Human Cathepsin O2, a Matrix Protein-degrading Cysteine Protease Expressed in Osteoclasts FUNCTIONAL EXPRESSION OF HUMAN CATHEPSIN O2 IN SPODOPTERA FRUGIPERDA AND CHARACTERIZATION OF THE ENZYME*

Cathepsin O2, a human cysteine protease predomi- nantly present in osteoclasts, has been functionally expressed in Spodoptera frugiperda Sf9 cells using the Autographa californica nuclear polyhedrosis virus. Following in vitro activation at pH 4.0 with pepsin, active enzyme with an apparent molecular weight of 29,000 was obtained. N-terminal sequencing revealed the typi- cal processing site for cysteine proteases of the papain family with a proline in the position adjacent to the N-terminal alanine residue. The S 2 P 2 subsite specificity of human cathepsin O2 is similar to cathepsin S but distinguished from cathepsins L and B. Similar to cathepsin S, cathepsin O2 is characterized by a bell- shaped pH activity profile and is stable at pH 6.5 for 30 min at 37 °C. Cathepsin O2 is further distinguished by its potent collagenolytic activity against Type I collagen between pH 5 and 6, and elastinolytic activity against insoluble elastin at pH 7.0. Its capacity to efficiently degrade Type I collagen and its high expression in osteoclasts suggest that cathepsin O2 may play a major role in human osteoclastic bone resorption. Bone tissue is constantly undergoing a process of formation and resorption. Osteoclasts are cells responsible for the bone resorbing process. Bone resorption includes demineralization and degradation of extracellular matrix proteins (Delaisse´ and Vaes, 1992). Type I collagen constitutes 95% of the organic -end of the preregion of human cathepsin O2 and a Bam HI site at the 3 (cid:57) -end of the mature region of the enzyme were placed. The amplification reaction was carried out with the Pfu DNA polymerase (Stratagene, La Jolla, CA). The 1.1-kb frag- ment was inserted into the Bgl II and Bam HI site of the pVL1392 transfer vector (PharMingen, San Diego, CA). Recombinant baculovirus into by with addition porcine (Sigma) of in a for 90 min at 40 °C at The was moni- tored using Z-FR-MCA (10 (cid:109) M ) as a fluorogenic substrate measured in 100 m M Tris-HCl buffer, pH 7.5. The lysate was adjusted to pH 7.0 with 2 M Tris base, clarified by centrifugation at 10,000 (cid:51) g , the supernatant was adjusted to 2.5 M ammonium sulfate at pH 5.5. After centrifugation at 16,000 (cid:51) g , the cleared supernatant was concentrated to 50 ml by ultrafiltration (YM10 Amicon). After additional centrifugation at 10,000

Bone tissue is constantly undergoing a process of formation and resorption. Osteoclasts are cells responsible for the bone resorbing process. Bone resorption includes demineralization and degradation of extracellular matrix proteins (Delaissé and Vaes, 1992). Type I collagen constitutes 95% of the organic matrix (Krane and Simon, 1994). In addition to the interstitial collagenase, the lysosomal cysteine proteases cathepsins B and L are thought to be involved in osteoclastic bone resorption (Delaissé and Vaes, 1992). Both enzymes are present in the lysosomes as well as in the acidified extracellular resorption lacuna of the osteoclast (Goto et al., 1993), and both proteases display the in vitro ability to degrade collagen Type I at acidic pH (Maciewicz et al., 1987;Delaissé et al., 1991). Cysteine protease inhibitors, such as E-64 and leupeptin, have been shown to prevent osteoclastic bone resorption (Delaisse et al., 1987;Everts et al., 1988). Cathepsin L is considered to be one of the main proteases involved in collagen degradation in bone (Maciewicz and Etherington 1988;Kakegawa et al., 1993). It is thought that the collagenolytic action of cysteine proteases is exerted preferentially in the most acidic part of the bone resorption lacuna close to the ruffled border at a pH around 3.5 or 4.5, whereas the zinc-containing collagenases are more active in the neutral environment at the interface between the demineralized and mineralized matrix (Delaisse and Vaes, 1992). Besides cathepsins L and B, a variety of cathepsin L-and B-like activities may participate in collagenolytic bone degradation. Page et al. (1992) isolated multiple forms of cathepsin B from osteoclastomas. These have an acidic pH optimum and the ability to degrade soluble and insoluble Type I collagen. Delaisse et al. (1991) identified a 70-kDa thiol-dependent protease in bone tissue, which is also capable of degrading Type I collagen.
More recently, a cDNA encoding a novel human cysteine protease was cloned independently by several groups (Shi et al., 1995;Inaoka et al., 1995;Brömme and Okamoto, 1995) and named cathepsin O, cathepsin K, and cathepsin O2. All three sequences are identical and represent the human equivalent of the OC2 gene cloned from rabbit osteoclasts (Tezuka et al., 1994). This novel human cathepsin is highly expressed in osteoclasts and displays an approximate 10-fold higher expression at the message level when compared with cathepsin L (Brömme and Okamoto, 1995). Saneshige et al. (1995) could show that cathepsin O2(K) is regulated at the transcriptional level by retinoic acid in osteoclasts. Retinoic acid is a potent regulator and enhances the proliferation and differentiation of many types of cells including osteoclasts. Shi et al. (1995) demonstrated an endopeptidase activity of this enzyme against fibrinogen at acidic pH when overexpressed in COS-7 cells. However, no data have been published describing the enzymatic properties of this protease and demonstrating its activity toward extracellular matrix proteins such as collagen and elastin. In this publication, we describe for the first time the functional expression of recombinant cathepsin O2, its purification, enzymology, and inhibitor profile as well as its ability to efficiently degrade matrix proteins.

EXPERIMENTAL PROCEDURES
Materials-Z-LR-MCA, 1 Z-VR-MCA, and Z-RR-MCA were obtained from Aminotech (Canada). Z-FR-MCA, Z-VVR-MCA, and Z-LLR-MCA were synthesized as described elsewhere (Brömme et al., 1989. Type I collagen was purchased from U. S. Biochemical Corp. Z-FF-CHN 2 and Z-FA-CHN 2 were obtained from Bachem Bioscience, Inc., Switzerland. Construction of Transfer Vector and Expression-Using the polymerase chain reaction, a BglII site at the 5Ј-end of the preregion of human cathepsin O2 and a BamHI site at the 3Ј-end of the mature region of the enzyme were placed. The amplification reaction was carried out with the Pfu DNA polymerase (Stratagene, La Jolla, CA). The 1.1-kb fragment was inserted into the BglII and BamHI site of the pVL1392 transfer vector (PharMingen, San Diego, CA). Recombinant baculovirus * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: Khepri Pharmaceuticals, Inc., 260 Littlefield Ave., South San Francisco, CA 94080. Tel.: 415-794-3511; Fax: 415-794-3599. was generated by homologous recombination following cotransfection of the baculovirus transfer vector and linearized AcNPV genomic DNA into Sf9 cells (PharMingen, San Diego, CA). Pure virus (AcNPVCO 2 ) was obtained by plaque purification. Sf9 cells were grown in 1 liter of Sf900II media (Life Technologies, Inc.) to a density of 2 ϫ 10 6 cells/ml and infected at a multiplicity of infection of 1. Total cell number and cellular as well as secreted activity of cathepsin O2 were monitored every 24 h. After 3.5 days, the cells were harvested at a cell density of 2.5 ϫ 10 6 /ml. Activation, Purification, and N-terminal Sequencing-The Sf9 cells were harvested from the production media by centrifugation at 2,000 ϫ g and were lysed in a Dounce homogenizer. The cell lysate containing inactive cathepsin O2 precursor was brought up to 100 ml with 100 mM sodium acetate buffer, pH 3.75, containing 0.5% Triton X-100, 5 mM dithiothreitol, and 2.5 mM Na 2 EDTA, and the pH was adjusted to 4.0. The conversion of proform into active enzyme was accomplished by treatment with pepsin. After addition of porcine pepsin (Sigma) at a final concentration of 0.4 mg/ml, the activation mixture was incubated in a shaker for 90 min at 40°C at 200 rpm. The activation was monitored using Z-FR-MCA (10 M) as a fluorogenic substrate measured in 100 mM Tris-HCl buffer, pH 7.5. The activated lysate was adjusted to pH 7.0 with 2 M Tris base, clarified by centrifugation at 10,000 ϫ g, and the supernatant was adjusted to 2.5 M ammonium sulfate at pH 5.5. After centrifugation at 16,000 ϫ g, the cleared supernatant was concentrated to 50 ml by ultrafiltration (YM10 Amicon). After additional centrifugation at 10,000 ϫ g, the cleared supernatant was loaded onto a butyl-Sepharose 4 Fast Flow column (Pharmacia Biotech Inc.), which was washed with an ammonium sulfate gradient (2.5-0 M) in 25 mM acetate buffer, pH 5.5. The activity was eluted at 0 M ammonium sulfate. The pooled and concentrated fractions were than applied to an fast protein liquid chromatography Mono S column (Pharmacia) and eluted with a linear NaCl gradient (0 -2 M) in 20 mM sodium acetate, pH 5.5. Electrophoretically homogeneous cathepsin O2 was eluted at 1.4 M NaCl. NH 2 -terminal sequencing was carried out by automated Edman degradation.
Recombinant human cathepsin S was expressed using the baculovirus expression sytem and purified as described elsewhere. 2 Recombinant human cathepsin L was kindly provided by Dr. J. S. Mort (Shriners Hospital for Crippled Children, Montreal, Canada). All cathepsins used were electrophoretically homogeneous, and their molarities were determined by active site titration with E-64 as described by Barrett and Kirschke (1981).
Antibodies to Human Cathepsin O2 and Immunohistochemistry-Polyclonal antibodies were made in New Zealand White rabbits to the proenzyme of human cathepsin O2. The cDNA encoding the proenzyme was amplified by polymerase chain reaction from a full-length clone of human cathepsin O2 cDNA using Pfu DNA polymerase (Promega). The primers used were made to the 5Ј-end of the proenzyme with an NheI site and to the 3Ј-end with a BamHI site. Human cathepsin O2 was cloned and expressed in E. coli (BL21(DE3)) in the pET11c vector from Novagen. Expression was induced with 0.4 mM isopropyl-1-thio-␤-Dgalactopyranoside at A 600 ϭ 0.6, and cells were harvested 2 h after induction. After collection, the expressed proteins were run on Novex 12% Tris/glycine SDS gels, which were Coomassie-stained and destained. The proenzyme band of cathepsin O2, which was confirmed by N-terminal sequencing, was cut out. The protein was electroeluted from the gel slices and concentrated on a Centricon 10, which was pretreated with 1 ϫ elution buffer. The antigen was brought up to 1 ml in 1 ϫ phosphate-buffered saline and used for immunization (EL Labs, Soquel, CA). The antibodies were purified from the whole serum with acetone powder made to an induced culture of BL21(DE3) and by affinity binding to and elution from the antigen on nitrocellulose. The purified antibodies are specific for human procathepsin O2 and for the mature enzyme, and do not exhibit cross-reactivity with human cathepsins S, L, and B in Western blot analysis at a 1:2000 dilution.
Formalin-fixed and paraffin-embedded osteoclastoma sections were deparaffinized in xylene and rehydrated with a 100, 95, 70, 50% alcohol series. The sections were washed with phosphate-buffered saline, pH 7.4, boiled in Citra antigen retrieval buffer as described (Biogenex, San Ramon, CA), and allowed to cool to room temperature. The sections were blocked with 5% normal goat serum, 0.1% (v/v) Triton X-100, phosphate-buffered saline, and 3% hydrogen peroxide, and incubated with control rabbit IgG (2.5 g/ml) or affinity-purified anti-cathepsin O2 antibodies (250 ng/ml). Specific immunoreactivity was detected with a biotin/streptavidin-horseradish peroxidase-based detection system (Biogenex). After development of the chromogen (3,3Ј-diaminobenzidine), the sections were washed with water and treated with a 50, 70, 95% alcohol series. The sections were counterstained with 0.025% (w/v) fast green FCF, 95% alcohol, further treated with 100% alcohol and xylene, and permanently mounted for microscopy.
Cathepsin O2 Assays with Methylcoumarylamide Substrates and Inhibitors-Initial rates of substrate hydrolysis were monitored in 1-cm cuvettes at 25°C in a Perkin-Elmer fluorimeter at excitation and emission wavelengths of 380 and 450 nm, respectively. Recombinant human cathepsin O2 was assayed at a constant enzyme concentration (1-18 nM) in 50 mM sodium acetate buffer, pH 6.5, containing 2.5 mM dithioerythreitol and 2.5 mM Na 2 EDTA.
The kinetic constants V max and K m were obtained by nonlinear regression analysis using the program Enzfitter (Leatherbarrow, 1987).
Inhibition of cathepsin O2 was assayed at a constant substrate (5 M Z-FR-MCA) and enzyme concentration (1 nM) in the presence of different inhibitor concentrations in the substrate assay buffer. Cathepsin O2 was preincubated with the inhibitors for 10 min, and the reaction was started with substrate. Residual activity was monitored, and percent inhibition was calculated from the uninhibited rate.
pH Activity Profile and pH Stability-Initial rates of substrate hydrolysis were monitored as described above. The pH activity profile of human cathepsin O2 was obtained at 1 M substrate (Z-FR-MCA) concentration ([S] ϽK m where the initial rate v 0 is directly proportional to the k cat /K m value). The following buffers were used for the pH activity profile: 100 mM sodium citrate (pH 2.8 -5.6) and 100 mM sodium phosphate, pH 5.8 -8.0. All buffers contained 1 mM EDTA and 0.4 M NaCl to minimize the variation in ionic strength. A three protonation model  was used for least square regression analysis of the pH activity data. The data were fitted to the following equation.
The pH stability of cathepsins O2, S, and L was studied at three different pH values. Recombinant human cathepsins O2, S, and L were incubated at 37°C in 100 mM sodium acetate buffer, pH 5.5, in 100 mM potassium phosphate buffer, pH 6.5, and in 100 mM Tris-HCl, pH 7.5, containing 5 mM dithiothreitol and 2.5 mM EDTA. Incubating for 0.5, 1, 2 and 4 h, the activity remaining was determined using 5 M Z-FR-MCA for cathepsin O2 (100 mM potassium phosphate buffer, pH 6.5) and cathepsin L (100 mM sodium acetate buffer, pH 5.5) and 5 M Z-VVR-MCA for cathepsin S (100 mM potassium phosphate buffer, pH 6.5).
Elastase, Collagenase, and Gelatinase Activity-[ 3 H]Elastin was prepared as described previously (Banda et al., 1987) and had a specific activity of 113,000 cpm/mg of protein. Elastin (2 mg) was incubated in 1 ml of buffer containing 2.5 mM dithiothreitol, 2.5 mM EDTA, and 0.05% Triton X-100 for the cathepsin O2, S, and L assays. Aliquots were withdrawn after 10, 20, 30, 50, 90, 120, and 180 min, centrifuged for 1 min at 14,000 ϫ g, and counted in a 24-well plate containing scintillation fluid with a liquid scintillation counter (1450 Microbeta Plus, Wallac/Pharmacia). Concentrations of human cathepsins O2, S, L, and bovine elastase in the elastin degradation assay were 65, 28, 80, and 80 nM, respectively. To determine the pH effect on protease activity, the digests were carried out at pH 4.5 and 5.5 (100 mM sodium acetate, 2.5 mM each dithiothreitol and EDTA, 0.05% Triton X-100), and at pH 7.0 (100 mM Tris-HCl, 2.5 mM each of dithiothreitol and EDTA, 0.05% Triton X-100). Pancreatic bovine elastase (Boehringer Mannheim) was assayed under the same conditions except that neither dithiothreitol nor EDTA was added to the incubation mixture.
Soluble calf skin Type I collagen was diluted to 0.4 mg/ml into 100 mM sodium acetate buffer, pH 4.0, 5.0, and 5.5, in 100 mM potassium phosphate buffer, pH 6.0 and 6.5, and 100 mM Tris-HCl, pH 7.0, containing 2 mM dithiotreitol, 2 mM EDTA. Human cathepsins O2, S, and L and bovine trypsin (Sigma) were incubated at concentrations of 100 nM enzyme concentration for 10 h at 28°C. To measure the gelatinase activity of cathepsins O2 and S, Type I collagen was heated for 10 min at 70°C prior to incubation with the proteases. In the presence of 0.1 nM cathepsin O2, 0.2 nM cathepsin L and 2 nM cathepsin S, respectively, the reaction mix was incubated for 30 min at 28°C. The samples were subjected to SDS-polyacrylamide electrophoresis using 4 -20% Tris/ glycine gels (Novex, San Diego, CA).

RESULTS AND DISCUSSION
Expression and Activation of the Precursor of Human Cathepsin O2-Sf9 cells infected with AcNPV[preproCatO2] were harvested 84 h postinfection. The majority of immunoreactive material of about 43 kDa was found within the infected cells. In contrast to the single product of 43 kDa in the culture medium, an additional slight band of 44 kDa was detected in the cellular extract (data not shown). We assume that the higher molecular weight band represents unprocessed preprocathepsin O2, whereas the 43-kDa protein is proenzyme. No activity was observed immediately after lysis of the cells nor during autoactivating conditions at 40°C between pH 4.0 and 4.5 in the presence of dithiothreitol using the synthetic substrate Z-FR-MCA at pH 7.5. The increase of an E-64 inhibitable activity under autoactivating conditions and measured at pH 5.5 was assigned to an endogenous Sf9 cysteine protease. 3 No processing of the cathepsin O2 precursor was observed with human cathepsin B incubated at pH 4.0 and 5.5 for 2 h at 37°C (data not shown).
The precursor of cathepsin O2 can be efficiently transformed into mature active enzyme in presence of pepsin at pH 4.0. The digest of crude cellular extract or of concentrated culture media supernatant results in a time-dependent disappearance of precursor and generation of mature enzyme (29 kDa) via an intermediate of 36 kDa (Fig. 1). In parallel with this process an increase of E-64 inhibitable activity measured at pH 7.5 is observed.
The purified enzyme is a single-chain enzyme and exhibits an apparent molecular mass of 29 kDa in a 4 -20% Tris/glycine SDS gel under reducing conditions. Treatment with endoglycosidases H and F as well as N-glycosidase F did not result in a shift in the molecular weight, which implies that the protease is not glycosylated (data not shown). Human cathepsin O2 has two potential glycosylation sites in its mature sequence (Brömme and Okamoto, 1995). However, both sites have either a proline residue adjacent to the asparagine or to the threonine, so that their use is unlikely (Gavel and von Heijne, 1990). Additionally, cathepsin O2 contains one putative glycosylation site in the propart close to the processing site between the mature enzyme and the propart. Again, no shift in molecular weight of the proenzyme was observed after overnight treatment with endoglycosidases H and F as well as N-glycosidase F. N-terminal sequencing of the mature protease processed in presence of pepsin revealed the typical processing site described for cysteine proteases of the papain family with a proline adjacent to the N-terminal alanine (NH 2 -APDS-VDYRKKGYVTPVKN) (Rowan et al., 1992). Since a cleavage by pepsin after arginine (position Ϫ1) is very unlikely, it is possible that this site was generated after an initial cleavage by pepsin in the proregion followed by an aminopeptidase activity present in the Sf9 cell extract or by an active intermediate of cathpsin O2 generated by pepsin. Most aminopeptidases will stop at an alanine residue (position 1 of the mature enzyme sequence) due to the subsequent proline residue. This mechanism of processing involving aminopeptidases is also discussed for procathepsin B (Rowan et al., 1992).
In contrast, autocatalytically activated cysteine proteases frequently have at their processing site an N-terminal extension of 3-6 amino acids from the propart . No activation of the precursor was observed by addition of purified active cathepsin O2 at pH 4.5 (data not shown), indicating that neither an initial cis nor trans autoactivation of cathepsin O2 within the lysosomes is likely. This contrasts related cysteine proteases such as papain and cathepsin S, which exhibit a potential autocatalytic activation pathway (Vernet et al., 1990;Brömme et al., 1993).
The activating enzyme of cathepsin O2 within the osteoclast could be the aspartyl protease cathepsin D, which is present in osteoclastic lysosomes but secreted at low levels into the resorption lacuna (Goto et al., 1993). S 2 P 2 Subsite Specificity of Recombinant Human Cathepsin O2-The S 2 P 2 specificity of human cathepsin O2 was characterized using synthetic substrates of the type Z-XR-MCA with X equal to F, L, V, or R. The S 2 subsite pocket of cysteine proteases is structurally well defined and determines the primary specificity of this protease class. For example, cathepsin B contains a glutamate (Glu-245) residue at the bottom of the S 2 subsite pocket, which favors the binding of basic residues like arginine. This glutamate residue is replaced by neutral residues in all other known human cathepsins resulting in a very low hydrolysis rate of the Z-RR-MCA substrate. Cathepsin O2 contains a leucine residue in position 205 which makes Z-RR-MCA a very poor substrate (Fig. 3). The specificity of cathepsin O2 toward P 2 residues resembles that of cathepsin S. Both enzymes prefer a leucine over a phenylalanine in this position, while cathepsin L is characterized by an inverse specificity (Table II, Fig. 3). Valine in position P 2 is relatively well 3 D. Brömme, unpublished results.  accepted by cathepsin O2, whereas the presence of this ␤-branched residue in P 2 results in a poor substrate for cathepsins L, S, and B. The catalytic efficiency (k cat /K m ) of cathepsin O2 toward dipeptide substrates is comparable with that of cathepsins S and B but is approximately 1 order of magnitude lower than that of cathepsin L. Interestingly, the K m values for cathepsin O2 are comparable with those determined for cathepsin L. The K m value reflects to some extent the affinity of the substrates for the protease. This trend is even more obvious for the tripeptide substrate, Z-LLR-MCA, which displays a K m value as low as 4 ϫ 10 Ϫ7 M (Table II). However, in contrast to cathepsins S and L, the k cat values are almost 2 orders of magnitude lower for cathepsin O2.
pH Activity Profile and pH Stability of Human Recombinant Cathepsin O2-Profiles of pH activity are sensitive measures of enzymatic functional and structural integrity. A comparison of pH profiles from different but related proteases reveals differences in intrinsic activity and stability of these proteases. Human cathepsin O2 displays a bell-shaped pH profile with flanking pK values of 4.0 and 8.13 (Table III; Fig. 4). Its pH optimum is between 6.0 and 6.5 and comparable with that observed for cathepsin S . The width of the pH profile, which mirrors the stability of the ion-pair , is 4.15 for cathepsin O2 but only 3.35 for cathepsin S . This parameter for human cathepsin O2 is more similar to that observed for the very stable papain, which displays a profile width of 3.91 .
Human cathepsin O2 is more stable than cathepsin L at slightly acidic to neutral pH values but less stable than cathepsin S (Table IV). Approximately 50% of the cathepsin O2 activity remains after 1 h at 37°C and pH 6.5, whereas essentially no cathepsin L activity could be observed under these conditions. However, it must be considered that the pH stability was determined without substrate protection, which usually increases the pH stability. In the [ 3 H]elastin degradation assay with cathepsin O2, an increase of solubilized 3 H fragments was still observed after 2 h at pH 7.0.
Inhibitor Profile of Human Cathepsin O2-Human cathepsin O2 displays a typical inhibitor profile of a cysteine protease. It is inhibited by cysteine protease inhibitors and by inhibitors of both cysteine and serine proteases (Table V). At concentrations above 0.1 M, peptide aldehydes, diazomethanes, E-64, and chicken cystatin completely inhibit enzyme activity. On the other hand, specific serine and aspartic protease inhibitors do not affect enzyme activity. No effect of EDTA at a concentration of 4 mM was observed on the activity of cathepsin O2. At higher concentrations (Ͼ5 mM) a partial nonspecific inhibition was observed.
Degradation of Type I Collagen and Elastin by Human Cathepsin O2-The elastinolytic activity of human cathepsin O2 was measured at pH 4.5, 5.5, and 7.0 against insoluble [ 3 H]elastin. Maximal activity was observed at pH 5.5. Cathep-   3.33 Ϯ 0.14 4.22 Ϯ 0.28 6.95 Ϯ 0.09 5.6 Papain c 3.58 Ϯ 0.29 4.54 Ϯ 0.29 8.45 Ϯ 0.02 6.5 a Calculated from (pK 1 ϩ pK 2 )/2. b From Brömme et al. (1993). c From Khouri et al. (1991). sin O2 has an elastinolytic activity between pH 4.5 and 7.0 that is 1.7-3.5 times higher than that of cathepsin S. Its elastinolytic activity at the pH optimum of cathepsin L (pH 5.5) and at neutral pH is almost 9 and 2.4 times higher when compared with cathepsin L and pancreatic elastase, respectively (Fig. 5). The values determined for cathepsin L and S are in good accordance with published data (Xin et al., 1992;Kirschke and Wiederanders, 1994). The ability of cathepsin O2 to degrade elastin may indicate an involvement in the pathogenesis of lung diseases. Northern blot analysis revealed relative high levels of expression in lung (Brömme et al., 1995). The high elastinolytic activity at neutral pH (twice the activity of porcine elastase and cathepsin S) may implicate an extracellular activity of cathepsin O2. Since cathepsin S displays a similar high elastinolytic activity at neutral pH and is highly expressed in lung macrophages Shi et al., 1992;Reddy et al., 1995), a concerted activity of both enzymes, cathepsin O2, and cathepsin S, is possible in lung emphysema.
The collagenolytic activity against Type I collagen was determined between pH 4.0 and 7.0. Cathepsin O2 extensively degrades Type I collagen between pH 5.0 and 6.0 at 28°C, whereas the degradation at pH 4.5 and 7.0 is much less pronounced (Fig. 6A). The primary cleavage seems to occur in the telopeptide region since the ␣ monomers released from the ␤ and ␥ components are slightly smaller. Additionally cleavage may also occur within the ␣ monomers. It is yet unclear whether the cleavage occurs in the intact helical region or in unraveled ␣ monomers. Major fragments of Type I collagen observed after cathepsin O2 action have the size of 70 -80 kDa.
Cathepsin L also cleaves in the telopeptide region, but essentially no small molecular weight fragments were detected under the conditions used. The effective pH range for the collagenolytic activity of cathepsin L is more acidic when compared with that observed for cathepsin O2 (between pH 4.0 and 5.5). Cathepsin S seems to reveal only a very weak collagenolytic activity. In contrast, tissue collagenases cleave the monomers into 3/4 and 1/4 fragments (Gross and Nagai, 1965). No degradation of Type I collagen was observed with trypsin at equal enzyme concentration compared with cathepsin O2, showing that the integrity of the triple helix of the collagen used was not impaired (data not shown).
In addition to its collagenase activity, cathepsin O2 displays a powerful gelatinase activity. At 0.1 nM concentration of the enzyme, denatured collagen is totally degraded within 30 min within a pH range of 5.0 -7.0. In contrast, cathepsin L displays its gelatinase activity only in the pH range between 4.5 and 5.5 (Fig. 6B). Cathepsin S is active between pH 4.0 and 7.0, but it displays a significantly weaker activity than that of cathepsins O2 and L.
Immunohistochemical Localization of Human Cathepsin O2 in Osteoclastoma Tissue-Expression of cathepsin O2 was previously detected by Northern blot analysis at high levels in osteoclastomas (Brömme and Okamoto, 1995). Immunostaining of an osteoclastoma revealed specific staining of multinucleated osteoclasts, whereas stromal cells displayed only a weak immunoreactivity (Fig. 7).
The Potential Role of Cathepsin O2 in Osteoclastic Bone Resorption-Current data suggest that osteoclastic bone resorption is mostly linked to the activity of cathepsins L and B (Everts et al., 1992;Delaisse and Vaes, 1992;Kakegawa et al., 1993), but the results presented here suggest human cathepsin O2 as a potential key player in bone remodelling. Message levels in human osteoclastoma preparations (Brömme and Okamoto, 1995) exhibit a manyfold higher level of expression of cathepsin O2 than cathepsin L. Immunohistochemical staining of multinucleated osteoclasts in osteoclastoma sections (Fig. 7) as well as osteoclasts in prenatal human bones 4 confirm the expression of the protease at the protein level. Also, demonstrated for the first time, cathepsin O2 is a highly active cysteine protease that is capable of hydrolyzing extracellular matrix proteins such as collagen and elastin. Type I collagen, the major structural protein component in bone, is efficiently hydrolyzed at pH values above 5.5 by cathepsin O2, a pH region where only a low or no activity is observed for cathepsin L.
The pH value in the subosteoclastic resorption zone underneath the ruffled border is approximately between pH 3.5 and 4.5 (Fallon, 1984;Silver et al., 1988). However, it can be assumed that in the lower part of the resorption lacuna, at the interface between the demineralized and mineralized matrix, a pH gradient toward a neutral pH value exists due to the buffering capacity of dissolved bone salts. This area of the resorption lacuna is the proposed compartment for the action of interstitial collagenase, which is only active at slightly acidic to slightly alkaline pH (Delaisse and Vaes, 1992). The results presented for the pH stability, pH activity profile as well as the in vitro collagenase activity of cathepsin O2, suggest that this novel cysteine protease is active at the site where interstitial collagenase is active and that both activities may act in a concerted manner on Type I collagen. Besides cathepsin S , cathepsin O2 is the second known human papain-like cysteine protease that displays a bell-shaped pH profile with a pH optimum at 6.0. Cathepsin L, due to its high instability at neutral pH  is excluded from this site and may be involved in a later degradation stage 4 A. Rinne and D. Brömme, unpublished data.

FIG. 6. SDS-PAGE of type I collagen (soluble calf skin collagen) after digestion with recombinant human cathepsins O2, L, and S.
A, collagenase activity. Digestion of soluble calf skin collagen at 28°C and at pH 4.0, 5.0, 5.5, 6.0, 6.5, and 7.0 by human cathepsins O2, S, and L (each 50 nM) for 12 h is shown. The reaction was stopped by the addition of 10 M E-64. Untreated soluble collagen was used as standard (S). B, gelatinase activity. Digestion of denatured soluble calf skin collagen (10 min heated at 70°C) at 28°C and at pH 4.0, 5.0, 5.5, 6.0, 6.5, and 7.0 by human cathepsin O2 (0.1 nM), human cathepsin L (0.2 nM) and human cathepsin S (2 nM) is shown. Molecular mass standards are indicated in the right lane.
FIG. 7. Immunohistochemical staining of human cathepsin O2 in a human osteoclastoma. A, osteoclastoma, stained with affinity-purified anti-cathepsin O2 antiserum, magnification 400ϫ. B, osteoclastoma, stained with control rabbit IgG, magnification 400ϫ. of collagen in the more acidified microenvironment of the resorption lacuna. Demonstration of cathepsin O2 secretion into the bone resorption lacuna will need to be achieved in order to verify this hypothesis.
In addition, cathepsin O2 is distinguished by a very potent gelatinase activity in the pH range between 5 and 7, whereas the gelatinase activity of cathepsin L is limited to the pH range between 4 and 5.5. Consequently, a rapid degradation of the collagen cleavage products released by the interstitial collagenase is possible in the pH micro environment of the tissue collagenase.
Studies with cysteine protease inhibitors, E-64, leupeptin, and diazo methanes, clearly demonstrate that cysteine protease inhibitors are capable of inhibiting bone resorption (Delaisse et al., 1980(Delaisse et al., , 1987Debari et al., 1995). Since all three types of inhibitors are very effective on cathepsin O2, it is likely that these inhibition studies reflect to a large extent the inhibition of cathepsin O2.
In conclusion, cathepsin O2, is highly expressed in osteoclasts. Cathepsin O2 is distinguished from classical human cysteine proteases like cathepsins L and B by its increased stability at neutral pH as well as its ability to efficiently hydrolyze Type I collagen and elastin at pH values above 6.0. The expression of cathepsin O2 protein in osteoclasts as well as its enzymatic properties imply a central role in normal bone remodelling as well as in pathological processes, such as osteoarthritis, osteoporosis, and multiple myeloma osteoclastomas. The design of potent cathepsin O2 inhibitors could be an important contribution in the efforts to arrest these pathological processes.