Mechanism of Ca2+-dependent activity of human neutrophil gelatinase B.

Progelatinase B can be activated in vitro by organomercurial compounds and by proteolytic enzymes such as trypsin, chymotrypsin, and stromelysin. Activation of the proenzyme by either 4-aminophenylmercuric acetate or chymotrypsin yielded proteins that absolutely required Ca2+ for activity, regardless of the pH of the reaction mixture. The trypsin- and stromelysin-activated gelatinases, on the other hand, did not require Ca2+ for activity at pH 7.5, but the activity of the trypsin-activated enzyme became Ca2+ dependent as the pH increased. The pH study revealed that an amino acid residue with an apparent pKa of 8.8 was involved in this process. The NH2-terminal analyses showed that trypsin- and stromelysin-activated enzymes had the same NH2 termini (Phe88), but 4-aminophenylmercuric acetate- and chymotrypsin-activated enzymes had Met75 and Gln89 or Glu92 as the NH2-terminal amino acid, respectively. These data, in conjunction with the x-ray crystal structure of collagenase, suggest that a salt linkage involving Phe88 is responsible for the Ca2+-independent activity of trypsin- and stromelysin-activated gelatinase. Replacing Asp432 in progelatinase with either Glu, Asn, Gly, or Lys resulted in the proteins that, upon activation by trypsin, required Ca2+ for activity. These substitutions did not significantly affect Km for the synthetic substrate but decreased the kcat and increased the half-maximal Ca2+ concentration required for enzyme activity (KCa) by severalfold. The effects on kcat and KCa depended on both charge and size of the side chains of the substituted amino acids. The decrease in kcat correlated well with the increase in KCa of the mutants. The orders of decrease in kcat and increase in KCa were wild type >/= D432E > D432N > D432G > D432K and wild type </= D432E < D432N < D432G < D432K, respectively. These data suggest that in trypsin- or stromelysin-activated enzyme, the NH2-terminal Phe88 forms a salt linkage with Asp432, rendering the enzyme Ca2+ independent. Ca2+ affects catalytic activity of the 4-aminophenylmercuric acetate- and chymotrypsin-activated enzymes by substituting for the salt linkage and interacting with Asp432. This interaction generates a similar, if not identical, conformational change to that generated by the salt linkage in the protein, leading to catalysis.

Matrix metalloproteinases (MMP) 1 are a family of zinc en-dopeptidases that have the ability to mount a concerted degradative attack on virtually all components of the extracellular matrix (1)(2)(3)(4). It is widely assumed that normal physiological processes depend on careful spatial and temporal regulation of the activities of these enzymes. The uncontrolled expression of MMP has been reported to be associated with many pathological processes such as angiogenesis, rheumatoid arthritis, and tumor metastasis (5)(6)(7)(8)(9)(10). These proteinases are often divided into three subgroups based on their substrate specificities: collagenases, gelatinases, and stromelysins. Gelatinase B, which specifically degrades types IV, V, and XI collagens as well as denatured collagens (gelatins), is believed to play an important role in both physiological and pathological processes. Several researchers have recently shown that cells bearing the gelatinase B cDNA are able to metastasize in nude mice, whereas inhibition of the gelatinase activity prevents metastasis (11)(12)(13)(14)(15), thereby providing direct evidence that this enzyme is involved in the destruction of the basement membrane associated with invasion and metastasis.
Gelatinase B activity is controlled at several levels (2,4), including proenzyme activation. Progelatinase B isolated from various cell lines is activated by autoproteolytic processing of both NH 2 -terminal propeptide and COOH-terminal hemopexin domain. Activation can be initiated in vitro by either a reaction with an organomercurial compound, such as 4-aminophenylmercuric acetate (APMA), or by limited proteolysis by other proteases such as trypsin, chymotrypsin, and stromelysin (16 -18). The autoprocessing of gelatinase B in the presence of APMA occurs with a stepwise truncation of the NH 2 -terminal propeptide to Met 75 , followed by a Ca 2ϩ -dependent loss of COOH-terminal domain (19), apparently to Ala 506 , and generates a 70-kDa active species (16). Activation of the enzyme by trypsin, however, generates Phe 88 as the amino terminus (19). It is noteworthy that the presence of amino acid residues 75-87, corresponding to the Met 75 -Arg-Thr-Pro-Arg-Cys-Gly-Val-Pro-Asp-Leu-Gly-Arg sequence in the APMA-activated gelatinase (E a ), renders the activity of the enzyme dependent on Ca 2ϩ , whereas removal of this fragment by trypsin abolishes the Ca 2ϩ dependence of the enzyme activity (19). This finding suggests that Ca 2ϩ affects the enzyme activity by interaction with this fragment. Ca 2ϩ has also been found to be indispensable for activity of other members of the MMP family (20 -23). As early as 1975, Seltzer et al. (20) demonstrated that Ca 2ϩ plays a dual role in collagenase catalysis. It acts as an activator (20) and a stabilizer (20,21) of collagenase and a stabilizer of stromelysin (22,23) at physiological temperature and pH. Recent crystallographic studies of the human collagenases (24 -27) have identified at least three Ca 2ϩ -binding sites. One of the Ca 2ϩ sites is sandwiched between the surface S-shaped double loop Arg 145 to Leu 160 and the surface of the ␤-sheet. Because this loop structure is rather compact, it is assumed that this Ca 2ϩ contributes to the stability of the collagenase structure. However, we have recently shown that Ca 2ϩ is only required for the stability of the trypsin-activated gelatinase B (E t ), whereas it is involved in both stability and activity of E a (19). These observations led us to propose that Ca 2ϩ affects gelatinase B catalysis by binding to the NH 2 -terminal region of E a , inducing a conformational change in the protein and rendering the enzyme active (Ref. 19; Fig. 1, A1 or A2). It has been suggested that in the catalytic domain of neutrophil collagenase, the presence of Phe 79 at the NH 2 terminus stabilizes the active-site conformation by forming a salt linkage with Asp 232 at the COOH-terminal region (27). Because the NH 2 -terminal amino acid of E t is Phe 88 (corresponding to Phe 79 in collagenase), one can envision that a similar salt linkage between Phe 88 and Asp 432 (corresponding to Asp 232 in collagenase) in gelatinase mimics the role of Ca 2ϩ in stabilizing the active-site conformation ( Fig. 1, T2).
In the present report, we have extended our previous study (19) and clearly demonstrated that a salt linkage between Phe 88 and Asp 432 exists in E t . The presence of the salt linkage is responsible for Ca 2ϩ -independent activity and not for stability of gelatinase. In addition, our study provides evidence to suggest that the side chain of Asp 432 can also ligand to Ca 2ϩ in the absence of the salt linkage, thereby generating a catalytically competent enzyme.
Mutagenesis, Expression, and Purification of Recombinant Progelatinase B-Mutagenesis of the gelatinase B cDNA (pETNG; Ref. 30) was performed by polymerase chain reaction, as described previously for creating the Asp 432 to Gly mutation (28). To substitute Glu, Asn, or Lys for Asp 432 , a degenerate mutagenic oligonucleotide, 5Ј-TTCACT-GAGGGGCCCCCCTTGCATAAGA(G)AG(T)GACGTGAATGGCATC-3Ј was designed to mutate the codon for Asp 432 (GAC) to that for Glu (GAG), Asn (AAT), or Lys (AAG). The underlined sequence is the mutated codons. All mutant cDNA was sequenced to verify that the desired substitution was the only one generated by polymerase chain reaction. Plasmids encoding for wild-type (WT) and mutant enzymes were introduced into Escherichia coli strain BL21(DE3)(pLys), the cells bearing the plasmids were induced by isopropyl ␤-D-thiogalactopyranoside, and the recombinant proteins were purified as described (30). Protein concentration was determined by the Bradford dye binding techniques (a standard Bio-Rad assay), using bovine serum albumin as a standard.
Activation of Recombinant Progelatinase B-WT and each mutant enzyme (ϳ15 g/ml) were activated in 50 mM Tris-HCl, 0.1 M NaCl, 0.5 M ZnCl 2 , and 5 mM CaCl 2 , pH 7.5 (activation buffer) either by APMA (1.0 mM) for 16 h, by trypsin (20 g/ml) for 4 h, by chymotrypsin (20 g/ml) for 4 h, or by stromelysin (1 unit/ml, 1 unit of stromelysin degrades 1 M fluorogenic peptide, Mca-PLGL(Dpa)AR-NH 2 , per min at 37°C) for 4 h at 37°C. Trypsin was removed by passage through a soybean trypsin inhibitor column (2 ml). Chymotrypsin-or stromelysinactivated enzyme was separated from chymotrypsin or stromelysin on a gelatin-agarose affinity column (1 ml) at 4°C. The column was washed thoroughly with the activation buffer to remove the activators. The activated gelatinase was then eluted with the activation buffer containing 15% dimethyl sulfoxide. Ca 2ϩ was removed from the activated enzymes, as described previously (19).
Enzyme Assays-Gelatinolytic activity of activated WT and mutants was determined by incubating the enzymes (1-23 g/ml) with 100 g of [ 14 C]gelatin (77,000 cpm/mg gelatin) in a reaction mixture containing 10 mM HEPES-NaOH, pH 7.5, 0.1 M NaCl, 0.5 M ZnCl 2 , and 0 -1.0 mM CaCl 2 in a final volume of 150 l at 37°C for 5 min to 2 h, as described (31). Duplicates were performed for each assay. The effect of pH on gelatinolytic activity of the enzymes was determined using the following buffers for the indicated pH ranges: 50 mM HEPES, pH 7.35-7.80; 50 mM CHES, pH 8.25-9.24; and 50 mM CAPS, pH 9.77.
The kinetic parameters of activated WT and mutant enzymes were obtained by assaying the enzymes (0.02 M) at 37°C against 0.5-20 M fluorogenic peptide, Mca-PLGL(Dpa)AR-NH 2 , as described previously (32). The initial rates of substrate hydrolysis were determined by measuring the time-dependent increase in fluorescence ( ex , 328 nm; em , 393 nm) in a reaction mixture containing 10 mM HEPES-NaOH, pH 7.5, 0.1 M NaCl, 0.5 M ZnCl 2, and 1 mM CaCl 2 in a final volume of 150 l. The assay was performed using an Aminco⅐Bowman Luminescence Spectrometer (SLM Instruments, Inc., Urbana, IL).

FIG. 1. Proposed mechanisms for
Ca 2؉ -dependent catalysis of gelatinase B. A, APMA activates progelatinase B in the presence of Ca 2ϩ , generating an active species (E a ) having NH 2 terminus Met 75 that still requires Ca 2ϩ for activity; A1, Ca 2ϩ binding to E a induces a conformational change in the NH 2 -terminal segment of E a , relieving its autoinhibition; A2, Ca 2ϩ binding to E a induces a conformational change and stabilizes the active site by interaction with the NH 2 -and COOH-terminal residues; T, trypsin activates progelatinase B in the presence of Ca 2ϩ , generating an active enzyme (E t ) having NH 2 terminus Phe 88 that does not require Ca 2ϩ for activity; TЈ, removal of the NH 2 -terminal segment of E a by trypsin generates E t ; T1, E t does not require Ca 2ϩ for activity due to permanently relieving the autoinhibition of the NH 2 -terminal segment of E a ; T2, E t does not require Ca 2ϩ for activity due to stabilizing the active site by a salt linkage between NH 2 -and COOH-terminal residues (dotted line).
Enzyme Stability-Stabilities of activated WT and mutants (2-5 g/ml) were determined by preincubating each enzyme species at 37°C in 10 mM HEPES (pH 7.5) or CAPS buffer (pH 9.8) containing 0.1 M NaCl and 0.5 M ZnCl 2 in the absence or presence of 1 mM Ca 2ϩ for various times (0 -60 min). The samples were then adjusted with 0.5 M HEPES (pH 7.5) and 10 mM Ca 2ϩ to a final concentration of 50 and 1 mM, respectively. The activities remaining were assayed using [ 14 C]gelatin or fluorogenic peptide as described above.
Other Methods-Gelatin zymography and NH 2 -terminal sequence analyses were described previously (19). The apparent affinity for Ca 2ϩ (K Ca ) and Michaelis-Menten kinetic parameters (K m and k cat ) of WT and each mutant enzyme were determined from Lineweaver-Burk plots using the ENZYME KINETICS program, version 1.0 (Trinity Software, Campton, NH).

Appearance of the NH 2 -terminal Phe 88 Correlates with the Loss of Ca 2ϩ Dependence of Gelatinase B Activity-We have
shown previously that APMA-activated gelatinase B (E a ) absolutely depends on Ca 2ϩ for activity, but the activity of trypsinactivated enzyme (E t ) is Ca 2ϩ independent (19). Because gelatinase B can be activated by other proteases, such as chymotrypsin and stromelysin (17), the effect of these activators on Ca 2ϩ -dependent activity of gelatinase was assessed. The activity of gelatinase B that had been activated by various activators (APMA, trypsin, chymotrypsin, and stromelysin) was determined by assaying the enzyme against [ 14 C]gelatin in the absence or presence of 1 mM Ca 2ϩ .
As shown in Table I, chymotrypsin-activated gelatinase B, similar to E a , had negligible activity in the absence of Ca 2ϩ . They were able to catalyze the degradation of gelatin substrate only in the presence of 1 mM Ca 2ϩ . Trypsin-or stromelysinactivated gelatinase B, on the other hand, exhibited significant activity in the absence of Ca 2ϩ (78 and 73% of maximal activity, respectively). Ca 2ϩ -independent activity of E t was not due to trypsin contamination, because the activity of the enzyme was inhibited by a zinc chelating agent, 1,10-phenanthroline (19). It has been demonstrated that stromelysin requires Ca 2ϩ for activity (21); thus, the possibility that Ca 2ϩ -independent activity of stromelysin-activated gelatinase B resulted from the contaminated stromelysin is unprecedented. Zymographic analyses of gelatinase B activated by each activator showed that the 57.5-kDa unglycosylated progelatinase B was converted to a 41.5-kDa species by APMA ( Fig. 2A, lane 2) and to a 40-kDa species by trypsin, chymotrypsin, or stromelysin ( Fig.  2A, lanes [3][4][5]. This finding agrees with the result obtained using the gelatinase B (92-kDa gelatinase) from HT 1080 human fibrosarcoma cells (17). The NH 2 -terminal sequence analyses of the activated species (Table I and Fig. 2B) indicated that gelatinase B activated by APMA, trypsin, chymotrypsin, or stromelysin had Met 75 , Phe 88 , Gln 89 , or Glu 92 , or Phe 88 as its NH 2 terminus, respectively. It is interesting to note that both the trypsin-and stromelysin-activated species that did not require Ca 2ϩ for activity had the same NH 2 terminus, Phe 88 , whereas APMA and chymotrypsin-activated enzymes that depended on Ca 2ϩ for activity displayed different NH 2 termini. These observations suggest that the loss of Ca 2ϩ requirement for the enzyme activity is associated with the presence of Phe 88 at the NH 2 terminus of the activated enzyme.
Effect of pH on Ca 2ϩ Dependence of Trypsin-activated Gelatinase B-X-ray crystal structure of the catalytic domain of collagenase (27) has shown that the ammonium group of the NH 2 -terminal Phe 79 makes a salt linkage with the side chain carboxylate group of Asp 232 to stabilize the active site. Although the x-ray structure of gelatinase B has not been determined, considering the homology between members of MMP, it is reasonable to assume that the enzyme has a tertiary structure similar to collagenase. As shown above, the presence of the NH 2 -terminal Phe 88 (corresponding to Phe 79 in collagenase) in gelatinase B correlates with the loss of Ca 2ϩ dependence of the enzyme. Therefore, it is possible that the formation of a salt linkage between Phe 88 and Asp 432 (corresponding to Asp 232 in collagenase) is responsible for the observed gelatinase activity in the absence of Ca 2ϩ . If a salt linkage that mimics the effect of Ca 2ϩ on enzyme activity exists in E t , disruption of the salt linkage at high pH should render activity of the enzyme to be Ca 2ϩ dependent.
The effect of pH on gelatinolytic activity of E t in the presence or absence of Ca 2ϩ was investigated. As pH increased, Ca 2ϩ

FIG. 2. Activation of gelatinase B by various activators.
A, zymographic analysis of activated gelatinase. Recombinant progelatinase B (ϳ15 g/ml) was activated with APMA (1 mM), trypsin (20 g/ml), chymotrypsin (20 g/ml), or stromelysin (1 unit/ml) as described under "Experimental Procedures." Trypsin was removed by passage through a soybean trypsin inhibitor column (200 l), and chymotrypsin was inactivated with 2 mM diisopropyl phosphorofluoride. Two-l samples were analyzed on a 10% gelatin zymogram gel. Arrows, activation sites. a, autoproteolytic cleavage site following activation by APMA; b, the site generated by trypsin or stromelysin activation. c, the sites generated by chymotrypsin activation. Boldface represents the "cysteine switch" sequence. dependence of gelatinolytic activity shifted gradually from being Ca 2ϩ independent to Ca 2ϩ dependent (Fig. 3). When the pH of the reaction mixture shifted from 7.4 to 9.8, the gelatinolytic activity of the enzyme decreased by about 35% in the presence of 1 mM Ca 2ϩ . However, about 95% of the activity was lost in the absence of Ca 2ϩ . This suggests that deprotonation of a positively charged group, probably associated with a salt linkage, is responsible for the loss of enzyme activity at pH 9.8 in the absence of Ca 2ϩ . The data also support the idea that disruption of the salt linkage at high pH can be compensated for by Ca 2ϩ binding to the enzyme.
Discrimination between the Effect of Ca 2ϩ on Stability and Catalysis of Gelatinase B-It is common knowledge that the role of Ca 2ϩ in MMPs is to stabilize the tertiary structures of the enzymes (20 -23). Thus, it is reasonable to assume that the putative salt linkage plays a role similar to Ca 2ϩ and stabilizes the conformation of E t at physiological pH in the absence of Ca 2ϩ . This was assessed by determining the effect of pH on the stability of E a and E t in the absence or presence of Ca 2ϩ . The enzymes were incubated for various times (0 -60 min) at pH 7.5 or 9.8 at 37°C in the presence or absence of 1 mM Ca 2ϩ prior to assaying under normal conditions (pH 7.5, 37°C, and 1 mM Ca 2ϩ ). Both enzymes had a similar stability at 37°C in the presence or absence of Ca 2ϩ , regardless of the pH of the solution (Fig. 4). They were essentially stable over 1-h preincubation at 37°C in the presence of 1 mM Ca 2ϩ at either pH 7.5 or pH 9.8, whereas they lost activity in the absence of Ca 2ϩ . The loss of activity at either pH followed first-order decay kinetics, and the half-life of both E a and E t was ϳ22 min. At pH 7.5, in which the salt linkage should still be intact in E t , the enzyme did not show higher stability than E a in the absence of Ca 2ϩ . This suggests that the Ca 2ϩ -independent activity of E t does not result from the stabilization effect of the putative salt linkage, and lack of activity of E a in the absence of Ca 2ϩ cannot be attributed to the instability of the enzyme. These data also imply that Ca 2ϩ can act as a reversible stimulator, as well as a stabilizer, of the enzyme. This idea was further supported by assaying E a and E t at pH 7.5 or pH 9.8 with or without Ca 2ϩ (1 mM) for only 5 min at 37°C. Under this condition, less than 15% of total activity of the enzymes was lost in the absence of Ca 2ϩ . E a was absolutely dependent on Ca 2ϩ for activity, regardless of the pH of the reaction mixture, indicating that the enzyme activity resulting from addition of Ca 2ϩ was due to the effect of Ca 2ϩ on activity per se rather than on stability (Fig. 5). E t , on the other hand, was not active at pH 9.8 in the absence of Ca 2ϩ . However, it regained about 60% of its activity upon the addition of 1 mM Ca 2ϩ . The increased activity was comparable to the activity of E a stimulated by Ca 2ϩ at pH 9.8. A decrease in pH from 9.8 to 7.5 apparently had a similar effect to that of Ca 2ϩ on enzymatic activity of E t . When E t was assayed at pH 7.5 in the absence of Ca 2ϩ , the enzyme displayed about 90% activity. Adding 1 mM Ca 2ϩ increased the activity to 100%. This might be the stabilization effect of Ca 2ϩ during the assay period. These data suggest that Ca 2ϩ not only stabilizes gelatinase B but also stimulates the enzyme activity by reversibly binding to a distinct "activation" site. Ca 2ϩ apparently affects catalysis by inducing a conformational change similar, if not identical, to the one generated by salt linkage in E t .
Formation of a Salt Linkage between the Side Chain of Asp 432 and the Ammonium Group of Phe 88 Is Responsible for Ca 2ϩ -independent Activity of Trypsin-activated Gelatinase B-The above observations in combination with the x-ray structure of collagenase (27) suggest that formation of a salt linkage between Phe 88 and Asp 432 is responsible for the loss of Ca 2ϩ dependence of the enzyme. To test this hypothesis, a series of mutations were generated at position 432 in the progelatinase molecule (D432E, D432N, D432G, and D432K) to assess the role of Asp 432 in Ca 2ϩ dependence of the enzyme activated by trypsin. All of the trypsin-activated mutant enzymes required Ca 2ϩ for activity (Table II). Even a subtle change at this site (D432E) caused the activity of the enzyme to depend completely on Ca 2ϩ . The percentage of the enzymatic activity of the mutant enzymes in the absence of Ca 2ϩ to their corresponding activities in the presence of Ca 2ϩ ranged from 0.4 to 1.6%. The Ca 2ϩ dependence of trypsin-activated Asp 432 mutants did not result from any major structural alteration in FIG. 3. Effect of pH on Ca 2؉ dependence of E t . Ca 2ϩ -free E t was prepared and assayed for gelatinolytic activity as described under "Experimental Procedures" in 50 mM HEPES (pH 7.35-7.8), CHES (pH 8.25-9.24), or CAPS (pH 9.77) buffer containing 4.3 g/ml enzyme and 0.67 mg/ml [ 14 C]gelatin (329,000 cpm/mg gelatin) in the presence (q) or absence (E) of 1 mM Ca 2ϩ at 37°C for 5 min. Inset, Ca 2ϩ dependence of E t at various pH levels is expressed as a ratio for the activity determined without Ca 2ϩ relative to that with 1 mM Ca 2ϩ .

FIG. 4. Effects of pH and Ca 2؉ on the stability of E a or E t .
Ca 2ϩ -free E a (5.4 g/ml) or E t (2.1 g/ml) was preincubated for up to 60 min in 10 mM HEPES (pH 7.5) or CAPS (pH 9.8) buffer containing 0.1 M NaCl and 0.5 M Zn 2ϩ in the presence (solid symbols) or absence (open symbols) of 1 mM Ca 2ϩ at 37°C. After adjusting the pH and Ca 2ϩ concentration to 7.5 and 1 mM, respectively, the remaining activity for each sample was determined using [ 14 C]gelatin as a substrate (see "Experimental Procedures") and expressed relative to zero-time incubation. q, E a , pH 7.5, ϩCa 2ϩ ; ࡗ, E a , pH 9.8, ϩCa 2ϩ ; å, E t , pH 7.5, ϩCa 2ϩ ; f, E t t, pH 9.8, ϩCa 2ϩ ; E, E a , pH 7.5, ϪCa 2ϩ ; छ, E a , pH 9.8, ϪCa 2ϩ ; Ç, E t , pH 7.5, ϪCa 2ϩ ; Ⅺ, E t , pH 9.8, ϪCa 2ϩ . the enzyme. The 57.5-kDa mutant proenzymes were processed to a 41.5-kDa species by APMA and to a 40-kDa species by trypsin, as judged by zymography (data not shown). The processing patterns were the same as those of WT ( Fig. 2A). Kinetic analysis demonstrated that the K m values for the synthetic substrate were not perturbed (ϳ7 M; Table III), and the K i values for the competitive inhibitor, hydroxamic acid HONHCOCH 2 CH(i-Bu)CO-L-Trp-NHMe, isomer 6A (GM 6001) were nearly identical for the D432G and WT enzyme (28). In addition, all mutants showed similar stability to that of WT in the absence or presence of 1 mM Ca 2ϩ (data not shown). NH 2terminal analysis revealed that the trypsin-activated WT and mutant enzyme D432N had the same amino terminus, Phe 88 . These observations strongly support the hypothesis that a salt linkage between the ammonium group of Phe 88 and the side chain carboxylate moiety of the Asp 432 may substitute for the Ca 2ϩ to generate a proper active-site conformation in E t . Introducing a mutation at position 432 apparently disrupted the salt linkage and resulted in activity of the enzyme to be dependent on Ca 2ϩ .
Identification of Asp 432 as a Putative Ca 2ϩ -binding Ligand-The above data in conjunction with our previous observation that Asp 432 is involved in Ca 2ϩ binding in E a (28) raised the possibility that in the absence of the salt linkage, this residue acts as a Ca 2ϩ -binding ligand. To assess this possibility, we determined the effect of substitutions at Asp 432 on Ca 2ϩ binding affinity of the enzymes activated by APMA. To eliminate the effect of Ca 2ϩ on stability of the enzymes, the enzymatic activities of WT, D432E, D432N, D432G, and D432K were measured as a function of Ca 2ϩ concentration for 5 min, during which time all enzymes were stable ( Fig. 4; some data not shown). The activities of the Ca 2ϩ -depleted WT and mutant enzymes were gradually restored by increasing Ca 2ϩ concentration (Fig. 6). Binding of Ca 2ϩ to these enzymes followed saturation kinetics. The half-maximal stimulation by Ca 2ϩ (K Ca , apparent affinity of Ca 2ϩ ) for WT, D432E, D432N, D432G, and D432K were 20.0, 36.5, 65.5, 121.0, and 194.0 M, respectively. The decrease in affinity for Ca 2ϩ apparently depended on the property of the substituted amino acid. Substitution of the positively charged Lys for the negatively charged Asp at position 432 decreased the Ca 2ϩ binding affinity of the enzyme by a factor of 10. Removing the negatively charged group increased the half-maximal Ca 2ϩ stimulation of D432G and D432N by factors of 6 and 3, respectively, whereas replacing Asp 432 with Glu, the conserved amino acid, did not have a significant effect on Ca 2ϩ binding affinity of the enzyme. Comparison of the k cat values revealed that efficiency of substrate hydrolysis by D432E, D432N, D432G, and D432K was reduced by 1.1-, 1.4-, ϳ3-, and ϳ4-fold, respectively. Interestingly, the order of the decrease in Ca 2ϩ binding affinity correlates well with that of the descending catalytic efficiency of the enzymes (Table III). These data strongly suggest that Asp 432 is directly involved in Ca 2ϩ binding, and decreased k cat values result from decreased affinities of the mutant enzymes for Ca 2ϩ . DISCUSSION We have demonstrated previously that the removal of 13 amino acids from the NH 2 terminus of APMA-activated gelatinase B (E a ) by trypsin was accompanied by the loss of Ca 2ϩ dependence of E a (19). Treating E a with trypsin generated a new NH 2 terminus, Phe 88 . These data led us to propose two possible mechanisms for Ca 2ϩ -dependent activity of gelatinase B (Fig. 1). (a) in the absence of Ca 2ϩ , the 13 amino acids at the NH 2 terminus of E a act as an autoinhibitory fragment, the inhibition of which could be released either by Ca 2ϩ binding to the enzyme (Fig. 1, A1) or by removal of this fragment from NH 2 terminus by trypsin (Fig. 1, T1). (b) the active site-conformation of E a is not stable in the absence of Ca 2ϩ . A proper active-site conformation could be created either by binding of Ca 2ϩ to the enzyme through interaction with NH 2 -and COOHterminal residues (Fig. 1, A2) or by a salt linkage between Phe 88 at the NH 2 terminus and Asp 432 at the COOH-terminal    c The assay conditions are described in Fig. 6. The half-maximal stimulation by Ca 2ϩ for each species (K Ca ) was determined from Lineweaver-Burk plots. region of the catalytic domain of the enzyme (Fig. 1, T2). In this study, we differentiate these proposals and provide sufficient evidence to conclude that the NH 2 -terminal Phe 88 contributes to a salt linkage leading to proper active-site conformation similar, if not identical, to the Ca 2ϩ -generated conformation favorable for catalysis.
Since the activation of progelatinase B by different activators is achieved by stepwise processing of both NH 2 -and COOHterminal peptides (16,17), it is reasonable to suggest that the Ca 2ϩ dependence of the activated species is governed by minor structural differences after processing of both ends. Although the tertiary structure of gelatinase B is unknown, according to known x-ray structures of the other MMP family members (24 -27), we would expect that overall structures of gelatinase activated by different activators should be identical. Therefore, the role of different NH 2 termini in Ca 2ϩ dependence of gelatinase was assessed by activating the proenzyme with trypsin, stromelysin, or chymotrypsin. Treating gelatinase B with these activators removes the putative autoinhibitory fragment (amino acid residues 75-87) from the NH 2 terminus of the enzyme and generates Phe 88 and Gln 89 or Glu 92 as the NH 2 termini, respectively (Fig. 2B). We have found that chymotrypsin-activated gelatinase B, similar to E a , absolutely depends on Ca 2ϩ for activity, whereas trypsin-and stromelysin-activated enzymes with Phe 88 as an NH 2 terminus do not require Ca 2ϩ for activity. The observed differences in Ca 2ϩ dependence of the activated enzymes do not result from differential processing of the COOH-terminal end of the enzyme, because the COOH terminus of both E a and E t were inferred to be identical (19). These data indicate that amino acid residues 75-87 do not constitute an autoinhibitory fragment and strongly support the idea that the presence of Phe 88 at the NH 2 terminus is responsible for the Ca 2ϩ -independent activity of trypsin-and stromelysin-activated gelatinase.
The possibility that Phe 88 is involved in the formation of a salt linkage was suggested by the x-ray structure of the catalytic domain of neutrophil collagenase (27). In this enzyme, the NH 2 terminus Phe 79 (corresponding to Phe 88 in gelatinase B) apparently forms a salt linkage with highly conserved residue Asp 232 (corresponding to Asp 432 in gelatinase B). That the presence of a salt linkage is responsible for the activity of E t in the absence of Ca 2ϩ is supported by the observation that the enzymatic activity of E t shifted from Ca 2ϩ independent to Ca 2ϩ dependent as the pH of the reaction mixture increased from 7.5 to 9.8. The enzyme lost about 95% of its activity at pH 9.5 in the absence of Ca 2ϩ . The loss of gelatinolytic activity of Ca 2ϩ -free E t at high pH could result from ionization of catalytically essential amino acid residues that are involved in catalysis or those that are responsible for structural integrity of the enzyme. The ionization of active-site groups directly involved in catalysis does not completely account for the loss of activity of E t in the absence of Ca 2ϩ at pH 9.8, because high pH would have impaired the enzyme activity whether Ca 2ϩ was present or not. A salt linkage in the protein, on the other hand, can play a role in stabilizing the enzyme structure and/or maintaining a proper active-site conformation and is sensitive to the pH of the solution (33). For example, the salt linkage formed between the ␣-ammonium group of Ile 16 and the side chain of Asp 194 in chymotrypsin (34) controls the active-site conformation of the enzyme. When the ␣-ammonium group is deprotonated at high pH, chymotrypsin is reversibly inactivated, displaying a pK a of 8.8. A pK a of ϳ8.8 reflecting the ionization state of the ␣-ammonium group of Phe 88 in E t could be extracted from Fig. 3. The pK a of free ammonium group of Phe is about 7.8, and it rises to about 10 when it is involved in a salt linkage (35). The observed pK a suggests that the superposition of both conformations causes this residue to titrate with a pK a of 8.8. These data suggest that a salt linkage involving Phe 88 , similar to the one in ␣-chymotrypsin (34), is present in E t and is able to substitute for Ca 2ϩ in maintaining the proper active-site conformation of the enzyme favorable for catalysis. The putative salt linkage in E t did not contribute to enzyme stability because the rate of inactivation of E t at pH 7.5 was similar to that at pH 9.8 in the absence of Ca 2ϩ . Furthermore, the activity of E t at different pH was determined for only 5 min. Under this condition, E t was essentially stable. Therefore, the loss of activity of Ca 2ϩ -free E t at high pH is apparently due to reversible disruption of the active-site conformation rather than from irreversible denaturation of the protein.
The involvement of Asp 432 in the salt linkage mimicking the effect of Ca 2ϩ on enzymatic activity is supported by the observation that replacing Asp 432 with Gly, Gln, or Lys, amino acids that conceivably could destroy the salt linkage, resulted in enzyme species the activities which were dependent on Ca 2ϩ . Even adding one more methylene group to the carboxylate side chain of amino acid 432 (D432E) resulted in an enzyme that required Ca 2ϩ for activity. The requirement of both correct processing of the NH 2 terminus and of exact length of the side chain at position 432 for Ca 2ϩ -independent activity also suggests the existence of a salt linkage between the two sites in the gelatinase molecule.
We and others (19 -23) have demonstrated that Ca 2ϩ acts as both an activator of MMPs and a stabilizer to prevent the enzymes from denaturation. In this study, we have shown that the function of Ca 2ϩ in catalytic activity of gelatinase B could be effectively performed by the salt linkage formed between the NH 2 and COOH termini of the catalytic domain. This result raised a possibility that Ca 2ϩ forms an intramolecular bridge between the NH 2 and COOH termini of the enzyme similar to the salt linkage to maintain a proper active-site conformation. Interestingly, a similar Ca 2ϩ -binding site has been identified in protein C (36). It was demonstrated that binding of Ca 2ϩ to a high affinity site in protein C involving Glu 70 and Glu 80 resulted in a conformational change that is required for activation by thrombin-thrombomodulin complex, the activator of protein C. Substituting Glu 80 with Lys created a salt linkage between Glu 70 and Lys 80 and led to a conformational change similar to that induced by Ca 2ϩ . Our study has strongly suggested that Asp 432 in the COOH-terminal helix functions as one of the Ca 2ϩ -binding ligands. The direct evidence for the participation of this residue in Ca 2ϩ binding was obtained from the observation that affinity of gelatinase for Ca 2ϩ was significantly affected by the side chain of the introduced amino acid residue at position 432. Replacing Asp 432 with either the isosteric amide group (D432N) or removing the side chain as in D432G decreased the affinity of the enzyme for Ca 2ϩ by factors 3 and 6, respectively, whereas the conserved substitution (D432E) had little effect on its affinity for Ca 2ϩ . Introducing a positively charged side chain (D432K), however, decreased the Ca 2ϩ affinity by a factor of 10. This finding indicates that both the size and charge of the side chain of Asp 432 is important for maintaining a proper geometry of the Ca 2ϩ -binding site and that the negatively charged carboxylate group of Asp 432 apparently contributes to Ca 2ϩ coordination. The involvement of Asp 432 and Ca 2ϩ binding in the catalysis of gelatinase B has been further suggested by the kinetic study of the WT and mutant enzymes. All of the substitution mutants had K m values for the synthetic substrate similar to the WT, but their k cat values were decreased significantly. The decreased Ca 2ϩ affinities are apparently responsible for the observed decrease in k cat values of the mutant enzymes. The order of decrease in k cat values for the mutants is WT Ն D432EϾ D432N Ͼ D432G Ͼ D432K, which correlates well with the order of decrease in Ca 2ϩ binding affinity of the mutant enzymes. The lack of effect of Asp 432 replacement on K m indicates that the active site geometry of the mutant enzymes are still intact. These observations strongly suggest that binding of Ca 2ϩ to Asp 432 triggers a minor conformational change at the active site and leads to proper orientation of the catalytically essential groups without significantly affecting the geometry of the entire active site.
The highly conserved sequence Lys 431 -Asp-Asp-Val-Asn 435 in gelatinase B containing Asp 432 has striking parallels with the Ca 2ϩ -binding site (Lys-Asp-Asp-Ile-Asn) in coagulation factors IX and X (37)(38)(39)(40), further supporting the idea that this sequence in gelatinase B is one of the Ca 2ϩ -binding site. On the basis of the x-ray crystal structure of the catalytic domain of neutrophil collagenase, this conserved sequence constitutes the basement of the active site of the enzyme (27); however, no Ca 2ϩ was detected at this particular region of the enzyme. This is not a surprising observation. Lebioda and Stec (41) have also reported that the "catalytic" metal ion in yeast enolase is not observed by x-ray diffraction. The concentration of Ca 2ϩ and the pH of an enzyme solution from which the crystals were grown are known to affect the number of Ca 2ϩ -binding sites seen in the protein. Pantoliano et al. (42) found that occupancy of a weak Ca 2ϩ -binding site in bacterial serine protease subtilisin is a function of the Ca 2ϩ concentration for a series of x-ray structures obtained for crystals grown with increasing amounts of Ca 2ϩ in the range of 0 -40 mM at pH 9.0. When crystals were grown in the presence of 10 mM Ca 2ϩ and at pH 5.0, the two low affinity regulatory sites of troponin C (K d Ϸ 10 Ϫ5 M at neutral pH) were devoid of Ca 2ϩ ion, whereas the higher affinity structural sites (K d Ϸ 10 Ϫ7 M at neutral pH) were occupied by Ca 2ϩ in the x-ray structure of the enzyme (43). Furthermore, it has been shown that the Ca 2ϩ binding affinity is significantly decreased by lowering the pH in stromelysin (44). The observed discrepancy between our data and the x-ray crystal structure of collagenase could be due to the fact that the reported x-ray structures of MMPs were obtained under different crystallization conditions. Lovejoy et al. (24) identified one Ca 2ϩ site in the truncated fibroblast collagenase when they crystallized the enzyme at pH 6.5 in the absence of added Ca 2ϩ . The same group (25) found three Ca 2ϩ sites in this enzyme only by increasing the pH to 7-9 and adding 1 mM Ca 2ϩ to the enzyme solution. Under the conditions of pH 6.0 and 5 mM Ca 2ϩ , Bode and co-workers (26,27) identified two Ca 2ϩ sites in the truncated neutrophil collagenase. These Ca 2ϩ -binding sites are thought to have a high affinity for Ca 2ϩ and be involved in stabilization of the enzyme. The site involving Asp 432 , however, may have a low affinity for Ca 2ϩ ; thus, it is likely that Ca 2ϩ is not able to bind to this site under the crystallization conditions used, or Ca 2ϩ binds to this region of the molecule only during catalysis.
In summary, our data provide sufficient evidence to conclude that Ca 2ϩ plays dual roles in gelatinase B. It is involved in catalysis by bridging NH 2 -and COOH-terminal regions of the catalytic domain and in structural stability by binding to several sites on the enzyme. The role of Ca 2ϩ in catalysis can be substituted by the salt linkage between the ammonium group of Phe 88 and the carboxylate side chain of Asp 432 . This interaction renders the enzyme independent of Ca 2ϩ for activity. Because all MMP family members require Ca 2ϩ for activity and contain these conserved amino acids, we anticipate that the proposed mechanism for gelatinase B will apply to other MMP members as well. For neutrophil collagenase we have demonstrated previously that the COOH-terminal region containing the conserved sequence Asp 232 -Asp-X-Asp is necessary for general catalytic activity of the enzyme (45). Replacing Asp 232 with Gly has also been demonstrated to reduce the catalytic activity of the APMA-activated collagenase. Whether the presence of a salt linkage in stromelysin-activated collagenase substitutes for Ca 2ϩ in activity remains to be determined.