Title: The Collagenolytic Activity of Cathepsin K Is Unique among Mammalian Proteinases
Abstract: Type I collagen fibers account for 90% of the organic matrix of bone. The degradation of this collagen is a major event during bone resorption, but its mechanism is unknown. A series of data obtained in biological models strongly suggests that the recently discovered cysteine proteinase cathepsin K plays a key role in bone resorption. Little is known, however, about the actual action of cathepsin K on type I collagen. Here, we show that the activity of cathepsin K alone is sufficient to dissolve completely insoluble collagen of adult human cortical bone. We found that the collagenolytic activity of cathepsin K is directed both outside the helical region of the molecule, i.e. the typical activity of cysteine proteinases, and at various sites inside the helical region, hitherto believed to resist all mammalian proteinases but the collagenases of the matrix metalloproteinase family and the neutrophil elastase. This property of cathepsin K is unique among mammalian proteinases and is reminiscent of bacterial collagenases. It is likely to be responsible for the key role of cathepsin K in bone resorption. Type I collagen fibers account for 90% of the organic matrix of bone. The degradation of this collagen is a major event during bone resorption, but its mechanism is unknown. A series of data obtained in biological models strongly suggests that the recently discovered cysteine proteinase cathepsin K plays a key role in bone resorption. Little is known, however, about the actual action of cathepsin K on type I collagen. Here, we show that the activity of cathepsin K alone is sufficient to dissolve completely insoluble collagen of adult human cortical bone. We found that the collagenolytic activity of cathepsin K is directed both outside the helical region of the molecule, i.e. the typical activity of cysteine proteinases, and at various sites inside the helical region, hitherto believed to resist all mammalian proteinases but the collagenases of the matrix metalloproteinase family and the neutrophil elastase. This property of cathepsin K is unique among mammalian proteinases and is reminiscent of bacterial collagenases. It is likely to be responsible for the key role of cathepsin K in bone resorption. matrix metalloproteinase polyacrylamide gel electrophoresis l-3-carboxy-trans-2,3-epoxypropionyl-leucyl-amido-(4-guanidino) butane type I collagen C-telopeptide breakdown product type I collagen N-telopeptide breakdown product enzyme-linked immunosorbent assay. The only mammalian proteinases that have been shown to attack the native triple helical region of type I collagen are the collagenases of the MMP1 family (1Birkedahl-Hansen H. Moore W.G.I. Bodden M.K. Windsor L.J. Birkedahl-Hansen B. DeCarlo A. Engler J.A. Crit. Rev. Oral Biol. Med. 1993; 4: 197-250Crossref PubMed Scopus (2691) Google Scholar, 2Aimes R.T. Quigley J.P. J. Biol. Chem. 1995; 270: 5872-5876Abstract Full Text Full Text PDF PubMed Scopus (838) Google Scholar, 3Ohuchi E. Imai K. Fujii Y. Sata H. Seiki M. Okada Y. J. Biol. Chem. 1997; 272: 2446-2451Abstract Full Text Full Text PDF PubMed Scopus (841) Google Scholar) and the neutrophil serine elastase (4Kafienah W. Buttle D.J. Burnett D. Hollander A.P. Biochem. J. 1998; 330: 897-902Crossref PubMed Scopus (93) Google Scholar) They cleave the type I collagen triple helix across all three chains (i.e. two α1 chains and one α2 chain) only at a specific point three-quarters of the way to the N-terminal end of the collagen molecule. Proteinases with broad specificity, such as cysteine proteinases, attack only the extrahelical regions that are located at either end of native collagen (telopeptides) and that represent only 4% of the molecule (5Murphy G. Reynolds J.J. Royce P.M. Steinmann B. Connective Tissue and Its Heritable Disorders. Wiley-Liss, Inc, New York1993: 287-316Google Scholar). Because the telopeptides are involved in intra- and intermolecular links, this attack may separate individual molecules. The latter proteinases may also attack destabilized triple helices, acting thereby as gelatinases. At 37 °C, such a destabilization may transiently affect a small proportion of collagen molecules, because the melting temperature of soluble collagen is only a few degrees higher. It has also been emphasized that when collagen molecules are cross-linked and arranged in insoluble fibers they become more resistant to proteolysis (6Eeckout Y. Delaissé J.-M. Ledent P. Vaes G. Glauert A.M. The Control of Tissue Damage. Elsevier Science, Amsterdam1988: 237-313Google Scholar). However, the co-operation of proteinases with distinct specificities toward the chemical bonds of collagen fibers has been shown to favor the efficiency of collagenolysis (7Danielsen C.C. Biochem. J. 1990; 272: 697-701Crossref PubMed Scopus (44) Google Scholar). Insoluble type I collagen fibers constitute 90% of the organic matrix of bone, and their degradation is necessary for bone resorption (8Delaissé J.-M. Vaes G. Rifkin B.R. Gay C.V. Biology and Physiology of the Osteoclast. CRC Press, Boca Raton, FL1992: 289-314Google Scholar). The test tube experiments that have been performed so far showed that it is difficult to achieve complete degradation of adult lamellar bone with a single bone proteinase (9Etherington D.J Connect. Tissue Res. 1977; 5: 135-145Crossref PubMed Scopus (44) Google Scholar, 10Etherington D.J. Birkedahl-Hansen H. Collagen Relat. Res. 1987; 7: 185-199Crossref PubMed Scopus (22) Google Scholar). On the other hand, various biological approaches have shown that both MMPs and cysteine proteinases participate in the bone resorption processes (8Delaissé J.-M. Vaes G. Rifkin B.R. Gay C.V. Biology and Physiology of the Osteoclast. CRC Press, Boca Raton, FL1992: 289-314Google Scholar, 11Delaissé J.-M. Eeckhout Y. Vaes G. Biochem. J. 1980; 192: 365-368Crossref PubMed Scopus (113) Google Scholar, 12Delaissé J.-M. Eeckhout Y. Sear C. Galloway A. McCullagh K. Vaes G. Biochem. Biophys. Res. Commun. 1985; 130: 483-490Crossref Scopus (84) Google Scholar, 13Everts V. Delaissé J.-M. Korper W. Niehof A. Vaes G. Beertsen W. J. Cell. Physiol. 1992; 150: 221-231Crossref PubMed Scopus (258) Google Scholar). Representatives of these two types of proteinases were identified in osteoclasts, the cells responsible for bone resorption. MMP-9 is the best established MMP in osteoclasts and is highly expressed by these cells (14Tezuka K. Nemoto K. Tezuka Y. Sato T. Ikeda Y. Kobori M. Kamashima H. Eguchi H. Hakeda Y. Kumegawa M. J. Biol. Chem. 1994; 269: 15006-15009Abstract Full Text PDF PubMed Google Scholar). However, since the recent discovery of the cysteine proteinase cathepsin K (15Tezuka K. Tezuka Y. Maejima A. Sato T. Nemoto K. Kamioka H. Hakeda Y. Kumegawa M. J. Biol. Chem. 1994; 269: 1106-1109Abstract Full Text PDF PubMed Google Scholar), a number of biological data have suggested that cathepsin K plays an unique role among the various osteoclast proteinases. Its expression correlates extremely well with bone resorption (16Littlewood-Evans A. Kokubo T. Ishibashi O. Inaoka T. Wlodarski B. Gallagher J.A. Bilbe G Bone. 1997; 20: 81-86Crossref PubMed Scopus (179) Google Scholar, 17Saneshige S. Mano H. Tezuka K. Kakudo S. Mori Y. Honda Y. Itabashi A. Yamada T. Miyata K. Hakeda Y. Biochem. J. 1995; 309: 721-724Crossref PubMed Scopus (95) Google Scholar, 18Kahudo S. Mano H. Shiokawa M. Mori Y. Kumegawa M. Hakeda Y. Biochem. Biophys. Res. Commun. 1997; 234: 600-604Crossref PubMed Scopus (12) Google Scholar, 19Mano H. Yuasa T. Kameda T. Miyazawa K. Nakamaru Y. Shiokawa M. Mori Y. Yamada K. Miyata H. Shindo H. Azuma H. Hakeda Y. Kumegawa M. Biochem. Biophys. Res. Commun. 1996; 223: 637-642Crossref PubMed Scopus (90) Google Scholar), and more importantly, it was shown that bone resorption is impaired in situations of cathepsin K deficiency (20Inui T. Ishibashi O. Inaoka T. Origane Y. Kumegawa M. Kokubo T. Yamamura T. J. Biol. Chem. 1997; 272: 8109-8112Abstract Full Text Full Text PDF PubMed Scopus (150) Google Scholar,21Gelb B.D. Shi G.-P. Chapman H.A. Desnick R.J. Science. 1996; 273: 1236-1238Crossref PubMed Scopus (867) Google Scholar), thus demonstrating the effective participation of cathepsin K in bone resorption. Among these situations is pycnodysostosis, a human recessive disease in which a mutation in the cathepsin K gene leads to decreased osteoclast function. An obvious question raised by the key role of cathepsin K in bone resorption concerns its efficiency in degrading bone collagen. Cleavage of soluble type I collagen has been shown in test tube experiments (22Brömme D. Okamoto K. Wang B.B. Biroc S. J. Biol. Chem. 1996; 271: 2126-2132Abstract Full Text Full Text PDF PubMed Scopus (386) Google Scholar,23Bossard M.J. Tomaszek T.A. Thompson S.K. Amegadzie B.Y. Hanning C.R. Jones C. Kurdyla J.T. McNulty D.E. Drake F.H. Gowen M. Levy M.A. J. Biol. Chem. 1996; 271: 12517-12524Abstract Full Text Full Text PDF PubMed Scopus (450) Google Scholar), but it is not known to what extent cathepsin K is able to degrade insoluble collagen of adult bone without the collaboration of other proteinases. It has not been investigated either whether cleavage of collagen by cathepsin K occurs through the mechanism typical of other cysteine proteinases. To address these questions, we investigated the ability of cathepsin K to degrade adult cortical bone and compared it to MMP-9, the other proteinase found to be highly expressed in osteoclasts, and we analyzed the fragmentation pattern of collagen by cathepsin K by using SDS-PAGE and amino acid sequencing. Active recombinant human cathepsin K was prepared as described (23Bossard M.J. Tomaszek T.A. Thompson S.K. Amegadzie B.Y. Hanning C.R. Jones C. Kurdyla J.T. McNulty D.E. Drake F.H. Gowen M. Levy M.A. J. Biol. Chem. 1996; 271: 12517-12524Abstract Full Text Full Text PDF PubMed Scopus (450) Google Scholar). Cathepsin L purified from rat liver lysosomes (24Kirschke H. Kembhavi A.A. Bohley P. Barret A.J. Biochem. J. 1982; 201: 367-372Crossref PubMed Scopus (205) Google Scholar) was generously provided by Dr. H. Kirschke (Martin Luther University; Halle Wittenbay, Germany). Recombinant human pro-MMP-9 was a kind gift of Dr. W. Stetler-Stevenson (National Cancer Institute, Bethesda, MD) and Dr. R. Fridman (Wayne State University, Detroit, MI); it was activated with 1 mm p-aminophenylmercuric acetate (Sigma) for 24 h at 34 °C. Recombinant human pro-MMP-13 was a kind gift of Drs. V. Knaüper and G. Murphy (University of East Anglia, Norwich, United Kingdom); it was activated with 1 mm p-aminophenylmercuric acetate for 30 min at 37 °C. Activated recombinant human MMP-1 and pro-MMP-1 were a generous gift of Dr. H. Nagase (University of Kansas Medical Center, Kansas City, KS); the pro-MMP-1 was activated with 1 mm p-aminophenylmercuric acetate for 4 h at 37 °C. Crude bacterial collagenase (Sigma type I) andl-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin were purchased from Sigma. The molar concentrations of active cathepsin L and K were determined by active site titration (25Knight C. Methods Enzymol. 1995; 248: 85-101Crossref PubMed Scopus (75) Google Scholar) using E-64 (Sigma), and that of the MMPs was determined by using BB94 ((4-(N-hydroxyamino)-2R-isobutyl-3-S-(thienylthiomethyl)-succinyl)-l-phenylalanine-N-methylamide) (kind gift of Dr. H. Van Wart, Roche, Palo Alto, CA). Normal cortical bone of a 70-year-old woman was finely ground in liquid nitrogen. The bone powder was defeated with ice-cold acetone for 2 h, rinsed with ice-cold water, and then demineralized in 10% formic acid for 3 days at 4 °C, with daily renewal of formic acid. The powder was then extensively washed with ice-cold water and freeze-dried. Ninety-five % of this powder consisted of collagen according to hydroxyproline determinations and considering that 1 mg of collagen contains 124 μg of hydroxyproline. The demineralized bone particles were then dispersed at a concentration of 4 mg/ml in 100 mm sodium acetate buffer, pH 5.5, 20 mm l-cysteine, and 5 mm EDTA for cathepsin K, or in 50 mm Tris-HCl buffer, pH 7.5, and 10 mmCaCl2 for MMPs and bacterial collagenase. Enzymatic assays were performed by mixing 0.5-ml aliquots of the suspension of bone particles (i.e. 2 mg of collagen) and 0.5 ml of proteinase diluted in the corresponding buffer so as to obtain a molar ratio of active enzyme:collagen of 1:30 (taking into account the active site titrations of the proteinases and the amount of triple helix of collagen in each test tube). Incubations were then performed in triplicates for 24, 48, 72, and 96 h at 37 °C under gentle agitation. Because the proteinases proved to lose their activity after 24 h, the incubation mixtures were supplemented daily with the initial dose of fresh enzyme, which was delivered in a volume of 50 μl. Blank assays were also performed by incubating the bone powder in buffer without proteinase. After incubation, the mixtures were centrifuged for 20 min at 1500 × g at 4 °C. Hydroxyproline was determined by a spectrophotometric method in the supernatant and in the pellet, after hydrolysis in 6 m HCl for 22 h at 110 °C (26Firschein H.E. Shill J.P. Anal. Biochem. 1966; 14: 296-304Crossref PubMed Scopus (116) Google Scholar). The amount of solubilized collagen was calculated for each incubation mixture from the hydroxyproline determinations in the supernatant and the corresponding pellet. The supernatants were also assessed for type I collagen C-terminal (CTX) and N-terminal (NTX) telopeptide breakdown products using the Crosslaps™ (Osteometer Biotech, Herlev, Denmark) (27Bonde M. Qvist P. Fledelius C. Riis B.J. Christiansen C. Clin. Chem. 1994; 40: 2022-2025Crossref PubMed Scopus (316) Google Scholar) and Osteomark™ (Ostex International, Seattle, WA) (28Hanson D.A. Weis M.A.E. Bollen A.M. Maslan S.H. Singer F.R. Eyre D.R. J. Bone Miner. Res. 1992; 7: 1251-1258Crossref PubMed Scopus (633) Google Scholar) ELISAs, respectively. Native acid-soluble guinea pig (generous gift of Dr. Y. Eeckout; ICP-UCL, Brussels, Belgium) or bovine (kind gift of Dr. Huc, Colletica, Lyon, France) acid-extracted type I collagen was used at a final concentration of 0.8 mg/ml (i.e. 2.67 μm when considering a molecular mass of 300 kDa, corresponding to nonpolymerized triple helices) in 100 mm sodium acetate buffer, pH 5.5, 20 mm l-cysteine, and 5 mm EDTA for cathepsins L and K, or in 50 mmTris-HCl buffer, pH 7.5, and 10 mm CaCl2 for MMP-1 and trypsin. This collagen was incubated with proteinases at concentrations ranging from 0.08 to 2.14 μm, during various times, at 15 or 25 °C. After incubation, the reactions of cathepsins K and L were stopped with 430 μm E-64, and those of MMP-1 were stopped with 50 mm EDTA. Samples were then subjected to SDS-PAGE, and the gels were stained with Coomassie Blue. To relate migration distances to molecular weights, it was taken into account that the electrophoretic mobility of the α1 chain and its fragments differs from that of the α2 chain and its fragments (29French M.F. Mookhtiar K.A. Van Wart H.E. Biochemistry. 1987; 26: 681-687Crossref PubMed Scopus (98) Google Scholar). The type I collagen fragments generated by cathepsin K were separated by SDS-PAGE, electroblotted on a polyvinylidene difluoride membrane (Immobilon, Millipore, Bedford, MA). The type I collagen fragment bands were cut, and the N-terminal sequencing was performed on a 477A protein sequencer with an on-line 120A analyzer (Applied Biosystems) and chemicals recommended by the manufacturer. The first 10–14 residues were analyzed and compared with known sequences of human and bovine type I collagen (SWISS-PROT™ Protein Data Bank). We first investigated to what extent cathepsin K can solubilize the collagen of adult cortical bone. Fig. 1 shows that cathepsin K solubilized 40% of the collagen after 1 day of incubation, about 80% after 3 days, and almost all of it after 4 days. This solubilization by cathepsin K proved to be almost as extensive as that obtained with crude bacterial collagenase, but it was slower. Several MMPs were tested at the same molar concentration as cathepsin K. MMP-9, a gelatinase that is highly expressed in osteoclasts, achieved only half the solubilization of cathepsin K at each time point, and after 72 h, the solubilization reached only 38% of the total amount. Collagenases such as MMP-1 and MMP-13 degraded about 8% of the collagen after 1 day and about 18% after 4 days. We also analyzed whether some of the reaction products might be recognized by antibodies directed against the N and C telopeptide fragments of type I collagen and that proved to be specific and sensitive reagents to assess bone resorption in vivo (27Bonde M. Qvist P. Fledelius C. Riis B.J. Christiansen C. Clin. Chem. 1994; 40: 2022-2025Crossref PubMed Scopus (316) Google Scholar,28Hanson D.A. Weis M.A.E. Bollen A.M. Maslan S.H. Singer F.R. Eyre D.R. J. Bone Miner. Res. 1992; 7: 1251-1258Crossref PubMed Scopus (633) Google Scholar). MMP-9, MMP-1, and MMP-13 did not release fragments that are recognized by these antibodies (not shown). In contrast, crude bacterial collagenase (not shown) and cathepsin K (Table I) released NTX and CTX antigens with a similar time dependence.Table ICTX and NTX telopeptide breakdown products generated from adult human cortical bone collagen by cathepsin KDuration of incubationCTXNTXhng/mg of collagenpmol/mg of collagen24124 ± 21367 ± 5948304 ± 12718 ± 11072861 ± 922226 ± 143Collagen was solubilized exactly as explained in Fig. 1 and analyzed by ELISA for CTX and NTX. The units for CTX and NTX are those given by the ELISA. These units are dependent on the calibration of each assay and thus cannot be used to directly deduce a molar ratio of CTX and NTX generated per molecule of collagen. The results are shown as mean ± S.D. of three incubations. Open table in a new tab Collagen was solubilized exactly as explained in Fig. 1 and analyzed by ELISA for CTX and NTX. The units for CTX and NTX are those given by the ELISA. These units are dependent on the calibration of each assay and thus cannot be used to directly deduce a molar ratio of CTX and NTX generated per molecule of collagen. The results are shown as mean ± S.D. of three incubations. To investigate the mechanism of collagen fragmentation by cathepsin K, incubations of soluble collagen were performed at 25 °C, a temperature at which it is believed that the triple helix cannot be cleaved by any proteinase, except the collagenases of the MMP family (1Birkedahl-Hansen H. Moore W.G.I. Bodden M.K. Windsor L.J. Birkedahl-Hansen B. DeCarlo A. Engler J.A. Crit. Rev. Oral Biol. Med. 1993; 4: 197-250Crossref PubMed Scopus (2691) Google Scholar, 2Aimes R.T. Quigley J.P. J. Biol. Chem. 1995; 270: 5872-5876Abstract Full Text Full Text PDF PubMed Scopus (838) Google Scholar, 3Ohuchi E. Imai K. Fujii Y. Sata H. Seiki M. Okada Y. J. Biol. Chem. 1997; 272: 2446-2451Abstract Full Text Full Text PDF PubMed Scopus (841) Google Scholar) and neutrophil elastase (4Kafienah W. Buttle D.J. Burnett D. Hollander A.P. Biochem. J. 1998; 330: 897-902Crossref PubMed Scopus (93) Google Scholar). Accordingly, unspecific proteinases like trypsin (not shown) or the cysteine proteinase cathepsin L both used at 2.14 μm (Fig. 2,II) did not cleave the triple helix, whereas MMP-1 cleaved it into the characteristic 3/4N-terminal (TCA) and 1/4C-terminal (TCB) fragments (Fig. 2,I). Cathepsin L, however, generated monomeric α chains, at the expense of the dimeric (β) and trimeric (γ) chains, as is typical of cysteine proteinases and due to a cleavage at the level of the telopeptides. Upon incubation with cathepsin K at a concentration 30 times lower than that of cathepsin L, a similar electrophoretic pattern was obtained. An important difference, however, was that the “monomer-like” bands generated by cathepsin K migrated slightly faster, suggesting a possible cleavage in the helical region (Fig. 2,II) (22Brömme D. Okamoto K. Wang B.B. Biroc S. J. Biol. Chem. 1996; 271: 2126-2132Abstract Full Text Full Text PDF PubMed Scopus (386) Google Scholar). When using a 30 times higher concentration of cathepsin K (i.e. the same concentration as the one used for cathepsin L), more bands were generated, suggesting further cleavage of the triple helix (Fig. 2, V). When incubating for a longer time, no more bands appeared, but the lower molecular weight bands became stronger at the expense of the higher molecular weight bands. This collagen fragmentation was not due to a peculiarity of the guinea pig collagen, as similar results were obtained with bovine skin collagen (Fig. 2, VI). E-64, an inhibitor of cathepsin K, blocked all the above described effects of cathepsin K (Fig. 2,III). We also analyzed the activity of cathepsin K in a situation where collagen triple helices and their fragments exhibit still higher stability, and we therefore performed incubations at 15 °C. Fig. 2 shows that a 10-min incubation was sufficient to weaken the dimer and trimer bands and increase monomer-like bands (Fig. 2, III); 2 h more allowed the generation of shorter fragments (Fig. 2, IV), and a further 2 h did not result in changing significantly the migration pattern. In order to analyze the cleavage sites further, similar experiments were repeated, and N-terminal amino acid sequencing was performed (Table II). First, we analyzed the high molecular weight monomer-like bands generated from the dimers and trimers of guinea pig collagen after a 30-min incubation at 15 °C. We found evidence for a cleavage inside the triple helix exactly 10 amino acids away from the N-terminal ends of the helical regions of both α1 and α2 chains. This thus may explain why monomer-like bands generated by cathepsin K migrate slightly faster than those generated by cathepsin L. Among the cleavage sites generated upon longer incubation at 15 °C, one was clearly identified 811 and 815 amino acids away from the N-terminal ends of the helical domains of the α1 and α2 chains, respectively (i.e. about 35 amino acids from the cleavage site of interstitial collagenases). Other cleavage sites proved to be difficult to determine, probably due to the same problem as the one encountered for the analysis of the fragmentation of type I collagen by bacterial collagenases (30French M.F. Bhown A. Van Wart H.E. J. Protein Chem. 1992; 11: 83-97Crossref PubMed Scopus (64) Google Scholar) and ascribed to the heterotrimeric nature of type I collagen and to the regularity of the repetitions of similar amino acid sequences in the α chains. However, in an effort to further demonstrate cleavage sites in the triple helical area of stable collagen molecules, we also performed incubations under other conditions. When using cathepsin K at 25 °C and at a concentration 30 times lower than in the above experiments, we found evidence for cleavage sites in the α1 and α2 chains, which are 97 and 100 amino acids, respectively, away from the N-terminal end of their helical domains. When using bovine type I collagen, we also found evidence for a cleavage site in the triple helix exactly 22 amino acids away from the N-terminal ends of the helical region of both the α1 and α2 chains. These data clearly establish the ability of cathepsin K to cleave the triple helix of type I collagen at multiple sites.Table IIN-terminal amino acid sequencing of fragments generated by the action of cathepsin K on type I collagenIncubation conditionSequenceCollagen sourceTemperatureCathepsin K: collagenTimeSpeciesα1 Chainα2 Chain° Cmol/molhGuinea pig150.80.5 and 2Guinea pig2-aExperimentally determined.X9X-10GLPGPPGAPGAQGFX 9 X-10GPPGAVGAPGPQGFHuman2-bFrom data bank.P9R-10GLPGPPGAPGPQGFP9R-10GPPGAAGAPGPQGFGuinea pig250.0316Guinea pig2-aExperimentally determined.X96X-97GAKGDAGPAGX99X-100GQPGAPGVKGEPGAHuman2-bFrom data bank.L96D-97GAKGDAGPAGL99K-100GQPGAPGVKGEPGAGuinea pig15 and 250.82Guinea pig2-aExperimentally determined.X 810 X-811GASGERGPPGPAX 814G-815ARGPPGNVGXPGVHuman2-bFrom data bank.P810S-811GASGERGPPGPMP814G-815ARGPPGAVGSPGVBovine250.86Bovine2-aExperimentally determined.P21Q-22GFQGPPGEPGEPP21Q-22GFQGPPGEPGEPBovine2-bFrom data bank.P21Q-22GFQGPPGEPGEPP21Q-22GFQGPPGEPGEPHuman2-bFrom data bank.P21Q-22GFQGPPGEPGEPP21Q-22GFQGPAGEPGEPGuinea pig or bovine fragments were generated by incubation with cathepsin K. Their N-terminal amino-acid sequences were determined as explained under “Materials and Methods” and are shown starting at the amino acid following the hyphen (P1′ position of the presumed cleavage site). They were compared to the closest resembling sequence of human or bovine α1 or α2 chains. These sequences were consistent with the electrophoretic mobility of the fragments they were derived from. The amino acids at the P1 and P2 positions of the presumed cleavage site are indicated to the left of the hyphen. The numbering of the residues starts from the first triplet of the triple helical domain of type I collagen. X represents an undetermined amino acid. Note that same cleavage sites were found for distinct incubation times or temperatures.2-a Experimentally determined.2-b From data bank. Open table in a new tab Guinea pig or bovine fragments were generated by incubation with cathepsin K. Their N-terminal amino-acid sequences were determined as explained under “Materials and Methods” and are shown starting at the amino acid following the hyphen (P1′ position of the presumed cleavage site). They were compared to the closest resembling sequence of human or bovine α1 or α2 chains. These sequences were consistent with the electrophoretic mobility of the fragments they were derived from. The amino acids at the P1 and P2 positions of the presumed cleavage site are indicated to the left of the hyphen. The numbering of the residues starts from the first triplet of the triple helical domain of type I collagen. X represents an undetermined amino acid. Note that same cleavage sites were found for distinct incubation times or temperatures. Type I collagen is one of the most abundant molecules in mammals, and its degradation is an important physiological issue. So far, only two collagenolytic mechanisms have been described in mammals (Fig. 3). One is used by interstitial collagenases cleaving the stable triple helices of collagen at a single site between their amino acids 775 and 776. The other is used by proteinases with broad specificity, such as cysteine proteinases, cleaving the nonhelical telopeptides of native type I collagen and also degrading denatured collagen (i.e. gelatin). The present work shows that there is a third mechanism reminiscent of bacterial collagenases (29French M.F. Mookhtiar K.A. Van Wart H.E. Biochemistry. 1987; 26: 681-687Crossref PubMed Scopus (98) Google Scholar, 30French M.F. Bhown A. Van Wart H.E. J. Protein Chem. 1992; 11: 83-97Crossref PubMed Scopus (64) Google Scholar, 31Lecroisey A. Keil B. Biochem. J. 1979; 179: 33-58Crossref Scopus (36) Google Scholar), the peculiarity of which is cleavage of native triple helices at multiple sites, and that this mechanism is used by cathepsin K, a recently discovered cysteine proteinase (15Tezuka K. Tezuka Y. Maejima A. Sato T. Nemoto K. Kamioka H. Hakeda Y. Kumegawa M. J. Biol. Chem. 1994; 269: 1106-1109Abstract Full Text PDF PubMed Google Scholar) that is essential for osteoclastic bone resorption (20Inui T. Ishibashi O. Inaoka T. Origane Y. Kumegawa M. Kokubo T. Yamamura T. J. Biol. Chem. 1997; 272: 8109-8112Abstract Full Text Full Text PDF PubMed Scopus (150) Google Scholar, 21Gelb B.D. Shi G.-P. Chapman H.A. Desnick R.J. Science. 1996; 273: 1236-1238Crossref PubMed Scopus (867) Google Scholar). The evidence for this new collagenolytic mechanism is based on the analysis of collagen fragments generated by cathepsin K at 15 and 25 °C. The stability of the helical region of the collagen molecules under these assay conditions was obvious from the absence of cleavage of the triple helix by unspecific proteinases. These included cathepsin L, the cysteine proteinase hitherto considered to have the highest telopeptidase and gelatinase activity (32Maciewicz R.A. Etherington D.J. Kos J. Turk V. Collagen Relat. Res. 1987; 7: 295-304Crossref PubMed Scopus (60) Google Scholar, 33Maciewicz R.A. Etherington D.J. Biochem. J. 1988; 256: 433-440Crossref PubMed Scopus (139) Google Scholar). That the α chains were not unraveled at the time of the attack by cathepsin K was further demonstrated by the fact that cleavage sites, at least two of them, were found exactly at the same position in the α1 and α2 chains. The fragmentation by cathepsin K was not an artifact, but really results from cysteine proteinase activity during incubation because it was completely blocked by a cysteine proteinase inhibitor. The cleavage sites inside the triple helix were clearly demonstrated not only because the size of the bands detected by electrophoresis was compatible with the size of fragments arising from the helical region, but also because the amino acid sequences of eight fragments (i.e. four from α1 chains and four from α2 chains) matched the amino acid sequences of the helical region of α1 and α2 chains reported in data bases. Interestingly, none of the identified cleavage sites corresponded to that of interstitial collagenase (1Birkedahl-Hansen H. Moore W.G.I. Bodden M.K. Windsor L.J. Birkedahl-Hansen B. DeCarlo A. Engler J.A. Crit. Rev. Oral Biol. Med. 1993; 4: 197-250Crossref PubMed Scopus (2691) Google Scholar) or to those of bacterial collagenases (30French M.F. Bhown A. Van Wart H.E. J. Protein Chem. 1992; 11: 83-97Crossref PubMed Scopus (64) Google Scholar). Thus, it appears that the specificity of the cleavage of collagen is determined by the type of proteinase, rather than by the position of hyperactive sites in the collagen molecule. Finally, it is worth noting that cathepsin K also generated collagen fragments that are recognized by the Crosslaps ELISA. Because these fragments are recognized by the ELISA only if there is a cleavage between Arg22 and Tyr23 of the α1 C-terminal telopeptide (34Bonde M. Garnero P. Fledelius C. Qvist P. Delmas P.D. Christiansen C. J. Bone Miner. Res. 1997; 12: 1028-1034Crossref PubMed Scopus (124) Google Scholar), this demonstrates a cleavage site in the C-terminal telopeptide. Thus, cathepsin K has telopeptidase activity like other cysteine proteinases, as previously suggested (22Brömme D. Okamoto K. Wang B.B. Biroc S. J. Biol. Chem. 1996; 271: 2126-2132Abstract Full Text Full Text PDF PubMed Scopus (386) Google Scholar); it has gelatinase activity (22Brömme D. Okamoto K. Wang B.B. Biroc S. J. Biol. Chem. 1996; 271: 2126-2132Abstract Full Text Full Text PDF PubMed Scopus (386) Google Scholar); and as shown above, it shares with bacterial proteinases the ability to cleave collagen at multiple points. It is of interest to consider the consequences of these properties for proteolysis of collagen in the physiological situation at 37 °C. Interstitial collagenases will cleave stable triple helices of all accessible collagen molecules but only at a single site. Cysteine proteinases other than cathepsin K will cleave the telopeptides and may also attack the small proportion of soluble collagen molecules that have a transiently destabilized helix. In addition to this, cathepsin K will attack any collagen molecule at multiple sites irrespective of its conformation. What makes cathepsin K unique among mammalian proteinases is the fact that its collagenolytic activity does not depend on destabilization of the triple helix in contrast to that of the other cysteine proteinases and that it cleaves native molecules at more sites than does interstitial collagenase. Thus, cathepsin K alone can both depolymerize collagen fibers and cleave triple helices, whereas it was previously thought that distinct proteinases were required to perform these activities and that their cooperation was necessary for an efficient degradation of insoluble cross-linked type I collagen (6Eeckout Y. Delaissé J.-M. Ledent P. Vaes G. Glauert A.M. The Control of Tissue Damage. Elsevier Science, Amsterdam1988: 237-313Google Scholar, 7Danielsen C.C. Biochem. J. 1990; 272: 697-701Crossref PubMed Scopus (44) Google Scholar, 8Delaissé J.-M. Vaes G. Rifkin B.R. Gay C.V. Biology and Physiology of the Osteoclast. CRC Press, Boca Raton, FL1992: 289-314Google Scholar). These different models of collagenolysis are consistent both with the present test tube experiments showing that cathepsin K can degrade completely the insoluble collagen of adult cortical bone in the absence of other proteinases, and with earlier reports on poor solubilization of bone collagen by other proteinases, whether interstitial collagenases or gelatinases (6Eeckout Y. Delaissé J.-M. Ledent P. Vaes G. Glauert A.M. The Control of Tissue Damage. Elsevier Science, Amsterdam1988: 237-313Google Scholar, 7Danielsen C.C. Biochem. J. 1990; 272: 697-701Crossref PubMed Scopus (44) Google Scholar, 8Delaissé J.-M. Vaes G. Rifkin B.R. Gay C.V. Biology and Physiology of the Osteoclast. CRC Press, Boca Raton, FL1992: 289-314Google Scholar, 9Etherington D.J Connect. Tissue Res. 1977; 5: 135-145Crossref PubMed Scopus (44) Google Scholar, 10Etherington D.J. Birkedahl-Hansen H. Collagen Relat. Res. 1987; 7: 185-199Crossref PubMed Scopus (22) Google Scholar). The latter observations were confirmed in the present study with two distinct interstitial collagenases and a gelatinase used at the same concentrations as cathepsin K. The progress of collagen solubilization in these assays was faster at the beginning and slower at the end, despite the additions of fresh proteinase. This decrease in the rate of solubilization may be due to the fact that highly susceptible collagen was degraded first and the remaining more resistant cross-linked collagen later (9Etherington D.J Connect. Tissue Res. 1977; 5: 135-145Crossref PubMed Scopus (44) Google Scholar). Although it is not possible to know how close are these test tube experiments to the conditions under which cathepsin K acts in the resorption zone, it is of interest to draw the attention to two aspects. First, concerning the molar ratio of cathepsin K to collagen of 0.03 used in our assays, it should be mentioned thatin vivo, collagen degradation in the resorption zone occurs as soon as the collagen is denuded by the demineralization processes (35Vaes G. Clin. Orthop. Relat. Res. 1988; 231: 239-271PubMed Google Scholar). The amount of demineralized collagen in the resorption zone is therefore quite low at any time, and this is a situation that should favor a relatively high concentration of cathepsin K as compared with that of collagen. Second, it is of interest that our assays were performed at pH 5.5, the pH for optimal activity of cathepsin K (22Brömme D. Okamoto K. Wang B.B. Biroc S. J. Biol. Chem. 1996; 271: 2126-2132Abstract Full Text Full Text PDF PubMed Scopus (386) Google Scholar), and that this pH is within the range of pH values of 4.7–6.8, measured in the resorption zone of osteoclasts (36Silver A. Murrills R.J. Etherington D.J Exp. Cell Res. 1988; 175: 266-276Crossref PubMed Scopus (761) Google Scholar). Cathepsin K alone can thus solubilize completely the major constituent of the bone matrix in conditions compatible with those of the resorption zone. MMP-9 is another proteinase that is highly expressed in osteoclasts (14Tezuka K. Nemoto K. Tezuka Y. Sato T. Ikeda Y. Kobori M. Kamashima H. Eguchi H. Hakeda Y. Kumegawa M. J. Biol. Chem. 1994; 269: 15006-15009Abstract Full Text PDF PubMed Google Scholar), and it has been speculated that it might be one of the MMPs (37Okada Y. Naka K. Kawamura K. Matsumoto T. Nakanishi I. Fujimoto N. Sato H. Seiki M. Lab. Invest. 1995; 72: 311-322PubMed Google Scholar) playing a role in collagen degradation in the resorption zone (13Everts V. Delaissé J.-M. Korper W. Niehof A. Vaes G. Beertsen W. J. Cell. Physiol. 1992; 150: 221-231Crossref PubMed Scopus (258) Google Scholar). It is of interest that in our test tube experiments, the level of collagen degradation by MMP-9 reached 50% of that obtained by cathepsin K when tested at pH 7.5, the optimal pH for MMP-9 activity. Taking into account the fact that at pH 6, MMP-9 still shows 90% of its activity at pH 7.5 (37Okada Y. Naka K. Kawamura K. Matsumoto T. Nakanishi I. Fujimoto N. Sato H. Seiki M. Lab. Invest. 1995; 72: 311-322PubMed Google Scholar), we thus have to consider that MMP-9 may contribute to some extent to the collagenolytic activity of the osteoclast. However, according to our data, the efficiency of cathepsin K is at least twice as important. In conclusion, this work demonstrates that cathepsin K degrades type I collagen by using a mechanism that has never been reported for mammalian proteinases. This unique collagenolytic mechanism allows interpretation of the key role of cathepsin K in bone resorption (16Littlewood-Evans A. Kokubo T. Ishibashi O. Inaoka T. Wlodarski B. Gallagher J.A. Bilbe G Bone. 1997; 20: 81-86Crossref PubMed Scopus (179) Google Scholar, 17Saneshige S. Mano H. Tezuka K. Kakudo S. Mori Y. Honda Y. Itabashi A. Yamada T. Miyata K. Hakeda Y. Biochem. J. 1995; 309: 721-724Crossref PubMed Scopus (95) Google Scholar, 18Kahudo S. Mano H. Shiokawa M. Mori Y. Kumegawa M. Hakeda Y. Biochem. Biophys. Res. Commun. 1997; 234: 600-604Crossref PubMed Scopus (12) Google Scholar, 19Mano H. Yuasa T. Kameda T. Miyazawa K. Nakamaru Y. Shiokawa M. Mori Y. Yamada K. Miyata H. Shindo H. Azuma H. Hakeda Y. Kumegawa M. Biochem. Biophys. Res. Commun. 1996; 223: 637-642Crossref PubMed Scopus (90) Google Scholar, 20Inui T. Ishibashi O. Inaoka T. Origane Y. Kumegawa M. Kokubo T. Yamamura T. J. Biol. Chem. 1997; 272: 8109-8112Abstract Full Text Full Text PDF PubMed Scopus (150) Google Scholar, 21Gelb B.D. Shi G.-P. Chapman H.A. Desnick R.J. Science. 1996; 273: 1236-1238Crossref PubMed Scopus (867) Google Scholar) at the molecular level. We are grateful to Anne-Cecilie Pedersen for expert technical assistance.