Osteoarthritic infrapatellar fat pad aggravates cartilage degradation via activation of p38MAPK and ERK1/2 pathways

This study aimed to investigate the biochemical effects of osteoarthritic infrapatellar fat pad (IPFP) on cartilage and the underlying mechanisms. Human IPFP and articular cartilage were collected from end-stage osteoarthritis (OA) patients during total knee arthroplasty. IPFP-derived fat-conditioned medium (FCM) was used to stimulate human primary chondrocytes and cartilage explants. Functional effect of osteoarthritic IPFP was explored in human primary chondrocytes and articular cartilage in vitro and ex vivo. Activation of relative pathways and its effects on chondrocytes were assessed through immunoblotting and inhibition experiments, respectively. Neutralization test was performed to identify the main factors and their associated pathways responsible for the effects of IPFP. Osteoarthritic IPFP-derived FCM significantly induced extracellular matrix (ECM) degradation in both human primary chondrocytes and cartilage explants. Several pathways, such as NF-κB, mTORC1, p38MAPK, JNK, and ERK1/2 signaling, were significantly activated in human chondrocytes with osteoarthritic IPFP-derived FCM stimulation. Interestingly, inhibition of p38MAPK and ERK1/2 signaling pathway could alleviate the detrimental effects of FCM on chondrocytes, while inhibition of other signaling pathways had no similar results. In addition, IL-1β and TNF-α instead of IL-6 in osteoarthritic IPFP-derived FCM played key roles in cartilage degradation via activating p38MAPK rather than ERK1/2 signaling pathway. Osteoarthritic IPFP induces the degradation and inflammation of cartilage via activation of p38MAPK and ERK1/2 pathways, in which IL-1β and TNF-α act as the key factors. Our study suggests that modulating the effects of IPFP on cartilage may be a promising strategy for knee OA intervention.


Introduction
Osteoarthritis (OA) is the most frequent joint disorder, characterized by structural alterations involving the whole joint. The pathogenesis of OA includes cartilage degradation, subchondral bone remodeling, osteophyte formation, and synovial inflammation, which leads to joint pain, tenderness, swelling, stiffness, and limitation of movement [1,2]. It has caused a huge socioeconomic burden worldwide due to its disabling characteristics and an unmet medical need [3,4]. Investigation into the pathogenesis of OA and searching for the novel prevention and treatment strategies are of great significance.
Infrapatellar fat pad (IPFP) is the largest local adipose tissue around the knee joint. Researchers found that IPFP released a variety of bioactive factors including proinflammatory and anti-inflammatory mediators as well as catabolic and anabolic components. It acts as an important source of cytokines in knee synovial fluids [5,6]. Our previous studies revealed that IPFP maximal area as well as a larger volume measured by magnetic resonance imaging had a protective role for the knee symptoms and cartilage degradation [7][8][9][10], while signal intensity alterations within IPFP facilitated the progression of OA [11][12][13][14]. However, there are still controversies about the role of IPFP in OA and further studies are warranted.
It has been reported that identified risk factors including aging, joint injury, genetics, and obesity initiate or promote OA by activating many different molecular pathways [15]. IPFP-derived factors contain IL-1β, TNF-α, IL-6, growth factors, adipocytokines, and other known or unknown components, which may also activate different pathways to interact with cartilage or other adjacent tissues in advanced OA [16]. Some of them are well investigated; however, the overall role of IPFP-released factors in the progression of OA is poorly clarified and the underlying biochemical mechanisms remain unclear.
In this study, we aimed to investigate the biochemical effects of osteoarthritic IPFP on cartilage by measuring its effects on the expression of cartilage degrading enzymes and inflammatory mediators, and to identify the leading cytokines as well as the potential signal transduction pathways.

Articular cartilage and IPFP collection
This study was approved by the Ethics Committee of Zhujiang Hospital of Southern Medical University (Guangzhou, China). Human IPFP and cartilage were aseptically collected as surgical waste from end-stage knee OA patients (n = 23) who had undergone total knee arthroplasty. Informed consent form was signed by each patient before the surgery. Further information about the individuals is summarized in the Supplementary Table 1. Human articular cartilage tissues were the undamaged area of cartilage from end-stage knee OA patients who had undergone total knee arthroplasty. All the experiments in this study were independently repeated for at least three times.

Preparation of infrapatellar fat pad-conditioned medium (FCM)
To generate FCM, the inner part of the IPFP was carefully separated from the synovium lining. Then, the fat pad, minced into pieces of approximately 50 mg, was immediately incubated in Dulbecco's Modified Eagle's Medium/ F12 (DMEM/F12) containing 0.75 mg/ml BSA, 100 U/ml penicillin, and 0.1 mg/ml streptomycin at a concentration of 100 mg/ml. After incubation for 2 h, the medium was refreshed to remove possible contamination. In another 24 h, the supernatants were harvested, centrifuged at 300 g for 8 min, and frozen at -80 ℃ until use.

Isolation and culture of chondrocytes
To isolate the human articular chondrocytes, knee cartilage was washed with phosphate-buffered saline and cut into small particles with a diameter about 1 mm. Subsequently, the cartilage tissues were digested in trypsin-EDTA solution containing 0.25% trypsin and 0.02% EDTA for 30 min at 37 °C with gentle agitation, followed by a digestion in 0.2% type 2 collagenase overnight. The digested suspension was then filtered through a 40-μm mesh and centrifuged at 300 g for 5 min. Afterward, cell pellets were resuspended and seeded in 25 cm 2 culture flasks. The medium was refreshed every 2 days.

RNA extraction and quantitative real-time polymerase chain reaction (qRT-PCR)
Total RNA was isolated from treated chondrocytes using an RNAiso Plus kit (Takara), and then was reverse transcribed into cDNA using a PrimeScript ™ RT Master Mix kit (Takara). Afterward, cDNA was amplified by qRT-PCR using TB Green ™ Premix Ex Taq ™ (Takara) and LightCycler 480 System (Roche) with specific primers. The glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as an internal control gene. Results were calculated by 2 −△△Ct formula and presented as fold changes relative to GAPDH. Primer sequences are listed in Supplementary Table 2.

Immunoblotting
Chondrocytes were lysed with RIPA lysis buffer (KeyGEN) supplemented with protease inhibitors. Total protein was collected and denatured. After a 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis, proteins were transferred to a polyvinylidene difluoride membrane (Millipore). Blots were further probed with primary antibodies (listed in Supplementary Table 3) and horseradish peroxidase-conjugated secondary antibodies (ABclonal). After washing, the membrane was visualized using an enhanced chemiluminescence kit (Ncm Biotech).

Histological and immunohistochemical analysis
Articular cartilage was diced into small pieces which were randomly and evenly divided into two groups: one was treated with autologous FCM and the other with the medium that was used to generate the FCM. The medium was changed every 2 days. After 14 days of culture, the cartilage explants were fixed and decalcified. Subsequently, the tissues were embedded in paraffin and sectioned at 4-μm thickness. The histological sections were stained to detect the histomorphology and extracellular matrix of cartilage explants with Safranin O/Fast Green (Sigma) as well as Toluidine blue (Solarbio). For immunohistochemical (IHC) analysis, the sections were stained overnight at 4 °C with the following primary antibodies: rabbit anti-MMP3 (1:50, Abcam, ab52915), rabbit anti-COX2 (1:200, Abcam, ab179800), and rabbit anti-Collagen II (1:200, Abcam, ab34712). More detailed information of the procedure is provided in supplementary materials.

Inhibitor experiments
The effect of relative pathway activation on chondrocytes was examined through pathway inhibition. To assess the pathway-inhibited effects of different inhibitors at appropriate concentrations, cells were pretreated for 2 h with the following pathway inhibitors: BAY 117,082 (BAY, 1 μM, an NF-κB inhibitor, Selleck), Rapamycin (Rapa, 10 nM, an mTORC1 inhibitor, Sigma), SB203580 (SB, 10 μM, a p38MAPK inhibitor, Selleck), SP 600,125 (SP, 5 μM, a JNK inhibitor, MCE), and U0126 (U0, 10 μM, an ERK inhibitor, MCE), respectively. Then, these chondrocytes were further stimulated with or without 50% FCM for 30 min in the presence of relative pathway inhibitors. Proteins were collected to do the next western blotting analysis.
To detect the influence of relative pathway inhibition on expressions of degrading enzymes and pro-inflammatory cytokines in chondrocytes, cells were treated with or without autologous 50% FCM for 24 h following pretreatment with relative pathway inhibitor for 2 h. And then, total RNA and proteins were extracted to do the further qRT-PCR or western blotting analysis.

Neutralization tests
An early study suggested that adipokines such as leptin and adiponectin secreted by IPFP may not be responsible for cartilage destruction and inflammation [17]. To identify the main factors that mediated the FCM-induced degradation and inflammation of cartilage, neutralization experiments were performed. Osteoarthritic IPFP could secret higher levels of inflammatory mediators such as IL-6 and TNF-α compared with paired subcutaneous adipose tissues, and IL-1β, IL-6, and TNF-α have long been viewed as the main inflammatory factors in OA pathophysiology [18]. To this end, we first chose neutralizing antibodies targeting these cytokines.
FCMs were incubated overnight with neutralizing antibodies targeting human IL-1β, IL-6, and TNF-α, respectively. FCM incubated with normal IgG was set as the negative control. Subsequently, chondrocytes were stimulated with relative FCM for 24 h, and then, the total RNA and proteins were extracted to further analyze the expressions of degrading enzymes and inflammatory mediators. The information of neutralizing antibodies is depicted in Supplementary Table 3.

Statistical analysis
All experiments were conducted at least three separate times independently. For comparison of parametric variables, differences between two groups were analyzed through Student's t test, while those among three or more groups were analyzed by one-way analysis of variance (ANOVA). To compare the nonparametric variables, the Kruskal-Wallis H test was conducted. All statistical analyses were performed with SPSS 24.0. Data were presented as the mean ± SD, and P < 0.05 was considered as significant.

Osteoarthritic IPFP-derived FCM induces catabolic and inflammatory phenotypes in human primary chondrocytes
The role of IPFP-derived factors on chondrocytes remains controversial. Therefore, we determined the effect of FCM to human articular chondrocytes first. FCM that was harvested by incubation of IPFP pieces, was subsequently added to human primary chondrocytes at different concentrations (Fig. 1A). Little effect of IPFP-derived FCM on cell viability was observed by CCK8 assay (Supplementary Figure S1). qRT-PCR indicated that FCM upregulated the expression of ECM catabolic markers including MMP1, MMP3, and ADAMTS4 in a dosedependent manner (Fig. 1B). Similarly, the expressions of inflammatory genes such as IL-1B, IL-6, and COX2 were increased consistently, especially in the 100% FCM group (Fig. 1C). Moreover, we examined the expression of chondrogenic-specific genes (such as SOX9, COL2A1, and ACAN). Only SOX9 and COL2A1 showed a trend of downregulation in the concentration 100% and little effects were detected on the expression of ACAN (Fig. 1D).
In consistent with qRT-PCR results, the protein levels of MMP1, MMP3, ADAMTS4, and COX2 in chondrocytes were significantly enhanced with the increase of FCM concentrations ( Fig. 1E-I). Taken together, our data indicate that osteoarthritic IPFP-derived FCM could induce a catabolic and inflammatory phenotype in human chondrocytes.

FCM promotes the degradation of cartilage matrix and aggravates the inflammation of cartilage ex vivo
To investigate the effects of osteoarthritic IPFP on cartilage tissue, cartilage explants were treated with or without IPFPderived FCM ex vivo for 14 days (Fig. 2A). Loss of proteoglycans was detected through safranin o/fast green-staining in FCM-treated cartilage explants but no significant cartilage erosion compared with control group (non FCM-treated group) (Fig. 2B). The similar result can be determined by toluidine blue staining of the cartilage slices (Fig. 2C). Subsequently, we performed IHC staining and observed a decrease in Collagen II in FCM-treated tissues (Fig. 2D). Furthermore, we also examined markers of inflammation and degradation through IHC in cartilage explants. The number of MMP3 or COX2 positive cells increased significantly Representative results are shown as mean ± SD, Statistical difference: ns not significant, *P < 0.05, **P < 0.01 and ***P < 0.001 compared with the CTRL. FCM fat-conditioned medium, IPFP infrapatellar fat pad, MMP matrix metalloproteinase, ADAMTS A disintegrin and metalloproteinase with thrombospondin-like motifs, IL interleukin, COX2 cyclooxygenase 2, ACAN aggrecan, GAPDH glyceraldehydephosphate dehydrogenase in the samples treated with FCM compared with the control (Fig. 2E-H). Taken together, our data suggest that osteoarthritic IPFP plays a detrimental role in cartilage degradation.

Signaling pathways associated with knee OA progression were activated within FCM-stimulated chondrocytes
Some classic signaling pathways, such as NF-κB, MAPK, and mTOR, have being reported to be associated with OA [19][20][21]. To explore the mechanisms behind the detrimental effects of FCM derived from advanced knee OA on cartilage, we detected these signaling in 0-120 min after FCM stimulation. Group 0 refers to the non-conditioned medium, while the other groups are 50% FCM. Both the non-conditioned medium and 50% FCM were formulated before the experiments and all experimental operations between the groups were independent and separate, so that there was no mechanical disturbance or change of newly added components. The results showed that p65, a member in NF-κB family which plays a pivotal role in inflammatory and immune responses, was significantly phosphorylated within 5-60 min after stimulation with FCM (Fig. 3A). S6, a downstream molecule of mTOC1, was also phosphorylated in the presence of FCM (Fig. 3B). At the same time, phosphorylation levels of markers in all three pathways of MAPK including P38, JNK, and ERK1/2 were enhanced by FCM at different time points (Fig. 3C-E). These results obtained by western blotting analysis suggest that the biochemical effects of FCM on cartilage are probably mediated by one or more pathways above.
Subsequently, human chondrocytes were cultured for 24 h with IPFP-derived FCM after pretreatment with or without inhibitors for 2 h. Total RNA and proteins were extracted to determine the expression of ECM catabolic and inflammatory markers. The mRNA expressions of ECM-degrading enzymes including MMP1, MMP3, and ADAMTS4 were substantially upregulated in chondrocytes co-cultured with FCM, while downregulated in the presence of SB or U0 except for ADAMTS4 (Fig. 4F). Inflammatory mediators including IL-1B, IL-6, and COX2 were also upregulated after stimulation by FCM. In the groups pretreated with SB or U0, the expression of IL-6 and COX2 was decreased, while IL-1B expression was downregulated only in the presence of U0 (Fig. 4G). These results were further supported by protein expressions including MMP1, MMP3, and COX2 (Fig. 4H). Taken together, these findings suggest that p38MAPK and ERK1/2 pathways rather than NF-κB, mTORC1 or JNK are involved in the adverse effects of FCM on articular chondrocytes.

Pro-inflammatory and pro-catabolic effects of FCM on articular chondrocytes are reduced after neutralization of IL-1β and TNF-α, not IL-6
To identify the main factors that mediate the inflammatory and degradative effect of FCM, human chondrocytes were cultured with FCM which was pretreated with or without a neutralizing antibody or a control IgG. Consistently, osteoarthritic IPFP-derived FCM enhanced the expressions of catabolic and inflammatory genes, which was almost unaffected by normal mouse IgG (Fig. 5A, B). As expected, neutralizing IL-1β or TNF-α in FCM resulted in the decrease of ECM-degrading enzymes and inflammatory markers, though only a trend for MMP1 mRNA after blocking IL-1β was observed. The decrease became even more pronounced when the multiple antibodies were used (Fig. 5A, B ). Surprisingly, the addition of IL-6 neutralizing antibody to FCM before stimulation of chondrocytes had little impact on the effects of FCM (Fig. 5A, B). Similarly, these results were further supported by western blotting analysis (Fig. 5C). Our data suggest that IL-1β and TNF-α, instead of IL-6, might play a decisive role in the pro-catabolic and pro-inflammatory effects of FCM.

Discussion
In the current study, we explored the biochemical effects of human osteoarthritic IPFP on cartilage by examining the responses of chondrocytes and cartilage explants Pro-inflammatory and degrading effects of FCM on chondrocytes diminished after neutralization of relative cytokines (IL-1β and TNF-α) in FCM due to the inhibition of p38MAPK signaling. FCM was pretreated with mouse normal IgG, mouse anti-human IL-1β, IL-6, TNF-α antibody, or multiple antibodies targeting IL-1β, IL-6, and TNF-α overnight, and then incubated with human chondrocytes for 24 h. Blank control group did not use FCM. qRT-PCR analysis of (A) degrading enzymes (MMP1, MMP3, and ADAMTS4) and (B) inflammatory mediators (IL-1B, IL-6, and COX2) expression (n = 3). C Western blotting analysis of the protein levels of MMP1, MMP3, COX2, and ADAMTS4 in different groups. D-G Western blotting analysis of the effects of blocking IL-1β (D), IL-6 (E), TNF-α (F), or all of them (G) in FCM on pathway activation including NF-κB, p38MAPK, and ERK1/2 signaling. Results expressed as mean ± SD. Statistical difference: ns, not significant, *P < 0.05, **P < 0.01 and ***P < 0.001 compared with the CTRL or 50% FCM group. CTRL, control group; FCM, fat-conditioned medium; IL-1β-Ab, IL-6-Ab, TNF-α-Ab, and Multi-Ab represented neutralizing antibodies targeting IL-1β, IL-6, TNF-α, and all of them, respectively to IPFP-derived FCM. Our data show that IPFP, at the end-stage of knee OA, has a pro-inflammatory and procatabolic phenotype for adjacent cartilage. The possible mechanism is that IPFP-released factors activate different molecular pathways, especially the p38MAPK and ERK1/2 signaling, among which IL-1β and TNF-α are the key drivers (Fig. 6).
A previous study found that IPFP adipocyte-released factors could induce a dose-dependent decrease in chondrocyte number and downregulate the mRNA level of chondrogenicspecific genes [22]. Similarly, the same study also indicated that co-culturing with IPFP adipocytes, chondrocytes showed an enhanced expression of genes including IL-1B, COX2, MMP1, and MMP3, suggesting that the inflammatory and matrix degrading effects of IPFP may largely depend on the adipocytes [22]. However, Bastiaansen-Jenniskens et al. reported that IPFP from patients with end-stage OA inhibited catabolic metabolism of bovine cartilage [23]. Interestingly, others did the similar experiments and came to the opposite conclusion [24]. Our data are in agreement with those presented by Gross et al. who found detrimental effects of IPFP-derived FCM on human chondrocyte in spite of the heterogeneity that might happen in human samples from different individuals [17]. Here, we examined the role of osteoarthritic IPFP not only through cellular tests but also experiments ex vivo, reaching the consistent conclusion. Moreover, the underlying mechanisms were explored. It is well established that the role of IPFP during OA includes the aspects of biomechanics, biochemistry, and damage repair [25,26]. However, this study mainly focused on its biochemical effects at the end-stage of knee OA. Further studies are warranted to assess the combined role of biomechanics and biochemistry within IPFP.
To the best of our knowledge, this is the first study to investigate the signaling pathways that mediate the interaction between osteoarthritic IPFP and chondrocytes. Our data indicate that p38MAPK and ERK1/2 might be the main pathways responsible for the pro-inflammatory and pro-catabolic effects of osteoarthritic IPFP, though all five examined pathways including NF-κB, mTORC1, p38MAPK, JNK, and ERK1/2 are activated by FCM in chondrocyte. NF-κB signaling has long been viewed as the essential pathway, which is involved in cartilage inflammation and matrix degradation during the process of OA [19]. Chen et al. reported that medium conditioned by IPFP adipocytes could induce the activation of NF-κB signaling, increasing the level of inflammatory markers in chondrocyte [22]. However, further reverse effect by blocking the pathway has not been verified in their study. In our data, NF-κB pathway seems not to be the main player that mediates the effects of IPFP, because NF-κB inhibition could not improve the inflammatory and degradative effects of FCM. This may also be attribute to the complexity of the components contained in FCM and the counteraction from other pathways. An earlier study has demonstrated that leptin combined with IL-1 activates MAPK and NF-κB signaling in chondrocytes [24]. This study supports our results, because both leptin and IL-1 are contained in FCM. mTORC1 pathway also plays an key Osteoarthritic IPFP-released factors activate different molecular pathways, especially the p38MAPK and ERK1/2 signaling, in which, IL-1β and TNF-α rather than IL-6 are the key players. IPFP, infrapatellar fat pad; IL, interleukin; TNF-α, tumor necrosis factor-α; MAPK, mitogenactivated protein kinase; ERK, extracellular signal-regulated kinase. SB203580, a p38MAPK inhibitor; U0126, an ERK1/2 inhibitor; MMP matrix metalloproteinase, ADAMTS A disintegrin and metalloproteinase with thrombospondin-like motifs; COX2 cyclooxygenase 2 role in cartilage metabolism and pathophysiology of OA [20]. The activation of mTORC1 in chondrocyte by gene modification could initiate OA in mice [27]. However, in the current study, we did not find any improvement in the inflammatory and chondro-destructive markers after inhibiting mTORC1 in FCM-induced chondrocytes, suggesting that mTORC1 might not be the main signaling that mediates IPFP-induced effects.
MAPK is a key upstream signaling leading to cartilage destruction, including members of p38MAPK, JNK, and ERK1/2 [28]. Sondergaard et al. reported both p38MAPK and ERK1/2 signaling were essential for MMPs expression and activity, while only ERK1/2 was essential for aggrecanase-mediated degradation [21]. This could explain why the expression of MMP1 and MMP3 was significantly downregulated when p38MAPK or ERK1/2 was inhibited in our study. Moreover, our results showed ERK1/2 rather than p38MAPK inhibition could downregulate ADAMTS4 expression, though only a trend was observed, which is also in accordance with the finding by Sondergaard et al. [21]. Although JNK signaling may function in the process of OA [29], an early study found that JNK pathway might be of little or no importance for IL-1β-induced IL-6 expression [30]. Hui et al. suggested JNK inhibition had no effect on MMP13 expression of chondrocyte stimulated by leptin, as well as cartilage collagen release induced by leptin and IL-1 [24]. Similarly, in our study, JNK inhibition showed no effect on the improvement of cartilage damage markers induced by FCM. Overall, our data highlighted that the role of JNK might not be as important as that of p38MAPK or ERK1/2 in FCM-induced cartilage damage. However, further studies are required to provide more information on the signaling mechanisms underlying the interaction between IPFP and cartilage.
IPFP-derived FCM is considered as a black box due to its diverse components [23]. An early publication showed that the leptin in IPFP-derived FCM might be able to induce the expression of MMPs in chondrocytes [24]; however, this may be limited by a small sample size and the bias in case selection. More recently, researchers demonstrated that the effect of IPFP on chondrocyte was not mediated by leptin or adiponectin [17]. Therefore, we speculate that the deleterious phenotype of osteoarthritic IPFP mainly mediated by other cytokines instead of adipocytokines. Because IL-1β, TNF-α, and IL-6 are widely considered as the main proinflammatory cytokines during OA progression and could be highly secreted by IPFP [5,16], we performed neutralization test to determine whether these factors are the leading players contributing to the biochemical deleterious effects of IPFP on cartilage. As a result, neutralizing IL-1β or TNF-α could attenuate the pro-inflammatory and pro-catabolic effects of FCM on chondrocytes. Surprisingly, blocking IL-6 did not change these adverse effects, suggesting that both IL-1β and TNF-α rather than IL-6 might play decisive roles in IPFP-induced damage. Therefore, we deduce that IL-6 may play a subordinate role in IPFP-derived FCM and other protective factors such as IL-10, adiponectin, and oxylipins secreted by IPFP may also counteract its effects. In addition, the signaling pathways especially p38MAPK and ERK1/2 responsible for the adverse effects of FCM showed no alteration after neutralizing IL-6 in FCM, which further suggested that IL-6 might not be a key player in the pro-inflammatory and pro-catabolic phenotype of IPFP. In our study, we did not detect all the factors that may mediate the damage effect of IPFP; however, what we need most is to identify the leading factors in IPFP, which is of the utmost importance to guide the prevention and treatment of knee OA.
In conclusion, osteoarthritic IPFP enhances the inflammation and degradation of cartilage through activating p38MAPK and ERK1/2. These findings suggest that modulating the effects of IPFP on cartilage may be a promising strategy for knee OA intervention.