(284f) Single-Cell Transcriptomic Analysis of Neural Development

Background: Recent technological breakthroughs in single-cell genomics have facilitated the interrogation of complex biological tissues at unprecedented molecular detail. Here, the ability to marry high-throughput measurements with data-driven machine learning algorithms enable the inference of genetic rules that underlie not only the diversity of cell types present in tissue, but also dynamical processes such as differentiation and disease pathogenesis. Our group is applying these approaches to understand developmental questions in the brain. A central question in neuroscience is how the brain’s diverse neuronal types – defined by location, connectivity, function and molecular profile – are established via developmental processes. The vertebrate retina has been a valuable model to address this question. Lineage tracing, where the fate of a mitotic progenitor cell is tracked using an indelible marker, has provided deep insights into when and how broad classes of neurons are generated. However, the diversification of a neuronal class into multiple closely related types happens postmitotically, and is inaccessible to classical lineage tracing methods.

Goals: To address this challenge, we have used single-cell RNA-sequencing to understand the postmitotic diversification of immature neuronal precursors in two parts of the visual system, the retina and the visual cortex. Here, the identity of a neuron is associated with a multi-dimensional molecular state (e.g. expression profiles of thousands of genes). By applying machine learning methods to tens of thousands of single-cell snapshots, we can not only identify molecular states that define cell types, but also infer their process of diversification and molecular maturation. I will discuss two examples in the retina that illustrate the power of this approach that bridge both basic and translational questions.

Results: First, I will describe the process by which retinal ganglion cells (RGCs), the sole output neurons of the retina, diversify into ~45 discrete transcriptomic types from immature postmitotic precursors. We find that the number of molecularly distinct groups of RGCs as well as their transcriptomic distinctiveness increases with age. By adapting a mathematical approach known as optimal transport, we obtained evidence that RGC types are not specified at the progenitor level but rather that multipotentiality persists at the postmitotic stage in precursor RGCs. These precursor RGCs become gradually and asynchronously restricted to specific types with developmental age. We identified temporally regulated modules of genes that correlate with, and likely regulate, multiple aspects of RGC development, ranging from differentiation and axon guidance to synaptic recognition and refinement. Diversification may in many cases occur in two steps, with precursors initially committing to subclasses, each defined by the selective expression of TFs in the adult. Subsequently, precursors within a subclass, become restricted to single types by a process we named “fate-decoupling”.

Second, using these results as a foundation, we analyzed the impact of disrupting vision-dependent activity on RGC diversification. We analyzed RGC transcriptomes in mice lacking visual experience due either to dark rearing from birth or to mutations that ablate bipolar or photoreceptor cells. We find, perhaps unsurprisingly, that visual activity is dispensable for diversification as > 98.5% of visually deprived RGCs can be unequivocally assigned to one of 45 types based on their transcriptomes. However, the number and magnitude of type-specific differentially expressed genes were decreased in VD RGCs, implying that activity influences RGC maturation or maintenance. Consistent with this notion, transcriptomic alternations in VD RGCs share a significant overlap with gene modules found in developing RGCs. Together, our results provide a resource for mechanistic analyses of RGC differentiation and maturation, and for investigating the role of activity in these processes.