NFIA regulates articular chondrocyte fatty acid metabolism and joint homeostasis

Abstract

Osteoarthritis (OA) is a joint disease with an etiology partially rooted in metabolic dysfunction, yet the underlying mechanisms in this context are not determined, limiting opportunities to develop therapeutic treatments. In this study, we used a multiomic approach combining RNA sequencing, ATAC-seq, MRE-seq, and metabolomics to reveal that OA articular chondrocytes induced by imbalanced transforming growth factor–β (TGF-β) and bone morphogenetic protein (BMP) signaling have increased fatty acid synthesis and oxidation processes regulated by nuclear factor I A (NFIA) up-regulation. Inhibition of NFIA suppressed the elevated gene expression of essential metabolic enzymes, including acetyl-CoA carboxylase A (ACACA) and carnitine palmitoyltransferase 2 (CPT2), leading to the restoration of fatty acid metabolism and cellular homeostasis in both murine and human OA articular chondrocytes. Obese mice displayed metabolic stress with elevated expression of NFIA, ACACA, and CPT2 in joint tissues, and they simultaneously developed profound synovitis, cartilage degeneration, subchondral bone sclerosis, and pain after joint injury. Both Nfia inhibition and pharmacological suppression of fatty acid metabolism in obese mice preserved joint integrity and mitigated synovitis and pain in the context of injury-induced OA settings. Overall, this work identifies a role for NFIA in the regulation of fatty acid metabolism and articular chondrocyte homeostasis and highlights fatty acid metabolism as a potential therapeutic target for OA treatment, particularly under obesity conditions.

INTRODUCTION

The rapid growth of the aging and obese population over the past decades has led to substantial global health challenges, particularly in the form of chronic degenerative disorders. Despite advances in therapeutics for many of these diseases, osteoarthritis (OA) remains one of the few major conditions without a disease-modifying treatment (1–3). Current treatments are limited to managing joint stiffness and pain to enhance patient mobility. Ultimately, patients with advanced OA undergo joint arthroplasty surgery. Although effective, joint arthroplasty is costly and may lead to complications, including infection, prosthetic loosening, and continued pain (4). There is a critical need to explore the mechanisms underlying OA and identify new therapeutic targets for the development of disease-modifying treatments.

OA is a heterogeneous disease with multiple etiological factors, including aging, obesity, and joint trauma, which complicate the development of effective treatments. OA has been recently recognized for its shared metabolic abnormalities in joint tissues (5–9), shedding light on potential new therapeutic approaches. Clinical investigations have revealed a distinct fatty acid signature in the synovial fluid of patients with OA compared with healthy individuals, characterized by increased concentrations of free fatty acids and phospholipids (10–12). Several lipid metabolites are further elevated as the disease progresses from early to late stages in patients with OA (10, 13–15). Furthermore, under conditions of systemic metabolic challenges, obese individuals are more than four times more likely to develop OA with worse pain and other symptoms, compared with individuals with normal body weight (16, 17). In animal models, obesity-induced systemic metabolic challenges can substantially affect the lipid metabolites in joints, thereby accelerating injury- and aging-induced OA progression (18). This accumulating evidence strongly implicates the potential effect of lipid metabolism on joint homeostasis and OA pathogenesis.

Lipid metabolism, particularly the synthesis and oxidation of fatty acid, has been recognized as a critical regulator of cellular homeostasis, which is frequently altered in disease conditions, such as diabetes and cancer (19, 20). Thus far, little is known about fatty acid metabolism in musculoskeletal diseases, particularly OA. Although OA affects the entire joint, articular chondrocytes are considered the primary responders to environmental stimuli at the cellular and molecular level. Imbalances in chondrocyte anabolic and catabolic activities lead to cartilage tissue degeneration and OA progression (21). Emerging research has begun to explore the connection between fatty acid metabolism and the homeostasis of chondrocyte and cartilage. Fatty acids and lipid metabolites, although present in small quantities, circulate through the joint in synovial fluid and are stored in the cartilage matrix as well as chondrocytes within cartilage by various fatty acid transporters (14, 22–25). Although articular chondrocytes have access to these fatty acids and lipids, they typically exhibit undetectable cellular fatty acid oxidation because of their low energy demands under healthy conditions (26). In OA, however, hypervascularization in the synovium and subchondral bone regions supplies increased oxygen and nutrients to articular chondrocytes (27–32), supporting their increased energy needs for the production of matrix-degrading enzymes and the synthesis of inappropriate or mechanically inferior fibrillar collagens, such as type I, III, and X collagen. Recent integrative genome-wide gene expression analyses show up-regulated oxidative phosphorylation pathways in OA tissues, including synovium (33) and articular cartilage (34). Furthermore, single-cell RNA sequencing (RNA-seq) analyses have identified increased fatty acid metabolism within the articular chondrocytes of patients with OA (35). On the molecular level, hormones such as leptin and insulin-like growth factor 1 (IGF1), which are key regulators of fatty acid metabolism, are found to be elevated in patients with OA and are implicated in OA pathogenesis. High leptin concentration in the serum and synovial fluid are associated with obesity-associated OA progression by promoting catabolic responses in cartilage (36). IGF1 has been linked to OA through the down-regulation of glucose transporter 1 (GLUT1), a critical enzyme for glucose and fatty acid metabolism (37, 38). Supporting this clinical observation, our recent investigations have demonstrated that the ablation of Glut1 impairs chondrocyte proliferation and maturation during embryonic development (39) and disrupts the balance between cartilage anabolism and catabolism in postnatal mice (40), underscoring the important role of fatty acid metabolism in cartilage health throughout the life span. Despite these associations between lipid metabolism and OA, the causal links between chondrocyte fatty acid metabolism, cartilage homeostasis, and OA pathogenesis remain largely unknown.

Using unbiased comprehensive sequencing of the genome and epigenome, as well as high-throughput measurement of metabolites and murine OA models, we identified nuclear factor I A (NFIA) as a key regulator of fatty acid metabolism in articular chondrocytes and defined a role for fatty acid metabolism in chondrocyte homeostasis and OA progression. We found that NFIA inhibition maintains cellular and maturational homeostasis in human OA articular chondrocytes and attenuates cartilage degeneration and pain progression in obese mice after joint trauma. Furthermore, suppression of fatty acid metabolism also attenuated joint injury–induced OA progression in obese mice. Collectively, our data revealed a role for NFIA-mediated fatty acid metabolism in the regulation of chondrocyte homeostasis, highlighting NFIA as a potential therapeutic target for OA treatment.

RESULTS

OA is associated with imbalanced TGF-β and BMP signaling in articular chondrocytes

Balanced signaling of the bone morphogenetic protein (BMP) and transforming growth factor–β (TGF-β) pathways is critical for maintaining cartilage tissue homeostasis and integrity (41, 42). OA cartilage has a relative increase in BMP pathway member activin receptor–like kinase 1 (ALK1) compared with TGF-β pathway member activin receptor–like kinase 5 (ALK5) (43). To validate this finding, we examined the ALK1/ALK5 ratio in human healthy or OA articular chondrocytes by quantitative polymerase chain reaction (qPCR) analysis. OA chondrocytes had a fivefold increase in the ALK1/ALK5 ratio, indicating a shift toward BMP signaling (Fig. 1A). This shift in OA chondrocytes was associated with induction of hypertrophic and catabolic genes (RUNX2 and MMP13) and reduction of cartilage matrix genes (COL2A1 and ACAN) (Fig. 1B). In human cartilage tissues, we evaluated the activation of the BMP pathway by staining for pSMAD1 and the TGF-β pathway by pSMAD2 immunohistochemistry (IHC). In OA-derived tissues, there was an elevation of pSMAD1 and a decrease in pSMAD2 (Fig. 1, C and D).

Fig. 1. Altered TGF-β and BMP signaling in human OA pathogenesis.

(A and B) Gene expression analysis by qPCR of ALK1 and ALK5 (a) and RUNX2, MMP13, COL2A1, and ACAN (B) from human healthy chondrocytes (HC) and OA articular chondrocytes. The gene expression was normalized to ACTB and then the HC group. n = 3. (C) IHC for pSMAD1 (left) and pSMAD2 (right) on human healthy cartilage and OA cartilage. high-magnification views of the boxed regions are shown below. Scale bars, 100 μm. (D) Quantification of the pSMAD1-positive and pSMAD2-positive chondrocytes. data in (A), (B), and (D) are presented as mean ± SD, and dots represent individual samples. Analysis by Student’s t test. For all panels, n = 3. *P < 0.05.

Because human OA pathogenesis has multiple etiologies, to further understand individual risk factors, we examined the TGF-β and BMP balance in cartilage from murine OA models induced by various individual risk factors. In murine arthritic cartilage induced by joint trauma [meniscus-ligament injury (MLI)], high-fat diet (HFD)–induced obesity, and the spontaneous aging process (fig. S1, A to C), we observed the down-regulation of the TGF-β pathway, as evidenced by decreased pSMAD2 staining, and an increase in BMP signaling, evidenced by an increase in pSMAD1 staining, in articular chondrocytes (fig. S1, D to F). These data suggest a common signaling imbalance in TGF-β and BMP in articular chondrocytes across etiologies in models of OA disease.

To mimic the clinical scenario and provide a cell culture model to explore the mechanism behind OA disease, we maintained murine articular chondrocytes in a homeostatic state with TGF-β1 and induced articular chondrocytes to become hypertrophic and diseased by culture with BMP2. TGF-β1 suppressed the disease-like hypertrophic differentiation of murine articular chondrocytes by down-regulating Runx2 and Mmp13 under chondrocyte maturation conditions (fig. S2, A and B). Conversely, BMP2 induced disease-like differentiation and the potential for matrix catabolism, evidenced by up-regulation of Runx2 and Mmp13 in murine articular chondrocytes cultured in basal medium (fig. S2, C and D). These findings align with our understanding of how TGF-β and BMP signaling balance modulates chondrocyte hypertrophy and matrix production through established transcriptional regulators, such as RUNX2 (44). Other mechanistically linked pathways, including those related to metabolism, are uncharacterized; thus, we used unbiased multiomic approaches to study TGF-β and BMP signaling balance in articular chondrocyte biology.

TGF-β1 and BMP2 alter fatty acid metabolism in articular chondrocytes

RNA-seq analyses were performed on TGF-β1– and BMP2-treated primary murine articular chondrocytes. We found that TGF-β1 and BMP2 substantially altered the transcriptome of murine articular chondrocytes (Fig. 2A and fig. S3A), with 2934 genes significantly up-regulated by TGF-β1 treatment (P < 0.05), annotating to essential housekeeping biological processes, such as ribosome and mitochondrial function. Conversely, BMP2 treatment significantly induced the expression of 1152 genes (P < 0.05), annotating to biological processes related to lipid metabolism (Fig. 2A). Compared with TGF-β1 treatment, BMP2 stimulated both fatty acid synthesis and oxidation processes in murine articular chondrocytes (Fig. 2B and fig. S3B). In particular, acetyl-CoA carboxylase A (Acaca) (45) and carnitine palmitoyltransferase 2 (Cpt2) (46), the rate-limiting enzymes for fatty acid synthesis and oxidation (47), were significantly up-regulated by BMP2 treatment in murine articular chondrocytes (P < 0.05) (Fig. 2B). BMP2 increased all other key genes involved in fatty acid metabolism, including acyl-CoA synthetase (Acsl) family members and acyl-CoA dehydrogenase (Acad) members as well as acetyl-CoA acyltransferase (Acaa) members and fatty acid synthase (Fasn) (fig. S4), suggesting overall enhancement of fatty acid synthesis and oxidation (Fig. 2B). Moreover, our RNA-seq and ATAC-seq analyses indicate that vehicle-treated murine articular chondrocytes spontaneously undergo hypertrophic differentiation in culture, exhibiting transcriptomic profiles, genetic functions (fig. S5, A to H), and accessible chromatin regions (fig. S6, A to C) similar to those observed with BMP2 treatment. Despite this spontaneous hypertrophy, BMP2 still up-regulated genes associated with fatty acid metabolism, including Acaca and Cpt2, compared with vehicle treatment (fig. S6D). In contrast, TGF-β1 treatment suppressed the expression of fatty acid metabolism–related genes in murine articular chondrocytes in comparison with vehicle and BMP2 treatments.

Fig. 2. TGF-β1 and BMP2 altered fatty acid metabolism in murine articular chondrocytes.

(A) Differentially expressed genes and enriched Gene Ontology terms associated with TGF-β1 or BMP2 treatment in murine articular chondrocytes. TGF-β1, blue; BMP2, orange. Dots represent differentially expressed genes. (B) A schematic illustration of fatty acid oxidation and synthesis processes and box plot graphs showing the gene expression of key enzymes responding to TGF-β1 (blue) and BMP2 (orange) treatment in murine articular chondrocytes. ACP, acyl carrier protein. (C) A schematic illustration of fatty acid synthesis traced by U-¹³C glucose. The black filled circles denote ¹³C, and the open circles denote ¹²C. α-KG, α-ketoglutarate; OAA, oxaloacetate. (D) The contribution of fatty acid synthesis was traced using U-¹³C glucose. The percentage of ¹³C-labeled and -unlabeled palmitoylcarnitine in murine articular chondrocytes treated with TGF-β1 and BMP2. (E) A schematic illustration of fatty acid synthesis traced by U-¹³C glutamine. The black filled circles denote ¹³C, and the open circles denote ¹²C. (F) The contribution of fatty acid synthesis was traced using U-¹³C glutamine. The percentage of ¹³C-labeled and -unlabeled palmitoylcarnitine in murine articular chondrocytes treated with TGF-β1 and BMP2. (G) A schematic illustration of fatty acid oxidation traced by U-¹³C palmitic acid. The black filled circles denote ¹³C, and the open circles denote ¹²C. (H) Fraction of citrate M + 0 and M + 2 in murine articular chondrocytes treated with TGF-β1 and BMP2. Data in (D), (F), and (H) are presented as mean ± SD, and dots represent individual samples. Analysis by unpaired Student’s t test. For all panels, n = 3. *P < 0.05.

To validate the functional effect of TGF-β1 and BMP2 on fatty acid metabolism in articular chondrocytes, we evaluated fatty acid synthesis and oxidation using stable isotope labeling and high-resolution liquid chromatography–mass spectrometry (LC-MS) technologies. Given that glucose and glutamine are two major precursors for fatty acid synthesis, we examined fatty acid synthesis in murine articular chondrocytes by culturing cells in either U-¹³C glucose or U-¹³C glutamine. Glucose-derived pyruvate enters the mitochondria and is metabolized to acetyl-CoA (AcCOA). AcCOA and oxaloacetate are used to produce citrate, which is transported into the cytosol and cleaved to produce lipogenic AcCOA for fatty acid synthesis (Fig. 2C). Palmitate is the typical end product generated by fatty acid synthase, directly reflecting the fatty acid synthesis. We measured palmitoylcarnitine to infer the labeling pattern of palmitate, from which the contributions of nutrients to fatty acid synthesis could be determined (48). The total ¹³C enrichment of palmitoylcarnitine was not changed in murine articular chondrocytes between TGF-β1 and BMP2 treatments (Fig. 2D). The total ¹³C enrichment of palmitoylcarnitine was less than 1%, indicating that glucose is not a major precursor for fatty acid synthesis in the murine articular chondrocyte cell context. In addition to glucose, oxidation or reductive carboxylation of glutamine also contributes to citrate and is then cleaved to lipogenic AcCOA for fatty acid synthesis (Fig. 2E). The total ¹³C enrichment of palmitoylcarnitine and other intermediate metabolites, such as phosphatidylcholine (PC), acylcarnitine (CAR), lysophosphatidylcholine (LPC), and lysophosphatidylethanolamine (LPE), was significantly higher in murine articular chondrocytes treated with BMP2 compared with TGF-β1 (P < 0.05) (Fig. 2F and fig. S7). Using the isotope labeling data, we also performed isotopomer spectral analysis (ISA) to quantify the biosynthetic rate. The fraction of newly synthesized palmitate [reflected as g(t)] in cells treated with BMP2 increased by ~40-fold compared with TGF-β1 treatment (table S1). These findings established that glutamine is the major precursor for fatty acid synthesis in murine articular chondrocytes, with BMP2 treatment up-regulating fatty acid synthesis in this context.

We next examined alterations in fatty acid oxidation in murine articular chondrocytes. Cells were labeled with either U-¹³C palmitate (saturated fatty acid) or U-¹³C oleate (unsaturated fatty acid). The transfer of ¹³C from palmitate and oleate to citrate, directly reflecting fatty acid oxidation (Fig. 2G), displayed a consistent pattern across conditions (fig. S8, A and B). BMP2 treatment resulted in a significantly higher enrichment of M + 2 citrate (P < 0.05) (Fig. 2H) and other tricarboxylic acid (TCA) cycle metabolites, including aspartate, glutamate, and fumarate, compared with TGF-β1 treatment (P < 0.05) (fig. S8, A and B). In addition, energy-related molecules in the TCA cycle, such as deoxyguanosine triphosphate (dGTP), nicotinamide adenine dinucleotide (oxidized form) (NAD⁺), reduced form of NAD⁺ (NADH), and adenosine triphosphate (ATP), were increased after BMP2 treatment (fig. S8, A and B), indicating enhanced fatty acid oxidation in BMP2-treated murine articular chondrocytes. Along with this increase in fatty acid oxidation, BMP2-treated murine articular chondrocytes exhibited reduced glycolysis, as reflected by a decrease in the extracellular acidification rate and reduced lactate production (fig. S9, A and B). Conversely, these cells showed greater mitochondrial mass and membrane potential (fig. S9C), as well as elevated oxygen consumption (fig. S9D). Consequently, BMP2 treatment led to an approximately fourfold increase in cellular ATP production relative to TGF-β1 (fig. S9E). Collectively, these metabolic changes, aligned with genomic profiles, suggest that BMP2 stimulated fatty acid synthesis and oxidation as well as cellular energy production during the transition of murine articular chondrocytes toward a diseased hypertrophic phenotype. In contrast, TGF-β1 maintains metabolic homeostasis and a healthy cellular phenotype.

Human OA articular chondrocytes display increased fatty acid metabolism

Given imbalanced TGF-β and BMP as a common molecular phenotype in OA, we then investigated whether increased lipid metabolism is also a common feature in OA chondrocytes. We isolated human articular chondrocytes from healthy individuals and patients with OA with similar body mass indices (BMIs; 31.51 ± 2.12 versus 32.77 ± 1.28; table S2) and used the same stable isotope labeling and high-resolution LC-MS methods to examine lipid metabolism. We cultured the human chondrocytes in medium containing U-¹³C glucose and found that the total enrichment of palmitoylcarnitine was increased in OA compared with healthy control articular chondrocytes (Fig. 3A). The low total ¹³C enrichment of palmitoylcarnitine (<2%) from glucose was consistent with our observations from murine articular chondrocytes. We then administered U-¹³C glutamine to human healthy and OA articular chondrocytes, revealing that the total ¹³C enrichment of palmitoylcarnitine as well as PC, CAR, LPC, and LPE was significantly increased in human OA articular chondrocytes compared with healthy control cells (P < 0.05) (Fig. 3B and fig. S10). The fraction of newly synthesized palmitate from U-¹³C glutamine by ISA was also significantly increased in human OA articular chondrocytes compared with healthy articular chondrocytes (P < 0.05) (table S3), indicating that OA chondrocytes have up-regulated fatty acid synthesis from glutamine. To investigate fatty acid oxidation, we labeled human healthy and OA articular chondrocytes with albumin-conjugated U-¹³C palmitate and U-¹³C oleate, respectively. We observed significantly higher enrichment of M + 2 citrate, aspartate, glutamate, and fumarate (P < 0.05) as well as energy-related molecules, such as dGTP, NAD⁺, NADH, and ATP (P < 0.05) in human OA compared with healthy articular chondrocytes (Fig. 3C and fig. S11, A and B), demonstrating that fatty acid oxidation is increased in human OA articular chondrocytes. At the molecular level, genes related to fatty acid metabolism, such as ACACA and CPT2, were significantly up-regulated in human OA articular chondrocytes (P < 0.05) (Fig. 3D). Human OA articular chondrocytes had increased mitochondrial mass and membrane potential (P < 0.05) (fig. S11C), as well as elevated oxygen consumption (P < 0.05) (fig. S11D). Cellular ATP was also higher in OA articular chondrocytes relative to healthy articular chondrocytes (P < 0.05) (fig. S11E).

Fig. 3. Proteomics revealed increased fatty acid metabolism in human OA articular chondrocytes.

(A and B) The percentage of ¹³C-labeled and ¹³C-unlabeled palmitoylcarnitine in human healthy chondrocytes and OA articular chondrocytes cultured with U-¹³C glucose (A) or U-¹³C glutamine (B). (C) Fraction of citrate M + 0 and M + 2 in human healthy chondrocytes and OA articular chondrocytes. (D) Gene expression of ACACA and CPT2 by qPCR from healthy and OA articular chondrocytes. (E) Heatmaps of differentially expressed genes in TGF-β1– and BMP2-treated murine articular chondrocytes as well as human control and OA articular chondrocytes. Color scale indicates the z-score transformed normalized gene expression values. Dendrograms show hierarchical clustering and correlation among the displayed genes. Key genes shared between murine and human datasets are highlighted in bold. Data in (A) to (D) are presented as mean ± SD, and dots represent individual samples. Analysis by unpaired Student’s t test. For all panels, n = 3. *P < 0.05.

Reanalysis of published RNA-seq data, including 6 healthy and 10 intermediate- or late-stage OA cartilage samples (49), revealed an increase in several key genes related to fatty acid metabolism in OA cartilage, similar to their murine homologs that were up-regulated by BMP2 stimulation (Fig. 3E). In addition, our omic analysis also revealed conservation of glucose, glutamine, and fatty acid transporters between murine and human articular chondrocytes, with some species-specific differences in expression. Glucose transporters Slc2a1 (murine) and SLC2A1 (human) were the most abundantly expressed in both murine and human articular chondrocytes. Similarly, glutamine transporters, Slc1a5/SLC1A5, Slc7a2/SLC7A2, and Slc38a2/SLC38A2, showed the highest expression. Among fatty acid transporters, Slc27a3/SLC27A3 and Slc27a4/SLC27A4 were highly expressed in both species (fig. S12). The high expression of these transporters supports the metabolic-related findings in murine articular chondrocytes as being relevant to human articular chondrocytes. Collectively, these findings suggest that increased fatty acid metabolism is associated with the OA disease and is a common feature of OA chondrocytes in preclinical models and diseased tissues.

Nuclear factor I is a key transcription factor regulating lipid metabolism in articular chondrocytes

To explore the molecular mechanisms by which lipid metabolism was regulated in articular chondrocytes, we performed ATAC-seq and integrated the analysis with RNA-seq and ATAC-seq data. At the genome-wide chromatin structural level, ATAC-seq identified 6630 and 5580 regions that became more accessible in responding to TGF-β1 and BMP2 treatment in murine articular chondrocytes, respectively (Fig. 4A). Consistent with RNA-seq findings, we discovered that genes located near BMP2-responsive regions were highly enriched in regulation of cell cycle, ossification, lipid metabolism, and catabolic process, whereas TGF-β1–induced accessible regions were associated with skeletal and connective tissue development (Fig. 4B). We then examined the binding motifs of transcription factors by motif analysis in these more accessible regions. As expected, the transcription factors identified in TGF-β1–induced accessible regions included SMAD2 and FOXO1, which has been reported previously (50, 51). Nuclear factor I (NFI) was one of the most enriched transcription factors in BMP2-induced accessible regions in murine articular chondrocytes (Fig. 4C and data file S1). To better understand the correlation between key enzymes in the regulation of fatty acid metabolism and evidence of enrichment of the transcription factor NFI in BMP2-treated murine articular chondrocytes, we analyzed the DNA sequence of open chromatin regions around genes belonging to the fatty acid metabolism pathway. Many of the genes involved in both fatty acid synthesis and oxidation contained NFI-binding motifs in cis-regulatory regions (Fig. 4D) (52). For instance, NFI-binding motifs were identified around cis-regulatory regions of the murine Acaca gene, and BMP2 treatment resulted in a more accessible NFI-binding site in Acaca cis-regulatory regions in murine articular chondrocytes, compared with that seen with TGF-β1 induction (Fig. 4E). Furthermore, Acaca cis-regulatory regions were conserved between human and mouse, and these conserved regulatory elements became more accessible in the arthritic cartilage from patients with OA (Fig. 4E) (53), implicating a conserved and common role for NFI-mediated fatty acid metabolism in OA disease.

Fig. 4. ATAC-seq identified NFI as a key regulator of fatty acid metabolism in murine articular chondrocytes.

ATAC-seq analysis was performed on murine articular chondrocytes treated with TGF-β1 and BMP2. (A) Volcano plot of chromatin accessibility changes induced by TGF-β1 and BMP2 treatment. (B) Enriched biological process terms around differentially accessible regions induced by TGF-β1 and BMP2. Fatty acid metabolism–related pathways are highlighted in red. (C) Left: ATAC-seq signal enrichment on DARs (differentially accessible regions) that associated with TGF-β1 and BMP2 treatment. Right: Transcription factor–binding motifs enriched in TGF-β1 and BMP2-specific DARs. (D) Existence of NFI-binding motifs in cis-regulatory regions of genes in regulation of fatty acid synthesis and oxidation process. (E) Left (genome browser view): Conserved NFI-binding pattern around Acaca cis-regulatory regions between mouse and human. Left top: ATAC-seq signal in TGF-β1 – and BMP2-treated murine articular chondrocytes, alongside the ChIP-seq signal of NFIA in murine cells. Left bottom: ATAC-seq signals in control and OA human articular chondrocytes, along with ChIP-seq signals of NFIA in human cells. Right (box plot): increased ATAC-seq signals at the Acaca cis-regulatory regions in BMP2-treated murine articular chondrocytes (top) and at the ACACA cis-regulatory regions in human OA cartilage (bottom).

We also explored the key epigenetic landscape, particularly DNA methylation alterations, by performing genome-wide methylation-sensitive restriction enzyme sequencing (MRE-seq) in murine articular chondrocytes. We observed 2689 and 2033 regions across the entire genome with reduced DNA methylation in response to TGF-β1 and BMP2 treatment, respectively (fig. S13A). Consistent with the transcriptomic and chromatin changes, genes located around TGF-β1–induced hypomethylated regions were highly enriched in skeletal development, particularly cartilage development, whereas BMP2-induced hypomethylated regions were highly enriched in lipid metabolism (fig. S13B). For instance, TGF-β1–induced hypomethylated regions included Foxo1, which is known to stimulate chondrocyte anabolism and homeostasis (54). The NFI-binding motif was highly enriched in BMP2-induced hypomethylated regions (fig. S13C). Together, these multiomic analytical approaches implicate NFI as a critical transcription factor in the regulation of fatty acid metabolism in BMP2-treated articular chondrocytes undergoing OA-like hypertrophic progression, raising the possibility that it is causally involved in OA pathogenesis in mice and humans.

NFIA loss of function restores articular chondrocyte lipid metabolism and cellular homeostasis

There are four members in the Nfi family—Nfia, Nfib, Nfic, and Nfix. In murine articular chondrocytes treated with BMP2, the RNA-seq showed significant up-regulation of Nfia and Nfib (P < 0.05), whereas Nfic and Nfix were not significantly changed (P > 0.05) (fig. S14A). In human OA articular chondrocytes, qPCR found that NFIA was the only NFI family member increased (Fig. 5A), which was further supported by immunofluorescent staining data showing increased NFIA expression in OA articular cartilage (Fig. 5B). We next examined the physical interaction between NFIA and its binding sites. Chromatin immunoprecipitation (ChIP) assays validated the interaction with its binding site in the 1-kb cis-regulatory regions of Acaca (fig. S14, B and C) and Cpt2 genes in cultured cells (fig. S14, D and E). Lentiviral-mediated overexpression of Nfia in murine articular chondrocytes induced the expression of both Acaca and Cpt2 (fig. S15A). In contrast, knockdown of Nfia expression by short hairpin RNA (shRNA) prevented up-regulation of Acaca and Cpt2 induced by BMP2 treatment in murine articular chondrocytes (fig. S15B).

Fig. 5. NFIA inhibition restored fatty acid metabolism in human OA articular chondrocytes.

(A) qPCR analyses for the gene expression of NFIA, NFIB, NFIC, and NFIX from healthy and OA articular chondrocytes. The gene expression was normalized to ACTB and then healthy cartilage. (B) Immunofluorescent staining for NFIA on human healthy cartilage and OA cartilage (n = 3). Scale bars, 100 μm. (C) qPCR analyses for the gene expression of NFIA, ACACA, and CPT2 from OA articular chondrocytes treated with control and lenti-shNFIA virus. The gene expression was normalized to ACTB and then control treatment. (D) The percentage of ¹³C-labeled and -unlabeled palmitoylcarnitine from U-¹³C glutamine in OA articular chondrocytes treated with control and lenti-shNFIA virus. (E) Fraction of citrate M + 0 and M + 2 in OA articular chondrocytes treated with control and lenti-shNFIA virus. (F) qPCR analyses for the gene expression of MMP13, RUNX2, COL2A1, and ACAN from OA articular chondrocytes treated with control and lenti-shNFIA virus. The gene expression was normalized to ACTB and then control treatment. Data in (A), (C), (D), (E), and (F) are means ± SD, and dots represent individual samples. Analysis by unpaired Student’s t test. For all panels, n = 3. *P < 0.05.

Similar to murine articular chondrocytes, the gene expression of ACACA and CPT2 was significantly decreased in human OA articular chondrocytes with knockdown of NFIA (P < 0.05) (Fig. 5C). The total ¹³C enrichment of palmitoylcarnitine from U-¹³C glutamine in human OA articular chondrocytes decreased to 5% (Fig. 5D), which was similar to the percentage found in healthy articular chondrocytes (Fig. 3B). The fraction of newly synthesized palmitate, as determined by ISA, was significantly decreased as well in human OA articular chondrocytes with NFIA inhibition (P < 0.05) (table S4). Other intermediate metabolites, including PC, CAR, LPC, and LPE, were also decreased by knockdown of NFIA in human OA articular chondrocytes (fig. S16A). Compared with the control treatment, NFIA knockdown led to a significant reduction in the enrichment of M + 2 citrate, glutamate, and fumarate derived from U-¹³C palmitate in human OA articular chondrocytes (P < 0.05) (Fig. 5E and fig. S16B). Energy-related molecules in the TCA cycle, such as dGTP, NAD⁺, NADH, and ATP, were markedly reduced as well after NFIA knockdown (fig. S16B). NFIA knockdown restored cellular homeostasis in human OA articular chondrocytes, evidenced by reduced expression of catabolic genes, such as RUNX2 and MMP13 as well as increased expression of cartilage matrix genes, such as ACAN and COL2A1 (Fig. 5F). Together, these results suggest that the process of fatty acid synthesis and oxidation can be restored in human OA articular chondrocytes after NFIA inhibition. Collectively, our results suggest that increased NFIA may, at least in part, explain the dysregulation of articular chondrocyte anabolism and catabolism and implicate NFIA as a potential therapeutic target to treat OA.

Nfia inhibition protects against metabolic stress–mediated OA in mice

Given the beneficial effect of NFIA knockdown on restoration of chondrocyte fatty acid metabolism and homeostasis in cell cultures, we next investigated the impact of Nfia inhibition on OA progression under systemic metabolic stress in mice. First, we confirmed that fatty acid metabolism in articular chondrocytes was increased in HFD-induced obese mice. Wild-type mice (C57BL/6J, 2 months old) were fed an HFD, leading to a significant increase in body weight over 8 weeks (P < 0.05) (fig. S17A). This increase was accompanied by substantial lipid droplet accumulation in the bone marrow, characterized by elevated PC and phosphatidylethanolamine (fig. S17B) (55). The blood glucose concentration and serum lipid metabolites, such as cholesterol, triglyceride, and free fatty acid, were also significantly increased (P < 0.05) (Fig. 6, A and B), confirming systemic metabolic stress in HFD-fed mice. NFIA expression was found to be up-regulated in the cartilage tissues of HFD-fed wild-type mice experiencing this metabolic stress. The expression of ACACA and CPT2 was also up-regulated in cartilage tissue IHC with HFD (Fig. 6C), implicating the increased fatty acid metabolism in cartilage under whole-body metabolic stress. Elevated expression of NFIA, ACACA, and CPT2 was also observed in the synovium (fig. S18). In line with the up-regulation of these metabolic enzymes in joint tissues, wild-type mice on an HFD exhibited enhanced fatty acid synthesis and oxidation in articular cartilage and synovium, as evidenced by significantly increased lysoPC and carnitine (P < 0.05) (fig. S19, A to D). Consistent with observations in animal models, we confirmed a significantly up-regulated expression of NFIA, ACACA, and CPT2 in osteoarthritic chondrocytes from obese individuals (BMI > 30) compared with normal weight patients with OA by qPCR analysis (fig. S20A). We also found reduced methylation in the 3-kb cis-regulatory regions of the NFIA gene in obese osteoarthritic chondrocytes (fig. S20B), which at least partially contributes to the up-regulation of NFIA.

Fig. 6. Obesity led to systemic metabolic stress, increased chondrocyte fatty acid metabolism, and injury-induced OA progression in mice.

(A) Glucose and (B) lipid metabolites were measured in sera isolated from control and obese mice. (C) IHC for NFIA, ACACA, and CPT2 on murine control and obese cartilage (n = 5). Scale bars, 100 μm. (D) Safranin O/fast green staining of knee joints of control and obese mice at 8 weeks post–MLI injury (n = 5). Yellow arrow, inflammatory synovial tissue. Scale bars, 100 μm. (E) OARSI score and synovitis score for the knee joints of control and obese mice at 8 weeks post-MLI injury. (F) Representative images of subchondral bone from control and obese mice at 8 weeks post–MLI injury (n = 5). Scale bars, 0.5 mm. (G) BV/TV of subchondral bone from medial tibia plateau of control and obese mice at 8 weeks post–MLI injury. (H) Pain assessment for control and obese mice at 8 weeks post–MLI injury. Data in (A), (B), (E), (G), and (H) are means ± SD, and dots represent individual mice. Analysis by unpaired Student’s t test [(A), (B), (E), and (G)] and two-way ANOVA with Tukey’s post hoc multiple comparisons test (H). For all panels, n = 5. *P < 0.05. FFA, free fatty acid.

Joint injury combined with obesity appeared to further promote fatty acid metabolism in joint tissues. Four weeks post–joint injury by MLI, an increase in fatty acid synthesis and oxidation was detected in both articular cartilage and synovium. Coincident with increased fatty acid metabolism in joint tissues, wild-type mice fed the HFD experienced a more severe OA phenotype 8 weeks post–joint injury (Fig. 6, D and E). In comparison with the mild OA phenotypes in cartilage, synovium, and subchondral bone seen in mice fed the control diet, HFD-fed mice had complete cartilage loss, severe synovial hyperplasia, and inflammation as well as medial tibia plateau thickening (Fig. 6, D to G). Mice fed the HFD were also more sensitive to mechanical allodynia at the plantar surface of the paw and the injured knee joint (P < 0.05) (Fig. 6H).

To investigate the mechanistic role of Nfia inhibition in the progression of OA under systemic metabolic stress, we generated AcanCre^ERT2;Nfia^f/f [Nfia loss of function (LOF)] mice to conditionally delete the Nfia gene in cartilage. Nfia gene ablation was confirmed in articular chondrocytes isolated from Nfia LOF mice with no compensatory up-regulation observed in other Nfi genes (fig. S21A). Histology and micro-CT (micro–computed tomography) analyses showed no detectable baseline alterations of the knee joint, including articular cartilage, synovium, and subchondral bone, in Nfia LOF mice (fig. S21, B to D). Two-month-old Nfia LOF mice were then fed the HFD for 8 weeks followed by joint trauma induced through MLI surgery in the right knee. Although the HFD led to body weight gain, lipid droplet accumulation, and elevated serum glucose and lipid metabolites, these metabolic phenotypes were comparable between control and Nfia LOF mice (Fig. 7, A and B, and fig. S22, A and B). Conditional deletion of Nfia suppressed fatty acid synthesis and oxidation in cartilage, as evidenced by reduced lysoPC and carnitine under HFD conditions (fig. S23, A to D). Meanwhile, Nfia knockdown largely mitigated joint injury–induced cartilage loss and synovial hyperplasia under the HFD condition, with significantly reduced OARSI scores and synovitis scores compared with control mice (P < 0.05) (Fig. 7, C and D). Micro-CT also revealed a reduced subchondral bone BV/TV in the medial tibia plateau of Nfia LOF mice fed the HFD and challenged with joint injury (Fig. 7, E and F). Moreover, Nfia LOF mice showed an improved pain response to mechanical stimuli at the paw and the injured knee joint, as reflected by the increased von Frey and SMALGO (small-animal analgesiometer) behavior measurements (Fig. 7G). Control mice displayed higher sensitivity to mechanical allodynia at both the plantar surface of the paw (von Frey) and the injured knee joint itself (SMALGO). Nfia LOF mice showed more tolerance and less pain response to these stimuli.

Fig. 7. Nfia conditional knockout protected against injury-induced OA progression in obese mice.

(A) Glucose and (B) lipid metabolites were measured in the sera isolated from control and Nfia LOF obese mice. (C) Safranin O/fast green staining of knee joints of control and Nfia LOF obese mice at 8 weeks post–MLI injury (n = 7). Yellow arrow, inflammatory synovial tissue. Scale bars, 100 μm. (D) OARSI score and synovitis score for the knee joints of control and Nfia LOF obese mice at 8 weeks post–MLI injury. (E) Representative images of subchondral bone from control and Nfia LOF obese mice at 8 weeks post–MLI injury (n = 7). Scale bars, 0.5 mm. (F) BV/TV of subchondral bone from medial tibia plateau of control and Nfia LOF obese mice at 8 weeks post–MLI injury. (G) Pain assessment for control and Nfia LOF obese mice at 8 weeks post–MLI injury. Data in (A), (B), (D), (F), and (G) are means ± SD, and dots represent individual mice. Analysis by unpaired Student’s t test [(A), (B), (D), and (F)] and two-way ANOVA with Tukey’s post hoc multiple comparisons test (G). For all panels, n = 7. *P < 0.05.

Fatty acid metabolism suppression attenuates metabolic stress–mediated OA progression in mice

Last, we examined the effect of fatty acid metabolism suppression in maintenance of joint homeostasis under obese conditions. Two-month-old wild-type mice were fed the HFD for 8 weeks to induce obesity and systemic metabolic stress. The obese mice then underwent MLI surgery to synchronize the OA progression under the continuous HFD feeding conditions. After MLI surgery, the obese mice were administered ND646 and R-AminoCarnitine for 8 weeks to suppress the fatty acid synthesis and oxidation by inhibition of ACACA and CPT2 activity systemically (56–58). ND646 and R-AminoCarnitine–treated mice maintained a lower relative body weight compared with HFD alone (fig. S24A). The metabolic stress induced by HFD was also significantly attenuated, as reflected by lower blood glucose and serum fatty acid metabolites (P < 0.05) (fig. S24, B and C). With these mitigated metabolic stresses, fatty acid metabolism suppression attenuated HFD-mediated OA progression in mice. In saline-treated obese mice, complete cartilage loss and synovial hyperplasia were observed 8 weeks postinjury. In contrast, cartilage integrity was better maintained, and synovial hyperplasia was mitigated in ND646 and R-AminoCarnitine–treated mice (Fig. 8A). Consistent with histological observations, cartilage grading confirmed that ND646 and R-AminoCarnitine–treated mice had a significantly lower OARSI score (P < 0.05) (Fig. 8B) and synovitis score (P < 0.05) (Fig. 8C). Moreover, micro-CT assessments demonstrated increased subchondral bone BV/TV in the medial tibia plateau of saline-treated obese mice by 8 weeks postsurgery. Conversely, ND646 and R-AminoCarnitine treatment at least partially protected against this phenotype, as indicated by a significantly reduced BV/TV in the medial tibia plateau (P < 0.05; Fig. 8, D and E). In addition, ND646 and R-AminoCarnitine–treated mice displayed less stimulus-evoked nociception on the basis of von Frey and SMALGO tests. Saline-treated mice displayed higher sensitivity to mechanical allodynia at both the plantar surface of the paw (von Frey) and the injured knee joint itself (SMALGO) (Fig. 8F). Together, these findings support that NFIA is a key regulator of cartilage fatty acid metabolism and chondrocyte homeostasis and implicate fatty acid metabolism inhibition as a candidate approach to treating OA disease on the basis of our observation of chondroprotection, reduced synovitis, correction of subchondral sclerosis, and alleviation of pain.

Fig. 8. Fatty acid metabolism suppression attenuates metabolic stress-mediated OA progression in mice.

(A) Safranin O/fast green staining of knee joints of saline as well as ND646 and R-AminoCarnitine treated obese mice at 8 weeks post–MLI injury (n = 5). Yellow arrow, inflammatory synovial tissue. Scale bars, 100 μm. (B) OARSI score for the knee joints of saline as well as ND646 and R-AminoCarnitine–treated obese mice at 8 weeks post–MLI injury. (C) Synovitis score for the synovium tissue of saline as well as ND646 and R-AminoCarnitine–treated obese mice at 8 weeks post–MLI injury. (D) Representative images of subchondral bone from saline as well as ND646 and R-AminoCarnitine–treated obese mice at 8 weeks post–MLI injury (n = 5). Scale bars, 0.5 mm. (E) BV/TV of subchondral bone from medial tibia plateau of saline as well as ND646 and R-AminoCarnitine-treated obese mice at 8 weeks post-MLI injury. (F) Pain assessment for saline as well as ND646 and R-AminoCarnitine–treated obese mice at 8 weeks post–MLI injury. Data in (B), (C), (E), and (F) are means ± SD, and dots represent individual mice. Analysis by unpaired Student’s t test [(B), (C), and (E)] and two-way ANOVA with Tukey’s post hoc multiple comparisons test (F). For all panels, n = 5. *P < 0.05.

DISCUSSION

Despite a growing understanding of the pathogenic signatures that lead to OA, which is based on studies from animal models and human patients with OA, it remains the most prevalent musculoskeletal disease without a clinically accepted disease-modifying therapeutic intervention. This is largely due to lack of comprehensive insights into the signaling and molecular mechanisms underlying OA initiation and progression. In this study, we applied an unbiased multiomic approach in combination with established murine OA models to demonstrate that (i) OA is a disease associated with increased fatty acid synthesis and oxidation in articular chondrocytes, (ii) NFIA is a pivotal transcription factor involved with the regulation of fatty acid metabolism in these cells, (iii) NFIA knockout restores homeostasis in human OA chondrocytes and attenuates OA progression in mice, and (iv) suppression of fatty acid metabolism protects against metabolic stress–mediated OA in mice.

NFIA mutations, including nonsense and missense variants, frame-shift mutations, and intragenic deletions, are predominantly reported in human patients with disorders affecting the central nervous system and urinary tract (59); however, recent reports have identified musculoskeletal defects in two patients with de novo NFIA mutations (60). In addition, NFIA has been identified as a critical regulator of articular cartilage differentiation and maintenance in chick embryonic development (61). Thus, our findings highlight fatty acid metabolism dysregulation as a conserved metabolic mechanism underlying OA disease and implicate NFIA and its regulated fatty acid metabolism as promising therapeutic targets for clinical intervention in OA.

Numerous studies have established the pivotal role of the TGF-β pathway in maintenance of articular cartilage during homeostasis and during OA development and progression (42). Mechanistically, these pathways signal either by receptor ALK5 to activate canonical SMAD2/3 (TGF-β pathway), maintain chondrocyte homeostasis, and prevent hypertrophic differentiation or by receptor ALK1 to stimulate SMAD1/5/9 (BMP pathway) and induce chondrocyte hypertrophy, matrix catabolism, and ultimately apoptosis. It follows that Alk5 gene expression has been consistently reported as reduced in OA chondrocytes. With this reduced signaling on the TGF-β pathway, findings have revealed initiation and rapid progression of OA in murine genetic models (62–64) and human patients (65, 66). Despite a report of decreased Alk1 gene expression in aged murine chondrocytes (67), ALK1 gene and protein expression were found to be increased in murine and human OA cartilage (43), and an increased ALK1/ALK5 ratio has been documented in OA chondrocytes from rodents (C57/BL1/6 mice), minipigs, and human patients, implicating a conserved and activated BMP pathway in OA pathogenesis (68, 69). We note that the phenotype of human OA articular chondrocytes, including anabolic and catabolic responses that are at least partially mediated by ALK1 and ALK5 alterations, highly depends on the disease stages (70, 71).

To more deeply examine the molecular mechanisms and phenotypes in OA chondrocytes, we used in vitro surrogates for healthy and OA chondrocytes to support the feasibility of multiomic approaches to uncover unappreciated pathways that drive disease. We used TGF-β1 to maintain murine primary articular chondrocytes in a nonhypertrophic and healthy state. Conversely, we used BMP2 to stimulate hypertrophic maturation in these murine cells, inducing an OA-like disease state. Through comprehensive transcriptome and epigenome screening, NFI-mediated fatty acid synthesis and oxidation up-regulation were identified to be correlated with disease (for example, under the influence of BMP2) in our murine articular chondrocyte model. Increased fatty acid synthesis and oxidation were also identified in human OA articular chondrocytes, indicating the tight association between fatty acid metabolism and articular chondrocyte homeostasis. The DNA methylation in the cis-regulatory regions of Foxo1 (50, 54), a well-known regulator of chondrocyte homeostasis, was decreased by TGF-β1 treatment, reinforcing that TGF-β1 preserved the murine articular chondrocytes in the healthy/homeostatic state through an epigenetic mechanism. Moreover, BMP2 treatment reduced methylation in the Nfi cis-regulatory regions, therefore maintaining murine articular chondrocytes in a state that is associated with increased Nfi expression and fatty acid metabolism. This finding is consistent with recent single-cell RNA-seq data showing that a distinct cluster of articular chondrocytes from the superficial layer of human OA cartilage displayed increased fatty acid metabolism (35). In addition, vehicle-treated articular chondrocytes displayed a genetic profile similar to BMP2-treated cells (figs. S22 and S23), which is consistent with the established concept that articular chondrocytes spontaneously undergo hypertrophic differentiation in monolayer cultures. BMP2 treatment synchronized the terminal hypertrophy status in articular chondrocytes, which showed an enhanced fatty acid metabolism signature even in comparison with the vehicle treatment.

OA is now widely appreciated as a metabolic disease, frequently associated with a metabolic syndrome involving obesity, type 2 diabetes mellitus, and elevated blood glucose and triglycerides (72). It has been revealed that altered glucose (40) or glutamine (73) metabolism causes inappropriate hypertrophy and production of matrix-degrading enzymes in chondrocytes and a resultant degeneration of articular cartilage and OA progression in mice. Here, we showed that besides glucose and glutamine metabolism, fatty acid metabolism regulates articular chondrocyte function and homeostasis as well. Given that articular chondrocytes are in an avascular environment, the nutrients together with oxygen are in limited supply compared with other cells, implying that articular chondrocytes are adapted to perform their maintenance functions with limited and likely specific metabolic requirements, for example, low oxygen demand and limited fatty acid availability. Consequently, articular chondrocytes maintain their activities, particularly related to extracellular matrix (ECM) synthesis, using a distinct metabolic profile characterized by extremely low fatty acid synthesis and oxidation processes (26) within the constraints of the cartilage and synovial joint environment. In OA, however, hypervascularization with increased blood flow occurs, especially in the synovium and subchondral bone regions (27–32), leading to an increased supply of oxygen and nutrients to the articular cartilage. As OA progresses, cartilage matrix degradation and tissue thinning further facilitate a more uniform distribution of oxygen throughout the cartilage (74). In this altered environment, diseased articular chondrocytes shift to a hypertrophic and matrix catabolic state, with increased energy demands to modify ECM composition and structural integrity (33, 34, 75). Diseased articular chondrocytes show elevated expression of matrix metalloproteinases (MMPs) and a disintegrin and metalloproteinase with thrombospondin motifs, which degrade the primary collagen backbone, type II collagen. Furthermore, diseased articular chondrocytes produce other types of collagens, such as type I, type III, and type X collagen, contributing to ECM composition changes characteristic of OA pathology (76). Mechanistically, we provide the evidence that NFIA is a crucial transcription factor governing this OA-related metabolic shift in murine articular chondrocytes through up-regulation of essential fatty acid metabolism–related genes, such as Acaca and Cpt2. Our in vitro rescue experiments demonstrate that inhibition of Nfia in both murine and human articular chondrocytes attenuates the increased expression of Acaca and Cpt2 under OA disease conditions, which in turn leads to suppression of fatty acid synthesis and oxidation in cells. Similarly, the indispensable role of Nfia has also previously been observed in preadipocytes, where gain of function promotes fatty acid formation in 3T3-L1 cells, whereas LOF using an Nfia shRNA inhibits fatty acid accumulation (77). In addition, NFIA-binding sites are identified in the cis-regulatory regions of the Runx2 gene, suggesting potential direct regulation of Runx2 expression by NFIA in articular chondrocytes. Given the established role of Runx2 in regulating articular chondrocyte catabolism and OA progression (44), Nfia activation in OA articular chondrocytes could stimulate both fatty acid metabolism by Acaca and Cpt2 and cellular catabolism by Runx2.

Our study further demonstrates that the up-regulation of NFIA and its downstream targets, ACACA and CPT2, is associated with whole-body metabolic stress induced by obesity. Previous studies have reported that obesity elevates glucose and lipid metabolite concentrations in serum and within the knee joint (78, 79) and increases fatty acid metabolism in OA articular chondrocytes (35). Our findings provide strong evidence that NFIA is at least one of the key transcription factors responsive to systemic metabolic challenges and therefore mediates increased fatty acid metabolism and catabolism in cartilage in mice. Although all experimental mice used in this study were of the C57BL/6J background, which carries the Nnt mutation potentially affecting cellular metabolism, this strain is commonly used in diet-induced obesity models to study metabolic challenges across various organs. C57BL/6J mice exhibit robust responses to diet-induced obesity comparable to other strains with the wild-type Nnt allele, such as C57BL/6NJ, C57BL/6JNomTac, C57BL/6JRj, Kunming, BALB/c, and ICR (80–82). In our study, C57BL/6J mice displayed similar changes in body weight, adipose tissue accumulation, and serum glucose and lipid concentrations. In addition, in vitro assays indicated normal fatty acid metabolism in primary articular chondrocytes isolated from C57BL/6J mice. These results, consistent with previous studies, underscore the strain’s suitability for metabolic studies. To evaluate the therapeutic potential of Nfia inhibition, we assessed its protective effect in an obesity-related OA model. We demonstrated that inhibiting Nfia in cartilage mitigated injury-induced OA progression in obese mice. Specifically, Nfia-deficient mice exhibited a less severe OA phenotype compared with littermate controls, including preserved articular cartilage integrity, reduced subchondral bone sclerosis and synovial inflammation, and alleviated knee joint pain. In addition to articular cartilage, fatty acid synthesis and oxidation were also up-regulated in the synovia of obese mice, as evidenced by elevated expression of NFIA, ACACA, and CPT2, along with increased intermediate metabolites, such as lysoPC and carnitine. These elevated fatty acid metabolites in cartilage and synovium may alter the intraarticular environment (83) particularly in OA, leading to M1 polarization and synovial inflammation (84, 85), as well as osteogenic differentiation and subchondral bone sclerosis (86–88). Collectively, these changes can affect the entire joint and exacerbate OA progression.

In our study, only adult male mice were used for the injury-induced OA model, given that male mice develop more severe OA phenotypes than females after surgical induction (89). Although it is a commonly accepted practice in the field, we acknowledge that additional studies are needed to evaluate the protective effects of NFIA inhibition in female mice. In addition, although NFIA is the only up-regulated NFI family member in human articular chondrocytes in OA, both Nfia and Nfib were found to be up-regulated by BMP2. Given the overlapping and distinct roles of Nfia and Nfib (77, 90), it would be valuable to also clarify the role of Nfib in articular chondrocytes and further delineate the molecular mechanism by which Nfia and Nfib coregulate the behavior of articular chondrocytes and thus could be coordinately involved in joint homeostasis and disease.

Overall, the data presented in this report support NFIA inhibition and fatty acid metabolism suppression as potential therapeutic strategies that have substantial translational promise. Restoration of fatty acid metabolism in articular chondrocytes may be disease modifying in patients with OA and could be even more beneficial in patients with OA suffering from comorbidities including obesity and other metabolic syndromes. This is supported by our observation of further elevated expression of NFIA, ACACA, and CPT2 in osteoarthritic chondrocytes from obese individuals, partially due to reduced methylation of the NFIA gene. These findings align with our previous report demonstrating that the DNA methylation enzyme DNMT3B is down-regulated in OA chondrocytes under obesity-associated conditions (91–93). There are now several medications available to target fatty acid metabolism to treat patients with cancer (94), which could be repurposed for use in patients with OA as well. Our findings support the development of medications that target fatty acid metabolism through manipulation of NFIA because the available cancer medications are designed to target other pathways, such as mTOR and AMPK (95), which may not have the same pivotal role in the metabolic activity of chondro-lineage cells.

MATERIALS AND METHODS

Study design

In this study, both in vitro cell culture and multiple animal models of OA were used to investigate the mechanisms underlying OA pathogenesis. A multiomic approach including RNA-seq, ATAC-seq, and metabolomics was applied to comprehensively assess the effect of NFIA-mediated fatty acid metabolism on the homeostasis of articular chondrocyte. All cell culture experiments were performed in three biological repeats to assess treatment-induced alterations. Animal sample sizes were determined through power analysis to achieve a statistical significance level of 0.05 for OARSI scores between control and experimental groups, targeting a 25% effect size with an anticipated within-group SD not exceeding 15%. On the basis of our recent published histological findings using the same obesity OA murine models (96) and using a commercial power analysis tool (97), we determined that a sample size of five to eight mice per group was sufficient to provide adequate power to detect the desired difference. Animals were randomly assigned to control and experimental groups. Histological evaluations and pain response assessments were performed by three independent, blinded observers. Animal studies were approved under protocol 24–0286 at Washington University in St Louis. Details of primer sequences for qPCR (table S5), instrument parameters for matrix-assisted laser desorption/ionization–MSI (mass spectrometry imaging) (table S6), human participant BMI in fig. S20 (table S7), and primer sequences for ChIP-qPCR (table S8) are reported in the Supplementary Materials.

Statistical analysis

In addition to RNA-seq, ATAC-seq, and MRE-seq, statistical analyses were performed using GraphPad Prism for all other datasets. Normality was assessed using the Shapiro-Wilk test before statistical analysis. Statistical significance between two groups was determined using a two-tailed Student’s t test with a 95% confidence interval. For comparisons among multiple groups, two-way analysis of variance (ANOVA) followed by Tukey’s post hoc test was applied. All data were presented as the means ± SD of at least three independent experiments. Differences between groups were considered statistically significant at a threshold of P < 0.05. All individual-level data are available in data file S2.

Data and materials availability:

All data associated with this study are present in the paper or data file S2. The RNA-seq, ATAC-seq, and MRE-seq data generated in this study have been deposited in the Gene Expression Omnibus (GEO) database under accession GSE188327. LC-MS data have been uploaded to the Metabolomics Workbench (98) under project identifier ST003598 (http://dx.doi.org/10.21228/M8WG15). Code used for analysis of RNA-seq, ATAC-seq, and MRE-seq data has been deposited in Zenodo (https://doi.org/10.5281/zenodo.15760321).