Autocatalytic Activation of Human Cathepsin K*

The in vitro activation of the recombinant purified human cathepsin K (EC 3.4.22.38) was examined by mutagenesis. Cathepsin K was expressed as a secreted proenzyme using baculovirus-infected Sf21 insect cells. Spontaneous in vitro activation of procathepsin K oc- curred at pH 4 and was catalyzed by exogenous mature cathepsin K. Three intermediates were identified as re- sulting from cleavages after Glu 19 , Ser 98 , and Glu 110 . The mature enzyme was composed of mixture of enzymes with N termini of Gly 113 , Arg 114 , and Ala 115 with varying ratios depending on the preparation. Molecular weight determinations were consistent with the absence of carbohydrate in the mature protein, while electrospray mass spectroscopy indicated that six of the eight cysteine residues were in disulfide linkage, and that the protein had Met 329 as the C-terminal residue. A mutant was constructed in which the active site Cys 139 was changed to Ser. [Ser 139 ,Ala 163 ]Procathepsin K (containing mutation C139S,S163A) failed to spontane- ously process and was only partially processed in the presence of 1% exogenous wild-type mature cathepsin K forming intermediates, which were identical to those observed in the activation of wild-type. [Ser 139 ,Ala- 163 ]Procathepsin K could be fully processed to mature enzyme by including one equivalent of wild-type proca- Sequence on ysis of the carbohydrate is located on Asn 103 in the propeptide domain, although our results do not rule out the possibility of O -gycosylation elsewhere in the propeptide domain. Our results contrast with those of Bro¨mme et al. (13), who observed no glycosylation on procathepsin K. The discrepancy may be a result of different expression conditions.

Bone remodeling is a constant process that involves bone resorption and rebuilding (for review, see Ref. 1). The resorption phase of this process is carried out by osteoclasts, which adhere to the surface of bone leading to the creation of an extracellular compartment termed the resorption pit. The resorption pit is maintained at an acidic pH, causing the dissolution of the mineral components of the underlying bone and exposure of the proteinaceous matrix to the action of proteolytic enzymes (2)(3)(4)(5)(6). The rebuilding phase of the remodeling process involves the recruitment of osteoblasts to the sites of prior bone resorption, where the layering of a new proteinaceous matrix occurs and becomes mineralized.
Activation of procathepsin K in vivo is likely to occur in the low pH environment of the resorption pit, via two possible mechanisms. The propeptide may be cleaved by another protease, such as cathepsin D as suggested by Brömme et al. or by an autocatalytic process, which is more consistent with the data presented by Bossard et al. (14).
To elucidate the mechanism of activation of cathepsin K, we constructed a mutant in which the presumed active site Cys at position 139 was changed to Ser. The kinetics of activation of mutant and wild-type cathepsin K were studied in vitro.
In this report we provide the following evidence for an autocatalytic activation mechanism. First, in vitro self-activation of wild-type procathepsin K occurs spontaneously at 4°C, pH 4 and is catalyzed by mature cathepsin K. Second, unlike wildtype enzyme, the [Ser 139 ,Ala 163 ]procathepsin K mutant lacks the ability to self-process into a mature active enzyme, but can be processed by addition of wild-type mature cathepsin K. Significantly, the intermediates observed in this trans-processing of [Ser 139 ,Ala 163 ]procathepsin K are identical to those observed in the spontaneous activation of wild-type procathepsin K. Finally, procathepsin K has trace proteolytic activity, suggesting autocatalysis may occur in vivo.

EXPERIMENTAL PROCEDURES
Materials-Z 1 -Phe-Arg-AMC was obtained from Bachem; L-cysteine⅐HCl was from Amresco; MES was from Calbiochem Corp.; EDTA and E64 were from Sigma; prestained molecular weight markers were obtained from Amersham Life Sciences; precast SDS-PAGE gels were purchased from Bio-Rad. Protein concentrations were estimated by the Bradford method using the Bio-Rad protein assay except where otherwise specified.
SpeI fragment containing the coding region of pre-procathepsin K was subcloned into pSelect301, a modified version of pSelect (Promega) designed to contain additional restriction sites within the multiple cloning region. 2 Single-stranded phagemid DNA was generated in KO7infected Escherichia coli strain JM101. The following oligonucleotides were used as primers for mutagenesis reactions: 5Ј-AGAGCTAAAAGC-CCagaGGAACCACACTGACC-3Ј and 5Ј-CACTAGGTTCTGGGGagc-CAGATTTAAGAGTTTGCC-3Ј. This changed Cys 139 to Ser and Ser 163 to Ala, respectively, creating a double mutant. Both the C139S and S163A mutations were confirmed by automated DNA sequencing (Applied Biosystems, Inc.). The mutagenized 1.36-kilobase BamHI/SpeI fragment was then subcloned into the baculovirus transfer vector pVL1393, which had been digested with BamHI and XbaI, generating the plasmid pBacMut1CatK.
For construction of recombinant viruses, Sf21 cells were co-transfected with purified AcNPV linear DNA (Pharmingen) and pBacMut1CatK using the method described to generate recombinant virus (vBacCatK) for native cathepsin K (14).
Expression of Wild-type or Mutant Cathepsin K in Recombinant Virus-infected Cells-Sf21 cells were infected with about 4 plaque-forming units of either wild-type vBacCatK or mutant vBacMut1CatK recombinant virus/cell at 27°C in serum-containing medium. Twenty-four hours after infection, the cells were pelleted, resuspended in a serumfree medium, and incubated for an additional 72 h. Western blot analyses were performed on aliquots of the medium as described previously elsewhere (14).
Purification of Wild-type and Mutant Procathepsin K-Baculovirus conditioned medium (10 liters) containing wild-type or [Ser 139 ,Ala 163 ]procathepsin K at pH 6.5 was loaded onto a 100-ml S-Sepharose Fast Flow column (Pharamacia XK50, 5 ϫ 5 cm) pre-equilibrated with 20 mM sodium phosphate, pH 6.9 (buffer A). The column was washed to base line with 10 column volumes of buffer A. Bound material was eluted with a series of NaCl steps (0.35 M, 0.5 M, and 1 M) in buffer A. Fractions were analyzed by SDS-PAGE and Western blot. Procathepsin K eluted in the 0.5 M NaCl fraction and was analyzed by N-terminal sequencing and MALDI-MS (see below).
Protein concentration was determined using a BCA (bicinchoninic acid) protein assay with bovine serum albumin standards (Pierce). Purified wild-type and mutant procathepsin K were concentrated to between 1 and 2.5 mg/ml using an Omegacell (Filtron Technology Corp.) with a 10-kDa molecular mass cut-off membrane.
E64 Treatment of Procathepsin K-E64 (70 nmol) was added to 0.2 ml of procathepsin K (2.5 mg/ml, 14 nmol) in 20 mM sodium phosphate, 0.5 M NaCl at pH 6.8. After 15 min at 25°C, the mixture was dialyzed against 5 liters of 20 mM sodium phosphate, 0.5 M NaCl, pH 6.8 overnight at 4°C using a a 12-14-kDa cutoff membrane (Spectra-Por, Spectrum Medical Industries, Inc.). The enzyme was diluted to 1 mg/ml with dialysis buffer prior to activation.
Activation of Wild-type Procathepsin K-Concentrated procathepsin K was incubated in activation buffer consisting of 0.2 M sodium acetate, 20 mM L-cysteine adjusted to pH 4.0 at 4°C on a rotary mixer. Where indicated, 1% (mass/mass) mature cathepsin K was added as seed after pH adjustment. The mature cathepsin K that was used as seed was initially obtained from 60°C induced activation of procathepsin K, a gift from M. Bossard (SmithKline Beecham Pharmaceuticals) (14). Subsequently, "seed" was made by the 4°C activation procedure described here. The extent of activation and processing were assessed hourly by measuring hydrolytic activity as described below and by SDS-PAGE, respectively. SDS-PAGE analysis was carried out using 15% Tris-glycine Ready Gels (Bio-Rad). When the specific activity stopped increasing (ϳ15-25 mol/min/mg), the reaction was stopped by the addition of E64 or by snap-freezing in a bath of dry ice in acetone.
Cathepsin K Activity Assay-Cathepsin K activity was determined using a fluorogenic substrate in a microtiter plate format. The reactions consisted of 0.04 -1 g of mature cathepsin K, 20 M Z-Phe-Arg-AMC in 100 mM sodium acetate, 20 mM L-cysteine, 5 mM EDTA at pH 5.5. Reactions were initiated by the addition of substrate-containing assay buffer to the enzyme sample. The assay followed Z-Phe-Arg-AMC hydrolysis, which was measured using a Dynatech microtiter plate reader with excitation at 365 nm and fluorescence emission at 530 nm. Under these conditions the sensitivity of the assay was typically 4.7 ϫ 10 Ϫ7 mol of AMC/fluorescence unit. Procathepsin K was assayed under the same conditions in the same buffer, but also included 100 mM MES and 100 mM HEPES so that the pH could be varied between 4 and 7.
Activation of [Ser 139 ,Ala 163 ]Procathepsin K by Wild-type Mature Ca-thepsin K or Pepsin-[Ser 139 ,Ala 163 ]procathepsin K was concentrated to 1 mg/ml and was used in three 0.1-ml reactions. All reactions contained activation buffer (0.2 M sodium acetate, 20 mM L-cysteine at pH 4) and were mixed on a rotary mixer at 4°C. No mature cathepsin K was added to the first reaction, whereas the second reaction included 1% wild-type mature cathepsin K (1 g, specific activity 21 mol/min/mg). The third reaction consisted of a mixture of 50 l of [Ser 139 ,Ala 163 ]procathepsin K (1 mg/ml), 50 l of wild-type procathepsin K (1 mg/ml) and 1% mature cathepsin K. One-microliter and 10-l samples were taken at 1-h intervals for activity and SDS-PAGE analysis, respectively. Proteolytic cleavage of the propeptide from [Ser 139 ,Ala 163 ]procathepsin K was also accomplished using a modification of the published procedure utilizing pepsin (13). Briefly, [Ser 139 ,Ala 163 ]procathepsin K (1 mg/ml) was incubated with pepsin (20 g/ml) in 0.2 M sodium acetate, 5 mM EDTA, 1 mM L-cysteine, and 5 mM dithiothreitol at 40°C for 45 min. N-terminal Sequence Analysis-Sequence analysis was performed on a Beckman LC-3400 TriCart gas-phase protein sequencer equipped with a Beckman 126/166 system for on-line phenylthiohydantoin analysis (Beckman Instruments, Inc., Fullerton, CA). Data was acquired using System Gold chromatography software. Samples were electroblotted onto polyvinylidene difluoride type supports (Problott), and standard Beckman optimized polyvinylidene difluoride sequencing cycles were used.
MALDI-Mass Spectrometry-MALDI-MS data were obtained on a Voyager RP laser desorption time-of-flight mass spectrometer (PerSeptive Biosystems, Framingham, MA). Protein samples were prepared for analysis by diluting analyte 1:5 with 3,5-dimethoxy-4-hydroxycinnamic acid (10 mg/ml in 2:1 0.1% trifluoroacetic acid/acetonitrile) for a final concentration of 1-10 pmol/l. Bovine ␤-lactoglobulin A (Sigma) was included as an internal calibrant (MH ϩ 18364 Da). Desorption/ionization was accomplished using photon irradiation from a 337-nm pulsed nitrogen laser and 25-keV accelerating energy. Spectra were averaged over ϳ100 laser scans. Calibrations were carried out using a customized version of IGOR Pro (WaveMetrics, Inc., Lake Oswego, OR) on a Macintosh personal computer.
Electrospray Mass Spectrometry-Both the apo-cathepsin K (32 M) and the E64-cathepsin K product (39 M) were supplied in 4 mM MES buffer also containing 10 mM NaCl and 0.4 mM L-cysteine at a pH of 6.06. These solutions were diluted with 10 l of 88:8:4 MeOH:water: formic acid, followed by 80 l of a 1:1:0.2% solution of MeOH:water: formic acid, resulting in a final cathepsin K concentration of ϳ1.5 M and a final volume of 100 l. Two microliters of this solution was loaded into a pulled glass capillary for ultralow volume electrospray (nanospray) mass spectrometry. Mass spectra were recorded on a PE-Sciex API III (PE-Sciex Instruments, Concord, Ontario, Canada) by repetitively scanning the m/z range of 1050 -1650 with a step size of 0.05 and a dwell of 1 ms, and averaging 5-10 min of accumulated data. Results for all charge states corresponding to [M ϩ nH] nϩ and [M ϩ(n Ϫ 1)H ϩ Na] nϩ between 15ϩ and 22ϩ were averaged together, producing a typical 95% confidence interval of Ϯ 1.5 Da.

Expression of Wild-type and Mutant Cathepsin K-Infection
of Sf21 cells with either recombinant wild-type cathepsin K (vBacCatK) (14) or mutant (vBacCatKMut1) virus resulted in the production of identically sized proenzyme protein bands of approximately 35,000 Da as determined by Western blot analysis using a previously generated antiserum (14). As was determined for the wild-type proenzyme (14), most of the mutant protein was also secreted into the medium with a smaller percentage of the expressed protein retained in the cell pellets (data not shown).
Purification of Wild-type Procathepsin K-The secreted proenzyme was purified to greater than 85% homogeneity, as described under "Experimental Procedures" (Fig. 1, lane 1). The N-terminal sequence of the purified proenzyme was LYPEEILD, which indicated cleavage of the secretion signal sequence occurred after Ala 15 . No secondary sequence was observed. Analysis the proenzyme by MALDI-MS yielded a mass of 36366 Da, which exceeded by 3% the theoretical mass, 35300.5 Da, calculated from the amino acid sequence. Edman sequencing data gave a very low yield for Asn 103 , the potential N-linked glycosylation site in the propeptide domain, consist-ent with glycosylation of this residue. Absence of glycosylation at Asn 161 in the mature form of the protein was indicated by MS (see below). Several attempts to further purify the enzyme by a variety of ion exchange and size exclusion chromatography resulted in poor recovery of procathepsin K.
Activation of Wild-type Procathepsin K: Catalysis by Mature Cathepsin K-The time courses of activation reactions containing zero (open circles) and 1% mature cathepsin K (filled circles) were determined (Fig. 2). Procathepsin K activated without mature cathepsin K at 4°C, pH 4. The reaction containing 1% mature cathepsin K had no apparent lag and required a shorter time to obtain full enzymatic activity. SDS-PAGE analysis of the proteolytic conversion to mature enzyme in the absence of seed (Fig. 1) indicated the accumulation of intermediates (25-35 kDa) and propeptide fragments (6.5-12 kDa). Full activity, typically 15-25 mol/min/mg, was determined at the end of the activation when an accurate protein concentration was determined. The reaction of 1% mature cathepsin K with procathepsin K that had been pretreated with E64 did not produce any activated mature cathepsin K (Fig. 2, open triangles), and Ͻ5% propeptide degradation was observed by SDS-PAGE (data not shown).
Early cleavage sites were determined by N-terminal sequencing of the protein bands isolated from a blot of a SDS-PAGE gel from a subsequent activation reaction using identical conditions to the one described above. The two largest intermediates resulted from cleavages which occurred immediately after Glu 19 and Ser 98 , respectively. The third and smallest intermediate was derived from cleavage after Glu 110 .
Characterization of Mature Cathepsin K-N-terminal sequence analysis indicated that mature cathepsin K was composed of a mixture of enzymes with N termini of Gly 113 RAP-, Arg 114 APD-and Ala 115 PD-, with varying ratios depending on the preparation. These results were consistently observed in four individual activation reactions. MALDI-MS of a pool of activated enzyme that exhibited a 1:1 ratio of Gly 113 and Arg 114 N termini yielded a M r of 23,696 for the unresolved protein components. Electrospray mass spectrometry (ESMS) for a preparation in which Arg 114 and Ala 115 dominated as the N termini yielded molecular masses of 23,646.7 Ϯ 1.5 Da and 23,590.1 Ϯ 1.5 Da for the two protein species (calculated ϭ 23,645.7 and 23,489.5, respectively, for the protein ending with Met 329 ). The electrospray data are consistent with six of the eight Cys residues being in disulfide linkage. Both the electrospray and MALDI-MS data indicate that the mature protein is not glycosylated.
ESMS of the complex formed by reaction of a preparation of activated, mature cathepsin K (Arg 114 , major and Ala 115 , minor N termini) with E64 yielded a determined M r for the major component of 24,001.8 (calculated ϭ 24,003.0). This mass is consistent with addition of the entire inhibitor into the protein via ring opening of the epoxide by the thiol of the active-site Cys residue. 139 ,Ala 163 ]Procathepsin K was purified to greater than 85% homogeneity as described under "Experimental Procedures" (Fig. 3A, lane 1). The Nterminal sequence of proenzyme indicated that cleavage of the secretion signal sequence occurred after the Ala 15 residue, consistent with that observed for wild-type protein. MALDI-MS analysis indicated the mutant procathepsin K contained approximately 2% glycosylation by mass.

Demonstration of Autoprocessing: Activation of [Ser 139 ,Ala 163 ]Procathepsin K-[Ser
The studies designed to distinguish between autocatalytic activation and activation catalyzed by a protease other than cathepsin K consisted of three different experiments. The first two experiments contained mutant procathepsin K, which was incubated in pH 4 buffer either in the absence or presence of one percent wild-type mature cathepsin K. The third experiment was done to observe full processing of the mutant procathepsin K and consisted of a 1:1 mixture of wild-type procathepsin K and mutant procathepsin K with 1% wild-type mature cathepsin K.
The result of the first experiment is shown in Fig. 3A. Mutant procathepsin K did not process to mature cathepsin K in the absence of wild-type mature cathepsin K seed, but was partially processed in the presence of a catalytic amount of wild-type mature cathepsin K as shown in Fig. 3B. There was no increase in activity (Fig. 4, open circles), but rather the activity of the seed decreased 15-fold during the first 2 h of the reaction. Under the reaction conditions the initially formed propeptide fragments were stable, which allowed for their characterization (see below). The third experiment consisting of a 1:1 mixture of mutant and wild-type cathepsin K, and one percent mature cathepsin K seed resulted in complete conversion to mature cathepsin K as shown in Fig. 3C. The specific activity of the resultant mature enzyme was 11 mol/min/mg, which was about one half that observed for the wild-type enzyme used in this experiment (25 mol/min/mg). The mutant enzyme was also processed to mature enzyme by pepsin (data not shown).
The intermediates and propeptide fragments (Fig. 3B) were  Fig. 2.   FIG. 2. Activity of wild-type mature cathepsin K during activation. Samples of purified procathepsin K were incubated at 4°C, in the presence of 0.2 M sodium acetate, 20 mM L-cysteine at pH 4.0 with no added mature cathepsin K (E), 1% mature cathepsin K (q), or procathepsin K that was preincubated with E64, dialyzed, and activated in the presence of 1% mature cathepsin K (Ç). The proteolytic activity was evaluated using the Z-Phe-Arg-AMC assay as described under "Experimental Procedures" and is expressed here in fluorescence units/minute (FU/min). S.A., specific activity. characterized by N-terminal sequencing and MALDI-MS. Cleavages occurred after Glu 19 , Ser 98 , and Glu 110 . Two fragments of the propeptide both had an N terminus starting at Glu 20 (shown better in Fig. 3C). The small fragment had an observed mass of 6756, which indicated the C-terminal would be Ala 74 . The large fragment had an observed mass of 9378, which would require the C terminus to be Ser 98 . A summary of the cleavages observed in both the wild-type and mutant activations is presented in Fig. 5.
Activity of Procathepsin K-Activation of procathepsin K in the absence of mature cathepsin K suggests that the proenzyme has proteolytic activity. Procathepsin K isolated from baculovirus medium, as described under "Experimental Procedures," was assayed for activity at pH 3.5-7.0 and was found to hydrolyze Z-PheArg-AMC having a maximal activity at pH 4. The specific activity, based upon estimates from initial velocity measurements, was 0.007-0.014 mol/min/mg, approximately 2000-fold lower than mature cathepsin K activity. DISCUSSION Bossard et al. (14) reported a successful small scale (80 g/ml) activation of semi-pure procathepsin K under two conditions: (a) brief exposure to elevated temperatures or (b) incubation at 4°C in the presence of a catalytic amount of preactivated mature cathepsin K. Their results suggested an autocatalytic mechanism for cathepsin K activation. Evidence against an autoactivation mechanism was provided by a study by Brömme et al. (13), in which no autoprocessing was observed and the activation could only be accomplished using pepsin (13). To examine the mechanism of cathepsin K activation, we examined the putative catalytic effect of preactivated mature cathepsin K.
In agreement with Bossard's observation, procathepsin K could be activated at 4°C and pH 4 in the absence of preactivated mature cathepsin K, and the activation was catalyzed by  Fig. 4. B, the [Ser 139 ,Ala 163 ]procathepsin K activation at 4°C, catalyzed by 1% mature cathepsin K. Lanes M, molecular size markers; lane 1, procathepsin K incubated at pH 4°in buffer at time ϭ 0.25 h. Lanes 2-5 are [Ser 139 ,Ala 163 ]procathepsin K incubated at 4°C in buffer at times ϭ 1, 3, 5.5, and 9.5 h, respectively. Samples from the time course correspond to the open circles in Fig. 4. The arrows point to intermediates, and the arrowhead points to propeptide fragments. C, the [Ser 139 ,Ala 163 ]procathepsin K activation at 4°C, containing one equivalent of wild-type procathepsin K catalyzed by 1% mature cathepsin K. Lanes M, molecular size markers; lane 1, procathpsin K, (mutant wildtype mixture) incubated at pH 4 in buffer at time ϭ 0.25 h. Lanes 2-5 are procathepsin K (mutant and wild-type mixture) incubated at 4°C in buffer at times ϭ 1, 3, 5.5, and 9.5 h, respectively. Samples from the time course correspond to the filled circles in Fig. 4. The arrow points to propeptide fragments that were characterized by sequencing and MALDI-MS. FIG. 4. Activity during the [Ser 139,Ala 163 ]procathepsin activation. Samples of [Ser 139 ,Ala 163 ]cathepsin K were incubated at 4°C in 0.2 M sodium acetate, 20 mM L-cysteine, pH 4.0, without preactivated mature wild-type cathepsin K (Ç), with 1% mature cathepsin K (E), and as a 1:1 mixture of wild-type procathepsin K and mutant procathepsin K with 1% wild-type mature cathepsin K (q). The proteolytic activity was evaluated using the Z-Phe-Arg-AMC assay as described under "Experimental Procedures" and is expressed here in fluorescence units/ minute (FU/min). S.A., specific activity. the addition of 1% mature cathepsin K. These results differ from those of Brömme et al.; however, there were several differences in the procedures employed. We activated enzyme that was greater than 85% pure and was at a much higher concentration in the activation reaction. Brömme et al. attempted to activate procathepsin K without purification in the baculovirus crude cell lysate. Additionally, we observed that the spontaneous activation in the absence of 1% mature cathepsin K could be inhibited by pretreating the procathepsin K with the cysteine protease inhibitor, E64, thereby demonstrating that activation was due to a cysteine protease.
Mature cathepsin K produced under the conditions described under "Experimental Procedures" contained a ragged N terminus, composed of a mixture of NH 2 -G 113 RAPD-, NH 2 -R 114 APDand NH 2 -A 115 PD-. The site for N-terminal cleavage for mature cathepsin K was predicted to be after NH 2 -A 115 based upon a sequence alignment of cathepsin S and L (14). The one or two extra N-terminal amino acids observed in the activation of cathepsin K were consistent with the N-terminal extension of several amino acids of the propeptide, which were observed in the autoprocessing of cathepsin B (16) and cathepsin S (17).
The MALDI-MS of wild-type procathepsin K indicated that approximately 3% of the mass was due to glycosylation at one or both of two potential glycosylation sites in the procathepsin K sequence, one of which was located in the propeptide domain (14). To determine if the mature enzyme was glycosylated, we analyzed mature cathepsin K and E64-inhibited mature cathepsin K by MALDI-MS and ESMS, which indicated the absence of glycosylation on the mature cathepsin K. These data, together with the Edman sequencing results, indicate that all of the carbohydrate is located on Asn 103 in the propeptide domain, although our results do not rule out the possibility of O-gycosylation elsewhere in the propeptide domain. Our results contrast with those of Brömme et al. (13), who observed no glycosylation on procathepsin K. The discrepancy may be a result of different expression conditions.
In vitro activation of procathepsin K in the absence of preactivated mature cathepsin K could be the result of limited proteolytic activity of procathepsin K under the reaction conditions, i.e. autoactivation, or by the activity of a contaminating baculovirus protease as suggested by Brömme. Autocatalytic processing mechanisms have been verified for propapain (15) and procathepsin B (16) through experiments that showed that an active site cysteine to serine mutation eliminated the proenzyme's ability to autoactivate. Analogously, we constructed a mutant in which the active site cysteine of procathepsin K, Cys 139 , was mutated to serine. We reasoned that such a mutant should lack the ability to autoprocess, and any processing observed under the activation conditions in the absence of mature cathepsin K would have to be due to a contaminating protease. The additional mutation at residue 163, serine to alanine, was incorporated to remove the potential glycosylation site in the mature enzyme.
The [Ser 139 ,Ala 163 ]procathepsin K did not process to the mature form at pH 4, but did undergo limited proteolytic cleavage when 1% wild-type mature cathepsin K was included in the reaction mixture. The peptide intermediates generated in this reaction match exactly the peptide intermediates observed in the activation of wild-type procathepsin K (Fig. 5). These results clearly demonstrated that mature cathepsin K could proteolytically process mutant procathepsin K and gives strong evidence in support of a trans-autoactivation process.
Additionally, the intermediates formed during the reaction had the ability to inhibit the processing of the mutant enzyme by decreasing the activity of the wild-type seed 15-fold during the first 2 h of the reaction (data not shown). Inhibition of cysteine proteases by their respective propeptides has been previously observed for cathepsin L (18), papain and papaya protease IV (19), and cathepsin B (20). Evidently, the inhibitory intermediates formed during activation of cathepsin K are proteolytically degraded during the activation of wild-type enzyme due to the concomitant generation of molar equivalents of mature, active enzyme. An active enzyme is not generated in the processing of [Ser 139 ,Ala 163 ]procathepsin K; hence, the reaction stops.
To see if the mutant could be completely processed to mature cathepsin K, we activated the mutant in the presence of one equivalent of wild-type pro-cathepsin K. Therefore, active enzyme would be formed in concert with the inhibitory intermediates. This reaction successfully converted the mixture of procathepsin K to mature cathepsin in a nearly quantitative conversion (Fig. 3C). Since mature cathepsin K is a relatively nonspecific protease, which would be expected to proteolyze an unfolded mutant protein, and extensive proteolysis of mutant procathepsin K was not observed, it seems likely that the mutant procathepsin K is folded properly.
The activation of wild-type procathepsin K consists of initial cleavage at three preferred sites, namely residues Glu 19 , Ser 98 , and Glu 110 as illustrated in Fig. 5. An additional cleavage site is observed in the activation of [Ser 139 ,Ala 163 ]procathepsin K after Ala 74 . Inspection of the preferred cleavage sites revealed few trends. Cleavage occurred one or two amino acids after each and every proline residue in the propeptide, and at an additional cleavage site, after Ala 74 , where proline was not present. This led us to postulate that even though cathepsin K has a somewhat broad substrate specificity (13,14), it may prefer sites that possess secondary structural features imparted by a nearby proline residue. A preference for sites containing proline would be conducive for activity with one of its natural substrate, type I collagen, which is proline-rich.
The unique site not containing proline lies within a putative consensus motif, Gly-X-Asn-X-Phe-X-Asp, (Fig. 5) found in papain, carcicain, and papaya protease IV, which was postulated to be the site of initial cleavage in the pH-dependent autoactivation of papain (21). The protonation of the aspartic acid residue was postulated to be a trigger for activation. Although the authors could not identify the site of cleavage they speculated it to be between X and Asp, since this would place the preferred phenylalanine residue in the S2 pocket. In the case of cathepsin K, leucine is a preferred residue for P2 (12,14), which predicts cleavage after Ala 74 , which is consistent with the observed cleavage within this motif in cathepsin K. More information about the structures around the proline residues and the putative consensus motif and their mechanistic implications in cathepsin K autoactivation await the determination of the three dimensional x-ray structure of procathepsin K.
If autoprocessing were to occur in vivo, procathepsin K would have to be able to initiate its activation without the aid of mature cathepsin K under the conditions in the bone resorption pit. The spontaneous activation of procathepsin K (Figs. 1 and  2), and the proteolytic cleavage of Z-Phe-Arg-AMC by procathepsin K demonstrates that such a process is possible.
In summary, the in vitro activation of cathepsin K is autocatalytic and does not require a different protease. We postulate a similar mechanism is occurring in vivo. Once procathepsin K is secreted into the resorption pit, it undergoes a conformational change induced by the lower pH, which unmasks the active site and makes the propeptide more vulnerable to endoproteolysis. The ensuing activation of procathepsin K in the resorption pit would be accelerated by the action of catalytic amounts of the newly formed mature cathepsin K.
Once the propeptide is fragmented at the preferred Pro-X-X sites, it is degraded further by endoproteolysis at less preferred sites, resulting in fully active mature cathepsin K.