KLHDC2 – The Next Level!

Started by rmoretti

rmoretti Staff Lv 1

You may remember that approximately a year ago we ran a series of puzzles which looked to find small molecules which bound to the KLHDC2 E3 ligase. Like VHL, (for which we've had success), molecules which bind KLHDC2 would be helpful in creating PROTACs, a new class of molecular regulators. But unlike VHL, there isn't a good druglike molecule which binds KLHDC2, just the naturally occurring peptide.

Collaborators at Boehringer Ingelheim have pored over the results of the first set of KLHDC2 puzzles. Due to logistics considerations, custom synthesis of the full designs wasn't feasible at the time. Additionally, the direct path from peptide to small molecule is a difficult and risky one. Instead, based on the first round results, medicinal chemists identified 17 core motifs which were common in the designed molecules, looked promising for further development, and which could be easily obtained. By confirming which compound cores which work well, we can have better confidence that derivative molecules would show activity.

BI experimentalists then used a technique called saturation transfer difference NMR, which is sensitive to even weak binding. They could confirm binding of a number of molecules, and could even confirm that they bound in the proper location by examining how the naturally occurring peptide competed with small molecule binding.

A listing of all the fragment molecule tested. The compounds with stars indicate those which showed binding activity.

With the knowledge of which compounds bound, we embarked on the sometimes difficult process of determining crystal structures of the bound complex. And we were successful! Crystals with compound 12 bound to the KLHDC2 protein were obtained, and they showed that binding happened in exactly the way we hoped.

Binding of the small molecule to the KLHDC2 protein. Left, the major interactions of the compound to the protein. Right, a comparison of the small molecule (in green) with the binding mode of the naturally occurring peptide (cyan).

Enter KLHDC2: the Next Level

Based on these exciting results, we're launching a new series of KLHDC2 puzzles. This time around, we want to start with the discovered small molecule. Keep the core structure and the existing hydrogen bonds, but build up the structure, making additional interactions with the protein.

Bletchley Park Lv 1

"Keep the core structure and the existing hydrogen bonds, but build up the structure, making additional interactions with the protein": Your discovered small molecule is not in the compound list, so anything we create based on it will likely not be in the compound list. How do you suggest we proceed ? Given that this is all hand-work and thus costs a lot of time I'd appreciate more guidance on what you expect. Script automation for the use of the compound list (upload, selection, mmf wiggle etc) would also be welcomed so we can write scripts to do the tedious work.

rosie4loop Lv 1

I think it'd be better if we can use substructure search with the compound library, instead of similarity search for keeping the fragment while expanding the structure. For example using Arthor instead of SmallWorld.

Also, I'd like to ask is there any additional filters or criteria when the compound library look for a similar compound? Or it's searching a custom database? I've tried to query compound 12 using both the publicly available Arthor and SmallWorld web interface to search the REAL database, then build the structure in foldit for a library search. However it's not in the library. Happens in previous rounds of puzzles too.

Although the substructure search of REAL database via Arthor doesn't give promising results using the default settings, at the first glimpse. Querying compound 12 with Arthor search gives many compounds extended via ester linkage at the carboxylate or cyclized in the 2nd position of carboxylate. Doesn't fit well in the pocket, and weren't found in compound library.

Puzzle 2310 is challenging particularly for the molecular weight limit of 300, typical for fragment-based drug design (design for later extension) instead of drug-like compounds.
Maybe easier for me to use pharmacophores to keep the hbond while putting the core elsewhere. I'm not good at designing drugs.

(edited for clarity, restoring contents of original post for record)

jeff101 Lv 1

I've been playing with the Zinc database today, as below:

I drew compound 12 from above in the box to the lower left at:
https://zinc20.docking.org/substances/home/
which told me it was O=C(O)Cc1ccoc1 in SMILES format.

After that, I think I clicked in the box listing the SMILES format to get:
https://zinc20.docking.org/substances/ZINC000000334130/
which looks like compound 12. Scrolling down gave me terms like
furan, carboxylate-sp3, & weak electrophile and links like below:
https://zinc20.docking.org/rings/furan/
https://zinc20.docking.org/patterns/carboxylate-sp3/
https://zinc20.docking.org/patterns/?type_name=weak-electrophile

Just doing https://zinc20.docking.org/patterns/ gives a long list
of symbols to help find compounds with different groups, for example,
https://zinc20.docking.org/substances?structure-contains=[OD1]-C(=O)[CX4]
shows many compounds containing sp3 carboxylates while
https://zinc20.docking.org/substances?structure-contains=C(=O)[OD1]
seems to find ones with more general types of carboxylates.
I think there are multiple pages of compounds for each of the above,
so don't just scroll to the bottom of what they show you and stop.
There is a way to get to page 2 and on. One compound I saw listed
(#331 or furaprofen) looks like it has the -COOH and an -O- placed
a bit like in our compound 12:
https://zinc20.docking.org/substances/ZINC000000000331/
Perhaps there are others.

There is a searchable wiki about the Zinc database too:
https://wiki.docking.org/index.php/Category:ZINC20

I hope this helps. If anyone finds a way to list compounds
containing both a furan ring and a -COOH group, please post
about it here. Also, if you can find a way to list compounds
with both an -O- and a -COOH group, please post about it
here.

Thanks, and good luck!

rosie4loop Lv 1

Maybe I'll provide more details on how I got these results in the previous post.

Although the substructure search on web doesn't give promising results using the default settings, at the first glimpse. Querying compound 12 with Arthor search gives many compounds extended via ester linkage at the carboxylate or cyclized in the 2nd position of carboxylate. Doesn't fit well in the pocket, and weren't found in compound library.

(Edit: was searching the REAL database. Other database may give better results.)

According to the devs discussion in this thread they're using SmallWorld search method to search the database host by ZINC. SmallWorld has the web interface of public database in the following link:
https://sw.docking.org
SmallWorld allows you to query different database and it's specialized in searching in ultra large database based on similarity.

However, for substructure search, the developers of ZINC suggest using the more accurate Arthor:
https://arthor.docking.org
It only search in smaller database, but easier to look for compounds with a certain core.

for the technical background behind SmallWorld and Arthor, see the original publication of ZINC 20 https://pubs.acs.org/doi/10.1021/acs.jcim.0c00675

AlphaFold2 Lv 1

Are we going to make spacepersons food for their trip to the moon and mars, i mean proteins are the name of the game

fdrehman Lv 1

Medicinal chemists can focus their efforts on modifying these core motifs to optimize their binding affinity and selectivity for KLHDC2. This step can increase the likelihood of finding small molecules that bind to KLHDC2 effectively, paving the way for developing PROTACs as molecular regulators.

It's also encouraging to know that Boehringer Ingelheim is collaborating on this project. Their expertise and resources in medicinal chemistry can be invaluable in advancing the research and development of potential KLHDC2 ligands.

As the project progresses, it will be fascinating to see how these core motifs are further developed and how they can be leveraged in the design of PROTACs for various therapeutic applications.