Measuring Site-specific Glycosylation Similarity between Influenza a Virus Variants with Statistical Certainty

Abstract

Influenza A virus (IAV) mutates rapidly, resulting in antigenic drift and poor year-to-year vaccine effectiveness. One challenge in designing effective vaccines is that genetic mutations frequently cause amino acid variations in IAV envelope protein hemagglutinin (HA) that create new N-glycosylation sequons; resulting N-glycans cause antigenic shielding, allowing viral escape from adaptive immune responses. Vaccine candidate strain selection currently involves correlating antigenicity with HA protein sequence among circulating strains, but quantitative comparison of site-specific glycosylation information may likely improve the ability to design vaccines with broader effectiveness against evolving strains. However, there is poor understanding of the influence of glycosylation on immunodominance, antigenicity, and immunogenicity of HA, and there are no well-tested methods for comparing glycosylation similarity among virus samples. Here, we present a method for statistically rigorous quantification of similarity between two related virus strains that considers the presence and abundance of glycopeptide glycoforms. We demonstrate the strength of our approach by determining that there was a quantifiable difference in glycosylation at the protein level between WT IAV HA from A/Switzerland/9715293/2013 (SWZ13) and a mutant strain of SWZ13, even though no N-glycosylation sequons were changed. We determined site-specifically that WT and mutant HA have varying similarity at the glycosylation sites of the head domain, reflecting competing pressures to evade host immune response while retaining viral fitness. To our knowledge, our results are the first to quantify changes in glycosylation state that occur in related proteins of considerable glycan heterogeneity. Our results provide a method for understanding how changes in glycosylation state are correlated with variations in protein sequence, which is necessary for improving IAV vaccine strain selection. Understanding glycosylation will be especially important as we find new expression vectors for vaccine production, as glycosylation state depends greatly on the host species. Influenza A virus (IAV) mutates rapidly, resulting in antigenic drift and poor year-to-year vaccine effectiveness. One challenge in designing effective vaccines is that genetic mutations frequently cause amino acid variations in IAV envelope protein hemagglutinin (HA) that create new N-glycosylation sequons; resulting N-glycans cause antigenic shielding, allowing viral escape from adaptive immune responses. Vaccine candidate strain selection currently involves correlating antigenicity with HA protein sequence among circulating strains, but quantitative comparison of site-specific glycosylation information may likely improve the ability to design vaccines with broader effectiveness against evolving strains. However, there is poor understanding of the influence of glycosylation on immunodominance, antigenicity, and immunogenicity of HA, and there are no well-tested methods for comparing glycosylation similarity among virus samples. Here, we present a method for statistically rigorous quantification of similarity between two related virus strains that considers the presence and abundance of glycopeptide glycoforms. We demonstrate the strength of our approach by determining that there was a quantifiable difference in glycosylation at the protein level between WT IAV HA from A/Switzerland/9715293/2013 (SWZ13) and a mutant strain of SWZ13, even though no N-glycosylation sequons were changed. We determined site-specifically that WT and mutant HA have varying similarity at the glycosylation sites of the head domain, reflecting competing pressures to evade host immune response while retaining viral fitness. To our knowledge, our results are the first to quantify changes in glycosylation state that occur in related proteins of considerable glycan heterogeneity. Our results provide a method for understanding how changes in glycosylation state are correlated with variations in protein sequence, which is necessary for improving IAV vaccine strain selection. Understanding glycosylation will be especially important as we find new expression vectors for vaccine production, as glycosylation state depends greatly on the host species. In the 2017–2018 influenza season, nearly 80,000 influenza-related deaths were reported in the U.S., the highest death toll in 40 years (1Estimated Influenza Illnesses, Medical visits, Hospitalizations, and Deaths in the United States — 2017–2018 influenza season | Seasonal Influenza (Flu) | CDC (2018).Google Scholar). Effective vaccines help to mitigate disease, but influenza A virus (IAV) mutates rapidly, resulting in antigenic drift that causes the need for frequent updates to the influenza seasonal vaccine strains. IAV mutates during viral evolution in two ways. First, antigenic drift causes changes in the protein sequence to accumulate over time, altering the antigenic reactivity of the virus. Such changes may add N-glycosylation sequons that cause antigenic shielding, alter receptor binding avidity, and change the interactions with innate immune system lectins (2Both G.W. Sleigh M.J. Cox N.J. Kendal A.P. Antigenic drift in influenza virus H3 hemagglutinin from 1968 to 1980: multiple evolutionary pathways and sequential amino acid changes at key antigenic sites.J. Virol. 1983; 48: 52-60Crossref PubMed Google Scholar). Second, antigenic shift occurs when IAV RNA segments recombine with zoonotic genes to create new pandemic strains (3Webster R.G. Laver W.G. Air G.M. Schild G.C. Molecular mechanisms of variation in influenza viruses.Nature. 1982; 296: 115-121Crossref PubMed Scopus (335) Google Scholar). New pandemic strains start out with very low glycosylation in the head group, reflecting different selective pressures on IAV in animal hosts than in humans. The antigenic character of the abundant IAV envelope protein hemagglutinin (HA) depends on protein sequence (4Daniels R.S. Douglas A.R. Skehel J.J. Wiley D.C. Analyses of the antigenicity of influenza haemagglutinin at the pH optimum for virus-mediated membrane fusion.J. Gen. Virol. 1983; 64: 1657-1662Crossref PubMed Scopus (131) Google Scholar, 5Hensley S.E. Das S.R. Bailey A.L. Schmidt L.M. Hickman H.D. Jayaraman A. Viswanathan K. Raman R. Sasisekharan R. Bennink J.R. Yewdell J.W. Hemagglutinin receptor binding avidity drives Influenza A virus antigenic drift.Science. 2009; 326: 734-736Crossref PubMed Scopus (345) Google Scholar) and glycosylation state (6Skehel J.J. Stevens D.J. Daniels R.S. Douglas A.R. Knossow M. Wilson I.A. Wiley D.C. A carbohydrate side chain on hemagglutinins of Hong Kong influenza viruses inhibits recognition by a monoclonal antibody.Proc. Natl. Acad. Sci. U S A. 1984; 81: 1779-1783Crossref PubMed Scopus (335) Google Scholar, 7Vigerust D.J. Shepherd V.L. Virus glycosylation: role in virulence and immune interactions.Trends Microbiol. 2007; 15: 211-218Abstract Full Text Full Text PDF PubMed Scopus (408) Google Scholar). There are defined antigenic regions of HA protein sequence that evolve rapidly (8Wiley D.C. Wilson I.A. Skehel J.J. Structural identification of the antibody-binding sites of Hong Kong influenza haemagglutinin and their involvement in antigenic variation.Nature. 1981; 289: 373-378Crossref PubMed Scopus (1151) Google Scholar). Antigenic regions must balance their ability to evade host adaptive immune response with maintaining receptor-binding function (9Wagner R. Wolff T. Herwig A. Pleschka S. Klenk H.D. Interdependence of hemagglutinin glycosylation and neuraminidase as regulators of influenza virus growth: a study by reverse genetics.J. Virol. 2000; 74: 6316-6323Crossref PubMed Scopus (215) Google Scholar). HA antigenicity can be assessed through hemagglutination inhibition (HI) data. Correlating HI with protein sequence information allows prediction of antigenicity based on protein sequence (10Cai Z. Zhang T. Wan X.-F. Antigenic distance measurements for seasonal influenza vaccine selection.Vaccine. 2012; 30: 448-453Crossref PubMed Scopus (23) Google Scholar, 11Sun H. Yang J. Zhang T. Long L.-P. Jia K. Yang G. Webby R.J. Wan X.-F. Using sequence data to infer the antigenicity of influenza virus.mBio. 2013; 4: e00213-e00230Crossref Scopus (63) Google Scholar, 12Han L. Li L. Wen F. Zhong L. Zhang T. Wan X.-F. Graph-guided multi-task sparse learning model: a method for identifying antigenic variants of influenza A(H3N2) virus.Bioinformatics. 2019; 35: 77-87Crossref PubMed Scopus (8) Google Scholar). HA surface antigenicity is affected strongly by the presence and types of N-glycosylation, which cannot be predicted from genetic sequence or 3D protein coordinates (13Sun X. Jayaraman A. Maniprasad P. Raman R. Houser K.V. Pappas C. Zeng H. Sasisekharan R. Katz J.M. Tumpey T.M. N-linked glycosylation of the hemagglutinin protein influences virulence and antigenicity of the 1918 pandemic and seasonal H1N1 influenza A viruses.J. Virol. 2013; 87: 8756-8766Crossref PubMed Scopus (73) Google Scholar). N-Glycosylation sequons accumulate in the protein sequence, giving rise to increasing glycosylation as the strain evolves, shielding of antigenic sites (8Wiley D.C. Wilson I.A. Skehel J.J. Structural identification of the antibody-binding sites of Hong Kong influenza haemagglutinin and their involvement in antigenic variation.Nature. 1981; 289: 373-378Crossref PubMed Scopus (1151) Google Scholar, 14Ohshima N. Iba Y. Kubota-Koketsu R. Asano Y. Okuno Y. Kurosawa Y. Naturally occurring antibodies in humans can neutralize a variety of influenza virus strains, including H3, H1, H2, and H5.J. Virol. 2011; 85: 11048-11057Crossref PubMed Scopus (91) Google Scholar, 15Wanzeck K. Boyd K.L. McCullers J.A. Glycan shielding of the influenza virus hemagglutinin contributes to immunopathology in mice.Am. J. Respir. Crit. Care Med. 2011; 183: 767-773Crossref PubMed Scopus (69) Google Scholar), influencing receptor binding (16Alymova I.V. York I.A. Air G.M. Cipollo J.F. S. T. A. Zeng H. S. McCullers J.A. changes in the head of influenza hemagglutinin binding virus PubMed Scopus Google Scholar, P. Y. S. C. G. of hemagglutinin and neuraminidase of Influenza A virus as for and among influenza Scopus (23) Google Scholar) and binding to innate immune system in New New Scholar, M. M. L. K.L. of innate immune proteins and neuraminidase J. PubMed Scopus Google Scholar). the of to HA N-glycans site-specifically K. N. L. Y. S. H. J. of site-specific glycosylation in in a PubMed Scopus Google Scholar, A. Li X. N-glycosylation and site-specific of influenza proteins in the H1N1 candidate PubMed Scopus Google Scholar, Y. L.M. M. Cipollo J.F. of seasonal influenza vaccine for and immune Virol. Scopus Google Scholar, L.M. Y. Cipollo J.F. of Influenza hemagglutinin with glycosylation for PubMed Scopus Google Scholar). However, HA in the were defined information the glycan Our is K. J.A. N. R.J. K.L. J. and for Influenza A virus glycan and host 15: Full Text Full Text PDF PubMed Scopus (69) Google Scholar). N-Glycosylation is is a of at sites may be to by with sites K. J.A. N. R.J. K.L. J. and for Influenza A virus glycan and host 15: Full Text Full Text PDF PubMed Scopus (69) Google Scholar). of glycan the and host at HA is a of HA The that are are To antigenicity and immunogenicity with rigorous comparison of glycosylation among strains is A. M. immunogenicity of Virol. 2012; PubMed Scopus Google Scholar). the viral expression that can the are influenza vaccine effectiveness is from in the years Influenza Vaccine | Seasonal Influenza (Flu) | CDC (2018).Google Scholar). effective vaccine antigenicity of the candidate vaccine virus must be to circulating strains, but K. T. J. Influenza Scholar). the similarity of HA strains by correlating protein sequence with antigenicity can be antigenic which is a method for the antigenic distance of related protein in a (10Cai Z. Zhang T. Wan X.-F. Antigenic distance measurements for seasonal influenza vaccine selection.Vaccine. 2012; 30: 448-453Crossref PubMed Scopus (23) Google Scholar, Z. Zhang T. Wan X.-F. A for influenza antigenic PubMed Scopus Google Scholar). rigorous quantification of antigenicity, we that the similarity of glycosylation state of the candidate virus HA to circulating strains must be as However, a quantitative of glycosylation similarity between two strains is a of the of and IAV host glycosylation the expression to the vaccine greatly the resulting glycosylation of the vaccine virus. a method for of viral glycosylation and L. M. L. S. R. J. M. Y. J.R. site-specific N-glycosylation of envelope PubMed Scopus (131) Google Scholar, L. Y. N. M. J.R. site-specific of PubMed Scopus Google Scholar). method and the to which a is by or comparison of glycosylation is necessary to quantify as as there are no well-tested methods for rigorous glycosylation similarity among virus samples. The challenge is that the ability to and quantify glycopeptide is by Here, we present a the similarity M. A. similarity and methods for 2019; PubMed Scopus Google Scholar, D.J. A for PubMed Scopus Google Scholar), for the quantification of statistically rigorous similarity between two related virus strains that considers the presence and abundance of glycopeptide glycoforms. We the of data on the ability to rigorous similarity Using the we demonstrate our approach to similarity between two of site-specifically and at the protein The was and the HA glycosylation likely be the as from However, the of IAV vaccines are is to the glycosylation changes between related strains in The for HA or a and neuraminidase genes of influenza A/Switzerland/9715293/2013 virus were by and the viruses with genes of influenza virus were by reverse as J. Y. J. H. F. C. Wan X.-F. from to at the of hemagglutinin contributes to the antigenic drift in influenza 2013; PubMed Scopus Google Scholar). The HA mutations and with the WT HA and we the mutant as A/Switzerland/9715293/2013 virus and the mutant virus were in at for in were a from of of The viruses were as M. and of influenza PubMed Scopus Google Scholar). were for in to and to in a were and with and for at were by with for at in the was with for at in the The of in was to at with the of a of or was at of and the were at the were by the for at A from was for and the were as were to and the were with and with acid were in were in and with were to and in acid for The were with for of at were the from and were in with and as the were with of for of at was to acid from The the and were two The first was as The was at with the as the neuraminidase and The with were and data for were in for glycopeptide were with at were by a in with a and and a Using A and were for at in and a of with the and for and were were at at of for were at at for of and and at 40 for The abundant were data were for and were first to L. M. M. F. J. A. S. L. N. M. F. P. H. for 2011; Full Text Full Text PDF PubMed Scopus Google Scholar) R. S. L. J. K. N. N. C. L. K. C. L.M. F. J. P. S. T. A. J. T. K. J. Katz M. P. A for and 2012; 30: PubMed Scopus Google Scholar) from with no data from were the function of the J. L. M. Zhang Zhang Z. for and 2012; Full Text Full Text PDF Scopus Google Scholar). was against the M. a of protein 2011; PubMed Scopus Google Scholar) with the to for and the variants of HA and neuraminidase from strain and IAV protein from to the host The and and can be in and were or was as the with of and were and were as and The was at and a at a of two were for the for We a of at the by are in including identification and protein of the proteins to the were by the for the and for a the and the of for results were in K. J.A. N. R.J. K.L. J. and for Influenza A virus glycan and host 15: Full Text Full Text PDF PubMed Scopus (69) Google Scholar) for were the a that from J.A. L. J. of a for and site-specific identification of Full Text Full Text PDF PubMed Scopus Google Scholar). the of a in from the and a glycan We the glycan glycan with and The glycan was for and IAV were for were by the of the and by for the glycopeptide of a were glycopeptide for and for including with are in The glycopeptide from the variations in resulting from or of or of the the of variants from a with the glycan were that for glycan were In with a were We a to the similarity of glycosylation between two A and are vectors to the the abundance of glycopeptide in the the are to low to low abundance to be in and including and are vectors the of to the is a distance to and is the distance between A and First, a of is by comparing of from A and The be the are A is by the of of from A with from The similarity is the from from A to of The of of the and the with which we can quantify the glycosylation similarity between the two The in the is of and at were of of the for and mutant IAV HA were on a different but with the and a we that there be in glycopeptide and for two between and were as samples. and were by in The similarity comparing and mutant WT HA, were for the proteins as a and HA amino acid from the WT HA, and are in and for the protein and in and and for the site-specific for are when is and when the is WT in a new for of and was for and abundance for at and In the the similarity is by the and are the and A of similarity is on the with at and at In the the glycopeptide were to be in and including for of HA and and abundance for and In the the similarity is by the and are the and A of similarity is on the with at and at In the the glycopeptide were to be in and including for of mutant HA and WT and abundance for and In the the similarity is by the and are the and A of similarity is on the with at and at In the the glycopeptide were to be in and including similarity between mutant and WT glycosylation HA was in the WT HA for was in the WT HA for in a new We the to glycosylation similarity between two strains of HA, the of the of glycan including in the in we the data by and of on a similarity The the The is the of the to in a The in was by the between of from and of were from the data the of be to or The was of comparing of from to of The similarity is the from the of to of The and and their are to and and the in the and we the and that the two are we to the we have statistically to that the are The of to the of the of between the two is to the or the and the to the the or A level of the is of the of the and that for of in the and we have statistically to are or and we are A of that the is to of the of the and A of than that our data are to a the similarity between samples. is of with and and between the The were and of and are of The in and is that are very of may be of poor data and we cannot that are is of that have or that were very from the resulting in of the that in the as that there may have with a data we can of from of from a over the and The in are the to the in and in the were and of data We can the and of the by the of to In the for A a of and a of the for was that for was our is that the be at as as is to be information can be in Our of the similarity depends on the and in to the our we of information to defined as by the of the two The be when a of or The in and than in and were than and we have to a their To the of the we including The of N-glycosylation a of M.J. of the glycosylation sites of J. PubMed Scopus Google Scholar). we by acid and the glycosylation of the which we as is to have similarity to the out to be the and nearly were by the of and The for and for the protein and a similarity of and with of and to and are in and for and are in and the abundance for at are in low in a with the We from and resulting in a very as as a that the data was poor for The for was for and are in is a and with which we demonstrate the of the IAV are the neuraminidase envelope protein acid that can from the host we the HA with to we as We from our on site-specific of K. J.A. N. R.J. K.L. J. and for Influenza A virus glycan and host 15: Full Text Full Text PDF PubMed Scopus (69) Google Scholar) that have be and the sites with of similarity and on the abundance of the by N-glycans that have be and low similarity by The comparing and data for the HA protein as a a similarity of with However, the comparison a which data In of poor we were to for the our a of the was a different the was we to the data of the of in the data we the data for are in and for and were and are in and for the glycosylation sites that were in the comparison and their can be in with comparing the abundance for glycoforms. the of N-glycans that and of N-glycans in and the to their sites in were and and of and the were for two at and a of and a similarity in the at and similarity of but poor data in very in the for at the be there was in the for and as a the for comparison was is to a of the similarity at A/Switzerland/9715293/2013 (SWZ13) is a strain in the influenza vaccine of influenza virus vaccines for in the influenza season, Scholar, of influenza virus vaccines for in the influenza season, Scholar). N-glycosylation We the WT and a mutant in for HA amino acid from the and of or a glycosylation To the of the to which changes the of the HA in the of we the similarity for and However, the data from the WT HA was on a different data The and for comparison and were of and the will be for the mutant WT HA the similarity between the two was with a of The was with and as as that even variations that change of the glycosylation sites can in a of and glycan the for the mutant WT HA comparison are in We which sites to the similarity between mutant and WT HA the and the and a of there was for WT HA and the be for and data and the and were with the of The low for was of the for mutant HA the between the and was we that the data can be the between the and that we can be in sites similarity and low similarity data our that the glycosylation on the of HA and a of glycosylation similarity and on the head of HA and similarity the head immune to the antigenic regions of the head genetic amino acid of new sequons in the H. Z. J.M. Stevens J. and receptor binding of A(H3N2) virus PubMed Scopus (69) Google Scholar). We a similarity of and at and in the which our However, the sites in the head present a of and have similarity and a similarity of binding sites are on the head and sites are strains of IAV to viral recognition of host head glycosylation sites are or the receptor binding and antigenic sites H. Z. J.M. Stevens J. and receptor binding of A(H3N2) virus PubMed Scopus (69) Google Scholar), which that there is competing at sites to the ability to host while variation of amino acid sequence to evade host Our data that competing pressures are in the of similarity of in to the amino acid sequence changes or We that and glycosylation which that sites may be to the of the receptor-binding which low glycosylation similarity between our two strains, may be to to to protein a that is of the innate immune system K. J.A. N. R.J. K.L. J. and for Influenza A virus glycan and host 15: Full Text Full Text PDF PubMed Scopus (69) Google Scholar, X. K.L. The role and of of protein in innate host against Influenza A PubMed Scopus Google Scholar). to innate immune may be glycosylation variation at in the for in the difference that contributes to the low similarity was in the of the and glycoforms. The mutant HA abundance of with and of that the mutations the of the HA to the of the glycan K. J.A. N. R.J. K.L. J. and for Influenza A virus glycan and host 15: Full Text Full Text PDF PubMed Scopus (69) Google Scholar, A. of in Gen. PubMed Scopus Google Scholar). we cannot on the results of between as is that the in identification of glycosylation sites and in to the that were with In may be that glycosylation sites are by We methods for to the highest of data to in similarity for of J.A. L. J. of a for and site-specific identification of Full Text Full Text PDF PubMed Scopus Google Scholar). is a for of is for of glycopeptide the of selection by results in with a of our ability to similarity with are low in abundance with the in the glycopeptide to the of We to at the of of glycopeptide in a selection is for are by the and are the for of the are by The quantification for is at the to the and to the However, the to is and there is a to how can be of the of the a were to than resulting in ability to among which is we on to glycopeptide for the of the of that a cannot be predicted from quantification of from is and methods including M.J. The for quantitative PubMed Scopus Google Scholar) a of to including and among R. in Scholar, M. A. C. F. Cox J. S. M. during PubMed Scopus Google Scholar, and of protein in PubMed Scopus Google Scholar, in Scholar, L. L. M. P. S. Li J. Y. Li M. The of in a Full Text Full Text PDF PubMed Scopus Google Scholar). The for quantitative which must be to a of at is

Topics & Concepts