High-throughput functional characterization of enhancers in totipotent-like cells

Abstract

Zygotic genome activation (ZGA) marks the initial transcription event in embryogenesis, yet the cis-regulatory mechanism remains unclear. Here, utilizing massively parallel reporter assays, we functionally dissect enhancers across mouse genome in DUX-induced 2C-like cells (2CLCs). Through integrated analysis with epigenomic and transcriptomic data from 2CLCs and 2C embryos, active enhancers in totipotent cells are depicted. Among them, a notable proportion of promoters exhibit enhancer activities, showing elevated active chromatin features and correlating with enhanced gene expression during ZGA. Transcription factors preferentially enriched at enhancer-promoter regions play critical roles in regulating ZGA and early embryonic development. Furthermore, only half of the MT2_Mm exhibit enhancer activities in 2CLCs. Notably, 2CLC enhancers augment transcription in 2C embryos. Finally, deleting enhancer regions in both 2CLCs and 2C embryos, together with dCas9-KRAB-MeCP2-mediated CRISPRi in vitro, collectively underscores their crucial role in facilitating transcription of ZGA genes. These findings advance our comprehension of the cis-regulatory mechanism governing ZGA process.

Zygotic genome activation marks the first wave of gene transcription in embryogenesis, yet its regulatory mechanisms remain unclear. Here, Yang et.al mapped and functionally validated active enhancers driving this process in totipotent cells.

Introduction

During embryogenesis, the zygote is formed by the fusion of two specified gametes¹. The zygotic genome is initially transcriptionally quiescent, allowing the maternal components in the oocyte to reset the genome into a naïve and globally accessible state². Subsequently, the zygotic genome is activated to establish a so-called totipotent state, enabling the resultant blastomeres with the potential to differentiate into all cell types of an organism, through a process known as the zygotic genome activation (ZGA)³. ZGA is a critical developmental milestone as it signifies the beginning of embryonic genome exerting control over its own development^4,5, analogous to the initiation of a precisely programmed genetic code. Elucidating the mechanisms by which specific genes are selected for activation during ZGA is fundamental to understand the initiation of life. Despite its importance, the regulatory mechanisms governing mammalian ZGA remain poorly understood due to the limited materials.

Interestingly, it was reported that a small population of cells within mouse naïve embryonic stem cells (ESCs) shows the activity of MERVL (murine endogenous retrovirus with leucine tRNA primer) elements and shares hundreds of transcripts that are specifically activated in 2-cell (2C) embryos. These cells are therefore referred to as 2C-like cells (2CLCs)^6–8. Moreover, 2CLCs also exhibit chromatin architecture resembling that of the 2C embryos^9–12. These 2CLCs provide an opportunity for addressing the mechanism of ZGA. With the aid of this in vitro pluripotent-to-totipotent transition model, several critical transcriptional factors (TFs) and signaling¹³ function in mammalian ZGA have been identified^9,14–16. Recently, several works reported that high proportion of 2CLCs can be generated from pluripotent cells by either ectopic overexpression of TFs¹⁷ or culturing in a certain medium which contained epigenetic modifiers or the agonist of retinoic acid (RA) signaling^18–21. These systems expand the application of 2CLCs to study the mechanism underlying ZGA.

Enhancers, as vital cis-regulatory elements, are fundamental in orchestrating the spatially and temporally precise regulation of gene expression programs during embryonic development²². In recent years, with the rapid development of low-input, high-throughput sequencing technology, the epigenetic modification profiles of early embryos are unveiled to predict the cis-regulatory modules^23,24. Nonetheless, the prediction based on epigenetic features may lead to false positive candidates. Putative enhancers need to be functionally validated through direct transcriptional reporter assays. Self-transcribing active regulatory region sequencing (STARR-seq) is a newly-developed massively parallel reporter assay which is able to quantitatively identify transcriptional enhancers directly based on their activity in a big scale by cloning sheared DNA between a minimal-promoter and a downstream polyA sequence^25,26. If the DNA fragment exhibits enhancer activity, this will result in transcription of the enhancer sequence. In this work, we perform STARR-seq across the whole murine genome in DUX-induced 2CLCs. Combining with the epigenetic modifications and transcriptome analysis in cells and 2C embryos, we depict the potential functional enhancers in 2CLCs and ZGA process.

Results

STARR-seq dissects functional enhancers across murine genome in DUX-induced 2CLCs

In order to obtain sufficient 2C-like cells to conduct the genome-wide enhancer reporter assay, we established an in vitro DUX-induced 2CLC model. An ESC line containing a doxycycline (Dox)-inducible Dux transgene and a MERVL-LTR (LTRs, long terminal repeats) driven EGFP-F2A-puroR cassette (an indicator of the 2C-like state) was generated, which referred as iDux-MGP cells (Supplementary Fig. 1a). Puromycin was used at 6 h after Dox supplementation to help eliminate cells in which MERVL activation did not occur (Supplementary Fig. 1a). Single cell derived clones from iDux-MGP cells were then picked up, and clone 3 which exhibited high 2CLC induction efficiency and normal karyotype was used for the following experiments (Supplementary Fig. 1b, c). To induce 2CLCs, doxycycline was added for 30 h (Supplementary Fig. 1a, b). Transcriptome analysis including mRNA and long non-coding RNA (lncRNA) in sorted DUX-induced 2C::EGFP-positive (GFP⁺) and -negative (GFP⁻) cells identified 4,994 transcripts upregulated in GFP⁺ cells, including 2C specific markers such as MERVL-int, MT2_Mm, Zfp352, and Zscan4 family (Supplementary Fig. 1d, e and Supplementary Data 1). To evaluate their totipotency potential in vivo, mCherry-expressing DUX-induced GFP⁺ or GFP⁻ cells were injected into 8-cell stage embryos. Immunostaining of E4.5 blastocysts revealed that GFP⁻ cells contributed exclusively to the epiblast (SOX2⁺), whereas GFP⁺ cells were capable of contributing to both the epiblast and trophectoderm (CDX2⁺) in 6 of 21 chimeric blastocysts, demonstrating that these GFP⁺ cells retain totipotent potential (Supplementary Fig. 1f, g).

To further assess the properties of cells with or without DUX induction in a single-cell level, we performed single-cell RNA-seq (scRNA-seq) in the iDux-MGP-Clone 3 cells with or without Dox treatment. By integrating our scRNA-seq data with mouse early embryos^27,28 and other published datasets from totipotent cell types, including reported DUX-induced 2C-like cells (2CLCs), totipotent-blastomere like cells (TBLCs) and chemically induced totipotent stem cells (ciTotiSCs)^17,20,21, we found that both DUX-induced 2CLCs and ciTotiSCs were mainly clustered with the cells of 2–4 cell (2C-4C) embryos, with a subset of DUX-induced 2CLCs close to an intermediate population reported previously¹⁷ (Supplementary Fig. 1h). In contrast, the -Dox cells mainly centered near epiblast and mESCs (Supplementary Fig. 1h). Furthermore, unsupervised clustering analysis identified six distinct clusters. It showed that 59.7% of the -Dox group cells were in Cluster 1, which epiblast and mESCs were located (Supplementary Fig. 1i). Only 4.51% of the -Dox cells were in Cluster 2, where the 2C-4C embryos and reported 2CLCs located. Conversely, 59.33% of the +Dox cells were located in Cluster 2, while only 4.46% were in Cluster 1 (Supplementary Fig. 1i). The gene expression UMAP plot indicated that GFP⁺ cells mainly located in Cluster 2 and enriched 2C markers such as Zscan4c, Zscan4d, and Ddit4l (Supplementary Fig. 1j). These data demonstrate that our induction system effectively generated a high proportion of 2CLCs, which can be used for the further studies.

To identify potential functional enhancers across the whole murine genome in totipotent-like cells, we constructed a STARR-seq plasmid library by inserting sonication-sheared genomic DNA at the average size of 700–1200 bp into the optimized STARR-seq vector²⁹. The plasmid library was then transfected into the cells at 24 h after the Dox addition and cells were collected at 16 h after the transfection (Fig. 1a). The plasmids and the mRNAs transcribed from the library were isolated respectively from the transfected cells for sequencing, and the plasmids were set up as the input (Fig. 1a). Two independent experiments were performed. Input DNA coverage analysis revealed adequate representation of the mouse genome in our library, with one replicate surpassing 4 times of the genome coverage and the other exceeding 11 times (Supplementary Fig. 2a). Moreover, Pearson’s correlation analysis indicated strong reproducibility between the two replicates of the STARR-seq data (Supplementary Fig. 2b). Fragment sequences which exhibited self-transcription levels at least 2-fold higher than those of the input were defined as STARR peaks, which were significantly enriched over the input compared to the randomly selected fragments from genome (Supplementary Fig. 2c). The sizes of STARR peaks corresponded with those of the input, predominantly falling within the range of 700–1200 bp, without interference from smaller fragment noise (Supplementary Fig. 2d). In addition, we generated ATAC-seq and ChIP-seq libraries (targeting H3K4me1, H3K4me3, and H3K27ac) in sorted GFP⁺ and GFP⁻ cells following DUX induction, aiming to combine the analysis of STARR peaks with the chromatin status and epigenetic modifications in 2CLCs (Supplementary Fig. 2e, f).

Fig. 1. STARR-seq dissects functional enhancers across murine genome in 2CLCs.

a Schematic of whole-genome STARR-seq in DUX-induced 2C::EGFP positive cells. b Heatmap showing the differential STARR peaks between 2CLCs and mESCs, and combined with indicated multi-omics analysis of chromatin characteristics in 2C::EGFP negative (GFP-) and 2C::EGFP positive (GFP + ) cells (see Supplementary Fig. 2e). Signal was computed on the STARR-seq peaks flanked by 10 kb. 2C.spe: specific STARR peaks in 2CLCs; ESC.spe: specific STARR peaks in mESCs; Share: STARR-peaks in both 2CLCs and mESCs. The number of peaks in each group is indicated. c Boxplot showing the overall signal intensity of indicated marks in Fig. 1b (2C.spe peaks: n = 3452; ESC.spe peaks: n = 11,668; Share peaks: n = 9908.) in indicated cell types. Boxplots denote the medians and the interquartile ranges (IQR). The whiskers of a boxplot are the lowest datum still within 1.5 IQR of the lower quartile and the highest datum still within 1.5 IQR of the upper quartile. Padj was from two-sided Mann-Whitney U test with Benjamini-Hochberg correction. d A genomic snapshot showing STARR-seq, ATAC-seq, ChIP-seq of the specified histone modification with input, and RNA-seq tracks at the loci of Zscan4 family in indicated cells. e Pie charts showing the genomic distribution of the STARR-peaks in Fig. 1b. f Boxplot showing the expression levels of the genes (n = 4059) flanked by 2C-specific (2C.spe) STARR peaks within 20 kb (± 20 kb) in 2C::EGFP negative (GFP-) and 2C::EGFP positive (GFP + ) cells. Data are presented as mean values ± SEM. The center line represents the median value, the lower and upper hinges correspond to the first and third quartiles (the 25th and 75th percentiles), the whisker extends from the hinge to the largest or smallest value at most 1.5 * IQR (inter-quartile range) of the hinge. P-values were from a two-sided Mann-Whitney U test.

To identify the 2C-specific enhancers, we conducted an integrative analysis of enhancers in 2CLCs and mESCs using a published STARR-seq dataset of mESCs (GSE143546)³⁰. Given that enhancers detected by transient-assay are based on exogenous plasmids, and chromatin is a complex structure, highly active enhancers typically reside in open and active chromatin regions to effectively regulate target genes³¹. Therefore, we firstly classified enhancers based on their STARR activities and chromatin opening status. Among them, putative enhancers exhibiting both STARR activities and locating in open chromatin regions in either 2CLCs or mESCs were defined as Cluster 1 (C1). In contrast, elements with STARR activities yet residing in close chromatin regions in the respective cell types were defined as C2, suggesting that their enhancer activities probably be suppressed or poised in cells³⁰. In addition, elements located in active chromatin regions without STARR activities were defined as C3, implying that these regions although had the active epigenetic modifications, might not act as enhancers in cells (Supplementary Fig. 3a). Therefore, our subsequent functional enhancer analysis in 2CLCs focused primarily on the STARR peaks in C1.

Subsequently, the differential STARR peaks were calculated between the 2CLCs and mESCs in cluster C1 (n = 25,028). Using a 2-fold threshold, we identified 3452 2C-specific enhancers (2C.spe), 11,668 ESC-specific enhancers (ESC.spe) and 9908 shared enhancers in 2CLCs and mESCs (Share) (Fig. 1b and Supplementary Fig. 3b, c). Integrated analysis with epigenomic data indicated that 2C-specific enhancers exhibited elevated signals of ATAC, H3K27ac, H3K4me1 and H3K4me3 in DUX-induced GFP⁺ cells, such as the STARR peaks within the Zscan4 family gene loci (Fig. 1b–d). Whereas, ESC-specific enhancers were associated with increased signals in GFP⁻ cells (Fig. 1b, c). Moreover, 2C-specific TFs, DUX and OBOX1/3/5, exhibited much higher binding signals at the 2C-specific and shared peaks than ESC-specific peaks (Supplementary Fig. 3d). These results suggested that the enhancer sequences identified via STARR-seq align with the chromatin features of enhancers and cell state in both cell types. Genome annotation revealed a significant enrichment of transposable element (TE) components in the three groups of enhancers (Fig. 1e). Notably, over 20% of enhancers in 2CLCs were located within promoter regions, while only 3.3% of ESC-specific enhancers were situated in promoter regions (Fig. 1e). Moreover, among the genes upregulated in DUX-induced GFP⁺ cells, those promoter regions with enhancer activities had higher levels of active epigenetic modification, P300 signals, and related gene expression, compared to those without enhancer activities (Supplementary Fig. 4a, b). Furthermore, we found that the expression levels of genes around 2C-specific enhancers (± 20 kb) were significantly higher in DUX-induced GFP⁺ cells (Fig. 1f), especially for non-coding RNAs (Supplementary Fig. 4c), suggesting their functional role in regulation of gene expression in 2CLCs.

Enhancers encompass the binding sites to recruit cell-type specific transcription factors (TFs) for facilitating the initiation of transcription³². We then conducted a motif analysis within each class of C1 enhancers to examine the TFs potentially involved in these interactions. It revealed that ESC-specific enhancers were enriched for pluripotency TFs, such as OCT4, SOX2 and members of KLF family. Shared enhancers were enriched some pan-TFs, such as PRDM14 and YY1. Whereas, 2C-specific enhancers were particularly enriched the TFs highly expressed during the ZGA process, including TEAD4, DUX and GATA2, with the latter two playing vital roles in activating ZGA genes in previous studies^9,14,33,34 (Supplementary Fig. 4d, e). These findings underscore the distinct characteristics between 2C-specific and ESC-specific enhancers. Taken together, we identified in total 13,360 enhancer elements in 2CLCs by using STARR-seq in conjunction with multi-omics analysis, and further examined the distinct characteristics between 2C-specific and ESC-specific enhancers.

Evaluation of the in vivo functional potential of enhancers identified in 2CLCs

To investigate the in vivo functional potential of the enhancers we identified in 2CLCs, we first examined the expression profiles of genes associated to the 2C-specific (2C.spe) enhancers (± 20 kb) in embryos, using previously published transcriptome data from different stages of embryos². The results showed that the associated genes of 2C-specific (2C.spe) enhancers exhibited a significant increase in expression at early 2C (E2C) stage, followed by a striking decrease in expression in ICM (Supplementary Fig. 5a). Subsequently, we conducted an integrated analysis of the three classes of C1 enhancers with the published ATAC-seq data in early embryos² (Fig. 2a). Interestingly, consistent with their differential activities in vitro, enhancers identified in 2CLCs (2C.spe and share) had higher levels of transposase-accessibility in the E2C and 2C stages compared to the ESC-specific enhancers. Whereas, the ATAC-seq signals associated with ESC-specific enhancers were much lower compared to the enhancers identified in 2CLCs in E2C and 2C stage. Interestingly, while the signals of 2C-specific enhancers progressively decreased to their lowest point in ICM after they reached their culmination at 2C stage, the ATAC-seq signals associated with ESC-specific enhancers dramatically increased at 8-Cell stage and maintained their high levels in the ICM of embryos (Fig. 2a). Specifically, 42% of 2C-specific enhancers and 22% of shared enhancers were in an open status in 2C embryos, while only 17% of ESC-specific enhancers were located in the open chromatin and with a lowest level of ATAC-seq signal in 2C embryos (Supplementary Fig. 5b, c). These results suggest the in vivo functional potential of the enhancers identified in 2CLCs.

Fig. 2. Evaluation of the in vivo functional potential of enhancers identified in 2CLCs.

a Heatmap showing ATAC-seq signals of the indicated groups of STARR peaks (±5 kb) during mouse early embryonic development. b Venn diagram illustrating the overlap between ZGA genes and genes flanked by 2C-specific (2C.spe) or shared (Share) STARR peaks within 20 kb. c The genomic distribution of 2C-specific (2C.spe) and shared (Share) STARR peaks in proximity to ZGA genes, defined with STARR peaks located within 20 kb of ZGA gene loci. d The expression levels of minor (top) and major (bottom) ZGA genes categorized by the presence (Promoter.STARR) or absence (Promoter.noSTARR) of STARR-seq signals at their promoter regions in 2CLCs (n = 1116 minor and 1392 major ZGA genes within Promoter.STARR; and n = 3628 minor and 2054 major ZGA genes within Promoter.noSTARR). Data are presented as mean ± SEM. The center line represents the median value, the lower and upper hinges correspond to the first and third quartiles, the whisker extends from the hinge to the largest or smallest value at most 1.5 * IQR of the hinge. Padj was from two-sided Mann-Whitney U test with Benjamini-Hochberg correction. e The CpG island distribution of the promoter sequences with or without STARR-seq signals. f The ATAC-seq, H3K27ac, H3K4me1 and H3K4me3 signals at the indicated promoter regions (TSS ± 5 kb) of minor (top) or major ZGA genes (bottom) in 2C embryos, categorized by the presence (Promoter.STARR) or absence (Promoter.noSTARR) of STARR-seq signals in 2CLCs. E2C: early 2-cell embryos; L2C: late 2-cell embryos. g Heatmap showing TF motifs identified from ZGA gene promoters (TSS ± 1 kb), categorized by the presence (Promoter.STARR) or absence (Promoter.noSTARR) of STARR-seq signals in 2CLCs. The color indicated the enrichment of the motifs (-log10 P-value). h Representative bright-field microscopic images of mouse embryos injected with indicated siRNAs. Scale bar: 100 µm. i The blastocyst formation rates (%) of mouse E4.5 embryos injected with negative control (siNC) or siRNAs targeting Nfya (siNfya), Gabpa (siGabpa), and Zbtb33 (siZbtb33). Values are means ± SD, n = 3 biological replicates. Two-sided unpaired t tests were performed. Source data are provided as a Source Data file.

To further investigate potential target genes of the identified 2C enhancers in 2C embryos, we first defined 4744 minor and 3446 major ZGA genes by analyzing differentially expressed genes between E2C/Zygotes and MII oocytes as well as between late 2C (L2C) and early 2C (E2C) embryos respectively with previously published data² (Supplementary Fig. 5d, e and Supplementary Data 2). In total, 3460 ZGA genes were associated to the enhancers identified in 2CLCs (2C.spe and share), up to 51% (3460/6779) of the total ZGA genes (Fig. 2b). Notably, genome annotation of the STARR-peaks in 2CLCs around ZGA genes (± 20 kb) revealed that 75.69% of the peaks were located in the gene promoter regions, with 59.58% of these genes being ZGA genes (Fig. 2c and Supplementary Data 3). Given that the 3D chromatin structure has not been fully established in mouse 2C embryos^35,36, which results in the weakening regulation of distal enhancers, especially in minor ZGA³⁷, those enhancer-promoters might be an alternative way in early embryos to promote gene transcription. Indeed, ZGA genes with enhancer-promoters had a significant higher expression level compared to those whose promoters had no enhancer activities (Fig. 2d). Previous works suggested that promoters of some housekeeping (HK) genes exhibit enhancer activities in flies and pluripotent stem cells (PSCs)^26,38. Nevertheless, with a reference of HK genes in mouse³⁹, our data revealed that 60.18% promoters of ZGA genes are non-HK genes. Notably, these genes included the critical TFs during ZGA process, such as Dux, Ddit4l, Zfp352 and Zscan4 family (Supplementary Fig. 5f and Supplementary Data 3). Moreover, GO analysis revealed that the non-HK ZGA genes with enhancer-promoters predominantly enriched in ribosome biogenesis and nuclear division, which were essential for cell survival and important for maternal-to-zygotic transition (MZT) (Supplementary Fig. 5g).

Unlike gene promoters, enhancers are cis-regulatory DNA elements that activate transcription of target genes in an orientation- and position-independent manner⁴⁰. As genomic fragments were inserted in a random orientation in our STARR library, the enhancer activities of both forward and reverse DNA fragments could be tested. We therefore performed strand-specific analysis of the proximal-promoter STARR peaks, particularly those related to ZGA genes. The results showed that the enhancer activities exhibited no detectable strand bias in either direction (forward and reverse), supporting an orientation-independent activity of these elements (Supplementary Fig. 6a). To investigate what distinguishes the enhancer-promoters from other promoters of ZGA genes, we first characterized their sequences. While both classes of promoters exhibited similar level of GC content (Supplementary Fig. 6b), 78% enhancer-promoters contained CpG islands (CGI), in contrast only 28% of promoters without STARR activities had CGI (Fig. 2e). Subsequently, we compared the chromatin status² and epigenetic modifications^23,24 between these two different groups of promoters. The results revealed that promoter regions with enhancer activities of both minor or major ZGA genes exhibited much higher levels of ATAC-seq signal in 2C embryos compared to promoters without STARR activities (Fig. 2f). Moreover, the enhancer-promoter regions displayed much higher levels of H3K27ac and H3K4me3 in E2C and L2C embryos than those without STARR peaks (Fig. 2f). Nevertheless, as the global H3K4me1 levels were weak in 2C embryos (Supplementary Fig. 6c), the levels of H3K4me1 were low in both promoters of ZGA genes in 2C embryos (Fig. 2f). Moreover, compared with promoters lacking STARR-seq activity, enhancer-promoters exhibited markedly higher occupancy signals for TFs known to play critical roles in ZGA in vivo, including DUX, OBOX and NFY family members (Supplementary Fig. 6d). These findings implicate these 2C-specific TFs in the regulation of enhancer-promoter activity.

To further explore the potential mechanism underlying the enhancer-promoters, a motif analysis was performed. Interestingly, Additional TFs preferentially enriched in the enhancer-promoters were identified⁴¹, including NRF1 and NFY, the two well-known transcription activators^42–44, as well as GABPA, P53 and DUX, the key regulators of ZGA genes^45,46 (Fig. 2g and Supplementary Data 4). Notably, most of these TFs specifically highly expressed in MII oocytes or during ZGA process (Supplementary Fig. 6e). The function of several enhancer-promoter-enriched TFs (NFYA, GABPA, and ZBTB33) was evaluated by injecting siRNA into zygotes. Consistent with previous reports, knockdown of Nfya (to nearly 50% of non‑targeting control levels) and Gabpa (to nearly 20% of control levels) impaired early embryonic development, reducing the blastocyst formation rate from 92.7% (control) to 23.8% and 57.1%, respectively (Fig. 2h, i and Supplementary Fig. 6f, g). Interestingly, knockdown of Zbtb33 also reduced the blastocyst formation rate to 66.7%, suggesting its previously undefined role in early embryonic development (Fig. 2h, i and Supplementary Fig. 6f, g). Furthermore, qPCR analysis of late 2‑cell embryos revealed that knockdown of Gabpa showed no significant effect on 2C‑specific gene expression. This aligns with published data indicating that GABPA degradation downregulates 589 genes at the 2‑cell stage involved in processes such as ribosome biogenesis and rRNA processing⁴⁶. In contrast, knockdown of Nfya and Zbtb33 downregulated 2C‑specific genes, including Dux and MERVL-pol (Supplementary Fig. 6g). Moreover, Zbtb33 downregulation also reduced the expression levels of Zscan4 and Zfp352, suggesting its potential role in the regulation of ZGA (Supplementary Fig. 6g).

MERVL elements with enhancer activities were dissected in totipotent cells

It was reported that transposable elements (TEs), especially MERVL elements, play an important cis-regulatory function in the regulation of ZGA genes, and therefore, are critical for embryonic development^{6–8,47–49}.To investigate the enhancer function of TEs, we first annotated the TE families within the 2C-specific and ESC-specific enhancers. The results showed that the ERVL (ERV3) family, such as MT2_Mm, MT2C_Mm and ORR1A1-int, were predominantly enriched in the 2C-specific enhancers (Fig. 3a–c). While ESC-specific enhancers were primarily enriched elements of ERVK and ERV1, like ETnERV, MMETn-int and LTRIS2 (Fig. 3a, b and Supplementary Data 5). These results indicated different TE clusters were adopted as enhancer elements in these two different cell types.

Fig. 3. MERVL elements with enhancer activities were dissected in totipotent cells.

a The enrichment of TEs in the 2C-specific (2C.spe) and ESC-specific (ESC.spe) STARR peaks in Fig. 1b. The fold enrichment was calculated by the ratio of the observed over expected counts of the modified TEs among randomly selected genomic regions. b Read count tag density pileups (in RPKM) of the STARR-seq signals for the selected TEs in 2CLCs and mESCs. The TEs were scaled to the same size, and the flanking 10 kb regions are shown. c The STARR-seq, ATAC-seq, ChIP-seq of the specified histone modification with input, and RNA-seq tracks at the indicated regions with genes and TEs in the indicated cells. The 2C-specific enhancers are highlighted (Yellow square). d The classification of all MERVL elements into three clusters based on STARR-seq and ATAC-seq signals in 2CLCs (left). MERVL.C1: cluster comprises MERVL elements showing both high STARR-seq signals and ATAC-seq signals in MERVL-GFP⁺ cells. MERVL.C2: cluster includes MERVL elements with STARR-seq signals and minimal ATAC-seq signals in MERVL-GFP^- cells. MERVL.C3: cluster consists of MERVL elements exhibiting low or no STARR-seq signals in 2CLCs. The right panel displays the indicated histone modifications associated with these three clusters of MERVL elements. e The expression levels of genes flanked by the MERVL.C1 elements in Fig. 3d within 20 kb. The X-axis represents the expression levels in GFP^- cells, while the Y-axis denotes the expression levels in MERVL-GFP⁺ cells. f The genome distribution of the MERVL.C1 elements in Fig. 3d. g Length distribution of the three clusters of MERVL elements in Fig. 3d. h Heatmap showing MT2_Mm from the corresponding MERVL clusters (as shown in Fig. 3d) were classified into four clusters based on their ATAC-seq signals in L2C embryos (left). The binding of indicated transcription factors (TFs) in these four clusters of MT2_Mm elements were also shown (right). i The differential expressed genes between MT2_Mm inhibited late 2-Cell (L2C) embryos and control embryos within the MERVL.C1 associated ZGA genes (n = 315) in Supplementary Fig. 7i. MT2_Mm were perturbed by injection of dCas9-KRAB with six MT2_Mm- targeting gRNAs. Control embryos were injected dCas9-KRAB with non-targeting gRNAs.

Considering that there are nearly 4,472 sequence-conserved copies of MERVL and their LTRs within the mouse genome according to the RepeatMasker, the locations and characteristics of the MERVL elements with enhancer activities are still enigmatic. To this end, we further dissected the MERVL elements including MERVL internal regions (MERVL-int), and their LTRs (MT2_Mm) in our STARR peaks. Although the sequences of MERVL elements are highly repetitive, we found that not all MERVL repeats had enhancer activities (Supplementary Fig. 7a). Consequently, we divided the MERVL elements into three clusters based on their STARR activities and chromatin openness in 2CLCs (Fig. 3d). Cluster 1 MERVL elements (MERVL.C1, n = 1684) exhibited high STARR activities, and co-localized with the transposase-accessible regions in 2CLCs, which were considered as functional enhancers in 2CLCs. Cluster 2 (MERVL.C2, n = 744) represented MERVL repeats with STARR-seq signals but low chromatin openness in 2CLCs, suggesting that although these repeats had enhancer potential, they were unlikely able to function as active enhancers in 2CLCs. Cluster 3 (MERVL.C3, n = 2044) represented MERVL repeats with nearly undetectable STARR-seq signals in 2CLCs, and were unlikely to function as enhancers (Fig. 3d). Consistent with their function in 2CLCs, MERVL.C1 had much higher levels of chromatin openness and active histone modifications, including H3K27ac, H3K4me1 and H3K4me3, in GFP⁺ cells compared to those of MERVL.C3 (Fig. 3d and Supplementary Fig. 7b). Moreover, 56% (629/1,115) of MERVL.C1 associated genes (± 20 kb) were upregulated in DUX-induced GFP⁺ cells, including Zscan4 and Usp17 families (Fig. 3e and Supplementary Data 6).

To characterize the MERVL elements with enhancer activities, genome annotation revealed that majority of them were situated in distal intergenic regions (59%) and introns (31.6%), with 5.2% of them locating in annotated gene promoters (Fig. 3f), suggesting their distal regulatory function. Size distribution analysis revealed that except a small fraction of elements measuring nearly 100 base pairs (bp), majority of MERVL.C1 and C2 had an average size of around 500 bp, a size which is similar to the MERVL LTRs (MT2_Mm), suggesting MT2_Mm represents the majority of MERVL repeats with enhancer activities (Fig. 3g). Indeed, MT2_Mm represented 73.04% (1230/1684) and 70.16% (522/744) of the MERVL.C1 and MERVL.C2 respectively, whereas 42.27% in MERVL.C3 (Supplementary Fig. 7c). Further analysis revealed that the truncated MERVL-int (~ 100 bp) in MERVL.C1 were predominantly located near the LTRs with STARR activities, with 89.1% nearby LTRs in MERVL.C1 and 8.9% in MERVL.C2 (Supplementary Fig. 7d, e). These suggest that the detection of their enhancer activities is attributed to their spatial association with MT2_Mm. In addition, MT2_Mm in MERVL.C1 and MERVL.C2 consisted of either solo-LTR or the 5′ and 3′ LTRs of full length MERVL with no obviously biased. Whereas, MERVL.C3 had more solo LTRs (Supplementary Fig. 7f).

As our analysis identified MT2_Mm as the primary elements of MERVL exhibiting enhancer functions, to investigate the enhancer potential of MERVL in vivo, we performed integrated analysis of MT2_Mm in the three MERVL clusters with published ATAC-seq in 2C embryos² (Fig. 3h). The results showed that almost all MT2-Mm in MERVL.C1 (designated as MT2_Mm.C1) had high chromatin openness in 2C embryos. Interestingly, despite MERVL.C2 exhibiting low ATAC-seq signal intensity in vitro, the majority of MT2_Mm within MERVL.C2 (designated as MT2_Mm.C2) displayed ATAC-seq signals in 2C embryos, a bit lower than those of MT2_Mm.C1, suggesting their potential functionality in vivo. In addition, albeit the absence of STARR activities of MERVL.C3 in 2CLCs, there were two distinct subpopulations of LTRs within MERVL.C3. Notably, 67.7% of these (designated as MT2_Mm.C3.1) showed chromatin accessibility in 2C embryos, while the other subpopulation (designated as MT2_Mm.C3.2) exhibited no chromatin accessibility in vivo (Fig. 3h and Supplementary Fig. 7g). Consistent with their chromatin accessibility in vivo, the LTRs which showed high intensity of ATAC-seq signals were also bounded by DUX in 2CLCs⁹ and OBOX1/3/5 in 2C embryos⁵⁰. As a control, they were not bound by the OBOX2 and DNA-binding mutant OBOX5 (R98E) (Fig. 3h and Supplementary Fig. 7g). Given the highly conserved sequences of MT2_Mm (MERVL-LTRs), it is noteworthy to investigate the reasons why the LTRs within MT2_Mm.C3.2 were not bound by DUX and OBOX. Sequence alignment analysis revealed that while MT2_Mm with chromatin accessibility in 2C embryos were of full length, the MT2_Mm within MT2_Mm.C3.2 were truncated mutants, with most of them containing less than the first 150 bp of LTRs, lacking the DUX and OBOX binding motifs (Supplementary Fig. 7h). This might explain why this part of LTRs (MT2_Mm.C3.2) had no active enhancer function and low chromatin accessibility. Furthermore, 315 ZGA genes were associated to the MERVL.C1 (within ± 20 kb) (Supplementary Fig. 7i and Supplementary Data 6). Notably, half of them were downregulated upon the MT2_Mm perturbed by dCas9-KRAB with 6 small-guide RNAs (sgRNAs)⁴⁸, suggesting a direct regulatory influence of MT2_Mm on the expression of these genes (Fig. 3i and Supplementary Data 6).

Verification of the 2C specific enhancers in vitro and in vivo

To further substantiate the function of the enhancers identified in 2CLCs in vivo, we performed an ultra-low-input STARR-seq (ULI-STARR-seq) in 2C embryos. This protocol had been specifically optimized from UMI-STARR-seq protocol⁵¹ for low-input samples with an enhanced RNA recovery method. Due to the restricted availability of early embryos, it is unfeasible to evaluate all enhancers in vivo. Therefore, 88 candidates from the enhancers identified in 2CLCs (2C.spe or share), with their location proximal to ZGA-related genes such as Zscan4f (seq19) and Usp12 (seq54) were synthesized and cloned into the STARR-seq plasmid (Fig. 4a and Supplementary Fig. 8a). Additionally, 9 DNA fragments without STARR and ATAC peaks in 2CLCs were cloned to serve as negative controls (Fig. 4a and Supplementary Fig. 8a). The constructed STARR-seq plasmid library containing both the candidate enhancers and negative controls, which was injected into the male pronucleus of zygotes, and self-transcriptional RNAs were then collected for sequencing at the L2C stage (Fig. 4a). Two independent experiments were conducted. It is noted that although activity was detected in two negative controls, it was not reproducible across replicates (Supplementary Data 7). The 7 remaining controls displaying no enhancer activities in both replicates were used for the following analysis. The results revealed that 69% of the candidate enhancers (61 out of 88) exhibited enhancer activities in both replicates, indicating their functional enhancer capacity in 2C embryos (Fig. 4b, c and Supplementary Fig. 8b and Supplementary Data 7). To cover the complexity of the fragments, in the library, the selected 88 candidates encompassed both TE and non-TE elements with diverse genome distribution (Supplementary Fig. 8c, d). Notably, a comparable proportion of LTRs, and enhancer-promoter candidates showed activities in 2C embryos, further validating their enhancer function in vivo (Supplementary Fig. 8e, f).

Fig. 4. Verification of the 2C specific enhancers in vitro and in vivo.

a Schematic of ULI-STARR-seq in mouse L2C embryos. b The fold enrichment of enhancer candidates relative to the negative control in L2C samples. The Y-axis shows the fold change divided by the median of negative control (yellow dots); values greater than 1.5-fold are considered positive enriched (green dots). c The intensity of STARR-seq signals for enhancer candidates (n = 88) and negative controls (n = 7) in 2CLCs (left) and L2C embryos (right). Heatmap of relative STARR-seq signals across 2CLC and L2C replicates, with enhancer candidates and negative controls as rows. Enhancer candidates (n = 61) that exhibited positive enrichment (+) in both replicates of the L2C samples are indicated. d The RT-qPCR analysis of the indicated gene expression levels in the DUX-induced GFP⁺ cells, including control (Ctrl) cells and different clones with homogeneous knockout of the potential enhancer regions highlighted in Supplementary Fig. 10a. eKO: enhancer knockout. Ctrl: cells transfected Cas9 plasmid with no gRNA. For each gene, Cas9 without gRNA from the same transfection batch was used as its control. Values are means ± SD, n = 3 biological replicates. Two-sided unpaired ttests were performed. e Schematic illustration of the procedure of nuclear transfer with control (Ctrl) mESCs or enhancer knockout (eKO) clones in Fig. 4d. f The RT-qPCR analysis of the indicated gene expression levels in mouse L2C embryos generated from nuclear transfer with control (Ctrl) mESCs or enhancer knockout (eKO) clones in Fig. 4d (indicated with same color). Ctrl: cells transfected Cas9 plasmid with no gRNA. Values are means ± SD, n = 3 biological replicates. Two-sided unpaired t tests were performed. g The RT-qPCR analysis of the indicated gene expression levels in the DUX-induced GFP⁺ cells, including control (Ctrl) cells and multiple cell lines that utilize the CRISPRi to interfere with the potential enhancers of corresponding genes. Ctrl: cells were transfected with the dCas9-KRAB-MeCP2 plasmid alone. eCRi: enhancer interference by CRISPRi. Values are means ± SD, n = 3 biological replicates. Two-sided unpaired t tests were performed. Source data are provided as a Source Data file.

To further validate the activities of the enhancers identified in the mini-STARR screening, fluorescence reporter plasmids were constructed. We first cloned 6 ESC-specific enhancers including Oct4 distal-enhancer (Oct4-DE) into a reporter plasmid containing a minimal promoter upstream of the EGFP coding sequence. The enhancer elements were inserted into the downstream of EGFP coding sequence. Upon transfection into mESCs, these ESC-specific enhancers markedly increased GFP intensity compared with the empty vector (Supplementary Fig. 9a, b). Nevertheless, no obvious GFP enhancement was observed in 2C embryos when each plasmid containing ESC-specific enhancers was injected into zygotes (Supplementary Fig. 9c, d). Subsequently, we evaluated the function of 2C-specific enhancers using the same fluorescence reporter assay. We inserted control elements and 10 candidate 2C-specific or shared enhancers into the reporter plasmid, including four enhancer-promoters (Supplementary Fig. 9e, f). Each plasmid along with a constitutively expressed mCherry tracer plasmid were co-injected into zygotes. Fluorescence was observed and quantified at 24 h post-microinjection. The results confirmed that all the 10 candidate enhancers drove significantly higher GFP expression compared to the negative controls or the promoter elements showing no STARR activities in 2CLCs (promoters of Duxbl1, Obox6, Klf4, Pou5f1) (Supplementary Fig. 9e, f). In contrast, transfection of vectors containing 2C-specific enhancers into mESCs did not yield obvious GFP enhancement, indicating their activity is unique to 2C embryos. (Supplementary Fig. 9g).

Next, we excised three enhancer regions located near the ZGA genes, Ercc4 (Excision repair cross-complementing rodent repair deficiency, complementation group 4), Stox1 (Storkhead box 1) and Obox6 (Oocyte specific homeobox 6), to assess their functional roles (Supplementary Fig. 10a). These three genes were not only highly expressed in DUX-induced GFP⁺ cells (Supplementary Fig. 10b), but also showed elevated expression during ZGA process (Supplementary Fig. 10c). For each enhancer region, a pair of sgRNAs was employed for the genomic deletion. Homozygous knockout (KO) clones derived from single cells for each enhancer were expanded and utilized for the further analysis (Supplementary Fig. 10d). RT-qPCR analysis revealed that deletion of these enhancers resulted in a significant reduction in the expression levels of their presumed target genes, Ercc4, Stox1, and Obox6 in 2CLCs compared to the control cells transfected with Cas9 alone. (Fig. 4d and Supplementary Fig. 10d). It is noted that the expression levels of Dux and MERVL-Pol remained unaffected, eliminating the possibility that the reduction in the expression of Ercc4, Stox1, and Obox6 was due to a failure of Dux-mediated 2CLC induction (Fig. 4d).

To further investigate whether the deletion of the three enhancers could affect the expression of their target genes in vivo, one of the homozygous knockout clones of each enhancer was selected to conduct nuclear transfer (NT). NT-L2C embryos were subsequently collected for gene expression analysis (Fig. 4e). While the developmental rate of Ercc4-/Stox1-/Obox6-eKO 2C embryos had no apparent difference, the expression levels of Ercc4 and Stox1 were significantly reduced in their respective enhancer knockout (eKO) embryos, compared to the NT embryos derived from control (ctrl) cells (Fig. 4f and Supplementary Fig. 10e, f). Whereas, the expression of Obox6 showed minimal changes (Fig. 4f). Considering that 2C embryos exhibit a notably higher expression level of Obox6 compared to 2CLCs⁵⁰, we speculated that compensatory mechanisms, probably involving other unidentified enhancers, are responsible for regulating Obox6 expression in vivo⁵⁰.

To further validate the function of these three enhancers, we deleted a 2C-specific enhancer that is not associated with any genes within at least ± 45 kb as a negative control (Supplementary Fig. 11a-c). RT-qPCR analysis showed no changes in the expression levels of either 2C marker genes or the genes Ercc4, Stox1, and Obox6 in this enhancer-deleted cells, both in vitro and in the NT-embryos in vivo (Supplementary Fig. 11a–c). Furthermore, we targeted 12 2C-specific distal enhancers, including enhancers of Ercc4 and Stox1 by gRNAs/dCas9-KRAB-MeCP2. We generated stable transgenic cell lines in DUX-inducible cells and set cells transfected with dCas9-KRAB-MeCP2 without gRNAs as control. GFP positive cells were collected and qPCR was performed to assess target gene expression. The results revealed that 10 of 12 enhancers targeting with gRNA/dCas9-KRAB-MeCP2 significantly reduced the expression levels of their predicted target genes, including Zfp809, Foxa1 and Spz1. Only two showed no observable effect on expression of the potential target genes (Lmx1a and Slc35e3) (Fig. 4g). These findings indicated that the enhancers identified in 2CLCs play an important role in regulating their target ZGA genes both in vitro and in vivo.

Discussion

In this study, we conducted a comprehensive analysis of the entire mouse genome to pinpoint the functional enhancers in DUX-induced 2CLCs. Although 2CLCs partially mimic 2C embryos, our integrated analysis of epigenetic modification and transcriptome during ZGA has enabled us to shed light on the cis-regulatory elements governing gene expression during the ZGA process.

After fertilization, the quiescent zygotic genome entered into an open status without detectable topologically associated domains (TADs)^2,12,35,36. Previous studies suggested that the first zygotic transcription is a promiscuous process⁵². Enhancers might be dispensable for transcription in zygotes³⁷ and become essential when chromatin is proposed to progressively adopt a repressive state during ZGA. Notably, in our data, we identified a group of promoters of minor ZGA genes, including Dux, Zfp352, Zscan4, have enhancer activities. Those enhancer-promoters have high levels of open chromatin-related histone modifications both in vitro and in vivo and their corresponding genes show higher expression compared to the other transcripts. Moreover, the introduction of STARR constructs and fluorescence reporter plasmids containing those enhancer-promoter regions into embryos demonstrated their role as enhancers in 2C embryos. Considering that minor ZGA is critical for the proper ZGA process, probably by producing TFs to recruit Pol II to major ZGA-related targets^4,53. Rather than merely a promiscuous process, our results support the presence of an alternative mechanism which is independent of long-distance enhancers, facilitates a burst in gene expression during ZGA. Interestingly, the ZGA specific enhancer-promoters exclusively enriched binding motifs of both canonical transcription activator (including NRF1, YY1 and NFY) and 2C-specific TFs. Previous work showed that NFYA, a sequence-specific DNA binding subunit of NFY complex which forming a histone-like structure binding to DNA, promotes establishment of DNase I-hypersensitive site (DHS) in 2C embryos and is essential for the ZGA^43,54,55, suggesting that those enhancer-promoters might play a role in establishing of the chromatin structure in early embryos. In the present study, we evaluated the roles of three enhancer-promoter-enriched TFs in ZGA and early embryonic development. Notably, in addition to NFYA⁴³ and GABPA⁴⁶, which have been previously reported in this context, we identified ZBTB33 as a potentially regulator of ZGA and early embryonic development. Zbtb33 (also known as Kaiso) encodes a BTB/POZ-domain zinc finger transcription factor and is expressed across a broad range of tissues⁵⁶. Accumulating evidence suggests that ZBTB33 plays important roles in embryonic development and tumor progression. Previous studies have shown that Zbtb33-null mice are viable and fertile⁵⁷, implying functional redundancy or compensatory mechanisms in certain developmental contexts⁵⁸. Nevertheless, the precise role of ZBTB33 in ZGA and early embryonic development — and its potential influence on litter size — remains to be elucidated. Furthermore, the mechanisms by which these enhancer-promoters drive target gene expression warrant further investigation.

Mouse endogenous retrovirus elements play an important role in ZGA^6,8,48,59. Among them, MERVL is specifically activated at the 2C stage concomitant with ZGA and its cis-acting functions are required for the correct ZGA process^8,48. In our work, we offered the direct evidence of the enhancer function of MERVL elements in totipotent cells, and further characterized those enhancer-like MERVL elements. We revealed that the long terminal repeats (LTRs) of MERVL were the major elements which exhibited enhancer functions, occurring either as solo copies or as part of full-length MERVL. Although LTRs have conserved sequences, only around half of MERVL-LTRs exhibited the enhancer potential in 2CLCs. In total, 1115 annotated genes were associated with those LTR-enhancers, with 56.4% of them (629/1,115) upregulating in 2C::EGFP positive cells. As for in vivo, those LTRs also exhibited high levels of chromatin accessibility, with 315 associated annotated genes activated during the ZGA process. It should be noted that the other LTRs which hardly exhibited enhancer potential in 2CLCs, including those with STARR activities but low chromatin accessibility (MT2_Mm.C2), or those full-length LTRs exhibiting undetectable STARR activities (MT2_Mm.C3.1), were also bound by ZGA-critical TFs, DUX and OBOX, and exhibited an open chromatin state in 2C embryos. Further investigation is required to determine whether these LTRs perform an enhancer function in vivo. Collectively, our findings provide valuable insights into the cis-regulatory mechanisms governing the ZGA process both in vitro and in vivo.

Methods

Animals

All animal experiments were approved by Institute of Health and Medicine, Hefei Comprehensive National Science Center (approval number: IHM-AP-2024-037). ICR and B6D2F1 (C57BL/6×DBA2) adult (eight to nine-week-old) mice were used in this study. The mice were kept in under controlled conditions, with a constant temperature and a 12-h light/12-hour dark cycle. The environmental temperature ranges from 22 °C to 24 °C, and the humidity is between 50 and 70%.

ESC culture and construction of the DUX-induced 2CLC model

The ES-E14Tg2a cells were cultured on 0.1% gelatin-coated (07903, Stem cell) plates with standard LIF/serum medium containing 15% FBS (10099141, Gibco), 1000 U/mL mouse LIF (ESG1107, Millipore), 0.055 mM β-mercaptoethanol (21985023, Gibco), 0.1 mM non-essential amino acids (11140050, Gibco), 2 mM GlutaMAX (35050061, Gibco), 1 mM sodium pyruvate (11360070, Gibco), and 100 µg/mL penicillin/streptomycin (15140122, Gibco). The ESC lines were cultured in a 37 °C humidified incubator with 5% CO₂. Medium was changed daily, and cells were routinely passaged every other day. The MERVL-LTR-EGFP-F2A-puroR reporter and doxycycline-inducible DUX overexpression cassette were cloned into PB-mCherry plasmid (Blasticidin resistance) and PB-3xFlag (Hygromycin resistance) respectively, and then transfected into E14Tg2a cells by Lipofectamine 3000 (L3000015, Invitrogen). ESCs with the 2C::EGFP reporter and doxycycline-inducible DUX overexpression cassette were selected simultaneously with blasticidin and hygromycin for one week. Single clones were then picked and expanded. Clone 3 which exhibited a high and stable proportion MERVL-EGFP positive cells upon doxycycline addition were chosen for the following experiments. 2 µg/mL doxycycline was used for the DUX induction.

Karyotype analysis

The chromosome preparations were made in a previous protocol⁶⁰. When cells were grown with 60% confluence, 0.2 mg/mL colchicine (A600322, Sangon Biotech) was added into the medium and incubated at 37 °C for 130 min (minutes). Then the cells were digested by 0.25% trypsin, and treated with 0.075 mol/L KCL (A501159, Sangon Biotech) at 37 °C for 28 min. After centrifuge at 250 g for 5 min, cells pellets were collected and incubated in fixative solution (methanol: glacial acetic acid=3:1) at 37 °C for 40 min. The cell suspension was then dropped on the slide and oven baken at 75 °C for 3 h. Subsequently, slides were incubated in the 0.25% trypsin for 10 s (seconds), and the trypsinization was terminated by saline solution. Slides were then stained by Giemsa dye solution for 10 min, and dried at room temperature (RT). Finally, Images were captured by Carl Zeiss imager Z2 (Carl Zeiss) and processed using Ikaros karyotyping system (MetaSystems).

Flow cytometry

Cells were dissociated with 0.05% trypsin in 37 °C for 5 min, then neutralized with the medium. After centrifuge at 250 g for 5 min, the cells were collected at the bottom of the tube and the supernatant was removed. The cells were resuspended in 300–400 µL PBS solution with 2% knockout serum (KSR, A3181501, Invitrogen) and 100 µg/mL penicillin/streptomycin (15140122, Gibco), then filtered with 35 mm strainer cap. Flow cytometry analysis was performed using the BD LSR Fortessa X-20 (BD), and cell sorting was performed on the BD FACSAria Fusion cell sorter (BD). Data and images were analysed and generated using FlowJo (V10) software.

Chimera formation assay

Following DUX induction, 6–8 MERVL GFP negative and MERVL GFP positive cells were microinjected into ICR embryos at the 8-cell stage. Injected embryos were cultured in KSOM medium for 48 h at 37 °C in a humidified 5% CO₂ incubator. The resulting chimeric embryos were then fixed for immunostaining.

Immunofluorescence and imaging of chimeric embryos

Embryos were fixed in 4% paraformaldehyde (PFA, MA0192, Meilunbio) for 30 min at room temperature, followed by three 5 min washes in PBS (SH30028.02, HyClone) containing 0.1% Triton X 100 (T824275, Macklin) (PBST). Samples were permeabilized in PBS with 0.3% Triton X 100 for 30 min and blocked in PBST supplemented with 3% donkey serum (D9663, Sigma) and 1% BSA (Sigma) for 2 h. Embryos were then incubated overnight at 4 °C with primary antibodies: anti CDX2 (1:200, MU392A 5UC, BioGenex), anti SOX2 (1:200, 14981182, ThermoFisher), and anti mCherry (1:100, ARG55723, Airgo).

After four 5 min washes with PBST, embryos were incubated for 2 h at room temperature with secondary antibodies (all at 1:500): Donkey anti mouse IgG (H + L) Alexa Fluor™ 647 (A31571, ThermoFisher), Donkey anti rat IgG (H + L) Alexa Fluor™ 488 (A21208, ThermoFisher), and Donkey anti goat IgG (H + L) Alexa Fluor™ 568 (A11057, ThermoFisher). Embryos were then washed three times with PBST, stained with DAPI (D9542, Sigma) for 5 min at room temperature, and washed again. Images were acquired using a Carl Zeiss LSM800 confocal microscope and processed with ZEN software (Carl Zeiss).

RNA isolation, RT-qPCR and total RNA-seq

Cellular total RNA was collected using a Trizol extraction method. The cells were lysed with 200 µL Trizol Reagent (15596026, Invitrogen) at RT for 5 min. The samples were added 40 µL chloroform and mixed well by vigorously upside down for 15 s, then put at RT for 15 min. After centrifuge at 13000 g, 4 °C for 15 min, the upper aqueous solution was drawn into the new tube and add 100 µL cold isopropyl alcohol (I811925, Macklin), then mixed gently and placed on ice for 5–10 min. After centrifuge at 13000 g, 4 °C for 10 min, the RNA was collected at the bottom of the tube and the supernatant was removed. Add 750 µL 75% ethanol (prepared with RNAase free water), and then reverse several times. After centrifuge at 13000 g, 4 °C for 5 min, the RNA was collected at the bottom of the tube and the supernatant was removed. Dry the RNA at RT until the precipitate changed from white to transparent. Subsequently, the RNA was dissolved in 10–50 µL RNAase free water at 60 °C for 10 min, and then the concentration of the RNA solution was measured. Complementary DNA was generated using PrimeScript II Reverse Transcriptase kit (2690 A, TAKARA) and RT-qPCR was performed using TB Green PremixExTaq Kits (RR420, TAKARA) on QuantStudio^TM 7 Flex (ThermoFisher Scientific). Relative quantification was performed using the comparative CT method with normalization to GAPDH. Primers used for RT-qPCR are listed in Supplementary Data 8.

The total RNA-seq library was prepared with rRNA depletion following stranded method. Briefly, the ribosomal RNA was depleted from total RNA using the Illumina Ribo-Zero plus rRNA Depletion kit (20037135, Illumina) following manufacturer’s instruction. RNA was then fragmented into 300 ~ 350 bp fragments, and first strand cDNA was reverse-transcribed using fragmented RNA and dNTPs (dATP, dTTP, dCTP and dGTP). RNA was degraded using RNase H, and second strand cDNA was synthesised using DNA polymerase I and dNTPs (dATP, dUTP, dCTP and dGTP). Remaining overhangs of double-strand cDNA were converted into blunt ends via exonuclease/polymerase activities. After adenylation of 3′ ends of DNA fragments, sequencing adaptors were ligated to the cDNA and the library fragments were purified. The template without U was enriched by PCR, and the PCR product was purified to obtain the final library. Finally, the total RNA-seq library use PE150 (paired-end 150nt) sequencing (Illumina) at Berry Genomics Co., Ltd.

10× Genomics single-cell transcriptome sequencing (ScRNA-seq)

2C::EGFP reporter mESCs with (+Dox) or without (-Dox) Dux induction were digested and re-suspended in 0.5% BSA-PBS. Then 5000–10,000 cells were collected under the microscope. The cells were counted using Countess (CellDropFL, Denvoix) and the concentration of single cell suspension was adjusted to 1,000 cells/µL. Cells were loaded according to standard protocol of the Chromium Next GEM Chip G Single Cell Kit (1000121, 10x Genomics) in order to capture 5000 cells to 10,000 cells/chip position (V3 chemistry). All the remaining procedures including the library construction were performed according to the standard manufacturer’s protocol. Single cell libraries were sequenced on Illumina HiSeqXTen instruments using 150 nt paired-end sequencing at Berry Genomics Co., Ltd.

Genome-Wide STARR-seq Plasmid pool Construction

The mouse genome was extracted from E14Tg2a mESCs and sonicated. Then sheared genomic DNA fragments of 700 ~ 1200 bp were size selected on 1% agarose gel to construct the STARR-seq plasmid pool following the STARR-seq library cloning protocol⁵¹. Illumina adapters were added to 5 μg of size-selected DNA fragments using NEBNext® Ultra™ II DNA Library Prep Kit for Illumina (NEB, #E7645S). The adapter-ligated fragments were amplified through 5 cycles across 10 reactions with library cloning primers⁵¹, which added a 15nt sequence to the adapters for Gibson assembly. The PCR products were purified with 0.8× AMPure XP beads (A63881, Beckman) and then assembled into the Age1/Sal1-digested hSTARR-seq_ORI vector (99296, Addgene) following the Gibson Assembly Kit instructions (RK21020, ABclonal). The ligation products were purified with MinElute kit (28004, Qiagen) and transformed into MegaX DH10B™ T1R Electrocomp™ Cells (Invitrogen, #C640003). The Plasmid Plus Giga Kit (Qiagen, #12991) was used to extract the plasmid pool for screening.

Genome-wide STARR-seq library preparation and sequencing

2C::EGFP reporter mESCs were pre-treated with 2 µg/mL doxycycline for 6 h, followed by 2 µg/mL doxycycline and 1 µg/mL puromycin for 18 h (as shown in Fig. 2a). Then 400 µg STARR plasmid pool was transfected into 400 million Dox-treated cells with Lipo3000 (L3000015, Invitrogen). After 16 h of transfection, approximately 150 million cells were harvested. The STARR-seq libraries and input were generated according to a previous UMI-STARR-seq protocol⁵¹. STARR-seq libraries were sequenced on Illumina HiSeqXTen instruments using 150 nt paired-end sequencing at Annoroad Co., Ltd.

ULI-NChIP-seq

For ULI-NChIP-seq, 5000 MERVL-GFP positive or MERVL-GFP negative cells were collected per reaction, and two replicates were performed for each stage. The cells were isolated and washed thoroughly in 0.5% BSA-PBS to avoid contamination. The ULI-NChIP procedure was performed according to a previous protocol⁶¹. One microgram of histone H3K27ac (39133, Active Motif), H3K4me1 (5326S, Cell Signaling Technology) or H3K4me3 (9751S, Cell Signaling Technology) antibody was used for each immunoprecipitation reaction. First, antibodies were coupled to protein A‑coated magnetic beads by incubation at 4 °C with rotation for at least 2 h or overnight. Next, cells were resuspended in 20 μl of nucleus extraction buffer (10 mM Tris‑HCl pH 8.5, 140 mM NaCl, 5 mM MgCl₂, 0.6 % NP‑40, 1 mM PMSF, PIC), and MNase master mix (1× MNase master buffer, 2 mM DTT, 5 % PEG6000, 30 U MNase) was added to each sample. After gentle vortexing to mix thoroughly, the samples were incubated at 25 °C for 8 min. The digestion was stopped by adding 5.5 μl of 100 mM EDTA, followed by addition of 5.5 μl of nuclear break buffer (1 % Triton, 1 % DOC solution) and mixing well. At this stage, the samples contained sheared chromatin and were ready for ChIP. For immunoprecipitation, the antibody-coated beads were collected using a magnetic rack, and 100 μl of diluted sheared chromatin was added to each tube, while 10 μl of diluted chromatin was reserved as input at 4 °C. The tubes were gently inverted several times to resuspend the beads, then incubated at 4 °C under constant rotation for 2 h or overnight. Subsequently, the beads were washed twice with 180 μL of ice-cold low-salt buffer (20 mM Tris-HCl pH 8.0, 0.1 % SDS, 1 % Triton X-100, 2 mM EDTA, 150 mM NaCl, PIC) and twice with high-salt wash buffer (20 mM Tris-HCl pH 8.0, 0.1 % SDS, 1 % Triton X-100, 2 mM EDTA, 500 mM NaCl, PIC). To elute DNA, the tubes were incubated at 65 °C for 2 h with rotation, and then 100 μl of DNA eluate along with an equal volume of phenol:chloroform:isoamyl alcohol were transferred to PhaseLock tubes. After vigorous vortexing, the tubes were centrifuged at 13,000 g for 5 min, and the upper aqueous phase was transferred to a fresh 1.5 mL tube. To precipitate the DNA, 10 μl of 3 M sodium acetate and 1 μl of glycogen were added, followed by 100 μl of ice-cold isopropanol and thorough mixing. The tubes were kept at −20 °C for 15 min or overnight to allow precipitation. The precipitated DNA was collected by centrifugation at 13,000 g for 15 min at 4 °C, the supernatant was discarded, and the pellet was washed with 800 μl of 80 % ethanol for 5 min. After a final centrifugation at 13,000 g for 5 min at 4 °C, the supernatant was removed, the pellet was air-dried at room temperature for 5 min, and then dissolved in DNA elution buffer. The sequencing libraries were prepared using the NEBNext^® Ultra™ II DNA Library Prep Kit for Illumina^® (E7645S, NEB) and the NEBNext^® Multiplex Oligos for Illumina^® (Index Primers Set 1) (E7335S, NEB) following the manufacturer’s instructions. Paired-end sequencing with a 150 nt read length was performed on a NovaSeq (Illumina) platform at Berry Genomics Co., Ltd.

CUT&Tag

For p300 (Santa Cruz Biotechnology, sc-48343) CUT&Tag, 200,000 MERVL-GFP positive or MERVL-GFP negative cells were collected per reaction. The CUT&Tag was using Hyperactive® Universal CUT&Tag Assay Kit for Illumina (Vazyme, TD903) following the manufacturer’s instructions.

ATAC-seq

For ATAC-seq, 50,000 MERVL-GFP positive or MERVL-GFP negative cells were collected per reaction, and two replicates were performed for each stage. The cells were isolated and washed thoroughly in 0.5% BSA-PBS to avoid contamination. The ATAC-seq libraries were generated according to a previous protocol with minor modifications². Briefly, samples were lysed in lysis buffer (10 mM Tris-HCl (pH 7.4), 10 mM NaCl, 3 mM MgCl₂ and 0.1% NP-40) for 10 min on ice to prepare the nuclei. Immediately after lysis, nuclei were spun at 500 g for 5 min to remove the supernatant. Then the nuclei were washed once with 200 µL 0.5% BSA-PBS and spun at 500 g for 5 min to remove the supernatant. Nuclei were then incubated with the Tn5 transposome and tagmentation buffer at 37 °C for 30 min (TD502, Vazyme). After the tagmentation, the stop buffer was added directly into the reaction to end the tagmentation at RT for 5 min. The DNA was extracted using MinElute PCR Purification Kit (28004, QIAGEN) following the manufacturer’s instructions. PCR was performed to amplify the library for 13 cycles using the following PCR conditions: 72 °C for 5 min; 98 °C for 45 s; and thermocycling at 98 °C for 15 s, 60 °C for 30 s and 72 °C for 1 min; following by 72 °C for 5 min. After the PCR reaction, libraries were screened and purified with 0.65× and 1.8× AMPure XP beads (A63881, Beckman). Paired-end sequencing with a 150 nt read length was performed on a NovaSeq (Illumina) platform at Annoroad Co., Ltd.

ULI-STARR-seq

We selected the 88 candidates basic based on their STARR-activities from the top. Meanwhile, we tried to cover different clusters in our assay, based on their locations or TE types, as we showed shown in Fig. 2c. The selected candidate enhancer and negative control sequences with Illumina adapters were synthesized and cloned into hSTARR-seq_ORI vector to construct the plasmid library (99296, Addgene). Then the STARR-seq plasmid library was purified and injected into the male pronucleus of zygotes (PN3-PN4, 20-21 h post-hCG injection). After 26 h, the late 2-cell embryos (46–48 h post-hCG injection) were lysed in hypotonic lysis buffer (S1550S, NEB) and the polyadenylated mRNAs were captured using the Magnetic mRNA Isolation Kit (S1550S, NEB). Then the sample was added URBO DNase (0.25 U /5 µL reaction system, AM2238, invitrogen) to digest genomic DNA at 37 °C for 30 min. Subsequently, the polyadenylated mRNAs were purified using 1.8× RNA Clean beads (N412-02, Vazyme). The reverse transcription and subsequent steps followed the previous UMI-STARR-seq protocol⁵¹. STARR-seq libraries were sequenced on Illumina HiSeqXTen instruments using 150 nt paired-end sequencing at Repugene Co., Ltd.

Mouse zygote collection and Injection

Eight- to nine-week-old B6D2F1 females were superovulated by injecting 10 international units (IU) pregnant mare’s serum gonadotropin (PMSG), followed by 14 IU human chorionic gonadotropin (hCG) 48 h (hour) later. The superovulated B6D2F1 female mice mated with B6D2F1 adult males after hCG injection. Zygotes were collected from oviducts and cultured in KSOM medium (Millipore, MR-16-D) at 37 °C under 5% CO₂.

For plasmid microinjection, the successfully constructed plasmids were diluted to 20 ng/μl and microinjected into the male pronuclei of mouse zygotes by using an Eppendorf FemtoJet 4i microinjector. The late-2-cell stage embryos were harvested 46–48 h after hCG injection.

Candidate enhancer and negative control plasmids were introduced by pronuclear injection, following the same procedure as described above. Purified candidate enhancer or negative control plasmids were co-injected with the mCherry reporter plasmid, with final concentrations of 4.5 ng/μl for the candidate enhancer or negative control constructs and 1.5 ng/μl for the mCherry plasmid. After pronuclear injection, zygotes were cultured under the same conditions, and fluorescence intensity was examined 24 h later using a Nikon fluorescence microscope (Ti2-U).

For siRNA-mediated knockdown experiments, small interfering RNAs (siRNAs) were microinjected into the cytoplasm of zygotes using an Eppendorf FemtoJet 4i. The siRNAs (RiboBio or Tsingke Biotechnology) were injected at a volume 5–10 pL and a working concentration of 50 μM. Zygotes injected with negative control siRNA at the same concentration were used as controls. All siRNA sequences used in this study are listed in the Supplementary Data 8. After microinjection, zygotes were cultured in KSOM medium at 37 °C under 5% CO2 to allow recovery and further development. Embryonic development was monitor until the blastocyst stage. Injected embryos were collected at the late 2-cell stage for subsequent knockdown efficiency analyses.

Nuclear transfer

All MII oocytes were collected from superovulated female B6D2F1 mice (8-9 weeks old) 14 h post-hCG injection and removed the cumulus cells in vitro by 1% hyaluronidase. The collected MII oocytes were firstly transferred to droplets of HEPES-CZB medium containing 5 μg/ml cytochalasin B (CB; Sigma, C6762). A piezo-assisted pipette with an inner diameter of 8  μm was then used to pierce zona pellucida and aspirate the chromatin spindle region with a minimum volume of oocyte cytoplasm. The enucleated oocytes were thoroughly washed and preincubated at 37 °C in KSOM medium with amino acids until nuclear transfer.

The mESCs were pretreated by culturing in medium containing 0.05 μg/ml demecolcine (Sigma, D1925) for 12 h to arrest the mESCs at M phase. Then pretreated mESCs were digested with 0.05% trypsin-EDTA, resuspended the cells in HEPES-CZB and stored at 4 °C before injection. The pretreated mESCs were gently injected into the enucleated oocytes. Around 1 h after nuclear transfer, the reconstructed embryos were activated for 6 h by Ca²⁺-free CZB containing 5 nM trichostatin A (TSA; Sigma, T8552) and 10 mM SrCl₂ (Sigma, 255521). Following activation, the reconstructed embryos were transferred into KSOM medium with amino acids containing 5 nM TSA for 4 h. Then all of the reconstructed embryos were subsequently cultured in KSOM medium with amino acids at 37 °C under 5% CO₂. Late 2-cell embryos were collected at 31.5–32 h (including 6 h activation) for the RT-qPCR.

RT-qPCR for 2C embryos

Zona pellucida-free embryos were harvested at the late 2-cell stage and transferred into RNase-free PCR tubes, followed by immediate storage at −80 °C. Total RNA isolation and reverse transcription were performed using the Single Cell Sequence Specific Amplification Kit (Vazyme, P621) according to the manufacturer’s protocol. The resulting cDNA was subsequently diluted fivefold with RNase-free Ultrapure H₂O. Quantitative PCR was performed using ChamQ Universal SYBR qPCR Master Mix (Vazyme, Q711) on a Bio-Rad CFX96 Real-Time PCR System. Each 20 μL reaction contained 10 μL 2×SYBR Master Mix, 0.5 μL each of forward and reverse primers, and 1 μL diluted cDNA template. Amplification was conducted under the following conditions: initial denaturation at 95 °C for 30 s, followed by 40 cycles of 95 °C for 10 s and 60 °C for 30 s. Samples were maintained at 4 °C after amplification. A minimum of three independent biological replicates was included for each experiment. All RT-qPCR primers were synthesized by Tsingke Biotechnology and are listed in Supplementary Data 8.

Enhancer knockout (eKO) in mESCs

The eKO mESCs were performed using CRISPR-Cas9 genome editing with a pair of guide RNAs (gRNAs) for each enhancer region. Online tool (Benchling) was used to design gRNAs. The top listed gRNAs were then cloned into the pSpCas9(BB)−2A-Puro (PX459) V2.0 vector (62988, Addgene) (sgRNAs listed in Supplementary Data 8). Similar to the above description, MERVL-LTR-EGFP reporter and Dox-inducible Dux expression cassette were transfected into E14Tg2a cells to generate the Dox-inducible Dux overexpression 2C::EGFP reporter mESCs without puromycin resistance (inducible Dux mCherry-MERVL-EGFP clone 28, iDux-mcmg-Clone 28). The 2C::EGFP reporter mESCs without puromycin resistance were then transfected with the Cas9/gRNA plasmids. Subsequently, Single cells were sorted after 2-day treatment with 1.5 μg/ml puromycin and amplified. Then the genomic DNA from single cell-derived clones was extracted for genotyping. Primers used for genotyping are listed in Supplementary Data 8. Homogenous eKO clones were used for the following analysis.

Enhancer interference in mESCs

CRISPR-dCas9-KRAB system was used to perform CRISPRi. The pB-Cas9m4-KRAB-MeCP2 plasmid (kind gift from Jiekai Chen’s lab) was first integrated into a pre‑established inducible Dux‑mCherry‑MERVL‑EGFP reporter cell line (iDux‑mcmg‑Clone 28) using the piggyBac transposon system, generating dCas9‑expressing mESCs. Guide RNAs (gRNAs) targeting each enhancer region were designed using the online tool Benchling (Benchling, [Biology Software]. (2026). Retrieved from https://benchling.com). The selected gRNAs were synthesized and cloned into the lenti‑U6‑guide‑puro vector to produce gRNA lentivirus (sgRNAs listed in Supplementary Data 8). The dCas9‑mESCs were then transduced with the gRNA lentivirus and selected with 1.5 μg/mL puromycin for at least 3 days. After selection, 2C‑like cells (2CLCs) were induced with 2 μg/mL doxycycline for 24–36 h, and GFP‑positive cells were sorted by flow cytometry. Total RNA was extracted from the sorted GFP+ cells, and RT‑qPCR was performed using the primers listed in Supplementary Data 8.

Bioinformatics analysis

Single-cell RNA-seq analysis

Single-cell RNA-seq reads were mapped to the mouse reference genome (mm10) using STAR (v2.7.6a). The cells with a minimum cut-off of 1500 genes or over 10% proportion of UMI counts from mitochondrial genes were removed. Cells with total UMI counts above 170000 were also filtered, resulting in a total of 4961 retained cells: 3209 cells with Dox (+Dox) and 1752 cells without Dox (-Dox). Subsequently, single-cell data integration was performed using Seurat’s canonical correlation analysis (CCA) integration tool (Seurat v4.0.4) with the FindIntegrationAnchors and IntegrateData functions. The data from this study and the scRNA-seq data of the in-house mouse preimplantation embryos, published data sets from GSE121459 (mESCs, intermediate cells, and 2C-like cells), GSE168728 (totipotent blastomere-like cells (TBLCs) and pluripotent stem cells (PSCs)), and GSE185005 (chemically induced totipotent stem cells, ciTotiSCs) were scaled using 2000 highly variable genes, and principal component analysis (PCA) was conducted using 30 principal components to compute UMAP coordinates with RunUMAP. Cell clusters were identified through a shared nearest neighbor (SNN) modularity optimization algorithm using the FindClusters function.

Total RNA-seq data analysis

The raw RNA-seq reads were processed using TrimGalore (v.0.6.4) (https://github.com/ FelixKrueger/TrimGalore) to remove adapter sequences and low-quality bases. The cleaned reads were then aligned to the mouse Gencode (vM15) transcriptome with RSEM (version 1.2.22)⁶². Gene expression values were normalized across samples using the variance modeling tool DESeq2 (version 1.30.0)⁶³. Differentially expressed genes were obtained using DESeq2 (version 1.30.0)⁶³. Gene ontology analysis was performed using clusterProfiler (version 3.18.0)⁶⁴.

ChIP-seq and ATAC-seq data analysis

Analysis of ATAC-seq and ChIP-seq data were performed essentially as previously described⁶⁵. The raw reads were processed using TrimGalore (v.0.6.4) to remove adapter sequences and low-quality bases. The cleaned reads were aligned to the mouse mm10 genome using bowtie2 (v2.2.5)⁶⁶, with the options: “-p 10 –very-sensitive –end-to-end –no-unal –phred33 –no-mixed -X 2000”. To analyze repetitive sequences, the multi-mapped reads were kept but only the best alignment was reported; if more than one equivalent best alignment was found, then one random alignment was reported. The PCR duplicates were removed by Picard (https://broadinstitute.github.io/picard/) with the MarkDuplicates function. Peaks were called by MACS3⁶⁷ by default settings. Alignment bam files were transformed into read coverage files (bigwig format) using deepTools (v2.5.4)⁶⁸ with the RPKM (Reads Per Kilobase per Million mapped reads) normalization method. Coordinates and annotations of TEs were downloaded from the UCSC Genome Browser (GRCm38/mm10) version of RepeatMasker (http://hgdownload.soe.ucsc.edu/goldenPath/mm10/database/rmsk.txt.gz). Heatmaps and pileups were generated using deepTools. Motif discoverywere performed by HOMER⁶⁹ with the command:”findMotifsGenome.pl input.bed mm10 output -p 20 -size given”. Peaks annotation was performed by HOMER with the command “annotatePeaks.pl input.bed mm10”. To increase statistical power and reduce noise, we merged the replicate samples by taking the mean expression (or signal) value per group. We then performed the two-sided Mann-Whitney U test on these averaged values. When multiple comparisons were invoved in the same analysis, we applied the Beniamini-Hochberg (BH) procedure to correct for multiple hypothesis testing and controlled the false discovery rate (FDR). In cases where only a single p-value was generated per comparison, no multiple-testing correction was applied, as it is not applicable.

STARR-seq data analysis

The raw reads of STARR-seq and input were processed using TrimGalore (v.0.6.4) to remove adapter sequences and low-quality bases. The cleaned reads were aligned to the mouse mm10 genome using bowtie2 (v2.2.5)⁶⁶, with the options: -p 48 --very-sensitive --end-to-end --no-unal --phred33 -X 2000’. To analyze repetitive sequences, the multi-mapped reads were kept but only the best alignment was reported; if more than one equivalent best alignment was found, then one random alignment was reported. The PCR duplicates were removed by Picard (https://broadinstitute.github.io/picard/) with the MarkDuplicates function. Alignment bam files were transformed into read coverage files (bigwig format) using deepTools (v2.5.4)⁶⁸ with the RPKM normalization method. Genome coverage was computed with deeptools using the pybigwig function. Enriched genomic peaks were called using a 500 bp sliding window algorithm, with a sliding step of 250 bp, using custom scripts. For every 500 bp window in the genome, the number of reads falling in that window for a given sample (normalized by RPKM method) was divided by the number of reads in that window for the corresponding input file (also normalized by RPKM method). Enriched regions were initially formed by taking all 500 bp windows whose signal-over-input value exceeds a 2-fold threshold. Finally, any remaining overlaps among regions were merged, to produce the final peak calls. To identify the differential STARR-seq signal between 2CLCs and ESCs, the peaks from those two samples were merged using the mergeBed command from bedtools⁶⁹. Then deeptools’ pybigwig was used to calculate the value for each region. When integrating analysis STARR-seq with ATAC-seq signals, a small subset of elements showed STARR activity in mESCs or 2CLCs yet were not Tn5-accessible in these cells. As these elements met neither the C1 nor C2 cluster criteria, they were excluded from our analysis (Supplementary Fig. 3a). Genome coverage was computed with deeptools using the pybigwig function. Regions where the value differed by more than 2-fold between 2CLCs and ESCs were considered to be differential peaks. Motif discovery were performed by HOMER⁶⁹ with the command:”findMotifsGenome.pl input.bed mm10 output -p 20 -size given”. Peaks annotation were performed by HOMER with the command “annotatePeaks.pl input.bed mm10”. Heatmap were made by deeptools⁶⁸. The CpG island (CGI) annotation were downloaded from UCSC browser.

Read 1 of the raw paired-end sequencing data was mapped to the mouse reference genome (mm10) using Bowtie2 with parameters -p 18 --very-sensitive --end-to-end --no-unal --phred33. Strand orientation was determined with samtools: alignments with flag -F 16 were considered as forward (Watson) strand, while those with flag -f 16 were considered as reverse (Crick) strand.

ULI-STARR-seq data analysis

First, we constructed a genome containing only our target sequences of interest. We then mapped the cleaned sequencing data back to this reference genome using Bowtie2 with the options: -p 48 --very-sensitive --end-to-end --no-unal --phred33 -X 2000’. Reads that aligned to identical genomic coordinates were identified as PCR duplicates, and only one representative read from each set of duplicates was retained for subsequent analysis. Samtools was used to count the number of sequencing reads mapped to each reference sequence. The ratio of reads mapped to a reference sequence in embryos compared to the input was considered the STARR-seq signal value for that sequence in the embryos. A reference sequence was considered enriched for STARR-seq signal in 2C embryos if the number of reads mapped to that sequence exceeded 1.5 times the average of the negative control.

Statistics & reproducibility

The two-sided Mann-Whitney U test was used to analyze the significant difference between different groups in Fig. 1f and Supplementary Fig. 4b, c. The Padj was from two-sided Mann-Whitney U test with Benjamini-Hochberg correction was used to analyze the significant difference between different groups in Figs. 1c, 2d. For the presented boxplots, error bars represent the Standard Error of the Mean (SEM). The two-sided Student’s t test was used to analyze the significant difference between different groups in Figs. 2i, 4d, f, g and Supplementary Figs. 6g, 10f. Error bars in the graphical data represent the Standard Deviation (SD). STARR-seq, ChIP-seq and RNA-seq experiments were performed two or three times for each group. For other experiments, the precise numbers of replicates and the data qualities were summarized in a Source Data file. No statistical method was used to predetermine sample size. For mouse embryonic experiments, embryos were random used for the experiments.

For the L2C ULI-STARR-seq assay, two data points were excluded from the final analysis due to the variability between replicates. These two fragments were part of the negative control set, and their exclusion did not compromise the integrity or interpretability of the control group, as sufficient negative controls (n = 7) remained for reliable normalization. The excluded data points are clearly indicated in Supplementary Data 7. The investigators were not blinded to allocation during experiments or outcome assessment.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

M.Z. conceived the project and designed the experiments; L.Y. and T.P. performed most of the experiments, with the help of B.H., H.Z., D.T., and L.L.; D.L., T.Z., Y.C., and Y.Z. performed embryo injection and nuclear transfer experiments; J.H. performed most of the bioinformatic data analysis, with the help of X.Z., L.Y., T.P., and M.Z.; M.Z., L.Y., T.P., J.H., and J.C. wrote the manuscript.

Peer review

Peer review information

Nature Communications thanks Chih-Jen Lin, Jun Xu and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by grants to M.Z. from the National Natural Science Foundation of China (32570958), the Pearl River Talent Recruitment Program (2021ZT09Y233), and Major Project of Guangzhou National Laboratory (grant no. GZNL2023A02005); grants to J.H. from the National Natural Science Foundation of China (32570664, 32370856), and the Guangdong Basic and Applied Basic Research Foundation (2023B1515020111); grants to T.P. from National Natural Science Foundation of China (grant no. 32400600) and Young Scientists Program of Guangzhou Laboratory (grant no. QNPG23-04); grants to D.L. from the National Natural Science Foundation of China (32671093) and the Anhui Province Fund for Excellent Young Scholars (grants no. 2508085Y045 and 2024AH030022).

Data availability

High-throughput sequencing data generated in this study are available in the Gene Expression Omnibus with accession number GSE283389. The scRNA-seq data were from GAS: CRA003171²⁸ GSE121459 [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE121451]¹⁷, GSE168728²¹ and GSE185005²⁰. The ATAC-seq and RNA-seq data for pre-implantation embryos were from GSE66390². The H3K4me1, H3K4me3 and H3K27ac ChIP-seq data for pre-implantation embryos were from GSE179941, GSE217970²⁴, GSE71434²³. The OBOX ChIP-seq data were from GSE215813⁵⁰, the DUX ChIP-seq data were from GSE95517⁹, the NFY ChIP-seq data were from GSE56839⁷⁰. Source data are provided with this paper. This paper does not report original code. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-77311-8.