Abstract
Objective. Uncontrolled proliferation of synovial fibroblasts is characteristic of the pathology of rheumatoid arthritis (RA). Since synovial tissues in the rheumatoid joints are hypoxic, we investigated how hypoxia affects RA synovial fibroblast (RASF) proliferation.
Methods. RASF were cultured at 2000 cells (low density culture) or at 5000 cells (high density, growth-inhibitory confluent culture) per microtiter well under hypoxic (10%, 3%, or 1% O2) or normoxic (21% O2) conditions. Some RASF were treated with recombinant human interleukin 1 receptor antagonist (IL-1ra), anti-tumor necrosis factor-α (TNF-α)-neutralizing antibodies, anti-N-cadherin-blocking antibodies, or MG132. 3H-labeled thymidine incorporation was quantified to assess their proliferation. Total RNA and cell lysates were prepared for real-time polymerase chain reaction and Western blot analyses.
Results. Hypoxia exerted no effect on proliferation of RASF cultured at low density. At high density, it abrogated contact-dependent growth inhibition of RASF, but not of human dermal fibroblasts. Addition of anti-TNF-α antibodies or IL-1ra did not affect the results. Upregulated expression of cyclin-dependent kinase inhibitor p27Kip1 was observed in the cells cultured at high density under normoxic conditions, but not under hypoxic conditions. Hypoxia decreased N-cadherin expression on RASF. Addition of anti-N-cadherin-blocking antibodies mimicked the effects of hypoxic culture; it promoted proliferation of RASF cultured at high density under normoxic conditions. This antibody treatment also downmodulated p27Kip1 expression.
Conclusion. Hypoxia downregulates N-cadherin expression on RASF, and thus prevents p27Kip1 upregulation for their contact inhibition. It is likely that hypoxia in rheumatoid synovial tissues contributes to rheumatoid pathology by augmenting proliferation of synovial fibroblasts.
Rheumatoid arthritis (RA) is a common autoimmune disease characterized by synovitis and subsequent destruction of cartilage and bone. In affected joints, inflammatory cells including lymphocytes and macrophages are recruited and activated to produce many cytokines at high concentrations. Among them, tumor necrosis factor-α (TNF-α) and interleukin 1 (IL-1) are established targets in treatment of RA, and stimulate RA synovial fibroblasts (RASF) to proliferate and to produce inflammatory mediators1. In the normal joint, the synovial lining layer is at most a few cells thick. In the RA inflamed joint, RASF proliferate to form stratified hyperplastic synovial tissues called pannus. When isolated and transferred to in vitro culture, they still proliferate as if they were transformed cells, which show no contact-dependent proliferative inhibition (contact inhibition)1.
Because of the uncontrolled synovial hyperplasia, capillary density becomes insufficient for oxygen demand by synovial cells in the rheumatoid joints2. Moreover, synovial fluid retention in affected joints increases the intraarticular pressure, leading to further reduction of blood perfusion3. Thus, rheumatoid synovial tissues support vigorous proliferation of synovial fibroblasts, and are paradoxically hypoxic3,4. Lund-Olesen reported that oxygen levels of synovial fluid in patients with RA are reduced to less than half of those in healthy controls; mean pO2 is 63 mm Hg in the normal joints and 27 mm Hg in the rheumatoid joints3.
Hypoxia regulates gene expression of various inflammatory mediators and proteinases involved in bone and cartilage destruction2,5,6, and which are suggested to contribute to rheumatoid inflammation. Effects of hypoxia on cellular proliferation depend on the cell types. Hypoxia promoted proliferation of endothelial cells7 and fibroblasts8,9. It down-regulated cyclin-dependent kinase inhibitor (CDKI) p21Cip1 protein expression9. It also extended the lifespan of vascular smooth-muscle cells by activating telomerase10. In contrast, it halted cell-cycle progression by upregulation of CDKI, p16INK4a in the CV-1P monkey kidney cell line11, and p21Cip1 and p27Kip1 in murine embryonic fibroblasts12,13. Thus, our objective was to determine how hypoxia affects the cell cycle of RASF.
Cell-cycle progression is largely regulated by kinase activity of cyclin/cyclin-dependent kinase complexes14. CDKI are intracellular molecules that halt cell-cycle progression via inhibition of cyclin/CDK kinase activities. We reported previously that gene transfer of CDKI inhibited RASF proliferation in vitro, and that intraarticular CDKI gene therapy ameliorated animal models of RA15,16. Thus, RASF proliferation is crucial in the pathology of RA, and could be a target of treatment for RA. We investigated the effect of hypoxia on RASF proliferation in the context of CDKI expression.
MATERIALS AND METHODS
Cell culture
Synovial tissues were derived from 10 patients with RA who had responded poorly to antirheumatic drugs and underwent joint replacement or synovectomy at Tokyo Medical and Dental University Hospital, Tokyo Metropolitan Bokuto Hospital, National Shimoshizu Hospital, or Shinshu University Hospital. All patients fulfilled the American College of Rheumatology criteria for classification of RA17. All gave their consent to procedures in our studies, which were approved by the ethics committees of Tokyo Medical and Dental University and RIEN Research Center for Allergy and Immunology.
RASF were isolated and cultured as described15. Adult normal human dermal fibroblasts (HDF) were purchased from Cambrex, East Rutherford, NJ, USA. They were used at early passages (passages 5 to 9). These fibroblasts were cultured under normoxic conditions (21% O2). Hypoxic conditions (10%, 3%, or 1% O2) were generated in a hypoxic chamber filled with CO2 (5%) and N2 (85%, 92%, or 94%) gas mixture. Oxygen concentration was monitored with an oxygen electrode (Cosmo or Jiko, Tokyo, Japan). Dissolved oxygen levels in culture supernatants were measured with a dissolved oxygen monitor (Central Kagaku, Tokyo, Japan). RASF and HDF were cultured in a microtiter-plate at 5000 cells per well (high density culture) or 2000 cells per well (low density culture). RASF for RNA, total protein, and nuclear extraction were cultured at 3.5 × 105 cells per 60 mm dish (high density culture).
Cell proliferation assay
RASF were incubated for 24–72 h in a hypoxic chamber or under normoxic conditions. Some RASF were treated with 100 ng/ml recombinant human IL-1 receptor antagonist (IL-1ra; Prospec, Rehovot, Israel), 2 μg/ml anti-TNF-α-neutralizing monoclonal antibody (mAb; J2D10, Lab Vision, Fremont, CA, USA), 80 μg/ml anti-N-cadherin-blocking mAb (GC-4; Sigma, St. Louis, MO, USA), control mouse IgG1 mAb (MOPC-31; BD Biosciences, San Diego, CA, USA), and up to 200 μM cobalt chloride (CoCl2; Sigma). During the last 24-h culture, 0.3 μCi of 3H-labeled thymidine was present for quantification of the incorporated radioactivities.
Real-time polymerase chain reaction (PCR)
Total RNA was isolated with an RNeasy kit (Qiagen, Valencia, CA, USA) and converted to cDNA with Superscript II (Invitrogen, Carlsbad, CA, USA) reverse transcriptase. Real-time PCR was carried out with iQ Syber Green supermix (Bio-Rad, Hercules, CA, USA) and a set of primers specific to N-cadherin18 or p27Kip1 cDNA (sense: GCT CTA GAT TTT TTG AGA GTG CGA GAG AG; antisense: GGG GTA GCC GCT TTT AGA GGC AGA TCA TT). Data were standardized with human 28S ribosomal RNA (sense: TTGAAA ATC CGG GGG AGA G; antisense: ACA TTG TTC CAA CAT GCC AG), and were analyzed with the cycle threshold method19.
Western blot analyses
RASF were cultured for 24–72 h. Some RASF were treated with dimethyl sulfoxide (DMSO); Sigma), or 2.5 μM carbobenzoxy-L-leucyl-L-leucyl-L-leucinal (MG132; EMD Chemicals, Darmstadt, Germany) during the last 24-h culture. Total protein extraction and sodium dodecyl sulfate-polyacrylamide gel electrophoresis were performed as described15. Nuclear extracts were prepared using a Nuclear Extract Kit (Active Motif, Carlsbad, CA, USA). Rabbit anti-human p16INK4a, p21Cip1, CDK4 (sc-468, sc-397, and sc-260; Santa Cruz Biotech, Santa Cruz, CA, USA), mouse anti-E-cadherin, N-cadherin, p27Kip1 (HECD-1, clone 32, and clone 57; BD-Biosciences), cadherin-11 (clone 283416; R&D Systems, Minneapolis, MN, USA), and hypoxia inducible factor-1α 67; Gene Tex, San Antonio, TX, USA) antibodies were used as primary antibodies. Horseradish peroxidase-conjugated anti-rabbit (NA-934; GE Healthcare Biosciences, Piscataway, NJ, USA) or anti-mouse IgG antibodies (6175-05; Southern Biotech, Birmingham, AL, USA) were used as secondary antibodies. Bound antibodies were visualized with ECL (GE Healthcare Biosciences). Signal intensities were quantified with ImageJ software (US National Institutes of Health, Bethesda, MD, USA).
ELISA
ELISA kits for IL-1α, IL-1ß, IL-6, and TNF-α (Biosource International, Camarillo, CA, USA) were employed to quantify protein levels in the culture supernatants.
Statistics
Student’s paired t test was used for statistical comparisons.
RESULTS
RASF proliferation accelerated by hypoxia in high density culture
The effect of hypoxia on RASF growth was studied with 3H-thymidine incorporation. RASF were first cultured at low density in a microtiter plate for logarithmic cell growth. They grew equally under normoxic (21% O2) and hypoxic (1% O2) conditions. When cultured at high density under normoxic conditions, they stopped growing and incorporated less thymidine. Although more cells were present in the high density culture under normoxic conditions, these cells incorporated an amount of 3H-thymidine comparable only to those in the low density culture. However, under hypoxic conditions, they incorporated more thymidine than those under normoxic conditions (Figure 1A). Thus, hypoxia attenuated growth suppression that was induced by high density culture.
Hypoxia-induced augmentation of fibroblast proliferation. A. RA synovial fibroblasts (RASF) were cultured in microtiter plates at high or low density for 72 h under normoxic (21% O2) or hypoxic (1% O2) conditions. RASF proliferation was assessed using 3H-thymidine incorporation in 5 wells per RASF sample, and the mean values were calculated. Columns and bars represent mean and SD values of 5 RASF samples. *p < 0.05. B. Effect of hypoxia on proliferation of human dermal fibroblasts was assessed in the same manner. Representative results of 1 of 2 experiments are shown. Columns and bars represent mean and standard deviations of 5 wells. C. The effects of hypoxia on the proliferation of RASF were assessed as in panel A. RASF were cultured in microtiter plates at high density for 72 hours under normoxic (21% O2) or hypoxic (10% and 3% O2) conditions. Columns and bars represent mean and SD of 3 RASF samples. D. RASF were cultured at high density for 24, 48, or 72 hours under normoxic (21% O2) or hypoxic (1% O2) conditions as described above. The effect of hypoxia on the proliferation of RASF was assessed in the same manner. Columns and bars represent mean and SD of 3 RASF samples. The relative 3H-thymidine incorporations by RASF cultured in hypoxic conditions standardized with cultures in normoxic conditions at the same timepoints are shown at the bottom. *p < 0.05. E. The effect of hypoxia on the proliferation of RASF cultured at high density was assessed as in A. RASF were treated with culture medium only (control), 100 ng/ml human recombinant IL-1ra, 2 μg/ml isotype IgG1 control antibody, or 2 μg/ml anti-TNF-α-neutralizing mAb. Representative results of 1 of 2 experiments are shown. Columns and bars represent mean and SD of 5 wells. *p < 0.05. cpm: counts per minute; ns: not significant.
Under normoxic (21% O2) and hypoxic conditions (1% O2), the dissolved oxygen concentrations in the culture supernatants of RASF cultured at low density were 8.01 ± 0.06 and 2.15 ± 0.14 mg/dl, and those of RASF cultured at high density were 8.00 ± 0.03 and 2.14 ± 0.07 mg/dl, respectively (mean ± SD of 3 samples). In these experiments, the cell density was found not to affect dissolved oxygen concentrations.
To determine if this was a general feature of fibroblasts, HDF were cultured in the same conditions. The results showed that hypoxia (1% O2) did not promote proliferation of HDF in high density culture (Figure 1B).
The mean pO2 is reported to be 63 mm Hg in normal joints and 27 mm Hg in rheumatoid joints3. To simulate oxygen supplies in normal and rheumatoid joints, we examined the effects of 10% O2 (mean O2 level in normal joints) or 3% O2 (mean O2 level in rheumatoid joints) on proliferation of RASF cultured at high density. RASF cultured under 3% O2 incorporated slightly more 3H-thymidine than fibroblasts under 10% O2, but the differences were not statistically significant (Figure 1C).
To elucidate the time course of accelerated RASF proliferation by hypoxia, we examined the 3H-thymidine incorporation by RASF cultured at high density under normoxic (21% O2) or hypoxic (1% O2) conditions at 24 hours, 48 hours and 72 hours. The relative 3H-thymidine incorporation by RASF under hypoxic conditions standardized with those under normoxic conditions at the same timepoints increased gradually with time, and it reached 2.64 ± 0.51 after 72 hours of culture (Figure 1D).
Effects of IL-1 and TNF-α on hypoxia-induced proliferation
When IL-1α, IL-1ß, and TNF-α in the supernatants of RASF culture were quantified with specific ELISA, they were all below detection limits (IL-1α and IL-1ß < 3.9 pg/ml, TNF-α < 1.7 pg/ml). This suggested that hypoxia does not stimulate RASF to produce these cytokines. Even when IL-1ra or anti-TNF-α-blocking mAb were included in the medium, RASF in the high density culture grew more under hypoxic conditions (Figure 1E). In separate experiments, the same concentrations of these blocking reagents suppressed the 3H-thymidine incorporation and matrix metalloproteinase-3 production by RASF stimulated with 10 pg/ml IL-1ß and 10 pg/ml TNF-α (data not shown).
It has been reported that IL-6 suppresses RASF proliferation20. The supernatants of RASF cultured at high density under normoxic conditions and those cultured under hypoxic conditions contained comparable levels of IL-6: 4.6 ± 4.9 ng/ml and 6.4 ± 6.5 ng/ml (mean ± SD of 3 samples), respectively. Thus, IL-6 was not responsible for the overgrowth of RASF under the hypoxic conditions.
Downregulation of CDKI P27Kip1 expression by hypoxia
In general, logarithmic cell growth of nontransformed cells can be inhibited in high density culture. This contact inhibition involves upregulation of the CDKI p16INK4a, p21Cip1, and/or p27Kip1, depending on the cell types21–23. Our previous studies revealed that RASF in confluent culture upregulated expression of these CDKI when they were incubated for more than 4 days15. We examined the effect of hypoxia in CDKI expression. When RASF were cultured at high density for 3 days, the protein level of p27Kip1 was upregulated. p16INK4a and p21Cip1 protein were not upregulated at this timepoint. However, the high density culture did not induce p27Kip1 upregulation under hypoxic conditions (Figure 2A, 2B). The protein level of p27Kip1 was not upregulated when RASF were cultured at low density for 3 days (data not shown). Quantitative PCR of p27Kip1 mRNA transcripts showed that the difference did not depend on alteration of the p27Kip1 mRNA expression (Figure 2C). It has been reported that the p27Kip1 protein is degraded via the ubiquitin-proteasome pathway. To elucidate whether this pathway is involved in the attenuation of p27Kip1 protein expression by hypoxia, a proteasome inhibitor, MG132, was added to the culture medium. MG132 upregulates p27Kip1 protein expression through inhibition of protein degradation24. MG132 2.5 μM upregulated p27Kip1 protein concentrations in RASF cultured at high density under hypoxia (1% O2; Figure 2D). This suggests that attenuation of p27Kip1 by hypoxia is regulated via the ubiquitin-proteasome pathway. These findings were in agreement with the fact that the protein level of p27Kip1 was primarily controlled at the posttranscriptional level, via the ubiquitin-proteasome pathway25. In contrast, when HDF were cultured at high density, the p27Kip1 protein was upregulated even in hypoxic conditions (Figure 2E).
Downregulation of p27Kip1 expression by hypoxia. A. RASF were cultured at high density under normoxic (21% O2) or hypoxic (1% O2) conditions, and whole-cell lysates were collected after indicated number of days. Expression of CDKI p16INK4a, p21Cip1, and p27Kip1 by RASF was immunodetected by Western blot. CDK4 was stained as loading control. Expression levels of p27Kip1 protein standardized with those of the CDK4 protein in the same blot are shown at the bottom. pc: positive control. B. p27Kip1 levels cultured under normoxic (21% O2) and hypoxic (1% O2) conditions for 3 days were standardized with levels of CDK4. Columns and bars represent mean and SD of 3 samples. *p < 0.05. C. RASF were cultured for 3 days at high density under normoxic (21% O2) or hypoxic (1% O2) conditions before extraction of total RNA. p27Kip1 mRNA expression was quantified by real-time PCR using 28S ribosomal RNA as an endogenous control. Columns and bars represent mean and SD of 3 samples. ns: not significant. D. RASF were cultured at high density under normoxic (21% O2) or hypoxic (1% O2) conditions, and whole-cell lysates were collected after 3 days. RASF were treated with culture medium alone (med), 0.1% DMSO, or 2.5 μM MG132 during the last 24 hours of culture. p27Kip1 expression by RASF was immunodetected by Western blot. Representative results of 2 samples are shown. Expression levels of p27Kip1 protein standardized with those of the CDK4 protein in the same blot are shown at the bottom. E. Human dermal fibroblasts were cultured for 3 days at high density under normoxic (21% O2) or hypoxic (1% O2) conditions before extraction of whole-cell lysates. Expression of p27Kip1 and CDK4 was assessed as in A. Representative blots of 2 independent experiments are shown.
Hypoxia-induced N-cadherin downregulation responsible for accelerated RASF proliferation
Cadherins are a group of cell-surface molecules that recognize direct cell-cell contact and mediate signal pathways into the cells. Homophilic interaction of surface N-cadherin and E-cadherin promotes p27Kip1 protein accumulation for contact inhibition26,27. Recently, it was reported that cadherin-11 is expressed by RASF and contributes to organization of the lining-like structure of RA synovial tissues28,29. RASF grown under normoxic conditions expressed N-cadherin and cadherin-11, but not E-cadherin. When the same cells were cultured under hypoxic conditions, their expression of N-cadherin, but not cadherin-11, was reduced (Figure 3A, 3B). Quantitation of mRNA with real-time PCR revealed that this reduction was regulated at the mRNA level (Figure 3B). These data suggested that hypoxia suppressed N-cadherin expression, resulting in augmentation of RASF proliferation at high density. In support of this, hypoxia did not alter the N-cadherin protein expression in HDF (Figure 3C).
Downregulation of N-cadherin expression by hypoxia. A. RASF were cultured at high density under normoxic (21% O2) or hypoxic (1% O2) conditions for 3 days before collection of whole-cell lysates. N-cadherin protein expression was immunodetected by Western blot. Representative blots of 3 samples are shown in the upper panel. Expression levels of N-cadherin protein were standardized with those of CDK4 protein (lower panel). Columns and bars represent mean and SD of the 3 samples. *p < 0.05. B. RASF were cultured as in A before extraction of total RNA. N-cadherin mRNA expression was quantified by real-time PCR using 28S as an endogenous control. Columns and bars represent mean and SD of the 3 samples. *p < 0.05. C. Expression of N-cadherin and CDK4 in human dermal fibroblasts cultured under normoxic (21% O2) or hypoxic (1% O2) conditions was assessed as in A. Representative blots of 2 independent experiments are shown. D. RASF were cultured at high density, and treated with anti-N-cadherin-blocking mAb (αN-cad) or isotype IgG1 control mAb. After 3 days, effect of N-cadherin blockade on RASF proliferation was examined by 3H-thymidine incorporation. Representative results of 3 samples are shown. Columns and bars represent mean and SD of 5 wells. *p < 0.05. E. Expression of p27Kip1 and CDK4 by RASF treated with αN-cad or IgG1 for 3 days were immunodetected by Western blot. Representative results of 2 independent experiments are shown. pc: positive control
Next, anti-N-cadherin blocking mAb was used to interfere with homophilic interaction of N-cadherins. When RASF were cultured with this mAb at high density under normoxic conditions, they proliferated more than those treated with control IgG (Figure 3D). In analogy to hypoxia, treatment with the blocking mAb downmodulated expression of p27Kip1 protein in the cultured RASF (Figure 3E).
DISCUSSION
Hypoxia promoted proliferation of RASF by attenuating their contact inhibition. The accelerated proliferation was mediated by suppressed expression of N-cadherin, which should otherwise upregulate p27Kip1 expression on their homophilic interaction. It has been reported that hypoxia stimulates RASF to produce several inflammatory mediators and proteinases that are involved in the pathology of RA2,5. We observed that hypoxia may contribute to RA pathology by promoting proliferation of RASF as well (Figure 4).
Abrogation of contact-dependent proliferative inhibition of RASF by hypoxia. Hypoxia downregulates N-cadherin and subsequently p27Kip1 expression to augment RASF proliferation. It also upregulates inflammatory mediators and proteinases involved in RA. These results suggest that hypoxia in rheumatoid synovial tissues should contribute to joint destruction through multiple pathways. MMP: matrix metalloproteinase; VEGF: vascular endothelial growth factor.
Hypoxia did not affect logarithmic-phase proliferation of RASF, but abrogated their contact inhibition by modulating N-cadherin and subsequent p27Kip1 expression. The same changes were not observed in HDF. Thus, hypoxia-induced augmentation of proliferation via N-cadherin and p27Kip1 might be one specific feature of RASF. RASF were found to be prone to express a few types of CDKI in longterm in vitro cultures14. Although the short-term hypoxic culture in our experiments disclosed only its effect on p27Kip1, we assume that expression of other CDKI might be affected by hypoxia.
Our results showed that RASF proliferation was augmented in the presence of 1% O2, but not in the presence of 3% O2. Treuhaft, et al reported that the knees of 30% of patients with RA exhibited profound hypoxia, with pO2 less than 10 mm Hg, and the lowest pO2 in a rheumatoid joint was 8 mm Hg4. Thus, 1% O2 corresponds to the lowest levels in rheumatoid joints. Further, it has been reported that movement of inflamed joints can reduce the blood flow in synovial capillaries, and thus further aggravate synovial hypoxia30. Hypoxia may augment RASF proliferation in such severely hypoxic rheumatoid joints, especially during exercise.
The time-course experiments revealed that the relative 3H-thymidine incorporation by RASF was increased gradually with time, and peaked after 72 hours of culture. Further, hypoxia downregulated the expression of p27Kip1 and N-cadherin to a statistically significant level. However, the absolute level of 3H-thymidine incorporation by RASF peaked after 48 hours, and decreased thereafter, i.e., at 72 hours, even under hypoxic conditions. These data suggest that hypoxia-induced augmentation of RASF proliferation does not completely overcome contact inhibition. After culture for 48 hours, hypoxia augmented RASF proliferation at a statistically significant level, and downregulated the protein expression of p27Kip1, but the latter effect was not significant. Thus, some mechanism other than the attenuation of p27Kip1 protein expression may be involved in the hypoxia-induced augmentation of RASF proliferation.
It is known that quite a few cellular responses to hypoxia are mediated by hypoxia inducible factor-1 (HIF-1)31. Its expression was upregulated in the rheumatoid synovial tissues, reflecting the hypoxic environment of the affected joints2. In a separate experiment, HIF-1 in cultured RASF was upregulated artificially with CoCl2 2. However, this treatment did not affect the RASF proliferation or the protein level of p27Kip1 (data not shown), suggesting that an HIF-1-independent pathway should operate in the hypoxiatriggered augmentation of RASF proliferation.
Cadherins maintain the integrity of multicellular structures. One of the intracellular outcomes of cadherin activation is induction of CDKI, including p21Cip1 and p27Kip1, that inhibit cell-cycle progression. E-cadherin interaction mediates contact-dependent proliferative inhibition by increasing the level of p27Kip1 expression in thyroid cancer cells32, while N-cadherin interaction induces contact inhibition through upregulation of p27Kip1 in CHO cells26. These data point to a link between cadherin signaling and CDKI protein expression. In addition, hypoxia suppresses E-cadherin expression on cancer cell lines33 and N-cadherin expression in cortex proximal renal tubules34. Thus, there is another link between hypoxia and cadherins.
RASF expressed N-cadherins, but not E-cadherins. Recently, it was reported that cadherin-11 is expressed by RASF and contributes to organization of the lining-like structure of synovial tissues28,29. A cadherin-11–Fc fusion protein and an anti-cadherin–11-blocking mAb prevented and reduced arthritis in mouse models35. Thus, cadherins, which should control proliferative reactions to hypoxia, appear to be crucial molecules for formation and maintenance of the rheumatoid synovial structure.
Acknowledgments
We thank Drs. T. Muneta, K. Goto and K. Taniguchi for providing synovial samples, Dr. I. Morita for monitoring of oxygen concentration, M. Toyomoto for technical assistance, and Dr. H. Tanaka for critical advice.
Footnotes
-
Supported by grants from the Ministry of Health, Labor and Welfare, Japan; the Ministry of Education, Culture, Sports, Science and Technology, Japan; the 21st Century Center of Excellence Frontier Research Program on Molecular Destruction and Reconstruction of Tooth and Bone; Kato Memorial Bioscience Foundation; and Japan Foundation for Applied Enzymology.
- Accepted for publication November 18, 2008.