A map of human embryo development that combines imaging, molecular, genetic and epigenetic data for comparisons to other species and across pathologies would be greatly beneficial for basic science and clinical applications. Here, we compared mRNA and protein expression of key mediators of DNA methylation and histone modifications between mouse and human embryos, embryos from fertile/infertile couples, and following growth factor supplementation. We observed that individual mouse and human embryos are characterized by similarities and distinct differences in DNA methylation and histone modification patterns especially at the single-cell level. In particular, while mouse embryos first exhibited sub-compartmentalization of different histone modifications between blastomeres at the morula stage and cell sub-populations in blastocysts, differential histone modification expression was detected between blastomeres earlier in human embryos at the four- to eight-cell stage. Likewise, differences in epigenetic mediator expression were also observed between embryos from fertile and infertile couples, which were largely equalized in response to growth factor supplementation, suggesting that select growth factors might prevent alterations in epigenetic profiles during prolonged embryo culture. Finally, we determined that reduced expression via morpholino technologies of a single histone-modifying enzyme, Rps6ka4/Msk2, resulted in cleavage-stage arrest as assessed by time-lapse imaging and was associated with aneuploidy generation. Taken together, data document differences in epigenetic patterns between species with implications for fertility and suggest functional roles for individual epigenetic factors during pre-implantation development.
Recent studies in mouse and human have explored molecular, genetic, epigenetic and imaging profiles of pre-implantation embryos at different stages during development (1–6). A fundamental component of mammalian pre-implantation development is the erasure and re-establishment of epigenetic marks (epigenetic reprogramming) (7). Following fertilization, the paternal and maternal genomes are extensively modified and reset prior to implantation, which is thought to be required to establish the totipotency of the newly formed embryo (8). The two main types of epigenetic mechanisms are DNA methylation and histone modifications, which work together to affect gene expression in a potentially heritable manner (without altering DNA sequence) and influence chromatin structure (9–11). DNA methylation is mediated by a family of DNA methyltransferases (DNMTs) that catalyze the transfer of a methyl group to the 5′-position of cytosine residues within CpG dinucleotides usually resulting in effective gene silencing (12). Although global DNA methylation patterns in pre-implantation development have been documented in several species, the elucidation of DNMT expression particularly in early human embryos is far from complete, with focus on just a few stages of pre-implantation development and/or particular DNMT family members (13–16). Histone modifications include, but are not limited to, the phosphorylation of serine residues, acetylation of lysine residues and the methylation of either lysine or arginine residues, all of which are mediated by different histone-modifying enzymes and may affect biological outcome (17). While some studies have analyzed a subset of histone modifications in pre-implantation embryos from different species, data remains limited, especially in the human (18–21).
In this study, we compared expression of key regulators of DNA methylation and histone modifications between the different stages of mouse and human pre-implantation development, between embryos from fertile and infertile couples, and following media supplementation with a growth factor cocktail. We then assessed function via reduction in expression of a particular epigenetic regulator implicated in both mouse and human pre-implantation development. Human embryos were obtained from a unique set that were cryopreserved at the one-cell stage prior to assessment of quality and thus, likely to be representative of ‘fresh’ embryos from conception cycles (22,23), which have been shown to have similar potential for successful development, implantation, pregnancy and delivery as previously described (24).
Differential DNMT and histone-modifying enzyme mRNA expression patterns in mouse and human embryos
We first examined which DNMTs were associated with different stages of pre-implantation development in mouse and human embryos (Fig. 1A and B). For this purpose, we evaluated DNMT1, DNMT2, DNMT3A, DNMT3A2, DNMT3B and DNMT3L expression in individual mouse or human embryos at the one-cell, two-cell, four-cell, eight-cell, morula and blastocyst stages by microfluidic quantitative RT-PCR (Q-PCR; Supplementary Material, Table S1). As shown in Figure 2A and Supplementary Material, Figure S1a, some differences in DNMT expression between mouse embryos were detected beginning at the two-cell stage when zygotic genome activation (ZGA) begins (25), until the eight-cell stage of development; less variation in DNMT expression patterns was observed between individual mouse one-cell, morula and blastocysts. In human embryos, greater variation in DNMT expression was detected between individual human embryos at the cleavage stage than in the mouse and this variation occurred later in development [between the four- and eight-cell stages, coincident with the major wave of embryonic genome activation (EGA; 3,26), and the morula stage; Fig. 2B and Supplementary Material, Fig. S1b]. Thus, no common DNMT expression pattern could be detected between mouse and human embryos, confirming the differences in global DNA methylation levels observed in each species (15,27–30).
We next examined the expression of enzymes that mediate particular histone modifications in both mouse and human pre-implantation embryos (Fig. 1A and B). To accomplish this, we focused on histone-modifying enzymes that are involved in the phosphorylation of serine residues [Aurora B Kinase (AURKB), Ribosomal Protein S6 kinase, 90 kDa, polypeptide 5 (RPS6KA5/MSK1) and RPS6KA4/MSK2], acetylation of lysine residues [activating transcription factor 2 (ATF2), K (lysine) acetyltransferase 5 (KAT5), Elongator complex protein 3 (ELP3)/KAT9 and Histone acetyltransferase 1 (HAT1)/KAT1] and the methylation of either lysine residues [euchromatic histone-lysine N-methyltransferase 2 (EHMT2)/G9A, SET domain containing 7 (SETD7)/KMT7 and SET domain bifurcated 1 (SETDB1)/KMT1E] or arginine residues [Protein arginine methyltransferase 5 (PRMT5)] (Supplementary Material, Table S1). Analogous to DNMT expression patterns, the levels and timing of expression of most histone-modifying enzymes differed between mouse and human embryos (Fig. 2A and B and Supplementary Material, Fig. S1a and b). Indeed, only ATF2, KAT5, RPS6KA4/MSK2, PRMT5 and SETDB1 were similarly expressed in mouse and human pre-implantation embryos. Moreover, the greatest differences in histone-modifying enzyme expression were detected in human embryos between the four- to eight-cell and morula stages. This variability in both DNMT and histone-modifying enzyme expression correlated with differences in the expression of the cell lineage regulators, POU class 5 homeobox 1 (POU5F1)/OCT-4 and caudal-type homeobox 2 (CDX2), markers that have been shown to be predictive of inner cell mass (ICM) and trophectoderm lineage patterning (4,31), respectively (Fig. 2A and B and Supplementary Material, Fig. S1a and b).
Single-cell analysis of epigenetic regulators in cleavage-stage mouse and human embryos
We next sought to evaluate expression of epigenetic regulators in individual blastomeres from the same embryo. For this purpose, Day 3 mouse and human embryos were disassembled to single cells at approximately the eight-cell stage, a time when the embryonic genomes of both species should be activated (3,25,26), and DNMT and histone-modifying enzyme expression was analyzed by Q-PCR (Fig. 1A and B). While some transcripts such as Dnmt3a2 were undetectable in individual blastomeres from mouse embryos, which may be due to less total RNA in mouse versus human embryos (6,32), both mouse and human blastomeres exhibited a relatively similar degree of variation in DNMT and histone-modifying expression within the same embryo. This variation was also reflected in blastomere differences between OCT-4 and CDX2 expression as well as other pluripotency regulators, including NANOG Homeobox (NANOG), sex determining region Y-box 2 (SOX2), YIN-YANG-1 (YY1) and myelocytomatosis oncogene (MYC)/c-MYC, and trophectoderm regulators such as eomesodermin (EOMES) (Fig. 2C and D and Supplementary Material, Fig. S1c and d). However, differences in the expression of pluripotency, trophectoderm and epigenetic regulators were not universal to all genes expressed at this stage of development since little or no variation was observed in the expression of the maternal-effect genes, NLR family, pyrin domain containing 5 (NLRP5)/MATER and zygote arrest 1 (ZAR1), or the zygotic/EGA genes, DEAD box polypeptide 20 (DDX20)/GEMIN3, KH domain containing 3, subcortical maternal complex member (Khdc3)/Filia and KH domain containing 3-like, subcortical maternal complex member (KHDC3L)/ECAT1 (Fig. 2C and D and Supplementary Material, Fig. S1c and d).
Analysis of histone modification sub-compartmentalization in mouse and human embryos
Using immunofluorescence and multi-color confocal microscopy analysis, we assessed changes in the expression of certain histone modifications in mouse and human embryos at the protein level (Fig. 1A and B). We focused our attention on the expression and localization of Histone H3 serine 10 phosphorylation (H3-S10P), H3-S28P, histone H4 lysine 16 acetylation (H4-K16acetyl) and histone H4 arginine 3 dimethylation (H4-R3me2) given that the mRNAs for the corresponding histone-modifying enzymes were expressed in mouse and human embryos. The expression of histone H3 lysine 4 tri-methylation (H3-K4me3) was also evaluated since H4-R3me2 can be both a transcriptionally active and repressive mark (33,34). We validated specificity of the antibodies by immunofluorescence in the undifferentiated human embryonic stem cell (hESC) lines, H9 (XX) and HSF1 (XY) (Fig. 3A). Negative controls included single-color antibody staining to discount potential interference between different confocal channels as well as isotype controls (Supplementary Material, Fig. S2a and b).
Analysis of expression and localization of histone modifications in mouse embryos indicated that each blastomere exhibited similar epigenetic marks during the early cleavage stages of development both by single-frame confocal imaging (Fig. 3B and Supplementary Material, Fig. S3a) and three-dimensional modeling of Z-stacked confocal images (Fig. 3D). However, there were differences in the expression levels of the same histone modification between blastomeres within the same embryo (indicated by white-dashed arrow), which is in accordance with previous findings (19). As Figure 3B and Supplementary Material, Figure S3a demonstrate, it was not until later in development (between the morula and blastocyst stages) that sub-compartmentalization of different epigenetic marks such as H3-S28P and H4-R3me2 was observed between blastomeres or cell populations within the same embryo (shown by white solid arrows). Notably, we also observed H4-R3me2 and H3-K4me3 expression in one or both of the pronuclei and H3-S28P expression in one of the polar bodies in mouse embryos at the one-cell stage (Fig. 3B and Supplementary Material, Fig. S3a and c), suggesting that H3-S28P may be a marker of the first polar body since H3-S28P is associated with mitosis and the first polar body can undergo cytokinesis at least in the mouse (35). Expression of H3-S10P appeared to localize to small structures in the cytoplasm of cleavage-stage mouse embryos (Fig. 3B and D), which may be explained by previous findings that histones can bind strongly to isolated mitochondria, the phosphorylation of histone H3 is associated with mitosis and mitotic progression is closely integrated with mitochondrial dynamics (36–38). We also note differences in the expression pattern of H3-S10P as well as H4-K16acetyl to that of previous reports (39,40) and this is likely due to differences in fixation. The previous studies used methods that included fixation with paraformaldehyde, which is known to cause DNA–protein crosslinking and the loss of organelles such as mitochondria when combined with Triton X-100 for permeabilization (41,42), while methanol fixation was used in the present study to avoid histone crosslinking.
In contrast to mouse embryos, human embryos first exhibited differential histone modification expression patterns between blastomeres by the four- to eight-cell stages when analyzed by single-frame confocal imaging (Fig. 3B and Supplementary Material, Fig. S3b) or three-dimensional modeling of Z-stacked confocal images (Fig. 3D). Individual human blastomeres expressed either completely different histone modification marks such as H3-S28P and H4-R3me2 or were entirely devoid of histone modification expression observed in other blastomeres within the same embryo (indicated by white solid arrows). Similar to the mouse, however, sub-compartmentalization of histone modifications was also detected in human embryos later in development between blastomeres at the morula stage and within the ICM and trophectoderm sub-populations at the blastocyst stage (Fig. 3B and Supplementary Material, Fig. S3b). Immunostaining human embryos for H3-K4me3 also revealed a differential histone modification expression pattern at the eight-cell stage of human development in addition to positive signals for H3-K4me3 and H3-S28P in one of the pronuclei and polar bodies at the one-cell stage, respectively (Fig. 3C and Supplementary Material, Fig. S3b).
Reduced Rps6ka4/Msk2 expression induces mitotic arrest and micronuclei formation in mouse embryos
In order to determine whether specific histone-modifying enzymes were implicated in the progression of pre-implantation development beyond ZGA, we microinjected morpholino oligonucleotides (MOs) at the one-cell stage and monitored embryo development by time-lapse imaging (Fig. 1A). To facilitate an understanding of both mouse and human development, we initially focused our attention on the histone-modifying enzymes that were similarly expressed in both mouse and human embryos. However, ATF2, KAT5, PRMT5 and SETDB1 had multiple transcript variants, a high percentage of GC base content around the first exon and/or an atypical start codon, making MO design difficult for these histone-modifying enzymes (43). Given that relatively little is known about the role of serine phosphorylation in mammalian pre-implantation development and the importance of mitosis at the cleavage stage, we focused our efforts on RPS6KA4/MSK2, a mitogenic factor that is known to be involved in the G1 phase of the cell cycle (44). We first sought to determine whether we could specifically knockdown Msk2 expression using a MO designed to target the translation start site of Msk2. To accomplish this, E8.5 mouse embryonic germ (mEG) cells, which express high levels of Msk2, were mock transfected or transfected with either a 3′-carboxyfluorescein-labeled standard control or 3′-carboxyfluorescein-labeled Msk2 morpholino (Msk2 MO) since analysis of Msk2 knockdown in pre-implantation embryos at the protein level would require several hundred embryos. Msk2 expression was analyzed by western blot 48–144 h later and showed that in contrast to the mock transfection or control MO, the nucleofected Msk2 MO knocked down Msk2 protein expression in 8.5mEGs (Fig. 4A).
Our next objective was to determine whether microinjection of the Msk2 MO using the injection of the 3′-carboxyfluorescein-labeled MO as a control had any effects on pre-implantation development. Initially, we determined that both non-injected and nuclease-free water-injected one-cell mouse embryos exhibited similar rates of blastocyst formation at ∼80%, while 75% blastocyst formation rates were observed in standard control injected embryos (Fig. 4B and C and Table 1) as demonstrated by time-lapse imaging (Supplementary Material, Movies S1 and S2). More importantly, when the gene expression profiles of non-injected and standard control injected blastocysts were compared, differences in the expression of only 2 out of the 50 genes tested could be detected (Supplementary Material, Fig. S4a). The two genes that were expressed at higher levels in non-injected embryos were the trophectoderm genes, Cdx2 and Eomes, which may be explained by the finding that a delay (approximately 12 hours) in blastocyst formation was observed in control injected over non-injected embryos (Supplementary Fig. S4b and c).
|Non-injected (NI)||Water-injected (DW)||3′-Carboxyfluorescein morpholino-injected (COF)||Msk2 morpholino-injected (Msk2 MO)||Msk2 morpholino + modified mRNA-injected (Msk2 MO + mRNA)|
|Number of independent experiments||7||2||7||7||3|
|Number of embryos reaching blastocyst stage||61||15||55||1||12|
|Total number of embryos||73||19||71||73||33|
|Non-injected (NI)||Water-injected (DW)||3′-Carboxyfluorescein morpholino-injected (COF)||Msk2 morpholino-injected (Msk2 MO)||Msk2 morpholino + modified mRNA-injected (Msk2 MO + mRNA)|
|Number of independent experiments||7||2||7||7||3|
|Number of embryos reaching blastocyst stage||61||15||55||1||12|
|Total number of embryos||73||19||71||73||33|
A table displaying the number of independent experiments performed, the number of embryos that reached the blastocyst stage out of the total number of embryos with the corresponding percentage in the different microinjection groups. Note that only one embryo became a blastocyst in the Msk2 MO-injected group and this Msk2-induced embryo arrest was partially rescued by co-injecting with a similar concentration (20 ng/µl) of modified Msk2 mRNA (Msk2 MO + mRNA). This table is an accompaniment to Figure 4B.
We then evaluated the affects of Msk2-specific knockdown by comparing similar concentrations of Msk2 and standard control injected MOs. Only 1.4% blastocyst formation rates were observed with the Msk2 MO-injected one-cell mouse embryos (Fig. 4B and Table 1) and co-injection with a modified Msk2 mRNA designed to avoid direct targeting of the Msk2 morpholino (Supplementary Fig. S5a and b) partially rescued this defect by significantly increasing blastocyst development to ∼36% (P < 0.001). Notably, of the ∼75 mouse embryos injected with the Msk2 MO, only a single embryo developed to the blastocyst stage (Fig. 4C and Table 1). This exception may be due to minor differences in the volume of MO injected into each embryo as demonstrated by the slight variation in fluorescent intensities between embryos following sequential injection (Fig. 4D). However, we observed that there was a decrease in the expression of Msk2 detected in the one Msk2 MO-injected embryo that reached the blastocyst stage compared with the level of Msk2 expression detected in non-injected, water or 3′-carboxyfluorescein-labeled MO-injected embryos (Fig. 4E).
Following our observation that the knockdown of Msk2 in mouse embryos significantly reduced blastocyst formation rates, we then examined the developmental timing of embryo arrest. A closer examination of each stage of pre-implantation development revealed that Msk2-medidated arrest most commonly occurred in three- to eight-cell embryos (Fig. 5A). More specifically, arrest was observed in ∼40% of three- to four-cell embryos and 51% of six- to eight-cell embryos following injection of the Msk2 MO (P < 0.001; Fig. 5B). In addition, we also determined that the few embryos that arrested at the one-cell stage was due to polyploidy (Fig. 5C; upper panel) and that a decrease in Msk2 expression was observed in non-polyploid Msk2 MO-injected two-cell mouse embryos prior to arrest in comparison with 3′-carboxyfluorescein-labeled MO-injected embryos at a similar stage (Fig. 5C; middle and lower panels). Finally, when we evaluated the epigenetic expression profiles in 3′-carboxyfluorescein-labeled MO- or Msk2 MO-injected embryos, we detected considerably lower expression levels of Dnmt1 and Aurkb, additional epigenetic regulators that have known mitotic functions (45,46), in Msk2 MO embryos (Fig. 5D), suggesting that Msk2 knockdown had adverse effects on other epigenetic mediators in developing embryos. Assessment of embryo behavior by time-lapse imaging also revealed that Msk2 MO-injected embryos exhibited the unusual phenotype of increased internal blastomere movement and lysis upon arrest (Fig. 5E and Supplementary Material, Movie S3), which was not observed in Msk2 mRNA rescued blastocysts (Supplementary Material, Movie S4). We note that this phenotype resembled the aberrant mitosis and cell death observed during mitotic catastrophe (47) and reasoned that perhaps the lysis of blastomeres induced by Msk2-mediated knockdown may constitute a mechanism for mouse embryos to avoid chromosomal instability. In support of this, we immunostained mouse embryos with Lamin-B1, a nuclear envelope marker, and observed the formation of Lamin-B1 encapsulated micronuclei in Msk2 MO-injected, but not 3′COF MO-injected embryos prior to lysis (Fig. 5F).
Association between MSK2 expression, mitosis and aneuploidy in human embryos
Given our findings that the embryo arrest in Msk2 MO-injected embryos was associated with mitotic arrest and micronucleation, our next aim was to determine whether MSK2 was involved in abnormal mitotic divisions and possible aneuploidy generation in human embryos. Our experimental design for this is as shown in Figure 1B and made use of previous studies. By measuring previously identified cell cycle parameters predictive of blastocyst formation prior to EGA (1), we recently demonstrated that blastomere behavior reflects human embryo ploidy by the four-cell stage (5) and that chromosome-containing micronuclei/fragments may contribute to the complex aneuploidy observed in cleavage-stage human embryos (5,48,49). Therefore, we similarly analyzed cell cycle parameter timing in human embryos cultured from the zygote to approximately the four-cell stage using time-lapse imaging and evaluated each embryo for the expression of MSK2 and LAMIN-B1 (Fig. 1B). We detected significantly lower MSK2 expression in blastomeres with visible micronuclei and abnormal cell cycle parameters, whereas high MSK2 expression was observed in blastomeres with intact primary nuclei and normal parameter timing (Fig. 6A and Supplementary Material, Fig. S6; P < 0.01). In those embryos with micronuclei, we also observed a lack of both H3-S10 and H3-S28 phosphorylation, two of the histone modifications that MSK2 mediates (50,51), but not H4-K16 acetylation, which has been shown to be involved in DNA repair and programmed cell death (52) (Fig. 6B). Thus, the absence of MSK2 expression and its corresponding histone modifications appears to be related to both abnormal mitotic divisions and micronuclei formation and may potentially contribute to aneuploidy generation in human embryos.
To further investigate the role of MSK2 in human pre-implantation development, we analyzed the expression of both full-length MSK2 (MSK2-2) as well an alternative MSK2 splice variant (MSK2-1; Supplementary Material, Table S2), which does not exist in the mouse (53), in single blastomeres from cleavage-stage embryos that were determined to be either euploid (Supplementary Material, Fig. S7a) or aneuploid (Supplementary Material, Fig. S7b) by array-comparative genomic hybridization (A-CGH). While both MSK2 isoforms were more highly expressed in the blastomeres of euploid embryos (n = 13), low to moderate levels of MSK2 expression were detected in aneuploid embryos (n = 20) to support a role for MSK2 in regulating human embryonic aneuploidy generation (Fig. 6C). Given that between 50 and 80% of cleavage-stage human embryos are chromosomally abnormal (5,48,49), this may also help explain why MSK2 expression was only detected in a few blastomeres from human embryos at the eight-cell stage (Fig. 2D). In addition, DNMT1, AURKA and AURKB, but not MSK1 were also expressed in a similar pattern as MSK2 expression in euploid and aneuploid embryos to suggest an association between MSK2, DNMT1, AURKA and AURKB function.
Growth factor supplementation restores epigenetic expression levels in human embryos
Based on previous studies suggesting that media supplementation of certain autocrine/paracrine factors can enhance oocyte maturation and early embryo development (54,55), our final objective was to determine whether the addition of defined growth factors to culture media had any effects on the expression of epigenetic regulators in human embryos by Q-PCR (Fig. 1B). For this purpose, we evaluated DNMT and histone-modifying enzyme expression in embryos cultured with a growth factor cocktail containing brain-derived neurotrophic factor (BDNF), insulin-like growth factor 1 (IGF-I), epidermal growth factor (EGF), granulocyte macrophage colony-stimulating factor (GM-CSF), basic fibroblast growth factor (bFGF)/FGF2 and glial cell line-derived neurotrophic factor (GDNF), which were selected based on the aforementioned studies (54,55) as well as the confirmation of their corresponding receptor expression in human oocytes (56). Although similar blastocyst formation rates (∼30%) were observed with human zygotes cultured in the absence or presence of the growth factor cocktail, comparison of gene expression between the two groups revealed that media supplementation induced changes in DNMT and histone-modifying enzyme expression beginning at the eight-cell stage (Supplementary Material, Fig. S8). Analysis of epigenetic regulator expression in blastocysts from both fertile and infertile patients also revealed differences in DNMT and histone-modifying enzyme expression levels (Fig. 6D). More importantly, when epigenetic regulator expression was compared in infertile, fertile and growth factor treated blastocysts, we determined that media supplementation either partially or fully restored the expression levels of several DNMTs and histone-modifying enzymes, including MSK2, to that observed in embryos from fertile patients (Fig. 6D). Thus, our data suggest that the addition of growth media may potentially reduce chromosomal instability and improve IVF outcomes by positively influencing the expression of epigenetic regulators in human embryos.
To our knowledge, this is the first study to directly compare DNA methylation and histone modification patterns in both the mouse and human, individual embryos and single cells, at the mRNA as well as protein level, in embryos from fertile and infertile couples and following growth factor supplementation. In addition, we further interrogated the function of a particular epigenetic regulator in mouse and human pre-implantation development. While DNMT expression patterns differed substantially between mouse and human embryos, mRNA expression levels and timing of the histone-modifying enzymes, ATF2, KAT5, MSK2, PRMT5 and SETDB1 were similar in these two species. Since both ATF2 and KAT5 are involved in the acetylation of lysine residues, MSK2 in the phosphorylation of serine residues, PRMT5 in the methylation of arginine residues and SETDB1 in the methylation of lysine residues (50,51,57–60), this suggests that expression of at least one member from each histone-modifying enzyme class is conserved between human and mouse development (Fig. 7). There also remains the possibility that homologous DNMT genes with analogous functions could be similarly expressed between mouse and human embryos (61–63).
The greatest variation in DNMT, histone modification and histone-modifying enzyme expression was observed between human embryos beginning at the four- to eight-cell stages, which we attributed, at least in part, to differences between single cells within the same embryo. We previously demonstrated by single-cell gene expression profiling that human blastomeres develop cell autonomously (1) and this is supported by the findings presented here since we show that individual blastomeres within the human embryo can simultaneously express different epigenetic marks. Given that some of these epigenetic marks are differentially associated with mitosis, DNA repair and transcriptional activation and/or silencing (33,34,51,52), findings also support the concept that blastomeres of the human embryo may have divergent developmental potential beginning at the four- to eight-cell stages (Fig. 7). Given the extension of epigenetics to microRNAs (miRNAs) and our previous findings of distinct differences in the expression of miRNA biogenesis genes in normal versus arrested one-cell and two-cell human embryos (1) and normal versus parthenogenetic morulas and blastocysts (56), the likely importance of epigenetic-miRNA regulation beyond the two-cell stage and especially in the context of aneuploidy should be further investigated.
In order to functionally validate our findings, we reduced expression of the histone-modifying enzyme, Msk2, and examined the effects in the mouse for potential correlation to human pre-implantation development. Although knockout mice for Msk2 are reported to be viable and fertile, it is not clear whether this conclusion was drawn from heterozygous or homozygous crosses as these details are not provided in the study (50). Moreover, we suspect that we observed such a profound phenotype in our study since the Msk2 morpholino may potentially target all Msk2 mRNAs whether embryonic, maternally or paternally derived, the latter of which may be associated with hypomethylation and histone-retained regions in sperm (64,65). Nonetheless, another study has shown that Msk1/Msk2 double knockout mice exhibit hypersentivity to endotoxins and prolonged inflammation (66) and thus, it will be important to assess fecundity in Msk1 and/or Msk2 mouse knockouts generated by site-specific enginneered nucleases (67,68) or in specific cells and tissues using conditional Msk2 knockout mice (69). Following injection of the Msk2 MO in one-cell embryos here, we observed almost complete embryo arrest at the three- to eight-cell stage and an unusual phenotype in which increased cell movement and eventual blastomere lysis was detected. Based on the observations that reduced Msk2 expression had effects on the expression of Dnmt1 and Aurkb, both of which are known to have important roles in mitosis and possibly ploidy status (45,46,70), this suggests that the embryo arrest observed after Msk2 injection is likely due to a defect in mitosis. In addition, we further suggest that the blastomere lysis observed following Msk2 MO injection resembles cellular events described during mitotic catastrophe (47), may represent a mechanism for embryos to avoid chromosomal instability and explain the low aneuploidy rates observed in mice (71). A correlation between MSK2 expression and human aneuploidy generation is supported by our findings of abnormal cell cycle parameter timing, ploidy status, micronuclei formation, reduced MSK2 expression and alterations in histone modifications mediated by MSK2, in human embryos (Fig. 7), the latter of which has similar indications in lower organisms (72). Whether low MSK2 expression is a cause or consequence of aneuploidy is unknown, but based on previous findings of a prolonged G1 phase in MSK2-deficient cells (44) and that a closely related family member is a kinetochore-associated protein that participates in the spindle assembly checkpoint (73), it is likely that reduced MSK2 expression prevents the localization and function of several proteins involved in cell cycle progression. Thus, further investigation is required to determine whether MSK2 prevents aneuploidy generation directly or indirectly via DNMT1, AURKA/AURKB and/or other cell cycle associated genes and the exact mechanism by which MSK2 mediates its function.
Finally, while we did observe differences in epigenetic expression profiles between fertile and infertile patient embryos, the incubation of embryos from infertile couples with a growth factor cocktail either partially or fully restored the expression levels of numerous epigenetic factors beginning at the eight-cell stage when the major wave of EGA occurs (3,26). This suggests that the addition of select growth factors to embryo culture media may prevent alterations in epigenetic profiles (Fig. 7) and potentially improve developmental competence of patient embryos subjected to prolonged culture. Altogether, our data suggest that although epigenetic mechanisms do not alter DNA sequence, they may have essential roles for pre-implantation development and possible pregnancy outcomes, especially in the context of IVF. There may also be more long-term implications of these findings, including an association with the fetal origins of adult disease and transgenerational inheritance of epigenetic changes or defects. The work described here contributes to our understanding of the epigenetic requirements for normal embryogenesis, in cases of human reproductive failure, and may be applied to clinical improvement of assisted reproduction.
MATERIALS AND METHODS
Sample source and procurement
Approximately 150 supernumerary human embryos subsequently donated for non-stem research were obtained with written informed consent from the Stanford University Regenerative Medicine through the Ethical procurement of Nonviable or Excess cellular Waste (RENEW) Biobank and the Reproductive Medicine Center at the University of Minnesota, which received embryos from the Lutheran General Hospital IVF Program (Park Ridge, IL, USA) when it closed in 2002. De-identification and molecular analysis was performed according to the Stanford University Institutional Review Board (IRB)-approved protocol #10466 entitled ‘The RENEW Biobank’ (74) and the University of Minnesota IRB-approved protocol #0306M49242 entitled ‘Stage-Specific Genomic Characterization of Human Pre-implantation Embryos’. No protected health information was associated with each of the embryos. The average maternal age was 34 years old and the most common cause of infertility was unexplained at 35% of couples. Embryos from patients classified as fertile was based on the use of donor materials for IVF cycles and the inclusion of embryos from fertile couples seeking gender selection or Human Leukocyte Antigen (HLA) typing.
Human embryo thawing and culture
Human embryos were thawed by a two-step rapid thawing protocol using Quinn's Advantage Thaw Kit (CooperSurgical, Trumbull, CT, USA) as previously described (1,5). In brief, either cryostraws or vials were removed from the liquid nitrogen and exposed to air before incubating in a 37°C water bath. Once thawed, embryos were transferred to a 20 µl microdrop of either Quinn's Advantage Cleavage Medium (CooperSurgical) supplemented with 10% SPS between Days 1–3 or Quinn's Advantage Blastocyst Medium (CooperSurgical) with 10% SPS after Day 3 under mineral oil (Sigma, St Louis, MO, USA). Approximately half of the embryos were also cultured in the presence of a growth factor cocktail containing 10 ng/ml BDNF (PeptroTech Inc., Rocky Hill, NJ, USA), 40 ng/ml IGF-I (Sigma-Aldrich, St Louis, MO, USA), 5 ng/ml EGF (R&D Systems, Inc., Minneapolis, MN, USA), 2 ng/ml GM-CSF (R&D Systems, Inc.), 0.5 ng/ml FGF2 (R&D Systems, Inc.) and 10 ng/ml of GDNF (R&D Systems, Inc.). All embryos were cultured at 37°C with 6% CO2, 5% O2 and 89% N2 and embryo development was monitored daily by microscope for up to 7 days.
Mouse embryo collection and culture
Three- to five-week-old wild-type C57BL6 × DBA/2 (B6D2F1) female F1 mice were obtained from Charles River (Hollister, CA, USA) and superovulated by intraperitonial injections of 10 IU of Pregnant Mare's Serum Gonadotropin (PMSG; Sigma) followed by 10 IU of human Chorionic Gonadotropin (hCG; Sigma) 48 h later and mated overnight with wild-type B6D2F1 male mice. Females were sacrificed by cervical dislocation ∼18 h after hCG injection and their oviducts, along with the adjacent uterine and ovarian tissue, were removed and transferred to EmbryoMax M2 Medium (Millipore, Billerica, MA, USA) for dissection. One-cell embryos were released from the oviducts and cumulus cells were removed from the embryos by hyaluronidase (Sigma) treatment and gentle pipetting. Two pronuclei stage embryos were recovered, pooled from 8 to 10 females in M2 media and similarly cultured in Quinn's Advantage Cleavage Medium with 10% SPS at 10 embryos per 20 µl microdrop. All procedures involving animals were performed under the Institutional Animal Care and Use Committee (IACUC) protocol #16146 entitled ‘Molecular Analysis of Embryogenesis and Gametogenesis,’ which was approved by the Administrative Panel on Laboratory Animal Care (APLAC) at Stanford University.
Single embryo and cell gene expression analysis
Gene expression was analyzed in mouse and human embryos using the BioMark Dynamic Array microfluidic system (Fluidigm Corp., So. San Francisco, CA, USA). The zona pellucida (ZP) was removed by treatment with Acidified Tyrode's Solution (Millipore) and ZP-free embryos were washed in Quinn's Advantage Medium with Hepes plus 5% SPS three times and then quick frozen in phosphate buffered solution (PBS; Invitrogen, Carlsbad) with 0.1% bovine serum albumin (BSA; Sigma-Aldrich) on dry ice for storage at −80°C until use. Cleavage-stage human embryos were dissembled to single cells in Quinns Advantage Medium with HEPES (Ca and Mg Free; Copper Surgical) plus 10% Human Albumin (Cooper Surgical), while mouse embryos were dissembled in 0.25% trypsin-EDTA (Gibco) plus 10% human albumin both at 37°C with gentle pipetting. Individual embryos and single cells were pre-amplified according to the manufacturer's protocol (Fluidigm Corp.) using the CellsDirect One-Step qRT-PCR kit (Invitrogen) and 20× TaqMan gene expression assays (Applied Biosystems, Foster City, CA, USA) or gene-specific primers designed to span exons using Primer-BLAST (NCBI). Pre-amplified cDNA was loaded into the sample inlets of a 96 × 96 dynamic array (DA; Fluidigm Corp.) and assayed in triplicate. The expression of between 6 and 10 housekeeping genes was analyzed as a control. Calculated normalized relative quantity (CNRQ) values were calculated and normalized to the two or three most stable housekeeping genes using the qBasePlus 1.3 analysis software (http://www.biogazelle.com) as previously described (75) and graphed using Gene-E (http://www.broadinstitute.org/cancer/software/GENE-E/).
The Histone H3-S10P mouse monoclonal antibody (clone 6G3) was purchased from Cell Signaling Technology, Inc. (Danvers, MA, USA), whereas the H3-S28P rat monoclonal (clone HTA28) and the H4-R3me2 rabbit polyclonal antibody (catalog #39275) were obtained from Thermo Fisher Scientific (Rockford, IL, USA) and Active Motif (Carlsbad, CA, USA), respectively. The Histone H3-K4me3 rabbit monoclonal antibody (clone MC315) was purchased from Millipore, while the Histone H4-K16acetyl goat polyclonal antibody (catalog #sc-8662) was obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA, USA). The Msk2 rabbit polyclonal (catalog # ab42282), Lamin-B1 rabbit polyclonal (catalog #ab16048) and Gapdh rabbit polyclonal (catalog #ab9485) antibodies were obtained from Abcam (Cambridge, MA, USA).
The Alexa Fluor donkey anti-mouse 405 antibody (custom synthesis), donkey anti-rat 488, donkey anti-sheep 488, donkey anti-goat 568 and donkey anti-rabbit 647 antibodies were purchased from Invitrogen (Carlsbad, CA, USA) for confocal imaging analysis. As controls, the mouse, rat, rabbit and goat IgG isotype antibodies were obtained from Vector Laboratories (Burlingame, CA, USA).
Confocal imaging analysis and three-dimensional modeling
ZP-free mouse and human embryos were obtained as described above and washed in PBS plus 0.1% BSA and 0.1% Tween-20 (PBS-T; Sigma-Aldrich) before fixation in either 100% cold methanol for 20 min at −20°C to avoid DNA-histone crosslinking or 4% paraformaldehyde in PBS (USB Corp., Cleveland, OH, USA) for 20 min at room temperature (RT) when visualizing non-histone associated proteins. Once fixed, the embryos were washed three times in PBS-T to remove any residual fixative and permeabilized in 1% Triton X-100 (Sigma-Aldrich) for 1–2 h at RT. Following permeabilization, the embryos were washed three times in PBS-T and then blocked in 4% normal donkey serum (Jackson ImmunoReasearch Laboratories, Inc., West Grove, PA, USA) in PBS-T overnight at 4°C. The embryos were incubated with primary antibodies in PBS-T with 1% donkey serum sequentially for 1 h each at RT at the following dilutions: mouse H3-S10P (1:200), rat H3-S28P (1:200), rabbit H4-R3me2 (1:1000), rabbit H3-K4me3 (1:200) and goat H4-K16Ac (1:200) and/or with rabbit Msk2 (1:400) and Lamin-B1 (1:1000). Primary signals were detected using the appropriate 405, 488, 568 or 647 conjugated donkey Alexa Fluor secondary antibody (Invitrogen) at a 1:250 dilution at RT for 1 h in the dark. Immunofluorescence was visualized by sequential imaging, whereby the channel track was switched each frame to avoid cross-contamination between channels, using a Zeiss LSM510 Meta inverted laser scanning confocal microscope described here: http://nisms.stanford.edu/Equipment/LSM510Meta01v01.html. The instrument settings, including the laser power, pinhole and gain, were kept constant for each channel to facilitate semi-quantitative comparisons between developmental stages and mouse and human embryos. Confocal sections were captured at 1 mm intervals throughout the whole embryo and processed in ImageJ (NIH) for Z-stack imaging analysis. Three-dimensional reconstructions of embryos were accomplished with IMARIS (Bitplane). Semi-quantitative analysis of MSK2 expression in human embryos was performed by measuring the mean pixel intensity of immunostaining in each blastomere using ImageJ.
Microinjection of antisense MOs
Initial concentrations of 0.05–0.6 mm of 3′-carboxyfluorescein-labeled morpholino were injected into mouse zygotes based on previous findings (76) and it was determined that 0.3 mm of the standard control was the maximum concentration that would allow normal rates of blastocyst development. Therefore, a similar concentration of Msk2 morpholino with the sequence 5′-CCTCGTCCTCATCCTCGTCTCCCAT-3′, which was designed to target the translation start site of Msk2 and labeled with 3′-carboxyfluorescein for visualization, was injected into the cytoplasm of each embryo using a CellTram vario (Eppendorf, Hauppauge, New York, USA), electronic microinjector (Narashige IM300, Japan) and Transferman NK 2 Micromanipulator (Eppendorf). Non-injected, water-injected and the injection of a 3′-carboxyfluorescein-labeled standard control MO that does not target any known sequence in the mouse genome or transcriptome served as controls in each experiment to assess enzyme-MO-mediated knockdown. Each MO was prepared in water and incubated at 65°C for 10 min prior to injection.
The data are represented as the average ± the standard deviation and analyzed for statistical significance (P < 0.05) using the one-way ANOVA with the Bonferonni correction. All experiments were repeated at least three times with similar results.
S.L.C. designed and performed experiments, analyzed data, wrote and edited the manuscript. S.L.M. performed experiments, analyzed data and wrote and edited the manuscript. N.L.B., M.V.R., D.E.L. and B.B. performed experiments and assisted in the writing and editing of the manuscript. C.J.D. and L.M.W. interpreted results and assisted in the writing and editing of the manuscript. R.A.R.P. assisted in the design of experiments, interpreted results and assisted in the writing and editing of the manuscript.
This work was supported, in part, by the California Institute for Regenerative Medicine (CIRM; RB3-02209 to R.A.R.P.) and a grant from the March of Dimes (6-FY10-351 to R.A.R.P.). Funding to pay the Open Access Publications Charges was provided by private funds.
The authors gratefully acknowledge Tasha Kalista, H. Austinn Freeman, Tracy Lindsay and Dharti Trivedi who managed the Stanford RENEW bank. We also thank Drs Yimin Shu, Kevin Loewke, Jinnuo Han, Connie Wong and Andrew Olson as well as Akshi Goyal, Jens Durruthy-Durruthy, Bahareh Haddad Derafshi and Ha Nam Nguyen for technical assistance with these studies.
Conflict of Interest statement. None declared.