Morgan Sheng and Casper Hoogenraad just published a through review of recent attempts to understand the synapse at a quantitative level using mass spectrometry and electron microscopy. The focus is on the receiving end of synaptic transmission (the post-synapse). There is a highly organized, disc-shaped protein architecture just inside the post-synaptic side of a synapse called the Post-Synaptic Density. This is where neurotransmitter receptors are anchored in the membrane along with various scaffolding and signaling molecules. We learn things such as the average size of the PSD (300-400 nanometers wide and 40 nanometers thick), the average number of AMPA receptors ~15. The number of molecules per average synapse, maybe up to 1000, seems almost manageable, as though we could perhaps one day do computations and make predictions. There are two major caveats associated with the type of studies reviewed though: 1) They rely on averages rather than quantitative studies of individual synapses. As Sheng and Hoogenraad point out, there are probably synapses with 30 AMPARs and 0 AMPARs, and this is very important for function. 2) The studies freeze the synapse in time and provide no indication of the dynamic nature of synaptic molecular architecture. Hence this electron microscope-generated post-synaptic mountain range should probably look more like molten lava:

Recent advances in imaging technologies and molecular genetics in vivo have allowed studies such as the one discussed here. The amount of diffusion and protein turnover makes one despair of ever getting a handle on post-synaptic architecture. This is captured in a rather apt analogy at the end of the review:

"Even as we struggle to reach a stoichiometric and geometric description of the PSD and its constituent proteins, it is clear that we are chasing a moving target that changes rapidly and substantially in response to neural activity and developmental experience (akin to taking a census of a fluid society)."

To fully appreciate the analogy one would have to ask how much use a census is for a society. We know how much good it did David. I think one of the major uses of the census is to identify global trends across decades. Here the metaphor fails a bit because we have to destroy the synapse to identify its molecular constituents, but it does point toward a useful experimental design. The census of synaptic components must be taken over time to observe the evolution of synapses in response to stimuli or development. Some developmental data exists, but I would very much like to see a fine scale time course of PSD content following LTP induction. Easy for me to say; if it were really easy it would've been done already.