No abstract

In pandemic and epidemic forms, influenza causes substantial, sometimes catastrophic, morbidity and mortality. Intense selection from the host immune system drives antigenic change in influenza A and B, resulting in continuous replacement of circulating strains with new variants able to re-infect hosts immune to earlier types. This 'antigenic drift' often requires a new vaccine to be formulated before each annual epidemic. However, given the high transmissibility and mutation rate of influenza, the constancy of genetic diversity within lineages over time is paradoxical. Another enigma is the replacement of existing strains during a global pandemic caused by 'antigenic shift'--the introduction of a new avian influenza A subtype into the human population. Here we explore ecological and immunological factors underlying these patterns using a mathematical model capturing both realistic epidemiological dynamics and viral evolution at the sequence level. By matching model output to phylogenetic patterns seen in sequence data collected through global surveillance, we find that short-lived strain-transcending immunity is essential to restrict viral diversity in the host population and thus to explain key aspects of drift and shift dynamics.

Eighteen codons in the HA1 domain of the hemagglutinin genes of human influenza A subtype H3 appear to be under positive selection to change the amino acid they encode. Retrospective tests show that viral lineages undergoing the greatest number of mutations in the positively selected codons were the progenitors of future H3 lineages in 9 of 11 recent influenza seasons. Codons under positive selection were associated with antibody combining site A or B or the sialic acid receptor binding site. However, not all codons in these sites had predictive value. Monitoring new H3 isolates for additional changes in positively selected codons might help identify the most fit extant viral strains that arise during antigenic drift.

We have studied the HA1 domain of 254 human inf luenza A(H3N2) virus genes for clues that might help identify characteristics of hemagglutinins (HAs) of circulating strains that are predictive of that strain's epidemic potential. Our preliminary findings include the following.

The hemagglutinin (HA) gene of influenza viruses encodes the major surface antigen against which neutralizing antibodies are produced during infection or vaccination. We examined temporal variation in the HA1 domain of HA genes of human influenza A (H3N2) viruses in order to identify positively selected codons. Positive selection is defined for our purposes as a significant excess of nonsilent over silent nucleotide substitutions. If past mutations at positively selected codons conferred a selective advantage on the virus, then additional changes at these positions may predict which emerging strains will predominate and cause epidemics. We previously reported that a 38% excess of mutations occurred on the tip or terminal branches of the phylogenetic tree of 254 HA genes of influenza A (H3N2) viruses. Possible explanations for this excess include processes other than viral evolution during replication in human hosts. Of particular concern are mutations that occur during adaptation of viruses for growth in embryonated chicken eggs in the laboratory. Because the present study includes 357 HA sequences (a 40% increase), we were able to separately analyze those mutations assigned to internal branches. This allowed us to determine whether mutations on terminal and internal branches exhibit different patterns of selection at the level of individual codons. Additional improvements over our previous analysis include correction for a skew in the distribution of amino acid replacements across codons and analysis of a population of phylogenetic trees rather than a single tree. The latter improvement allowed us to ascertain whether minor variation in tree structure had a significant effect on our estimate of the codons under positive selection. This method also estimates that 75.6% of the nonsilent mutations are deleterious and have been removed by selection prior to sampling. Using the larger data set and the modified methods, we confirmed a large (40%) excess of changes on the terminal branches. We also found an excess of changes on branches leading to egg-grown isolates. Furthermore, 9 of the 18 amino acid codons, identified as being under positive selection to change when we used only mutations assigned to internal branches, were not under positive selection on the terminal branches. Thus, although there is overlap between the selected codons on terminal and internal branches, the codons under positive selection on the terminal branches differ from those on the internal branches. We also observed that there is an excess of positively selected codons associated with the receptor-binding site and with the antibody-combining sites. This association may explain why the positively selected codons are restricted in their distribution along the sequence. Our results suggest that future studies of positive selection should focus on changes assigned to the internal branches, as certain of these changes may have predictive value for identifying future successful epidemic variants.