Synthetic Route Suggestions

[TomkUCL]

Hi GH community,

We are looking for proposed synthetic routes toward the following target molecules. Due to the similarity of these compounds and in the interest of saving time, a general synthetic route has been planned involving simple amidation and Boc/de-Boc chemistry, however, if better routes can be suggested then please feel free to add them here. We are also looking for synthetic routes toward commercially unavailable / high-cost building blocks (coloured in blue and red). I have given a couple of my suggestions here, however again please feel free to offer your suggestions.

Thanks, Tom

A general route example:

Proposed unavailable building block routes:

Alternatively, a Staudinger reduction might offer more chemoselectivity in the reduction step so that amidation and subsequent nitrile reduction can take place: Reference: WO2014059902A1·2014-04-24 [0336] https://worldwide.espacenet.com/patent/search/family/050487575/publication/WO2014059902A1?q=WO2014059902A1

[TomkUCL]

@ahsgc has kindly offered the following route suggestions for the fluoropyridine acid motif and similar analogues that is ubiquitous in all of the top ten scoring generative Glide hits from @kipUNC. I am currently pursuing the route towards acid 1 below using CuO and CuI catalysed malonation and decarboxylation as per reported patent routes as the commercial cost of the acid starting material is ~$2000 per gram and is required on multi-gram scale for forward chemistry.

[mattodd]

OK, great @TomkUCL, so just to step back a sec (in case people were not at the last monthly meeting (#26):

The molecules in the top pic are some of the most promising that have been predicted to bind our target site on SARS-CoV-2 Nsp13. The predictions were made by @kipUNC using generative methods. Some compounds from this set are being made by Pyramal (ping @ahsgc), but those above involve commercially unavailable building blocks. Specifically, we need routes for these:

...where 1 and 2 are both interesting and useful, and 3 and 4 are tricky and maybe for these we should just ask @kipUNC if he can design them away (given that this is a generative approach).

So if you're a chemist and have a nice route to either 1 or 2, please suggest (ignore the need to control stereochem for now). Either post below, tweet us or post somewhere that you can link to.

[kipUNC]

...where 1 and 2 are both interesting and useful, and 3 and 4 are tricky and maybe for these we should just ask @kipUNC if he can design them away (given that this is a generative approach).

@mattodd I would gladly help if you clarify what is exactly the task. Maybe we can have a meeting?

[mattodd]

All: @kipUNC and I are going to talk this through in person here at UNC. Ultimately, since we're comparing commercial vs generative approaches we can't include new analogs without checking first that those analogs are predicted to work.

@TomkUCL - @ahsgc likes the idea of coupling the N-oxide last, to avoid carrying it through several synthetic steps. It's an interesting point we can interrogate. Doesn't change the building blocks much I don't think. Just the order.

[cedricgr]

Hi all (@TomkUCL, @mattodd ),

We (myself, Rosie Street-Jenkins and Luccas Sanches) discussed some routes for 1 and 2 in our weekly WCAIR Med. Chem. problem sessions here at Dundee. We came up with some possible routes for you.

Route for 1 is based on the patent WO2021050992, using 2-cyanobenzyl bromide as starting material and diethyl acetaimidomalonate in a Claisen-like reaction. We only fear that maybe the Boc group will not stay intact during the amide reduction (LiAlH4 is known to sometimes remove Boc as well)

Route for 2 is based on a Mannich reaction with indanone followed by a reductive amination (some refs. in SciFinder point to use hydroxylamine as nitrogen source instead of ammonia. Edit: amine chains were kinda wrong :)

We hope it helps!

[mattodd]

Thanks, @cedricgr ! I agree some of the selectivity in the reductions in the first synthesis could be tricky. I'm liking the second one, though. Well, except that the amine chains are the wrong way round ;)

[cedricgr]

Thanks, @cedricgr ! I agree some of the selectivity in the reductions in the first synthesis could be tricky. I'm liking the second one, though. Well, except that the amine chains are the wrong way round ;)

Issue corrected!

[TomkUCL]

I am currently working on this building block (~$2000 / 1 g), which is needed on a multigram (~10 g scale) due to its prevalence in the 10 compounds that I am currently pursuing: Compound list:

To me, the simplest route seems to be Cu-catalyzed alpha-arylation of diethyl malonate using 5-fluoro-2-bromopyridine as substrate. I have tried three conditions so far; Pd/enolate chemistry *(no reaction), CuO catalysed (Pfizer patent, 20% yield reported - no reaction), and CuI-catalysed (61% yield reported, some success reproducing this one - see patent link: https://patents.google.com/patent/US20200275661 and ELN link: https://uk-mynotebook.labarchives.com/share/Thomas%2520Knight/MTI0LjgwMDAwMDAwMDAwMDAxfDEwNDQxLzk2L1RyZWVOb2RlLzM1ODI4MzU2Mzl8MzE2Ljc5OTk5OTk5OTk5OTk1 for further details.

I think this latest patent from 2020 sounds the most plausible, and my initial trial reaction on 500 mg scale seems to have been somewhat successful by LCMS (waiting for NMR to confirm), so now I am trying to get to a point where this can be scaled up. Schlenk techniques and dry solvents/reagents are needed due to air/moisture sensitivity of the copper catalyst during this Ullman-type coupling. From the crude LCMS it looks as though some decarboxylation has taken place in situ before the hydrolysis step:

Here is the Biotage Selekt report:

• the first green peak appears to be diethyl malonate reagent (strange this didn't show up in the initial crude LCMS?)
• the second yellow peak is a mixture of the desired product and malonate intermediate (ratio at 254 nm ~ 1 : 3)
• waiting on data to come back on whatever is dragging off the column, but it seems to be highly polar.

[jamesday100]

hi @TomkUCL here's an idea for you, for the seven membered ring target.

[TomkUCL]

Great @jamesday100 - the starting material is a little expensive, but 2-Benzylsuccinic acid is cheap and widely available in the UK;

[jamesday100]

Great, here's another route for you. Starting material is expensive, but available. You might be able to make it more cheaply. NB: this with give the racemic cis isomer by the looks of it.

I think the route to this target using the Mannich reaction is good. Just to note we usually do not advise testing 4 isomers at once, so hopefully you can separate out the diastereomers (cis and trans) and couple them up.