Follow up for the cluster 15 fragment hit for MurE Ligase

[drc007]

This fragment is binding close to the ADP site.

Details of the fragment screen results

[drc007]

Update. We now have a few more crystal structures and another fragment binding at this site.

[drc007]

You can now view these fragments using NGLviewer

I've also recorded a very quick intro movie here

[danaklug]

Hi, I'm a new postdoc working with Mat at UCL. Focusing on fragment 0206, I think there are some analogs that can be made fairly quickly that would probe the SAR around key parts of the compound.

For the pyrimidine: It would be nice to know whether both nitrogens and both methyls are essential to activity, or if we can pare it back a bit. On the flip side, it would also be good to know where some bulk is tolerated. Unfortunately, all of these analogs require different starting materials.

For the cyclobutanol: I would like to start by probing various HBD/HBA, and the flexibility of the cyclobutane ring. Several of these compounds can be made from the same starting material.

For initial synthesis, four of the proposed analogs can be made using two commercially available starting materials. The synthesis should be straighforward, involving some manipulation of protecting groups.

This should hopefully be enough to get started!

[danaklug]

I also have a few questions about the initial screen:

• Were there any structurally related compounds included in the screen that did not bind?
• Is there a number associated with binding for fragment 0206 (for example an IC50, Kd, etc)? It would be nice to compare numerically back to the original hit.
• I assume a cis/trans mixture was used in the original screen but could anyone just confirm this?

Thanks!

[drc007]

@danaklug great to have you on board. The fragment screen was run at Diamond at mM concentrations using X-ray detection which means it is very much a yes/no assay. It is likely that initial fragments will bind too weakly for detection in a biological assay, so no IC50 etc.

Currently there is not a way to structure-based searching via the wiki but I have profiled the Diamond fragment collection here and I've compile a list of all 768 fragments used in the screen here. To aid exploration of the list I've implemented a javascript real time text-based searching (simply type "bromo" into the text box top left on the page). I had suggested trying to write a substructure search facility into the web page but the folks at Diamond felt it would be a conflict with their plans.

[danaklug]

Gotcha. I'll have a look through the database to see if I can find any related compounds - might give us a head start on the SAR.

Thanks!

[drc007]

Regarding the stereochemistry the stereochemistry of the sample in the fragment collection is undefined (I think), however the X-ray structure shows only a single isomer shown below (which could have been selected by the protein from a mixture).

It would of course be useful to confirm this with authentic samples

[danaklug]

Below are related compounds included in the screen that were non-binders. My main observations are that the orientation of the HBD and the presence of both H-bonds are important for binding - nothing we couldn't have surmised by looking at the crystal structure but good to have confirmed.

I have also played around with the docked ligand structure and to me it seems like the methyl at the 2-position is pointing into the protein towards a potential pocket, while the 6-position methyl is solvent-exposed. However, I don't have much experience with interpreting crystal structures so Chris, perhaps you have some insight as to whether that's a reasonable interpretation? If so, is there anything known about the identity of that pocket (hydrophobic, polar, etc) that we could use in compound design?

My main takeaway from all this would be to deprioritize any initially proposed analogs without both the HBA and HBD in the short term, and in the long term a) look at ways to expand the structure at the 2-position, and b) keep the 6-position in mind as a potential handle for attaching solubilizing groups. Thoughts?

[drc007]

@danaklug I've attached a schematic of the binding interactions. The key interactions are hydrogen bonds between Asn311 and Ser310 with the hydroxyl, and a pi-stacking interaction with Phe307.

As you say the 6-position methyl is pointing towards solvent whilst the 2-position points into a fairly large pocket containing a number of water molecules. I've added an image below, green is lipophilic regions, purple is polar and red is solvent exposed.

Building from the 2-position looks like worthwhile effort, simple alkyl substituents could access the more lipophilic regions and displace waters which might be advantageous. There is also Asn308 at the back of the pocket which could contribute potential hydrogen bonding interactions.

I can't see any particular interactions with the nitrogens of the pyrimidine, however it may be that the electronics of the ring favour the pi-stacking interaction. It might be useful to look at the corresponding pyridines.

[danaklug]

Cool! Yes, I've got the pyridines on my list for synthesis. I was also thinking it would be good to prioritize some of the cyclobutyl replacements since that building block is a bit pricey... might be nice if we could replace it with something like a cyclohexyl or phenyl instead.

I think that Suzuki or Buchwald reactions at the 2-position are synthetically feasible... I was also wondering if docking would be a helpful exercise here? I've got some ideas on how to build out but maybe docking some potential ligands would help prioritize the synthesis?

[drc007]

I'm happy look at docking but in my experience when looking at small fragments you can only really use it to rule out molecules that are very unlikely to bind. It is not possible to rank order binding affinities.

[danaklug]

Ah okay. Sounds like it's probably more useful to synthesize a small library and see what happens, I'll look into the synthesis there as well.

[drc007]

Just checked the Enamine record for this compound the stereochemistry appears not to be defined.

I'll ask the folks at Diamond to look at the electron density in more details to check their assigned stereochemistry.

[danaklug]

Okay, great. I've ordered the mixture for now just so I can get some chemistry going but synthesis of the two pure isomers is high on my priority list.

[tkrojer]

Hi @danaklug, just a general comment on the initial fragment screen:
The DSI fragment collection consists of 768 compounds, but we soaked and collected only 114 of them (most of them as duplicates). The selection of fragments was completely random at this point and was mostly dictated by crystal availability and time constraints.

The electron density for the bound ligand is not brilliant, but it seems that only the cis isomer can fit.

[danaklug]

@tkrojer Thanks! That's really helpful - I'll focus my synthetic efforts on the cis isomer. I'll also make a pure sample of the trans just to confirm.

[mattodd]

@drc007 you had previously wondered whether the two compounds here (MUREECA-x0206_1 and MUREECA-x0209_1) also bound in MurD. Apologies if I missed it, but was that question resolved?

[drc007]

@drc007 you had previously wondered whether the two compounds here (MUREECA-x0206_1 and MUREECA-x0209_1) also bound in MurD. Apologies if I missed it, but was that question resolved?

@mattodd I don't think so.

[danaklug]

Below are structures of compounds included in the 2/12/2019 shipment (#12), grouped by design rationale. Hopefully some interesting results to follow!

[rhanson1046]

Dana:
One comment on your proposed synthesis. Be careful about trying to alkylate the exocyclic amino group of 2-/4-aminopyridine/pyrimidines. Usually alkylation under neutral conditions will occur on the ring nitrogen (most nucleophilic site). At best you get a mixture. If you start with the alkylated amine and do the displacement of the 2-/4-chloro group, then you are OK.