Back

Danuglipron and orforglipron engage GLP-1R through materially different interfaces: an unbiased molecular dynamics study

Oztekin, A.; Turker, S.

2026-10-06 biophysics
10.64898/2026.09.30.754652 bioRxiv
Show abstract

GLP-1R is among the most heavily pursued targets in metabolic disease, but its natural agonist is a 30-residue peptide that must be injected. Two oral non-peptide agonists have reached late-stage trials. Cryo-EM studies have already shown that danuglipron and CHU-128, a close analogue of orforglipron, occupy the receptor differently, but not how either interface behaves over time. We ran 500 ns of unbiased all-atom molecular dynamics on GLP-1R bound to GLP-1, danuglipron (PF-06882961) and orforglipron (LY3502970), one trajectory each, in a POPC bilayer with the G protein removed. Although CHU-128 has been simulated before, its analogue, orforglipron, has not. Scored against contacts computed from the deposited structures, under a force field that shares no parameters with previous oral agonist studies, we recovered 80%, 86% and 79% above 50% occupancy. Indexing contact by chemical group, rather than by receptor residue, and normalising by molecular size, separates the two oral compounds: orforglipron carries ring systems dedicated almost entirely to one receptor domain, one of them directing 99.7% of its contact to the extracellular domain, whereas every group of danuglipron engages both. Danuglipron is correspondingly the outlier on measures that do not depend on a chosen threshold: only 33% of its contacts exceed 90% occupancy against 72% and 84%, and it is the only system whose occupancies remain unconverged at 100 ns. With single trajectories, these are observations rather than tested claims, but two drugs of the same class and comparable size occupy this receptor in structurally distinct ways.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.