NR TFs regulate mouse early development

To identify putative key TFs functioning between ZGA and the first lineage segregation in mouse early development, we firstly investigated the regulatory elements from our previous ATAC-seq data,17,28 which showed strong enrichment of motifs of NR TFs, including NR5A2 and RARG, from the 2–8C stages (Fig. 1a). Consistently, these two TFs were highly activated upon ZGA but their expression declined after the 8C stage (Fig. 1b, right). As a comparison, pluripotency TFs NSO’s motifs were enriched in ICMs and mESCs, paralleled by their expression dynamics (Fig. 1a). NR5A2 protein was undetectable in oocytes under our condition, consistent with its weak transcription levels, but was strongly induced at the 2C stage using immunofluorescence (Supplementary information, Fig. S1a). Given that NR family members often share similar motif sequences (Fig. 1b, left),29 we asked whether the NR family TFs may play redundant roles during early development. RNA-sequencing (RNA-seq) analysis showed that two additional NR family TFs Nr1h3 and Nr2c2 were also highly expressed from 2–8C, and exhibited similar binding motifs (Fig. 1b). Therefore, we first performed quadruple knockdown (4KD) against these four NR factors by injecting combined siRNAs into zygotes (Supplementary information, Fig. S1b, left). Remarkably, the 4KD embryos developed to 8C at a lower rate and barely developed to blastocysts, with most embryos arrested at the morula stage (Supplementary information, Fig. S1b, middle and right). RNA-seq confirmed the KD efficiency of these NR TFs at the 4C and 8C stages (Supplementary information, Fig. S1c). In particular, 4C-specifically activated genes and 8C-specifically activated genes (Materials and Methods; Supplementary information, Tables S1 and S2) were substantially downregulated upon KD of the 4 NR TFs (Supplementary information, Fig. S1d). Taken together, these data indicate that NR family TFs play important roles in early development and transcription regulation at the 4–8C stages.

Fig. 1: Both KD and KO of Nr5a2 led to morula arrest.

a TF motifs identified from distal ATAC-seq peaks17,28 in mouse early embryos. Sizes of circles indicate the –LogP value. Expression levels of TFs are color-coded. Epi, E6.5 Epiblast; Ect, E7.5 ectoderm. b Sequence logos of the binding motifs of NR5A2, RARG, NR1H3, and NR2C2 (left), and RNA expression (FPKM) of Nr5a2, Rarg, Nr1h3, and Nr2c2 (right). VE, visceral endoderm. The error bars denote the standard deviations of two biological replicates of RNA-seq. c Bar charts showing the expression of Nr5a2, Rarg, Nr1h3, and Nr2c2 in the negative control (NC) group (injected with negative control siRNA) and siRNA targeting group based on RNA-seq. The error bars denote the standard deviations of two biological replicates of RNA-seq. d Immunofluorescence of NR5A2 (red) and DAPI (blue) in mouse 8C embryos after KD of Nr5a2 (left). Scale bars: 20 μm. Quantification of NR5A2 signal intensity (relative to DAPI) and P values (t-test, two-sided) are shown (right, n = 12–13 embryos). Each dot represents a single blastomere. e Bar plots showing the developmental rates of NC group, Nr5a2 KD group, Rarg KD group, Nr1h3 KD group, and Nr2c2 KD group at the blastocyst stage (E4.5). MO, morula; BL, blastocyst. f Bar charts showing the percentage of WT Nr5a2 (blue) and base-edited Nr5a2 (red) reads from RNA-seq at the 8C stage after injection of Nr5a2 sgRNA only, base-editor (BE) mRNA only, and both sgRNA and BE mRNA. Each bar represents a single embryo. g Immunofluorescence of NR5A2 (red) and DAPI (blue) in mouse 8C embryos after injection of sgRNA only, BE mRNA only, and both sgRNA and BE mRNA (left). Scales bars: 20 μm. Quantification of NR5A2 signal intensity (relative to DAPI) and P values (t-test, two-sided) are shown (right, n = 6–8 embryos). Each dot represents a single blastomere. h Bar plots showing the developmental rates at the blastocyst stage (E4.5) after injection of sgRNA only, BE mRNA only, and both sgRNA and BE mRNA.

KD or KO of Nr5a2 leads to developmental arrest at the morula stage

To ask which NR TF plays a more critical role in early development, we removed each one of the 4 siRNAs in the combined siRNAs and evaluated the developmental rates at E4.5 (blastocyst) (Supplementary information, Fig. S2a, b). Remarkably, when knocking down these 4 NR TFs, leaving out the siRNA targeting Nr5a2, but not Nr1h3 or Nr2c2, substantially “rescued” the morula arrest phenotype of the 4KD embryos (Supplementary information, Fig. S2c). Leaving out Rarg siRNA had a moderate “rescue” effect (Supplementary information, Fig. S2c). Consistently, leaving out Nr5a2 siRNA also restored normal 4C and 8C transcriptional programs (Supplementary information, Fig. S2d), indicating that NR5A2 is the main TF responsible for the 4KD phenotype. To further confirm this result, we performed a single KD for each of the 4 NR TFs (Fig. 1c). Indeed, knocking down Nr5a2 alone recapitulated the transcription defects and morula arrest phenotype in 4KD embryos (Fig. 1d, e; Supplementary information, Fig. S2e–g). Knocking down Rarg led to a milder developmental defect, while depletion of Nr1h3 or Nr2c2 had negligible effects in preimplantation development (Fig. 1e). To rule out possible off-target effects of Nr5a2 KD, we attempted to mutate Nr5a2 via base editing in embryos (Supplementary information, Fig. S3a).30 The mutation of Nr5a2 mRNA and the depletion of protein was confirmed by RNA-seq and immunofluorescence, respectively, at the 8C stage (Fig. 1f, g). Consistently, these mutant embryos largely arrested at the morula stage (Fig. 1h; Supplementary information, Fig. S3b), thus phenocopying the Nr5a2 KD embryos. Taken together, these results suggest that among the four NR TFs activated upon ZGA, NR5A2 is the dominant factor in regulating early development and transcription.

NR5A2 is largely dispensable for global ZGA but is required for the 4–8C gene program

We then asked how these NR factors including NR5A2 regulate transcription programs in early development. Knocking down Nr5a2 led to the strongest transcription changes among the 4 NR TFs (Supplmentary information, Fig. S4a), with 976 downregulated and 806 upregulated genes (Fig. 2a; Supplmentary information, Table S3). Furthermore, only Nr5a2 KD recapitulated the failed activation of 4–8C genes (Fig. 2b). By contrast, the gene dysregulation at the 2C stage was much weaker compared to that at 8C (Fig. 2a, b; Supplementary information, Table S4), with 128 genes downregulated and 115 genes upregulated. A similar observation was made for Nr5a2 BE KO embryos (Fig. 2c), consistent with the observations that after Nr5a2 KD or BE KO, embryos progressed beyond 2C normally (Supplementary information, Figs. S2g and S3b). We confirmed the depletion or mutation of Nr5a2 mRNA and protein at this stage (Fig. 2d, e). Among all examined embryos at the 2C stage (14 embryos), 6 embryos displayed nearly completely mutated Nr5a2 mRNA reads (96.9%–100%) (Supplementary information, Fig. S4b, the 1st–6th KO embryos), while the remaining 8 embryos exhibited high mutation rates ranging from 66.7% to 87.5% (Supplementary information, Fig. S4b, the 7th–14th KO embryos). The presence of wild-type (WT) transcripts may arise from either unedited alleles or low levels of oocyte-deposited Nr5a2. The zygotic genome was globally activated in both Nr5a2 KD and BE KO embryos (Fig. 2f, g). When using a  defined ZGA gene list (n = 1170, Materials and Methods), relatively small and comparable numbers of ZGA genes were upregulated (n = 24 and 13, respectively) and downregulated (n = 36 and 19, respectively) in KD and BE KO embryos (Fig. 2g; Supplementary information, Fig. S4b and Table S5). In sum, these data suggest that NR5A2 is largely dispensable for the initiation of ZGA but plays a key role in transcriptional regulation especially during the 4–8C stages.

Fig. 2: NR5A2 is largely dispensable for global ZGA but is required for 4–8C gene activation.

a Bar plot showing the numbers of upregulated (red) and downregulated (blue) genes after Nr5a2 KD in mouse 2C and 8C embryos. b Box plots showing the average expression levels of stage-specifically activated genes for NC groups (blue), Nr5a2 KD groups (red), Rarg KD groups (green), Nr1h3 KD groups (brown), and Nr2c2 KD groups (yellow) at the 8C stage. c Box plots showing the average expression levels of stage-specifically activated genes after injection of sgRNA only (blue), BE mRNA only (green), and both sgRNA and BE mRNA (red) at the 8C stage. d Bar charts showing the expression of Nr5a2 in NC and Nr5a2 KD group at the 2C stage (left), with P values (t-test, two-sided) indicated (top). Bar charts show the percentages of WT Nr5a2 (blue) and base edited Nr5a2 (red) reads from RNA-seq after injection of Nr5a2 sgRNA only, BE mRNA only, and both sgRNA and BE mRNA at the 2C stage (bottom). e Immunofluorescence of NR5A2 (red) and DAPI (blue) in mouse 2C embryos after Nr5a2 KD (top, left). Immunofluorescence of NR5A2 (red) and DAPI (blue) in mouse 2C embryos after injection of sgRNA only, BE mRNA only, and both sgRNA and BE mRNA (bottom, left). Scale bars: 20 μm. Quantification of NR5A2 signal intensity (relative to DAPI) in mouse 2C (n = 4–7 embryos) and 8C (n = 6–8 embryos) embryos and P values (t-test, two-sided) are shown (right). Each dot represents a single blastomere. f Box plots showing the average expression levels of stage specifically activated genes for NC groups (blue) and Nr5a2 KD groups (red) at the 2C stage (top). Box plots show the average expression levels of stage-specifically activated genes after injection of sgRNA only (blue), BE mRNA only (green), and both sgRNA and BE mRNA (red) at the 2C stage (bottom). g Scatter plot comparing gene expression of NC group and Nr5a2 KD group in mouse 2C embryos (top). Scatter plot comparing gene expression of NC group and Nr5a2 base edited group in mouse 2C embryos (bottom). ZGA genes are colored in red. The Pearson correlation coefficients are also shown.

Genome-wide mapping of NR5A2 chromatin occupancy in mouse early embryos

To further investigate the direct targets of NR5A2 during the mouse early development, we sought to capture its chromatin binding landscape using CUT&RUN.31 We identified a sensitive NR5A2 antibody that can detect its binding using as few as 1000 WT mESCs, but not in Nr5a2 KO mESCs (Supplementary information, Fig. S5a, b). NR5A2 motif was the top 1 enriched motif in its binding peaks (Supplementary information, Fig. S5c), thus validating the data. Next, we performed NR5A2 CUT&RUN in the 2C and 8C embryos, which were highly reproducible (Fig. 3a; Supplementary information, Fig. S5d and Table S6). We could not detect NR5A2 binding in ICM likely due to its low expression (Fig. 1b) and the limited cells we collected. Therefore, we included mESCs for comparison instead. As a strong validation, the NR5A2 motif was again enriched in 2C and 8C binding peaks as the top 1 motif (Fig. 3b; Supplementary information, Table S7). At the 2–8C stages, NR5A2 showed distinct binding compared to that in mESCs and was preferentially enriched at the promoter of 2–8C specifically expressed genes (Fig. 3c; see Fig. 3d for example). In distal regions, NR5A2 also occupied sites specifically marked by H3K27ac and open chromatin,17 suggesting that NR5A2 preferentially bound active regulatory regions (Fig. 3e). Stage-specific binding analysis showed that 2C-specific and 2–8C shared NR5A2 distal bindings were enriched near genes involved in housekeeping activities including DNA repair, RNA processing, histone modification, and DNA methylation (Fig. 3e, “C1” and “C4”). Interestingly, 8C-specific NR5A2 binding tended to enrich near genes functioning in blastocyst formation and stem cell maintenance, indicating the possible involvement of NR5A2 in lineage regulation at this stage (Fig. 3e, “C2”). By contrast, mESC-specific binding was found near genes functioning in epithelial cell maturation (Fig. 3e, “C3”). LIF-responding genes also showed higher enrichment of NR5A2 in mESC (Fig. 3e, “C5”). Notably, NR5A2 also bound the putative enhancers near Nr5a2 itself (Fig. 3d), suggesting a positive feedback regulatory loop. Taken together, these data demonstrate that we successfully captured NR5A2 binding in 2C and 8C embryos.

Fig. 3: NR5A2 binding dynamics in mouse early embryos.

a The UCSC browser view showing NR5A2 CUT&RUN signals in 2C embryos, 8C embryos, and mESCs (two biological replicates). NR5A2 ChIP-seq from a reference dataset24 is also shown. b TF motifs identified from NR5A2 distal binding peaks in 2C embryos, 8C embryos, and mESCs. Sizes of circles indicate –LogP value. Expression levels of TFs are color-coded. c Heat maps showing stage-specific gene expression based on a reference dataset17 and their promoter NR5A2 binding enrichment. d The UCSC browser views and heat maps showing NR5A2 enrichment and gene expression of representative genes, respectively. e Heatmaps showing the NR5A2 binding, ATAC,17 and H3K27ac signals at the stage-specific and shared NR5A2 binding peaks (left). The gene ontology (GO) terms are also shown for different clusters (right).

NR5A2 targets are preferentially downregulated upon its deficiency

We then asked whether NR5A2 regulates its binding targets by primarily focusing on the 8C stage when the impact of NR5A2 on transcription was evident. NR5A2 binding was preferentially enriched at the promoters of downregulated but not upregulated genes upon Nr5a2 KD (Fig. 4a, left), suggesting that it functions primarily as an activator. Consistently, distal NR5A2 binding peaks also preferentially resided near the downregulated genes (Fig. 4a, right). Supporting a direct activation role of NR5A2, genes with NR5A2 binding and more NR5A2 motifs at their promoters were more downregulated in Nr5a2 KD embryos (Fig. 4b). Among 976 downregulated genes at the 8C stage in Nr5a2 KD embryos, a significant fraction (27%, n = 263, P < 0.001) showed NR5A2 binding at promoters. These data demonstrated that NR5A2 directly regulates gene activation at the 8C stage.

Fig. 4: NR5A2 regulates the expression of its binding targets in early development.

a Box plots showing the average enrichment of NR5A2 binding signals at the promoters (TSS (transcription starting site) ± 2.5 kb) (in WT embryos) of downregulated, upregulated, expressed, and all genes identified in Nr5a2 KD 8C embryos (left), with P values (t-test, two-sided) indicated. The cumulative distributions of downregulated, upregulated, expressed, and all genes with defined distances (x-axis) between their TSSs and nearest distal 8C NR5A2 binding peaks are shown (right). b Box plots showing the fold changes of gene expression in Nr5a2 KD 8C embryos for all expressed genes based on the NR5A2 binding states and the numbers of motifs at promoters, with P values (t-test, two-sided) indicated. c Heatmap showing the NR5A2 binding, enrichment of ATAC-seq signals in WT reference,17 NC, Nr5a2 KD 2C and 8C embryos. Average plots show enrichment of ATAC-seq signals in NC and Nr5a2 KD 8C embryos at the 2C-specific, 8C-specific, and 2–8C shared NR5A2 distal binding peaks. d The UCSC browser views showing NR5A2 binding, ATAC-seq signals of representative genes in 2C and 8C NC, and Nr5a2 KD embryos (top). Bar plots shows the expression fold-changes of representative genes between Nr5a2 KD and control 8C embryos of two biological replicates of RNA-seq (bottom). e Heatmap showing NANOG binding in E3.5 blastocyst and NSO binding in mESC at 2C-specific, 8C-specific, and 2–8C shared NR5A2 distal binding peaks. f Venn diagrams showing the overlap among 8C NR5A2 binding peaks, 8C-specific NR5A2 binding peaks, and E3.5 blastocyst NANOG peaks (top). Random peaks shuffled from the genome with matched lengths are also shown (bottom). g Volcano plot showing the TF footprints in differentially accessible regions in ATAC-seq upon Nr5a2 KD at the 8C stage. Example footprints in regions with increasing and decreasing accessibilities are colored in red and green, respectively. P values for their enrichment are also shown on the y-axis.

To further investigate whether NR5A2 regulates chromatin accessibility at putative enhancers, we performed ATAC-seq in the negative control (NC) and Nr5a2 KD embryos. We classified distal NR5A2 peaks into 2C-specific, 8C-specific, and 2–8C shared peaks (Fig. 4c). At the 2C stage, chromatin accessibility was globally unaltered (Fig. 4c, “2C ATAC” and Fig. 4d, “Rangap1”). The closing of 2C-specific peaks at the 8C stage was unaffected in Nr5a2 KD 8C embryos. 2–8C shared NR5A2 occupied peaks remained accessible (Fig. 4c, d, “Rarg”). However, the opening of 8C-specific NR5A2 peaks was specifically impaired (Fig. 4c, “8C ATAC” and Fig. 4d “Satb1”). Among 8C-specific ATAC-seq peaks, 73.9% showed reduced chromatin accessibility upon Nr5a2 KD. The affected regions were relatively more enriched in distal regions (78.6%) compared to those that were not affected (63.5%). These data indicate that NR5A2 is required for opening 8C-specific putative enhancers but not those opened at the 2C stage (Fig. 4c, d), consistent with NR5A2 being required for 4–8C, but not 2C, global gene activation.

NR5A2 opens a subset of regions that are bound by pluripotency TFs at later stages

NR5A2 is closely linked to pluripotency regulation.22,24,25 Therefore, we asked whether NR5A2 binding at the 8C stage may be related to the future binding of pluripotency factors such as NSO. We analyzed the binding of NANOG in blastocysts32 and the binding of NSO in mESC.33,34 Intriguingly, 8C-specific, but not 2C-specific NR5A2 binding preferentially enriched for NANOG binding in blastocyst and NSO binding in mESC (Fig. 4e). There was a small group of 2–8C shared ATAC-seq peaks enriched in Nanog binding in the blastocysts (Fig. 4e). About 1042 8C-specific NR5A2 peaks were also occupied by NANOG in blastocysts (Fig. 4f). These regions became less open upon Nr5a2 KD in the 8C embryos (Fig. 4c, “8C ATAC”). As we could not determine NANOG and OCT4 binding at the 8C embryos due to its low expression (NANOG) and the lack of sensitive antibodies that could be applied to low-input samples, we searched for footprints of TFs within the differentially accessible sites using TOBIAS.35 As a validation, the NR5A2 motif showed a significant decrease in ATAC-seq upon Nr5a2 KD. Consistently, motifs of pluripotency TFs, including NANOG and OCT4, and TEAD4 (a TE regulator) showed decreased ATAC-seq signals upon Nr5a2 KD (Fig. 4g). In addition, it was reported that ESRRB and NR5A2, which share similar binding motifs and binding sites in mESCs, redundantly regulate NSO binding and pluripotency in mESCs.27 Notably, while NR5A2 alone is dispensable for mESC self-renewal,27 it plays a much more essential role in embryos likely due to its much higher gene expression (Supplementary information, Fig. S6a). In Nr5a2/Esrrb double KO mESCs, the binding of NSO in regions that were co-occupied by ESRRB and NR5A2 (n = 2599) was severely disrupted (Supplementary information, Fig. S6b). As a comparison, NSO binding to regions that exhibited weak or no ESRRB/NR5A2 binding was less affected (Supplementary information, Fig. S6b). The dependence of NSO binding on NR5A2/ESRRB in mESCs extended to the regions that were preferentially bound by NR5A2 both in 8C embryos and mESCs, as well as by NANOG in blastocysts (Supplementary information, Fig. S6c). These data raise a possibility that NR5A2 may open a subset of regions at the 8C stage, possibly in anticipation of future binding of pluripotency TFs at later stages.

NR5A2 regulates early ICM genes and other lineage regulator genes at the 8C stage

We then sought to investigate whether NR5A2 may directly regulate the pluripotency program at the early stage. Globally, we identified 360 genes that were specifically expressed in ICM compared to TE using a reference dataset,17 and among them 62% (n = 224) were already expressed at the 8C stage (fragments per kilobase of exon per million mapped fragments (FPKM) > 1), which we termed as early ICM genes (Fig. 5a). The rest of them were termed as late ICM genes (Fig. 5a). The expected lineage genes in corresponding classes (Early ICM: Nanog, Pou5f1; Late ICM: Sox2) validated the gene list (Fig. 5a). We then examined the enrichment of NR5A2 binding near the early and late ICM genes. Intriguingly, NR5A2 showed strong binding both at promoters (Fig. 5b) and in distal regions (Supplementary information, Fig. S7a) near early ICM genes, but not near late ICM genes in 2C and 8C embryos. This was also true in mESCs, where both early and late ICM genes were expressed (Fig. 5a). Consistently, the NR5A2 motif was more enriched at the promoter or the nearby regions of early ICM genes compared to late ICM genes (Fig. 5c; Supplementary information, Fig. S7b), suggesting that NR5A2 preferentially regulates early wave of ICM genes. In line with this notion, 45 out of 224 (20%) early ICM genes were downregulated in Nr5a2 KD 8C embryos, compared to 929 out of 9744 (9.5%) non-early ICM genes expressed at the 8C stage (Fig. 5d). The downregulated, but not the upregulated, early ICM genes were more enriched for NR5A2 binding at their promoters or in their neighbor distal regions (Fig. 5e; Supplementary information, Fig. S7c), indicating that they are part of direct targets of NR5A2. NR5A2 peaks near the early ICM genes also became less accessible upon Nr5a2 KD (Supplementary information, Fig. S7d). For example, it bound pluripotency genes Oct4, Nanog, and Tdgf1 which were all downregulated in Nr5a2 KD 8C embryos (Fig. 5f). NANOG and OCT4 proteins were both decreased in Nr5a2 KD and, to a lesser extent, Nr5a2 KO 8C embryos (Fig. 5f; Supplementary information, Fig. S8a). Moreover, 8C-specific NR5A2 binding peaks with future NANOG binding were preferentially present near early ICM genes, which also exhibited decreased accessibility upon Nr5a2 KD (Fig. 4c; Supplementary information, Fig. S8b), as exemplified in Id3 (Supplementary information, Fig. S8c). Early ICM genes tended to be downregulated rather than upregulated upon Nr5a2 KD at the 8C stage, an effect that was more evident for those with NR5A2 binding than those not bound by NR5A2 (Supplementary information, Fig. S8b). Together, these results suggest that NR5A2 directly binds and promotes transcription of early ICM genes at the 8C stage including a subset of key pluripotency genes. Of note, a few key PrE regulator genes Gata6 and Fgfr1/2, and TE regulator genes, including Tead4, Klf5, Tfap2c and Gata3 were also bound by NR5A2 and downregulated in Nr5a2 KD 8C embryos (Fig. 5g), suggesting that NR5A2 may also regulate TE and PrE lineages.

Fig. 5: NR5A2 regulates the early ICM, TE and PrE genes at the 8C stage.

a Heatmap showing the gene expression levels in 2C, 4C, and 8C embryos, E3.5 ICM, TE, and mESC for the early/late ICM genes. Early ICM gene: FPKM ≥ 1 in 8C embryos. Late ICM gene: FPKM < 1 in 8C embryos. b Box plots showing the average enrichment of NR5A2 and NSO binding signals at the promoters (TSS ± 2.5 kb) of early or late ICM genes at each stage. c Box plots showing the motif numbers of NR5A2 and NSO at the promoters (TSS ± 2.5 kb) of early or late ICM genes. d Bar plots showing the percentages of downreuglated and upregulated genes among early, late ICM, or other expressed genes in Nr5a2 KD 8C embryos. e Box plots showing the average enrichment of NR5A2 signals at the promoters (TSS ± 2.5 kb) (in WT embryos) of downregulated or upregulated genes identified in Nr5a2 KD 8C embryos for early ICM genes and other 8C-expressing genes, with P values indicated. All genes were similarly analyzed and are shown as controls. f, g  The UCSC browser views showing NR5A2 binding of representative ICM/epiblast (f) or ICM/PrE and TE genes (g) in WT embryos and mESCs (top). Bar plots show expression fold change (Nr5a2 KD/control) for representative genes in 8C embryos of two biological replicates of RNA-seq (middle). Immunofluorescence of DAPI, NANOG (n = 11–12 embryos), OCT4 (n = 9–12 embryos), GATA6 (n = 13–14 embryos) or TEAD4 (n = 12–14 embryos) in mouse 8C embryos after Nr5a2 KD (scale bars: 20 μm) and quantification of signal intensity (relative to DAPI) as well as P values (t-test, two-sided) are shown. Each dot represents a single blastomere.

Interestingly, NR5A2 alone is dispensable for mESC pluripotency and self-renewal.22 Consistently, Nr5a2 KO affected limited ICM-specific genes in mESC, unlike that for SOX2 (Supplementary information, Fig. S8d), which echoed the absence of NR5A2 motif in ATAC-seq peaks and its downregulated expression in ICM and mESC (Fig. 1a, b). We reasoned that the pluripotency transcription network is likely taken over by the pluripotency TFs NSO in ICM and mESC based on the motif analysis from ATAC-seq data (Fig. 1a). Indeed, the motifs of NSO were more enriched near late ICM genes compared to early ICM genes (Fig. 5c; Supplementary information, Fig. S7b). Interestingly, while these factors clearly bound late ICM genes in blastocyst (for NANOG) or ESCs (NSO), they also bound early ICM genes as well (Fig. 5b). To ask to what extent pluripotency TFs directly regulate the early and late ICM genes, we fused FKBP12F36V sequence to the C-terminus of the endogenous SOX2 protein in 2i mESC so that SOX2 protein can be rapidly degraded by the addition of dTAG.36 We then performed RNA-seq at 24 h and 48 h after SOX2 degradation (Supplementary information, Fig. S8d). The result showed that 15.5% early and 14.6% late ICM genes were significantly downregulated upon SOX2 depletion within 1 day, and the number increased to 23.5% early and 32.3% late ICM genes within 2 days, which contrasted a negligible role of NR5A2 in transcription in mESC (Supplementary information, Fig. S8d). Together, these data indicate that while NR5A2 preferentially regulates early ICM genes, NSO regulate both early and late ICM genes. NR5A2 not only activates the transcription of pluripotency factor genes like Oct4 and Nanog, but also opens a subset of sites that will be bound by NSO at the later stage.

NR5A2 predominantly occupies the B1 repeats in the 2C and 8C embryos

One interesting question is why NR5A2 binds many early-stage genes in embryos but not in mESCs. Repetitive elements are known to harbor the repertoire of TF binding sites.37 Several classes of transposable elements such as B1, B2, and MERVL are specifically activated in mouse preimplantation embryos.17,38,39,40,41 Interestingly, NR5A2 binding sites at the 2C and 8C stages, both for those at promoters and distal regions, overwhelmingly overlapped with the short-interspersed element (SINE) family, especially with the B1 repeats (Fig. 6a; Supplementary information, Fig. S9a). For example, 84.8% and 73.3% of NR5A2 binding peaks contained B1 at the 2C and 8C stages, respectively, compared to 18.9% of random peaks (Supplementary information, Fig. S9a). B1 was abundantly expressed42 and was indicated to play critical roles in mouse early embryogenesis.40 B1 was preferentially present near 2–8C gene promoters, consistent with previous work,43 and correlated with NR5A2 binding at the 2C and 8C stages (Fig.