Identification of Factors Promoting HBV Capsid ... - ACS Publications

Jan 8, 2018 - imidines (HAPs) are a family of antivirals that target the HBV capsid protein ... surrounded by a capsid core protein (Cp) and an extern...
0 downloads 0 Views 2MB Size
Article Cite This: J. Chem. Inf. Model. XXXX, XXX, XXX−XXX

pubs.acs.org/jcim

Identification of Factors Promoting HBV Capsid Self-Assembly by Assembly-Promoting Antivirals Soumya Lipsa Rath, Huihui Liu, Susumu Okazaki, and Wataru Shinoda* Department of Materials Chemistry, Nagoya University, Nagoya 464-8603, Japan S Supporting Information *

ABSTRACT: Around 270 million individuals currently live with hepatitis B virus (HBV) infection. Heteroaryldihydropyrimidines (HAPs) are a family of antivirals that target the HBV capsid protein and induce aberrant self-assembly. The capsids formed resemble the native capsid structure but are unable to propagate the virus progeny because of a lack of RNA/DNA. Under normal conditions, self-assembly is initiated by the viral genome. The mode of action of HAPs, however, remains largely unknown. In this work, using molecular dynamics simulations, we attempted to understand the action of HAP by comparing the dynamics of capsid proteins with and without HAPs. We found that the inhibitor is more stable in higher oligomers. It retains its stability in the hexamer throughout 1 μs of simulation. Our results also show that the inhibitor might help in stabilizing the Cterminus, the HBc 149−183 arginine-rich domain of the capsid protein. The C-termini of dimers interact with each other, assisted by the HAP inhibitor. During capsid assembly, the termini are supposed to directly interact with the viral genome, thereby suggesting that the viral genome might work in a similar way to stabilize the capsid protein. Our results may help in understanding the underlying molecular mechanism of HBV capsid self-assembly, which should be crucial for exploring new drug targets and structure-based drug design.



INTRODUCTION Hepatitis B virus (HBV) belongs to the Hepadnaviridae family and specifically infects liver tissues. HBV-affected individuals are at high risk for developing liver cirrhosis and chronic hepatitis. HBV is also a leading cause of hepatocellular carcinoma. More than 257 million people are infected with the virus globally. In 2015, 887 000 deaths occurred as a result of HBV infection.1,2 Thus, in recent years researchers worldwide have investigated a variety of approaches for developing effective therapeutic strategies to cure HBV infections. HBV contains 3.2 kb of partially double-stranded DNA surrounded by a capsid core protein (Cp) and an external lipid envelope containing surface proteins.3 Electron micrographs show that the viral cores are uniform in nature. They exist most commonly in T=4 symmetry and are made up of 240 monomeric Cp units. However, cores with T=3 symmetry and 180 Cp monomers have also been well-documented in the literature.3 The primary role of the HBV capsid is to aid in the packaging of pregenomic RNA (pgRNA), an intermediate © XXXX American Chemical Society

genomic entity formed during HBV replication. Apart from that, HBV cores are involved in intracellular trafficking, interacting with nuclear import machinery, regulating reverse transcription, signaling, RNA chaperoning, and envelope acquisition.4 The basic building block of the HBV capsid is a homodimeric subunit. Each dimer is formed by monomers containing 183 residues (Cp183). Residues 1−149 form the N-terminal helixrich assembly domain, which is sufficient to drive capsid assembly. The C-terminal arginine-rich domain comprises 34 residues (150−183) and is primarily involved in RNA binding.5−7 In vitro and in vivo studies have shown that the N-terminal 149 residues (Cp149) alone are sufficient to drive capsid assembly and that the C-terminus primarily aids in genome packaging. The difference between the Cp149 and Cp183 lies in the capsid wall thickness, intraparticle space, and 2 nm increase in the capsid diameter with Cp183.8,9 The Cp Received: August 6, 2017 Published: January 8, 2018 A

DOI: 10.1021/acs.jcim.7b00471 J. Chem. Inf. Model. XXXX, XXX, XXX−XXX

Article

Journal of Chemical Information and Modeling

Figure 1. (a) Structure of HBV dimer, with the important domains highlighted in different colors. The inhibitor is shown in licorice and colored by name, with C in yellow, N in blue, O in red, F in cyan, Br in pink, and S in orange. (b) HBV capsid structure (shown in gray) highlighting the hexamer (shown in red).

dimers are held together by a four-helix bundle, with two helices being contributed by each monomer. These bundles (residues 49−109) form spikes that project out from the virus surface (Figure 1). Generally, dimers are linked to each other by disulfide linkages between residues Cys61 and Cys61′ that aid in stability.10 Although the four-helix bundles formed by dimers are stable, flexibility has been observed around the spikes (residues 63−94), the fulcrum that mechanically connects the other two subdomains (residues 10−25), and the interdimer contact region (residues 111 to the C-terminus) (Figure 1).11,12 The end loops of the dimer, distal to the fourhelix bundle, are responsible for interdimer contacts by fitting into the helix−loop structure at the assembly domain of a neighboring unit.13 Each of the dimers associates with a weak contact energy, such that incorrectly or weakly associated dimers could be removed by thermodynamic editing.14−16 Capsid assembly is essential for viral propagation. In vivo, the HBV capsid assembly process is facilitated by the pgRNA, which acts as a scaffold to bring the capsid subunits together.17 Surprisingly, in vitro, capsid assembly has also been observed in the absence of pgRNA upon modulation of the experimental conditions.15,17 However, for full-length Cp183 without any artificial truncation at the C-terminus, in vitro capsid assembly can be successfully and efficiently induced by polyanions, including poly(glutamic acid), dextran sulfate, oligonucleotides, and nonbiological polymers.15,17−19 In the absence of RNA, antiviral molecules such as heteroaryldihydropyrimidines (HAPs) have been found to accelerate capsid assembly.20−24 Binding of HAPs to capsid proteins does not alter the morphology of the protein but does enhance the interdimer contact energy. The first step in capsid assembly is the formation of a nucleus consisting of a trimer of dimers (Figure 1b).25 Although the presence of pentameric and other smaller subunits upon dispersion of HBV capsid has been reported,26

their formation is assembly-competent and low in abundance. Thus, using hexameric subunits as the nucleation state that has been established27,28 is more acceptable. According to the kinetic model, during capsid assembly, which proceeds via nucleation, faster nucleus formation accelerates the assembly kinetics. Under normal conditions, equilibrium exists between nucleation and the formation of viral capsids.26 Many crystal structures of HAP-bound HBV subunits have been resolved. These molecules bind at the dimer−dimer interface, burying a large hydrophobic surface area. Strangely, binding of HBV subunits to different HAPs results in the formation of geometrically different virus capsid structures.29 Very recently, Klumpp et al.30 crystallized HBV hexamers bound to six HAP molecules (named NVR-010-001-E2) (Figure S1). When bound to this inhibitor, the generated capsids bear a striking resemblance to native HBV capsid structures. This compound varies slightly from the existing HAP inhibitors such as BAY 41-4109, GLS4, and others by possessing greater thermal stability and a lower EC50 value. According to their reports, the presence of a morpholino group increased the number of interactions between the dimers, thereby burying a significantly larger hydrophobic surface area and producing more stable HBV capsids.30 However, this raises an intriguing question of the role of RNA in promoting capsid assembly because RNA does not bind at the dimer−dimer interface.17 The global structural changes that are mimicked by these assembly-promoting antivirals remain unknown. Stray et al.21 found HAP-1 to be an allosteric effector of capsid assembly by enhancing morphological changes in the dimer structure to an active assembly state. Packianathan et al.11 observed allostery to be a general phenomenon for self-assembling viruses. Although diverse capsid structures were observed upon subtle changes in HAP structure, it is still not clear how HAPs might allosterically influence capsid self-assembly. B

DOI: 10.1021/acs.jcim.7b00471 J. Chem. Inf. Model. XXXX, XXX, XXX−XXX

Article

Journal of Chemical Information and Modeling

constrained by the LINCS algorithm. All of the simulations were performed using GROMACS 5.0.7 and the CHARMM36 force field.32−36 Multimodal Dynamic Cross-Correlation and Dynamic Cross-Correlation Analyses. The dynamic cross-correlation (DCC) and multimodal dynamics cross-correlation (MDCC) analyses were carried out using the mDCC_tools software designed by Kasahara et al.41,42 The DCC analysis to quantify the correlation between the ith and jth atoms in the system is given by

Molecular dynamics (MD) simulations have emerged as an essential tool for understanding the structure and dynamics of large macromolecular structures such as proteins. In this work, we implemented all-atom MD simulations at nanosecond to microsecond time scales to study the impact of inhibitor binding on various Cp subunits. We addressed how HAP binding might influence HBV protein subunits. Apart from the HAP binding site at the interdimer interface, is there an underlying allosteric phenomenon that mimics the effect of RNA binding and promotes capsid self-assembly? To address this question, we first compare inhibitor binding to monomers, dimers, and hexamers and then extensively discuss the impact of inhibitor binding to hexamers, the first stable unit that acts as the nucleus for capsid assembly.

DCC(i , j) =



⟨Δri(t ) ·Δrj(t )⟩t ⟨∥Δri(t )∥2 ⟩t ⟨∥Δrj(t )∥2 ⟩t

where ri(t) is the vector of the ith atom’s coordinates as a function of time t, ⟨···⟩ denotes the ensemble average, and

METHODS The coordinates for the hexameric HBV capsid subunit bound to the inhibitor NVR-001-010-E2 were obtained from the recent crystal structure by Klummp et al. (PDB ID 5E0I).30 On the basis of this crystal structure, monomer, dimer, and tetramer systems were set up (Table S1). Although the Cterminus is largely intrinsically disordered, we added the missing residues of the hexamer according to the crystal structure data to find out whether the residues contribute toward the hexamer stability. The missing residues (chain A 149−151; chain B 149−151; chain C 75−82, 149−151; chain D 73−80, 149−151; chain E 72−81, 149−151; chain F 72−88, 149−151) were modeled using MODELLER version 9.0.31 GROMACS version 5.07 was used for initial system setup and further MD simulation.32−34 The CHARMM36 all-atom force field with CMAP terms was used for proteins.35,36 Each monomer contained residues 1−149 (Cp149). The disulfide bonds between the dimers were maintained between Cys61 and Cys61′ in the dimer, tetramer, and hexamer systems. Each system was solvated in a periodic cubic box made up of CHARMM TIP3P water molecules.37 The force field parameters for the ligand NVR-001-010-E2 were obtained using the VMD version 1.9.2 Force Field Development Toolkit (FFTK) (described in the Supporting Information).38 The systems were neutralized by the addition of sodium ions. After neutralization, the physiological concentration of 150 mM NaCl was maintained by the addition of Na+ and Cl− ions. The initial box size was defined such that the minimum distance between two periodic protein images was longer than 2 nm. The systems were initially energy-minimized for 50 000 steps. The minimization was initially performed with position restraints on the protein atoms (1000 kJ mol−1nm−2) and subsequently without restraints, first by steepest descent followed by the conjugate gradient algorithm. After minimization, each system was equilibrated without restraints for 100 ps in an NVT ensemble and then for 1 ns under NPT conditions. The simulation was prolonged further to generate production data for analysis. We could generate 25 ns/day for the largest system using the supercomputer. A temperature of 310 K and pressure of 1 bar were used. During equilibration and throughout the production run, temperature was controlled by the V-rescale thermostat (velocity rescaling).39 The Andersen barostat40 was used to maintain the desired pressure of 1 bar. The Lennard-Jones interactions were truncated at 1.2 nm by applying a force switching function in the range of 1.0− 1.2 nm. The long-range electrostatics were calculated by the particle mesh Ewald method using a Fourier grid spacing of 0.1 nm. In order to allow a time step of 2 fs, all of the bonds were

Δri(t ) = ri(t ) − ⟨ri(t )⟩t

DCC analysis has a definitive explanation only when the distribution of atomic coordinates is unimodal because the average coordinate is uniquely determined for each atom. However, atoms may show multimodal movements during long-time simulations. This is especially plausible for the side chains of proteins, whose rapid flips are important for interface contact. The mDCC analysis can quantify these correlative motions that are invisible in DCC analysis. The mDCC calculation basically involves two steps. In the first step, a Gaussian mixture distribution is built on the basis of the atomic coordinates obtained from the MD trajectory. From the distribution, the maximum number of modes K is determined. In the second step, the correlation mDCC between fluctuations of the ith atom from mode k and the jth atom from mode l is determined using the following equation: mDCC(i , j ; k , l) =

⟨wi , j; k , l(t )(Δri , k(t ) ·Δrj , l(t ))⟩t ⟨wi , j; k , l(t )∥Δri , k(t )∥2 ⟩t ⟨wi , j; k , l(t )∥Δrj , l(t )∥2 ⟩t

where wi,j;k,l(t) is the common weight coefficient when the ith and jth atoms belong to modes k and l, respectively. Further details about mDCC can be found in ref 41.



RESULTS AND DISCUSSION We first studied the impact of HAP binding to different capsid subunits, such as monomers, dimers, and hexamers. In vitro and in vivo experiments have shown the hexamer to be the nucleating unit of assembly. We used biological assembly no. 1 assigned by the authors of ref 30, which is identical to the asymmetric unit for our study. Upon nucleus formation, Cp dimers are added to generate the fully formed HBV capsid structure.30 For comparison, we considered each system with and without a HAP inhibitor. HBV monomers have a single inhibitor binding site, while dimers can bind two inhibitors and hexamers can bind six inhibitors. Binding of HAP to Hexamers Is More Stable than Binding to Monomers or Dimers. After initial thermalization and equilibration, the systems were subjected to a production run under NPT conditions (for details, see Methods). To see the structural stability and dynamics of the systems during the whole simulation time, the root-meansquare deviations (RMSDs) of simulations with respect to the starting structures were calculated and plotted (Figure 2). Our C

DOI: 10.1021/acs.jcim.7b00471 J. Chem. Inf. Model. XXXX, XXX, XXX−XXX

Article

Journal of Chemical Information and Modeling

ejection of the inhibitor from monomers and dimers within 400 ns while they remained in the hexamer. This has been elaborated in later sections. According to recent studies by Zlotnick,27 minor differences in inhibitor structure impact the global capsid structure to a very large extent. Thus, although subtle differences were observed in the monomer and dimer structures upon the binding of inhibitor (Figure 2a,b), the hexamer showed larger variation (Figure 2c) The differences in residue-wise fluctuations of monomer, dimer, and hexamer complexes with inhibitors were calculated by their root-mean-square fluctuations (RMSFs), representing an ensemble average of RMSF with standard deviation. Most residues showed similar RMSFs among all three complexes, both bound to inhibitors and without inhibitors (Figure S3). However, a remarkable difference in the fluctuation of residues around the spike region, defined as the region comprising the α3 and α4 helices (Figure 3) was observed for inhibitor-bound

Figure 3. Average RMSFs of the monomer (black), dimer (red), and hexamer (blue) bound to inhibitor with the respective standard deviations. The five different helices are shown at the top of the figure for clarity.

(Figure 3) and apo complexes (Figure S4). This region encompasses the four-helix bundle formed by the helix−helix association and the disulfide linkage and plays a significant role in complex stability, as previously mentioned. The spike region, especially around residues 78 to 82, constitutes part of the “hole” through which molecules can freely travel into and out of the capsid.43 Nevertheless, this further strengthens the fact that capsid dimers are stable HBV subunits. The residues around the inhibitor binding site did not display large variations in any of the three HBV subunits. After determining the stability and residue-wise fluctuations in the protein, we analyzed the inhibitor stability in each of the complexes. In order to determine the stability, snapshots of representative proteins and inhibitors were taken at 100 ns intervals for monomer− and dimer−inhibitor complexes and at 250 ns intervals for the hexamer−inhibitor complex. As can be seen in Figure 4, the inhibitor escapes the monomer before 400 ns of simulation, and one of the two inhibitors is immediately released from the dimer before 100 ns of simulation. To confirm that this phenomenon was not an artifact of our simulation, we simulated replicates of the inhibitor-bound monomer and dimer systems. Figure 4b shows the percentage occupancy of the inhibitors in the inhibitor binding pocket. Occupancy was calculated as an estimate of the time during

Figure 2. (a, b) Time evolution of the RMSDs of the (a) monomer and (b) dimer without inhibitor (black) and with inhibitor(s) (various colors). (c) Time evolution of the RMSD of the hexamer with (red) or without (black) inhibitor.

calculations show that the monomer and dimer attained converged RMSDs in the initial 60 ns of simulation time. Moreover, the presence or absence of inhibitor(s) did not bring about drastic changes in the RMSDs. Additionally, the dimer showed smaller RMSDs than the monomer. The dimer is a complex of two monomers with a Cys61−Cys61′ disulfide bond, which reduces thermal fluctuations.10 The hexamer also showed the convergence of RMSDs in the initial 40 ns of simulation time. However, after 400 ns of run time, the fluctuations in the hexamer without inhibitor were prominently observed (Figure 2c). While the production data for the monomer and dimer were generated for 400 ns, 1 μs simulations were run for the hexamer. This was due to the D

DOI: 10.1021/acs.jcim.7b00471 J. Chem. Inf. Model. XXXX, XXX, XXX−XXX

Article

Journal of Chemical Information and Modeling

the inhibitors were still bound to the pocket (Figure 4b). The 100% occupancy of the residues at the interface (shown as (2) in Figure 4b) was not unexpected because of the interaction of the inhibitor with residues from both dimers. However, unlike the monomer and dimer cases, the relatively exposed inhibitors (shown as (1) in Figure 4b) occupy the inhibitor binding pocket throughout. These inhibitors interact only with residues from one of the dimers. Moreover, the differences in the dynamics of the residues in the dimer and hexamer systems are indicated. In order to determine whether the difference persists for a tetramer (a dimer of dimers), we ran a simulation of a tetramer Cp complex bound to four inhibitors in two replicates. Surprisingly, except for the interface inhibitor, which showed 100% occupancy, no consistent pattern could be observed for the stability of the remaining inhibitors (Figure S5). Thus, the long-term stability of the exposed inhibitor in the hexamer complex might be due to the allosteric influence of more than one dimer, which will be discussed later. Thus, from the above calculations it is clear that although the inhibitor binding pocket is largely similar in the different HBV subsystems, its effect is more pronounced in the hexameric form. Besides, inhibitor occupancy occurs for a much longer time in the hexamer complex than in the monomer, dimer, or tetramer systems. In vitro, this inhibitor instability in dimers and tetramers could be overcome as a result of the crowding effect of dimers in the systems, which could readily lead to hexamer formation.28 Because of the readily available Cp dimers, upon association the complex might gain rapid thermodynamic and kinetic stability in the presence of inhibitors.29 HBV Hexamers Are More Stable in the Presence of Inhibitors. After determining that inhibitor binding is more stable in the HBV hexamer, we determined differences in hexamer dynamics with and without inhibitors by performing principal component analysis (PCA). PCA extracts the most important elements in the data using a covariance or correlation matrix constructed from atomic coordinates. The correlated variables are decomposed into a reduced set of independent variables by eigenvalue decomposition of the matrix. The orthogonal collective modes are known as eigenvectors, each of which has a corresponding eigenvalue (or variance). Often the first few eigenvectors, with larger eigenvalues, are sufficient to describe the overall protein motion.44,45 Thus, using PCA one can easily capture the large concerted motions in hexamer structures with a few eigenvectors. We applied PCA on the Cα atoms of each system to capture the essential motion of the protein during the 1 μs simulation time. First, we analyzed the contributions of the individual eigenvectors to the overall protein motion by determining their eigenvalues. Figure S6 shows that the first 10 eigenvectors are sufficient to explain ∼80% of the protein motion. The original data were then projected onto the two most prominent eigenvectors (principal components), which account for more than ∼40% of the protein motion. For clarity, we projected the PCs for each dimer. Although a significant overlap was observed in the movements of each dimer, the inhibitorbound hexamer showed restricted fluctuations. Figure S6b,d suggests that the dimers bound to inhibitors explore a different region of conformational space than the apo proteins. Figure S6c clearly shows the confined dynamics of one of the dimers when it is bound to inhibitor. To understand the motions even more clearly, porcupine plots along the direction of PC1 are presented in Figure 5.

Figure 4. (a) Snapshots of monomer, dimer, and hexamer HBV subunits (green) and inhibitor (magenta) during specific time intervals during the simulation. (b) Inhibitor occupancies in the pocket for the monomer, dimer, and hexamer. The average occupancies of the inhibitors with error bars are shown for Monomer (Monomer), Dimer with one inhibitor (Dimer (1)), Dimer with two inhibitors Dimer (2), solvent exposed inhibitors of hexamer (Hexamer (1)) and interface inhibitors of hexamer (Hexamer (2)).

which the inhibitor remained within 5 Å of its original position, allowing for minor fluctuations in inhibitor position during the simulation. Two of the three replicates of monomer bound to inhibitor indicated that the inhibitor left the binding pocket. From Figure 4a, we can deduce that inhibitor binding is not very stable in the capsid protein monomer. The dimer system has two binding pockets and hence the propensity to bind two inhibitors. Thus, we carried out one 400 ns simulation for the dimer bound to one inhibitor and three replicates for the dimer bound to two inhibitors. In the single-inhibitor system, the inhibitor left the pocket almost immediately after being subjected to the production run, resulting in a significantly low occupancy value of 12%. For the dimer bound to two inhibitors, in all instances we observed one inhibitor remaining and the other leaving the pocket (Figure 4b). However, there was no pattern in the inhibitor release to identify which was a better binding site in the symmetrical HBV dimer. Thus, one could say that inhibitor binding and unbinding are completely stochastic in smaller oligomers but larger assemblies help in inhibitor stability. The above observations imply that inhibitor stability is low in monomer and dimer systems. Since the hexamer systems were relatively large, we evaluated their stabilities for a longer time, up to 1 μs. Interestingly, out of all the HBV subunits, the conformation that provided nearly 100% inhibitor occupancy was that of the hexamer. Three of the six inhibitors that were located at the interface of two dimer subunits showed 100% occupancy, while the three others that were relatively solventexposed showed slightly less than 100% occupancy, although E

DOI: 10.1021/acs.jcim.7b00471 J. Chem. Inf. Model. XXXX, XXX, XXX−XXX

Article

Journal of Chemical Information and Modeling

Figure 5. Porcupine plots of a hexamer (a) without and (b) with inhibitor.

Figure 6. (a, b) Average structures of a hexamer (a) without and (b) with inhibitor. (c) Structures obtained from clustering analysis showing changes in the C-terminus of the hexamer (i) without and (ii) with inhibitor.

movement of interdimer terminals 1−2 and 2−3 at the interface are pointed in a similar direction, while the movement of 3−1 is negligible, indicating complex stability. Without the inhibitor, only dimer 1−2 movements are directed toward the similar path, while no such synchronous movement could be seen for 2−3 and 3−1 dimers. Comparison of the average hexamer structures obtained after 1 μs simulations helped us determine the effect of inhibitor binding on the hexamer structure. Figure 6 shows the arrangement of the dimers in the hexamer with and without inhibitor. An interesting structural change found in the hexamer structure without inhibitor was a consistent C-terminal overlap between adjacent dimers that was missing in the apo complex (Figure 6a). At the molecular level, we found that the Cterminus was indeed captured between the helix−loop structure at the terminal region. Stability was further promoted by polar interactions between Ser121 and Pro144′, Thr114 and Lys150′, and Glu117 and Lys150′ between the helix from one dimer and the C-terminal loop of another, and also between Thr147 and

Here, each cone represents the magnitude and direction of movement of the Cα atoms for the extreme conformations.44 In Figure 5, reduced hexamer fluctuations can be observed when inhibitor is bound (Figure 5b). More and comparatively longer vectors in the porcupine plots of hexamers without inhibitor suggest the lower stability of the apo hexamer. It is well-documented in the literature that the hexamer forms the nucleus for capsid formation. Thus, in the absence of genetic material such as RNA/DNA or assembly-promoting antivirals, even though the HBV hexamer might form, the stability of the hexamer complex is highly uncertain. Hexamer stability is essential for viral capsid formation.21,22,30 Thus, increased hexamer stability shifts the equilibrium favorably toward capsid formation. Figure 5 shows the three different dimers colored in cyan, pink, and green. Overall fluctuations without the inhibitor are larger, especially in the spike region of the protein. Normally, one might assume stronger dimer−dimer association if the movement of each dimer is concerted or very nominal.44 This is seen in the hexamer−inhibitor complex, where F

DOI: 10.1021/acs.jcim.7b00471 J. Chem. Inf. Model. XXXX, XXX, XXX−XXX

Article

Journal of Chemical Information and Modeling

Figure 7. Conformations of residues around the inhibitor binding pocket with (colored) and without (white) inhibitor. (a, b) Residues around the exposed inhibitor binding site. (c) Changes in the positions of residues around the interfacial inhibitor binding site. (d) Interdimer interactions highlighting the C-terminal-interacting residues in the hexamer bound to inhibitor.

Ser141′ between the loop of the first dimer and the C-terminal loop of another (Figure 7). Apart from polar interactions, hydrophobic interactions between Val124, Val120, Val115, and Leu143 from one dimer and Leu143′, Val148′, Val149′, and Val151′ of the C-terminus effectively stabilize the interdimer overlapping region (data not shown). Surprisingly, a drastic difference was encountered in the apo complex, where the Cterminus and the helix−loop domain face away from each other and therefore do not interact with each other (Figure 6c). The RMSF of this region is shown in Figure S4c. We measured the angles between the three dimers at the beginning and end of the simulations for the hexamer with and without inhibitor and found them to be 55°, 46°, 72° and 51°, 46°, 72°, respectively, with inhibitor and 56°, 47°, 70° and 57°, 54°, 63°, respectively, without inhibitor. This shows gradual distortion of the hexamer symmetry in the absence of inhibitor. This raises a question regarding the impact of inhibitors on the stability of the C-terminus. To evaluate the changes, we superimposed the inhibitor binding regions of the apo and inhibitor-bound hexamers. HAPs promote dimer stability by burying a large surface area or by increasing the hydrophobic interaction energy in the interdimer region. However, the conformational changes in the inhibitor binding residues were very small. We found that although the interdimer interacting region residues Ala132, Arg133, and Pro134 effectively move closer to each other after inhibitor removal, the overall

positions of the residues are largely consistent (Figure 7). However, the C-terminal residues 146−149 are brought into close proximity with the inhibitor, thereby effectively stabilizing the C-termini in the hexamer. In the interface region, interdimer interactions of Ile139 and Leu140 with the inhibitor stabilize the C-terminus, further aided by polar and hydrophobic interactions as previously mentioned. Surprisingly, when the inhibitor was relatively solvent-exposed, van der Waals interactions between the morpholino group of the inhibitor and C-terminal residues 146−149 were observed. To further validate the importance of the C-terminus, we carried out a 1 μs MD simulation of the hexamer without the C-terminal residues in the presence of inhibitors. Although no significant changes in the RMSFs and RMSDs were detected, we found that the exposed inhibitor occupancy was largely affected: the average occupancy of the exposed inhibitors decreased to 50.18 ± 38.88% and that of the interface inhibitors was 99.63 ± 0.09% (Figure S7). HAP binding at the dimer−dimer interface not only buries a large hydrophobic surface area but also directly affects the Cterminal stability of the hexamer. It has been suggested that the C-termini of HBV capsids are directly involved during capsid assembly.2,5−7 They interact directly with viral genetic material, acting as a scaffold. Even though we used a truncated capsid monomer, the enhanced stability of the C-terminus upon inhibitor binding is evident. From these observations, we can G

DOI: 10.1021/acs.jcim.7b00471 J. Chem. Inf. Model. XXXX, XXX, XXX−XXX

Article

Journal of Chemical Information and Modeling

Figure 8. Multimodal dynamic cross-correlation (upper triangles) and dynamic cross-correlation (lower triangles) plots for the hexamer (a) without and (b) with inhibitor to show differences in atomic-level dynamics. The monomers in the hexamer are labeled as A−F.

spike region, the C-terminus, the interdimer contact region, and parts of the N-terminus show anticorrelated movements (Figure S8). Similarly, the dynamics of monomer D is also anticorrelated with those of E and F. Interestingly, while the monomer−monomer contact region in each dimer shows positive correlations, the C-termini are negatively correlated in the apo hexamer complex. This implies decreased stability of the hexamer complex and hints at a higher probability of dissociation over a longer time. Anticorrelated domain movements are destabilizing for large structural entities such as virus capsids, whereas several other proteins have anticorrelated motions to support their function. For the HBV hexamer, large correlated movements help in keeping the structure intact. This was further promoted by the binding of HAP inhibitor, which reduces anticorrelated movements between dimers and helps in hexamer stability. The inhibitor binds at the junction of the interdimer contact region and the base of the spike region (see Figure 1) and interacts with the residues from the C-terminus of the dimer. In addition, the inhibitor directly or indirectly promotes Cterminal stability through interdimer interactions. However, from the DCC analysis we also found that the inhibitor actually decreases anticorrelated movements between dimers.

conclude that RNA/DNA might also help in effectively stabilizing capsid C-termini, in turn promoting capsid stability and self-assembly.5−7 Inhibitor Binding Promotes Capsid Assembly Allosterically. As discussed above, inhibitor binding directly promotes assembly by stabilizing the C-termini of HBV capsid dimers. However, HAPs also promote capsid assembly allosterically.11,20 Thus, we performed dynamic cross-correlation (DCC) analysis to characterize the correlated and anticorrelated movements of the protein backbone with and without inhibitor. For a long simulation time of 1 μs and about 200 000 atoms in the system, it is likely that some of the correlated motions in the protein might not be effectively captured by the DCC method alone. The calculation depends on the displacement from the average structure alone, which is a single-mode displacement. However, during the simulation atoms may exhibit multimodal behavior, especially due to sidechain movement. In order to capture multimodal dynamics, we also performed multimodal DCC (mDCC) along with DCC (for details see Methods).41 Figure 8 shows the DCC (lower triangles) and mDCC (upper triangles) values for both the apo hexamer and the hexamer bound to inhibitor. The mDCC values are higher than the DCC values because the maximum values are taken from a combination of all modes. In Figure 8, each monomer within the hexamer is named from A−F, and the important domains are colored according to Figure 1 for clarity. The figure shows that all of the dimers (A− B, C−D, and E−F) exhibit positively correlated motions, although the correlation is much stronger for the inhibitor− hexamer complex. The collective motions of the protein assembly help us determine whether the individual protein domains are moving in a synchronous or asynchronous manner. When the domains move asynchronously or are anticorrelated, their structural assembly is compromised. Figure 8a shows that the multimodal motions between the dimers of the apo hexamer are anticorrelated. In the inhibitor−hexamer complex (Figure 8b), except for the spike region between dimers AB and CD, most proteins still show a positive correlation in the mDCC plot (colors range from green to dark red). Since the spike region protrudes from the capsid surface and is not directly associated with the base of the capsid, most parts of the hexamer moved in correlated motions and therefore were stable. In the apo complex without the assembly promoters, anticorrelations were observed between all dimers (shown in blue). For example, between dimers AB and EF, apart from the



CONCLUSIONS Hepatitis B virus (HBV) infects the human liver tissues. It primarily causes liver cirrhosis, chronic hepatitis, or hepatocellular carcinoma. One of the important therapeutic strategies for curbing the growth of HBV is targeting the viral capsid. In recent years, antiviral molecules such as heteroaryldihydropyrimidines (HAPs) have been found to bind to the capsid protein and deregulate its assembly process. HBV capsid assembly is generally initiated by its genomic material in vivo, and thus, it is interesting to understand the molecular mechanism of capsid assembly by nonconventional molecules such as HAPs. HAPs increase the stability of empty HBV capsids and help in growth of hexameric capsid units under experimental conditions. Simulations of HAPs with various oligomers were carried out to understand the basis of structural stability. We found that inhibitor occupancy was greater in the hexamer than in the monomer, dimer, or tetramer. The drug unbinds randomly from all oligomer−inhibitor complexes except the hexamer−inhibitor complex. Reduced dynamics of larger complexes facilitates the stability of HAPs. Apart from that, lower occupancies of inhibitors in isolated monomer, dimer, and tetramer systems point toward the importance of H

DOI: 10.1021/acs.jcim.7b00471 J. Chem. Inf. Model. XXXX, XXX, XXX−XXX

Article

Journal of Chemical Information and Modeling

Computer” (Building Innovative Drug Discovery Infrastructure through Functional Control of Biomolecular Systems). This research used computational resources of the K computer provided by the RIKEN Advanced Institute for Computational Science through the HPCI System Research Project (Projects hp150269, hp160223, and hp170255) and supercomputers at Nagoya University.

macromolecular crowding of Cp dimers in HBV assembly. As observed in earlier studies, macromolecular crowding might increase the likelihood of more stable protein−protein and protein−inhibitor assemblies. A recent study by Perilla et al.46 suggested that HAP binding produces global structural changes. This was observed earlier by Zlotnick27 as well. In our study, we saw pronounced changes in the hexamer structure compared with the monomer and dimer structures. We found that even though we used a truncated capsid protein, the stability of the C-terminus upon inhibitor addition was evident. There was overlapping of the C-terminus between the dimers, which was missing in the absence of inhibitors. We also found that while the inhibitor was stabilized by interactions with the adjoining residues, the capsid stability was due to inhibitor-assisted interactions between dimers in the hexamer. Moreover, the inhibitor actually decreased anticorrelated movements in the hexamer by stabilizing interactions. These observations may suggest that in a similar way, RNA/ DNA could also help in effectively stabilizing the C-terminus of the capsid protein by allosterically restricting the anticorrelated movements among the dimers. The drug binds to residues in the drug binding pocket and also to residues in the C-terminus. The C-terminus is largely structureless and random in nature.47 It mostly interacts with the RNA and hence is expected to lie inward in the capsid structure and not be exposed. Therefore, it does not interact with other dimer/hexamer structures completely. However, in our work we found that a portion of it might help bind the hexamer structure. The difference between the hexamer structures with and without inhibitor is important because the probability of the hexamer structure remaining intact is lesser on its own without RNA/drug. Therefore, capsids are not formed even if dimers are readily available, which is nature’s way of regulating virus formation. Once we add a drug or in the presence of RNA, hexamers and gradually capsids can be generated.





(1) Ganem, D.; Prince, A. M. Hepatitis B Virus InfectionNatural History and Clinical Consequences. N. Engl. J. Med. 2004, 350, 1118− 29. (2) Global Hepatitis Report, 2017; World Health Organization: Geneva, 2017. (3) The Hepatitis B and Delta Viruses; Seeger, C., Locarnini, S. A., Eds.; Cold Spring Harbor Perspectives in Medicine; Cold Spring Harbor Laboratory Press: Cold Spring Harbor, NY, 2015. (4) Zlotnick, A.; Venkatakrishnan, B.; Tan, Z.; Lewellyn, E.; Turner, W.; Francis, S. Core Protein: A Pleiotropic Keystone in the HBV Lifecycle. Antiviral Res. 2015, 121, 82−93. (5) Pasek, M.; Goto, T.; Gilbert, W.; Zink, B.; Schaller, H.; Mackay, P.; Leadbetter, G.; Murray, K. Hepatitis B Virus Genes and Their Expression In E. Nature 1979, 282, 575−79. (6) Dryden, K. A.; Wieland, S. F.; Whitten-Bauer, C.; Gerin, J. L.; Chisari, F. V.; Yeager, M. Native Hepatitis B Virions and Capsids Visualized by Electron Cryomicroscopy. Mol. Cell 2006, 22, 843−50. (7) Wingfield, P. T.; Stahl, S. J.; Williams, R. W.; Steven, A. C. Hepatitis Core Antigen Produced in Escherichia coli: Subunit Composition, Conformation Analysis, and in Vitro Capsid Assembly. Biochemistry 1995, 34, 4919−32. (8) Newman, M.; Suk, F. M.; Cajimat, M.; Chua, P. K.; Shih, C. Stability and Morphology Comparisons o Self-Assembled Virus-lke Particles fom Wild-Type and Mutant Human Hepatitis B Virus Capsid Proteins. J. Virol. 2003, 77, 12950−60. (9) Su, P. Y.; Yang, C. J.; Chu, T. H.; Chang, C. H.; Chiang, C.; Tang, F. M.; Lee, C. Y.; Shih, C. HBV Maintains Electrostatic Homeostasis by Modulating Negative Charges from Phosphoserine and Encapsidated Nucleic Acids. Sci. Rep. 2016, 6, 38959. (10) Zhou, S.; Standring, D. N. Cys Residues of the Hepatitis B Virus Capsid Protein Are Not Essential for the Assembly of Viral Core Particles but Can Influence Their Stability. J. Virol. 1992, 66, 5393−98. (11) Packianathan, C.; Katen, S. P.; Dann, C. E.; Zlotnick, A. Conformational Changes in the Hepatitis B Virus Core Protein Are Consistent with a Role for Allostery in Virus Assembly. J. Virol. 2010, 84, 1607−15. (12) Wynne, S. A.; Crowther, R. A.; Leslie, A. G. The Crystal Structure of the Human Hepatitis B Virus Capsid. Mol. Cell 1999, 3, 771−80. (13) Ceres, P.; Stray, S. J.; Zlotnick, A. Hepatitis B Virus Capsid Assembly Is Enhanced by Naturally Occurring Mutation F97L. J. Virol. 2004, 78, 9538−43. (14) Zlotnick, A. Distinguishing Reversible from Irreversible Virus Capsid Assembly. J. Mol. Biol. 2007, 366, 14−8. (15) Ceres, P.; Zlotnick, A. Weak Protein−Protein Interactions Are Sufficient To Drive Assembly of Hepatitis B Virus Capsids. Biochemistry 2002, 41, 11525−31. (16) Katen, S.; Zlotnick, A. The Thermodynamics of Virus Capsid Assembly. Methods Enzymol. 2009, 455, 395−417. (17) Zlotnick, A.; Porterfield, J. Z.; Wang, J. C. To Build a Virus on a Nucleic Acid Substrate. Biophys. J. 2013, 104, 1595−604. (18) Newman, M.; Chua, P. K.; Tang, F.-M.; Su, P. Y.; Shih, C. Testing an Electrostatic Interaction Hypothesis of Hepatitis B Virus Capsid Stability by Using an in Vitro Capsid Disassembly/Reassembly System. J. Virol. 2009, 83, 10616−10626. (19) Kegel, W. K.; van der Schoot, P. Competing Hydrophobic and Screened-Coulomb Interactions in Hepatitis B Virus Capsid Assembly. Biophys. J. 2004, 86, 3905−13.

ASSOCIATED CONTENT

* Supporting Information S

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jcim.7b00471. Description and figures for ligand parametrization and principal component analysis (PDF)



REFERENCES

AUTHOR INFORMATION

Corresponding Author

*Tel: (+) 81-52-789-5288. Fax (+) 81-52-789-5118. E-mail: w. [email protected]. ORCID

Wataru Shinoda: 0000-0002-3388-9227 Author Contributions

S.L.R., S.O., and W.S. designed the research; S.L.R. performed the research; S.L.R., H.L., S.O., and W.S. analyzed the data; S.L.R. and W.S. wrote the paper. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We thank Profs. Yasuto Tanaka and Katsumi Omagari and Drs. Xibing He and Kota Kasahara for their useful suggestions. This research was supported by MEXT as “Priority Issue on Post-K I

DOI: 10.1021/acs.jcim.7b00471 J. Chem. Inf. Model. XXXX, XXX, XXX−XXX

Article

Journal of Chemical Information and Modeling (20) Zlotnick, A.; Mukhopadhyay, S. Virus Assembly, Allostery and Antivirals. Trends Microbiol. 2011, 19, 14−23. (21) Stray, S. J.; Bourne, C. R.; Punna, S.; Lewis, W. G.; Finn, M. G.; Zlotnick, A. A Heteroaryldihydropyrimidine Activates and Can Misdirect Hepatitis B Virus Capsid Assembly. Proc. Natl. Acad. Sci. U. S. A. 2005, 102, 8138−43. (22) Katen, S. P.; Chirapu, S. R.; Finn, M. G.; Zlotnick, A. Trapping of Hepatitis B Virus Capsid Assembly Intermediates by Phenylpropenamide Assembly Accelerators. ACS Chem. Biol. 2010, 5, 1125− 36. (23) Liu, N.; Zhao, F.; Jia, H.; Rai, D.; Zhan, P.; Jiang, X.; Liu, X. Non-Nucleoside Anti-HBV Agents: Advances in Structural Optimization and Mechanism of Action Investigations. MedChemComm 2015, 6, 521−35. (24) Zhou, Z.; Hu, T.; Zhou, X.; Wildum, S.; Garcia-Alcalde, F.; Xu, Z.; Wu, D.; Mao, Y.; Tian, X.; Zhou, Y.; Shen, F.; Zhang, Z.; Tang, G.; Najera, I.; Yang, G.; Shen, H. C.; Young, J. A.; Qin, N. Heteroaryldihydropyrimidine (Hap) and Sulfamoylbenzamide (SBA) Inhibit Hepatitis B Virus Replication by Different Molecular Mechanisms. Sci. Rep. 2017, 7, 42374. (25) Zlotnick, A.; Johnson, J. M.; Wingfield, P. W.; Stahl, S. J.; Endres, D. A Theoretical Model Successfully Identifies Features of Hepatitis B Virus Capsid Assembly. Biochemistry 1999, 38, 14644−52. (26) Holmes, K.; Shepherd, D. A.; Ashcroft, A. E.; Whelan, M.; Rowlands, D. J.; Stonehouse, N. J. Assembly Pathway of Hepatitis B Core Virus-Like Particles from Genetically Fused Dimers. J. Biol. Chem. 2015, 290, 16238−45. (27) Zlotnick, A. Theoretical Aspects of Virus Capsid Assembly. J. Mol. Recognit. 2005, 18, 479−90. (28) Smith, G. R.; Xie, L.; Lee, B.; Schwartz, R. Applying Molecular Crowding Models to Simulations of Virus Capsid Assembly. Biophys. J. 2014, 106, 310−20. (29) Venkatakrishnan, B.; Katen, S. P.; Francis, S.; Chirapu, S.; Finn, M. G.; Zlotnick, A. Hepatitis B Virus Capsids Have Diverse Structural Responses to Small-Molecule Ligands Bound to the Heteroaryldihydropyrimidine Pocket. J. Virol. 2016, 90, 3994−4004. (30) Klumpp, K.; Lam, A. M.; Lukacs, C.; Vogel, R.; Ren, S.; Espiritu, C.; Baydo, R.; Atkins, K.; Abendroth, J.; Liao, G.; Efimov, A.; Hartman, G.; Flores, O. A. High-Resolution Crystal Structure of a Hepatitis B Virus Replication Inhibitor Bound to the Viral Core Protein. Proc. Natl. Acad. Sci. U. S. A. 2015, 112, 15196−201. (31) Webb, B.; Sali, A. Comparative Protein Structure Modeling Using MODELLER. Curr. Protoc. Bioinf. 2014, 47, 5.6.1−5.6.32. (32) 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 2013, 29, 845−54. (33) Pall, S.; Abraham, M. J.; Kutzner, C.; Hess, B.; Lindahl, E. Tackling Exascale Software Challenges in Molecular Dynamics Simulations with GROMACS. In Solving Software Challenges for Exascale; Markidis, S., Laure, E., Eds.; Springer International Publishing: Cham, Switzerland, 2015; pp 3−27. (34) Abraham, M. J.; Murtola, T.; Schulz, R.; Páll, S.; Smith, J. C.; Hess, B.; Lindahl, E. Gromacs: High Performance Molecular Simulations through Multi-Level Parallelism from Laptops to Supercomputers. Softwarex 2015, 1−2, 19−25. (35) Best, R. B.; Zhu, X.; Shim, J.; Lopes, P. E.; Mittal, J.; Feig, M.; Mackerell, A. D., Jr Optimization of the Additive CHARMM All-Atom Protein Force Field Targeting Improved Sampling of the Backbone ϕ, ψ and Side-Chain χ1 and χ2 Dihedral Angles. J. Chem. Theory Comput. 2012, 8, 3257−73. (36) Mackerell, A. D., Jr.; Feig, M.; Brooks, C. L., III Extending the Treatment of Backbone Energetics in Protein Force Fields: Limitations of Gas-Phase Quantum Mechanics In Reproducing Protein Conformational Distributions In Molecular Dynamics Simulations. J. Comput. Chem. 2004, 25, 1400−15.

(37) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L. Comparison of Simple Potential Functions for Simulating Liquid Water. J. Chem. Phys. 1983, 79, 926−35. (38) Mayne, C. G.; Saam, J.; Schulten, K.; Tajkhorshid, E.; Gumbart, J. C. Rapid Parameterization of Small Molecules Using the Force Field Toolkit. J. Comput. Chem. 2013, 34, 2757−70. (39) Leekumjorn, S.; Sum, A. K. Molecular Simulation Study of Structural and Dynamic Properties of Mixed DPPC/DPPE Bilayers. Biophys. J. 2006, 90, 3951−65. (40) Parrinello, M.; Rahman, A. Polymorphic Transitions in Single Crystals: A New Molecular Dynamics Method. J. Appl. Phys. 1981, 52, 7182−90. (41) Kasahara, K.; Fukuda, I.; Nakamura, H. A Novel Approach of Dynamic Cross Correlation Analysis on Molecular Dynamics Simulations and Its Application to Ets1 Dimer−DNA Complex. PLoS One 2014, 9, e112419. (42) Kasahara, K.; Mohan, N.; Fukuda, I.; Nakamura, H. MDCC_Tools: Characterizing Multi-Modal Atomic Motions in Molecular Dynamics Trajectories. Bioinformatics 2016, 32, 2531−33. (43) Selzer, L.; Kant, R.; Wang, J. C.; Bothner, B.; Zlotnick, A. Hepatitis B Virus Core Protein Phosphorylation Sites Affect Capsid Stability and Transient Exposure of the C Terminal Domain. J. Biol. Chem. 2015, 290, 28584−93. (44) Berendsen, H. J. C.; Hayward, S. Collective Protein Dynamics in Relation to Function. Curr. Opin. Struct. Biol. 2000, 10, 165−69. (45) Wan, H.; Hu, J.-P.; Li, K.-S.; Tian, X.-H.; Chang, S. Molecular Dynamics Simulations of DNA-Free and DNA-Bound Tal Effectors. PLoS One 2013, 8, e76045. (46) Perilla, J. R.; Hadden, J. A.; Goh, B. C.; Mayne, C. G.; Schulten, K. All-Atom Molecular Dynamics of Virus Capsids as Drug Targets. J. Phys. Chem. Lett. 2016, 7, 1836−44. (47) Shepherd, D. A.; Holmes, K.; Rowlands, D. J.; Stonehouse, N. J.; Ashcroft, A. E. Using Ion Mobility Spectrometry−Mass Spectrometry To Decipher the Conformational and Assembly Characteristics of the Hepatitis B Capsid. Biophys. J. 2013, 105, 1258−67.

J

DOI: 10.1021/acs.jcim.7b00471 J. Chem. Inf. Model. XXXX, XXX, XXX−XXX