Ureidopeptide GLP-1 analogues with prolonged activity in vivo via signal bias and altered receptor trafficking

This study demonstrates the efficacy of single α-amino acids substitution with ureido residues to design long lasting peptides.


Statistical analysis.
GraphPad Prism 7.0 and 8.0 were used for all analyses. Replicate measurements were taken from different samples. Statistical analyses (two-way t-test, two-way ANOVA and Bonferroni post-test, one-way ANOVA with Dunnett's multiple comparison test, two-way ANOVA with Tukey's test, oneway ANOVA with Sidak's multiple comparison test) were applied when indicated. P values lower than 0.05 were considered significant.

Synthesis of oligomers 2-14 and S1-S12
General Procedure P1 Oligomers 2-14 and S1-S12 were synthesized using the following general procedure as previously described: [1] Sieber resin (≈ 160mg, loading 0.62mmol/g) was swelled in DMF (3 mL) for 30 min. All steps were performed under microwave irradiation. The synthesis were conducted with microwave irradiation using the Liberty Blue TM microwave peptide synthesizer from CEM.

A1: Fmoc deprotection
The N-Fmoc protecting group was removed with 20% piperidine in DMF (3 ml) with the standard liberty blue methods. [2] A2: Coupling of Fmoc-amino acid N-Fmoc-α amino acid (5 eq. relative to the resin loading) were coupled with PyBOP or DIC/Oxyma (5 eq. relative to the resin loading) and DIEA (10 eq. relative to the resin loading) as coupling reagent using the standard liberty blue methods. [2] A3: Coupling of activated N 3 -building bloc Each activated monomer (3 eq. relative to the resin loading) was coupled twice using DIEA (10 eq. relative to the resin loading) under microwave irradiation (70°C, 50W, 20 min) in DMF (4 mL).
A5: Cleavage from the resin After completion of the synthesis, the resin was transferred into a syringe with a frit, and washed three times with DMF, three times with CH 2 Cl 2 and three times with Et 2 O. Cleavage from the resin was performed using 95% TFA with 2.5% triisopropylsilane and 2.5% water (3 mL). After 2h the resin was filtered and discarded. Diethyl ether was added to precipitate the oligomer and the solid was triturated and filtrated.
Purification and characterization: Analytical RP-HPLC analyses were performed on a Dionex U3000SD using a Macherey-Nagel Nucleodur C18ec column (4 x 100 mm, 3 µm) at a flow rate of 1 mL/min with UV detection at 200 nm. The mobile phase was composed of 0.1% (v/v) TFA-H 2 O (Solvent A) and 0.1% (v/v) TFA-CH 3 CN (solvent B).

S7
Semi preparative purification of all compound was performed on a Dionex U3000SD using a Macherey-Nagel Nucleodur C18ec column (10 x 250 mm, 5 µm) at a flow rate of 4 mL/min with UV detection at 200 nm. The mobile phase was composed of 0.1% (v/v) TFA-H 2 O (Solvent A) and 0.1% (v/v) TFA-CH 3 CN (solvent B).
LC-MS analyses were carried out on a UHPLC (Agilent 1290 Infinity) coupled to a ESI -MS Tof (Agilent 6230 ESI).
Electrospray ionization mass spectrometry (ESI-MS) experiments were performed on an Agilent 6560 DTIMS-Q-TOF spectrometer (Agilent Technologies, Santa Clara, CA), with the dual-ESI source operated in positive ion mode.
The oligomer was synthesized using procedure P1. Then the resin was transferred in a 10 mL syringe, 5 mL of DCM was added and the Alloc group was removed using Pd(Ph 3 ) 4 (30 mg, 0.25 equiv relative to the resin loading) and phenylsilane (135 µL, 1.1 equiv relative to the resin loading) at room temperature for 45 min. After filtration and washes with DMF (2x) and DCM (3x), DCM (5 mL), N-Fmoc-Glu-OtBu (222 mg, 5 equiv relative to the resin loading), PyBop (260 mg, 5 equiv relative to the resin loading), and DIEA (93 µL, 5 equiv relative to the resin loading) were loaded on the resin and it was shaken for 2 hours at room temperature. Fmoc group was removed with piperidine in DMF (20%), 2 times 20 min. The resin was washed with DMF (2x) and DCM (3x), then palmitoyl chloride (93 µL), and DIEA (260 µL) were loaded on the resin and it was shaken for 2 hours at room temperature. The cleavage, the purification and the characterization were the same as procedure P1.

Cyclic AMP measurements
In vitro pharmacology: Affinity (IC 50 ) The affinity of oligomers at the mouse GLP-1 receptor endogenously expressed in βTC6 cells was determined in a radioligand binding assay (performed by Cerep S.A., catalog 2015, ref. 0228) as previously described. [4] Cell membrane homogenates (20 µg protein) are incubated for 120 min at 37°C with 0.025 nM [ 125 I]GLP-1  in the absence or presence of the test compound in a buffer containing 50 mM Hepes-NaOH (pH 7.4), 120 mM NaCl, 5 mM KCl, 1.2 mM MgSO 4 , 1 mM EDTA, 0.025% bacitracin and 1% BSA. Nonspecific binding is determined in the presence of 1 µM GLP-1(7-36). Following incubation, the samples are filtered rapidly under vacuum through glass fiber filters (GF/B, Packard) presoaked with 0.3% PEI and rinsed several times with an ice-cold buffer containing 50 mM Tris-HCl and 500 mM NaCl using a 96-sample cell harvester (Unifilter, Packard). The filters are dried then counted for radioactivity in a scintillation counter (Topcount, Packard) using a scintillation cocktail (Microscint 0, Packard). The results are expressed as a percent inhibition of the control radioligand specific binding. The standard reference compound is exenatide (9), which is tested in each experiment at several concentrations to obtain a competition curve from which its IC 50 is calculated using Graphpad Prism.
In vitro pharmacology on HEK-293T cells (EC 50 : Method C1) The agonist activity of compounds at the human GLP-1 receptor exogenously expressed in HEK-293T cells (Multispan inc., lot#.DC1267-062017) was determined by measuring their effects on cAMP production using the HTRF detection method (performed by UREkA SARL). The cells were distributed in 384-well microplates at a density of 1.0x10 4 cells/well. Stock solutions of the test compounds or the reference agonist were prepared at a concentration of 1 mM in DMSO. Then, compounds to be tested were diluted in cell culture media and transferred to the plate containing the cells to reach final assay concentrations in the range 10 -8 to 10 -15 M. The cells were incubated for 10 min at 5% CO 2 at 37°C. The reaction was then stopped by addition of fluorescence donor (anti-cAMP antibody labeled with europium cryptate) and the fluorescence acceptor (D2-labeled cAMP) mixed in lysis buffer (cAMP Gs-Dynamic kit, Cisbio). After 120 min at room temperature, a microplate reader (F500 Tecan) was used to measure the fluorescence transfer at λex = 337 nm and λem = 620 nm and 665 nm. The ratio of the signal measured at 665 nm on signal measured at 620 nm is used to determine the cAMP concentration. The results are given as a percent of the control response to 10 nM Forskolin. The standard reference agonist is GLP-1-G2-NH2 (peptide 1), which is tested in each experiment at several concentrations to generate a concentration response curve from which its EC50 value is calculated using GraphPad Prism.

In vitro pharmacology on immortalised beta cells (EC 50 : Method C2)
The agonist activity of compounds at the mouse GLP-1 receptor endogenously expressed in βTC6 cells was determined by measuring their effects on cAMP production using the HTRF detection method (performed by Cerep S.A., catalog 2015, ref. 2181) as previously described. [1,3] The cells were distributed in microplates at a density of 1.5x10 4 cells/well and incubated for 10 min at room temperature in the presence of HBSS (basal control), the test compound or the reference agonist. Following incubation, the cells were lysed and the fluorescence acceptor (D2-labeled cAMP) and fluorescence donor (anti-cAMP antibody labeled with europium cryptate) are added. After 60 min at room temperature, the fluorescence transfer is measured at λex= 337 nm and λem= 620 nm and 665 nm using a microplate reader (Rubystar, BMG). The cAMP concentration is determined by dividing the signal measured at 665 nm by that measured at 620 nm (ratio). The results are expressed as a percent of the control response to 10 nM GLP-1(7-37). The standard reference agonist is GLP-1(7-37), which is tested in each experiment at several concentrations to generate a concentration-response curve from which its EC50 value and SE are calculated using GraphPad Prism.
In vitro pharmacology on immortalised CHO-GLP-1R cells (EC 50 : Method C3) PathHunter CHO-GLP-1R cells were stimulated for 30 min at 37°C with indicated concentration of agonist in serum free medium without phosphodiesterase inhibitors. The reaction was terminated by addition of cAMP detection reagents (cAMP Dynamic 2, Cisbio) and read by HTRF.

β-arrestin-2 recruitment assay
PathHunter CHO-GLP-1R cells were stimulated for 30 min at 37°C with indicated concentration of agonist in serum free medium. The reaction was terminated by addition of PathHunter detection reagents (DiscoverX) and luminescent signal recorded.

Calculation of signal bias
Biased signalling was quantified using a modified form of the operational model. [6] LogR values were calculated for each ligand in each assay, and ∆LogR were determined by subtracting from the corresponding values for URK-062. Bias was then determined as ∆∆LogR by subtracting ∆LogR for cAMP from β-arrestin-2. As all ligands were run in parallel in both pathways, calculations were done on a per-assay basis 3 . Bias was compared statistically by determining the difference between LogR for cAMP and β-arrestin-2 for each ligand and performing a randomised-block 1-way ANOVA on this value.

Insulin secretion in vitro
In vitro insulin secretion was performed as previously described. [7] INS-1 832/3 cells were primed at 3 mM glucose overnight before the assay. After washing, cells were added in suspension to the normal growth medium at 11 mM glucose, with or without agonist. In 96 well plates. After overnight incubation, a sample of supernatant was removed and analysed for insulin by HTRF (Insulin Hi Range, Cisbio).

Surface SNAP-GLP-1R labelling and internalization
SNAP-GLP-1R-expressing INS-1 832/3 cells were labeled at 37°C with 1 μM of SNAP-Surface 549 fluorescent probe (New England Biolabs) in full media prior to being stimulated with 10 nM ligands for 30 minutes, fixed in 4% paraformaldehyde, mounted in Prolong Diamond antifade reagent with 4,6diamidino-2-phenylindole (Life Technologies), and imaged with a Zeiss LSM-780 inverted confocal laser-scanning microscope in a 63x/1.4 numerical aperture oil-immersion objective from the Facility for Imaging by Light Microscopy (FILM) at Imperial College London, and analyzed in Fiji.

Mouse plasma stability (Method MP1)
Stock solutions of the oligomers were prepared at a concentration of 250 µM in mQ water. The oligomer was then diluted 1/50 with a solution of plasma/PBS pH 7.4 (1/1) to afford a final concentration of 5µM and incubated at 37°C. (4µL of the stock solution was diluted with 196 µL of plasma/PBS, pH 7.4, 1/1) Each compound was also incubated in the absence of plasma (196 µL of H2O/PBS, pH 7.4, 1/1)). At the indicated time (20min and 60min) an aliquot of 70 µL was removed from each experimental reaction and pipetting into 175 µL of Acetonitrile at 0°C to quench the reaction. (t = 0 min was determined using the reaction without plasma). The samples were frozen at -80°C before analysis. The frozen sample were defrosted, stirred with a vortex 5 min and finally centrifuged 5 min at 16°C. The supernatant was analyzed by LC-MS. The time course of oligomer degradation was determined by integrating the area of the peak in the extracted ion chromatogram.

Mouse plasma stability (Method MP2)
Stock solutions of the oligomers were prepared at a concentration of 250 µM in DMSO. The oligomer was then diluted 1/50 with a solution of plasma/PBS pH 7.4 (1/1) to afford a final concentration of 5µM and incubated at 37°C. (4µL of the stock solution was diluted with 196 µL of plasma/PBS, pH 7.4, 1/1) Each compound was also incubated in the absence of plasma (196 µL of H2O/PBS, pH 7.4, 1/1)). At the indicated time (20min and 60min) an aliquot of 70 µL was removed from each experimental reaction and pipetting into 175 µL of Acetonitrile at 0°C to quench the reaction. (t = 0 min was determined using the reaction without plasma). The samples were frozen at -80°C before analysis. The frozen sample were defrosted, stirred with a vortex 5 min and finally centrifuged 5 min at 16°C. The supernatant was analyzed by LC-MS. The time course of oligomer degradation was determined by integrating the area of the peak in the extracted ion chromatogram.

Mouse plasma stability (Method MP3)
Stock solutions of the oligomers were prepared at a concentration of 250 µM in DMSO. The oligomer was then diluted 1/50 with a solution of plasma/PBS pH 7.4 (1/1) to afford a final concentration of 5µM and incubated at 37°C. (4µL of the stock solution was diluted with 196 µL of plasma/PBS, pH 7.4, 1/1) Each compound was also incubated in the absence of plasma (196 µL of H2O/PBS, pH 7.4, 1/1)). At the indicated time (20min and 60min) an aliquot of 70 µL was removed from each experimental reaction and pipetting into 10 µL of HCL 12N solution. The remaining mixture was then pipetting into 165 µL of Acetonitrile at 0°C to quench the reaction. (t = 0 min was determined using the reaction without plasma). The samples were frozen at -80°C before analysis. The frozen sample were defrosted, stirred with a vortex 5 min and finally centrifuged 5 min at 16°C. The supernatant was analyzed by LC-MS. The time course of oligomer degradation was determined by integrating the area of the peak in the extracted ion chromatogram.

Animals.
For the pharmacodynamics studies performed by Physiogenex S.A.S., mice were housed in ventilated and enriched housing cages (310 x 125 x 127 mm³) throughout the experimental phase. The mice were housed in groups of 3 animals during the study, on a normal 12 hours light cycle (at 8:00 pm lights off), 22 ± 2 °C and 50 ± 10 % relative humidity. A standard chow diet (RM1 (E) 801492, SDS) and tap water were provided ad libitum. All animal protocols done by Physiogenex S.A.S were reviewed and approved by the local (Comité régional d'éthique de Midi-Pyrénées) and national (Ministère de l'Enseignement Supérieur et de la Recherche) ethics committees (protocol number 05049-06). For the dose-response study, animal procedures were approved by British Home Office under the UK Animal (Scientific Procedures) Act 1986 (Project Licence 70/7596). Lean male C57Bl/6J mice (6-8 weeks of age, body weight 25-30 g, obtained from Charles River) were maintained at 21-23°C and light-dark cycles (12:12h schedule, lights on at 07:00). Ad libitum access to water and normal chow (RM1, Special Diet Services) was provided unless otherwise stated. Mice were housed in groups of five. For the pharmacokinetics studies performed by TechMedILL service (PCBIS platform, CNRS UMS3286), mice were housed in polycarbonate cages (PCT2L12SHT, Allentown) enriched with play tunnels throughout the experimental phase. The mice were housed in groups of 9 animals during the study, under controlled environment (22 ± 1 °C) with a relative humidity (50 ± 10 %) and a normal 12 hours light cycle (at 8:00 pm lights off). A standard chow diet (A04, SAFE, France) and tap water were provided ad libitum. All animal protocols done by TechMedILL service were reviewed and approved by the agriculture ministry regulating animal research in France (Ethics regional committee for animal experimentation Strasbourg, APAFIS 1341#2015080309399690).

IPGTT experiments in healthy mice (IPGTT 3 h and 6 h).
After the acclimation period of 5 days, 20-25 g male C57Bl/6J mice 8 weeks old (Charles River laboratories) were randomized into groups (6 mice per group) according to their body weight. The mice were acutely treated with vehicle or 1 µg per mouse (10 nmol kg -1 ) of oligomers via i.v. injections (formulation: 4 µg mL -1 in PBS 1×) IPGTTs (glucose 2 g kg -1 i.p.) were performed after 3 or 6 hours of dosing with a fasting period of 6 hours. Blood glucose was measured before the administration of glucose and after at 30, 60, 90 and 120 min.

Fasted Blood Glucose and IPGTT experiments in healthy mice (IPGTT 9 h).
After the acclimation period of 5 days, 20-25 g male C57Bl/6J mice 8 weeks old (Charles River laboratories) were randomized into groups (6 mice per group) according to their body weight. The mice were acutely treated with vehicle or 1 µg per mouse (10 nmol kg -1 ) of oligomers via i.v. injections (formulation: 4 µg mL -1 in PBS 1×). The mice were fasted and blood glucose was measure before dosing (T0) and after 2, 4, 6, 8 and 9 hours. An IPGTT (glucose 2 g kg -1 i.p.) was performed after 9 hours of dosing with a fasting period of 9 hours. Blood glucose was measured before the administration of glucose and after 30, 60 and 90 min.

Fed Blood Glucose experiments in healthy mice (30 h).
After the acclimation period of 5 days, 20-25 g male C57Bl/6J mice 8 weeks old (Charles River laboratories) were randomized into groups (6 mice per group) according to their body weight. The mice were acutely treated with vehicle or 1 µg per mouse (10 nmol kg -1 ) of oligomers via i.v. injections (formulation: 4 µg mL -1 in PBS 1×) Fed blood glucose was measure before dosing (T0) and after 3, 6, 18, 24, 27 and 30 hours. An IPGTT (glucose 2 g kg -1 i.p.) was performed after 9 hours of dosing with a fasting period of 9 hours. Blood glucose was measured before the administration of glucose and after 30, 60 and 90 min.

Dose-response study in healthy mice.
Mice were lightly fasted (4 hours) before IP administration of varying doses of each agonist or vehicle (100 µL saline). After a further 6 hours, blood glucose was measured (GlucoRx) from the tail followed by IP administration of 2 g/kg 20% glucose. Blood glucose was monitored at 20, 40 and 60 minute timepoints. To derive the ED 50 (dose required to give a 50% maximal response), 3-parameter fitting was performed on the area-under-curve for each set of data.

Study in db/db mice (15 days).
After the acclimation period of 5 days, male db/db mice 8 weeks old (Charles River laboratories) were randomized into groups (n=10) according to their body weight, HOMA-IR and HbA1C measured after an overnight fast and 8 non diabetic mice were excluded. Mice were chronically treated during 15 days via s.c. route (once daily) at around 8AM with the vehicle or with 100 µg kg -1 (25 nmol kg -1 ) of the oligomers. Formulation: 20 µg mL -1 in PBS 1×. On the 7 th day of treatment, fed blood glucose was measured at 1, 2, 3, 6 and 24 h after dosing. On the 12 th day of treatment, mice were fasted 6 hours and an OGTT (glucose 1 g kg -1 p.o.) was performed 6 hours (2 PM) after the dosing. Blood glucose was measure 30 and 0 min before the glucose injection and after 15, 30, 60, 90, and 120 min. Plasma insulin was measure 30 min before the glucose injection and 15 min after. On the 15 th day of treatment, mice were fasted 6 hours and an ITT (insulin 2 U kg -1 i.p.) was performed 6 hours (2 PM) after the dosing. Blood glucose was measure before insulin injection and after 15, 30, 60, 90, and 120 min. Blood samples were collected for measuring HbA1c just before the start of the treatments and at the end of the treatment period.

Pharmacokinetics.
Fifteen mice (male C57BL/6J mice (Janvier Labs, France) 9 weeks old, 20-25 g) were treated with GLP-1 analogues via i.v. injections (2 mg kg -1 ) formulated at 1 mg mL -1 in PBS 1×. After different time points mice were sacrificed and blood sample were collected. The plasma was separated by centrifugation and the samples were frozen at -80°C before analysis. A volume of 400 µL of each sample of plasma was mixed with 1 ml of acetonitrile to precipitate the protein and extract the compound. The sample were then vortexed and centrifuged (15 000 × g, 5 min, 16 °C) to sediment the precipitated protein.
The supernatant was analysed by LC-MS/MS using a UHPLC coupled to LC-MS 8030 Shimadzu triple quadrupole.

Molecular dynamics simulation of GLP-1R in complex with ureidopeptide 3
A three dimensional model of human GLP-1 receptor (GLP-1R) was obtained from the cryo-EM structure of rabbit GLP1-R in complex with human GLP-1 (PDB ID 5VAI) [8] . Human GLP-1R which shares 92% sequence identity with the rabbit template was directly obtained by mutating corresponding residues in SYBYL X2.1.1 (Cetera Inc, Princeton, NJ). GLP-1 bound peptide was then modified to ureido peptide 3 by substituting Ala u for Ala2. Dihedral angle values for main chain atoms of Ala u 2 were manually assigned to that of a right-handed 2.5 helix, in agreement with our previous work on the NMR structure elucidation of an ureido heptamer in methanol [9] . Hydrogen atoms were added with AMBER16 (University of California, San Francisco). Previously developed AMBER parameters were used to model ureido alanine [9] . The CHARMM-GUI interface [10] was used to orient the receptorpeptide complex in a 114 * 113 *136 Å hydrated phospholipid bilayer (60 cholesterol + 120 POPC + molecules ( Figure S8). The full system containing 125,694 atoms was first refined in AMBER16 using the ffSB14 force field [11] and the lipid14 parameter set. [12] A harmonic positional restraint of 20 kcal.mol -1 A -2 was first set on GLP-1R and peptide 3 for 10,000 steps of steepest descent followed by 20,000 steps of conjugate gradient minimization. A non-bonded cut off value of 10 Å was used to compute non-bonded interactions with the particle-mesh Ewald (PME) to handle long-range electrostatic interactions with periodic boundaries. A second minimization step similar to the first one was performed while reducing the positional restraints to a value of 7 kcal.mol -1 A -2 for the receptor-peptide atoms, excepted for the His1-Ala u 2 N-terminal part of the ureido peptide for which the restraint was looser (1.5 kcal.mol -1 A -2 ). A last fully unrestrained minimization protocol (10,000 steps of steepest-descent + 20,000 steps of conjugate gradient) was then applied to the full system.
All atoms were then submitted to a molecular dynamics (MD) protocol consisting of 10 iterative equilibrium steps of 25 fs each using a time step of 1 fs, a Langevin thermostat and a collision frequency of 1 ps -1 . In the first three steps, a constant volume was applied and the temperature gradually increased from 100 to 300K while restraining differently the protein-peptide complex and the ions (restraint of 10 kcal.mol -1 A -2 ) from the phospholipid bilayer and water molecules (restraint of 2.5 kcal.mol -1 A -2 ). In the next four steps, a constant pressure of 1 atm was applied with a Berendsen barostat, a isotropic position scaling, and a pressure relaxation time of 2.0 ps. The SHAKE algorithm was applied to constrain bonds involving hydrogen atoms and harmonic positional restraints were gradually decreased from 5 to 0.5 kcal.mol -1 A -2 (protein + peptide), and from 2.5 to 0.1 kcal.mol -1 A -2 for the hydrated phospholipid bilayer. In the last three equilibrium steps of NPT dynamics, the phospholipid bilayer was free to move, while the restraints on the protein-peptide complex were decreased from 1 to 0.5 kcal.mol -1 A -2 and lastly removed. The system was next simulated for three final steps of 2 ns NPT dynamics each, where the time step was set to 2 fs and the skinnb parameter set to a value of 5 Å. Last, 5 independent production NPT dynamics trajectories of 60 ns each were run to sample the conformational space of the receptor-peptide complex in a hydrated phospholipid bilayer. All simulations were run on a NVIDIA Tesla K80 GPU accelerator using the CUDA version of AMBER PMEMD module.
Recording statistics along the entire trajectory for density, temperature, pressure and total energy shows that all simulations were well equilibrated ( Figure S9). Figure S9: Time course and average value for density, temperature, pressure and total energy along the 60 ns production molecular dynamics trajectories of GLP1-R in complex with the ureido peptide 3 in a fully hydrated phospholipid bilayer.
Atomic coordinates were saved every 20 ps, therefore producing a set of 3,000 snapshots for each trajectory out of which the last 10 ns (50-60 ns simulation time) were kept for producing a timeaveraged structure ( Figure S10). Root-mean square deviations (rmsd) of main chain atomic coordinates to that of the starting conformation shows that neither the GLP-1R nor the peptide main chain atoms encounter major conformational changes (rmsd < 2.0 Å), excepted for the fifth run for which the ureido peptide is drifting away from its initial positions, mostly at its C-terminal end ( Figure S11). Altogether, the first four trajectories are remarkably stable considering the large size of the simulated system (125,964 atoms). Receptor-peptide non-bonded interactions were computed with the in-house developed IChem toolkit [13] and converted into a receptor-ligand interaction fingerprint [14] for ease of comparisons of MD structures with respect to the GLP-1R/GLP-1 cryo-EM template (Figure S12) Analysis of the interaction fingerprints reveals that most of key interactions seen in the reference cryo-EM structure between GLP-1R and GLP-1 are conserved in the five MD trajectories of the ureido peptide bound to the same receptor. As to be expected, interactions of the deeply buried N-terminal residues to the TMD are much better conserved than that of the more flexible C-terminal end to the GLP-1R ECD. Notably key hydrogens bonds to Tyr152, Arg190 and Arg299 [8] are strictly conserved across the five MD trajectories. Specific interactions, not observed with GLP-1, are also detected upon binding of the ureido peptide. The most significant concerns the N-terminal tripeptide residues (His1-Ala u 2-Glu3) and are a direct consequence of the Ala u 2 to Ala substitution: (i) an ionic bond between the N-terminal ammonium of His1 and Glu364, (ii) an hydrogen bond of the N nitrogen of Ala u 2 to Glu387, (iii) an hydrogen bond of Glu3 side chain to Tyr152.

S24
The solvent was half evaporated. The organic phase was washed with KHSO 4 (1M), brine, dried over MgSO 4 and concentrated. The mixture was triturated in hexane to afford the monomer M4 as a white powder (4.1 g, 12.5 mmol, 50% yield).      a: Boc-L-Ala-OH (10.0 g, 52.9 mmol) was dissolved in dry THF (100 ml) at -10 °C, (ice + salt bath) under N 2 . 4-Methymorpholine (6.4 mL, 58.1 mmol) was added. Isobutyl chloroformate (7.2 mL, 55.5 mmol) was added dropwise via an addition funnel. The mixture was stirred 45 min at -10 °C. The white precipitate was filtered and washed with THF. b: Sodium borohydride (4.0 g, 105.7 mmol) was dissolved in 8 ml of H 2 O and the previous filtrate was added dropwise at 0 °C. The mixture was stirred overnight at room temperature. THF was evaporated. The compound was solubilized in EtOAc and washed with KHSO 4 (1M), NaHCO 3 (sat) and brine (sat), dried over MgSO 4 and concentrated. c: I20 (9.2 g, 52.9 mmol) was dissolved in TFA (40 ml). The mixture was stirred 3h at room temperature. TFA was evaporated and coevaporated with cyclohexane. The compound was dissolved in Et 2 O and extracted with H 2 O. The aqueous phase was lyophilized to afford I21. d: I21 (6.2 g, 32.7 mmol) was dissolved in CH 3 CN/H 2 O (50:50, 100 ml). Potassium carbonate (10.01 g, 72.4 mmol), Copper(II) sulfate pentahydrate (82.0 mg, 330 umol) and imidazolium salt (8.2 g, 39.2 mmol) were added. The mixture was stirred 1.5 h at room temperature. EtOAc was added and the phase were separated. Aqueous phase was extracted with EtOAc. Organic phases were combined and washed with KHSO 4 and brine. The solvent was evaporated and the compound was purified by flash column chromatography on silica gel. Eluent Cyclohexane/ EtOAc, 90:10. e: I22 (0.57 g, 5.6 mmol), was dissolved in DCM under N 2 . Pyridine (0.5 ml, 6.2 mmol) was added and the reaction mixture was cooled to 0°C. 4-nitrophenylchloroformate (2.27 g, 11.3 mmol) dissolved in DCM (5ml) was added dropwise and the reaction mixture was stirred 5h. The organic phase was washed with NaHCO 3 (1M), dried over MgSO 4