Crucial role of iron in epigenetic rewriting during adipocyte differentiation mediated by JMJD1A and TET2 activity

Abstract

Iron metabolism is closely associated with the pathogenesis of obesity. However, the mechanism of the iron-dependent regulation of adipocyte differentiation remains unclear. Here, we show that iron is essential for rewriting of epigenetic marks during adipocyte differentiation. Iron supply through lysosome-mediated ferritinophagy was found to be crucial during the early stage of adipocyte differentiation, and iron deficiency during this period suppressed subsequent terminal differentiation. This was associated with demethylation of both repressive histone marks and DNA in the genomic regions of adipocyte differentiation-associated genes, including Pparg, which encodes PPARγ, the master regulator of adipocyte differentiation. In addition, we identified several epigenetic demethylases to be responsible for iron-dependent adipocyte differentiation, with the histone demethylase jumonji domain-containing 1A and the DNA demethylase ten-eleven translocation 2 as the major enzymes. The interrelationship between repressive histone marks and DNA methylation was indicated by an integrated genome-wide association analysis, and was also supported by the findings that both histone and DNA demethylation were suppressed by either the inhibition of lysosomal ferritin flux or the knockdown of iron chaperone poly(rC)-binding protein 2. In summary, epigenetic regulations through iron-dependent control of epigenetic enzyme activities play an important role in the organized gene expression mechanisms of adipogenesis.

Graphical Abstract

Crucial role of iron in epigenetic rewriting during adipocyte differentiation mediated by JMJD1A and TET2 activity.

Figure 1.

Ferritinophagy is induced during the early stage of adipocyte differentiation. (A) Adipocyte differentiation of 3T3-L1 cells was induced with the addition of bafilomycin A1 at the indicated concentrations for the first 2 days, and ORO staining was performed on Day 8. (B) The lysosomal flux of ferritin was measured by immunoblot analysis on Days 0, 1 and 2 using whole cell lysates from 3T3-L1 cells. Cells were treated with DMSO or 100 nM bafilomycin A1 for 24 h prior to the collection of whole cell lysates (n = 3 biological replicates). Ferritin levels were normalized to the total protein level quantified by TGX stain-free gel (left). Ferritin flux to the lysosomes was calculated by subtracting the ferritin level in DMSO-treated cells from that in bafilomycin A1-treated cells (right). Data are shown as the mean ± s.e.m. The two-tailed Student's t-test (left) or one-way ANOVA followed by the Tukey-Kramer test (right) was performed for statistical analysis. *P < 0.05. The immunoblot images are shown in Supplementary Figures S1B and S10. (C) On the indicated day of differentiation, 3T3-L1 cells were treated with 200 μM 2,2-bipyridyl for indicated hours (top) or 200 μM 2,2-bipyridyl and either 100 nM bafilomycin A1 or DMSO for 3 h (bottom). Ferritin levels in whole cell lysates were quantified by immunoblot analysis, and normalized to the GAPDH level (n = 3 biological replicates) (top). The change in the ferritin level in response to 2,2-bipyridyl treatment was calculated by subtracting the level before treatment from the level 3 h after treatment (n = 3 biological replicates) (bottom). Data are shown as the mean ± s.e.m. One-way ANOVA followed by the Tukey–Kramer test (top) or the two-tailed Student's t-test (bottom) was performed for statistical analysis. *P < 0.05, **P < 0.01. The immunoblot images are shown in Supplementary Figures S1C, D, and S10. (D, E) The lysosomal flux of NCOA4 (D) and LC3-II (E) was measured by immunoblot analysis as performed in (B) (n = 3 biological replicates). Data are shown as the mean ± s.e.m. The two-tailed Student's t-test (D, left, E, left) or one-way ANOVA followed by the Tukey-Kramer test (D, right, E, right) was performed for statistical analysis. *P < 0.05. The immunoblot images are shown in Supplementary Figures S1C, D, and S10. (F, G) On Days 0, 1 and 2, 3T3-L1 cells were treated with 100 nM bafilomycin A1 for 6 h, and immunostained for LC3 and ferritin (F), or LC3 and NCOA4 (G). Scale bar, 10 μm. For evaluating the colocalization of LC3 and ferritin (F) or NCOA4 (G), Pearson's R-values in the cytoplasmic region of each cell were calculated in 50 to 80 cells (F, right, G, right). The horizontal line indicates the median value. The Kruskal–Wallis test followed by the Steel–Dwass test was performed for statistical analysis. *P < 0.05. The immunoblot images are shown in Supplementary Figures S1B and S10. (H, I) The NCOA4-KD cell line established by stably expressing an shRNA against Ncoa4 mRNA (sh-Ncoa4), and its control cell line (sh-Ctrl) were induced to undergo adipocyte differentiation. The lysosomal flux of ferritin was measured on Days 0 and 2 by immunoblot analysis, as performed in (B) (n = 3 biological replicates) (H). ORO staining was performed on Day 8 (I). Data are shown as the mean ± s.e.m. The two-tailed Student's t-test (H, left) or one-way ANOVA followed by the Tukey–Kramer test (H, right) were performed for statistical analysis. *P < 0.05. The immunoblot images of (H) are shown in Supplementary Figures S1F and S10. (J) Whole cell lysates of the NCOA4-KD cell line (sh-Ncoa4) and the control cell line were subjected to immunoblot analysis using an anti-transferrin receptor (TfR) antibody on Days 0 and 2 (n = 3 biological replicates). TfR levels were normalized to the total protein level quantified by TGX stain-free gel. Data are shown as the mean ± s.e.m. One-way ANOVA followed by the Tukey–Kramer test was performed for statistical analysis. *P < 0.05. The immunoblot images are shown in Supplementary Figure S10.

Figure 2.

Transient iron depletion during the early stage of adipocyte differentiation suppresses terminal differentiation of 3T3-L1 cells. (A) Cells were treated with vehicle (−) or DFO (100 μM) for the indicated times during adipocyte differentiation as schematically illustrated (left). Cells were stained with ORO on Day 8, and a scanned picture of the cell culture plate (top right) and bright field microscopy images (bottom right) are shown. (B) Cells were treated with the indicated concentrations of DFO for the first 2 days of adipocyte differentiation, and stained with ORO on Day 8. (C) Cells were differentiated into adipocytes with (+) or without (−) 100 μM DFO. Protein levels of ferritin and β-actin at the indicated time points were determined by immunoblot analysis using whole cell extracts. Uncropped images are shown in Supplementary Figure S10. (D) RNA-seq heatmap depicting the comparison of gene expression patterns at each time point (Days 0, 1, 2, 4 and 8) for cells treated with vehicle (−) or DFO (100 μM) for the first two days during adipocyte differentiation. For each gene, z-scores were calculated based on log2-transformed mean TPM values, and are shown along the color scale. Expression patterns of the genes, except for genes with a coefficient of variation (CV) of <0.2, were classified by hierarchical clustering. A total of 4907 expressed genes were classified into four groups (cluster 1: 1742 genes; cluster 2: 901 genes; cluster 3: 1692 genes; cluster 4: 572 genes). Both Pparg and Cebpa were in cluster 3 as indicated. (E) The top 3 canonical pathways ranked by P-value for DEGs that were upregulated (positive z-score) or downregulated (negative z-score) by DFO treatment on Day 1. (F) Transcriptional changes of adipogenic regulatory genes (Pparg, Cebpa, and Cebpb) determined by RNA-seq. Data are presented in TPM and circles indicate two biological replicates. Insets show the changes from Day 0 to Day 1. (G) Protein levels of PPARγ, C/EBPα and β-actin at the indicated time points during 3T3-L1 cell differentiation were determined by immunoblot analysis using whole cell extracts. Uncropped images are shown in Supplementary Figure S10.

Figure 3.

Iron-dependent demethylation of histones and DNA correlates with the expression of adipocyte differentiation-associated genes. ChIP-seq analysis against repressive histone marks (H3K9me2, H3K9me3, and H3K27me3) was performed. 3T3-L1 cells were induced to differentiate with [DFO(+)] or without [DFO(−)] the addition of 100 μM DFO during the first two days of differentiation, and samples were collected at the indicated time points. (A) The ChIP-seq signals of H3K9me2, H3K9me3, and H3K27me3 were calculated as the normalized count per million (CPM) in the TSS ±5 kb region of each DEG with upregulated expression from Day 0 to Day 2 in the DFO(−) condition. The patterns of histone modifications among the three groups, i.e. Day 0, DFO(−) on Day 2, and DFO(+) on Day 2, were classified into seven clusters based on the z-score calculated from the mean CPM of duplicate cultures. (B) GO enrichment analysis was performed across the 7 clusters in (A) for the terms selected as relevant to adipocyte biology using the ubiquitously distributed ‘ribosome biogenesis’ (highlighted in gray) as a control. For reference, both a color intensity scale of the adjusted P-value and a size scale of the gene ratio are included. (C) Venn diagram depicting the number of overlapping genes in cluster 6 of H3K9me2, H3K9me3, and H3K27me3 in (A). (D) The normalized CPM of H3K9me2 ChIP-seq signals in the TSS ± 5 kb of Pparg. The circles represent two individual replicates for each treatment. (E) WGBS was performed in triplicate using 3T3-L1 cells differentiated into adipocytes with or without 100 μM DFO treatment for the first two days of differentiation. Metilene software detected 2521 DMRs as significantly decreased on Day 8. The enriched motifs among the DMRs are shown. (F) Using the WGBS data in (E), the DNA methylation level of the region within 1 kb upstream from the TSS of each gene was analyzed. CpGs that differ in methylation level by more than 50% on Day 8 [DFO(−) or DFO(+)] compared with Day 0 were identified, and the number of regions with 3 or more such CpGs is shown in the Venn diagram. (G) Top 5 upstream regulators of 330 iron-dependent demethylated genes in (F). (H) The 330 iron-dependent demethylated regions during adipocyte differentiation in (F) were annotated into their flanking genes. RNA-seq heatmap depicting their expression profiles throughout adipocyte differentiation classified into four clusters by Ward's hierarchical clustering method.

Figure 4.

JMJD1A mediates iron-dependent demethylation of H3K9me2 in the Pparg region. (A) Pparg mRNA levels on Day 2 in a series of 3T3-L1 cell lines stably expressing either sh-Jmjd1a, sh-Jmjd2b, sh-Jhdm1d, sh-Jmjd2c, sh-Jmjd2d, sh-Utx, sh-Phf2, sh-Jmjd2a, or sh-Jmjd3 were determined by qPCR (n = 3 biological replicates). The mRNA levels normalized to the Cyclophilin B level in the cell lines stably expressing an shRNA for each target enzyme (shRNA #1 or shRNA #2) is shown as a ratio to the level of the corresponding control cell line (sh-Ctrl). Data are shown as the mean ± s.e.m. The two-tailed Student's t-test was performed for the cell lines expressing either sh-Jmjd2c, sh-Jmjd2d, or sh-Phf2. One-way ANOVA followed by the Dunnett test was performed for the other cell lines. *P < 0.05. (B) Each knockdown line was retrovirally transduced with the corresponding enzyme with or without the indicated mutation in the iron-binding site. Pparg mRNA levels on Day 2 were measured by qPCR (n = 3 biological replicates). The full-length mouse sequence of the corresponding enzyme gene was used for overexpression, except for the partial sequence encoding amino acids (a.a.) 71 to 940 of JHDM1D and the human version for UTX. Data are shown as the mean ± s.e.m. One-way ANOVA followed by the Tukey-Kramer test was performed for statistical analysis. *P < 0.05. (C) ChIP-qPCR analysis of H3K9me2 on Pparg and Actb genes in JMJD1A-KD cells (sh-Jmjd1a #1) and its control cell line (sh-Ctrl) on Day 2 (mean ± s.e.m. of three biological replicates). ChIP signals were presented as a percentage of input DNA. The two-tailed Student's t-test was performed for statistical analysis. *P < 0.05, ***P < 0.001, ****P < 0.0001, N.S., not significant. (D) ORO staining was performed on Day 8 in JMJD1A-KD cells (sh-Jmjd1a #1), or the JMJD1A-KD cells (sh-Jmjd1a #1) that stably express mouse JMJD1A harboring an shRNA-resistant mutation with [JMJD1A (H1122A-sh^R)] or without [JMJD1A (WT-sh^R)] additional mutations in the iron-binding site. (E) JMJD1A-KD cells (sh-Jmjd1a #1) overexpressing either lacZ [Ctrl (lacZ)], JMJD1A (WT-sh^R), or JMJD1A (H1122A-sh^R) were induced to differentiate, and mRNA levels of Jmjd1a and Pparg on Day 2 were measured by qPCR (n = 3 biological replicates). Data are shown as the mean ± s.e.m. One-way ANOVA followed by the Tukey-Kramer test was performed for statistical analysis. *P < 0.05. (F) ChIP-qPCR analysis of H3K9me2 on Pparg and Actb genes. JMJD1A-KD cells (sh-Jmjd1a #1) overexpressing either lacZ [Ctrl (lacZ)], JMJD1A (WT-sh^R), or JMJD1A (H1122A-sh^R) were differentiated with 100 μM DFO [DFO(+)] (bottom) or vehicle [DFO(−)] (top). ChIP-qPCR was performed on Day 2 (mean ± s.e.m. of three biological replicates). One-way ANOVA followed by the Tukey-Kramer test was performed for statistical analysis. *P < 0.05; N.S.: not significant. (G) JMJD1A-KD cells (sh-Jmjd1a #1) overexpressing either lacZ [Ctrl (lacZ)], JMJD1A (WT-sh^R), or JMJD1A (H1122A-sh^R) were induced to differentiate, and protein levels of PPARγ and β-actin on Day 2 of differentiation were determined by immunoblot analysis using whole cell extracts. Uncropped images are shown in Supplementary Figure S10. (H) Adipocyte differentiation of 3T3-L1 cells was induced with or without treatment of 100 μM DFO for the first 2 days, and mRNA levels of Jmjd1a during adipocyte differentiation were analyzed by RNA-seq, as described in Figure 2D. Data are presented in TPM and circles indicate two biological replicates.

Figure 5.

Iron regulates the H3K9me2 demethylase activity of JMJD1A. (A) Histone demethylase activity of JMJD1A was determined by the HTRF demethylation assay using the recombinant JMJD1A protein and its substrate (biotinylated-H3K9me1 or biotinylated-H3K9me2). The vertical axis represents DF%: DF% = ([665 nm/620 nm of Enzyme (+) condition] / [665 nm/620 nm of Enzyme (−) condition] −1) ×100). Data are shown as the mean ± SD of 3 technical replicates. (B) Cells transfected with pCAG-HA-Jmjd1a were cultured in the presence or absence of 100 μM DFO for 2 days, and immunostained with anti-H3K9me2 and anti-HA antibodies. Counterstaining was performed using Hoechst. Dashed circles indicate HA-JMJD1A-positive cells. Scale bar, 50 μm. (C) JMJD1A activity was determined by the in vitro demethylation assay in the presence or absence of DFO. Flag-Twin-Strep-tagged-WT-JMJD1A or Flag-Twin-Strep-tagged-H1122A-JMJD1A were affinity purified from 3T3-L1 preadipocytes overexpressing the respective proteins. Acquired purified WT-JMJD1A or H1122A-JMJD1A was reacted with the synthesized H3K9me2 peptide in a buffer (50 mM HEPES–KOH, pH 7.5, 1 mM α-KG, 2 mM ascorbic acid, and 70 μM ferrous ammonium sulfate) with or without 100 μM DFO (top). Flag-Twin-Strep-tagged-WT-JMJD1A WT or Flag-Twin-Strep-tagged-H1122A-JMJD1A was affinity purified from 3T3-L1 cells induced to differentiate with or without 100 μM DFO for 2 days. The resultant purified WT-JMJD1A or H1122A-JMJD1A was reacted with the synthesized H3K9me2 peptide in an iron-free buffer (50 mM HEPES–KOH, pH 7.5, 1 mM α-KG, and 2 mM ascorbic acid) (bottom). Protein levels were determined by performing immunoblot analysis using anti-H3K9me2 and anti-FLAG. Uncropped images are shown in Supplementary Figure S10. (D) Eight-week-old male mice were intraperitoneally injected with DFO (100 mg/kg BW/day) for 2 weeks under a high-fat, high-cholesterol diet, and H3K9me2 levels in epididymal WAT were determined by ChIP-qPCR (mean ± s.e.m. of three biological replicates). ChIP signals were presented as a percentage of input DNA. The two-tailed Student t-test was performed for statistical analysis. *P < 0.05, **P < 0.01

Figure 6.

TET2 mediates iron-dependent DNA demethylation during adipocyte differentiation. (A) Pparg mRNA levels on Day 2 in a series of 3T3-L1 cell lines with knockdown of either TET1, TET2 or TET3 determined by qPCR (n = 3 biological replicates). The mRNA levels normalized to the Cyclophilin B mRNA level in the cell lines stably expressing shRNA for each target enzyme (shRNA #1 or shRNA #2) are shown as ratios to the level of the corresponding control cell line (sh-Ctrl). Data are shown as the mean ± s.e.m. The two-tailed Student's t-test was performed for the cell line stably expressing sh-Tet3. One-way ANOVA followed by the Dunnett test was performed for the cell lines stably expressing either sh-Tet1 or sh-Tet2. *P < 0.05. (B) The 3T3-L1 cell lines in which the expression of each demethylase was knocked down were retrovirally transduced with the corresponding enzyme with or without the indicated mutation in the iron-binding site. mRNA levels on Day 2 were measured by qPCR (n = 3 biological replicates). Data are shown as the mean ± s.e.m. One-way ANOVA followed by the Tukey-Kramer test was performed for statistical analysis. *P < 0.05. (C) Bisulfite sequence analysis of methylated CpGs in the promoter region of Pparg2 (NM_011146) was performed using the TET2-KD cell line (sh-Tet2 #2) and its control cell line (sh-Ctrl) on Days 0, 2 and 8. The positions of the CpGs are shown at the top. Methylated and unmethylated CpGs are presented as closed and open circles, respectively (16 clones in each group). The levels of methylated CpGs under each condition are shown as percentages. (D) TET2-KD cell lines stably expressing either lacZ [Ctrl (lacZ)] or mouse TET2 harboring an shRNA-resistant mutation with (H1295Y/D1297A-sh^R) or without (WT-sh^R) additional mutations in the iron-binding site were differentiated into adipocytes. Bisulfite sequence analysis of methylated CpGs in the promoter region of Pparg2 were performed as in (C). (E) TET2-KD cells (sh-Tet2 #2) overexpressing either lacZ [Ctrl (lacZ)], TET2 (WT-sh^R), or TET2 (H1295Y/D1297A-sh^R) were induced to differentiate, and protein levels of PPARγ and β-actin on Day 2 of differentiation were determined by immunoblot analysis using whole cell extracts. Uncropped images are shown in Supplementary Figure S10. (F, G) ORO staining of the cell lines presented in (C) and (D) was performed on Day 8 and presented in (F) and (G), respectively. (H) Cells transfected with pCAG-HA-Tet2 were cultured in the presence or absence of 100 μM DFO for 2 days, and immunostained with anti-5mC and anti-HA antibodies. Counterstaining was performed using Hoechst. Dashed circles indicate HA-TET2-positive cells. Scale bar, 20 μm. (I) Integrated analysis of WGBS data and ChIP-seq data of repressive histone marks (H3K9me2, H3K9me3 and H3K27me3) to determine the colocalization of histone demethylation and DNA demethylation. Aggregation plots showing changes in DNA and histone methylation levels around the DMRs that become less methylated during adipocyte differentiation. Methylation levels at the indicated time points compared with Day 0 are shown in the DMRs (±3 kb from center). (J) 3T3-L1 cells were induced to differentiate into adipocytes with 25 nM bafilomycin A1. ChIP-qPCR was performed using an anti-H3K9me2 antibody on Day 2 (mean ± s.e.m. of three biological replicates) (left). Bisulfite sequence analysis of methylated CpGs in Pparg2 was performed on Day 0, 2 and 8 (right). The two-tailed Student's t-test was performed for statistical analysis. **P < 0.01, ***P < 0.001, ****P < 0.0001

Figure 7.

Iron chaperone PCBPs translocate to the nucleus and are crucial for adipocyte differentiation. (A) Nuclear PCBP2 levels in 3T3-L1 cells on Days 0, 2, 4 and 8 were quantified by immunostaining using an anti-PCBP2 antibody. The normalized PCBP2 fluorescence intensity to the DAPI signal is shown. The horizontal line indicates the median value of 100 nuclei. The Kruskal–Wallis test followed by the Steel–Dwass test was performed for statistical analysis. *P < 0.05. Microscopic images are shown in Supplementary Figure S7A. (B) The adipocyte differentiation of 3T3-L1 cells was induced together with the addition of either 100 μM DFO, 5 μM PIK-III, or DMSO. Nuclear fractions on the indicated day of differentiation were subjected to immunoblot analysis using anti-PCBP1, anti-PCBP2, and anti-TATA-binding protein (TBP) antibodies. Data are shown as fold change of the band intensities normalized to TBP (n = 4–6 biological replicates, as indicated in the dot plots). Data are shown as the mean ± s.e.m. One-way ANOVA followed by the Tukey-Kramer test was performed for statistical analysis. *P < 0.05, N.S., not significant. Uncropped immunoblot images are shown in Supplementary Figure S10. (C) ChIP-qPCR was performed using the anti-H3K9me2 antibody in PCBP2-KD cells (sh-Pcbp2) and control cells (sh-Ctrl) on Day 2 (mean ± s.e.m. of three biological replicates) (left). Bisulfite sequence analysis of methylated CpGs in the Pparg2 promoter region were performed in PCBP2-KD cells and control cells on Day 2 (right). The two-tailed Student's t-test was performed for statistical analysis. **P < 0.01, ***P < 0.001. (D) Pparg mRNA levels on Day 2 in 3T3-L1 cell lines with single or double knockdown of PCBP1 and PCBP2, determined by qPCR (n = 3 biological replicates). Pparg mRNA levels normalized to Cyclophilin B level in the cell lines stably expressing shRNA (sh-Pcbp1, sh-Pcbp2, or both) are shown as ratios to the level of the corresponding control cell line (sh-Ctrl). Data are shown as the mean ± s.e.m. One-way ANOVA followed by the Tukey–Kramer test was performed for statistical analysis. *P < 0.05. (E) Differentiated 3T3-L1 cell lines with single or double knockdown of PCBP1 and PCBP2 were stained with ORO on Day 8. Scanned images of the cell culture plate (top) and bright-field microscopy images (bottom) are shown.