(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



Genetic, developmental, and neural changes underlying the evolution of butterfly mate preference [1]

['Nicholas W. Vankuren', 'Department Of Ecology', 'Evolution', 'The University Of Chicago', 'Chicago Illinois', 'United States Of America', 'Nathan P. Buerkle', 'Department Of Organismal Biology', 'Anatomy', 'Chicago']

Date: 2025-03 

Many studies have linked genetic variation to behavior, but few connect to the intervening neural circuits that underlie the arc from sensation to action. Here, we used a combination of genome-wide association (GWA), developmental gene expression, and photoreceptor electrophysiology to investigate the architecture of mate choice behavior in Heliconius cydno butterflies, a clade where males identify preferred mates based on wing color patterns. We first found that the GWA variants most strongly associated with male mate choice were tightly linked to the gene controlling wing color in the K locus, consistent with previous mapping efforts. RNA-seq across developmental time points then showed that seven genes near the top GWA peaks were differentially expressed in the eyes, optic lobes, or central brain of white and yellow H. cydno males, many of which have known functions in the development and maintenance of synaptic connections. In the visual system of these butterflies, we identified a striking physiological difference between yellow and white males that could provide an evolutionarily labile circuit motif in the eye to rapidly switch behavioral preference. Using single-cell electrophysiology recordings, we found that some ultraviolet (UV)-sensitive photoreceptors receive inhibition from long-wavelength photoreceptors in the male eye. Surprisingly, the proportion of inhibited UV photoreceptors was strongly correlated with male wing color, suggesting a difference in the early stages of visual processing that could plausibly influence courtship decisions. We discuss potential links between candidate genes and this physiological signature, and suggest future avenues for experimental work. Taken together, our results support the idea that alterations to the evolutionarily labile peripheral nervous system, driven by genetic and gene expression differences, can significantly and rapidly alter essential behaviors.

Funding: This work was supported by National Science Foundation (NSF) EAPSI 1515295 and a Dubner Fellowship to NPB, a University of Chicago Big Ideas Generator seed award to SEP, a University of Chicago BSD Pilot Funds Award to SEP and MRK, and National Institute of General Medical Sciences award R35 GM131828, and NSF grants IOS-1452648 and IOS-1922624 to MRK. The funders played no role in the study design, data collection, data analysis, decision to publish, or preparation of the manuscript. Funder information can be found at https://www.nsf.gov/ and https://nih.gov .

Here, we took an integrative approach to attempt to identify the causes of H. cydno mate choice variation. Using a combination of genome-wide association (GWA), developmental transcriptomics, and PR electrophysiology, we sought to comprehensively characterize variation in the Heliconius visual system. We found genetic and gene expression variation associated with mate choice behavior on at least four chromosomes, with the largest effect genes in the K locus. Furthermore, we identified striking differences in inter-PR inhibition between H. cydno males that were strongly correlated with their mate preference. Overall, our results suggest that one outcome of K locus variation is differences in inter-PR inhibition of UV PRs that contribute to variable mate choice behavior, and provide important hypotheses about the links between genetic variation, peripheral visual system variation, and a critical visual behavior.

Mate choice behavior thus appears to be associated with one or a few major-effect loci, but how this genetic variation manifests as differences in the underlying neural circuitry remains unknown. Butterfly visual systems vary extensively across species both in eye organization [ 28 , 29 ] and in the size of the brain [ 30 ], highlighting that causal differences could occur anywhere along the visual processing pathway ( Fig 1C ). Differences in eye organization are especially well characterized within Papilio and Heliconius, with variation in screening pigments [ 31 – 33 ], opsin expression [ 34 , 35 ], and co-expression of multiple opsins within a single photoreceptor (PR) [ 36 , 37 ] contributing to at least six distinct ommatidial types in Heliconius alone ( Fig 1D ) [ 29 , 38 , 39 ]. In Heliconius, these opsins include a long wavelength sensitive opsin (LW), a blue sensitive opsin, and paralogous UV sensitive opsins (UV1 and UV2) tuned to ~ 355 nm and ~ 390 nm, respectively [ 28 , 40 ]. Additionally, physiological and anatomical data from other butterfly species have revealed inhibitory inter-PR synaptic connections between cells with different spectral sensitivities [ 23 – 25 , 41 – 43 ]. These interactions between PRs generate color-opponent-like receptive fields, although the impact on downstream visual computations and behavior remains unexplored.

( A ) The cydno complex and its sister taxon, Heliconius melpomene. ( B ) Male preference, i.e., the proportion of courts directed at white females [ 15 , 16 ]. Each point represents preference of one male; red points are means. ( C ) Visual perception of wing color and organization of the Heliconius visual system. Light received through hundreds of individual ommatidia in the retina is transmitted through multiple layers of the optic lobe to the central brain. ( D ) The eye is organized into ommatidia containing nine photoreceptors (left), with distinct ommatidial types typically defined by the opsin expression patterns in the R1 and R2 cells (right). R1/R2 axons bypass the lamina (LA) and project directly to the medulla (ME), while R3-8 project only to the lamina where they can also make inter-photoreceptor synaptic connections with R1/R2 axons [ 23 , 24 , 25 ]. ( E ) Spectral reflectance of white and yellow H. cydno wings primarily differ in the UV region. Raw data and code used to generate panels A and E can be found in Dryad repository dryad.z8w9ghxjz.

Here, we investigated the genetic and neurobiological basis of mate choice variation within the H. cydno clade of butterflies ( Fig 1 ). Heliconius pachinus and Heliconius cydno galanthus are yellow- and white-winged sister species in which males have strong preference for females with conspecific wing colors. Previous mapping experiments showed that both color and mate choice variation are most strongly associated with the K locus, a narrow region on chromosome 1 [ 15 ]. In contrast, Heliconius cydno alithea is polymorphic for yellow and white wings and males display variable mate choice behavior: while yellow H. c. alithea males have strong preference for yellow females, white H. c. alithea males court yellow and white females equally ( Fig 1B ). This latter result is consistent with the fact that most white alithea are heterozygous at the K locus, and mirrors both the behavior and genetics of galanthus/pachinus F1 hybrids [ 15 , 16 ]. We previously showed that aristaless-1, a gene located in the K locus, controls the switch between recessive yellow and dominant white wing colors [ 26 , 27 ]. However, the gene(s) controlling mate choice variation remained unknown.

The co-evolution of wing color and mate choice in Neotropical Heliconius butterflies presents an excellent system to integrate multiple approaches to studying behavioral evolution. Heliconius butterflies evolved myriad bold wing color patterns that warn predators of their toxicity and mediate mimicry, but also serve as the primary signals used for mate choice [ 14 – 19 ]. Visual perception of wing color patterns is the critical first step in mate choice, as males preferentially court females with the same wing color [ 14 , 15 , 20 ]. The genes controlling most color pattern variation have been identified and extensively studied, but only recent mapping studies have begun to identify the genetic loci associated with mate choice variation [ 15 , 18 , 19 ]. These studies provided two important observations. First, loci associated with mate choice variation are often tightly linked to the loci that control color patterns used for mate choice [ 15 , 16 , 19 – 22 ]. Second, Rossi and colleagues [ 19 ] showed that genetic and expression variation of regucalcin1, a gene expressed throughout the nervous system, was associated with species-specific mate choice behavior in Heliconius melpomene, Heliconius cydno, and Heliconius timareta. Together, these studies suggest where to look for putative mate choice genes in the genome and visual circuits. However, the direct links between genetic variation, expression variation, development, and behavior remain elusive.

Behavioral evolution requires genetic variation that ultimately alters the neural circuits mechanistically responsible for generating differences in behavior. Many studies have mapped the genetic basis for behavioral evolution [ 1 – 4 ] or associated differences in neural physiology with divergent behavior [ 5 – 7 ], but the links between each of these layers of variation are often missing. The peripheral nervous system appears to be an especially labile target for evolutionary modification [ 8 – 11 ]. Shifts in receptor sensitivity can avoid the potentially deleterious effects associated with changing complex brain circuits while still enabling changes in the perception and distinguishability of sensory stimuli. However, simple shifts in receptor sensitivity may be insufficient to enact large behavioral changes, instead requiring more significant changes in how downstream circuits process sensory information [ 5 , 6 , 12 , 13 ]. Understanding how these more complex changes emerge requires an integrative approach combining genetics, transcriptomics, and neurobiology that can reveal how genetic variation affects the functional organization of the brain.

Results

Heliconius cydno alithea mate choice variation is strongly associated with the K locus Previous quantitative trait locus mapping in H. c. galanthus and H. pachinus showed that both wing color and male preference variation were linked to the ~ 2 Mb K locus on chromosome 1 [15]. Although color and preference were also strongly correlated in H. c. alithea, the genetic architecture of preference variation in this subspecies, and whether it mirrored the situation H. c. galanthus and H. pachinus, remained unknown [15,16]. We began mapping color and preference variation in H. c. alithea by performing a GWA for color and male mate choice using 1,529 courtship events from 57 yellow and 56 white males (Figs 2 and S1 and S2; S1 Table) [16]. These males were tested for their preference by Chamberlain and colleagues (2009), then re-sequenced for this study. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Genetic architecture of Heliconius cydno alithea male mate choice variation. (A, B) Genome-wide association (GWA) for H. c. alithea mate choice and forewing color genome-wide (A) and at the K locus (B) using 1,529 courtship events from 57 yellow and 56 white males. Chromosome numbers are shown between the plots. Lambda values (λ) are genomic inflation factors for each analysis, indicating no p-value inflation in the dataset despite potential residual population stratification (S1 Fig). K locus peak (KP) labels referenced in the text are shown above the plot. DEGs: differentially expressed genes—see Fig 3. KP2 was 25 kb downstream of the ortholog of Drosophila melanogaster Hasp; KP3 was 15 kb upstream of the transcription factor senseless-2 and RNA pseudouridylate synthase domain containing protein 2 (rpusd2); and KP4 was within an rpusd2 intron. FDR: Benjamini–Hochberg false discovery rate cutoff. (C) Posterior inclusion probability for each variant in the chromosome 1, 7, 9, and 11 peaks, calculated using SuSiE-RSS [44]. PIP values indicate the probability that each variant is predictive of the input phenotypes (choices in this case). Higher PIP values indicate higher predictive value. We calculated PIP values for top variants in the K locus as well as the peaks on chromosomes 7, 9, and 11. (D) Pairwise linkage disequilibrium between top color and preference variants relative to the empirical LD in the K locus at each distance. We only show values between the top color variant and each of the 34 choice variants at FDR < 0.01. Patterns were similar between the second and third ranked color variants and are not shown. K locus variants are shown in the left panel while the chromosome 7, 9, and 11 variants are shown in subsequent panels relative to empirical unlinked marker LD. Raw data, code, and further exploration of the data can be found in S1 and S2 Tables, S1 and S2 Figs, and Dryad repository dryad.z8w9ghxjz. https://doi.org/10.1371/journal.pbio.3002989.g002 Consistent with our previous study, H. c. alithea forewing color variation was strongly associated with a single narrow region within the K locus, 21 kb downstream of al-1 (Fig 2) [26]. In contrast, H. c. alithea mate choice variation was associated with genetic variation on four chromosomes at a false discovery rate (FDR) of 1% (Fig 2) [15,46]. The most significant associations were found in the K locus, while secondary peaks were found on chromosomes 7, 9, and 11. Within the K locus, significant variants localized to four discrete K locus peaks (KP): KP1 was near, but not coincident with, the top color variants while KP2, KP3, and KP4 were 240 kb, 520 kb, and 790 kb away, respectively (Fig 2B). In addition to the K locus peak, mate choice was also strongly associated with variation on chromosomes 7, 9, and 11 (Figs 2 and S2). Top variants on chr7 fall within an intron of diacyl glycerol kinase, a gene encoding an enzyme involved in cell membrane homeostasis and intracellular signaling. The chr11 variants span 350 kb and 19 genes, with the top variants falling within introns of a potential acyl transferase (CG17707) and collagen 11A1 and 2.5 kb downstream of ubiquitin conjugating enzyme 2M (UBE2M; S2 Fig). Although the chr7 and chr11 genes had no immediately obvious links to vision or courtship behavior, we were intrigued to find that the top variant on chr9 falls within an intron of the transcription factor spineless, which controls stochastic choice of opsin expression in R1/R2 cells in Papilio swallowtail butterfly eyes (Fig 1) [47]. Thus, plausible mate choice genes may also be found outside of the major effect gene(s) within the K locus. Consistent with the K locus having the main effect on mate choice variation, white alleles at the top KP2 variant had the highest marginal effects on male mate choice (0.11, 95%CI 0.05–0.18) followed by 0.08 (0.04–0.12), 0.07 (0.03–0.11), and 0.02 (−0.02 to 0.06) at the top chr9, chr11, and chr7 variants, respectively. Although these estimates may be somewhat inflated due to the Beavis effect [48], they suggest that each white allele in KP2 reduced the probability of choosing white females by ~ 11%. These results were altogether consistent with previous mapping studies linking mate choice to the K locus in H. c. galanthus and H. pachinus, but the increased resolution from this population-level analysis strongly suggested that separate, discrete genetic loci control color and behavior variation.

Color and choice are associated with physically separate, but genetically linked variants To better estimate which of the significant GWA variants was potentially causal for mate choice variation, we performed fine-mapping of the K locus, chr7, chr9, and chr11 variants using SuSiE-RSS [49]. This approach calculates the likelihood of each variant contributing to mate choice variation given the GWA coefficient estimates and linkage disequilibrium (LD) between those variants, outputting statistically credible sets of putative causal sites. SuSiE-RSS identified 160 variants in the final credible set, out of 61,824 analyzed. These results largely mirrored the GWA, but strongly suggested that the top KP2 variant, a C/T single-nucleotide polymorphism (SNP) 240 kb away from the top color variant, was most predictive of male mate choice (Fig 2C) [44]. The top mate choice and color variants were not coincident, suggesting that physically separate loci control these two traits. Yet, theory predicts that the correlation between loci controlling mate preference and cue will rapidly decay in the absence of mechanisms that maintain LD between them [50]. Despite the top color and choice K locus variants being up to 790 kb apart, most pairs of color and choice variants were in extremely high LD relative to empirical levels of LD in the K locus (Fig 2D). LD between the top color variant and KP4 variants, for example, was over 10 times higher than the K locus average of 0.06 for similarly spaced variants (Fig 2D). LD across the K locus, measured in sliding windows of r2 or D′, was not generally increased relative to the genome-wide average; however, and we found no evidence for inversions or other genomic features that could easily explain this high pairwise LD (Figs 2D and S3 and S4). Together, these results suggested that the genomic basis for genetic coupling in H. cydno comprises physically separate, but genetically linked variants that control wing color and male color preference.

Variability in UV photoreceptor spectral sensitivity Gene expression analysis uncovered differential expression throughout the eye and brain, suggesting many possible neural loci where variability could be important for modulating courtship preference behavior (Figs 1D and 3). We opted to begin our investigation on the neural correlates of courtship behavior in the eye for several reasons. First, gene modules that were most affected by differential expression contained many genes involved in eye development, including herzog and mtg. Second, the peripheral nervous system is known to be an evolutionarily labile hotspot for modification [8–11]. Finally, the Heliconius eye displays extreme diversity across the genus [28,29], but there is limited knowledge about the specific organization of the eye across H. cydno butterflies (but see [59]). Thus, understanding the potentially variable sensory information available to the butterflies was an important first step towards describing the circuits associated with this vision-based courtship behavior. To characterize the encoding of sensory information in the retina, we measured the spectral tuning and response properties of single PRs using sharp intracellular electrodes while presenting brief flashes of monochromatic light. In addition to recording from 5 groups of H. cydno butterflies separated by species and wing color, we also included H. melpomene as a closely related outgroup. Based on previous results [59] and analysis of the data presented here, we observed a strong sexual dimorphism where male eyes varied with species and wing color while female eyes did not. We therefore grouped all females together, regardless of species or wing color, and treated them as a single group separate from males. Across all butterflies, we recorded from a total of 503 PRs with spectral sensitivities that segregated into five classes broadly consistent with cell types previously described in other Heliconius species [29,39]. These included UV sensitive, blue sensitive, and three types of LW sensitive PR (Figs 4 and S7). The three types of LW sensitive comprised a green sensitive cell type that fit the expected tuning of the LW opsin, a red-shifted variant that is likely derived from a combination of the LW opsin and a red screening pigment, and a broadband sensitive variant that is likely derived from co-expression of the blue and LW opsin (S7 Fig) [39]. For each PR, we estimated the wavelength of peak sensitivity (λ Max ) by fitting the measured spectral sensitivity with a rhodopsin tuning template [60]. The spectral tuning of blue and LW sensitive PRs did not vary across groups (S7 Fig). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. Variable co-expression of UV1 and UV2 within single photoreceptors cannot explain mate choice variability. (A) Spectral sensitivity (mean ± SEM) of UV photoreceptors measured for H. c. alithea males (left) and all females (right). Dotted lines indicate the expected sensitivity for UV photoreceptors expressing either the UV1 or UV2 opsin. (B) λ Max was estimated for each cell with a rhodopsin tuning template and separated into groups based on species, sex, and wing color (n = 43 cells/12 individuals, 40/14, 18/9, 19/9, 8/4, 30/16, 22/10). Asterisks above indicate significant differences between groups assessed using pairwise GLME models (t-statistic, p < 0.05) with negligible random effects due to grouping by individual animal, and asterisks below indicate a significant difference from the expected tuning of both UV1 and UV2 (t test, p < 0.05 with Bonferroni correction). (C) Representative anti-UV1 and anti-UV2 antibody stains in three representative white H. c. alithea males showing consistent expression of UV2 and variable expression of UV1 across individuals. Additional stains can be found in reference [59], which comprehensively characterizes this variation. (D) Overlay of predicted opsin absorption from template tuning curve and wing reflectance (left) and normalized convolution between the two (right) shows that differences in spectral tuning cannot explain differences in courtship preference. The data underlying this figure can be found in Dryad repository dryad.z8w9ghxjz. https://doi.org/10.1371/journal.pbio.3002989.g004 In contrast, the spectral sensitivity of UV sensitive PRs varied significantly across groups, with λ Max ranging continuously between 345 and 404 nm across all cells (Figs 4A, 4B, and S8). For H. melpomene males and all females, λ Max was not significantly different from the expected tuning of UV1. For all H. cydno males; however, UV spectral tuning was significantly different from the expected tuning of both UV1 and UV2 opsins (p < 0.05). Within each group, the distribution of λ Max was unimodal (S8 Fig), so this variability cannot be explained as differences in the proportion of cells expressing UV1 or UV2. Instead, we previously showed that single PRs can co-express both UV1 and UV2 [59], and the relative expression of each opsin within a cell may function to tune the specific λ Max . Consistent with this interpretation, we observed substantially more within-group variability in UV cell λ Max values than blue cell λ Max values (S8 Fig). This effect was most prominent in H. c. alithea, where antibody staining revealed that UV2 staining was strong in every male, while UV1 staining qualitatively varied from strong to weak to absent (Fig 4C). Across all groups, we observed a weak correlation between λ Max and male courtship preferences (Fig 4B). Groups with increasing preference for yellow females tended to have λ Max at longer wavelengths, indicative of stronger UV2 expression. To compare UV spectral tuning across groups, we used a generalized linear mixed-effects (GLME) model to account for recording from multiple PRs in some individuals. This model showed that predicted male courtship preference (white, yellow, or none) was a significant predictor of λ Max (p < 0.001). However, this apparent relationship is unlikely to play a causal role in preference for white or yellow females for two main reasons. First, a GLME specifically comparing λ Max in white and yellow H. c. alithea revealed no significant differences (p = 0.98), in contrast to the observed behavioral differences (Figs 1B and 4B). Second, the primary difference between white and yellow wings is the presence or absence, respectively, of reflectance below ~ 420 nm (Fig 4D). Convolving wing reflectance measurements with opsin absorption showed that white wings strongly excite while yellow wings weakly excite UV PRs regardless of the specific λ Max (Fig 4D). This means that wing reflectance evokes nearly identical primary sensory responses in the eye of every butterfly, despite differences in UV spectral tuning. Together, these results show that courtship behavior cannot be explained by small changes in sensory reception, instead likely requiring modifications to downstream neural circuits that can modulate the cognitive perception of the two distinct wing colors.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002989

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/