Cellular profiling of a recently - evolved social behavior

Social behaviors are essential for survival and reproduction and vary strongly among individuals, species, and heritable brain diseases. The biological bases of this variation are poorly resolved, and discovering them is necessary to understand how neural circuit and behavioral functions - and dysfunctions - vary in social contexts. Here we integrate single nuclei RNA-sequencing (snRNA-seq) with comparative genomics and automated behavior analysis to investigate the neurobiology of castle-building, a recently-evolved social, spatial, goal-directed, and repetitive construction behavior in Lake Malawi cichlid fishes. We simultaneously control for and analyze two biological variables correlated with castle-building behavior: quivering, a courtship "dance" behavior, and relative gonadal mass. We demonstrate 1) distinct cell type-specific signatures of building-, quivering-, and gonadal-associated neuronal excitation, gene expression, and neurogenesis; 2) converging evidence for the involvement of estrogen, TrkB, and CCK signaling systems, as well as pallial and hippocampal-like excitatory neuronal subpopulations, in castle-building behavior; and 3) evidence that castle-building has evolved in part through genomic divergence in a gene module that is selectively expressed in stem-like quiescent radial glia cells (RGCs) lining the dorsal telencephalon. This RGC subpopulation exhibits signatures of a building-associated departure from quiescence, which in turn is associated with neuronal rebalancing in the putative fish homologue of the hippocampus. Our work supports an unexpected role for glia and neurogenesis in the evolution of social behavior, and more broadly shows how snRNA-seq can be used to rapidly generate testable mechanistic models for the genetic, molecular, and cellular regulation of previously unstudied social behaviors in new species systems.