growth in malignant gliomas, calling for the elucidation of factors that drive their expression. The present data suggest that increased EGFR signaling, a hallmark of malignant gliomas, may augment proteolytic invasion of glioma cells by causing overexpression of MT1-MMP.
Established cell lines and short-term cultures were frozen in freezing medium composed of complete medium supplemented with 20% fetal bovine serum and 5% to 10% sterile dimethyl sulfoxide (DMSO) as a cryoprotectant. The high-passage cell line U251MG, initially derived from a glioblastoma specimen, originated from a European source. The IPSB-18 cell line was derived from a WHO grade III malignant glioma tissue specimen and has been extensively characterized (Knott and Pilkington, 1990; Knott et al., 1990, 1991).
Cultured cells were maintained in tumor growth medium at 37°C, 5% CO2, 95% humidity in a tissue culture incubator. Growth medium was composed of a base of Dulbecco’s Modified Eagles Medium (DMEM) with high glucose, pyridoxine hydrochloride, L-glutamine, and 110 μg/ml sodium pyruvate (Invitrogen/Gibco BRL Life Technologies, Rockville, Md.), supplemented with 1% penicillin-streptomycin solution (100 units/ml penicillin, 100 μg/ml streptomycin, final concentration; Life Technologies), and 10% fetal bovine serum (Life Technologies). Serum-free medium was DMEM/F12 Ham mix (50:50, Life Technologies), supplemented with 1% antibiotic solution. Subculturing was accomplished by washing cells once with Hanks’ Balanced Salt Solution for 5 min, applying approximately 5 ml of a trypsin–ethylenediamine triacetate (EDTA) solution (0.05% trypsin, 0.53 mM EDTA-4Na) per 107 adherent cells, incubating for 5 min at 37°C, and centrifuging cell suspensions at 1000 rpm for 5 min. Supernatants were discarded, prior to resuspending cells in 10% fetal calf serum–supplemented DMEM as above.
Recombinant human epidermal growth factor (rhEGF) was obtained commercially (Fisher Scientific, Pittsburgh, Penn.), reconstituted in sterile serum-free growth medium, and frozen in aliquots at – 20°C until use. Stock solutions of EGF (100 μg/ml) were diluted to various concentrations in serum-free medium, mixed, and added directly to culture dishes prior to the addition of suspended cells. Cells were incubated for 24 h, followed by cell lysis and collection of proteins for analysis. For time-course experiments, incubation was for the times indicated, followed by one brief wash with sterile phosphate-buffered saline (PBS), pH 7.4, and protein collection.
The selective pharmacological inhibitor to EGFR autophosphorylation, tyrphostin AG1478, and the general tyrosine kinase inhibitor, genistein, were obtained from Calbiochem (La Jolla, Calif.). Both were diluted in 1 ml sterile DMSO upon receipt and stored at −20°C in light-protected containers until use. AG1478 was applied to cell lines under serum-free conditions, at 10 to 50 nM (for EGFR, reported IC50 = 5 nM, obtained from the manufacturer). The phosphatidylinositol (PI) 3-kinase inhibitor LY294002 was also obtained from Calbiochem. LY294002 was reconstituted in 1 ml sterile DMSO and stored as stock solutions at – 20°C until use. LY294002 was diluted to 10 μM (IC50 = 1.4 μM) for a pretreatment period of 15 min prior to EGF stimulation. Genistein was also used at 10 μM (Calbiochem).
Glioma cells were treated with recombinant human EGF as described, prior to RNA extraction using Trizol Reagent, according to the manufacturer’s protocol (Invitrogen, Gaithersburg, Md.). Probes and primer sets were designed to span intronexon junctions by using Gen-Bank sequences. Prepared RNA samples were tested in triplicate over several trials. Fluorescence signals for MT1-MMP were measured against 18S RNA expression levels. Polymerase chain reactions (PCRs) were run under the following conditions. The experiments were performed in the ABI Prism 7900 Sequence Detection System (Applied Biosystems, Foster City, Calif.) using the TaqMan One-Step RT-PCR Master Mix Reagents Kit (Applied Biosystems, P/N 4309169). All the samples were tested in triplicate under the conditions recommended by the fabricant. The cycling conditions were as follows: 48°C for 30 min; 95°C for 10 min; and 40 cycles of 95°C for 15 s and 60°C for 1 min. The cycle threshold was determined to provide the optimal standard curve values (0.98 to 1.0). The probes and primers were designed by using a software program designed for this purpose (Primer Express, version 2.0, Applied Biosystems). The probes were labeled in the 5' end with FAM (6-carboxy-fluorescein) and in the 3' end with TAMRA (6-carboxytetramethyl-rhodamine). Ribosomal RNA (18S rRNA) from the Pre-developed TaqMan Assay Reagents (P/N 4310893E) was used as endogenous control. The reactions and the synthesis of the probes and primers were performed in the Virginia Commonwealth University Nucleic Acid Research Facilities.
Cell lysates were prepared in a modified radioimmunoprecipitation assay (RIPA) buffer containing 50 mM tris(hydroxymethyl)aminomethane (Tris)-HCl, pH 7.4, 1% NP-40, 0.25% Na deoxycholate, 150 mM NaCl, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM activated sodium orthovanadate, and 1 μg/ml each of aprotinin, leupeptin, and pepstatin (all Sigma, St. Louis, Mo.). Monolayers were rinsed for 5 min in PBS pH 7.4, and 0.5 to 1 ml of chilled RIPA was added for every 107 cells. Cells were lysed for 30 min at 4°C with agitation, scraped with sterile cell scrapers and supernatants, and procured following centrifugation for 10 min at 14,000 rpm in an Eppendorf 1513 C micro-centrifuge (Eppendorf, Westbury, N.Y.) at 4°C. Lysates were assayed for protein concentration by using the Coomassie Plus Protein Assay (Pierce Chemical Company, Rockville, Ill.).
The RIPA lysates were pre-cleared by adding 100 μl protein A-agarose bead slurry (Roche Molecular Biochemicals, Indianapolis, Ind.) for every 1 mg of protein and incubating for 10 min on a rotator at 4°C. Beads were removed by brief centrifugation at 14,000 rpm for 10 s. Precleared lysates were diluted to 1 μg/ml following protein quantification, and 500 μl lysate was used to precipitate the protein of interest. EGFR antisera were added directly to each sample and incubated at 4°C on a rotator overnight. Antibody-protein complexes were captured by adding 40 μl protein A agarose bead slurry to each sample and rotating for 1 to 3 h. Collection of complexes was accomplished by centrifugation at 14,000 rpm for 10 s, followed by two washes with RIPA and a final wash with PBS. Collected bead-IgG-protein complexes were then boiled for 5 min in 2× Laemmli sample buffer (Biorad Laboratories, Hercules, Calif.). One final centrifugation at 14,000 rpm for 2 min was used to remove the beads. The supernatants were analyzed by sodium dodecyl sulfate-poly-acrylamide gel electrophoresis (SDS-PAGE) and Western blotting.
Total protein was analyzed in 30-μg samples via Western blot. Briefly, samples were diluted in Nanopure water (Barnstead International, Dubuque, Iowa), and 10× Laemmli sample buffer was added. Samples were boiled 5 min and loaded directly into prepoured Tris-HCl-glycine SDS-PAGE gels along with prestained molecular-weight standards (Broad Range Protein Ladder; Calbiochem). Electrophoresis was performed in Tris/glycine/SDS running buffer (25 mM Tris, 192 mM glycine, 0.1% sodium dodecyl sulfate) at a constant 100 V. Gels were transferred to 0.2-μm-pore nitrocellulose paper (Biorad Laboratories) on ice at 290 mA constant current for 3 to 4 h, followed by 375 mA for 45 min for higher molecular weight proteins. Following transfer, blots were briefly transferred to Tris-buffered saline Tween (TBST) solution (10 mM tris, 150 mM NaCl, and 0.5% Tween-20, pH 8.0), blocked for 1 h at room temperature (RT) in TBST, and supplemented with 5% nonfat dry milk, and the primary antibody was applied for a further 1 h at RT in fresh blocking buffer. Following four 10-min washes with TBST, species-specific horseradish peroxidase-conjugated secondary antiserum (Rockland, Gilbertsville, Penn.) was applied for 1.5 h at RT with agitation, washed for 1 h with frequent buffer changes, treated with enhanced chemiluminescence reagents (Amersham Lifesciences, Piscataway, N.J.) for 1 min at RT, and subjected to autoradiographic film (Marsh Bio Products, Inc., Rochester, N.Y.). Film was developed on a Kodak X-Omat automatic film processor and densitometry performed, including analysis with Imagequant software (Amersham Biosciences).
Mouse monoclonal antiphosphotyrosine antiserum clone 4G10 was obtained from Upstate Biotechnology (Lake Placid, N.Y.). Mouse monoclonal antisera against MT1-MMP catalytic domain (clone IM39L, 1:50) and cyclin D1 (clone Ab-3, 1:100) were obtained from Calbiochem. Rabbit polyclonal MMP-2 antiserum (clone AB809; 1:2000) and rabbit polyclonal antibody against the hinge region of MT1-MMP (AB815; 1:2000) were obtained from Chemicon (Temecula, Calif.). Goat polyclonal EGFR antibody (clone sc-03-G; 1:250) and mouse monoclonal antihuman β-actin (1:5000) were obtained from Santa Cruz Biotechnology (Santa Cruz, Calif.) and Sigma, respectively.
Samples of serum-free tumor-conditioned medium (CM) were collected following EGF treatment of cell cultures in sterile 15-ml tubes, and debris was removed by centrifuging samples for 5 min at 2000 rpm (805 × g). Samples were transferred to fresh sterile tubes and stored at −20°C until use. CM was analyzed by SDS-PAGE under nonreducing conditions. Polyacrylamide SDS-PAGE gels (10%) co-polymerized with purified gelatin were used for analysis of gelatinase activity (Biorad Laboratories). Protein samples were run for 1.5 h at 100 V. After a brief wash in Nanopure water, proteins were renatured by incubation in 2.5% Triton X-100 (Rohm and Haas, Philadelphia, Penn.) twice for 15 min, further rinsed briefly in Nanopure water, and incubated at 37°C for 24 h in developing buffer (50 mM Tris, pH 7.6, 10 mM CaCl2, 0.1% NaN3, 1 μM ZnCl2). After development, gels were stained with Commassie blue R-250 (Biorad Laboratories, 0.5% solution in water:methanol:acetic acid 40:50:10) for 4 h at RT with agitation and destained twice for 30 min in 40:50:10 solvent, and twice for 30 min with 10% methanol, 7% acetic acid. Protein clearing in the stained gel, indicative of protease activity specific to the substrate, was photographed by using a gel documentation camera and standard autoradiograph lightbox.
Polycarbonate filters (9 mm, 8.0-μm pore size) were pre-coated with 200 μl of Matrigel (Becton Dickinson, Bed-ford, Mass.), diluted to 1 mg/ml in serum-free medium (DMEM/F12), and applied for 30 min at 37°C until gelled, after which cell suspension was applied. Glioma cells were grown under standard conditions (see above) and weaned from 10% fetal calf serum to serum-free medium in 2 steps to 5% and 0%, over a 24-h period. Following trypsinization and cell counting using trypan blue dye exclusion and a Brightline hemocytometer, cells were diluted such that 104 cells were added per well. The lower chamber was filled with 700 μl serum-free medium supplemented with EGF (10 ng/ml). Cells under various treatments were incubated for 48 h and then washed once (5 min) with PBS (pH 7.4), fixed for 5 min in methanol (1 ml per chamber), air-dried briefly, and stained with eosin Y and methylene blue. The index of invasiveness (invasive fraction = fraction of total cells migrated to lower chamber) was calculated by counting total cell number/field (10 fields), removing the cells from the upper side of the filter, and then recounting the cells that migrated to the lower side of the filter (10 fields each). Average cell number below the filter, divided by average total cell number, gave the invasive fraction (also given as percent by multiplying by 100). Three separate wells were used per test condition, and these experiments were performed 2 or more times. Statistical analysis was performed to compare means from different test conditions, using the entire data set (Sigma Stat, Student’s t-test).
To characterize the influence of EGFR activation on MT-1 MMP expression in glioma cells, we used rhEGF to stimulate cultures that had been previously serum deprived for 12 h. The glioma cell lines U251MG, derived from GBM, and IPSB-18, derived from anaplastic astrocytoma, were treated with 50 ng/ml rhEGF for 24 h followed by RNA isolation. Quantitative reverse transcriptase PCR (RT-PCR) analysis revealed that MT1-MMP transcript levels were significantly increased in each cell line in comparison with untreated controls when normalized to 18S RNA (Figs. 1 A and B). Both cell lines tested demonstrated a significant augmentation in MT1-MMP levels, each with a roughly 5-fold increase in MT1-MMP mRNA expression relative to unstimulated controls (P < 0.001, Student’s t-test). Because expression of other MT-MMPs has been shown in brain tumors, and may play a function similar to that of MT1-MMP in protease activation and invasion in gliomas, we next tested whether a similar augmentation was seen in transcript levels of three other MT-MMPs (Van Meter et al., 2001; reviewed in Fillmore et al. ). When MT2-MMP, MT3-MMP, and MT5-MMP mRNA were examined following EGF treatment, results indicated that, although there were slight differences in their mRNA expression when compared to untreated controls, none of the differences observed compare to that of MT1-MMP induction following EGF treatment (Fig. 2).
To examine the effects of EGF stimulation on MT1-MMP protein expression, we prepared protein samples from different time points following EGF stimulation using RIPA lysis buffer, supplemented with phosphatase and proteinase inhibitors. Immunoprecipitation was performed to collect the EGFR protein and to examine the EGFR tyrosine phosphorylation state, indicating its activation. As shown in Fig. 3A, U251MG cells stimulated with EGF demonstrate a marked increase in EGFR phosphorylation, with robust detection at 5 min and decreasing thereafter. Unstimulated controls show very low levels of EGFR phosphorylation at both the 5-min and 18-h time points. Densitometric analysis of protein bands indicates EGFR phosphorylation was increased 12.2-fold at 5 min and then decreased to 5-fold by 20 min, relative to levels in the unstimulated controls. In contrast, the amount of EGFR protein in these samples remained constant in cell lysates, but appeared to diminish by 18 h. In parallel studies, IPSB-18 cells showed a similar time course of EGFR activation (data not shown). When MT1-MMP protein levels were assessed in the same lysates (Fig. 3B), an induction was observed in the MT1-MMP protein level, apparent at the 18-h time point. Densitometric comparison of MT1-MMP protein bands from 5 min (basal levels) and 18 h (induced levels) after EGF stimulation showed a 2.5-fold increase after normalization to β-actin loading control.
These initial studies utilized an anti-MT1-MMP antibody (IM39L) specific for the catalytic domain, but that does not recognize the major 45-kDa cleavage fragment. A polyclonal antiserum (AB815) raised against the recombinant hinge region of MT1-MMP, which is retained in the 45-kDa breakdown product of MT1-MMP, has been used to assess MT1-MMP protein expression more fully by demonstrating the pro- and active forms, as well as the 45-kDa cleavage product (Overall et al., 2000). Similar time-courses of MT1-MMP protein induction were demonstrated when this antiserum was used in studies of U251MG and IPSB-18 cells (data not shown). Data from studies that are reported in Figs. 4 and and55 were also performed with the hinge-region-directed antiserum (AB815).
Because it has been reported that tumor cell lines cultured over time can result in increases in MMP expression, we performed identical experiments with a low-passage (passage 12) cell line, IPSB-18 obtained from a high-grade glioma (WHO grade III). IPSB-18 cells express moderate levels of EGFR and have been shown to be highly invasive in response to EGF (Knott et al., 1990; Koochekpour et al., 1995; Merzak et al., 1995; Rooprai et al., 2000). Additionally, U251MG expresses high levels of wild-type EGFR, whereas IPSB-18 cells express moderate levels (our own unpublished data). EGF stimulation resulted in an increase in MT1-MMP protein, as seen in Western blot analysis of IPSB-18 cell lysates (Fig. 4A). The dose-dependent increase in MT1-MMP protein was observed with increasing concentrations of EGF for a 24-h treatment period.
Entry into the cell cycle by EGF-stimulated mitogenesis is indicated by an increase in cyclin D1 protein expression, increasing 1.5- to 2.9-fold over basal (untreated) levels. MT1-MMP protein was increased 3.5- to 5.1-fold over basal levels in the IPSB-18 cell line, a slightly greater induction than the 2.5-fold increase observed for U251MG. Figure 4B represents densitometric analysis of protein, relative to β-actin loading control. Total MT1-MMP and MMP-2 levels were analyzed by summing the results of each band, performing a background correction, and dividing this value by a similar value obtained for the corresponding band detected for the loading control.
MMP-2 activation following EGF stimulation was also studied by the more sensitive gelatin zymography technique using equal loading of CM collected from IPSB-18 under each treatment condition. As shown in Fig. 4C, the latent, 72-kDa pro-form of MMP-2 was detected in each of the samples tested, roughly paralleling the results obtained from Western blot analysis of MMP-2. In contrast, detection of the lower molecular weight, 68-kDa activated form of MMP-2 demonstrates an appreciable increase in enzymatic clearance of the gelatin substrate, a result that closely parallels the levels of induced MT1-MMP protein shown in Fig. 4A.
Following the demonstration that EGF-stimulation leads to an increase in MT1-MMP expression in glioma cells, we sought to test the specificity of the EGF receptor in mediating this induction, using pharmacological inhibitors targeting EGFR tyrosine kinase activity and subsequent downstream signaling. U251MG cells were serum starved overnight (12 h) and treated with rhEGF (10 ng/ml) for 24 h with and without a 15-min preincubation with the EGFR-selective tyrphostin inhibitor AG1478 (10 nM) or PI 3-kinase inhibitor LY294002 (10 μM). EGF stimulation induced MT1-MMP expression and autoproteolytic turnover, evident in the strong increase in the 45-kDa species of MT1-MMP (1.5-fold overall, 3.3-fold for the 45-kDa species; Fig. 5). This induction was prevented by inhibition of EGFR tyrosine kinase activity using AG1478. In addition, inhibition of one of the key signaling intermediates of EGFR, PI 3-kinase, using the specific inhibitor LY294002, prevented the induction of MT1-MMP by EGF. Densitometric analysis of MT1-MMP protein levels quantifies this effect (Fig. 5B) and reveals a fold decrease in MT1-MMP expression below basal levels when cells were cotreated with EGF and either AG1478 (0.7- and 0.4-fold of control, for 60- and 45-kDa forms, respectively) or LY294002 (0.5- and 0.2-fold of control, for 60- and 45-kDa forms, respectively). This suggests that intrinsic EGFR activation and PI 3-kinase activity contribute to basal MT1-MMP protein expression in U251MG. Similar results were seen in IPSB-18 cells under the same treatments (data not shown).
The EGFR signaling inhibitors AG1478 and LY294002, used as described above to inhibit induction of MT1-MMP by EGF, were next tested for their ability to inhibit in vitro invasion of glioma cells through Matrigel in Transwell (Corning, Corning, N.Y.) assays. EGF was used to stimulate chemotactic invasion of glioma cells through Matrigel-coated Transwell filters, tested under serum-free conditions. Treatment with AG1478 (10 nM) and LY294002 (10 μM), shown to inhibit MT1-MMP protein induction by EGF and basal expression, caused a significant decrease in invasion of EGF-stimulated U251MG cells (P = 0.05 and 0.002, respectively; Fig. 6). In addition, the general tyrosine kinase inhibitor genistein was used in this assay to examine the specificity of the EGFR-mediated effect. Genistein (10 μM) also significantly decreased EGF-stimulated invasion of the U251MG cells (P = 0.02).
In the present study we have shown that EGFR signaling can lead to an increase in MT1-MMP mRNA and protein levels. After establishing the effect of recombinant EGF in causing the activation of EGFR in glioma cell lines, we performed quantitative RT-PCR analysis and Western blots to identify changes in MT1-MMP mRNA and protein expression. Our results demonstrate a 2.5-to 5-fold induction of MT1-MMP mRNA and protein following EGF stimulation. These results are consistent with a report that demonstrated a 10-fold increase in MT1-MMP mRNA expression in lung fibroblasts treated with EGF (Kheradmand et al., 2002). It is of additional interest that none of the other MT-MMPs examined showed an increase of similar magnitude with EGF stimulation, providing additional evidence for the specificity of EGF’s stimulation of MT1-MMP expression.
To examine changes in protein activity, we performed Western blots and analyzed molecular-weight forms of MT1-MMP. The 43- to 45-kDa form of MT1-MMP has been reported to be a proteolytic breakdown product that is formed by autoproteolytic cleavage of the extra-cellular catalytic domain of MT1-MMP following activation and activity at the cell surface (Hernandez-Barrantes et al., 2000; Overall et al., 2000). Densitometric analysis of this molecular-weight species indicates a much greater fold increase than the controls following EGFR activation, inducing as much as a 16.4-fold increase in 45-kDa MT1-MMP levels at 10 ng/ml of EGF (Fig. 4).
Pro-MMP-2 protein (72 kDa), which is catalytically activated by MT1-MMP, did not appear to increase with EGF treatment. An increase of activated MMP-2, however, has been noted in previously reported zymographic studies of growth factor–treated glioma cells in vitro (Rooprai et al., 2000) and is demonstrated in Fig. 4C. The increase in MMP-2 activity, but not pro-MMP-2, following EGF treatment in the human glioma cell lines is again consistent with studies reported by Kheradmand et al. (2002) in which the authors examined mRNA levels of MT1-MMP and MMP-2, as well as MMP-2 activity, in developing lung from either wild-type or EGFR-/-mice. Their results demonstrated that the levels of active MMP-2 were significantly lower in lungs from the EGFR-/- mice compared to the wild-type control and that this was due to lower levels of MT1-MMP.
In addition to its role in activating proMMP-2, MT1-MMP may have significant intrinsic proteolytic activity for a variety of molecules in vivo, a feature that has been demonstrated for many ECM proteins in vitro, including tenascin-C, collagens, and proteoglycans (d’Ortho et al., 1997). EGFR-mediated upregulation of MT1-MMP could therefore influence invasiveness independent of MMP-2 levels. Other membrane-type MMPs may also be induced by the EGF-EGFR signaling axis in gliomas, a subject that remains to be examined.
We further explored the role of EGF stimulation in MT1-MMP induction using the selective pharmacological inhibitors AG1478, blocking EGFR signaling, and LY294002, which has been shown to block PI 3-kinase activity resulting from EGFR activation (Maity et al., 2000). The induction of MT1-MMP seen with EGF stimulation was prevented by both inhibitors (Fig. 5). Loading controls demonstrate even levels of the housekeeping protein β-actin, suggesting that this decrease is not due to cell loss or growth inhibition. In addition, cells were monitored for morphological signs of apoptosis, and no differences were observed.
Finally, the ability of EGF to stimulate invasion through an ECM barrier (Matrigel) was tested by using Transwell invasion assays. The ability of EGF-stimulated U251MG cells to invade through the ECM barrier was significantly reduced in cells treated with either the EGFR signaling blocker AG1478 (10 nM, P = 0.05) or the PI 3-kinase blocker LY294002 (10 μM; P = 0.002), compared to control cells. The soy isoflavone genistein (10 μM) was also used in this assay as an additional general tyrosine kinase inhibitor, which also showed a significant decrease from controls (EGF, no inhibitor, P = 0.02). A group of controlled experiments are currently under way to selectively downregulate MT1-MMP using an antisense strategy and form the subject of another manuscript now in preparation.
The predominance of EGFR signaling as a receptor system capable of driving glioma malignancy is most evident in primary, or de novo, glioblastoma, in which EGFR amplification or mutation is present in as much as 50% of cases (Ekstrand et al., 1994). Mutation in the EGFR has been well characterized in gliomas, with the most common alteration being the vIII variant, wherein all or part of exons 2 through 7, encoding a portion of the extracellular domain, are deleted by aberrant splicing (Ekstrand et al., 1992). Mutations involving deletion of the C-terminal tail have also been noted. Several studies have shown that the N-terminally truncated mutant EGFR vIII is constitutively active and does not require ligand binding for tyrosine kinase activity and signaling (Nishikawa et al., 1994). This type of receptor mutant has been shown to confer an oncogenic growth advantage in 3T3 fibroblasts and in astrocytoma cells expressing the wild-type receptor. This effect was observed in both in vitro and in vivo studies (Nishikawa et al., 1994; Prigent et al., 1996).
EGFR has been shown to bind multiple ligands, including EGF, heparin-binding EGF, and TGFα, and concomitant expression of these elements has been shown to be a frequent occurrence in human glioblastoma tissues (Ekstrand et al., 1991; El-Obeid et al., 1997) and by the tumor cells themselves (von Deimling et al., 1992). Autocrine signaling is thought to be a potent stimulator of malignant behavior in EGFR-overexpressing tumors, with constant signaling leading to unchecked growth. Our data, shown in Fig. 3, are consistent with the findings of others in demonstrating rapid EGFR autophosphorylation and then downregulation of total EGFR protein levels upon stimulation with EGF (Burke et al., 2001). This is thought to occur via endocytosis in a feedback mechanism involving clathrin-coated pits and degradation by the lysomal pathway, after sorting through the early endosome.
Our results demonstrate that increases in EGFR signaling can lead to augmented MT1-MMP expression in cultured glioma cells. That EGFR tyrosine kinase inhibition and PI 3-kinase inhibition can prevent the EGF stimulated response supports the argument for EGFR signaling as a causative factor driving MT1-MMP expression. What remains to be proven is whether amplified EGFR, or specific EGFR mutations, can drive similar increases in MT1-MMP expression in vivo. Therefore, a larger examination relating gene amplification and mutation of EGFR with expression of EGFR protein and MT1-MMP in vivo in patient samples is warranted. Of particular interest in this regard are recent studies by Okada et al. (2003), in which amplification of EGFR was most evident in the invasive rim of human gliomas, further implicating EGFR in glioma invasiveness in vivo.
The present work suggests that EGFR and MT1-MMP may be linked as causative factors mediating glioma malignancy. MT1-MMP has recently been shown to promote angiogenesis by increasing maturation of vascular endothelial growth factor isoforms, by directly cleaving integrins and CD44, and by pericellular fibrinolysis as well (d’Ortho et al, 1997; Galvez et al., 2001; Kajita et al., 2001). Therefore, interference with MT1-MMP activity is an attractive target for both anti-invasive and antiangiogenic therapies. In addition, selective inhibitors of the EGFR tyrosine kinase presently in clinical trials may have the additional benefit of preventing MT1-MMP-dependent aspects of glioma malignancy.
The high-passage cell line U251MG, originally derived from a glioblastoma specimen, was a gift of Dr. Derek McCormick of The Queen’s University of Belfast, U.K. We are extremely grateful to the Hord, Cullather, and Crone families for their support of this work.
1This work has been supported by a project grant from the British Brain Tumour Association awarded to G.J.P. and by the Hord-Crone-Cullather Predoctoral Research Fellowship in Neurosurgery at Virginia Commonwealth University awarded to T.E.V. to support collaborative work between the two centers involved.
3Abbreviations used are as follows: CM, conditioned medium; DMEM, Dulbecco’s Modified Eagles Medium; DMSO, dimethyl sulfoxide; ECM, extracellular matrix; EDTA, ethylenediamine triacetate; EGF, epidermal growth factor; EGFR, epidermal growth factor receptor; GBM, glioblastoma; MMP, matrix metalloproteinase; MT, membrane type; PBS, phosphate-buffered saline; PCR, polymerase chain reaction; PI, phosphatidylinositol; rhEGF, recombinant human epidermal growth factor; RIPA, radioimmunoprecipitation assay; RT, room temperature; RT-PCR, reverse transcriptase PCR; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; TBST, Tris-buffered saline Tween; TGFα, transforming growth factor α; Tris, tris(hydroxymethyl)aminomethane.