Adaptive introgression of a visual preference gene

Abstract

Visual preferences are important drivers of mate choice and sexual selection, but little is known of how they evolve at the genetic level. In this study, we took advantage of the diversity of bright warning patterns displayed by Heliconius butterflies, which are also used during mate choice. Combining behavioral, population genomic, and expression analyses, we show that two Heliconius species have evolved the same preferences for red patterns by exchanging genetic material through hybridization. Neural expression of regucalcin1 correlates with visual preference across populations, and disruption of regucalcin1 with CRISPR-Cas9 impairs courtship toward conspecific females, providing a direct link between gene and behavior. Our results support a role for hybridization during behavioral evolution and show how visually guided behaviors contributing to adaptation and speciation are encoded within the genome.

Fig. 1. Parallel visual preferences are controlled by the same genomic region in the Heliconius melpomene-cydno group.

(A) H. melpomene (dotted orange line) co-occurs with H. cydno (blue) in Central America and South America to the west of the Eastern Cordillera in the Andes, while H. melpomene co-occurs with H. timareta (orange) to the east of the Eastern Cordillera. H. melpomene and H. timareta share red warning patterns even though the latter is more closely related to the white/yellow H. cydno. (B) Proportion of courtship time directed towards red H. timareta females relative to white H. cydno females by males of the three species. Point size is scaled to the number of total minutes a male responded to either female type (a custom swarmplot was used to distribute dots horizontally). Estimated marginal means and their 95% confidence intervals are displayed with black bars. (C) Crossing design for producing backcross hybrid individuals to H. cydno segregating at the behavioral QTL region on chromosome 18. (D) Relative courtship time directed towards red H. timareta females by F1 hybrid and backcross to H. cydno hybrid males. Orange points represent individuals that are heterozygous (i.e., ‘cydno-timareta’) and blue points represent individuals that are homozygous for H. cydno alleles at the QTL peak/optix region on chromosome 18. Note that although we observe evidence of recombination in our crosses, the QTL peak/optix region on chromosome 18 often segregates with warning pattern (see supplementary Materials and Methods (46)). (E) Differences in estimated marginal means for relative courtship time between butterfly types tested in Colombia (this study) and in Panama (11). T = H. timareta, M = H. melpomene, C = H. cydno, Backcross = backcross to H. cydno hybrids.

Fig. 2. Different genomic signatures support both divergence and adaptive introgression at the regucalcin locus.

Left, from top to bottom: Admixture proportion values (20kb windows) between H. melpomene and H. timareta at the behavioral QTL region on chromosome 18 (x-axis indicates physical position) for Colombian (black) and Peruvian (gray) populations, with recombination rate overlaid in blue; topology weightings (proportions of a particular phylogenetic tree over all possible rooted trees) for the “species” (blue) and “introgression” (orange) trees (50 SNPs windows, a loess smoothing function across 150kb windows was applied). H. numata was used as outgroup; composite likelihood ratio (CLR) of a selective sweep in H. timareta (50 SNPs windows); fixation index (F_ST) and d_xy, measures of genetic differentiation and divergence, between H. timareta and H. cydno. The gene coordinates of the candidate gene for behavioral difference regucalcin1 as well as the color pattern gene optix (~550 kb apart) are highlighted by vertical light blue dotted lines, and the putative regulatory regions of optix affecting color pattern are indicated by gray shading. Note that the QTL confidence region contains 200 genes (47). Panel to the right zooms into the region containing candidate behavioral genes. M, T, C and N denote H. melpomene, H. timareta, H. cydno and H. numata, respectively; subscripts _P and _C denote Panama and Colombia, respectively.

Fig. 3. Cis-regulated expression differences of regucalcin1 are associated with visual preference and regucalcin1 is expressed in the visual pathways.

(A) Regucalcin1 is differentially expressed between red-preferring and white-preferring butterflies. Histogram heights represent the value and bars the standard error of the (base 2) logarithmic fold change in expression between red-preferring and white-preferring Heliconius subspecies (comparisons conducted only between butterflies raised in the same insectary locations). The dashed red line indicates the threshold for a 2-fold change in mRNA expression. M, T, C denote H. melpomene, H. timareta and H. cydno, respectively; subscripts _{P, C} and _Pe denote Panama, Colombia and Peru, respectively. (B) Allele specific expression analyses indicate that differences in expression of regucalcin1 in the brains of red and white preferring population are cis-regulated. Points indicate the value and bars the standard error of the log2 (fold change) in expression between parental species (vertical) and the alleles in F1 hybrids (horizontal), for regucalcin1. Dashed red lines indicate the threshold for a 2-fold change in expression for the genes in the species (horizontal), and for the alleles in the hybrids (vertical). Regucalcin1 is largely cis-regulated (indicated by proximity to y = x). (C) Regucalcin1 is widely expressed in Heliconius melpomene brains, including the visual pathways, and eye (Fig. S8). On top, immunostaining of the right hemisphere, from left to right: counterstaining of somata with neurotrace and of the neuropil with synapsin, center: staining against regucalcin1, right: merged image. Below, enlargement of somata (i, iii, iv), where the signal is particularly strong in nuclei, and neuropil (ii) along the visual pathways.

Fig. 4. Disruption of regucalcin1 with CRISPR/Cas9 impairs male courtship behavior.

(A) Left: schematic representation of the regucalcin1 locus with the target sites of the small guide RNAs and resulting CRISPR/Cas9-mediated deletion of ~1300bp. Right: gel electrophoresis of PCR-amplified regucalcin1 fragments from individuals without (ND) and with deletion (mKO) at regucalcin1. (B) Schematic representation of courtship trials. Experimental (i.e., a mKO or ND) males that passed our ‘drop test’ were paired with a wildtype (WT) male and introduced into a cage with a wildtype virgin H. melpomene female. This paired design allowed us to control for both the injection procedure, as well as prevailing conditions that might potentially influence male behavior. (C) Proportion of time spent flying or feeding by experimental (‘exp’) males, i.e. those injected but non-deletion (ND) males or regucalcin1 mosaic knock-out (mKO) males, relative to wildtype (WT) males (left panel); proportion of courtship time directed towards the same H. melpomene female by injected but non-deletion (ND) males (left) and regucalcin1 mosaic knock-out (mKO) males relative to wildtype (WT) males (right panel). Point size is scaled to the number of total minutes a male flew/fed or courted during the experiments.