Discovery of Novel Allosteric Modulators Targeting an Extra-Helical Binding Site of GLP-1R Using Structure- and Ligand-Based Virtual Screening

biomolecules

Article Discovery of Novel Allosteric Modulators Targeting an Extra-Helical Binding Site of GLP-1R Using Structure- and Ligand-Based Virtual Screening

Qingtong Zhou 1 , Wanjing Guo 2,3, Antao Dai 2,4, Xiaoqing Cai 2,4,Márton Vass 5,†, Chris de Graaf 5,‡ , Wenqing Shui 6,7, Suwen Zhao 6,7,*, Dehua Yang 2,3,4,* and Ming-Wei Wang 1,2,3,4,7,*

1 School of Basic Medical Sciences, Fudan University, Shanghai 200032, China; [email protected] 2 The National Center for Drug Screening, Shanghai Institute of Materia Medica, Chinese Academy of Sciences, Shanghai 201203, China; [email protected] (W.G.); [email protected] (A.D.); [email protected] (X.C.) 3 University of Chinese Academy of Sciences, Beijing 100049, China 4 The CAS Key Laboratory of Receptor Research, Shanghai Institute of Materia Medica, Chinese Academy of Sciences, Shanghai 201203, China 5 Amsterdam Institute for Molecules, Medicines and Systems, Division of Medicinal Chemistry, Faculty of Sciences, Vrije Universiteit Amsterdam, 1081 Amsterdam, The Netherlands; [email protected] (M.V.); [email protected] (C.d.G.) 6 iHuman Institute, ShanghaiTech University, Shanghai 201210, China; [email protected] 7 School of Life Science and Technology, Shanghai Tech University, Shanghai 201210, China * Correspondence: [email protected] (S.Z.); [email protected] (D.Y.); [email protected] (M.-W.W.) † Current address: BenevolentAI, 4-8 Maple Street, London W1T 5HD, UK. ‡ Current address: Sosei Heptares, Steinmetz Granta Park Building, Great Abington, Citation: Zhou, Q.; Guo, W.; Dai, A.; Cambridge CB21 6DG, UK. Cai, X.; Vass, M.; de Graaf, C.; Shui, W.; Zhao, S.; Yang, D.; Wang, M.-W. Abstract: Allosteric modulators have emerged with many potential pharmacological advantages as Discovery of Novel Allosteric they do not compete the binding of agonist or antagonist to the orthosteric sites but ultimately affect Modulators Targeting an downstream signaling. To identify allosteric modulators targeting an extra-helical binding site of Extra-Helical Binding Site of GLP-1R Using Structure- and Ligand-Based the glucagon-like peptide-1 receptor (GLP-1R) within the membrane environment, the following Virtual Screening. Biomolecules 2021, two computational approaches were applied: structure-based virtual screening with consideration of 11, 929. https://doi.org/10.3390/ lipid contacts and ligand-based virtual screening with the maintenance of speciﬁc allosteric pocket biom11070929 residue interactions. Veriﬁed by radiolabeled ligand binding and cAMP accumulation experiments, two negative allosteric modulators and seven positive allosteric modulators were discovered us- Academic Editor: Pedro J. Ballester ing structure-based and ligand-based virtual screening methods, respectively. The computational approach presented here could possibly be used to discover allosteric modulators of other G protein- Received: 26 April 2021 coupled receptors. Accepted: 16 June 2021 Published: 23 June 2021 Keywords: GLP-1R; virtual screening; allosteric modulator; drug discovery; molecular docking

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional afﬁl- 1. Introduction iations. G protein-coupled receptors (GPCRs) inﬂuence virtually every aspect of human phys- iology [1,2] and are one of the most successful therapeutic targets with over 500 approved drugs [3,4]. To mediate transmembrane signal transduction, GPCR has the following two spatially distant but conformationally linked regions: the extracellular agonist-binding Copyright: © 2021 by the authors. site and the intracellular transducer binding site [5,6]. The agonist binding in the extra- Licensee MDPI, Basel, Switzerland. This article is an open access article cellular side promotes the transmembrane domain (TMD) of GPCR to undergo extensive distributed under the terms and conformational changes and eventually activates the intracellular transducers, such as conditions of the Creative Commons G protein and β-arrestin. Besides the orthosteric sites where the endogenous agonists Attribution (CC BY) license (https:// bind, recent structural and pharmacological studies highlight the fact that ligands can bind creativecommons.org/licenses/by/ spatially and topologically to distinct (allosteric) sites on receptors and modulate GPCR- 4.0/). mediated signaling simultaneously through conformational cooperativity [6–12]. Allosteric

Biomolecules 2021, 11, 929. https://doi.org/10.3390/biom11070929 https://www.mdpi.com/journal/biomolecules Biomolecules 2021, 11, 929 2 of 14

modulators that enhance the functional response of an orthosteric agonist are regarded as positive allosteric modulators (PAMs), while those that inhibit or negatively modulate the action of an orthosteric ligand are called negative allosteric modulators (NAMs) [13]. The glucagon-like peptide-1 (GLP-1) receptor (GLP-1R) belongs to the class B1 GPCR subfamily and plays a crucial role in glucose homeostasis. As a successful therapeutic target for type 2 diabetes and obesity, many peptidic analogs of GLP-1 are on the market. Meanwhile, continuous efforts in small-molecule drug discovery resulted in several nonpeptidic agonists such as Boc5 [14], TT-OAD2 [15], PF-06882961 [16], RGT1383 [17] and HD-7671 [18], as well as dozens of small molecule allosteric modulators that interact with distinct regions of GLP-1R. There are at least four reported allosteric sites in class B1 GPCRs including: (i) deep inside the helical bundle observed in the corticotropin-releasing factor receptor type 1 receptor (CRF1R) [19]; (ii) extracellular helical bundle at the TM helices 1 and 2 interface found in GLP-1R [8]; (iii) TM helices 3−4−5 interface with receptor activity-modifying protein 1 (RAMP1) seen in the calcitonin gene related peptide receptor (CGRP) [20]; and (iv) outside of the helical bundle of TM helices 5, 6 and 7 at the lipid interface of GLP-1R and glucagon receptor (GCGR) [7,21]. Interestingly, two PAM agonists of GLP-1R, 6,7-dichloro-3-methanesulfonyl-2-tert-butylamino-quinoxaline (compound 2) and 4- (3-benzyloxyphenyl)-2-(ethylsulfinyl)-6-(trifluoromethyl) pyrimidine (BETP) [6,7,22], have been found to covalently link with C3476x36b (Wootten numbering in superscript) [23] at the intracellular end of TM6, sharing a similar region as the NAMs (such as NNC0640 and PF-0637222) [7,21]. In the case of GLP-1R, a variety of allosteric modulators have been described [24,25], including compound 19 [26], ZINC19797057 [27], compound 3286 [28], CD3878-F005 [29], HTL26119 [30] and ZINC00702587 [31], thereby greatly enhancing our knowledge about the small molecule allosterism of this important drug target. Assisted by advances in algorithm, protocol and software development, computer- aided drug design (CADD) has shown promise in the discovery of novel drug candidates targeting GPCRs [32,33]. Structure-based virtual screening (SBVS) and ligand-based virtual screening (LBVS) are two distinct CADD approaches. SBVS employs three-dimensional (3D) structural information of a target protein and performs molecular docking to identify potent binders, while LBVS utilizes known active ligands to establish a structure-activity relationship (SAR) for guiding subsequent lead discovery and optimization. Given that the scoring function in SBVS is developed for soluble proteins without considering the membrane environment, and the number of known allosteric modulators for LBVS is limited, virtual screening that aims to discover allosteric modulators for GPCRs is still challenging. Here, we report the discovery of allosteric modulators of GLP-1R by computational screening and experimental validation. Focused on an extra-helical allosteric site (outside of the intracellular half of TMs 5-6-7 at the lipid interface) where the NAM PF-0637222 and PAM compound 2 bind, we developed a new virtual screening strategy to discover allosteric modulators at the lipid interface of GPCR TMD: SBVS takes the lipid interactions into consideration and LBVS retains the specific residue contacts involved in the allosteric site. Verified using cAMP accumulation and radiolabeled ligand binding assays, two NAMs and seven PAMs were discovered using these two methods, respectively.

2. Materials and Methods 2.1. Ligand Database Chemical structures from Enamine, ChemDiv, Vitas-M, ChemBridge and TimTec (~540,000 compounds) as well as the Chinese National Compound Library (CNCL, ~760,000 compounds) were collected for virtual screening. The reported NAMs (PF- 06372222, NNC0640 and MK0893) that interact with the extra-helical binding sites of GLP-1R or GCGR share similar chemotypes and one anionic end (carboxylic acid or tetra- zole) inserts into a polar cleft between TM6 and TM7. After being protonated at a pH of 7.4 and its properties calculated using ChemAxon (cxcalc 5.1.4, Budapest, Hungary), the compound collection was ﬁltered with the following criteria: (i) 18 ≤ HAC (number of heavy atoms) ≤ 36; (ii) 1 ≤ HBA (number of hydrogen bond acceptors) ≤ 10; (iii) 0 ≤ HBD Biomolecules 2021, 11, 929 3 of 14

(number of hydrogen bond donors) ≤ 5; (iv) 0 ≤ clogP ≤ 5; and (v) no positively charged atoms and at least one negatively charged atom or containing an acid isostere [34]. A substructure search was then performed using the ChemAxon (cxcalc 5.1.4), resulting in 27,939 compounds having negatively charged atoms, 16,103 compounds with isostere from CNCL, as well as 195,831 and 87,873 compounds from commercially available libraries, respectively. In addition, 583 CNCL compounds and 11,779 commercially available compounds were identiﬁed as they have alternative chirality. Three-dimensional conformations of these compounds were generated using Corina (4.1.0, Erlangen, Germany) [35] with default parameters, except for the maximum number of generated stereoisomers per molecule, which was set to 1 to restrict the number of output stereoisomers. Finally, a tailored library consisting of 340,108 compounds was prepared for virtual screening.

2.2. Protein Preparation The crystal structure of human GLP-1R TMD in a complex with two negative allosteric modulators [7], PF-06372222 (PDB code: 5VEW) and NNC0640 (PDB code: 5VEX), were obtained from the Protein Data Bank [36]. The missing side chains and hydrogens were added and optimized using Protein Preparation Wizard (New York, NY, USA) [37] in Schrödinger Suite 2017-3. Given the lipid environment for the extra-helical binding site of GLP-1R, we embedded the receptor within the palmitoyl oleoyl phosphatidyl choline (POPC) bilayer using PyMOL (v1.7, New York, NY, USA) [38,39] and performed short-time molecular dynamics (MD) simulation to relax the protein and lipid molecules. The force field parameters of two ligands (PF-06372222 and NNC0640) were modelled with ACPYPE (Wilmington, DE, USA) [40], while the CHARMM36-CAMP force field [41] was applied to the receptor and lipids. MD simulations were conducted using Gromacs (5.0.2, Groningen, The Netherlands) [42]. All bonds involving hydrogen atoms were constrained using LINCS algorithm [43]. The particle mesh Ewald (PME) method was used to treat long-range electrostatic interactions with a cutoff of 10 Å. The entire system was first relaxed using the steepest descent energy minimization, followed by equilibration steps of 5 ns in total to equilibrate the lipid bilayer and the solvent, while the position restraints to the protein and the ligand were retained. As we focused on the extra-helical binding site of GLP-1R, these POPC molecules within 5 Å of the ligand were kept, while the rest of lipids, water and ions were removed.

2.3. Structure-Based Virtual Screening Receptor grids were generated for the two complexes using the Receptor Grid Gen- eration tool (New York, NY, USA) in the Glide module [44] of Schrödinger Suite 2017-3. The grid boxes were defined as a 10 × 10 × 10 Å3 region centered at negative allosteric modulators PF-06372222 or NNC0640. For virtual screening, the prepared 340,108 compounds were subjected to the following two levels of docking using the relevant workflow in Glide [44]. After a standard precision (SP) docking stage, compounds with top 10% docking scores went through an extra precision (XP) docking process by adopting a more accurate scoring function essential to reduce false positives. To further relax the complexes and approximately calculate the binding free energies, the XP docking poses with the top 1,000 docking scores were subjected to a Prime MM-GBSA calculation [45]. All residues within 7 Å of the docked ligands were relaxed with the sampling method “Minimize”. Finally, by visual inspection of the optimized docking pose and considering the XP docking scores, calculated binding free energies (MM-GBSA dG Bind) and chemotype diversity, 45 compounds were selected and purchased from TargetMol (Boston, MA, USA) with a specified purity of >95% for experimental validation (Table S1).

2.4. Ligand-Based Virtual Screening Ligand-based dockings were performed using Protein-Ligand ANT System (PLANTS) [46] version 1.2 (Konstanz, Germany), post-processed and ranked by Interaction Fingerprints (IFP) [47]. PLANTS employs an ant-colony-optimization algorithm for the prediction of Biomolecules 2021, 11, 929 4 of 14

binding poses of small molecules in a protein structure and an empirical scoring function, ChemPLP, for grading the resultant binding poses. The docking site of GLP-1R was defined by all residues within a 5 Å radius around the co-crystallized ligands (PF-06372222 and NNC0640) [7], and for each ligand, 25 poses were produced (speed setting 2) and scored using the ChemPLP scoring function [46]. IFP evaluates a (predicted) binding mode of a compound in a protein structure by annotating the presence or absence of different types of interactions (hydrophobic, aromatic, H-bond, ionic) between each pocket residue and the molecule based on a set of rules [47]. This results in a molecular interaction fingerprint representing all interactions between the molecule and the protein in bit-string, allowing for easy comparison and scoring (using the Tanimoto coefficient) of the similarity of multiple IFPs. In our previous virtual screening toward orthosteric sites of the histamine H1 receptor (H1R) [48], the IFP score (≥0.75) and PLANTS score (≤−90) cutoffs are able to discriminate H1R ligands from decoys with the discovery of a chemically diverse set of novel fragment- like H1R ligands. A similar screening strategy was applied to β2-adrenoceptor with a hit rate of 53% and hits with up to nanomolar affinities and potencies [49]. Considering that the allosteric modulator-binding site in the present study is on the lipid facing surface where the ligands have more freedom in the docking simulation, a lower PLANTS score (≤−75) and IFP similarity (IFP score of ≥ 0.6) cutoff were applied to obtain a reasonable number of top hits from the virtual screening. The allosteric GLP-1R pocket was defined by 41 pocket residues within 6.5 Å around the co-crystallized ligands [7] (R2x46b,H2x50b,L3x54b, F5x54b,V5x57b,I5x58b,V5x61b,V5x62b,L5x65b, M340, K342, D6x33b,I6x34b,K6x35b,F6x36b,R6x37b, L6x38b,A6x39b,K6x40b,S6x41b,T6x42b,L6x43b,T6x44b,L6x45b,I6x46b,P6x47b,L6x48b,F7x44b,L7x51b, M7x52b,I7x55b,L7x56b,Y7x57b,C7x58b,F7x59b,V7x60b,N8x47b,N8x48b,E8x49b,V8x50b and Q8x51b). Ionic interaction distance cutoff was set to 6.5 Å. In the GLP-1R retrospective validation, the binding mode of the co-crystallized compound for each respective crystal structure [7] was used for calculation of a reference IFP. These reference IFPs were subsequently used to score the docking poses. Filtering was performed by applying two filters (polar interactions with S6x41b and N8x47b, and contacts with at least one of V5x61b,F6x36b and L6x43b), the PLANTS (PLANTS score of ≤−75 according to the best IFP pose) and IFP cutoff (only compounds with an IFP score of ≥0.6 according to the best PLANTS pose). Compounds were visually clustered based on scaffold similarity to ensure structural diversity, and those with buried polar groups located in hydrophobic parts of the receptor binding site were discarded by visual inspection. Finally, 42 compounds, including one compound shared by LBDD (ZINC72191544), were chosen and purchased from TargetMol (Boston, MA, USA) with a specified purity of >95% for experimental validation (Table S2).

2.5. cAMP Accumulation Assay cAMP accumulation was measured using the LANCE Ultra cAMP kit (PerkinElmer, Boston, MA, USA) according to the manufacturer’s instructions. CHO-K1 cells stably expressing wild-type (WT) or mutant GLP-1R were digested by 0.2% (w/v) EDTA and washed once with PBS. Cells were then resuspended with assay buffer (DMEM, 0.1% BSA, 5 mM HEPES) and seeded onto 384-well plates (6 × 105/mL, 5 µL/well). Transfected cells were incubated for 40 min with 20 µM allosteric modulators and different concentrations of GLP-1. After the addition of 5 µL of Eu-cAMP tracer and ULight-anti-cAMP, the reactions were stopped. The plates were left for 1 h at room temperature to measure time-resolved FRET signals at excitation wavelengths of 620 nm and 665 nm by EnVision (PerkinElmer, Waltham, MA, USA). The cAMP response is depicted relative to the maximal response of GLP-1 at the WT receptor (100%).

2.6. Whole Cell Binding Assay CHO-K1 cells stably expressing WT or mutant GLP-1R were seeded into 96-well 4 ◦ plates at a density of 3 × 10 cells/well and incubated overnight at 37 C and 5% CO2. For homogeneous binding, cells were washed twice and incubated with a blocking buffer (F12 supplemented with 33 mM HEPES, and 0.1% (w/v) BSA, pH 7.4) for 2 h at 37 ◦C. Biomolecules 2021, 11, 929 5 of 14

Then, radiolabeled 125I-GLP-1 (40 pM, PerkinElmer, Waltham, MA, USA) and unlabeled compounds were added and reacted with the cells in binding buffer (F12 supplemented with 33 mM HEPES, and 0.1% (w/v) BSA, pH 7.4) at 4 ◦C overnight. Cells were washed three times with ice-cold PBS and lysed using 50 µL of lysis buffer (PBS supplemented with 20 mM Tris–HCl and 1% Triton X-100, pH 7.4). Subsequently, the plates were counted for radioactivity (counts per minute, CPM) in a scintillation counter (MicroBeta2 Plate Counter, PerkinElmer, Waltham, MA, USA) using a scintillation cocktail (OptiPhaseSuper- Mix; PerkinElmer, Waltham, MA, USA).

3. Results 3.1. Virtual Screening Workflow A schematic workflow of the virtual screening toward the allosteric modulator for GLP-1R is presented in Figure1. The high-resolution X-ray structures of the human GLP-1R in complex with negative allosteric modulators PF-06372222 and NNC0640 were adopted [7,37]. After analyzing the reported allosteric modulators, we screened a focused library with loosely defined physicochemical properties that were filtered as described above [34]. SBVS and LBVS were performed independently with an identical input (chemical library and receptor file). For SBVS, three stages were adopted. In the first stage, the compounds were docked using Glide (New York, NY, USA) standard precision (SP) [44] where the top 10% of good scoring compounds were retained. Then, these compounds were docked with Glide extra precision (XP) where the top 1000 hits of the best docking scores were kept. Finally, these hits were processed by Prime MM-GBSA [45] to further relax the complex and calculate the binding free energy. For LBVS, three filtering protocols were applied. Firstly, these compounds were docked to the receptor with PLANTS [50,51], which predicts and scores the binding poses in a protein structure by a combination of an ant colony optimization algorithm with an empirical scoring function. The resultant docking poses of 10,000 compounds were subjected to molecular interaction fingerprint (IFP) calculations, where seven different interaction types (negatively charged, positively charged, H-bond acceptor, H-bond donor, aromatic face-to-edge, aromatic face-to-face and hydrophobic interactions) were adopted to define the IFP. The top 100 compounds with the best IFP scores were further screened with the pan-assay interference compound filter (PAINS) [52] to exclude reactive functional groups and promiscuous chemotypes. Based on our previous LBVS experience [48,49] and the consideration that the docked ligands in the allosteric modulator-binding site may have more freedom in the lipid environment, the PLANTS score (≤−75) and IFP similarity (IFP score of ≥0.6) cutoff were applied. To maximize the chemotype diversity, these compounds were classified into 42 clusters using spectral clustering, and the potency of the compounds in each cluster was further assessed by visual inspection of their interactions with the residues that constructed the allosteric extra-helical binding site of GLP-1R. Eventually, 45 compounds identified using SBVS and 42 compounds found using LBVS were purchased and underwent experimental validation (Tables S1 and S2).

3.2. MD Simulation Studies Given that the accuracy of molecular docking is fundamental for high-quality virtual screening, it is important to redock NNC0640 and PF-06372222 to the allosteric site of GLP-1R. Two different ligand-sampling methods, score in place (not docking and use the original position for scoring) and flexible docking (generate conformations internally and then perform molecular docking and scoring), were applied during Glide XP docking, thereby reflecting the scoring accuracy and sampling efficiency, respectively. As shown in Figure2, the Glide XP (New York, NY, USA) docking scores of the crystal pose of NNC0640 in GLP-1R [7] by score in place and flexible docking are −2.9 and −4.1 kcal/mol (a more negative docking score indicates stronger binding), respectively. A similar observation was made for PF-06372222, whose docking scores are −4.8 and −4.6 kcal/mol, respectively. Moreover, the root-mean square deviation (RMSD) of the redocked poses relative to the Biomolecules 2021, 11, 929 6 of 14

crystal pose [7] is 5.5 Å for NNC0640 and 5.2 Å for PF-06372222 (Figure2A), suggesting that traditional molecular docking failed to reproduce the crystal poses [7]. In addition, Prime MM-GBSA calculation demonstrates that both ligands showed weak binding energies (MM-GBSA dG Bind, around −50 kcal/mol). As a comparison, the orthosteric ligands

Biomolecules with2021, 11, nanomolarx FOR PEER REVIEW binding afﬁnity in class A GPCRs generally have a docking6 of score 15 and

MM-GBSA dG Bind lower than −10.0 and −100.0 kcal/mol, respectively.

FigureFigure 1. Schematic 1. Schematic representation representation of the GLP-1R allosteric of the modulator GLP-1R discovery allosteric process modulator with structure-based discovery virtual process with screening (SBVS) and ligand-based virtual screening (LBVS). By analyzing the reported allosteric modulators (PF-0637222, NNC0640structure-based and MK0893), a tailored virtual library screening consisting (SBVS) of 340,108 and compounds ligand-based underwent virtual the following screening two different (LBVS). proto- By analyzing cols: structure-the reported and ligand-based allosteric virtual modulators screening. (PF-0637222, NNC0640 and MK0893), a tailored library consisting Biomolecules 2021of, 340,108 11, x FOR compoundsPEER REVIEW underwent the following two different protocols: structure- and ligand-based7 of 15 3.2. MD Simulation Studies virtual screening. Given that the accuracy of molecular docking is fundamental for high-quality virtual screening, it is important to redock NNC0640 and PF-06372222 to the allosteric site of GLP-1R. Two different ligand-sampling methods, score in place (not docking and use the original position for scoring) and flexible docking (generate conformations internally and then perform molecular docking and scoring), were applied during Glide XP docking, thereby reflecting the scoring accuracy and sampling efficiency, respectively. As shown in Figure 2, the Glide XP (New York, NY, USA) docking scores of the crystal pose of NNC0640 in GLP-1R [7] by score in place and flexible docking are −2.9 and −4.1 kcal/mol (a more negative docking score indicates stronger binding), respectively. A similar observation was made for PF-06372222, whose docking scores are −4.8 and −4.6 kcal/mol, respectively. Moreover, the root-mean square deviation (RMSD) of the redocked poses relative to the crystal pose [7] is 5.5 Å for NNC0640 and 5.2 Å for PF-06372222 (Figure 2A), suggesting that traditional molecular docking failed to reproduce the crystal poses [7]. In addition, Prime MM-GBSA calculation demonstrates that both ligands showed weak binding energies (MM-GBSA dG Bind, around −50 kcal/mol). As a comparison, the orthosteric ligands with nanomolar binding affinity in class A GPCRs generally have a docking score and MM-GBSA dG Bind lower than −10.0 and −100.0 kcal/mol, respectively. To mimic the lipid environment of the extra-helical binding site of GLP-1R during molecular docking, we performed short-time MD simulations for NNC0640 or PF- 06372222 bound GLP-1R placed in a POPC bilayer that was retained within 5 Å of the ligands (Figure 2A). With the assistance of POPC, the Glide XP docking scores of both were improved to −11.1 and −10.9 kcal/mol, respectively (Figure 2B), significantly better than that without lipids. In addition, their docked poses are almost identical to that observed in crystal structures [7] with RMSDs of 1.1 Å and 0.3 Å, respectively. A similar improvement was noted with the Prime MM-GBSA binding free energy, whose values were −84.9 and −86.3 kcal/mol for NNC0640 and PF-06372222, respectively (Figure 2B). Collectively, these results demonstrate the key role of lipids in generating correct docking conformations and reasonable docking scores.

Figure Figure2. Evaluation 2. Evaluation of molecular of docking molecular protocols docking on re-doc protocolsked negative on re-docked allosteric modulators negative NNC0640 allosteric and modulators PF- 06372222 to the allosteric extra-helical binding site of GLP-1R [7]. (A) Comparison of docking poses of NNC0640 in GLP- 1R by NNC0640docking protocol and with PF-06372222 or withoutto the the consideration allosteric extra-helicalof lipids. The crystal binding structures site of of GLP-1R human GLP-1R [7]. (A TMD) Comparison in complexof with docking PF-06372222 poses (PDB of NNC0640 code: 5VEW) in [7] GLP-1R and NNC0640 by docking (PDB code: protocol 5VEX) [7] with were or obtained without from the the considerationProtein Data Bankof lipids. [36]. CarbonThe atoms crystal of structuresNNC0640 in ofcrystal human structur GLP-1Re [7], traditional TMD in Glide complex XP docking with and PF-06372222 Glide XP docking (PDB code: within lipids are colored orange, cyan and magenta, respectively. (B) Comparison of docking scores and calculated binding free5VEW) energies [ 7(MM-GBSA] and NNC0640 dG Bind) (PDB of NNC0640 code: 5VEX) and PF- [706372222] were obtainedto GLP-1R from with theor without Protein considering Data Bank the [ 36lipid]. Carbon contribution.atoms Molecular of NNC0640 docking in and crystal binding structure free energy [7], calculation traditional were Glide performed XP docking using Glide and and Glide Prime XP MM-GBSA docking within modules in Schrödinger Suite 2017-3, respectively. Docking score and MM-GBSA dG Bind are in kcal/mol.

3.3. Identification of New NAMs In SBVS, 113 compounds have Glide XP (New York, NY, USA) docking scores that are better than −10.0 kcal/mol, and 11 of them are lower than −11.0 kcal/mol, comparable to that of NNC0640 (−11.1 kcal/mol) and PF-06372222 (−10.9 kcal/mol). However, none have a better binding free energy (MM-GBSA dG Bind) than that of NNC0640 (−84.9 kcal/mol) or PF-06372222 (−86.3 kcal/mol). Experimental validation found that two of them, Z21 (ZINC254697034) and Z42 (ZINC16949012), negatively modulated GLP-1-elicited cAMP accumulation (Figures 3 and 4A, Table 1), where Z21 has an EC50 of 76 μM. In the binding assay, Z21 decreased the binding of GLP-1 to GLP-1R (Figure 4B). Figure 3 shows that Z21 and Z42 were predicted to occupy the extra-helical binding site in a manner similar to NNC0640 and PF-06372222. These four compounds share a common negatively charged terminus that inserts into a positively charged cleft between TM6 and TM7 and forms multiple hydrogen bonds or a salt bridge with the polar residues (such as R1762x46b, N4068x47b and N4078x48b). In terms of other segments, Z21 and Z42 make weaker interactions with GLP-1R compared to NNC0640 or PF-06372222. The residues in TM6 form at least two hydrogen bonds with both PF-06372222 and NNC0640 via S3526x41b and T3556x44b, but only one hydrogen bond was observed for Z21 (via S3526x41b) or Z42 (via T3556x44b). Moreover, PF-06372222 and NNC0640 have extensive hydrophobic contacts with I3285x58b, V3315x61b, A3506x39b and L3546x43b through their extended arms pointing to Biomolecules 2021, 11, 929 7 of 14

lipids are colored orange, cyan and magenta, respectively. (B) Comparison of docking scores and calculated binding free energies (MM-GBSA dG Bind) of NNC0640 and PF-06372222 to GLP-1R with or without considering the lipid contribution. Molecular docking and binding free energy calculation were performed using Glide and Prime MM-GBSA modules in Schrödinger Suite 2017-3, respectively. Docking score and MM-GBSA dG Bind are in kcal/mol.

To mimic the lipid environment of the extra-helical binding site of GLP-1R during molecular docking, we performed short-time MD simulations for NNC0640 or PF-06372222 bound GLP-1R placed in a POPC bilayer that was retained within 5 Å of the ligands (Figure2A ). With the assistance of POPC, the Glide XP docking scores of both were improved to −11.1 and −10.9 kcal/mol, respectively (Figure2B), signiﬁcantly better than that without lipids. In addition, their docked poses are almost identical to that observed in crystal structures [7] with RMSDs of 1.1 Å and 0.3 Å, respectively. A similar improvement was noted with the Prime MM-GBSA binding free energy, whose values were −84.9 and −86.3 kcal/mol for NNC0640 and PF-06372222, respectively (Figure2B). Collectively, these results demonstrate the key role of lipids in generating correct docking conformations and reasonable docking scores.

3.3. Identiﬁcation of New NAMs In SBVS, 113 compounds have Glide XP (New York, NY, USA) docking scores that are better than −10.0 kcal/mol, and 11 of them are lower than −11.0 kcal/mol, comparable to that of NNC0640 (−11.1 kcal/mol) and PF-06372222 (−10.9 kcal/mol). However, none have a better binding free energy (MM-GBSA dG Bind) than that of NNC0640 (−84.9 kcal/mol) or PF-06372222 (−86.3 kcal/mol). Experimental validation found that two of them, Z21 (ZINC254697034) and Z42 (ZINC16949012), negatively modulated GLP-1-elicited cAMP accumulation (Figures3 and4A, Table1), where Z21 has an EC 50 of 76 µM. In the binding assay, Z21 decreased the binding of GLP-1 to GLP-1R (Figure4B). Figure3 shows that Z21 and Z42 were predicted to occupy the extra-helical binding site in a manner similar to NNC0640 and PF-06372222. These four compounds share a common negatively charged terminus that inserts into a positively charged cleft between TM6 and TM7 and forms multiple hydrogen bonds or a salt bridge with the polar residues (such as R1762x46b, N4068x47b and N4078x48b). In terms of other segments, Z21 and Z42 make weaker interactions with GLP-1R compared to NNC0640 or PF-06372222. The residues in TM6 form at least two hydrogen bonds with both PF-06372222 and NNC0640 via S3526x41b and T3556x44b, but only one hydrogen bond was observed for Z21 (via S3526x41b) or Z42 (via T3556x44b). Moreover, PF-06372222 and NNC0640 have extensive hydrophobic contacts with I3285x58b, V3315x61b, A3506x39b and L3546x43b through their extended arms pointing to TM5. Such contacts are much weaker for Z21 (short arm) and Z42 (without arm). The differences in the binding modes and efficiencies between potent NAMs (PF-06372222 and NNC0640) and weak NAMs (Z21 and Z42) suggest that the polar interaction contributed by TM6 and the hydrophobic contacts from TM5 are essential for negative allosteric modulation, an observation supported by our mutagenesis studies (Figure4B), where C347F enhanced the ligand binding of Z42. These data are valuable to structure-guided lead optimization.

3.4. Identiﬁcation of New PAMs In LBVS, 44 compounds have PLANTS scores better than −75, 18 of them are lower than −100 and 11 display an IFP similarity better than 0.7, suggesting their interactions with GLP- 1R resembled NNC0640 and PF-06372222. Experimental validation showed that compounds C10 (EN_Z1424437838), C11 (EN_Z1445206940), C13 (EN_Z18696867), C22 (EN_Z26483797), C23 (EN_Z28052152), C26 (EN_Z317215770) and C31 (VM_STL480883) positively modulated GLP-1 potency (Figure5, Tables2 and3). Figure6A indicates that C22, C26, C31, C39, C13 and C11 enhanced ligand binding, with C22 being the most potent (IC50 = 0.63 µM). They also facilitated GLP-1-induced intracellular cAMP accumulation to various extents (Figure6C ), e.g., 20 µM C22 increased the GLP-1 potency from 34 pM (EC50) to 9.8 pM. Biomolecules 2021, 11, x FOR PEER REVIEW 8 of 15

TM5. Such contacts are much weaker for Z21 (short arm) and Z42 (without arm). The differences in the binding modes and efficiencies between potent NAMs (PF-06372222 and NNC0640) and weak NAMs (Z21 and Z42) suggest that the polar interaction contributed by TM6 and the hydrophobic contacts from TM5 are essential for negative allosteric mod- Biomolecules 2021, 11, 929 ulation, an observation supported by our mutagenesis studies (Figure 4B), where C347F 8 of 14 enhanced the ligand binding of Z42. These data are valuable to structure-guided lead optimization.

FigureFigure 3. Comparison 3. Comparison of binding of binding modes modes of of representative representative hitshits from SBVS SBVS and and PF-06372222. PF-06372222. Carbon Carbon atoms atoms of PF-06372222, of PF-06372222, Z21 andZ21 Z42 and are Z42 colored are colored orange, orange, green green and an yellow,d yellow, respectively. respectively. DockingDocking score score and and MM-GBSA MM-GBSA dG dG Bind Bind are arein kcal/mol. in kcal/mol.

FigureFigure 4. Experimental 4. Experimental validation validation of selectedof selected compounds compounds identiﬁedidentified using using SBVS. SBVS. (A ()A Dose-dependent) Dose-dependent inhibition inhibition curves curves of of Z21 andZ21 Z42 and on Z42 cAMP on cAMP activity activity induced induced by by constant constant concentration concentration of of GLP-1 GLP-1 (0.08 (0.08 nM nM for for Z21, Z21, 0.05 0.05 nM nMfor Z42) for Z42)in GLP-1R in GLP-1R expressingexpressing CHO-K1 CHO-K1 cells. cells. (B) Binding (B) Binding of Z21 of Z21 or Z42 or Z42 to GLP-1R to GLP-1R or itsor its mutants mutants in in competition competition with with radiolabeled radiolabeled GLP-1.GLP-1. Data Biomolecules 2021, 11, x FOR PEER REVIEW 9 of 15 Biomolecules Data 2021 are, 11presented, x FOR PEER as REVIEWmeans ± S.E.M. of three independent experiments. 9 of 15 are presented as means ± S.E.M. of three independent experiments.

Table 1. Experimentally validated new NAMs based on SBVS approach a. Table 1. Experimentally validated new NAMs based on SBVS approach a. Table 1. Experimentally validated new NAMs based on SBVS approach a. cAMPcAMP Accumulation Accumulation Binding Binding IDID NameName Chemical Structure cAMP Accumulation Binding ID Name ChemicalChemical Structure b b pEC50± ± SEM Emax (% WTb ) pIC50 ± SEM ± Span ± SEM ± pECpEC50 ± SEMSEM EmaxE (%max WT(% WT) pIC) 50 ± SEMpIC 50 SpanSEM ± SEM Span SEM

Z21 Z49584845 4.12 ± 0.38 97.02 ± 0.81 4.72 ± 0.24 75.3 ± 12.57 Z21Z21 Z49584845Z49584845 4.12 ±± 0.380.38 97.02 97.02 ± 0.81± 0.81 4.72 ± 0.24 4.72 ± 75.30.24 ± 12.57 75.3 ± 12.57

Z42 STL446272 N.A.c c 99.72 ± 0.80 N.B. d d N.B. d d Z42Z42 STL446272STL446272 N.A.N.A. c 99.7299.72 ± 0.80± 0.80 N.B. d N.B. N.B. d N.B.

a All data were fitted with a three-parameter logistic curve to obtain pEC50 and pIC50 values. Data represent means ± S.E.M. a a All data were fitted with a three-parameter logistic curve to obtain pEC50 and pIC50 values. Data represent means ± S.E.M. ± All dataof wereat least ﬁtted three withindependent a three-parameter experiments logisticperformed curve in duplicate.; to obtain b pECWT, 50wild-type;and pIC c50 N.A.,values. not active.; Datarepresent d N.B., no binding. means S.E.M. of at least three independentof at least three experiments independent performed experiments in performed duplicate.; in bduplicate.;WT, wild-type; b WT, wild-type;c N.A., not c N.A., active.; not active.;d N.B., d noN.B., binding. no binding. 3.4. Identification of New PAMs 3.4. Identification of New PAMs In LBVS, 44 compounds have PLANTS scores better than −75, 18 of them are lower In LBVS, 44 compounds have PLANTS scores better than −75, 18 of them are lower than −100 and 11 display an IFP similarity better than 0.7, suggesting their interactions than −100 and 11 display an IFP similarity better than 0.7, suggesting their interactions with GLP-1R resembled NNC0640 and PF-06372222. Experimental validation showed that with GLP-1R resembled NNC0640 and PF-06372222. Experimental validation showed that compounds C10 (EN_Z1424437838), C11 (EN_Z1445206940), C13 (EN_Z18696867), C22 compounds C10 (EN_Z1424437838), C11 (EN_Z1445206940), C13 (EN_Z18696867), C22 (EN_Z26483797), C23 (EN_Z28052152), C26 (EN_Z317215770) and C31 (VM_STL480883) (EN_Z26483797), C23 (EN_Z28052152), C26 (EN_Z317215770) and C31 (VM_STL480883) positively modulated GLP-1 potency (Figure 5, Tables 2 and 3). Figure 6A indicates that positively modulated GLP-1 potency (Figure 5, Tables 2 and 3). Figure 6A indicates that C22, C26, C31, C39, C13 and C11 enhanced ligand binding, with C22 being the most potent C22, C26, C31, C39, C13 and C11 enhanced ligand binding, with C22 being the most potent (IC50 = 0.63 μM). They also facilitated GLP-1-induced intracellular cAMP accumulation to (IC50 = 0.63 μM). They also facilitated GLP-1-induced intracellular cAMP accumulation to various extents (Figure 6C), e.g., 20 μM C22 increased the GLP-1 potency from 34 pM various extents (Figure 6C), e.g., 20 μM C22 increased the GLP-1 potency from 34 pM (EC50) to 9.8 pM. (EC50) to 9.8 pM.

Figure 5. Proposed binding modes of representative hits from LBVS. (A,B) Carbon atoms of C22 and C26 are colored green Figure 5. Proposed binding modes of representative hits from LBVS. (A,B) Carbon atoms of C22 and C26 are colored green and gray, respectively. and gray, respectively.

Biomolecules 2021, 11, x FOR PEER REVIEW 9 of 15

Table 1. Experimentally validated new NAMs based on SBVS approach a.

cAMP Accumulation Binding ID Name Chemical Structure pEC50 ± SEM Emax (% WT b) pIC50 ± SEM Span ± SEM

Z21 Z49584845 4.12 ± 0.38 97.02 ± 0.81 4.72 ± 0.24 75.3 ± 12.57

Z42 STL446272 N.A. c 99.72 ± 0.80 N.B. d N.B. d

a All data were fitted with a three-parameter logistic curve to obtain pEC50 and pIC50 values. Data represent means ± S.E.M. of at least three independent experiments performed in duplicate.; b WT, wild-type; c N.A., not active.; d N.B., no binding.

3.4. Identification of New PAMs In LBVS, 44 compounds have PLANTS scores better than −75, 18 of them are lower than −100 and 11 display an IFP similarity better than 0.7, suggesting their interactions with GLP-1R resembled NNC0640 and PF-06372222. Experimental validation showed that compounds C10 (EN_Z1424437838), C11 (EN_Z1445206940), C13 (EN_Z18696867), C22 (EN_Z26483797), C23 (EN_Z28052152), C26 (EN_Z317215770) and C31 (VM_STL480883) positively modulated GLP-1 potency (Figure 5, Tables 2 and 3). Figure 6A indicates that Biomolecules 2021, 11, 929 C22, C26, C31, C39, C13 and C11 enhanced ligand binding, with C22 being the most potent9 of 14 (IC50 = 0.63 μM). They also facilitated GLP-1-induced intracellular cAMP accumulation to various extents (Figure 6C), e.g., 20 μM C22 increased the GLP-1 potency from 34 pM (EC50) to 9.8 pM.

BiomoleculesBiomolecules 2021 2021, ,11 11, ,x x FOR FOR PEER PEER REVIEW REVIEWFigure 5. Proposed binding modes of representative hits from LBVS. (A,B) Carbon1010 ofatoms of 15 15 of C22 and C26 are colored green Biomolecules 2021, 11, x FOR PEER REVIEWFigure 5. Proposed binding modes of representative hits from LBVS.10 of 15 (A,B) Carbon atoms of C22 and Biomolecules 2021,, 11,, xx FORFOR PEERPEER REVIEWREVIEWand gray, respectively. 10 of 15 C26 are colored green and gray, respectively. TableTable 2. 2. Chemical Chemical structures structures and and receptor receptor interaction interaction fingerpr fingerpr intsints of of the the compounds compounds iden identifiedtified by by LBVS LBVS approach. approach. Table 2. Chemical structures and receptor interaction fingerprints of the compounds identified by LBVS approach. TableTable 2. Chemical2. Chemical structures structures and receptor and receptorinteraction fingerpr interactionintsints ofof thethe ﬁngerprints compoundscompounds ideniden oftifiedtified the byby compounds LBVSLBVS approach.approach. identiﬁed by LBVS approach. PLANTSPLANTS IFPIFP IDID NameName Chemical Chemical Structure Structure PLANTS IFP SS6x41b6x41b NN8x47b8x47b TT6x44b6x44b LL6x48b6x48b LL6x43b6x43b ID Name Chemical Structure PLANTSScoreScore ScoreScoreIFPIFP S6x41b N8x47b T6x44b L6x48b L6x43b IDIDID NameNameName Chemical Chemical Chemical Structure Structure PLANTSScore Score Score S6x41b6x41b IFP NN Score8x47b8x47b TT6x44b6x44b SLL6x41b6x48b6x48b LL6x43b6x43b N8x47b T6x44b L6x48b L6x43b Score Score

C22C22 EN_Z26483797EN_Z26483797 −−104.19104.19 0.610.61 11 11 00 00 11 C22 EN_Z26483797 −104.19 0.61 1 1 0 0 1 C22C22 EN_Z26483797 EN_Z26483797 −104.19−104.19 0.61 1 0.611 0 0 1 1 1 0 0 1

C26C26 EN_Z317215770EN_Z317215770 −−90.2890.28 0.690.69 11 11 11 00 11 C26 EN_Z317215770 −90.28 0.69 1 1 1 0 1 C26 C26EN_Z317215770 EN_Z317215770 −90.28− 90.280.69 1 0.691 1 0 1 1 1 1 0 1

C10C10 EN_Z1424437838EN_Z1424437838 −−105.18105.18 0.750.75 11 11 11 00 11 C10 EN_Z1424437838 −105.18 0.75 1 1 1 0 1 C10 C10EN_Z1424437838 EN_Z1424437838 −105.18−105.18 0.75 1 0.751 1 0 1 1 1 1 0 1

C11C11 EN_Z1445206940EN_Z1445206940 −−107.69107.69 0.720.72 11 11 00 00 11 C11 EN_Z1445206940 −107.69 0.72 1 1 0 0 1 C11 C11EN_Z1445206940 EN_Z1445206940 −107.69−107.69 0.72 1 0.721 0 0 1 1 1 0 0 1

C13C13 EN_Z18696867EN_Z18696867 −−92.0392.03 0.610.61 11 11 11 00 00 C13C13 EN_Z18696867 EN_Z18696867 −92.03− 92.030.61 1 0.611 1 0 1 0 1 1 0 0 C13 EN_Z18696867 −92.03 0.61 1 1 1 0 0

C23C23C23 EN_Z28052152 EN_Z28052152EN_Z28052152 −−75.4575.45− 75.450.630.63 11 0.6311 11 1 1 00 1 1 1 0 C23 EN_Z28052152 −75.45 0.63 1 1 1 1 0 C23 EN_Z28052152 −75.45 0.63 1 1 1 1 0

C31 C31C31VM_STL480883 VM_STL480883VM_STL480883 −−95.2395.23− 95.230.630.63 11 0.6311 11 0 10 11 1 1 0 1 C31 VM_STL480883 −95.23 0.63 1 1 1 0 1 C31 VM_STL480883 −95.23 0.63 1 1 1 0 1

Biomolecules 2021, 11, 929 10 of 14

Table 3. Experimentally validated new PAMs based on LBVS approach a.

cAMP Accumulation Binding ID b pEC50 ± SEM Emax (% WT ) pIC50 ± SEM Span ± SEM C22 6.3 ± 0.29 96.9 ± 4.56 6.2 ± 0.11 −135.7 ± 8.4 C26 5.4 ± 0.31 89.4 ± 3.72 5.85 ± 0.32 −82.09 ± 13.44 C10 3.3 ± 0.47 92.93 ± 4.4 4.34 ± 0.23 76.17 ± 16.21 C11 N.A. c N.A. c 5.05 ± 0.22 −46.37 ± 5.83 C13 4.8 ± 0.35 76.6 ± 6.24 5.83 ± 0.24 61.32 ± 7.59 C23 4.8 ± 0.21 98.5 ± 3.6 N.B. d N.B. d C31 5.0 ± 0.81 91.0 ± 0.95 5.89 ± 0.13 −102.5 ± 6.85 a AllBiomolecules data were 2021 ﬁtted, 11, with x FOR a three-parameterPEER REVIEW logistic curve to obtain pEC50 and pIC50 values. Data represent means ± S.E.M. of at11 least of 15

three independent experiments performed duplicate; b WT, wild-type; c N.A., not active.; d N.B., no binding.

Figure 6.FigureExperimental 6. Experimental validation validation of selected of selected compounds compounds identiﬁed identified using using LBVS LBVS on on GLP-1R. GLP-1R. ((AA)) Binding of of hit hit compounds compounds to GLP-1Rto or GLP-1R its mutants or itsin mutants competition in competition with radiolabeled with radiolabeled GLP-1. GLP-1. (B) cAMP (B) cAMP activity activity induced induced by different by different GLP-1 GLP-1 concentrations concentrations in the presence of hit compounds (20 μM) in GLP-1R expressing CHO-K1 cells. (C) Binding of four hit compounds µ C in the presenceto GLP-1R of mutants hit compounds (C347F and (20 T355A)M) in competition GLP-1R expressing with radiolabeled CHO-K1 GLP-1. cells. (D () Effects) Binding of binding-pocket of four hit compounds mutations to GLP-1R(C347F mutants and (C347F T355A) andon the T355A) alloster inic modulation competition of withselected radiolabeled compounds GLP-1.in cAMP ( Daccumulati) Effectson of elicited binding-pocket by 0.02 nM mutations GLP- (C347F and1. Data T355A) are presented on the allosteric as means modulation± S.E.M. of three of selectedindependent compounds experiments. in cAMP accumulation elicited by 0.02 nM GLP-1. Data are presented as means ± S.E.M. of three independent experiments.

Biomolecules 2021, 11, x FOR PEER REVIEW 12 of 15

Table 3. Experimentally validated new PAMs based on LBVS approach a.

cAMP Accumulation Binding ID pEC50 ± SEM Emax (% WT b) pIC50 ± SEM Span ± SEM C22 6.3 ± 0.29 96.9 ± 4.56 6.2 ± 0.11 −135.7 ± 8.4 C26 5.4 ± 0.31 89.4 ± 3.72 5.85 ± 0.32 −82.09 ± 13.44 C10 3.3 ± 0.47 92.93 ± 4.4 4.34 ± 0.23 76.17 ± 16.21 C11 N.A. c N.A. c 5.05 ± 0.22 −46.37 ± 5.83 C13 4.8 ± 0.35 76.6 ± 6.24 5.83 ± 0.24 61.32 ± 7.59 C23 4.8 ± 0.21 98.5 ± 3.6 N.B. d N.B. d C31 5.0 ± 0.81 91.0 ± 0.95 5.89 ± 0.13 −102.5 ± 6.85 Biomoleculesa All data2021 ,were11, 929 fitted with a three-parameter logistic curve to obtain pEC50 and pIC50 values. Data represent means ± S.E.M.11 of 14 of at least three independent experiments performed duplicate; b WT, wild-type; c N.A., not active.; d N.B., no binding. C22 and C26 were initially predicted to locate at an extra-helical binding site similar to thatC22 of andNNC0640 C26 were and initially PF-06372222. predicted With to excellent locate at PLANTS an extra-helical and IFP binding scores, site they similar were ableto that to penetrate of NNC0640 into and the PF-06372222.TM6-TM7 cleft With usin excellentg one arm PLANTS with the and formation IFP scores, of at they least were two hydrogenable to penetrate bonds (S into6x41b and the TM6-TM7N8x47b), while cleft pointing using oneto the arm TM5-TM6 with the interface formation with of massive at least hydrophobictwo hydrogen interactions bonds (S6x41b usingand another N8x47b arm.), while In addition, pointing there to the is TM5-TM6one additional interface hydrogen with betweenmassive hydrophobicC26 and T355 interactions6x44b. It is interesting using another that the arm. intracellular In addition, region there of is TM6 one appears additional to regulatehydrogen the between functionality C26 and of T355 interacting6x44b. It is ligands: interesting NAMs that theincluding intracellular NNC0640 region and of TM6 PF- 06372222appears torestrict regulate the theoutward functionality movement of interacting of the TM6, ligands: a key feature NAMs of including receptor activation, NNC0640 whereasand PF-06372222 compound restrict 2 and theBETP outward covalently movement bind to C347 of the6x36b TM6, and aact key as ago-allosteric feature of receptor mod- ulatorsactivation, (ago-PAMs) whereas [7,22,53,54]. compound 2To and verify BETP this covalently hypothesis, bind we to mutated C3476x36b twoand key act residues as ago- atallosteric TM6, C347 modulators6x36b and T355 (ago-PAMs)6x44b, and [ 7found,22,53 that,54]. the To verifyeffects thisare variable hypothesis, and we ligand-depend- mutated two ent.key residuesThe compounds at TM6, tested C3476x36b retainedand T355 their6x44b binding, and affinities found that to thethe effects C347F are mutant variable (Figure and 6C),ligand-dependent. which is different The to compounds the abolishment tested shown retained by compound their binding 2 [7], affinities indicating to the that C347F they aremutant not covalently (Figure6C), bound which to isC347 different6x36b, consistent to the abolishment with our observations shown by made compound in the cAMP 2 [ 7], 6x36b accumulationindicating that assay they are(Figure not covalently 6D). bound to C347 , consistent with our observations madeTo in explore the cAMP the accumulationSAR of the nine assay allosteric (Figure 6modulatorsD). discovered in this study (two NAMsTo and explore seven the PAMs), SAR of we the performed nine allosteric a structural modulators similarity discovered search in in the this Chinese study (two Na- tionalNAMs Compound and seven PAMs),Library wedatabase performed and aidentified structural 54 similarity compounds search with in thea high Chinese 2D Tan- Na- imototional Compoundsimilarity (>0.85) Library to databasethe nine anddiscov identifiedered compounds. 54 compounds These with compounds a high 2D were Tanimoto sub- jectedsimilarity to both (>0.85) cAMP to theaccumulation nine discovered and whole-cell compounds. binding These assays compounds (Table 4). were CD3532-B002 subjected andto both CD3559-D005, cAMP accumulation two analogues and whole-cellof C11 without binding one negatively assays (Table charged4). CD3532-B002 atom, displayed and aCD3559-D005, similar enhancement two analogues of ligand of binding C11 without as well one as negatively GLP-1-induced charged intracellular atom, displayed cAMP accumulation.a similar enhancement JK1719-D011 of ligand weakened binding the as ligand well asbinding, GLP-1-induced which is intracellulardifferent from cAMP that observedaccumulation. with its JK1719-D011 analogue C13, weakened especially the considering ligand binding, their whichstructural is different difference from is only that limitedobserved to withthe length its analogue of the C13,linker. especially A similar considering phenomenon their was structural also seen difference with CD1652- is only limited to the length of the linker. A similar phenomenon was also seen with CD1652-A009 A009 and its analogue C26. These results indicate that the subtle difference in the chemical and its analogue C26. These results indicate that the subtle difference in the chemical structures of allosteric modulators may impact their pharmacological profiles. structures of allosteric modulators may impact their pharmacological profiles. Table 4. Experimentally validated analogues of new PAMs a. Table 4. Experimentally validated analogues of new PAMs a. cAMP Accumulation Binding Name Chemical Structure cAMP Accumulation Binding b Name Chemical Structure pEC50 ± SEM Emax (% WT ) pIC50 ± SEM Span ± SEM pEC ± SEM E (% WT b) pIC ± SEM Span ± SEM Biomolecules 2021, 11, x FOR PEER REVIEW 50 max 50 13 of 15 Biomolecules 2021, 11, x FOR PEER REVIEW 13 of 15

Biomolecules 2021, 11, x FOR PEER REVIEW 13 of 15

c c CD3532-B002CD3532-B002 4.8 ±4.80.01 ± 0.01 92.2 92.2± 3.9 ± 3.9 N.B. N.B.c N.B.N.B.c

CD3559-D005 5.1 ± 0.07 78.2 ± 13.6 5.62 ± 0.63 35.25 ± 11.50 CD3559-D005 5.1 ± 0.07 78.2 ± 13.6 5.62 ± 0.63 35.25 ± 11.50 CD3559-D005CD3559-D005 5.1 5.1± 0.07 ± 0.07 78.2 78.2± 13.6 ± 13.6 5.62 5.±620.63 ± 0.63 35.25 35.25± 11.50± 11.50

JK1719-D011 4.2 ± 0.28 126.1 ± 4.2 4.91 ± 0.37 59.69 ± 15.13 JK1719-D011JK1719-D011 4.2 ±4.20.28 ± 0.28 126.1 126.1± 4.2 ± 4.2 4.91 ± 4.0.3791 ± 0.37 59.69 59.69± 15.13 ± 15.13 JK1719-D011 4.2 ± 0.28 126.1 ± 4.2 4.91 ± 0.37 59.69 ± 15.13

CD1652-A009CD1652-A009 4.8 ±4.85.3 ± 5.3 114.9 114.9± 8.6 ± 8.6 5.39 5.39± 0.33 ± 0.33 51.92 51.92± 9.60± 9.60 CD1652-A009 4.8 ± 5.3 114.9 ± 8.6 5.39 ± 0.33 51.92 ± 9.60 CD1652-A009 4.8 ± 5.3 114.9 ± 8.6 5.39 ± 0.33 51.92 ± 9.60

a b c a AllAll data data were were fitted ﬁtted with aathree-parameter three-parameter logistic logistic curve curve to obtain to obtain pEC50 andpEC pIC50 and50 values; pIC50 values;WT, wild-type; b WT, wild-type;N.B., no binding. c N.B., no a All data were fitted with a three-parameter logistic curve to obtain pEC50 and pIC50 values; b WT, wild-type; c N.B., no binding. a Allbinding. data were fitted with a three-parameter logistic curve to obtain pEC50 and pIC50 values; b WT, wild-type; c N.B., no binding. 4. Conclusions 4. Conclusions 4. ConclusionsThe current study was conducted with an aim of developing computational screen- The current study was conducted with an aim of developing computational screening protocols to discover new allosteric modulators of GLP-1R in the lipid environment. ing Theprotocols current to study discover was new conducted allosteric with modulators an aim of of developing GLP-1R in computational the lipid environment. screen- By adopting MD simulation-derived lipids, we successfully built a membrane environ- ingBy protocols adopting to MD discover simulation-derived new allosteric lipids, modulators we successfully of GLP-1R built in the a lipidmembrane environment. environment rich of lipids for molecular docking and SBVS. Through generating a ligand-receptor Byment adopting rich of MD lipids simulation-derived for molecular docking lipids, an dwe SBVS. successfully Through built generati a membraneng a ligand-receptor environ- interaction pattern, LBVS identified seven PAMs. mentinteraction rich of lipidspattern, for LBVS molecular identified docking seven and PAMs. SBVS. Through generating a ligand-receptor interaction pattern, LBVS identified seven PAMs. Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1, Table S1: Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1, Table S1: List of 45 compounds identified using SBVS; Table S2: List of 42 compounds identified using LBVS. SupplementaryList of 45 compounds Materials: identified The following using SBVS; are availableTable S2: onlineList of at42 www.mdpi.com/xxx/s1,compounds identified using Table LBVS. S1: ListAuthor of 45 Contributions: compounds identified Conceptualization, using SBVS; Q.Z., Table S.Z. S2: andList C.d.G.;of 42 compounds methodology, identified Q.Z., C.d.G.,using LBVS. W.S., Author Contributions: Conceptualization, Q.Z., S.Z. and C.d.G.; methodology, Q.Z., C.d.G., W.S., S.Z. and D.Y.; software, Q.Z., M.V., D.Y. and W.G.; validation, W.G., A.D. and X.C.; formal analysis, AuthorS.Z. and Contributions: D.Y.; software, Conceptualization, Q.Z., M.V., D.Y. and Q.Z., W.G.; S.Z. validation, and C.d.G.; W.G., methodology, A.D. and X.C.; Q.Z., formal C.d.G., analysis, W.S., Q.Z., M.V., D.Y. and W.G.; investigation, M.-W.W., S.Z. and C.d.G.; resources, D.Y., M.-W.W., S.Z. S.Z.Q.Z., and M.V., D.Y.; D.Y. software, and W.G.; Q.Z., investigation,M.V., D.Y. and M.-W.W., W.G.; validation, S.Z. and W.G.,C.d.G.; A.D. resources, and X.C.; D.Y., formal M.-W.W., analysis, S.Z. and C.d.G.; data curation, Q.Z., M.V., D.Y. and W.G.; writing—original draft preparation, Q.Z., D.Y. Q.Z.,and M.V.,C.d.G.; D.Y. data and curation, W.G.; investigation,Q.Z., M.V., D.Y. M.-W.W., and W.G. S.Z.; writing—original and C.d.G.; resources, draft preparation, D.Y., M.-W.W., Q.Z., S.Z. D.Y. and W.G.; writing—review and editing, M.-W.W. and S.Z.; visualization, Q.Z., D.Y., W.G. and M.V.; andand C.d.G.; W.G.; datawriting—review curation, Q.Z., and M.V., editing, D.Y. M.-W.W. and W.G. and; writing—original S.Z.; visualization, draft Q.Z. preparation,, D.Y., W.G. Q.Z., and D.Y.M.V.; supervision, D.Y., M.-W.W., S.Z. and C.d.G.; project administration, D.Y.; funding acquisition, D.Y., andsupervision, W.G.; writing—review D.Y., M.-W.W., and S.Z. editing, and C.d.G.; M.-W.W. project and administration,S.Z.; visualization, D.Y.; Q.Z. funding, D.Y., acquisition,W.G. and M.V.; D.Y., M.-W.W., S.Z. and C.d.G. All authors have read and agreed to the published version of the manu- supervision,M.-W.W., S.Z. D.Y., and M.-W.W., C.d.G. All S.Z. auth andors C.d.G.; have project read and administration, agreed to the D.Y.; published funding version acquisition, of the manu-D.Y., script. M.-W.W.,script. S.Z. and C.d.G. All authors have read and agreed to the published version of the manuscript.Funding: This research was partially supported by National Key Research and Development Pro- Funding: This research was partially supported by National Key Research and Development Pro- gram of China grants 2018YFA0507000 (S.Z. and M.-W.W.) and 2016YFC0905900 (S.Z.); National Funding:gram of ThisChina research grants was2018YFA0507000 partially supported (S.Z. and by M.-W.W.)National Key and Research 2016YFC0905900 and Development (S.Z.); National Pro- Mega R&D Program for Drug Discovery grants 2018ZX09711002-002-005 (D.Y.) and 2018ZX09735- gramMega of R&D China Program grants 2018YFA0507000for Drug Discovery (S.Z. grants and 2018ZX09711002-002-005M.-W.W.) and 2016YFC0905900 (D.Y.) and (S.Z.); 2018ZX09735- National 001 (M.-W.W); National Natural Science Foundation of China grants 21704064 (Q.Z.), 81573479 Mega001 (M.-W.W);R&D Program National for Drug Natural Discovery Science grants Foundation 2018ZX09711002-002-005 of China grants 21704064(D.Y.) and (Q.Z.), 2018ZX09735- 81573479 (D.Y.), 81773792 (D.Y.), 81872915 (M.-W.W.), 82073904 (M.-W.W.) and 31971178 (S.Z.); Shanghai Sci- 001(D.Y.), (M.-W.W); 81773792 National (D.Y.), 81872915Natural Science(M.-W.W.), Foundation 82073904 of (M.-W.W.) China grants and 31971178 21704064 (S.Z.); (Q.Z.), Shanghai 81573479 Sci- ence & Technology Development Fund grant 16ZR1407100 (A.D.); Novo Nordisk-CAS Research (D.Y.),ence &81773792 Technology (D.Y.), Development 81872915 (M.-W.W.), Fund grant 82073904 16ZR1407100 (M.-W.W.) (A.D.); and 31971178 Novo Nordisk-CAS (S.Z.); Shanghai Research Sci- Fund grant NNCAS-2017-1-CC (D.Y.) and annual overhead support from ShanghaiTech University enceFund & grantTechnology NNCAS-2017-1-CC Development (D.Y.) Fund and grant annual 16ZR1407100 overhead support(A.D.); Novo from ShanNordisk-CASghaiTech UniversityResearch and Chinese Academy of Sciences. Fundand grantChinese NNCAS-2017-1-CC Academy of Sciences. (D.Y.) and annual overhead support from ShanghaiTech University andInstitutional Chinese Academy Review Board of Sciences. Statement: Not applicable. Institutional Review Board Statement: Not applicable. InstitutionalInformed Consent Review Statement: Board Statement: Not applicable. Not applicable. Informed Consent Statement: Not applicable. InformedData Availability Consent Statement: Statement: Data Not applicable.are contained within the article or supplementary material. Data Availability Statement: Data are contained within the article or supplementary material. DataAcknowledgments: Availability Statement: We thank Data M. Wu are andcontained J. Tong within for their the technical article or assistance. supplementary material. Acknowledgments: We thank M. Wu and J. Tong for their technical assistance. Acknowledgments:Conflicts of Interest: We The thank authors M. Wu declare and J. no Tong conflict for their of inte technicalrest. The assistance. funders had no role in the Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manu- Conflictsdesign of of the Interest: study; inThe the authors collection, declare analyses, no conflict or interpretation of interest. of The data; funders in the writinghad no ofrole the in manu- the script or in the decision to publish the results. designscript ofor the in thestudy; decision in the to collection, publish the analyses, results. or interpretation of data; in the writing of the manuscript or in the decision to publish the results.

Biomolecules 2021, 11, 929 12 of 14

4. Conclusions The current study was conducted with an aim of developing computational screening protocols to discover new allosteric modulators of GLP-1R in the lipid environment. By adopting MD simulation-derived lipids, we successfully built a membrane environment rich of lipids for molecular docking and SBVS. Through generating a ligand-receptor interaction pattern, LBVS identiﬁed seven PAMs.

Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/biom11070929/s1, Table S1: List of 45 compounds identified using SBVS; Table S2: List of 42 compounds identified using LBVS. Author Contributions: Conceptualization, Q.Z., S.Z. and C.d.G.; methodology, Q.Z., C.d.G., W.S., S.Z. and D.Y.; software, Q.Z., M.V., D.Y. and W.G.; validation, W.G., A.D. and X.C.; formal analysis, Q.Z., M.V., D.Y. and W.G.; investigation, M.-W.W., S.Z. and C.d.G.; resources, D.Y., M.-W.W., S.Z. and C.d.G.; data curation, Q.Z., M.V., D.Y. and W.G.; writing—original draft preparation, Q.Z., D.Y. and W.G.; writing—review and editing, M.-W.W. and S.Z.; visualization, Q.Z., D.Y., W.G. and M.V.; supervision, D.Y., M.-W.W., S.Z. and C.d.G.; project administration, D.Y.; funding acquisition, D.Y., M.-W.W., S.Z. and C.d.G. All authors have read and agreed to the published version of the manuscript. Funding: This research was partially supported by National Key Research and Development Program of China grants 2018YFA0507000 (S.Z. and M.-W.W.) and 2016YFC0905900 (S.Z.); National Mega R&D Program for Drug Discovery grants 2018ZX09711002-002-005 (D.Y.) and 2018ZX09735-001 (M.-W.W); National Natural Science Foundation of China grants 21704064 (Q.Z.), 81573479 (D.Y.), 81773792 (D.Y.), 81872915 (M.-W.W.), 82073904 (M.-W.W.) and 31971178 (S.Z.); Shanghai Science & Technology Development Fund grant 16ZR1407100 (A.D.); Novo Nordisk-CAS Research Fund grant NNCAS-2017-1-CC (D.Y.) and annual overhead support from ShanghaiTech University and Chinese Academy of Sciences. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: Data are contained within the article or supplementary material. Acknowledgments: We thank M. Wu and J. Tong for their technical assistance. Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript or in the decision to publish the results.

References 1. Venkatakrishnan, A.J.; Deupi, X.; Lebon, G.; Tate, C.G.; Schertler, G.F.; Babu, M.M. Molecular signatures of G-protein-coupled receptors. Nature 2013, 494, 185–194. [CrossRef] 2. Katritch, V.; Cherezov, V.; Stevens, R.C. Structure-Function of the G Protein-Coupled Receptor Superfamily. Annu. Rev. Pharmacol. Toxicol. 2013, 53, 531–556. [CrossRef] 3. Hauser, A.S.; Chavali, S.; Masuho, I.; Jahn, L.J.; Martemyanov, K.A.; Gloriam, D.E.; Babu, M.M. Pharmacogenomics of GPCR Drug Targets. Cell 2018, 172, 41–54.e19. [CrossRef] 4. Hauser, A.S.; Attwood, M.M.; Rask-Andersen, M.; Schioth, H.B.; Gloriam, D.E. Trends in GPCR drug discovery: New agents, targets and indications. Nat. Rev. Drug Discov. 2017, 16, 829–842. [CrossRef][PubMed] 5. Zhou, Q.; Yang, D.; Wu, M.; Guo, Y.; Guo, W.; Zhong, L.; Cai, X.; Dai, A.; Jang, W.; Shakhnovich, E.I.; et al. Common activation mechanism of class A GPCRs. eLife 2019, 8, e50279. [CrossRef][PubMed] 6. Wootten, D.; Miller, L.J. Structural Basis for Allosteric Modulation of Class B G Protein-Coupled Receptors. Annu. Rev. Pharmacol. Toxicol. 2020, 60, 89–107. [CrossRef][PubMed] 7. Song, G.; Yang, D.; Wang, Y.; de Graaf, C.; Zhou, Q.; Jiang, S.; Liu, K.; Cai, X.; Dai, A.; Lin, G.; et al. Human GLP-1 receptor transmembrane domain structure in complex with allosteric modulators. Nature 2017, 546, 312–315. [CrossRef] 8. Bueno, A.B.; Sun, B.; Willard, F.S.; Feng, D.; Ho, J.D.; Wainscott, D.B.; Showalter, A.D.; Vieth, M.; Chen, Q.; Stutsman, C.; et al. Structural insights into probe-dependent positive allosterism of the GLP-1 receptor. Nat. Chem. Biol. 2020, 16, 1105–1110. [CrossRef] 9. Hua, T.; Li, X.; Wu, L.; Iliopoulos-Tsoutsouvas, C.; Wang, Y.; Wu, M.; Shen, L.; Brust, C.A.; Nikas, S.P.; Song, F.; et al. Activation and Signaling Mechanism Revealed by Cannabinoid Receptor-Gi Complex Structures. Cell 2020, 180, 655–665.e18. [CrossRef] [PubMed] Biomolecules 2021, 11, 929 13 of 14

10. Shao, Z.; Yan, W.; Chapman, K.; Ramesh, K.; Ferrell, A.J.; Yin, J.; Wang, X.; Xu, Q.; Rosenbaum, D.M. Structure of an allosteric modulator bound to the CB1 cannabinoid receptor. Nat. Chem. Biol. 2019, 15, 1199–1205. [CrossRef][PubMed] 11. Liu, X.; Masoudi, A.; Kahsai, A.W.; Huang, L.Y.; Pani, B.; Staus, D.P.; Shim, P.J.; Hirata, K.; Simhal, R.K.; Schwalb, A.M.; et al. Mechanism of beta2AR regulation by an intracellular positive allosteric modulator. Science 2019, 364, 1283–1287. [CrossRef] [PubMed] 12. Hori, T.; Okuno, T.; Hirata, K.; Yamashita, K.; Kawano, Y.; Yamamoto, M.; Hato, M.; Nakamura, M.; Shimizu, T.; Yokomizo, T.; et al. Na(+)-mimicking ligands stabilize the inactive state of leukotriene B4 receptor BLT1. Nat. Chem. Biol. 2018, 14, 262–269. [CrossRef][PubMed] 13. Wold, E.A.; Chen, J.; Cunningham, K.A.; Zhou, J. Allosteric Modulation of Class A GPCRs: Targets, Agents, and Emerging Concepts. J. Med. Chem. 2019, 62, 88–127. [CrossRef] 14. Chen, D.; Liao, J.; Li, N.; Zhou, C.; Liu, Q.; Wang, G.; Zhang, R.; Zhang, S.; Lin, L.; Chen, K.; et al. A nonpeptidic agonist of glucagon-like peptide 1 receptors with efficacy in diabetic db/db mice. Proc. Natl. Acad. Sci. USA 2007, 104, 943–948. [CrossRef] 15. Zhao, P.; Liang, Y.L.; Belousoff, M.J.; Deganutti, G.; Fletcher, M.M.; Willard, F.S.; Bell, M.G.; Christe, M.E.; Sloop, K.W.; Inoue, A.; et al. Activation of the GLP-1 receptor by a non-peptidic agonist. Nature 2020, 577, 432–436. [CrossRef] 16. Saxena, A.R.; Gorman, D.N.; Esquejo, R.M.; Bergman, A.; Chidsey, K.; Buckeridge, C.; Griffith, D.A.; Kim, A.M. Danuglipron (PF-06882961) in type 2 diabetes: A randomized, placebo-controlled, multiple ascending-dose phase 1 trial. Nat. Med. 2021, 27, 1079–1087. [CrossRef] 17. Ma, H.; Huang, W.; Wang, X.; Zhao, L.; Jiang, Y.; Liu, F.; Guo, W.; Sun, X.; Zhong, W.; Yuan, D.; et al. Structural insights into the activation of GLP-1R by a small molecule agonist. Cell Res. 2020, 30, 1140–1142. [CrossRef] 18. Baek, M.; Kim, D.; Kim, C.H.; Kim, D.Y.; Choi, H.S.; Kang, S.J.; Lee, H.; Lee, K. 939-P: HD-7671, a Nonpeptide Small Molecule for Orally Available GLP-1 Receptor Agonist, Improves Glucose Tolerance by Increasing Insulin Secretion. Diabetes 2020, 69 (Suppl. 1), 939-P. [CrossRef] 19. Hollenstein, K.; Kean, J.; Bortolato, A.; Cheng, R.K.; Dore, A.S.; Jazayeri, A.; Cooke, R.M.; Weir, M.; Marshall, F.H. Structure of class B GPCR corticotropin-releasing factor receptor 1. Nature 2013, 499, 438–443. [CrossRef] 20. Liang, Y.L.; Khoshouei, M.; Deganutti, G.; Glukhova, A.; Koole, C.; Peat, T.S.; Radjainia, M.; Plitzko, J.M.; Baumeister, W.; Miller, L.J.; et al. Cryo-EM structure of the active, Gs-protein complexed, human CGRP receptor. Nature 2018, 561, 492–497. [CrossRef] 21. Jazayeri, A.; Dore, A.S.; Lamb, D.; Krishnamurthy, H.; Southall, S.M.; Baig, A.H.; Bortolato, A.; Koglin, M.; Robertson, N.J.; Errey, J.C.; et al. Extra-helical binding site of a glucagon receptor antagonist. Nature 2016, 533, 274–277. [CrossRef][PubMed] 22. Nolte, W.M.; Fortin, J.P.; Stevens, B.D.; Aspnes, G.E.; Griffith, D.A.; Hoth, L.R.; Ruggeri, R.B.; Mathiowetz, A.M.; Limberakis, C.; Hepworth, D.; et al. A potentiator of orthosteric ligand activity at GLP-1R acts via covalent modification. Nat. Chem. Biol. 2014, 10, 629–631. [CrossRef][PubMed] 23. Wootten, D.; Simms, J.; Miller, L.J.; Christopoulos, A.; Sexton, P.M. Polar transmembrane interactions drive formation of ligand- specific and signal pathway-biased family B G protein-coupled receptor conformations. Proc. Natl. Acad. Sci. USA 2013, 110, 5211–5216. [CrossRef] 24. Smelcerovic, A.; Lazarevic, J.; Tomovic, K.; Anastasijevic, M.; Jukic, M.; Kocic, G.; Anderluh, M. An Overview, Advantages and Therapeutic Potential of Nonpeptide Positive Allosteric Modulators of Glucagon-Like Peptide-1 Receptor. ChemMedChem 2019, 14, 514–521. [CrossRef] 25. Willard, F.S.; Bueno, A.B.; Sloop, K.W. Small molecule drug discovery at the glucagon-like peptide-1 receptor. Exp. Diabetes Res. 2012, 2012, 709893. [CrossRef][PubMed] 26. Mendez, M.; Matter, H.; Defossa, E.; Kurz, M.; Lebreton, S.; Li, Z.; Lohmann, M.; Lohn, M.; Mors, H.; Podeschwa, M.; et al. Design, Synthesis, and Pharmacological Evaluation of Potent Positive Allosteric Modulators of the Glucagon-like Peptide-1 Receptor (GLP-1R). J. Med. Chem. 2020, 63, 2292–2307. [CrossRef][PubMed] 27. Redij, T.; Ma, J.; Li, Z.; Hua, X.; Li, Z. Discovery of a potential positive allosteric modulator of glucagon-like peptide 1 receptor through virtual screening and experimental study. J. Comput. Aided Mol. Des. 2019, 33, 973–981. [CrossRef] 28. Qin, S.; Meng, M.; Yang, D.; Bai, W.; Lu, Y.; Peng, Y.; Song, G.; Wu, Y.; Zhou, Q.; Zhao, S.; et al. High-throughput identification of G protein-coupled receptor modulators through affinity mass spectrometry screening. Chem. Sci. 2018, 9, 3192–3199. [CrossRef] 29. Wan, F.; Zhu, Y.; Hu, H.; Dai, A.; Cai, X.; Chen, L.; Gong, H.; Xia, T.; Yang, D.; Wang, M.W.; et al. DeepCPI: A Deep Learning-based Framework for Large-scale in silico Drug Screening. Genom. Proteom. Bioinform. 2019, 17, 478–495. [CrossRef][PubMed] 30. O’Brien, A.; Andrews, S.P.; Baig, A.H.; Bortolato, A.; Brown, A.J.H.; Brown, G.A.; Brown, S.H.; Christopher, J.A.; Congreve, M.; Cooke, R.M.; et al. Identification of a novel allosteric GLP-1R antagonist HTL26119 using structure- based drug design. Bioorg. Med. Chem. Lett. 2019, 29, 126611. [CrossRef] 31. Redij, T.; Chaudhari, R.; Li, Z.; Hua, X.; Li, Z. Structural Modeling and in Silico Screening of Potential Small-Molecule Allosteric Agonists of a Glucagon-like Peptide 1 Receptor. ACS Omega 2019, 4, 961–970. [CrossRef] 32. Stein, R.M.; Kang, H.J.; McCorvy, J.D.; Glatfelter, G.C.; Jones, A.J.; Che, T.; Slocum, S.; Huang, X.P.; Savych, O.; Moroz, Y.S.; et al. Virtual discovery of melatonin receptor ligands to modulate circadian rhythms. Nature 2020, 579, 609–614. [CrossRef] 33. Weiss, D.R.; Ahn, S.; Sassano, M.F.; Kleist, A.; Zhu, X.; Strachan, R.; Roth, B.L.; Lefkowitz, R.J.; Shoichet, B.K. Conformation guides molecular efficacy in docking screens of activated beta-2 adrenergic G protein coupled receptor. ACS Chem. Biol. 2013, 8, 1018–1026. [CrossRef] Biomolecules 2021, 11, 929 14 of 14

34. Meanwell, N.A. Synopsis of some recent tactical application of bioisosteres in drug design. J. Med. Chem. 2011, 54, 2529–2591. [CrossRef][PubMed] 35. CH, S. Conformations and 3D pharmacophore searching. Drug Discov. Today Technol. 2010, 7, e203–e270. 36. Goodsell, D.S.; Zardecki, C.; Di Costanzo, L.; Duarte, J.M.; Hudson, B.P.; Persikova, I.; Segura, J.; Shao, C.; Voigt, M.; Westbrook, J.D.; et al. RCSB Protein Data Bank: Enabling biomedical research and drug discovery. Protein Sci. 2020, 29, 52–65. [CrossRef] [PubMed] 37. Sastry, G.M.; Adzhigirey, M.; Day, T.; Annabhimoju, R.; Sherman, W. Protein and ligand preparation: Parameters, protocols, and influence on virtual screening enrichments. J. Comput. Aided Mol. Des. 2013, 27, 221–234. [CrossRef][PubMed] 38. Liu, C.H.; Gong, Z.; Liang, Z.L.; Liu, Z.X.; Yang, F.; Sun, Y.J.; Ma, M.L.; Wang, Y.J.; Ji, C.R.; Wang, Y.H.; et al. Arrestin-biased AT1R agonism induces acute catecholamine secretion through TRPC3 coupling. Nat. Commun. 2017, 8, 14335. [CrossRef][PubMed] 39. Cordomi, A.; Caltabiano, G.; Pardo, L. Membrane Protein Simulations Using AMBER Force Field and Berger Lipid Parameters. J. Chem. Theory Comput. 2012, 8, 948–958. [CrossRef][PubMed] 40. Sousa da Silva, A.W.; Vranken, W.F. ACPYPE—AnteChamber PYthon Parser interfacE. BMC Res. Notes 2012, 5, 367. [CrossRef] 41. Klauda, J.B.; Venable, R.M.; Freites, J.A.; O’Connor, J.W.; Tobias, D.J.; Mondragon-Ramirez, C.; Vorobyov, I.; MacKerell, A.D., Jr.; Pastor, R.W. Update of the CHARMM all-atom additive force field for lipids: Validation on six lipid types. J. Phys. Chem. B 2010, 114, 7830–7843. [CrossRef][PubMed] 42. Pronk, S.; Pall, S.; Schulz, R.; Larsson, P.; Bjelkmar, P.; Apostolov, R.; Shirts, M.R.; Smith, J.C.; Kasson, P.M.; van der Spoel, D.; et al. GROMACS 4.5: A high-throughput and highly parallel open source molecular simulation toolkit. Bioinformatics 2013, 29, 845–854. [CrossRef][PubMed] 43. Hess, B.; Bekker, H.; Berendsen, H.J.C.; Fraaije, J.G.E.M. LINCS: A linear constraint solver for molecular simulations. J. Comput. Chem. 1997, 18, 1463–1472. [CrossRef] 44. Halgren, T.A.; Murphy, R.B.; Friesner, R.A.; Beard, H.S.; Frye, L.L.; Pollard, W.T.; Banks, J.L. Glide: A new approach for rapid, accurate docking and scoring. 2. Enrichment factors in database screening. J. Med. Chem. 2004, 47, 1750–1759. [CrossRef] [PubMed] 45. Genheden, S.; Ryde, U. The MM/PBSA and MM/GBSA methods to estimate ligand-binding affinities. Expert Opin. Drug Discov. 2015, 10, 449–461. [CrossRef][PubMed] 46. Korb, O.; Stutzle, T.; Exner, T.E. Empirical scoring functions for advanced protein-ligand docking with PLANTS. J. Chem. Inf. Model. 2009, 49, 84–96. [CrossRef] 47. Marcou, G.; Rognan, D. Optimizing fragment and scaffold docking by use of molecular interaction fingerprints. J. Chem. Inf. Model. 2007, 47, 195–207. [CrossRef] 48. de Graaf, C.; Kooistra, A.J.; Vischer, H.F.; Katritch, V.; Kuijer, M.; Shiroishi, M.; Iwata, S.; Shimamura, T.; Stevens, R.C.; de Esch, I.J.; et al. Crystal structure-based virtual screening for fragment-like ligands of the human histamine H(1) receptor. J. Med. Chem. 2011, 54, 8195–8206. [CrossRef] 49. Kooistra, A.J.; Vischer, H.F.; McNaught-Flores, D.; Leurs, R.; de Esch, I.J.; de Graaf, C. Function-specific virtual screening for GPCR ligands using a combined scoring method. Sci. Rep. 2016, 6, 28288. [CrossRef] 50. Korb, O.; Stützle, T.; Exner, T.E. An ant colony optimization approach to flexible protein–ligand docking. Swarm Intell. 2007, 1, 115–134. [CrossRef] 51. Korb, O.; Stutzle, T.; Exner, T.E. PLANTS: Application of ant colony optimization to structure-based drug design. Lect. Notes Comput. Sc. 2006, 4150, 247–258. 52. Baell, J.B.; Holloway, G.A. New substructure filters for removal of pan assay interference compounds (PAINS) from screening libraries and for their exclusion in bioassays. J. Med. Chem. 2010, 53, 2719–2740. [CrossRef] 53. Graaf, C.; Donnelly, D.; Wootten, D.; Lau, J.; Sexton, P.M.; Miller, L.J.; Ahn, J.M.; Liao, J.; Fletcher, M.M.; Yang, D.; et al. Glucagon- Like Peptide-1 and Its Class B G Protein-Coupled Receptors: A Long March to Therapeutic Successes. Pharmacol. Rev. 2016, 68, 954–1013. [CrossRef][PubMed] 54. Cong, Z.T.; Chen, L.N.; Ma, H.L.; Zhou, Q.T.; Zou, X.Y.; Ye, C.Y.; Dai, A.T.; Liu, Q.; Huang, W.; Sun, X.Q.; et al. Molecular insights into ago-allosteric modulation of the human glucagon-like peptide-1 receptor. Nat. Commun. 2021.[CrossRef]