Neurotranscriptomic signatures of natural variation in mate preference learning in two subspecies of Heliconius melpomene butterflies

This study identifies transcriptomic differences in neural and sensory tissues between two *Heliconius melpomene* subspecies that explain their natural variation in aversive mate-preference learning, suggesting that selection on these genetic networks could drive reproductive isolation and speciation.

The Gist

The Big Picture: The "Rejection" Test

Imagine you are at a dance. You ask someone to dance, they say "no," and you walk away. Now, imagine you try again with someone else. Do you dance with the same enthusiasm, or do you feel a bit discouraged and dance less?

In the world of butterflies, specifically the Heliconius melpomene, this is exactly what happens. Scientists found that two different "cousin" subspecies of these butterflies react very differently to being rejected:

• The "Smart Learners" (H. m. malleti): If a male tries to court a female and she rejects him (or he fails to mate), he learns his lesson. Two days later, he stops trying to court females as much. He has learned to avoid wasted energy.
• The "Stubborn Learners" (H. m. rosina): If a male tries to court and gets rejected, he doesn't care. He keeps dancing with the same energy, ignoring the lesson.

The big question was: Why? What is happening inside their brains that makes one group learn from rejection and the other ignore it?

The Investigation: A Molecular Detective Story

The researchers acted like detectives, but instead of looking for fingerprints, they looked for genetic footprints (RNA) in three specific parts of the butterfly's body:

1. The Brain: The command center.
2. The Eyes: The cameras.
3. The Antennae: The smell sensors.

They compared the "Smart Learners" and the "Stubborn Learners" in two scenarios:

• The Control Group: Just hanging out, no rejection.
• The Training Group: Just got rejected after trying to court a female.

The Findings: It's All in the Brain

Here is what they discovered, broken down simply:

1. The Brain is the Boss
Even before the rejection happened, the two subspecies were already slightly different in their genes. But, the biggest difference showed up after the rejection experience.

• Analogy: Think of the eyes and antennae as the "sensors" on a robot. They are different between the two models, sure. But the Brain is the "CPU." When the robot gets a "rejection" signal, the CPU of the Smart Learner rewrites its software immediately. The CPU of the Stubborn Learner just ignores the error message. The study found the most dramatic genetic changes happened in the brain, proving that the processing of the rejection is what drives the behavior change.

2. The "Magic" Connection
This is the coolest part. In butterflies, there are "Magic Loci." These are special spots in the DNA that control two things at once:

• What they look like: Their wing colors and patterns (which help them survive by looking like poisonous butterflies).
• Who they like: Their preference for those same wing colors.

The researchers found that the genes responsible for learning and memory are physically linked to these "Magic" wing-color genes.

• Analogy: Imagine a factory assembly line. Usually, the machine that paints the car (wing color) is in one building, and the machine that teaches the driver how to drive (learning) is in a totally different building. But in these butterflies, the "Paint Machine" and the "Driving School" are built right next to each other on the same lot. Because they are neighbors, if nature selects for a specific paint job (wing color), the "Driving School" genes get dragged along for the ride. This means the ability to learn from rejection might be evolutionarily tied to the color of their wings.

3. The "Zinc" Factor
The study found that genes related to Zinc were very active in the brains of the learners.

• Analogy: Think of Zinc as the "oil" for the brain's gears. In the Smart Learners, the brain pumped out more "Zinc oil" to help the gears turn smoothly and store the memory of the rejection. In the Stubborn Learners, this oil wasn't being used in the same way, so the memory didn't stick.

Why Does This Matter?

This study explains how new species are born.

• If a group of butterflies evolves a new wing color (to survive), and the genes for that color are linked to the genes for "learning to avoid the wrong mates," then that group will stop mating with the other group.
• Over time, they become two completely different species that never mix.

The Takeaway

Nature is efficient. It doesn't build a separate system for "learning" and a separate system for "mating." Instead, it links them together. In the malleti butterflies, the rejection of a mate triggers a brain chemical change that says, "Stop, I learned something!" In the rosina butterflies, that switch is either broken or turned off, so they keep dancing to the same tune, even when it doesn't work.

It's a perfect example of how a single genetic "neighborhood" can control how a butterfly looks, who it likes, and how it learns from its mistakes.

Technical Summary

1. Problem Statement

While social learning is a widespread adaptive trait, individual and population-level variation in learning ability exists. In the butterfly Heliconius melpomene, two allopatric subspecies exhibit distinct behaviors regarding aversive mate preference learning:

• H. m. malleti: Males decrease courtship after a failed copulation attempt (learning to avoid that phenotype).
• H. m. rosina: Males do not alter courtship behavior following the same experience.

The study aims to identify the neurogenomic mechanisms underlying this natural variation. Specifically, it investigates whether differences in mate preference learning are driven by:

1. Differential gene expression in neural (brain) vs. sensory (eye, antennae) tissues.
2. The involvement of conserved learning/memory pathways.
3. The physical linkage or pleiotropy of learning genes with "magic traits" (loci controlling wing color/pattern and mate preference simultaneously), which could drive speciation through sexual selection.

2. Methodology

Study Design & Subjects:

• Species: Two subspecies of Heliconius melpomene: malleti (Brazil/Ecuador) and rosina (Panama/Costa Rica).
• Experimental Groups: 10-day-old sexually mature males were subjected to two treatments:
  • Training: 90-minute exposure to a conspecific female (preventing copulation).
  • Control: 90-minute isolation.
• Tissues Analyzed: Brain (BR), Eyes (EY), and Antennae (AN) were dissected immediately post-treatment.

Transcriptomic Workflow:

• Sequencing: RNA-seq (Illumina NovaSeq 6000, 100bp single-end).
• Alignment: Reads aligned to the H. melpomene v2.5 genome using STAR.
• Differential Expression Analysis: Performed using DESeq2 with a generalized linear model (y ~ phenotype + treatment + phenotype:treatment).
  • Comparisons included: Subspecies differences (Control), Training effects within subspecies, and the interaction (Subspecies $\times$ Training).
• Network Analysis: Weighted Gene Co-expression Network Analysis (WGCNA) was used to identify gene modules associated with specific traits (learning vs. non-learning).
• Functional Annotation: Gene Ontology (GO) enrichment via Blast2GO; identification of "magic loci" (wing patterning genes like optix, cortex, WntA, aristaless), pheromone pathways, and locomotion genes.

3. Key Results

A. Baseline Transcriptomic Differences

• Even in the absence of social exposure (Control), significant baseline differences exist between the subspecies:
  • Brain: 1,608 Differentially Expressed Genes (DEGs).
  • Eyes: 1,301 DEGs.
  • Antennae: 1,398 DEGs.
• 62–72% of these DEGs were unique to the control condition, suggesting inherent differences in sensory processing and neural development ("umwelt") between the populations.

B. Learning-Associated Transcriptomic Shifts

• Tissue Specificity: The greatest transcriptomic divergence between the "good learner" (malleti) and "bad learner" (rosina) occurred in the brain (1,929 DEGs), followed by eyes (1,201) and antennae (834).
• Unique DEGs: 53–68% of the DEGs in the training condition were unique to the learning context (not present in controls).
• Key Genes:
  • Potassium voltage-gated channel protein Shab (Shab): Downregulated in trained malleti across all tissues.
  • Regucalcin: Upregulated in sensory tissues but downregulated in the brain of malleti.
  • Neurobeachin (rugose): Downregulated in trained malleti brains; known to be essential for short-term olfactory memory in Drosophila.

C. Functional Enrichment & Network Modules

• Brain Modules: WGCNA identified three brain modules (blue, brown, lightyellow) associated with learning differences.
  • Blue Module: Enriched for "zinc ion binding" (critical for synaptic plasticity and memory).
  • Brown Module: Enriched for 53 GO terms related to learning and memory.
• Sensory Tissues: The eye showed one significant module, but the antennae showed no significant training-associated modules, suggesting the brain is the primary site of processing for this specific aversive learning task.

D. "Magic Loci" and Pleiotropy

• The study found that genes controlling wing color/pattern ("magic traits") and their adjacent genes are differentially expressed in the context of learning:
  • Optix Locus: Regucalcin 1 & 2 and GRIK2 (glutamate receptor) were DE. Regucalcin is a known visual preference gene; GRIK2 is linked to NMDA-mediated learning.
  • Cortex Locus: Wash and Dome (JAK/STAT pathway) were DE in the brain. Dome is implicated in long-term olfactory memory in flies.
  • L Locus (Locomotion): Genes Col4a1 and Vkg (wing beat frequency) were upregulated in trained malleti brains.
• Pheromone Pathway: Genes involved in male pheromone synthesis (GGPS1, wat, ARD1) were differentially expressed, suggesting a link between chemical signaling and learning.

4. Key Contributions

1. Neurogenomic Basis of Learning Variance: Demonstrates that natural variation in mate preference learning is driven primarily by central nervous system (CNS) processing differences rather than just sensory input differences, though sensory tissues also show divergence.
2. Conserved Pathways: Identifies that learning differences involve conserved pathways across animals, specifically zinc ion binding and glutathione transferase activity (energetic demands of learning).
3. Linkage of Learning and "Magic Traits": Provides strong evidence that genes controlling learning and memory are physically linked to or pleiotropic with genes controlling wing patterning and mate preference.
   • This supports the hypothesis that selection on wing patterns (natural/sexual selection) can "hitchhike" learning genes, or vice versa, facilitating rapid speciation.
4. Mechanism of Learning Suppression: Suggests that in H. m. rosina, social rejection (mate failure) may actively repress learning genes, whereas in H. m. malleti, these genes are activated.

5. Significance

• Speciation Mechanism: The study offers a molecular framework for how assortative mating evolves. If learning genes are linked to the traits they help select (e.g., wing color), selection on the trait reinforces the learning mechanism, creating a feedback loop that drives reproductive isolation.
• Evolution of Plasticity: It highlights that the capacity to learn is itself a selectable trait. The divergence between malleti and rosina suggests that in some lineages, the cost of learning (energy, time) may outweigh the benefits, leading to the evolutionary loss of specific learning behaviors.
• Multimodal Integration: The findings suggest that mate preference learning in Heliconius is not purely visual or olfactory but involves a complex integration of visual, olfactory, and locomotor processing, all potentially regulated by a shared genomic architecture.

Conclusion:
The paper concludes that the genetic underpinnings of mate preference learning involve a complex interplay between neural processing and sensory tissues. Crucially, the physical linkage of learning genes to "magic trait" loci suggests that selection on mating traits can directly drive the evolution of learning behaviors, accelerating the process of speciation in Heliconius butterflies.