PPARα−ACOT12 axis is responsible for maintaining cartilage homeostasis through modulating de novo lipogenesis

Here, in Ppara−/− mice, we found that an increased DNL stimulated the cartilage degradation and identified ACOT12 as a key regulatory factor. Suppressed level of ACOT12 was observed in cartilages of OA patient and OA-induced animal. To determine the role and association of ACOT12 in the OA pathogenesis, we generated Acot12 knockout (KO) (Acot12−/−) mice using RNA-guided endonuclease. Acot12−/− mice displayed the severe cartilage degradation with the stimulation of matrix MMPs and chondrocyte apoptosis through the accumulation of acetyl CoA. Delivery of acetyl CoA-conjugated chitosan complex into cartilage stimulated DNL and cartilage degradation. Moreover, restoration of ACOT12 into human OA chondrocytes and OA-induced mouse cartilage effectively rescued the pathophysiological features of OA by regulating DNL. Taken together, our study suggested ACOT12 as a novel regulatory factor in maintaining cartilage homeostasis and targeting ACOT12 could contribute to developing a new therapeutic strategy for OA.

C hondrocytes embedded within the extracellular matrix maintain cartilage homeostasis by stimulating anabolic and catabolic pathways to regulate production, turnover, and degradation of cartilage matrix proteins, such as type II collagen and proteoglycan 1 . During the pathogenesis of osteoarthritis (OA), chondrocytes produce extracellular matrixdegrading proteins such as matrix metalloproteinases (MMPs) as well as pro-inflammatory cytokines such as interleukin (IL)-1 and tumor necrosis factor (TNF) leading to the loss of chondrocyte cellularity and cartilage degeneration [2][3][4][5][6] . OA is a complex disease complicated with several metabolic diseases including dyslipidemia, hyperglycemia, and hypertension [7][8][9] . Epidemiological studies suggested that obese individuals have significantly increased the incidence of OA [10][11][12] . Increased levels of total free fatty acid (FFA) have been reported in OA patients [13][14][15] and prior to the appearance of histopathological OA features, accumulation of substantial LD in articular cartilage has been reported 9,16,17 . The elevated level of FFA in chondrocytes may contribute to the accumulation of LDs in articular cartilage 18 . Increased level of saturated fatty acids (SFAs) is closely related to the progression of cartilage degradation 19,20 . High susceptibility to SFA-rich diet is associated with increased OA-like pathophysiological changes 20,21 and SFA deposits in the cartilage alter the cartilage metabolism to be more prone to damage 20 . Recently, our laboratory also reported that regardless of body mass index (BMI), abnormal lipid accumulation in OA chondrocytes may responsible for the development and progression of OA 22 . Moreover, regardless of obesity, an increased level of serum cholesterol has been reported to be closely associated with the OA progression 14,23 . Although there is an increasing evidence that excessive lipid deposition (LD) could be associated with OA pathogenesis, the functional role and regulatory mechanism of LD during OA pathogenesis has not been well established. Therefore, understanding how LDs are regulated and accumulated in chondrocytes and how LD affects the pathogenesis of OA could have a great impact on developing a novel approach or techniques for treating OA.
De novo lipogenesis (DNL) and fatty acid (FA) oxidation, which regulates the breakdown and synthesis of fatty acids, are important determinants of lipid accumulation 24,25 . Lipogenesis is the process that synthesizes the FA chain by combining the acetyl group which adds carbons to a growing FA chain 26 . DNL is a highly regulated pathway and begins with acetyl-CoA as a principal building block for de novo synthesis of fatty acids 26,27 . Acetyl-CoA is a key metabolic intermediate in DNL. Acetyl-CoA, a key indicator of the metabolic state is regulated by adenosine triphosphate (ATP)-citrate lyase (ACLY), acetyl-CoA synthetase 1 (ACSS1), acetyl-CoA synthetase 2 (ACSS2), and acyl-CoA thioesterases (ACOTs). ACLY, ACSS1, and ACSS2 are involved in the production of acetyl-CoA and ACOTs hydrolyze fatty acyl-CoA into free FA and CoA, contributing to pyruvate metabolism and peroxisomal and mitochondrial fatty acyl-CoA oxidation [28][29][30][31][32] . Among them, ACOT12, also known as StarD15 or cytosolic acetyl-CoA hydrolase, is the major cytoplasmic enzyme that preferentially hydrolyzes the thioester bond of acetyl-CoA and generates acetate and CoA 27,33 . Since ACOT12 may responsible for determining the rate of degradation of cytosolic acetyl-CoA by controlling their dispositions towards oxidation versus complex lipid synthesis, the elucidation of the biological role and function of ACOT12 could provide a molecular basis and mechanism for deregulation of lipid homeostasis involved in the OA pathogenesis and could contribute in developing a new therapeutic approach for OA pathogenesis. In this study, we found that lipid accumulation increased by PPARα deficiency stimulates cartilage degradation through the modulation of ACOT12. In addition, the significant suppression of ACOT12 in the cartilage of human OA patient and OA animal models, indicating a requisite role of ACOT12 in maintaining articular cartilage homeostasis. We have also demonstrated that ACOT12 deficiency is involved in the LD accumulation of the stimulation of de novo lipogenesis (DNL) through the accumulation of acetyl-CoA, leading to the stimulation of cartilagedegrading enzymes and apoptosis of chondrocyte.

Results
PPARα is responsible for the accumulation of obeseindependent LD in the OA articular chondrocytes. The cartilage of OA patients undergo total knee replacement (TKR) surgery was divided as relatively healthy (non-OA) or severely damaged (OA) region and cartilage degeneration in OA cartilage assessed by Safranin O or Alcian blue staining (Fig. 1a) and Osteoarthritis Research Society International (OARSI) grading 34,35 (Fig. 1b). We also confirmed the increased expression of matrix metalloproteinase (MMP) 13 in OA cartilage. The LD accumulation in chondrocytes was assessed as staining with BODIPY, a fluorescent probe for neutral lipid or transmission electron microscopy (TEM) analysis ( Fig. 1c, d). In this study, to avoid the possible influence or effects of metabolic diseases, sex, and aging in cartilage environment, we studied cartilage tissues in normal BMI (<25) and 50-60-year-old male without hypertension and diabetes mellitus. Consistent with this observation, exposure of interleukin (IL)-1β into human normal articular chondrocyte (HN-AC) cell line and immature murine articular chondrocytes (iMACs) also dramatically induced LD accumulation (Fig. 1e). We also observed a significant increase in the number of BODIPY-positive cells in the cartilage of OA mice induced by destabilization of medial meniscus (DMM) surgery 36 compared to control cartilage at 8 weeks post surgery (Supplementary Fig. 1). The profile of genes in lipid metabolism using human chondrocytes and iMACs treated with IL-1β (Fig. 1f) as well as GSE dataset (GSE16464; human OA chondrocytes, GSE64394; human OA chondrocytes, GSE104793; IL-1β-treated primary mouse articular chondrocytes, GSE104794; HIF-2α-overexpressed primary mouse articular chondrocytes, GSE104795; ZIP8overexpressed primary mouse articular chondrocytes) (Fig. 1g) indicated the involvement of peroxisome proliferator-activated receptor (PPAR) signaling in OA pathogenesis. The expression level of PPARα was significantly decreased in chondrocytes and cartilages of OA patients (Fig. 1h, i) and DMM-induced mouse cartilages at post surgery of 4-and 6 weeks ( Supplementary Fig. 2).
To confirm the involvement of PPARα in the OA pathogenesis, iMACs were treated with fenofibrate, a potent agonists for PPARα, or GW6471, the specific PPARα inhibitor, or introduced by small interference RNA specific to PPARα (siPpara) in the presence or absence of IL-1β. Exposure of fenofibrate into iMACs significantly reduced the expression level of matrix metalloproteinase (Mmp)-9, Mmp13, a disintegrin and metalloproteinase with thrombospondin motifs (Adamts)-4, and -5 and increased cartilage matrix synthesis with a significant reduction of LD accumulation ( Supplementary Fig. 3a-c). In addition, LD accumulation in OA chondrocytes was also significantly reduced by the exposure of fenofibrate ( Supplementary Fig. 3d). On the other hand, exposure of GW6471 or introduction of siPpara into IL-1β-treated iMACs was stimulated the expression of Mmp9, Mmp13, Adatms4, or Adamts5. LD accumulation was also significantly increased by treatment of GW6471 or the introduction of siPpara in IL-1β-treated iMACs. Global deletion of PPARα (Ppara −/− ) did not change the skeletal development as assessed by alcian blue/alizarin red staining on postnatal day 0 (P0) and the thickness of tibial plateau compared to the cartilage of Ppara +/+ mice ( Supplementary Fig. 4a, b). The presence of sulfated proteoglycans in the culture of Ppara −/− iMACs was not different from the culture of Ppara +/+ iMACs ( Supplementary  Fig. 4c). However, a significant increase in LD accumulation and cartilage degradation and (Fig. 2a) and increased levels of MMP13 and ADAMTS4 (Fig. 2b) were observed in DMMinduced Ppara −/− cartilage. Restoration of Ppara using lentiviruses containing Ppara (Lenti-Ppara) into joint cavity of Ppara −/− mouse reduced the cartilage degradation, decreased number of BODIPY-positive cells, and decreased expression level of MMP13 and ADAMTS4 in Ppara −/− cartilage (Fig. 2c, d).
Decreased level of ACOT12 is responsible for the cartilage degradation by Ppara deficiency. To elucidate the responsible factors in cartilage degradation induced by PPARα deficiency, we analyzed the alteration of expression levels in downstream target genes of PPARα using Ppara −/− iMACs and IL-1β-treated Ppara +/+ iMACs and identified ACOT12 as a key regulator factor in this (Fig. 3a). The expression level of Acot12 was significantly increased by the introduction of siPpara into iMACs of Ppara +/+ mice as well as in iMACs of Ppara −/− mice (Fig. 3b, c). Moreover, decreased ACOT12 level and decreased number of ACOT12-positive cells were observed in OA chondrocytes and OA cartilage (Fig. 3d, e). The protein level of ACOT12 in articular chondrocyte was confirmed by western blotting using chondrocyte progenitor cell (CPC) and iMACs ( Supplementary  Fig. 5a). Consistent with our data, analysis of GSE8077 of rat articular OA chondrocytes also showed the decreased level of ACOT12 in OA chondrocytes (Fig. 3f). Exposure of IL-1β into human normal chondrocytes (HN-AC) or iMACs significantly reduced the expression level of ACOT12 (Fig. 3g). Decreased ACOT12-positive cells were observed in the cartilage of DMMinduced mice ( Fig. 3h and Supplementary Fig 5b). Introduction of short hairpin RNA against ACOT12 (shAcot12) into HN-AC accumulated free fatty acid and LD with significant increases of MMP13, ADAMTS4, and ADAMTS 5 level ( Fig. 3j-l). These data suggest that ACOT12 may act as a key regulator in the cartilage degradation induced by PPARα deficiency.
Accumulation of acetyl-CoA by ACOT12 deficiency stimulates DNL during OA pathogenesis. Because articular cartilage does not have blood vessels, nerves, or lymphatics unlike most tissue, chondrocytes mainly depend on anaerobic metabolism using glucose as an important fuel to maintain cartilage homeostasis. Several reports suggested that the glycolytic metabolic pathway converting glucose into lactate when limited amounts of oxygen are available, or pyruvate which then enters the Krebs cycle is upregulated during OA pathogenesis 7,37,38 . In silico analysis of GSE57218, GSE104794 and GSE104795 indicated a significant increase in glycolysis during OA pathogenesis ( Supplementary  Fig. 10a). Enrichment plot of KEGG and GDEA analysis from RNA-seq data of OA patient (OA vs non-OA) also suggested glycolysis as one of the enriched signaling pathways in OA pathogenesis ( Supplementary Fig. 10b). Consistent with these data, Acot12 −/− iMACs showed activated glycolysis pathway and increased level of glycolysis genes such as hexokinase 2 (HK2), phosphofructokinase (PFKP), phosphoglycerate mutase 2 (PGAM2) and phosphoenolpyruvate carboxykinase (PCK) (Supplementary Fig. 10c, d). Inhibition of glycolysis by 2-deoxy-D-glucose (2DG) suppressed LD accumulation in the presence of IL-1β without significant differences in chondrocyte apoptosis ( Supplementary Fig. 11). These data suggested that LD accumulated in OA chondrocytes may In addition, analysis from RNA-seq data of OA patient (OA vs non-OA) showed that expression levels of genes participating in lipogenic pathways such as FA synthesis (ACLY, ACSS2, and ACCL), FA elongation (SCD1, ELOVL5, and ELOVL6), and FA desaturation (FADS2 and FADS3) were significantly increased in OA patient chondrocytes ( Supplementary Fig. 12) suggesting that LD in OA chondrocytes is possibly due to activation of de novo lipogenic pathway (DNL). Since DNL initiated with the conversion of acetyl-CoA into malonyl-CoA, these data suggested that increased acetyl-CoA level by ACOT12 deficiency could activate DNL. Among three major enzymes, ACLY, ACSS2, or ACOT12, involved in the regulation of acetyl-CoA pool 33,39,40 ( Supplementary Fig. 13a), expression levels of ACLY and ACSS2 which are involved in the production of cytosolic acetyl-CoA were significantly increased whereas expression level of ACOT12 which is involved in the hydrolysis of cytosolic acetyl CoA was significantly decreased by IL-1β-treated HC-N cells (Supplementary Fig. 13b) and DMM-induced cartilage ( Supplementary   Fig. 13c) indicating the accumulation of acetyl-CoA during OA pathogenesis. To investigate the effect of acetyl-CoA accumulation on cartilage homeostasis, iMACs were treated with acetate in a dose-dependent manner. With increasing the concentration of acetate, increased cellular level of acetyl-CoA, FFA, and TG (Fig. 6b), increased expression level of genes in DNL such as Acaca, Fasn, and Scd1 (Fig. 6c) and in cartilage-degrading enzyme such as Mmp13, Adamts4, and −5 (Fig. 6d) were observed. Furthermore, acetyl-CoA-conjugated chitosan complex (Supplementary Fig. 14a, b) was delivered into articular chondrocyte twice per week by intra-articular injection after DMM surgery 41,42 and delivery efficiency was confirmed by fluorescein isothiocyanate (FITC)-labeled chitosan (Supplementary Fig. 14c). In the cartilage injected with acetyl CoA-conjugated chitosan complex (Chitosan-AcCoA), cartilage degradation as assessed by safranin O staining and the number of BODIPY-and TUNELpositive cells were significantly increased compared to cartilages injected chitosan alone (Chitosan-Con) and sham-operated cartilage (Supplementary Fig. 15). Increased cartilage degradation and increased number of TUNEL-and MMP13-positive cells (P = 0.0008) in Acot12 +/+ and Acot12 −/− iMACs treated with 5 ng/ml IL-1β (n = 3). d Safranin O and BODIPY staining in cartilage of DMM-induced Acot12 +/+ and Acot12 −/− mice at 8 weeks. The degree of cartilage degradation is quantified according to OARSI grade (n = 7), P < 0.0001. Scale bar, 100 μm. BODIPY-positive cells are indicated by bar-dot plot (n = 6), P < 0.0001. e Immunohistochemistry of MMP13 and ADAMTS4 and TUNEL staining in the cartilage section of Acot12 +/+ and Acot12 −/− mice (n = 7). MMP13 (P < 0.0001), ADAMTS4 (P = 0.0001), and TUNEL-positive cells (P < 0.0001) were indicated by bar-dot plot. Unpaired Student's t test (a, b, d, e) or multiple t test followed by Holm-Sidak method (c) were used for statistical analysis. **P < 0.01, ***P < 0.001, ****P < 0.0001. Fig. 15a, b) as well as the increased number of BODIPY-, FASN-, and SCD1-positive cells were observed with the delivery of Chitosan-AcCoA into Acot12 +/+ mice twice per week by intra-articular injection (Fig. 6e, f).

(Supplementary
ACOT12 overexpression prevents OA pathogenesis induced by PPARα deficiency. To confirm the involvement of ACOT12 in the OA pathogenesis induced by PPARα deficiency, lentivirus containing HA-tagged ACOT12 (HA-Acot12) were transfected into Ppara −/− iMACs or injected into the cartilage of Ppara −/− mouse. In Ppara −/− iMACs, cartilage matrix was increased and LD accumulation was decreased by introduction of ACOT12 same as shown in the restoration of PPARα (Fig. 7a). Increased level of FFA and acetyl-CoA by IL-1β was also significantly decreased with introduction of ACOT12 as well as the restoration of PPARα (Fig. 7b). Moreover, decreased levels of MMP13, ADAMTS-4, and −5 were observed with introduction of ACOT12 into Ppara −/− iMACs ( Supplementary Fig. 16a). Consistent with this, injection of HA-Acot12 into the cartilage of Ppara −/− mice significantly reduced the degradation of cartilage ( Fig. 7c and Supplementary Fig. 16b). Increased lipid accumulation and the number of TUNEL-, MMP13-, ADAMTS4-positive cells in Ppara −/− catilage were significantly decreased by introduction of HA-Acot12 (Fig. 7d, e). These data suggest that ACOT12 is a major regulatory factor in the OA pathogenesis induced by PPARα deficiency and introduction of exogenous ACOT12 could effectively alleviate cartilage degradation induced by PPARα deficiency.

Discussion
Fine regulation of metabolism is important for maintaining cartilage homeostasis and impaired metabolism could trigger various pathophysiological situations through stimulation of proinflammatory cytokines and biomechanical stress 7,43 . Thus, metabolic reprogramming in chondrocyte from pathological metabolic state to normal metabolic state by certain enzymes could be one of the possible alternative approaches to OA therapeutic intervention. Several metabolic pathways including glycolysis and fatty acid oxidation are associated with OA pathogenesis 7 . Peroxisome proliferator-activated receptor (PPAR) s, members of the nuclear hormone receptor superfamily in mediating the physiological actions of FA and FA derivatives are known to be involved in the OA pathogenesis 44 . Among three PPAR isoforms, PPARα, PPARβ/δ, and PPARγ, PPARγ is the most studied member in OA pathogenesis. PPARγ1, predominantly expressed in human cartilage, is known to have a protective role against OA 45 . Agonists of PPARγ downregulate inflammatory response and prevent cartilage degradation in OA Recently several studies have reported that PPARα also plays an important role in cartilage homeostasis 46,47 . PPARα activation by its agonist, WY14643, decreases inflammatory and cartilage destructive responses in OA cartilage 47 suggesting its protective effect on articular cartilage against OA pathogenesis. Here, in this study, we found the protective role of PPARα against cartilage degradation via regulating acetyl-CoA pool through ACOT12 modulation. we also observed that fenofibrate, the agonist for PPARα, reduced the typical characteristics of OA such as increased levels of cartilage-degrading enzymes such as MMP-9, −13, ADMATS-4, and −5 and LD accumulation. Furthermore, we also observed the increase in cartilage degradation and LD accumulation in DMM-induced Ppara −/− cartilage.
important regulator for acetyl-CoA accumulation in OA chondrocytes 58 . In this study, we identified ACOT12 as one of the crucial factors in maintaining cartilage homeostasis by regulating a novel metabolic circuit via acetyl-CoA. We observed a suppressed level of ACOT12 in human OA cartilages and of OAinduced animal cartilages. Germline deletion of Acot12 spontaneously developed articular cartilage lesions without affecting bone development. Cartilage degradation was exacerbated due to a decreased chondrocyte cellularity by activating Mmp-13 and increasing lipid depositions (LD) in chondrocytes, particularly through acetyl-CoA accumulation in DMM-induced Acot12 −/− mice.
Then, where do lipid droplets in Acot12 −/− cartilage come from Intracellular lipids can accumulate from an increased lipid influx in blood or stimulation of autophagy or can be synthesized de novo 59 . In this study, we found that de novo lipogenesis (DNL) might be the one key regulatory process for LD accumulation induced by Acot12 deficiency. In Acot12 −/− cartilage, an increased level of genes in DNL contributing lipid accumulation in chondrocytes was observed and this could lead lipotoxicity and metabolic stress to chondrocytes. Acetyl-CoA is a crucial component in lipid metabolism, as it is a requisite carbon donor in the de novo synthesis of fatty acids and used by acetyl-CoA carboxylase (ACC) to catalyze the synthesis of malonyl-CoA. In Acot12 −/− cartilage, increased level of acetyl-CoA was observed and increased cellular level of acetyl-CoA by treating acetate or delivery of Chitosan-AcCoA into cartilage increased DNL in cartilage and stimulated cartilage degradation by unbalancing between catabolic and anabolic reaction. The recent study also suggests the importance of acetyl-CoA in the OA pathogenesis. Reduced level of acetyl-CoA in cartilage by ACLY deficiency showed a chondroprotective action and resulted less cartilage damage 58 .
Taken together, our study suggested ACOT12 is the main key regulatory factor in PPARα-mediated cartilage homeostasis through regulating acetyl-CoA pool. Increased level of acetyl-CoA by ACOT12 deficiency stimulates DNL pathway and resulted in the stimulation of cartilage degradation. Therefore, targeting

Methods
Human OA specimens and primary chondrocyte culture. Human articular cartilage specimens were obtained from patients undergoing total knee replacement (TKR) surgery. Specimens from osteoarthritic cartilages were classified as relatively heathy (non-OA) or severe damaged (OA) region. Human articular cartilage specimens were cut into pieces, and some pieces were immediately fixed by 10% neutral buffered formalin and others used for primary chondrocyte culture. For primary human chondrocyte culture, pieces of each cartilage (OA or non-OA) were digested using 0.06% collagenase (Sigma) and cultured in high glucose DMEM (Gibco) supplemented with 10% fetal bovine serum (FBS), 1× antibiotic-antimycotic (Gibco). The cells were cultured at 37°C in a humidified atmosphere of 5% CO 2 .
Animals. Wild-type C57BL/6N mice were purchased from Samtako BioKorea Inc. (Osan, Korea). Eight-week-old male C57BL/6N mice used in DMM surgery and 5days-old pups used in iMACs culture. All mice were housed at 23 ± 1°C with 12-h light/dark cycles and a relative humidity of 50 ± 5% with food and water available ad libitum.
Experimental OA and histology of OA cartilage. Experimental OA was induced by destabilization of the medial meniscus (DMM) surgery using 8-weeks-old male mice. Knee joints were analyzed 8 weeks after DMM surgery. Cartilages were processed for histological and biochemical analyses. Briefly, knee joints were fixed in 4% paraformaldehyde, decalcified in 14% EDTA (pH 7.4), and embedded in paraffin. The paraffin blocks were sectioned at 5-μm thickness and sectioned joints were performed safranin O staining. The degree of cartilage degradation was scored 0-6 grade using OARSI scoring system and images were acquired by EVOS FL Auto software v.1.7.
Generation of Acot12 −/− mice. Global Acot12 knockout mice were generated by germline transmission of an RGEN-induced mutant allele. Acot12 specific guide RNA and Cas9 protein (each 50 ng/μl) were injected into the cytoplasm of C57BL/ 6N mouse eggs and transferred into the pseudo-pregnant foster female mice. The absence of Acot12 in the mutants was confirmed by routine tail DNA genotyping and western blotting. The absence of Acot12 was confirmed by standard PCR genotyping. Genotyping primers were designed to distinguish Acot12 +/+ or Acot12 +/− or Acot12 −/− alleles: forward: 5'-agccaggacgatggagtcga-3'; reverse: 5'ggtgtccatccacttgagca-3'.
Transmission electron microscopy. Chondrocytes fixed in 2.5% glutaraldehyde solution were post-fixed with 0.1 M OsO 4 and embedded with Epoxy Embedding Medium Kit (Sigma-Aldrich). The epoxy embedded block was ultramicrosectioned using EM UC7 (Leica) and stained with uranyl acetate and lead citrate. The microstructure image was analyzed with 120 kV transmission electron microscope (HITACHI).
Alcian blue staining. Cells were rinsed with PBS and fixed in 4% paraformaldehyde for 20 min. Fixed cells were stained in 1% Alcian blue in 0.1 N HCl for 24 h at room temperature, rinsed with 0.1 N HCl, and captured bright-field image by EVOS FL Auto software v.1.7. Stained Alcian blue was extracted by 6 M guanidine HCl for 2 h and measured absorbance at 600 nm.
Whole-mount skeletal staining. Postnatal 0 pups were euthanized and removed skin, visceral organ, eye, adipose tissue, and other excess tissue. Specimens were fixed by 95% EtOH for 24 h. Specimens were submerged in alcian blue solution overnight at room temperature and destained in 95% EtOH. To clear the tissue, solution was replaced by 2% KOH solution for 12 h and replaced with Alizarin red solution for 24 h at RT. Stained specimens were replaced by 20% glycerol/1% KOH, 50% glycerol/1% KOH, and 80% glycerol/1% KOH for 24 h each and transferred to 100% glycerol. Biochemical analysis. The cellular level of FFA and acetyl-CoA were measured using Free Fatty Acid Assay Kit (Dogen; DG-FFA100) and acetyl-CoA Assay Kit (Biovision; K317-100), respectively according to the manufacturer's instructions.
TUNEL assay. TUNEL assay was performed using In Situ Cell Death Detection Kit (Roche) on the deparaffinized cartilage section according to the manufacturer's instructions. Nuclear was stained with 4',6-diamidino-2-phenylindole, Dihydrochloride (DAPI). Fluorescence images were acquired using fluorescent microscopy.
Neutral lipid staining. Fixed cells or cryosection of human and mouse cartilages were stained using BODIPY 493/503 (Thermo Fisher) for 20 min and mounted with DAPI mounting medium (Vector Laboratories). Fluorescence images were acquired with cartilage images were acquired by EVOS FL Auto software v.1.7.
Lipid peroxidation staining. Live cells were stained using BODIPY 581/591 (Invitrogen) probe followed by manufacturer's recommendation, and fluorescence intensity was quantified by image J 1.51j8.
Cell apoptosis assay. Cells were harvested and incubated with annexin V and propidium iodide (PI) solution from Muse Annexin V & Dead Cell Kit (Luminex) according to the manufacturer's instructions. The fluorescence of annexin V and PI were analyzed by Muse Cell Analyzer (Merk Millipore).
Western blot. ACOT12 KO efficiency was confirmed by western blot using mouse liver. Mouse liver protein were extracted using RIPA buffer (Cell Signaling) with 2 mM PMSF. Thirty micrograms of protein was separated by 10% polyacrylamide gel electrophoresis containing 0.1% SDS and transferred to nitrocellulose membranes (GE Healthcare). The membranes were incubated for 1 h at RT in TBS-T with 5% skim milk and probed with the following primary antibodies: ACOT12 (1:1000 dilution, Mybiosource; #MBS273137), GAPDH (1:10,000 dilution, Bioworld; #AP0066). The blots were developed with a HRP-conjugated anti-rabbit secondary antibody (1:2000 dilution, Bethyl Laboratories) and visualized with SuperSignal West Pico PLUS Chemiluminescent Substrate (Thermo Fisher).
Quantitative real-time (qRT)-PCR. Total RNA was isolated from liver tissue using RNAiso Plus (TaKaRa) according to the manufacturer's instructions. Then, 1 μg RNA was reverse transcribed using the 5× All-In-One RT Master Mix (ABM). Real-time PCR was performed with ABI StepOnePlus instrument (Applied Biosystems) using AMPIGENE qPCR Green Mix Hi-ROX (Enzo). The cycling protocol was 40 cycles of 95°C for 10 s, 57°C for 15 s, and 72°C for 10 s. All qRT-PCR reactions were performed in triplicate. The relative expression level of each gene was normalized to 18 S rRNA expression level. Forward and reverse primer sequences were listed in Supplementary Tables S1 and S2 (used in Fig. 1f), S3 (used in Fig. 3a).
Synthesis of fluorescein isothiocyanate (FITC)-labeled chitosan (chitosan-FITC). Chitosan (50 mg, 267.4 μmol of NH2) was dissolved in 1% (v/v) acetic acid solution (2.5 mL). FITC (10.4 mg, 26.7 μmol) in 0.5 ml methanol was slowly added onto the chitosan and reacted for 4 h in the dark. The obtained product was purified by dialysis (MWCO: 12-14 kDa, SpectraPor, USA) against an HCl solution (pH 5) for 2 days, followed by distilled and deionized water (DDW) for 2 days under darkness. The final product was lyophilized and held in the dark.
Preparation of chitosan-FITC/acetyl-CoA complexes. A chitosan-FITC stock solution was prepared by dissolving chitosan-FITC (1 mg) in 0.05 N HCl solutions (1 ml). A reaction solution was prepared by making a 1:10 dilution of stock solution with pH 7.4 PBS. Ten microliters of acetyl-CoA (50 mM) were added into the reaction solution and reacted for 6 h to ensure the formation of chitosan-FITC/ acetyl-CoA complexes. The final product was purified by dialysis (MWCO: 12-14 kDa, SpectraPor, USA) against the DDW for 1 day.
Ethical approval. All animal studies were approved by the Wonkwang University Animal Care and Use Committee (#WKU18-23, WKU19-09, WKU20-61) and were in compliance with the institutional guidelines. Human cartilage tissue collection was approved by the Human Subjects Committee of Wonkwang University Hospital (WKUH 201605-HRBR-041) and studies were performed in compliance with the institutional guidelines. Written informed consent was obtained from all adult patients or at least one guardian of each patient prior to the start of the experiment.
Statistical analyses. Results are expressed as the mean ± SD. The mean values for RNA level and biochemical data of two groups were compared by unpaired twotailed Student's t test. For more than two groups, one-or two-way ANOVA for multiple comparisons were used. Statistical tests with P < 0.05 was considered significant and significance was defined as *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. All statistical tests were performed with the software GraphPad Prism 6 (GraphPad).
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.