Targeting ferroptosis suppresses osteocyte glucolipotoxicity and alleviates diabetic osteoporosis

Abstract

Diabetic osteoporosis (DOP) is the leading complication continuously threatening the bone health of patients with diabetes. A key pathogenic factor in DOP is loss of osteocyte viability. However, the mechanism of osteocyte death remains unclear. Here, we identified ferroptosis, which is iron-dependent programmed cell death, as a critical mechanism of osteocyte death in murine models of DOP. The diabetic microenvironment significantly enhanced osteocyte ferroptosis in vitro, as shown by the substantial lipid peroxidation, iron overload, and aberrant activation of the ferroptosis pathway. RNA sequencing showed that heme oxygenase-1 (HO-1) expression was notably upregulated in ferroptotic osteocytes. Further findings revealed that HO-1 was essential for osteocyte ferroptosis in DOP and that its promoter activity was controlled by the interaction between the upstream NRF2 and c-JUN transcription factors. Targeting ferroptosis or HO-1 efficiently rescued osteocyte death in DOP by disrupting the vicious cycle between lipid peroxidation and HO-1 activation, eventually ameliorating trabecular deterioration. Our study provides insight into DOP pathogenesis, and our results provide a mechanism-based strategy for clinical DOP treatment.

Fig. 1

The diabetic microenvironment induced osteocyte ferroptosis in vivo. a A schematic illustration of the DOP model established by HFD feeding with injection of low-dose STZ. b Gross images of CTL and STZ&HFD mice. c Representative micro-CT (scale bar: 300 μm) and X-ray (scale bar: 2 μm) images showing the microarchitecture of the distal femur. Tibial paraffin sections stained with H&E (scale bar: 50 μm) and subjected to TUNEL (scale bar: 50 μm) showed empty lacunae and dead osteocytes. Quantitative analysis of the BV/TV ratio (d) trabecular separation (Tb.Sp, e), trabecular number (Tb. N, f), trabecular thickness (Tb. Th, g), number of empty lacunae with respect to bone area (N. Empt. Lc./B. Ar., h), fraction of empty lacunae (Frac. Empt. Lc., i) and number of TUNEL-positive cells (j). k Tibial paraffin sections stained for GPX4 and PTGS2 (scale bar: 50 μm). l The expression level of GPX4 in tibial bone tissue was determined by WB analysis. m Semiquantitative analysis of GPX4 expression based on WB analysis. Concentration of MDA in serum (n) and bone tissue (o). Serum concentrations of OCN (p), FFAs (q), and PA (r). *P < 0.05; **P < 0.01. Each group contained six mice

Fig. 2

HGHF-induced osteocyte ferroptosis in vitro. a Osteocytes were treated with BSA (5.5 mmol·L⁻¹ glucose) or various concentrations of PA (25.5 mmol·L⁻¹ glucose) for different times and evaluated with a CCK‐8 assay. b CCK-8 assay of osteocytes pretreated with DMSO (solvent), Z-VAD-FMK (an apoptosis inhibitor), VitE (a ROS scavenger), Fer-1 (a ferroptosis inhibitor), Nec-1 (a necroptosis inhibitor) or 3-MA (an autophagy inhibitor) and then subjected to HGHF treatment for 24 h. c CCK-8 assay of osteocytes pretreated with various concentrations of Fer-1 and then subjected to BSA or HGHF treatment for 24 h. d Osteocytes treated with BSA, HGHF or HGHF + Fer-1 for 24 h. Representative images of the TUNEL assay (scale bar: 100 μm), C11-BODIPY staining (scale bar: 25 μm), FerroOrange staining (scale bar: 100 μm), and TEM (scale bar: 0.5 μm) images are presented. e Semiquantitative analysis of TUNEL-positive cells. Semiquantitative analysis of the fluorescence intensity of lipid peroxides (f) and ferrous iron (g). h WB analysis of the expression of GPX4 and ACSL4 in osteocytes. i Total iron, ferrous iron, and ferric iron levels in osteocytes were quantitatively determined using an iron assay kit. j MDA levels in osteocytes were quantitatively determined using an MDA assay kit. “*” indicates comparison between the two indicated groups, and “#” indicates comparison with the BSA group at 24 h. *P < 0.05; “**” and ^##P < 0.01. All data are from n = 3 independent experiments

Fig. 3

HGHF treatment activated ferroptosis and promoted HO-1 expression in osteocytes. a Schematic illustration of the experimental design and procedure for sample preparation for RNA sequencing. b Heatmap showing differentially expressed genes between the BSA and HGHF groups of osteocytes. Red: high expression levels. Blue: low expression levels. c GSEA enrichment plot for the ferroptosis pathway (normalized enrichment score = 2.11; FDR q value < 0.001) based on RNA sequencing. d KEGG pathway enrichment analysis of the differentially expressed genes between the BSA and HGHF groups. e KEGG database visualization of ferroptosis pathway-associated genes affected by HGHF treatment in osteocytes. Red: high expression levels. Green: low expression levels. f The mRNA levels of several ferroptosis-related genes were measured using RT–qPCR in BSA-treated and HGHF-treated osteocytes. The β-actin gene was used as the internal reference gene. ^#P < 0.01. All data are from n = 3 independent experiments

Fig. 4

HO-1 was essential for HGHF-induced ferroptosis. a Representative images of HO-1-stained bone sections from CTL mice and STZ&HFD mice (scale bar: 50 μm). b Quantification of HO-1 expression in vivo. c Western blot showing HO-1 expression under various treatments. d Quantification of HO-1 expression based on WB analysis. e RT–qPCR showing HO-1 expression under various treatments. The β-actin gene was used as the internal reference gene. C11-BODIPY (scale bar: 100 μm) and FerroOrange (scale bar: 100 μm) staining of ZnPP-treated osteocytes (f) and HO-1-overexpressing osteocytes (g). h, i Quantification of fluorescence intensity. j Osteocyte viability was assessed by a CCK-8 assay. k The MDA concentration was assessed using an MDA assay kit. **P < 0.01. “NS” indicates nonsignificant. All data are from n = 3 independent experiments

Fig. 5

HGHF enhanced the NRF2–c-JUN interaction. a Effects of HGHF on the levels of MAPK signaling pathway proteins, including p-MKK4, p‐p38/p38, p‐JNK/JNK, p‐ERK/ERK, and c-JUN. b Representative images of bone sections stained with an anti-c-JUN antibody (scale bar: 50 μm). c Effects of HGHF on nuclear import of NRF2. d Effects of HGHF on NRF2 upstream signaling pathway proteins, including LC3, P62 and KEAP1. β-Actin was used as the internal control for cytoplasmic proteins, and Lamin B1 was used as the internal control for nuclear proteins. e Representative images of bone sections stained with an anti-NRF2 antibody (scale bar: 50 μm). f Co-IP results for NRF2 and c-JUN in osteocytes transfected with the NRF2-HA and c-JUN-Myc plasmids. g Co-IP results for endogenous NRF2 and c-JUN in osteocytes. h IF showing that NRF2 colocalized with c-JUN in the nucleus in osteocytes under HGHF treatment (scale bar: 50 μm). i 3D binding structure of NRF2 and c-JUN determined via molecular modeling and docking studies. j Co-IP results for ΔNRF2 and Δc-JUN in osteocytes transfected with plasmids expressing the NRF2-HA and c-JUN-Myc truncations. All data are from n = 3 independent experiments

Fig. 6

HGHF-induced HO-1 expression was dependent on the NRF2–c-JUN interaction. HGHF-induced HO-1 expression was dependent on the NRF2–c-JUN interaction. a WB results for osteocytes transfected with the indicated siRNA or scrambled siRNA. b Semiquantitative analysis of WB results. c Hmox1 mRNA expression under the indicated treatments was determined using RT–qPCR. d Hmox1 promoter activity under treatment with various concentrations of HGHF was assessed using a dual-luciferase reporter assay. e Hmox1 promoter activity under the indicated treatments was assessed using a dual-luciferase reporter assay. f Schematic representation of the mouse Hmox1 promoter region, including the AREs. g Osteocytes were transfected with the −4 100/+50 promoter construct (WT) or a mutant −3 900/+50 construct (with deletion of the AREs, ΔAREs), and Hmox1 promoter activity was assessed using a dual-luciferase reporter assay. h Osteocytes were treated as indicated for 24 h, and ChIP assays were performed with an anti-NRF2 antibody or normal IgG to detect AREs. i Schematic illustration of the proposed molecular mechanism underlying osteocyte ferroptosis in DOP. HGHF-induced HO-1 expression was dependent on the NRF2–c-JUN interaction. *P < 0.05; **P < 0.01. “NS” indicates nonsignificant. All data are from n = 3 independent experiments

Fig. 7

Targeting ferroptosis rescued DOP and osteocyte death in diabetic mice. a, b Representative micro-CT (scale bar: 300 μm) and X-ray (scale bar: 2 μm) images showing the microarchitecture of the distal femur. c Quantitative analysis of the micro-CT parameters. d H&E staining of tibial paraffin sections showing empty lacunae (scale bar: 50 μm). e TUNEL assay of tibial paraffin sections showing dead osteocytes (scale bar: 50 μm). f, g Tibial paraffin sections stained for HO-1 and GPX4 (scale bar: 50 μm). h, i Quantitative analysis of the number of empty lacunae with respect to the bone area (N. Empt. Lc./B. Ar.) and the fraction of empty lacunae (Frac. Empt. Lc.) based on H&E staining. j Quantitative analysis of TUNEL-positive cells based on a TUNEL assay. k, l Quantitative analysis of HO-1 and GPX4 expression in vivo based on IHC staining. *P < 0.05; **P < 0.01. Each group contained six mice