The present study demonstrates that adiponectin increased NO and three MMPs production in human OA chondrocytes mainly via the AMPK-JNK pathway in vitro and that adiponectin-induced NO and MMPs lead to accelerated degradation of OA cartilage matrix ex vivo.
Our in vitro
findings indicate that adiponectin is a potential catabolic mediator in OA. This is in line with the previous findings that adiponectin induces iNOS, MMP-3, MMP-9, and MCP-1 in murine chondrocytes [13
]. More important, increased cartilage degradation products after adiponectin treatment further supports that in vitro
catabolic activity induced by adiponectin is relevant to cause cartilage degradation. Our result is in parallel with the result of a recent study indicating that the synovial fluid levels of adiponectin are correlated with aggrecan degradation markers in patients with knee OA [19
]. However, Chen et al
] reported that adiponectin did not alter the expression levels of MMP-3 and MMP-13 mRNA. The contrasting results regarding the effect of adiponectin might be due to experimental conditions. Chen et al
. used chondrocytes from the OA knees with diverse severities and evaluated the effects in monolayered cells at passages 3 to 7 [12
], whereas we isolated chondrocytes from the OA knees with Kellgren-Lawrence grade 3 or 4 and grew them in suspension at passage 0. Because OA chondrocyte behavior and phenotypes can be affected by the surrounding matrix state, culture methods, and passage numbers [20
], this might have contributed to the difference of adiponectin-induced responses in each study.
Another possibility is a different composition of adiponectin isoforms due to a different biologic source from which adiponectin is produced. Native adiponectin has a multimeric structure and circulates in blood as trimers, hexamers, and high-molecular-weight (HMW) complexes [14
]. Biologic effects mediated by adiponectin have been considered to be isoform dependent. HMW adiponectin has pro-inflammatory effects [21
], whereas the low-molecular-weight (LMW) isoform has antiinflammatory functions in human leukocytes and monocytic cells [23
]. We used HEK293 cell-derived full-length adiponectin, the most abundant isoforms of which are hexamers and HMW forms, followed by trimers [25
]. This composition is similar to that of human OA synovial fluid in which hexamers and HMW forms are the most abundant isoforms [12
]. Conversely, full-length adiponectin derived from Escherichia coli
lacks HMW forms [25
]. Morevoer, adiponectin of the same isoform could display a different potency to induce a biologic response depending on whether it is E. coli
derived or mammalian cell derived [25
]; adiponectin produced in mammalian cells seems to be functionally more potent than bacterially produced adiponectin because the HMW form is a predominantly active form. Because it is believed that E. coli
-derived adiponectin was used in the previous studies [12
], pro-inflammatory effects of adiponectin might not have been fully developed in those studies.
Biologic effects of adiponectin are mediated mainly through two receptors, AdipoR1 and AdipoR2, and these two receptors are believed to activate different signaling pathways; AdipoR1 activates the AMPK pathway, whereas AdipoR2 is linked more closely with the peroxisome proliferator-activated receptor α (PPAR-α) pathway in the liver [27
]. Chen et al
] showed that human cartilage expressed only AdipoR1. However, our study showed that both AdipoR1 and AdipoR2 are expressed in human cartilage tissue, consistent with the results of Lago et al.
]. A heterogeneous distribution of AdipoR1 and AdipoR2 on chondrocytes might be a factor that explains the difference between our results and those of the others. In our study, the expression of AdipoR2 was higher in terms of the immunostaining intensity as well as the percentage of stained cells, but the increase rate of AdipoR1 was as twice as high as that of AdipoR2, when nonlesional and lesional cartilage areas were compared. This finding suggests that the change of AdipoR1 expression might better reflect the cartilage catabolic state than that of AdipoR2, and that the AdipoR1-AMPK pathway could be associated with cartilage catabolism.
It has been well established that adiponectin activates AMPK [14
]. Lago et al
] reported that the AMPK/Akt signaling pathway is involved in iNOS and MMP-3 induction by adiponectin in the murine chondrocyte ATDC5 cell line. In addition, adiponectin activated the AMPK/p38/NF-κB axis in human synovial fibroblasts to induce IL-6 production [15
]. Conversely, in our study, AMPK/JNK pathways are the major signaling pathway involved in adiponectin-mediated induction of iNOS and MMPs in human OA chondrocytes, whereas the AMPK/Akt or AMPK/p38 pathway is partially involved in MMP-13 or MMP-3 induction, respectively. The JNK pathway is one of the signaling intermediates activated by adiponectin [28
], and adiponectin-induced JNK activation has been shown to follow AMPK activation [30
]. Furthermore, JNK is involved in MMPs and iNOS expression in human articular chondrocytes [32
]. Therefore, we expect that adiponectin induces iNOS and MMP expression via JNK downstream to AMPK in human chondrocytes and that the AMPK/JNK axis is a major signaling system responsible for the adiponectin-induced degradation of cartilage matrix.
Because NO can upregulate the expression or activity of MMPs [17
], we determined whether NO mediates adiponectin-induced synthesis of MMPs. Unexpectedly, the expression of MMPs was further increased by adiponectin after pretreatment with a nonspecific NOS and a specific iNOS inhibitor. This finding is consistent with the previous observation by Hattori et al
] in which adiponectin-induced NF-κB activation was further enhanced by a nonspecific NO inhibitor, L-NMMA, in human umbilical vein endothelial cells. Interestingly, LY294002, a PI3-K/Akt kinase inhibitor, significantly suppressed NO production, whereas it caused a higher MMP-3 production in adiponectin-treated ATDC5 cells in the study of Lago et al.
]. In this context, we are tempted to speculate that NO serves as a negative-feedback regulator of adiponectin action in cartilage, and that this negative feedback may lead to the delayed effects of adiponectin on the OA cartilage catabolism when compared with those of IL-1β in our study. The role of NO as a catabolic mediator has been controversial. The protective effect of NO on cartilage degradation has been shown by several studies [38
], in which the treatment with NOS inhibitors accelerated the proteoglycan breakdown by increasing MMP levels in culture media [38
]. Thus, the exact role of NO in cartilage homeostasis seems to be complex. Further studies on the effect of NO on AMPK or JNK activation in chondrocytes will elucidate the mechanisms by which NO influences adiponectin-induced MMP production.
We used the highest dosage (30 μg/ml) of adiponectin with maximal biologic activity to investigate the full catabolic potential of adiponectin. Because adiponectin concentrations in OA synovial fluid are typically lower (1 to 5 μg/ml) than the doses used in our study [11
], a possibility exists that the catabolic effect of adiponectin is overemphasized in our study. However, the human OA joint tissues including cartilage were reported to release adiponectin in ex vivo
culture study [11
], and ATDC5 cells have been shown to express adiponectin themselves in an autocrine manner [41
]. Therefore, the actual concentrations of adiponectin might be higher in the microenvironment surrounding chondrocytes than those measured in OA synovial fluid.