Department News



(l to r) Xin Duan, Arjun Krishnaswamy, and Josh Sanes

The aim of developmental neurobiology is to learn how the nervous system is built, and to find the developmental anomalies that are thought to underlie some brain diseases.  The past few decades has seen incredible progress in elucidating the mechanisms by which neurons form and differentiate; now attention is turning to the ways in which they wire up to assemble the complex circuits responsible for all of our mental activities and behaviors. Our lab uses the mouse retina to study this issue. As MCB Professor John Dowling has noted, it is a particularly accessible part of the brain.  In addition, we actually know what the retina does, so we can relate developmental processes to adult function –quite different from, for example, the cortex, cerebellum and hippocampus, whose functions remain obscure.
In the retina, the output processes of approximately 70 types of interneurons intermingle with the input processes (dendrites) of approximately 30 types of retinal ganglion cells (RGC) in a narrow synaptic layer called the inner plexiform layer (IPL).  During the first few postnatal weeks, these processes make stereotyped and specific choices among potential partners to form the complex circuits that process visual input, endowing each type of RGC with the ability to detect a specific feature, such as a color contrast or an edge.  The RGCs then send this information to the rest of the brain for further analysis. How are these choices made? One clue is the observation that the IPL is divided into ~10 narrow sublaminae, with processes of most cell types restricted to just one or a few of them. In this sense, the IPL has been likened to a club sandwich, with different ingredients in each of its many layers. The restriction of processes to sublaminae limits their choice of synaptic partners to a fraction of the total, thereby contributing to specific connectivity.
In their new study, postdoctoral fellows Xin Duan and Arjun Krishnaswamy asked how laminar restriction occurs for inputs to RGCs that respond selectively to motion in particular directions. First, they identified two subtypes of interneurons, Type 2 and Type 5 bipolar cells, that deliver visual input to these RGCs.  Next, they devised genetic methods that allowed them to selectively mark these cells with a fluorescent protein or to selectively manipulate them by expressing genes of interest within them.  Then, they went looking for recognition molecules that could mediate their specific connections. Work by a former graduate student, Irina de la Huerta (a co-author of the paper) led them to two, called cadherin 8 and cadherin 9:  Type 2 bipolars express Cadherin 8 and Type 5 bipolars express cadherin 9.  By deleting the cadherins from the cells in which they are normally expressed (using “knock-out” mutant mice) or by expressing them in the “wrong” cell type, they found that the cadherins instruct the bipolars to send their axons to appropriate sublaminae. The cadherins may also be required for forming functional connections once the laminae form. Finally, they showed that mistargeting affected visual function:  the direction-selective RGCs could no longer respond properly when the cadherins were deleted, and the bipolar cells made synapses on improper targets when the cadherins were misexpressed.
The results of Duan and Krishnaswamy are interesting in several respects.  First, the family of around 20 cadherins have been known to exhibit complex, dynamic and combinatorial patterns of expression in many parts of the brain, leading to the idea that they comprise an “adhesive code” underlying specific connectivity.  This new work provides strong support for this attractive but hitherto speculative hypothesis, and paves the way for tests in other parts of the brain, where these and other cadherins are expressed.  Second, returning to the value of the retina for analyzing circuit assembly: it is already clear that laminar specificity is not the whole story.  For example, neurons must make further choices within a sublamina. It will now be possible to ask how the cadherins interact with other recognition molecules to establish connectivity.  Finally, there is a growing conviction among neuroscientists that some behavioral disorders such as autism and schizophrenia are developmental disorders that arise from some sort of miswiring of brain circuits.  As we learn how circuits wire normally, we will finally be in a position to ask how they miswire, first in animal models of behavioral disorders and ultimately in people.  

Read more in Cell or download PDF

View Joshua Sanes’s Faculty Profile