Divergent sensory transcriptomic profiles in positive and negative learning in <em>Bicyclus anynana</em> butterflies

Mate preference learning, where individuals learn to prefer or avoid specific phenotypes during mate selection, is pervasive across animal taxa and influences reproductive isolation and trait evolution. Despite its significance, the genetic basis underlying the associated valence attribution (whether to prefer or avoid) remains largely unclear. Both in terms of what genes are associated with attributing valence, and whether valence attribution is associated with transcriptional changes in both sensory tissues and neural circuits in the brain, or is restricted to the brain. Here, we investigate the neurogenomic basis of positive and negative mate preference learning using Bicyclus anynana butterflies. We put females in positive or negative learning scenarios, as well as a naïve control, and compared their transcriptomic profiles across three tissues: antennae, eyes, and brain using RNA-sequencing. Our results reveal tissue-specific transcriptional responses, with the antennae showing higher transcriptomic changes during negative learning and the eyes showing higher transcriptomic changes during positive learning, relative to the naïve control, indicating that valence attribution during learning may not be restricted to transcriptional changes in the brain. We identified a subset of genes in each tissue whose expression patterns changed with valence, as well as genes previously linked to classical conditioning pathways, supporting the hypothesis that imprinting-like learning and classical conditioning share molecular mechanisms. Our findings suggest that valence attribution in mate preference learning involves tissue-specific transcriptional responses in sensory and brain tissues, emphasising the role of peripheral sensory systems in modulating learned mate preferences. Methods To determine the gene expression associated with different types of social learning, we exposed female B. anynana to different training/ exposure treatments for one hour (n = 10 per treatment) from August to October 2022. Exposure assays were conducted one hour after sunrise. Every individual used in our exposure assays were size- and age-matched: females were Day 0, ensuring sexual immaturity (Costanzo and Monteiro 2007), and males were Day 3, when their sex pheromones are known to be attractive to females (Nieberding et al. 2012). Naïve females were isolated and were not exposed to training stimuli (i.e. males), females that were exposed to a randomly selected 4-spotted male with undisrupted male sex pheromones were considered given a “positive training exposure”, and females that were exposed to a randomly selected 2-spotted male with disrupted male sex pheromones were considered given a “negative training exposure”, based on prior studies (Westerman et al. 2012; Westerman and Monteiro 2013). After one hour of exposure (training), the females were decapitated with RNA-free scissors. Each head was placed into individual RNA-free 1.5ml LoBind tubes, immediately flash frozen in liquid nitrogen, and stored in -80⁰C until dissection. Sensory (antennae and eyes) and brain tissues were dissected from these females, RNA was extracted, libraries prepared, and sent for RNA-seq. Sequencing data was analysed using DESeq2 analyses and permutation tests.