A key method for studying evolution is to compare genome sequences of many species; one can ask which genes are conserved, lost, duplicated, or altered. Genes, of course, code for proteins, and the proteins build cells. The cells can be grouped into types, which differ among and within tissues. In the mouse brain, for example, the latest estimate is that there are over 5000 cell types, each with a unique molecular architecture, structure, function and distribution. One might therefore imagine that, by analogy to genes, one could follow the evolution of cell types among species, asking which are conserved, which are new, and just how conserved the conserved types are. This has been a challenging problem, but recent advances in high throughput RNA sequencing (scRNAseq) has provided a new way to tackle it. In this method, gene expression is measured in thousands of cells at relatively low cost and high speed; computational methods are then used to group the cells into types; the validity of the classification can then be assessed in a variety of ways.
In their new study, Hahn et al. use this method to study the evolution of cell types in the retina, a particularly accessible and well-studied part of the brain. The retina is a good subject for evolutionary analysis because its overall structure is highly conserved. In all vertebrates studied to date, there are 5 neuronal classes arranged in three cellular layers separated by two synaptic layers, with information flowing from outside (photoreceptors) through interneurons to retinal ganglion cells (RGCs), which send their axons through the optic nerve to the rest of the brain. Some of these classes are divided into subclasses and all are divided into types – a total of around 130 in mice. There is also an endogenous glial type, and other glia that migrate into the retina during development.
This shared plan allowed a comparison of cell types within classes and among species. The work was led by Aboozar Monavarfeshani, formerly a postdoctoral fellow in the Sanes lab and currently a senior scientist at Kate Therapeutics; Karthik Shekhar, a former a postdoctoral fellow and now an Assistant Professor at UC Berkeley; and Joshua Hahn, a graduate student in the Shekhar lab. They used single cell transcriptomics to generate retinal cell atlases from 17 vertebrate species, including humans, monkeys, pigs, ferrets, squirrels, mice, chickens, lizards, zebrafish and lamprey. They then compared cell types across species, revealing numerous orthologous types shared across phylogeny.
The comparison led to several novel insights. One is that cell types of the classes that populate the outer retina (photoreceptors, horizontal cells and bipolar cells) are more conserved than RGCs, suggesting that evolution acts preferentially on those cells that pass information from the eye to the rest of the brain. These differences may help explain how the retina of each species is optimized for its visual needs.
Another discovery illuminated the origins of the major human retinal ganglion cell (RGC) types. These cells, called “midget RGCs” because of their small size, comprise 90% of human RGCs with the other 15 types together making up only 10%. Until now, no orthologues of midgets had been found outside of primates, leading to the suspicion that they are a primate innovation. This of course has limited the ability to analyze them in, for example, mouse models of human retinal diseases such as glaucoma, which is the leading cause of irreversible blindness worldwide and results from loss of RGCs. However, our evolutionary comparison revealed orthologues in most mammalian species. Remarkably, the mouse orthologues, called “sustained alpha RGCs,” are large and comprise only 2% of mouse RGCs. The striking differences between alphas and midgets presumably explains why the relationship hadn’t been noted. Tracing their evolution suggests that these types became smaller and more numerous in parallel to the enlargement of the visual cortex, which is the predominant center of visual processing in humans. This correspondence suggests that retina and cortex evolved in parallel to provide primates with high acuity vision.
A next step will be to dig deeper into the gene expression patterns that characterize shared and unique retinal cell types. An initial result is that numerous transcription factors that have been implicated in retinal cell type specification in mice are highly conserved, suggesting that developmental mechanisms are evolutionarily ancient. In a continuing collaborative effort between the Sanes and Shekhar groups, we are now looking for genes that account for the remarkable alterations in size and abundance of RGC types across phylogeny. More generally, methods developed in this work may be useful for tackling cell type evolution in other brain regions.
by Josh Sanes
