Late Stage Functionalisation of Series 4 Compounds Through Biosynthesis

[mattodd]

Another overdue post (that I thought I’d already made public), on some lovely new data related to the use of Aldehyde Oxidase (AO) as a synthetic tool for late stage functionalisation. See #485 for background and #512 for parallel work on the AO lability assay for these compounds.

The summary of email-based correspondence between me and Scott Obach (Pfizer) to date is as follows:

In March: Scott carried out a test of systems for biosynthesizing metabolites of the four compounds sent (see below for structures). He tried ten different batches of liver microsomes from various species and four cytosols. He obtained MS data that suggests region, but not exact site of metabolism. Scott’s initial observations on these compounds:

MMV897709
Scott: saw two metabolites in liver microsomes – O-demethyl and the corresponding ketone. No reaction in cytosol. Are the alcohol and ketone of any interest?
Me: So no AO conversion. The alcohol and ketone are things we're making from scratch, so I guess not too interesting as an output of your work, no.
Scott: OK I won’t bother with this compound any more.

MMV670246
Scott: generated a major +16 (OH) on the azabicyclic—interestingly both liver microsomes and cytosol generated the same product, which is possible but unusual. This should be very makeable. Also saw a “metabolite” in monkey cytosol that had the same MS as the parent—does the amide bond exist as the enol tautomer and would these exist in separable rotational isomers? Not important for biosynthesis; just something odd I noticed and only in monkey cytosol.
Me: This was the compound with the electron-deficient core that we thought might be vulnerable to ring functionalisation, so it's good that you might be seeing something attach there. It'd be nice to see where on the ring, of course, and whether the compound is active. Tautomers - I wouldn't expect that, no. Interesting and odd. When you say it has the same MS, do you mean the same molecular weight, or is the implication that the molecule is not a constitutional isomer?
Scott: this new peak with the same MS1 and same MS2 fragments (in different abundances) was minor and only in monkey liver microsomes—just a point of curiosity. Certainly the main peak is the O on the azabicyclic.

MMV670944
Scott: generated two +16 metabolites in good conversion, especially in dexamethasone-induced rat liver microsomes. One is on the azabicyclic; the other is probably also there but MS data could not rule out the difluoromethoxyphenyl. One metabolite was also observed in cytosol.
Me: also an amide, so we'd again expect azaaromatic functionalisation. In this compound there is an additional pendant pyridine, so I'd assumed the O was going there? Other side of the N from the CF3?
Scott: just one +16 was in cytosol and the MS data suggests on the azabicylic and not the pyridine. The other was generated in liver mics either on the azabicylic or the phenoxy. But of course we’ll know for sure after doing biosynthesis and getting the NMR data.

MMV693155
Scott: generated major +16 metabolite on the benzyl (either on the ring or the benzyl position). Also saw the O-dealkylation which is probably not of any interest. This all happened in liver microsomes from several species. No conversion seen in cytosol.
Me: So no AO conversion. The O-dealkylation in microsomes is something we've seen before. The two current strategies to combat this are blocking or, more promisingly perhaps, improving solubility.
Scott: while there was some O-dealkylation, I am confident I could make the product where the OH is on the phenyl (or benzyl) position. Would there be interest in probing he SAR on that region?
Me: Yes, interested in the hydroxy product here. This is a potent compound with a primary OH. Very interested in any chance of derivatives with potentially further increased solubility, and the SAR around the benzyl position is of top priority for us synthetically at the moment, in part because it's such a strong determinant of potency.

Following this work, in May, Scott updated me with this exciting news (paraphrased):

“Using liver microsomes via P450s I’ve made an N-oxide analogue of MMV670944 (at 4.8 mM) and a phenol of MMV693155 (at 0.85 mM). Using cytosol (presumably AO) I’ve made a carbonyl metabolite of MMV670246.”

Structures:

These are very exciting results. We’ll seek approval ASAP to get these three molecules evaluated by Dundee, which means registering them in ScienceCloud to get MMV numbers. @edwintse could you please take care of this given that you’re doing this right now for the next batch of compounds to be sent anyway? Please post the MMV numbers here when done so Scott can use them in labeling his vials, and then ship.

AO Late Stage Functionalisation Results.cdxml.zip

[edwintse]

The 3 new compounds have been added to the master list and MMV numbers have now been assigned as below:

[mattodd]

Irene just sent the potency data in. Very interesting. We have a potent metabolite, at 9 nM. We'll need to get more of this compound, to confirm.

(Just as an experiment, I'm sharing the data from Irene here on Labarchives).

Question, then: Are there other compounds, related to MMV693155, that we should subject to a similar biofunctionalisation approach? I believe Scott may be willing to look at those.

[drc007]

@mattodd It would be interesting to find out if MMV1557865 is now chiral. It would not be totally unexpected if the enzymes treated the enantiomers differently.

[mattodd]

Jeez @drc007 that's a great point. That's going to be easier to address with a scale-up of the bio route following chem re-synthesis of the rac.

[spadavec]

@mattodd FWIW, my model predicts OSM-S-412 to have a potency of ~90 nM. It also shows that the meta-hydroxy is roughly equipotent, and the para substituted alcohol is likely ~250 nM. Interestingly, the 2-fluro-3-hydroxy is predicted to be ~2x more potent than OSM-S-412, but that may very well be an artifact of the neural net.

[mcoster]

Can anyone link me to relevant discussion of the target properties for series 4 compounds? What needs improvement most? Potency? Water solubility? Half-life? Off-target effects? Ie. is OSM-S-412 a good candidate to base future structures on, or do the phenol and primary alcohol groups have negative DMPK consequences?

[mcoster]

@spadavec Does your model make any predictions about potency of the R vs the S enantiomer?

[mattodd]

Hey @mcoster start with #388 #390 ? Solubility is key, and is thought to be a way to improve clearance. Potency then needs to follow. Ideally. Primary alcohol should be OK. Phenol? Not sure. Can anyone comment on that?

[spadavec]

@mcoster my model doesn't take into account chirality in this case; I could probably implement that fairly quickly though...

[mcoster]

Thanks @mattodd & @spadavec - speaking of chirality, which I love but does add complexity, have there been oxetane derivatives proposed like this before: (I couldn't find them by searching in InChIKeys)

Super expensive to buy the building blocks, but not that hard to make. More of a ketone isostere than a primary alcohol. Gets rid of the chiral centre, which sometimes makes life easier...

"phenyloxetane"
QFPOSUUWHVGDFZ-UHFFFAOYSA-N
FC(F)OC1=CC=C(C=C1)C1=NN=C2C=NC=C(OCC3(COC3)C3=CC=CC=C3)N12

"4-hydroxyphenyloxetane"
WKTGMMQLGDTFEJ-UHFFFAOYSA-N
OC1=CC=C(C=C1)C1(COC2=CN=CC3=NN=C(N23)C2=CC=C(OC(F)F)C=C2)COC1

[MFernflower]

I wonder how a prodrug approach would work? @mattodd @mcoster

[mattodd]

Irene Hallyburton has confirmed the activity of the phenol - actually slightly improved potency to 3 nM, which is promising. Data, sent by email on Aug 2nd:

Irene Hallyburton Obach Evaluation n2.pdf

Also posted here.

Confirms the importance of resynthesis in #525 . @david1597 can you please start a new issue on which other existing OSM compounds we should send to Scott for similar biofunctionalisation?

By email, Scott says:
"Whether or not a phenol itself offers everything you want without other things you don’t want (e.g. increased clearance via conjugation pathways, decreased membrane permeability), it opens the door to tolerance to a bit of PSA and/or H bond donor characteristic in that region so that you can consider some bioisosteres too in your chemical syntheses."

If @david1597 's synthesis works, we should consider other groups here. @MFernflower has alread mentioned the prodrug idea. What of phenol bioisosteres?

"The comment someone ( @drc007 ) made on the stereochemistry is a good insight. While the P450 enzymes are rarely stereospecific, there may be some stereoselectivity such that this could be a mixture where one enantiomer is more potent than 9 nM and the other less so. When you try chemical synthesis on scale and chiral separation, that answer can be had ."

Quick note: @mcoster re oxetanes. Yes, tried, as shown here, MMV670438. That's an inherited data point for which we have no primary evidence (of e.g. the molecule's identity or purity. As it stands, it's a group that's tolerated, but gives lower potency than the related alcohol. OK metabolic parameters though, as shown further down the wiki page.

[MFernflower]

I have a totally insane idea but it might not be that looney considering how potent alcohols are!! - isosorbide ether anyone?

Another (simpler idea) is to produce a compound like this: @edwintse do you have this alcohol in stock?

[drc007]

@mattodd @mcoster The solubility of MMV693155 is 12-50 ug/mL depending on assay which is not impossible to work with. Do we know if solubility compromises oral absorption? I'd expect the phenol to have better solubility, but it may increase clearance. Classic bioisostere for phenol would be chlorophenyl, aniline, amide, sulphonamide and cyclic versions as shown here http://www.cambridgemedchemconsulting.com/resources/bioisoteres/aromatic_bioisosteres.html.
Solubility can be improved by converting phenyl rings to pyridazine .
@MFernflower prodrugs are interesting, but what issue are you trying to address? If solubility then you need to design a prodrug that contains a ionisable group preferable a basic amine.

[MFernflower]

@drc007 Dr.Swain, What would be an example synthetic route to say 6‐(difluoromethoxy)pyridazine‐3‐carbaldehyde ? - I really like the idea but it seems like it might be a bit of a fuss to actually make real

[drc007]

@MFernflower 3,6-Dichloropyridazine is known, sequential displacement by suitable nucleophiles might be an option?

[david1597]

With the recent activity regarding the identity of the third compound in this list, I've just noticed that the potency data was never added to the master chemical list (MCL).

Pfizer I potencies.zip

The first two compounds for which we know the structure have just had their potencies added to the MCL. The third, for now, remains unknown and this issue remains open until we know it. Whatever it is, at 5.8 nM, it is the most potent compound tested from series 4 to-date so it'd sure be good if we could figure that out (currently under discussion at OpenSourceMalaria/Series4#33). (We've also often quoted this value as 9 nM - this was the result of the first run only. The second repeat was 2.5 nM, hence 5.8 nM average)