Generating neuronal diversity

Listen to audio

The Drosophila optic lobes receive retinotopic inputs from photoreceptors specialized in motion vision (lamina), or color and polarized light vision (medulla). At least 100 types of neurons in the optic lobes process these inputs for extracting visual information. They are organized in 800 columns corresponding to the 800 unit eyes in the retina (ommatidia). How is this variety of neurons generated and how is retinotopy established?

Neural stem cells that produce neurons in the medulla, the main part of the optic lobes, sequentially express six transcription factors in a temporal manner. Different neurons emerge in each temporal window, therefore generating a series of 800 neurons of each type: These 'Uni-columnar neurons' are generated throughout the neuroepithelium and have a 1:1 stoichiometry with the photoreceptors that innervate the medulla. The less numerous 'multi-columnar' neurons that have larger receptor fields and are present at a lower stoichiometry with photoreceptors emerge from the same neural stem cells but differ in distinct regions of the medulla neuroepithelium. In spite of their restricted origins, these neurons still contribute to the entire retinotopic map through dispersion of their cell bodies. Therefore, the generation of 80 cell types involves the integration of temporal and spatial patterning that preserves retinotopy of neurons present at different stoichiometry.

Once generated in the larva, neurons acquire their adult functional characteristics in response to combinations of transcription factors that are expressed in the 100 types of optic lobe adult neurons. To identify these transcription factors, we usedsingle-cell mRNA sequencing to identify the transcriptome of >80,000 individual cells. We were able to group these transcriptomes into 90 clusters, 60 of which could be assigned to a known cell type. To identify the transcription factors responsible for inducing each of their specific terminal differentiation features, we generated a 'random forest' model and we could show that one transcription factors was required in many, but not all, cholinergic (Apterous) or glutamatergic neurons (Traffic-jam). In fact, the same terminal characters are often regulated by different transcription factors in different cell types, arguing for extensive phenotypic convergence. These data provide a deep understanding of the developmental and functional specification of a complex brain structure.

Once neurons are generated, they must incorporate into a neuronal network that relies on precise connectivity between neurons. Understanding the genetic control of the complexity of brain wiring during development is one of the greatest challenges in neurobiology. In the optic lobe, the direction-selective T4 and T5 neurons represent a conspicuous example of such complex retinotopic organization. Within each column, there are four subtypes of T4 and of T5 neurons that receive their inputs in the medulla (T4, which respond to the motion of a bright edges) or lobula (T5, which respond to the motion of a dark edges). Each T4 and T5 subtype responds to one direction of local motion along each of the four cardinal directions (vertical up-down and down-up, and horizontal front-to-back and back-to-front). We investigated how the identity of the four subtypes of T4 and of T5 neurons is specified and how their retinotopic organization is established. I will show that T4 and T5 neurons are produced by a specific type of neurogenesis: Vertical and horizontal motion sensitive T4/T5 neurons originate from two distinct neuroprogenitors. These divide in an asymmetric Notch-dependent manner to produce two distinct intermediate progenitors that generate neurons sensitive to opposite directions (e.g. front-to-back or back-to-front). The intermediate progenitors then divide one last time, also in a Notch-dependent manner, producing sibling T4 (^NotchOFF) and T5 (^NotchON) neurons of the same subtype. I will propose a model in which retinotopy results from patterns established by neuronal birth order. This illustrates how a complex neuronal organization can be implemented by simple developmental rules.