We divide optic lobe intrinsic neurons into four broad classes: columnar, local interneuron, cross-neuropil tangential and cross-neuropil amacrine (Fig. 1a–c). Cells of the columnar class (Fig. 1a) have axons oriented parallel to the main axis of the visual columns (‘axon’ is defined in the Methods). Following a previous study6, the arbour of a columnar neuron is allowed to be wider than a single column; what matters is the orientation of the axon, not the aspect ratio of the arbour. Photoreceptor cells are columnar but are not intrinsic to the optic lobe, strictly speaking, because they enter from the retina. Nevertheless, they will sometimes be included with intrinsic types in the following analyses.
Interneurons constitute just 17% of optic lobe intrinsic neurons, but the majority of cell types
Distinguishing cell types using connectivity Features based on connectivity (Fig. 2a) enabled us to discriminate between cell types that stratify in very similar neuropil layers. Stratification constrains connectivity, because neurons cannot connect with each other unless they overlap in the same layers1. However, stratification does not completely determine connectivity, because neurons in the same layer may or may not connect with each other. Classical neuroanatomy, whether based on Golgi or genetic staining, relied on stratification because it could be seen with a light microscope. Now that we have electron microscopy data, we can rely on connectivity for cell typing, rather than settle for stratification as a proxy2.
That being said, the present study used only connectivity at the final stage of cell typing, which was seeded by the morphological types identified during the first and second stages (Methods). It was possible to demonstrate self consistency of the final cell types using connectivity-based features only. We expect that it should be possible to eliminate all dependence on morphological typing, and base the approach on connectivity from start to finish. This challenge is left for future work.
Eventually morphology became insufficient for further progress. Expert annotators, for example, struggled to classify Tm5 cells into the three known types, not knowing that there would turn out to be six Tm5 types. At this point, we were forced to transition to connectomic cell typing. In retrospect, this transition could have been made much earlier. As mentioned above, connectomic cell typing must be seeded with an initial set of types, but the seeding did not have to be as thorough as it ended up. We leave for future work the challenge of extending the connectomic approach so it can be used from start to finish.
Hanchuan Peng
CBLAST heminbrain Scheffer, L. K. et al. A connectome and analysis of the adult Drosophila central brain. eLife 9, e57443 (2020).
Wan, Y. et al. BlastNeuron for automated comparison, retrieval and clustering of 3D neuron morphologies. Neuroinformatics 13, 487–499 (2015).
Gouwens, N. W. et al. Integrated morphoelectric and transcriptomic classification of cortical GABAergic cells. Cell 183, 935–953 (2020).
Zhang, M. et al. Molecularly defined and spatially resolved cell atlas of the whole mouse brain. Nature 624, 343–354 (2023).
Kalmbach, B. E. et al. Signature morpho-electric, transcriptomic, and dendritic properties of human layer 5 neocortical pyramidal neurons. Neuron 109, 2914–2927 (2021).
Neuronal cell-type classification: challenges, opportunities and the path forward. Nat. Rev. Neurosci.
Morphological diversity of single neurons in molecularly defined cell types. https://www.nature.com/articles/s41586-021-03941-1 Hanchuan Peng, Hongkui Zeng, nature 2021