T-cell epitope content comparison (EpiCC) of swine H1 influenza A virus hemagglutinin

Abstract

Background: Predicting vaccine efficacy against emerging pathogen strains is a significant problem in human and animal vaccine design. T-cell epitope cross-conservation may play an important role in cross-strain vaccine efficacy. While influenza A virus (IAV) hemagglutination inhibition (HI) antibody titers are widely used to predict protective efficacy of 1 IAV vaccine against new strains, no similar correlate of protection has been identified for T-cell epitopes.

Objective: We developed a computational method (EpiCC) that facilitates pairwise comparison of protein sequences based on an immunological property-T-cell epitope content-rather than sequence identity, and evaluated its ability to classify swine IAV strain relatedness to estimate cross-protective potential of a vaccine strain for circulating viruses.

Methods: T-cell epitope relatedness scores were assessed for 23 IAV HA sequences representing the major H1 swine IAV phylo-clusters circulating in North American swine and HA sequences in a commercial inactivated vaccine (FluSure XP^® ). Scores were compared to experimental data from previous efficacy studies.

Results: Higher EpiCC scores were associated with greater protection by the vaccine against strains for 23 field IAV strain vaccine comparisons. A threshold for EpiCC relatedness associated with full or partial protection in the absence of cross-reactive HI antibodies was identified. EpiCC scores for field strains for which FluSure protective efficacy is not yet available were also calculated.

Conclusion: EpiCC thresholds can be evaluated for predictive accuracy of protection in future efficacy studies. EpiCC may also complement HI cross-reactivity and phylogeny for selection of influenza strains in vaccine development.

Keywords: T-cell epitope content comparison; T-cell epitope prediction; computational immunology; hemagglutinin; influenza A viruses; swine influenza H1 viruses; swine leukocyte antigen; vaccine efficacy.

Figure 1

Phylogenetic tree for the Hemagglutinin (HA) amino acid sequences of influenza A field and FluSure vaccine (FS) viruses representing H1 phylo‐cluster in the North American swine. Bootstrap test results are shown next to the branches

Figure 2

Comparison of scores of shared and unique epitopes across strains. Scores of shared, vaccine‐unique, and strain‐unique swine leukocyte antigen class I and II epitopes were determined for the comparison of hemagglutinin (HA) sequences from vaccine viruses and field (challenge) strains. Note that y‐axes show different scales. Solid connecting lines are included only for visualization purposes. P‐values of comparisons were calculated using 1‐tailed Wilcoxon matched‐pairs signed rank test (**P < .001). HA vaccine sequences had higher scores for shared epitopes with strains belonging to the same H1 cluster or the same HA lineage. In general, scores of class II shared and unique epitopes were significantly higher than those of class I. Viruses are sorted by nucleotide identity relative to H1γ FS. Strain numbers on the x‐axis are described in detail in the legend below

Figure 3

EpiCC score comparisons between hemagglutinin (HA) sequences of FluSure vaccine (FS) viruses and field viruses. Each axis corresponds to the HA sequence of 1 virus. EpiCC score = Scores of shared epitopes ‐ Scores of strain‐ and vaccine‐unique epitopes. HA sequences in the same cluster had the highest EpiCC scores. For each vaccine virus, class I and II EpiCC scores were significantly different from each other (P < .05). Note that for comparisons where the score for unique epitopes was greater than the score for shared epitopes, EpiCC scores were below zero

Figure 4

Relationship between EpiCC scores and hemagglutinin (HA) amino acid identity. The second‐order polynomial relationship between class I (top) and II (bottom) EpiCC scores and amino acid identity for each FS virus is shown. R₂ values of regression models are shown. H2N3, H5N1, H5N2, H3N2, H4N6, and a random sequence were included in this analysis to represent the lower end of the identity range. Interestingly, there were instances where viruses had low EpiCC scores despite high identity (eg, class II epitope content of IA00 H1γ FS compared to SD15 H1α)

Figure 5

Definition of threshold for prediction of vaccine efficacy prediction. Total EpiCC scores (sum of class I and II EpiCC scores; blue line) for the comparison of H1γ FS and each viral strain are shown. The FS γ‐cluster vaccine strain was protective or partially protective against challenge with viruses annotated as (P) or (PP), respectively. The rest of the viruses were not tested as challenge strains. Protection and partial protection thresholds (black lines) defined 3 areas shown in white (protection; total EpiCC score above −0.001), light gray (partial protection), and dark gray (no protection). Viruses used to set the thresholds are marked with an asterisk (*). We hypothesize that FS would confer at least partial protection against challenge with viruses that had EpiCC scores outside the darker gray region