Evolutionary rewiring of regulatory topology refines butterfly wing patterns

Phenotypes evolve through changes in gene regulatory networks (GRNs), but how changes in network topology contribute to phenotypic evolution is still poorly understood. Butterfly wing color patterns provide a tractable system to answer this question. Here, we characterized spatiotemporal expression of four color pattern genes, WntA, pdm3, ivory, and Optix, in five nymphalid butterflies, and tested their regulatory connections in WntA-associated patterns through genetic perturbations in three species spanning 80 million years of divergence. We found broadly conserved 1) regulatory hierarchy linking early pattern instruction via WntA signaling to subsequent color readout via ivory and Optix, 2) positive autoregulation in WntA signaling, and 3) positive regulation from WntA signaling to pdm3. The WntA-pdm3 axis, however, produced detectable color pattern readouts only in Vanessa cardui. In this species, additional negative feedback from pdm3 to WntA signaling closes a reciprocal WntA-pdm3 feedback loop that refines spatially separated pattern elements. We show that developmental programs can evolve by rewiring internal interactions among pre-existing GRN nodes, emphasizing the importance of network topology in phenotypic evolution.