Enhancing drug residence time by shielding of intra-protein hydrogen bonds: a case study on CCR2 antagonists. Supplementary Information

Transcription

1 Enhancing drug residence time by shielding of intra-protein hydrogen bonds: a case study on CCR2 antagonists Aniket Magarkar, Gisela Schnapp, Anna-Katharina Apel, Daniel Seeliger & Christofer S. Tautermann * Medicinal Chemistry, Boehringer Ingelheim Pharma GmbH & Co. KG, Birkendorfer Str. 65, D Biberach a.d. Riss, Germany KEYWORDS: CCR2, GPCR, water networks, drug residence time, buried hydrogen bonds Supplementary Information

2 Results: Section : Analysis of hydrogen bond in CCR2 and ligands MK-82 & 5a. A: Hydrogen bond : Glu29and MK82 Number of H-bonds Tme (ns) Number of H-bonds B: Hydrogen bond : Glu29 and 5a Tme (ns) Supplementary Information Figure : Interactions of ligands in the CCR2 binding pocket: The salt bridge between Glu29 and the ligands remains stable throughout the simulation. Section 2: Comparison of binding site occupied volume vs time for MK-82, 5a and 8 In order to understand the behavior of ligands MK-82, 5a and 8, we analyzed the volume occupied by ligands in the CCR2 binding site. First we calculated distance in the center of mass of protein and center of mass of ligand, as shown in Supporting Information Figure 2A, all of the three ligands are present in the binding site, throughout the simulation. (The error bars denote standard deviation calculated from all simulations). Then we plotted the volume of the ligands in CCR2 binding site vs time for three ligands. As can be seen in supporting Information Figure 2B, 5a occupies higher volume in binding site as compared to 8 and MK-82. Comparing volume occupied by 5a to 8, where the only difference in the structure is absence of Br in 8, it can be concluded that, higher volume of 5a is responsible for the broken water network in the CCR2 binding site. This disruption of water network in turn imparts rigidity to CCR2 binding site rigidity and slower off rate of for 5a.

3 B. MK-82 Ligand 5a Ligand 8 Volume (nm 3 ) Time (ns) Supporting Information Figure 2: (A) Distance between center of mass of protein and ligands (B) Occupied volume by ligands in the CCR2 binding site vs time. Section 3: Biased simulation In order to validate the hypothesis that extra water density is responsible for forming a dynamic H-bonding network and repeatedly disrupting the Tyr2-Glu29 H-bond, we set up a biased simulation, where the water molecules present near Tyr2, for MK-82 and 8 are treated in a different way than rest of the waters. The partial charges on these extra waters are set to zero (shown in Supporting Information Figure 3B in grayscale), to ensure that these water molecules will not be able to participate in H-bonding with other water molecules as well as Tyr2. The MK-82 and 8 systems are then simulated for 5 ns with 5 replicates, with position restraints in z-direction (perpendicular to the membrane) on selected water molecule s oxygen atoms as shown in Supporting Information Figure 3. The position restraints applied (see method section) ensure that the selected water molecules can t be replaced by other normal water molecules, which have H-bonding abilities. Analysis of these trajectories showed that the Tyr2-Glu29 H-bond remained stable throughout the simulation time. This finding allows the interpretation that the Tyr2-Glu29 interaction is usually part of a highly dynamic H-bond network, which loses all dynamics when taking out the additional water molecule.

4 A Tyr2 Glu29 Charged waters B Tyr2 Glu29 Supporting Information Figure 3: Effect of water on the Tyr2-Glu29 H-bond (A) For MK-82 and 8, water molecules form a network of H-bonds, which disrupts the Tyr2-Glu29 interaction (B) When the charges on these molecules were turned off, they are no longer able to influence Tyr2-Glu29 interaction and Tyr2-Glu29 H-bonds are stable throughout 5 ns of simulations for MK-82 (C) and 8 (D) Section 4: Choice of compound 8 over compound 5b for hypothesis validation In order to test the our hypothesis that, extra water density near Tyr2 as shown in Figure 3A (main text) (highlighted in the green circle), is responsible for short RT of MK-82, by disrupting the rigidity of the binding pocket and keeping it flexible. In order to test this hypothesis with further simulations, we pick an additional ligand from the study by Vilums et al. with following criteria: ) chemical structure should be very similar to 5a, 2) CCR2 affinity should be very similar to 5a and 3) RT should be much shorter than for 5a, if possible closer to the RT of MK-82. Compound 8 (Figure ) from their dataset is chosen as it fulfills all of the above-mentioned criteria as the only change in structure is the deletion of the Br in 5a. Although 5b, a diastereomer of 5a has much shorter residence time as compared to 5a, fulfills the criteria no., the difference in affinity is much larger as compared to that of 5a. For this reason, ligand 8 was chosen over 5b. Section 5: Methods Number of H-bonds Initial system preparation: C MK-82 Tyr2-Glu29 H-bond Time (ns) Uncharged waters In total we have performed MD simulations for 5 systems as shown in Supporting Information Table. To start with, the recently solved structure of CCR2/MK-82, (PDB id: 6GPS) was used as a starting point with minor modifications, to resemble the wild type structure of CCR2. Specifically the rubredoxin fusion on intracellular loop 3 was manually removed, still keeping residues R23 and R24. The missing loop between these residues was modeled and energy minimized employing standard homology modeling tools as implemented in MOE (Molecular Operating Environment (MOE), 28.; Chemical Computing Group). Also the Asn75 mutant introduced for thermostabilisation was back-mutated to Gly D 8 Tyr2-Glu29 H-bond Time (ns)

5 to match the wild-type form of CCR2. The final structure was protonated by the Protonate3D procedure as implemented in MOE. Based on the structural similarity of ligands, 5a and 8 with MK-82, they were manually placed in the binding pocket to obtain initial configurations for the MD systems. Tethered energy minimization protocols were applied (first minimization: only H atoms relaxed, second minimization: tethering heavy atoms to.5, third minimization: tethering the protein backbone to.5 all with the Amber:EHT force field as implemented in MOE) to relax ligand poses and remove minor clashes. These structures were then embedded in a membrane bilayer consisting of 86 POPC molecules with 82 water molecules, and Cl - ions were added to neutralize the charges. The system was then equilibrated in following 2 stages. ) Position restraints on protein heavy atoms (fc= KJ mol - nm - ) and ligand (fc= KJ mol - nm - ) and lipid headgroup (fc= KJ mol - nm - ) for ns 2) Position restraints on protein backbone (fc= KJ mol - nm - ) for ns Following this procedure, as shown in supporting table ns or 5 ns of production run was performed for the respective systems. Simulation parameters: All simulations were performed with the GROMACS 5..4 package 2. The protein was parameterized using the AMBER ff99sb-ildn 3 force field, water molecules were described with the SPC/E 4 watermodel, and for lipids S-lipids 5 parameters were used. Newton s equations of motion were integrated by employing the leap-frog algorithm with a time step of 4 fs with virtual sites 6,7. A cutoff of. nm was applied to short-range electrostatic interactions while long-range electrostatics was calculated using the particle mesh Ewald method 8. Van der Waals interactions were truncated at. nm. Covalent bonds containing hydrogen atoms were constrained by the LINCS algorithm 9, and water molecules were held rigid by the SETTLE algorithm. The temperature of the system was maintained at 3 K by the velocity rescaling thermostat and the Parrinello Rahman barostat was used to keep the system pressure at bar. The water network analysis was performed by in-house developed protocol as described in previous studies 2. All visualizations were performed in VMD 3. For the biased system setup for systems 4 and 5, charges on the selected waters were set to by defining a variant of SPC/E water molecules in a separate topology file. A force constant of 5 KJ mol - nm - were applied in z-direction in order to waters to maintain the position and not to be replaced by normal SPC/E water molecules. For H-bond analysis Gromacs analysis tools were used with default parameters, that is, distance cutoff of.35nm. System Ligand Replicas Simulation Length MK-82 5 Unbiased ns 2 5a 5 Unbiased ns Unbiased ns 4 MK-82 5 Biased (uncharged waters) 5 ns Biased (uncharged waters) 5 ns Supporting Information Table : Summary of simulated systems

6 Section 6: Calculation of free energies ΔG based on K i and residence time Supporting Information Figure 4: Energy profiles of two similar compounds with different K i and residence time values. Assuming Boltzmann equilibrium, the change in the free energy of the bound states of two ligands ( G BS) is directly related to the ratio of their K i-values via: G #$ = R T ln, K.[ligand ] K. [ligand 2] 8 For the change in the activation barrier ( G TS), the deviation of the residence time (RT) from G BS-driven behavior is calculated by: G 9$ = R T ln : K.[ligand ] ] ; R T ln :RT[ligand K. [ligand 2] RT[ligand 2] ; with R being the molar gas constant and T stands for the temperature.

7 References: () Vilums, M.; Zweemer, A. J. M.; Barmare, F.; van der Gracht, A. M. F.; Bleeker, D. C. T.; Yu, Z.; de Vries, H.; Gross, R.; Clemens, J.; Krenitsky, P.; Brussee, J.; Stamos, D.; Saunders, J.; Heitman, L. H.; IJzerman, A. P. When Structure Affinity Relationships Meet Structure Kinetics Relationships: 3- ((Inden--Yl)Amino)--Isopropyl-Cyclopentane--Carboxamides as CCR2 Antagonists. Eur. J. Med. Chem. 25, 93, (2) Pronk, S.; Páll, S.; Schulz, R.; Larsson, P.; Bjelkmar, P.; Apostolov, R.; Shirts, M. R.; Smith, J. C.; Kasson, P. M.; van der Spoel, D.; Hess, B.; Lindahl, E. GROMACS 4.5: a High-Throughput and Highly Parallel Open Source Molecular Simulation Toolkit. Bioinformatics 23, 29 (7), (3) Lindorff-Larsen, K.; Piana, S.; Palmo, K.; Maragakis, P.; Klepeis, J. L.; Dror, R. O.; Shaw, D. E. Improved Side-Chain Torsion Potentials for the Amber ff99sb Protein Force Field. Proteins 2, 78 (8), (4) Berendsen, H. J. C.; Grigera, J. R.; Straatsma, T. P. The Missing Term in Effective Pair Potentials. J. Phys Chem. 987, 9 (24), (5) Jämbeck, J. P. M.; Lyubartsev, A. P. Another Piece of the Membrane Puzzle: Extending Slipids Further. J Chem. Theory Comput. 22, 9 (), (6) Feenstra, K. A.; Hess, B.; Berendsen, H. J. C. Improving Efficiency of Large Time-Scale Molecular Dynamics Simulations of Hydrogen-Rich Systems. J. Comp. Chem. 999, 2 (8), (7) Melcr, J.; Bonhenry, D.; Timr, Š.; Jungwirth, P. Transmembrane Potential Modeling: Comparison Between Methods of Constant Electric Field and Ion Imbalance. J Chem. Theory Comput. 26, 2 (5), (8) Essmann, U.; Perera, L.; Berkowitz, M. L.; Darden, T.; Lee, H.; Pedersen, L. G. A Smooth Particle Mesh Ewald Method. J. Chem. Phys. 995, 3 (9), (9) Hess, B.; Bekker, H.; Berendsen, H. LINCS: a Linear Constraint Solver for Molecular Simulations. J. Comput. Chem. 997, 8 (2), () Jämbeck, J. P. M.; Lyubartsev, A. P. An Extension and Further Validation of an All-Atomistic Force Field for Biological Membranes. J. Chem. Theory Comput. 22, 8 (8), () Parrinello, M.; Rahman, A. Polymorphic Transitions in Single Crystals: a New Molecular Dynamics Method. J. Appl. Phy. 98, 52 (2), (2) Tautermann, C. S.; Seeliger, D.; Kriegl, J. M. What Can We Learn From Molecular Dynamics Simulations for GPCR Drug Design? Comput. Struct. Biotechnol. J. 25, 3, 2. (3) Humphrey, W.; Dalke, A.; Schulten, K. VMD: Visual Molecular Dynamics. J. Mol. Graph. 996, 4 (),