Tumor cell invasion of von Hippel Lindau renal cell carcinoma cells is mediated by membrane type-1 matrix metalloproteinase

In previous studies [23], we used human RCC cells derived from a VHL null parental cell line (786-0), which was stably transfected with the pRc/CMV vector expressing a wild-type copy of VHL (WT8) or with the vector alone (pRc-9) [16]. Notably, these cells only express HIF-2α and do not express the HIF-1α isoform [17]. We determined that the lack of VHL tumor suppressor activity stabilized HIF-2α protein and increased MT1-MMP expression through direct transcriptional transactivation by HIF-2α (summarized in Table 1) [23]. These data provide a link between the loss of VHL function and the overexpression of an enzyme responsible for pericellular proteolysis of type I collagen and the activation of pro-MMP-2 [23, 35, 36]. In addition, our studies showed that MT1-MMP is the primary type I collagenolytic enzyme expressed by the VHL null cells. MMP-2 and MMP-9, which degrade type IV collagen in basement membrane, are also upregulated in these cells [23, 34]. Since metastatic tumor cells must invade through both interstitial collagen in the stromal environment and basement membrane [33], we hypothesize that MT1-MMP may contribute to the invasiveness of renal cell carcinoma in the context of VHL inactivation by mediating type I collagen invasion.

To begin our investigations, we characterized the WT8 and pRc-9 cells for differences in tumor cell invasion and matrix degradation that would suggest a role for MT1-MMP [24, 30, 35]. We first compared the ability of the WT8 and pRc-9 cells to invade through a type I collagen matrix using a Transwell^® insert assay system. Neither cell line invaded towards serum-free media. However, the pRc-9 cells, mutant for VHL, invaded approximately twice as effectively towards a serum chemoattractant when compared to the WT8 cells (Figure 1A). These data suggest that VHL expression suppresses tumor cell invasion through type I collagen. In similar in vitro invasion assays using Matrigel or type IV collagen, RCC cells mutant for VHL were also shown to be more invasive than RCC cells expressing wild-type VHL [34, 37, 38].

Invasive tumor cells have the ability to migrate and to remodel the surrounding, restrictive ECM using enzymes, such as MT1-MMP, for pericellular proteolysis [36, 39]. To ask whether the observed increase in cell invasion by the pRc-9 cells involves increased matrix degradation, we measured the ability of these cells to degrade a type I collagen matrix using an in vitro collagen degradation assay. This assay is based upon the fact that when a fibrillar collagen gel is degraded by cells embedded in it, culture medium is liberated. This medium can be recovered and weighed, and the amount of medium liberated provides a simple, but accurate, measure of matrix destruction [40, 41]. Indeed, we found that collagen degradation was increased by approximately 25% (P < 0.005) in the VHL null cell line, pRc-9, as compared to the WT8 cells expressing wild-type VHL, indicating that the lack of VHL function promotes type I collagen degradation (Figure 1B). A similar increase in collagen degradation was observed in the VHL null 786-0 parental line (data not shown).

Although the most well-described function of VHL is the negative regulation of HIF-α isoforms, VHL has many HIF-independent functions as well, including assembly of actin filaments [42], fibronectin matrix assembly [38, 43, 44], regulation of PKC isotypes [37], and endocytosis of FGFR1 [45]. Each of these VHL functions has been implicated in either RCC cell migration (cell movement in the absence of matrix) or invasion (cell movement through a matrix). To test whether the increased invasion and degradation of type I collagen observed in VHL null cells was due to the overexpression of HIF-2α target genes or to the loss of HIF-independent functions of VHL, we overexpressed HIF-2α in the context of wild-type VHL. Previously, we showed that transient overexpression of pCMV-HIF-2α in the WT8 cells was sufficient to drive transcription of a reporter construct containing HIF binding sites even in the presence of wild-type VHL [23]. Therefore, we introduced a pCMV-HIF-2α expression construct into the WT8 cells by transient transfection and measured the ability of these transfectants to degrade a collagen matrix. As shown in Figure 2A, overexpression of HIF-2α in the WT8 cells increased collagen degradation of these cells by ~30% (P < 0.0001), similar to the difference in collagen destruction between the WT8 and the pRc-9 cells (see Figure 1B). These results indicate that HIF-2α expression is sufficient to increase the collagen-degrading ability of cells that are wild-type for VHL, thereby suggesting that the increased invasion and matrix degradation measured in the pRc-9 cells may be due to aberrant expression of a HIF target gene(s) rather than the dysregulation of other VHL targets. These data support our hypothesis that as a target of HIF-2, MT1-MMP may contribute to VHL RCC tumor cell invasion.

To further test this hypothesis, we transiently transfected an MT1-MMP expression construct (MTpc3SE) into the WT8 cells and measured the ability of these transfectants to degrade type I collagen. This experimental design allowed us to study the effects of MT1-MMP expression on collagen invasion in the context of wild-type VHL and HIF-2α instability, thereby eliminating the contribution of other factors related to VHL mutations. Increased expression of MT1-MMP in the MTpc3SE transfected cells was confirmed by real-time RT-PCR, and MT1-MMP mRNA levels were approximately six-fold greater than the empty vector control transfectants (Figure 2B). The difference in MT1-MMP levels in these transfectants is comparable to the difference in endogenous MT1-MMP levels in the WT8 and pRc-9 cells previously published [23]. Furthermore, overexpression of MT1-MMP increased collagen degradation of the WT8 cells by approximately 30% (P < 0.0001; Figure 2C), again resembling the increase in degradation measured in the pRc-9 cells as well as the HIF-2α transfected WT8 cells (see Figures 1B, 2A). These data indicate that MT1-MMP may be the HIF target gene responsible for the increased collagen degradation by the VHL mutant cells since, to our knowledge, no other HIF target with the ability to cleave type I collagen has been identified in these cells.

Next, we tested whether MT1-MMP expression was sufficient to promote collagen invasion of the WT8 cells by transient expression of the MTpc3SE construct in these cells and measured the ability of the transfectants to invade through a type I collagen matrix using the assay system as described in Figure 1. Expression of MT1-MMP in the WT8 cells increased collagen invasion approximately two-fold compared to empty vector transfected cells, showing that exogenous MT1-MMP expression alone drives the invasiveness of cells that retain VHL function (P < 0.005; Figure 2D). Importantly, the difference in invasiveness between the WT8 transfectants mimics the difference in collagen invasion observed between the WT8 and pRc-9 cells (see Figure 1A). Together with our previous finding that MT1-MMP expression is regulated by HIF-2α in VHL null cells [23], our data suggest that one mechanism by which VHL may inhibit tumor invasion is by suppressing MT1-MMP expression through the negative regulation of HIF-2α protein.

To extend these findings, we asked if MT1-MMP was necessary for VHL RCC cell invasion of collagen using RNAi to specifically inhibit MT1-MMP expression. We could not use the pRc-9 cell line for these experiments due to the sensitivity of these cells to transfection and subsequent manipulations. Although expression of MT1-MMP was effectively reduced by three different targeting siRNAs in the pRc-9 cells (data not shown), once transfected, the cells did not survive trypsinization and culturing on top of collagen for use in the invasion assays.

Alternatively, we performed the siRNA experiments using the WT8 cells transfected to express either MT1-MMP or HIF-2α in order to mimic the ability of the pRc-9 cells to degrade collagen (see Figures 1B, 2A, 2C). Specifically, we co-transfected the WT8 cells with MTpc3SE and one of three specific MT1-MMP siRNAs or a non-specific control siRNA and measured MT1-MMP mediated collagen invasion. GFP expression was used to measure transfection efficiency and to track tumor cell invasion using FluoroBlok™ membranes as described below. A dose response analysis showed that the concentration of siRNA oligos that reduced MT1-MMP mRNA by >50% was 25 nM (data not shown). The reduction in MT1-MMP protein levels by the specific siRNAs was confirmed with an ELISA assay and was >60% as compared to the control siRNA (P < 0.005; Figure 3A).

In addition to its function as a matrix-degrading enzyme, MT1-MMP also acts as a cell surface receptor for pro-MMP-2 and promotes MMP-2 activation [46‐48]. Specifically, MT1-MMP, TIMP-2 and pro-MMP-2 exist in a trimolecular complex at the cell membrane [48]. When MT1-MMP levels are in excess of TIMP-2 and pro-MMP-2, a free MT1-MMP molecule cleaves the propeptide sequence of pro-MMP-2 (72 kDa), thereby generating the intermediate (69 kDa) form [49]. Fully active (67 kDa) MMP-2 is achieved by either intermolecular autolytic cleavage or by the activity of other enzymes [46, 47, 50]. WT8 cells express low levels of MMP-2 as well as TIMP-2 [23, 34]. To confirm that the reduction in MT1-MMP protein by siRNA was reflected in an inhibition of MT1-MMP enzyme activity, we measured the presence of pro, intermediate, and active forms of MMP-2 in the conditioned media collected from the transfectants using gelatin zymography. Expression of MTpc3SE in the WT8 cells resulted in the presence of the intermediate and active forms of MMP-2 (Figure 3B). Inhibition of MT1-MMP activity by each MT1-MMP siRNA was demonstrated by the absence of the MMP-2 intermediate and active species in these samples as compared to the media from the siRNA control transfectants (Figure 3B). The only other MMPs constitutively expressed by these cells in addition to MT1-MMP are MMP-2 and MMP-9 [23]. Therefore, it is important to note that the MT1-MMP siRNAs have no effect on the expression levels of MMP-2 or MMP-9 as seen by the unchanged levels of the bands representing these proteins in Figure 3B, confirming the specificity of these siRNAs against MT1-MMP.

Next, the ability of the siRNA transfectants to invade through a type I collagen matrix was measured using a similar in vitro invasion assay system described in the previous Figures. For these invasion assays, we used type I collagen-coated FluoroBlok™ membrane transwell inserts. FluoroBlok™ membranes block the transmission of light; therefore, non-invasive cells are not detectable, whereas invasive, pCMV-eGFP transfected cells can be counted by fluorescence imaging. By only visualizing GFP-expressing cells that have invaded, we could specifically quantitate the number of transfected cells that had invaded collagen while discounting the non-transfected cells. As before (see Figure 2D), WT8 cells expressing exogenous MT1-MMP displayed increased collagen invasion (P < 0.005; Figure 3C). Co-transfection of the control siRNA oligo had no significant effect on this increased invasion by MT1-MMP, whereas each of the three specific MT1-MMP siRNA oligos inhibited collagen invasion of the MT1-MMP transfectants by greater than 85% (Figure 3C). Taken together, these data suggest that MT1-MMP is important for tumor cell invasion by RCC cells.

We previously identified MT1-MMP as a target of HIF-2α [23], and our current data show that expression of HIF-2α in a wild-type VHL background induces the ability of the RCC cells to degrade type I collagen (see Figure 2A). Together these results suggest that the increase in collagen destruction mediated by HIF-2α overexpression in the WT8 cells is due to the induction of MT1-MMP expression. To test this hypothesis, we transfected the WT8 cells with pCMV-HIF-2α and either the control siRNA oligo or one of the three MT1-MMP siRNAs. As shown in Figure 4A, MT1-MMP protein levels were reduced by the target siRNAs in the HIF-2α transfectants by at least 50%. To ensure that the inhibition of MT1-MMP expression by the siRNAs was not due to non-specific effects of the siRNAs on HIF-2α protein levels, we measured HIF-2α protein in the transfectants. The WT8 cells do not endogenously express detectable levels of HIF-2α protein due to the presence of VHL [17] as shown in the empty vector control (Figure 4B); however, HIF-2α protein is detectable in these cells upon transfection of CMV-HIF-2α, and the MT1-MMP siRNAs had no effect on HIF-2α protein. Next, we employed the FluoroBlok™ invasion assay system described for Figure 3C to measure the ability of these transfectants to invade type I collagen. Inhibition of MT1-MMP blocked HIF-2α mediated invasion of these RCC cells by >90% (Figure 4C). To our knowledge, the only other HIF-2α target gene identified in playing a role in RCC tumor cell migration and invasion is the chemokine receptor, CXCR4 [51]. No effect of the MT1-MMP siRNAs on the expression of CXCR4 was detected in the HIF-2α transfectants (data not shown); thus, HIF-2α mediated cell invasion is blocked by specific inhibition of MT1-MMP. Together, these data imply that MT1-MMP is the HIF-2α target gene responsible for the increased tumor cell invasion observed in VHL mutant RCC cells.

Cell lines were maintained in Dulbecco's Modified Eagle's Medium (Mediatech, Inc., Herndon, VA) supplemented with 10% fetal bovine serum (FBS) (Hyclone, Logan, UT), penicillin [100 U/mL], streptomycin [100 μg/mL], and L-glutamine and cultured at 37°C, 5%CO₂. For serum-free conditions, DMEM supplemented with lactalbumin hydrosylate (2%), penicillin [100 U/mL], streptomycin [100 μg/mL], and L-glutamine was used. The WT8 and pRc-9 cell lines (kindly provided by William Kaelin, Dana-Farber Cancer Institute, Boston, MA) represent stable subclones of the 786-0 renal cell carcinoma cell line transfected with pRc/CMV-HA-VHL or pRc/CMV-empty, respectively [16]. The 786-0 cell line (American Type Culture Collection (Manassas, VA) has a single VHL allele harboring a frameshift mutation at codon 104 resulting in a truncated, non-functional protein [16, 67]. The WT8 and pRc-9 cells were cultured under continuous selection with the addition of [1 mg/mL] G418 sulfate. Cells were washed with either Hank's Balanced Salt Solution (HBSS) (Mediatech, Inc.) or 1× PBS (13.7 mM NaCl, 0.27 mM KCl, 1.2 mM phosphate buffer, pH 7.4) (National Diagnostics, Atlanta, GA) as indicated in the different experimental procedures.

The pCMV-HIF-2α expression construct containing full-length cDNA of HIF-2α was a generous gift of Richard Bruick (University of Texas, Southwestern Medical Center, Dallas, TX) [68]. The MT1-MMP expression construct, MTpc3SE, was a kind gift of Jouko Lohi (University of Helsinki, Helsinki, Finland), and contains full-length MT1-MMP cDNA downstream of a CMV promoter [69]. The pCMV-eGFP expression construct is commercially available from BD Biosciences (San Jose, CA).

WT8 cells were plated in 6-well dishes at a density of 1.5 × 10⁵ cells/well in DMEM+10% FBS in the absence of antibiotics or selection. At this density, the cells were ~90% confluent the following day and ready for transfection. Cells were then transiently transfected with 1 μg of DNA using Lipofectamine 2000™ Transfection Reagent (Invitrogen, Carlsbad, CA) following manufacturer's instructions for at least 12 hours before being washed three times with HBSS and switched to serum-free conditions for additional time depending on the experiment. Transfection with an empty vector (pRc/CMV) was used as a control for mock transfected cells [23]. All transfections were performed in triplicate. Separate plates of cells were transfected with pCMV-eGFP to control for transfection efficiency under the given conditions. Transfection efficiency was determined as a percentage of GFP-expressing cells counted from 3 fields of 3 separate transfections, and the average transfection efficiency was approximately 50% in all experiments. Cells were harvested for either 1) functional assays as described under 'Collagen invasion assays' or 'Collagen degradation assay', 2) total RNA using the RNeasy kit as described under 'Real-time RT-PCR', or 3) protein expression analyses as described under 'ELISA' or 'Immunoblotting'.

Three siRNA duplex oligoribonucleotides targeting the MT1-MMP coding region (NM_604995) were designed using the Block-iT™ RNAi Designer program from the Invitrogen website. The Stealth™ RNAi Negative Control Duplex (cat# 12935-300; Invitrogen, Carlsbad, CA) is a proprietary non-targeting sequence with medium G/C content, which is similar to the G/C content of the target Stealth™ siRNAs. The target siRNA sequences are as follows: siRNA(1): 5'-AAUUUGCCAUCCUUCCUCUCGUAGG-3'; siRNA(2): 5'-AAGAGAGCAGCAUCAAUCUUGUCGG-3'; siRNA(3): 5'-AAUGAUGAUCACCUCCGUCUCCUCC-3'. WT8 cells were plated in 6-well dishes at a density of 1.5 × 10⁵ cells/well as described under 'Transient transfections'. The following day, cells were transfected with 0.5 μg of pCMV-eGFP, 1 μg MTpc3SE or pCMV-HIF-2α, and 25 nM of either the control siRNA or one of the MT1-MMP target siRNAs using Lipofectamine 2000™ Transfection Reagent (Invitrogen) following manufacturer's instructions for co-transfection of plasmid DNA and siRNA oligos. Cells were transfected in the presence of serum for 24 hours before being washed three times with HBSS and switched to serum-free conditions for an additional 24 hours. Transfection efficiency was determined by GFP expression as described above before the cells were harvested for either protein expression or the collagen invasion assay as described under 'Fluoroblok™ invasion assay'. The average transfection efficiency was approximately 50% in all experiments. An empty vector control (pRc/CMV) was used to balance the amount of plasmid DNA in co-transfections [23]. All transfections were performed in triplicate.

Total cellular RNA was purified using the RNeasy kit with on-column DNase I treatment (Qiagen, Valencia, CA). Reverse transcription and real-time PCR reactions were performed using the Taqman Reverse Transcription Reagent Kit and Syber Green Master Mix, respectively, following the manufacturer's protocol and as described previously (Applied Biosystems, Foster City, CA) [23]. For each experiment, triplicate samples were obtained, and the cDNA for each was assayed in duplicate using a MJ Research DNA Engine Opticon thermal cycler. Data are presented as the average pg of MT1-MMP mRNA per ng of GFP mRNA and are representative of three or more experiments. Standard curves were included in each assay. Standards were prepared from serial log dilutions of plasmids carrying the appropriate cDNA as follows: MTpc3SE plasmid, 10 pg-0.001 pg; pCMV-eGFP plasmid, 1 ng-0.001 ng. Primer sequences are as previously published for MT1-MMP [23] and for eGFP [41].

The Matrix Metalloproteinase-14 (MMP-14) Biotrak Activity Assay System (GE Healthcare BioSciences Corp., Piscataway, NJ), a quantitative measure of MT1-MMP activity as a direct measure of MT1-MMP protein, was used to analyze MT1-MMP protein expression. The assay was conducted according to the manufacturer's protocol. A standard curve ranging from [0.125 ng/mL] to [8 ng/mL] was used to quantitate protein expression. Because the assay is colorimetric and does not require quenching, absorbance readings were taken at 6 hr, 9 hr, and 12 hrs after incubation with the substrate. The data presented represent the concentration of MT1-MMP in the samples at the timepoint at which the sample values fit the standard curve most appropriately. Total protein from the extracts was quantitated using the Bradford Assay (Bio-Rad, Hercules, CA) and used to normalize the MT1-MMP protein concentration in each sample.

Whole cell lysates were harvested from transfection experiments by washing the cells twice with cold 1× PBS, adding 100 μL of SDS reducing buffer (60 mM Tris-HCl, pH 6.8, 2% SDS, 14.4 mM β-mercaptoethanol, 25% glycerol, 0.1% bromphenol blue), and boiling for 5 minutes. Proteins were resolved by SDS-PAGE and electrotransferred to Immobilon-P PVDF membranes (Millipore Corp., Bedford, MA). Membranes were blocked with 5% milk in Tris-buffered saline 0.1% Tween-20 at room temperature for 1–2 hours. Primary antibodies, HIF-2α polyclonal antibody, 1:1000 (Novus Biologicals, Littleton, CO) and actin monoclonal antibody, 1:5000 (Oncogene, Cambridge, MA), were diluted in blocking buffer and incubated with the membranes at room temperature for 2 hours. Appropriate secondary antibodies were diluted in blocking buffer and incubated with the membrane at room temperature for 1 hour. Proteins were visualized by Western Lightning Chemiluminescence Reagent (Perkin Elmer, Boston, MA).

Transfectants were cultured in serum-free media for 24 hours, and conditioned media was concentrated approximately 30 fold using BioMax 30 K NMWL membrane ultrafree filters (Millipore Corp., Bedford, MA) as per manufacturer's instructions. An equal volume of non-reducing buffer (60 mM Tris-HCl, pH 6.8, 25% glycerol, 0.1% bromphenol blue) was added to the concentrated media before being applied to a 10% SDS-PAGE gel impregnated with [2.8 mg/mL] gelatin. After electrophoresis, the gel was washed two times for 30 min in 50 mM Tris-HCl (pH 7.5), 5 mM CaCl₂, 5 μM ZnCl₂, plus 2.5% Triton X-100 followed by an overnight incubation at 37°C in the same buffer without Triton X-100. Gels were stained with Coomassie Brilliant Blue R-250 and then destained in 20% methanol, 10% glacial acetic acid. Activity was detected as transparent bands. Recombinant human pro-MMP-2 (R&D Systems, Minneapolis, MN) was used as a control for gelatinolytic activity at [1 ng/μL]. Active MMP-2 was obtained by incubation with 20 μM 4-Aminophenylmercuric acetate (APMA, Sigma, St. Louis, MO) at 37°C for 45 min.

Cells were serum-starved for at least 8 hours before being harvested for the assay. Cells were embedded as a mixture of fibrillar collagen and media in a 12-well assay format. Collagen was prepared from purified bovine type I collagen (Cohesion Technologies, Palo Alto, CA) following manufacturer's instructions. Briefly, collagen was neutralized to pH 7.4 with the addition of 10× PBS and 0.1 N NaOH. Next, 3.2 × 10⁵ cells were mixed with 132 μL of neutralized collagen, and serum-free media was added to bring the total volume to 635 μL per well. The final concentration of collagen in the mixture was [0.5 mg/mL]. The collagen was allowed to gel for 1 hour at 37°C after which 635 μL of serum-free media was added to the top of the gel. Assays were harvested at 48 hours, and the overlying media was removed and weighed. The specific gravity of serum-free media was determined to be 1 mg/mL. Therefore, collagen degradation is reported as the volume of liberated media calculated by the difference of the weight of total media removed and weight of the original volume added (635 μg or 635 μL). This assay provides an accurate measure of collagen degradation in vitro [40, 41].

Statistical significance was calculated using the student's t-test available online [70] and are represented as +/- standard deviation (S.D.) of the mean. Significance was assigned to P values < 0.05.