Cross-species single-cell transcriptomic atlas of female fetal gonadal development

(A) Schematic overview of the study design and developmental stages analyzed. Single-cell RNA sequencing was performed on bovine fetal gonadal tissues. Human and mouse fetal gonadal single-cell RNA-seq datasets were obtained from publicly available databases. Samples span key developmental stages including sex determination, mitotic expansion, entry into meiosis, and early oocyte differentiation. Corresponding embryonic or post-conception stages are indicated for each species. (B) UMAP showing clustering of major cell types in the integrated dataset. Twelve cell types were identified in total, 11 shared across species and one specific to cattle. PV indicates perivascular cells. Bar plots show the relative proportion of major cell types across developmental time points for each species, illustrating dynamic changes in cellular composition during ovarian development. (C) Feature plots showing the expression of canonical germ cell and granulosa cell markers projected onto the integrated UMAP. DAZL and DDX4 mark germ cells, while KITLG and AMHR2 mark granulosa cells. (D) UMAPs showing cell population distributions across three species and developmental time points.

Cross-species dynamics of epigenetic regulators during early germ cell development

(A) Expression dynamics of DNA methylation–related regulators (DNMT1, HELLS, TET1, UHRF1) across developmental stages in bovine (E38-E112), human (PCW6–PCW16) and mouse (E11.5–E18.5) germ cells. (B) Expression dynamics of chromatin remodeling and histone modification–related regulators (EZH2, KDM3A, KDM6B, SMARCA1, SUZ12) across developmental stages in bovine (E38–E112), human (PCW6–PCW16), and mouse (E11.5–E18.5) germ cells. Violin plots show the distribution of log2(TPM + 1) expression levels for each gene at each developmental stage, with embedded boxplots indicating median and interquartile range. Colors represent developmental stages as indicated. Species are denoted by icons on the left (bovine, human, and mouse).

Conserved and species-specific transcriptional dynamics during early germ cell development across species

(A) Heatmaps showing gene expression dynamics along pseudotime in germ cells from bovine, human, and mouse. Cells are ordered from early to late pseudotime, and gene expression is scaled by gene (z-score). Pluripotency-associated genes and meiosis-related genes are highlighted, illustrating a progressive downregulation of pluripotency programs and activation of meiosis-associated transcriptional programs during germ cell development across species. (B) Venn diagram summarizing the overlap of dynamically expressed genes identified along pseudotime in bovine, human, and mouse germ cells. (C) Gene Ontology (GO) enrichment analysis of species-specific dynamic genes identified along pseudotime. (D) Transcription factor activity analysis inferred by pySCENIC. Dot size represents scaled expression levels, and color intensity indicates scaled regulon activity (AUC scores).

Conserved and species-specific transcriptional dynamics during early granulosa cell development across species

(A) Heatmaps showing gene expression dynamics along pseudotime in granulosa cells from bovine, human, and mouse. Cells are ordered from early to late pseudotime, and gene expression is scaled by gene (z-score). Representative genes associated with early supporting states, signaling pathways, and granulosa cell differentiation are highlighted, revealing progressive transcriptional changes during granulosa cell development across species. (B) Venn diagram summarizing the overlap of dynamically expressed genes identified along granulosa cell pseudotime in bovine, human, and mouse. (C) Gene Ontology (GO) enrichment analysis of species-specific dynamic genes identified along granulosa cell pseudotime. (D) Transcription factor activity analysis inferred by pySCENIC. Dot size represents scaled expression levels, and color intensity indicates scaled regulon activity (AUC scores).

Cross-species cell–cell communication networks involving germ cells

(A) Global cell-cell communication networks inferred from single-cell transcriptomes of female fetal gonads in bovine (left), human (middle), and mouse (right). Nodes represent annotated cell types, and edges represent inferred ligand–receptor–mediated interactions. Edge thickness reflects the overall strength of communication between cell types. (B) Dot plots showing inferred ligand–receptor interactions with germ cells specified as the source cell population. Each panel summarizes outgoing signaling from germ cells to major somatic cell types across species. Dot size and color indicate the communication probability for each ligand–receptor pair. (C) Dot plots showing inferred ligand-receptor interactions with germ cells specified as the target cell population. Incoming signaling from surrounding somatic cell types, including endothelial, epithelial, granulosa, mesothelial, and PV cells, is shown across species. Dot size and color represent communication probability.

Cross-species comparison of cell types using a support vector machine (SVM)–based classifier

(A) Distribution of predicted classification probabilities for each cell type when a human-trained SVM classifier was applied to the mouse dataset. Each point represents an individual cell, and boxplots summarize the distribution of prediction probabilities for each annotated cell type. Higher prediction probabilities indicate greater transcriptional similarity to the corresponding human cell type. The vertical dashed red line indicates a reference probability threshold. (B) Distribution of predicted classification probabilities for each cell type when the human-trained SVM classifier was applied to the bovine dataset. Visualization and interpretation are as in panel (A). (C) UMAP visualization of representative cell types illustrating cross-species similarity and divergence. Mesenchymal cells show substantial overlap across species, whereas granulosa cells display increased separation, consistent with reduced cross-species classification confidence. (D) Species-specific transcriptional features of granulosa cells identified by SVM. An additional SVM classifier was trained to distinguish granulosa cells from bovine, human, and mouse. The top weighted genes contributing to species-specific classification are shown, with positive and negative weights indicating relative enrichment in each species.