Fig 1.
Male-specific scales in the dorsal wings of Colias eurytheme.
A. Dorsal views of male and female butterflies, highlighting the difference in width and patterning of the marginal bands, and the male-specific UV-iridescence (magenta; an UV-photograph was colored in magenta and overlaid on a normal image). Alba morphs are a female-limited phenotype characterized by pterin-pigment deficiency in their scales. B, C. Scanning electron micrographs of representative non-dimorphic scales (here in B, left: female dorsal forewing melanic scales of the discal spot; right: male ventral orange scales), and male-specific scales (here in C: interface between UVI and spatulate scales, dorsal hindwing). D. Mean apical ridge distance in individual scales from melanic discal spots, medial regions, or male marginal bands, highlighting the derived ultrastructure of male-specific UVI and spatulate scales. Each category features N = 25 scales—except for spatulate scales (N = 20). The measurements originate from electron micrographs of dorsal wing surfaces in a total of three individuals: an orange female, an Alba female, and a male (Supplementary Methods in S1 Text and Data 10 at https://osf.io/yjvkc/). E–H. Apical views of the ultrastructure of representative scale types (same scale types as in panels B, C), featuring the longitudinal ridges (horizontal structures) and the transversal microribs (vertical). E: female dorsal cover black scale; F: male ventral cover scale (orange, non-UVI); G: male UVI dorsal cover orange scale; H: male spatulate dorsal cover margin scale. I. Position of male-specific cell types on the dorsal wings. The cell bodies of scale precursors (blue) are transient structures that do not occur in adults. J. Schematic representation of the ultrastructures (viewed as transversal cross-sections) from the four main scale types—canonical (orange and black), UVI, and spatulate. Scale bars; B and C = 50 μm; D = 1 mm; E–H = 2 μm.
Fig 2.
DsxF blocks UV-iridescence in females in a cell-autonomous fashion.
A. Effects of Dsx mosaic knock-outs (mKO), as seen in comparisons of female and male WT individuals with representative Dsx G0 crispants. The sex of each crispant was determined by concordant genital morphology and genotyping. Phenotypic effects are limited to the dorsal surfaces and most visible in the marginal region. B. UV-photography (320–400 nm) of UV-iridescent dorsal patterns in the same mutants, showing an ectopic gain of UVI scales in females (top). In males (bottom), the feminization of the marginal region triggers a regression of the male UVI pattern distal border. C–C′. Close-up views of the central forewing regions (same individuals as in panels A, B), including overlays of UV photographs (C′: magenta false-color) over visible light images, and highlighting the intermediate states of marginal patterns. In females: gain of UVI scales, regression of the melanic marginal patterns. In males: gain of female yellow spots in males (arrowheads), extension of the melanic marginal patterns (white arrow), and regression of the UVI field (black arrow). D. Cell autonomy of UVI scale gains in female Dsx mKOs, as shown by continuous UVI mutant clones with sharp boundaries (KO). Superimposed views of the same region, taken in visible and ultraviolet light, are shown across each diagonal line. E. RNA interference effects of Dsx siRNA electroporation in male and female forewings. Dotted lines mark the approximate areas that were electroporated. Dsx knockdown results in ectopic UVI (top), and in a feminization of the marginal pattern in males, including with a regression of medial UV iridescence at its distal border. F. High-magnification view of the female wing shown in panel E, at the interface of the treated (ectopic UVI scales) and untreated area. Scale bars: C, E = 1 mm; D = 100 μm.
Fig 3.
DsxF medial expression activates Bab in the female dorsal cover scales.
A, B. High-magnification views of central dorsal wings, showing dorsal cover-scale (dcs) specific transformation of non-UVI to UVI scales in female Dsx KOs. Males are unaffected by Dsx mKOs in the central region. Dashed lines indicate small patches of the wing where ground scales (gs) are exposed. C–D′. Immunofluorescent detection of the DsxDBD antigen in the dorsal layer of female and male wings between at 40% of development. The Bab antigen (green) marks both dorsal cover scales (dcs) and ground scales (gs) in females (C′), and only the ground scales in males. Forewings are shown in C′ and D′ insets. E, F. Immunofluorescence of Bab (green) in the control (E, contralateral left wing, horizontally flipped for re-orientation) and treated (F) female forewings following electroporation with Dsx siRNAs, shown at the 40% pupal stage. DAPI DNA stainings (blue) highlight rows of scale nuclei interspersed by smaller epithelial nuclei. Dsx RNAi results in loss of Bab expression in dorsal cover scales (dcs, dotted lines), while maintaining Bab expression in ground scales (gs, arrowheads). Scale bars: A, B = 100 μm; C, D = 1 mm; C′, D′, E, F = 10 μm.
Fig 4.
Doublesex controls the specification of male-specific spatulate scales.
A. Magnified views of the dorsal hindwing marginal bands of mosaic Dsx G0 crispants. KO clone boundaries are visible with yellow/orange color shifts in the medial region, and in the marginal band, with shifts from canonical black scales to the spatulate scales in females, and vice-versa in males. B, C. Complete, reciprocal shifts in scale composition (B) and melanic scale identities (C) in both female and male Dsx crispants. Scale bars: A–C = 100 μm.
Fig 5.
Dsx and Bab control sexually dimorphic scale fates.
A. Working hypothesis for the effects of Dsx on sexually dimorphic wing traits, including cell non-autonomous patterning effects on marginal patterns, cell-autonomous requirements in the specification of spatulate vs. canonical margin scales, and female-specific repression of UV-iridescence in dorsal cover scales. B–D. Bab RNAi knockdowns on the dorsal forewing phenocopy mosaic KO effects [29]. Bab-expressing cells acquire a UVI state upon Bab perturbation, with the exception of the spatulate scales that are unaffected. This includes canonical melanic scales that acquire UV-iridescence while maintaining melanism (D, cyan arrowheads), resulting in a dark blue iridescent phenotype in the visible spectrum (panel D), as previously described in Bab mosaic KO experiments [29]. Ground scales are transformed to yellow UVI scales in each sex (e.g., in the medial region: C, white arrowheads), except in the female forewing marginal band where they convert from melanic to blue iridescent (combination of melanic and UVI features). Star: inset features yellow vein scales (UV-negative in WT and controls) that acquire a UVI fate upon Bab RNAi, indicating successful knockdown in this area. E, F. Summary of Dsx expression, Bab expression, and perturbation assays in dorsal wing surfaces. Scale bars: B = 5 mm; C, D = 100 μm.
Fig 6.
Single-nucleus transcriptomic profiles of major cell types in the Colias eurytheme male 40% pupal hindwing.
A. UMAP plot representing the 9 clusters resolved across a total of 2,961 cells. B–B′. Violin plots of the number of unique genes (B) and RNA read counts (B′) per nucleus. C–C′. Violin and heatmap plots highlighting differences in mtDNA content across non-scale and scale clusters. D–G. Heatmap plots showing the expression of key marker genes (bold: see main text for citations). Serpent and Rala (arrowheads) in two small unannotated clusters, Stubble (epithelial cells), neuromusculin (nrm, socket cells), rhophilin-2 (wing epithelium), breathless (tracheal epithelium), Notch (non-scales), and shaven (sv, scales). H–H′. Dot plots featuring the expression profiles of 150 differentially expressed genes, chosen among top markers of cluster or cell type identity (Data 1 at https://osf.io/yjvkc/). Panel H includes top markers for non-SOP clusters, socket cells, and scale cells; panel H′ focuses on differentiators between scale sub-types. Dot size reflects the percentage of cells in which gene expression was detected. Color coding allows a relative comparison of gene expression levels within a cluster (horizontal comparisons), but is not proper for vertical comparisons between clusters. Gene names in bold: see main text for references. Feature count matrices and Seurat objects used for the generation of all panels are available at the Open Science Framework Repository [78].
Fig 7.
Diverging gene expression profiles of male scale cell precursors at 40% development.
A. UMAP plot representing the 8 sub-clusters of scale cell precursors across a total of 1,550 cells. B. Violin plots for scale markers (sv, peb, Osi24) and developmental factors showing differential expression patterns across scale subtype (see text) markers (all scales). C. Violin plot of Antp expression showing enrichment in Scale1d, and Antp immunofluorescence (Antp 4C3 antigen) localization to unpigmented wing coupling scales [81]. D. Violin plots for the unpigmented scales marker Bar-H1 [25], far-posterior region marker mirror [82], and proximal region marker homothorax [80]. E. Violin plots for Scale4 marker genes Cut, previously localized to large wing nuclei and wing margin [36,83], and apterous-A, a marker of dorsal-specific structures [79]. F. Immunofluorescent localization of the Cut 2B10 antigen (gray) shows bright marking of the wing margin, and nuclear signal interspersed in the wing epithelium and in tracheal lumen. G. Cut signal in tracheal lumen likely corresponds to chains of differentiating neurons. H–I′. Large nuclei with Cut signal likely correspond for dorsal wing hair, restricted to the proximal side of the discal crossvein (dotted line). J–J′. Expression of Cut in wing margin scales likely corresponds to the precursors of elongated margin scales. Cyan: staining of the basal wing membrane, here shown using the non-specific signal from animmunofluorescence assay using a guinea pig polyclonal anti-Dve antibody. K. Violin plot of nubbin expression. L. Antigenicity of Nubbin 2D4 revealing repression of Nubbin in dorsal scale cells (dcs) and lower signal in ventral scale cells (vcs). Scale bars: Scale bars: C, G, H, H′, I′, J′ = 100 μm; F = 1 mm; I, L = 20 μm. Feature count matrices and Seurat objects used for the generation of Fig 7A–7E, 7K are available at the Open Science Framework Repository [78].
Fig 8.
Transcriptional divergence of UVI and spatulate scale cell precursors.
All experiments correspond to male wing tissues at 40% pupal development. A–A′. Heatmap plot of the 145 most-significant differentially expressed genes in clusters Scale2 and Scale3 relative to the remaining scale cell clusters (Data 2-3 at https://osf.io/yjvkc/). The top-left panel (A) shows genes downregulated in Scale2 and Scale3. The remaining panels (A–A′) show genes that are enriched in either or both of these clusters. Bold: see text for details. Blue: see panels C–E for spatial expression. B. Violin plots for Bab and marker genes used for the spatial identification of the Scale2 and Scale3 cluster. C. HCR localization of Dsx mRNA (exons 1-2) in male dorsal wings. D. HCR localization of Arylsulfatase (green) and Abp1 (magenta), respectively, tested as positive and negative markers of the Scale2 cluster, in male dorsal wings. Arylsulfatase is found in the dorsal cover scales of the male marginal region. Arrowheads: ground scale expression of Abp1 in the marginal region, without overlap with Arylsulfatase. E. HCR localization of Sulfatase1 mRNA, tested as a marker of the Scale3 cluster, in the medial region of a male dorsal hindwing. Expression is restricted to alternating scale precursor cells corresponding to the presumptive UVI scales, while ground scale precursor cells (gs; dotted lines) are negative. Scale bars: C-D (top) = 1 mm; C, D insets (bottom) = 100 μm; E = 10 μm. Feature count matrices and Seurat objects used for the generation of Fig 8A, 8B are available at the Open Science Framework Repository [78].
Fig 9.
ChIP-seq profiling of Bab identifies potential target genes involved in the differentiation of Bab– scale types.
A. MEME-ChIP predicted motif for Bab occupancy as inferred from the Bab ChIP pupal wing dataset (top), resembling a previous binding profile for a Bab fly ohnologue (bottom). B. Summary of the position of all predicted Bab ChIP-seq binding sites across two datasets, at the 40% and 60% pupal stages (Data 4-5 at https://osf.io/yjvkc/). Inner circles: HOMER classification of imputed binding sites relative to gene annotation features. Outer rings: overlap with genes that are among the differentially expressed (DE) genes in Scale2 (spatulate) and/or Scale3 (UVI) relative to other scale types. C. Venn diagram featuring the numbers of genes identified as Bab-bound and differentially expressed (DE) in the Scale2-3 clusters, resulting in an intersection of 77 (up) and 53 (down) DE genes with binding sites identified across two stages. D, E, F, H, J. Genomic intervals featuring the Bab ChIP-seq profiles across replicates and peak calls at each stage, including at the promoters of Scale2-3 markers (bottom, buff color). G–G′, I–I″, K-K′. Violin plots of snRNAseq scale cluster expression for the same marker genes.
Fig 10.
Potential targets of Bab regulation in the UVI scale type.
A. Box and whiskers plot of probability of differential expression, binned by number of called Bab ChIP sites in proximity to each gene (Wilcoxon rank sum tests; *: p < 0.01; ***: p < 1e−15). B. Expression dot plot visualization of selected UVI-DE genes (log2FC > 1.8, adjusted p < 0.01) with at least 3 Bab ChIP peaks and potential roles in ultrastructural specialization (Data 5 at https://osf.io/yjvkc/). C, D. Volcano plots of UVI-DE genes (Data 7-8 at https://osf.io/yjvkc/), colored by number of Bab ChIP sites (C), and by the presence of selective sweeps in a sympatric population of C. eurytheme x C. philodice (D). These two morphospecies hybridize and show genome-wide admixture across autosomes (dendrograms reproduced from [29], resulting in a population polymorphic for UV iridescence (present in C. eurytheme, absent in C. philodice).