Sema3A Restores Bone Homeostasis in Peri-Implantitis by Activating Osteoblasts and Suppressing Osteoclasts via PI3K/AKT Authors: Chenjiang He, Fang Dai, Jingting Wu, Libin Deng, Meixiu Jiang, Yuting Yang, Li Li, Fancheng Xu, Li Song Published: 2026 ABSTRACT AbstractBackgroundPeri-implantitis, a major cause of implant failure, is driven by disrupted bone homeostasis. Semaphorin 3A (Sema3A) bidirectionally regulates bone remodelling, yet its role in peri-implantitis remains poorly defined. This study aimed to elucidate the mechanisms by which Sema3A modulates bone homeostasis in peri-implantitis.Materials and methodsSingle-cell transcriptomic analysis (GSE171213) was performed on a periodontitis-derived periodontal tissue dataset to characterize bone-related cell communication. A murine peri-implantitis model was established via silk ligation, and bone loss was assessed by micro‑CT. In vitro, MC3T3-E1 osteoblasts were exposed to Porphyromonas gingivalis LPS and transduced with Sema3A‑overexpressing lentivirus. Proliferation, migration, osteogenic differentiation, and the PI3K/AKT pathway were evaluated. An osteoblast–osteoclast coculture system was used to assess osteoclastogenesis.ResultsReanalysis of a periodontitis-derived single-cell dataset showed reduced SEMA3A expression and a lower osteogenic-lineage score in sparsely represented osteogenic-lineage-like cell populations under inflammatory conditions. Peri‑implantitis mice exhibited reduced SEMA3A, decreased P1NP, elevated CTX‑1, and significant bone loss (P < .05). Sema3A overexpression significantly rescued LPS‑impaired osteoblast proliferation, migration, and osteogenic differentiation (P < .01). RNA‑seq identified PI3K/AKT pathway enrichment; Sema3A increased phosphorylated AKT, and the PI3K/AKT inhibitor Palomid 529 abolished these protective effects. In coculture, Sema3A‑overexpressing osteoblasts suppressed osteoclast differentiation, reducing TRAP⁺ cells, Nfatc1, and Ctsk (P < .01).ConclusionsSema3A restores osteoblast function under inflammatory conditions, partly through PI3K/AKT signalling, and suppresses osteoclast differentiation indirectly in an osteoblast–osteoclast coculture system. Notably, Sema3A overexpression alone had no effect under normal conditions but specifically counteracted inflammatory damage. These findings suggest Sema3A is a potential therapeutic target for peri‑implantitis.Clinical significanceSema3A simultaneously enhances osteogenesis and inhibits bone resorption specifically under inflammatory conditions, while preserving basal bone homeostasis, thereby representing a targeted bone-protective strategy for peri-implantitis that may improve implant survival and reduce the need for invasive surgical interventions.Key words: Peri-implantitis, Semaphorin-3A, Homeostasis, Porphyromonas gingivalis Introduction Implant-supported prostheses are clinically regarded as one of the most reliable treatment modalities for missing teeth. It has been shown that the success rate of implant denture restorations is 86.2% to 83.8% in a 15- to 25-year follow-up of implant patients.1 However, peri-implantitis is increasingly recognized as a leading cause of implant failure.2 Wada et al3 reported that peri-implantitis was present in 4.8% to 23% of implants and in 8.9% to 45% of implant patients. The pathogenesis of peri-implantitis is highly complex, involving interactions among microbial biofilms, occlusal overload, and the host immune response.4,5 Despite considerable research attention devoted to its pathogenesis, elucidating the underlying mechanisms and developing effective mechanism-based therapies remain a major clinical challenge. Bone homeostasis is intricately maintained by a balance between bone formation and bone resorption.6 In recent years, several studies have emphasized the central role of osteoblasts and osteoclasts in peri-implantitis bone homeostasis.7,8 In particular, the phenotypic changes of osteoblasts and osteoclasts play an important role in bone homeostasis, such as proliferation, migration, osteogenic differentiation,9 and osteoclastic differentiation.10 Although several studies have demonstrated differences in various cytokines between peri-implantitis and healthy implants, their relevance to bone homeostasis, effects on cellular phenotypes, and associated intracellular signalling mechanisms remain incompletely understood. Semaphorin 3A (Sema3A) is a secreted member of the class 3 semaphorin family and is structurally composed of four parts: the semaphorin domain, the Plexin-Semaphorin-integrin domain, the immunoglobulin-like domain, and the short basic domain.11 In recent years, it has been shown that Sema3A plays a critical role in the regulation of bone homeostasis.12 Sema3A has a bidirectional role in regulating bone homeostasis: on the one hand, it promotes osteogenic differentiation through the classical Wnt/β-catenin signalling pathway; on the other hand, osteoclast formation is inhibited through the natural killer-activated receptor-associated protein 12 (DAP12)-dependent immunoreceptor tyrosine-based activation motif signalling pathway and the RhoA signalling pathway.13,14 This further implies a potential link between Sema3A and peri-implantitis bone homeostasis. Although several studies have shown the paramount importance of Sema3A in immunomodulation and alveolar bone remodelling in periodontitis (PD),15,16 its role in peri-implantitis bone homeostasis and related mechanism has not been established. In this study, we aimed to discuss the regulatory mechanisms of Sema3A signalling in bone homeostasis and explore its function in the bone homeostasis of peri-implantitis. We conducted the analysis of single-cell data with clinical samples and then constructed in vivo animal model of peri-implantitis and in vitro coculture model of osteoblasts and osteoclasts overexpressing Sema3A. We observed the phenotypes and detected the relevant pathways using RNA-seq. Materials and methods Analysis of single-cell data The publicly available single-cell RNA sequencing dataset GSE171213 was obtained from the Gene Expression Omnibus database.17 This dataset was originally generated to characterize the osteoimmunological microenvironment of human PD and comprised periodontal tissues from 12 participants, including four healthy controls (HC), five patients with untreated severe chronic PD, and three patients with severe chronic PD after initial periodontal therapy (PDT). The collected specimens contained periodontal tissue components, including gingival tissue, periodontal ligament, and alveolar bone. In the present study, only the HC and untreated PD samples were included to investigate inflammation-associated changes without potential treatment-related confounding; the PDT samples were excluded from the comparative analysis. Quality control and downstream analyses were performed using Seurat in R. Cells with fewer than 200 detected genes or more than 40% mitochondrial transcripts were excluded. After quality control, 34,683 cells and 36,163 genes were retained. Gene-expression data were normalized using SCTransform, and Harmony was applied to reduce intersample batch effects.18,19 Dimensionality reduction and graph-based clustering were performed using the RunPCA, RunUMAP/RunTSNE, FindNeighbors, and FindClusters functions.20 Major cell populations were annotated using canonical lineage markers. Because osteogenic-lineage cells were relatively sparse and showed transcriptional continuity with fibroblast/stromal populations, the fibroblast and myeloid compartments were separately reclustered and conservatively annotated as cell states rather than fully differentiated cell types. Fibroblast-derived populations were classified as fibroblasts, MSC-like stromal cells, pre-OB-like cells, and OB-like cells according to the expression patterns of DCN, COL1A1/COL1A2, CD55, PDGFRA, RUNX2, SP7, and IBSP. Myeloid populations were classified as monocytes, macrophages, OC-like cells, dendritic cells, and proliferative cells according to S100A9, C1QA, ACP5, CD1C, LAMP3, CLEC9A, JCHAIN, CD5, and MKI67 expression. An osteogenic-lineage module score was calculated using RUNX2, SP7, ALPL, COL1A1, COL1A2, IBSP, BGLAP, and SPP1. CellChat was used to explore predicted ligand–receptor interactions among six bone-homeostasis-associated populations: MSC-like stromal cells, pre-OB-like cells, OB-like cells, monocytes, macrophages, and OC-like cells. Pathway-specific network roles were further evaluated for TGF-β, WNT, TNF, and CSF signalling.21 Spearman correlation analysis was performed to examine the association between SEMA3A expression and the osteogenic-lineage score. Given that GSE171213 was generated from periodontal rather than peri-implant tissues, the single-cell analysis was used as an exploratory, hypothesis-generating component and was not interpreted as direct evidence of peri-implantitis-specific cellular alterations. Construction of animal model of peri-implantitis Four-week-old SPF male C57BL/6 mice (GemPharmatech) were acclimated for 1 week and used to establish a peri-implantitis model, approved by the Nanchang University Experimental Animal Ethics Committee (NCULAE-20221031158). All animals were housed under specific pathogen-free conditions with controlled temperature, humidity, and a 12-hour light/dark cycle and had free access to water and a soft diet throughout the experimental period. After acclimatization, the mice were randomly allocated to the nonligated control group and the ligature-induced peri-implantitis group, with five animals in each group. All surgical procedures were performed under microscopic magnification by a trained operator. General anaesthesia was induced and maintained using inhaled isoflurane, and ophthalmic lubricant was applied to prevent corneal drying. The left maxillary first molars were carefully luxated and extracted using a dental explorer and microsurgical forceps. Haemostasis was achieved by applying sterile cotton pellets to the extraction sockets. Postoperatively, the mice received analgesic treatment and were provided with a soft diet. Amoxicillin was administered in the drinking water during the postextraction healing period. Four weeks after tooth extraction, the healed alveolar ridge was exposed through a mesiodistal incision, and full-thickness buccal and palatal flaps were elevated. An osteotomy approximately 1.0 mm in depth was prepared manually using a 0.3-mm-diameter carbide microdrill. A custom-made, screw-shaped, smooth-surface Ti-6Al-4V implant measuring 1.0 mm in length and 0.5 mm in diameter (Appendix Figure 6) was inserted into the healed extraction site in the region of the left maxillary first molars, with one implant placed per animal. Postoperative analgesic and antibiotic treatments were provided, and the implants were allowed to osseointegrate for 4 weeks. Clinical implant stability and complete soft-tissue healing were confirmed before peri-implantitis induction. After the 4-week osseointegration period, the animals were randomly maintained in either the control or peri-implantitis group. In the peri-implantitis group, a 6-0 silk ligature was positioned immediately apical to the implant head and secured with a double knot to facilitate plaque accumulation and induce peri-implant inflammation. The control group underwent no ligature placement. Ligatures were examined every 2 days and were replaced when lost. The ligatures were maintained for 4 weeks, after which the animals were subjected to endpoint evaluation.22,23 At endpoint, blood was collected for serum ELISA, then mice were euthanized by CO₂ asphyxia and maxillae harvested. High-resolution micro-CT (15 μm) was used to evaluate alveolar bone loss; Mimics V21.0 measured the implant platform-to-bone distance and performed 3D reconstruction, while BoneAnalysis quantified bone volume and total volume. Maxillae were decalcified in 10% EDTA (pH 7.6) for 4 to 8 weeks, embedded in paraffin, and sectioned. H&E staining followed standard dewaxing, rehydration, and staining. For immunofluorescence, sections underwent antigen retrieval, primary antibody incubation overnight at 4°C, Alexa Fluor 488‑conjugated secondary antibody, and DAPI counterstaining. Imaging was done with a 3DHISTECH digital slide scanner; antibody details are in the Supplementary Materials. Cell cultures, osteogenic induction and overexpression lentivirus infection MC3T3-E1 preosteoblasts were cultured in MEMα medium with 10% FBS and 1% penicillin-streptomycin.24 Porphyromonas gingivalis lipopolysaccharide (P. gingivalis LPS) was added at 10 μg/mL to create an inflammatory environment, while osteogenic induction solution was used for differentiation. The Sema3A overexpression MC3T3-E1 cell line was created using lentivirus LV-Sema3A. MC3T3-E1 cells were seeded in six-well plates and cultured until they reached approximately 40% to 50% confluence. The cells were then transduced with either the Sema3A-overexpressing lentivirus (LV-Sema3A) or the corresponding empty-vector lentivirus (LV0) at a multiplicity of infection of 10 in the presence of 8 μg/mL polybrene. After 24 hours of incubation, the virus-containing medium was replaced with fresh complete culture medium. Lentiviral transduction efficiency was evaluated by fluorescence microscopy 72 hours after infection. Stable transduced cells were selected using 2 μg/mL puromycin for 48 to 72 hours, followed by maintenance in medium containing 1 μg/mL puromycin for approximately 2 weeks. Sema3A overexpression was subsequently confirmed by real-time quantitative polymerase chain reaction (qRT-PCR) and western blot analysis before the cells were used in further experiments. Total mRNA extraction and fluorescent qRT-PCR and protein immunoblotting RNA was extracted from MC3T3-E1 and RAW264.7 cells using TRIzol reagent, reverse transcribed, and analysed for target gene expression using SYBR Green qRT-PCR kit and 2−ΔΔCT method. Primers for target sequences are listed in Appendix Table 1. Cellular proteins were extracted using RIPA lysate and their concentration determined by BCA method. SDS-PAGE electrophoresis and PVDF membrane transfer were performed. Membranes were sealed with 5% BSA and incubated with specific primary antibodies overnight at 4°C. HRP-labelled secondary antibodies were used, and imaging was done with ECL luminescent reagent. The grey values of protein bands were analysed using ImageJ software. Cell proliferation and migration analyses MC3T3-E1 cell proliferation was measured using the CCK-8 assay in 96-well plates with absorbance values at 450 nm. Cell proliferation was also observed under a fluorescence microscope using EDU and Hoechst 33342 staining. Cell migration was assessed using Transwell assay. Treated cells were placed in the upper chamber and cultured for 24 hours. The number of cells crossing the basement membrane was counted after fixation and staining. Scratch healing on 6-well plates was monitored at 0, 8, and 24 hours using ImageJ software. Analysis of cellular osteogenic differentiation MC3T3-E1 cells infected with empty-vector lentivirus and LV-Sema3A virus were cultured for 14 days in osteogenic induction solution with or without 10 μg/mL P. gingivalis LPS. For osteogenic induction, MC3T3-E1 cells were cultured in α-MEM supplemented with 10% foetal bovine serum, 1% penicillin–streptomycin, 10 mM β-glycerophosphate, 50 μg/mL ascorbic acid, and 100 nM dexamethasone. The osteogenic induction medium was replaced every 2 to 3 days. Where indicated, P. gingivalis LPS was added to the induction medium at the specified concentration and replenished at each medium change. Total cellular proteins were extracted on days 7 and 14, and ALP activity was measured. Cell mineralization was assessed at day 14 using alizarin red S staining and quantified by measuring absorbance at 550 nm after extraction with 10% ethylpyridinium chloride. RNA sequencing and functional analysis of screened genes MC3T3-E1 cells were divided into the LV0 + LPS and LV-Sema3A + LPS groups. Cells in both groups were exposed to P. gingivalis LPS under the same experimental conditions, after which total RNA was extracted using TRIzol reagent according to the manufacturer’s instructions. RNA concentration and purity were assessed using a spectrophotometer, and RNA integrity was evaluated using an automated electrophoresis system. Only RNA samples meeting the quality-control criteria were used for library construction. Messenger RNA was enriched using poly(A) selection, followed by fragmentation, reverse transcription, second-strand cDNA synthesis, end repair, adapter ligation, and library amplification. The constructed libraries were subjected to paired-end sequencing on an Illumina sequencing platform. Raw sequencing reads were assessed using FastQC, and adapter sequences and low-quality reads were removed before downstream analysis. Clean reads were aligned to the mouse reference genome using a standard splice-aware alignment algorithm, and gene-level read counts were generated based on the corresponding genome annotation file. Gene-expression levels were normalized, and differential expression analysis between the LV-Sema3A + LPS and LV0 + LPS groups was performed using DESeq2. Differentially expressed genes (DEGs) were defined using an adjusted P value <.05 and an absolute log2 fold change >1. The Benjamini–Hochberg method was used to control the false discovery rate associated with multiple comparisons. Kyoto Encyclopedia of Genes and Genomes pathway-enrichment analysis was performed to identify biological pathways associated with the DEGs. Enrichment results with an adjusted P value <.05 were considered statistically significant.25 Analysis of osteoclast differentiation An osteoblast–osteoclast coculture system was established using MC3T3-E1 and RAW264.7 cells in Transwell inserts with 0.4-μm pore-size membranes. In the coculture process, 1 mL of medium with 50 ng/mL RANKL was added to the upper chamber and 2 mL of medium to the lower chamber, with changes every 2 days. After 7 days, osteoclasts in the lower chamber were stained using the TRAP Staining Kit and observed under an inverted microscope. For quantitative analysis, TRAP-positive multinucleated cells containing three or more nuclei were defined as osteoclast-like cells. Five randomly selected, nonoverlapping microscopic fields were captured from each well at the same magnification. The number of TRAP-positive multinucleated cells was counted using ImageJ software by an investigator blinded to the experimental grouping. The mean number of TRAP-positive multinucleated cells per field was calculated for each well. All experiments were independently repeated at least three times. Statistical analysis Statistical analyses were performed using GraphPad Prism 8.0 and R software. Data are presented as mean ± standard deviation. Each mouse was treated as an independent biological replicate (n = 5 per group). In vitro experiments were independently repeated at least three times. Two-group comparisons were performed using an unpaired two-tailed Student’s t test as appropriate. Multiple-group comparisons were analysed using one-way ANOVA with Tukey’s post hoc test. Two-way ANOVA with Tukey’s test was used for experiments with two factors. Time-course data were analysed using two-way repeated-measures ANOVA. Spearman’s correlation was used to assess associations. For single-cell analyses, the biological donor was treated as the statistical unit. RNA-seq data were analysed using DESeq2 with Benjamini–Hochberg correction; adjusted P < .05 and |log2 fold change| > 1 were considered significant. All tests were two-sided, and P < .05 was considered statistically significant. Results Reanalysis of a PD-derived single-cell dataset identifies inflammation-associated alterations in osteogenic-lineage and myeloid cell states To explore inflammation-associated changes in periodontal cell populations related to bone homeostasis, we reanalysed the publicly available scRNA-seq dataset GSE171213. The original dataset comprised periodontal tissues from HC, patients with untreated severe chronic PD, and patients after initial PDT.17 Because the present analysis focused on untreated inflammatory changes, only the HC and PD samples were included. Importantly, the specimens were obtained from periodontal rather than peri-implant tissues; therefore, this analysis was regarded as exploratory and was used to generate hypotheses for subsequent validation in peri-implantitis models. After quality control, 34,683 cells were retained and assigned to 10 broad cellular compartments, including mast cells, endothelial cells, monocytes, fibroblasts, T cells, B cells, neutrophils, plasma cells, epithelial cells, and an unassigned population, based on canonical lineage markers (Figure 1A). Cell-type-specific transcriptional differences between HC and PD samples were observed across multiple cellular compartments (Supplementary Figure 1A). Reanalysis of a periodontitis-derived single-cell dataset reveals inflammation-associated changes in osteogenic-lineage and myeloid cell states. (A) t-SNE visualization of 34,683 cells from healthy control (HC) and untreated severe chronic periodontitis (PD) samples, coloured according to the broad cell-type annotations. The dot plot shows the expression of representative canonical markers used to define the major cell populations. Dot size represents the percentage of cells expressing the indicated gene, and colour intensity represents the mean expression level within each population. (B) UMAP visualization of the separately reclustered fibroblast/stromal compartment (upper panels) and myeloid compartment (lower panels), displayed according to HC and PD status. The fibroblast/stromal compartment was conservatively classified into fibroblasts, MSC-like stromal cells, pre-OB-like cells, and OB-like cells. The myeloid compartment was classified into monocytes, macrophages, OC-like cells, dendritic cells, and proliferative cells. (C) CellChat analysis of predicted intercellular communication among MSC-like stromal cells, pre-OB-like cells, OB-like cells, monocytes, macrophages, and OC-like cells. Circle plots show the inferred number of ligand–receptor interactions (left) and aggregate interaction weights/strengths (right). Edge thickness indicates the relative number or strength of predicted interactions, and edge colour denotes the source-cell population. (D) Violin and box plots showing the osteogenic-lineage module score in HC- and PD-derived osteogenic-lineage cells. (E) Violin and box plots showing normalized SEMA3A expression in HC- and PD-derived osteogenic-lineage cells. The analysis included four HC donors and five untreated PD donors. HC, healthy control; MSC, mesenchymal stromal cell; OB, osteoblast; OC, osteoclast; PD, untreated severe chronic periodontitis. *P < .05; ****P < .0001. Because cells displaying osteogenic features were relatively sparse and transcriptionally continuous with fibroblast/stromal cells, the fibroblast compartment was separately reclustered and conservatively annotated into fibroblasts, MSC-like stromal cells, pre-OB-like cells, and OB-like cells rather than being classified as fully differentiated osteogenic cell types. These annotations were supported by the distributions of DCN, CD55, PDGFRA, RUNX2, SP7, and IBSP (Figure 1B and Supplementary Figure 1B). Similarly, reclustering of the myeloid compartment identified monocytes, macrophages, OC-like cells, dendritic cells, and proliferative cells, supported by the expression patterns of S100A9, C1QA, ACP5, CD1C, LAMP3, CLEC9A, JCHAIN, CD5, and MKI67 (Figure 1B and Supplementary Figure 1C). Quantitative assessment further demonstrated the limited representation of mature osteogenic-lineage cells. Among 2034 cells in the reclustered fibroblast/stromal compartment, 1289 cells were classified as fibroblasts, 390 as MSC-like stromal cells, 307 as pre-OB-like cells, and only 48 as OB-like cells. The combined MSC-like, pre-OB-like, and OB-like populations accounted for 36.63% of this compartment. Within the reclustered myeloid compartment, 553 monocytes, 571 macrophages, and 291 OC-like cells were identified, together accounting for 63.94% of the analysed myeloid cells; dendritic and proliferative cells accounted for the remaining populations (Supplementary Figure 1E). To explore potential communication among bone-homeostasis-associated cell states, CellChat analysis was restricted to MSC-like stromal cells, pre-OB-like cells, OB-like cells, monocytes, macrophages, and OC-like cells. The global interaction network indicated extensive predicted communication among the selected stromal/osteogenic and myeloid populations (Figure 1C). Pathway-specific network-role analysis suggested prominent involvement of stromal and osteogenic-lineage populations in TGF-β and WNT signalling, whereas myeloid and OC-like populations showed greater involvement in TNF- and CSF-related networks (Supplementary Figure 1D). The osteogenic-lineage module score was significantly lower in PD-derived cells than in HC-derived cells (Figure 1D). SEMA3A expression within the reannotated osteogenic-lineage compartment was also modestly but significantly reduced in PD samples (Figure 1E). In addition, SEMA3A expression showed a statistically significant positive correlation with the osteogenic-lineage score across the analysed cells (Supplementary Figure 1F). Collectively, these results suggest that periodontal inflammation is associated with attenuation of the osteogenic transcriptional program and reduced SEMA3A expression in sparse osteogenic-lineage-like cell populations. These observations provided exploratory support for subsequently examining Sema3A in peri-implantitis-specific in vivo and in vitro models. Reduced Sema3A expression accompanies bone homeostasis imbalance in peri‑implant tissues of a peri‑implantitis animal model To further validate the relationship between the inflammatory environment, dysregulated Sema3A expression, and disruption of bone homeostasis, we established an experimental peri‑implantitis mouse model (implants placed 4 weeks after tooth extraction, then silk‑ligated 4 weeks later). Micro‑CT revealed obvious peri‑implant bone resorption – increased platform‑to‑bone distance and decreased bone volume/total volume ratio – and three‑dimensional reconstruction showed discontinuous bone margins with marked local destruction (Figure 2A, B). Serum ELISA demonstrated significantly lower osteogenesis‑related P1NP and SEMA3A, and higher bone resorption‑related CTX‑126 in the peri‑implantitis group (Figure 2C). qPCR of gingival tissue confirmed decreased Sema3a mRNA, upregulated inflammatory genes (Mmp9, Il-6, Tnf-α), downregulated osteogenic genes (Runx2, Col1a1), and upregulated osteoclastic genes (Nfatc1, Ctsk) (Figure 2D, E). Histologically, Masson staining showed loose fibrous tissue in peri‑implantitis vs abundant mature collagen in controls, and H&E staining confirmed successful modelling (Figure 2F). IF staining revealed markedly weaker SEMA3A and OPG signals, with stronger RANKL signal,27 indicating a pro‑resorptive microenvironment (Figure 2G). Collectively, persistent low Sema3A expression in peri‑implantitis closely associates with suppressed bone formation, enhanced bone resorption, and disrupted bone homeostasis. Bone homeostasis imbalance around implants in a peri‑implantitis animal model is associated with reduced Sema3A expression. (A) Establishment of the peri‑implantitis animal model. (B) Representative Micro‑CT images in sagittal, coronal, and transverse planes, along with 3D reconstructions of the maxillary alveolar bone in the control and peri‑implantitis groups. The red area indicates the distance from the platform to the bone, providing a clear visual reference for specimen identification (n = 5). (C) ELISA detection of serum levels of SEMA3A, P1NP, and CTX‑1 in the peri‑implantitis animal model. (D) qPCR analysis of Sema3a, Mmp9, Il-6, and Tnf-α mRNA in gingival tissue. (E) qPCR analysis of Runx2, Col1a1, Nfatc1, and Ctsk mRNA in gingival tissue. (F) Representative H&E and Masson staining images of the control and peri‑implantitis groups. (G) Representative IF staining images showing the expression and localization of SEMA3A (red), OPG (red), RANKL (green), and DAPI (blue) in gingival tissue sections from the control and peri‑implantitis groups. Statistical significance between the two groups was calculated using unpaired two‑tailed Student’s t test. Data are presented as mean ± SD. *P < .05, **P < .01, ***P < .001. Sema3A increases cell migration and proliferation of osteoblast in the inflammatory microenvironment Having demonstrated, through single‑cell sequencing and in vivo experiments, that significant bone homeostasis imbalance exists in peri‑implantitis tissues and that this imbalance correlates with reduced Sema3A expression, we next sought to gain a deeper understanding of how the Sema3A protein affects osteoblasts under an inflammatory microenvironment. The mouse preosteoblast MC3T3-E1 cell line was used to simulate the inflammatory microenvironment in vitro using LPS, and a Sema3A-overexpressing strain was constructed. A concentration-response experiment was performed to select the LPS concentration. Treatment with 10 μg/mL P. gingivalis LPS induced a pronounced inflammatory response without the marked cytotoxicity observed at 50 μg/mL. Therefore, 10 μg/mL LPS was used in subsequent experiments (Supplementary Figure 3). The results showed that Sema3A overexpression alone had no significant effect on osteoblast activity and proliferation, but under inflammatory conditions, Sema3A overexpression increased osteoblast activity and proliferation capacity (Figure 3A, C, D). Sema3A overexpression enhances proliferation and migration of MC3T3-E1 osteoblasts under Porphyromonas gingivalis LPS-induced inflammatory conditions. (A and C) EdU staining showed no significant difference in proliferation between LV0 and LV-Sema3A groups, while proliferation was significantly increased in the LV-Sema3A + LPS group compared with LV0 + LPS. (D) CCK-8 assays demonstrated no significant differences among groups at 24 hours. No significant difference was observed between LV0 and LV-Sema3A at any time point, whereas LV-Sema3A + LPS showed significantly higher proliferation than LV0 + LPS at 48, 72, and 96 hours. (B and E) Scratch assays revealed no significant difference in wound closure between LV0 and LV-Sema3A groups, while LV-Sema3A + LPS exhibited significantly enhanced migration compared with LV0 + LPS at 8 and 24 hours. (F) Transwell assays further confirmed that migration was comparable between LV0 and LV-Sema3A groups, whereas significantly more cells migrated in the LV-Sema3A + LPS group than in the LV0 + LPS group. ns, difference was not statistically significant, *P < .05, **P < .01, ***P < .001. Meanwhile, in the scratch assay and Transwell assay, Sema3A overexpression had no effect on the migration of osteoblasts in the normal environment, but in the LPS-mimicked inflammatory microenvironment, Sema3A overexpression accelerated the healing of osteoblast scratches (Figure 3B, E) and increased the number of cells passing through the lower compartment of the Transwell (Figure 3F, Supplementary Figure 2). Overexpression of Sema3A reverses the inhibition of osteogenic differentiation in the inflammatory microenvironment To assess the effect of Sema3A on osteogenic differentiation of peri-implant osteoblasts, we analysed alizarin red staining, ALP staining, and activity, as well as protein and gene expression levels of relevant osteogenic markers in each experimental group. The results showed that overexpression of Sema3A enhanced cellular mineralization of osteoblasts as well as osteoblast ALP activity, both in normal and inflammatory microenvironments (Figure 4A, B). Meanwhile, overexpression of Sema3A also rescued the reduction of protein, gene expression levels of osteogenic markers in the inflammatory microenvironment (Figure 4C, D). Sema3A overexpression promotes osteogenic differentiation of MC3T3-E1 cells under both normal and Porphyromonas gingivalis LPS-induced inflammatory conditions. (A) Representative Alizarin Red S and ALP staining images. For each staining assay, the upper panels show whole-well images observed macroscopically, whereas the lower panels show representative microscopic images. LPS markedly reduced mineralized nodule formation and ALP staining in LV0 cells, whereas Sema3A overexpression enhanced osteogenic differentiation and partially reversed the inhibitory effects of LPS. (B) Quantitative analysis of Alizarin Red S staining and ALP activity. Alizarin Red S staining was quantified after extraction of the mineralized nodules, and ALP activity was measured on days 7 and 14. LPS significantly reduced mineralization and ALP activity in LV0 cells, whereas Sema3A overexpression rescued these reductions. Sema3A overexpression also increased ALP activity under basal conditions. (C and D) Western Blot experiment demonstrated the inhibition of osteogenic markers by inflammatory microenvironment, while overexpression of Sema3A could increase the expression of osteogenic markers. qRT-PCR experiments similarly suggested that at the gene expression level, overexpression of Sema3A was able to rescue the inhibition of osteogenic markers in osteoblasts by the inflammatory microenvironment. ns, difference was not statistically significant, *P < .05, **P < .01, ***P < .001, ****P < .0001. Sema3A signalling activates the PI3K/AKT signalling pathway in the inflammatory microenvironment RNA-seq analysis of blank control and Sema3A overexpressing MC3T3-E1 cells in the inflammatory microenvironment identified 172 DEGs, of which 127 were up-regulated, and 45 were down-regulated. KEGG pathway analysis showed that these DEGs were mainly enriched in 65 pathways, including the PI3K-AKT signalling pathway (Figure 5A), which is important for osteoblast proliferation, migration, and differentiation.28 Sema3A signalling activates the PI3K/AKT signalling pathway in the inflammatory microenvironment. (A) KEGG pathway analysis of 172 DEGs from both experimental groups. (B) Representative Alizarin Red S and ALP staining images following treatment with the PI3K/AKT pathway inhibitor P529. For each staining assay, the upper panels show whole-well images observed macroscopically, whereas the lower panels show representative microscopic images. Sema3A overexpression restored mineralized nodule formation and ALP staining under LPS-induced inflammatory conditions, whereas P529 attenuated these protective effects. (C) Western blot analysis. Sema3A overexpression restored LPS-suppressed p-AKT but did not alter p-AKT under basal conditions. P529 did not affect Sema3A expression levels, but blocked the Sema3A‑mediated upregulation of osteogenic markers. (D) CCK-8 assay showing that P529 attenuated the Sema3A-mediated restoration of osteoblast viability and proliferation under LPS-induced inflammatory conditions. (E) Scratch-wound assay showing that P529 inhibited the Sema3A-mediated restoration of osteoblast migration under inflammatory conditions. Data are presented as mean ± SD from independent experiments. ns difference was not statistically significant, *P < .05, **P < .01, ***P < .001. It was further demonstrated that Sema3A overexpression under inflammatory conditions could enhance AKT protein phosphorylation, increase the expression of osteogenic markers, and promote cell proliferation, migration and mineralization. However, when Palomid 529,29 was added (Supplementary Figure 4), the Sema3A-induced improvements in cell function were blocked (Figure 5B-E). Overexpression of Sema3A in osteoblasts in the inflammatory microenvironment regulates osteoclast differentiation To further investigate the effect of Sema3A on bone homeostasis, we established an osteoblast-osteoclast coculture system. Osteoblasts from different experimental groups were placed in the upper chamber of Transwell, and osteoclasts RAW264.7 were placed in the lower chamber,30 and the medium could be exchanged freely to observe the changes of osteoclasts in the lower chamber (Figure 6A). Overexpression of Sema3A in osteoblasts in the inflammatory microenvironment regulates osteoclast differentiation. (A) Schematic diagram of osteoblast-osteoclast coculture. (B) TRAP staining of osteoclasts in the coculture system after 7 days of RANKL induction. (C) Quantitative analysis of TRAP-positive multinucleated cells in the coculture system. The number of TRAP-positive multinucleated cells was significantly reduced in the LV-Sema3A group compared with the LV0 group and in the LV-Sema3A + LPS group compared with the LV0 + LPS group. (D and E) Quantitative analysis of the expression of osteoclast-specific proteins NFATC1 and CTSK in osteoclasts in the coculture system after 7 days of RANKL induction. (F) Quantitative analysis of the expression of osteoclast-specific genes Nfatc1 vs Ctsk in osteoclasts in the coculture system by qRT-PCR after 7 days of RANKL induction. ns difference was not statistically significant, *P < .05, **P < .01, ***P < .001. After 7 days of RANKL induction, we found that osteoclasts with enlarged size, deep red cytoplasmic staining and increased number of nuclei appeared in all groups, among which the total number of TRAP-positive multinucleated cells was the lowest in the LV-SEMA3A group, and the number of TRAP-positive multinucleated cells was lower in the LV-SEMA3A + LPS group than that in the LV0 + LPS group (Figure 6B, C and Supplementary Figure 5). Meanwhile, the expression of osteoclast-specific genes and proteins Nfatc1 and Ctsk in the LV-SEMA3A group was lower than that in the LV0 group under both normal and LPS conditions (Figure 6D-F). Discussion In this study, we investigated the potential role of Sema3A in peri-implantitis-associated bone homeostasis by integrating exploratory single-cell transcriptomic analysis with an experimental peri-implantitis model and in vitro functional assays. Reanalysis of the PD-derived GSE171213 dataset indicated that osteogenic-lineage cells were sparsely represented and exhibited transcriptional continuity with fibroblast/stromal populations. Accordingly, these populations were conservatively annotated as MSC-like stromal, pre-OB-like, and OB-like cell states rather than as fully differentiated osteogenic cell types. Within this reannotated compartment, periodontal inflammation was associated with a reduced osteogenic-lineage score and modestly decreased SEMA3A expression, while SEMA3A expression showed a statistically significant positive association with the osteogenic transcriptional program. CellChat analysis further suggested extensive predicted communication among stromal/osteogenic and myeloid cell states, particularly involving TGF-β, WNT, TNF, and CSF signalling. The observed reduction in SEMA3A-associated osteogenic activity provided a biological rationale for subsequent validation in the peri-implantitis animal model and in LPS-stimulated osteoblast–osteoclast experimental systems. Sema3A, the most extensively studied dual‑function bone‑protective member of the Semaphorin family, promotes osteogenic differentiation while inhibiting osteoclastogenesis.31,32 Its classic mechanism involves binding to the Nrp1/PlexinA receptor complex, activating β‑catenin signalling in osteoblast lineage cells to upregulate OPG expression, and interfering with immunoreceptor tyrosine-based activation motif‑mediated signalling in osteoclast precursors to block RANKL‑induced osteoclast differentiation. Our findings in the peri‑implantitis model are highly consistent with this: immunofluorescence staining showed that the Sema3A fluorescent signal in the peri‑implant tissues of the inflammatory group was markedly reduced, accompanied by decreased OPG expression and increased RANKL expression; serum levels of the osteogenic marker P1NP and Sema3A were decreased, while the bone resorption marker CTX‑1 was increased. qPCR analysis of gingival tissue further confirmed a significant decrease in Sema3a mRNA expression, downregulation of the osteogenic regulators Runx2 and Col1a1, and upregulation of the osteoclastogenic regulators Nfatc1 and Ctsk. Meanwhile, histomorphological and imaging results also indicated that the local chronic inflammatory pathological changes, discontinuous bone margins, and decreased bone density around the implant were highly consistent in time and space with the changes in Sema3A and bone metabolic markers. Together, these findings suggest that a reduction in Sema3A disrupts the local ‘bone protection barrier’, shifting the osteoblast‑osteoclast coupling towards a pro‑resorptive state. Previous studies have shown that Sema3A protein expression is low in PD and apical PD.33 Sema3A has been shown to prevent bone loss in orthodontic tooth movement and PD to some extent.34,35 However, the significance and application of Sema3A in peri‑implantitis have not yet been investigated. Therefore, we aimed to further explore through in vitro experiments how Sema3A signalling regulates bone homeostasis in peri‑implantitis and the precise role of this signalling in bone homeostasis. In subsequent in vitro experiments, we found that the inflammatory microenvironment modelled by P. gingivalis LPS was able to reduce Sema3A expression in osteoblasts. These findings were consistent with the reduced serum and peri-implant tissue Sema3A levels observed in the experimental peri-implantitis model. Therefore, we conclude that the inflammatory microenvironment of peri-implantitis can affect Sema3A signalling, and this effect may be closely related to peri-implant bone homeostatic imbalance. In parallel, Sema3A overexpression partially reversed the LPS-induced impairment of osteoblast proliferation, migration, and osteogenic differentiation. In the noncontact osteoblast–osteoclast coculture system, RAW264.7 cells cocultured with Sema3A-overexpressing osteoblasts exhibited reduced formation of TRAP-positive multinucleated cells and decreased expression of osteoclast-related markers. These findings suggest that Sema3A overexpression in osteoblasts may indirectly suppress osteoclast differentiation through paracrine signalling. Collectively, Sema3A appears to contribute to the restoration of osteoblast–osteoclast coupling under inflammatory conditions and may represent a potential therapeutic target for re-establishing peri-implant bone homeostasis. Thus, we wanted to find the specific pathway by which Sema3A affects osteoblasts. RNA-seq and KEGG analyses indicated that Sema3A overexpression under inflammatory conditions was closely associated with activation of the PI3K/AKT signalling pathway. Consistently, Sema3A increased AKT phosphorylation and enhanced osteoblast proliferation, migration, and osteogenic differentiation, whereas these effects were significantly attenuated by the PI3K/AKT inhibitor Palomid 529. These findings suggest that the protective effects of Sema3A on osteoblast function in the inflammatory microenvironment are at least partly mediated through PI3K/AKT signalling. Moreover, Sema3A overexpression suppressed osteoclast differentiation in the osteoblast–osteoclast coculture system, as evidenced by reduced TRAP-positive multinucleated cells and decreased expression of Nfatc1 and Ctsk. These results indicate that Sema3A contributes to the maintenance of bone homeostasis by simultaneously promoting osteogenesis and inhibiting osteoclastogenesis under inflammatory conditions. Several limitations should be acknowledged. GSE171213 was derived from periodontal rather than peri-implant tissues, and osteogenic-lineage cells, particularly mature OB-like cells, were sparsely represented; therefore, the single-cell findings should be interpreted as exploratory rather than as direct evidence of peri-implantitis-specific alterations. The in vitro inflammatory model relied on P. gingivalis LPS and murine cell lines, which cannot fully reproduce the polymicrobial, immune, and mechanical complexity of human peri-implantitis. Although osteoclast differentiation was suppressed in the noncontact coculture system, the specific osteoblast-derived mediators responsible for this effect were not identified. Moreover, PI3K/AKT signalling was primarily examined in osteoblasts, while its direct involvement in osteoclast suppression and the contribution of upstream receptors such as NRP1/PlexinA remain to be clarified. The animal sample size was relatively limited, and an in vivo Sema3A therapeutic rescue experiment was not performed. Further studies using peri-implantitis-specific clinical samples, primary human cells, larger animal cohorts, and pathway-specific genetic or rescue approaches are warranted. Conclusion Our research has provided novel insights into the regulatory mechanism of Sema3A in maintaining bone homeostasis in peri-implantitis. By integrating single-cell data analysis with in vivo and in vitro experiments, we have demonstrated that Sema3A enhances the proliferative, migratory, and osteogenic differentiation capacities of osteoblasts, while inhibiting osteoclast differentiation through cell-cell signalling between osteoblasts and osteoclasts in the inflammatory microenvironment. Additionally, our findings suggest that the restoration of osteoblast function was partly associated with PI3K/AKT signalling, whereas the suppression of osteoclastogenesis occurred indirectly through osteoblast-derived paracrine signals. Hence, we suggest that the role of Sema3A in the prevention, diagnosis, and treatment of peri-implantitis can continue to be explored and applied. Ethics statement This research protocol has been approved by the Experimental Animal Ethics Committee of Nanchang University (Ethical Approval Number: NCULAE-20221031158). Author contributions Formal analysis, validation, visualization, writing – original draft: He. Investigation, formal analysis: Dai, Wu, Deng, and Jiang. Validation, visualization: Yang, Li. Data curation: Xu. Conceptualization, funding acquisition, supervision, writing – review and editing: Song. Data availability The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Funding This work was supported by the National Natural Science Foundation of China (No. 8246031044 to L.S.), Key Project of Jiangxi Education Department (No. GJJ180010), Double Thousand Talents Project of Jiangxi Province (No. jxsq2023201045 to S.L.) and High-Level and High-Skilled Leading Talent Training Project of Jiangxi Province (No. G/Y3034 to S.L.). Conflict of interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article. Acknowledgements We sincerely thank all the researchers who generously shared the data used in this study. At the same time, we also appreciate the convenience provided by GEO to the public. Figure 2A was partly generated using Figdraw. Footnotes Supplementary material associated with this article can be found in the online version at doi:10.1016/j.identj.2026.111100. Supplementary materials REFERENCES Associated Data Supplementary Materials Data Availability Statement The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.