Matrix metalloproteinases: Fold and function of their catalytic domains

Biochimica et Biophysica Acta 1803 (2010) 20–28

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / b b a m c r

Review

Matrix metalloproteinases: Fold and function of their catalytic domains Cynthia Tallant 1, Aniebrys Marrero 1, F.Xavier Gomis-Rüth ⁎ Proteolysis Lab, Molecular Biology Institute of Barcelona, CSIC, Barcelona Science Park, Helix Building, c/Baldiri Reixac, 15-21, E-08028 Barcelona, Spain

a r t i c l e

i n f o

Article history: Received 24 February 2009 Received in revised form 7 April 2009 Accepted 7 April 2009 Available online 15 April 2009 Keywords: Zinc metalloproteinase Metzincin MMP Vertebrate collagenase Three-dimensional structure Catalytic domain X-ray crystal structure Metallopeptidase

a b s t r a c t Matrix metalloproteinases (MMPs) are zinc-dependent protein and peptide hydrolases. They have been almost exclusively studied in vertebrates and 23 paralogs are present in humans. They are widely involved in metabolism regulation through both extensive protein degradation and selective peptide-bond hydrolysis. If MMPs are not subjected to exquisite spatial and temporal control, they become destructive, which can lead to pathologies such as arthritis, inflammation, and cancer. The main therapeutic strategy to combat the dysregulation of MMPs is the design of drugs to target their catalytic domains, for which purpose detailed structural knowledge is essential. The catalytic domains of 13 MMPs have been structurally analyzed so far and they belong to the “metzincin” clan of metalloendopeptidases. These compact, spherical, ∼ 165-residue molecules are divided by a shallow substrate-binding crevice into an upper and a lower sub-domain. The molecules have an extended zinc-binding motif, HEXXHXXGXXH, which contains three zinc-binding histidines and a glutamate that acts as a general base/acid during catalysis. In addition, a conserved methionine lying within a “Met-turn” provides a hydrophobic base for the zinc-binding site. Further earmarks of MMPs are three α-helices and a five-stranded β-sheet, as well as at least two calcium sites and a second zinc site with structural functions. Most MMPs are secreted as inactive zymogens with an N-terminal ∼ 80-residue pro-domain, which folds into a three-helix globular domain and inhibits the catalytic zinc through a cysteine imbedded in a conserved motif, PRCGXPD. Removal of the pro-domain enables access of a catalytic solvent molecule and substrate molecules to the active-site cleft, which harbors a hydrophobic S1′pocket as main determinant of specificity. Together with the catalytic zinc ion, this pocket has been targeted since the onset of drug development against MMPs. However, the inability of first- and second-generation inhibitors to distinguish between different MMPs led to failures in clinical trials. More recent approaches have produced highly specific inhibitors to tackle selected MMPs, thus anticipating the development of more successful drugs in the near future. Further strategies should include the detailed structural characterization of the remaining ten MMPs to assist in achieving higher drug selectivity. In this review, we discuss the general architecture of MMP catalytic domains and its implication in function, zymogenic activation, and drug design. © 2009 Elsevier B.V. All rights reserved.

1. Physiological background MMPs, also called matrixins, vertebrate collagenases, and matrix metalloproteinases, are secreted or membrane-bound zinc-dependent protein and peptide hydrolases, which constitute a separate family within the “metzincin” clan of metallopeptidases (MPs) [1–6]. They

Abbreviations: ADAM, a disintegrin and metalloprotease; ADAMTS, ADAM with thrombospondin-like repeats; CTS, C-terminal sub-domain; ECM, extracellular matrix; MP, metallopeptidase; MMP, matrix metalloproteinase alias vertebrate collagenase and matrixin; MT-MMP, membrane-type MMP; NTS, N-terminal sub-domain; PDB, Protein Data Bank (http://www.pdb.org) access code for three-dimensional structure coordinates; RECK, reversion-inducing cysteine-rich protein with Kazal motifs; S1, S2,… and P1, P2,…, denote protease active-site cleft and substrate subsites, respectively, N-terminally of the scissile peptide bond; S1′, S2′,… and P1′, P2′,…, at the C-terminus of the scission, in accordance with [37]; TIMP, tissue inhibitor of metalloproteinases ⁎ Corresponding author. Tel.: +34 934020186; fax: +34 934034979. E-mail address: [email protected] (F.X. Gomis-Rüth). 1 These authors contributed equally and share first authorship. 0167-4889/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.bbamcr.2009.04.003

were discovered 47 years ago as the agents responsible for tail resorption during frog metamorphosis [7–9] and have since been identified as the main processors of extracellular matrix (ECM) components [2]. As such, they participate in physiological processes involving tissue turnover and repair during blastocyst implantation, ovulation, postpartum and post-lactational involution, bone resorption, wound healing, etc. Such processing capacity is also required during embryogenesis and angiogenesis. More recently, MMPs have also been implicated in more sophisticated processes than mere ECM turnover [10,11]. These include the activation or inactivation of other proteins through limited proteolysis of selected bonds, as well as the shedding of membrane-anchored forms into circulation. Substrates include other (pro-)proteases, protease inhibitors, clotting factors, antimicrobial peptides, chemotactic and adhesion molecules, and growth factors, hormones, cytokines, as well as their receptors and binding proteins. In such shedding functions, MMPs overlap in substrate specificity, spatial and temporal location, and in their proclivity to cleave upstream of hydrophobic residues, typically

C. Tallant et al. / Biochimica et Biophysica Acta 1803 (2010) 20–28

21

leucine, with members of another metzincin family, the ADAMs/ adamalysins (ADAM is an acronym for a disintegrin and metalloprotease; [12–14]). If this formidable proteolytic potential of MMPs is not subjected to fine-tuned control, dysregulation can give rise to degradation and ensuing pathologies. This happens during inflammation, ulceration, arthritis, periodontitis, cardiovascular disease, fibrosis, and emphysema. Further consequences of deregulated MMP activity include cancer-associated processes like tumorigenesis and tumor neovascularisation, invasion and metastasis, as well as apoptosis, intestinal defense activation, and pathologies of the nervous system, such as stroke and Alzheimer's disease [15].

Table 2 Access codes corresponding to experimental three-dimensional structures containing at least the catalytic domain of MMPs as deposited with the Protein Data Bank (www. ebi.ac.uk/msd/index.html or www.pdb.org; closing date: January 26, 2009).

2. Classification and modular organization

Gelatinase B (MMP-9) (H. sapiens) Stromelysin-2 (MMP-10) (H. sapiens) Stromelysin-3 (MMP-11) (Mus musculus) Macrophage elastase (MMP-12) (H. sapiens)

Open reading frames with high sequence similarity to MMP catalytic domains are present across almost all kingdoms of life, from archaebacteria to sea urchins, and from viruses to worms and plants [3]. In humans, there are 23 paralogs (numbered 1 to 3, 7 to 17, 19 to 21, and 23 to 28 for historical reasons; there are two identical forms for MMP-23, encoded by two genes, mmp-23a and mmpP-23b, see http://degradome.uniovi.es/met.html#M10; [16,17]). MMPs can be divided into: (i) true collagenases, which cut triple-helical collagen at a single site across the three chains, giving rise to products 3/4 and 1/4 of the length of the original molecule; (ii) gelatinases, which target denatured collagens and gelatins; and (iii) stromelysins, which have broad specificity and may degrade proteoglycans. Another classification identifies matrilysins, archetypal MMPs, convertase-activatable MMPs, and gelatinases [11,15,18–23]. MMPs are mosaic proteins, each constituted by a modular combination of inserts and domains. These may include, from N- to C-terminus, a signal peptide for secretion; a ∼ 80-residue zymogenic pro-peptide; a ∼ 165-residue zinc- and calcium-dependent catalytic MP domain (Table 1); a 15-to-65-residue linker region; and a ∼ 200-residue hemopexin-like domain for collagen binding, pro-MMP activation, and dimerization [24,25]. Further possible insertions include fibronectin type-II-related domains for interaction with collagens

Fibroblast collagenase (MMP-1) (Homo sapiens ; Sus scrofa) Gelatinase A (MMP-2) (H. sapiens) Stromelysin-1 (MMP-3) (H. sapiens)

Matrilysin (MMP-7) (H. sapiens) Neutrophil collagenase (MMP-8) (H. sapiens)

Collagenase-3 (MMP-13) (H. sapiens; M. musculus) MT1-MMP (MMP-14) (H. sapiens) MT3-MMP (MMP-16) (H. sapiens) Enamelysin (MMP-20) (H. sapiens)

966c 1ayk 1cge 1cgf 1cgl 1fbl 1hfc 1mnc 2ayk 2tcl 3ayk 4ayk 1su3 2j0t 2clt 1ck7 1hov 1qib 1gxd 1eak 1b3d 1b8y 1biw 1bm6 1bqo 1c3i 1caq 1ciz 1cqr 3usn 1uea 2d1o 1c8t 2jt6 2jt5 2jnp 1d5j 1d7x 1d8f 1d8m 1g05 1g49 1g4k 1hfs 1hy7 1sln 1slm 1qia 1qic 1ums 1umt 1usn 2srt 2usn 1 mmp 1 mmq 1 mmr 2ddy 1jan 1a85 1a86 1bzs 1i73 1i76 1jao 1jap 1jaq 1jh1 1jj9 1 kbc 1 mmb 1mnc 1zvx 2oy4 1zs0 1zp5 2oy2 1jh1 3dpe 3dpf 3dng 1gkc 1gkd 1l6j 2ow0 2ow1 2ow2 2ovz 2ovx 1q3a 1hv5 1jiz 1jk3 2z2d 2k2g 3ba0 2oxu 2oxz 2oxw 2hu6 1z3j 1ycm 1y93 1rmz 1os9 1os2 2poj 1utz 1utt 1ros 2w0d 3ehx 3ehy 1zrg 2jpv 3f15 3f16 3f17 3f18 3f19 3f1a 2k9c 2jxy 456c 830c 1cxv 1eub 1fls 1fm1 2pjt 2ozr 2ow9 2e2d 2d1n 1ztq 1you 1xur 1xud 1xuc 3elm 1bqq 1buv 1rm8 2jsd

and gelatins, uniquely found in gelatinases A (MMP-2) and B (MMP9); a collagen type-V-like insert; a vitronectin-like insertion domain; a cysteine-rich, proline-rich and interleukin-1 receptor-like domain; an immunoglobulin-like domain; a glycosyl phosphatidylinositol linkage signal; a type-I or type-II trans-membrane domain; a stem or “linker 2” region downstream of the hemopexin-like domain; and a cytoplasmic tail [3,20,25,26]. Those MMPs bearing a membrane anchor give rise to the MT-MMP subfamily (MT1-MMP to MT6-MMP; in plain MMP nomenclature they are, respectively, MMP-14 to-17, MMP-24, and MMP-25). 3. Activity regulation

Table 1 Statistics on the pro- and catalytic domains of structurally characterized MMPs. MMP

Access code⁎

Catalytic domain

Catalytic domain length

NTS residues+

CTS residues++

Pro-domain length⁎⁎

1 2 3 7 8 9 10 11 12 13 14 16 20

P03956 P08253 P08254 P09237 P22894 P14780 P09238 P24347 P39900 P45452 P50281 P51512 O60882

F100–G261 Y110–G446# F100–G264 Y95–G259 F99–G262 F107–G444# F99–G263 F98–G258 F100–G263 Y104–G267 Y112–G284 Y120–G291 Y108–G271

162 160 165 165 164 160 165 161 164 164 173 172 164

126 124 126 127 126 124 126 130 126 126 135 134 126

36 36 39 38 38 36 39 31 38 38 38 38 38

80 80 82 77 79 74 81 66 83 84 91 88 85

⁎ Sequence entry according to UniProt (www.uniprot.org). The extent of the prodomain as annotated in the entry does not always correspond to the experimentally determined sequence stretch. Only the human orthologs were considered. ⁎⁎ Based on the estimated length of the signal peptide and the N-terminus of the mature protease. + Residues between the N-terminus and the glycine of the zinc-binding consensus sequence. ++ Residues between the glycine of the zinc-binding consensus sequence and the C-terminal glycine at the end of helix αC. # Gelatinases A (MMP-2) and B (MMP-9) have three fibronectin type-II repeats totaling 177 and 178 residues, respectively, inserted into their catalytic domains within LβVαB.

MMPs are regulated via modulation of gene expression, compartmentalization, and inhibition by protein inhibitors. Most MMPs are not constitutively transcribed, but are expressed after external induction by cytokines and growth factors. In addition, some MMPs are stored in inflammatory cell granules, which restrict their compass of action. Another regulatory mechanism is provided by zymogenicity. All MMPs except MMP-23 comprise a pro-domain upstream of the catalytic domain, and activation proceeds through its removal (see Section 7.). This is catalyzed by furin-like pro-hormone convertases, by members of the plasminogen/plasmin cascade or by other MMPs [27–31]. Once MMPs have been released to the extracellular space or anchored to the membrane and activated, they are kept in check by α2-macroglobulin (non-specifically) and by their endogenous tissue inhibitors, the co-secreted or ECM-anchored 22to-29-kDa tissue inhibitors of metalloproteinases (TIMPs; [2,32]). Four different forms (TIMP-1 to TIMP-4) have been described to date. They inhibit active MMPs with relatively low selectivity, forming tight 1:1 complexes and also participate in pro-MMP activation; suppression of tumour-growth, invasion and metastasis; morphology modulation; cell-growth promotion; matrix binding; inhibition of angiogenesis; and induction of apoptosis. They further exhibit growth factor-like, mitogenic, and steroidogenic activities [33–35]. In addition to MMPs, TIMPs can inhibit ADAMs and ADAMTSs, which contain thrombospondin (TS)-like domains. In particular TIMP-3 is a major global regulator of MP activity (see [3] and references therein). More recently, other MMP inhibitors have been

22

C. Tallant et al. / Biochimica et Biophysica Acta 1803 (2010) 20–28

suggested, such as β-amyloid precursor protein; the reversioninducing cysteine-rich protein with Kazal motifs, RECK; tissuefactor-pathway-inhibitor 2; and procollagen C-terminal proteinase enhancer. However, the hypothetic mechanisms of action of the latter inhibitors are still unknown [2]. 4. Overall structure of MMP catalytic domains MMPs have been object of a number of structural studies for the last 15 years [25,36]. To date, X-ray or NMR structures comprising at least the catalytic domains, isolated or in complexes with inhibitors, are available for MMP-1-to-MMP-3, MMP-7-to-MMP-14, MMP-16, and MMP-20 (see Table 2). The catalytic domain structures are very similar, in the shape of a sphere with a diameter of ∼ 40 Å (Fig. 1A). A shallow active-site cleft lies on the front surface, which entails that substrates bind in an approximately extended conformation according to the standard orientation of MPs used here. This orientation entails that a substrate binds horizontally from left (N-terminal non-primed side) to right (C-terminal primed side) of the catalytic metal ion [37,38]. The cleft bifurcates the molecule asymmetrically into an upper N-terminal (NTS, 127 residues on average; see Table 1) and a lower Cterminal sub-domain (CTS; 37 amino acids on average). No disulfide bonds are present in any structure and, as foreseeable in a protein with preceding and downstream domains, both termini are surface located, at the lower left of the molecule (Fig. 1A, B).

Depending on the construct employed, the N-terminal α-amino group of the mature enzymes is either salt-bridged to Asp232 (numbering and description hereafter correspond to MMP-8 unless otherwise stated; see Protein Data Bank (PDB) entry 1jan; [39]), the first of two conserved vicinal aspartate residues at the beginning of αhelix αC, the “C-terminal helix” (construct Phe79–Gly242 in the case of MMP-8; see Fig. 1A) or free and undefined by electron density (e.g., until Pro86 in MMP-8 construct Met80–Gly242; [40]). In the first case, which represents the physiologically relevant form, the stabilization provided by the salt-bridge renders a “superactive” form, with fourfold-higher activity than the shorter construct in the case of MMP-8 [39]. In several MMP structures, this salt-bridge is not formed, thus giving rise to disparate locations of their N-termini (see Fig. 3A in [25]). In any case, the polypeptide chain subsequently creeps upwards along the molecular surface to enter the NTS and a five-stranded twisted β-sheet that parallels and delimits the active-site cleft on its top. All strands except the fourth one are parallel and run left to right (βI–βV; connectivity − 1x,+2x,+2,− 1; see Fig. 1A, B). The sheet accumulates a clockwise twist of ∼70° from back to front between βII and βIV, but is only slightly arched. Two helices (αA, the “backing helix”, and αB, the “active-site helix”; Fig. 1A, B) nestle into the concave side of the sheet. They make contact at Ala116 from αA and Ala196 from αB and their axes lie in the same plane, rotated ∼35° away. The residues at the interface between the sheet and the helices are mainly hydrophobic and give rise to an extended central

Fig. 1. MMP catalytic domain structure. (A) Stereographic Richardson-plot of the catalytic domain of human MMP-8 (Phe79–Gly242) shown in standard orientation (PDB 1jan; [39]). The repetitive secondary structure elements (orange arrows for β-strands, βI–βV; cyan ribbons for α-helices, αA-αC) and the four cations (two zinc ions in magenta and two calcium ions in red) are depicted. The side chains of the zinc-binding histidines, the general base/acid glutamate, the Met-turn methionine, and residues engaged in key electrostatic interactions (grey dots) within the CTS are shown as stick models with yellow carbons and labeled. A substrate of sequence Pro-Leu-Gly-Leu-Ala, modeled based on published inhibitor structures (see [40]), is further shown as a stick model with grey carbons. Additional relevant chain segments are shown in distinct colors and labeled (Met-turn in green; specificity loop in red; S1′-wall-forming segment in blue; S-loop in purple; and bulge-edge segment in magenta). (B) Topology scheme of MMP-8 in the same orientation as in (A). (C) Close-up view of (A) depicting the side chains engaged in zinc binding and those shaping the specificity pocket, which are labeled. The figure was prepared with programs CARRARA4 and SETOR [89].

C. Tallant et al. / Biochimica et Biophysica Acta 1803 (2010) 20–28

23

Fig. 2. Catalytic mechanism of MMPs. Scheme for the cleavage mechanism proposed for MMPs, with the catalytic zinc ion as a sphere and hydrogen bonds as dashed lines. The three histidine ligands are represented by sticks. One conceivable alternative is that the second proton is transferred directly from the gem-diolate to the leaving amine in II. and not via the general base/acid glutamate. This proton transfer could hypothetically occur before or after scissile-bond cleavage.

hydrophobic core. On the convex side of the sheet, three elements protrude from the molecular surface: the loop connecting strands βII and βIII (LβIIβIII), LβIIIβIV, and LβIVβV. These elements are characteristic for MMPs and distinguish them from other metzincins [3,4,6]. LβIIβIII and LβIVβV are at the top left and LβIIIβIV at the top right of the molecule. LβIVβV participates in an octahedral calciumbinding site made up by three main-chain carbonyl oxygen atoms, two solvent molecules, and a carboxylate oxygen from an aspartate residue, Asp173. LβIIIβIV, called the “S-loop” (see Fig. 1A; [25]), spans 16 residues and meanders around two further ion-binding sites: First, a structural zinc ion is tetrahedrally co-ordinated by three histidines and, monodentately, by an aspartate (His147, His162, His175, and Asp149). This cation is indispensable, as a double mutant disabling metal co-ordination was shown to be inactive. The second half of the S-loop is engaged in a second octahedral calcium-binding site (Fig. 1A). Three main-chain carbonyl oxygens and the side chains of Asp154, Asp177, and Glu180 contribute to the ligand sphere in a monodentate manner. All residues participating in the three ionbinding sites are conserved among structurally characterized MMPs (see Fig. 2 in [25]). In some structures, a third, trigonal-bipyramidal calcium site is found trapped by LβIαA and LβVαB under participation of two main-chain carbonyl oxygens and the carboxylate groups of two acidic groups [25]. After the S-loop, the chain enters the lowermost, antiparallel strand of the β-sheet, βIV, which delimits the active-site cleft (see below). After the last strand, βV, the chain enters a large loop, LβVαB, which is highly variable among MMPs and contributes to substrate specificity (see below). This loop leads to the active-site helix, αB, which comprises the first half of an extended zinc-binding consensus sequence, HEXXHXXGXXH (amino acid oneletter code; X stands for any residue). The helix includes two histidine ligands of the catalytic zinc ion, separated by a single helical turn, which allows a concerted approach to the metal, and the general base/ acid glutamate required for catalysis (see Section 6.). The motif is conserved not only among MMPs but also across metzincins [3,4]. The NTS ends after αB at the glycine of the consensus sequence and enters the CTS. This residue allows for a sharp turn in the trajectory of the polypeptide to reach the third histidine of the motif (in some metzincins it is replaced by an aspartate; [3]), which is also a zincbinding ligand. The values of the main-chain angles of the glycine (Φ = 108°; Ψ = 10° in MMP-8) indicate that this residue is required at this position, as any other amino acid would be in a high-energy conformation. After the third zinc ligand, a short, structurally conserved loop adopts the shape of a wide, right-handed spiral. It

features a conserved residue within MMPs, serine or threonine in all structures analyzed (Ser208 in MMP-8), which is engaged in a strong electrostatic interaction with the residue that anchors the N-terminus, Asp232 (Fig. 1A). After this loop, the chain leads to a tight 1,4-turn of type I, termed the “Met-turn” due to the invariable residue found at the third turn position (Met215; see Fig. 1A–C and Section 5.). This turn is stabilized by a double main-chain interaction with Asp233 of αC (Fig. 1A). Between the Met-turn and the last helix αC, all MMPs have a “specificity loop”, which is important for substrate specificity (see Section 5.). This segment runs along the molecular surface from left to right, before it penetrates the molecular moiety and turns to the left on the back surface to join αC. In this conserved scaffold, some MMPs show specific structural elements. For example, one insertion, found only in MT-MMPs within LβIIβIII, termed “MT-loop”, could have an exosite function during proMMP-2 activation. Moreover, in MMP-2 and MMP-9 three fibronectin type-II repeats are found within LβVαB. They are important for substrate interaction [25]. 5. Active-site cleft and zinc-binding site The active-site top is delimited by strand βIV and the last part of the preceding LβIIIβIV, the S-loop. The former outlines the upper-rim or northern-wall of the active-site crevice and binds a substrate mainly on its non-primed side through antiparallel β-ribbon like interactions (see Section 6. and Fig. 2). The latter element was termed “bulge-edge segment” by K. Maskos and W. Bode [41] and is of great importance for substrate and inhibitor binding. The cleft bottom is closed on its primed side by the specificity loop, which is responsible for a hydrophobic S1′-substrate binding site or specificity pocket (see Section 7.). It displays the largest variations of all the MMPs, which accounts for the variety in depth and nature of the pocket. The specificity loop also contributes to substrate binding through its initial segment, Pro217-Asn-Tyr219 in MMP-8, termed the “S1′-wall-forming segment” [41]. The catalytic zinc ion lies at the bottom of the cleft center (Fig. 1A,C). It is co-ordinated by the Nɛ2 atoms of the three consensus histidines (His197, His201, and His207) and a catalytic solvent molecule (substituted by other ligands in the enzyme/ inhibitor complexes) in an approximately tetrahedral manner, with binding distances of ∼ 2.1 Å. Below the zinc site, the Met-turn features an element conserved (even for the conformation of the methionine side chain) in position and orientation across all metzincins studied despite disparate upstream and downstream chain traces [3,4]. The

24

C. Tallant et al. / Biochimica et Biophysica Acta 1803 (2010) 20–28

Met-turn forms a hydrophobic pillow for the catalytic zinc ion, although the methionine–sulfur is too far away from the metal (6.3 Å in MMP-8), and its two lone electron pairs do not point towards the metal. The reason for the strict conservation of the Met-turn is not clear. Some mutation studies suggested a role in folding and stability of metzincins but others did not [42–45]. 6. Mechanism of peptide-bond cleavage Based mainly on studies on the distantly related MPs, thermolysin and carboxypeptidase A, as well as on a number of MMP complexes, a

single-step catalytic mechanism of action was suggested for MMPs [4,40,46–53]. This mechanism comprises the nucleophilic attack of a catalytic solvent molecule, polarized by the general base/acid glutamate and the catalytic zinc ion, on the scissile peptide bond at close-to-neutral pH values. For this to happen, a substrate must be bound and form a Michaelis complex (Fig. 2, step I). Binding occurs in an extended conformation through the S1′-wall-forming segment and the bulge-edge segment on the primed side and through upper-rim strand βIV on the non-primed side of the active-site cleft. The scissile carbonyl group co-ordinates the catalytic zinc ion, which further ligands the three protein histidines of the zinc-binding consensus

Fig. 3. Zymogenic structure of MMPs. (A) Stereographic Richardson-plot of pro-MMP-2 (PDB 1eak) without the fibronectin-related domains approximately in standard orientation. The pro-domain is shown in white and the catalytic moiety in yellow. The magenta arrow points to the beginning of the mature protease domain (Tyr110). This zymogen displays a unique disulfide-bond (orange sticks), not present in other pro-MMPs (see Fig. 2 in [25]). (B) Close-up view of (A) showing electrostatic interactions (grey dots) between pro-domain segment Lys99-Asp106 (stick model with light-grey carbon atoms) and residues of the protease moiety (sticks with yellow carbon atoms). The structure reported has the general base/acid glutamate replaced with glutamine. (C) Superimposition of the structures of pro-MMP-1 (PDB 1su3, [61]; pro-domain in violet and protease domain in purple), pro-MMP-2 (PDB 1eak, [58]; pink/magenta; without the fibronectin-related domains), pro-MMP-3 (PDB 1slm, [60]; cyan/blue), and pro-MMP-9 (PDB 1l6j, [62]; yellow/orange). Discontinuities in the chain trace reflect undefined regions. Structure superposition was computed with program RAPIDO [90] and figures were prepared with programs SETOR and TURBO-FRODO [91].

C. Tallant et al. / Biochimica et Biophysica Acta 1803 (2010) 20–28

25

Fig. 4. Inhibitor complexes of MMP catalytic domains. Close-up view similar to Fig. 1C of MMP catalytic domains in complexes with small-molecule inhibitors. (A) Complex of MMP-8 with BB-1909 (N3-[3S-hydroxy-4-{N-hydroxyamino}-2R-isobutylsuccinyl]amino-1-azacyclotridecan-2-one). This compound displays IC50 values of 20 nM against MMP-8; 30 nM against MMP-1; 20 nM against MMP-2; 500 nM against MMP-3, and 200 nM against MMP-7 (PDB 1 kbc;[92]). (B) Detail of the complex of MMP-13 with n,n′-bis(4-fluoro-3methylbenzyl)pyrimidine-4,6-dicarboxamide, which evinces an IC50-value of 8 nM against MMP-13 and no detectable affinity at 100 μM for MMP-1, -2, -3, -7, -8, -9, -10, -12 -14 and 16. (PDB 1xud; [74]). (C) Same region of MMP-8 as (A) with 5-carbamoylmethyl-4-oxo-3,4,5,6,7,8-hexahydro-benzo[4,5]thieno[2,3-d]pyrimidine-2-carboxylic acid (3-oxo-3,4dihydro-2H-benzo[1,4]oxacin-6-yl-methyl)amide. This compound presents an IC50-value of 7.4 nM against MMP-8,b25 nM against MMP-13, and N2500 nM against MMP-1, -2, -3, -7, -9, -12, and -14 (PDB 1dng; [75]). The left panel shows the structure after a horizontal 90°-rotation. The figure was prepared with program SETOR.

sequence and a solvent molecule in a pentameric fashion. In some MMP structures, two additional, somewhat labile solvent molecules were found bound to the zinc [51]. The pentameric zinc ligand sphere is retained throughout the reaction, although the ion may undergo local charge transitions during the process [53]. The solvent is further bound to the glutamate of the zinc-binding consensus sequence, which acts first as a general base and then as an acid during catalysis. This arrangement entails the solvent becoming extremely polarized between the glutamate base and the zinc Lewis acid [50]. In addition, the zinc further exerts Lewis acid activity on the scissile carbonyl oxygen, thus enhancing the electrophilicity of the carbon [54]. The scissile nitrogen hydrogen bonds to a main-chain carbonyl of the upper-rim strand and the hydrophobic P1′-chain is accommodated in the cognate specificity pocket. The glutamate base abstracts a proton from the water (Fig. 2, step I), thus giving rise to a hydroxide that attacks the scissile carbonyl carbon following a tetrahedral mechanism [55]. This leads the planar carbonyl group to become a negativelycharged tetrahedral reaction intermediate centered on an sp3hybridized gem-diolate group (Fig. 2, step II). The latter interacts in a bidentate manner with the zinc ion, with one of its hydroxyls occupying the position of the catalytic solvent in the substratedepleted enzyme. The glutamate subsequently acts as a general acid catalyst, delivering the proton captured from the solvent to the scissile-bond nitrogen, which becomes a secondary ammonium [56]. The remaining gem-diolate proton is transferred to the secondary ammonium directly or via the glutamate (Fig. 2, step III) [57]. This second proton-shuffling event results in the reaction intermediate collapsing to two products containing new carboxylate and αammonium groups, respectively. The two cleavage products remain initially bound to the enzyme in the form of a double-product complex [51]. Biophysical studies on a functionally-related MP, ADAM-17, revealed that product release is actually the rate-limiting step of the reaction [53]. The newly formed C-terminus interacts with the

catalytic zinc, possibly in a monodentate manner, and the new Nterminus hydrogen bonds to the glutamate. Subsequently, an incoming solvent molecule binds the zinc ion and separates the new Cterminus from the general base/acid glutamate carboxylate group. This facilitates release of the non-primed-side product half, while the primed-side half remains still bound to the enzyme. Repulsion between the zinc ion and the new N-terminus leads to a subtle rearrangement of the bound product half and to reaccommodation of the P1′ side chain within the specificity pocket. This entails that the main-chain interactions of the product with the enzyme are weakened, eventually leading to product release. 7. Zymogen structure and activation Except for MMP-23, MMPs are kept under control through prodomains and, to date, structures of pro-MMP-1, -2, -3, and -9 have been reported [58–62] (Fig. 3A–C). They show that the catalytic domains are already preformed in the zymogen and that pro-domains, which span between 66 and 91 residues among structurally characterized MMPs (see Table 1), shield the active-site clefts, thus preventing substrate access. Such shielding of an already functional protease moiety also occurs in metallocarboxypeptidases like funnelins [47]. In contrast to the latter, however, the MMP pro-domain is not required for (re)folding of the catalytic moiety [63]. In pro-MMP-2 (PDB 1EAK; [58]), taken hereafter as an example, the 80-residue domain is globular, with an ellipsoidal shape (Fig. 3A, B). The Nterminus resides on top of the third fibronectin type-II attachment of the MMP-2 moiety (see Section 4.). It runs in two segments of extended conformation preceding the globular part. A similar arrangement of the N-terminal segment is observed for pro-MMP-9 but not pro-MMP-1 or pro-MMP-3, which lack the fibronectin type-II domains found in gelatinases (see Section 4. and Fig. 3C). The globular part is maintained by an extensive hydrophobic core including side

26

C. Tallant et al. / Biochimica et Biophysica Acta 1803 (2010) 20–28

chains of a three-helix bundle that creates the scaffold to position a peptide in extended conformation to block the active-site cleft. This peptide starts in the outermost right edge of the cleft with Pro100 and it encompasses a conserved “cysteine-switch” or “velcro” motif [30,31,64], PRCGXPD (Pro100-Arg-Cys-Gly-Asn-Pro-Asp106 in proMMP-2), which runs parallel to the upper cleft rim, i.e., in the opposite direction to a bound substrate (Fig. 3B). This segment interacts with the bulge-edge, the upper-rim, and the S1′-wall-forming segments through inter-main chain and side-chain/main-chain interactions. The central cysteine, Cys102, binds the catalytic zinc ion via its Sγ atom, which replaces the catalytic solvent molecule (see Section 6.). Together with the three histidine (His403, His407, and His413) Nɛ2 atoms, it gives rise to a tetrahedral metal co-ordination. In addition, Cys102 Sγ hydrogen binds the general acid/base glutamate and the pro-domain main-chain at Asn104 N (Fig. 3B). Two residues downstream, the latter asparagine (other pro-MMPs display a valine at this position) partially occupies the S1 pocket and links both main-chain atoms of upper-rim Ala194 via its side chain. A subsequent proline residue leads to an invariant aspartate (Asp106), whose double saltbridge with Arg101 maintains the required conformation of this inhibitory sequence (Fig. 3B). At this point, the chain turns back and approaches the molecular surface of the enzyme. Disparate chain traces within the distinct pro-MMP structures lead to the beginning of the mature enzyme moiety (Fig. 3A, C) [58,60–62,65]. Activation of pro-MMPs occurs through removal of the zymogenic domain by mercurial compounds, chaotropic agents, oxidants, disulfide compounds, alkylating agents, and several proteinases such as trypsin, plasmin, and other MMPs [66]. All these reagents cause a conformational change in the molecule that pulls out the cysteine residue and enables a solvent molecule to enter the zinc co-ordination sphere and, thus, to generate a functional active site. This, in turn, enables the enzymes to undergo autoproteolysis to completely remove the pro-domain. This mechanism may be shared with other metzincin families, in which conserved cysteine-switch motifs have been found, such as the ADAMs/adamalysin (conserved motif PKMCGV, [67,68]; although this is controversial [69]), leishmanolysin (motif HRCIHD) and pappalysin families (motif CG; [70]). The resulting mature catalytic domains have different N-termini depending on the means of activation and they display differences in activity [39,66]. The physiologically most relevant variant is achieved through proteolytic processing of the Asn109–Tyr110 peptide bond in pro-MMP-2 (His99–Phe100 in pro-MMP-3, Gln99–Phe100 in proMMP-1, Gly98–Phe99 in pro-MMP-8, and Arg106–Phe107 in proMMP-9) to render forms that can establish the salt-bridge between the mature N-terminus and an aspartate of helix αC (Asp436 in MMP-2; see also Section 4.). In the case of pro-MMP-2, the position of the new N-terminus entails a ∼ 20 Å-displacement downwards with respect to the zymogen. A special case is provided by some MMPs such as MT-MMPs, in which a specific recognition sequence (RX{K/R}R) is inserted before the start of the catalytic domain. This facilitates intracellular activation by Golgi-associated Kex-2/furinlike convertases of the subtilisin serine proteinase family prior to secretion [71]. 8. Specificity pocket and inhibitor design In MMPs, the main subsites for substrate recognition are the specificity pocket, S1′, and, to a lesser extent, S2 [72]. The former is shaped by elements from helix αB (Leu193, Val194, and His197 in MMP-8), the Met-turn (Leu214 in MMP-8), and the specificity loop (Tyr219, Ala220, Arg222, Tyr227, Leu229, and Pro230 in MMP-8) (Fig. 1C). The specificity loop greatly diverges in length among MMPs and encompasses from nine (in MMP-1, -9, -11, and -23) to 13 residues (in MMP-17 and -25) (see Fig. 1 in [72]). The S1′-pocket has been considered for the design of small-molecule drugs. Most first and second generation inhibitors were peptide-derived and bound MMPs

on their primed side (see Section 9.). They contained hydroxamic acid to firmly co-ordinate the catalytic zinc and were broad-spectrum inhibitors of MMPs and other MPs. Examples included batimastat, marimastat, prinomastat, ilomostat, Ro32-3555, etc. [36,72] (see Fig. 4A). More recently, new strategies have been developed to increase selectivity among MMPs including alternative, less potent zincbinding groups such as phosphinic compounds [73]. They take advantage of differences in the specifity loop and, thus, the S1′-pocket [72,74–76]. In some cases, compounds merely targeting the specificity pocket rather than the zinc ion were successful (Fig. 4B, C). Cumulatively, these strategies have given rise to both potent and specific inhibitors of MMP-2/9, -12, -8/13, and -13 ([74–79]; see Fig. 4). An alternative but still incipient strategy consists of targeting the general base/acid through covalent modification. The proof of concept was provided in 1992 by the elucidation of the crystal structure of a funnelin [47], bovine carboxypeptidase A, in complex with 2-benzyl3,4-epoxybutanoic acid. Its oxirane group acted as a suicide inhibitor and covalently modified one of the two glutamate carboxylate oxygens [80,81]. More recently, this strategy has led to a new generation of “mechanism-based inhibitors”, which contain a thiirane group to target the glutamate. They were shown to display significant selectivity for MMP-2 over other MMPs [82,83]. 9. Prospects Since the finding more than 30 years ago that MMPs were capable of degrading essentially all structural components of the ECM and that they were massively up-regulated in a wide spectrum of pathologies, they were heralded as promising targets for the design of new drugs. Initial steps in search for an orally bioavailable small-molecule MMP inhibitor were undertaken in the late 1970s for the treatment of arthritis [84]. Later, the results of preclinical studies of MMP inhibitors were so successful that their design was rapidly optimized and they soon entered human clinical trials. Particularly promising scenarios were cancer and metastasis, arthritis, and atherosclerosis [76,85]. However, these trials failed to reach their end points as severe side effects were observed. The main reasons adduced were that while some MMPs promote pathology, others have a protective effect [86,87]. In addition, most of the inhibitors assayed were not selective, and they inhibited several MMPs simultaneously with similar affinity, even other MPs such as ADAMs/adamalysins [76]. As a result, only periostat (doxycycline hyclate) received FDA approval in 2001, and then only for the treatment of periodontitis [36]. Therefore, efforts have since been redirected a more detailed examination of the temporary and spatial roles of each MMP in the distinct pathologies [15,72]. Another approach contemplated the development of extremely selective inhibitors to reliably distinguish between MMPs, and included the consideration of exosites [25,88]. However, despite considerable chemical efforts in recent years, only a few highly selective synthetic inhibitors have been identified so far [76]. That is why detailed knowledge of the atomic structure of the particular MMPs to be targeted—ten MMPs still await structural characterization of their catalytic domains—and of their complexes with inhibitors is an urgent requirement. Acknowledgements This study was supported by grants from Spanish ministries (BIO2006-02668, BIO2008-04080-E, BIO2009-10334, and CONSOLIDER-INGENIO 2010 Project “La Factoría de Cristalización” (CSD200600015)). Additional funding was obtained from the European Union through EU FP6 Strep Project LSHG-2006-018830 “CAMP” and EU FP7 Collaborative Project 223101 “AntiPathoGN”. We thank Robin Rycroft for helpful suggestions to the manuscript and Giorgio Pochetti for Fig. 4C.

C. Tallant et al. / Biochimica et Biophysica Acta 1803 (2010) 20–28

References [1] R. Visse, H. Nagase, Matrix metalloproteinases and tissue inhibitors of metalloproteinases: structure, function, and biochemistry. Circ. Res. 92 (2003) 827–839. [2] G. Murphy, H. Nagase, Progress in matrix metalloproteinase research, Mol. Aspects Med. 29 (2008) 290–308. [3] F.X. Gomis-Rüth, Structural aspects of the metzincin clan of metalloendopeptidases. Mol. Biotech. 24 (2003) 157–202. [4] F.X. Gomis-Rüth, Catalytic domain architecture of metzincin metalloproteases, J. Biol. Chem. 284 (23) (2009) 15353–15357. [5] W. Bode, F.X. Gomis-Rüth, W. Stöcker, Astacins, serralysins, snake venom and matrix metalloproteinases exhibit identical zinc-binding environments (HEXXHXXGXXH and Met-turn) and topologies and should be grouped into a common family, the ‘metzincins’, FEBS Lett. 331 (1993) 134–140. [6] W. Stöcker, F. Grams, U. Baumann, P. Reinemer, F.X. Gomis-Rüth, D.B. McKay, W. Bode, The metzincins — topological and sequential relations between the astacins, adamalysins, serralysins, and matrixins (collagenases) define a superfamily of zinc-peptidases, Prot. Sci. 4 (1995) 823–840. [7] J. Gross, C.M. Lapière, Collagenolytic activity in amphibian tissues: a tissue culture assay. Proc. Natl. Acad. Sci. U. S. A. 48 (1962) 1014–1022. [8] J. Gross, How tadpoles lose their tails: path to discovery of the first matrix metalloproteinase. Matrix Biol. 23 (2004) 3–13. [9] C.M. Lapière, Tadpole collagenase, the single parent of such a large family. Biochimie 87 (2005) 243–247. [10] M. Egeblad, Z. Werb, New functions for the matrix metalloproteinases in cancer progression. Nat. Rev., Cancer 2 (2002) 161–174. [11] C.E. Brinckerhoff, L.M. Matrisian, Matrix metalloproteinases: a tail of a frog that became a prince, Nat. Rev., Mol. Cell Biol. 3 (2002) 207–214. [12] D.R. Edwards, M.M. Handsley, C.J. Pennington, TheADAM , metalloproteinases, Mol. Aspects Med. 29 (2008) 258–289. [13] G. Murphy, The ADAMs: signalling scissors in the tumour microenvironment, Nat. Rev., Cancer 8 (2008) 929–941. [14] C.M. Overall, C.P. Blobel, In search of partners: linking extracellular proteases to substrates. Nat. Rev., Mol. Cell Biol. 8 (2007) 245–257. [15] C.M. Overall, C. López-Otín, Strategies for MMP inhibition in cancer; innovations for the post-trial era, Nat. Rev., Cancer 2 (2002) 657–672. [16] X.S. Puente, L.M. Sánchez, C.M. Overall, C. López-Otín, Human and mouse proteases: a comparative genomic approach. Nat. Rev., Genet. 4 (2003) 544–558. [17] V. Quesada, G.R. Ordoñez, L.M. Sánchez, X.S. Puente, C. López-Otn, The Degradome database: mammalian proteases and diseases of proteolysis, Nucl. Acids Res. 37 (Database issue) (2009) D239–D243. [18] J.A. Uría, C. López-Otín, Matrilysin-2, a new matrix metalloproteinase expressed in human tumors and showing the minimal domain organization required for secretion, latency, and activity. Cancer Res. 60 (2000) 4745–4751. [19] T.L. Blundell, Metalloproteinase superfamilies and drug design. Nat. Struct. Biol. 1 (1994) 73–75. [20] J.F. Woessner Jr., H. Nagase, Matrix Metalloproteinases and TIMPs (Protein Profile Series- Ed. P. Sheterline), Oxford Univ. Press, New York, 2000. [21] C.M. Overall, Matrix metalloproteinase substrate binding domains, modules and exosites. Overview and experimental strategies, Meth. Mol. Biol. 151 (2001) 79–120. [22] H. Birkedal-Hansen, W.G. Moore, M.K. Bodden, L.J. Windsor, B. Birkedal-Hansen, A. DeCarlo, J.A. Engler, Matrix metalloproteinases: a review. Crit. Rev. Oral. Biol. Med. 4 (1993) 197–205. [23] H. Nagase, J.F. Woessner Jr., Matrix metalloproteinases. J. Biol. Chem. 274 (1999) 21491–21494. [24] F.X. Gomis-Rüth, Hemopexin domains. in: A. Messerschmidt, W. Bode, M. Cygler (Eds.), Handbook of Metalloproteins., vol. 3, John Wiley and Sons, Ltd., Chichester (UK, 2004, pp. 631–646. [25] K. Maskos, Crystal structures of MMPs in complex with physiological and pharmacological inhibitors, Biochimie 87 (2005) 249–263. [26] H. Nagase, K. Fushimi, Elucidating the function of non catalytic domains of collagenases and aggrecanases. Connect. Tissue Res. 49 (2008) 169–174. [27] H. Nagase, Substrate specificity of MMPs. in: N.J. Clendeninn, K. Appelt (Eds.), Matrix Metalloproteinase Inhibitors in Cancer Therapy. Humana Press, Totowa, NJ, 2000, p. 39. [28] G. Murphy, H. Stanton, S. Cowell, G. Butler, V. Knäuper, S. Atkinson, J. Gavrilovic, Mechanisms for pro-matrix metalloproteinase activation. APMIS 107 (1999) 38–44. [29] H. Nagase, Activation mechanisms of matrix metalloproteinases. Biol. Chem. 378 (1997) 151–160. [30] E.B. Springman, E.L. Angleton, H. Birkedal-Hansen, H.E. Van Wart, Multiple modes of activation of latent human fibroblast collagenase: evidence for the role of a Cys73 active-site zinc complex in latency and a “cysteine switch” mechanism for activation. Proc. Natl. Acad. Sci. U. S. A. 87 (1990) 364–368. [31] H.E. Van Wart, H. Birkedal-Hansen, The cysteine switch: a principle of regulation of metalloproteinase activity with potential applicability to the entire matrix metalloproteinase gene family. Proc. Natl. Acad. Sci. U. S. A. 87 (1990) 5578–5582. [32] W.G. Stetler-Stevenson, D.W. Seo, TIMP-2: an endogenous inhibitor of angiogenesis, Trends Mol. Med. 11 (2005) 97–103. [33] W.C. Duncan, P.J. Illingworth, H.M. Fraser, Expression of tissue inhibitor of metalloproteinases-1 in the primate ovary during induced luteal regression. J. Endocrinol. 151 (1996) 203–213. [34] K. Kikuchi, T. Kadono, M. Furue, K. Tamaki, Tissue inhibitor of metalloproteinase 1 (TIMP-1) may be an autocrine growth factor in scleroderma fibroblasts. J. Invest. Dermatol. 108 (1997) 281–284.

27

[35] T. Hayakawa, K. Yamashita, K. Tanzawa, E. Uchijima, K. Iwata, Growth-promoting activity of tissue inhibitor of metalloproteinases-1 (TIMP-1) for a wide range of cells. A possible new growth factor in serum, FEBS Lett. 298 (1992) 29–32. [36] B.G. Rao, Recent developments in the design of specific matrix metalloproteinase inhibitors aided by structural and computational studies. Curr. Pharm. Des. 11 (2005) 295–322. [37] I. Schechter, A. Berger, On the size of active site in proteases. I. Papain, Biochem. Biophys. Res. Commun. 27 (1967) 157–162. [38] F.X. Gomis-Rüth, W. Stöcker, R. Huber, R. Zwilling, W. Bode, Refined 1.8 Å X-ray crystal structure of astacin, a zinc-endopeptidase from the crayfish Astacus astacus L. Structure determination, refinement, molecular structure and comparison with thermolysin, J. Mol. Biol. 229 (1993) 945–968. [39] P. Reinemer, F. Grams, R. Huber, T. Kleine, S. Schnierer, M. Piper, H. Tschesche, W. Bode, Structural implications for the role of the N terminus in the ‘superactivation’ of collagenases. A crystallographic study, FEBS Lett. 338 (1994) 227–233. [40] F. Grams, M. Crimmin, L. Hinnes, P. Huxley, M. Pieper, H. Tschesche, W. Bode, Structure determination and analysis of human neutrophil collagenase complexed with a hydroxamate inhibitor. Biochemistry 34 (1995) 14012–14020. [41] K. Maskos, W. Bode, Structural basis of matrix metalloproteinases and tissue inhibitors of metalloproteinases. Mol. Biotechnol. 25 (2003) 241–266. [42] T. Hege, U. Baumann, The conserved methionine residue of the metzincins: a sitedirected mutagenesis study. J. Mol. Biol. 314 (2001) 181–186. [43] M. Pieper, M. Betz, N. Budiša, F.X. Gomis-Rüth, W. Bode, H. Tschesche, Expression, purification, characterization, and X-ray analysis of selenomethionine 215 variant of leukocyte collagenase. J. Protein Chem. 16 (1997) 637–650. [44] G.S. Butler, E.M. Tam, C.M. Overall, The canonical methionine 392 of matrix metalloproteinase 2 (gelatinase A) is not required for catalytic efficiency or structural integrity: probing the role of the methionine-turn in the metzincin metalloprotease superfamily, J. Biol. Chem. 279 (2004) 15615–15620. [45] A.E. Oberholzer, M. Bumann, T. Hege, S. Russo, U. Baumann, The metzincin's canonical methionine is responsible for the structural integrity of the zinc-binding site, Biol. Chem. 390 (9) (2009) 875–881. [46] D.S. Auld, Catalytic mechanisms for metallopeptidases. in: A.J. Barrett, N.D. Rawlings, J.F. Woessner Jr. (Eds.), Handbook of Proteolytic Enzymes., vol. 1, Elsevier Academic Press, London, 2004, pp. 268–289. [47] F.X. Gomis-Rüth, Structure and mechanism of metallocarboxypeptidases. Crit. Rev. Biochem. Mol. Biol. 43 (2008) 319–345. [48] B.W. Matthews, Structural basis of the action of thermolysin and related zinc peptidases, Acc. Chem. Res. 21 (1988) 333–340. [49] V. Pelmenschikov, P.E.M. Siegbahn, Catalytic mechanism of matrix metalloproteinases: two-layered ONIOM study, Inorg. Chem. 41 (2002) 5659–5666. [50] L.L. Johnson, A.G. Pavlovsky, A.R. Johnson, J.A. Janowicz, C.F. Man, D.F. Ortwine, C.F. Purchase, 2ndA.D. , WhiteD.J. , HupeA , rationalization of the acidic pH dependence for stromelysin-1 (Matrix metalloproteinase-3) catalysis and inhibition, J. Biol. Chem. 275 (2000) 11026–11033. [51] I. Bertini, V. Calderone, M. Fragai, C. Luchinat, M. Maletta, K.J. Yeo, Snapshots of the reaction mechanism of matrix metalloproteinases, Angew. Chem. Int. Ed. Engl. 45 (2006) 7952–7955. [52] B. Lovejoy, A.M. Hassell, M.A. Luther, D. Weigl, S.R. Jordan, Crystal structures of recombinant 19-kDa human fibroblast collagenase complexed to itself. Biochemistry 33 (1994) 8207–8217. [53] A. Solomon, B. Akabayov, A. Frenkel, M.E. Milla, I. Sagi, Key feature of the catalytic cycle of TNF-α converting enzyme involves communication between distal protein sites and the enzyme catalytic core. Proc. Natl. Acad. Sci. U. S. A. 104 (2007) 4931–4936. [54] B.L. Vallee, D.S. Auld, Active-site zinc ligands and activated H2O of zinc enzymes. Proc. Natl. Acad. Sci. U. S. A. 87 (1990) 220–224. [55] J. March, Advanced Organic Chemistry — Reactions, Mechanisms and Structure, 3rd ed. John Wiley and Sons, Inc., New York, 1985. [56] A.F. Monzingo, B.W. Matthews, Binding of N-carboxymethyl dipeptide inhibitors to thermolysin determined by X-ray crystallography: a novel class of transitionstate analogues for zinc peptidases. Biochemistry 23 (1984) 5724–5729. [57] D.G. Hangauer, A.F. Monzingo, B.W. Matthews, An interactive computer graphics study of thermolysin-catalyzed peptide cleavage and inhibition by N-carboxymethyl dipeptides. Biochemistry 23 (1984) 5730–5741. [58] E. Morgunova, A. Tuuttila, U. Bergmann, M. Isupov, Y. Lindqvist, G. Schneider, K. Tryggvason, Structure of human pro-matrix metalloproteinase-2: activation mechanism revealed. Science 284 (1999) 1667–1670. [59] E. Morgunova, A. Tuuttila, U. Bergmann, K. Tryggvason, Structural insight into the complex formation of latent matrix metalloproteinase 2 with tissue inhibitor of metalloproteinase 2. Proc. Natl. Acad. Sci. U. S. A. 99 (2002) 7414–7419. [60] J.W. Becker, M.A.I.L.L. Rokosz, M.G. Axel, J.J. Burbaum, P.M. Fitzgerald, P.M. Cameron, C.K. Esser, W.K. Hagmann, J.D. Hermes, J.P. Springer, Stromelysin-1: three-dimensional structure of the inhibited catalytic domain and of the C-truncated proenzyme, Prot. Sci. 4 (1995) 1966–1976. [61] D. Jozic, G. Bourenkov, N.H. Lim, R. Visse, H. Nagase, W. Bode, K. Maskos, X-ray structure of human proMMP-1: new insights into procollagenase activation and collagen binding, J. Biol. Chem. 280 (2005) 9578–9585. [62] P.A. Elkins, Y.S. Ho, W.W. Smith, C.A. Janson, K.J. D'Alessio, M.S. McQueney, M.D. Cummings, A.M. Romanic, Structure of the C-terminally truncated human ProMMP9, a gelatin-binding matrix metalloproteinase, Acta Cryst. D58 (2002) 1182–1192. [63] D.R. Wetmore, K.D. Hardman, Roles of the propeptide and metal ions in the folding and stability of the catalytic domain of stromelysin (matrix metalloproteinase 3), Biochemistry 35 (1996) 6549–6558.

28

C. Tallant et al. / Biochimica et Biophysica Acta 1803 (2010) 20–28

[64] B.L. Vallee, D.S. Auld, Zinc coordination, function, and structure of zinc enzymes and other proteins. Biochemistry 29 (1990) 5647–5659. [65] W. Bode, K. Maskos, Structural studies on MMPs and TIMPs. Meth. Mol. Biol. 151 (2001) 145–177. [66] J. Cha, M.V. Pedersen, D.S. Auld, Metal and pH dependence of heptapeptide catalysis by human matrilysin. Biochemistry 35 (1996) 15831–15838. [67] F. Loechel, M.T. Overgaard, C. Oxvig, R. Albrechtsen, U.M. Wewer, Regulation of human ADAM 12 protease by the prodomain. Evidence for a functional cysteine switch, J. Biol. Chem. 274 (1999) 13427–13433. [68] F. Grams, R. Huber, L.F. Kress, L. Moroder, W. Bode, Activation of snake venom metalloproteinases by a cysteine switch-like mechanism, FEBS Lett. 335 (1993) 76–80. [69] M.E. Milla, P.E. Gonzáles, J.D. Leonard, The TACE zymogen: re-examining the role of the cysteine switch, Cell Biochem. Biophys. 44 (2006) 342–348. [70] C. Tallant, R. García-Castellanos, J. Seco, U. Baumann, F.X. Gomis-Rüth, Molecular analysis of ulilysin, the structural prototype of a new family of metzincin metalloproteases. J. Biol. Chem. 281 (2006) 17920–17928. [71] P. Osenkowski, M. Toth, R. Fridman, Processing, shedding, and endocytosis of membrane type 1-matrix metalloproteinase (MT1-MMP), J. Cell. Physiol. 200 (2004) 2–10. [72] C.M. Overall, O. Kleifeld, Towards third generation matrix metalloproteinase inhibitors for cancer therapy. Br. J. Cancer 94 (2006) 941–946. [73] V. Dive, D. Georgiadis, M. Matziari, A. Makaritis, F. Beau, P. Cuniasse, A. Yiotakis, Phosphinic peptides as zinc metalloproteinase inhibitors. Cell Mol. Life Sci. 61 (2004) 2010–2019. [74] C.K. Engel, B. Pirard, S. Schimanski, R. Kirsch, J. Habermann, O. Klingler, V. Schlotte, K.U. Weithmann, K.U. Wendt, Structural basis for the highly selective inhibition of MMP-13, Chem. Biol. 12 (2005) 181–189. [75] G. Pochetti, R. Montanari, C. Gege, C. Chevrier, A.G. Taveras, F. Mazza, Extra binding region induced by non-zinc chelating inhibitors into the S1' subsite of matrix metalloproteinase 8 (MMP-8), J. Med. Chem. 52 (4) (2009) 1040–1049. [76] A. Yiotakis, V. Dive, Synthetic active site-directed inhibitors of metzincins: achievement and perspectives. Mol. Aspects Med. 29 (2008) 329–338. [77] G. Rosenblum, S.O. Meroueh, O. Kleifeld, S. Brown, S.P. Singson, R. Fridman, S. Mobashery, I. Sagi, Structural basis for potent slow binding inhibition of human matrix metalloproteinase-2 (MMP-2), J. Biol. Chem. 278 (2003) 27009–27015. [78] A.C. Dublanchet, P. Ducrot, C. Andrianjara, M. O'Gara, R. Morales, D. Compere, A. Denis, S. Blais, P. Cluzeau, K. Courte, J. Hamon, F. Moreau, M.L. Prunet, A. Tertre, Structure-based design and synthesis of novel non-zinc chelating MMP-12 inhibitors. Bioorg. Med. Chem. Lett. 15 (2005) 3787–3790.

[79] L. Devel, V. Rogakos, A. David, A. Makaritis, F. Beau, P. Cuniasse, A. Yiotakis, V. Dive, Development of selective inhibitors and substrate of matrix metalloproteinase-12. J. Biol. Chem. 281 (2006) 11152–11160. [80] M. Yun, C. Park, S. Kim, D. Nam, S.C. Kim, D.H. Kim, The x-ray crystallographic study of covalently modified carboxypeptidase A by 2-benzyl-3,4-epoxybutanoic acid, a pseudomechanism-based inactivator. J. Am. Chem. Soc. 114 (1992) 2281–2282. [81] D.H. Kim, Chemistry-based design of inhibitors for carboxypeptidase A, Curr. Top. Med. Chem. 4 (2004) 1217–1226. [82] P.D. Brown, A.H. Davidson, A.H. Drummond, A. Gearing, M. Whittaker, Hydroxamic acid matrix metalloproteinase inhibitors. in: N.J. Clendeninn, K. Appelt (Eds.), Matrix Metalloproteinase Inhibitors in Cancer Therapy. Humana Press, New Jersey, 2001, pp. 113–142. [83] M. Ikejiri, M.M. Bernardo, S.O. Meroueh, S. Brown, M. Chang, R. Fridman, S. Mobashery, Design, synthesis, and evaluation of a mechanism-based inhibitor for gelatinase A. J. Org. Chem. 70 (2005) 5709–5712. [84] J.A. Hodgson, Remodeling MMPIs. Biotechnology 30 (1995) 554–557. [85] L.M. Coussens, B. Fingleton, L.M. Matrisian, Matrix metalloproteinase inhibitors and cancer: trials and tribulations. Science 295 (2002) 2387–2392. [86] P.A. Konstantinopoulos, M.V. Karamouzis, A.G. Papatsoris, A.G. Papavassiliou, Matrix metalloproteinase inhibitors as anticancer agents. Int. J. Biochem. Cell Biol. 40 (2008) 1156–1168. [87] C. López-Otín, L.M. Matrisian, Emerging roles of proteases in tumour suppression. Nat. Rev. Cancer 7 (2007) 800–808. [88] D.T. Puerta, J.A. Lewis, S.M. Cohen, New beginnings for matrix metalloproteinase inhibitors: identification of high-affinity zinc-binding groups. J. Am. Chem. Soc. 126 (2004) 8388–8389. [89] S.V. Evans, SETOR: hardware lighted three-dimensional solid model representations of macromolecules, J. Mol. Graphics 11 (1993) 134–138. [90] R. Mosca, T.R. Schneider, RAPIDO: a web server for the alignment of protein structures in the presence of conformational changes. Nucleic Acids Res. 36 (2008) W42–W466. [91] C. Carranza, A.-G. Inisan, E. Mouthuy-Knoops, C. Cambillau, A. Roussel, Turbo-Frodo., AFMB Activity Report 1996–1999, CNRS-UPR 9039, Marseille, 1999, pp. 89–90. [92] M. Betz, P. Huxley, S.J. Davies, Y. Mushtaq, M. Pieper, H. Tschesche, W. Bode, F.X. Gomis-Rüth, 1.8-Å crystal structure of the catalytic domain of human neutrophil collagenase (matrix metalloproteinase-8) complexed with a peptidomimetic hydroxamate primed-side inhibitor with a distinct selectivity profile, Eur. J. Biochem. 247 (1997) 356–363.