Cystic fibrosis (CF) is a cruel disease whose genetic cause has been known since the late 1980s. The CFTR protein is a chloride channel, a transport protein that moves chloride ions out of epithelial cells (and whose actions also inhibit the uptake of sodium ions by another such channel). There are hundreds of known mutations in it, many of which lead to cystic fibrosis of varying severity. Probably the most common is “delta-508”, where a phenylalanine residue is skipped entirely. The resulting CFTR protein doesn’t even fold properly, so its function is severely diminished. This leads to the classic CF phenotype, thickened mucus in the lungs due to the dysfunctional epithelial cells in the airway lining that should be moving it along and clearing it, but can’t. There are other problems, including altered pancreatic function, but the main thing that everyone associates with cystic fibrosis is lung trouble. Lung infections that lead to permanent damage are a common problem, and over the years one of the standard treatments, if you can call it that, has been to help clear the airways by flipping the patient head down and beating them on the back with a stick.
This is what we call “unmet medical need” in this business. But despite the pinpointing of the exact protein responsible, coming up with a better treatment based on this knowledge took until about 2012, with the advents of “potentiator” and “corrector” molecules from Vertex. (That’s an object lesson in why knowing the target is not the same as being able to do anything about the disease). The potentiators cause the channels to show increased function, and the correctors increase the number of CFTR proteins that make it to the cell membrane at all. Around the time that the potentiator ivacaftor came to market, its mechanism was being worked out – it binds directly to the phosphorylated CFTR protein at an allosteric site in a way that increases the chances that its chloride “gate” region is open and functional.
But the mechanism for the corrector molecules (such as lumacaftor) has been harder to pin down. This new paper, though, looks like the answer (and includes a long list of references from the past few years from other groups that have worked on the problem). The authors show that (as many had suspected) these compounds do bind directly to the CFTR protein. By some very nice cryo-EM structural biology work, they show that the binding site is in the transmembrane domain of the channel, and that when a corrector molecule fills this spot it stabilizes four transmembrane helices that are otherwise not energetically favored to stay in the right conformation. This happens early in the protein’s lifetime and allows it to continue being processed in the endoplasmic reticulum, instead of piling up there in misfolded forms and being degraded. At right is an illustration from the paper; the lines are roughly where the cell membrane sits when the protein is in its correct location, and you can see a lumacaftor molecule over on the right-hand side in that region.
When you step back and look at the disease and at these therapies, it’s a remarkable picture. There are uncounted thousands of mutations that can spring up in the proteome that are completely silent – all of us have them. But in this case, loss of a single amino acid in a single protein is enough to lead to a terrible, life-shortening disease, but a small molecule drug – if it binds in just the right place at the right time – is able to tip the thermodynamic balance for that far larger protein over to a state where it has a chance to actually be functional again. In practice, most patients get a dual-therapy regime of both a potentiator and a corrector, and the fact that both of these can work at the same time in patients is not something that anyone could have taken for granted, either. It has to be emphasized that these compounds were arrived at by relentless screening efforts and a great deal of chemical optimization – there is really no way at present that one could have predicted ab initio that either mechanism would work, or that either mechanism even existed at all.
This new paper is also something of a triumph for cryoelectron microscopy, too. Cryo-EM has been growing in importance over the last few years, and when you see something like this you can understand why. This is a transmembrane protein, as mentioned, and structures for these have traditionally been very difficult indeed to determine by x-ray crystallography (practically impossible, in many cases). Cryo-EM looks at single protein particles, though, one at a time, and assembles these data into structures, so a regular crystalline arrangement isn’t even part of the workflow. In this case, this group looked over hundreds of thousands of individual CFTR protein particles (or rather, their software did!) and binned these into different views as the protein sat on the solid surface. They obtained thousands of detailed data sets from these, and were able to fit all this into coherent structural models, the best of which are below 3Å resolution. Finding small-molecule ligands in cryo-EM protein structures is never a sure thing (at least it isn’t in 2022), so being able to pick out the lumacaftor density and work out its interactions with the protein is quite a feat. And the hardware and software just keep on improving. . .
