Structure-based compound design for dihydroorotate dehydrogenase - how to leverage protein-ligand interaction patterns for scaffold hopping idea generation
Have you already got familiar with the 3decision protein-ligand interaction search feature? This summer, we updated our software with a novel analytical tool that allows you to mine the entire 3decision protein structure repository for 3D complexes using interaction patterns. To understand how it works from the user perspective, I performed a small theoretic experiment with the aim of generating ideas for scaffold hopping using this feature.
As usual, I will describe the usecase step by step and share findings, so keep reading until the end of this article for interesting results coming from 3decision.
This feature complements the 3decision sub-pocket similarity search released last year in June. In case you missed it and would like to know more have a look at one of our previous webinars.
Introduction
Dihydroorotate dehydrogenase (DHODH) is a metabolic enzyme involved in the de novo synthesis pathway of pyrimidine (necessary for the production of nucleic acids). DHODH catalyzes the conversion of dihydroorotate to orotate in the fourth reaction of the 6-step pyridine biosynthetic pathway. In the human malaria parasite, Plasmodium falciparum, pyrimidine metabolism is essential for its survival. Therefore, targeting DHODH has been a clinically validated approach for anti-malarial drug development and is especially important since more efficient therapies are urgently needed to overcome the rapid increase of resistance to current drugs. (1)
Interesting fact: The human homolog of this enzyme has been associated with various types of malignancies. Its inhibition blocks the upregulated synthesis of pyrimidine-based nucleotides and decreases rapid cell proliferation, which affects many biological processes. Therefore, besides malaria, DHODH is an attractive target against human cancer. (2)
Here I focused on the structure of PfDHODH in complex with an active compound bearing a 4 – hydroxy – 1,2,5 - thiadiazol amide scaffold (PDB ID: 6I55). The compound is designed byLolli et al. and identified as the most potent and selective against PfDHODH compared to other compounds described in their publication.
For this experiment, I’ve used the latest 3decision interaction search feature to find additional compound ideas by replacing the thiadiazol amide scaffold.
The 3D structure of an active compound co-crystallized with PfDHODH (PDB ID: 6I55). The scaffold that I decided to replace is circled in red.
New ideas for scaffold hopping
The protein-ligand interaction-based search is a perfect analytical tool for generating ideas on how to modify the original scaffolds. It allows you to mine the entire PDB for protein-ligand complexes with similar interaction patterns. This way, you can quickly gather other scaffolds that bind the same/different protein target with a desired binding mode. The search is taking geometrical orientation into account and can be refined with a set of parameters, such as interaction type, length, and orientation tolerance, for a fuzzier match.
I will not dive into technical details here, but if you are curious to know how this feature works have a look at this article and don’t hesitate to ask questions using this form.
The first step in this experiment was to prepare the query by selecting interactions that the original scaffold - 4 – hydroxy – 1,2,5 - thiadiazol amide – forms with the target. I identified a total of 3 interactions that I would like to keep. Those are:
Hydrogen bond between the scaffold hydroxy group and the His185 side chain
Hydrogen bond between scaffold carbonyl and Arg265 side chain
Aromatic interaction between the thiadiazole ring and the Gly181 backbone carbonyl group
I then finetuned the query by modifying what 3decision defines as atom-aromatic interaction – the latest in the list above. By default, the settings indicated an aromatic interaction of the carbonyl group on the target side with any atom on the ligand side. To expand this search, I changed these settings by allowing the software to retrieve complexes with aromatic interaction between any group on the target side – not just carbonyl – and aromatic group on the ligand side - instead of any.
3D representation of amide scaffold in complex with PfDHODH (PDB ID: 6I55). Protein-ligand interaction for the 3decision search circled in red.
2D representation of interactions used as query generated by the thiadiazole amide scaffold in complex with PfDHODH (PDB ID: 6I55).
Some of these steps are presented in the 3decision interaction search How-to video. Press button below.
Next, I launched the search. After just a couple of minutes, 3decision returned 7 protein-ligand complex hits with matching interaction patterns. Each hit has a score (0-1, where 1 is the best) based on how well it matches the query. To analyze the results, I automatically superposed hit compounds to the query in one simple click and compared how the scaffolds and their interaction patterns overlap in the 3D Viewer.
It was interesting to see that the human mitochondrial branched-chain aminotransferase (BCATM) in complex with a triazolopyrimidinone compound (PDB ID: 5BWU) forms interactions that match very well my query (score 0.76). Namely:
Hydrogen bond of the NH of pyrimidine moiety with water that matches H-bond interaction of hydroxy group with His185 of the query
Hydrogen bond of carbonyl group of pyrimidine moiety with water molecules matches polar interaction of the carbonyl of the amide moiety with the side chain of Arg265
Aromatic interaction of triazolo moiety with Thr240 matches Atom-aromatic interaction of the thiadiazole moiety with carboxy group on Gly181
View fullsize
2D representation of query interactions of original compound with the PfDHODH (left) and matched interactions of hit ligand with BCATM (right).
3decision interaction-based search results: Comparison of the interaction patterns between the query and the hit scaffolds. Query interactions are presented in white; overlay is circled in red.
Inspired by the hit compound, I could guide the modification of my ligand towards closing the aromatic ring formed by the hydroxy and amide group and replace the thiadiazol ring with a triazole.
Et voilà! In a few simple steps, I have already got an idea of how to design a new compound by replacing the core scaffold.
New compound design
Without exiting the program, I tested the idea by first checking if the hit scaffold fits in the binding site of PfDHODH. Using a mesh surface representation of the pocket while compounds are still superposed, I confirmed that the scaffold on the hit ligand could be accommodated in the pocket of my target. Therefore, it makes sense to design a derivative with the novel core by “hopping” fragments as mentioned above. To sketch the derivative and re-dock it in my target I used the Ligand Docking and Design feature of 3decision.
If you would like to get familiar on how the 3decision docking feature works check out this 2-minutes How To video.
Derivative obtained from the original scaffold with novel core inspired by the hit compound coming from the 3decision interaction-based search.
The docking results in 3decision are visually represented with the protein-ligand contact coloring by statistics that facilitate the analysis of docked poses. (Quick reminder: green atom color indicates common protein-ligand contact in the database, whereas orange to red is less frequently present. For more info, check out the previous usecase). Below you can see the best pose of the docking that not only maintains the desired interaction but also fits well in the binding site. All of this was in favor of scaffold alteration initially proposed by 3decision.
The best docking pose of the derivative, query interactions are shown in white.
Conclusion
To highlight the new 3decision release, I have presented a theoretical usecase using the exclusive protein-ligand interaction-based search. In a couple of simple steps, I generated an idea on how to modify the core scaffold of the active compound in PfDHODH (PDB ID: 5I66) to potentially improve its drug-like properties. The next step would be to test the novel compound experimentally.
To sum up, the main strength of the new 3decision interaction-based analytic tool is its ability to mine a huge amount of 3D protein-ligand complexes for systems containing specific interaction patterns. It allows you to quickly and easily get out-of-the-box scaffolds without losing the desired binding modes by relying on real structural data. You can use our innovation to facilitate the ideation process of novel chemical compounds for various applications in early drug design projects: identify a new intellectual property, avoid toxicity, modify ADMET properties and improve drug selectivity.
Like this usecase? For more details and examples for this feature, have a look at our webinar “Protein-ligand interaction search: Discover 3decision’s latest feature and generate ideas for scaffold hopping”.
References:
Pippione AC, Sainas S, Goyal P, Fritzson I, Cassiano GC, Giraudo A, Giorgis M, Tavella TA, Bagnati R, Rolando B, Caing-Carlsson R, Costa FTM, Andrade CH, Al-Karadaghi S, Boschi D, Friemann R, Lolli ML. Hydroxyazole scaffold-based Plasmodium falciparum dihydroorotate dehydrogenase inhibitors: Synthesis, biological evaluation and X-ray structural studies. Eur J Med Chem. 2019 Feb 1;163:266-280. doi: 10.1016/j.ejmech.2018.11.044. Epub 2018 Nov 22. PMID: 30529545.
Zhou, Y., Tao, L., Zhou, X. et al. DHODH and cancer: promising prospects to be explored. Cancer Metab9, 22 (2021). https://doi.org/10.1186/s40170-021-00250-z