Human Adaptation of Ebola Virus during the West African Outbreak

Abstract

The 2013-2016 outbreak of Ebola virus (EBOV) in West Africa was the largest recorded. It began following the cross-species transmission of EBOV from an animal reservoir, most likely bats, into humans, with phylogenetic analysis revealing the co-circulation of several viral lineages. We hypothesized that this prolonged human circulation led to genomic changes that increased viral transmissibility in humans. We generated a synthetic glycoprotein (GP) construct based on the earliest reported isolate and introduced amino acid substitutions that defined viral lineages. Mutant GPs were used to generate a panel of pseudoviruses, which were used to infect different human and bat cell lines. These data revealed that specific amino acid substitutions in the EBOV GP have increased tropism for human cells, while reducing tropism for bat cells. Such increased infectivity may have enhanced the ability of EBOV to transmit among humans and contributed to the wide geographic distribution of some viral lineages.

Keywords: Ebola virus; Makona; adaptation; bat; epistasis; evolution; human; pseudovirus; tropism.

Graphical abstract

Figure 1

Schematic Maximum Likelihood Phylogenetic Tree of 1,610 Complete EBOV Makona Genomes The tree is color coded according to lineage, and the lineage-defining amino acid substitutions in the viral GP studied here are marked. To enhance data display, monophyletic groups of sequences were collapsed, and the tree was vertically compressed in multiple sections. The tree was rooted using the earliest Kissidougou-C15 sequence, and all horizontal branch lengths are drawn to a scale of nucleotide substitutions per site. An expanded tree is presented in Data S1.

Figure 2

Schematic of the 17 Lineage-Defining Amino Acid Combinations in the EBOV GP Amino acid changes identified on the EBOV Makona phylogeny presented in Figure 1 are shown in black. The GP scheme is drawn to scale. SP: signal peptide; RBD: receptor binding domain; MLD: mucin-like domain; IFL: internal fusion loop; HR: heptad repeat; TM: transmembrane domain. Also indicated to the left of the alignment is the assigned variant name, where A and B denote lineage A (82A) and B (82V) backgrounds, respectively. Prime (′) indicates variants not sampled during the outbreak and AB′ has been used to identify variants generated to investigate the impact of 82A background on lineage-B-defined substitutions. On the right-hand side, the specific combination of amino acid substitutions compared to the reference strain. Colors relate to the lineages identified in Figure 1.

Figure 3

Differential Infectivity of Pseudoviruses Supplemented with EBOV Makona GP Mutants in Human Cells (A–C) Relative infectivity of each glycoprotein was expressed as a proportion (%) of that observed for the Kissidougou-C15 strain in HuH7 (A), BEAS-2B (B), and A549 (C) cells. Histogram bar colors correspond to the lineage color-coding shown in Figure 1, with gray bars indicating variants not sampled during the outbreak. These data are the means ± 1 SD of either two (non-sampled variants) or three (sampled variants) independent experiments, each performed in triplicate. Differences in the mean infectivity of each outbreak-associated mutant compared to the Kissidougou-C15 EBOV strain and the lineage B viruses to the A82V mutant were assessed using repeated-measures one-way ANOVA with Dunnett’s multiple comparison test and indicated in the table inset; ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001; n.s., not significant. (D–F) Plots of normalized infectivity for the three human cell lines and correlations were calculated using Pearson’s correlation test. See also Figure S2.

Figure 4

Differential Infectivity of Pseudoviruses Supplemented with EBOV Glycoprotein Mutants in Different Bat Cell Lines (A and B) Relative infectivity of each glycoprotein was expressed as a proportion (%) of that observed for the Kissidougou-C15 strain in HypLu/45.1 (A) and HypNi/1.1(B) cells. Histogram bar colors correspond to the lineage color coding shown in Figure 1, with gray bars indicating variants not sampled during the outbreak. These data are the means ± 1 SD of either two (non-sampled variants) or three (sampled variants) independent experiments, each performed in triplicate. Differences in the mean infectivity of each outbreak-associated mutant compared to the Kissidougou-C15 EBOV strain and the lineage B viruses to the A82V mutant were assessed using repeated-measures one-way ANOVA with Dunnett’s multiple comparison test and indicated in the table inset; ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001; n.s., not significant. See also Figure S1.

Figure 5

Negative Correlation of GP Mutant Infectivity between Hypsignathus Monstrosus Cells and Human Cells (A–F) Normalized infectivity data for each GP mutant were plotted for HypLu/45.1 cells versus HuH7 (A), BEAS-2B (B), and A549 (C) and HypNi/1.1 cells versus HuH7 (D), BEAS-2B (E), and A549 (F). Infectivity data are the same as that shown in Figures 3 and 4. Correlations were determined using Pearson’s correlation test.

Figure 6

Residue 82 in the GPcl/NPC1-C Complex (A) Cartoon showing the organization of the GP complex as anchored in the viral membrane, colored according to domains as labeled (left). The right panel shows the cleaved GP (GPcl), i.e., what remains on the viral membrane after cathepsin cleavage in the endosome, which eliminates the glycan cap and the mucin-like domain (MLD) from the GP trimer. The endosomal membrane and the multiple transmembrane spanning protein NPC1 (Gong et al., 2016) are also shown, with domain C, which is bound by cleaved GP, in orange, contacted by the RBD of GP. (B) Crystal structure of NPC1-C domain in complex with GPcl (PDB 5F1B [Wang et al., 2016]). NPC1-C is in orange, and residues that are different in bat NPC1-C displayed in green and with sticks. GPcl is shown in yellow, with the GP2 moiety in cyan (as in the cartoon in A). Overlaid is the structure of the GP complex in its pre-fusion form (gray, PDB 3SCY [Lee et al., 2008]), which lacks the MLD but still contains the glycan cap. The side chains of GP1 Y232 and NPC1-C Y134 are shown in sticks. The α1 helix, which moves upon complex formation, is indicated. (C) Zoom of the region framed in (B), rotated to better display the interactions. The ring of P80, the first residue on the α1 helix, packs against the phenol ring of Y232 within GP1 prior to cathepsin cleavage. Removal of the 191–503 region after cathepsin cleavage frees the α1 helix to interact with NPC1-C, where Y134 takes the same place. The helix packs against the side chains of W86 and Y109, which remain relatively unchanged, while A82 glides downward, accompanying the movement of P80 to maintain the interaction with NPC1-C Y134 in the complex. Because the environment is different in bat NPC1, where residue 127 (labeled in green in (B)) changes from a charged lysine to a hydrophobic isoleucine, our data point to a more favorable interaction with the human NPC1 by acquiring a valine at position 82. (D and E) Same as (C) but showing, for clarity, only the individual structures in the same orientation.

Figure S1

Infectivity of Pseudoviruses Supplemented with EBOV GP Mutants Observed in Kidney Cells from Two Different Fruit-Bat Species, Related to Figures 4 and S2 (A and B) Relative infectivity of each glycoprotein was expressed as a proportion (%) of that observed for the Kissidougou-C15 strain in Epomops buettikoferi kidney cells (EpoNi/22.1; A) and Rousettus aegyptiacus kidney cells (RoNi/7.1; B). Histogram bar colors correspond to the lineage color-coding shown in Figure 1. These data are the means ± 1 SD of three independent experiments, each performed in triplicate. Differences in the mean infectivity of each mutant compared to the Kissidougou-C15 EBOV strain and the lineage B viruses to the A82V mutant were assessed using repeated-measures one-way ANOVA with Dunnett’s multiple comparison test and indicated in the table inset; ^∗ p < 0.05; ^∗∗ p < 0.01; ^∗∗∗ p < 0.001; n.s., not significant.

Figure S2

Negative Correlation of GP Mutant Infectivity in Epomops Buettikoferi Kidney, Rousettus Aegyptiacus Kidney, and Human Cells, Related to Figures 3 and S1 (A–F) Normalized infectivity data for each GP mutants were plotted for Epomops buettikoferi kidney cells (EpoNi/22.1) versus HuH7 (A), BEAS-2B (B), and A549 (C) and Rousettus aegyptiacus kidney cells (RoNi/7.1) versus HuH7 (D), BEAS-2B (E), and A549 (F). Infectivity data are the same as that shown in Figures 3 and S1. Correlations were determined using Pearson’s correlation test.