Probing the dynamics of agonist drug interaction with Cys-loop channels by single-molecule recording

Ion channels are proteins that act as "nanoswitches" to translate voltage or chemical stimuli into electrical currents. They are essential to many bodily functions, including information processing in neurones and cell-to cell communication at synapses outside the brain which allow nerve impulses to move muscles and regulate blood pressure and heart rate. Unsurprisingly, inherited channel mutations can produce human disease, from cystic fibrosis to neurological conditions.

Channels are targeted by many drugs. The nicotinic-type channels we work on mediate the effects of sleeping pills, drugs for epilepsy, nicotine, alcohol, insecticides and antiparasitic drugs. If we were better at designing drugs to activate or modulate nicotinic channel function, they could be useful in other hard-to-treat conditions, from chronic pain to spasticity after stroke, especially if we could exploit the great diversity of channel subtypes to design agents selective for single subtypes.

To this day, drugs are discovered by large scale screening of chemicals to find if they are effective on a human drug receptor. The process is expensive and wasteful, and few new drugs become available every year. Instead of that, it would be ideal to be able to design chemicals to have a specific action on a particular human channel. For this we need to understand two steps: how a chemical binds to the channel and how it then changes the shape and function of the channel protein. Part of the problem is that, working as a switch, the channel itself changes shape and we don't know how this affects the binding of the drug.

This is what we want to find out. At UCL we perfected a technique to see and interpret the tiny current (more than a billion times smaller than the current in a kettle) produced by one channel protein. This analysis is the only one that can tell us how tightly the drug binds to different states of the protein and how quickly the protein moves between these different states, with and without the drug.
We will use in our work the glycine channel as a model for the nicotinic family. This channel has ideal properties for single molecule recording and has recently allowed us to see why some drugs are less effective than others in turning the receptor on, a result we found to be applicable to other nicotinic channels. We will extend our work and obtain these measurements for chemicals that activate the receptor, and differ from each other in their chemical structure in a systematic way. We will also change the protein itself, by mutating appropriate positions. Combining this information will allow us to see where the drugs "touch" the protein most closely.

Glycine channels are also the mammalian channel that is closest to a well-resolved X-ray structure (that of an invertebrate channel, GluCl, published in June 2011). This makes it possible to use the structural data in relation to channel function. At Oxford we will model the structure of the glycine channel by homology to GluCl by computer calculations and use this work to plan and interpret the experiments on channel function in terms of channel 3-D shape.

Ultimately our work should lead us to understand what features in a chemical determine its affinity and efficacy for a nicotinic channel and how the different parts of the channel move with activation. It should give us indications on how the structure of drugs should be modified, in order to make them more effective. Hence this fundamental research will be useful to lay the basis for future drug development and hopefully enable rational drug design, in the glycine receptor itself (a therapeutic orphan) and in the nicotinic superfamily as a whole. Our results will also help our drug industry colleagues interpret their data that come from the quick assay techniques used by in high-throughput screening of libraries of compounds, such as binding and macroscopic functional measurements.

We will carry out ultra-low noise single channel recording of recombinant glycine channels activated by a series of agonists (with and without appropriate channel mutations). Kinetic analysis by time course idealization and global fitting of detailed activation mechanisms will be used to obtain microscopic equilibrium and rate constants, including the agonist microscopic affinity for the different states of the receptor.

We will use in silico computational modelling, both to help us choose the agonists and the mutations to be tested and to interpret our results. Our homology model will be based on the structure of GluCl, a nematode channel 34% homologous to the glycine channel. We will use molecular dynamics to explore the conformational behaviour of the receptor, particularly the binding site. We will also dock series of ligands to the receptor to examine the conformational dependence of the bound states.

Docking trials, whole-cell recordings and maximum open probability measurements will be used to select agonists suitable for full characterization, focussing on series of compounds, each series introducing small systematic structural differences in one moiety of the agonist molecule. If contaminated by glycine, agonists will be purified before experimental use.

We will also mutate the channel, choosing, from the literature and our in silico work, binding site residues likely to interact with specific parts of the agonist molecules, and repeat the agonist experiments with these mutant channels. Thermodynamic cycle analysis will be carried out on equilibrium dissociation constant values obtained by single channel recording to map the direct interactions between agonist molecule and channel at rest and after activation. Should a particularly informative agonist or mutation give rise to data beyond experimental bandwidth, we shall engineer appropriate compensating background mutations in domains distant from the binding site to allow data collection.

Understanding how ion channel molecules function is basic research: this does not mean that it has no impact on our economy or health, but simply that this impact takes time to come to fruition. There are many reasons why the work we plan will be useful to the future well-being and wealth of our society.

Our work on the nicotinic group of channels has already produced important information about how drugs act, and why some agonists are more effective than others on the same target (it is to do with the initial conformational change in the protein). We plan to continue and exploit our findings, extending them to a series of chemically-related agonists, to understand channel activation and drug-protein interaction in greater detail and depth.

We expect the main non-academic beneficiary of our work to be the pharmaceutical industry (development of new drugs, improving the selectivity of existing ones) and the agro-chemical industry (control of insect and nematode pests with nicotinoids and avermectins). Nicotinic channels are the target of many therapeutic drugs (sleeping pills, neuromuscular blockers, antiepileptics, ondansetron, antiparasitic drugs such as ivermectin, drugs to facilitate smoking cessation) and of drugs of abuse, as they mediate the effects of nicotine and some of those of alcohol.

In time our results will help in the design of new drugs, hopefully achieving greater specificity for particular receptor subtypes. At the moment, drugs are developed at vast expense, mostly by screening large numbers of compounds and using large numbers of experimental animals. It is getting more and more expensive and difficult. Only a handful of drugs have been discovered by designing them to fit a particular protein target. For channels, the problem is not that we don't know which target to go for, but that we don't understand channel function well enough, especially with regards to how they change in shape when they are functioning in the body, be it healthy, be it sick. It will take a long time to get to design drugs mostly in silico, but our sort of data is precisely what is needed in order to make it eventually possible. The UK has a big tradition and massive knowledge in drug discovery, partly because much of the basic science discoveries occurred here.

Because of the reasons outlined above, the basic research we plan will in the long term benefit our society through its impact both on human health and well-being (drug discovery, better understanding of physiological and pathological processes) and on economic productivity (pest control in agriculture).
Naturally, the timescale of these outcomes is long, given that the development of a drug takes ten years or more, as shown by an example from the work of one of us (LGS). In 1996 the PI participated to the discovery of the sodium channel NaV 1.8, by carrying out its first electrophysiological characterization (Akopian, Sivilotti & Wood, Nature 379, 257-262). Because of the discrete expression of this protein, in nociceptive pathways, it was obvious that blockers of this channel could be selective analgesics. It took 10-11 years for selective blockers of this channel subtype to be described. Work in rodents now shows that this channel is indeed a good target for analgesics, but a drug suitable for humans has not emerged yet.