Macrophage metabolic reprogramming via HIF-1α–glycolysis drives osteoblast ferroptosis and bone loss through an IL-6–STAT3–dependent redox axis Authors: Yifan Gu, Kun Wang, Yicong Wang, Ziru Wang, Yiheng Li, Lei Li, Shuai Jiang, Yu Zheng, Run Feng, Min Yang Published: 2026 ABSTRACT AbstractBackgroundPostmenopausal osteoporosis (PMOP) is characterized by exacerbated bone resorption and inadequate bone formation, with macrophage-driven inflammation playing a key role. However, how immunometabolic reprogramming of macrophages modulates osteoblast fate remains unknown.MethodsUsing integrated single-cell and bulk transcriptomics, we identified a hypermetabolic macrophage subpopulation in PMOP marrow reliant on HIF-1α-glycolysis. We pharmacologically disrupted this axis with the HDAC inhibitor valproic acid (VPA) and validated its function using the HIF-1α stabilizer DMOG. The paracrine effects on osteoblasts were assessed via conditioned medium, focusing on ferroptosis and differentiation. Therapeutic efficacy was tested in ovariectomized rats.ResultsVPA upregulated HIF1AN, enhancing its binding to HIF-1α and promoting its degradation. This suppressed glycolytic flux and M1 polarization, reducing IL-6 secretion. The altered secretome protected osteoblasts from ferroptosis by inhibiting the IL-6/p-STAT3/HIF-1α/TFRC axis and rebalancing GPX4/ACSL4. Osteogenic differentiation was restored. In OVX rats, VPA improved bone mass and microstructure, effects abolished by DMOG.ConclusionWe unveil a macrophage-centric immunometabolic checkpoint that is linked to osteoblast ferroptosis via IL-6/STAT3 signaling. Targeting this HIF-1α-glycolysis axis, exemplified by VPA, represents a novel therapeutic strategy for PMOP.Keywords: Osteoporosis, immunometabolism, ferroptosis, HIF-1α, glycolysis, macrophage polarization Introduction Osteoporosis (OP) imposes a profound global health burden, driven by progressive microarchitectural deterioration and an elevated risk of fragility fractures [1]. In postmenopausal osteoporosis (PMOP), oestrogen withdrawal disrupts skeletal homoeostasis, causing osteoclast-mediated bone resorption to outpace osteoblast-driven bone formation [2]. While traditionally viewed through an endocrine lens, PMOP pathogenesis is heavily exacerbated by chronic low-grade inflammation [3–7]. Because current frontline therapies—including bisphosphonates and anabolic agents—often suffer from waning long-term efficacy and dose-limiting side effects [8], there is a pressing need to identify and therapeutically target the upstream inflammatory drivers of bone loss. Within the bone marrow niche, highly plastic macrophages serve as critical immune regulators of skeletal turnover [9–11]. Oestrogen depletion skews these cells toward a pro-inflammatory (M1) phenotype, triggering a sustained release of cytokines—such as TNF-α and IL-6—that simultaneously hyperactivate osteoclasts and impair osteoblast function [12]. Attempts to simply neutralise these downstream cytokines often yield limited skeletal benefits due to cytokine redundancy. Instead, modulating the intrinsic metabolic engines that dictate macrophage polarisation offers a more fundamental therapeutic approach [13]. M1 activation is metabolically wired to aerobic glycolysis, a process governed largely by hypoxia-inducible factor-1α (HIF-1α) [14–19]. Since this glycolytic shift drives the subsequent inflammatory secretome, intercepting the HIF-1α–glycolysis axis presents a compelling, upstream immunomodulatory strategy for PMOP [20,21]. On the bone-forming side of the remodelling equation, osteoblast failure is increasingly linked to ferroptosis, an iron-dependent, lipid peroxidation-driven form of regulated cell death [22–25]. Following menopause, systemic iron accumulation coupled with local oxidative stress renders osteoblasts particularly vulnerable to ferroptotic death, directly compromising bone formation [26,27]. This cell death cascade hinges on the balance between anti-ferroptotic enzymes like glutathione eroxidase 4 (GPX4) and pro-ferroptotic drivers such as acyl-CoA synthetase long-chain family member 4 (ACSL4) [28]. Importantly, exposure to pro-inflammatory cytokines like IL-6 can sensitise cells to ferroptosis through STAT3 activation [29,30]. These observations lead to a central hypothesis: the metabolically reprogrammed state of bone marrow macrophages dictates a specific inflammatory secretome that contributes to osteoblast ferroptosis. To test this macrophage-osteoblast functional circuit, we utilised valproic acid (VPA). While classically characterised as a histone deacetylase (HDAC) inhibitor with osteogenic properties [31–33], VPA exerts biological effects well beyond epigenetic regulation. Emerging evidence indicates that VPA can induce profound metabolic reprogramming [34,35], partly by upregulating HIF1AN (Factor Inhibiting HIF-1) to promote HIF-1α degradation. Building on our recent findings that VPA mitigates bone loss by suppressing osteoblastic ferroptosis [36], we reasoned that VPA could serve as an ideal pharmacological probe to disrupt the HIF-1α–glycolysis axis in macrophages. Using an integrative approach encompassing transcriptomics, metabolic profiling, and both in vitro and in vivo models, this study addresses three core aims. First, we delineate the immunometabolic landscape of PMOP macrophages and define how VPA disrupts the HIF-1α–glycolysis axis. Second, we establish whether this metabolic reprogramming alters the macrophage secretome to protect osteoblasts from IL-6/STAT3-mediated ferroptosis. Finally, we validate the therapeutic efficacy of targeting this macrophage-centric checkpoint in an ovariectomized rat model of osteoporosis (Scheme 1). Proposed pathophysiological framework of ovariectomy (OVX)-induced osteoporosis and the protective effects of valproic acid (VPA). (Top) OVX disrupts bone homoeostasis, characterised by sustained osteoclast overactivation and compromised osteoblast survival. Concurrently, the accumulation of M1-polarised macrophages is associated with inflammatory infiltration, contributing to exacerbated bone loss. (Bottom) The M1 macrophage secretome, enriched in pro-inflammatory cytokines (e.g. IL-6), creates a microenvironment that supports osteoclastogenesis and contributes to ferroptosis in osteoblasts. VPA counteracts these pathological shifts by targeting HIF-1α to restrict glycolysis-driven M1 polarisation in macrophages. This attenuation of M1 polarisation correlates with reduced IL-6 secretion and the subsequent downregulation of the p-STAT3/HIF-1α/TFRC signalling axis, ultimately alleviating iron overload-associated oxidative stress. Materials and methods Bioinformatic analysis of single-cell RNA sequencing data The single-cell RNA sequencing dataset GSM4423510 (GSE147287) was retrieved from the Gene Expression Omnibus (GEO) database. Subsequent analysis was performed using the Seurat package (v5.3.0) in R. Cells were filtered based on quality thresholds: those with unique feature counts below 200 or exceeding 5000, or with mitochondrial gene content higher than 25%, were excluded. Normalisation was carried out using the ‘LogNormalize’ method. Dimensionality reduction was conducted via principal component analysis (PCA), followed by uniform manifold approximation and projection (UMAP) for visualisation. Cell clusters were identified with the FindClusters function and annotated according to established marker genes. Macrophage subclusters were extracted for re-clustering, and Gene Ontology (GO) enrichment analysis was conducted using the clusterProfiler package (v4.14.6). Cell Culture and treatment We maintained the mouse monocyte-macrophage cell line RAW264.7 (RRID: CVCL_0493) and the pre-osteoblastic cell line MC3T3-E1 subclone 14 (RRID: CVCL_5437) at 37 °C in a humidified 5% CO₂ atmosphere. Both cell lines were procured from BDBIO (Hangzhou, China). RAW264.7 cells were cultured in high-glucose DMEM (BDBIO, #L100500) supplemented with 10% (v/v) foetal bovine serum (FBS; BDBIO, #F800500) and 1% (v/v) penicillin-streptomycin (BDBIO, #A200100). MC3T3-E1 cells were cultured in complete α-MEM (BDBIO, #L1051500) with identical FBS and antibiotic supplementation. To induce inflammatory polarisation, RAW264.7 cells were exposed to 100 ng/mL lipopolysaccharide (LPS; MedChemExpress, #HY-D1056) for 24 h. Valproic acid (VPA; 1 mM, Macklin Biochemical, #S830714) and the HIF-1α stabiliser DMOG (1 mM; MedChemExpress, #HY-15893) were applied 24 h prior to LPS stimulation. To establish HIF1AN dependency, RAW264.7 cells were transiently transfected with specific small interfering RNA against HIF1AN (si-HIF1AN) or a scrambled negative control (si-NC) using Lipofectamine 3000 (Invitrogen) for 48 h prior to pharmacological treatments. For targeted osteoblast interventions, MC3T3-E1 cells were treated with the ferroptosis inhibitor Ferrostatin-1 (Fer-1; 10 µM, MedChemExpress, #HY-100579), an IL-6 neutralising antibody (1 µg/mL; Invitrogen, #16-7061-81), or the STAT3 specific inhibitor Stattic (5 µM; MedChemExpress, #HY-13818). As a strict negative control to rule out direct endotoxin cytotoxicity, an independent group of MC3T3-E1 cells was treated directly with 100 ng/mL LPS in complete α-MEM. Macrophage metabolic profiling To evaluate the metabolic shift in RAW264.7 macrophages prior to conditioned medium collection, the extracellular acidification rate (ECAR)—a real-time indicator of glycolytic flux—was recorded using a Seahorse XFe96 Analyser equipped with a Glycolysis Stress Test Kit (Agilent Technologies, #103344-100). In parallel, lactate production and glucose consumption within the macrophage culture supernatants were biochemically quantified using specific assay kits (Beyotime, #S0208S and #S0554S, respectively). To ensure rigorous cross-group comparability, all metabolic readout values were strictly normalised to the total cellular protein content extracted from the corresponding culture wells. Preparation and application of macrophage conditioned medium (CM) Conditioned media (CM) were generated from RAW264.7 cultures across different treatment groups. Following 24 h of designated treatments, macrophages were thoroughly washed twice with PBS to remove residual drugs and LPS, and subsequently incubated in serum-free α-MEM for an additional 24 h. The supernatants were collected, centrifuged (1000 × g, 10 min, 4 °C), and passed through 0.22 µm filters. To definitively exclude the confounding possibility that observed osteoblast ferroptosis was a direct consequence of residual endotoxin, an independent functional control group was established (as detailed in Section 2.2), wherein MC3T3-E1 cells were treated directly with an equivalent concentration of LPS (100 ng/mL). For all subsequent functional assays (including viability, fluorescence imaging, and osteogenesis evaluation), the CM was diluted 1:1 (v/v) with complete α-MEM and applied to MC3T3-E1 cells for 24 h unless otherwise specified. Osteoblast viability assessment Following the application of the various CM, the viability of MC3T3-E1 osteoblasts was quantitatively determined using a CCK-8 assay kit (Servicebio, #G1613). Briefly, MC3T3-E1 cells were seeded in 96-well plates, allowed to adhere, and subsequently exposed to the designated CM for 24 h before absorbance measurement. To provide morphological validation of cell survival, a Calcein-AM/PIE Live/Dead Cell Viability Assay Kit (Beyotime, #C2015M) was employed, enabling the visual distinction between live (green fluorescence) and dead (red fluorescence) osteoblasts. Evaluation of intracellular redox homoeostasis, ferroptosis, and ultrastructure Lipid peroxidation, the biochemical hallmark of ferroptosis, was assessed utilising the BODIPY 581/591 C11 fluorescent probe (Servicebio, #G1733). Cells were incubated with the probe (2 µM) at 37 °C for 30 min in the dark. For morphological evaluation, fluorescence images were captured using an inverted microscope (Leica DMi8). To acquire robust quantitative data, stained cells were harvested, washed, and subjected to flow cytometry (Beckman CytoFLEX). The oxidation-induced shift from red to green fluorescence was recorded in the FITC channel, and results were analysed as mean fluorescence intensity (MFI). Intracellular labile iron pools were visualised using the Phen Green SK probe (1 µM; MedChemExpress, #HY-126823), where the quenching of green fluorescence directly correlates with intracellular iron accumulation. Mitochondrial membrane potential (ΔΨm) was evaluated via JC-1 staining (Solarbio, #M8650). The loss of ΔΨm was quantified by calculating the fluorescence ratio of red J-aggregates to green monomers. To provide a gold-standard biochemical measure of the antioxidant defence system, the intracellular ratio of reduced to oxidised glutathione (GSH/GSSG) was determined using a GSH/GSSG Quantification Kit (Beyotime, #S0053) according to the manufacturer’s instructions. For ultrastructural visualisation of mitochondrial damage, cells were fixed in specialised electron microscopy fixative (Servicebio, #G1102), post-fixed in 1.5% (v/v) osmium tetroxide, and embedded in epoxy resin. Ultrathin sections (50 nm) were stained with uranyl acetate and lead citrate. The morphological features of ferroptosis, including mitochondrial shrinkage and cristae reduction, were imaged utilising a Hitachi H-7650 transmission electron microscope (Tokyo, Japan). In vitro osteogenic differentiation MC3T3-E1 cells were seeded in 6-well plates and grown to full confluence. The standard culture medium was then replaced with an osteogenic induction medium (Servicebio, #G4119), which was replenished every 2–3 days. After 7 days of induction, early osteogenic differentiation was evaluated using a BCIP/NBT Alkaline Phosphatase (ALP) Colour Development Kit (Beyotime, #C3206). Following 14 days of induction, extracellular matrix mineralisation was detected utilising an Osteoblast Mineralisation Nodule Staining Kit (Alizarin Red S method; Beyotime, #C0148S). RNA sequencing, RT-qPCR, western blotting, and co-immunoprecipitation For transcriptomic profiling, total RNA was extracted from RAW264.7 macrophages using an RNA Extraction Kit (Servicebio, #G3013). Libraries were prepared using the NEBNext® Ultra™ RNA Library Prep Kit and sequenced on an Illumina NovaSeq 6000 platform (Novogene Co., Ltd.) generating 150 bp paired-end reads. Raw data were processed via fastp, aligned using HISAT2, and quantified with FeatureCounts. Differential expression and Gene Set Enrichment Analysis (GSEA) were executed using the DESeq2 and clusterProfiler packages, respectively. For real-time quantitative PCR (RT-qPCR), cDNA was synthesised (Servicebio, #G3329) and amplified using SYBR Green Premix (Servicebio, #G3328) on a CFX Connect Real-Time PCR System (Bio-Rad). Target gene expression was calculated using the 2−ΔΔCt method against the Gapdh internal control. All primer sequences, alongside their validated NCBI accession numbers confirming target specificity via Primer-BLAST, are comprehensively detailed in Supplementary Table 1. Total protein was extracted using RIPA lysis buffer (Servicebio, #G2002) containing PMSF (Servicebio, #G2008). Equal amounts of protein were separated by SDS-PAGE, transferred to PVDF membranes, and probed with primary antibodies (Supplementary Table 2) overnight at 4 °C. Protein bands were visualised using ECL Prime substrate (Bio-Rad). To verify the physical interaction between HIF1AN and HIF-1α, co-immunoprecipitation (Co-IP) was performed. Cell lysates were incubated overnight with an anti-HIF1AN antibody (Santa Cruz Biotechnology, #sc-271780), captured by Protein A/G Magnetic Beads (Bio-LinkedIn, #IK1204), and subsequently analysed via Western blotting. In vivo animal model, VPA administration, and toxicity evaluation All animal procedures rigorously complied with the ARRIVE 2.0 guidelines and were approved by the Animal Ethics Committee of Wannan Medical University (Approval No.: WNMC-AWE-2024141). Following a one-week acclimatisation in an SPF-grade facility, 24 female Sprague-Dawley rats (8 weeks old; Jiangsu Qinglongshan Biotechnology Co., Ltd.) were randomly assigned to four groups (n = 6 per group) utilising a computer-generated randomisation sequence: Sham, OVX, OVX + VPA, and OVX + VPA + DMOG. Bilateral ovariectomy was performed under general anaesthesia to induce osteoporosis; the Sham group underwent identical surgical exposure without ovary excision. Pharmacological interventions commenced one-week post-surgery. Based on established therapeutic windows, VPA (Macklin Biochemical, #S830714) was administered orally at 100 mg/kg/day, and the HIF-1α stabiliser DMOG (MedChemExpress, #HY-15893) was injected intraperitoneally at 20 mg/kg/day. Treatments were maintained for 8 consecutive weeks. To thoroughly evaluate the in vivo safety profile of chronic VPA administration, body weights were monitored weekly. Upon sacrifice, major metabolic organs (liver and kidney) were harvested and subjected to H&E staining to definitively rule out structural damage or hepatorenal toxicity. During all subsequent micro-CT and histological assessments, investigators remained strictly blinded to the group allocations to minimise observer bias. At the end of the experimental period, all rats were euthanized via anaesthetic overdose followed by cervical dislocation to ensure death, in strict accordance with the approved animal ethical guidelines. Micro-CT, histology, and serum biomarker analysis Rat femurs were fixed in 4% (w/v) paraformaldehyde and scanned utilising an NMC-200 micro-CT system (Pingsheng Medical Technology) at an isotropic resolution of 2 µm. Morphometric analysis within the distal femoral metaphysis yielded parameters including bone mineral density (BMD), bone volume fraction (BV/TV), trabecular number (Tb.N), and trabecular separation (Tb.Sp). Following decalcification in 10% (w/v) EDTA, the femurs were paraffin-embedded and sectioned (4 µm). Sections underwent H&E and TRAP staining (Servicebio, #G1076 and #G1050, respectively) to visualise trabecular architecture and osteoclast activity. For immunohistochemistry (IHC), sections were probed with specific primary antibodies (Supplementary Table 3). To maintain strict spatial consistency during analysis, the average optical density (AOD) of osteoblastic targets (HIF-1α, GPX4, TFRC, 4-HNE) was exclusively quantified within the osteoblasts lining the trabecular bone surfaces, whereas CD86 expression was evaluated specifically within the macrophages scattered throughout the bone marrow cavity. Systemic biochemical and inflammatory statuses were evaluated using collected serum. Commercial assay kits were utilised according to the manufacturers' protocols to measure oxidative stress indicators (SOD, Beyotime #S0101; MDA, #S0131; GSH, #S0053), bone turnover markers (PINP, Elabscience #EEL-R1414; CTX-I, #EEL-R1456; TRAP, Beyotime #P0332), and inflammatory cytokines (TNF-α, Beyotime #PT516; IL-6, #PI328). Statistical analysis All quantitative data are expressed as the mean ± standard deviation (SD). Statistical evaluations were executed using GraphPad Prism software (v9.0). To ensure the appropriateness of parametric testing, data distribution normality was rigorously assessed via the Shapiro-Wilk test, and homogeneity of variance was confirmed using the Brown-Forsythe test. For normally distributed datasets satisfying variance homogeneity, differences between two independent groups were analysed utilising the unpaired two-tailed Student’s t-test. Multi-group comparisons were performed via one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test. In instances where data deviated from a normal distribution, the non-parametric Kruskal-Wallis’s test was applied. A two-sided P-value < 0.05 was considered statistically significant. Results Single-cell transcriptomics uncovers a metabolically hyperactive macrophage subpopulation in human PMOP To delineate the immunometabolic landscape of PMOP, we analysed public single-cell RNA sequencing data derived from human postmenopausal osteoporotic bone marrow. Uniform Manifold Approximation and Projection (UMAP) visualisation identified distinct immune cell populations, including neutrophils, monocytes, and macrophages (Figure 1A, B). Re-clustering of the macrophage lineage revealed four transcriptionally distinct subpopulations (Figure 1C). Gene Ontology (GO) enrichment analysis demonstrated that the predominant C0 subcluster was highly enriched for terms associated with mitochondrial organisation and oxidative phosphorylation, whereas the C1 subcluster aligned with classical immune-regulatory pathways (Figure 1D, E). To provide a more robust quantitative assessment of this metabolic shift, we applied targeted module scoring. The C0 subcluster exhibited significantly elevated Glycolysis Module Scores alongside a robust HIF1A expression signature (Figure 1F, G), directly confirming the existence of a metabolically hyperactive, HIF-1α-driven macrophage subpopulation within the PMOP bone marrow niche. Single-Cell Transcriptomic Analysis Reveals Macrophage Heterogeneity in PMOP. (A) UMAP visualisation illustrating the immune cell composition in PMOP. (B) Quantitative analysis of macrophage proportions. (C) UMAP plot displaying macrophage subclusters (4 subtypes). (D) GO enrichment analysis of subcluster C0 (highlighting metabolic pathways). (E) GO enrichment analysis of subcluster C1 (highlighting immune regulatory pathways). (F) Glycolysis Module Score by Cluster. (G) HIF1A Expression Level by Cluster. Bulk transcriptomics identifies the HIF-1α–glycolysis axis as a central target of VPA Guided by these single-cell insights, we performed bulk RNA sequencing on RAW264.7 macrophages to evaluate the transcriptomic impact of VPA on LPS-induced M1 polarisation. Principal component analysis (PCA) revealed distinct transcriptional clustering among the Control, LPS, and LPS + VPA groups (Figure 2A). VPA pre-treatment successfully reversed the broad pro-inflammatory gene expression signature induced by LPS (Figure 2B). Crucially, KEGG pathway and Gene Set Enrichment Analysis (GSEA) highlighted a marked suppression of the ‘Glycolysis/Gluconeogenesis’ and ‘HIF-1 signalling’ pathways following VPA intervention (Figure 2C-F). Transcriptional evaluation of the HIF pathway constituents revealed that rather than altering Hif1a mRNA levels directly, VPA robustly upregulated Hif1an (encoding FIH-1), a recognised negative regulator that facilitates HIF-1α degradation (Figure 2G). RNA-Seq Confirms the Reversal of Macrophage Metabolic Reprogramming by VPA. (A) Principal component analysis (PCA) of the Control, LPS, and LPS + VPA groups. (B) Hierarchical clustering heatmap of inflammation-related Genes. (C–D) KEGG pathway enrichment analysis, highlighting the reversal of key pathways such as MAPK, HIF-1α, and TNF signalling. (E–F) Gene Set Enrichment Analysis (GSEA) plots for the glycolysis and HIF-1 pathway, demonstrating its reversal by VPA treatment. (G) Heatmap showing RNA-seq results of HIF1A and HIF1AN expression, revealing VPA's targeting of the HIF-1α pathway. VPA suppresses macrophage glycolysis and M1 polarisation via HIF1AN-mediated HIF-1α degradation We next sought to functionally validate this metabolic target. In LPS-stimulated macrophages, VPA profoundly suppressed the extracellular acidification rate (ECAR), a dynamic readout of glycolytic flux (Figure 3A). This metabolic blunting corresponded with reduced lactate production, diminished glucose consumption, and the targeted downregulation of key glycolytic effectors including Glut1, Hk2, and Ldha (Figure 3B-D). Co-immunoprecipitation assays confirmed that VPA strongly enhanced the physical interaction between endogenous HIF1AN and HIF-1α (Figure 3H, I), driving the subsequent proteasomal degradation of the HIF-1α protein (Figure 3E-G). VPA suppresses macrophage glycolysis by promoting HIF-1α degradation. (A) Real-time extracellular acidification rate (ECAR) measured following sequential injections of glucose, oligomycin, and 2-deoxyglucose (2-DG). (B, C) Lactate production and glucose uptake in different treatment groups. (D) mRNA expression levels of key glycolysis-related genes. (E-G) Representative Western blot images of HIF-1α and HIF1AN, with corresponding quantitative densitometric analysis. (H, I) Co-IP assay and quantitative results of the endogenous interaction between HIF1AN and HIF-1α. Data are presented as mean ± SD (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 versus the Model group. To rigorously establish HIF1AN as the indispensable mediator of this metabolic reprogramming, we performed targeted siRNA knockdown. Transfection with si-HIF1AN completely abolished VPA’s capacity to degrade HIF-1α. Concurrently, the loss of HIF1AN reversed VPA's suppressive effects on lactate production and IL-6 secretion, rendering the cells refractory to the drug's immunometabolic benefits (Supplementary Fig. S1). Consistent with the reliance of M1 polarisation on active glycolysis, VPA downregulated the expression of classic M1 markers (iNOS, TNF-α, IL-6), reduced the CD86-positive cell population, and blunted inflammatory cytokine secretion (Figure 4A-I). These anti-inflammatory effects were entirely negated by the HIF-1α stabiliser DMOG, further supporting the critical role of the HIF-1α pathway in mediating VPA's immunometabolic benefits, although the broad action of DMOG as a pan-hydroxylase inhibitor suggests that parallel cooperative mechanisms cannot be completely ruled out. VPA Suppresses M1 Polarisation in an HIF-1α-Dependent Manner. (A-D) Representative western blot images of classic M1 polarisation markers and their quantitative analysis. (E, F) Representative flow cytometry plots and quantification of CD86-positive cells. (G) Immunofluorescence staining images of CD86. (H, I) Concentrations of TNF-α and IL-6 in the conditioned medium from each group. Data are presented as the mean ± SD (n = 3 per group). Statistical significance compared to the Model group: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. The reprogrammed macrophage secretome shields osteoblasts from ferroptosis via the IL-6/STAT3 signalling cascade We then investigated how this macrophage metabolic shift dictates osteoblast fate via paracrine signalling. Conditioned medium derived from VPA-treated macrophages (VPA-CM) rescued MC3T3-E1 osteoblast viability to a degree comparable with the specific ferroptosis inhibitor Fer-1 (Figure 5A). Crucially, direct exposure of MC3T3-E1 cells to 100 ng/mL LPS failed to induce comparable cytotoxicity (Figure 5B), definitively proving that the observed osteoblast death was driven by the pathological macrophage secretome rather than residual endotoxin. VPA-reprogrammed macrophage secretome rescues osteoblasts from ferroptosis via the IL-6/STAT3 axis. (A) CCK-8 assay assessing the viability of MC3T3-E1 cells treated with the indicated conditioned media or targeted inhibitors. (B) Viability of MC3T3-E1 cells treated with 100 ng/mL LPS for 24 h, to rule out direct endotoxin cytotoxicity. (C, D) Representative merged fluorescence images of JC-1 staining and corresponding quantitative analysis of the red/green fluorescence intensity ratio for evaluating mitochondrial membrane potential.(E) Representative merged fluorescence images of intracellular lipid peroxidation detected using the BODIPY 581/591 C11 probe.(F) Representative fluorescence images of the intracellular labile iron pool detected using the Phen Green SK probe.(G, H) Flow cytometry analysis of oxidised BODIPY 581/591 C11 (FITC channel), with representative stacked histograms and corresponding quantification of mean fluorescence intensity. (I) Intracellular GSH/GSSG ratio measured via biochemical assay. All quantitative data are presented as mean ± SD (n = 3 independent biological replicates). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 versus the LPS-CM group. Flow cytometric quantification of oxidised BODIPY C11 alongside fluorescence imaging demonstrated that VPA-CM profoundly suppressed LPS-CM-induced lipid peroxidation (Figure 5E, G, H) and intracellular labile iron accumulation (Figure 5F). VPA-CM also preserved mitochondrial membrane potential (Figure 5C, D) and restored the intracellular GSH/GSSG ratio (Figure 5I), effectively halting the ferroptotic cascade. Ultrastructural analysis further confirmed the protection of mitochondrial integrity against typical ferroptotic shrinkage and cristae loss (Figure 6A). Ultrastructural and mechanistic validation of the IL-6/STAT3-driven ferroptosis axis in osteoblasts. (A) Representative transmission electron microscopy (TEM) images of MC3T3-E1 osteoblasts exposed to the indicated treatments. Red arrows indicate typical ferroptotic mitochondria, characterised by reduced volume, increased membrane electron density, and disrupted cristae. (B, C) Representative Western blot images and corresponding quantitative densitometric analysis of p-STAT3 and total STAT3, demonstrating that IL-6 neutralising antibody (anti-IL-6) blocks upstream STAT3 activation. (D–H) Representative Western blot images and corresponding quantifications of ferroptosis-related proteins (TFRC, HIF-1α, GPX4, and ACSL4), revealing that the STAT3 inhibitor Stattic effectively reverses the molecular signature of ferroptosis. All quantitative data are presented as mean ± SD (n = 3 independent biological replicates). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 versus the LPS-CM group. To delineate the precise molecular executioners linking the altered secretome to osteoblast ferroptosis, we targeted the IL-6/STAT3 pathway. Western blot analysis revealed that LPS-CM triggered robust STAT3 phosphorylation in osteoblasts, driving the subsequent accumulation of HIF-1α and the iron importer TFRC, while severely skewing the GPX4/ACSL4 balance toward lipid peroxidation (Figure 6B-H). Elegantly, either neutralising IL-6 directly in the conditioned medium (Anti-IL-6) or selectively inhibiting intracellular STAT3 activation (Stattic) completely phenocopied the molecular and phenotypic rescue observed with VPA-CM. Ultimately, this protection from ferroptosis translated into the successful restoration of osteoblast differentiation capacity, evidenced by rescued ALP activity, matrix mineralisation, and osteogenic gene transcription (Figure 7A, B). VPA-Induced Alterations in the Macrophage Secretome Rescue the Compromised Osteogenic Capacity under Inflammatory Conditions. (A) Osteogenic differentiation of MC3T3-E1 cells treated with CM from the indicated groups. (B) RT-qPCR analysis of key transcripts involved in osteogenesis. All data are shown as mean ± SD (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 versus the LPS-CM group. In vivo targeting of the macrophage HIF-1α checkpoint mitigates PMOP by suppressing local osteoblastic ferroptosis To evaluate the translational validity of this immunometabolic circuit, we utilised an ovariectomized (OVX) rat model. Systemic administration of VPA substantially mitigated OVX-induced trabecular bone loss, preserving bone mineral density (BMD) and microarchitectural integrity while curbing excessive osteoclast activity (Figure 8A-D). The concurrent administration of DMOG entirely abolished these skeletal benefits, proving the in vivo reliance on HIF-1α destabilization. VPA treatment ameliorates bone loss in ovariectomized (OVX) rats. (A) Representative three-dimensional micro-CT reconstruction images of the distal femurs. (B) Quantitative analysis of bone morphometric parameters, including Bone Mineral Density (BMD) and Trabecular Number (Tb.N). (C) Representative H&E-stained sections of the distal femur, revealing the trabecular bone architecture. (D) TRAP-stained sections of the distal femur, showing mature osteoclasts (red). Black arrows indicate typical TRAP-positive multinucleated osteoclasts lining the trabecular bone surfaces. Data are presented as mean ± SD (n = 6 per group). Statistical significance versus the Model (OVX) group: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Crucially, spatially resolved immunohistochemistry (IHC) confirmed the precise cellular targets of this systemic rescue. While VPA successfully diminished the population of CD86-positive M1 macrophages scattered within the bone marrow cavity, it simultaneously reversed the pathological ferroptotic signatures—specifically, the upregulation of HIF-1α, TFRC, and the lipid peroxidation marker 4-HNE, alongside the depletion of GPX4—exclusively within the osteoblasts lining the trabecular bone surfaces (Figure 9A-F). This targeted spatial validation was corroborated by systemic biochemical improvements, including reduced serum markers of oxidative stress (MDA, SOD, GSH) and systemic inflammation (TNF-α, IL-6) (Figure 9G-N). Collectively, these in vivo data support the notion that pharmacological disruption of macrophage glycolysis is associated with the rescue of osteoblasts from iron-dependent cell death in the osteoporotic microenvironment. In vivo validation of VPA-mediated suppression of local osteoblastic ferroptosis and rescue of systemic oxidative stress and inflammatory dysfunction. (A) Representative immunohistochemistry (IHC) staining panels for CD86, HIF-1α, TFRC, GPX4, and 4-HNE in the distal femurs of the indicated groups. Red arrows highlight the typical positively stained cells: macrophages scattered within the bone marrow cavity for CD86, and osteoblasts lining the trabecular bone surfaces for HIF-1α, TFRC, GPX4, and 4-HNE. (B–F) Corresponding quantitative analysis of the average optical density for IHC staining of CD86, HIF-1α, TFRC, GPX4, and 4-HNE.(G–N) Systemic biochemical and ELISA analyses of serum biomarkers, including oxidative stress indicators, bone turnover markers and inflammatory cytokines. All quantitative data are presented as mean ± SD (n = 6 biologically independent mice per group). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 versus the OVX group. Discussion The present study identifies the HIF-1α–glycolysis axis in bone marrow macrophages as a pivotal immunometabolic checkpoint in the pathogenesis of PMOP. By integrating transcriptomic profiling with targeted functional assays, we provide compelling pharmacological evidence supporting a novel pathophysiological framework: macrophage metabolic reprogramming modulates a pro-inflammatory secretome, which in turn contributes to osteoblast dysfunction via ferroptosis. While our in vivo findings utilising systemic VPA administration provide robust translational support for this model, they primarily serve as a pharmacological proof-of-concept, generating a strong hypothesis for a macrophage-to-osteoblast paracrine circuit that operates within the osteoporotic bone marrow niche. A major mechanistic insight from our in vitro data is the delineation of VPA’s precise impact on macrophage immunometabolism. Historically characterised as a broad-spectrum HDAC inhibitor [31,37–39], VPA exerts highly pleiotropic effects across multiple cell lineages. Our data reveal that within inflammatory macrophages, VPA specifically upregulates HIF1AN to accelerate HIF-1α proteasomal degradation, corroborating HIF1AN's canonical role as a negative regulator that facilitates HIF-1α breakdown via asparaginyl hydroxylation [38]. The strict reliance on this specific node was confirmed by our in vitro siRNA knockdown experiments, where the loss of HIF1AN completely abrogated VPA’s suppressive effects on glycolytic flux and IL-6 secretion. This robust functional coupling further reinforces the established paradigm that the HIF-1α-glycolysis pathway operates as an indispensable driver of M1 macrophage polarisation [20]. Nevertheless, considering the systemic administration route in our animal model, the observed skeletal benefits in vivo likely reflect a composite therapeutic effect. VPA’s classical epigenetic activities undoubtedly act in concert with this newly identified metabolic reprogramming to comprehensively restore bone homoeostasis [33–35]. Beyond macrophage intrinsic metabolism, we bridge the fields of osteoimmunology and redox biology by mapping a defined paracrine execution axis. Our neutralisation and inhibitor studies establish IL-6/STAT3 signalling as a principal driver linking the macrophage secretome to osteoblastic iron overload and lipid peroxidation, consistent with the canonical role of IL-6 as a potent upstream instigator of STAT3 phosphorylation to dictate cell fate across diverse pathological microenvironments [40]. It is important to acknowledge, however, that the osteoporotic marrow constitutes a highly complex inflammatory milieu. While IL-6 clearly plays a dominant role in our isolated in vitro system, classical M1-derived mediators such as TNF-α and IL-1β likely act cooperatively in vivo. These additional cytokines may synergistically amplify intracellular STAT3 activation or progressively deplete the osteoblast glutathione pool, thereby collectively lowering the cellular threshold for ferroptotic death [41–46]. From a translational perspective, identifying osteoblast ferroptosis as a downstream consequence of macrophage dysfunction offers renewed therapeutic opportunities. Our data indicate that pharmacologically rewiring macrophage metabolism can achieve an osteoprotective effect analogous to direct ferroptosis inhibition. When contextualising these findings for clinical PMOP, the choice of animal model warrants consideration. Although the young ovariectomized rat faithfully recapitulates oestrogen-withdrawal bone loss, it inherently lacks the chronological aging component characteristic of the human disease. Normal aging independently drives a state of chronic, low-grade basal inflammation—often termed ‘inflammaging’—which severely exacerbates macrophage metabolic dysfunction. Concurrently, senescent osteoblasts exhibit a progressive functional decline in endogenous antioxidant defence systems, rendering them inherently more susceptible to lipid peroxidation [22,26,47–49]. Consequently, the HIF-1α/glycolysis/ferroptosis axis identified here may exert an even more profound pathological impact within the elderly human skeleton. This study possesses several acknowledged limitations. First, while our systemic VPA intervention effectively halted bone loss, definitively proving a macrophage-exclusive mechanism requires future investigations employing myeloid-specific Hif1a conditional knockout murine models. Second, our analytical focus remained primarily on the suppression of M1 inflammatory activation; the potential capacity of VPA to simultaneously promote pro-resolutive M2 macrophage polarisation—a phenotype highly beneficial for bone repair—warrants comprehensive future exploration [50,51]. Lastly, the specific HDAC isoform responsible for mediating the transcriptional upregulation of HIF1AN remains to be distinctly elucidated. Conclusion In summary, this study provides compelling pharmacological evidence supporting a novel immunometabolic framework for PMOP. We demonstrate that the HIF-1α–glycolysis axis serves as a critical metabolic engine underlying macrophage inflammatory polarisation. Pharmacological disruption of this axis via VPA, mediated through HIF1AN upregulation, profoundly alters the macrophage secretome. In vitro, this metabolic rewiring effectively shields osteoblasts from IL-6/STAT3-driven ferroptosis, a finding robustly supported by systemic skeletal preservation in an OVX animal model. While future genetic models are necessary to isolate cell-autonomous functions, our findings strongly suggest that targeting macrophage immunometabolism represents a highly promising therapeutic avenue for addressing the inflammatory and redox dysfunctions inherent to osteoporosis. Supplementary Material Author Checklist 260326.pdf Author Checklist 260326.pdf Supplementary Tables 260326.docx Supplementary Tables 260326.docx Supplementary Figures 260326.docx Supplementary Figures 260326.docx Acknowledgements We would like to express our gratitude to the Central Laboratory of Yijishan Hospital and the Anhui Provincial Key Laboratory for Non-coding RNA Translational Research on Major Diseases for providing the necessary instruments and facilities for this study. Funding Statement This research was supported by Talent Introduction Project of Yijishan Hospital, Wannan Medical University (Grant No. KY29240723), Anhui Province Higher Education Science Research Project (Grant Nos. 2025AHGXZK31385, 2023AH040265), Anhui Clinical Medical Research Transformation Project (Grant No. 202304295107020007), Talented Scholars of Wannan Medical University (Grant No. YR202226), Technology Mountaineering Programme of Yijishan Hospital, Wannan Medical University (Grant No. PF2019005), and Research project of Anhui Provincial Health Commission (Grant No. AHWJ2023A10149). Disclosure statement No potential conflict of interest was reported by the author(s). Data availability statement Data will be made available on request. Ethics approval statement The animal protocols were approved by the Experimental Animal Welfare and Ethics Committee of Wannan Medical University, Wuhu city (No. WNMC-AWE-2024141) and Anhui Province Higher Education Science Research Project (Grant No. 2025AHGXZK31385). Consent for publication Not applicable. Clinical trial number Not applicable. Competing interests 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 paper. Supplemental material Supplemental data for this article can be accessed at https://doi.org/10.1080/13510002.2026.2667673. References Associated Data Supplementary Materials Author Checklist 260326.pdf Author Checklist 260326.pdf Supplementary Tables 260326.docx Supplementary Tables 260326.docx Supplementary Figures 260326.docx Supplementary Figures 260326.docx Data Availability Statement Data will be made available on request.