Osteoarthritis is a debilitating and painful disease that is anticipated to affect 59 million Americans by 2020.1 In the past, osteoarthritis was often thought to be a noninflammatory form of arthritis completely separate from rheumatoid arthritis, but results of recent studies indicate that they share more similarities than previously thought. Populations of clonally expanded T-cells are found in the synovium of osteoarthritis patients,2 and factors that promote the proliferation and differentiation of adaptive immunity, including IL-2 and interferon-γ, were found in the synovium of 50% of osteoarthritis patients examined. Tumor necrosis factor α and IL-Iβ appear to be the primary inflammatory mediators of the initiation and perpetuation of degradative pathways in osteoarthritis.3 These 2 pleiotropic cytokines directly or indirectly promote a multitude of inflammatory responses involved in osteoarthritis, including iNOS, COX-2, and PGE2. It has been reported that PPARγ agonists do attenuate, at least in part, some of the inflammatory pathways associated with osteoarthritis. Fahmi et al4 reported that PGJ2 decreases IL-lβ—induced production of NO and MMP-13 in human articular cartilage. In the same study4 PGJ2 inhibited IL-1, IL-6, tumor necrosis factor-β, IL-8, monocyte chemotactic protein-1, NO, and MMP-9 production by synovial macrophages and IL-2 from T-cells in a dose-dependent manner. Similar effects associated with PGJ2 have been reported by several other researchers investigating osteoarthritis and rheumatoid arthritis.5–7
Previously, no synthetic PPARγ agonists have had anti-inflammatory effects similar to those reported for PGJ2, the endogenous form of PPARγ agonist. However, pioglitazone, an FDA-approved PPARγ agonist commonly used in diabetic patients to combat insulin resistance, has recently been reported to have chondroprotective effects in guinea pigs8 and dogs with osteoarthritis.5
Because pioglitazone is a commercially available pharmaceutical, its potential use as a novel therapeutic for osteoarthritis provides an attractive focus for continued research. However, information regarding safety and efficacy of PGJ2 and pioglitazone with respect to cartilage health is still limited. The purpose of the study reported here was to evaluate the effects of PPARγ agonists on tissue metabolism in cartilage and synovial explants from dogs. We hypothesized that PGJ2 and pioglitazone would substantially alleviate the detrimental effects of IL-1β known to consistently occur in cartilage and synovial explants.9
Materials and Methods
Cartilage-synovium coculture—Cartilage and synovium were harvested from 12 dogs euthanized for reasons not related to this study within 2 hours after euthanasia, and 4-mm-diameter explants were created from the tissues by use of a dermal punch. Two cartilage explants and 1 synovial explant were cocultured by use of 24-well tissue culture plates and 0.4-μm filter insertsa with the cartilage explants in the wells and the synovial explants in the filters. Explants were cultured with 1 mL of Dulbecco modified Eagle medium supplemented with insulin, transferrin, selenium,b L-glutamine (2mM), penicillin (1 U/mL), streptomycin (100 μg/mL), amphotericin B (2 μg/mL), sodium pyruvate (1mM), modified Eagle medium nonessential amino acids (0.1mM), ascorbic acid (50 μg/mL), and 100 ng/mL of recombinant human IL-1β (100 ng/mL). For this study, there were 6 culture groups: no IL-1β (negative control), IL-lβ alone (positive control), PJG2 low concentration (15μM and IL-1β), PJG2 high concentration (30μM and IL-1β), pioglitazone low concentration (2.5μM and IL-1β), and pioglitazone high concentration (4.2μM and IL-1β). One coculture per dog was used for each group, resulting in a sample size of 12 for each culture group. The concentrations for pioglitazone were derived from standard formulation amounts for oral administration. The physiologic concentrations in the body were calculated relative to the oral dosages on the basis of the pharmacokinetic data for dogs10 and extrapolated to the cell culture system. Power analysis revealed that the sample size of 12/group was sufficient to achieve significance (P < 0.05). The media were changed on days 3, 7, 9, 12, 15, 18, and 21 and stored at −20°C for analysis. On day 21, the cartilage explants alone were weighed to determine the wet weight, lyophilized, and weighed again to determine the dry weight and water content of the tissue. The dried cartilage tissues were digested to completion in 1 mL of papain (0.5 mg/mL)c in 20mM sodium phosphate buffer (pH, 6.8) containing 1mM EDTA at 60°C overnight. The digested tissue was used to determine the proteoglycan and collagen concentration of the cartilage tissue.
DMMB assay—The GAG concentrations in cartilage explants after 21 days of culture were determined via DMMB assay.11 Briefly, a 5-μL aliquot of the tissue digest solution was assayed for total GAG concentration (dry weight) by addition of 245 μL of DMMB solution and spectrophotometric determination of absorbance at 525 nm with known concentrations of chondroitin sulfate as standards. Results were standardized to correct for differences in sample weight. Total GAG concentration for samples was reported in micrograms per milligram.
The GAGs in liquid media were also quantitated via DMMB assay. The GAG concentration in liquid medium samples (5 μL) was assayed by addition of 245 μL of DMMB solution and spectrophotometric determination of absorbance at 525 nm. The GAG loss from the explant was determined by calculating GAG concentration in the liquid media as a percentage of total GAG in the explant for each sample.11
Hydroxyproline analysis—For collagen matrix evaluation, total collagen concentrations of the cartilage explants were estimated by measurement of hydroxyproline by use of a described colorimetric procedure.12 Briefly, a 50-μL aliquot of the tissue digest solution was mixed with 50 μL of 4N NaOH, and the mixture was autoclaved for 20 minutes at 121°C to hydrolyze the sample. The sample was then mixed with chloromine T reagent and incubated at 25°C for 25 minutes. The sample was mixed with Ehrlich aldehyde reagent, and the chlorophore was developed at 65°C for 20 minutes. After development, the absorbance was read at 550 nm and the samples were compared with a hydroxyprolinec standard to determine the hydroxyproline concentration of the sample. Results were standardized to tissue dry weight and are reported as micrograms of hydroxyproline per milligram of tissue (dry weight).
NO assay—The NO concentration of the media was obtained by determining NO2 — concentration according to manufacturer's instructions.d
PGE2 determination via ELISA—The PGE2 concentration of the media was determined via ELISA according to manufacturer's instructions. Because the concentration of PGE2 in the samples was beyond the range of the assay, the media samples were diluted in plain Dulbecco modified eagle medium to obtain concentrations within the range of the assay. Samples were initially tested at a dilution of 1:250. If needed, further dilutions were made to obtain appropriate concentrations. The maximum dilution conducted in this study was 1:1,500.
Liquid chromatography-mass spectrometry—Samples of media were submitted to the University of Missouri mass spectrometry core facility for analysis for metabolites. The liquid chromatography-mass spectrometry analyses were performed by use of electrospray ionization on a triple-quadrupole mass spectrometerf with an integrated liquid chromatography systemf consisting of a quaternary liquid chromatography pump and vacuum degasser, an autosampler, and a diode-array detector. A columng with dimensions of 4.6 × 150 mm and 5-μm packing was used for the separation.
The mobile phase was composed of high-performance liquid chromatography-grade solvents and consisted of 0.1% formic acid in water (A) and methanol (B) flowing at 1 mL/min; the gradient was as follows: 0 minutes, 100% A; 16 minutes, 100% B; 20 minutes, 100% B; 21 minutes, 100% A; and 25 minutes, 100% A. The photodiode array detector was set to acquire spectra over the 190- to 420-nm range. On the mass spectrometer, the electrospray needle voltage was 4.5 kV and the heated inlet capillary was set at 250°C. Nitrogen sheath and auxiliary gases were provided by use of liquid nitrogen. All other voltages and settings were optimized to maximize ion transmission and minimize unwanted fragmentation as determined during the regular tuning and calibration of the instrument.
Statistical analysis—All statistical analyses were performed by use of a computer software program.h A 1-way ANOVA was performed to determine significant differences among treatment groups with respect to each assay at each collection time. When significant differences among groups were detected, an all-pairwise multiple comparison (Tukey test) was performed. If the data failed a normality test, a Mann-Whitney rank sum test was performed to determine significance among groups. Significance was set at P < 0.05.
Results
To determine the in vitro effects of PPARγ agonists on cartilage health, tissue GAG and hydroxyproline concentrations were examined in cartilage explants. The GAG concentration (dry weight) of the explants cultured in IL-lβ (100 ng/μL) plus PGJ2 (30μM and 15μM) was significantly higher than that of all other groups. There were no significant differences in tissue hydroxyproline concentrations in PPARγ-treated groups, compared with the negative controls (media alone) or the positive controls (media plus IL-lβ alone; Figure 1).
To further delineate the nature of higher tissue GAG concentration in PGJ2 groups, media GAG concentrations were analyzed. The GAG concentration in the 30μM PGJ2 group was significantly lower than the concentration in all other groups on days 3 to 18, except for the 15μM PGJ2 group on day 18. The 15μM PGJ2 group had significantly lower GAG concentration than the positive control group on days 15 and 18 (Figure 2). The ratios of the total tissue GAG concentration (explants) on day 21 to the sum of the tissue GAG concentration on day 21 and the total media GAG concentration over all time points for each respective group revealed that the groups treated with either concentration of PGJ2 had significantly greater retention of GAG in the tissue, compared with all other groups (Figure 3). The retention of GAG for the 30μM PGJ2 group was significantly greater than that for the 15μM PGJ2 group (P < 0.001).
The PGE2 concentrations were significantly (P ≤ 0.001) lower in the negative control, compared with the positive control and treatment groups, at all time points except for the 30μM PGJ2 group on day 18 (Figure 4). The 30μM PGJ2 treatment group had significantly lower PGE2 than did the positive control group at all time points except on days 12 and 18. The 30μM PGJ2 treatment group also had significantly lower PGE2 than the 15μM PGJ2 group on days 7, 9, and 15 and the 4.2μM and 2.5μM pioglitazone groups on days 3 to 9, 15, and 21. The PGE2 concentrations of the 4.2μM and 2.5μM pioglitazone groups were not significantly different from the positive control group at any time point.
To determine whether pioglitazone was being adequately metabolized in the coculture system, samples of media from pioglitazone treatment groups were analyzed for metabolites. No metabolites were detected.
Nitric oxide and PGE2 were examined as mediators of inflammation. The NO concentration of the negative control was significantly lower than that of the positive control group at all time points (Figure 5). The NO concentration of the 30μM PGJ2 treatment group was significantly lower than that of the negative control group on day 3 and days 12 to 21 and the positive control group at all time points (days 3 to 21 [P < 0.001]).
Figure 1—Hydroxyproline concentrations (mean ± SD) measured as an indication of collagen concentration in cartilage explants from dogs. Explants were untreated (Control) or treated with IL-1β alone (positive control) or IL-1β plus PGJ2 or pioglitazone
Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1142
Figure 1—Hydroxyproline concentrations (mean ± SD) measured as an indication of collagen concentration in cartilage explants from dogs. Explants were untreated (Control) or treated with IL-1β alone (positive control) or IL-1β plus PGJ2 or pioglitazone
Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1142
Figure 1—Hydroxyproline concentrations (mean ± SD) measured as an indication of collagen concentration in cartilage explants from dogs. Explants were untreated (Control) or treated with IL-1β alone (positive control) or IL-1β plus PGJ2 or pioglitazone
Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1142
Media GAG concentration (mean ± SD) associated with canine cartilage explant cultures in the same groups as in Figure 1. *Significantly (P < 0.05) different from positive control value. †Significantly (P < 0.05) different within treatment group at that time point.
Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1142
Media GAG concentration (mean ± SD) associated with canine cartilage explant cultures in the same groups as in Figure 1. *Significantly (P < 0.05) different from positive control value. †Significantly (P < 0.05) different within treatment group at that time point.
Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1142
Media GAG concentration (mean ± SD) associated with canine cartilage explant cultures in the same groups as in Figure 1. *Significantly (P < 0.05) different from positive control value. †Significantly (P < 0.05) different within treatment group at that time point.
Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1142
The NO concentration in the 15μM PGJ2 treatment group was significantly lower than that in the positive control group at all time points except day 7 and greater than the negative controls at days 3, 7, and 9. Nitric oxide concentration in the 4.2μM pioglitazone treatment group was significantly lower than that in the positive control and the 2.5μM pioglitazone groups at days 7 to 21.
Ratios (mean ± SD) of total tissue GAG concentration in explants on day 21 to the sum of the tissue GAG concentration on day 21 and the total media GAG concentration over all time points for explants groups as in Figure 1. See Figure 2 for key.
Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1142
Ratios (mean ± SD) of total tissue GAG concentration in explants on day 21 to the sum of the tissue GAG concentration on day 21 and the total media GAG concentration over all time points for explants groups as in Figure 1. See Figure 2 for key.
Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1142
Ratios (mean ± SD) of total tissue GAG concentration in explants on day 21 to the sum of the tissue GAG concentration on day 21 and the total media GAG concentration over all time points for explants groups as in Figure 1. See Figure 2 for key.
Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1142
Amounts of PGE2 in media of the same groups as in Figure 1.
Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1142
Amounts of PGE2 in media of the same groups as in Figure 1.
Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1142
Amounts of PGE2 in media of the same groups as in Figure 1.
Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1142
Concentrations of NO in media of the same groups as in Figure 1.
Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1142
Concentrations of NO in media of the same groups as in Figure 1.
Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1142
Concentrations of NO in media of the same groups as in Figure 1.
Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1142
Discussion
For quantitation of the GAG concentration of the media and cartilage explants, the DMMB assay was used. In a previous study9 the GAG concentration of the positive controls decreased significantly compared with the negative control. In the present study an overall decrease occurred but it was not significant, possibly because the positive control values were more variable. Nevertheless, the GAG concentration was significantly greater in both PGJ2 treatment groups, compared with all other groups, in explants harvested on day 21.
To gain a better understanding of the nature of the GAG retention in the explants, ratios of the total tissue GAG concentration (explants) on day 21 to the sum of the tissue GAG concentration on day 21 and the total media GAG concentration over all time points for each respective group were examined. This revealed that both PGJ2 groups had significantly greater retention of GAG, compared with all other groups (Figure 3). This indicated that PGJ2, but not pioglitazone, contributed to the maintenance of GAG within the cartilage explant. Also, there was a significantly decreased media GAG concentration in PGJ2 groups, which suggested that the maintenance of cartilage GAG concentration in these groups might be attributed to inhibition of proteoglycan degradation, although definitive statements regarding GAG synthesis or degradation cannot be made.
Nitric oxide and PGE2 are potent inflammatory mediators in osteoarthritis. Excess production of NO has been linked with inhibition of collagen synthesis, proteoglycan synthesis, and IL-1 receptor antagonist production as well as interference with integrin signaling, induction of chondrocyte apoptosis, stimulation of metalloproteinase production and activation, and inactivation of tissue inhibitors of MMPs.13 The PPARγ agonists inhibit NO, and the mechanism for inhibition of NO has been reported to be associated with effects on mitogen-activated protein kinase kinase kinase-1 induction of iNOS promoter activity.4 Fahmi et al4 reported that PGJ2 inhibited delta MEKK-1-induced activator protein-1 and nuclear factor-kappa B-luciferase reporter plasmid activation through a PPARγ-dependent pathway. Colville-Nash et al14 reported that PGJ2 may also affect the stress protein, heme oxygenase-1, and in doing so inhibit iNOS.
In the present study, the NO concentration measured via the Griess reaction in the positive control group was greatly increased, compared with that in the negative control group, consistent with previous findings.9 The 4.2μM pioglitazone treatment group had significantly reduced NO concentration, compared with the positive group and 2.5μM pioglitazone group, whereas the NO concentration in the 2.5μM pioglitazone group was not significantly different from the positive control group, suggesting a possible dose-related effect of pioglitazone. The PGJ2 treatment groups also had a similar dose relationship. The lower NO concentration in the high-concentration PGJ2 group was of particular note because it was less than that of the negative control group even though it was treated with IL-1β.
Although it is unclear whether the alteration of NO production in the PGJ2 groups caused the changes in GAG concentration, NO production was greatly reduced by administration of PPARγ agonists. Not only did the higher concentration of the PGJ2 treatment (30μM) appear to be more efficient at reducing NO production than did the lower concentration treatment (15μM), but it also induced particularly striking results by reducing the NO concentration to less than that of the negative controls. This also supported previous findings that iNOS induction is suppressed in a dose-dependent manner by PGJ2.4 Interestingly, the 4.2μM pioglitazone treatment also appeared to consistently reduce NO production, compared with the positive controls, but the 2.5μM pioglitazone treatment did not.
The significant reduction in PGE2 in association with PPARγ treatment could also prove beneficial in osteoarthritis by reducing inflammation and related degradation.15 Several mechanisms have been proposed for the action of PPARγ agonists as anti-inflammatory drugs. In 2002, Fahmi et al4 examined human osteoarthritis chondrocytes and noted that PGJ2 did not inhibit PGE2 through a COX-2 pathway. Recently, it was reported that PGJ2 can act directly on microsomal PGE synthase type 2 to prevent PGE2 production in colorectal cells.16
In the present study, the PGE2 concentration of the positive control group was greatly increased, compared with the negative control group, consistent with previous findings.9 The 30μM PGJ2 treatment group had significantly lower PGE2 concentration, compared with the IL-1β positive control group at most time points and all other PPARγ treatments at multiple time points. The 4.2μM and 2.5μM pioglitazone groups did not have significantly reduced PGE2 concentration, compared with the positive control group, at any time point. The data confirmed previous findings in regard to PGJ2, but it was still unclear why pioglitazone had no observed suppressive effect.
The major limitation of this study was its in vitro nature, and although the system has been determined to closely mimic in vivo conditions, it would be presumptuous to assume that the interactions in a culture system completely explain the complex conditions in vivo. One notable difference in the experimental system used in this study, compared with the previous study,9 is that it incorporated not only cartilage but also synovium in an attempt to resemble the entire joint environment as closely as possible.
We hypothesized that PGJ2 and pioglitazone would be both anti-inflammatory and anticatabolic in the coculture system. The PGJ2 reduced NO and PGE2 production while preserving tissue GAG concentration, compared with the positive control. Pioglitazone reduced NO production, compared with the positive control, but failed to preserve GAG concentration or reduce PGE2 production. Therefore, despite apparent beneficial effects of PGJ2 on joint health, pioglitazone did not markedly reduce inflammatory responses and matrix alterations in cartilage exposed to IL-1β. Pioglitazone did not cause detrimental effects in the cartilage, compared with the negative control, but it appears that pioglitazone does not have substantial chondroprotective effects in this in vitro system despite recent notable reports5,8 in animal models. This could be attributable to the fact that metabolism to more active forms may be required because pioglitazone is a prodrug and must be bioconverted to active forms.10 This bioconversion is not facilitated in a culture system where proper metabolic pathways for bioconversion, usually found in the liver and kidneys,17 may not be sufficient or even present. To definitively answer the question of whether pioglitazone was metabolized in the coculture model, we conducted liquid chromatography-mass spectrometry analysis of multiple samples of media in both pioglitazone concentration groups and found only trace amounts of metabolized pioglitazone. This finding confirmed that the metabolites were not produced in substantial amounts, and we believe this is the most likely reason that results for the pioglitazone groups did not more closely resemble those for the PGJ2 groups.
Taken together, these data suggest that targeting PPARγ would be a potential treatment strategy for osteoarthritis. The endogenous PPARγ agonist, PGJ2, appears to induce anti-inflammatory and chondroprotective effects in a simulated osteoarthritis joint environment; however, we were not able to conclude that the synthetic PPARγ agonist, pioglitazone, has analogous effects. Further investigation is needed to explain this finding and elucidate the actions of PPARγ agonists in osteoarthritis.
Abbreviations
| COX-2 | Cyclooxygenase 2 |
| DMMB | 1,9-dimethylmethylene blue |
| GAG | Glycosamingoglycan |
| IL | Interleukin |
| iNOS | Inducible nitric oxide synthase |
| MMP | Matrix metalloproteinase |
| NO | Nitric oxide |
| PGE2 | Prostaglandin E2 |
| PGJ2 | 15-deoxy-Δ 12, 14-prostaglandin J2 |
| PPARγ | Peroxisome proliferator-activated receptor gamma |
Becton-Dickinson, Franklin Lakes, NJ.
BD Biosciences, San Jose, Calif.
Sigma-Aldrich, St Louis, Mo.
Promega Corp, Madison, Wis
Caymen Chemical, Ann Arbor, Mich.
ThermoFinnigan, San Jose, Calif.
GL Sciences Inc, Tokyo, Japan.
SigmaStat, Jandel Scientific, San Rafael, Calif
References
- 1↑
Lawrence RCHelmick CGArnett FC, et al. Estimates of the prevalence of arthritis and selected musculoskeletal disorders in the United States. Arthritis Rheum 1998;41:778–799.
- 2↑
Sakkas LIPlatsoucas CD. The role of T cells in the pathogenesis of osteoarthritis. Arthritis Rheum 2007;56:409–424.
- 3↑
Stadler JStefanovic-Racic MBilliar TR, et al. Articular chondrocytes synthesize nitric oxide in response to cytokines and lipopolysaccharide. J Immunol 1991;147:3915–3920.
- 4↑
Fahmi HDi Battista JAPelletier JP, et al. Peroxisome proliferator-activated receptor gamma activators inhibit interleukin-lbeta-induced nitric oxide and matrix metalloproteinase 13 production in human chondrocytes. Arthritis Rheum 2001;44:595–607.
- 5↑
Boileau CMartel-Pelletier JFahmi H, et al. The peroxisome proliferator-activated receptor gamma agonist pioglitazone reduces the development of cartilage lesions in an experimental dog model of osteoarthritis: in vivo protective effects mediated through the inhibition of key signaling and catabolic pathways. Arthritis Rheum 2007;56:2288–2298.
- 6
Sabatini MBardiot ALesur C, et al. Effects of agonists of peroxisome proliferator-activated receptor? on proteoglycan degradation and matrix metalloproteinase production in rat cartilage in vitro. Osteosteoarthritis Cartilage 2002;10:673–679.
- 7
Koufany MMoulin DBianchi A, et al. Anti-inflammatory effect of antidiabetic thiazolidinediones prevents bone resorption rather than cartilage changes in experimental polyarthritis. Arthritis Res Ther 2008;10:R6.
- 8↑
Kobayashi TNotoya KNaito T, et al. Pioglitazone, a peroxisome proliferator-activated receptor gamma agonist, reduces the progression of experimental osteoarthritis in guinea pigs. Arthritis Rheum 2005;52:479–4687.
- 9↑
Cook JLKuroki KStoker AM, et al. Review of in vitro models and development and initial validation of a novel co-culture model for the study of osteoarthritis. Curr Rheumatol Rev 2007;3:172–182.
- 11↑
Farndale RWButtle DJBarrett AJ. Improved quantitation and discrimination of sulphated glycosaminoglycans by use of dimethylmethylene blue. Biochim Biophys Acta 1986;883:173–177.
- 12↑
Reddy GKEnwemeka CS. A simplified method for the analysis of hydroxyproline in biological tissues. Clin Biochem 1996;29:225–229.
- 13↑
Lotz M. The role of nitric oxide in articular cartilage damage. Rheum Dis Clin North Am 1999;25:269–282.
- 14↑
Colville-Nash PRQureshi SSWillis D, et al. Inhibition of inducible nitric oxide synthase by peroxisome proliferator-activated receptor agonists: correlation with induction of heme oxygenase 1. J Immunol 1998;161:978–984.
- 15↑
Abramson SBAttur MAmin AR, et al. Nitric oxide and inflammatory mediators in the perpetuation of osteoarthritis. Curr Rheumatol Rep 2001;3:535–541.
- 16↑
Schröder OYudina YSabirsh A, et al. 15-deoxy-delta 12, 14-prostaglandin J2 inhibits the expression of microsomal prostaglandin E synthase type 2 in colon cancer cells. J Lipid Res 2006;47:1071–1080.
- 17↑
Kiyota YKondo TMaeshiba Y, et al. Studies on the metabolism of the new antidiabetic agent pioglitazone. Identification of metabolites in rats and dogs. Arzneimittelforschung 1997;47:22–28.
