Dynamics of glycoprotein charge in the evolutionary history of human influenza

Abstract

Background: Influenza viruses show a significant capacity to evade host immunity; this is manifest both as large occasional jumps in the antigenic phenotype of viral surface molecules and in gradual antigenic changes leading to annual influenza epidemics in humans. Recent mouse studies show that avidity for host cells can play an important role in polyclonal antibody escape, and further that electrostatic charge of the hemagglutinin glycoprotein can contribute to such avidity.

Methodology/principal findings: We test the role of glycoprotein charge on sequence data from the three major subtypes of influenza A in humans, using a simple method of calculating net glycoprotein charge. Of all subtypes, H3N2 in humans shows a striking pattern of increasing positive charge since its introduction in 1968. Notably, this trend applies to both hemagglutinin and neuraminidase glycoproteins. In the late 1980s hemagglutinin charge reached a plateau, while neuraminidase charge started to decline. We identify key groups of amino acid sites involved in this charge trend.

Conclusions/significance: To our knowledge these are the first indications that, for human H3N2, net glycoprotein charge covaries strongly with antigenic drift on a global scale. Further work is needed to elucidate how such charge interacts with other immune escape mechanisms, such as glycosylation, and we discuss important questions arising for future study.

Figure 1. Timecourses of total charge of the major antigens HA and NA.

Shown for three main human subtypes of influenza A. Sizes of dots represent frequencies, as indicated in the legend. Increase in positive charge for both H3N2 glycoproteins is statistically significant (p<0.0001 in both cases).

Figure 2. Comparison of changes in charge in the two subunits of the H3N2 HA molecule.

Solid lines indicate mean trends. The side-bar illustrates these regions on the HA molecule (diagram source: RCSB Protein Data Bank).

Figure 3. Charge data for the influenza NS1 protein.

Restricted to the H3N2 subtype.

Figure 4. Charge trends when summed separately over ‘group 1’ and ‘group 2’ amino acids (see table 1 ).

This approach shows that certain sites, which reverted charge state between 1985 and 2000, may account for the ‘plateau’ in total HA charge observed over this time. The right-hand panel locates the amino acids on the HA structure, showing the receptor binding site in green; group 1 positions showing a positive transition (in blue); group 1 positions showing a negative transition (in red); and group 2 positions, showing a transition followed by a reversion in a subsequent year (in magenta).

Figure 5. Estimated timing of major charge and glycosylation events.

In the lower panel, grey bars denote the presence of an N-glycosylation sequon in at least 10% of sequences in a year. Note that where there are only few samples in a year, as depicted in Figure S1, estimates in both panels become less reliable.