Selective pressure to increase charge in immunodominant epitopes of the H3 hemagglutinin influenza protein

Trace Reconstruction Problems in Computational Biology

The problem of reconstructing a string from its error-prone copies, the trace reconstruction problem, was introduced by Vladimir Levenshtein two decades ago. While there has been considerable theoretical work on trace reconstruction, practical solutions have only recently started to emerge in the context of two rapidly developing research areas: immunogenomics and DNA data storage. In immunogenomics, traces correspond to mutated copies of genes, with mutations generated naturally by the adaptive immune system. In DNA data storage, traces correspond to noisy copies of DNA molecules that encode digital data, with errors being artifacts of the data retrieval process. In this paper, we introduce several new trace generation models and open questions relevant to trace reconstruction for immunogenomics and DNA data storage, survey theoretical results on trace reconstruction, and highlight their connections to computational biology. Throughout, we discuss the applicability and shortcomings of known solutions and suggest future research directions.

Implications of localized charge for human influenza A H1N1 hemagglutinin evolution: Insights from deep mutational scans

Seasonal influenza A viruses of humans evolve rapidly due to strong selection pressures from host immune responses, principally on the hemagglutinin (HA) viral surface protein. Based on mouse transmission experiments, a proposed mechanism for immune evasion consists of increased avidity to host cellular receptors, mediated by electrostatic charge interactions with negatively charged cell surfaces. In support of this, the HA charge of the globally circulating H3N2 has increased over time since its pandemic. However, the same trend was not seen in H1N1 HA sequences. This is counter-intuitive, since immune escape due to increased avidity (due itself to an increase in charge) was determined experimentally. Here, we explore whether patterns of local charge of H1N1 HA can explain this discrepancy and thus further associate electrostatic charge with immune escape and viral evolutionary dynamics. Measures of site-wise functional selection and expected charge computed from deep mutational scan data on an early H1N1 HA yield a striking division of residues into three groups, separated by charge. We then explored evolutionary dynamics of these groups from 1918 to 2008. In particular, one group increases in net charge over time and consists of sites that are evolving the fastest, that are closest to the receptor binding site (RBS), and that are exposed to solvent (i.e., on the surface). By contrast, another group decreases in net charge and consists of sites that are further away from the RBS and evolving slower, but also exposed to solvent. The last group consists of those sites in the HA core, with no change in net charge and that evolve very slowly. Thus, there is a group of residues that follows the same trend as seen for the entire H3N2 HA. It is possible that the H1N1 HA is under other biophysical constraints that result in compensatory decreases in charge elsewhere on the protein. Our results implicate localized charge in HA interactions with host cells, and highlight how deep mutational scan data can inform evolutionary hypotheses.

The hemagglutinin (HA) surface protein of influenza A viruses evolves rapidly to evade host immunity, leading to sizable yearly epidemics in human populations. Previous transmission experiments with H1N1 in mice have tied immune escape to an increase in HA avidity for cellular receptors, mediated by electrostatic charge. Furthermore, retrospective sequence analyses from a previous study confirmed that the HA of circulating global H3N2 has increased in net charge, yet surprisingly, that of H1N1 has not varied significantly. How is a stable net charge related to local patterns of H1N1 HA charge in response to selection? To elucidate the role of local electrostatic charge in host-virus interactions, we investigate characteristics of local charge on the H1N1 HA using functional data from deep mutational scan experiments. Combining measures of functional selection and expected charge at each site on DMS data from a 1933 H1N1 HA yields a striking visual pattern that identifies three groups of sites that have different biophysical properties and that prove to have distinct evolutionary patterns in natural human sequences. Essentially, we find evidence for an increase in charge near the receptor binding site and for compensatory changes elsewhere on the protein. Thus, our findings may reconcile disparate results from transmission experiments and natural sequence analyses, and highlight the importance of local properties of the HA protein. Overall, our findings further support the hypothesis of immune escape due to increased HA avidity to cellular receptors mediated by electrostatic charge. More generally, our work illustrates a novel method of leveraging deep mutational scans in conjunction with natural sequences to shed light on virus-host interactions.

Immunodominance of Antigenic Site B in the Hemagglutinin of the Current H3N2 Influenza Virus in Humans and Mice

The hemagglutinin protein of H3N2 influenza viruses is the major target of neutralizing antibodies induced by infection and vaccination. However, the virus frequently escapes antibody-mediated neutralization due to mutations in the globular head domain. Five topologically distinct antigenic sites in the head domain of H3 hemagglutinin, A to E, have been previously described by mapping the binding sites of monoclonal antibodies, yet little is known about the contribution of each site to the immunogenicity of modern H3 hemagglutinins, as measured by hemagglutination inhibition activity, which is known to correlate with protection. To investigate the hierarchy of antibody immunodominance, five Δ1 recombinant influenza viruses expressing hemagglutinin of the A/Hong Kong/4801/2014 (H3N2) strain with mutations in single antigenic sites were generated. Next, the Δ1 viruses were used to determine the hierarchy of immunodominance by measuring the hemagglutination inhibition reactivity of mouse antisera and plasma from 18 human subjects before and after seasonal influenza vaccination in 2017-2018. In both mice and humans, mutations in antigenic site B caused the most significant decrease in hemagglutination inhibition titers compared to wild-type hemagglutinin. This study revealed that antigenic site B is immunodominant in the H3N2 influenza virus strain included in the current vaccine preparations.IMPORTANCE Influenza viruses rapidly evade humoral immunity through antigenic drift, making current vaccines poorly effective and antibody-mediated protection short-lived. The majority of neutralizing antibodies target five antigenic sites in the head domain of the hemagglutinin protein that are also the most sequence-variable regions. A better understanding of the contribution of each antigenic site to the overall antibody response to hemagglutinin may help in the design of improved influenza virus vaccines.

Development of a novel peptide aptamer-based immunoassay to detect Zika virus in serum and urine

Zika virus (ZIKV) has been identified as a cause of adverse outcomes of pregnancy, including microcephaly and other congenital diseases. Most people infected with ZIKV do not show any symptoms. Development of a method to discriminate dengue virus (DENV) and ZIKV infections has been challenging, and efficient assays for patient management are limited, attributable to high levels of cross-reactivity among co-circulating Flaviviruses. Thus, there is an urgent need for a specific high-throughput diagnostic assay to discriminate ZIKV infections from other Flavivirus infections. Methods: A novel epitope peptide of the ZIKV envelope protein was predicted using three immune epitope database analysis tools and then further modified. A molecular docking study was conducted using three-dimensional structures of the ZIKV envelope and peptide. Experimentally, interactions between the selected peptides and virus were assessed via a fluorescence-linked sandwich immunosorbent assay (FLISA), and performance of peptide-linked sandwich FLISA was evaluated in virus-spiked human serum and urine. Results: The Z_10.8 peptide (KRAVVSCAEA) was predicted to be a suitable detector, with a higher binding affinity than other candidates based on four criteria (binding affinity, root mean square deviation, position of amine residue of lysine at the N-terminus, and interactive site) in a docking study. Z_10.8 was significantly more efficient at detecting ZIKV than the other two peptides, as shown in the direct FLISA (P < 0.001). Further, the equilibrium dissociation constant (K_d) for the Z_10.8 peptide was 706.0 ± 177.9 (mean ± SD, nM), with specificity to discriminate ZIKV from DENV. The limit of detection for the sandwich FLISA was calculated as 1×10⁴ tissue culture infective dose (TCID)₅₀/mL. The presence of serum or urine did not interfere with the performance of the Z_10.8-linked sandwich FLISA. Conclusion: Four criteria are suggested for the development of an in silico modeled peptide aptamer; this computerized peptide aptamer discriminated ZIKV from DENV via immunoassay.

Effects of egg-adaptation on receptor-binding and antigenic properties of recent influenza A (H3N2) vaccine viruses

Influenza A virus (subtype H3N2) causes seasonal human influenza and is included as a component of influenza vaccines. The majority of vaccine viruses are isolated and propagated in eggs, which commonly results in amino acid substitutions in the haemagglutinin (HA) glycoprotein. These substitutions can affect virus receptor-binding and alter virus antigenicity, thereby, obfuscating the choice of egg-propagated viruses for development into candidate vaccine viruses. To evaluate the effects of egg-adaptive substitutions seen in H3N2 vaccine viruses on sialic acid receptor-binding, we carried out quantitative measurement of virus receptor-binding using surface biolayer interferometry with haemagglutination inhibition (HI) assays to correlate changes in receptor avidity with antigenic properties. Included in these studies was a panel of H3N2 viruses generated by reverse genetics containing substitutions seen in recent egg-propagated vaccine viruses and corresponding cell culture-propagated wild-type viruses. These assays provide a quantitative approach to investigating the importance of individual amino acid substitutions in influenza receptor-binding. Results show that viruses with egg-adaptive HA substitutions R156Q, S219Y, and I226N, have increased binding avidity to α2,3-linked receptor-analogues and decreased binding avidity to α2,6-linked receptor-analogues. No measurable binding was detected for the viruses with amino acid substitution combination 156Q+219Y and receptor-binding increased in viruses where egg-adaptation mutations were introduced into cell culture-propagated virus. Substitutions at positions 156 and 190 appeared to be primarily responsible for low reactivity in HI assays with post-infection ferret antisera raised against 2012–2013 season H3N2 viruses. Egg-adaptive substitutions at position 186 caused substantial differences in binding avidity with an insignificant effect on antigenicity.

Punctuated evolution of influenza virus neuraminidase (A/H1N1) under opposing migration and vaccination pressures

Influenza virus contains two highly variable envelope glycoproteins, hemagglutinin (HA) and neuraminidase (NA). The structure and properties of HA, which is responsible for binding the virus to the cell that is being infected, change significantly when the virus is transmitted from avian or swine species to humans. Here we focus first on the simpler problem of the much smaller human individual evolutionary amino acid mutational changes in NA, which cleaves sialic acid groups and is required for influenza virus replication. Our thermodynamic panorama shows that very small amino acid changes can be monitored very accurately across many historic (1945–2011) Uniprot and NCBI strains using hydropathicity scales to quantify the roughness of water film packages. Quantitative sequential analysis is most effective with the fractal differential hydropathicity scale based on protein self-organized criticality (SOC). Our analysis shows that large-scale vaccination programs have been responsible for a very large convergent reduction in common influenza severity in the last century. Hydropathic analysis is capable of interpreting and even predicting trends of functional changes in mutation prolific viruses directly from amino acid sequences alone. An engineered strain of NA1 is described which could well be significantly less virulent than current circulating strains.

The antigenic drift molecular basis of the H5N1 influenza viruses in a novel branch of clade 2.3.4

H5N1 subtype influenza A virus has evolved into many HA clades since late 1990 s. Six circulating H5N1 influenza viruses clustered to a novel branch in clade 2.3.4 and could escape vaccine protection, indicating their antigenic drift. Eleven amino acids substitutions in three antigenic sites of the hemagglutinin of these isolates were found when compared with the hemagglutinin of the primary viruses in clade 2.3.4. On the backbone of the novel isolates A/chicken/Northern China/k0602/2010, we generated a panel of recombinant viruses with HA mutations of restoring the primary vaccine strain Re-5's amino acid and homologous antisera to determine the role of these substitutions. The results of cross-HI assay, micro-neutralization assay and the antigen map of the mutated recombinant viruses showed that three substitutions in antigenic site B, especially D205K, are the major contributors to the antigenic drift of the novel branch of clade 2.3.4. Our study highlights the importance of surveillance of antigenic drift of H5N1 viruses for the control and preparedness of pandemic threats.

Phylogenetic, molecular and drug-sensitivity analysis of HA and NA genes of human H3N2 influenza A viruses in Guangdong, China, 2007-2011

Annual H3N2 subtype influenza outbreaks in Guangdong, China are a severe public health issue and require ongoing monitoring of emerging viral variants. The variation and evolution of haemagglutinin (HA) and neuraminidase (NA) genes of influenza isolates from Guangdong during 2007-2011 and others from GenBank were analysed using Lasergene 7.1 and MEGA 5.05, and serological analysis of antigens was determined by haemagglutination inhibition (HI). Susceptibility to antiviral drugs was correlated with genetic mutations. Phylogenetic analysis and alignment of HA and NA genes were performed on 18 Guangdong isolates and 26 global reference strains. The non-synonymous (dN) evolutionary rate of HA1 was 3.13 times that of HA2. Compared with the A/Perth/16/2009 vaccine HA gene, homologies of Guangdong isolates were between 98.8-99.7% and 98.0-98·4% in 2009 and 2010, respectively. Amino-acid substitutions were found in five epitopes of HA1 from Guangdong isolates between 2007 and 2011, especially in epitopes B (N160K) and D (K174R/N). The K189E/N/Q and T228A mutations in the receptor-binding site (RBS) occurred in the 2010 strains, which affected the antigenicity of HA1. The antigenicity of the epidemic H3N2 isolates in 2010 was somewhat different from that of A/Perth/16/2009. The Guangdong H3N2 isolates were determined to be oseltamivir-resistant with IC50 of 0.396 ± 0.085 nmol/l (n=17) and zanamivir-resistant with IC50 of 0.477 ± 0.149 nmol/l (n=18). Variations were present in epitopes B and D, two sites in the RBS and two glycosylation sites in the Guangdong H3N2 HA1 gene. The majority of the Guangdong H3N2 isolates were sensitive to oseltamivir and zanamivir. Compared to the World Health Organization 2012 vaccine strains, Guangdong H3N2 strains varied genetically and antigenically to some degree.

Immunodominance of antigenic site B over site A of hemagglutinin of recent H3N2 influenza viruses

H3N2 influenza viruses have now circulated in the human population for 43 years since the pandemic of 1968, accumulating sequence changes in the hemagglutinin (HA) and neuraminidase (NA) that are believed to be predominantly due to selection for escape from antibodies. Examination of mutations that persist and accumulate led to identification of antigenically significant mutations that are contained in five antigenic sites (A–E) mapped on to the H3 HA. In early H3N2 isolates, antigenic site A appeared to be dominant while in the 1990s site B seemed more important. To obtain experimental evidence for dominance of antigenic sites on modern H3 HAs, we have measured antibodies in plasma of human subjects who received the 2006–07 trivalent subunit influenza vaccine (H3 component A/Wisconsin/67/05) or the 2008–09 formulation (H3 component A/Uruguay/716/07). Plasmas were tested against expressed HA of Wisconsin-like influenza A/Oklahoma/309/06 and site-directed mutants in antigenic site A (NNES121-124ITEG, N126T, N133D, TSSS135-138GSNA, K140I, RSNNS142-146PGSG), and antigenic site B (HL156-157KS, KFK158-160GST, NDQI189-192QEQT, A196V). “Native ELISA” analysis and escape mutant selection with two human monoclonal antibodies demonstrated that antibody E05 binds to antigenic site A and 1_C02 binds to site B. We find that most individuals, after vaccination in seasons 2006–07 and/or 2008–09, showed dominance of antigenic site B recognition over antigenic site A. A minority showed dominance of site A in 2006 but these were reduced in 2008 when the vaccine virus had a site A mutation. A better understanding of immunodominance may allow prediction of future antigenic drift and assist in vaccine strain selection.

Subtype- and antigenic site-specific differences in biophysical influences on evolution of influenza virus hemagglutinin

Background

Influenza virus undergoes rapid evolution by both antigenic shift and antigenic drift. Antibodies, particularly those binding near the receptor-binding site of hemagglutinin (HA) or the neuraminidase (NA) active site, are thought to be the primary defense against influenza infection, and mutations in antibody binding sites can reduce or eliminate antibody binding. The binding of antibodies to their cognate antigens is governed by such biophysical properties of the interacting surfaces as shape, non-polar and polar surface area, and charge.

Methods

To understand forces shaping evolution of influenza virus, we have examined HA sequences of human influenza A and B viruses, assigning each amino acid values reflecting total accessible surface area, non-polar and polar surface area, and net charge due to the side chain. Changes in each of these values between neighboring sequences were calculated for each residue and mapped onto the crystal structures.

Results

Areas of HA showing the highest frequency of pairwise changes agreed well with previously identified antigenic sites in H3 and H1 HAs, and allowed us to propose more detailed antigenic maps and novel antigenic sites for H1 and influenza B HA. Changes in biophysical properties differed between HAs of different subtypes, and between different antigenic sites of the same HA. For H1, statistically significant differences in several biophysical quantities compared to residues lying outside antigenic sites were seen for some antigenic sites but not others. Influenza B antigenic sites all show statistically significant differences in biophysical quantities for all antigenic sites, whereas no statistically significant differences in biophysical quantities were seen for any antigenic site is seen for H3. In many cases, residues previously shown to be under positive selection at the genetic level also undergo rapid change in biophysical properties.

Conclusions

The biophysical consequences of amino acid changes introduced by antigenic drift vary from subtype to subtype, and between different antigenic sites. This suggests that the significance of antibody binding in selecting new variants may also be variable for different antigenic sites and influenza subtypes.

Detecting patches of protein sites of influenza A viruses under positive selection

Influenza A viruses are single-stranded RNA viruses capable of evolving rapidly to adapt to environmental conditions. Examples include the establishment of a virus in a novel host or an adaptation to increasing immunity within the host population due to prior infection or vaccination against a circulating strain. Knowledge of the viral protein regions under positive selection is therefore crucial for surveillance. We have developed a method for detecting positively selected patches of sites on the surface of viral proteins, which we assume to be relevant for adaptive evolution. We measure positive selection based on dN/dS ratios of genetic changes inferred by considering the phylogenetic structure of the data and suggest a graph-cut algorithm to identify such regions. Our algorithm searches for dense and spatially distinct clusters of sites under positive selection on the protein surface. For the hemagglutinin protein of human influenza A viruses of the subtypes H3N2 and H1N1, our predicted sites significantly overlap with known antigenic and receptor-binding sites. From the structure and sequence data of the 2009 swine-origin influenza A/H1N1 hemagglutinin and PB2 protein, we identified regions that provide evidence of evolution under positive selection since introduction of the virus into the human population. The changes in PB2 overlap with sites reported to be associated with mammalian adaptation of the influenza A virus. Application of our technique to the protein structures of viruses of yet unknown adaptive behavior could identify further candidate regions that are important for host–virus interaction.

Evolution of H3N2 influenza virus in a guinea pig model

Studies of influenza virus evolution under controlled experimental conditions can provide a better understanding of the consequences of evolutionary processes with and without immunological pressure. Characterization of evolved strains assists in the development of predictive algorithms for both the selection of subtypes represented in the seasonal influenza vaccine and the design of novel immune refocused vaccines. To obtain data on the evolution of influenza in a controlled setting, naïve and immunized Guinea pigs were infected with influenza A/Wyoming/2003 (H3N2). Virus progeny from nasal wash samples were assessed for variation in the dominant and other epitopes by sequencing the hemagglutinin (HA) gene to quantify evolutionary changes. Viral RNA from the nasal washes from infection of naïve and immune animals contained 6% and 24.5% HA variant sequences, respectively. Analysis of mutations relative to antigenic epitopes indicated that adaptive immunity played a key role in virus evolution. HA mutations in immunized animals were associated with loss of glycosylation and changes in charge and hydrophobicity in and near residues within known epitopes. Four regions of HA-1 (75-85, 125-135, 165-170, 225-230) contained residues of highest variability. These sites are adjacent to or within known epitopes and appear to play an important role in antigenic variation. Recognition of the role of these sites during evolution will lead to a better understanding of the nature of evolution which help in the prediction of future strains for selection of seasonal vaccines and the design of novel vaccines intended to stimulated broadened cross-reactive protection to conserved sites outside of dominant epitopes.

Predicting Fixation Tendencies of the H3N2 Influenza Virus by Free Energy Calculation

Influenza virus evolves to escape from immune system antibodies that bind to it. We used free energy calculations with Einstein crystals as reference states to calculate the difference of antibody binding free energy (ΔΔG) induced by amino acid substitution at each position in epitope B of the H3N2 influenza hemagglutinin, the key target for antibody. A substitution with positive ΔΔG value decreases the antibody binding constant. On average an uncharged to charged amino acid substitution generates the highest ΔΔG values. Also on average, substitutions between small amino acids generate ΔΔG values near to zero. The 21 sites in epitope B have varying expected free energy differences for a random substitution. Historical amino acid substitutions in epitope B for the A/Aichi/2/1968 strain of influenza A show that most fixed and temporarily circulating substitutions generate positive ΔΔG values. We propose that the observed pattern of H3N2 virus evolution is affected by the free energy landscape, the mapping from the free energy landscape to virus fitness landscape, and random genetic drift of the virus. Monte Carlo simulations of virus evolution are presented to support this view.