Heparan Sulfate Proteoglycans as Extracellular Docking Molecules for Matrilysin (Matrix Metalloproteinase 7)*

Many matrix metalloproteinases (MMPs) are tightly bound to tissues; matrilysin (MMP-7), although the smallest of the MMPs, is one of the most tightly bound. The most likely docking molecules for MMP-7 are heparan sulfate proteoglycans on or around epithelial cells and in the underlying basement membrane. This is established by extraction experiments and confocal microscopy. The enzyme is extracted from homogenates of postpartum rat uterus by heparin/heparan sulfate and by heparinase III treatment. The enzyme is colocalized with heparan sulfate in the apical region of uterine glandular epithelial cells and can be released by heparinase digestion. Heparan sulfate and MMP-7 are expressed at similar stages of the rat estrous cycle. The strength of heparin binding by recombinant rat proMMP-7 was examined by affinity chromatography, affinity coelectrophoresis, and homogeneous enzyme-based binding assay; theK D is 5–10 nm. Zymographic measurement of MMP-7 activity is greatly enhanced by heparin. Two putative heparin-binding peptides have been identified near the C- and N-terminal regions of proMMP-7; however, molecular modeling suggests a more extensive binding track or cradle crossing multiple peptide strands. Evidence is also found for the binding of MMP-2, -9, and -13. Binding of MMP-7 and other MMPs to heparan sulfate in the extracellular space could prevent loss of secreted enzyme, provide a reservoir of latent enzyme, and facilitate cellular sensing and regulation of enzyme levels. Binding to the cell surface could position the enzyme for directed proteolytic attack, for activation of or by other MMPs and for regulation of other cell surface proteins. Dislodging MMPs by treatment with compounds such as heparin might be beneficial in attenuating excessive tissue breakdown such as occurs in cancer metastasis, arthritis, and angiogenesis.

Remodeling of the extracellular matrix is critical in many physiological processes such as embryonic development, bone growth, nerve outgrowth, ovulation, uterine involution, and wound healing; the family of matrix metalloproteinases (MMPs) 1 plays a major role in such remodeling (1,2). These proteases also have a prominent role in pathological processes such as cancer invasion and metastasis (3), arthritis (4), and atherosclerosis (5). The matrixin family comprises 20 members that differ in domain structure and specificity (6). Much is known about the regulation of these enzymes at the level of transcription (7), activation of proenzyme (8), and inhibition by natural inhibitors such as the tissue inhibitors of metalloproteinases (9). However, an important aspect of these enzymes has generally been overlooked, namely, how they are anchored outside the cell.
An instructive example is found in articular cartilage: compression drives out fluid into the synovial space, and release of pressure leads to fluid imbibition. This tissue contains MMP-1 (collagenase 1), MMP-2 (gelatinase A), MMP-3 (stromelysin), MMP-8 (collagenase 2), and MMP-13 (collagenase 3) (4), yet the continual slow movement of fluid does not dislodge or wash away the bulk of these enzyme activities. MMPs are commonly studied by tissue culture methods, a situation in which there is a vast overproduction of enzyme and a paucity of matrix; here the MMPs appear to be readily soluble. However, early workers tried to extract MMPs from various tissues such as skin, cartilage, and uterus but had success only with chaotropic agents such as 5 M urea (10) or 2 M guanidine HCl (11). It was thought that MMP-1 bound to its substrate (12), but subsequent studies showed that the proMMP-1 could not bind collagen (13) but was still difficult to extract.
Anchoring MMPs to the cell surface or extracellular matrix would not only prevent them from rapidly diffusing away but would also enable the cell to keep them under close regulatory control. If there were specific binding sites, the cell could monitor these sites by receptors or integrins and respond by altering the synthesis of MMP. If the MMPs bind to the matrix, this could provide a reservoir for subsequent rapid tissue degradation. If they bind to the cell surface, they could be positioned for activation as in the case of gelatinase A (8), for interaction with cell surface adhesion molecules or receptors, for regulating the turnover of these molecules by shedding activity, or for a directed attack on the matrix as illustrated by MMP-14 bound on invadopodia (14).
To explore the binding of MMPs to cells or matrix, we have selected the smallest member of the family, matrilysin (MMP-7), which has only two domains: the propeptide (9 kDa) and the catalytic domain (19 kDa). This enzyme is greatly elevated in the postpartum rat uterus but is quite difficult to extract; it requires the use of 0.1 M calcium salt and 60°C heat (15). This method was originally developed for rat uterine collagenase 3 (16), an enzyme that was shown to bind tightly to heparin-Sepharose (17). This suggested that sulfated glycosaminoglycans might provide anchoring sites for MMP-7. Our present findings indicate that heparan sulfates on the cell surface and/or in the matrix may be the major binding site of MMP-7 and may also bind other MMPs such as MMP-1, -2, -9 (gelatinase B), and -13.

Extraction of MMPs from Postpartum Rat Uterus-Pregnant
Sprague-Dawley rats were purchased from Harlan. Uteri were collected 1 day postpartum when MMP-7 levels are highest, weighed (ϳ2 g), washed three times with cold 50 mM Tris, pH 7.5, 0.03% sodium azide, and homogenized in 20 ml of this buffer containing 0.1% Triton X-100 with a Polytron for 6 min at 4°C. The mixture was centrifuged at 11,000 rpm for 20 min. The pellet was washed twice, resuspended in the same volume of cold 50 mM Tris, pH 7.5, 0.03% sodium azide, 50 M ZPCK, 50 M 4-(2-aminoethyl)-benzenesulfonyl fluoride. This was divided at 0.5 ml/tube. Extractants (Sigma) were added, held at 4°C for 30 min, and then centrifuged at 14,000 rpm for 10 min. For heat extraction, see Ref. 16.
Western Immunoblotting-The protein samples were separated by SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes (Bio-Rad). Then 0.1% Tween-20 (TTBS) containing 5% nonfat milk was used to block the membrane for 3 h at 24°C. The first antibody (e.g. RM7-P) was applied at 4°C overnight. After washing three times with TTBS/milk, the second antibody (e.g. goat anti-rabbit IgG-alkaline phosphatase) was applied for 2 h. Then the transblot was washed three times with TTBS/milk and stained with NBT/BCIP (Pierce). If horseradish peroxidase-conjugated secondary antibody was used, the transblot was visualized by chemiluminescence (DuPont Renaissance kit) Rat Uterus Tissue Section Preparation-Pieces of mature rat uterus at various points in the estrous cycles (staged by cervical smears) were fixed in paraformaldhyde and embedded. Sections 0.5 m thick were cut, dewaxed in xylene for 5 min, and fixed in 100% methanol for 20 min. Sections were washed with PBS three times and then submitted to directly immunofluorescence staining or pretreatment of sections for glycosaminoglycan digestion. Digestion conditions were 0.2 units/ml of heparinase III (bacterial, Sigma) and/or chondroitin ABC lyase (Sigma) in buffer containing 0.05% sodium azide 100 M zinc chloride, 5 mM CaCl 2 /MgCl 2 , 10 mM phenylmethylsulfonyl fluoride, 0.1 mM ZPCK/ TPCK/TLCK, 50 M BB94 (British Biotech), 0.1 mM antipain 0.15 M NaCl acetate buffer (pH 6.5) at 37°C for 18 h. Controls were incubated with buffer only. Samples were gently washed twice in PBS, then post-fixed with 2% formaldehyde in PBS with 0.1 mM CaCl 2 and 1 mM MgCl 2 for 10 min, and washed twice with PBS for 10 min.
Immunofluorescence Double Staining and Confocal Microscopy-The tissue sections were blocked with 3% bovine serum albumin and 10% goat serum at room temperature in PBS for 40 min and subsequently exposed to first antibodies: rabbit polyclonal antibodies (IgG) to rat MMP-7 (RM7-C) and mouse monoclonal antibodies (IgM) to heparan sulfate (Sagaku) for 1 h at room temperature. Sections were washed three times in PBS for 45 min. Primary antibodies were detected by using goat polyclonal antibodies to rabbit IgG and mouse IgM conjugated to either fluorescein isothiocyanate or Texas Red, respectively (Jackson ImmunoResearch Labs., Inc.). Sections were washed three times in PBS for 45 min. The samples were mounted on slides and analyzed by confocal microscopy (inverted Nikon with Multiprobe 2001, Molecular Dynamics). Excitation was at 488 nm, and emission was at 530 and 590 nm, for fluorescein isothiocyanate and Texas Red, respectively. Images were captured at 0.6 nm increments along the z axis and converted to composite images by ImageSpace 3.10 software (Molecular Dynamics).
Expression, Purification, and Folding of Recombinant Rat proMMP-7 from Escherichia coli BL21(DE3)-The full-length rat proMMP-7 coding region (19) was cloned into the NdeI and BamHI sites of PET3a vector (Novagen). The cDNA insert of the proMMP-7 expression construct was completely sequenced by using the Sequenase version 2.0 kit (U. S. Biochemical Corp.). The expression plasmid was transfected into E. coli BL21(DE3) cells (Novagen). Cells were cultured for 6 h (A 600 ϭ 0.2ϳ0.4) followed by 2 h with isopropyl-1-thio-␤-D-galactopyranoside. Cells were collected, passed through a French press, and centrifuged at 13,000 rpm for 20 min. The pellets were washed three times with 10 ml of inclusion body wash solution (50 mM Tris, pH 7.5, 10 mM dithiothreitol, 1% Triton X-100). The pellets suspended in 8 M urea, 50 mM Tris, pH 7.5, 0.02% sodium azide. After 48 h at 4°C, the sample was centrifuged. The soluble fraction was passed through a PD10 column (Amersham Pharmacia Biotech) to remove dithiothreitol and then applied to a zinc chelate-Sepharose 6LB column (Amersham Pharmacia Biotech). The bound proMMP-7 was eluted by stepwise decreasing pH. The product had a purity greater than 95%. The proMMP-7, still in 8 M urea, was diluted dropwise 10-fold in ice-cold refolding buffer (20 mM acetate, pH 5.6, 10 mM CaCl 2 , 1 M ZnCl 2 , and 0.05% Triton X-100). After 30 min it was centrifuged at 14,000 rpm for 10 min to remove the insoluble portion. Storage was at Ϫ70°C in 18% glycerol. Proper folding was shown by aminophenyl mercuric acetate activation of latent form, specific activity, and specificity of bond cleavage in gelatin and transferrin.
The E198Q mutant of proMMP-7 was prepared by the Quick Change site-directed mutagenesis method (Stratagene). The PET3a.proMMP-7 expression vector (above) was used with the inserts of rat proMMP-7 as templates and a pair of synthetic oligonucleotide primers containing the desired mutation site (GAA to CAA), the 3Ј-oligonucleotide primer (Ϫ)strand) 5Ј-ACCCAGAGAGTGGCCAAGTTGATGAGTGGC-3Ј and 5Јoligonucleotide primer (ϩ)-strand) 5Ј-GCCACTCATCAACTTGGCCAC-TCTCTGGGT-3Ј for PCR amplification. The PCR product was digested with DpnI endonuclease. This PCR extended nicked vector was then transformed into E. coli, Epicurian Coli XL1-Blue supercompetent cells (Stratagene). The ampicillin resistant colonies were screened by PCR and NdeI/BamHI digestion, and the positive plasmids were purified and transformed into the expression host BL21(DE3) strain. The total sequence including the desired mutation site was confirmed by DNA sequencing. Purification and folding was as described above.
Heparin-Agarose Chromatography-rproMMP-7 in 8 M urea was diluted 1:10 in refolding buffer (20 mM acetate buffer, pH 5.6, 10 mM CaCl 2 , 1 M ZnCl 2 , and 0.05% Triton X-100). Heparin-agarose beads (Sigma, H0402) were suspended in 50 mM Tris, 5 mM CaCl 2 , pH 7.8. A mixture of 1 ml of enzyme plus 4-ml beads (0.5 mg) was stirred at room temperature for 4 h. The mixture at pH 7.5 was poured into a small column and then washed with 50 mM Tris, pH 7.5, plus 0.15 M NaCl. Elution was then carried out in steps of 0.5 ml with increasing concentrations of NaCl or heparin in Tris buffer. The enzyme did not bind to agarose beads lacking heparin.
Homogeneous DNase II-based Binding Assay-This assay system is based on the method of Guo et al. (22). The binding of compounds to heparin was assessed by using fresh standard solutions of these substances in pH 4.8 acetate buffer ϩ 5 mM dithiothreitol. The concentrations of enzyme and substrate were monitored at 280 and 405 nm, respectively. Heparin concentration was adjusted to inhibit 1.1 M DNase II by 90%; compounds binding heparin reversed this inhibition. The percentage of inhibition observed in the presence of both the test species and heparin was plotted versus log concentration to yield the dose-response curve for the given species.

RESULTS
Heparin, Heparan Sulfate, Sulfated Compounds, and Heparinase III Can Release Bound MMP-7 from Rat Uterus-Various sulfated compounds were tested to see whether they could extract MMP-7 from the rat uterus at 1 day postpartum (a rich source of enzyme). Fig. 1A illustrates that heparin and heparan sulfates are effective extractants at 0.2 mg/ml, but chondroitin and dermatan sulfates are less effective. Sulfated compounds such as pentosan polysulfate and suramin also liberate MMP-7, but higher concentrations (2 mg/ml) are required (Fig.  1A, lanes 7 and 8). The heparin antagonist, protamine, which is highly positively charged, can also extract MMP-7 from tissues (lane 9). Blotting with a polyclonal antibody specific for a segment of the propeptide (RM7-P, see "Experimental Procedures") established that the zymogram bands correspond to proMMP-7 (data not shown). The major forms of MMP-7 that appear are the 28-kDa proform and a partially activated form of about 25 kDa. Active enzyme did not appear where expected (18 kDa), but there is evidence, obtained by use of antibody RM7-W to the whole enzyme, that it aggregates to produce some of the upper bands on the zymogram. Pro-and active MMP-2, proMMP-9, and proMMP-13 were released by the same extractants as for MMP-7 and could be detected by gelatin zymography (data not shown).
Next, the effect of heparin concentration on MMP-7 extraction was studied. Because high levels of heparin (Ͼ2 mg/ml or 20 g/gel lane) affect the resolution and enzyme activity on the subsequent CM-transferrin zymography, samples above 1 mg/ml were diluted to that final concentration. The optimum extraction of proMMP-7 from the uterus is with 2-4 mg heparin/ml based on the dilution factor multiplied by band intensity measured by UVP image analysis of Fig. 1B (left panel). The polyclonal antibody RM7-P confirmed that the 28-kDa band in Fig. 1B (right panel) was due to proMMP-7. The levels of heparin (0.2 mg/ml) used in Fig. 1A were suboptimal but were chosen to emphasize the differences among the various compounds tested. The numbers at the top indicate the concentration (mg/ml) of heparin used for extraction of rat uterus. 10 l of each extract was mixed with sample buffer and applied to CM-transferrin zymographic gels and Western blot. The higher concentration extracts were first diluted to give a final concentration of 1 mg/ml before the 10-l aliquot was taken. The rabbit antibody RM7-P was used for the Western blot. C, scheme for extraction of proMMP-7 from rat uterus. Seven rat uteri were pooled and homogenized in Triton X-100; quadruplicate samples were carried through the following steps. The homogenates were centrifuged, and the pellet was washed with Tris buffer (see "Experimental Procedures") and divided into five portions. Heparinase III (0.4 unit/ml, Sigma) and chondroitinase ABC were incubated 18 h 37°C (controls were incubated with buffer alone); heparin (5 mg/ml), EDTA (10 mM), and SDS (2%) were kept on ice for 30 min. Centrifugation gave supernatants that were applied to transferrin zymography after adjusting their heparin contents to 1 mg/ml (heparin enhances activity on the zymogram). The gels were scanned and analyzed by UVP/Gelbase program, and the results are in arbitrary density units Ϯ S.E. (n ϭ 4).
Heparin (5 mg/ml) and EDTA (10 mM) extracted maximum amounts of proMMP-7 and direct extraction with SDS also released most of the enzyme (Fig. 1C). EDTA presumably leads to enzyme unfolding because of removal of stabilizing calcium and zinc. SDS will unfold the enzyme and also swamp out positive charges that might bind heparin. SDS treatment following heparin extraction did not release any further enzyme. Heparinase III (heparitinase I) digestion released 80 -85% of the enzyme, but 15-20% was also released by chondroitin lyase ABC (Sigma), similar to results with chondroitin sulfates (Fig.  1A). During digestion, serine protease inhibitors, ZPCK, phenylmethylsulfonyl fluoride (0.1 mM), and 1 mM Zn 2ϩ (to inhibit MMPs) were used to inhibit proteolytic release of enzymes, and also blanks were incubated without heparinase III or chondroitinase (not shown). The results of the five treatments are consistent with the hypothesis that native proMMP-7 binds to sulfated proteoglycans, particularly to heparan sulfate.
Coexpression and Colocalization of MMP-7 and Heparan Sulfate on the Apical Surface of the Epithelial Layer of the Rat Uterus in Estrous Stage II-The physiological relevance of the extraction data was investigated by studying the expression and localization of heparan sulfate and MMP-7 in the hormon-ally regulated tissue of the rat uterus. The expression of MMP-7 in the estrous cycle was most prominent in the early proliferative (estrous II) and late secretory (dioestrous) stages and was restricted to the endometrial glandular and lining epithelial cells (Fig. 2). Its secretion is bidirectional, apical and basolateral, but is predominant on the apical surface and some appears to be in the lumen. Interestingly, the epithelial surface heparan sulfate appears predominantly in the same estrous stages as MMP-7 does (Fig. 2). MMP-7 and heparan sulfate are colocalized at the apical surface of epithelial cells lining the lumen (Fig. 3C). The hypothesis that heparan sulfate and MMP-7 are directly interacting was tested by washing tissue sections with heparin sulfate or digesting with heparinase III or chondroitinase ABC. Heparin sulfate completely washed MMP-7 from tissue sections (Fig. 3E), consistent with the tissue extraction experiments (Fig. 1C), but did not displace heparan sulfate as shown by the remaining positive stain (Fig. 3D). Greater than 80% of MMP-7 in tissue sections was released by heparinase III treatment (Fig. 3I); however, chondroitinase did not release appreciable amounts of MMP-7 from tissue sections (Fig. 3G). The combined action of the two enzymes completely removed MMP-7 (Fig. 3K). The results support the hypothesis Rat Recombinant proMMP-7 and Human Active MMP-7 Bind to Heparin-Agarose with High Affinity-Further investigation of the binding of rat proMMP-7 to heparin was facilitated by using recombinant methods to obtain large amounts of enzyme from E. coli (see "Experimental Procedures"). Human active MMP-7 lacking the propeptide (clone from Dr. S. Sha-piro, St. Louis) was also expressed in E. coli BL21(DE3) cells, purified, and refolded. Rat rproMMP-7 was bound (Ͼ95%) to heparin-agarose beads as described under "Experimental Procedures" and then washed with Tris buffer to remove unbound material. Stepwise increasing salt concentration from 0.2 to 2.0 M NaCl failed to elute the bound enzyme (Fig. 4A). Heparin (20 mg/ml) gave complete elution (Fig. 4A, lane 14); 2 mg/ml (lane 11) did not release all enzyme. The active form of rat rMMP-7 can be eluted by 0.6 -0.8 M NaCl and the human active form, by 0.4 -0.5 M (data not shown). EDTA also elutes rproMMP-7 efficiently but in a state that forms dimers and aggregates on SDS-polyacrylamide gel electrophoresis (data not shown).
K D Determined by Affinity Coelectrophoresis-Because rproMMP-7 undergoes spontaneous autoactivation at high concentrations, we prepared the E198Q mutant by the same methods described for wild type rproMMP-7. This mutation is known to reduce activity of the active form of the enzyme by 400-fold (23) and should retard self-activation. The purified enzyme showed no significant autolysis in 8 h at 24°C as shown by Western blot with the polyclonal antibody RM7-W, which reacts with all fragments of the enzyme (Fig. 4B, lanes 4  and 5). Zymography at the usual enzyme levels showed no activity (Fig. 4B, lanes 1 and 2). When this mutant enzyme was embedded at various concentrations in gel slabs and a radioiodinated heparin probe was electrophoresed through the gel, retardation of heparin by the enzyme was visualized by radiography (Fig. 4C). K D can be estimated from the protein concentration at which the heparin is half-shifted from being fully mobile to being maximally retarded. The K D is approximately 5-10 nM; at 10 nM rproMMP-7, the heparin is 80% retarded, and at 5 nM, heparin is fully mobile. Because of heterogeneity of the heparin, the migration front in Fig. 4C is fractionated into two populations: one binds tightly to MMP-7 and is retarded at the top of the gel, and the other appears half way down the lane; the excess unbound heparin migrates to the bottom half of the gel. All of these affinity retardation effects were eliminated by adding 500 g/ml of cold heparin as a competitor (data not shown).
First, various sulfated compounds can release the enzyme from homogenates of postpartum rat uterus; the most potent agents are heparin and heparan sulfate. Heparin is more effective, probably because of its higher degree of sulfation, and its action is comparable with direct solubilization of enzyme in 2% SDS. Heparin does not extract MMP-7 by competing for positively charged binding sites in the tissue because treatment with bacterial heparinase III, which can specifically degrade heparan sulfate, also releases the enzyme. Positively charged protamine supports this conclusion, it appears to release MMP-7 by competing for negatively charged binding sites as in the case of acetylcholinesterase release (25).
Second, confocal microscopy with immunofluorescence staining for heparan sulfate and MMP-7 in rat uterus shows colocalization of heparan sulfate proteoglycans and MMP-7, which is most pronounced at the apical ends of epithelial cells (Fig.  3C). The enzyme is released to the extent of 80% or more by heparinase III treatment and is completely released by a heparin wash of tissue sections. Matrilysin is typically expressed in glandular epithelial cells and carcinoma cells of epithelial origin (26) and in phagocytic monocytes (27). A possible candidate for MMP-7 docking is the GPI-anchored glypican (28) commonly found on the apical surfaces of epithelial cells. Other possibilities are epican or syndecan on the basolateral surface (29) and perlecan in the underlying basement membrane (30).
The staining for MMP-7 and for heparan sulfate was examined as a function of the rat estrous cycle. MMP-7 showed maximal staining at estrous, which compares with the maximal concentration of mRNA for MMP-7 at the same stage in the mouse (31). It is interesting that heparan sulfate staining is also maximal at estrous, where it appears concentrated in the basal lamina. There is also heavy diffuse staining at dioestrous. The only previous literature study of heparan sulfate proteoglycan changes in the cycling uterus showed that syndecan staining was strong at the basal side of mouse epithelial cells at estrous and basolaterally at proestrous (32). Our stain would detect any heparan sulfate chains regardless of the type of protein core.
Rat matrilysin zymogen form binds more tightly to heparin than the active form as shown by the elution of active MMP-7 from a heparin column at 0.7-0.8 M NaCl and the failure to elute proMMP-7 from the column even at 2 M NaCl (Fig. 4A). Binding to tissue sites requires the correct three-dimensional form of the enzyme: EDTA removes calcium, and possibly zinc, from the enzyme structure, releasing it from the tissue (Fig.  1C). Binding of proMMP-7 was quantified by affinity coelectrophoresis and the DNase II homogeneous binding assay. The former indicated a K D of 5-10 nM (Fig. 4C). The binding assay (Fig. 5) gave a 50% reversal of DNase II inhibition by heparin at a concentration of 1 M proMMP-7. Collagen binds even tighter, although the reason is not clear because literature values for collagen K D are 60 -80 nM (33). Perhaps collagen also interacts directly with DNase II to interfere with its heparin binding. Two putative binding peptides were identified based on the analysis of the rat MMP-7 sequence by use of the 18-residue amphipathic helical wheel diagram (34). These peptides, one each from near the C and N termini of proMMP-7, bound to heparin, but with 10 and 100 times less avidity than proMMP-7. This indicates that the strength of binding of proMMP-7 to heparin is not explained by the Arg/Lys residues in these two segments considered separately.
The zymogen forms of rat MMP-2, -9, and -13 and active MMP-2 were also extracted from the uterus by heparin, although none bound as tightly as proMMP-7 (data not shown). It was noted long ago (35) that heparin and dextran sulfate could extract MMP-13 from rat bone. Heparin-Sepharose TM has frequently been used to purify MMPs. Active enzyme can be eluted with NaCl as follows: MMP-1, 0.  (41). No one has studied both forms of one enzyme. There is a wide variation in affinity for heparin, and not all MMPs can bind. It is further known that MMP-1, -2, and -9 bind heparin through their hemopexin domains (42)(43)(44), and this promotes interactions between two MMPs during proenzyme activation. Therefore, the docking to heparan glycosaminoglycans is not restricted to MMP-7, but rat proMMP-7 is one of the most tightly binding MMPs.
Molecular modeling was used to examine how the distribution of positive residues might account for the tight binding of rat active MMP-7 and why it binds tighter than human MMP-7. X-ray structure is available only for the active human form, so the model sheds no light on the contribution of the rat propeptide to binding. As seen in Fig. 7 human active MMP-7 shows Arg and Lys residues only at the periphery of the enzyme when viewed with the active site at the center (Fig. 7A). However, the reverse side (Fig. 7, B and C) shows a double track of Arg/Lys residues that resembles the cradle for heparin binding seen with fibronectin module III-13 (45). There is a further cluster of three Arg residues to the "north" of this track. The rat enzyme has a similar number of positive residues, but the distribution is somewhat different; there is a long continuous line of positive residues involving Lys 260 , Lys 264 , Arg 139 , Lys 135 , Arg 127 , Lys 199 , and Arg 165 , plus additional residues to the right side that produce a partial double track (Fig. 7, E and F; note that residue numbers are adjusted to match the human collagen sequence (46)). The x-ray picture stops at residue 264 because the remaining seven residues are not fixed, but the next residue in the rat is Arg 265 and is likely to fall between 264 and 139. These anionic residues involve many different strands of the folded molecule rather than a single amphipathic helical motif found in many heparin-binding proteins (34). A surprisingly similar distribution of basic residues is found in the primary sequence of cobra cardiotoxin M3, which binds heparin with K D ϭ 20 nM; these basic residues are distributed over a number of ␤ strands (47). The similarity extends over 57 residues: KX 2 KX 6 KX 16 KKX 13 KX 15 K for M3 and KX 3 RX 4 RX 17 KRX 13 KX 15 R for rat MMP-7 residues 108 -165 (Browner numbering); there are additional basic residues within each sequence that do not match. Fig. 7 does not illustrate the Asp and Glu residues; however, in the tracks discussed here there is only one Asp residue (131) lying to the left of Lys 151 and potentially forming a salt bridge. The rat MMP-7 propeptide, which is expected to lie on the "front" of the molecule overlying the active center, has a further 12 Arg/Lys residues, some of which probably interact with heparin (as in Fig. 5), so that it is FIG. 7. Computer modeling of the active form of human and rat matrilysin. The 1.8 Å coordinates for human MMP-7 catalytic domain from the Protein Data Bank were viewed with the Rasmol program. The rat enzyme is 70% identical to the human; it was modeled by inserting its Arg and Lys residues in place of human residues of the same number and by replacing any human Arg or Lys residues not found in rat with Ala. The numbering of residues starts from Met-1 of the signal peptide and is further adjusted by adding 2 to the residue number to match the collagen structure (46); thus the N-terminal Tyr 98 of the active enzyme becomes 100. The rat numbering used elsewhere in this paper numbers the corresponding rat Tyr residue 78. A, human matrilysin: viewed with active center HEXXHXXXXXH at the center of the field. B, opposite face of molecule turned 180°around vertical axis and up 30°about the horizontal axis. C, opposite face but rotated down 90°from B about the horizontal axis. D-F, corresponding views of the model of rat MMP-7. Yellow, Arg or Lys; Purple, His; Red, Glu at active center. Blue numbers, human sequence position of Arg and Lys and rat residues corresponding to the human in position and charge. Red numbers, positively charged residues in rat that do not correspond exactly to the human sequence (five residues, typically displaced by three or four positions). In addition, there are three (human) or two (rat) positively charged residues in the C-terminal peptide that cannot be visualized by x-ray crystallography. much more difficult to dislodge the proenzyme than the active enzyme.
The x-ray structures (not shown) of the catalytic domains of human MMP-1, MMP-2, and MMP-13 show similar putative heparin-binding cradles, but none is so well organized as that of MMP-7. Jeffrey (48) comments on the high affinity of both active and latent rat MMP-13 for heparin, in the absence of obvious binding motifs. However, we note that MMP-13 has a similar number of Lys/Arg residues as MMP-7 and that 7 of 11 in the catalytic domain and 6 of 12 in the propeptide are exact positional matches (49).
Does heparin binding have further effects on MMP-7 in addition to anchoring it in the extracellular space? Several enzymological roles might be postulated. Modeling suggests that heparan sulfate chains might wrap around the proenzyme. This could protect the enzyme from exposure to activating proteases, or it might exert a force on the propeptide that would tend to pull it away from the active center, favoring activation. Heparin greatly enhances the activity of proMMP-7 in zymography (Fig. 6). This could be due to facilitating refolding after SDS removal, enhancing enzyme activity, or helping to anchor the protein in the gel so that it is not eluted during overnight incubation. We find that adding heparin to refolded rproMMP-7 enhances stability during prolonged storage. We suggest that heparin may have an important effect on MMP-7 stability, activation, and enzymatic activity in vivo, but this point remains to be tested.
Confocal microscopy indicates that MMP-7 is anchored to heparan sulfates on cell surfaces and possibly in the basement membrane in vivo. Such anchoring of proMMP-7 to either cell surface or nearby basement membrane would retain the enzyme near the cell while it awaits proteolytic activation and the binding of the pro-and active forms would prevent loss of enzyme by diffusion and permit continuing control by the cell. Elaboration of these points is presented in the Introduction.
More importantly, if the enzyme is anchored to the cell surface, this will permit the cell to focus proteolysis on a particular region next to the cell. The role of pericellular proteolysis has been reviewed (50). Proteolysis of the extracellular matrix and the cell surface molecules is a critical requirement in processes such as morphogenesis, alteration of cell function, migration of cells, tissue repair, tumorigenesis, and cell death. These processes are best controlled if the proteases are on the cell surface. One function of binding MMPs to heparan sulfate might be the formation of invadopodia, specialized extensions used by the cell to move in a specified direction through the matrix (51). By bringing together proenzymes and their activators in an organized fashion and in high concentration, proteolysis at the cell surface can proceed even in the presence of high concentrations of inhibitors. MMP-7 may have specialized functions in its main location, the apical region of glandular epithelial cells. Here it may serve to maintain patency of the glandular lumen, to provide host defense, and to activate or turn over cell surface proteins. On the other hand, in matrix remodeling, cell turnover, and injury response MMP-7 secretion may be basal (26). In either case, anchoring of the enzyme to the appropriate surface of the cell or to the underlying basement membrane may be critical in controlling these processes.
If MMPs such as MMP-1, -2, -7, -9, and -13 are bound to heparan sulfate in the matrix or on the cell surface, this could have important therapeutic implications; dislodging enzyme from cells or matrix could attenuate destruction of tissue. The sulfated compounds pentosan polysulfate and suramin have been used to treat osteoarthritis and cancer in animals (52,53).
Their activity has been attributed to various effects, in particular to reducing MMP activity or blocking cell surface proteases such as guanidinobenzoatase. However, the present study suggests that dislodging enzymes from the tissue might be a mechanism of action. One should not be misled by the high concentrations of heparin used in the present study (2-4 mg/ ml) because only 30 min was allowed for elution in the cold. In the study of MMP-13 binding to heparin columns, it was possible to get good elution with prolonged washing using only 5 nM dextran sulfate (17). In conclusion, this study emphasizes the importance of considering MMP anchoring in the regulation of processes of tissue remodeling.