Transcriptomic profiles reveal the characteristics of oocytes and cumulus cells at GV, MI, and MII in follicles before ovulation

Upon obtaining ethical approval, we collected a total of 139 samples from 47 patients undergoing preimplantation genetic testing (PGT). The patients had an average age of 35.1 years (Supplementary Tables 1 and 2) for RNA sequencing (Fig. 1A), including 97 CCs (GV: 35; MI: 30; MII: 32) and 42 oocytes (GV: 22; MI: 11; MII: 9). The quality of the sequencing results from all samples met the requirements for further analysis.

The first principal component (PC1) was found to be mainly influenced by differences between oocytes and CCs. PC2, on the other hand, reflected differences among the CC cohorts, amounting to 71.6% and 4.7%, respectively (Fig. 1B). This suggests that oocytes and CCs can be clustered into two distinct populations that are not dependent on the maturation stage. The average amount of raw data per sample was about 3 GB. Low expression data and individual-specific expression data were filtered out to obtain significantly differentially expressed genes (DEGs) with less bias. To identify transcripts that are specifically enriched at each maturation stage, we considered genes with a log2-fold change ≥ 2 and a p‐value < 0.05 as DEGs. We identified a total of 7460 DEGs between the two cell populations, among which 3982 were significantly highly expressed in oocytes, and 3478 in CCs (Supplementary Fig. 1). Next, the expression levels of known cell type-specific markers for oocytes (FIGLA, GDG9, SYCP3, ZP3) [3] and CCs (ADAMTS1, AREG, COX20, STAR) [5] were analyzed. The results showed that the marker genes for oocytes were significantly highly expressed in oocytes, while the marker genes for CCs were significantly highly expressed in CCs (Fig. 1C, D).

Principal component analysis (PCA) showed that all GV oocytes and MII oocytes were well separated, while MI oocytes were clustered into two groups. Five of the MI oocytes were clustered with GV oocytes, while the remaining six were in the middle of the GV and MII oocytes population (Fig. 2A). This was further confirmed by Pearson correlation, which showed that the five MI oocytes were highly correlated with GV oocytes, and the remaining six were positioned between GV oocytes and MII oocytes. This suggests that MI oocytes can be classified into two subgroups: GV-like MI oocytes and MII-like MI oocytes (Fig. 2B). To represent the transcriptional profile of MI oocytes more accurately, we used the six MII-like-MI oocytes as the sample for subsequent analysis. We analyzed DEGs that were highly expressed in oocytes at different stages. Genes are organized into three clusters. Custer1 shows high expression in the MII cohort, while cluster2 is expressed in the GV cohort. Cluster 3 appears to be a gene set that is intermediate between the two (Fig. 2C). Genes that are highly expressed in GV oocytes are annotated to mitochondrial translation, mitochondrial gene expression, and mitochondrial translation elongation. Genes that are upregulated in MI populations are associated with processes such as chromosome segregation, gene silencing, and cell division. Genes that are upregulated in MII populations are involved in covalent chromatin modification, regulation of gene expression, epigenetics, and histone modification (Fig. 2D). Given the significant role of the mitochondrial translation pathway, we conducted a screening for genes that are enriched in this pathway to observe their expression at three different stages. We then normalized the expression in the GV stage. The results showed that these genes were highly expressed during the GV stage, which decreased during the MI stage, and had the lowest expression during the MII stage (Fig. 2E). This indicates that mitochondria are active during the GV stage, and the GV stage oocytes require a large amount of energy to support their completion of meiosis [14]. As oocytes mature to the MII stage, mitochondrial function and energy substrate utilization gradually decline. Some of the unused energy substrates may be stored [15].

To investigate the key regulators and establish a transcriptional regulatory network during the maturation of human preovulatory oocytes, we utilized the ARACNe method to analyze all 1665 known transcription factors (TFs) from the Human Transcription Factor Database (TFDB v3.0) [16, 17]. We observed the upregulation of MYSM1, ATF2, and BCLAF1 in MI oocytes, which indicates that these TFs may play a critical role in the transition from the GV to the MI stage. BCLAF1, SON, ETV3, TMF1, MLLT10, SSRP1, PBX3, ZBTB22, and ZNF292 were found to be highly expressed in the oocytes from MII oocytes compared to those from MI oocytes (Fig. 3), implying that they are likely the regulators of the MI-to-MII stage transition. Interestingly, we found that BCLAF1 was upregulated in oocytes during both the GV-to-MI and MI-to-MII transitions. These may suggest a potential role for BCLAF1 in regulating transcription networks during human oocyte maturation before ovulation. It is worth noting that the transcriptional regulatory network between the three stages of CCs cannot be constructed due to the limited number of differentially expressed genes. This may require additional investigation.

When CC samples were analyzed using PCA, the results showed that all CC samples clustered loosely according to the development period (Supplementary Fig. 2A). Heatmaps of all genes in the CCs showed no transcripts that were specifically expressed during different stages of follicle maturation (Supplementary Fig. 2B).

However, upon comparing GV CCs and MII CCs, there were more DEGs (log2-fold change > 0.5 and p‐value < 0.05) enriched in functional pathways. The comparative analysis of DEGs abundance between CCs at the GV stage and CCs at the MII stage showed that 148 transcripts were significantly downregulated in the CCs at the MII stage. KEGG and GO enrichment analyses showed that the DEGs in GV CCs were mainly related to negative regulation of plasminogen activation, regulation of the MAPK signaling pathway cascade, regulation of protease activity, regulation of the transcriptional growth factor β (TGF-β) receptor signaling pathway cascade, transmembrane receptor protein serine/threonine, tyrosine kinase signaling, and the classical Wnt signaling pathways (positive regulation). These genes are responsible for cell–cell interactions [18‐20] Fig. 4A).

In addition, although we did not find any association between glucose metabolism in the annotated signaling pathways enriched in the DEGs, we did find that two genes associated with glucose transport were highly expressed in GV CCs. Solute Carrier Family 2 Member 1 (SLC2A1) is responsible for the facilitated diffusion of glucose and interacts with Angiopoietin-like 4 (ANGPTL4), which is also highly expressed in the GV phase. Furthermore, we discovered that among the few DEGs in GV CCs vs. MII CCs, THAP4, HOMER1, and HOMER3, which are related to glucose homeostasis, were significantly highly expressed in the GV CC cohort. This suggests that these genes are involved in glucose metabolism during the early stages of follicular development (Fig. 4C). Similarly, some terms concerning amino acid metabolism were found to be enriched in DEGs in the GV phase. The transcriptomes of CCs revealed the contributions of amino acid and glucose metabolism contribute to follicular metabolism during the early stage of oocyte maturation.

To further investigate the roles of transcripts of CCs in the later stages of oocyte maturation, we also conducted an enrichment analysis of 117 genes upregulated in MII CCs. The GO and KEGG analyses revealed many pathways related to lipid metabolism, including cholesterol anabolism and FA metabolism (Fig. 4B, D). MII CC cohort genes that were differentially up-regulated were annotated to cholesterol biosynthesis and the steroid and alcohol metabolic pathways, indicating that sterol metabolism of CC is a significant factor in oocyte development which is consistent with previous animal studies [21‐25]. Subsequently, we conducted a cluster heatmap analysis of 17 transcripts involved in cholesterol metabolism. We observed a significant increase in the expression levels of these genes as oocyte development progressed (Fig. 4E). We investigated the expression profiles of several key genes involved in the cholesterol metabolism pathway during the GV, MI, and MII phases. These genes exhibited low transcriptional activity during the GV stage but were upregulated in MI CCs, eventually reaching a high level of expression in MII CCs (Supplementary Fig. 2C).

The GO analysis also showed that MII CCs exhibited enrichment in transcripts associated with fatty acid (FA) metabolism (Fig. 4B). To obtain a comprehensive understanding of changes in FA metabolism, we examined biological pathway terms related not only to FA metabolism but also to catabolism. Additionally, we analyzed terms related to triglyceride (TG) and acyl-CoA metabolism (Fig. 4F). Next, we analyzed the expression levels of some FA synthetase-encoding genes, including FADS2, FABP3, ELOVL5, ELOVL6, ACSS2, and AACS, at the GV, MI, and MII stages. The expression of all these genes gradually increased as the developmental stage progressed and reached its peak in the preovulatory follicles (Fig. 4G). The qPCR results of several genes confirmed our findings (Supplementary Fig. 3). We observed an increase in FA synthesis in the follicles as they matured, indicating that FA accumulation peaks before ovulation.

To determine the energy source utilized during early oocyte maturation, we analyzed the expression of genes involved in two main biological energy production processes: glycolysis, and FA β-oxidation, in oocytes and CCs. We discovered that the abundance of genes associated with the glycolysis pathway was significantly higher than that of genes associated with the FA β-oxidation pathway in the GV and MI stages. We also found that the abundance of genes associated with the glycolysis pathway in CCs is much higher than in oocytes (Fig. 5A,B), implying that the primary energy source of oocyte maturation is provided by CCs’ glycolysis.

A total of 42 oocytes and 97 corresponding CCs were collected voluntarily after obtaining written informed consent from 47 donor couples, who were undergoing Preimplantation Genetic Testing for Aneuploidies (PGT-A) at the Reproductive & Genetic Hospital of CITIC-XIANGYA, using standard antagonist protocols. Patients with a history of chromosomal abnormalities or a family history of genetic abnormalities, or who have been found to have abnormal karyotypes were excluded. Couples who donated oocytes were informed that the donation posed a potential risk to their fertility success for that cycle. The MII oocytes were donated by couples who had > 20 oocytes derived from the same IVF or ICSI cycle. No financial benefit was involved in the donation process. This study was approved by the Ethics Committee of CITIC XIANG-YA Reproductive and Genetic Hospital (NO: LL-SC-2019–005).

Each cumulus-oocyte complex (COC) was placed in a single droplet, digested with 80 IU/ML hyaluronic acid droplets in a pre-warmed Falcon 3002 Petri dish, and the CCs were mechanically dissociated. Due to hyaluronidase’s strong digestion ability, it was unsuitable for excessive digestion, generally not more than 1 min, to reduce damage to CCs and oocytes. Rapid transfer of COC to be separated into pre-prepared ICSI operating fluid, repeat gently a few more times until completely separated. Then, the naked oocyte removed was quickly transferred to the operating dish for marking, and the maturation stage was assessed and recorded by observing the nucleus of the oocyte. Oocytes with an extruded polar body were deemed MII oocytes, oocytes with an intact germinal vesicle were deemed GV oocytes, and oocytes without an observable germinal vesicle or an extruded polar body was deemed MI oocytes. A total of 42 oocytes (GV:22, MI:11, MII:9) and 97 CCs (GV:35, MI:30, MII:32; randomly selected 10 cells from individually COC per CC samples to avoid bias due to the different number of CCs in each follicle) were getting. After cleaning, each sample was placed in a separate EP tube that had been added to the RNA lysate for lysis (or stored in a -80℃ refrigerator).

Single-cell was lysed by RNA lysates, and RNA was released and converted into cDNA using the Smart RNA-seq2 protocol as previously described [46], the sequencing technology widely used for its high sensitivity and specificity. Briefly, Superscript II reverses transcriptase, and oligo-dT primers are used to specifically reverse RNA with the poly (A) tail. Once reversed to the 5′ ends of the RNA molecule, reverse transcriptase adds a C base to the 3′ ends of cDNA. The template-switching oligonucleotide modified with Locked Nucleic Acid is then used to pair with C, and the second strand is synthesized. After template conversion, the resulting cDNA contains the complete sequence of the mRNA 5 end and the anchor sequence for the synthesis of the second cDNA strand, which is then amplified. The library was amplified to make a DNA nanoball (DNB) which had more than 300 copies of one molecular. The DNBs were loaded into the patterned nanoarray, and single end 50 bases reads were generated in the way sequenced by combinatorial Probe-Anchor Synthesis(cPAS) (BGI, Shenzhen).

According to the results of differential gene analysis, the R software packages hclust and heatmap were used to cluster analysis of Pearson correlation among the samples, and the visualized heatmap and unsupervised principal component analysis (PCA) diagram were used for transcriptome differences of these samples.

According to the results of differential gene analysis, R software package hclust, and Pheatmap were used for clustering analysis of Pearson correlation between samples, and visual heatmap and principal component analysis diagram were output.

Differential gene function enrichment analysis and data visualization: R software package ClusterProfile: GO enrichment analysis of differential genes. The list of biological process (BP), molecular function (MF), and cellular component (CC) of differential genes were output respectively. Metascatpe online database for gene function analysis:https://​metascape.​org/​gp/​index.​html#/​main/​step1 The website integrates GO Term, KEGG, Reactome, and BioCarta, as well as MSigDB multiple pathways. The above two databases enriched the genes with significant differences and drew corresponding annotation maps.