Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2019 Feb 5;27(2):253-267.e8.
doi: 10.1016/j.str.2018.10.009. Epub 2018 Nov 21.

Mechanism of Enhanced Immature Dengue Virus Attachment to Endosomal Membrane Induced by prM Antibody

Free PMC article

Mechanism of Enhanced Immature Dengue Virus Attachment to Endosomal Membrane Induced by prM Antibody

Melissa Wirawan et al. Structure. .
Free PMC article


Dengue virus (DENV) particles are released from cells in different maturation states. Fully immature DENV (immDENV) is generally non-infectious, but can become infectious when complexed with anti-precursor membrane (prM) protein antibodies. It is unknown how anti-prM antibody-coated particles can undergo membrane fusion since the prM caps the envelope (E) protein fusion loop. Here, we determined cryoelectron microscopy (cryo-EM) maps of the immDENV:anti-prM complex at different pH values, mimicking the extracellular (pH 8.0) or endosomal (pH 5.0) environments. At pH 5.0, there are two structural classes with fewer antibodies bound than at pH 8.0. These classes may represent different maturation states. Molecular simulations, together with the measured high-affinity pr:antibody interaction (versus the weak pr:E interaction) and also the low pH cryo-EM structures, suggest how antibody:pr complex can dislodge from the E protein at low pH. This exposes the E protein fusion loop enhancing virus interaction with endosomes.

Keywords: cryo-EM; dengue virus; enhancement of infection; human antibody; immature dengue virus; maturation.

Conflict of interest statement


The authors declare no competing interests.


Figure 1.
Figure 1.. Anti-prM HMAb 1H10 enhanced the attachment of immDENV to anionic liposome and enhanced infection of immDENV in monocytic THP-1 cell.
(A) Fab 1H10 enhanced attachment of the fully immDENV to anionic liposomes which composition mimic that of the endosome. Presence of virus was detected in different fractions of the sucrose gradient in a virus-liposome co-floatation assay by anti-E antibody. At all pH 8.0 conditions, with or without liposome and Fab, immDENV remained in the middle fractions where the viruses were loaded initially onto the gradient. At pH 5.0, immDENV co-floated with the liposomes in the top 4 fractions indicating some interactions with the liposomes. When Fab 1H10 was added to the virus-liposome mixture at pH 5.0, all immDENV was detected only in the topmost fraction indicating enhanced interaction with liposomes. The assay was repeated three times and showed consistent results. (B) IgG 1H10 enhanced infection of immDENV3 in THP-1 cells at equal or higher than molar ratio of 1 antibody per viral prM molecule. See also Figure S2.
Figure 2.
Figure 2.. Samples for cryoEM studies and resolution of the cryoEM maps.
(A) Micrographs of the uncomplexed immDENV3 and the immDENV3:Fab 1H10 complexes at pH 8.0 and 5.0. The uncomplexed immDENV3 at pH 8.0 micrograph showed the sample contains mostly spiky fully immature virus. At pH 5.0, the uncomplexed virus becomes unstable and the surface of particles looked smoother. In the immDENV3:Fab 1H10 complex samples, at both pH 8.0 and 5.0, particles appear to have a bigger diameter and spikier, indicating the binding of Fab molecules. (B) [Top] Fourier Shell Correlation (FSC) curve of immDENV:Fab 1H10 at pH 8.0 cryoEM map determined by gold-standard FSC procedure. The 12 Å resolution of the map is estimated by using FSC 0.143 correlation cut-off. [Bottom] Fourier shell correlation curve of class I and II immDENV:Fab 1H10 at pH 5.0 cryoEM maps. The FSC curve was plotted from two reconstructed maps of two half-datasets of the final iteration step and the resolution was estimated by using 0.5 cut-off. Class I and II pH 5.0 complexed structures were determined to a resolution of ~25Å, suggesting the structures were flexible. See also Figure S1.
Figure 3.
Figure 3.. The 12 Å resolution cryoEM structure of immDENV3:Fa b 1H10 complex at pH 8.0.
(A) Surface of the cryoEM map of immDENV3:Fab 1H10 complex (left) and a quarter of the center-section of the map (right). The map is colored according to its corresponding radius (red: 0–30 Å, orange: 31–160 Å, yellow: 161– 180 Å, green: 181–260 Å, cyan: 261–280Å and blue : >281 Å). (B) The structure of immDENV3:Fab 1H10 which was fitted into the cryoEM map. E, prM proteins and Fab surfaces are colored in grey, cyan and yellow, respectively. We observed 180 copies of Fabs binding to prM molecules on the virus surface. (C) The 1H10 epitope (pink) on prM consists of a and c strands, and the b-c loop. One of the E:prM complexes within a trimeric spike is colored with the two proteins in blue and cyan, respectively while the other two E:prM complexes are colored in grey. (D) Open-book representation of the surface potential of the prM:Fab interacting interfaces. The boundaries of the epitope and the paratope are marked by black solid lines. Positive and negative charges are colored in blue and red, respectively. (E) Sequence comparison of the Fab 1H10 epitope (green box) across four DENV serotypes show high similarities consistent with the ability of antibody to cross-react with all serotypes. White and red letters represent identical residues, whereas black letter indicates non-conserved residues. See also Figures S1 and S9.
Figure 4.
Figure 4.. Class I and II cryoEM structures of immDENV3:Fab 1H10 complex at pH 5.0.
(A) CryoEM maps of the Class I (left) and Class II (right) particles of immDENV3:Fab 1H10 complex at pH 5.0. The map is colored according to its radius (red: 0–30Å, orange: 31–160Å, yellow: 161–180Å, green: 181–260Å, cyan: 261–280Å and blue: >281Å). (B and C) Comparison of the class I and II cryoEM maps with two models – (B) structures of immDENV:Fab 1H10 complex at pH 8.0 (the structure before maturation), and (C) immDENV2 at pH 6.0 (Yu et al., 2009) (structure after completion of maturation) superimposed with the pr:Fab 1H10 structure. The Fabs and the prM:E molecules of the pH 8.0 model are colored in pink and red, respectively, while that of the pH 6.0 model are in light blue and blue. The height of densities (transparent grey surface) corresponding to the E protein:pr:Fab 1H10 complex in the class I and class II cryoEM maps is similar to that of the virus:Fab 1H10 pH 8.0 and pH 6.0 models, respectively. This suggests that the class I and II structures may represent the early and late stages of the low pH-induced structural change during the maturation process, respectively. See also Figures S1 and S3
Figure 5.
Figure 5.. The arrangement of the surface proteins of the class I and II immDENV:Fab 1H10 complex at low pH shows dissociation of some Fab:pr and identifies four stages of the maturation process.
At pH 8.0, Fab 1H10 is bound to all pr-molecules (indicated by filled star). The transition from pH 8.0 to pH 5.0 in the class I structure causes some of the pr:Fab 1H10 complexes to dissociate from the E protein red molecule (indicated by half-filled star). Further structural rearrangements of the E proteins in the class II pH 5.0 structures result in all pr:Fab complexes dissociating from the same E protein red molecules (indicated by open star). This suggests that the red molecule likely experiences more clashes compared with the other E proteins (green and blue) in the asymmetric unit during maturation. The class I and II structures may also represent intermediate steps of the maturation process (stage II and III, respectively) that are partially stabilized by Fab binding. The class I structure (stage II) exists in a state closer to stage I of the maturation process, while the class II (stage III) structure is closer to the completion of the maturation process (stage IV). Comparison of the class I (stage II) structure with the immDENV:Fab 1H10 complex at pH 8.0 (stage I) shows mainly translational movements of all three individual E proteins in an asymmetric unit. On the other hand, in the class II structure (stage III), the E proteins nearly form two dimers (red:blue, green:green molecules) that were previously observed in the immature virus at pH 6.0 (Yu et al., 2009) or the mature virus pH 8.0 (Kostyuchenko et al., 2013) structures (stage IV). See also Figure S4.
Figure 6.
Figure 6.. Molecular insights into the stage I-II transition using molecular dynamics simulations.
(A) Representative snapshots extracted from targeted molecular dynamics simulations and their corresponding simulation frame number. E proteins from molecular simulations are colored in lighter colors than those of the Stage I and II cryoEM structures. Level of pr-Fab occupancies are shown as filled or partially filled stars. (B) Pr molecules interacting with the red E proteins were blocked by the neighboring blue E protein molecules within a trimeric spike during the structural rearrangement from stage I to II. Plot showing number of pr molecules on the red E proteins that are blocked by neighboring blue E protein molecules during the structural arrangement (blue line) against frame number. The number of dissociated pr proteins from the E proteins (right y-axis) is also shown in brown. The grey dashed lines indicate the frames of the snapshots in panel (A). (C) Snapshots of the motions of pr:E complexes within a trimeric spike at different frames, showing clashes (black outlined arrow in frames 25 and 85) mainly occurring between the pr:blue E protein complex, with the pr molecule on the red E protein leading to the dissociation of the pr from the red E protein molecule. The E proteins from the molecular simulations are colored in lighter shades than those from Stage I and II cryoEM structures. (D) Plot of number of Fab-Fab clashes between the Fab on the pr-blue E protein and that on the pr-red E protein (red line) throughout the simulation. At around frame 25, severe clashes between the Fab:pr:E red molecules and Fab:pr:E blue molecules were observed when Fabs were superimposed onto the simulated pr-E virus molecule. See also Figures S5, S6, S7 and S8.
Figure 7.
Figure 7.. Possible E protein structural rearrangement between stage III-IV determined using molecular dynamics simulations.
E protein movements from targeted molecular dynamics simulations are shown at different frame numbers along the conformational transition pathway. See also Figure S7.
Figure 8.
Figure 8.. Fab 1H10 remains bound to pr-molecule at pH range of 5.0 to 8.0.
(A) ELISA assay showing the IgG and Fab bind equally well to immDENV over a pH range of 5.0 to 8.0 when pH is kept constant throughout the assay (Top panels). The same is observed when binding of Fab was done at pH 8.0 and then washed with other pH buffers (bottom panel). NS indicates results are not-significantly different analyzed by one-way ANOVA. (B) Kinetics of HMAb 1H10 binding to recombinant prM:E linked protein determined by BLI analysis. When association and dissociation of the prM:E protein to the captured IgG were done at a constant pH, all are in the nM range suggesting high affinity interactions at all pH conditions. When association was done at pH 7.4 and then dissociation at either pH 6.0 or 5.0, results suggest that once the antibody is bound at neutral pH, exposure of the complex to low pH does not result in the detachment of HMAb 1H10 from the prM:E protein. The values summarized in (A) and (B) are average and standard deviation from three independent experiments. (C) The Fab 1H10-pr interacting interface is largely the same when Fab-prM:E complex samples that were formed at pH 7.4 were then incubated at pH 5.0, 6.0 or 7.4, determined by HDX-MS. Plot of the detectable pr peptides from the prM:E:Fab complex samples that were also obtained in the uncomplexed prM:E protein controls. The peptides that have deuterium exchange difference of <−0.5 when subtracting between the Fab complexed and the uncomplexed prM-E at the same pH conditions are boxed with a black line. These peptides indicate the regions that are more buried on the pr-molecule upon Fab binding, consistent with the cryoEM epitope (pink). Fab remains bound to the epitope at all pH values. (D) HDX-MS identified epitope (spheres) shown on the structure of prM:E complex with that identified by cryoEM (pink).
Figure 9.
Figure 9.. The binding footprint of Fab 1H10 on recombinant prM:E protein was the same at pH 7.4, 6.0, and 5.0 as determined by HDX-MS.
(A) Differences in the number of deuterons between pepsin proteolyzed fragments of prM:E protein and prM:E:Fab1H10 complex after 30 sec of deuterium exchange at pH 7.4, pH 6.0 and pH 5.0. Each node represents a pepsin proteolyzed peptide and is listed according to its position from the N-terminus. Differences in deuterium exchange below −0.5 deuterons are considered significant (red dashed line). Standard error for each peptide at each pH condition is indicated by the shaded regions along the X-axis and peptide residue numbers are colored according to the legend in each difference plot. The standard error for each peptide represent standard deviations observed across all the HDX-MS measurements from three independent HDX-MS experiments and the standard error for a given peptide represent the sum of the sigma standard deviations of each of the two conditions being compared. (B) Plot of the M and E protein peptides detected in both prM:E and prM:E:Fab 1H10 complex samples. The peptides that have deuterium exchange difference of < −0.5 deuterons are boxed in thick black outline. These regions are more buried upon Fab binding, however comparison across pHs is difficult as there are no common peptides with exactly the same length detected.

Similar articles

See all similar articles

Cited by 5 articles

Publication types