SMCHD1 terminates the first embryonic genome activation event in mouse two-cell embryos and contributes to a transcriptionally repressive state

Abstract

Embryonic genome activation (EGA) in mammals begins with transient expression of a large group of genes (EGA1). Importantly, entry into and exit from the 2C/EGA state is essential for viability. Dux family member genes play an integral role in EGA1 by activating other EGA marker genes such as Zscan4 family members. We previously reported that structural maintenance of chromosomes flexible hinge domain-containing protein 1 (Smchd1) is expressed at the mRNA and protein levels in mouse oocytes and early embryos and that elimination of Smchd1 expression inhibits inner cell mass formation, blastocyst formation and hatching, and term development. We extend these observations here by showing that siRNA knockdown of Smchd1 in zygotes results in overexpression of Dux and Zscan4 in two-cell embryos, with continued overexpression of Dux at least through the eight-cell stage as well as prolonged expression of Zscan4. These results are consistent with a role for SMCHD1 in promoting exit from the EGA1 state and establishing SMCHD1 as a maternal effect gene and the first chromatin regulatory factor identified with this role. Additionally, bioinformatics analysis reveals that SMCHD1 also contributes to the creation of a transcriptionally repressive state to allow correct gene regulation.

INTRODUCTION

After oocyte fertilization, embryos undergo an extensive reprogramming of their genomes in preparation for initiating gene transcription and switch from oocyte to embryonic control of development. Indeed, neither the ability to perform gene transcription nor the ability to regulate it exist at the time of fertilization. In mice, the first round of DNA replication is needed to initiate transcription, and the second round establishes a repressive chromatin structure in which enhancers are needed for high rates of gene transcription (34). Moreover, the first half of the mouse one-cell stage is transcriptionally repressive, with a transition to transcriptional permissiveness arising in the late one-cell stage (37).

Embryonic genome activation occurs in multiple waves of transcriptional activation, followed by waves of gene repression as embryos move through subsequent cleavage stages (26). This is accompanied by distinct profiles of proteins produced at different times during cleavage (35). In mice, a small amount of gene transcription is detectable at the late one-cell stage. An initial, transient burst of transcription of >800 genes comprises the first wave of genome activation (EGA1). After EGA1 is terminated, thousands of other genes are activated at the late two-cell stage (EGA2), followed by additional rounds of activation at the eight-cell, morula, and blastocyst stages. In nonrodent species, the timing is different but basic principles apply, with multiple waves of transcriptional activation starting at about the six- to eight-cell stages (40).

Previously, it was thought that EGA1 may reflect “accidental” transcriptional activation of genes, such as endogenous retroviruses, occurring due to the lack of a chromatin structure that enables robust gene regulation. Recently, however, essential functions of the EGA1 have been reported in mice and humans, and a similar “2C” state has been reported in a fraction of proliferating human and mouse embryonic stem cells (ESCs) in culture (2, 12, 13, 16, 19, 27, 33). During that 2C/EGA1 state, Dux gene family members are activated by the Dppa2 and Dppa4 genes and in turn activate other genes such as Zscan4 (13, 15, 27). In embryos, Zscan4 expression is essential for viability, but correct downregulation of ZScan4 and correct amount of expression are also required (19, 33). In some cell types, ectopic DUX gene expression is deleterious (17). During the 2C/EGA1 state in ESCs, genome-wide DNA hypomethylation occurs, and cells enter a totipotent state, so that the 2C/EGA1 state must be terminated to restore a normal ESC pluripotent state (16, 33). These observations have provided new insight into a vital role played by the 2C/EGA1 state in embryos and ESCs and also highlight the need for correct regulation, and especially timely termination, of the 2C/EGA1 state.

The structural maintenance of chromosomes flexible hinge domain-containing protein 1 gene (Smchd1) is expressed in the oocyte at both mRNA and protein levels (9, 50). Smchd1 expression has been detected by RNA sequencing data at both the germinal vesicle and MII oocyte stages (Table 1). The Dux mRNA was undetected at both stages, whereas multiple mRNAs from the Zscan4 family were detectable at very low levels (Table 1). These results confirm Smchd1 as a maternally expressed gene and Dux and Zscan4 as not highly expressed in oocytes. This gene regulates hundreds of genes across the genome (3, 5, 6, 8, 23, 28, 30, 39, 47, 52). SMCHD1 deficiency contributes to facioscapulohumeral muscular dystrophy, arhinia, and cancer in humans (24, 39, 54, 57). Deficiency in mice disrupts X chromosome inactivation, imprinted gene regulation, and regulation of other gene clusters and leads to selective lethality in female embryos on embryonic day 10.5 (3, 6, 8, 22, 23, 28, 30, 47, 48, 52). Ablation of both maternal and embryonic Smchd1 mRNA using siRNA knockdown at the zygote stage compromises blastocyst formation and term development, in conjunction with disruption in specialization of the inner cell mass (51). Because Smchd1 is expressed initially as an oocyte factor and because the 2C/EGA1 state is so important in maintaining ESC potency, we tested whether Smchd1 plays a role in the EGA1 state in mouse two-cell stage embryos. We show here that Smchd1 deficiency disrupts the normal downregulation of the EGA1 marker genes Dux and Zscan4. Additional bioinformatics analyses further indicate a role for SMCHD1 in establishing a transcriptionally repressive chromatin state to allow correct gene regulation. These results reveal additional key developmental roles for this powerful chromatin regulator in the early embryo.

Table 1.

RNAseq gene expression data (FPKM) for study genes

Gene	Multiple Study Average*		Ramskold et al. (53) and Deng et al. (14)
	GV	MII	Zygote	Early 2-cell	Mid 2-cell	Late 2-cell	4-Cell
Dux	ND†	ND	6.703	23.215	2.554	0.282	0.000
Smchd1	2.771	6.357	5.230	5.370	12.096	5.869	12.148
Zscan4c	0.109	0.179	0.000	0.334	23.361	14.969	0.499
Zscan4d	1.155	0.736	0.004	2.331	60.629	32.983	1.046
Zscan4f	ND	ND	0.000	0.832	34.182	12.096	0.660

Dux, double homeobox; FPKM, fragments per kilobase of exon per million mapped reads; GV, germinal vesicle; MII, metaphase II; Smchd1, structural maintenance of chromosomes flexible hinge domain-containing protein 1; Zscan4c, -4d, and -4f, zinc finger and SCAN domain containing 4c, 4d, and 4f, respectively.

^*Average GV and MII FPKM expression values from published studies (18, 19, 38, 52, 59).

^†Not detected.

MATERIALS AND METHODS

Mice and embryo culture.

All animal use was approved by the Institutional Animal Care and Use Committee of Michigan State University and performed in accordance with committee guidelines. Mice were sustained on 12 h light-dark cycles and housed in Optimice caging systems. Oocytes and embryos from 8- to 12-wk-old female mice were obtained [C57BL/6 × DBA/2 (B6D2F1; The Jackson Laboratory, Bar Harbor, ME)] and were either unmated or mated to B6D2F1 males. Superovulation was stimulated in these animals using intraperitoneal injections of pregnant mare serum gonadotropin (PMSG; Sigma or ProspecBio HOR-272, 5 IU) and human chorionic gonadotropin (HCG; Millipore Sigma 9002-61-3, 5 IU) 48 h later. Oocytes were isolated at the second meiotic metaphase (MII), and embryos were isolated at the zygote stage in M2 medium (Millipore, Billerica, MA). Cumulus cells were removed using hyaluronidase (Sigma-Aldrich, St. Louis, MO) as described (11). Embryos were then cultured in KSOM medium (18) in humidified atmospheres containing 5% O₂ and 5% CO₂ at 37°C and collected at various time points during development.

Immunofluorescence confocal microscopy, imaging and image quantitation.

Once embryos reached the desired point of development (unsynchronized early, mid, or late two-cell stages or 0, 3, 6, 9, 12, 15, 18, or 21 h post-cleavage), zona pellucida were removed using acidified Tyrode’s buffer, followed by a brief wash in M2 medium as described (10). Embryos were washed in two drops of ∼50 μL of 0.4% polyvinyl alcohol (PVA) in phosphate-buffered saline (PBS). Fixation was achieved by moving embryos into 3.7% paraformaldehyde and incubating at room temperature for 30 min. Embryos were washed in three drops of ∼50 μl of of M2 and permeated in 0.1% Triton X-100 in PBS for 30 min. Embryos were blocked in 2% BSA for 45 min and placed in drops of primary antibodies (rabbit α‐SMCHD1 polyclonal antibody, ab122555; Abcam, Cambridge, MA; or rabbit‐α‐ZSCAN4 polyclonal antibody, LS-Bio, LS-B5074) overnight at 4°C. The next morning, embryos were washed 3× 5 min in 0.01% Triton X-100 in PBS, reacted with secondary antibodies (donkey-anti-rabbit 594, ab150076, or donkey-anti-rabbit 488, ab150073; Abcam), washed three times for 5 min in 0.01% Triton X-100 in PBS, and mounted using VectaShield mounting medium with DAPI (Vector, Burlingame, CA). The specificity of the SMCHD1 antibody was demonstrated by siRNA knockdown and Western blotting in our earlier study (50). The specificity of the ZSCAN4 antibody was demonstrated by Western blotting and immunofluorescence microscopy in a published study (38). An additional no-primary antibody control for two-cell embryos yielded no detectable staining (not shown).

Immunofluorescence confocal microscopy (IFCM) was performed on an Olympus FluoView FV1000 filter-based CLSM confocal microscope with a ×40 oil immersion objective with additional ×2 zoom using Olympus FluoView FV1000 Advanced Software (FV10-ASW) version 4.2. Optical sections were taken at ∼3-µm intervals and were compiled into Z-stacks. Image analysis was conducted using ImageJ Software (55). For each embryo, we quantified the nucleus:cytoplasmic ratio using Z-stacks images, which was composed of the all z-slices complied together. To quantify nuclear protein expression, the two-cell embryo images were first split by wavelength and the color corresponding to the secondary antibody fluorescence wavelength analyzed. Using the free hand selection tool in ImageJ, the shape of the nucleus was traced and measured. Cytoplasm and image background intensities were measured as the average of three regions equal in size to the nuclear region, yielding a total of three measurements (nucleus, cytoplasm, and background) for each individual blastomere within the embryo. After background subtraction, the nucleus/cytoplasm ratio was calculated for each stage of development (0, 3, 6, 9, 12, 15, and 18 h post-cleavage). Normalized nucleus/cytoplasmic ratios were calculated to correct for any out-of-plane fluorescence or differences in total expression between embryos. Also, due to a change in equipment involving a change in lasers, background-corrected intensity values were normalized before calculation of the nucleus/cytoplasm ratio. Measurements were derived from six to nine embryos collected in three replicate experiments. Statistical testing of mean differences was conducted using a t-test with a Holm-Bonferroni correction for multiple testing.

siRNA knockdown.

Zygotes were isolated at 20 h post-hCG and incubated for 2 h in KSOM under 5% O₂, 5% CO₂, and 37°C in a humidified chamber. After a brief 5 s treatment with acidified Tyrode’s buffer to thin the zona pellucida, zygotes were washed in M2 for 5 min, transferred to Opti-MEM (Life Technologies, Waltham, MA) medium, and then placed in 4 μL droplets of 200 $\mu$M SMCHD1 siRNA (ON-TARGET plus SMART POOL, L040501-01-0005; Dharmacon, Lafayette, CO) or 200 $\mu$M of control siRNA (ON-TARGET plus Non-targeting Pool, D-001810-10-05; Dharmacon) situated between the electrodes of a BTX electroporation microscope slide (BTX Harvard Apparatus 45-0104). The siRNAs were then electroporated using a BTX 2001 electroporator set to deliver two sequential 30V pulses of 1-ms pulse and 1,000 ms. Zygotes were oriented in a single line for electroporation. After electroporation, zygotes were washed in fresh Opti-MEM and then washed three times in KSOM before culturing as above for two-cell data. For the eight-cell knockdown data, we used cytoplasmic microinjection of zygotes as described (50).

Parthenogenetic activation and α-amanitin treatment.

Diploid parthenogenetic embryos were obtained by activating MII stage oocytes in 10 mM strontium chloride in calcium-free CZB medium (7) supplemented with 5.5 mM glucose (CZBG) and with 5 μg/ml cytochalasin B (Sigma-Aldrich, St. Louis, MO) to prevent polar body extrusion. Activated parthenotes were divided into two pools per mouse and cultured in standard KSOM or KSOM + 24 μg/mL α-amanitin (A2263-1MG; Millipore Sigma, Burlington, MA). Parthenotes were lysed at the late one-cell stage (15 h post-hCG), early two-cell stage (0 h post-cleavage), or late two-cell stage (18 h post-hCG) and processed for quantitative RT-PCR analysis as described below.

Quantitative reverse transcription-polymerase chain reaction.

Pools of oocytes or embryos were processed by removing the zona pellucidae as above and washing in M2 medium and then immediately lysed in 40 μL of extraction buffer (PicoPure RNA Isolation, ThermoFisher KIT0204), and the lysate was incubated at 42°C for 30 min as per the manufacturer’s protocol. For developmental expression studies, ∼14 embryos were pooled for each time point and the genes analyzed in three bioreplicates. For siRNA knockdown experiments, paired pools of 16–19 embryos corresponding to control and knockdown treatments (equal number of embryos within sample pair) were used, and three bioreplicates were performed. For parthenogenetic studies, pools of 10 control and treated embryos were used (3 bioreplicates). RNA isolation and purification were performed using the PicoPure isolation kit as stated above. cDNA was synthesized using the qScript cDNA SuperMix kit (Quanta 95048-100). qPCR assays employed the Taqman system buffers and gene expression assay probes from Thermo-Fisher (Table 2). These included probes for Smchd1 (NM_028887.3), Zscan4 family (Zscan4f: NM_001110316.2; Zscan4c: NM_001013765.2; and Zscan4d: NM_001100186.1), and Duxf3 [NM_001081954.1; corresponding to the Dux gene from previous studies (13, 27), henceforth referred to here as Dux] and probes for endogenous standards Rpl18 (NM_009077.2) and Pgk1 (NM_008828.3) (Table 2). Each plate was loaded on ice and contained wells of 10 µL of cDNA (5 µL, 1:5 dilution with DI water) and buffer plus probe mixes (5 µL, 1:10, probe/buffer) and four to five technical replicates run per sample plus probe combination. Upon loading, each well was mixed six times and the plate was spun for 30 s before placement in a QuantStudio 7 (Thermofisher).

Table 2.

Taqman primers and assay IDs used for quantitative RT-PCR

Gene	TaqMan Assay ID	Accession ID	Job Specification
Smchd1	Mm00512253_m1	NM_028887.3	Target gene
Dux	Mm03038727_g1	NM_001081954.1	Target gene
Zscan4	Mm01234988_g1	NM_001110316.2, NM_001013765.2, and NM_001100186.1	Target gene
Rpl18	Mm02745785_g1	NM_009077.2	Internal standard for knockdown and developmental series
Pgk1	Mm00435617_m1	NM_008828.3	Internal standard for developmental series
Kif18a	Mm01327661_m1	NM_139303.1	Internal standard for α-amanitin experiments

Dux, double homeobox; Kif18a, kinesin family member 18A; Pgk1, phosphoglycerate kinase 1; Rpl18, ribosomal protein L18; Smchd1, structural maintenance of chromosomes flexible hinge domain-containing protein 1; Zscan4, zinc finger and SCAN domain containing 4.

qPCR statistical testing.

All qPCR testing was conducted using the the ddCT R package with standard settings for relative quantification, differing only in the internal control. For studies of α-amanitin-treated parthenotes, expression levels were normalized to Kif18a (NM_139303.1). Studies of Smchd1 knockdown effects used Rpl18 as a standard. For developmental time series studies, the averaged Ct values of two housekeeping genes (Rpl18 and Pgk1) were used for normalization. Statistical testing was conducted using a t-test, with a Holm-Bonferroni correction for multiple testing.

Determining EGA1 genes as potential targets of DUX or SMCHD1.

Because previous studies did not specifically address genes with transiently elevated expression (i.e., EGA1), we identified a set of 552 EGA1 genes using data from a previously published study (14, 53); RNA-sequencing samples were selected from the cell stages of MII, zygote, early two-cell, mid-two-cell, four-cell, eight-cell, 16-cell, and mid-blastocyst. It is noted that the data was derived from Mus castaneus and mixed Mus castaneus × Mus musculus mice, as opposed to Mus musculus mice used here. The raw fastq sequencing data were retrieved and reprocessed. Low-quality reads were trimmed using TrimGalore (https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/), and read alignment was conducted using Hisat2 (32) to the Ensembl Mus musculus genome (GRCm38.p6, genome annotation relase-91) (1, 65). Read counts were generated with featureCounts (42). Differential expression was calculated at the early and mid two-cell stages with the DESeq2 software package (44). They were independently compared with a combination of pre two-cell stages (MII oocyte and 1-cell stages). This was followed by a comparison to the four-cell stage. EGA1 genes were classified as those showing a significant increase from pre-two-cell to mid or early two-cell stage, followed by a significant decrease at the four-cell stage. Addressing later expression changes, differential calculations were conducted comparing four-cell, eight-cell, 16-cell, and mid-blastocyst. A false-discovery rate was set at 0.05. To identify prospective targets of DUX and SMCHD1, data for ChiP and effects of Smchd1 or Dux modulation (knockdown/knockout) studies were used (8, 13, 17, 27, 31, 50, 52, 60). ChiP peak calls were retrieved from the supplemental data from those studies and gene annotation was applied using ChiPseeker (64). For a de novo binding motif analysis, sequences from ChiP peak calls were extracted and submitted to the MEME software (4). Non-mononucleotide resultant motifs were submitted to FIMO to identify potential occurrences (25). FIMO results were filtered with a false discovery rate threshold of P < 0.0001. A gene was considered a potential target if it was present in any one of the above studies and/or de novo motif analysis.

Dux, Smchd1, and Zscan4 expression at germinal vesicle and MII stages from published RNAseq data.

The expression of Dux, Smchd1, and Zscan4 mRNAs was calculated at GV and MII stages by taking the average fragments per kilobase of exon per million mapped reads (FPKM) values from five publicly available (20, 21, 41, 56, 63) (Table 1). The raw sequencing data were obtained and processed in the same manner as utilized in the identification of EGA1 genes above. Each study was processed in parallel, and ultimately calculating the mean FPKM for each gene at GV and MII stages.

RESULTS

Gene expression patterns in oocytes and early embryos.

Because EGA1 is a rapid and transient event during the two-cell stage, it is vital to have as clear of an understanding as possible of the relative timing of expression of genes at the mRNA and protein levels to dissect potential regulatory mechanisms. To achieve this, we first established a detailed mRNA temporal expression profile for MII stage oocytes, late one-cell stage embryos, and synchronous cohorts of two-cell stage embryos at sequential 3 to 6 h time points postcleavage (obtained by the “pick-off” method of collecting embryos cleaving within 1 h intervals). Analysis by RT-qPCR revealed that the Dux mRNA was already high in abundance in late one-cell embryos and at the start of the two-cell stage and rapidly declined thereafter, reaching a baseline level by 9 h after pick-off (Fig. 1A). To assess transcription dependence of Dux mRNA expression at the late one-cell stage, parthenogenetic embryos were employed as a readily available source of synchronized embryos (because pick-off is not an option for 1-cell embryos). Comparing MII stage oocytes, parthenogenetic control and α-amanitin-treated embryos revealed that the increased expression of Dux mRNA at the late one-cell and early two-cell stages compared with the MII stage was sensitive to α-amanitin, indicating transcriptional dependence of this upregulation (Fig. 1B). The low level of Zscan4 mRNA detected at the late one-cell stage was not sensitive to α-amanitin (Fig. 1B). The Zscan4 mRNA increased rapidly between 0 and 3 h postcleavage (Fig. 1A). At the early two-cell stage, Zscan4 mRNA expression was sensitive to α-amanitin in parthenogenetic embryos confirming transcription dependence (Fig. 1B). This timing of Zscan4 mRNA expression was consistent with it being transcriptionally activated by the earlier increase in Dux expression. The Smchd1 mRNA was detected in MII oocytes as expected and in late one-cell embryos and early two-cell embryos, declining sharply in abundance by 9 h postcleavage, with further decline thereafter. Despite an apparently higher Smchd1 mRNA abundance in one- and early two-cell embryos compared with MII stage oocytes by RT-qPCR (Fig. 1A), Smchd1 mRNA expression was not sensitive to α-amanitin treatment at these stages (Fig. 1B), indicating that expression at these stages is attributable to maternally derived oocyte expression. These data demonstrate that Smchd1 is expressed as a maternal mRNA in oocytes and early embryos and that Dux and Zscan4 are transcriptionally activated at the late one-cell and early two-cell stages, respectively.

Fig. 1.

Log fold change expression levels of double homeobox (Dux), structural maintenance of chromosomes flexible hinge domain-containing protein 1 (Smchd1), and zinc finger and SCAN domain containing 4 (Zscan4) mRNAs measured by quantitative (q)RT-PCR. A: analysis of expression in oocytes and fertilized embryos. Two separate qPCR series were conducted. The first series included samples for metaphase II (MII), one-cell, and two-cell embryos at 0 h after pick-off. The second series included samples of two-cell stage embryos collected at the indicated times after pick-off (hours 0, 3, 9, 12, 15, and 18). Results for the two-cell 0h time were used as the reference (denoted as “b” and set to a value of 1) for normalization to allow the two series to be plotted together. The averaged Ct values of both Rpl18 and Pgk1 were used as “housekeeping genes” for normalization. Statistical testing was conducted using a t-test with a Holm–Bonferroni correction for multiple testing. The “a” denotes time points of max SMCHD1 intensity staining from Figs. 2 and 3. Statistical significance: *P < 0.05; **P < 0.01; ***P < 0.001. B: analysis of transcription dependence of expression in parthenogenetic embryos. qPCR was used to determine the effect of α-amanitin treatment on the expression of mRNAs in parthenogenetic embryos at late one-cell and early two-cell stages. Expression was normalized to the endogenous control of Kif18a for each gene. Statistical testing was conducted using a t-test with a Holm–Bonferroni correction for multiple testing. Statistical significance: *P < 0.05, **P < 0.01. For both experiments, pools of ∼14 oocytes or embryos were used for each time point and included 3 bioreplicates.

Next, because these proteins exert functions as nuclear proteins, we assessed the relative temporal profiles of nuclear localization of SMCHD1 and ZSCAN4 by IFCM. Because a suitable antibody for DUX was not available, its expression profile was not examined. SMCHD1 was shown previously to display periodic peaks in nuclear staining intensity during preimplantation development, including at the mid two-cell stage (50). To define the nuclear staining profile more precisely for the two-cell stage, we analyzed synchronous cohorts of embryos (Fig. 2). SMCHD1 nuclear staining increased during the first half of the two-cell stage and reached a peak at 9 h after pick-off. A peak in ZSCAN4 nuclear staining arrived earlier, at ∼3 h after pick-off, the same time as its mRNA peak expression. ZSCAN4 nuclear staining then declined sharply between 6 and 9 h after pick-off, in synchrony with SMCHD1 reaching its peak in nuclear expression at 9 h after pick-off.

Fig. 2.

Expression of structural maintenance of chromosomes flexible hinge domain-containing protein 1 (SMCHD1) and zinc finger and SCAN domain containing 4 (ZSCAN4) proteins in mouse embryos by immunofluorescence. A: normalized nuclear fluorescence intensities by immunofluorescence confocal microscopy (IFCM) imaging of SMCHD1 and ZSCAN4 at indicated times during the two-cell stage. Cytoplasm and image background intensities were measured as the average of three regions equal in size to the nuclear region, yielding a total of three measurements (nucleus, cytoplasm, and background) for each individual blastomere within the embryo, using ImageJ software. After background subtraction, the nucleus-to-cytoplasm ratio was calculated for each stage of development (0, 3, 6, 9, 12, 15, and 18 h postcleavage). Measurements were derived for 6–9 embryos collected in 3 replicate experiments using an Olympus FluoView FV1000 filter-based CLSM confocal microscope for both targets. Statistical testing was conducted using a t-test with Holm–Bonferroni correction for multiple testing. Statistical significance: **P < 0.01; ***P < 0.001; ****P < 0.0001. B: representative IFCM images for late one-cell, early two-cell, and late two-cell stages.

Effects of zygotic Smchd1 siRNA knockdown on Dux and Zscan4 mRNA expression.

Because SMCHD1 negatively regulates the DUX gene family in humans, DUX activates EGA1 genes, and peak SMCHD1 nuclear staining coincided with the downregulation of Dux and Zscan4 mRNA and protein expression, we tested whether Dux and Zscan4 expression in two-cell stage embryos is indeed sensitive to Smchd1 expression. We used zygotic Smchd1 siRNA knockdown, followed by RT-qPCR assessment of Dux and Zscan4 mRNA expression at the two-cell and eight-cell stages (Fig. 3). Smchd1 siRNA treatment reduced Smchd1 mRNA expression in late two-cell stage embryos by an average of 72% and in eight-cell embryos by an average of 91%, indicating effective knockdown of expression, consistent with previous success in other studies (50). Conversely, Dux mRNA expression was increased 1.5-fold and 1.83-fold at these two stages. Additionally, Zscan4 mRNA was elevated 1.75-fold at the two-cell stage. Zscan4 mRNA was not reliably detected at the eight-cell stage in either control or siRNA knockdown embryos (not shown).

Fig. 3.

Quantitative (q)RT-PCR testing of the impact of structural maintenance of chromosomes flexible hinge domain-containing protein 1 (Smchd1) knockdown at the mid two-cell (A) and eight-cell (B) stages. A: mid two-cell embryos were collected from both control and siRNA knockdown for SMCHD1 after electroporation. B: eight-cell embryos were injected with siRNA for SMCHD1 or scrambled control by cytoplasmic microinjection. For both experiments, paired pools of 16–19 embryos corresponding to control and knockdown treatments were used, and three bioreplicates were performed for RT-qPCR testing. Relative expression of SMCHD1, double homeobox (DUX), and zinc finger and SCAN domain containing 4 (ZSCAN4) was measured. Expression was normalized to the endogenous control of Rpl18 for each gene at each stage. Statistical testing was conducted using t-test, with a Holm–Bonferroni correction for multiple testing. Statistical significance: *P <0.05; ***P < 0.001; ****P < 0.0001.

Bioinformatics analysis of SMCHD1 and DUX target genes in early embryos.

To further assess the potential role for SMCHD1 in regulating genes at the two-cell stage, we evaluated available evidence for DUX and SMCHD1 effects, genes that display increasing mRNA abundance during the two-cell stage. The analysis took into account chromatin immunoprecipitation data and DNA sequence motif analysis (8, 17, 27, 31, 60) and data for effects of gene ablation or knockdown (8, 13, 17, 31, 50, 52, 60) to identify putative DUX and SMCHD1 target genes (Supplemental Table S1; Supplemental data for this article can be found online at https://doi.org/10.6084/m9.figshare.8956724). First, we assessed effect on a set of 552 EGA1 genes, as described in materials and methods. Overall, 78% of the EGA1 genes are putative DUX targets, 62% are putative SMCHD1 targets, and 53% are putative targets of both factors (Fig. 4A). Importantly, 89% of the EGA1 genes are targets of either SMCHD1 or DUX, indicating that SMCHD1 can regulate the vast majority of EGA1 genes either directly or via DUX. Next, we examined the regulation of two sets of genes, comprising a portion of the total EGA2 class (3,393 genes total). These are EGA2 genes that display activation between the zygote and mid two-cell stages or between the mid two-cell and four-cell stages (denoted as “R” in Fig. 4, A and B). These genes display limited expression in two-cell embryos with higher levels of expression at later stages, indicating continued transcriptional repression during the two-cell stage. A majority (465/579; 80%) of these genes are also putative targets of SMCHD1, indicating a broader role for SMCHD1 than just terminating EGA1.

Fig. 4.

Structural maintenance of chromosomes flexible hinge domain-containing protein 1 (SMCHD1) is a putative regulator of a large proportion of embryonic genome activation 1 (EGA1) and EGA2 genes. Genes in the EGA1 and EGA2 classes were identified as described in materials and methods, and their overall patterns of regulation are summarized here. EGA1 and EGA2 genes are noted in the figure. A positive slope denotes a significant increase (false discovery rate < 0.05), a negative slope denotes a significant decrease, and horizontal lines indicate no significant change between indicated stages. The number of genes regulated are indicated for each line, and the numbers of putative SMCHD1 target genes are indicated in accompanying parentheses. A: genes with mRNA abundances increase from the one-cell to either early or mid two-cell stages. The * indicates a decrease in expression with a magnitude exceeding 2.5-fold change defining the EGA1 gene set. B: genes with mRNA abundances increase during the latter half of the two-cell stage. In both A and B, EGA2 genes are indicated, and those EGA2 genes that display partial repression during the two-cell stage (i.e., upregulated after the two-cell stage) are indicated by “R.”

DISCUSSION

Previous studies of mouse EGA1 revealed some factors that mediate transcriptional activation in two-cell stage mouse embryos (15, 29, 49, 58, 61, 62). To date, no factor mediating the termination of EGA1 has been identified. Our results demonstrate a novel maternal effect of SMCHD1 in repressing two genes (Dux and Zscan4) that initiate and maintain expression of EGA1 genes. As such, increasing expression of SMCHD1 from maternally expressed mRNA likely plays a key role in terminating EGA1.

Our data provide a detailed timeline of Dux, Zscan4, and Smchd1 mRNA and protein expression and demonstrate by siRNA knockdown that SMCHD1 represses Dux in mouse two-cell stage embryos, as seen in human muscle tissues (5, 47, 59), as well as repressing Zscan4. We noted that although the Zscan4 mRNA was delayed in its decline, it did not persist to the eight-cell stage after Smchd1 siRNA knockdown. One explanation for this is that the small amount of Smchd1 mRNA remaining encoded a sufficient supply of SMCHD1 to eventually terminate EGA1. Another explanation is that additional regulators may contribute to the timely termination of EGA1 but are unable to drive this termination at the correct time without SMCHD1 present at the normal level. Additionally, our analysis of 552 EGA1 genes revealed that SMCHD1 regulates both DUX-sensitive and DUX-insensitive EGA1 genes, indicating that activation of EGA1 genes is likely not accounted for by DUX activation alone, and that SMCHD1 likely impacts EGA1 genes by DUX-independent mechanisms as well. We note that 33 of the 38 known SMCHD1-interacting proteins are expressed in oocyte RNAseq data (data not shown), further suggesting the potential for SMCHD1 to regulate EGA1 in part through DUX-independent mechanisms. Additionally, that analysis indicated that some EGA1 genes are not regulated by either SMCHD1 or DUX, indicating that additional factors likely contribute to EGA1 activation and termination. Finally, we note that the peak in nuclear SMCHD1 staining occurred after Smchd1 mRNA displayed its maximal expression. This suggests possible regulation at the level of mRNA translation or nuclear SMCHD1 import.

SMCHD1 regulates a wide range of autosomal and X chromosome-linked genes, many in gene clusters (8, 23, 28, 30, 31, 39, 43, 47). Its function is mediated by its ability to dimerize with itself and other factors and by promoting DNA methylation, and its targets may include genes regulated by CCCTC-binding factor (30). SMCHD1 expression can either positively or negatively regulate embryonic gene expression (50), indicating diverse actions across the genome. Further study of SMCHD1 function should provide significant new insight into the control of EGA1.

In addition to terminating EGA1, the embryonic genome undergoes a transition to a transcriptionally repressive state that includes emergence of enhancer dependence in gene transcription (36, 45, 46). This transition contributes to subsequent correct gene regulation. As a chromatin regulator with broad actions in repressing genes across the genome, SMCHD1 is ideally suited to contribute to this vital process. Indeed, we observe that 80% of EGA2 genes, identified as being only partially activated during the two-cell stage and upregulated later, are putative SMCHD1 targets. We also note that over half (57%) of the >12K genes that are modulated at the four-cell or later stages during preimplantation development are also SMCHD1 targets (Fig. 5). This suggests that the events set in motion by oocyte-expressed SMCHD1 are extended by embryonically expressed SMCHD1. Consistent with this, our previous studies revealed a long-term effect of zygotic Smchd1 siRNA knockdown on embryo viability, including a nearly 40% reduction in birth rate (50), although Smchd1 mRNA levels in that study were recovered partially between the morula and blastocyst stage due to embryonic expression. Previous studies revealed that timely regulation of EGA in the embryo is vital and that sustained expression of Zscan4 can compromise development beyond the blastocyst stage (33). The early effect of Smchd1 knockdown on EGA1 indicates that a lack of timely EGA1 termination could contribute to the long-term effects of Smchd1 knockdown on viability through a mechanism similar to the effects of sustained Zscan4 expression and that embryonic expression is unable to fully substitute for the lack of oocyte-expressed SMCHD1.

Fig. 5.

Structural maintenance of chromosomes flexible hinge domain-containing protein 1 (SMCHD1) is a putative regulator of a large proportion of embryonic genome activation 3 (EGA3) and EGA4 genes. Genes in the EGA3 and EGA4 classes were identified as described in materials and methods and Fig. 4 applied to four-cell through blastocyst stage embryos. The number of genes regulated are indicated for each line, and the numbers of putative SMCHD1 target genes are indicated in accompanying parentheses.

More recently, an EGA1-like state in ESCs driven by DUX and contributing to developmental potency was reported (2, 12, 13, 15, 16, 19, 27, 33). Our previous study of zygotic Smchd1 knockdown effects revealed a reduction in the number of inner cell mass cells and increased expression of a number of trophoblast lineage genes (50). The observations here for two-cell stage EGA1 regulation suggest a potential role for SMCHD1 in regulating the EGA1-like in ESCs and early stem cells. This may account for the effects seen on inner cell mass and term development. Interestingly, female homozygous-null mutants for Smchd1 undergo midgestation arrest by embryonic day 10.5, and homozygous males are viable (6). This sexually dimorphic effect is due to SMCHD1’s role in X chromosome inactivation and a female-specific defect in trophoblast cells and placenta. The viability of homozygous embryos, both males (term) and females (embryonic day 10.5), appears at first glance to be inconsistent with an early role for SMCHD1 in embryos, the inner cell mass, ESCs, or derivatives. However, the homozygous-null progeny produced by heterozygous crosses may have benefited from maternal oocyte-derived Smchd1 expression. Whether the EGA1-like state in ESCs is regulated in the same manner as at the two-cell stage and whether other factors can compensate for SMCHD1 deficiency in ESCs, remain to be determined.

Collectively, the data presented here demonstrate another previously unknown role for SMCHD1 in the early embryo as the first transcription factor yet identified contributing to the termination of EGA1 and a continuing role for maternally expressed SMCHD1 in controlling EGA2. This novel role is distinct from any role previously described for SMCHD1 but appears to work in part through a known mediator of SMCHD1 function and a known mediator of EGA1, i.e., DUX. Moreover, we discovered that Dux is transcribed at the late one-cell stage, distinguishing it as one of the earliest known transcription factors identified to be regulated with a known function in EGA1. This suggests that the mechanism leading to EGA1 is initiated at the late one-cell stage rather than at the two-cell stage. This coincides with the acquisition of a transcriptionally permissive state demonstrated using nuclear transplantation methods (34, 37). The increase in SMCHD1 nuclear localization, peaking in concert with Dux and Zscan4 downregulation just 9 h after cleavage, must be subject to fine temporal control in the two-cell embryo. Our discovery of this novel role for SMCHD1 in the two-cell embryo thus provides insight into the complexity of mechanisms that regulate early embryonic genome activation. Additionally, these observations provide further understanding of oocyte quality and early gene regulatory events that define long-term embryo viability and the exquisite and complex nature of controlling mechanisms that provide for fine control of these events.

GRANTS

This work was supported in part by National Institutes of Health, Eunice Kennedy Shriver National Institute of Child Health and Human Development Grants R01 HD075903, R01 HD043092, and T32 HD087166, MSU AgBioResearch, and Michigan State University.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

K.E.L. conceived and designed research; K.A.V. and K.W. performed experiments; M.L.R., P.Z.S., and K.E.L. analyzed data; P.Z.S. and K.E.L. interpreted results of experiments; M.L.R., K.A.V., P.Z.S., and K.E.L. prepared figures; M.L.R., K.A.V., P.Z.S., and K.E.L. drafted manuscript; M.L.R., K.A.V., P.Z.S., K.W., and K.E.L. edited and revised manuscript; M.L.R., K.A.V., P.Z.S., K.W., and K.E.L. approved final version of manuscript.