Serotonin is perhaps the only neurotransmitter molecule that you could find named in a random poll, thanks to its association with antidepressants. (That association is far messier than popular opinion realizes, but that’s another topic). It’s a complicated one to have embraced, that’s for sure. There are 13 subtypes of GPCR serotonin receptor in six different families, and one (5-HT3) that’s actually an ion channel instead. All of these are scattered in what must be quite meaningful patterns across different cell and tissue types, if only we understood what they meant, and in various (and definitely non-random) associations with other neurotransmitter receptors at the cell surface.
As far as I know, this is the only GPCR ligand family that has an ion channel in it as well (update: forgot about the GABA, glutamate, and acetylcholine ones! Being an idiot is hard work). 5-HT3 is involved, from what we know so far, in processes as varied as memory, anxiety, learning, and nausea, and there’s a whole list of drugs that are known to bind to it. But the physical picture of that binding hasn’t been too clear. And while there are other ion channels broadly like this one (pentameric ligand-gated), and there are other structures known, they aren’t enough to really say what’s going on.There seem to be multiple binding modes and multiple states of the ion channels to go along with them. Well, put this down as another success for cryo-electron microscopy: a new paper reports four (!) structures with various ligand combinations.
This isn’t the microcrystal electron diffraction (microED) sort of experiment; this is the blast-single-protein-particles-in-different-poses one. (As a side note, we’re going to have to make sure that these two experimental techniques don’t all end up being called “cryo-EM”, because they really are different). The first of these was done with the antagonist tropisetron, and this (the “T form”) came out pretty similar to the known X-ray structure for the unliganded channel. The ligand fits into a pocket, but it doesn’t change the protein conformation much.
How about an agonist, then? How about. . .serotonin itself? This experiment shows off one of the real advantages of cryo-EM. As the team analyzed the particles, scattered over the EM grid in all sorts of arrangements, they realized that they were looking at a mixture of two different ligand-bound forms. Binning these and solving them as separate problems gave each of them as individual protein structures. If you had to grow a crystal under these conditions, you’d almost certainly be in trouble. The extracellular part of the receptor looks pretty similar to the T form, and serotonin is in the expected neurotransmitter binding site. But the transmembrane domains are very different. It looks like one of the forms (the “I form”) is an intermediate on the way to the other, the full agonist “F form”. The intermediate form has some transmembrane helix positions shifted from the T form, but the F form is a wholesale rearrangement, with bundles of helices sliding over to new positions. In addition, the intracellular part of the protein becomes too disordered to get a good read on it; that’s apparently a very flexible region.
This subtype is known to show a lot of allosteric effects as well – signaling that’s modulated somehow by molecules that aren’t binding at the canonical ligand site and changing the protein conformation. The team tried out the receptor protein in the presence of both serotonin and TMPPAA, a known (and confusing) compound that can display both agonist-like and antagonist-like behavior, depending on the conditions. This structure gives the clearest electron density of any of the four structures, from what I can see, and serotonin is very well defined in its binding site. The extracellular domain of the protein looks like the two previous serotonin-bound forms, while the transmembrane region resembles the intermediate form, but with some differences.
The tricky/annoying part is that there’s no assignable electron density for the TMPPAA molecule, even though it seems to be having an effect on the structure. Good old single point mutation analysis suggests that the compound binds in a region between several protein subunits, down in the transmembrane domain – it would have been nice to nail that down by EM, but it was not to be. It looks like there are several potential allosteric pockets, whose sizes and shapes vary as the transmembrane helices slide around, so getting detailed structures on those will be a tall order.
Comparing these four structures, though, gives new insights into how the receptor rearranges as it goes between unbound, antagonist, and agonist-bound states. Remember, the whole point of these multimeric ion channels is that there’s a pore in the middle of those subunits, and it can open and close in response to ligand stimuli. These studies are certainly not going to be the last word – as the authors put it, “The challenge of matching structures to states without ambiguity transcends the present study and pertains to the whole field of (pentameric ligand-gated ion channel) structures“. That it does! And that’s the tricky part about X-ray, EM, and ED structures in general – you can see what’s moved and what’s next to what, but you have to infer what’s going on from those static pictures. It ain’t easy. It’s like trying to reconstruct the NFL rule book from looking at lots of static photos of football plays. (In fact, it’s so much like that, that I can’t believe that analogy just dropped into my head – feel free to use it at any time).
But even so, comparing these results to the other ion channel structures begins to make some major motifs clear. There are particular twists and rearrangements in the transmembrane domain that seem to be broadly conserved (things that evolution hit on a while ago?), and piling up information like this is the only way we’re going to be able to figure out what’s going on. That is, until someone invents the single-molecule femtosecond X-ray laser device and we can just start snapping frames for the video presentation (not holding my breath on that one, but hey). But when I think of what was known about such things when I started in medicinal chemistry, what we have already is pretty impressive, and there’s more to come. . .