As neurons develop in the central nervous system, they encounter many potential synaptic partners, and choose among them to form stable connections with only a few. The specific patterns of connectivity that emerge from these choices generate the complex and stereotyped circuits that underlie our mental activities and behaviors. A leading hypothesis for how this specificity arises is that there are recognition molecules on the surface of neurons that hook up in a “lock-and-key” fashion to bias synapse formation in favor of appropriate partners. In previous work, Masahito Yamagata, in Joshua Sanes’ lab, identified two such molecules, called Sidekick1 (Sdk1) and Sdk2, and went on to demonstrate their role in formation of connections in the retina (Yamagata et al., Cell, 2002; Yamagata and Sanes, Nature, 2008; Krishnaswamy, Yamagata et el., Nature, 2015). The Sdks are so-called homophilic molecules (Sdk1 binds specifically to Sdk1) so in this case, the Sidekicks are both lock and key, promoting specific connections between neuronal types that bear the same label.
In their latest study, the team investigated the molecular basis of the specific recognition (Goodman, Yamagata et al., eLife, 2016). They collaborated with structural biologists Kerry Goodman, Barry Honig and Larry Shapiro at Columbia University to obtain x-ray crystallographic structures of Sdk1 and Sdk2, revealing the Sdk domains that account for binding and specificity. They found that Sdks interact in ways that are similar to yet distinct from distant relatives such as the contactins and dscams. The surprise came when they compared the binding of pure molecules with that observed when Sdks are in their native habitat on the cell surface. Whereas binding is strictly homophilic in cells, Sdk1 and Sdk2 can bind to each other in solution. The affinity of this “heterophilic” interaction was substantially lower than that of homophilic interactions, but seemed incompatible with the results in cells. Rather than dismiss the discrepancy, the authors tried to explain it, and in so doing, they found an unexpected new contributor to specificity. It turns out that Sdk molecules on the surface of a single cell can bind to each other (so-called cis interaction), using the same domains that they use in binding to Sdks on other cells (the trans interactions that promote synaptic recognition). In many cases the cis interactions will be completely homophilic, because any given cell type generally expresses at most one of the Sdks. So, to generate a productive cell-cell interaction, a Sdk molecule on an adjoining cell has to compete with a Sdk’s cis partners, and this competition can enhance[RE1] specifity. Thus, although Sdk1 on Cell A can bind to Sdk2 on Cell B, the interaction is too weak to pull the Sdk2 away from its cis partner; only another Sdk2 molecule can do that. Thus, the combination of structural and cellular studies has revealed a mechanism of “cis/trans competition” that is well-suited for enhancing specific connectivity in the crowded, complex neuropil of the central nervous system, where cells expressing the two Sdks are intermingled. More generally, the authors speculate that other recognition molecules may use a similar mechanism to allow cellular interactions to be highly selective even when the recognition molecules are slightly promiscuous.
[RE1]Instead of enhances