A panopticon of the fruit fly optic lobe

To create the atlas, a team of researchers at New York University in the US, Umeå University in Sweden, and New York University Abu Dhabi in the United Arab Emirates sequenced the full range of gene readouts, or transcriptomes, of about 110,000 cells from the optic lobes of adult Drosophila, and another 165,000 optic lobe cells spread across five pupal stages. “This is a highly significant resource, as it covers almost all cell types and developmental cell states in the optic lobes of the brain,” says Stein Arts of KU Leuven in Belgium, who wasn’t involved in the study.

The researchers used a clustering analysis to separate the transcriptomes into 193 clusters in the adult transcriptomes. The clusters could be thought of as representing cell types, with 172 of them corresponding to optic lobe neurons, and 19 to glia, the supporting tissue of the nervous system. The researchers were able to match 61 neuronal clusters to cell types with published transcriptomes. They used the clusters to train a machine learning tool to classify the transcriptomes of the pupae. 

Analysis of these transcriptomes revealed several interesting features of neuronal development. “As a developmental biologist, being able to see how the neurons are generated and how their transcriptome changes with time was very exciting,” says Claude Desplan, of New York University, who led the study. The team found that some neuronal subtypes with different morphologies and synaptic connections have transcriptomes that are nearly identical in adult cells, but differ during development, when they established their connections.

The analysis also showed that neurons of a given type converge to a common transcriptome profile within 15 hours of birth, after which they can't be distinguished based on gene expression profiles. 

Finally, the team identified two clusters of neurons that appear transiently in the optic lobe during development, but die before adulthood. The transient clusters wrap around the dorsal and ventral side of dense networks of nerve fibres known as neuropils, but their function isn’t yet clear. “Almost everyone focuses on adult neurons, as the research tools are all made for functional studies. Our approach gave us a view during development and our algorithm was able to identify these cells,” says Desplan.

Desplan hopes their findings will be more broadly useful by giving researchers working on vertebrates a clearer picture of the events and processes to look for and study during neuronal development. “But more importantly, this might allow us to find the ‘rules’ that allow each pair of neurons to connect with one another,” he says.

Aerts sees great value in the resource generated by this study, saying that “future single-cell analyses of this system, for example using disease models or genetic perturbations, will greatly benefit from this reference atlas.”

Sign-up to receive our e-alert update every two weeks to keep up with everything new on the portal.