**3.2. Results and discussion**

Hydrogen bonds which form the primary interactions between oseltamivir and H5N1/ H1N1pdm neuraminidases were shown in figure 4 and figure 5 accordingly. While the SA binding sites of H5N1 and H1N1pdm appear to differ mainly in the sequence of loop residue 347, it is not well understood whether antiviral drugs bind to each protein in the same manner, or if the drug resistant H274Y and N294S mutations disrupt critical hydrogen bonds. To address this question, the hydrogen bonds which form between oseltamivir and the residues lining the SA binding pockets of H5N1 and H1N1pdm were calculated for all simulation trajectories. R292 and R371 were observed to hydrogen bond with oseltamivir's carboxyl moiety, and E119 and D151 with oseltamivir's amino group (NH3+). The H274Y mutation, however, appeared to disrupt the hydrogen bonding of oseltamivir's acetyl group with R152, an interaction which was seen in the wild type and N294S systems for both simEQ1 and simEQ2. Prior analyses of crystallographic data alone suggested that Y347 forms a stable hydrogen bond with oseltamivir's carboxyl group and is the source of oseltamivir-resistance in the N294S mutant.

Our MD simulations however, reflect statistics collected from long timescale simulations which produce a dynamic picture of molecular interactions in greater detail and resolution than can be seen from a static crystal structure. In our simulations oseltamivir's carboxyl group primarily forms hydrogen bonds with R292 and R371, having little involvement with Y347. In fact, residue Y347 undergoes rotation to interact strongly with residue W295. Therefore, the speculation from previous studies, that the N294S mutation in the case of H5N1 actually destabilizes the hydrogen bonding between oseltamivir and Y347 to induce drug resistance, is not supported in our simulations.

The notable difference between H5N1 and H1N1pdm neuraminidases is the replacement of Y347 by N347 at the drug binding pocket. No conserved drug-protein hydrogen bond was observed for N347 in any of the three H1N1pdm simulations. Given the transient nature of even the N294S mutant induced hydrogen bond involving residue 347 in the case of H5N1, and the lack of interaction with residue 347 in any of the other simulations, it is highly unlikely that the single residue change (Y347 to N347), between the H5N1 and H1N1pdm strains significantly alters the drug-protein stability in regard to the hydrogen bond network involved. H274Y mutation induced disruption of the stable hydrophobic packing of oseltamivir's pentyl group in both H5N1 and H1N1pdm neuraminidases.

168 Bioinformatics

electrostatics were approximated using the multilevel summation method (MSM), which uses nested interpolation of the smoothed pairwise interaction potential, with computational work that scales linearly with the size of the system. The calculation was performed using the molecular visualization program VMD that provides a graphics processing units (GPU) accelerated version of MSM to produce the electrostatic potential map. The GPU acceleration of MSM ( Hardy et al., 2009) provided a significant speedup over conventional electrostatic summation methods such as the Adaptive Poisson Boltzman Solver (APBS), achieving a benchmark processing time of 0.2s per frame versus 180 seconds per frame (on a conventional CPU) using APBS69 for a 35,000 atom system, offering a speedup factor of about 900. The use of GPU acceleration enabled averaging the electrostatic potential field over all frames of the simulation trajectories. Hydrogen bond analysis utilized

Hydrogen bonds which form the primary interactions between oseltamivir and H5N1/ H1N1pdm neuraminidases were shown in figure 4 and figure 5 accordingly. While the SA binding sites of H5N1 and H1N1pdm appear to differ mainly in the sequence of loop residue 347, it is not well understood whether antiviral drugs bind to each protein in the same manner, or if the drug resistant H274Y and N294S mutations disrupt critical hydrogen bonds. To address this question, the hydrogen bonds which form between oseltamivir and the residues lining the SA binding pockets of H5N1 and H1N1pdm were calculated for all simulation trajectories. R292 and R371 were observed to hydrogen bond with oseltamivir's carboxyl moiety, and E119 and D151 with oseltamivir's amino group (NH3+). The H274Y mutation, however, appeared to disrupt the hydrogen bonding of oseltamivir's acetyl group with R152, an interaction which was seen in the wild type and N294S systems for both simEQ1 and simEQ2. Prior analyses of crystallographic data alone suggested that Y347 forms a stable hydrogen bond with oseltamivir's carboxyl group and is the source of

Our MD simulations however, reflect statistics collected from long timescale simulations which produce a dynamic picture of molecular interactions in greater detail and resolution than can be seen from a static crystal structure. In our simulations oseltamivir's carboxyl group primarily forms hydrogen bonds with R292 and R371, having little involvement with Y347. In fact, residue Y347 undergoes rotation to interact strongly with residue W295. Therefore, the speculation from previous studies, that the N294S mutation in the case of H5N1 actually destabilizes the hydrogen bonding between oseltamivir and Y347 to induce

The notable difference between H5N1 and H1N1pdm neuraminidases is the replacement of Y347 by N347 at the drug binding pocket. No conserved drug-protein hydrogen bond was observed for N347 in any of the three H1N1pdm simulations. Given the transient nature of even the N294S mutant induced hydrogen bond involving residue 347 in the case of H5N1, and the lack of interaction with residue 347 in any of the other simulations, it is highly unlikely that the single residue change (Y347 to N347), between the H5N1 and H1N1pdm

a distance and angle cut-offs of 3.5 Å and 60 degrees, respectively.

**3.2. Results and discussion** 

oseltamivir-resistance in the N294S mutant.

drug resistance, is not supported in our simulations.

**Figure 3.** Network and occupancy of hydrogen bonds stabilizing oseltamivir in the SA binding pocket of wild type and drug-resistant mutant avian H5N1 neuraminidases, in simEQ1, simEQ3, and simEQ5.(A) shows histograms of the percent of hydrogen-bond occupancies for interactions between oseltamivir and residues E119, D151, R152, R292, Y347, and R371 across each simulation run. (B) through (D) are schematic views depicting the orientation of protein sidechains which form proteindrug hydrogen bonds.

Incorporating Molecular Dynamics Simulations

into Rational Drug Design: A Case Study on Influenza a Neuraminidases 171

solvent accessible surface area (SASA) of oseltamivir's pentyl group for all simulation trajectories. While there was no significant change to the pentyl group SASA (henceforth referred to as PG-SASA) in the wild type and N294S mutant for either H5N1 or H1N1pdm neuraminidases, an outward rotation of the pentyl group was observed in the H274Y mutant simulations, visible in oseltamivir's binding pose and evident in a notably higher calculated PG-SASA. The PG-SASAs for all simulated systems are plotted in Figure 5, with inset images of oseltamivir's binding pose in simEQ2 (Figure 5A) and simEQ4 (Figure 5B) illustrating the rotation of the pentyl group towards the open mouth of the binding pocket.

This image cannot currently be displayed.

**Figure 5.** The solvent accessible surface area of oseltamivir's pentyl group (PG-SASA) in H5N1 and H1N1pdm wild type and mutant simulations. The PG-SASA in the N294S mutant simulation (simEQ5 and simEQ6) did not vary significantly from wild type simulations (simEQ1 and simEQ2). However, there was a huge loss of hydrophobic interaction between drug and protein in the case of the H274Y

Previously published MD simulations performed over relatively short time scales (3 to 6 ns) have suggested two possible mechanisms: 1) that the H247Y mutation reduces the size of the hydrophobic pocket within the SA binding pocket near Oseltamivir's pentyl moiety (Malaisree et al., 2008), and 2) that the H274Y mutation breaks a critical salt bridge between E276 and R224 to disrupt drug binding( Wang et al., 2009). Our longer (40ns) simulations were able to corroborate the former suggested mechanism (as shown by the increase in PG-SASA in the case of the H274Y mutant simulations) but not the latter. In fact, in all six of our simulations, E276 maintains stable charge-charge interactions (salt bridging) with R224 despite displacement of the drug from the protein I222-R224-A246-E276 pocket in the case of H274Y mutants. This drug displacement increases water penetration into the I222-R224-A246-E276 pocket (Figure 5B). Evidence from our simulations therefore supports predictions from earlier

mutant simulations (simEQ3 and simEQ4), resulting in an increase in measured PG-SASA.

**Figure 4.** Network and occupancy of hydrogen bonds stabilizing oseltamivir in the SA binding pocket of wild type and drug-resistant mutant avian H1N1pdm neuraminidases, in simEQ2, simEQ4, and simEQ6. (A) are histograms of the percent of hydrogen-bond occupancies for interactions between oseltamivir and residues E119, D151, R152, R292, N347, and R371 across each simulation run. (B) through (D) are schematic views depicting the orientation of protein sidechains which form proteindrug hydrogen bonds.

Beyond disrupting the drug-protein hydrogen-bonding network, another mechanism through which protein mutations may induce drug resistance is by disruption of the hydrophobic packing of the drug into the protein binding pocket. Through inspection of the static crystal structures of the H274Y and N294S mutants of H5N1, it has been speculated that the mutations disrupt favourable hydrophobic packing interactions necessary for strong binding of oseltamivir. In our wild type simulations for both the H5H1 and H1N1pdm systems, the packing of oseltamivir's pentyl moiety tended to favour close association with residues I222, R224, A246, and E276. To test the effect of mutations H274Y and N294S on hydrophobic interactions of oseltamivir's pentyl group with the proteins, we calculated the solvent accessible surface area (SASA) of oseltamivir's pentyl group for all simulation trajectories. While there was no significant change to the pentyl group SASA (henceforth referred to as PG-SASA) in the wild type and N294S mutant for either H5N1 or H1N1pdm neuraminidases, an outward rotation of the pentyl group was observed in the H274Y mutant simulations, visible in oseltamivir's binding pose and evident in a notably higher calculated PG-SASA. The PG-SASAs for all simulated systems are plotted in Figure 5, with inset images of oseltamivir's binding pose in simEQ2 (Figure 5A) and simEQ4 (Figure 5B) illustrating the rotation of the pentyl group towards the open mouth of the binding pocket.

170 Bioinformatics

drug hydrogen bonds.

**Figure 4.** Network and occupancy of hydrogen bonds stabilizing oseltamivir in the SA binding pocket of wild type and drug-resistant mutant avian H1N1pdm neuraminidases, in simEQ2, simEQ4, and simEQ6. (A) are histograms of the percent of hydrogen-bond occupancies for interactions between oseltamivir and residues E119, D151, R152, R292, N347, and R371 across each simulation run. (B) through (D) are schematic views depicting the orientation of protein sidechains which form protein-

Beyond disrupting the drug-protein hydrogen-bonding network, another mechanism through which protein mutations may induce drug resistance is by disruption of the hydrophobic packing of the drug into the protein binding pocket. Through inspection of the static crystal structures of the H274Y and N294S mutants of H5N1, it has been speculated that the mutations disrupt favourable hydrophobic packing interactions necessary for strong binding of oseltamivir. In our wild type simulations for both the H5H1 and H1N1pdm systems, the packing of oseltamivir's pentyl moiety tended to favour close association with residues I222, R224, A246, and E276. To test the effect of mutations H274Y and N294S on hydrophobic interactions of oseltamivir's pentyl group with the proteins, we calculated the

**Figure 5.** The solvent accessible surface area of oseltamivir's pentyl group (PG-SASA) in H5N1 and H1N1pdm wild type and mutant simulations. The PG-SASA in the N294S mutant simulation (simEQ5 and simEQ6) did not vary significantly from wild type simulations (simEQ1 and simEQ2). However, there was a huge loss of hydrophobic interaction between drug and protein in the case of the H274Y mutant simulations (simEQ3 and simEQ4), resulting in an increase in measured PG-SASA.

Previously published MD simulations performed over relatively short time scales (3 to 6 ns) have suggested two possible mechanisms: 1) that the H247Y mutation reduces the size of the hydrophobic pocket within the SA binding pocket near Oseltamivir's pentyl moiety (Malaisree et al., 2008), and 2) that the H274Y mutation breaks a critical salt bridge between E276 and R224 to disrupt drug binding( Wang et al., 2009). Our longer (40ns) simulations were able to corroborate the former suggested mechanism (as shown by the increase in PG-SASA in the case of the H274Y mutant simulations) but not the latter. In fact, in all six of our simulations, E276 maintains stable charge-charge interactions (salt bridging) with R224 despite displacement of the drug from the protein I222-R224-A246-E276 pocket in the case of H274Y mutants. This drug displacement increases water penetration into the I222-R224-A246-E276 pocket (Figure 5B). Evidence from our simulations therefore supports predictions from earlier

studies of a possible mechanism for the H274Y mutation-induced drug resistance through water infiltration and destabilization of favourable drug packing. However, because no change in PG-SASA was observed during our simulations in the case of N294S for either protein, N294S induced drug resistance is probably due to a different mechanism.

Incorporating Molecular Dynamics Simulations

into Rational Drug Design: A Case Study on Influenza a Neuraminidases 173

electrostatic potential of the binding pocket wall, which exhibits a weak negative charge. In order to understand whether charge-charge interactions may play a role in mutation induced drug resistance, the electrostatic potentials were calculated and mapped onto the surfaces of the proteins. Even though prior studies have investigated the possible role of the electrostatic surface potential for drug binding in neuraminidases, the extensive electrostatic calculations required for fully understanding role of this phenomenon over a long simulation trajectory has not yet been done. The electrostatic surface potentials of the equilibrated systems were calculated and averaged across every trajectory frame in six equilibrium simulations. Extensive electrostatic calculations for the six systems show that the mouth of the binding pocket is positively charged, except for a narrow pathway of negative surface charge that seems to direct a possible binding pathway through which the drug may access the binding pocket. In Figure 6, the electrostatic surface potentials calculated from all six simulations is shown for the SA binding pocket with oseltamivir

Mutation H274Y in the H1N1pdm virus shares the same sources of oseltamivir resistance as H5N1, including loss of hydrogen bonds with R152 and reduction of hydrophobic interaction of oseltamivir's pentyl group. This leads to our suggestion to replace the pentyl group by a more hydrophilic one by adding a few hydroxyl groups on this bulky hydrophobic group. Our results do not support previous suggestions that the N294S mutation of H5N1 destabilizes the hydrogen bonding between oseltamivir and Y347, or that it disrupts the hydrophobic pocket, leading to drug resistance. Sources of drug resistance in mutant N294S remain unclear. Quantitative analysis of drug binding to the pocket is surrounded by a highly positive potential ring (colored blue). Simulations revealed the presence of a negatively charged pathway at the mouth of the binding pocket which may play a role for drug binding and mutation induced resistance, as the position of residue 294 maps directly onto the pathway while position of residue 274 is positioned on or adjacent( Le et al., 2009). WT and the mutants will be importantly complementary to obtained results in this chapter. Since the two mutations are nonactive-site, endpoint interactions alone cannot account for all drug resistance. Kinetics of drug binding is therefore also important for drug resistant mechanism. Especially, based on the decrease in association rate constants (Kon) of oseltamivir with H5N1 neuraminidase mutants (2.52 μM−1s−1 in the WT, 0.24 2.52 μM−1s−1 in H274Y and 1.1 2.52 μM−1s−1 in N294S mutants), we speculate that mutations also prevent drugs from entering the SA binding site. It is hoped that our observation of a possible binding pathway for oseltamivir will encourage further investigations that test the hypothesis posed

here, by identifying whether the actual drug binding pathway follows this route.

**shed light for further study on drug binding and drug resistance** 

**4.1. Characteristics of the electrostatic surface potential of N1 neuraminidases** 

Up to this point, all proposed mechanisms for oseltamivir resistance have focused on the effects of the mutations on the SA binding site but little has been known about the effects of

**4. Identification of drug binding pathway** 

bound.

**Figure 6.** Electrostatic surface potential of avian H5N1 and swine H1N1pdm neuraminidases in oseltamivir-bound simulations, revealing a positively charged pathway into the binding pocket. (A) through (C) show oseltamivir bound to H5N1 wild type, H274Y, and N294S drug-resistant mutant structures, respectively. (D) through (F) show oseltamivir bound to wild type, H274Y, and N294S drugresistant mutant structures, respectively. The outer columns show a close-up view of the binding pocket, highlighted as a subset of the enter protein shown in the central column. The positions of the mutant residues are shown in green for residue 274 and 294 in the mutant systems

While alterations in hydrophobic sidegroup packing may explain in part the mechanism behind H274Y mutation-induced drug resistance, they fail to shed light on the role that the N294S mutation plays for oseltamivir binding inhibition. Initial investigations of the electrostatic surface potentials of drug-bound N1 neuraminidases have proposed that drug binding affinity is closely related to favourable charge-charge interactions with the electrostatic potential of the binding pocket wall, which exhibits a weak negative charge. In order to understand whether charge-charge interactions may play a role in mutation induced drug resistance, the electrostatic potentials were calculated and mapped onto the surfaces of the proteins. Even though prior studies have investigated the possible role of the electrostatic surface potential for drug binding in neuraminidases, the extensive electrostatic calculations required for fully understanding role of this phenomenon over a long simulation trajectory has not yet been done. The electrostatic surface potentials of the equilibrated systems were calculated and averaged across every trajectory frame in six equilibrium simulations. Extensive electrostatic calculations for the six systems show that the mouth of the binding pocket is positively charged, except for a narrow pathway of negative surface charge that seems to direct a possible binding pathway through which the drug may access the binding pocket. In Figure 6, the electrostatic surface potentials calculated from all six simulations is shown for the SA binding pocket with oseltamivir bound.

Mutation H274Y in the H1N1pdm virus shares the same sources of oseltamivir resistance as H5N1, including loss of hydrogen bonds with R152 and reduction of hydrophobic interaction of oseltamivir's pentyl group. This leads to our suggestion to replace the pentyl group by a more hydrophilic one by adding a few hydroxyl groups on this bulky hydrophobic group. Our results do not support previous suggestions that the N294S mutation of H5N1 destabilizes the hydrogen bonding between oseltamivir and Y347, or that it disrupts the hydrophobic pocket, leading to drug resistance. Sources of drug resistance in mutant N294S remain unclear. Quantitative analysis of drug binding to the pocket is surrounded by a highly positive potential ring (colored blue). Simulations revealed the presence of a negatively charged pathway at the mouth of the binding pocket which may play a role for drug binding and mutation induced resistance, as the position of residue 294 maps directly onto the pathway while position of residue 274 is positioned on or adjacent( Le et al., 2009). WT and the mutants will be importantly complementary to obtained results in this chapter. Since the two mutations are nonactive-site, endpoint interactions alone cannot account for all drug resistance. Kinetics of drug binding is therefore also important for drug resistant mechanism. Especially, based on the decrease in association rate constants (Kon) of oseltamivir with H5N1 neuraminidase mutants (2.52 μM−1s−1 in the WT, 0.24 2.52 μM−1s−1 in H274Y and 1.1 2.52 μM−1s−1 in N294S mutants), we speculate that mutations also prevent drugs from entering the SA binding site. It is hoped that our observation of a possible binding pathway for oseltamivir will encourage further investigations that test the hypothesis posed here, by identifying whether the actual drug binding pathway follows this route.
