Pycnodysostosis (Pycno) is an autosomal recessive osteosclerotic skeletal dysplasia that is caused by the markedly deficient activity of cathepsin K. This lysosomal cysteine protease has substantial collagenase activity, is present at high levels in osteoclasts, and is secreted into the subosteoclastic space where bone matrix is degraded. In vitro studies revealed that mutant cathepsin K proteins causing Pycno did not degrade type I collagen, the protein that constitutes 95% of organic bone matrix. To determine the in vivo effects of cathepsin K mutations on bone metabolism in general and osteoclast-mediated bone resorption specifically, several bone metabolism markers were assayed in serum and urine from seven Pycno patients. Two markers of bone synthesis, type I collagen carboxy-terminal propeptide and osteocalcin, were normal in all Pycno patients. Tartrate-resistent acid phosphatase, an osteoclast marker, was also normal in these patients. Two markers that detect type I collagen telopeptide cross-links from the N and C termini, NTX and CTX, respectively, were low in Pycno. A third marker which detects a more proximal portion of the C terminus of type I collagen in serum, ICTP, was elevated in Pycno, a seemingly paradoxical result. The finding of decreased osteoclast-mediated type I collagen degradation as well as the use of alternative collagen cleavage sites by other proteases, and the accumulation of larger C-terminal fragments containing the ICTP epitope, established a unique biochemical phenotype for Pycno.
Pycnodysostosis (Pycno), an autosomal recessive sclerosing skeletal dysplasia, recently was shown by a positional cloning strategy to result from the deficient activity of the lysosomal cysteine protease, cathepsin K (EC 220.127.116.11).(1) The disease is characterized by reduced stature, osteosclerosis, acro-osteolysis of the distal phalanges, frequent fractures, clavicular dysplasia, and skull deformities with delayed suture closure.(2,3) The fact that Pycno resulted from deficient cathepsin K activity proved the important physiological role of this lysosomal cysteine protease in bone matrix degradation and implied that inhibition of cathepsin K activity might be therapeutic for bone diseases characterized by excessive bone degradation, such as osteoporosis and certain forms of arthritis.
The cathepsin K gene, which was cloned originally from rabbit osteoclasts(4) and subsequently from several human tissues,(5–8) was highly expressed in osteoclasts, the site of Pycno pathology.(9) Immunohistochemical and fluorescence microscopic studies detected cathepsin K protein in intracellular vacuoles which were fused with the ruffled border as well as in the subosteoclastic space,(10,11) the putative site of organic bone matrix degradation during bone resorption. Cathepsin K gene expression and protein were not detected in other bone cells such as osteoblasts in mature or developing bones.(10,12,13) Biochemical characterization of purified recombinant human cathepsin K revealed that the mature enzyme was a monomeric protein with an apparent molecular mass of about 29 kDa.(14) The enzyme had a broad bell-shaped pH activity profile, with an optimum at 6.1, and strong collagenolytic, elastase, and gelatinase activities, exceeding those of cathepsins S or L.
Among the cathepsin K mutations that have been identified in Pycno patients, all appeared to obliterate its enzymatic function.(1,15) Several of the genetic lesions were nonsense mutations, predicting termination of the protein prior to the critical cysteine or histidine residues in the active cleft which are required for proteolysis. A base substitution obliterated the stop codon and extended the protein by 19 residues at its C terminus, but was not immunologically detectable when expressed in vitro.(1) More recently, expression of five Pycno missense mutations in yeast revealed that four mutant cathepsin K proteins lacked all protease activity, while the fifth, Y212C, retained protease activity but had no detectable collagenase activity.(15) Thus, all cathepsin K mutations identified to date eliminated activity of the enzyme toward type I collagen.
While the impact of Pycno mutations on cathepsin K function has been studied in vitro, the specific effects of cathepsin K deficiency on organic bone matrix degradation in vivo have not been elucidated. To gain insights into those effects, markers of bone metabolism were assayed in serum and urine from Pycno patients. In this communication, we report significant decreases in the quantities of the cross-linked N- and C-telopeptides of type I collagen (NTX and CTX, respectively) in urine, and the paradoxical increase of the more proximal portion of the C terminus of type I collagen in the serum (ICTP) of Pycno patients. The identification of these unique bone metabolic abnormalities established the first biochemical phenotype for this lysosomal disorder, providing insights into the normal degradation of type I collagen by cathepsin K.
MATERIALS AND METHODS
Patients and specimens
Blood and urine samples were obtained with informed consent from patients with Pycno, aged 6–21 years, all of who met the standard clinical criteria for this skeletal dysplasia.(16) For those patients whose cathepsin K defects had not be previously characterized, mutations were determined using exon and intron boundary amplification from genomic DNA and subsequent DNA sequencing on an ABI377 Sequencer (Perkin-Elmer Corp., Norwalk, CT, U.S.A.) as previously described.(15) Putative mutations were confirmed using restriction assays on PCR products amplified from genomic DNA; the possibility that these changes were polymorphisms was eliminated by documenting their absence in at least 100 normal alleles. Serum and urine samples were stored at –80°C until the various assays were performed.
Assays of bone metabolism markers
Bone metabolism markers that were determined included: osteocalcin (OC), the carboxy-terminal propeptide of type I procollagen (PICP), tartrate-resistant acid phosphatase (TRAP), ICTP, NTX, CTX, as well as urine pyridinoline (Pyr), and urine deoxypyridinoline (Dpyr). OC, PICP, TRAP, Pyr, and Dpyr were determined by the Nichols Institute (San Juan Capistrano, CA, U.S.A.) for the New York patients and by SRL Inc. laboratory (Tokyo, Japan) for the Japanese patients. OC was measured either by an immunoradiometric assay that detects both intact OC and the 1–43 OC fragment (Nichols Institute) or by a sandwich immunoassay that detected intact OC (SRL, Inc.).(17,18) PICP was measured in serum using a radioimmunoassay (Orion Diagnostica, Espoo, Finland). TRAP was measured within 1 week of blood drawing by enzymatic methods using either α-naphthyl phosphate (Nichols Institute) or p-nitrophenyl phosphate (SRL, Inc.) as substrate.(19,20) Pyr and Dpyr were assayed on morning urine void samples using high-performance liquid chromatography (HPLC). ICTP values were assayed from serum in the Renal Metabolism Laboratory at Children's Memorial Hospital (Chicago, IL, U.S.A.; courtesy of Dr. C.B. Langman) and at SRL using an radioimmunoassay produced by Orion Diagnostica. NTX values for the three New York patients as well as a large cohort of Israeli Arab Pycno patients were determined at the University of Washington (Seattle, WA, U.S.A.) as described previously,(21) while those for the Japanese patients were assayed by Mochida Pharmaceutical Co., Ltd. (Tokyo, Japan). Urine excretion of CTX was assayed at Fuji Revio, Inc. (Tokyo, Japan) (enzyme-linked immunosorbent assay [ELISA]; Osteometer A/S, Copenhagen, Denmark). NTX and CTX values were normalized to urine creatinine concentrations determined by the Jaffe procedure using a commercially available kit (Sigma Diagnostics, St. Louis, MO, U.S.A.) or the standard colorimetric procedure.
The bone marker values were compared with age-appropriate normative values available in the laboratories performing the assays or, when unavailable, from the literature. Statistical comparisons for NTX values for the Pycno patients, and normal individuals were made using the unpaired two-tailed Student's t-test with significance level set at p ≤ 0.05.
Among the seven Pycno patients who were studied with several markers of bone metabolism, sequencing of the seven coding exons of cathepsin K revealed two novel mutations (L9P and W298R) and three previously reported lesions (Fig. 1).(15) The genotypes of four Japanese patients included homoallelism for L9P in one patient, homoallelism for A277 V in two patients, and heteroallelism for L9P and W298R in one patient. The genotypes of three patients studied in New York were heteroallelism for R241X and Y212C in a Spanish patient and homozygosity for R312G in two siblings of Honduran descent.(15) For the larger cohort of Pycno patients for whom only NTX concentrations were measured, mutation analysis revealed 16 Israeli Arab patients who were homoallelic for X330 W,(1) a Canadian patient homoallelic for R241X, a previously reported Mexican-American boy with heteroallelism for G146R and R241X,(1) and two Swiss siblings homoallelic for A311P, a novel mutation affecting the mature enzyme.
To assess type I collagen synthesis rates and osteoblast activity, serum PICP and OC levels were determined. The values for both markers were within their respective normal ranges for six of seven Pycno patients, the sole exception being a 12-year-old girl of Honduran descent who had elevated levels (Table 1). These results were consistent with the view that the osteosclerosis observed in Pycno patients arises solely from inadequate bone resorption by osteoclasts.
Table Table 1. Markers of Bone Metabolism in Pycnodysostosis Patients
To document the effects of the reduced bone resorption in Pycno, serum TRAP activity and ICTP were assayed, and the urine levels of Pyr, Dpyr, NTX, and CTX were determined. As shown in Table 1, the TRAP activities were normal in all seven Pycno patients, suggesting normal osteoclast numbers and differentiation. Urine Pyr and Dpyr levels were normal in four of the five patients studied; the fifth patient, a 6-year old Japanese Pycno patient had low levels of both. In contrast, NTX concentrations were below the normal range for five of six pediatric Pycno patients (with most being strikingly low), but a normal value was obtained for the sole adult Pycno patient. CTX concentrations were all extremely low in the four patients studied. Finally, ICTP values were greater than twice the upper level of normal in six Pycno patients. Thus, the NTX and CTX concentrations were low in Pycno, while ICTP was elevated. Urine Pyr and Dpyr were normal in the Pycno patients.
To document further the effects of cathepsin K deficiency on bone matrix degradation, the NTX assay was performed on a larger cohort which primarily included pediatric and adult Pycno patients from the large Israeli Arab family who participated in the positional cloning studies of this disorder.(1,22,23) As shown in Table 2, the mean NTX concentration in 17 affected children (aged 5–16 years) was significantly lower than the comparable mean value in 9 age-matched normal individuals. Similarly, the mean NTX concentration for six adult Pycno patients (aged 18–59 years) was significantly decreased compared with normal adults. Of note, there was some overlap in the NTX concentrations of the Pycno patients with the lower end of the age-appropriate normal ranges (data not shown), which may be attributable in part to the broad width of those ranges. Nonetheless, average NTX concentrations were decreased significantly in the pediatric and adult Pycno patients, documenting the role of cathepsin K in bone matrix degradation by osteoclasts during the relatively more active phase of linear growth as well as for baseline bone turnover in the mature skeleton.
Table Table 2. NTX Values in Pediatric and Adult Pycnodysostosis Patients
The measurement of several noninvasive markers of bone metabolism demonstrated apparently normal osteoblastic activity but deranged bone matrix degradation in patients with Pycno. The urine NTX concentrations, normalized to urine creatinine, were decreased in growing children and adults with Pycno, indicating reduced bone resorption. The mouse monoclonal antibodies used for the NTX ELISA recognize the N-telopeptide–to-helix pyridinoline cross-linking site of the α2 chain of human type I collagen, the major protein in organic bone matrix (Fig. 2).(21) Epitope recognition is conformation dependent (i.e., the peptide must be trivalent), and the antibody does not bind to comparable portions of type II collagen. In vitro studies have documented that cathepsin K cleaves type I collagen in the N-telopeptide as well as near the N terminus of the helical region of the α1 chain (Fig. 2).(24) This gives cathepsin K potent activity for releasing the NTX epitope from type I collagen, an activity that is significantly greater than that observed with cathepsin B, L, or S.(24,25) The demonstration of decreased NTX concentrations among Pycno patients, both adults and children, was consistent with a decreased rate of type I collagen degradation from bone secondary to the cathepsin K deficiency in osteoclasts, and confirmed the sensitivity of this assay for detecting decreases in bone resorption rates as has been documented with other metabolic bone disorders.(26,27)
The normalized urine CTX concentration, which measured the C-telopeptide of type I collagen,(28) also was low in Pycno patients. This assay was performed with rabbit polyclonal antibodies that were raised against a synthetic peptide of the eight C-terminal residues of the type I collagen α1 chain, including the lysine residue that is part of the C-terminal cross-link (Fig. 2). It is believed that the CTX assay detects both cross-linked and non–cross-linked C-terminal telopeptides. As with the NTX assay, in vitro studies have shown that cathepsin K activity releases the CTX epitope from type I collagen.(24) The CTX results were consistent with those obtained for NTX, providing further evidence that type I collagen degradation in bone is diminished in Pycno. These findings also were consistent with the ultrastructural studies of bone from a Pycno patient which documented the presence of incompletely degraded collagen fibrils in vacuoles within osteoclasts.(9) While vacuolar transcytosis from the apical to the basolateral surface of osteoclasts has been shown to be the normal pathway for release of resorbed bone matrix proteins,(29,30) the observation of identifiable collagen fibrils within the Pycno osteoclasts is pathological.
Unlike patients with osteopetrosis (Albers–Schonberg disease) who appear to have absent or thoroughly ineffective osteoclasts, Pycno patients possess osteoclasts that have some degradative capacity. Cathepsin K–deficient osteoclasts appear to differentiate normally from their monocyte-macrophage precursors and possess an intact mechanism for demineralization.(9) The appendicular skeletal bones of young Pycno patients grow longer and wider, albeit more slowly than normal, and fracture healing generally proceeds normally. This bone-resorptive capacity occurs despite the fact that all Pycno cathepsin K mutations studied to date have no residual activity against type I collagen. Thus, cathepsin K–deficient osteoclasts are capable of degrading bone matrix, presumably with other proteases, either enzymes that normally contribute to bone resorption or others that are recruited for this purpose when cathepsin K is deficient. The aforementioned ultrastructural study of Pycno bone indicated that bone matrix resorption resulted from osteoclast-produced proteases, and not from other cell types, a finding substantiated by in vitro studies of Pycno osteoclast-like cells generated from circulating precursors (D. Brömme, D. Dempster, and B.D. Gelb, unpublished results).
In this context, the paradoxical finding that serum ICTP values were elevated in Pycno patients, while the NTX and CTX concentrations were reduced, provides insights into the function of cathepsin K in bone matrix degradation by osteoclasts. The ICTP assay uses polyclonal antiserum which was raised against cross-linked C-terminal fragments of type I collagen that had been digested with bacterial collagenase or trypsin.(31) The assay detects type I collagen cross-links that are trivalent and include a hydrophobic phenylalanine-rich domain of the two α1-chains located between the helix and the cross-link near the C terminus (Fig. 2).(32) The assay provides a sensitive indication of high or low bone turnover for some bone diseases,(33) but has not proven reliable in detecting the effects of treatments that affect bone density such as bisphosphonates for osteoporosis.(34,35) Since the collagen fragment detected by the ICTP assay has a molecular weight of around 10 kDa,(31) small enough to permit filtration at the glomerulus, significant renal dysfunction can also result in elevated serum ICTP values. Since Pycno patients have normal renal function and are not subject to excessive degradation of bone or soft tissue, the most likely explanation for the elevated serum ICTP values must be the generation of increased amounts of the C-terminal epitope detected by this assay despite an overall reduction in bone resorption. This implies that cathepsin K hydrolyzes collagen at its C terminus beyond the hydrophobic domain recognized by the ICTP assay. Alternatively, the C terminus may be liberated normally by another protease and then subsequently hydrolyzed further by cathepsin K. Isolation and amino acid sequencing of the type I collagen C-terminal fragments derived from cathepsin K–deficient osteoclasts could identify the cleavage sites, thereby providing insights into the substrate specificity of the residual or compensatory protease(s).
Urine Pyr and Dpyr concentrations were normal in nearly all of the Pycno patients, findings that were unexpected for a disorder with the decreased bone resorption by osteoclasts. Pyr and Dpyr, measured by the HPLC assay used for this study, are excreted in the urine in two forms, as free amino acids (hydroxylysylpyridinoline and lysylpyridinoline, respectively) and as small peptides containing those amino acids.(36) Pyr and Dpyr derive from the post-translational hydroxylation of lysine residues at sites near the N and C termini of collagen that become cross-linked. In urine, ∼67% of Pyr and Dpyr are in the form of small peptides while the remainder is free. In vitro studies have documented that osteoclasts release Pyr and Dpyr only as small peptides(37); degradation to the free forms is believed to occur in the liver and, possibly, the kidney.(38) Although collagens from several tissues have modified cross-linked lysine residues, urine Pyr and Dpyr are believed to derive principally from bone collagen. This is supported by the observation that the Pyr-to-Dpyr ratio in human bone is 3.5:1, similar to that found in urine and markedly different from collagens in other tissues which have considerably lower amounts of Dpyr (ratios > 9:1).(39) In Pycno patients, the Pyr-to-Dpyr ratio averaged 4.56, suggesting that urine Pyr and Dpyr were reflecting bone collagen resorption. Since both the NTX and CTX values were markedly reduced for these patients, these findings indicated that excretion of the free forms of Pyr and Dpyr constituted a larger percentage of the total. Consistent with the ICTP results, this suggests that cathepsin K–deficient osteoclasts degrade bone collagen into abnormally large fragments that are not filtered at the glomerulus but rather undergo extensive proteolysis in the liver and kidney. It is also inferred that the degree of impairment of bone matrix degradation in Pycno is not as severe as the NTX and CTX values implied, a view compatible with the observed skeletal phenotype.
In summary, the in vivo effects of cathepsin K deficiency on bone metabolism were examined in the osteosclerotic bone dysplasia, Pycno. Assays for bone formation, OC and PICP, indicated normal osteoblast activity in this disorder. Two urine markers of bone resorption, NTX and CTX, had decreased concentrations, consistent with diminished osteoclast-mediated bone matrix degradation due to cathepsin K deficiency. Another marker of bone resorption, serum ICTP, was increased, suggesting that the type I collagen degradation in bone resulted from the actions of other protease(s) using alternative cleavage sites near the C terminus. By characterizing these bone metabolism markers of type I collagen degradation in patients with Pycno, the decreased urine NTX and CTX and increased serum ICTP concentrations defined the biochemical phenotype of cathepsin K deficiency.
The authors thank the patients for their participation, Dr. C.B. Langman (Children's Memorial Hospital, Chicago, IL, U.S.A.) for performing ICTP assays, Mr. T. Matsuura (Biosciences Research Laboratory, Mochida, Japan) and Mr. N. Suzuki (Fuji Revio Inc., Hachiouji, Japan) for performing urine NTX and CTX assays, as well as Drs. Kumiko Araki (Kochi Medical School, Kochi, Japan) and Masaaki Yoshimoto (Children's Clinic Yoshimoto, Nagasaki, Japan) for enrolling patients. This work was supported in part by National Institutes of Health research grants (AR44231 to B.D.G., DK34045 to R.J.D., and AR37318 to D.E.E.), a grant from the National Center for Research Resources for the Mount Sinai General Clinical Research Center (RR00071), a grant for the Mount Sinai Child Health Research Center (HD28822), a grant from Ostex International, Inc. to D.E.E., and a grant from the Foundation for the Growth Science in Japan to Y.N.