Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2024 Sep 26;128(46):11335–11344. doi: 10.1021/acs.jpcb.4c02700

Ebola Virus Matrix Protein VP40 Single Mutations G198R and G201R Significantly Enhance Plasma Membrane Localization

Michael D Cioffi †, Tej Sharma †, Balindile B Motsa ‡, Nisha Bhattarai ⊥, Bernard S Gerstman †,§, Robert V Stahelin ‡,∥, Prem P Chapagain †,§,*
PMCID: PMC11586905  PMID: 39326870

Abstract

graphic file with name jp4c02700_0009.jpg

Viral proteins frequently undergo single or multiple amino acid mutations during replication, which can significantly alter their functionality. The Ebola virus matrix protein VP40 is multifunctional but primarily responsible for creating the viral envelope by binding to the inner leaflet of the host cell plasma membrane (PM). Changes to the VP40 surface cationic charge via mutations can influence PM interactions, resulting in altered viral assembly and budding. A recent mutagenesis study evaluated the effects of several mutations and found that mutations G198R and G201R enhanced VP40 assembly at the PM and virus-like particle budding. These two mutations lie in the loop region of the C-terminal domain (CTD), which directly interacts with the PM. To understand the role of these mutations in PM localization at the molecular level, we performed both all-atom and coarse-grained molecular dynamics simulations using a dimer–dimer configuration of VP40, which contains the CTD–CTD interface. Our studies indicate that the location of mutations on the outer surface of the CTD regions can lead to changes in membrane binding orientation and degree of membrane penetration. Direct PI(4,5)P2 interactions with the mutated residues seem to further stabilize and pull VP40 into the PM, thereby enhancing interactions with numerous amino acids that were otherwise infrequently or completely inaccessible. These multiscale computational studies provide new insights at the atomic and molecular level as to how VP40–PM interactions are altered through single amino acid mutations. Given the high case fatality rates associated with Ebola virus disease in humans, it is essential to explore the mechanisms of viral assembly in the presence of mutations to mitigate the severity of the disease and understand the potential of future outbreaks.

1. Introduction

Viruses are capable of undergoing numerous genetic mutations, which can give rise to highly efficient variants with epidemic and pandemic potential. Ebola virus is one of the deadly viruses accounting for thousands of deaths and a 60 to 90% fatality rate.1−6 The virus’ genome encodes for seven structural proteins, and genetic sequencing during the outbreak revealed several naturally occurring mutations in these proteins.7−9 Genetic mutations can cause various changes in protein function and level of interaction, potentially leading to drastic changes in virulence.9 Gaining insights into mutation-induced variability in protein function and interaction is essential for isolating drug target areas to effectively combat the pathogen. For the Ebola virus, and specifically the Zaire species, the matrix protein VP40 is one such target as it is responsible for a variety of functions in the viral life cycle.8 VP40 consists of two main domains: the N-terminal domain (NTD) and the C-terminal domain (CTD), responsible for structural formation, membrane association, oligomerization, and budding.8,10−14 After replication in the host cell, VP40 is trafficked to the inner leaflet of the cell membrane where it will assemble filamentous, linear oligomeric structures through electrostatic interactions with the lipid bilayer14 as shown in Figure 1a. After many VP40 proteins assemble at the PM, the budding will be induced using the host lipid bilayer as the virus particle envelope. The creation of this lipid envelope and a successful budding process are essential for the virus particle to spread.14−16 The cell membrane consists of several lipid types,17,18 and VP40's ability to robustly bind to the PM is dependent on accessible amino acids, whereby changes to these amino acids through mutations can significantly affect the overall membrane association. Mutations located within VP40, especially C-terminal domain (CTD) regions, can influence how the protein assembles and interacts at the PM due to variations in structure and charge.13,14,16,19 It has been shown that the VP40 mutations C311A and C314A can cause increased phosphatidylserine (PS) lipid binding due to an increase in loop flexibility.20 PS lipids and phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2 or PIP2) are two PM lipids known to play key roles in the association of VP40 due to their strong, anionic headgroups in which they promote protein recruitment and protein stability, respectively.16,20−22 Consequentially, these minor amino acid mutations can result in large differences in lipid binding abilities due to the accessibility and abundance of cationic VP40 residues.

Figure 1.

Figure 1

Open in a new tab

(a) Schematic representation of the VP40 dimer association at the plasma membrane during viral assembly and budding. Created with BioRender.com. (b) Dimer–dimer structure of VP40 in the WT, G198R, and G201R configurations. The locations of residues 198 and 201 are shown in each structure and are labeled based on being in their WT or mutated forms. The WT residues are colored magenta; the G198R mutation, blue; and the G201R mutation, light blue.

Effects of single amino acid mutations in the 198–201 loop region have recently been characterized at the cellular level by the Stahelin group.23 That study found significant effects on membrane localization and virus-like particle (VLP) budding due to the single amino acid substitutions of G198R or G201R. Oligomerization and assembly at the PM are essential for budding to occur,24 and amino acid mutations that cause changes in the interactions with essential membrane lipids like PS and PIP2 may enhance the efficiency of this process through more stable binding sites or deeper membrane penetration.13,21,25 Oligomer forms have different functions,14 and VP40 structural arrangements were updated in 2020,24 so computational studies that extend beyond monomer and dimer interactions need to be further evaluated for a comprehensive analysis of these processes. With the monomeric and dimeric forms of VP40, CTD–CTD interface interactions with the PM are missed. In recent work, we computationally examined the hexamer–lipid interactions26 with the updated dimer–dimer–dimer arrangement of VP40. However, the effects of amino acid mutations in oligomeric VP40 have not yet been examined at the molecular level. In this study, we performed both all-atom (AA) and coarse-grained (CG) molecular dynamics (MD) simulations to further assess how membrane interactions due to cationic single amino acid mutations in the loop region can influence the association dynamics of dimer–dimer VP40 (Figure 1b). The AA MD simulations are beneficial for examining initial hydrogen bond interactions at the nanosecond time scale, whereas the CG MD simulations are beneficial for examining lipid–protein dynamics at the microsecond time scale. These computational studies provide molecular-level details of how single amino acid replacement can significantly alter oligomer membrane association, and the information can be valuable for developing preemptive strategies to combat future rapid viral mutations and inhibit VP40 function.

2. Computational Methods

2.1. VP40 Dimer–Dimer Model and Mutations

We obtained the crystal structures of the VP40 dimer from the Protein Data Bank (PDB) [PDB ID: 7JZJ], which also supplies the structure on the CTD–CTD interface for the end-to-end association of dimers. Ebola virus VP40 dimer–dimer structures were used for these studies because not many computational studies have been done on more than a single dimer VP40 since the Ebola structural configurations were updated in 2020.24 This structural arrangement also allows us to see how CTD–CTD interfaces influence membrane association and how minor mutations in the CTD regions may affect localization. The method of dimer-concatenation was the same as was done in ref (26), and the missing residues were inserted using Modeler.27,28 To determine the effects of single amino acid mutations on VP40–PM interactions, two cationic mutations were examined: G198 to R198 (referred to as G198R) and G201-R201 (termed G201R) are both located within the C-terminal domain of VP40. In addition to the wild-type (WT) dimer–dimer, we built two additional dimer–dimer structures each containing one of the mutations. The relative location of these mutations in the dimer–dimer structure can be seen in Figure 1b.

2.2. VP40 Plasma Membrane System

Using this dimer–dimer structure, we investigated three different protein–membrane complexes—VP40 WT-PM (referred to as just “WT” for the rest of the paper), VP40 G198R-PM (referred to as just “G198R” for the rest of the paper), and VP40 G201R-PM (referred to as just “G201R” for the rest of the paper)—in both the all-atom (AA) and coarse-grained (CG) resolution for a total of six different protein–membrane systems. All six systems were built with the CHARMM-GUI Web server.29 The first three AA systems (WT, G198R, G201R) were built with the Membrane Bilayer Builder plugin30,31 and the CHARMM36 force field.32,33 The second three CG systems (WT, G198R, G201R) were built with the Martini Maker plugin.34 The ElNeDynP22 force field was used for the CG systems, which include Martini 2.2 polar amino acids with an elastic network model, Martini 2.0 lipids, and polarizable water.35,36 The lipid bilayers were built to represent a typical mammalian PM composition17,37,38 of cholesterol (CHOL), phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylserine (PS), phosphatidylinositol 4,5-bisphosphate (PIP2), and sphingomyelin (SM). Each AA system had a lipid bilayer containing CHOL/PC/PE/PS/PIP2/SM in a ratio of 20:41:8:4:4:23 for the upper leaflet and 21:11:37:16:10:5 for the lower leaflet. For the CG systems, the upper leaflet of the PM has the same composition as the AA systems. For the lower leaflets, we slightly reduced the total percentage of PIP2 in the CG systems compared with that of the AA systems due to the favorability of electrostatic interactions in the CG model. Therefore, the CG system had a lipid bilayer containing CHOL/PC/PE/PS/PIP2/SM in a ratio of 20:41:8:4:4:23 for the upper leaflet and 22:13:38:17:6:5 for the lower leaflet. The VP40 dimer–dimer structure was initially placed ∼10 Å below the inner leaflet of the PM, and the C-terminal domain (CTD) regions were oriented toward the inner leaflet of the human PM.39 The initial VP40–PM configuration used for all systems can be seen (in AA and CG resolutions) in Figure 2.

Figure 2.

Figure 2

Open in a new tab

Initial configuration of the VP40 (WT) dimer–dimer structure and PM composed of various lipid types in the (a) AA and (b) CG representations. For the CG system, lipid headgroup PO4 beads are colored distinctly (PIP2 (red), PS (orange), PC (blue), PE (green), SM (purple), and CHOL (brown)).

2.3. All-Atom Molecular Dynamics Simulations

Each AA system was solvated in TIP3 water molecules and neutralized with 0.15 M KCl ions (∼950 K+ and ∼400 Cl–). The total lipids in the upper and lower leaflets of the PM were 800 and 782, respectively. The total number of atoms in each system was ∼730,000. The SHAKE algorithm was used to constrain covalent bonds involving hydrogen atoms. Energy minimization was completed over 10,000 steps followed by NPT (constant pressure/temperature) equilibration and production runs. Nonbonded interactions had a pair list distance of 1.6 nm and were updated every 10 steps. Force switching was used for van der Waals interactions where it was turned on at 1.0 nm and cut off at 1.2 nm. Long-range electrostatic interactions were treated with the particle mesh Ewald (PME) method. The pressure was maintained at 1 bar using the Langevin piston Nosé–Hoover method with a piston period and decay of 50 and 25 fs, respectively. Temperature was controlled at 303.15 K using Langevin dynamics temperature coupling and a friction coefficient of 1 ps–1. For production runs, timesteps of 2 fs were employed, and trajectories were collected for a total of 100 ns for all three systems. Replica simulations were completed for all systems by using the same parameters. Periodic boundary conditions were applied in all directions. These AA MD simulations were performed using NAMD3.40

2.4. Coarse-Grained Molecular Dynamics Simulation

Each CG system was solvated in water (∼52,000 CG water beads) and neutralized with NaCl ions (∼1200 Na+ and ∼600 Cl–). The total lipids in the upper and lower leaflet of the PM were 800 and 861, respectively. Each system consisted of ∼172,000 total CG beads. Minimization, equilibration, and production runs were completed using GROMACS.41,42 Energy minimization was completed for 10,000 total steps using the steepest decent algorithm. Equilibration was done over five iterations with an NPT ensemble. The timesteps were 0.2 0.5, 1, 0.15, and 2 fs, respectively, for a total number of steps of 500,000, 200,000, 100,000, 50,000, and 50,000, respectively, followed by NPT production runs. The Verlet cutoff scheme was used, updating the grid neighbor list every 20 steps. Periodic boundary conditions were applied in all directions. Reaction field electrostatics were employed with a Coulomb cutoff at 1.1 nm and a relative dielectric constant of 2.5. The van der Waals potential had a straight cutoff of 1.1 nm. The temperature was maintained at 303.15 K where temperature coupling was done using a velocity-rescaling thermostat. Pressure coupling was completed using a Parrinello–Rahman barostat, where the pressure was semi-isotropic and maintained at 1 bar. The leapfrog algorithm was used to integrate Newton’s equations of motion with a time step of 2 fs. For each system, trajectories were collected for a total of 20 μs. Replica simulations were completed for all systems using the same parameters. All trajectories, as well as system images, were visualized using Visual Molecular Dynamics (VMD).43

3. Results

3.1. Initial Association Dynamics

In each simulation, we made the initial conditions such that the VP40 dimer–dimer structures rapidly interact with the PM. We first analyzed the three AA systems to determine atomic-level details of the initial interactions of the VP40–PM equilibrated complexes. To quantify the results, we calculated the number of hydrogen bonds over time and the number of contacts over time between VP40 and PM for each of the three systems. Each lipid type interacts with the amino acids in different ways; some contribute to recruitment, whereas others facilitate strong binding.13 Therefore, we also determined the percent contribution of each lipid type to the total number of hydrogen bonds for the last 50 ns of the simulation. Each of these results is shown in Figure 3.

Figure 3.

Figure 3

Open in a new tab

Time evolution of (a) hydrogen bonds and (b) contacts between VP40 and the PM for all three systems (WT in magenta, G198R in blue, and G201R in light blue). The number of hydrogen bonds was calculated using VMD with an angle and distance criteria of 30° and 3.5 Å. The number of contacts was also calculated by using VMD with a distance criterion of 3.5 Å between any VP40 atom and any lipid atom. For each system type, the solid lines indicate the average number of hydrogen bonds or contacts between the three replica runs. The standard deviations of these averages are displayed in the lighter-shaded regions. (c) Total hydrogen bond contribution (%) of each lipid type for the last 50 ns of the all-atom simulations. Results are averaged between three replica runs for each system type, and the standard deviations are depicted by the error bars. Cholesterol interactions for this simulation were negligible and therefore were not included in this chart.

In the 150 ns simulations, all three systems have similar overall hydrogen bonding percentages. Noted by the seemingly “leveled off” number of hydrogen bonds and contacts around 75–100 ns, we considered the protein to be locally equilibrated with the membrane from 100 to 150 ns for all three systems. Figure 3c shows that PE, PS, and PIP2 lipid types dominate most of the hydrogen bonding from 100 to 150 ns, with PIP2 interactions slightly enhanced in G198R and G201R compared to WT. It is important to note that these time scales only represent the early dynamics during the association and that the average hydrogen bonds seem to be similar in all systems, considering the margin of error. The variation in initial hydrogen bonding with PIP2 is seemingly due to this local charge difference associated with the mutated amino acids, as described in ref (23) for single dimer systems. However, all-atom simulations limited to short time scales due to the larger system sizes of the dimer–dimer systems are not able to sufficiently sample the protein interactions at the membrane surface. Therefore, we further investigated lipid dynamics and interactions at a longer time scale by utilizing CG MD simulations.

3.2. Microsecond Timescale Association and Lipid Headgroup Interactions

To investigate the role of these single amino acid mutations on the dynamics of dimer–dimer VP40–PM interactions on a longer time scale, we performed 20 μs CG MD simulations. Because of the short distance between VP40 and the inner leaflet of the PM, interactions occur immediately. Our interests were in the longer time scale dynamics; therefore, these initial conditions were deemed sufficient. There were immediate visual differences in the number of nearby lipid contacts among the three systems. To quantify this, we calculated the total number of contacts over time between all lipids in the system (excluding CHOL) and VP40 residues. We determined that an interaction occurred if any lipid headgroup bead was within 7 Å45,46 of any VP40 residue, as was done in refs (26), (38), and (47). Similar to that of ref (48), lipid headgroups for PE and PS were chosen to be the phosphate groups (PO4 beads) plus the ethanolamine (NH3 bead) and serine (CNO bead) moieties, respectively. The phosphate group (PO4 bead) plus choline moiety (NC3 bead) was chosen for PC and SM. For PIP2, all phosphate groups (PO4, P1, and P2 beads) plus the inositol moiety (C1, C2, and C3 beads) were chosen. The number of contacts over time is shown in Figure 4.

Figure 4.

Figure 4

Open in a new tab

Time evolution of lipid–protein interactions. (a) Initial configuration of the VP40 dimer–dimer structure below the PM and the three systems at the end of the 20 μs simulation for each system. Phosphate group beads (PO4) for all lipid types (excluding CHOL) are shown in orange for the initial state (left) and dark orange for the final state (right). (b) Number of contacts (within 7 Å) between all lipid headgroup beads (excluding CHOL) and VP40 residues over time with a 10-frame moving average over 2000 total frames to smooth the data. For each system type, the average of two individual runs is depicted in the solid lines, whereas the results for the individual two runs are depicted in the lighter-shaded regions. The kernel density estimate plot on the right side indicates the distribution of contacts for each system type over the last 15 μs.

From Figure 4a, it is visually clear that the G198R system and especially the G201R system have more lipids within the interaction distance of VP40 than the WT system. From Figure 4b, the interactions in the WT and G198R systems seemed to settle around 3–4 μs, whereas they slightly increased in G201R beyond 5 μs. From 5 to 20 μs, the average and standard deviation of the WT system, G198R system, and G201R system were 74.4 ± 3.9, 86.0 ± 5.1, and 94.0 ± 3.7, respectively. We note that the number of interactions here is obtained for the contacts with CG beads, whereas they were obtained for atomic contacts in the all-atom systems in Figure 3. Between the two replica runs for each system type, the deviations are quite low in relation to the overall number of contacts. In the mutated systems, there is a clear increase in the average number of contacts between lipid headgroups and VP40 amino acids, especially for that of G201R.

3.3. Membrane Penetration

From visual analysis of our trajectories, the mutated structures (especially G201R) had more efficiently embedded themselves in the membrane than the WT structure. This effect can moderately be seen in Figure 4a, as the upper portion of the CTD regions in the mutated systems is more “hidden” by the lower leaflet phosphate groups as the lipids appear densely clustered. To quantify this, we calculated the center of mass (COM) distance in the z direction (direction perpendicular to the PM) between the backbone (BB) beads of VP40 and the lower leaflet phosphate group (PO4) beads for each system. The results are listed in Figure 5.

Figure 5.

Figure 5

Open in a new tab

Distance between the center of mass of VP40 backbone (BB) beads and the center of mass (COM) of lower leaflet lipid phosphate group (PO4) beads. The results for the WT system are in magenta, those for the G198R system are in blue, and those for the G201R system are in light blue. For each system, the solid lines show the average of the two individual runs with a 10-frame moving average over 2000 frames, and the lighter-shaded regions are the results of the individual runs. The kernel density estimate plot on the right side indicates the distribution of contacts for each system type in the last 10 μs.

The COM distance can give us an idea of the protein’s positioning relative to the membrane. We calculated the average and standard deviation of the COM distance for the last 15 μs of our simulations. In our simulations, the WT and G198R systems have an average COM distance between the PO4 and BB beads of 3.6 ± 0.2 and 3.5 ± 0.2 nm, respectively. The G201R system has a notably lower average COM distance of 3.2 ± 0.2 nm. The G201R structure seems to penetrate the PM lower leaflet more than that of the WT (Movie S1) or even G198R. Although the CG elastic network model for the protein prevents major conformational or structural changes, there is still a variance in the positioning of the protein and the PM between the WT and G201R systems. This penetration effect is likely caused by enhanced interactions with the PM due to the specific location of the mutated amino acid.

3.4. Individual Amino Acid Interactions

To determine which specific amino acids were contributing to the increased number of contacts (and the decreased COM distance) in the G201R system, we calculated the interaction frequency between any VP40 residue and any lipid headgroup (excluding CHOL). The lipid headgroups are the same as was defined in Section 3.2, and the calculation was completed from 10 to 20 μs for each system. Results for all lipid types (excluding CHOL) are shown in Figure 6.

Figure 6.

Figure 6

Open in a new tab

Bar graphs showing the percentage of time any VP40 residue is within 7 Å of any (a) PS, (b) PE, (c) PIP2, (d) SM, or (e) PC headgroup bead. The results are averaged between the two replica runs for each different system. The lipid type is indicated in the upper right of each graph, and the different systems are indicated by distinct colors: WT (magenta), G198R (blue), and G201R (light blue). The y axis shows the percentage of time any lipid headgroup bead was within 7 Å of any VP40 amino acid, the x axis shows the amino acid in contact, and the error bars depict the respective standard deviations. For clarity, only the residues with frequencies exceeding 20% in at least one system were considered. The red brackets highlight enhanced interactions in the mutated systems compared to WT. This calculation was completed for the last 10 μs of the simulations, after the systems were relatively stable. The p values were calculated between the WT and the mutants using Python with the Wilcoxon signed-rank test package. The p values for PS, PE, PIP2, SM, and PC between the WT and G198R system were 0.910, 0.470, 0.009, 0.008, and 0.055, respectively, showing a statistically significant difference (p < 0.05) for PIP2 and SM lipids between the two systems. The p values for PS, PE, PIP2, SM, and PC between the WT and G201R system were 0.004, 0.001, 0.021, 0.004, and 0.004, respectively, showing a statistically significant difference (p < 0.05) for all lipid types between the two systems.

Figure 6 displays the percentage of time (for the last 10 μs) that any lipid headgroup is within 7 Å (interaction distance) of any VP40 residue. Only residues with frequencies exceeding 20% in at least one system are displayed. Many of the same residues appear throughout the various lipid types. There are also many patches of lipid–residue interactions that are conserved (based on >20% interaction frequency) throughout each system type, such as S222–S228, K224-A229, and K274-T277 (extends from T266-N280 for PIP2 interactions). These residues are mostly located within basic loop 1 (219–233) or basic loop 2 (274–283), highlighting the importance of the basic loop regions for VP40 whether a mutation is present or not. However, when either mutation, especially the G201R mutation, is present, the interaction frequency increases for almost all amino acids listed for all given lipid types. Additionally, many enhanced interactions were found for each lipid type in the mutated systems, mainly for the G201R system. When considering an “enhanced interaction” as a difference of 20% or more between the mutated and WT systems, for PS lipids, there are enhanced interactions with K225, N227, and S228. For PE, there are enhanced interactions with K225, G226, S228, S233, K275, V276, and T277. For PC, there are enhanced interactions with K224, K225, G226, N227, and S228. For PIP2 lipids, there are many enhanced interactions. All residues from 195–202 (excl. 196) have a higher frequency of interaction (>20% difference from the WT), T232, S233, N280, and K326. In the G198R system, the interaction frequency with the patch of residues 195–200 (excluding 196) is much greater; notably, this is right where the mutated residue is located. The same goes for the 199–202 patch in the G201R system, located adjacent to the mutated residue. The mutated amino acids in their respective systems (R198 in G198R system and R201 in G201R system) are both enhanced interactions, and both interact with PIP2 for a large portion of the last 10 μs of the simulation. These mutated amino acids appear to create binding sites for PIP2 lipids.

3.5. Binding Mechanisms and Residue Flexibility

From contact analyses (Figures 4 and 6), it is evident that the G201R mutation causes significant changes in the way VP40 interacts with the PM. Additionally, even though the COM distance over time (Figure 5) changes are similar between the WT and the G198R systems, the contacts over time (Figure 4) increased, and the interaction frequency (Figure 6) was larger for almost all amino acids. This suggests that the connection between VP40 and the PM is more stable when the G198R mutation is present even if the COM distance is relatively unchanged. When comparing the two mutated systems to the WT, there were many enhanced lipid–protein interactions with >20% frequency (especially in the G201R system). Based on Figure 6, this increase in protein–membrane binding ability seems to be largely influenced by PIP2 lipids interacting with the mutated residues (R198 and R201) and occasionally the associated adjacent residues (T195-A202), resulting in a slightly deeper and more stable connection. PIP2 has a larger and longer headgroup than any of the other lipids located in this simulated PM. The headgroup also has a strong negative charge in the Martini model of −5. From mutating glycine (no charge) to arginine (+1 charge), as PIP2 lipids diffuse, they will eventually attach to these mutated residues by way of electrostatic interactions. A snapshot of PIP2 interacting with the mutated residues (R198 and R201) and the resulting enhanced lipid interactions, deeper membrane penetration, and altered binding orientation can be seen in Figure 7.

Figure 7.

Figure 7

Open in a new tab

Lipid interactions within 7 Å of VP40 at ∼18 μs. PIP2 is shown in red, PS in orange, PE in yellow, SM in green, and PC in purple. Lipid phosphate groups (PO4 beads) near VP40 are shown in brown. VP40 is shown in silver. (a, b) Lipids interacting with the WT and G198R systems at the CTD–CTD interface. For visual clarity of the major interaction exhibited by the R198 mutation, only the two middle monomers of VP40 are displayed. The mutated residue in panel b is shown in blue. (c, d) Side view of lipids interacting with the WT and G201R systems. For visual clarity of the major interaction exhibited by the R201 mutation, only one of the end monomers of VP40 is displayed. The mutated residue in panel d is shown in light blue. Note the clear increase in the number of lipid–protein interactions and slightly deeper membrane penetration in the mutated systems (b and d) as well as the minor change in binding orientation due to the mutation in panel d.

The direct PIP2 interactions with the mutated residues are very stable, as was quantified in Figure 6 by the interaction frequencies of these residues being very high. Because of the location of the G201R mutation in the CTD region, electrostatic interactions occur first with PIP2, which alter the VP40 binding orientation (similar to the orientational change found in ref (37) with mutated monomers), allowing other lipid interactions to take place. Although the location of both of these mutated residues enhances overall lipid interactions, this binding orientation effect is especially exaggerated in the G201R system due to residue 201’s relative location, slightly higher and more accessible on the outer surface compared to G198R, causing a more drastic change in COM distance after the interaction occurs. Because there is not much of a change in binding orientation due to the G198R mutation, the COM distance does not differ significantly from the WT. However, the direct interactions between the G198R residue and the PM cause an increase in the overall number of contacts, as described by Figures 4 and 6, due to the longer residency time offered by the direct R198–PIP2 interaction. Generally, both of these minor single amino acid mutations seem to cause more interfacial contacts between VP40 and the PM, therefore showing enhancements in lipid–protein binding. In the G201R system, the enhanced lipid–protein binding is due to a minor change in binding orientation and a longer protein–membrane residency time caused by PIP2 lipids directly interacting with the mutated residues. In the G198R system, this is only due to the direct PIP2 interactions with the mutated residues and the resulting longer protein–membrane residency time, increasing interactions but not to the extent of the G201R system because there is little or no change in binding orientation.

With the longer residency time of VP40, other important anionic lipids with smaller headgroups such as PS can make their way to suitable binding sites. Although almost all lipids had enhanced binding sites, PS lipids (and PIP2 as described above) were of special interest due to their essential role in VP40 localization and assembly at the PM.21,49 The more PS and PIP2 lipids interact with the protein, the greater localization, penetration, or curvature we should see as well as the already obvious enhanced interactions. From the interaction frequencies (Figure 6), it was clear that there was an increase in the influence of PS lipids on PM binding in the mutated systems compared to that in the WT, specifically for the G201R system. Notably, all residues interacting with PS at high frequency (>20%) are located adjacent to each other (residues 224–229) in the basic loop patch. As mentioned above, this amino acid patch (residues 224–229) appears to be important for interactions with all lipid types (Figure 6), not just PS, and must largely contribute to the increase in the number of contacts over time in the mutated systems. This patch of residues is also located higher in the CTD regions (closer to the inner leaflet of the PM). Through visual investigation, it was noticed that this amino acid patch was more accessible to various PM lipids due to the added PIP2 interactions anchoring VP40 to the membrane. PIP2 lipids interacting with these mutated residues, the associated 224–229 binding patch, and the root-mean-square fluctuations (RMSFs) of all residues can be seen in Figure 8.

Figure 8.

Figure 8

Open in a new tab

(a) Snapshot of each system at ∼15 μs. The binding amino acid patch (residues 224–229) is colored lime green, all PIP2 lipids within 7 Å of VP40 are colored red, all PS lipids within 7 Å of the 224–229 binding patch are colored orange, all lipid phosphate group (PO4) beads are colored brown, and the mutations (for their respective systems) are colored blue (R198) and light blue (R201). (b) RMSF of all residues per monomer for the last 10 μs of the simulation. The WT system is colored magenta, the G198R system is blue, and the G201R system is light blue. For each system, the solid lines show the average of the two individual runs. Error bars are shown in lighter color, which represent the standard deviations. From left to right, monomers labeled 1 and 4 are the “end” monomers, whereas monomers labeled 2 and 3 are the “middle” monomers (CTD–CTD interface). The PM-binding residue patches (residues 224–229) are highlighted in lime green.

On average, the coordinate fluctuations of the residues in the 224–229 patch (highlighted lime green in Figure 8b) for the last 10 μs of the simulation are higher than in other PM-binding regions, indicating residue excitations and coordinate fluctuations. Because of the transient nature of PS, PC, PE, and SM interactions (unlike that of PIP2 interactions), the higher average RMSF values of the amino acid patch describe the excited motion of those residues, briefly interacting with these transient lipids as they approach and relaxing as they diffuse. This patch is more affected in the G201R system due to the change in binding orientation, making it more accessible to the diffusing lipids. Both the G198R and G201R mutations are located further down the CTD region (further away from the PM) than in the 224–229 patch. When the direct PIP2 interactions take place with the mutated residues, this helps in anchoring the protein structure and the basic patch comes within interaction distance of the diffusing smaller anionic lipid headgroups like that of PS (example in Figure 8a). This patch sufficiently interacting with PM lipids appears to be essential for VP40 stability at the PM, and the effect is particularly enhanced in the G201R system.

4. Conclusions

Amino acid mutations can occur rapidly in viral proteins, causing significant changes in assembly and function. In this study, we investigated the molecular level interactions of single amino acid mutations in Ebola VP40. These mutations were recently shown to enhance interactions with the PM and increase assembly and budding of VP40 derived VLPs.23 Details of VP40 oligomer interactions with the PM needed further mechanistic evaluation, so we used a dimer–dimer structure of VP40 that includes the CTD–CTD interface. Our results indicate that mutating glycine to arginine in the CTD region can cause considerable changes to membrane localization by increasing the degree of interaction with PM lipids. Although this effect occurs with the G198R mutation present, it is especially noticeable with the G201R mutation. Many enhanced amino acid interactions occur in the mutated systems as well as higher interaction frequencies with those amino acids that already interact with the membrane in the WT system. This effect occurs with all of the phospholipids present in the system. The enhanced interactions, and greater frequency with those that already occur, seem to be due to direct electrostatic interactions between the mutated amino acids (R198 and R201) and PIP2 lipids. We find that PIP2 lipids interact with these mutated amino acids for most of the entire second half of the simulations (from 10 to 20 μs). As this takes place, the protein binding orientation slightly changes, and the COM distance between the protein structure (especially for G201R) and the PM is decreased. The decreased COM distance between VP40 and PM, especially with G201R, is indicative of a more efficient and stable binding process. As PS, SM, PE, and PC lipids diffuse, they are now able to interact more strongly with the patch of amino acids from residue 224–229, which they previously had less interactions with. The direct interactions between the mutated residues and PIP2 headgroups allowed for more efficient access to these binding sites because the VP40 structure in that region was closer to the membrane interface. The location of the mutation seems to be essential for this enhancement to take place. Both of these mutations are located on the outer surfaces of the CTD regions of VP40, allowing for access to the large PIP2 headgroups. G201R is slightly more accessible on the outer surface, “higher” in the CTD, and therefore shows even greater binding potential across all lipid types than G198R. Given the significant role that PIP2 plays in the VP40 interactions in the mutated systems and with known limitations of the general Martini 2.2 model,50 more accurate electrostatic interactions of PIP2 may be explored with the more refined Martini 3 model51 and the scFix flag in future investigations. Single amino acid mutations can cause substantial changes to protein function. Identifying amino acid sites in VP40 that are dramatically susceptible to change can be valuable for developing efficient and comprehensive therapeutics to inhibit VP40 assembly at the PM.

Acknowledgments

This research was funded by the NIH (AI158220) to R.V.S. and P.P.C. T.S. acknowledges the Dissertation Year Fellowship support from the University Graduate School at Florida International University.

Data Availability Statement

The PDB structure file for the dimer–dimer assembly (modeled with PDB 7JZJ) and the input files for the AA and CG simulations are included in Zenodo (https://zenodo.org/records/13377174). The CHARMM36m force field files used in this work are available from the MacKerell Lab (https://www.charmm.org/archive/charmm/resources/charmm-forcefields/) or the CHARMM-GUI web server. The Martini force field files are available from the Martini Web site (http://cgmartini.nl/) or the CHARMM-GUI web server. The simulation trajectory files (dcd and xtc files) are available upon request. A movie (Movie S1) is included in the Supporting Information.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jpcb.4c02700.

  • Movie S1 shows the 20 μs trajectories of the WT (left) and G201R (right) systems, with PIP2 (red), PS (orange), PE (yellow), PC (purple), and SM (green) lipids within 7 Å of VP40 shown. The location of the G201R mutation is colored light blue (MPG)

  • PDB structure file for the dimer–dimer assembly (modeled with PDB 7JZJ) (PDB)

  • A folder containing the input files for AA and CG simulations, which can be used to run the simulations or generate additional systems in CHARMM-GUI (ZIP)

Author Contributions

M.D.C. performed the molecular dynamics simulations and data analysis. T.S. assisted with all-atom simulations. All authors contributed to the analysis and interpretation of the results. B.S.G., R.V.S., and P.P.C. supervised the project. M.D.C. wrote the manuscript, and all authors contributed to editing.

The authors declare no competing financial interest.

Supplementary Material

jp4c02700_si_001.mpg (35.1MB, mpg)
jp4c02700_si_002.pdb (665.1KB, pdb)
jp4c02700_si_003.zip (41MB, zip)

References

  1. Johnson K. M.; et al. Isolation and partial characterisation of a new virus causing acute haemorrhagic fever in Zaire. Lancet 1977, 1 (8011), 569–571. 10.1016/S0140-6736(77)92000-1. [DOI] [PubMed] [Google Scholar]
  2. Zheng H.; et al. Ebolavirus classification based on natural vectors. DNA Cell Biol. 2015, 34 (6), 418–28. 10.1089/dna.2014.2678. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Feldmann H.; Geisbert T. W. Ebola haemorrhagic fever. Lancet 2011, 377 (9768), 849–62. 10.1016/S0140-6736(10)60667-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Baize S.; et al. Emergence of Zaire Ebola virus disease in Guinea. N Engl J. Med. 2014, 371 (15), 1418–25. 10.1056/NEJMoa1404505. [DOI] [PubMed] [Google Scholar]
  5. Wilson J. A.; Bosio C. M.; Hart M. K. Ebola virus: the search for vaccines and treatments. Cell. Mol. Life Sci. 2001, 58 (12–13), 1826–41. 10.1007/PL00000821. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Kadanali A.; Karagoz G. An overview of Ebola virus disease. North. Clin. Istanbul 2015, 2 (1), 81–86. 10.14744/nci.2015.97269. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Gire S. K.; et al. Genomic surveillance elucidates Ebola virus origin and transmission during the 2014 outbreak. Science 2014, 345 (6202), 1369–72. 10.1126/science.1259657. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Madara J. J.; et al. The multifunctional Ebola virus VP40 matrix protein is a promising therapeutic target. Future virology 2015, 10 (5), 537–546. 10.2217/fvl.15.6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Wong G.; et al. Naturally Occurring Single Mutations in Ebola Virus Observably Impact Infectivity. J. Virol. 2018, 93 (1), e01098-18 10.1128/jvi.01098-18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Adu-Gyamfi E.; et al. A loop region in the N-terminal domain of Ebola virus VP40 is important in viral assembly, budding, and egress. Viruses 2014, 6 (10), 3837–54. 10.3390/v6103837. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Jasenosky L. D.; et al. Ebola virus VP40-induced particle formation and association with the lipid bilayer. Journal of virology 2001, 75 (11), 5205–5214. 10.1128/JVI.75.11.5205-5214.2001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Dessen A.; et al. Crystal structure of the matrix protein VP40 from Ebola virus. EMBO J. 2000, 19 (16), 4228–36. 10.1093/emboj/19.16.4228. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Adu-Gyamfi E.; et al. The Ebola virus matrix protein penetrates into the plasma membrane. J. Biol. Chem. 2013, 288 (8), 5779–5789. 10.1074/jbc.M112.443960. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Bornholdt Z. A.; et al. Structural rearrangement of ebola virus VP40 begets multiple functions in the virus life cycle. Cell 2013, 154 (4), 763–774. 10.1016/j.cell.2013.07.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Soni S. P.; et al. The Ebola virus matrix protein deeply penetrates the plasma membrane: an important step in viral egress. Biophys. J. 2013, 104 (9), 1940–9. 10.1016/j.bpj.2013.03.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Stahelin R. V. Membrane binding and bending in Ebola VP40 assembly and egress. Front. Microbiol. 2014, 5, 300. 10.3389/fmicb.2014.00300. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Zachowski A. Phospholipids in animal eukaryotic membranes: transverse asymmetry and movement. Biochem. J. 1993, 294 (Pt 1), 1. 10.1042/bj2940001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. van Meer G.; Voelker D. R.; Feigenson G. W. Membrane lipids: where they are and how they behave. Nat. Rev. Mol. Cell Biol. 2008, 9 (2), 112–24. 10.1038/nrm2330. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. McCarthy S. E.; et al. Role for amino acids 212KLR214 of Ebola virus VP40 in assembly and budding. J. Virol 2007, 81 (20), 11452–60. 10.1128/JVI.00853-07. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Johnson K. A.; et al. Cysteine Mutations in the Ebolavirus Matrix Protein VP40 Promote Phosphatidylserine Binding by Increasing the Flexibility of a Lipid-Binding Loop. Viruses 2021, 13 (7), 1375. 10.3390/v13071375. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Adu-Gyamfi E.; et al. Host cell plasma membrane phosphatidylserine regulates the assembly and budding of Ebola virus. Journal of virology 2015, 89 (18), 9440–9453. 10.1128/JVI.01087-15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Johnson K. A.; et al. The Ebola Virus matrix protein, VP40, requires phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2) for extensive oligomerization at the plasma membrane and viral egress. Sci. Rep. 2016, 6, 19125. 10.1038/srep19125. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Motsa B. B.; et al. Minor electrostatic changes robustly increase VP40 membrane binding, assembly, and budding of Ebola virus matrix protein derived virus-like particles. J. Biol. Chem. 2024, 300, 107213 10.1016/j.jbc.2024.107213. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Wan W.; et al. Ebola and Marburg virus matrix layers are locally ordered assemblies of VP40 dimers. Elife 2020, 9, e59225 10.7554/eLife.59225. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Scianimanico S.; et al. Membrane association induces a conformational change in the Ebola virus matrix protein. EMBO J. 2000, 19 (24), 6732–41. 10.1093/emboj/19.24.6732. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Cioffi M. D.; et al. Role of phosphatidic acid lipids on plasma membrane association of the Ebola virus matrix protein VP40. Biochim Biophys Acta Mol. Cell Biol. Lipids 2024, 1869 (3), 159464 10.1016/j.bbalip.2024.159464. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Fiser A.; Do R. K.; Sali A. Modeling of loops in protein structures. Protein Sci. 2000, 9 (9), 1753–73. 10.1110/ps.9.9.1753. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Webb B.; Sali A. Comparative Protein Structure Modeling Using MODELLER. Curr. Protoc. Protein Sci. 2016, 86, 2.9.1–2.9.37. 10.1002/cpps.20. [DOI] [PubMed] [Google Scholar]
  29. Jo S.; et al. CHARMM-GUI: a web-based graphical user interface for CHARMM. Journal of computational chemistry 2008, 29 (11), 1859–1865. 10.1002/jcc.20945. [DOI] [PubMed] [Google Scholar]
  30. Wu E. L.; et al. CHARMM-GUI Membrane Builder toward realistic biological membrane simulations. J. Comput. Chem. 2014, 35 (27), 1997–2004. 10.1002/jcc.23702. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Lee J.; et al. CHARMM-GUI Membrane Builder for Complex Biological Membrane Simulations with Glycolipids and Lipoglycans. J. Chem. Theory Comput 2019, 15 (1), 775–786. 10.1021/acs.jctc.8b01066. [DOI] [PubMed] [Google Scholar]
  32. Mackerell A. D. Jr. Empirical force fields for biological macromolecules: overview and issues. J. Comput. Chem. 2004, 25 (13), 1584–604. 10.1002/jcc.20082. [DOI] [PubMed] [Google Scholar]
  33. Huang J.; MacKerell A. D. Jr. CHARMM36 all-atom additive protein force field: validation based on comparison to NMR data. J. Comput. Chem. 2013, 34 (25), 2135–45. 10.1002/jcc.23354. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Qi Y.; et al. CHARMM-GUI martini maker for coarse-grained simulations with the martini force field. J. Chem. Theory Comput. 2015, 11 (9), 4486–4494. 10.1021/acs.jctc.5b00513. [DOI] [PubMed] [Google Scholar]
  35. Marrink S. J.; et al. The MARTINI force field: coarse grained model for biomolecular simulations. J. Phys. Chem. B 2007, 111 (27), 7812–7824. 10.1021/jp071097f. [DOI] [PubMed] [Google Scholar]
  36. Monticelli L.; et al. The MARTINI coarse-grained force field: extension to proteins. J. Chem. Theory Comput. 2008, 4 (5), 819–834. 10.1021/ct700324x. [DOI] [PubMed] [Google Scholar]
  37. Yusoff M. A. M.; et al. Interaction of monomeric Ebola VP40 protein with a plasma membrane: a coarse-grained molecular dynamics (CGMD) simulation study. J. Mol. Graphics Modell. 2018, 82, 137–144. 10.1016/j.jmgm.2018.04.010. [DOI] [PubMed] [Google Scholar]
  38. Gc J. B.; et al. The Ebola virus protein VP40 hexamer enhances the clustering of PI(4,5)P(2) lipids in the plasma membrane. Phys. Chem. Chem. Phys. 2016, 18 (41), 28409–28417. 10.1039/C6CP03776C. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Del Vecchio K.; et al. A cationic, C-terminal patch and structural rearrangements in Ebola virus matrix VP40 protein control its interactions with phosphatidylserine. J. Biol. Chem. 2018, 293 (9), 3335–3349. 10.1074/jbc.M117.816280. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Phillips J. C.; et al. Scalable molecular dynamics on CPU and GPU architectures with NAMD. J. Chem. Phys. 2020, 153 (4), 044130 10.1063/5.0014475. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Van Der Spoel D.; et al. GROMACS: fast, flexible, and free. Journal of computational chemistry 2005, 26 (16), 1701–1718. 10.1002/jcc.20291. [DOI] [PubMed] [Google Scholar]
  42. Abraham M. J.; et al. GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 2015, 1, 19–25. 10.1016/j.softx.2015.06.001. [DOI] [Google Scholar]
  43. Humphrey W.; Dalke A.; Schulten K. VMD: visual molecular dynamics. J. Mol. Graphics 1996, 14 (1), 33–38. 10.1016/0263-7855(96)00018-5. [DOI] [PubMed] [Google Scholar]
  44. Alessandri R.; et al. Pitfalls of the Martini Model. J. Chem. Theory Comput 2019, 15 (10), 5448–5460. 10.1021/acs.jctc.9b00473. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Atilgan A. R.; et al. Anisotropy of fluctuation dynamics of proteins with an elastic network model. Biophys. J. 2001, 80 (1), 505–15. 10.1016/S0006-3495(01)76033-X. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Chao E.; et al. Structure-Based Simulation and Sampling of Transcription Factor Protein Movements along DNA from Atomic-Scale Stepping to Coarse-Grained Diffusion. J. Visualized Exp. 2022, (181), e63406 10.3791/63406. [DOI] [PubMed] [Google Scholar]
  47. Duncan A. L.; Corey R. A.; Sansom M. S. P. Defining how multiple lipid species interact with inward rectifier potassium (Kir2) channels. Proc. Natl. Acad. Sci. U. S. A. 2020, 117 (14), 7803–7813. 10.1073/pnas.1918387117. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Soni S. P.; Stahelin R. V. The Ebola virus matrix protein VP40 selectively induces vesiculation from phosphatidylserine-enriched membranes. J. Biol. Chem. 2014, 289 (48), 33590–7. 10.1074/jbc.M114.586396. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Herzog F. A.; et al. Improved Side Chain Dynamics in MARTINI Simulations of Protein-Lipid Interfaces. J. Chem. Theory Comput 2016, 12 (5), 2446–58. 10.1021/acs.jctc.6b00122. [DOI] [PubMed] [Google Scholar]
  50. Souza P. C. T.; et al. Martini 3: a general purpose force field for coarse-grained molecular dynamics. Nat. Methods 2021, 18 (4), 382–388. 10.1038/s41592-021-01098-3. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

jp4c02700_si_001.mpg (35.1MB, mpg)
jp4c02700_si_002.pdb (665.1KB, pdb)
jp4c02700_si_003.zip (41MB, zip)

Data Availability Statement

The PDB structure file for the dimer–dimer assembly (modeled with PDB 7JZJ) and the input files for the AA and CG simulations are included in Zenodo (https://zenodo.org/records/13377174). The CHARMM36m force field files used in this work are available from the MacKerell Lab (https://www.charmm.org/archive/charmm/resources/charmm-forcefields/) or the CHARMM-GUI web server. The Martini force field files are available from the Martini Web site (http://cgmartini.nl/) or the CHARMM-GUI web server. The simulation trajectory files (dcd and xtc files) are available upon request. A movie (Movie S1) is included in the Supporting Information.

Articles from The Journal of Physical Chemistry. B are provided here courtesy of American Chemical Society

RESOURCES