Natural and directed antigenic drift of the H1 influenza virus hemagglutinin stalk domain

Abstract

The induction of antibodies specific for the influenza HA protein stalk domain is being pursued as a universal strategy against influenza virus infections. However, little work has been done looking at natural or induced antigenic variability in this domain and the effects on viral fitness. We analyzed human H1 HA head and stalk domain sequences and found substantial variability in both, although variability was highest in the head region. Furthermore, using human immune sera from pandemic A/California/04/2009 immune subjects and mAbs specific for the stalk domain, viruses were selected in vitro containing mutations in both domains that partially contributed to immune evasion. Recombinant viruses encoding amino acid changes in the HA stalk domain replicated well in vitro, and viruses incorporating two of the stalk mutations retained pathogenicity in vivo. These findings demonstrate that the HA protein stalk domain can undergo limited drift under immune pressure and the viruses can retain fitness and virulence in vivo, findings which are important to consider in the context of vaccination targeting this domain.

Figure 1

Amino acid usage across influenza H1N1 HA proteins. Analysis of HA protein sequence variability among the 4827 H1N1 strains publically available in Influenza Resources Database. (A) 3D structure of A/California/04/2009 hemagglutinin protein. Each aa is colored by its conservation frequency, meaning the percent of viruses that have the dominant (most common) amino acid at that position. (B) Number of unique aa found at each position of the HA protein (566 residues). HA head region is indicated in gray, stalk in yellow. (C) Histogram of the number of strains for the range of unique aa found in (B). (D) Average number of aa used at each position for head or stalk region. P-value <0.001 using a Welch’s t-test. (E) Shannon entropy (unpredictability) at each aa position across HA protein.

Figure 2

Variability of influenza H1N1 HA protein. (A) dN/dS analysis showing positive selection of codons across the H1 HA proteins of viruses that circulated since 1918 (top) and those that circulated since the 2009 pandemic (bottom). Values greater than one (bar) are considered positively selected (see Supplementary Tables 1 and 2 for exact values). (B) Principal component analysis (PCA) of protein amino acid changes between the stalk regions of influenza strains. HA proteins from viruses that circulated from 1918–2015, colored as indicated in legend. (C) PCA of only 1918-pandemic lineage viruses; viruses that circulated from 1918–2009. (D) PCA of 2009-pandemic lineage viruses that circulated from 2009–2015.

Figure 3

Antigenic variation in the stalk region of historical influenza viruses. ELISA titers were measured using seven monoclonal antibodies reactive to the stalk region of HA protein (6F12, RA5-22, CM2S3, CR9114, C179, and FB75) against recombinant HA from 6 H1N1 viruses and 5 other subtypes (H2, H9, H5, H3, and influenza B). The assays were performed in duplicates, twice, and the averages are shown. Purified HA proteins from the following strains were used: H1N1 strains A/California/04/2009 (CA09), A/South Carolina/11/1918 (SC18), A/Puerto Rico/8/1934 (PR8), A/New Caledonia/20/1999 (NC99), A/Solomon Islands/3/2006 (SI06), A/Brisbane/59/2007 (BR07), and H2N2 A/Singapore/1/1957 (H2), H9N2 A/Hong Kong/33982/2009 (H9), A/Indonesia/05/2005 (H5), H3N2 A/Brisbane/10/2007 (H3), and B/Brisbane/60/2008 (B).

Figure 4

Amino acid locations of escape mutations. 3D structure simulation of the HA protein reconstructed using the 2009-pandemic virus protein sequence. Canonical B cell epitopes are colored as indicated in legend, mutations are colored in red. The fusion peptide in the HA2 domain is colored in orange. The antigenic sites are colored differently (see legend). *Indicates this is not the specific position for the aa 526, since it is in the conformationally unstable linker domain (the crystal structures of HA all end at residue 510).

Figure 5

Microneutralization assays for the viruses incorporating mutations in the HA protein stalk domain. The original virus or the viruses incorporating the mutations A388V (A), V466I/R526G (B), or V41I (C and D) were grown in the presence of the mAbs 6F12 (A), and CR9114 (B), or in the presence of serum from subject 298 directly (C) or pre-absorbed with the head domain of the HA protein from A/California/04/2009 (D). Virus growth in the presence of different concentrations of mAb/serum was compared to virus growth in the absence of mAbs/serum, and represented as % growth (upper panels) or as virus titers (in FFU/ml, lower panels). Experiments were repeated three independent times in duplicate, with similar results. Data represents the means and standard deviations (SD) from the three experiments. *p-values <0.05 for comparison of wt versus V41I, A388V or V466I/R526G mutants, using a Student’s t-test.

Figure 6

Microneutralization assays for the recombinant viruses incorporating the mutations in the stalk domain. The wt (original) virus or the recombinant viruses incorporating the mutations A388V (A), V466I/R526G alone or in combination (B), or V41I (C and D) were grown in the presence of the mAbs 6F12 (A), and CR9114 (B), or in the presence of serum from subject 298 directly (C) or pre-absorbed with the head domain of the HA protein from A/California/04/2009 (D). Virus growth in the presence of the mAb/serum was compared to virus growth in the absence of mAb/serum, and represented as % growth (top panels) or as virus titers (in FFU/ml, bottom panels). Experiments were repeated three independent times in duplicate, with similar results. Data represents the means and SDs from the three experiments from the three experiments. *p-values <0.05 for comparison of wt versus V41I, A388V, I466/G526 and V466/G526 viruses, using a Student’s t-test.

Figure 7

Co-infection of wt and viruses encoding the mutations selected under immune pressure. The wt virus and viruses incorporating the mutations V41I (A), A388V (B) and V466I/R526G (C) were used to co-infect cells at a ratio 2: 1 in the absence (0) or presence of two different concentrations of 298 serum (A), mAb 6F12 (B) or mAb CR9114 (C). When cytopathic effect was approximately 10%, viral and cellular RNA from infected cells was extracted and the haemagglutinin (HA) sequence was obtained. Chromatograms showing the sequences for amino acids 41 (A), 388 (B) and 466/526 (C) are shown. Letter represent the nucleotide sequence for codons 41 (GTA, encoding V, to ATA, encoding I, in A), 388 (GCC, encoding A, to GAC, encoding V, in B), and 466/526 (GTA/AGG, encoding V/R, to ATA/GGG, encoding I/G, in C).

Figure 8

Effect of mutations V41I, A388V, and R526G on the binding of mAbs/sera to the virus. ELISA titers specific for the whole virus were measured using the sera from patient 298 pre-absorbed for anti-head antibodies (A), mAb 6F12 (B), and mAb CR9114 (C). Concentrated virions from the following recombinant viruses were used: WT and V41I (A), WT and A388V (B) and WT, V466I, R526G and V466I/R526G (C). ELISAs were performed twice in duplicates, with similar results. Means and standard deviations from the duplicates are shown.

Figure 9

Recombinant viruses growth kinetics in vitro. Canine MDCK (A) and human A549 (B) cells were infected in duplicates with recombinant A/California/04/2009/E3 viruses incorporating the mutations in the HA protein stalk domain at the indicated MOIs. Virus titers of infected cells supernatants were determined at different times pi by immunofocus assay. Experiments were repeated three independent times in duplicate, with similar results. Data represents the results from the three experiments combined. *p-values <0.05 for comparison of wt versus I466 virus, using a Student’s t-test. P-values using the same Student’s t-test were >0.05 (not significant) for comparison of wt versus I466/G526, wt versus G526, wt versus V388, and wt versus I41 viruses.

Figure 10

Virulence of recombinant A/California/04/2009/E3 viruses containing mutations in the HA stalk domain. Groups (N=8) of 7- to-8-week-old C57BL/6 female mice were infected (500 FFU/mice) with recombinant A/California/04/2009/E3 viruses containing mutations in the HA stalk domain. (A) Mice were monitored daily for weight loss. Data is plotted as the percent of the starting weight averaged for each group of mice. To determine significance, the weight loss over time (days) was non-parametrically modeled by penalized regression splines, assuming different curves for each virus. Bonferroni corrected confidence intervals for the A388V and V466I were significantly less than the other groups, including the wt virus (p < 0.003). (B) Mice survival was analyzed during 14 days.