PostEra

Second design call: focus on covalents and merging with non-covalents

COVID Design
#1

The response to our first call for designs was truly overwhelming - clearly this initiative hit a nerve, and a lot of careful thinking is clearly evident in the designs.

One thing though that was hugely underrepresented was designs of covalent compounds - and yet we see this as one of the major opportunities afforded by the experiment results.

Therefore, for the second call, we ask contributors to focus on designs that build on the covalent hits, especially ways of merging them with with the non-covalent hits, preferably multiple ones at the same time.

If you haven’t already, read Resnick et al 2019 for the full rationale and some examples of what we mean. The general premise is that, if whatever is attached to the covalent warhead is itself pretty potent, then the warhead can be very unreactive (and thus selective and non-toxic),et still form the covalent bond and thus inhibit with high efficiency.

Figure 5E and F in Resnick et al 2019 capture what we hope to achieve; in that paper, a single merging step led to a dramatic improvement in potency.
image

Another variation is to replace the warhead with some other weakly reactive covalent group; feel free to suggest ideas acade we will be (computationally) exploring some replacements anyway, but feel free

(The covalent compounds are clearly marked in Excel file near the top of the download page of Diamond website. )

Thanks again for your enthusiasm…
Frank

About the Design category
#2

Hello Frank and others,

regarding warhead replacements, switching from chloroacetamides to chlorofluoroacetamides was shown in one case to lower reactivity of the warhead. Should impart minimal additional steric demand (H to F). Would be nice to consider this computationally, if possible.

Lit: https://www.ncbi.nlm.nih.gov/pubmed/30643284

Also, can we have a summary list of submissions ordered into covalent/non-covalent? To avoid redundancy, inspire further designs and perhaps manually assess synthetic tractability?

Best
Paul

#3

Hi Frank,
This is a great challenge, and the Fragalysis tool is really user friendly. Just a quick suggestion - is there an easier way in the fraganlysis tool to tell which fragments are covalent and which are non-covalent? This would certainly help in designing covalent + non-covalent merge ideas. Right now it seems the only way is to go via the links for the prefiltered selections on the diamond website, and as soon as one starts removing or adding fragments the info on which is Covalenet and which is non-covalent is lost…

As an example, is it possible to create different binding site IDs for covalent and non covalent? Right now for example site 1 and site 11 have both covalent and non covalent fragments. Could you split these into (e.g.) Site 1nc and 1c; 11c and 11 n and so on? Or perhaps have some kind of signifier in the fragment ID so for example change X0736_0 into C_X0736_0 (since it is covalent)…

Just some thoughts to help make this a little simpler…
Ben

#4

Dear Frank
I do support the way forward keeping efforts in covalent bonding.
Have a look at the link below
https://covid.postera.ai/covid/submissions/64047f27-b27c-468c-87a8-63d24bd5bb43
Have your comment on 640-2 and 640-3

Dr GVK

#5

Hi Ben - yes we are trying - it’s technically awkward because of small design decision innocently made long ago… but maybe we can nail it quickliy. So thanks for the prod.

#6

1-Cyano-1-cyclopropane-carboxamides are also useful warheads

1 Like
#7

Good plan. I guess all Cysteine protease inhibtors that ever went to clinical trials had covalent warhead. That’s why all my earlier designs had a covalent warhead. May I particularly place a link to this design, using some warheads that were successful for CatK inhibitors at erstwhile Organon
https://covid.postera.ai/covid/submissions/b2fae8f6-2e08-4265-94ae-ede5e98fea0c

1 Like
#8

Can I also place a link to work of my former great colleagues in Merck Canada, who developed this nitrile containing warhead. It was for Cathepsin K, also a Cysteine protease, like the Coronavirus protease, and went all the way to Phase III, proving that covalent warheads can be drug like. The drug’s name was odanacatib.
https://pubs.acs.org/doi/abs/10.1021/jm800610j

1 Like
#9

My thoughts around project strategy –

I believe the fastest approach to a potent and selective Mpro inhibitor is to focus, right from the start, on the discovery of covalently acting compounds. From what I heard at the team meeting yesterday, it sounds like a non-covalent fragment strategy is not feasible. And if you only select higher MW proposed targets, you’ll have to be awfully lucky to find something IMHO. The covalent interaction will buy a lot of potency. That will speed things up.

I would forget chloromethyl ketones outright as targets – relatively reactive and potentially unselective. I think the best “warheads” to work on will be nitriles, acrylamides and possibly alpha keto esters.

Nitriles, as covalent warheads have been very successfully applied as DPPIV inhibitors, for example saxagliptin and vildagliptin. Granted, DPP4 is a serine peptidase, but it is well known that nitrile-based compounds can inhibit cysteine proteases quite potently. I note in a recent review the cathepsin K (a cys protease) inhibitor odanacitib has a nitrile warhead. This compound was advanced into Phase by Merck. One issue with nitriles is they will not allow interaction with prime-side pockets.

Acrylamides are now well established as drugs against cysteine kinases – BTK, JAK3, EGFR. Not personally aware of any drugs with this type of warhead vs Cys proteinases, but that doesn’t mean they won’t work. They are potentially reactive and non-selective, but the structural features that can make them more reactive are well known and can be controlled.

I haven’t looked into alpha-keto amides yet. But, like nitriles, they are likely to be slowly reversible, thereby potentially mitigating tox risk. The reactive ketone may make them susceptible to metabolic reduction.

So going forward, I would look for designs (based on the crystal structure) that try to incorporate these warheads. I wouldn’t put ease of synthesis as a first tier filter for selecting designs. Best used to decide between equally compelling ideas.

Anyway that’s my 2 cents.

3 Likes
#10

There are no acrylamide covalent fragment hits so far, to the best of my knowledge. A change from Cl-acetamide fragment hits towards acrylamides will change overall geometry significantly since the electrophilic carbon atom is C1-homologated. This will complicate rational design.

#11

I agree with Ralph - the most practical covalent warheads for incorporation - those that we can advance the fastest to tox. studies - are warheads that are already incorporated in approved drugs.
These will include:
Acrylamides (e.g. in Ibrutinib)
Methylpropiolamides (e.g. in Acalabrutinib)
Nitriles (e.g. in Vildagliptin)

Re. the suggestion above from @JoostU - I think nitrile conjugated to a heterocycle, will likely be too reactive, unless you know on an example to the contrary (that passed PhI).
1-Cyano-1-cyclopropane-carboxamides however (@stentor) could work (as in Odanacatib).

Two additional options, though perhaps for “edgy” are chlorofluoroacetamides, as reported here https://www.nature.com/articles/s41467-018-04343-0 and suggested on Twitter. The advantage in these is that we can install them directly on the chloroacetamides we already have - so we know the geometry would work.

And lastly as Ralph mentioned alphaketoamides might do the trick as the one in the recently reported science paper on Mpro: https://science.sciencemag.org/content/early/2020/03/20/science.abb3405

Looking forward to your brilliant covalent designs!
Nir.

1 Like
Screening cascade - fast track and regular track of designs
#12

There are no structures with acrylamide hits, but if you look at the covalent screen results here: https://www.diamond.ac.uk/dam/jcr:9fdc4297-15b6-47e2-8d53-befb0970bf7c/COVID19-summary-20200324.xlsx

There are a few promising acrylamides - namely:
PCM-0102835
PCM-0102610
PCM-0102682
PCM-0102844
PCM-0102635
PCM-0102730

2 Likes
#13

Nice, should be a good starting point

#14

Nir, should we try again to get them into structures? Diamond is back up today, and we’ll be doing more Mpro experiments.

#15

Dear All,

I would first like to commend this entire effort. It really is remarkable what has already been accomplished and the plans ahead.

In regards to this “Second design call” and from reading through this full thread, here are my thoughts:

Considering the initial hits involved chloroacetamides, geometrically it would make the most sense to switch to fluoracetamides (weakly reactive, smaller than the Cl) however largely different reactivity so potency may take a hit. There are concerns of completely changing the warhead, however there are several options (see image below from a review article we just published).

If the acrylamides are not hits (possibly due to a lack of sufficient reactivity), we can move to a more reactive variant (Vinyl sulfonamides). Or the propiolamide seen in Acalabrutinib.

A second strategy (if we have the space within the pocket), we can utilize a technique typically seen in covalent kinase inhibitors. Whereby you use a poorly reactive warhead (i.e. acrylamide) and append a basic nitrogen group (dimethylamine) to this electrophile. This basic group accelerates Cys reactivity by possibly creating the thiolate nucleophile. It also acts as a good solubility handle (improving PK). See examples such as Afatinib.

Another approach, I have seen the use of 2-fluoropyridines are good warheads which are Cys reactive via a Nucleophilic aromatic substitution mechanism.

So that would be my suggestion, either:

  1. Fluoroacetamides
  2. More reactive variants of acrylamides (acrylamides with dimethylamines appendages)
  3. More reactive variants of acrylamides (vinyl sulfonamides)
  4. More reactive variants of acrylamides (propriolamides)
  5. SnAr-reactive variants (2-fluoropyridines)

I do have covalent docking capabilities as well and would be more than happy to run some calculations if that would be helpful with these proposed structures. I am just having trouble finding the two fragments of interest.

Lit: https://pubs.rsc.org/en/content/articlehtml/2020/cs/c9cs00720b

1
11920Ă—984 192 KB

1 Like
#16

So my magic resident geek fxed it today - it really is better this way, thanks for the kick in the backside. Go Rachael Skyner!

#17

@pgz , let me work on this now!

#18

@pgz I have now updated the list, any of the submissions that mentions a covalent fragment will now have a covalent flag. Note that this is prone to false positives if people submit many molecules (some of which are not based on covalents) as a part of a bigger submission.

1 Like
#19

Hi Frank,
I went to check the updated version and now it seems buggy - I’m getting to the fragalysis site via the links here:
https://www.diamond.ac.uk/covid-19/for-scientists/Main-protease-structure-and-XChem/Interactive-views.html

When it opens up all of the compounds from sites 1, 11 and 8 docked, and the list in the hit navigator. If I then unselect these sites in the cluster selector, and then reselect a site from the hit cluster selector, nothing shows up in the hit navigator (see screenshot below). Also I can’t see how the different fragments have been classified as covalent or not… all I see if that the sites have been renamed with an additonal “C_of_m” after each site

image
image634Ă—761 37 KB

Is it possible that I’m not following the correct link to the fragalysis snapshot?

#20

Hi all

This is a most welcome escape from online teaching and recording lectures and workshops and re-organising exam papers! Thanks!

Wow, a nice set of compounds for testing. Well done on that. Red on the Excel chart, I guess (many of mine included!) means not good? For what reason? Poor docking? Mine were by eye as I think our Glide expired and I don’t have a licence at home and I’m pretty hopeless with docking anyway!

Just a couple of additional points, perhaps if you have some really nice Cys-covalent dockers (caveat, you’ve probably already got 1 million +1 things already to contend with!);

SO2F and similar warheads e.g. Synfacts 2019; 15(12): 1436; *ACS Med. Chem. Lett. 2018, 9, 7, 584-586

oxoboroles (Benkovic/Anacor type) might be useful warheads to envisage:

DMARDs are interesting compounds and might warrant an investigation e.g. Leflunomide, especially its active metabolite: https://en.wikipedia.org/wiki/Teriflunomide?? Probable Michael acceptor, small and repurposable?

Thinking beyond in vitro, in vivo, a number of thiophenes and the like can be bioactivated to form a S-S link (via thiophene oxidation, ring opening). Think of Plavix and omeprazole? I know these won’t show up in a frag screen but, if you get hits, incorporating bioactivated warheads might be useful?

Comments now on the fragments are included.

I’d just be careful with some of the ureas (solubility) and a number of CF3 are very high logP. If these were fragments they’d be very poor leads in terms of LipE.

Anyway, comments attached. Please keep me posted. Well done-you all deserve a medal or some kind of recognition in acting on trying to make the world a safer place!

Stay safe.

Thanks

Best Wishes
John

2 Likes