Many of us who started to study neurobiology before the domination of the molecular approach will surely remember the classic electron micrographs of synaptic contacts, which taught us about their asymmetry, and about the distinction between pre- and postsynaptic compartments. The electron-dense postsynaptic density (PSD) characterized the postsynaptic bouton, whereas the presence of synaptic vesicles was indicative of the presynaptic element. In addition, a conspicuous element of the presynaptic terminal was always present in the diagrams of that age: a series of 'pyramids' linked by some kind of web, which were supposed to dock synaptic vesicles near their fusion sites. Many years later, the morphological analysis gave way to the identification of the molecular components of the synapse, and those old textbook diagrams were replaced by more detailed depictions that incorporated the actual molecules of the PSD and the synaptic vesicles. But the pyramids faded into the background and were never really incorporated into our modern view of the synapse, mainly because little was learned about them during that period. Phillips et al. might trigger a revival of the interest in the presynaptic web after their report in Neuron on new data about its molecular composition.

In synaptosomal preparations, the pre- and postsynaptic elements can remain attached by links that are poorly characterized. Phillips et al. identified conditions in which this macromolecular complex remained assembled after membrane solubilization. The resulting particles contained presynaptic material with the characteristics of the presynaptic web, which was connected to the PSD through protein filaments. They went on to show that it was possible to disassemble and reconstitute the presynaptic web by varying pH. Using this system, the authors identified some of the proteins that formed the web, and found several molecules that are known to participate in synaptic vesicle fusion and recycling — clathrin, dynamin, NSF and UNC18. They also found that it contained adhesion and adhesion-related molecules, such as N-cadherin and β-catenin; these molecules might mediate the physical interaction between the presynaptic web and the PSD.

So, the synaptic junction is much more than just two membranes in close proximity; it is a complex protein assembly in which pre- and postsynaptic elements are bridged by a discrete molecular scaffold. Just as the molecular analysis of the PSD has shaped our understanding of the postsynaptic compartment, the findings of Phillips et al. should help us to develop a more sophisticated view of the presynaptic terminal. We can finally start to do justice to a familiar figure that appears in some of our oldest textbooks.