Mur Ligase Online Research Meeting February 2022

[mattodd]

Date: Feb 8th Time: 2pm UK time (other times)
Place: https://ucl.zoom.us/j/91379419977
Recording: https://youtu.be/tTzBa5hv3H8 Previous Meeting: #54

From Last Time:

As mentioned in #62 the two key experiments are the first two items here:

1) Elaborated Fragments Inhibiting Two Murs

• [x] @Rebecca-Steventon to complete dose-response experiments with the @danaklug compounds that have been found to inhibit two mur ligases. In meeting: @Rebecca-Steventon presented these data, that suggested of the compounds, OSA759 was the most promising. Follow-ups:

• [x] @Rebecca-Steventon to post the new data below.

• [ ] @KatoLeonard @edwintse to check whether related structures were found to be inactive, i.e. that the inhibitory compounds are likely inhibiting via a specific, rather than a non-specific interaction, and what this tells us about the SAR. e.g. compounds with 3- and 2-pyridyls?

• [ ] Potential follow-up experiments would be to determine if these are competitive inhibitors.

• [x] @LizbeK and @LauraDS1 to look into how best to obtain structures of the @danaklug fragments bound to enzyme

In meeting: @LauraDS1 presented these data, indicating that no protein structures with bound ligand have yet been found in these soaking experiments, though protein with added compounds display conformational changes suggesting binding. A partial structure was seen for MurE and OSA749, but the molecule appeared to be binding in the wrong place.

• [ ] @LauraDS1 to pursue co-crystallisation experiments.

• [ ] @KatoLeonard to trial structure determination experiments with newly-synthesised compounds in #68.

• [x] Are the hits (and non-hits) available to @jhjensen2 for inclusion in the predictive work?

• [x] @LauraDS1 was purifying E coli MurD. @LauraDS1 to update on current status of protein available - was looking promising in #63. Update Jan 13th 2022: "I have gotten very nice looking EcMurD apo crystals. I’m currently assessing their reproducibility and hopefully sending some crystals next week for testing diffraction and soaking. I also got some EcMurD micro crystals in the presence of ADP. I have set up follow-up experiments and I’m waiting for the crystals to grow. Hopefully, they can also be sent to Diamond next week."

• [ ] @KatoLeonard and @ZigBu have been designing/making the next round of elaborated fragments on the assumption that the binding site/pose has not changed: #58. @KatoLeonard 's structures shown in #68 and will be taken to Oxford shortly.

• [x] Preliminary results from @LauraDS1 soaking experiments mentioned in #68 will be discussed.

In meeting: Adrian Lloyd and Anita Catherwood presented data on the development of a high throughput mur ligase assay and on the inhibitory values for a known AstraZeneca compound as well as the known inhibitor ADPCP. The IC50 values for the AZ compound were quite different from those published (here, 1.2 uM, in lit 8 nM vs MurC), and it will be important to compare equivalent assays. The AZ compound is clearly not a dual inhibitor, but some effectiveness vs MurE was seen, and was synergistic with ADPCP.

• [ ] @chrisdowson1 it would be very helpful to have those slide decks for posting here. Could you please pass on?

2) Atomwise Hits

• [ ] @LauraDS1 to complete SPR experiments (MurC, MurD in presence and absence of ADP/AMPPNP) on Atomwise compounds (#55 and on wiki), following the previous crystal structures of several compounds bound (see also #52).
• [ ] @LauraDS1 @Rebecca-Steventon @chrisdowson1 to obtain inhibition data with the Atomwise compounds, to verify whether binding gives inhibition.
• [x] @mattodd to contact Atomwise to obtain clarification on whether the compounds were designed to bind in this way (!)
• [ ] Can we check whether the "target site" Atomwise used included residues in the allosteric pocket where the compounds were found to bind? (see comments) @LauraDS1 @LizbeK

3) Variants of AZ Compounds

• [ ] @Yuhang-CADD to complete synthesis of AZ5595 and the amino derivative of that compound (intended to demonstrate activity vs wild type bacteria) - some updates have been posted e.g. #60. As described in #64 there was a slight risk that the compound made is isomeric with AZ5595, but the chances are low. Some compounds have been shipped to the Rosalind Franklin Institute to obtain microED structures #65
• [ ] @Rebecca-Steventon was going to assay @Yuhang-CADD 's compound here, shipped in August.
• [ ] @Yuhang-CADD compound structure (AZ cmpd 4) to be overlayed with AZ5595 (PDB 6X9N) and AZ8074 (PDB 6X9F)
• [ ] @Yuhang-CADD to ship AZ5595 to SSGCID
• [x] @Yuhang-CADD: Give all his structures OSA compound code and ensure in the OSA compound masterlist.
• [ ] @chrisdowson1 to work on compound transfer from SSGCID to Warwick and update the group.

4) New Protein Structures

Nov 2021: Pseudomonas aeruginosa MurD, SSGCID code PsaeA.17938.a, PDB 7SY9. Jan 2022: Acinetobacter baumannii MurD in complex with ADP, SSG code AcbaC.17938.a, PDB 7TI7.

For the 7TI7 structure, Jan Abendroth pointed out "there are distinct conformational changes between the apo and the ADP-bound structure. RMSD is at 3Å, and especially the C-term is hard to overlay. A curious observation is the carbamylation / carboxylation of Lys 198, now named KCX 198. I have seen this modified Lys in only a few structures within SSGCID. It typically serves as an ‘elongated Glu’ that tends to ligate active site waters or ions." Could Jan please present this and discuss?

More recent updates: #67

Jan was (1/22) collecting a data set on a Psae MurD crystal grown in the presence of ADP. Clearly have ADP bound.

5) De Novo Computational Modelling

• [ ] @jhjensen2 to get docking poses on top scorers. Last data were here and @cstein was going to look for primary amines with good docking scores using genetic algorithms. Note @eyermanncj comments on salt bridges. Updated structures proposed here.
• [ ] @mattodd @edwintse @Yuhang-CADD to examine these compounds for purchase/synthesisability and determine a way forward.

6) Other Potential Starting Points

• [ ] @loriferrins to update on Cyclica collaboration
• [ ] @mattodd @chrisdowson1 to update on status of Euro Lead Factory screen, following discussion with staff at Lygature.
• [ ] @eyermanncj and @chrisdowson1 to speak to the benefits of ordering the Enamine kinase library.
• [ ] @chrisdowson1 to dig up any and all details of the compounds developed/evaluated in the LifeARC project

In meeting: @chrisdowson1 mentioned that data from LifeArc is likely incoming re hits from screens that were performed.

7) Misc

• [ ] Comparative analysis was done by @ZigBu of mur ligase ATP binding sites (#56). Any learnings?
• [ ] @ZigBu also posted (#57) on a paper highlighting that mur domain movement may not be dependent on substrate binding. Again, anything we need to consider?
• [ ] @eyermanncj posted "As a reminder the dynamics of E. coli murD is well established. Here is a “recent” NMR study on E. coli murD."
• [ ] @mattodd to consult on meeting timings - would 2 hours later work for most people? Better than an hour earlier?

8) Mothballs

• [ ] @Rebecca-Steventon has posted data of @LizbeK fragment screen (https://github.com/opensourceantibiotics/murligase/wiki/Pocket-binding-site - but @drc007 would like structures to be added).
• [ ] @LauraDS1 to liaise with Peter and @LizbeK to see if a soak of the @eyermanncj Enamine compounds is possible at Diamond.

[JZuegg]

Hi Mat, Just wondering if any of the compounds have been screened for MIC. In case I'd like to extend the offer for any of the compounds to be screened for MIC at CO-ADD ([co-add.org]).
We have a couple of E.coli mutant (d(LpxC) and d(TolC)) which might be of interets in this context.
Cheers Johannes

[Yuhang-CADD]

Synthetic Plans

[Yuhang-CADD]

Help Needed

Could anyone help dock the following AZ5595 purine core derivatives into the binding pocket of Pae. MurC (PDB: 6x9n) please? We would like to see if these two have a potential to make as the purine core is way cheaper than the original one!

[mattodd]

Hi @JZuegg - thanks for the offer. This is something we will keep in mind for all compounds suspected of having biochemical potency. The compounds @Yuhang-CADD is making displayed excellent potency but zero potency vs the pathogen, and it is in order to improve accumulation that he's making the primary amine derivatives above. Some of the other compounds you may be seeing at the moment are elaborated fragments, and it's too early to tell if they are doing what we hope.

[mattodd]

Great! @Yuhang-CADD @eyermanncj - could one of you please v briefly (since we've a packed schedule) talk us through the result of this docking in 30 seconds on-screen?

[Yuhang-CADD]

Great! @Yuhang-CADD @eyermanncj - could one of you please v briefly (since we've a packed schedule) talk us through the result of this docking in 30 seconds on-screen?

As far as I understand, the result showed that the amine derivative of AZ5595 could be a promising binder as the best binding position (provided by Joe) was overlapping with the AZ5595 (RMSD=0.188) while the strong interactions with key residues remained.

[mattodd]

Great - I think this tallies with what @dayang-us found and is the reason I'm so continously excited/impatient about the molecule you're making, @Yuhang-CADD

[Yuhang-CADD]

Help Needed

Could anyone help dock the following AZ5595 purine core derivatives into the binding pocket of Pae. MurC (PDB: 6x9n) please? We would like to see if these two have a potential to make as the purine core is way cheaper than the original one!

Thanks for Joe's help, the docking results of AZ5595 amino derivative (the top right structure in the orange panel) has been uploaded here DOCKING RESULT

However, we still need two more structures to be docked (down right two structures in the pink panel) just to see if we could get more promising and cheaper analogues.

[mattodd]

Update on the Atomwise discussions: I've made four attempts to contact Atomwise about the nature of their prediction (the X-ray data shows several molecules binding to an allosteric site that we're pretty sure was not identified as the target site). Obviously this slience is a shame (I had hoped for follow-up), but I wonder if we can double-check this ourselves in the meantime? The original workplan document clearly describes that the target site is where substrate (UMA) is bound in PDB 3UAG. However, the diagram at the end of the workplan, reproduced below, indicates a wide net of potential target residues and these are also listed in the document. Is it possible to see whether these residues overlap with the residues interacting with the bound compounds in the subsequent Xstal structures you obtained @LauraDS1 @LizbeK ?

[LauraDS1]

@mattodd Thanks Mat. Yes, I can check those residues.

[LauraDS1]

@mattodd

I found 3 overlaps: 157, 159, 177, and there are couple of residues near residue 180.

There is a difference in the residues numbering between the XChem and the 3UAG models. Those residues in the XChem models are: 184, 186, 204 and 207. I will look into this residue numbering difference. For now, those residues don't show strong interactions except 204 (resi 177 in 3UAG) for some of the hits.

[Yuhang-CADD]

Difference between WYH9-2-P and AZ5595

Hi Anita, thanks for the presentation (really nice results)! Here is the comparison. Sorry for confusing you! Thanks!

[eyermanncj]

Compound W from the AZ paper

[jhjensen2]

Here are the updated #slides Poses.pdf

[eyermanncj]

The purine scaffold hop looks risky as there is a very bad contact with the Y246 OH in that scaffold versus a H-bond interaction between the AZ inhibitor which forms a H-bond to the Y246 OH. Also note that in the AZ inhibitor there is a close contact between the CH andthe carbonyl oxygen of A292. Note that this aromatic CH to carbonyl oxygen has been observed in a number of kinase inhibitors interacting with a kinase hinge carbonyl oxygen.

If the chemistry is realtively easy maybe make a matched pair to confirm the above hypothesis?

@Yuhang-CADD @mattodd @drc007 @LauraDS1 @loriferrins @chrisdowson1

[Yuhang-CADD]

Thanks, Joe, I think I would like to try the chemistry of it and make a matched pair. (if not working then I would stop right there to save time)

Also, is there a chance that the two residues (1 and 2 marked in yellow circles) would rotate/move around to make the purine analogue fit into the pocket and likely to have two hydrogen bonds with the proximal nitrogen (3, marked in a red circle, serving as an electron donor) on the purine core?

[eyermanncj]

@Yuhang-CADD @mattodd @drc007 @LauraDS1 @loriferrins @chrisdowson1

Pae murC Y246 is not able to move much - it is 'sandwiched' between a helix and a beta-sheet. Using Schrodinger, I did a Prime minimization of the purine analog allowing the protein residues within 8 Angstroms of the ligand to be flexible. In the figure below the crystal structure of the AZ compound is gold and the purine analog is yellow. The purine analog does move and the N (your 3) moves to get closer to the Y246 OH but the movement doesn't alleviate the very poor VDW contact with the purine C.

Given this is your PhD work, making the purine is a reasonable objective if the synthesis is reaktively easy. As for it being as active as the AZ compound, I suspect that is unlikely.

[mattodd]

Nice work @eyermanncj @Yuhang-CADD - if the chemistry is simple (should be) then I agree it's worth a punt. Joe, we were obviously wondering whether this purine analog had been part of the original AZ campaign, but not published? Any way that we could find out, e.g. by requesting an internal AZ SMILES search..? An hour in the archive saves you a week in the lab.

[Yuhang-CADD]

Dear Joe @eyermanncj

May I also ask for help with docking the following two structures into MurC binding pocket and see if they would fit? Just to see if these two structures have the potential to sit in the pocket (we would also like to test them against MurC as well just for fun). If it worked, then we can build up molecules with this structure (as they were the cores of p38α kinase inhibitor doramapimod) For example we can build up more complicated structures on the methyl side and the chloride side. Many thanks, Yuhang

[Rebecca-Steventon]

The dose response curves carried out with @danaklug compounds can be seen below with the Hill Slope and IC50 of each compound being shown in the figure legend.

[KatoLeonard]

Hi @Rebecca-Steventon, just a quick question: I know that there is no dose response curve for compound 788 because it precipitated, but what was the reason that compound 749 was not tested? Was it because of the same reason, or was it maybe not considered a dual inhibitor because the MurE inhibition was slightly lower than for the other dual inhibitors? Thank you in advance!