Here’s some medicinal chemistry combined with synthetic biology for you. Many people are used to thinking in terms of finding small-molecule probes for various cell targets, and those are valuable things. But what if you want to control a certain population of (for example) ion channels, but there aren’t any compounds that will do the job potently and selectively? One course of action is to find better compounds (and indeed, in a drug discovery project, that’s what you’d be spending your time doing). But if you’re after test systems for in vitro and in vivo biology, you have another option: if the receptors you want to target don’t respond the right way to any of the compounds you have, then find a good compound and modify the receptors so they they respond to it instead.
That sounds somewhat crazy, and what’s more, it sounds like even more work than finding a better compound. But the modularity of ion channels lets you get away with it. A team led out of the HHMI Janelia campus (among others) has been working on this idea for some years now, mixing and matching ligand-binding domains and ion-pore domains to engineer new functional ion channels that respond to different small-molecule ligands than they were originally evolved for. This latest paper focuses on mutated forms of nicotinic acetylcholinergic ion channel ligand-binding domains as the front end of the system with the 5-HT3 ion pore domain as the back end. The group screened a large number of compounds that might be good candidates (for their selectivity and pharmacokinetics) against these unnatural chimeric ion channels, and the smoking-cessation compound varenicline stood out.
A few more rounds of mutation screens got its affinity down to about a 1.6 nM EC50 for one particular LBD mutant, and they identified changes (particularly replacing a specific leucine residue in the binding pocket) that no only increased affinity for varenicline but decreased it (by up to five orders of magnitude!) for the endogenous agonist (acetylcholine) and the other standard (nicotine itself). The potency at the mutant receptor versus the wild-type nicotinic receptor was improved by almost 500x, and it was now 160x more potent at the mutant receptor than it was at its actual target (the α4β2 nicotinic receptor subtype) and 880x more potent at the mutant than at its major off-target, the natural 5-HT3 ion channel. This gives you a new and orthogonal way to manipulate ion-channel behavior in cells, using concentrations of a compound that are too low to do much of anything else, with an engineered receptor that doesn’t seem to respond to any other ligands it will encounter.
The team generated mutant-gated forms of the glycine and 5-HT3 ion channels and showed that these express in mouse brain, and that they showed the expected effects on exposure to varenicline. The chimeric glycine receptor, for example, strongly repressed neuron firing due to the intracellular changes in chloride concentration brought on by receptor activation. To their credit, though, they also tried to improve the small-molecule side of this system, synthesizing a series of varenicline analogs to try to get even higher potency and selectivity. Alkoxy substitution on on the quinazoline ring, as it turned out, let to a series of compounds with even better properties (that 160x improvement turned into a >10,000 improvement, for example). Never neglect the chemistry side of the equation, it can work wonders; although good luck getting us to tell you when any specific wonder is scheduled to show up. These compounds were clean in a screen of other GPCRs and ion channels, show good PK properties, and look to be very useful tools in the engineered cells and animals.
So this sort of thing should open up some studies of neuronal function, tied to whole-animal effects, that are basically impossible by other means. The paper goes on to speculate that one might even try therapeutic applications by introducing expression of the mutant receptors in vivo (targeted, perhaps, during a surgical procedure and aimed at localized pain, movement disorders, etc.) That’s a long way off, but the basic research applications of this sort of work are already in hand – and honestly, those are enough by themselves without wondering about eventual direct therapies. Neuron activities are enough of a black box that any new tools to help pry it open are welcome.