Genotype-specific features reduce the susceptibility of South American yellow fever virus strains to vaccine-induced antibodies

Abstract

The resurgence of yellow fever in South America has prompted vaccination against the etiologic agent, yellow fever virus (YFV). Current vaccines are based on a live-attenuated YF-17D virus derived from a virulent African isolate. The capacity of these vaccines to induce neutralizing antibodies against the vaccine strain is used as a surrogate for protection. However, the sensitivity of genetically distinct South American strains to vaccine-induced antibodies is unknown. We show that antiviral potency of the polyclonal antibody response in vaccinees is attenuated against an emergent Brazilian strain. This reduction was attributable to amino acid changes at two sites in central domain II of the glycoprotein E, including multiple changes at the domain I-domain II hinge, which are unique to and shared among most South American YFV strains. Our findings call for a reevaluation of current approaches to YFV immunological surveillance in South America and suggest approaches for updating vaccines.

Keywords: 17D; South America; antibody response; emerging virus; flavivirus; glycoprotein; immunological response; neutralizing antibodies; vaccine; yellow fever virus.

Figure 1.. Antiviral potency and breadth of the neutralizing antibody response elicited by YF-17D vaccination.

(A) Serum neutralizing titers (half-maximal inhibitory concentration values from dose-response curves [IC₅₀]) for three YF-17D–vaccinated U.S. donors over time against the indicated RVPs. Means±SEM, n=9–15 from 3–5 independent experiments. (B) Serum neutralizing titers for the three donors in panel A at days 14 and 360 post-vaccination. Means, n=9–15 from 3–5 independent experiments. (C) Serum neutralizing titers (half-maximal inhibitory concentration values in a focus reduction neutralization test [FRNT₅₀]) for five YF-17DD–vaccinated Brazilian donors over time against the indicated authentic viruses. Means, n=3. (D) Serum neutralizing titers (FRNT₅₀) for the five donors in panel C at days 14 and 360 post-vaccination. Means, n=3. (E) Serum neutralizing titers for a U.S. vaccinee cohort (n=16 donors) against the indicated RVPs. n=9–12 from 3–4 independent experiments. (F) Serum neutralizing titers (FRNT₅₀) for a Brazilian vaccinee cohort (n=24 donors against the indicated authentic viruses. n=3. In panels E–F, lines indicate group medians. Groups in E were compared by two-way ANOVA followed by Tukey’s correction for multiple comparisons. Groups in F were compared by the Wilcoxon matched-pairs signed-rank test. **, P<0.002. ****, P<0.0001. LOD, limit of detection.

Figure 2.. Binding and neutralization breadth of a panel of nAbs isolated from YF-17D vaccinees.

(A) 99 nAbs were tested for their binding response to recombinant, soluble E (sE) proteins from the indicated viruses by biolayer interferometry. (B) Neutralizing activities of selected nAbs (10 nM) against the indicated RVPs. Means, n=6–9 from three independent experiments, (C) Neutralizing activities of selected mAbs against the indicated authentic viruses. (D) A subset of nAbs comprising DII binders were evaluated for sE binding as in panel A. (E–F) Neutralizing activities of DII-specific nAbs against the indicated RVPs (E) and authentic viruses (F), as in panels B–C. (G–I) Binding (G) and neutralization activities (H–I) of DIII-specific nAbs, as in panels B–C. In all panels, lines indicate group medians. Groups in A, C, D, F, G, and I were compared by the Wilcoxon matched-pairs signed-rank test. Groups in B, E, H were compared by two-way ANOVA followed by Tukey’s correction for multiple comparisons. **, P<0.002. ****, P<0.0001. ns, not significant. Only the significant comparisons are shown in panels B, E, H. LOD, limit of detection.

Figure 3.. Neutralizing activities of vaccinee sera against RVPs bearing 17D/ES-504 domain-swapped E proteins.

(A) Schematic of the 17D/ES-504 E protein chimeras. (B–C) Serum neutralizing titers for the three U.S. donors (see Figure 1A) at the indicated times post-vaccination against the indicated RVPs. Means±SEM, n=6–12 from 2–4 independent experiments. (D–E) Serum neutralizing titers for a U.S. vaccinee cohort (n=16 donors; see Figure 1E) against the indicated RVPs. n=6–12 from 2–4 independent experiments. Parental RVP_17D and RVP_ES-504 controls are from Figure 1E. Groups in D–E were compared to WT by two-way ANOVA followed by Tukey’s correction for multiple comparisons. **, P<0.002. ****, P<0.0001. Only the significant comparisons are shown. In all panels, lines indicate group medians. LOD, limit of detection.

Figure 4.. YFV strain-dependent sequence polymorphisms in DII of the E protein.

(A) Amino acid sequence alignments of the portion of E corresponding to DII are shown for the indicated sequences. The colored residues in the alignment correspond to positions where residue identities are not identical in all five sequences. Polymorphic sites 1 and 2 are indicated below the alignment. (B) Amino acid positions with unique residue identities in YFV-ES-504 E are shown on a YFV virion, which was constructed in Chimera (Pettersen et al., 2004) using the atomic coordinates of YFV E (Lu et al., 2019) and the symmetry matrix of the dengue virus 2 virion (Zhang et al., 2004). (C) Close-up view displaying two neighboring dimers. Highlighted surface area in the lower dimer corresponds to the nAb 5A binding interface as dictated by PISA (relative buried surface area ≥ 10%) (Krissinel and Henrick, 2007). Residue numbering corresponds to that of the mature YFV-17D E protein.

Figure 5.. Effects of polymorphisms at Sites 1 and 2 on the neutralizing activities of vaccinee nAbs and sera.

(A) Schematic of YFV-17D and -ES-504 E proteins chimerized at Sites 1 and 2. (B–C) Neutralizing activities of selected DII-specific nAbs (10 nM) against the indicated RVPs. Means, n=6–12 from 2–4 independent experiments, Parental RVP_17D and RVP_ES-504 controls are from Figure 2. (D–E) Serum neutralizing titers for a U.S. vaccinee cohort (n=16 donors; see Figure 1E) against the indicated RVPs. Means, n=6–12 from 2–4 independent experiments. Parental RVP_17D and RVP_ES-504 controls are from Figure 1E. Groups were compared to WT by two-way ANOVA followed by Tukey’s correction for multiple comparisons. **, P<0.002. ****, P<0.0001. ns, not significant. In all panels, lines indicate group medians. LOD, limit of detection.

Figure 6.. Effects of polymorphisms at subsite 2a and 2b on N–linked glycosylation of YFV E and neutralization sensitivity to vaccinee sera.

(A) Occupancy and relative abundance of the indicated glycan structures at residue Asn 269 in recombinant soluble E proteins (sE) from YFV-Asibi and YFV-ES-504 E were determined by mass spectrometry. (B) E protein immunoprecipitates from RVP_Asibi and RVP_ES-504, and postnuclear lysates from authentic virus ES-504–infected cells were subjected to PNGase F treatment and detected by SDS-PAGE and Western blot. Viral particles bearing the Ebola virus glycoprotein (GP) were used as an internal control for PNGase F activity. Representative blots are shown. (C) Schematic of YFV-17D E protein chimerized at subsites 2a and 2b. (D) Serum neutralizing titers for a U.S. vaccinee cohort, n=15, against the indicated RVPs. Means, n=6 from 2 independent experiments. Parental RVP_17D controls are from Figure 1E. Groups were compared to WT by two-way ANOVA followed by Tukey’s correction for multiple comparisons. ***, P<0.0002. ****, P<0.0001. In all panels, lines indicate group medians. LOD, limit of detection.

Figure 7.. Genetic makeup of YFV strains at polymorphic sites 1 and 2.

Amino acid-based tree depicting E proteins from representative YFV strains (scale bar represents the number of amino acid substitutions per site). Colored ranges denote viral origin and clade(s): i) Old World (light blue), ii) SA1 (South American genotype 1, brown). SA2 (South American genotype 2, beige). Black stars indicate YFV sequences used in this study. The inner ring is color-coded according to the continent of origin. Note that all sequences of apparent Asian or European origin are ex-African (if clustering with Old World) or ex-South American (if clustering with SA1). N/A, not applicable. Sequence features are represented with a square (Site 1: H67 A83), circle (Site 2a: predicted glycan at N269), or star (Site 2b: K272). Features present in sequence are represented with filled shapes. The outer char bar denotes the number of strains being represented at each branch of the tree.