Calcium regulates tertiary structure and enzymatic activity of human endometase/matrilysin-2 and its role in promoting human breast cancer cell invasion

Abstract

Human MMP-26 (matrix metalloproteinase-26) (also known as endometase or matrilysin-2) is a putative biomarker for human carcinomas of breast, prostate and other cancers of epithelial origin. Calcium modulates protein structure and function and may act as a molecular signal or switch in cells. The relationship between MMPs and calcium has barely been studied and is absent for MMP-26. We have investigated the calcium-binding sites and the role of calcium in MMP-26. MMP-26 has one high-affinity and one low-affinity calcium binding site. High-affinity calcium binding was restored at physiologically low calcium conditions with a calcium-dissociation constant of 63 nM without inducing secondary and tertiary structural changes. High-affinity calcium binding protects MMP-26 against thermal denaturation. Mutants of this site (D165A or E191A) lose enzymatic activity. Low-affinity calcium binding was restored at relatively high calcium concentrations and showed a K_d2 (low-affinity calcium-dissociation constant) value of 120 μM, which was accompanied with the recovery of enzymatic activity reversibly and tertiary structural changes, but without secondary structural rearrangements. Mutations at the low-affinity calcium-binding site (C3 site), K189E or D114A, induced enhanced affinity for the Ca²⁺ ion or an irreversible loss of enzymatic activity triggered by low-affinity calcium binding respectively. Mutation at non-calcium-binding site (V184D at C2 site) showed that C2 is not a true calcium-binding site. Observations from homology-modelled mutant structures correlated with these experimental results. A human breast cancer cell line, MDA-MB-231, transfected with wild-type MMP-26 cDNA showed a calcium-dependent invasive potential when compared with controls that were transfected with an inactive form of MMP-26 (E209A). Calcium-independent high invasiveness was observed in the K189E mutant MDA-MB-231 cell line.

INTRODUCTION

MMPs (matrix metalloproteinases) are a family of proteolytic enzymes that degrade components of the ECM (extracellular matrix) such as collagen, gelatin, fibronectin, vitronectin and laminin [1,2]. MMPs are known to participate in reproduction, angiogenesis, development, morphogenesis and tissue remodelling, as well as pathological conditions, including arthritis, cardiovascular disease, and cancer progression and metastasis [1–3]. MMP-26 (also known as endometase/matrilysin-2) is the smallest member of the MMP family, comprising only pro- and catalytic domains like MMP-7 (matrilysin), but without a haemopexin-like domain [4–7]. According to peptide library screening and kinetic studies, MMP-26 has a substrate specificity similar to that of other MMPs, such as MMP-2 and MMP-9 [8]. MMP-26 promotes the invasion of a highly metastatic and tumorigenic prostate cancer cell line through its activation of pro-MMP-9 (pro-gelatinase B) [9]. Expression level of MMP-26 is significantly higher in pre-invasive human breast DCIS (ductal carcinoma in situ) [10,11] and HGPIN (high-grade prostatic intraepithelial neoplasia) [12] compared with that in normal human breast tissue samples and non-neoplastic ducts. Level of MMP-26 expression is also high in early stages (I and II) of invasive carcinoma, whereas in stage III invasive carcinomas the level of MMP-26 decreases [10,12]. This is similar to the discovery that a high level of MMP-26 expression was detected in low-grade SCC (squamous cell cancer), but not in dedifferentiated grade III tumours [13]. Thus MMP-26 could be a marker of early-stage cancer [10–13].

Ca²⁺ ions have numerous critical roles in biological systems with involvement in cell-cycle regulation, proliferation, cell migration, angiogenesis and apoptosis, with enzymes such as calmodulin, calbindin and calpain [14–20]. Ca²⁺ ions also participate as signalling mediators in biological systems [21]. Ca²⁺ ions regulate divergent calcium cascade pathways in cancer cells compared with normal cells, as a result of differing protein expression patterns in these cells [22]. For example, cancer cells respond to hormones via Ca²⁺-regulatory mechanisms that result in Ca²⁺-mediated apoptosis, whereas normal cells do not [18]. Normally, very low Ca²⁺ ion concentrations are maintained in the cytoplasm (300–1200 nM) [23]. When the cell is activated by an extracellular signal, Ca²⁺ ions are released from the endoplasmic reticulum or enter through voltage-gated Ca²⁺ channels in the cell membrane, resulting in cytosolic calcium concentrations of approx. 5 μM [23]. This fluctuation of calcium concentrations can alter cellular physiology by means of structural changes induced by calcium-binding proteins, such as the calcium-mediated electrostatic switch mechanism for m-calpain [24]. Binding of Ca²⁺ ions to proteins may confer structural stability in which reagent-induced and thermal denaturation [25,26] are less probable, or modulate substrate binding [27].

On the basis of crystallographic studies of the catalytic domains of fibroblast collagenase/MMP-1 [28], stromelysin-1/MMP-3 [29], matrilysin/MMP-7 [30], neutrophil collagenase/MMP-8 [31], stromelysin-3/MMP-11 [32], macrophage metalloelastase/MMP-12 [33] collagenase-3/MMP-13 [34] and MT1-MMP (membrane-type-1 MMP) [35], at least two calcium-binding sites have been identified. Previous studies have shown that Ca²⁺ ions stabilize MMP-1 [25], mediate signal transduction in the expression of MMP-2 [36], facilitate the activation of pro-MMP-3 [37], and induce cleavage of both N- and C-termini, resulting in activation and C-terminal truncation of pro-MMP-9 from the complex of pro-MMP-9 and chondroitin sulfate proteoglycans [38], and that Ca²⁺ ion influx inhibits MT1-MMP processing [39]. However, little research has focused on the structural and functional roles of Ca²⁺ ions in MMPs. We have recently observed that Ca²⁺ ions can modulate the catalytic activity of MMP-26 through conformational changes in the active site. We investigated the characteristics of calcium binding in wild-type and mutant forms of MMP-26 by means of calcium titration monitoring fluorescent peptide hydrolysis, by fluorescence spectroscopy with ANS (8-anilino-1-naphthalenesulfonate) and Quin2 [2-({2-[bis(carboxymethyl)amino]-5-methylphenoxy}methyl)-6-methoxy-8-[bis-(carboxymethyl)amino]quinoline] and CD spectroscopy. Cellular invasion assays lend additional support to the consequential role of Ca²⁺ ions in enzymatic function. In the present study we investigated the characteristics of calcium binding in MMP-26.

MATERIALS AND METHODS

Construction of pro-MMP-26 and mutant forms

Human endometase was cloned and ligated into pET-20b(+) and subsequently expressed in Escherichia coli as described previously [4]. cDNA of pro-MMP-26 was amplified by PCR using forward primer, containing a HindIII restriction site (underlined), 5′-CCCAAGCTTGTTCCAGTGCCCCCTGCTGCAGACCAT-3′ and reverse primer, containing a BglII restriction site (underlined), 5′-GAAGATCTAGGTATGTCAGATGAA- CATTTTTCTCC-3′, and subcloned into the pFLAG-CTS expression vector (Sigma) between the HindIII and BglII sites. In order for cDNA to be aligned with the ATG translational start codon, an extra nucleotide following the OmpA Signal Peptide (signal sequence for secretion of C-terminal FLAG fusion proteins to periplasmic space) was deleted using forward primer, 5′-GCTACCGTTGCGCAAGCTGTTCCAGTGCCCCCTGCT-3′, and reverse primer, 5′-AGCAGGGGGCACTGGAACAGCTTGCGCAACGGTAGC-3′. Site-directed mutagenesis of putative calcium-binding sites, D114A, D165A, V184D, K189E and E191A, and catalytic Glu²⁰⁹, E209A, was accomplished using PCR with the primers given in Table 1. Mutant constructs were transformed into E. coli DH5α cells for amplification and purification of the pFLAG-CTS/pro-MMP26 vector. Sequences were confirmed by DNA sequencing using an N-26 sequencing primer (Sigma), 5′-CATCATAACGGTTCTGGCAAATATTC-3′ for pFLAG-CTS/pro-MMP-26. Resulting constructs were transformed into E. coli BL21 cells for expression and purification of the protein. cDNA of pro-MMP-26 was also subcloned into the p3xFLAG-CMV™-13 expression vector (Sigma) between the HindIII and XbaI sites. The PCR primer for the 5′ HindIII restriction site was the same as the pFLAG-CTS subcloning primer, and for the XbaI restriction site (underlined), 5′-GC-TCTAGAAGGTATGTCAGATGAACATTTTTCTCC-3′.Site-directed mutagenesis of the putative calcium-binding site (K189E) and catalytic Glu²⁰⁹ (E209A) was accomplished using PCR with the same primers as above for p-FLAG-CTS. The p3xFLAG-CMV™-13/pro-MMP-26 vector was amplified and purified using the above methods. Sequences were confirmed by DNA sequencing using an N-CMV sequencing primer (Sigma), 5′-AATGTCGTAATAACCCCGCCCCGTTGACGC-3′, for p3xFLAG-CMV™-13/pro-MMP-26.

Table 1. Primers used for mutagenesis.

Mutant	Primer	Sequence
D114A	Forward	5′-CAATTACCCACATGCTATGAAGCCATCCG-3′
	Reverse	5′-CGGATGGCTTCATAGCATGTGGGTAATTG-3′
D165A	Forward	5′-GATGGTTGGCCCTTTGCTGGGCCAGGTGGTATC-3′
	Reverse	5′-GATACCACCTGGCCCAGCAAAGGGCCAACCATC-3′
V184D	Forward	5′-GGAAATCCTGGAGATGTCCATTTTGAC-3′
	Reverse	5′-GTCAAAATGGACATCTCCAGGATTTCC-3′
K189E	Forward	5′-GTTGTCCATTTTGACGAGAATGAACACTGG-3′
	Reverse	5′-CCAGTGTTCATTCTCGTCAAAATGGACAAC-3′
E191A	Forward	5′-CCATTTTGACAAGAATGCACACTGGTCAGCTTCAG-3′
	Reverse	5′-CTGAAGCTGACCAGTGTGCATTCTTGTCAAAATGG-3′
E209A	Forward	5′-CCTGGTTGCAACTCATGCGATTGGGCATTCTTTGG-3′
	Reverse	5′-CCAAAGAATGCCCAATCGCATGAGTTGCAACCAGG-3′

Expression of MMP-26 in E. coli and refolding of the denatured protein

Expression of the catalytic domain of MMP-26, but not its prodomain, caused improper folding and resulted in an inactive enzyme (results not shown). Therefore the prodomain is necessary to chaperone active enzyme formation. The activation mechanism of MMP-26 is still unclear, but is likely to involve auto-activation [8,40]. Active MMP-26 was prepared as described previously [4]. In brief, MMP-26 was expressed in the form of inclusion bodies from E. coli BL21 cells. The inclusion bodies were isolated and purified using B-PER™ (Pierce, Rockford, IL) bacterial protein extraction reagent according to the manufacturer's instructions. The insoluble protein was dissolved in 8 M urea and 25 mM Tricine at approx. 2.5 mg/ml and then refolded by dialysis. During dialysis, pro-MMP-26 was auto-activated. Folding and activation patterns were determined by electrophoresis followed by Western blotting with an anti-FLAG M2 monoclonal antibody (Sigma). SeeBlue Plus2 pre-stained standard (Invitrogen) was used to determine the molecular mass of MMP-26. The protein was purified using an anti-FLAG M2 affinity column (Sigma). The enzyme concentration was measured with a molar absorption coefficient, ϵ₂₈₀, of 57130 M⁻¹·cm⁻¹ using GCG (Genetics Computer Group) software as described previously [4].

Removal of low-affinity, or both high- and low-affinity, Ca²⁺ ions

Sequence alignment and crystal structural analysis revealed the possibility of both low- and high-affinity calcium-binding sites for MMP-26. To remove the low-affinity Ca²⁺ ions (i.e. Ca²⁺ ions bound to low-affinity binding sites), the enzyme was dialysed three times in 0.01% Brij-35 (polyoxyethlene dodecyl ether), 10 mM Hepes/NaOH, pH 7.5, 10 mM NaCl and 0.1 μM ZnSO₄ for 8 h at 4 °C. This was followed by dialysis in the presence of 0.1% Chelex 100 (Sigma). The affinity of Chelex 100 for Ca²⁺ ions is not particularly high, with a binding constant of 4.6×10³ M⁻¹. Therefore this method removed only the low-affinity Ca²⁺ ions from the enzyme [41]. Dialysis in the presence or absence of Chelex 100 did not alter further experiments. Adding Chelex 100 ensured the removal of low-affinity Ca²⁺ ions from the enzyme. To eliminate the high-affinity Ca²⁺ ions (i.e. Ca²⁺ ions bound to high-affinity binding sites), the enzyme was dialysed three times in the presence of 0.01% Brij-35, 2 mM EGTA, 10 mM Hepes/NaOH, pH 7.5, 0.1 M NaCl and 0.1 μM ZnSO₄ for 8 h at 4 °C, followed by dialysis three times in 0.01% Brij-35, 10 mM Hepes/NaOH, pH 7.5, 10 mM NaCl and 0.1 μM ZnSO₄ for 8 h at 4 °C, in order to remove EGTA. For CD spectroscopy, 10 mM Tris/HCl was substituted for 10 mM Hepes/NaOH. For ANS-binding assays, Brij-35 was omitted from the buffer to avoid interactions between ANS and Brij-35.

Fluorescence titration with Quin2

A Quin2 (Sigma) stock solution (approx. 500 μM) was prepared by dissolution in buffer (Chelex 100-treated 10 mM Hepes/NaOH, pH7.5, and 10 mM NaCl). The concentration of the Quin2 stock solution was determined by using the molar absorption coefficient, ϵ₂₄₀, of 4.2×10⁴ M⁻¹·cm⁻¹, following a 1:100 dilution of stock solution in the presence of 10 mM CaCl₂ [42]. LACD (low-affinity-Ca²⁺-ion-depleted) enzymes used for the Quin2 titration experiments were prepared by dialysis as described above. Titration was performed at 25 °C in a quartz cell (1 cm×0.5 cm). All measurements were performed in 10 mM Hepes/NaOH, pH 7.5, and 10 mM NaCl at 25 °C. Titration was performed by adding 5 μl aliquots of 300 μM Quin2 to 1500 μl of a 2.5 μM enzyme solution in a quartz cell. Time-dependence assays of Quin2 binding to Ca²⁺ showed no change in Quin2 fluorescence after MMP-26 was mixed with a 20-fold excess of Quin2 at 25 °C in 10 mM Hepes buffer, after 5–40 min (results not shown). At each titration point, the sample was incubated for 10 min before measuring the emission spectra from 450 to 530 nm with 5 nm excitation and emission slit width at an excitation wavelength of 339 nm. The emission was scanned at a rate of 2 nm/s. Fluorescence spectra intensity changes of the Quin2–Ca²⁺ complex were monitored at 490 nm. Background fluorescence spectra of Quin-2 were measured by using either water or buffer in the presence of various concentrations of Quin2. Background fluorescence spectra of water and buffer were identical and showed no calcium contamination.

Calculation of the high-affinity calcium-dissociation constant, K_d1, following fluorescence titration with Quin2

Calcium-dissociation constants (K_dCa²⁺) for proteins can be calculated using several methods [42,43], assuming that the increase in fluorescence intensity is proportional to the concentration of the Quin2–Ca²⁺ complex [44]. In the present study, in the presence of the calcium-binding protein, the apparent dissociation constant of the Quin2–Ca²⁺ complex, K_d(QCa), was calculated by fitting experimental data to the following equation [42]:

(1)

where f_i is the intensity of fluorescence at each point, V is the maximum of the fluorescence, [Ca]_T is the total calcium concentration, and [Q]_i is Quin2 concentrations at each point. K_d(QCa) may then be applied to the following equation in order to obtain the high-affinity calcium-dissociation constant of the enzyme, K_d1:

(2)

where [E] is the enzyme concentration, and K_Q is the calciumdissociation constant for Quin2 (5.3 nM) at low salt concentrations [43].

Calcium titration by monitoring enzyme activity

The fluorescence-quenched peptide substrate Mca [(7-methylcoumarin-4-yl)acetyl]-Pro-Leu-Gly-Leu-Dap (2,3-diaminopropyl)-Dnp (2,4-dinitrophenyl)-Ala-Arg-NH₂ (Bachem) was used to measure the rate of substrate hydrolysis. Enzyme activity was measured using a PerkinElmer LS 50B Luminescence Spectrometer connected to a constant-temperature water bath at 25 °C. Release of the fluorescent cleavage product was monitored by measuring fluorescence (λ_excitation=328 nm and λ_emission=393 nm), where the increase of fluorescence intensity is proportional to substrate hydrolysis. Assays were performed in 0.01% Brij-35, 10 mM Hepes/NaOH, pH 7.5, and 10 mM NaCl, with various Ca²⁺ concentrations at 25 °C. Different concentrations of Ca²⁺ stock solutions were prepared by serial dilutions from 1.6 M to 20 μM. Enzymes (final concentration of 0.1 μM) were incubated in 0.01% Brij, 10 mM Hepes/NaOH, pH 7.5, and 10 mM NaCl with varying concentrations of Ca²⁺ solution in a quartz cell (3 mm×3 mm) for an incubation period of 20 min. Assays commenced with the addition of a substrate stock solution to a final concentration of 1 μM. Calcium reagent (Sigma) contained less than 0.0005% (w/w) Zn²⁺ ions. The EC₅₀ value of Ca²⁺ was 0.12 mM, resulting in less than 0.37 nM of possible Zn²⁺ ion contamination. Denatured MMP-26 from the bacterial extract was refolded in the presence of 0.1 μM Zn²⁺ to produce folded and activated MMP-26. Therefore incubation of LACD-MMP-26 with 0.1 μM Zn²⁺ is an appropriate zinc concentration to restore Zn²⁺ ions, if lost. However, incubation of LACD-MMP-26 with a final concentration of 0.1 μM Zn²⁺ had no effects on enzymatic activity, leading to the conclusion that dialysis did not remove Zn²⁺ ions from enzyme and Zn²⁺ ions were unlikely to have affected the experimental outcomes.

Calculation of the low-affinity calcium-dissociation constant, K_d2

If a single Ca²⁺ ion regulates enzymatic activity, then the calcium-dissociation constant, K_d2, can be expressed as follows:

when enzymatic activity reaches 50% of the maximum

Therefore K_d2=[Ca²⁺]_Total−0.5[E]_Total. If it is assumed that [Ca²⁺]_Total is much greater than [E]_Total, then K_d2=[Ca²⁺]_Total at 50% enzymatic activity, the EC₅₀ value.

For MMP-26, E stands for LACD-MMP-26 and Ca²⁺ stands for the low-affinity Ca²⁺ ion. We assume that the high-affinity Ca²⁺ ion was not dissociated from LACD-MMP-26 owing to its low value of dissociation constant, K_d1.

ANS binding

An ANS (Sigma) stock solution of 1 mM was prepared by dissolving ANS in 10 mM Hepes, pH 7.5, and 10 mM NaCl. ANS concentration was determined by using the molar absorption coefficient, ϵ₃₅₀, of 5000 M⁻¹·cm⁻¹ [45]. LACD-MMP-26 utilized for the ANS-binding experiments were prepared by dialysis as described above in the absence of Brij-35. All measurements were performed in 10 mM Hepes, pH 7.5, and 10 mM NaCl, with various Ca²⁺ concentrations, at 25 °C in a quartz cell (3 mm×3 mm) after 20 min of incubation. To saturate the hydrophobic sites of the enzyme with ANS, 100 μM ANS was added to 3.4 μM of wild-type MMP-26 or 2 μM of mutant-form MMP-26 solution, constituting a 30-fold molar excess. ANS emission spectra were obtained from 405 to 560 nm with 5 nm excitation and emission slit widths at an excitation wavelength of 395 nm. Changes in ANS fluorescence spectra, owing to the binding of ANS to protein, were monitored at 480 nm.

CD spectroscopy

CD was used to monitor changes in the secondary structure of the enzyme. Protein samples for CD spectroscopy experiments were prepared by dialysis as described above. All measurements were performed in 10 mM Tris/HCl, pH 7.5, 10 mM NaCl and 0.01% Brij-35. CD spectra were obtained after the enzyme had been incubated for 2 h in the presence of 2 mM EGTA (400-fold excess) or in 10 mM CaCl₂ solution. CD spectroscopic measurements were performed on an Aviv 62 ADS CD spectrometer equipped with a 0.1-cm-pathlength cell and a thermoelectric cuvette holder and connected to a model CFT-33 refrigerated recirculator (NESLAB). CD spectra were acquired by scanning from 260 to 185 nm in 0.5 nm increments with a 1 nm bandwidth. Triplicate scans were recorded, averaged and analysed. Buffer scans were subtracted, and the data were converted into molar ellipticity ([θ], degrees·cm²·dmol⁻¹). Thermal denaturation transitions were measured by continuously monitoring ellipticities at a fixed wavelength of 218 nm while the temperature of the sample was increased. Data were collected at 3 °C increments from 4 °C to 61 °C. The apparent fractional extent of unfolding, f_app, was calculated from the molar ellipticity values ([θ]) at 218 nm using the following equation:

where [θ]_N and [θ]_U represent ellipticity values of the native and non-native states respectively [46].

Computer modelling of the MMP-26 catalytic domain

The homology modelling structure of the MMP-26 catalytic domain was constructed previously [47] using the SwissModel program [48]. The crystal structure of MMP-12 [33], which exhibits high similarity to MMP-26, was used as a template (PDB accession code 1JK3). Minimization calculations were performed with an Amber forcefield modified to include parameters for zinc and calcium [47]. Both of the putative low-affinity Ca²⁺ ions at the C2 and C3 sites were removed, and the protein structure was minimized to examine the resultant structural changes. The catalytic domain structures of the K189E and V184D mutants were then generated and minimized to investigate any structural changes resulting from these mutations. RMSD (root mean square deviation) values were reported after the comparison of structures.

Cell culture

MDA-MB-231 (A.T.C.C., Manassas, VA, U.S.A.), an established human breast carcinoma cell line, was routinely grown in high-glucose DMEM (Dulbecco's modified Eagle's medium) (Gibco Invitrogen Corporation) supplemented with 10% (v/v) FBS (fetal bovine serum) (Hyclone), 100 units/ml penicillin and 100 μg/ml streptomycin (Cambrex) in a humidified atmosphere containing 5% CO₂ at 37 °C.

Transfection of MDA-MB-231 cells and isolation of cell lines expressing wild-type, and K189E and E209A mutant forms of MMP-26

MDA-MB-231 cells were transfected with wild-type, and K189E and E209A mutant forms of MMP-26 containing p3xFLAG-CMV™-13 vectors using Transfectol (GeneChoice). Transfectol-mediated DNA transfections into MDA-MB-231 cells were performed following the instructions provided by GeneChoice. Transfected cell lines were maintained in the presence of 550 μg/ml Geneticin (G-418 sulfate) (Fisher Scientific) and then screened on the basis of FLAG and MMP-26 expression. MMP-26, which was overexpressed from MDA-MB-231 cells, tends to secrete out owing to the signal peptide of the p3x-FLAG vector. Serum-free medium from the cell culture was used for screening colonies. Selected stably transfected cell lines were expanded in the presence of 550 μg/ml G-418 due to rapid loss of vector in the absence of G-418. Cell lines expressing wild-type, and K189E and E209A mutant forms of MMP-26 were analysed for their invasiveness capabilities in modified Boyden chamber (8 μm pore size) (BD Biosciences) invasion assays.

Modified Boyden chamber invasion assays

The invasiveness of MDA-MB-231 cells transfected with wild-type, and K189E and E209A mutant forms of MMP-26 through reconstituted ECM was determined as described previously [9]. Briefly, invasion inserts containing polycarbonate filters with 8 μm pores were coated with 70 μl of 250 μg/ml Type IV collagen (Sigma). The Boyden chambers were dried and sterilized in a laminar flow hood under ultraviolet radiation overnight. To commence the assay, 500 μl of high-glucose DMEM containing 10% (v/v) FBS was added to the lower chambers, and 300 μl of prepared cell suspensions (1.5×10⁵ cells/ml) in Eagle's MEM (minimum essential medium) with or without calcium (Mediatech) was added to each insert for calcium-plus or calcium-minus conditions respectively. In either instance, serum is only present in the medium outside of the invasion inserts. The cells housed inside the invasion inserts have no direct contact with serum at the beginning of the experiment. After 12 h of incubation, invasive cells that had passed through the filter to the lower surface of the membrane were stained using a 0.1% (w/v) Crystal Violet solution. Cells remaining inside the chamber were removed with a cotton swab, and the filters were removed and mounted on a microscope slide. The membranes were photographed with an Olympus DP10 digital camera under a Nikon FX microscope. Cells were counted by IMA (integrated morphometry analysis). For statistical analyses, the most aggressively invasive cell line, expressing the K189E mutant form of MMP-26, was assumed to reflect 100% cell invasion. The ratio of the number of invaded cells of each cell lines to the invaded cells of the K189E mutant line was used for subsequent comparative analyses by ANOVA.

RESULTS

Sequence alignment and homology modelling of MMP-26

Crystallographic studies demonstrate that catalytic domains of the MMPs share similar folding topology and metal-binding structures [33]. Thus sequence alignment and homology-modelling structures can be useful tools to predict the calcium-binding sites of MMPs. We used the same sequencing numbering as given by Park et al. [4]. A homology modelling structure of MMP-26 was constructed based on the crystal structure of the catalytic domain of MMP-12 (cdMMP-12) [47]. A comparison of the MMP-26 structure with cdMMP-12 predicted three putative calcium-binding sites, designated C1, C2 and C3 in Figure 1(A). The C1 site of MMP-26 is composed of three side-chain carboxy groups originating from amino acid residues (Asp¹⁶⁵, Asp¹⁸⁸ and Glu¹⁹¹) and three carbonyl groups from the backbone (Gly¹⁶⁶, Gly¹⁶⁸ and Ile¹⁷⁰). The C2 site of MMP-26 consists of three carbonyl groups originating from the backbone (Asp¹⁴⁸, Gly¹⁸⁰ and Pro¹⁸²), which is not feasible for calcium binding. The C3 site is composed of one carboxy group from an amino acid residue (Asp¹¹⁴) and two carbonyl groups from the backbone (Lys¹⁸⁹ and Glu¹⁹¹).

Figure 1. Modelled structure of MMP-26 and sequence alignment data of several MMPs.

(A) A modelled structure of MMP-26 based on the structure of MMP-12. Green spheres represent Ca²⁺ ions. C1, C2, and C3 represent calcium-binding sites. Small dark blue spheres represent Zn²⁺ ions. Each calcium-binding site is magnified to demonstrate the binding moieties. The RMSD value between the backbones of the minimized and the homology modelled structure of MMP-26 is 1.40 Å. (B) Sequence alignment of the MMP26 catalytic domain sequence with the catalytic domains of several MMPs. Conserved calcium-binding residues are boxed. Residues in dark grey correspond to the MMP C1 sites, residues in medium grey correspond to MMP C2 sites, and residues in light grey correspond to MMP C3 sites. Co-ordinating residues, shared by the C1 and C3 sites, are denoted in the black-lined box.

Sequence alignment of MMP-26 with the available crystallographic structures of other MMPs was used to re-examine the possible ligands at each calcium-binding site (Figure 1B). The possible Ca²⁺ co-ordination at each site for each MMP was summarized (Table 2). Asp¹⁶⁵, Gly¹⁶⁶, Asp¹⁸⁸ and Glu¹⁹¹ are highly conserved for the C1 site. The calcium at the C1 site is co-ordinated by three carboxy groups and three carbonyl groups. The calcium at the C2 site is co-ordinated by one carboxy group and three carbonyl groups from the backbone with conserved Asp¹⁴⁸ and Asp¹⁸⁴, except in MMP-26 where the latter residue is Val¹⁸⁴. The C3 site is the most varied among these sites. The calcium at the C3 site is co-ordinated by two carboxy and two carbonyl groups in MMP-1 [28], MMP-3 [29], MMP-7 [30] and MMP-12 [33] with conserved Asp¹¹⁴, Glu¹⁸⁹ (Asp¹⁸⁹ for MMP-3) and Glu¹⁹¹ residues. However, the calcium at the C3 site of cdMMP-8 [31] and cdMMP-14 [35] is not found in their crystal structures, perhaps due to the absence of the carboxy group at this site. The calcium at the putative C3 site of MMP-26 is most likely ligated to the carboxy group of Asp¹¹⁴ and carbonyl groups of Lys¹⁸⁹ and Glu¹⁹¹.

Table 2. Comparison of crystal structures of several MMPs.

MMP-26 showed one less carboxylic acid residue at both putative low-affinity calcium-binding sites, C2 and C3 with PDB accession codes 1HFC (MMP-1) [28], 1B3D (MMP-3) [29], 1MMQ (MMP-7) [30], 1A86 (MMP-8) [31], 1JK3 (MMP-12) [33], 1BQQ (MT1-MMP) [35]. MMP-8 and MMP-14 show no calcium ion binding at the C3 site, from which Glu¹⁸⁹ is absent. Likewise, MMP-26 is negative for Glu¹⁸⁹ at the C3 site. Unlike other MMPs, MMP-26 has Val¹⁸⁴ instead of aspartic acid.

		Co-ordinating residues
Name	Calcium co-ordination	C1	C2	C3
MMP-1	Carboxy	Asp165, Asp188, Glu191	Asp184	Asp114, Glu189
	Carbonyl	Gly166, Gly168, Asn170	Asp148, Gly180, Gly182	Glu189, Glu191
			Two water molecules
MMP-3	Carboxy	Asp165, Asp188, Glu191	Asp184	Asp114, Asp189
	Carbonyl	Gly166, Gly168, Val170	Asp148, Gly180, Asn182	Asp189, Glu191
			Two water molecules	Two water molecules
MMP-7	Carboxy	Asp165, Asp188, Glu191	Asp184	Asp114, Glu189
	Carbonyl	Gly166, Gly168, Thr170	Asp148, Gly180, Gly182	Glu189, Glu191
			Two water molecules
MMP-8	Carboxy	Asp165, Asp188, Glu191	Asp184	Gln114 (amide)
	Carbonyl	Gly166, Asn168, Ile170	Asp148, Gly180, Gly182	Ala189, Glu191
MMP-12	Carboxy	Asp165, Asp188, Glu191	Asp184	Asp114, Glu189
	Carbonyl	Gly166, Gly168, Gly170	Asp148, Gly180, Gly182	Glu189, Glu191
			Two water molecules	Two water molecules
MMP-14	Carboxy	Asp165, Asp188, Glu191	Asp184	Lys114 (amide)
	Carbonyl	Gly166, Gly168, Phe170	Asp148, Asn180, Gly182	Ser189, Glu191
MMP-26	Carboxy	Asp165, Asp188, Glu191	Val184	Asp114
	Carbonyl	Gly166, Gly168, Ile170	Asp148, Gly180, Pro182	Lys189, Glu191

Analyses of sequence homology and the modelled MMP-26 structure indicate that cdMMP-26 may contain two possible binding sites with different calcium-binding affinities. The C3 site of MMP-26 may possess modest affinity when compared with the C1 site, while the C2 site may not represent a true calcium-binding site, as these studies predict only three carbonyl groups participating in the possible co-ordination of calcium at this site, analogous to the non-binding C3 sites of cdMMP-8 [31] and cdMMP-14 [35].

Refolding and removal of low-affinity Ca²⁺ ions from recombinant wild-type and mutated forms of MMP-26

Wild-type pro-MMP-26 was auto-activated during folding by dialysis [4,8,40]. Enzymatic activity measurements of the wild-type, and D114A, V184D and K189E mutants showed high rates of fluorescent synthetic peptide hydrolysis, while the D165A and E191A mutants showed very low rates of hydrolysis that were less than 5% of that seen for wild-type MMP-26. Inactive MMP-26 (E209A), with mutation at the catalytic glutamic acid residue, showed no hydrolysis activity at all. Figure 2 shows that the wild-type MMP-26 preceeded auto-cleavage of the prodomain resulting in one band around 30 kDa (pro-form) and three bands around 20 kDa (activated forms). D114A, V184D and K189E mutations resulted in similar activation patterns. However, D165A and E191A mutants remaining largely intact even show large amounts of precipitation during refolding by dialysis, and mutation at the catalytic glutamic acid residue (E209A) completely abrogated activation (Figure 2).

Figure 2. Western blotting of wild-type and mutant forms of MMP-26 after refolding.

The pro-form of MMP-26 is slightly less than 36 kDa, while the auto-activated cleavage product is seen at approx. 22 kDa. Only the E209A mutant form is observed as a single band. The cleaved forms of wild-type, and K189E, V184D and D114A mutant forms represent activated MMP-26, although the cleaved forms of D165A and E191A demonstrate less than 5% activity for the fluorescence-quenched substrate hydrolysis, when compared with that of wild-type MMP-26. An anti-FLAG tag antibody was used to detect MMP-26.

Removal of low-affinity Ca²⁺ ions by dialysis in the presence of Chelex 100 resulted in a dramatic decreases of catalytic activity for the wild-type and D114A, V184D and K189E mutant forms of MMP-26. An interesting aspect of the K189E mutant is its higher enzymatic activity than wild-type MMP-26 in the absence of low-affinity Ca²⁺ ions. The activity of LACD-MMP-26 was restored in the presence of 10 mM Ca²⁺ ions within 20 min for the wild-type, and V184D and K189E mutant forms of MMP-26, but the activity of the D114A mutant was not restored (results not shown).

Titration of high-affinity Ca²⁺ ions by Quin2

Quin2 has been utilized to probe high-affinity calcium-binding given its own high affinity (K_Q=1.9×10⁸ M⁻¹) for Ca²⁺ ions and the changes in its fluorescent properties after calcium binding [42]. Quin2 fluoresces with an emission maximum at 492 nm, which is far away from that of proteins, and its fluorescence is enhanced approx. 6-fold by the formation of calcium complexes [44]. This fluorescent property of Quin2 can be exploited to calculate the dissociation constants of high-affinity Ca²⁺ ions within the range 0.1 nM–0.1 μM [49].

To determine the true number of high-affinity Ca²⁺ ions in MMP-26, low-affinity Ca²⁺ ions were removed by repeated dialyses as described in the Materials and methods section. When LACD-MMP-26 was titrated with Quin2, the release of the high-affinity Ca²⁺ ion from MMP-26 and formation of the Quin2–Ca²⁺ complex were observed via increases in emission spectra around 490 nm with excitation at 339 nm. Fluorescence intensities at 490 nm following each addition of Quin2 were collected to estimate the total calcium concentration and the dissociation constant of the Quin2–Ca²⁺ complex, K_d(QCa) (Figure 3A). Substituting these data into eqn (1) yielded K_d(QCa) and the calcium concentration. Application of K_d(QCa) and the enzyme concentration to eqn (2) then yielded the calcium-dissociation constant of the enzyme, K_d1. The total calcium concentration estimated by these calculations was comparable with the total MMP-26 concentration, [Ca²⁺]/[MMP-26]=1.4, with the K_d1 value of 62.9 nM, which indicated that approximately one high-affinity binding site for calcium exists in MMP-26.

Figure 3. Titration of high- and low-affinity Ca²⁺ ion for measuring calcium-dissociation constants, K_d1 and K_d2.

(A) Measuring high-affinity calcium-dissociation constant, K_d1. Fluorescence intensity of Quin2 binding to Ca²⁺ ions from wild-type, and V184D and K189E mutant form of LACD-MMP-26 at 490 nm. ○, 2.5 μM wild-type LACD-MMP-26; ▲, 2.5 μM V184D mutant form of LACD-MMP-26; ▼, 5 μM K189E mutant form of LACD-MMP-26; □, background Quin-2 spectrum with water or buffer. (B) Measuring low-affinity calcium-dissociation constant, K_d2. Enzymatic activity was measured by monitoring fluorescence-quenched substrate hydrolysis after incubation with various concentrations of Ca²⁺ ions. LACD-MMP-26 (100 nM) was incubated with various concentrations of Ca²⁺ ions in 0.01% Brij, 10 mM Hepes/NaOH, pH 7.5, and 10 mM NaCl for 20 min at 25 °C. Maximal activity was taken as 100%. ○, wild-type MMP-26; ▲, V184D mutant form of LACD-MMP-26; ▼, K189E mutant form of LACD-MMP-26. Approximate negative logarithm values of the low-affinity calcium-dissociation constant for wild-type, and V184D and K189E mutants are 3.93, 3.6 and 4.55 respectively.

The C2 and C3 residues unique to MMP-26, when compared with other MMPs, were mutated separately to introduce a new carboxy group and enhance Ca²⁺ co-ordination. These mutations facilitated exploration of the possibility of enhanced binding affinity at these sites and thus their possible titration by Quin2. The V184D (C2 site) and K189E (C3 site) mutants yielded K_d1 values of 63.4 nM and 58.5 nM for the C1 high-affinity Ca²⁺-binding site respectively, without changing the calcium/enzyme concentration ratio (Table 3). Therefore these mutations at putative low-affinity calcium-binding sites support the presence of a single high-affinity Ca²⁺ ion at the C1 site.

Table 3. Calcium dissociation constants.

High- and low-affinity calcium dissociation constants, K_d1 and K_d2 for wild-type, and V184D and K189E mutant forms of MMP-26.

	Ca2+ dissociation constants
	Kd1	Kd2
Wild-type	62.9±13.0 nM	0.12 mM
V184D	63.4±16.2 nM	0.24 mM
K189E	62.9±13.0 nM	0.028 mM

Titration of low-affinity Ca²⁺ ions

LACD-MMP-26 was then titrated with Ca²⁺ ions, and its activity was monitored to estimate the low-affinity calcium-dissociation constant, K_d2. To determine the number of low-affinity calcium-binding sites, the K189E and V184D mutants were similarly titrated and monitored. As both mutants contained an additional carboxy group, either mutation could be expected to result in enhanced calcium-binding affinity. As the calcium concentrations at which 50% enzymatic activity was observed were much greater than the enzyme concentrations, the necessary assumption for calculations of the low-affinity calcium-dissociation constant, K_d2, was satisfied. The resultant K_d2 value is equal to the EC₅₀ value. The K_d2 values for wild-type, and V184D and K189E mutant forms were 0.12 mM, 0.24 mM and 0.028 mM respectively (Figure 3B and Table 3). The K189E mutant showed increased affinity for Ca²⁺ ions, whereas the V184D mutant showed the opposite result. The K189E mutation caused an approx. 4-fold decrease in the K_d2 value, whereas the V184D mutation displayed an approx. 2-fold increase in the K_d2 value when compared with the wild-type. According to the effects of calcium concentration on the activity of the wild-type and mutant forms, the putative C2 low-affinity calcium-binding site, if it exists, did not appear to be involved in enzymatic catalysis, and only a single low-affinity calcium-binding site, C3, affects enzymatic activity.

ANS binding to MMP-26

ANS is a useful indicator of hydrophobic regions in a protein. Used as a fluorescent probe, ANS shows an increase in the quantum yield of fluorescence and a blue shift of the emission spectrum by interaction with the hydrophobic regions in a protein. A common application of ANS is the detection of folding intermediates or unfolded states of protein [50]. In buffer, ANS has little fluorescent intensity, and incubation with tightly folded globular proteins such as carbonic anhydrase, which lacks specific binding sites for ANS, results in no change of the wavelength of maximum fluorescence and intensity. However, incubation with lipid-binding proteins such as apolipoprotein A-I induces an enhancement in ANS fluorescence intensity together with a 35 nm blue shift in its wavelength of maximum emission [50]. Wild-type, and K189E and V184D mutant forms of LACD-MMP-26 were prepared as described in the Materials and methods section. Following the addition of ANS, the formation of ANS–enzyme complexes generated spectra revealing not only an increase in fluorescence intensity, but also blue shifts of the maximum emission peak when compared with background spectra, thereby providing evidence for the existence of hydrophobic areas in MMP-26 (background is shown for wild-type MMP-26, Figure 4A, d). Moreover, conformational changes of enzymes upon binding to Ca²⁺ resulted in increased hydrophobic surface area and a subsequent increase in fluorescence spectra as ANS binds to the enzyme (Figure 4A, a–c). These observations indicated that the hydrophobic regions of MMP-26 accessible to ANS could be altered by the presence or absence of Ca²⁺ ions, which resulted in conformational changes. Monitoring spectra at 480 nm demonstrated a difference in low-affinity calcium effects among the wild-type and mutant forms (Figure 4B). The points of 50% tertiary structural changes were similar for the K189E and V184D mutant forms compared with wild-type MMP-26 following the addition of Ca²⁺ ions; however, recovery patterns were different. In comparison with wild-type MMP-26, the K189E mutant showed a greater response at low calcium concentrations and smooth changes of its structure, whereas the V184D mutant showed a smaller response at low calcium concentrations and sudden changes of its structure. LACD-MMP-26 was treated with 2 mM EGTA to remove high-affinity Ca²⁺ ion (Figure 4A, d). It revealed that ANS spectra with high-affinity calcium-associated MMP-26 (solid line) are identical with ANS spectra with high-affinity calcium-dissociated MMP-26 (dashed-line), with no changes in fluorescence intensity. This suggested that loss of the high-affinity Ca²⁺ ion did not alter the hydrophobic surface area of MMP-26.

Figure 4. ANS binding to MMP-26; depending on low-affinity Ca²⁺ ions.

(A) Fluorescence spectra from the mixtures of ANS and protein: 3.4 μM wild-type LACD-MMP-26 and 100 μM ANS in the presence of various concentrations of Ca²⁺ ions (a); 2 μM K189E mutant form of LACD-MMP-26 and 100 μM ANS in the presence of various concentrations of Ca²⁺ ions (b); 2 μM V184D mutant form of LACD-MMP-26 and 100 μM ANS in the presence of various concentrations of a calcium solution (c); 2 μM LACD-MMP-26 and 100 μM ANS before (solid line) or after (broken line) addition of 2 mM EGTA (d). Background spectra were taken from MMP-26 only or ANS only. (B) Relative fluorescence intensity at 480 nm. ○ and solid line, wild-type MMP-26; ▲ and dashed line, V184D mutant form of MMP-26; ▼ and dashed-dotted line, K189E mutant form of MMP-26.

Effects of Ca²⁺ on secondary structure and thermal denaturation by CD spectroscopy

The effects of Ca²⁺ ions on the secondary structure of MMP-26 were explored by measuring CD spectra in the presence or absence of low- and low/high-affinity Ca²⁺ ions. LACD-MMP-26 was prepared, and then CD spectra were measured with calciumfree buffer (Figure 5A, ●), buffer containing 10 mM EGTA as a Ca²⁺ ion chelator (Figure 5A, ○), or 10 mM CaCl₂ (Figure 5A, □). No secondary-structural changes were detected when calciumassociated MMP-26 was compared with calcium-dissociated MMP-26 at 25 °C. To study the effects of Ca²⁺ ions on MMP-26 stability, CD spectra were obtained at 218 nm as increasing temperatures were utilized to induce thermal denaturation. Increased temperature results in an increase of the apparent fractional extent of unfolding, f_app, and melting points were measured when f_app was 0.5 (Figure 5B). Results showed that the melting points of high-affinity calcium-associated and -dissociated MMP-26 were similar, at 32 and 30 °C respectively. However, the magnitude of the denaturation caused by increased temperature was greater for high-affinity calcium-dissociated MMP-26 (Figure 5B, ○) than for high-affinity calcium-associated MMP-26 (Figure 5B, ●). The 30–32 °C melting temperature range is lower than the known range of melting temperatures for other MMPs [51]. This might be caused by the absence of low-affinity Ca²⁺ ion in MMP-26. Unfortunately, the addition of Ca²⁺ ions resulted in precipitation during the temperature adjustments, and the effects of the low-affinity Ca²⁺ ions against thermal denaturation could not be determined. CD spectroscopy shows that the high-affinity Ca²⁺ ion did not play a pivotal role in stabilizing the secondary structure of MMP-26 at 25 °C and might be involved in the stability of MMP-26 during folding.

Figure 5. CD spectroscopy of MMP-26 associated or dissociated with calcium binding and thermal denaturation.

(A) CD spectroscopy of MMP-26 associated or dissociated with Ca²⁺ ion binding. LACD-MMP-26 (5 μM) in 10 mM Tris/HCl, pH 7.5, 10 mM NaCl and 0.01% Brij-35 (●) was incubated with 2 mM EGTA (○) or 10 mM CaCl₂ (□). CD spectroscopy revealed no chages in secondary structure associated or dissociated with Ca²⁺ ion binding. (B) Thermal denaturation of LACD-MMP-26 associated or dissociated with high-affinity Ca²⁺ ion binding. ●, 10 μM LACD-MMP-26 dialysed in calcium-free buffer; ○, 10 μM LACD-MMP-26 treated with 2 mM EGTA for 4 h. f_app is the apparent fractional extent of unfolding. Melting points for high-affinity calcium-associated and -dissociated LACD-MMP-26 are 32 and 30 °C respectively.

Modified Boyden chamber invasion assay

Three stably transfected cell lines expressing wild-type, and K189E and E209A mutant forms of MMP-26 invaded through type IV collagen in vitro during cell invasion assays in the presence or absence of Ca²⁺ ions (Figure 6). For the wild-type MMP-26-transfected cell line under calcium-minus conditions there was a significant (P=0.001) decrease in invasion, while the K189E- and E209A-transfected cell lines showed no significant changes in invasive potential (P=0.84 and 0.61 respectively). Thus, while Ca²⁺ ions increased the invasiveness of a wild-type MMP-26-transfected cell line, they failed to alter the invasive potential of either the K189E or E209A MMP-26 mutant-form-transfected cell line. Comparisons of invasiveness between the wild-type MMP-26-transfected cell line under calcium-minus conditions with that of the E209A MMP-26 mutant-form-transfected cell line under calcium-minus or calcium-plus conditions showed insignificant differences (P=0.89 and 0.52 respectively). Comparison of invasiveness between the wild-type MMP-26-transfected cell line under calcium-minus conditions with the K189E MMP-26 mutant-form-transfected cell line showed a significant increase of invasive potential (P=0.0002). Cells transfected with the K189E mutant form demonstrated the high invasive potential in either case, independently of the calcium-plus or calcium-minus conditions.

Figure 6. Invasive potential of wild-type, and K189E and E209A mutant forms of MMP-26-transfected cell lines.

Invasion assays were performed with modified Boyden chambers, and the percentage of invading cells was quantified as described in the Materials and methods section. Invasion assays were performed under calcium-plus or calcium-minus conditions. Results are means±S.D. for four separate experiments for each group.

DISCUSSION

The EC₅₀ value of MMP-26 for calcium titration is several hundred times higher than physiological calcium concentrations ranges, which vary from 300 to 1200 nM inside normal cells [26], and therefore normal intracellular conditions serve to maintain MMP-26 in an inactive state. It is when MMP-26 is secreted from the cell or exposed to a transient intracellular calcium influx that it can adopt an active conformation.

When combined with data acquired from inhibition kinetics studies, a previously built MMP-26 homology-modelling structure was useful in predicting the presence of an intermediate-size S₁′ pocket at the MMP-26 active site [47]. This MMP-26 structure also revealed a high-affinity calcium-binding site, C1, and putative C2 and C3 low-affinity calcium-binding sites (Figure 1A). The lack of calcium at the analogous C3 site of MMP-8 and MMP-14 as revealed by the crystal structures suggested that carbonyl oxygen atoms alone might not possess sufficient affinity to retain Ca²⁺ ions in the catalytic domain of the MMPs (Figure 1B and Table 2). This suggested further that the C2 site in MMP-26 might not co-ordinate with Ca²⁺ ions. Hence homology modelling and sequence alignment predicted that MMP-26 might have one high-affinity and one low-affinity calcium-binding site at C1 and C3 respectively.

Essential for constructing an enzymatically active structure during the folding of MMP-26 is its ability to bind high-affinity Ca²⁺ ions. Mutations at the C1 site (removal of one carboxylic acid ligand, D165A or E191A) resulted in a loss of enzymatic activity using a fluorescence-quenched peptide hydrolysis assay, suggesting that C1 is a true high-affinity calcium-binding site. Quin2 titration of wild-type, and V184D (C2 site) and K189E mutant (C3 site) forms of LACD-MMP-26 supported the presence of a single high-affinity calcium-binding site, C1 (Table 3). Quin2 titration of V184D and K189E mutant forms also suggested that mutations at the putative C2 and C3 low-affinity calcium-binding sites did not generate fluctuations in the high-affinity calcium-binding constant, nor did the addition of a carboxylic acid moiety at these sites fortify calcium affinity at these mutated sites to the levels observed at the high-affinity calcium-binding site. Therefore mutations of the putative low-affinity calcium-binding sites seemed to leave high-affinity calcium binding at the C1 site unaffected. Removal of high-affinity Ca²⁺ ions from LACD-MMP-26 did not induce conformational changes related to the hydrophobic surface area of the protein (Figure 4A, d), nor did it affect the secondary structure (Figure 5A). However, it stabilized secondary structures as observed in thermal denaturation (Figure 5B). Hence, although high-affinity Ca²⁺ ions might be involved in the folding process of MMP-26, apparently they were not essential for maintenance of its secondary structures.

Low-affinity calcium binding to MMP-26 is a reversible and equilibrium state. Low-affinity calcium association and dissociation not only regulated enzymatic activity, but also modulated its tertiary structure, with resultant changes in total hydrophobic surface area. Computer modelling revealed geometric changes at the catalytic site following removal of the low-affinity Ca²⁺ ion, which was actually quite distant from the catalytic site. Figure 7(A) shows the superimposition of two structures with or without the low-affinity Ca²⁺ ions. The elimination of the low-affinity Ca²⁺ ion induced a shift in the active site (small grey spheres at the centre stand for catalytic Zn²⁺ ions). As a result, the structures of the substrate-binding region, and of the catalytic site specifically, were perturbed. Molecular modelling thus supported the assertion that changes at the low-affinity calcium-binding site could result in distortion of the MMP-26 active site.

Figure 7. Modelled structures of MMP-26 with or without low-affinity Ca²⁺ ions, and superimposed structures of wild-type/V184D and wild-type/K189E mutant MMP-26.

(A) Modelled structure of MMP-26 with or without low-affinity Ca²⁺ ions. The grey solid ribbon structure represents the minimized MMP-26 structure in the presence of low-affinity Ca²⁺ ions, while the dark-grey solid ribbon structure represents that in the absence of these ions resulting in LACD-MMP-26. The large grey spheres, which were designated C2 and C3 and were eliminated for further minimization, are putative low-affinity Ca²⁺ ions. The dark-grey sphere, designated C1, is a high-affinity Ca²⁺ ion. Small grey and dark-grey spheres at the centre of MMP-26 are the catalytic Zn²⁺ ions before and after elimination of the low-affinity Ca²⁺ ions respectively. Small grey and dark-grey spheres at the top of MMP-26 are the structural Zn²⁺ ions. The RMSD value between the backbones of the wild-type and LACD-MMP-26 structure is 1.23 Å. (B) Superimposition of the wild-type and V184D mutant MMP-26 structures. The grey ribbon structure represents the minimized structure of wild-type MMP-26 and the black ribbon structure represents the minimized structure of the MMP-26 mutant forms. The inhibitor, GM6001, is bound to the catalytic site of MMP-26. The V184D mutation alters both the C2 site and the active site of MMP-26 when compared with wild-type MMP-26, resulting in a perturbed orientation of the bound inhibitor. The RMSD value between the backbones of the wild-type and V184D mutant MMP-26 structure is 0.56 Å. (C) Superimposed structure of wild-type/K189E mutant MMP-26. Protein representation follows the same way of V184D mutant representation above. The K189E mutation does not alter the C3 site or the active site of MMP-26 when compared with the wild-type, and there is no change in orientation of the bound inhibitor (GM6001). The RMSD value between the backbones of the wild-type and K189E mutant MMP-26 structure is 0.15 Å.

A mutation at the C2 site (V184D), which adds a carboxylic acid ligand and leads to co-ordinating residues similar to the C2 sites of other MMPs, resulted in an increase in the low-affinity calcium-dissociation constant (K_d2, 0.24 mM). Therefore V184D mutation revealed a higher calcium requirement for enzymatic activity when compared with wild-type MMP-26. ANS spectra revealed that the calcium concentrations at which tertiary structural changes within the V184D mutant were first induced is higher than that for wild-type MMP-26, but this resistance to change was overcome in the presence of additional Ca²⁺ ions. This might be the result of competition between the C3 and artificial C2 sites for Ca²⁺ ions following this mutation. Molecular modelling of the V184D mutant showed that the introduction of a carboxylic acid moiety (Asp¹⁸⁴) at the C2 site might initiate calcium binding with subsequent changes at the C2 site and the catalytic site of MMP-26, resulting in a new binding orientation for the inhibitor, GM6001 (Figure 7B). Because the C2 and C3 sites share the same β-sheet as the artificial Asp¹⁸⁴ residue, these sites are co-operative. Therefore calcium binding at the C2 site in the absence of calcium binding at the C3 site might increase perturbation of the catalytic site, which persisted until calcium binding at the C3 site relieves this distortion. This structural distortion might be the reason that calcium-binding affinity enforcement at the C2 site by the additional Ca²⁺ co-ordinate (Asp¹⁸⁴) increases the Ca²⁺ requirement for enzymatic activity and rapid tertiary structural changes induced by calcium binding. These data combined with sequence analysis suggest that C2 is not a true low-affinity calcium-binding site.

D114A mutation at the C3 site resulted in irreversible loss of enzymatic activity triggered by low-affinity calcium-binding. As long as Ca²⁺ ions were present during folding, the D114A mutant form of MMP-26 retained enzymatic activity, but once the Ca²⁺ ions were removed, enzymatic activity could not be recovered even if the Ca²⁺ ions were restored. K189E mutation at the C3 site resulted in a decrease of the low-affinity calcium-dissociation constant (K_d2, 0.028 mM). Therefore a K189E mutation at C3 site revealed a lower Ca²⁺ requirement for enzymatic activity compared with the wild-type. Moreover ANS-binding spectra showed that the tertiary structure of the K89E mutant changed initially at lower calcium concentrations than that for wild-type MMP-26. Therefore the K189E mutation successfully fortified calcium-binding affinity at C3 site. According to molecular modelling of the K189E mutant, addition of a carboxylic acid moiety (from Glu¹⁸⁹) by mutation did not alter either the C3 calcium-binding site or the catalytic site of MMP-26, resulting in no change in orientation of the inhibitor, GM6001. Nevertheless, the addition of the Ca²⁺ co-ordinate (Glu¹⁸⁹) at the C3 site did result in enhanced calcium binding (Figure 7C).

Results of the invasion assays corresponded to the kinetics data. As the E209A mutation resulted in a loss of enzymatic activity owing to mutation of the catalytic Glu²⁰⁹, a cell line transfected with the E209A mutant form of MMP-26 showed that decreased invasive potential was independent of Ca²⁺ ions, resulting in an invasive potential similar to that of the wild-type MMP-26-transfected cell line under calcium-minus conditions. As Ca²⁺ ions restored enzymatic activity for wild-type MMP-26, Ca²⁺ ions also increased the invasive potential of the wild-type MMP-26-transfected cell line, but this was not true for the E209A-transfected cell line. In contrast with these findings, cells transfected with the K189E mutant form of MMP-26 showed a high invasive potential that was independent of Ca²⁺ ions.

In summary, MMP-26 has one high-affinity calcium-binding site, designated here as C1, and one low-affinity calcium-binding site, designated C3, with 62.9 nM and 120 μM dissociation constants respectively. High-affinity Ca²⁺ ion binding stabilizes secondary-structural features during folding, whereas low-affinity Ca²⁺ ion binding serves to regulate enzymatic activity through changes in enzyme tertiary structure. The D165A or E191A mutations at the C1 site show a loss of enzymatic function, while the D114A mutation at the C3 site demonstrates an irreversible loss of enzymatic function triggered by low-affinity calcium binding. The K189E mutation at C3 showed lower calcium requirements for regulation of enzymatic activity and changes in tertiary structure when compared with wild-type MMP-26, whereas the V184D mutation at the C2 site showed opposite effects. Invasion assays support a role for Ca²⁺ ions in the regulation of ECM degradation and cell invasion. These results suggest that the enzymatic activity of MMP-26 might be regulated by calcium concentrations in cellular systems.