FBXO31-mediated ubiquitination of OGT maintains O-GlcNAcylation homeostasis to restrain endometrial malignancy

Abstract

Protein O-GlcNAcylation is a post-translational modification coupled to cellular metabolic plasticity. Aberrant O-GlcNAcylation has been observed in many cancers including endometrial cancer (EC), a common malignancy in women. However, clinical characterization of dysregulated O-GlcNAcylation homeostasis in EC and interrogating its molecular mechanism remain incomplete. Here we report that O-GlcNAcylation level is positively correlated with EC histologic grade in a Chinese cohort containing 219 tumors, validated in The Cancer Genome Atlas dataset. Increasing O-GlcNAcylation in patient-derived endometrial epithelial organoids promotes proliferation and stem-like cell properties, whereas decreasing O-GlcNAcylation limits the growth of endometrial cancer organoids. CRISPR screen and biochemical characterization reveal that tumor suppressor F-box only protein 31 (FBXO31) regulates O-GlcNAcylation homeostasis in EC by ubiquitinating the O-GlcNAc transferase OGT. Downregulation of O-GlcNAcylation impedes EC tumor formation in mouse models. Collectively, our study highlights O-GlcNAcylation as a useful stratification marker and a therapeutic vulnerability for the advanced, poorly differentiated EC cases.

Fig. 1. Correlative analysis of O-GlcNAcylation level with clinical parameters.

a A flowchart illustrating the process of clinical sample selection, data collection, and analysis. All samples were derived from patients receiving their initial treatment, and none of the patients had concurrent or previous tumors. Paraffin-embedded (FFPE); Immunohistochemistry (IHC). b, c Representative images depicting IHC staining of O-GlcNAcylation (RL2) and OGT in EC tumoral and peritumoral tissues in the FFPE tissue array. O-GlcNAcylation (RL2) and OGT immunostaining were intense in the glandular epithelium of the tumor. Scale bars: 50 µm. d, e Quantitative analysis of the levels of O-GlcNAcylation (RL2) and OGT in the EC tissue arrays. The levels of O-GlcNAcylation and OGT were assessed semi-quantitatively based on both the intensity and area of the staining. The product of proportion and intensity score was used as the final IHC score (0–12). Tumoral tissue (n = 31); peritumoral tissue (n = 23). The results are presented as mean ± SD. Statistical significance was calculated using unpaired two-tailed Student’s t-test. f Representative images of IHC staining showing varying levels of O-GlcNAcylation in serial sections of EC tissues of different histologic grades (well differentiated G1, moderately differentiated G2, and poorly differentiated G3. G1, n = 71; G2, n = 106; G3, n = 42). Scale bar: 50 µm. g Percentage of samples with high or low level of O-GlcNAcylation in different histologic grade groups. High and low categories were determined using a scoring system (high score: 8–12; low score: 0–6). (G1, n = 71; G2, n = 106; G3, n = 42). Statistical significance between groups was calculated using two-sided Fisher’s exact test. h, i Kaplan–Meier survival curves of PFS and OS of the EC patients stratified by the level of O-GlcNAcylation derived from their IHC scores. (Patients in high-RL2 group, n = 89; Patients in low-RL2 group, n = 115). Statistical significance was determined by the log-rank test. The source data for (d-e, g, h, i) are provided in the Source Data file.

Fig. 2. Validation with TCGA endometrial cancer dataset.

a Heatmap displaying the expression profiles of the 1000 O-GlcNAcylation correlated genes in the TCGA UCEC RNA-seq dataset (n = 589). The EC samples are annotated by clinical parameters, including Body Mass Index (BMI), menopause status, diabetes, histologic grades, molecular subtypes (integrative cluster), International Federation of Gynecology and Obstetrics (FIGO) stage, age, and primary diagnosis. Patients were categorized into O-GlcNAcylation high or O-GlcNAcylation low group using the median of the calculated O-GlcNAc index as the threshold. The symbol (*) indicates a statistically significant difference of the calculated O-GlcNAc index among the patients’ groups according to the indicated clinical parameter. Statistical significance was determined by two-sided Wilcoxon test, **p < 0.01, ****p < 0.0001. b–e The O-GlcNAc index in different EC groups stratified by histologic grade, FIGO stage, integrative cluster, or age in the TCGA UCEC dataset. For histologic grade (b): G1 (n = 99), G2 (n = 119), and G3 (n = 324). For FIGO stage (c): Stage I (n = 335), Stage II (n = 51), Stage III (n = 127), and Stage IV (n = 29). For integrative clusters (d): POLE (n = 17), copy number low (n = 90), microsatellite unstable (n = 65), and copy number high (n = 61). For age (e), n = 177 and 362. The box bounds the interquartile range divided by the median, with the whiskers extending to a maximum of 1.5 times the interquartile range beyond the box. Outliers are shown as dots. Statistical significance was determined by two-sided Wilcoxon test. f, g Kaplan–Meier survival curves for Progression-free interval (PFI) and OS of EC groups with high or low O-GlcNAc index in the TCGA UCEC dataset. Statistical significance was determined by the log-rank test.

Fig. 3. Increase of O-GlcNAcylation level promotes proliferation and stemness.

a Bright-field images of endometrial epithelial organoids (EE-Os) depicting responses to Thiamet G (TMG) at day 1 and day 3. Representative images from control dimethyl sulfoxide (DMSO) and 10 µM TMG treated EE-O groups are presented. Scale bar: 50 µm. b Immunoblot with RL2 antibody assessing O-GlcNAcylation level in EE-Os treated with DMSO, 5 µM, or 10 µM TMG for 48 h. Actin was used as the loading control. c Comparison of the EE-O numbers at day 3 of culture after 10 µM TMG or DMSO treatment. d Measurement of cross-sectional area of EE-Os at day 3 after treatment with 10 µM TMG versus DMSO. (Organoids derived from 6 biological replicates, n = 94 and 165 organoids). e Representative immunofluorescence images of control and TMG-treated EE-Os stained with PH3 (red), Tubulin (green), DAPI (blue), and F-actin labeled by Phalloidin (magenta). Scale bar: 5 µm. f Quantification of phospho-histone H3 (PH3) positive cells in each EE-O. (n = 31 and 22 organoids). g Representative immunofluorescence images of control and TMG-treated EE-Os. Ciliated epithelium is labeled by acetylated alpha-tubulin (Ac-tubulin, green), secretory cells by PAEP (red), DAPI (blue), and F-actin (magenta). Scale bar: 50 µm. Insets show magnification of the area in the white box, scale bar: 5 µm. h Quantification of the number of ciliated cells (Ac-tubulin + ) in each EE-O (n = 23 and 25 organoids). i qPCR analysis of stemness markers’ expression in EE-Os treated with TMG or DMSO, normalized to actin mRNA level. j Minimum-Distortion Embedding (MDE) projection of scRNA-seq data of DMSO and TMG treated EE-Os. k Subclustered epithelial populations of EE-Os (left), and the proportion of each subcluster in control and TMG-treated groups (right). Results in (a, b) show a representative example from n = 3 independent experiments. Results in (c, d) were derived from n = 6 biologically independent experiments, and results in (f–i) were derived from n = 3 biologically independent experiments, with p-values calculated by unpaired two-tailed Student’s t-test and data presented as mean ± SD. The source data for (b–d, f, h, i, k) are provided in the Source Data file.

Fig. 4. Decrease of O-GlcNAcylation level induces differentiation and cell death.

a Representative bright-field images of endometrial cancer organoids (EC-Os) treated with 50 µM OSMI-1 at day 1 and day 3. Scale bar: 50 µm. b Immunoblot assessing O-GlcNAcylation level in EC-Os treated with 25 µM or 50 µM OSMI-1 for 48 h. Actin was used as the loading control. c Comparison of the numbers of EC-Os at day 3 after treatment with OSMI-1 versus the control DMSO. d Cross-sectional area of EC-Os at day 3 after OSMI-1 treatment compared to control (DMSO). (n = 101 and 57 organoids). e TUNEL staining showing apoptotic cells in EC-Os after 50 µM OSMI-1 treatment. Nuclei are visualized with DAPI (blue). Scale bar: 50 µm. f Representative immunofluorescence images of control and OSMI-1 treated EC-Os. Mitotic cells (PH3, red), tubulin (green), DAPI-labeled nuclei (blue), and Phalloidin labeled F-actin (magenta). Scale bar: 10 µm. g Quantification of the number of PH3+ cells. (n = 42 and 46 organoids). h Representative immunofluorescence images of control and OSMI-1 treated EC-Os showing ciliated epithelial cells (Ac-tubulin, green), secretory cells (PAEP, red), DAPI (blue), and F-actin (magenta). Scale bar: 50 µm. Insets show magnification of the area in the white box, scale bar: 5 µm. i Quantification of the number of Ac-tubulin+ cells. (n = 40 and 28 organoids). (j–l) qPCR analyses of stemness (j), EMT (k), and angiogenesis (l) markers in EC-Os treated with OSMI-1 or DMSO, normalized to actin mRNA. m Measurement of cell viability in patients-derived endometrial organoids with the indicated inhibitors. RLU represents relative light units. 5 replicates per each patient-derived organoid. Results in (a, b) show a representative example from n = 3 independent experiments. Results in (c, d) represent n = 6 biologically independent experiments, and results in (g, i-l) represent n = 3 biologically independent experiments, with p-values calculated by unpaired two-tailed Student’s t-test and data presented as mean ± SD. The source data for (b–d, g, i–m) are provided in the Source Data file.

Fig. 5. Screen for TSGs that regulate O-GlcNAcylation homeostasis.

a Schematic representation of the FACS-based genome-wide CRISPR-Cas9 screen for putative regulators of O-GlcNAcylation homeostasis. Fluorescence-Activated Cell Sorting (FACS); Immunofluorescence (IF); CRISPR: Clustered Regularly Interspaced Short Palindromic Repeats; Human genome-wide CRISPR/Cas9 knockout (GeCKO). The elements in this figure were created using BioGDP.com (https://BioGDP.com). b Validation of the sensitivity of RL2 staining (red) with 293T cells transfected with OGT (green). Nuclei are labeled with DAPI (blue). Scale bar: 5 µm. c Genes plotted according to their relative ranking analysis (RRA) enrichment scores, with known O-GlcNAcylation regulators highlighted in red and blue. Knockdown (KD). d Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis showing enrichment of putative O-GlcNAcylation regulators in the indicated pathways (statistical analysis was performed using a hypergeometric test to calculate p-values). Analysis was performed on the 1038 high-confidence genes (p < 0.05). e Venn diagram showing the overlap between the 526 human UCEC Tumor Suppressor Genes (TSGs) and the 1038 high-confidence genes from the O-GlcNAcylation screen. Source data are provided in the Source Data file. f Immunofluorescent detection of O-GlcNAcylation level by RL2 (red) in WT (Wild Type) and FBXO31-KO (FBXO31-Knockout) 293T cells. Nuclei were stained with DAPI (blue). Scale bar: 5 µm. g Kaplan-Meier analysis of the OS of the EC patients stratified by the expression level of FBXO31 (http://kmplot.com/analysis/). EC cases were stratified using the median cut-off, and statistical significance was determined using the log-rank test. Results in (b, f) show a representative example from n = 3 independent experiments.

Fig. 6. FBXO31 interacts with and ubiquitinates OGT.

a Immobilized recombinant GST-OGT protein but not GST control absorbed GFP-FBXO31 from 293T cell lysates. GST and GST-OGT were detected by Coomassie brilliant blue (CBB) staining, and FBXO31 was detected by western blotting with FBXO31 antibody. b Co-immunoprecipitation of GFP-FBXO31 with Flag-OGT in 293T cell lysates. The presence of MG132 enhanced the interaction between Flag-OGT and GFP-FBXO31. c Western blotting assessing the protein level of OGT as well as the global O-GlcNAcylation (RL2) level in 293T cells transfected with increasing amount of GFP-FBXO31. d Western blotting quantification of the protein level of endogenous OGT in 293T cells transfected with GFP-FBXO31. MG132 was added to inhibit the ubiquitination-mediated proteasome degradation. e Western blotting detecting the protein level of endogenous OGT and its ubiquitination in 293T cells transfected with different amount of HA-Ub and GFP-FBXO31. f In vitro ubiquitination of His-OGT by the SCF complex together with FBXO31. HA-tagged SCF components (Skp1, Cul1, and Roc1) and HA-FBXO31 were affinity-purified using anti-HA-conjugated magnetic beads from 293T cell lysates. The purified protein complex was incubated with E1 (UBA1), E2 (UBE2D1), Ub, and His-OGT in ubiquitination buffer. The reaction was halted by the addition of SDS sample buffer, and the samples were subjected to western blotting using the indicated antibodies. g In vivo ubiquitination assay was performed to evaluate the ubiquitination levels of exogenous Flag-OGT in 293T cells transfected with HA-tagged Ub and GFP-FBXO31 or its F-box domain deletion mutant GFP-FBXO31ΔF. h Western blotting detecting the O-GlcNAcylation (RL2) and OGT levels in WT and FBXO31-KO 293T cells. i Western blotting quantitation of OGT protein level following cycloheximide (CHX) treatment in WT and FBXO31-KO 293T cells. Results in (d, i) represent n = 3 independent experiments, with p-values calculated by unpaired two-tailed Student’s t-test and data presented as mean ± SD. Samples derive from the same experiment and gels were processed in parallel. (a–c, e–h) show representative examples from n = 3 independent experiments. The source data for results in (a–i) are provided in the Source Data file.

Fig. 7. Loss of FBXO31 increases O-GlcNAcylation level in clinical samples.

a Representative IHC images of FBXO31 in EC and peritumoral tissues from an FFPE tissue array. Scale bar: 50 µm. b Quantitative analysis of FBXO31 levels in the EC tissue array. FBXO31 expression was semi-quantified based on staining intensity and area. Tumoral tissue (n = 31); peritumoral tissue (n = 23). Results are presented as mean ± SD. Statistical significance was calculated using unpaired two-tailed Student’s t-test. c Percentage of samples with high or low FBXO31 level by IHC in the two different O-GlcNAcylation level groups (Patients in high-RL2 group, n = 83; Low-RL2 group, n = 38). Statistical significance was calculated using two-sided Fisher’s exact test. d Spearman two-sided correlation analysis between the calculated virtual O-GlcNAc index and the expression of FBXO31 (Transcripts Per Million, TPM), n = 542. e Protein levels of OGT and FBXO31 were assessed by western blotting in EC-Os and EE-Os derived from different patients. f Immunofluorescence detection of O-GlcNAcylation (RL2, green) and FBXO31 (red) in control and shFBXO31 infected EE-Os. The nuclei were stained with DAPI (blue) and F-actin with Phalloidin (magenta). Scale bar: 50 µm. g qPCR analysis of stemness markers’ expression in control shNT and shFBXO31 infected EE-Os, normalized to actin mRNA level. h Quantification of organoid numbers of the control and shFBXO31 infected EE-Os after 3D culture. Representative bright-field images are provided on the left. Scale bar: 300 µm. i Quantification of organoid numbers in shFBXO31 treated EE-Os, with OSMI-1 or DMSO treatment at day 3. Representative bright-field images are shown on the left. Scale bar: 150 µm. j Quantification of organoid numbers of EC-Os overexpressing GFP or GFP-FBXO31. Bright-field and fluorescent images of the treated EC-Os are shown on the left. Scale bar: 50 µm. Results in (g) represent n = 3 biologically independent experiments, and results in (h–j) represent n = 6 biologically independent experiments, with p-values calculated by unpaired two-tailed Student’s t-test and data presented as mean ± SD. e, f show a representative example from n = 3 independent experiments. The source data for results in (b–e, g–j) are provided in the Source Data file.

Fig. 8. Decrease of O-GlcNAcylation by inhibition of OGT limits tumor formation of EC cells in mouse models.

a Schematic representation of the treatment schedule in the Ishikawa cells xenograft model. On day 10 after subcutaneous injection of EC cells, mice were treated daily with DMSO, TMG, or OSMI-1 for 15 days. Tumor growth and survival were monitored till the endpoint. The mouse elements in this figure were created using BioGDP.com (https://BioGDP.com). b Tumor growth curves of Ishikawa xenografts in different treatment groups as indicated (DMSO group, n = 8 mice; TMG group, n = 7 mice; OSMI−1 group, n = 8 mice). The results are presented as mean ± SEM. c Survival curves for mice bearing Ishikawa xenografts across different treatment groups. (DMSO group, n = 8 mice; TMG group, n = 7 mice; OSMI-1 group, n = 8 mice). Statistical significance was determined by log-rank test. d Representative Hematoxylin and Eosin (HE) and IHC staining of mouse tumor tissues from different treatment groups. Scale bar: 50 µm. e Schematic representation of the treatment schedule in the WT and FBXO31-KO Ishikawa cells xenograft model. On day 10 of tumor growth, mice with WT or FBXO31-KO cells xenografts received daily treatment with DMSO or OSMI-1 for 15 days. Tumor growth was assessed till the endpoint. The mouse elements in this figure were created using BioGDP.com (https://BioGDP.com). f Tumor growth curves of WT and FBXO31-KO Ishikawa cells xenografts in different treatment groups as indicated (WT DMSO control group, n = 10 mice; WT OSMI-1 treatment group, n = 9; FBXO31-KO DMSO group, n = 10 mice; FBXO31-KO OSMI-1 treatment group, n = 9 mice). The results are presented as mean ± SEM. Statistical significance was calculated using unpaired two-tailed Student’s t-test. g Photograph of the excised tumors from different treatment groups as indicated. h Immunofluorescence detection of CD31 (red) in tumor tissues from the indicated treatment groups. Nuclei were stained with DAPI (blue). Scale bar: 50 µm. i Quantitative analysis of CD31-positive blood vessel areas. The results are presented as mean ± SD. Statistical significance was calculated using unpaired two-tailed Student’s t-test, n = 6 mice. The source data for results in (b–i) are provided in the Source Data file.

Fig. 9. Working model.

FBXO31-mediated ubiquitination of OGT maintains a relatively low level of O-GlcNAcylation in the non-cancerous endometrium. Inactivation of FBXO31 in endometrial cancer tissues results in accumulation of OGT and concurrent increase of O-GlcNAcylation that promote endometrial malignancy. The uterus elements in this figure were created using BioGDP.com (https://BioGDP.com).