High glucose induces osteoblasts ferroptosis via the TXNIP/Trx-1/GPX4 pathway in type 2 diabetic osteoporosis

Abstract

Type 2 diabetic osteoporosis (T2DOP) is characterized by impaired bone formation and increased fracture susceptibility; however, the molecular mechanisms linking hyperglycemia to osteoblast dysfunction remain insufficiently defined. In this study, we demonstrate that high glucose (HG) triggers ferroptosis in osteoblasts through activation of the thioredoxin-interacting protein (TXNIP)/thioredoxin-1 (Trx-1)/glutathione peroxidase 4 (GPX4) pathway, leading to reduced cell viability, bioenergetic failure, and impaired osteogenic differentiation. HG exposure induced hallmark ferroptotic features—including excessive intracellular ferrous iron (Fe²⁺), elevated reactive oxygen species (ROS), lipid peroxidation, increased 4-hydroxynonenal (4-HNE), and acyl-CoA synthetase long-chain family member 4 (ACSL4), together with depletion of glutathione (GSH), GPX4, and ferritin heavy chain 1 (FTH1). Transcriptomic profiling identified TXNIP as a key upstream regulator under hyperglycemic conditions. Mechanistically, TXNIP directly interacts with and inhibits Trx-1, suppresses thioredoxin reductase-1 (TrxR1) activity, disrupts redox homeostasis, and ultimately inactivates GPX4, thereby promoting ferroptosis and osteogenic dysfunction. Silencing TXNIP restored antioxidant capacity, mitochondrial respiration, and mineralization in vitro. In a T2DOP rat model, systemic TXNIP knockdown significantly reduced oxidative injury, improved trabecular architecture and bone mineral density, and increased biochemical markers of bone formation. These findings identify TXNIP-dependent ferroptosis as a central mechanism of osteoblast impairment in diabetes and highlight the TXNIP/Trx-1/GPX4 axis as a promising therapeutic target for type 2 diabetic osteoporosis.

Introduction

Type 2 diabetes mellitus (T2DM) is a chronic metabolic disorder characterized by persistent hyperglycemia, insulin resistance, and relative or absolute insulin deficiency [1]. China has the largest T2DM population worldwide, and the prevalence continues to rise with aging demographics and lifestyle changes [2]. Nearly 70% of patients develop complications—including cardiovascular disease, nephropathy, retinopathy, neuropathy, and osteoporosis—reflecting the systemic impact of long-term metabolic imbalance [3]. Osteoporosis, although common, is often overlooked, and is defined by decreased bone mass, disrupted microarchitecture, and an increased risk of fragility fractures [4]. In type 2 diabetic osteoporosis (T2DOP), factors such as chronic glucose–insulin dysregulation, accumulation of advanced glycation end-products (AGEs), oxidative stress, and diabetes-related microvascular damage collectively impair bone quality and contribute to skeletal fragility [5]. As T2DOP substantially reduces quality of life and poses a growing public health and economic burden [6], its rising incidence highlights the urgent need for targeted preventive and therapeutic strategies, which are still lacking [7].

Ferroptosis is an iron-dependent form of regulated cell death, mechanistically distinct from apoptosis and autophagy, and is characterized by mitochondrial shrinkage, impaired membrane integrity, and excessive accumulation of lipid reactive oxygen species (ROS) [8–10]. Growing evidence indicates that ferroptosis contributes to the pathogenesis of T2DOP. For instance, the accumulation of AGEs under chronic hyperglycemia can trigger ferroptosis in osteoblasts, an effect that the iron chelator deferoxamine effectively reverses. [11]. In T2DOP rat models, overexpression of activating transcription factor 3 (ATF3) aggravates bone loss by suppressing glutathione peroxidase 4 (GPX4)—a central regulator of the ferroptotic defense system—while ATF3 inhibition promotes osteogenic activity and partially restores bone metabolism [12]. Furthermore, melatonin has been shown to attenuate high glucose-induced ferroptosis through activation of the nuclear factor erythroid 2-related factor 2 (Nrf2)/heme oxygenase-1 (HO-1) antioxidant pathway, highlighting the importance of redox homeostasis in osteoblast survival [13]. Collectively, these findings suggest that ferroptosis is not merely a downstream consequence of diabetic metabolic stress but represents a mechanistically meaningful contributor to diabetic bone deterioration and a promising therapeutic target for T2DOP.

Oxidative stress, resulting from excessive ROS production, leads to protein modification, lipid peroxidation, and cellular dysfunction, playing a central role in various diseases [14–16]. The thioredoxin system, particularly Thioredoxin-1 (Trx-1), is vital in maintaining redox homeostasis by eliminating ROS and protecting cells from oxidative injury [17–20]. Thioredoxin-interacting protein (TXNIP), identified through yeast two-hybrid screening, binds to and inhibits Trx-1, thereby impairing its antioxidant function and promoting oxidative stress, inflammation, and cell death [21–25]. Elevated TXNIP expression has been implicated in multiple diabetic complications, as well as in the regulation of ferroptosis. For example, studies have shown that the TXNIP/Trx-1/GPX4 axis mediates ferroptosis in hippocampal neurons after hypoxic-ischemic injury [26, 27]. In our previous research, we found that high glucose significantly upregulates TXNIP in osteoblasts, while suppressing Trx-1 and GPX4, suggesting activation of ferroptosis pathways. Transcriptomic analysis revealed enrichment of ferroptosis-related genes in the high glucose group. Based on these findings, we hypothesize that high glucose induces osteoblast ferroptosis through the TXNIP/Trx-1/GPX4 pathway, leading to impaired bone formation and progression of T2DOP. This study aims to elucidate the molecular mechanism of this pathway and provide insights into potential therapeutic targets for type 2 diabetic osteoporosis.

Materials and methods

Cell culture and treatments

The human osteoblast cell line hFOB1.19 was acquired from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China, RRID: CVCL_3700) and cultured under conditions recommended by the supplier: DMEM/F-12 medium (Gibco, Grand Island, NY, USA) supplemented with 10% FBS (Biological Industries, Kibbutz Beit HaEmek, Israel), maintained at 34 °C with 5% CO₂. After stabilization and expansion, cells were transferred to standard conditions (37 °C, 5% CO₂) and exposed to normal glucose (NG, 17.5 mM) or high glucose (HG, 35 mM) for 48 h. To assess ferroptosis, Ferrostatin-1 (Fer-1, Sigma-Aldrich, RRID: SCR_016623) (5 µM), a specific ferroptosis inhibitor, was added to the cultures for 12 h.

Cell transcriptome sequencing and analysis

PolyA-tailed mRNA was enriched using Oligo (dT) beads, and total RNA was fragmented to ~ 300 bp by ion disruption. First-strand cDNA was synthesized using reverse transcriptase and random hexamer primers, followed by second-strand synthesis using the first strand as a template. The resulting cDNA library was amplified by PCR, and fragments around 450 bp were selected. Library quality was assessed with an Agilent 2100 Bioanalyzer to determine concentration and integrity. Indexed libraries were pooled, diluted to 2 nM, and denatured to obtain single-stranded templates. Sequencing was performed using paired-end (PE) mode on an Illumina platform. Raw data were processed with the platform’s software, and downstream analyses—including heatmaps, volcano plots, and KEGG enrichment—were conducted using R for data interpretation and visualization.

Measurement of intracellular iron ions (Fe.²⁺)

Intracellular iron ions (Fe²⁺) levels were measured using the Cell Ferrous Iron Colorimetric Assay Kit (E-BC-K881-M, Elabscience, Wuhan, China) according to the manufacturer’s instructions. After cell lysis and centrifugation, the supernatant was mixed with the chromogenic reagent, and absorbance was measured at 593 nm using a microplate reader. Fe²⁺ concentrations were calculated from a standard curve generated with known iron standards. All measurements followed the kit protocol to ensure accuracy and reproducibility.

TXNIP knockdown via short hairpin RNA (shRNA) transfection

hFOB1.19 cells were infected with lentiviral particles containing TXNIP-shRNA (Sense strand: 5′-CAUCCUUCGAGUUGAAUAUTT-3′; Antisense strand: 5′-AUAUUCAACUCGAAGGAUGTT-3′), supplied by GeneChem Corporation in Shanghai, China. Prior to infection, 1× 10⁶ cells were plated in 6-well dish 24 h earlier. Next day, cells were transfused using 40 µL transfer agent from Beyotime Biotechnology, also located in Shanghai, along with strict lentiviral pellets at a multiplicity of infection (MOI) of 20. 72 h post-transfection, the efficiency of the transfection was assessed by measuring the fluorescent intensity microscopically. Subsequently, stable transfectants were selected by exposing them to 2 µg/mL of puromycin, sourced from Solarbio in Beijing, China. The effectiveness of the knockdown was further verified through Western blot assay. Cells were treated with 1 μmol/L 1-methylpropyl 2-imidazolyl disulfide (PX-12, RRID: CHEBI:132052) for 12 h before protein extraction to assess its effects on GPX4 regulation under TXNIP knockdown.

RNA extraction and qRT-PCR

Total RNA was extracted and quantified using a NanoDrop spectrophotometer (Thermo Fisher, USA). cDNA was synthesized using the PrimeScript RT Reagent Kit (TaKaRa, Japan), and qRT-PCR was performed with SYBR Premix Ex Taq II (TaKaRa) on an Applied Biosystems 7500 system. Relative expression was calculated using the 2^{^−ΔΔCt} method with 18S rRNA as the internal control. PCR conditions were 95 °C for 30 s, followed by 45 cycles of 95 °C for 15 s and 58 °C for 34 s. Primer sequences were as follows: TXNIP, forward 5′-CTTAGTGTAACCAGCGGCGT-3′, reverse 5′-CTGAGGAAGCTCAAAGCCGA-3′; Trx-1, forward 5′-CTTGGACGCTGCAGGTGATA-3′, reverse 5′-TCCTGACAGTCATCCACATCT-3′.

Cell counting Kit-8 (CCK8) assay

Cells were cultured at 5 × 10³ per well in 96-well plate, with CCK-8 provided by Beyotime Biotech to assess their viability. Following incorporating 10 μL CCK-8 reagent per well, these plates were incubated at 37 °C for a duration of 4 h. Optical density (OD) at 450 nm was then determined with a microwell plate reader manufactured by Bio-Tek Instruments in Winooski, Vermont, USA.

Lactate dehydrogenase (LDH) cytotoxicity assay

Cytotoxicity in hFOB1.19 cell was evaluated with the LDH Cytotoxicity Assay Kit from Beyond Biotechnology. Following cell lysis, 60 µL of LDH working solutions was spiked into this mixture, then cultivated over 30 min at 25 °C in darkness. OD at 490 nm was measured utilizing a Bio-Tek Instruments microplate reader (Winooski, Vermont, USA).

EdU proliferation assay

hFOB1.19 cells in the logarithmic phase were seeded into 6-well plates and incubated overnight at 37 °C. Cells were then treated with 10 µM EdU for 2 h, washed with PBS, and fixed in 4% formaldehyde for 30 min. After permeabilization with 0.5% Triton X-100 for 10 min, EdU staining was performed in the dark for 30 min, followed by PBS washes. Nuclei were counterstained with Hoechst 33342 for 30 min in the dark. Fluorescence images were captured using a fluorescence microscope.

Alizarin Red S (ARS) staining

hFOB1.19 cells were induced in osteogenic medium for 14 days, with medium refreshed every 2 days. At the end of induction, cells were washed with PBS and fixed in 4% paraformaldehyde for 15 min. After fixation, they were rinsed with distilled water and stained with 40 mM Alizarin Red S (Sigma-Aldrich, Germany) solution (pH 4.2) for 5 min at room temperature. Excess dye was removed by thorough washing with distilled water until the background was clear. Images of mineralized nodules were captured under a phase-contrast microscope, and mineralization was semi-quantitatively analyzed using ImageJ software.

Agilent Seahorse XF cell mito stress test assay

Mitochondrial respiration of hFOB1.19 cells was assessed using the Agilent Seahorse XFe96 Analyzer (Agilent Technologies, USA). Cells were seeded into XF 96-well microplates at defined density. Prior to the assay, culture medium was replaced with CO₂-free, pH 7.4 assay medium pre-warmed to 37 °C. After instrument calibration, oligomycin, FCCP, and rotenone/antimycin A were sequentially injected. OCR was recorded and mitochondrial parameters including basal respiration, maximal respiration, and ATP production were calculated using the XF Cell Mito Stress Test Report Generator.

Malondialdehyde (MDA) and 4-hydroxynonenal (4-HNE) measurement

MDA and 4-HNE are commonly used as indicators of lipid peroxidation and oxidative stress, respectively. To measure lipid peroxidation in hFOB1.19 cells, MDA assay kit (Beyotime Biotechnology) and 4-HNE assay kit (Cambridge, MA, USA) were utilized following manufacturer’s guidelines. Subsequently, these OD values were recorded via microplate readers produced by Bio-Tek Instrument in Winooski, VT, USA, at wavelengths of 532 and 450 nm.

Lipid peroxidation fluorescence probe detection

Lipid peroxidation levels were assessed using the C11 BODIPY 581/591 fluorescence probe. After incubating these cells at 37 °C under dark environment for 30 min, fluorescence image was obtained on a LSCM manufactured by Olympus in Tokyo, Japan. Dual wave length excitations at 488 nm and 565 nm, combined with detection through 530/30 and 590/30 transmission filters, respectively, were employed during the imaging process.

Intracellular ROS measurement

ROS generation was measured with a DCFH-DA fluorescence probe (Sigma-Aldrich, Germany). Cells were treated with DCFH-DA working fluid at 37 °C with 45 min. Subsequently, fluorometric intensity was recorded with an LSCM, manufactured by Olympus in Tokyo.

Glutathione (GSH) level detection

hFOB1.19 cells were cultured into 6-well plates at 2 × 10⁵ density per well. A protein stripping reagent was incorporated into culture medium, followed by two freeze–thaw cycles of cells utilizing liquefied nitrogen and 37 °C water baths. Following a 5-min incubation at 4 °C, cells were centrifuged at 10,000 × g over 10 min, with ultrastructure collected for total GSH measurements. To evaluate the level of oxidized glutathione (GSSG), a GSH removal auxiliary solution and operating solution were incorporated into supernatant. Absorption was subsequently tested at 412 nm utilizing microtiter plates reader.

Western blot analysis

Samples were lysed in RIPA buffer (Beyotime Biotechnology) and centrifuged at 12,000 × g for 30 min at 4 °C to collect the protein-containing supernatant. Protein concentration was determined and normalized using the BCA assay. Equal amounts of protein (30 µg) were separated by 10% SDS-PAGE and transferred onto PVDF membranes (GE Healthcare, Madison, USA) at 110 mA for 60 min. After blocking, membranes were incubated with specific primary and secondary antibodies. Protein bands were visualized using the EC3 Imaging System (UVP, Upland, CA, USA) and quantified by ImageJ (NIH, Bethesda, MD, USA). Primary antibodies included TXNIP (ab188865, Abcam, RRID:AB_2923036), Trx-1 (A4024, ABclonal, RRID:AB_2863174), GPX4 (ab125066, Abcam, RRID:AB_10973901), Osteoprotegerin (OPG) (ab8448, Abcam, RRID:AB_306566), Osteocalcin (OCN) (ab93876, Abcam, RRID:AB_10675660), Thioredoxin reductase 1 (TrxR1) (ab124954, Abcam, RRID:AB_10975643), c-Jun N-terminal kinase (JNK) (ab179461, Abcam, RRID:AB_2744672), Phosphorylated c-Jun N-terminal kinase (p-JNK) (ab4821, Abcam, RRID:AB_2141012), p38 mitogen-activated protein kinase (p38) (ab170099, Abcam, RRID:AB_3083680), Phosphorylated p38 mitogen-activated protein kinase (p-p38) (abab195049, Abcam, RRID:AB_2576214), Glucose transporter 1 (GLUT1) (ab115730, Abcam, RRID:AB_10903230), Glucose transporter 4 (GLUT4) (ab33780, Abcam, RRID:AB_2191441), Acyl-CoA synthetase long-chain family member 4 (ACSL4) (ab155282, Abcam, RRID:AB_2714020), Ferritin heavy chain 1 (FTH1) (ab65080, Abcam, RRID:AB_10564857), and β-actin (ab115777, Abcam, RRID:AB_10899528), and secondary antibodies were from Cell Signaling Technology (Danvers, USA, RRID:AB_2099233).

Co-immunoprecipitation (Co-IP) assay

Co-immunoprecipitation was performed to detect protein–protein binding in hFOB1.19 cells. After treatment, cells were washed with cold PBS and lysed on ice using NP-40 lysis buffer (Beyotime, P0013F) containing protease and phosphatase inhibitors (Beyotime, P1045). Lysates were cleared by centrifugation (14,000 × g, 15 min, 4 °C), precleared with Protein A/G agarose beads (Thermo Fisher, 20421) for 1 h, and incubated overnight at 4 °C with primary antibody. Normal rabbit IgG (CST, 2729, RRID:AB_1031062) was used as the negative control. Protein A/G beads were added for 3 h to pull down immune complexes, which were washed five times, eluted by boiling in 2 × SDS loading buffer, and analyzed by Western blot with ECL detection (Thermo Fisher, 32106).

TrxR1 activity assay

TrxR1 activity in hFOB1.19 cells was measured using a colorimetric Thioredoxin Reductase Assay Kit (Beyotime, S0052). After treatment, cells were washed with cold PBS and lysed in the kit lysis buffer on ice. Lysates were centrifuged at 12,000 × g for 10 min at 4 °C, and protein concentration was determined by BCA assay (Beyotime, P0012). Equal protein amounts were mixed with the TrxR reaction buffer, DTNB, and NADPH working solutions, with inhibitor-treated wells used as background controls. Absorbance at 412 nm was recorded at 0 and 10 min at 37 °C, and TrxR1 activity was calculated from the change in absorbance after subtracting the inhibitor control.

Experimental animals

Forty 8-week-old male pathogen-free SD rats were acquired from the Experimental Animals Department at Dalian Medical University. Among them, twenty of them were designated as controls, whereas the other twenty were utilized for establishing T2DOP model. Rats were fed a high-fat diet for 8 weeks and then received a single intraperitoneal injection of streptozotocin (STZ, Sigma-Aldrich S0130, RRID: SCR_015773, 30 mg/kg). Three days after STZ administration, fasting blood glucose (FBG) and insulin sensitivity index (ISI) were evaluated. Consistent with widely accepted criteria, rats with FBG > 7.8 mmol/L accompanied by a reduced ISI were considered successfully diabetic. Only rats meeting these criteria progressed to the extended high-fat feeding phase to induce T2DOP. After 16 weeks, TXNIP-siRNA (Sense strand: 5′-GAGACCUGGAAACAAAUAUTT-3′; Antisense strand: 5′-AUAUUUGUUUCCAGGUCUCTT-3′, 200 µL, JinTuoSi Biotechnology Co., Ltd, Wuhan, China) was administered in T2DOP rats by tail vein injection every 3 days for two consecutive weeks (five injections in total) to create a TXNIP interference model. Following an additional 20 weeks of feeding, the rats were euthanized for further analysis. The experimental timeline is shown in Fig. 6A. Control rats were maintained on a standard diet and administered 200 µL NC-siRNA 16 weeks later.

Fig. 6.

TXNIP silencing attenuates ferroptosis and bone loss in T2DOP rats. Forty male SD rats were randomly allocated into four groups (Ctrl, Ctrl + si-TXNIP, T2DOP, and T2DOP + si-TXNIP; n = 10/group). A A schematic illustrates the T2DOP induction protocol, including high-fat diet feeding, STZ administration, and tail vein delivery of TXNIP-siRNA. B TXNIP protein expression in femoral tissue was assessed by Western blot. C, D T2DOP rats exhibited elevated serum iron and MDA levels, both of which were reduced by TXNIP silencing. E Serum GSH levels were significantly decreased in T2DOP rats and partially restored following si-TXNIP treatment. F, G Serum ALP and P1NP levels were measured as indicators of osteogenic activity. H, I Micro-CT analysis revealed severe deterioration of femoral trabecular microarchitecture in T2DOP rats, with improvements in BMD, Tb.N, Tb.Th, and BV/TV following TXNIP knockdown. J Western blot analysis showed downregulation of Trx-1 and GPX4 in T2DOP rats and restoration of both proteins after TXNIP silencing. K Immunohistochemical staining of TXNIP, Trx-1, and GPX4 in femoral sections (scale bar = 100 μm), and L corresponding quantitative analyses further confirmed elevated TXNIP expression and reduced Trx-1/GPX4 levels in T2DOP rats, with these changes reversed by si-TXNIP. Data are presented as mean ± SD. **P < 0.01, ***P < 0.001, ****P < 0.0001

Microcomputed tomography (micro-CT) assessment

Using micro-CT, rat femur microstructure was evaluated, as outlined in our previous study [28]. The images were reconstructed and analyzed for these parameters: bone mineral density (BMD), trabecular number (Tb.N), trabecular thickness (Tb.Th), and trabecular bone volume fraction (BV/TV).

Immunohistochemistry (IHC)

Bone tissue was fixed, decalcified, embedded, and sectioned. The slides were then de-paraffinized with xylene and rehydrated via an array of ethanol solutions. After antigen retrieval, sections were stopped with 10% goose serum. Primary antibodies—anti-TXNIP (1:200; ab188865, Abcam, RRID:AB_2923036), anti-Trx-1 (1:200; A4024, ABclonal, RRID:AB_2863174), and anti-GPX4 (1:200; ab125066, Abcam, RRID:AB_10973901)—were appended and incubated at 4 °C overnight. On second day, slices were handled for 2 h using secondary antibody, then stained over 5 min at 0.1% DAPI. Following rinsing with PBS, pictures were captured together utilizing microscope, and semiquantitatively analyzed by ImageJ software.

Western blot analysis of bone tissue proteins

TXNIP, GPX4, and Trx-1 expression in rat femurs was analyzed by Western blot. After euthanasia, femurs were collected and cleaned, and the distal metaphysis was snap-frozen in liquid nitrogen and ground into powder. Bone powder was lysed in RIPA buffer (Beyotime, P0013B) with protease and phosphatase inhibitors (Beyotime, P1045), followed by centrifugation at 14,000 × g for 15 min at 4 °C. Protein concentration was measured using a BCA kit (Beyotime, P0012). Equal amounts of protein were separated by SDS-PAGE, transferred to PVDF membranes (Millipore, IPVH00010), blocked with 5% BSA, and incubated overnight at 4 °C with primary antibodies against TXNIP (ab188865, Abcam, RRID:AB_2923036), Trx-1 (A4024, ABclonal, RRID:AB_2863174), GPX4 (ab125066, Abcam, RRID:AB_10973901), and β-actin (ab115777, Abcam, RRID:AB_10899528). After incubation with HRP-conjugated secondary antibodies, signals were developed using ECL substrate (Thermo Fisher, 32106) and quantified with ImageJ.

Determination of serum iron ions, MDA, and GSH levels

Serum rats were sampled for measuring the concentration of ferric ions, MDA, and GSH. Following the manufacturer’s guidelines (Nanjing Jian-cheng Bioengineering Institute, China), OD values, respectively, were determined at 520, 532, and 412 nm.

Statistical analyses

Statistics were performed with GraphPad Prism 9.0 software (GraphPad Software, Inc., La Jolla, California, USA). All trials were done in triple copies, with figures presented as mean ± standard error. Variations among two groups were calculated utilizing Student's t-test, while one-way ANOVA was applied for comparisons among multiple groups. P-values < 0.05 were interpreted as remarkable.

Results

High glucose induces ferroptosis in osteoblasts

hFOB1.19 cells were exposed to normal glucose (NG, 17.5 mM) or high glucose (HG, 35 mM) to determine the early cellular response to high glucose. HG stimulation induced a progressive decline in cell viability, with a significant reduction evident at 48 h (Fig. 1A). Intracellular ferrous iron levels were markedly elevated under HG conditions, whereas treatment with the ferroptosis inhibitor Fer-1 substantially diminished this increase (Fig. 1B). HG also triggered pronounced lipid peroxidation, as reflected by increased MDA and 4-HNE levels, both of which were effectively mitigated by Fer-1 (Fig. 1C, D). Consistent with these biochemical findings, C11-BODIPY 581/591 staining revealed robust lipid ROS accumulation in HG-treated osteoblasts, which was significantly reduced following Fer-1 administration (Fig. 1F). In parallel, HG exposure markedly enhanced general intracellular ROS production, as detected by DCFH-DA fluorescence, and this elevation was attenuated by Fer-1 (Fig. 1G). HG also disrupted cellular redox homeostasis, evidenced by a substantial decline in intracellular GSH levels, while Fer-1 partially restored GSH content (Fig. 1E). At the protein level, HG upregulated ACSL4—a key lipid-metabolism enzyme promoting ferroptotic sensitivity—and concurrently downregulated GPX4 and FTH1, indicating impaired antioxidant defense and reduced iron-storage capacity. These alterations were largely reversed by Fer-1 (Fig. 1H–J). Collectively, these results demonstrate that high glucose induces a ferroptotic phenotype in osteoblasts, characterized by iron overload, excessive lipid and general ROS accumulation, GSH depletion, and dysregulation of ACSL4/GPX4/FTH1, whereas Fer-1 effectively counteracts these effects.

Fig. 1.

High glucose induces ferroptosis in hFOB1.19 osteoblasts. hFOB1.19 cells were cultured under normal glucose (NG) or high glucose (HG) for 48 h, and 5 μM Fer-1 was added to the HG group during the last 12 h. A Cell viability was evaluated at the indicated time points using the CCK-8 assay. B Intracellular Fe²⁺ levels were quantified using an iron assay kit. C, D Lipid peroxidation was assessed by measuring MDA and 4-HNE. E Intracellular GSH levels were determined to evaluate antioxidant capacity. F Lipid ROS accumulation was visualized using the C11-BODIPY 581/591 fluorescence probe (scale bar = 20 μm). G Intracellular ROS generation was measured using the DCFH-DA probe (scale bar = 20 μm). H–J Western blot analysis was performed to detect the expression of ferroptosis-related proteins, including ACSL4, GPX4, and FTH1. All data represent mean ± SD from three independent experiments. **P < 0.01, ***P < 0.001, ****P < 0.0001

High glucose-induced ferroptosis inhibits osteoblast activity and mitochondrial function

hFOB1.19 cells were cultured under NG or HG conditions for 48 h. Fer-1 (5 μM) was added during the final 12 h of HG exposure to assess its protective effects. HG treatment significantly reduced cell viability, as evidenced by decreased CCK-8 activity (Fig. 2A) and increased LDH release (Fig. 2B), indicating enhanced cellular damage. Western blot analysis showed downregulation of OCN and OPG, which was partially reversed by Fer-1 (Fig. 2C). EdU staining revealed impaired proliferation under HG, also alleviated by Fer-1 (Fig. 2D). Alizarin Red S staining demonstrated reduced mineralization in HG-treated cells, which improved with Fer-1 (Fig. 2E). Mitochondrial function, assessed by Seahorse XF analysis, showed that HG markedly suppressed basal respiration, maximal respiration, and ATP production, all of which were restored by Fer-1 (Fig. 2F–I). These findings indicate that high glucose impairs osteoblast viability, differentiation, and mitochondrial function via ferroptosis, while Fer-1 confers protective effects.

Fig. 2.

High glucose suppresses osteoblast activity, osteogenic differentiation, and mitochondrial function. hFOB1.19 cells were cultured in NG or HG for 48 h. After 36 h of HG exposure, Fer-1 (5 μM) was added for an additional 12 h to establish the HG + Fer-1 group. After treatment, A Cell activity was estimated using the CCK-8 assay, and B LDH activity was assessed. C Western blot analysis of OCN and OPG expression in osteoblasts from each experimental group. D Proliferation was evaluated with the EdU test. Scale bar = 200 μm. E Extracellular matrix mineralization was observed using Alizarin Red S staining to assess osteogenic fractionation after two weeks. Scale bar = 200 μm. F Seahorse XF analysis of mitochondrial function in osteoblasts under high glucose conditions. G–I Measurement of osteoblast mitochondrial ATP production, basal respiration, and maximal respiratory capacity. All data are stated as mean ± SD of three individual tests. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

TXNIP induction under high glucose inhibits the Trx-1/TrxR1 system and drives ferroptotic signaling in osteoblasts

We first performed transcriptomic profiling of NG- and HG-treated hFOB1.19 cells to investigate the molecular mechanisms by which high glucose induces ferroptosis in osteoblasts. Hierarchical clustering revealed a distinct transcriptional signature in the HG group (Fig. 3A), and the volcano plot identified TXNIP as one of the most significantly upregulated genes under hyperglycemic conditions (Fig. 3B). KEGG pathway enrichment further demonstrated that ferroptosis-related pathways were among the most prominently activated pathways in response to HG (Fig. 3C). Consistent with the sequencing data, both TXNIP mRNA and protein levels were markedly elevated following HG exposure (Fig. 3D, E), whereas Trx-1 expression was significantly reduced at both the transcript and protein levels (Fig. 3F, G). Co-immunoprecipitation analysis confirmed an enhanced interaction between TXNIP and Trx-1 under HG conditions, suggesting functional suppression of Trx-1 by TXNIP (Fig. 3H).

Fig. 3.

High glucose upregulates TXNIP and promotes ferroptosis by suppressing Trx-1–mediated antioxidant defense in osteoblasts. A Heatmap illustrating global transcriptional alterations in hFOB1.19 cells following NG or HG treatment. B Volcano plot identifying TXNIP as one of the most significantly upregulated genes under HG exposure. C KEGG pathway enrichment analysis revealing marked activation of ferroptosis-associated signaling pathways in the HG group. D, E qRT-PCR and Western blot analyses demonstrating a robust increase in TXNIP mRNA and protein expression in HG-treated cells. F, G HG exposure significantly decreased Trx-1 expression at both the mRNA and protein levels. H Co-immunoprecipitation confirming enhanced TXNIP–Trx-1 interaction under HG conditions. I Western blot verifying efficient TXNIP silencing in hFOB1.19 cells. J TXNIP knockdown restored Trx-1 and GPX4 expression in HG-treated osteoblasts. K Western blot showing that TXNIP silencing reversed the HG-induced suppression of TrxR1 protein. L Measurement of TrxR1 activity demonstrating reduced enzymatic activity under HG conditions and partial recovery upon TXNIP knockdown. M Western blot analysis indicating that HG downregulated GLUT1 and GLUT4, whereas TXNIP knockdown significantly restored their expression. N Western blot analysis of MAPK signaling showing HG-induced phosphorylation of p38 and JNK, with TXNIP knockdown attenuating these effects. O Treatment with the Trx-1 inhibitor PX-12 abrogated the GPX4 upregulation induced by TXNIP silencing, supporting the requirement of Trx-1 activity in TXNIP-mediated ferroptosis regulation. Data are presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

To validate the functional consequences of TXNIP induction, TXNIP expression was silenced using shRNA, as confirmed by Western blotting (Fig. 3I). TXNIP knockdown restored Trx-1 protein expression and significantly increased GPX4 levels under HG conditions (Fig. 3J). Similarly, the decrease in TrxR1 induced by HG was reversed following TXNIP silencing (Fig. 3K). In agreement with these findings, HG markedly reduced TrxR1 enzymatic activity, whereas TXNIP knockdown partially rescued TrxR1 function (Fig. 3L), indicating restoration of the thioredoxin antioxidant system. Moreover, HG downregulated GLUT1 and GLUT4, while TXNIP knockdown significantly restored both glucose transporters (Fig. 3M), suggesting a link between TXNIP signaling and glucose metabolic impairment.

We next examined whether Mitogen-activated protein kinase (MAPK) signaling participates in TXNIP-mediated responses. HG stimulation increased phosphorylation of p38 and JNK, and TXNIP silencing blunted these activations (Fig. 3N), indicating that TXNIP may amplify HG-induced oxidative stress through MAPK pathway activation. Finally, treatment with the Trx-1 inhibitor PX-12 abolished the GPX4 upregulation induced by TXNIP knockdown (Fig. 3O), confirming that the protective effects of TXNIP silencing depend on Trx-1 activity. Collectively, these findings demonstrate that high glucose upregulates TXNIP, enhances TXNIP–Trx-1 interaction, suppresses the thioredoxin antioxidant axis, and consequently drives ferroptosis in osteoblasts.

Knockdown of TXNIP expression significantly attenuates ferroptosis in osteoblasts under high glucose conditions

To further determine whether TXNIP mediates high glucose–induced ferroptosis in osteoblasts, we examined ferroptosis-related biochemical and molecular markers following TXNIP knockdown. High glucose markedly increased intracellular Fe²⁺ accumulation, whereas TXNIP silencing significantly reduced iron overload (Fig. 4A). Consistently, lipid peroxidation—as indicated by elevated MDA and 4-HNE levels—was strongly induced by high glucose but substantially attenuated in the presence of TXNIP-shRNA (Fig. 4B, C). High glucose also caused a pronounced decline in intracellular GSH content, while TXNIP knockdown partially restored GSH levels, suggesting improved redox buffering capacity (Fig. 4D). C11-BODIPY 581/591 fluorescence imaging further confirmed excessive lipid ROS generation under hyperglycemic conditions, which was markedly suppressed by TXNIP silencing (Fig. 4E). Similarly, general intracellular ROS levels detected using DCFH-DA were elevated by high glucose and significantly reduced upon TXNIP knockdown (Fig. 4F).

Fig. 4.

TXNIP knockdown alleviates high glucose-induced ferroptosis in osteoblasts. To determine the protective effects of TXNIP silencing against ferroptotic injury, hFOB1.19 cells were treated with NG, NG + TXNIP-shRNA, HG, or HG + TXNIP-shRNA. A Intracellular Fe²⁺ levels were markedly elevated under HG conditions and significantly reduced by TXNIP knockdown. B, C Lipid peroxidation, assessed by MDA and 4-HNE, was increased in HG-treated cells and attenuated following TXNIP silencing. D High glucose markedly decreased intracellular GSH, whereas TXNIP knockdown partially restored GSH levels. E C11-BODIPY 581/591 fluorescence imaging revealed robust lipid ROS accumulation under HG, which was substantially reduced upon TXNIP knockdown (scale bar = 20 μm). F General intracellular ROS levels, detected using the DCFH-DA probe, were similarly elevated by HG and mitigated by TXNIP silencing (scale bar = 20 μm). G Western blot analysis showed that HG-induced upregulation of ACSL4 was reversed following TXNIP knockdown. H GPX4 expression, markedly suppressed by HG, was restored by TXNIP silencing. I FTH1 was reduced under HG conditions but increased following TXNIP knockdown, indicating improved intracellular iron buffering. Data are presented as mean ± SD from three independent experiments. **P < 0.01, ***P < 0.001, ****P < 0.0001

At the protein level, high glucose increased ACSL4 expression—an enzyme that sensitizes cells to ferroptosis—while TXNIP knockdown reversed this effect (Fig. 4G). Conversely, GPX4, the key lipid peroxidase responsible for suppressing ferroptosis, was downregulated under high glucose but restored following TXNIP silencing (Fig. 4H). Expression of FTH1, a critical iron-storage protein, was similarly reduced under hyperglycemic conditions and recovered upon TXNIP knockdown, indicating improved intracellular iron sequestration (Fig. 4I). Together, these findings demonstrate that TXNIP knockdown effectively mitigates high glucose–induced ferroptosis in osteoblasts by reducing iron overload, suppressing lipid and general ROS accumulation, restoring antioxidant capacity, and normalizing ferroptosis-related protein expression.

TXNIP knockdown restores osteoblast function under high glucose

hFOB1.19 cells were transfected with TXNIP-shRNA and cultured in NG or HG for 48 h. TXNIP knockdown significantly improved cell viability (CCK-8 assay) and reduced LDH release compared to HG alone, indicating decreased cytotoxicity (Fig. 5A, B). Western blotting showed increased expression of osteogenic markers OCN and OPG in the HG + TXNIP-shRNA group (Fig. 5C). EdU incorporation assays demonstrated enhanced proliferation, and Alizarin Red S staining after two weeks of osteogenic induction revealed greater mineralized nodule formation upon TXNIP knockdown (Fig. 5D–E). Mitochondrial function assessed by Seahorse XF analysis showed that TXNIP knockdown restored ATP production, basal respiration, and maximal respiratory capacity under HG conditions (Fig. 5F–I). These results indicate that TXNIP knockdown mitigates HG-induced osteoblast dysfunction by promoting viability, proliferation, osteogenic differentiation, and mitochondrial bioenergetics.

Fig. 5.

TXNIP knockdown alleviates HG-induced osteoblast dysfunction. A Cell vitality was evaluated with CCK-8 assay. B Cytotoxicity was measured through the LDH release test. C Western blot analysis of OCN and OPG expression in osteoblasts from each experimental group. D Proliferation was evaluated with the EdU test. Scale bar = 200 μm. E Extracellular matrix mineralization was observed using Alizarin Red S staining to assess osteogenic fractionation after two weeks. Scale bar = 200 μm. F Seahorse XF analysis of mitochondrial function in osteoblasts under high glucose conditions. G–I Measurement of osteoblast mitochondrial ATP production, basal respiration, and maximal respiratory capacity. All data are stated as mean ± SD of three individual tests. **P < 0.01, ***P < 0.001, ****P < 0.0001

TXNIP downregulation protects against bone loss and ferroptosis in T2DOP mouse model

We established a T2DOP rat model and treated the animals with TXNIP-siRNA, aiming to clarify the in vivo function and therapeutic relevance of TXNIP in diabetic osteoporosis. As shown in Fig. 6A, rats receiving a high-fat diet combined with low-dose STZ developed typical metabolic abnormalities and bone deterioration characteristic of T2DOP. Western blot analysis confirmed a marked upregulation of TXNIP in femoral tissue from T2DOP rats, whereas TXNIP-siRNA effectively suppressed its expression (Fig. 6B). Consistent with enhanced ferroptotic activity, T2DOP rats exhibited significantly elevated serum iron and MDA levels, together with a marked reduction in GSH content; all three biochemical abnormalities were substantially ameliorated following TXNIP silencing (Fig. 6C–E).

We next assessed bone metabolism markers and skeletal microarchitecture. Serum Alkaline Phosphatase (ALP) and Procollagen Type I N-Terminal Propeptide (P1NP) levels, both key indicators of osteogenic activity, were significantly reduced in T2DOP rats and partially restored by TXNIP knockdown (Fig. 6F, G). Micro-CT analysis revealed profound trabecular bone loss in the distal femur of T2DOP rats, reflected by decreased BMD, Tb.N, Tb.Th, and BV/TV; notably, these structural deficits were markedly improved after TXNIP-siRNA treatment (Fig. 6H, I). At the molecular level, T2DOP rats displayed substantial downregulation of Trx-1 and GPX4—two major components of the antioxidant and anti-ferroptotic defense systems—while TXNIP silencing restored their expression (Fig. 6J). Immunohistochemical staining further corroborated these findings, showing increased TXNIP and diminished Trx-1/GPX4 in T2DOP bone tissue, with the opposite pattern observed after knockdown (Fig. 6K, L). Collectively, these results demonstrate that TXNIP contributes to ferroptosis-associated bone loss in T2DOP, and its inhibition effectively restores antioxidant capacity, mitigates ferroptotic damage, and improves bone microarchitecture in vivo.

Discussion

In this study, we demonstrate that high glucose induces ferroptosis in osteoblasts through activation of the TXNIP/Trx-1/GPX4 pathway, leading to impaired cellular viability, mitochondrial dysfunction, and suppressed osteogenic differentiation. Ferroptosis has recently gained attention as a key pathogenic mechanism in diabetic osteoporosis and has been implicated in osteoblast impairment in both clinical samples and experimental models [12, 13]. Whereas most previous studies have emphasized oxidative stress and apoptosis as major contributors to diabetic bone loss, our findings provide direct evidence that ferroptosis constitutes a distinct and previously underappreciated mode of osteoblast death under chronic hyperglycemia. Additionally, the observed downregulation of GLUT1 and GLUT4 under high glucose conditions suggests impaired glucose transport, which may exacerbate metabolic stress and diminish antioxidant capacity during ferroptotic progression. This metabolic vulnerability may partly explain why diabetic bone exhibits reduced material quality despite preserved or even elevated bone mineral density.

Our findings further extend the established role of TXNIP as a negative regulator of the thioredoxin antioxidant system [14–16] and identify it as a critical upstream inducer of ferroptotic injury in osteoblasts. TXNIP overexpression inhibited Trx-1 activity, reduced TrxR1 enzymatic function, and disrupted redox homeostasis, ultimately leading to GPX4 depletion—one of the core biochemical hallmarks of ferroptosis [17–20]. Conversely, TXNIP silencing restored TrxR1 activity and GPX4 expression, whereas the Trx-1 inhibitor PX-12 abolished this protective effect, demonstrating that Trx-1 activity is essential for maintaining GPX4-dependent ferroptosis resistance. We also observed activation of MAPK signaling under high glucose conditions, a pathway known to participate in oxidative stress responses and ferroptosis modulation. Although MAPK activation may not serve as the primary trigger in this context, it may amplify TXNIP-mediated redox imbalance and increase ferroptotic susceptibility. Together, these findings validate the TXNIP/Trx-1/GPX4 axis as a mechanistically coherent ferroptotic cascade and highlight the convergence of metabolic stress, disrupted redox defense, and ferroptotic injury in osteoblasts [21–25]. These results are consistent with accumulating evidence linking TXNIP upregulation to ferroptosis and metabolic dysfunction across multiple pathological conditions [23, 24, 35–40].

Beyond mechanistic insight, this study also carries translational relevance. TXNIP has emerged as a promising therapeutic target, and several TXNIP modulators—including verapamil and naturally occurring compounds such as resveratrol—have shown benefits in diabetes and related disorders by suppressing TXNIP expression or restoring metabolic balance [26]. Given that ferroptosis integrates oxidative stress, iron dysregulation, and glucose metabolism, strategies aimed at preserving the TXNIP/Trx-1/GPX4 axis may provide dual metabolic and cytoprotective benefits in diabetic osteoporosis. Moreover, several small-molecule TXNIP inhibitors currently under development have demonstrated efficacy in attenuating oxidative damage and ferroptosis in diverse tissues [23, 24, 27–34]. These findings suggest that therapeutic repurposing, combination therapy, or bone-targeted delivery systems may represent feasible approaches for clinical translation. Future research optimizing metabolic modulation (e.g., GLUT1/GLUT4 restoration), or targeting upstream regulators such as MAPK, may further enhance the efficacy of TXNIP-directed interventions.

This study has several limitations. The contribution of high glucose cannot be fully excluded due to the lack of an osmotic control. The in vivo findings are limited to male rats, leaving potential sex-specific differences unexplored. While TXNIP knockdown restored TrxR1 and GPX4, mechanistic validation through Trx-1 overexpression is needed to solidify the TXNIP/Trx-1 hierarchy. Furthermore, the long-term efficacy and safety of TXNIP silencing require evaluation. Finally, the interplay between ferroptosis and other cell death pathways in diabetic bone disease remains to be elucidated.

In conclusion, this work identifies TXNIP-dependent ferroptosis as a key pathogenic mechanism linking hyperglycemia to osteoblast dysfunction and establishes the TXNIP/Trx-1/GPX4 axis as a mechanistically rational therapeutic target for type 2 diabetic osteoporosis. These findings not only enhance current understanding of diabetic bone fragility but also provide a foundation for developing ferroptosis-based translational strategies.

Author contributions

Y. Z. and D.X. performed the experiments and wrote the main manuscript text; M. Z. and Y.G. prepared Figs. 1–6; A.H. and C.R. conducted statistical analysis on the experimental data; Y.D. designed the research project and drafted the manuscript.All authors reviewed the manuscript.

Funding

This study was funded by the Dalian Medical Science Research Program Project (2211004).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with national and institutional guidelines and was approved by the Animal Experimental Ethical Committee of Central Hospital of Dalian University of Technology (Ethical review approval number: YN2023-147-03).

Competing interests

The authors declare no competing interests.