An aptamer specifically targeting mCRP ameliorates experimental arthritis

Abstract

Background

Recent evidence highlights the important role of the liver–bone axis in the development of arthritis, particularly rheumatoid arthritis (RA) and osteoarthritis (OA). The liver secretes various factors that impact joint health, one of which is C-reactive protein (CRP), elevated in RA and OA patients. Traditionally regarded as an inflammatory marker, the causal role of CRP in arthritis development remains a topic of debate due to the existence of its two isoforms with opposing functions: native pentameric CRP (nCRP) and monomeric CRP (mCRP).

Methods

We generated hepatocyte-specific CRP knockout mice to investigate the causal role of CRP in RA and OA mouse models. In vitro experiments were conducted to assess the effects of mCRP and nCRP on phenotypic changes in effector cells common to RA and OA, including fibroblast-like synoviocytes (FLSs), monocytes/macrophages, and chondrocytes. Using systematic evolution of ligands by exponential enrichment (SELEX), we screened nucleic acid aptamers targeting mCRP rather than nCRP. We determined the neutralizing effects of the selected aptamers on mCRP in vitro and explored their therapeutic potential and safety in RA and OA mouse models.

Results

Hepatocyte-specific knockout of CRP significantly reduced disease severity in RA and OA mouse models. mCRP promoted in vitro pathological changes in FLSs, monocytes/macrophages, and chondrocytes, while nCRP exhibited minimal or slightly protective effects. We identified an aptamer, ApmCRP3, which effectively inhibited mCRP-induced pathological changes of RA and OA effector cells in vitro. In mouse models of RA and OA, ApmCRP3 displayed strong therapeutic effects and a favorable safety profile.

Conclusion

This study identifies hepatocyte-derived mCRP as a contributor to RA and OA pathogenesis and highlights ApmCRP3 aptamer as a promising therapeutic candidate.

The translational potential of this article

This study highlights the therapeutic potential of ApmCRP3 in attenuating mCRP-driven pathology and controlling arthritis progression.

Graphical abstract

1. Introduction

Arthritis, a condition characterized by pain and inflammation in one or multiple joints, is prevalent worldwide and encompasses more than 100 types [1]. The two most common types are rheumatoid arthritis (RA) and osteoarthritis (OA). Etiologically, RA is a poly-articular systemic autoimmune disorder that mainly harms the feet and hands, while OA usually begins in an isolated joint and is caused by external force-induced trauma or age-related wearing down of cartilage [2,3]. Notably, RA and OA share several overlapping pathological features, such as immune cell infiltration, synovial swelling, and cartilage destruction, which are mediated by diverse effector cells, including monocytes/macrophages, fibroblast-like synoviocytes (FLSs), and chondrocytes [4,5]. Despite these similarities, the molecular mechanisms underlying the initiation and progression of both RA and OA are still not well elucidated, and there is an urgent need for therapeutic strategies to address the unmet clinical demand [6,7].

The pathological features observed in RA and OA are increasingly understood to be modulated by inter-organ communication [8,9]. Emerging evidence suggests that bone serves not only as a structural organ but also as a central hub that integrates signals from distant tissues and secretes regulatory factors to coordinate immune responses [10]. Organs such as the liver, adipose tissue, muscle, and brain engage in bidirectional communication with bone through the exchange of circulating factors, including osteocalcin, sclerostin, lipocalin-2, cytokines, adipokines, myokines, and neurohormones [[11], [12], [13], [14], [15], [16]]. The inter-organ talks between liver and bone, known as the liver–bone axis, have drawn significant attention due to their potential influence on the progression of RA and OA. It has been reported that hepatocytes produce circulating factors, including interleukin-6 (IL-6), serum amyloid A (SAA), insulin-like growth factor-binding proteins (IGFBPs), and bile acids, which can influence synovial hyperplasia, cartilage remodeling, and bone immunity [[17], [18], [19], [20], [21], [22]].

C-reactive protein (CRP) is a highly conserved acute-phase protein synthesized primarily by hepatocytes in response to pro-inflammatory cytokines, particularly IL-6 [23]. CRP exists in two distinct forms: native pentameric CRP (nCRP) and monomeric CRP (mCRP). The monomeric form arises from the dissociation of nCRP within an inflamed microenvironment and exhibits biological activities that are opposite to those of nCRP [23]. While nCRP demonstrates weak anti-inflammatory properties by affecting complement activation and inducing phagocytosis, mCRP promotes inflammation via enhancing endothelial activation, monocyte chemotaxis, and leukocyte recruitment [23]. CRP is a well-established marker and potential therapeutic target for cardiovascular diseases, with elevated levels indicating atherosclerotic progression and increased adverse cardiovascular events [24]. However, despite CRP levels reliably correlating with disease activity in RA and OA, the significance of CRP elevation, whether as a driver of pathology or mere epiphenomenon, continues to be debated [[25], [26], [27]].

In this study, we generated hepatocyte-specific CRP knockout mice and established experimental models for RA and OA, demonstrating that the deletion of hepatocyte-derived CRP significantly suppressed arthritis in these mouse models. We determined the in vitro effects of mCRP and nCRP on effector cells common to RA and OA, including FLSs, monocytes/macrophages, and chondrocytes, and showed that mCRP promoted pathogenic phenotypes of these cells, while nCRP exhibited minimal or even opposing protective effects. We employed a systematic evolution of ligands by exponential enrichment (SELEX) strategy to identify a single-stranded nucleic acid aptamer, AptmCRP3, that specifically binds to mCRP without affecting nCRP. AptmCRP3 efficiently neutralized the pathological effects of mCRP, substantially alleviating disease progression in our experimental models for RA and OA. These findings clarify the controversial role of CRP in arthritis via the liver–bone axis and suggest that targeting mCRP with aptamers may be a promising therapeutic strategy for both RA and OA.

2. Results

2.1. Hepatocyte-specific CRP knockout alleviates arthritic progression in CAIA mice

To investigate whether CRP plays a causal role in the development of RA and OA within the liver–bone axis framework, we generated transgenic mouse model expresses a floxed CRP with loxP sites flanking exons 1 and 2 (refer to as CRP^flox/flox mice) and then crossed them with commercially available Albumin-Cre-ER^T2 (Alb-Cre-ER^T2) mice to generate offspring carrying both the CRP^flox/flox allele and the inducible Alb-Cre-ER^T2 transgene (referred to as CRP^flox/flox; Alb-Cre-ER^T2 mice). At two months of age, these mice were subjected to five daily injections of tamoxifen (TAM) to create hepatocyte-specific CRP knockout mice (referred to as CRP^ΔHep/ΔHep mice) (Fig. S1A). In contrast, the CRP^flox/flox; Alb-Cre-ER^T2 mice injected with corn oil served as wild-type controls (referred to as CRP^+/+ mice). The collagen antibody-induced arthritis (CAIA) mouse model is commonly employed to study the efferent phase of RA [28]. This model is induced by administering a cocktail of antibodies against collagen type II (CII) and recapitulates key pathogenic features similar to RA, such as pannus formation, cellular infiltration, synovitis, and cartilage destruction [28]. A notable advantage of the CAIA model is its ability to induce arthritis in mouse strains that are typically resistant to traditional collagen-induced arthritis (CIA) methods, such as the C57BL/6J strain [28]. Given that both CRP^ΔHep/ΔHep and CRP^+/+ mice were maintained on a C57BL/6J background, we examined arthritis symptoms following CAIA induction in these mice (Fig. 1A). As CRP is synthesized in the liver as the native nCRP isoform and secreted into circulation, subsequently dissociating into the mCRP isoform upon reaching sites of inflammation [29], we collected liver, serum, and paw tissues post-induction and assessed CRP expression by immunofluorescence (IF) staining and enzyme-linked immunosorbent assay (ELISA). Our results demonstrated a marked reduction in nCRP expression in hepatocytes and serum, as well as decreased accumulation of mCRP in the inflamed paws of CRP^ΔHep/ΔHep mice compared to CRP^+/+ mice (Fig. S1B–S1D). Our results demonstrated that CRP^ΔHep/ΔHep mice showed reduced arthritis scores compared to CRP^+/+ mice (Fig. 1B). ELISA revealed that serum levels of pro-inflammatory cytokines interleukin (IL)-6 and IL-1β were significantly lower in CRP^ΔHep/ΔHep mice than in CRP^+/+ mice (Fig. 1C and D). Paw edema was less severe in CRP^ΔHep/ΔHep mice (Fig. 1E). Histological analysis demonstrated that CRP^ΔHep/ΔHep mice had reduced synovial hyperplasia and cartilage erosion compared to CRP^+/+ mice (Fig. 1F and G). IF staining and quantitative analysis showed significantly reduced levels of vimentin (a marker of FLSs), CD86 (a marker for M1 macrophages), and matrix metalloproteinase (MMP)-13 (a cartilage-degrading enzyme) in the paws of CRP^ΔHep/ΔHep mice (Fig. 1H–K). Neither CRP^ΔHep/ΔHep nor CRP^+/+ mice developed spontaneous arthritis without CAIA induction (Fig. 1B–K).

Fig. 1.

Development of CAIA in hepatocyte-specific CRP knockout mice. (A) Schematic diagram illustrating the development of CAIA in CRP^+/+ and CRP^ΔHep/ΔHep mice. (B) Arthritis scoring of CRP^+/+ and CRP^ΔHep/ΔHep mice with or without CAIA induction. Mice without CAIA served as non-immunized (NI) controls. (C and D) Serum IL-6 (C) and IL-1β (D) levels in CRP^+/+ and CRP^ΔHep/ΔHep mice with or without CAIA induction, as determined by ELISA. (E) Photographs of hind paws of CRP^+/+ and CRP^ΔHep/ΔHep mice with or without CAIA induction. (F) H&E staining of paw sections of CRP^+/+ and CRP^ΔHep/ΔHep mice with or without CAIA induction. Scale bar = 300 μm. (G) SO&FG staining of paw sections of CRP^+/+ and CRP^ΔHep/ΔHep mice with or without CAIA induction. Scale bar = 300 μm. (H) IF staining of paw sections of CRP^+/+ and CRP^ΔHep/ΔHep mice with or without CAIA induction, using an anti-Vimentin, an anti-CD86, or an anti-MMP-13 antibody. Scale bar = 50 μm. (I–K) Quantification of expression of Vimentin (I), CD86 (J), and MMP-13 (K) on IF-stained sections. Data are represented as mean ± SD of n = 5 per group. P-values from one-way ANOVA (C, D, I-K) or two-way ANOVA (B): ∗∗p < 0.01, ∗∗∗p < 0.001.

2.2. Genetic deletion of hepatocyte-derived CRP reduces OA pathology in DMM mice

The most widely used mouse model of OA is the destabilization of the medial meniscus (DMM). To investigate the role of hepatocyte-derived CRP in OA pathogenesis, we performed DMM surgery on the knee joints of CRP^+/+ or CRP^ΔHep/ΔHep mice (Fig. 2A). Following the DMM surgery, IF staining and ELISA confirmed reduced levels of nCRP in the serum and mCRP in the joints of CRP^ΔHep/ΔHep mice compared to CRP^+/+ mice (Fig. S1E and S1F). Micro-computerized tomography (μCT) imaging revealed that CRP^ΔHep/ΔHep mice exhibited reduced osteophyte formation compared to CRP^+/+ mice (Fig. 2B). Histological analysis demonstrated less synovial hyperplasia and cartilage damage in CRP^ΔHep/ΔHep mice than in CRP^+/+ mice (Fig. 2C and D). IF staining indicated lower levels of vimentin (a marker of FLSs) and CD86 (a M1 macrophage marker), and increased level of COL2A1 (collagen type II α1 chain, a chondrogenic marker) in the knee joints of CRP^ΔHep/ΔHep mice compared to those of CRP^+/+ mice (Fig. 2E). Quantitative assessments, including osteoarthritis research society International (OARSI) score, cartilage area, osteophyte score, synovitis score, vimentin-positive cells, CD86-positive cells and COL2A1-positive cells, consistently suggested that the genetic ablation of hepatocyte-derived CRP significantly attenuated OA progression in DMM mice model (Fig. 2F–L). Neither CRP^ΔHep/ΔHep nor CRP^+/+ mice developed spontaneous OA following sham surgery (Fig. 2B–L).

Fig. 2.

Establishment of surgical DMM in hepatocyte-specific CRP knockout mice. (A) Schematic diagram illustrating the establishment of surgical DMM in CRP^+/+ and CRP^ΔHep/ΔHep mice. (B) μCT scans of knee joints of CRP^+/+ and CRP^ΔHep/ΔHep mice after sham or DMM surgery, with black arrows indicating the osteophyte formation. Scale bar = 1 mm. (C) H&E staining of joint sections of CRP^+/+ and CRP^ΔHep/ΔHep mice after sham or DMM surgery. Scale bar = 300 μm. (D) SO&FG staining of joint sections of CRP^+/+ and CRP^ΔHep/ΔHep mice after sham or DMM surgery. Scale bar = 300 μm. (E) IF staining of joint sections of CRP^+/+ and CRP^ΔHep/ΔHep mice after sham or DMM surgery, using an anti-vimentin, an anti-CD86, or an anti-COL2A1 antibody. Scale bar = 50 μm. (F–I) Quantification of OARSI score (F), cartilage area (G), osteophyte score (H), and synovitis score (I) of CRP^+/+ and CRP^ΔHep/ΔHep mice after sham or DMM surgery. (J–L) Quantification of expression of vimentin (J), CD86 (K), and COL2A1 (L) on IF-stained sections. Data are represented as mean ± SD of n = 5 per group. P-values from one-way ANOVA (F–L): ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

2.3. mCRP, rather than nCRP, induces pathological changes in RA and OA effector cells

FLSs drive the transition of the joint synovium from a healthy state to a pathological one in RA and OA [30,31]. In the inflamed microenvironment, FLSs become activated and acquire tumor-like phenotypes, including resistance to apoptosis, increased proliferation, and enhanced invasion and migration. Additionally, these activated FLSs secrete pro-inflammatory cytokines (e.g., IL-6 and IL-1β), CXC motif chemokines (CXCLs), and matrix-degrading MMPs, leading to bone and cartilage erosion [30,31]. We dissociated nCRP into mCRP via urea/EDTA treatment [32] and validated mCRP formation by non-denaturing electrophoresis (Fig. 3A and B), as denaturing conditions would disrupt the native pentameric structure of nCRP, resulting in migration patterns similar to mCRP and thus precluding isoform discrimination [32]. Upon incubating human primary FLSs with either mCRP or nCRP, the RT-qPCR assay demonstrated that mCRP upregulated mRNA expression of IL-6, CXCL12, MMP-3, and IL-1β (Fig. 3C). CCK-8 assay showed that mCRP enhanced the proliferation of FLSs (Fig. 3D). Transwell assay revealed that mCRP increased migratory and invasive capacities of FLSs (Fig. 3E and F). Flow cytometry revealed that mCRP decreased the apoptosis of FLSs (Fig. 3G). Conversely, nCRP exhibited minimal or opposite effects on the pro-inflammatory and tumor-like phenotypes of FLSs (Fig. 3C–G).

Fig. 3.

Effects of mCRP and nCRP on FLSs and monocytes/macrophages in vitro. (A) Schematic diagram of mCRP preparation from nCRP. (B) Non-denaturing electrophoresis of nCRP and mCRP. (C) RT-qPCR analysis for the mRNA levels of IL-6, CXCL12, MMP-3, and IL-1β in FLSs after treatment with mCRP or nCRP at the indicated concentrations for 24 h. (D) CCK-8 assay for the viability of FLSs after 4-day treatment with mCRP or nCRP at the indicated concentrations daily. (E and F) Transwell migration (E) and invasion (F) of FLSs after treatment with mCRP or nCRP at the indicated concentrations for 48 h. Scale bar = 100 μm. (G) Flow cytometry for detecting the serum starvation-induced apoptosis of FLSs after 6-day treatment with mCRP or nCRP at the indicated concentrations daily. (H) RT-qPCR analysis for the mRNA levels of IL-12, IL-6, TNF-α, and IL-1β in THP-1 cells after treatment with mCRP or nCRP at the indicated concentrations for 12 h. (I) Flow cytometry for detecting the M1 polarization of PMA-pretreated THP-1 cells after treatment with mCRP or nCRP at the indicated concentrations for 24 h. Each experiment was repeated three times. Data were presented as mean ± SD. P-values from one-way ANOVA (C, E-I) or two-way ANOVA (D): ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Monocytes play a critical role in RA and OA pathogenesis by differentiating into pro-inflammatory M1 macrophages within the synovial microenvironment [33]. Once polarized, M1 macrophages secrete pro-inflammatory cytokines such as IL-12, TNF-α, IL-1β, and IL-6, which amplify synovitis, recruit additional immune cells, and promote cartilage and bone destruction [31,34]. RT-qPCR assay demonstrated that exposure to mCRP significantly upregulated IL-12, IL-6, TNFα, and IL-1β mRNA expression in the human monocytic THP-1 cell line (Fig. 3H). We differentiated THP-1 cells into naïve macrophage-like cells (M0) using phorbol 12-myristate 13-acetate (PMA) [35], prior to mCRP or nCRP treatment. Flow cytometry analysis showed that mCRP induced M1 polarization, as evidenced by increased surface expression of M1 macrophage markers CD86 and CD68 (Fig. 3I). However, nCRP either had no effects or reduced the mRNA expression of these pro-inflammatory cytokines or M1 markers in the monocytes/macrophages (Fig. 3H and I).

Chondrocytes, the sole resident cells of articular cartilage, are essential for maintaining the integrity and homeostasis of the extracellular matrix (ECM) [36]. Under physiological conditions, chondrocytes ensure the synthesis of cartilage matrix components, particularly aggrecan (ACAN) and COL2A1, which are critical for cartilage structure and function [37]. However, when exposed to pro-inflammatory stimuli, chondrocytes can shift toward a catabolic phenotype, characterized by the increased expression of matrix-degrading enzymes, such as MMPs and a disintegrin and metalloproteinase with thrombospondin motifs-5 (ADAMTS-5) [38,39]. We incubated mouse primary chondrocytes with mCRP or nCRP and revealed that mCRP downregulated the expression of ACAN and COL2A1, while upregulating MMP-13 and ADAMTS-5 levels in chondrocytes. In contrast, nCRP displayed no effects on the expression of these anabolic or catabolic markers (Fig. S2A and S2B).

We then treated FLSs, monocytes, and chondrocytes with equimolar concentrations of mCRP and nCRP, as nCRP is a pentamer composed of five identical mCRP subunits. Consistently, RT-qPCR analysis showed that mCRP significantly increased the expression of IL-6, CXCL12, MMP-3, and IL-1β in FLSs (Fig. S2C). In monocytes, mCRP elevated the levels of IL-12, IL-6, TNFα, and IL-1β (Fig. S2D). In chondrocytes, mCRP upregulated the catabolic markers MMP-13 and ADAMTS-5 while suppressing the anabolic markers ACAN and COL2A1 (Fig. S2E). In contrast, nCRP exerted either minimal or opposite effects, suggesting a potential anti-inflammatory role (Fig. S2C–S2E).

2.4. FcγRI is involved in mCRP-induced pathological changes in RA and OA effector cells

Fc gamma receptor I (FcγRI), the high-affinity receptor for the Fc region of immunoglobulin G (IgG), is predominantly expressed on monocytes/macrophages where it mediates immune complex recognition and inflammatory activation [40]. Moreover, under inflammatory conditions, FcγRI expression can be upregulated in non-immune stromal cells, such as FLSs and chondrocytes, enabling these cells to respond to Fc-mediated signals [[41], [42], [43]]. Emerging evidence suggests that FcγRI may function as a binding receptor for CRP [42]. To examine whether FcγRI mediates mCRP-driven pathogenic responses across diverse effector cells in RA and OA. We treated FLSs, THP-1 cells, and chondrocytes with mCRP in the presence of either a control IgG or an anti-FcγRI blocking antibody. FcγRI inhibition reduced mCRP-induced pro-inflammatory mediator production, including IL-6, IL-1β, and CXCL12, as well as MMP-3 upregulation in FLSs (Fig. S3A). Similarly, in THP-1 cells, FcγRI blockade attenuated mCRP-stimulated expression of IL-12, TNF-α, IL-1β, and IL-6 (Fig. S3B). Furthermore, FcγRI inhibition reversed mCRP-mediated downregulation of anabolic markers ACAN and COL2A1 while suppressing the induction of catabolic markers MMP-13 and ADAMTS-5 in chondrocytes (Fig. S3C).

To further elucidate the FcγRI-mediated downstream effects of mCRP stimulation, we performed RNA sequencing on FLSs, monocytes, and chondrocytes treated with mCRP. In FLSs, a total of 858 differentially expressed genes (DEGs) were identified by a volcano plot, and enrichment analysis revealed robust activation of pattern recognition receptor (PRR) signaling and type I interferon (IFN) pathways (Fig. S4A and S4B), consistent with emerging literature suggesting that FcγRI activation triggers similar PRR- and IFN-driven innate immune programs [44,45]. In monocytes, 638 DEGs were detected, with mCRP inducing a pronounced pro-inflammatory transcriptional profile characterized by NLRP3 inflammasome assembly, IL-18 production, and NOD-like receptor signaling, together with suppression of the anti-inflammatory cytokine IL-10 (Fig. S4C and S4D), which aligns well with the known role of FcγRI in mediating priming and activation of inflammasomes in myeloid cells [46,47]. In chondrocytes, mCRP treatment resulted in 1323 DEGs, characterized by suppression of anabolic genes and upregulation of IL-17, IL-1, and innate immune response pathways (Fig. S4E and S4F), which is in line with a prior study [48]. Collectively, these data suggest that mCRP drives pathogenesis in joint-resident effector cells via FcγRI-dependent mechanisms.

2.5. Screening of therapeutic aptamer targeting mCRP

Aptamers are single-stranded DNA (ssDNA) or RNA molecules generated through SELEX screening that fold into unique three-dimensional structures, enabling them to bind target proteins with high affinity and specificity [49,50]. Using His-tagged mCRP immobilized on magnetic beads as the target protein, we conducted positive selection to enrich ssDNA sequences targeting mCRP during the SELEX process. Magnetic beads coupled with His-tagged nCRP were used for negative selection (Fig. 4A). To evaluate the binding affinity, mCRP or nCRP proteins were immobilized on magnetic beads and incubated with biotin-labeled ssDNA libraries from different SELEX rounds. As the selection progressed, the enriched libraries exhibited increased binding to mCRP, whereas no detectable binding was observed with nCRP (Fig. 4B and C). To further validate these findings, we immobilized the ssDNA libraries on beads and incubated them with mCRP or nCRP. Consistently, mCRP binding increased with SELEX progression (Fig. 4D), while nCRP binding remained minimal (Fig. 4E). From the most enriched ssDNA library, we identified four aptamer candidates (ApmCRP1, ApmCRP2, ApmCRP3, and ApmCRP4). Among them, ApmCRP3 demonstrated the highest binding affinity for mCRP compared to a negative control sequence (NC) or other candidates (Fig. 4F). Importantly, none of the aptamer candidates showed binding to nCRP (Fig. 4G), which is in line with a dot blotting assay (Fig. 4H). Given the superior binding performance of ApmCRP3, we then applied surface plasmon resonance (SPR) analysis to determine the physical interaction between ApmCRP3 and mCRP or nCRP. SPR analysis revealed concentration-dependent binding of ApmCRP3 to mCRP, with a K_D of 400 nM, while no detectable binding to nCRP was observed (Fig. 4I and J). Based on aptamer sequences, we predicted that the four aptamer candidates formed distinct stem-loop structures (Fig. 4K).

Fig. 4.

Screening of aptamers targeting mCRP by SELEX. (A) Schematic diagram of SELEX. Briefly, His-tagged mCRP immobilized on magnetic beads was used as the target protein in positive selection, and His-tagged nCRP immobilized on magnetic beads was employed for negative selection. After multiple rounds of selection, aptamer candidates were cloned and sequenced. (B and C) Binding of enriched pools with His-tagged mCRP (B) or nCRP (C) immobilized on magnetic beads. (D and E) Binding of mCRP (D) or nCRP (E) with biotin-labeled enriched pools immobilized on magnetic beads. (F) Binding of NC or four aptamer candidates (ApmCRP1, ApmCRP2, ApmCRP3, and ApmCRP4) with His-tagged mCRP immobilized on magnetic beads. (G) Binding of NC or four aptamer candidates with His-tagged nCRP immobilized on magnetic beads. (H) Binding of NC or four aptamer candidates with mCRP, as determined by the dot blotting assay. (I) SPR for detecting the binding affinity between ApmCRP3 and mCRP. (J) SPR for detecting the binding affinity between ApmCRP3 and nCRP. (K) Predicted secondary structures of four aptamer candidates by the RNAstructure software. Each experiment was repeated three times. Data were presented as mean ± SD. P-values from one-way ANOVA (B and D) or two-way ANOVA (F): ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

2.6. ApmCRP3 effectively inhibits mCRP-induced pathological effects in vitro

To evaluate the therapeutic potential of ApmCRP3, we investigated whether ApmCRP3 could neutralize mCRP to attenuate the pathological changes in RA and OA effector cells induced by mCRP in vitro. After incubating FLSs with mCRP in the presence of NC or ApmCRP3, RT-qPCR results demonstrated that mCRP-induced expression of MMP-3, IL-6, IL-1β, and CXCL12 was significantly decreased by ApmCRP3 when compared to NC (Fig. 5A). CCK-8 assay revealed that ApmCRP3 inhibited mCRP-stimulated proliferation of FLSs (Fig. 5B). Additionally, transwell and flow cytometry assays showed that ApmCRP3 attenuated mCRP-induced migration, invasion, and resistance to apoptosis of FLSs (Fig. 5C–E). In THP-1 cells, ApmCRP3 treatment markedly reduced mCRP-induced expression of inflammatory mediators, including IL-12, IL-6, TNFα, and IL-1β, and inhibited M1 macrophage polarization (Fig. 5F and G). Furthermore, in chondrocytes exposed to mCRP, ApmCRP3 increased the expression of anabolic markers ACAN and COL2A1, while decreasing the levels of catabolic markers MMP-13 and ADAMTS-5 (Fig. S5).

Fig. 5.

Neutralizing effects of ApmCRP3 on mCRP-induced pathological changes in vitro. (A) RT-qPCR analysis for the mRNA levels of MMP-3, IL-6, IL-1β, and CXCL12 in FLSs after treatment with 5 μg/mL mCRP in the presence of NC or ApmCRP3 at a concentration of 1000 nM for 24 h. (B) Viability of FLSs after 4-day treatment with daily 5 μg/mL mCRP in the presence of NC or ApmCRP3 at a concentration of 1000 nM. (C and D) Transwell migration (C) and invasion (D) of FLSs after treatment with 5 μg/mL mCRP in the presence of NC or ApmCRP3 at a concentration of 1000 nM for 48 h. Scale bar = 100 μm. (E) Flow cytometry for detecting the serum starvation-induced apoptosis of FLSs after 6-day treatment with daily 5 μg/mL mCRP in the presence of NC or ApmCRP3 at a concentration of 1000 nM. (F) mRNA levels of IL-12, IL-6, TNF-α, and IL-1β of THP-1 cells after treatment with 5 μg/mL mCRP in the presence of NC or ApmCRP3 at a concentration of 1000 nM for 24 h. (G) Flow cytometry for detecting the M1 macrophage polarization of PMA-pretreated THP-1 cells after treatment with 5 μg/mL mCRP in the presence of NC or ApmCRP3 at a concentration of 1000 nM for 24 h. Each experiment was repeated three times. Data were presented as mean ± SD. P-values from one-way ANOVA (A, C-G) or two-way ANOVA (B): ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

2.7. ApmCRP3 inhibits arthritic development in CIA mice

CIA serves as a traditional RA mouse model [51]. Unlike the aforementioned CAIA model, CIA can be elicited in susceptible mouse strains, such as DBA/1J, through immunization with CII [52]. This results in polyarthritis that shares clinical and histological features with RA [51]. We assessed the therapeutic efficacy of ApmCRP3 in vivo using the CIA mouse model (Fig. 6A). Treatment with ApmCRP3 effectively reduced the arthritis score in CIA mice compared to those treated with PBS or NC (Fig. 6B). μCT analysis demonstrated that ApmCRP3 attenuated bone erosion in CIA mice (Fig. 6C). Histological staining demonstrated less synovial hyperplasia and cartilage damage in CIA mice treated with ApmCRP3 (Fig. 6D and E). IF staining and quantitative data revealed that ApmCRP3 significantly reduced levels of vimentin, CD86, and MMP-13 compared to PBS or NC (Fig. 6F–I). It has been reported that 1,6-bis(phosphocholine)-hexane (1,6-bisPC) is a synthetic phosphocholine analog that stabilizes the pentameric structure of nCRP and prevents its dissociation into the mCRP [53]. We observed that administration of 1,6-bisPC also showed a therapeutic effect in CIA mice, but to a lesser extent than ApmCRP3 (Fig. 6B–I). Serum biochemical assays were performed to test levels of liver and kidney function parameters, including alanine aminotransferase (ALT), aspartate aminotransferase (AST), and blood urea nitrogen (BUN). The mice treated with ApmCRP3 showed no significant change in levels of ALT, AST, and BUN compared to the mice in other treatment groups (Fig. S6A). Histological analysis of major organs, including the heart, liver, spleen, lung, and kidney, demonstrated no obvious tissue damage resulting from ApmCRP3 treatment (Fig. S6B and S6C).

Fig. 6.

Therapeutic effects of ApmCRP3 in CIA mice. (A) Schematic diagram of experimental design using CIA mice. Briefly, CIA mice were treated with PBS, 1,6-bisPC, NC, or ApmCRP3 every 3 days for 45 days. NC or ApmCRP3 was intravenously injected at a dose of 25 mg/kg, and 1,6-bisPC was intraperitoneally administered at a dose of 3 mg/kg. (B) Arthritis scoring of NI or CIA mice from different treatment groups. (C) μCT scans of hind paws of NI or CIA mice from different treatment groups, with black arrows indicating the bone erosion. (D) H&E staining of paw sections of NI or CIA mice from different treatment groups. Scale bar = 300 μm. (E) SO&FG staining of paw sections of NI or CIA mice from different treatment groups. Scale bar = 300 μm. (F) IF staining of paw sections of NI or CIA mice from different treatment groups, using an anti-vimentin, an anti-CD86, or an anti-MMP-13 antibody. Scale bar = 50 μm. (G–I) Quantification of expression of Vimentin (G), CD86 (H), and MMP-13 (I) on IF-stained sections. Data were represented as mean ± SD of n = 6 per group. P-values from one-way ANOVA (G–I) or two-way ANOVA (B): ∗p < 0.05, ∗∗∗p < 0.001.

2.8. ApmCRP3 attenuates OA progression in DMM mice

To assess the therapeutic potential of ApmCRP3 in the DMM mouse model of OA, we administrated DMM mice with PBS, 1,6-bisPC, NC, or ApmCRP3 (Fig. 7A). μCT analysis demonstrated that ApmCRP3 attenuated osteophyte formation in DMM mice compared to PBS- or NC-treated mice (Fig. 7B). Histological staining showed decreased synovial hyperplasia and cartilage erosion in DMM mice treated with ApmCRP3 (Fig. 7C and D). IF staining revealed that ApmCRP3 lowered levels of vimentin and CD86, while increasing the expression of COL2A1 (Fig. 7E). Quantitative data of OARSI score, cartilage area, osteophyte score, synovitis score, vimentin-positive cells, CD86-positive cells, and COL2A1-positive cells consistently suggested that ApmCRP3 significantly attenuated DMM-induced OA progression (Fig. 7F–L). Notably, although 1,6-bisPC also exhibited therapeutic efficacy compared to PBS or NC, its overall impact on DMM-associated pathology was less pronounced than the effects achieved by ApmCRP3 (Fig. 7B–L). Serum biochemical analysis showed that ApmCRP3 treatment did not significantly alter ALT, AST, or BUN levels compared to other groups (Fig. S7A). Histological examination of major organs revealed no apparent tissue damage induced by ApmCRP3 administration (Fig. S7B and S7C).

Fig. 7.

Therapeutic effects of ApmCRP3 in DMM mice. (A) Schematic diagram of experimental design using DMM mice. Briefly, DMM mice were treated with PBS, 1,6-bisPC, NC, or ApmCRP3 every 3 days for 49 days. NC or ApmCRP3 was intravenously injected at a dose of 25 mg/kg, and 1,6-bisPC was intraperitoneally administered at a dose of 3 mg/kg. (B) μCT scans of knee joints of sham or DMM mice from different treatment groups, with black arrows indicating the osteophyte formation. (C) H&E staining of joint sections of sham or DMM mice from different treatment groups. Scale bar = 300 μm. (D) SO&FG staining of joint sections of sham or DMM mice from different treatment groups. Scale bar = 300 μm. (E) IF staining of joint sections of sham or DMM mice from different treatment groups, using an anti-vimentin, an anti-CD86, or an anti-COL2A1 antibody. Scale bar = 50 μm. (F–I) Quantification of OARSI score (F), cartilage area (G), osteophyte score (H), and synovitis score (I) of sham or DMM mice from different treatment groups. (J–L) Quantification of expression of vimentin (J), CD86 (K), and COL2A1 (L) on IF-stained sections. Data were represented as mean ± SD of n = 6 per group. P-values from one-way ANOVA (F–L): ∗∗∗p < 0.001.

3. Discussion

The liver–bone axis is increasingly recognized as a vital mediator in the pathogenesis of arthritis, particularly in RA and OA. This axis plays a pivotal role in orchestrating the complex interplay between systemic inflammation and local joint degradation [8,9]. In our study, we employed a combination of genetic and pharmacological strategies, including a hepatocyte-specific knockout mouse model and aptamer technology, to underscore the critical involvement of CRP within this axis. Our findings reveal that the distinct forms of CRP (mCRP and nCRP) might exert divergent effects on the disease progression of RA and OA. Mechanistically, mCRP directly drives pathogenic remodeling of key effector cells in RA and OA, including FLSs, monocytes/macrophages, and chondrocytes, through FcγRI-mediated signaling within the inflammatory joint microenvironment. In contrast, nCRP demonstrated negligible protective effects. This corroborates previous literature that highlights the pro-inflammatory nature of mCRP and the modest anti-inflammatory functions of nCRP [48,[54], [55], [56], [57]]. Structurally, nCRP forms a cyclic pentamer composed of five identical subunits in a discoid arrangement. mCRP is generated through non-covalent dissociation of nCRP under conditions such as low pH, oxidative stress, or exposure to urea/EDTA, leading to the exposure of neoepitopes with distinct binding properties and biological activities [23]. Functionally, mCRP activates pro-inflammatory pathways via FcγRI, FcγRIIIb, and lipid raft microdomains [46,47,58,59]. We focused specifically on the FcγRI axis due to its high affinity for mCRP and predominant expression on key effector cells in arthritis pathogenesis [[40], [41], [42], [43]]. In contrast, nCRP signals mainly through FcγRII and often exerts limited or even anti-inflammatory effects [58,60,61], which aligns with our findings that nCRP treatment resulted in lower pro-inflammatory cytokine mRNA levels compared to PBS. However, the precise molecular pathways and regulatory mechanisms governing this duality are still not fully elucidated and warrant further investigation.

CRP levels are a hallmark of elevation in both OA and RA [62], yet its role within arthritis remains a subject of intense debate. Conflicting data from various animal models have emerged over the past decades. For instance, whole-body CRP knockout mice have shown exacerbated arthritis, suggesting a protective function for CRP [63]. Similarly, systemic overexpression or exogenous CRP administration in mice has demonstrated therapeutic potential in alleviating arthritis symptoms [64,65]. Conversely, whole-body CRP knockout in rat models has been reported to reduce arthritis severity, implying a potentially detrimental role for CRP [66,67]. We propose that these inconsistent findings might be due to the absence of liver-specific CRP knockout models and the failure to differentiate between CRP isoforms. Taken together, our study addresses these discrepancies by utilizing a hepatocyte-specific knockout strategy alongside aptamer-directed targeting of mCRP, affirming the causal role of CRP, especially mCRP, in the pathogenesis of RA and OA. These insights pave the way for refining CRP-targeted therapeutic strategies, offering valuable perspectives into its role in arthritic conditions.

The therapeutic landscape for CRP inhibition has traditionally focused on small molecules like 1,6-bisPC, which prevents the conversion of nCRP to mCRP, primarily within cardiovascular contexts [68]. However, this approach is limited by challenges such as short half-life and low binding affinity [69]. Recent efforts have attempted to develop monoclonal antibodies against mCRP, yet these are not commercially available [55]. This limitation opens a promising opportunity for nucleic acid aptamers, often referred to as chemical antibodies [49]. Aptamers offer significant advantages, including high specificity and stability, ease of modification, low immunogenicity, and cost-effective synthesis, positioning them as viable alternatives to traditional antibodies and small molecules for therapeutic interventions [50]. In our study, we developed an aptamer, ApmCRP3, specifically targeting mCRP rather than nCRP. ApmCRP3 demonstrated robust efficacy in neutralizing mCRP in vitro and effectively curtailing arthritis progression in both RA and OA mouse models. Furthermore, ApmCRP3 provided superior therapeutic benefits compared to 1,6-bisPC.

Our study has several limitations. First, chemical modifications such as 2′-O-methylation, 2′-fluoro substitution, locked nucleic acids, phosphorothioate linkages, PEGylation, or cholesterol conjugation are commonly employed to enhance the nuclease resistance and pharmacokinetic properties of aptamers [49,50]. However, no chemical modifications were introduced to ApmCRP3, as the current work was intended primarily as a proof-of-concept study. Future efforts will explore such modifications to optimize its pharmacological potential. Second, replenishing the deleted protein in knockout models is a well-established strategy to verify the causal role of a specific molecule. Although CRP-deficient mice in our study exhibited reduced arthritis severity, we did not reintroduce mCRP to functionally confirm its pathogenic role. Future studies will incorporate intra-articular or systemic mCRP administration to address this mechanistic question. Third, a dose of 25 mg/kg was chosen for in vivo administration with reference to previously published aptamer studies [70,71]. Nevertheless, a comprehensive pharmacokinetic and dose–response evaluation remains necessary to define the optimal therapeutic window and ensure safety prior to clinical translation.

In conclusion, our study reveals the significance of targeting the liver–bone axis, specifically through mCRP, as a promising therapeutic strategy in RA and OA. The development of aptamer-based interventions offers a proof-of-concept approach with substantial potential for improving clinical outcomes of arthritis.

4. Materials and methods

4.1. Cell culture

Human primary FLSs were obtained from Jennio Biotech (China) and maintained in a DMEM medium supplemented with 20 % fetal bovine serum and 1 % penicillin-streptomycin. The human monocytic cell line, THP-1 (CL-0233), was obtained from Procell (China) and cultured in a 1640 medium supplemented with 10 % fetal bovine serum and 1 % penicillin-streptomycin. Mouse primary chondrocytes were isolated from neonatal C57BL/6 mouse articular cartilage (postnatal day 3–5) and maintained in a DMEM/F12 medium supplemented with 10 % fetal bovine serum and 1 % penicillin-streptomycin. All cells were incubated at 37 °C with 5 % CO₂ and 95 % humidity and tested negative for mycoplasma contamination.

4.2. Animals

CRP^flox/flox and Alb-Cre-ER^T2 mice were obtained from CyagenBiosciences Co., Ltd. All transgenic mice were bred on a C57BL/6J background. We bred CRP^flox/flox mice with the Alb-Cre-ER^T2 mice to generate CRP^flox/flox; Alb-Cre-ER^T2 mice. Following TAM (100 mg/kg body weight) induction, CRP^ΔHep/ΔHep were generated and used to investigate the in vivo roles of CRP in CAIA and DMM models. Additionally, wild-type male C57BL/6J and DBA/1J mice (6–8 weeks old, 20–25 g) were purchased from GemPharmatech Co., Ltd., and were used for the establishment of CIA and DMM models. All the mice were housed in the Laboratory Animal Center at Southern University of Science and Technology, in a specific pathogen-free (SPF) mouse facility (room temperature, 20–22 °C; room humidity, 40–60 %) with food/water access ad libitum under a 12 h light/dark cycle. The research protocols conducted in this study were approved by the Institutional Animal Care and Use Committees of the Southern University of Science and Technology. Primers for the validation of Alb-CreER^T2, CRP^flox/flox, and CRP^ΔHep/ΔHep mice were listed in Table S1.

4.3. CAIA mouse model

The CAIA mouse model was induced following a previously established protocol [28]. 5 mg/mouse of the CAb cocktail was injected on day 0, followed by an LPS injection (50 μg) on day 3 into C57BL/6J mice. Arthritis progresses rapidly, and acute inflammation peaks on day 7–10 and persists for almost 2 weeks. Two independent, blinded observers performed the scoring to evaluate the severity of arthritis. The scoring system ranged from 0 to 4: 0 = normal; 1 = mild but definite redness and swelling of the ankles or redness and swelling of any degree in any single digit; 2 = moderate to severe redness and swelling of the ankles; 3 = redness and swelling of the entire foot including the digits; and 4 = maximally inflamed limb, with involvement of multiple joints. A score of one or above was considered indicative of arthritis [72].

4.4. DMM mouse model

DMM surgery was performed on the right knee to induce OA, as previously described [3]. Briefly, mice were anesthetized using 1.2 % tribromoethanol. The anteromedial meniscotibial ligament and the medial collateral ligament were transected to induce joint instability. Sham-operated mice served as controls, in which no ligament transection was performed [73].

4.5. CIA mouse model

The CIA mouse model was induced following a previously established protocol [74]. Bovine CII (Chondrex, USA) was emulsified with an equal volume of Complete Freund's adjuvant (CFA). The DBA/1J mice were immunized with a single subcutaneous injection of 100 μL emulsion (containing 100 μg of collagen) at the base of the tail. The emulsion also contained a final concentration of 2 mg/mL of Mycobacterium tuberculosis. Two independent, blinded observers performed the scoring to evaluate the severity of arthritis. The scoring system ranged from 0 to 4: 0 = normal; 1 = redness and/or swelling in one joint; 2 = redness and/or swelling in more than one joint; 3 = redness and/or swelling in the entire paw; and 4 = deformity and/or ankylosis. A score of one or above was considered indicative of arthritis [75].

4.6. μCT and histological analysis

The knee and paw tissues were dissected to remove surrounding soft tissues and fixed in 10 % formalin overnight. Subsequently, the analysis was conducted using Skyscan 1276 high-resolution (10 μm) μCT scanner (Bruker, Germany) with 60 kVp source and 100 μAmp currents. To generate three-dimensional (3D) images of the joints, the μCT data from each group were processed using the same thresholds. For histological analysis, the joints were fixed in 10 % buffered formalin, followed by decalcification in 12 % EDTA solution. After decalcification, knee and paw tissues were embedded in paraffin and sectioned at a thickness of 5 μm. H&E staining, as well as SO&FG staining, was performed to visualize and analyze the histological characteristics of the samples [35].

4.7. H&E staining

The isolated tissues and organs, including the knees, paws, hearts, lungs, livers, kidneys, and spleens, were fixed in 10 % neutral-buffered formalin followed by embedding in paraffin (knee and paw samples were decalcified using 10 % EDTA (pH 7.4) at 4 °C for 40 days prior to embedding). 5-μm sections were made, mounted on glass slides, deparaffinized, and subjected to H&E staining using standard protocols. After mounting with coverslips, the specimens were viewed and blindly analyzed by a pathologist under a light microscope (HAMAMATSU, NanoZoomer S60, Japan) [76].

4.8. SO&FG staining

The isolated knees and paws were fixed in 10 % neutral-buffered formalin and decalcified using 10 % EDTA (pH 7.4) at 4 °C for 40 days prior to paraffin embedding. After embedding, 5-μm sections were prepared, mounted on glass slides, and deparaffinized. SO&FG staining was then performed according to standard protocols to assess cartilage integrity. Briefly, the sections were stained with Weigert's iron hematoxylin, followed by Fast Green for background staining, and counterstained with Safranin O to visualize proteoglycan-rich cartilage. After dehydration and mounting, stained sections were observed under a light microscope (HAMAMATSU, NanoZoomer S60, Japan) and blindly evaluated by a pathologist [77].

4.9. Histological qualitative analysis

The qualitative analysis was performed as previously reported. Briefly, the histological grading systems were applied for the semi-quantitative analysis of the damages in the sections of the heart (e.g., morphological changes in myocardial cells), liver (e.g., the changes in hepatocytes and hepatocyte rows, as well as the presence of steatosis, fibrosis, necrosis, macrophages, and lymphocytes), spleen (e.g., inflammation, necrosis/abscess formation and thrombus formation), lung (e.g., the inflammatory changes of the lung tissue, hemorrhage in the alveolar cavity, and edema and thickening of the alveolar wall), and kidney (e.g., the changes in glomeruli, tubular structure, epithelial cells, and medulla, as well as the presence of fibrin deposition, vascular dilation, inflammatory cells, fibrosis and necrosis), respectively [70].

4.10. Serum biochemistry assays

Serum was obtained by centrifugation of whole blood at 13000×g at 4 °C for 10 min. AST, ALT, and BUN levels were detected using an automatic biochemistry analyzer (MS-480, MedicalSystem) following the manufacturer's instructions [70].

4.11. IF staining

5 μm sections of the samples were permeabilized using 0.2 % Triton X-100 (Beyotime, CN). Subsequently, the sections were blocked with 1 % BSA, followed by another incubation with 0.2 % Triton X-100. Next, the sections were stained overnight at 4 °C with either anti-CRP antibody (Proteintech, Cat# 24175-1-AP, CN), anti-vimentin antibody (Proteintech, Cat# 10366-1-AP, CN), anti-CD86 antibody (Proteintech, Cat# 13395-1-AP, CN), anti-COL2A1 antibody (Proteintech, Cat# 28459-1-AP, CN), or MMP-13 antibody (Proteintech, Cat# 18165-1-AP, CN). After rinsing with PBS for 15 min, the sections were incubated with a goat anti-rabbit Alexa Fluor 488 secondary antibody (Abcam, Cat# A11008, UK) for 1 h at room temperature. To visualize nuclei, the slides were mounted using a mounting medium containing DAPI (Beyotime, CN). Images were captured using a confocal fluorescence microscope (Zeiss LSM980, Germany) [78].

4.12. ELISA

Quantitative sandwich ELISA kits for IL-6 (Epizyme, Cat# HJ064/HJ182, CN), IL-1β (Epizyme, Cat# HJ177, CN) and CRP (Abcam, Cat# ab222511, UK) were used according to the instructions of the manufacturers. Briefly, assay buffer, samples, and antibody cocktail were added into a pre-coated 96-well microplate and incubated at room temperature. After several washes, the detection solution and stop solution were added sequentially. The absorbance was then measured at 450 nm using a microplate reader (PerkinElmer, EnSpire®, USA) [79].

4.13. nCRP and mCRP preparation

nCRP was purchased from Sino Biological Co., Ltd. nCRP was converted to mCRP using a urea/EDTA-based denaturation method as previously described with minimal modifications. Briefly, nCRP was incubated in a denaturing solution containing 8 M urea and 10 mM EDTA (pH 8.0) at 4 °C overnight to disrupt its pentameric structure. To ensure complete denaturation, the urea/EDTA solution was replaced with fresh solution every 3–4 h during incubation. The denatured protein solution was subsequently subjected to stepwise dialysis against PBS (pH 7.4) at 4 °C to eliminate residual urea and promote the stabilization of the monomeric CRP conformation. The resulting mCRP was immediately used for downstream applications or stored at 4 °C for short-term use. To verify the conformational change, samples were analyzed by non-denaturing polyacrylamide gel electrophoresis, and the mobility shift between nCRP and mCRP was assessed [80].

4.14. RT-qPCR

Total RNA was isolated using the TransZol Up Plus RNA Kit (TransGen, Cat# ER501-01-V2, CN). cDNA was synthesized by PrimeScript™ RT reagent Kit (Takara, Cat# RR037A, USA). Then, the gene expression was measured by the CFX96 Touch instrument (Bio-Rad, US) with the Green qPCR SuperMix (TransGen, Cat# AQ601-01-V2, CN) [35]. The specific primers used are listed in Table S2.

4.15. Cell viability assay

FLSs (2 × 10³ cells/well) were first seeded in the 96-well plate overnight to measure cell viability. Cells were treated with mCRP, nCRP, NC, or ApmCRP3 at an appropriate concentration. Then, 10 μL cell counting kit-8 (CCK-8) (MCE, Cat# HY-K0301, USA) solution and 90 μL DMEM (Corning, Cat# 10-013-CVRC, USA) were added to every well at an appropriate time. After incubating at 37 °C for 2 h, the absorbance was measured at 450 nm using a spectrometer (PerkinElmer, EnSpire®, USA) [70].

4.16. Transwell assay

FLSs were harvested and resuspended in DMEM without FBS. Cells (5 × 10³ cells/well) were added to the upper chambers of 24-well transwell plates with 8 μm pore polyester membranes (Jet, Cat# TCS013024, CN). For invasion assays, cells (8 × 10³ cells/well) were seeded into the top chamber pre-coated with Matrigel (Corning, Cat# 354234, USA). DMEM containing 20 % FBS was added to the lower chamber to induce cell migration or invasion. Cells were treated with mCRP, nCRP, NC, or ApmCRP3 at an appropriate concentration. After 48 h incubation, cells that had migrated through the membrane were fixed with 4 % paraformaldehyde (Beyotime, Cat# P0099, CN), stained with 0.1 % crystal violet (Aladdin, Cat# C110703, CN), and visualized using an inverted light microscope (Nikon ECLIPSE Ts2, Japan) [77].

4.17. Apoptosis assay

FLSs were seeded in 6-well culture plates. Cells were treated with mCRP, nCRP, NC, or ApmCRP3 at an appropriate concentration. After 5 days, the cells were harvested and stained using the Annexin V-FITC Apoptosis Detection Kit (Beyotime, Cat# C1062L, CN) according to the manufacturer's instructions. Flow cytometry was performed using a BD FACSCanto SORP flow cytometer (BD Biosciences, USA), and data were analyzed with FlowJo software (Tree Star, USA) [77].

4.18. M1 polarization

THP-1 monocytes were cultured in DMEM with 10 % FBS and differentiated into M0 macrophages using 100 ng/mL PMA (Sigma–Aldrich, USA) for 72 h. After serum starvation for 12 h, cells were treated with mCRP, nCRP, NC, or ApmCRP3 for 24 h. Cells were then harvested and blocked with Fc buffer. Surface marker CD86 was stained directly with anti-human CD86 antibody (Proteintech, Cat# 13395-1-AP, CN). For intracellular CD68 staining, cells were fixed with 4 % paraformaldehyde, permeabilized with 0.1 % Triton X-100, and incubated with anti-human CD68 antibody (Proteintech, Cat# 28058-1-AP, CN). After washing, cells were incubated with APC-conjugated secondary antibodies (Abcam, Cat# ab130820, UK) and FITC-conjugated secondary antibodies (Abcam, Cat# ab6858, UK) and analyzed using a BD FACSCanto SORP flow cytometer (BD Biosciences, USA). Data were processed with FlowJo software (Tree Star, USA) [35].

4.19. Western blotting

Total protein was quantified using a BCA assay kit (Thermo Scientific, Cat# 23225, USA). Protein samples were separated by SDS-PAGE and transferred onto PVDF membranes (Millipore, Cat# IPVH00010, USA). Membranes were blocked with 5 % skim milk (BD, Cat# 232100, USA) and incubated overnight at 4 °C with the following primary antibodies: anti-ACAN (Proteintech, Cat# 13880-1-AP, CN), anti-COL2A1 (Proteintech, Cat# 28459-1-AP, CN), anti-MMP-13 (Proteintech, Cat# 18165-1-AP, CN) and anti-ADAMTS-5 (Proteintech, Cat# 11865-1-AP, CN). After washing, the membranes were incubated with HRP-conjugated secondary antibodies (ABclonal, Cat# AS003/AS014, CN). Immunodetection was performed using an enhanced chemiluminescence (ECL) kit (ABclonal, Cat# RM00021P, CN), and signals were visualized using a chemiluminescence imaging system (Tanon, Multi5200, CN) [81].

4.20. RNA sequencing

RNA sequencing was conducted by Sangon Biotech (Shanghai) Co., Ltd. (Shenzhen, CN). Total RNAs (∼1 μg) extracted from each cell sample were used for RNA-seq library preparation. The raw sequencing data were first evaluated using FastQC for quality assurance. To obtain high-quality reads, the data were trimmed using Trimmomatic. These high-quality reads were then aligned to the reference genome using HISAT2, and mapping statistics were compiled. Subsequently, featureCounts was integrated into the pipeline to quantify gene expression. It quickly and accurately counted the number of reads mapped to each gene or feature, using the aligned read files as input and the gene annotation in GTF format to assign reads to the corresponding genes. Finally, for the differential expression analysis, the read counts generated by featureCounts were used to perform volcano plots and enrichment analysis, specifically focusing on differential mRNA in an R or Python environment for statistical computing and graphics.

4.21. SELEX

Protein-SELEX was performed as previously described [82]. Briefly, His-tagged mCRP was first incubated with the initial ssDNA library for positive selection. After discarding unbound sequences, the bound sequences were eluted by heat treatment. For negative selection, nCRP beads were incubated with the eluted sequences to remove nonspecific binders. The unbound sequences were then amplified by PCR to generate a new ssDNA library for the next round of selection. After multiple rounds of selection, the enriched ssDNA pool was cloned into TOP10 chemically competent Escherichia coli using a TA cloning kit (Invitrogen, Cat# K450002, USA), and sequenced by Sangon Biotech Co., Ltd. (Shanghai, CN). The binding assay of enriched ssDNA to mCRP was performed using a BD FACSCanto SORP flow cytometer (BD Biosciences, USA). Finally, four candidate aptamers targeting mCRP were obtained and listed in Table S3. Secondary structures were predicted using the RNAstructure software.

4.22. Binding assay

For aptamers binding to CRP, briefly, 100 pmol of mCRP or nCRP protein was immobilized onto NHS-activated magnetic beads (Thermo Fisher Scientific) according to the manufacturer's protocol. After blocking with 3 % BSA, the beads were incubated with 500 pmol of enriched ssDNA libraries (Round 0, 4, 8, 12, and 16) or gradient concentrations of aptamer candidate sequences (0–1000 pmol) in binding buffer (PBS supplemented with 1 mM MgCl₂ and 0.05 % Tween-20) for 60 min at room temperature with gentle rotation. Following incubation, the beads were washed three times with binding buffer to remove unbound sequences. Bound DNA was then eluted by heating at 95 °C for 5 min and quantified using a Qubit ssDNA assay kit (Invitrogen). For CRP binding to aptamers, 500 pmol of biotin-labeled ssDNA aptamer library from SELEX Rounds 0, 4, 8, 12, and 16 was immobilized on streptavidin-coated magnetic beads (Thermo Fisher) in binding buffer (PBS containing 1 mM MgCl₂ and 0.05 % Tween-20). After incubation and blocking, 500 pmol of either mCRP or nCRP protein was added and incubated at room temperature for 60 min with gentle rotation. Following incubation, the beads were washed thoroughly to remove unbound protein. The amount of CRP bound to the aptamer was then quantified using a human CRP ELISA kit (Abcam) according to the manufacturer's instructions.

4.23. Dot blotting

Dot blot assay was performed using a chemiluminescent biotin-labeled nucleic acid detection kit (Beyotime, Cat# D3308, CN) according to the manufacturer's instructions. mCRP was diluted to 200 ng and spotted onto a nitrocellulose membrane. The membrane was air-dried, blocked with 5 % non-fat dry milk for 1 h at room temperature, and incubated with biotin-labeled aptamer overnight at 4 °C. After washing and re-blocking, the membrane was incubated with streptavidin-HRP conjugate. Signal detection was performed using the BeyoECL Moon A and B solutions (Beyotime, CN), and chemiluminescence was visualized using a Tanon Multi5200 imaging system (Tanon, CN) [83].

4.24. Statistical analysis

All numerical data are expressed as the mean ± SD. A two-sided unpaired t-test was employed to compare the two independent groups. Comparisons among three or more independent groups were analyzed using one-way ANOVA followed by Dunnett's test. To compare the mean differences between groups that have been split on two independent variables, a two-way ANOVA followed by Tukey's multiple comparisons test was performed. P < 0.05 was considered statistically significant. All statistical analyses were performed with GraphPad Prism 9 software. We chose representative images based on the average or median level of the data for each group. For the in vivo experiments, the sample size was pre-determined by a power calculation. The mice were grouped randomly and blindly by researchers. The mice in poor body condition before the experiments were excluded.

Ethics statement

All animal experiments were approved by the Ethics Committee of Southern University of Science and Technology (SUSTech-JY202107022).

Author contributions

C.L. and Y.D. conceived and designed the study. Z.W., D.X., and P.Z. performed key cell experiments and animal administration. C.C. investigated the related literature. J.G. performed the aptamer screening. Z.W. wrote the manuscript. C.L. revised the manuscript. C.L. and A.L. provided supervision for the study.

Declaration of competing interest

The authors declare the following competing financial interest(s): A patent application has been filed related to this work by The Shenzhen LingGene Biotech Co., Ltd.

Abbreviations

ACAN

aggrecan

ADAMTS-5

a disintegrin and metalloproteinase with thrombospondin motifs-5

Alb

albumin

ALT

alanine aminotransferase

ANOVA

analysis of variance

ApmCRP3

aptamer targeting monomeric C-reactive protein 3

AST

aspartate aminotransferase

BUN

blood urea nitrogen

CAIA

collagen antibody-induced arthritis

CCK-8

Cell Counting Kit-8

CIA

collagen-induced arthritis

COL2A1

collagen type II α1 chain

CRP

C-reactive protein

DAPI

4′,6-diamidino-2-phenylindole

DMM

destabilization of the medial meniscus

ECL

enhanced chemiluminescence

EDTA

ethylenediaminetetraacetic acid

ELISA

enzyme-linked immunosorbent assay

FBS

fetal bovine serum

FcγRI

Fc gamma receptor I

FLSs

fibroblast-like synoviocytes

IF

immunofluorescence

IgG

immunoglobulin G

IL

interleukin

IP

immunoprecipitation

mCRP

monomeric C-reactive protein

MMP

matrix metalloproteinase

nCRP

native C-reactive protein

OA

osteoarthritis

PBS

phosphate-buffered saline;

PMA

phorbol 12-myristate 13-acetate

PVDF

polyvinylidene fluoride;

RA

rheumatoid arthritis

RT-qPCR

reverse transcription-quantitative polymerase chain reaction

SAA

serum amyloid A

SD

standard deviation

SELEX

systematic evolution of ligands by exponential enrichment

SO&FG

safranin O and fast green

TAM

tamoxifen

TLR

Toll-like receptor

TNF

tumor necrosis factor

μCT

micro-computed tomography

Appendix A. Supplementary data

The following are the Supplementary data to this article.

Data availability

The datasets generated during the current study are not publicly available due to the data also forming part of our ongoing study but are available from the corresponding author on reasonable request.