Generating neuronal diversity in fly eyes and human retinal organoids

A central challenge in developmental neurobiology is to understand how the incredible diversity of neuronal cell types is generated. My lab studies this question in the visual systems of flies and humans. By studying highly divergent organisms, we aim to identify fundamental mechanisms that diversify neuronal function during development.

The random mosaic of photoreceptors in the fly eye

“I, at any rate, am convinced that He does not play dice.” These famous words were spoken by Albert Einstein about the existence of an underlying reality and the predictability of the universe. A simple look at a pair of twins would back the notion that nature and biology can be amazingly reproducible. However, one would only have to look into the eyes of these twins to see something remarkable: the different types of cells that detect light are randomly distributed and thus each twin is unique. Only recently have we begun to understand the mechanisms controlling such stochastic developmental events. It appears that noise in molecular processes creates variation, and this variation can be exploited to diversify the functions of cells. Stochastic specification is critical for generating a wide range of cell types, from human cone cells that detect colors, to olfactory neurons that sense odors, to B cells required for immune responses. Beyond its role in normal development, stochastic gene regulation can determine whether a person with a mutated gene will suffer from disease. Despite its importance, very little is understood about how this fundamentally different strategy operates.

We are using the fly eye as a paradigm to elucidate the mechanisms controlling stochastic gene expression during development. Similar to the human color vision system, the photoreceptors of the fly eye randomly express several light-detecting Rhodopsin proteins (Rhodopsin3 =blue; Rhodopsin4 = red). The fly eye is an ideal system to study this phenomenon because it provides a simple binary output for stochastic gene expression, the general mechanisms of cell-fate specification are well-understood, and a vast array of genetic and transgenic tools are available to manipulate cis-regulatory inputs and upstream trans-acting factors.

061714 website fly eye

The transcription factor Spineless is the critical regulator controlling the random mosaic pattern of photoreceptor subtypes in the fly eye. Each allele of spineless makes its own random expression choice independent of the other. Stochastic on/off expression of spineless is determined by general activation coupled with random repression requiring combinatorial inputs from cis-regulatory elements acting at long range. Through interchromosomal communication, the two alleles coordinate their expression state.

Project 1. Stochastic photoreceptor specification (Caity Anderson, Liz Urban, Luke Voortman, Mini Yuan, Grace Gu): How does a gene randomly decide to be on or off? To address this question, we are studying how DNA looping, nuclear architecture, chromatin state, and transcriptional regulators control stochastic expression of spineless. We found that a two-step mechanism involving transcriptional priming and chromatin compaction determines stochastic expression of a transcription factor that controls patterning in the retina. This mechanism may represent a general paradigm for gene regulation during development.

Project 2. Nuclear architecture and interchromosomal gene regulation (Liz Urban, Jeong Han, Adrienne Chen): Very little is known about how long range interactions of regulatory DNA elements control stochastic gene expression. Our studies of nuclear architecture focus on the pairing of homologous chromosomes in somatic cells, which enables gene regulation between chromosomes. We found that chromatin structures called topologically associating domains (TADs) and clusters of DNA-looping insulators button chromosomes together to promote interchromosomal gene regulation. Our findings highlight how distinct elements in the genome drive physical interactions between chromosomes to regulate gene expression.

Project 3. Innate color preference (Natalie Roberts): Flies have innate attraction to different colors. We are investigating how natural variation in color photoreceptor specification and environmental perturbations impact innate color preference.

The mosaic of photoreceptors in the human eye

The human retina is composed of three cone cell subtypes. These cone cell subtypes are defined by expression of color-detecting opsin proteins: L/red, M/green, and S/blue opsins. Despite their importance in color and daytime vision, very little is understood about how these three photoreceptor subtypes are generated. Since mice and fish have dramatically different patterns of cone cells, we address this question in developing human tissue. We are studying the question of cone cell specification by differentiating human stem cells  into retinal cups. These retinal organoids contain all the cell types of the human retina including opsin-expressing cone photoreceptors (green = L/M opsins; blue = S opsin).


Project: Human cone subtype choice (Kiara Eldred, Sarah Hadyniak, Kasia Hussey, Christina McNerney, Aki Sogunro, Raphi Chernoff): In humans, three subtypes of cone photoreceptors enable trichromatic color and high acuity vision. To overcome the challenges associated with studies of human development, we utilized a human organoid system that recapitulates retinal development and photoreceptor specification. Human stem cell-derived retinal organoids are an ideal system to study cone subtype specification due to their limited number of cone cell subtypes, arrangement in a monolayer for simple quantitative analysis, thin tissue depth allowing live visualization, and tractability for CRISPR-mediated genome engineering. We found that spatiotemporal regulation of thyroid hormone and retinoic acid signaling specifies cone subtypes in human retinal organoids. Our studies advanced human retinal organoids as a model for revealing mechanisms of human development, with promising utility for therapeutics and vision repair.

Retinal ganglion cell (RGC) subtype specification

Retinal ganglion cells are the neurons that connect the retina to the brain. The advent of stem cell-based organoid technology has enabled the interrogation of human biology at the molecular level for the first time. Direct study of developing human tissue will increase our understanding of the mechanisms that control normal organ development, what goes wrong in developmental disorders and disease, and how to manipulate these systems for therapeutic applications. Here, we study in retinal organoids to understand the generation and death of RGCs. Elucidating the molecular and developmental mechanisms that specify retinal ganglion cells will be critical for understanding and treating glaucoma.

One of the goals of stem-cell derived organoid studies is to influence specification pathways to enrich for specific cell types, while taking advantage of endogenous developmental mechanisms. Organoids could provide a rich source of RGCs, but we must first understand how to drive these fates and ensure that these cell types are properly generated to provide the highest probability of successful transplantation/integration.

Direct study of RGC death in humans has been limited by the absence of a model of developing human retinal tissue. As in glaucoma, RGCs die in human retinal organoids, likely due to innervation problems. We propose that human retinal organoids provide a model system to study RGC death on the cellular and molecular level due to their amenability to genetic and chemical perturbations. These studies will not only yield information about how RGCs die, but also potentially provide strategies to specify and maintain RGC populations for regenerative applications.

Project: RGC subtype choice (Brian Guy, Yang Liu, Joyce Wang): We have learned a great deal about the generation of RGCs through the study of model organisms. Human retinal organoids enable, for the first time, the study of RGC specification in developing human tissue. Retinal organoids generate RGCs, yet they are poorly characterized. Our preliminary data suggest that RGCs are the first neurons specified in organoids, consistent with observations in human retinas. Interestingly, recent studies have suggested that there are 20 subtypes of RGCs in humans. Do retinal organoids generate all 20 subtypes? How? If not, why? We are using fluorescence-based cell sorting combined with single cell RNA sequencing to address these questions.