Background
Influenza virus continues to have a major impact on global health and is responsible for millions of cases of respiratory illness and hundreds of thousands of hospitalizations annually in the United States alone [
1]. The envelope glycoproteins, hemagglutinin (HA) and neuraminidase (NA), mediate host cell attachment and release, respectively, and are the primary targets of the protective antibody-mediated immune response. HA has functionally defined immunodominant antigenic sites that primarily map to the globular domain of the glycoprotein and surround the receptor binding site (RBS) [
2]. Circulating influenza viruses gradually accumulate HA mutations, primarily in the antigenic sites targeted by neutralizing antibodies, and these changes frequently allow escape from the antibody-mediated memory immune response. This process is known as antigenic “drift” and is likely driven by selection imposed by prevailing immunity in the host population, resulting in the need to periodically update the vaccine strains. Influenza virus can escape the antibody response through substitutions that induce conformational changes in the antigenic sites (epitopes), thus limiting antibody binding. Moreover, the modulation of viral HA receptor binding avidity can also lead to antigenic change and escape from antibody neutralization [
3,
4].
Many of the antiviral drug products that are either FDA-approved or in development for prophylaxis or treatment of influenza virus infection target the HA and/or NA glycoproteins and they include NA inhibitors (NAIs), monoclonal antibodies (mAbs), and vaccines. The activity of these drugs and vaccines may be affected by changes in the dynamic HA and NA molecules selected by the clinical use of these therapeutic agents. For example, influenza viruses with amino acid substitutions and/or deletions associated with reduced susceptibility to NAIs have been identified in cell culture selection studies, NAI-treated patients, as well as in circulating viruses from untreated individuals [
5‐
11]. Genetic analysis showed that reduced susceptibility to NAIs is associated with mutations in the viral NA and/or HA proteins and many of these mutations are listed in the NAI package inserts [
12‐
14]. Although the mechanistic basis for NAI treatment-emergent mutations in HA has yet to be defined, it is likely that their predicted effect of lowering receptor binding avidity compensates for reduced NA activity [
5‐
11]. The link between HA antibody escape and occurrence of compensatory NA mutations that result in acquisition of increased NAI resistance has been documented [
15]. However, it is not clear if HA mutations associated with clinical use of NAIs correlate with decreased immune reactivity to anti-influenza antibodies. The present study demonstrates that NAI treatment-emergent HA mutations can result in altered antigenic profiles and may potentially impact antibody-mediated virus inhibition.
Discussion
The relationship between antigenic drift and concomitant changes in HA receptor binding specificity/avidity has been well documented [
3,
4,
33‐
35]. Due to the proximity of the antigenic sites to the RBS, antigenic changes selected by neutralizing antibodies are often accompanied by changes in HA receptor binding properties [
3,
33,
36]. Immune pressure elicited by infection and/or immunization may favor influenza HA substitutions that facilitate antibody escape both by altering antigenicity and by increasing HA avidity for cell surface receptors. Subsequent absence of immune pressure may further induce compensatory substitutions that reduce avidity often by increasing the negative charge of the HA region and alter antigenicity [
3,
33,
34]. On the other hand, HA mutations that alter viral receptor affinity/specificity can contribute to NAI resistance by allowing efficient virus release from infected cells without the need for significant NA activity [
5,
37]. Although the effect of NAI treatment-emergent HA mutations on NAI susceptibility in humans remains uncertain, they are usually listed in the package inserts for NAI drug products. The possibility exists that NAIs may be associated with development of HA mutations that permit escape from natural or vaccine immunity by changing HA receptor avidity. The selective pressures mediated by NAIs and acquired immunity may also work together to accelerate the selection of beneficial compensatory HA mutations affecting receptor specificity, antigenicity, and/or functional compatibility with the NA protein.
Indeed, most of the NAI treatment-emergent HA1 mutations studied here involved changes in receptor binding avidity and specificity and, therefore, demonstrated drug-dependent phenotype in Calu-3 cells. The G155E mutation, which maps to the antigenic site Sa, and the S183P mutation, which is located within the RBS and overlaps with antigenic site Sb [
36,
38], have been shown to significantly increase HA receptor binding to α2,6-linked sialyl receptors [
38]. According to the previous studies, the S183P mutation enhanced virulence by altering binding to sialyl receptors in a mouse animal model [
3,
35,
37]. Moreover, the high frequency of the S183P mutation in contemporary H1N1 viruses in 2017–2018 (~ 28%) indicates that this mutation is being strongly selected for in humans. The D222G mutation, which maps to the antigenic site Ca [
17,
31,
39], was shown to be closely associated with the enhanced virulence of A(H1N1)pdm09 virus through the increased binding affinity to α2,3-linked sialyl receptors, while maintaining α2,6 specificity [
40‐
43]. In addition, S183P and D222G altered receptor binding avidity of the A/Puerto Rico/8/34 (H1N1) strain; however, whether a specific mutation results in increased or decreased receptor binding avidity depends on its genetic context [
3,
31,
34].
There is considerable uncertainty regarding the impact of HA mutations both on NAI susceptibility in in vitro systems and on virus inhibition and clinical response to NAIs in vivo. As noted in several NAI drug product labels, the impact of HA mutations on antiviral activity of NAIs is not well characterized in humans and is likely to be influenza virus strain-dependent. While some of the HA mutations within the RBS have been associated with a drug-dependent phenotype [
5,
7], the NAI-reduced susceptibility phenotype associated with HA mutations demonstrated in vitro has not yet been directly correlated with increased drug resistance in humans. Reduced NAI susceptibility associated with HA mutations in cell culture may not necessarily translate to reduced susceptibility in vivo because of differences in sialyl acid distribution patterns and HA binding requirements [
5,
37]. It is often difficult to define the selective pressure that leads to the emergence of HA substitutions that circulate in humans or emerge in NAI-treated patients. Thus, many HA mutations are pleiotropic in that they may be selected by NAIs, anti-HA antibodies, or changes in sialyl receptor distributions in a new host.
The clinical relevance of the studied HA mutations on susceptibility to NAIs remains unknown and they may represent cell culture and host adaptations. Several mutations did emerge during clinical use of NAIs or have been observed in clinical specimens. For example, the HA1 S162 N mutation was selected during intravenous zanamivir treatment [
23]. Subjects with the HA1 D222G mutation had significantly longer ICU stays: 22.8 days vs 14.0 days for those without this substitution. The D222G substitution was also found with considerable frequency in fatal and severe cases but was virtually absent among clinically mild cases [
18,
40]. Consistent with the previous reports, our data showed that recombinant virus containing D222G resulted in ~ 2 log
10PFU/mL higher titers over 72 h of replication in cell culture.
We found that several of the HA mutations associated with reduced susceptibility to NAIs correlate with decreased immune reactivity to polyclonal goat or ferret anti-influenza CA/04 antisera as measured by HI assay. Viruses containing two specific HA mutations (G155E and D222G) also demonstrated significantly decreased inhibition by anti-CA/04 mAbs and human sera collected from three different donors in MN assay. It is worth noting that these HA mutations were shown to be associated with antigenic drift in previously circulating H1N1 viruses and were also selected in escape mutants under mAb selective pressure [
17,
31,
39,
44]. Our findings suggest that chronic use of NAIs such as oseltamivir, zanamivir, or peramivir may promote acquisition of HA mutations that correlate with reduced sensitivity to inhibition by anti-influenza antibodies. The potential exists that NAIs selective pressure can contribute to the HA antigenic evolution.
Acknowledgments
We are grateful to Dr. Robert G. Webster (St. Jude Children’s Research Hospital, Memphis, TN) for providing plasmids for influenza A/California/04/09 (H1N1) virus and Dr. Dorothy Scott (FDA, CBER, Silver Spring, MD) for providing convalescent human serum samples from donors who were infected with the A(H1N1)pdm09 influenza virus. We thank Simone E. Adams (FDA, CDER, Silver Spring, MD) for excellent technical assistance. We also thank Drs. Jeffrey Murray, Debra Birnkrant, and Edward Cox at FDA CDER for helpful discussions and editorial suggestions.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.