May 2018 Compounds for Metabolic/Physchem Evaluation

[edwintse]

The following 9 compounds have just been sent to Sue Charman's lab at Monash University for evaluation of their metabolic and physical chemical properties, namely human liver microsome clearance/half-life, kinetic solubility, logD (pH 7.4) and, if time permits, rat hepatocyte.

This data set aims to answer a number of lingering questions: 700 was originally proposed to deplanarise and improve the solubility of the compounds. 784 is an extension of this with the carborane. 565, 784 and 246 are part of the IVIVC set of compounds in which some, but not all of the aforementioned data has been obtained. 336 is the recently synthesised hERG Evador with the primary alcohol 155 as the comparison. 341 is the potent RHS indole compound recently made by @maratsydney 100 is the lead Series 3 compound made by @mbhebhe

Results are posted below.

Monash 2018 Chemdraw.zip

[MFernflower]

See issue #18 for a paper I dug up about possible off target carborane interactions

[edwintse]

UPDATE: The results have come back and are drawn up below. Measurements are the standard human and mouse microsomes. Rat cryopreserved hepatocyte was also obtained for this set.

Observations:

1. There is a range of human and mouse microsomal clearance rates. Good half lives for MMV669784, MMV1579336, MMV670246 and MMV025100
2. Generally good correlation between LogD and cLogP values, with the exception of MMV897700 and MMV1576784.
3. Solubilities are variable. Good levels for MMV1579336. Poor levels for MMV897700 and MMV1576784.

Data will be entered into the Master List and the image posted to the temporary biology page on the wiki.

RESULTS Monash 2018 Chemdraw.zip

[drc007]

I think the cLogP for MMV897700 is incorrect, difficult to see how it can be nearly 3 log units lower than LogD.

[edwintse]

I thought it was a big difference too. That's the value on the report we got (using JChem to calculate). DataWarrior has it as 2.5 and ChemDraw has it as 2.9 though.

[drc007]

Sounds like an error within JChem. You should report it to ChemAxon. Since there are no ionisable groups LogP and LogD should be very similar.

[MFernflower]

Just for comparison this is what http://swissadme.ch spits out for the cubane compound: @edwintse @drc007

[drc007]

Those numbers look more reasonable

[MFernflower]

Perhaps an in-lab switch to swissadme.ch for predicted solubility is needed until MarvinSketch is fixed? @mattodd @edwintse

[mcoster]

I co-supervised a project at USyd with Michael Kassiou, where the student, Hendra Gunosewoyo (now an academic at Curtin University) made cubane substituted compounds, and during that project we noticed that different software came up with significantly varying values for cLogP, one of which matched the mearured logD fairly well. From Gunosewoyo, H., et al. Bioorg. Med. Chem. Lett. 2008, 18, 3720: Perhaps it is less an 'error' and more that some algorithms don't deal with cubanes well?

[MFernflower]

Totally unrelated but since it's late-night where I am going to just jot this down:

@edwintse @mcoster

Those cubyl amide painkillers gave me an interesting molecule idea but they are going to be rather greasy and a rather time consuming synthesis:

SMILES STRINGS:

FC(OC1=CC=C(C=C1)C1=NN=C2N1C(=CN=C2)C(=O)NC21C3C4C5C3C2C5C14)F
FC(OC1=CC=C(C=C1)C1=NN=C2N1C(=CN=C2)C(=O)NCC21C3C4C5C3C2C5C14)F

[david1597]

The cLogP calculator choice is tricky with no ideal solution. We (Ed, me etc) use DataWarrior when we quote values because 1) that's where all our data feeds into and what we use further analysis 2) it's freely available for anyone to use and 3) "it gives more accurate values" (over chemdraw) (however I don't have hard evidence at hand to back that claim up right now). It looks like DataWarrior is quite off for the cubanes however. We've also noticed a very large difference with the carboranes, although that may be the way they are drawn/represented.

[drc007]

None of the LogP calculators are perfect, but in my experience ChemDraw is perhaps more susceptible to differences in the way structures are drawn. The most important thing however is to not to use values from different packages in any analysis. Stick to one package within any analysis and make sure you include details of the package used to generate the values (ideally including version number).

[mattodd]

Hi all, good data, good discussion. Quick points:

1) @edwintse Can you insert link above to raw data on your ELN?

2) Seems like the phenyl isosteres have a negative impact on microsomal clearance and have not made a big impact on solubility. I had thought the impact of deplanarisation would have been bigger. Perhaps we're thinking about this in the wrong way. For the clearance, @edwintse , are these results expected? Do we have other data, e.g. from Craig Williams' review, that things like cubanes don't necessarily solve such problems. If the phenyl is not the site of metabolic change, then why should it affect that, after all?

3) The hERG evador, containing the acid, has very nice parameters. An unusual molecule. I wonder if we could improve its potency.

4) 155 is not amazing, but we need to investigate the OHOH compound, given that benzylic oxidation is most likely clearance mechanism (see low clearance for 246, 784). If this molecule is indeed potent, we must send it for the same measurement set.

5) It's likely that 341, the indole, will need to be similarly fixed re clearance. What are the tricks to achieve that with indoles?

6) Jeez, 100 is a fantastically good compound all round, again supporting more future interest in Series 3 @mbhebhe

I'm hoping to have an online OSM meeting next week so we can discuss this and next steps for the coming 2 months. More in another issue.

Note on software for calculating solubility. We have previously discussed this (OpenSourceMalaria/OSM_To_Do_List#333). Seems like we only really have an issue at the moment with unusual structures like cubanes, rather than more generally. @MFernflower you requested access to edit the wiki for this? It'd help a lot if you could add a note to these discussions in the TechOps wiki - I just added a link to an empty page called "The Platform, Software". We recently had a request on Twitter for a description of how we do things, and it'd be good to point people at a single page that describes what we use. Software for calculations (plus links to any deliberations associated with those choices) would be v useful.