How many different types of neurons are there? Neurons can be categorized in myriad ways—by transmitter type, for example, including glutamatergic, GABAergic and dopaminergic. Or by brain region: cortex, cerebellum, retina, claustrum, thalamus, septum, tectum, and on and on. Or by morphology, connectivity or electrophysiological properties. This heterogeneity is crucial to the function of neural circuits, but defining types and understanding how they are generated is a daunting task.
In this interview, I talked with Oliver Hobert, professor of biological sciences at Columbia University, who has been studying this question for many years in the simple nervous system of the nematode Caenorhabditis elegans. Even though these tiny worms have only 302 neurons, these represent 118 distinct types; Hobert and his colleagues have been trying to figure how all these types are specified.
This work began with a variety of behavioral mutants—worms that had specific defects in chemosensation, mechanosensation or other particular behaviors. When researchers tracked down the underlying genetic variants, it turned out that many of them encoded transcription factors of the “homeodomain” family. This is where the simplicity of the worm’s nervous system really pays off. Some behavioral responses to specific stimuli, such as withdrawal from gentle touch on the snout, for example, rely on the activity of single neurons. Mutating specific homedomain transcription factors changes the “identity” of these neurons—they no longer express the full profile of “nuts and bolts” genes that make them functionally what they are. These cells are still neurons—just no longer the right type. In many cases, mutating these transcription factors is functionally equivalent to ablating the neuron altogether, which highlights just how crucial these cell-type specifications are.
Hobert has called these transcription factors “terminal selector genes.” Importantly, they don’t work alone. There isn’t a dedicated terminal selector gene for each neuron type. They work in a kind of combinatorial code, with each cell type specified by the simultaneous expression of two or three proteins from this family. This is obviously a much more efficient way to encode a large diversity of types. A different subfamily of transcription factors is responsible for making neurons more generally—the terminal selector genes act at a later stage of development to specify the type. Notably, expression of these factors is required not just when the neurons are born but throughout the life of the neuron to maintain its postmitotic identity.
This kind of encoding is also highly evolvable. By analyzing patterns of gene expression and connectivity in closely and more distantly related species of nematodes, Hobert and his colleagues have been able to track how changes in the expression of terminal selector genes relate to changes in the structural and functional organization of the nervous system. In particular, analysis of a predatory nematode, Pristionchus pacificus, revealed that even though this species has effectively the same transcription-factor-defined cell types as C. elegans, the wiring of the nervous system is extensively altered. This includes changes in the projections of axons and dendrites, directly altering connectivity, but also changes in profiles of receptors for neurotransmitters and neuropeptides, altering who is listening to whom.
Understanding how different types of neurons are specified thus illuminates the logic of how these different types and their patterns of interconnections can be tweaked by evolution. The relative simplicity of the worm has made it a powerful system for working out these principles. Not surprisingly, the same mechanisms and the same gene families are at play in the more complex nervous systems of insects and vertebrates, which have thousands of types of neurons. It therefore seems likely that each type of neuron in your own brain has its own combination of terminal selector genes, making sure it knows who it is.