A non-canonical mechanism of GPCR activation

The goal of designing safer, more effective drugs has led to tremendous interest in molecular mechanisms through which ligands can precisely manipulate signaling of G-protein-coupled receptors (GPCRs), the largest class of drug targets. Decades of research have led to the widely accepted view that all agonists—ligands that trigger GPCR activation—function by causing rearrangement of the GPCR’s transmembrane helices, opening an intracellular pocket for binding of transducer proteins. Here we demonstrate that certain agonists instead trigger activation of free fatty acid receptor 1 by directly rearranging an intracellular loop that interacts with transducers. We validate the predictions of our atomic-level simulations by targeted mutagenesis; specific mutations which disrupt interactions with the intracellular loop convert these agonists into inverse agonists. Further analysis suggests that allosteric ligands could regulate signaling of many other GPCRs via a similar mechanism, offering rich possibilities for precise control of pharmaceutically important targets.

Supplementary Figure 1.AP8 has minimal effect on the conformation of key transmembrane helices in simulation.(a) Distributions of distances describing conformation of transmembrane helices under different simulation conditions: blue started from AP8-bound crystal structure (PDB 5TZY), red started from AP8-absent crystal structure (PDB 5TZR), and orange started from AP8-bound crystal structure with AP8 removed.Distances between transmembrane helices were measured using indicated pairs of Cɑ on the intracellular or extracellular end of each helix.Dotted vertical lines show distances in the initial crystal structures.Plots include simulation frames after 500 ns from 10 independent simulations for each condition.(b) As a positive control, we applied the same analysis to simulations with and without FFAR1 partial agonist MK-8666, which binds closer to "canonical" orthosteric site within the helical bundle.MK-8666 does have a substantial effect on key transmembrane helices which shifted by 2-6 angstroms.

Supplementary Figure 2. Differences in conformation of transmembrane helices in crystal structures may be a result of crystal packing, not ligand binding. (a)
The conformation of TM5 differs in the AP8-bound and AP8absent crystal structures, as pictured at left.The receptor is aligned to TM4 to show that TM5 is shifted 3.3.angstroms downward in the AP8-absent crystal structure.The intracellular ends of TM3, 5, and 6 are shifted up to 2.5 angstroms between the AP8-bound and AP8-absent crystal structures, as pictured at right.The receptor was aligned to TM1, 2 3, and 4. (b) The AP8-absent crystal structure has a different type of crystal packing, with neighboring subunits pressing TM5 (red) downward.The AP8-bound crystal structure lacks this contact.(c) In simulations started from the AP8-absent crystal structure (no AP8 present), TM5 quickly shifts upward to the same conformation as in the AP8-bound crystal structure.This suggests this shift is primarily due to the difference in crystal packing, rather than AP8 binding.Simulation trajectories show the TM5 shift for 5 independent simulations (see methods for details of the metric).Dashed line indicates the distance in the AP8-absent structure while the solid line is the distance in the AP8-bound crystal structure.

Supplementary Figure 3. No favorable changes in interactions between AP8 and TM residues, with and without G-protein. (a)
Comparison of cryo-EM structure of the FFAR1-Gq complex (PDB 8EIT, yellow) with the inactive, AP8-bound FFAR1 crystal structure (PDB 5TZY, purple).TM helices surrounding the AP8 binding site undergo little change upon receptor activation; the root mean square deviation of aligned residues shown in images is 0.65 Å (using residues 122-133, 94-106, and 190-201).(b) We used in-place docking (Schrodinger's Glide software) to calculate interaction metrics between AP8 and each residue in the binding site, using simulations frames.Two simulation conditions were used: the inactive WT receptor with AP8 bound (light grey) and the active, G-protein complex with AP8 bound (blue) based on our model.Note that our active-state model is very similar to the recently published cryo-EM structure around the AP8 binding site.We used 5 independent simulations per condition with 10 frames (every 100 ns) extracted from each simulation for analysis.The mean values for each metric and each residue were computed for the two conditions.The error bars show the 68% CI.
Supplementary Figure 4: AP8 modulates ICL2 angle in simulations.The ICL2 helix angle vs. time is shown for each independent simulation trial.The ICL2 angle was measured using the rotation of L112 and Y114 around the ICL2 helix axis for frames where ICL2 remained helical (see methods).All simulations were the same length, though helix unfolding could occur before that (trace ends early).For clarity only the first 1000 ns of the simulations are shown; the average helix unfolding time is 1050 ns with AP8-bound and 3500 ns with AP8-absent.Angles corresponding to the two stable states are labeled with dotted lines.The top line at 0 degrees is the PR state, the bottom line at -40 degrees is the NR state.Simulations of the receptor with AP8 bound (blue) favor the PR state while simulations without AP8 (orange) favor the NR state.All simulations were started from the crystal structure with AP8bound; AP8 was removed in orange traces.Thick traces are a 15 ns moving average.

Supplementary Figure 5. Quantifying folding free energy of ICL2 helix with adaptively biased MD (ABMD). (a)
Reaction coordinate chosen was the RMSD (root mean square deviation) of ICL2 backbone to the ICL2 helix.Using ABMD with this reaction coordinate led to more uniform sampling of the reaction coordinate and more frequent transitions between states as expected (right trace).This is in contrast to standard MD (left trace) which shows limited sampling of the conformational space.(b) The free energy landscape across the helix folding reaction coordinate was calculated with and without AP8 bound.The results are the average of 5 simulations for each condition.The shaded area shows the standard error of the mean.(c) The net folding free energy was not significantly different with and without AP8 bound.To get the net folding free energy difference, we first calculated the relative probability of the folded vs. unfolded states by integrating the free energy landscape from (b) using the partition function, and then transforming the relative probability to an energy difference.Error calculated from min and max curves of the energy landscape.

Supplementary Figure 8. ICL2 forms 25-33% of the interface between the GPCRs and G proteins. (top) M1
(pink) and G11 (cyan) cryo-EM structure is shown as a cartoon.Spheres show atoms on the receptor that contact the G protein (defined by an atom center-center distance of < 3 angstroms).The contact atoms are colored orange if they are part of ICL2.23% of contacts occur on ICL2.(bottom) MOR (pink) and Go (cyan) cryo-EM structure is shown as a cartoon.31% of contacts occur on ICL2.