Expression of three membrane-type matrix metalloproteinases (MT-MMPs) in rat vascular smooth muscle cells and characterization of MT3-MMPs with and without transmembrane domain.

Matrix metalloproteinases (MMPs) produced by rat smooth muscle cells (SMCs) were investigated. SMCs expressed three kinds of membrane-type MMP, MT1-MMP, MT2-MMP, and MT3-MMP, and the MT-MMP expression was stimulated by the presence of serum. MT3-MMP was characterized further by cloning its cDNA. A rat MT3-MMP cDNA encoding 607 amino acids and a cDNA for its transmembrane domainless variant MT3-MMP-del were cloned from a rat SMC cDNA library; a human MT3-MMP cDNA was cloned from a fetal brain cDNA library. Human brain MT3-MMP was similar but not identical to the previously reported human placenta MT3-MMP (94.4% homology). When the MT3-MMP cDNA was expressed in COS-7 cells, endogenous progelatinase A was processed to the mature form. The transfection of rat MT3-MMP-del efficiently converted progelatinase A to the intermediate form but not to the mature one, indicating that the transmembrane domain is important for the complete processing of progelatinase A to maturation. Both MT3-MMP-del and MT3-MMP hydrolyzed gelatin and casein, indicating their broad substrate specificity. Results of experiments with a synthetic MMP inhibitor suggested that MT3-MMP-del and MT3-MMP are rapidly degraded immediately after maturation. The present study suggests that multiple forms of MMPs including MT3-MMP are involved in the matrix remodeling of blood vessels.

connective tissues as physical barriers. Among the MMP family, gelatinase A (MMP-2) and gelatinase B (MMP-9) are critical in the invasion of tumor cells and other cells into basement membranes because of their strong activity against type IV collagen, the major component of basement membranes (4 -7). Most MMPs are secreted in a latent form (pro-MMP) and activated by serine proteinases or some activated MMPs. The activities of activated MMPs are regulated by natural inhibitors called tissue inhibitors of metalloproteinases (TIMPs) (1)(2)(3).
In tumor tissues, gelatinase A is often present on tumor cell surfaces (8,9), although its mRNA is mainly expressed by surrounding stromal cells (10 -12). Many studies have shown that progelatinase A is activated by a metalloproteinase bound to cell membrane (13)(14)(15)(16)(17). Using reverse transcription-polymerase chain reaction (RT-PCR), Sato and co-workers (18,19) recently identified a novel membrane-type MMP, named MT-MMP (MT1-MMP), which is responsible for the activation of progelatinase A on the cell surface. MT1-MMP mRNA is expressed at a high level in various cancer tissues compared with corresponding normal tissues or benign tumors (18,20). Therefore, MT1-MMP is believed to play a key role in the spatially regulated proteolysis by invasive tumor cells. Very recently, three other MT-MMPs, MT2-MMP (21), MT3-MMP (22), and MT4-MMP (23), have been identified. However, at present, these new MT-MMPs are relatively poorly characterized compared with MT1-MMP.
Like invasive tumor cells, vascular smooth muscle cells (SMCs) are known to migrate and proliferate under certain pathological conditions such as intimal hyperplasia after arterial injury (24), vascular grafting (25), and atherosclerosis (26). These processes are presumed to require partial degradation of the vascular basement membrane and extracellular matrix surrounding the cells (27,28). However, less is known about MMP species involved in these processes and their roles. Some previous studies have shown that SMCs produce interstitial collagenase (29,30), gelatinases A and B (31,32), and stromelysin (32). In the present study, we analyzed MMP, especially MT-MMP mRNAs expressed in rat vascular SMCs and normal tissues, using the RT-PCR method and Northern analysis. In addition, the cDNA for two forms of MT3-MMPs with and without the transmembrane domain were cloned, and their enzymatic functions were investigated.

EXPERIMENTAL PROCEDURES
Materials-cDNA libraries constructed from human placenta and fetal brain and plasmid vector pBluescript SK(ϩ) were purchased from Stratagene (La Jolla, CA). [␣-32 P]dCTP was from Amersham (Backinghamshire, United Kingdom). Rat tissue and human fetal tissue multiple Northern blots were from Clontech Laboratories (Palo Alto, CA).
Mammalian expression vector pGM, which had been constructed from pCDL-SR␣296 (33), was a kind gift from N. Ohkura (Terumo Research and Development Center, Kanagawa, Japan). Enzymes for DNA digestion and modification were purchased from Takara Shuzo (Shiga, Japan) and Toyobo (Osaka, Japan). A synthetic hydroxamic acid inhibitor for MMPs, KB8301, was a generous gift from Dr. K. Yoshino (Kanebo Institute for Cancer Research, Osaka, Japan).
Cell Culture-Vascular SMCs were isolated from rat thoracic aorta by the explant method of Chamley-Cambell (34). SMCs were grown in Dulbecco's modified Eagle's medium (Nissui, Tokyo, Japan) supplemented with 10% fetal calf serum, 2 mM glutamine, and 30 g/ml gentamycin under humidified 5% CO 2 conditions. SMCs were repeatedly subcultured by trypsinization and used for experiments between the 5th and 20th passages. COS-7 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and 2 mM glutamine.
RT-PCR of MMP cDNAs-Total RNA was extracted from SMCs by the acid-guanidinium phenol chloroform method (35) and reverse transcribed in the presence of random hexanucleotides. Synthesized cDNA was amplified by PCR. A degenerated primer set for PCR was designed from two highly conserved sequences of known MMPs, the cysteine switch (PRCGVPD) and the Zn 2ϩ binding site (AAHELGH). Taking account of the codon usage in rat (36), two 20-mer oligonucleotides, CCI C/AGI TGT GGI GTI CCT/A GA for the sense primer (cysteine switch) and TG ICC IAG CTC ATG IGC AGC for the antisense primer (Zn 2ϩ binding site) were synthesized. PCR was performed with 35 cycles of heat denaturation at 94°C for 1-2 min, annealing at 50°C for 2 min, and polymerization at 72°C for 3 min. Resultant amplified DNA fragments were electrophoretically separated and cloned into a pUC118 plasmid vector (Takara Shuzo).
cDNA Library Construction-Poly(A) ϩ RNA was isolated from total RNA of rat SMCs using oligo(dT)-coupled latex beads (OligoTex dT; Takara Shuzo). A cDNA library that contained 3.2 ϫ 10 6 recombinant clones was constructed from 5 g of the rat SMC poly(A) ϩ RNA by the method of Gubler and Hoffman (37), using a ZAP-cDNA cloning kit (Stratagene).
Screening and Isolation of cDNA Clones-cDNA libraries were screened by the standard plaque hybridization method. Digoxigeninlabeled probes were prepared by PCR using digoxigenin-11-dUTP (Boehringer Mannheim, Germany), heat denatured, and used for hybridization. The hybridized probe molecules on the transferred membranes were visualized by the Dig-ELISA (enzyme linked-immunosorbent assay) non-radioisotopic nucleic acid detection system (Boehringer Mannheim) with an anti-digoxigenin antibody conjugated with alkaline phosphatase. Alternatively, a 32 P-labeled probe was used to screen a cDNA library from human fetal brain with the sequence of rat MT3-MMP. From the isolated clones, insert cDNAs were subcloned to an appropriate plasmid vector.
DNA Sequencing-DNA sequences were determined by the nonradioisotopic dideoxynucleotide chain termination method with a LI-COR model 4000L DNA sequencer (Lincoln, NE). Determined DNA sequences were analyzed by DNASIS software (Hitachi Software Engineering, Kanagawa, Japan) and Genetyx Mac software (Software Development Co. Ltd., Tokyo).
Northern Blot Analysis-Total RNAs were electrophoresed on 1% agarose-formaldehyde gels and transferred onto nylon membranes by capillary elution. The nylon membranes were hybridized with 32 Plabeled probes and washed by the standard method. The hybridized signals were visualized by autoradiography or a bio-imaging analyzer BAS-2000II (Fuji Film, Tokyo).
Transfection of Cloned cDNAs into Mammalian Cells-Both fulllength cDNAs of human MT3-MMP and rat MT3-MMP-del, a transmembrane domainless variant of MT3-MMP, were inserted into a mammalian expression vector pGM for transient expression. The recombinant plasmids were transfected into COS-7 cells by the standard lipofection method. To obtain serum-free conditioned medium (CM), the transfected cells were incubated in serum-free medium for 2 days. The resultant CM was collected, dialyzed against H 2 O, lyophilized, and dissolved in a small volume of 10 mM Tris-HCl (pH 7.5) buffer containing 0.05% Brij-35. To analyze intracellular enzymes, the transfected cells were directly dissolved in sodium dodecyl sulfate-sample buffer and used.

Screening of MMP mRNAs Expressed in Rat SMCs by RT-
PCR-MMP mRNAs expressed in cultured rat SMCs were screened by RT-PCR with a degenerated primer set, which had been designed from the two highly conserved sequences of known MMPs, the Zn 2ϩ binding active site and the regulatory site cysteine switch. Total RNA purified from SMCs was reverse transcribed, and the products were subjected to PCR. Agarose gel electrophoresis of the RT-PCR products reproducibly detected amplified DNA fragments of 750, 520, 450, and 400 bp (Fig. 1). Each DNA fragment was isolated, cloned into a plasmid vector, and sequenced. The sequence analysis showed that the amplified fragments of 750 and 520 bp were nonspecific amplification products. The 400-bp fragment contained two different sequences known as rat stromelysin 1 (39) and rat collagenase, the latter of which had recently been identified as the homolog of human collagenase 3 (40) (data not shown). On the other hand, three similar but distinct sequences were identified from the electrophoretically single 450-bp fragment. One of the sequences was identical to that of rat membrane type 1 MMP (MT1-MMP) reported by Okada et al. (20). The others were highly homologous to human MT2-MMP and MT3-MMP, respectively, which had been reported recently by different groups (21,22). Thus the two sequences were regarded as rat homologs of MT2-MMP and MT3-MMP. These results indicate that rat SMC expresses three types of MT-MMPs simultaneously.
When the CM of cultured SMCs was analyzed by gelatin zymography, a high level of progelatinase A and a low level of its activated form were detected (data not shown).
Molecular Cloning of MT3-MMP cDNAs-To date, no normal cells producing MT3-MMP have been reported, and this enzyme has been characterized poorly compared with MT1-MMP. Therefore, we attempted to clone rat MT3-MMP cDNA. A rat SMC cDNA library was screened with the 450-bp MT3-MMP cDNA fragment as a probe, and three positively hybridized clones were isolated. These clones contained an insert cDNA of approximately 3.5 kilobase pairs with a typical poly(A) ϩ tail and a polyadenylation signal. Therefore, the longest clone, originally identified RT-PCR product of 450 bp. The MT3-MMP cDNA encodes a protein of 607 amino acids, and its molecular weight is calculated to be 69,622. The deduced amino acid sequence of the MT3-MMP contains a carboxyl-terminal hydrophobic stretch reported as a transmembrane sequence of MT3-MMP (see Fig. 3).
In addition to the MT3-MMP cDNA clone, another MT3-MMP cDNA was cloned from the same cDNA library of rat SMC. This cDNA differed from the above cDNA in two nucleotides at positions 2070 and 2071 as follows: CCA CCA TGA in the second cDNA and CCA GAT GA in the first cDNA. These nucleotide addition and replacement in the second clone produce the termination signal TGA just before the transmembrane domain, and hence the cDNA encodes a putative transmembrane domainless MT3-MMP variant, MT3-MMP-del, which is shorter than the normal MT3-MMP by 60 amino acids (see Fig. 3). When the presence of the two sequences in cDNAs from cultured SMC and the testis and brain of normal rats was examined by RT-PCR, only the MT3-MMP sequence with the putative transmembrane domain was obtained, suggesting that the MT3-MMP-del was either a very minor variant or an artificial product resulting from the misreading of reverse transcriptase in the construction of the cDNA library. When the rat SMC cDNA library was screened with the rat MT3-MMP probe, three weakly hybridized clones were also obtained. Analysis of their partial nucleotide sequences indicated that they corresponded to rat MT1-MMP. One of the clones, termed pratMT1-MMP, was used as a probe for Northern blotting analysis of rat MT1-MMP.
To examine the presence of MT3-MMP-del in humans, we also attempted to clone human MT3-MMP. Because Northern blotting analysis showed that MT3-MMP mRNA was most predominantly expressed in human fetal brain, the cDNA library from the tissue was screened with the rat cDNA as the probe. From eight positively hybridized clones, a cDNA of 2,052 bp containing a whole open reading frame was obtained. As shown in Fig. 3, this open reading frame encodes 607 amino acids, in accordance with rat MT3-MMP, and the predicted amino acid sequence contains the putative transmembrane domain and is 98.0% homologous to that of rat MT3-MMP. All of the seven other clones had the same sequence. The predicted primary structure of the human brain MT3-MMP is similar but not identical in both amino acid sequence and number of total amino acids to that of the MT3-MMP cloned from human placenta by Takino  (94.4% homology). The differences are localized in two regions (residues 273-289 and 501-527). The amino acid sequences in these regions of human brain MT3-MMP are identical to those of rat MT3-MMP (Fig. 2). The homology of the human brain MT3-MMP to other human MMPs is as follows: MT1-MMP (54.3%), gelatinase A (43.3%) (6), gelatinase B (41.6%), stromelysin (39.7%) (41), stromelysin 3 (36.0%) (42), and matrilysin (39.4%) (43).
Using the same method as for MT3-MMP, we also cloned a cDNA for MT2-MMP from a human placental cDNA library. The sequence of human MT2-MMP cDNA was identical to that of MT2-MMP reported by Will and Hinzmann (21).
Expression of MT-MMP mRNAs in Rat SMCs and Tissues-Expression of three MT-MMPs in rat SMCs and rat tissues was examined by Northern blotting analysis. Fig. 3 shows the expression of three MT-MMP mRNAs in SMCs cultured in the presence or absence of 10% fetal calf serum. The three MT-MMP genes were expressed predominantly in SMCs in the presence of serum, indicating that their expression was upregulated by serum factors. Expression of three MT-MMP mRNAs in eight tissues from normal rats is shown in Fig. 4. MT3-MMP mRNA was strongly detectable in the lung and brain, weakly detectable in the spleen and liver, but undetectable in the heart, skeletal muscle, and kidney. In the testis, a smaller sized transcript was observed. This seemed to be the product of the MT3-MMP gene that had been alternatively spliced or that suffered some other modification. It should be noted that there is a big difference in the size of mRNA between rat MT3-MMP (3.5 kilobases) and human MT3-MMP (12 kilobases) (22), although their protein products contain the same number of amino acid residues. MT2-MMP mRNA was most abundant in the lung and detectable in the brain, liver, skeletal muscle, and kidney but undetectable in the heart, spleen and testis. MT1-MMP mRNA was most abundant in the lung and liver and detectable in the other tissues tested except for the heart.
Properties of MT3-MMP and MT3-MMP-del-To verify enzymatic activity, the cDNAs for human brain-derived MT3-MMP and rat SMC-derived MT3-MMP-del were individually transfected into COS-7 cells by the lipofection method. As shown in Fig. 5A These results indicate that MT3-MMP is able to process progelatinase A regardless of the presence or absence of the transmembrane sequence, but the sequence is important for the complete activation of progelatinase A. On the other hand, activation of progelatinase B (92 kDa) was not seen in any transfectants (Fig. 5A).
The gelatin zymography of the CM of the MT3-MMP-del transfectants specifically showed faint and broad gelatinolytic activity at 45-50 kDa in addition to the activities due to gelatinases A and B (Fig. 5A). The broad activity appeared to consist of a major 45-kDa and a minor 50-kDa band and was more prominent in the casein zymography (Fig. 5B). The proteolytic activity of 45-50 kDa was Ca 2ϩ -dependent, and its electrophoretic mobility was hardly affected by the treatment with p-aminophenylmercuric acetate (data not shown), indicating that it has already been activated or can not be activated by p-aminophenylmercuric acetate. The molecular size (45-50 kDa) of the unidentified enzyme is compatible with the predicted size of MT3-MMP-del. In other experiments, recombinant rat MT3-MMP-del expressed in Escherichia coli also degraded gelatin and casein in zymography and was resistant to activation by p-aminophenylmercuric acetate (data not shown). Taken together, we considered that the activity of 45-50 kDa was due to the secreted MT3-MMP-del expressed transiently in the cultured cells.
Rapid Turnover of Two Forms of MT3-MMPs-To investigate autolytic processing of the MT3-MMPs with and without the transmembrane domain, a synthetic inhibitor of MMPs, KB8301, was used. When 10 M KB8301 was added to the culture of human MT3-MMP and rat MT3-MMP-del transfectants, the activation of progelatinase A was blocked completely in both transfectants (Fig. 6A). In addition, the treatment with KB8301 markedly increased the gelatinolytic activity of 45-50 kDa in the CM of MT3-TM-del transfectants, suggesting that KB8301 inhibited the autolytic degradation of the secreted MT3-MMP-del. When the cell lysates of the transfectants were analyzed by casein zymography, a caseinolytic activity of 62 kDa was detectable in the MT3-MMP transfectants (Fig. 6B). The treatment of the MT3-MMP transfectants with KB8301 produced an additional caseinolytic activity of 55 kDa. The 62and 55-kDa activities seemed to correspond to the proform and the mature form of membrane-bound MT3-MMP, respectively. The cell lysate of the MT3-MMP-del transfectants showed a major 55-kDa activity and a minor 50-kDa activity regardless of the KB8301 treatment. It seemed likely that the 55-, 50-, and 45-kDa bands in the MT3-MMP-del transfectants corresponded to the proform, intermediate form, and the mature form of MT3-MMP-del, respectively. These results strongly suggest that both MT3-MMP and MT3-MMP-del are degraded rapidly after intracellular activation. DISCUSSION The present study showed that rat SMCs express at least three kinds of MT-MMPs, MT1-MMP, MT2-MMP, and MT3-MMP, in addition to gelatinase A, collagenase 3, and stromelysin 1. We cloned cDNAs for rat MT3-MMP, its transmembrane domainless variant MT3-MMP-del, and human MT3-MMP. Human MT3-MMP cDNA was cloned previously by Takino et al. (22) from a human placental cDNA library as a membrane-bound activator of progelatinase A. Human brain MT3-MMP in this study was similar to the previously reported MT3-MMP but differed in 33 amino acid residues including the insertion of 3 amino acids (94.4% homology). These differences between human brain MT3-MMP and human placenta MT3-MMP were completely conserved between rat SMC and human brain MT3-MMPs. In addition, the sequences of eight positively hybridized clones obtained from the fetal brain cDNA library matched with the brain MT3-MMP but not with the reported placenta MT3-MMP. Therefore, it is concluded that the MT3-MMP that is expressed mainly in the brain and SMCs has the structure determined in this study. At present it is not clear whether the structural differences between the two human MT3-MMPs are the result of some experimental error, gene multiplicity, or microheterogeneity of a single gene.
Similarly, little is known about rat MT3-MMP-del, which lacks the transmembrane domain. Such MT-MMP variants have not been reported elsewhere. Despite repeated cDNA cloning, we could not obtain this variant cDNA. Therefore, MT3-MMP-del seems to be a very minor variant of rat MT3-MMP or an artificial product resulting from the misreading of the reverse transcriptase. However, MT3-MMP-del may prove a useful tool for understanding the enzymatic activity of MT3-MMP. MT1-MMP and MT3-MMP activate progelatinase A when their cDNAs are transfected into progelatinase A-producing cells (18,22). MT2-MMP also has membrane-dependent progelatinase-A-activating activity (our unpublished data). Cao et al. (44) reported that MT1-MMP lacking the COOH-terminal transmembrane domain could neither generate the active form of gelatinase A nor hydrolyze other substrates. However, recent studies showed that recombinant MT1-MMP from which the membrane-spanning domain had been truncated was able to activate progelatinase A and to degrade gelatin, casein, and some extracellular matrix proteins such as fibronectin or vitronectin (45)(46)(47). They also showed that soluble MT1-MMP expressed in E. coli was activated by autolysis or partial digestion with trypsin but not by p-aminophenylmercuric acetate treatment. In the present study, we showed that rat MT3-MMP-del processed progelatinase A to the intermediate form even more efficiently than human brain MT3-MMP when the respective cDNA was transiently expressed in COS-7 cells. The efficient processing of progelatinase A by MT3-MMP-del seems due to the high availability of progelatinase A in culture medium. However, the complete activation of progelatinase A hardly occurred with MT3-MMP-del. The conversion of the partially activated gelatinase A to the mature form is known to involve intermolecular autolytic processing (48). Apparently the membrane-bound MT3-MMP is more suited than the soluble enzyme to this process because the intermediate form of gelatinase A can be concentrated on the cell surface in the former enzyme. In the case of MT3-MMP-del, TIMP-2 may quickly bind to the partially activated gelatinase A in solution, blocking the autolytic processing. These results suggest that the membrane binding of MT3-MMP is not essential for the initial processing of progelatinase A but important for the autolytic conversion of the intermediate form to the mature form.
The present study also showed that the treatment of the cDNA transfectants with the synthetic inhibitor KB8301 increased the amount of MT3-MMP-del in culture medium and MT3-MMP in cell membrane, suggesting that these enzymes are rapidly degraded by autolysis after maturation. This seems to be an important regulatory mechanism to prevent excess activity of MT-MMPs because these enzymes, unlike most other MMPs, are intracellularly activated by furin or a similar serine proteinase (49).
SMCs are known to migrate and proliferate actively under various physiological and pathological conditions. Zempo et al. (27) have shown that the activities of gelatinases A and B are increased in rat carotid artery after balloon catheter injury. In the present study, the expression of MT-MMPs, especially MT3-MMP, in cultured SMCs was strongly induced by the presence of serum. These facts strongly suggest that the three MT-MMPs are also involved in the migration and proliferation of SMCs under various physiological and pathological conditions. MT3-MMP-del had weak gelatinolytic and relatively high caseinolytic activities in the zymography assay. The MT3-MMP with the transmembrane domain also showed caseinolytic activity in zymography. These results suggest that MT-MMPs probably hydrolyze various extracellular matrix proteins and other functional proteins on the cell surface without depending on the activation of progelatinase A. They might play an important role in the matrix remodeling of blood vessels by activating progelatinase A and/or directly degrading matrix proteins.
We detected only a negligible amount of the active form of gelatinase A in the culture of rat SMCs, even though all three kinds of MT-MMPs were expressed. This implies that the progelatinase A processing by MT-MMPs is negatively regulated by some other factors including TIMP-2 in cultured SMCs. Northern blotting analysis showed that MT1-MMP was distributed widely to normal rat tissues, although its expression was especially high in the lung and liver. The expression of MT3-MMP gene was more restricted to specific organs such as the lung and brain than was the expression of MT1-MMP and MT2-MMP genes. This suggests some specific role of MT3-MMP in these organs.