L565M mutation in HIV-1 glycoprotein 41 stabilizes the coiled-coil structure.

N-terminal and C-terminal heptad repeats (NHR and CHR) of HIV type 1 (HIV-1) glycoprotein 41 are known to be regions directly related to cell fusion during virus attack, and their complex core constructs a coiled-coil structure in the fusion process. In our recent studies, MT-4/17-3-6, a strain of HIV-1, showed the strong resistance to peptide fusion inhibitors compared with other strains such as MT-4/LAI, L-2 and CU98-26, and had a distinctive L565M mutation in the central region of NHR. To investigate the relationship between the mutation and resistance, we performed a molecular modeling of the coiled-coil of MT-4/17-3-6 by using energy minimization and molecular dynamics simulation based on the MT-4/LAI X-ray structure. As a result, we found that H564 in the NHR was pushed to the outer side by this mutation, and three hydrogen bond bridges of Y638-H564-E560-Q650 could be formed, enclosing the coiled-coil. The binding of peptide inhibitors would be disturbed by the structural stabilization of these bridges in MT-4/17-3-6.

The ectodomain of human immunodeficiency virus type 1 (HIV-1) glycoprotein 41 (gp41) has been investigated as a part of the spike protein directly related to the cell fusion of virus, and it consists of fusion protein (FP), N-terminal heptad repeats (NHR), interdomain fusion protein (IFP), C-terminal heptad repeats (CHR), and other regions [1]. Studies by electron microscopy [2], [3], [4] demonstrated that the spike protein of HIV-1 formed a trimer complex in fusion process. In addition, molecular structure analyses [5], [6], [7] revealed that NHR and CHR had similar helical conformations packed in an anti-parallel manner, and that the three pairs of NHR and CHR in a trimer gp41 were arranged tightly in a trimeric coiled-coil. Three NHR regions were coiled in the inner side and the three CHR regions covered the NHR coil on the outer side. This coiled-coil complex could be constructed by conformational changes of glycoprotein 120 interacting with CD4 and CXCR4 coreceptor [8], [9]. Formation of the coiled-coil causes the viral envelope to approach the target cell membrane via tight binding, and allows membrane fusion associated with rapid entry of the viron core into the target cell [1], [10], [11]. Thus, the NHR–CHR interactions are very important for investigating the fusion mechanism of the virus entry process and for designing the inhibitors against the cell fusion.

One complicating fact is that HIV-1 is genetically variable [12]. Among HIV-1 genome, env gene has been widely used for the classification by phylogenetic analysis [13]. Based on the sequence of complete viral genomes [14], HIV-1 nucleotide sequences circulating in the world have been divided into three groups: M (major); O (outlier); and N (non-M or O). Within the most prevalent group, M, there are at least eight discrete subtypes (A, B, C, D, F, G, H, and J) and 14 circulating recombinant forms. Different HIV-1 subtypes and recombinant forms are generally associated with different countries and continents [15], [16], and the similarities among these subtypes are distributed widely [17], [18]. It was also reported that homologous synthetic peptides of the NHR and CHR regions, called N34 and C34, had binding activities to CHR and NHR, respectively [19], [20], [21]. Therefore, these synthetic peptides play an important role as experimental tools for studies of cell fusion and its inhibition because of the affinities to gp41.

We have studied the fusion mechanism to identify the important factors for the design of inhibitors, and reported that strain MT-4/17-3-6 [22] from a Japanese patient with acquired immunodeficiency syndrome (AIDS) showed stronger resistance against both N34 and C34 peptides than other strains, MT-4/LAI, L-2, and CU98-26 [23]. In this article, to investigate the basis of the resistance of MT-4/17-3-6, an NHR–CHR complex model was predicted stereo-chemically by molecular modeling with energy minimization and dynamics simulation. We compared it with MT-4/LAI model, especially focusing the central region of the coiled-coil, which was not markedly affected by any other parts of the envelope protein. This central region includes the M565 of MT-4/17-3-6, which is mutated distinctively from the L565 of MT-4/LAI, L-2 and CU98-26. The amino acid sequences of NHR and CHR of these strains are summarized in Table 1 .

Table 1

Amino acid sequences of the NHR and CHR regions of HIV-1s used in our previous research [23]

HIV-1	Subtype	Amino acid sequencea
		NHR	CHR

A dash (-) indicates identity with the corresponding MT-4/LAI amino acid residue.

Materials and methods

Molecular modeling of MT-4/17-3-6 NHR–CHR complex. The initial model for the MT-4/17-3-6 complex was constructed by using of residue replacements on the coordinate sets 1AIK in Protein Data Bank [5], a NHR–CHR complex X-ray model of MT-4/LAI. Structural optimization was performed for this model by an energy minimization method with an MMFF94 force field [24] considering water molecules within 10 Å around protein atoms. The energy cutoff distance was set at 10 Å and the dielectric constant was distance dependent based on the value of 1.0 for the protein atoms and 80.0 for the solvent atoms. The 1AIK X-ray model of MT-4/LAI was optimized for comparison with the MT-4/17-3-6 model by the same method.

100 ps molecular dynamics simulation was performed for each optimized model with the water molecules at 300 K by using a 0.002 ps time step, the MMFF94 force field and NVT method (number of particles, volume, and temperature were fixed) [25], [26], [27]. Before these equilibrium iterations, 1 ps heating iterations were employed to consider the stable equilibration. The potential energy of each total molecular system after the 100 ps simulation was −5.85 × 104  kcal/mol and −5.67 × 104  kcal/mol, including 3233 and 3252 water molecules for MT-4/17-3-6 and MT-4/LAI, respectively. Hydrogen bond analyses were based on 100 conformations sampled every 0.5 ps during the equilibrium iterations from 50 to 100 ps, in which the potential energy seemed to settle down clearly.

All operations and calculations were performed using a graphical package for molecular structure analyses, Molecular Operating Environment (CCG, http://www.ccg.com/).

Results and discussion

A possible structure of the MT-4/17-3-6 NHR–CHR complex predicted from energy minimization is shown in Fig. 1 A. We found three sets of hydrogen bond bridges which seemed to fix the complex effectively. The bridge consisted of three hydrogen bonds of Y638Oη–H564Nδ1, H564Nε2–E560Oδ1 and E560Oδ2–Q650Nε2, and three of the bridges enclosed the complex on its surface. The alignment of functional groups in the bridge was seemed to be similar to the charge relay system of serine protease active sites. In the structure of the MT-4/LAI NHR–CHR complex, shown in Fig. 1B, only E560Oδ2–Q650Nε2 was formed and Y638Oη–H564Nδ1 and H564Nε2–E560Oδ1 were not formed. Especially, the phenyl ring of Y638 and the imidazole ring of H564 existed side by side, but Oη and Nδ1 or Nε2 were not placed where they could form a hydrogen bond. The sidechain of H564 in the MT-4/17-3-6 complex shifted slightly to the outer side of the complex compared with the MT-4/LAI complex, and it was thought that the formation of hydrogen bonds of Y638Oη–H564 Nδ1 and H564Nε2–E560Oδ1 was enable by this shift. The total interaction energy between NHR and CHR was −4.52 × 102  kcal/mol in MT-4/17-3-6 and −4.05 × 102  kcal/mol in MT-4/LAI; thus, the NHR and CHR of MT-4/17-3-6 were combined more strongly than these of MT-4/LAI.

Fig. 1

Optimized models of MT-4/17-3-6 and MT-4/LAI NHR–CHR complex. The optimized structures of the NHR–CHR regions of MT-4/17-3-6 and MT-4/LAI are shown by tube models with ball and stick models of Y638, H564, E560, and Q650 in A and B respectively. The helical structures of the NHR core coiled-coil are colored by red, blue and green, and the helical structures of CHR are colored by purple, light blue and yellow. Possible hydrogen bonds are shown by black dotted lines.

The statistical analyses of the formation of these hydrogen bonds during the later iterations of the molecular dynamics simulations are summarized in Table 2 , and a superimposed stereo view of the most stable structure in the simulations is shown in Fig. 2 . In the MT-4/17-3-6 complex, hydrogen bonds were formed in the conformation with probabilities of 82.3% and 59.7% for Y638Oη–H564Nδ1 and H564Nε2–E560Oδ1 respectively, which are very high probabilities compared with these for MT-4/LAI. Just beside H564, the L565M mutation was found in MT-4/17-3-6 compared with MT-4/LAI. In the MT-4/LAI complex, as shown by dark blue and green sticks in Fig. 2, the L565 sidechain contacted CHR sidechains of I635 and I642 by hydrophobic interaction, and this interaction was fixed tightly by the existence of a hydrogen bond between Q567 and T639. On the other hand, in the MT-4/17-3-6 complex, as shown by orange and yellow sticks, a longer hydrophobic sidechain of M565 existed instead of the shorter sidechain of leucine. Two isoleucines, glutamine and threonine in the MT-4/17-3-6 complex remained at the same positions in the MT-4/LAI complex and fixed the sidechain of methionine. In addition, the I635 residue, especially its sidechain, was close to the sidechain of M565 because of loss of the branched structure in the sidechain L565.

Table 2

Probability of hydrogen bond formation during the molecular dynamics simulation

Models	Probability of hydrogen bond formation (%)a
	Y638Oη–H564Nδ1	H564Nε2–E560Oδ1	E560Oδ2–Q650Nε2
MT-4/17-3-6	82.3	59.7	74.3
MT-4/LAI	1.0	0.3	76.7

Hydrogen bond analyses were based on 100 conformations sampled every 0.5 ps during the equilibrium iterations from 50 to 100 ps. The criteria for hydrogen bond formation were as following; bond length of N–O < 3.6 Å and bond angle of N–H–O > 110°.

Fig. 2

A superimposed stereo view of the molecular models of MT-4/17-3-6 and MT-4/LAI. The most stable structures in our molecular dynamics simulations of MT-4/17-3-6 and MT-4/LAI were superimposed by a fitting of all mainchain atoms, and only parts related to the hydrogen bond bridges of MT-4/17-3-6 are shown. The NHR and CHR regions of MT-4/17-3-6 are shown by orange and yellow sticks, and these of MT-4/LAI by dark blue and green sticks, respectively. Oxygen, nitrogen, and sulfur atoms are colored by red, light blue, and white. Each hydrogen bond and its length is shown by a green line and numerical value, and the atoms involved in these hydrogen bonds and sidechains of the 565th residues are shown by a ball and stick model.

We found that H564 in the NHR was pushed to the outer side in the MT-4/17-3-6 complex, and three hydrogen bond bridges of Y638–H564–E560–Q650 could be formed and seen to enclose the coiled-coil. As shown in detail in Fig. 2, the structural difference from the MT-4/LAI complex in this region was that the M565 mainchain and the H564 of NHR were pushed to the outer side (from dark blue to orange) and the neighboring Y638 and T639 of CHR were shifted to the opposite direction (from green to yellow). As a result, the imidazole ring of H564 in the MT-4/17-3-6 complex was placed in the position and direction where it could be stabilized by hydrogen bonds with both Y638 in the CHR region and E560 in the same NHR region.

We tentatively optimized another virtual model with mutation of only L565M from MT-4/LAI, and also found similar Y638–H564 hydrogen bonds as in the MT-4/17-3-6 model (data not shown). As shown in Table 1, E630Q, S641G, and N651I mutations were found distinctively in the CHR region of MT-4/17-3-6, but all of these residues were placed on the outer side, which had no direct interaction with the inner NHR coils. S641G and N651I were adjacent to Y638 and Q650. It seemed that these mutations might contribute to the formation of hydrogen bond bridges by increasing the flexibilities of the neighboring mainchain of Y638 and decreasing the hydrogen bonding to the Q650 sidechain via water molecules, respectively.

Therefore, it was thought that the L565M mutation would be a major factor affecting the coiled-coil stabilization by these bridges, and the binding of peptide fusion inhibitors might be distinctively disturbed by this stabilization in MT-4/17-3-6. The parallel stacking of two rings of H564 and Y638 was also found in the X-ray structure [7], but no hydrogen bond between these sidechains was shown in the structure, because the two rings were placed too closely, as was also true in the MT-4/LAI model. We think that the hydrogen bond between H564 and Y638 fixes the sidechain of H564 in the position that makes it easy to form the next hydrogen bond with E560, and that the positional relationship of the two rings is very important for the total formation of the hydrogen bond bridges shown in the model for MT-4/17-3-6.

Coiled-coil trimeric structures are commonly found in the spike proteins of not only HIV but also other corona viruses, including influenza virus, Ebola virus, and SARS virus [28]. In these coiled-coil structures, it seemed that HIV had the most tightly combined complex of NHR and CHR, and therefore the mode of interaction between NHR and CHR of HIV may be most effective for the trimeric formation at the time of virus attack. We found in our previous research that the differences of gp41 resistance to peptide fusion inhibitors were larger for intra-subtype comparisons, such as MT-4/LAI and MT-4/17-3-6, than for inter-subtype comparisons, such as MT-4/LAI and CU98-26 (subtypes are shown in Table 1) [23]. These large intra-subtype differences can be explained by our model, which showed that the L565M mutation could stabilize the coiled-coil without affecting the structure of the protein widely. We previously reported that the resistance to peptide fusion inhibitors differed markedly between MT-4/LAI and L-2 in spite of the fact that they had the same sequences of NHR and CHR (Table 1). It is not thought that all of the differences between these virus strains could be explained in here clearly, because the interactions between the NHR–CHR complex and the other regions of the spike protein were not considered in our analyses. However, by the focusing on the L565M mutation in the inner and central moiety, which would be not affected by the other regions, we found that this mutation might be an important factor for the resistance of MT-4/17-3-6 to peptide inhibitors with respect to coiled-coil formation and cell fusion.