Single Mutation Effects on Conformational Change and Membrane

Jun 16, 2010 - Chinese Academy of Sciences, P.O. Box 800-204, Shanghai 201800, China, IBM Thomas J. Watson Research. Center, Yorktown Heights, New ...
0 downloads 0 Views 3MB Size
Download PDF
J. Phys. Chem. B 2010, 114, 8799–8806

8799

Single Mutation Effects on Conformational Change and Membrane Deformation of Influenza Hemagglutinin Fusion Peptides Jingyuan Li,†,‡,⊥ Payel Das,§ and Ruhong Zhou*,§,| Department of Physics, Zhejiang UniVersity, Hangzhou, 310027, China, Shanghai Institute of Applied Physics, Chinese Academy of Sciences, P.O. Box 800-204, Shanghai 201800, China, IBM Thomas J. Watson Research Center, Yorktown Heights, New York 10598 and Department of Chemistry, Columbia UniVersity, New York, New York 10027 ReceiVed: April 1, 2010; ReVised Manuscript ReceiVed: June 3, 2010

The single mutation effect on the conformational change and membrane permeation of influenza hemagglutinin fusion peptides has been studied with molecular dynamics simulations. A total of seven peptides, including wild-type fusion peptide and its six single point mutants (G1E, G1S, G1V, G4V, E11A, and W14A, all with no fusion or hemifusion activity) are examined systematically, which covers a wide range of mutation sites as well as mutant residue types (both hydrophobic and hydrophilic). The wild-type shows a kink structure (inversed V-shape), which facilitates the interaction between the fusion peptide and the lipid bilayer, as well as the interaction between the two arms of the fusion peptide. All mutants show a strong tendency toward a linear R-helix conformation, with the initial kink structure in the wild-type broken. More interestingly, one of the key hydrophobic residues around the initial kink region, Phe-9, is found to flip away from the membrane surface in most of these mutants. This conformational change causes a loss of key interactions between the original two arms of the inversed V-shape of the wild-type, thus disabling the kink structure, which results in the stabilization of the linear R-helix structure. The fusion peptides also display significant impact on the membrane structure deformation. The thickness of the lipid bilayer surrounding the wild-type fusion peptide decreases significantly, which induces a positive curvature of lipid bilayer. All the single mutations examined here reduce this membrane structural deformation, supporting the fusion activity data from experiments. Introduction Enveloped viruses, such as influenza A viruses, enter and infect cells by a process involving the fusion between viral and cellular membranes. This membrane fusion is mediated by viral envelope glycoproteins that act as membrane fusion proteins. A common feature of all viral influenza fusion proteins is that they bear a highly conserved hydrophobic fusion peptide (20 amino acids: GLFGA-IAGFI-ENGWE-GMIDG), which is most often located at the extreme N-terminus of the fusion-mediating polypeptide chain.1-4 The importance of the fusion peptide (or fusion domain) has long been recognized, as conservative point mutations in this domain dramatically alter fusion phenotypes;5,6 however, the exact molecular mechanism for these drastic changes due to single mutations is largely unknown.7 The unusually high content of glycine residues and their distribution in the fusion peptide sequence are particularly the focus of extensive studies that investigate the role of conserved spacing of glycine residues within the first 10 amino acids. The mutations of the glycine residues at the N-terminus of fusion peptide have been shown to induce a large effect on the fusion activity of hemagglutinin (HA). Many experimental studies have examined the effect of mutation in the glycine ridge of the fusion peptides Gly1 and Gly4. Both hydrophobic and hydrophilic residue substitutions on Gly1 can cause the impairment of fusion * Corresponding author. E-mail: [email protected]. † Zhejiang University. ‡ Chinese Academy of Sciences. § IBM Thomas J. Watson Research Center. | Columbia University. ⊥ Current address: Department of Chemistry, Columbia University, New York, NY 10027.

activity. For example, mutation of Gly1 to Val, Glu, Gln, or Lys completely impairs the fusion activity, whereas mutation to Ser induces the hemifusion.6,8-10 Only mutation to Ala does not change much of the fusion activity of HA.6,8-10 Meanwhile, for the second glycine, Gly4, the mutation G4V can impair the fusion,5 whereas the mutation G4E induces hemifusion.11 Recently, Han and Tamm linked the HA fusion peptide of strain X31 to a polar peptide (seven amino acids: -GCGKKKK) to prevent aggregation.12 Using this synthetic peptide P20H7, Han et al. successfully determined the structure of the HA fusion peptide in detergent micelles of dodecylphosphocholine (DPC) by NMR.13 They showed that these 20 amino acids fold into an inverted V-shape structure at both pH 5 and pH 7.4, with its hydrophobic pocket (Ile6, Phe9, Ile10, Trp14, Met17, and Ile18) facing the hydrophobic region of micelles. The kink structure has been further confirmed by the recently published NMR research work on fusion peptides.14 Although the kink in the wild-type structure appears to be quite fixed according to the NMR structures, it could also be dynamic as demonstrated by EPR spectroscopy.15 As shown in the experimental studies, the fusion peptide mutants with reduced fusogenic activities all seem to have this original kink structure disrupted to various extents. For example, G1V shows an approximately linear amphipathic helix; the kink structures of G1S and F9A are disrupted; and the structure of W14A is changed to a flexible kink that points away from the membrane instead of pointing inward.9,16,17 Why this kink structure is important for the activity of fusion peptide is still not very clear. It is also of great interest why mutations in the N-terminus, such as Gly1, can propagate into the kink region of the fusion peptide. The fusogenic activity also strongly depends on the peptide’s oblique insertion to the membrane and

10.1021/jp1029163  2010 American Chemical Society Published on Web 06/16/2010

8800

J. Phys. Chem. B, Vol. 114, No. 26, 2010

on the depth of its insertion.18,19 The active fusion peptide analogues insert obliquely into the phospholipid bilayer, whereas the inactive peptides adopt an angle more parallel to the bilayer surface.8,19-21 This oblique insertion mode is also shared by some other peptides with fusogenity, such as human immunodeficiency virus (HIV) and simian immunodeficiency virus (S1V) fusion peptides.15,19 On the other hand, the reduced insertion depth of fusion peptide mutants can cause the fusogenic activity to decrease.7,8,22 The stable conformation of the wild-type fusion peptide has been confirmed by molecular dynamics (MD) simulations.23 Herrmann and co-workers performed MD simulations on the fusion peptide in DPMC lipid bilayers, ∼1.5 ns MD simulation with an all-atom model (CHARMM/TIP3P), and a consecutive 18 ns simulation with a united-atom model (Gromos/SPC), and found that the V-shaped conformation is the most stable conformation.23 On the other hand, Lazaridis and co-workers used an implicit lipid model to study the structure and orientation of fusion peptides and found that the kinked structure is not stable for a monomeric fusion peptide (in addition, only its trimer form can induce the helices to insert deep into the membrane and at a steeper angle with respect to the membrane plane).24 Roux and co-workers have studied the wild-type fusion peptide in two different sizes of micelles and a lipid bilayer.25 By examining the average structure of the fusion peptide in these different membrane mimics, they found the peptide structures (mostly kinked) are similar, i.e., not sensitive to these different environments. Both Herrmann and Roux groups found that the fusion peptide can significantly affect the structure of the surrounding lipid bilayer.23 Skehel and co-workers also studied the mutation effects on the membrane and mutant structures.7 For the wild-type and other fusion peptides from H9 subtype (H9 influenza HA) and H2B (influenza B), the N-terminus 11 residues are helical, and they insert into the membrane with a tilt angle of 30°. This tilt insertion leads to bilayer thinning and disorder of alkyl chains. While for the nonfusogenic mutant peptides, such as G1E, G4V, and G1L, no tilted conformations were found, but instead, the peptides equilibrate parallel to the membrane surface. In addition, Woolf and co-workers have studied systems with various initial peptide structures at different locations and orientations across the membrane. From the different fusogenic behavior with different conformations, they found that the free energy of insertion can be expressed as a function of the depth of immersion.26 Although, these studies using both experimental and theoretical approaches offer valuable information about the structural and functional aspects of the fusion peptide, some important questions still remain unanswered. First of all, the effect of mutation in the glycine ridge on the kink structure of fusion peptide is unclear: Is (and by how much) the stability of the kink structure in various mutants reduced? Is there some particular residue that plays a major role in the disruption of the kink structure, and if yes, how? In addition, the impact of fusion peptide on the structure of lipid bilayer remains unresolved: Do the various configurations of mutant fusion peptides (e.g., the conformation, insertion angle, and insertion depth) impair the thinning of lipid bilayer? Despite extensive research performed on the configuration of nonfusogenic mutants, the direct connection between the structure of the mutant fusion peptide, its activity, and its adverse effect on lipid structure at atomic detail is still unavailable. In this paper, we use all-atom MD simulations to systematically study the single mutation effects of the fusion peptide in explicit lipid/water environment, with a total of seven peptides

Li et al. examined that covers a wide range of mutation sites as well as mutant residue types (both polar and nonpolar). These include six single mutants, G1E, G1S, G1V, G4V, E11A, and W14A, plus the wild-type one. The wild-type fusion peptide shows a stable inverted V-shape structure (kink structure), while all mutants show a strong tendency toward a linear helix conformation, with the initial kink structure broken. We find that the hydrophobic residues around the kink region in the wild-type display a unique pattern, in which residues Ile-6, Phe-9, Ile-10, and Trp-14 form the “hydrophobic core” facing the lipid bilayer. The conformational changes in the mutants of G1X (X ) E, S, V) and G4V share a similar behavior: the changes in the glycine ridge always cause the side-chain of Phe-9 to flip across the peptide plane. The change in the side-chain orientation of Phe-9 further disrupts the interaction among side-chains of the hydrophobic pocket, thus inducing the rearrangement of these residues. By studying the process of conformational change triggered by various mutations, we further probe how the mutations in the glycine ridge induce the conformational change of entire fusion peptide, which allows us to propose a general mechanism by which the changes in the N-terminus propagate into the kink region. This mechanism helps us to understand the significance of the kink structure and the structural flexibility of the fusion peptide, which is crucial to its function. The fusion peptides are also found to significantly impact the membrane structure. With the use of relatively large lipid bilayers in our simulations, the effect of the fusion peptide on the lipid bilayer at farther distances can be studied. As observed in previous simulations,23 the thickness of the lipid bilayer surrounding the wild-type fusion peptide decreases significantly, which induces a positive curvature of the lipid bilayer. It is interesting to note that the fusion peptide can induce the thinning of lipid bilayer even in the region with distance more than 40 Å from the fusion peptide. This finding provides an important path to understand how the fusion peptides disturb the membrane structure in a cooperative way, and how the various mutants reduce this longrange effect. System and Methods In this paper, we simulate the 20 N-terminal residues of the hemagglutinin protein of influenza strain X31 (GLFGA-IAGFIENGWE-GMIDG) with MD27-32 to study the effect of different mutations on its interaction with a dipalmitoylphosphatidylcholine (DPPC) bilayer. For this purpose, mutants G1E,5,7 G1S, G1V,9 G4V,5,7 E11A, and W14A16 were studied in addition to the wild type. The starting configuration of wild-type fusion peptide was taken from PDB entry 1IBN.13 All the mutations were designed based on the initial wild-type structure (Figure 1). The peptides were simulated with unprotonated Glu and Asp residues. [At pH 5, it is unclear what the protonation states should be for Asp and Glu. There is no definite data showing the protonation states of Glu-11, Glu-15, and Asp-19 in the fusion peptide. Considering that in a normal protein environment, the pKa for Asp is ∼3.9 and the pKa for Glu is ∼4.3,33 we followed the same unprotonated approach that Herrmann and co-workers used previously.23] In order to study the effect of fusion peptide on the lipid bilayer in a wide range and to eliminate the asymmetry of bilayer induced by the fusion peptide, the peptides were inserted in a previously equilibrated 512 DPPC lipid bilayer. The peptides were first placed in bilayer at the position and orientation suggested by Han et al.,13 i.e., the C atom of residue N12 was placed at the average position of the lipid phosphate groups. In addition, the first helix formed by residues 2-10 was positioned

Single Mutation Effects on Influenza HA Fusion Peptides

J. Phys. Chem. B, Vol. 114, No. 26, 2010 8801

Figure 1. The initial structure of wild-type fusion peptide system. (A) Side view. (B) Top view. The backbone of the peptide is in cyan, the hydrophobic side chains are in blue, and the charged side chains are in red. For clarity of the peptide, water molecules and some lipids are removed in B.

at an angle of 37° with the bilayer plane. Nine lipid molecules from the upper leaflet were removed to accommodate the peptide. The system was then solvated with water in a 146 × 143 × 65 Å3 box, resulting in a total of 112 111 atoms including 6 Na+ and 3 Cl- ions. To our knowledge, this is the largest system with the fusion peptide on DPPC bilayers simulated in explicit solvent. To relax the conformation of the peptide and eliminate bad contacts, the initial system was subjected to 20 000 steps of energy minimization. Next, a 50 ps long simulation was performed on the system with velocities initialized at 200 K, and then gradually increasing the temperature up to 300 K. The system was then equilibrated for 900 ps, during which atomic constraints were employed as follows: (i) first, both the peptide backbone and lipid bilayer were restrained, and the penetration of water was prevented by harmonic forces; (ii) next, harmonic forces were used to restrain only the peptide and the water, whereas those forces to lipids were released; (iii) then, the restraint on the water was reduced; (iv) subsequently, the peptide was progressively released from its initial configuration. The final structures were submitted to up to 15 ns MD simulations. All simulations were performed in NPT ensemble at 1 atm and 300 K using NAMD234 software. The CHARMM force field (c32b1 parameter set35) and the TIP3P water model were used. Electrostatic interactions were calculated using the particle mesh Ewald (PME) method with a distance cutoff of 12 Å. A switching function over the range of 10-12 Å was used for the truncation of the Lennard-Jones interactions. Result and Discussion Overall Conformational Change. Figure 2 shows the final conformations of the wild-type and mutants after 15 ns simulations, as well as a comparison of the time dependence of the backbone root-mean-square deviation (rmsdbb). Clearly, the major change in the overall conformation is the loss of the wildtype kink structure in all mutants. All the mutants adopt a linear helical structure. This is also evident from the comparison of the time evolution of rmsdbb of the wild-type peptide with that

Figure 2. Left: time evolution of the backbone rmsd of fusion peptide (wild-type, black; G1E, red; G1S, green; G1V, blue; G4V, yellow; E11A, brown; W14A, cyan). All the mutants quickly lose their initial structure (rmsd become larger than 3 Å) within 2.5 ns. Right: the corresponding final configuration of fusion peptides.

of various mutants (Figure 2). Our simulations indicate that the rmsdbb for the wild-type peptide remains nearly constant around 2 Å during the entire 15 ns MD simulation. On the other hand, rmsdbb for all the mutants increase to more than 3 Å within 2.5 ns, indicating the quick loss of the initial kink structure (see below for more analysis on the kink angle). The immediate conformational change in the mutants is likely due to the relatively short length of the peptides. Once the kink structure is lost, the mutants are found with the stretched linear helical structure that is fairly stable during the rest of the simulation. Compared to the wild-type fusion peptide, the stability of the kink structure is reduced in all the mutants we studied, while the final configuration of the mutants after 15 ns simulations beginning with the inverse V-shape might not properly reflect the structures of these mutants. For example, as reported in the NMR experiment, the kink structure is still largely retained in G1S.9 Kink/Tilt Angles. To further quantify the structural changes due to a mutation on the fusion peptide, the kink angles of fusion peptide and the tilt angles of oblique insertion were calculated. The kink angle is defined as the angle between the two vectors, denoted as vector2-10 and vector14-20, that are constructed by the backbone of the two helical parts of the fusion peptide

8802

J. Phys. Chem. B, Vol. 114, No. 26, 2010

Li et al.

Figure 3. Time evolution of the kink angle θ, the angle between the helix axes of residues 2-10 and residues 14-20 (A,B), and tilt angle δ, the angle between the helix axis of residues 2-10 and bilayer plane (C,D) of the wild-type, G1E, G1S, G1V, G4V, E11A, and W14A. Compared to the wild-type fusion peptide, all the mutants exhibit the preference for more linear configuration and less oblique insertion to various extents.

(residue 2-10 and 14-20). Consequently, the tilt angle is calculated from the angle between vector2-10 and bilayer plane (x-y plane). As shown in Figure 3A,B, the kink angle θ of the wild-type peptide fluctuates between 90° and 120°, with the angle remaining close to the original value of 95°. The results suggest that the initial kink structure is greatly retained. In contrast, the kink angles of all mutants quickly increase to higher values (larger than 160°), indicating the loss of the initial inversed V-shape. A close visual inspection of Figures 2 and 3A,B indicates that the mutants reach the linear configuration (kink angle starts increasing) and lose the initial kink structure within ∼4 ns (due to the relatively short peptide length, the loss of the kink structure only causes its rmsdbb to above 3 Å, while the wild-type maintains below 2 Å). For the convenience of the following study of unfolding mechanism of mutants and the identification of the critical events happening a priori to the conformational change, here we define the unfolding time as the first occasion when rmsdbb becomes larger than 3 Å (this might be somewhat arbitrary, but we are just using this as a way for comparison based in Figure 2). Accordingly, the unfolding times are ∼2.3 ns for G1E, ∼0.8 ns for G1S, ∼2.3 ns for G1V, ∼1.6 ns for G4V, ∼1.2 ns for E11A, and ∼4.2 ns for W14A. These results strongly suggest that, upon these single mutations, the stability of the initial kink structure is reduced, in good agreement with existing experimental data for fusion peptide structure. This is also consistent with the fusion activity

data, given that the kink structure is highly correlated with the fusion activity. Figure 3C,D reports the tilt angle of insertion of wild type and all mutants, which shows that the tilt angle δ of the wildtype fluctuates between 20° and 50°. This observation is in agreement with the experimental values of 25° and 38°, as obtained from the electron paramagnetic resonance (EPR) data on single spin-label peptides and the mapping of NMR data onto the best-fit EPR data correspondingly.9,12,13,36,37 On the contrary, most mutants tend to orient roughly parallel to the bilayer surface with significantly smaller tilt angles ( 120°), which adversely disrupts the original hydrophobic core formed by Ile-6, Phe-9, Ile-10, and Trp-14. In other words, those four mutants share a common feature in the conformational change, during which the Phe-9 side-chain flips across the plane upon the mutation at the glycine-ridge which disrupts the packing of the hydrophobic pocket. This result also indicates that, besides the hydrogen bonds of CO(G8) · · · NH(N12) and CO(F9) · · · NH(W14), the tight packing of the hydrophobic core (hydrophobic side-chains orientations), which is different from a regular linear helix, also plays an important role for the stability of the kink structure. Another interesting finding from previous experiment studies17 is that the mutant F9A largely retains the kink wild-type structure. In addition, both single mutants F9A and I10A are fusogenic. However, the double mutant F9A/I10A can reduce the cell-cell fusion activity.17 Taking together these experimental observations and our simulation results, the removal of the hydrophobic Phe-9 side-chain alone does not affect the stability of kink structure, whereas a large twist of Phe-9 sidechain (flip to the opposite side of the kink structure plane) or the removal of both Phe-9 and Ile-10 side-chains has a significant impact on the hydrophobic core formed by Ile-6, Phe9, Ile-10 and Trp-14, thus triggering the conformation change of the peptide. A closer look at the Phe-9 side-chain orientation also indicates its special role as compared to other important hydrophobic residues: Phe-9 is the only hydrophobic residue on the positive side along the normal of the peptide plane, while Ile6, Ile-10, and Trp-14 are all on the negative (or opposite) side (see Figure 6). Clearly, among residues comprising the hydrophobic core, Phe-9 is the most sensitive one to the environment. The special

pattern of hydrophobic side-chains surrounding the kink region indicates that they are sensitive to various environments or mutations. Together with the asymmetry of polarity along the fusion peptide (polar residues all locate in the middle of the fusion peptide), this pattern significantly contributes to the formation of the kink configuration of the fusion peptide. As shown in Figure 5B, a mutation in the glycine ridge can cause the Phe-9 side-chain to flip across the peptide plane, thus disrupting the original core formed by residues Ile6, Ile-10, and Trp-14. When time moves on, the flip can be so large that eventually the Phe-9 side chain points to the same side as Ile6, Ile10, and Trp-14, resulting in a final linear R-helix conformation of the fusion peptide. It is important to mention here that, in an ordinary amphipathic R-helix, the orientations of hydrophobic side-chains are almost parallel to each other on the same side.38,39 We think it is this special pattern of hydrophobic side-chains (i.e., Phe-9 on the opposite side of other hydrophobic residues) in the wildtype fusion peptide that plays a significant role in the formation of the inverse V-shaped kink structure and its interaction with lipid bilayers. Interaction with Membrane. It has been widely accepted that, the insertion of the HA fusion peptide affects the structure of the bilayer. One way to explore the effect of mutations on the function of fusion peptide is to examine the change in the bilayer thickness upon mutations. As described above, the mutations can significantly affect the conformation of the fusion peptide, as well as the oblique insertion to the membrane and its insertion depth. However, it is still not clear about the quantitative relationship between the conformation and the position of fusion peptide in lipid bilayer and its effect on the lipid bilayer structure. The use of relatively large lipid bilayers in this study allows us to probe this relationship to better understand the fusion activity. The average position of the phosphate groups was used to characterize the thickness of the bilayer. In the starting configuration, the phosphate to phosphate distance between leaflets is 37.6 Å. To examine how the peptide insertion affects the surrounding DPPC bilayer thickness, in each leaflet we define two groups of lipids: (i) lipids in the nearby region that are within 6 Å distance from the fusion peptide; and (ii) lipids in the distant region with a distance of >40 Å from the fusion peptide. To each group, the lipid molecules were assigned from each frame of the last 5 ns of MD trajectories. The z-positions of the lipid phosphates in the nearby and distant groups were used to calculate the density profile. For the wild-type system,

Single Mutation Effects on Influenza HA Fusion Peptides

J. Phys. Chem. B, Vol. 114, No. 26, 2010 8805

TABLE 1: Thickness of Lipid Bilayer in the Nearby Region and Distant Region Regarding the Fusion Peptide, and the Relative Thickness Reduced Compared to the Initial Configuration in Wild-Type (WT), G1E, G1S, G1V, G4V, E11A, and W14A Peptides

WT G1E G1S G1V G4V E11A W14A

nearby region (Å)

thickness reduced (Å)

distant region (Å)

thickness reduced (Å)

32.8 33.3 34.5 35.0 34.2 33.6 33.4

4.8 4.3 3.1 2.6 3.4 4 4.2

35.4 36 35.9 36.2 36.1 36.7 37.1

2.2 1.6 1.7 1.4 1.5 0.9 0.5

the phosphate-to-phosphate distance in the nearby region becomes 32.8 Å, with a decrease of 4.8 Å compared to the starting configuration. This observation indicates that the lipids surrounding the fusion peptide undergo a displacement along the axis normal to the bilayer. The rearrangement of the peptide surrounding lipids and the perturbation of the bilayer thickness might reflect the deformation and destabilization of the membrane essential for virus membrane fusion. On the other hand, in all the mutants, the decrease in the membrane thickness in the nearby region is significantly reduced (see Table 1). For example, the decrease in the DPPC bilayer thickness in G1V peptide is only half (or less than two-thirds for G1S and G4V peptides) of the value in the wild-type fusion peptide. We also investigate the phosphate-to-phosphate distance of the membrane in the distant region. It is important to note that the use of such a relatively large lipid bilayer with the size of 130 × 130 Å2 in the x-y direction enables us to observe the effect of the fusion peptide on the thickness of the membrane far from the peptide (as compared to much smaller membrane systems in previous simulations). For the wild-type system, the phosphate-to-phosphate distance in the distant region of the fusion peptide decreases from 37.6 Å (initial configuration) to 35.4 Å (a decrease of 2.2 Å). In the mutant systems, the decrease in the thickness also gets moderated (Table 1). Especially, for the E11A and W14A mutants, this value is only 0.9 Å and 0.5 Å, respectively. As mentioned above, the mutations can decrease the insertion depth of the fusion peptide to various extents (Figure 4). To quantify the insertion depth of the fusion peptide, we calculated the z-position of Leu-2 (center of mass of its sidechain). We interestingly find an almost linear relationship between the insertion depth and thinning of the lipid bilayer in the distant region (see Figure 7). This finding provides direct evidence for the effect of the insertion depth on the fusogenic activity of the fusion peptide: the deeper the fusion peptide gets into the lipid bilayer, the stronger is the effect on the structure of the lipid bilayer, thus facilitating the succeeding fusion of the viral membrane. This finding is in agreement with the study by Woolf and co-workers,26 in which the position of the wildtype peptide was artificially adjusted to investigate the effect on lipid bilayer thinning. The fusion peptide can also induce the membrane to adopt a positive curvature (positive second-order derivatives) that can facilitate the actual fusion by forming the stalk between two membranes.25,40 Using a relatively large lipid bilayer system, we have also studied the relative position of the central plane of the membrane in the nearby and distant regions by calculating the averaged position of last two carbon atoms of both hydrophobic tails of the assigned DPPC molecules. It is interesting to note that the z-positions of the central plane in the nearby region (within 6 Å from the fusion peptide) of all

Figure 7. The relationship between the z-position of Leu-2 and the thinning of a lipid bilayer in the distant region. The z-position of the center of mass of the Leu-2 side chain was used to represent the insertion depth of the fusion peptide.

TABLE 2: The Displacement of the Position of the Central Plane in the Nearby Region with Respect to the Distant Region in Wild-Type (WT), G1E, G1S, G1V, G4V, E11A, and W14A Peptides displacement (Å) WT G1E G1S G1V G4V E11A W14A

3.3 1.4 2.3 2.9 1.9 2.3 2.7

studied systems are found to be higher than that of the distant region (30 Å away from the fusion peptide), as one would expect for a positively curvatured membrane. For the wild-type system, the positive curvature with the central plane in the nearby region is 3.3 Å higher than that of the distant region. However, in several mutant systems, e.g., G1E, G1S, G4V, and E11A, this tendency toward higher positive curvature in the nearby region gets significantly weakened, as shown in Table 2. Taken together, the fusion peptide has a great effect on the structure of the surrounding DPPC bilayer, both by decreasing the thickness and inducing the positive curvature of membrane. On the basis of the extensive analysis of our simulation data, we propose that such a change in membrane structure is linked with the kink structure of the peptide. The deformation of the membrane structure becomes moderate in mutant systems compared to the wild-type system. For example, the behavior of membrane thickness in both the nearby and distant regions around G1E is similar to that of the wild-type system, but the tendency toward the positive curvature is greatly reduced. On the other hand, G1V has positive curvature similar to that of the wild-type system, but the effect on the membrane thickness is significantly reduced. Conclusion In this paper, we study the conformational change of the wildtype HA fusion peptide and its six mutants (G1E, G1S, G1V, G4V, E11A, and W14A) in the DPPC bilayer in addition to the effect of those mutations on the structure of the DPPC bilayer. The initial kink structure of the wild-type disappears in each of these six single mutants. All mutants show a strong tendency toward some linear R-helix conformations, and the time-dependent behavior of the kink angle is similar to that of the rmsd from the wild-type, indicating the kink angle variation

8806

J. Phys. Chem. B, Vol. 114, No. 26, 2010

dominates the process of the conformational change. On the other hand, the wild-type maintains the kink conformation during the entire simulation (15+ ns), and maintains both kink and tilt angles in agreement with previous experimental results. We also find that the special pattern of the side chain orientation of the hydrophobic residues around the kink region can facilitate the interaction with the lipid bilayer, as well as the interaction between residues belonging to the two arms of the fusion peptide, which is critical to the structural stability of the inverse V-shaped kink structure. The effect of mutant and wild-type fusion peptides on the structure of the lipid bilayer has also been examined. The thickness of the lipid bilayer around the wild-type fusion peptide decreases significantly. It is interesting to note that the lipids as far as 40 Å away from the fusion peptide also get affected. Also, the fusion peptide induces a positive curvature of the lipid bilayer. On the other hand, for the nonfusogenic mutants, these adverse effects on the lipid bilayer structure are significantly reduced in both nearby and distant regions, along with the positive curvature. These findings from our large-scale simulations not only complement recent experimental results but also provide new insights into the mechanism behind the influenza fusion peptide interaction with host cell membranes, which might be useful in developing future therapeutics for the influenza virus. Acknowledgment. We would like to thank Gary Whittaker, Alan Grossfield, Ian Wilson, Peter Palese, David Topham, Bruce Berne, and Ajay Royyuru for helpful discussions. We would also like to acknowledge the contributions of the BlueGene/L hardware, system software, and science application teams whose efforts and assistance made it possible for us to use the BlueGene/L supercomputer at the IBM Watson Center. References and Notes (1) Harter, C.; James, P.; Bachi, T.; Semenza, G.; Brunner, J. J. Biol. Chem. 1989, 264, 6459–6464. (2) Stegmann, T.; Delfino, J. M.; Richards, F. M.; Helenius, A. J. Biol. Chem. 1991, 266, 18404–18410. (3) Tsurudome, M.; Gluck, R.; Graf, R.; Falchetto, R.; Schaller, V.; Brunner, J. J. Biol. Chem. 1992, 267, 20225–20232. (4) Durrer, P.; Galli, C.; Hoenke, S.; Corti, C.; Gluck, R.; Vorherr, T.; Brunner, J. J. Biol. Chem. 1996, 271, 13417–13421. (5) Steinhauer, D. A.; Wharton, S. A.; Skehel, J. J.; Wiley, D. C. J. Virol. 1995, 69, 6643–6651. (6) Qiao, H.; Armstrong, R. T.; Melikyan, G. B.; Cohen, F. S.; White, J. M. Mol. Biol. Cell 1999, 10, 2759–2769. (7) Vaccaro, L.; Cross, K. J.; Kleinjung, J.; Straus, S. K.; Thomas, D. J.; Wharton, S. A.; Skehel, J. J.; Fraternal, F. Biophys. J. 2005, 88, 25– 36. (8) Wu, C. W.; Cheng, S. F.; Huang, W. N.; Trivedi, V. D.; Veeramuthu, B.; Assen, B. K.; Wu, W. G.; Chang, D. K. Biochim. Biophys. Acta 2003, 1612, 41–51. (9) Li, Y.; Han, X.; Lai, A. L.; Bushweller, J. H.; Cafiso, D. S.; Tamm, L. K. J. Virol. 2005, 79, 12065–12076. (10) Nieva, J. L.; Agirre, A. Biochim. Biophys. Acta 2003, 1614, 104– 115.

Li et al. (11) Rafalski, M.; Ortiz, A.; Rockwell, A.; van Ginkel, L. C.; Lear, J. D.; DeGrado, W. F.; Wilschut, J. Biochemistry 1991, 30, 10211–10220. (12) Han, X.; Tamm, L. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 13097– 13102. (13) Han, X.; Bushweller, J. H.; Cafiso, D. S.; Tamm, L. Nat. Struct. Biol. 2001, 8, 715–720. (14) Sun, Y. D. W. J. Am. Chem. Soc. 2009, 131, 13228–13229. (15) Tamm, L. K.; Lai, A. L.; Li, Y. Biochim. Biophys. Acta 2007, 1768, 3052–3060. (16) Lai, A. L.; Park, H.; White, J. M.; Tamm, L. K. J. Biol. Chem. 2006, 281, 5760–5770. (17) Lai, A. L.; Tamm, L. K. J. Biol. Chem. 2007, 282, 23946–23956. (18) Efremov, R. G.; Nolde, D. E.; Volynsky, P. E. V.; Chemyavsky, A. A.; Dubovskii, P. V.; Arseniev, A. S. FEBS Lett. 1999, 462, 205–210. (19) Bradshaw, J. P.; Darkes, M. J. M.; Harroun, T. A.; Katsaras, J.; Epand, R. M. Biochemistry 2000, 39, 6581–6585. (20) Lins, L.; Charloteaux, B.; Thomas, A.; Brasseur, R. Proteins: Struct., Funct., Genet. 2001, 44, 435–447. (21) Nguyen, N.; Tabruyn, S. P.; Lins, L.; Lion, M.; Cornet, A. M.; Lair, F.; Rentier-Delrue, F.; Brasseur, R.; Martial, J. A.; Struman, I. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 14319–14324. (22) Qiang, W.; Sun, Y.; Weliky, D. P. Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 15314–15319. (23) Huang, Q.; Chen, C.-L.; Herrmann, A. Biophys. J. 2004, 87, 14– 22. (24) Sammalkorpi, M.; Lazaridis, T. Biochim. Biophys. Acta 2007, 1768, 30–38. (25) Lagu¨e, P.; Roux, B.; Pastor, R. W. J. Mol. Biol. 2005, 354, 1129– 1141. (26) Jang, H.; Michaud-Agrawal, N.; Johnston, J. M.; Woolf, T. B. Proteins: Struct., Funct., Bioinf. 2008, 72, 299–312. (27) Zhou, R.; Berne, B. J.; Germain, R. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 14931–14936. (28) Liu, P.; Huang, X.; Zhou, R.; Berne, B. J. J. Phys. Chem. B 2006, 110, 19018–19022. (29) Zhou, R.; Eleftheriou, M.; Royyuru, A. K.; Berne, B. J. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 5824–5829. (30) Zhou, R. J. Mol. Graphics Modell. 2004, 22, 451–463. (31) Zhou, R.; Huang, X.; Margulis, C. J.; Berne, B. J. Science 2004, 305, 1605–1609. (32) Krone, M. G.; Hua, L.; Soto, P.; Zhou, R.; Berne, B. J.; Shea, J. E. J. Am. Chem. Soc. 2008, 130, 11066–11072. (33) Creighton, T. E. Proteins: Structure and Molecular Properties; 2nd ed.; W. H. Freeman and Company: New York, 1993. (34) Kumar, S.; Huang, C.; Zheng, G.; Bohm, E.; Bhatele, A.; Phillips, J. C.; Yu, H.; Kale, L. V. IBM J. Res. DeV.: Appl. MassiVely Parallel Syst. 2008, 52, 177–188. (35) MacKerell, A. D.; Bashford, D.; Bellott, M.; Dunbrack, R. L., Jr.; Evanseck, J. D.; Field, M. J.; Fischer, S.; Gao, J.; Guo, H.; Ha, S.; JosephMcCarthy, D.; Kuchnir, L.; Kuczera, K.; Lau, F. T. K.; Mattos, C.; Michnick, S.; Ngo, T.; Nguyen, D. T.; Prodhom, B.; Reiher, III, W. E.; Roux, B.; Schlenkrich, M.; Smith, J. C.; Stote, R.; Straub, J.; Watanabe, M.; Wiorkiewicz-Kuczera, J.; Yin, D.; Karplus, M. J. Phys. Chem. B 1998, 102, 3586–3616. (36) Macosko, J. C.; Kim, C.; Shin, Y. J. Mol. Biol. 1997, 267, 1139– 1148. (37) Lu¨neberg, J.; Martin, I.; Nu¨ssler, F.; Ruysschaert, J.-M.; Herrmann, A. J. Biol. Chem. 1995, 270, 27606–27614. (38) Tossi, A.; Sandri, L.; Giangaspero, A. Biopolymers 2000, 55, 4– 30. (39) Khandelia, H.; Langham, A. A.; Kaznessis, Y. N. Biochim. Biophys. Acta 2006, 1758, 1224–1234. (40) Kuzmin, P. I.; Zimmerberg, J.; Chizmadzkev, Y. A.; Cohen, F. A. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 7235–7240.

JP1029163