Transmitter Co-Expression Reveals Key Organizational Principles of Local Interneuron Heterogeneity in the Olfactory System

Abstract

AbstractHeterogeneity of individual neurons within a population expands the computational power of the entire neural network. However, the organizing principles that support heterogeneity within a neuronal class are often poorly understood. Here, we focus on a highly heterogeneous population of local interneurons whose traits co-vary seemingly at random. We asked if local interneurons (LNs) in the antennal lobe (AL) of Manduca sexta express fixed, predictable combinations of neurotransmitters, or if transmitter co-expression can be explained by random probability. We systematically determined the co-expression of neuropeptides and GABA by LNs and found variable patterns of co-expression for all neuropeptides, except for tachykininergic LNs which exhibited highly stereotyped co-expression on a neuron-by-neuron basis. To test if observed patterns of co-expression were random, we used a computational model and found that the probabilities of transmitter co-expression cannot be explained by independent expression of each transmitter. We also determined that setting a single rule in the model, while leaving the rest of the co-expression up to random probability, allowed the model to replicate the overall heterogeneity of transmitter co-expression across antennal lobe LNs. This implies that certain co-expression relationships contribute to the ground plan of the AL, but that otherwise, transmitter expression amongst LNs may be random, allowing heterogeneous co-expression patterns to emerge. Furthermore, neuropeptide receptor expression suggests that peptidergic signaling from LNs may simultaneously target olfactory receptor neurons, LNs and projection neurons, and thus the effects of different peptides do not segregate based on principal AL cell type. Our data suggest that while specific constraints may partially shape transmitter co-expression in LNs, a large amount of flexibility on a neuron-by-neuron basis produces heterogeneous network parameters.