Preventing an Antigenically Disruptive Mutation in Egg-Based H3N2 Seasonal Influenza Vaccines by Mutational Incompatibility

Abstract

Egg-based seasonal influenza vaccines are the major preventive countermeasure against influenza virus. However, their effectiveness can be compromised when antigenic changes arise from egg-adaptive mutations on influenza hemagglutinin (HA). The L194P mutation is commonly observed in egg-based H3N2 vaccine seed strains and significantly alters HA antigenicity. An approach to prevent L194P would therefore be beneficial. We show that emergence of L194P during egg passaging can be impeded by preexistence of a G186V mutation, revealing strong incompatibility between these mutations. X-ray structures illustrate that individual G186V and L194P mutations have opposing effects on the HA receptor-binding site (RBS), and when both G186V and L194P are present, the RBS is severely disrupted. Importantly, wild-type HA antigenicity is maintained with G186V, but not L194P. Our results demonstrate that these epistatic interactions can be used to prevent the emergence of mutations that adversely alter antigenicity during egg adaptation.

Keywords: antigenicity; egg-adaptive mutation; epistasis; hemagglutinin; influenza virus; receptor binding; vaccine.

Graphical abstract

Figure 1

Egg-Adaptive Mutations in Human H3N2 Virus (A) Occurrence frequencies of seven major egg-adaptive mutations in egg-passaged human isolates from different years are shown. Mutations are numbered according to H3 numbering. The absolute residue numberings (from the first Met in the HA protein as 1) for residues 156, 183, 186, 194, 219, and 246, are 172, 199, 202, 210, 235, and 262, respectively. (B) The Cα positions of the major egg-adaptive mutations are shown on the HA structure as teal spheres. The binding site for the sialic acid is shaded by a yellow oval. (C) For each of the nine representative vaccine strains listed in Table S2, the amino-acid identities at residues 156, 183, 186, 194, 219, and 246 (H3 numbering) are shown. Egg-adaptive mutations are highlighted in yellow. Of note, the egg-adaptive mutation G186E in A/South Australia/55/2014 (IVR-175) was seldom observed during egg adaptation. Two egg-adapted strains that were derived from A/Brisbane/10/2007, namely NYMC X-171 (GISAID: EPI_ISL_23300) and IVR-147 (GISAID: EPI_ISL_20693), are also included here for comparison. (D) A/Kansas/14/2017 is the H3N2 vaccine strain for the 2019–2020 northern hemisphere influenza season. The amino-acid identities at residues 156, 183, 186, 194, 219, and 246 are shown for four egg-adapted strains that were derived from A/Kansas/14/2017. The sequences were downloaded from GISAID with the following accession numbers: IVR-195 (GISAID: EPI_ISL_346482), X-327 (GISAID: EPI_ISL_346457), CBER-22C (GISAID: EPI_ISL_346456), and CBER-22B/CDC19A (GISAID: EPI_ISL_346455). Egg-adaptive mutations are highlighted in yellow.

Figure 2

Incompatibility of Egg-Adaptive Mutations HA L194P and G186V (A) A network diagram was constructed that describes the co-occurrence of egg-adaptive mutations and was based on the sequence analysis of egg-passaged human H3N2 viruses in the GISAID database. Each node represents an egg-adaptive mutation. Two mutations that co-occur more than once are connected by an edge. The width of the edge is proportional to the co-occurrence frequency. The occurrence frequency of each egg-adaptive mutation is color coded. L183 is not shown because it does not co-occur with other major egg-adaptive mutations in more than one isolate. (B) The fitness effects of different mutants were examined by a virus rescue experiment. The titer was measured by median tissue culture infectious dose (TCID₅₀). Error bars indicate the standard deviation of three independent experiments. (C and D) Bris07 HA G186V mutant virus (C) and Bris07 HA L194P mutant virus (D) were passaged for five rounds in eggs. The emergence of egg-adaptive mutations in the receptor-binding domain (HA1 residues 117–265) was monitored by next-generation sequencing. Only those mutations that reached a minimum of 10% occurrence frequency are plotted. Three independent passaging experiments were performed for the Bris07 HA G186V mutant virus in (C) and one passaging experiment was performed for the Bris07 HA L194P mutant virus in (D). Passage 0 indicates the input virus.

Figure 3

Structural Comparison of IVR-165 HA and Vic11 HA (A) The HA RBS conformations of IVR-165 HA and Vic11 HA were compared by aligning their receptor-binding subdomains (HA1 residues 117–265). The slight expansion of the RBS due to the backbone shifts of the 190-helix and 220-loop is indicated by the arrows. IVR-165 HA that was used for structural determination was expressed recombinantly in insect cells (see STAR Methods). (B) A zoomed-in view of the RBS backbone shift. (C) The distances between the phenolic oxygen of Tyr 98 (OH₉₈) and the Cα of residue 190 (Cα₁₉₀) in different H3 strains were measured: Wy03: PDB 6BKN (Wu et al., 2018), Fin04: PDB 2YP2 (Lin et al., 2012), HK05: PDB 2YP7 (Lin et al., 2012), Bris07: PDB 6AOR (Wu et al., 2017b), Bris07 (L194P): PDB 6AOP (Wu et al., 2017b), Vic11: PDB 4O5N (Lee et al., 2014), and Mich14: PDB 6BKP (Wu et al., 2018).

Figure 4

Structural Analysis of IVR-165 HA in Complex with Glycan Receptor Analogs (A and B) HA crystal structures of IVR-165 in complex with 3′SLNLN (A) and IVR-165 in complex with 6′SLNLN (B). The apo form is aligned on the complexes (blue) and colored in whitish gray. Glycan receptor analogs (3′SLNLN and 6′SLNLN) are colored in yellow and shown in stick representation. Hydrogen bonds are represented by black dashed lines. (C) The HA RBS conformations of IVR-165 HA and Bris07-L194P HA in complex with 3′SLNLN are compared. Structural alignment here and in (D) was performed using the receptor-binding subdomain (HA1 residues 117–265). (D) The RBS conformations of IVR-165 HA and Bris07 HA-L194P in complex with 6′SLNLN are compared. (E) The distance between equivalent Cαs of IVR-165 and Cαs of Bris07-L194P for each residue in the receptor-binding subdomain (HA1 residues 117–265) is shown. This analysis was performed on the structural alignment of IVR-165 and Bris07-L194P in complex with 3′SLNLN (upper panel) and in complex with 6′SLNLN (bottom panel). The major differences are seen in the 190-helix (shaded region).

Figure 5

Electron Density of the 190-Helix Final 2Fo-Fc electron density maps for the HA receptor-binding site of the L194P mutant (top) and G186V/L194P double mutant (bottom) are represented in a blue mesh and contoured at 0.8 σ. Apo form (left panel), complex with 3′SLNLN (middle panel), and complex with 6′SLNLN (right panel). Final 2Fo-Fc electron density maps and coordinates for L194P mutant were retrieved from PDB 6AOP (apo form), PDB 6AOS (in complex with 3′SLNLN), and PDB 6AOT (in complex with 6′SLNLN) (Wu et al., 2017b). The HA of the Bris07 G186V/L194P double mutant was expressed recombinantly in insect cells (see STAR Methods).

Figure 6

Antigenic Characterization of G186V and L194P (A) A total of 6 mice were immunized with Bris07 WT virus (6:2 reassortant on a PR8 backbone). Sera from immunized mice were tested for binding to WT, G186V mutant, and L194P mutant of recombinant Bris07 HA using ELISA. Percentage of WT binding was computed as binding level of the mutant at 1:1000 dilution of the serum sample/binding level of WT at 1:1000 dilution of the serum sample. Each line represents one serum sample. Statistical significance was determined using the paired Student’s t test. (B) Biolayer interferometry (BLI) was used to measure the binding kinetics of anti-RBS C05 IgG against recombinant HAs of Bris07 WT, G186V mutant, and L194P mutant. Gray lines represent the response curve and colored lines represent the 1:1 binding model.