One would like to be able to reach into a cell and mess around with its functions in real time. Thanks to CRISPR and other gene-editing technologies, we can (more or less selectively) tweak individual genes, to a wide number of interesting effects. What if that gene just disappears? What if it gets expressed even more? How about replacing it with a protein whose active site has been modified? These are excellent experiments, but they’re still not real-time enough for many applications, although certainly easier to interpret than the mouse-knockout Cre/Lox technology of the older days. With those knockouts, you were raising mice all the way from embryos, letting the entire system compensate as best it could for your disruption. CRISPR, etc. can at least be done with individual cell lines, which lets you isolate many more variables, but it’s still a change back at the genetic level. What about something a little bit closer to the downstream machinery?
Small molecules can let us do that in some cases – it’s like pointing at a particular binding site and suddenly occupying it (usually clogging it up) and watching the effects. Vast amounts of information have been obtained that way, of course, especially since such inhibition can be put to multiple uses. You can gum up the cell’s ability to clear out proteins (proteasome inhibitors), mess around with various phosphorylation signals (kinase inhibitors), and so on. But we’re getting the ability to do other sorts of changes. Targeted protein degradation, for example, is like pointing at a particular protein in the working cell and saying “Hey you – disappear for a bit and let’s see what happens”. That obviously can have different effects than just a small molecule binder, since proteins have many different functions and binding partners, and have different effects than a CRISPR knockdown, since TPD comes out of the blue from the cell’s perspective, rather than being something that it’s had to grow up with, so to speak.
Then there are techniques for suddenly making a protein (or some of its activities) suddenly appear. Here’s an example of that in a new paper. A group at the University of Washington demonstrates a new system for “chemically disrupted proximity”, using a system imported from the hepatitis C virus. That viral protease (NS3a) is known to bind with high affinity to a short protein inhibitor (called ANR by this team, who developed it for these sorts of uses), and this interaction in turn is disrupted by the small-molecule protease inhibitor drugs such as danoprevir, grazoprevir, and asunaprevir. (A look at those structures will show that none of them are particularly small, but in comparison to the protease itself, anything we medicinal chemists make is small).
So what you do is express some cellular protein of interest with that short ANR peptide coming off one end of it, and you also splice in the viral NS3 protease. That gives you cells with your target protein stuck to NS3, and you can release it by adding the small-molecule protease compound. If you have membrane-targeted NS3 protease, that drags your target protein along with it, and it only breaks free to find its usual place on addition of the drug. You can fuse the NS3 protein, via a linker, to the target protein itself (and have the ANR on the other end of it, so that it’s biting its own tail) or have them separate. Using a variant of the NS3 protein that localizes to mitochondria (and an appropriate ANR variant that binds to it), they were able to show that addition of the small molecule ligand disrupts the system within five minutes. Localized protein on the cell membrane was just about as fast, while the nuclear-localized system took longer (but still within an hour).
And in the same way that the other methods for protein function disruption can be adapted to other processes, the team tried messing around with transcription by using the nuclear-localized system with ANR attached to a transcriptional activator (VPR). In this case, the cells that had a fluorescent protein (mCherry) reporter gene, the localization of that ANR-VPR to the nucleus actually cranked up transcription of the red protein (the two were functional when colocalized), and addition of the small-molecule ligand then sent that transcription back down to background levels when the VPR was no longer being forcibly recruited. A further experiment showed that this same technique could be applied to CRISPR enzymes, giving you another method of controlling that system.
This adds to a growing list of such chemogenetic controls. The NS3 system has been a popular proving ground for that sort of thing – some groups have hijacked the protease activity itself for this purpose in increasingly complex ways. Using a completely different system, there are also reports of using light-induced protein affinities to control activity and association. There will be more. We’re heading for just that wish list mentioned in the opening paragraph – the ability to reach into cells and move turn all those controls, those thousands and thousands of small knobs and switches, at whatever level we want: from-the-ground-up genetic, real-time functional, etc. The amount of information that all this will generate is going to be staggering, and there will be ridiculous complications along the way for sure, but the end of it may be a complete mechanistic picture of a living cell. What a sight that will be. . .