Changes in energy molecule metabolism contribute to cartilage breakdown and osteoarthritis

To investigate the role of NAD⁺ in OA, we first determined the level of NAD⁺ biosynthesis in human OA and a mouse ageing model. Following our previous work,11 G1 (grade 1) cartilage was isolated from the femur surface with no evidence of cartilage damage, whereas G4 (grade 4) cartilage was derived from the damaged surface with a significantly higher modified Mankin score (Figure 1A and B). We found a significant decrease in NAD⁺ levels and total NAD (tNAD, the sum of both the NAD⁺ and NADH species) in human G4 cartilage when compared to G1 cartilage (Figure 1C), while NADH and the calculated NAD⁺/NADH ratio were non‐significant between G1 and G4 (Figure S1). NAD⁺ is primarily synthesised in mammalian cells via the de novo NAD⁺ synthesis pathway, the Preiss–Handler pathway via NA, and the salvage pathway (Figure 1D). Real‐time quantitative PCR (RT‐qPCR) analysis revealed that NAMPT gene expression was significantly increased in G4 cartilage, whereas other enzymes showed no significant change (Figure 1E–J). The increased expression of NAMPT in G4 cartilage was confirmed by Western blotting and immunohistochemical (IHC) staining (Figure 1K–N). Following that, we measured metabolites from the kynurenine pathway (Figure 1O–Q), with no change observed. Instead, we observed an increase in the NAD⁺ precursors nicotinamide (NAM) and nicotinamide mononucleotide (NMN), along with a decrease in NAD⁺ in G4 cartilage (Figure 1R–T). Together, the accumulation of these NAD⁺ precursors, along with a decline in NAD⁺ levels, suggests declining rates of NAD⁺ biosynthesis at the level of NMN adenosyltransferase (NMNAT) enzymes, which catalyse the conversion of NMN to NAD⁺. To simulate the natural progression of age‐related knee OA in mice, we next assessed cartilage integrity in three age groups of male C57/BL6 mice: young (1–3 months), middle‐aged (10–14 months), and elderly (18–24 months) (Figure 1U).12 Histological findings from these mice demonstrated increased NAMPT expression with age, which correlated with the natural age‐related cartilage degeneration in OA progression (Figure 1V–X). Taken together, these findings suggest that despite increased levels of the NAD⁺ biosynthetic enzyme NAMPT, OA development includes a decline in NAD⁺ levels, which is most likely due to declining activity of the subsequent step in NAD biosynthesis, catalysed by NMNAT enzymes.

Given the decline in NAD⁺ levels during OA, the next key question is whether this is due to an increased rate of NAD⁺ breakdown, and if so, which enzyme(s) were responsible for this. We therefore examined the expression levels of NAD⁺‐consuming enzymes in cartilage, including CD38, PARPs, and Sirtuins, by analysing our previous RNA sequencing data. While PARP1, PARP4, PARP9, PARP14, and SIRT2 were all expressed at a relatively high level in cartilage, PARP14 was the NAD⁺ consuming enzyme among them that was significantly up regulated in G4 cartilage (Figure 2A and B). These RNA‐seq data were confirmed by RT‐qPCR, showing an increase in the level of PARP14 in G4 cartilage (Figure 2C). Immunostaining indicated that PARP14 expression was increased in G4 chondrocytes (Figure 2D and E). We next examined the level of PARP14 in knee cartilage during mouse ageing. As with humans, PARP14 increased significantly in 13‐ and 25‐month mice knee cartilage compared to 6 weeks (Figure 2F and G). Similarly, PARP14 levels were increased in the OA model of rat medial collateral ligament‐medial meniscus (MCL‐MM) surgery (Figure 2H and I). Altogether, these results suggest an increase in the expression of the NAD⁺ consuming enzyme PARP14 in human OA, along with an increase in expression during age‐associated OA in mice, and in a rat OA model.

Given the decline in NAD⁺ levels and elevation in the NAD⁺ consuming enzyme PARP14, we next sought to test whether altered NAD⁺ levels played a causative role in mediating OA. To start with, bovine cartilage explants were cultured with/without IL‐1β and NMN for 48 h in the inflammation ex vivo OA model. Our results indicated that IL‐1β treatment significantly increased glycosaminoglycan (GAG) release, whereas NMN supplementation inhibited GAG release (Figure 3A and B). We next used treatment with the NAD⁺ precursors NMN and NR, which have been frequently used to elevate NAD⁺ levels in preclinical and clinical studies.13, 14 Wistar rats were subjected to sham or MCL‐MM surgery for 8 weeks, followed by daily treatment with NR (200 mg/kg, oral gavage) or vehicle control (Figure 3C). MCL‐MM surgery significantly damaged the medial tibia cartilage, whereas NR supplementation slowed the progression of OA and decreased the Mankin score (Figure 3D and E). Next, we sought to test these findings in the absence of surgical injury, instead examining these changes in the more physiologically relevant context of natural ageing. NAD⁺ levels were significantly reduced in 13‐ and 25‐month‐old mice (Figure 3F). Given this decline in NAD⁺ levels, we established a separate cohort of female C57BL/6 mice to examine the effect of NAD⁺ precursor supplementation. Mice at 13 months of age were randomly assigned to receive either vehicle (regular drinking water) or NMN administered via drinking water at a concentration of 2 g/L for four consecutive weeks (Figure 3G).15 NMN treatment led to a dose‐dependent reduction in the severity of cartilage degeneration and decreased the Mankin score compared to controls (Figure 3H and I). In addition to these histological improvements, NMN supplementation in mice reduced the expression of genes associated with OA degeneration while increasing the expression of a chondrogenic protein, Collagen II (COL2) (Figure 3H and J–M). Our in vitro data indicated that NMN had a similar effect on 3D cultured primary chondrocytes from old donors (Figure S1), whereby NMN treatment suppressed the expression of cartilage degeneration genes such as Mmp1, Mmp9, Mmp13 (Figure S2D–G), Dipen and Nitege (Figure S1A and B) while increasing Col2 chondrogenic gene expression. Our earlier data demonstrated an increase in levels of the NAD⁺ biosynthetic gene NAMPT (Figure 1J–M and V), yet a decrease in NAD⁺. This discrepancy suggests a possible bottleneck at the level of NMNAT enzymes, which catalyse the final step in NAD⁺ biosynthesis. To test this, we next used transgenic animals which over‐express the nuclear localised NMNAT enzyme NMNAT1 (Figure 3G).15 NMNAT1 plays a pivotal role in the biosynthesis of NAD⁺, functioning as a key enzyme in the final step of NAD⁺ synthesis.16 Overexpression of NMNAT1 has been shown to significantly elevate intracellular NAD⁺ levels, thereby modulating cellular metabolic and redox states.17, 18 These animals were naturally aged to 13 months, and tissues collected for measurement of age‐associated OA. Notably, NMNAT1‐tg mice were protected against cartilage damage and had a significant reduction in the Mankin score compared to their wild‐type (WT) littermate controls (Figure 3H and I). NMNAT1 overexpression was associated with decreased expression of cartilage degeneration proteins and increased expression of chondrogenic proteins (Figure 3H and J–M). Taken together, these pharmacological and transgenic interventions demonstrate that elevating NAD⁺ levels mitigates cartilage degeneration at the histological and molecular levels in OA models.

These results suggested that increased PARP14 activity in OA could lead to a reduction in NAD⁺ levels, which play a causative and reversible role in OA disease progression. To gain a better understanding of the mechanisms underlying PARP14 and chondrocyte function, we examined whether PARP14 affected the energy metabolism of chondrocytes. Three separate PARP14 siRNAs were transfected into primary human chondrocytes (Figure 4A). The effect of PARP14 on chondrocyte viability was then determined using a cell counting kit‐8 (CCK‐8) assay. Interestingly, our results indicated that silencing PARP14 expression had no effect on the viability of chondrocytes after 72 h (Figure S3). In line with the idea that increased PARP14 activity could drive reductions in NAD⁺ levels, PARP14 siRNA silencing significantly increased NAD⁺ levels in chondrocytes after 24 h (Figure 4B). Given the critical role of NAD⁺ as a cofactor for metabolism, we next examined whether PARP14 impacted glycolysis and oxidative phosphorylation. Comparison of untreated controls with IL‐1β–treated cells showed that IL‐1β markedly increased glucose uptake and lactate production, consistent with a glycolytic shift (Figure 4C–E). In chondrocytes exposed to an inflammatory environment, PARP14 silencing resulted in decreased glucose consumption and lactate production compared to the control group, resulting in a higher pyruvate/lactate ratio (Figures 4C–E and S4). Seahorse analysis further confirmed that IL‐1β enhanced glycolytic flux, as indicated by elevated extracellular acidification rate (ECAR), whereas PARP14 silencing suppressed this effect and restored metabolic balance (Figure 4F–H). Mitochondrial stress testing revealed that PARP14 silencing under IL‐1β stimulation did not result in a significant increase in oxygen consumption rate (OCR), but in a substantial increase in reserve capacity and maximal OCR (Figure 4I–L). Together, these findings indicate that PARP14 plays an important role in mediating changes to chondrocyte metabolism including glycolysis during inflammation.

To test whether these impacts of PARP14 on cell metabolism were relevant to the pathophysiology of OA, we silenced PARP14 in primary chondrocytes by siRNA transfection to test for impacts on the extracellular matrix. PARP14 expression was silenced using siRNA transfection, and cells were subsequently exposed to IL‐1β to mimic an inflammatory environment. As expected, IL‐1β treatment alone markedly increased the expression of matrix‐degrading enzymes, including MMP1 and MMP13, while concomitantly reducing the expression of chondrogenic genes such as aggrecan (Figure 5A and B). When PARP14 was silenced, RT‐qPCR analysis confirmed a robust reduction in PARP14 mRNA expression and revealed a significant downregulation of multiple catabolic and hypertrophic markers, including Collagen X (COL10), A disintegrin and metalloproteinase with thrombospondin motifs 4 (ADAMTS4), ADAMTS5, NOS2, MMP1, MMP9, MMP13 (Figure 5C–K). In contrast, anabolic markers such as collagen type II (COL2) were significantly upregulated (Figure 5D). These results demonstrate that suppression of PARP14 not only dampens matrix catabolism but also helps preserve chondrogenic gene expression, suggesting that PARP14 is a key regulator of the balance between catabolic and anabolic processes in cartilage.

Together, these results indicate that NAD⁺ and PARP14 are involved in OA pathogenesis, and their underlying mechanism is illustrated in Figure 6.

Clinical femur cartilage specimens were obtained from patients undergoing primary total knee arthroplasty. Human ethics approval for this project was granted by the Queensland University of Technology (QUT) and the St Vincent Private Hospital Ethics Committees, and informed consent was obtained from all participants (ethics number: #1400001024). Following the Kellgren and Lawrence grading system, the cartilage tissues collected from the relatively smooth region were classified into the grade 1 (G1, non‐OA) group, and the cartilage collected from the severely damaged region were classified into the grade 4 cartilage (G4, OA) group (n = 19; age: 72.5 ± 5.72 years, gender: 9 males and 10 females; BMI: 29.05 ± 5.31 kg/m²). Both G1 and G4 cartilages were harvested from each participant.

For cell culture, cartilage was cut into small pieces and washed 3–4 times with phosphate‐buffered saline (PBS; Oxoid, Thermo Fisher Scientific, Victoria, Australia, CAT. OXBR0014G) to remove the blood debris. Chondrocytes were isolated by digestion with 1% type II collagenase (Gibco, Thermo Fisher Scientific, CAT. 17101015) for 6–10 h and resuspended in low glucose Dulbecco's Modified Eagle's Medium (DMEM; Gibco, CAT. 11885084) with 10% Fetal Bovine Serum (FBS; Gibco, CAT. A5256701) and 1% penicillin–streptomycin (Gibco, CAT. 15140122) (complete medium) as described previously.28 For 3‐dimensional (3D) pellet culture, 5 × 10⁵ of primary chondrocytes were centrifuged to form a 3D pellet as described previously.28 Pellets were cultured in chondrogenic culture medium (DMEM), high (4.5 g/L) glucose (DMEM; Gibco, CAT. 11965092) supplemented with 1% inulin–transferrin–selenium (Gibco, CAT. 41400045), 1.25 mg/mL bovine serum albumin (Sigma‐Aldrich, CAT. A3311),.1 µM dexamethasone (Sigma‐Aldrich, CAT. D4902),.1 mM ascorbic acid (Sigma‐Aldrich, CAT. A92902), 1% penicillin–streptomycin, 10 mM hydroxyethyl piperazineethanesulphonic acid (HEPES) (Gibco, CAT. 11560496),.1 mM L‐proline (Sigma‐Aldrich, CAT. P0380),.1 mM MEM nonessential amino acids (Gibco, CAT. 11140050), and 10 ng/mL transforming growth factor‐β1 (TGF‐β1, Gibco, CAT. 10021100UG) for 14 days according to our previous study.28

Ageing mice model was performed using young (6 weeks), middle‐aged (13 months), and aged (25 months) male C57/BL6 mice. Transgenic mice models were performed using wild type (WT, 13 months) and NMNAT1 transgenic C57/BL6 mice (NMNAT1 tg, 13 months). Mice were obtained from Dr Lindsay Wu's Laboratories (ethic number: 18/133A). For nicotinamide mononucleotide (NMN) treatment, 13‐month‐old mice were given both.5 and 2 g/L of NMN in drinking water for 4 weeks. For OA surgery model, 8‐week‐old male Wistar rats were purchased from the Animal Resources Centre (Perth, WA, Australia) and divided into four groups: Sham, Sham + nicotinamide riboside (NR, Tru Niagen, CAT. 856029008554), medial collateral ligament‐medial meniscus (MCL‐MM), and MCL‐MM + NR (ethic number: 1900000938, approved by University Animal Ethics Committee, QUT). Right knee of each rat was operated sham or MCL‐MM surgery.29 After that, Sham + NR and MCL‐MM + NR groups were given 200 mg/kg NR oral drop supplement daily for 8 weeks. Mice and rats were group‐housed (6 mice or 3 rats/cage) in ventilated double‐decker cages in a temperature‐controlled room on 12‐h light/dark cycles with ad libitum access to food and water and routine veterinary assessment. Animals were randomly assigned to experimental groups using a random number generator to minimise bias.

Total RNA from G1/G4 cartilage (n = 10) was isolated and tested for integrity with Agilent's Fragment Analyser 5200 (Agilent Technologies, California, United States). RNA‐Seq libraries were then constructed using Qubit 4.0 Fluorometer (Thermo Fisher Scientific) with Qubit dsDNA High Sensitivity Assay Kit (Thermo Fisher Scientific, CAT. Q32851) and checked for quality with the Fragment Analyzer (Agilent Technologies). These were then pooled based on equimolar concentrations and sequenced on the Novaseq 6000 Sequencing System (Illumina) with an S1 PE100 sequencing kit (Illumina). High‐quality reads were then mapped onto the Human Genome Reference Consortium build 38 using Spliced Transcripts Alignment to a Reference (STAR, version 2.7.9a) and feature counts extracted for human Ensembl annotated genes (release 103). Differential gene expression (DE) analysis was conducted using DESeq2 (Bioconductor, version 1.32.0) with an FDR < .05.

Staining was performed following the methods from our previous publication.30 All histological scoring and outcome assessments were performed by investigators blinded to group allocation. Both cartilage and knee joints were fixed in 4% paraformaldehyde (PFA; Sigma‐Aldrich, CAT. P6148), decalcified in 10% ethylenediaminetetraacetic acid (EDTA; Sigma‐Aldrich, CAT. E5134) and embedded in paraffin. All paraffin samples were cut into 5 µm sections. For Safranin O/Fast Green staining, samples were stained with.1% Safranin O solution (Sigma‐Aldrich, CAT. TMS‐009‐C) and 1% Fast Green solution (Sigma‐Aldrich, CAT. F7252). Modified mankin score was used to grade the severity of cartilage degeneration by two observers blinded to group‐identifying information. For Masson's Trichrome staining, samples were stained with Masson's composition solution (Electron Microscopy Sciences, CAT. 26367) and aniline blue solution (Electron Microscopy Sciences, CAT. 50‐317‐77). The percentage of collagen area was calculated using Image J imaging software (version 1.51n). For IHC staining, samples were stained with primary monoclonal antibodies and secondary antibody (EnVision Dual Link, Dako, CAT. K4065). Positively stained cells were evaluated using Image J. The primary antibodies used were as follows: anti‐NAMPT antibody (1:200, Abcam, Cambridge, United Kingdom, ab45890), anti‐DIPEN antibody (1:500, kind gift from Professor Amanda Fosang, Murdoch Childrens Research Institute, Melbourne, VIC, Australia), anti‐NITEGE antibody (1:500, kind gift from Professor Amanda Fosang, Murdoch Childrens Research Institute, Melbourne, VIC, Australia), anti‐COL2 antibody (1:200, Abcam, ab34712), anti‐MMP13 antibody (1:200, Abcam, ab39012), anti‐PARP14 antibody (1:200, Santa Cruz Biotechnology, Texas, United States, sc‐377150).

Cartilage sample was grounded into powder in liquid nitrogen and homogenised with NAD/NADH Extraction Buffer. The supernatant was collected and filtered through a 10 kD spin column (Abcam, CAT. ab93349). NAD⁺ and total NAD (tNAD) levels were measured using NAD/NADH Assay Kit (Colorimetric; Abcam, CAT. ab65348) according to the manufacturer's instructions. Reading was taken at Optical Density (OD) 450 nm.

Total RNA was extracted using TRIzol reagent following our previous publication.30 Reverse transcription of mRNAs to cDNA was performed using miScript PCR Starter Kit (Qiagen, Hilden, Germany, CAT. 339320). RT‐qPCR was used to determine the mRNA expression levels using the gene‐specific primers (Sigma‐Aldrich) with SYBR Green (Life Technologies, California, United States, CAT. A25742) on QuantStudio Real‐Time PCR system (Applied Biosystems, Massachusetts, United States). The relative mRNA expression levels were quantified using the 2^−ΔΔCt method. All RT‐qPCR experiments were performed in triplicate. All primers are listed in Table S1.

Western blot was performed following our previous study.28 SDS‐PAGE separated total cellular protein, then transferred to a Nitrocellulose Transfer membrane (.2 µm pore size, BioTrace, Arizona, United States, CAT. 27376–991). The membranes were blocked and then incubated with primary antibodies overnight at 4°C, and subsequently incubated with secondary antibodies for 1 h. α‐tubulin served as the internal standard, and Image J was used to quantify blots intensities. The antibodies used were as follows: anti‐NAMPT antibody (1:1000), anti‐KNYU antibody (1:1000, Abcam, ab225916), anti‐IDO1 antibody (1:1000, Abcam, ab55305), anti‐QPRT antibody (1:1000, Abcam, ab171944), anti‐3‐HAAO antibody (1:1000, Abcam, ab106436), anti‐NMNAT1 antibody (1:1000, Abcam, ab118270), anti‐α‐tubulin antibody (1:1000, Abcam, ab7291).

50 mg of each cartilage sample was grounded into powder in liquid nitrogen and homogenised using Omni TH Homogenizer (Omni, Inc, Georgia, United States). The supernatant was collected, and 1 mL of chloroform (Sigma‐Aldrich, CAT. C2432) was added, incubated on ice and centrifuged for 10 min at 13 000 × g at 4°C. The supernatant was then collected, freeze‐dried, and resuspended in 90 µL of 1% ACN in water for LC‐MS/MS analysis.

Kynurenine pathway metabolites were quantified by LC‐MS/MS. In brief, a targeted LC‐MS/MS metabolomics analysis for kynurenine pathway metabolites was performed using a Shimadzu UHPLC System coupled to a Shimadzu 8060 triple quadrupole (QqQ) mass spectrometer. The UHPLC (Nexera X2, Shimadzu Corp., Kyoto, Japan) consisted of LC‐30AD pump units, DGU‐20ASR degassing units, a SIL‐30AC autosampler, a CTO‐20AC column oven, a CBM‐20A communications BUS module and an FCV‐20AH2 diverter valve unit. LC was performed using a Phenomenex Gemini‐NX C18 column (150 × 2 mm, 3 µm, 110 A, PN: 00F‐4453‐B0) with a guard column (Phenomenex SecurityGuard Gemini‐NX C18, 4 × 2 mm, PN: AJ0‐8367). Solvents A and B were.1% v/v formic acid (Optima LC/MS Grade, Fisher Chemicals) in high purity water (Milli‐Q Reference Water Purification System, Merck) and in acetonitrile (Lichrosolv, PN: 1142914000, Merck), respectively. A flow rate of 300 µL/min was used to run the HPLC gradient, which was as follows: 2% B from 0–1 min, 2%–20% B from 1–22 min, 20%–98% B from 22–28 min, 98% B continued to 34 min, then 2% B at 35–40 min. Samples were kept at 4°C in the autosampler, and the column was operated at 50°C in a column oven.

The Shimadzu 8060 QqQ system had an electrospray ion source (ESI), which used nitrogen (> 99.999 vol % BOC, NSW, Australia) as drying gas, and argon (> 99.999 vol %, UN1006, Coregas Pty Ltd, NSW, Australia) as collision gas. Further instrument detail includes drying gas flow: 10 L/min, nebulising gas flow: 3.0 L/min, heating gas flow: 10 L/min, desolvation line: 250°C, heat block temperature: 400°C, CID gas: 270 kPa, and interface temperature: 300°C. Data were collected from 1–20 min, while the flow eluent was diverted to waste at 0–1 min and 22–40 min during column clean up and re‐equilibration.

Multiple reaction monitoring (MRM) transitions were optimised on either the negative or positive ionisation mode (m/z‐H or m/z+H). All analytical standards were purchased from Sigma‐Aldrich and stock solutions were prepared using high purity water. Details of the compound‐ and instrument‐specific parameters are shown in Table S2.

Cartilage explants were obtained from the femoral condyle head of fresh young bovine knee joints (3–5 months old) using a 3.0 mm biopsy punch (Kai Medical, Gifu, Japan, CAT. BP‐30F).31 After that, explants were cultured in a chondrogenic culture medium at 37°C for 72 h before further experiments. For NMN treatment, cartilage explants were washed with 1× PBS 3–4 times, and the culture medium was changed to chondrogenic culture medium with/without 1 mM NMN and 10 ng/mL IL‐1β for 48 h.

Glycosaminoglycan (GAG) was measured following the protocol of DMMB assay with purified shark chondroitin sulphate (Sigma‐Aldrich, CAT. C6737) as a standard.32 In brief, for GAG release, 30 µL of cartilage explants culture media was added to 96‐well plates with 300 µL of DMMB dye solution (Sigma‐Aldrich, CAT. 341088). All samples and standards were read at a 525 nm wavelength. Cartilage explants were dried and weighed. The percentage of GAG release was calculated as GAG release per tissue weight.

The PARP14 siRNAs and negative control vector were transfected into primary chondrocytes, C28/I2 chondrocytes cell line and 3D pellets for 24–72 h using the Lipofectamine RNAiMAX (ThermoFisher Scientific, CAT. 11668027) according to the recommended protocol by the manufacturer. The target sequences of siRNAs were as follows: siPARP14 (#1), 5′‐GGCTGAATGGTGCTTGGTACA‐3′; siPARP14 (#2), 5′‐ GCTGAATGGTGCTTGGTACAA‐3′; siPARP14 (#3), 5′‐ GCCTCATTTCATCGTTTATCT‐3′.

Chondrocytes were seeded at a density of 5 × 10³ cells/well in 96‐well plates. The next day, cells were transfected with siPARP14 or negative control vector for 24, 48, and 72 h. CCK‐8 assay (Abcam, CAT. ab228554) was performed according to the manufacturer's instructions. After each time point, 10 µL of WST‐8 solution was added to each well and incubated for 1 h at 37°C in the dark. Absorbance was measured at 460 nm.

2 × 10⁶ of primary P0, G1, and G4 chondrocytes were seeded on each T75 culture flask with a complete medium. On the next day, cells were transfected using siPARP14 or negative control vector with/without 10 ng/mL IL‐1β for 48 h. 1 mL of culture medium was collected in each flask after that. The quantification of glucose and lactate was performed using a previously published method with minor modifications. Glucose, pyruvate, and lactate were quantified by ion‐exclusion chromatography using an Agilent 1200 high‐performance liquid chromatography (HPLC) system with UV and RI detectors (Agilent Technologies) and an Agilent Hi‐Plex H column (300 × 7.7 mm; Agilent Technologies) with a guard column (SecurityGuard Carbo‐H, 4 × 3 mm, Phenomenex, New South Wales, Australia).33 Chromatograms were integrated using the Chromeleon software (version 7.3, Thermo Scientific Dionex).

Chondrocytes were seeded at a density of 5 × 10³ cells/well in Seahorse XF96 Cell Culture Microplates (Agilent Technologies, CAT. 103794‐100). The next day, cells were transfected with siPARP14 or negative control vector with/without 10 ng/mL IL‐1β for 48 h. After that, the Glycolysis Stress test (Agilent Technologies, CAT. 103020–100) and Mito Stress test (Agilent Technologies, CAT. 103015–100) were performed according to the manufacturer's instructions.28 For the Glycolysis Stress test, cells were sequentially treated with 10 mM glucose, 1 µM oligomycin, and 50 mM 2‐deoxyglucose (2DG). For the Mito Stress test, cells were sequentially treated with 1.5 µM oligomycin, 1 µM carbonyl cyanide‐4 (trifluoromethoxy) phenylhydrazone (FCCP), and a combination of.5 µM rotenone/antimycin A. The following methodology was used to collect each set of measurements: mix for 3 min, measure for 3 min, repeat cycle 3–4 times. All cells were plated in triplicates. Each assay was performed twice. Prior to analysis, zeros were removed following Agilent's guidelines. We averaged the remaining replicates. Calculations were made using the average of all measurements per injection and normalised to total protein content using BCA protein assay (Pierce, Thermo Fisher Scientific, CAT. 23227). Data were analysed using Wave Desktop 2.6 (Agilent Technologies).

Chondrocyte pellets and C28/I2 cells were fixed in 4% PFA for 30 min. After that, pellets were embedded in paraffin, sectioned into 5 µm of slides and dewaxed. Then, sections were incubated with proteinase K and 3% H₂O₂ for 20 min, respectively. C28/I2 cells were covered with 1× PBS containing.25% Triton X‐100 for 15 min. Next, both pellets sections and cells were sealed with 1% BSA for 1 h. After that, sections and cells were incubated with primary antibodies overnight at 4°C and secondary antibodies (Goat Anti‐Rabbit, Alexa Fluor 488, Abcam, ab150077) for 1 h at room temperature. Nuclei were restained with DAPI (Thermo Fisher Scientific, CAT. D3571) for 5 min. Images were captured using a confocal laser scanning microscope (Nikon A1R Confocal, Amsterdam, The Netherlands). Fluorescence intensity was measured using Image J. The antibodies used were as follows: anti‐Aggrecan antibody (1:200, Abcam, ab186414), anti‐MMP1 antibody (1:200, Abcam, ab52631), anti‐MMP13 antibody (1:200).

All experiments were performed in triplicate using different human samples collected from 5 different donors (5 independent experiments) except 6 donors for NAD⁺/tNAD measurement and RT‐qPCR, 10 donors for RNA sequencing, and 3 donors for siRNAs‐related experiments. All animal experiments were performed using 6 animals per group except 3 samples per group for bovine explants experiments. The statistical analysis was performed using GraphPad Prism (version 8, California, United States), graphically testing for normal distribution, and using either t‐test or one‐way ANOVA with Tukey's post hoc analysis. The data are presented as mean ± standard deviation (SD) for all variables. For all figures, differences were considered significant with p < .05.