Sex-specific chromatin remodelling safeguards transcription in germ cells

Sex-specific chromatin remodelling safeguards transcription in germ cells 1 2 Tien-Chi Huang1,2, Yi-Fang Wang1*, Eric Vazquez-Ferrer1,2*, Cristina E. Requena1,2, Courtney 3 Hanna3,4, Gavin Kelsey,3,4 and Petra Hajkova1,2 † 4 5 1MRC London Institute of Medical Sciences (LMS), Du Cane Road, London W12 0NN, UK. 6 2Institute of Clinical Sciences (ICS), Faculty of Medicine, Imperial College London, Du Cane 7 Road, London, W12 0NN, UK 8 3Epigenetics Programme, Babraham Institute, Cambridge, UK 9 4Centre for Trophoblast Research, University of Cambridge, Cambridge, UK 10 11 *these authors contributed equally 12 13 †Correspondence: [email protected] 14 Tel: +44 (0)20 83838264 15 16 17 18 19 Abstract: 20 Stability of the epigenetic landscape underpins maintenance of the cell type specific transcriptional 21 profile. DNA methylation as one of the main repressive epigenetic systems has been shown to be 22 important for long term gene silencing; its loss leading to ectopic and aberrant transcription in 23 differentiated cells and cancer1. Interestingly, the developing mouse germ line endures global 24 changes in DNA methylation in the absence of widespread transcriptional activation. Using an ultra-25 low input native (ULI-n) ChIP approach we show that following DNA demethylation the gonadal 26 primordial germ cells (PGCs) undergo remodelling of repressive histone modifications resulting in a 27 sex specific signature. We further demonstrate that polycomb plays a central role in transcriptional 28 control in the newly hypomethylated germline genome as the genetic loss of Ezh2 leads to aberrant 29 transcriptional activation, retrotransposon derepression and dramatic loss of developing female 30 germ cells. Last but not least, we show that the base composition of promoters determines 31 H3K27me3 enrichment following the loss of DNA methylation. Overall, our study provides 32 unprecedented insight into the dynamic interplay between repressive chromatin modifications in 33 the context of a developmental reprogramming system. 34 35 36 DNA methylation is functionally associated with transcriptional silencing of tissue-specific genes and 37 repression of transposable elements (TEs). Although globally relatively stable in the soma, 38 development of the mammalian germ line is associated with profound changes in genomic DNA 39 methylation 2-4. Following the specification of nascent primordial germ cells (PGCs) at embryonic day 40 E6.25 in the mouse, global changes in the germline epigenome encompass both reprogramming of 41 histone modifications as well as erosion of genome-wide DNA methylation 5,6. Global DNA 42 methylation drops gradually during germ cell migration and further decreases extensively between 43 E10.5 and E12.5 once PGCs have entered the genital ridge, installing the most hypomethylated state 44 ever observed during normal development 7. The gonadal reprogramming directly precedes sex 45 differentiation followed by the onset of meiosis in female PGCs and proliferation arrest in the male 46 germ line 8 (Fig. 1a). 47 Several lines of evidence indicate that the loss of DNA methylation and TET1-mediated 48 transcriptional activation are crucial for the initiation of the expression of key germline genes at this 49 developmental stage 9. However, the molecular mechanisms that enable germ cells to maintain and 50 co-ordinate the progressive gene activation and prevent precocious or aberrant wide-spread 51 transcription that could be triggered by genome-wide DNA demethylation are unclear. 52 We have previously shown that the global loss of DNA methylation in the gonadal PGCs temporally 53 overlaps with other major changes in chromatin structure and global histone modifications 6. In 54 order to understand the temporal changes and to address whether remodeling of the 55 heterochromatin silencing could functionally compensate for the global loss of DNA methylation to 56 provide the means of transcriptional control at this critical stage of PGC development we assessed 57 H3K27me3 and H3K9me3 using an ultra-low input native chromatin immunoprecipitation (ULI-58 nChIP-seq) 10,11. The analysis was carried out using 1000 PGCs isolated from E10.5 (pre-gonadal 59 reprogramming) and E13.5 (post-gonadal reprogramming) male and female embryos derived from 60 C57BL/6 X ΔPE-Oct4-GFP crosses 12 (Extended Data Fig.1a, 1b,1c). Similar to previous reports in ES 61 cells 13, more than 50% of H3K27me3 peaks overlapped with promoters and gene body (Fig. 1b) with 62 30% of H3K27me3 peaks located within 1kb of the transcription start sites (TSS) (Extended Data Fig. 63 1d, 1e ). In contrast, 75% of H3K9me3 peaks localised to intergenic regions, remote from the 64 annotated TSSs (Fig. 1b, Extended Data Fig. 1d, 1e) with clear peak enrichment at repetitive 65 sequences (58% at E10.5, 64% in E13.5 female and 68% for E13.5 male, respectively) (Extended Data 66 Fig 1f). 67 68 Our ULI-nChIP Seq analysis revealed relatively low correlation between E13.5 male and female PGCs 69 for both H3K27me3 and H3K9me3 documenting sex specific differences at this stage of germline 70 development (Spearman correlation coefficient 0.43 for H3K27me3, 0.5 for H3K9me3) (Fig 1c). 71 Furthermore, for both H3K9me3 and H3K27me3 the correlation between E10.5 and E13.5 PGC 72 samples is even lower than it is between sexes suggesting that the repressive chromatin is globally 73 remodelled between E10.5 and E13.5 alongside the global wave of DNA demethylation (Fig. 1c). In 74 further support of this notion, the peak number of H3K27me3 was greatly reduced from E10.5 to 75 E13.5 in both male and female germ cells (Fig. 1d). Although the loss of H3K27me3 peaks is more 76 pronounced in male PGCs (6933 peaks in the female PGCs vs 2740 identified peaks in the male PGCs), 77 even female cells had only 40% of the peaks persisting from E10.5 with the majority of peaks being 78 newly generated during the reprogramming process (Fig. 1d). Sex specific re-configuration of the 79 signal is also evident upon analysis of the H3K9me3 signal. Despite the comparable number of 80 enriched loci, H3K9me3 seems to be remodelled in female PGCs between E10.5 and E13.5; while the 81 male specific loss of H3K27me3 peaks described above seems to be compensated by a dramatic (2.7 82 fold) increase in the number of H3K9me3 peaks observed in male E13.5 PGCs (Fig. 1e). Of note, the 83 majority of male H3K9me3 peaks are relatively short, compared with peaks observed at E10.5 84 (Extended Data Fig. 1h). Despite a significant change in peaks number, the average ChIP intensity 85 relative to input within peaks are comparable between stages (Extended Data Fig. 1g). Furthermore, 86 the observed sex specific remodelling of heterochromatin is not underpinned by profound changes 87 in the abundance of relevant histone modifications, or differential expression of the relevant histone 88 methyltransferases (Fig. 1f, Supplementary Table 3); but potentially linked to a difference in 89 targeting of histone modifying complexes. Indeed, we observed sex specific expression of the 90 polycomb repressive complex 2 (PRC2) auxiliary subunits AEBP2 and PHF19, known to be implicated 91 in polycomb targeting (Extended Fig. 6a, 6b)14,15 . Collectively, our analysis shows that global loss of 92 DNA methylation in gonadal germ cells coincides with genome-wide reprograming of repressive 93 histone modifications, resulting in a sex specific heterochromatin configuration in PGCs at E13.5. 94 95 H3K27me3 is catalysed by PRC2 that shows clear preference for binding to unmethylated CG-rich 96 sequences in mammalian cells16 . We thus asked whether the H3K27me3 remodelling observed in 97 PGCs is due to PRC2 recruitment in response to the global loss of DNA methylation. Our whole 98 genome bisulphite sequencing (WGBS) datasets 9 show that promoter methylation (5mC+5hmC) 99 significantly decreases from E10.5 to E12.5 in both sexes (Extended Data Fig. 2a). However, this does 100 not lead to an overall global increase in H3K27me3 deposition (Extended Data Fig. 2b, 2c). 101 To understand the potential relationship better, we categorised promoters based on the pattern of 102 dynamic changes between DNA methylation and H3K27me3 from E10.5 to E13.5 (See Methods). We 103 observed 4 different patterns: Group A: loss of 5mC, gain of H3K27me3; Group B: loss of 5mC, 104 continuous H3K27me3 enrichment; Group C: devoid of 5mC, continuous high level of H3K27me3; 105 Group D: loss of 5mC, devoid of H3K27me3 (Fig. 2a, Extended Data Fig 3). Of note, all promoters 106 showed general H3K9me3 reduction between E10.5 and E13.5 alongside global DNA demethylation 107 (Extended Data Fig 3a). Promoters showing consistently high H3K27me3 enrichment (Group C) are 108 characterised by high CpG density, high C+G content and lack of DNA methylation at E10.5 (Fig. 2b), 109 consistent with the known characteristics of polycomb targets in mammalian cells16. Promoters with 110 lower CpG density contain DNA methylation at E10.5 (Groups A, B and D); however, following 111 methylation loss, CpG density and G+C content determine the polycomb recruitment. As an 112 example, both group A and group D lose DNA methylation, but compared with Group A that attracts 113 H3K27me3 (CpG obs/exp ratio=0.27, GC percentage =46.6%), Group D has significantly lower 114 frequency of CpG (CpG obs/exp ratio=0.22) and G+C (44.8%) (P<2X10-16, pairwise Wilcox test) (Fig. 115 2b). This suggests that critical threshold of CpG density and G+C enrichment is required to render 116 promoters capable of recruiting PRC2 once DNA methylation is removed (Fig. 2c). Interestingly, this 117 relationship is less obvious in the male E13.5 PGCs (Extended Data Fig. 2d, 2e), suggesting a key role 118 for PRC2 subunits that are differentially expressed between sexes in the targeting of polycomb. 119 Promoters in group B have medium to low levels of DNA methylation at E10.5 and are enriched for 120 H3K27me3 even before gonadal DNA demethylation. This group thus likely represents an 121 intermediate state. Similar to group A, group B has higher CpG density and higher GC content, 122 compared with group D (P<2X10-16, pairwise Wilcox test), but higher GC enrichment than group A. 123 Interestingly, promoters in group A and B are enriched for germline-specific genes, such as Stra8, 124 Sycp1, Sycp2, Dnmt3l and Tex101 (Extended Data Fig. 3a), whereas promoters in the group C 125 generally are linked to development and lineage-specific regulators (Extended Data Fig. 3a). 126 In order to functionally validate the importance of the observed dynamic changes in H3K27me3, we 127 depleted H3K27me3 in early PGCs using Blimp1-Cre (Prdm1) driven Ezh2 deletion 17. (Extended Data 128 Fig. 4b, 4c, 4d). EZH2 is highly expressed throughout the early germline development, compared 129 with surrounding somatic cells (Extended Data Fig. 4a). The expression of Blimp1-Cre begins at E7.0, 130 shortly after PGC specification, allowing us to deplete EZH2 before gonadal reprogramming occurs17. 131 Consistent with the reported high recombination efficiency of Blimp1-Cre18, EZH2 protein was absent 132 in gonadal Ezh2 Δ/Δ, Blimp1-Cre PGCs (Fig. 3a) (Ezh2 conditional knock out is henceforth referred to 133 as Ezh2 CKO). This coincided with the loss of H3K27me3 in Ezh2 CKO PGCs, confirming that EZH2 is 134 the main enzyme responsible for catalysing H3K27me3 at this stage of PGC development (Fig. 3b, 135 Extended Data Fig. 6c). Of note, the loss of H3K27me3 did not affect global levels of H2A119ub or 136 H3K9me3, as evaluated by immunofluorescence (Extended Data Fig. 5a, 5b), or the global erasure of 137 DNA demethylation that occurs in gonadal PGCs (Extended Data Fig. 5c, 5d, 5e). 138 Although the observed loss of H3K27me3 did not have any apparent effect on the germ cell numbers 139 in both sexes at E13.5 (Extended Data Fig. 5f), RNA-Seq analysis showed clear separation by sex and 140 genotype (Ctrl vs CKO Ezh2-/- PGCs) (Fig. 3c, Extended Data Fig. 6d). As anticipated based on our 141 H3K27me3 ChIP-Seq analysis (Fig. 1d), loss of EZH2 had a more profound impact on the 142 transcriptome of female PGCs (Fig. 3c, 3d) with 679 and 290 genes differentially expressed in female 143 and male Ezh2-/- PGCs, respectively (adj. P<0.05 FC>2). Consistent with the role of PRC2 as a 144 transcriptional repressor, more than 80% of differentially expressed genes in male (239) and female 145 (565) were upregulated (Fig. 3d) with clear enrichment for cell differentiation, developmental 146 processes and gamete generation using Gene Ontology (GO) analysis (Fig. 3e, Extended Data Fig. 7a). 147 Importantly, loss of EZH2 affected specifically genes with promoters enriched in H3K27me3 in 148 control E13.5 PGCs (Extended Data Fig. 3a, 3b), documenting that following DNA demethylation, 149 promoter H3K27me3 is required for maintaining correct transcriptional regulation of these genes in 150 the developing germ line. 151 Amongst the genes upregulated in female E13.5 Ezh2 CKO PGCs, we noticed an enrichment of genes 152 associated with reproduction and meiosis; out of 104 previously identified genes associated with 153 female meiotic prophase 19, 31 genes were differentially expressed following Ezh2 deletion (P=4X10154 14) (Fig. 3f, Extended Data Fig. 7c). Similarly, depletion of EZH2 led to the transcriptional 155 derepression of 23 of the previously identified germline reprogramming-responsive (GRR) genes 156 (Extended Data Fig. 7g) 9. H3K27me3 increased on the promoters of many meiosis-related genes 157 following the loss of DNA methylation between E10.5 to E13.5 (group A promoters), suggesting that 158 H3K27me3 deposition is required to modulate the expression of germline-specific, DNA methylation-159 sensitive genes in the hypomethylated germ line (Extended Data Fig. 3a, 3b,Extended Data Fig 7d). 160 Although H3K27me3 enrichment on promoters of meiotic genes is observed in both male and 161 female PGCs (Fig. 3f, Extended Data Fig 7d), female PGCs show a much more profound 162 transcriptional response following EZH2 loss (Extended Data Fig. 7c). We reasoned that the presence 163 of signalling-induced, sex-specific transcription factors might be required to drive ectopic or 164 precocious transcription once the functional repressive chromatin mark is removed. To this end, the 165 presence of retinoic acid (RA) is essential for the transcriptional activation of Stra8 and entry into 166 meiosis in the female germline. In contrast, RA is degraded in male gonads leading to diversion from 167 the meiotic programme 20. In agreement with our hypothesis, we found that RA-related transcription 168 factor binding motifs, such as RARB, RXRA and RARG are enriched in genes upregulated in female, 169 but not in male E13.5 Ezh2 CKO PGCs. (Extended Data Fig. 7e). Furthermore, as expected, the 170 expression levels of these transcription factors are higher in the female germ cells than in the male 171 counterparts at E13.5 (Extended Data Fig. 7f). Taken together, RA signalling and the related 172 transcription factors present in the female germ line at this stage of development potentiate the 173 transcriptional response of the DNA hypomethylated genome upon removal of EZH2. 174 Although the loss of EZH2 led to precocious transcriptional upregulation of meiotic genes in female 175 Ezh2 CKO PGCs, we did not observe any profound effect on the initial progression of the meiotic 176 prophase (Extended Data Fig. 6e, 8a, 8b). However, despite normal PGC number at E13.5, germ cell 177 numbers in Ezh2 f/-, Blimp1-Cre ovaries at E16.5 and E18.5 were significantly reduced (Fig. 3g). The 178 accumulation of γH2A.X signal (Fig. 4a) specifically in female Ezh2 CKO PGCs suggested the presence 179 of DNA damage, prompting us to examine the transcriptional regulation of transposable elements 180 (TEs) that when re-activated could compromise genome integrity. This revealed that many TEs 181 subfamilies were upregulated predominantly in the E13.5 female Ezh2 CKO PGCs (Fig. 4b), while only 182 very few differentially expressed TE subfamilies were identified in male Ezh2 CKO PGCs (Extended 183 Data Fig. 9a). The same trend is observed when analysing uniquely mapped, single copy TEs (Fig. 4c, 184 Extended Data Fig. 9b). The derepression of TEs in female Ezh2 CKO PGCs is further documented by 185 the upregulation of IAP-GAG and LINE1-ORF1 proteins (Fig. 4d) and the enrichment of P53 and 186 interferon-alpha response pathways in Gene Set Enrichment Analysis (GSEA) (Extended Data Fig. 9c, 187 9d). Both of these pathways have been previously linked to retrotransposon activation 21,22. In 188 addition, while most of RAD51 foci disappeared at E18.5, filament-like RAD51 signal was identified in 189 female Ezh2 CKO germ cells, suggesting that unrepaired DNA damage persisted along the 190 chromosomes at this developmental stage (Extended Data Fig. 8b). 191 To understand the molecular underpinnings of the TE activation induced by EZH2 loss, we analysed 192 the repressive chromatin profile. This revealed that between E10.5 and E13.5 the levels of H3K9me3 193 increased on many evolutionarily young and potentially transcriptionally active LTR and non-LTR 194 retrotransposons, such as L1Md_A, IAP-d-int, and MMERVK10C, (Fig. 4e). In fact, more than 90% of 195 de novo H3K9me3 peaks identified in E13.5 PGCs are located within 1kb to TEs (Extended Data Fig. 196 9e), suggesting that H3K9me3 provides transcriptional repression of TEs in the absence of DNA 197 methylation. TEs tend to gain more H3K9me3 in the male germline, with newly identified peaks 198 often representing the spreading of a repressive domain (Fig. 4e, Extended Data Fig. 9f). 199 Interestingly, many H3K9me3-marked ERVK, ERV1 and LINE1 subfamilies, including L1Md_Gf, T, A, 200 IAP-d-int and RLTR10, also show an increase in H3K27me3 signal between E10.5 and E13.5 (Fig. 4e). 201 H3K9me3 and H3K27me3 do not usually overlap in somatic cells; however, the dynamic changes 202 observed in PGCs indicate that PRC2 is recruited to these TEs following global DNA demethylation 203 (after E10.5), suggesting a possible functional compensation for the DNA methylation loss. The TEs 204 solely enriched in H3K9me3 tend to be evolutionarily young, transcriptionally competent ERVs 205 (IAPEz.int, IAPLTR1.Mm), while the H3K27me3 enriched TEs tend to be TE relics that lost their 206 transcriptional potential (Charlie, Tigger elements) (Extended Data Fig. 10a). Further analysis at the 207 level of single uniquely mapped TEs revealed three types of histone modification enrichment: 208 H3K9me3 and H3K27me3 are enriched separately on different copies of TEs (Extended Data Fig. 10b); 209 with the copies showing pronounced 5mC loss attracting H3K27me3 while TEs associated with 210 higher levels of DNA methylation showing H3K9me3 enrichment (Extended Data Fig. 10b). However, 211 we also detected a surprising co-enrichment of H3K27me3 and H3K9me3 on some copies 212 particularly of LINE1 elements (Extended Data Fig. 10c) as previously suggested23. Importantly, the 213 pattern of repressive histone modifications is specific to an individual integrated copy of TE rather 214 than being representative of the whole TE subfamily. 215 216 Soon following their entry into the developing genital ridge, mouse PGCs undergo a wave of global 217 DNA demethylation yielding the most hypomethylated state of genome ever observed during 218 development 9. Surprisingly, however this loss of DNA methylation does not lead to widespread 219 transcriptional activation raising an important question regarding the regulation of transcriptional 220 programme and fidelity in the almost near absence of this major repressive epigenetic system9. 221 Using an ULI-nChIP approach, we have investigated the dynamics of alternative epigenetic silencing 222 marks H3K27me3 and H3K9me3 in the context of dynamic global DNA methylation changes 223 occurring in gonadal PGCs. Our results show that genome-wide DNA demethylation is accompanied 224 by global remodelling of both of these repressive histone modifications, resulting in a sex specific 225 heterochromatin signature observed in post-reprogramming E13.5 PGCs (Fig. 1d, 1e). Using a genetic 226 loss of function model, we further show that EZH2mediated H3K27me3 is required to maintain 227 transcriptional repression in hypomethylated gonadal PGCs (Extended Data Fig. 3), reinstating that 228 PRC2 is necessary to prevent aberrant transcriptional activation when other stable silencing 229 mechanisms, such as DNA methylation, are absent. This is particularly apparent for germline 230 associated and developmental genes (Fig.3, Extended Data Fig. 3, Extended Data Fig. 10f); and 231 especially in the context of the developing early post reprogramming female germ line, where the 232 presence of RA signalling promotes entry into meiosis. Given that robust transcriptional activation 233 has been shown to override polycomb-mediated repression directly or indirectly 24,25, it is thus 234 plausible that the balance between promoter associated H3K27me3 and cell signalling induced 235 transcription factor repertoire regulates the correct timing of meiotic programme initiation in the 236 female germ line at this stage of development. 237 PRC1 has been previously reported to regulate meiosis-related gene expression in male and female 238 germ cells26. Intriguingly, only a small number of previously identified PRC1 targets are dysregulated 239 in our Ezh2 knockout system pointing to the non-redundant roles of PRC1 and PRC2 in the germ line 240 (Extended Data Fig. 10d). 241 Our study importantly reveals that next to silencing of coding genes, EZH2-mediated H3K27me3 is in 242 post reprogramming hypomethylated germ cells required to repress potentially active transposable 243 elements (Fig. 4, Extended Data Fig. 10f). Using a combination of WGBS and ULI-nChIP-Seq data, we 244 show that H3K27me3 is enriched on many TE subfamilies only after gonadal DNA demethylation (Fig. 245 4e) providing an alternative mode of transcriptional silencing. Loss of H3K27me3 results in profound 246 TE derepression, leading to DNA damage and eventually germ cell loss in the female germ line (Fig. 247 3g). Consistent with this, the elevation of TEs has been linked to foetal oocyte attrition 27. The 248 severity of the TE derepression in the female PGCs upon EZH2 loss is likely due to the sex specific 249 heterochromatin configuration observed in the developing germ line (Fig. 1d,e, 4e). Following the 250 gonadal 5mC loss, the global DNA methylation remains low for a long period of time during 251 oogenesis; whereas de novo DNA methylation ensues soon (from E15.5 onwards) in male germ cells 252 28. Additionally, the TE associated H3K9me3 enrichment is much more pronounced in male than in 253 female E13.5 PGCs (Fig. 4e), suggesting that the male germ line utilises H3K9me3 and DNA 254 methylation, in connection with SETDB1 and piRNA pathway29,30 to restrain TEs. In contrast, 255 H3K27me3 is indispensable for the repression of hypomethylated TEs in the female developing germ 256 cells. Our results thus show diversification of the epigenetic repressive systems in controlling the 257 integrated TEs (Extended Data Fig 10e) and explain the previously reported sex specific effect of the 258 Setdb1 deletion (Liu et al,). Our observation is also conceptually consistent with the previous report 259 in cultured mouse ES cells indicating that H3K27me3 relocalise and silence certain groups of 260 retrotransposons upon rapid and extensive loss of DNA methylation31; although different groups of 261 TEs seem to be responsive to the H3K27me3 loss in our in vivo system (Extended Data Fig. 10e). 262 Intriguingly, the role of polycomb in controlling parasitic DNA is likely to be ancestral as documented 263 by the recent discovery of the role of the Enhancer-of-zeste like protein Ezl1 in the repression of 264 transposable elements in Paramecium32 . 265 Group D: Loss of DNA methylation, Low H3K27me3 at E13.5. 497 ℎ10.5 > 0.2, ℎ10.5 − 12.5 > 0 ((13.53273)/(ℎ)) < 0.5 498 Non-classified: the rest of promoters. 499 500 CpG obs/exp ratio, percentage of CpG dinucleotide and percentage of C or G were 501 calculated within 2kb window as previously reported55. 502 503 Statistical analysis 504 All statistical analysis were performed using R or Graphpad software and are described in 505 the figure legends. Biological replicates for all experiments were based on embryos from 506 independent litters. Specifically, P values were calculated by Wilcoxon rank sum test 507 (Extended Data Fig. 2c, Fig. 3b, 7c). Adjusted P values were calculated by Wilcoxon rank sum 508 test and adjusted by Bonferroni correction (Fig. 2). P values were calculate by two-tailed 509 unpaired Student’s t test (Fig.3g and Fig. 4a, 4d). Adjusted P values were calculated by 510 ANOVA and Tukey’s post-hoc multiple comparison test (Extended Data Fig. 5e). 511 Box plots were plotted using Tukey’s method. The upper and lower hinges represent the 512 first and the third quartiles. The central line represents the median. The upper end of the 513 whisker represents the lowest value among either the third quartile plus 1.5 X IQR or the 514 maximum value from the data set. The lower end of the whisker represents the largest 515 value among either the first quartile minus 1.5 X (IQR) or the minimum value from the data 516 set. 517 Box plots were generated by R package ggplot2 with argument outlier.shape = NA in (Fig. 2b 518 and extended data Fig.3b). 519 Data availability 520 ChIP-seq and RNA-seq data have been deposited in Gene Expression Omnibus (GEO) under 521 GSE141182. 522 523 524 525 526 527 528 529 Figure Legends 530 Figure 1: Sex-specific remodelling of repressive histone modifications during genome-wide 531 DNA demethylation in PGCs. 532 a, Overview of chromatin dynamics during mouse PGC development. *: The assessment is 533 based on immunofluorescence staining. b, Genomic distribution of H3K27me3 and 534 H3K9me3 peaks. c, Genome-wide correlation (Spearman’s correlation coefficient, 2 kb bins, 535 whole genome) of H3K27me3 and H3K9me3 enrichment after normalised to the input. d, e, 536 Total H3K27me3 and H3K9me3 peak numbers at each PGC stage. Venn diagram showing 537 peaks overlapping between E10.5, E13.5 male and E13.5 female PGCs. Bar charts show 538 retained and de novo peaks. f, Western blot analysis of H3K27me3 (left) and H3K9me3 (right) 539 abundance based on 5000 mESCs and 5000 GFP-positive E13.5 PGCs. MVH: germ cell marker. 540 P value was calculated from 3 independent experiments by two-tailed Student’s t test. Error 541 bar: standard deviation. 542 Figure 2: Base composition determines H3K27me3 enrichment during gonadal DNA 543 demethylation. 544 a, Heat map depicting the dynamics of DNA methylation and the enrichment of H3K27me3 545 at promoters. 4 groups of promoters were identified based on the dynamic patterns 546 between DNA methylation and H3K27me3 (See Methods). b, Sequence characteristics of 547 the promoters described in (Fig. 2a). adj. P values were calculated by pairwise comparison 548 using Wilcoxon rank sum test and adjusted by Bonferroni correction. ***: P<0.001. c, 549 Dynamics of DNA methylation and H3K27me3 at all promoters with respect to their 550 sequence characteristics. Methylation rate and H3K27me3 enrichment are shown by the 551 colour gradient. Distribution of each dot’s value is shown using rug plot along x axis and y 552 axis. 553 554 Figure 3: Conditional Ezh2 KO leads to wide-spread transcriptional derepression and loss 555 of germ cells in the female germ line. 556 a, b, Immunofluorescence staining of EZH2 and H3K27me3 in gonadal sections (n=2). MVH: 557 germ cell marker. Yellow arrowheads indicate PGCs. DNA stained by DAPI (blue). Scale bar: 558 10 um. c, Principal component analysis (PCA) of RNA-seq data from E13.5 male and female 559 Ctrl and Ezh2 CKO (Ezh2-/-, Blimp1-Cre) PGCs at E13.5. d, Number of differentially expressed 560 (DE) genes using different fold change (FC) thresholds. adj. P<0.05. e, Gene ontology (GO) 561 terms associated with DE genes (CKO vs Ctrl). f, Heat map of gene expression and chromatin 562 dynamics of promoters associated with meiotic prophase genes. P value was calculated 563 using Fisher exact test. g, Representative immunostaining of E18.5 Ctrl and Ezh2 CKO female 564 gonad sections. DNA was stained by DAPI. Scale bar: 100 um. Dot plot shows the total 565 number of germ cells per ovary. Each dot represents one biological replicates. Error bars 566 indicate standard deviation. P values were calculated by two-tailed unpaired Student’s t test. 567 Figure 4 EZH2-mediated H3K27me3 regulates retrotransponson repression. 568 a, Immunofluorescence staining for γH2AX (biological replicates n=2). Quantification of the 569 staining per PGC nucleus shown on the right. Numbers indicate the numbers of analysed 570 PGC nuclei. DNA was stained by DAPI. Scale Bar: 10 um. b, Expression of TEs (Multiple 571 mapped plus uniquely mapped reads) in E13.5 PGCs (Ezh2 CKO vs Ctrl). Significantly 572 upregulated TEs are labelled in red. (Fold change > 1, FDR< 0.1). c, Dot plot represents the 573 expression levels of individual TE copies. Each dot represents a single element in indicated 574 subfamilies. Only uniquely mapped reads were considered. d, Immunofluorescence staining 575 (top) and signal quantification (bottom) of IAP GAG and LINE1 ORF1 proteins. Biological 576 replicates n=2. Statistical analysis was carried out using two-tailed unpaired Student’s t test. 577 e, Heat map showing H3K9me3 and H3K27me3 enrichment of TE subfamilies ranked by DNA 578 methylation levels at E10.5. Each row represents one TE subfamily. Both multiple mapped 579 and uniquely mapped reads were considered. 580 Extended data Figure 1 581 Summary of ULI-nChIP-seq and genomic distribution of H3K27me3 and H3K9me3 582 enrichment. 583 a, Experimental scheme of PGC isolation using ∆PE Oct4-GFP mice (GOF 18∆PE-GFP)12. 584 ~1000 PGCs were used for ULI-nChIP-seq. b, Characteristics of the H3K27me3 and H3K9me3 585 ULI-nChIP-seq. Fraction of paired-end reads based on the mappability. Uniquely aligned 586 (dark green). Uniquely aligned duplicates (light green). Multiple aligned (blue). Unaligned 587 (grey). c, Scatter plot showing the correlation between 2 biological replicates using 2kb 588 window. Higher variability observed between E13.5 samples relates to slight difference in 589 developmental progression between different embryos and litters (embryos from 590 independent litters used as biological replicates). Pearson correlation coefficient is shown 591 on the top left. d, Genomic distribution of H3K27me3 and H3K9me3 peaks. More than 60% 592 of H3K27me3 peaks are associated with promoters or gene bodies. H3K9me3 peaks are 593 located mostly in distal intergenic regions and introns. e, Distribution of H3K27me3 and 594 H3K9me3 peaks relative to transcription start site (TSS). f, Bar chart showing proportion of 595 H3K27me3 and H3K9me3 peaks associated with TEs in the genome. g, Violin plot showing 596 the distribution of peak intensity in E10.5, E13.5 male and female PGCs. Peaks were 597 identified by MACS2 peak calling pipeline with broad peak setting (See Methods). h, 598 Distribution of H3K27me3 and H3K9me3 peak length. The number of short peaks increases 599 at E13.5 male PGCs, compared with peaks at E10.5. 600 601 602 603 Extended Data Figure 2 604 Dynamics of DNA methylation and H3K27me3 at promoters during gonadal 605 reprogramming. 606 a, Whole-genome bisulphite sequencing (WGBS) data from E10.5, E12.5 female and E12.5 607 male PGCs9. Density plot depicting DNA methylation levels at all promoters. b, Density plot 608 depicting H3K27me3 enrichment at all promoters. c, Violin plot of H3K27me3 levels at 609 promoters that lost DNA methylation (DNA methylation >0.2 at E10.5). d, Dynamics of DNA 610 methylation and H3K27me3 at all promoters. DNA methylation and H3K27me3 enrichment 611 are shown by colour gradient. Distribution of each dot’s value is shown using rug plot along 612 x axis and y axis. e, Box plot shows H3K27me3 enrichment of low CpG density (CpG <4.1%) 613 promoters which gained H3K27me3 in the female PGCs following global loss of DNA 614 methylation. Box plots were presented by Tukey method. P values were calculated by 615 Wilcoxon rank sum test. ***: P <0.001. 616 Extended data Figure 3 617 Dynamics of DNA methylation, H3K9me3 and H3K27me3 at promoters during gonadal 618 reprogramming. 619 a, Heat map depicting the H3K27me3 and H3K9me3 enrichment (ULI-nChIP-seq), and DNA 620 methylation rate (WGBS) at promoters. The promoters were grouped based on the pattern 621 of dynamic change between DNA methylation and H3K27me3 (See also Fig. 2). Group A: loss 622 of DNA methylation, gain H3K27me3 at E13.5. Group B: median loss of DNA methylation, 623 high H3K27me3. Group C: low DNA methylation, High H3K27me3. Group D: loss of DNA 624 methylation, low H3K27me3 at E13.5. Promoters that did not meet the criteria were 625 grouped into non-classified. The expression levels of promoter-associated genes from RNA-626 seq are presented by TPM (Transcripts Per Kilobase Million) or z-score. The total number of 627 promoters in each group are shown on the left. b, Box plot showing the quantitative 628 measurement in each category, female and male PGCs, respectively. P values were 629 calculated by Wilcoxon rank sum test. c, Venn diagram showing the number of overlapped 630 promoters between male and female PGCs. 631 Extended Data Figure 4 632 Generation of the germline specific Ezh2 conditional knock out. 633 a, Representative immunofluorescence (IF) staining for EZH2 using cryosectioned genital 634 ridges. EZH2 is highly expressed during PGC development, compared with surrounding 635 somatic cells. Biological replicates n=3. OCT4: PGC marker. DAPI indicates DNA. Scale bar: 636 10um. b, Functional domains of EZH2 protein and targeting strategy of Ezh2 allele. Open 637 boxes: exons. Black arrowhead: loxP sites c, Breeding scheme for germline Ezh2 knockout. 638 Ezh2 Δ/ Δ, Tg (Blimp1-Cre) refers to CKO in the figures. f: allele flanked by loxP sites (floxed). 639 Δ: Deleted allele generated using Cre-mediated recombination. Tg (Blimp1-Cre): transgenic 640 mice express Cre recombinase under the control of Blimp1 (Prdm1) promoter. d, Deleted 641 alleles were confirmed by genotyping using the primers shown by black arrows in (b). #1: 642 Ezh2f/∆, Tg (Blimp1-Cre)+/-. #2: Ezh2f/+ 643 Extended Data Figure 5 644 Global H3K9me3, H2A119ub and DNA methylation are not altered in PGCs following the 645 loss of EZH2. 646 a, b, c, Representative IF staining for H3K9me3, H2A119ub and TET1 using cryosectioned 647 genital ridges. MVH: PGC marker. Biological replicates n=2. d, IF staining for 5-648 methylcytosine (5mC). 5mC is enriched in pericentromeric regions in the nucleus of somatic 649 cells but depleted in both Ctrl and Ezh2 CKO germ cells. e, Global 5mC and 5hmC levels were 650 measured by LC-MS/MS. Each dot represents one biological replicate. Mean values of 651 5mdC/dG or 5hmdC/dG are shown. adj. P values were calculated using ANOVA and Tukey’s 652 post-hoc multiple comparison test. f, Representative IF images of E13.5 Ctrl and Ezh2 CKO 653 female and male gonads. DNA was stained by DAPI. Scale bar: 100 um. The bar chart shows 654 the total number of germ cells per female gonad. Error bars indicate standard deviation. 655 Biological replicates n=2. 656 Extended data Figure 6 657 Transcriptome analysis of Ctrl and Ezh2 CKO PGCs. 658 a, Catalytic core and accessory subunits of mammalian PRC2. b, RNA expression of PRC2 659 components during germ cell development. c, RNA expression of Ezh1 in the Ctrl and Ezh2 660 CKO germ cells. d, Sample distance matrix of RNA-seq samples by non-supervising cluster 661 (see Methods). e, PCA Plot shows the distance of transcriptomes from different PGC 662 developmental stages. Dash line circle indicates samples of the same developmental stage. 663 Extended Data Figure 7 664 Sex-specific transcription factor repertoire determines transcriptional activation upon loss 665 of EZH2. 666 a, Gene ontology (GO) terms associated with E13.5 ♂ DE genes (Ctrl vs CKO). b, Integrative 667 Genomics Viewer (IGV) plot shows the H3K27me3 enrichment and RNA-seq read counts of 668 Stra8. Mouse genome: mm9. c, Heat map depicting gene expression and the chromatin 669 dynamics at promoters of meiotic differentially expressed (DE) genes. Box plot shows the 670 H3K27me3 enrichment and RNA expression (TPM) of meiotic DE genes in male and female 671 PGCs. Z scores were calculated for male and female separately. P values were calculated by 672 Wilcoxon rank sum test. ***: P<0.01 d, Bar chart showing the odds ratio of 104 meiosis 673 prophase genes19 in each groups of promoters (Fig. 2). P values were calculated by Fisher 674 exact test. *: P<0.05. **: P<0.01. ***: P<0.001. e, Promoters of upregulated genes in female 675 Ezh2 CKO are significantly enriched for transcription factor motifs that relate to retinoid acid 676 signalling pathway. Motif analysis was performed using Bioconductor package PWMEnrich. f, 677 Heat map shows the relative gene expression of identified transcription factors in male and 678 female PGC samples. g, Heat map shows gene expression of 45 Germline Reprogramming 679 Responsive (GRR) genes in Ctrl and Ezh2 CKO samples. Differentially expressed genes are 680 shown on the top (adj. P<0.05). 681 Extended Data 8 682 Loss of Ezh2 does not lead to precocious meiotic prophase 683 a, Representative IF images of the meiotic specific synaptonemal complex protein SCP3 in 684 embryonic ovaries. Axial element alignment was observed in female E18.5 germ cells. b, 685 Representative IF images of cryosectioned gonads. γH2AX signal shows DNA double strand 686 breaks (DSBs) occurring during homologous recombination. Accumulation of γH2AX signal in 687 E16.5 ctrl and Ezh2 CKO germ cells, this is greatly reduced at E18.5. A number RAD51 foci 688 can be identified at E16.5 but greatly decreases at E18.5. Filament-like, RAD51-positive 689 structure was identified in the Ezh2 CKO germ cells but not in Ctrl germ cells. Germ cells are 690 indicated by yellow arrowhead and are positive for MVH. DAPI indicates DNA. Scale bar: 691 10um. 692 Extended Data Figure 9 693 EZH2-mediated H3K27me3 regulates TE repression. 694 a, Differential expression analysis of TE expression in E13.5 male PGCs (CKO vs Ctrl). 695 Significantly upregulated TEs are labelled in red. Fold change > 1, FDR< 0.1. b, 696 Multidimensional scaling (MDS) plot showing distance of Ctrl and Ezh2 CKO samples based 697 on TE expression. Only uniquely mapped reads were considered. c, Gene Set Enrichment 698 Analysis (GSEA) of DE genes( E13.5♀ CKO vs Ctrl). Number of genes enriched in each gene 699 set is shown by the circle size. d, GSEA plot showing genes upregulated in female Ezh2 CKO 700 PGCs are enriched in p53 pathway and interferon alpha response. FDR q value < 0.25 was 701 considered significant. NES: normalized enrichment score. e, Analysis of distance between 702 H3K9me3 de novo peaks, transcription start sites (TSS) and transposable elements (TEs).f, 703 Representative IGV plot showing H3K9me3 enrichment on IAP Ez elements. 704 Extended Data Figure 10 705 H3K9me3 enrichment on TEs and co-localisation with H3K27me3. 706 a, TE subfamilies enriched predominantly for H3K9me3 or H3K27me3. Each row represents 707 one TE subfamily. Multiple mapped and uniquely mapped reads were taken into account. b, 708 c, Heat map showing DNA methylation, H3K9me3 and H3K27me3 enrichment at individual 709 copies of IAPLTR2_Mm and L1Md_Gf. Each row represents one uniquely mapped, single TE 710 copy belonging to the respective TE subfamily. d, Venn diagram depicting overlap of genes 711 upregulated in the Ezh2 CKO female PGCs and the Rnf2 CKO PGCs26. e, Venn diagram 712 showing overlap between the EZH2 and SETDB1 regulated TEs in mouse germ cells and 713 between the EZH2 regulated TEs in germ cells and the EED regulated TEs in mESCs 23,31. f, 714 Model depicting the relationship between DNA demethylation and heterochromatin 715 changes in gonadal PGCs undergoing epigenetic reprogramming. 716 717 718 Acknowledgements 719 We thank J. Elliot, T.Adejumo and B. Patel for help with the fluorescence activated cell sorting, C. 720 Whilding for the help with microscopy and IF data analysis, L. Game for the help with next-721 generation sequencing, Z. Agate-Bacon and G. Zimmerman for mouse husbandry, T. Carell (LMU 722 Munich) for providing isotopically labelled deoxynucleoside standards and the members of the 723 Hajkova laboratory for discussions and revisions of the manuscript. Work in the Hajkova laboratory is 724 supported by MRC funding (MC_US_A652_5PY70) and an ERC grant (ERC-CoG-648879–725 dynamicmodifications) to P.H. P.H. acknowledges membership in the EMBO Young Investigator 726 Programme. T.C.H. is a recipient of Imperial College London/Taiwan Top University Strategic Alliance 727 PhD Scholarship. 728 729 Author Contributions 730 T.C.H and P.H. conceived the study; T.C.H performed the experiments and analysed the data; Y.F.W. 731 analysed the next-generation sequencing data; E.V-F. carried out the ULI-nChIP with the help of C.H. 732 and G.K.; C.E.R. carried out LC–MS/MS; T.C.H. and P.H. wrote the manuscript. 733 734 735 736 1 Robertson, K. D. DNA methylation and human disease. Nat Rev Genet 6, 597-610, 737 doi:10.1038/nrg1655 (2005). 738 2 Guibert, S., Forne, T. & Weber, M. Global profiling of DNA methylation erasure in mouse 739 primordial germ cells. Genome research 22, 633-641, doi:10.1101/gr.130997.111 (2012). 740 3 Hackett, J. A. et al. Germline DNA demethylation dynamics and imprint erasure through 5-741 hydroxymethylcytosine. Science 339, 448-452 (2013). 742 4 Seisenberger, S. et al. The Dynamics of Genome-wide DNA Methylation Reprogramming in 743 Mouse Primordial Germ Cells. Molecular Cell 48, 849-862 (2012). 744 5 Seki, Y. et al. Cellular dynamics associated with the genome-wide epigenetic reprogramming 745 in migrating primordial germ cells in mice. Development 134, 2627-2638, 746 doi:10.1242/dev.005611 (2007). 747 6 Hajkova, P. et al. Chromatin dynamics during epigenetic reprogramming in the mouse germ 748 line. Nature 452, 877-881, doi:nature06714 [pii] 749 10.1038/nature06714 (2008). 750 7 Hajkova, P. Epigenetic reprogramming in the germline: towards the ground state of the 751 epigenome. Philos Trans R Soc Lond B Biol Sci 366, 2266-2273, doi:10.1098/rstb.2011.0042 752 (2011). 753 8 Spiller, C., Koopman, P. & Bowles, J. Sex Determination in the Mammalian Germline. Annu 754 Rev Genet 51, 265-285, doi:10.1146/annurev-genet-120215-035449 (2017). 755 9 Hill, P. W. S. et al. Epigenetic reprogramming enables the transition from primordial germ 756 cell to gonocyte. Nature 555, 392-396, doi:10.1038/nature25964 (2018). 757 10 Brind'Amour, J. et al. An ultra-low-input native ChIP-seq protocol for genome-wide profiling 758 of rare cell populations. Nat Commun 6, 6033, doi:10.1038/ncomms7033 (2015). 759 11 Hanna, C. W. et al. MLL2 conveys transcription-independent H3K4 trimethylation in oocytes. 760 Nature structural & molecular biology 25, 73-82, doi:10.1038/s41594-017-0013-5 (2018). 761 12 Yoshimizu, T. et al. Germline-specific expression of the Oct-4/green fluorescent protein (GFP) 762 transgene in mice. Dev Growth Differ 41, 675-684 (1999). 763 13 Juan, A. H. et al. Roles of H3K27me2 and H3K27me3 Examined during Fate Specification of 764 Embryonic Stem Cells. Cell Rep 18, 297, doi:10.1016/j.celrep.2016.12.036 (2017). 765 14 Li, H. et al. Polycomb-like proteins link the PRC2 complex to CpG islands. Nature 549, 287-766 291, doi:10.1038/nature23881 (2017). 767 15 Wang, X. et al. Molecular analysis of PRC2 recruitment to DNA in chromatin and its inhibition 768 by RNA. Nature structural & molecular biology 24, 1028-1038, doi:10.1038/nsmb.3487 769 (2017). 770 16 Mendenhall, E. M. et al. GC-rich sequence elements recruit PRC2 in mammalian ES cells. 771 PLoS genetics 6, 1-10 (2010). 772 17 Ohinata, Y. et al. Blimp1 is a critical determinant of the germ cell lineage in mice. Nature 436, 773 207-213, doi:10.1038/nature03813 (2005). 774 18 Li, Z. et al. The Sm protein methyltransferase PRMT5 is not required for primordial germ cell 775 specification in mice. EMBO J 34, 748-758, doi:10.15252/embj.201489319 (2015). 776 19 Soh, Y. Q. et al. A Gene Regulatory Program for Meiotic Prophase in the Fetal Ovary. PLoS 777 genetics 11, e1005531, doi:10.1371/journal.pgen.1005531 (2015). 778 20 Bowles, J. & Koopman, P. Retinoic acid, meiosis and germ cell fate in mammals. 779 Development 134, 3401-3411, doi:10.1242/dev.001107 (2007). 780 21 De Cecco, M. et al. L1 drives IFN in senescent cells and promotes age-associated 781 inflammation. Nature 566, 73-78, doi:10.1038/s41586-018-0784-9 (2019). 782 22 Jackson-Grusby, L. et al. Loss of genomic methylation causes p53-dependent apoptosis and 783 epigenetic deregulation. Nat Genet 27, 31-39, doi:10.1038/83730 (2001). 784 23 Liu, S. et al. Setdb1 is required for germline development and silencing of H3K9me3-marked 785 endogenous retroviruses in primordial germ cells. Genes and Development 28, 2041-2055, 786 doi:10.1101/gad.244848.114 (2014). 787 24 Riising, E. M. et al. Gene silencing triggers polycomb repressive complex 2 recruitment to 788 CpG islands genome wide. Mol Cell 55, 347-360, doi:10.1016/j.molcel.2014.06.005 (2014). 789 25 Schmitges, F. W. et al. Histone methylation by PRC2 is inhibited by active chromatin marks. 790 Mol Cell 42, 330-341, doi:10.1016/j.molcel.2011.03.025 (2011). 791 26 Yokobayashi, S. et al. PRC1 coordinates timing of sexual differentiation of female primordial 792 germ cells. Nature 495, 236-240, doi:10.1038/nature11918 (2013). 793 27 Malki, S., van der Heijden, G. W., O'Donnell, K. A., Martin, S. L. & Bortvin, A. A role for 794 retrotransposon LINE-1 in fetal oocyte attrition in mice. Dev Cell 29, 521-533, 795 doi:10.1016/j.devcel.2014.04.027 (2014). 796 28 Saitou, M., Kagiwada, S. & Kurimoto, K. Epigenetic reprogramming in mouse pre-797 implantation development and primordial germ cells. Development 139, 15-31, doi:139/1/15 798 [pii] 799 10.1242/dev.050849 (2012). 800 29 Pezic, D., Manakov, S. A., Sachidanandam, R. & Aravin, A. A. piRNA pathway targets active 801 LINE1 elements to establish the repressive H3K9me3 mark in germ cells. Genes and 802 Development 28, 1410-1428, doi:10.1101/gad.240895.114 (2014). 803 30 Inoue, K., Ichiyanagi, K., Fukuda, K., Glinka, M. & Sasaki, H. Switching of dominant 804 retrotransposon silencing strategies from posttranscriptional to transcriptional mechanisms 805 during male germ-cell development in mice. PLoS genetics 13, e1006926, 806 doi:10.1371/journal.pgen.1006926 (2017). 807 31 Walter, M., Teissandier, A., Perez-Palacios, R. & Bourc'his, D. An epigenetic switch ensures 808 transposon repression upon dynamic loss of DNA methylation in embryonic stem cells. Elife 809 5, doi:10.7554/eLife.11418 (2016). 810 32 Frapporti, A. et al. The Polycomb protein Ezl1 mediates H3K9 and H3K27 methylation to 811 repress transposable elements in Paramecium. Nat Commun 10, 2710, doi:10.1038/s41467-812 019-10648-5 (2019). 813 33 Wachter, E. et al. Synthetic CpG islands reveal DNA sequence determinants of chromatin 814 structure. Elife 3, e03397, doi:10.7554/eLife.03397 (2014). 815 34 Ehrlich, M. DNA hypomethylation in cancer cells. Epigenomics 1, 239-259, 816 doi:10.2217/epi.09.33 (2009). 817 35 Jones, P. A., Issa, J. P. & Baylin, S. Targeting the cancer epigenome for therapy. Nat Rev 818 Genet 17, 630-641, doi:10.1038/nrg.2016.93 (2016). 819 36 Su, I. H. et al. Ezh2 controls B cell development through histone H3 methylation and Igh 820 rearrangement. Nature immunology 4, 124-131, doi:10.1038/ni876 (2003). 821 37 Malki, S., Tharp, M. E. & Bortvin, A. A Whole-Mount Approach for Accurate Quantitative and 822 Spatial Assessment of Fetal Oocyte Dynamics in Mice. Biology of reproduction 93, 113, 823 doi:10.1095/biolreprod.115.132118 (2015). 824 38 Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: a flexible trimmer for Illumina sequence 825 data. Bioinformatics 30, 2114-2120, doi:10.1093/bioinformatics/btu170 (2014). 826 39 Kim, D. et al. TopHat2: accurate alignment of transcriptomes in the presence of insertions, 827 deletions and gene fusions. Genome Biol 14, R36, doi:10.1186/gb-2013-14-4-r36 (2013). 828 40 Liao, Y., Smyth, G. K. & Shi, W. The Subread aligner: fast, accurate and scalable read mapping 829 by seed-and-vote. Nucleic Acids Res 41, e108, doi:10.1093/nar/gkt214 (2013). 830 41 Liao, Y., Smyth, G. K. & Shi, W. featureCounts: an efficient general purpose program for 831 assigning sequence reads to genomic features. Bioinformatics 30, 923-930, 832 doi:10.1093/bioinformatics/btt656 (2014). 833 42 Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for 834 RNA-seq data with DESeq2. Genome Biol 15, 550, doi:10.1186/s13059-014-0550-8 (2014). 835 43 Young, M. D., Wakefield, M. J., Smyth, G. K. & Oshlack, A. Gene ontology analysis for RNA-836 seq: accounting for selection bias. Genome Biol 11, R14, doi:10.1186/gb-2010-11-2-r14 837 (2010). 838 44 Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for 839 interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 102, 15545-15550, 840 doi:10.1073/pnas.0506580102 (2005). 841 45 Risso, D., Ngai, J., Speed, T. P. & Dudoit, S. Normalization of RNA-seq data using factor 842 analysis of control genes or samples. Nat Biotechnol 32, 896-902, doi:10.1038/nbt.2931 843 (2014). 844 46 Li, H. & Durbin, R. Fast and accurate short read alignment with Burrows-Wheeler transform. 845 Bioinformatics 25, 1754-1760, doi:10.1093/bioinformatics/btp324 (2009). 846 47 Consortium, E. P. An integrated encyclopedia of DNA elements in the human genome. 847 Nature 489, 57-74, doi:10.1038/nature11247 (2012). 848 48 Amemiya, H. M., Kundaje, A. & Boyle, A. P. The ENCODE Blacklist: Identification of 849 Problematic Regions of the Genome. Sci Rep 9, 9354, doi:10.1038/s41598-019-45839-z 850 (2019). 851 49 Ramirez, F. et al. deepTools2: a next generation web server for deep-sequencing data 852 analysis. Nucleic Acids Res 44, W160-165, doi:10.1093/nar/gkw257 (2016). 853 50 Li, B. & Dewey, C. N. RSEM: accurate transcript quantification from RNA-Seq data with or 854 without a reference genome. BMC Bioinformatics 12, 323, doi:10.1186/1471-2105-12-323 855 (2011). 856 51 Zhang, Y. et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol 9, R137, 857 doi:10.1186/gb-2008-9-9-r137 (2008). 858 52 Yu, G., Wang, L. G. & He, Q. Y. ChIPseeker: an R/Bioconductor package for ChIP peak 859 annotation, comparison and visualization. Bioinformatics 31, 2382-2383, 860 doi:10.1093/bioinformatics/btv145 (2015). 861 53 Robinson, M. D., McCarthy, D. J. & Smyth, G. K. edgeR: a Bioconductor package for 862 differential expression analysis of digital gene expression data. Bioinformatics 26, 139-140, 863 doi:10.1093/bioinformatics/btp616 (2010). 864 54 Krueger, F. & Andrews, S. R. Bismark: a flexible aligner and methylation caller for Bisulfite-865 Seq applications. Bioinformatics 27, 1571-1572, doi:10.1093/bioinformatics/btr167 (2011). 866 55 Gardiner-Garden, M. & Frommer, M. CpG islands in vertebrate genomes. J Mol Biol 196, 867 261-282, doi:10.1016/0022-2836(87)90689-9 (1987). 868 869 Promoter Regions: (n=33158) Group A (n=764) Group B (n=821) Group C (n=2900) Group D (n=3114) Non-classified (n=25559) RNA-seq ChIP-seq WGBS 5mC+5hmC H3K27me3 H3K9me3 WT RNA Ezh2 CKO Group A (n=646) Group B (n=890) Group C (n=2998) Group D (n=3236) Non-classified (n=25388) RNA-seq ChIP-seq WGBS Extended Data Figure 3 Ezh2 Ctrl Brdt Ctcfl Rhox13 Syce3 Sycp1 Taf7l Tdrd9 Tex19.2 Mei1 Stra8 Sycp2 Tdrd1Tex101 Dnmt3l Foxa2 Gata4 Eomes Otx2 Runx1 Gli1 Pax6 Lefty1 Wnt3 Hoxa10 Bmp8b Twist1 Shh Stat5a Myod1 E10.5 E14.5♀ E12.5♀ E10.5 E12.5♀ E10.5 E13.5♀ E10.5 E13.5♀ E13.5♀ E13.5♀ Methylation rate 0 0.2 0.4 0.6 0.8 1 −2 −1 0 1 2 −2 −1 0 1 2 WT.RNAseq (TPM) 0 1000 2000 3000 4000 Ezh2 RNA-Seq (Z−score) −1 −0.5 0 0.5 1 H3K27me3 log2(ChIP/Input) H3K9me3 log2(ChIP/Input) Promoter Regions: (n=33158) 5mC+5hmC H3K27me3 H3K9me3 WT RNA Ezh2 CKO Ezh2 Ctrl E10.5 E14.5♂ E12.5♂ E10.5 E12.5♂ E10.5 E13.5♂ E10.5 E13.5♂ E13.5♂ E13.5♂ All.Promoters Non-classified GroupD GroupC GroupB GroupA E10.5 E12.5♀ 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6 Methylation Methylation rate All.Promoters Non-classified GroupD GroupC GroupB GroupA E10.5 E13♀ −4 −2 0 2 −4 −2 0 2 −4 −2 0 2 −4 −2 0 2 −4 −2 0 2 −4 −2 0 2 H3K27me3 H3K27me3 log2(ChIP/Input) All.Promoters Non-classified GroupD GroupC GroupB GroupA E10.5 E12.5♀ E14.5♀ 0 5 10 0 5 10 0 5 10 0 5 10 0 5 10 0 5 10 Gene expression log2(TPM) All.Promoters Non-classified GroupD GroupC GroupB GroupA −1.0 −0.5 0.0 0.5 1.0 −1.0 −0.5 0.0 0.5 1.0 −1.0 −0.5 0.0 0.5 1.0 −1.0 −0.5 0.0 0.5 1.0 −1.0 −0.5 0.0 0.5 1.0 −1.0 −0.5 0.0 0.5 1.0 Gene expression zscore(rlog) All.Promoters Non-classified GroupD GroupC GroupB GroupA 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6 Methylation rate All.Promoters Non-classified GroupD GroupC GroupB GroupA −4 −2 0 2 −4 −2 0 2 −4 −2 0 2 −4 −2 0 2 −4 −2 0 2 −4 −2 0 2 log2H3K27me3(ChIP/Input) All.Promoters Non-classified GroupD GroupC GroupB GroupA 0 5 10 0 5 10 0 5 10 0 5 10 0 5 10 0 5 10 All.Promoters Non-classified GroupD GroupC GroupB GroupA −1.0 −0.5 0.0 0.5 1.0 −1.0 −0.5 0.0 0.5 1.0 −1.0 −0.5 0.0 0.5 1.0 −1.0 −0.5 0.0 0.5 1.0 −1.0 −0.5 0.0 0.5 1.0 −1.0 −0.5 0.0 0.5 1.0 RNA RNA E13.5♀ Ctrl E13.5♀ CKO E10.5 E12.5♂ Methylation E10.5 E13♂ H3K27me3 E10.5 E12.5♂ E14.5♂ RNA RNA E13.5♂Ctrl E13.5♂ CKO Gene expression log2(TPM) Gene expression zscore(rlog) ♀ ♂ 429 335 311 358 463 427 739 2161 837 323 2791 445 Group A Group B Group C Group D ♂♀ ♂♀ ♂♀ ♀ ♂ ♂♀ P < 2.2e-16 P < 2.2e-16 P < 2.2e-16 P < 2.2e-16 P = 1.318e-10 P < 2.2e-16 P < 2.2e-16 P < 2.2e-16 P < 2.2e-16 P < 2.2e-16 P = 0.008619 P = 1.038e-09 a b c E10.5 E11.5 E12.5 ♂ E12.5 ♀ E13.5 ♂ E13.5 ♀ DAPI OCT4 EZH2 Merge Extended Data Figure 4 SANT SANT CXXC SET EZH2 746 a.a. Ezh2 Floxed locus 17 18 1916 20 15 Loxp Loxp Exons a ♀Ezh2 f/f ♂Tg(Blimp1-Cre)+/- ♂Ezh2 f/+,Tg(Blimp1-Cre)+/- ♂Ezh2 f/+,Tg(Blimp1-Cre)+/-♀Ezh2 f/f Ezh2 f/+,Tg(Blimp1-Cre)+/- Ezh2 f/▲,Tg(Blimp1-Cre)+/- Ezh2 f/+ Ezh2 f/▲ X X Fetus genotype Ezh2▲/▲,Tg(Blimp1-Cre)+/- Germline deletion Floxed allele Wt allele) Deleted allele( ) Blimp1-Cre #1 #2 440 bp 410 bp 370 bp 200 bp c Ezh2 Floxed locus 20 15 Loxp Exons 5’ left loxP 3’ loxP gen-Enx1 F gen-Enx1 R Cre-mediated deletion b d Ctrl E13.5 ♀ CKO Ctrl CKO MVHH3K9me3 Merge E13.5 ♂ a E13.5 ♂ CKO Ctrl H2AK116ub MVH Merge TET1 MVH Merge E13.5 ♂ CKO Ctrl c b Extended Data Figure 5 PI Mvh 5mC 5mC CKO Ctrl E13.5 ♀ d e 0.0 0.5 1.0 1.5 2.0 2.5 P<0.001 P<0.001 *** P<0.001 *** P<0.001 *** P=0.995 P=0.971 0.00 0.05 0.10 0.15 P=0.272 P=0.354 P=0.076 P=0.002 ** P>0.999 P=0.824 n=2 n=2 CKOCtrl E13.5 ♀ DAPI MVH Total number of germ cells/ Ovary f 1.734 0.253 0.316 0.244 0.365 E13.5 ♀ CKO E13.5 ♂ Ctrl E13.5♀ Ctrl E13.5 ♂ CKO E10.5 WT 5mdC/dG(%) 5hmdC/dG(%) E13.5 ♀ CKO E13.5 ♂ Ctrl E13.5♀ Ctrl E13.5 ♂ CKO E10.5 WT 0.076 0.065 0.065 0.047 0.056 E13.5♀ Ctrl E13.5 ♀CKO 0 1000 2000 3000 4000 5000 6000 *** E13.5 ♂ E10.5 E12.5 E14.5 0 50 100 150 200 Eed E10.5 E12.5 E14.5 0 20 40 60 80 Suz12 E10.5 E12.5 E14.5 0 50 100 150 200 250 300 350 400 Rbbp7 E10.5 E12.5 E14.5 0 50 100 150 200 Rbbp4 E10.5 E12.5 E14.5 0 20 40 60 80 Jarid2 E10.5 E12.5 E14.5 0 5 10 15 Aebp2 E10.5 E12.5 E14.5 0 20 40 60 80 100 Mtf2 E10.5 E12.5 E14.5 0 5 10 Phf1 E10.5 E12.5 E14.5 0 20 40 60 80 Phf19 E10.5 E12.5 E14.5 0 2 4 6 8 10 Ezhip TPM E10.5 E12.5 E14.5 0 20 40 60 80 Ezh1 E10.5 E12.5 E14.5 0 20 40 60 80 Ezh2 ♂ PGC ♀ PGC EZH1/2 EED SUZ12 PCL RBBP4/7 EZHIP PCL1/PHF1 PCL3/PHF19 PCL2/MTF2 JARID2 AEBP2 EZHIP PRC2.1 PRC2.2 E10.5 E12.5 E14.5 0 2 4 6 8 10 Epop E10.5 E12.5 E14.5 0 2 4 6 8 10 12 14 Lcor/Pali1 LCOR/PALI1 EPOP EZH1/2 EED SUZ12 RBBP4/7 ● ● ● ● ● ● ● ● ● ● ●●●● ●● ●● ●● ● ● ● ● ● ● ● −100 −50 0 50 100 −50 0 50 100 PC1: 21.09% variance PC2: 20.83% variance E10.5 WT E12.5 ♂ WT E12.5 ♀ WT E14.5 ♂ WT E14.5 ♀ WT E13.5 ♂ E13.5 ♀ c WT PCA Sample Distance Matrix E13.5 ♀ CKO E13.5 ♂ Ctrl E13.5♀ Ctrl E13.5 ♂ CKO E13.5 ♂ Ctrl E13.5♀ Ctrl E13.5 ♂ CKO E13.5♀ CKO E13.5 ♀ CKO E13.5 ♂ Ctrl E13.5♀ Ctrl E13.5 ♂ CKO d e 0 10 20 30 40 Ezh2 Ctrl Ezh2 CKO TPM E13.5♂ E13.5♀ Ezh1 a b Extended Data Figure 6 TPM ♂ PGC ♀ PGC a c Extended Data Figure 7 P-valueName Motif RARB 9.42E-13 RXRA 3.83E-12 RARG 4.82E-12 ESRRA 2.22E-11 MSX2 2.69E-09 CREB1 0.00017 CEBPB 0.00187 CEBPD 0.0291 CDX2 0.043 MSX1 0.047 Rarb Rxra Rarg Esrra Msx2 Creb1 Cebpb Cebpd Cdx2 Msx1 z−score −2 −1 0 1 2 d e E13.5 ♀ Ctrl E13.5 ♀ CKO E10.5 WT E13.5 ♀ WT E13.5 ♂ WT RNA H3K27me3 chr6:34880000-34890000 Stra8 E13.5 ♂ Ctrl E13.5 ♂ CKO 0 1 0 5000 E13.5 ♀ Ctrl E13.5 ♀ CKO E13.5 ♂ Ctrl E13.5 ♂ CKO 01234 spermatid development phosphatidylinositol 3-kinase signaling multi-organism reproductive process spermatogenesis multi-organism process 9/115 7/73 22/588 15/348 34/1168 -log10(P-value) E13.5 ♂ CKO vs Ctrl DE (adj. P<0.05, FC>2) z−score −2 −1 0 1 2 adj. P<0.05 Phf17 Hsf5 Mael Sec1 4933416C03Rik Taf7l Adad1 1700013H16Rik Slc25a31 4930467D21Rik Sycp3 Tex12 Gm16270 Fkbp6 4921515J06Rik Tex15 Ly6k Trim52 1700018B24Rik Sycp2 Taf9b Rhox13 Asz1 Brdt Naa11 Rpl10l Rad51c Fam178b Dazl Sycp1 Pnldc1 8030474K03Rik Ddx4 Tdrd1 Mov10l1 Gm7061 D1Pas1 Hormad1 Dpep3 Stk31 Hsf2bp Gm13718 Gm2382 1700029P11Rik AU022751 E13.5 ♀ Ctrl E13.5 ♀ CKO GRR genes Tsc22d3 Mei1 Taf9b Taf7l H2−K1 Tex12 Cdkl2 Tex101 Spdya Slc25a31 Tktl1 4933416C03Rik 1700013H16Rik Rhox13 Sycp2 Figla Syce3 Aspa BC051142 4933427D06Rik Tex15 Fmr1nb Wbp2nl Larp1b Syn2 Sycp3 Dopey1 Ccdc36 Inca1 Syngr4 Ccdc79 H3K27me3 WT RNA E14.5♂ E12.5♂ E10.5 E12.5♀ E12.5♂ E10.5 E14.5♀ E10.5 E13.5♀ E13.5♂ E12.5♀ E13.5 ♀ Ctrl E13.5 ♀ CKO E13.5 ♂Ctrl E13.5 ♂ CKO Methylation rate 0 0.2 0.4 0.6 0.8 1 −2 −1 0 1 2 WT.RNAseq (TPM) 0 1000 2000 3000 4000 Expression (z−score)H3K27me3 log2(ChIP/Input) −1 −0.5 0 0.5 1 5mC+5hmC f ● −1 −2 −3 0 1 2 3 E13.5 ♀ E13.5 ♂ E10.5 H3K27me3 log2(ChIP/Input) P= 0.3064 p = 0.6147 P= 0.493 Gene expression (log2(TPM)) E13.5 ♀ Ctrl E13.5 ♀ CKO E13.5 ♂ Ctrl E13.5 ♂ CKO ● ● ● ● ● ● ● ● ● ● ● ● ● ● −5 0 5 10 P = 0.008534 P = 0.2868 Meiotic DE genes ** 012 Group A Group B Group C Group D 345 Odds Ratio ♀104 meiosis prophase genes 01234 Group A Group B Group C Group D ♂ Odds Ratio P=7.89E-07 *** P=0.001104 ** P=0.971044 P=0.012323 * P=2.66E-05 *** P=0.069418 P=0.961657 P=0.033084 * g b E13.5 ♀ CKOCtrl E15.5 ♀ CKOCtrl E16.5 ♀ CKOCtrl E18.5 ♀ CKOCtrl SCP3 SCP3 a b DAPI SCP3 MVH DAPI SCP3 MVH γH2AX RAD51 DAPI γH2AX MVH DAPI RAD51 MVH E16.5 ♀ CKOCtrl E18.5 ♀ CKOCtrl Zoom-in Extended Data Figure 8 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● L1MA6 L1P5 ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● IAP−d−int MER34 −2 0 2 0 3 6 9 0 3 6 9 log2FC − log10 P-Value LINE LTR E13.5♂ CKO vs Ctrl Extended Data Figure 9 RNA-seq: retrotransposon Uniquely mapped reads only Leading logFC dim 2 E13.5 ♂ Ctrl E13.5♀ Ctrl E13.5 ♂ CKO E13.5♀ CKO ● ● ● ●● ●● ● ● ● ● ● −1.5 −1.0 0.5 1.0 −1.0 −0.5 0 0.5 1.0 −0.5 0 a b c Leading logFC dim1 No. of genes FDR q value P53_PATHWAY INTERFERON_ALPHA_RESPONSE P53 pathway Interferon alpha response d e 50 40 30 20 10 TE 10 20 30 40 50 Percentage (%) (5'−>3') E13.5♀ De novo peaks E13.5♂ De novo peaks 0−1kb 1−3kb 3−5kb 5−10kb 10−100kb >100kb Distance of H3K9me3 peaks to TEs 50 40 30 20 10 TSS 10 20 30 40 50 Distance of H3K9me3 peaks to TSS 0−1kb 1−3kb 3−5kb 5−10kb 10−100kb >100kb E13.5♀ De novo peaks E13.5♂ De novo peaks Percentage (%) (5'−>3') E10.5 WT E13.5 ♀ WT E13.5 ♂ WT Peak calling Peak calling Peak calling 0 1 IAP Ez-int chr3:125830kb-125840kb f Extended Data Figure 10 a IAPLTR2_Mm L1Md_Gf b H3K9me3 enrichedH3K27me3 enriched Co-localised (uniquly mapped, single copy TEs) Mutually exclusive (uniquly mapped, single copy TEs) 5mC+5hmC H3K9me3 E10.5 E12.5♀ E12.5♂ E10.5 E13.5♀ E13.5♂ E10.5 E13.5♀ E13.5♂ H3K27me3 5mC+5hmC H3K9me3 E10.5 E12.5♀ E12.5♂ E10.5 E13.5♀ E13.5♂ E10.5 E13.5♀ E13.5♂ H3K27me3 0 0.2 0.4 0.6 0.8 1 −4 −2 0 2 4 −1 −0.5 0 0.5 1 Methylation rate H3K9me3 log2(ChIP/Input) H3K27me3 log2(ChIP/Input) IAPLTR1.Mm IAPLTR1a.Mm IAPEz.int SYNREP.MM GSAT.MM LTRIS2 MMETn.int LTRIS4 RLTR13A2 MER50 RLTR13A3 RLTR13C1 RLTR13D6 MER101B RMER16.int L1M3b RLTR47.MM LTR59 RLTR13D1 MER72B MMSAT4 ORR1D1 ORR1C1 ORR1A0.int ORR1C2 BGLII ORR1B1 ORR1B2 ORR1A4 HAL1.2a.MD RMER16 BGLII.Mus RLTR19C L1M1 L1.Mur2 ORR1C2.int ORR1A1.int RLTR25B MusHAL1 ORR1A3.int RLTR33 RMER6A MER106B ORR1A4.int ORR1B1.int RMER3D.int MYSERV.int RLTR21 RMER4A B1.Mus1 MER63C RLTR9A LTR68 Lx7 MurERV4.int Lx9 Lx8 L1.Rod MTC RMER4B FordPrefect MYSERV6.int RLTR23 RLTR24 ORR1D.int RLTR15 L1.Mur3 ORR1D1.int LTR55 MERVL.2A.int MYSERV16.I.int ORR1A2.int MLT1F1.int MER90a RLTR40 RLTR12A MT2B L1MEa RMER12 RLTR11A2 RLTR19.int RLTR11A HAL1.3A.ME RLTR19B RLTR11B BGLII.B RLTR22.Mur RLTR12B RLTR9B2 RLTR43B MADE2 RLTR14.int MLT1.int LTR16D2 MER88 RLTR18 MLT1E2.int MER73 RLTR22.Mus RLTR9B RLTR30 MER76 L1M3a MER67D HERVL40.int RMER10A MER4B.int RMER10B MER70.int MER72 LTR52.int MLT1H1.int SUBTEL.sa MER66B Cheshire.Mars LTR58 Arthur1A Ricksha.a MLT1D.int MLT1C.int L1M2b Charlie17a LTR88c Kanga1d MamRep1894 ORR1E MTEb MER95 ERVL.int LTR75 MLT1J Charlie24 MER106A MER47A MLT2B2 ZP3AR MLT1H2 MLT2B1 MER44C LTR90A B4A MLT1I MER45B LTR29 Charlie22a Charlie10 MLT1A0 ID2 MLT1A1.int MER89 MER31A MLT1C MLT1G MLT1L MLT1G3 MER77 MLT1F1 MLT1F2 MLT1H1 Arthur1B MLT2C2 RMER12B LTR16B1 LTR40b Tigger9b MLT1A MLT1I.int Kanga1a MLT2E LTR53 LTR33 MLT1F.int LTR85a MER45A LTR16D1 MER70B MLT1J.int L2b Charlie13a MLT1E3 MER45C MLT1G1 MER110A LTR16A1 RLTR42 LTR31 LTR85c LTR16B2 MLT1J1 Tigger15a MER81 LTR16A Zaphod2 MER34.int MLT1M MLT1J1.int LTR16C MamRep1161 MamGypLTR2b Charlie10b MamGypLTR1c MER47C LTR86A1 LTR64 MamRep38 LTR84a LTR50 LTR16D MER91A MamGypLTR1a AmnSINE1 MLT1E3.int LTR41B LTR83 LTR45 L1M2a1 MLT1E.int 5mC+5hmC H3K9me3 E10.5 E12.5♀ E12.5♂ E10.5 E13.5♀ E13.5♂ E10.5 E13.5♀ E13.5♂ H3K27me3 5mC+5hmC H3K9me3 E10.5 E12.5♀ E12.5♂ E10.5 E13.5♀ E13.5♂ E10.5 E13.5♀ E13.5♂ H3K27me3 H3K9me3 enriched H3K27me3 enriched 20 3 14 IAPLTR1_Mm IAPEz.int MMERVK10C.int Ezh2 CKO Upregulated TEs Setdb1 CKO upregulated TEs E13.5♀ PGCs E13.5♀ PGCs 18 5 18 IAPEz.int L1Md_A L1Md_Gf L1Md_T MMERVK10C.int Ezh2 CKO Upregulated TEs Eed KO upregulated TEs at D13 E13.5♀ PGCs ES cells 2i+VitC c d 0 0.2 0.4 0.6 0.8 1 −4 −2 0 2 4 −1 −0.5 0 0.5 1 Methylation rate H3K9me3 log2(ChIP/Input) H3K27me3 log2(ChIP/Input) H3K9me3 EZH2 H3K27me3 H3K9me3 E13.5♀ PGCs TE derepression Germ cell reduction H3K9me3 EZH2 H3K9me3 E10.5 PGCs Phenotype TE repression DNA damage Ezh2 Ctrl Ezh2 CKO Transposable elements ? Retrotransposons 5mC Unmethylated C EZH2 H3K27me3 E10.5 PGCs E13.5♀ PGCs Wide-spread derepression Transcriptional outcomes Silencing or Low level trancription RA signalling Transcription factors Ezh2 Ctrl Ezh2 CKO EZH2 Meiosis related genes 5mC Promoters Unmethylated C RA signalling Transcription factors f 450 74 851 Rnf2 CKO up FC1.5, P<0.05 Ezh2 CKO up P<0.05 e