Furin-independent Pathway of Membrane Type 1-Matrix Metalloproteinase Activation in Rabbit Dermal Fibroblasts*

We investigated the gene expression and intracellular activity of processing protease furin and its involvement in the process of membrane type 1-matrix metalloproteinase (MT1-MMP) activation in rabbit dermal fibroblasts. When the rabbit fibroblasts were treated with concanavalin A (ConA), pro-MMP-2 was converted to an active 62-kDa MMP-2 through the appearance of a 64-kDa intermediate MMP-2. The ConA-induced pro-MMP-2 activation resulted from increasing the gene expression and production of MT1-MMP in the rabbit fibroblasts. Reverse transcriptase-polymerase chain reaction demonstrated that in rabbit dermal fibroblasts furin mRNA was detected and, unlike MT1-MMP, was not increased by ConA. These findings are further supported by the fact that the intracellular furin activity also was constitutively detected and was unchanged by the ConA treatment. Very similar phenomena were also observed in human uterine cervical fibroblasts, which are known to produce MT1-MMP by ConA stimulation. These results suggest that the expression of the furin gene and the intracellular activity are not regulated by ConA. On the other hand, neither a synthetic furin inhibitor, decanoyl-RVKR-CH2Cl (25–100 μm) nor a furin antisense oligonucleotide (40 μm) inhibited the MT1-MMP-mediated pro-MMP-2 activation in ConA-treated rabbit dermal fibroblasts, whereas these compounds interfered with pro-MMP-2 activation in ConA-treated human uterine cervical fibroblasts. Nonetheless, the furin antisense oligonucleotide completely suppressed furin gene expression in both rabbit and human fibroblasts. These results suggest that furin does not participate in the process of MT1-MMP activation induced by ConA in rabbit dermal fibroblasts.

MT1-MMP is expressed on the cell surface as an active form in most cell culture systems (18). The reason for this is considered to be that pro-MT1-MMP is activated by an intracellular processing protease furin (9,19), which is a serine protease classified in the family of subtilisin/Kex 2p-like proprotein convertases and widely expressed in various tissue and cell species (20 -23). MT1-MMP, as well as other MT-MMPs, possesses a furin recognition sequence RRKR in the propeptide (9, 24 -26), suggesting that an autoproteolytic cleavage by furin may be programmed in eliciting MT1-MMP functions such as pro-MMP-2 activation and ECM breakdown by itself (27,28). On the other hand, Cao et al. (29) reported that MT1-MMP processing by furin is not a prerequisite for pro-MMP-2 activation in a transfection study using COS-1 cells. Therefore, it is not fully understood whether furin would substantially participate in MT1-MMP processing to activate pro-MMP-2. Furthermore, it is of interest whether furin activity and its production could be regulated together with MT1-MMP production.
Recently, rabbit MT1-MMP cDNA has been isolated from rabbit osteoclasts (30) and arterial smooth muscle cells (31). The deduced amino acid sequence has the furin recognition sequence in the propeptide, presuming that rabbit pro-MT1-MMP activation may be furin-dependent. However, there is no evidence of the regulatory mechanism of rabbit MT1-MMP activation and production to augment the processing of pro-MMP-2. In the present study we demonstrated that concanavalin A (ConA) induced the gene expression and production of MT1-MMP, and thereby pro-MMP-2 activation was augmented. Furin mRNA and its enzymic activity were constitutively expressed but not influenced by ConA in rabbit dermal fibroblasts. On the other hand, neither a furin synthetic inhibitor nor a furin antisense oligonucleotide (F-AS) inhibits the MT1-MMP-mediated pro-MMP-2 activation in the rabbit fibroblasts, suggesting that rabbit MT1-MMP activation is furin-independent.

EXPERIMENTAL PROCEDURES
Cell Culture and Treatments-Rabbit dermal fibroblasts were prepared from the back of male Japanese White rabbits by the explant culture method with some modifications (32) and cultured in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% (v/v) fetal bovine serum. Human uterine cervical fibroblasts were cultured in 10% (v/v) fetal bovine serum/minimum essential medium (33,34). The confluent cells were washed once with 0.2% (w/v) lactalbumin hydrolysate/Dulbecco's modified Eagle's medium or minimum essential medium and then treated with or without ConA (10 g/ml) (Seikagaku Kogyo, Tokyo, Japan) in the same medium for 24 h. In the experiment with synthetic furin inhibitors, the cells were pretreated with a specific furin inhibitor, decanoyl (Dec)-RVKR-CH 2 Cl, or a poor furin inhibitor, Dec-RAIR-CH 2 Cl (35), for 6 h and then treated with these inhibitors together with ConA (10 g/ml) for 18 h. In the experiment with furin antisense and sense oligonucleotides, a similar treatment was performed with a phosphorothioate-linkage human furin antisense (F-AS) or sense (F-S) oligonucleotide (Ϫ8 to 8 bp) (5Ј-AGC TCC ATG GGG GGG A-3Ј and 5Ј-TCC CCC CCA TGG AGC T-3Ј, respectively) (36). The harvested culture medium was stored at Ϫ20°C until use. The cells were subjected to the preparation of cell lysate and RNA isolation. In this series of experiments, rabbit dermal fibroblasts and human uterine cervical fibroblasts were used at the 1st to the 6th passage level and the 1st to the 11th, respectively.
Preparation of Cell Lysate and Western Blot Analysis for MT1-MMP-To monitor the production of MT1-MMP, the cells were washed with ice-cold phosphate-buffered saline (Ϫ) once and then scraped into 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM CaCl 2 , 0.05% (w/v) Briji 35 with a rubber policeman. After centrifugation at 800 ϫ g for 5 min at 4°C, the cells were resuspended in the same buffer, sonicated 6 times at 4°C for 10 s, and then centrifuged at 8,000 ϫ g for 15 min at 4°C. The supernatant was subjected to Western blot analysis for detection of MT1-MMP. The protein concentration in the samples was determined according to the method of Lowry et al. (38). An aliquot (100 g) of the precipitate was mixed with SDS-polyacrylamide gel electrophoresis sample buffer containing 2-mercaptoethanol (39) and then run on 10% (w/v) acrylamide gel. The proteins separated in the gel were electrotransferred onto a nitrocellulose membrane. The membrane was reacted with human MT1-MMP monoclonal antibody (mAb), which was then complexed with alkaline phosphatase-conjugated rabbit antimouse IgG. Immunoreactive MT1-MMP was visualized with 5-bromo-4-chloro-3-indolyl phosphate and nitro blue tetrazolium as described previously (33,34).
In Vitro Pro-MMP-2 Activation by Plasma Membranes-Plasma membranes (10 g of protein) prepared from the ConA (10 g/ml)treated rabbit fibroblasts were incubated with or without 50 g of human MT1-MMP mAb in a total volume of 50 l of 50 mM Tris-HCl (pH 7.5), 0.15 M NaCl, 10 mM CaCl 2 at 37°C for 30 min and then reacted with purified human pro-MMP-2 (30 ng) at 37°C for 24 h as described previously (40). To monitor pro-MMP-2 processing, the reaction mixture was centrifuged at 10,000 ϫ g for 10 min, and the resultant supernatant (10 l) was subjected to gelatin zymography as described above.
RNA Isolation and Northern Blot Analysis-Cytoplasmic RNA was isolated with ISOGEN (Nippon Gene Co., Toyama, Japan) according to the manufacturer's instructions. Isolated RNA (15 g) was denatured and electrophoresed on 1.0% (w/v) formaldehyde-denatured agarose gel and then transferred onto a nylon membrane (GeneScreen; NEN Life Science Products). The membrane was hybridized with a 32 P-labeled MT1-MMP or glyceraldehyde-3-phosphate dehydrogenase cDNA probe (CLONTECH Laboratories, Inc., Palo Alto, CA) as described previously (40).
Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR)-A total of 3 g of isolated RNA was subjected to first strand cDNA synthesis with Moloney-murine leukemia virus reverse transcriptase, RNase inhibitor (Roche Molecular Biochemicals), and oligo(dT) [12][13][14][15][16][17][18] primer (Life Technologies, Inc.) for 1 h at 37°C according to the manufacturer's instructions. One-tenth of the cDNA generated from the reverse transcriptase reaction was used for PCR amplification in the presence of Taq DNA polymerase (Roche Molecular Biochemicals) and oligonucleotide primers, 5Ј-CAG ATC TTC GGG GAC TAT TAC CAC-3Ј (sense) and 5Ј-CCT GTT GTC ATT CAT CTG TGT GTA CC-3Ј (antisense). The sense and antisense primers were designed corresponding to human furin cDNA (391-414 and 770 -795 bp, respectively), since furin is substantially conserved among mammalian tissues and cell lines (21,36). The amplification was performed at 92°C for 45 s, 54°C for 60 s, and 72°C for 90 s after initial denaturation at 92°C for 2 min. We also confirmed that the PCR product was lineally produced between 35 and 40 cycles. The amplified PCR product (405 bp) was analyzed with 1% (w/v) agarose gel and visualized by ethidium bromide staining. The 405-bp PCR product was subcloned into pGEM-T vector (Promega, Madison, WI), and then the cDNA sequence was confirmed with a Sequenase version 2.0 DNA sequencing kit (U. S. Biochemical Corp.) according to the manufacturer's instructions.
Intracellular Furin Activity-Intracellular furin activity was measured with a furin-specific substrate, t-butyloxycarbonyl (Boc)-RVKR-7amino-4-methylcoumarin (MCA) (41,42). Briefly, the cell lysate (20 g of protein) prepared as described above was incubated with 1.5 M Boc-RVKR-MCA in a 96-well plate for 6 h at 37°C. The reaction was terminated by adding 2 mM ZnCl 2 , and then the fluorescence was measured by excitation at 365 nm and emission at 450 nm. The intracellular furin activity was calculated from a standard curve made with authentic MCA (0.125-2 M). The specificity of the furin activity measured was estimated by comparing it with that of a poor furin substrate, Boc-LKR-MCA (42).
Statistical Analysis-Data were analyzed by Student's t test; p Ͻ 0.05 was considered to be statistically significant.

ConA-induced Pro-MMP-2 Activation and MT1-MMP Production in Rabbit Dermal
Fibroblasts-When rabbit dermal fibroblasts were treated with ConA (10 g/ml), a spontaneously produced 72-kDa pro-MMP-2 was converted to a 62-kDa active MMP-2 with the appearance of a 64-kDa intermediate form (Fig. 1A). The production of MT1-MMP was also augmented by ConA, whereas the enzyme was not detected in untreated rabbit fibroblasts (Fig. 1B). The molecular weight of MT1-MMP detected in rabbit dermal fibroblasts was the same as that of human MT1-MMP expressed in ConA-treated human fibrosarcoma HT-1080 cells and normal human fibroblasts (data not shown). In addition, the augmented MT1-MMP production resulted from increasing its steady-state mRNA expression in ConA-treated rabbit fibroblasts (Fig. 1C). These results suggest that ConA increases pro-MMP-2 activation by augmenting the production of MT1-MMP in rabbit dermal fibroblasts and further that rabbit MT1-MMP may be expressed as an active form on the cell surface.
Rabbit MT1-MMP Can Activate Pro-MMP-2 in Vitro-To clarify whether the immunoreactive rabbit MT1-MMP was able to activate pro-MMP-2, an in vitro pro-MMP-2 activation experiment was performed with the membrane fraction prepared from the ConA-treated rabbit dermal fibroblasts. As shown in  Regulation of Furin mRNA Expression and Intracellular Furin Activity-Since MT1-MMP contains a furin recognition sequence RRKR in the propeptide (9,19), we investigated whether the gene expression of furin was modified by ConA together with MT1-MMP production. The semiquantitative RT-PCR analysis showed that furin mRNA was constitutively expressed in rabbit dermal fibroblasts, and the transcript level was barely changed by ConA (10 g/ml) (Fig. 3A). Similar results were observed in human uterine cervical fibroblasts, which are characterized as MT1-MMP-producing cells when treated with ConA (40) (Fig. 3A). Furthermore, sequence analysis revealed that the amplified 405-bp cDNA in both cells was exactly the same as the human furin sequence (36) (data not shown). These results suggest that rabbit furin is at least partially identical to human furin and that the gene expression is not influenced by ConA.
To support further the fact that furin expression is not influenced by ConA in rabbit or human fibroblasts, we measured intracellular furin activity with a furin-specific synthetic substrate, Boc-RVRR-MCA (41,42). As shown in Fig. 3B, furin activity was constitutively detected in both rabbit dermal and human uterine cervical fibroblasts, and the enzymic activity in the rabbit fibroblasts was relatively weaker than that in the human fibroblasts. In addition, like furin gene expression, the intracellular furin activity was almost unchanged by ConA (10 g/ml) in both types of fibroblasts, and furin activity could no longer be detected with a nonspecific furin substrate, Boc-LKR-MCA (42) (data not shown). Furthermore, we confirmed that the furin activities in both rabbit and human fibroblasts were inhibited by high concentrations of diisopropyl fluorophosphate (10 -20 mM), EDTA (5 mM), and dithiothreitol (1 mM) as described previously (41-43) (data not shown), indicating that the enzymic activity detected was substantially that of furin. These results therefore suggest that furin is constitutively produced in MT1-MMP-producing cell lines, and its production is not regulated by ConA.
Synthetic Furin Inhibitor and Furin Antisense Oligonucleotide Do Not Inhibit MT1-MMP-mediated Pro-MMP-2 Activation in Rabbit Dermal Fibroblasts-We demonstrated that the activation of pro-MMP-2 was augmented by MT1-MMP in ConA-treated rabbit dermal fibroblasts (see Fig. 2) and human uterine cervical fibroblasts (40). In addition, since pro-MMP-2 is one of the substrates for MT1-MMP, the monitoring of pro-MMP-2 activation is adequate to interpret whether MT1-MMP is intracellularly activated and expressed as an active form in the cultured cells. Therefore, to clarify the involvement of furin in rabbit pro-MT1-MMP activation, we investigated the effect of a furin inhibitor and an antisense oligonucleotide on the activation of pro-MMP-2. As shown in Fig. 4A, the ConA-induced pro-MMP-2 activation was not inhibited, even when rabbit dermal fibroblasts were pretreated with a furin synthetic inhibitor, Dec-RVKR-CH 2 Cl, followed by the ConA (10 g/ml) stimulation (lanes 3-5). In contrast, Dec-RVKR-CH 2 Cl inhibited the augmentation of pro-MMP-2 activation in the ConA-treated human uterine cervical fibroblasts in a dose-dependent manner (25-100 M) (Fig. 4B, lanes 3-5). Similar experiments were performed with a poor furin inhibitor, Dec-RAIR-CH 2 Cl (100 M), the inhibitory activity of which is 100fold less than the specific inhibitor (35), and there was no change in the ConA-induced pro-MMP-2 activation in the human fibroblasts (Fig. 4B, lane 6). Furthermore, another inhibitory experiment with F-AS and F-S showed that the ConAinduced pro-MMP-2 activation in rabbit dermal fibroblasts was not inhibited by F-AS (40 M) or F-S (Fig. 5A, lanes 3 and 4). In human uterine cervical fibroblasts, however, F-AS (40 M) but not F-S (40 M) interfered with the ConA-induced pro-MMP-2 activation (Fig. 5B, lanes 3 and 4). To confirm further whether F-AS could specifically suppress the production of furin, we monitored the level of furin mRNA in the F-AS-and F-Streated cells. As shown in Fig. 6A, the furin mRNA expression was completely abolished in both rabbit and human F-AStreated fibroblasts (lane 1), but F-S did not influence the gene

FIG. 3. Furin mRNA expression and its intracellular activity in rabbit dermal fibroblasts and human uterine cervical fibroblasts.
A, RT-PCR for furin. Confluent rabbit dermal fibroblasts (RDF) at the 6th passage and human uterine cervical fibroblasts (HUCF) at the 9th passage were treated with or without ConA (10 g/ml) for 24 h. Isolated total RNA (3 g) was subjected to RT-PCR at the 38 cycles as described under "Experimental Procedures." The PCR products were electrophoresed and then visualized by ethidium bromide staining. The three independent experiments were reproducible, and typical data are shown. B, furin activity. Confluent rabbit dermal fibroblasts at the 5th passage and human uterine cervical fibroblasts at the 8th passage were similarly treated with ConA (10 g/ml) for 24 h, and then the furin activity in the cellular fraction (20 g protein) was assayed with a specific furin substrate at 37°C for 6 h as described under "Experimental Procedures." Fluorescence was determined by excitation at 365 and emission at 450 nm. The data are indicated as the mean Ϯ S.D. for triplicate wells. expression (lane 2). In addition, there was no difference in the level of rRNA stained with ethidium bromide in either F-AS-or F-S-treated cells (Fig. 6B), indicating that F-AS specifically suppressed the expression of furin mRNA in both types of fibroblasts. These results therefore suggest that furin is substantially involved in MT1-MMP processing to activate pro-MMP-2 in human uterine cervical fibroblasts but not in rabbit dermal fibroblasts. DISCUSSION MT1-MMP production and pro-MMP-2 activation are accelerated by ConA in normal human fibroblasts and some human tumor cells (44 -46). Okada et al. (11) also reported that ConA stimulates rat fibroblasts to augment the production of MT1-MMP and the activation of pro-MMP-2. In the present study, we confirmed that ConA augmented the gene expression and production of MT1-MMP and the subsequent pro-MMP-2 activation in rabbit dermal fibroblasts (Fig. 1). It is therefore likely that MT1-MMP production and pro-MMP-2 activation increased by ConA are commonly observed in various cell species and origins.
Human MT1-MMP mAb and cDNA recognized rabbit MT1-MMP protein and mRNA, respectively (Fig. 1). It is therefore of interest whether the rabbit MT1-MMP recognized by human MT1-MMP mAb would functionally participate in pro-MMP-2 activation. From this point of view, we demonstrated that in vitro pro-MMP-2 activation induced by plasma membranes from ConA-treated rabbit dermal fibroblasts was effectively inhibited by exogenously adding human MT1-MMP mAb (Fig.  2). A similar inhibitory effect of MT1-MMP mAb was observed in the case of the membrane fraction from the ConA-treated human uterine cervical fibroblasts (40). On the other hand, rabbit MT1-MMP cDNA was recently characterized in rabbit osteoclasts (30) and arterial smooth muscle cells (31), and the deduced amino acid sequence shows 96% identity to human MT1-MMP, but only 57, 52, and 38% identity to human MT2-MMP, MT3-MMP, and MT4-MMP, respectively. Taken together, these results strongly suggest that rabbit MT1-MMP is functionally identical to human MT1-MMP in its pro-MMP-2 activation.
Since the propeptide of pro-MT1-MMP contains RRKR which is a recognition sequence of the intracellular processing enzyme furin, it is considered that the activation of proMT1-MMP by furin is automatically accomplished in the trans-Golgi network (9,19). Our results showed that furin mRNA and its enzymic activity were detected in both rabbit dermal and human uterine cervical fibroblasts (Fig. 3). Furthermore, neither furin mRNA expression nor its intracellular activity was altered by ConA in both fibroblast types (Fig. 3). Santavicca et al. (47) reported that the gene expression of furin in human diploid HFL1 fibroblasts is not regulated by phorbol ester, which is an inducer of MT1-MMP production in human fibroblasts and human fibrosarcoma HT-1080 cells (48,49). Taken together, these results suggest that, unlike MT1-MMP production, the furin gene expression and intracellular activity cannot be substantially regulated by stimuli such as ConA and phorbol ester.
Mouse and human furin cDNAs have been cloned, and the homologies of their cDNA and amino acid sequences are 96 and 94%, respectively (21,36). We demonstrated that furin cDNA (405 bp) amplified in rabbit dermal fibroblasts was the same as the cDNA sequence of human furin (data not shown). These results strongly suggest that the furin gene might be conserved over cell species. Furthermore, this is the first evidence that rabbit furin cDNA is identical to human furin, at least partially.
It has been reported that a specific furin inhibitor, Dec-RVKR-CH 2 Cl, inhibits the activation of human pro-MMP-11 (47) and pro-MT1-MMP (50). We demonstrated that the ConAinduced pro-MMP-2 activation in human uterine cervical fibroblasts was inhibited by Dec-RVKR-CH 2 Cl but not by the nonspecific furin inhibitor Dec-RAIR-CH 2 Cl (Fig. 4B). In addition, F-AS but not F-S inhibited the ConA-induced pro-MMP-2 activation in the human fibroblasts (Fig. 5B). Our results coincided with previous reports indicating that human MT1-MMP processing is mediated by furin. In contrast, neither the specific furin inhibitor nor F-AS inhibited the MT1-MMP-mediated pro-MMP-2 activation in ConA-treated rabbit dermal fibroblasts (Figs. 4A and 5A) regardless of whether a furin recognition sequence RRKR was contained in the propeptide of rabbit MT1-MMP (30,31). We further demonstrated that F-AS specifically and completely suppressed the expression of furin mRNA in both rabbit and human fibroblasts (Fig. 6). These results strongly suggest that furin is not required for the intracellular processing of MT1-MMP in rabbit dermal fibroblasts, and therefore the involvement of furin in MT1-MMP activation may be dependent on the animal species.
Furin belongs to the family of subtilisin/Kex 2p-like proprotein convertases, and six distinct convertases (PC2, PC1/PC3, PACE4, PC4, PC5/PC6, and LPC/PC7/PC8/SPC7) in this family have been identified in mammalian species. It has also been reported that furin, PACE4, PC/5PC6, and LPC/PC7/PC8/ SPC7 are expressed in a broad range of tissues and cell lines (for review, see Ref. 51). Furthermore, the expression patterns of these convertases are different in some physiological conditions such as embryogenesis and gestation (52)(53)(54). In the present study, our finding that furin is not required for MT1-MMP activation in rabbit dermal fibroblasts regardless of the detectable enzymic activity suggests that other subtilisin/Kex 2p-like convertases may be involved in rabbit MT1-MMP activation. This hypothesis can be supported by previous studies. Other convertases such as PACE4, PC5/PC6, and LPC/PC7/ PC8/SPC7 have been shown to possess enzymic properties similar to those of furin (51). Santavicca et al. (47) also reported that the activation of human pro-MMP-11 is mediated specifically by furin, but mouse pro-MMP-11 activation is accomplished by both furin and PACE4.
In conclusion, we demonstrated for the first time that in both rabbit and human fibroblasts ConA increased MT1-MMP production and thereby augmented pro-MMP-2 activation, whereas the furin gene and its enzymic activity was constitutively expressed and not influenced by ConA. Furthermore, the increased pro-MMP-2 activation was specifically inhibited by the synthetic furin inhibitor and F-AS in the ConA-treated human fibroblasts but not in rabbit fibroblasts. These results suggest a cell type-specific involvement of furin in the process of MT1-MMP activation.