JMJD3 acts in tandem with KLF4 to facilitate reprogramming to pluripotency

The interplay between the Yamanaka factors (OCT4, SOX2, KLF4 and c-MYC) and transcriptional/epigenetic co-regulators in somatic cell reprogramming is incompletely understood. Here, we demonstrate that the histone H3 lysine 27 trimethylation (H3K27me3) demethylase JMJD3 plays conflicting roles in mouse reprogramming. On one side, JMJD3 induces the pro-senescence factor Ink4a and degrades the pluripotency regulator PHF20 in a reprogramming factor-independent manner. On the other side, JMJD3 is specifically recruited by KLF4 to reduce H3K27me3 at both enhancers and promoters of epithelial and pluripotency genes. JMJD3 also promotes enhancer-promoter looping through the cohesin loading factor NIPBL and ultimately transcriptional elongation. This competition of forces can be shifted towards improved reprogramming by using early passage fibroblasts or boosting JMJD3’s catalytic activity with vitamin C. Our work, thus, establishes a multifaceted role for JMJD3, placing it as a key partner of KLF4 and a scaffold that assists chromatin interactions and activates gene transcription.

(a) RT-qPCR for Jmjd3 and Utx in P2 MEFs, ESCs (OG2 ESCs) and an OSKM-reprogramming time course. (b) Western blotting for FLAG showing exogenous FLAG-tagged JMJD3 or UTX expression in OSKM-reprogrammed P2 MEFs. ACTIN serves as the loading control. (c) Growth curves of OSKM-reprogrammed MEFs with Empty or JMJD3 in medium with or without Vc at the indicated time points. (d) Number of Oct4-GFP + and Dppa5a-dTomato + colonies (left panel) and percentage of positive cells for each, measured by flow cytometry (right panel) in OSKM-reprogrammed dual-reporter P2 MEFs with Empty or JMJD3. (e,f) Silencing of the OSKM transgenes (e) and activation of endogenous pluripotency genes (Oct4 and Nanog; f) in iPSC clones produced with exogenous JMJD3. MEFs were used as negative control and OSKM-D6 (cells at day 6 of OSKM reprogramming; in e) and ESCs (OG2 ESCs; in f) as positive controls. Data were plotted from one experiment with three technical replicates. (g,h) Phase contrast, Oct4-GFP fluorescence, NANOG immunofluorescence, chimeric mice, germline transmission (g) and karyotype analysis (h) of JMJD3-C5 iPSCs. Scale bars, 50 μm.
(i) Number of AP + and GFP + colonies in OSKM-reprogrammed P2 MEFs with Empty or UTX at day 16 (without Vc, left panel) and day 11 (with Vc, right panel).
(j) Number of GFP + colonies at day 7 in OSKM-reprogrammed P2 MEFs with Empty or JMJD3 in iSF1 medium.
(k) Flow cytometry analysis of cell surface markers CD44 and ICAM1 in OSKM-reprogrammed P2 MEFs with Empty or JMJD3 at days 5 (early) and 10 (late), with MEFs and ESCs (OG2 ESCs) as controls.
Error bars represent the s.e.m.. Data are presented as mean ± s.e.m. from n=3 (a,c,d,i-k) biologically independent experiments. Statistical analyses were performed using a two-tailed unpaired Student's t-test (*P < 0.05; **P < 0.01; ***P < 0.001  Figure 2 The effect of JMJD3 on PHF20 during reprogramming. (a) Western blotting for PHF20 and p16 INK4A in P2 MEFs transduced with Empty or JMJD3 in MEF medium with or without Vc (left panels), and for PHF20 in MEFs transduced with Empty or OSKM with or without JMJD3 in reprogramming medium with Vc (right panels). The quantification and statistics for PHF20 (normalized to ACTIN) in three independent experiments is shown. (b) Western blotting for FLAG-tagged JMJD3 and PHF20 using FLAG antibody. (c) Number of GFP + colonies in OSKM-reprogrammed P2 MEFs with the indicated genes. All error bars throughout the figure represent the standard error of mean (s.e.m.). Data are presented as mean ± s.e.m. from n=3 (a,c) biologically independent experiments. Statistical analyses were performed using a two-tailed unpaired Student's t-test (a), or one-way ANOVA followed by a Holm-Sidak multiple comparison test (c) (*P < 0.05; **P < 0.01; ***P < 0.001). P values: 0.0005, 0.0007, 0.0018, 0.0102 (a); 0.0022, 0.0392, 0.0001, 0.0013 (JMJD3+PHF20 vs JMJD3), 0.0002 (JMJD3+PHF20 vs PHF20) (c). Experiments were performed once (b). Source data are provided as a Source Data file. (a) Western blotting, quantification and statistics for H3K27me3 in MEFs, OSKM reprogramming MEFs (day 5 and day 10), pre-iPSCs and OSK-iPSCs. (b) Immunofluorescence showing the effect of exogenous JMJD3 compared to Empty on H3K27me3 levels in OSKM-reprogrammed P2 MEFs. Scale bar, 20 μm. Arrows mark cells without exogenous JMJD3 and arrowheads mark cells with exogenous JMJD3.

Supplementary
Error bars represent the s.e.m.. Data are presented as mean ± s.e.m. from n=3 (a) biologically independent experiments. Statistical analyses were performed using a two-tailed unpaired Student's t-test (*P < 0.05). P values: 0.0225, 0.0328 (a). Experiments were repeated independently three times (b) with similar results. Source data are provided as a Source Data file.          (a) Number and percentage of significantly (q value < 0.1, fold change > 1.5) upregulated and downregulated genes at days 5 and 10 upon JMJD3 OE or knockdown. (b,c) Venn diagram analysis of genes significantly upregulated upon JMJD3 OE at days 5 and 10 (b), and genes significantly upregulated upon JMJD3 OE and downregulated by shJmjd3 at days 5 and 10 (c  (a) Schematic representation of the ChIP-seq workflow with reference genome normalization due to the global change of H3K27me3 by JMJD3 OE. S2 cells from Drosophila melanogaster (DM) were used as reference. (b) Venn diagram analysis of H3K27me3 peaks downregulated upon JMJD3 OE (relative to empty vector) in WT and Jmjd3 KO P2 MEFs transduced with OSKM in medium with Vc at days 5 and 10. (c) Genome distribution of H3K27me3 peaks. (d) Analysis of the dataset (GSE90895) showing violin plots of the levels of H3K27me3 (upper panels) and H3K9me3 (middle panels), and the dataset (GSE106525, GSE112520 and GSE56986) showing violin plots of the level of DNA methylation (lower panels) at TSS and ESC-enhancers of PSC-enriched genes upregulated upon JMJD3 OE from day 5 and day 10, respectively, in the indicated samples. (e-g) ChIP-seq data (e,f) and ChIP-qPCR confirmation (g) for H3K27me3 at epithelial (Cdh1 and Epcam) and pluripotency (Sall4, Tdh and Sall1) genes, and the Ink4a/Arf locus in P2 MEFs transduced with OSKM and Empty or JMJD3 at days 5 and 10. MED1 peaks in ESCs (from GSE22562) were used as reference to indicate potential enhancer sites. SE, super-enhancer; TE, typical enhancer. Shaded regions represent TSS or enhancer. Error bars represent the s.e.m.. Data are presented as mean ± s.e.m. from n=3 (g) biologically independent experiments. Statistical analyses were performed using a two-tailed unpaired Student's t-test (*P < 0.05; **P < 0.01; ***P < 0.001).   Figure 7 H3K27me3 demethylation by JMJD3 at pluripotency loci.

Supplementary Figure 4
(a) Analysis of our H3K27me3 and H3K27ac ChIP-seq dataset in OSKM-reprogrammed P2 MEFs with Empty or JMJD3 at day 5 and/or day 10, and analysis of published datasets for H3K27me3 and H3K9me3 in MEFs (GSE90895) and DNA methylation in MEFs, iPSCs and ESCs (GSE106525, GSE112520 and GSE56986), at the indicated pluripotency gene loci. SE, super-enhancer; TE, typical enhancer. Shaded regions represent TSS or enhancer. (b) De novo DNA sequence motifs of many pluripotency factors detected in H3K27me3 demethylated regions by HOMER. (c) Correlation between H3K27me3 demethylation induced by JMJD3 OE and the binding of the indicated pluripotency factors in ESCs (from GSE11431, GSE19019 and GSE25409). Binding density of the indicated factors was measured using a moving window.  (a) ChIP-qPCR for H3K27me3 at the indicated loci in P2 MEFs reprogrammed with OSM and low or high level of KLF4 plus Empty or JMJD3. (b) RT-qPCR for epithelial genes (Cdh1, Epcam and Ocln) and Ink4a at day 5 (left panels), and pluripotency genes (Sall4, Tdh, Nanog and Utf1) at day 10 (right panels) in P2 MEFs reprogrammed with OSM and no KLF4, low KLF4, or high KLF4 plus Empty or JMJD3. (c) Western blotting for E-cadherin, KLF4 (day 5) and NANOG (day 10) in P2 MEFs transduced with OSM and low or high level of KLF4 plus Empty or JMJD3. (d) Number of GFP + colonies at day 19 (without Vc) and day 14 (with Vc) in OSK-reprogrammed P2 MEFs with Empty or JMJD3. (e) RT-qPCR for epithelial genes (Cdh1, Epcam and Ocln) and Ink4a in P2 MEFs transduced with KLF4 and JMJD3, either alone or in combination, at day 5. (f) RT-qPCR for epithelial genes (Cdh1, Epcam and Ocln) in P2 MEFs transduced with KLF4 and shLuc or shJmjd3 at day 5.
(g) ChIP-qPCR for H3K27me3 at the Ink4a/Arf locus in P2 MEFs transduced with KLF4 and JMJD3, either alone or in combination, at day 5 in medium with Vc.