Hydrogel microspheres loaded with sinomenine and Drynaria rhizome enhance the treatment of rheumatoid arthritis via immune regulation and promoting bone repair

Abstract

Rheumatoid arthritis (RA) is an autoimmune disease characterized by chronic synovial inflammation, cartilage destruction, and bone loss. Current therapeutic approaches are often limited by short drug half-life, insufficient local drug release, and substantial systemic side effects.In this study, we developed a composite thermosensitive hydrogel system that integrates in situ gelation, sustained drug release, and multitarget therapeutic effects for localized precision treatment of RA.The system consists of a thermosensitive hydrogel matrix composed of hydroxypropyl methylcellulose (HPMC), hyaluronic acid (HA), and glycerol, in which gelatin methacryloyl (GelMA) hydrogel microspheres are embedded. The microspheres efficiently encapsulate Drynaria rhizome–derived extracellular vesicles (DR-EVs), while sinomenine is incorporated into the thermosensitive hydrogel to enhance anti-inflammatory activity. Characterization by transmission electron microscopy (TEM), scanning electron microscopy (SEM), nanoparticle tracking analysis (NTA), rheological measurements, and Fourier-transform infrared spectroscopy (FTIR) confirmed the intact morphology of DR-EVs, the uniform porous structure of the microspheres, and the favorable thermoresponsive gelation behavior and controllable degradation properties of the composite system. Functional assays revealed that, in vitro, the system effectively suppressed TH17 cell proliferation, promoted Treg cell differentiation, and inhibited M1 macrophage polarization.Meanwhile, it upregulated osteogenesis-related genes (Runx2, BMP2) and inhibited osteoclast formation. In a collagen-induced arthritis (CIA) rat model, the system significantly alleviated joint swelling, restored cartilage and bone architecture, and suppressed the progression of synovial inflammation. In summary, this composite thermosensitive hydrogel system possesses injectability, thermoresponsive behavior, prolonged release capability, and multiple biological activities, offering a safe, efficient, and controllable novel strategy for localized precision therapy of RA.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-026-04063-4.

Introduction

RA is a chronic autoimmune disease characterized by the presence of specific autoantibodies, persistent synovitis, and progressive destruction of bone and cartilage [1]. RA affects approximately 0.5–1% of the global population and is one of the leading causes of disability and loss of work capacity worldwide [2]. The lack of a comprehensive understanding of RA’s complex pathogenesis has significantly limited the development of effective clinical therapies. A considerable proportion of RA patients exhibit poor responses or intolerance to conventional disease-modifying antirheumatic drugs (DMARDs), and even with the advent of biologics and small-molecule targeted therapies, over 30% of patients fail to achieve satisfactory disease control [3]– [4]. Therefore, there is an urgent need to develop novel and effective therapeutic strategies for RA.

Persistent synovitis is a hallmark of RA, with lymphocytes and macrophages playing central roles in the immunopathological microenvironment [5]. Increasing evidence indicates that the imbalance between T helper 17 (Th17) cells and regulatory T (Treg) cells is a critical factor in sustaining chronic inflammation in RA [6]. Th17 cells exacerbate inflammation through secretion of proinflammatory cytokines such as IL-17 and IL-22, while Treg cells exert immunosuppressive effects by producing IL-10 and TGF-β. In addition, monocytes and macrophages are actively involved in immune regulation and bone remodeling within the RA synovium [7]. Synovial macrophages in RA can be polarized by cytokines in the joint microenvironment into pro-inflammatory (M1) and anti-inflammatory (M2) phenotypes. The M1/M2 ratio in the synovial fluid of RA patients is significantly higher than in patients with osteoarthritis (OA) [8]. M1 macrophages are major sources of TNF-α, IL-6, IL-1β, and the chemokine CCL2, which initiate and sustain inflammatory responses, and they exhibit greater potential for differentiation into osteoclasts (OCs) compared to M2 macrophages [9]. Conversely, anti-inflammatory cytokines such as IL-4, IL-10, and TGF-β promote M2 polarization, favoring a Th2 immune response and mitigating inflammation [10]. Thus, immunomodulation through rebalancing immune cell populations represents a promising strategy for RA therapy.

Alongside synovitis, destruction of cartilage and subchondral bone is another major pathological feature of RA and remains a significant therapeutic challenge [11]. RA-induced bone erosion is primarily driven by overactivated osteoclasts that are stimulated by inflammatory cytokines (e.g., TNF-α, IL-1, IL-17) to upregulate RANKL expression, thereby enhancing bone resorption. Simultaneously, the inflammatory microenvironment inhibits the Wnt/β-catenin signaling pathway, suppressing osteoblast (OB) differentiation and disrupting the balance of bone remodeling, further aggravating skeletal damage [12]. In cartilage, inflammatory stimuli impair matrix synthesis, promote matrix degradation, and accelerate chondrocyte apoptosis, collectively leading to cartilage loss and damage. Although current therapies such as biologics and DMARDs have demonstrated efficacy in controlling inflammation, their ability to promote bone regeneration and protect cartilage remains limited [13]. Therefore, the development of advanced biomaterials that can both modulate inflammation and enhance osteochondral repair is of great clinical importance.

Natural products have emerged as promising sources of novel therapeutics due to their broad bioactivities and relatively low toxicity. Recent studies have shown significant potential for natural compounds in RA treatment [14]. For instance, sinomenine (SIN), an isoquinoline alkaloid with potent anti-inflammatory and immunosuppressive properties, has been approved in China as an RA therapeutic agent [15]. SIN has been shown to regulate immune responses by targeting a variety of immune cells (T cells, macrophages, dendritic cells, mast cells) and immune-related mediators (cytokines, reactive oxygen species, NF-κB, adhesion molecules) [16]. Another traditional Chinese medicine, Drynariae Rhizoma (DR), is widely used for its analgesic and bone-healing properties. Its active components, including flavonoids and steroidal compounds, have demonstrated the ability to promote osteogenic differentiation of mesenchymal stem cells, stimulate osteoblast proliferation, and inhibit osteoclast activity, thereby ameliorating RA-related bone destruction [17–19]. However, both SIN and DR suffer from poor bioavailability, low solubility, and non-selective tissue distribution. To overcome the limitations associated with crude plant extracts, plant-derived extracellular vesicles (EVs) have been identified as a promising alternative. These EVs offer superior stability, enhanced cellular uptake efficiency due to their lipid bilayer structure, and deeper penetration into the dense cartilage matrix, making them a more potent therapeutic candidate [20].

Hydrogels, owing to their excellent biocompatibility, controllable drug release capability, and favorable structural tunability, have emerged as ideal carriers for rheumatoid arthritis (RA) drug delivery systems [21]. Among them, gelatin methacryloyl (GelMA) is a naturally derived hydrogel with both cell-adhesive properties and biodegradability, capable of mimicking the natural extracellular matrix (ECM) and providing a favorable microenvironment for osteogenesis and chondrogenesis [22]. Hyaluronic acid (HA), a key component of synovial fluid and cartilage, possesses remarkable lubrication, cushioning, and anti-inflammatory properties [23]. Through synergistic combination with hydroxypropyl methylcellulose (HPMC) and glycerol, HA can acquire thermosensitive gelation properties, rapidly forming a gel at 34 °C, thereby enabling localized drug sustained release and retention [24].

In this study, we developed a novel thermosensitive composite hydrogel composed of HA-encapsulated GelMA microspheres, loaded with sinomenine and Drynariae Rhizoma (HPA@SIN-GM@DR). This hydrogel system was designed for localized, sustained release of active ingredients to restore immune balance in the RA joint microenvironment while promoting bone regeneration. The composite hydrogel exhibits excellent injectability, biodegradability, and targeted retention at inflamed joints. More importantly, it enables multidimensional regulation of the RA pathological microenvironment, effectively suppressing inflammation and facilitating osteochondral repair. As an innovative biomaterial-based therapeutic approach, this system offers a promising and practical solution for the treatment of RA-related bone and cartilage damage, and opens new avenues for precision medicine in RA.

Materials and methods

Materials

Glycerol, and liquid paraffin were purchased from Macklin (Shanghai, China). Hyaluronic acid (HA), hydroxypropyl methylcellulose (HPMC), Methacrylic anhydride (MA), Span 80, collagenase (Type II) and gelatin were obtained from Aladdin Company (Shanghai, China). All other chemicals and reagents used were of analytical grade unless otherwise specified.

For characterization, a transmission electron microscope (TEM, Talos F200S G2, Thermo Fisher Scientific, USA), a scanning electron microscope (SEM, Zeiss Sigma 300, German), a Fourier transform infrared spectrometer (FT-IR, Bruker Tensor 27, German), a nanoparticle size and Zeta potential analyzer (Malvern Zetasizer Nano As, Spectris plc, England) and a rotational rheometer (TA Instruments, DE, UK) were used.

Extraction and characterization of Drynaria roosii-derived extracellular vesicles (DR-EVs)

As illustrated in Fig. 1, fresh Drynaria roosii was homogenized with pure phosphate-buffered saline (PBS, 1:10, g/mL) using a blender at high speed for 10 min, followed by filtration and differential centrifugation to remove impurities. The supernatant was sequentially subjected to differential centrifugation at 500 ×g for 20 min, 2,000 ×g for 30 min, and 10,000 ×g for 50 min at 4 °C to eliminate cellular debris and large particles. Extracellular vesicles were subsequently pelleted by ultracentrifugation at 100,000 ×g for 70 min at 4 °C, resuspended in sterile PBS, and further purified by an additional ultracentrifugation cycle. The final DR-EVs were filtered through a 0.22 μm membrane and stored at −80 °C until use. The morphology of DR-EVs was observed by TEM (Talos F200S G2, Thermo Fisher Scientific, USA), particle size and zeta potential were determined by nanoparticle tracking analysis (Malvern Zetasizer Nano As, Spectris plc, England), and Western blotting for TET8, HSP70 (positive markers), and Calnexin (negative marker) was performed to verify vesicle origin and purity.

Fig. 1.

Schematic depicting the role of HPA@SIN-GM@DR in sustained release of SIN and DR-EVs to suppress inflammation and facilitate RA bone repair

Preparation of GelMA@DR-EVs microspheres

Gelatin was dissolved in PBS at 50 °C, followed by the slow addition of methacrylic anhydride (MA) while maintaining the pH at 8.0. After 3 h of stirring, the reaction was terminated, dialyzed for purification, and lyophilized to obtain the GelMA precursor. The precursor was dissolved in sterile PBS (10% w/v) containing 0.5% (w/v) lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP), and DR-EVs were incorporated at a final concentration of 100 µg/mL with gentle mixing on ice. A microfluidic emulsion method was applied, where the GelMA@DR-EVs solution and the oil phase (liquid paraffin containing 3% Span 80) were combined within a microfluidic chip (flow rate ratio: oil phase 4.8 mL/h, aqueous phase 0.48 mL/h) to generate uniform droplets, the droplets were collected and immediately crosslinked under 365 nm UV irradiation (10 mW/cm²) for 30 s to ensure complete gelation. The cured microspheres were washed alternately with petroleum ether and PBS, and stored at 4 °C until further use.

Preparation of HPA-SIN hydrogel precursor

Hyaluronic acid (HA) at different concentrations (1% or 2% w/v) was dissolved in deionized water, followed by the gradual addition of hydroxypropyl methylcellulose (HPMC, 7% w/v) and continuous stirring for 30 min at 80 °C. The mixture was allowed to hydrate and degas overnight at 4 °C. Glycerol (10% or 20% v/v) was added to regulate rheological properties, and sinomenine (SIN, 2 mg/mL) was incorporated and ultrasonically dispersed to obtain the SIN-loaded thermosensitive hydrogel precursor. Gelation conditions of HPA-based hydrogels with different formulations were assessed using a thermostatic water bath.

Construction of the composite hydrogel system (HPA@SIN-GelMA@DR-EVs)

GelMA@DR-EVs microspheres were dispersed in the cold (4 °C) SIN-loaded thermosensitive hydrogel precursor and gently mixed to achieve uniform distribution. Upon exposure to 36 °C, rapid gelation occurred, resulting in the formation of the composite hydrogel system.

Fourier transform infrared (FT-IR) spectroscopy

FT-IR spectra were recorded using a spectrometer (Bruker Tensor 27, German) within the range of 400–4000 cm⁻¹ to characterize the characteristic absorption peaks of HPA-based thermosensitive hydrogels and the composite system, thereby confirming chemical structures and intermolecular interactions.

Scanning electron microscopy (SEM)

The surface morphology and porous structure of GelMA microspheres were examined using SEM (Zeiss Sigma 300, German) after sputter coating with gold.

Rheological analysis

The storage modulus (G′) and loss modulus (G″) as a function of temperature (20–70 °C) were evaluated using a rotational rheometer (Discovery HR-2 Rheometer, TA Instruments, DE, UK).

Swelling and degradation studies

The swelling behavior of GelMA microspheres was examined by immersing lyophilized microspheres in PBS at 37 °C and periodically recording the weight to calculate the swelling ratio. Degradation studies were conducted in PBS containing 1 U/mL collagenase, with mass loss monitored over time. The degradation rate was calculated using the following formula: W_remaining (%) = (W_t/W₀) ×100%, where W_t is the wet weight at the measured time point and W₀ is the initial wet weight. The degradation of the composite hydrogel system was evaluated in the same manner.

In vitro drug release assay

The in vitro release profiles of SIN and DR-EVs from the composite hydrogel were evaluated in PBS (pH 7.4) at 37 °C. Briefly, 200 µL of HPA@SIN-GM@DR hydrogel was injected into the bottom of a release vial, followed by the addition of 2 mL of PBS. The vials were incubated in a shaker at 37 °C and 100 rpm. At predetermined time points, 1mL of the supernatant was collected and replaced with an equal volume of fresh medium. The concentration of released SIN was quantified using High-Performance Liquid Chromatography (HPLC) with a UV detector set at a detection wavelength of 262 nm. The amount of released DR-EVs was determined using a BCA Protein Assay Kit (Thermo Fisher, USA) by measuring absorbance at 562 nm, and the particle integrity was verified by NTA. Cumulative release percentages were calculated and plotted against time.

Primary cell isolation and culture

Spleens were aseptically harvested, mechanically dissociated, filtered through a 70-µm cell strainer, and treated with ACK lysis buffer to remove red blood cells. CD4⁺ T cells were isolated using a rat CD4⁺ T cell negative selection magnetic bead kit (Miltenyi Biotec), and resuspended in RPMI-1640 supplemented with 10% fetal bovine serum (FBS). Peritoneal macrophages were obtained by PBS lavage of the peritoneal cavity, centrifuged, and cultured in DMEM with 10% FBS at 37 °C for 2 h. Non-adherent cells were removed by washing three times with PBS to obtain adherent macrophages with a purity > 95%. Bone marrow cells were flushed from femurs and tibias, filtered through a 70-µm strainer, and centrifuged. Bone marrow-derived mesenchymal stem cells (BMSCs) were directly seeded into α-MEM containing 15% FBS. Non-adherent cells were removed after 48 h, and adherent cells were passaged every 3 days using 0.25% trypsin. Passages 3 to 5 were used for subsequent experiments. Bone marrow-derived macrophage precursors were seeded in DMEM with 10% FBS and used for experiments after cells adherence.

Cell treatments

Macrophage polarization: Peritoneal macrophages were seeded in 6-well plates (1 × 10⁶ cells/well in DMEM with 10% FBS). After adherence, cells were divided into the following groups: Control (Ctrl; medium only), LPS (100 ng/mL LPS for 24 h), and intervention groups (LPS plus HPA-GM, HPA@SIN-GM, HPA-GM@DR, or HPA@SIN-GM@DR). Cells and supernatants were collected for analysis. T cell differentiation: CD4⁺ T cells were seeded in anti-CD3 pre-coated (5 µg/mL, overnight at 4 °C) 24-well plates (1 × 10⁶ cells/well in RPMI-1640 with 10% FBS). All groups were stimulated with plate-bound anti-CD3 (5 µg/mL). Grouping was as follows: Ctrl (anti-CD28 antibody 2 µg/mL), Th17/Treg induction group (anti-CD28 2 µg/mL + TGF-β 2.5 ng/mL + IL-6 30 ng/mL) [25], and intervention groups (induction conditions plus HPA-GM, HPA@SIN-GM, HPA-GM@DR, or HPA@SIN-GM@DR). Cells were collected for analysis. Osteoclast differentiation: Macrophages were seeded in 12-well plates (4.5 × 10⁴ cells/mL), and replaced with osteoclast induction medium (OCM: 30 ng/mL M-CSF +50 ng/mL RANKL; OriCell^®). Groups: Ctrl (medium only), OCM group, and intervention groups (OCM + HPA-GM, HPA@SIN-GM, HPA-GM@DR, or HPA@SIN-GM@DR). On day 7, TRAP-positive multinucleated cells (≥ 3 nuclei) were counted after TRAP staining. Osteoblast differentiation: BMSCs were seeded in 12-well plates (2 × 10⁴ cells/cm²) and induced using osteoblast induction medium (OBM; MCE). Groups: Ctrl (medium only), OBM group, and intervention groups (OBM plus HPA-GM, HPA@SIN-GM, HPA-GM@DR, or HPA@SIN-GM@DR). ALP staining was performed on day 7 and ARS staining on day 14 under light microscopy.

Cell viability and Live/Dead staining assay

To ensure the reliability of in vitro experiments, the effects of HPA-GM, HPA@SIN-GM, HPA-GM@DR, and HPA@SIN-GM@DR on cell viability were assessed using a CCK-8 assay (Dojindo) before induction. Blank, control, and test wells were set. Isolated CD4⁺ T cells (1 × 10⁵/well), peritoneal macrophages (5 × 10⁴/well), and BMSCs (5 × 10³/well) were seeded into 96-well plates. CD4⁺ T cells and macrophages were tested immediately, while BMSCs were tested after 24 h of adherence. Each well received 10 µL of CCK-8 reagent, and absorbance at 450 nm (reference at 650 nm) was measured after incubation. Cell viability was calculated as: Cell viability (%) = (ODtest − ODblank)/(ODcontrol − ODblank) × 100%.

To visually assess cell survival, a Live/Dead staining assay was performed. BMSCs and macrophages were seeded in 24-well plates and treated with the extraction medium of HPA@SIN-GM@DR for 24 h. Subsequently, the cells were stained using a Calcein-AM/Propidium Iodide (PI) Double Stain Kit (Yeasen, Shanghai, China) according to the manufacturer’s instructions. After incubation at 37℃ for 15 min in the dark, the cells were washed with PBS and imaged using a fluorescence microscope (Leica, Germany). Live cells emitted green fluorescence (Calcein-AM), while dead cells exhibited red fluorescence (PI).

Experimental animals

Male Sprague Dawley (SD) rats (aged 6–8 weeks, SPF grade, weight 180–220 g) were purchased from the Animal Experimental Center of Southern Medical University. Prior to the experiment, all rats were acclimated to the laboratory environment for one week. All animal procedures conformed to the ARRIVE guidelines and were approved by the Experimental Animal Ethics Committee of Southern Medical University (NO. IACUC-LAC-20230210-004).

Induction of Collagen-Induced arthritis (CIA) in rats

CIA was induced as previously described [26]. Bovine type II collagen was emulsified with an equal volume of complete Freund’s adjuvant (final concentration: 1 mg/mL collagen +2 mg/mL inactivated Mycobacterium tuberculosis) and injected intradermally at the base of the tail (0.1 mL/rat) as a primary immunization. On day 14, a booster immunization was given using an emulsion of collagen and incomplete Freund’s adjuvant (0.1 mL/rat). Rats were randomly assigned into six groups: Ctrl, CIA, HPA-GM, HPA@SIN-GM, HPA-GM@DR, and HPA@SIN-GM@DR. On day 14 (post-primary immunization), rats in the treatment groups received a single intra-articular injection of 50 µL of the respective hydrogel formulations into the right ankle joint using a 27G needle. The specific dosages were 2 mg/mL for SIN and 100 µg/mL for DR-EVs within the hydrogel system.

Body weight and arthritis score monitoring

Starting from the day of booster immunization (day 14 post-primary immunization), body weight and arthritis severity were monitored every 3 days [27]. Each paw was scored from 0 to 4 based on severity (0: no swelling; 1: swelling in one toe joint; 2: swelling in multiple toe joints or the paw; 3: swelling up to the ankle; 4: severe swelling/deformity of the entire paw). Total arthritis score per rat was the sum of scores from all four paws. On day 35, rats were euthanized, and ankle joints were harvested for further analysis.

In vivo biodegradation assay

To monitor the degradation profile of the hydrogel in vivo, Cyanine 5 (Cy5) amine-labeled hydrogels were prepared. Rats were anesthetized, and the dorsal hair was removed. Cy5-labeled MicroGel, HPA, and HPA-MicroGel (50µL) were injected subcutaneously into the dorsal region of the rats (n = 3). The fluorescence signal was monitored at predetermined time points (Day 0, 1, 3, 7, 14, and 21) using an In Vivo Imaging System (IVIS). The fluorescence intensity was quantified to evaluate the degradation kinetics of the implanted materials.

Histological staining

Ankle joints were fixed in 4% paraformaldehyde (prepared in PBS), decalcified, paraffin-embedded, and sectioned at 5 μm. Sections were stained with hematoxylin and eosin (H&E), tartrate-resistant acid phosphatase (TRAP), and Safranin O–Fast Green (SOFG). Additionally, H&E staining was performed on sections of the heart, liver, spleen, lung, and kidney tissues.

Micro-CT analysis

Fixed ankle joints were scanned using a high-resolution Micro-CT system (Skyscan 1176). Three-dimensional reconstructions were performed using CTAn software to analyze bone volume/total volume (BV/TV) and other parameters.

Immunofluorescence

Synovial tissues were fixed in 4% paraformaldehyde, decalcified, embedded, and sectioned continuously. After antigen retrieval with citrate buffer and permeabilization with 0.1% Triton X-100, sections were blocked with 5% BSA for 1 h. Primary antibodies against CD86 and RORγt (diluted 1:200) were applied overnight at 4 °C. The next day, sections were incubated with fluorophore-conjugated secondary antibodies for 1 h, counterstained with DAPI, and mounted. Fluorescence was observed under a confocal laser scanning microscope.

Flow cytometry

Differentiated macrophages and T cells were stained for surface or intracellular markers. Macrophage phenotypes were assessed using FITC-conjugated CD86 and PE-conjugated CD206 antibodies to distinguish M1 (CD86⁺) and M2 (CD206⁺) subsets. Th17 cells were identified using CD4 and intracellular IL-17 A staining; Treg cells were identified using CD4, CD25, and intracellular FoxP3 staining. Stained samples were analyzed on a flow cytometer (e.g., BD FACSCanto II) and data processed with FlowJo software.

ELISA

Supernatants from macrophage polarization experiments were collected, and concentrations of TNF-α, IL-6, IL-1β, and IL-10 were measured using ELISA kits (Shanghai Enzyme-linked Biotechnology) according to the manufacturer’s instructions.

Western blot

Treated macrophages were lysed using pre-chilled RIPA buffer (with 1% protease and phosphatase inhibitors) on ice for 30 min. Lysates were centrifuged (12,000 rpm, 10 min, 4 °C), and supernatants were collected. Protein concentrations were determined using a BCA assay, and equal amounts of protein were separated by 10% SDS-PAGE and transferred to PVDF membranes. After blocking with 5% skim milk, membranes were incubated overnight at 4 °C with primary antibodies, followed by HRP-conjugated secondary antibodies for 2 h. Signals were visualized and analyzed using the AIWBwell system (Servicebio).

RT-qPCR

Total RNA from cells or tissues was extracted using TRIzol, purified, and reverse-transcribed into cDNA. Real-time PCR was performed using SYBR Green chemistry. For in vitro assays, expression levels of osteoclast-related genes (TRAP, Ctsk, ITGB3, MMP9) and osteoblast-related genes (ALP, RUNX2, BMP2, ERα) were analyzed. For in vivo analysis, ankle joint synovial tissues were used to assess the expression of inflammatory cytokines (TNF-α, IL-6, IL-1β) and anti-inflammatory cytokines (IL-10). β-actin was used as an internal reference, and gene expression was calculated using the 2^–ΔΔCt method.

Network Pharmacology

Active ingredients and targets of SIN and DR were retrieved from the Traditional Chinese Medicine Systems Pharmacology Database (TCMSP), applying the criteria of oral bioavailability > 30% and drug-likeness ≥ 0.20. RA-related targets were obtained from GeneCards, OMIM, and PharmGKB databases using the keyword “rheumatoid arthritis” Intersection targets were input into STRING (https://string-db.org/) to construct a protein–protein interaction (PPI) network. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were conducted using the clusterProfiler package.

RNA sequencing (RNA-Seq) and bioinformatics analysis

Ankle tissues were isolated from HPA@SIN-GM@DR intervention group vs. CIA group rats respectively. RNA was extracted using Trizol reagent (Invitrogen, Carlsbad, USA) according to the manufacturer’s instructions. Total RNA quantity and purity were analyzed using a Life Invitrogen Qubit 3.0 fluorescence quantitative analyzer (Pinuofei, China) and a NanoDrop spectrophotometer (Thermo, USA). Construct an RNA seq library and sequence it using Illumina PE150 (Pinuofei, China).

RNA-seq read quality was assessed, and low-quality reads were trimmed with fastp (v.0.23.2). Reads were mapped to the Rat genome (mRatBN7.2) using STAR (v.2.7.10a) and quantified with featureCounts (v.2.0.1). Read counts were normalized and differential gene expression was quantified with DESeq2 (v.1.34.0). A log (fold change) larger than one and a false discovery rate cutoff of 5% was used to select significantly over-represented and under-represented genes. Using clusterProfiler (v. 4.2.2) of gene ontology (GO, http://www.geneontology.org/) and the Kyoto encyclopedia (KEGG) genes and genomes pathways (http://www.kegg.jp) for gene set enrichment analysis. Draw the volcano graph using GraphPad Prism (v.8.0). Combined bubble-bar plots were generated in R using the ggplot2 package (v.3.5.1) for graphical rendering with multi-panel composition facilitated by the patchwork package (v.1.3.0).

Statistical analysis

All statistical analyses were performed using GraphPad Prism (version 8.0.2). Data are presented as mean ± standard deviation (Mean±SD). Comparisons between two groups were performed using unpaired Student’s t-test, while one-way ANOVA was used for multiple group comparisons. A p-value < 0.05 was considered statistically significant.

Results

Characterization of DR-EVs, GelMA microspheres, and thermosensitive hydrogel

DR-EVs were successfully isolated by differential ultracentrifugation and subsequently characterized in terms of morphology and physicochemical properties (Fig. 1) [28]. TEM images clearly showed the vesicles with a typical round or cup-shaped morphology, smooth surface, and intact bilayer membrane structure, consistent with the general features of extracellular vesicles (Fig. 2A). NTA results demonstrated a narrow and monodisperse size distribution with an average particle size of 67.8 nm, indicating good uniformity (Fig. 2B). In order to evaluate the membrane integrity, DR-EVs were treated with increasing concentrations of Triton X-100. The particle concentration gradually decreased in a concentration-dependent manner, with the remaining vesicles significantly reduced to 32.1% at 0.5% Triton X-100 compared to untreated control (Fig. 2C). Western blot analysis confirmed the presence of exosomal marker proteins HSP70 and TET8 while the negative marker Calnexin was not detected, suggesting a high purity of the isolated vesicles without significant cellular contamination (Fig. 2D). In addition, zeta potential analysis showed a surface potential with a major peak at around −40mV, implying that DR-EVs possessed good colloidal stability under physiological conditions (Fig. 2E).

Fig. 2.

Schematic illustration and characterization of DR-EVs, GelMA microspheres, and HPA hydrogel. (A) TEM image of DR-EVs showing the characteristic cup-shaped morphology. (B) NTA analysis displaying size distribution of DR-EVs. (C) Particle concentration changes after Triton X-100 treatment, confirming membrane integrity. (D) Western blot analysis showing positive exosomal markers (HSP70, TET8) and absence of the endoplasmic reticulum negative marker (Calnexin). (E) Zeta potential distribution of DR-EVs. (F) Schematic illustration of the thermoresponsive sol–gel transition of HPA hydrogel. (G) Optical photographs of HPA hydrogel precursor solution at 26 °C (injectable state) and 36 °C (gel state). (H) FTIR spectra of ungelled and gelled HPA hydrogel, confirming hydrogen bonding and physical crosslinking. (I) Rheological analysis of HPA hydrogel showing the gelation temperature. (J) SEM image of GelMA microspheres exhibiting surface porosity. (K) Cumulative in vitro release profiles of SIN and DR-EVs from the composite hydrogel. (L) In vitro degradation profiles of MicroGel, HPA, and HPA-MicroGel in PBS containing collagenase

The thermosensitive hydrogel system was constructed using HPMC, HA, and glycerol (Gl) as the basic matrix [29]. Different formulations exhibited variable gelation times and temperatures, and the optimal composition (7% HPMC, 2% HA, and 20% Gl) was identified from comparative studies (Table S1) [30]. Subsequently, our biocompatibility assays (Fig.S2 C, D) and in vivo organ histology (Fig.S2 B) confirmed that this formulation does not induce cytotoxicity or systemic toxicity. At 26 °C, the mixture remained as a low-viscosity liquid, facilitating injectability, while at 36 °C it rapidly transformed into a stable three-dimensional hydrogel network without apparent fluidity (Fig. 2F, G). Rheological measurements confirmed this transition, with the crossover of G′ and G″ occurring at ~34 °C, and G′ subsequently increasing sharply to reach a stable state of approximately 10 kPa, significantly higher than G″ (Fig. 2H), indicating the formation of a robust hydrogel network suitable for joint movement. FTIR spectra further confirmed that gelation was a physical crosslinking process rather than a chemical reaction. Both gelled (line a) and ungelled (line b) samples showed characteristic peaks of HPMC, HA, and glycerol without new absorption peaks, but obvious peak shifts were observed (Fig. 2I). The O–H stretching band shifted from 3342 cm⁻¹ to 3296 cm⁻¹, suggesting enhanced hydrogen bonding. The C = O band shifted from 1648 to 1611 cm⁻¹, while COO⁻ symmetric stretching moved from 1419 to 1376 cm⁻¹, indicating reorganization of hydrogen bonding and ionic interactions. Moreover, C–O–C and C–O vibrations shifted from 1209, 1109 cm⁻¹ to 1320, 1060 cm⁻¹, reflecting increased chain packing order [31]. These changes demonstrated that hydrogen bonding networks were strengthened and water molecule states were altered during gelation, which was consistent with the rheological observation of increased G′. The XRD pattern showed broad peaks around 15°and 20°without sharp reflections, indicating that the hydrogel is mainly amorphous, consistent with the non-crystalline nature of HA, HPMC, and glycerol (Fig. S1A).

GelMA microspheres were prepared using a microfluidic-based water-in-oil emulsion method combined with UV crosslinking [32]. The resulting microspheres exhibited uniform spherical morphology with an average diameter of 400 μm (Fig. S1B). SEM images further revealed a porous surface architecture, suggesting interconnected pores of 5–10 μm, providing channels for molecular loading and release (Fig. 2J).

To determine the degree of substitution (DS) of GelMA, 1 H-NMR spectroscopy was performed, revealing a methacrylation degree of approximately 41% (Fig.S1C). Furthermore, the in vitro release profiles (Fig. 2K) demonstrated a distinct sequential release pattern designed to match the RA therapeutic window. SIN was primarily released within the first week, allowing for rapid suppression of acute inflammation. In contrast, the DR-EVs encapsulated in the microspheres exhibited a delayed and sustained release behavior, which became more predominant after one week. This temporal lag ensures that osteogenic factors are delivered precisely when the inflammatory storm has subsided, facilitating effective bone repair.

The degradation profiles differed significantly among groups (Fig. 2L). HPA alone degraded rapidly, with residual mass of ~60–70% at day 7, decreasing to ~20–25% at day 14 and nearly complete loss by day 21. MicroGel exhibited slower degradation (~85–90% at day 7, ~ 45–50% at day 14, approaching complete degradation by day 21). In contrast, the composite HPA-MicroGel displayed an intermediate degradation profile. It degraded slower than the pure HPA hydrogel due to the presence of stable microspheres, but exhibited faster mass loss compared to pure MicroGels alone, attributed to the rapid dissolution of the outer HPA matrix. By day 21, the residual mass declined close to zero. Swelling studies confirmed the water uptake behavior of MicroGel and GM@DR-Evs, and there was no significant difference observed between the two groups (Fig. S1D).

Evaluation of the in vitro biocompatibility, Immunomodulatory activity, and anti-inflammatory effects of HPA@SIN-GM@DR

To evaluate the in vitro biocompatibility of HPA@SIN-GM@DR, we first examined its effects on the viability of BMSCs, CD4⁺ T cells, and macrophages. CCK-8 assay results showed that after 24 h and 48 h of HPA@SIN-GM@DR treatment, all groups of cells maintained high viability without significant cytotoxicity, indicating good in vitro cytocompatibility of HPA@SIN-GM@DR (Fig. 3A). Live/Dead staining further corroborated these findings. As shown in Fig.S2 C and D, the majority of BMSCs and macrophages in the HPA@SIN-GM@DR group exhibited green fluorescence (live), with negligible red fluorescence (dead) observed, which was comparable to the control group. This visual evidence reinforces the excellent cytocompatibility of the composite hydrogel system.

Fig. 3.

Evaluation of the in vitro immunomodulatory and anti-inflammatory effects of HPA@SIN-GM@DR. (A) Cell viability of macrophages, BMSCs, and CD4⁺ T cells at 24 h and 48 h, measured by CCK-8 assay. (B, C) Flow cytometry analysis of M1 (CD86⁺), M2 (CD206⁺) macrophages and Th17 (IL-17 A⁺), Treg (Foxp3⁺) T cells under different treatments. (D) ELISA detection of TNF-α, IL-1β, IL-6, and IL-10 secreted by treated macrophages. (E, F) Representative images of NF-κB and JAK signaling pathway protein levels in differently treated macrophages. All the data are presented as mean±SD from three independent experiments (n = 3). *p < 0. 05, **p < 0. 01, ^#p > 0. 05

Immune cell imbalance and excessive activation are key factors that drive and sustain the inflammatory response in RA [10, 32]. Therefore, we assessed the immunomodulatory effects of HPA@SIN-GM@DR.We established in vitro models of murine splenic CD4⁺ T cells and RAW264.7 macrophages, and applied different pretreatment regimens before stimulation with TGF-β + IL-6 or LPS, respectively [30, 31]. Flow cytometry results showed that stimulation of CD4⁺T cells with TGF-β + IL-6 increased the Th17/Treg ratio, whereas HPA@SIN-GM@DR intervention significantly reversed this process. Meanwhile, HPA@SIN-GM@DR intervention inhibited the M1 polarization of macrophages induced by LPS (Fig. 3B, C). In the RA synovial microenvironment, Th17 cells, Treg cells, and macrophages regulate inflammation by secreting cytokines with distinct functions; therefore, we further measured the levels of pro-inflammatory cytokines (TNF-α, IL-6, and IL-17) and the anti-inflammatory cytokine (IL-10) in the culture medium. ELISA results showed that HPA@SIN-GM@DR treatment significantly suppressed the expression of TNF-α, IL-6, and IL-17, and increased IL-10 secretion (Fig. 3D). Notably, in both in vitro cell models, the therapeutic efficacy of HPA@SIN-GM@DR was superior to that of SIN alone.

In addition, we specifically focused on the NF-κB and JAK signaling pathways, which are excessively activated in RA pathogenesis. Western blot results revealed that HPA@SIN-GM@DR intervention markedly inhibited the LPS-induced increases in the phosphorylation of p65, IκBα, JAK1, and STAT3 in macrophages (Fig. 3E, F). Similarly, the therapeutic effects of HPA@SIN-GM@DR were superior to those of SIN or D.R. alone. The above data suggest that HPA@SIN-GM@DR can regulate the Th17/Treg cell ratio while inhibiting the polarization of M1 macrophages, and exert anti-inflammatory effects by downregulating the NF-κB and JAK-STAT signaling pathways, showing good potential for alleviating RA inflammatory responses.

In vitro effects of HPA@SIN-GM@DR on osteoclast and osteoblast regulation

In the middle and late stages of RA progression, bone and cartilage destruction may occur, ultimately leading to joint deformity and functional impairment. In RA patients, osteoclast formation at the pannus–bone junction and subchondral bone is abnormally accelerated, while osteoblast differentiation is weakened, together disrupting the balance between bone formation and bone resorption [33]. Under physiological conditions, osteoclasts (OCs) originating from monocyte–macrophage lineage and osteoblasts (OBs) derived from mesenchymal stem cells jointly mediate the dynamic balance between bone resorption and bone formation in the human body [34]. Based on this, we first used BMSC-derived osteoblasts to examine the effects of HPA@SIN-GM@DR on bone formation.ALP and ARS staining results revealed that HPA@SIN-GM@DR treatment significantly increased ALP activity and mineralized nodule formation in osteoblasts (Fig. 4A, B). Furthermore, we explored the potential of HPA@SIN-GM@DR in promoting osteogenesis at the molecular level.PCR results demonstrated that HPA@SIN-GM@DR treatment significantly upregulated the expression of osteogenesis-related genes (RUNX2, BMP2, ERα and Smad signaling pathway) (Fig. 4C, Fig.S3 B). These findings suggest that HPA@SIN-GM@DR exerts a favorable effect on promoting osteoblast proliferation in vitro.

Fig. 4.

In vitro regulatory effects of HPA@SIN-GM@DR on osteoblast and osteoclast differentiation. (A) ALP staining of BMSCs after 7 days of treatment. Scale bar = 100 μm. (B) ARS staining of BMSCs after 14 days of treatment. Scale bar = 50 μm. (C) RT-qPCR detection of ALP, RUNX2, BMP2, and ERα mRNA expression levels in BMSCs. (D) TRAP staining of macrophages after 7 days of treatment. Scale bar = 100 μm. (E) RT-qPCR detection of Ctsk, ITGB3, and MMP9 mRNA levels in treated macrophages. All the data are presented as mean±SD from three independent experiments (n = 3). *p < 0. 05, **p < 0. 01, ^#p > 0. 05

Next, we established an in vitro osteoclast model using RAW264.7 cells to evaluate the effects of HPA@SIN-GM@DR on osteoclast differentiation. TRAP staining showed that HPA@SIN-GM@DR treatment markedly reduced the number of TRAP-positive multinucleated cells, suggesting an inhibitory effect on osteoclastogenesis (Fig. 4D). Meanwhile, PCR analysis revealed that HPA@SIN-GM@DR treatment significantly suppressed the expression of genes related to osteoclast differentiation and bone resorption functions (TRAP, Ctsk, ITGB3, MMP9 and NFATc1) (Fig. 4E, Fig.S3 C). In summary, these data demonstrate that HPA@SIN-GM@DR possesses favorable anti-bone destruction and bone formation–promoting properties in vitro.

In vivo therapeutic effects of HPA@SIN-GM@DR in an RA animal model

Next, we used a RA rat model, collagen-induced arthritis (CIA), which has similar pathogenesis and arthritis symptoms to human RA, to evaluate the therapeutic potential of HPA@SIN-GM@DR for RA in vivo [26]. The modeling and intervention process is shown in the figure (Fig. 5A). Prior to evaluating the therapeutic efficacy, we assessed the stability and fate of the implanted material in vivo. Using Cy5 fluorescence imaging to monitor the biodegradation process following subcutaneous injection, we observed that the fluorescence intensity of the HPA@SIN-GM@DR hydrogel gradually attenuated over time and became negligible by Day 21 (Fig.S2 A). This degradation timeframe of approximately 3 weeks perfectly aligns with the sequential release of the loaded drugs, ensuring sufficient material retention during the critical window of inflammation suppression and subsequent bone repair, while avoiding long-term foreign body accumulation. Following the booster immunization (day 14), rats in the CIA group exhibited progressive body weight loss and developed typical arthritic symptoms such as ankle swelling and restricted mobility, indicating successful induction of arthritis.Notably, HPA@SIN-GM@DR treatment markedly attenuated body weight loss and reduced the arthritis index in CIA rats, with greater efficacy than SIN or DR alone, suggesting that HPA@SIN-GM@DR could more effectively slow the progression of arthritis (Fig. 5B, C).

Fig. 5.

Evaluation of in vivo therapeutic effects of HPA@SIN-GM@DR in CIA rats (A) Schematic diagram showing CIA modeling and treatment protocol. (B, C) Body weight and arthritis scores were assessed every three days post-booster across groups (n = 3). (D, E, F) Representative H&E, TRAP, and SOFG stained sections of ankle joints (n = 3). Scale bar = 100 μm. Data are presented as mean±SD, *p < 0. 05, **p < 0. 01, ^#p > 0. 05

Furthermore, we conducted histopathological examinations of the joint tissues in each group. Hematoxylin–eosin (HE) staining revealed pronounced synovial hyperplasia, massive inflammatory cell infiltration, and marked destruction of the articular cartilage in the CIA group. In contrast, the HPA@SIN-GM@DR group exhibited markedly reduced synovial proliferation and inflammatory infiltration, with relatively well-preserved joint architecture (Fig. 5D, Fig.S3 A). In addition, Safranin O–Fast Green (SOFG) and Masson staining were used to assess the effects of HPA@SIN-GM@DR on cartilage and bone [35]. In the CIA group, cartilage destruction was evident, along with subchondral bone sclerosis and thickening, osteophyte formation, and increased bone marrow fibrosis.In contrast, the HPA@SIN-GM@DR group showed intact cartilage structure and well-preserved collagen fiber staining, suggesting a significant improvement in bone metabolism and prevention of bone destruction in CIA rats (Fig. 5E, F). The preservation of cartilage, as evidenced by the maintained proteoglycan content, is likely a consequential benefit of the mitigated synovial inflammation and stabilized subchondral bone, both of which are key drivers of cartilage degradation in RA [11, 36]. Collectively, these data demonstrate that HPA@SIN-GM@DR effectively alleviates joint inflammation and protects bone integrity in CIA rats, exhibiting synergistic anti-inflammatory and bone-protective effects, with superior therapeutic efficacy compared to SIN or DR alone. To address potential concerns regarding the immunogenicity and systemic toxicity of the gelatin-based carrier, we performed histological examinations of major organs (heart, liver, spleen, lung, and kidney) at the end of the treatment period. H&E staining results (Fig.S2 B) revealed no significant pathological abnormalities, inflammatory cell infiltration, or tissue damage in the HPA@SIN-GM@DR group compared to the control group, confirming the excellent in vivo biocompatibility and biosafety of the composite system.

HPA@SIN-GM@DR inhibits inflammatory responses and improves bone metabolism in vivo

To further elucidate the role of HPA@SIN-GM@DR in alleviating bone destruction in RA, we performed micro-computed tomography (Micro-CT) three-dimensional reconstruction and analysis of the ankle joints in each group of rats. The results revealed that rats in the CIA group exhibited marked bone structural damage, characterized by trabecular fracture and sparsity, subchondral bone erosion, and cortical bone defects, accompanied by a significant reduction in overall bone volume.In contrast, following HPA@SIN-GM@DR intervention, the integrity of joint bone structure was substantially improved, as evidenced by more regularly arranged trabeculae, reduced bone erosion, restoration of cortical bone continuity, and a marked increase in bone mineral density. Quantitative analysis demonstrated that critical bone parameters, such as bone volume fraction (BV/TV), were significantly higher in the HPA@SIN-GM@DR-treated group compared with the model group (Fig. 6A).

Fig. 6.

In vivo anti-inflammatory and bone-metabolism-improving effects of HPA@SIN-GM@DR. (A) Representative Micro-CT images of ankle joints and BV/TV quantification (n = 3). (B) Immunofluorescence staining of CD86 and RORγt in ankle joints and quantification of MFI (n = 3). Scale bar = 100 μm. (C) RT-qPCR detection of TNF-α, IL-1β, IL-6, and IL-10 mRNA expression levels in ankle joints (n = 3). Data are presented as mean±SD, *p < 0. 05, **p < 0. 01, ^#p > 0. 05

Moreover, immunofluorescence staining of ankle joint tissue sections revealed that the number of CD86⁺ M1-type macrophages in the synovium of the CIA group was significantly increased compared with the blank group, whereas HPA@SIN-GM@DR intervention markedly reduced the number of M1 macrophages. Similarly, the number of RORγt⁺ Th17 cells in the ankle joints of CIA rats was significantly elevated compared with the blank group, whereas HPA@SIN-GM@DR treatment significantly reduced Th17 cell numbers (Fig. 6B). These findings suggest that HPA@SIN-GM@DR can improve the synovial inflammatory microenvironment in CIA rats by inhibiting excessive activation of M1 macrophages and Th17 cells.

To verify the immunomodulatory effect of HPA@SIN-GM@DR in vivo at the molecular level, we detected the transcription levels of inflammatory factors in the ankle joint tissues of rats in each group by qPCR.The results showed that mRNA expression levels of pro-inflammatory cytokines (TNF-α, IL-6, and IL-1β) were significantly upregulated in the CIA model group, whereas the anti-inflammatory cytokine IL-10 was markedly downregulated. Following HPA@SIN-GM@DR intervention, mRNA expression of pro-inflammatory cytokines (TNF-α, IL-6, IL-1β) was significantly suppressed, while IL-10 mRNA expression was markedly increased (Fig. 6C). The above results indicate that HPA@SIN-GM@DR improves arthritis in CIA rats through its excellent immunomodulatory and anti-inflammatory effects.

Network Pharmacology and bioinformatics analysis of the therapeutic mechanisms of HPA@SIN-GM@DR in RA

Finally, bioinformatics approaches were employed to further elucidate the potential mechanisms of HPA@SIN-GM@DR in the treatment of RA. We first screened the active compounds of Rhizoma Drynariae and sinomenine, along with their corresponding targets, from the TCMSP database; after merging and removing duplicates, a total of 165 drug-related target genes were obtained. Subsequently, RA-related targets were retrieved from the GeneCards and OMIM databases, yielding a total of 1,031 disease-associated genes.By employing the Venny 2.1 analysis platform, 67 overlapping targets between drug-related and disease-related genes were identified (Fig. 7A). These 67 intersecting targets were then imported into the STRING database to construct a protein–protein interaction (PPI) network, and visualized using Cytoscape software.The analysis results showed that multiple key inflammatory and immune regulatory factors were located at the core of the network, such as IL-6, TNF, MMP9, CASP3, AKT1, ESR1, BCL2, TP53, PPARG, PTGS2, etc., suggesting that HPA@SIN-GM@DR intervention can act on multiple links of RA inflammatory response and bone metabolism (Fig. 7B, C). Furthermore, GO and KEGG pathway enrichment analyses of the intersecting targets were performed using the DAVID database.GO functional enrichment analysis indicated that these targets were mainly involved in biological processes such as inflammatory response, apoptosis, immune regulation, and the modulation of bone resorption and bone formation (Fig. 7D). KEGG pathway enrichment analysis revealed that the target genes were significantly enriched in classical RA-associated inflammatory pathways, including TNF, NF-κB, IL-17, and JAK–STAT signaling, as well as in pathways closely related to the regulation of bone destruction and bone formation, such as AGE–RAGE, HIF-1, and Relaxin signaling pathways (Fig. 7E).

Fig. 7.

Network pharmacology and transcriptomic analysis of HPA@SIN-GM@DR mechanisms in RA treatment. (A-E) Network pharmacology analysis of Qingteng alkaloids and bone fragment supplements in the treatment of RA. (A) Venny plot of drug (SIN, DR) targets and RA targets. (B, C) Construction and visualization of PPI network with common targets. (D, E) GO and KEGG enrichment analysis of common targets. (F-H) Detecting CIA group and through RNA seq HPA@SIN-GM @Changes in gene expression of ankle joint tissue in DR group rats. (F) Volcanic diagram of differentially expressed genes (DEGs) between groups (HPA@SIN-GM@DR group vs. CIA group, n = 3). (G, H) Perform GO and KEGG enrichment analysis on DEGs

We further investigated the molecular targets of HPA@SIN-GM@DR intervention in CIA rats through transcriptome sequencing.RNA-seq analysis identified a total of 680 differentially expressed genes (DEGs) between the HPA@SIN-GM@DR intervention group and the PBS control group, of which 161 genes were significantly upregulated and 519 genes were significantly downregulated (Fig. 7F). GO enrichment analysis of the DEGs demonstrated that the target genes were predominantly associated with immune regulation, inflammatory responses, and the regulation of bone homeostasis (Fig. 7G). KEGG pathway enrichment analysis revealed that the DEGs under HPA@SIN-GM@DR intervention were significantly enriched in multiple RA-related signaling pathways, including MAPK, JAK–STAT, IL-17, and NF-κB pathways, which are classical cross-talk pathways between inflammation and bone metabolism. These findings suggest that HPA@SIN-GM@DR may modulate RA pathogenesis by coordinately regulating the dynamic balance between osteogenesis and osteoclastogenesis, thereby alleviating bone destruction induced by RA (Fig. 7H).Collectively, these results indicate that HPA@SIN-GM@DR exerts therapeutic effects on RA via a multi-pathway, multi-target mechanism that modulates immune-inflammatory responses and bone metabolism, thereby reducing synovial inflammation and joint structural damage.

Conclusion

In this study, we developed a multifunctional, injectable, thermosensitive composite hydrogel system (HPA@SIN-GM@DR), structurally designed with a mixed matrix of hyaluronic acid (HA) and hydroxypropyl methylcellulose (HPMC) (HPA) encapsulating GelMA microspheres, thereby enabling the co-delivery of sinomenine (SIN) and Drynaria rhizome–derived extracellular vesicles (DR-EVs). Distinct from conventional hydrogel delivery systems that often rely on the simple physical blending of a single therapeutic agent, our work represents a significant advancement in RA treatment strategies. While previous studies have focused primarily on either anti-inflammatory or osteogenic approaches in isolation, our system innovatively integrates a “microsphere-in-hydrogel” architecture to achieve a programmable “dual-source” therapy. This design not only overcomes the limitations of burst release common in traditional hydrogels but also establishes a synergistic hierarchy: Sinomenine provides rapid upstream immunomodulation to quench the cytokine storm, while the plant-derived DR-EVs offer a novel, cell-free approach to directly promote downstream bone regeneration. This precise “temporal and spatial” coordination constitutes the core scientific superiority of our approach over existing single-target therapies. The GelMA microspheres provide an excellent platform for drug loading and cell adhesion, while the HPA matrix exhibits superior thermoresponsive gelation and synovium-mimicking properties, ensuring rapid in situ gelation at body temperature, enhancing local retention at the lesion site, and prolonging therapeutic efficacy. In vitro, HPA@SIN-GM@DR effectively inhibited M1 macrophage polarization, restored the Th17/Treg immune balance, and downregulated the activity of key inflammatory signaling pathways such as NF-κB and JAK/STAT. Meanwhile, the material significantly enhanced the osteogenic differentiation potential of bone marrow mesenchymal stem cells (BMSCs), inhibited osteoclast differentiation and resorptive activity, thereby enabling remodeling regulation of bone metabolism. Collectively, the in vitro evidence delineates a coherent mechanistic link: the immunomodulatory actions (upstream) effectively suppress the inflammatory drivers of bone erosion, while the direct effects on osteoblasts and osteoclasts (downstream) actively rectify the imbalance in bone remodeling. This “dual-tracks” strategy ensures a comprehensive intervention in the RA pathological cascade. The GelMA microsphere structure offers a three-dimensional scaffold and sustained drug release function, while the thermosensitive HPA encapsulation ensures efficient intra-articular localization following injection. In vivo, HPA@SIN-GM@DR markedly improved clinical signs and joint histopathology in CIA rat models, increased bone mineral density, and preserved cartilage integrity. Transcriptomic and network pharmacology analyses further revealed that it could exert synergistic regulation through multiple inflammatory and osteogenic signaling pathways-including IL-17, NF-κB, JAK-STAT, MAPK, and AGE-RAGE-thereby achieving comprehensive intervention in the pathological progression of RA.

In conclusion, this study proposes an innovative strategy integrating intelligent material design, bioactive extracellular vesicle delivery, and bidirectional therapeutic functions.HPA@SIN-GM@DRs exhibit favorable injectability, biocompatibility, targeting capability, and therapeutic efficacy, offering a safe, efficient, and translationally promising treatment platform for RA and other inflammatory bone-destructive diseases.

Author contributions

Li Cai, Juan Li^7*, Shixian Chen, Ji Li and Yang Chen designed the experiments; Li Cai, Jian Gao, Kai Zhang, Bing Xiao and Sijia Xu performed experiments. Wei Zhao, Juan Li¹, Yanli Zhou, Wenying Zhu, Shuyuan Liu, Tingting Pei and Junhua Li analyzed data. Li Cai, Jian Gao, Kai Zhang and Bing Xiao wrote the paper. All authors reviewed and edited the manuscript.

Funding

This work was supported in part by funding from the National Natural Science Foundation of China (82304928) to Li Cai and (82374206) to Juan Li^7*, Guangzhou Traditional Chinese Medicine Major Science and Technology Project (2025QN003) to Shixian Chen. Guangdong Basic and Applied Basic Research Foundation (2024A1515011263) to Junhua Li and (2025A1515012141) to Tingting Pei. 1 + N Special Support Plan of Guangzhou Medical University Affiliated Hospital of Traditional Chinese Medicine (2024-TZ-16) and Joint Funding from Guangzhou Science and Technology Bureau, School (Hospital) and Enterprise in 2025 (2025A03J3281) to Ji Li.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.