Identification of the compound made in the original bio-transformation experiment

[david1597]

Background

Earlier last year, a bio-transformation experiment created a product which exhibited a potency of <10 nM. At the time this was believed to be the para-phenol transformation. The compound with this structure was synthesised in the lab, and was found to have potency around 3.4 µM. Subsequent comparison of the NMR spectra revealed differences between the two and we now believe that the original structure proposed was incorrect. We need to identify what the original product was. This Issue is to discuss the NMR spectra of the original biosynthesis, compare to other related compounds made in our labs, and propose the structure of that original compound.

All spectra

downloadable as a MestraNova file: Pfizer compound comparison.zip

Observations and deductions of the NMR spectra

1. COSY indicates a para-substitution in the Pfizer compound.
   

2. The chemical shifts of the LHS aromatics in the Pfizer compound are significantly different to the para substituted phenol (and aniline), potentially suggesting an electron withdrawing substituent in this para position.
   

3. There is a peak(s) at around 11.3 in the Pfizer compound, which integrates to 2. This could indicate a carboxylic acid, or potentially two distinct acids. This would explain the observed chemical shift of the pair of aromatics, but I am unsure how this bio transformation would occur.
   

4. There appears to be no benzylic CH in the Pfizer compound.
   

5. The splitting of the CH2 directly attached to the core is simpler, each H is a doublet rather than doublet of doublets as observed in the phenol and aniline This is to be expected if 4 is true.
   

6. The triplet from the -OCHF2 proton is shifted (more shielded) in the Pfizer compound, whereas it occurs at a near identical position for the phenol, aniline and phenyl and hERG evador.
   

7. Is it possible that an intramolecular H-bond interaction is occurring between the proposed benzylic OH and the -OCHF2, causing this shift?

Proposed compound:

Based on the above, we are propose that the benzylic carbon was hydroxylated, and that an event also occurred in the para position of the LHS.

What else is currently happening?

• The original bio-transformation experiment will be repeated. We will be shipping more material this week.
• We will check on any mass spec data available
• Updates will be posted here

[MFernflower]

Was the biosynth done in liver cells? They have a odd way of dehydrating substrates! A good example is how paracetamol is processed:

Do you have any clue what happened to the section marked R on your proposed molecule? Did a quinone form? Was the para hydroxyl sulfated?

@david1597

[MFernflower]

@mattodd Have we ever gotten liver tox screens done in mice or rats? I suspect the electron deficient core might be a bit hard on the liver

[david1597]

@MFernflower Details on biosynth should be in OpenSourceMalaria/OSM_To_Do_List#513; it's various liver microsomes.

For identity of the R group - as of right now, I don't know so I'm leaving it as 'R' and that way I can't be wrong! I don't think it's a quinone however - we wouldn't be able to have the benzylic hydroxylation, and the chemical shifts of the pair of doublets would likely be more 6-7 ppm if it were a quinone. Will need to wait on more info - and expertise, from Scott and team. They're gonna rerun the original experiment as well which is fantastic

[MFernflower]

@david1597 If it's of any help for your NMR work the distance between the O of the OCHF2 and the hydrogen of the added hydroxyl group is predicted by MMFF94 to be 492 pm - I think that's a bit far for shielding to happen????

Do you see any traces of sulfur in the NMR? I suspect the R might have been sulfated

[david1597]

This structure would also be consistent with the 1H NMR region shown above, as the broad singlets at 4.6 and 4.95 may be the two x OH.

[MFernflower]

How hard would this be to make in a lab as a sort of control to verify potency of the dihydroxy motif??? @david1597 @mattodd

[david1597]

Update 10/4/18

@mattodd, @edwintse and myself had a conference call with the team from Pfizer this morning.

• we are confident with the assignment and identity of the phenol made in Sydney
• the identity of the original biosynthesis is still not confirmed. They are working with incredibly small amounts of material so it is not as easy to do with this sample (eg signal to noise in NMR)
• their team received our shipment of compounds today, which includes the original starting point for a repeat biosynthesis, and our synthesised phenol which can be compared through HPLC, and also subjected to the conditions in their assays to look out for any transformations
• mass spec data of their compound indicates the addition of just one oxygen: only +16 is observed
• their simplified ABq at 4.5 ppm in the 1H NMR suggests the loss of the adjacent benzylic CH, however we may have also expected to see a shift in the position of these protons (which we don't) if the neighbouring site now had a hydroxyl attached
• the difference in chemical shift observed in the two pairs of aromatic doublets may be due to salt formation, or
• the phenolic OH may have been substituted with something without a proton, Scott indicated there are rare reports of phosphate esters being generated when phosphate buffers are present, which there are. There is no sign of this in the MS, but it could be fragmenting before detection.
• there are additional non-drug peaks in their NMR (between 0.5-1.5 ppm and also at 11.5 ppm) - they will check the other two compounds biosynthesied at the same time (both of which were inactive) to see if these same impurities are present. If they are this likely rules out these extras as the cause of any activity.
• also a reminder to ourselves - we should ensure we are getting generic cytotox for our active hits
• they're keen to look into this soon, and we may have updates from them within a couple of weeks which is fantastic.

Thanks to Scott, Greg and Raman for the call, and for the interest in working on this intriguing issue!

[mcoster]

Just trying to catch up on latest OSM developments... It's been a busy and productive month or two! Great detective work @david1597 - I particularly appreciated being able to download and look at the NMRs via your MestreNova file, and the presentation of all those stacked NMR traces is fantastic. Complete details of the structure seem so close! I don't have anything significant to put in the mix, but do have a couple of minor comments:

• it would be great to get the raw MS spectrum on GitHub - does the +16 peak observed look like a parent ion? Was it -ve or +ve mode or both?
• I wouldn't be concerned about the lack of shift in the ABq at 4.5 ppm - these chain aliphatic protons should be the most susceptible to changes in shielding by neighbouring aromatic groups, which could be due to conformational and/or aromatic substitution changes, eg. the hERG evador shows a pretty big upfield shift where I would have expected inductive effects to cause a downfield shift.
• I wouldn't be surprised if the shift of the OCHF2 proton is a concentration effect or subtle pH effect. All the other samples would have been run at "synthetic chemists" concentrations and purity level, whereas the biotransformation sample was no doubt run at much lower concentration and with inevitably higher levels of trace contaminants. This often leads to a shift in exchangeable or H-bond active protons. You see this a lot in comparing natural vs synthetic NMR spectra for a natural product - everything lines up nicely except exchangeable, H-bonding and pH sensitive regions of the molecule.
• the ring size for intramolecular H-bonding between the putative tertiary alcohol and the OCHF2 group is really big and therefore unlikely, especially for a weak interaction. I think intermolecular H-bonding is more likely - see previous point.

Once again, great work!

[david1597]

Thanks @mcoster - comments, whether minor or major are always welcomed and appreciated!

Small update:

1. they really did get on this quick and the metabolite has been remade. Scott says this is +16 on the MS (I've not got a copy of the raw MS yet, answering your question above) which from fragments has occurred in the NW region. It does not coelute with the phenol compound I made - so definitely something different. NMR is the next step.
2. I'm going to make a test compound where the benzylic position is hydroxylated, The OHOH ( Synthesis of "The OHOH" - a benzylic hydroxylation product #36).

[mcoster]

@david1597 thanks for the update - quick results are a beautiful aspect of having industry collaborators! Just another thought occurred to me... If the new compound is a phosphate ester or some other highly ionizable species, I would expect the hplc properties to be drastically different to the other compounds in the series. Certainly, I would expect it to move more than just an extra hydroxyl or two. Do we have details of retention times for this compound vs starting materials and other congeners?

[mattodd]

True @mcoster . Email came in overnight from the Pfizer team (@david1597 will I'm sure update when he's a moment away from busy lab work at the moment) suggesting that the compound has been re-made in the biosynthesis experiment, and extra data acquired, that supports a product with a benzylic oxidation (as per the OHOH). Awaiting the raw data to confirm, which we'll post as soon as we get it. (The Pfizer team are not (yet) on GH, so we have to relay like this at present)

[MFernflower]

ICP-OES would tell us the if phosphorylation happened but that involves destruction of the sample and I do not know how much would be needed for ICP testing @david1597 @mattodd

[david1597]

OK, a little more info - as Mat said above, the Pfizer team got back within 24 h with a further update. (We jumped the NMR queue on this one!)

• the same product was made as in the first biosynthesis but a better sample allowed 1D, COSY, HSQC and HMBC to be obtained
• this confirmed that it is the benzylic OH, not the p-OH product. There is no para substitution.
• this also confirms the data from mass spec, in that only one +16 is observed
• they will be sending this compound to Dundee

The actual data should follow next week, this is right up to the minute news - but worth sharing now - and some time is needed to write it up first.

The structure is shown below, it has been assigned MMV and OSM numbers. And, as it happens, I set about making this exact molecule earlier this week - details on that synthesis at #36.

FC(F)OC(C=C1)=CC=C1C2=NN=C3C=NC=C(OCC(CO)(O)C4=CC=CC=C4)N32
InChI=1S/C21H18F2N4O4/c22-20(23)31-16-8-6-14(7-9-16)19-26-25-17-10-24-11-18(27(17)19)30-13-21(29,12-28)15-4-2-1-3-5-15/h1-11,20,28-29H,12-13H2
LVBNVRWXODMMAV-UHFFFAOYSA-N
MMV1580315
OSM-S-541
C(F)OC(C=C1)=CC=C1C2=NN=C3C=NC=C(OCC(CO)(O)C4=CC=CC=C4)N32
InChI=1S/C21H18F2N4O4/c22-20(23)31-16-8-6-14(7-9-16)19-26-25-17-10-24-11-18(27(17)19)30-13-21(29,12-28)15-4-2-1-3-5-15/h1-11,20,28-29H,12-13H2
LVBNVRWXODMMAV-UHFFFAOYSA-N
MMV1580315
OSM-S-541

[drc007]

Is this a single enantiomer? Perhaps try a chiral shift reagent?

[MFernflower]

@drc007 compound currently under resynthesis is going to be racemic

[drc007]

But what about compound from biotransformation?

[MFernflower]

@mattodd do we know if the stuff made in vivo was racemic or single isomer?

[drc007]

@mattodd @cdsouthan When we have unambiguously confirmed the structure what is the mechanism for correcting all the web pages, tweets, identifiers, PubChem?

[cdsouthan]

@mattodd @drc007 good question i.e. how do we keep our open records straight when these are revised/corrected?

1. Since it was not in the latest sub-100nM list from David I am submitting to PubChem https://mail.google.com/mail/u/0/#inbox/162b3c38e741f4d9 and LVBNVRWXODMMAV is Google -ve so far, we can hope no erroneous results have "got out" to globally propagate

2. We need to maintain the primacy of the master-list (in the absence of proper registration system "sigh") and purge structure<>activity relationships from that sheet that are experimentally confirmed as erroneous/misleading. Along the same lines I would ague that replicated re-tests could also displace older discordant values but this would depend on assay technicalities

3. For the internal stuff I don't have any immediate ideas as how to back-purge data and relationships that are found retrospectively to be erroneous (e.g. in ELNs). I suggest the practitioners on the ground may have some suggestions here, but I'm thinking adding explanatory notes and forward pointers to the corrections might be better than erasing the original records

4. For the soft info whizzing around here and in tweets I don't think anyone would take us to task if some of it turned out to be wrong. It emphasises the fact that, at least the way we practice open science is, of necessity, somewhat rough-and-tumble

5. (crossing over with comment below) When this interesting story really is locked-down (i.e. with re-syns, re-tests and all that) I suggest David and Scott et al should then write a stand-alone paper (ACS Omega? JMedChem?) with all the gory details. That would ensure the permanent public record is as straight as we can make it

[drc007]

Sounds good, if we are working in the "open" we have an extra responsibility to ensure we don't contaminate the literature.

[david1597]

@drc007 @MFernflower Scott indicated to us that the biotransformation would be a single enantiomer.

[david1597]

@drc007 @cdsouthan Excellent points regarding how we correct everything.

1. The revised benzylic-OH structure has been assigned a new MMV/OSM number. I propose to contact the MMV data specialist for advice, and getting the initial testing results from the first biosynthesis transferred to this new identification. The current sample, currently en route, has already been assigned these new numbers, so no problem there. There is currently nothing incorrect in the Master List.

2. My synthesis of the phenol was assigned the original MMV number, that is correctly linked to that structure. As the structure matches the sample, that can be left as it is, with the biology results already correctly linked to it.

3. For my current racemic synthesis of the benzylic-OH compound I'll assign a fresh MMV number, as it will be of a different composition to the likely enantiopure biotransformation sample.

4. Agreed with @cdsouthan point 3, we shouldn't erase anything. There's not too much, if anything, in the ELN's anyway, as they generally just use our internal synthesis codes.

5. There will be some inconsistencies here on GitHub. A note and/or forward pointer in places we find necessary would be the best way I think. For example, any issues where we have discussed. Or on the wiki where we have the potency results - currently, it's showing the incorrect structure alongside that first Pfizer potency result.

[cdsouthan]

1. Comments like that @david1597 could get you a beer, especially if @drc007 buys both of us one, and maybe @mattodd buys all three of us at least one....

2. AYK, if you generate a new MMV numbers make sure MMV then synch these back into their system with the identical structures so we can ensure OSM:MMV:CIDs are 1:1:1

3. AY also K we can submit racemates and enatiomers to PubChem (if you exactly specify the R/S centre in the SMILES). Note also, since the issue has come up before - no we cant specify enantiomeric purity in the CID but we can as a comment line in the SID - but using these as batch numbers may be a step too far just now

[cdsouthan]

Innit just a small world. Article on the Obach team metabolic transformation work popped into my Twitter feed

https://cen.acs.org/pharmaceuticals/drug-discovery/Mixtures-liver-enzymes-improve-drug/96/i17