Cresset Discovery provides molecular modeling expertise utilizing the application of proprietary in silico methods which, when combined with in vitro assays up to in vivo assays, drives efficient hit finding and optimization workflows for drug discovery. A recent paper¹ describes such a workflow in collaboration with Epics Therapeutics. In this article, we summarize Cresset Discovery’s analysis of protein-ligand interactions, generation of new designs through fragment replacement and optimization, and building of predictive models to prioritize each iteration, accelerating the development cycle.

Abstract

The RNA methyltransferase enzyme METTL3 processes mRNA by writing the epigenetic marker N⁶-methyladenosine (m⁶A), which is linked to the development of acute myeloid leukemia (AML) and various solid tumor cancers, including Ovarian (SK-ov-3), head and neck (FaDu) and lung cancer (A549, NSCLC). This paper describes the discovery and optimization process that resulted in the selection of compound EP652 as an in vivo proof-of-concept molecule, developed for its inhibitory action against METTL3. Systematic tuning of the PK properties allowed the demonstration of efficacy of EP652 in preclinical oncology models, via IP dosing. Thus, the pharmacological suppression of this (m⁶A) epigenetic marker was confirmed as a valid approach towards new cancer therapeutics.

Introduction

The methyltransferase-like 3 (METTL3) enzyme is of interest as a primary modifier of mRNA sequences and specifically a writer of m6A epigenetic markers that have been linked to the fate of oncology-related modulatory proteins and ultimately to the development of tumors. This enzyme acts as part of a heterodimer wherein METTL3 is the catalytic domain and another component of the heterodimer, methyltransferase-like 14 (METTL14), serves as a substrate recognition element for RNA binding. This enzyme, like many other methyltransferase enzymes, operates by facilitating the simultaneous binding of both the co-factor S-adenosyl methionine (SAM, a synthon for ‘Me+’) and substrates. Once a suitable geometry for the delivery of a methyl group is achieved, S-adenosyl homocysteine (SAH) and methylated substrate, in this case an m⁶A labelled RNA sequence, can be liberated. This protein target is clearly druggable, as demonstrated by the natural product small molecule inhibitor sinefungin, a non-selective inhibitor of METTL3, that nonetheless sets a precedence for the potential development of small molecules that might target this methyltransferase.

Hit identification and optimization approach

Our initial view after processing the available X-ray data for METTL3 was that in silico SAM mimicry alone might be challenging for providing target selectivity thereby necessitating a combined assessment of both SAM and substrate mimicry, the challenge then being navigating inherent loop flexibility in the protein. In addition to this design approach, both HTS screening and novel virtual screening efforts were pursued. Unfortunately, neither of the latter approaches provided anything useful as starting points for drug development, for various reasons not discussed further here.
We returned to our initial idea after the first (non-nucleoside) inhibitors of METTL3 (Figure 1) were reported by Caflisch², and latterly further publications^3-5 and reports and patents from Storm Therapeutics^6-8 and reviews⁹, which demonstrated the utility of our original thinking. Deviation from a pure SAM competition inhibitory mode is evident in that the core region (cyan in Figure 1) overlaps with substrate. Ultimately, early data and X-ray crystallography of such inhibitors revealed critical ‘adapted loop configurations’ that allowed interpretation of the geometrical requirements of the whole current inhibitory repertoire, facilitating the adoption of an in silico ligand design approach.

As detailed in Figure 2, we analyzed the SAR of known inhibitors and their crystallography. We conducted protein alignment and ligand alignment protocols using Cresset’s proprietary software Flare™,¹¹ to build predictive QSAR models and used this as part of a prioritization vehicle for ligand designs. Ligand ideas from structure-based design and ligand-based design efforts were guided by in silico scoring metrics such as docking score¹² and Electrostatic Complementarity™.¹³ Fragment replacement, with Cresset’s proprietary tool Spark™,¹⁴ provided alternative scaffolds. All ideas were checked via in silico physiochemical property profiling as a multiparameter optimization to define the most appropriate designs. Finally, chemical route scouting assessments, defining synthetic expediency, was conducted by the medicinal chemistry team as the final arbiter of selection that preceded each round of “Design-Make-Test-Analyze” (DMTA) cycles.

Progression of the chemical hit series towards the required target cellular (and latterly, in vivo) activity profile was an exercise in systematic chemical exploration and multiparameter optimization, balancing cellular activity (i.e. intracellular uptake) and competing factors such as clearance and metabolic stability. For this, the Kasumi-1 cellular readout combined with Log P/logD as lipophilic ligand efficiencies (LLE) was a very useful optimization metric.

Results and discussion

Applying the combined in-silico design/medicinal chemistry workflow provided eight diverse chemical structures that were progressed to synthesis and testing. From this initial set of compounds, all achieved measurable activity in the enzyme SPA assay in the 2nM to 2.4µM range. The most promising of these was compound 1, at 2 nM, for the preferred R enantiomer. Compound 1 was subsequently profiled for activity and PK properties and ultimately submitted for X-ray crystallography as shown in Figure 3.

This initial hit, whilst potent in the SPA enzymatic assay and moderate in ATPlite cellular read-out, suffered rapid metabolism in rat liver microsomes. Also, as a potent inhibitor of p450 enzymes (CYP3A4, 0.8 µM), it was unlikely to progress without further modification. The ensuing medicinal chemistry effort (detailed in the full paper) culminated in the discovery of an optimized candidate: EP652, that demonstrated activity following intraperitoneal dosing in mouse models of AML and selected solid tumors.

Adherence to the 3D shape and electronic signature of the pre-existing chemotypes (illustrated by Figure 4) shows the utility of building from this foundation to explore new avenues through in silico design combined with the iterative results stemming from the medicinal chemistry effort to discover and prioritize new chemotypes in lead optimization.

Summary

We have demonstrated that the use of a highly efficient in silico workflow, that incorporates medicinal chemistry expertise, was pivotal in hit finding and optimization, ultimately enabling rapid progression towards a drug development candidate.

For more details, please refer to the original publication¹: J. Med. Chem. 2025, 68, 3, 2981–3003.

References

1. Dutheuil, G., Oukoloff, K., Korac, J., Lenoir, F., El Bousmaqui, M., Probst, N., Lapin, A., Nakhabina, G., Sorlet, C., Parmentier,N., Karila, D., Ghavtadze, N., Casault, P., Claridge, S., Sapmaz, S., Slater, M.J., and Fraser, G.L. Discovery, Optimization, and Preclinical Pharmacology of EP652, a METTL3 Inhibitor with Efficacy in Liquid and Solid Tumor Models, J Med Chem. 2025, Vol 68, Issue 3, 2981-3003 https://pubs.acs.org/doi/10.1021/acs.jmedchem.4c02225
2. Moroz-Omori, E. V., Huang, D., Bedi, R.K., Cheriyamkunnel, S. J., Bochenkova, E., Dolbois, A., Rzeczkowski, M.D., Li, Y., Wiedmer, L., Caflisch, A. METTL3 inhibitors for epitranscriptomic modulation of cellular processes. ChemMedChem. 2021, 16, 3035-3043.
3. Bedi, R.K., Huang, D., Eberle, S.A., Wiedmer, L., Śledź, P., Caflisch, A. Small-molecule inhibitors of METTL3, the major human epitranscriptomic writer. ChemMedChem 2020, 15 (9), 744-748.
4. Bedi, R.K., Huang, D., Li, Y., Caflisch, A. Structure-based design of inhibitors of the m6A-RNA writer enzyme METTL3. ACS Bio. Med. Chem. Au 2023, 3 (4), 359–370.
5. Dolbois, A., Bedi, R.K., Bochenkova, E., Müller, A., Moroz-Omori, E.V., Huang, D., Caflisch, A. 1,4,9-Triazaspiro[5.5]undecan-2-one derivatives as potent and selective METTL3 inhibitors. J. Med. Chem. 2021, 64 (17), 12738-12760; Caflisch, A., Dolbois, A., Huang, D., Zalesak F. N6-adenosine-methyltransferase inhibitors in cancer treatment. PCT International Application WO 2022/243333 A1, May 17, 2022; Bedi, R.K., Huang, D., Li, Y., Caflisch, A. Structure-based design of inhibitors of the m6A-RNA writer enzyme METTL3. ACS Bio. & Med. Chem. Au 2023, 3 (4), 359-370.
6. Yankova, E., Blackaby, W., Albertella, M., Rak, J., De Braekeleer, E., Tsagkogeorga, E., Pilka, E.S., Aspris, D., Leggate, D., Hendrick, A.G., Webster, N.A., Andrews, B., Fosbeary, R., Guest, P., Irigoyen, N., Eleftheriou, M., Gozdecka, M., Dias, J.M.L., Bannister, A.J., Vick, B., Jeremias, I., Vassiliou, G.S., Rausch, O., Tzelepis, K., Kouzarides, K. Small molecule inhibition of METTL3 as a strategy against myeloid leukemia. Nature 2021, 593, 597−601.
7. Guirguis, A.A., Ofir-Rosenfeld, Y., Knezevic, K., Blackaby, W., Hardick, D., Chan, Y., Motazedian, A., Gillespie, A., Vassiliadis, D., Lam, E.E.N., Tran, K., Andrews, B., Harbour, M.E., Vasiliauskaite, L., Saunders, C.J., Tsagkogeorga, G., Azevedo, A., Obacz, J., Pilka, E.S., Carkill, M., MacPherson, L., Wainwright, E.N., Liddicoat, B., Blyth, B.J., Albertella, M.R., Rausch, O., Dawson, M.A. Inhibition of METTL3 results in a cell-intrinsic interferon response that enhances antitumor immunity. Cancer Discov. 2023, 13 (10), 2228–2247.
8. Blackaby, W., Hardick, D., Thomas, E., Bookfield, F.A., Bubert, C., Shepherd, J., Ridgill, M.P. Preparation of imidazopyridines and related derivatives as MeTTL3 inhibitors. WO 2020/201773 A1, Oct 08, 2020; Blackaby, W., Hardick, D., Thomas, E., Bookfield, F.A., Shepherd, J., Bubert, C., Ridgill, M.P. Preparation of polyheterocyclic compounds as METTL3 inhibitors. WO 2021/111124 A1, Jun 10, 2021; Hardick, D., Blackaby, W., Thomas, E., Bookfield, F.A., Shepherd, J., Bubert, C., Ridgill, M.P. Preparation of heteroaromatic compounds as METTL3 inhibitory compounds. WO 2022/074379 A1, Apr 14, 2022; Hardick, D., Blackaby, W., Thomas, E., Bookfield, F.A., Shepherd, J., Bubert, C., Ridgill, M.P. Preparation of nucleosides as inhibitors of N6-adenosine-methyltransferase METTL3. WO 2022/074391 A1, Apr 14, 2022; Bubert, C., Ridgill, M.P., Canelli, T., Hardick, D. Processes for the preparation of inhibitory compounds. WO 2022/254218 A1, Dec 08, 2022; Bayle, E.D., Bucknell, S.J., Thomas, E.J.Hardick, D.J., Harper C., Byrne, D., Shepherd, J., Ridgill, M.P.Inhibitory compounds. WO 2024/126897 A1, Dec 12 2024.
9. Fiorentino, F., Menna, M., Rotili, D., Valente, S., Mai, A. METTL3 from Target Validation to the First Small-Molecule Inhibitors: A Medicinal Chemistry Journey, J. Med. Chem. 2023, 66 (3), 1654-1677; Li, H., Zhang, Q., Feng, Q., You, Q., Guo, X. The development of small molecules targeting methyltransferase-like 3, Drug Discovery Today 2023, 28 (4), 103513.
10. https://www.stormtherapeutics.com/site/assets/files/1460/storm-asco-poster-72×42-may24-v5b_final.3w27wb.pdf
11. Flare™, version 10, Cresset®, Litlington, Cambridgeshire, UK; https://cresset-group.com/flare/; Cheeseright T.; Mackey, M.; Rose, S.; Vinter, A. Molecular Field Extrema as Descriptors of Biological Activity: Definition and Validation. J. Chem. Inf. Model. 2006, 46, 665-676.
12. Lead Finder™, BioMolTech®, Toronto, Ontario, Canada; https://cresset-group.com/software/leadfinder.
13. Bauer, M. R.; Mackey, M. D. Electrostatic Complementarity as a Fast and Effective Tool to Optimize Binding and Selectivity of Protein-Ligand Complexes. J. Med. Chem. 2019, 62, 3036-3050.
14. Spark™, Cresset®, Litlington, Cambridgeshire, UK; https://cresset-group.com/software/spark/; Cheeseright, T., Mackey, M., Rose, S., Vinter, A. Molecular field extrema as descriptors of biological activity: definition and validation. J. Chem. Inf. Model. 2006, 46 (2), 665-676.