Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Mur Ligase Online Research Meeting 2pm July 6th 2021 #47

Open
4 of 11 tasks
mattodd opened this issue Jul 5, 2021 · 7 comments
Open
4 of 11 tasks

Mur Ligase Online Research Meeting 2pm July 6th 2021 #47

mattodd opened this issue Jul 5, 2021 · 7 comments
Assignees
@mattodd
Labels
Meetings Meetings

Comments

@mattodd
Copy link
Contributor

mattodd commented Jul 5, 2021 •
edited

Meeting follows on from June 2021 meeting at #44.
When: Tue July 6th 2pm London (BST).
Where: https://ucl.zoom.us/j/91614633419
Recording: https://youtu.be/lAWCUQGDGtY

Actions from Last Time

  • @WieselHenson to complete the synthesis of the AZ alcohol compound. Question: when this compound is sent to @bartrum, which enzyme? @eyermanncj says MurD aceto, pseudo. @mattodd asked about the value of a negative control (compound known to be inactive). Useful for machine learning. Would need Pseudo MurC biochemical assay to verify the inhibition (currently don't have).
  • @Rebecca-Steventon to post data of @LizbeK fragment screen (Done: https://github.com/opensourceantibiotics/murligase/wiki/Allosteric-Pocket - but @drc007 would like structures to be added, pretty please) and perform screen of @danaklug elaborated fragments.
  • Laura to continue crystallisation trials of MurD and MurE in Warwick, with the plan to take crystals to Diamond. Update: MurE is faster. MurD is 1-2 months.
  • @eyermanncj, Chris and @bartrum to pursue AZ compound shipment for crystallisation trials.
  • [x ] @jhjensen2 to extract in vitro inhibition data from published literature and use towards the construction of a predictive model. See below. Predictive model has been started based on docking. This action not pursued further for now.
  • Peter Horanyi to continue crystallisation of e.g. MurC with Enamine compounds sent by @eyermanncj. (Structures here from @loriferrins)
  • Chris Dowson to liaise with others to establish protein needs, and what's needed from SSGCID.
  • @mattodd and Chris D to interact with Alphafold. With Laura they'll come up with a defined piece of work and circulate.
  • @bartrum to liaise with David Baker lab about potential protein structure prediction work. <- would suggest this is now moot given Protein Structure Prediction Tools More Generally Available #48.
  • Laura to liaise with Peter and Lizbe to see if a soak of the @eyermanncj Enamine compounds is possible at Diamond.
  • Following some discussion at May meeting, @eyermanncj asked whether there might be the possibility of a biochemical screen against another non-ATP site, and Chris D would chat to Adrian about that. Marking as complete because @Rebecca-Steventon has indeed started biochemical screening.
  • June meeting posted to Youtube.

Additional items for discussion:

  • New docking studies vs MurC described in Docking studies on MurC ligase #46 discussed by @jhjensen2 @drc007 @dkoes and @cstein. The approach aims to find new molecules that are predicted to bind better than commercially-available molecules (which are predicted first, as a baseline <-- indicating what scores to beat). The structures shown in the meeting are commercial. The genetic algorithm (which is expected to find things that are not commercially available) was running at the time of the meeting. Relatively simple things to predict (on an input set size of ca. 400k compounds) like logP (and even synthetic accessibility) can be included in the algorithm. @chrisdowson1 was hoping for the application of an Entryway filter, to aim for compounds that might accumulate intracellularly. Question was whether the Entryway web tool could easily deal with batches - we're not 100% sure. Could in theory do a post- or pre-screen. Hergenrother could provide code, to allow us to implement? Action on @mattodd to loop in Jan and Paul H to see what can be done. Aside from the genetic algorithm, the existing predictions show predicted strong binders from Zinc. How do we source these Zinc compounds? They are not necessarily "in stock": MolPort suggests only two of them are in stock, though the ones suggested do not look difficult to make. (note in this case the search was done with a smaller set of molecules from Enamine vs a larger set from Zinc). Could use a larger set (250k) from Enamine; advantage being all purchasable. The genetic algorithm compounds will need to be made, and at that point we'd likely benefit from an automated retrosynthesis algorithm that Jan can implement.). Jan is hoping for experimental validation of the genetic algorithm approach (to show it's worth doing). Follow-up Q1: It would be useful if we could generate the pictures of the poses of the best predicted binders, to learn about what may be productive places to bind to the enzyme. @cstein may have these. Follow-up Q2: Could we re-dock the e.g. top 100 hits vs another mur ligase, as part of the multi-targeting approach (or even against the same enzyme in another species...)? @jhjensen2 said yes, and even better might be to have the genetic algorithm optimise simultaneously vs two Mur targets. Caveat from Jan: this is docking ("guessing with style", said Chris). Peter Horanyi asked whether we could screen out compounds that are similar to others, thereby increasing diversity; Jan said yes via calculating a similarity score (and eliminating any with low scores that are similar to ones with high scores), though Jan does not want to triage compounds before the medchemist sees them (i.e. maybe we just need clustering of similar structures in a tree). Machine learning can be done too, which can kick in better when we have ca 50 experimental data points; for validation of the docking we'd need fewer datapoints. For visualization of the compounds in this project we can all use Datawarrior to display structures and cluster by similarity? (sdf files are useful for this). @Yuhang-CADD asked how many samples we can test biologically, to validate the approach. This is a resources question... Yuhang also asked whether we could score these suggestions with other software; Jan said yes, this is possible, but you need expertise in those platforms; SMILES are available, so anyone can have a go at such consensus modelling and provide inputs (i.e. need people to drive those comparisons).

  • Possible publicising of the challenge..?

  • @Rebecca-Steventon's data on Lizbe's fragments. Where are the structures? Action on @LizbeK to dig those up and put somewhere. Also new data from Becca on Dana's elaborated fragments. Biochemical hits found at 1 mM. 78 compounds tested. 11 compounds gave >80% inhibition vs 2.5 ug/mL Streptococcus agalactiae(?) MurD, and those were run against Streptococcus agalactiae MurE with some inhibiting both (i.e. dual inhibition!). Most compounds inactive (i.e. not the case that, at those high concentrations, everything is active) and some compounds hit one enzyme better than the other.

  • How to gain structural insights, i.e. motifs in the actives that are not in the inactives, etc? @chrisdowson1 suggested sending the actives to Jan to verify predicted binding? Useful, but fragments are even more difficult to predict. Co-crystal formation desirable if the hits confirmed - at the moment Laura was hoping for a better crystal system for MurD. Currently Streptococcus agalactiae MurD and E. coli MurE.

  • @eyermanncj suggested that we need full 10-point dose response curves to be sure of the inhibition observed. Dual inhibitors are the priority. @Rebecca-Steventon said this was doable if targeting the small number of compounds exhibiting dual inhibition (not all 78) (Action on Becca). The assay is functional, though the predicted binding site is allosteric (not ATP-competitive) (need to point to where (on the wiki) these are binding). Question: how's the conservation of the site out in the allosteric position across ESKAPE/other pathogens? Status of MurC (pseudomonas) screen in Warwick?

  • Shipment of Yuhang's AZ compound will take place as soon as the identity is confirmed by NMR. e.g. to Bart first. Peter has Acetobacter and Psuedomonas MurC.

  • Atomwise shipment - nearing the end of the legal quest, hopefully.

  • Status of ELF Screen application?

  • Note that experimental screening is done in the presence of metal ions (e.g. Mg2+); crystallography - can't tell what is a water and what is a magnesium ion; computational docking removes waters and metal ions on the assumption that small molecules would displace those.

Overall Aims remain:
i) Structures of two mur ligases bound to the same small molecule.
ii) Identify new starting points for inhibitors.

Future logistical aim: devise platform legal agreements within OSA in order to simplify the MTA-related paperwork, which is proving problematic. (Could we transform OSA into a charity (though since none of us would work for this charity, it's not clear how much that might help)).

Next meeting: Tuesday 3rd August 2pm BST London
Location: https://ucl.zoom.us/j/98754959307
@mattodd mattodd added the Meetings Meetings label Jul 5, 2021
@mattodd mattodd self-assigned this Jul 5, 2021
@mattodd mattodd mentioned this issue Jul 5, 2021
Open
@drc007
Copy link
Contributor

drc007 commented Jul 6, 2021 via email

@jhjensen2
Copy link

I'll try to make the meeting

@mattodd
Copy link
Contributor Author

mattodd commented Jul 6, 2021

@drc007 yes, both good ideas and I've added to the agenda. @jhjensen2 would be very good if you could come along and describe possible next steps.

@Yuhang-CADD
Copy link
Contributor

//Synthesis update://
image

@Yuhang-CADD
Copy link
Contributor

Yuhang-CADD commented Jul 6, 2021 •
edited

//Next steps://
image

@chrisdowson1
Copy link

chrisdowson1 commented Jul 6, 2021 via email

@jhjensen2
Copy link

Here's the summary of our (@cstein and myself) tasks as I understand them. Let me know if I missed something

  • Find more purchasable compounds with better docking scores than AZ8074. Since the purchasable ZINC compounds found by Molport are from Enamine, the best course of action is probably to dock the larger libraries from Enamine.
  • Dock to another ligase. What x-ray structure should we use? Are we looking for molecules that inhibit both, or one but not the other?
  • Compute "entryway" scores. Get access to an off-line version or make our own based on the underlying data.

@mattodd mattodd mentioned this issue Jul 27, 2021
Open
24 tasks
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@mattodd mattodd

Labels
Meetings Meetings
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
5 participants
@jhjensen2 @mattodd @drc007 @chrisdowson1 @Yuhang-CADD