ZDHHC11-mediated palmitoylation alleviates chondrocyte senescence and serves as a therapeutic target for osteoarthritis

Abstract

Osteoarthritis (OA) is a whole-joint disorder that interferes with the quality of life in older individuals. Here we report that ZDHHC11 is highly expressed in articular chondrocytes but is downregulated in the degenerated cartilage of aged mice and patients with OA. ZDHHC11 prevents chondrocyte senescence and promotes cartilage anabolism, culminating in an improved OA phenotype. The deletion of Zdhhc11 in mice (Zdhhc11^fl/fl) exacerbates OA progression in a destabilized medial meniscus model. Specifically, we identify ZDHHC11 as a key palmitoyltransferase whose depletion leads to a GNB2-dependent E3 ubiquitin ligase-mediated proteasomal degradation of APOD. Mechanistically, ZDHHC11-mediated palmitoylation alleviates OA progression by deactivating the GATA4–P65 signaling pathway. We also propose an original lipid nanoparticle-based platform for Zdhhc11 mRNA delivery to rejuvenate impaired cartilage by specifically targeting chondrocytes in vivo. Collectively, ZDHHC11-dependent palmitoylation is essential for ameliorating OA, and the targeted delivery of ZDHHC11 may serve as a promising strategy for future OA treatment.

Wang, He, Gong and colleagues identify an age-related decline in the palmitoyltransferase ZDHHC11 in chondrocytes that leads to senescence and the pathogenesis of osteoarthritis, highlighting the potential of targeted ZDHHC11 delivery as a therapeutic strategy for osteoarthritis.

Main

Osteoarthritis (OA) is a pervasive, degenerative whole-joint disorder that affects the aging population. Approximately 595 million people worldwide have OA, which interferes with the quality of life and incurs high societal costs^1,2. OA is characterized primarily by cartilage destruction, osteophyte formation and synovial inflammation, which can subsequently result in pain and deformities^3,4. Currently, the pharmacotherapeutic use of nonsteroidal anti-inflammatory drugs, corticosteroid treatment and joint replacement surgery are recommended; however, there are no approved disease-modifying drugs for the treatment of OA^5,6.

While certain risk factors for OA have been extensively reviewed, the precise molecular mechanisms underlying OA pathogenesis remain poorly understood^7,8. Although mechanical overload and joint overuse contribute to the development of OA, a plethora of studies have shown that a chronic state of low-grade inflammation, aging and microenvironmental alterations have crucial roles in disease progression^9,10. Proinflammatory cytokines are involved in the senescence-associated secretory phenotype (SASP), contributing to cellular senescence during age-related OA¹¹. The accumulation of senescent cells within joint tissues can lead to mitochondrial dysfunction, excessive oxidative stress and elevated levels of pathophysiological mediators, including reactive oxygen species, all of which are believed to trigger damaging metabolic disturbances in chondrocytes^12,13.

Post-translational modifications have been implicated in multiple diseases^14–18. Recent studies have revealed the post‐translational addition of lipids to proteins, termed ‘protein lipidation’, which includes N‐myristoylation, S‐farnesylation and S‐palmitoylation^19,20. S-palmitoylation, also referred to as S-acylation, is a reversible post-translational modification that occurs in all eukaryotes²¹. Given that S-palmitoylation is a dynamic modification process, this molecular switch can determine the functioning of proteins, thus governing a broad spectrum of biological processes^22,23. The enzymatic regulation of S-palmitoylation is mediated by the palmitoyl acyltransferases (PATs), which contain a conserved aspartate–histidine–histidine–cysteine (DHHC) domain that is essential for their catalytic activity²⁴. There are 23 PATs in humans (ZDHHC1–9, ZDHHC11–24), while there are 24 PATs in mice (ZDHHC1–9, ZDHHC11–25)²⁵. Accumulating evidence shows that the aberrant regulation and functional sequelae of palmitoylation are involved in the development of various pathological conditions^26–28. For decades, the extent of protein S-palmitoylation has been undervalued, and the regulatory effects of those proteins are now beginning to surface. The limited number of studies has prompted us to investigate the role of S-palmitoylation in the pathogenesis of OA.

In this study, we identify a chondroprotective role for ZDHHC11 in mitigating cellular senescence and modulating extracellular matrix (ECM) metabolism. Mechanistically, ZDHHC11-mediated palmitoylation attenuates OA progression by competitively inhibiting ubiquitination and regulating the GATA4–P65 signaling pathway. Notably, we introduce an original engineered Zdhhc11 mRNA lipid nanoparticle (mRNA-LNP)–chondrocyte affinity peptide (CAP) complex that specifically targets chondrocytes and exerts functional effects in retarding OA in preclinical models. Together, these findings suggest that targeting ZDHHC11 may restore articular homeostasis in OA.

Results

ZDHHC11 expression is downregulated in senescent chondrocytes and degenerative OA tissues

The identification of distinct palmitoyltransferases (23 in humans and 24 in mice) has revealed the tissue-specific enrichment of certain enzymes²⁹. However, the expression profile of palmitoyltransferases in chondrocytes has yet to be fully characterized. To explore this, we conducted a series of experiments to reveal the effects of palmitoylation on chondrocytes. Specifically, we isolated primary human chondrocytes (HCs) and treated them with interleukin-1β (IL-1β), a proinflammatory cytokine known to induce an inflammatory environment. As shown in Fig. 1a and Extended Data Fig. 1a, we observed a marked reduction in palmitoylation levels within chondrocytes in response to IL-1β stimulation. To further investigate the expression profile of palmitoyltransferases in HCs and mouse chondrocytes (MCs), we performed Southern blotting and identified five common palmitoyltransferases with relatively high abundance (ZDHHC5, ZDHHC11, ZDHHC13, ZDHHC20 and ZDHHC21) in both HCs and MCs (Extended Data Fig. 1b). Based on these findings, we hypothesized that one or more of these enzymes might regulate chondrocyte function. To test this hypothesis, we silenced their expression using specific siRNAs and assessed the effects on catabolic, anabolic and cellular senescence-related activities (Extended Data Fig. 1c,d). Among the five ZDHHCs examined, only the knockdown of ZDHHC11 demonstrated consistent and notable effects on the expression of genes associated with catabolism, anabolism and cellular senescence at both the mRNA and protein levels. These findings suggest a protective effect of ZDHHC11 against chondrocyte senescence and catabolism.

Fig. 1. ZDHHC11 expression is downregulated in aged patients with OA and in senescent HCs.

a, Flowchart for screening potential PATs that mediate palmitoylation in chondrocytes. b, Heatmap visualization of known ZDHHC genes in HCs treated with or without 10 ng ml⁻¹ IL-1β for 48 h. n = 3 independent experiments per group. c, Representative images of Safranin O/Fast Green staining and immunofluorescence of MMP13, P16, COL2A1 and ZDHHC11 comparing the younger group (40–65 years old) to the older group (older than 70 years). Scale bars, 100 μm. d, SA-β-Gal staining in P0- and P4-generation HCs. Scale bars, 50 μm. e, Western blotting of ZDHHC11 and P53 expression in P0- and P4-generation HCs. n = 3. f, Representative images of the immunofluorescence of ZDHHC11 in HCs with or without H₂O₂ stimulation for 48 h. Scale bars, 100 μm. g, Pearson correlation analysis between ZDHHC11 expression and K–L grades of OA. n = 5. Error bands are presented as a two-tailed 95% CI. In c–f, experiments were repeated at least three times independently, with similar results. Two-sided Student’s t-test was used for b. NS, not significant (P > 0.05), *P < 0.05, **P < 0.01, ***P < 0.001. The exact P values are reported as source data.

Source data

Extended Data Fig. 1. Identification and characterization of ZDHHC11 in human chondrocytes.

(a) Detection of total palmitoylated proteins in HCs treated with or without 10 ng/mL IL-1β for 48 h by the ABE assay. HAM: hydroxylamine. n = 3. (b) Southern blotting for detecting known ZDHHCs in human chondrocytes (HCs) and mouse chondrocytes (MCs). (c) RT- qPCR of P21, SOX9 and ADAMTS4 levels in HCs transfected with the negative control or ZDHHCs siRNAs. (n = 6 independent experiments per group, mean ± SD). (d) Western blotting of P21, SOX9 and ADAMTS4 expression in HCs transfected with the negative control or ZDHHCs siRNAs. n = 3. (e) Heatmap visualization of known Zdhhcs in MCs treated with or without 10 ng/mL IL-1β for 48 h. n = 3 independent experiments per group. (f) Quantification of COL2A1 staining intensity shown as mean ± SD. n = 9 independent experiments per group. (g) Quantification of MMP13 staining intensity shown as mean ± SD. n = 9 independent experiments per group. (h) Quantification of P16 staining intensity shown as mean ± SD. n = 9 independent experiments per group. (i) Quantitative data of ZDHHC11 positive cells shown as mean ± SD. n = 9 independent experiments per group. (j) Quantitative data of SA-β-Gal positive cells shown as mean ± SD. n = 6 independent experiments per group. (k) Quantification of ZDHHC11 staining intensity shown as mean ± SD. n = 9 independent experiments per group. (l) Representative images of SA-β-Gal staining in HCs with or without H₂O₂ stimulation. Scale bars, 50 μm. Quantitative data shown as mean ± SD. n = 6. (m) Quantitative data of SA-β-Gal positive cells shown as mean ± SD. n = 6 independent experiments per group. (n) Representative images of immunofluorescence of ZDHHC11 in patients with different K-L grades. n = 5. Scale bars, 100 μm. In a, d, l and n, experiments were repeated at least three times independently with similar results. One-way ANOVA with Tukey’s correction for multiple comparisons is used for (c, f, g, h and i) and two-sided Student’s T-test for (e, j, k and m). ns (p > 0.05), *p < 0.05, **p < 0.01, ***p < 0.001. The exact p-values are reported in the Source data file for Extended Data Fig. 1.

Source data

Given the observed downregulation of palmitoylation levels in chondrocytes under inflammatory conditions, we further investigated whether the expression of ZDHHC11 is consistent with this change. First, we treated both HCs and MCs with IL-1β in vitro. As expected, we observed a reduction in ZDHHC11 expression at the mRNA level (Fig. 1b and Extended Data Fig. 1e). Next, we used OA samples to uncover the expression of ZDHHC11 in the inflammatory environment in vivo. As shown in Fig. 1c, Safranin O/Fast Green staining indicated severe degradation in the medial tibial plateau cartilage, particularly in older individuals with OA. Senescent samples also exhibited elevated expression of matrix metalloproteinase 13 (MMP13) and P16, alongside a reduction in collagen II (COL2A1) and ZDHHC11 expression, as illustrated by immunofluorescence staining (Fig. 1c and Extended Data Fig. 1f–i). To explore the expression of ZDHHC11 in the senescent condition, HCs were then passaged and analyzed after senescence-associated β-galactosidase (SA-β-Gal) staining (Fig. 1d and Extended Data Fig. 1j). Moreover, western blotting results confirmed a reduced abundance of ZDHHC11 in P4-generation HCs, indicating that the protein levels of ZDHHC11 decrease as chondrocytes undergo senescence (Fig. 1e). This reduction was further confirmed with H₂O₂ treatment, which induces oxidative stress and accelerates senescence. The quantification of SA-β-Gal-positive areas and immunofluorescence staining of ZDHHC11 in H₂O₂-treated HCs validated the above results (Fig. 1f and Extended Data Fig. 1k–m). Moreover, the fluorescence intensity of ZDHHC11 exhibited an inverse correlation with the Kellgren–Lawrence (K–L) grades, suggesting that ZDHHC11 could serve as an indicator of disease severity in OA (Fig. 1g and Extended Data Fig. 1n).

ZDHHC11 attenuates senescence-related OA pathological phenotypes

Following the preliminary identification of the relationship between ZDHHC11 expression and both inflammation and senescence, we further investigated the effect of ZDHHC11 on senescence-related OA through knockdown and overexpression experiments. ZDHHC11 knockdown in HCs resulted in a reduction of COL2A1 and SOX9 expression at both the mRNA and protein levels, accompanied by increased expression of ADAMTS4, MMP3, P53, P21 and P16 (Extended Data Fig. 2a,b). Additionally, ZDHHC11 knockdown led to decreased glycosaminoglycan deposition and enhanced senescence-associated staining (Extended Data Fig. 2c–f), whereas its overexpression had the opposite effect (Extended Data Fig. 2g–l).

Extended Data Fig. 2. ZDHHC11 modulates ECM metabolism and senescent phenotype in HCs.

(a) RT- qPCR of P53, P21, MMP3, IL-6, COL2A1, SOX9, ADAMTS4 and ZDHHC11 levels in HCs transfected with the negative control or ZDHHC11 siRNAs. (n = 9 independent experiments per group, mean ± SD). (b) Western blotting of COL2A1, SOX9, ADAMTS4, MMP3, P53, P21, P16 and ZDHHC11 expression in HCs transfected with the negative control or ZDHHC11 siRNAs. n = 3. (c) Quantification of relative Alcian blue intensity in HCs transfected with the negative control or ZDHHC11 siRNAs shown as mean ± SD. n = 3 independent experiments per group. (d) Representative images of Alcian blue staining in HCs transfected with the negative control or ZDHHC11 siRNAs. (e) Representative images of SA-β-Gal staining in HCs transfected with the negative control or ZDHHC11 siRNAs. Scale bars, 50 μm. (f) Quantitative data of SA-β-Gal positive cells transfected with the negative control or ZDHHC11 siRNAs shown as mean ± SD. n = 6 independent experiments per group. (g) Quantification of relative Alcian blue intensity in HCs infected with the ZDHHC11 overexpression lentivirus or control lentivirus shown as mean ± SD. n = 3 independent experiments per group. (h) Representative images of Alcian blue staining in HCs infected with the ZDHHC11 overexpression lentivirus or control lentivirus. (i) Representative images of SA-β-Gal staining in HCs infected with the ZDHHC11 overexpression lentivirus or control lentivirus. Scale bars, 50 μm. (j) Quantitative data of SA-β-Gal positive cells infected with the ZDHHC11 overexpression lentivirus or control lentivirus shown as mean ± SD. n = 6 independent experiments per group. (k) Western blotting of COL2A1, SOX9, ADAMTS4, MMP3, P53, P21, P16 and ZDHHC11 expression in HCs infected with the ZDHHC11 overexpression lentivirus or control lentivirus. n = 3. (l) RT- qPCR of P53, P21, MMP3, IL-6, COL2A1, SOX9, ADAMTS4 and ZDHHC11 levels in HCs infected with the ZDHHC11 overexpression lentivirus or control lentivirus. (n = 9 independent experiments per group, mean ± SD). In b, d, e, h, i and k, experiments were repeated at least three times independently with similar results. Two-sided Student’s T-test is used for (e, g and j) and one-way ANOVA with Tukey’s correction for multiple comparisons for (a, c and f). ***p < 0.001. The exact p-values are reported in the Source data file for Extended Data Fig. 2.

Source data

Based on the in vitro results, and given that ZDHHC11 has a distinct chondroprotective role in cellular senescence and ECM remodeling, we aimed to clarify whether these effects could be replicated in vivo. Thus, we performed a destabilized medial meniscus (DMM) operation on aged (24-month-old) and control (3-month-old) C57BL/6J mice, followed by the intra-articular injection of an adeno-associated virus (AAV) vector that carries Zdhhc11 (Zdhhc11-AAV) after 14 days. All the animals were killed 8 weeks after the injury (Fig. 2a and Extended Data Fig. 3a). In 24-month-old mice, we confirmed that increased P16 expression was accompanied by reduced ZDHHC11 expression (Extended Data Fig. 3b,e,f). To assess the microstructure of the subchondral bone, we scanned the samples using micro-computed tomography (micro-CT). Overall, in contrast to the mice in the sham group, the DMM-operated mice exhibited typical OA manifestations, especially in the aged group. Notably, compared with the DMM-operated group, the group receiving ZDHHC11 supplementation exhibited a reduced number of osteophytes (Fig. 2b,c). Hematoxylin and eosin (H&E) staining also revealed that the DMM + Zdhhc11-AAV group presented a lower synovium score, suggesting that ZDHHC11 alleviated the age-related OA severity in the mouse model of post-traumatic OA development (Fig. 2d,e). To rule out the possibility that ZDHHC11 directly contributes to synovial inflammation, we performed immunofluorescence staining. Here, tissue sections were stained with vimentin, a marker of fibroblast-like synoviocytes (FLS), and ZDHHC11 (Extended Data Fig. 3c). The results showed that the relative immunofluorescence intensities of ZDHHC11 and vimentin remained unchanged following DMM surgery (Extended Data Fig. 3g). We also overexpressed ZDHHC11 in FLS and measured the protein levels of proinflammatory mediators, including inducible nitric oxide synthase (iNOS) and cyclooxygenase-2 (COX-2) (Extended Data Fig. 3d). Western blot analysis revealed that ZDHHC11 did not induce inflammation in FLS, as the expression of iNOS and COX-2 was similar between the two groups (Extended Data Fig. 3d). Thus, we can conclude that the effect of ZDHHC11 on the synovium was more a consequence of OA rather than an independent change. Safranin O/Fast Green staining also revealed notable cartilage destruction in the DMM group, an effect that was attenuated after Zdhhc11-AAV injection. Moreover, we calculated the OA Research Society International (OARSI) histological grade and found lower OARSI scores in Zdhhc11-AAV-injected mice with DMM-induced OA at 24 months of age (Fig. 2f,g). Next, immunohistochemical (IHC) staining for the ECM-related proteins COL2A1 and MMP3, the SASP marker IL-6, and the aging-associated proteins P21 and P16 was performed in each group. ZDHHC11 mitigated the pathological features of OA model mice, as illustrated by increased COL2A1 abundance and decreased expression of MMP3, IL-6, P21 and P16 following DMM surgical induction (Fig. 2f and Extended Data Fig. 3h). Moreover, Zdhhc11-AAV-injected 24-month-old DMM mice exhibited decreased mechanical hypersensitivity, as indicated by a greater withdrawal threshold in the hind paw (Fig. 2h). In summary, the results of our preclinical animal studies support the therapeutic use of ZDHHC11 for the treatment of senescence-related OA.

Fig. 2. ZDHHC11 attenuates age-related OA pathological phenotypes in vivo.

a, Schematic depicting the DMM operation and Zdhhc11 intra-articular injection in 3- and 24-month-old male C57BL/6J mice (3-month-old mice: sham group (n = 6), DMM group (n = 6), DMM + Zdhhc11-AAV group (n = 6); 24-month-old mice: sham group (n = 6), DMM group (n = 6), DMM + Zdhhc11-AAV group (n = 6)). w, weeks. b, Micro-CT images of knee joint osteophytes from 3- and 24-month-old male C57BL/6J mice. Scale bars, 2.5 mm. c, Quantification of osteophyte numbers in each group (n = 6 animals per group). d, Representative images of H&E staining of the synovium in each group. Scale bars, 100 μm. e, Quantification of synovial hyperplasia according to the synovitis score in each group (n = 6 animals per group). f, Representative images of Safranin O/Fast Green staining and IHC staining for COL2A1, MMP3, IL-6 and P21 in each group. Scale bars, 50 μm. g, Quantification of cartilage damage according to the OARSI scoring system in each group (n = 6 animals per group). h, Results of the Von Frey assay conducted at the indicated times after DMM surgery in each group (n = 6 animals per group). One-way ANOVA with Tukey’s correction for multiple comparisons was used for c and h (mean ± s.d.). Two-sided Mann–Whitney U test was used for e and g (mean ± 95% CI). **P < 0.01, ***P < 0.001. The exact P values are reported as source data. Panel a created with BioRender.com.

Source data

Extended Data Fig. 3. Zdhhc11 ameliorates synovial inflammation in naturally aged mice.

(a) Representative images of immunofluorescence of GFP-Zdhhc11 from different groups. 3-month-old mice: sham group (n = 6), DMM group (n = 6), DMM + Zdhhc11 AAV group (n = 6); 24-month-old mice: sham group (n = 6), DMM group (n = 6), DMM + Zdhhc11 AAV group (n = 6). Scale bars, 50 μm. (b) Representative images of immunofluorescence of P16 and Zdhhc11 from 3-month-old and 24-month-old male C57BL/6 J mice. Scale bars, 50 μm. (c) Representative images of immunofluorescence of Zdhhc11 and Vimentin on sham and DMM-induced mouse models. (d) Westerning blotting of iNOS and COX2 expression in fibroblast-like synoviocytes (FLS) infected with the control lentivirus or ZDHHC11 overexpression lentivirus. n = 3. (e) Quantification of P16 staining intensity from 3-month-old and 24-month-old male C57BL/6 J mice shown as mean ± SD. n = 6 independent experiments per group. (f) Quantification of Zdhhc11 staining intensity from 3-month-old and 24-month-old male C57BL/6 J mice shown as mean ± SD. n = 6 independent experiments per group. (g) Quantification of relative immunofluorescence intensity of Zdhhc11/Vimentin on sham and DMM-induced mouse models shown as mean ± SD. n = 6 independent experiments per group. (h) Representative images of IHC of P16 in each group. Scale bars, 50 μm. In a, b, c, d and h, experiments were repeated three times independently with similar results. Two-sided Student’s T-test is used for (e, f and g). ns (p > 0.05), **p < 0.01, ***p < 0.001. The exact p-values are reported in the Source data file for Extended Data Fig. 3.

Source data

ZDHHC11 attenuates the aggressive OA phenotype in Zdhhc11 cKO-DMM mice

On the basis of the above experimental studies, we conducted rescue experiments to investigate the effects of ZDHHC11 on OA and senescent phenotypes in conditional knockout (cKO) mice. As schematically shown in Fig. 3a and Extended Data Fig. 4a–d, Zdhhc11^fl/fl; Aggrecan^CreERT2 (cKO) mice were successfully generated and subjected to DMM surgery at week 12, followed by the intra-articular injection of Zdhhc11-AAV. All animals were killed 8 weeks after the DMM operation. Strikingly, ZDHHC11 supplementation reversed the severity of OA, as demonstrated by reduced osteophyte formation in cKO-DMM mice (Fig. 3b,e). H&E staining indicated that synovial inflammation in cKO-DMM mice was attenuated by Zdhhc11-AAV injection (Fig. 3c,f). Safranin O/Fast Green staining also revealed that ZDHHC11 protected mice from cKO-induced articular cartilage degeneration. In cKO-DMM mice, representative images illustrated the upregulation of the senescence-associated markers IL-6, MMP3, P21 and P16, along with the concurrent downregulation of the cartilage anabolism marker COL2A1. However, the overexpression of ZDHHC11 prominently reduced cellular senescence and OA phenotypes (Fig. 3d and Extended Data Fig. 4e). The OARSI grade indicated that ZDHHC11 mitigated OA manifestations in vivo (Fig. 3g). Moreover, ZDHHC11 reduced the DMM-induced OA pain, which was manifested as a facilitatory effect on mechanical withdrawal thresholds (Fig. 3h). In summary, we revealed that ZDHHC11 overexpression mitigated disease severity and reversed the senescent phenotypes exacerbated by Zdhhc11 knockout in a DMM mouse model.

Fig. 3. ZDHHC11 mitigates the aggressive OA phenotype in Zdhhc11 cKO-DMM mice.

a, Schematic depicting the DMM operation and AAV intra-articular injection in Zdhhc11 cKO male C57BL/6J mice (WT-sham group, n = 6; cKO-sham group, n = 6; cKO-DMM group, n = 6; cKO-DMM + Zdhhc11-AAV group, n = 6). TAM, tamoxifen. b, Micro-CT images of knee joint osteophytes in each group. Scale bars, 2.5 mm. c, Representative images of H&E staining of the synovium in each group. Scale bars, 100 μm. d, Representative images of Safranin O/Fast Green staining and IHC staining for COL2A1, MMP3, IL-6 and P21 in each group. Scale bars, 50 μm. e, Quantification of osteophyte numbers in each group (n = 6 animals per group). f, Quantification of synovial hyperplasia according to the synovitis score in each group (n = 6 animals per group). g, Quantification of cartilage damage according to the OARSI scoring system in each group (n = 6 animals per group). h, Results of the Von Frey assay conducted at the indicated times after DMM surgery in each group (n = 6 animals per group). One-way ANOVA with Tukey’s correction for multiple comparisons was used for e and h (mean ± s.d.). Two-sided Mann–Whitney U test was used for f and g (mean ± 95% CI). **P < 0.01, ***P < 0.001. The exact P values are reported as source data. Panel a created with BioRender.com.

Source data

Extended Data Fig. 4. Knockout of Zdhhc11 augments osteoarthritis progression in a murine model.

(a) Schematic illustration of gRNA design in Zdhhc11 knockout mice. (b) Southern blotting for detecting Zdhhc11 knockout efficiency. (c) Representative images of immunofluorescence of Zdhhc11 from wildtype and Zdhhc11 cKO C57BL/6 J mice. Scale bars, 50 μm. (d) Representative images of immunofluorescence of GFP-Zdhhc11 from different groups: WT-sham group (n = 6); cKO-sham group (n = 6), cKO-DMM (n = 6); cKO-DMM + Zdhhc11 AAV group (n = 6). Scale bars, 50 μm. (e) Representative images of IHC of P16 in each group. Scale bars, 50 μm. In c–e, experiments were repeated three times independently with similar results.

ZDHHC11 functions as a palmitoyltransferase responsible for APOD palmitoylation

Based on the findings from our in vitro and in vivo experiments, we sought to further investigate the underlying mechanisms through which ZDHHC11 mediates antisenescence effects. To this end, we performed immunoprecipitation combined with mass spectrometry (IP–MS) in ZDHHC11-overexpressing chondrocytes and identified 4,596 coprecipitated proteins (Fig. 4a). In addition, we simulated the inflammatory environment of OA by treating chondrocytes with IL-1β, resulting in the identification of 127 differentially expressed proteins (|log₂(fold change (FC))| > 2). To further explore proteins associated with OA, cell senescence and palmitoylation, we selected 721 proteins using GeneCards (https://www.genecards.org/). Notably, three proteins—apolipoprotein D (APOD), CYBA and DDX39B—were identified as potential targets modulated by ZDHHC11 through a Venn diagram analysis of overlapping molecules (Fig. 4b). Considering that ZDHHC11 is a palmitoyltransferase, we hypothesized that its interacting proteins might undergo the palmitoylation process. Palmitoylation of candidate proteins was subsequently predicted using the motif-based predictor CSS-Palm 4.0 software (https://gpspalm.biocuckoo.cn/), with APOD (C185) achieving the highest score (Fig. 4c). The ectopic coexpression of APOD and ZDHHC11 in HEK293T cells confirmed the interaction of these two proteins through a co-IP assay (Fig. 4d). Molecular docking analysis further demonstrated that the ZDHHC11 domains interact with the C185 site of APOD (Fig. 4e). This indicates the molecular interface between ZDHHC11 and APOD, supporting the hypothesis that this interaction is essential for the regulatory effects of ZDHHC11. Next, we validated this interaction by assessing the palmitoylation of APOD using acyl-biotin exchange (ABE) assays. The results confirmed that APOD undergoes palmitoylation, which was markedly suppressed in the presence of 2-bromopalmitate (2-BP) (Fig. 4f).

Fig. 4. ZDHHC11 acts as a palmitoyltransferase for APOD palmitoylation.

a, Schematic illustration of the co-IP assay for MS in C28/I2 cells infected with the ZDHHC11 overexpression (OE) lentivirus. b, Venn diagram analysis identified the following: proteins coprecipitated with ZDHHC11 (n = 4,596); differentially expressed proteins in IL-1β-stimulated chondrocytes, detected using MS (n = 127); proteins associated with OA, cell senescence and palmitoylation in GeneCards (n = 721); and three overlapping proteins (APOD, CYBX and DDX39B). c, Palmitoylation sites of APOD, CYBA and DDX39B predicted by CSS-Palm 4.0 software. d, The interaction between ZDHHC11 and APOD was confirmed by a co-IP assay. n = 3. e, Molecular docking predicted the interaction between ZDHHC11 and APOD. f, ABE assay to detect palmitoylated APOD in HCs treated with or without 50 μM 2-BP for 24 h. n = 3. HAM, hydroxylamine. g, t-SNE projection displaying nine cell subpopulations from normal (n = 3) and OA (n = 4) human single-cell RNA-seq data. Cells are colored by clusters. EC, effector chondrocytes; FC, fibrocartilage chondrocytes; HTC, hypertrophic chondrocytes; HomoC_RegC, homeostatic chondrocytes/regulatory chondrocytes; PreHTC, prehypertrophic chondrocytes; ProC, proliferative chondrocytes; RepC, reparative chondrocytes; SPP1+C, SPP1-positive chondrocytes. h, Violin plot showing the senescence score across distinct cell subpopulations (arranged in ascending order). i, Dot plot showing the correlation between the senescence score and normalized APOD expression within each subpopulation. Two-sided Pearson correlation analysis without multiple comparisons was used for i. P = 2.2 × 10⁻¹⁶. In d and f, experiments were repeated at least three times independently, with similar results. Panel a created with BioRender.com.

Source data

Additionally, we conducted a series of experiments to validate the role of APOD in regulating cellular senescence. Western blot analysis revealed reduced APOD abundance in P4-generation HCs (Extended Data Fig. 5a), and immunofluorescence staining indicated significantly decreased APOD expression in both Zdhhc11 cKO mice and 24-month-old mice (Extended Data Fig. 5b–e). These findings suggest that reduced APOD expression could be a hallmark of senescent chondrocytes, potentially linking ZDHHC11 modulation to cellular aging. Further, we annotated chondrocytes into nine distinct subtypes based on canonical gene markers to analyze the articular cartilage at a single-cell level³⁰ (Extended Data Fig. 5f). As shown in Fig. 4g, the t-distributed stochastic neighbor embedding (t-SNE) projection of the atlas indicated the relative chondrocyte composition in patients with OA and healthy controls. Specifically, the senescence-associated gene score (SenMayo) was applied to assess the senescence status within each subpopulation³¹. The different subpopulations were arranged in ascending order according to their scores (Fig. 4h). Notably, the expression of APOD exhibited a negative correlation with the SenMayo score across various subpopulations (Fig. 4i). Moreover, downregulation of APOD was detected in patients with OA (Extended Data Fig. 5g). In summary, we identified limited expression of APOD in association with aging-related OA, and its antiaging function may be governed by ZDHHC11.

Extended Data Fig. 5. Identification and characterization of APOD in chondrocytes.

(a) Western blotting of APOD and P53 expression in P0 and P4 generation HCs. n = 3. (b) Representative images of immunofluorescence of Apod from wildtype and Zdhhc11 cKO C57BL/6 J mice. Scale bars, 50 μm. (c) Quantification of Apod staining intensity in each group shown as mean ± SD. n = 6 independent experiments per group. (d) Representative images of immunofluorescence of Apod from 3-month-old and 24-month-old male C57BL/6 J mice. Scale bars, 50 μm. (e) Quantification of Apod staining intensity in each group shown as mean ± SD. n = 6 independent experiments per group. (f) Dot plot showing representative marker genes across cell clusters. Dot size is proportional to the fraction of cells expressing specific genes. Color intensity corresponds to the relative expression of specific genes. (g) t-SNE presenting the APOD expression among subpopulations in normal or OA chondrocytes. In a, b and d, experiments were repeated three times independently with similar results. Two-sided Student’s T-test is used for (c and e). ***p < 0.001. The exact p-values are reported in the Source data file for Extended Data Fig. 5.

Source data

APOD palmitoylation at the C185 site inhibits ECM catabolism and chondrocyte senescence

To further explore the biological significance of APOD palmitoylation, we examined the specific site (C185) involved in this modification (Fig. 4c,d). Consistent with the highly conserved C185 site in APOD across species (Fig. 5a), wild-type (WT) APOD, rather than the C185A mutant, underwent palmitoylation (Fig. 5b). ZDHHC11 knockdown prominently reduced the APOD palmitoylation status (Fig. 5c); however, this could be reversed by ZDHHC11 overexpression (Fig. 5d). Moreover, PAT assay revealed that ZDHHC11 could catalyze the palmitoylation of WT APOD but not that of the C185A mutant (Fig. 5e). These results underline the importance of the C185 site in ZDHHC11-catalyzed APOD palmitoylation. To investigate the role of APOD on senescence-mediated OA progression, we transfected APOD-specific siRNAs (si-APOD) into primary HCs. Reverse transcription followed by qPCR (RT–qPCR) showed that COL2A1 and SOX9 expression was reduced (Extended Data Fig. 6a). In contrast, APOD knockdown induced notable production of ADAMTS4, P21 and P53, as well as the SASP markers MMP3 and IL6 at the mRNA level (Extended Data Fig. 6a). Consistent with these findings, western blotting confirmed a decrease in COL2A1 and SOX9 protein expression, accompanied by upregulated expression of ADAMTS4, P53, P21, MMP3 and P16 (Extended Data Fig. 6b). Furthermore, APOD inhibition diminished Alcian blue staining, indicative of decreased glycosaminoglycan content (Extended Data Fig. 6c,d). APOD knockdown also significantly induced cellular senescence, as evidenced by an increase in SA-β-Gal-positive areas in HCs (Extended Data Fig. 6e,f). Notably, APOD overexpression in chondrocytes promoted COL2A1 and SOX9 expression while reducing ADAMTS4, MMP3, P53 and P21 expression at both the mRNA and protein levels. In contrast, overexpression of the mutant form of APOD (APOD(C185A)) intensified OA and aging phenotypes (Fig. 5f,g). Additionally, SA-β-Gal and Alcian blue staining confirmed that APOD overexpression enhanced glycosaminoglycan accumulation and reduced cellular senescence, whereas APOD(C185A) overexpression produced opposing effects (Fig. 5h,i and Extended Data Fig. 6g,h). Taking these results together, we propose that the C185 site of APOD is essential for ZDHHC11-mediated S-palmitoylation, which may mitigate OA progression and cellular senescence phenotypes.

Fig. 5. APOD palmitoylation at the C185 site inhibits ECM catabolism and chondrocyte senescence.

a, Schematic illustration of the conservation of the C185 site on APOD across different species. b, Palmitoylation levels of APOD and the C185A mutant in HCs, detected using the ABE assay. n = 3. c, Palmitoylation levels of APOD in HCs transfected with either the negative control or si-ZDHHC11, detected using the ABE assay. n = 3. d, Palmitoylation levels of APOD in HCs infected with either the ZDHHC11 overexpression lentivirus or the control lentivirus, detected using the ABE assay. n = 3. e, Detection of palmitoylated APOD in the presence of 10 μM palmitoyl azide-CoA, using the in vitro PAT assay. n = 3. f, Heatmap visualization of P53, P21, MMP3, IL6, COL2A1, SOX9, ADAMTS4 and APOD levels in HCs infected with either the APOD overexpression lentivirus or the control lentivirus. Asterisks (*) denote statistical significance between control and OE-APOD, while pound keys (#) denote statistical significance between control and OE-APOD(C185A). n = 3 independent experiments per group. g, Western blotting of COL2A1, SOX9, ADAMTS4, MMP3, P53, P21, P16 and APOD expression in HCs infected with either the APOD overexpression lentivirus or the control lentivirus. n = 3. h, Representative images of SA-β-Gal staining in HCs infected with either the APOD overexpression lentivirus or the control lentivirus. Scale bars, 50 μm. i, Representative Alcian blue staining in HCs infected with the APOD overexpression lentivirus or control lentivirus (the ‘shift’ images were obtained by inverting the original blue image using ImageJ software). In b–i, experiments were repeated at least three times independently, with similar results. One-way ANOVA with Tukey’s correction for multiple comparisons was used for f. NS (P > 0.05), *P < 0.05, **P < 0.01, ***P < 0.001. The exact P values are reported as source data. Panel a created with BioRender.com.

Source data

Extended Data Fig. 6. APOD facilitates ECM anabolism and inhibits chondrocyte senescence.

(a) Heatmap visualization of P53, P21, MMP3, IL-6, COL2A1, SOX9, ADAMTS4 and APOD levels in negative control or APOD siRNAs group. Pound key (*) denotes statistical significance between control and si-APOD#1, asterisk (#) denotes statistical significance between control and si-APOD#2. n = 3 independent experiments per group. (b) Western blotting of COL2A1, SOX9, ADAMTS4, MMP3, P53, P21, P16 and APOD expression in HCs transfected with the negative control or APOD siRNAs. n = 3. (c) Representative Alcian blue staining in HCs transfected with the negative control or APOD siRNAs (the “shift” images were obtained by inverting the original blue image using Image J software). (d) Quantification of relative Alcian blue intensity in HCs transfected with the negative control or APOD siRNAs shown as mean ± SD. n = 3 independent experiments per group. (e) Quantitative data of SA-β-Gal positive cells transfected with the negative control or APOD siRNA shown as mean ± SD. n = 9 independent experiments per group. (f) Representative images of SA-β-Gal staining in HCs transfected with the negative control or APOD siRNAs. Scale bars, 50 μm. (g) Quantitative data of SA-β-Gal positive cells infected with the APOD overexpression lentivirus or control lentivirus shown as mean ± SD. n = 9 independent experiments per group. (h) Quantification of relative Alcian blue intensity in HCs infected with the APOD overexpression lentivirus or control lentivirus shown as mean ± SD. n = 3 independent experiments per group. In b, c and f, experiments were repeated at least three times independently with similar results. One-way ANOVA with Tukey’s correction for multiple comparisons is used for (a, d, e, g and h). ns (p > 0.05), *p < 0.05, **p < 0.01, ***p < 0.001. The exact p-values are reported in the Source data file for Extended Data Fig. 6.

Source data

APOD depalmitoylation triggers proteasomal degradation through the E3 ubiquitin ligase GNB2

Palmitoylation is known to have a crucial role in protein stability³². To further investigate the impact of palmitoylation on APOD, we treated chondrocytes with the general palmitoylation inhibitor 2-BP. In the presence of 2-BP, the expression of APOD decreased in a time- and dose-dependent manner (Fig. 6a,b). Immunofluorescence staining confirmed that depalmitoylation at the C185A site inhibited the stabilization of APOD (Extended Data Fig. 7a). Interestingly, blockade of palmitoylation further accelerated APOD degradation when HCs were treated with the protein synthesis inhibitor cycloheximide alone (Fig. 6c). Targeted protein degradation in eukaryotes occurs mainly through the lysosomal pathway and the proteasome system³³. Hence, we incubated chondrocytes with chloroquine and bortezomib as specific pathway inhibitors for lysosomal and proteasomal degradation, respectively. The results revealed that bortezomib, rather than chloroquine, notably prevented the proteasome-mediated degradation of depalmitoylated APOD (Fig. 6c). Notably, the ubiquitin–proteasome system predominates as the primary proteasome-associated pathway in protein degradation³⁴.

Fig. 6. APOD depalmitoylation leads to proteasomal degradation through the E3 ubiquitin ligase GNB2.

a, Western blotting of APOD expression in HCs incubated with 50 μM 2-BP for 0–48 h. n = 3. b, Western blotting of APOD expression in HCs treated with 2-BP at the indicated concentrations for 24 h. n = 3. c, The degradation of APOD in HCs was evaluated using cycloheximide (CHX) chase assay in the presence of inhibitors of palmitoylation (2-BP), proteasomes (bortezomib (Borz)) and lysosomes (chloroquine (CHQ)). n = 3. d, Schematic illustration of the intersection between the MYC-APOD IP–MS and UbiBrowser prediction. e, The interaction between GNB2 and APOD was confirmed by a co-IP assay. n = 3. Bortezomib treatment for 6 h. f, Effect of GNB2 overexpression on APOD ubiquitination. n = 3. Bortezomib treatment for 6 h. g, Effect of ZDHHC11 knockdown on the interaction between GNB2 and APOD in HCs. n = 3. Bortezomib treatment for 6 h. h, Molecular docking predicted the interaction between GNB2 and APOD. i, Molecular docking predicted the interaction between GNB2 and palm-APOD. j, Effect of APOD(C185A) mutant overexpression or 50 μM 2-BP on APOD ubiquitination. n = 3. Bortezomib treatment for 6 h. k, Effect of ZDHHC11 knockdown on APOD ubiquitination. n = 3. Bortezomib treatment for 6 h. l, Representative images of the immunofluorescence of MMP13, COL2A1 and P16 in C28/I2 cells stably expressing APOD in the indicated groups: untreated (left), 50 μM 2-BP (middle) and 50 μM 2-BP plus GNB2 knockdown (right). Scale bars, 20 μm. In a–c, e–g and j–l, experiments were repeated at least three times independently, with similar results.

Source data

Extended Data Fig. 7. GNB2 is identified as an E3 ubiquitin ligase of APOD.

(a) Representative images of immunofluorescence of APOD in APOD-overexpressed or C185A-overexpressed HCs treated with or without 50 μM 2-BP for 12 h. Scale bars, 20 μm. (b) Coomassie Blue staining of protein pulled down by APOD co-IP assay. (c) The potential interaction between E3 ligase and APOD predicted by the UbiBrowser website. (d) The interaction between GNB2 and APOD was confirmed by co-IP assay. n = 3. Borz treatment for 6 h. (e) Representative images of immunofluorescence of APOD and GNB2 colocalization in HCs. Scale bars, 50 μm. (f) Western blotting of GNB2 and ZDHHC11 expression in HCs transfected with the ZDHHC11 siRNAs or infected with the ZDHHC11 overexpression lentivirus. n = 3. (g) Quantification of MMP13 staining intensity shown as mean ± SD. n = 9 independent experiments per group. (h) Quantification of P16 staining intensity shown as mean ± SD. n = 9 independent experiments per group. (i) Quantification of COL2A1 staining intensity shown as mean ± SD. n = 9 independent experiments per group. In a, d, e and f, experiments were repeated at least three times independently with similar results. One-way ANOVA with Tukey’s correction for multiple comparisons is used for (g, h and i). ***p < 0.001. The exact p-values are reported in the Source data file for Extended Data Fig. 7.

Source data

In this context, we sought to identify the underlying E3 ubiquitin ligases responsible for regulating APOD stability by conducting IP–MS on APOD-overexpressing chondrocytes (Extended Data Fig. 7b). Additionally, an integrated online bioinformatics platform, UbiBrowser (http://ubibrowser.ncpsb.org.cn), was used for predictions (Extended Data Fig. 7c). Following the IP–MS and bioinformatics analyses, a Venn diagram revealed that GNB2 was the only E3 ubiquitin ligase responsible for APOD degradation (Fig. 6d). Thus, GNB2 was selected as a potential candidate for APOD degradation. As expected, we observed an interaction between GNB2 and APOD (Fig. 6e and Extended Data Fig. 7d) and confirmed their colocalization in the cytosol (Extended Data Fig. 7e). Consistent with this, overexpression of GNB2 led to an increase in the ubiquitination levels of APOD (Fig. 6f), supporting the hypothesis that GNB2 facilitates APOD degradation through the ubiquitin–proteasome pathway. Next, we explored the relationship between palmitoylation and ubiquitination. As illustrated in Fig. 6g, western blot analysis demonstrated increased aggregation of GNB2 with APOD following ZDHHC11 knockdown. Furthermore, neither ZDHHC11 knockdown nor ZDHHC11 overexpression directly influenced GNB2 at the protein level, indicating that the ZDHHC11-mediated ubiquitin-specific degradation of APOD was not due to the endogenous accumulation of GNB2 (Extended Data Fig. 6f). To further investigate steric inhibition, we constructed computer homology models for protein docking. Here, we added a palmitoyl group to the cysteine residue of the APOD protein and assessed its interaction with GNB2. As shown in Fig. 6h,i, the addition of a palmitoyl group led to a lower docking score between palm-APOD and GNB2, accompanied by a reduction in hydrogen bonds, indicating a less favorable binding interaction. Either the transfection of the APOD(C185A) mutant alone or the combined treatment with APOD and 2-BP resulted in increased ubiquitination of APOD (Fig. 6j). Finally, we found that ZDHHC11 is a bona fide palmitoyltransferase that competes with GNB2 for APOD degradation (Fig. 6k). To study the effects of palmitoylation and ubiquitination on the biological function of APOD, we performed immunofluorescence experiments. The blockade of palmitoylation inhibited the protective effect of APOD, while the knockdown of GNB2 promoted APOD-mediated chondroprotective effects in HCs (Fig. 6l and Extended Data Fig. 7g–i). In summary, we elucidated that ZDHHC11-mediated palmitoylation inhibits the ubiquitination and subsequent degradation of APOD through a mechanism regulated by the E3 ubiquitin ligase GNB2.

ZDHHC11-mediated palmitoylation alleviates OA progression by suppressing the GATA4–P65 signaling pathway

Next, we unraveled the molecular mechanism through which ZDHHC11-mediated palmitoylation governs ECM metabolism and cellular senescence. Hence, we selectively knocked down ZDHHC11 or APOD in HCs and performed RNA sequencing (RNA-seq) (Extended Data Fig. 8a–d). Analysis of commonly differentially expressed genes (|log₂(FC)| > 2.5) revealed that GATA4, AFF2 and NANOS1 were potential responders triggered by either ZDHHC11 or APOD knockdown (Fig. 7a). RT–qPCR further validated that only the expression of GATA4 exhibited consistent alterations under the regulation of ZDHHC11 or APOD (Fig. 7b,c). Additionally, KEGG (Kyoto Encyclopedia of Genes and Genomes) enrichment and Circos plot analyses illustrated an upregulation of the cellular senescence pathway after silencing APOD in HCs (Extended Data Fig. 8e,f). Notably, GATA4 was the most highly upregulated senescence-related gene in APOD-knockdown chondrocytes (Extended Data Fig. 8f). Previous work has clarified that GATA4 has a distinct capacity to control the proinflammatory secretory niche associated with aging by activating downstream nuclear factor-κB (NF-κB) signaling³⁵. Similarly, several OA-related pathways, including the NF-κB signaling pathway, were notably activated (Fig. 7d). Based on these findings, we focused on the GATA4–NF-κB axis for further investigation. RT–qPCR analysis revealed a drastic decrease in GATA expression following ZDHHC11 or APOD overexpression (Fig. 7e and Extended Data Fig. 8g). Western blotting results indicated enhanced expression of the GATA protein and phosphorylation of the key protein P65 (p-P65) after depalmitoylation; however, these effects were reversed by ZDHHC11 or APOD overexpression (Fig. 7f,g and Extended Data Fig. 8h,i). Moreover, APOD or ZDHHC11 knockdown promoted P65 phosphorylation and nuclear translocation, an effect that was abolished by GATA4 silencing (Fig. 7h,i and Extended Data Fig. 8j,k). Western blotting experiments on human samples revealed a notable increase in the expression levels of GATA4 and p-P65 in senescent chondrocytes. These findings suggest that the GATA4–P65 axis has a key role in regulating chondrocyte senescence (Extended Data Fig. 8l). To elucidate the role of the intricate relationship between palmitoylation and ubiquitination in the progression of OA, we designed several treatment groups (Fig. 7j). Our results demonstrated that depalmitoylation of APOD through 2-BP treatment induced P65 phosphorylation and upregulated GATA4 expression. Furthermore, coexpression of APOD and GNB2, rather than APOD expression alone, yielded similar results to those observed with APOD depalmitoylation (Fig. 7j). Finally, we aimed to investigate whether APOD supplementation could reverse ZDHHC11 knockdown-triggered chondrocyte senescence and ECM alterations. Consistent with our hypothesis, APOD overexpression notably promoted the expression of COL2A1 and SOX9, accompanied by the suppression of ADAMTS4, MMP3, P53, P21 and P16 expression in chondrocytes treated with si-ZDHHC11 (Fig. 7k). Therefore, these results confirm that ZDHHC11-mediated palmitoylation of APOD alleviates OA progression by competitively inhibiting GNB2-mediated ubiquitination, resulting in the downregulation of the senescence-related GATA4–P65 signaling pathway.

Extended Data Fig. 8. RNA-sequencing reveals that ZDHHC11-mediated APOD palmitoylation inactivates GATA4-P65 signaling.

(a) Histogram of DEGs after APOD knockdown by RNA-seq. (b) Heatmap visualization after APOD knockdown by RNA-seq (FDR ≤ 0.05 and |log2 (fold change)| > 1). (c) Histogram of DEGs after ZDHHC11 knockdown by RNA-seq. (d) Heatmap visualization after ZDHHC11 knockdown by RNA-seq (FDR ≤ 0.05 and |log2 (fold change)| > 1). (e) Bubble diagram of KEGG enrichment analysis after APOD knockdown by RNA-seq. (f) Circos plot of the DEGs enriched in KEGG pathway after APOD knockdown. (g) RT- qPCR of GATA4 levels in HCs infected with the APOD overexpression lentivirus or control lentivirus. (n = 9 independent experiments per group, mean ± SD). (h) Western blotting of GATA4, p-P65, P65 and APOD expression in HCs infected with the control lentivirus or APOD overexpression lentivirus. n = 3. (i) Western blotting of GATA4, p-P65, P65 and APOD expression in HCs transfected with the negative control or APOD siRNAs. n = 3. (j) Western blotting of GATA4, p-P65 and P65 expression in HCs transfected with the APOD siRNAs or GATA4 siRNAs. n = 3. (k) Quantification of p-P65 staining intensity shown as mean ± SD. n = 9 independent experiments per group. (l) Western blotting of GATA4, p-P65 and P65 expression between younger group and older group. n = 3. In h, i, j and l, experiments were repeated at least three times independently with similar results. Two-sided Student’s T-test is used for (g) and one-way ANOVA with Tukey’s correction for multiple comparisons for (k). ***p < 0.001. The exact p-values are reported in the Source data file for Extended Data Fig. 8.

Source data

Fig. 7. ZDHHC11-mediated palmitoylation alleviates OA progression by suppressing the GATA4–P65 signaling pathway.

a, Schematic illustration of the intersection between the RNA-seq results after ZDHHC11 or APOD knockdown. b, RT–qPCR of GATA4, AFF2 and NANOS1 levels in HCs transfected with either the negative control or si-ZDHHC11 (n = 3 independent experiments per group, mean ± s.d.). c, RT–qPCR of GATA4, AFF2 and NANOS1 levels in HCs transfected with either the negative control or si-APOD (n = 3 independent experiments per group, mean ± s.d.). d, Lollipop chart showing the differences in signaling pathways after the knockdown of ZDHHC11 or APOD. Left: Gene Ontology enrichment analysis after ZDHHC11 knockdown. Right: Gene Ontology enrichment analysis after APOD knockdown. e, RT–qPCR of GATA4 levels in HCs infected with either the ZDHHC11 overexpression lentivirus or the control lentivirus (n = 9 independent experiments per group, mean ± s.d.). f, Western blotting of GATA4, p-P65, P65 and ZDHHC11 expression in HCs transfected with the negative control or ZDHHC11-specific siRNAs. n = 3. g, Western blotting of GATA4, p-P65, P65 and ZDHHC11 expression in HCs infected with either the control lentivirus or the ZDHHC11 overexpression lentivirus. n = 3. h, Representative images of the immunofluorescence of p-P65 in HCs transfected with si-GATA4, si-APOD or si-ZDHHC11. Scale bars, 100 μm. i, Western blotting of GATA4, p-P65, P65 and ZDHHC11 expression in HCs transfected with si-ZDHHC11 or si-GATA4. n = 3. j, Western blotting of GATA4, p-P65, P65 and APOD expression in the indicated treatment groups. n = 3. k, Western blotting of GATA4, COL2A1, SOX9, ADAMTS4, MMP3, P53, P21, P16 and ZDHHC11 expression in HCs transfected with si-ZDHHC11 or infected with the APOD overexpression lentivirus. n = 3. In f–k, experiments were repeated at least three times independently, with similar results. Two-sided Fisher’s exact test with the Benjamini–Hochberg procedure for multiple comparisons was used for d. Two-sided Student’s t-test was used for e. ***P < 0.001. The exact P values are reported as source data.

Source data

Zdhhc11 mRNA-LNP restores cartilage integrity in OA and notably mitigates joint senescence in rats

To date, mRNA-based therapies have shown promise as an alternative platform for delivering therapeutic proteins³⁶. Due to their positive charge and resistance to nuclease degradation, LNPs have emerged as a leading nanocarrier platform for therapeutic mRNA delivery^37,38. Furthermore, precise drug delivery and minimization of off-target effects are essential for advancing the clinical potential of nucleic acid therapeutics. Previous studies have demonstrated that conjugating CAPs to the exosome surface enables targeted delivery to chondrocytes^39,40. In this context, we propose assembling selective cell-targeting LNPs by fusing CAPs with liposomes to specifically target chondrocytes in the treatment of OA in a rat model (Fig. 8a).

Fig. 8. Zdhhc11 mRNA-LNP rescues OA cartilage degeneration and effectively attenuates senescence in rat joints.

a, Schematic depicting the assembly of LNPs with Zdhhc11 mRNA. b, Representative TEM images of mRNA-LNP and mRNA-LNP–CAP. Scale bars, 100 nm. c, Hydrodynamic diameter of LNPs (n = 3 independent experiments per group, mean ± s.d.). d, Zeta potential of LNPs (n = 3 independent experiments per group, mean ± s.d.). e, Encapsulation efficiencies (EE) of LNPs (n = 3 independent experiments per group, mean ± s.d.). f, Micro-CT images of knee joint osteophytes in each group: sham (n = 6), DMM (n = 6), DMM + LNP (n = 6), DMM + mRNA-LNP (n = 6) and DMM + mRNA-LNP–CAP (n = 6). Scale bars, 5 mm. g, Quantification of osteophyte numbers in each group (n = 6 animals per group). h, Representative images of H&E staining of the synovium in each group. Scale bars, 100 μm. i, Representative images of Safranin O/Fast Green staining and the immunofluorescence of GATA4 and p-P65 in each group. Scale bars, 50 μm. j, Quantification of synovial hyperplasia according to the synovitis score in each group (n = 6 animals per group). k, Quantification of cartilage damage according to the OARSI scoring system in each group (n = 6 animals per group). l, A model of the proposed mechanism by which ZDHHC11-mediated palmitoylation ameliorates chondrocyte senescence. One-way ANOVA with Tukey’s correction for multiple comparisons was used for i (mean ± s.d.). Two-sided Mann–Whitney U test was used for j and k (mean ± 95% CI). *P < 0.05, **P < 0.01, ***P < 0.001. The exact P values are reported as source data. Panels a and l created with BioRender.com.

Source data

Subsequently, the hybrid LNPs were characterized to assess particle morphology, alterations in zeta potential and mRNA encapsulation efficiency. Transmission electron microscopy (TEM) analysis revealed that hybrid LNP–CAPs displayed larger particle sizes compared with mRNA-LNPs, with both types exhibiting a near-spherical morphology (Fig. 8b,c). The hybridization of mRNA and CAP within the LNPs induced a shift in zeta potential close to neutrality (Fig. 8d). The encapsulation efficiency of mRNA-LNPs was quantified using a fluorescent RiboGreen assay, yielding an average efficiency of 95.7%. Notably, losses of LNPs during the modification and purification processes resulted in LNP–CAPs achieving an encapsulation efficiency of approximately 50% (Fig. 8e). To confirm that mRNA-LNP–CAP specifically targeted chondrocytes, we performed western blotting and found that the hybrid LNPs could effectively enter primary chondrocytes and translate into functional proteins (Extended Data Fig. 9a). Additionally, we conducted immunofluorescence staining on chondrocytes and FLS at the indicated time points. As shown in Extended Data Fig. 9b,c, our self-developed mRNA-LNP–CAP demonstrated an evident targeting effect in chondrocytes as early as 2 h after treatment, compared with mRNA-LNP. Furthermore, a higher density of green fluorescence signals was observed within the chondrocytes, indicating a greater uptake of mRNA-LNP–CAP by chondrocytes compared with FLS.

Extended Data Fig. 9. Selection of optimal volume used for Zdhhc11 mRNA@LNP in treating OA.

(a) Western blotting of COL2A1, ADAMTS4 and ZDHHC11 expression in chondrocytes treated with the EGFP@LNP or Zdhhc11 mRNA@LNP. n = 3. (b) Quantitative data of EGFP positive cells incubated with either EGFP mRNA@LNP or EGFP mRNA@LNP-CAP in chondrocytes and FLS at indicated time points shown as mean ± SD. n = 6 independent experiments per group. (c) Representative images of immunofluorescence of EGFP in chondrocytes and FLS at indicated time points. Scale bars, 100 μm. (d) Representative images of Safranin O/Fast green staining in mRNA@LNP or mRNA@LNP-CAP group with different drug dosages. Scale bars, 50 μm. (e) Quantification of cartilage damage according to OARSI scoring system in mRNA@LNP or mRNA@LNP-CAP group (n = 6 animals per group). In a, c and d and h, experiments were repeated at least three times independently with similar results. One-way ANOVA with Tukey’s correction for multiple comparisons is used for (b), Means ± SD. Two-sided Mann-Whitney U test is used for (e), Means ± 95% CI. ns (p > 0.05), **p < 0.01, ***p < 0.001. The exact p-values are reported in the Source data file for Extended Data Fig. 9.

Source data

To validate the in vivo results, we evaluated the potential of hybrid LNPs in ameliorating OA-related phenotypes in a rat model. Similarly, therapeutic interventions included weekly intra-articular injections administered over a 6-week period after DMM surgery (Fig. 8a). To compare the protective effects of Zdhhc11 mRNA-LNP–CAP and Zdhhc11 mRNA-LNP, we administered different volumes of LNPs into model rats. We found that Zdhhc11 mRNA-LNP–CAP notably attenuated cartilage destruction at the tested concentration. Zdhhc11 mRNA-LNP exhibited minimal protective effects at a dose of 20 μl, whereas Zdhhc11 mRNA-LNP–CAP conferred robust protection against cartilage degradation even at a low dose (Extended Data Fig. 9d,e). Based on these observations, we selected 20 μl as the optimal dosing volume for subsequent experiments. In this context, micro-CT analysis was used to evaluate alterations in subchondral bone tissue following different treatments in DMM rats. Imaging of the knee joints revealed a significant reduction in osteophyte formation following treatment with Zdhhc11 mRNA-LNP–CAP, compared with the DMM control condition and treatment with Zdhhc11 mRNA-LNP (Fig. 8f,g). To assess the therapeutic efficacy of mRNA-LNP–CAP at the histological level, we performed a series of routine staining procedures on cartilage tissue sections. Synovial inflammation was assessed using H&E staining. Elevated levels of cell infiltration into the synovial lining were observed following DMM surgery, whereas hybrid LNP–CAP treatment markedly reduced the thickness of the synovial lining cell layer, as evidenced by lower synovitis scores (Fig. 8h,j). In DMM rats treated with mRNA-LNP–CAP, Safranin O/Fast Green staining demonstrated mitigated degeneration of the cartilage layer and enhanced staining intensity (Fig. 8i). The mRNA-LNP–CAP treatment group also exhibited notably lower OARSI scores (Fig. 8k). IHC staining of COL2A1 and MMP3 was performed to evaluate the effects of intra-articular administration of mRNA-LNP–CAP on cartilage ECM metabolism. Consistent with expectations, mRNA-LNP–CAP treatment markedly upregulated COL2A1 expression and downregulated MMP3 levels, indicating a suppression of ECM catabolic activity alongside an enhancement of anabolic processes (Extended Data Fig. 10a). Additionally, the treatment resulted in decreased expression of the cellular senescence markers P21 and P16, further supporting the antisenescence effects of mRNA-LNP–CAP (Extended Data Fig. 10a). This conclusion was further reinforced through immunofluorescence analysis of GATA4 and p-P65 (Fig. 8i and Extended Data Fig. 10b,c). Collectively, these results indicate that Zdhhc11 mRNA-LNP–CAP promotes cartilage repair and exhibits antisenescence properties, with the effects particularly enhanced by chondrocyte-targeted hybrid LNP–CAPs.

Extended Data Fig. 10. Intra-articular Zdhhc11 mRNA@LNP can attenuate the synovial inflammation of surgery-induced OA.

(a) Representative images of IHC of P16, COL2A1, MMP3 and P21 in each group. Scale bars, 50 μm. (b) Quantitative data of GATA4 in each group shown as mean ± SD. Sham group (n = 6 animals per group), DMM group (n = 6 animals per group), DMM + LNP group (n = 6 animals per group), DMM + mRNA@LNP group (n = 6 animals per group), DMM + mRNA@LNP-CAP group (n = 6 animals per group). (c) Quantitative data of p-P65 in each group shown as mean ± SD. n = 6 animals per group. In a, experiments were repeated six times independently with similar results. One-way ANOVA with Tukey’s correction for multiple comparisons is used for (b, c), Means ± SD. ns (p > 0.05), **p < 0.01, ***p < 0.001. The exact p-values are reported in the Source data file for Extended Data Fig. 10.

Source data

Discussion

Osteoarthritis, a prevalent arthritic condition that affects more than 50% of the older adult population, leads to a progressive deterioration of joint function⁴¹. Generally, OA notably limits mobility and diminishes the quality of life in affected individuals^42,43. With age emerging as the predominant risk factor for disease development, the exact mechanisms linking senescence and OA pathology remain unknown. Jeon et al.⁴⁴ reported that the local clearance of senescent cells inhibited the development of post-traumatic OA, further supporting the notion that OA is an age-related disease. In this work, we demonstrate the chondroprotective property of ZDHHC11 in OA progression. Notably, the overexpression of ZDHHC11 inhibits catabolism and the senescent phenotype in HCs. Mechanistically, ZDHHC11-mediated palmitoylation regulates chondrocyte senescence by inhibiting the ubiquitination-dependent degradation of APOD through the downregulation of the GATA4–P65 signaling pathway (Fig. 8l).

Post-translational modifications influence protein function through the regulation of protein abundance or activity^45–47. To investigate the role of ZDHHC11 in vivo, we intra-articularly injected Zdhhc11-AAV into a mouse model of DMM-induced OA. Compared with the DMM group, micro-CT results indicated that ZDHHC11 supplementation markedly reduced the number of osteophytes. The synovium and OARSI scores were lower, and the cartilage layer was thinner in the DMM group than in the Zdhhc11-AAV DMM group. Zdhhc11-AAV also promoted the expression of COL2A1 while reducing the expression of IL-6, P21 and MMP3. Here, we used 3- and 24-month-old mice to investigate the underlying relationship between senescence and OA. However, one limitation of our study is that we used DMM surgery to model post-traumatic OA. This model may not fully replicate the molecular and cellular processes underlying age-associated OA, which occur naturally in humans. Moreover, Zdhhc11^fl/fl; Aggrecan^CreERT2 (cKO) mice were successfully generated, subjected to DMM surgery and intra-articularly injected with Zdhhc11-AAV. As expected, Zdhhc11 cKO-DMM mice exhibited quintessential OA phenotypes. Genetic ablation of Zdhhc11 in chondrocytes notably aggravated articular cartilage degeneration, as indicated by higher OARSI grades, increased osteophyte formation and enhanced infiltration of FLS. IHC staining further confirmed that cKO mice exhibited upregulation of MMP3, P21 and IL-6, accompanied by a decrease in the expression of COL2A1. Thus, we conclude that ZDHHC11 acts as an antisenescence agent and attenuates OA cartilage damage.

Mounting evidence has revealed a close relationship between protein S-palmitoylation and ubiquitination^48,49. ZDHHC5 catalyzes the S-palmitoylation of CLOCK at Cys194 by preventing ubiquitination-dependent degradation⁵⁰. Likewise, programmed death ligand 1 (PD-L1) palmitoylation stabilizes PD-L1 by internalizing it into recycling endosomes and prevents its lysosomal degradation³². APOD is a multifunctional molecule that affects lipid metabolism, inflammation and neuroprotection^51,52. In recent decades, studies have unraveled altered APOD expression in age-related diseases and neurological disorders^53–55. Interestingly, we classified nine articular chondrocyte subtypes by annotating canonical gene expression profiles at the single-cell level. A senescence-associated gene dataset was also used to evaluate the senescence scores across these subpopulations. Of note, we found that APOD expression was negatively correlated with the senescence score. In vitro studies also confirmed that APOD could attenuate joint cartilage degeneration by protecting chondrocytes from senescence.

Importantly, we discovered that GNB2 was the only E3 ubiquitin ligase that could act directly on APOD based on experimental and bioinformatics analyses. Sterically, palmitoylation competitively hinders GNB2 from interacting with APOD. Specifically, we demonstrated that depalmitoylation weakened the chondroprotective effect of APOD, which was reversed by GNB2 silencing. Disruption of APOD palmitoylation led to increased amounts of MMP13 and P16 proteins, accompanied by a reduced level of COL2A1 in chondrocytes. Furthermore, depalmitoylation of APOD or overexpression of GNB2 notably activated the GATA4–P65 signaling pathway. Overall, we elucidated that the ZDHHC11–APOD–GNB2 complex dynamically regulates chondrocyte senescence in OA.

In recent decades, mRNA-based synthetic chemical therapies have demonstrated substantial promise in the treatment of OA⁵⁶. This is because mRNA does not necessitate nuclear localization and offers a safe and effective alternative⁵⁷. LNPs function as cationic delivery systems, while the principal molecular constituents of the extracellular cartilage—proteoglycans—possess a negative charge. This charge differential endows LNPs with intrinsic chondrocyte-targeting capabilities. Hybrid LNPs preserve the cell-specific targeting attributes of CAPs, as evidenced by our findings. To gain more insights into the therapeutic use of ZDHHC11, we developed a chondrocyte-targeting LNP–CAP that encapsulates Zdhhc11 mRNA and validated its efficacy in vivo. The Zdhhc11 mRNA-LNP notably alleviated the OA phenotypes induced by DMM surgery. Before our findings can be translated into clinical practice, several challenges must be addressed. First, compromised encapsulation efficiency during the CAP modification and hybrid LNP purification processes may result in reduced mRNA encapsulation rates. Second, as the notion of the Zdhhc11 mRNA-LNP–CAP delivery system was only recently reported, further evaluation is crucial to better understand its effectiveness across different species.

In conclusion, previous studies have revealed that the age-dependent deterioration of chondrocyte function undermines cartilage function in OA^58,59. This study explores palmitoylation-mediated protein modification in the context of OA. Furthermore, the innovative ZDHHC11 mRNA-LNP–CAP delivery system represents a promising tool for targeting chondrocytes, offering an alternative therapeutic approach for the treatment of OA.

Methods

Human tissue specimens

This study was performed in accordance with the experimental protocol approved by the Ethics Committee of Sir Run Run Shaw Hospital (approval no. 20230756). We collected human knee samples of various ages from patients with primary OA who underwent total knee replacement surgery. The severity of knee OA was graded according to the K–L grading system⁶⁰ (Supplementary Table 6). The patients were divided into two groups according to age: the younger group (40–65 years old) and the older group (older than 70 years). In both groups, we then divided all cartilage samples into two additional groups: lateral OA and medial OA. Patient information is summarized in Supplementary Table 1. These cartilage specimens were subsequently used to isolate chondrocytes or perform histological experiments. All patients provided written informed consent before enrolling in this study, and none were aware of the study design, conduct or reporting.

Animal studies

All animal experiments were conducted in accordance with institutional guidelines (Regulations for the Administration of Affairs Concerning Experimental Animals issued by the State Science and Technology Commission) and were approved by the Laboratory Animal Welfare and Ethics Committee of Zhejiang University (approval no. ZJU20240052). The mice were housed under specific pathogen-free conditions and maintained in a controlled environment with a 12-h light–dark cycle, an ambient temperature of 25 °C and air humidity ranging from 50% to 65%.

Zdhhc11^fl/fl mice were purchased from Cyagen Biosciences and crossed with Aggrecan^CreERT2 mice to establish the Zdhhc11-knockout mouse (cKO) line.

For the inducible deletion of Zdhhc11, 2-month-old male Zdhhc11^fl/fl; Aggrecan^CreERT2 mice were administered five daily intraperitoneal injections of tamoxifen (Sigma, cat. no. T5648) dissolved in corn oil, at a dose of 100 μg per gram of body weight.

As shown in Figs. 2, 3 and 8, we randomly divided the male C57BL/6 mice and SD rats into different groups (n = 6 per group). Animals were anesthetized with an intraperitoneal injection of 1% sodium pentobarbital before being subjected to DMM surgery⁶¹. Then, we used microsurgical scissors to cut the medial meniscotibial ligament, followed by flushing the area with sterile saline and closing the incision with sutures. In the sham group, the skin was cut and sutured directly without removing the medial meniscotibial ligament.

In the AAV treatment experiments⁶², mice were subjected to intra-articular injections using a 10-μl microsyringe with a total volume of 10 μl (approximately 1 × 10¹² vector genomes per milliliter) 2 weeks after DMM surgery. Apod-AAV and Zdhhc11-AAV were constructed and packaged by Hanbio.

In the LNP treatment experiments, 10-week-old rats were subjected to intra-articular injections using a 25-μl microsyringe with a total volume of 20 μl mRNA-LNP 2 weeks following DMM modeling. Injections were administered weekly for a total of six doses, and all animals were killed 8 weeks after the DMM operation.

Primary cell culture

The collected human cartilage tissue was cut into pieces in a sterile environment and washed with PBS three times. The pieces were then digested overnight in a 37 °C culture incubator with 0.2% collagenase type II (Sigma, cat. no. 1148090). On the next day, the supernatant was filtered using a 100-μm cell strainer (Biosharp, cat. no. BS-100-XBS) and centrifuged at 300g for 5 min. The cells were then washed three times with PBS, cultured in DMEM supplemented with 10% FBS, and maintained in a humidified incubator set at 37 °C and 5% CO₂.

We chose 5-day-old C57BL/6 mice to obtain mouse articular cartilage⁶³. Briefly, the hind limbs were dissected and washed three times with PBS. Skin and soft tissues were removed to expose the articular tissues. The articular cartilage on the femoral condyle and tibial plateau was then removed using ophthalmic microsurgery. The subsequent procedures were performed similarly to the isolation of HCs.

Preparation and characterization of mRNA-LNP

Zdhhc11 mRNA was purchased from CYNBIO (cat. no. CYNBIO-YX-S-240097-ZD-SYF).

The LNP was formulated using ionizable lipids, including SM-102, DSPE-PEG-MAL, DSPC and cholesterol, at a molar ratio of 50:1.5:10:38.5. mRNA (25 μl, 1,000 ng μl⁻¹), dissolved in 50 μl Tris buffer (20 mM) at pH 4.0, was mixed with lipids dissolved in ethanol within a microfluidic device, maintaining a volume ratio of 3:1 for mRNA to lipids. One minute later, the mixture was incubated at room temperature for 15 min. After incubation, a 1.5-fold volume of 1× PBS was added to further reduce the ethanol concentration.

The purification steps were as follows. (1) Column preparation: a 5-ml desalting column was prepacked, connected to a syringe containing 25 ml ultrapure water, and flushed at a rate of 120 drops per minute before being sealed at the base, followed by hydration of the top interface. (2) Equilibration: the column was equilibrated with 25 ml equilibration buffer at 120 drops per minute; after sealing, a small volume of buffer was added to the top interface. (3) Sample loading: the LNP sample was loaded using a syringe at 120 drops per minute, after which the column was sealed and the top interface was hydrated. (4) Elution: the elution buffer (25 ml) was passed through the column at 120 drops per minute, with the first 5 ml discarded; the target eluate (approximately twice the sample volume) was collected before sealing and topping with buffer. (5) Column cleaning: the column was sequentially washed with 25 ml elution buffer, ultrapure water and 20% ethanol; then, it was sealed, topped with 20% ethanol and stored at 2–8 °C.

The obtained nanoparticles were filtered using a 0.22-μm filter and stored at 4 °C until further use. TEM was used to capture images. The hydrodynamic diameters and zeta potentials of the LNP samples were quantified using a Zetasizer Nano ZS90 analyzer (Malvern Instruments). Additionally, the encapsulation efficiency of mRNA was assessed using a modified Quant-iT RiboGreen RNA assay (Invitrogen).

Drug treatments

For in vitro experiments, chondrocytes were incubated with the following drugs: H₂O₂ (200 μM), IL-1β (10 ng ml⁻¹; Proteintech, cat. no. HZ-1164), bortezomib (250 nM; Selleck, cat. no. S1013), chloroquine (50 μM; Selleck, cat. no. S6999), cycloheximide (10 μg ml⁻¹; Selleck, cat. no. S7418) and 2-BP (0–100 μM; Sigma-Aldrich, cat. no. 238422).

Western blotting

The prepared protein samples were separated using an SDS–PAGE gel and transferred to a PVDF membrane (Bio-Rad, cat. no. 1620177). Then, the membrane was blocked with 5% skim milk at room temperature for 1 h, followed by incubation with the primary antibody at 4 °C overnight (Supplementary Table 2). Following the initial incubation, the cells were subjected to three washes with TBST before being incubated with secondary antibodies at room temperature for 1 h. Finally, antibody binding was detected using an ECL substrate (Fude Biological Technology, cat. no. FD8020) and a chemiluminescence system (Bio-Rad).

Acyl-biotin exchange

The ABE procedure followed the established methods⁶⁴. Briefly, we mixed the treated cells with lysis buffer (in distilled water: 1% IGEPAL CA-630, 50 mM Tris–HCl, pH 7.5, 150 mM NaCl, 10% glycerol, 10 mM PMSF and one tablet of a protease inhibitor cocktail) at 4 °C for 30 min. The lysate was then transferred into a 1.5-ml centrifuge tube for ultrasonic treatment and centrifuged at 13,000g for 10 min. We then transferred the supernatant into another clean 1.5-ml centrifuge tube and used the IP-ABE Palmitoylation Kit (Aimsmass, cat. no. AM10422) to prepare palmitoyl proteins following the manufacturer’s instructions. First, we blocked unmodified cysteine thiol groups using N-ethylmaleimide and used hydroxylamine to specifically cleave the palmitoylated cysteine’s thiol group. Then, biotin was added to react with free sulfhydryl groups, and finally, the biotin-labeled proteins were enriched by incubation with streptavidin magnetic beads (Beyotime, cat. no. P2151). After incubation, the palmitoyl proteins were released through heat denaturation and detected by western blotting.

MS analysis

The treated cell samples were suspended in protein lysis buffer (8 M urea, 1% SDS) containing protease inhibitors, followed by protein quantification. Proteins (100 μg) were reduced with 10 mM TCEP (37 °C, 60 min) and alkylated with 40 mM iodoacetamide (room temperature, 40 min in darkness). After centrifugation at 10,000g at 4 °C for 20 min, the pellet was resuspended in 100 mM TEAB and digested overnight with trypsin (1:50, 37 °C). Based on the results of peptide quantification, the peptides were analyzed using a Vanquish Neo ultra-high-performance liquid chromatography system coupled with an Orbitrap Astral mass spectrometer (Thermo Fisher Scientific) at Majorbio. Spectronaut software (version 18) was used to search the DIA raw data. Six peptides per protein and three daughter ions per peptide were selected for quantitative analysis. Bioinformatics analysis of proteomic data was performed using the Majorbio Cloud platform (https://cloud.majorbio.com).

Single-cell RNA-seq analysis

The single-cell RNA-seq dataset (GSE169454) was obtained from the Gene Expression Omnibus database³⁰, with chondrocyte subpopulations identified using established canonical markers^65,66. Single-cell RNA-seq data were analyzed by Majorbio. Briefly, raw FASTQ files were processed using Cell Ranger (version 7.1.0), aligning reads to the human genome (STAR algorithm) and generating gene-barcode matrices through unique molecular identifier counting and barcode filtering⁶⁷. The gene-barcode matrix was analyzed in Seurat, where datasets from multiple samples were integrated using the FindIntegrationAnchors and IntegrateData functions. Then, the data were scaled, reduced to 30 principal components and visualized using t-SNE for cluster identification. To evaluate the senescence-associated gene score, we used a gene set (SenMayo) consisting of 125 previously identified senescence/SASP-associated factors, using the AddModuleScore function³¹.

RNA extraction and RT–qPCR

The AG RNAex Pro reagent (Accurate Biotechnology, cat. no. AG21101) was applied to extract total cellular RNA, and the PrimeScript RT reagent kit (Accurate Biotechnology, cat. no. AG11706) was used to synthesize cDNA from 500 ng of total RNA. To evaluate the expression levels of candidate genes, RT–qPCR was performed using an ABI 7500 Sequencing Detection System (Applied Biosystems) with a SYBR Green PCR kit (Yeasen, cat. no. 11204ES). The above steps were performed according to the manufacturer’s instructions. The 2^−ΔΔCT method was applied to calculate the relative mRNA expression levels normalized to the housekeeping gene β-actin. The primer sequences used in this experiment are listed in Supplementary Table 3.

RNA-seq analysis

Total RNA from control and si-APOD-transfected HCs (n = 3 per group) was processed using poly(A) selection with oligonucleotide(dT) beads, followed by fragmentation and double-stranded cDNA synthesis (SuperScript kit, random hexamer primers). Then, the cDNA underwent end repair, phosphorylation, ‘A’ base addition and size selection (300-bp fragments on an agarose gel). Libraries were PCR-amplified (15 cycles with Phusion polymerase) and quantified using Qubit 4.0. After that, paired-end sequencing (2 × 150 bp) was performed on the NovaSeq 6000 (Majorbio). The participant information is listed in Supplementary Table 1. Genes exhibiting P < 0.05 and absolute log₂(FC) > 1 were identified as differentially expressed genes.

Southern blotting

Total RNA was extracted from chondrocytes, and cDNA was synthesized as mentioned above. Gene-specific PCR primers validated by BLAST (https://blast.ncbi.nlm.nih.gov/) were used to analyze mRNA expression with PCR. PCR amplifications used the 2× Rapid Taq Master Mix (Vazyme Biotech, cat. no. P222) with the following program: one step of 3 min at 95 °C, 35 cycles of 15 s at 95 °C, 15 s at 60 °C and 5 s at 72 °C, followed by 5 min at 72 °C. PCR-amplified products were electrophoresed in 1–1.5% TAE gels for 30 min at 150 V and stained with YeaRed (Yeason, cat. no. 10202) for visualization using a chemiluminescence system (Bio-Rad). Supplementary Table 4 lists all sequences used for Southern blotting.

SiRNA transfection

Custom-designed siRNAs were synthesized by GenePharma to knock down specific mRNAs. The cells were transfected with the siRNAs using Lipofectamine iMax (Invitrogen, cat. no. 13778150) at a ratio of 1 μl per 10⁵ cells for 48 h. Supplementary Table 5 lists all siRNA sequences.

Virus infection

Overexpression plasmids of APOD and APOD(C185A) were designed and constructed by Tsingke Bio. ZDHHC11 and GNB2 were provided by Miaoling Bio. After plasmid extraction, each plasmid was cotransfected with the packaging plasmids (pMD2.G and psPAX2) into HEK293T cells using Lipofectamine 3000 (Invitrogen, cat. no. L3000075). The medium was changed 6 h after transfection and was collected 24 h later with polybrene (10 μg ml⁻¹, Solarbio, cat. no. H8761). The medium containing viruses was subsequently added to the target cells.

Coimmunoprecipitation

We mixed the treated cells with RIPA buffer (Fude Biological Technology, cat. no. FD011) and a protease inhibitor cocktail at 4 °C overnight. Then, the lysate was incubated with magnetic beads (Yeason, cat. no. 36417ES) that bind specific IP antibodies at 4 °C overnight (Supplementary Table 2). After incubation, the IP proteins in the complex were released by heat denaturation and detected by western blot or MS (Supplementary Table 9).

In vitro PAT assay

We mixed the recombinant ZDHHC11 (HUABIO) with the recombinant APOD (HUABIO) in 25 μl reaction buffer at 25 °C for 1 h. The mixture was incubated with biotin picolyl alkyne (50 μM) and loaded onto a 30-kDa spin column (Beyotime, cat. no. FUF053) to remove the free biotin picolyl azide. Finally, the biotin-labeled proteins were enriched by incubation with streptavidin magnetic beads (Beyotime, cat. no. P2151). After incubation, the palmitoyl proteins were released by heat denaturation and detected by western blotting.

Molecular docking

The molecular docking program HADDOCK (version 2.4), an information-driven flexible docking method for modeling biomolecular complexes, was used in this study^68,69. HADDOCK combines a variety of experimental and/or bioinformatics data to drive modeling, allowing for more complex processing of conformational flexibility to focus searches on relevant parts of the interaction space. PyMOL (version 3.0.3) software was used for result analysis and drawing operations.

Immunofluorescence

Chondrocytes were fixed in a 4% paraformaldehyde solution for 15 min, permeabilized with 0.1% Triton X-100 in PBS for 15 min, incubated with 5% goat serum for 1 h at room temperature and then incubated with primary antibodies (Supplementary Table 2) overnight at 4 °C. Then, the cells were incubated with DyLight 488 AffiniPure goat anti-rabbit secondary antibodies (Fdbio, cat. no. FD0136) or DyLight 594 AffiniPure goat anti-rabbit IgG secondary antibodies (Fdbio, cat. no. FD01296) for 1 h at room temperature. The nuclei were stained with DAPI (Fdbio, cat. no. FD9637) for 5 min. All images were acquired randomly using a fluorescence inverted microscope (Leica).

Histology and immunohistochemistry

Cartilage specimens were fixed in a 4% paraformaldehyde solution for 2 days, after which they were decalcified, paraffin-embedded and sectioned to a thickness of 5 μm. The sections were sequentially stained with 1% Fast Green solution (Sigma, cat. no. S7258) for 5 min, 1% acetic acid solution for 5 s and 1% Safranin O solution (Sigma, cat. no. S8884) for 10 min. For immunohistochemistry, the following steps were performed: (1) antigen retrieval with 0.1% trypsin (37 °C, 1 h); (2) endogenous peroxidase blocking (3% H₂O₂, room temperature, 10 min); (3) blocking with 5% goat serum (37 °C, 1 h); (4) primary antibody incubation (4 °C overnight; antibodies are listed in Supplementary Table 2); (5) incubation with HRP-conjugated secondary antibody (37 °C, 1 h); (6) DAB chromogenic development; and (7) hematoxylin counterstaining. All images were acquired randomly with a light microscope (Nikon).

Histological assessment

After Safranin O/Fast Green and H&E staining, cartilage sections from the medial tibial plateau were evaluated using the OARSI system⁷⁰ (Supplementary Table 7). Synovial sections were evaluated using the synovitis score⁷¹ (Supplementary Table 8). Each specimen was evaluated by five independent researchers who remained blinded to the experimental design.

SA-β-Gal staining

The treated cells were fixed in a 4% paraformaldehyde solution at room temperature for 30 min, followed by three washes with PBS. Then, the cells were incubated with SA-β-Gal solution (Beyotime, cat. no. C0602) at 37 °C overnight (pH 6.0), and images were randomly acquired using a microscope (Olympus, CX33TRF).

Alcian blue staining

The treated chondrocytes were fixed in a 4% paraformaldehyde solution at room temperature for 30 min. Then, the chondrocytes were incubated with a 1% Alcian blue solution (Sigma, cat. no. A5268) at room temperature for 1 h. Images were randomly acquired using a microscope (Olympus, CX33TRF). The original images of Alcian blue staining were then shifted using ImageJ (version 1.5.3) with the same parameters, followed by quantification.

Micro-CT

The knee joint specimens from mice were fixed in a 4% paraformaldehyde solution for 48 h, followed by washing and storage in 70% ethanol. Then, the samples were scanned using a high-resolution micro-CT scanner (Skyscan 1072, Skyscan) in 17-mm scanning tubes with an 11-mm³ volume at 180 mA, 50,000 V and an acquisition time of 5 min. After acquiring the CT images, a three-dimensional reconstruction was generated using Skyscan software (VGStudio MAX).

Von Frey assay

We used the Electronic Von Frey system (Ugo Basile, cat. no. 38450) to measure neuropathic pain in the OA model mice every 2 weeks⁷². First, we placed the mouse in a modular animal enclosure with a metal grid floor. We allowed the mouse to sit for half an hour before the experiment began. Then, a rigid stimulator filament in nitinol was used to successively stimulate the hind limbs of the animal. The value shown on the instrument when the hind limbs retreated rapidly was recorded. The test was repeated three times, and the mean value was calculated.

Statistics and reproducibility

The Olympus cellSens Standard software (version 1.18) was used for collecting images of histological staining. The Skyscan system (Skyscan 1072) was used for micro-CT scanning. The ABI 7500 system (version 2.0.6) was used to collect qPCR data. Positive cells in the images were analyzed using ImageJ (version 1.5.3). Statistical analysis was performed using GraphPad Prism software (version 8.0.1). Micro-CT three-dimensional reconstruction was performed using Skyscan software (VGStudio MAX). No statistical methods were used to predetermine sample sizes, but our sample sizes are based on our previous publications^73,74. The data distribution was assumed to be normal, but this was not formally tested. Data from two groups were analyzed using a two-sided Student’s t-test for RT–qPCR and the percentages of positive cells. A one-way analysis of variance (ANOVA) was used for multigroup comparisons. For nonparametric data, the Mann–Whitney U test was used for the OARSI grade and synovitis score. Data based on the clinical correlation of ZDHHC11 were obtained using Pearson correlation analysis. Parametric data are presented as means ± s.d., along with the calculated 95% confidence intervals (CIs) for nonparametric data. A difference of P < 0.05 between groups was considered statistically significant (*P < 0.05, **P < 0.01, ***P < 0.001). No randomization method was used. The investigators were blinded to group allocation during data collection, experimental procedures and analysis. No data were excluded from the analyses.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

S.S., M.J. and X.Z. designed the experiments. K.W. and Z.G. conducted most of the experiments. J.G., T.G., D.W. and Z.L. helped conduct the animal experiments. J.G., Y.Y. and N.P. helped collect the materials. S.S., M.J. and X.Z. supervised the experiments. K.W. and W.H. analyzed the results and wrote the paper. K.W., D.W. and W.H. revised the paper. S.S. acquired the funding.

Peer review

Peer review information

Nature Aging thanks Sundeep Khosla and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Data availability

All data needed to support the conclusions of the paper are included in the article and/or the supplementary materials. RNA-seq data have been deposited in the Sequence Read Archive (https://www.ncbi.nlm.nih.gov/sra) under accession number PRJNA1222457. Proteomics data have been deposited in the ProteomeXchange Consortium via the PRIDE partner repository (https://www.ebi.ac.uk/pride/archive) under accession numbers PXD063537, PXD063840, PXD063859 and PXD063864. Single-cell RNA-seq data (GSE169454) were obtained from the Gene Expression Omnibus database³⁰. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Extended data

is available for this paper at 10.1038/s43587-025-00968-1.

Supplementary information

The online version contains supplementary material available at 10.1038/s43587-025-00968-1.