Hi, I'm Corin, CEO and co-founder of Rowan. In this video, I'm going to walk through tautomers: what they are, how to run a tautomer search on Rowan, and how to interpret the results.
First off, what are tautomers? So tautomers are groups of molecules that can readily interconvert under fairly normal conditions. Most commonly, this means compounds which can interconvert between each other by switching protons. These are prototropic tautomers, like ketones and enols. There are other forms of tautomerism, like the ring–chain tautomerism exhibited by some sugars, but we're not going to talk about that more in this video. That's much less relevant for most cases.
One example of a nice tautomeric molecule here is 4-pyrimidone. So you see the molecule as drawn is in one of the N–H tautomers because the proton is sitting on an N. But we can also imagine the proton sitting on this other N and shuffling this double bond around, or even sitting on this O to form sort of the O–H tautomer. And when you're faced with a molecule like this, in this case, we can sort of guess that this is the relevant tautomer because it's the one that's on the Wikipedia article. But it's not very obvious from first principles how to necessarily guess where the protons will prefer to sit. It can be pretty sensitive to the exact steric or electronic effects in a given molecule. So it's nice to be able to run a calculation to try to get an estimate for what will happen.
So let's try it out on this compound since we already sort of know the right answer. We can copy and paste the SMILES string and head over to Rowan, where we can click the “new tautomer search” button from the main page. We can input the SMILES string and we can say “4-pyrimidone”, submit and there's our molecule, just like we saw it on Wikipedia, with the proton sitting on, I suppose this is N3 in the nomenclature. Now we have very little to do to run the tautomer search. The only thing we have to select is the mode, which essentially tells us how we're going to generate the conformers of each tautomer and in most cases careful is appropriate, unless you're really trying to run this on a big library. So we'll say “submit tautomers.”
Let's talk about what's happening behind the scenes while this runs. The Rowan tautomer search code is going to iterate through all potential tautomeric sites: it's going to try to find the places that are reasonable to add or remove protons, and then it's going to run a conformer search and geometry optimization on each of these sites, and then compare the energy of all the different tautomers. In the end, this gives us an output like this. So we see the lowest energy tautomer is number one here, which indeed puts the proton on N3, like we saw on Wikipedia. The second highest energy tautomer is the O–H tautomer, and then the highest energy tautomer is this N1 tautomer. It's labeled as N5 here, but if we follow the proper sort of pyrimidine nomenclature, this should be nitrogen #1. And so we can see from this output that the relative energy is shown here along with the weight. So this lowest energy tautomer has a relative energy of 0 because it's the lowest, the O–H tautomer is like 1.3 kcal/mol above, so that's about a 9 to 1 ratio that's predicted. And then this N1 tautomer is very high in energy, almost 5 kcal/mol above the ground state, and is predicted to have really almost no population whatsoever in solution. So if we were trying to engineer a drug in which we wanted one of these tautomers, it'd be helpful to have these results. If we think there's a key interaction that this N–H is making, for instance, we could feel pretty confident that this is an accessible tautomer, whereas if we expected this N–H to make an interaction in a binding site, we might be pretty disappointed by the results we got back.
So let's try this on another compound just to showcase how different electronic effects can impact tautomeric populations. So we can take this molecule, and we can resubmit it. We'll resubmit as a tautomer search and now let's say we'll do 2-fluoro-4-pyrimidone, so we'll throw a fluorine in there, which we should sort of expect will probably have some sort of impact on the different stabilities of these different tautomers, although we might not be able to guess ab initio what this might have. So actually, if you want to pause the video right now and think about what effect you'd expect this to have, go for it and we can see if your intuition was right or not.
Okay, I'll go ahead and submit this, and it's gonna run right away which is lovely. Yeah and so now we can see in this case we resubmitted from this totomer sort of the N1 tautomer but it doesn't matter because Rowan will search through all the tautomers no matter the input structure, so it's not important to resubmit from any given tautomer. And there we go, and here's the answer. Did you get it right? So now the lowest energy tautomer—by a lot—is predicted to be this O–H tautomer. The tautomer which was previously favored, this N3 tautomer, is now about 4 kcal/mol higher in energy. And the N1 tautomer is now insanely unstable, so almost 8 kcal/mol above the ground state. And it's worth thinking about why this might be.
So this perhaps is a little bit of a surprising result. But I think what we can think about in trying to interpret this is that in this case, OH is not a super electron-withdrawing group. It sort of depends on exactly where you are, but in general we think of O–H or O–R substituents as somewhat electron-donating. But in these other two cases, we have this CO, this carbonyl, which we do think of as quite an electron-withdrawing group. And so my interpretation of this is that when we add this extra very electronegative atom that we're adding on here, we're actually destabilizing other electron-hungry groups on the ring. And so, as a result we destabilize anything that has a C=O double bond and stabilize anything that has a C–O single bond, just because everyone's able to get their preferred electron density a little bit more easily in this case. I'm not sure if that's the right way to think about it: there might be a more intelligent or more mathematical way to think about it, but that's how I interpret these results. Anyhow, thanks for watching!