|Home | About | Journals | Submit | Contact Us | Français|
Mineralization of growth plate cartilage is a critical event during endochondral bone formation, which allows replacement of cartilage by bone. Ankylosis protein (Ank), which transports intracellular inorganic pyrophosphate (PPi) to the extracellular milieu, is expressed by hypertrophic and, especially highly, by terminally differentiated mineralizing growth plate chondrocytes. Blocking Ank transport activity or ank expression in terminally differentiated mineralizing growth plate chondrocytes led to increases of intra- and extracellular PPi concentrations, decreases of alkaline phosphatase (APase) expression and activity, and inhibition of mineralization, whereas treatment of these cells with the APase inhibitor levamisole led to an increase of extracellular PPi concentration and inhibition of mineralization. Ank-overexpressing hypertrophic nonmineralizing growth plate chondrocytes showed decreased intra- and extracellular PPi levels; increased mineralization-related gene expression of APase, type I collagen, and osteocalcin; increased APase activity; and mineralization. Treatment of Ank-expressing growth plate chondrocytes with a phosphate transport blocker (phosphonoformic acid [PFA]) inhibited uptake of inorganic phosphate (Pi) and gene expression of the type III Na+/Pi cotransporters Pit-1 and Pit-2. Furthermore, PFA or levamisole treatment of Ank-overexpressing hypertrophic chondrocytes inhibited APase expression and activity and subsequent mineralization. In conclusion, increased Ank activity results in elevated intracellular PPi transport to the extracellular milieu, initial hydrolysis of PPi to Pi, Pi-mediated upregulation of APase gene expression and activity, further hydrolysis and removal of the mineralization inhibitor PPi, and subsequent mineralization.
Mineralization of growth plate cartilage plays a crucial role during endochondral bone formation and allows the cartilage elements to be replaced by bone. Mineralization is restricted to a few layers of chondrocytes close to the chondro-osseous junction. Because of the tight restriction of mineralization to certain areas, the process has to be highly regulated. Uncontrolled (pathological) mineralization can have severe consequences. For example, mineral deposition is detected in articular cartilage of up to 40% of elderly people. These mineral deposits are calcium pyrophosphate dihydrate (CPPD), consisting of calcium and inorganic pyrophosphate (PPi), and/or basic calcium phosphate (BCP) crystals, consisting of calcium and inorganic phosphate (Pi) (36). These mineral deposits can lead to acute joint pain and eventually to cartilage destruction. Furthermore, mineral crystals have been shown to stimulate proliferation of synovial fibroblasts and expression of matrix metalloproteinase in these cells (37).
Recently, the progressive ankylosis gene (ank) was shown to encode a multiple-pass transmembrane protein that regulates PPi transport from the cytoplasm to the extracellular milieu (11, 30). A mouse mutation of ank creating a nonfunctional Ank protein led to a marked increase of intracellular PPi concentration and a decrease of extracellular PPi levels in homozygous mice (11). The homozygous mice showed severe joint mineralization and arthritis (11). Studies investigating the expression of ank showed ank mRNA expression in the developing articular cartilage of various joints, in hypertrophic chondrocytes of the mouse growth plate (where physiological mineralization occurs), in cells of the perichondrium/periosteum, and in tendons (11, 41). Low ank mRNA expression was detected in normal healthy human articular cartilage, whereas ank mRNA was robustly expressed in all zones of osteoarthritic human cartilage, where pathological mineralization occurs (10, 17). Detailed histological analysis of various skeletal and joint tissues revealed three abnormal processes in these homozygous ank mutant mice: (i) increased calcification in the joints, (ii) increased proliferation of synovial cells and osteophyte formation, and (iii) increased degeneration starting in tendons and ligaments and later in articular cartilage (39, 42). However, none of these studies investigated the effects of this mutation during early bone development.
Previous findings showing that PPi is a potential inhibitor of mineralization that suppresses hydroxyapatite crystal propagation (5, 44) are consistent with the increased joint mineralization in the homozygous ank mice, and these results suggest that BCP crystal formation occurs in these mice because of the lack of a sufficient concentration of the mineralization inhibitor PPi. These notions were supported by previous studies showing that PPi directly inhibits the capacity of growth plate chondrocytes and osteoblasts to deposit BCP crystals in the extracellular matrix (15, 44). However, these suggestions are contradictory to findings of high expression of ank mRNA in areas where BCP crystal deposition occurs (e.g., hypertrophic growth plate and osteoarthritic cartilage) (10, 17, 41). In contrast, PPi supersaturation of the matrix in articular hyaline cartilage, meniscal fibrocartilage, and certain tendons and ligaments can lead to CPPD crystal deposition (9, 24, 44).
Mutations in human ank have been shown to cause CPPD crystal deposition in articular cartilage. In contrast to the mouse mutation, these mutations lead to more Ank activity, resulting in increased extracellular PPi concentration and eventually CPPD crystal deposition in articular cartilage (31). Other mutations in human ank lead to craniometaphyseal dysplasia, a disease characterized by an overgrowth and excessive BCP mineral formation of craniofacial bones (30, 34). These studies clearly establish an important function of Ank in the regulation of physiological and pathological mineralization. Recent studies showing high expression of Ank in osteoarthritic cartilage, where BCP crystals also were found, suggest that high expression of Ank or Ank PPi transport properties activating mutations can on one hand lead to CPPD crystal formation and on the other hand lead to BCP crystal formation (10, 17). Therefore, the understanding of how Ank regulates physiological and pathological mineralization processes is of great importance.
Physiological mineralization of growth plate cartilage and bone is accompanied by high alkaline phosphatase (APase) activities. APase is an enzyme that hydrolyzes phospho-compounds, including PPi, and generates Pi (1). APase deficiency (hypophosphatasia) leads to defective bone mineralization (osteomalacia) and increased extracellular PPi concentrations (4, 8, 16, 29, 48). APase is enriched on the outer membrane surface of matrix vesicles (1). Matrix vesicles are released from the plasma membrane of mineralization-competent skeletal cells, and they have the critical role of initiating the mineralization process (1, 19). Therefore one can speculate that the major function of matrix vesicle- and cell surface-attached APase might be to remove a potent inhibitor of mineralization (PPi) and to provide Pi required for BCP crystal formation. Interestingly, Pi is not only required for BCP mineral formation but is also a modulator of cell differentiation and gene expression. For example, Pi has been shown to induce apoptosis in chondrocytes and osteoblasts, and it regulates expression of a variety of genes (2, 3, 23, 26).
A previous study has demonstrated the expression of ank mRNA in a subset of hypertrophic chondrocytes close to the primary and secondary ossification centers during murine development, suggesting a possible role of Ank in bone formation and physiological mineralization (41). However, little is known about the role of Ank in regulating physiological mineralization of growth plate cartilage and bone. Based on our and other findings that Ank is highly expressed in terminally differentiated mineralizing growth plate chondrocytes in vitro and in vivo (41), we hypothesized that a rapid increase of the extracellular PPi concentration and APase activity regulated by Ank is required for promotion of the mineralization process in growth plate cartilage. To test this hypothesis, we determined how suppressing ank expression by using small interfering RNA (siRNA) or blocking Ank PPi transport activities with probenecid, which inhibited PPi transport activity in articular chondrocytes (35), affects mineralization of terminally differentiated mineralizing growth plate chondrocytes. In addition, we overexpressed Ank in hypertrophic nonmineralizing growth plate chondrocytes and determined the effect of Ank overexpression in these cells on their terminal differentiation and mineralization.
Full-length chicken ank cDNA was amplified from total chicken brain RNA by nested reverse transcription-PCR (RT-PCR). The primers were designed according to homologous regions of the mouse and human ank genes by using Primers Express software (Applied Biosystems, Foster City, Calif.). Total RNA was isolated from embryonic (19-day) chicken brain by using an RNeasy minikit (Qiagen, Stanford, Calif.). One microgram of total RNA was reverse transcribed by using an Omniscript reverse transcription kit (Qiagen). The nested PCRs were performed by using a PCR master mix kit (Qiagen) with 30 cycles of 94°C for 30 s, 45 or 57°C for 1 min, and 72°C for 2 min with the following nested PCR primers: forward external primer, 5′-TGAGTGTGGGGTCAGCCCAC-3′; reverse external primer, 5′-ATCCCCAGTATGCTAGAGAAT-3′; forward internal primer, 5′-ATGGTGAAATTCCCGGCGCTC-3′; and reverse internal primer, 5′-TTACTCATTTTCTTCTCTCAT-3′. The final PCR product amplified with the pair of internal primers was cloned into the pCR2.1-TOPO plasmid vector by using a TOPO TA cloning kit (Invitrogen, Carlsbad, Calif.) and was sequenced.
Chondrocytes were isolated from the hypertrophic zone of 19-day embryonic chick tibial growth plate cartilage as described previously (19). Briefly, the sliced hypertrophic zone of growth plate cartilage was digested with 0.25% trypsin and 0.05% collagenase for 5 h at 37°C. Cells were plated at a density of 3 × 106 into 100-mm-diameter tissue culture dishes and grown in monolayer cultures in Dulbecco's modified Eagle's medium (Life Technologies, Gaithersburg, Md.) containing 5% fetal calf serum (HyClone, Logan, Utah), 2 mM l-glutamine (Life Technologies), and 50 U of penicillin and streptomycin (Life Technologies) per ml (complete medium). After cultures reached confluence, chondrocytes were cultured in the presence of 1.5 mM phosphate (untreated) and in the absence or presence of (i) 35 nM retinoic acid (RA) (Sigma Chemical Co., St. Louis, Mo.), (ii) 2.5 mM probenecid (Sigma Chemical Co.), (iii) 35 nM RA and 2.5 mM probenecid, or (iv) 35 nM RA and 0.8 mM levamisole (Sigma Chemical Co.). For Pi uptake studies, cells were cultured in the presence of 1.5 mM β-glycerophosphate (GP) (Sigma Chemical Co.) and in the absence or presence of (i) 35 nM RA or (ii) 35 nM RA and 1 mM phosphonoformic acid (PFA) (Sigma Chemical Co.) Intra- and extracellular PPi levels and the degree of mineralization in these cultures were determined after 6 days of treatment. Pi uptake and APase gene expression and activity were determined after 3 days of treatment.
Primary chicken embryonic fibroblasts were obtained from the dorsal region of 10-day embryonic chickens by dissociating pieces of skin in 0.3% collagenase for 1 h at 37°C. Cells were plated at 2 × 106/60-mm-diameter culture dish and cultured in complete medium until 90% confluence.
We used a Silencer siRNA construction kit from Ambion, Inc. (Austin, Tex.) to synthesize siRNA. Two pairs of oligonucleotides encoding the desired sense and antisense siRNA strands were designed according to the cloned chicken ank sequence by using a computer program (Ambion, Inc.). Oligonucleotides were designed to include an 8-base sequence complementary to the 5′ end of the T7 promoter primer included in the kit. The first pair of oligonucleotides has the sense sequence 5′-ACAGTAAGAGAGACAGGACCCCTGTCTC-3′ and the antisense sequence 5′-AAGGTCCTGTCTCTCTTACTGCCTGTCTC-3′. The second pair of oligonucleotides has the sense sequence 5′-AAACACAAGTACAGTTTCCTGCCTGTCTC-3′ and the antisense sequence 5′-AACAGGAAACTGTACTTGTGTCCTGTCTC-3′. The oligonucleotides were annealed to the T7 promoter primer, and a fill-in reaction with Klenow fragment generated a double-stranded template that was ready for use in the in vitro transcription reaction with T7 RNA polymerase. After transcription, the reaction products were combined to permit annealing of the two siRNA strands. The siRNA preparation were then treated with DNase to remove template, followed by RNase to polish the ends of the double-stranded RNA, and then column purified. The siRNAs were first tested in 10-day chicken embryonic dorsal fibroblasts to select the most effective one for use in growth plate chondrocytes. Primary hypertrophic growth plate chondrocytes were first treated for 1 day with RA, followed by transfection of 200 nM siRNA by using the Lipofectamine 2000 transfection reagent according to the protocol of the manufacturer (Invitrogen). After transfection, RA treatment was continued for a total of 6 days. Four days after transfection, Ank protein levels were determined by immunoblotting.
Full-length chicken ank cDNA was first cloned into an adaptor vector, SLAX-myc, which contained a 10-amino-acid epitope of human c-Myc tag fused to the COOH-terminal end of the recombinant protein and then subcloned into a replication-competent, nontransforming Rous sarcoma virus-based expression vector (RCASBP) (12, 32). The plasmid constructs and RCASBP containing no insert were used to transfect chicken embryonic dorsal fibroblasts by using the Lipofectamine 2000 transfection reagent according to the protocol of the manufacturer (Invitrogen) to produce high-titer retroviral stocks. High-titer virus stocks in a small volume (5 × 106 CFU/106 cells in less than 1 ml of medium) were incubated with chondrocytes isolated from the hypertrophic zone of 19-day embryonic chicken growth plate cartilage for 4 h. Thereafter, cells were cultured in Dulbecco's modified Eagle's medium containing 5% fetal calf serum until ~90% of chondrocytes were infected (~1 week of culture). The transfection efficiency was assessed by immunostaining with fluorescein-labeled antibodies specific for the c-Myc tag (Covance Research Products, Philadelphia, Pa.).
Total RNA was isolated from untreated and RA-treated hypertrophic growth plate chondrocytes from day 1 to 3 as described above. Ten micrograms of total RNA was denatured, fractionated on 1% agarose gels, and transferred to Hybond-N membranes. Blots were hybridized with a 0.3-kb cDNA ank probe, which was labeled with a BrightStar psoralen-biotin nonisotopic labeling kit (Ambion, Inc.). Blots were hybridized and washed by using Northern Max and BrightStar BioDetect kits (Ambion, Inc.). Blots were exposed to Kodak radiographic films. Blots were also stained with 0.04% methylene blue to verify that each sample had been transferred efficiently.
Total RNA was isolated from untreated, RA-treated, RA-treated and ank-specific siRNA-infected, RCASBP-infected, and ank/RCAS-infected chondrocytes and from ank/RCAS-infected and levamisole-treated and ank/RCASBP-infected and PFA-treated growth plate chondrocytes by using the RNeasy minikit (Qiagen). Gene expression of Pit-1 and Pit-2 was analyzed by RT-PCR, while gene expression of ank, APase, type I and II collagen, and osteocalcin was quantified by real-time PCR as described previously (46). Briefly, 1 μg of total RNA was reverse transcribed by using an Omniscript RT kit (Qiagen). PCR was then performed with Pit-1 and Pit-2 primers generated from the mouse sequences. The primer sequences were as follows: Pit-1 forward primer, 5′-GATGAAATGGAGACGCTGAC-3′; Pit-1 reverse primer, 5′-AGGAACTGGAAGAGAGAAGGGA-3′; Pit-2 forward primer, 5′-GGCTTCCTATGGACGGGCAC-3′; and Pit-2 reverse primer, 5′-CAGCCACTGCGTTGCAGTAG-3′. PCR was performed with an annealing temperature of 51°C, and the number of cycles was adjusted to 30. Actin was amplified at the same time and was used as an internal control.
A 1:100 dilution of the resulting cDNA was used as the template to quantitate the relative content of mRNA by real-time PCR (ABI PRISM 7700 sequence detection system; Applied Biosystems) with the respective primers and SYBR Green. The following primers were used for real-time PCR analysis: APase forward primer, 5′-CCCTGACATCGAGGTGATCCT-3′; APase reverse primer, 5′-GGTACTCCACATCGCTGGTGTT-3′; collagen type I forward primer, 5′-CAGCCGCTTCACCTACAGC-3′; collagen type I reverse primer, 5′-TTTTGTATTCAATCACTGTCTTGCC-3′; collagen type II forward primer, 5′-GGCCCTAGCAGGTTCACGTACA-3′; collagen type II reverse primer, 5′-CGATAACAGTCTTGCCCCACTT-3′; osteocalcin forward primer, 5′-TCGCGGCGCTGCTCACATTCA-3′; and osteocalcin reverse primer, 5′-TGGCGGTGGGAGATGAAGGCTTTA-3′. RT-PCRs were performed with a TaqMan PCR Master Mix kit (Applied Biosystems), with 40 cycles of 50°C for 2 min, 95°C for 10 min, 95°C for 15 s, and 60°C for 1 min. The 18S RNA was amplified at the same time and used as an internal control. The cycle threshold (Ct) values for 18S RNA and the samples were measured and calculated by computer software. Relative transcript levels were calculated as x = 2−ΔΔCt, in which ΔΔCt = ΔE − ΔC, ΔE = Ctexp − Ct18S, and ΔC = Ctctl − Ct18S.
Intra- and extracellular PPi concentrations in growth plate chondrocytes were determined by using a coupled enzymatic and fluorometric assay as described previously (22). After 6 days of treatment or culture of infected chondrocytes, the medium was collected for subsequent analysis of the extracellular PPi concentration. For analysis of the intracellular PPi concentration, chondrocytes were washed with ice-cold phosphate-buffered saline (PBS) (pH 7.4) and collected. The cell pellet was resuspended thoroughly in 1 M perchloric acid-PBS solution. After centrifugation, the supernatant was neutralized to a pH of 7.0 to 8.0 by adding 5 M KOH and was subjected to the coupled enzymatic and fluorometric PPi assay. Protein content was determined by using the bicinchoninic acid protein assay. Intra- and extracellular PPi concentrations were normalized to the protein concentration. To determine the extracellular PPi concentration, the collected medium was centrifuged at 350 × g for 10 min, followed by centrifugation at 14,000 × g for 1 h at 4°C (Sorvall Discovery 90 SE ultracentrifuge and T-1270 rotor; Kendro Laboratory Products, Newtown, Conn.). The supernatant was subjected to the enzymatic and fluorometric PPi assay. Since we measured the extracellular PPi concentration in the medium, it is possible that the actual extracellular PPi concentration might be slightly higher, especially in mineralized RA-treated and Ank-overexpressing cell cultures, because some extracellular PPi might adsorb to the mineral phase and therefore might not have been released into the medium. However, it is unlikely that large amounts of the mineralization inhibitor PPi adsorbed to the forming mineral phase in RA-treated or Ank-overexpressing growth plate chondrocytes, because these cultures showed a high rate of mineralization.
The intracellular Pi concentration of growth plate chondrocytes after 3 days of treatment was measured by using the PiPer phosphate assay kit (Molecular Probes, Eugene, Oreg.). Cells were washed, and the cytoplasmic fraction was obtained by ultracentrifugation as described previously (46). Ten microliters of the cytoplasmic fraction was used, and the Pi concentration was determined according to the manufacturer's instructions.
APase activity was determined as described previously (46). APase activity was normalized to the total protein concentration.
To determine the degree of mineralization, chondrocyte cultures were stained with alizarin red S as described previously (46). Briefly, chondrocyte cultures were washed twice with PBS for 5 min, fixed with 70% ethanol for 10 min, and then stained with 0.5% alizarin red S solution (pH 4.0) for 5 min at room temperature. To quantify the staining intensity, 100 mM cetylpyridinium chloride solution was added and left for 1 h to solubilize and release calcium-bound alizarin red into solution. The absorbance of the released alizarin red S staining was measured at 570 nm with a spectrophotometer. Data are expressed as units of alizarin red S released per milligram of protein in each culture.
A recombinant region of chicken Ank (amino acids 74 to 180) containing one extracellular domain, one intracellular domain, and three transmembrane domains (30) was prepared by using the pGEX expression vector (Amersham Biosciences, Piscataway, N.J.) as described previously (18). Recombinant Ank-glutathione S-transferase fusion protein was expressed in Escherichia coli DH5αF′ and purified. Rabbits were injected with 200 μg of purified Ank-glutathione S-transferase fusion protein three times (Cocalico Biologicals Co., Denver, Pa.). Total immunoglobulin G from the antisera was obtained by affinity chromatography with a protein A-Sepharose column (Amersham Biosciences). Preimmune immunoglobulin G isolated from the same rabbits before immunization with Ank fusion protein was used as a control. The specificity of the immune immunoglobulin G for Ank was determined by enzyme-linked immunosorbent assays (not shown) and immunoblotting.
Monoclonal antibodies specific for chicken type X collagen and human APase were obtained from the Developmental Studies Hybridoma Bank (University of Iowa, Iowa City) and have been described by Schmid and Linsenmayer (40) and Lawson et al. (21). Polyclonal antibodies specific for chicken APase were obtained from Ellis E. Golub, University of Pennsylvania School of Dental Medicine (7). Mouse monoclonal antibodies specific for chicken actin were obtained from Chemicon International (Temecula, Calif.).
To determine the localization of Ank, APase, and type X collagen in the chicken growth plate and human articular cartilage, immunohistochemical analysis using antibodies specific for Ank, APase, and type X collagen was performed as described previously (33). Samples of human osteoarthritic knee cartilage from medial and lateral femoral condyles (from 56- to 85-year-old donors) were collected from patients undergoing knee arthoplasty at the time of surgery. Samples of articular cartilage without signs of osteoarthritis were obtained from human knees (34- to 59-year-old donors) within 12 to 24 h after death. The study protocol was approved by institutional board review. After fixation in 4% paraformaldehyde and decalcification with 0.2 M EDTA pH 7.4), 10-μm-thick paraffin sections from 19-day chick embryonic tibial growth plate cartilage and human articular cartilage were cut longitudinally. After deparaffinization and rehydration, sections were incubated with sheep testicular hyaluronidase (2 mg/ml) (Sigma Chemical Co.) in PBS for 30 min at 37°C. Immunostaining was performed with the Histostain-SP kit (Zymed Laboratories Inc., San Francisco, Calif.) according to the manufacturer's protocol. Briefly, after incubation with a blocking solution for 10 min at room temperature, sections were incubated with primary antibodies for 3 h at room temperature and then with biotinylated secondary antibodies for 10 min at room temperature. Sections were stained with a streptavidin-peroxidase conjugate for 10 min at room temperature and then with a solution containing diaminobenzidine (chromogen) and 0.03% hydrogen peroxide for 5 min at room temperature and were counterstained with methylene green. Control sections were incubated with nonimmune rabbit serum. Specimens were viewed and analyzed under a microscope (Olympus Optical Company, Ltd., Tokyo, Japan).
To determine the degree of Ank expression in variously treated or infected chondrocyte cultures, cells were collected and incubated in 200 μl of lysis solution (50 mM Tris-HCl [pH 8.0], 150 mM NaCl, 1 mM EDTA, 1.2% Triton X-100) on ice for 20 min as described elsewhere (6). After centrifugation, the supernatant was collected and equal amounts of protein were dissolved in 4× NuPAGE sodium dodecyl sulfate (SDS) sample buffer containing a reducing agent (Invitrogen), denatured at 70°C for 10 min, and analyzed by electrophoresis in 10% bis-Tris polyacrylamide gels. Samples were electroblotted onto nitrocellulose filters after electrophoresis. After blocking with a solution of low-fat milk protein, blotted proteins were immunostained with primary anti-Ank immunoglobulin G and then peroxidase-conjugated secondary antibody, and the signal was detected by enhanced chemiluminescence (Pierce Chemical Co., Rockford, Ill.).
Numerical data are presented as means ± standard deviations (SD) (n > 3), and statistical significance between groups was identified by using the two-tailed Student t test (P values are reported in the figure legends).
We cloned and sequenced the full-length coding sequence of the chicken ank gene. The chicken ank gene is predicted to encode a 493-amino-acid protein with an expected molecular mass of 54.5 kDa and an isoelectric point of 8.15. Protein sequence alignment showed a high degree of conservation among a wide variety of species, suggesting that the function of Ank is of major importance. For example, the chicken Ank sequence shows 95.25% identity to the human Ank sequence (Fig. (Fig.1).1). However, chicken Ank contains one amino acid residue (asparagine) more at the COOH-terminal end than do the human, mouse, and rat counterparts.
An antibody made against Ank (amino acids 74 to 180)-glutathione S-transferase fusion protein was used to examine the localization of Ank in sections of 19-day chicken embryonic tibial growth plate cartilage. Its localization was compared with the localizations of APase and type X collagen, which are considered markers for chondrocyte hypertrophy (25, 40). Positive immunostaining for Ank was detected in the zones of reserved or prearticular chondrocytes, early hypertrophic chondrocytes, and hypertrophic and terminally differentiated mineralizing chondrocytes, whereas the zone of proliferating chondrocytes showed no staining (Fig. (Fig.2).2). Immunostaining for APase was restricted to the prehypertrophic, hypertrophic, and terminally differentiated zones, whereas type X collagen staining was detected in the zone of hypertrophic and terminally differentiated chondrocytes (Fig. (Fig.2A).2A). These findings revealed a colocalization of Ank, APase, and type X collagen in the zones of hypertrophic and mineralizing growth plate cartilage. These results confirm previous findings showing ank mRNA expression in hypertrophic chondrocytes of the embryonic murine growth plate (41).
Enhanced immunostaining for Ank was also detected in human osteoarthritic cartilage compared to normal healthy human articular cartilage (Fig. (Fig.2B).2B). Immunostaining with antibodies specific for APase revealed the absence of staining for APase in normal healthy human articular cartilage (Fig. (Fig.2B)2B) but staining of chondrocytes in human osteoarthritic cartilage (Fig. (Fig.2B),2B), similar to the colocalization of Ank and APase in embryonic growth plate cartilage.
Based on the intense immunostaining for Ank in the hypertrophic and mineralizing (terminally differentiated) zones of growth plate cartilage, we determined the possible role of Ank in growth plate cartilage mineralization by using different approaches: (i) interfering with Ank PPi transport activity in terminally differentiated, mineralizing growth plate chondrocytes by using probenecid; (ii) interfering with ank expression in terminally differentiated, mineralizing growth plate chondrocytes by using siRNA; and (iii) overexpressing Ank in hypertrophic, nonmineralizing growth plate chondrocytes by using a retroviral expression vector (RCASBP). As shown in findings from our laboratory and others, treatment of growth plate chondrocytes isolated from the hypertrophic zone of 19-day embryonic chicken tibial growth plate cartilage with RA led to an induction of terminal differentiation and mineralization (13, 46). Interestingly, Northern blot analysis revealed that ank gene expression was markedly upregulated in RA-treated cultures after 1, 2, and 3 days of treatment compared with the expression levels in untreated cultures (Fig. (Fig.3A).3A). Moreover, Ank protein expression was also notably upregulated after 6 days of treatment with RA compared with the levels in untreated cells (Fig. (Fig.3B).3B). Probenecid, an anion channel blocker that has been shown to block Ank PPi transport activity (11, 35), did not affect the levels of Ank protein expression in RA-treated cultures (Fig. (Fig.3B3B).
Since probenecid is an organic anion channel blocker that may also exert other effects (35), we used siRNA specific for ank to suppress expression of Ank in terminally differentiated growth plate chondrocytes. Using ank sequence-specific siRNA, we suppressed Ank protein expression in RA-treated terminally differentiated growth plate chondrocytes by more than 70% compared to the levels in RA-treated cells (Fig. (Fig.4A).4A). The levels of Ank protein expression in siRNA-transfected and RA-treated cells were similar to the levels in untreated hypertrophic nonmineralizing chondrocytes (Fig. (Fig.4A).4A). Alternatively, we overexpressed Ank in hypertrophic growth plate chondrocytes by using a replication-competent, nontransforming Rous sarcoma virus-based expression vector (RCASBP). Hypertrophic growth plate chondrocytes infected with RCASBP containing cDNA encoding full-length ank showed markedly increased Ank protein expression compared to the levels in uninfected or RCASBP-infected cell cultures (Fig. (Fig.4B).4B). Staining of the Western blots with antibodies specific for actin was performed to demonstrate equal loading of the gel and to show that siRNA transfection or RCAS infection of growth plate chondrocytes specifically affected Ank expression and is not toxic to the cells (Fig. (Fig.44).
Analysis of the intracellular PPi concentration in these cell cultures revealed that the intracellular PPi concentration was significantly (P ≤ 0.01) reduced in RA-treated and ank/RCAS-infected growth plate chondrocytes compared with the concentration in untreated or RCAS-infected cells (Fig. (Fig.5A).5A). Blocking Ank transport activity with probenecid led to a significant (P ≤ 0.01) increase of the intracellular PPi concentration in hypertrophic chondrocyte cultures and cultures treated with RA compared with the levels in untreated and RA-treated cells (Fig. (Fig.5A).5A). Treatment with levamisole, an inhibitor of APase activity (45), had no effect on the intracellular PPi concentration in hypertrophic growth plate chondrocytes and RA-treated growth plate chondrocytes (Fig. (Fig.5A).5A). These results indicate that increased Ank activity led to a lower intracellular PPi concentration. Analysis of the extracellular PPi concentration revealed that the extracellular PPi concentration was also significantly (P ≤ 0.01) lower in RA-treated and ank/RCAS-infected cultures than in untreated or RCAS-infected cultures (Fig. (Fig.5B).5B). Levamisole treatment led to a significant (P ≤ 0.01) increase of the extracellular PPi concentration in RA-treated cultures but did not affect the extracellular PPi concentration in cells which were treated only with levamisole (Fig. (Fig.5B).5B). Interestingly, probenecid also led to an increase of extracellular PPi concentration in RA-treated cultures (Fig. (Fig.5B);5B); however, the probenecid-caused increase was markedly less than the increase caused by levamisole (Fig. (Fig.5B).5B). These findings showing that increased expression of Ank led to decreased intracellular and extracellular PPi concentrations suggest that extracellular PPi is being hydrolyzed by APase and that increased Ank PPi transport activity led to increased APase activity.
To test the hypothesis that increased Ank expression and PPi transport activity affect APase expression and activity, real-time PCR analysis of APase expression in ank-specific siRNA-transfected and RA-treated terminally differentiated growth plate chondrocytes or in Ank-overexpressing nonmineralizing hypertrophic growth plate chondrocytes was performed. Suppression of Ank expression in RA-treated growth plate chondrocytes by using ank-specific siRNA led to a notable decrease in APase gene expression compared to the expression level in RA-treated growth plate chondrocytes (Fig. (Fig.6A).6A). In addition, osteocalcin, another mineralization-related gene product, was also notably downregulated in RA-treated and ank-specific siRNA-transfected growth plate chondrocytes compared to the expression levels in RA-treated cells (Fig. (Fig.6A).6A). In contrast, overexpression of Ank in hypertrophic nonmineralizing growth plate chondrocytes led to a ~5-fold increase of APase gene expression compared to the levels in RCAS-infected and uninfected cells (Fig. (Fig.6B).6B). Expression of other mineralization-related genes, such as those for type I collagen and osteocalcin, was increased ~7- and ~3-fold, respectively, in ank/RCAS-infected cells compared to the levels in uninfected or RCAS-infected cell cultures (Fig. (Fig.6B).6B). Type II collagen gene expression was reduced in ank/RCAS-infected cells compared to the levels in uninfected or RCAS-infected cells (Fig. (Fig.6B),6B), suggesting that mineralization-related genes are upregulated in Ank-overexpressing cells, whereas type II collagen expression is downregulated in these cells.
Transfection of RA-treated growth plate chondrocytes with ank-specific siRNA led not only to a decrease in APase gene expression but also to a significant reduction of APase activity compared to the levels in RA-treated cells (Fig. (Fig.7).7). APase activity was measured after 3 days of treatment. Inhibition of Ank PPi transport activities by probenecid treatment also markedly reduced APase activity in RA-treated cultures (Fig. (Fig.7).7). Treatment of hypertrophic nonmineralizing growth plate chondrocytes with probenecid alone had no effect on their APase activity, which was significantly lower (P ≤ 0.01) than the activity in RA-treated cells (Fig. (Fig.7).7). These findings suggest that downregulation of Ank expression and/or activity led to a decreased transport of intracellular PPi to the extracellular milieu and decreased APase activities.
Next, we determined the degree of mineralization in RA-treated, Ank-overexpressing, and RA-treated and ank-specific siRNA-transfected growth plate chondrocytes. We also determined the effect of probenecid or levamisole treatment on mineralization of growth plate chondrocytes expressing high levels of Ank. Mineral deposits were detected and quantitated in chondrocyte cultures after 6 days of treatment and stained with alizarin red S. Alizarin red S stain was released by incubation with cetylpyridinium chloride, and the degree of mineralization was expressed as released units of alizarin red S per milligram of total protein. RA treatment and Ank-overexpression markedly increased the degree of mineralization compared to the degree in untreated, probenecid-treated, or RCAS-infected growth plate chondrocytes (Fig. (Fig.8).8). Downregulation of Ank expression or treatment with probenecid notably reduced the rate of mineralization in RA-treated cultures (Fig. (Fig.8).8). Levamisole also markedly decreased the rate of mineralization in RA-treated or Ank-overexpressing growth plate chondrocytes (Fig. (Fig.88).
To gain insight into the mechanisms of how enhanced expression and activity of Ank regulates APase expression and activity, we determined whether Pi resulting from hydrolysis of extracellular PPi by APase is involved. Previous studies have shown that Pi enters through type III Na+/Pi cotransporter growth plate chondrocytes (23). PFA is a competitive inhibitor of type III Na+/Pi cotransporters (2, 23). First, we measured Pi uptake by growth plate chondrocytes after 3 days of treatment. The intracellular Pi concentration was significantly higher (P ≤ 0.01) in growth plate chondrocytes treated with RA and GP (a substrate of APase) than in cells treated with GP only (Fig. (Fig.9A).9A). Cotreatment of cells with RA, GP, and PFA significantly (P ≤ 0.05) reduced the intracellular Pi concentration compared to the concentration in RA- and GP-treated cells (Fig. (Fig.9A).9A). RT-PCR analysis showed that hypertrophic nonmineralizing growth plate chondrocytes express both Pit-1 (Glvr-1) and Pit-2 (Ram-1) (Fig. (Fig.9B).9B). Ank overexpression in these cells resulted in an upregulation of Pit-1 and Pit-2 gene expression compared to the expression levels in RCAS-infected or uninfected cells (Fig. (Fig.9B).9B). Treatment of Ank-overexpressing growth plate chondrocytes with either levamisole or PFA led to a marked reduction in Pit-1 and Pit-2 gene expression (Fig. (Fig.9B),9B), suggesting that Pi influx into growth plate chondrocytes regulates Pit-1 and Pit-2 expression.
Overexpression of Ank in hypertrophic nonmineralizing growth plate chondrocytes resulted in upregulation of APase protein expression and activity compared to the levels in uninfected or RCAS-infected cells (Fig. (Fig.10).10). Treatment of Ank-overexpressing chondrocytes with the APase activity inhibitor levamisole led not only to a reduction of APase activity (as expected) (Fig. 10B) but also to decreased APase protein expression (Fig. 10A), suggesting that Pi generated by APase is involved in further upregulation of APase expression. Treatment of Ank-overexpressing cells with PFA also resulted in a decrease in APase protein expression (Fig. 10A) and activity (Fig. 10B), suggesting that interfering with Pi transport into chondrocytes affects APase expression and activity. APase expression and activity were measured after 3 days of treatment. At that time the cultures showed no sign of mineralization (data not shown), and therefore PFA treatment was solely affecting Pi transport into cells and not interfering with matrix mineral formation and, as a possible consequence, the chondrocyte phenotype.
Physiological mineralization of skeletal tissues is a complex and highly regulated process. Uncontrolled or pathological mineralization can have severe consequences and can lead to morbidity and mortality. Recently, Ank was shown to be a transmembrane protein, which transports intracellular PPi to the extracellular milieu (11, 30). Recent studies have demonstrated that mutations in human Ank lead to abnormal mineralization in articular cartilage and bone (11, 30, 34). Furthermore, other studies have shown that ank mRNA is highly expressed by hypertrophic chondrocytes in the murine growth plate and that Ank expression is upregulated in osteoarthritic cartilage, where mineralization also occurs (10, 17, 41). However, little is known about the regulatory functions of Ank during physiological and pathological mineralization. The results of this study show that Ank is a crucial regulator of PPi-Pi homeostasis and BCP mineral formation in growth plate cartilage. Ank expression is markedly upregulated in terminally differentiated growth plate chondrocytes undergoing mineralization. Suppression of ank gene expression by using siRNA or inhibition of Ank PPi transport activities in terminally differentiated growth plate chondrocytes notably reduced their rate of mineralization. In contrast, overexpression of Ank in hypertrophic nonmineralizing growth plate chondrocytes accelerates their terminal differentiation and mineralization. These findings seem to be contradictory to the notion that PPi is an inhibitor of BCP mineral formation (5, 44). However, our study also demonstrates that enhanced Ank PPi transport activities lead to an upregulation of APase expression and activity, resulting in the rapid hydrolysis of the mineralization inhibitor PPi and the generation of Pi, which is required as a signaling molecule and for BCP mineral formation. Consequently, inhibition of APase activity in growth plate chondrocytes expressing high levels of Ank leads to an increase of extracellular PPi concentration and inhibition of BCP mineralization, consistent with previous findings showing that PPi is an inhibitor of BCP mineralization (5, 24, 31, 36, 44).
Our findings that Ank is highly expressed in hypertrophic and mineralizing growth plate chondrocytes in vitro and in vivo are consistent with previous findings showing ank mRNA localization in hypertrophic chondrocytes during endochondral bone formation in the mouse (41). In addition, our sequence analysis between the chicken ank sequence and ank sequences from a variety of other species reveals a high sequence similarity, suggesting that the function of Ank within different species is highly conserved and important. Interestingly, increased expression of Ank, APase, and other terminal differentiation markers was also detected in human osteoarthritic cartilage (10, 17, 20, 33 (see also Fig. Fig.2B),2B), suggesting that Ank may play a similar regulatory role in pathological BCP mineral formation and terminal differentiation in articular cartilage during osteoarthritis.
How does Ank upregulate APase gene expression and activity? Our findings demonstrating (i) that chondrocytes expressing large amounts of Ank show a lower extracellular PPi concentration than chondrocytes expressing less Ank; (ii) that the intracellular Pi concentration in growth plate chondrocytes treated with RA and GP is higher than the concentration in cells treated with GP only, and this increase of intracellular Pi concentration is inhibited by the type III Na+/Pi cotransport inhibitor PFA; (iii) that type III Na+/Pi cotransporter Pit-1 and Pit-2 gene expression is increased in Ank-overexpressing growth plate chondrocytes, and this increase is inhibited by APase activity inhibitor levamisole or by PFA; and (iv) that APase expression and activity in Ank-overexpressing hypertrophic growth plate chondrocytes is inhibited by either levamisole or PFA suggest that APase-mediated hydrolysis of PPi to Pi acts as feedback loop to further stimulate APase expression and activity. Furthermore, these findings suggest that extracellular Pi being transported through the type III Na+/Pi cotransporters, Pit-1 and Pit-2, into chondrocytes acts as an intracellular signaling molecule, which further stimulates APase and other mineralization-related gene expression (Fig. (Fig.11).11). Other studies have also demonstrated that extracellular Pi not only is required for the formation of BCP crystals but also acts as a signaling molecule and affects cell differentiation (2, 23). Microarray analysis of osteoblasts cultured in the presence of different concentrations of Pi or β-glycerophosphate revealed up- and downregulation of a variety of genes, including transcriptional regulators, membrane transport proteins (including Pit-1), signaling molecules, and extracellular matrix proteins (3). Other studies have shown that high concentrations of extracellular Pi stimulate apoptosis of osteoblasts and chondrocytes (23). These studies have also demonstrated that osteoblasts and growth plate chondrocytes mainly express type III Na+/Pi cotransporters and that blocking these transporters by PFA or other agents inhibited Pi-mediated events, including alterations of gene expression and apoptosis (2, 3, 23). These findings are consistent with our results demonstrating that interfering with type III Na+/Pi cotransport systems in terminally differentiated or Ank-overexpressing growth plate chondrocytes inhibits upregulation of APase expression and activity. However, our results also demonstrate that besides the transport of Pi into growth plate chondrocytes, other mechanisms which regulate APase gene expression and activity exist. We and others have shown that RA increases Cbfa1 expression and activity in growth plate chondrocytes (14, 46). Cbfa1 is a transcription factor which has been shown to stimulate hypertrophic and terminal differentiation events and which activates the expression of mineralization-related genes, including the APase gene (43). Another activator of mineralization-related gene expression, including that of APase, is annexin-mediated Ca2+ influx into growth plate chondrocytes. We have previously shown that annexins II, V, and VI mediate Ca2+ influx into terminally differentiated growth plate chondrocytes, leading to increased cytoplasmic Ca2+ concentration and upregulation of Cbfa1, APase, type I collagen, and osteocalcin gene expression. Chelation of intracellular Ca2+ with 1,2-bis(2-amino phenoxy)ethane-N,N,N′,N′-tetraacetic acid acetoxy methylester (BAPTA-AM) or a specific annexin channel activity blocker inhibited this upregulation (46, 47). In addition, it might be possible that APase gene expression is also regulated by nuclear RA receptors. Therefore, PPi/Pi homeostasis regulated by Ank likely acts as a feedback loop, which further stimulates APase expression and activity and allows sufficient Pi to be generated during the initiation of mineralization.
Interestingly, even higher concentrations of APase are found on the outer membrane surface of matrix vesicles than on the plasma membrane of growth plate chondrocytes (19). These vesicles are released from the plasma membrane of mineralization-competent growth plate chondrocytes, and they have the critical role of initiating the mineralization process (1, 19). We and others have shown that the first mineral forms within the vesicles and that annexins also form Ca2+ channels in matrix vesicles, allowing influx of Ca2+ into the vesicle lumen (1, 19). Other studies have demonstrated that these vesicles also contain type III Na+/Pi cotransporters allowing the influx of Pi into the vesicles (27). Therefore, high APase expression and activities are required for the initiation of mineralization, and Ank-regulated PPi/Pi homeostasis seems to play a critical role in providing sufficient APase activity required for this process.
A recent study has shown that increased APase expression leads to an enhanced production of plasma cell membrane glycoprotein-1, a member of the nucleoside triphosphate pyrophosphohydrolases, which is expressed by osteoblasts, hypertrophic growth plate chondrocytes, and articular chondrocytes and which generates PPi by hydrolysis of its major substrate ATP (16). Therefore, it is possible that an interrelated series of upregulations of extracellular PPi-generating and hydrolyzing protein expression and activities is required for effective and controlled mineralization of growth plate cartilage and other skeletal tissues. Any disturbance to this well-regulated system leads to uncontrolled or defective mineralization. For example, hypophosphatasia as a consequence of deactivating mutations in the APase gene is characterized by poorly mineralized growth plate cartilage (rickets) and bones (osteomalacia) and by elevated levels of PPi, which probably causes poor mineralization (4, 29, 48). CPPD crystal deposits in articular cartilage due to elevated levels of extracellular PPi are eminent in patients with adult hypophosphatasia. Tiptoe-walking mice have a nonsense mutation in plasma cell membrane glycoprotein-1, leading to excessive BCP mineralization in ligaments, tendon, and articular cartilage because of a lack of extracellular PPi (28, 38). A nonsense mutation in ank mice leads to a similar phenotype, again because of the lack of extracellular PPi (11). On the other hand, mutations in human ank that activate Ank PPi transport activity lead to CPPD crystal deposits in articular cartilage and subsequent osteoarthritis (because of supersaturation of PPi), whereas other mutations in human ank, which were also suggested to be Ank PPi transport-stimulating mutations, result in overgrowth and overmineralization (BCP crystals) of craniofacial bones (likely because of the antagonistic effects of APase and Ank) (30, 31).
As discussed above, excessive extracellular PPi can lead to either CPPD or BCP crystal formation. As shown in this study, increased levels of extracellular PPi in hypertrophic and terminally differentiated growth plate chondrocytes lead to increased APase activity, subsequent hydrolysis of PPi to Pi, and BCP mineralization. In contrast, patients with CPPD deposition disease resulting from either upregulated expression of ank or Ank PPi transport-activating mutations have CPPD crystal deposits in articular cartilage because of a supersaturation of extracellular PPi (10, 31). In the case of growth plate chondrocytes, these cells respond to the increased extracellular PPi concentration by hydrolysis of PPi to Pi followed by further upregulation of APase expression and activity. In contrast, in articular chondrocytes, which do not express APase and seem not to upregulate APase by increased levels of extracellular PPi, supersaturation with PPi leads to CPPD crystal formation. The fact that articular chondrocytes, in contrast to growth plate chondrocytes, do not upregulate APase expression and activity in response to increased Ank activities further confirms our earlier notion that Pi (resulting from the hydrolysis of PPi) and not PPi is the signaling molecule that leads to the upregulation of expression of the APase gene and other genes and possibly to apoptosis. However, results from this study and others have shown that articular chondrocytes can undergo hypertrophic and terminal differentiation events similarly to growth plate chondrocytes, including upregulation of Ank and APase expression (10, 17, 20, 33). In this case, the Ank-mediated increase of extracellular PPi and initial hydrolysis of PPi to Pi by APase may lead to a further stimulation of APase and other mineralization-related gene expression and subsequent BCP mineral formation in osteoarthritic cartilage, similar to the events occurring in growth plate cartilage. BCP crystals will then further accelerate cartilage destruction.
In conclusion, our study demonstrates that upregulated expression and activities of Ank during growth plate chondrocyte hypertrophy and terminal differentiation, and possibly in articular osteoarthritic cartilage, play a crucial regulatory element in BCP mineralization in these tissues. Upregulated Ank expression and activity result in elevated levels of extracellular PPi. The initial APase activity present in growth plate chondrocytes and osteoarthritic chondrocytes then hydrolyzes PPi, thereby removing an inhibitor of mineralization and providing Pi required for further upregulation of APase and other mineralization-related gene expression. Therefore, the coordinated regulation of expression and activities of Ank and other proteins involved in PPi and Pi generation and the control of a precise extracellular PPi level are absolutely crucial for normal skeletal development and mineralization.
We thank E. E. Golub for providing us with the antibodies specific for chicken alkaline phosphatase, and we thank J. A. Katzman and T. F. Linsenmayer for developing the monoclonal antibodies specific for chicken type X collagen, which were obtained from the Developmental Studies Hybridoma Bank maintained by the Department of Biological Sciences, University of Iowa, Iowa City.
This work was supported by grants from the National Institutes of Health (National Institute of Arthritis and Musculoskeletal and Skin Diseases, grant AR 46245) and the Arthritis Foundation (to T.K.). The Developmental Studies Hybridoma Bank is maintained under contract N01-HD-7-3263 from the National Institute of Child Health and Human Development.