Co-repressor CBFA2T2 regulates pluripotency and germline development

Abstract

Developmental specification of germ cells lies at the heart of inheritance, as germ cells contain all of the genetic and epigenetic information transmitted between generations. The critical developmental event distinguishing germline from somatic lineages is the differentiation of primordial germ cells (PGCs), precursors of sex-specific gametes that produce an entire organism upon fertilization. Germ cells toggle between uni- and pluripotent states as they exhibit their own 'latent' form of pluripotency. For example, PGCs express a number of transcription factors in common with embryonic stem (ES) cells, including OCT4 (encoded by Pou5f1), SOX2, NANOG and PRDM14 (refs 2, 3, 4). A biochemical mechanism by which these transcription factors converge on chromatin to produce the dramatic rearrangements underlying ES-cell- and PGC-specific transcriptional programs remains poorly understood. Here we identify a novel co-repressor protein, CBFA2T2, that regulates pluripotency and germline specification in mice. Cbfa2t2(-/-) mice display severe defects in PGC maturation and epigenetic reprogramming. CBFA2T2 forms a biochemical complex with PRDM14, a germline-specific transcription factor. Mechanistically, CBFA2T2 oligomerizes to form a scaffold upon which PRDM14 and OCT4 are stabilized on chromatin. Thus, in contrast to the traditional 'passenger' role of a co-repressor, CBFA2T2 functions synergistically with transcription factors at the crossroads of the fundamental developmental plasticity between uni- and pluripotency.

Extended Data Figure 1. Biochemical interaction between PRDM14 and CBFA2T2

a, Mass spectrometry peptide counts from FLAG affinity purification from NCCIT control cells and stable lines expressing FLAG-HA-PRDM14 (PRDM14-F), and FLAG-HA-CBFA2T2 (CBFA2T2-F). b, Characterization of in-house human PRDM14 antibody. Western blot performed using 30 µg of NCCIT and KH2 mESC cell lysate. Human PRDM14 antibody is specific and does not cross-react with mouse PRDM14. c, Immunoprecipitation using antibodies against the indicated endogenous proteins in mESC cells. d, Western blot of Superose 6 column fractionation of FLAG-purified CBFA2T2 complex in NCCIT cells stably expressing FLAG-HA-CBFA2T2. e, Chromatin immunoprecipitation analysis using the indicated antibodies in 293T-REx harboring a UAS-TK-Luciferase transgene. Fold enrichment represents the ratio of enrichment by ChIP-qPCR upon induction of GAL4-PRDM14 expression via doxycycline addition. Positions of the primer set are indicated by small arrows in the schematic. qPCR source data are included in SI Source Data Table3.

Extended Data Figure 2. PRDM14 and CBFA2T2 exhibit an overlapping and interdependent distribution on chromatin in NCCIT cells

a, Heatmap depicting PRDM14, CBFA2T2, RING1B and SUZ12 read density across a 5kb window centered about the PRDM14 (top) or RING1B (bottom) SERs identified in NCCIT cells. b, Representative genome browser tracks depicting SERs at the indicated genomic loci. c, Gene Ontology analysis of PRDM14/CBFA2T2 common target genes. d, Western blot analysis of PRDM14 and CBFA2T2 protein levels in knockdown experiments (Fig. 1d, 1e).

Extended Data Figure 3. Characterization of KO ESC mutants and quantification of human iPSC reprogramming efficiency

a,b, Strategy for generating Prdm14 and Cbfa2t2 knockout mESCs via CRISPR/Cas9 genome-editing. Sequencing chromatograms confirming homozygous disruption of the locus are depicted. c, Cbfa2t2 and Prdm14 KO ESCs require 2i to maintain growth. ESC lines generated under FBS+LIF+2i conditions were continuously cultured in FBS+LIF+2i (top, middle), or switched to FBS+LIF (bottom). Eight days after 2i withdrawal (FBS+LIF), well-formed ESC colonies were undetectable; instead, mutant ESCs appeared to be differentiated. Scale bar: 100 µm. d, Proliferation rates of WT and mutant KO ESCs as sescibed in c. Data were obtained from 3 biological replicates. Please note error bars were shown in the plots. Due to the logarithmic scale used here, some error bars are very small and might be invisible. e, RNA-seq MA plot in the indicated ESCs. Data is representative of three biological replicate experiments for each line. Mean abundance is plotted on the x-axis and enrichment (both in log 2 scale) is plotted on the y-axis. Genes depicted in red are differentially expressed with a FDR < 0.0001. f, Heatmap showing relative expression of all differentially expressed genes as sescribed in Fig. 2c. Only difference is now the heatmap is centered on CBFA2T2 differentially expressed genes, rather than PRDM14 DEGs. g, Scheme of human fibroblast reprogramming to induced pluripotent stem cells. Fibroblasts were transduced with lentiviruses expressing polycistronic Oct4/Klf4/Sox2/Myc (OKSM) and either Prdm14 or Cbfa2t2. Three weeks later, bright-field images of successfully reprogrammed colonies (left) and live TRA-1-81 staining (right) were recorded. Scale bar: 500 µm. h, Quantification of human iPSC reprogramming efficiency based on TRA-1-81 staining with secondary antibody conjugated with peroxidase HRP and substrate DAB. Error bars are based on 4 biological replicates of each condition. The source data is included in SI Source Data Table 4.

Extended Data Figure 4. Cbfa2t2^−/− mouse genotypes and sperm defects

a, One representative Cbfa2t2^−/− mouse genotype wherein a 7 bp fragment is deleted. b, Testes of multiple WT (n=4) and Cbfa2t2^−/− (n=4) male mice at 8 weeks old were dissected and weighed. c, Number of sperm in the epididymis of Cbfa2t2^+/+ (n=4) and Cbfa2t2^−/− (n=4) mice is shown with SEM. p-value was determined by student’s t-test. d, Near loss of gonocytes in Cbfa2t2 KO mutant P0 testes by DDX4 (MVH) staining. Visualization of MVH (red) positive gonocytes in Cbfa2t2^+/− (upper panels) or Cbfa2t2^−/− (lower panels) testis at P0 stage. The merged images with Hoechst (left panels; white) are shown on the right panels. Scale bars, 100 mm. e, Numbers of AP2γ–positive PGCs in Cbfa2t2^+/+ (black), Cbfa2t2^+/− (gray) and Cbfa2t2^−/−(red) embryos at the indicated embryonic stages. LS, late-streak stage; EB, MB, and LB, early-, mid-, and late-bud stage; EHF, early-head fold stage; 2 st., 2 somites stage. t-test: *p=0.03, **p=0.003. f, Numbers of AP2γ–positive PGCs in Cbfa2t2^+/+ (black), Cbfa2t2^+/− (gray) and Cbfa2t2^−/−(red) embryos at the indicated embryonic stages. LS, late-streak stage; 0B, EB, MB, and LB, zero-, early-, mid-, and late-bud stage; EHF and LHF, early- and late-head fold stage; 2 st., 2 somites stage. g, Left panel, expression of SOX2 (red) in AP2γ (green)-positive PGCs in Cbfa2t2^+/+ (upper panels) or Cbfa2t2^−/− (lower panels) embryo at mid-bud (MB) stage, E7.25, shown as z-projection images of posterial confocal sections. Arrow indicates a minor PGC with relatively normal activation of SOX2. Scale bar, 50 mm. Right panel, %SOX2 (red)-positive cells in AP2γ (green)-positive PGCs in the indicated genotypes of Cbfa2t2 at E7.0-7.25 (zero- to mid-bud stage) are shown with statistical significance (t-test: *p=0.0006, **p=0.0001; Cbfa2t2^+/+, n=7; Cbfa2t2^+/−, n=5; Cbfa2t2^−/−, n=5).

Extended Data Figure 5. Cbfa2t2 m7 mutant characterization and the related mechanism

a, Gene Ontology analysis of PRDM14 ChIP-seq target genes. PRDM14 target genes are enriched in histone methyltransferase activities by DAVID analysis. b, Cbfa2t2 m7 mutant genotyping. The mutant 7 amino acids are in red and corresponding WT residues are highlighted in blue in the displayed protein sequences. c, Bright-field images of WT and Cbfa2t2-m7 mESCs. Scale bar: 100 µm. d, Western Blot analysis of PRDM14, CBFA2T2, and OCT4 protein levels in Prdm14 KO, Cbfa2t2 KO or m7 mutant ESCs under feeder-free FBS+LIF+2i condition. Nonspecific bands are denoted with an asterisk. e, ChIP-qPCR using antibodies against PRDM14, CBFA2T2, or OCT4 at selected Oct4 target genes. Occupancy is compared between WT, Cbfa2t2-KO and Cbfa2t2-m7 mESCs. ChIP-qPCR primer sequences are included in Supplementary Table 3. f, RT-qPCR quantification of Ehmt1 mRNA levels in WT and mutant lines. P values are 0.004 (**) and 0.0142 (*).The source data is included in SI Source Data Table 5. g, Mass spectrometry quantification of histone H3K9 modifications in WT and mutant lines. p values are 0.00956, 0.04248 (*). The source data is included in SI Source Data Table 5. h, i, additional Immunofluorescence analysis of H3K9me2 (red) of AP2γ-positive (green; arrowheads) PGCs in Cbfa2t2^+/− and ^−/− embryos at E8.75 as described in Fig 4i.

Figure 1. PRDM14 and the co-repressor protein CBFA2T2 interact and bind to chromatin inter-dependently

a, Immunoprecipitation using antibodies against the indicated endogenous proteins in NCCIT cells. For all Western blots, source gel data are included in SI Figure 1. b, Venn Diagram depicting the overlap of PRDM14 and CBFA2T2 target genes as identified by ChIP-seq. c, Genome browser tracks showing PRDM14 and CBFA2T2 at their respective genomic loci. d, e, ChIP-qPCR at SERs found near the 11 indicated genes in NCCIT cells with siRNAs against CBFA2T2 (e) or short hairpin RNAs against PRDM14 (f) (n=3 biological replicates). Error bars, s.d. qPCR source data are included in SI Source Data Table 1.

Figure 2. PRDM14 and CBFA2T2 regulate pluripotency

a, Western blots confirming loss of PRDM14 or CBFA2T2 expression in KO mESCs. Nonspecific bands are denoted with an asterisk. b, Venn diagram depicting the overlap of differentially expressed genes upon deletion of Prdm14 or Cbfa2t2 after removal of feeders from FBS+LIF+feeders culture. c, Heatmap showing relative expression of all differentially expressed genes identified with a false discovery rate less than 1E-3 between WT and either Prdm14 or Cbfa2t2 KO mESCs.

Figure 3. Cbfa2t2^−/− mice are defective in their germline

a, Schematic of Cbfa2t2 KO mouse generation by CRISPR zygotic injection. b and d, Image of dissected ovaries (n=8) and testes (n=4), respectively, in Cbfa2t2 KO mice. Scale bars, 1 mm. c and e, Histological sections of ovaries and testes, respectively, stained by H&E. Scale bars, 100 µm. f. Genital ridges of Cbfa2t2^+/+ and Cbfa2t2^−/− embryos at E11.5 stained by alkaline phosphatase (AP). Scale bar, 1mm. g, AP staining of PGCs of E8.75 (9 somites) embryos is shown (n=3). Arrowheads point to the boundary of the developing hindgut. pm, para-axial mesoderm. Scale bar, 100 µm. PGC numbers in each embryo were plotted in the right panel, with the following values: +/+, 93±5; +/−, 87±5; −/−, 38±1.

Figure 4. Mechanism of CBFA2T2/PRDM14 complex chromatin binding and direct regulation of PGC epigenetic reprogramming

a, Heatmap depicting CBFA2T2, PRDM14 and OCT4, SOX2, NANOG ChIP-seq read density centered about the top 299 CBFA2T2 SERs in mESCs. b, Representative genome browser tracks at the indicated Ehmt1 locus in mESCs. c, Domain annotation of WT, Cbfa2t2 KO and oligomerization mutant m7 proteins. The 7 amino acids mutated in Cbfa2t2-m7 are depicted as lines within NHR2. d, Immunoprecipitation against the indicated proteins in Cbfa2t2-m7 mESCs followed by Western blot. e, f, g, ChIP-qPCR using antibodies directed against CBFA2T2 (e), PRDM14 (f) or OCT4 (g) at SERs found near the indicated genes (n=3). Error bars, s.d. qPCR source data are included in SI Source Data Table 2. h, GLP (EHMT1) expression (red) in AP2γ-positive PGCs (green, arrowheads) in Cbfa2t2^+/− and ^−/− embryos at E8.0, Late Head-fold (LHF) stage. i. Immunofluorescence analysis of H3K9me2 (red) of AP2γ-positive (green; arrowheads) PGCs in Cbfa2t2^+/− and ^−/− embryos at E8.75. Line plot analysis on yellow arrowed-area are shown on the right. Scale bars, 10 µm. Data are representative of 3 indepenent expeiments. j. Model depicting the co-repressor CBFA2T2 oligomerizing to stabilize associated TFs (PRDM14 and OCT4) on chromatin.