JOURNAL OF VIROLOGY, July 2008, p. 6200–6208
Copyright © 2008, American Society for Microbiology. All Rights Reserved.
Vol. 82, No. 13
Comparative Efficacy of Neutralizing Antibodies Elicited by
Recombinant Hemagglutinin Proteins from Avian
H5N1 Influenza Virus?
Chih-Jen Wei,1† Ling Xu,1† Wing-Pui Kong,1Wei Shi,1Kevin Canis,2James Stevens,3‡
Zhi-Yong Yang,1Anne Dell,2Stuart M. Haslam,2Ian A. Wilson,3and Gary J. Nabel1*
Vaccine Research Center, NIAID, National Institutes of Health, Bldg. 40, Room 4502, MSC-3005, 40 Convent Drive, Bethesda,
Maryland 20892-30051; Division of Molecular Biosciences, Faculty of Natural Sciences, Imperial College London,
London SW7 2AZ, United Kingdom2; and Department of Molecular Biology & Skaggs Institute for
Chemical Biology, The Scripps Research Institute, 10550 North Torrey Pines Road,
BCC206, La Jolla, California 920373
Received 25 January 2008/Accepted 1 April 2008
Although the human transmission of avian H5N1 virus remains low, the prevalence of this highly pathogenic
infection in avian species underscores the need for a preventive vaccine that can be made without eggs. Here,
we systematically analyze various forms of recombinant hemagglutinin (HA) protein for their potential efficacy
as vaccines. Monomeric, trimeric, and oligomeric H5N1 HA proteins were expressed and purified from either
insect or mammalian cells. The immunogenicity of different recombinant HA proteins was evaluated by
measuring the neutralizing antibody response. Neutralizing antibodies to H5N1 HA were readily generated in
mice immunized with the recombinant HA proteins, but they varied in potency depending on their multimeric
nature and cell source. Among the HA proteins, a high-molecular-weight oligomer elicited the strongest
antibody response, followed by the trimer; the monomer showed minimal efficacy. The coexpression of another
viral surface protein, neuraminidase, did not affect the immunogenicity of the HA oligomer, as expected from
the immunogenicity of trimers produced from insect cells. As anticipated, HA expressed in mammalian cells
without NA retained the terminal sialic acid residues and failed to bind ?2,3-linked sialic acid receptors. Taken
together, these results suggest that recombinant HA proteins as individual or oligomeric trimers can elicit
potent neutralizing antibody responses to avian H5N1 influenza viruses.
Since 1889, at least five influenza virus pandemics have oc-
curred, the most catastrophic of which was the Spanish influ-
enza of 1918, which resulted in 20 to 50 million deaths world-
wide (4, 8). Today, an average of about 200,000 influenza
virus-related hospitalizations and about 36,000 influenza virus-
related deaths occur in a typical winter-seasonal epidemic in
the United States (14). First appearing in 1997, the highly
pathogenic avian influenza H5N1 virus continues to spread
globally (19). The current global outbreak of H5N1 avian in-
fluenza virus among domestic and wild birds, and its potential
adaptation to humans, has accelerated influenza H5N1 virus
research and pandemic preparedness. More than 300 cases of
human H5N1 influenza virus infection had been confirmed. Of
these cases, nearly 200 individuals have died as a consequence
of infection (22). Although a few instances of human-to-human
H5N1 influenza virus transmission have been documented, the
current H5N1 virus has not yet acquired the ability to spread
efficiently within the human population, and most human cases
of H5N1 avian influenza virus are strongly associated with
exposure to infected domestic fowl (21).
Effective vaccination is a critical tool that supports public
health efforts to reduce influenza virus morbidity and mortal-
ity. Each year, the World Health Organization selects three
influenza virus strains as targets for inactivated vaccine devel-
opment. While the trivalent inactivated influenza virus vac-
cines currently used in the United States are manufactured
using embryonated eggs, it will be difficult to rapidly scale up
this technology for the mass production of vaccine in the event
of a potential pandemic (18). Recently, a new cell culture-
based approach for influenza virus vaccine development, in-
volving the production of influenza virus in cell culture fol-
lowed by virus inactivation and purification, has been proposed
and tested (1). While offering advantages over egg-based ap-
proaches, e.g., cell culture technology can be scaled up in
shorter periods of time, cell culture-based approaches for
H5N1 manufacture still require the production of a potentially
hazardous virus (1).
It has been demonstrated that protection provided by the
trivalent influenza virus vaccine is mediated primarily by anti-
hemagglutinin (HA) neutralizing antibodies. Thus, a recombi-
nant protein-based approach utilizing purified HA proteins
expressed in different mammalian systems offers another alter-
native for influenza virus vaccine development. This platform
provides advantages over current approaches, including well-
described technologies for mass production and reduced bio-
hazards during manufacturing. Various prototypes produced
* Corresponding author. Mailing address: Vaccine Research Center,
NIAID, National Institutes of Health, Bldg. 40, Room 4502, MSC-
3005, 40 Convent Drive, Bethesda, MD 20892-3005. Phone: (301)
496-1852. Fax: (301) 480-0274. E-mail: email@example.com.
† These authors contributed equally to this work.
‡ Present address: Molecular Virology and Vaccines Branch, Influ-
enza Division, NCIRD, CCID, Centers for Disease Control and Pre-
vention, 1600 Clifton Rd., Mail stop G-16, Atlanta, GA 30333.
?Published ahead of print on 16 April 2008.
in a baculovirus-insect cell expression system have proven safe
and effective in clinical studies for both H1N1 and H3N2
influenza viruses (7, 10, 11, 15–17). In this study, we systemat-
ically tested various recombinant HA proteins as alternatives
to egg-based vaccine candidates against influenza virus infec-
tion. H5N1 HA proteins were expressed and purified from
either insect or mammalian cells. The immunogenicity of dif-
ferent recombinant HA proteins was evaluated by antibody
neutralization. The data suggest that stable, trimeric viral
spikes serve as the optimal protein immunogens to elicit neu-
tralizing antibodies against H5N1 isolates, an approach that
may be applicable to seasonal influenza and other viruses.
MATERIALS AND METHODS
Genes and expression vectors. Based on H3 numbering (20), a cDNA corre-
sponding to residues 11 to 500 of the HA from A/Thailand/KAN-1/2004 (KAN-1;
GenBank accession no. AAS65615) was synthesized using human-preferred
codons as described previously (6) by using GeneArt (Regensburg, Germany).
This construct terminates at the bromelain cleavage site (12). Alternatively, the
14 amino acids (EISGVKLESIGIYQ) between the bromelain cleavage site and
the transmembrane domain of HA were inserted to produce the ?TM construct.
The original viral protease cleavage site PQRERRRKKRG was changed to
PQRETRG in order to retain the uncleaved and unprocessed proteins. The
purified protein contains additional residues at the C terminus (ISGRLVPRGS
PGSGYIPEAPRDGQAYVRKDGEWVLLSTFLGHHHHHH), in which the
thrombin cleavage site is in italics, the bacteriophage T4 fibritin foldon trimer-
ization sequence is underlined, and the His tag is in boldface (12). The inserts
were cloned into the cytomegalovirus/human T-cell leukemia virus type 1 repet-
itive sequence (CMV/R) 8?B expression vector for efficient expression in mam-
malian cells (6) or into the baculovirus transfer vector pAcGP67A (BD Bio-
sciences, Bedford, MA). Genes for NA(KAN-1)(H5N1) and NA(New
Caledonia/99)(H1N1) (GenBank accession nos. AY555150 and AJ518092,
respectively) also were synthesized using human-preferred codons (GeneArt,
Regensburg, Germany) and were cloned into the expression vector CMV/R 8?B.
Baculovirus production. HA proteins were produced by the cotransfection of
baculovirus transfer vector with BaculoGold-linearized baculovirus DNA (BD
Biosciences, Bedford, MA) into Spodoptera frugiperda (Sf9) cells (Invitrogen,
Carlsbad, CA) using the BaculoGold transfection buffer set (BD Biosciences,
Bedford, MA) and subsequently was amplified in the same cells according to the
Protein expression and purification. Plasmids expressing a secreted HA were
transfected into the human embryonic kidney cell line 293F using 293fectin
(Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions. 293F
cells were cultured in Freestyle 293 expression medium (Invitrogen, Carlsbad,
CA), and supernatant was collected 72 to 96 h posttransfection and cleared by
centrifugation and filtration. HA proteins were purified as previously described
(6), with minor modifications. Briefly, HA was recovered from the cell superna-
tant by metal affinity chromatography using Ni Sepharose high-performance
resin (GE Healthcare, Piscataway, NJ). Fractions containing HA were combined
and subjected to ion-exchange chromatography using a MonoQ HR10/10 column
(GE Healthcare, Piscataway, NJ). HA oligomers, trimers, and monomers then
were separated by gel filtration chromatography using a Hi-Load 16/60 Superdex
200-pg column (GE Healthcare, Piscataway, NJ). To remove the foldon se-
quence and His tag, HA proteins were subjected to thrombin digestion (EMD
Chemicals, Inc., San Diego, CA) at 3 U/mg at 4°C overnight. Insect-expressed
HA proteins (KAN-1) were purified as previously described (12). Trichoplusia ni
(Hi5) cells were infected at a multiplicity of infection of 10 and cultured in
Express Five serum-free medium (Invitrogen, Carlsbad, CA). The cell culture
was maintained at 27°C with gentle shaking. The culture suspension was col-
lected 96 h after infection, and the HA proteins were purified using the same
method as that described for mammalian cell-expressed proteins, except that,
after using the MonoQ column, HA protein was left overnight at 4°C to precip-
itate ferritin (12). HA protein from A/Vietnam/1203/2004 (VN1203) was purified
as previously described (12). The expression of the HA proteins was confirmed
by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and
Western blotting using a mouse monoclonal anti-Penta His antibody (Qiagen,
Hilden, Germany), mouse monoclonal anti-HA antibody 10D10 (24), or a rabbit
polyclonal anti-HA antibody (Immune Technology, New York, NY). Protein
purity also was examined by dynamic light scattering using a DynaPro plate
reader (Wyatt Technology, Santa Barbara, CA). Additionally, mammalian cell-
expressed HAs were produced by the cotransfection of a 1/10 ratio (wt/wt) of
NA(KAN-1) or NA(New Caledonia/99) expression vector for the glycan array
analysis, the mass spectrometry (MS) analysis of HA N-glycan composition, or
the NA-coexpressed HA proteins that also were used for immunization. The
molecular weights of the HA oligomer, trimer, or monomer proteins were de-
termined by density gradient sedimentation as previously described (9).
Vaccination. Female BALB/c mice (6 to 8 weeks old; Jackson Laboratories)
were immunized intramuscularly with 20 ?g of inactivated influenza virus sub-
virion vaccine [rgA/Vietnam/1203/2004 (H5N1); Biodefense and Emerging In-
fections Research Resources Repository, NIAID, NIH) or 20 ?g of purified
protein in 50 ?l of phosphate-buffered saline (PBS) (pH 7.4) and mixed with 50
?l of Ribi adjuvant (Sigma, St. Louis, MO) in PBS, pH 7.4, as recommended, at
weeks 0 and 3. Blood was collected 14 days after each immunization, and serum
was isolated. Animal experiments were conducted in full compliance with all
relevant federal regulations and NIH guidelines.
ELISA and isotyping of anti-HA antibodies. The mouse anti-HA immuno-
globulin G (IgG) and IgM enzyme-linked immunosorbent assay (ELISA) titers
were measured by a previously described method (23). Purified trimeric HA was
used to coat the plate, and anti-HA antibodies were detected by peroxidase-
conjugated goat anti-mouse IgG and IgM antibody (Jackson ImmunoResearch,
West Grove, PA). The subclasses of anti-HA antibodies also were determined by
ELISA using antibodies to IgA, IgG1, IgG2a, IgG2b, IgG3, and IgM (Calbio-
chem, Gibbstown, NJ).
Production of pseudotyped lentiviral vectors and measurement of neutralizing
antibodies. The recombinant lentiviral vectors expressing a luciferase reporter
gene were produced as previously described (6, 24).
Glycan array analysis. The glycan microarray analysis of the HA proteins was
performed as previously described (24).
MS of HA N-glycans. The HA N-glycans were prepared for MS analysis as
previously described (5). Briefly, purified HA glycoproteins were reduced, car-
boxymethylated, and digested with L-1-tosylamido-2-phenylmethyl chloromethyl
ketone (TPCK) bovine pancreas trypsin (EC 220.127.116.11). The N-glycans were
enzymatically released by digestion with PNGase F (EC 18.104.22.168; Roche Molec-
ular Biochemicals) and purified by reverse-phase C18Sep-Pak (Waters Corp.)
chromatography. Prior to MS analyses, the released N-glycans were permethyl-
ated and purified using a reverse-phase C18Sep-Pak (Waters Corp.). Matrix-
assisted laser desorption ionization–time of flight (MALDI-TOF) data were
acquired on a Voyager-DE STR mass spectrometer (PerSeptive Biosystems,
Framingham, MA) in the reflectron mode with delayed extraction. Permethyl-
ated samples were dissolved in 10 ?l of 80% (vol/vol) methanol in water, and 1
?l of dissolved sample was premixed with 1 ?l of matrix (10 mg/ml 2,5-dihy-
droxybenzoic acid [DHB] in 80% [vol/vol] aqueous methanol) before being
loaded onto a metal plate. MALDI-TOF/TOF experiments were performed on
a 4800 Proteomics Analyzer (Applied Biosystems, Framingham, MA) operated
in the reflectron positive ion mode.
Expression, purification, and characterization of HA pro-
teins. To evaluate the efficacy of recombinant HA proteins as
potential vaccine candidates, cDNAs encoding the ectodomain
of HA (A/Thailand/KAN-1/2004) were cloned into a baculo-
virus transfer vector, pAcGP67A, or a mammalian expres-
sion vector, CMV/R 8?B (6), to allow the efficient secretion
of HA proteins (Fig. 1A). The multibasic protease cleavage
site PQRERRRKKRG between HA1 and HA2 was mutated
to PQRETRG to reduce the efficiency of processing. To
stabilize the trimeric conformation of HA proteins, a bac-
teriophage-trimerizing foldon sequence was engineered into
the constructs, and a His tag was introduced at the COOH
terminus for purification purposes (12). A thrombin cleav-
age site was inserted between the HA and foldon sequence
to ensure the cleavage of the foldon and His tag, if neces-
sary. After the generation of baculovirus vectors expressing
HA proteins, the expression of HA proteins was carried out
by either infecting Hi5 cells or transfecting 293F cells with
mammalian expression vectors. The expression of secreted
VOL. 82, 2008INFLUENZA VACCINATION WITH RECOMBINANT HA PROTEINS6201
proteins was first confirmed by Western blotting using anti-
His tag or anti-HA antibodies. The secreted HA proteins
then were purified using a nickel affinity and MonoQ anion-
exchange column, followed by using a Superdex200 gel fil-
tration column to separate HA oligomer, trimer, and mono-
mer. In Hi5 cells, HA was expressed as two major species, a
high-molecular-weight oligomer and an uncleaved trimer
(Fig. 1B). The molecular sizes of insect-expressed HA oligo-
mer and trimer were estimated to be 1,321 and 214 kDa,
respectively, as determined by density gradient sedimenta-
tion. When the foldon sequence was removed by thrombin
digestion, the majority of the HA proteins appeared as a
cleaved trimer, and a small fraction of cleaved monomer
also was present (Fig. 1B). The expression of insect-ex-
pressed HA also was confirmed by SDS-PAGE (Fig. 1C) and
by Western blot analysis using an antibody against HA, and
this revealed a slightly lower molecular size after the cleav-
age of the trimerization motif and His tag (Fig. 1C). Its
removal was confirmed by Western blotting using an anti-
His tag antibody (Fig. 1C). The mammalian cell-expressed
HA also appeared as a high-molecular-mass oligomer and
trimer (Fig. 1B) with molecular masses of 1,394 and 222
kDa, respectively, as determined by density gradient sedi-
mentation. In contrast to the insect-produced protein, the
peak size of the trimer showed a higher molecular mass due
to the more extensive glycosylation in mammalian cells. In
addition, unlike its insect-expressed counterpart, the high-
molecular-mass oligomer remained intact after thrombin
FIG. 1. H5 HA expression vectors, biochemical purification, and characterization of insect- and mammalian cell-expressed HA proteins.
(A) Vectors encoding H5 HA (A/Thailand/KAN-1/2004) with a mammalian codon preference were prepared in the mammalian expression vector
CMV/R 8?B or the baculovirus transfer vector pAcGP67A. The cleavage site between HA1 and HA2 was mutated in order to obtain full-length
proteins. The clones were engineered to contain a trimeric foldon sequence and hexa-His tag (see Materials and Methods) at the COOH terminus,
which could be removed by thrombin digestion. These vectors were introduced into the 293F renal epithelial cells or Hi5 cells, as described in
Materials and Methods, to generate recombinant protein that was purified in monomeric, trimeric, or oligomeric forms for immunization studies.
(B) The HA proteins purified from insect or mammalian cells existed as a high-molecular-mass oligomer (?700 kDa) and trimer (uncleaved trimer;
?200 kDa) after gel filtration (red line). After thrombin digestion, HA protein eluted as a trimer (cleaved trimer) and a small fraction of monomer
(cleaved monomer) also could be detected (black line). The figures present superimposed elution profiles of insect- or mammalian cell-expressed
HA proteins overlaid with calibration standards (blue lines). (C) Insect or mammalian cell-derived proteins purified in the oligomer, trimer, and
cleaved trimer forms were analyzed by SDS-PAGE (left). Uncleaved trimer and cleaved trimer also were analyzed by Western blotting using a
polyclonal anti-HA antibody (middle), and the removal of the foldon domain and His tag from the trimer was confirmed by the decrease in the
size of the HA band and by using an anti-His tag antibody (right). HA-F, uncleaved HA with a foldon trimerization domain and His tag.
6202 WEI ET AL.J. VIROL.
cleavage, although trimeric and monomeric species were
detected (Fig. 1B). The expression of HA proteins of the
expected size also was confirmed by SDS-PAGE and West-
ern blot analysis (Fig. 1C). For subsequent immunogenicity
studies, only the peak fractions of each species were col-
lected in 2-ml aliquots. An analysis of these fractions by
dynamic light scattering confirmed that each immunogen
was of ?97% homogeneity (data not shown).
Humoral immune responses elicited by insect-expressed re-
combinant HAs. Humoral immunity in mice was evaluated by
vaccination with different forms of HA expressed in insect
cells. Mice were immunized with 20 ?g of oligomers, trimers,
or monomers in Ribi adjuvant twice at an interval of 3 weeks.
Antisera from the immunized animals were collected 14 days
after the second immunization and analyzed for neutralization
activity using a previously described HA/neuraminidase (NA)-
pseudotyped, lentiviral vector assay (24). To analyze the ability
of insect-expressed HA to induce neutralizing antibodies, HA/
NA-pseudotyped reporters were incubated with antisera from
immunized mice, and the neutralizing antibody activity was
measured using a luciferase assay. Sera from animals immu-
nized with insect-expressed oligomer and uncleaved trimer in-
hibited pseudovirus entry effectively, indicating the presence of
neutralizing antibodies to H5 HA (Fig. 2A). The thrombin-
digested trimer elicited only a modest level of neutralizing
antibody against KAN-1 HA (Fig. 2A), although a cleaved
trimer from a closely related strain (VN1203) elicited levels
comparable to those observed with uncleaved trimer (Fig. 2A).
These differences likely were related to the relative stability of
the cleaved HA trimer of these two strains when mixed with
adjuvant. The total HA antibodies were measured by ELISA
(Fig. 2D) and used to determine the ratio of neutralizing to
FIG. 2. Ability of the trimeric HAs from mammalian or insect cells to elicit neutralizing antibodies: higher neutralizing antibody responses were
elicited by oligomers than by trimers, and the titers increased with a repeat immunization. (A) The neutralization by antisera from five mice
immunized with insect-expressed HAs prepared from insect cells was assessed by the incubation of mouse sera with HA/NA-pseudotyped lentiviral
reporter vectors encoding luciferase. The percent neutralization was calculated by the reduction of luciferase activity relative to the values achieved
in the absence of sera. Among the insect cell-expressed proteins, oligomer, uncleaved trimer, and cleaved trimer from the VN1203 strain all elicited
potent neutralizing antibodies against KAN-1 HA/NA pseudovirus, while the cleaved trimer induced only modest neutralizing activity. (B) Mam-
malian high-molecular-mass oligomers induced the highest titer of neutralizing antibody response, followed by cleaved trimer, uncleaved trimer,
and cleaved monomer. Statistically significant differences were observed between oligomer and trimer (P ? 0.0001), oligomer and cleaved trimer
(P ? 0.0001), and oligomer and cleaved monomer (P ? 0.0001) forms. (C) Mice received a single injection of mammalian cell-expressed oligomer
or a secondary boost, as indicated, 3 weeks after the initial injection. The neutralizing antibody responses were measured in serum samples
collected 14 days after each vaccination using the HA/NA-pseudotyped lentivirus reporter assay. All animals were immunized with 20 ?g/injection
of HA protein in 50 ?l PBS and an equal volume of Ribi as the adjuvant. (D) Total HA antibodies in mice immunized with insect- or mammalian
cell-expressed HA proteins were measured by ELISA. (E) Antisera from mice immunized with insect- or mammalian cell-expressed HA proteins
were characterized for IgG1, IgG2a, and IgM responses. Similarly to mice immunized with inactivated H5N1 subvirion vaccine (inact. H5 vaccine),
both insect- and mammalian cell-expressed oligomeric and trimeric HA induced the highest levels of IgG2a, followed by IgG1. Naı ¨ve mouse serum
was used as a negative control. IgG3 and IgA also were tested but were negative for all sera (not shown).
VOL. 82, 2008INFLUENZA VACCINATION WITH RECOMBINANT HA PROTEINS6203
total binding antibodies. Among the insect cell-expressed pro-
teins, uncleaved trimer elicited the highest percentage of neu-
tralizing antibodies, followed by oligomer and cleaved trimer.
Isotypes of antibodies elicited by insect-expressed oligomer
and trimer were mostly IgG1 and IgG2a and were similar
whether the immunogen was generated in mammalian or
insect cells (Fig. 2E).
Comparison of the immunogenicity of mammalian cell-ex-
pressed oligomers, trimers, and monomers. Mammalian cell-
expressed HA oligomers, trimers, and monomers were ana-
lyzed for their ability to elicit neutralizing antibody against H5
HA. Vaccination in mice was performed similarly, and antisera
were collected and analyzed for their neutralizing activities.
Neutralizing antibody titers were significantly higher for ani-
mals immunized with oligomers (Fig. 2B). Cleaved trimers also
elicited a neutralizing antibody response, although the titer
was lower than that of the oligomers in this lentiviral neutral-
ization assay (Fig. 2B). While uncleaved trimers also induced
neutralizing antibody, no detectable antibody responses were
found in the animals immunized with cleaved monomers (Fig.
2B). When the ratio of neutralizing to total HA binding anti-
bodies (Fig. 2D) was calculated, antisera from mice immunized
with mammalian cell-expressed oligomer showed the highest
percentage of neutralizing antibodies, suggesting that these
complexes better preserve the physiologic trimeric spike struc-
ture. The ratios of neutralizing to total HA binding antibodies
in mice immunized with uncleaved and cleaved trimer were
?70% lower than those of mice immunized with oligomer.
While a single dose of mammalian cell-expressed oligomers
elicited only modest levels of neutralizing antibody, as shown
by the lentiviral neutralization assay, neutralizing titers against
H5 (KAN-1) pseudovirus were enhanced substantially after a
secondary boost 3 weeks after the initial injection (Fig. 2C).
Like insect-expressed HA, mammalian cell-expressed oligomer
and trimer injected with Ribi adjuvant elicited antibodies of
IgG1 and IgG2a subclasses that were similar to the antibodies
elicited by the inactivated influenza H5N1 subvirion vaccine
NA is not required for HA-elicited neutralizing antibody
responses. NA has been shown to play a role in viral release
from cells (2). This viral enzyme cleaves terminal sialic acid
residues from carbohydrate moieties on the surfaces of in-
fected cells and, therefore, promotes the release of progeny
viruses (2). NA also cleaves sialic acid residues from HA,
thereby preventing the aggregation of viruses (2). HA proteins
made with or without NA coexpression behaved differently in
the glycan array binding analysis (Fig. 3). HA trimers ex-
pressed without NA showed no prominent binding to any of
the glycans tested (Fig. 3A). In contrast, mammalian cell-ex-
pressed HA trimers that coexpressed NA preferentially bound
to ?2,3-linked sialic acid oligosaccharides (Fig. 3B).
The effect of the coexpression of NA on HA glycosylation
was investigated by MS analysis (Fig. 4, Table 1). The N-linked
glycans from HA (Fig. 4A) and HA coexpressed with NA-
(KAN-1)(H5N1) (Fig. 4B) or NA(New Caledonia/99)(H1N1)
(Fig. 4C) were purified, derivatized, and subjected to MALDI-
TOF (MS). The spectra derived from HA without the coex-
pression of NA showed predominantly complex type N-glycans
(m/z 1835 to 4586; Hex3HexNAc4Fuc-NeuAc4Hex7HexNAc6
Fuc) (Fig. 4), which is consistent with bi-, tri-, and tetra-anten-
nary core fucosylated structures. The sialylation of the complex
glycans is a predominant feature, with fully sialylated bi-, tri-,
and tetra-antennary structures being observed at m/z 2966,
3776, and 4586, respectively. The spectra derived from the
N-glycans of HA coexpressed with NA(KAN-1)(H5N1) again
showed compositions consistent with those of bi-, tri-, and
tetra-antennary core fucosylated structures (m/z 1835 to 3503;
Hex3HexNAc4Fuc-NeuAc1Hex7HexNAc6Fuc) (Fig. 4). How-
ever, the sialylation of these glycans has been greatly reduced,
as clearly observed by a comparison of the core fucosylated
tetra-antennary glycans. Without the coexpression of NA, core
fucosylated tetra-antennary glycans with 1, 2, 3, and 4 sialic
acid residues were observed in abundance at m/z 3503, 3864,
4225, and 4586, respectively (Fig. 4A, Table 1). The coexpres-
sion of NA caused a loss of the signals at m/z 3864, 4225, and
4586 and a concurrent increase in the abundance of the non-
sialylated core fucosylated tetra-antennary glycans at m/z 3142
(Fig. 4A, Table 1). Similarly, a dramatic reduction in HA
N-glycan sialylation was observed by the coexpression of NA
(New Caledonia/99)(H1N1) (Fig. 4C, Table 1).
Given that the mammalian cell-expressed oligomers elic-
ited the strongest neutralizing antibody response, we tested
whether the inclusion of NA during protein expression
would affect its immunogenicity. In the construct that ter-
minates at the bromelain cleavage site, the inclusion of NA
during protein expression reduced the ability of HA to elicit
antibodies that neutralize HA/NA-pseudotyped virus (Fig.
5A). Since these proteins appeared similarly on purification
after gel filtration, this difference most likely was due to
their stability in the Ribi adjuvant. In particular, the coex-
pression of NA with the bromelain site HA construct was
noted to reduce protein precipitation. In contrast, the in-
clusion of NA in the ?TM construct, which has 14 additional
amino acids between the bromelain cleavage site and the
transmembrane domain, did not affect the neutralizing an-
tibody response (Fig. 5B), suggesting that the additional
sequence stabilized the protein in the presence of adjuvant.
The rationale for testing the ?TM construct was that it
contained more of the HA protein ectodomain and might,
therefore, represent a more native protein with additional
determinants as an immunogen. However, the ?TM con-
struct primarily formed oligomers that lacked trimers, sug-
gesting that the inclusion of the additional amino acids
destabilized the trimeric form of the protein. Taking these
results together, the coexpression of NA is not required to
elicit neutralizing antibodies by transmembrane-deleted,
stabilized oligomers of the HA protein, suggesting that this
form of HA serves as a preferred immunogen.
Although egg-based vaccines have been used to combat sea-
sonal flu, the lengthy production cycles and limited manufac-
turing capacity of egg-based vaccines are not conducive to
facilitating a rapid response during a potential influenza virus
pandemic. Several clinical studies have shown that recombi-
nant HA-based vaccines purified from baculovirus expression
systems are safe and effective against H1N1 and H3N2 influ-
enza viruses (7, 10, 11, 15–17). The protein-based approach
represents an attractive alternative to egg-based technology,
6204WEI ET AL. J. VIROL.
since it uses HA proteins as antigens and does not require the
production of potentially dangerous live virus. Also, with this
approach, vaccine production is not limited by the supply of
eggs and can be easily scaled up in a good-manufacturing-
Recently, a recombinant HA vaccine against avian H5N1
influenza virus has demonstrated tolerability in humans (16).
However, this vaccine only induced protective neutralizing an-
tibody titers in 50% of the subjects receiving the highest dose
(two doses of 90 ?g vaccine). Since it has been previously
reported that recombinant HA proteins expressed in insect
cells tend to form monomers (13), the suboptimal immunoge-
nicity of this H5 HA vaccine may be due in part to recombinant
HA protein not being presented in its native trimeric confor-
mation. In this study, we cloned the ectodomain of HA from an
H5N1 virus (KAN-1) and expressed the HA proteins in mam-
malian or insect cells. HA proteins initially were purified using
a nickel affinity column followed by anion-exchange and gel
filtration chromatography. The entire purification process can
be completed in 2 to 3 days, and protein production can easily
be scaled up. In both Hi5 insect cells and 293F mammalian
cells, HA proteins were expressed as high-molecular-weight
oligomers and stabilized trimers, demonstrating that the tri-
merizing foldon sequence indeed prevented the HA from dis-
sociating into monomers. Upon the removal of the foldon
sequence by thrombin digestion, only trimers and monomers
were present in the insect-expressed proteins, whereas in the
mammalian cell-expressed proteins only a small portion of
monomer was observed after the removal of the foldon se-
quence. The discrepancy may be due to the different glyco-
sylation states of proteins derived from insect cells versus pro-
teins produced in mammalian cells, although it is certainly not
We then evaluated the immunogenicity of these different
FIG. 3. Glycan array analysis of the specificity of HA trimers expressed with or without NA. Glycan microarray analysis of H5 HA proteins
expressed without (A) or with (B) the coexpression of NA. Glycans with related linkages are grouped by color and include selected glycoproteins
(orange), predominantly ?2,3 sialosides (yellow) or ?2,6 sialosides (green), ?2,8 ligands (blue), and others (purple). This analysis was performed
by the Core H of the Consortium for Functional Genomics at Emory University.
VOL. 82, 2008 INFLUENZA VACCINATION WITH RECOMBINANT HA PROTEINS6205
forms of HA derived from either insect or mammalian cells
using an HA/NA-pseudotyped lentiviral system (24). In this
assay, the neutralization activity can be determined easily by
measuring the ability of antisera from mice immunized with
recombinant HA proteins to inhibit pseudovirus entry. It has
been shown that this pseudotype inhibition assay correlates
highly with traditional microneutralization and hemagglutina-
tion inhibition assays (6, 24) and can be easily performed in a
conventional biosafety level 2 laboratory with biosafety level 3
practices. Among the proteins produced from mammalian
cells, high-molecular-weight oligomers elicited the highest ti-
ters of neutralizing antibody, followed by the cleaved trimers
and uncleaved trimers. Cleaved monomers failed to induce
significant neutralizing antibodies against H5N1 virus, even
though anti-H5 antibodies were detected by ELISA. This may
be due to the preferential induction of antibodies against
epitopes present in the monomeric form and not in the trimer,
similarly to that observed with human immunodeficiency virus
type 1 gp120 monomers and trimers (reviewed in reference 3).
It also is possible that the monomeric form is less immunogenic
than the trimer/oligomer forms of the same protein. In a sep-
arate study, antisera from animals immunized with mammalian
cell-expressed oligomers or cleaved trimers were examined,
and their ability to elicit neutralizing antibodies against differ-
ent H5N1 strain pseudoviruses was similar (unpublished data),
though we cannot exclude the possibility of differences in their
fine specificity. It should be noted that, although Ribi adjuvant
does not contain any denaturants or reducing agents, its effect
on the stability and conformation of HA proteins is unknown.
We attempted to analyze this effect biochemically but were
unable to extract HA proteins from the lipid-rich components
of this adjuvant. Although NA plays an essential role in viral
replication and infection, the trimming of terminal sialic acid
from the HA proteins by NA did not affect the immunogenicity
of recombinant HA oligomers. However, the addition of NA
did prevent the precipitation of purified protein and facilitated
the production of the HA oligomers (data not shown). The
removal of terminal sialic acids by NA appeared to be impor-
tant for the receptor binding of HA. Glycan binding analyses of
HA expressed in the insect cell, which lacks sialic acids, have
revealed a similar ?2-3 specificity (12) to the NA-coexpression
mammalian HA protein, which bound to ?2,3-linked sialic acid
oligosaccharides. These findings are consistent with the obser-
vation that insect-produced, stabilized trimers elicited substan-
tial levels of neutralizing antibodies (Fig. 2A) despite the lack
of the sialylation of HA in this cell type.
Although previous studies have shown that recombinant HA
proteins derived from insect cells elicit immune responses (7,
10, 11, 15–17), our data provide evidence that oligomeric or
trimeric HA produced in mammalian cells are comparable or
slightly better in eliciting neutralizing antibodies against avian
H5N1 virus. Further testing will be required to determine
FIG. 4. MALDI-TOF mass spectra of permethylated N-glycans from HA trimers expressed with or without NA. Shown are profiles of N-glycans
from HA (A), HA coexpressed with NA isolated from the KAN-1 strain (B), or HA coexpressed with NA isolated from the New Caledonia (NC)
strain (C) from the 50% (vol/vol) acetonitrile fraction from C18Sep-Paks. All molecular ions are [M ? Na]?. Structural assignments are based on
monosaccharide composition, MALDI-TOF/TOF (MS/MS) analysis, and knowledge of N-glycan biosynthetic pathways.
FIG. 5. HA-elicited antibody response is independent of NA coex-
pression in oligomers from transmembrane-deleted, but not bromelain
site-truncated, HA. Neutralizing antibody titers in sera from mice
immunized with HA proteins produced with or without the coexpres-
sion of NA were examined. (A) In the bromelain site construct, the
inclusion of NA [bromelain site oligomer (?NA)] reduced HA’s ability
to induce neutralizing antibodies compared to that of the bromelain
site construct containing no NA [bromelain site oligomer (?NA)].
(B) In ?TM oligomers, coexpression with NA [oligomer (?TM,
?NA)] did not alter the immunogenicity of the HA proteins compared
to ?TM oligomer expression without NA [oligomer (?TM, ?NA)].
TABLE 1. Assignments of major molecular ions observed in
MALDI spectra of permethylated N-glycans from HA
trimers expressed with or without NA
aND, not determined.
VOL. 82, 2008 INFLUENZA VACCINATION WITH RECOMBINANT HA PROTEINS6207
whether other adjuvants, such as alum, QS-21, or MF-59, can Download full-text
improve the immunogenicity of recombinant HA proteins. Not
only could the potency of these adjuvants differ but also their
effects on the stability of the trimer may vary. These results
eventually will require validation with the most active and
manufacturable forms in human clinical trials. Nonetheless,
our data demonstrate that recombinant mammalian cell- or
insect-expressed trimeric HA proteins represent a promising
approach to the development of vaccines relevant to seasonal
and pandemic influenza virus.
We thank Ati Tislerics and Hamani Henderson for manuscript prep-
aration, Brenda Hartman and Michael Cichanowski for the prepara-
tion of figures, and members of the G.J.N. laboratory for helpful advice
and discussions. We acknowledge The Consortium for Functional Gly-
comics, funded by the NIGMS GM62116, and David F. Smith, Emory
University School of Medicine, Atlanta, GA, for the glycan array
analysis. We thank Jacob Lebowitz for performing the sedimentation
equilibrium measurements. Monovalent influenza virus subvirion vac-
cine, rgA/Vietnam/1203/2004 (H5N1), NR-4143, was obtained through
the NIH Biodefense and Emerging Infections Research Resources
Repository, NIAID, NIH.
This research was supported by the Intramural Research Program,
Vaccine Research Center, NIAID, NIH, and by the Biotechnology and
Biological Sciences Research Council (BBSRC) of the Wellcome Trust
(A.D. and S.M.H.). A.D. was supported as a BBSRC Professorial
1. Alymova, I. V., S. Kodihalli, E. A. Govorkova, B. Fanget, C. Gerdil, and R. G.
Webster. 1998. Immunogenicity and protective efficacy in mice of influenza
B virus vaccines grown in mammalian cells or embryonated chicken eggs.
J. Virol. 72:4472–4477.
2. Colman, P. M. 1994. Influenza virus neuraminidase: structure, antibodies,
and inhibitors. Protein Sci. 3:1687–1696.
3. Douek, D. C., P. D. Kwong, and G. J. Nabel. 2006. The rational design of an
AIDS vaccine. Cell 124:677–681.
4. Hampson, A. W. 1997. Surveillance for pandemic influenza. J. Infect. Dis.
5. Jang-Lee, J., S. J. North, M. Sutton-Smith, D. Goldberg, M. Panico, H.
Morris, S. Haslam, and A. Dell. 2006. Glycomic profiling of cells and tissues
by mass spectrometry: fingerprinting and sequencing methodologies. Meth-
ods Enzymol. 415:59–86.
6. Kong, W.-P., C. Hood, Z.-Y. Yang, C. J. Wei, L. Xu, A. Garcia-Sastre, T. M.
Tumpey, and G. J. Nabel. 2006. Protective immunity to lethal challenge of
the 1918 pandemic influenza virus by vaccination. Proc. Natl. Acad. Sci. USA
7. Lakey, D. L., J. J. Treanor, R. F. Betts, G. E. Smith, J. Thompson, E.
Sannella, G. Reed, B. E. Wilkinson, and P. F. Wright. 1996. Recombinant
baculovirus influenza A hemagglutinin vaccines are well tolerated and im-
munogenic in healthy adults. J. Infect. Dis. 174:838–841.
8. Morens, D. M., and A. S. Fauci. 2007. The 1918 influenza pandemic: insights
for the 21st century. J. Infect. Dis. 195:1018–1028.
9. Pancera, M., J. Lebowitz, A. Schon, P. Zhu, E. Freire, P. D. Kwong, K. H.
Roux, J. Sodroski, and R. Wyatt. 2005. Soluble mimetics of human immu-
nodeficiency virus type 1 viral spikes produced by replacement of the native
trimerization domain with a heterologous trimerization motif: characteriza-
tion and ligand binding analysis. J. Virol. 79:9954–9969.
10. Powers, D. C., J. E. McElhaney, O. A. Florendo, Jr., M. C. Manning, C. M.
Upshaw, D. W. Bentley, and B. E. Wilkinson. 1997. Humoral and cellular
immune responses following vaccination with purified recombinant hemag-
glutinin from influenza A (H3N2) virus. J. Infect. Dis. 175:342–351.
11. Powers, D. C., G. E. Smith, E. L. Anderson, D. J. Kennedy, C. S. Hackett,
B. E. Wilkinson, F. Volvovitz, R. B. Belshe, and J. J. Treanor. 1995. Influenza
A virus vaccines containing purified recombinant H3 hemagglutinin are well
tolerated and induce protective immune responses in healthy adults. J. In-
fect. Dis. 171:1595–1599.
12. Stevens, J., O. Blixt, T. M. Tumpey, J. K. Taubenberger, J. C. Paulson, and
I. A. Wilson. 2006. Structure and receptor specificity of the hemagglutinin
from an H5N1 influenza virus. Science 312:404–410.
13. Stevens, J., A. L. Corper, C. F. Basler, J. K. Taubenberger, P. Palese, and
I. A. Wilson. 2004. Structure of the uncleaved human H1 hemagglutinin from
the extinct 1918 influenza virus. Science 303:1866–1870.
14. Thompson, W. W., D. K. Shay, E. Weintraub, L. Brammer, N. Cox, L. J.
Anderson, and K. Fukuda. 2003. Mortality associated with influenza and
respiratory syncytial virus in the United States. JAMA 289:179–186.
15. Treanor, J. J., R. F. Betts, G. E. Smith, E. L. Anderson, C. S. Hackett, B. E.
Wilkinson, R. B. Belshe, and D. C. Powers. 1996. Evaluation of a recombi-
nant hemagglutinin expressed in insect cells as an influenza vaccine in young
and elderly adults. J. Infect. Dis. 173:1467–1470.
16. Treanor, J. J., B. E. Wilkinson, F. Masseoud, J. Hu-Primmer, R. Battaglia,
D. O’Brien, M. Wolff, G. Rabinovich, W. Blackwelder, and J. M. Katz. 2001.
Safety and immunogenicity of a recombinant hemagglutinin vaccine for H5
influenza in humans. Vaccine 19:1732–1737.
17. Wang, K., K. M. Holtz, K. Anderson, R. Chubet, W. Mahmoud, and M. M.
Cox. 2006. Expression and purification of an influenza hemagglutinin–one
step closer to a recombinant protein-based influenza vaccine. Vaccine 24:
18. Webby, R. J., and R. G. Webster. 2003. Are we ready for pandemic influ-
enza? Science 302:1519–1522.
19. Webster, R. G., M. Peiris, H. Chen, and Y. Guan. 2006. H5N1 outbreaks and
enzootic influenza. Emerg. Infect. Dis. 12:3–8.
20. Wilson, I. A., J. J. Skehel, and D. C. Wiley. 1981. Structure of the haemag-
glutinin membrane glycoprotein of influenza virus at 3 A resolution. Nature
21. World Health Organization. 2006. Basic health information on human avian
influenza A/H5N1. http://www.wpro.who.int/information_sources/databases
22. World Health Organization. 2007. Cumulative number of confirmed human
cases of avian influenza A/(H5N1) reported to WHO, 15 June 2007. http:
23. Yang, Z.-Y., B. K. Chakrabarti, L. Xu, B. Welcher, W.-P. Kong, K. Leung, A.
Panet, J. R. Mascola, and G. J. Nabel. 2004. Selective modification of
variable loops alters tropism and enhances immunogenicity of human im-
munodeficiency virus type 1 envelope. J. Virol. 78:4029–4036.
24. Yang, Z.-Y., C. J. Wei, W.-P. Kong, L. Wu, L. Xu, D. F. Smith, and G. J.
Nabel. 2007. Immunization by avian H5 influenza hemagglutinin mutants
with altered receptor binding specificity. Science 317:825–828.
6208WEI ET AL.J. VIROL.