A single residue in the yellow fever virus envelope protein modulates virion architecture and antigenicity

Abstract

Yellow fever virus (YFV) is a re-emerging flavivirus that causes severe hepatic disease and mortality in humans. Despite being researched for over a century, the structure of YFV has remained elusive. Here we use a chimeric virus platform to resolve the first high resolution cryo-EM structures of YFV. Stark differences in particle morphology and homogeneity are observed between vaccine and virulent strains of YFV, and these are found to have significant implications on antibody recognition and neutralisation. We identify a single residue (R380) in the YFV_17D envelope protein that stabilises the virion surface, and leads to reduced exposure of the cross-reactive fusion loop epitope. The differences in virion morphology between YFV strains also contribute to the reduced sensitivity of the virulent YFV virions to vaccine-induced antibodies. These findings have significant implications for YFV biology, vaccinology and structure-based flavivirus antigen design.

Fig. 1. Structural and antigenic characterisation of bYFVs.

a Representative cryo-EM micrographs of purified bYFV_17D, bYFV_ES504 and bYFV_Asibi particles incubated at 4 °C prior to freezing. b Two-dimensional class averages of bYFV_17D and bYFV_ES504 particles. The displayed class distribution is shown as a percentage of the total number of particles. Cryo-EM density map of bYFV_17D (c) and bYFV_ES504 (d) with I3 symmetry applied. The maps are radially coloured according to the following: 0-150 Å red, 151-195 Å yellow, 196-210 Å green, 211-225 Å cyan, 226-240 Å blue Å. e IC50 values of recombinant hIgG1 anti-YFV or anti-flavivirus mAbs against bYFV_17D, bYFV_ES504 and bYFV_Asibi. Each symbol represents a technical replicate from three biological replicates (n = 6). f Serum neutralising titres of human YFV_17D vaccinees (n = 14) against bYFV_17D, bYFV_ES504 and bYFV_Asibi. In both (e) and (f), neutralisation was determined via FRNTs on C6/36 (Aedes albopictus) cells. Lines indicate group medians. Significance was determined via Kruskal-Wallis tests with Dunn’s multiple comparisons test on GraphPad Prism 9.0. ****p < 0.0001, ***p ≤ 0.0002, **p ≤ 0.002, *p ≤ 0.03. For 2C9: p = 0.006 (ES504 vs. Asibi), 2A10G6: p = 0.040 (17D vs. ES504) and p = 0.002 (17D vs. Asibi), 6B6C−1: p = 0.034 (17D vs. ES504) and p = 0.002 (17D vs. Asibi), 864: p = 0.0015 (17D vs. ES504 and 17D vs. Asibi), human vaccinee sera: p = 0.0001 (17D vs ES504) and p = 0.046 (17D vs Asibi). LOD = limit of detection. Source data is provided as a Source Data file.

Fig. 2. Cryo-EM of bYFV_17D and bYFV_ES504 complexed with 5A and 2C9 Fabs.

a Representative micrographs of bYFV_17D and bYFV_ES504 with and without complexing with 5A or 2C9 Fab. Viruses and Fabs were combined in a 1:1 ratio and incubated overnight at 4 °C prior to vitrification. b Cryo-EM density maps of bYFV_17D and bYFV_ES504 complexed with either 5A Fab or 2C9 Fab with I3 symmetry applied. Maps are radially coloured according to the following: 0-150 Å red, 151-195 Å yellow, 196-210 Å green, 211-225 Å cyan, 226-240 Å blue Å, 241-255 Å orchid. c Cryo-EM density map and atomic model of bYFV_17D complexed with 2C9 Fab. The density map was obtained via symmetry expanded ASU reconstruction and is shown in grey (E) and blue (variable region of 2C9 Fab), with the atomic model shown in red (E-DI), yellow (E-DII), blue (E-DIII) and grey (2C9 Fab). The N151 glycan is shown in green. d Top view of bYFV_17D:2C9 E dimer atomic model. e Comparison of the bYFV_17D:2C9 cryo-EM near-atomic resolution model and the YFV_17D X-ray crystal structure (PDB:6IW4) in pink. Dotted lines are illustrating the E protein curvature differences between these two models.

Fig. 3. Structural and antigenic characterisation of bYFV prME chimeras.

a Schematic of the prM and E genes of the bYFV_17D/ES504 prME chimeras. b Representative micrographs of purified bYFV_17D/ES504 prME chimera particles at 4 °C. c IC50 values of recombinant hIgG1 anti-YFV or anti-flavivirus mAbs against the bYFV DIII chimeras. Neutralisation was determined via FRNTs on C6/36 (Aedes albopictus) cells. Each symbol represents a technical replicate from three biological replicates (n = 6) and lines indicate group medians. Parental bYFV_17D and bYFV_ES504 controls are from Fig. 1E. The bYFV DIII chimeras were compared with parental viruses by two-tailed Mann-Whitney tests on GraphPad Prism 9. **p ≤ 0.002, *p ≤ 0.03. For 2A10G6: p = 0.002 (17D vs. 17D/DIII_ES504) and p = 0.015 (ES504 vs. ES504/DIII_17D), 6B6C−1: p = 0.002 (17D vs. 17D/DIII_ES504) and p = 0.009 (ES504/DIII_17D), 864: p = 0.002 (17D vs. 17D/DIII_ES504) and p = 0.002 (ES504 vs. ES504/DIII_17D). Source data is provided as a Source Data file. LOD limit of detection.

Fig. 4. Cryo-EM of bYFV_ES504/DIII_17D and bYFV_Asibi/DIII_17D.

a Cryo-EM density map of bYFV_ES504/DIII_17D with I3 symmetry applied and coloured according to its local resolution. b Cryo-EM reconstruction and atomic model of the bYFV_ES504/DIII_17D ASU. c Cryo-EM density map of bYFV_ES504/DIII_17D:2C9 with I3 symmetry applied and coloured according to its local resolution. d Cryo-EM reconstruction and atomic model of bYFV_ES504/DIII_17D:2C9 ASU. e Cryo-EM density map of bYFV_Asibi/DIII_17D with I3 symmetry applied and coloured according to its local resolution. f Cryo-EM reconstruction and atomic model of bYFV_Asibi/DIII_17D ASU. In (e) and (f) the pentagons, triangles, and ovals represent the icosahedral five-, three- and two-fold axes, respectively. In (b), (d), and (f) the density map was obtained via symmetry expanded ASU reconstruction using cisTEM2 and is shown in either grey (E) or blue (2C9 Fab), with the atomic models shown in red (E-DI), yellow (E-DII), blue (E-DIII) and grey (2C9 Fab). g IC50 values of recombinant hIgG1 anti-YFV or anti-flavivirus mAbs against the bYFV DIII chimeras. Neutralisation was determined via FRNTs on C6/36 (Aedes albopictus) cells. Lines indicate group medians, and each symbol represents a technical replicate from three biological replicates. For bYFV_17D/DIII_Asibi and bYFV_Asibi/DIII_17D against 5A n = 3, and for all other groups n = 6. Parental bYFV_17D and bYFV_Asibi controls are from Fig. 1E. bYFV DIII chimeras were compared to parental viruses by two-tailed Mann-Whitney tests on GraphPad Prism 9. **p ≤ 0.002. For 2A10G6: p = 0.002 (17D vs 17D/DIII_Asibi) and p = 0.004 (Asibi vs Asibi/DIII_17D), 6B6C−1 and 864: p = 0.002 (17D vs 17D/DIII_Asibi and Asibi vs. Asibi/DIII_17D). LOD = limit of detection. Source data is provided as a Source Data file.

Fig. 5. Structural and antigenic characterisation of bYFV DIII mutant chimeras.

a Schematic of the YFV genome illustrating the four amino acids in DIII where YFV_17D differs to both YFV_ES504 and YFV_Asibi. b Representative micrographs of the bYFV DIII mutant chimeras at 4 °C. c IC₅₀ values of recombinant hIgG1 anti-YFV or anti-flavivirus mAbs against the bYFV DIII mutant chimeras. Neutralisation was determined using FRNTs on C6/36 (Aedes albopictus) cells. Each symbol represents a technical replicate from three biological replicates (n = 6) and lines indicate group medians. Parental bYFV_17D and bYFV_ES504 controls are from Fig. 1E. Significance was determined via Kruskal-Wallis tests with Dunn’s multiple comparisons test on GraphPad Prism 9.0. ***p ≤ 0.0002, **p ≤ 0.002, *p ≤ 0.03. For 5A: p = 0.035 (17D F305S vs. 17D R380T), 2A10G6: p = 0.002 (all 17D viruses vs. 17D R380T), p = 0.004 (ES504 M299I vs. ES504 T380R), p = 0.016 (ES504 S305F vs. ES504 T380R), p = 0.01 (ES504 P325S vs. ES504 T380R), 6B6C−1: p = 0.0004 (17D, 17D I299M and 17D F305S vs 17D T380R), p = 0.003 (17D S325P vs 17D R380T), p = 0.002 (ES504 M299I vs. ES504 T380R), p = 0.005 (ES504 S305F vs. ES504 T380R), p = 0.040 (ES504 P325S vs. ES504 T380R), 864: p = 0.029 (17D vs. 17D F305S), p = 0.002 (17D vs. 17D S325P), p = 0.01 (17D I299M vs. 17D F305S), p = 0.0005 (17D I299M vs. 17D S325P). Source data is provided as a Source Data file. LOD limit of detection.

Fig. 6. Characterisation of bYFV_ES504 T380R.

a Cryo-EM reconstruction of bYFV_ES504 T380R with I3 symmetry applied and coloured according to its local resolution. b Cryo-EM reconstruction and atomic model of bYFV_ES504 T380R ASU. The density map was obtained by symmetry expanded ASU reconstruction using cisTEM2 and is shown in grey, with the atomic model shown in red (E-DI), yellow (E-DII) and blue (E-DIII). In (a) and (b) the pentagons, triangles, and ovals represent the icosahedral five-, three- and two-fold axes, respectively. c Atomic model of bYFV_ES504 T380R E monomer with R380 highlighted in dark pink and the fusion loop highlighted in orange. d Contacts between the five-fold (blue), three-fold (dark purple) and two-fold (light purple) axes E protein R380 residues and the adjacent E proteins. Bonds are depicted by blue dotted lines. e Comparison of the bYFV_ES504 T380R and bYFV_ES504/DIII_17D (in grey) ASU atomic models, aligned at the five-fold axis E protein. Below is a comparison of the three-fold axis E protein. f Serum neutralising titres of human YFV_17D vaccinees (n = 14) against the bYFV DIII chimeras. Neutralisation was determined using FRNTs on C6/36 (Aedes albopictus) cells. Significance was determined using two-tailed Wilcoxon matched-pairs signed rank test on GraphPad Prism 9.0. ***p ≤ 0.0002, **p ≤ 0.002, *p ≤ 0.03. p = 0.0001 (17D vs 17D/DIII_ES504), p = 0.0004 (17D vs. 17D R380T), p = 0.009 (ES504 vs. ES504/DIII_17D), p = 0.017 (ES504 vs. ES504 T380R). LOD = limit of detection. Source data is provided as a Source Data file. LOD limit of detection.