Peptides were synthesized based on the human SOST (GenBank: AAK16158.1) fragment encompassing amino acids 24 to 211, excluding the signal peptide, yielding approximately 30-amino-acid segments. An indirect enzyme-linked immunosorbent assay (ELISA) was performed to identify a high-affinity SOST peptide fragment for ROMO. The polypeptides were synthesized by Sangon Biotech (Shanghai, China).

96-well plates (Abcam, Cambridge, MA, USA) were coated with 100 µL of 1 µg/mL human SOST-synthesized polypeptides in the provided coating buffer overnight at 4 °C. Following coating, the plates were washed with 1× washing buffer and subsequently blocked using 1× blocking buffer. Diluted ROMO (AntibodySystem, Schiltigheim, France) was then added to the wells and incubated for 2 hour at 37 °C. After additional washing, 100 µL of 1:10,000 diluted horseradish peroxidase (HRP)-labeled goat anti-human antibodies (Bioss, Beijing, China) was introduced to the wells, followed by another 2-hour incubation at 37 °C. Finally, 100 µL of TMB was dispensed into each well, and after a 20-minute incubation at 37 °C, the absorbance was measured at 450 nm.

The high-affinity SOST peptide fragment for ROMO was then identified using ELISA. The binding affinity of the high-affinity SOST fragment to the heavy and light chains of ROMO was assessed using the HawkDock server (13), while interactions between the SOST fragment and ROMO were analyzed with PDBsum (14).

Prediction of major histocompatibility complex (MHC) class I-restricted cytotoxic T lymphocyte (CTL) epitopes for the high-affinity SOST fragment was performed using NetMHCpan 4.1 EL tool (15). A comprehensive analysis of 9-mer epitopes was conducted across the A1, A2, A3, A24, and B7 supertypes. Epitopes with a percentage rank (< 0.5%) were categorized as strong binders (SB), whereas those with a percentage rank (< 2%) were classified as weak binders (WB). In parallel, the identification of helper T lymphocyte (HTL) epitopes, comprising 15-mer peptides that bind to MHC class II, was accomplished using the NetMHCIIpan 4.1 EL server (16), with strong binders defined as having a percentage rank (< 1%) and weak binders defined as having a percentage rank (< 5%).

To predict linear B cell epitopes within the high-affinity SOST fragment, we employed BepiPred2.0 tool (17), applying a default filtering threshold of 0.5. Additionally, conformational B cell epitopes for both the screening peptide and the designed vaccines were predicted using the ElliPro application (18), with a minimum score threshold set at 0.5 and a maximum distance of 6 Angstroms.

The high-affinity SOST fragment was selected as the target antigen for the development of a recombinant subunit vaccine. Based on our previous experimental experience (10), the DTT fragment (amino acids 203–378; WP_371890660.1) was chosen as immune scaffold to facilitate conjugation with SOST fragment, thereby enhancing immune recognition and promoting antibody production. The DTT scaffold was conjugated with varying copy numbers of SOST peptide, ranging from 0 to 5, to generate a series of chimeric molecules. These conjugations were connected via a (GGGGS)₂ flexible linker to ensure optimal conformational flexibility. The resulting vaccine candidates were designated as DDT-SOST_(131-163)0 (DS₀) to DDT-SOST_(131-163)5 (DS₅), with the numerical suffix denoting specific number of SOST peptide fragments fused to DTT scaffold.

Tertiary structure of candidate vaccines was predicted using AlphaFold2 server (19), based on their amino acid sequences. Five structural models were generated, demonstrating close alignment with experimental accuracy. Top-ranked model was selected for further analysis. The quality of vaccine structure was assessed using the predicted Local Distance Difference Test (pLDDT).

To evaluate the immunogenic potential of candidate vaccines, we employed the C-ImmSim server (20), a platform capable of simulating immune responses. This computational tool mimics the activation of B and T lymphocytes following hypothetical vaccine administration, allowing for the exploration of immune response dynamics. The simulation parameters were configured as follows: Random Seed = 12,345, Simulation Volume = 10, Simulation Steps = 240, and HLA selections: A0101, B0702, and DRB1_0101. The simulation framework was designed to include three administrations of 400 antigens, each spaced by a two-week interval. Each time step was delineated to represent an elapsed duration of 8 hours in real-world time, leading to time periods set at 1, 42, and 84. The simulation predicted the cellular immune responses provoked by candidate vaccines, encompassing antibody production, B cell and T cell activation, and cytokine release. The vaccine demonstrating the highest titer of IgG2 antibody were subsequently selected for further analysis, given that the IgG2 subtype is known to mediate the function against SOST in ROMO (21).

Immunological properties of the selected vaccines were systematically evaluated. Allergenicity assessments were performed using AllerTOP v.2.1 server (22), while antigenic potential was analyzed with VaxiJen v2.0 server (23). To assess solubility of the vaccines, SOLpro server (24) was employed. Additionally, the physicochemical properties—including chemical formula, total atom count, molecular weight, theoretical isoelectric point (pI), half-life, instability index, aliphatic index, and the grand average of hydropathicity (GRAVY)—were predicted using ExPASy ProtParam tool (25).

Secondary structure elements of the selected vaccines, including α-helices, extended strands, β-turns, and random coils, were predicted using SOPMA server (26) and PSIPRED web server (27). For these predictions, all parameters were maintained at their default settings. Additionally, the solubility characteristics of the selected vaccines were assessed using Protein-Sol server (28).

Top-ranked model of tertiary structure for the selected vaccines, generated by AlphaFold2, was refined using GalaxyRefine web server (29). This refinement yielded reliable core structures based on multiple templates, while less reliable loops and terminal regions were constructed through optimization-based modeling. The structural quality of the refined vaccine model was further assessed using ProSA-web (30), ERRAT (31) and PROCHECK (32).

Molecular docking analyses were conducted using HawkDock server (13) to evaluate the interactions between vaccine candidates and Toll-Like Receptor 2 (TLR2) immune receptor (PDB ID: 6NIG). This platform organizes docking models based on surface complementarity and clustering characteristics. The highest-ranking model derived from the docking evaluations was selected for further analysis and visualized with PyMOL software. Binding energy and interaction surfaces within the docking complex were assessed using Prodigy (33), PDBePISA (34), and PDBsum (14).

Molecular dynamics simulations of the vaccine-TLR2 docking complex were performed utilizing the internal coordinate normal mode analysis server (iMODS) (35). This platform employs Normal Mode Analysis (NMA) in internal coordinates to identify collective motions that are critical for the functional dynamics of macromolecules. iMODS provides interactive tools for visualizing these modes, including vibration analysis, motion animations, and morphing trajectories.

Codon-optimized cDNA sequences for the selected vaccine candidates were generated in silico using the Java Codon Adaptation Tool (JCAT) (36). Optimized sequences were then cloned into pSmartI plasmids. Following cloning, the recombinant plasmids were transformed into Escherichia coli BL21 (DE3) for protein expression. The resulting recombinant proteins were purified through a two-step chromatography process, which involved ion exchange chromatography followed by gel filtration chromatography. Purity and quality of protein products were assessed using 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (Sangon Biotech, Shanghai, China).

For immunization studies, C57BL/6J mice (purchased from Guangdong Medical Laboratory Animal Center, China) received two subcutaneous injections of 200 µg of each vaccine candidate, administered two weeks apart(n=3). Freund’s Complete Adjuvant was used for the initial dose, while Freund’s Incomplete Adjuvant (Sigma, USA) was employed for the booster injection. Blood samples were collected five weeks after the final immunization, and anti-SOST antibody titers were measured using ELISA.

The antiserum from vaccine-immunized mice was obtained. The titers of specific anti-SOST antibodies were assessed using an indirect ELISA. In brief, 1 µg/mL of human SOST protein (MedChemExpress Inc.) was coated onto the wells of MaxiSorp microtiter plates (Thermo Fisher Scientific Inc.) and incubated overnight at 4°C. Mouse serum samples were diluted 1:200 in sample dilution buffer and added to the pre-coated plates, followed by incubation at room temperature for 2 hours. After washing the plates with washing buffer, bound IgG was detected using a horseradish peroxidase-conjugated goat anti-mouse IgG antibody (1:10,000, Abcam). The absorbance was measured at 450 nm using a Multiskan FC microplate reader (Thermo Fisher Scientific, San Jose, USA).

Splenocytes were isolated immediately post-euthanasia via mechanical dissociation of the spleen tissue. Mononuclear cells were then separated using density gradient centrifugation with a murine spleen mononuclear cell isolation kit (Solarbio, Beijing, China). Isolated cells were resuspended at a concentration of 1×10^6 cells/mL in RPMI 1640 medium supplemented with 10% fetal bovine serum. Cells were subsequently stimulated in vitro with recombinant SOST protein (100 ng/mL), vaccine protein (100 ng/mL), or PBS as a control, and incubated for 48 hours at 37 °C in a humidified atmosphere containing 5% CO₂. Post-incubation, culture supernatants were collected and analyzed for cytokine concentrations (IL-4, IL-10, and IFN-γ) using ELISA kits (Meimian Industrial Co., Ltd., China), following the manufacturer’s instructions.

Functional activity of anti-SOST antiserum from vaccine-immunized mice was evaluated using primary osteoclasts and osteoblasts. Mice were euthanized via CO₂ inhalation, starting with a flow rate of 10% of chamber volume per minute to gradually increase CO₂ concentration to 30%, inducing unconsciousness. Once righting reflex was lost, the flow rate was increased to 30% per minute to maintain a CO₂ concentration of ≥70% for 5 minutes, ensuring humane euthanasia. All procedures complied with animal welfare guidelines. Primary osteoclasts were isolated from tibiae and femora of 8-week-old C57BL/6J mice. Bone marrow mononuclear cells were extracted using an isolation kit, filtered, and cultured in α-MEM supplemented with 50 ng/mL M-CSF and 80 ng/mL sRANKL (PeproTech) for 4–6 days to induce differentiation. During the second medium change, anti-SOST antiserum and 100 ng/mL recombinant SOST protein (Novoprotein) were added. Osteoclast differentiation was confirmed by TRAP staining (Servicebio). Primary osteoblasts were obtained from bone marrow stromal cells, and the MC3T3-E1 subclone 14 osteoblast cell line (purchased from Pricella Biotechnology Co., Ltd.) was cultured in osteogenic medium. Co-cultures of osteoblasts with anti-SOST antiserum and 200 ng/mL SOST were established, and mineralization was evaluated on day 21 via Alizarin Red S staining (Solarbio). The culture medium was refreshed every 2–3 days throughout the experiment.

Data are presented as means ± standard deviation (SD). Differences between two independent groups were analyzed using one-way ANOVA followed by Tukey’s multiple comparisons test. Data visualization was conducted utilizing GraphPad Prism software version 10 (GraphPad Software, San Diego, CA, USA). A p-value of less than 0.05 was considered statistically significant.

To identify potential interaction sites of SOST with ROMO for development of recombinant subunit vaccines, we initially fragmented SOST protein into six peptides, each comprising approximately 30 amino acids. Screening through ELISA pinpointed two peptides, SOST_114–143 and SOST_144-173, that exhibited high-affinity binding to ROMO (Figure 1A). We further dissected the identified region (amino acids 114-173) into four peptides based on their biological properties (Figure 1B). A subsequent ELISA revealed that SOST_131–163 peptide served as a specific epitope with substantial affinity for ROMO (Figure 1C), located within loop3 domain of SOST (Figure 1D-a).

Molecular docking studies indicated that SOST_131–163 fragment interacts with the light chain of ROMO’s variable domain, establishing 4 hydrogen bonds and 105 non-bonded contacts, resulting in a binding free energy of -25.8 kcal/mol and an interface area of 712.9 Ų (Figure 1D-b, Supplementary Figure S1A). Furthermore, SOST_131–163 demonstrated affinity for the heavy chain’s variable domain, forming 2 salt bridges and 90 non-bonded contacts, with a binding free energy of -33.19 kcal/mol and an interface area of 451.6 Ų (Figure 1D-c, Supplementary Figure S1B). These docking results corroborate that SOST_131–163 is a critical and distinctive peptide for ROMO, aligning with our ELISA observations.

Identification of immunodominant epitopes is pivotal for effective vaccine design. In this study, NetMHCpan 4.1 EL tool was utilized to predict six cytotoxic T lymphocyte (CTL) epitopes in SOST_131–163 fragment, consisting of four weak binders and two strong binders (Figure 1E). Additionally, NetMHCIIpan 4.1 EL server was employed to forecast five helper T lymphocyte (HTL) epitopes, comprising four weak binders and one strong binder (Figure 1F). For the prediction of B cell epitopes, linear epitopes were analyzed using BepiPred 2.0 tool, resulting in the identification of two distinct epitopes within SOST_131–163 fragment (Figure 1G). Furthermore, conformational B cell epitopes were evaluated using ElliPro tool, which yielded two additional epitopes (Figure 1H).

To enhance vaccine efficacy in inducing antibodies, we conjugated DTT scaffold, which contains substantial T-helper epitopes capable of disrupting immune tolerance to autoantigens, with varying quantities of repetitive SOST_131–163 epitopes via a (GGGGS)₂ linker. Six recombinant vaccines were constructed using this method, labeled DS₀ to DS₅, corresponding to the incorporation of 0 to 5 copies of SOST_131–163 epitopes into DTT scaffold (Figure 2A). Tertiary structures of these candidate vaccines were predicted using AlphaFold2 server, which generated five structural models for each vaccine. The top-ranked model for each vaccine was selected based on the highest predicted Local Distance Difference Test (pLDDT) score, and the resulting structures are presented (Figure 2B).

To evaluate immune-stimulating potential of the top-ranked model for each candidate vaccine, we employed C-IMMSIM online server. The results demonstrated that all vaccine candidates elicited relatively high antibody titers following three immunization injections. Notably, the DS₃ and DS₅ vaccines stimulated the production of multiple antibody isotypes, including IgM, IgG1 and IgG2, while other candidate predominantly induced IgM and IgG1 responses (Figure 2C). Importantly, the production of the IgG2 subtype is particularly significant as it is the functional antibody associated with ROMO. Since the candidate vaccines are designed to elicit an IgG2 antibody response that closely resembles that of ROMO (21), we selected the DS₃ and DS₅ vaccines for further analysis and investigation.

Safety and efficacy are critical criteria for evaluating vaccines. Analysis using AllerTOP v.2.1 revealed that both DS₃ and DS₅ vaccines are non-allergenic. Antigenicity of these vaccines was assessed through VaxiJen v2.0, yielding scores of 0.7434 for DS₃ and 0.7948 for DS₅, both surpassing the threshold value of 0.5. These results indicate that DS₃ and DS₅ vaccines are not only safe but also exhibit high immunogenicity. Solubility assessments conducted via SolPro server produced favorable scores of 0.713 for DS₃ and 0.925 for DS₅. Additional physicochemical parameters were predicted using ExPASy ProtParam server. Both vaccines are classified as recombinant proteins, with molecular weights of 31.8 kDa for DS₃ and 40.3 kDa for DS₅, and isoelectric points of 9.14 and 9.61, respectively. The total atom counts were recorded as 4463 for DS₃ and 5657 for DS₅. Estimated half-lives for both vaccines are approximately 30 hours in mammalian reticulocytes, over 20 hours in yeast, and exceeding 10 hours in Escherichia coli. The instability indices were calculated to be 42.19 for DS₃ and 39.93 for DS₅, while aliphatic indices measured 82.72 and 79.38, respectively. The grand average of hydropathicity (GRAVY) values were -0.257 for DS₃ and -0.294 for DS₅ (Table 1).

Secondary structure compositions of DS₃ and DS₅ vaccine were analyzed using SOPMA server. DS₃ candidate exhibited a secondary structure composed of 50.33% α-helices (152/302), 37.09% random coils (112/302), and 12.58% extended strands (38/302) (Figure 3A). Additionally, DS₃ demonstrated enhanced solubility, with a Protein-Sol score of 0.574, surpassing the baseline threshold of 0.45 (Figure 3B). Confidence in the secondary structure predictions for DS₃ was further assessed using the PESIPRED web server, which provided favorable results (Figure 3C). In contrast, DS₅ candidate displayed a distinct secondary structure profile, characterized by 11.66% α-helices (45/386), 72.28% random coils (279/386), and 16.06% extended strands (62/386) (Figure 3D). Similarly, DS₅ exhibited enhanced solubility, with a Protein-Sol score of 0.647, exceeding the baseline value of 0.45 (Figure 3E). Confidence of the secondary structure predictions for DS₅ was also assessed using PESIPRED web server, yielding favorable results (Figure 3F).

Top-ranked models for DS₃ and DS₅ vaccines, predicted using AlphaFold2, exhibited pLDDT scores of 37.1 and 45.1, respectively. The pLDDT scores, which range from 0 to 100, serve as an indicator of model confidence, where values above 80 reflect high confidence in the accuracy of residue structure, and scores below 50 suggest the presence of disordered regions. Both DS₃ and DS₅ displayed pLDDT scores below the confidence threshold of 50, warranting further refinement.

Refinement was conducted using GalaxyRefine server, resulting in the generation of five refined models for each vaccine candidate. Optimal model quality is characterized by higher Global Distance Test High Accuracy (GDT-HA) and Ramachandran values, and lower root-mean-square deviation (RMSD), MolProbity scores, clash scores, and counts of poor rotamers. For DS3, Model 1 demonstrated the most favorable refinement metrics, achieving GDT-HA of 0.9007, RMSD of 0.580 Å, MolProbity score of 1.459, clash score of 3.7, poor rotamer count of 0.4, and Ramachandran favored percentage of 95.7% (Figure 4A, Supplementary Table S1). Similarly, Model 1 of DS5 exhibited optimal refinement results, with GDT-HA of 0.9424, RMSD of 0.480 Å, MolProbity score of 1.772, clash score of 10.2, poor rotamer count of 0.0, and Ramachandran favored of 96.4% (Figure 4E, Supplementary Table S2).

Ramachandran plots generated via PROCHECK indicated that 94.3% of DS₃ residues reside in the most favored regions, with 5.7% in additional allowed regions, while no residues were found in generously allowed or disallowed regions (Figure 4B). Conversely, for DS₅, 93.7% of residues were in the most favored regions, 6.3% in additional allowed regions, with none in generously allowed or disallowed regions (Figure 4F). The Z-scores calculated from ProSA-web were -5.69 for DS₃ model and -7.33 for DS₅ model (Figures 4C, G), both of which fall within the acceptable range for conformational scores typical of experimentally validated protein structures. DS₃ vaccine exhibited a quality factor of 90.8451 according to ERRAT, while DS₅ vaccine received a quality factor of 63.2867. Furthermore, energy plots corresponding to the amino acid compositions for both DS₃ (Figure 4D) and DS₅ (Figure 4H) were analyzed, providing additional evidence for structural integrity of the refined models.

Analysis using IEDB reveals that DS₃ vaccine exhibits a high density of T-cell epitopes, comprising 13 strong and 34 weak CTL binders, as well as 3 strong and 47 weak HTL binders (Supplementary Table S3). Additionally, the ElliPro tool identifies 14 conformational B-cell epitopes associated with DS₃ vaccine (Figure 5A, Table 2). In contrast, DS₅ vaccine also demonstrates a rich repertoire of T-cell epitopes, featuring 17 strong and 42 weak CTL binders, along with 5 strong and 55 weak HTL binders (Supplementary Table S3). Moreover, a total of 10 conformational B-cell epitopes are predicted for DS₅ vaccine, as indicated by ElliPro (Figure 5B, Table 2).

Molecular docking and molecular dynamics simulations were performed to elucidate the interactions between the DS₃ and DS₅ vaccines and Toll-like receptor 2 (TLR2). For molecular docking analysis, HawkDock was conducted to generate ten docking models for each vaccine. The optimized docking models revealed binding scores of -6618.72 kcal/mol for DS₃ and -8177.20 kcal/mol for DS₅, suggesting a stronger interaction between DS₅ and TLR2. Further structural analysis of DS₃-TLR2 and DS₅-TLR2 complexes indicated distinct binding energies and interface areas. Specifically, DS₃-TLR2 complex demonstrated a binding energy of -46.21 kcal/mol with an interface area of 1285.9 Ų (Figure 6A), whereas DS₅-TLR2 complex exhibited a higher binding energy of -79.05 kcal/mol and a surface area of 1245.0 Ų (Figure 7A). DS₃-TLR2 complex was characterized by the presence of 1 salt bridge, 6 hydrogen bonds, and 137 non-bonded contacts (Figure 6B), while DS₅-TLR2 complex featured 6 salt bridges, 7 hydrogen bonds, and 142 non-bonded contacts (Figure 7B).

Both complexes demonstrated stability, as evidenced by their B-factors (Figures 6C, 7C), deformability profiles (Figures 6D, 7D), and low eigenvalues (8.476709e-06 for DS₅-TLR2 and 1.077629e-05 for DS₃-TLR2) (Figures 6E, 7E), as well as variance analyses (Figures 6F, 7F). Covariance matrix evaluations highlighted correlations, as well as uncorrelated and anti-correlated motions among the residues within the complexes (Figures 6G, 7G). Elastic network analysis illustrated spring-like interactions between atoms, with darker gray depicting stiffer springs (Figures 6H, 7H). Collectively, these findings suggest that both DS₃ and DS₅ vaccines effectively engage TLR2, potentially eliciting robust immune responses.

To assess the adaptive immune responses elicited by DS₃ and DS₅ vaccines, we employed C-IMMSIM server to simulate in vivo immune reactions. Our analysis revealed an increase in the total B cell population, including B-memory cells and IgM isotypes, which contributed to a significant rise in activated B cells in the host (Figures 8, 9A, B). Following the second immunization, the total count of T helper (TH) cells exhibited a rapid increase, peaking after the third immunization (Figures 8, 9C). Both activated and resting TH cell populations surged after each injection, primarily comprising TH1 cells (Figures 8, 9D, E), suggesting effective antibody maturation processes. Notably, the population of anergic (y2) T cells remained stable throughout the duration of the study (Figures 8, 9F). In contrast, the count of activated T cytotoxic (TC) cells showed a transient increase followed by a decline, while resting TC cells exhibited the opposite trend (Figures 8, 9G).

Additionally, we evaluated the effects of DS₃ and DS₅ vaccines on innate immune cell populations. Natural killer (NK) cells, dendritic cells (DCs), and active epithelial cell populations demonstrated a relatively stable response upon immunization (Figures 8, 9H, J, K). Upon initial immunization, there was a marked increase in both active and resting macrophage populations in a short time, which subsequently reached a peak and stabilized (Figures 8, 9I). Approximately four weeks after the third immunization, we observed a decline in the number of active macrophages, coinciding with a rapid increase in resting macrophages (Figures 8, 9I). Following the administration of DS₃ and DS₅ vaccines, there was an activation of downstream inflammatory mediators, with significant elevations in levels of IFN-γ and IL-2 (Figures 8, 9L). Collectively, these findings indicated that the DS₃ and DS₅ vaccines effectively stimulate both innate and adaptive immune responses, highlighting their potential as effective vaccine candidates.

Codons for the optimized DS₃ and DS₅ sequences were successfully cloned into pSmartI plasmids at XhoI restriction sites (Figures 10A, D). The recombinant plasmids were confirmed through PCR amplification, as shown by agarose gel electrophoresis (Figures 10B, E). Subsequently, the recombinant plasmids were transformed into Escherichia coli BL21(DE3), leading to the successful expression and purification of DS₃ and DS₅ vaccine proteins, which exhibited molecular weights of 31.8 kDa and 40.3 kDa, respectively (Figures 10C, F).

To evaluate immunogenic potential of DS₃ and DS₅ vaccines, mice were immunized with DS₃ and DS₅ proteins at a dose of 200 µg per mouse, with a two-week interval between doses (n=3). Serum was collected at the seventh week to assess antibody production against SOST (Figure 10G). Our results demonstrated that both DS₃ and DS₅ effectively elicited a significant immune response, resulting in the production of high titers of anti-SOST antibodies in the immunized mice. The antibody titers induced by both vaccines were significantly higher than those observed in PBS control group (Figures 10H).

To assess T cell responses to vaccine stimulation, cytokine production associated with Th1 and Th2 responses was quantified using DS₅ as a representative antigen. Specifically, levels of IL-4 and IL-10 (Th2 markers) and IFN-γ (Th1 marker) were measured in splenocyte cultures stimulated with PBS, SOST, or DS₅ (Figures 10I–K). IFN-γ levels did not differ significantly among the PBS and DS₅ groups, indicating that DS₅ does not elicit a robust Th1-mediated cytotoxic response (Figure 10K). In contrast, IL-4 and IL-10 secretion were reduced following stimulation with both SOST and DS₅ (Figures 10I, J). Notably, cytokine levels in DS₅-treated splenocytes remained higher than in SOST-treated cells, suggesting that the DS₅ vaccine induces a moderated Th2 response that may support B cell-mediated anti-SOST antibody production.

The antiserum obtained from vaccinated mice, which exhibited the highest antibody titer, was selected for in vitro functional assays using primary osteoclasts and osteoblasts. To model the in vivo role of SOST, which promotes osteoclast differentiation and inhibits osteoblast maturation, recombinant SOST was co-cultured with the respective cell types. Results showed that SOST supplementation had no sinificantly effect on osteoclast differentiation and maturation, however, the addition of the antiserum significantly attenuated osteoclast differentiation and maturation, leading to a marked reduction in osteoclastogenesis (Figures 11 A, B). In primary osteoblasts (Figures 11C, D) and the MC3T3-E1 subclone 14 cell line (Figures 11E, F), SOST partially suppressed differentiation and mineralization; however, the presence of antiserum mitigated these inhibitory effects, thereby restoring osteoblast mineralization capacity (Figures 11C–F). These findings demonstrate that the vaccine-induced antiserum effectively inhibits osteoclast activity and enhances osteoblast function, confirming its functional efficacy. Moreover, these results provide preliminary evidence supporting the vaccine’s potential as a therapeutic strategy for osteoporosis.