Collagenase 2 (MMP-8) Expression in Murine Tissue-remodeling Processes

Neutrophil collagenase or collagenase 2 (MMP-8) is unique among the family of matrix metalloproteinases (MMPs) because of its exclusive pattern of expression in inflammatory conditions. At present, no evidence of the occurrence of this enzyme in tissues other than human has been reported. In this work, we have cloned the murine homologue of human collagenase 2. The isolated cDNA contains an open reading frame coding for a polypeptide of 465 amino acids, which is 74% identical to its human counterpart. The mouse collagenase 2 exhibits the domain structure characteristic of several MMPs, including a signal sequence, a prodomain with the cysteine residue essential for enzyme latency, an activation locus with the Zinc-binding site, and a COOH-terminal fragment with sequence similarity to hemopexin. It also contains the three conserved residues (Tyr-209, Asp-230, and Gly-232) located around the Zinc-binding site and are distinctive of the collagenase subfamily. Northern blot analysis of RNAs isolated from a variety of mouse tissues revealed that collagenase 2 is expressed at late stages during mouse embryogenesis, coinciding with the appearance of hematopoietic cells. In addition, collagenase 2 was highly expressed in the postpartum uterus starting at 1 day postpartum and extending up to 5 days. Enzymatic analysis revealed that matrilysin, another MMP overexpressed in uterine tissue, is able to activate murine procollagenase 2. These data suggest that both enzymes could form an activation cascade resulting in the generation of the collagenolytic activity required during the process of massive connective tissue resumption occurring in the involuting uterus.

The matrix metalloproteinases (MMPs) 1 form a group of structurally related endopeptidases that mediate the degradation of the different macromolecular components of the extracellular matrix and basement membranes (1,2). These proteolytic enzymes have been implicated in the connective tissue remodeling occurring in normal processes such as embryonic development, angiogenesis, bone growth, or wound healing (3,4). In addition, abnormal expression of these enzymes may contribute to a variety of pathological processes including atherosclerosis (5), pulmonary emphysema (6), immune complexinduced alveolitis (7), rheumatoid arthritis (8), and tumor invasion and metastasis (9). At present, the family of human MMPs is composed of 16 members that can be classified into four different subfamilies: collagenases, gelatinases, stromelysins, and membrane-type MMPs, although there are some enzymes like macrophage metalloelastase (10), stromelysin 3 (11), MMP-19 (12), and enamelysin (13) that do not belong to any of these groups.
The collagenase subfamily of human MMPs is composed of three members: fibroblast collagenase (collagenase 1 or MMP-1), neutrophil collagenase (collagenase 2 or MMP-8), and collagenase 3 (MMP-13). An additional collagenase called collagenase 4 (MMP-18) has been cloned from metamorphosing Xenopus laevis tadpoles, but no human homolog of this enzyme has been described yet (14). The collagenases share the unique ability to cleave the native helix of fibrillar collagens at a single peptide bond, generating fragments of about three-fourths and one-fourth the size of the intact molecule (15)(16)(17). However, the three human collagenases characterized to date show distinct substrate preferences toward the different fibrillar collagens. Thus, collagenase 1 preferentially cleaves type III collagen (18), and collagenase 2 is more active against type I collagen (19), whereas collagenase 3 is especially effective against type II collagen (20,21). These data have led us to propose that the three human collagenases have evolved as specialized enzymes to degrade tissues with different collagen composition, thereby performing different functional roles in the human body (20). The verification of this hypothesis as well as the precise functional characterization of these enzymes would be facilitated by the availability of animal models in which their activity could be selectively manipulated. However, and somewhat surprisingly, only the murine homologue of collagenase 3 has been characterized (22,23). In fact, all previous attempts to verify the presence of collagenases 1 and 2 in murine tissues have been unsuccessful, which has suggested that collagenase 3 may be the only interstitial collagenase occurring in murine tissues. One possibility to explain the lack of detection of the remaining collagenases in murine tissues is that their expression is highly restricted and circumscribed to very specific normal or patho-logical conditions. This may be especially the case of collagenase 2. This member of the MMP gene family was originally cloned from mRNA extracted from the peripheral leukocytes of a patient with chronic granulocytic leukemia (16,24). Further studies have shown that collagenase 2 is a 75-80-kDa glycoprotein that is synthesized as a latent proenzyme during the myelocyte stage of neutrophil development (25,26). However, in marked contrast to the other collagenases that are secreted immediately after synthesis, collagenase 2 is stored intracellularly within specific granules, being released upon chemotactic stimulation in vitro or during inflammatory conditions in vivo (27,28). Once released and activated through proteolytic or oxidative mechanisms, the enzyme appears to play a major role in the connective tissue turnover occurring in inflammatory processes (29). In addition to its degrading activity on connective tissue components, collagenase 2 also has the ability to proteolytically inactivate the ␣ 1 -proteinase inhibitor, which has suggested that this enzyme may be involved in the development of pulmonary emphysema (30,31). Finally, recent studies have shown that, besides its expression in neutrophils, human articular chondrocytes, rheumatoid synovial fibroblasts, and endothelial cells are also a source of collagenase 2 (32,33). The expression in articular cartilage is significantly enhanced in osteoarthritic patients and can be strongly upregulated in primary cartilage explants treated with interleukin-1␤ (34). These data, together with the observation that purified collagenase 2 can cleave the interglobular domain of aggrecan at the aggrecanase site (Glu 373 -Ala 374 ), have suggested that this MMP may be an important mediator of cartilage destruction in arthritic diseases (35).
Because of the relevance of collagenase 2 in both normal and pathological processes, we have examined the possibility that a putative homologue of this enzyme is produced by mouse tissues. In this work, we describe the molecular cloning and nucleotide sequence of a cDNA coding for mouse collagenase 2. We also show that, in addition to its expected secretion from polymorphonuclear cells during inflammatory processes induced in the mouse, it is strongly expressed in the postpartum uterus where it may play an essential role in the degradation of the collagen fibers during uterine involution.
Isolation of a Mouse Collagenase 2 Probe-The WI/MIT mouse genomic YAC library from the Human Genome Mapping Resource Center (Cambridgeshire, UK) was screened by PCR searching for clones containing the mouse collagenase 3 gene as well as other MMP genes presumably clustered in the same region of the genome. After amplification with two specific oligonucleotides, 5Ј-ATGCATTCAGCTATCCT-GGCCACCTTC and 5Ј-AGCCTGTCAACTGTGGAGGTCACT, derived from the murine collagenase 3 sequence, a positive YAC clone (I139A1) was isolated. DNA from this clone was then used as a template for a PCR amplification experiment with degenerate oligonucleotides 5Ј-CC(A/C/T)(A/C)GNTG(C/T)GGNGT(C/G/T)CC and 5Ј-TCNGANAC(T/ C)TTNGT(A/G)AANGT as primers. A number of fragments were amplified, cloned in pUC18, and sequenced. One of the clones contained a 870-bp sequence with the highest degree of similarity to human collagenase 2, encompassing 84 bp from exon 2, 675 bp of intron 2, and 110 bp of exon 3. This fragment was used as a probe for screening a mouse 129/SvJ library (Stratagene, La Jolla, CA) and for obtaining specific oligonucleotides to be used for PCR amplification of mouse cDNAs.
cDNA Cloning of Mouse Collagenase 2-Oligonucleotides RCol1 (5Ј-TGACTCTGGTGATTTCTTGCTAA) and RCol3 (5Ј-GTGAAGGTCAGG-GGCGATGC) deduced from the coding exons 2 and 3 of the previously isolated genomic sequence were used as primers for RT-PCR amplification of RNA from mouse tissues using the RNA-PCR kit from Perkin-Elmer. All PCR assays were carried out in a GeneAmp 2400 PCR system from Perkin-Elmer. A fragment of the expected size (164 bp) was amplified from 17.5-day-old mouse embryo mRNA. A nucleotide sequence of this fragment confirmed its identity to the sequence corresponding to exons 2 and 3 of the genomic clone. The 5Ј-and 3Ј-extension of this cDNA for murine collagenase 2 was then carried out by successive cycles of rapid amplification of cDNA ends (RACE) using RNA from mouse embryo and the Marathon TM cDNA amplification kit (CLON-TECH Laboratories, Palo Alto, CA) essentially as described by the manufacturer. Each cycle of RACE allowed the extension of approximately 60 -200 bp of new collagenase 2 cDNA in either 5Ј or 3Ј direction. After cloning and sequencing the amplified products, new specific oligonucleotides were synthesized and used for the next RACE experiment. Final assembly of the complete cDNA was also performed as described by the manufacturer of the Marathon TM cDNA amplification kit.
Nucleotide Sequence Analysis-All DNA fragments were inserted in the polylinker region of pUC18 vector and sequenced by the dideoxy chain termination method using either M13 universal primer or cDNA specific primers and the Sequenase Version 2.0 kit (U. S. Biochemicals Corp.). All nucleotides were identified in both strands. Computer analysis of DNA and protein sequences was performed with the GCG software package from the University of Wisconsin Genetics Computer Group (36).
Northern Blot Analysis-Northern blots containing 20 g of total RNA of different mouse tissues were prehybridized at 42°C for 3 h in 50% formamide, 5 ϫ SSPE (1 ϫ ϭ 150 mM NaCl, 10 mM NaH 2 PO 4 , 1 mM EDTA, pH 7.4), 10 ϫ Denhardt's solution, 2% SDS, and 100 g/ml denatured herring sperm DNA and then hybridized with appropriate radiolabeled probes for 20 h under the same conditions. All probes used for analyzing murine mRNAs were of murine origin. Filters were washed with 0.1 ϫ SSC and 0.1% SDS for 2 h at 50°C and exposed to autoradiography. RNA integrity and equal loading were assessed by hybridization with an actin probe.
Collection and Culture of Mouse Peritoneal Polymorphonuclear Cells-Cell suspensions enriched in primary mouse polymorphonuclear cells were collected by washing out the peritoneal cavity with 8 ml of sterile phosphate-buffered saline 16 h after intraperitoneal injection with 2.5 ml of 4% thioglycollate broth. For in vitro activation, cell suspensions were resuspended at 3 ϫ 10 6 cells/ml in Dulbecco's modified Eagle's medium without fetal calf serum, and 1 ml/well was placed into 6-well plates. The cells were incubated with 10 Ϫ6 M TPA for 30 min to 3 h, and after incubation, culture supernatants were collected, concentrated by centrifugation through Amicon 10 filters, and analyzed by Western blot. Cells were collected by adding 0.3 ml of guanidinium thiocyanate solution and processed for RNA purification and Northern blot analysis.
Western Blot Analysis-Proteins from concentrated conditioned medium were separated by SDS-polyacrylamide gel electrophoresis under denaturing and reducing conditions and transferred to nitrocellulose membranes (Hybond-ECL, Amersham International). The membrane was blocked with 5% low fat dried milk and incubated with anti-human collagenase 2 monoclonal antibody 115-13D2 (5 g/ml), followed by goat-anti-mouse-IgG coupled to horseradish peroxidase (diluted 1:20,000). The bound antibody was detected by chemiluminescence (ECL system, NEN Research Products, Boston, MA).
Immunohistochemical Analysis-For light microscopic immunohistochemistry, fresh fragments of rat uterine horn obtained at different times postpartum were embedded in Tissue-Tek OCT compound and snap-frozen. Cryostat sections, 5 m thick, were set on Superfrost Plus slides (Menzel-Glaser, Germany) and stored at Ϫ20°C until used. Slides were quickly warmed to room temperature, rinsed with phosphate-buffered saline and treated with 0.3% hydrogen peroxide in methanol for 30 min to block endogenous peroxidase activity, and then incubated with 1% horse serum for 30 min. They were then placed in a humid chamber and sequentially incubated with rabbit antiserum antihuman collagenase 2 (50 g/ml in phosphate-buffered saline) for 12 h at 4°C, then with biotinylated goat anti-rabbit antibody (Biomeda Corp., Foster City, CA) and peroxidase-conjugated streptavidin (Biomeda Corp.) for 30 min at room temperature, washing with phosphate-buffered saline between incubations. After washes, sections were treated with a solution containing 0.66 M 3,3Ј-diaminobenzidine and 2 mM hydrogen peroxide in 50 mM Tris-HCl, pH 7.6. Sections were finally counterstained with hematoxylin, dehydrated, and mounted with Eukitt. The source and characteristics of the anti-human collagenase 2 have been previously described (37). Specificity of staining was determined using controls that involved incubation of tissues with buffer alone or with an equal amount of IgG from nonimmunized animals. In both cases there was no significant staining. In addition, immunostaining was completely abolished by preincubation of antiserum with equimolecular amounts of purified recombinant human collagenase 2. Identification of immunoreactive cells as neutrophils was performed by incubating slides containing consecutive tissue sections with naphthol AS-D chloroacetate esterase staining (Sigma).
Generation of an Escherichia coli Expression Vector for the Mouse Procollagenase 2 (proMMP-8) Catalytic Domain and Refolding of the Recombinant Proenzyme-An expression vector for the catalytic domain of mouse procollagenase 2 was generated by PCR using the following oligonucleotides: 5Ј-GCCAATGCGGATCCAGTACCTGAACACCTGGA-AGAG and GGAATTCTCAAGGCTTTGGGTGTGCTGGGCCCAGTA-GG, which introduced a BamHI site in the forward primer and a stop codon with an EcoRI site just following the codon for Ala 278 in the reverse primer. The PCR product was cleaved with BamHI and EcoRI and ligated into frame of the pRSET B expression vector (Invitrogen) previously cleaved with the above restriction enzymes. The expression vector encodes a fusion protein consisting of the His-tag sequence of the vector followed by the mouse procollagenase 2 catalytic domain and hinge sequence motifs with the alteration of a Phe residue to Asp in the original sequence because of the introduction of the BamHI site. Thus, the expressed protein would have the amino-terminal sequence Asp-Pro-Val-Pro-Glu-His. Expression was achieved using the BL21 (DE3)pLysS E. coli strain in the presence of 1 mM isopropyl-1-thio-␤-Dgalactopyranoside. Production was monitored by SDS-PAGE, and inclusion bodies were generated following overnight expression at 37°C. Purified inclusion bodies were solubilized in 20 mM Tris/HCl, pH 8.0, 5 M guanidine hydrochloride, and 5 mM ␤-mercaptoethanol and diluted into refolding buffer (20 mM Tris/H 2 SO 4 , pH 7.5, 5 mM CaSO 4 , 100 mM Na 2 SO 4 , 0.5 M ZnSO 4 , 2% glycerol, 0.05% NaN 3 ), followed by purification using a Ni-NTA agarose column.
Activation of the Mouse Procollagenase 2 (proMMP-8) Catalytic Domain by Active Matrilysin-Human promatrilysin, produced as described previously (38), was activated by p-aminophenylmercuric acetate for 1 h at 37°C. p-Aminophenylmercuric acetate was removed by gel filtration using a G-25 desalting column. Mouse procollagenase 2 catalytic domain (1 M) was incubated with 0.2 M active matrilysin, and aliquots of the reaction mixture were removed at the time intervals indicated and analyzed by activity assay using the quenched fluorescent substrate Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH 2 and by SDS-PAGE.

RESULTS
Identification, Molecular Cloning, and Nucleotide Sequencing of a cDNA Encoding Mouse Collagenase 2-Because all previous attempts to identify the murine homologues of human collagenases 1 and 2 were unsuccessful, we approached an alternative strategy based on analysis of genomic libraries rather than on the screening of cDNA libraries prepared from RNA of tissues in which murine collagenases could be overexpressed. To this purpose, we hypothesized that mice could have a MMP gene cluster similar to that found in humans located on chromosome 11q22 and containing 8 different MMP genes, including those encoding collagenases 1, 2, and 3 (13,39,40). To isolate murine genomic clones containing this hypothetical gene cluster, we screened a YAC mouse genome library by PCR amplification of distinct YAC pools with specific oligonucleotides for murine collagenase 3. Following this strategy, several collagenase 3-containing YAC clones were obtained, all of them being large enough to contain the coding information for other MMPs presumably clustered to the collagenase 3 gene. Then, by using degenerate oligonucleotides derived from DNA sequences conserved in the collagenase subfamily of MMPs, an 870-bp genomic fragment was PCR-amplified from DNA of YAC clone I139A1. Nucleotide sequence analysis of this fragment allowed the identification of two putative exons whose sequence was distinct from those previously determined for any known murine MMP gene but significantly similar to the se-quence of the exons 2 and 3 previously identified in some human MMP genes. A more detailed comparative analysis revealed that the percentage of identities of the two identified murine exons was about 79% with those of human collagenase 2, 60% with human collagenase 1, and 55% with human collagenase 3. Taken together, these results strongly suggested that we had isolated a fragment of the murine homologue of collagenase 2. To obtain a complete cDNA sequence for this novel murine MMP, we performed successive cycles of 5Ј-and 3Ј-RACE using specific oligonucleotides derived from the sequences determined for the putative exons of mouse collagenase 2 and cDNA from 17.5-day-old mouse embryo as template. After following the strategy depicted under "Experimental Procedures," a complete cDNA sequence for this murine MMP was finally obtained (Fig. 1). Analysis of this sequence revealed the presence of an open reading frame coding for a protein of 465 amino acids with a predicted molecular mass of 53.1 kDa. Amino acid sequence comparisons of the identified sequence with that of human collagenase 2 revealed a high percentage of identities between them (74%), whereas the percentage of identities with the other human collagenases was significantly lower (56 and 50% with collagenase 1 and collagenase 3, respectively). These data strongly suggest that the cloned cDNA corresponds to mouse collagenase 2 (Fig. 2).
The deduced amino acid sequence for murine collagenase 2 presents all the structural features characteristic of members of this family of proteolytic enzymes, including the three domains that are conserved among all of them: the predomain encoding the hydrophobic sequence that targets these proteins to the secretory pathway, the prodomain containing the conserved PRCGVPD motif (at positions 89 -95) involved in maintaining the latency of these enzymes, and the catalytic domain with the consensus sequence HEXGHXXGXXH (at positions 217-227) containing the three histidine residues involved in the coordination of the zinc atom at the active site. The murine collagenase 2 sequence also contains the hemopexin-like domain found in the COOH-terminal region in all MMP family members with the exception of matrilysin and the soybean leaf MMP (41,42). A more detailed analysis of the deduced sequence for mouse collagenase 2 revealed that it displays all specific features characteristic of the collagenase subfamily of MMPs. Thus, it contains the three residues close to the zincbinding site (Tyr-209, Asp-230, and Gly-232) that are conserved in all collagenases characterized to date and are never present in stromelysins (Fig. 2) (17). In addition, murine collagenase 2 contains the 16-amino acid sequence of the hinge region essential to determine the specific action of collagenases on triple helical collagen (43,44). By contrast, it lacks the 9-residue insertion present in this region in stromelysins, and whose introduction in human collagenase 2 results in complete loss of the collagenolytic activity of the chimeric enzyme (43). Finally, the structural analysis of the amino acid sequence of murine collagenase 2 confirmed that, like its human homologue, it lacks the fibronectin-like domain present in all gelatinases, the COOH-terminal extension rich in hydrophobic residues characteristic of MT-MMPs, and the furin activation sequence (RX(R/X)R) mediating the intracellular activation of MT-MMPs and stromelysin 3 (45)(46)(47). Taken together, these structural data strongly suggest that the cloned cDNA codes for the murine homologue of collagenase 2. In addition, the availability of this cDNA opens the possibility to perform further studies on tissue distribution and functional roles of this proteolytic enzyme. During revision of this manuscript, Lawson et al. (48) reported a cDNA sequence very similar to that described here for murine collagenase 2. The differences between both sequences can be because of naturally occurring polymor-phisms or variations originated during PCR amplifications or reverse transcription experiments. Nevertheless, in virtually all positions in which differences are found, the sequence reported here for murine collagenase 2 (Figs. 1 and 2) coincides with the amino acids present at equivalent positions in its human counterpart.
Induction of Mouse Collagenase 2 Expression during Inflammatory Processes-To elucidate whether the newly identified murine collagenase 2 is a MMP released by neutrophil cells during inflammatory processes as its human counterpart, we examined its expression by mouse polymorphonuclear cells under inflammatory conditions. In this regard, it is well known that thioglycollate injection into the mouse peritoneum results in an inflammatory response with accumulation and activation of neutrophils and lymphocytes within the first 16 h followed by later recruitment of macrophages (49). To evaluate the possibility that collagenase 2 was one of the proteinases present in this inflammatory process, a neutrophil-enriched cell population was isolated from the peritoneal cavity of thioglycollate-treated mice, plated, and stimulated in vitro with TPA to trigger degranulation (50). Then, RNA from treated cells was isolated and analyzed by Northern blot using the full-length murine collagenase 2 cDNA as a probe. As can be seen in Fig.  3A, a single mRNA transcript of about 3.3 kb was detected in all cases, indicating that collagenase 2 mRNA is present in the isolated cells collected from the inflamed mouse peritoneal cavity. The same filter was reprobed with specific probes derived from cloned cDNAs for murine collagenase 3 and human collagenase 1, but no hybridizing band was detected in any case (data not shown), suggesting that other collagenase genes were not significantly expressed at these conditions. To examine the presence of collagenase 2 at the protein level in these peritoneal cells, conditioned medium from TPA-treated cells was concentrated and analyzed by Western blot using a monoclonal antibody directed against human recombinant collagenase 2. As shown in Fig. 3B, medium from cells stimulated for degranulation with TPA for 30 min or 3 h contained a protein that was detected by the monoclonal antibody 115-13D2 and was not present in conditioned medium from unstimulated cells. The detected band had an estimated molecular mass of about 60 kDa, which is higher than that calculated from the amino acid sequence, suggesting that murine collagenase 2 is glycosylated in some of the three potential sites for N-linked glycosylation present in its amino acid sequence (at positions 55, 112, and 119). In addition, hybridization of the same membrane with antibodies against murine collagenase 3 or human collagenase 1 did not recognize any specific band (data not shown). In summary, these results provide evidence that the mouse collagenase 2 gene identified herein is transcribed and translated at least in inflamed peritoneal cells. Furthermore, the comparative expression analysis with the genes encoding other collagenases allows us to conclude that murine collagenase 2 is the major collagenase expressed in these inflammatory conditions generated by thioglycollate injection into the mouse peritoneum.

Collagenase 2 Is Expressed at High Levels during Mouse Embryogenesis and in the Involuting Postpartum Uterus-To
examine the possibility that murine collagenase 2 was expressed in processes other than inflammatory responses, we undertook an extensive analysis of its mRNA levels in a series of physiological conditions involving an extensive connective tissue remodeling, such as embryonic development and reproductive processes. To this purpose, we first performed RT-PCR amplification with oligonucleotides specific of mouse collagenase 2 and RNA prepared from a variety of normal tissues. This analysis revealed that collagenase 2 is expressed during fetal development as well as in ovaries from a pregnant mouse. Similarly, a positive amplification band, whose identity was confirmed by nucleotide sequencing, was detected in kidney, muscle, and uterus but not in other tissues that produce different MMPs such as the mammary gland and brain (Fig. 4A). Collagenase 2 expression during mouse embryogenesis was also confirmed by Northern blot analysis. As shown in Fig. 4B, a transcript of about 3.3 kb was detected in RNA from 17.5-  2 and 3, respectively) under serum-free conditions. Total cellular RNA was obtained by guanidinium/acid phenol extraction and equal amounts of RNA were subjected to Northern blot analysis with mouse collagenase 2 cDNA as the probe (left). Ethidium bromide stain of the same gel shows equal content of the loaded RNA samples (right). B, Western blot analysis of collagenase 2 production by mouse peritoneal inflammatory cells induced with TPA. Conditioned medium from the above described plated cells was concentrated 25-fold and analyzed by SDS-PAGE along with recombinant purified human collagenase 2 (100 ng). After electrophoresis, proteins were transferred to nitrocellulose membranes and incubated with monoclonal antibody 115-13D2 against collagenase 2. Immunoblots were developed with a chemiluminescence detection reagent. Lane 1, medium from cells unstimulated. Lanes 2 and 3, medium from cells treated with TPA for 30 min and 3 h, respectively. Lane H, recombinant purified human collagenase 2.

FIG. 2. Comparison of the amino acid sequence of mouse collagenase 2 (Col-2 Mm) with human collagenases.
The amino acid sequences of human collagenase 1 (Col-1 Hs), collagenase 2 (Col-2 Hs), and collagenase 3 (Col-3 Hs) were extracted from the GenBank TM data base, and the multiple alignment was performed with the PILEUP program of the GCG package (36). Regions corresponding to the activation locus and catalytic domain are gray-shadowed, and collagenase-specific conserved residues are shown with an asterisk. day-old mouse embryos but not in 12.5-day-old or earlier embryos ( Fig. 4B and data not shown). Expression of collagenase 2 was still present at birth, although the intensity of the detected band was weaker. The level of mouse collagenase 2 expression in tissues like kidney and muscle, previously detected by RT-PCR, was below Northern blot sensitivity (Fig.  4B). Similarly, we could not observe by Northern blot analysis any collagenase 2 expression in other tissues like placenta, cartilage, or intestine. By contrast, a clear hybridizing band was observed in postpartum uterus. This finding was particularly interesting because remodeling of uterine tissue during postpartum involution is a process that requires a potent collagenolytic activity to perform the breakdown of large amounts of type I collagen. Therefore, we decided to investigate in more detail the possibility that collagenase 2 could be involved in the extensive remodeling of connective tissue during postpartum involution of the uterus.
To do that, we prepared tissue samples of the uterus at different pre-and postpartum stages and analyzed collagenase 2 expression by Northern blot. In addition, we performed a comparative analysis with the expression levels in these tissues of both matrilysin, whose participation in the postpartum uterine involution has been widely studied, and collagenase 3, which has also been proposed to play a role in this process (51)(52)(53)(54). As can be observed in Fig. 5, collagenase 2 expression, detected as a hybridizing band of 3.3 kb, started in a 1-day postpartum uterus and extended up to 5 days postpartum. By contrast, no hybridizing band was detected in uterine tissue samples taken during the estrous cycle or in prepartum or partum uteri as well as in RNA from ovarian tissue. When the same Northern blot was hybridized with a probe corresponding to full-length matrilysin cDNA, a very strong hybridization band was present in uterine tissue during partum, reaching maximal levels at 24 h postpartum and declining thereafter. In marked contrast to the above results on collagenase 2 expression, matrilysin levels were undetectable by day 5 postpartum. Also in contrast with the collagenase 2 results, matrilysin was significantly expressed in the uterus from a mature nonpregnant rat as well as in RNA from ovaries of immature rats. Finally, we hybridized the same blots under the same conditions with a 2-kb probe specific for murine collagenase 3. However, as can be seen in Fig. 5, collagenase 3 mRNA was undetectable by Northern blot analysis in any uterine tissue. By contrast, and as previously reported (55), a clear positive band was observed in RNA from ovaries at different estrous cycle stages, confirming that the absence of significant collagenase 3 expression in uterine samples is not because of technical problems during the hybridization process. These results were also confirmed at the protein level because we were unable to detect any collagenase 3-specific signal after immunohistochemical analysis of uterine tissue sections taken at different times postpartum. By contrast, strong and specific signals were detected after incubation of tissue sections from rat ovary with the same anti-collagenase 3 antiserum (55 and data not shown).
To examine the identity of cells responsible for the produc-

FIG. 4. Expression of collagenase 2 in mouse tissues.
A, RT-PCR was performed on 1 g of RNA from the indicated samples in a volume of 100 l with the mouse collagenase 2-specific oligonucleotides Rcol1 and Rcol3 as primers. 20 l of the final product were separated on a 2% agarose gel and then transferred to nylon filter. Filters were hybridized with a 5Ј end-labeled internal specific oligonucleotide as probe and autoradiographed. Blank lane shows RT-PCR performed without added template. B, samples of 20 g of total RNA from mouse tissues were separated by agarose gel electrophoresis under denaturing conditions, blotted onto nylon filters, and analyzed by hybridization with the fulllength cDNA for mouse collagenase 2. Filters were exposed to autoradiography at Ϫ70°C for 7 days. Hybridization with an actin probe was performed to verify sample loading. The positions of 28 and 18 S RNA are indicated.
FIG. 5. Northern blot analysis of RNA from rat ovaries taken at different phases of the estrous cycle and from uterine tissue during pregnancy, labor, and postpartum. Total RNA (20 g) of each sample was separated by agarose gel electrophoresis under denaturing conditions, blotted onto nylon filters, and hybridized consecutively with radiolabeled probes corresponding to mouse collagenase 2 full-length cDNA, mouse collagenase 32-kb probe, 800-bp murine matrilysin cDNA probe, and finally with an actin probe. Filters were exposed to autoradiography at Ϫ70°C for 7 days after hybridization with the collagenase 2 and with the collagenase 3 probes, for 10 h after hybridization with the matrilysin probe, and for 48 h after hybridization with the actin probe. tion of collagenase 2 in the involuting uterus, we carried out an immunohistochemical analysis on tissue sections taken at days 1, 5, and 7 postpartum, using antibodies raised against purified human collagenase 2. At day 1 postpartum (Fig. 6, A and B), immunoreactivity was mainly observed in the cytoplasm of scattered cells located in the endometrial stroma closed to the columnar epithelium as well as in the luminal space. Collagenase 2-immunopositive cells were also found in the uterus at day 5 postpartum (Fig. 6, C and D), but marked differences were observed in the relative distribution of these cells when compared with those detected at day 1 postpartum. Thus, at day 5 positive cells appeared widely distributed throughout the different uterine layers being found frequently among the muscle cells of the myometrium. Finally, at day 7 postpartum, positive cells were virtually absent (Fig. 6E). In all cases of immunopositive staining, the immunoreactivity was completely abolished by preincubation of the anti-collagenase 2 antibodies with purified recombinant human collagenase 2 (data not shown). It is also worthwhile mentioning that collagenase 2 immunoreactive cells showed histological characteristics of circulating leukocytes escaped from vessels and moved into the endometrial stroma in an inflammatory-like process. The identity of these collagenase 2 immunoreactive cells as neutrophils was confirmed by naphthol chloroacetate esterase staining, a method specifically designed to identify neutrophils (Fig. 6F). An evident colocalization of collagenase 2 and naphthol chloroacetate esterase positive cells was observed.
Activation of Murine Procollagenase 2 (proMMP-8) by Recombinant Active Matrilysin (MMP-7)-To evaluate whether the coexpression of collagenase 2 and matrilysin (MMP-7) during mouse uterine postpartum involution is of physiological significance, we initially performed activation experiments of procollagenase 2 by active matrilysin. Recombinant murine procollagenase 2 (1 M) was incubated with 0.2 M active matrilysin for the time intervals indicated in Fig. 7. As can be seen, matrilysin was able to convert mouse procollagenase 2 to the active enzyme in a time-dependent fashion. Analysis by SDS-PAGE revealed that the propeptide of the mouse procollagenase 2 catalytic domain (32,000) was cleaved by matrilysin through two intermediate forms showing apparent molecular masses of 28,000 and 24,000 (Fig. 7). These intermediate forms were converted to the final active form of 22,000 after 500 min of incubation. Earlier data using the human procollagenase 2 had already revealed that human procollagenase 2 was cleaved at the Phe 80 -Leu 81 peptide bond by matrilysin-specific cleavage rendering active human collagenase 2 with a specific collagenolytic activity of 1450 units/mg active enzyme. 2

DISCUSSION
In this work we describe the identification of the murine homologue of collagenase 2 (MMP-8) and report that it is a protease associated with inflammatory conditions. We also propose that collagenase 2 may play a role in the massive resorption of connective tissue that takes place during postpartum involution of the uterus. This finding suggests an additional function for this proteolytic enzyme whose presence had been so far described exclusively in human cells, but whose participation in this involuting process was unknown.
The identification and cloning of the cDNA for this murine protease was the result of studies directed to clarify the debated question about the number and identity of collagenolytic enzymes present in murine tissues. To date, only the murine homologue of human collagenase 3 had been identified, whereas successive attempts carried out by different groups including ours had failed to clone the homologues of collag-enases 1 and 2. By using a combined strategy involving the search of YAC genomic clones containing a hypothetical MMP gene cluster followed by PCR amplification of positive YAC clones with degenerate oligonucleotides encoding conserved regions in human collagenases, we identified a partial genomic sequence encoding a putative mouse collagenase 2. This genomic information was used to design the specific oligonucleotides required for RT-PCR and RACE procedures, which finally led us to obtain a collagenase 2 full-length cDNA from RNA of mouse embryos. Pairwise comparisons with the sequences of previously described MMPs strongly suggest that the obtained sequence corresponds to the murine homologue of human collagenase 2. This finding demonstrates that collagenase 3 is not the only collagenolytic enzyme encoded in the genome of rodents and emphasizes the necessity of performing further studies to clarify whether a homologue of collagenase 1 is also present in murine tissues.
The availability of the cDNA for murine collagenase 2 provided a specific reagent to perform functional studies in animal models directed to elucidate the tissue distribution and physiological role of this proteolytic enzyme. In this work, we show that collagenase 2 is expressed by polymorphonuclear leukocytes released under inflammatory conditions. The expression analysis performed in this work also revealed the presence of collagenase 2 at late stages during mouse embryogenesis likely associated with the appearance of mature polymorphonuclear cells in the developing mouse. This observation is consistent with studies of location and temporal appearance of hematopoietic cells during murine embryogenesis showing that totipotent stem cells first originate in the yolk sac, then migrate to the fetal liver, and finally colonize the bone marrow shortly before birth (56). Nevertheless, the most striking finding in this collagenase 2 expression analysis in murine tissues was its presence at large levels in the postpartum uterus. A likely role for collagenase 2 in this context would be the degradation of collagen fibers during the process of massive connective tissue resumption occurring immediately after parturition. In support of this proposed role, collagenase 2 preferentially degrades type I collagen (19), which is a predominant fibrillar collagen in the uterine tissue. Interestingly, previous studies have shown that another collagenolytic enzyme, collagenase 3, can be detected in the rat postpartum uterus (53,54). However, the presence of collagenase 3 transcripts was only detected at very low levels and, usually, these transcripts could only be visualized after using highly sensitive RT-PCR procedures (53), suggesting a secondary role of this enzyme in the involuting process when compared with other abundantly expressed MMPs such as matrilysin and collagenase 2. Similarly, over the last year a series of in vitro studies have reported the ability of cultured uterine smooth muscle cells to produce murine collagenase 3 upon stimulation with serotonin (57,58). However, none of these studies have shown the in vivo presence of RNA transcripts for collagenase 3 in the involuting uterus, thus hampering the extrapolation of the above results with cultured cells to the in vivo situation. Finally, it is remarkable that although the present study has confirmed that matrilysin is expressed at very high levels shortly after parturition, it should be taken into account that this enzyme is unable to cleave fibrillar collagens suggesting that its role in the involuting process may be distinct from that of collagenase 2 (51,52). One possibility is that both enzymes could form part of an activation cascade with the ability to generate the collagenolytic activity required in this resorptive process. In fact, we have shown that matrilysin is able to activate murine procollagenase 2 in a time-dependent manner, providing an indirect support for this hypothesis. In addition, the dynamics of temporal appearance of both enzymes during the involutive process is consistent with the presence of matrilysin upstream of collagenase 2 in this putative activation cascade. Further studies will be required to provide a more definitive evidence on the putative collaborative role of both metalloproteases in the process of postpartum involution of the uterus.
The proposed role of collagenase 2 in the process of uterine resumption could also explain previous observations of the impaired reproductive ability of transgenic female animals having a mutant type I collagen that cannot be cleaved into the characteristic three-fourths and one-fourth fragments (59,60). In fact, the uterine wall of these transgenic mice accumulates multiple nodules consisting of large collagen aggregations that reflect the impaired collagen degradation during the postpartum period. The maintenance of this high collagen content in the postpartum uterus would be a consequence of the inability of collagenase 2, the predominant collagenolytic enzyme in this process, to remove a mutant collagen carrying alterations in the collagenase cleavage site. The availability of genomic and cDNA clones for murine collagenase 2 opens the possibility to generate animals in which the expression of this gene can be deregulated or ablated. These animal models will be essential to further extend the proposed functional relevance of this enzyme in both inflammatory and reproductive processes.