Fig 2: Knockdown NuRD complex subunits compromised JGES mediated iPS induction.a Bar plot of numbers for Oct4 GFP positive iPS colonies of MEF cells reprogrammed by JGES or individual factor dropout in iCD3 medium for 7 days. Data are mean ± s.d., two-sided, unpaired t-test; n = 3 independent experiments, ****p < 0.0001. b Principal-component analysis (PCA) of RNA-seq for MEF, JGES, individual factor dropout reprogramming and ESC samples. Solid line with arrows represents one reprogramming sample at sequential time. c Volcano plots of SALL4 significantly enriched proteins of reprogramming samples at day3 performed in triplicate for SALL4 and IgG pull down followed by MS analysis. IP-MS experiments were performed in triplicates and a two-sided t-test was applied. P = 0.05 and fold change = 1.5 were used as threshold. d Bar plot of numbers for Oct4 GFP positive iPS colonies by knockdown NuRD subunits. Data are mean ± s.d., two-sided, unpaired t-test; n = 3 independent experiments, *p < 0.05, **p < 0.01, ***p < 0.001. e PCA analysis of RNA-seq for the roadmap of MEF cell towards ESC by knockdown Gatad2b or Chd4. A zoomed-in snapshot of all day5 and day7 samples was placed on the button of the PCA map. f Heat map showing the differential expressing genes in reprogramming with knocking down Gatad2b and Chd4 in JGES. Genes were clustered into 6 groups according to change pattern. Bar plot showing the gene ontology (GO) analysis of the category. G1-6 indicates group1–6. The data in the heat map were clustered using minisom (a minimalistic and Numpy based implementation of the Self Organizing Maps (SOM)). The p-value in GO results were calculated by hypergeometric test. g Heat map showing the change of chromatin accessibility during reprogramming in knockdown Luciferase, Gatad2b or Chd4 in the reprogramming of JGES. Regions were divided into open-close (OC) state change or close-open (OC) state change with the sequence of occurrence in control. Source data related to Fig. 1a, d are provided as a Source Data file.

Fig 3: MDB3 ChIP-seq experiment in WT and SP140 KO Ramos cells. (A) Profile and tornado plots of MBD3 ChIP-seq signal in wild type (WT) and SP140 KO Ramos cells. (B) Features distribution of MDB3 peaks in WT and SP140 KO Ramos cells. (C) Genomic distribution of MDB3 binding loci relative to TSS in WT and SP140 KO Ramos cells. (D) Venn diagram showing that the absence of SP140 affects MDB3 recognition of 47% of its target genes. (E) Venn diagram showing the intersection between SP140 target genes and MDB3 targets which are lost upon SP140 depletion. (F) EnrichR analysis of genes coregulated by both NuRD and SP140.

Fig 4: SP140 interactome in Daudi cells. (A) Comparison of SP140 abundance by Western blot in nuclear extract, upon sonication or benzonase treatment. (B) Volcano plot showing SP140 interactors under mild conditions and their relative enrichment. Interactors belonging to NuRD complex and PRC2 complex are highlighted. The curve is derived at FDR < 0.05, and s0= 1. (C) Ontology analysis (Enrichr) of the interactors significantly enriched in SP140 immunoprecipitation. (D) Western blot validation of SP140 interactors involved in chromatin regulation. Members of the NuRD (MTA1) and PRC2 complexes (RBBP4, SUZ12 and EED), and DNA binding proteins as CTCF and IKZF3 are specifically enriched by SP140 immunoprecipitation. (E–F) iBAQ values for NuRD and PRC2 components identified by SP140 immunoprecipitation. (G) Volcano plot showing SP140 interactors under stringent conditions and their relative enrichment. Interactors belonging to NuRD complex and PRC2 complex are highlighted. (H) Volcano plot showing SP140 interactors upon crosslinking and their relative enrichment. Interactors belonging to NuRD complex and PRC2 complex are highlighted. (I) Comparison of the intensity ratio (LFQ ratio) of SP140 interactors enriched upon SP140 immunoprecipitation in mild, stringent and crosslinking conditions.

Fig 5: The Dmrta2-Pax6 interaction requires the recruitment of the HDAC-NuRD complex by Zfp423. A, B, Coimmunoprecipitation assays as indicated with Zfp423, Flag-Dmrta2, and a Flag-Dmrta2 mutant lacking the DM domain (Flag-Dmrta2 ΔDM) overexpressed in HEK293T cells as indicated. Note in A that Flag-Dmrta2 pulled down Zpf423 (line 1–2) and that, conversely, Zpf423 coprecipitated Flag-Dmrta2 (line 3–4). Note in B that Flag-Dmrta2 ΔDM does not pull down Zfp423 (line 4). n = 3. Segments from the same blot have been spliced together to show side by side the results of the WT and the ΔDM Dmrta2 mutant. C, Full-length mouse Myc-tagged Zfp423 was synthesized in vitro and incubated with GST alone or with GST fusion proteins bound to glutathione-agarose beads as indicated. Bound Myc-tagged Zfp423 was detected by immunoblot with an anti-Myc antibody. A 2% input sample was loaded for comparison. The corresponding Coomassie-stained gel is shown. n = 2. D, Coimmunoprecipitation assays as indicated with HEK293T cells transfected with a Flag-Dmrta2 expression construct, alone or together with increasing doses of a Zfp423 expression construct. Note that Dmrta2 immunoprecipitates some NuRD subunits and that the binding of Zfp423 to Dmrta2 increases the amount of coimmunoprecipitated HDAC1/2 and MBD3. Densitometric quantification of the western blot results is shown in Extended Data Figure 6-1. n = 3. E, Reporter assays in P19 cells transfected with a Pax6 E60 tk-luc reporter vector, or an “empty” tk-luc reporter vector as indicated, together with a pCS2Myc-Pax6 expression vector and/or a pCS2Flag-Dmrta2 expression vector as indicated, in the absence (white bars) or presence of increasing doses of the HDAC1 inhibitor romidepsin (gray bars). Note that romidepsin leads to a stronger increase of luciferase activity reaching significance in the presence of Dmrta2 but not in its absence. The mean activity of the Pax6 E60 enhancer reporter construct with cotransfected Pax6 is set to 1. NS, not significant. *p < 0.05, one-way ANOVA test. Results of similar reporter assays performed in P19 cells, in the presence or absence of Zfp423 are presented in Extended Data Figure 6-2. F, Reporter assays in HEK293T cells show that both Gal4-Dmrta2 and the Gal4-Dmrta2 (126–531) fusion construct lacking the DM domain required for Zfp423 interaction has strong repression activity on the 5XUAS-tk-luc reporter construct and that Zfp423 slightly increases the repressive activity of Gal4-Dmrta2 but not of the Gal4-Dmrta2 (126–531) construct. In each condition, 200 ng of the 5XUAS-tk-luc reporter was transfected, together with 25 ng of the pCMV-Gal4-Dmrta2 or the pCMV-Gal4-Dmrta2 (126–531) and different doses (200, 400 and 600 ng) of pCDNA3-Myc-Zfp423 expression plasmids. Values represent the mean ± SD of one transfection done in triplicate. A Western blot showing the expression levels of the overexpressed factors is shown below. Reporter assays showing that the Gal4-Dmrta2 fusion protein represses in a UAS-dependent manner the activity of the 5XUAS-tk-luc reporter are presented in Extended Data Figure 6-3. Reporter assays in HEK293T cells showing that Zfp423 does not increase the modest repression observed when an expression vector encoding the Gal4 DNA-binding domain alone is cotransfected with the 5XUAS-tk-luc reporter are presented in Extended Data Figure 6-4.