Skip to main content
NCBI home page
FASEB BioAdvances logoLink to FASEB BioAdvances
. 2025 Jul 9;7(7):e70036. doi: 10.1096/fba.2025-00075

Single‐Cell Analysis of Alternative Splicing and Gene Regulatory Network Reveals Remarkable Expression and Regulation Dynamics During Human Early Embryonic Development

Jiwei Chen 1,2, Gaigai Wei 3, Fangge Sun 2,4, Yunjin Li 1, Shuang Tang 2,✉, Geng Chen 1,✉
PMCID: PMC12239682  PMID: 40641848

ABSTRACT

Single‐cell RNA‐seq (scRNA‐seq) technologies greatly revolutionized our understanding of cell‐to‐cell variability of gene expression, but few scRNA‐seq technologies were used to describe the expression dynamics at the isoform and exon levels. Although the current expression profile of early embryos was studied focusing on the expression changes at the gene level, systematic investigation of gene expression dynamics of human early embryonic development remains insufficient. Here we systematically explored the gene expression dynamics of human early embryonic development integrating gene expression level with alternative splicing, isoform switching, and expression regulatory network. We found that the genes involved in significant changes in these three aspects are all gradually decreased along embryonic development from E3 to E7 stage. Moreover, these three types of variations are complementary for profiling expression dynamics, and they vary significantly across embryonic development as well as between different sexes. Strikingly, only a small number of genes exhibited prominent expression level changes between male and female embryos in the E3 stage, whereas many more genes showed variations in alternative splicing and major isoform switching. Additionally, we identified functionally important specific gene regulatory modules for each stage and revealed dynamic usage of transcription factor binding motifs (TFBMs). In conclusion, this study provides informative insights into gene dynamic characteristics of human early embryonic development by integrating gene expression level with alternative splicing, isoform switching, and gene regulatory networks. A systematic understanding of gene dynamic alteration features during embryonic development not only expands knowledge on basic developmental biology but also provides fundamental insights for regenerative medicine and developmental diseases.

Keywords: alternative splicing, differential expression, early embryonic development, gene regulatory network, major isoform switching, single cell RNA‐seq

By integrating single‐cell RNA‐seq data, this study reveals dynamic changes in gene expression, alternative splicing, and isoform switching during human early embryonic development (E3–E7). These regulatory layers are complementary and stage‐specific, with sex‐dependent differences especially evident at E3. Gene regulatory networks further uncover key transcription factors and dynamic motif usage, offering new insights into early developmental regulation.

graphic file with name FBA2-7-e70036-g004.jpg

1. Introduction

To date, scRNA‐seq has been widely applied to explore cell subtypes, gene expression dynamics and heterogeneity of diverse tissues in many species [1, 2, 3, 4, 5, 6, 7, 8], which have revealed various meaningful results. With the innovation of scRNA‐seq, an increasing number of different scRNA‐seq protocols have been developed, but only the full‐length transcript capturing scRNA‐seq technologies (such as Smart‐seq2 [9]) allow the analysis of alternative splicing rather than the 3′‐end capturing protocols (e.g., drop‐seq [10]) [8]. However, the vast majority of published scRNA‐seq data was generated using the 10× 3′ scRNA‐seq method, which primarily focuses on gene‐level expression changes, without emphasizing the expression dynamics at the isoform and exon levels.

Gene expression dynamics of human early embryonic development and underlying regulatory mechanisms remain unclear. Previously, Yan et al. investigated the transcriptome landscape of human early embryos from oocyte to blastocyst, where they revealed dynamic expression patterns across different developmental stages [11]. Then Xue et al. further compared the gene expression profiles between humans and mice from oocyte to morula, and highlighted the evolutionary conservation of some expression networks in early embryonic development [12, 13]. Recently, Petropoulos et al. provided a single‐cell transcriptome profiling of a number of human preimplantation embryos across five developmental stages (E3–E7) [14]. However, these studies did not explore the alternative splicing and gene regulatory profiles of early embryos. Both of the changes in gene expression level (gene‐level) and alternative splicing (exon‐level) are crucial for understanding the gene expression dynamics of early embryonic development. Moreover, isoform switching (isoform‐level) is another important aspect of gene expression since most human genes could encode multiple different isoforms, but none of these studies investigated and compared these different‐level changes. Besides, a network view of dynamic gene expression regulations during embryonic development for unraveling developmental regulatory changes is still lacking. Consequently, it is necessary to further explore the expression of genes from different aspects (gene‐level, isoform‐level, exon‐level, and regulatory network) to better and more comprehensively understand the gene expression dynamics and the underlying mechanisms of embryonic development.

Here we comprehensively analyzed the gene expression profile of the human early embryonic development process from distinct levels of gene, isoform, exon, and regulatory network based on the scRNA‐seq data of 1529 cells from 88 male and female embryos. Besides differential gene expression analysis, we also investigated the changes of alternative splicing and major isoform switching along embryonic development as well as the sex differences between male and female embryos. Furthermore, we compared the genes and related pathways involved in differential expression, significant alternative splicing, and major isoform switching, and revealed large inconsistency among these aspects of gene expression changes. Finally, we constructed a gene expression regulatory network of embryonic development and identified stage‐specific regulatory modules as well as enriched motifs bound by corresponding transcription factors. Our systematic analyses revealed interesting gene expression dynamics from diverse aspects, which deepen the understanding of underlying processes and mechanisms of human early embryonic development.

2. Materials and Methods

2.1. Single‐Cell RNA‐Seq Data Processing and Sex Determination of Early Embryos

A total of 1529 scRNA‐seq data of 88 human male and female early embryos (from stage E3 to E7) were first downloaded from the ArrayExpress database (https://www.ebi.ac.uk/arrayexpress/) with accession number E‐MTAB‐3929. The paper by Petropoulos et al. [14] provides information on both the origin of the embryos and how their ages were determined. Here's a summary of the relevant details:

The embryos were human preimplantation embryos derived from in vitro fertilization (IVF) procedures. The authors used donated human embryos from IVF clinics, following ethical guidelines and informed consent from the donors. These embryos were cultured in standard IVF conditions. The embryo age was determined based on the number of cell divisions. In particula, the authors focused on early‐stage embryos, from the 2‐cell to the 8‐cell stage, up to the blastocyst stage. The ages of the embryos were tracked by their developmental stages (such as 2‐cell, 4‐cell, etc.) and by timing during the culture period, as embryos from IVF procedures are typically cultured for specific time intervals (e.g., 3 days for cleavage‐stage embryos, and 5–6 days for blastocysts). Then the scRNA‐seq data were mapped to the human reference genome GRCh38 using HISAT2 [15] (version 2.1.0) with default parameters. Expression of genes and transcripts were quantified in Transcripts Per Kilobase Million (TPM) by StringTie [16] (version 1.3.3b) with parameters of “‐e ‐A” based on the human gene annotation file of the Ensembl database in GTF format (version 91). In order to determine the sex of each cell, the sum of TPM values for seven Y chromosome genes (DDX3Y, EIF1AY, KDM5D, PRKY, RPS4Y1, UTY and ZFY) was used to distinguish male embryos (cut off: ≥ 60 TPM) from female embryos (cut off: ≤ 40 TPM) (see Figure S1) [14].

2.2. Differential Expression Calling of scRNA‐Seq Data

Differential gene expression analyses between adjacent embryonic stages or between male and female embryos were conducted by employing the tool of Single‐Cell Differential Expression (SCDE, version 2.8.0) [17]. Then differentially expressed genes (DEGs) were determined using the criteria of |cZ| > 1.96, which corresponds to FDR < 0.05.

2.3. Differential Alternative Splicing Analysis of Human Early Embryos

To identify the alternative splicing events that significantly differ between distinct conditions, we used the software specifically designed for scRNA‐seq data of BRIE [18] (version 0.2.0) to analyze the differential alternative splicing events between male and female cells or between adjacent stages. Then differential alternative splicing genes (DASGs) were selected with the threshold of Bayes factor > 10.

2.4. Major Isoform Switching Analysis

To investigate the genes that switched the isoform with the highest expression (major isoform) in different conditions, the transcript expressions of each multi‐isoform gene in each sample were ranked from large to small according to their expression level. Considering that each embryonic stage or each gender of a certain stage contains many single cells, we defined the transcript that has the highest expression among all isoforms of a gene in at least 60% of the cells for a given condition as the major isoform. Then the major isoform of each multi‐transcript gene in each condition was determined. The genes that switched major isoforms were denoted as major isoform switching genes (MISGs).

2.5. Gene Functional Enrichment Analysis of GO Terms and Pathways

Gene Ontology (GO) enrichment analysis of biological processes and KEGG pathway enrichment analysis were carried out for the corresponding gene set using the R package clusterProfiler [19] (version 3.8.1). Only those GO terms and pathways with p value < 0.05 were considered statistically significant.

2.6. Construction of Gene Expression Regulatory Network

Gene expression regulatory network inference was conducted by SCENIC [20] (version 1.0.1.1) following the standard analysis pipeline. After the filtering process of SCENIC, the count matrix of genes was used as input to perform co‐expression analysis based on GENIE3 [21] (version 1.4.0) and the co‐expressed genes potentially regulated by transcription factors (TFs) were identified. The TF co‐expression modules were then confirmed by RcisTarget [20] (version 1.2.0) using the human GRCh38/hg38 database. The significant TF‐modules identified by SCENIC were defined as regulons.

2.7. Identification of Stage Specific Regulons and Motifs

In order to identify the regulons and motifs specific to each stage, we first performed gene expression enrichment analysis through the analysis of variance (ANOVA) based on the gene expression profile of human early embryos. ANOVA followed by Tukey's range test (Tukey's honest significance differences) was employed to detect the genes with significantly higher expression in one stage compared to other stages (fold change > 2 and adjusted p value < 0.01). Then the TFs with significantly enriched expression (EEGs) in each stage were obtained. Combining with the regulons identified by the SCENIC pipeline, the regulons of TFs that specifically expressed in each stage were determined and the motifs bound by those TFs were defined as stage‐specific motifs.

3. Results

3.1. Gene Expression Level, Alternative Splicing and Major Isoform Switching Are Very Dynamic Across Embryonic Development

To systematically investigate the gene expression changes from different levels during embryonic development, we first explored the changes in gene expression level, alternative splicing, and major isoform switching between adjacent embryonic stages based on 1529 scRNA‐seq data of human early embryos from E3 to E7 stages. The software SCDE was employed to carry out differential gene expression calling for a total of 58,302 genes (Ensembl gene annotation). Using the threshold of FDR < 0.05, we detected thousands of DEGs in each group of adjacent stage comparisons (Figure 1A). Interestingly, the number of stage‐related DEGs gradually decreased along embryonic development, where the largest number of genes showed significant expression changes between E3 and E4 stages, whereas the lowest count of DEGs was identified in E6 versus E7 (Figure 1A). Moreover, more genes changed in expression level in male embryos compared to those of female embryos during embryonic development, except E6 versus E7. In adjacent stage comparisons, a large portion of DEGs shared between male (47.8%–72.4%) and female (27.7%–76.3%) embryos. Surprisingly, only 127 DEGs were common among those neighboring stage comparisons, leaving most of the DEGs different between any two groups (Figure 1B).

FIGURE 1.

FIGURE 1

Open in a new tab

Distribution and comparison of DEGs, DASGs and MISGs for adjacent stage comparisons. (A) Bar plot showing the number of differentially expressed genes (DEGs) identified between each pair of adjacent embryonic stages (E3–E4, E4–E5, E5–E6, E6–E7), highlighting dynamic transcriptomic changes across early development. (B) Venn diagram illustrating the overlap among DEGs across adjacent stage comparisons. (C) Count distribution of DASGs between each neighboring stage pair, indicating stage‐specific regulation at the splicing level. (D) Venn diagram summarizing the relationships among DASGs for nonoverlapping stage comparisons. (E) Bar plot presenting the number of MISGs for each stage‐to‐stage transition, reflecting isoform dynamics throughout embryonic progression. (F) Venn diagram depicting the degree of overlap among MISGs across nonadjacent stages.

We further explored the alternative splicing changes between adjacent developmental stages using BRIE. Similar to the trend of stage‐related DEGs, the amount of DASGs was also gradually reduced during embryonic development (Figure 1C). Specifically, the counts of DASGs between neighboring stages range from 2415 (E3 vs. E4) to 1183 (E6 vs. E7). More genes also exhibited differential alternative splicing in male embryos compared to those of female embryos in each neighboring stage comparison. We also observed that a large portion of stage‐related DASGs was common between male (28.8%–57.3%) and female (24.2%–51.8%) embryos, whereas the percentages of shared DASGs were smaller than those of corresponding stage‐related DEGs. Those stage comparison groups only have 96 common DASGs, whereas each group possesses hundreds of unique DASGs.

Considering that the major isoform (the isoform with the highest expression of a gene) switching is also an important aspect of gene expression changes, we then examined the isoform switching events between adjacent embryonic stages. Strikingly, the number of MISGs was gradually decreased with embryonic development as well, which was consistent with that of DEGs and DASGs (Figure 1E). Except for the comparison of E3 versus E4, a bigger number of genes switched their major isoforms in male embryos than that of female in other stage comparisons. Notably, only four MISGs (SLC39A7, TMC7, CLDN6, and DDR1) shared among distinct comparison groups (Figure 1F). SLC39A7 is a zinc transporter that plays a critical role in regulating cell growth and death. Yan et al. [22] found SLC39A7 is required for eye, brain, and skeleton formation during early embryonic development in zebrafish. The function of TMC7 is not very clear. A recent study by Cheng et al. [23] revealed that TMC7 is overexpressed in pancreatic carcinoma and contributes to tumor progression and metastasis. CLDN6 is one of the earliest molecules expressed in embryonic stem cells committed to epithelial differentiation in both murine and human tissues [24, 25]. DDR1 has been proved to be functionally important in differentiation, cell motility, collagen synthesis, and signaling [26]. Consequently, our results show that the changes in gene expression level, alternative splicing, and major isoform switching are all decreased during early embryonic development, and the genes involved in these three types of changes varied greatly across different embryonic stages.

3.2. Male and Female Embryos Have Large Differences in Gene Expression, Alternative Splicing, and Major Isoform Usage

To further check the expression changes associated with sex differences during early embryonic development, we further explored the DEGs, DASGs, and MISGs between male and female embryos in each stage. The E6 stage had the greatest amount of DEGs, whereas the E3 stage exhibited the fewest number (only 47) of genes changed in expression for gender comparison (Figure 2A). Notably, 32 of those 47 DEGs (68.1%, 20 X chromosome genes and 12 Y chromosome genes) were from sex chromosomes, suggesting that most of the autosome genes do not vary in expression level in the E3 stage. In contrast, a number of genes were significantly changed in alternative splicing and major isoform switching between male and female embryos of E3. Specifically, 1038 sex‐related DASGs were detected in E3, which is the second largest compared to other stages (Figure 2B), and the largest number of sex‐related MISGs (1248) were identified in E3 (Figure 2C). Thus, although only a small fraction of genes exhibited significant changes in expression level between male and female embryos of the E3 stage, great differences exist in alternative splicing and major isoform switching. Only 17 DEGs, 13 DASGs, and 5 MISGs were shared among different stages of sex comparison, and a large portion of unique DEGs, DASGs, and MISGs were identified in each stage (Figure 2D–F). Interestingly, all these 17 DEGs are from sex chromosomes (six of them come from X chromosome and the remaining 11 are from Y chromosome), and 12 of the 13 DASGs are encoded by autosomes, and four of the 5 MISGs come from sex chromosomes. Accordingly, the genes involved in expression level changes are largely distinct from those significantly changed in alternative splicing and isoform switching for gender comparison in early embryonic development.

FIGURE 2.

FIGURE 2

Open in a new tab

Distribution and comparison of DEGs, DASGs and MISGs for sex comparisons of different stages. (A–C) Bar plots representing the numbers of DEGs, DASGs, and MISGs between male and female embryos at each individual stage (E3–E7), enabling identification of sex‐biased transcriptomic and splicing patterns. (D–F) Venn diagrams showing the overlap among DEGs, DASGs, and MISGs in male vs. female comparisons for each stage, which uncover common and unique sex‐specific regulatory events during early embryogenesis.

Moreover, we found that a significant portion of sex‐related DEGs are from sex chromosomes, but the great majority of sex‐related DSAGs and MISGs are expressed by autosomes. Only 47 DEGs were detected in the E3 stage between male and female; 68.1% of these sex‐related DSAGs were from sex chromosomes (20 X chromosome genes and 12 Y chromosome genes), and the E4 stage had 52.75% of sex‐related DEGs encoded by sex chromosomes, whereas a much smaller portion of sex‐related DEGs were generated from sex chromosomes for E5 (17.49%), E6 (10.81%) and E7 (13.16%) stages, indicating that many autosome genes mainly exhibited gene expression level changes between male and female embryos after the E4 stage. By contrast, lots of autosome genes showed significant alternative splicing variations and major isoform switching between different sexes from the E3 stage, where only 2.28%–4.07% of DASGs and 4.36%–5.95% of MISGs were from sex chromosomes (see Table S1). Therefore, the expression level profile of genes is much different from that of alternative splicing and major isoform switching between male and female embryos during early embryonic development.

3.3. Gene Expression Level, Alternative Splicing and Isoform Switching Are Complementary in Profiling Gene Expression Variations

Then we compared the DEGs, DASGs, and MISGs resulted from the comparisons of embryonic development and different sexes. Strikingly, only a small portion of genes were common between any two types of DEGs, DASGs, and MISGs for stage comparisons (Figure 3A–D), which is similar to the result of sex comparisons for distinct stages (Figure 3E–I). This could be explained by the fact that DEGs, MISGs, and DASGs mainly reflect the expression changes at levels of gene, isoform, and exon, respectively. Exon level variation in expression may not affect the whole expression at gene and isoform levels, and isoform level changes do not necessarily result in the expression change at gene level as well. Moreover, for the genes shared among DEGs, DASGs, and MISGs, the percentages for stage comparisons ranged from 3.9% to 11.4%, whereas the ratios were smaller for sex comparisons (0.5%–5.3%). Combining the three types of genes, DEGs occupied the largest portion of genes, followed by DASGs and MISGs along the embryonic development (Figure 3A–D). However, distinct results were observed between male and female embryos. The DEGs detected from sex comparisons took up the smallest fractions among these three types of genes in both E3 and E4 stages but occupied the largest portions in sex comparisons of E5, E6, and E7 stages (Figure 3E–I). Therefore, the genes involved in changes of expression level, alternative splicing, and major isoform switching are largely different across distinct embryonic developmental stages as well as between disparate sexes. Our results also show that DEGs, DASGs, and MISGs reflect the expression changes of genes from distinct aspects, which are complementary for dissecting gene expression dynamics.

FIGURE 3.

FIGURE 3

Open in a new tab

Comparison of stage and sex related DEGs, DASGs and MISGs. (A–D) Overlap and intersection analysis of DEGs, DASGs, and MISGs associated with developmental stage transitions, facilitating dissection of shared and distinct regulatory signatures during temporal progression. (E–I) Comparative analysis of DEGs, DASGs, and MISGs between sexes at each stage (E3–E7), revealing the extent and nature of sex‐based differences in gene regulation at the expression and isoform levels.

3.4. DEGs, DASGs, and MISGs Are Involved in Distinct But Important Pathways Related to Embryos

To gain insights into the functions of those DEGs, DASGs, and MISGs, we first conducted gene functional enrichment analysis for stage‐related genes using clusterProfiler [19]. As expected, a notable portion of the significantly enriched pathways was distinct among stage‐related DEGs, DASGs, and MISGs. Specifically, the stage‐related DEGs are mainly involved in thermogenesis, oxidative phosphorylation, and adherens junction (Figure 4A), whereas stage‐related DASGs are mainly enriched in autophagy, ubiquitin‐mediated proteolysis, and phosphatidylinositol signaling system (Figure 4B), and the top enriched pathways for stage‐related MISGs are ribosome, cell cycle, and HIF–1 signaling pathway. Intriguingly, we found that most of the significantly enriched pathways for stage‐related MISGs and DASGs are common with those of stage‐related DEGs (Figure 4D, p value < 0.05). But only 10 significantly enriched pathways are shared among the DEGs, DASGs, and MISGs of stage comparisons (such as endocytosis, autophagy‐animal, mitophagy, mTOR signaling pathway, Fc gamma R‐mediated phagocytosis, endometrial cancer, choline metabolism in cancer, tight junction, AMPK signaling pathway and ubiquitin mediated proteolysis), and each type of stage‐related gene is enriched in some unique pathways (Figure 4D, p value < 0.05).

FIGURE 4.

FIGURE 4

Open in a new tab

Enriched pathways for DEGs, DASGs and MISGs of stage and sex comparisons. (A–C) Bar plots showing the significantly enriched KEGG pathways associated with stage‐related DEGs, DASGs, and MISGs, indicating key biological processes dynamically regulated across development. (D) Comparison chart illustrating common or unique pathways enriched among stage‐related gene sets, emphasizing divergence in functional consequences of transcriptional and splicing regulation. (E–G) Similar bar plots of enriched pathways for sex‐related DEGs, DASGs, and MISGs at various stages, offering insights into the biological impact of sex‐specific gene regulation. (H) Summary comparison of significantly enriched pathways between male and female embryos, highlighting sex‐dependent functional enrichment patterns.

Next, we carried out functional enrichment analysis for those DEGs, DASGs, and MISGs identified in gender comparisons. The enriched pathways differ greatly among these three types of genes, which show larger differences with that of stage comparisons. The top enriched pathways for sex‐related DEGs are oxidative phosphorylation, protein processing in endoplasmic reticulum, and thermogenesis (Figure 4E), whereas sex‐related DASGs are mainly involved in ubiquitin‐mediated proteolysis, autophagy–animal, and phosphatidylinositol signaling system (Figure 4F). The sex‐related MISGs are mainly enriched in spliceosome, ribosome, cell cycle, and RNA transport (Figure 4G). Surprisingly, no significantly enriched pathways are common among sex‐related DEGs, DASGs, and MISGs, suggesting that these three types of genes mainly function in distinct pathways.

3.5. Functionally Important Regulatory Modules Are Identified in Early Embryonic Development

To explore the gene expression regulations during early embryonic development, we further constructed the gene expression regulatory network in early embryos by employing SCENIC [20]. A total of 506 regulons that the modules are with significant motif enrichment of upstream regulating TFs were identified. Each regulon contains a transcription factor (TF) and the target genes likely regulated by this TF. Among these 506 TFs of regulons, 277 were stage‐related DEGs, 80 were stage‐related DASGs, and 70 were stage‐related MISGs. Most of those TFs were involved in at least one level change of gene expression, alternative splicing, and major isoform switching. Interestingly, the embryonic developmental process from E3 to E7 can be clearly revealed based on those 506 regulons in t‐SNE (t‐Distributed Stochastic Neighbor Embedding), suggesting that the gene regulatory networks vary greatly among different developmental stages (Figure 5A). The male and female cells from the same embryonic stage mixed together in t‐SNE, indicating that male and female embryos share a similar regulatory network in the same stage. In order to identify stage‐specific regulons, we further examined the genes with significantly EEGs in each stage using the method of analysis of variance (ANOVA). With the criteria of fold change > 2 and adjusted p value < 0.01, E3 had the largest number (2977) of genes with EEGs, whereas E6 possessed only 16 EEGs, which was the smallest. For E4, E5, and E7 stages, 530, 50, and 193 EEGs were identified, respectively (Figure 5B). Then we compared these EEGs with those 506 regulons and found that 56, 9, 1, and 9 specific regulons were separately detected in E3, E4, E5, and E7 stages. No specific regulons were identified in the E6 stage, which could be the result of the fact that the gene expression profile in E6 was similar to that of E7 since the smallest amount of DEGs, DASGs, and MISGs were detected in E6 versus E7 compared to other adjacent stage comparisons.

FIGURE 5.

FIGURE 5

Open in a new tab

Gene expression regulatory network of early embryos. (A) t‐SNE plot of human early embryonic cells (E3–E7) based on the activity of 506 regulons, clustering cells by stage and suggesting stage‐specific transcriptional control. (B) Bar graph of representative genes with significantly higher expression at each stage (fold change > 2, adjusted p < 0.05), revealing stage‐specific markers. (C–F) Examples of representative stage‐specific regulons, showing key transcription factors and their target networks that dominate gene regulatory programs at stages E3, E4, E5, and E7, respectively.

The regulon of TF CREB1 is one of the specific regulons in E3 stage, where 602 genes potentially regulated by CREB1 were involved (Figure 5C). CREB1 is a critical TF that regulates ~25% of the eukaryotic genome and plays an important role in the synchronization of circadian rhythmicity, the differentiation of adipose cells, the initial development of the nervous system, memory formation, and neuronal protection [27, 28, 29]. Some studies indicate that CREB1 mediates signals essential for maintaining cell viability during early embryonic development of mouse [30, 31]. Gene functional enrichment analysis showed that those 602 genes were mainly involved in the biological processes of histone modification, covalent chromatin modification, and chromosome segregation (p value < 0.05). The regulon of TF ATF3 and its 461 target genes were mainly expressed in E4 stages (Figure 5D). ATF3 is induced by a variety of signals, including cell cycling, neutrophil migration, and sexual differentiation [32, 33, 34]. Cheng et al. revealed that ATF3 has an important role in regulating human endometrial receptivity and embryo attachment in vitro via upregulation of leukemia inhibitory factor [35]. Those 461 genes were enriched in the biological processes of anion transmembrane transport, response to corticosteroid, and response to glucocorticoid (p value < 0.05). The only one specific regulon of E5 stage contains TF NFKB2 and its 42 target genes (Figure 5E). A central role for the NFKB2 gene is the maintenance of the peripheral B‐cell population, humoral immune responsiveness, and splenic architecture [36]. But no previous studies show the association between NFKB2 and embryonic development, suggesting that NFKB2 is a novel TF specifically expressed in E5 stage. Those 42 targeting genes of NFKB2 were mainly involved in epithelial tube morphogenesis, protein autophosphorylation, and regulation of smooth muscle cell proliferation (p value < 0.05). In E7 stage, TF GCM1 coupled with 38 genes formed a specific regulon (Figure 5F). Proteins encoded by GCM1 are crucial for mediating the differentiation of trophoblast cells along both the villous and extravillous pathways in placental development [37]. Those 38 targeting genes were mainly enriched in leukocyte migration and cell chemotaxis (p value < 0.05).

3.6. The Usage of Transcription Factor Binding Motifs for TFs Is Very Dynamic Along Embryo Development

Since the binding sequences of TFs are usually conserved, we further investigated the transcription factor binding motifs (TFBMs) for those 506 regulons by employing RcisTarget [20]. Interestingly, the counts of enriched TFBMs for those 506 regulons varied tremendously (range from 1 to 2601). It is notable that TF GABPA has the largest number (2601) of TFBMs. GABPA belongs to the E‐twenty‐six (ETS) family of DNA‐binding factors and regulates a broad range of genes involved in cell cycle control, apoptosis, differentiation, hormonal regulation, and other critical cellular functions [38]. Most of the TFs in those regulons possessed 1–50 TFBMs (see Figure S2). Over 91% (461 out of 506) of TFs had at least two TFBSs, suggesting that these TFs exhibit dynamic TFBS usage in early embryonic development. Specifically, the aforementioned stage‐specific TFs CREB1, ATF3, NFKB2, and GCM1 for E3, E4, E5, and E7 stages possessed 6, 223, 48, and 8 TFBMs, respectively. Some of those TFBMs for the stage‐specific TFs are shown in Figure 6A–D. Thus, our results show that the TFBM usage for those 506 TFs of regulons is very dynamic in early embryonic development and distinct TFBMs could be used in different developmental stages.

FIGURE 6.

FIGURE 6

Open in a new tab

Stage‐specific transcription factor binding motifs (TFBMs). (A–D) Representative transcription factor binding motifs (TFBMs) enriched in E3, E4, E5, and E7 stages, respectively, suggesting dynamic shifts in the transcription factor landscape as development progresses.

4. Discussion

In this study, we systematically explored the gene expression profile of human early embryos from the changes of gene‐level (differential expression calling), isoform level (major isoform switching), exon‐level (alternative splicing analysis) and expression regulatory network inference. Intriguingly, a common trend among DEGs, DASGs, and MISGs is that their quantity is gradually decreased along the developmental stages, indicating that the expression changes are reduced during embryonic development. But for the comparison between male and female embryos of each stage, we did not observe a similar trend as stage comparisons. The number of sex‐related DEGs, DASGs, and MISGs detected in different stages varies greatly as well. An interesting result is that only 47 sex‐related DEGs were detected in the E3 stage; however, 1038 DASGs and 1284 MISGs were identified between male and female comparisons of E3. The result indicates that although the expression has changed little between different sexes at the gene level in the E3 stage, great variations existed in alternative splicing and isoform switching of genes. Moreover, the stage‐related DEGs, MISGs, and DASGs had another similar tendency that the majority of each type of genes were distinct among the comparisons of E3–E4, E4–E5, E5–E6, and E6–E7, suggesting that the expression changes varied widely across different levels of gene, isoform, and exon. Gene functional enrichment analysis results showed that those DEGs, DASGs, and MISGs were mainly enriched in different pathways, which is reasonable since the large differences of distinct types of genes. Therefore, combining expression level, alternative splicing, and isoform switching can reveal a broader range of gene expression changes.

Alternative splicing enables the multi‐exon genes to generate different protein‐coding and/or noncoding isoforms, which largely increase the diversity of the transcriptome and can influence both the gene expression level and isoform switching events [39]. When a gene executes a specific role/function in distinct conditions, it may employ disparate isoforms and change the expression levels of corresponding isoforms, resulting in the switching of the major isoform [39]. In fact, the gene expression level reflects the total expression changes of all the isoforms encoded by this gene, whereas isoform switching and alternative splicing separately show the expression variations of isoforms and exons. Therefore, the genes involved in expression changes among the levels of gene, isoform, and exon could vary tremendously. Currently, most scRNA‐seq studies mainly focus on the expression level changes of genes, which miss a large number of genes that significantly change in alternative splicing and isoform switching [40]. Our results indicate that the genes significantly changed in these three aspects are much different in early embryonic development, and they are complementary for elucidating the dynamic expression profile of genes. To the best of our knowledge, our study is the first to simultaneously explore the expression variations from gene expression level, alternative splicing, and isoform switching.

Additionally, we identified 506 functionally important regulons related to early embryonic development, including some stage‐specific regulons (such as E3: MSX1, E4: SOX2, and E7: TFEB), and revealed dynamic usages of TFBMs for those involved TFs. MSX1 encodes transcription factors that control organogenesis and tissue interactions during embryonic development [41]. Systemic deletion of MSX1 in mice results in perinatal lethality [42]. SOX2 is necessary for the maintenance of pluripotency in epiblast and embryonic stem cells, and knockout of SOX2 is early embryonic lethal [43, 44]. Later in development, SOX2 is required in various tissue stem cells and early progenitors, especially in the nervous system [45]. Eiríkur et al. showed that TFEB plays a critical role in the signal transduction processes required for normal vascularization of the placenta [46]. Therefore, the stage‐specific regulons identified by us are functionally important for embryonic development.

In conclusion, our study gains insights into the dynamic changes of gene activities during human early embryonic development from gene expression level, alternative splicing, isoform switching, and gene expression regulatory network. The systematic analyses of gene expression profile from different aspects can deepen the understanding of gene expression variations and the underlying molecular mechanisms.

Author Contributions

Shuang Tang, Jiwei Chen, and Geng Chen conceived and designed the research; Jiwei Chen, Gaigai Wei, and Yunjin Li performed the research; Jiwei Chen, Gaigai Wei, and Yunjin Li analyzed and interpreted the data; Shuang Tang, Jiwei Chen, and Geng Chen wrote the manuscript; Fangge Sun assisted with data analysis, proofreading, and responding to reviewers' comments during the revision; all authors were involved in revising the manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1

FBA2-7-e70036-s001.pdf (198.8KB, pdf)

Figure S2

FBA2-7-e70036-s003.pdf (226.3KB, pdf)

Table S1

FBA2-7-e70036-s002.pdf (229.5KB, pdf)

Acknowledgments

This work was supported by the National Natural Science Foundation of China (grant no. 82404082) and the Science and Technology Commission of Shanghai Municipality (grant no. 24ZR1412000).

Chen J., Wei G., Sun F., Li Y., Tang S., and Chen G., “Single‐Cell Analysis of Alternative Splicing and Gene Regulatory Network Reveals Remarkable Expression and Regulation Dynamics During Human Early Embryonic Development,” FASEB BioAdvances 7, no. 7 (2025): e70036, 10.1096/fba.2025-00075.

Funding: The authors received no specific funding for this work.

Jiwei Chen and Gaigai Wei contributed equally to this study.

Contributor Information

Shuang Tang, Email: tangshuang@fudan.edu.cn.

Geng Chen, Email: gchen@bio.ecnu.edu.cn.

Data Availability Statement

The data that support the findings of this study are available in the Sections 2, 3, and Supporting Information of this article.

References

  • 1. Loo L., Simon J. M., Xing L., et al., “Single‐Cell Transcriptomic Analysis of Mouse Neocortical Development,” Nature Communications 10 (2019): 134. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2. Hammond T. R., Dufort C., Dissing‐Olesen L., et al., “Single‐Cell RNA Sequencing of Microglia Throughout the Mouse Lifespan and in the Injured Brain Reveals Complex Cell‐State Changes,” Immunity 50 (2019): 253–271. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3. Gupta I., Collier P. G., Haase B., et al., “Single‐Cell Isoform RNA Sequencing Characterizes Isoforms in Thousands of Cerebellar Cells,” Nature Biotechnology 36 (2018): 1197–1202. [DOI] [PubMed] [Google Scholar]
  • 4. Paik D. T., Tian L., Lee J., et al., “Large‐Scale Single‐Cell RNA‐Seq Reveals Molecular Signatures of Heterogeneous Populations of Human Induced Pluripotent Stem Cell‐Derived Endothelial Cells,” Circulation Research 123 (2018): 443–450. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5. Davie K., Janssens J., Koldere D., et al., “A Single‐Cell Transcriptome Atlas of the Aging Drosophila Brain,” Cell 174 (2018): 982–998. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6. Sebe‐Pedros A., Saudemont B., Chomsky E., et al., “Cnidarian Cell Type Diversity and Regulation Revealed by Whole‐Organism Single‐Cell RNA‐Seq,” Cell 173 (2018): 1520–1534. [DOI] [PubMed] [Google Scholar]
  • 7. Zhong S., Zhang S., Fan X., et al., “A Single‐Cell RNA‐Seq Survey of the Developmental Landscape of the Human Prefrontal Cortex,” Nature 555 (2018): 524–528. [DOI] [PubMed] [Google Scholar]
  • 8. Chen G., Ning B., and Shi T., “Single‐Cell RNA‐Seq Technologies and Related Computational Data Analysis,” Frontiers in Genetics 10 (2019): 317. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9. Picelli S., Faridani O. R., Bjorklund A. K., Winberg G., Sagasser S., and Sandberg R., “Full‐Length RNA‐Seq From Single Cells Using Smart‐Seq2,” Nature Protocols 9 (2014): 171–181. [DOI] [PubMed] [Google Scholar]
  • 10. Macosko E. Z., Basu A., Satija R., et al., “Highly Parallel Genome‐Wide Expression Profiling of Individual Cells Using Nanoliter Droplets,” Cell 161 (2015): 1202–1214. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11. Yan L., Yang M., Guo H., et al., “Single‐Cell RNA‐Seq Profiling of Human Preimplantation Embryos and Embryonic Stem Cells,” Nature Structural & Molecular Biology 20 (2013): 1131–1139. [DOI] [PubMed] [Google Scholar]
  • 12. Xue Z., Huang K., Cai C., et al., “Genetic Programs in Human and Mouse Early Embryos Revealed by Single‐Cell RNA Sequencing,” Nature 500 (2013): 593–597. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Xue L., Cai J. Y., Ma J., et al., “Global Expression Profiling Reveals Genetic Programs Underlying the Developmental Divergence Between Mouse and Human Embryogenesis,” BMC Genomics 14 (2013): 568. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14. Petropoulos S., Edsgard D., Reinius B., et al., “Single‐Cell RNA‐Seq Reveals Lineage and X Chromosome Dynamics in Human Preimplantation Embryos,” Cell 167 (2016): 285. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Kim D., Langmead B., and Salzberg S. L., “HISAT: A Fast Spliced Aligner With Low Memory Requirements,” Nature Methods 12 (2015): 357–360. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Pertea M., Pertea G. M., Antonescu C. M., Chang T. C., Mendell J. T., and Salzberg S. L., “StringTie Enables Improved Reconstruction of a Transcriptome From RNA‐Seq Reads,” Nature Biotechnology 33 (2015): 290–295. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Kharchenko P. V., Silberstein L., and Scadden D. T., “Bayesian Approach to Single‐Cell Differential Expression Analysis,” Nature Methods 11 (2014): 740–742. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18. Huang Y. and Sanguinetti G., “BRIE: Transcriptome‐Wide Splicing Quantification in Single Cells,” Genome Biology 18 (2017): 123. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19. Yu G., Wang L. G., Han Y., and He Q. Y., “clusterProfiler: An R Package for Comparing Biological Themes Among Gene Clusters,” Omics: A Journal of Integrative Biology 16 (2012): 284–287. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20. Aibar S., Gonzalez‐Blas C. B., Moerman T., et al., “SCENIC: Single‐Cell Regulatory Network Inference and Clustering,” Nature Methods 14 (2017): 1083–1086. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21. Huynh‐Thu V. A., Irrthum A., Wehenkel L., and Geurts P., “Inferring Regulatory Networks From Expression Data Using Tree‐Based Methods,” PLoS One 5 (2010): e12776. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22. Yan G., Zhang Y., Yu J., et al., “Slc39a7/zip7 Plays a Critical Role in Development and Zinc Homeostasis in Zebrafish,” PLoS One 7 (2012): e42939. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23. Cheng Y., Wang K., Geng L., et al., “Identification of Candidate Diagnostic and Prognostic Biomarkers for Pancreatic Carcinoma,” eBioMedicine 40 (2019): 382–393. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24. Morita K., Furuse M., Fujimoto K., and Tsukita S., “Claudin Multigene Family Encoding Four‐Transmembrane Domain Protein Components of Tight Junction Strands,” Proceedings of the National Academy of Sciences of the United States of America 96 (1999): 511–516. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25. Turksen K. and Troy T. C., “Claudin‐6: A Novel Tight Junction Molecule Is Developmentally Regulated in Mouse Embryonic Epithelium,” Developmental Dynamics: An Official Publication of the American Association of the Anatomists 222 (2001): 292–300. [DOI] [PubMed] [Google Scholar]
  • 26. Faraci‐Orf E., McFadden C., and Vogel W. F., “DDR1 Signaling Is Essential to Sustain Stat5 Function During Lactogenesis,” Journal of Cellular Biochemistry 97 (2006): 109–121. [DOI] [PubMed] [Google Scholar]
  • 27. Dworkin S., Heath J. K., deJong‐Curtain T. A., et al., “CREB Activity Modulates Neural Cell Proliferation, Midbrain‐Hindbrain Organization and Patterning in Zebrafish,” Developmental Biology 307 (2007): 127–141. [DOI] [PubMed] [Google Scholar]
  • 28. Impey S., McCorkle S. R., Cha‐Molstad H., et al., “Defining the CREB Regulon: A Genome‐Wide Analysis of Transcription Factor Regulatory Regions,” Cell 119 (2004): 1041–1054. [DOI] [PubMed] [Google Scholar]
  • 29. Zhang X., Odom D. T., Koo S. H., et al., “Genome‐Wide Analysis of cAMP‐Response Element Binding Protein Occupancy, Phosphorylation, and Target Gene Activation in Human Tissues,” Proceedings of the National Academy of Sciences of the United States of America 102 (2005): 4459–4464. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30. Jin X. L. and O'Neill C., “cAMP‐Responsive Element‐Binding Protein Expression and Regulation in the Mouse Preimplantation Embryo,” Reproduction 134 (2007): 667–675. [DOI] [PubMed] [Google Scholar]
  • 31. Bleckmann S. C., Blendy J. A., Rudolph D., Monaghan A. P., Schmid W., and Schutz G., “Activating Transcription Factor 1 and CREB Are Important for Cell Survival During Early Mouse Development,” Molecular and Cellular Biology 22 (2002): 1919–1925. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32. Liu B., Agras K., Willingham E., Vilela M. L., and Baskin L. S., “Activating Transcription Factor 3 Is Estrogen‐Responsive in Utero and Upregulated During Sexual Differentiation,” Hormone Research 65 (2006): 217–222. [DOI] [PubMed] [Google Scholar]
  • 33. Boespflug N. D., Kumar S., McAlees J. W., et al., “ATF3 Is a Novel Regulator of Mouse Neutrophil Migration,” Blood 123 (2014): 2084–2093. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Thompson M. R., Xu D., and Williams B. R., “ATF3 Transcription Factor and Its Emerging Roles in Immunity and Cancer,” Journal of Molecular Medicine 87 (2009): 1053–1060. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35. Cheng X., Liu J., Shan H., et al., “Activating Transcription Factor 3 Promotes Embryo Attachment via Up‐Regulation of Leukemia Inhibitory Factor In Vitro,” Reproductive Biology and Endocrinology: RB&E 15 (2017): 42. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36. Heusch M., Lin L., Geleziunas R., and Greene W. C., “The Generation of nfkb2 p52: Mechanism and Efficiency,” Oncogene 18 (1999): 6201–6208. [DOI] [PubMed] [Google Scholar]
  • 37. Baczyk D., Drewlo S., Proctor L., Dunk C., Lye S., and Kingdom J., “Glial Cell Missing‐1 Transcription Factor Is Required for the Differentiation of the Human Trophoblast,” Cell Death and Differentiation 16 (2009): 719–727. [DOI] [PubMed] [Google Scholar]
  • 38. Perdomo‐Sabogal A., Nowick K., Piccini I., et al., “Human Lineage‐Specific Transcriptional Regulation Through GA‐Binding Protein Transcription Factor Alpha (GABPa),” Molecular Biology and Evolution 33 (2016): 1231–1244. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39. Chen G., Shi T., and Shi L., “Characterizing and Annotating the Genome Using RNA‐Seq Data,” Science China. Life Sciences 60 (2017): 116–125. [DOI] [PubMed] [Google Scholar]
  • 40. Peng G. and Jing N., “The Genome‐Wide Molecular Regulation of Mouse Gastrulation Embryo,” Science China. Life Sciences 60 (2017): 363–369. [DOI] [PubMed] [Google Scholar]
  • 41. Alappat S., Zhang Z. Y., and Chen Y. P., “Msx Homeobox Gene Family and Craniofacial Development,” Cell Research 13 (2003): 429–442. [DOI] [PubMed] [Google Scholar]
  • 42. Satokata I. and Maas R., “Msx1 Deficient Mice Exhibit Cleft Palate and Abnormalities of Craniofacial and Tooth Development,” Nature Genetics 6 (1994): 348–356. [DOI] [PubMed] [Google Scholar]
  • 43. Masui S., Nakatake Y., Toyooka Y., et al., “Pluripotency Governed by Sox2 via Regulation of Oct3/4 Expression in Mouse Embryonic Stem Cells,” Nature Cell Biology 9 (2007): 625–635. [DOI] [PubMed] [Google Scholar]
  • 44. Avilion A. A., Nicolis S. K., Pevny L. H., Perez L., Vivian N., and Lovell‐Badge R., “Multipotent Cell Lineages in Early Mouse Development Depend on SOX2 Function,” Genes & Development 17 (2003): 126–140. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45. Pevny L. H. and Nicolis S. K., “Sox2 Roles in Neural Stem Cells,” International Journal of Biochemistry & Cell Biology 42 (2010): 421–424. [DOI] [PubMed] [Google Scholar]
  • 46. Steingrimsson E., Tessarollo L., Reid S. W., Jenkins N. A., and Copeland N. G., “The bHLH‐Zip Transcription Factor Tfeb Is Essential for Placental Vascularization,” Development 125 (1998): 4607–4616. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Figure S1

FBA2-7-e70036-s001.pdf (198.8KB, pdf)

Figure S2

FBA2-7-e70036-s003.pdf (226.3KB, pdf)

Table S1

FBA2-7-e70036-s002.pdf (229.5KB, pdf)

Data Availability Statement

The data that support the findings of this study are available in the Sections 2, 3, and Supporting Information of this article.

Articles from FASEB BioAdvances are provided here courtesy of Wiley

RESOURCES