Comparative single-nucleus transcriptomics reveals asymmetric evolution of the Drosophila male and female germlines
Plos.org·July 20, 2026
AI Summary
Researchers conducted single-nucleus transcriptomic analysis on germline tissues from two Drosophila species to understand evolutionary changes in reproductive organs. The study reveals that evolutionary divergence occurs asymmetrically between male and female germlines, with specific cell types showing differential rates of genetic change across species.
Reproductive organs vary widely across species yet share conserved cell types that produce gametes, sustaining species’ perpetuation. However, tissue-level comparisons mask critical differences among cell types, obscuring where evolutionary divergence occurs even between closely related species. We quantified expression divergence at cell-type resolution between two sibling species, Drosophila melanogaster and D. simulans, while disentangling adaptive and nonadaptive evolutionary mechanisms. We built a comparative single-nucleus transcriptomic atlas of over 100,000 nuclei from testes and ovaries of both species. Our analysis revealed sharply heterogeneous divergence across testis cell types, contrasting with a broader conservation across ovary cell types. Notably, in both organs, ~40% of genes showing interspecific differences did so in only one cell type. In the testis, spermatogonia were largely conserved, whereas divergence peaked in primary spermatocytes with extensive rewiring of coexpression modules linked to microtubule and mitochondrial functions. In the ovary, expression was largely conserved, except in early germline and late follicle cells, which showed shifts in oogenesis and cell-cycle-related coexpression modules. Divergent cell types in both tissues were enriched for evolutionarily young genes with narrow expression breadth and faster protein evolution rates. Additionally, the ovary exhibited a faster-X effect consistent with adaptive evolution. These findings reveal a fundamental asymmetry in how male and female germlines evolve, with functional constraints relaxed in specific testis cell types but broadly maintained across the ovary. Our work provides an evolutionary framework explaining how core reproductive functions are safeguarded during species diversification while identifying germline cells that drive evolutionary change.
Citation: Hariyani IE, Das S, Le EM, Gamero-Castano C, Soroudi T, Choi J, et al. (2026) Comparative single-nucleus transcriptomics reveals asymmetric evolution of the Drosophila male and female germlines. PLoS Biol 24(7): e3003869. https://doi.org/10.1371/journal.pbio.3003869
Full story reconstructed from Plos.org. Formatting and media may differ from the original.
Data Availability: Raw sequencing data, demultiplexed by sample for each sublibrary sequenced, were deposited as part of the NCBI BioProject PRJNA1273697. Further results are available in the supplementary files (Supporting information), and the processed expression data are browsable in a web app at https://ranzlab-server.bio.uci.edu/Drosophila_snRNA-seq_Gene_Expression_Browser_Testis/ and https://ranzlab-server.bio.uci.edu/Drosophila_snRNA-seq_Gene_Expression_Browser_Ovary/. The custom scripts used for this study are available at Zenodo [147] (https://doi.org/10.5281/zenodo.16622709).
Funding: This work was supported by a National Science Foundation (https://www.nsf.gov/) Award to J.M.R (MCB-2129845), and a National Institutes of Health (https://www.nih.gov/) grant to V.S. (5R01AG071683-03). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Abbreviations: ART, aligned rank transform; CC, cyst cells; C12, choriogenic MBFCs stage 12; C14, choriogenic MBFCs stage 14; DE, differentially expressed; DEGs, differentially expressed genes; DME, differential module eigengene; EC, epithelial cells; EFC, early follicle cells; ES, early spermatogonia; EpS, epithelium somatic; ESp, early spermatocytes; FCA, Fly Cell Atlas; FSC/preFCs, follicle stem cells and pre-follicle cells; G, germline; GO, gene ontology; GS, germarium somatic; G2a-2b, germarium region 2a and 2b cells; G2b-3, germarium region 2b and 3 cells; GSC/G1-2a, germline stem cells and germarium region 1 and 2a cells; HC, hub cells; LS, late spermatogonia; LSp, late spermatocytes; MBFCs, main-body follicle cells; Me, meiotic/post-meiotic; MEs, module eigengenes; MFC, mitotic follicle cells stage 1-5; Mi, mitotic; MSp, mid spermatocytes; MPSp, maturing primary spermatocytes; MSCI, meiotic sex chromosome inactivation; O, oviduct; OSM, ovarian sheath muscle; PCA, Principal component analysis; PMFC, post-mitotic follicle cells stage 6; ROIs, regions of interest; S, somatic; S, spermatids; SC, stretch cells; SPC, stalk and polar cells; TCLC, terminal corpus luteum cells; TE, transposable element; TOM, topological overlap matrix; UMAP, uniform manifold approximation and projection.; Un, unannotated; V7, vitellogenic main-body follicle cells (MBFCs) stage 7; V8, vitellogenic MBFCs stage 8; V9-10A, vitellogenic MBFCs stage 9-10A.
Gene expression changes are a key driver of species adaptation and phenotypic diversification [1–4]. While early comparative transcriptomic studies using microarrays or bulk RNA-sequencing helped characterize large-scale expression differences at the tissue-level, they lacked cell type resolution and were confounded by differences in tissue composition between species [5,6]. Single-cell and single-nucleus RNA sequencing (scRNA-seq and snRNA-seq, respectively) overcome these limitations by enabling direct comparisons of homologous cell types at unprecedented granularity within complex tissues [7,8]. These approaches have led to more precise evolutionary inferences—for example, in mammalian tissues such as the testis [5,9] and various regions of the brain [10,11]—by clarifying which cell types are most conserved versus those contributing disproportionately to expression divergence.
The gonads are central to understanding expression evolution, as they harbor most sex-biased genes [12] and exhibit distinct patterns of gene expression [13–15]. Drosophila melanogaster has served as a premier model organism for studying gametogenesis, primarily using bulk transcriptomic data, often derived from whole-body samples [16–21]. More recently, single-cell studies in Drosophila have revealed stage- and cell-type-specific expression patterns across both testis and ovary [7,22–30]. For example, DNA damage response genes are enriched in early germline stages of both gonadal tissues [22,27], and late spermatogenesis is marked by an enrichment of de novo genes [22]. However, reliance on single-strain datasets has limited our ability to capture the intraspecific genetic diversity of D. melanogaster [31–33], which is crucial for understanding functional constraints and identifying expression changes that truly contribute to species divergence [34,35]. Key questions remain unresolved: which gonadal cell types drive interspecific divergence? Do the testis and ovary follow parallel evolutionary trajectories, or do their constraints and modes of evolution fundamentally differ? What coexpression networks and functional pathways underlie this divergence? And do chromosomal location or genomic features explain putative cell-type-specific patterns?
Despite evolutionary divergence, gametogenesis in Drosophila and mammals shares conserved developmental features, including similar differentiation stages and core molecular pathways [36–38]. In mammals, transcriptome divergence is not uniform across the testis: it peaks in meiotic and post-meiotic germ cells of the testis, while early germline cells remain highly conserved—likely due to strong functional constraints tied to their pleiotropic roles [9,39,40]. In the mammalian ovary, the theca cells—endocrine somatic cells that surround the follicle—accumulate greater expression divergence between species compared to other ovarian somatic cell types [41]. Whether Drosophila exhibits similar patterns of stage- or cell-type-specific divergence in either gonadal tissue remains unknown. Comparative single-cell analyses of closely related Drosophila species remain scarce, making it difficult to discern whether previously reported expression divergence reflects true biological differences or methodological artifacts such as confounding or compensatory effects [13,16–18]. If substantial interspecific expression differences exist in Drosophila, it is unclear whether they are broadly distributed across cell types or concentrated in terminally differentiated stages, similar to patterns observed in the mammalian testis.
To address these questions, we used snRNA-seq to compare the testis and ovary transcriptomes of three strains spanning two closely related species, D. melanogaster and D. simulans, which diverged ∼1.4 mya [42]. We characterized expression divergence across cell types by integrating two complementary analyses: differential expression at the level of individual genes and network divergence at a systems level. We also examined how differentiation trajectories in spermatogenesis and oogenesis relate to pleiotropic constraint, using proxies such as expression breadth and phylogenetic gene age. Our results reveal a dichotomy in the tempo and mode of transcriptome evolution between the male and female germline. Testis transcriptomes evolve rapidly and in a highly cell-type-specific manner, with divergence in meiotic cells driven by a combination of adaptive and nonadaptive mechanisms. In contrast, the ovary exhibits more conserved expression properties, consistent with stronger pleiotropic constraint—except in a subset of germline and late-stage follicle cells, which show pronounced divergence. We further show that the cell types accumulating divergent genes also tend to have narrow expression breadth and younger phylogenetic age, especially in the testis, and that divergent genes are more likely to be on the X chromosome in the ovary. Together, these findings uncover how distinct evolutionary mechanisms shape gene expression programs across cell types in the male and female germlines.
To characterize gene expression divergence between closely related Drosophila species at cell-type resolution, we performed single-nucleus RNA-sequencing (Parse Biosciences Evercode v2 [43]) on gonadal tissues from two strains of Drosophila melanogaster (A4, ISO1) and one strain of D. simulans (w501), selected to capture intra- and interspecific divergence within the D. melanogaster species subgroup. We profiled testes from 1 to 2-day-old naïve males and ovaries from 3–day-old virgin females using split-pool combinatorial barcoding [43]. After quality control and doublet removal, we retained 106,328 high-quality nuclei–40,274 nuclei from testis and 66,054 nuclei from ovary–across 12 biological samples (3 strains × 2 tissues × 2 biological replicates). To minimize mapping artifacts due to nucleotide differences, we aligned sequencing reads to strain-specific genome assemblies. We then integrated cross-strain data for each tissue using Harmony [44], which best preserved cell-type and species structure among the tools tested (Methods). With an average of 44,742 reads per nucleus (S1 Table), our dataset is comparable to or exceeds those of previous studies (S2 and S3 Tables), enabling high-resolution comparisons of conserved and divergent expression patterns within and between species.
We identified conserved germline and somatic cell populations in both testis and ovary, whose proportions and marker expression were highly conserved across all three strains. Using uniform manifold approximation and projection (UMAP) and graph-based clustering, we found 10 testis and 17 ovary cell types and annotated them using known marker genes (S4 Table). In the testis, we identified cells undergoing mitosis (Mi; GSC/early spermatogonia (ES) and late spermatogonia (LS)) and meiosis (Me; early spermatocytes (ESp), mid spermatocytes (MSp), late spermatocytes (LSp), maturing primary spermatocytes (MPSp), and post-meiotic spermatids (S)), along with three somatic populations (S; epithelial (EC), cyst (CC), and hub (HC) cells). In the ovary, we distinguished early germline cells in the germarium (G; germline stem cells and germarium region 1 and 2a (GSC/G1-2a), germarium region 2a and 2b (G2a-2b), and germarium region 2b and 3 cells (G2b-3)) from somatic cells in the germarium (GS; follicle stem cells (FSC/pre-FCs), early follicle cells (EFC), and stalk and polar cells (SPC)), and multiple somatic populations in the epithelium (EpS; main body follicle cells (V7, V8, V9-10A, C12, C14), oviduct (O), ovarian sheath muscle (OSM), stretch cells (SC), and terminal corpus luteum cells (TCLC)). These somatic cells play essential roles in stem cell maintenance, germ cell differentiation and survival, and sex-specific germline development [45–47]. UMAP visualization faithfully recapitulated the expected spatial arrangement of these cell types, despite the absence of spatial data (Fig 1A and 1B). We validated these annotations using developmental trajectory reconstruction, which recovered the expected progression of germline differentiation (S1 and S2 Figs).
UMAP visualization of transcriptomic profiles from the (A) testis and (B) ovary, respectively, colored by annotated cell types. Each cluster represents a distinct cell type that corresponds to a particular stage during spermatogenesis and oogenesis, respectively, combined for all three strains (above) and individually in D. melanogaster strains A4 and ISO1, and D. simulans strain w501 (below). Stacked bar plot showing the relative fraction of identified cell types for each strain in the (C) testis and (D) ovary, respectively. Stacked bar plot showing the proportion of all cells annotated for each identified cell type across the three strains in the (E) testis and (F) ovary, respectively. All annotated cell types were included after quality control filtering, regardless of number of nuclei. Testis cell types: EC, epithelial cells; HC, hub cells; CC, cyst cells; ES, germline stem cells and early spermatogonia; LS, late spermatogonia; ESp, early spermatocytes; MSp, mid spermatocytes; LSp, late spermatocytes; MPSp, maturing primary spermatocytes; and S, spermatids. Ovary cell types: GSC/G1-2a, germline stem cells and germarium region 1 and 2a cells; G2a-2b, germarium region 2a and 2b cells; G2b-3, germarium region 2b and 3 cells; SPC, stalk and polar cells; FSC/preFCs, follicle stem cells and pre-follicle cells; EFC, early follicle cells; MFC, mitotic follicle cells stage 1-5; PMFC, post-mitotic follicle cells stage 6; V7, vitellogenic main-body follicle cells (MBFCs) stage 7; V8, vitellogenic MBFCs stage 8; V9-10A, vitellogenic MBFCs stage 9-10A; C12, choriogenic MBFCs stage 12; C14, choriogenic MBFCs stage 14; TCLC, terminal corpus luteum cells; SC, stretch cells; OSM, ovarian sheath muscle; and O, oviduct. Un, unannotated.
https://doi.org/10.1371/journal.pbio.3003869.g001
The testis and ovary cell types displayed strong signatures of evolutionary stability. The different strains exhibited similar overall contributions to each cell type (Fig 1C and 1D), with minimal differences in relative cell type composition (Figs 1E, 1F, and S3; S5 Table and S1 Text). This resulted in highly correlated cell proportions across strains (S6 Table). Furthermore, marker gene expression remained remarkably consistent across strains for all annotated cell types (S4 and S5 Figs). Principal component analysis revealed that expression profiles clustered by species and by broad cell type category—mitotic, meiotic/post-meiotic, and somatic in the testis, and along a developmental trajectory continuum in the ovary (Fig 2A and 2B). Biological replicates consistently clustered together, and the two strains of D. melanogaster always grouped more closely with each other than with D. simulans, reflecting their phylogenetic relatedness and confirming the validity of our sample preparation and sequencing pipeline. Together, these findings indicate the conserved developmental architecture of gametogenesis, enabling cell-type-level comparisons of gene expression evolution across species.
Principal component analysis (PCA) of cell-type pseudo-bulks. (A) Testis. PC1 separates the mitotic cells from the meiotic and somatic cells. PC2 shows separation between species. (B) Ovary. PC1 shows the separation between species. PC2 shows cell progression along the oogenesis trajectory. Correlation matrices depicting Pearson’s correlation coefficients between the expression levels of genes found expressed in common between any two cell types across the three strains assayed in (C) testis and (D) ovary. A4 and ISO1, D. melanogaster strains; w501, D. simulans strain. In this correlation analysis, only cell types with at least 50 cells per strain and 350 cells across strains were included; as a result, the TCLC and SC ovarian cell types were excluded. Genes were required to be expressed in at least 1% of cells with a minimum average expression of 0.01. The order of cell types along both axes is identical, with broadly defined and precise cell types indicated on the x-axis and y-axis, respectively. Broad categories: somatic (S), mitotic (Mi), and meiotic (Me) categories for testis; and germline (G), germarium somatic (GS), and epithelium somatic (EpS) for ovary. Testis cell types: EC, epithelial cells; HC, hub cells; CC, cyst cells; ES, germline stem cells and early spermatogonia; LS, late spermatogonia; ESp, early spermatocytes; MSp, mid spermatocytes; LSp, late spermatocytes; MPSp, maturing primary spermatocytes; and S, spermatids. Ovary cell types: GSC/G1-2a, germline stem cells and germarium region 1 and 2a cells; G2a-2b, germarium region 2a and 2b cells; G2b-3, germarium region 2b and 3 cells; SPC, stalk and polar cells; FSC/preFCs, follicle stem cells and pre-follicle cells; EFC, early follicle cells; MFC, mitotic follicle cells stage 1-5; PMFC, post-mitotic follicle cells stage 6; V7, vitellogenic main-body follicle cells (MBFCs) stage 7; V8, vitellogenic MBFCs stage 8; V9-10A, vitellogenic MBFCs stage 9-10A; C12, choriogenic MBFCs stage 12; C14, choriogenic MBFCs stage 14; TCLC, terminal corpus luteum cells; SC, stretch cells; OSM, ovarian sheath muscle; and O, oviduct.
https://doi.org/10.1371/journal.pbio.3003869.g002
The conserved cellular architecture of the gonads extended to the number of genes expressed per cell type, although we noted sharp within-tissue variation (S6 Fig). In the testis, late and maturing primary spermatocytes (LSp and MPSp, respectively), exhibited the highest median number of expressed genes, whereas mitotic cell types (ES and LS) had the fewest (Kruskal-Wallis chi-squared, P < 0.001 for each strain; S7 Table for post-hoc tests). In the ovary, germline stem cells and stage 1 and 2a cells in the germarium (GSC/G1-2a), and stage 12 choriogenic MBFCs (C12) expressed the highest median number of genes (Kruskal-Wallis chi-squared, P < 0.001 for each strain; S7 Table for post-hoc tests). These patterns remained stable across all three strains (S6 Fig).
Testis and ovary cell types displayed distinct trajectories of transcriptome differentiation. We quantified this differentiation at three levels: within strain, within species (A4 versus ISO1), and between species (D. melanogaster versus D. simulans). To do so, we calculated the average expression of each expressed gene for each gonadal cell type in each strain (S1 and S2 Data) and constructed gene expression correlation matrices for each tissue (Fig 2C and 2D). Within strains, median correlation values among broad cell-type categories differed significantly in both the testis and ovary, although this variation was larger in the testis (S7 Fig). This pattern indicates that transcriptome differentiation is more cell-type-specific in the testis than in the ovary (S8 Table). In the testis, median correlation values were significantly higher among mitotic cell types (Mi-Mi), among meiotic cell types (Me-Me), and between post-meiotic and meiotic categories (PMe-Me) (Kruskal-Wallis rank sum test; S7 Fig; S8 Table). In the ovary, the highest correlation values were found among germline cell types (G-G) (Kruskal-Wallis rank sum test; S7 Fig; S8 Table). When all pairwise cell type comparisons within each tissue were examined, within-strain median correlation values were significantly lower in the testis than in the ovary (two-way aligned rank transform (ART) ANOVA; S8 Fig; S9 Table).
Between strains, median expression correlations were also significantly higher among ovary cell types than among testis cell types, both within and between species (two-way ART ANOVA; S9 Fig; S10 Table). A closer examination by tissue revealed that the testis harbored the widest range of expression correlation values (min = 0.31, somatic cells, A4 versus w501; max = 0.94, meiotic cells, A4 versus ISO1). Among broad cell-type categories, meiotic cells exhibited the largest and the only significant difference between intra- and interspecific contrasts. In contrast, mitotic cell types–germline stem cells/ early spermatogonia (ES) and late spermatogonia (LS)–displayed similar median correlation values within and between species, suggesting a more conserved early transcriptional program. Somatic cell populations exhibited significantly lower correlation values than all other broad cell categories (two-way ART ANOVA; S10 Fig; S11 Table). By contrast, the ovary featured high intra- and interspecific correlation values (>0.75) across all broad cell-type categories, indicating more conserved expression patterns across both germline and somatic categories. Within each broad cell-type category, intraspecific median correlation values were significantly higher than interspecific ones (Padj < 0.05). However, for any given pair of strains, median correlation values did not differ significantly among broad cell-type categories (two-way ART ANOVA; S11 Fig; S12 Table).
Together, these results show that expression correlations are highest within strain, lower within species, and lowest between species, but that this decline is highly tissue- and stage-specific. The ovary is more uniformly conserved, whereas the testis spans a much wider range of transcriptome differentiation, with meiotic cells showing the largest differences between intra- and interspecific comparisons.
To identify the genes driving patterns of expression correlation within and between species, we performed differential expression analyses, focusing on 11,481 one-to-one orthologous genes [48]. These analyses included 9,218 genes expressed in testis cell types and 8,533 genes expressed in ovary cell types, with 8,084 genes expressed in both (S3 and S4 Data). We defined differentially expressed genes (DEGs) using a 1% FDR and a log2 fold-change ≥ |1|, based on pairwise comparisons at both the intraspecific (ISO1 versus A4) and interspecific (ISO1 versus w501, and A4 versus w501) levels for each cell type and tissue. Consistent with previous work on the interspecific evolution of sex-biased expression [16,17,20,21], the testis showed greater expression divergence than the ovary (3,253 DEGs or 35.3% versus 1,748 DEGs or 20.5% of the expressed genes in each tissue, respectively; 2-sample test for equality of proportions with continuity correction or 2STEP, χ2 = 479.2, d.f. = 1, P < 2.2 × 10−16) (Fig 3A and 3E).
Venn diagrams showing the number of differentially expressed genes (DEGs) unique to and shared between D. melanogaster and D. simulans in the testis: (A) at the inter- and intraspecific levels; and (B) across the three pairwise comparisons involving the D. melanogaster strains A4 and ISO1, and the D. simulans strain w501. (C) Patterns of differential expression across pairwise comparisons and testis cell types. Strict Inter (pattern 110): DEGs in both interspecific contrasts in the same cell type without being differentially expressed at the intraspecific level. Strict Intra (pattern 001): DEGs at the intraspecific level without being differentially expressed at the interspecific level. (D) Percentage of DEGs per testis cell type between the D. melanogaster strains A4 and ISO1, shown relative to the D. simulans strain w501. DEGs are categorized based on whether they show consistent differential expression in both interspecific comparisons (A4 vs. w501 and ISO1 vs. w501) or are unique to only one of the comparisons. (E–H) Equivalent plots for the ovary. Genes were required to be expressed in at least 1% of cells with a minimum average expression of 0.01. Asterisks, cell types showing the lowest and highest deviation relative to the random expectation for DEGs consistent across both interspecific contrasts (S14 Table). Testis cell types: EC, epithelial cells; HC, hub cells; CC, cyst cells; ES, germline stem cells and early spermatogonia; LS, late spermatogonia; ESp, early spermatocytes; MSp, mid spermatocytes; LSp, late spermatocytes; MPSp, maturing primary spermatocytes; and S, spermatids. Ovary cell types: GSC/G1-2a, germline stem cells and germarium region 1 and 2a cells; G2a-2b, germarium region 2a and 2b cells; G2b-3, germarium region 2b and 3 cells; SPC, stalk and polar cells; FSC/preFCs, follicle stem cells and pre-follicle cells; EFC, early follicle cells; MFC, mitotic follicle cells stage 1-5; PMFC, post-mitotic follicle cells stage 6; V7, vitellogenic main-body follicle cells (MBFCs) stage 7; V8, vitellogenic MBFCs stage 8; V9-10A, vitellogenic MBFCs stage 9-10A; C12, choriogenic MBFCs stage 12; C14, choriogenic MBFCs stage 14; TCLC, terminal corpus luteum cells; SC, stretch cells; OSM, ovarian sheath muscle; and O, oviduct.