Characterization of changes in the hemagglutinin that accompanied the emergence of H3N2/1968 pandemic influenza viruses

The hemagglutinin (HA) of A/H3N2 pandemic influenza viruses (IAVs) of 1968 differed from its inferred avian precursor by eight amino acid substitutions. To determine their phenotypic effects, we studied recombinant variants of A/Hong Kong/1/1968 virus containing either human-type or avian-type amino acids in the corresponding positions of HA. The precursor HA displayed receptor binding profile and high conformational stability typical for duck IAVs. Substitutions Q226L and G228S, in addition to their known effects on receptor specificity and replication, marginally decreased HA stability. Substitutions R62I, D63N, D81N and N193S reduced HA binding avidity. Substitutions R62I, D81N and A144G promoted viral replication in human airway epithelial cultures. Analysis of HA sequences revealed that substitutions D63N and D81N accompanied by the addition of N-glycans represent common markers of avian H3 HA adaptation to mammals. Our results advance understanding of genotypic and phenotypic changes in IAV HA required for avian-to-human adaptation and pandemic emergence.


52
Wild aquatic birds represent the major natural reservoir of IAVs, which occasionally transmit, Alteration of the HA sensitivity to protease digestion that accompany acid-induced conformational 159 transition was determined using a solid-phase receptor binding assay as described previously (fet-HRP) was determined and expressed in percentages of binding to low-pH-exposed virus with 167 respect to that of the virus exposed to pH 7. Binding-versus-pH curves were plotted, and pH values that 168 corresponded to HA inactivation by 50% (pH50) were determined by linear interpolation.   Receptor-binding specificity of the viruses was characterized using soluble synthetic 219 sialoglycopolymers (SGPs) (GlycoNZ, Auckland, New Zealand) (Tuzikov et al., 2021). The SGPs 220 contained 20 mol% of sialyloligosaccharide moieties and 5 mol% of biotin attached to either the low-221 molecular-mass (20-kDa) or high-molecular-mass (1000-kDA) poly-N-(2-hydroxyethyl)acrylamide 222 backbone. 223 The structures of the sialyloligosaccharide moieties and designations of SGPs are shown below. The binding of the viruses to SGPs was determined in a direct solid-phase binding assay as 232 described previously (Matrosovich and Gambaryan, 2012 (Fig. 2c). R2 and R7 were more sensitive than HK and R5 to inhibition by NH4Cl, 412 confirming that R2 and R7 require a lower pH in endosomes for fusion and cell entry.

413
Acid stability of the HA correlates, at least partially, with HA resistance to heat and denaturing  human-type amino acids in HK). We concluded that these amino acids are not involved in substantial 503 epistatic interactions.

504
Because of the relatively weak binding of HK, R5 and point mutants to Neu5Acα2-3Gal-505 containing receptors, we could not reliably quantify binding of these IAVs to corresponding 20-kDa 506 SGPs. Using more sensitive but less discriminative 1-MDa SGPs, we found that R5 bound stronger 507 than HK to both 3'SLN and SLe c and that most single-point mutants did not significantly differ in this 508 respect from the parental IAVs (Fig. 4b,c). These results suggested that single-point substitutions had a 509 weaker effect than their combination on HA binding to 3'SLN and SLe c .

510
To further characterize receptor-binding properties of the HA mutants, we studied inhibition of  The following conclusions could be made from the binding data. A combination of human-type 524 substitutions separating HK from R5 reduced HA binding to both Neu5Acα2-6Gal-and Neu5Acα2-525 3Gal-containing receptors. Three of these substitutions, namely, R62I, N193S and either D81N or 526 D63N, were primarily responsible for the reduction of binding avidity. The other three substitutions 527 either had a weak binding-enhancing effect (N92K and A144G) or no effect [F(-2)L]).

528
The observed reduction of the avidity for human-type receptors during HA evolution from its  Fig. S4a,b) as did their MDCK-grown counterparts (Fig. 4a,d). The HTBE-grown HK, 538 R5 and two glycosylation mutants R5-81 and R5-63 also showed the same relative binding avidity 539 (supplementary Fig. S4c) as did corresponding MDCK-grown variants (Fig. 4d). These results indicated In MDCK cells, R5 formed smaller plaques than did HK indicative of a less efficient multicycle 553 replication of the former virus (Fig. 6). The effects of point substitutions in the HA on plaque size was 554 studied separately for HK mutants and R5 mutants (Fig. 6). Although the resolving power of the assay  . We therefore compared single-and multicycle replication of HK and R5 in the same 586 experiment (Fig. 7b). This experiment confirmed that R5 infected more cells in the first round of 587 infection in HTBE cultures, whereas HK produced more viral progeny after multiple infection cycles.

588
To infer which of the substitutions separating HK from R5 contributed to more efficient multicycle 589 replication of HK in HTBE cultures, we compared replication of HK and its single-point mutants under 590 competitive conditions. In the first experiment (Fig. 8), we focused on substitutions at positions 81 and 591 193 as they showed the major effect on receptor-binding properties of HK and R5. We also studied the The H3 HA sequences available from the GISAID EpiFlu database include sequences of avian 658 IAVs (predominantly from wild aquatic birds) and of several stable mammalian-adapted viral lineages 659 that emerged from the avian reservoir via interspecies transmission (Fig. 11a, supplementary Fig. S5).  infected less cells than R5 during initial inoculation into the HTBE cultures, HK produced more 796 infectious virus particles after multicycle replication (Fig. 7), suggesting that it outperforms R5 during

814
Reduced avidity of the HK HA was primarily associated with the avian-to-human substitutions 815 R62I, N193S and either D81N or D63N (Fig. 4). The substitution R62I is located relatively far from the    dN≠dS. When significant result is obtained, negative selection is inferred if dN/dS < 1, otherwise diversifying positive selection is inferred. d Site-level characterization of episodic positive selection in the corresponding group inferred using the MEME method. Asterisks show significance levels for the test that dN>dS for some fraction of branches. The number of individual branches where episodic selection may have acted is indicated as well (empirical Bayes factor ≥ 100) e Site-level characterization of directional selection using the DEPS/FADE method. The residue that is the putative target of directional selection is indicated, with empirical Bayes factor supporting directional selection shown. f Site-level characterization of differences in selective pressures between branch groups using the Contrast-FEL method. Asterisks show significance levels for the test that dN/dS differ between pairs of branch groups, or among all branch groups (overall). Notation like H>A means that dN/dS for "human" branches is greater than dN/dS for "avian" branches. Only significant results are listed.