SAR analysis for 3-aminopyridine-like inhibitors

One question of potential interest to COVID Moonshot designers is whether the potency benefits of isoquinoline at P1 relative to pyridine are maintained when the P1 heterocycle is linked by CH2 (as opposed to NH). In terms of potency, an isoquinoline at P1 needs to ‘pay its way’ since it’s bicyclic (naphthalene is less aromatic than benzene and therefore more reactive) as well as being bulkier and more lipophilic than pyridine. Here is some SAR analysis which suggests that isoquinoline at P1 may be less beneficial (relative to pyridine) when linked to carbonyl by CH2. Let me know if anything is not clear and/or if you spot any errors. This analysis has implications for lipophilicity management in what I’ll call the ‘benzotriazole series’ (isoquinoline has been substituted for benzotriazole in this series) and I’ll mention @mc-robinson @edgriffen @alphalee.

The starting point for the analysis is to note that ‘reversing’ the acetamide linker has a minimal effect on potency (f: fluorescence; RF: RapidFire) for methylpyridine at P1.


While ADA-UCB-6c2cb422-1 has been assayed, the other compound required for making the comparison for quinoline at P1 has not yet been synthesized although it was submitted as PET-UNK-8922bd3c-1 (and has been ordered). However, the ‘reversed acetamide’ DAR-DIA-23aa0b97-19 has been assayed and I’ll mention @Daren_Fearon who designed it. The isoquinolinyl amide with 3-cyanobenzyl at P2 has not been synthesized although its potency can be estimated as shown below.

The estimate for the effect of ‘reversing’ the acetamide linker can be derived as shown below. While ‘reversing’ the acetamide linker of TRY-UNI-714a760b-6 results in a small decreases in potency, applying the same structural transformation to ADA-UCB-6c2cb422-1 is predicted to result in a 1.2 log unit decrease in potency in the the fluorescence assay. This suggests that a significant proportion of the potency benefit (relative to pyridine) of the P1-isoquinoline will be lost when the acetamide linker is ‘reversed’.


what about a urea? These tend to be problematic and insoluble for a start…

It’d be interesting to see where the ureas fit in if the relevant data are available and I’ll take a look. On average, N-methylation of acyclic secondary amides leads to increased aqueous solubility as discussed in this article. I’d guess that RN(H)C(=O)N(H)R will pack better than RN(H)C(=O)R although I’m not aware that anybody has done matched molecular pair analysis to explore this possibility. My understanding is that tertiary ureas (e.g. tetramethylurea) are non-planar.

@pwkenny I believe this loss of potency is due to the hybrid sp2 nature of the amide bond, which make the aromatic bound to it more restricted in terms of the number of conformations it can adopt to interact with the active site (in this case with His163). And when you change NH for CH2, by losing this feature, you increase the conformationally flexibility of the isoquinoline and that might affect the number of possible conformers to proper interact with His163. You can argue that when you “reverse” it, you are just changing the hybrid sp2 nature from one place to another and that is true. But we have found that the interaction with His163 demands a higher degree of spatial orientation of the HBA, whereas the other side of the molecule interacts in the His41 pocket that is more “flexible”.

This is true, when you alkylate amides or similar groups, their water solubility increases due to less packing (the melting points reflect that perfectly, you will notice that every amide has a higher m.p than its alkyl counterpart). Another problem that you also need to be aware of is the dramatic conformational change that an amide alkylation produces. We have seen this very clearly with N-acylhydrazones.