Cooperative binding of Ca2+ to human interstitial collagenase assessed by circular dichroism, fluorescence, and catalytic activity.

Dissociation of Ca2+ from human interstitial collagenase induced either by chelation with EGTA or by dilution resulted in loss of enzyme activity, a red shifted emission maximum from 334 to 340 nm and quenching of protein fluorescence by 10% at 340 nm. Circular dichroism indicated that secondary structure was unaffected by EGTA. Ca2+ binding to the EGTA-treated enzyme as assessed by fluorescence was cooperative (Hill coefficient, 2.9; 50% saturation at 0.4 mM Ca2+). The dependence of catalytic activity on [Ca2+] was also cooperative (Hill coefficient, 1.7-2.0; midpoint [Ca2+], 0.2 mM). The Ca2+-reconstituted protein was indistinguishable from the untreated enzyme by activity and fluorescence measurements. These results demonstrate that removal of Ca2+ from full-length collagenase generates a catalytically incompetent, partially unfolded state with native secondary structure but altered tertiary structure characterized by exposure of at least one tryptophyl residue to a more polar environment.

Dissociation of Ca 2؉ from human interstitial collagenase induced either by chelation with EGTA or by dilution resulted in loss of enzyme activity, a red shifted emission maximum from 334 to 340 nm and quenching of protein fluorescence by 10% at 340 nm. Circular dichroism indicated that secondary structure was unaffected by EGTA. Ca 2؉ binding to the EGTA-treated enzyme as assessed by fluorescence was cooperative (Hill coefficient, 2.9; 50% saturation at 0.4 mM Ca 2؉ ). The dependence of catalytic activity on [Ca 2؉ ] was also cooperative (Hill coefficient, 1.7-2.0; midpoint [Ca 2؉ ], 0.2 mM). The Ca 2؉ -reconstituted protein was indistinguishable from the untreated enzyme by activity and fluorescence measurements. These results demonstrate that removal of Ca 2؉ from full-length collagenase generates a catalytically incompetent, partially unfolded state with native secondary structure but altered tertiary structure characterized by exposure of at least one tryptophyl residue to a more polar environment.
The matrix metalloproteinases mediate remodeling of extracellular matrix in healthy and diseased tissue (1,2). Sequence analysis of the MMPs, 1 also referred to as the metzincin proteinase family (3), reveals a characteristic domain structure. The smallest MMP, matrilysin, consists of 19-kDa catalytic domain with an active site zinc ion coordinated within a conserved HEXXHXXGXXH motif. Proteolytic latency is maintained by an N-terminal propeptide of approximately 10 kDa in which a conserved cysteine is coordinated to the catalytic zinc (4). The larger members of the family, including interstitial collagenases, stromelysins, and gelatinases, possess additional functional and structural domains. In particular, collagenases have a C-terminal domain of about 30 kDa that exhibits sequence homology with hemopexin and vitronectin (5). This so-called pexin domain is essential for the expression of collagenolytic activity; without it, the enzyme retains general peptidase activity but does not hydrolyze native collagen (6). Matrilysin and mutant forms of collagenase and stromelysin possessing only a catalytic domain contain a second zinc ion (7)(8)(9). This structural zinc is absent from the full-length forms of stromelysin-1 and gelatinase A (10).
Calcium ions are also an essential components of the MMPs (11,12). Crystallographic analysis of the catalytic domain reveals three bound Ca 2ϩ in fibroblast collagenase (13,14) and two in the neutrophil enzyme (15). Crystallography of fulllength porcine MMP-1 also shows three Ca 2ϩ in the catalytic portion and two Ca 2ϩ in the pexin domain (16). Ca 2ϩ undoubtedly binds to similar sites in the pexin domain of human collagenase because the binding sites are conserved. Previous studies suggest that Ca 2ϩ stabilizes an active conformation of the MMP that is more stable to denaturants (17) and less susceptible to proteolysis (18). However, the precise stereochemical basis for Ca 2ϩ stabilization is unknown because a three-dimensional model of a Ca 2ϩ -free MMP is not available for comparison.
To provide additional insight into the relationship between MMP structure and calcium binding, we correlated the effect of Ca 2ϩ on catalytic activity and protein structure using intrinsic protein fluorescence and circular dichroism as structural indicators. Ca 2ϩ concentration was manipulated either by the chelating agent EGTA or by dilution. EGTA was chosen because its affinity for Ca 2ϩ is comparable with that of EDTA, 2 although its affinity for Zn 2ϩ is much lower than that of EDTA (19). The following data document that Ca 2ϩ binding to collagenase is cooperative and associated with changes in tertiary but not secondary structure. Furthermore, these changes were independent of the method used to alter calcium concentration.
Methods-MMP activity was measured with DnpS, TPS, or collagen as substrate. Hydrolysis of DnpS by rMMP-1 was assessed by high pressure liquid chromatography as described (22). Hydrolysis of TPS by HUVEC MMP-1 was measured in a continuous spectrophotometric * This study was supported in part by National Institutes of Health Grant AM 39733. The CD spectrophotometer was purchased with support from National Science Foundation Grant BIR-91-19404 and funds from the Graduate School of the University of Louisville. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ Recipient of an interdisciplinary predoctoral fellowship from the Graduate School of the University of Louisville.

RESULTS
To assess the effect of Ca 2ϩ binding on rMMP-1 secondary structure, we recorded the low UV CD spectrum of proMMP-1 in the presence and the absence of the chelating agent, EGTA. As shown in Fig. 1A, EGTA had no effect on the CD spectrum of the enzyme. The addition of Ca 2ϩ to the EGTA-treated protein also elicited no observable change in CD spectrum. Similar results were obtained with the recombinant MMP-1 (data not shown). Thus, we conclude Ca 2ϩ binding to MMP-1 resulted in no global changes in secondary structure.
We also measured the emission spectrum of rMMP-1 in the presence of 5 mM CaCl 2 before and after the addition of 6 mM EGTA to assess Ca 2ϩ -dependent changes in tertiary structure. Fig. 1B shows that the addition of EGTA resulted in decreased fluorescence intensity of 10% at 340 nm and a red shift in the emission maximum from 334 to 340 nm (compare spectra 1 and 2). Similar changes in emission intensity were observed when HUVEC procollagenase was diluted into Ca 2ϩ -free buffer (Fig.  1B, spectrum 4) and then supplemented with added Ca 2ϩ (Fig.  1B, spectrum 3). This suggests that the fluorescence changes in spectra 1 and 2 result from dissociation of Ca 2ϩ induced by the chelating agent. Lack of an obvious shift in emission maximum in comparing spectra 3 and 4 may result from the presence of the propeptide. This portion of the protein contains one tryptophan, which could have obscured the shift in emission maximum associated with Ca 2ϩ dissociation from the enzyme lacking the propeptide.
The emission characteristics of the quenched residue (shown in Fig. 1B, spectrum 5) were assessed by subtracting the spectrum measured in the presence of EGTA (spectrum 2) from that obtained in its absence (spectrum 1). The maximum in the difference spectrum 5 at 320 nm indicates that the quenched fluorophores are predominantly tryptophyl residues that reside in a hydrophobic environment (28).
The contribution of tyrosyl residues to rMMP-1 fluorescence was ascertained by subtracting the emission spectrum obtained with excitation at 297 nm (where Trp alone absorbs) from the spectrum obtained with excitation at 280 nm (Trp and Tyr absorb). The resulting difference spectra (Fig. 2) allow comparison of the emission properties of tyrosyl residues in the Ca 2ϩ -free and Ca 2ϩ -bound forms of the enzyme. Each spectrum exhibits a maximum near 315 nm, which is strongly red shifted from that expected for tyrosine in aqueous solution (28). A small decrease in emission intensity was observed, suggesting that the Ca 2ϩ -dependent conformational differences are sensed by tyrosyl as well as tryptophyl residues.
Titration data illustrating the dependence of fluorescence and catalytic activity on [Ca 2ϩ ] are shown in the Hill plots of Fig. 3. Ca 2ϩ binding as assessed by changes in fluorescence was cooperative (Hill coefficient, 2.9; 50% saturation at 0.43 mM Ca 2ϩ ). When the Ca 2ϩ dependence of rMMP-1 activity was assayed with a peptide substrate, the resulting Hill plot also indicated cooperative Ca 2ϩ binding (Hill coefficient, 1.7; 50% saturation at 0.23 mM [Ca 2ϩ ]). When the [Ca 2ϩ ] was manipu-  2. Effect of EGTA on the fluorescence emission of tyrosyl residues of rMMP-1. Tyrosyl emission was estimated by subtracting the contribution of the tryptophyl residues from the total emission. To accomplish this, emission spectra of rMMP-1 in the presence (ϩ) and the absence (Ϫ) of 6 mM EGTA were recorded with excitation either at 280 nm to excite both tyrosyl and tryptophyl residues or at 297 nm to excite tryptophyl residues alone. The pairs of spectra (ϮEGTA) were normalized at 380 nm, where emission is solely from tryptophan. The spectra obtained with excitation at 297 nm were subtracted from those obtained with excitation at 280 nm to yield the tyrosyl difference spectra. Experimental conditions were the same as described in the legend to Fig. 1. lated by dilution rather than EGTA, cooperative Ca 2ϩ binding to native HUVEC MMP-1 was also observed (Hill coefficient, 2.0; 50% saturation at 0.20 mM Ca 2ϩ ). Because similar results were obtained whether [Ca 2ϩ ] was altered by EGTA or dilution, we conclude that the effects of the chelating agent result from Ca 2ϩ dissociation from the enzyme rather than removal of catalytic Zn 2ϩ .
To determine if the effects of Ca 2ϩ on activity and fluorescence were reversible, we compared the catalytic activity of the recombinant enzyme with DnpS and collagen prior to Ca 2ϩ removal, after Ca 2ϩ removal with EGTA, and after reconstitution with Ca 2ϩ . Previous reports (11,12) indicate that prolonged incubation of collagenase with EDTA resulted in removal of zinc and irreversible denaturation of the protein. The data in Table I show that activity against both DnpS and collagen was completely lost on treatment with EGTA. Loss of activity and fluorescence changes were reversed by adding Ca 2ϩ , even 24 h after removal of the metal ion.

DISCUSSION
The present studies, conducted with full-length rMMP-1 that includes both the catalytic and pexin domains, demonstrate that acquisition of catalytic activity by collagenase is coupled to protein conformational changes induced by Ca 2ϩ binding. Binding was cooperative as assessed by both structural and activity changes. The observation that a lower Ca 2ϩ concentration is required for catalytic activity compared with development of the fluorescence changes suggests that fewer binding sites influence catalysis, whereas the fluorescence changes involve sites other than those necessary for catalytic efficiency. Our studies with the full-length protein extend previous studies conducted with the catalytic domain alone. Lowry et al. (17) showed that both Ca 2ϩ and Zn 2ϩ stabilized a recombinant collagenase catalytic domain against denaturation by GdnHCl, and Housley et al. (18) reported that Ca 2ϩ stabilized the catalytic domain of recombinant stromelysin against heat denaturation. In addition, activation of prostromelysin-1 by an organomercurial in the presence of 0.1 mM Ca 2ϩ resulted in autolysis that could be prevented with 5 mM Ca 2ϩ (18). These data indicate that Ca 2ϩ stabilizes a compact protein structure that is less susceptible to denaturation and proteolysis. Our results show that Ca 2ϩ binding to full-length rMMP-1 results in local conformational change(s) that do not detectably alter the secondary structure of the protein but do change tertiary structure as reflected in the environment of tryptophyl and tyrosyl residues. The change in tertiary structure in the absence of secondary structural changes suggests that Ca 2ϩ removal may produce a "molten globule-like" structure (20).
The emission maximum of denatured MMP-1 is at 356 nm (20), characteristic of tryptophan in water, whereas the emission maximum of the Ca 2ϩ -deficient enzyme shifts only to 340 nm. Thus the change in tryptophyl environment brought about by Ca 2ϩ release is less drastic than complete unfolding. In our study of GdnHCl-induced denaturation of rMMP-1, we observed a folding intermediate with an emission maximum at 340 nm at 1 M denaturant (20). However, it is unlikely that the Ca 2ϩ -free form of MMP-1 in the present study is the same as this intermediate because in 1 M GdnHCl, there was an increase, rather than a decrease, in emission intensity.
MMP-1 contains one tryptophyl residue in the propeptide, three in the catalytic domain and four in the pexin domain. Assignment of the fluorescence changes to particular residues is of course impossible from the data at hand. Lovejoy et al. (13) suggested that in 19-kDa collagenase one of the calcium ions and the structural zinc may stabilize a surface loop by fastening it to two ␤ strands. Neither the loop nor the two strands contain tyrosyl or tryptophyl side chains that might be directly affected by Ca 2ϩ . Thus, it is likely that any conformational change induced by Ca 2ϩ binding at this site is propagated to other regions of the structure. The two Ca 2ϩ in the pexin domain of pig collagenase also appear to link structural domains together: ␤-sheet-1 to sheet 3 and ␤-sheet-4 to sheet 2 (16). Two of the four pexin tryptophyl residues are located within these sheets; Trp 349 is in sheet 2, and Trp 398 is in sheet 3. Thus, it is reasonable to assign at least a portion of the Ca 2ϩ -dependent fluorescence change to them. In addition, both tryptophyls are close in the primary structure to tyrosyl residues [YW 349 A and YW 398 RY (29)]. Variation in the relationship between these groups should influence the efficiency of excitation energy transferred from tyrosyl to tryptophyl residues. However, the quenching of tyrosyl emission observed on Ca 2ϩ

TABLE I
Effect of EGTA and Ca 2ϩ on the activity of rMMP-1 rMMP-1 (0.2 M) in 150 mM Tris-Cl, 200 mM NaCl, 0.05% Brij 35, 5 mM CaCl 2 was treated with EGTA under the indicated conditions. Fluorescence emission spectra were recorded, after which aliquots of the enzyme solutions were assayed with either collagen or DnpS as substrate under the same conditions. The emission spectra were identical to those in Fig. 1  dissociation was accompanied by decreased, rather than increased, tryptophyl emission, as would be expected if altered transfer efficiency were the sole basis for the altered fluorescence.