Harvard University COVID-19 updates

Department News



(L to R)Joshua R. Sanes and Jeremy Kay

The retina is a lot more complicated than a simple camera. In fact, it acts more like a parallel processing computer, with specialized neural circuits devoted to extracting particular aspects of the visual scene such as color, the flickering of light, or motion in a particular direction.

To extract all this information, the retina has over 70 subtypes of neurons that each plays a specific role in one or more of these parallel circuits. We are studying how the developing retina makes all of these different neuronal subtypes, and how they are wired together properly so they can do their appropriate job.

As a first step we developed a molecular-genetic database that allowed us to see and study many subtypes of retinal neurons.

In initial studies, we used the database to identify markers for closely related subtypes (Kay et al., J. Neuroscience, 2011) and find factors that promote their differentiation (Kay et al., Nature Neuroscience, 2011).

Now, we have gone on to address a long-standing mystery about how retinal cells are distributed in the retina. Over 30 years ago, scientists showed that neurons of any given subtype are evenly spaced over the retinal surface in pattern called “mosaics.” This makes sense – if cells of a particular type were missing from patches of the retina, we would not, for example, be able to see upward motion in the corresponding part of the visual field. So how do cells of the same subtype recognize cells of the same subtype so they can form a mosaic, while ignoring all other cells?

We now report in Nature the discovery of a pair of closely related proteins, called MEGF10 and MEGF11, that do the job for 2 of the 70 types. One cell type, the starburst cell, is required for detection of motion in specific directions. A starburst signals its neighbors by presenting MEGF10 to neighboring starburst. When MEGF10 is detected, one starburst can conclude that the neighboring cell is another starburst, so they move away from each other. The closely related protein MEGF11 works alongside MEGF10 to do this same job for a different neuron subtype, the horizontal cell. Together our results help answer the thirty-year-old question of how mosaics form during development. Because neurons throughout the brain show non-random arrangements, we speculate that similar mechanisms could be widely deployed.


Read more in Nature

[March 22, 2012]


View Joshua Sanes’s Faculty Profile