TET enzymes regulate skeletal development through increasing chromatin accessibility of RUNX2 target genes

Abstract

The Ten-eleven translocation (TET) family of dioxygenases mediate cytosine demethylation by catalyzing the oxidation of 5-methylcytosine (5mC). TET-mediated DNA demethylation controls the proper differentiation of embryonic stem cells and TET members display functional redundancy during early gastrulation. However, it is unclear if TET proteins have functional significance in mammalian skeletal development. Here, we report that Tet genes deficiency in mesoderm mesenchymal stem cells results in severe defects of bone development. The existence of any single Tet gene allele can support early bone formation, suggesting a functional redundancy of TET proteins. Integrative analyses of RNA-seq, Whole Genome Bisulfite Sequencing (WGBS), 5hmC-Seal and Assay for Transposase-Accessible Chromatin (ATAC-seq) demonstrate that TET-mediated demethylation increases the chromatin accessibility of target genes by RUNX2 and facilities RUNX2-regulated transcription. In addition, TET proteins interact with RUNX2 through their catalytic domain to regulate cytosine methylation around RUNX2 binding region. The catalytic domain is indispensable for TET enzymes to regulate RUNX2 transcription activity on its target genes and to regulate bone development. These results demonstrate that TET enzymes function to regulate RUNX2 activity and maintain skeletal homeostasis.

Fig. 1. Loss of Tet family genes leads to cleidocranial dysplasia-like phenotype.

a Gross images of indicated genotype mice. Scale bar = 5 mm. b The whole mount skeleton staining of WT and indicated genotype mice by Alcian blue and Alizarin red S at postnatal day 0 (P0). The quantification data of the unmineralized area was annotated below the cranial bones. scale bar = 5 mm. c–g Quantification of the length of clavicle (c), femur (d), tibia (e), radius (f), and humerus (g) of indicated genotype mice. *P < 0.05, **P < 0.01. Ordinary one-way ANOVA. Data are presented as mean ± s.d., WT mice = 7, TCKO mice = 5, Prx1-Cre; Tet1 ^f/+ Tet2 ^f/f Tet3 ^f/f mice=3, Prx1-Cre; Tet1 ^f/f Tet2 ^f/+ Tet3 ^f/f mice = 3, Prx1-Cre; Tet1 ^f/f Tet2 ^f/f Tet3 ^f/+ mice = 4 biologically independent animals. h 3D μ-CT images of cranial bones isolated from indicated genotype mice at postnatal day 10 (P10). Representative images are from two independent samples. i Quantification of the unmineralized area of cranial bones in h. n = 2 biologically independent animals.

Fig. 2. Loss of Tet family genes impairs bone formation.

a, b Immunofluorescence assay of 5mC and 5hmC in limb at embryonic day 16.5 (E16.5) (a) and calvarial bone at P0 (b) of WT and TCKO mice. Scale bar = 20 μm or 500 μm. Representative images for three independent samples. The images in the center were zoomed in from the flanking images. c Representative images of Von-kossa staining in the cranial bones from WT and TCKO mice at P0. Scale bar = 500 μm. d Immunofluorescence of bone formation marker Osteopontin (OPN) in the cranial bones from WT and TCKO mice at P0. Scale bar = 500 μm. The lines with endpoint indicated the unmineralized region in the calvarial bone. Representative images for three independent samples. e ALP staining and Alizarin red S staining after osteoblast differentiation for 7 days (top) and 21 days (bottom), respectively. Scale bar = 1 mm. Representative images for five independent samples. f ALP activity quantification was measured by phosphatase substrate assay as A405/Ala.Blue. **P < 0.01. Two-tailed Student’s t test. Data are presented as mean ± s.d., n = 5 independent cell supernatants. g RT-qPCR analysis of Tet1, Tet2, Tet3, Runx2, Sp7, Alpl, Col1α1, and Bglap expression after osteoblast differentiation for 7 days, cells were from WT and TCKO mice. *P < 0.05, **P <0.01. Two-tailed Student’s t test. Data are presented as mean ± s.d., n = 4 independent samples. h Representative images of Von-kossa staining in the femurs of WT and TCKO mice at E16.5. Scale bar = 500 μm.

Fig. 3. Tet-mediated bone formation is dependent on its catalytic domain.

a ALP staining and Alizarin red S staining after osteoblast differentiation for 7 days (up) and 21 days (bottom), respectively. Cells were isolated from calvarial bones of WT and TCKO mice, and cells from TCKO mice were infected with indicated lentivirus. Scale bar = 1 mm. b ALP activity quantification was measured by phosphatase substrate assay as A405/Ala. Blue. *P < 0.05, **P < 0.01. Ordinary one-way ANOVA. Data are presented as mean ± s.d., n = 4 independent cell supernatants. c The whole mount skeleton staining of WT and indicated genotype mice by Alcian blue and Alizarin red S at P0. Scale bar = 5 mm. d–h Quantification of the length of clavicle (d), femur (e), tibia (f), radius (g), and humerus (h) of indicated genotype mice. *P < 0.05, **P < 0.01. Two-tailed Student’s t test. Data are presented as mean ± s.d., WT mice = 7, Prx1-Cre; Tet1 ^HD/+ Tet2 ^f/f Tet3 ^f/f mice = 3, Prx1-Cre; Tet1 ^f/f Tet2 ^f/f Tet3 ^f/f mice = 4, Prx1-Cre; Tet1 ^HD/f Tet2 ^f/f Tet3 ^f/f mice = 3 biologically independent animals. i ALP staining and Alizarin red S staining after osteoblast differentiation for 7 days (up) and 21 days (bottom), respectively. Scale bar = 1 mm. j ALP activity quantification was measured with by phosphatase substrate assay as A405/Ala. Blue. *P < 0.05, **P < 0.01. Ordinary one-way ANOVA. Data are presented as mean ± s.d., n = 4 independent cell supernatants. k RT-qPCR analysis of Sp7, Alpl, Col1α1, and Bglap expression after osteoblast differentiation for 7 days, cells were from indicated genotype mice. *P < 0.05, **P < 0.01. Ordinary one-way ANOVA. Data are presented as mean ± s.d., n = 4 independent samples.

Fig. 4. Genome-wide differentially gene expression and DNA methylation patterns between TCKO and WT cells.

a Volcano plot of differential expressed protein-coding genes (Fold Change greater than 2 and FDR less than 0.05). b, c Enriched GO term (b) and KEGG pathway (c) analysis of downregulated genes in TCKO cells. d Barplot of genome-wide CG methylation level in TCKO and WT cells. e Metaplot of CG methylation level over protein-coding genes in TCKO and WT cells. f Pie chart of hypermethylated and hypomethylated CG-DMR counts in TCKO and WT cells. TSS transcription start site, TTS transcription termination site.

Fig. 5. Association analysis of RNA-seq, WGBS, 5hmC-Seal, ATAC-seq, and RUNX2 ChIP-seq data suggests TET and RUNX2 proteins could synergistically promote osteogenesis.

a Heatmap and metaplot of CG methylation level, 5hmC-Seal, and ATAC-seq signals over hyper-CG-DMR in TCKO cells. b Venn diagram of hyper-CG-DMR and chromatin Open-to-Close-Region (oC-R) in TCKO cells. c Top five enriched known motifs and their potential transcription factors in overlapped regions of hyper-CG-DMR and chromatin oC-R in TCKO cells. d Heatmap and metaplot of RUNX2 ChIP-seq, CG methylation level, and ATAC-seq signals of TCKO and WT cells over RUNX2 binding sites. e RAD analysis for regions shared by hyper-CG-DMRs in TCKO cells and RUNX2 binding sites and DEGs in TCKO cells compared with WT. *P < 0.01, the exact P values were provided in the source data; hypergeometric test. f Enriched GO terms of downregulated genes in TCKO cells, which had a distance less than 100 kb from overlapped regions of hyper-CG-DMRs in TCKO cells and RUNX2 binding sites.

Fig. 6. TET proteins associate with RUNX2 and promote osteogenesis with RUNX2 synergistically.

a Co-immunoprecipitation (Co-IP) of Flag-TET1 or Flag-TET2 or Flag-TET3 with HA-RUNX2. HA-RUNX2 expressing plasmid was co-transfected with Flag-TET1 or Flag-TET2 or Flag-TET3 in 293 T cells. Whole cell lysate was used for immunoprecipitation and then blotted with indicated antibodies. Representative images for two independent samples. b Co-immunoprecipitation (Co-IP) of Flag-TET1-CD or Flag-TET2-CD or Flag-TET3-CD with HA-RUNX2. HA-RUNX2 expressing plasmid was co-transfected with Flag-TET1-CD or Flag-TET2-CD or Flag-TET3-CD in 293 T cells. Whole cell lysate was used for immune precipitation and then immunoblotting with indicated antibodies. Representative images for 3 independent samples. c ALP staining and Alizarin red S staining after osteoblast differentiation for 7 days (left) and 21 days (right), respectively. Cells were infected with Runx2-lentivirus and Tet2-adenovirus separated or combined. Scale bar = 1 mm. d ALP activity quantification was measured by phosphatase substrate assay as A405/Ala.Blue. *P < 0.05, **P < 0.01. Ordinary one-way ANOVA. Data are presented as mean ± s.d., n = 3 independent cell supernatants. e–i RT-qPCR analysis of Runx2 (e), Tet2 (f), Col1α1 (g), Alpl (h), and Bglap (i) expression after osteoblast differentiation for 7 days, cells were from WT and TCKO mice. *P < 0.05, **P < 0.01. Ordinary one-way ANOVA. Data are presented as mean ± s.d., n = 4 independent samples.

Fig. 7. RUNX2 recruits TET proteins to the promoter of target genes to facilitate transcription by modification of DNA methylation.

a The strategy for promoter luciferase assay. OSE2: Osteocalcin-specific element 2. DMR: −2327bp to −1702bp of Col1α1 promoter. b, c The measurement of luciferase activity for 6XOSE2 (b) and 6XOSE2 + DMR (c) promoter stimulated by RUNX2. **P < 0.01. Two-tailed Student’s t test. Data are presented as mean ± s.d., n = 3 independent cell samples. d The measurement of luciferase activity for 6XOSE2 + DMR treated with TET2 or RUNX2 separated or combined. **P < 0.01. Ordinary one-way ANOVA. Data are presented as mean ± s.d., n = 3 independent cell samples. e–g ChIP-qPCR for RUNX2 over the promoter regions of osteogenic genes including Col1α1 (e), Col1α2 (f), Alpl (g). h–j ChIP-qPCR for RNA Polymerase II (RNA Pol II) over the promoter regions of osteogenic genes including Col1α1 (h), Col1α2 (i), Alpl (j). **P < 0.01. Two-tailed Student’s t test. Data are presented as mean ± s.d., n = 4 independent samples. k Based on our results, we proposed the model that RUNX2 could bind to the promoter regions of osteogenic genes. By recruiting TET proteins via RUNX2, DNA hypomethylation would be induced over those osteogenic genes and thus promote their expression by providing a more accessible chromatin landscape for transcriptional machinery such as RNA Pol II. Loss of all TET proteins would result in repression of osteogenic genes targeted by RUNX2, and thus lead to bone development failure.