Closed edwintse closed 1 year ago
Hi all, Here are some initial ideas for a synthetic strategy towards the first target molecule (TM-001), with a nod to this years Nobel prize with an organocatalysed oxidation in step 4! Of course if anyone can see any potential problems or a shorter route, feel free to leave feedback.
6 step synthesis of TM-001.pdf
The synthesis can be broken down into two main parts, 1) alcohol synthesis and 2) amide coupling:
Amide coupling:
Based on the reaction with 4-chloroaniline (Reference here: https://onlinelibrary.wiley.com/doi/10.1002/jhet.5570390616) , we propose in-situ acyl chloride generation of the thiophene carboxylic acid (step 5), followed by amide coupling with 3,4-dichloroaniline (step 6). There was no reported reaction with 3,4-dichloroaniline, but for 4-chloroaniline reacting with 2-thiophenecarboxylic acid, a 66% yield was reported. Based on an input of 100mg of acid, we’d hope to get around 167mg of ‘alcohol’ amide based on this similar aniline.
Alcohol synthesis:
Using the commercially available 2-(4-bromothien-2-yl)-1,3-dioxolane (https://store.apolloscientific.co.uk/product/2-4-bromothien-2-yl-13-dioxolane), we could either do a lithium-bromide exchange followed by lithiation of ethylene oxide OR a standard Mg Grignard. I prefer Grignard as a paper in the Chinese Journal of Pharmaceuticals (https://ucl-new-primo.hosted.exlibrisgroup.com/primo-explore/fulldisplay?docid=RS_610018255331lopidogrelhydrogensulfate&context=SP&vid=UCL_VU2&lang=en_US , ISSN: 1001-8255 , 1001-8255; CODEN: ZYGZEA) reported 90% for their reaction using 2-bromothiophene (CAS Reaction Number: 31-063-CAS-8489980). In either case it would be best to start from the commercial acetal-protected thiophene (CAS 58267-85-9) to avoid any side chemistry between the Grignard or nBuli with the carboxyl group (acidic proton destroying the Grignard or attack on the carbonyl), then deprotect the acetal and amide couple at a later stage.
Keen to hear your thoughts if you have anything to add!
Great @TomkUCL - you should be able to drag pictures into issues so that they appear. And if you're referring to journals, you can link directly to them so that people can click through. As we talked about, do similarity searches on key combined frags to see if people have made them before, and fuzzy searches on Scifinder to see if people have made similar bits.
Also a general design criterion - we don't necessarily need these exact compounds, since we're still in exploratory phase. If suggestions are tough, we can re-think! For example we can replace the cyclobutane in one of our suggestions with something else. If 3,4-dichloro is no good, we can replace with other halogens, maybe. That kind of thing.
We're also aiming to make stuff that is not commercially available. So keep an eye on Manifold, MCule, etc etc to ensure that we're not making stuff we can easily buy.
(I guess we need a "how-to" list of things like this in the wiki, right? Want to add a few notes in the wiki to take care of this workflow?)
What does Scifinder tell us about the most similar molecules to the pharmacophore that have been made (but are not commercially available). We should be checking this too! @edwintse @TomkUCL
We are currently planning routes based around the site 16 pharmacophore designed by Hebba (see below). Notably, we want to focus on the specific coloured interactions shown in the structure.
.
The fragments shown below in red have now been made through a 2-step amide coupling / alkyl lithiation route. NOTE; this is a much shorter route than the initial brainstorm, much to the relief of any process chemists out there.
Amide coupling: Yields for 3,4-dichloro aniline amide coupling using sulfonyl chloride were slightly better (~30%, 101 mg) than with direct coupling using EDAC (21%, 53 mg), however still disappointingly low. This low yield, along with significant amounts of the carboxylic acid / EDAC intermediate indicated that the poor nucleophilicity of the dihalogenated aniline can be problematic. However, adopting an amide coupling method designed specifically for electron-poor anilines would likely result in a higher conversion to amide (as indicated by Jemima's increased yield from 39% to 64% in the amide coupling step for the 3-CF3 analogue using this HOBt / DMAP / DIPEA coupling approach, see ref: https://doi.org/10.1016/j.tetlet.2020.152719)
Going forward, we want to extend the alkyl chain by one CH2 group and try using the pyrrole (rather than thiophene) analogue to exploit the pocket and N503 interactions, respectively, shown in the pharmacophore model (see example fragment below). This also gives us options to easily add ether groups to the free alcohol in order to explore the new pocket, as well as substitute in different amines/ anilines to evaluate the aromatic interactions shown in Heba’s pharmacophore
To do this, we propose an amidation (using the HOBt/DMAP approach for electron-poor aniline amide couples), see ref: https://doi.org/10.1016/j.tetlet.2020.152719 This will be followed by N-Boc protection on the pyrrole, followed by the previously successful Li/Br exchange and subsequent alkyl lithiation with cyclooxabutane, which has reported 69% for the simpler derivative (see below). Note lithiation with the amide was previously achieved successfully. Subsequent N-Boc deprotection should give the desired product shown above.
The starting material is cheap (£82 / 5g) from Apollo UK, and having troubleshot some of the previous issues (poor aniline nucleophilicity, lack of Grignard reactivity, purification methods), we believe this molecule and its aryl (right hand side) and ether (left hand side) analogues can be made quickly.
Here's a different angle on the HO-alkylation reaction for the pyrrole that is reported, since the Li/Br exchange - alkyl lithiation route hasn't been reported as far as I can tell. Maybe we can try this route if our alkyl lithiation doesn't work on oxetane. https://worldwide.espacenet.com/patent/search/family/043857159/publication/WO2011044394A1?q=WO2011044394A1
UPDATE 26/11/2021:
These conditions look to give full conversion to the desired amide intermediate, without the need to purify by column chromatography in between - product is currently on high-vac and awaiting H-NMR to confirm this. If clean, I will continue with the alkyl lithiation route successfully used for the bromothiophene/epoxide reaction (see below).
I am currently synthesising this pyrrole analogue of the previous thiophene molecule:
The pyrrole bromo-amide needed for the follow on C-alkylation with oxetane has been synthesised in 76% yield as a pale-brown solid, confirmed by 1H-, 13C NMR and LCMS. Consumption of the starting acid was complete, so the loss of yield is assumed to be during the workup (multiple washes). LabArchive link here: https://uk-mynotebook.labarchives.com/share/Thomas%2520Knight/MTguMnwxMDQ0MS8xNC9UcmVlTm9kZS8xNDQ5NTA0NDQ5fDQ2LjE5OTk5OTk5OTk5OTk5Ng==
I am currently N-Boc protecting the pyrrole nitrogen so that the C-alkylation can be performed at the bromo group via lithium/bromo exchange using nBuLi (1.35 M in hexanes), followed by Lewis-acid (BF3.Et2O, 48wt%) assisted alkyl lithiation, leading to opening of the oxetane to give the free terminal OH group needed to explore the new pocket.
I will attempt the alkylation using the previously tried lithiation route, then try Mg-Grignard conditions if unsuccessful:
13mg of synthesised novel molecule #1 confirmed by 1H and 13C NMR:
84% pure by 1H-NMR; some aliphatic impurities remain despite wash with 1:1 MeCN / Hexanes to remove plasticizers. These are suspected from the nBuLi step in which a plastic rather than glass syringe was used. Glass syringe will be used in all future lithiations.
H NMR (400 MHz, MeOD) δ 8.00 (d, J = 2.5 Hz, 1H), 7.74 (d, J = 3.8 Hz, 1H), 7.60 (dd, J = 8.8, 2.5 Hz, 1H), 7.48 (d, J = 8.7 Hz, 1H), 6.97 (d, J = 3.8 Hz, 1H), 4.59 (s, 1H), 3.82 (t, J = 6.5 Hz, 2H), 3.08 (t, J = 6.5 Hz, 2H).
H-NMR in agreement with proposed structure base on the following: Five aromatic signals; 2x thiophene protons identified at 7.74 (d, J = 3.8 Hz, 1H) and 6.97 (d, J = 3.8 Hz, 1H). 3x phenyl protons identified at 8.00 (d, J = 2.5 Hz, 1H), 7.60 (dd, J = 8.8, 2.5 Hz, 1H), and 7.48 (d, J = 8.7 Hz, 1H). 2x aliphatic CH2 signals identified at 3.82 (t, J = 6.5 Hz, 2H) and 3.08 (t, J = 6.5 Hz, 2H). 1x OH proton identified at 4.59 (s, 1H). Amide NH proton faintly visible at 10.09 (s).
13C NMR (101 MHz, MeOD) δ 162.80, 150.55, 139.86, 138.11, 133.29, 131.51, 130.67, 128.01, 127.57, 123.28, 121.34, 63.45, 34.63.
Carbon NMR in agreement with product, with thirteen unique 13C environments identified. 2x aliphatic CH2 identified at 63.45 (CH2OH) and 34.63 (CH2CH2). 1x amide carbonyl identified at 162.80 (C=O(NHR)). 10x aromatic carbons at 150.55, 139.86, 138.11, 133.29, 131.51, 130.67, 128.01, 127.57, 123.28, 121.34.
[M+H]+ : measured = 315.9981, theoretical = 315.9966.
IR data to follow shortly.
Brainstorm ideas for molecule #3 (right hand side); reference here: J. Med. Chem. 2015, 58, 18, 7286–7309 https://pubs-acs-org.libproxy.ucl.ac.uk/doi/full/10.1021/acs.jmedchem.5b00560
Microwave-promoted formation of 2-acylpyrroles in good yields is iminyl radical cyclizations, terminated by trapping with TEMPO, affording functionalized adducts without using toxic and hazardous reagents and using alkynes as radical acceptors: https://pubs.acs.org/doi/abs/10.1021/ol5035047?casa_token=DuSsH_EEp3cAAAAA:Pcx_ot1QMTBt5G2tRrcPxBD33mmKEGnyvZHxGGNmHQMB35Xvuvu7lGkPjMdGGEDEvZIAjO0bkKm1vc6x
So far the plan looks like this:
Could this also be a possible route? It allows addition of the alkyl group to the pyrrole in the presence of an EWG (ester group) which could subsequently be used to introduce the amide maybe? But would be missing a CH2 group between the pyrrole and amide Reference: https://doi.org/10.1002/ange.201301154
@jemimahaque good spot! This could be a useful alkylation method. As for the ester group, this isn’t necessarily a problem since the pharmacophore model actually preferred the carbonyl at position 2, so we could have that extra CH2 on the NH side of the amide bond instead (I.e. an extra CH2 on the amine that we are coupling with). This also means we could lengthen / change that RHS of the molecule by first hydrolysing the ester to the carboxylic acid, then swapping in different amines for the amide coupling to make analogues, as we have just done for molecule #1.
@mattodd @edwintse I wonder how important it is that the hydrophobic chain is in the 3-position on the pyrrole? especially given that one of those products looks suspiciously similar to our pharmacophore…
@jemimahaque good spot! This could be a useful alkylation method. As for the ester group, this isn’t necessarily a problem since the pharmacophore model actually preferred the carbonyl at position 2, so we could have that extra CH2 on the NH side of the amide bond instead (I.e. an extra CH2 on the amine that we are coupling with). This also means we could lengthen / change that RHS of the molecule by first hydrolysing the ester to the carboxylic acid, then swapping in different amines for the amide coupling to make analogues, as we have just done for molecule #1.
Yep, that sounds good.
A possible method for adding the CH2?:
Hmm you could.. although I’m not sure how keen I’d be to use dioxane, thionyl chloride and diazomethane in a synthesis! Also a good reminder why eating in the lab is never a good idea! Maybe we’ll consider as a last resort.
Great work guys. Nice ideas, nice lit work. @TomkUCL that molecule 3cd looks interesting. If we pretend for a moment that we can access that in quantity, and on the assumption that that building block is not commercially available, we could think about enumerating all the molecules we can make with it (then see if they fit well). So the building block is where the carboxylic acid is free (coupling to amines) and the other end is the free OH. Can you just outline the synthetic scheme for it so we can present to the UNC team and see how we might enumerate the possibilities?
Sure thing, here is the general scheme with references for each step that would allow us to access a range of pyrroles mimicking Heba's pharmacophore:
1a --> 1b ; Angew. Chem. Int. Ed. 2013, 52, 6080 –6083. https://doi.org/10.1002/anie.201301154
1b --> 1c ; Chemistry - A European Journal, 23(14), 3300-3320; 2017. https://scifinder.cas.org/scifinder/view/link_v1/reaction.htmll=T5OKcO0Ri0Vxs5SgbYOjGpsSliqn_MPPwSTko7IUq3sAuA71bfxW0bF_gtFKqvXv
1c --> 1d (aliphatic amines) ; J. Med. Chem.2015, 58, 7286−7309. http://dx.doi.org/10.1021/acs.jmedchem.5b00560
1c --> 1e (electron-poor anilines) ; Tetrahedron Letters 63 (2021) 152719. https://doi.org/10.1016/j.tetlet.2020.152719
As for the building block, we should be able to use this route:
From the scope of the 1a --> 1b paper it looks like there are a lot of known variations for that aliphatic side chain we could try already:
I will start looking for which starting materials we have and what we would need to order to try these here:
Starting pyrrole ester is cheap with 120 vendor choices, can be shipped in 1 week from Apollo UK for £7.50 / 5g. https://store.apolloscientific.co.uk/product/ethyl-1h-pyrrole-2-carboxylate
Looks like Enamine can make the free-OH / EtOOC building block, https://www.enaminestore.com/catalog
Currently checking on analogues of this molecule
Great. So can we generalise to a building block (or a generic building block with R groups) that we know we can make but which is not commercially available? i.e. if we could access it (if it can be included in "make on demand" libraries, then we could access new chemical space, which we can ask Kostya to enumerate. See what I mean?
Yeah, so we can either make the COOH / free OH building block , then enumerate from their with different ethers and amides... or we can alkylate first with different alkyl chains, then hydrolyse the ester to the carboxylic acid, which we can then enumerate with different amines. There's numerous options, so I suppose it just depends on what pocket interactions we think are more important.
Nitrile building block is not available on enamine or SciFinder, but can likely be made using the Pd-catalysed alkylation approach above using the Apollo building starting material.
Looks like the COOH/ CN intermediate is not made by enamine and the final small molecule is novel.
Here's some novel compounds which we could make using the aforementioned method, which are also not commercially available from Enamine:
We can also explore routes to making other heterocycle analogues.
Might be a nice quick way of adding alkyl groups to either side of those pyrroles, then further functionalising (e.g. the nitrile and ester FGs):
Synlett 2005(16): 2425-2428 DOI: 10.1055/s-2005-87268 https://www.thieme-connect.de/products/ejournals/html/10.1055/s-2005-872688?update=true
Notes from today's discussion
I think this xanthate radical chemistry method is also useful: https://doi.org/10.1039/B306966D
I don't think the scope of this has been fully defined, but it could be worth seeing if it works with different substituents on the heterocycle? e.g.
If it works, the ester can be hydrolysed and ketone can be selectively reduced using Zn/HCl to give an alkyl chain, and we can access this building block - which you cannot buy on Enamine:
Also, the ester group on the xanthate may be replaced with a CN group: https://doi.org/10.1039/C2OB25169H
Since molecule no. 3 (pictured below) is proving a little trickier, I have a slightly modified structure that we should have an easier time synthesising quickly (see below). References are as follows:
1) Angew. Chem. Int. Ed. 2013, 52, 6080 –6083 https://doi.org/10.1002/anie.201301154
2) Synthesis 2019; 51(21): 4085-4105 DOI: 10.1055/s-0037-1611904
3) Ishihara, K., Ohara, S., & Yamamoto, H. (2000). Direct Condensation of Carboxylic Acids with Alcohols Catalyzed by Hafnium(IV) Salts. Science, 290(5494), 1140–1142. http://www.jstor.org/stable/3078404
This paper might be useful later for adding some space-filling alkyl groups on either side: https://pubs-acs-org.libproxy.ucl.ac.uk/doi/full/10.1021/cn500054n
(1H-PYRROL-2-YL)-ACETIC ACID (79673-53-3) is available from Apollo at £150 / 1g with a lead time of 1-2 weeks, which would allow for a simple amidation using adamantylamine (which we have already). https://store.apolloscientific.co.uk/product/1h-pyrrol-2-yl-acetic-acid?enquiry=true . However it is unlikely that the C-alklyation with (bromomethyl)cyclobutane would be successful without the EWG (ethyl ester) on the other side.
Whilst the first round of compound soaking is taking place, we are still in pursuit of the following compound:
N-protection on the pyrrole is likely to be needed before subsequent C-alkylation at the Beta-position due to the delocalised structure of the pyrrole-Grignard. https://pubs.acs.org/doi/pdf/10.1021/ja00901a048
A variety of N-protecting groups can be used, as discussed in the following article:https://www.sciencedirect.com/science/article/pii/S004040200601341X?via%3Dihub Notably, N-Boc protection provides a reliable protective step, whilst allowing access to a 1-step Boc-deprotection/ Suzuki procedure for pyrrole-arylation.
Alternatively, N-sulfonyl protection provides tunability of the electronic properties of the pyrrole Beta-carbon at which we wish to perform the C-alkylation step. The introduction of a sulfonyl group on the nitrogen of pyrrole is generally accomplished by reacting the pyrrolyl anion with the corresponding sulfonyl halide. The counter ion, the solvent, and the electrophile are all important for avoiding electrophilic substitution at the 2-position. Some of the common bases and solvents for N-sulfonyl protection include metal hydrides (NaH or KH) in THF, DCM or DMF, as well as trialkylamines in DCM or MeCN.
Notably, N-sulfonyl pyrroles (e.g. SES group) can be removed at room temperature with TBAF in THF at room temperature. https://reader.elsevier.com/reader/sd/pii/S0040403997106530?token=41DA92CC3DB06A559B86462F1CB990ACD963BB35D1072735917E4292BB178941914276C2AF97E2F339FA33C223881459&originRegion=eu-west-1&originCreation=20211206100937
Since we already have methanesulfonyl chloride (MsCl) in the lab, this was demonstrated on an alpha-ester pyrrole analogue (35) and this has the highest atom economy of the protecting groups, we will try the N-Ms protection first, followed by Mg-Grignard formation and oxetane alkylation.
A quick Scifinder search leads me to believe that benzene sulfonyl chloride will result in a higher yield and under milder reaction conditions, and BsCl is readily available in our lab.
VS
https://worldwide.espacenet.com/patent/search/family/064069225/publication/CN108794379A?q=CN108794379A
Based on Heba's most recent docking (see thread https://github.com/StructuralGenomicsConsortium/CNP4-Nsp13-C-terminus-B/issues/2#issuecomment-988953912), the following 4-step synthesis should be manageable from available starting materials:
.
@lindapatio do you think you would be able to try docking these two compounds? where n = 0 or 1. Then we can see if these are worth chasing these further.
We are currently thinking standard amidation conditions using coupling reagents currently available in the lab (HATU, HMBU), however mechanochemistry, microwave and BTFFH should also be considered:
BTFFH: Fluoro-N,N,N ,N -bis(tetramethylene)formamidinium hexafluorophosphate = 99.0 T 164298-25-3 (sigmaaldrich.com).
It may also be possible to form the azide from the halo precursor in-situ in a single azidation/click step, as shown here:
Other potential linkers?:
Reference: https://doi.org/10.1016/j.tetlet.2018.03.013
With more derivatives:
Reference: https://doi.org/10.1021/jm2017144
Reference: 10.1002/chem.200701769
Synthetic accessibility is definitely an important thing, but let's get, say, 5 potential structures together and send over the SMILES to @H-agha and @kipUNC and @lindapatio so that they can see if the structures minimise to something that looks like it might bind properly. Then we can work with that and refine our structures.
Synthetic accessibility is definitely an important thing, but let's get, say, 5 potential structures together and send over the SMILES to @H-agha and @kipUNC and @lindapatio so that they can see if the structures minimise to something that looks like it might bind properly. Then we can work with that and refine our structures.
@mattodd Do these look reasonable?
Corresponding SMILES are: O=C1C2=CC=CC=C2CN3N=NC(C4=C(N1)C=NN4C)=C3 O=C1C2=CC(C3=CC=CC(O)=C3)=CC=C2CN4N=NC(C5=C(C1)C=NN5C)=C4 O=C(NC1=C2N(C)N=C1)C3=CC(C(F)F)=CC=C3CCN4C(C2)=CC=C4 CN1N=CC(OC(C2=CC(C3(C4)CC4(C(F)(F)F)C3)=CC=C2CC5=C(O6)C=CC=C5)=O)=C1C6=O O=S1(C2=CC(OCC3(C#N)COC3)=CC=C2CN4N=NC(C5=C(N1)C=NN5C)=C4)=O
Hi @TomkUCL, I can definitely try docking those two compounds. Or should I work with the five compounds you just posted?
Well, I'm wondering instead whether we keep the binding bit simple for now, and we vary linkers in existing structures. Like this
then we model these (and ones based on the other cyclic systems above) to see what the optimal connectivity might be to present the protein-interacting groups in the best way (i.e. along the most similar vectors to those seen in the frags). Then we worry about how to make the promising ones.
@mattodd Ok so that would give 9 unique structures to be docked by @lindapatio as you have drawn them, then @jemimahaque and I can look into routes for making the top 3 scorers?
Well, 9 just for the scaffold with the triazole, @TomkUCL . But add in the others you drew and you'd have, what, 100? That's more the kind of number we'd want to be docking, and then looking to make the top 3? Something like that?
Other potential linkers?:
Reference: https://doi.org/10.1016/j.tetlet.2018.03.013
With more derivatives:
Reference: https://doi.org/10.1021/jm2017144
Reference: 10.1002/chem.200701769
Here are the SMILES for these linkers @H-agha :
CN1C=C(NC(C2=C3C=CC=C2)=O)C(C4=CNN=C4CC3)=N1
CN1C=C(NC(C2=C3C=CC=C2)=O)C(CC4=CNN=C4C3)=N1
CN1C=C(NC(C2=C3C=CC=C2)=O)C(C4NCCN4CC3)=N1
CN1C=C(NC(C2=C3C=CC=C2)=O)C(CC4=CC3=NN4)=N1
CN1C=C(NC(C2=C3C=CC=C2)=O)C(CC4=C(CC)C(CC)=C3N4)=N1
CN1C=C(NC(C2=C3C=CC=C2)=O)C(CC4=NC3=NO4)=N1
CN1C=C(NC(C2=C3C=CC=C2)=O)C(CC4=NN=C3O4)=N1
Well, 9 just for the scaffold with the triazole, @TomkUCL . But add in the others you drew and you'd have, what, 100? That's more the kind of number we'd want to be docking, and then looking to make the top 3? Something like that?
@lindapatio Here are the first set of compounds based on @H-agha molecule 1C but with a triazole linker for your docking experiments, with the SMILES codes below. Each molecule is labelled as "Linker type - no. of x carbons in linker - no. of y carbons in linker" e.g. A10 = triazole - 1x CH2 - 0x CH2. Hope that makes sense? I will do the same for the other compounds discussed above, but that should hopefully be sufficient to start with. As @mattodd discussed, the motif optimised for binding is unchanged in these, it is just the linker that varies. Thanks!
SMILES:
A00 : O=C1C2=CC=CC=C2N3N=NC(C4=C(N1)C=NN4C)=C3
A10 : O=C(NC1=C(C2)N(C)N=C1)C3=CC=CC=C3N4N=NC2=C4
A20 : O=C(NC1=C(CC2)N(C)N=C1)C3=CC=CC=C3N4N=NC2=C4
A01 : O=C1C2=CC=CC=C2CN3N=NC(C4=C(N1)C=NN4C)=C3
A11 : O=C(NC1=C(C2)N(C)N=C1)C3=CC=CC=C3CN4N=NC2=C4
A21: O=C(NC1=C(CC2)N(C)N=C1)C3=CC=CC=C3CN4N=NC2=C4
A02 : O=C1C2=CC=CC=C2CCN3N=NC(C4=C(N1)C=NN4C)=C3
A12 : O=C(NC1=C(C2)N(C)N=C1)C3=CC=CC=C3CCN4N=NC2=C4
A22 : O=C(NC1=C(CC2)N(C)N=C1)C3=CC=CC=C3CCN4N=NC2=C4
Here is the glide score for the first set of cyclic compounds that Tom suggested yesterday. I added the 2D interactions and some 3D photos. Generally, scores are still low and the highest score is -3.89 Tom- cyclic- GLIDE SCORE.docx
Here is the glide score for the first set of cyclic compounds that Tom suggested yesterday. I added the 2D interactions and some 3D photos. Generally, scores are still low and the highest score is 3.89 Tom- cyclic- GLIDE SCORE.docx
Great, thanks Heba. It's a shame about the low scores, but your comments should provide some useful insights.
Great, thanks Heba. It's a shame about the low scores, but your comments should provide some useful insights.
We need to design more, different scaffolds to improve the scores. Set A seems to be better. Here is the glide scores and 2d interactions for Set A- cyclic structures with different linker length. Set A- cyclic with different linkers.docx
@TomkUCL Here is the minimised affinity scores for A10 - A22. For some reason A00 was not docking, I tried to input the SMILES first but came up as an error. Then I tried to redraw the compound manually but it still came up as an error, apparently the compound is not "readable". But the rest of the compounds had no issues. Compound A10 looks quite good in terms of results. Hope this helps! Cyclic Compounds A00-A22.pdf
Here is the glide score for the first set of cyclic compounds that Tom suggested yesterday. I added the 2D interactions and some 3D photos. Generally, scores are still low and the highest score is 3.89 Tom- cyclic- GLIDE SCORE.docx
Great, thanks Heba. It's a shame about the low scores, but your comments should provide some useful insights.
Just to make sure, those 3.89 actually -3.89 correct?
Here is the glide score for the first set of cyclic compounds that Tom suggested yesterday. I added the 2D interactions and some 3D photos. Generally, scores are still low and the highest score is 3.89 Tom- cyclic- GLIDE SCORE.docx
Great, thanks Heba. It's a shame about the low scores, but your comments should provide some useful insights.
Just to make sure, those 3.89 actually -3.89 correct?
Yes, it is -3.89. I corrected it.
@TomkUCL Here is the minimised affinity scores for A10 - A22. For some reason A00 was not docking, I tried to input the SMILES first but came up as an error. Then I tried to redraw the compound manually but it still came up as an error, apparently the compound is not "readable". But the rest of the compounds had no issues. Compound A10 looks quite good in terms of results. Hope this helps! Cyclic Compounds A00-A22.pdf
For A10, what happens to the mean binding affinity if the amide is swapped for a ketone, as you discussed with @jemimahaque on the thiophene compounds?
Great, thanks Heba. It's a shame about the low scores, but your comments should provide some useful insights.
We need to design more, different scaffolds to improve the scores. Set A seems to be better. Here is the glide scores and 2d interactions for Set A- cyclic structures with different linker length. Set A- cyclic with different linkers.docx
Based on your feedback here are some additional scaffolds - i've added the furan at the top to try and also exploit pi-stacking interactions in the upper TRP / PHE pocket..
SMILES:
T-3-1: O=C1C2=CC(C(F)F)=CC=C2CCN3C=CC=C3CC4=C(C=NN4CC)N1
T-3-2: O=C1C2=CC(C(F)F)=CC=C2CCN3C=CC=C3CC4=C(C=NN4CO)N1
T-3-3: O=C1C2=CC(C(F)F)=CC=C2CCN3C=CC=C3CC4=C(C=NN4CF)N1
T-3-4: O=C1C2=CC(CC3=CC=CO3)=CC=C2CCN4C=CC=C4CC5=C(C=NN5CC)N1
T-3-5: O=C1C2=CC(C3=CC=CO3)=CC=C2CCN4C=CC=C4CC5=C(C=NN5CO)N1
T-3-6: O=C1C2=CC(C3=CC=CO3)=CC=C2CCN4C=CC=C4CC5=C(C=NN5CF)N1
T-2-1: CN1C(C(O)C2=CN(CC3=CC(C4=CC(O)=CC=C4)=CC=C3C5=O)N=N2)=CC(C5)=N1
Based on your feedback here are some additional scaffolds - i've added the furan at the top to try and also exploit pi-stacking interactions in the upper TRP / PHE pocket..
I think it is better to use F rather than CF2 or CF3. Compounds that has CF2 are outside the pocket. Furan seems to be a good idea. It sets inside the pocket and form interaction with Arg503.
Compounds that showed good scores interact with Pro593, Asn503, and Arg502. In addition, if we extend the compound to have interaction with Glu 591, and maybe Trp506 it could be better.
Based on your feedback here are some additional scaffolds - i've added the furan at the top to try and also exploit pi-stacking interactions in the upper TRP / PHE pocket..
I think it is better to use F rather than CF2 or CF3. Compounds that has CF2 are outside the pocket. Furan seems to be a good idea. It sets inside the pocket and form interaction with Arg503.
Compounds that showed good scores interact with Pro593, Asn503, and Arg502. In addition, if we extend the compound to have interaction with Glu 591, and maybe Trp506 it could be better.
I am not quite sure what the distances are like in the pocket, but I have this structure based on the highest scoring hit from Set T-3-Cyclic and some of your suggestions.
SMILES: O=C1C2=CC(C3=CC(COC(N)=O)=CO3)=C(CCO)C=C2CCN4C=CC=C4CC5=C(C=NN5C6=CC(O)=CC=C6)N1
I'm also curious as to whether the cyclic structure is required given that it doesn't seem to show any interactions, so i've also included the SMILES for this one:
SMILES:
O=C(NC1=CN(C2=CC(O)=CC=C2)N=C1)C3=CC(C4=CC(COC(N)=O)=CO4)=C(CCO)C=C3
@TomkUCL
mean minimised affinity = -5.31 (vs -5.32 for original) range = -4.5 to -6.33
Other potential linkers?:
![]()
Reference: https://doi.org/10.1016/j.tetlet.2018.03.013 With more derivatives:
![]()
Reference: https://doi.org/10.1021/jm2017144
Reference: 10.1002/chem.200701769
Here are the SMILES for these linkers @H-agha :
CN1C=C(NC(C2=C3C=CC=C2)=O)C(C4=CNN=C4CC3)=N1
CN1C=C(NC(C2=C3C=CC=C2)=O)C(CC4=CNN=C4C3)=N1
CN1C=C(NC(C2=C3C=CC=C2)=O)C(C4NCCN4CC3)=N1
CN1C=C(NC(C2=C3C=CC=C2)=O)C(CC4=CC3=NN4)=N1
CN1C=C(NC(C2=C3C=CC=C2)=O)C(CC4=C(CC)C(CC)=C3N4)=N1
CN1C=C(NC(C2=C3C=CC=C2)=O)C(CC4=NC3=NO4)=N1
CN1C=C(NC(C2=C3C=CC=C2)=O)C(CC4=NN=C3O4)=N1
Hi @H-agha, here are a few other suggestions for the cyclic structures that I think may be synthetically accessible?
Update slides from 18/02/2022.
This route is based upon exploiting the innate 2- and 4-position reactivity of pyrrole. The Fenton-type radical addition of the iodoalkane ester is based on a previous reaction using pyrrole as the substrate. Since the mechanistic pathway proceeds via a pyrrole cation intermediate, I expect the 4-position reactivity to be enhanced even more so in this substrate over pyrrole. This is because the added alkyl group is electron-donating and (thereby carbocation-stabilizing) because the electrons around the neighbouring carbons are drawn towards the nearby positive charge, thus slightly reducing the electron poverty of the positively-charged carbon.
Given that the entire premise of this reaction is i) radical generation from an alkyl halide, and b) SET oxidation of a stabilized-radical intermediate, I suspect this would be an ideal reaction to try using organic electrochemistry or even electrocatalysis, thereby cutting out the need for hydrogen peroxide and DMSO.
Slide 2: Current target fragment. Slide 3: Current route (in-progress). Slides 4-9: Trial reactions with links to ELN entries.
Update 21/02/2022:
In addition to the shorter (but riskier) route above, I have started preparing the following backup route for the third fragment (asymmetric pyrrole). Once the acid is made, amide coupling should be trivial. A 69% yield has been reported for the bromination reaction in step 2 (CAS RN 27807-85-8).
Advice/ discussion on the following would be appreciated:
https://pubs.acs.org/doi/full/10.1021/ja021284l > 3000-fold rate increase with Et3N and LiHMDS.
https://pubs.acs.org/doi/abs/10.1021/ja030168v > 3000-fold rate increase with nBu3N and LiHMDS.
In step 4, does the 1,3-dicarbonyl require protection as the silyl enol ether BEFORE the enolate alkyl bromide SN2 step to prevent deprotonation by base residual from the lithium enolate-forming step?
In step 5, is MW-assisted pyrrole synthesis recommended to expedite the synthesis? If so, is it robust?
@edwintse does anything here seem like a potential issue?
OK, closing this issue for now because effort has re-focussed on compounds that are not derivatives of the fragments. This issue is clearly linked from the Story So Far and the Desirable Compounds Not Yet Synthesised page.
Some of the cyclic structures remain interesting targets, as are some of the novel pyrrol-containing building blocks that were planned.
Just briefly, based on the analysis of the fragments done by Heba, we have decided on 3 target compounds to make based on her pharmacophore/merging of overlapping fragments (more background on this will follow shortly on the Wiki).
Specific issues for the planning of synthetic routes for each compound will follow. @TomkUCL