Analysis by metadynamics simulation of binding pathway of influenza virus M2 channel blockers

ABSTRACT

M2 protein of influenza A virus is a proton channel spanning the viral envelope. Activity of this proton channel is required for uncoating of viral particles and equilibrating the pH across the trans Golgi apparatus, which prevents conformational change in hemagglutinin. Amantadine, an anti‐influenza A virus drug, inhibits M2 proton channel activity by binding to the channel pore; however, most currently circulating influenza A viruses are amantadine‐resistant. The most prevalent resistant mutation is a substitution from Ser31 to Asn31 in M2. Further atomistic analysis of ligand‐M2 complexes is needed to provide new approaches for the design of novel M2 channel blockers. Here, the free energy profiles of the binding kinetics of M2 channel blockers were examined by well‐tempered metadynamics simulations and it was found that amantadine first binds to Asp24 of S31 M2 and forms a metastable conformation. In contrast, the free energy profiles of adamantyl bromothiophene dual inhibitor with either S31 M2 or N31 M2 are broad funnel‐shaped curves, suggesting that adamantyl bromothiophene does not form metastable complexes with M2. The trajectory of well‐tempered metadynamics simulations revealed that steric hindrance between adamantyl bromothiophene and S31 M2 interrupts formation of a metastable conformation at Asp24 and that a halogen bond between the bromine atom and N31 is responsible for pulling down the ligand to the channel pore of N31 M2 in the absence of a metastable state. Binding pathways of M2 channel blockers to M2 are here proposed on the basis of these findings; they may provide new approaches to designing further M2 channel blockers.

Abbreviations

L1

metastable state

L2

local minimum

MD

molecular dynamics

NMR

nuclear magnetic resonance

POPC

1‐palmitoyl‐2‐oleoyl‐sn‐glycero‐3‐phosphocholine

RMSD

root mean‐square deviation

TS

transition state

INTRODUCTION

Influenza viruses are enveloped negative‐stranded RNA viruses with segmented genomes. After endocytosis of viral particles, the low pH of the endosomes activates M2 viral protein, which is a proton channel spanning the viral envelope 1, acidifying the interior of viral particles, which results in their uncoating and release of the viral genome into the cytoplasm 2. This proton channel activity is also required for equilibration of the intralumenal pH of the trans Golgi network with that of the cytoplasm, thus preventing conformational change of hemagglutinin to the low‐pH‐induced form 3-5. The M2 transmembrane domain consists of 19 amino acid residues and is assembled in a homo‐tetrameric helical bundle to form the proton channel pore 6. The M2 Val27 residue acts as an N‐terminal channel gate. The main chain carbonyl oxygen atoms of Ala30 and Gly34, and the side chain of Ser31 form hydrogen bonds with the water molecules in the channel pore 7, 8. The M2 His37 residue functions as a pH sensor 9 and selectivity filter that allows conduction of protons but not of other cations 10, 11 and the indole side chain of Trp41 acts as a channel gate in concert with His37 12-15. The channel is closed at high pH because His37 is not charged and Trp41 obstructs the pore near its cytoplasmic end. At low pH, His37 is charged and associates with Trp41 via interaction between the protonated imidazole and the pi electrons of indole, allowing proton flow 16-20.

Amantadine (1‐adamantanamine hydrochloride) is an anti‐influenza A virus drug 21 that inhibits viral replication by binding specifically to the channel pore of M2 and preventing proton conduction 22-24. On the basis of various experimental and computational studies 25, 26, it has been proposed that the hydrophobic adamantane ring interacts with hydrophobic residues lining the channel pore (e.g., Val27 and Ala30) and that the amino group of amantadine orients down toward the C‐terminus and shares water‐bridged hydrogen bonds with the carbonyl oxygen of Ala30 and the imidazole group of His37 7, 8. However, the high prevalence of influenza viruses with amantadine‐resistant mutations in M2 has limited the effectiveness of amantadine 27. A single amino acid substitution of Ser31 to Asn (S31N) in M2 confers resistance to amantadine and is currently the most prevalent resistant mutation in influenza A 28. Conventional MD and stirred MD simulation studies have shown that, although amantadine specifically binds to S31 M2 through amino acid residues such as Val27, Ala30 and His37, the amino group of amantadine oppositely faces up toward the N‐terminus and does not have specific binding residues in the channel pore of N31 M2; thus, water molecules occupy space in the channel pore and proton conduction is not impaired 29-31. Conventional and multiple‐walkers well‐tempered metadynamics simulations have revealed that amantadine transiently binds to Asp24 during the unbinding pathway of amantadine from the channel pore of S31 M2 32, suggesting that Asp24 is one of the metastable binding sites located at the entrances of channel pores. However, the details of the kinetics of binding of amantadine to M2 remain unclear.

Amantadine derivatives are thought to be high‐profile drug candidates because M2 is a proven drug target and essential for influenza virus replication. Recently, adamantyl bromothiophene ((N‐[5‐bromothiophen‐2yl]methyl) amandatan‐1‐amine) was reported to be a dual inhibitor, targeting both S31 M2 and N31 M2. Although the inhibitory effect of adamantyl bromothiophene is relatively weak (EC₅₀ = 4.6 μM for S31, EC₅₀ = 1.8 μM for N31) compared with amantadine (EC₅₀ = 0.3 μM for S31) 33, adamantyl bromothiophene reportedly presents a higher genetic barrier to drug resistance than amantadine 34. Solution NMR structures revealed that the adamantane ring in adamantyl bromothiophene is oriented up toward the N‐terminus of S31 M2, but down toward the C‐terminus of N31 M2 33. In S31 M2, the bromothiophene moiety is close to Ala30 and Ile33, and the adamantane ring close to Val27 residues, whereas in N31 M2, the bromothiophene group is close to the Val27 methyl groups and the adamantane ring is near Ile33. Conventional MD simulations have shown that adamantyl bromothiophene stably interacts with S31 M2 and N31 M2 channel pores within the 200 ns simulation time as observed in the solution NMR structure 33, suggesting that the orientation of adamantyl bromothiophene may be determined during its binding process with M2.

Molecular dynamics simulation has provided insights into the structure and dynamics of ligand–protein complexes at the atomistic level, thus facilitating focusing on the affinity and selectivity of a ligand for its target when designing drugs. However, it is known that binding of ligands to proteins is often restricted by metastable states that are separated by high free energy barriers from stable states, leading to kinetic bottlenecks 35, 36. Here, we examined the entry mechanism of M2 channel blockers into both S31 M2 and N31 M2 proteins by well‐tempered metadynamics simulation. Metadynamics simulation accelerates exploration of new binding pathways by pushing the ligand away from local free energy minima according to history‐dependent bias potentials.

The binding trajectories of amantadine in the S31 M2 channel pore indicate that amantadine forms a metastable binding state with Asp24. In contrast, such an apparently separated free energy state has not been observed for adamantyl bromothiophene in complexes with either S31 M2 or N31 M2. Further, we found that the hydrogen bond between the amino group of adamantyl bromothiophene and Asn31 is required for the flipping of adamantyl bromothiophene in N31 M2.

MATERIALS AND METHODS

Initial M2 models

Initial structures of apo form of S31 M2 were modeled from S31 M2‐amantadine complex (PDB: 2KQT) 37 and S31 M2‐adamantyl bromothiophene complex (PDB: 2MUW) 33 by removing the cognate ligands. Ser31 in the 2KQT‐ and 2MUW‐based models was replaced with Asn for N31 M2 initial models using the molecular builder program in Maestro (Schrödinger, LLC, New York, NY, USA). To incorporate the chloride ion, 2KQT‐ and 2MUW‐based S31 M2 and N31 M2 models were superposed on the x‐ray structure of M2 with chloride ion (PDB: 5C02) 38. All initial structures with chloride ion were refined for well‐tempered metadynamics simulations using the Protein Preparation Wizard Script within Maestro.

Ligand binding simulation using well‐tempered metadynamics simulations

Binding of amantadine and adamantyl bromothiophene to S31 M2 and N31 M2 was examined by ten 100 ns independent well‐tempered metadynamics simulations using Desmond 39 version 2.3 (Schrödinger, LLC). Metadynamics simulation is a widely used enhanced sampling method that allows sampling of free energy landscapes. In this simulation, biasing collective variables were defined as the distance between the center of mass of the ligand molecule and the ligand‐binding residues; namely, Val27, Ala30, Ser31/Asn31 and Gly34 for amantadine binding simulation and Val27 and Ala30 for adamantyl bromothiophene binding simulation. The initial Gaussian hill height and well‐tempered parameters were set at 0.1 and 2.4 kcal/mol, respectively. An Optimized Potentials for Liquid Simulations (OPLS3) force field was used for the simulations. Each M2 structure was placed in a large POPC bilayer and TIP3P water molecules were solvated with 0.15 M NaCl. Ligand molecules in the extracellular solvent region were placed 15 Å far from the center of mass of the ligand‐binding residues. The initial internal conformations of amantadine and adamantyl bromothiophene were based on the NMR structures of 2KQT and 2MUW, respectively. After minimization and relaxation of the model, MD phase production was performed for 100 ns simulations in the isothermal–isobaric (NPT) ensemble at 300 K and 1 bar using a Langevin thermostat. Long‐range electrostatic interactions were computed using the Smooth Particle Mesh Ewald method. All system setups were performed using Maestro (Schrödinger, LLC). The data were processed utilizing Metadynamics Analysis and Simulation Interaction Diagram Tools (Schrödinger, LLC) to calculate free energies and protein‐ligand interactions.

RESULTS

Comparison between well‐tempered metadynamics simulations and NMR structures for ligand positions in M2 channels

To examine the binding kinetics of M2 channel blockers, the trajectories of amantadine and adamantyl bromothiophene during entry into the channel pore of M2 were analyzed. Well‐tempered metadynamics simulations of amantadine with either S31 M2 (PDB 2KQT) or N31 M2, and adamantyl bromothiophene with either S31 M2 (PDB 2MUW) or N31 M2 were performed in POPC phospholipid bilayers. The ligand positions in the channel pore after the well‐tempered metadynamics simulation were compared to each ligand position in the original NMR structure by calculating the RMSD (Fig. 1a–d). Replica simulations that showed smaller RMSDs in 10 runs are shown by the solid lines in Figure 1a–d, and were used as representative replicas in further analyses.

Figure 1

RMSDs of well‐tempered metadynamics simulation structures of amantadine and adamantyl bromothiophene in M2 channel pores. The ligand positions in the channel pores observed in the well‐tempered metadynamics simulations were compared with their positions in (a) the original PDB structures, (b) 2KQT, (c) 2MUW, and (d) 2MUV by calculating the RMSD. Replicas that showed smaller RMSDs in 10 runs are shown as solid lines and the others as dotted lines.

Well‐tempered metadynamics studies of M2‐amantadine complexes

Binding free energy profiles of representative replicas (rep5, rep6 and rep8) were determined from metadynamics simulations of the entry pathway of amantadine into S31 M2 channels (Fig. 2a). The entry pathway was found to consist of three steps; a metastable state (L1), a transition state (TS), and a large local minimum (L2). L1 was about 12 Å (−11 kcal/mol) and L2 about 0.5 Å (−17 kcal/mol) from the amantadine‐binding site of 2KQT PDB structure. These two local free energy minima were separated by a TS region with a relatively high free energy barrier (−4 kcal/mol). Representative snapshots of amantadine‐S31 M2 obtained by well‐tempered metadynamics simulation (rep6) are shown in Figure 2b. Along the entry of amantadine to the channel pore of S31 M2, the positively charged amino group of amantadine first binds to two of the four Asp24 residues via salt bridges. Asp24 guides the ligand to the entrance of the channel pore, and amantadine forms the metastable conformations for several nanoseconds (L1). The hydrophobic interactions between the adamantane ring and Val27 residues transfer amantadine into the channel pore (TS). Finally, amantadine forms stable bindings with the carbonyl oxygen atom and the side chain of Ala30 via hydrogen bonds and hydrophobic interactions (L2).

Figure 2

Well‐tempered metadynamics simulations of amantadine binding in S31 M2. (a) Binding free energy profiles (kcal/mol) of representative replicas as determined by well‐tempered metadynamics simulations of the entry pathway of amantadine into S31 M2. The steps in the entry pathway were designated as follows: L1, a metastable state; TS, a transition state; and L2 a large local minimum. (b) Representative replica (rep6) snapshots from the binding trajectory of amantadine in S31 M2. Asp24 charged residues and Val27 hydrophobic residues are shown in ball and stick models. Amantadine is shown as green stick models, and the M2 backbones as thin tubes. Amantadine from the original PDB structure (2KQT) is depicted by gray lines. The amantadine entry pathway has three steps: L1, two salt bridges (pink dotted lines) between the amantadine amino group and two Asp24 residues bind amantadine; TS, hydrophobic interactions between the adamantane ring and Val27 stabilized amantadine inside the pore; and L2, amantadine stably binds to the carbonyl oxygen atom of Ala30 and the side chain.

The binding free energy profiles of representative replicas (rep3, rep6, rep8 and rep10) determined from the binding kinetics of amantadine with N31 M2 show a single steep energy minimum (Fig. 3a), suggesting that amantadine mainly remains in the N31 M2 channel pore about 2 Å from the amantadine‐binding site of 2KQT PDB structure (S1), which is a 1.5 Å shift from the L2 state in S31 M2 (Fig. 2b). The amino group of amantadine was found to form direct hydrogen bonds with the side chains of Asn31 during the whole simulation time and to be oriented toward the N‐terminus of M2 (Fig. 3b).

Figure 3

Well‐tempered metadynamics simulations of amantadine binding in N31 M2. (a) Binding free energy profiles (kcal/mol) of representative replicas (rep3, rep6, rep8 and rep10) as determined by well‐tempered metadynamics simulations of the amantadine entry pathway into N31 M2. The structure in the free energy minimum was designated S1. (b) Snapshot of amantadine in N31 M2 (S1). The Asn31 polar residue is shown as a ball and stick model, simulated amantadine as a green stick model and the M2 backbone as thin tubes. Amantadine in the original NMR structure is depicted by gray lines (2KQT). Hydrogen bonds between the amino group of amantadine and the side chains of Asn31 are shown as black dotted lines.

Well‐tempered metadynamics studies of adamantyl bromothiophene‐M2 complexes

Next, the focus was placed on the binding kinetics of adamantyl bromothiophene, which is an amantadine derivative that inhibits both S31 M2 and N31 M2 proteins. Adamantyl bromothiophene binds to S31 M2 with the bromothiophene group oriented toward the C‐terminus of M2, but it binds to N31 M2 with the bromothiophene group facing up toward the N‐terminus of M2 33. To investigate the binding kinetics leading to the dual binding mode, free energy profiles of the representative replicas (rep5, rep7 and rep10) of adamantyl bromothiophene in S31 M2 and N31 M2, as determined by well‐tempered metadynamics simulations, were analyzed. The free energy profiles of adamantyl bromothiophene‐S31 M2 complexes showed broad funnel‐shaped curves and no metastable state was identified (Fig. 4a). These characteristics differ significantly from those of the free energy profiles of amantadine‐S31 M2 complexes (Fig. 2a). The timeline of the interactions between adamantyl bromothiophene and residues in the S31 M2 channel pore is shown in Figure 4b (rep7). Representative snapshots of adamantyl bromothiphene‐S31 M2 complexes observed from the N‐terminal end (Fig. 4c) and from the lateral side (Fig. 4d) are shown. In the first 10 ns, adamantyl bromothiophene binds to the entrance of the channel pore of S31 M2 through interaction of the amino group with one of the four Asp24 residues by a salt bridge (S1 in Fig. 4a,c,d). Adamantyl bromothiophene then enters the channel pore through hydrophobic interactions between the adamantane ring and Val27 (S2 in Fig. 4a,c,d). In contrast with amantadine, the amino group of adamantyl bromothiophene only forms a hydrogen bond with one of the four Asp24 residues because of steric hindrance. Thus, the interaction of adamantyl bromothiophene with Asp24 in S31 M2 may not be apparent as a metastable state comparable to that of the interaction of amantadine with S31 M2. After adamantyl bromothiophene enters the channel pore, the bromothiophene group oriented toward the C‐terminus of S31 M2 and adamantyl bromothiophene remains stable in the channel pore by means of water‐bridged hydrogen bonds with the carbonyl oxygen atoms of Ala30 (S3 in Fig. 4a,c,d).

Figure 4

Well‐tempered metadynamics simulations of adamantyl bromothiophene binding in S31 M2. (a) Binding free energy profiles (kcal/mol) of representative replicas as determined by well‐tempered metadynamics simulations of the adamantyl bromothiophene entry pathway of adamantyl bromothiophene into S31 M2. The steps in the entry pathway were designated S1, S2 and S3. (b) Timeline of the entry of a representative replica (rep7) showing the interactions and contacts between adamantyl bromothiophene and S31 M2 residues. Contacts between M2 residues and adamantyl bromothiophene are shown as orange lines. Key residues for adamantyl bromothiophene‐S31 M2 interaction are highlighted. The one charged amino acid is highlighted in pink, the polar amino acids in blue and the hydrophobic amino acids in green. (c) Representative replica (rep7) snapshot of the binding trajectory of adamantyl bromothiophene in the channel pore of S31 M2 observed from the N‐terminal end. Key M2 residues (Asp24, Val27 and Ala30), which interact with adamantyl bromothiophene, are shown as ball and stick models, whereas the corresponding amino acid residues in other M2 helices not interacting with adamantyl bromothiophene are shown as thin tubes. Adamantyl bromothiophene is shown as a green stick model. The quaternary S31 M2 backbones shown in green, blue, pink and yellow represent A, B, C and D chains in Fig. 4b, respectively. (S1) The amino group of adamantyl bromothiophene forms a salt bridge with S31 M2 Asp24 (pink dotted line). (S2) Hydrophobic interactions between the adamantane ring with Val27 stabilize amantadine inside the channel pore. (S3) Adamantyl bromothiophene is kept stable in the channel pore by water‐bridged hydrogen bonds with the carbonyl oxygen atom of Ala30. (d) Representative replica (rep7) snapshot of the binding trajectory of adamantyl bromothiophene in the channel pore of S31 M2 observed from the lateral side. Key M2 residues (Asp24, Val27 and Ala30), which interact with adamantyl bromothiophene, are shown as ball and stick models. Adamantyl bromothiophene is shown as a green stick model and the M2 backbone as thin tubes. Adamantyl bromothiophene from the original NMR structure (2MUW) is depicted by gray lines.

Figure 5a shows a representative replica (rep4) binding free energy profile calculated from well‐tempered metadynamics simulation of adamantyl bromothiophene‐N31 M2 complex. This profile shows a broad funnel‐shaped curve with four shallow local minima, (L1, L2, L3 and L4) about 13 Å (−10 kcal/mol), 2 Å (−15 kcal/mol), 10 Å (−13 kcal/mol), and 5 Å (−18 kcal/mol) from the adamantyl bromothiophene binding site of 2MUV PDB structure, respectively. The timeline of the contacts between adamantyl bromothiophene and N31 M2 (Fig. 5b) and the snapshots of the complexes in the four shallow local minima are shown (Fig. 5c,d). Adamantyl bromothiophene‐N31 M2 complex is shown from the N‐terminal end (Fig. 5c) and lateral side (Fig. 5d). Adamantyl bromothiophene was found to enter the channel pore through a halogen bond with Asn31 of N31 M2 (Fig. 5b and L1 in Fig. 5c,d). The amino group of adamantyl bromothiophene then interacts with the side chain of Asn31 through a transient hydrogen bond for 30–40 ns (L2 in Fig. 5c,d). Because the hydrogen bond between the amino group and Asn31 is relatively unstable because of the lack of a high free energy barrier (Fig. 5a), there is repeated binding and unbinding of adamantyl bromothiophene with Asn31. In addition, the bromothiophene group is oriented toward the N‐terminus of N31 M2 (L3 in Fig. 5c,d) and adamantyl bromothiophene forms stable bonds in the channel pore through hydrophobic interactions between the bromothiophene group and Ala30 and the hydrogen bond between the amino group and Asn31 (L4 in Fig. 5c,d). The binding topology of the amantadine‐M2 N31 complex revealed by well‐tempered metadynamics simulations (Fig. 3b) is consistent with direct hydrogen bonds between the amino group and Asn31 pushing the adamantane ring toward the C‐terminus of M2 and stabilizing that orientation. Therefore, it is likely that the hydrogen bonds between the amino group and Asn31 are important in determining the orientation of adamantyl bromothiophene in the channel pore.

Figure 5

Well‐tempered metadynamics simulations of adamantyl bromothiophene binding in N31 M2. (a) Binding free energy profiles (kcal/mol) of a representative replica (rep4) determined from the well‐tempered metadynamics simulations of the adamantyl bromothiophene entry into N31 M2. The entry steps were designated L1, L2, L3 and L4. (b) Timeline of the entry of a representative replica (rep4) showing the interactions and contacts between adamantyl bromothiophene and N31 M2 residues. Contacts between adamantyl bromothiophene and N31 M2 residues are shown as orange lines. Key residues for adamantyl bromothiophene‐N31 M2 interaction are highlighted. Polar amino acids are highlighted in blue and hydrophobic amino acids in green. (c) Representative replica (rep4) snapshots of the binding trajectory of adamantyl bromothiophene in N31 M2 observed from the N‐terminal end. Key N31 M2 residues (Val27, Ala30 and Asn31), which interact with adamantyl bromothiophene, are shown as ball and stick models, whereas the corresponding amino acid residues in other M2 helices not interacting with adamantyl bromothiophene are shown as thin tubes. Adamantyl bromothiophene is shown as a green stick model. The quaternary N31 M2 backbones shown in green, blue, pink and yellow represent A, B, C and D chains in Fig. 5b, respectively. L1: adamantyl bromothiophene enters the channel pore through a halogen bond with Asn31 (violet dotted line). L2: the amino group of adamantyl bromothiophene interacts with the side chain of Asn31 through a hydrogen bond (black dotted line). L3: the bromothiophene group is oriented toward the N‐terminus of N31 M2 and the hydrogen bond moved from the D to the B chain. (L4) Adamantyl bromothiophene is kept stably bound to the channel pore by hydrophobic interactions between the bromothiophene group and Val27 and Ala30 and through the hydrogen bond of the amino group with Asn31. (d) Representative replica (rep4) snapshots of the binding trajectory of adamantyl bromothiophene in N31 M2 observed from the lateral side. Key N31 M2 residues (Val27, Ala30 and Asn31), which interact with adamantyl bromothiophene, are shown as ball and stick models. Adamantyl bromothiophene is shown as a green stick and the M2 backbone as thin tubes. Adamantyl bromothiophene from the original PDB structure (2MUV) is depicted by gray lines.

DISCUSSION

In this study, we examined the binding kinetics of M2 channel blockers in S31 M2 and N31 M2 by well‐tempered metadynamics simulations. The binding trajectories of the M2 channel blockers in the channel pore of S31 M2 showed that amantadine first binds to two of the four Asp24 residues via salt bridges to yield a metastable conformation separated by a high free energy barrier. Previous conventional and multiple‐walkers well‐tempered metadynamics simulations of amantadine unbinding pathway from S31 M2 have shown that Asp24 interacts with amantadine and this binding mediates the amantadine flipping at the entrance of channel pore 32. Therefore, it is possible that these data support the hypothesis that the metastable binding of amantadine to Asp24 is important in forming a thermodynamically favored binding of amantadine to the channel pore of S31 M2. In the N31 M2 channel pore, the amino group of amantadine oriented toward the N‐terminus of M2 during the whole simulation time and the direct hydrogen bonds between its amino group and Asn31 residues appeared to push the adamantine ring to a deeper position than in the S31 M2 channel pore, in agreement with previous reports using conventional 29, 30 and steered MD simulations 31. It has been reported that, although amantadine can bind to the N31 M2 channel pore, amantadine in the N31 M2 channel is more mobile than that in the S31 M2 channel; as a consequence, amantadine cannot block proton conductance 29, 30. It should be noted that the well‐tempered metadynamics simulations in this study were carried out without a constant pulling force, which is required to overcome the steepest free energy barrier along the binding pathway; however, such a strong force may be excessive for analyzing local minima.

In contrast to amantadine, adamantyl bromothiophene could only form one salt bridge with Asp24 of S31 M2 because of steric hindrance; thus, the metastable binding state of adamantyl bromothiophene at Asp24 is more unstable than that of amantadine. This may explain why the antiviral activity of adamantyl bromothiophene is weaker than that of amantadine. The binding of adamantyl bromothiophene to Asp24 is not metastable in N31 M2, possibly because of the halogen bond between the bromine atom of adamantyl bromothiophene and Asn31 given that halogen bonds tend to be more directional than hydrogen bonds 40. Drugs, which bind to the target protein, shuttle between metastable binding sites and the binding pocket. Thus, separation of metastable binding sites from the binding pocket by transition states with high free energy barriers is important for rational control of drug binding kinetics 36, 41, 42.

In summary, our simulation indicates that Asp24 is a metastable binding site for amantadine. In contrast, adamantyl bromothiophene does not form metastable complexes with M2 because of steric hindrance of adamantyl bromothiophene or the halogen bond of the bromine atom to Asn31. Because formation of metastable complexes is critical for effective M2 inhibition, these findings may be helpful in developing new approaches to the design of further M2 inhibitors.

ACKNOWLEDGEMENTS

We thank to Dr. M. Kita (Nagoya University) for his helpful discussion. This research was supported in part by grants‐in‐aid from the Ministry of Education, Culture, Sports, Science, and Technology of Japan (16H05192 to A.K., 24115002 to K.N.) and the NOMURA Microbial Community Control Project of ERATO of the Japan Science and Technology Agency (A.K.). This research was also supported by the Platform Project for Supporting Drug Discovery and Life Science Research (Platform for Drug Discovery, Informatics, and Structural Life Science) from the Japan Agency for Medical Research and Development (H.T.).

DISCLOSURE

The authors declare that they have no conflicts of interest.