Symmetrically dimethylated histone H3R2 promotes global transcription during minor zygotic genome activation in mouse pronuclei

Paternal genome reprogramming, such as protamine–histone exchange and global DNA demethylation, is crucial for the development of fertilised embryos. Previously, our study showed that one of histone arginine methylation, asymmetrically dimethylated histone H3R17 (H3R17me2a), is necessary for epigenetic reprogramming in the mouse paternal genome. However, roles of histone arginine methylation in reprogramming after fertilisation are still poorly understood. Here, we report that H3R2me2s promotes global transcription at the 1-cell stage, referred to as minor zygotic genome activation (ZGA). The inhibition of H3R2me2s by expressing a histone H3.3 mutant H3.3R2A prevented embryonic development from the 2-cell to 4-cell stages and significantly reduced global RNA synthesis and RNA polymerase II (Pol II) activity. Consistent with this result, the expression levels of MuERV-L as minor ZGA transcripts were decreased by forced expression of H3.3R2A. Furthermore, treatment with an inhibitor and co-injection of siRNA to PRMT5 and PRMT7 also resulted in the attenuation of transcriptional activities with reduction of H3R2me2s in the pronuclei of zygotes. Interestingly, impairment of H3K4 methylation by expression of H3.3K4M resulted in a decrease of H3R2me2s in male pronuclei. Our findings suggest that H3R2me2s together with H3K4 methylation is involved in global transcription during minor ZGA in mice.

Fertilised oocytes undergo various and essential events for preimplantation development, such as Ca 2+ oscillations, meiosis resumption, second polar body extrusion, protamine-histone exchange, and epigenetic reprogramming 1,2 . In particular, the male genome undergoes protamine-histone exchange immediately after fertilisation and is actively demethylated by TET3-dependent and -independent mechanisms at the 1-cell stage, whereas the female genome mainly undergoes DNA replication-dependent passive demethylation up to the morula stage [3][4][5][6][7] . Histone variant H3.3 is predominantly incorporated into male genomes by histone chaperone HIRA at an early pronuclear (PN) stage before DNA replication in zygotes. A deficiency of HIRA prevents the male PN formation and results in incomplete nucleosome assembly 8 , indicating that the recruitment of H3.3 is essential for further events such as reconstitution of paternal genome followed by global transcription. Thereafter, zygotic transcripts are globally increased in a process referred to as zygotic genome activation (ZGA). A large number of protein-coding genes are transcribed at major ZGA during the 2-cell stage, whereas transcription is already initiated at minor ZGA during the late 1-cell stage 7,[9][10][11] . In this reprogramming process, some histone modifications occur during major initial events for embryonic development and genomic stability. A previous report showed that methylation of paternal histone H3K4 was required for minor ZGA 12 . In addition, paternal pericentric regions showed the absence of H3K9me3 and the accumulation of H3K27me3 and H3K9me1, whereas H3K9me2/3 was enriched at maternal pericentric regions 13,14 . HP1-beta is associated with H3K9me1,

Results
The accumulation of H3.3R2A in male pronuclei leads to developmental arrest at the 2-cell stage. To clarify the role of histone arginine methylation catalysed by PRMT5 and PRMT7 in preimplantation embryos, we examined the effect of forced expression of unmethylatable mutants of arginine residues of H2AX, H3.3, and H4: H2AXR3A, H3.3R2A, H3.3R8A, H4R3A on mouse embryonic development (Fig. 1a). Because PRMT5 and PRMT7 are abundantly expressed in unfertilised mouse oocytes 20,22 , we hypothesised that the corresponding histone post-transcriptional modifications could be important for early embryonic development prior to implantation. Based on this hypothesis, we injected mRNA for point-mutated, enhanced green fluorescent protein (EGFP)-tagged histones into mouse oocytes before fertilisation (Fig. 1a). First, we evaluated the optimal concentration of mRNA for microinjection as 50 ng/µL for each mRNA ( Fig. S1 and Table S1). Moreover, we confirmed that the EGFP signals (H2A.X-EGFP, H3.3-EGFP, H4-EGFP, H2A.XR3A-EGFP, H3.3R2A-EGFP, H3.3R8A-EGFP, and H4R3A-EGFP) were observed in MII oocytes and the pronuclei of zygotes at similar levels (Fig. S2). Next, we observed the effects of these mutations on preimplantation development. The forced expression of H3.3R2A-EGFP resulted in developmental arrest at the 2-cell stage unlike in other mutants (Figs. 1b, c, S3 and Table S2). H3.3-EGFP and H3.3R2A-EGFP were preferentially recruited to the male genome at fertilisation (Fig. S4) and were localised in both male and female pronuclei at 10 hpi, PN4 stage (Figs. 1d, e and S5), consistent with the localisation of forced H3.3 expression in previous reports 12,23,24 . We also observed that the incorporation of these H3.3R2A mutants into male pronuclei resulted in diminished H3R2me2s at late PN stages, whereas a marginal difference was detected in female pronuclei (Figs. 1d, e and S5). These results indicate that the reduction of methylation of H3.3R2 in male pronuclei is associated with embryonic development from the 2-cell stage onwards.
H3R2me2s is localised in the euchromatic region during zygote development. Next, we investigated the localisation profile of H3R2me2s in mouse zygotes. Endogenous H3R2me2s was observed on MII chromosomes and was localised in both pronuclei of zygotes (Fig. 2a). Moreover, the signals were clearly detected in decondensed sperm heads in fertilised oocytes at PN0 but not in sperm heads in the perivitelline space (Fig. 2a). These results indicate that H3R2me2s is incorporated into male pronuclei immediately after fertilisation. Previous reports have shown that H3R2me2s is localised in DAPI-weak euchromatin but not in DAPI-dense heterochromatin, including the chromocenter in Drosophila polytene chromosomes 21,[25][26][27] . Consistent with these reports, H3R2me2s signals in mouse zygotes were enriched in DAPI-weak regions but hardly  (blue bar with square box), H3.3-EGFP-(green bar with round box), H3.3R2A-EGFP-(green bar  with triangular box), H3.3R8A-EGFP-(green bar with square box), H4-EGFP-(red bar with round box), and H4R3A-EGFP-(red bar with square box) expressed embryos. A chi-squared test was performed for analysing embryonic development through to the blastocyst stage. The asterisk indicates statistical significance (***p < 0.001). (c) Representative images of H3.3-EGFP-expressed, H3.3R2A-EGFP-expressed, and untreated embryos at 24 hpi and 96 hpi. Scale bar = 100 µm. (d) Immunofluorescent images of H3R2me2s (red) and DAPI (blue) in H3.3-EGFP-expressed, H3.3R2A-EGFP-expressed, and untreated zygotes. EGFP (green) shows successful injection. Key: ♂, male pronuclei; ♀, female pronuclei. Scale bar = 20 µm. (e) Quantification of H3R2me2s signal intensities in the pronuclei of H3.3-EGFP-expressed, H3.3R2A-EGFP-expressed, and untreated zygotes. Each dot plot represents a single zygote. Red bars indicate the mean values. The mean value of male pronuclei in untreated zygotes was set as 1. One-way ANOVA and the Tukey-Kramer test were performed to evaluate statistical significance. *Significantly different from the control (**p < 0.01, ***p < 0.001); n.s.: Not significantly different from the control. The number of zygotes was 15 for the H3.3-EGFP group, 15 for the H3.3R2A-EGFP group, and 13 for the untreated group. The pronucleus used for quantification are presented in Fig. S5. www.nature.com/scientificreports/ detected in centromere/pericentromere regions and the nuclear periphery (Fig. 2b), which are heterochromatin regions [28][29][30] . Conversely, asymmetrically dimethylated H3R2 (H3R2me2a), which has been reported as a transcriptional repression marker, was hardly observed in male pronuclei, while it was slightly observed at low levels in a part of the centromere/pericentromere regions (Fig. S6). To summarize, H3R2me2s is localised in euchromatic regions in mouse zygotes in which it may play a role in transcriptional activation rather than H3R2me2a.
Incorporation of the H3.3R2A mutant into male pronuclei impairs global transcription during minor ZGA. A recent study showed that the inhibition of minor ZGA by treatment with 5,6-dichloro-1-β-D-ribofuranosylbenzimidazole (DRB), an inhibitor of Pol II elongation, resulted in developmental arrest at the 2-cell stage 31 , implying that a failure in minor ZGA is associated with the 2-cell arrest phenotype. We also observed that embryos that expressed H3.3R2A, in which H3R2me2s was diminished, were arrested at the 2-cell stage (Fig. 1b, c). Therefore, we hypothesised that there might be some relationships between H3R2me2s and minor ZGA. To test this possibility, we examined whether the inhibition of H3R2me2s results in impaired global transcription at minor ZGA. Intranuclear BrUTP signals, which indicate active RNA synthesis, were significantly decreased in the male pronuclei of H3.3R2A-EGFP-expressed zygotes compared with H3.3-EGFPexpressed and untreated zygotes (Figs. 3a, b and S7a). Furthermore, the signals for Pol II phosphorylated at Ser2 (Pol II Ser2P), which indicate elongation of Pol II, were significantly decreased in the H3.3R2A-EGFP-expressed zygotes (Figs. 3c, d and S7b), while pan-Pol II signals were equally detected in the H3.3R2A-EGFP-expressed zygotes compared with those in untreated zygotes ( Fig. S7c-e). Interestingly, the localisation of pan-Pol II was significantly accumulated under the forced expression of H3.3-EGFP compared to that in the others (Fig. S7ce). We next validated the expression levels of MuERV-L, which is one of the earliest transcribed retrotransposon using RT-PCR analysis 11,31 . The expression of MuERV-L was clearly downregulated in H3.3R2A-EGFP-expressed early 2-cell (24 hpi) embryos, which is the stage just before major ZGA (Figs. 4 and S8). These results suggest that methylation of H3.3R2 is involved in global transcription via activation of Pol II during minor ZGA.
Formation of H3R2me2s, catalysed by PRMT5 and PRMT7, is associated with minor ZGA. A previous report showed that PRMT5 and PRMT7 are abundantly expressed in MII oocytes and preimplantation embryos by proteomic analyses 20 . First, to confirm whether endogenous H3R2me2s is catalysed by PRMT5 and PRMT7, we used the PRMT5-PRMT7 dual inhibitor DS-437. Because H3R2me2s signals were already observed in MII oocytes (Fig. 2a), we treated MII oocytes with DS-437 to interfere with de novo symmetric dimethylation of H3R2 (Fig. 5a). H3R2me2s signals were significantly decreased in male and female pronuclei of DS-437-treated zygotes compared with DMSO-treated zygotes as a control (Figs. 5b, c and S9a). These results indicate that endogenous H3R2me2s is reduced by DS-437 treatment during zygote development. Next, we examined the effect of DS-437 treatment on global RNA synthesis in zygotes at minor ZGA. The treatment of DS-437 resulted in the impairment of BrUTP incorporation and decreased Pol II Ser2P signals, particularly in male pronuclei (Figs. 5d-g and S9b, c). Finally, we attempted to determine the arginine methyltransferases that are important for the symmetrical dimethylation of H3R2 in zygotes. To this purpose, each Prmt5 and Prmt7 siRNA was injected into MII oocytes (Fig. 6a), and knockdown oocytes were obtained (Fig. 6b). A decrease in H3R2me2s and Pol II Ser2P signals was only observed by co-injection of Prmt5 and Prmt7 siRNAs (Figs. 6c-e,   H3K4 methylation associates with H3R2me2s accumulation in male pronuclei at minor ZGA. A previous report showed that methylation of H3K4 in male pronuclei is required for minor ZGA in early mouse embryo 12 . In addition, H3K4me3 and H3R2me2s co-exist at transcription start sites in mouse B cells 32 . The H3K4me3 signal appears in male pronuclei at late PN stage (PN4) 12,24,33 , whereas our results showed that H3R2me2s was detected during all PN stages and was implicated in minor ZGA. These results prompted us to examine whether H3R2me2s is involved in H3K4me3 accumulation during minor ZGA. To show the relationship between H3K4me3 and H3R2me2s, we first investigated the effect of inhibiting H3R2me2s on H3K4me3 accumulation in zygotes at the PN4 stage by expressing H3.3R2A-EGFP and treating with DS-437. Paternal H3K4me3 signals showed no difference between either of the H3R2me2s inhibited zygotes and the control zygotes (Figs. 7a, b and S11a-d). However, H3K4me3 signals were significantly increased in female pronuclei of H3.3-EGFP and H3.3R2A-EGFP expressed zygotes compared with an untreated control (Figs. 7a, b and S11a). These results suggest that H3K4me3 deposition does not rely on H3R2me2s accumulation in male pronuclei. We next investigated the effect of preventing H3K4 methylation on H3R2me2s by forced expression of H3.3K4M-EGFP. We confirmed the expression of H3.3K4M-EGFP in MII oocytes and its localisation at the pronuclei of zygotes (Fig. S12a, b). Additionally, H3K4me3 signals were clearly diminished in male pronuclei, consistent with a previous report (Fig. S12c) 12 . Interestingly, the deposition of H3.3K4M caused a decrease in H3R2me2s signals in male pronuclei (Figs. 7c, d, and S13). These results suggest that H3R2me2s relies on H3K4 methylation in zygotes.

Discussion
After fertilisation, protamines packaged around the sperm genome are immediately replaced with maternal histones, and the male genome undergoes chromatin remodelling for a transcriptionally active regions 34 . Here, we suggested that the formation of H3R2me2s, catalysed by PRMT5 and PRMT7, in the male genome is important for the first wave of transcription during zygote development. Generally, H3R2me2s is reported as a transcriptionally active marker 21,32 . Consistent with this, H3R2me2s incorporated into the male genome is detected in euchromatic regions during zygote development (Fig. 2). In addition, we have recently shown that H3R17me2a is associated with protamine-histone exchange and active DNA demethylation in zygotes 16 . These data suggest that histone arginine methylations in the paternal genome play an indispensable role for initial reprogramming events in zygotes. Endogenous H3R2me2s was detected on maternal chromosomes and in female pronuclei in oocytes and zygotes, respectively (Fig. 2a). In female pronuclei, the efficient reduction of H3R2me2s signals at late PN stages was not observed, although the expressed signals of the unmethylatable mutant H3.3R2A-EGFP were clearly detected (Figs. 1d, e and S5). It is known that maternal chromatin largely maintains its histone methylations before and after fertilisation 3,35,36 . Indeed, endogenous H3R2me2s was clearly detected on the chromosomes in MII oocytes and was constantly observed throughout the PN stages (Fig. 2a). Thus, for reducing H3R2me2s in the female genome, the histone mutant would need to be expressed at an earlier stage during oocyte maturation. H3R2me2a was detected at low levels in a part of the centromere region compared with H3R2me2s localised in the euchromatic regions (Fig. S6). In general, euchromatin allows gene expression, whereas heterochromatin, which is a highly compacted state, is transcriptionally repressive. Thus, based on immunofluorescent staining at the late PN stage, H3R2me2s would be associated with global transcription during minor ZGA rather than H3R2me2a in zygotes. Consistent with this hypothesis, the inhibition of PRMT5 and PRMT7, which catalyse symmetric www.nature.com/scientificreports/ dimethylation, using a dual inhibitor and their siRNAs resulted in a decrease of transcriptional activity in zygotes (Figs. 5d-g and 6c, e). In the previous report, inhibiting Pol II elongation by DRB treatment at the late 1-cell stage led to downregulation of the M-phase promoting genes, resulting in developmental arrest at the 2-cell stage 31 , suggesting that these gene expressions might be downregulated by PRMT5 and PRMT7 inhibition. However, we could not exclude the possibility of H3R2me2a involvement in early embryonic arrest at the 2-cell stage. Further studies, such as knockdown experiments of PRMT6, which catalyses H3R2me2a, are required to clarify these functions in early embryonic development. Interestingly, the dual inhibitor had more effect on BrUTP level than H3.3R2A forced expression; however, reduction in H3R2me2s levels was similar under both the conditions (Figs. 3a, b, and 5d, e), suggesting the possibility that PRMT5 and PRMT7 have other targets for transcription. Although Pol II Ser2P accumulated at the same level between untreated and H3.3-EGFP-expressed zygotes (Fig. 3c-d), pan-Pol II signals were significantly increased under the forced expression of H3.3-EGFP compared with those in untreated and H3.3R2A-EGFP-expressed zygotes (Fig. S7c-e). According to previous reports, both H3R2me2s and H3K4me3 are recognised by WDR5, which is a component of the MLL complex, and these proteins interact with the Pol II complex 21,37 , suggesting that these modifications on H3.3-EGFP might contribute to the excessive recruitment of Pol II into the pronucleus. Another report indicates that excess H3 by the forced expression is not associated with chromatin 38 . Indeed, the global RNA levels and Pol II Ser2P were not different between untreated and H3.3-EGFP-expressed zygotes (Fig. 3a-d). Thus, there would be no effect of the global transcription activity, even though H3R2me2s accumulation was significantly increased in H3.3-EGFP-expression zygotes (Fig. 1d-e).
We hypothesised that H3R2me2s might regulate H3K4 methylation in mouse zygotes, since H3R2me2s and H3K4me3 are co-localised on the genome of mouse B cells 32 . However, our results showed that the forced expression of H3.3K4M results in a decrease of H3R2me2s level in male pronuclei (Fig. 7c, d), whereas H3.3R2A overexpression did not alter the level of H3K4me3 (Fig. 7a, c). These results suggest that the deposition of H3.3K4 methylation is related to H3R2me2s accumulation. Some studies have showed that H3K4me1 starts to localise at male pronuclei in PN1-2 zygotes 24 and double knockdown of Mll3 and Mll4, which are methyltransferases of H3K4me1, resulting in impairment of minor ZGA 12 . Conversely, accumulation of H3K4me3 in male pronuclei occurs from the PN4 stage onward 12,24 . Thus, there might be a zygote-specific relationship between H3K4me1 and H3R2me2s during minor ZGA. Since target genes for H3R2me2s have still not been reported, additional studies such as RNA-seq analyses would be required to uncover which genes are regulated by H3R2me2s.
The forced expression of H3.3-EGFP and H3.3R2A-EGFP resulted in an increase of maternal H3K4me3 compared with an untreated control at the late PN stage (Fig. 7a, b). In general, maternal H3K4me3 signals are clearly detected throughout zygote development 24,33 . Exogenous H3.3 is incorporated in female pronuclei later than in male pronuclei 24 . Therefore, we might detect the H3K4me3 signals of both chromatin incorporated endogenous H3 and free H3.3-EGFP and H3.3R2A-EGFP in female pronuclei. Because endogenous H3K4me3 accumulation in male pronuclei starts from the PN4 stage 24,33 , paternal H3K4me3 signals in these forced expression zygotes might not be increased at this stage.
The peptidylarginine deiminase (PADI) family positively converts charged arginine residues to citrulline and is associated with gene activation [39][40][41] . Recently, studies on early embryos have shown that the inhibition of PADI results in developmental arrest at the 2-or 4-cell stage 42 , and the loss of PADI1 leads to the impairment of genome activation at the 2-cell stage 43 . Interestingly, nuclear PADI1 is observed from the 2-cell stage but not the 1-cell stage 43 . These results might suggest that a genome activation mechanism at the 2-cell stage is required for the citrullinisation of histone arginine methylation unlike in the zygote. However, in the zygote and 2-cell embryo, the function of another histone arginine methylation is still unclear, and additional study is required to uncover the mechanisms by knockout of Prmt and Padi family.
In conclusion, our study reported that paternal H3R2me2s promotes the phosphorylation of Pol II Ser2 and RNA synthesis on the paternal genome during minor ZGA in mouse zygotes. Our results will contribute to elucidating the relationship between epigenetic modifications and transcriptional dynamics in zygotes. Collection of spermatozoa and oocytes. Collection of spermatozoa and oocytes was performed as described in previous studies [44][45][46][47][48][49][50][51][52] . Spermatozoa were collected from the cauda epididymis of male mice. The sperm suspension was incubated in HTF medium (Ark Resource Co., Ltd., Kumamoto, Japan) for 1-1.5 h to allow for capacitation at 37 °C under 5% CO 2 in air. Oocytes were collected from the excised oviducts of female mice (2-3 months old) that were superovulated with pregnant mare serum gonadotropin (PMSG; Serotropin, Teikoku Zoki, Tokyo, Japan) and 48 h later with human chorionic gonadotropin (hCG; Puberogen, Sankyo, Tokyo, Japan). Cumulus-oocyte complexes were recovered into pre-equilibrated HTF medium.
Image analysis. Images were analysed using ZEN imaging software 2.3 (Carl Zeiss) and ImageJ software (National Institutes of Health). To quantify the pronuclear signal of H3R2me2s, BrUTP, Pol II Ser2P, and H3K4me3, all focal planes that covered the pronuclear region were merged by average intensity projection, followed by selecting the whole PN, the average intensities were measured.
RT-PCR analysis. Total RNA was isolated from 30 pooled oocytes and 20 pooled embryos followed by synthesis of cDNA using the RNeasy Plus Micro Kit (50) (QIAGEN; 74,034). A random 6 mer (Takara Bio Inc.; RR037A) was used as the first strand primer. Armoured RNA control provided from Cells-to-cDNA II Kit (Thermo Fisher Scientific; AM1723) as an external control was spiked into the cell lysate containing total RNA samples before the RT reaction. Half of the cell lysate without the RT enzyme was used as a negative control. cDNA was synthesised using the Superscript III RT First-Strand system (Life Technologies; 18,080,051). Following the RT reaction, the prepared cDNA samples were amplified and analysed using PCR. The primers used are described in Table S4. Amplifications were run in a Thermal Cycler Dice TP600 (Takara Bio Inc.) and Takara Taq Hot start version (Takara Bio Inc.; R007A). The reaction parameters were as follows: one cycle at 98 °C for 2 min and 40 cycles for Armored RNA control or 26 cycles for MuERV-L at 98 °C for 10 s, 58 °C for 30 s, and 72 °C for 25 s. The RT-PCR products were electrophoresed on 2% agarose gels with TBE and visualised using Midori Green Advanced (1:10,000; NIPPON Genetics Co., Ltd; NE-MG04). Three independent experiments were performed for each.