Topic automatically created for discussing the designs at:
https://covid.postera.ai/covid/submissions/EDJ-MED-3c65e9ce
Hi @edgriffen the isoquinoline analog ALP-POS-6747fa38-1 of the first design of the submission (aka BEN-DND-93268d01-8) was about 70 μM. Crystal structures are available for both inhibitors as x11417 and x11541 and there are significant differences in the two binding modes. I would not expect either piperazine N to be protonated under assay conditions.
Indeed - in fact this is the problem. The pyridine is potent in one assay and has brilliant solubility and low human microsomal clearance, but the isoquinoline is dead. The crystal structure of the pyridine BEN-DND-93268d01-8) shows a movement in the methionine at the bottom of the P2 pocket. Currently inactive in antiviral assays, but that could be because of the very low log D (clogP ~ -0.5) so make a very limited number of closer analogues where the logD is higher and it’s substituted pyridines where the SAR should give potency. If we could get potency in this logP space it would be a very attractive start point.
The rather striking discrepancy between the two enzyme inhibition assay results suggests that it might have been an idea to repeat both assays before expending the slots in the antiviral, sol and microsomal clearance assays (as far as I can tell the IC50 values were not measured in replicate). The basic N of the piperazine is not likely to be significantly protonated at neutral pH (I recall a figure of 5 - 5.5 in discussions with @joe) and I’d want to be confident that the two crystal structures had been determined at the same pH. The quinoline conformation is closer what is seen for in the CSD for neutral substructure models such as HODQUP and UWUJAZ (basic nitrogen eclipses amide NH). In contrast, the bound conformation of the pyridine analog has the basic nitrogen eclipsing the amide C=O