With all the work going into targeted protein degradation now (recent review), we’re discovering a lot of things about it that weren’t apparent at first. To pick an obvious one, these things have several steps in their mechanism (binding to the target protein, binding to a ubiquitin ligase to form a ternary complex, ubiquitination of the target, and its subsequent degradation), and there can be interesting complications at several of these. To make matters more interesting, it’s not like we have solid assays to monitor every step along the way, either.
And here’s another: resistance to degradation. Yep, especially if you’re targeting a bunch of fast-growing tumor cells, you can expect some of them to stumble into ways to get around your TPD strategy and become the foundation for a whole new generation of cells that will ignore your therapeutic strategy. This paper from Roche is a look at this process in action (edit: see also this one from AbbVie from July). They used some (by now well-established) bifunctional degrader molecules for the BRD4 protein target, with the JQ1 compound as target ligand and with tail groups recruiting both CRBN and VHL (by far the two most common ubiquitin ligase complexes that have been used in the field). And just as BRD4 degraders have been a solid proof-of-concept proving ground for the mechanism when they work, they should be illustrative of what can go wrong as well.
They started off by doing an siRNA screen targeting dozens of genes that are believed to function in ubiquitination pathways and looked to see which of these impaired the efficacy of the compounds. This turned up several that affected CRBN or VHL degraders separately, as well as proteins such as RBX1 that had effects on both. And while some of these genes also had effects on cell viability or proliferation when they were knocked down, some of them didn’t, suggesting that they could be relatively expendable for a tumor cell.
The next step was taking MV4-11 cells, an AML tumor line that had already been shown to be about equally sensitive to both varieties of degrader, and expose them long-term to gradually increasing amounts of either compound. They ended up with three cell lines that were resistant to the CRBN-based degrader dBet1, one cell line resistant to another CRBN-targeted compound (dBet6), and one that was resistant to the VHL degrader MZ1. Sequencing these lines showed that the one of the dBet1-resistant lines and the dBet6-resistant one as well had picked up nonsense mutations in exon 4 of CRBN, and showed impaired production of mRNA and/or protein for cereblon itself.
The other two dBet1-resistant ones didn’t have an obvious mutation in their genomes, but showed very impaired protein production of UBE2G1, an E2 ubiquitin-conjugated enzyme that’s part of the relevant machinery. Similarly, the Mz1-resistant cells didn’t have an obvious red flag in the sequencing, but showed impaired levels of CUL2, which is known to be a key part of its active protein complex. Consistent with that, these cells had much higher levels of HIF1-alpha, what with the VHL degradation pathway being bollixed up, and for more on that see Monday’s Nobel Prize post! That’s exactly what you’d expect if you mess up the “continuous degradation” regulatory setup that HIF1-alpha lives under.
So all of these genes had been in the set that they tested up front, which is reassuring, since it means that we may well have a decent handle on the moving parts of these systems (and the components most likely to fail under pressure!) The paper goes on to profile a long list of tumor cell lines looking for those that might be more intrinsically resistant to one degrader or another, and finds evidence of such differences. This is an area of research that we can expect to see more of as we continue to make more degraders and send more of them into the clinic: you not only want to target proteins that are important in some particular disease, but you want to make sure that your targeted protein is optimally degradable by the system you’re choosing. It will also be quite interesting to stratify clinical-trial patients out there in the real world once we get better handles on these issues. Stay tuned!