Metabolic remodeling during murine early embryo development

During early mammalian embryogenesis, dynamic changes in cell growth and proliferation are tightly linked to the underlying genetic and metabolic regulation. However, our understanding of metabolic reprogramming and its impact on epigenetic regulation in early embryo development remains elusive 1 . Here, we pro�led metabolomes of embryos from the 2-cell and blastocyst stages, and their in vitro counterpart 2-cell like cells and ES cells, and reconstructed their metabolic landscape through the transition from totipotency to pluripotency. Our integrated metabolomics and transcriptomics analysis showed that 2-cell embryos favor methionine, polyamine and glutathione metabolism and stay in a more reductive state, whereas blastocyst embryos have higher mitochondrial TCA cycle metabolites and are in a more oxidative state. Moreover, we identify a reciprocal relationship between α-ketoglutarate (α-KG) and the competitive inhibitor of α-KG-dependent dioxygenases L-2-hydroxyglutarate (2-HG) 2 , namely, higher L-2-HG in the 2-cell embryos inherited from oocytes and 1-cell zygotes, and higher α-KG in the blastocyst. Supplementing 2-HG or knocking down L2hgdh, a gene encoding the 2-HG consuming enzyme L-2-hydroxyglutarate dehydrogenase 3 impeded erasure of global histone methylation markers 4–6 . Together, our data demonstrate dynamic and interconnected metabolic, transcriptional and epigenetic network remodeling during murine early embryo development.


Introduction
Metabolic reprogramming and its regulation during mammalian early embryo development are hardwired into the complex program of development 7 . Limitations in obtaining su cient sample amounts and availability of reliable and scalable methods have hindered a deeper understanding of dynamic changes of metabolites, metabolic pathways and their upstream regulating factors during embryogenesis. Classical analysis on whole embryos' physiology has focused on nutrition uptake, respiratory activity, and by-product secretion 7 . For instance, embryos utilize pyruvate earlier and later transition to glucose, and the knowledge led to the development of optimal media conditions currently used in in vitro fertilization clinical practice [8][9][10][11][12][13] . Moreover, blastocysts exhibit a respiratory burst during development 14,15 , but the underlying metabolic remodeling is still elusive. Recent omics analysis using low-input or single cell RNAseq and chromatin analysis has provided new insights into molecular regulatory circuitry governing early embryo development 6,16−20 . But metabolic genes are often labelled as "housekeeping" genes and metabolic alterations have been broadly neglected. Thus, a comprehensive picture of metabolic gene expression, their epigenetic status, and metabolite dynamics in early murine embryo development has yet to be uncovered.
Our limited knowledge of the metabolism molecular wiring in embryos is primarily built on studies with embryonic stem cells (ES cells), and their various pluripotent states representing different embryo stage counterparts [21][22][23] , and studies with these cultured cells have provided key insights of their catabolic and anabolic metabolism 24,25,26 , and how metabolic networks are integrated into the well-established pluripotency networks such as Esrrb, Stat3, Lin28 and Zic3 in energy metabolism [27][28][29] , or c-Myc and mTor in anabolic metabolism 30,31 . However, there is a great need to con rm the physiological importance of these ndings from cultured ES cells in the in vivo counterparts of embryo systems.
Epigenetic remodeling, especially histone methylation, is highly dynamic during pre-implantation embryogenesis 4-6, 19,20,32,33 . How metabolites are involved in this process is largely unknown. Our ndings for the rst time provide insights into the role of reciprocal changes of a pair of competitive metabolites α-KG and 2-HG in the dynamic histone methylation erasure during early embryogenesis.

Results
Metabolomics pro ling revealed an elevated mitochondrial TCA cycle metabolic program during the 2-cell to blastocyst stage transition In order to understand the dynamic metabolic remodeling during pre-implantation embryo development, we employed mass spectrometry-based metabolomics to directly measure metabolite abundance in embryos. Due to di culties in obtaining a large number of embryos and the limited detection range of current metabolomics technologies, we rst need to optimize a targeted metabolomics approach to detecting metabolites using a small number of cells. We titrated a range of mouse ES cells from 2.5k to 80k cells (Extended Data Fig. 1a), or a range of zygotes from 15 to 240 embryos (Extended Data Fig. 1b), and we were able to achieve strong correlations between the number of cells/embryos and the mass spectrometry signal for most of the metabolites even at lower input ranges of the gradients (Extended Data Fig. 1c-e), indicating good quanti cation of those metabolite with this approach. Then we collected 100 2-cell stage embryos and 100 blastocyst stage embryos, which represent the totipotent state when zygotic genome activation takes place and the pluripotent state when ES cells can be derived 34 , and applied the targeted metabolomics method established above, each with three biological replicates (Fig. 1a). PCA analysis on targeted metabolite levels showed that 2-cell embryos were readily separated from blastocysts (Fig. 1b). Top differential metabolites include citrate, α-ketoglutarate (α-KG), succinate and glutamine and malate that are higher in the blastocyst stage, and 2-hydroxyglutarate(2-HG), Sadenosyl-methionine, GSSG, GSH, and spermidine that are higher in the 2-cell embryos ( Fig. 1c and Extended Data Table 1). Notably, almost all the TCA cycle intermediate metabolites were more abundant in blastocyst embryos including citrate, succinate and α-KG (p < 0.05), but the abundance of the competitive inhibitor of α-KG-dependent dioxygenases 2 , the 2-HG, was higher in 2-cell embryos (p < 0.001, Fig. 1d, e). Indeed, Metabolite Set Enrichment Analysis demonstrated "Citrate Acid Cycle" was among the top enriched metabolite sets in blastocyst embryos, whereas "Methionine Metabolism", "Spermidine and Spermine Biosynthesis" and "Nicotinate and Nicotinamide Metabolism" were among the top enriched metabolite sets in 2-cell embryos, respectively (Fig. 1f, g). Detailed analysis of the above metabolic pathways including the detected metabolites and the corresponding metabolic genes also revealed the dynamic metabolic signatures of the TCA cycle and purine metabolism pathways in the blastocyst embryos (Extended Data Fig. 2, 3), and one carbon metabolism as well as polyamine/glutathione/nicotinamide pathways related to the redox state in 2-cell embryos (Extended Data Fig. 4). For instance, higher level of the TCA cycle pathway metabolites in blastocysts is consistent with higher expression of the TCA cycle genes Aco1, Idh2, Sucla2, Sdha, Fh1 and Mdh2 ( Figure S2), and higher level of GSH, GSSG, GSH/GSSH, spermidine and nicotinamide is associated with higher expression of the redox-related genes such as G6pdx (Extended Data Fig. 4).
To further analyze the metabolomics data in a systemic way, metabolites were mapped to a metabolic network based on our previous published method in genome-scale metabolic network modeling 26 (Extended Data Fig. 5a), and differential genetic deletion sensitivity was predicted (Extended Data Fig. 5b). Consistently, 2-cell embryos are more sensitive to deletions of genes in redox-related "Glutathione metabolism", while blastocyst embryos are more sensitive to deletion of genes in "Oxidative phosphorylation" (Extended Data Table 2). In summary, our embryo metabolomics analysis revealed unique metabolic characteristics of 2-cell and blastocyst embryos.
It has been reported that the in vitro cultured ES cells have a spontaneously emerged 2-cell like cell (2CLC) population that resemble 2-cell embryos in their transcriptional signature 21 . To examine whether the metabolomes between embryos in vivo and their cultured counterparts in vitro are also similar, we applied the same metabolomics approach to ES cells and 2CLCs. After transduced with a 2C reporter, 2C::tdTomato, the 2CLCs can be isolated from ES cell culture 21 . The same ES cell culture was also transduced with a Nanog::GFP reporter to indicate the pluripotent state ES cells (Extended Data Fig. 1f). We thus sorted 10,000 tdTomato positive 2CLCs or GFP positive ES cells each with three biological replicates for metabolomics analysis. In general, PCA analysis of common metabolites revealed that cultured cells clustered more closely to each other (Extended Data Fig. 5c), indicating that the in vitro culturing condition has a strong impact on cellular metabolome. To lent further support, we performed metabolomics with another 2CLC system, a 2C gene Dux inducible ES cell line 35 , to enrich the tdTomato positive 2CLCs. The ratio of 2CLCs reached to 27.2% upon Dux induction (Extended Data Fig. 5d,e), and the 2C genes were strongly activated (Extended Data Fig. 5f). We observed the similar patterns of the PCA analysis as the sorted 2CLC metabolomics from wild-type cells: globally, 2-cell and BC embryos clustered away from 2CLC and ES cells as they represented in vivo and in vitro samples (Extended Data Fig. 5g). Further comparing these 2CLCs and ES cells in the heatmap showed GSH, GSSG, GSH/GSSG, and spermidine are higher in 2CLC (Extended Data Fig. 5h-j), consistent with the analysis from embryos (Extended Data Fig. 1g and 4d,f,g). For better visualization, we overlapped metabolites higher in 2-cell embryos (compared to BC) and higher in 2CLC (compared to ES) (Fig. 1h,i), and they were speci cally enriched in spermidine, methionine and glutathione metabolism, etc (Fig. 1j,k).Together, our embryo and cultured cell metabolomic analyses demonstrate that the embryo metabolome is distinguishable from that of their in vitro counterpart cells. Selective metabolic signatures show consistent trends, such as a more reduced metabolism in 2-cell embryos and 2CLCs, and a more oxidative state in blastocyst embryos and ES cells.
Common and unique metabolic signatures revealed by integrated transcriptional and metabolic analysis RNA-seq and ChIP-seq analyses of early embryos have shed light on the dynamic genetic and epigenetic programs of these stages, including genes in metabolism. To further examine embryo metabolic features from a genetic regulation perspective, we re-analyzed publicly available bulk and scRNA-seq data in mouse embryos of different stages 20,36 , and identi ed "Energy Metabolism" and "Translation" as the two most drastically and consistently altered gene categories in early embryo development ( Fig. 2a and Extended Data Fig. 6a). We then examined the ~ 3000 metabolism-associated genes including metabolic enzymes or transporters 37 , and observed contrasting stage-speci c gene expression patterns, and these metabolic genes were classi ed into 6 stage-speci c clusters by k-means clustering for downstream analysis (Fig. 2b). Pathway enrichment analysis of the bulk RNA-seq data revealed that oxidative phosphorylation and citrate cycle (TCA cycle)-related genes were enriched in the blastocyst stage (equivalent to the ICM or Inner Cell Mass stage) gene cluster, consistent with the embryo metabolomics data, and phosphatidylinositol signaling-related genes were enriched in the MII-Oocyte and 2-cell stage clusters (Fig. 2c). Similarly, scRNA-seq data showed robust enrichment patterns for OxPhos and TCA cycle in the blastocyst stage, and phosphatidylinositol signaling in the zygote and 2-cell stage embryos (Extended Data Fig. 6b-c), together demonstrating that dynamic transcriptome with metabolic alterations are associated with early embryo development stages. When we examined the 2CLC and ES cell transcriptomes 21 , we also found metabolic genes globally were not clustered with an embryo stage (Fig. 2b). However, GSEA analysis on selected gene sets such as"TCA cycle" and "Oxidative phosphorylation" showed higher expression in ES cells than 2CLCs (Extended Data Fig. 6d), consistent with the gene expression features in embryos.
To further validate the signature of "TCA cycle" and "OxPhos" (Extended Data Fig. 6e), we tested the ratio of NADH/NAD + during embryo development with a SoNar biosensor system by injecting in vitro transcribed SoNar mRNA into zygote embryos 38 . The ratio of NADH/NAD + from 2-cells to the blastocysts slightly increased, suggesting an increased mitochondrial TCA cycle and production of NADH (Extended Data Fig. 6f). This is consistent with the metabolomic analysis indicating that the mitochondrial TCA cycle oxidative metabolism is elevated in the blastocyst embryo (Fig. 1f). To validate a strong signature of "Translation" (Fig. 2a and Extended Data Fig. 6a) revealed by an increase of ribosome gene expression (Extended Data Fig. 6g), we determined nascent protein synthesis rate by OPpuromycin staining and found a marked increase from zygotes to morula/blastocyst embryos (Extended Data Fig. 6h), indicating an increased anabolic metabolism program as the embryos develop, which is critical to support the increased proliferation of embryonic cells. Interestingly, ribosome gene expression and translation activity peaked at 8-cell to morula stage, and slightly dropped at the blastocyst or ICM stage (Extended Data Fig. 6g), with a concomitant peak in OxPhos gene expression (Extended Data Fig. 6e), suggesting a unique state of slightly braked anabolic program, likely for embryos to be primed for the subsequent implantation process.
Materials from early embryos can be maternally inherited from oocytes. To distinguish the contribution of maternally-inherited versus zygotically transcribed metabolic signature, we also examined the chromatin status of the ~ 3000 metabolism-associated genes using previously published ATAC-seq 17 . ICM stage showed highest chromatin accessibility of ICM-speci c metabolic genes (or Cluster 5 genes in Fig. 2b) determined by ATAC-seq peaks at transcription start sites (TSS), suggesting these genes are zygotically transcribed at the blastocyst stage (Extended Data Fig. 6i). In contrast, 2-cell stage-speci c metabolic genes (or Cluster 2 genes in Fig. 2b) did not show signi cantly higher signals at TSS than ICM (Extended Data Fig. 6i), suggesting that their presence may not re ect active transcription, as most of genes at this stage are likely maternally inherited 34 . Importantly, our metabolomics analysis also identi ed 2-cell stagespeci c metabolic pathways that were not revealed by gene expression analysis such as methionine metabolism (Fig. 1g, 2c). Comparing oocytes with 2C embryos showed that oocytes are also enriched in metabolites in this pathway such as methionine (Extended Data Fig. 1h), suggesting that certain metabolites from 2-cell stage embryos may be maternally inherited from oocytes. Together, our transcriptional analysis showed common and unique signatures compared with metabolomics analysis, demonstrating the importance of the integrated analysis to obtain a comprehensive understanding of embryo metabolism.

The dynamic metabolic network is integrated into the developmental TF network in early embryogenesis
The embryo development process is tightly governed by transcription regulation from developmental transcription factors (TF). We performed a TF binding motif enrichment analysis (Extended Data Fig. 6j) across embryo stage-speci c clusters of metabolic genes (Fig. 2b). This analysis revealed that TFs like Esrrb, Klf, Myc and Nr5a2 are enriched in 8-cell/morula or ICM-speci c metabolic genes' upstream regulators (Fig. 2d, Extended Data Fig. 6k). Next, co-expression pro les derived from RNA-seq and ChIPseq binding information were integrated to build a high con dence TF-metabolic gene regulatory network (Extended Data Fig. 6l) for each embryo stage-speci c gene cluster (Fig. 2b). Increasingly complex regulatory networks were established from 2-cell/4-cell stages to 8-cell/morula and ICM stages (Extended Data Fig. 6l). Particularly, many mitochondrial TCA cycle and OxPhos-related genes such as Ndufa13, Ndufs2, Ndufc1, Ndufa1, Sdhc, Atp4a and Dlat were among the ICM TF targets. These regulatory networks are consistent with the transcriptome and metabolomics results implying the TCA cycle and OxPhos metabolism are highly activated during the blastocyst/ICM stage, but, more importantly, they demonstrate that the metabolic program is regulated by the classical developmental transcription factors. We also validated these ndings with moues ES cell ChIP-seq data of key developmental transcription factors 39 , and found Esrrb, Myc and Klf genes were largely involved in the regulation of the TCA cycle and OxPhos genes with many shared target genes (Fig. 2e), in contrast to Sox2, Stat3 and Nanog which appeared to have very little or no in uence in the TCA cycle and OxPhos metabolism gene regulation (Fig. 2e). Together, these results demonstrate that metabolic program are integrated into a speci c set of upstream developmental TFs' regulatory network.
A reciprocal relationship between α-KG and 2-HG and a reduction of L-2-HG facilitating global erasure of histone methylation during early embryogenesis Next, to understand whether metabolites are directly involved in gene regulation and embryo development, we chose α-KG and 2-HG which have established roles in epigenetic regulations and cell fate decision 2,40,41 . α-KG is required for the enzyme activity of many dioxygenases such as histone demethylases, whereas D-2-HG is an oncometabolite produced by mutant IDH1/2, and it acts as an antagonist of α-KG to inhibit histone demethylation 2 . There are two enantiomers of 2-HG, D-2-HG and L-2-HG, and the L type 2-HG was recently found to be produced under certain physiological condition 42,43 .
The detection of high 2-HG in early embryos when global epigenetic reprogramming take place made us to further explore the subtype and absolute concentration of this metabolite in MII oocytes, 1-cell zygotes and 2-cell embryos. To our surprise, we found that L-2-HG, but not D-2-HG ( Fig. 3a and Extended Data Fig. 7a), had the highest abundance in MII oocytes, zygotes and 2-cell embryos, and it steadily decreased during the embryo development (Fig. 3b,c). The absolute concentration of L-2-HG we determined is in the millimole range, comparable to some cases of D-2-HG reported in IDH-mutation cancer cells or higher than L-2-HG reported in certain physiological conditions (Extended Data Fig. 7b,c) [44][45][46][47][48][49] . On the contrary, the absolute concentration of α-KG in blastocysts is more than 10-fold higher than 2-cell embryos (Fig. 3e). Even though α-KG is also slightly decreased from MII oocytes to 2-cell embryos, the ratio of L-2-HG/α-KG is markedly decreased throughout the early embryo development, from ~ 6-fold to less than 1fold (Fig. 3f,g). A decreasing L-2-HG concentration and decreasing L-2-HG/α-KG ratio after fertilization suggests that it might allow erasure of various histone methylation during this stage 4,6 . To test this hypothesis, we treated embryos with a permeable L-2-HG octyl-L-2-HG 50 during their in vitro development. Global erasure of H3K4me3 and H3K9me3 were indeed impeded or aberrant hyper-methylation was observed in the zygote to 4-cell stage embryos when supplementing ectopic L-2-HG from the zygote stage ( Fig. 3h-k), with a non-toxic L-2-HG concentration that did not affect viability of 2-cell or 4-cell embryos (Extended Data Fig. 7d,e).
We also found delay of development process and embryo morphological abnormalities (Fig. 4a,b), such as the formed blastocyst embryos tended to collapse or not be able to hatch normally upon L-2-HG treatment (Fig. 4c,d), and the cavity area or averaged cell number per blastocyst decreased (Fig. 4e,f). Moreover, this effect at the blastocyst stage was more obvious when the treatment was restricted to earlier stages (Fig. 4g,h). RNA-seq analysis with L-2-HG-treated embryos that reached to early blastocyst stage showed a reduction in the TCA cycle and pluripotency genes (Fig. 4i). On the contrary, treatment with a permeable and embryo tolerable concentration of dimethyl-α-KG 41 showed increased expression in the TCA cycle and pluripotent genes (Fig. 4j). Real-time PCR also validated increased Nanog expression upon α-KG treatment and decreased Nanog expression upon 2-HG treatment in blastocyst embryos (Fig. 4k,l). α-KG also rescued the L-2-HG effects on blastocyst formation rate, but could not fully rescue the morphological abnormalities (Fig. 4m,n), suggesting these two metabolites have certain opposing effects, but may also have other non-antagonizing effects 51 .
We next explored where the L-2-HG is from and how it is cleared off. It has been reported that malate dehydrogenase (MDH) or lactate dehydrogenase (LDH) can produce L-2-HG under certain physiological conditions 42,43,52 , and L-2-hydroxyglutarate dehydrogenase (L2HGDH) can consume L-2-HG 3 . We thus determined the expression levels of these enzymes with RNA-seq (Extended Data Fig. 8a), real-time PCR (Fig. 5a and Extended Data Fig. 8b) and western blotting analysis (Extended Data Fig. 8c). For both the mRNA and protein levels, Ldhb/LDHB is highly expressed in the oocytes, 1-cell and 2-cell embryos, with a gradual decrease during embryo development, and reaches to the lowest level at the blastocyst (ICM) stage. The other Ldh genes and Mdh2 do not show high expression of their mRNA at or before the 2-cell stage of the early embryo development. Simply knocking down a maternal factor Ldhb from zygote embryos may not interfere with L-2-HG abundance as it may be produced during oocyte development and inherited maternally from oocytes. Interestingly, mRNA level of L2hgdh increased from 2-cell, and reaches to the highest level at 4-cell, and decreased afterwards (Fig. 5a). Its protein expression is also the highest at the 2-cell and 4-cell stages and decreases afterwards, suggesting it might have a role in consuming L-2-HG during these stages. When knocking down L2hgdh by injecting siRNA in the zygote (Fig. 5b), we found an increase of L-2-HG in the 4-cell embryos (Fig. 5c). We also found H3K4me3 hyper-methylation in the 4-cell embryos, and in the presence of L-2-HG, this effect was further ampli ed (Fig. 5d-f), suggesting accumulation of L-2-HG without a consumption enzyme deteriorated the histone methylation erasure process. Together, these data demonstrate that a reduction of L-2-HG after fertilization and during preimplantation embryo development is required for the global erasure of histone methylation, and that accumulation of L-2-HG impedes this epigenetic remodeling process and in uences embryo development.

Discussion
Mammalian embryogenesis is a complex process with coordinated regulation at multiple molecular levels. Using pooled mouse early embryos and optimized mass spectrometry based ultra-sensitive targeted metabolomics for a small number of cells we previously developed 53 , here we provided a comprehensive and quantitative metabolomics pro ling of key stages in murine early embryo development. By integrating this data with transcriptome, chromatin states, and single cell pro ling data on ~ 3000 metabolism-associated genes, and by performing metabolic network analysis, we delineated the rst metabolic landscape of murine pre-implantation embryo development. Notably, snapshots of metabolomics cannot directly indicate metabolic pathway activity, and isotope tracing analyses on embryos merit further investigations.
Our comprehensive metabolomics and genomics analysis unraveled a redox state transition during preimplantation embryogenesis, which is consistent with previous functional analysis on redox or respiration 14,54,55 . Transition from a reduced state to an oxidative state is consistent with the high L-2-HG to high α-KG transition. We found this decrease of L-2-HG is required for histone erasure during early embryogenesis, establishing a role of metabolic regulation on embryo epigenome remodeling.
Interestingly, the phenotypes of H3K4 hypermethylation and embryo collapse defects revealed by L-2-HG supplementation is reminiscent of embryos with H3K4me3 demethylase genes Kdm5a/b knocked down, suggesting these histone modi cations may play a role in the L-2-HG effect 6 . However, the effects of decreased L-2-HG in facilitating demethylation might be broader than one class of histone demethylases.
For instance, we also observed increased H3K9me3 and DNA methylation upon L-2-HG supplementation, and because the global active demethylation of H3K4me3 at the early embryogenesis is well characterized 6,19,56 , we focused on H3K4me3 here. How L-2-HG's impact on various methylation marks collectively in uences embryo development, and how the downstream speci city on different modi cations and genes merit further investigation. Importantly, metabolism in early embryos have also been linked to histone acetylation and zygotic genome activation, implicating the broad range of interaction between metabolic and epigenetic regulation during early embryo development 57 .
The 2-cell and blastocyst metabolic signatures also indicate that embryos transition from a quiescent state to metabolically active state with high anabolic metabolism and mitochondrial TCA cycle to provide intermediate metabolites 58 . The OxPhos and TCA cycle peak at blastocyst embryo stage, which also produce high level of α-KG that can in turn facilitates pluripotent gene expression and developmental process 41 . Along with the activation of the TCA and OxPhos metabolism genes, many pluripotency/developmental transcriptional factors are induced. In contrast to Nanog and Sox2, transcription factors such as Esrrb, c-Myc and Klf family members share many binding targets in the TCA and OxPhos genes, and they might coordinately contribute to activation of these genes. This neglected metabolic regulation adds to the developmental TFs' well-established roles in regulating pluripotency networks 59,60 , and thus these TFs can be de ned as a new set of developmental regulators with dual roles in establishing both pluripotency and metabolic networks.      The absolute level of L-/D-2-HG in MII-oocytes, zygote and 2-cell embryos. Data is presented as the mean ± SEM of three biological replicates. **p<0.002, ***p<0.001, (n=3, unpaired t-test). (c) The absolute level of L-2-HG and D-2-HG in 2-cell and blastocyst embryos; Data is presented as the mean ± SEM of three biological replicates. *** p<0.001, (n=3, unpaired t-test). (d) The absolute level of α-KG in MII-oocytes, zygote and 2-cell embryos. Data is presented as the mean ± SEM of three biological replicates. ***p<0.001, (n=3, unpaired t-test). (e) The absolute level of α-KG in 2-cell and blastocyst embryos. Data is presented as the mean ± SEM of three biological replicates. ***p<0.001, (n=3, unpaired t-test). (f) The ratio of 2-HG/α-KG in MII-oocytes, zygote and 2-cell embryos. Data is presented as the mean ± SEM, * p<0.033, (n=3, unpaired t-test). (g) The ratio of 2-HG/α-KG in 2-cell and blastocyst embryos. Data is presented as the mean ± SEM, ***p<0.001, (n=3, unpaired t-test).