PSMD14 Stabilizes SLC7A11 to Ameliorate Glucocorticoid-Induced Osteoporosis by Suppressing Osteocyte Ferroptosis

Abstract

Glucocorticoid-induced osteoporosis (GIOP) remains the most prevalent complication compromising bone health in patients undergoing glucocorticoid (GC) therapy. Despite its clinical significance, osteocyte death, a pivotal initiator of GC-driven bone metabolic imbalance, has received insufficient attention. This study identifies ferroptosis, an iron-dependent regulated cell death mechanism, as a novel pathological phenotype of osteocytes in GC microenvironments. Utilizing GPX4 conditional knockout mice and pharmacological ferroptosis inhibitors, this work demonstrates that osteocyte ferroptosis exacerbates GIOP progression. Metabolomic profiling reveals cystine insufficiency and glutathione depletion in GC-treated osteocytes. Mechanistically, GCs directly impede the deubiquitinase PSMD14 from binding to SLC7A11, thereby promoting SLC7A11 ubiquitination and proteasomal degradation, which sharply diminishes cystine uptake. Bone-targeting adeno-associated virus-mediated PSMD14 overexpression stabilized SLC7A11, attenuating both osteocytic ferroptosis and bone loss in GIOP mice. Through high-throughput virtual screening, this work identifies Pantethine as a potent PSMD14 activator that enhances deubiquitinase activity, restores SLC7A11 expression in osteocytes, and mitigates osteoporosis. Collectively, this study elucidates the role and mechanism of osteocyte ferroptosis in GIOP pathogenesis and proposes PSMD14-targeted therapy as a viable clinical strategy.

Keywords: PSMD14; SLC7A11; ferroptosis; glucocorticoid‐induced osteoporosis; ubiquitination.

Figure 1

The glucocorticoid microenvironment induced osteocyte ferroptosis in vivo. A) Schematic illustration of the preparation of cortical bone samples for proteomic analysis. 20‐week‐old wild‐type C57BL/6 mice were subcutaneously injected with 2.5 mg kg⁻¹ DEX (or PBS as control) three times a week for 8 weeks to establish the GIOP (or control) group. B) Volcano plot showing differentially expressed proteins from proteomics. C) KEGG analysis of the differentially expressed proteins. D) GSEA enrichment plot evaluating the changes of ferroptosis‐related proteins. E) Heatmap of representative proteins related to ferroptosis pathway between the Control and GIOP groups of mouse cortical bone. F) X‐ray imaging, H&E staining, and GPX4 histochemical staining of femoral tissues from normal and GIOP patients. G) Quantitative analysis of GPX4‐positive osteocytes in human femoral based on IHC staining (n = 8 per group). H) MDA content in human femoral tissue (n = 8 per group). I) Schematic showing the experimental protocol for 8‐weeks of DMSO/Fer‐1 injections in mice. J) Micro‐CT 3D restruction, H&E staining, and GPX4 histochemical staining of the distal femur of mice in each group. The processing details of each group are shown in (I). K) Distal femur BV/TV, Tb.Th, Tb.Sp, Tb.N, and BMD of mice in each group were measured by micro‐CT (n = 6 per group). Quantitative analysis of the empty lacunae in cortical bone (Number of empty lacunae with respect to bone area, N. Empt. Lc./B. Ar. per mm²) based on H&E staining (n = 6 per group). L) Quantification of GPX4‐positive osteocytes in mouse cortical femurs based on IHC staining (n = 6 per group). M) MDA content in tibia tissue of mice in each group (n = 6 per group). N) Maximum load and Maximum deflection of femoral cortical bone evaluated by the three‐point bending test (n = 6 per group). Data are expressed as mean ± SD, with biologically individual data points shown. P values were determined by unpaired two‐tailed Student's t test (G,H) and two‐way ANOVA test with Tukey's multiple comparisons (K–N), * p <  0.05, ** p <  0.01.

Figure 2

Knockout of GPX4 in osteocytes triggered ferroptosis and bone loss upon DEX exposure. A) Schematic showing the experimental protocol for 8‐weeks of DEX injections in transgenic mice. Control: GPX4^fl/fl mice. cKO (conditional osteocytic GPX4 knockout): Dmp1‐Cre; GPX4^fl/fl mice. B) Micro‐CT 3D restruction, H&E staining, and GPX4 histochemical staining of the distal femur of mice in each group. The processing details of each group are shown in (A). C) Distal femur BV/TV, Tb.Th, Tb.Sp, Tb.N, and BMD of mice in each group were measured by micro‐CT (n = 6 per group). Quantitative analysis of the empty lacunae in cortical bone (Number of empty lacunae with respect to bone area, N. Empt. Lc./B. Ar. per mm²) based on H&E staining (n = 6 per group). D) Quantification of GPX4‐positive osteocytes in mouse cortical femurs based on IHC staining (n = 6 per group). E) MDA content in mouse tibia tissue (n = 6 per group). F) Maximum load and Maximum deflection of femoral cortical bone evaluated by the three‐point bending test (n = 6 per group). G) CCK‐8 assay was performed on siRNA‐transfected MLO‐Y4 cells (50 nM si‐NC or si‐GPX4 for 24 h). After transfection was completed, MLO‐Y4 cells were treated with DEX (100 µM) for 48 h while adding the ferroptosis inhibitors (Fer‐1 1µM or DFO 100µM) (n = 5 per group). H–K) Western blot and quantitative analysis of GPX4 protein (H), MDA concentration detection (I), C11‐BODIPY 581/591 staining (J) and quantitative analysis (K) were performed on siRNA‐transfected MLO‐Y4 cells (50 nM si‐NC or si‐GPX4 for 24 h). After transfection was completed, MLO‐Y4 cells were treated with PBS or DEX (100 µM) for 48 h (n = 5 per group). L) Schematic showing the experimental protocol for 8‐weeks of Fer‐1/DFO injections in GIOP transgenic mice. M) Micro‐CT 3D restruction and H&E staining of the distal femur of mice in each group. The processing details of each group are shown in (K). N) Distal femur BV/TV, Tb.Th, Tb.Sp, Tb.N, and BMD of mice in each group were measured by micro‐CT (n = 6 per group). Quantitative analysis of the empty lacunae in cortical bone (Number of empty lacunae with respect to bone area, N. Empt. Lc./B. Ar. per mm²) based on H&E staining (n = 6 per group). O) MDA content in tibia tissue of mice in each group (n = 6 per group). P) Maximum load and Maximum deflection of femoral cortical bone evaluated by the three‐point bending test (n = 6 per group). Data are expressed as mean ± SD, with biologically individual data points shown. p values were determined by one‐way ANOVA test with Tukey's multiple comparisons (G–I,K,N–P) and two‐way ANOVA test with Tukey's multiple comparisons (C–F), ns, p > 0.05, * p <  0.05, ** p <  0.01.

Figure 3

DEX accelerated UPS‐dependent degradation of SLC7A11 and restricted cystine/glutathione metabolism. A) Schematic illustration of the preparation of cell samples for targeted metabolomics. MLO‐Y4 cells were treated with PBS or DEX (100 µM) for 48 h. B) Heatmap showing significantly differently expressed metabolites (p < 0.05) between PBS and DEX groups. C) KEGG analysis of the differentially expressed metabolites. D) Illustration of the cystine/glutathione metabolism pathway and the relative levels of cystine, cysteine, glutamine, glutamate, and glycine (n = 6 per group). E) GSH level detection was performed on MLO‐Y4 cells treated with PBS or DEX (100 µM) for 48 h (n = 9 per group). F) Cystine uptake assay was performed on MLO‐Y4 cells treated with PBS, DEX (100 µM) or Erastin (10 µM, for positive control) for 48 h (n = 9 per group). G) CCK‐8 assay was performed on MLO‐Y4 cells treated with DEX (100 µM) and Cystine (100 µM) for 48 h (n = 5 per group). H) Histochemical staining and quantitative analysis of SLC7A11 in human femoral from normal and GIOP patients (n = 8 per group). I) Histochemical staining and quantitative analysis of SLC7A11 in the distal cortical femurs of mice treated with PBS or DEX (n = 6 per group). J) Slc7a11 mRNA levels in femurs of mice treated with PBS or DEX (n = 6 per group). K) Slc7a11 mRNA levels in MLO‐Y4 cell treated with PBS or DEX (100 µM) for 48 h (n = 5 per group). L) Western blot and quantitative analysis of SLC7A11 protein was performed on MLO‐Y4 cells treated with different doses of DEX (0, 10, 50, 100 µM) for 8 h (Top). Western blot and quantitative analysis of SLC7A11 protein was performed on MLO‐Y4 cells treated with DEX (100 µM) for different time periods (0, 2, 4, 8 h) (Bottom) (n = 5 per group). M) Western blot and quantitative analysis of SLC7A11 protein were performed on MLO‐Y4 cells treated with CHX (50 µg mL⁻¹) and PBS/DEX (100 µM) for the indicared time (0, 2, 4, 8 h) (n = 5 per group). CHX: cycloheximide. N) Western blot and quantitative analysis of SLC7A11 protein were performed on MLO‐Y4 cells treated with MG132 (10 µM) with or without DEX (100 µM) for 8 h (n = 5 per group). O) Immunoprecipitation of SLC7A11 ubiquitination were performed on 10 µM MG132‐treated MLO‐Y4 cells supplemented with or without DEX (100 µM) for 8 h. P,Q) Western blot of GPX4 and SLC7A11 proteins (P) and MDA concentration detection (Q) were performed on AAV‐transfected MLO‐Y4 cells. After AAV‐NC/SLC7A11 transfection was completed, MLO‐Y4 cells were treated with PBS or DEX (100 µM) for 8 h (n = 5 per group). AAV: adeno‐associated virus. Data are expressed as mean ± SD, with biologically individual data points shown. p values were determined by unpaired two‐tailed Student's t test (D–F,H–K), one‐way ANOVA test with Tukey's multiple comparisons (G,L) and two‐way ANOVA test with Tukey's multiple comparisons (M,N,P,Q), ns, p > 0.05, * p <  0.05, ** p <  0.01.

Figure 4

PSMD14 acts as a major regulator for SLC7A11 stability in response to DEX stimulation. A) Schematic illustration of the preparation of SLC7A11‐binding protein sample in MLO‐Y4 cells for LC‐MS/MS analysis to identify the deubiquitinase. B) After proofreading using the UbiBrowser library, the SLC7A11‐bound deubiquitinases were ranked according to sequence coverage. C) LC‐MS/MS analysis of SLC7A11‐bound deubiquitinase eluted from MLO‐Y4 cell lysate. PSM: Peptide Spectrum Match. D) Immunoprecipitation of SLC7A11 or control IgG was performed on MLO‐Y4 cells treated with PBS or DEX (100 µM) for 8 h. E) Immunoprecipitation of PSMD14 or control IgG was performed on MLO‐Y4 cells. F) Immunoprecipitation of the interaction between SLC7A11 and PSMD14 was performed on plasmid‐transfected MLO‐Y4 cells with the indicated antibodies. After transfection was completed, MLO‐Y4 cells were treated with MG132 (10 µM) for 8 h while adding PBS or DEX (100 µM). G) Western blot and quantitative analysis of Flag‐PSMD14 and SLC7A11 protein was performed on MLO‐Y4 cells transfected with indictaed plasmid (n = 5 per group). H) Western blot and quantitative analysis of PSMD14 protein was performed on MLO‐Y4 cells transfected with si‐RNA (50 nM si‐NC or si‐PSMD14 #1‐3 for 24 h) (n = 5 per group). I) Western blot and quantitative analysis of PSMD14 and SLC7A11 proteins were performed on siRNA‐transfected MLO‐Y4 cells (50 nM si‐NC or si‐PSMD14 for 24 h). After transfection was completed, MLO‐Y4 cells were treated with PBS or DEX (100 µM) for 8 h (n = 5 per group). J) Immunoprecipitation of SLC7A11 ubiquitination and its binding to PSMD14 were performed on siRNA‐transfected MLO‐Y4 cells (50 nM si‐NC or si‐PSMD14 for 24 h). After transfection was completed, MLO‐Y4 were treated with MG132 (10 µM) for 8 h while adding PBS or DEX (100 µM). K) Predicted binding complex model of SLC7A11 and PSMD14 (left). The picture showed the hydrogen bonds in the protein interaction region and the corresponding amino acid residues (right). L) Binding energy analysis of the PSMD14‐SLC7A11 complex and the PSMD14‐SLC7A11‐DEX complex based on molecular dynamics simulations. Data are expressed as mean ± SD, with biologically individual data points shown. p values were determined by one‐way ANOVA test with Tukey's multiple comparisons (H) and two‐way ANOVA test with Tukey's multiple comparisons (G,I), * p <  0.05, ** p <  0.01.

Figure 5

PSMD14 maintains SLC7A11 expression by cleaving K48‐linked polyubiquitin chains from SLC7A11. A) Schematic diagram of the PSMD14‐SLC7A11 protein complex and the PSMD14 mutant plasmids (residues 1–233 and 234–310) used in subsequent immunoprecipitation assays. B) Immunoprecipitation of the interaction between SLC7A11 and each PSMD14 fragment was performed on plasmid‐transfected MLO‐Y4 cells. C) Immunoprecipitation of the interaction between SLC7A11 and PSMD14 (234‐310) fragment was performed on plasmid‐transfected MLO‐Y4 cells. After transfection was completed, MLO‐Y4 cells were treated with PBS or DEX (100 µM) for 8 h. D,E) Immunoprecipitation was performed to identify the type of polyubiquitination of SLC7A11 in MLO‐Y4 cell. MLO‐Y4 cells were cotransfected with Myc‐SLC7A11 plasmid, Flag‐PSMD14 plasmid and HA‐Ub K48/K63 or HA‐Ub K48R/K63R plasmid. After transfection was completed, MLO‐Y4 cells were treated with MG132 (10 µM) for 8 h. F) Immunoprecipitation of SLC7A11 ubiquitination and its binding to PSMD14 was HA‐Ub K48/K48R plasmid‐treated MLO‐Y4 cells. After transfection was completed, MLO‐Y4 cells were treated with MG132 (10 µM) and supplemented with DEX (100 µM) and/or THL (2 µM) for 8 h. G) Stable PSMD14 knockout MLO‐Y4 cells were constructed by CRISPR‐Cas9 strategy. H) Deubiquitination of SLC7A11 requires the intact PSMD14 fragment. PSMD14 knockout MLO‐Y4 cells treated with HA‐Ub K48 plasmid, Myc‐SLC7A11 plasmid, and PSMD14 full‐length or deletion mutant plasmids upon MG132 (10 µM) treatment for 8 h. I) Deubiquitination of SLC7A11 requires the intact PSMD14 fragment. Stable PSMD14 knockout MLO‐Y4 cells were treated with HA‐Ub K48 plasmid, Myc‐SLC7A11 plasmid, and Flag‐PSMD14 or PSMD14 mutants upon MG132 (10 µM) treatment for 8 h. The experiments were repeated three times independently with similar results.

Figure 6

Activation of PSMD14 with Pantethine suppresses osteocyte ferroptosis and bone loss. A) Schematic diagram of screening and identification of PSMD14 agonist. B) Docking scores of the top 20 candidates based on virtual screening. C) CCK‐8 assay was performed on MLO‐Y4 cells treated with different 20 candidates (10 µM) and DEX for 48 h (n = 5 per group). D) Chemical structures of five drug candidates (CA, TA, TB, PT and PC). E) Western blot and quantitative analysis of GPX4 protein were performed on MLO‐Y4 cells treated with DEX (100 µM) supplemented with or wihtout candidates (CA, TA, TB, PT or PC 10 µM) for 48 h. F) Cystine uptake assay was performed on MLO‐Y4 cells treated with DEX (100 µM) supplemented with or wihtout candidates (CA, TA, TB, PT or PC 10 µM) for 48 h (n = 9 per group). G) Immunoprecipitation of SLC7A11 ubiquitination and its binding to PSMD14 were performed on MG132 and DEX‐exposed MLO‐Y4 cells (MG132 10µM and DEX 100 µM). MG132 and DEX‐expousred MLO‐Y4 cells were treated with PT (10 µM) and/or THL (2 µM) for 8 h. H) Binding affinity of PT with recombinant PSMD14 was determined using an SPR assay (K_D = 5.14 µM). I) Recombinant PSMD14 were incubated with PT, followed by the measurement of the absorbance at OD 445 nm to detect PSMD14 activity using Ubiquitin‐AMC assay (n = 5 per group). J) CCK‐8 assay was performed on DEX‐exposed MLO‐Y4 cells treated with PT (0–100 µM) and/or THL (2 µM) for 48 h (n = 5 per group). K–N) Western blot and quantitative analysis of GPX4 protein (K), MDA concentration detection (L), C11‐BODIPY 581/591 staining (M) and quantitative analysis (N) were performed on DEX‐exposed MLO‐Y4 cells treated with PT (100 µM) and/or THL (2 µM) for 48 h (n = 5 per group). O) Schematic showing the experimental protocol for 8‐weeks of PT / PT + THL injections in GIOP mice. P) Micro‐CT 3D restruction and H&E staining of the distal femur of mice in each group. The processing details of each group are shown in (O). Q) Distal femur BV/TV, Tb.Th, Tb.Sp, Tb.N, and BMD of mice in each group were measured by micro‐CT (n = 6 per group). Quantitative analysis of the empty lacunae in cortical bone (Number of empty lacunae with respect to bone area, N. Empt. Lc./B. Ar. per mm²) based on H&E staining (n = 6 per group). R) GPX4, SLC7A11 and PSMD14 IHC staining of the distal femur of mice in each group. The processing details of each group are shown in (O). S) Quantification of GPX4, SLC7A11 and PSMD14‐positive osteocytes in mouse cortical femurs based on IHC staining (n = 6 per group). T) MDA content in tibia tissue of mice in each group (n = 6 per group). U) Maximum load and Maximum deflection of femoral cortical bone evaluated by the three‐point bending test (n = 6 per group). Data are expressed as mean ± SD, with biologically individual data points shown. p values were determined by one‐way ANOVA test with Tukey's multiple comparisons (C,E,F,H,I,L,N,Q,S–U) and two‐way ANOVA test with Tukey's multiple comparisons (J,K), ns, p > 0.05, * p <  0.05, ** p <  0.01.

Figure 7

Schematic diagram illustrating the mechanism of GC‐mediated osteocyte ferroptosis. Top: Under physiological conditions, PSMD14 binds to and deubiquitinates SLC7A11 to stabilize SLC7A11 expression and the cystine uptake capacity of osteocytes, thereby ensuring the GSH content and GPX4 activity in osteocytes to maintain cellular function and vitality. Down: During DEX exposure, SLC7A11 is degraded due to limited binding with PSMD14, leading to insufficient cystine in osteocytes and triggering ferroptosis. Combined with virtual screening, we identified PT as a PSMD14 agonist that stabilizes SLC7A11 expression against DEX‐mediated ferroptosis.