EGFR is a growth-factor receptor protein that’s well known as a cancer target, and there are a number of drugs that target its kinase activity in order to shut it down. But as is also well known, many cancer cells are rather genomically unstable, and throw off mutations constantly. One of the most common problems with EGFR kinase inhibitor therapy is the development of the T790M mutation, where a methionine gets substituted for threonine in the kinase domain’s ATP binding pocket. That’s right where the inhibitors bind, and it makes the resulting EGFR protein resistant to their action. Up to 50% of the resistant tumors that develop have this change – it’s unfortunately quite effective, which is why we see it so much.
That’s led to efforts to target the mutated receptor, of course, and in late 2015 the FDA approved AstraZeneca’s Tagrisso (osimertinib) for just this indication. It’s one of the newer wave of covalent inhibitors – there’s a Michael addition acceptor hanging off one end of the molecule, which reacts (more or less irreversibly) with a cysteine residue on the protein. Ibrutinib (targeting another protein, BTK, in Burkitt’s lymphoma) was one of the first compounds of this type, and it’s done very well indeed , as have neratinib and afatinib for their respective targets. (Update: caterinib was an early entry that fell out of the clinic, and dacomtinib is still in trials). Mechanistically, this idea has been very well validated; it’s a clear success story.
But as usual, there’s more to that story. That’s the take-away from this paper from Cravatt lab at Scripps, in collaboration with a team at Pfizer. They’re using their expertise in covalent profiling across entire proteomes (and they’re some of the best in the world at this) to look at several covalent EGFR compounds – osimertinib, Clovis’ recently abandoned compound rociletinib, and PF-06747775, which is still in clinical trials. The group prepared versions of these with acetylene reporter groups (in order to find them after they’d labeled their protein targets, which is standard chemical genomics practice and yet another reason why the whole bio-orthogonal chemistry field is a plausible candidate for a Nobel). And things aren’t as simple as you might have thought:
Our chemical proteomic studies reveal that, despite the highly engineered EGFR mutant inhibition profile achieved by all three third-generation inhibitors and their shared unsubstituted acrylamide reactive group, the inhibitors exhibited strikingly distinct proteome-wide reactivity profiles in human cancer cells. More in-depth characterization of the specific off-targets for each third-generation inhibitor revealed that inhibitor 1 (osimertinib – DBL) reacts at high stoichiometry with multiple cathepsins in cell and animal models due to lysosomal accumulation of the drug. That these off-target interactions for 1 were not observed in vitro underscores the importance of performing chemical proteomic studies on drug action directly in living systems.
That is very sound advice. Many studies by now have shown that covalent labeling is a very context-dependent thing. Proteins have different surfaces exposed (and hidden) in living cells as opposed to cell lysates, for example. And this effect is another big one – compartmentalization inside the cell. The whole point of cellular architecture is to make various regions very different from each other and to keep them separated, with defined roles and interactions, and drug molecules can find themselves barred from some of them and accumulating in others. At the whole organism level, such effects are known even to many people outside the biomedical field (the idea that some drugs get into the brain and some don’t, for example), but the same principles apply within cells as well.
These findings build on those from previous Cravatt-group papers and others, and suggest that drug targeting covalent cysteine interactions really need to be profiled carefully across the entire proteome. Cysteine is a unique and important amino acid residue, one that shows up in many binding sites and in the active sites of enzymes. Its free SH group can vary widely in reactivity depending on its environment in a given protein, from not-so-reactive to “red hot”. Which of these you react with when you give a covalent-acting compound will depend on the particular drug structure that your covalent warhead is attached to – it’s quite reasonable to assume that every one of them will have a different fingerprint, to some degree.
In this case, some of these off-targets definitely merit a closer look. One of them, CHEK2, has been variously reported as a tumor suppressor or tumor promoter, depending on the cell line, so you’d really want to see if your drug is hitting it under clinical conditions and what that might be doing. The cathepsins (mentioned in that extract above) are an important class of enzyme themselves, and hitting one of these in particular (cathepsin C) could have direct consequences for the immune system.
Importantly, if you just screen osimertinib against cathepsin C in vitro, it doesn’t look like there would be a problem at all. It’s not that reactive under normal assay conditions, but in the cell, when the compound piles up in the lysosomes, things are different. Lysosomal accumulation is a known behavior for many compounds, and can be a bug or a feature depending on the situation. Tyrosine kinase inhibitors in particular are known to accumulate in this way, an effect that is quite possibly of clinical relevance, so covalent ones need to be checked in real living cells to make sure that they’re not going to do more than you want. Covalent drug developers should take note.