Human knee joint sample were collected from patients undergoing primary TJA and revision TJA due to aseptic loosening at the General Hospital of Ningxia Medical University. The collection and use of human sample were conducted in strict compliance with ethical guidelines and were approved by the Institutional Review Board of the General Hospital of Ningxia Medical University (Ethical Approval No. 2020-913).

Primary TJA sample were collected from patients undergoing primary TJA for end-stage osteoarthritis or other degenerative joint diseases, served as control. Revision TJA sample were collected from patients undergoing revision TJA due to aseptic loosening caused by wear particle-induced osteolysis. Immediately after collection, the clinical sample were rinsed with sterile phosphate-buffered saline (PBS) to remove blood and debris. The tissue was snap-frozen in liquid nitrogen and stored at −80°C for RNA isolation.

Titanium alloy particles (Ti–6Al–4V, the medium particle size 5 μm, range 4–6.35 µm) used in this project were generated by the Zimmer Corporation (Indiana, United States). Sterilization was achieved by heating the particles at 180°C for 45 min. They were then cleaned with three ethanol rinses to remove contaminants, followed by three phosphate-buffered saline (PBS) rinses to neutralize the surface. Endotoxin levels were verified to be below 0.1 EU/mL using the E-Toxate kit from Sigma. For in vivo applications, a suspension was made by dispersing 10 mg of particles in 1 mL of normal saline (Yu et al., 2013).

siRNA oligos (5′-GGT​TGC​CTC​TGT​CTC​AGA​ATT-3′) targeting TNF-α were synthesized and annealed by Sangon Biotech (Shanghai, China). Then, siRNA oligos were cloned into the pAdEasy-U6-CMV-EGFP vector. The full-length Wnt3a cDNA from the pCDNA3.1 vector was cloned into the pAdEasy-EF1-MCS-3flag-CMV-EGFP vector using KpnI and XhoI restriction sites. Positive clones were confirmed by PCR and sequencing. Recombinant adenoviral vectors were created by recombination between pAdEasy-EF1-Wnt3a-3flag-CMV-EGFP and pAdeasy-1 in E. coli BJ5183. For dual expression of Wnt3a and TNF-α-siRNA, pHBHD-U6-TNF-α-siRNA-CMV-Wnt3a was linearized with PacI and co-transformed with pAdeasy-1 into E. coli BJ5183. Adeno-associated virus packaging and purification were conducted by Hanheng Biological (Shanghai, China).

60 SPF-grade BALB/c mice (10-week-old, 28–30 g) were purchased from Chongqing Enswell Biotechnology (Chongqing, China) and maintained under standard conditions for an additional 2 weeks. The mice were divided into the following five groups, with 12 mice in each group: negative control group (PBS injection + titanium nails), positive control group (titanium particle suspension + titanium nails), TNF-α-siRNA treatment group (titanium particle suspension + titanium nails + TNF-α-siRNA), Wnt3a overexpression group (titanium particle suspension + titanium nails + Wnt3a overexpression) and combination therapy group (titanium particle suspension + titanium nails + TNF-α-siRNA-Wnt3a overexpression).

Prior to surgery, mice were fasted but provided water ad libitum. Anesthesia was induced with an intraperitoneal injection of 10% chloral hydrate at 30 mg/kg (Liu et al., 2019). The surgical site’s fur was removed, and the skin was disinfected with povidone-iodine solution. A 5 mm medial parapatellar incision was made to expose the patellar ligament and tibial plateau. A bone tunnel was drilled into the tibial plateau using a 0.8 mm dental burr, extending 5 mm into the medullary cavity from the center of the plateau.

Negative control group mice received a 10 µL PBS injection into the tunnel, while Positive control group and the gene therapy group was treated with a 10 µL suspension of titanium particles at 10 mg/mL. Artificial prosthetic titanium nails were then implanted in all mice, ensuring the nail cap was flush with the tibial plateau’s articular surface. Post-implantation, joint mobility was assessed for range of motion and stability, and the incision was closed in layers. Postoperative X-rays confirmed the prosthesis’s positioning within the bone marrow cavity.

All mice were screened for common pathogens, including mouse hepatitis virus, minute virus of mice, mouse parvovirus, and others, to ensure specific pathogen-free conditions. The animal study was approved by the Animal Research-Animal Care Committee of the General Hospital of Ningxia Medical University (Ethical Approval No. 2020-913, same as the human study).

Tibiae with implanted titanium nails underwent high-resolution micro-computed tomography (micro-CT) using the SuperNova®CT SNC-100 system (PINGSEGN). The scanning parameters were set at 15 μm per layer with a voltage of 45 kV and a current of 435 μA. A region of interest (ROI), 1.7 mm in diameter and adjacent to the tibial growth plate, was selected for analysis. Three-dimensional (3D) reconstruction of the images was performed to assess morphometric parameters, including bone mineral density (BMD) and bone volume fraction (BV/TV).

Starting from the second week postoperatively, mice in the study received a series of biweekly intra-articular injections of 40 µL into the operated limbs. Both the negative and positive control groups were injected with PBS. The treatment group was administered TNF-α-siRNA, OE-Wnt3a, and TNF-α-siRNA-OE-Wnt3a adeno-associated virus at a concentration of 2 × 10^10 pfu/mL. This injection regimen was continued for 16 weeks post-surgery, with all mice maintained on a standard diet throughout the study period.

Tissue samples were collected from areas surrounding the implants at 4, 8, 12, and 16 weeks post-implantation for TNF-α, Wnt3a and urine samples were obtained at 16 weeks for urinary deoxypyridinoline (DPD). TNF-α level was measured using ELISA kit from Ruixinbio (Quanzhou, China). Wnt3a and DPD level was measured using ELISA kit from COIBO BIO (Shanghai, China). All procedures were performed following the manufacturers' instructions.

Total RNA was extracted from peri-prosthetic tissue, which included both bone tissue and associated soft tissues surrounding the prosthesis. The tissues were homogenized in TRIzol reagent (Invitrogen, United States) using a grinding machine (Cebo, CEBO-24) with steel beads. RNA was isolated following the protocol, which involved phase separation with chloroform, RNA precipitation with isopropanol, and washing with 75% ethanol. The RNA was dissolved in RNase-free water, and its concentration and purity were measured using a NanoDrop spectrophotometer (Thermo Fisher Scientific, United States), with A260/A280 ratios between 1.8 and 2.0 indicating high-quality RNA.

Subsequently, 1 μg of RNA was reverse transcribed into complementary DNA (cDNA) using the HiScript II Reverse Transcriptase Kit (Noweizan, China). q-PCR was then performed on the ABI 7900 real-time PCR system to analyze the cycle threshold values. The specific sequences of the PCR primers used in this study are detailed in Supplementary Table S1.

Peri-prosthetic tissues, including both bone and associated soft tissues, were homogenized using a grinding machine (Cebo, CEBO-24) with steel beads in RIPA buffer supplemented with protease inhibitors. The homogenization involved grinding, followed by vortexing for 1 min and incubation on ice for 10 min. The homogenate was centrifuged at 13,000 rpm at 4°C for 20 min, and the supernatant was collected. Proteins (500 µg) were mixed with 5× SDS loading buffer in a 4:1 ratio and denatured at 100°C for 6 min. Samples (30 µg per lane) were resolved by 12% SDS-PAGE and transferred onto the PVDF membrane, which was blocked with 5% non-fat milk for 1 h at room temperature. The membrane was incubated overnight at 4°C with primary antibodies against RANKL (abclon, A2550), osteocalcin (OCN) (abclon, A20800), osteoprotegerin (OPG) (abclon, A2100), tartrate-resistant acid phosphatase (TRAP) (abclon, A0962), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (abclon, A19056) at a 1:1000 dilution. After washing, the membrane was incubated with Goat anti-Rabbit IgG secondary antibody (Abclonal, AS014) for 1 h at room temperature. Protein bands were visualized using an ECL detection reagent (Yeasen, China) for 1 min and imaged using Fusion FX6 Spectra imaging system.

Periprosthetic tissues (including both bone and associated soft tissues) and major organs (liver, spleen, kidney, and brain) were collected and fixed in 4% paraformaldehyde. For bone tissues, decalcification was performed using 10% formic acid for 7–10 days, with the solution changed daily, until the bones were sufficiently softened. After decalcification, the bone tissues were rinsed thoroughly with distilled water to remove residual acid. Subsequent processing for both bone and soft tissues included dehydration through a graded ethanol series (70%, 80%, 90%, and 100%), clearing in xylene, and paraffin embedding. The samples were then sectioned, deparaffinized, and rehydrated before being stained with hematoxylin and eosin (H&E) using a commercial kit from Servicebio (Wuhan, China). Following staining, the slides were dehydrated, cleared, and mounted with neutral resin. Histological evaluation was conducted using an Mshot MF53 microscope.

Tibial bone specimens were collected from mice at 16 weeks post-surgery and processed for Goldner staining using a kit from Servicebio (Wuhan, China) following the manufacturer’s instructions. The specimens were fixed in 4% paraformaldehyde for 48 h and then subjected to a series of histological procedures including dehydration, clearing, wax impregnation, and paraffin embedding to prepare paraffin sections. These sections were deparaffinized and rehydrated before staining.

The staining protocol involved incubating the slides with an equal volume mixture of Goldner A/B solutions for 20 min, followed by differentiation in 1% acid alcohol. Subsequently, the slides were immersed in Goldner C solution for 10 min, rinsed briefly in 0.2% acetic acid, treated with Goldner D solution for 3 min, and finally stained with Goldner E solution for 5 min. After staining, the sections were dehydrated through a graded ethanol series and xylene, mounted with neutral resin, and visualized using a 3DHISTECH Pannoramic MIDI slide scanner and corresponding software.

Tibial bone sections were prepared for Von Kossa staining, paralleling the process for Goldner staining. The Von Kossa staining solution from Servicebio (Wuhan, China) was applied, and the Rehydrated sections were covered and exposed to ultraviolet (UV) light for 15 min. The slides were then washed extensively with distilled water. Following this, the sections were stained with hematoxylin for 3 min, differentiated, and then rinsed with running water to develop the blue color. A series of dehydration steps in graded ethanol was performed, after which eosin staining was applied. The sections were then dehydrated through anhydrous ethanol and xylene before being mounted with neutral resin. The stained sections were visualized using a 3DHISTECH Pannoramic MIDI slide scanner.

After the humane sacrifice of the animal, the mouse limb retaining the intact implant was extracted by disarticulating the knee joint. Subsequently, all surrounding soft tissues were meticulously dissected to reveal the surface of the implanted pin and the proximal tibia. The exposed tissue was securely mounted onto a custom fixture. Multiple nylon threads were then wound around the titanium nail cap, with their opposite ends encircling a screw, which was anchored to the fixture of a tensile testing machine. The setup was meticulously aligned to ensure that the pin’s longitudinal axis was parallel to the direction of extraction and perpendicular to the tension applied by the nylon threads. A controlled tension test was then executed at a displacement rate of 1.0 mm/min, continuing until the titanium nail was completely extracted. Throughout this process, the load in newtons (N) and displacement in millimeters (mm) were meticulously documented for subsequent analysis.

Statistical analysis and graph plotting were performed using SPSS 20.0 and Prism 8.0 software. Each experiment included three biological replicates, and data are presented as mean ± standard deviation. One-way ANOVA followed by Sidak’s post hoc test was used for multiple group comparisons with homogeneity of variances, and Dunnett’s T3 test was used when variances were not homogeneous. Statistical significance was accepted at p < 0.05 for all analyses.

To elucidate the roles of TNF-α and Wnt3a in aseptic loosening of prostheses, we collected tissue samples from patients undergoing primary total joint arthroplasty (TJA) and those experiencing aseptic loosening requiring revision TJA (Figure 1A). q-PCR analysis revealed significantly elevated TNF-α expression in revision TJA tissues compared to primary TJA tissues. In contrast, WNT3A gene expression was lower in revision TJA tissues (Figure 1B; Supplementary Figure S1A). These findings suggest a potential involvement of TNF-α and WNT3A in wear-related processes associated with aseptic loosening.

We then developed a murine model of aseptic loosening of knee prostheses and conducted gene therapy to explore the roles of TNF-α and Wnt3a in this condition (Figure 1C; Supplementary Figure S1B). Postoperative X-ray and μCT examination confirmed the accurate implantation of the titanium nail into the medullary cavity (Figures 1D, E). q-PCR analysis showed that TNF-α expression in the positive control group was significantly higher than in the negative control group, while Wnt3a expression was markedly lower (Figure 1F). These results were corroborated by ELISA results (Figure 1G). Briefly, these results indicate that our model mouse reflects the expression of TNF-α and Wnt3a genes observed in clinical samples.

To investigate the effect of TNF-α on osteoclast differentiation, we injected AAV carrying TNF-α-siRNA into the knee joint cavity and detected the expression of TNF-α in tissue surrounding the prosthesis. ELISA data revealed a significant suppression of TNF-α gene expression in the TNF-α-siRNA-treated group compared to the positive control group at different time points (Figure 2A). Additionally, to confirm the specificity of TNF-α knockdown and rule out potential off-target effects, we assessed the mRNA levels of the top predicted off-target genes in the RAW264.7 cell model stimulated with titanium particles. As shown in Supplementary Figure S1C, no significant changes in the expression levels of these off-target genes were observed following TNF-α knockdown, confirming the specificity of the siRNA-mediated TNF-α suppression. Histological examination through H&E staining of peri-prosthetic tissues showed that at 12 weeks, the negative control group displayed minimal inflammatory cell infiltration, no distinct boundary membranes, and well-preserved bone structures (Figure 2B). In contrast, the positive control group exhibited extensive inflammatory cell infiltration, substantial membrane thickening, and significant bone destruction. Notably, the TNF-α-siRNA treated group demonstrated minimal inflammatory cell infiltration, thin membrane formation, and no significant bone destruction (Figure 2B). Over time, the positive control group showed an escalation in inflammatory cell infiltration, while the TNF-α-siRNA treatment group showed a relative decrease in inflammatory cells and a reduction in boundary membrane thickness.

Given that RANKL and TRAP are pivotal in osteoclastogenesis, we next evaluated the effects of TNF-α knockdown on their expression. Western blot analysis indicated that TNF-α knockdown resulted in reduced levels of both RANKL and TRAP (Figures 2C,D; Supplementary Figure S2). Since TNF-α is known to regulate the production of other inflammatory cytokines, including IL-6, we also measured IL-6 expression post-therapy. Consistent with our hypothesis, IL-6 expression was downregulated following TNF-α knockdown (Supplementary Figure S3). TRAP staining, used to visualize differentiated osteoclasts, showed a significantly higher number of TRAP-positive cells in the positive control group, while the TNF-α knockdown and negative control groups displayed fewer TRAP-positive cells (Figures 2E, F). In line with these findings, genes associated with osteoclast formation and function, such as nuclear factor of activated T cells, cytoplasmic, calcineurin dependent 1 (Nfatc1), Trap, osteoclast associated receptor (Oscar), matrix metallopeptidase 9 (Mmp9), and cathepsin K (Ctsk), were also downregulated following TNF-α knockdown (Supplementary Figure S3). Collectively, these results suggest that AAV-mediated TNF-α knockdown effectively reduces osteoclast formation in prosthetic model mice.

To elucidate the role of Wnt3a in osteoblast differentiation, we employed an AAV vector to overexpress Wnt3a in mouse models. We subsequently measured Wnt3a gene expression in peri-prosthetic tissues at various time points following gene therapy. ELISA results demonstrated a significant upregulation of Wnt3a in the treatment group compared to the positive control, validating the efficacy of our AAV-mediated approach in enhancing Wnt3a expression levels (Figure 3A). Subsequent H&E staining around the prosthesis revealed minimal inflammatory cell infiltration and thin fibrous membrane formation in the Wnt3a treatment group at 12 weeks, with no significant bone destruction observed. By 16 weeks, a relative decrease in inflammatory cells and reduced membrane thickness was noted, indicative of a positive therapeutic response (Figure 2B).

Given the critical roles of OPG and OCN in osteoblast differentiation and bone development, we assessed their expression levels following Wnt3a overexpression. Western blot analysis confirmed an increase in both OPG and OCN levels (Figures 3B,C; Supplementary Figure S2). Consistent with these findings, a panel of osteogenic genes, including Ocn, Opg, runt related transcription factor 2 (Runx2), alkaline phosphatase, liver/bone/kidney (Alpl), and transcription factor 4 (Tcf4), were upregulated post-Wnt3a overexpression (Figure 3D). In brief, these results suggest that AAV-mediated Wnt3a overexpression promotes osteoblast formation in prosthetic model mice.

The above findings demonstrated the therapeutic effects of TNF-α-siRNA and Wnt3a gene therapies individually. Building on these results, we investigated the synergistic potential of combining these two therapeutic strategies for the treatment of aseptic loosening. We constructed an AAV vector carrying both the TNF-α interference and Wnt3a overexpression sequences (Figure 4A). ELISA assays confirmed significant downregulation of TNF-α in the combination gene therapy group, comparable to the TNF-α-siRNA monotherapy group and markedly lower than the positive control group (Figure 2A). Interestingly, at 16 weeks, the combination gene therapy group showed even lower TNF-α expression than the TNF-α-siRNA monotherapy group (Figure 2A). Concurrently, the combination gene therapy group exhibited Wnt3a expression levels similar to the Wnt3a monotherapy group, both significantly higher than the positive control group, with elevated Wnt3a expression observed as early as 12 weeks in the combination group (Figure 3A). These results indicate the successful development of an AAV vector capable of knocking down TNF-α and overexpressing Wnt3a.

The combination gene therapy group demonstrated thinner boundary membranes, reduced inflammatory cell infiltration, and enhanced bone formation at 12 weeks compared to the single gene therapy groups. Over the course of treatment, the combination gene therapy group showed superior therapeutic effects, with significant reductions in RANKL and TRAP expression and increases in OCN and OPG protein levels compared to the control group (Figures 2C, D). Notably, differential expression levels of RANKL, OCN, and OPG genes were observed between the combination and single gene therapy groups (Figures 2C, D). Consistent with these findings, the combination gene therapy group had a lower number of TRAP-positive cells (Figures 2E, F).

Micro-CT scanning and three-dimensional reconstruction of the mouse knee prosthesis revealed pronounced osteolysis in the positive control group, while the gene therapy groups showed significantly reduced osteolysis compared to the positive control. The combination gene therapy group had significantly higher bone volume fraction and bone density than the single gene therapy group (Figures 4B, C). Biomechanical testing further supported our primary findings, with the ultimate shear strength value in the positive control group being lower than the negative control group, indicating prosthesis loosening post-implantation (Figure 4D). Gene therapy partially restored this value, with the combination gene therapy group showing higher ultimate shear strength, apparent shear stiffness, and total energy absorption values compared to the single gene therapy group, further highlighting the efficacy of the combination gene therapy.

The therapeutic efficacy of combination gene therapy targeting TNF-α knockdown and Wnt3a overexpression was further evaluated at the 16-week mark, with a focus on its ability to alleviate bone resorption. Von Kossa staining was utilized to visualize calcium deposits, revealing more pronounced dark black staining in the negative control group, indicating robust mineralization. In contrast, the positive control group showed less black-brown staining, suggesting reduced mineral deposition and increased bone resorption due to the influence of titanium particles on bone metabolism (Figure 5A). Post-gene therapy, a significant enhancement in mineralization was observed, with quantitative analysis showing a substantial difference in the mean area of calcium salt deposition islands between the single gene therapy group and the positive control group (Figure 5B). Notably, the combination gene therapy group exhibited a significantly greater mean area of calcium salt deposition islands compared to the single gene therapy group, and a similar trend was observed in the quantification of the area fraction of calcium salt sedimentary islands.

To gain a comprehensive understanding of the gene therapy’s impact, Goldner trichrome staining was performed (Figure 5C). The negative control group displayed a greater proportion of green-stained areas, indicating less mineralization, while the positive control group showed a higher proportion of orange-red stained areas, suggesting severe non-mineralization. Image analysis using IPP software was conducted to quantify the bone-implant contact rate and bone volume fraction based on the staining results. The average bone-prosthesis contact rate in the positive control group was below 50%, whereas the gene therapy group showed a contact rate exceeding 80% (Figure 5D). The combination gene therapy group demonstrated an even higher bone-prosthesis contact rate compared to the single gene therapy group. Additionally, the bone volume percentage surrounding the prosthesis in the gene therapy group significantly increased compared to the control group, with the combination gene therapy group showing an even higher value.

Furthermore, we assessed urinary DPD levels, a well-recognized biomarker of collagen degradation released during osteoclast-mediated bone resorption, to complement our histological staining. The urine DPD concentrations in the positive control group were significantly elevated compared to the negative control group. Following gene therapy, a reduction in DPD levels was observed, with the combination gene therapy group showing a particularly pronounced decrease (Figure 5E). These findings, along with the DPD data, collectively demonstrate that the combination gene therapy involving Wnt3a overexpression and TNF-α knockdown can effectively mitigate bone resorption, offering a promising approach for the treatment of periprosthetic osteolysis.

To ensure a thorough assessment of the safety profile of our gene therapy approach, we conducted a series of evaluations monitoring mouse survival and potential systemic effects. Initially, we tracked the survival status of the mice at various time points post-treatment. Figure 6A illustrates that while a few early deaths occurred post-modeling, the implementation of gene therapy did not lead to any additional increase in mortality, suggesting that our gene therapy strategy did not adversely impact the overall survival rate of the mice. Furthermore, to determine if gene therapy might induce a systemic inflammatory response, we quantified the levels of the inflammatory cytokine TNF-α in peripheral blood (Figure 6B). The data revealed no significant differences in TNF-α levels across the experimental groups, with all maintaining low TNF-α levels, thereby confirming that gene therapy did not trigger significant systemic inflammatory responses.

After a 16-week treatment period, mice were euthanized, and organs including the heart, brain, spleen, and liver were harvested and weighed. The organ weights showed no significant differences between the treatment and control groups, with all organs appearing normal in size (Figure 6C). Subsequent histological examination using H&E staining was performed on these major organs to assess tissue morphology. The histological analysis demonstrated preserved normal tissue architecture across all examined organs, with prominent nucleoli and no observable pathological abnormalities (Figure 6D). Collectively, these findings suggest that the intra-articular administration of the adenoviral vector into the knee joint cavity does not induce systemic toxicity in other organs, supporting the safety of our gene therapy approach.