Here’s one that I certainly wouldn’t have thought about doing. This recent paper in ACS Medicinal Chemistry Letters describes changing the substituents off the core of a PAK1 inhibitor. Nothing too interesting about that, you’d think: the inhibitor itself is one of your standard-looking kinase inhibitor sausage strings of heteroaromatics. And that’s the problem, because those things can tend to have lousy properties when you dose them in vivo. Live by the Suzuki (and Buchwald-Hartwig) couplings, and die by them when you see the pharmacokinetics. So you often see polar stuff hanging off of these cores – it’s not there to help at the kinase binding site, but to drag everything along out of the gut, through the bloodstream, and to help escape the shredding machines in the liver. A classic example is Tarceva (erlotinib), which has those ethylene-glycol-based chains hanging off it mainly to keep it from being such a sun-baked brick after oral dosing. You can see the same sort of thing in Iressa (gefitinib), Caprelsa (vandetanib) and many others.
In this case, the molecule is FRAX1036, shown here. And that piperidine is the polar add-on, but it turned out to be more trouble than it was worth. Some of the medicinal chemists in the audience will be saying “hERG“, and they’re right. Piperidines and piperazines are great structures, and medicinal chemistry would be a lot poorer without them, but they can set off hERG activity, which can (in some cases) lead to heart rhythm problems, which can (in pretty much all cases) lead to your drug wiping out. And maybe taking you with it, depending on just how bad things got. hERG is one of those things that we didn’t used to understand very well, and arguably still don’t, but we understand just enough about it to make trouble for ourselves. Many older compounds made it onto the market with hERG activity in them, but no one is going to feel comfortable developing a compound with strong activity on that ion channel now, for fear of what might happen in the clinic. Even if it were to make it through, no one would feel comfortable sending it off for approval, for fear of what the FDA would make of it, because everyone would be worried about what might happen when the stuff hit the general population. Some hERG stuff can be dealt with via more predictive animal models (dogs wired for telemetry, with constant monitoring of heart rate), but a compound with lotsa hERG may not even get that far.
So the Genentech folks who were working on this structure knew from their previous work on this target that an amine pointing off in a different vector in the binding site (off the lactam nitrogen) could be beneficial. So they merged the two structures (along the way dropping one ring nitrogen, which didn’t seem to be needed). Adding this amine did indeed help the potency, by interacting with some polar side chains in the ribose-binding part of the ATP pocket, and several different things could be used to bring it down there – cyclohexyl, plain old n-butyl, what have you. But hERG activity was high, and cellular permeability was marginal, so something had to be done to make the overall compound less basic, but not so greasy that the PK properties got even worse. Putting that side-chain amine in position via a morpholine analog seemed to help a bit, but not enough. How about another oxygen? That’ll attenuate the basicity even more, while not greasing things up.
The solution is shown – compound G-5555 at right. And that’s a group that I wouldn’t have picked – even the paper describes it as “atypical”. Organic chemists in general may be raising their eyebrows a bit at it, because that two-oxygens-on-one-carbon group (an acetal, in this case) is known to be labile to acid conditions. Y’know, like in the stomach, which is where your drug is going to go when someone eats it. In this case, though, the dioxolane really performed well – in the same way that it reduces the basicity of the amine, the amine makes that acetal more stable. The team tried a number of acid conditions, trying to force the compound to degrade, but it held up. In rodents and monkeys, it showed low clearance and high bioavailability.
I’m going to have to readjust my thinking about these groups. I’ve seen them show up before, but I’ve generally made a face and figured that they were the work of desperate chemists. But any side chain that can turn around a compound this thoroughly deserves a more friendly welcome. Congratulations to whoever thought to put that group in there, because they probably got some of the same reactions that I would have given it, and we doubters would have been completely wrong. (But I still hold on to some of my deep-seated prejudices about other structures – there’s a limit, you know).
Update: corrected the structures, which I had left in a higher oxidation state due to not being fully awake.