Pre-implantation embryo metabolism identified by PEMA reveals endogenous lactate insufficiency contributes to pre-implantation development arrest

Abstract

Metabolism is a major regulator of pre-implantation embryonic development, but pre-implantation development arrest (PIDA) due to abnormal metabolisms remains ambiguous. By estimating a weight for each metabolic reaction using the translatome data, we developed a computational tool to perform pre-implantation embryos metabolic analysis (PEMA), and uncovered that the embryos exist in high activities of lactate dehydrogenase and lactate synthesis when major zygotic genome activation (ZGA) occurs in humans and mice. The insufficiency of endogenous lactate was predicted by PEMA in human 8-cell PIDA and corrected human tripronuclear (ch3PN) embryos, which was coordinated with failed H3K18lac and major ZGA. In transcriptomes, the human PIDA embryos resembled endogenous lactate-deprived mouse embryos. Lac-CoA addition could promote H3K18lac, major ZGA, and ch3PN pre-implantation development. Collectively, our study decodes the metabolic abnormalities in human PIDA embryos. The experimental validation suggests that PEMA offers an opportunity to study metabolic heterogeneity in early embryos.

Graphical abstract

1. Introduction

Mammalian pre-implantation development, from zygotes to blastocysts, undergoes a series of crucial biological events, including zygotic genome activation (ZGA), which is the first transcription event in life [1,2]. In mice, the initial ZGA occurs between the S phase of the zygote and G1 of the early 2-cell embryo, and is designated as minor ZGA to discriminate it from the burst of transcription that occurs during the mid-to-late 2-cell stage, which is designated as major ZGA [3,4]. In humans, ZGA marks the major wave of transcription around the 4- to 8-cell stages [3]. ZGA is essential for the ensuing subsequent development. Aberrant ZGA leads to substantial pre-implantation development arrest (PIDA) [[5], [6], [7]]. PIDA is a common reason for recurrent failure of assisted reproductive technologies (ART) and leads to female infertility [8,9]. If all human embryos generated by in vitro fertilization (IVF) or intracytoplasmic sperm injection (ICSI) are taken into consideration, about 10% of them are arrested at the cleavage stages [10]. The embryos from patients with genetic mutations in TUBB8 [11], CHK1 [12] or FBXO43 [13] mostly undergo developmental arrest. Recently, by integrative multi-omics analysis, Wang et al. [7] have revealed that AURKA repression is a reason for human embryonic arrest [7]. However, the mechanisms of PIDA due to non-genetic reasons remain ambiguous.

Cellular metabolism underlies all biological activities [[14], [15], [16]]. The significance of cellular metabolism in controlling pre-implantation embryonic development was recognized several decades ago through identification of the conditions that allow early embryos to grow outside of the oviduct [17]. The metabolism of mammalian early embryos has traditionally been studied using radiolabeled substrates in combination with inhibitors, and exogenous substrates at specific stages [18]. Recently, with the advancement in high-performance liquid chromatography coupled with mass spectrometry technology, research has begun to shed light on the metabolomic profiles of early mammalian embryos [5,19,20]. Zhao et al. have established the metabolomic profiling of mouse 2-cell embryos and blastocysts, highlighting the role of reciprocal changes in a pair of competitive metabolites (α-KG and 2-HG) in the dynamic erasure of histone methylation during blastocyst formation in mice [19]. In the most recent work, they have established the dynamic lipid landscapes during mouse and human pre-implantation embryonic development, showing that lipid desaturases are required for in vitro blastocyst development and implantation [20]. We have reported a metabolome profile of pre-implantation embryos of mice from the zygote to blastocyst stages [5]. Our findings have shown that the combination of the prominent metabolic cofactor NAD⁺ and the deacetylase SIRT1 removes zygotic H3K27ac in the late 2-cell stage for the exit of the minor ZGA, which is essential for mouse and human pre-implantation embryonic development [5]. Our study has also shown that the inhibition of lactate production and uptake results in developmental arrest at the 2-cell stage, major ZGA failure, and loss of lactate-derived H3K18lac, which is enriched on the promoter regions of most major ZGA genes and correlates with their expressions [6]. These results highlight that metabolic reprogramming is hardwired into the complex program of early mammalian embryonic development, directly involving alterations in chromatin and DNA states. However, the identification of metabolic networks for pre-implantation embryos is still in its infancy because of the relatively low throughput and sensitivity.

The advent of RNA-seq and single-cell RNA-seq provided an opportunity to systemically interrogate transcriptomic profiles to reveal metabolic reprogramming [21,22]. Those studies often employ statistical methods that score mRNA expression levels of genes in a predefined metabolic pathway (e.g., KEGG) [[23], [24], [25]]. One of the most common methods is flux balance analysis (FBA), which describes the potential flux over the topological structure of a metabolic network, with a set of equations governing the mass balance at a steady state [26]. Based on FBA, Nir Yosef et al. have presented Compass, which maximizes the coherence between single-cell transcriptomics and predicted flux in solution space of FBA [27]. However, the method is ill-suited for pre-implantation embryos, due to a large amount of dormant mRNAs stored in the blastomere [[28], [29], [30]]. Therefore, in the study, by estimating a weight for each metabolic reaction using Ribo-seq data during pre-implantation development, we developed a computational tool, pre-implantation embryos metabolic analysis (PEMA; available at https://github.com/summus-kong/PEMA), to characterize metabolic states of human and mouse pre-implantation embryos comprehensively and precisely. We showed that the embryos exist in high activities of lactate dehydrogenase (LDH) and lactate synthesis state when major ZGA occurs in humans and mice, and human PIDA and ch3PN embryos displayed insufficiency of endogenous lactate, leading to impaired major ZGA and aberrant H3K18lac patterns. Supplement of Lac-CoA could promote H3K18lac established at the promoter regions of human 8-cell embryo signature genes, and improve major ZGA and pre-implantation development of ch3PN embryos.

2. Materials and methods

2.1. Human early embryos

The patients who donated embryos are under 35 years old without a history of genetic diseases or smoking. Pituitary downregulation with gonadotrophin-releasing hormone (GnRH) agonist (Triptorelin Acetate, Ipsen Pharma, France) and controlled ovarian stimulations with recombinant follicle stimulating hormone (rFSH) (Puregon; Organon, Netherlands or Gonal-F, Merck Serono, Switzerland) were performed. Human chorionic gonadotropin (HCG) (Ovidrel, Merck Serono, Italy) was administered when at least three follicles measured > 18 mm. Transvaginal oocyte retrieval was performed 36 h after HCG injection. MII oocytes were obtained and fertilized using IVF or ICSI. Normal fertilized zygotes with two-pronuclear (2PN) were cultured in G-TL (Vitrolife Sweden, Gothenburg, Sweden) medium at 37 °C in 6% CO₂ and 5% O₂. The 8-cell and 16-cell PIDA embryos were collected on Day 4 and Day 5, respectively. The numbers of PIDA embryos used were as follows: 2 8-cell PIDA and 2 16-cell PIDA embryos were used for IF; 18 8-cell PIDA embryos were subjected to single-embryo RNA-seq; 5 8-cell PIDA and 2 16-cell PIDA embryos were used for scCUT&Tag.

Three-pronuclear (3PN) zygotes were collected 16–18 h after fertilization. According to the previously published protocols [31], the removal of the extra male pronucleus of the 3PN zygote was performed, producing the ch3PN embryos. These embryos were then cultured in G-TL (Vitrolife Sweden, Gothenburg, Sweden) medium at 37 °C in 6% CO₂ and 5% O₂. The number of 3PN embryos used was as follows: 2 8-cell ch3PN embryos were used for IF; 2 8-cell ch3PN embryos were used for FiLa analysis; 5 8-cell ch3PN embryos were used for single-embryo RNA-seq; 7 8-cell ch3PN and 6 16-cell ch3PN embryos were used for scCUT&Tag; 53 ch3PN embryos were used for the statistics of embryo development. The study used embryos that donors have been informed not to affect their normal IVF process and have written informed consent.

2.2. Mice

Mice were maintained and handled according to the protocol approved by Laboratory Animal Ethics Committee of Wenzhou Medical University (xmsq2023–0118). Male and female ICR mice (6–8 wk old) were purchased from Charles River (Beijing, China).

2.3. Ethics statement

This study was approved by the Institutional Review Board (IRB) of Chongqing Health Center for Women and Children (2022-RGI-01), China, in accordance with the measures of the People's Republic of China on the Administration of Human Assisted Reproductive Technology and the Helsinki Declaration. The research followed the principles of the Human Embryonic Stem Cell Ethics issued by the MOST and MOH and was regularly reviewed by the Medical Ethics Committee of Chongqing Health Center for Women and Children.

2.4. Embryo collection

Embryos were collected from superovulated female ICR mice treated with 6.5 IU pregnant mare serum gonadotropin (PMSG); after 47–48 h, they were treated with 5 IU human chorionic gonadotropin (hCG) and crossed with male ICR mice. The in vivo 2-cell embryos were collected through flushing the fallopian tubes and uterine horns with M2 medium (M7167; Merck). Embryos in the control group were cultured in modified KSOM medium (95 mM NaCl, 2.5 mM KCl, 0.35 mM KH₂PO₄, 0.20 mM MgSO₄, 25 mM NaHCO₃, 1.71 mM CaCl₂, 0.01 mM EDTA, 0.20 mM glucose, 10 mM lactate, and 0.20 mM pyruvate) devoid of amino acids and BSA. Medium osmolarity was maintained using 0.1% polyvinyl alcohol under mineral oil at 37 °C in a 5% CO₂ incubator.

2.5. Treatment of embryos with inhibitors

We prepared solutions of GNE-140 (MedChemExpress, Cat HY-100742A) or GSK2837808A (MedChemExpress, Cat HY-100,681) in dimethyl sulfoxide (DMSO). NMN (Sigma, Cat N3501) solutions were prepared in water. Solutions were diluted to the desired final concentration in medium (GNE-140, 5 µM; GSK2837808A, 1 µM; NMN, 50 µM). Embryos were cultured in modified KSOM medium containing the indicated doses of inhibitors.

2.6. Microinjection

For FiLa analysis, pAAV-CMV-MCS-FiLa, and pLVX-Nuc-FiLa [32] were linearized and in vitro transcribed, and 200 ng/µL mRNA was microinjected into human and mouse early zygotes. In Lacetyl-CoA rescued experiment, Ch3PN zygotes were injected with 1 mM Lacetyl-CoA (Shanghai Nafu Biotechnology Co., LTD., 4625–32–5) through microinjection protocol, and the control group was injected with RNase-free Water (9012; Takara, Kusatsu, Japan). Microinjection was performed using a FemtoJet 4i Microinjector (Eppendorf, Hamburg, Germany) and ELIPSE Ti Micromanipulators (Nikon, Tokyo, Japan). For injection, a glass capillary Femtotip (Eppendorf) was loaded with the mixtures using a Microloader (Eppendorf) and injected into cytoplasm in a drop of M2 medium. The injection volume was approximately 2–5 pL. The injection pressure was 250 hPa, compensation pressure was 60 hPa, and injection time was 0.7 s. Immediately after microinjection, embryos were cultured at 37 °C in 5% CO₂.

2.7. Immunofluorescence staining

After removing the zona pellucida using acidic operating fluid, mouse and human embryos were fixed in 4% PFA (FB002; Invitrogen) for 40 min at room temperature, followed by permeabilization in 1% Triton X-100 (93,443, 100 mL; Sigma) for 20 min at room temperature. Embryos were blocked in blocking solution consisting of 1% bovine serum albumin (BSA) in phosphate-buffered saline (PBS) for 1 h at room temperature after three washes in washing solution (0.1% Tween-20, 0.01% Triton X-100 in PBS). Embryos were incubated with the indicated antibodies against H3K18lac (PTM-1406RM, PTM BIO), LDH (Abcam, ab52488) overnight at 4 °C. The next day, the embryos were washed in washing solution and incubated with the secondary antibodies (A10040; Invitrogen) for 1 h at room temperature. After staining with Hoechst, the embryos were washed in washing solution and imaged using an inverted confocal microscope (TCS SP8; Leica, Wetzlar, Germany); image analysis was performed using LAS X software (Leica).

2.8. Single-embryo RNA-seq

RNA-seq libraries were prepared as described previously [5]. In brief, Embryos were washed three times in 0.5% BSA-PBS solution. cDNA was amplified using the Phusion Hot Start II High-Fidelity PCR Master Mix (F-565S; Thermo Fisher). Library preparation was performed using the TruePrep DNA Library Prep Kit (TD501; Vazyme Biotech, Nanjing, China) according to the manufacturer's instructions. Libraries were sequenced using the NovaSeq 6000 (Illumina, San Diego, CA) according to the manufacturer's instructions.

2.9. RNA-seq data quality control, processing, and analysis

For paired-end RNA-seq analysis of embryos, FastQC was performed for Illumina reads. The fastp software was used to discard low-quality reads, trim adaptor sequences, and eliminate poor-quality bases. Clean reads were aligned to the mouse reference genome (mm39) or the human reference genome (hg38) using HISAT2 v2.2.1. The gene-level quantification approach was used to aggregate raw counts of mapped reads using the featureCounts tool. Gene expression was quantified by calculating the normalized fragments per kilobase of transcript per million mapped reads (FPKM) value. Next, the R package DESeq2 was used to identify differentially gene expressed genes between different groups in mice and humans. GSEA analysis was performed using R package clusterProfiler, which provided normalized enrichment scores and nominal p-values. Major ZGA gene set used for this analysis collected from a previously published study [33]. KEGG functional annotation was performed by the KOBAS online tool (http://bioinfo.org/kobas). For human single-cell RNA sequencing data, transcriptional activity of each embryo was defined as mean expression levels of all genes in each single embryo, and karyotype was estimated using R package scploid (https://github.com/MarioniLab/Aneuploidy2017).

2.10. PEMA framework

We developed a comprehensive statistical framework named PEMA to quantify and characterize the metabolic activity of embryos during mammalian pre-implantation development. The PEMA framework was inspired by flux balance analysis, with the input of normalized expression level of genes from RNA-seq and Ribo-seq data. The modeling procedure is delineated through the four sequential steps, with an overarching framework overview depicted in Fig. 1A.

Fig. 1.

Metabolic profiles of human and mouse pre-implantation embryos predicted by PEMA. (A) Computational framework of PEMA for the prediction of metabolic activity during pre-implantation development in mice and humans. (B) Heatmap showing the Pearson correlation of Compass or PEMA with our metabolic profiles in mouse pre-implantation embryos. Purple and grey squares representing a higher or lower correlation than Compass respectively. (C and D) Violin plot and heatmap illustrating the dynamics of predicted reaction scores of glycolysis during mouse pre-implantation development. **P < 0.01; ***P < 0.001. (E) Dynamics of metabolic activities predicted by PEMA across human pre-implantation embryos. ***P < 0.001. (F) Bubble diagram showing the reaction scores of glycolysis in human pre-implantation embryos. The colors and sizes of dots indicate the normalized reaction scores. (G) Pearson correlation of lactate synthesis and LDH activity with gene expressions in mice and humans. Purple bars represent major ZGA genes in mice or 8-cell signature genes in humans, and yellow bars represent other coding genes. Error bars are SEM. ***P < 0.001. Differences between means were calculated using Student's t -test.

Step1, input data standardization. Normalized mRNA gene expression FPKM from RNA-seq of mammalian early embryos was prepared as a gene expression matrix {G}_{g × s}, which columns (s) represent the embryonic development stages and rows (g) indicate number of genes in a given transcriptome dataset. For Ribo-seq data, Ribo-lite signals were tidied as matrix {T} _{g × s}, columns (g) are the same RNA libraries as in the gene expression, rows (s) were in line with mRNA gene expression matrix.

Then, we estimated the embryonic metabolic activity based on genome-scale metabolic model (GSMM) [34]. Highly curated metabolic reactions were extracted from the Recon2 database, which includes reaction stoichiometry, biochemical constraints and gene-enzyme-reaction associations. The metabolic flux is the instantaneous rate at which a chemical reaction happens, and it is calculated in units of mol X gDW⁻¹ X hr⁻¹, where gDW represents gram dry weight [26,27]. The concentrations of all metabolites are represented by the vector x, and vector v indicate the metabolic fluxes. Stoichiometric matrix S is a sparse matrix indicating a set of possible metabolic reactions in the metabolic network. Then:

$$S·v=\frac{dx}{dv}$$

The system of mass balance equations at metabolic steady state according to FBA [26]:

$$S·v=0$$

And, the optimal points were identified within a constrained space. Using FBA algorithm, mRNA gene expression matrix {G}_{g × s} was transformed into a matrix {U}_{r × s} of scores where columns (s) are the same RNA libraries as in the gene expression, rows (r) indicate the metabolic reactions.

Subsequently, metabolism-associated genes were obtained from Recon2 database and mapped to matrix {T} _{g × s} of Ribo-seq data. The mean expression level of metabolism-associated enzymes from Ribo-seq data was calculated and normalized per reaction and then was converted into {C}_{r × s} of weighted values where columns and rows are in accordance with matrix U.

Finally, metabolic activity matrix U predicted using RNA-seq data was optimized by weighted value matrix C per developmental stage and obtained the reaction score matrix {R} _{r × s} of metabolism.

RNA-seq and Ribo-seq data of mouse zygote, 2-cell, 4-cell, 8-cell and blastocyst embryos from GSE165782 and GSE209648 was used to develop the PEMA framework. For evaluating the feasibility of PEMA, our previously published metabolomic profiles was used as a benchmark, and the Pearson correlation between predicted metabolic activity of Compass or PEMA and our metabolic profiles was calculated. The correlation between the reaction scores of LDH activity and lactate synthesis predicted by PEMA and each gene respectively was calculated in mouse 2-cell and human 8-cell embryos.

2.11. CUT&Tag

CUT&Tag was performed using the Hyperactive Universal CUT&Tag Assay Kit for Illumina (TD903; Vazyme Biotech, Nanjing, China). Embryos were treated with 10 µl of pre-washed ConA beads. To this, 50 µl of antibody buffer and 0.5 µg of the primary antibody were added. The solution was then left to incubate overnight at 4 °C. After two washes using dig-wash buffer, the embryos were incubated in 100 µl of the same buffer, now containing 1 µl of the secondary antibody, for an hour at room temperature. Following another pair of washes with 800 µl dig-wash buffer, we added 2 µl of pG-Tn5 and 98 µl of dig-300 buffer to the samples and incubated them for an hour at room temperature. This was followed by two more washes with 800 µl of dig-wash buffer. Subsequently, 300 µl of tagmentation buffer was added, and the samples were incubated at 37 °C for an hour. To terminate the reaction, we added 10 µl of 0.5 M EDTA, 3 µl of 10% SDS, and 2.5 µl of 20 mg/ml proteinase K. The samples then underwent phenol-chloroform extraction and ethanol precipitation. PCR amplification of the libraries was carried out under these conditions: an initial step at 72 °C for 3 min, denaturation at 98 °C for 30 s, followed by 20 cycles of 98 °C for 10 s, and annealing at 60 °C for 30 s. This was concluded with an extension step at 72 °C for 1 min before holding the samples at 4 °C. The subsequent clean-up post-PCR involved the addition of 1.5 × volume of DNA Clean Beads (N411; Vazyme Biotech). Libraries were allowed to interact with the beads for 10 min at room temperature. After a gentle rinse with 80% ethanol, the libraries were eluted in 20 µl of elution buffer. Sequencing of all libraries was performed on the Illumina NovaSeq 6000 platform, adhering to the manufacturer's guidelines.

For single-cell CUT&Tag of human embryos, CUT&Tag was performed in well of a microplate. After tagmentation, single human embryos were transferred to 200 µl tubes containing 5% Proteinase K. After embryos lysis, PCR buffer, enzyme and adaptors were directly added to tubes, normal PCR cycle was performed. Post-PCR clean-up was performed by adding a 1.5 × volume of DNA Clean Beads. Libraries were sequenced on the Illumina NovaSeq 6000 platform according to the manufacturer's instructions.

2.12. CUT&Tag data analysis

For Paired-end CUT&Tag reads of H3K18lac, FastQC was performed for Illumina reads. The fastp software was used to discard low-quality reads, trim adaptor sequences, and eliminate poor-quality bases. Clean reads were aligned to the mm39 mouse or hg38 human reference genome downloading from UCSC database using Bowtie2 v2.5.0 software. Non-uniquely mapped reads were removed using Picards. The H3K18lac signal peaks were called using MACS2 v2.2.7.1 software for each sample. Peak annotation was performed using the R package ChIPseeker with the default parameters. For quantitative analysis, we normalized the read counts by computing the RPKM for the BAM file of each sample with the bamCoverage tool in deepTools software. The washU epigenome browser (http://epigenomegateway.wustl.edu/browser/) was used to visualize H3K18lac CUT&Tag data.

2.13. Clustering of pre-implantation embryos

The hierarchical clustering and dimensionality reduction analysis for RNA-seq embryo samples were performed in R by hclust() and Rtsne() function with FPKM values and visualized using ggtree and ggplot2 package. For the CUT&Tag clustering analysis, RPKM of promoters of the ZGA genes (defined as ± 3 kb around the transcription start site) was calculated by deepTools for each bin in each sample. For each sample, log2 normalization was performed and used as input for dimensionality reduction analysis and hierarchical clustering analysis. For the integrated dimensionality reduction analysis of mouse and human sequencing data, mouse gene names were converted to those of human-based genes on the published annotation information to interpret the data.

2.14. Statistical analysis

Statistical analysis was performed using R v4.2.1 software (R Development Core Team, Vienna, Austria). Data are expressed as means ± standard errors of the mean (SEMs). Differences between means were evaluated using Student's t-test. Asterisks indicate significant differences as follows: *P < 0.05, **P < 0.01, and ***P < 0.001.

3. Results

3.1. Comprehensive characterization of pre-implantation embryo metabolism in humans and mice by PEMA

We reasoned that mRNA expressions of enzymes may make a bias when scoring their metabolic activities owing to a huge amount of dormant mRNAs retained in mammalian pre-implantation embryos, and expected that Ribo-seq data might somewhat correct the projected metabolism. Therefore, using the transcriptome and translatome of each stage as input, we developed a computational framework called PEMA (see Methods for details) based on FBA to study the metabolic states throughout mammalian pre-implantation embryonic development (Fig. 1A). Specifically, for each stage using transcriptional profile, we scored each metabolic reaction activity, by generalizing the mRNA expression levels of associated functional genes. The translational levels of metabolism-associated genes from Ribo-seq data showed significantly higher than those of other coding genes (Fig. S1A). Thus, subsequently, we defined a weight for each reaction using Ribo-seq data to optimize each metabolic reaction activity. The PEMA and Compass predictions were performed correlation analysis with our metabolomic profiles during mouse pre-implantation development [5], and PEMA showed a higher correlation with the metabolomic profiles, compared to Compass (Fig. S1B; P < 0.05). In addition, across 29 metabolic pathways, PEMA achieved higher correlation in 22 of the cases (Fig. 1B). These results indicate the feasibility of PEMA on the prediction of embryonic metabolism.

Using PEMA, the metabolic scores of the majority of reactions in four metabolic pathways including glycolysis, TCA cycle, fatty acid oxidation and amino acid metabolism, were examined and exhibited higher activities in blastocysts, consistent with previous reports [18] (Figs. 1C, D and S1C; Table S1). Furthermore, we noticed that the high activity of glycolysis also happens in mouse 2-cell embryo (Fig. S1D). PEMA predicted LDH activity and lactate synthesis scores were high in mouse 2-cell embryos (Fig. 1D). We then applied the framework to multi-omic data of human pre-implantation embryos from GSE197265 [30] to describe the metabolic activities. PEMA, but not Compass, showed the dynamics of global metabolic activity during human pre-implantation development, revealing high metabolic activity in inner cell mass (ICM) of blastocysts (Figs. 1E and S1E). We also noticed the metabolic scores of LDH activity and lactate synthesis were high at the 8-cell stage (Fig. 1F; Table S2). ZGA occurs in mouse 2-cell and human 8-cell embryos [3]. Definitely, we found that the reaction scores of LDH activity and lactate synthesis were highly correlated with the expressions of major ZGA and 8-cell signature genes, respectively in mice and humans (Fig. 1G), suggesting LDH activity and lactate synthesis may contribute to mammalian major ZGA. Taken together, the results suggest that the output of PEMA is a quantitative profile for the metabolic states of pre-implantation embryos.

3.2. LDH inhibition reduces endogenous lactate level in mouse 2-cell embryo

In glycolysis, pyruvate produces endogenous lactate by LDH [18]. The expressions of Ldha and Ldhb were high in mouse 2-cell embryo [6]. Thus, we inhibited LDH activity by Gne140 or GSK2837808A, dual LDHA/B inhibitors, and nicotinamide mononucleotide (NMN) was administered to rescue the loss of NAD⁺ during the in vitro culture at the late zygote stage [5], which resulted in a significant development block to blastocyst (Fig. 2A, B).

Fig. 2.

Endogenous lactate level is reduced in LDH-inhibited embryos in mice. (A and B) Blastocyst rate of embryos in the G&N groups. Data are from three independent experiments. Error bars are SEM. *P < 0.05. Scale bars, 50 μm. (C) RNA-seq analysis of 2-cell embryos from the Control and G&N groups. Yellow and purple dots indicate upregulated and downregulated genes, respectively. (D) KEGG analysis of downregulated genes in the G&N group. (E) The expressions of genes in glycolysis pathway in the control and G&N groups. (F) Representative IF images of LDH in the control, G&N, and GSKA&N groups. Scale bars, 50 μm. (G) Normalized reaction scores in glycolysis predicted by PEMA in the control and G&N groups. Blue line and dots represent the scores of lactate synthesis, red line and dots represent the scores of LDH activity. (H) Fluorescence images of mouse embryos at the 2-cell stage from the control, G&N and GSKA&N groups by FiLa detection. Scale bars, 20 μm. DMSO, mKSOM culture medium plus DMSO; G&N, mKSOM culture medium plus Gne140 and NMN; GSKA&N, mKSOM culture medium plus GSK2837808A and NMN. Differences between means were calculated using two-tailed Student's t -test.

To understand the involvement of endogenous lactate with the development, RNA sequencing (RNA-seq) was performed on 2-cell embryos with addition of Gne140 and NMN (G&N group). Unsupervised hierarchical clustering (UHC) could distinguish between the G&N and control embryos (Fig. S2A). Differential expression analysis showed 760 and 793 significantly up- and down-regulated genes, respectively (Fig. 2C, |Fold Change| > 2 and P < 0.05; Table S3). Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis revealed that the downregulated genes in the G&N group were prominent in transcription and glycolysis pathways (Fig. 2D). We further observed the inhibitors could repress not only the activity of LDH, but also the expressions of Ldha and Ldhb, as confirmed by immunofluorescence (IF) staining (Fig. 2E, F).

Then, we applied PEMA to uncover the changes in metabolic states of G&N group, and found that the activity of glycolysis was significantly impaired scored in TOP10 (Fig. S2B, C), and the reaction scores of LDH activity and lactate synthesis were lowered (Fig. 2G). Previously, we have demonstrated that lactate is highly enriched in the nuclei of mouse 2-cell embryos by FiLa-sensor test [6]. Here, we observed that the two inhibitors led to a low enrichment of lactate in the nuclei of mouse 2-cell embryos (Fig. 2H), suggesting that the large amount of exogenous lactate in the culture medium can not compensate for the loss of endogenous lactate led by LDH inhibition. Therefore, we propose that the insufficiency of endogenous lactate may be associated with the developmental arrest.

3.3. Deprivation of endogenous lactate results in reduced H3K18lac and failed major ZGA in mice

In our previous study, we demonstrate that lactate regulates major ZGA by H3K18lac in mouse 2-cell embryos [6]. To investigate whether the deprivation of endogenous lactate influences mouse major ZGA, gene set enrichment analysis (GSEA) was performed and downregulated genes in the G&N group were significantly enriched in major ZGA (Fig. 3A; NES = -1.65, P = 2.85e-07). Definitely, we identified that 1,492 major ZGA genes were decreased in the G&N embryos (Fig. 3B). By IF analysis, we observed that the H3K18lac levels were markedly reduced in the 2-cell embryos treated by the LDH inhibitor (Fig. 3C; P < 0.001). To further confirm the observation, we performed cleavage under targets and tagmentation (CUT&Tag) on the G&N embryos at the 2-cell stage. The results showed the decreased H3K18lac enrichment on the promotor regions of coding genes in the G&N group (Fig. 3D). The combined analysis of RNA-seq and CUT&Tag data showed that the decreased H3K18lac enrichment on the promoter regions of majority of major ZGA genes was positively correlated with their decreased expression levels (Figs. 3E, S2D), such as Elk1 (Fig. 3F). Together, our findings show that major ZGA is regulated by endogenous lactate, alternatively, by H3K18lac enrichment on the promoters of major ZGA genes.

Fig. 3.

Endogenous lactate-deprived mouse embryos show impaired major ZGA and H3K18lac at the 2-cell stage. (A) GSEA of major ZGA genes showing preferential downregulation in 2-cell embryos of the G&N group. (B) Heatmap showing the changes of major ZGA genes expressions between control and G&N groups, and major ZGA genes were classified into three clusters. Clusters 1, 2, and 3 represent downregulated, unchanged and upregulated genes in the G&N group respectively. (C) Representative IF images of 2-cell embryos showing that H3K18lac was significantly reduced in the G&N and GSKA&N groups compared to the DMSO group. Right panel shows the quantification of H3K18lac fluorescence intensities in the control, G&N and GSKA&N embryos. Each dot represents a single nucleus. Scale bars, 50 μm. Error bars are SEM. ***P < 0.001. (D) Metaplots of H3K18lac enrichments at the promoters of coding genes in the DMSO and G&N group at the 2-cell stage. (E) Heatmap (left panel) of H3K18lac signals ranked by their relative changes in the control and G&N groups. Heatmap (right panel) showing the expression ratios of major ZGA genes between the groups indicated. (F) Genome browser snapshot of H3K18lac signals and RNA expression at the Elk1 locus of 2-cell embryos from the control and G&N groups. DMSO, mKSOM culture medium plus DMSO; G&N, mKSOM culture medium plus Gne140 and NMN; GSKA&N, mKSOM culture medium plus GSK2837808A and NMN. Differences between means were calculated using two-tailed Student's t -test.

3.4. Human PIDA embryos fail in the activation of 8-cell signature genes

The above results demonstrate that endogenous lactate insufficiency results in mouse embryonic development arrest. Whether the metabolic abnormality is associated with human PIDA is worth to be further addressed. We first observed >1,200 ART embryos of the Chinese Han population by Time-lapse, and statistical analysis revealed that approximately 40% of embryos happened PIDA (Fig. S3A), in line with the previous studies [7,35]. We found nearly half of the PIDA embryos were developmentally blocked on day 3, and mainly happened at the 6–8 cell stage (Fig. 4A). Then, we focus on the human 8-cell PIDA embryos arrested with non-genetic reasons, without cell division for 1 day and exhibiting normal morphology (Fig. 4B).

Fig. 4.

ZGA fails in human PIDA embryos. (A) Fraction of PIDA embryos across human pre-implantation development. (B) Representative images of normal and PIDA embryos at the 8-cell stage by Time-lapse. Scale bars, 50 μm. (C) Dimensionality reduction analysis of the transcriptomes of the 8-cell, 8-cell PIDA and reported 8-cell arrested embryos. (D) Heatmap showing the expressions of 8-cell signature genes in normal, 8-cell PIDA and reported 8-cell arrested embryos.

To uncover the molecular signatures of the PIDA embryos, we performed single-cell RNA-seq on a small number of human PIDA embryos that we defined above, through a fully consented and documented process, individually, and transcriptome analysis, compared to publically available RNA-seq data on normal and arrested human embryos at 8-cell stage. The normalized gene expressions were projected into hierarchical clustering and dimensionality reduction analysis, and expectedly, we found that the PIDA embryos in the study were clustered with the reported 8-cell arrested embryos, and they were clearly diverged from the normal embryos (Figs. 4C and S3B). PIDA embryos displayed no significant change in transcriptional activity (Fig. S3C), but tended to abnormal karyotypes examined by R package scploid, compared to normal ones, showing that 7 of the 18 PIDA embryos were projected to be aneuploid (Fig. S3D and E). Thus, we argue that aneuploidy may be a contributing cause to PIDA.

To further investigate the major ZGA program of PIDA embryos, we classified the human 8-cell embryo signature genes identified by a previous study [33] into two groups with higher expressions in 4-cell (cluster I) and 8-cell (cluster II) embryos, respectively (Fig. S3F), and found that the 8-cell embryo signature genes were significantly downregulated in the PIDA embryos and the reported 8-cell arrested embryos (Figs. 4D and S3 G). Collectively, these results show that the major ZGA program is damaged in human 8-cell PIDA embryos.

3.5. PEMA predicts insufficiency of endogenous lactate in human 8-cell PIDA embryos

Dimensionality reduction analysis revealed that the transcriptomes of the human PIDA embryos and the lactate-deprived mouse embryos were clustered together, and they were separated from those of normal human and mouse embryos, implying the similar transcriptional defects happen on the arrested human and mouse embryos (Fig. S4A). Therefore, we propose that the insufficiency of endogenous lactate may also occur in the human PIDA embryos. As expected, PEMA predicted the impaired glycolysis scored in TOP10 in PIDA embryos (Fig. S4B, 4C), and the reaction scores of LDH activity and lactate synthesis were reduced in most PIDA embryos (Fig. 5A), with the lower expressions of LDHA and LDHB (Fig. 5B), suggesting that the PIDA embryos may be short of endogenous lactate.

Fig. 5.

PEMA predicts insufficiency of endogenous lactate in human 8-cell PIDA embryos. (A) Heatmap showing the reaction scores in glycolysis in human normal, 8-cell PIDA and published 8-cell arrested embryos. (B) Lollipop graphs showing the lower expressions of LDHA and LDHB in 8-cell PIDA and reported 8-cell arrested embryos than those in normal embryos. (C) Quantification of H3K18lac fluorescence intensities in 8-cell normal and PIDA embryos. Two embryos for each group were used. Each dot represents a single nucleus. Error bars are SEM. ***P < 0.001. (D) Dimensionality reduction analysis of the H3K18lac profiles of the 8-cell and 16-cell normal embryos, and the 8-cell and 16-cell PIDA embryos. (E) Fraction of the genome covered by H3K18lac signals in the normal and PIDA embryos at the 8-cell and 16-cell stages, and percentages of H3K18lac peaks assigned to the promoter, intron, exon, and intergenic regions. (F) Heatmaps of H3K18lac signals (left panel) in the promotors of 8-cell signature genes in the normal and PIDA embryos indicated; right panel, expression changes of the corresponding genes. Differences between means were calculated using two-tailed Student's t-test.

The insufficiency of endogenous lactate can lead to decreased H3K18lac modification. Definitely, by IF, we observed that H3K18lac was significantly decreased in the 8-cell and 16-cell PIDA embryos compared with normal ones (Figs. 5C and S4D). Thus, we sought to check whether H3K18lac is associated with the abnormal ZGA in the PIDA embryos. Using single-cell CUT&Tag (scCUT&Tag), an approach we previously developed [6], we generated genome-wide H3K18lac modification maps for 8-cell and 16-cell PIDA embryos and compared them to those of normal 8-cell and 16-cell embryos we previously published. The generally high quality of our CUT&Tag data was validated (Fig. S4E, F). We quantified the H3K18lac signals across all samples and performed dimensionality reduction analysis, and observed that the normal and PIDA embryos could be separated clearly (Fig. 5D). The genomic coverage analysis revealed that the H3K18lac modification levels were decreased in the PIDA embryos compared with the normal embryos, especially the peaks at the promoter regions (Fig. 5E). In addition, most of the 8-cell signature genes, which are highly expressed in the normal 4-cell and 8-cell embryos and markedly decreased in the PIDA embryos, exhibited lower H3K18lac signals at their promoter regions (± 3 kb around TSS; Figs. 5F and S4 G). These observations suggest that the insufficiency of endogenous lactate may happen on human 4-cell and 8-cell PIDA embryos, resulting in the loss of H3K18lac at the promoter regions of 8-cell signature genes and impaired major ZGA.

3.6. Insufficiency of endogenous lactate happens on human ch3PN embryos

Tripronuclear (3PN) zygote, forming by two sperms entering into one oocyte, occurs at a rate of 2%–5% after conventional IVF [[36], [37], [38]]. The corrected human tripronuclear (ch3PN) zygote obtained by removal of the extra pronucleus from 3PN can develop into a genetically normal embryo, but with a significantly lower ratio to the blastocyst stage compared to normal fertilized embryos. However, the molecular mechanisms underlying the developmental obstacle remain unknown. Next, we wanted to evaluate the metabolic states of the ch3PN embryos. Single-cell RNA-seq was performed on 8-cell ch3PN embryos on day 3. By R package scploid, we found the karyotypes of ch3PN embryos were normal (Fig. S5A), suggesting that the low developmental competence of ch3PN embryos may not be attributed to abnormal karyotype. RNA-seq data analysis found that a total of 1037 8-cell signature genes were co-reduced in the PIDA and ch3PN embryos compared with normal embryos (Fig. 6A), revealing major ZGA is also impaired in ch3PN embryos. In line with the PIDA embryos, PEMA predicted that the reaction scores of LDH activity and lactate synthesis were also reduced in some ch3PN embryos (Fig. 6B). In addition, the expressions of LDHA and LDHB were significantly decreased in the ch3PN embryos (Fig. 6C). By using FiLa sensor, we found decreased level of lactate in the nuclei of the ch3PN embryos at the 8-cell stage (Figs. 6D and S5B), compared to the normal embryos, validating the loss of endogenous lactate in the ch3PN embryos.

Fig. 6.

Human ch3PN embryos show low lactate levels. (A) The scatter plots showing a total of 1,037 co-downregulated 8-cell signature genes in the human 8-cell PIDA and ch3PN embryos. (B) Heatmap showing the reaction scores in glycolysis predicted by PEMA in the 8-cell normal and ch3PN embryos. (C) The expressions of LDHA and LDHB were reduced in the 8-cell ch3PN embryos than those in the normal embryos. Error bars are SEM. *P < 0.05. (D) Quantification of Fila fluorescence intensities in the 8-cell normal and ch3PN embryos. Two ch3PN embryos were used. Each dot represents a single nucleus. Error bars are SEM. ***P < 0.001. (E) Dimensionality reduction analysis of the H3K18lac profiles of the normal, PIDA and ch3PN embryos at the 8-cell and 16-cell stages. (F) Fraction of the genome covered by H3K18lac signals in the normal and ch3PN embryos indicated, and percentages of H3K18lac peaks assigned to the promoter, intron, exon, and intergenic regions. (G) Heatmaps of H3K18lac signals (left panel) in the promotors of 8-cell signature genes in the normal and ch3PN embryos indicated; right panel, expression changes of the corresponding genes. Differences between means were calculated using two-tailed Student's t-test.

Then, we used scCUT&Tag to detect whether the insufficiency of endogenous lactate affects H3K18lac profiles for 8-cell and 16-cell ch3PN embryos. Dimensionality reduction analysis showed that the H3K18lac profiles for the ch3PN embryos were more similar with those of PIDA embryos (Fig. 6E). Also, the H3K18lac modification levels in the ch3PN embryos were decreased (Fig. 6F). In addition, most of the 8-cell signature genes, with decreased expressions in the ch3PN embryos, exhibited lower H3K18lac signals at the promoter regions in the ch3PN embryos than in the normal embryos (Fig. 6G). Together, these findings show that insufficiency of endogenous lactate happens on the 8-cell ch3PN embryos, resulting in loss of H3K18lac and impaired major ZGA, which is related to the developmental barrier of ch3PN embryos.

3.7. Lac-CoA can rescue the insufficiency of endogenous lactate-induced defects on H3K18lac, major ZGA and developmental competence of ch3PN embryos

Lac-CoA is the substrate for histone lactylation [39]. Next, we wanted to test whether Lac-CoA addition could rescue the insufficiency of endogenous lactate-induced loss of H3K18lac, to improve major ZGA and developmental competence of ch3PN embryos. Therefore, 1 mM Lac-CoA was injected into ch3PN zygotes (Fig. 7A). Firstly, we observed nearly half of the ch3PN zygotes could not reach to the 2-cell stage, but, even if they could come to the 2-cell stage, their developmental competence to the 8-cell stage was seriously impaired (Fig. 7B). By addition of Lac-CoA, more ch3PN embryos could develop into blastocysts than the control group (20.0% vs 3.03%), and, notably, the developmental competence to the 4-cell and 8-cell stage was improved remarkably (Fig. 7B).

Fig. 7.

Lac-CoA supplement can rescue the defects of ch3PN embryos. (A) Schematic presentation of the Lac-CoA injection experimental protocol. (B) Bar plots showing the fraction of human embryos at the different developmental stages of control (n = 20) or Lac-CoA addition (n = 33). (C) Heatmap of the expressions (left panel) of 8-cell signature genes in the 8-cell, ch3PN, and Lac-CoA-injected ch3PN embryos; right panel, mean expressions of the corresponding genes. (D) Immunostaining of H3K18lac of the 8-cell ch3PN and Lac-CoA-injected ch3PN embryos. Right panel shows the quantification of H3K18lac fluorescence intensities in the embryos indicated. Each dot represents a single nucleus. Scale bars, 50 μm. Error bars are SEM. *P < 0.05. (E) Fraction of the genome covered by H3K18lac signals in the embryos indicated, and percentages of H3K18lac peaks assigned to the promoter, intron, exon, and intergenic regions. (F) Heatmaps of H3K18lac signals (left panel) in the promotors of 8-cell signature genes in the embryos indicated; right panel, expression changes of the corresponding genes. Differences between means were calculated using two-tailed Student's t-test.

By analysis of RNA-seq data on Lac-CoA-added ch3PN embryos, we found that their karyotypes were normal (Fig. S5A). Notably, RNA-seq data showed that the expressions of a large portion of 8-cell signature genes were enhanced by Lac-CoA (Fig. 7C). Next, IF analysis confirmed that the H3K18lac levels were increased by Lac-CoA (Fig. 7D). In addition, we utilized scCUT&Tag to check the H3K18lac maps of Lac-CoA-treated 8-cell ch3PN embryos, which appeared higher correlation with those of normal embryos (Fig. S5C). And, the genome-wide coverage of H3K18lac levels was enhanced in Lac-CoA-treated ch3PN embryos, especially at the promoter regions (Fig. 7E). Importantly, we found that, along with the increase in the H3K18lac levels at the promoter regions, the expressions of 1,939 8-cell signature genes were enhanced (Fig. 7F). Based on these findings, we reveal that the defects induced by insufficiency of endogenous lactate in ch3PN embryo could be rescued by Lac-CoA.

4. Discussion

In the article, we developed PEMA to study metabolic programs of mammalian pre-implantation embryos, and showed that the embryos exist in a high glycolysis state when major ZGA occurs. More importantly, human PIDA embryos displayed insufficiency of endogenous lactate, leading to impaired major ZGA and aberrant H3K18lac pattern. We deciphered the metabolic reprogramming defects and molecular mechanisms that underlie PIDA.

It is supported that the transcriptomic-based methods predicting cellular metabolic states in a mammalian system are powerful [27]. However, it is ill-suited for studying metabolic programs of pre-implantation embryos, due to a large amount of translationally inactive mRNAs stored in the blastomere [40,41]. To overcome this challenge, we interrogated translational activation of mRNAs of metabolism-associated genes, which reflect the metabolic potential. Using Ribo-seq data, a weight was assigned to each metabolic reaction to correct bias for forecasting metabolic activities employing mRNA expressions of enzymes. By the optimization, the metabolic networks predicted using PEMA were highly consistent with mass spectrometry method. Furthermore, PEMA correctly predicted the role of LDH activity and lactate synthesis in human and mouse embryos when major ZGA happens, revealing that the endogenous lactate was largely produced by LDHs.

PIDA is a common phenotype of recurrent failure of ART [8,9]. However, because of technical difficulties and the scarcity of study materials, investigations involving the pre-implantation human embryo development are limited, and the underlying molecular mechanisms behind human PIDA are largely unrevealed. By multi-omics analysis, our results and others underscored that arrested embryos due to non-genetic reasons displayed major ZGA disorder, increased genomic instability and multiple epigenetic reprogramming defects. Aneuploidies have been indicated to lead to embryonic arrest. In 2024, Wang et al. have shown that aberrant chromosome segregation during mitosis results in aneuploidy formation in arrested embryos using the DNA-part dataset from single-cell multi-omics sequencing [7]. Yang et al. have reported that around 30% of the arrested embryos are predicted to be aneuploid based on RNA-seq data [42]. In the study, we examined the karyotype using R package scploid based on RNA-seq data, and our results found 7 of 18 8-cell arrested embryos showed aneuploidy. Thus, we think the aneuploidy may not comprehensively explain the reasons for the human PIDA. In addition, our results also found that the PIDA embryos were mostly failed to activation of glycolysis, and Yang et al. also showed that the arrested embryos were failed to upregulate activities of glycolysis or fatty acid oxidation [42]. The results indicate that the metabolic disorders may be the main causes contributing to the human PIDA. However, the role of metabolic reprogramming in PIDA is still lack of systematical analysis. Though impaired glycolysis has been observed in arrested embryos, using PEMA, we predicted PIDA embryos showed insufficient endogenous lactate. And, more importantly, PIDA embryos were mimicked by LDHs inhibition of mouse arrested embryos in terms of transcriptome, indicating that endogenous lactate insufficiency is a cause of PIDA. The results reveal the prediction accuracy, robustness, and biological interpretability of PEMA.

There is a large amount of lactate in the culture medium [43]. However, given that removal of exogenous lactate from the medium has no significant effect on pre-implantation embryo development, and the inhibition of LDHs led to PIDA, we propose that the endogenous lactate is indispensable for mammalian pre-implantation embryo development. Previously, we have demonstrated that lactate-derived H3K18lac correlates with the expressions of 8-cell embryo signature genes in humans [6]. Here, we further show the loss of H3K18lac at the promoter regions of 8-cell embryo signature genes in the PIDA embryos, highlighting the role of H3K18lac in the ZGA program. The ZGA program is regulated very complexly. Epigenetic reprogramming after fertilization enables transcriptionally quiescent maternal and paternal chromatin to acquire a permissive state for subsequent ZGA, such as H3K27ac, which marks genomic regions primed for ZGA [44]. Thus, the major ZGA genes, whose expressions were upregulated upon loss of H3K18lac, might be activated by other epigenetic modifications. Furthermore, we utilized ch3PN embryos to determine the role of H3K18lac in ZGA and pre-implantation development of human embryos. The ch3PN embryos are reliably corrected by removing the specific extra pronucleus from 3PN zygotes, and developing into a genetically normal embryo, but their pre-implantation development is severely comprised. In the study, it was at a rate of 3.03% developed into blastocyst with a majority of embryos arrested before 8-cell stage. And, the transcriptomic analysis of the 8-cell arrested ch3PN embryos showed that the badly impaired ZGA occurs in the embryos, suggesting that the removal of extra pronucleus cannot correct the ZGA program. In addition, the previous reports have also indicated that the removal cannot correct the epigenetic status, such as DNA methylation and H3K9me3 patterns, which are much higher in the ch3PN embryos than in the normal fertilized embryos [31,45]. Here, the aberrant H3K18lac levels were also detected in the ch3PN embryos, suggesting that the normal development of an embryo requires not only a correct diploid genome but also normal epigenetic patterns, which facilitate the ZGA program. Importantly, we found that the injection of lac-CoA effectively promoted H3K18lac levels on the promoter regions of human 8-cell embryo signature genes, and facilitated ZGA and increased the developmental competence of ch3PN embryos. Along with that of our prior study, we propose the conserved role of H3K18lac in regulating ZGA and pre-implantation development in humans and mice.

5. Conclusion

PEMA is a powerful tool to improve our understanding of metabolism in mammalian embryos. Using it, we systematically unveil the mechanism of PIDA led by metabolic disorders, and demonstrate the role of endogenous lactate in major ZGA by H3K18lac. We also found that the use of lac-CoA can promote the pre-implantation development of human ch3PN embryos. Lac-CoA is derived from lactate. Thus, with the promising results, we propose that future studies should test to enhance endogenous lactate by activating LDHs in the developmental microenvironment in vivo or in vitro, to improve human embryo quality and viability.

Data availability

Publicly available datasets analyzed in this work are available in GEO, the GSE ID shown below: GSE101571 and GSE36552 (RNA-seq datasets of human early embryos), HRA003366 (single-cell RNA-seq datasets of human 8-cell arrested embryos), GSE197265 (RNA-seq and Ribo-seq datasets of human early embryos), GSE165782 and GSE209648 (RNA-seq and Ribo-seq datasets of mouse early embryos). RNA-seq and CUT&Tag data of mouse early embryo derived from our previously published data (GSE234027). All scRNA-seq and scCUT&Tag data of human PIDA embryos generated in this study have been deposited in GSA-Human under accession HRA006017.

Code availability

The computational pipeline named PEMA for characterizing metabolic states of human and mouse pre-implantation embryos using RNA-seq and Ribo-seq data is available at https://github.com/summus-kong/PEMA.

Declaration of competing interest

The authors declare that they have no conflicts of interest in this work.

Biographies

Jingyu Li is currently a professor and director of Chongqing Key Laboratory of Human Embryo Engineering and Precision Medicine, Women and Children's Hospital of Chongqing Medical University, working on genetic, metabolic and epigenetic regulation of gametogenesis, fertilization and early embryonic development. He has published > 30 papers in international journals including Cell Discovery, National Science Review, Cell Reports and EMBO Reports as the corresponding or first author.

Qingran Kong received his PhD in Developmental Biology from Northeast Agricultural University in 2010. He is a full professor in School of Laboratory Medicine and Life Sciences of Wenzhou Medical University from 2020, and major in metabolism and epigenetic regulation in mammalian early embryo development based on multi-omics data, publishing in Cell Discovery, National Science Review, Cell Death and Discovery and JBC, etc.