Potential antigenic explanation for atypical H1N1infections among middle-aged adults during the2013–2014 influenza seasonSusanne L. Lindermana,b,1, Benjamin S. Chambersa,c,1, Seth J. Zosta,c,1, Kaela Parkhousea, Yang Lia,c, Christin Herrmanna,c,Ali H. Ellebedyd,e, Donald M. Carterf, Sarah F. Andrewsg, Nai-Ying Zhengg, Min Huangg, Yunping Huangg,Donna Straussh, Beth H. Shazh, Richard L. Hodinkai,j, Gustavo Reyes-Teránk, Ted M. Rossf, Patrick C. Wilsong,Rafi Ahmedd,e, Jesse D. Blooml, and Scott E. Hensleya,b,c,2

aWistar Institute, Philadelphia, PA 19104; bInstitute for Immunology, cDepartment of Microbiology, and iDepartment of Pathology and Laboratory Medicine,Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA 19104; jClinical Virology Laboratory, Children’s Hospital of Philadelphia,Philadelphia, PA 19104; dEmory Vaccine Center and eDepartment of Microbiology and Immunology, Emory University School of Medicine, Atlanta, GA 30322;fDepartment of Vaccine and Viral Immunity, Vaccine and Gene Therapy Institute of Florida, Port St. Lucie, FL 34987; gDepartment of Medicine, Universityof Chicago, Chicago, IL 60637; hNew York Blood Center, New York, NY 10065; kCenter for Research in Infectious Diseases, National Institute of RespiratoryDiseases, 14080 Mexico City, Mexico; and lDivision of Basic Sciences and Computational Biology Program, Fred Hutchinson Cancer Research Center,Seattle, WA 98109

Edited by Peter Palese, Icahn School of Medicine at Mount Sinai, New York, NY, and approved September 30, 2014 (received for review May 16, 2014)

Influenza viruses typically cause the most severe disease in childrenand elderly individuals. However, H1N1 viruses disproportionatelyaffected middle-aged adults during the 2013–2014 influenza sea-son. Although H1N1 viruses recently acquired several mutations inthe hemagglutinin (HA) glycoprotein, classic serological tests usedby surveillance laboratories indicate that these mutations do notchange antigenic properties of the virus. Here, we show that oneof these mutations is located in a region of HA targeted by anti-bodies elicited in many middle-aged adults. We find that over 42%of individuals born between 1965 and 1979 possess antibodiesthat recognize this region of HA. Our findings offer a possibleantigenic explanation of why middle-aged adults were highly sus-ceptible to H1N1 viruses during the 2013–2014 influenza season.Our data further suggest that a drifted H1N1 strain should be in-cluded in future influenza vaccines to potentially reduce morbidityand mortality in this age group.

influenza | antigenic drift | hemagglutinin | antibody | vaccine

Seasonal H1N1 (sH1N1) viruses circulated in the humanpopulation for much of the last century and, as of 2009, most

humans had been exposed to sH1N1 strains. In 2009, an anti-genically distinct H1N1 strain began infecting humans and causeda pandemic (1–3). Elderly individuals were less susceptible to2009 pandemic H1N1 (pH1N1) viruses because of cross-reactiveantibodies (Abs) elicited by infections with older sH1N1 strains(3–7). pH1N1 viruses have continued to circulate on a seasonalbasis since 2009. Influenza viruses typically cause a higher dis-ease burden in children and elderly individuals (8) but pH1N1viruses caused unusually high levels of disease in middle-agedadults during the 2013–2014 influenza season (9–12). For ex-ample, a significantly higher proportion of individuals aged 30-to 59-y-old were hospitalized in Mexico with laboratory-con-firmed pH1N1 cases in 2013–2014 relative to 2011–2012 (11).Most neutralizing influenza Abs are directed against the hem-

agglutinin (HA) glycoprotein. International surveillance labora-tories rely primarily on ferret anti-influenza sera for detectingHA antigenic changes (13). For these assays, sera are isolatedfrom ferrets recovering from primary influenza infections. Sea-sonal vaccine strains are typically updated when human influenzaviruses acquire HA mutations that prevent the binding of pri-mary ferret anti-influenza sera. Our laboratory and others havedemonstrated that sera isolated from ferrets recovering fromprimary pH1N1 infections are dominated by Abs that recognizean epitope involving residues 156, 157, and 158 of the Sa HAantigenic site (14, 15). The pH1N1 component of the seasonal

influenza vaccine has not been updated since 2009 because veryfew pH1N1 isolates possess mutations in residues 156, 157, and158. The majority of isolates from the 2013–2014 season havebeen labeled as antigenically similar to the A/California/07/2009vaccine strain (9).It is potentially problematic that major antigenic changes of

influenza viruses are mainly determined using antisera isolatedfrom ferrets recovering from primary influenza infections. Un-like experimental ferrets, humans are typically reinfected withantigenically distinct influenza strains throughout their life (16).In the 1950s, it was noted that the human immune system pref-erentially mounts Ab responses that cross-react to previouslycirculating influenza strains, as opposed to new Ab responsesthat exclusively target newer viral strains (17). This process,which Thomas Francis Jr. termed “original antigenic sin,” hasbeen experimentally recapitulated in ferrets (14, 18), mice (19–21), and rabbits (22). Our group and others recently demon-strated that the specificity of pH1N1 Ab responses can be shapedby prior sH1N1 exposures (14, 23–26). We found that ferrets

Significance

Influenza viruses typically cause a higher disease burden inchildren and the elderly, who have weaker immune systems.During the 2013–2014 influenza season, H1N1 viruses causedan unusually high level of disease in middle-aged adults. Here,we show that recent H1N1 strains possess a mutation thatallows viruses to avoid immune responses elicited in middle-aged adults. We show that current vaccine strains elicit im-mune responses that are predicted to be less effective in somemiddle-aged adults. We suggest that new viral strains shouldbe incorporated into seasonal influenza vaccines so that properimmunity is elicited in all humans, regardless of age and pre-exposure histories.

Author contributions: S.L.L., B.S.C., S.J.Z., T.M.R., P.C.W., R.A., J.D.B., and S.E.H. designedresearch; S.L.L., B.S.C., S.J.Z., K.P., Y.L., C.H., A.H.E., D.M.C., S.F.A., N.-Y.Z., M.H., Y.H., D.S.,B.H.S., J.D.B., and S.E.H. performed research; D.S., B.H.S., R.L.H., and G.R.-T. contributednew reagents/analytic tools; S.L.L., B.S.C., S.J.Z., R.L.H., G.R.-T., T.M.R., P.C.W., R.A., J.D.B.,and S.E.H. analyzed data; and S.L.L., B.S.C., S.J.Z., and S.E.H. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.1S.L.L., B.S.C., and S.J.Z. contributed equally to this work.2To whom correspondence should be addressed. Email: shensley@wistar.org.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1409171111/-/DCSupplemental.

15798–15803 | PNAS | November 4, 2014 | vol. 111 | no. 44 www.pnas.org/cgi/doi/10.1073/pnas.1409171111

sequentially infected with sH1N1 and pH1N1 viruses mount Abresponses dominated against epitopes that are conserved be-tween the viral strains (14). These studies indicate that primaryferret antisera may not be fully representative of human in-fluenza immunity.It has been proposed that increased morbidity and mortality of

middle-aged adults during the 2013–2014 influenza season isprimarily a result of low vaccination rates within these pop-ulations (27). An alternative explanation is that recent pH1N1strains have acquired a true antigenic mutation that has beenmislabeled as “antigenically neutral” by assays that rely on pri-mary ferret antisera. Here we complete a series of experimentsto determine if recent pH1N1 strains possess a mutation thatprevents binding of Abs in middle-aged humans who have beenpreviously exposed to different H1N1 strains.

ResultsRecent pH1N1 Strains Possess a Mutation That Prevents Binding ofHuman Antibodies. Antisera isolated from ferrets recovering fromprimary pH1N1 infections are highly specific for an epitope in-volving residues 156, 157, and 158 of the Sa HA antigenic site(14, 15). Very few pH1N1 isolates possess mutations in these Saresidues (Fig. S1); however, pH1N1 viruses recently acquireda K166Q HA mutation, which is located at the interface of theSa/Ca (28) antigenic sites (Fig. 1A). The K166Q HA mutationfirst arose during the 2012–2013 season and is now presentin over 99% of pH1N1 isolates (Fig. 1 B and C). Based onexperiments using primary antisera isolated from infected fer-rets, surveillance laboratories have reported that pH1N1 viruseswith the K166Q HA mutation are antigenically indistinguishablefrom the A/California/7/2009 pH1N1 vaccine strain (9).To address if human Abs are capable of recognizing pH1N1

viruses with the K166Q HA mutation, we performed hemag-glutination-inhibition (HAI) assays using sera from healthy

humans collected during the 2013–2014 influenza season inthe United States. Remarkably, 27% of sera from individualsborn between 1940 and 1984 possessed Abs specific for an epi-tope involving K166 (Fig. 2A and Table S1). Over 42% of indi-viduals born from 1965 to 1979 had K166 HA-specific Abs intheir sera (n = 54 individuals). Sera isolated from individualsborn between 1985 and 1997 (n = 49 individuals) did not havedetectable levels of K166 HA-specific Abs. Differences in K166HA-specificity were statistically significant between sera iso-lated from individuals born between 1965 and 1979 and indi-viduals born after 1985 (Fisher’s exact test; P < 0.0001). Similarresults were obtained when we analyzed sera from healthyhumans collected during the 2013–2014 influenza season inMexico (Fig. S2 and Table S2).It is remarkable that HAI assays, which are relatively in-

sensitive, were able to reproducibly detect K166-specific Abs in somany individuals in our experiments. Fig. 2A and Fig. S2 showpercentages of donors that had at least a twofold reduction in HAItiter using the K166Q HA mutant virus in three independentassays. It is worth pointing out that many sera samples had overfourfold reduced HAI titers using a pH1N1 virus engineered topossess the single K166Q HAmutation compared with the pH1N1vaccine strain (Table S1). Age-related differences in K166 HA-specificity among United States donors remained statistically sig-nificant using a fourfold reduction in HAI titer as a cut-off(Fisher’s exact test; P < 0.05 comparing donors born between 1965and 1979 and individuals born after 1985). Sera that had K166HA-specificity based on HAI assays failed to efficiently neutralizeK166Q-possessing viruses in in vitro neutralization assays (TableS3). K166 HA-specific sera were also unable to recognize a pri-mary viral isolate collected in 2013 (A/CHOP/1/2013) that pos-sesses a K166Q HA mutation (Table S3).

The K166 Epitope Is Shielded by a Glycosylation Site Present in sH1N1Viruses Circulating After 1985. Original antigenic sin Abs areoriginally primed by influenza strains that circulated in the past (14,17, 18, 22, 29). We propose that K166 HA-specific Ab responseswere likely primed by sH1N1 viruses circulating in humans before1985 and then boosted by the 2009 pH1N1 virus. Sera isolated fromindividuals born between 1965 and 1979 had the highest K166 HA-specificity (both in percent and titer) (Fig. 2A and Table S1). sH1N1viruses that circulated in the late 1970s and early 1980s share ex-tensive homology with pH1N1 viruses in the vicinity of K166 (Fig.2B). sH1N1 viruses were absent from the human population from1957 to 1976 and began infecting humans again in 1977. Therefore,humans born between 1957 and 1976 likely had their first H1N1encounter with a sH1N1 virus that shared homology with pH1N1viruses in the vicinity of K166.In 1986, sH1N1 viruses acquired a new glycosylation site at

HA amino acid 129 that is predicted to shield the epitope in-volving K166 (Figs. S3 and S4). The absence of K166 HA-specific responses in individuals born after 1985 is likely becausesH1N1 viruses glycosylated at HA amino acid 129 fail to primeK166 HA-specific responses. The lower number of K166 HA-specific responders born in the 1950s might also be attributed tounique glycosylation sites in sH1N1 viruses that circulated duringthis time period (Fig. S3), although precise glycosylation statusesof viruses circulating before 1977 are uncertain because of lim-ited numbers of sequenced viruses. Although we did not examinesera from very elderly individuals, it is possible that they alsohave immunodominant K166 HA-specific responses, becausea recent study reported that a mAb isolated from a survivor ofthe 1918 H1N1 pandemic binds to pH1N1 in an epitope in-volving K166 (6). There is considerable homology between the1918 H1N1 and the 2009 H1N1 in the vicinity of K166 (6).To experimentally address if glycosylation sites present in pre-

vious sH1N1 strains shield the epitope involving K166, we usedreverse-genetics to produce pH1N1 viruses that had glycosylation

B

res idue 166K

Q

other

1.0

A/Colorado/3514/2013

A/California/04/2009

A

KQother

amino acid 166

Year

Frac

tion

166

A/California/4/2009

A/Colorado/3514/2013

C

Fig. 1. pH1N1 viruses rapidly acquired HA mutation K166Q during the2013–2014 influenza season. (A) Residue K166 (red) is shown on theA/California/04/2009 HA trimer [PDB ID code 3UBN (6)]. (B) Plotted is the fre-quency of different amino acid identities at HA residue 166 in pH1N1 HAsequences as a function of time. Nearly all pH1N1 possessed K166 from 2009to mid-2012, but most isolates possessed Q166 by the 2013–2014 season. (C)A phylogenetic tree of pH1N1 viruses with branches colored according toamino acid identity at site 166 illustrates the rapid fixation of K166Q in re-cent pH1N1 isolates.

Linderman et al. PNAS | November 4, 2014 | vol. 111 | no. 44 | 15799

MICRO

BIOLO

GY

sites that were either present in sH1N1 strains from 1977 to 1985(sites 131+163) or 1986–2008 (sites 129+163). Western blot analysisrevealed that residues 129 and 131, but not residue 163, were gly-cosylated in our reverse-genetics derived viruses (Fig. S4B). Con-sistent with the hypothesis that the K166 epitope is shielded byglycosylation sites present in 1986–2008 sH1N1 viruses, K166 HA-specific human sera had reduced titers to pH1N1 viruses with the129 glycosylation site but normal titers to pH1N1 viruses with the

131 glycosylation site (Fig. S4C). As a control, we also completedHAI assays with sera from donors that were born in the 1970s whodid not have detectable levels of K166 HA-specific sera Abs. Asexpected, these sera did not have reduced titers to pH1N1 viruseswith the 129 glycosylation site, but interestingly, these sera did havereduced titers to pH1N1 viruses with the 131 glycosylation site (Fig.S4C). We previously demonstrated that Ab responses focused on anepitope near the 131 glycosylation site can be elicited by sequentialinfections with a sH1N1 virus from the early 1990s and the pH1N1virus (14). We speculate that donors in the “non-K166 HA-specific”group were previously infected with antigenically distinctsH1N1 strains compared with donors in the K166 HA-specificgroup (ie: A/Singapore/06/1986-like strain vs. A/Chile/01/1983-likestrain). Taken together, these data suggest that glycosylationsites on previously circulating sH1N1 viruses shield epitopesand influence the development of subsequent Ab responsesagainst pH1N1 virus.

Vaccination with Current pH1N1 Vaccine Strain Elicits K166 HA-Specific Abs. The pH1N1 vaccine strain has not been updatedsince 2009. We determined whether this vaccine strain, whichpossesses an HA that has K166, elicits K166 HA-specific Abs inhumans. First, we analyzed sera from individuals vaccinated in2009. All of the individuals in this cohort were born before 1984and most did not have pH1N1 Ab titers before vaccination(Table S4). Sera from 5 of 17 individuals possessed detectablelevels of K166 HA-specific Abs following vaccination (Fig. 3Aand Table S4). Sera from all five of these individuals had <1:40HAI titers against the K166Q HA mutant pH1N1 virus (TableS4). One K166 HA-specific individual (subject #1) possessedK166 HA-specific Abs before vaccination (Fig. 3A and TableS4). It is possible that this individual was naturally infected withpH1N1 before vaccination. All of the K166 HA-specific donorshad detectable prevaccination Ab titers against sH1N1 virusesfrom 1977 and 1983; however, we also found titers against thesestrains in some donors that did not have detectable levels ofK166 HA-specific serum Abs (Table S4). We also measuredbinding of 42 HA head-specific mAbs isolated from 12 adultdonors (born 1949–1985) that were vaccinated against thepH1N1 strain in 2009. Strikingly, 23% of these mAbs had re-duced binding to pH1N1 engineered to have the K166Q muta-tion (Fig. 3B). This finding is consistent with a previous reportthat identified several K166 HA-specific mAbs derived froma donor that was born before 1977 (30).We passively transferred a K166 HA-specific mAb (SFV009-

3F05) or a control mAb that binds equally to WT and K166Q-HApH1N1 viruses (SFV015-1F02) to BALB/c mice 12 h beforeinfecting with a lethal dose of WT or K166Q-HA pH1N1 viruses.Control animals that did not receive a mAb before infectionrapidly lost weight and died or needed to be euthanized (Fig.3C). Mice receiving the control SFV015-1F02 mAb before in-fection with WT or K166Q-HA pH1N1 viruses all survived withminimal weight loss (Fig. 3C). Mice receiving the K166 HA-specificSFV009-3F05 mAb survived following infection with WT pH1N1but rapidly lost weight and died or needed to be euthanized fol-lowing infection with K166Q-HA pH1N1 (Fig. 3C). These datasuggest that K166 HA-specific Abs can be less efficient at pre-venting disease in a mouse model following infection with apH1N1 virus possessing K166Q HA.

Can K166 HA-Specific Immunity Be Recapitulated in Ferrets forSurveillance Purposes? Current surveillance efforts rely heavilyon antisera isolated from ferrets recovering from primary in-fluenza virus infections. Ferret antisera could potentially bemore reflective of human immunity if isolated from animalssequentially infected with antigenically distinct viral strains.We attempted to elicit K166 HA-specific Abs in ferrets by

0/49 K166-specific

*A

23/54 K166-specific

1940

-44

1945

-49

1950

-54

1955

-59

1960

-64

1965

-69

1970

-74

1975

-79

1980

-84

1985

-89

1990

-94

1995

-97

0

10

20

30

40

50

Year born

% S

pecif

ic to

ant

igen

ic sit

ein

volvi

ng K

166

donors from United States

Homology betweenA/Chile/01/1983 andA/California/07/2009

B

166

Fig. 2. Adult humans possess Abs that bind to a region of HA that wasrecently mutated in pH1N1. (A) Sera were isolated from healthy donors (n =195) from the state of New York during the 2013–2014 influenza season. HAIassays were performed using viruses with either WT A/California/07/2009 HAor A/California/07/2009 HA with a K166Q HAmutation. For each sera sample,we completed three independent HAI assays. Raw HAI data are reported inTable S1. Percentages of samples that had at least a twofold reduction in HAItiter using the mutant virus in three independent experiments are shown.K166-specificity of sera from individuals born between 1965 and 1979 isstatistically significant compared with K166-specificity of sera from individ-uals born after 1985 (Fisher’s exact test; *P < 0.0001). (B) Homology betweenthe A/Chile/01/1983 sH1 and the A/California/04/2009 pH1 are shown usingthe crystal structure of the A/California/04/2009 HA [PDB ID code 3UBN (6)].Residue K166 is colored green. Amino acids that differ between A/Chile/01/1983 and A/California/04/2009 are shown in red. The glycan receptor isshown in black.

15800 | www.pnas.org/cgi/doi/10.1073/pnas.1409171111 Linderman et al.

sequentially infecting animals with older sH1N1 strains and thenthe A/California/07/2009 pH1N1 strain.We initially infected animals with a sH1N1 virus that circu-

lated in 1977 (A/USSR/90/1977), a sH1N1 virus that circulatedin 1983 (A/Chile/01/1983), or a sH1N1 virus that circulated in1986 (A/Singapore/06/1986). After 84 d, we reinfected ani-mals with the A/California/07/2009 pH1N1 strain. As controls,we infected some animals twice with A/California/07/2009 andother animals only once with A/California/07/2009. Three ofeight of the ferrets sequentially infected with A/Chile/01/1983and A/California/07/2009 mounted K166 HA-specific Abs de-tectable in HAI assays (Fig. 4 and Table S5). The 22 ferrets inthe other experimental groups did not mount detectable levelsof K166 HA-specific Abs. The difference in K166 HA-specificityis statistically significant comparing the A/Chile/01/1983-A/California/07/2009 group with the rest of the groups (3 of 8 vs.0 of 22; Fisher’s exact test P < 0.05). K166 HA-specific Abs werelikely not elicited in the A/Singapore/06/1986-A/California/07/2009 group because the K166 HA epitope is predicted to beshielded by a glycosylation site at residue 129 of A/Singapore/06/1986 (Fig. S4). It is interesting that K166 HA-specific Abs werenot elicited by A/USSR/90/1977-A/California/07/2009 sequentialinfections. This result is likely because of variation at residue125, which is close to residue 166 (Fig. S5). A/Chile/01/1983 andA/California/07/2009 both possess S125, whereas A/USSR/01/1977 possesses R125 (Fig. S5).

DiscussionOur studies show that recent pH1N1 viruses have acquired asignificant antigenic mutation that prevents binding of Abs eli-cited in a large number of middle-aged humans. For this reason,we propose that the pH1N1 vaccine strain should be updated.

Conventional serological techniques used by most surveillancelaboratories have failed to recognize the K166Q HA mutation asantigenically important (9). HAI assays are based on serial seradilutions and can only detect large antigenic changes. Manysurveillance-based laboratories ignore twofold reductions in HAItiter because these laboratories typically process thousands ofsamples, which prohibit the experimental precision that isrequired to reliably detect twofold differences in these assays.However, a true twofold reduction in HAI titer against a mu-tated strain indicates an extremely immunodominant Ab re-sponse. Although we reproducibly detected as low as twofoldHAI differences using K166Q HA mutated viruses (Tables S1and S2 show results from three independent HAI experiments),our HAI assays likely underestimate the number of individualsthat possess K166Q Abs. For example, we were able to isolateK166 HA-specific mAbs from pH1N1-vaccinated individualswhose sera yielded similar HAI titers using WT and K166Qmutated pH1N1 viruses. We also identified many human serasamples that had >fourfold reductions in HAI titer using pH1N1viruses with the K166Q HA mutation, and it is worth noting thatthese results would likely have been missed if we pooled humansera samples or simply compared overall geometric means ofHAI data with mutant viruses.We attempted to recapitulate K166 HA-specific immunity

in ferrets by sequentially infecting with sH1N1 strains and theA/California/07/2009 pH1N1 strain. Only three of eight ferretssequentially infected with A/Chile/01/1983 and A/California/07/2009 mounted levels of K166 HA-specific Abs that could bedetected by HAI assays. Outbred ferrets were used in theseexperiments, and the overall percentage of ferrets with K166HA-specificity (Fig. 4) is similar to the overall percentage ofhumans born in the 1970s with K166 HA-specificity (Fig. 2A).

A

WT+/K166Q+WT+/K166Q -

B

#6

10204080

160

HAI

tite

r

#1

pre post

10204080

160H

AI ti

ter

#7

10204080

160

HAI

tite

r

#13

10204080

160

HAI

tite

r

#21

10204080

160

HAI

tite

r

pre post pre post pre post pre post

60

70

80

90

100

70

80

90

100

60

70

80

90

100

7 9 1153 11971 53 11971 53160

wei

ght

(% o

rigin

al)

wei

ght

(% o

rigin

al)

(% o

rigin

al)

days (post infection)

PBS SFV009-3F05 mAb SFV015-1F02 mAbmAb reactivity C

wei

ght

WTK166Q

WTK166Q

≤ ≤ ≤ ≤ ≤

Fig. 3. Vaccination of middle-aged adults with the current pH1N1 vaccine strain elicits Abs that bind to a region of HA that is now mutated in most pH1N1isolates. (A) Healthy adult volunteers were vaccinated with a monovalent pH1N1 vaccine in 2009. Sera were isolated prevaccination and 30 d postvaccinationand HAI assays were performed using viruses with either WT A/California/07/2009 HA or A/California/07/2009 HA with a K166Q HA mutation. Shown are HAItiters for donors that possessed K166 HA-specific Abs following vaccination. Data are representative of three independent experiments. Raw HAI titers for alldonors are shown in Table S4. (B) ELISAs were completed using mAbs isolated from healthy adult volunteers that were vaccinated with a monovalent pH1N1vaccine in 2009. ELISAs were coated either with A/California/07/2009 (WT) or A/California/07/2009 with a K166Q HA mutation. Shown are percentage of mAbsthat bound to both viruses and percentage of mAbs that bound to the WT virus but not the mutant virus (n = 42 mAbs). Data are representative of twoindependent experiments. (C) A K166 HA-specific mAb (SFV009-3F05) or a mAb that recognizes both WT and K166Q-HA pH1N1 (SFV015-1F02) were injectedinto BALB/c mice (n = 4 per group). Twelve hours later, mice were then infected with 20,000 TCID50 of WT or K166Q-HA virus and weight loss and survivalwere recorded for 11 d. Data are representative of two independent experiments.

Linderman et al. PNAS | November 4, 2014 | vol. 111 | no. 44 | 15801

MICRO

BIOLO

GY

We speculate that variation in K166 HA-specificity in humansis because of variations in pre-exposure histories and geneticdifferences that impact B-cell repertoires. Studies are ongoing todetermine if genetic differences in B-cell repertoires amongferrets influence K166 HA-specificity.Our results offer a possible antigenic explanation for the increased

disease burden in middle-aged adults during the 2013–2014 in-fluenza season. Given that the specificity of Ab responses is alteredby pre-exposures, we propose that conventional serological techni-ques used to identify antigenically novel viruses should be reeval-uated. The usefulness of arbitrary HAI titer cutoffs and dependenceon antisera generated in previously naïve ferrets (31, 32) should bereconsidered. Although we believe that the pH1N1 vaccine shouldbe updated immediately, it is not clear if a pH1N1 vaccine strainwith Q166 HA will be able to break the original antigenic sin thatcurrently exists in some middle-aged individuals. Further studiesshould be designed to determine if an updated H1N1 vaccine strainwith Q166 HA elicits more effective Ab responses in different agedhumans with distinct sH1N1 exposure histories.

Materials and MethodsHuman Donors. Studies involving human adults were approved by the In-stitutional Review Boards of Emory University, Vaccine and Gene Therapy In-stitute of Florida, the National Institute of Respiratory Diseases of Mexico, andtheWistar Institute. Informed consentwas obtained. For all experiments, HAI andin vitro neutralization assays were completed at the Wistar Institute usingpreexisting and de-identified sera. We analyzed several sera panels in this study.We analyzed sera fromhealthy donors collected at theNewYork BloodCenter inFebruary of 2014. We analyzed sera from healthy donors collected at the Centerfor Research in Infectious Diseases at the National Institute of Respiratory Dis-eases in Mexico. We analyzed sera and mAbs derived from healthy donorsvaccinatedwithamonovalentpH1N1vaccine in 2009aspreviously described (23).

Viruses. Viruses possessing WT pH1N1 HA or K166Q pH1N1 HA were gen-erated via reverse-genetics using HA and NA genes from A/California/07/2009and internal genes from A/Puerto Rico/08/1934. All of these viruses wereengineered to possess the antigenically neutral D225G HA mutation (15),which facilitates viral growth in fertilized chicken eggs. Viruses were grownin fertilized chicken eggs and the HA genes of each virus stock were se-quenced to verify that additional mutations did not arise during propaga-tion. sH1N1 strains (A/USSR/90/1977, A/Chile/01/1983, A/Singapore/06/1986, A/Texas/36/1991, A/New Caledonia/20/1999, and A/Solomon Is-lands/03/2006) were also grown in fertilized chicken eggs. We isolated

a pH1N1 virus from respiratory secretions obtained from a patient from theChildren’s Hospital of Philadelphia in 2013 (named A/CHOP/1/13 in this re-port). For this process, de-identified clinical material from the Children’sHospital of Philadelphia Clinical Virology Laboratory was added to Madin-Darby canine kidney (MDCK) cells (originally obtained from the NationalInstitutes of Health) in serum-free media with L-(tosylamido-2-phenyl) ethylchloromethyl ketone (TPCK)-treated trypsin, Hepes, and gentamicin. Viruswas isolated from the MDCK-infected cells 3 d later. We extracted viral RNAand sequenced the HA gene of A/CHOP/1/13. We also used reverse-geneticsto introduce glycosylation sites into A/California/07/2009 (pH1N1) HA. Theconsensus sequence for N-linked glycosylation (N-x-S/T) was added at HAresidues 129, 131, and 163 by making the mutations D131T, D131N andN133T, and K163N, respectively. Similar results were obtained in HAI assayswhen we used glycosylation mutants grown in eggs or MDCK cells. Glyco-sylation at residues 129 and 131 was confirmed by treating concentratedvirus with PNGase-F (New England Biolabs) under denaturing conditions.The CM1-4 anti-HA1 antibody was used as a primary antibody, and a donkeyanti-mouse fluorescent secondary antibody (Licor) was used. Blots wereimaged using the Licor Odyssey imaging system at 800nm (secondary anti-body) and 700 nm (molecular weight marker).

Animal Experiments. Murine experiments were performed at the Wistar In-stitute according to protocols approved by the Wistar Institute InstitutionalAnimal Care and Use Committee. BALB/cmice (Charles River Laboratories) wereinjected with 25 μg of mAb intraperitoneally and then infected intranasalywith 20,000 TCID50 of WT or K166Q-HA pH1N1 virus 12 h later. As controls,some mice received an intraperitoneal injection of PBS before infection.Weight loss and survival was recorded for 11 d. Severely sick mice were eu-thanized. Ferret experiments were performed at the Vaccine and Gene TherapyInstitute of Florida in accordance with the National Research Council’s Guide forthe Care and Use of Laboratory Animals (33), the Animal Welfare Act, and theCDC/NIH Biosafety in Microbiological and Biomedical Laboratories handbook.Fitch ferrets (Marshall Farms) were infected with 1 × 106 PFU of sH1N1 virus andbled 14 and 84 d later. These ferrets were then infected with the A/California/07/2009 pH1N1 strain and bled 14 d later. Some ferrets were sequentiallyinfected with A/California/07/2009 (84 d between infections) and other ferretswere infected with only A/California/07/2009 and bled 14 d later.

HAI Assays. Sera samples were pretreated with receptor-destroying enzyme(Key Scientific Products or Sigma-Aldrich) and HAI titrations were performedin 96-well round bottom plates (BD). Sera were serially diluted twofold andadded to four agglutinating doses of virus in a total volume of 100 μL. Turkeyerythrocytes (Lampire) were added [12.5 μL of a 2% (vol/vol) solution]. Theerythrocytes were gently mixed with sera and virus and agglutination wasread out after incubating for 60 min at room temperature. HAI titers wereexpressed as the inverse of the highest dilution that inhibited four agglu-tinating doses of turkey erythrocytes. Each HAI assay was performed in-dependently on three different dates. Sera that had at least twofoldreduced HAI titers using K166Q HA mutant viruses in three independent HAIassays were labeled as “K166 HA-specific.”

ELISAs. Viruses for ELISAs were concentrated by centrifugation at 20,000 RPMfor 1 h using a Thermo Scientific Sorvall WX Ultra 80 Centrifuge witha Beckman SW28 rotor. Concentrated viruses were then inactivated byB-Propiolactone (BPL; Sigma Aldrich) treatment. Viruses were incubatedwith 0.1% BPL and 0.1M Hepes (Cellgro) overnight at 4C followed by a 90-min incubation at 37 °C. The 96-well Immulon 4HBX flat-bottom microtiterplates (Fisher Scientific) were coated with 20 HAU per well BPL-treated virusovernight at 4 °C. Each human mAb was serially diluted in PBS and added tothe ELISA plates and allowed to incubate for 2 h at room temperature. Asa control, we added the 70-1C04 stalk-specific mAb to verify equal coating ofWT and K166Q HA virus. Next, peroxidase conjugated goat anti-human IgG(Jackson Immunoresearch) was incubated for 1 h at room temperature. Finally,Sureblue TMB Peroxidase Substrate (KPL) was added to each well and the re-action was stopped with addition of 250 mM HCl solution. Plates were ex-tensively washed with water between each ELISA step. Affinities weredetermined by nonlinear regression analysis of curves of 6 mAb dilutions(18 μg/mL to 74 ng/mL) using Graphpad Prism. mAbs were designated as K166-specific if they had a Kd at least four times greater for the K166Q mutant thanfor the WT virus.

In Vitro Neutralization Assays. Sera were serially diluted and then added to100 TCID50 units of virus and incubated at room temperature for 30 min. Thevirus-sera mixtures were then incubated with MDCK cells for 1 h at 37 °C.Cells were washed and then serum-free media with TPCK-treated trypsin

USSR/77 Chile/83 Sing/86 Cal/09 none

10

20

30

40%

ferr

ets

with

K16

6 H

A s

peci

ficity

Cal/09 Cal/09 Cal/09 Cal/09 Cal/09 first infection second infection

0/8

3/8

0/8 0/3 0/3

Fig. 4. Ferrets sequentially infected with A/Chile/01/1983 and A/California/07/2009 develop K166 HA-specific Abs. Ferrets were infected with a sH1N1virus and then reinfected 84 d later with the A/California/07/2009 pH1N1virus. Sera were collected 14 d after the second infection and HAI assayswere completed using WT and K166Q-HA pH1N1 viruses. Shown are per-centages of samples that had at least a twofold reduction in HAI titer usingthe K166Q HA mutant virus in three independent experiments. Raw HAItiters are shown in Table S5. The difference in K166 HA-specificity is statis-tically significant comparing the A/Chile/01/1983-A/California/07/2009 groupwith the rest of the groups (3 of 8 vs. 0 of 22; Fisher’s exact test P < 0.05).

15802 | www.pnas.org/cgi/doi/10.1073/pnas.1409171111 Linderman et al.

was added. Endpoints were determined visually 3 d later. Data are expressedas the inverse of the highest dilution that caused neutralization. All sampleswere repeated in quadruplicate and geometric mean titer is reported.

Structural Modeling of HA Glycosylation Sites and Computational and PhylogeneticAnalyses of HA Sequences. Glycans were modeled using the GLYCAM-WebGlycoprotein Builder (www.glycam.org) as described in SI Materials andMethods. Computational and phylogenetic analyses of HA sequences werecompleted after downloading all full-length human pH1N1 sequencespresent in the Influenza Virus Resources as of February 23, 2014 as describedin SI Materials and Methods.

Statistical Analyses. For all sera experiments, we excluded samples that didnot have positive pH1N1 HAI titers. All samples that were pH1N1 HA-WT HAI-negative were also pH1N1 HA-K166Q HAI-negative. Samples were allocated

to specific groups based on age of donor. The year of birth of each samplewasavailable during the experiment, but this information was not assessed untileach experiment was completed. Variance of raw HAI titers was similarbetween different age groups. Fisher’s exact tests were completed using SASv9.3 software.

ACKNOWLEDGMENTS. Research reported in this publication was sup-ported by the National Institute of Allergy and Infectious Diseases ofthe National Institutes of Health under Awards 1R01AI113047 (to S.E.H.),1R01AI108686 (to S.E.H.), 1P01AI097092-01 (to P.C.W.), 1U19AI090023-03(to P.C.W.), and HSN266200700006C Center of Excellence for InfluenzaResearch and Surveillance (to R.A.), and the National Institute of GeneralMedical Sciences of the National Institutes of Health under AwardR01GM102198 (to J.D.B.). Support for shared resources used in this studywas provided by Cancer Center Support Grant P30CA010815 (to theWistar Institute).

1. Dawood FS, et al.; Novel Swine-Origin Influenza A (H1N1) Virus Investigation Team(2009) Emergence of a novel swine-origin influenza A (H1N1) virus in humans. N EnglJ Med 360(25):2605–2615.

2. Smith GJ, et al. (2009) Origins and evolutionary genomics of the 2009 swine-originH1N1 influenza A epidemic. Nature 459(7250):1122–1125.

3. Garten RJ, et al. (2009) Antigenic and genetic characteristics of swine-origin 2009 A(H1N1) influenza viruses circulating in humans. Science 325(5937):197–201.

4. Manicassamy B, et al. (2010) Protection of mice against lethal challenge with 2009H1N1 influenza A virus by 1918-like and classical swine H1N1 based vaccines. PLoSPathog 6(1):e1000745.

5. Skountzou I, et al. (2010) Immunity to pre-1950 H1N1 influenza viruses confers cross-protection against the pandemic swine-origin 2009 A (H1N1) influenza virus.J Immunol 185(3):1642–1649.

6. Xu R, et al. (2010) Structural basis of preexisting immunity to the 2009 H1N1 pan-demic influenza virus. Science 328(5976):357–360.

7. Jacobs JH, et al. (2012) Searching for sharp drops in the incidence of pandemic A/H1N1influenza by single year of age. PLoS ONE 7(8):e42328.

8. Mertz D, et al. (2013) Populations at risk for severe or complicated influenza illness:Systematic review and meta-analysis. BMJ 347:f5061.

9. Arriola CS, et al.; Centers for Disease Control and Prevention (CDC) (2014) Update:Influenza activity—United States, September 29, 2013-February 8, 2014.MMWRMorbMortal Wkly Rep 63(7):148–154.

10. Ayscue P, et al.; Centers for Disease Control and Prevention (CDC) (2014) Influenza-associated intensive-care unit admissions and deaths—California, September 29,2013–January 18, 2014. MMWR Morb Mortal Wkly Rep 63(7):143–147.

11. Dávila J, et al. (2014) Substantial morbidity and mortality associated with pandemic A/H1N1 influenza in Mexico, winter 2013–2014: Gradual age shift and severity. PLoSCurr 6:6.

12. Epperson S, et al.; Influenza Division, National Center for Immunization and Re-spiratory Diseases, CDC (2014) Influenza activity—United States, 2013–14 season andcomposition of the 2014–15 influenza vaccines. MMWR Morb Mortal Wkly Rep63(22):483–490.

13. Stöhr K, Bucher D, Colgate T, Wood J (2012) Influenza virus surveillance, vaccine strainselection, and manufacture. Methods Mol Biol 865:147–162.

14. Li Y, et al. (2013) Immune history shapes specificity of pandemic H1N1 influenzaantibody responses. J Exp Med 210(8):1493–1500.

15. Chen Z, et al. (2010) Generation of live attenuated novel influenza virus A/California/7/09 (H1N1) vaccines with high yield in embryonated chicken eggs. J Virol 84(1):44–51.

16. Miller MS, et al. (2013) Neutralizing antibodies against previously encountered in-fluenza virus strains increase over time: A longitudinal analysis. Sci Transl Med 5(198):198ra107.

17. Davenport FM, Hennessy AV, Francis T, Jr (1953) Epidemiologic and immunologicsignificance of age distribution of antibody to antigenic variants of influenza virus.J Exp Med 98(6):641–656.

18. Webster RG (1966) Original antigenic sin in ferrets: The response to sequential in-

fections with influenza viruses. J Immunol 97(2):177–183.19. Virelizier JL, Postlethwaite R, Schild GC, Allison AC (1974) Antibody responses to

antigenic determinants of influenza virus hemagglutinin. I. Thymus dependence of

antibody formation and thymus independence of immunological memory. J Exp Med

140(6):1559–1570.20. Virelizier JL, Allison AC, Schild GC (1974) Antibody responses to antigenic determi-

nants of influenza virus hemagglutinin. II. Original antigenic sin: A bone marrow-

derived lymphocyte memory phenomenon modulated by thymus-derived lympho-

cytes. J Exp Med 140(6):1571–1578.21. Kim JH, Skountzou I, Compans R, Jacob J (2009) Original antigenic sin responses to

influenza viruses. J Immunol 183(5):3294–3301.22. De St Groth F, Webster RG (1966) Disquisitions on original antigenic sin. II. Proof in

lower creatures. J Exp Med 124(3):347–361.23. Li GM, et al. (2012) Pandemic H1N1 influenza vaccine induces a recall response in

humans that favors broadly cross-reactive memory B cells. Proc Natl Acad Sci USA

109(23):9047–9052.24. Wrammert J, et al. (2011) Broadly cross-reactive antibodies dominate the human B

cell response against 2009 pandemic H1N1 influenza virus infection. J Exp Med 208(1):

181–193.25. Carter DM, et al. (2013) Sequential seasonal H1N1 influenza virus infections protect

ferrets against novel 2009 H1N1 influenza virus. J Virol 87(3):1400–1410.26. Pica N, et al. (2012) Hemagglutinin stalk antibodies elicited by the 2009 pandemic

influenza virus as a mechanism for the extinction of seasonal H1N1 viruses. Proc Natl

Acad Sci USA 109(7):2573–2578.27. Catania J, Que LG, Govert JA, Hollingsworth JW, Wolfe CR (2014) High intensive care

unit admission rate for 2013–2014 influenza is associated with a low rate of vacci-

nation. Am J Respir Crit Care Med 189(4):485–487.28. Caton AJ, Brownlee GG, Yewdell JW, Gerhard W (1982) The antigenic structure of the

influenza virus A/PR/8/34 hemagglutinin (H1 subtype). Cell 31(2 Pt 1):417–427.29. De St Groth F, Webster RG (1966) Disquisitions of original antigenic sin. I. Evidence in

man. J Exp Med 124(3):331–345.30. Krause JC, et al. (2011) Epitope-specific human influenza antibody repertoires di-

versify by B cell intraclonal sequence divergence and interclonal convergence. J Immunol

187(7):3704–3711.31. Smith DJ, et al. (2004) Mapping the antigenic and genetic evolution of influenza virus.

Science 305(5682):371–376.32. Koel BF, et al. (2013) Substitutions near the receptor binding site determine major

antigenic change during influenza virus evolution. Science 342(6161):976–979.33. National Research Council (2011) Guide for the Care and Use of Laboratory Animals

(National Academies Press, Washington, DC), 8th Ed.

Linderman et al. PNAS | November 4, 2014 | vol. 111 | no. 44 | 15803

MICRO

BIOLO

GY

Supporting InformationLinderman et al. 10.1073/pnas.1409171111SI Materials and MethodsStructural Modeling of HA Glycosylation Sites. Glycans were mod-eled onto positions 129 and 131 in the A/Solomon Islands/03/2006HA crystal structure (PDB ID code 3SM5) using the GLYCAM-Web Glycoprotein Builder (www.glycam.org). The particular glycanused for modeling was an N-linked glycan with a trimannosyl core(DManpa1-6[DGlcpNAcb1-2DManpa1-3]DManpb1-4DGlcpNAcb1-4DGlcpNAcb1-OH in Glycam notation), and default rotamer set-tings were used for modeling. To model the 131 glycosylation site,a T131N mutation was introduced using the PyMol structure viewerbefore the structure was uploaded to the GLYCAM-Web server.

Computational and Phylogenetic Analyses of HA Sequences. Theoccurrence of different amino acid identities at HA residues 166,156, 157, and 158 (H3 numbering) was analyzed by downloadingall full-length human pandemic H1N1 sequences present in theInfluenza Virus Resource (1) as of February 23, 2014. After purgingsequences that were less than full length, contained ambiguousnucleotide identities, lacked full (year, month, day) isolationdates, or were otherwise anomalous, the sequences were aligned.

Each calendar year was broken into four equal partitions be-ginning with January 1, and the frequencies of different aminoacids at each residue of interest for each partition was calculatedand plotted. Only amino acids that reached a frequency of atleast 1% in at least one of the year partitions are labeled in thelegend to the plot. For construction of phylogenetic trees, thesequence set was randomly subsampled to 10 sequences perquarter-year partition. BEAST (2) was then used to samplefrom the posterior distribution of phylogenetic trees with re-constructed sequences at the nodes, after date stamping the se-quences, using a JTT (3) with a single rate category with anexponential prior, a strict molecular clock, and relatively un-informative coalescent-based prior over the tree. Fig. 1C showsa maximum clade credibility summary of the posterior distribu-tion with branches colored according to the reconstructed aminoacid identity at site 166 with the highest posterior probability attheir descendent nodes. The tree was visually rendered usingFigTree. The input data and computer code used for this anal-ysis can be found on GitHub at github.com/jbloom/pdmH1N1_HA_K166_mutations.

1. Bao Y, et al. (2008) The influenza virus resource at the National Center for Bio-technology Information. J Virol 82(2):596–601.

2. Drummond AJ, Suchard MA, Xie D, Rambaut A (2012) Bayesian phylogenetics withBEAUti and the BEAST 1.7. Mol Biol Evol 29(8):1969–1973.

3. Jones DT, Taylor WR, Thornton JM (1992) The rapid generation of mutation datamatrices from protein sequences. Comput Appl Biosci 8(3):275–282.

Linderman et al. www.pnas.org/cgi/content/short/1409171111 1 of 8

a

Frac

tion

Year

Frac

tion

Frac

tion

Frac

tion

Year

Year

Year

b

c

d

residue 156

residue 157

residue 158

residue 166

Fig. S1. Sequence variation of pH1N1 HA. The residues in the dominant antigenic site recognized by primary ferret anti-sera (residues 156, 157, and 158 of theSa antigenic site) are highly conserved in pH1N1 (A–C). No variation greater than 1% occurred at residue 156 and very little variation occurred at residues 157and 158. For comparison, residue 166 (D; also shown in Fig. 1B) has undergone a complete change in the last year.

Linderman et al. www.pnas.org/cgi/content/short/1409171111 2 of 8

donors from Mexico

1970

-1984

1985

-1990

0

5

10

15

20

25

% S

peci

fic to

ant

igen

ic s

itein

volv

ing

K16

6

Year born

n=24

n=21

*

Fig. S2. Mexican donors born before 1985 possess Abs that bind to the region of HA that was recently mutated in pH1N1. Sera were isolated from healthydonors (n = 45) at the Center for Research in Infectious Diseases at the National Institute of Respiratory Diseases in Mexico during the 2013–2014 influenzaseason. HAI assays were performed using viruses with either WT A/California/07/2009 HA or A/California/07/2009 HA with a K166Q HA mutation. For each serasample, we completed three independent HAI assays. Raw HAI data are reported in Table S2. Percentages of samples that had at least a twofold reduction inHAI titer using the mutant virus in three independent experiments are shown. (Fisher’s exact test; *P < 0.05).

158

131129 163

165166 166

sH1N1 (A/Solomon Islands/2006) pH1N1 (A/California/2009)a b

cc

1918 1950s 1940s 1977-1985 2009-2014

none 129 only 131 only 165 only none

131 only 163 only 129+158+163 131+158+163 131+163

1986-2008

129+163 none

1930s

none 131 only 165 only 131+165

Fig. S3. Glycosylation status of H1N1 viruses. The crystal structures of sH1N1 (A) and pH1N1 (B) HAs are shown (PDB ID codes 3SM5 and 3UBN). Glycosylationsites that have appeared from 1918 to 2008 in sH1N1 viruses are highlighted in orange and residue 166 is shown in green. (C) Glycosylation status of H1N1viruses circulating during different time periods is shown as reported in Wei et al. (1). Very few viral sequences are available for 1930–1950 viruses andvariability of glycosylation sites in these viruses likely relates to egg adaptations. The majority of sH1N1 viruses circulating between 1977 and 1985 have the 131and 163 glycosylation sites and the majority of sH1N1 viruses circulating between 1986 and 2008 have the 129 and 163 glycosylation sites. Of note, althoughresidues 129 and 131 are very close in the linear sequence, they are located on opposite sides of the Sa/Sb ridge. The 131 glycosylation site is not expected tocover the K166 epitope, whereas the 129 glycosylation site potentially shields the K166 epitope.

1. Wei CJ, et al. (2010) Cross-neutralization of 1918 and 2009 influenza viruses: Role of glycans in viral evolution and vaccine design. Sci Transl Med 2(24):24ra21.

Linderman et al. www.pnas.org/cgi/content/short/1409171111 3 of 8

K166 HA-specific samplessample # WT K166Q 131+163-GLY 129+163-GLY WT/K166Q WT/131+163-GLY WT/129+163-GLY

70-23 160 <20 120 60 >8.0 1.3 2.770-36 240 40 160 80 6.0 1.5 3.070-52 120 <20 80 <20 >6.0 1.5 >6.070-84 160 40 160 80 4.0 1.0 2.070-86 160 <20 160 60 >8.0 1.0 2.7

non-K166HA-specific samplessample # WT K166Q 131+163-GLY 129+163-GLY WT/K166Q WT/131+163-GLY WT/129+163-GLY

70-16 160 160 160 160 1 1 170-17 160 160 80 160 1 2 170-18 160 160 60 160 1 2.7 170-24 160 160 <20 160 1 >8 170-44 160 160 40 160 1 4 1

a

129

131

166

b

c

WT

129+

163G

LY13

1+16

3GLY

163G

LYMW(KDa)

75

WT

129+

163G

LY13

1+16

3GLY

163G

LYMW(KDa)

37

PNGase - PNGase +

HAI titers Ratio (WT/mutant)

HAI titers Ratio (WT/mutant)

Fig. S4. Modeling 129 and 131 glycosylation sites. (A) Modeling of the putative glycosylation sites at residues 129 and 131 on the HA of A/Solomon Islands/6/2006 (PDB ID code 3SM5) was completed using Glycam software (Materials and Methods). Residues 129 and 131 are shown in orange and predicted sugarsare shown in black. Residue 166 is shown in green. The 129 glycosylation site present in most sH1N1 isolates circulating after 1985 is predicted to shieldthe antigenic site involving residue 166. The 131 glycosylation site is not predicted to shield the antigenic site involving residue 166. (B) Viruses possessingA/California/07/2009 HA with different putative glycosylation sites were created by reverse-genetics. HA from viruses with putative glycosylation sites introducedat residues 129+163 and 131+163 migrated slower compared with unmodified HA and HA from viruses with a putative glycosylation site at only residue 163.This finding indicates that residues 129 and 131, but not residue 163 was glycosylated in the reverse-genetics derived viruses. HA from all viruses migratessimilarily following PNGase treatment. PNGase treatment was completed under reducing conditions, so HA migrated faster compared with no PNGasetreatment. (C) HAI assays were completed using the reverse-genetics derived viruses possessing A/California/07/2009 with different glycosylation sites. FiveK166 HA-specific samples and five non-K166 HA-specific human samples were tested.

166125

Fig. S5. Homology betwen A/USSR/90/1977, A/Chile/01/1983, and A/California/07/2009. HA residues that differ between A/Chile/01/1983 and the A/California/07/2009 are shown in red. A few additional HA residues differ between the HAs of A/USSR/90/1977 and A/California/7/2009, and these are colored yellow. Ofnote, A/Chile/01/1983 and A/California/07/2009 both possess S125 whereas A/USSR/90/1977 possess R125. Residues 125 and 166 are next to each other in thestructure. PDB ID file 3UBN (A/California/04/2009 HA) was used to make this figure.

Linderman et al. www.pnas.org/cgi/content/short/1409171111 4 of 8

Table S1. HAI titers using sera from healthy donors from the United States

Sample ID YOB

Exp. 1 Exp. 2 Exp. 3

HAI titers Ratio HAI titers Ratio HAI titers Ratio

WT K166Q WT/K166Q WT K166Q WT/K166Q WT K166Q WT/K166Q

40–60 1940 160 80 2.00 160 80 2.00 160 60 2.6740–49 1943 80 <20 >4.00 60 <20 >3.00 80 <20 >4.0040–05 1948 160 80 2.00 160 80 2.00 160 80 2.0040–02 1948 120 60 2.00 60 30 2.00 80 30 2.6740–44 1948 320 80 4.00 320 80 4.00 320 80 4.0040–14 1949 240 80 3.00 240 80 3.00 240 80 3.0040–16 1949 240 <20 >12.00 160 <20 >8.00 160 <20 >8.0050–55 1952 160 80 2.00 160 60 2.67 160 80 2.0050–11 1952 120 40 3.00 80 30 2.67 80 30 2.6750–43 1953 80 40 2.00 80 40 2.00 80 40 2.0050–06 1959 120 60 2.00 80 40 2.00 80 30 2.6760–04 1963 80 40 2.00 80 40 2.00 80 40 2.0060–10 1963 160 80 2.00 240 80 3.00 160 60 2.6760–13 1964 240 120 2.00 320 120 2.67 240 120 2.0060–25 1964 60 20 3.00 80 30 2.67 60 20 3.0060–14 1967 160 80 2.00 120 60 2.00 120 60 2.0060–23 1967 320 120 2.67 320 160 2.00 320 160 2.0060–44 1968 160 80 2.00 160 80 2.00 160 80 2.0060–28 1969 320 160 2.00 320 160 2.00 320 160 2.0060–41 1969 120 60 2.00 160 80 2.00 160 80 2.0060–63 1969 120 60 2.00 120 60 2.00 120 60 2.0060–35 1969 80 40 2.00 80 30 2.67 80 30 2.6770–56 1971 80 40 2.00 80 30 2.67 80 40 2.0070–52 1971 160 <20 >8.00 120 <20 >6.00 80 <20 >4.0070–60 1972 160 80 2.00 160 80 2.00 160 80 2.0070–74 1972 160 80 2.00 240 80 3.00 240 80 3.0070–36 1972 240 60 4.00 240 30 8.00 320 30 10.6770–84 1973 120 40 3.00 160 30 5.33 160 40 4.0070–86 1973 160 40 4.00 160 30 5.33 160 40 4.0070–13 1975 640 320 2.00 640 320 2.00 640 320 2.0070–54 1975 160 80 2.00 80 30 2.67 120 60 2.0070–25 1975 160 80 2.00 80 30 2.67 120 40 3.0070–10 1975 160 30 5.33 80 30 2.67 80 30 2.6770–23 1975 160 <20 >8.00 120 <20 >6.00 160 <20 >8.0070–82 1976 160 80 2.00 160 60 2.67 160 80 2.0070–50 1976 240 80 3.00 160 60 2.67 120 60 2.0070–80 1978 320 160 2.00 320 160 2.00 320 160 2.0070–19 1978 80 <20 >4.00 80 <20 >4.00 80 <20 >4.0080–16 1980 80 30 2.67 60 20 3.00 80 40 2.0080–48 1983 640 120 5.33 640 80 8.00 640 80 8.00

Sera were isolated from 195 healthy donors from the state of New York during the 2013–2014 influenzaseason. HAI assays were performed using viruses with either WT A/California/07/2009 HA or A/California/07/2009HA with a K166Q HA mutation. For each sera sample, we completed three independent HAI assays. Shown aresamples that had at least a twofold reduction in HAI titer using the mutant virus in three independent experi-ments. YOB, year of birth.

Linderman et al. www.pnas.org/cgi/content/short/1409171111 5 of 8

Table S2. HAI titers using sera from healthy donors from Mexico

Exp. 1 Exp. 2 Exp. 3

HAI titers Ratio HAI titers Ratio HAI titers Ratio

Sample ID YOB WT K166Q WT/K166Q WT K166Q WT/K166Q WT K166Q WT/K166Q

N-33 1973 160 60 2.67 120 40 3.00 80 30 2.67N-42 1975 40 <10 >4.00 30 <10 >3.00 30 <10 >3.00N-93 1977 120 30 4.00 60 30 2.00 60 20 3.00N-72 1979 30 <10 >3.00 30 <10 >3.00 20 <10 >2.00N-70 1980 160 80 2.00 120 60 2.00 120 40 3.00

Sera were isolated from 45 healthy donors from Mexico during the 2013–2014 influenza season. HAI assayswere performed using viruses with either WT A/California/07/2009 HA or A/California/07/2009 HA with a K166QHA mutation. For each sera sample, we completed three independent HAI assays. Shown are samples that had atleast a twofold reduction in HAI titer using the mutant virus in three independent experiments.

Table S3. Characterization of K166 HA-specific sera

HAI titers In vitro neutralization titers

Sample ID YOB A/CAL/07/09-WT HA A/CHOP/1/2013 A/CAL/07/09-WT HA A/CAL/07/09-K166Q HA

40–49 1943 20 <10 24 1240–44 1948 160 30 453 9540–16 1949 80 <10 160 2070–52 1971 40 <10 190 2870–36 1972 120 10 381 9570–84 1973 40 <10 160 5770–86 1973 60 20 190 6770–23 1975 80 <10 226 2070–19 1978 30 <10 95 1080–48 1983 320 30 1,076 8080–10 1986 160 160 761 1,81080–26 1986 80 80 113 16080–74 1986 40 40 135 11380–30 1987 20 10 67 11380–34 1987 160 160 381 53880–56 1988 20 <10 17 3490–01 1990 40 10 48 8090–24 1991 60 20 34 9590–12 1992 160 60 190 22690–13 1992 80 80 190 320

We further characterized several sera samples that had immunodominant K166 HA-specific Abs (light gray:K166 HA-specific sera from individuals born before 1985; darker gray: sera from individuals born after 1985). HAIassays were completed using viruses with the A/California/07/2009 HA-WT and a primary pH1N1 virus isolatedfrom the Children’s Hospital of Philadelphia (CHOP) in 2013. Additional in vitro neutralization assays werecompleted using the reverse-genetics derived viruses. Data are representative of two independent experiments.For each neutralization assay, each sample was titered four times, and geometric mean of these quadruplicatesamples is reported.

Linderman et al. www.pnas.org/cgi/content/short/1409171111 6 of 8

Table S4. Vaccination elicits K166 HA-specific responses

Visit 1 (prevaccine) Visit 2 (postvaccine)

ID YOB USSR/77 Chile/83 Texas/91 NC/99 SI/06 CAL/09 CAL/09-K166Q CAL/09-WT CAL/09-K166Q

K166-specific donors

1 1979 240 160 320 60 80 120 20 160 206 1966 160 120 240 120 240 <10 <10 40 107 1966 40 60 30 10 60 <10 <10 120 <1013 1982 80 40 80 40 60 10 <10 160 1021 1958 60 30 80 40 80 <10 <10 60 20

Other donors2 1976 80 80 120 <10 <10 60 60 80 605 1949 10 <10 <10 10 40 10 <10 40 408 1961 40 <10 30 80 120 <10 <10 80 809 1977 80 80 80 <10 <10 <10 <10 120 8010 1983 80 <10 320 240 320 <10 <10 240 32015 1961 <10 <10 <10 <10 <10 <10 <10 80 8017 1956 40 <10 30 80 120 10 <10 20 2019 1961 80 40 160 80 80 <10 <10 20 2020 1945 40 <10 <10 <10 <10 <10 <10 240 32022 1956 30 30 60 <10 <10 <10 <10 240 24023 1975 40 <10 240 80 120 <10 <10 160 16024 1968 <10 <10 30 <10 <10 <10 <10 60 60

HAI assays were completed using sera isolated from healthy donors before and after vaccination with the monovalent pH1N1 virus in2009. Postvaccination sera were collected 30 d after vaccination. HAI assays using postvaccine sera were completed three independenttimes and assays using prevaccine sera were completed two independent times. Red indicates HAI titers equal to or less than 10 andyellow indicates HAI titers greater than 10.

Linderman et al. www.pnas.org/cgi/content/short/1409171111 7 of 8

Table S5. Sera from ferrets sequentially infected with sH1N1 viruses and pH1N1

Day 14 Day 84Day 98 (14 d aftersecond infection)

First infection Second infection Group Animal no. WT K166Q WT K166Q WT K166Q

USSR/90/77 A/California/07/09 Group 1 7982 <10 <10 <10 <10 800 800Group 1 7983 <10 <10 <10 <10 800 800Group 1 7984 <10 <10 <10 <10 300 300Group 1 7985 <10 <10 <10 <10 800 600Group 1 7991 <10 <10 <10 <10 2,400 2,400Group 1 7993 <10 <10 <10 <10 800 1,200Group 1 7994 <10 <10 <10 <10 600 600Group 1 7995 <10 <10 <10 <10 400 600

A/Chile/01/83 A/California/07/09 Group 2 7987 <10 <10 <10 <10 1,200 800Group 2 7988 <10 <10 <10 <10 800 400Group 2 7989 <10 <10 <10 <10 800 400Group 2 7990 <10 <10 <10 <10 1,600 1,600Group 2 7996 <10 <10 <10 <10 1,200 400Group 2 7997 <10 <10 <10 <10 240 160Group 2 8000 <10 <10 <10 <10 600 800Group 2 8001 <10 <10 <10 <10 800 800

A/Singapore/06/86 A/California/07/09 Group 3 1487 <10 <10 80 120 6,400 6,400Group 3 1488 <10 <10 <10 <10 3,200 3,200Group 3 1481 <10 <10 <10 <10 2,400 3,200Group 3 1504 <10 <10 60 80 4,800 6,400Group 3 8002 <10 <10 <10 <10 1,600 2,400Group 3 8009 <10 <10 <10 <10 1,600 2,400Group 3 8010 <10 <10 <10 <10 4,800 6,400Group 3 8012 <10 <10 <10 <10 1,600 2,400

A/California/07/09 A/California/07/09 Group 4 8013 400 400 200 200 160 160Group 4 8014 800 800 400 400 2,400 2,400Group 4 8015 800 800 300 200 2,400 1,600

None A/California/07/09 Group 5 1486 ND ND ND ND 1,600 2,400Group 5 1497 ND ND ND ND 1,600 1,600Group 5 1492 ND ND ND ND 1,200 1,600

Ferrets were infected with a sH1N1 virus and bled 14 and 84 d later. Animals were then infected with a pH1N1 strain and bled 14 d later (98 d post firstinfection). Sera were isolated and HAI assays were completed using WT and K166Q-HA pH1N1 viruses. Data are representative of three independent HAIassays. The three ferrets with K166 HA-specific Ab responses had >twofold changes in HAI titer using the K166Q virus in three independent experiments. Redhighlights twofold or greater reductions in HAI titer using K166Q-HA pH1N1 virus compared to WT-HA pH1N1 virus. ND, not determined.

Linderman et al. www.pnas.org/cgi/content/short/1409171111 8 of 8