A large deletion at the cortex locus eliminates butterfly wing patterning

Abstract

As the genetic basis of natural and domesticated variation has been described in recent years, a number of hotspot genes have been repeatedly identified as the targets of selection, Heliconius butterflies display a spectacular diversity of pattern variants in the wild and the genetic basis of these patterns has been well-described. Here, we sought to identify the mechanism behind an unusual pattern variant that is instead found in captivity, the ivory mutant, in which all scales on both the wings and body become white or yellow. Using a combination of autozygosity mapping and coverage analysis from 37 captive individuals, we identify a 78-kb deletion at the cortex wing patterning locus, a gene which has been associated with wing pattern evolution in H. melpomene and 10 divergent lepidopteran species. This deletion is undetected among 458 wild Heliconius genomes samples, and its dosage explains both homozygous and heterozygous ivory phenotypes found in captivity. The deletion spans a large 5' region of the cortex gene that includes a facultative 5'UTR exon detected in larval wing disk transcriptomes. CRISPR mutagenesis of this exon replicates the wing phenotypes from coding knock-outs of cortex, consistent with a functional role of ivory-deleted elements in establishing scale color fate. Population demographics reveal that the stock giving rise to the ivory mutant has a mixed origin from across the wild range of H. melpomene, and supports a scenario where the ivory mutation occurred after the introduction of cortex haplotypes from Ecuador. Homozygotes for the ivory deletion are inviable while heterozygotes are the targets of artificial selection, joining 40 other examples of allelic variants that provide heterozygous advantage in animal populations under artificial selection by fanciers and breeders. Finally, our results highlight the promise of autozygosity and association mapping for identifying the genetic basis of aberrant mutations in captive insect populations.

Keywords: Lepidoptera; cortex; deletion; development; domestication; evolution; structural variant.

Fig. 1.

Phenotypes of H. melpomene BWPK butterflies. a) Ivory phenotypes in the H. melpomene “Piano Keys” in the UT Austin stock (BWPK). The 3 color states (from left to right “Dark BWPK,” “Pale BWPK,” and “ivory”) depend on the allelic dosage of a codominant mutation. Ivory homozygotes are inviable and only found among offspring from 2 pale heterozygotes. b) Magnified view of the forewing red band region. In the [ivory WT/−] state, abnormal scales have formed in the red region, with red pigment in granules and scales curled, while in [ivory −/−], all scales are yellow or white. c) Magnified view of a central hindwing region that is black in [ivory WT/WT]. The [ivory WT/−] wing has all cover scales as yellow or white while all ground scales remain black. All scales are yellow or white in [ivory −/−]. Complete replacement of melanic scales by yellow-white scales in the abdomen (d) and legs (e) from [ivory −/−] homozygotes. f) Cartoon of image locations for (b–e). Scale bars: b–c = 50 µm; d = 500 µm; e = 100 µm.

Fig. 2.

Genome-wide association and autozygosity mapping in H. melpomene BWPK. a) Cartoons of the allelic variation present in the H. m. BWPK stock. b) The hindwing veins, showing the composite effects of different alleles depicted in (a). c) GWAS for the Dennis element; Manhattan plot of Wald test P-values, with fixed SNPs in red, and a magnified annotation of the region around the gene optix. d) GWAS for ivory, with fixed SNPs in blue and a magnified annotation of the cortex region, with the point of the arrow at the annotated TSS. e) Comparison of red scales in [ivory WT/WT], [ivory −/−], and cortex crispant wings. f) Some ivory butterflies have a small red dot at the base of the ventral hindwing, the only scales on the entire butterfly that are not white or yellow. They share the atypical phenotype of red scales from cortex crispants (e).

Fig. 3.

Ivory is associated with a large deletion in the gene cortex (a) mean coverage plot for the region around cortex, with [ivory WT/WT] in green, [ivory WT/−] in red, and [ivory −/−] in blue. b) Alignments of de novo scaffolds, indicating the precise breakpoints of the ivory deletion. Colors indicate the orientation of sequences relative to the reference scaffold depicted in (a).

Fig. 4.

Knockout of the distal promoter phenocopies cortex protein-coding knockouts. a) Mapping of RNAseq reads at cortex suggests that there are 2 promoters and TSS. Numbers of intron-spanning reads are indicated (see Sashimi plots in Supplementary Figs. 2 and 3). Knockout of the distal promoter in H. erato causes a transformation of black scales to white/yellow in a phenocopy of cortex protein coding knockouts. b) A butterfly with extensive clones but that emerged poorly, indicating pleiotropic effects not observed in the ivory mutants, while c) much smaller clones similar to those reported by Livraghi et al. (2021), with clone positions indicated by pink arrows. d) High magnification images of mutant clones. Scale bars: 100 µm.

Fig. 5.

Phylogenetic clustering and weighting imply a mixed origin for H. melpomene BWPK. Clustering by PCA from both whole-genome data (a) and from the cortex locus using additional selective sequencing data (b) show similar geographic clustering to that previously reported by Martin et al. (2019); c) with the addition of separation between H. melpomene from East of the Andes and H. melpomene from French Guiana, Suriname, and Brazil (here termed “Atlantic”). Heliconius m. BWPK form a distinct cluster, closest to the West cluster of H. melpomene. Topology weighting was performed using TWISST (Martin and Van Belleghem 2017), with 5 defined groups (West of Andes, Ecuador, East of Andes, Atlantic, and BWPK) (c). d) Genome-wide, the most common topologies were 14 (BWPK as outgroup), 11 (BWPK clustered with East of Andes), and 1 (BWPK clustered with Ecuador). In contrast to this in (e), TWISST results at the cortex locus showed that topology 1 was most common, supporting an Ecuadorian origin for the cortex variants in H. m. BWPK.