Genome-wide mapping of histone modification H3K4me3 in bovine oocytes and early embryos

Reprogramming of histone modifications is critical to safeguard correct gene expression profile during preimplantation development. Of interest, trimethylation of lysine 4 on histone 3 (H3K4me3) exhibits a unique and dynamic landscape with a potential species-specific feature. Here, we address how it is reprogrammed and its functional significance during oocyte maturation and early embryonic development in cows. Notably, the overall signal of H3K4me3 decreased sharply during embryonic genome activation (EGA). By using low input ChIP-seq technology, we find widespread broad H3K4me3 domains in oocytes and early cleaved embryos. The broad domains are gradually removed after fertilization, which is obviously seen during EGA. Meanwhile, H3K4me3 become enriched at promoter regions. Interestingly, the gene expression level displays a positive correlation with the relative H3K4me3 signal of their promoters when embryos reach 16-cell stage. Importantly, disruption of H3K4me3 demethylases KDM5A-5C increases H3K4me3 level, decreases the embryonic developmental rate and results in dysregulation of over a thousand genes. Meanwhile, KDM5 deficiency causes a re-destribution of H3K4me3 across genome. In particular, the positive correlation between promoter H3K4me3 enrichment and gene expression level disappear. Overall, we describe the genomic reprogramming of H3K4me3 in a greater resolution during bovine preimplantation development and propose that KDM5-mediated re-distribution of H3K4me3 plays an important role in modulating oocyte-to-embryonic transition.


31
The journey of a new life begins with the combination of oocyte and sperm in mammals.

32
Upon fertilization, the highly differentiated gametes are magically reprogrammed to a 33 totipotent embryo, when oocyte-to-embryo transition (OET) is completed (Rivera and Ross,  (Graf et al., 2014;Misirlioglu et al., 2006). 39 Epigenetic modifications are dramatically reprogrammed and play an important role in 40 orchestrate gene expression profiles during OET (Santos et al., 2002). As a hallmark of gene activation, the trimethylation of lysine 4 on histone H3 (H3K4me3) intensity is significantly 42 reduced in accordance with EGA in various species, including mice, humans (Zhang et al.,43 2012), cows (Huang et al., 2015). However, the details of H3K4me3 reprogramming are just 44 emerging thanks to the advent of low input epigenome methods, including ULI-NChIP-seq 45 and CUT&RUN. Impressively, non-canonical broad H3K4me3 domains are enriched in gene 46 bodies and intergenic regions during oogenesis and greatly removed shortly after fertilization 47 in mice. Furthermore, the dysregulation of H3K4me3 removal appears detrimental for EGA 48 and subsequent development (Dahl et al., 2016;Liu et al., 2016;Zhang et al., 2016a). 49 Surprisingly, the genomic distribution of H3K4me3 in oocytes is quite different between 50 human and mouse oocytes, suggesting species-specific role of H3K4me3 (Xia et al., 2019). 51 Nevertheless, the landscape of H3K4me3 throughout bovine oocyte maturation and 52 preimplantation embryogenesis as well as its functional role have yet to be determined. 53 In this study, we mapped genome-wide distribution of H3K4m3 in bovine oocytes and 54 preimplantation embryos by taking advantage of low input ChIP-seq. Broad H3K4me3 55 domains were prevalent in bovine oocytes, and gradually erased after fertilization. Upon H3K4me3 and blastocysts formation failure. In summary, we propose that timely removal of 61 H3K4me3 is required for bovine early embryonic development.

63
Broad H3K4me3 domains are erased upon embryonic genome activation in cattle. 64 To measure the global level of H3K4me3, we first collected bovine oocytes and 65 preimplantation embryos and performed immunofluorescence (IF). Initially, relatively high 66 level of H3K4me3 was observed in both GV and MII oocytes, but decreased significantly in 67 2-cell embryos. Astonishingly, H3K4me3 was barely seen at 16-cell stage (Fig 1a, 1b), when 68 the major embryonic genome activation (EGA) occurs in cattle (Graf et al., 2014;Misirlioglu 69 et al., 2006). The coincidence of H3K4me3 demethylation and EGA is also found in mouse, 70 human and porcine embryos (Huang et al., 2015). Afterwards, H3K4me3 maintained low abundance in morulae and became brighter at blastocyst stage (Fig 1a, 1b). These results 72 suggest H3K4me3 undergoes a substantial reprogramming after fertilization and, in particular, 73 suffers a dramatical loss during EGA, which is conserved among species. 74 In order to explore the genomic distribution of H3K4me3 in detail, we next performed 75 ultra-low-input native chromatin immunoprecipitation (ULI-NChIP) in bovine oocytes and 76 early embryos (Brind'Amour et al., 2015). To validate the protocol used here, we performed 77 ULI-NChIP against H3K4me3 using 30 mouse morulae. Results showed that both biological 78 replicates exhibited strong correlations, and recapitulated results from published data 79 (R>0.80), indicating a robust reproducibility (Liu et al., 2016) (Fig S1a, S1b). Then, bovine 80 oocytes and preimplantation embryos of 2-cell, 8-cell, 16-cell and morula stage were 81 subjected to ULI-ChIP-seq (Fig 1c), and the data were reproducible between replicates (Fig 82 S1c, S1d). Interestingly, we found broad/non-canonical H3K4me3 domains in both GV and 83 MII oocytes, and the broad domains also existed in 2-cell and 8-cell embryos with slightly 84 weaker signals but were almost completely removed in 16-cell embryos (Fig 1d). Besides, the 85 broad H3K4me3 seemed to be more prevalent at gene body and intergenic regions (Fig 1d). 86 Meanwhile, H3K4me3 signal at promoter region appeared to increase from 8-cell stage (Fig   87   1d). This removal of broad H3K4me3 domains is reminiscent of the similar H3K4me3 88 dynamics in mouse embryos (Dahl et al., 2016;Liu et al., 2016;Zhang et al., 2016a), but in 89 sharp contrast to that in human embryos (Xia et al., 2019).

92
To test if broad H3K4me3 is truly prevalent at gene body and intergenic regions, we 93 identified H3K4me3 peaks in oocytes and embryos of different stages, and classified them 94 into gene body, intergenic and promoter peaks. Notably, the intergenic H3K4me3 distribution 95 is very broad, which makes up ~ 13% of the genome in MII oocytes, however, the coverage 96 decreased significantly after fertilization (Fig 2a). Moreover, the percentage of intergenic to 97 total peaks reached its maximum value in MII oocytes and dropped constantly afterwards 98 (Fig 2b). By scanning the genome with a 5 kb sliding window, we identified H3K4me3-lost 99 and -gained regions in MII oocytes and embryos by comparing with the former stage.

104
To comprehensively investigate the genome-wide changes of H3K4me3, we then merged 105 distal intergenic peaks together to calculate H3K4me3 enrichment at all potential intergenic 106 regions. Distinctly, the signal of intergenic H3K4me3 peaks enhanced from GV to MII 107 oocytes and became weaker at 2-cell and 8-cell stage, then continued to decrease in 16-cell 108 embryos and morulae (Fig 2c, 2f). Similar with intergenic H3K4me3, gene body H3K4me3 109 also accumulated from GV to MII oocytes and was removed after fertilization (Fig 2e, 2f).

116
H3K4me3 is well known as a hallmark of active promoters (Barski et al., 2007). Thus, we 117 asked if H3K4me3 deposition is correlated with the abundance of transcripts in bovine early 118 embryos. Based on the signal of promoter H3K4me3, the genes were divided into 2 clusters 119 (Fig 3a). Genes in cluster 1 displayed increased enrichment of promoter H3K4me3 during 120 8/16-cell stage, while genes of cluster 2 lost promoter H3K4me3 at 16-cell stage (Fig 3a, 3b).

121
Simultaneously, we quantified the average expression level of genes in the 2 clusters using 122 published RNA-seq data (Graf et al., 2014). However, we found no correlation between 123 promoter H3K4me3 enrichment and transcripts abundance in the 2 clusters (Fig 3c),  We thus wondered if promoter H3K4me3 enrichment is correlated with gene expression 128 within stages. We divided all genes into 10 groups by their expression abundance at each 129 stage, and calculated the enrichment of promoter H3K4me3 in the corresponding group. In 130 oocytes and early embryos prior to 8-cell stage, promoter H3K4me3 exhibits no or weak correlation with transcript abundance (Fig 3d-3f). Surprisingly, there was obvious positive 132 correlation between the signal of promoter H3K4me3 and transcript abundance in 16-cell 133 embryos (Fig 3g). Furthermore, the signal of H3K4me3 at gene body regions appeared 134 negative correlation with gene expression in GV and MII oocytes (Fig S3), suggesting that 135 H3K4me3 at gene body regions might be involved in transcriptional repression of oocytes. demethylation. Thus, we microinjected a cocktail of siRNA targeting all these three genes 145 into zygotes for interfering the removal of H3K4me3 (Fig S4b). qPCR results indicated a 146 robust knockdown efficiency (Fig 4a). As predicted, immunostaining data showed a 147 significant increase of H3K4me3 in KDM5 co-knockdown (KD) embryos at both 8-cell and 148 blastocyst stage (Fig 4b, 4c). Importantly, the proportion of 8/16-cell on day 3 post 149 insemination was reduced in KD group (Fig 4d) and the blastocyst rate for KD group 150 decreased significantly compared with NC group (Fig 4e). In all, the timely removal of 151 H3K4me3 mediated by KDM5 is required for bovine early embryonic development.

152
Blocked the removal of H3K4me3 dysregulated gene expression during EGA. 153 In order to explore the effect of H3K4me3 removal on gene expression, we collected embryos 154 from NC and KD groups at 8/16-cell stage to perform RNA-seq ( Fig S4b). 1488 genes were 155 differentially expressed in KD embryos, with 1310 up-regulated and 178 down-regulated (Fig   156   5a). We categorized the up-regulated and down-regulated genes into clusters based on their 157 expression pattern in wild-type embryos. Among 178 down-regulated genes, 85 in cluster 1 158 are supposed to be highly expressed especially at 16-cell stage, and 50 in cluster 2 should be 159 moderately expressed at 16-cell stage then highly expressed in blastocysts (Fig 5b). Only 24% 160 (43/178) down-regulated genes (cluster 3) act as maternal genes, which are repressed during EGA (Fig 5b). For the 1310 up-regulated genes, 322 in cluster 5 are developmental genes 162 which are prematurely expressed at 8/16-cell stage, and up to 826 in cluster 6 are maternal 163 genes (Fig 5c). The remaining up-regulated genes (12%, 122/1310) can be classified as EGA 164 genes (Fig 5c). The down-regulated genes were significantly enriched to several GO terms 165 related to positive regulation of transcription activity as well as cell differentiation (Fig 5d).  Precise distribution of H3K4me3 is essential for transcription activity during EGA. 171 We then tested that if the dysregulation of gene expression is attributed to the genomic H3K4me3-lost regions (Fig 6a), which is in consistent with the IF results (Fig 4b). Further 178 analysis revealed that more than 90% of the H3K4me3-gained regions were gene-body or 179 intergenic regions (Fig 6b, 6c), indicating the removal of broad H3K4me3 in these regions 180 was largely impaired. Indeed, gene body or intergenic H3K4me3 peaks covered more 181 genome in the KD embryos ( Fig S6f).

182
Genome annotation of H3K4me3-gained regions showed ~40% of them were located in 183 promoter regions (Fig 6b, 6d), reflecting the inadequate establishment of promoter H3K4me3. we found prevalent distribution of broad H3K4me3 at gene body and intergenic regions in 210 oocytes, which was removed sharply during EGA (Fig 7). Importantly, in contrast with the 211 overall loss of H3K4me3 detected by IF, ChIP-seq analysis revealed enhanced H3K4me3 at 212 promoter regions during EGA (Fig 7). This reminds that when studying dynamics of 213 epigenetic modification, the combination of IF and ChIP-seq is very important to obtain 214 accurate and comprehensive information.

215
H3K4me3 has been considered as a hallmark of actively transcribed genes. However, its  genes. This result is consistent with our recent finding that H3K4me3 level is not a reliable 227 marker to distinguish active and repressive genes in mouse early embryos (Dang et al., 2021).

228
The crosstalk between H3K4me3 and other histone modifications such as histone acetylation 229 H3K27ac may be a potential mechanism that recruits other chromatin factors to contribute to 230 the eventual transcription activation.

231
In summary, we characterize the genome-wide distribution of H3K4me3 in a greater 232 resolution using the cutting-edge ChIP-seq approach and reveal the dynamic changes of  Immunofluorescence was also performed to validate knockdown effects at the protein level.   324 The genome was binned into 5000bp windows using makewindows utility of Bedtools  The annotated distal intergenic regions were intergenic peaks, and peaks within 3kb of TSS 338 were promoter peaks.

339
the H3K4me3 distribution at promoter regions in corresponding group was normalized to 342 RPKM. 344 To compare H3K4me3 distribution between groups, bovine genome was scanned using a 345 sliding window of 5 kb and step size of 1 kb. H3K4me3 signal for each window was 346 calculated and normalized to RPKM. Next, RPKM of H3K4me3 was compared in parallel 347 between groups. The H3K4me3-gained or -lost regions were identified with following  Barski, A., Cuddapah, S., Cui, K., Roh, T.Y., Schones, D.E., Wang, Z., Wei, G., Chepelev, I., and Zhao, K. (2007).