Remodeling of senescent macrophages in synovium alleviates trauma- and aging-induced osteoarthritis

Abstract

Osteoarthritis (OA) is one of the most concerned aging-related diseases in the worldwide, yet the investigation of immune senescence in joint and related therapies are still poorly identified. Single-cell sequencing analysis and immunofluorescence of OA synovium reveal increased senescent macrophages in trauma-induced OA compared to controls. Importantly, senescent macrophages in OA synovium showed enhanced M1 polarization, mitochondrial damage and impaired efferocytosis, which could lead to increased senescence-associated secretory phenotypes (SASPs) in the joint and further exacerbate OA. Hence, a novel senotherapeutic nanoparticle is developed using chloroquine (CQ)-bearing polymers (pCQ) for targeted delivery of superoxide dismutase (SOD) to synovial macrophages, termed as pCQ/SOD. The nanoparticle achieves efficient intracellular delivery of SOD to synovial macrophages. RNA-seq results reveal that pCQ/SOD nanoparticle inhibits macrophage senescence via p53 and cellular senescence signaling pathway, further reprograms M1-to-M2 repolarization. Furthermore, the delivered SOD inhibits BAX-dependent mitochondrial outer membrane permeabilization (MOMP) which further reduces mitochondrial DNA (mtDNA) release and SASP secretion, while pCQ promotes macrophage efferocytosis against apoptotic cells via STAT3/ADAM17/MerTK signaling. As a result, intraarticular injection of pCQ/SOD nanoparticles in mice successfully alleviates not only trauma-induced OA, but aging-induced OA as well. The developed senotherapeutic nanoparticle in this study offers an effective approach for remodeling of senescent macrophages in synovium and a promising therapeutic strategy for OA treatment.

Graphical abstract

Highlights

• •In this study, we found senescent macrophages exist in OA synovium, which, importantly, show functional alterations such as enhanced SASP secretion, increased mitochondrial damage and impaired efferocytosis.
• •We for the first time constructed immuno-targeted therapy of synovial senescent macrophages for OA treatment via chloroquine-bearing polymers for delivery of superoxide dismutase.
• •The constructed therapy achieves therapeutic efficacy for treating not only trauma-induced OA, but aging-induced OA as well.

1. Introduction

Osteoarthritis (OA) is the most common aging-related disorder of joint, characterized by irreversible cartilage degeneration, synovitis and subchondral bone remodeling, affecting more than 654 million people worldwide [[1], [2], [3]]. However, current material-based, cell-based or drug-based treatments could not completely restore the structure and function of hyaline cartilage and effectively slow disease progression in OA treatment [[4], [5], [6]]. Thus, early-stage of OA is the window of opportunity for intervention due to the progressive course [7]. Immune homeostatic imbalance is one of the most concerned pathological changes in early-stage OA, in which synovial macrophages play a vital role in the occurrence and development of OA [8]. M1 macrophages secret proinflammatory cytokines, such as interleukin 1 (IL1), IL6, tumor necrosis factor-α (TNFα), to amplify inflammation and cartilage erosion, while M2 macrophages exhibit anti-inflammatory property via secretion of cytokines for tissue healing, such as IL10 [9]. Modulation of M1/M2 polarization is the hotspot in the field of OA intervention, but still compromised by unsatisfactory efficiency [10]. Thus, deeper understanding of the mechanisms underlying macrophage functions in OA is urgently needed to be elucidated.

Immunosenescence is closely correlated with aging-related diseases [11],characterized by inflammaging, impaired new antigen response, stress responses, and accumulation of senescent cells [11,12]. One of the hallmarks of immunosenescence is the systemic state of chronic, low-grade inflammation characterized by upregulated pro-inflammatory cytokines [13], which is also a typical feature of OA [14]. Previous studies revealed that macrophage senescence, characterized by high expression of senescent markers, such as Cdkn2a and Cdkn1b [15,16], results in enhanced inflammation via secretion of senescence-associated secretory phenotype (SASP) [[17], [18], [19]]. This implies senescent macrophages tend to perform as pro-inflammatory phenotype [20]. However, the relationship of macrophage senescence and OA is still largely elusive, and precisely remodeling of senescent immune cells remains a challenge in clinic [21].

OA progression is associated with reactive oxygen species (ROS) [22], produced by either activated macrophages or chondrocytes under inflammatory conditions [23,24], which could lead to chondrocyte senescence and enhanced p21 expression of macrophage [25,26], indicating ROS as a senescence-inducer in OA microenvironment. Therefore, reducing the level of ROS in macrophages is expected to rescue macrophage senescence. Superoxide dismutase (SOD) has long been used as ROS-scavenger, which can perform a cascade reaction removing both superoxide anion and hydrogen peroxide [21,27]. As a result, SOD could protect the oxidative damage and delay the aging of organism [28]. Deletion of SOD1 leads to increased level of p16, p21 and SASPs, indicating the anti-inflammatory and anti-aging property of SOD [29]. However, SOD, as a protein with characteristics of instability and membrane impermeability, could not directly enter macrophages to clear intracellular ROS. Consequently, intracellular protein delivery systems are of great significance for alleviating macrophage senescence by SOD in OA treatment [30].

For intracellular protein delivery, a variety of strategies for constructing polymer carriers have been proposed in our previous work [[31], [32], [33]]. Polymers with appropriate positive charge and hydrophobicity balance are expected to achieve efficient intracellular delivery of proteins. Chloroquine (CQ) is well-known as an anti-malarial agent, structurally characterized by protonable amine groups and hydrophobic quinoline rings, which could serve as a building block for delivery vehicles. Moreover, as a bioactive substance, CQ is closely related to aging and inflammation, since low-dose CQ has recently been proved to prolong lifespan and repress inflammation in aged rats [34]. The usage of CQ could also significantly decrease the induction of β-galactosidase [34], indicating the anti-aging property of CQ. Besides, CQ is one of the DMARDs for the treatment of rheumatoid arthritis via inhibition of production of pro-inflammatory cytokines, such as IL-1, IFNα and TNF [35], suggesting the anti-inflammatory property of CQ. Mechanistically, previous study identified hydroxychloroquine as an inactivator of STAT3 [36], which is a critical factor inducing the onset of senescence [37] and the impairment of macrophage efferocytosis [38]. Hence, CQ is a feasible anti-aging and anti-inflammatory material for intracellular delivery of functional proteins.

Herein, we firstly identified macrophage senescence in OA via single-cell sequencing analysis of GEO dataset [39] and co-immunofluorescence. We found significantly up-regulated expression of senescence-related biomarkers, such as p53, Cdkn1b, Rb1 and β-galactosidase, and reduced expression of anti-aging gene, Lmnb1 in OA synovial macrophages compared to controls [40,41]. Importantly, OA senescent macrophages showed enhanced M1 polarization, increased mitochondrial damage and impaired efferocytosis, which are critical for OA exacerbation. Based on this, remodeling of macrophage senescence in OA is expected to become a novel therapeutic strategy. We further synthesized CQ-bearing polymers (pCQ) as a carrier to achieve efficient intracellular delivery of SOD (pCQ/SOD) into macrophages and relieved macrophage senescence. Mechanistically, we observed pCQ/SOD could not only inhibit BAX-dependent mitochondrial outer membrane permeabilization (MOMP) as ROS scavenger, but could enhance macrophage efferocytosis via STAT3/ADAM17/MerTK signaling, both benefit for the rescue of macrophage senescence in local microenvironment [42,43]. Amazingly, the prepared senotherapeutic nanoparticle could not only reprogram M1-to-M2 repolarization via inhibiting macrophage senescence by p53 and cellular senescence signaling, but also alleviate both trauma-induced OA and aging-induced OA via intraarticular injection. The developed pCQ-based nanoparticles in this study prove the efficacy and importance of targeting macrophage senescence in OA (Fig. 1).

Fig. 1.

Schematic illustration of the mechanisms of senotherapeutic nanoparticle in the treatment of OA. Based on the evidence that synovial macrophages in OA showed an increased expression of senescent biomarkers and functional decline, senotherapeutic nanoparticles containing pCQ and SOD termed as pCQ/SOD were prepared. Functionally, pCQ/SOD led to M1-to-M2 repolarization, reduced ROS-mediated mtDNA release and SASPs secretion to block cellular senescence, and rescued the impaired efferocytosis against apoptotic fibroblast-like synoviocytes (FLS) via STAT3/ADAM17/MerTK signaling to remodel local aging microenvironment. As a result, pCQ/SOD nanoparticles efficiently inhibited macrophage senescence and relieved aging-induced and trauma-induced OA in vivo.

2. Results

2.1. OA synovial macrophages exhibit increased senescent phenotypes

The senescence of synovial macrophages in OA synovial tissue is our first concern. We firstly examined a typical senescent biomarker, p53, in macrophages of human synovium from OA patients and controls. Co-immunofluorescence (IF) of p53 (red) and macrophage marker, CD68 (green) revealed increased senescent synovial macrophages in OA compared to the controls (Fig. 2a), indicated by significantly increased mean IF intensity of p53 (Fig. 2c). To better confirm the result, we then constructed destabilization of the medial meniscus (DMM) surgery-induced OA in mice and also found significantly enhanced p53 expression in CD68 positive synovial macrophages in OA group compared to sham group (Fig. 2b and d), suggesting macrophage senescence exists in OA synovium. We further analyzed single-cell sequencing data of murine synovium in the NCBI dataset [39], in which clusters of synovial macrophages were identified in sham group and anterior cruciate ligament rupture-induced OA (ACLR-OA) group. As expected, we found the expression of a series of senescent biomarkers increased in macrophages of ACLR-OA group compared to sham group, such as Cdkn1a (p21), Trp53 (p53), Rb1, and Glb1 (β-galactosidase), while the expression of Lmnb1, negatively correlated with senescence, decreased in ACLR-OA group [[44], [45], [46], [47]] (Fig. 2e and f). Hence, we confirmed that senescent macrophages exist in OA synovium.

Fig. 2.

Examination of macrophage senescence in OA synovium. (a) Confocal laser scanning images of co-immunofluoresence of DAPI (blue), CD68 (green) and p53 (red) in synovium of OA patients (n = 7) and controls (n = 3). (Scale bar, 20 μm) (b) Confocal laser scanning images of co-immunofluoresence of DAPI (blue), CD68 (green) and p53 (red) in synovium of OA mice induced by DMM surgery and sham groups (n = 4). (Scale bar, 20 μm) (c, d) Quantification of mean IF intensity of p53 in OA synovium and control from human and mice. (e) Expression of Lmnb1, Rb1, Trp53, Cdkn1a, and Glb1 in clusters of murine synovial macrophages in sham group and ACLR-OA group. (f) UMAP plot of murine synovial macrophages in sham group and ACLR-OA group, and feature plots of Cdkn1a. (g) Senescent macrophages exist in OA synovial tissues and exhibit functional decline. (h) Gene expression correlated with macrophage functions in OA between senescent and non-senescent macrophages. Data are shown as the mean ± s.d. *p < 0.05, **p < 0.01, ***p < 0.001, between the indicated groups. p values were determined using Student's t tests.

To investigate the mechanisms of OA senescent macrophages in disease progression, we specifically analyzed functional alterations between OA senescent macrophages and non-senescent macrophages. Senescent macrophage cluster was distinguished from non-senescent macrophages in OA synovium via enhanced expression of senescence-related genes, such as p21, Trp53, Rb1, Glb1 (Fig. 2h). SASP-related genes, such as Csf2ra, Tnfrsf1a, Tnfrsf11a, Tnfrsf1b, Igfbp7, S100a6 [48], significantly increased in senescent macrophages, representing elevated SASP secretion of OA synovial senescent macrophages. We then investigated whether senescence drives functional decline which are closely related to OA progression, such as macrophage polarization, mitochondrial reprogramming, and efferocytosis [8,10,49]. As shown, M1-related gene, TLR4, increased in senescent macrophages, while M2-related gene, CD163, decreased in senescent macrophages compared to non-senescent macrophages, implying macrophage senescence might lead to M2-to-M1 repolarization which could worsen OA symptoms [50,51] (Fig. 2h). Besides, we found increased expression of BAX, BAK1, Fis1 and Opa1 in senescent macrophages compared to non-senescent macrophages [[52], [53], [54]], indicating senescence drives mitochondrial reprogramming which might further increase SASP secretion to accelerate cartilage erosion (Fig. 2h). Importantly, we focused on efferocytosis in senescent macrophages, which is important for maintenance of local microenvironment via elimination of SASP-secreting apoptotic cells [55]. CD47, Jak1, Stat3, and Adam17 significantly increased in senescent macrophages compared with non-senescent macrophages in OA synovium, indicating the efferocytosis of senescent macrophages is impaired [38,56] (Fig. 2h). To conclude, senescent macrophages secreting SASP exist in OA synovium which tend to polarize to M1 state, exhibit properties of mitochondrial damage and impaired efferocytosis, further leading to exacerbation of osteoarthritis (Fig. 2g).

Since synovial macrophages are more likely to polarize to M1 state, the relationship between M1 macrophages and senescent macrophages in OA synovium needs to be identified. We found that M1 macrophages in murine OA synovium partly showed p53 expression via co-immunofluorescence (Fig. S1a), however, successful induction of M1 polarization via lipopolysaccharide (LPS) of Raw264.7 cells could not drive cellular senescence indicated by the transcriptive levels of p21 and p53 (Fig. S1b). These results implied although some of OA M1 macrophages exhibit senescent phenotypes, these cells are potentially polarized from senescent macrophages.

2.2. Construction of pCQ for intracellular delivery of SOD to macrophages

Motivated by the above results, we next aimed to construct a senotherapeutic approach for remodeling senescent macrophages for OA treatment. Superoxide dismutase (SOD) was reported to scavenge ROS, benefit for inhibiting senescence and repair ROS-induced mitochondrial damage [27]. What's more, SOD also contributes to M1-to-M2 repolarization [57], which is promising for senotherapeutic strategy. To obtain the carrier for efficient intracellular delivery of SOD to macrophages, we synthesized a novel cationic copolymer, pCQ, by Reversible Addition-Fragmentation Chain Transfer (RAFT) polymerization from 2-(Diethylamino)ethyl methacrylate (DEMEMA) and a choloroquine (CQ)-bearing monomer (MACQ) (Fig. 3a, S2a), for CQ shows anti-aging property and also enhances phagocytotic ability [58], which is expected for repairment of efferocytosis. The synthesis and characterization of pCQ are shown in Fig. S2b. To confirm whether the grafting of CQ onto the polymer chains would affect the release behavior, we configured an acidic aqueous solution of deuterium (pH = 4), the polymers were dissolved and characterized by ¹H NMR (500 MHz) after 0 h, 2 h, 4 h and 10 h. As shown in Fig. S3, no significant degradation of pCQ was observed during our detection, and no signal of small molecules was observed in the spectrum. In previous studies, CQ-containing polymers have been used as bioactive polymers to achieve the biological functions of small molecules CQ [59,60], indicating CQ obtains bioactivity without the need of releasing from polymers.

Fig. 3.

pCQ for SOD delivery into OA synovial macrophages. (a) pCQ enables targeted intracellular delivery of SOD into synovial macrophages. (b) EGFP, pCQ/EGFP, SOD-Cy5.5 and pCQ/SOD-Cy5.5 were incubated with Raw264.7 cells for 12 h, respectively. The representative images of confocal laser scanning of EGFP (green) and SOD-Cy5.5 (red) in Raw264.7 cells were shown. (Scale bar, 25 μm) (c) Mean fluorescence intensity of Raw264.7 cells treated with EGFP, pCQ/EGFP, SOD-Cy5.5 and pCQ/SOD-Cy5.5 for 12 h, respectively. The concentrations of SOD and pCQ were 20 and 32 μg/mL, respectively. (d) pCQ/SOD-Cy5.5 was intraarticularly injected in the joints of 4-mon-old male mice and specimens were obtained 12 h after injection (n = 6). Representative images of confocal laser scanning of CD68 (green) and SOD-Cy5.5 (red) of joint synovium and cartilage were shown. (e) Representative images of confocal laser scanning of Vimentin (yellow) and SOD-Cy5.5 (red) of joint synovium were shown. Scale bar, 25 μm. Data are shown as means ± SD. *p < 0.05, **p < 0.01, ***p < 0.001. p values were determined using Student's t tests.

We then tested the intracellular delivery efficiency of pCQ on Raw264.7 cells using enhanced green fluorescent protein (EGFP) as a model protein and labeled SOD with Cyanine5.5 (SOD-Cy5.5) to track intracellular proteins. As shown in Fig. 3b and c, mean fluorescence intensity of Raw264.7 cells treated with pCQ/EGFP complexes were significantly increased compared with EGFP only, pCQ/SOD-Cy5.5 also exhibited significantly enhanced cell uptake on Raw264.7 cells compared with SOD-Cy5.5 only. We characterized the complexes formed by pCQ and EGFP or SOD using dynamic light scattering (DLS) and transmission electron microscopy (TEM), as shown in Fig S4b, d and e. The complexes are solid, dense nanoparticles with fluid dynamics diameters of about 350 nm and 140 nm for pCQ/EGFP and pCQ/SOD, respectively. The zeta potential of the pCQ/EGFP and pCQ/SOD nanoparticles is 8.3 mV and 9.9 mV, respectively (Fig. S4c and f). The protein loading efficiency (PLE) and protein loading content (PLC) of nanoparticles were about 30 % and 20 %, respectively (Fig. S5a). We further evaluated the stability and protein release of the nanoparticles in PBS for different times. The protein rapidly released nearly 60 % within 1 h and slowly released until complete within the following 8 h, which may be correlated with the trend of particle size changes (Fig. S5b and c).

Moreover, we noticed that there are many clustered fluorescent spots in Fig. 3b. We would like to further confirm whether these fluorescent spots are from proteins trapped in lysosomes. As shown in Fig. S6, we treated Raw264.7 cells with pCQ/SOD-Cy5.5 and pCQ/EGFP for 12 h respectively, labeled lysosomes with Lyso Tracker Red, and performed colocalization statistics by Image J. The trend of fluorescence intensity changes in the red and green channels is not completely consistent along the same path (Fig. S6a–b). Pearson's correlation coefficient (PCC) of both pCQ/SOD-Cy5.5 and pCQ/EGFP was below 0.5, which is sufficient to prove that intracellular proteins are not retained in lysosomes (Fig. S6d). In addition, we analyzed the endosomal escape ability of pCQ/protein nanoparticles via a Gal8-YFP recruitment assay. Gal8-YFP recruitment resulted in punctate yellow fluorescent spots on the inner luminal side of disrupted endosomal membranes [61]. The endosomal disruption by pCQ/protein nanoparticles could be evaluated according to the area of yellow spots. As shown in Fig. S6e–f, compared with model protein bovine serum albumin (BSA), cells treated with pCQ/BSA for 8 h showed significant endosomal membranes disruption, indicating that pCQ has strong endosomal escape ability.

We then investigated synovial macrophage-targeted ability of pCQ/SOD in vivo via intraarticular injection. As shown in Fig. 3d, SOD-Cy5.5 (red) was effectively delivered into CD68 positive (green) macrophages in synovium, while were merely found in cartilage, indicating the excellent targeted delivery efficacy of pCQ/SOD into synovial macrophages rather than into chondrocytes. Considering the abundance of both macrophages and synovial fibroblasts within the synovium, we nextly examined the uptake of pCQ/SOD by synovial fibroblasts and found that Vimentin positive synovial fibroblasts showed little uptake of pCQ/SOD (Fig. 3e). For detection of cellular uptake of pCQ/SOD by different subtypes of macrophages, we pretreated Raw264.7 cells with 100 ng/mL LPS, 1000 U/mL IL4, 10 g/L D-galactose (D-gal) and 800 μM hydrogen peroxide (H₂O₂), respectively in vitro for 24 h to induce M1, M2 and senescent macrophages, and then treated cells with pCQ/SOD-Cy5.5 for another 12 h. As shown in Fig. S7a–b, cellular uptake did not differ significantly between different subtypes of macrophages, and uptake by H₂O₂-induced senescent macrophages was even slightly higher than that by M0 macrophages. Based on these results, we speculate that the selectivity of pCQ/SOD may primarily attributable to the inherently strong phagocytic capacity of macrophages. Further mechanistic study revealed that the uptake of pCQ/SOD-Cy5.5 should be mediated by caveolae-mediated and macropinocytosis endocytosis pathways (Fig. S7c).

2.3. pCQ/SOD inhibits macrophage senescence and reprogramms M1-to-M2 polarization

To assess the role of pCQ/SOD in relieving macrophage senescence, we constructed in vitro models for induction of macrophage senescence. Firstly, we pretreated Raw264.7 cells with 10 g/L D-gal in vitro for 24 h to induce macrophage senescence [62], and then treated cells with pCQ/SOD for another 24 h. Treatments with pCQ or SOD were established as controls and no treatments in cells was set as blank. Macrophage senescence was successfully induced via D-gal stimulation, and pCQ/SOD significantly inhibited the senescent phenotypes of macrophages with better effects than other controls, indicated by β-gal staining and p53 mRNA level (Fig. 4a–c). Previous studies proved that senescent macrophages, identified by enhanced p16 and p21 expression, showed a proinflammatory phenotype which indicates macrophage senescence drives M1 polarization rather than M2 polarization [63]. To further assess whether pCQ/SOD could modulate macrophage polarization via relieving macrophage senescence, we pretreated senescent macrophages (induced by D-gal) with pCQ, SOD or pCQ/SOD, and further induced M1 polarization via lipopolysaccharide (LPS) or induced M2 polarization via IL4. We observed that senescent macrophages tended to polarize to M1 state other than M2 state, while pCQ/SOD treatment could efficiently limit M1 polarization and enhance M2 polarization of senescent macrophages, indicated by immunofluorescence and transcriptive levels of iNOS (M1 marker) and CD206 (M2 marker) (Fig. 4d–i). To better confirm this result, flow cytometry was performed for detection of CD86 positive M1 macrophages or CD206 positive M2 macrophages. As expect, pCQ/SOD treatment significantly reduced the amount of CD86 positive cells and increased CD206 positive cells in senescent macrophages (Fig. 4j), suggesting senescence inhibition via pCQ/SOD contributes to reprogramming of M1-to-M2 repolarization (Fig. 4m).

Fig. 4.

The effects of pCQ/SOD nanoparticles on macrophage senescence and polarization in vitro. (a) Representative images of Raw264.7 cells treated with pCQ, SOD or pCQ/SOD, respectively after 10 g/L D-gal induction for 24 h (Scale bar, 250 μm) (b) Relative quantifications of β-gal staining. (c) Transcriptive levels of p53 of Raw264.7 cells measured via qPCR. (d) Raw264.7 cells were treated with pCQ, SOD or pCQ/SOD, respectively after 10 g/L D-gal induction for 24 h, and then induced to M1-polarized macrophages via 100 ng/mL LPS. Representative images of iNOS (red) were shown. (Scale bar, 100 μm). (e) Relative quantifications of IF intensity of iNOS. (f) Transcriptive levels of iNOS of Raw264.7 cells measured via qPCR. (g) Raw264.7 cells were treated with pCQ, SOD or pCQ/SOD, respectively after 10 g/L D-gal induction for 24 h, and then induced to M2-polarized macrophages via 1000 U/mL IL4. Representative images of CD206 (green) were shown. (Scale bar, 100 μm). (h) Relative quantifications of IF intensity of CD206. (i) Transcriptive levels of CD206 of Raw264.7 cells measured via qPCR. (j) Flow cytometry and related quantification of CD86 and CD206 of macrophages. (k) KEGG enrichment of differentially expressed genes. (l) Heatmap of genes correlated with p53 signaling pathway and cellular senescence, and BAX-related genes in RNA-seq. (m) Schematic illustration of the regulatory effect of pCQ/SOD on polarization of senescent macrophages. Data are shown as means ± SD. *p < 0.05, **p < 0.01, ***p < 0.001. p values were determined using ANOVA tests. The significant differences are determined by comparison with the senescence induction group.

We also constructed a H₂O₂-induced senescence model to verify the universal anti-aging property of pCQ/SOD [64]. We pretreated Raw264.7 cells with 800 μM H₂O₂ in vitro for 24 h, and then treated cells with pCQ/SOD for another 24 h. The control groups were set up as that in the D-gal-induced model. As expected, pCQ/SOD could also efficiently rescue the H₂O₂-induced senescence and reprogram M1 to M2 macrophages after senescence induction (Fig. S8). Considering the pro-inflammatory OA microenvironment might influence the reprogramming of M2 macrophages via pCQ/SOD, we then tested the stability of the M2 phenotype when exposed to LPS. Raw264.7 cells were treated with IL-4 (1000 U/mL) for 24 h to induce M2 polarization. M2-polarized macrophages were then treated with or without LPS (100 ng/mL) for 24 h. Results showed that treatment with LPS in M2 macrophages did not significantly decrease the CD206 level compared to M2 macrophages with no treatment (Fig. S8h), indicating the stability of M2 phenotype when exposed to LPS and the efficacy of M2 induction by pCQ/SOD in OA microenvironment.

Based on the anti-aging activity of pCQ/SOD on macrophages, we further investigated the underlying mechanisms. We performed RNA-sequencing of aged bone-marrow derived macrophages (aged BMDMs, distracted from 18-month-old mice) treated with or without pCQ/SOD for 24 h. We observed 2200 up-regulated genes and 1667 down-regulated genes after treatment with pCQ/SOD (Fig. S9a). GO enrichment represented a series of significant changes related to immunomodulation, such as innate immune response, cytokine production (in which SASPs such as TNF, Il1a, Il1b are differentially expressed), and immune response, implying pCQ/SOD contributed to the immune regulation of aged macrophages (Fig. S9b). Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment revealed that p53 signaling pathway and cellular senescence pathway were significantly influenced by treatment of pCQ/SOD, indicating pCQ/SOD modulated senescence-related pathways of macrophages (Fig. 4k). We further investigated differentially expressed genes of p53 signaling pathway and cellular senescence pathway in RNA-seq. Aging-related genes (Ccnb2, Serpine1, Ccnd1, Ccne2, Chek1, Cdk1) greatly decreased in aged BMDMs treated with pCQ/SOD compared to control, while anti-aging genes (Gadd45b, and Mdm2) significantly increased in BMDMs after pCQ/SOD treatment (Fig. 4l). qPCR of related genes was tested to confirm the results (Fig. S9c). In addition, we analyzed the modulation of macrophage polarization by pCQ/SOD via Gene Set Enrichment Analysis (GSEA) of specific pathways. M2 polarization-related pathways, including Toll-like receptor signaling pathway, PPAR signaling pathway, NOD-like receptor signaling pathway, are more expressed in pCQ/SOD-treated BMDMs compared to control, while M1 polarization-related pathway, such as mTOR signaling pathway, showed less expression in pCQ/SOD-treated BMDMs (Fig. S9d). These results confirmed that pCQ/SOD contributes to rejuvenation of macrophage senescence and consequently modulated M1-to-M2 repolarization (Fig. 4m).

2.4. pCQ/SOD inhibits MOMP and enhances macrophage efferocytosis against senescence

In order to investigate pCQ/SOD-mediated alterations in senescent macrophages, we specifically focus on mitochondrial function, which are impaired in OA senescent macrophages and most likely to be influenced by SOD in ROS-mediated cellular senescence [65]. Widespread MOMP has been proved as a feature of cellular senescence caused by ROS, which calls for the participation of BAX macropores for the release of mtDNA, followed by upregulation of SASPs and exacerbation of cellular senescence [48,66]. Amazingly, we found that the transcriptive level of Transmembrane BAX Inhibitor Motif (Tmbim) proteins, Tmbim4 and Tmbim6, significantly increased after treatment of pCQ/SOD of aged BMDMs in RNA-seq results, which represented that pCQ/SOD might decrease BAX expression [67] (Fig. 4l). We further confirmed the results via qPCR and found that pCQ/SOD could exactly decrease the expression of BAX (Fig. S9c). These results indicated excellent anti-aging property of pCQ/SOD, which might be regulated via decreased BAX-dependent MOMP against oxidative stress of ROS. As shown in Fig. 5a–c and Fig. S10a–c, ROS detection and flow cytometry indicated that pCQ/SOD showed increased ROS scavenging activities compared to free pCQ or SOD after senescence induction via D-gal or H₂O₂. As expected, pCQ/SOD also rescued the significant decrease in mitochondrial membrane potential (mΔψ) caused by senescence, as evidenced by decreased monomers (green) and increased aggregates (red) of JC-1 staining, which represented pCQ/SOD decreased MOMP in senescent macrophages compared to controls [66] (Fig. 5d, e and Fig. S10a and d). Taken together, pCQ/SOD serves as a ROS scavenger and contributes to inhibit BAX-dependent MOMP, which limits mtDNA release and senescent phenotypes of macrophages.

Fig. 5.

The mechanisms of pCQ/SOD modulating macrophage senescence. (a–c) Representative images, with and without bright field (BF), and relative quantifications of Dichlorodihydrofluorescein (DCFH) detection for ROS production (Scale bar, 100 μm) of Raw264.7 cells which were treated with pCQ, SOD or pCQ/SOD respectively after 10 g/L D-gal induction for 24 h. The significant differences are determined by comparison with the D-gal treated group. (d–e) Representative images and relative quantifications of JC-1 staining (Scale bar, 10 μm) of Raw264.7 cells which were treated with pCQ, SOD or pCQ/SOD respectively after 10 g/L D-gal induction for 24 h. The significant differences are determined by comparison with the D-gal treated group. (f) Transcriptive level of MerTK was measured via qPCR. (g–i) Representative images and quantifications of staining (Scale bar, 20 μm) and flow cytometry of co-culture of AF647-labeled NIH-3T3 cells (red) and CFSE-labeled Raw264.7 cells (green). The significant differences are determined by comparison with the D-gal treated group. (j) Protein levels of p-STAT3, t-STAT3, ADAM17 and β-actin of Raw264.7 cells pretreated with or without D-gal in different conditions. (k, l) Quantifications of p-STAT3/t-STAT3 and ADAM17/β-actin protein levels. The significant differences are determined by comparison with the D-gal treated group. (m) Schematic illustration of the mechanisms of pCQ/SOD modulating macrophage senescence via regulating MOMP and inducing efferocytosis. Data are shown as means ± SD. *p < 0.05, **p < 0.01, ***p < 0.001. p values were determined using ANOVA tests.

In addition to the oxidative stress triggered by ROS and mitochondrial alterations, OA senescent macrophages also exhibit significant impairment in macrophage efferocytosis. This could lead to the accumulation of apoptotic cells which secret more SASPs, further exacerbates cellular senescence in turn and worsen OA symptoms [18,43]. Hence, enhancing macrophage efferocytosis is an effective approach to relieve macrophage senescence via remodel local aging microenvironment. qPCR examination revealed that pCQ/SOD significantly upregulated the transcriptive level of Mer tyrosine kinase receptor (MerTK) in senescent macrophages, which is a typical efferocytosis-related receptor of macrophage [68] (Fig. 5f). To further evaluate the efferocytosis of senescent macrophages, we pretreated fibroblasts (NIH-3T3 cells) and macrophages (Raw264.7 cells) with 10 g/L D-gal (Fig. 5g) or 800 μM H₂O₂ (Fig. S11) for 24 h, respectively [69,70]. The apoptosis of NIH-3T3 cells was identified via Terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) assay performed by flow cytometry (Fig. S11d and e). After treatment with pCQ/SOD for another 24 h, CFSE-labeled Raw264.7 cells (green) were co-cultured with AF647-labeled NIH-3T3 cells (red) in a ratio of 1:1 for detection of efferocytosis. We observed a decreased proportion of fibroblasts that were localized within senescent macrophages compared to control, while pCQ or pCQ/SOD treatment significantly enhanced the co-localization of fibroblasts and macrophages, indicating that senescent macrophages engulfed more fibroblasts after pCQ/SOD treatment (Fig. 5g, h, S11). We also assessed the proportion of fibroblasts in macrophages using flow cytometry. Our results confirmed that pCQ or pCQ/SOD could greatly enhance efferocytosis of senescent macrophages, indicated by the increased proportion of FITC-, AF647- double positive cells in AF647 positive cells (Fig. 5g–i, S11).

Based on the above results, pCQ or pCQ/SOD could greatly promote macrophage efferocytosis by upregulating the expression of MerTK in senescent macrophages. In this context, we further explored the mechanisms underlying the upregulation of MerTK expression. We noticed one of the efferocytosis-related signals, JAK-STAT signaling pathway, was enriched in the KEGG enrichment of RNA-seq (Fig. 4k), which was proved negatively regulated by CQ and negatively correlated with macrophage efferocytosis in previous studies [36,38]. Consistently, protein level of p-STAT3 was significantly decreased while treated with CQ, pCQ and pCQ/SOD in Raw264.7 cells pretreated with D-gal, compared to D-gal-induced senescent group (Fig. 5j and k). Previous reports have indicated that the phosphorylation of STAT3 promotes the expression of ADAM17. Consequently, ADAM17 enhances the cleavage of MerTK and inhibits macrophage efferocytosis [38,71]. As expected, the protein level of ADAM17 was significantly decreased by CQ, pCQ or pCQ/SOD, indicating pCQ/SOD increased macrophage efferocytosis via STAT3/ADAM17/MerTK signaling (Fig. 5j–l). To conclude, pCQ/SOD exhibited excellent abilities of remodeling senescent macrophages via reducing BAX-dependent MOMP and improvement of efferocytosis (Fig. 5m).

2.5. Intraarticular injection of pCQ/SOD alleviates DMM-induced OA and ACLT-induced OA

Building on the above in vitro results, we performed an intraarticular injection of pCQ/SOD to evaluate the therapeutic potential via targeting macrophage senescence for OA treatment. In vivo assessment of retention of pCQ/SOD was tested via examination of Cy5.5 which was labeled on SOD after 3, 5, 7 d intraarticular injection of pCQ/SOD-Cy5.5 in mice. Results showed that at day 7 nearly all of the pCQ/SOD (labeled with Cy5.5) could not be detected, indicating 7 days as the frequency of intraarticular treatment of pCQ/SOD (Fig. S12a). In vivo safety was confirmed via HE staining of liver and kidney of mice 7 days after intraarticular injection with or without pCQ/SOD (Fig. S12b). We then established DMM surgery of 3-mon-old mice, and injected pCQ/SOD into murine joints once per week from the 3rd day after surgery for 4 weeks. Sham group, injection with pCQ, or SOD respectively after DMM surgery were established as controls (Fig. 6a). pCQ/SOD treatment efficiently reduced cartilage erosion compared to DMM group, indicated by more safranine O area, less matrix metalloproteinase 13 (MMP13) positive cells and more aggrecan (ACAN) positive cells, which exhibited better therapeutic effects than treatment with pCQ, or SOD (Fig. 6b–e). Furthermore, we constructed OA model via anterior cruciate ligament transection (ACLT) surgery of 3-mon-old mice, and injected pCQ/SOD into murine joints once per week from the 3rd day after surgery for 4 weeks. Sham group, injection with pCQ, or SOD respectively after ACLT surgery were established as controls (Fig. 6f). Consistent with DMM model, pCQ/SOD treatment also showed significant efficacy of treating ACLT-induced OA (Fig. 6g–j), suggesting that the proposed senotherapeutic strategy is effective in both DMM and ACLT-induced OA.

Fig. 6.

Treatment of DMM or ACLT-induced OA with pCQ/SOD. (a) DMM surgery was established in 3-mon-old mice. 3 days after DMM surgery, pCQ, SOD or pCQ/SOD were intraarticularly injected respectively weekly and sacrificed after 4 weeks for joint specimens. (n = 6) (b) Representative images of safranine O staining (Scale bar, 50 μm), immunohistochemistry of MMP13 and ACAN (Scale bar, 20 μm) of joint cartilage. (c) Relative quantifications of Osteoarthritis Research Society International (OARSI) scores. (d) Relative quantifications of percentage of MMP13 positive cells. (e) Relative quantifications of percentage of ACAN positive cells. (f) ACLT surgery was established in 3-mon-old mice. 3 days after ACLT surgery, pCQ, SOD and pCQ/SOD were intraarticularly injected respectively weekly and sacrificed after 4 weeks for joint specimens. (g) Representative images of safranine O staining (Scale bar, 50 μm), immunohistochemistry of MMP13 and ACAN (Scale bar, 20 μm) of joint cartilage. (h) Relative quantifications of OARSI scores. (i) Relative quantifications of percentage of MMP13 positive cells. (j) Relative quantifications of percentage of ACAN positive cells. Data are shown as means ± SD. *p < 0.05, **p < 0.01, ***p < 0.001. p values were determined using ANOVA tests. The significant differences are determined by comparison with the DMM/ACLT group.

To better confirm the effects of pCQ/SOD in vivo, we further examined macrophage polarization in murine synovium in these groups. Our findings revealed that intraarticular injection of pCQ/SOD led to a reduction in the number of iNOS positive cells and an increase in CD206 positive cells both in the DMM and ACLT models (Fig. S13a and b). This observation clearly demonstrates the outstanding immunomodulatory capacity of pCQ/SOD. Moreover, pCQ/SOD treatment greatly reduced apoptotic cells in DMM or ACLT-induced OA, indicated by decreased c-cas3 positive cells in synovium (Fig. S13c). Co-immunofluorescence of CD68 positive macrophages and p53 positive senescent cells indicated that pCQ/SOD acts as an effective senotherapeutic method which could, to a great extent, decrease macrophage senescence in DMM or ACLT-induced OA in vivo (Fig. S14a and b). As confirmed, it is proved that targeting macrophage senescence via pCQ/SOD exactly made contribution to the relief of trauma-induced OA symptoms.

2.6. Remodeling of synovial senescent macrophages by pCQ/SOD contributes to relief of aging-induced OA symptoms

Given the demonstrated efficacy of pCQ/SOD in trauma-induced OA, we further examined whether pCQ/SOD could also relieve aging-related OA via rescuing macrophage senescence. pCQ/SOD was intraarticularly injected weekly into 18-mon-old mice (aged mice) for 6 weeks. 3-mon-old mice (young mice), and injection with pCQ or SOD respectively in aged mice were established as controls (Fig. 7a). As expected, cartilage erosion of aged mice was much more serious than young mice, indicated by decreased safranine O area, more MMP13 positive cells and less ACAN positive cells (Fig. 7b). In addition, aged mice showed more iNOS positive cells and less CD206 positive cells in joint synovium, indicating macrophage senescence spontaneously results in enhanced M2-to-M1 polarization, which is consistent with exacerbated OA symptoms in aged mice (Fig. 7b). Aged mice treated with pCQ, SOD or pCQ/SOD all exhibited varying degrees of alleviation of aging-induced OA symptoms, among which pCQ/SOD showed the best efficacy, indicated by significant decrease of OARSI scores, decreased MMP13 positive cells, increased ACAN positive cells, and enhanced M1-to-M2 repolarization (Fig. 7b–g).

Fig. 7.

Treatment of aging-induced OA with pCQ/SOD. (a) 18-mon-old mice were intraarticularly injected with pCQ, SOD or pCQ/SOD respectively weekly and sacrificed after 6 weeks for joint specimens. 3-mon-old mice were established as young control. (n = 6) (b) Representative images of safranine O staining (Scale bar, 50 μm), immunohistochemistry of MMP13 and ACAN (Scale bar, 20 μm) of joint cartilage, and confocal laser scanning images of iNOS (red) and CD206 (green) of joint synovium (Scale bar, 20 μm). (c–g) Relative quantifications of OARSI scores, percentage of MMP13 positive and ACAN positive cells, mean IF intensity of iNOS expression and CD206 expression. (h) Confocal laser scanning images of co-immunofluoresence of DAPI (blue), CD68 (green) and p53 (red) in synovium of aged mice and aged mice treated with pCQ/SOD (Scale bar, 20 μm). (i) Relative quantifications of mean IF intensity of p53 expression and P values were determined using Student's t tests. Data are shown as means ± SD. *p < 0.05, **p < 0.01, ***p < 0.001. The significant differences in (c)–(g) are determined by comparison with the aged group and p values were determined using ANOVA tests.

We further performed co-immunofluorescence of the macrophage marker, CD68 (green) and senescence marker, p53 (red) to investigate whether pCQ/SOD could rescue macrophage senescence in vivo. Interestingly, pCQ/SOD treatment could not only inhibit p53 expression in synovial macrophage in aged mice, but limit macrophage infiltration in synovium as well (Fig. 7h and i). Moreover, pCQ/SOD treatment decreased c-cas3 positive apoptotic cells in synovium of aged mice via enhancing macrophage efferocytosis (Fig. S13c). To conclude, this engineered senotherapeutic strategy exhibited excellent ability of anti-aging and anti-inflammation for treatment of aging-related OA.

3. Discussion

OA is one of the most common aging-related diseases with complex immunological pathogenesis. Synovial macrophages, which are the most involved immune cells in OA pathology, have been long proved to function via polarization to different states [8]. Yet in the complex microenvironment of OA, the clustering of synovial macrophages and related functions are urgently needed to be investigated more deeply, far more than M1/M2 macrophages. For example, Nature reported that CX3CR1+tissue-resident macrophages are distinct subcluster of synovial macrophages providing a tight-junction-mediated shield for intra-articular structures, indicating various functions of synovial macrophages [72]. Notably, senescent cells play a critical role in the OA pathology, including senescent chondrocytes and senescent fibroblasts [73], while immunosenescence, playing a vital role in various aging-related diseases, has merely been focused on in OA. In this study, we firstly identify macrophage senescence in OA synovium via analysis of single cell sequencing and immunofluorescence of OA samples. OA senescent macrophages exhibit enhanced SASP secretion, increased M1 polarization, mitochondrial damage, and impaired efferocytosis, which lead to OA progression. Hence, we propose a novel CQ-containing polymer for the intracellular delivery of SOD to target macrophage senescence and related functional decline. The prepared pCQ/SOD nanoparticle functions as an efficient ROS scavenging reagent which inhibits oxidative stress-induced MOMP via SOD, and promotes macrophage efferocytosis through STAT3/ADAM17/MerTK pathway via pCQ. This senotherapeutic nanoparticle strategy effectively reverses the senescence of synovial macrophages, further successfully modulates M1-to-M2 repolarization. Notably, the M2 induction of pCQ/SOD did not decrease the efferocytosis of macrophages, but enhance the efferocytotic ability which is consistent with previous studies [74,75]. Ultimately, this immunosenescence-targeting strategy relieves symptoms of both trauma- and aging-induced OA.

To conclude, our study not only reveals synovial macrophage senescence in OA progression, but establishes an effective and promising treatment for OA via the intervention of immunosenescence. It is believed that macrophage-targeted therapies possess more precise targeting ability due to the physical property of synovium and phagocytosis of macrophages, while chondrocyte-targeted therapies call for effective penetration of cartilage via material-based or drug-based therapies [76]. Importantly, the study reveals inadequate understanding of OA macrophages and opens new avenues for precise functional regulations of immunocytes in early-stage OA.

4. Methods

4.1. Materials

Methacryloyl chloride (95 %), DEAEMA (98 %), Triethylamine (TEA, 99 %) and 2,2′-Azobis(2-methylpropionitrile) (AIBN, 98 %) were purchased from Shanghai Titan Technology (Shanghai, China). Hydroxychloroquine (HCQ, 99.77 %), Chloroquine diphosphate (CQ, 98 %) and 4-cyano-4-(thiobenzoylthio)pentanoic acid (CPADB, 99 %) were purchased from Bide Pharmatech (Shanghai, China). SOD was purchased from Yuanye Bio (Shanghai, China). Hoechst 33342 and WST-8 were purchased from Beyotime (Shanghai, China). The expression and purification of EGFP was carried out according to a previous reported study [77].

4.2. Synthesis of MACQ

HCQ (2.0 g, 5.95 mmol) and TEA (1.8g, 17.86 mmol) were dissolved in dichloromethane under ice-cooling. Methacryloyl chloride (0.7g, 7.15 mmol) was added to the HCQ solution dropwise with stirring at 0 °C. The mixture was stirred overnight at room temperature, followed by washing with saturated sodium carbonate and saturated salt water sequentially. The crude product obtained by evaporating solvent under vacuum was further purified by silica gel chromatography (50 : 1 dichloromethane: methanol) to obtain MACQ (2.0g, 84.6 %) as light yellow oily liquid. ¹H NMR (500 MHz, CDCl₃, ppm, δ) (Fig. S1a). 8.48 (d, J = 5.5 Hz, 1H), 7.94 (d, J = 2.5 Hz, 1H), 7.68 (d, J = 9.0 Hz, 1H), 7.33(dd, J = 9.0, 2.0 Hz, 1H), 6.40 (d, J = 4.5 Hz, 1H), 6.06 (m, 1H), 5.50 (m, 1H), 5.12 (d, J = 7.5 Hz, 1H), 4.20(td, J = 6.5, 2.0 Hz, 2H), 3.69 (m, 1H), 2.73(td, J = 6.5, 1.0 Hz, 2H), 2.56 (q, J = 7.0 Hz, 2H), 2.50 (t, J = 7.0 Hz, 2H), 1.90 (m, 3H), 1.77–1.54 (m, 4H), 1.29 (d, J = 7.0 Hz, 3H), 1.0 (t, J = 7.0 Hz, 3H).

4.3. Synthesis of pCQ

pCQ was synthesized by RAFT polymerization. Specifically, the feeding ratio of [DEAEMA]/[MACQ]/[CPADB]/[AIBN] was fixed as 100:83:1:0.3. DEAEMA (213.5 mg, 1.152 mmol), MACQ (386.4 mg, 0.957 mmol), CPADB (3.2 mg, 0.011 mmol) and AIBN (0.57 mg, 0.003 mmol) were dissolved in N,N-dimethylformamide, and the solution was further encapsulated in a Schlenk line and fully deoxygenated with nitrogen for three times. After polymerized at 70 °C for 24 h, the solution was placed in an ice bath to quickly terminate the polymerization. The solution was dialyzed in N,N-dimethylformamide for 24 h, followed by 48 h in water (membrane cut-off M_W: 3500). pCQ as a white solid was obtained after freeze-drying, the product was characterized by ¹H NMR (500 MHz, CH₃OD) and gel permeation chromatography (GPC, HLC-8320). For the needs of subsequent experiments, a small amount of 1 M hydrochloric acid was added during dialysis in water to adjust the pH and obtain a clear and transparent polymer solution.

4.4. Polymer characterization

The synthesized polymer was characterized by ¹H NMR (500 MHz) and gel permeation chromatography (GPC, HLC-8320). Specifically, the molar ratio of [DEAEMA]/[MACQ] (m:n) in pCQ could be calculated by equation (1). Where m and n refer to the degree of polymerization of DEAEMA and MACQ, respectively, while I_d,e and I_l refer to the integrated areas of d,e peaks (H_d,e in Fig. S1b) and l peak (H_l in Fig. S1b), respectively.

$$\frac{6m+6n}{n}=\frac{I_{d,e}}{I_l}$$ (1)

Molecular weight of the synthesized polymer was analyzed by GPC operated in tetrahydrofuran (THF). The proportion of polymerization units and molecular weights of polymers were shown in Fig. S1b.

4.5. Polymer/protein complexes preparation and characterization

Briefly, proteins were mixed with pCQ in phosphate buffer saline (PBS) or cell culture medium, and incubated at room temperature for 30 min to form polymer/protein complexes. The complexes were further diluted before characterization. The size and particle dispersion (PDI) of the nanoparticles were measured by DLS (Malvern, UK) at 25 °C. The morphology and size of the nanoparticles were characterized by TEM (Hitachi, JPN).

4.6. Determination of protein loading efficiency (PLE) and protein loading content (PLC)

Fluorescein Isothiocyanate (FITC) was labeled on SOD to quantitatively detect the protein concentration in the system by fluorescence spectrophotometer. Briefly, the pCQ/SOD-FITC solution was centrifuged at 15000 rpm for 15 min and the unbound proteins in the supernatant solution were quantified. SOD-FITC only was also centrifuged as a control to avoid the effect of centrifugation on free proteins. PLE and PLC were defined as follows: PLE = (weight of loaded proteins/weight of proteins in feed) × 100 %, PLC = (weight of loaded proteins/total weight of loaded proteins and carrier) × 100 %.

4.7. Determination of protein release ratio

We measured and calculated the protein release ratio of pCQ/SOD-FITC nanoparticles at different times using centrifugation and fluorescence spectrophotometer. Briefly, FITC was labeled on SOD to quantitatively detect the protein concentration in the system by fluorescence spectrophotometer. The pCQ/SOD-FITC solution was centrifuged at 15000 rpm for 15 min and the unbound proteins in the supernatant solution were quantified. SOD-FITC only was also centrifuged as a control to avoid the effect of centrifugation on free proteins. At different intervals, the supernatants of the nanoparticles after centrifuged were taken to detect the free protein release, and the weight of protein loaded in the nanoparticles at 0 h was used as the total loaded protein. The formulation for calculating protein release ratio is as follows: (the weight of total loaded protein - the weight of loaded protein at different times)/the weight of total loaded protein.

4.8. Cell culture and intracellular protein delivery

Mouse mononuclear macrophages cells (Raw264.7 cells), mouse embryonic fibroblast cells (NIH-3T3 cells) were incubated in DMEM cell culture medium (GIBCO) at 37 °C under 5 % CO₂. The culture medium contains 10 % fetal bovine serum (FBS), penicillin (100 units/mL), and streptomycin (100 μg/mL).

For intracellular protein delivery, cells were seeded in 24-well plates and cultured overnight. Cargo proteins were complexed with pCQ in 100 μL culture medium for 30 min, then diluted with 400 μL medium containing 10 % serum. The cells were washed with PBS and treated with the polymer/protein complexes for 12 h at 37 °C. After that, the treated cells were washed with PBS, subsequently, the protein delivery efficiency were analyzed. The cells were observed by laser scanning confocal microscopy (Leica, TCS SP8, Germany) and the delivery efficacy was quantitatively analyzed by flow cytometry (BD, FACSCelesta, USA).

4.9. Cellular uptake pathway investigation

To investigate the internalization mechanism of the polymer/protein complexes, RAW264.7 cells were pre-incubated with different endocytosis inhibitors respectively, including cytochalasin D (10 μM), ethylisopropylamiloride (EIPA, 50 μM), Wortmannin (100 nM), Genistein (700 μM), Nystatin (25 μg/mL), Chlorpromazine (20 μM), Dynasore (20 μM), methyl-β-cyclodextrin (MβCD, 10 mM) and NaN₃ (0.5 mg/mL) for 1 h before intracellular protein delivery. Moreover, RAW264.7 cells were pre-incubated at 4 °C for 1 h and the protein delivery experiment was conducted at 4 °C to investigate whether the cellular uptake process is dependent on the energy. The cells treated with the complexes at 37 °C without any inhibitor were tested as a control. The cells were analyzed 3 h after incubation to quantify the fluorescence level of SOD-Cy5.5 by flow cytometry.

4.10. Cell viability assay

The cytotoxicity of pCQ was measured by the standard cell counting kit-8 (CCK-8) assay. Briefly, Raw264.7 cells were seeded in 96-well plates and cultured until 40–50 % confluence. The cells were then incubated with pCQ at concentrations ranging from 5 to 40 μg/mL for 24 h, subsequently refreshed with medium containing 10 % FBS, and further incubated for another 24 h. The CCK-8 solution was added to the treated cells and incubated for 1 h. Absorbance at the wavelength of 450 nm was measured to calculate cell viability.

4.11. Animals

C57BL/6 mice were purchased from the Shanghai Laboratory Animal Center (Shanghai, China). All animal experiments were performed in a pathogen-free animal facility at Xinhua Hospital affiliated with Shanghai Jiaotong University of Medicine. Institutional Ethics Committee of Xinhua Hospital approved related experiments and the animal allowance number of Xinhua Hospital is SYXK-2018-0038.

4.12. In vivo evaluation of targeting ability

4-mon-old male C57BL/6 mice were obtained for detection of targeting ability of pCQ/SOD. SOD was labeled with Cy5.5 (SOD-Cy5.5). The concentration of pCQ for intraarticular injection was 1.6 μg/μL pCQ/SOD-Cy5.5 (30 U/μL) was intraarticularly injected in the murine joints (SOD: 6200 U/mg, each injection for 0.024 mg SOD). Knee joints were collected 12 h after injection for further examination.

4.13. Experimental models

Aging-induced spontaneous OA in mice were constructed in 18-month-old male mice and 3-month-old male mice were established as young control (n = 6). Knee joint samples were collected after 6 weeks of intervention. Trauma-induced OA were maintained by DMM or ACLT surgery in 3-month-old male mice (n = 6), in which the medial meniscotibial ligament (for DMM) or the anterior cruciate ligament (for ACLT) was transected in the murine knee joint after anesthetized. Sham operations were done on independent mice with same gender and age. Knee joints were collected 4 weeks after DMM or ACLT surgery. The concentration of pCQ for intraarticular injection was 1.6 μg/μL. To test the efficacy of pCQ/SOD, 5 μL PBS, free pCQ, free SOD (30 U/μL) or pCQ/SOD (30 U/μL) was injected into murine joints (SOD: 6200 U/mg, each injection for 0.024 mg SOD).

4.14. Human samples

We collected human synovium samples from 7 OA patients who needed to perform total knee replacement surgery and 3 patients requiring joint repair who never suffered osteoarthritis in their history. Institutional Ethics Committee of Xinhua Hospital has already approved this study.

4.15. Pathological staining and immunochemical techniques

Safranin O and fast green staining were applied to examine the extent of cartilage degradation in OA models. CD68 (Cell Signaling Technology), Vimentin, p53, iNOS and CD206 (Abclonal) in pathological sections were examined via immunofluorescence and obtained using Zeiss microscope, while MMP13 and ACAN (Abclonal) were examined via immunohistochemistry. Joint samples and synovium were collected for preparation of paraffin sections. After treated with pepsin to repair antigen, the sections were blocked using BSA to decrease non-specific bindings of antibody. Furthermore, primary antibodies with appropriate concentrations (according to protocols) were added on the surface of slices for 10 h in 4 °C refrigerator. After primary antibodies (rabbit source) incubation, we next used secondary antibody (anti-rabbit) from IHC kit (MaxVision, China) to incubate slices for 10 min. Slices were then washed using TBST solution and stained for immunohistochemical staining according to the protocol of DAB Plus kit or for immunofluorescence according to the protocol of abs50012, absin. Related quantifications were measured via ImageJ.

4.16. Macrophage culture and assay

Bone marrow cells were collected via flush of murine bone marrow cavity, and then treated with 10 ng/mL mouse M-CSF in DMEM medium with 10 % FBS for 12 days for development of BMDMs. Raw 264.7 cells were induced by LPS (100 ng/mL) for 24 h to induce M1 polarization and were induced by IL-4 (1000 U/mL) treatment for 24 h to induce M2 polarization. Macrophage senescence was induced in vitro by treatment of 800 μM H₂O₂ or 10 g/L D-gal for 24 h in Raw264.7 cells [78,79]. β-gal staining was established by usage of β-Gal Staining Kit. ROS detection was established by usage of ROS Assay Kit with CM-H2DCFDA. JC-1 detection was established by usage of Mitochondrial membrane potential assay kit with JC-1. These kits were all purchased from Beyotime, China. Lipopolysaccharide (LPS; E. coli 055: B5) was purchased from Sigma-Aldrich, and recombinant mouse IL-4 was purchased from Peprotech. D-gal was purchased from MedChemExpress.

4.17. Flow cytometry

Raw264.7 cells were treated in different conditions and labeled with the following reagents for identification of M1 or M2 macrophages: PE/Cyanine5 anti-mouse CD86 Antibody (Biolegend, 105015), FITC anti-mouse CD206 (MMR) Antibody (Biolegend, 141703). Related controls are PE/Cyanine5 Rat IgG2a, κ Isotype Ctrl Antibody (Biolegend, 400509), FITC Rat IgG2a, κ Isotype Ctrl Antibody (Biolegend, 400505). After suspension of Raw264.7 cells, they were incubated with related reagents for 30 min, and analyzed via Guava Easycyte 12HT (Luminex) after flushed by PBS.

4.18. Western blot assay

Proteins of cells were obtained after lysed in RIPA buffer for 30 min on ice and quantified via BCA assay.

After SDS-PAGE of proteins, the samples were transferred to NC membranes. After blocked with 5 % BSA for 1 h and incubated with antibodies (such as p-STAT3, STAT3, GAPDH from Cell Signaling Technology, and ADAM17 from Abclonal) overnight at 4 °C, the appropriate HRP-coupled secondary antibody was then added and detected through chemiluminescence (Millipore). Related quantifications were measured via ImageJ.

4.19. qPCR analysis

We used TRIzol reagent (Invitrogen, Waltham, MA) for extraction of RNA and quantify by Nanodrop 2000 (Thermo Fisher Scientific). We then used PrimeScript RT Master Mix Kit (Takara Bio Inc., China) to obtain cDNA from related mRNA. The transcriptive levels of certain genes were then tested by qPCR. Housekeeping gene GAPDH was established as control. Sequences of primers for the genes were listed as below:

GAPDH-F: TGACCTCAACTACATGGTCTACA.

GAPDH-R: CTTCCCATTCTCGGCCTTG

iNOS-F: GTTCTCAGCCCAACAATACAAGA

iNOS-R: GTGGACGGGTCGATGTCAC.

CD206-F: CTCTGTTCAGCTATTGGACGC.

CD206-R: TGGCACTCCCAAACATAATTTGA

p53-F: CCCCTGTCATCTTTTGTCCCT

p53-R: AGCTGGCAGAATAGCTTATTGAG.

BAX-F: AGACAGGGGCCTTTTTGCTAC.

BAX-R: AATTCGCCGGAGACACTCG.

MerTK-F: TGCGTTTAATCACACCATTGGA.

MerTK-R: TGCCCCGAGCAATTCCTTTC.

4.20. Statistical analysis

For calculating statistical significance, analysis of variance (ANOVA) were used between multiple groups and two-tailed Student's t-test were used between two groups. P < 0.05 represents statistical significances.

CRediT authorship contribution statement

Yuhang Liu: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Jianan Duan: Writing – review & editing, Writing – original draft, Visualization, Validation, Resources, Project administration, Methodology, Investigation. Yifan Dang: Validation, Project administration, Investigation. Ruihan Hao: Visualization, Validation, Software, Resources, Methodology, Investigation, Data curation. Hui Wang: Resources, Project administration, Methodology. Echuan Tan: Visualization, Validation, Methodology. Ruijue Wang: Resources, Project administration, Methodology. Yuhan Li: Validation, Resources, Project administration, Methodology. Song Zhang: Resources, Project administration, Methodology. Yuanchi Wang: Resources, Project administration, Methodology. Jia Lv: Writing – review & editing, Visualization, Validation, Resources. Yuxin Qi: Resources, Methodology. Xiaoling Zhang: Writing – review & editing, Visualization, Validation, Supervision, Resources, Project administration, Methodology, Funding acquisition, Conceptualization. Yiyun Cheng: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Resources, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization.

Ethics approval and consent to participate

Institutional Ethics Committee of Xinhua Hospital approved related animal experiments with allowance number: SYXK-2018-0038, and approved collection of human samples with allowance number: XHEC-NSFC-2018-208. Written informed consent were obtained from all participants prior to sample collection.

Funding

This work was supported by grants from National Key R&D Program of China (2024YFA1210100), National Natural Science Foundation of China (22135002, 81830078), Science and Technology Commission of Shanghai Municipality (23141901200), Health Commission of Shanghai Municipality (2022JC029), Biomaterials and Regenerative Medicine Institute Cooperative Research Project, Shanghai Jiaotong University School of Medicine (2022LHA11), the East China Normal University Multifunctional Platform for Innovation (004), and the Instruments Sharing Platform of School of Life Sciences, East China Normal University.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article: