AU7778500A - Substituted phosphinate based peptide derivatives - Google Patents

Description

WO 01/25264 PCT/EPOO/09173 1 SUBSTITUTED PHOSPHINATE BASED PEPTIDE DERIVATIVES 5 The present invention relates to substituted phosphinate based peptide derivatives of formula (I) which are useful in the treatment of metalloproteinase-mediated conditions such as metabolic bone diseases, including but not limited to osteoporosis and bone metastasis. 10 R5 R2 R 3 Xa kb R 8 R R9 (I) 15 The invention also relates to a method of using these compounds in a pharmaceutical composition suitable therefore in the treatment of the above diseases in mammals, especially humans. In addition, the compounds may be used in combination therapy with e.g., radiotherapy, gene therapy, or drugs for 20 hormone replacement therapy, bone anabolic agents, cytotoxic drugs, and analgesics. Novel compounds within formula I are described which have both similar and broader utility. Human bone is constantly undergoing remodelling. The fine balance between bone resorption and bone formation is regulated 25 by local and systemic factors and by physical forces acting on various cells including, in the bone environment, the osteoclast and the osteoblast as well as specialised forms of the latter such as the bone lining cell and the osteocyte. However, in WO 01/25264 PCT/EPOO/09173 2 several metabolic bone diseases including most importantly osteoporosis and osteolytic bone metastasis, the balance is disturbed resulting in a sustained pathological net bone resorption. 5 Osteoporosis is a systemic skeletal disease characterised by low bone mass and microarchitectural deterioration of bone tissue, with a subsequent increase in bone fragility and susceptibility to fracture. Post-menopausal osteoporosis is a chronic disease which affects millions of women throughout the 10 world and it has an enormous economical and social impact on society. Bone metastasis results from spreading of primary tumours to bone, where the cancer cells can interfere with the normal bone remodelling process through local regulation of osteoblast 15 and/or osteoclast activities. This may subsequently lead to acute focal excess of bone formation over bone resorption or vice versa as in the case of osteosclerotic and osteolytic metastasis, respectively. Breast cancer is the most frequent non-skin cancer in the female population of industrialised 20 countries with approx. 500,000 new cases per year. Between 25% and 75% of these patients develop osteolytic bone metastases, and 90% of those patients who die of breast cancer have bone metastases. This severe prognosis makes treatment mandatory, and in most cases preventive measures are taken already at the time 25 of identification of the primary tumour. Other primary tumours that frequently spread to cause osteolytic bone metastasis include those associated with prostate cancer and myelomatosis. Reduction of bone resorption is believed to be an appropriate way to prevent and treat several metabolic bone 30 diseases, including osteoporosis and osteolytic bone metastasis. Agents such as calcitonin and bisphosphonates are able to suppress bone resorption and have been used for prevention and WO 01/25264 PCT/EPOO/09173 3 treatment of osteoporosis and/or osteolytic bone metastasis. Furthermore, the use of steroid hormones (especially oestrogen) in hormone replacement therapy is an established prophylactic method for post-menopausal osteoporosis. However, these 5 therapeutic agents fail to achieve satisfactory effects in some cases, due to subject limitation or uncertain efficacy, and particularly for preventive medication in osteoporosis risk groups compliance is low. Furthermore, there is not yet a curative treatment for bone metastasis, and all currently used 10 measures fo.r bone metastatic patients are of palliative type. There is therefore need for a new prophylactic/therapeutic method for preventing or treating accentuated bone resorption. Removal of the mineralised osseous substance, i.e. organic matrix embedded in deposits of calcium phosphate salts, is a 15 complicated process. Though still a controversial subject, it appears that osteoclasts are the only cells capable of bone resorption. The progressing bone loss in patients with osteoporosis as well as the acute local bone loss at metastatic sites in bones of patients with osteolytic metastasis are caused 20 by increased osteoclast activity. The expected life cycle of osteoclasts involves the following major phases: 1. recruitment of early osteoclast precursors from 25 haematopoietic stem cells, 2. proliferation of osteoclast precursors, 3. differentiation of osteoclast precursors and maturation including fusion of mononuclear into multinuclear cells, 4. migration to bone, often through a layer of bone lining 30 osteoblasts to reach the resorptive bone surface, 5. attachment to and polarisation on the resorptive bone surface, WO 01/25264 PCT/EPOO/09173 4 6. removal of mineralised osseous substance through secretion of protons and proteolytic enzymes into the sub-osteoclastic resorption zone, and 7. death by apoptosis, necrosis or a more random process. 5 These phases are, however, not always temporally (or spatially) separate events, e.g. differentiation might take place during migration to the resorptive surface and fusion might take place on the bone surface. All these phases represent 10 possibilities for intervention in order to regulate the level of bone resorption. Traditionally, proteolytic enzymes have been known to play a role in the degradation of the organic matrix of bone, and in particular in the removal of its type I collagen fibres. 15 Therefore, the speculations about the biological roles of proteinases in bone have almost entirely focused on proteinases of osteoclast origin and their potential ability as to degrade organic bone matrix in the sub-osteoclastic resorption zone. However, we show here that, though proteinases of osteoclast 20 origin are of central importance, proteinases produced by neighbouring cells such as osteoblasts and bone-metastasising cancer cells also have important influences on bone metabolism. Furthermore, our recent findings have shown that proteolytic enzymes are not only acting in the resorption zone but are also 25 very important for the migration and attachment of osteoclasts to the resorptive surface (Blavier & Delaiss6, 1995, Sato et al, 1998). In addition, proteinase-dependent migration of immature osteoclasts appears to be associated with the maturation and fusion into active bone-resorbing osteoclasts, i.e. osteoclast 30 differentiation processes. Interference with an early phase of the osteoclast life cycle e.g., osteoclast migration and/or attachment in the WO 01/25264 PCT/EPOO/09173 5 recruitment process by a proteinase inhibitor might be more effective in impairing bone resorption than inhibition of a proteinase involved directly in the resorptive process. The former type of interference will also be easier to accomplish 5 since the secreted enzymes of the migrating osteoclasts or osteoclast-precursors are not protected from inhibition as they are when secreted into the tightly sealed resorption zone which is formed when the active polarised osteoclasts attach to bone. Furthermore, according to our novel findings, proteolytic 10 enzymes of osteoblasts, including bone lining cells and osteocytes, are involved in the regulation of the passage of osteoclasts to the resorptive bone surface and in the removal of proteins, in particular collagen fibres, which remain in the resorption lacunae after the osteoclastic resorption process has 15 ended. Thus, inhibition of proteinases produced by osteoblasts including bone lining cells and osteocytes may be an effective method for regulation of bone metabolism in general, and for reducing osteoclastic bone resorption in particular. The knowledge about proteolytic enzymes involved in bone 20 resorption mainly comes from studies of the effects of natural and particularly synthetic enzyme inhibitors in bone cell and tissue cultures. Furthermore, histochemical and immunocytochemical characterisation of enzymes expressed by bone cells and tissues in vitro or in vivo, as well as more recently 25 identification of enzyme-encoding mRNA in osteoclasts and other bone cells has increased the information about the particular proteolytic enzymes involved in bone resorption. Recently, the development of several strains of transgenic mice deficient for different particular proteinases has improved the possibility to 30 clarify their roles in bone both in vivo and after cell and tissue isolation and culture.

WO 01/25264 PCT/EPOO/09173 6 The proteolytic enzymes of major relevance to osteoclastic bone resorption are members of the families of cysteine proteinases and metalloproteinases, of which some of the most important appear to be cathepsin K and L (EP-A-0611756) as well 5 as several of the Zn-containing matrix metalloproteinases (MMPs), respectively. Of the currently known approx. 15 structurally related, but genetically different members of the human MMP-family, at least MMP-9, and the membrane-type 1 MMP (MT1-MMP or MMP-14 as it is also called) and possibly also MMP-1 10 and- MMP-12 are expressed by osteoclasts. Furthermore, the most important MMPs in osteoblasts appear to be MMP-2, MMP-13 and possibly MMP-1, MMP-11 and MMP-14. The different members of the MMP-family play distinct roles in degradation of extracellular molecules under both physiological and pathological conditions. 15 They are all produced as latent proenzymes and most MMPs, except MMP-11, MMP-23 and the transmembrane MMPs (including apart from MMP-14, MT2-MMP or MMP-15, MT3-MMP or MMP-16, and MT4-MMP or MMP-17), are activated extracellularly by an auto-catalytic mechanism and/or in complex cascade reactions by other 20 proteinases including MMPs. In addition to their structural relatedness, there is a large homology between the primary sequences of the various MMPs. This is in particular the case for the well-conserved catalytic domains of the various members of the MMP-family; a fact that renders the development of 25 inhibitors selective for one or a subset of MMPs difficult. Most of the synthetic proteinase inhibitors developed over the last 10 years have been based on a substrate-mimicking peptide or pseudo-peptide framework. In the case of MMPs, the majority of novel inhibitors have included a Zn-binding group, 30 such as a hydroxamate or carboxylate. These MMP inhibitors have, however, mainly been able to interact with either the unprimed (P-), or more frequently the primed (P'-) side of the catalytic WO 01/25264 PCT/EPOO/09173 7 cleft of the proteinase, but not with both sides. This limitation does not seem to reduce the opportunity to produce high affinity MMP inhibitors with Ki-values in the low or even sub-nanomolar range, but does reduce the possibilities of 5 creating a selective inhibition of. just one particular or a subset of the MMPs. Furthermore, the most widely studied and very potent hydroxamate-type MMP inhibitors such as batimastat (BB-94) and galardin (GM6001) display poor solubility, unfavourable pharmacokinetics and/or toxic side effects. 10 - Phosphinate-based peptide derivatives of the formula (I) represent an alternative way to prepare molecules, which mimic the peptidic conformation of metalloproteinase substrates and chelate its catalytic Zn. In these derivatives, a peptide bond (-CO-NH-) susceptible to enzymatic hydrolysis has been replaced 15 by a phosphinate group (-P(O) (OH) -CH 2 -) . There are several advantages associated with these structures compared to other types of metalloproteinase inhibitors such as peptido-mimicking hydroxamates or carboxylates. The transition state of a peptide bond undergoing hydrolysis by a metalloproteinase is typically 20 depicted as -C(0) (OH)-N+H 2 - or -C(OH) 2 -NH- indicating its structural resemblance to the phosphinate group. A further indication, that phosphonamidate-type as well as phosphinate type metalloproteinase inhibitors act as transition state analogues, rather than multisubstrate ground state analogues, 25 was given by Bartlett and Marlow (1983), who showed that the kcat/Km values, but not the Km values, of peptidic thermolysine substrates, were highly correlated to the Ki values of the corresponding phosphonamidate-type inhibitors. Furthermore, the use of a phosphinate bond, in contrast to hydroxamates and 30 carboxylates, allows substrate mimicking at both the P- and P' corresponding sides of the inhibitor, and thereby an improved opportunity for increasing its selectivity towards particular WO 01/25264 PCT/EPOO/09173 8 proteinases. Finally, the dipeptido-mimetic nature of the Pl-Pl' corresponding part of the molecule, i.e. according to the nomenclature of formula (I), the dipeptido-mimetic, -NH-CH(R 4

)

P (Xa) (X'-R) -C (R') (R7) -CH(R8)-CO-, does not only allow the use of 5 traditional methods for peptide synthesis in the construction of individual inhibitors but also the use of combinatorial inhibitor libraries by including the dipeptido-mimetic in a protected building block format in the same way as amino acids are used. 10 - Alternatively, the phosphinic dipeptido-mimetic may be incorporated into peptides by solid phase synthesis using a direct approach, in which two building blocks are used, one containing Xa, Xb, R 4 and R 5 , and one containing R 6 , R 7 and R 8 . This methodology provides. the possibility for the generation of 15 compounds of formula (I) directly on the solid phase. Phosphinic acid derivatives of a traditional peptide framework (EP-0276436, US 5,776,903) have been described earlier, but not as regulators of metabolic bone diseases. More sophisticated substituted phosphinate based peptide derivatives 20 having a substituted aryl in the Pl'-corresponding side chain (US 5,579,700) or at the P-corresponding side of the inhibitor rather than a peptide or peptide-like sequence (W098/03516) have been claimed for general use in the treatment of diseases mediated by MMP-3 and diseases characterised by MMP-activity, 25 respectively. The design of synthetic inhibitors has traditionally been based on the cleavage site of peptide substrates. The optimisation process aiming at increased inhibitor potency (characterised by reduction of the Ki-values) and/or increased 30 inhibitor selectivity (characterised by Ki-ratios several orders of magnitude from 1) has mainly been done by simple side chain substitutions typically in the form of amino acid permutations WO 01/25264 PCT/EPOO/09173 9 in peptide derivatives (i.e., single mutation experiments). However, the subsites of proteinase inhibitors are not independent, e.g. two mutations which individually add beneficially to the characteristics of an inhibitor very often 5 lead to less than additive effects if combined. Thus, it is very difficult to develop potent and selective proteinase inhibitors through repeated and combined single mutation syntheses. The recent progress in development of synthetic compound libraries containing a large number of different compounds 10 prepared by e.g. combinatorial chemistry has provided efficient alternatives for the development of proteinase substrates and inhibitors. We have developed permeable resins consisting of polyethylene glycol-polyacrylamide copolymers (so-called PEGA resins, see W093/16118), which when prepared in the form of 15 small spherical beads with a typical diameter of approx. 0.1 mm, can be used for solid-phase combinatorial (pseudo-)peptide synthesis of large numbers of potential proteinase substrates and inhibitors (Meldal et al, 1994; and Renil et al, 1998). These PEGA bead libraries allow the identification of the most 20 potent among several hundred thousand compounds in a single step incubation with a few ml of approx. 10-100 nM proteinase. The identification of a suitable synthetic proteinase inhibitor such as a phosphinate based peptide derivative may be followed by appropriate modification of this compound to assure 25 its use as a medicament for the treatment of metabolic bone disease. Several characteristics are necessary, particularly sufficient uptake and stability in the living organism to assure a beneficial effect, sufficient tissue or cell specific action to assure maximal effects at the target site of the organism 30 relative to effects at non-target sites including acceptable levels of side effects, and a pharmacologically acceptable dose and time-response to the treatment.

WO 01/25264 PCT/EPOO/09173 10 Administration of proteins, peptides and peptide-like substances to animals and humans requires protective routes of administration and/or protective formulation of the peptide in order to avoid degradation of the compound. Though protective 5 encapsulation for oral administration of peptides and peptide like agents is a technology currently undergoing significant improvement, stabilisation of the agent itself prior to administration is advantageous. For peptide-mimicking MMP inhibitors this has been possible by chemical modification of an 10 initially identified compound apparently without important changes in. its inhibitory capacity (Brown & Giovazzi, 1995 and P. D. Brown personal communications June 1996). Targeting of a proteinase inhibitor to a particular cell type, e.g. osteoclasts, or particular tissue, e.g. bone, can be 15 obtained by two general means. One, is if the inhibitor due to its intrinsic specificity selectively reacts with the proteinase associated with this cell type or tissue either because the proteinase at this target cell type or tissue is particularly available to the inhibitor (due to e.g. the localisation of the 20 cell or tissue, the localisation of the proteinase in the cell or tissue, or simply by a local high concentration of the proteinase) or because the proteinase when associated with this cell type or tissue is different from the corresponding proteinase as it is expressed when associated with other cell 25 types and tissues (due to e.g. immobilisation or post translational modifications). The other way to obtain a specificity is by making hybrid molecules or conjugates combining one part of the agent having proteinase-inhibitory characteristics with another part having antibody or ligand 30 specificity for the particular cell type or tissue. These hybrids can be made by recombinant expression of fusion-proteins after cloning of a hybrid cDNA. E.g., a piece of cDNA encoding the osteoclast-specific ligand calcitonin (or a receptor-binding part thereof) can be ligated to another piece of cDNA encoding a 35 peptide inhibitor for an osteoclast proteinase. Hybrids can also be conjugates of two compounds, e.g. by chemically linking an WO 01/25264 PCT/EPOO/09173 11 amino-bisphosphonate, which has high affinity for hydroxyapatite in bone, or an antibody specific for a component exposed in the osteoclast membrane, such as the calcitonin receptor with a peptide or peptide-mimicking proteinase inhibitor. 5 The present invention includes compounds of the formula (I): R2 3 Xa kb

R

8 R R9 10 (I) or a pharmaceutically acceptable salt thereof, where: 15

R

1 is 1. a hydrogen atom 2. an amino protecting group such as a group Ria-0-CO- in which Ria is 20 2.1. an optionally substituted alkyl group (preferably Ci to C 10 more preferably C1 to CE and optionally substituted with e.g. hydroxy halogen (e.g. chlorine or fluorine), trifluoromethyl, or hydroxy alkyl) such as methyl, ethyl, propyl, isopropyl, butyl, isobutyl, sec-butyl, tert 25 butyl, pentyl, isopentyl, neopentyl, tert-pentyl, 1 ethylpropyl, norbonyl, hexyl, isohexyl, 1,1 dimethylbutyl, 2,2-dimethylbutyl, 3,3-dimethylbutyl, 1 adamantyl, 2-adamantyl, 2-ethylbutyl, heptyl, octyl, nonyl, decyl or an alkyl group, preferably having from 1 WO 01/25264 PCT/EPOO/09173 12 to 10 (more preferably 1 to 6) carbon atoms in the alkyl moiety), optionally substituted in the aryl moiety by alkyl (preferably Ci to C6), hydroxy, alkoxy (preferably Ci to C 6 ) such as benzul, 2,4,6-trimethylbenzyl, 4-tert 5 butylbenzyl, 4-tert-butoxybenzyl, 4-hydroxybenzyl, 4 methoxybenzyl, 2,4-dimethoxybenzyl, 3,4-dihydroxybenzyl, 3,4-dimethoxybenzyl, 9-fluorenylmethyl, etc. 2.2. an alkenyl group (preferably Ci to Cio, more preferably C 1 to C) such as vinyl, allyl, 10 - isopropenyl, 1-propenyl, 2-methyl-l-propenyl, 1 butenyl, 2-butenyl, 3-butenyl, 2-ethyl-l-butenyl, 3 methyl-2-butenyl, 1-pentenyl, 2-pentenyl, 3 pentenyl, 4-pentenyl, 4-methyl-3-pentenyl, 1 hexenyl, 2-hexenyl, 3-hexenyl, 4-hexenyl and 5 15 hexenyl, 2-cyclopenten-1-yl, 3-cyclopenten-1-yl, 2 cyclohexen-1-yl and 3-cyclohexen-1-yl, etc. 2.3. an alkynyl group such (preferably Ci to C 10 , more preferably C1 to C 6 ) such as ethynyl, 1-propynyl, 2 propynyl, 1-butynyl, 2-butynyl, 3-butynyl, 1-pentynyl, 2 20 pentynyl, 3-pentynyl, 4-pentynyl, 1-hexynyl, 2-hexynyl, 3-hexynyl, 4-hexynyl and 5-hexynyl, etc. 2.4. an aryl group such as phenyl, naphthyl, anthryl, phenanthryl, acenaphthylenyl, fluorenyl. 2.5. an aryl group such as phenyl, naphthyl, anthryl, 25 phenanthryl, acenaphthylenyl, fluorenyl which is fluorinated or chlorinated in one or more positions. 3. an alkyl, alkenyl, alkynyl or aryl group as Ria 4. a group R l-NH-CRcR ld-CO- in which 4.1. R l is 30 4.1.1. a hydrogen atom. 4.1.2. a natural or non-natural ax-amino acid such as alanine, arginine, asparagine, aspartic acid, WO 01/25264 PCT/EPOO/09173 13 cysteine, glutamine, glutamic acid, glycine, histidine, isoleucine, leucine, norleucine, lysine, methionine, phenylalanine, proline, hydroxyproline, hydroxylysine, serine, threonine, tryptophan, 5 tyrosine, valine or a peptide consisting of the same. 4.1.3. an alkyl, alkenyl, alkynyl or aryl group as Ria. 4.2. R1C and R independently of each other are 4.2.1. a hydrogen atom. 4.2.2. a radical corresponding to natural and non 10 - natural a-amino acids such as alanine, arginine, asparagine, aspartic acid, cysteine, glutamine, glutamic acid, glycine, histidine, isoleucine, leucine, norleucine, lysine, methionine, phenylalanine, proline, hydroxyproline, hydroxylysine, 15 serine, threonine, tryptophan, tyrosine, valine, nitrophenyl alanine, 3-nitrotyrosine, homoarginine, thiazolidine, dehydroproline, homocysteine, a aminobutyric acid, a-aminoisobutyric acid, 2 aminobenzoic acid, 4-aminobenzoic acid, homoalanine, 20 norvaline, ornithine, phenylglycine, pyroglutamic acid, sarcosine, etc. la 4.2.3. an alkyl, alkenyl, alkynyl or aryl group as R 5. a group Ri*-CO- in which R * is 5.1. a hydrogen atom. 25 5.2. an alkyl, alkenyl, alkynyl or aryl group as Ria 6. a group R"-SO 2 - in which Rl is an alkyl, alkenyl, alkynyl or aryl group as Ria 2 3 48 R2, R3, R 4 and R 8 independently of each other are 30 1. a group as Ric. 2. a group R 2 a -CH 2 - in which R 2 a is WO 01/25264 PCT/EPOO/09173 14 2.1. an aryl group such as phenyl, naphthyl, anthryl, phenanthryl, acenaphthylenyl, fluorenyl 2.2. a heteroalicylic or heteroaromatic group such as pyrrolidyl, piperidyl, morpholino, furyl, thienyl, 5 pyrrolyl, oxazolyl, isoxazolyl, thiazolyl, isothiazolyl, imidazolyl, pyrazolyl, 1,2,3-oxadiazolyl, 1,2,4 oxadiazolyl, 1,3,4-oxadiazolyl, furazanyl. 1,2,3 thiadiazolyl, 1,2,4-thiadiazolyl, 1,3,4-thiadiazolyl, 1,2,3-triazolyl, 1,2,4-triazolyl, tetrazolyl, pyridyl, 10 pyridazinyl, pyrimidinyl, pyrazinyl and triazinyl, benzofuranyl, isobenzofuranyl, benzo(b)thienyl, indolyl, isoindolyl, 1H-indazolyl, benzimidazolyl, benzoxazolyl, 1,2-benzisoxazolyl, benzothiazolyl, 1,2-benzisothiazolyl, 1H-benzotriazolyl, quinolyl, isoquinolyl, cinnolinyl, 15 quinazolinyl, quinoxalinyl, phthalazinyl, naphthylizinyl, purinyl, pteridinyl, carbazolyl, a-carbolinyl, B carbolinyl, g-carbolinyl, acricinyl, phenoxazinyl, phenothiazinyl, phenazinyl, phenoxthinyl, thianthrenyl, phenanthridinyl, phenanthrolinyl, indolizinyl, etc. 20

R

5 is 1. a hydrogen atom. 2. a group as Ria 25 R 6 and R 7 independently of each other are 1. a hydrogen atom. 2. an alkyl group such as methyl, ethyl, propyl, isopropyl, butyl, isobutyl, sec-butyl, tert-butyl, pentyl, etc. 30 R 9 is 1. a group R 3a-X- in which 1.1. R 3 a is WO 01/25264 PCT/EPOO/09173 15 1.1.1. a group as R". 1.1.2. a heteroalicylic or heteroaromatic group such as pyrrolidyl, piperidyl, morpholino, furyl, thienyl, pyrrolyl, oxazolyl, isoxazolyl, thiazolyl, 5 isothiazolyl, imidazolyl, pyrazolyl, 1,2,3 oxadiazolyl, 1,2,4-oxadiazolyl, 1,3,4-oxadiazolyl, furazanyl. 1,2,3-thiadiazolyl, 1,2,4-thiadiazolyl, 1,3,4-thiadiazolyl, 1,2,3-triazolyl, 1,2,4-triazolyl, tetrazolyl, pyridyl, pyridazinyl, pyrimidinyl, 10 - pyrazinyl and triazinyl, benzofuranyl, isobenzofuranyl, benzo(b)thienyl, indolyl, isoindolyl, 1H-indazolyl, benzimidazolyl, benzoxazolyl, 1,2 benzisoxazolyl, benzothiazolyl, 1,2-benzisothiazolyl, 1H-benzotriazolyl, quinolyl, isoquinolyl, cinnolinyl, 15 quinazolinyl, quinoxalinyl, phthalazinyl, naphthylizinyl, purinyl, pteridinyl, carbazolyl, a carbolinyl, B-carbolinyl, g-carbolinyl, acricinyl, phenoxazinyl, phenothiazinyl, phenazinyl, phenoxthinyl, thianthrenyl, phenanthridinyl, 20 phenanthrolinyl, indolizinyl, etc. 1.1.3. a group as R 2 . 1.2. Xc is 0, S or NH. Xa and X is: 25 1. 0 2. S 3. NH One genus of this embodiment is the compounds of the 30 formula (II): WO 01/25264 PCT/EPOO/09173 16 R5 R2 R3 Xa b Bu R 1 R 9 "N H (II) or a pharmaceutically acceptable salt, ester or amide thereof, 1 2 3 5 9 a 5 where R1, R , R , R , R , Xa amd Xb are as defined in Formula I, i.e. R1 is 1. a hydrogen atom 10 2. a group R a-0-CO- in which Ria is 2.1. an alkyl such as methyl, ethyl, propyl, isopropyl, butyl, isobutyl, sec-butyl, tert-butyl, pentyl, isopentyl, neopentyl, tert-pentyl, 1-ethylpropyl, norbonyl, hexyl, isohexyl, 1,1-dimethylbutyl, 2,2 15 dimethylbutyl, 3,3-dimethylbutyl, 1-adamantyl, 2 adamantyl, 2-ethylbutyl, heptyl, octyl, nonyl, decyl, benzyl, 2,4,6-trimethylbenzyl, 4-tert-butylbenzyl, 4 tert-butoxybenzyl, 4-hydroxybenzyl, 4-methoxybenzyl, 2,4 dimethoxybenzyl, 3,4-dihydroxybenzyl, 3,4 20 dimethoxybenzyl, 9-fluorenylmethyl, etc. 2.2. an alkenyl group such as vinyl, allyl, isopropenyl, 1-propenyl, 2-methyl-1-propenyl, 1-butenyl, 2-butenyl, 3 butenyl, 2-ethyl-1-butenyl, 3-methyl-2-butenyl, 1 pentenyl, 2-pentenyl, 3-pentenyl, 4-pentenyl, 4-methyl-3 25 pentenyl, 1-hexenyl, 2-hexenyl, 3-hexenyl, 4-hexenyl and 5-hexenyl, 2-cyclopenten-1-yl, 3-cyclopenten-1-yl, 2 cyclohexen-1-yl and 3-cyclohexen-1-yl, etc.

WO 01/25264 PCT/EPOO/09173 17 2.3. an alkynyl group such as ethynyl, 1-propynyl, 2 propynyl, 1-butynyl, 2-butynyl, 3-butynyl, 1-pentynyl, 2 pentynyl, 3-pentynyl, 4-pentynyl, 1-hexynyl, 2-hexynyl, 3-hexynyl, 4-hexynyl and 5-hexynyl, etc. 5 2.4. an aryl group such as phenyl, naphthyl, anthryl, phenanthryl, acenaphthylenyl, fluorenyl. 2.5. an aryl group such as phenyl, naphthyl, anthryl, phenanthryl, acenaphthylenyl, fluorenyl which is fluorinated or chlorinated in one or more positions. 10 3. an alkyl, alkenyl, alkynyl or aryl group as Ria. 4. a group R l-NH-CR cRid-CO- in which 4.1. Rib is 4.1.1. a hydrogen atom. 4.1.2. a natural a-amino acid such as alanine, 15 arginine, asparagine, aspartic acid, cysteine, glutamine, glutamic acid, glycine, histidine, isoleucine, leucine, norleucine, lysine, methionine, phenylalanine, proline, hydroxyproline, hydroxylysine, serine, threonine, tryptophan, tyrosine, valine or a 20 peptide consisting of the same. 4.1.3. an alkyl, alkenyl, alkynyl or aryl group as Ria 4.2. RiC and Rid independently of each other are 4.2.1. a hydrogen atom. 4.2.2. a radical corresponding to natural and non 25 natural a-amino acids such as alanine, arginine, asparagine, aspartic acid, cysteine, glutamine, glutamic acid, glycine, histidine, isoleucine, leucine, norleucine, lysine, methionine, phenylalanine, proline, hydroxyproline, hydroxylysine, 30 serine, threonine, tryptophan, tyrosine, valine, nitrophenyl alanine, 3-nitrotyrosine, homoarginine, thiazolidine, dehydroproline, homocysteine, a- WO 01/25264 PCT/EPOO/09173 18 aminobutyric acid, ax-aminoisobutyric acid, 2 aminobenzoic acid, 4-aminobenzoic acid, homoalanine, norvaline, ornithine, phenylglycine, pyroglutamic acid, sarcosine, etc. 5 4.2.3. an alkyl, alkenyl, alkynyl or aryl group as Ri. 5. a group Ri*-CO- in which Rl* is 5.1. a hydrogen atom. la 5.2. an alkyl, alkenyl, alkynyl or aryl group as Ri 6. a group Rlf-S0 2 - in which R"f is an alkyl, alkenyl, alkynyl 10 or aryl group as Ria

R

2 and R 3 independently of each other are 1. a group as Ric. 2. a group R 2a-CH 2 - in which R2a is 15 2.1. an aryl group such as phenyl, naphthyl, anthryl, phenanthryl, acenaphthylenyl, fluorenyl 2.2. a heteroalicylic or heteroaromatic group such as pyrrolidyl, piperidyl, morpholino, furyl, thienyl, pyrrolyl, oxazolyl, isoxazolyl, thiazolyl, isothiazolyl, 20 imidazolyl, pyrazolyl, 1,2,3-oxadiazolyl, 1,2,4 oxadiazolyl, 1,3,4-oxadiazolyl, furazanyl. 1,2,3 thiadiazolyl, 1,2,4-thiadiazolyl, 1,3,4-thiadiazolyl, 1,2,3-triazolyl, 1,2,4-triazolyl, tetrazolyl, pyridyl, pyridazinyl, pyrimidinyl, pyrazinyl and triazinyl, 25 benzofuranyl, isobenzofuranyl, benzo(b)thienyl, indolyl, isoindolyl, 1H-indazolyl, benzimidazolyl, benzoxazolyl, 1,2-benzisoxazolyl, benzothiazolyl, 1,2-benzisothiazolyl, 1H-benzotriazolyl, quinolyl, isoquinolyl, cinnolinyl, quinazolinyl, quinoxalinyl, phthalazinyl, naphthylizinyl, 30 purinyl, pteridinyl, carbazolyl, a-carbolinyl, B carbolinyl, g-carbolinyl, acricinyl, phenoxazinyl, WO 01/25264 PCT/EPOO/09173 19 phenothiazinyl, phenazinyl, phenoxthinyl, thianthrenyl, phenanthridinyl, phenanthrolinyl, indolizinyl, etc. R 5 is 5 1. a hydrogen atom. 2. a group as Ria

R

9 is 1. a group R 3 a-Xc- in which 10 1.1. R 3 a is 1.1.1. a group as Ria 1.1.2. a heteroalicylic or heteroaromatic group such as pyrrolidyl, piperidyl, morpholino, furyl, thienyl, pyrrolyl, oxazolyl, isoxazolyl, thiazolyl, 15 isothiazolyl, imidazolyl, pyrazolyl, 1,2,3 oxadiazolyl, 1,2,4-oxadiazolyl, 1,3,4-oxadiazolyl, furazanyl. 1,2,3-thiadiazolyl, 1,2,4-thiadiazolyl, 1,3,4-thiadiazolyl, 1,2,3-triazolyl, 1,2,4-triazolyl, tetrazolyl, pyridyl, pyridazinyl, pyrimidinyl, 20 pyrazinyl and triazinyl, benzofuranyl, isobenzofuranyl, benzo(b)thienyl, indolyl, isoindolyl, 1H-indazolyl, benzimidazolyl, benzoxazolyl, 1,2 benzisoxazolyl, benzothiazolyl, 1,2-benzisothiazolyl, 1H-benzotriazolyl, quinolyl, isoquinolyl, cinnolinyl, 25 quinazolinyl, quinoxalinyl, phthalazinyl, naphthylizinyl, purinyl, pteridinyl, carbazolyl, a carbolinyl, B-carbolinyl, g-carbolinyl, acricinyl, phenoxazinyl, phenothiazinyl, phenazinyl, phenoxthinyl, thianthrenyl, phenanthridinyl, 30 phenanthrolinyl, indolizinyl, etc. 2 1.1.3. a group as R 1.2. Xc is 0, S or NH.

WO 01/25264 PCT/EPOO/09173 20 Xa and Xb is: 1. 0 2. S 5 3. NH Compounds of Formula II are novel per se and will find utility in regulation of bone metabolism, regulation of metalloproteinases, regulation of MMPs, regulators of other 10 proteinases, and of proteinase mediated diseases. Exemplifying the invention are the 91 compounds (named 01-01 to 01-07, 02-01 to 02-22, 03-01, 04-01 to 04-21, and 05-01 to 05 40) each in two stereoisomeric forms (named A and B) listed in 15 Table 1 of Example 5, Table 2 of Example 6, Table 3 of Example 8, and Table 4 of Example 9. This invention also relates to a pharmaceutical composition for treatment of metalloproteinase-mediated metabolic bone diseases, including but not limited to 20 osteoporosis and bone metastasis incorporating an amount of a compound of formula (I) or a pharmaceutically acceptable salt thereof effective in such treatments and a pharmaceutically acceptable carrier. The present invention also provides the use of an agent in 25 the manufacture of a medicament for the treatment of metalloproteinase-mediated metabolic bone diseases, including but not limited to osteoporosis and bone metastasis in a mammal, including a human, comprising administering to said mammal an amount of a compound of formula (I) or a pharmaceutically 30 acceptable salt thereof effective in such treatments and a pharmaceutically acceptable carrier.

WO 01/25264 PCT/EPOO/09173 21 The invention will be further described and illustrated but not limited by the details thereof in the examples which follow and the appended drawings in which: Figure 1 shows the reaction scheme for preparation of the 5 building block III used to prepare compounds of the formula (I) containing a phosphinic dipeptido-mimetic moiety by solid phase synthesis. Reactions are Step A: Michael addition of an a-amino phosphinic acid to an acrylic ester to form the phosphorus carbon bond, Step B: esterification of the phosphinic acid, Step 10 C: -protective group manipulations. RN may be any amino protective group, preferably Fmoc, Alloc, Cbz or Boc. R 4 , R 5 , R 6 R , and R 8 may be any group according to the descriptions of the corresponding groups of compounds of the formula (I) . Rc may be any carboxy protective group, preferably an ethyl or any other 15 group included in Ria of compounds of the formula (I). Figure 2 shows the general reaction scheme for preparation by solid phase synthesis of those compounds of the formula (I) in which R' and R 9 represent amino acids or peptides.. Reactions are: Step A: attachment of the 4-hydroxymethylbenzoyl linker, 20 Step B: coupling of amino acids by conventional peptide synthesis, Step C: coupling of the building block III, Step D: coupling of amino acids by conventional peptide synthesis, Step E: cleavage of the compound off the solid support. R', R 2 , R 3 , 4 5 6 8 R R , R , R 7 , and R8 may be any group according to the 25 descriptions of the corresponding groups of compounds of the formula (I). R 9 and R 10 may be any group as R 2; R" may be any group as R a; RN may be any amino protective group, preferably Fmoc, Alloc, Cbz or Boc. Figure 3 shows the general reaction scheme for the 30 preparation of compounds of formula (I) by the direct approach on solid phase. Reactions are: Step A: solid phase peptide synthesis, Step B: acryloylation of the N-terminus, Step C: WO 01/25264 PCT/EPOO/09173 22 addition of a phosphinic acid to the acrylamide, Step D: removal of RN and subsequent solid phase peptide synthesis, Step E: cleavage of the compound from the solid support. R1, R 2 , R , R 4 , R , R 6, R , and R may be any group according to the descriptions 5 of the corresponding groups of compounds of the formula (I). R 9 and R1 may be any group as R2; R" may be any group as R ; RN may be any amino protective group, preferably Fmoc, Alloc, Cbz or Boc. Figure 4 shows the reaction scheme for preparation of a 10 one-bead-two-compounds solid phase combinatorial library of phosphinate based peptidic compounds of the formula (II) containing the phosphinic Gly-Leu dipeptido-mimetic moiety for screening against matrix metalloproteinases. One-letter abbreviations for amino acids are used. Reactions are Step A: 15 introduction of a functional biantenna consisting a free amino group and an Alloc protected amino group, Step B: attachment of a photolabile linker (Holmes and Jones, 1995) and a mass spacer hexapeptide sequence which is common for all beads, Step C: library synthesis (X 1 to X 6 represent amino acids) by the split 20 and combine methodology using 10% ladder capping, Step D: coupling of the building block III, Step C repeated, Step E removal of the Alloc group and coupling of a protected general MMP-substrate. Figure 5 shows the fragments formed by ladder capping in 25 the synthesis of the library. X 1 to X 6 represent amino acids. From the differences between the masses of these fragments the sequence of an active inhibitor was determined. Figure 6 shows the frequencies of the different amino acids in the different subsites of the inhibitors found on the 30 82 dark PEGA beads after incubation with MMP-12. Figure 7 schematically shows the sequential incubations of the one-bead-two-compounds solid phase combinatorial library of WO 01/25264 PCT/EPOO/09173 23 phosphinate based peptidic compounds of the formula (II) . The trapezoids represent batch incubations with either 370 nM MMP-9, 100 nM MMP-13 or 100 nM MMP-14 for either 24 hrs (primary proteinase incubation), or with 200 nM MMP-9, 100 nM MMP-13 or 5 100 nM MMP-14 for 1 hr (secondary and tertiary incubations). The arrows indicate transfer of dark beads. The numbers in brackets at the base of each trapezoid representing a secondary and tertiary incubation, e.g. (+9+13-14), designate the name of the group of selected beads, and indicate the selectivity of the 10 phosphinate. based peptidic compounds belonging to that group. E.g., compounds on beads belonging to group (+9+13-14) seem to be capable of inhibiting both MMP-9 and MMP-13, but not MMP-14. The order of the numbers in the bracket describes the order by which the sequential incubations were made. The numbers in bold 15 below indicate the total number of fluorescent beads that were isolated for each group. E.g., group (+9+13-14) included 12 beads that remained dark after incubation with first MMP-9 and then MMP-13, but turned fluorescent when subsequently incubated with MMP-14. All 209 selected beads belonging to the 12 groups 20 were used for MALDI-TOF sequence analysis of the corresponding phosphinate based peptide derivatives. However, the 78 sequences belonging to the two groups (+14-9) and (+14-13) were possibly false positives and not used for further analysis. Among the remaining 131 sequences, 4 sequences for each of the other 10 25 groups were selected for synthesis. Figure 8 shows the effect of different proteinase inhibitors on osteoclast invasion through a type I collagen matrix in vitro. All synthetic MMP inhibitors (at 10 ptM), including the hydroxamate-type inhibitors, RP59794, BB-94 (Ki = 30 0.3 nM for MMP-9), and GM6001 (Ki = 1 nM), and the novel phosphinate-type inhbitors, 01-02B (Ki = 70,000 nM for MMP-9), 01-01A (Ki = 300 nM), 01-07B (Ki = 80 nM), and 01-07A (Ki = 1.2 WO 01/25264 PCT/EPOO/09173 24 nM) , as well as the natural TIMP-2, reduced the invasion of osteoclasts. In contrast neither the cysteine proteinase inhibitors, E64 and EST, nor the serine proteinase inhibitor aprotinin (at 10-40 pM) showed any inhibitory effect, indicating 5 the specific role of MMPs in the invasive process. Figure 9 shows the pericellular collagenolytic activity exerted by osteoclasts seeded on a type I collagen surface in the presence or absence of 10 pM of the hydroxamate-type MMP inhibitor, GM6001. 10 Figure 10 shows the significant effect of 50 pM of the hydroxamate-type MMP inhibitor, GM6001, on the migrated distance of osteoclasts seeded on a type I collagen surface and observed for 4 to 12 hours. *: p<0.05, **: p<0.01. Figure 11 shows the inhibitory effect on bone resorption 15 exerted by a bone lining cell layer. A confluent cell layer of osteoblastic lining cells shielded the bone and thereby reduced the access of osteoclasts to the bone surface. The resorption was further reduced by addition of the hydroxamate-type MMP inhibitor, GM6001. 20 Figure 12 shows the significant inhibition (p<0.001) of the TGF-P induced elongation of osteoblasts in a confluent cell layer by the hydroxamate-type MMP inhibitor, GM6001 (10 pM) . In contrast the inhibitors of serine proteinases, cysteine proteinases, and aspartic proteinases, i.e. aprotinin, E-64, 25 and pepstatin (all 10 iM) , respectively, did not af fect the osteoblast elongation when compared to the TGF-P treated control without proteinase inhibitor. All values are Mean+SD (n=9) of the relative cell free surface area of the culture dish in each of 3 experiments. 30 Figure 13 shows the significant reduction (p<0.0001) in the TGF-0 induced elongation of osteoblast in a confluent cell WO 01/25264 PCT/EPOO/09173 25 layer by the hydroxamate-type MMP inhibitor, GM6001 (10 pM), and by the novel phosphinate-type MMP inhibitor, 01-07A (10 tM) , when compared to the TGF-3 treated control without proteinase inhibitor. All values are Mean+SD (n=9) of the relative cell 5 free surface area of the culture dish. Figure 14 shows the inhibition by the MMP inhibitor GM6001 of the TGF-P induced increase in bone resorption when the bone lining cell layer on bone slices was treated simultaneously with 2.5 ng/ml TGF-P and 10 pM GM6001 before fixation and seeding of 10 osteoclasts. All values are Mean+SD (n=4) in each of 2 experiments. Figure 15 shows that decalcification was significantly stimulated by 2.5 ng/ml TGF-P in cultures of calvariae from 18 day-old mouse foetuses. All values are Mean+SD (n=5). 15 Figure 16 shows that the novel phosphinate-type MMP inhibitor, 01-07A (10 pM) , when added simultaneously with 2.5 ng/ml TGF-P to cultures of calvariae from 18 day-old mouse foetuses markedly reduce their decalcification. All values are Mean+SD (n=5). 20 Figure 17 shows the inhibitory effect of the hydroxamate type MMP inhibitor, BB-94, on the decalcification of cultured metatarsals and tibiae. In metatarsals, osteoclasts must migrate before bone resorption can take place, whereas osteoclast migration is not a prerequisite for bone resorption 25 in tibiae in these tissue cultures. BB-94 dose-dependently reduced the 45 Ca release in non-stimulated (A, B) and 10-8 M PTH-stimulated (C, D) metatarsals (A, C) and tibiae (B, D) cultured for seven days. The result (T/C) is the ratio between the active 45Ca release in BB-94 treated bones and in paired 30 vehicle treated controls. All values are Mean+SD (n=4).

WO 01/25264 PCT/EPOO/09173 26 Significance levels, *: p < 0.05, **: p < 0.01, ***: p < 0.001. Figure 18 shows the inhibitory effect of the hydroxamate type MMP inhibitor, GM6001, on the decalcification of cultured 5 metatarsals and tibiae. In metatarsals, osteoclasts must migrate before bone resorption can take place, whereas osteoclast migration is not a prerequisite for bone resorption in tibiae in these tissue cultures. GM6001 dose-dependently reduced the 45 Ca release in tibiae (white bars) and metatarsals 10 (grey bars) cultured for seven days. The result (T/C) is the ratio between the active 45Ca release in GM6001 treated bones and in paired vehicle treated controls. All values are Mean+SD (n=4). Figure 19 shows the correlation between the individual 15 IC 50 values of a series of phosphinate based peptide derivatives and their Ki values against either MMP-7, MMP-9, MMP-12, MMP-13, MMP-14, or MMP-20. Figure 20 shows the dose-dependent reduction by the novel phosphinate-inhibitor 01-07A of the 100 nM PTH-induced release 20 of 45 Ca from cultures of pre-labelled foetal mouse calvariae. The effect at 10 pM 01-07A was similar to the effect of 10 pM GM6001. All values are Mean+SD (n=6). Figure 21 shows the significantly (p<0.05) increased retention of 3 H-labelled tetracycline in tibiae and femurs 25 isolated from mice treated in vivo with the MMP inhibitor RP59794 (200 pg s.c. bid) relative to vehicle (saline s.c. bid) treated littermates. Also the established inhibitors of bone resorption, clodronate (Cl2MBP), pamidronate (APD) and E-64 were efficient. All values are Mean+SD of respectively 4, 3, 4, 1 and 30 4 experiments each involving at least 5 mice per experimental group (RP59794, tibiae and femurs; clodronate, pamidronate, and E-64, tibiae).

WO 01/25264 PCT/EPOO/09173 27 Figure 22 shows the reduction in the development of osteolytic metastases in mice treated with the MMP inhibitor BB 94 at Day 19-28 after inoculation of human breast cancer cells (at day 0). The mice were X-rayed at Day 19 and those with 5 osteolytic metastasis were randomised into two groups receiving either vehicle or BB-94 (60 mg/kg/day) by i.p. injection once daily until Day 28. There was no significantly difference in area of osteolytic lesions between the two groups before the treatment was initiated (white bars), and the area of osteolytic 10 lesions was significantly increased (# p < 0.00001) for both group of mice at Day 28 (grey bars) . However, the increase was significantly lower in BB-94 (*** p < 0.005) . All values are Mean+SEM (n = 9, vehicle and n = 12, BB-94). Figure 23 shows that the MMP inhibitor, GM6001 (100 15 mg/kg/day i.p. by single daily injections) , was well tolerated by mice according to weight curves, and that it reduced the development of osteolytic metastases when given from Day -3 to 28 or Day -3 to 7 to mice inoculated with human breast cancer cells at Day 0. The increase in body weight was not 20 significantly different between the three groups of animals (A). The area of osteolytic lesions was significantly suppressed (*: p < 0.05) in mice receiving GM6001 for the whole experimental period and also reduced in mice treated with GM6001 for just 11 days (B). All values are Mean+SEM (n=7). 25 Figure 24 shows the significantly increased survival (Wilcoxon rank t-test, p < 0.0005) of mice treated with the MMP inhibitor, GM6001 (100 mg/kg/day by single daily injection, "protease inhibitor") after cancer cell inoculation at Day 0. Three out of nine mice treated with GM6001 group survived until 30 planned sacrifice at Day 49, whereas, all vehicle treated animals (n = 9) died within Day 36.

WO 01/25264 PCT/EPOO/09173 28 Figure 25 shows the significant reduction (p<0.001) in the number of MC3T3-El osteoblasts after 6 days of culture in 3 dimensional collagen gels induced by the hydroxamate-type MMP inhibitor, GM6001 (10 pM). In contrast, the inhibitors of serine 5 proteinases, cysteine proteinases, and aspartic proteinases, i.e. aprotinin, E-64, and pepstatin (all 10 iM), respectively, did not affect the osteoblast number when compared to the vehicle-treated control. All values are Mean+SD (n=4) in each of 3 experiments. 10 Figure 26 shows the significant reduction (p<0.001) in the number of primary foetal mouse osteoblast after 6 days of culture in 3-dimensional collagen gels induced by the hydroxamate-type MMP inhibitors, GM6001 and BB-94 (10 pM). In contrast, the novel phosphinate-type MMP inhibitor, 01-07A did 15 not affect the osteoblast number when compared to the vehicle treated control. All values are Mean+SD (n=4) in each of 2 experiments. Figure 27 shows the dose-dependent reduction in the number of MC3T3 osteoblastic cells when treated with the hydroxamate 20 type MMP inhibitor, GM6001, and the persistency of this osteoblast cell line when treated similarly with the novel phosphinate-type MMP inhibitor, 01-07A. All values are Mean+/-SD (n=4) in each of 2 experiments. Figure 28 shows the dose dependent reduction of bone 25 nodule formation by the hydroxamate-type MMP inhibitor, GM6001. The maximal effect was reached at 400 nM. All values are Mean+/ SD (n=4) in each of 2 experiments. Figure 29 shows the significant reduction (p<0.05) in bone nodule formation by differentiated MC3T3-El cells when cultured 30 in the presence of 10 p.M GM6001. In contrast, the novel phosphinate-type MMP inhibitor, 01-07A did not affect the bone WO 01/25264 PCT/EPOO/09173 29 formation when compared to the untreated control. All values are Mean+SD (n=4). Compounds of the formula (I) in which R 1 and R 9 represent amino acids or peptides hereof are conveniently prepared by a 5 solid phase peptide synthesis technique. Solid phase peptide synthesis has been standardised and peptides containing the phosphinic dipeptido-mimetic moiety (i.e., according to formula (I) : -NH-CH (R 4 ) -P (Xa) (Xb-Rs) -C (R5) (R 7 ) -CH(R 8 ) -CO-) were prepared using conventional protected amino acids in combination with a 10 specially designed building block to incorporate the phosphinic acid moiety. The syntheses of the building blocks, III, were carried out in solution and in analogy to a literature report concerning similar building blocks (Yiotakis et al, 1996). As shown in Example 1, the selected Gly-Leu mimicking phosphinate 15 building block had the following structure: Fmoc Glyy{P(O) (OAd) -CH 2 }Leu-OH (i.e. III in which RN = Fmoc, R 4 = R6 = R = H, Xa = Xb = 0, R = 1-Ad, R' = Bu') . The structure of this building block was selected due to the fact that the Gly-Leu moiety is a preferred cleavage site for some MMPs in certain 20 natural and synthetic substrates. We have found that both MMP-12 and MMP-14 readily cleave the extracellular matrix protein, osteopontin (OPN), at the particular Gly-Leu bond positioned between the amino acids Gly(159) and Leu(160) in OPN of bovine origin (boOPN): .. . -Val-Ala-Tyr-Gly 15 9

-

1 oLeu-Lys-Ser-Arg-... 25 Furthermore, the probably most widely used and very sensitive quenched fluorogenic synthetic peptide-like MMP-substrate, i.e. Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2 (M1895, Bachem, Switzerland) where Mca is (7-methoxycoumarin-4-yl)acetyl and Dpa is (2',4' dinitrophenyl)-L-2,3-diaminopropionyl, is cleaved by MMPs at the 30 Gly-Leu bond. In order to study the influence of particular side chains in formula (I) on potency and selectivity towards various MMPs, WO 01/25264 PCT/EPOO/09173 30 and with the expectation of thereby developing selective MMP inhibitors, several phosphinate based peptide derivatives were produced by solid phase peptide synthesis as shown in Example 2 and by using the aforementioned Gly-Leu mimicking phosphinate 5 building block, Fmoc-Glyy(P (O) (OAd) -CH 2 )Leu-OH. Among these were the seven substrate-mimicking peptide derivatives, named 01-01 to 01-07, and the 22 PEGA bead inhibitory library derived peptide derivatives, 02-01 to 02-22. The building block was not resolved and -consequently all peptide derivatives were produced 10 in two diastereomeric forms (named the A- and B-forms), which were separated by reversed phase HPLC. The phosphinate based peptide derivatives, 01-01 to 01-07 (see Table 4 of Example 9) were designed in analogy to the sequences of both natural bone-related protein substrates of 15 MMPs and a small synthetic peptide substrate of MMPs in which the Gly-Leu (or Ala-Ile) moiety of the substrate cleavage site was replaced by the -Glyy{P(O) (OH) -CH 2 }Leu- moiety. The specific cleavage site of MMP-12 and MMP-14 that we have recently identified in boOPN: . . .- Val-Ala-Tyr-Glyis _1iLeu-Lys-Ser-Arg-. . . 20 and the equal site: . .. -Val-Val-Tyr-Gly 6

_

68 Leu-Arg-Ser-Lys-... in human OPN (huOPN), was used in the design of derivatives 01 01 to 01-03. The cleavage site of MMP-12 and -14 in bovine bone sialoprotein, boBSP: . . .Gly-Leu-Ala-Ala1-13s1le-Trp-Leu-Pro-... and the equal site: . ..Gly-Leu-Ala-Ala1 2 9 -1 30 Ile-Gln-Leu-Pro-... 25 in human BSP (huBSP) was used in the design of derivatives 01-04 to 01-06. For the latter three derivatives, however, the inserted -Glyy{P (O) (OH) -CH 2 )Leu- moiety was expected to mimic the -Ala-Ile- moiety of the substrate cleavage site. Our formerly developed quenched fluorogenic peptide MMP-substrate, 30 H-Abz -Gly- Pro -Leu-Gly-Leu- Tyr (NO2)-Ala-Arg-OH (Renil et al, 1998), where Abz is the fluorogenic group, 2-aminobenzoic acid, WO 01/25264 PCT/EPOO/09173 31 and Tyr(N0 2 ) is the fluorescence quencher, 3-nitro-tyrosine, was used for the design of derivative 01-07. Instead of using a preformed phosphinic dipeptide building block (III), the phosphinic dipeptido-mimetic moiety may be 5 formed directly during solid phase peptide synthesis, e.g. by reacting an N-terminally acryloylated resin-bound peptide,

C(R

6 ) (R 7 ) =CH (R) -CO-peptide, with a protected 1 aminoalkylphosphinic acid, RN-NH-CH (R 4 ) - PH (Xa) (Xb-H), in the presence of. a trimethylsilylating agent. The acryloylated 10 peptide can be obtained, e.g. by reacting the N-terminus of a peptide with a 2-alkylacrylic acid chloride. This alternative, direct approach for the solid phase synthesis of compounds of formula (I) is shown in Example 3 by synthesis of the phosphinate based peptide derivative, LMF-Alay(P (0) (OH) -CH 2 }Gly 15 FAPFFG. The presence of a starting peptide on the solid support is not essential. For instance an amino group may be provided on the solid support otherwise than by a peptide, e.g. by the use of a non-peptide photolabile linker. This can be used for: 1) 20 making a combinatorial library of the form R-P2-Pl(PC)Pl' (i.e. with just the P1' position occupied in the C-terminal site) or 2) cleaving off the resulting phosphinate of the form: Rl-NH-C(R2) (R3) -CO-NH -CHR4-P(Xa) (XbR5) -C(R6) (R7) -CH(R8) -C(O) (OH) (=R-P2-P1 (PC) Pl' . 25 Also, such methods of synthesis may in principle be practised in solution rather than using a solid support. The phosphinate based peptide derivatives synthesised in the PEGA bead combinatorial library were of the general structure, Xl-X2_X3-Glyy

(PO

2

-CH

2 ) Leu-X 4 -X-X'-, in which X1 to X6 30 represent amino acids. As shown in Example 4, it was constructed as a one-bead-two-compounds library, i.e. each bead contained a different phosphinate based peptide derivative and a quenched WO 01/25264 PCT/EPOO/09173 32 fluorogenic substrate for MMPs which was common for all beads. During incubation of such a library with a metalloproteinase, e.g. MMP-9, MMP-12, MMP-13 or MMP-14, the substrate is either cleaved or it is not. For each bead, the amount of substrate 5 cleaved is inversely proportional to the potency of its particular inhibitor. Because of the incorporation of the special donor-acceptor pair of intramolecular resonance energy transfer (Abz and Tyr(N0 2 )), into the part of the substrate proximal and.distal to the bead, respectively, it is possible by 10 use of fluorescence microscopy to distinguish between beads containing substrate that has been cleaved and beads containing substrate that has not been cleaved to a considerable degree. In the beads containing substrate that has been cleaved to a considerable degree during incubation with the proteinase, the 15 3-nitrotyrosine will diffuse away from the 2-aminobenzoic acid, which will remain immobilised on the bead. Consequently, the bead will appear fluorescent when analysed by fluorescence microscopy. Thus, those beads that appear dark under a fluorescence microscope after incubation of the PEGA bead 20 library with e.g., MMP-9, MMP-12, MMP-13 or MMP-14, contains a putative inhibitor and can be manually collected and sequenced. As shown in Example 4, mixtures of Fmoc- and Boc-amino acids in the ratio 9:1 were used for each coupling cycle of the synthesis of the phosphinate based peptide derivatives in the 25 library. Since only the Fmoc group was removed during the subsequent deprotection step, "a ladder" of fragments of all possible lengths with N-terminal Boc groups were present after the completion of the library synthesis. After removal of the acid labile protective groups (among these Boc), the PEGA bead 30 library was incubated with proteinase. The ladder of fragments was released from each of the collected dark beads by cleavage of the light sensitive linker. This ladder-mixture consisted of WO 01/25264 PCT/EPOO/09173 33 the full length phosphinate based peptide derivative and all possible fragments that were shorter because they did not include one or more of the N-terminal residues. The sequences of the mixture of phosphinate based peptide derivatives and 5 peptides released from each dark bead were determined in a single MALDI-TOF mass spectrum. Each sequence was deduced directly from the differences between peaks of the mass spectrum. It should be noted that a drawback of this method is that - the amino acids Leu, Ile and Hyp, as well as Lys and Gln 10 have the same molecular weight and can therefore not be distinguished by this type of sequencing. For beads having one or more of these amino acids N-terminally to the phosphinate building block, conventional Edman degradation made it possible to distinguish between these residues. 15 The sequences present on active beads found by screening a solid phase combinatorial library with an MMP can be used directly for the design of soluble phosphinate based peptide derivatives of formula (I), which may then be analysed for their putative MMP inhibitory potential. Example 5 shows the sequences 20 of 16 putative inhibitors (04-02 to 04-17) that were designed according to sequences determined by MALDI-TOF mass spectrum analysis of phosphinate based peptide derivatives positioned on beads that remained dark after incubation with MMP-12. Furthermore, sequences present on the active beads found 25 by screening a solid phase combinatorial library with an MMP can be analysed statistically to determine a consensus sequence that contains the most frequently occurring amino acids in each subsite. Frequently occurring combinations of two or more amino acids should also be noticed, since the subsites cannot be 30 considered as completely independent. As shown in Example 6, incubation of a PEGA bead inhibitor library with MMP-12 resulted in the determination of putative inhibitor sequences from 82 WO 01/25264 PCT/EPOO/09173 34 dark beads, which were then used as a basis for establishment of the consensus sequence. The resulting consensus sequence as well as the sequences of 21 selected single amino acid substitutions in one of the positions corresponding to the P4 to P2 and the 5 P2' to P4' positions were used for synthesis according to Example 2 of the derivatives 02-01 and 02-02 to 02-22, respectively. Furthermore, truncated forms of the consensus sequence 02-01 (04-01 and 04-21) and of 02-16 (04-20) and 02-22 (04-19) were. designed for analysis of the importance of the P4, 10 P3, P3', P4', and P5' positions. Finally, a "second generation" consensus sequence (04-18) was designed after kinetic analyses of 02-01 to 02-22. This sequence contained those amino acids, which for each subsite gave the most potent inhibitor in the previous single amino acid substitutions of the consensus 15 sequence. This was mainly done in order to test for additivity of the effects of substitutions. As an alternative to incubation of the PEGA bead inhibitor library with just a single MMP, sequential incubations with different proteinases can be used in order to search for 20 selective inhibitors. Example 7 shows the experimental method, and the number of sequences obtained, by sequential incubations with MMP-9, MMP-13, and MMP-14 of a library analogous to the one used for incubation with MMP-12 alone (see above) . The library was split into three portions for the first incubation with 25 either MMP-9, MMP-13 or MMP-14. The resulting three portions of dark beads were separated from the fluorescent beads, and each split into two portions. Each of these two portions were incubated with a second MMP different from the MMP used in the first incubation. The resulting dark beads of the second 30 incubation were incubated with the third MMP in a final incubation.

WO 01/25264 PCT/EPOO/09173 35 Phosphinate based peptide derivatives isolated from beads that were identified as fluorescent after either the second or later incubations with a proteinase but remained dark after the first or more incubations can be expected to be selective in 5 their putative inhibitory activities. Example 8 shows the design of 40 sequences representing 10 of the 12 groups from the sequential incubations with MMP-9, MMP-13, and MMP-14 described in Example 7. The sequences were mainly based on the results from MALDI-TOF mass spectroscopy and used for the synthesis of 10 soluble, putatively selective MMP inhibitors. The kinetic characterisation of the soluble phosphinate based peptide derivatives of the present invention was primarily done in order to determine their Ki-values for a number of MMPs. A low Ki-value, preferably in the low or even subnanomolar 15 range, shows that the inhibitor is potent against the particular MMP. The ratio between the Ki-value of an inhibitor for one particular MMP and the Ki-value of the same inhibitor for another MMP describes its selectivity. A ratio higher than 1 shows that the inhibitor is more potent against the second MMP, 20 and a ratio below 1 shows that the inhibitor is more potent against the first inhibitor. The Ki-values of the 7 substrate mimicking phosphinate based peptide derivatives (01-01 to 01-07) in both their A and B diastereomeric forms are described for MMP-1, -3, -9, -12, and -14 in Example 9. Neither the 3 25 inhibitor sequences (01-01 to 01-03) mimicking the MMP-12 cleavage site in boOPN or huOPN, nor the 3 inhibitor sequences (01-04 to 01-06) mimicking the MMP-12 cleavage site in boBSP or huBSP, had very low Ki-values for the MMPs tested. In contrast, the inhibitor 01-07 with the sequence, Ala-Gly-Pro 30 Leu-Glyy{P0 2

-CH

2 }Leu-Tyr-Ala-Arg-Gly, mimicking a sensitive quenched fluorogenic synthetic MMP-substrate, was very potent against MMP-2, MMP-9, and MMP-13 (Ki 1.2 to 2.2 nM for the A- WO 01/25264 PCT/EPOO/09173 36 form) and also moderately strong against MMP-1, MMP-7, MMP-14, and MMP-20 (10 to 75 nM), but weak towards MMP-12 (200 nM) and MMP-3 (2,200 nM). This phosphinate based peptide derivative therefore represents a potent and somewhat selective inhibitor 5 for MMP-9. MMP-9 is one of the most abundant proteinases in osteoclasts and probably plays a major role in osteoclast invasion and migration, and 01-07A therefore could be an important regulator of bone metabolism. The soluble phosphinate based peptide derivative (03-01), 10 whi-ch was synthesized by the direct approach (see Example 3) and contains -Alay{P(O) (OH)-CH 2 }Gly- at its Pl-Pl' position was mainly prepared for documentation of the alternative route of synthesis and for comparison to phosphinate based inhibitors of the -Glyy[{P(O) (OH) -CH 2 }Leu- containing type. As shown in Example 15 9, and as expected, 03-01A was not a very efficient MMP inhibitor according to kinetic analyses with MMP-9, MMP-12, MMP 13 and MMP-20 (all Ki values above 10 pM). The sixteen phosphinate based peptide derivatives (04-02 to 04-17), which were designed directly according to sequences 20 found on active beads from a solid phase combinatorial library incubated with MMP-12 (see Example 5), were synthesised in soluble form according to Example 2. As shown in Example 10 the MMP-12 inhibitory activities of thirteen of these phosphinic peptides (04-03A to 04-05A, 04-07A, and 04-09A to 25 04-17A) were confirmed, showing Ki values of 1 ptM or less. Considering that the Km value for the substrate used in the library is in the low pM range, all beads containing phosphinic peptides with Ki values of 1 pM or less will appear as dark (positive) beads. In three of the phosphinic peptides 30 (04-02A, 04-06A, and 04-08A) the MMP-12 inhibitory activity could not be confirmed by enzymatic assay.

WO 01/25264 PCT/EPOO/09173 37 Twenty-two phosphinate based peptide derivatives (02-01 to 02-22), synthesised according to the consensus sequence determined from the result obtained by incubation of the X1-X 2 _

X

3 -Glyy (PO 2

-CH

2 ) Leu-X 4

-X

5

-X

6 -designed PEGA-bead library with MMP 5 12, were characterised by kinetic analyses. As shown in Example 10, most of them were, in their respective diastereomeric A-form, potent inhibitors of MMP-12 (Ki-values in the low nanomolar range) and some of them (e.g., 02-13A and 02-16A) were very selective according to comparative analyses 10 with MMP-9. and MMP-13 (Ki-values in the micromolar range), whereas others (e.g., 02-06A and 02-07A) were almost as, or even more, potent for MMP-9 and/or MMP-13. Thus, both selective and non-selective inhibitors of MMP-12 were developed, and since MMP-12 appears to be an MMP which is 15 mainly present in cells of the macrophage/osteoclast lineage, where it is hypothesised to play a role in invasion, these inhibitors are promising compounds for regulation of bone metabolism. Of the 40 soluble phosphinate based peptide derivatives 20 representing 10 of the 12 possible combinations of three sequential incubations of the PEGA bead inhibitor library with MMP-9, MMP-13, and MMP-14 groups, just a minor fraction reacted as expected when undergoing kinetic analysis with the relevant MMPs. As shown in Example 11, several of the 25 synthesized phosphinic peptides had Ki values in the nanomolar range, but their pattern of reaction with MMP-9, MMP-13, and MMP-14, more often than not was unlike the pattern of reactivities that was observed for the corresponding bead. Some of the surprising results may have been due to the lack 30 of purification of the 40 compounds resulting in testing of the mixed distereomeric form (A/B) rather than the individual A- (and B-)form(s). Still, some interesting selective MMP- WO 01/25264 PCT/EPOO/09173 38 inhibitors have been identified already by use of this new method, and more may be obtained through continuous screening among the total of 131 to 208 sequences identified by the sequential incubations or by repeating the sequential 5 incubations under more fierce conditions using a freshly prepared combinatorial library. We have previously shown that natural as well as synthetic hydroxamate-type MMP inhibitors are able to inhibit the invasion of osteoclasts through a type I collagen matrix, 10 the-reby reducing bone resorption in vitro (Sato et al, 1998) In contrast, inhibitors of serine, cysteine and aspartic proteinases did not inhibit osteoclast invasion through collagen (Sato et al, 1998) . The analyses described in Example 12 show that the novel phosphinate based peptide derivatives 15 are functional MMP inhibitors under biological test conditions, that they can regulate osteoclast invasion through type I collagen, and that their effect on osteoclast invasion is proportional to their potency towards MMP-9 determined by kinetic analyses. Thus, the novel phosphinate-type MMP 20 inhibitors, such as 01-07A, appear to be promising compounds for regulation of osteoclast invasion. The recruitment of osteoclasts to their future site of resorption requires motility of the osteoclast and therefore also interaction with the extracellular matrix. We have 25 observed that not only when invading through a type I collagen layer, but also when migrating over a type I collagen surface, MMPs appear to be of central importance for collagen degradation and osteoclast motility. In Example 13, the collagenolysis observed in the vicinity of individual 30 osteoclasts was correlated to their distance of migration, and it was shown that inhibition of the MMP-activity under these conditions does reduce both degradation of type I collagen and WO 01/25264 PCT/EPOO/09173 39 osteoclast migration. Thus, MMP-inhibition appears to be a suitable way to regulate osteoclasts migration. When osteoclasts migrate to their future site of resorption they must usually pass through a shielding layer of 5 bone lining osteoblasts. In order to mimic the natural situation, we have developed a model which includes a confluent bone lining cell layer protecting the bone surface from degrading osteoclast. As shown in Example 14, inhibition of the MMP-activity during culture of osteoclasts seeded on this layer 10 of- bone lining cells reduced bone resorption. Thus, MMP inhibition appears to be a suitable way to regulate osteoclasts invasion through the shielding layer of bone lining cells. The efficacy of the bone lining cell barrier to osteoclast passage is regulated by certain hormones and growth factors, 15 including PTH and TGF-$, respectively. Bone lining cells retract quickly and intermittently when exposed to PTH (or PTHrP), which in part could explain the induction of bone resorption by this hormone. We have found that also TGF-P induces bone resorption by regulation of the access of 20 osteoclasts to the bone surface through the layer of bone lining cells. However, the morphological change of bone lining cells induced by TGF-3 is different from that of PTH. The cells undergo a contracting elongation by intracellular actin rearrangement and thereby expose some of the otherwise covered 25 bone surface. When studied in cultures of osteoblasts, the maximum accessible surface appears already after 1 hour of exposure to PTH, but after 24 hours of exposure to TGF-3. In contrast, recovery of the bone lining cell layer takes place within 1 day of PTH exposure, even if PTH is added 30 continuously, whereas the recovery after TGF-P exposure is much slower and will not happen as long as the growth factor is WO 01/25264 PCT/EPOO/09173 40 still present. However, we have found that the regulation by TGF-$ of the morphology of bone lining cells depends on MMP activity. As shown in Example 15, the 2-3 fold increase in 5 accessible surface area which was induced by TGF-P treatment of the bone lining cell layer was significantly reduced by MMP inhibition, but not by inhibition of serine, cysteine or aspartic proteinases. Furthermore, this reduction was as strong for the MMP-9 selective phosphinate based peptide derivative, 10 01-07A, as .it was for two potent general MMP inhibitors of the hydroxamate-type. When the bone lining cell layer was instead established on slices of a bone substratum, the treatment with TGF-P resulted, as expected, in a reduced resistance to subsequent osteoclast invasion and thereby to increased bone 15 resorption. However, simultaneous treatment with an MMP inhibitor counteracted the TGF-P stimulation and reduced bone resorption. Thus, MMP-inhibition appears to be a suitable way to regulate the growth factor controlled bone lining cell barrier to osteoclast access. 20 In contrast to analyses of MMP-inhibition in bone cell cultures, the use of bone tissue cultures allows the study of complex interaction between many cell types which are spatially and temporally organised as under in vivo conditions. When the experiments of Example 15 were extended to include cultures of 25 calvariae which were treated in the same way as the bone lining cell cultures, i.e. with TGF-P in the absence or presence of an MMP inhibitor, such as the phosphinate based peptide derivative, 01-07A, they responded similarly. Thus, bone resorption in the calvariae was increased by TGF-P, but the 30 increase was abolished by simultaneous treatment with the novel phosphinate-type MMP inhibitor, 01-07A.

WO 01/25264 PCT/EPOO/09173 41 An established tissue culture method for characterisation of the interactive processes of osteoclast recruitment, invasion and bone resorption in foetal mouse metatarsal and tibia cultures was used in order to characterise the effects of 5 phosphinate based peptide derivatives. At this stage in foetal development, mononuclear osteoclasts are confined to the periosteum surrounding the newly calcified cartilage in metatarsals. During in vivo development as well as in culture, these osteoclasts mature, invade and resorb the calcified 10 car-tilage forming a primitive bone marrow cavity in the metatarsals. In contrast, mature bone resorbing osteoclasts are already present in the primitive bone marrow cavity of tibiae of the 17-day-old mouse foetuses. Thus, the bone resorption measured in tibia cultures reflects predominantly 15 resorptive activity per se of mature osteoclasts and to a lesser extent their maturation and invasion. As shown in Example 16, the two potent, hydroxamate-type MMP inhibitors, BB-94 and GM6001, which have low or sub-nanomolar Ki-values and little selectivity towards a variety of MMPs, including MMP-1, 20 -2, -3, -7, -9, -12, -13, -14, and -20, as well as several non-MMP metalloproteinases, including some of the ADAMs, not only inhibited the recruitment and migration but also reduced the resorptive activity of osteoclasts in bone cultures. In contrast, the potent MMP-9 selective phosphinate-type 25 inhibitor, 01-07A (see Table 4 of Example 9), selectively inhibited the recruitment and migration but did not affect the resorptive activity of osteoclasts. Thus, the selectivity for MMP-9 of some of the novel phosphinate-type MMP inhibitors, such as 01-07A, appears to confer a specific effect on 30 osteoclast recruitment and migration. This finding is strongly supported by our recent investigations of metatarsals and tibiae isolated from transgenic mice specifically deficient WO 01/25264 PCT/EPOO/09173 42 for the MMP-9 gene (i.e., so-called MMP-9 knockout mice) (Chen et al, 1998) . When male and female mice heterozygous for the functional MMP-9 gene were mated, approx. 25% of the foetuses in a litter were MMP-9 knockouts. Metatarsals isolated from 5 17-day-old foetuses that were MMP-9 knockouts had an approx. 50% reduction in decalcification during culture compared to their heterozygous and wild-type littermates. When these metatarsals were cultured in the presence of an MMP inhibitor, the -decalcification was completely blocked no matter whether 10 the bones were isolated from foetuses with a functional MMP-9 gene or not. Furthermore, an impeded and delayed recruitment and invasion of osteoclasts into the central part of freshly isolated metatarsals from MMP-9 knockout mouse foetuses was observed by histological inspections. In contrast, the 15 decalcification in cultured tibiae was similar in bones from MMP-9 knockout foetuses and littermates with a functional MMP 9 gene. Thus, an MMP-9 selective inhibitor such as 01-07A which act specifically on osteoclast recruitment and invasion both in invasion assays based on osteoclast and bone lining 20 cell cultures as described in Examples 12 and 15, respectively, and in the metatarsal assay based bone tissue cultures described in Example 16, is likely to have the same effect under in vivo conditions. It is commonly believed that cathepsin K is the most 25 important proteinase used by osteoclasts to solubilise the bone matrix in the subosteoclastic resorption zone. However, it was shown recently that the levels of cathepsin K in osteoclasts of calvariae are low compared to those of long bones (Everts et al, 1999). Accordingly, calvariae of 30 cathepsin K knockout mice have no apparent phenotype, whereas their long bones show deficient resorption (Saftig et al, 1998). These observations raise the question of what WO 01/25264 PCT/EPOO/09173 43 proteinases are then used by the osteoclast to solubilise the bone matrix in calvariae. As shown by the quantitative ultrastructural studies of demineralised collagen fibres in the subosteoclastic resorption zone of PTH-stimulated calvaria 5 cultures described in Example 17, not only cysteine proteinases but also MMPs play a rate limiting role in the solubilisation of bone matrix by osteoclasts. In the same cultures, it was also interesting to notice that the areas of demineralised matrix in those resorption pits from which the 10 ost-eoclasts had disappeared were much larger when the calvariae were cultured in the presence of an MMP inhibitor than when they were cultured with a cysteine proteinase inhibitor. The ultrastructural inspections described in Example 17 showed that after the osteoclast had left its resorption pit, bone lining 15 cells occupied this pit and exhibited phagocytic activity towards the collagen remnants left by the osteoclast. In the absence of a proteinase inhibitor or in the presence of a cysteine proteinase inhibitor, the areas of collagen remnants dropped to 20% of the initial values due to the lining cell 20 proteolysis, whereas this collagen degradation was almost completely blocked by MMP-inhibition. These observations show (1) that in situations where osteoclastic collagenolysis is inhibited, bone resorption proceeds in 2-steps: the osteoclasts first demineralise the bones, and the bone lining cells then 25 degrade the demineralised collagen; and (2) that MMP inhibitors can efficiently inhibit the latter process, whereas cysteine proteinase inhibitors cannot. As shown in Example 18, the regulation of bone resorption in calvariae by MMP inhibitors depended on whether the process 30 was stimulated or not. Whereas there was no effect of either a hydroxamate-type MMP inhibitor or the phosphinate based peptide derivative, 01-07A, on the 45Ca release from non-stimulated WO 01/25264 PCT/EPOO/09173 44 calvariae, both compounds almost completely inhibited the increase in bone resorption induced by 100 nM PTH. Thus, novel phosphinate-type MMP inhibitors, such as 01-07A, may be suitable for reduction of stimulated bone resorption. 5 As indicated above, there is strong evidence supporting a role of MMPs in bone resorption, but this evidence is largely based on bone cell and tissue culture models that, by definition do not precisely mimic in vivo situations. As shown in Example 19, we have developed an experimental model allowing 10 the~ evaluation of the in vivo effect of MMP inhibitors and other compounds on bone resorption in newborn mice. The model is based on observations showing that bone resorption rates can be assessed by measuring 3H-tetracycline retention in prelabelled bones (Klein et al, 1990) . Tibiae and femurs of 15 mice treated with the hydroxamate-type MMP inhibitor, RP59794, had significantly elevated contents of 3H compared to the corresponding bones of non-treated control group of mice, and as high levels of 3 H as bisphosphonate and cysteine proteinase inhibitor treated groups. Thus, MMP-inhibition appears to be a 20 suitable way to inhibit bone resorption in vivo. Acute and local pathological bone metabolism is observed in patients with bone metastasis, and, as shown in Example 20, also in an experimental model of bone metastasis, which is based on intracardial inoculation of human breast cancer cells into 25 nude mice. In this model, the cancer cells metastasise to bone and induce osteoclast-dependent osteolysis. A significant reduction in the number and size of osteolytic metastases was observed when mice inoculated with breast cancer cells were treated with an MMP-inhibitor either after radiographical 30 demonstration of osteolytic lesions, or continuously from around the time of inoculation. According to daily inspections of the animals, the general condition of the MMP inhibitor treated mice WO 01/25264 PCT/EPOO/09173 45 appeared normal as for the vehicle treated control mice until approx. 3 weeks after cancer cell inoculation and then was clearly better than in the control due to reduced or lack of cachexia. Furthermore, the survival was significantly prolonged 5 for mice treated with MMP inhibitor. The major effect of the MMP inhibitor seemed to be a reduction of osteoclast recruitment and bone resorption activity, which restricted the expansion of the metastatic areas. Thus, MMP-inhibition appears to be a suitable way to both prevent the development of bone metastases and 10 treat manifested osteolytic metastases. The use of an MMP inhibitor for treatment of a particular disease or disorder introduces an inevitable risk of side effects, particularly due to the fact that most MMPs have a multitude of physiological functions in many tissues. 15 Furthermore, most potent synthetic MMP inhibitors are strong inhibitors of several, if not all, members of the MMP-family as well as of several other related metalloproteinases such as the ADAMs (i.e., proteinases belonging to the "A Disintegrin And Metalloproteinase" family) . There are several ways to reduce the 20 potential side effects of MMP inhibitors, including the use of local administration near the site(s) of pathological excessive MMP activity, and the use of a restrictive treatment regimen, e.g. through intermittent dosing. However, a more efficient way to reduce the side effects is to improve the selectivity of the 25 MMP inhibitor without compromising its potency towards the MMPs involved in pathogenesis. Phosphinate based peptide derivatives spanning both the P- and P'-corresponding sides are particularly suitable for optimisation of selectivity, since they can interact with both the S- and S'-pockets of the target MMP(s), 30 and become repelled by non-target proteinases due to one or more of the amino acid or amino acid-mimicking residues of the pseudopeptidic compound being unable to dock and/or fit into the WO 01/25264 PCT/EPOO/09173 46 catalytic site. For phosphinate based peptide derivatives which are developed in order to act as regulators of bone metabolism, it seems particularly important that their inhibition is selective for MMP(s) which are directly or indirectly involved 5 in catabolic bone processes, in contrast to MMP(s) directly or indirectly involved in anabolic bone processes. Thus, certain MMPs of osteoclast origin, such as MMP-9 and MMP-14, which appears to be involved in osteoclast recruitment and invasion, as well as some MMPs of tumour cell or osteoblast origin, such 10 as -MMP-2 and MMP-13, which may be involved in upregulation of bone resorption, are preferred targets for selective inhibition. As shown in Examples 21 and 22, two potent hydroxamate-type MMP inhibitors which have low or sub-nanomolar Ki-values and little selectivity towards a variety of MMPs, including MMP-1, -2, 15 3, -7, -9,. -12, -13, -14, and -20, as well as several non-MMP metalloproteinases, including some of the ADAMs, have an adverse effect on the number of osteoblasts and also reduce the bone formation by osteoblasts in vitro. These side effects were not observed for the phosphinate-type MMP-9 selective inhibitor, 20 01-07A. Thus, treatment with selective MMP inhibitors, such as 01-07A, will be less prone to cause severe side effects than MMP inhibitors which act on many different proteinases. Further detailed description of methods suitable for use in this invention appear in Buchardt et al, "Phosphinic Peptide 25 Matrix Metalloproteinase -9 Inhibitors by Solid-Phase Synthesis Using a Building Block Approach', published after the priority date hereof, Chem. Eur. J. 1999, 5, No.10, 2877-2884.

WO 01/25264 PCT/EPOO/09173 47 Example 1 Synthesis of the O-Adamantyl P-(9-fluorenylmethyloxy-carbonyl aminomethyl)-P-(2-isobutylpropionic acid-3-yl)phosphinate, 5 building block, III (RN = Fmoc, R4 = R = R7 = H, Xa = X = 0, R5 = 1-Ad, R = Bu i, re: formula (III) in Fig. 1). See also the reaction scheme in Fig. 1. 10 Step A: Synthesis of P-(benzyloxycarbonylaminomethyl) -P- (ethyl 2 isobutylpropionate-3-yl)phosphinic acid (IV, R = Cbz, R = R = R 7 = H, R' = Bu', Rc = Et): A mixture of benzyloxycarbonyl 1-aminomethyl phosphinic 15 acid (5.00 g, 21.8 mmol) (Baylis et al, 1984) and hexamethyldisilazane (23 ml, 109 mmol, 5 eq.) was stirred under Ar in a dried 250 ml round-bottomed flask at 1150C for a period of 2 hours after which the temperature was lowered to 950C over a period of 35 min. Ethyl a-isobutylacrylate (4.76 20 ml, 28.3 mmol, 1.3 eq.) was added dropwise to the opaque mixture in 35 min. After stirring for 3.5 h the mixture was clear and colourless. The temperature was lowered to 70'C and EtOH (65 ml) was carefully added. Cooling to room temperature and subsequent concentration in vacuo yielded a white solid 25 which was dissolved in ethyl acetate (50 ml) and washed with 1 M HCl (50 ml). The aqueous phase was extracted with ethyl acetate (3x25 ml) and the combined extracts were washed with water (2x50 ml), saturated brine (50 ml), dried with Na 2

SO

4 and concentrated dryness at high vacuum to give a white solid. 30 Yield: 8.03 g (96%).

WO 01/25264 PCT/EPOO/09173 48 Step B: O-Adamantyl P- (benzyloxycarbonyl-aminomethyl) -P- (ethyl 2 isobutylpropionate-3-yl)phosphinate (V, RN = Cbz, R' = R 6 = R 7 H, R' = 1-Ad, R' = Bu , R: = Et): 5 A mixture of P-(benzyloxycarbonylaminomethyl)-P-(ethyl 2 isobutylpropionate-3-yl)phosphinic acid (IV, 820 mg, 2.12 mmol) and Ag 2 0 (986 mg, 4.25 mmol, 2 eq.) was refluxed under Ar in anhydrous chloroform (3 ml) for 15 min after which a chloroform solution of AdBr (503 mg, 2.34 mmol, 1.1 eq.) was 10 added dropwise to the refluxing suspension over a period of 30 min. Reflux was continued for 1 h and the mixture was stirred overnight. The crude mixture was filtered through Celite, concentrated and purified by vacuum liquid chromatography using toluene:ethyl acetate 3:1 as eluent to obtain a highly 15 viscous syrup. Yield: 977 mg (89%). Step C: O-Adamantyl P-(9-fluorenylmethyloxycarbonyl-aminomethyl) P-(2-isobutylpropionic acid-3-yl) phosphinate (III, RN= Fmoc, 20 R 4 = R 6 = R 7 = H, R 5 = 1-Ad, R 8 = Bu): Aqueous NaOH (11.8 ml, 4M, 47.2 mmol, 4 eq.) was added dropwise to a solution of O-Adamantyl P-(benzyloxycarbonyl aminomethyl)-P-(ethyl 2-isobutylpropionate-3-yl)phosphinate (V, 5.55 g, 11.8 mmol) in EtOH (100 ml). The resulting opaque 25 solution was stirred for 24 h after which it was concentrated, the residue mixed with ethyl acetate (100 ml) and water (50 ml), cooled to 0*C and HCl (aq., 20 ml 2.4 M, then 8 ml 1 M) was slowly added to adjust pH to 2. The aqueous phase was extracted twice with ethyl acetate (50 ml), the combined 30 extracts were dried with MgSO 4 and concentrated to dryness, affording a white solid carboxylic acid. The crude carboxylic acid (200 mg, 407 pLmol) was dissolved in ethyl WO 01/25264 PCT/EPOO/09173 49 acetate+MeOH+water (11+6+0.5 ml) and hydrogenated at atmospheric pressure in the presence of Pd (5% on activated carbon, 86 mg, 41 pmol, 0.1 eq.), NaHCO 3 (171 mg, 2.0 mmol, 5 eq.) and Fmoc-OSu (206 mg, 610 mmol, 1.5 eq.) for a period of 5 80 min. Vacuum (10 mmHg) was then applied for 30 min and the mixture was left stirring overnight. Filtration through Celite and concentration gave the crude product as a sticky solid (411 mg) which was purified by vacuum liquid chromatography using chloroform:methanol 30:1 as eluent to give a solid foam. 10 Yield: 154 mg (65%). Example 2 Synthesis of compounds of formula (I) by solid phase 15 synthesis, exemplified by synthesis of the phosphinate based peptide derivative, H-Ala-Gly-Pro-Leu-Gly!{ PO 2

H-CH

2 )Leu-Tyr Ala-Arg-Gly-OH (01-07) using the building block (III). See also the general reaction scheme in Fig. 2. 20 Step A: A pre-activated (10 min) mixture of hydroxymethylbenzoic acid (657 mg, 4.32 mmol, 3 eq.), O-(benzotriazol-1-yl) N,N,N',N'-tetramethyluronium tetrafluoroborate (1.33 g, 4.15 25 mmol, 2.88 eq.) and N-ethylmorpholine (727 pl, 662 mg, 5.76 mmol, 4 eq.) in anhydrous DMF (25 ml) was added to the resin

(PEGA

8 oo-resin, 0.48 ptmol/mg, 3.01 g, 1.44 mmol) swelled in DMF. After 2 h the resin was washed with DMF, dichloromethane and lyophilised overnight. 30 WO 01/25264 PCT/EPOO/09173 50 Step B: The resin was swelled in anhydrous dichloromethane (40 ml) and Fmoc-G-OH was coupled using 2,4,6-mesitylenesulfonyl 3-nitro-1,2,4-triazolide (2 couplings, 65 min and 75 min) 5 (Blankemeyer-Menge et al, 1990): Fmoc-G-OH (1.28 g, 4.32 mmol, 3 eq.) was dissolved in dry DCM (20 ml) together with N-methyl imidazole (258 p.1, 266 mg, 3.24 mmol, 2.25 eq.) and when dissolved 2,4,6-mesitylenesulfonyl-3-nitro-1,2,4-triazolide (1.28 g, 4.32 mmol, 3 eq.) was added. The mixture was 10 immediately added to the resin. After the second reaction the resin was washed with DCM, lyophilised and 75 mg (36 pmol) of the resin was weighed out for further synthesis. Couplings of the amino acids Arg(Pmc), Ala and Tyr(Bu t ) was carried out using aN-Fmoc protected amino acid pentafluorophenyl esters (3 15 eq.) in anhydrous DMF and 3,4-dihydro-3-hydroxy-4-oxo-1,2,3 benzotriazine (1 eq.) as catalyst. After coupling of an amino acid the resin was washed with DMF, followed by removal of the Fmoc group by treatment with piperidine (20% in DMF) for 2 and 10 min. After washing with DMF the cycle was repeated with the 20 next amino acid. Step C: Coupling of the building block III (RN= Fmoc, R 4 = R 6 = R 7 = H, R'= 1-Ad, R'= Bu) (1.5 eq., 31.3 mg) was accomplished by 25 activation with O-(benzotriazol-1-yl)-N,N,N',N'-tetramethyl uronium tetrafluoroborate (1.44 eq.) as above (Step A), and was completed in 4.5 h. Step D: 30 After attachment of the building block, Fmoc was removed and Leu, Pro, Gly and Ala was attached by peptide synthesis as described above for Arg(Pmc), Ala and Tyr(Bu t ) (Step B). The WO 01/25264 PCT/EPOO/09173 51 resin was then washed with DMF and DCM, dried by lyophilisation for 1.5 h and treated with a cocktail composed of trifluoroacetic acid:dichloromethane:H 2 0: MeSPh:

(CH

2

SH)

2 :triisopropylsilane 66.5:20:5:5:2.5:1 for 2.5 h to 5 remove peptide side-chain protective groups. The resin was washed with AcOH, DMF, diisopropylethyl amine (5% in DMF), DMF, dichloromethane and lyophilized (1 h). Step E: 10 The deprotected peptide was cleaved from the support by treatment with NaOH (0.1 M, 2x1.5 ml each, 1 h) and the resin was washed with H 2 0 (0.5 ml) and MeOH (0.5 ml) . The resulting peptide solution (-5 mM) was neutralised with HCl (0.1 M, 2.5 ml), lyophilised and purified by reversed phase HPLC. Yield: 15 12 mg (35%), split between two diastereomeric compounds. Example 3 Synthesis of compounds of formula (I) by solid phase synthesis 20 using a direct approach, exemplified by the synthesis of the phosphinate based peptide derivative, H-Leu-Met-Phe-Alay{PO2H

CH

2 }Gly-Phe-Ala-Pro-Phe-Phe-Gly-OH (03-01) See also the general reaction scheme in Fig. 3. 25 Step A: POEPS-3 resin (1.5 g, 0.3 mmol) (Buchardt and Meldal, 1998) was swelled in anhydrous DCM (50 ml) and drained after a period of 10 min. A solution of Fmoc-Gly-OH (267 mg, 0.9 mmol) 30 and N-methylimidazole (54 p.l, 0.675 mmol) in anhydrous DCM (30 ml) was mixed with 2,4,6-mesitylenesulfonyl-3-nitro-1, 2 ,4 triazolide (267 mg, 0.9 mmol) and added to the resin. After 60 WO 01/25264 PCT/EPOO/09173 52 min the resin was drained and washed with anhydrous DCM x 3 and another coupling of Fmoc-Gly-OH/2,4,6-mesitylenesulfonyl 3-nitro-1,2,4-triazolide/N-methylimidazole was performed for 60 min. The resin was drained and washed with DCM, DMF x 3, 5 (5% diisopropylethylamine in DMF) x 2, DMF x 3, DCM and lyophilized. Loading: 0.20 mmol/g. A fraction of the resin (840 mg, 167 imol) was treated with piperidine to remove Fmoc, washed with DMF, and treated with a solution of Fmoc-Phe-OPfp (278 mg, 501 imol) and 3,4-dihydro-3-hydroxy-4-oxo-1,2,3 10 benzotriazine (27 mg, 167 pmol) in DMF for 1 h. Similarly, Fmoc protected amino acid Pfp esters of Phe, Pro, Ala, and Phe were coupled using 3,4-dihydro-3-hydroxy-4-oxo-1,2,3 benzotriazine until a negative Kaiser ninhydrin test was observed (amounts, Fmoc-Pro-OPfp: 253 mg, Fmoc-Ala-OPfp: 240 15 mg). For the particular example, the product of this step was: Fmoc-FAPFFG-O-[POEPS-3 resin]. Step B A fraction of the resin (200 mg, 40 pmol) was treated 20 with piperidine and washed with DMF and DCM. Neat Et 3 N (111 ptl, 800 pmol) was added to the resin followed by a solution of the glycine analogue, acryloyl chloride (33 pl, 400 pmol) in DCM. The reaction was complete within 5 min according to Kaiser ninhydrin test. The resin was drained, washed with DCM and 25 MeOH x 3, and lyophilised. For the particular example, the product of this step was: CH 2 =CH-CO-FAPFFG-O-[POEPS-3 resin]. Step C A fraction of the resin (50 mg, 10 pmol) was placed in a 30 5 ml reaction vial and a solution of the alanine analogue, 1 (allyloxycarbonylamino)ethylphosphinic acid (23.2 mg, 120 i WO 01/25264 PCT/EPOO/09173 53 mol) and bis(trimethylsilyl)-acetamide (90 ptl, 360 pmol) in degassed 1,2-dichloroethane (1 ml) was added to the resin. The resin slurry was purged with Ar for 1 min, the vial was closed and heated to 100 0 C for 4 h after which the resin was cooled 5 to room temperature and washed with DCM and lyophilised. For the particular example, the product of this step was: Alloc

NH-CH(CH

3 )-P(0) (OH) -CH 2

-CH

2 -CO-FAPFFG-0-[POEPS-3 resin], or in short: Alloc-AlayI{PO 2

H-CH

2 }Gly-FAPFFG-O-[POEPS-3 resin]. 10 Step D A solution of Pd(PPh 3

)

4 (32 mg, 30 pmol) in CHCl 3 :AcOH:N ethylmorpholine 92.5:5:2.5 (800 il) was added to the resin and left for 15 min. The resin was washed with CHCl 3 , DMF x 3, (0.5% Et 2

NCS

2 Na in DMF) x 3 and DMF. A solution of Fmoc-Phe 15 OPfp (33 mg, 60 pimol) and Dhbt-OH (0.6 mg, 10 pmol) in DMF was added to the resin and allowed to react for a period of 16 h. Similarly, Fmoc protected amino acid Pfp esters of Met and Leu were coupled using 3,4-dihydro-3-hydroxy-4-oxo-1,2,3 benzotriazine (amounts Met: 16 mg, 30 pimol and Leu: 16 mg, 30 20 ptmol). After the final coupling the resin was washed with DMF, treated with piperidine to remove Fmoc and washed with DMF and MeOH. For the particular example, the product of this step was: H-LMF-Alay{PO 2

H-CH

2 }Gly-FAPFFG-O-[POEPS-3 resin]. 25 Step E The product was cleaved from the resin by treatment with NaOH (400 il, 0.1 M), the resin washed with water and MeOH. Aqueous HCl (400 pl, 0.1 M) was added to the combined filtrate 30 and the solvents were removed in vacuo. The solid residue (10 mg) was redissolved in 50 % aqeous MeCN (2 ml) and purified by WO 01/25264 PCT/EPOO/09173 54 preparative reversed phase HPLC. Fractions containing the product were collected and concentrated to dryness in vacuo to obtain, for the particular example, the phosphinic undecapeptide H-Leu-Met-Phe-Alay{ PO 2

H-CH

2 }Gly-Phe-Ala-Pro-Phe 5 Phe-Gly-OH as a solid mixture of two diastereomers (5 mg, 45 % based on the initial resin loading or 90 % based on loading of Fmoc-FAPFFG-O-[POEPS-3 resin]). Example 4 10 Preparation and screening of a one-bead-two-compounds solid phase combinatorial library of phosphinic peptides. Synthesis of the library (see also the reaction scheme in Fig. 15 4). Step A: A solution of Fmoc-Lys(Alloc)-OH (2 eq., 119.5 mg), 0 (benzotriazol-1-yl)-N,N,N',N'-tetramethyluronium tetrafluoro 20 borate (1.92 eq, 81.4 mg) and N-ethylmorpholine (2.67 eq., 44.5 il) in anhydrous DMF was added to PEGA 1900 -resin (0.11 mmol/g, 1.20 g, 0.132 mmol) and allowed to react for a period of 3.5 h. The resin was washed with DMF, Fmoc was removed (see Example 2), and the resin washed with DMF. A solution of Fmoc 25 Lys(Boc)-OPfp (3 eq., 251.5 mg) in anhydrous DMF was added to the resin and after a period of 1 h reaction was complete. The resin was washed with DMF, Fmoc was removed, and the resin was washed with DMF. The resin was treated with a solution of 3,4 dihydro-3-acetoxy-4-oxo-1,2,3-benzotriazine (1.5 eq., 40.6 mg) 30 in anhydrous DMF for a period of 18 h. It was washed with DMF, dichloromethane, and treated with trifluoroacetic acid: water WO 01/25264 PCT/EPOO/09173 55 19:1 for a period of 1 h after which it was washed with AcOH, dichloromethane, DMF, 5% diisopropylethylamine in DMF and DMF. Step B: 5 The photolabile linker (3 eq., 206 mg) (Holmes and Jones, 1995) was coupled using 0- (benzotriazol-1-yl) -N, N, N', N' tetramethyluronium tetra-fluoroborate (2.88 eq, 122 mg) and N ethylmorpholine (4 eq., 67 il) in anhydrous DMF and coupling was complete in 17 h. All resin manipulations beyond this 10 point were.carried out avoiding direct ceiling- and sunlight, and reaction vessels were wrapped in aluminium foil. Fmoc was removed and after wash with DMF the amino acids Ile, Thr(Bu t ), Arg(Pmc), Ser(Bu t ), and Ile were attached to resin using Fmoc amino acid pentafluorophenyl esters (3 eq., see Example 2) and 15 afterwards Thr (a mixture of 2.7 eq. Fmoc-Thr(Bu t )-OH and 0.3 eq. Boc-Thr(Bu t ) -OH) was coupled using 0- (benzotriazol-1-yl) N,N,N',N'-tetramethyl-uronium tetrafluoroborate (2.88 eq, 122 mg) and N-ethylmorpholine (4 eq., 67 pl). The resin was washed with DMF and dichloromethane and lyophilised. After reswelling 20 the resin in DMF it was placed in a 20 well multiple column peptide synthesiser together with excess DMF and distributed equally between the wells. Steps C and D: 25 The combinatorial synthesis was carried out by coupling 9:1 mixtures of 20 different Fmoc- and Boc-amino acids (Ala, Arg(Pmc), Asn(Trt), Asp(But), Gln(Trt), Glu(Bu t ), Gly, His(Trt), Ile, Leu, Lys(Boc), Met, Phe, Pro, Ser(Bu t ), Thr(Bu t ), Tyr(Bu t ), Val, Hyp and Hyl(Boc)) by the cyclic 30 protocol below. 1. Coupling of amino acids. Each amino acid in a specific well. Stock solutions of Fmoc- + Boc-amino acids were made and WO 01/25264 PCT/EPOO/09173 56 before each coupling 250 pl DMF was added to each well to avoid leaking from the wells. Mixtures of amino acids, 0 (benzotriazol-1-yl)-N,N,N',N'-tetramethyluronium tetrafluoro borate and N-ethylmorpholine was then added to the resin. 5 2. Wash with DMF. 3. Mixing the resin. Excess DMF was added to the synthesiser, which was capped with the lid and shaken thoroughly for 1 h. 4. Removal of Fmoc (see Example 2) 5. Wash with DMF. 10 6. Repeat from 1. After three couplings the building block III (R N= Fmoc,

R

4 = R 6 = R 7 = H, Rs= 1-Ad, R'= Bul) was coupled (as described in Example 2) and three more combinatorial couplings were performed by the above protocol. After the last coupling the 15 resin was transferred to a flat-bottomed syringe with Teflon fritte and washed with DMF, split into three equal portions which were washed with dichloromethane and lyophilised. Step E: 20 One third of the resin was treated with Pd(PPh 3

)

4 (3 eq., 139 mg) in a degassed mixture of chloroform:AcOH:N ethylmorpholine 92.5:5:2.5 for a period of 4.5 h and washed with chloroform, DMF, 0.5% Et 2

NCS

2 Na in DMF and DMF. A solution of the protected peptide MMP-substrate, Boc-Ala-Tyr(N0 2 )-Gly 25 Pro-Leu-Gly-Leu-Tyr(Bu t ) -Ala-Arg(Pmc) -Lys (Abz (Boc) ) -Gly-OH (3 eq., 264 mg), 0-(benzotriazol-1-yl)-N,N,N',N'-tetramethyl uronium tetrafluoroborate (2.88 eq., 40.5 pl) and N-ethyl morpholine (4 eq., 22.5 pl) in anhydrous DMF was added to the resin and allowed to react for a period of 14 h, where after 30 the resin was washed with DMF, Fmoc was removed, and the resin washed with DMF and dichloromethane. The resin was dried by lyophilisation for 1.5 h and treated with a cocktail composed WO 01/25264 PCT/EPOO/09173 57 of trifluoroacetic acid:dichloromethane:H 2 0:MeSPh: (CH 2

SH)

2 :tri isopropylsilane in 66.5:20:5:5:2.5:1 for 3 h to remove peptide side-chain protective groups. The resin was washed with AcOH, dichloromethane, methanol, water, 0.1 M NaOH (aq.), treated 5 with 0.1 M NaOH (aq.) for a period of 40 min, and washed with water, methanol, dichloromethane, and lyophilised. Screening of the library. 10 After. lyophilisation the dried resin (approx. 400 mg corresponding to 165,000 beads/different compounds) was added purified recombinant rabbit catalytic domain MMP-12 (8 ml, 100 nM in buffer 50 mM tris pH=7.5, 150 mM NaCl, 10 mM CaCl 2 , 50 iM ZnSO 4 , 0.05% (v/v) Brij-35). Incubation was carried out for a 15 period of 22.5 h at room temperature after which the resin was washed with water, 10 mM TFA, water, saturated NaHCO 3 , water. An amount of water just enough to swell it was added to the resin and approx. 1/50 of resin was placed on a glass plate (8 x 8 cm) . The other side of the glass plate had been furnished 20 with a line made with a yellow, fluorescent highlighting pen and the line was covered with a piece of transparent tape. The resin was spread out in a row beside the yellow line and while looking in a fluorescence microscope the beads were sorted in such a way that only those appearing to be approx. as dark as 25 the black support underneath were pushed (with a steel needle) to the side to lie on top of the yellow, fluorescent line. After sorting the batch of resin all dark beads were transferred to another glassplate and sorting was continued with a new batch. Approximately 1,000 dark beads were 30 collected manually by inspection under a fluorescence microscope and from these 112 of the most persistently dark beads were selected and washed with MeCN:water 7:3. Each bead WO 01/25264 PCT/EPOO/09173 58 was placed on a MALDI-TOF target and irradiated for 1 h with Hg lamp. Water (0.2 pl) and matrix solution (10 mg ca-cyano-4 hydroxycinnamic acid in 1 ml MeCN:water 7:3, 0.2 tl) were added to each bead and they were allowed to dry out before the 5 mass spectrum was acquired. The sequences of these active sequences were determined by the mass differences between the peaks from the ladder fragments (Fig. 5). Example 5 10 Direct design of inhibitor sequences from active sequences identified in the PEGA bead inhibitor library. Sixteen sequences of phosphinate based peptide 15 derivatives of formula (I) were resolved from some of those beads in the PEGA bead inhibitory library that remained dark after incubation with MMP-12. The sequences were used for the direct design of the soluble putative MMP-inhibitors shown in Table 1 (04-01 through 04-16).

WO 01/25264 PCT/EPOO/09173 59 Table 1. Sequences of soluble phosphinate based peptide derivatives designed directly from sequences identified on isolated beads from the solid phase combinatorial library of 5 phosphinic peptides that remained dark after incubation with MMP-12 Derivative P4 P3 P2 P2' P3' P4' 04-03 Thr Leu Tyr Leu Asp Gly 04-04 Val Leu Tyr Thr Leu Ser 04-05 Ile Met Tyr Val Lys Phe 04-07 Thr Leu Tyr Arg Ala Ile 04-09 Thr Leu Arg Leu Phe Phe 04-10 Ile Leu Arg Met Ala Pro 04-11 Ser Leu Phe Arg Asp Ile 04-12 Leu Met Phe Tyr Leu Ser 04-13 Ile Met Tyr Tyr Met Thr 04-14 Lys Phe Tyr Leu Tyr Ala 04-15 Tyr Ile Tyr Thr Met Pro 04-16 Ser Met Ala Tyr His Gly 04-17 Ile Met Arg Leu Ser Glu 04-02 Ile Leu Leu Asn Leu Ile 04-06 Leu Ile Glu Arg Lys Gly 04-08 Glu Phe Tyr Lys Tyr Asn In all cases P1-P1' consisted of the -Gly'{PO 2

H-CH

2 )Leu moiety. 10 In all cases except 04-03, 04-16 and 04-06, which terminate at P4', Gly constituted P5'.

WO 01/25264 PCT/EPOO/09173 60 Example 6. Design of inhibitor sequences from a consensus sequence. 5 Reliable sequences were resolved in 82 out of 112 dark beads after incubation of the PEGA bead inhibitory library with MMP-12. The frequencies of occurrence of the different amino acids -in the different subsites were determined giving 10 the distributions shown in Fig. 6. From this it is seen that, in subsite P4 the most predominantly observed amino acid was Leu/Ile/Hyp and Lys/Gln, in P3 it was Met and Leu/Ile/Hyp, in P2 it was Tyr, Arg and Phe, in P2' it was Tyr and Leu/Ile/Hyp, in P3' it was Ala, Tyr, Met and Leu/Ile/Hyp, and in P4' it was 15 Pro and Leu/Ile/Hyp. Based on these statistical observations as well as supporting observations of frequently occurring combinations of two or more amino acids, it was decided to use the sequence H-Leu-Met-Tyr-Gly{PO 2

H-CH

2 }Leu-Tyr-Ala-Pro-OH as a consensus sequence. It should be noted that Leu in P4 was 20 chosen irrespectively of the fact that it might as well have been the isobaric Ile or Hyp. However, when Edman degradation was used for sequencing of the three amino acids positioned N terminally to the -GlyT{PO 2

H-CH

2 }Leu- moiety of the phosphinate based pseudopeptides of five selected beads, it was observed 25 that Leu and Ile, but not Hyp, were present in P4 and P3. Thus, Leu (or Ile) seemed a more likely choice for the P4 position in the consensus sequence. By systematically exchanging single amino acids in the different subsites of the consensus sequence or by truncation 30 a further 25 sequences were designed (Table 2), that could give information about the amino acid preference in the difference subsites. In all cases P1-P1' consisted of the -Gly WO 01/25264 PCT/EPOO/09173 61

P{PO

2

H-CH

2 }Leu- moiety and except for the truncated sequences (04-01, 04-19, 04-20, and 04-21) a Gly constituted P5'. In the case of Gln in the P4-position, the N-terminal was acetylated due to problems with partial formation of pyro-Glu in 02-04. 5 Peptide 02-22 was prepared in order to study the mechanism of binding of phosphinate based peptide derivative inhibitors to MMPs (see Example 10). Table 2. The consensus sequence obtained from the PEGA-bead 0 library after incubation with MMP-12, and substitutions of this consensus sequence Derivative P4 P3 P2 P2' P3' P4' Consensus 02-01 Leu Met Tyr Tyr Ala Pro sequence Substitution P4 02-02 Ile Met Tyr Tyr Ala Pro P4 02-03 Lys Met Tyr Tyr Ala Pro P4 02-04 PGlu/Gln Met Tyr Tyr Ala Pro P4 02-21 Ac-Gln Met Tyr Tyr Ala Pro P3 02-05 Leu Leu Tyr Tyr Ala Pro P3 02-06 Leu Ile Tyr Tyr Ala Pro P3 02-07 Leu Pro Tyr Tyr Ala Pro P2 02-08 Leu Met Arg Tyr Ala Pro P2 02-09 Leu Met Phe Tyr Ala Pro P2' 02-10 Leu Met Tyr Leu Ala Pro P2' 02-11 Leu Met Tyr Ile Ala Pro P2' 02-12 Leu Met Tyr Hyp Ala Pro P3' 02-13 Leu Met Tyr Tyr Leu Pro P3' 02-14 Leu Met Tyr Tyr Ile Pro P3' 02-15 Leu Met Tyr Tyr Hyp Pro P3' 02-16 Leu Met Tyr Tyr Met Pro WO 01/25264 PCT/EPOO/09173 62 P3' 02-17 Leu Met Tyr Tyr Tyr Pro P3' 02-22 Leu Met Tyr Tyr Trp Pro P4' 02-18 Leu Met Tyr Tyr Ala Leu P4' 02-19 Leu Met Tyr Tyr Ala Ile P4' 02-20 Leu Met Tyr Tyr Ala Hyp Truncated 04-01 - - Tyr Tyr - Truncated 04-19 - - Tyr Tyr Trp Truncated 04-20 - - Tyr Tyr Met Truncated *04-21 - Met Tyr Tyr - P4-P4' 04-18 Ile Leu Phe Leu Met le In all cases Pl-Pl' consisted of the -GlyT{P0 2

H-CH

2 )Leu- moiety In all cases except 04-01, 04-19, 04-20, and 04-21, a Gly 5 constituted P5'. Example 7 Sequential screening of a solid phase combinatorial library of 10 phosphinic peptides (see also the schematic representation of the experimental set-up in Fig. 7). The dried resin (approx. 0.8 g, corresponding to 330,000 beads) was swelled in water, split in three equal sized portions 15 of approx. 265 mg each and lyophilised. After lyophilisation, 2 ml of either MMP-9, MMP-13 or MMP-14 in 10 mM Tris, 0.1 M NaCl, 10 mM CaCl2, 50 mM ZnCl2, 0.05 w/v % Brij, pH 7.50 was added per 100 mg of beads. Incubations was carried out at ambient temperature, in the dark for 24 hours. A small volume of highly 20 concentrated proteinase was added twice for MMP-9 during the 24 hrs incubation. The total proteinase concentrations were 370 nM WO 01/25264 PCT/EPOO/09173 63 for MMP-9, 170 nM for MMP-13, and 115 nM for MMP-14. After incubation, the resin was washed using 2 % TFA, water, TFA, water, buffer, water and buffer. An amount of water enough to swell the resin was added and approx. 1/50 of the resin was 5 placed on a glass plate (8.5 x 8 cm) . The backside of the plate was furnished with a line made by a yellow flourescent highlightening pen covered by transparent tape. A small fraction of the beads were placed under the microscope and sorted by the use of a steel needle while looking in the fluorescence 10 microscope. Dark beads were pushed to lie on top of the flourescent line. All the selected dark beads (apart from the beads obtained from the primary incubation with MMP-14) were inspected once more and persistently dark beads were split in two and transferred to plastic syringes equipped with a filter 15 stopper. The subsequent secondary incubation was done in these syringes. Approximately 1,000 dark beads were selected from each of the primary incubations. The resin was drained of water and 200-400 pl 200 nM MMP-9, or 100 nM MMP-13 or MMP-14 was added to the resin. Incubations were carried out for one hour at ambient 20 temperature, in the dark. After incubation, the beads were washed and sorted as described above, except that this time fluorescent beads were saved for sequence analysis by MALDI-TOF as described in Example 4. The dark beads were transferred into a new syringe with a filter stopper and incubated a third time 25 by the same procedure as used for the secondary incubations. Beads that were fluorecent after the third incubation were also saved for sequence analysis by MALDI-TOF.

WO 01/25264 PCT/EPOO/09173 64 Example 8 Design of sequences from active beads identified after 5 sequential incubations of the PEGA bead inhibitor library. Phosphinate based peptide derivatives from all of the 209 beads belonging to one of the twelve groups of beads that were fluorescent after the second or third MMP incubation (see Fig. 10 7) were sequence analysed by MALDI-TOF. For each of ten of the twelve groups, four sequences were selected as shown in Table 3 (05-01 to 05-40). None of the 79 sequences belonging to the last two groups of beads were selected, since these beads though remaining dark after the primary incubation with MMP-14 and 15 becoming fluorescent during the secondary incubations with MMP-9 (+14-9) or MMP-13 (+14-13) may have been false positives. Decisions to choose between Leu, Ile and Hyp or between Lys and Gln which have the same molecular mass and therefore cannot be distinguished by mass spectrometry analysis, included 20 comparisons of the 131 sequences from the sequential incubations with the sequences shown in Table 1 and Table 2. Furthermore, Hyp was chosen if Pro and/or Hyp was present in that particular position in either several of the sequences from the sequential incubations or in compounds that showed Ki values of less than 1 25 tM against the enzyme of interest. If not, Leu was chosen. The choice between Lys and Gln was made on the basis on the appearance of Asn and Arg in the same position. In ambiguous cases, both the Lys and Gln form of the same sequence were synthetised.

WO 01/25264 PCT/EPOO/09173 65 Table 3. Sequences of soluble phosphinate based peptide derivatives designed directly from sequences identified on beads isolated from the solid phase combinatorial library of 5 phosphinic peptides and representing different groups after sequential incubations with MMP-9, MMP-13 and MMP-14. Group Compound P4 P3 P2 P2' P3' P4' +13-9 05-01 Thr Ala Ser Met Phe Gly 05-02 Met Tyr Thr Tyr Lys Leu +13-14 05-03 Thr Arg Lys Ser Glu Leu 05-04 Thr Arg Gln Ser Glu Leu +13+14-9 05-05 Ser Met Leu Tyr Ala 05-06 Leu Ala Ala Tyr Phe Tyr +13+9-14 05-07 Glu Ser Asn Tyr Tyr Gly 05-08 Hyp Val Ala Ser Thr Gly +14+9-13 05-09 Hyp Tyr Met Leu Gln Leu 05-10 Hyp Tyr Met Leu Lys Leu +14+13-9 05-11 Val Phe Lys Met Ala Lys 05-12 Asn Arg Ala Phe Gln Ala +9-13 05-13 Arg Val Ser Asn Tyr Gly 05-14 Gly Hyp Lys Tyr Ans Arg +9-14 05-15 Gly Hyp Phe Glu Ser Leu 05-16 Val Ser His Ala Thr Phe +9+13-14 05-17 Tyr Pro Glu Ser Ala Ser 05-18 Hyp Met Val Leu Gln Phe +9+14-13 05-19 His Phe Lys Gln Gly Phe 05-20 Gln Pro His Phe Tyr Asp +13-9 05-21 Ser Hyp Asp Gly Val Glu 05-22 Arg Hyp Asp Thr Leu Hyp WO 01/25264 PCT/EPOO/09173 66 +13-14 05-23 Arg Pro Pro Leu Leu Gly 05-24 Lys Tyr Phe Gly Pro Met +13+14-9 05-25 Gly Met Gly Pro Phe Leu 05-26 Thr Asn Pro Asn Val Glu +13+9-14 05-27 Gly Thr Val Ala Lys Gln 05-28 Hyp Leu Leu Phe Hyp +9-13 05-29 Lys Thr Met Val Gln Leu 05-30 Lys Thr Met Val Gln Leu +9-14 *05-31 Tyr Met Arg His Ser Gly 05-32 Asn Val Val Tyr Leu Glu +9+13-14 05-33 Asp Ala His Asp Phe Gly 05-34 Thr Pro Leu Glu Ala Asp +9+14-13 05-35 Ala Pro Ala Leu Ala Gln 05-36 Arg Pro Ala Gln Met Arg +14+9-13 05-37 Tyr Ala Tyr Lys Tyr Glu 05-38 Tyr Ala Tyr Gln Tyr Glu +14+13-9 05-39 Thr Hyp Glu Val Ala Gly 05-40 Val Ala Lys Gln Arg Gly In all cases Pl-P1' consisted of the -GlyT'{PO 2

H-CH

2 }Leu moiety. The compounds that have Gly in position P4' terminates at this 5 position. In all other compounds, Gly constituted P5'.

WO 01/25264 PCT/EPOO/09173 67 Example 9 Kinetic analyses of novel substrate-mimicking phosphinate-tVe 5 MMP inhibitors and of a phosphinate-type MMP inhibitor synthesized on solid phase by a direct approach. Analyses of substrate mimicking phosphinate based peptide derivatives . 10 The 14 substrate-mimicking phosphinate based peptide derivatives 01-01A to 01-07B (see Table 4) were analysed kinetically by first pre-incubation in various concentrations with either MMP-1, MMP-3, MMP-9, MMP-12 or MMP-14 for 30 min at 37 OC and then incubation of the mixture with the quenched 15 fluorogenic peptide substrate, Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg

NH

2 (#M1895, Bachem, Switzerland), where Mca is (7 methoxycoumarin-4-yl) acetyl and Dpa is (2',4'-dinitrophenyl)

-L

2,3-diaminopropionyl. The rate of reaction was compared to the rate without inhibitor, and if possible the ratios were used for 20 calculation of the Ki-values for each combination of inhibitor and MMP (Table 4). For all of the 7 phosphinate based peptide derivatives, the diastereomeric A-form, which eluted first in HPLC purification, was more potent than the corresponding B-form, 25 which eluted last in HPLC. 01-07A, which mimics the very sensitive synthetic MMP-substrate, #M1895, was the most potent inhibitor of MMP-1, MMP-9 and MMP-14, and together with 01-04A the most potent inhibitor of MMP-3 and MMP-12. A Ki-value of 1.2 nM was observed for 01-07A with MMP-9. Similar values were 30 obtained for 01-07A with MMP-2 (2.2 nM) and MMP-13 (1.5 nM), whereas it showed increasing Ki-values with MMP-14 (10 nM) , MMP 20 (30 nM), MMP-1 (60 nM), MMP-7 (75 nM), MMP-12 (200 nM) and MMP-3 (2,200 nM) . Also, the affinity towards a member of the MMP related ADAM family, ADAM-17 or TACE (TNF converting enzyme) 35 was measured, giving a Ki value of 7,000 nM. These data indicate WO 01/25264 PCT/EPOO/09173 68 that compound 01-07A is a rather selective inhibitor of gelatinases and collagenases. A sub-micromolar Ki-value was observed for a few combinations of MMP and natural substrate mimicking inhibitor (01-01 to 01-06): 01-01A with MMP-9 and MMP 5 12, 01-03A with MMP-9, and 01-04A with MMP-3 and MMP-12, but it was never below 250 riM. Table 4. Kinetic analyses of substrate-analogue phosphinate based peptide derivatives No. Sequence MMP-l MMP-3 MMP-9 MMP-12 MMP-14 01-01A VAYGPcLKSRG 8.0 10-6 9.9 10-6 2.9 10-7 7.9 10~7 3.8 10~ 01-01B 8.3 10-5 1.4 10~4 7.8 10-6 3.6 10-5 6.1 10~5 01-02A AYGPcLKSG 2.1 10~' 5.7 10-s 3.5 10~' 8.6 10~6 2.0 10~5 01-02B 2.7 10~4 3.7 10~4 7.0 10-s 2.0 10~3 n.i. 01-03A VYGcLRSG 4.5 10- 2.4 10-s 6.8 10-7 6.0 10~6 6.3 10~ 01-03B 6.7 10~5 2.2 10~4 1.6 10-s 1.4 10~4 2.5 10~5 01-04A GLAGPcLWLPG 2.2 10~4 5.7 10~7 1.7 10-5 4.4 10-7 3.2 106 01-04B 3.5 10-4 1.9 10-5 3.9 10-4 1.6 10~5 1.1 10-4 01-05A LAGPcLWLG 2.1 10-s 3.8 10-6 1.1 10-5 1.6 10-6 1.7 10-5 01-05B 1.0 10- 9.4 10-s 1.5 10~4 9.1 10-s n.i. 01-06A LAGecLQLG 2.5 10~4 7.7 10-6 8.4 10-6 8.5 10~6 1.6 10~5 01-06B 2.4 10-4 3.1 10-4 2.9 10-5 9.8 10-4 n.i. 01-07A AGPLGPcLYARG 6.0 10-8 2.2 10-6 1.2 10~9 2.0 10~7 1.0 10~8 01-07B 2.8 10~7 8.2 10-6 3.4 10~9 3.2 10-6 6.2 10~7 10 n.i.: no measurable inhibition.

WO 01/25264 PCT/EPOO/09173 69 Analyses of a phosphinate based peptide derivative synthesized on solid phase bv a direct approach 5 The mean+/-SD of the Ki values of 03-01A were determined to be 86+/-17, 40+/-10, 19+/-2 and 27+/-4 ptM for MMP-9, MMP-12, MMP 13, and MMP-20, respectively, thus, indicating that phosphinate based peptide derivatives containing an -Alay(P(O) (OH) -CH 2 }Gly moiety may not be so efficient MMP-inhibitors. 10 Example 10 Kinetic analyses of phosphinate-type MMP inhibitors derived from a PEGA-bead library incubated with MMP-12 15 Sequences of phosphinate based peptide derivatives found on dark beads isolated after incubation of the Xl-X 2

_X

3 -Glyy (P0 2 CH 2 )Leu-X 4 -Xs-X 6 -designed PEGA-bead inhibitor library with MMP 12, were synthesized (04-02 to 04-17, see Table 1) and tested 20 for inhibitory activity towards MMPs. Also the consensus sequence LMYGy (PO 2

CH

2 )LYAPG (02-01) and selective amino acid substitutions thereof (02-02 to 02-22, see Table 2), as well as truncated forms of the consensus sequence (04-01 and 04-21) and of 02-22 (04-19) and of 02-16 (04-20), were synthesized and 25 tested. In preliminary kinetic analyses, MMP-12 and MMP-9 were preincubated with various concentrations of each of the two diastereomeric forms of the phosphinate based peptide derivatives in order to determine the approximate IC 5 0 -values. 30 As was the case for all of the 7 substrate-mimicking inhibitors, 01-01 to 01-07, all of the PEGA-bead derived inhibitors, except for the almost inactive 02-12, were more potent in the diastereomeric A-form (which elutes first in HPLC purification) than in the corresponding B-form (which elutes last in HPLC). 35 Thus, all further analyses were made with just the A-forms of the phosphinate based peptide derivatives of series 02 and 04.

WO 01/25264 PCT/EPOO/09173 70 Various MMPs (1-15 nM) were preincubated for 60 min at 37 oC with 8 or 14 different concentrations of each phosphinate based peptide derivative. Furthermore, each MMP was incubated alone in order to determine the rate of reaction of the 5 uninhibited proteinase. The concentrations of the phosphinate based peptide derivatives were from 0.1 to 10 times the Ki value, determined from preliminary experiments. The incubation buffer used for all MMPs, except MMP-3 contained 10 mM Trizma Base, 10 mM CaCl 2 , 50 ptM ZnCl 2 , 0.1 M NaCl with the pH adjusted 10 to 7.50 by addition of HCl. A 10 mM MES, 5 mM CaCl 2 , 100 pLM ZnCl 2 , pH 5.0 buffer was used for MMP-3 measurements. In experiments with ADAM-17/TACE a 20 mM Trizma Base, 10 mM CaCl 2 , 50 pM ZnCl 2 , pH 8.0 buffer was used. The preincubations were carried out with 0.05 % w/v Brij 35 added to the buffer. The 15 rate of reaction was measured using the quenched fluorogenic peptide substrate, Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2 (#M1895) from Bachem, Switzerland, where Mca is (7-methoxycoumarin-4 yl)acetyl and Dpa is (2',4'-dinitrophenyl)-L-2,3-diamino propionyl in a final concentration of 5 p.M (MMP-7 or MMP-12), 3 20 p.M (MMP-1 or MMP-14), 2 p.M (MMP-13) or 1 p.M (MMP-2 or MMP-9) . For MMP-3, a final concentration of 2 pM Mca-Arg-Pro-Lys-Pro Val-Glu-Nva-Trp-Arg-Lys(Dnp)-NH2 (#M2110 from Bachem) was used, while 0.5 pM Mca- Pro- Leu-Ala -Gln-Ala -Val -Dap (Dnp) -Arg- Ser- Ser Ser-Arg-NH 2 (#M2255 from Bachem) was used for assaying ADAM 25 17/TACE. The reaction was initiated by mixing the preincubated enzyme-inhibitor solution 1:1 with the prediluted substrate solution. The Ki-value for each pair of MMP and phosphinate based peptide derivative was calculated from plots of the rate of 30 reaction obtained in presence of inhibitor relative to the rate of reaction in absence of inhibitor: Iaei,[1] = (I[13o x [I]) /(Ki + [I]) 35 where [I] is the concentration of the inhibitor, and IRei,[I] and I[13. are the relative inhibition at [I] and the asymptotically WO 01/25264 PCT/EPOO/09173 71 measured relative inhibition at indefinitely high concentration of I, respectively. Of the sixteen phosphinate based peptides designed directly from sequences found in the library after incubation 5 with MMP-12, five (04-09A, 04-10A, 04-13A, 04-15A, and 04-17A) showed strong inhibition of MMP-12 with Ki values in the low nanomolar range. Another eight compounds showed intermediate inhibiton of MMP-12 with Ki values of less than 1 pIM, while the apparent inhibitory activity observed while on solid phase for 10 three compounds (04-02A, 04-06A, and 04-08A) could not be confirmed when they were tested in soluble form. A few of the compounds showed selective inhibition of MMP-12, with 04-15A being the most prominent example (40 to 2,000 times higher Ki values against MMP-7, -9, -13, and -14). 15 Table 5. Kinetic analyses of soluble phosphinate based peptide derivatives designed directly from sequences identified on beads isolated from the solid phase combinatorial library after incubation with MMP-12 Phosphinic Ki MMP-12 Ki MMP-9 Ki MMP-13 Ki MMP-14 Ki MMP-7 nM nM nM NM nM Peptide 04-03A 66±2 49,000±10,000 800±100 130±4 400±140 04-04A 985±315 270,000±50,000 39,000±3,000 4,800±500 1,200±500 04-05A 74±9 31,000±5,000 4,900+800 116±6 75±24 04-07A 875±465 32,000±5,000 2,200±200 1,350±10 4,600±300 04-09A 22±5 1,130±170 335±90 24±2 260±95 04-10A 25±4 56±4 20±10 20±5 455±70 04-11A 762±8 180,000±12000 1,500±1,400 1,500±300 4,700±400 04-12A 129±5 24,000±6,000 5,100±2,000 190±100 42±2 04-13A 8±4 7,600+1,200 1,500±100 450±120 100±35 04-14A 150±18 4,400±300 4,200±500 514±5 265±65 04-15A 19±15 35,000±13,000 3,000±1,000 4,500±1,700 750±90 04-16A 303±9 6,000±400 2,700±1,000 10,800±1,000 175±25 04-17A 23±8 400±80 204±51 190±33 1,700±200

-------------------------------------------------------------------------------------------------------------

WO 01/25264 PCT/EPOO/09173 72 04-02A 23,000±1, 100,000 311,000±56,000 > 75,000 29,000±11,000 000 04-06A >389,000±38 16,800±800 32,0005 ,000 2 ,100+200 2.300±1,700 ,000 04-08A 3,700±500 00,000±50,000 128,000±10,000 77,000±15,000 32,000i12,000 With a few exceptions, all of the soluble phosphinate based peptide derivatives (02-01A to 02-22A, 04-01A, 04-19A to 04-21A, and 04-18A), which were synthesized according to their 5 relationship to the consensus sequence (02-01), showed strong inhibition of MMP-12, with Ki-values in the low nanomolar range (Table 6). The exceptions were the truncated forms (04-01A and 04-19A to 04-21A), which were 150 to 20,000 times weaker than their corresponding full length forms, and the 2,500 fold 10 reduction in potency, which was observed when the Tyrosine in the P2' position of the consensus sequence was substituted by a Hydroxyproline. The Ki-values were also determined for each of the derivatives by incubation with MMP-9, MMP-13, MMP-14, MMP-7, and 15 MMP-20 (Table 6) . The selectivity of each inhibitor towards MMP 12 was calculated as the ratio between the Ki-value for MMP-12 and the Ki-value for any other MMP. From these selectivity numbers, it is particularly striking that the P3'-site provides a possibility of increasing the selectivity towards MMP-12 20 versus MMP-9 and MMP-13, when the Alanine of 02-01A is substituted by a Leucine (02-13A), Isoleucine (02-14A), Methionine (02-16A), Tyrosine (02-17A) or Tryptophan (02-22A). Furthermore, the Ki-values of 02-07A show that MMP-9 and MMP-13 favours a backbone bend introduced by substituting Methionine in 25 02-01A with Proline in the P3-site, whereas this substitution does not increase the inhibition of MMP-12. In inhibitors 02-12A and 02-1SA a Hydroxyproline was introduced in positions P2', and P3', respectively. Inhibitor 02-12A shows that neither MMP-12 WO 01/25264 PCT/EPOO/09173 73 nor MMP-9 or MMP-13 can accommodate a backbone bend in P2', whereas the 02-15A inhibitor shows that inhibition of MMP-12 is only moderately reduced when replacing the Alanine in P3' of 02 01A with Hydroxyproline. The affinity of MMP-9 and MMP-13 for 5 02-15A is more severely impeded by this change, which also reflects that small amino acids like Alanine are favoured very much by MMP-9 and MMP-13 in this position. For this position, it is also interesting to notice that MMP-14 is inhibited equally when this Alanine is substituted by a larger amino acid like 10 Leucine. In fact the substitution to Methionine (02-16A) increases the affinity towards MMP-14. A "second generation" consensus sequence compound (04-18A) was synthesized according to the results of the kinetic analyses of 02-01A to 02-22A. Compound 04-18A contained those amino 15 acids, which for each subsite gave the most potent inhibitor in the previous single amino acid substitutions of the consensus sequence 02-01A. However, when compared to the original consensus sequence, 04-18A was not a more powerful inhibitor of MMP-12 and only marginally better towards MMP-14 (and MMP-7), 20 whereas it was 40 to 100 times less efficient towards MMP-9 and MMP-13. Probably, these changes were mainly due to the substitution of Alanine in P3' with Methionine, as discussed above, and the pattern of inhibition by 04-18A resembled 02-16A very much. Thus, as has been suggested previously (Nagase et al, 25 1994) for petide-like substrates of MMPs, the additivity principle is not easily applicable for single position amino acid substitutions.

WO 01/25264 PCT/EPOO/09173 74 0 .0 4114 x + 1 + 1 + 1+1 +1 +1 +1 +1 0 + 0 n M r- ( 0 CN o 0N +1 0 0 -W +1 0 +1 In CN ~ N ( ON~ 0 4N 01 ON > 4J -4 N4 r N 11 00 -IV N n CN N N1 0; +1 +H +1 +1 +1 H +1 +1 +1- +1 +1 +Ir +1N0 +1 +1 N- 01 +i w 00 0) 0 N- + 0 In CN a, 0 N- M N) U) 0 o N N 0 4J4 04H 0 In 0Nl , m L -j LN +n m rn I n 0 + 1 + +i +-4 H' +1 +1 -H I N +1 +1 + N +1 0 wl 01 NN 0+1~ t 0 On Lnm W N -4 aC 0 N '0 4J.) 04 0 M '. N N H- N N 0 ~r + + : +I +1 +1 +1 +1 +1 +1 +1 +1 C.0 +10 0 .0 Z 0 a, In N +1 +1 '.0 N1 w~ w' O NN H- CN 0 0 0 C0 4Ia,0 a, 0 In 0 0 r 00 N+14 . N r-I kD r- 4 In Hn +1 N +1 +1i +1 +1 +1 +1 +1 +1 +1 +1 +1 0 +1 M40 fn~ IT In 0 m~ 0 C0 0 0 C0 Nn U)'~ '0~ iI N 0 0 0) 0 .- I -4-4.H N CN 0i Cd 0 0 0 H 0ac N Ln r- L Ln N N H N 4J~ +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 + 1 + r, M . ' . m N Go 0 H~ 0 00 m H- H4 H N CN H- CN Nq CN ' -4 a)m1 IH H: m: H: H H: W: c- H: N: H: : H: H 04 .11 -) N 0 . 0 ' c 4 In '. i 0 - N0 i-,4 0 0 04 O N, 0 0 0 N,4 H - q J. 04 0 0 0 0 00000000 WO 01/25264 PCT/EPOO/09173 75 1 +1 -1 (r N (\] LA +1 C) +1 + 1 +1 +1 +1 ' 0 ' 0 ' o 0 C) N '.D m' Ln w 0O- ~~ (n IV 14 I H N H 0 Ln r-4 - +1 1 +1C) i C) C) r- "' "'H H + H +1 + +1 +1 +1 +1 +1 + rn +1 + +1 +1 0~ H' t ' N D m A Hn 0 0 0 LA m (N H r- Hq r- C r- 0 (N C) C) C) o 0) C)l -W C)) N m cq +1 +1 H rn H I + 1 0N + +1 C1 +1 +++ ' 1 +1 0 0 + o oo 0 0A 0 C 0 C) C 0 0 co (N 0 z r- HN Hn - ( r C) c C) o C)C) C) C 0) C C) ) C) 0 0C) C) ) C) C) w~~~~ ~ ~ HC)n w 1 o q m+ +1 +1 +1 +1 +1 +1 HN +I +1 0 ) C 0C C> 0 + C) C 0 D C) C) C) C C) oC) r: w Co 1w C) C)% +1 +1 H HD LA +1 +1 +1 +1 + + +1 w i + +1 +1 +1 00 '0 '0 C) H w o C) M) C) 0 0 LAn 0H C) a' (Nw0r. H H H (N (N H (N H 0) i CC) 0~C C) C) 00 WO 01/25264 PCT/EPOO/09173 76 To investigate the binding mechanism of phosphinate based peptide derivatives to MMP-12, 02-22A was applied as a model inhibitor, since it contains a Tryptophan in P3'. The Tryptophan residue provides a fluorescent probe which is known to change 5 its fluorescence upon burial in a protein moiety. The kinetics of the binding can thus be measured directly as change of fluorescence in the inhibitor-enzyme solution, as opposed to the other inhibitors where binding is measured indirectly by the activity towards a substrate. Furthermore, 02-22A has 120 times 10 higher affinity towards MMP-12 than MMP-9, which allows the study of the binding mechanism of both a strong and a relatively weak inhibitor. Here only data for the binding of 02-22A to MMP 12 are shown. The incubations were performed in 10 mM TrizmaBase, 0.1 M NaCl, 10 mM CaCl 2 , 50 piM ZnCl 2 , pH 7.5 at 37 0 C. 15 The binding was followed by monitoring the change of fluorescence in a stopped-flow instrument (2 msec dead time) using 280 nm excitation and a 335 nm cut-off filter in the detector. The pseudo-first order rate of binding showed saturation at high concentrations of inhibitor, and the 20 amplitude remained constant over the concentration range examined. This suggest a two-step binding mechanism: Ki k2 E+I E-I El* 25 where E is enzyme, I is inhibitor, E-I is an intermediate complex rapidly formed with relatively low affinity and EI* is the final complex, which is the one giving the change in fluorescence. The value of the first dissociation constant K, is 93+13 30 ptM, and k 2 is 2.6+0.2 s~1. The value of k- 2 (5.9±0.7)x10- 4 s was too small to be calculated directly, and was thus calculated from the overall Ki-value (Ki = KixK 2 ; K2 = k- 2 /k 2 ) shown in Table 3 (21 + 6 nM) . These data suggest that the slow conversion from WO 01/25264 PCT/EPOO/09173 77 E-I to EI* involves a conformational change of either the inhibitor or the enzyme or both. Example 11. 5 Kinetic analyses of phosphinate-type MMP inhibitors derived from a PEGA-bead library sequentially incubated with MMP-9, MMP-13, and MMP-14. 10 Forty phosphinate based peptide derivatives, with sequences selected among 131 sequences identified by sequential incubations of the PEGA-bead libray with MMP-9, MMP-13 and MMP 14, were analysed (see Table 7) in soluble form for inhibitory activity against these three MMPs as described in Example 10. 15 Apparently, compound 05-34 belonging to the group (+9+13-14) was the only phosphinic peptide that gave the anticipated pattern of inhibiton by showing moderate Ki values of 560 nM and 100 nM for MMP-9 and MMP-13, respectively, and an approx. 100 times higher Ki value for MMP-14) . Furthermore, compounds 05-09, 05-10, 05-20 20 and 05-35 showed inhibition patterns that lived up to some, but not all of the expectations. Compounds 05-20 and 05-35 (both from the +9+14-13 group) showed relatively strong inhibition of all three MMPs, while 05-09 and 05-10 (both from the +14+9-13 group) inhibited MMP-14 quite selectively. Thus, according to 25 the selection procedure 05-09, 05-10, 05-20 and 05-35 should all be able to inhibit MMP-14 and MMP-9 but not MMP-13. Kinetic analyses of PEGA-bead library phosphinic peptides from sequential incubations. It is worth noting that the Series 05 phosphinic peptides 30 were tested as unpurified products, i.e. containing both the A and B diastereomeric form in a mixture as well as small amounts of impurities and NaCl. Further purification may lead WO 01/25264 PCT/EPOO/09173 78 to more convincing results, and at least will reduce the K 1 values, since the concentrations of the MMP-inhibitory A-form constitute just half (or slightly less) than the total amounts of the A/B-mixtures, which were used for the calculations in 5 Table 7. Table 7. Kinetic analyses of soluble phosphinate based peptide derivatives designed directly from sequences identified on beads isolated from the solid phase combinatorial library 10 after sequential incubations with MMP-9, MMP-13, and MMP-14. Phosphinic K, MMP-9 Kr MMP-13 Kr MMP-14 Group peptide pM PM pM +13-9 05-OAB 6±1 6.9±0.8 8±1 05-02AB 22±2 48±4 6.1±0.9 +13-14 05-03AB 250 58±5 260±45 05-04AB 500±117 340±26 87±7 +13+14-9 05-05AB 24±5 21±1 1.4±0.3 05-06AB 0.080±0.010 0.11±0.04 1.7±0.1 +13+9-14 05-07AB 105±18 73±4 0.56±0.05 05-08AB 5.1±0.8 3.3±0.1 4±1 +14+9-13 05-09AB 17±3 27±2 0.042±0.009 05-10AB 14±5 42±21 0.210±0.010 +14+13-9 05-11AB 6±1 6.5±0.7 0.8±0.3 05-12AB 12±5 22±4 1.6±0.3 +9-13 05-13AB 19±5 22±2 14±6 05-14AB 534±112 960±100 140±27 +9-14 05-15AB 2480 106±15 166±56 05-16AB 113±26 39±2 9±1 +9+13-14 05-17AB 1.7±0.6 0.850±0.028 1.0±0.3 05-18AB 240 NSI 39±4 1.1±0.5 WO 01/25264 PCT/EPOO/09173 79 +9+14-13 05-19AB 15±3 22i3 6±1 05-20AB 0..19±0.03 0.13±0.03 0.60±0.15 +13-9 05-21AB 33±8 2.1±0.9 2100 05-22AB 26±4 1.4±0.4 225 +13-14 05-23AB 26±8 32±8 50±8 05-24AB 270±40 100±65 50 +13+14-9 05-25AB 36±18 0.65±0.24 225 05-26AB 27±10 0.8±0.4 225 +13+9-14 .05-27AB 117±95 7±9 225 05-28AB 33±14 2.3±0.6 25 +9-13 05-29AB 228±54 30±39 8±3 05-30AB 178±15 29±8 20 +9-14 05-31AB 3.6±0.6 11±4 12±3 05-32AB nd nd Nd +9+13-14 05-33AB 66±6 25 25 05-34AB 0.56±0.08 0.100±0.02 32±2 +9+14-13 05-35AB 0.039±0.008 0.011±0.003 0.13±0.01 05-36AB 6±2 4.6±0.5 8±2 +14+9-13 05-37AB 2 0.8 ±0.3 18±9 05-38AB 2.5 1.0±0.3 12±2 +14+13-9 05-39AB 25 2.8±0.8 250 05-40AB 423±95 235±70 250 Example 12. The collagen invasion assay. Inhibition of osteoclast invasion through tvpe I collagen. 5 In order to study the role of osteoclast-derived MMPs in the recruitment and invasion of osteoclasts, we have developed a method which combines osteoclast purification and characterisation of their proteinase-dependent invasion 10 through type I collagen. Briefly, purified or non-purified WO 01/25264 PCT/EPOO/09173 80 osceoclasts were seeded on the membrane of tissue culture inserts (Costar, 12 pm pore size) which was coated with a type I collagen gel (Nitta Collagen, Japan; 10 pl/cm 2 of 3 mg/ml type I collagen pre-incubated at 37 0 C). Osteoclast migration 5 to the lower surface of the membrane was studied by quantitative microscopy after an overnight culture in the absence or presence of MMP inhibitors. We found that both purified and non-purified osteoclast could extend cell processes into the pores of the membranes and spread over the 10 lower membrane surface. We have previously shown that this migration process is inhibited by the general hydroxamate-type MMP inhibitors, RP59794 and BB-94, and by the natural MMP inhibitor, TIMP-2, but not by inhibitors of non-MMP proteinases (Sato et al, 1998) . This indicates that osteoclasts themselves 15 can overcome a collagen barrier by migrating through it without the participation of other cells and via a metalloproteinase dependent pathway. We have applied the novel phosphinate based peptide derivatives, 01-02B, 01-01A, 01-07A and 01-07B, in this model 20 and found a strict correlation between the Ki of the novel inhibitors towards MMP-9, and their ability to inhibit osteoclast invasion through type I collagen gel (Fig. 8 and Table 4 of Example 9) . When used in a concentration of 10 piM, the most potent of the phosphinate-type inhibitors, 01-07A, (Ki 25 = 0.6 nM towards MMP-9), reduced invasion with approx. 80%, which corresponds well to the potency of the previously studied hydroxamate-type MMP inhibitors. Thus, the phosphinate-type MMP inhibitors are at least as potent as the previously known general MMP inhibitors in 30 reducing osteoclast invasion through collagen.

WO 01/25264 PCT/EPOO/09173 81 Example 13. The pericellular collagenolysis assay Inhibition of osteoclastic pericellular collagenolysis. 5 In order to study the osteoclastic interaction with and movement over a surface of type I collagen, rabbit or murine bone cells were seeded and cultured on glass coverslips coated with type I collagen. The cells were cultured in the absence or presence of the hydroxamate-type MMP inhibitor GM6001. After 10 overnight culture, the collagen coating was studied by immunolocalisation with antibodies against type I collagen. Collagen-free zones surrounding the osteoclasts were evidence for a pericellular collagenolytic activity of the osteoclast. The areas of the collagen-free zones were significantly reduced 15 (approx. 80%) in the presence of GM6001 (Fig. 9). To investigate whether the effect of MMP inhibitors is also reflected in cell motility, time-lapse camera assisted studies of osteoclasts seeded on a two-dimensional type I collagen coating were performed. The migratory distance of the 20 osteoclasts was significantly reduced in the presence of GM6001. Furthermore, there was a strict correlation between the migrated path of the osteoclast and the collagen free zones (Fig. 10). 25 Example 14. The bone lining cell invasion assay Inhibition of osteoclast invasion through bone lining cells In order to study the invasion of osteoclasts through a 30 layer of bone lining cells, MC3T3-El cells (a murine calvarial osteoblast-like cell line) were seeded at a density of 3x10 4 WO 01/25264 PCT/EPOO/09173 82 per slice of bovine cortical bone (diameter 5.8 mm, thickness 150 ptm in a 0.32 cm 2 plastic well of a 96 well tissue culture plate) and cultured for 24 hrs. This resulted in a bone surface covered by a confluent layer of osteoblasts mimicking the 5 natural bone lining cells. For control purposes similar bone slices were prepared without seeding of cells. Mixed bone cells constituting osteoclasts or purified osteoclast were subsequently seeded onto the bone slices in the absence or presence of a proteinase inhibitor. Bone resorption 10 was quantified by staining for resorption pits and determination by computer-assisted microscopy of the resorbed surface area. Bone slices which were covered by an established layer of MC3T3-El cells had 40% less resorption than slices which were not covered by lining cells (Fig. 11). When bone was 15 covered by lining cells, the MMP inhibitor, GM6001, reduced bone resorption with approx. 50% (Fig. 11), whereas it had no effect when the bone slices were not covered (data not shown here, but described by us in Sato et al, 1998) . Thus metalloproteinases play an important role in osteoclast 20 penetration of an established protecting bone lining cell layer. Example 15. The bone lining cell elongation assay 25 Inhibition of TGF-D induced changes in bone lining cell morphology In order to mimic the bone lining cell layer, osteoblasts were cultured until confluence on either plastic surface or on 30 a bone substratum in tissue culture plates.

WO 01/25264 PCT/EPOO/09173 83 In brief, MC3T3-El cells were seeded in serum-free a-MEM on plastic in a density of 3x10 4 per cm 2 . After reaching confluence, the osteoblasts were cultured for 24 hours in the presence of 2.5 ng/ml TGF-3 and in the absence or presence of 5 10 pM proteinase inhibitor. The cell-free area was quantified by computer-assisted microscopy. Varying from experiment to experiment 2.5 ng/ml TGF-3 induced a 2-3 fold increase in cell free area compared to untreated controls. The induced osteoblast elongation could be reduced by more than 50% by 10 10 pM of the hydroxamate-type MMP inhibitor, GM6001 but not by 10 p.M of inhibitors of serine proteinases (aprotinin), cysteine proteinases (E-64), and aspartic proteinases (pepstatin) (Fig. 12). To clarify if the observed reduction in TGF-P induced osteoblast elongation was a general feature of MMP inhibitors 15 or restricted to hydroxamates, the novel phosphinate-type MMP inhibitor 01-07A was compared to GM6001. At 10 pM, the two inhibitors were equally efficient (Fig. 13). Thus, metalloproteinases are important for the induction of morphological change in bone lining cells by TGF-P. 20 The relevance of TGF-3 induced osteoblast elongation for bone resorption, and the effect of MMP inhibitors on this process, was studied by seeding 4.5x10 4 MC3T3-El cells on slices of bovine cortical bone (diameter 5.8 mm, thickness 150 pm) . The lining cells were cultured for 24 hrs in the absence 25 or presence of 2.5 ng/ml of TGF-3 and in the absence or presence of proteinase inhibitors. Bone slices without lining cells were prepared for control purposes. The bone slices were then treated with fixative (95% EtOH) for 20 min and washed carefully with culture medium. Subsequently, osteoclasts 30 isolated from long bones of 10-day-old rabbits were seeded on WO 01/25264 PCT/EPOO/09173 84 the bone slices and cultured for 72 hrs. Bone resorption was quantified by staining for resorption pits and determination of the resorbed surface area by computer-assisted microscopy. The presence of bone lining cells reduced pit formation by 50% 5 compared to bone slice which were not covered by cells (Fig. 14). Stimulation of the bone cell layer with 2.5 ng/ml TGF-3 almost recovered the bone resorption to the level observed for slices without cells. Stimulation of the lining cells with TGF 1 in the presence of 10 pM GM6001 blocked the bone resorption 10 to a level comparable to the one observed for slices with untreated lining cells (Fig. 14). Thus, bone resorption induced by TGF-P can be controlled specifically by MMP inhibitors through reduction of the elongation of bone lining cells. The hypothesis that TGF-P induced cells shape changes in 15 bone lining cells will lead to stimulated osteoclast resorption, was further tested in mouse calvariae cultures. In brief, 45 Ca labelled calvariae from 18-day-old mouse foetuses were cultured for 6 days in the presence of 2.5 ng/ml TGF-0, which significantly stimulated decalcification compared to non-treated 20 controls (Fig. 15) . In order to investigate whether MMP inhibition would reduce this stimulation of bone resorption in tissue culture, the novel phosphinate-type MMP inhibitor, 01-07A (10 ptM) was added simultaneously with TGF-P. Under this condition, decalcification was markedly reduced by 01-07A (Fig. 25 16).

WO 01/25264 PCT/EPOO/09173 85 Example 16. The metatarsal and tibia assays Inhibition of osteoclast migration and resorptive activity in metatarsal and tibia cultures 5 Pregnant NRMI mice were injected with 100 pCi asCa s.c. at day 16 of gestation and sacrificed 24 hrs later. The three middle metatarsals (i.e., the "triad") and tibiae were isolated from the hind limbs of the 17-day-old foetuses in 10 PBS, pH 7.4. In tissue culture, the right triad or tibia from individual foetuses was used for treatment with MMP inhibitor and the corresponding left triad or tibia from the same animal was used as paired vehicle treated control. One triad and one 15 tibia from each litter was killed at the start of the experiments by three freeze-thawing cycles and used for measuring the level of passive physico-chemical release of matrix bound radiolabelled calcium. Isolated triads and tibias were placed on pieces of filter paper (Millipore, 10 ptm pore 20 size) and allowed to preincubate for 2 hrs in 0.4 ml BGJb medium, supplemented with NaHCO 3 (2,200 mg/l), NaCl (900 mg/l), ascorbic acid (50 ptg/l), and albumax (1 g/l), before the medium was replenished with medium containing MMP inhibitor or vehicle. The bones were cultured for up to seven days with 25 conditioned media being replaced and collected at day 1, 2, 4 and 7 for analysis of 45 Ca release. At the last day of cultivation, triads and tibiae were demineralised in 5% acidic acid to recover the remaining 45Ca. The result is expressed as the ratio between the 45Ca 30 actively released in treated (T) bone (in %) and the corresponding 4 5 Ca actively released in the paired vehicle WO 01/25264 PCT/EPOO/09173 86 treated control (C) bone (in %) during the same period, using equation I listed below:

(%

4 5 CaT - % 45 CaK) 5 4SCa Release (T/C) = (I)

(%

4 sCac - % 4 5 CaK) The individual mean % 45 Ca value from treated (T), untreated control (C) and killed control (K) bones is 10 calculated as the ratio (in %) between the release of 45 Ca into the conditioned medium (in cpm) during the particular culture period and the total 4 5 Ca available in the individual bone (in cpm) at the initiation of the particular period, using equation II listed below: 15 %4 Ca = (4sCa (cpm) / 4 5 Ca (cpm)totai)x 100% (II) For histological inspections, the triads were cultured for two or four days and then rinsed twice in PBS before 20 fixation in 4% neutral-buffered formaldehyde for 3 hrs at 4 0 C, decalcification in 5% formic acid and 5% formaldehyde for 3 hrs at 4 0 C. Triads were then stained for TRAP by pararosanilin as coupler and 10 mM L(+)-tartratic acid as inhibitor, dehydrated in a graded series of ethanol and embedded in 25 paraffin. Serial 5 pLm sections were obtained by a microtome, counterstained with Ehrlich's haematoxylin for 30 seconds and finally mounted before quantification. The hydroxamate-type MMP inhibitor, BB-94, dose dependently reduced 4 5 Ca release in triads of metatarsals as 30 compared to paired vehicle-treated triads (Fig. 17) . This was seen both when triads were cultured under non-stimulated and WO 01/25264 PCT/EPOO/09173 87 PTH-stimulated conditions where the mean + SEM half inhibitory concentration (ICso) for BB-94 was calculated to be 11 - 3 nM (n = 4) and 23 + 5 nM (n = 4), respectively, with Hill slopes of -1.1 + 0.3 and -0.8 + 0.2, respectively. As demonstrated by 5 sections of triads stained for TRAP+ osteoclasts, BB-94 inhibited migration of maturing osteoclasts from the periosteal layer into the calcified matrix when administered at a concentration (3 piM) that completely blocks 45Ca release (Table 8). 10 Table 8. Histomorphometric analysis on paired triads treated with vehicle or BB-94. Day 2 Day 4 OCs out OCs in % 45 Ca OCs out OCs in % 45 Ca Vehicle 8.8+0.9 10+1.3 18+1.7 9.2+1.2 14+0.5 47+4.2 BB-94 10+1.3 0.9±0.44 3.5+1.2** 14+1.0* 0 5.0±1.1** Triads were cultured with vehicle as control or 3 pM BB-94 for 2 or 4 days 15 when bones were fixed, decalcified and stained for TRAP-positive osteoclasts (OCs) . The mean + SEM number of osteoclasts outside (OCs out) or inside (OCs in) the calcified matrix was counted in three non-adjacent sections from each metatarsal (n = 16 from 4 independent experiments). The bone resorption is expressed as % 45 Ca, which is the mean + SEM calcium 20 release in percentage of total calcium available corrected for killed control. Significance levels * p < 0.05, ** p < 0.005, # p < 0.00001. In the tibia assay, BB-94 dose-dependently, but less potently than in the triads, reduced the level of 45 Ca release 25 during basal and PTH-stimulated cultivation with calculated WO 01/25264 PCT/EPOO/09173 88 IC5o values of 280 ± 71 nM and 55 + 26 nM , (n = 4, for each condition) (Fig. 17) . Furthermore, the inhibition of 4 sCa release in the tibiae was incomplete reaching 68 + 7% and 59 + 4% at 3 jtM BB-94 under non-stimulated and PTH-stimulated 5 conditions, respectively. Similar results were observed for another hydroxamate type MMP inhibitor, GM6001, which dose-dependently reduced the 4 sCa release from both metatarsal and tibia cultures with IC 50 values of 22 + 6 nM and 75 + 8 nM, respectively and with Hill 10 slopes of 0.93 + 0.20 and 1.13 ± 0.10, respectively (Fig. 18). As seen for BB-94 the migration of osteoclasts from the periosteal layer into the calcified cartilage in metatarsals was totally abolished in the presence of 3 pIM GM6001 (data not shown) . In contrast to its complete inhibition in metatarsals, 15 the inhibition in tibiae by GM6001 (like for BB-94) was not complete reaching a maximal 75% at micromolar concentrations (Fig. 18). Also some of the novel phosphinate based peptide derivatives were efficient, dose-dependent inhibitors of 20 decalcification in the metatarsal model (Table 9A). Among these, the most potent inhibitor of MMP-9, 01-07A (Ki = 1.2 nM) , was able to inhibit the 4 Ca release in metatarsals with an IC 50 value of 200 nM. As for BB-94 (Ki = 0.3 nM for MMP-9,

IC

50 in metatarsals approx. 15 nM) and GM6001 (Ki = 1.0 nM for 25 MMP-9, ICs 0 in metatarsals approx. 20 nM), the migration of osteoclasts from the periosteal layer into the calcified cartilage according to histological inspection of the metatarsals was completely blocked in the presence of 2 ptM 01 07A, thus strongly indicating that both hydroxamate-type and 30 phosphinate based MMP-inhibitors can block the migration of osteoclasts. In order to increase the understanding of the WO 01/25264 PCT/EPOO/09173 89 importance of individual MMPs in the decalcification of metatarsals, the correlations between the individual TC5o value of phosphinate based MMP-inhibitors in these tissue cultures and their Ki values towards various MMPs were investigated (Fig 5 19). This correlation study is a simplification of the probable complex interactions between various MMPs, but still it is interesting to notice that the Ki values of the inhibitors towards MMP-9 and MMP-13 are good predictors of their ability to reduce decalcification in the metatarsal 10 cultures, whereas the seems to be no or even inverse proportionality for Ki values towards MMP-7, MMP-12, MMP-14, and MMP-20. When applied to tibia cultures at 10 pM or even higher concentrations, none of the hitherto tested phosphinate-type 15 inhibitors affected 4 5 Ca release (Table 9B), whereas the IC 50 values of BB-94 and GM6001 in this assay were approx. 55 nM and 75 nM, respectively. This indicates, that the hydroxamate type inhibitors also directly affect the removal of mineralised osseous substance by osteoclasts, whereas the 20 phosphinate based peptide derivatives do not. The two potent, hydroxamate-type MMP inhibitors BB-94 and GM6001, which have little selectivity (low or sub-nanomolar Ki values) for a variety of MMPs including MMP-1, -2, -3, -7, -9, -12, -13, -14, and -20, inhibited the migration and reduced 25 the resorptive activity of osteoclasts in bone cultures, whereas the potent MMP-9 selective phosphinate-type inhibitor, 01-07A (see Table 4 of Example 9), selectively inhibited the migration but did not affect the resorptive activity of osteoclasts. 30 WO 01/25264 PCT/EPOO/09173 90 Table 9 Effect of phosphinate based peptide derivatives on decalcification in bone tissue cultures A. Metatarsals Cultures - Migration and resorption IC50 in pM Compound GM6001 0.02 +/- 0.006 01-07A 0.20 01-07B 23 02-05A 10 02-06A 5 02-07A 0.9 +/- 0.2 02-08A 5 02-10A 14 +/- 8 02-11A 15 02-16A 23 +/- 6 04-18A 10 +/- 1 04-09A 7 04-10A 4 +/- 3 05-06A/B 7 05-10A/B 14 +/- 1 05-20A/B 11 +/- 2 05-23A/B No Inhibition 05-24A/B No Inhibition 05-35A/B 5 WO 01/25264 PCT/EPOO/09173 91 B. Tibia cultures - ResorDticn Compound IC50 in piM GM6001 0.1 +/- 0.01 01-07A > 30 02-07A No Inhibition 02-08A > 75 02-16A No Inhibition 04-18A No Inhibition 04-10A > 75 5 Example 17 The demineralised collagenolysis assay Inhibition of removal of collagen fibres in calvaria cultures 10 In order to investigate the roles of MMPs and cysteine proteinases in stimulated bone resorption, we cultured calvariae isolated from 5-day-old mice in the presence of 100 nM PTH, and in the absence or presence of various MMP inhibitors (the two hydroxamate-type inhibitors, RP59794 and 15 CT1166; and the carboxylate-type inhibitor, CI-1) or cysteine proteinase inhibitors (E-64 or EST) for 1 day, and analysed the ultrastructure of the subosteoclastic resorption compartment. We made ultrathin sections of these calvariae, stained them with uranyl and lead, and examined them by 20 electron microscopy. When the calvariae were cultured in the absence of a proteinase inhibitor, the front line of the ruffled border of the osteoclast was almost against the mineralised matrix, whereas a large amount of demineralised WO 01/25264 PCT/EPOO/09173 92 collagen fibres were observed between the ruffled border of the osteoclast and the mineralised matrix in calvariae cultured in the presence of either an MMP inhibitor or a cysteine proteinase inhibitor, thereby indicating that the 5 bone could be demineralised to some extent, but that collagen degradation was impaired by both types of proteinase inhibitor. In order to quantify this effect, pictures were taken and the surface of demineralised bone matrix between osteoclasts and mineralised bone was measured using a 10 computerised X-Y tablet. The areas of demineralised matrix were 5 to 7 times larger in the calvariae treated with a proteinase inhibitor than in the controls, the effects being most dramatic and arising fastest when cysteine proteinases were inhibited. These data shows that in PTH-stimulated 15 calvariae, not only cysteine proteinases but also metalloproteinases play a rate limiting role in the solubilisation of bone matrix by osteoclasts. In the same cultures, it was interesting to notice that the areas of demineralised matrix in those resorption pits from 20 which the osteoclasts had disappeared were much larger when the calvariae were cultured in the presence of an MMP inhibitor than when cultured with a cysteine proteinase inhibitor. In order to investigate the proteolytic events occurring in the latter situation, we precultured the calvariae for 1 day with a 25 cysteine proteinase inhibitor. Some cultures were stopped at this stage for determination by electron microscopy of the initial amount of demineralised collagen fibres in the calvariae, and the remaining cultured for a further 18 hours in the absence or presence of an MMP inhibitor or a cysteine 30 proteinase inhibitor, and then processed for electron microscopy. We saw that after the osteoclast had left its WO 01/25264 PCT/EPOO/09173 93 resorption pit, bone lining cells occupied this pit and showed phagocytic activity towards the collagen remnants left by the osteoclast. During the 18-hour cultures performed in the absence of inhibitor or in the presence of a cysteine 5 proteinase inhibitor, the areas of collagen remnants dropped to 20% of the initial values, whereas in the presence of an MMP inhibitor, this collagen degradation was almost completely inhibited. 10 Example 18 The calvarial decalcification assay Inhibition of 45 Ca release from foetal mouse calvaria cultures. 4 SCa labelled calvariae from day 18 mouse foetuses were 15 carefully isolated and subsequently cultured for 6 days in the absence or presence of 100 nM PTH, and in the absence or presence of either 10 iM of the hydroxamate-type MMP inhibitor, GM6001, or various concentrations of the novel phosphinate-type MMP inhibitor, 01-07A. Resorption was measured by the relative 20 release of 4 5 Ca into the conditioned medium compared to the total amount of 45Ca in the corresponding calvaria. The PTH stimulated release of 4 5Ca was inhibited dose-dependently by 01 07A (ICso: approx. 2 iM) and to a level similar to that obtained by GM6001 (Fig. 20). 25 Example 19 The tetracycline-labelled bone resorption assay Inhibition of bone resorption in vivo 30 In order to investigate the influence of MMP inhibitors on bone resorption in vivo, bones of newborn mice were labelled by WO 01/25264 PCT/EPOO/09173 94 tetracycline. Briefly, pregnant mice were injected with 160 mCi 3 H-tetracycline 1 day before parturition. Between 3 and 6 days of age, half of the mice of each litter were injected s.c. twice a day with a test substance, the other half were injected 5 with saline. The mice were sacrificed at the age of 7 days, their tibiae and femurs were isolated, the 3H-tetracycline was released by demineralisation of the bones in acid, and the level of 3H-tetracycline was determined. The test substances included the MMP inhibitor, RP59794 (200 pg bid) , as well as 10 some established inhibitors of bone resorption: the bisphophonates, clodronate (Cl 2 MBP) and pamidronate (APD) (both at 100 tg bid) , and the cysteine proteinase inhibitor, E-64 (100 pLg bid) . Tibiae and femurs of mice treated with the MMP inhibitor had significantly elevated contents of 3 H compared to 15 the corresponding bones of the non-treated control group of mice, and as high levels of 3 H as the bisphosphonate and cysteine proteinase inhibitor treated groups (Fig. 21) . This shows that, more 3H-tetracycline was retained in the bones of mice treated with MMP inhibitors or established inhibitors of 20 bone resorption, and that the rate of bone resorption was decreased by these treatments. We conclude that the inhibitory activity exerted by MMP inhibitors on bone resorption can also be detected in vivo. 25 Example 20 The bone metastasis assay Inhibition of experimental bone metastasis When breast cancer cells are inoculated intracardially 30 into nude mice the cells will metastasise to bone and induce osteoclast-dependent osteolysis.

WO 01/25264 PCT/EPOO/09173 95 Briefly, the oestrogen-receptor negative breast cancer cell line MDA-MB-231 (MDA-231/P) characterised by a high metastatic potential was cultured in E-MEM supplemented with penicillin (100 U/ml) , streptomycin (100 mg/ml) , and 5% (v/v) 5 FCS. An in vivo selected bone-seeking subclone of the MDA-231/P cells was isolated in our laboratory from bone metastases in a nude mice 28 days after cardiac inoculation. This subclone is hereafter referred to as MDA-231/B. MDA-231/B cells were maintained for up to 6 passages under the same conditions as 10 MDA-231/P. A suspension of 10s MDA-231/B cells in 0.1 ml were injected into the left cardiac ventricle of 4-week old female BALB/c-nu/nu mice (M&B, Denmark) at Day 0. Three days before (Day -3), a slow releasing pellet containing 0.72 mg of 17 9 oestradiol (Innovative Research of America Inc., Florida, USA) 15 had been implanted s.c. in each of the mice. The pellet was removed at day 7. X-ray inspection was done at Day 19 and 28, and the radiographs of mice were captured by a video camera and projected to a computer monitor, where areas of osteolytic metastases in limbs and pelvis were measured using the 20 stereological program CastGrid (Olympus A/S, Denmark). The mice were usually sacrificed at Day 28, and the tibiae and femurs were isolated and fixed in 4% formaldehyde (41C for 48 hrs), before decalcification in 15% EDTA-solution, pH 7.4, for 3 weeks. Five-mm sections of paraffin embedded tibiae or 25 femurs placed on Superfrost PLUS slides (Menzel-Gl&ser, Germany) were air-dried overnight at 37 0 C. Slides were deparaffinised in two changes of toluene and rehydrated in decreasing concentrations of methanol. The demonstration of tartrate resistant acid phosphatase (TRAP) positive cells (i.e., 30 osteoclasts) in bone sections was performed by incubation of the slides for 2 hrs in 0.7% (w/v) sodium acetate, 0.05% Naphtol-AS- WO 01/25264 PCT/EPOO/09173 96 I-phosphate, 0.74% (w/v) 5,5-diethyl-barbituric-acid, 0.23% (w/v) di-sodium tartrate, 0.16% (w/v) NaNO 2 , 0.16% (w/v) pararosaniline. The slides were subsequently rinsed in running tap water for 10 min before counterstaining with Mayers 5 haematoxylin for 1 min. Finally, the slides were rinsed in running tap water for 10 min, dehydrated in increasing concentrations of ethanol and toluene and coversliped with DPX Mountant (Fluka, Neu-Ulm, Switzerland). In one. experiment, the nude mice, which at Day 19 had 10 osteolytic lesions according to radiography, were randomly divided into two groups. The mice of the two groups were i.p. injected daily for 10 consecutive days (Day 19-28) with vehicle or the hydroxamate-type MMP inhibitor, BB-94 (60 mg/kg) and then X-rayed again at Day 28. At Day 28, the area of tumour lesions 15 in both groups had increased significantly from the equal levels at Day 19, but the increase was significantly higher in vehicle than BB-94 treated mice (Fig. 22) . When sections of areas with osteolytic metastases were inspected histologically and compared for vehicle and BB-94 treated mice, vascular damage manifested 20 by central haemorrhages in disintegrating tumours and areas of necrosis were observed for the BB-94 treated mice. Moreover, bone surfaces near these necrotic areas lacked the normal appearance of numerous actively resorbing osteoclasts. In a second experiment, the nude mice were randomised 25 three days before cancer cell inoculation (Day -3) to receive either vehicle for 32 days (until Day 28) or GM6001 (100 mg/kg weight) for either 11 days (until Day 7) or 32 days (until Day 28) by i.p. injections once daily. The GM6001 was well tolerated by the animals as judged by the animal weight curves (Fig 23A). 30 The number (data not shown) and area (Fig. 23B) of osteolytic lesions was significantly higher in the vehicle treated group WO 01/25264 PCT/EPOO/09173 97 than in the group treated with GM6001 group for 33 days, and a clear but non-significant reduction was also observed in the group treated with GM6001 for 11 days (Fig. 23B). In a third experiment, the nude mice were randomised three 5 days before cancer cell inoculation (Day -3) to receive either vehicle or GM6001 (100 mg/kg weight) by i.p. injections once daily for a total of up to 53 days (Day -3 to 49). The end-point measurement was animal survival. A highly significant increase in survival time was observed for the GM6001 treated mice (Fig. 10 24). Example 21 The osteoblast persistence assay Persistence of osteoblasts in 3-dimensional collagen gels 15 Mouse osteoblasts were seeded in a 3-dimensional collagen gel by preparation of a single cell suspension of 2,5*105 MC3T3-El cells per ml in solubilised collagen (Nitta, Japan, 2.4 mg/ml of type I collagen at 4*C) . The collagen was gelified 20 in 48-well culture plates (Costar, The Netherlands, 0.3 ml per 0.8 cm2 well at 37'C) and the cultures continued for 6 days by addition of 0.4 ml/well of cx-MEM containing 5% FCS. The osteoblasts were cultured in the absence or presence of 10 yM of inhibitors of the different classes of proteolytic enzymes: 25 the serine proteinase inhibitor, aprotinin; the cysteine proteinase inhibitor, E-64; the aspartic proteinase inhibitor, pepstatin; and the hydroxamate-type matrix metalloproteinase inhibitor GM6001. The number of viable osteoblasts at Day 6 was assessed by staining with AlamarBlueTM (Accumed, USA) (Fig. 25). 30 To clarify if the observed reduction in osteoblast number was a general feature of MMP inhibitors or restricted to WO 01/25264 PCT/EPOO/09173 98 hydroxamates, the novel phosphinate based MMP inhibitor 01-07A was compared to GM6001 and another well characterised hydroxamate-type MMP inhibitor, BB-94. Furthermore, to ensure that the previously observed GM6001-induced reduction in 5 osteoblast number was not limited to the cell line MC3T3-E1, the experiment was also performed in a primary osteoblast culture. Primary osteoblasts were obtained from mouse calvariae from 18-day-old foetuses. After 1 passage, cells were seeded at 10 1*105 cells per ml collagen suspension as described above and cultured for 4 to 6 days in the absence or presence of 10 pM MMP inhibitor. The reduction in osteoblast numbers induced by the two hydroxamate inhibitors, GM6001 and BB94, was not mimicked by the phosphinate-type inhibitor 01-07A (Fig. 26). 15 Similar results were obtained in a dose-response experiment, when using MC3T3-El cells instead of primary osteoblasts (Fig. 27) . Thus, the adverse effect on osteoblast number observed for hydroxamate-type MMP inhibitors are not observed for a phosphinate-type MMP inhibitor or inhibitors of non-MMP 20 proteinases. Examole 22 The in vitro bone formation assay Osteoblast differentiation and bone formation in vitro 25 The influence of MMP inhibitors on cell differentiation and bone nodule formation was investigated in osteoblast cultures. Differentiation of MC3T3-El osteoblasts was induced by 30 addition of 50 pg/ml ascorbic acid and 10 mM -glycerol phosphate to a-MEM supplemented with 10% FCS. Cell WO 01/25264 PCT/EPOO/09173 99 proliferation was monitored by staining with AlamarBlue r (Accumed, USA). Osteoblast differentiation was assessed by quantification of alkaline phosphatase activity in the conditioned medium (a marker for the early phase of osteoblast 5 differentiation) and by quantification of bone nodule formation according to Alizarin Red staining (a marker for the late phase of osteoblast differentiation) . MC3T3-El cells were cultured for 21 days in the absence or presence of increasing concentrations of the hydroxamate-type MMP inhibitor, GM6001 10 (Fig. 28), or at a fixed concentration of 10 ptM of either GM6001 or the novel phosphinate-type MMP inhibitor, 01-07A (Fig. 29) . No significant effects of the MMP inhibitors were observed for osteoblast proliferation or early phase differentiation, (i.e., alkaline phosphatase activity) (data 15 not shown), whereas GM6001 but not 01-07A caused a significant reduction in late phase differentiation (i.e., bone formation). Thus, the adverse effect on bone formation observed for the hydroxamate-type MMP inhibitor was not seen for the phosphinate-type MMP inhibitor. 20 WO 01/25264 PCT/EPOO/09173 100 References. Bartlett PA, Marlowe CK. Phosphonamidates as transition-state analogue inhibitors of thermolysine. Biochem 22: 4618-4624. 5 1983. Baylis EK, Campbell CD, Dingwall JG. 1-aminoalkyl-phosphonous acids. Part 1. Isosters of the protein amino acids. J Chem Soc Perkin Trans 1: 2845-2853. 1984. 10 Blankemeyer-Menge B, Nimitz M, Frank R. An efficient method for anchoring Fmoc-amino acids to hydroxyl-functionalized solid supports. Tetrahedron Lett 31: 1701-1704. 1990. 15 Blavier L, Delaissd JM. Matrix metalloproteinases are obligatory for the migration of preosteoclasts to the developing marrow cavity of primitive long bones. J Cell Sci 108: 3649-3659. 1995. 20 Brown PD, Giavazzi R. Matrix metalloproteinase inhibition: a review of anti-tumour activity. Annals Oncol 6: 967-974. 1995. Buchardt J, Meldal M. A chemically inert hydrophilic resin 25 for solid phase organic synthesis. Tetrahedron Lett 39: 8695 8698. 1998. Chen QJ, Lund L, Lenhard T, Engsig M, Winding B, Therkildsen B, Pedersen AC, Larsen D, Werb Z, Foged NT, Delaiss6 JM. MMP- WO 01/25264 PCT/EPOO/09173 101 9 is a regulator of osteoclast recruitment as demonstrated by targeted mutagenesis. Bone 23 (Suppl.): S548 (SA086). 1998. Everts V, Korper W, Jansen DC, Steinfort J, Lammerse I, Heera 5 S, Docherty AJ, Beertsen W. Functional heterogeneity of osteoclasts: matrix metalloproteinases participate in osteoclastic resorption of calvarial bone but not in resorption of long bone. FASEB J 13: 1219-1230. 1999. 10 Holmes CP, Jones DG. Reagents for combinatorial organic synthesis: Development of a new o-nitrobenzyl photolabile linker for solid phase synthesis. J Org Chem 60: 2813-2819. 1995. 15 Klein L, Li QX, Donovan CA, Powell AE. Variation of resorption rates in vivo of various bones in immature rats. Bone Miner 8:169-175. 1990. Meldal M, Svendsen I, Breddam K, Auzanneau FI. Portion-mixing 20 peptide libraries of quenched fluorogenic substrates for complete subsite mapping of endoprotease specificity. Proc Natl Acad Sci 91:3314-3318. 1994. Nagase H, Fields CG, Fields GB. Design an characterization of 25 a fluorogenic substrate selectively hydrolysed by stromelysin 1 (matrix metalloproteinase-3). J Biol Chem 269: 20952-20957. 1994.

WO 01/25264 PCT/EPOO/09173 102 Renil M, Ferreras M, Delaiss6 JM, Foged NT, Meldal M. PEGA supports for combinatorial peptide synthesis and solid-phase enzymatic library assays. J Peptide Sci 4: 195-210. 1998. 5 Saftig P, Hunziker E, Wehmeyer 0, Jones S, Boyde A, Rommerskirch W, Moritz JD, Schu P, von Figura K. Impaired osteoclastic bone resorption leads to osteopetrosis in cathepsin-K-deficient mice. Proc Natl Acad Sci USA 95:13453 13458. 1998. 10 Sato T, Foged NT, Delaiss6 JM. The migration of purified osteoclasts through collagen is inhibited by matrix metalloproteinase inhibitors. J. Bone Miner Res 13: 59-66. 1998. 15 Yiotakis A, Vassiliou S, Jiracek J, Dive V. Protection of the hydroxyphosphinyl function of phosphinic dipeptides by adamantyl. Application to the solid-phase synthesis of phosphinic peptides. J Org Chem 61: 6601-6605. 1996. 20 Buchardt et al, "Phosphinic Peptide Matrix Metalloproteinase 9 Inhibitors by Solid-Phase Synthesis Using a Building Block Approach". Chem. Eur. J. 1999, 5, No.10, 2877-2884.

WO 01/25264 PCT/EPOO/09173 103 Cited patents and patent aoplications. W098/03516 Reiter LA. Phosphinate based inhibitors of matrix 5 metalloproteinases. Pfizer Inc. Filed Jun. 30, 1997. WO93/16118 Meldal MP. Polyethylene or polypropylene glycol containing polymer. Carlsberg A/S. Filed Feb. 13, 1992. 10 EP-A-0,611,756 Sohda T, Fujisawa Y, Yasuma T, Mizoguchi J, Kori M, Takizawa M. Alcohol or aldehyde derivatives as cathepsin L inhibitor and bone resorption inhibitor. Takeda Chemical Ind. Ltd. 15 Filed Feb. 17, 1994. US-A-5,579,700 Caldwell CG, Durette PL, Goulet JL, Hagmann WK, Sahoo P. Substituted phosphinic acid-containing peptidyl derivatves as 20 antidegenerative agents. Merck & Co, Inc. Filed Nov. 9, 1995. US-A-5,776,903 Dive V, Jiracek J, Yiotakis A. Peptide derivatives usable as zinc endopeptidase 24-15 inhibitors. Commisariat A l'Energie 25 Atomique. Filed Jan. 24, 1996 EP-A-0276436 WO 01/25264 PCT/EPOO/09173 104 Broadhurst MJ, Hando BK, Johnson WH, Lawton G, Machin PJ. Phosphinic acid derivatives. Hoffmann-La Roche. Filed Dec. 10, 1987. 5

Claims (21)

1. 2. an amino-protecting group,
2. 3. an alkyl, alkenyl, alkynyl or aryl group,
3. 4. a group R -NH-CR"R id-CO- in which 4.1. R is 20 4.1.1. a hydrogen atom, 4.1.2. a natural or unnatural a-amino acid, or a peptide consisting of the same, 4.1.3. an alkyl, alkenyl, alkynyl or aryl group, 4.2. Ric and Rid independently of each other are WO 01/25264 PCT/EPOO/09173 106 4.2.1. a hydrogen atom, 4.2.2. a radical corresponding to a side chain of a natural and non-natural a-amino acid, 4.2.3. an alkyl, alkenyl, alkynyl or aryl group, 5 5. a group R "-CO- in which R" is
4. 5.1. a hydrogen atom. 5.2. an alkyl, alkenyl, alkynyl or aryl group,
5. 6. a group R -SO 2 - in which RM is an alkyl, alkenyl, alkynyl or aryl group. 10 R 2 , R 3 , R 4 and R 8 independently of each other are 1. a group as defined for Ri", 2. a group R 2a-CH 2 - in which R2a is 2.1. an aryl group, or 15 2.2. a heteroalicylic or heteroaromatic group, R 5 is 1. a hydrogen atom, or 2. an alkyl, alkenyl, alkynyl or aryl group, 20 R 6 and R 7 independently of each other are 1. a hydrogen atom, or 2. an alkyl group, 25 R, is 1. a group R 3aXc- in which 1.1. R 3 a is WO 01/25264 PCT/EPOO/09173 107 1.1.1. an alkyl, alkenyl, alkynyl or aryl group, 1.1.2. a heteroalicylic or heteroaromatic group, or 1.1.3. a group as defined for R 2 1.2. Xc is 0, S or NH, 5 Xa and Xb are: 1. 0 2. S 3. NH. 10 2. The use as claimed in Claim 1, wherein the compound of formula (I) or a pharmaceutically acceptable salt thereof acts by inhibition of the production or action of a metalloproteinase. 15 3. The use as claimed in Claim 2, wherein the compound of formula (I) or a pharmaceutically acceptable salt thereof acts by inhibition of the production or action of a matrix metalloproteinase. 20 4. The use as claimed in Claim 1, wherein the compound of formula (I) or a pharmaceutically acceptable salt thereof in a concentration of 50 pM or less is able to reduce significantly (p<0.05 in the appropriate statistical test) and by more than 50% compared to the appropriate vehicle 25 treated control, one or more of the following activities: the osteoclast invasion in the collagen invasion assay or in the bone lining cell invasion assay, the osteoclastic pericellular collagenolysis or the distance of migration in WO 01/25264 PCT/EPOO/09173 108 the pericellular collagenolysis assay, the TGF-P induced increase in accessible surface area of a culture of bone lining cells, or the osteoclastic bone resorption induced by treatment with TGF-P of a bone lining cell layer seeded on a 5 bone substratum, or the TGF-P induced decalcification of cultured foetal mouse calvariae in the bone lining cell elongation assay, the decalcification or the number of invading osteoclast in a culture of foetal mouse metatarsals in the metatarsal assay, the removal in calvarial cultures 10 of demineralised collagen fibres by osteoclasts in the subosteoclastic resorption zone or by bone lining cells in the resorption pits left by the osteoclasts in the demineralised collagenolysis assay, or the release of 45 Ca in the calvarial decalcification assay, or in a daily dose 15 of 100 mg/kg or less is able to reduce significantly (p<0.05 in the appropriate statistical test) and by more than 20% compared to the appropriate vehicle treated control, one or more of the following activities: the release of 3 H in the tetracycline-labelled bone resorption assay, or the number, 20 area or mortality rate in the bone metastasis assay. 5. The use as claimed in Claim 1, wherein the compound of formula (I) has a Ki-value of 100 nM or less with one or more of the MMPs, MMP-2, MMP-9, MMP-12, MMP-13, MMP-14 or 25 MMP-20, and a Ki-value at least 100 times higher than the lowest observed Ki-value for an MMP with two or more of the MMPs, MMP-1, MMP-2, MMP-3, MMP-7, MMP-8, MMP-9, MMP-10, MMP 11, MMP-12, MMP-13, MMP-14 or MMP-20. WO 01/25264 PCT/EPOO/09173 109 6. A compound of the formula (II): R5 R2 R3 Xa Xb Bu R R9 H 0 0 5 or a pharmaceutically acceptable salt thereof, wherein: R , R2, R ,R , R , and Xa and X are as defined in claim 1. 10 7. A compound as claimed in Claim 6, or a pharmaceutically acceptable salt thereof for use in a method of medical treatment by therapy or prophylaxis.
6. 8. The use of a compound- of claim 6 or a pharmaceutically 15 acceptable salt thereof in the manufacture of a medicament for the treatment of metabolic bone disease.
7. 9. A compound as claimed in Claim 6 or Claim 7 or the use thereof as claimed in Claim 8, wherein the compound of 20 formula (II) or a pharmaceutically acceptable salt thereof acts by inhibition of the production or action of a metalloproteinase. WO 01/25264 PCT/EPOO/09173 110
8. 10. A compound as claimed in Claim 6 or Claim 7 or the use thereof as claimed in Claim 8, wherein in the compound of formula (II) or a pharmaceutically acceptable salt thereof acts by inhibition of the production or action of a matrix 5 metalloproteinase.
9. 11. A compound as claimed in Claim 6 or Claim 7 or the use thereof as claimed in Claim 8, wherein the compound of formula (II) or a pharmaceutically acceptable salt thereof in a 10 concentration of 50 p.M or less is able to reduce significantly (p<0.05 in the appropriate statistical test) and by more than 50% compared to the appropriate vehicle treated control, one or more of the following activities: the osteoclast invasion in the collagen invasion assay or in the bone lining cell 15 invasion assay, the osteoclastic pericellular collagenolysis or the distance of migration in the pericellular collagenolysis assay, the TGF-3 induced increase in accessible surface area of a culture of bone lining cells, or the osteoclastic bone resorption induced by treatment with TGF 20 of a bone lining cell layer seeded on a bone substratum, or the TGF-P induced decalcification of cultured foetal mouse calvariae in the bone lining cell elongation assay, the decalcification or the number of invading osteoclast in a culture of foetal mouse metatarsals in the metatarsal assay, 25 the removal in calvarial cultures of demineralised collagen fibres by osteoclasts in the subosteoclastic resorption zone or by bone lining cells in the resorption pits left by the osteoclasts in the demineralised collagenolysis assay, or the WO 01/25264 PCT/EPOO/09173 111 release of 45 Ca in the calvarial decalcification assay, or in a daily dose of 100 mg/kg or less is able to reduce significantly (p<0.05 in the appropriate statistical test) and by more than 20% compared to the appropriate vehicle treated 5 control, one or more of the following activities: the release of 3 H in the tetracycline-labelled bone resorption assay, or the number, area or mortality rate in the bone metastasis assay. 10 12. A compound as claimed in Claim 6 or Claim 7 or the use thereof as claimed in Claim 8, wherein the compound of formula (II) has a Ki-value of 100 nM or less with one or more of the MMPs, MMP-2, MMP-9, MMP-12, MMP-13, MMP-14 or MMP-20, and a Ki-value at least 100 times higher than the lowest 15 observed Ki-value for an MMP with two or more of the MMPs, MMP-1, MMP-2, MMP-3, MMP-7, MMP-8, MMP-9, MMP-10, MMP-11, MMP 12, MMP-13, MMP-14 or MMP-20.
10. 13. A compound having or comprising a sequence described in 20 Table 10 or a pharmaceutically acceptable salt thereof, or a sequence related thereto by the substitution of one or more amino acids without decreasing the effectiveness or without decreasing the selectivity of the compound as a metalloproteinase inhibitor, or a compound acting as a 25 molecular mimic of any such compound in interacting with a metalloproteinase: WO 01/25264 PCT/EPOO/09173 112 Table 10: Selected phosphinate based peptide derivatives 'hosphinate based peptide PS P4 P3 P2 G'L P2'23'P4'P5' derivative 01-01 V A Y K S R G 01-02 A Y K S G 01-03 V Y R S G 01-04 G L A W L P G 01-05 L A W L G 01-06 L A Q L G 01-07 A G P L Y A R G 02-01 L M Y Y A P G 02-02 I M Y Y A P G 02-03 K M Y Y A P G 02-04 Z M Y Y A P G 02-21 Q M Y Y A P G 02-05 L L Y Y A P G 02-06 L I Y Y A P G 02-07 L P Y Y A P G 02-08 L M R Y A P G 02-09 L M F Y A P G 02-10 L M Y L A P G 02-11 L M Y I A P G 02-12 L M Y J A P G 02-13 L M Y Y L P G 02-14 L M Y Y I P G 02-15 L M Y Y J P G 02-16 L M Y Y M P G 02-17 L M Y Y Y P G 02-22 L M Y Y W P G 02-18 L M Y Y A L G 02-19 L M Y Y A I G 02-20 L M Y Y A J G 04-01 Y Y 04-19 Y Y W 04-20 Y Y M 04-21 M Y Y 04-18 I L F L M I G 04-03 T L Y L D G 04-04 V L Y T L S G WO 01/25264 PCT/EPOO/09173 113 04-05 I M Y V K F G 04-07 T L Y R A I G 04-09 T L R L F F G 04-10 I L R M A P G 04-11 S L F R D I G 04-12 L M F Y L S G 04-13 I M Y Y M T G 04-14 K F Y L Y A G 04-15 Y I Y T M P G 04-16 S M A Y H G 04-17 I M R L S E G 04-02 I L L N L I G 04-06 L I E R K G 04-08 E F Y K Y N G 05-01 T A S M F G 05-02 M Y T Y K L G 05-03 T R K S E L G 05-04 T R Q S E L G 05-05 S M L Y A G 05-06 L A A Y F Y G 05-07 E S N Y Y G 05-08 J V A S T G G 05-09 J Y M L Q L G 05-10 J Y M L K L G 05-11 V F K M A K G 05-12 N R A F Q A G 05-13 R V S N Y G G 05-14 G J K Y N R G 05-15 G J F E S L G 05-16 V S H A T F G 05-17 Y P E S A S G 05-18 J M V L Q F G 05-19 H F K Q G F G 05-20 Q P H F Y D G 05-21 S J D G V E G 05-22 R J D T L J G 05-23 R P P L L G 05-24 K Y F G P M G 05-25 G M G P F L G 05-26 T N P N V E G 05-27 G T V A K Q G 05-28 J L L F J G 05-29 K T M V Q L G WO 01/25264 PCT/EPOO/09173 114 05-30 K T M V K L G 05-31 Y M R H S G 05-32 N V V Y L E G 05-33 D A H D F G 05-34 T P L E A D G 05-35 A P A L A Q G 05-36 R P A Q M R G 05-37 Y A Y K Y E G 05-38 Y A Y Q Y E G 05-39 T J E V A G 05-40 V A K Q R G GPcL is an abbreviation for -Glyy(PO 2 -CH 2 )Leu-.
11. 14. A compound as claimed in Claim 13, or a pharmaceutically 5 acceptable salt thereof for use in a method of medical treatment by therapy or by prophylaxis.
12. 15. The use of a compound of Claim 13 or a pharmaceutically acceptable salt thereof in the manufacture of a medicament for 10 the treatment of metabolic bone disease.
13. 16. A compound as claimed in Claim 13 or Claim 14 or the use as claimed in Claim 15, wherein the compound of Claim 13 or a pharmaceutically acceptable salt thereof acts by inhibition 15 of the production or action of a metalloproteinase.
14. 17. A compound as claimed in Claim 13 or Claim 14 or the use as claimed in Claim 15, wherein the compound of Claim 13 or a pharmaceutically acceptable salt thereof acts by inhibition 20 of the production or action of a matrix metalloproteinase. WO 01/25264 PCT/EPOO/09173 115
15. 18. A compound as claimed in Claim 13 or Claim 14 or the use as claimed in Claim 15, wherein the compound of Claim 13 or a pharmaceutically acceptable salt thereof in a concentration of 50 pM or less is able to reduce significantly (p<0.05 in the 5 appropriate statistical test) and by more than 50% compared to the appropriate vehicle treated control, one or more of the following activities: the osteoclast invasion in the collagen invasion assay or in the bone lining cell invasion assay, the osteoclastic pericellular collagenolysis or the distance of 10 migration in the pericellular collagenolysis assay, the TGF-P induced increase in accessible surface area of a culture of bone lining cells, or the osteoclastic bone resorption induced by treatment with TGF-P of a bone lining cell layer seeded on a bone substratum, or the TGF-P induced decalcification of 15 cultured foetal mouse calvariae in the bone lining cell elongation assay, the decalcification or the number of invading osteoclast in a culture of foetal mouse metatarsals in the metatarsal assay, the removal in calvarial cultures of demineralised collagen fibres by osteoclasts in the 20 subosteoclastic resorption zone or by bone lining cells in the resorption pits left by the osteoclasts in the demineralised collagenolysis assay, or the release of 45Ca in the calvarial decalcification assay, or in a daily dose of 100 mg/kg or less is able to reduce significantly (p<0. 0 5 in the appropriate 25 statistical test) and by more than 20% compared to the appropriate vehicle treated control, one or more of the following activities: the release of 3 H in the tetracycline- WO 01/25264 PCT/EPOO/09173 116 labelled bone resorption assay, and the number, area or mortality rate in the bone metastasis assay.
16. 19. A compound as claimed in Claim 13 or Claim 14 or the use 5 as claimed in Claim 15, wherein the compound of claim 13 has a Ki-value of 100 nM or less with one or more of the MMPs, MMP 2, MMP-9, MMP-12, MMP-13, MMP-14 or MMP-20, and a Ki-value at least 100 times higher than the lowest observed Ki-value for an MMP with two or more of the MMPs, MMP-1, MMP-2, MMP-3, MMP 10 7, MMP-8, MMP-9, MMP-10, MMP-11, MMP-12, MMP-13, MMP-14 or MMP-20.
17. 20. An anti-bone resorption agent comprising a compound of formula (I) operatively linked to a ligand targeting the 15 compound of formula (I) to a proteinase involved in bone resorption or to the environment of the proteinase or incorporating such a ligand in or as one of the defined groups R', R 2 , R 3 , R 5 or R 9 of the formula. 20 21. An anti-bone resorption agent comprising a compound of Claim 6 operatively linked to a ligand targeting the compound of Claim 6 to a proteinase involved in bone resorption or to the environment of the proteinase or incorporating such a ligand in or as one of the defined groups R', R 2 , R 3 , R 5 or R 9 25 of the formula.
18. 22. An anti-bone resorption agent comprising the compound of claim 13 operatively linked to a ligand targeting the compound WO 01/25264 PCT/EPOO/09173 117 of claim 13 to a proteinase involved in bone resorption or to the environment of the proteinase.
19. 23. A method for the synthesis of a compound of formula I 5 which comprises carrying out a reaction to form a phosphorus to carbon bond between the phosphorus atom in a compound in which phosphorus bears at least the substituents Xa, -Xb -R 5 , and -CH(R 4 )NH and a carbon atom in a compound in which said carbon atom 10 bears the substituents R 6 , R 7 and -C (R') -C (O) R'.
20. 24. A method as claimed in Claim 23, wherein the said compound containing the carbon atom is attached to a solid support. 15 25. A method as claimed in Claim 24, comprising attaching said compound containing the carbon atom to said solid support by acylation by said compound of a primary amino group attached to said support. 20 26. A method as claimed in Claim 25, further comprising solid phase peptide synthesis on said support to provide said primary amino group.
21. 27. A method as claimed in any one of Claims 24 to 26, further 25 comprising solid phase peptide synthesis from said -NH- group attached to said -CH(R 4 )NH- substituent.