Differential receptor binding affinities of influenza hemagglutinins on glycan arrays

Article abstract of DOI:10.1021/ja104657b

A library of 27 sialosides, including seventeen 2,3-linked and ten 2,6-linked glycans, has been prepared to construct a glycan array and used to profile the binding specificity of different influenza hemagglutinins (HA) subtypes, especially from the 2009 swine-originated H1N1 and seasonal influenza viruses. It was found that the HAs from the 2009 H1N1 and the seasonal Brisbane strain share similar binding profiles yet different binding affinities toward various ±2,6 sialosides. Analysis of the binding profiles of different HA subtypes indicate that a minimum set of 5 oligosaccharides can be used to differentiate influenza H1, H3, H5, H7, and H9 subtypes. In addition, the glycan array was used to profile the binding pattern of different influenza viruses. It was found that most binding patterns of viruses and HA proteins are similar and that glycosylation at Asn27 is essential for receptor binding.

Full text of DOI:10.1021/ja104657b

Published on Web 09/30/2010
Differential Receptor Binding Affinities of Influenza
Hemagglutinins on Glycan Arrays
Hsin-Yu Liao,†,‡ Che-Hsiung Hsu,†,§,| Shih-Chi Wang,†,‡ Chi-Hui Liang,†
Hsin-Yung Yen,†,⊥ Ching-Yao Su,†,⊥ Chien-Hung Chen,† Jia-Tsrong Jan,†
Chien-Tai Ren,† Chung-Hsuan Chen,† Ting-Jen R. Cheng,*,† Chung-Yi Wu,*,† and
Chi-Huey Wong*,†,§
Genomics Research Center, Academia Sinica, 128 Academia Road, Section 2, Nankang,
Taipei 115, Taiwan, Institute of Biochemistry and Molecular Biology, National Yang-Ming
UniVersity, Taipei 112, Taiwan, Chemical Biology and Molecular Biophysics, Taiwan
International Graduate Program, Academia Sinica, Taiwan, Institute of Bioinformatics and
Structural Biology, National Tsing-Hua UniVersity, Hsin-Chu 300, Taiwan and Institute of
Biochemical Sciences, National Taiwan UniVersity, Taipei, Taiwan
Received May 28, 2010; E-mail: tingjenc@gate.sinica.edu.tw (T.-J.R.C.); cyiwu@gate.sinica.edu.tw
(C.-Y.W.); chwong@gate.sinica.edu.tw (or wong@scripps.edu) (C.-H.W.)
Abstract: A library of 27 sialosides, including seventeen 2,3-linked and ten 2,6-linked glycans, has been
prepared to construct a glycan array and used to profile the binding specificity of different influenza
hemagglutinins (HA) subtypes, especially from the 2009 swine-originated H1N1 and seasonal influenza
viruses. It was found that the HAs from the 2009 H1N1 and the seasonal Brisbane strain share similar
binding profiles yet different binding affinities toward various R2,6 sialosides. Analysis of the binding profiles
of different HA subtypes indicate that a minimum set of 5 oligosaccharides can be used to differentiate
influenza H1, H3, H5, H7, and H9 subtypes. In addition, the glycan array was used to profile the binding
pattern of different influenza viruses. It was found that most binding patterns of viruses and HA proteins
are similar and that glycosylation at Asn27 is essential for receptor binding.
Introduction
neuraminic acid (Neu5Ac) residue, the most abundant derivative
of sialic acid found at the terminal end of glycoproteins or
Pandemic influenza outbreaks pose a significant threat to
public health as highlighted by the recent emergence of highly
pathogenic avian influenza H5N11 and the latest 2009 outbreaks
of swine-oriented H1N1 viruses (2009 H1N1).2,3 Influenza virus
infection is initiated by virus attachment to cell-surface sialoside
receptors via influenza hemagglutinin (HA). The HA binding
is then followed by viral neuraminidase cleavage of the terminal
sialic acid on receptors, and receptor-mediated endocytosis
accompanied by pH-induced conformational changes of HA to
allow virus-cell membrane fusion and entry for replication.4
Influenza HA is a glycoprotein that forms a trimer on the
virus surface.5 In contrast to protein-protein interactions, the
binding of HA to cell-surface receptors is dominated by
protein-oligosaccharide interactions mediated by the N-acetyl
glycolipids on the cell surface.6 Extensive studies of various
influenza virus subtypes and recombinant HAs have established
a correlation between sialoside binding preferences and viral
species: R2,6 linkage to galactose is preferred by the HA from
human isolates, R2,3 linkage to galactose is preferred by the
HA from avian isolates, and either R2,6 or R2,3 linkage to
galactose or glucose is recognized by the HA from swine
viruses.7-11 It is generally believed that acquirement of the R2,6
sialosides recognition ability of an influenza virus is a prereq-
uisite to its transmissions among humans. Therefore, under-
standing the receptor binding specificity of influenza viruses
would lead to development of sensitive and fast diagnostic tools
to detect and differentiate various subtypes of influenza viruses.
Glycan arrays have become a powerful tool for use to help
identify virus strains by monitoring changes in the receptor
† Genomics Research Center, Academia Sinica.
‡ Institute of Biochemistry and Molecular Biology, National Yang-Ming
University.
(6) Varki, A. Nature 2007, 446, 1023–1029.
§ Taiwan International Graduate Program, Academia Sinica.
| Institute of Bioinformatics and Structural Biology, National Tsing-Hua
University.
(7) Rogers, G. N.; Pritchett, T. J.; Lane, J. L.; Paulson, J. C. Virology
1983, 131, 394–408.
(8) Gambaryan, A. S.; Karasin, A. I.; Tuzikov, A. B.; Chinarev, A. A.;
Pazynina, G. V.; Bovin, N. V.; Matrosovich, M. N.; Olsen, C. W.;
Klimov, A. I. Virus Res. 2005, 114, 15–22.
⊥ Institute of Biochemical Sciences, National Taiwan University.
(1) Ferguson, N. M.; Fraser, C.; Donnelly, C. A.; Ghani, A. C.; Anderson,
R. M. Science 2004, 304, 968–969.
(9) Rogers, G. N.; Paulson, J. C. Virology 1983, 127, 361–373.
(10) Kumari, K.; Gulati, S.; Smith, D. F.; Gulati, U.; Cummings, R. D.;
Air, G. M. Virol. J. 2007, 4, 42.
(2) Garten, R. J.; et al. Science 2009, 325, 197–201.
(3) Smith, G. J. D.; Vijaykrishna, D.; Bahl, J.; Lycett, S. J.; Worobey,
M.; Pybus, O. G.; Ma, S. K.; Cheung, C. L.; Raghwani, J.; Bhatt, S.;
Peiris, J. S.; Guan, Y.; Rambaut, A. Nature 2009, 459, 1122–U107.
(4) Skehel, J. J.; Wiley, D. C. Annu. ReV. Biochem. 2000, 69, 531–569.
(5) Wilson, I. A.; Skehel, J. J.; Wiley, D. C. Nature 1981, 289, 366–373.
(11) Yen, H. L.; Aldridge, J. R.; Boon, A. C.; Ilyushina, N. A.; Salomon,
R.; Hulse-Post, D. J.; Marjuki, H.; Franks, J.; Boltz, D. A.; Bush, D.;
Lipatov, A. S.; Webby, R. J.; Rehg, J. E.; Webster, R. G. Proc. Natl.
Acad. Sci. U.S.A. 2009, 106, 286–291.
9
10.1021/ja104657b  2010 American Chemical Society
J. AM. CHEM. SOC. 2010, 132, 14849–14856 14849

A R T I C L E S
Liao et al.
binding profile12-17 and by dissecting the binding energy
contribution via quantitative analysis.17 This new tool may
provide an alternative approach to the current methods based
on RT-PCR (which is highly accurate but time-consuming)18
and ELISA based on antibody-nucleoprotein interaction (which
is quick but less accurate).19 A recent significant discovery by
using glycan array is that Tumpey’s group examined the glycan
binding properties of the 2009 A(H1N1) HA and found that
the soluble form of CA/04 HA exhibited specific preference to
longer R2,6 glycan (6’SLN-LN) structures. This binding
behavior showed a significant difference from seasonal influenza
H1N1 viral HA.15 Moreover, Feizi’s group also utilized a glycan
array to evaluate the binding specificity of different influenza
viruses and found that Cal/09 and Ham/09 H1N1 viruses bound
not only to the R2,6-linked sialosides but also to a considerable
range of the R2,3-linked sialyl sequences, and Mem/96 bound
exclusively to the R2,6-linked sequences.16
site and study their secondary structure and receptor binding
using circular dichroism and a designer synthetic glycan array.
Results and Discussion
Preparation of Sialosides Array for HA Binding Evaluation.
In naturally occurring sialosides, Neu5Ac is linked to other
sugars through an R-linkage. For example, Neu5Ac is commonly
found as a terminal sugar linked to galactosides through the
R2,3 or R2,6 linkage in both N-linked and O-linked glycopro-
teins and also to N-acetyl-galactosamine through the R2,6
linkage in O-linked glycoproteins. In addition, Neu5Ac can be
linked to another Neu5Ac residue via the R2,8 or R2,9 linkage
as a constituent of glycoproteins and glycolipids.22,23 Though
enzymatic sialylation provides R-linked sialosides stereospe-
cifically, the enzymes are not all readily available and are often
limited to, though with some exceptions, the synthesis of
naturally occurring sialosides due to their substrate specificity.24,25
This limitation precludes the synthesis of a wide range of
sialosdies. We have therefore developed an R-specific sialylation
donor (D) for the synthesis of various R-sialosides, including
the design of “sialylated disaccharide” as building blocks for
the one-pot synthesis of sialylated oligosaccharides.26 We have
further extended this strategy to the preparation of a sialoside
array. By using disaccharide A27-29 as building block, seventeen
R2,3-linked sialosides (1-17) of disaccharides to hexasaccharide
were synthesized, and, using the disaccharide B as building
block, several R2,6-linked sialosides (21-30) were prepared
(Table S1 and Schemes S2-S13 of the Supporting Information).
These glycans were spotted onto the NHS-activated glass slide
via amide bond formation12 to create an array of sialosides for
HA binding evaluation.
In contrast to the well-known binding preferences of various
HA subtypes, the effect of HA glycosylation on receptor binding
is still not well understood. The addition and processing of
N-linked oligosaccharides on a protein has been shown to be
important in maintaining protein conformation and stability and
in modulating biological activities.20 In addition, glycosylation
may protect proteins from clearance and proteolysis. In the case
of virus glycoproteins, glycans may also protect the virus from
immune attacks and modulate receptor recognition during
infection. In the case of influenza viruses, numerous reports have
demonstrated that loss of carbohydrates on HA can affect its
biological functions. For example, Deom et al.21 reported that
deletion of a complex oligosaccharide on the tip of the HA
would increase virus-cell interaction and make it easier for
the mutant viruses to survive in the infected cells. The increased
interaction between less-glycosylated HA and sialosides was
also demonstrated in our recent studies using sugar microar-
ray:17 the more outer glycans removed, the higher receptor
binding affinity and immunogeneity observed, leading to the
development of more potent HA vaccines with broader neu-
tralization activities.17 To further understand how individual
glycans of a full-length HA affect receptor binding, we report
here the expression of HA variants with a deleted glycosylation
Establishment of a Platform for Quantitative Analysis of
Sialosides Binding (KD Determination). Recombinant full-length
HAs with a C-terminal Strep-tag and a (His)6 tag were prepared
from human HEK293 cells for array detection. Strep-tag is an
eight-residue minimal peptide sequence with neutral charges
and exhibits intrinsic affinity toward streptavidin.30 The Strep-
tag was reported to not to hamper protein folding nor secretion
and it usually does not interfere with protein function.30 Prior
to analyze binding specificity of HA toward sialosides, a general
(12) Blixt, O.; et al. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 17033–17038.
(13) Stevens, J.; Blixt, O.; Glaser, L.; Taubenberger, J. K.; Palese, P.;
Paulson, J. C.; Wilson, I. A. J. Mol. Biol. 2006, 355, 1143–1155.
(14) Stevens, J.; Blixt, O.; Chen, L.-M.; Donis, R. O.; Paulson, J. C.;
Wilson, I. A. J. Mol. Biol. 2008, 381, 1382–1394.
(15) Maines, T. R.; Jayaraman, A.; Belser, J. A.; Wadford, D. A.; Pappas,
C.; Zeng, H.; Gustin, K. M.; Pearce, M. B.; Viswanathan, K.; Shriver,
Z. H.; Raman, R.; Cox, N. J.; Sasisekharan, R.; Katz, J. M.; Tumpey,
T. M. Science 2009, 325, 484–487.
(16) Childs, R. A.; Palma, A. S.; Wharton, S.; Matrosovich, T.; Liu, Y.;
Chai, W.; Campanero-Rhodes, M. A.; Zhang, Y.; Eickmann, M.; Kiso,
M.; Hay, A.; Matrosovich, M.; Feizi, T. Nat. Biotechnol. 2009, 27,
797–799.
(22) Varki, A. Glycobiology 1993, 3, 97–130.
(23) Muthana, S.; Yu, H.; Cao, H.; Cheng, J.; Chen, X. J. Org. Chem.
2009, 74, 2928–2936.
(24) Koeller, K. M.; Wong, C. H. Nature 2001, 409, 232–240.
(25) Hsu, C. H.; Chu, K. C.; Lin, Y. S.; Han, J. L.; Peng, Y. S.; Ren,
C. T.; Wu, C. Y.; Wong, C. H. Chem.sEur. J. 2010, 16, 1754–60.
(26) Komba, S.; Galustian, C.; Ishida, H.; Feizi, T.; Kannagi, R.; Kiso, M.
Angew. Chem., Int. Ed. 1999, 38, 1131–1133.
(17) Wang, C. C.; Chen, J. R.; Tseng, Y. C.; Hsu, C. H.; Hung, Y. F.;
Chen, S. W.; Chen, C. M.; Khoo, K. H.; Cheng, T. J.; Cheng, Y. S.;
Jan, J. T.; Wu, C. Y.; Ma, C.; Wong, C. H. Proc. Natl. Acad. Sci.
U.S.A. 2009, 106, 18137–18142.
(18) Vinikoor, M.; Stevens, J.; Nawrocki, J.; Singh, K. J. Clin. Microbiol.
2009, 47, 3055.
(27) Komba, S.; Yamaguchi, M.; Ishida, H.; Kiso, M. Biol. Chem. 2001,
382, 233–240.
(19) de Boer, G. F.; Back, W.; Osterhaus, A. D. Arch. Virol. 1990, 115,
47–61.
(28) Yamaguchi, M.; Ishida, H.; Kanamori, A.; Kannagi, R.; Kiso, M.
J. Carbohydr. Chem. 2004, 23, 201–215.
(20) Skropeta, D. Bioorg. Med. Chem. 2009, 17, 2645–2653.
(21) Deom, C. M.; Caton, A. J.; Schulze, I. T. Proc. Natl. Acad. Sci. U.S.A.
1986, 83, 3771–3775.
(29) Stevens, J.; Corper, A. L.; Basler, C. F.; Taubenberger, J. K.; Palese,
P.; Wilson, I. A. Science 2004, 303, 1866–1870.
(30) Schmidt, T. G. M.; Skerra, A. Nature Protocols 2007, 2, 1528–1535.
9
14850 J. AM. CHEM. SOC. VOL. 132, NO. 42, 2010

Sugar Binding Profiling of Hemagglutinin
A R T I C L E S
Figure 1. Sialoside structures to be spotted on glass slide for creating an array of sialosides for the HA binding studies.
platform needs to be established. Therefore, detection methods
using Cy3-conjugated HA for direct measurement,17,31 anti-His
antibodies followed by Cy3-conjugated secondary antibodies,13,14
and Cy3-conjugated streptavidin were parallelly compared and
evaluated for the feasibility in binding analysis. For the detection
using antibodies or streptavidin, a precomplexation strategy that
is incubation of HA with the Cy-3 conjugated partners in defined
ratios prior to applying the whole complexes onto the slide as
used.13,14 The results showed that: 1) three detection methods
resulted in no significant differences in binding patterns nor
binding intensities, 2) Cy3-conjugated HA showed binding
activities, indicating that HA forms functional trimers and the
binding detected using the other two approaches was not due
to multimerization caused by the antibodies or streptavidin, and
3) preliminary qualitative binding analysis revealed similar
binding affinities for the reported antibody-based approach
(KD,surf ) 150 for 15 in the case of H5) versus the streptavidin-
based strategy (KD,surf ) 145 ( 44 for 15 in the case of H5).
We therefore used the HA labeled with Cy3-SA for the
following binding analysis. An additional advantage of this
strategy is that it can provide a clear and consistent binding
profile for experiments using low concentrations of HA, which
benefits the application of quantitative biochemical analysis.
Differential Receptor Binding of HAs from Seasonal and
Pandemic Influenza Viruses. Our ultimate goal is to differentiate
influenza virus subtypes by only a specific set of glycans. In
this respect, recombinant HAs from the seasonal and pandemic
viruses were examined with respect to their receptor binding
specificity. The results of binding profiles for HAs from both
pandemic H1N1 (California/07/2009) (part A of Figure 2) and
seasonal H1N1 Brisbane/59/2007 (Br/59/07) (part B of Figure
2) displayed a similar pattern, with both higher binding activities
toward longer R2,6 sialosides. It was noticed that the maximum
binding affinity of the 2009 pandemic H1 reached with R2,6
sialoside containing 5 or 7 sugar units. Yet the H1 from Brisbane
strains showed the highest binding affinity toward the R2,6
sialoside containing 7 sugar units. The surface dissociation
constant values (KD, surf) were further determined using glycan
microassay based on the Langmuir isotherms.32 The monovalent
HA-sialoside binding is weak, exhibiting solution dissociation
constants in the millimolar range (KD ) 2.5 × 10-3 M) if
competition-based experiments were conducted.33 HA, however,
is involved in multivalent interactions with sialosides on the
host cell surface, which can be seen in the quantitative array
profiling.17,32 By the analysis of KD,surf (Table 1) of both strains,
the result revealed stronger binding capability of H1 from Br/
59/07 than the H1 from 2009 pandemic strain toward R2,6
sialosides, and this observation was also supported by the
phenomena that Br/59/07 H1 showed a high binding affinity
even when the protein was used as low as nanomolar concentra-
tions for sugars 29 and 30, making it difficult for the KD
determination. Compared to the earlier circulating strain H1N1/
New Caledonia/1999 (NC/99) (part C of Figure 2), it was shown
that recent H1N1 strains showed strong binding affinities toward
specific long R2,6 sialosides such as 29 and 30, implying
possibility that broader receptor specificity necessitate efficient
transmission of influenza virus. On the other hand, the H3 from
Brisbane/10/2007 (Br/10/07) showed a narrower binding profile
toward only two R2,6 sialosides, 28 and 30. It was surprising
that binding can only be observed with sialoside 30, the glycan
contains three repeats of LacNAc, but not 29 with two LacNAc
repeats (part D of Figure 2).
The same array was also used to profile the binding pattern
of avian flu H5 (H5N1/Vietnam/1194/2004) (part E of Figure
2), avian flu H7 (A/H7N7/Netherlands/219/03) (part F of Figure
2), and human flu H9 (A/H9N2/Hong Kong/1073/99) (part G
of Figure 2). As expected, the HAs from human and avian
viruses showed respective binding profiles. The results also
suggested that binding to 28 and 30 is unique to human viruses
(32) Liang, P. H.; Wang, S.-K.; Wong, C.-H. J. Am. Chem. Soc. 2007, 129,
11177–11184.
(31) Srinivasan, A.; Viswanathan, K.; Raman, R.; Chandrasekaran, A.;
Raguram, S.; Tumpey, T. M.; Sasisekharan, V.; Sasisekharan, R. Proc.
Natl. Acad. Sci. U.S.A. 2008, 105, 2800–2805.
(33) Sauter, N. K.; Bednarski, M. D.; Wurzburg, B. A.; Hanson, J. E.;
Whitesides, G. M.; Skehel, J. J.; Don, C. Biochemstry; Wiley: New
York, 1989; 28, p 8388-8396
9
J. AM. CHEM. SOC. VOL. 132, NO. 42, 2010 14851

A R T I C L E S
Liao et al.
Figure 2. Differential binding patterns of HA from H1N1, H3N2, H5N1, H7N7, and H9N2 viruses.
Table 1. KD of HA from SOV and Seasonal Flu Towards
disaccharides 22 and 23 may imply strong foothold among
human populations. These sugars, together with R2,3-trisac-
charides, can be used to differentiate HA subtypes and thus have
the potential to provide a method of quick test upon emergence
of an influenza outbreak.
R2,6-Sialosides
Binding Profiles of Real Viruses. To understand the relation-
ship of real viruses and HA proteins toward sialosides binding,
receptor-binding profiles of four isolates of the influenza virus
by glycan array analysis were compared directly by using the
same sialosides array. A clear distinction among the receptor-
binding repertoire of the Cal/09 H1N1, Brisbane H1N1,
Brisbane H3N1, RG14 H5N1, and B/Lee/40 was observed
(Figure 3). The Cal/09 H1N1 (part A of Figure 3) and Brisbane
H1N1 viruses (part B of Figure 3) bound not only to the majority
of R2,6 linked sialyl sequences but also to a considerable range
of R2,3 linked sialyl sequences. In contrast, H5N1 (part D of
Figure 3) bound exclusively to R2,3 linked sialyl sequences.
Similar to recombinant H3 proteins, the H3N2 influenza viruses
showed a preferential binding to R2,6 linked and R2,3 linked
sialyl sequences with strongest binding toward 28 and 30 (part
C of Figure 3). Interestingly, influenza B showed a similar
binding profile to both R2,3 and R2,6 sialosides (part E of Figure
a Standard symbols are used. Blue circle, Glc; blue square, GlcNac;
red diamond, Neu5Ac; yellow circle, Gal; yellow square, GalNac; green
circle, Man; red triangle, Fuc. b High binding activities but no concentration-
dependence was observed.
but not avian viruses. Furthermore, H1 binds glycan 24 (parts
A-C of Figure 2), whereas H3 shows no binding (part D of
Figure 2), thereby providing diagnostic potential for the dif-
ferentiation of H1 from H3. In addition, binding to the
9
14852 J. AM. CHEM. SOC. VOL. 132, NO. 42, 2010

Sugar Binding Profiling of Hemagglutinin
A R T I C L E S
Figure 3. Glycan array analyses of the five viruses investigated. The binding signals are shown as means of duplicate spots at 100 µM per spot. Each
experiment was repeated twice. Arrays consisted of twenty seven sialylated oligosaccharide probes, printed on NHS-coated glass slides (NHS: N-hydroxy
succinimide). The various types of terminal sialic acid linkage are indicated by the colored panels as defined at the bottom of the figure.
3). The broader specificity, namely, the ability to bind to R2,3
in addition to R2,6 linked receptors was also pertinent to the
greater virulence of the pandemic virus, and its capacity to cause
severe and fatal disease in humans. Binding to R2,3 linked
receptors is thought to be associated with the ability of influenza
viruses to infect the lower respiratory tract where there is a
greater proportion of R2,3 vs R2,6 linked sialyl glycans.
Differences in receptor binding among the viruses may therefore
formulate a good candidate for classifying the serotype of
influenza viruses.
The binding preference of RG14 was the same as that of
recombinant H5. In the case of H1N1 virus, the binding profile
using the whole virus is slightly different from the profile
obtained with recombinant proteins. Like recombinant HAs,
viruses showed the strongest binding toward long R2,6 sialo-
sides. However, the viruses also showed significant binding to
R2,3 sialosides, which was unusual for recombinant HAs. The
intrinsic binding affinity of sialosides for the hemagglutinin is
dominated by polyvalent interactions at the cell surface.
Therefore, weak monovalent binding may become significant
in multivalent interactions, and protein presentation, such as HA
orientation and density, on the cell surface may have a major
impact in receptor recognition. Furthermore, the tip of the
globular region harbors the receptor-binding pocket, which is
known to be crucial for the process of virus binding to its
receptor. The orientation, quantity, and structure of N-glycans
neighboring the receptor-binding pocket appear to be important
regulators of receptor specificity, which may also cause the
differences in binding preferences between recombinant HA and
whole virus.
host receptors to mediate virus entry. Glycosylation sites on
certain regions of HA are highly conserved.34 It was also
reported that changes in the HA glycosylation pattern may be
advantageous or detrimental to HA binding and eventually to
the survival of the virus.35 To elucidate the effects of site-specific
glycosylation on the HA binding to sialoside receptors, we
produced H5 (consensus HA, CHA536) mutants in which the
individual predicted glycosylation sequon Nx(S/T) was changed
to NxA to prevent N-glycosylation at the specific site (part A
of Figure 4). Except for ∆26, ∆170, ∆209, and ∆559, remove
of each of the HA glycosylation sites showed an increased
rate of gel mobility on SDS-PAGE compared with WT, corre-
sponding loss of glycosylation on the mutant (data not shown).
These glycosylation sites were consistent with the results from mass
sequencing analysis, where carbohydrates were identified to attach
to the sites N26/27, N39, N170, N181, N302, N500, and N559
(Data not shown). The subsequent sugar binding analysis suggested
that glycosylation at position 39, 302, 500, or 209 did not
significantly change sialoside binding (part B of Figure 4). In
contrast, the mutation to block the glycosylation on N27 appeared
to be detrimental for receptor binding activities.
The glycosylation on the amino-terminal N-glycosylation
sites, that is N27, had been implicated in protein transport and
folding. The loss of N20/21 of H2 was demonstrated to have a
decrease in hemadsorption and cell fusion activities.37 It was
also reported that HA without glycosylation on amino-terminal
(34) Inkster, M. D.; Hinshaw, V. S.; Schulze, I. T. J. Virol. 1993, 67, 7436–
7443.
(35) Gunther, I.; Glatthaar, B.; Doller, G.; Garten, W. Virus Res. 1993,
27, 147–160.
Effects of Site-Specific HA Glycosylation on Sialoside
Receptor Binding. It is believed that influenza HA needs to be
glycosylated to have proper function, that is to bind sialylated
(36) Chen, M. W.; Cheng, T. J.; Huang, Y.; Jan, J. T.; Ma, S. H.; Yu,
A. L.; Wong, C. H.; Ho, D. D. Proc. Natl. Acad. Sci. U.S.A. 2008,
105, 13538–43.
9
J. AM. CHEM. SOC. VOL. 132, NO. 42, 2010 14853

A R T I C L E S
Liao et al.
Figure 4. Binding profiles of HA glycosylation mutant. (A) Nine glycosylation sites were predicted using H5 as an example. The predicted glycosylation
sites in gray boxes were confirmed to be glycosylated using LCMS. The sites in white boxes were confirmed to be unglycosylated and the one in the hatched
box may be partially glycosylated. The glycosylation mutant were created by changing Nx(T/S) to NxA. (B) Sugar binding patterns of individual glycosylation
mutants.
sites would be trapped in endoplasmic reticulum and cannot be
transported to cell membrane.38 Our results showed that removal
of the N27 glycosylation site almost demolished the sugar
binding activity of WT HA. To further find out if the loss of
binding affinity is due to improper folding as other reports
suggested, circular dichroism (CD) spectroscopy analysis was
conducted. The results showed that the secondary structure of
∆27 changed, whereas another glycosylation mutant with similar
binding (∆170) had the same spectrum as the wild-type HA
(Figure 5). The results indicated that deletion of glycosylation
at position 27 had a major impact on HA structure, implying
that N-glycosylation on N27 is important for proper HA folding.
However, although the glycans found on the receptor binding
domain were shown to modulate the antigenic properties,
hemadsorption activities, and cell fusion activities,39 our array
studies showed no major change in the binding patterns for
glycosylation mutants. The binding affinities, however, de-
creased for glycosylation mutants (Table 2). Consistent with
previous results where soluble HAs were used for array
analysis,13,17 our results showed that high affinity toward sulfated
trisaccharides that contains LacNAc can be observed.
Figure 5. CD spectra showed that ∆170 exhibits similar secondary structure
with the wild-type HA, whereas the secondary structure of ∆27 changed.
Conclusions
Using the sialylated disaccharide strategy, 27 sialosides have
been synthesized to prepare a sialoside microarray on glass slides
which can be used to profile not only the receptor binding
specificity of various HA subtypes but also HAs with different
glycosylation patterns and real influenza viruses.
(37) Tsuchiya, E.; Sugawara, K.; Hongo, S.; Matsuzaki, Y.; Muraki, Y.;
Nakamura, K. J. Gen. Virol. 2002, 83, 3067–3074.
(38) Roberts, P. C.; Garten, W.; Klenk, H. D. J. Virol. 1993, 67, 3048–
3060.
The sugar binding results of recombinant HAs (Figure 2) and
influenza viruses (Figure 3) suggested that a minimum set of
(39) Tsuchiya, E.; Sugawara, K.; Hongo, S.; Matsuzaki, Y.; Muraki, Y.;
Li, Z. N.; Nakamura, K. J. Gen. Virol. 2002, 83, 1137–1146.
9
14854 J. AM. CHEM. SOC. VOL. 132, NO. 42, 2010

Sugar Binding Profiling of Hemagglutinin
A R T I C L E S
Table 2. Sialosides on the Array and the KD Values (nM) of H5
were kindly provided by the NIH Biodefense and Emerging
Infections Research Resources Repository, NIAID, NIH, Cy3-
labeled antimouse secondary antibodies, and Cy3-labeled strepta-
vidin were purchased from Jackson ImmunoResearch. Standard
chemicals and reagents were purchased from commercial suppliers
and used as received.
(CHA536) Glycosylation Mutants for Represented Sialoside
Construction of Hemagglutinin Expression Plasmids. The full-
length genes encoding HA from H1N1 Influenza A Virus including
pandemic California/07/2009, Brisbane/59/2007, New Caledonia/
20/1999 (ABF21272.1), H3N2 Brisbane/10/2007 (ABW23422.1),
H5N1 Vietnam/1194/2004 (ABP51976.1), H7N7 Netherland/219/
03 (AAR02640.1), or H9N2 HongKong/1073/99 (CAB95856.1)
fused with a C-terminal Strep (Trp-Ser-His-Pro-Gln-Phe-Glu-Lys)30
and (His)6 tag were cloned into pcDNA (Invitrogen) for expression
in human 293T cells. The sequences were confirmed by DNA
sequencing and prepared in high quality for expression. The genes
were cloned into pcDNA for expression in human 293T cells.
Construction of Hemagglutinin Glycosylation Mutants. The
nucleotide sequence of consensus hemagglutinin H5 (CHA537) was
cloned into pcDNA with a C-terminal Strep-Tag, and the resulting
plasmid was used as the templates for site-directed mutagenesis.
The N-glycosylation sites of CHA5 were predicted with The
NetNGly server provided by Expasy (www.expasy.org). The
putative N-X-S/T sequons were then mutated to N-X-A using site-
directed mutagenesis with QuikChange II Site-Directed Mutagenesis
Kit (Stratagene, La Jolla, CA, USA). The mutagenesis was
confirmed by DNA sequencing, and the glycosylation knockouts
were confirmed by SDS-PAGE analysis.
a Standard symbols are used. Blue circle, Glc; blue square, GlcNac;
red diamond, Neu5Ac; yellow circle, Gal; yellow square, GalNac; green
circle, Man; red triangle, Fuc, R ) (CH2)5NH2.
oligosaccharides containing 4, 5, 24, 29, 30 might be useful
for influenza subtyping. Our results showed that avian influenza
virus and recombinant HAs displayed no difference with their
glycan array profiling results, which preferentially bound R2,3
sialosides with relatively no binding was observed for R2,6
sialosides (parts E and F of Figure 2 and part D of 3), providing
a unique pattern for differentiating between avian and human
viruses. However, broader specificity toward both R2,3 and R2,6
sialosides was shown as a general theme of binding for real
human virus, comparing to major R2,6 binding of recombinant
human virus HAs. (parts A-C of Figures 2 and parts A-C of
3). Another interesting observation is that use of sialyl mono-,
di-, and trilactosamine glycans, which are 24, 29, 30 respectively
seems to be able to differentiate recombinant human virus HAs
such as pandemic H1, seasonal H1, and H3 (parts A-C of
Figure 3), wherein subtle binding of H3 to 29 comparing to
strong binding of both seasonal and pandemic H1 appeared to
be a good hint. Human influenza type A and B also displayed
different binding patterns in our real virus glycan array profiling
results. Influenza type A virus showed stronger binding to R2,6
sialosides than R2,3 sialosides, whereas type B virus showed
similar preference to both categories of sialosides. The strong
binding to R2,6 disaccharide 22 implicate high transmission
possibilities because seasonal H1 is uniquely binding to it
comparing to other H1 HAs.
It has been known that the glycans of influenza hemagglu-
tinins have some important biological functions.40 For example,
the glycans in the head region of hemagglutinins block the
antigenic site. Also, the HA receptor binding specificity was
affected by the absence of a complex glycan chain near the
receptor binding site.35 To address how glycosylation affects
the specificity of HA binding, we adopted the powerful glycan
microarray technology to profile HAs produced from different
expression systems which generate different N-glycosylation
patterns on the protein surface. Mutations of certain N-
glycosylation sequons on HA further suggested that certain
glycosylation sequons such as N27ST have detrimental effects
on HA function. The studies that use site-specific glycosylation
mutants to elucidate the relationship between HA glycosylation
and its receptor binding specificity underscore the importance
of HA glycosylation.
Expression of Recombinant Full-Length Hemagglutinin
from Expressed Cells. For expression in human 293T cells,
pcDNA carrying the gene of interest were prepared in high quality
and transfected with cation lipids DOTAP/DOPE (1:1) (Avanti
Lipids) for transient expression. The expression of hemagglutinin
was confirmed with immunoblots using anti(his)6 antibodies
(Qiagen) or specific anti-hemagglutinin antibodies and the horserad-
ish peroxidase-conjugated secondary antibodies (PerkinElmer).
Purification of Recombinant Hemagglutinins. The cells con-
taining recombinant full-length HA were lysed in 20 mM Hepes
buffer (pH 7.4) using a microfluidizer. The lysates were centrifuged
at 12,000 rpm for 10 min and the supernatant were collected for a
second centrifugation at 40 000 rpm for 1 h. Next, the pellets were
extracted with 20 mM Hepes buffer containing 1% dodecyl
maltoside for 2 h followed with a brief centrifugation at 12 000
rpm. The supernatant were then passed through an affinity column
packed with Nickel Sepharose Hi Performance (GE Healthcare).
After washes with 20 mM Hepes buffer containing 0.5% dodecyl
maltoside, the recombinant full-length HA was eluted with 500 mM
imidazole in 20 mM Hepes buffer. The purified proteins were
concentrated using Amicon (Millipore) and stored at 4 °C. The
concentrations were determined using immunoblots with anti-
hemagglutinin and horseradish peroxidase (HRP)-conjugated sec-
ondary antibodies.
Microarray Analysis of Sugar Binding Activities of Hemag-
glutinin. Microarrays were prepared by printing (AD3200, BioDot)
the glycan with an amide tail to the NHS-activated glass slide
(Nexterion H) by robotic pin (SMP2B, TeleChem International
Inc.). Nexterion H slides were spotted with solutions of sugar 1-17
and 21-30 at 100 µM from bottom to top with 12 replicates
horizontally in each grid and dried under vacuum. The spotted slides
were blocked with ethanolamine in sodium borate for 1 h just before
use followed with three washes of 0.05% Tween 20 in PBS buffer
(pH 7.4) (PBST). A solution of hemagglutinin at 50 µg/mL in PBST
was premixed with Cy3-labled streptavidin in 1:1 molar ratio for
1 h prior to incubation of the preformed complexes with the slides
for another hour. After six washes with PBST, one wash with PBS,
and three washes with distilled water, the slides were air-dried and
scanned with a 532 laser using a microarray fluorescence scanner
(GenePix 4000B, Molecular Devices). The PMT gain was set to
600. The resulting images were analyzed with GenePix Pro 6.0
Experimental Section
Materials. NHS-coated glass slides were obtained from SCHOTT
(Nexterion H), monoclonal antibodies to influenza hemagglutinin
(40) Vigerust, D. J.; Shepherd, V. L. Trends Microbiol. 2007, 15, 211–
218.
9
J. AM. CHEM. SOC. VOL. 132, NO. 42, 2010 14855

A R T I C L E S
Liao et al.
(Molecular Devices) to locate and quantify the fluorescence intensity
of all of the spots on the grid. The median of fluorescence intensity
of each spot was taken to calculate the median value of binding
activities toward each sugar (12 replicates for each sugar). The
medians from at least three independent experiments were averaged
for the figures. For KD determination, the preformed complexes
were serially diluted for binding reaction,31 and the binding
intensities were quantified at various concentrations of complexes
and fitted to the Langumir isotherms using the Prism (GraphPad,
San Diego, CA).41
Virus Preparation. Samples of various viruses are collected
from the Center for Disease Control and Prevention in Taiwan.
All viruses were propagated in 10-day-old embryonated specific-
pathogen free chicken eggs. Purified viruses were done by
centrifugation at 100 000 g for 30 min and resuspended in PBS.
As previously reported,16 differences in the degree of virus purity
did not influence significantly the pattern of receptor binding.
Viruses were inactivated by treatment with ꢀ-propiolactone (BPL;
0.05% v/v) for 60 min at 33 °C, and resuspended in 0.01 M
phosphate buffered saline pH 7.4 (PBS) and stored at -80 °C.
Comparison of samples of live and inactivated virus showed that
BPL inactivation did not alter receptor binding specificity.
Virus Binding Assay Procedure. Influenza virus A/Vietnam/
1194/2004 RG14 (H5N1), A/California/7/2009 (H1N1), A/Brisbane/
10/2007 (H1N1), and A/Brisbane/10/2007 (H3N2) were from
Taiwan CDC. Influenza virus B/Lee/40 was obtained from ATCC
(Manassas, VA, USA) and propagated. Printed slides were analyzed
without any further modification. Inactivated whole virus was
applied at a concentration around 107 virus/mL in PBS buffer
containing the neuraminidase inhibitor Oseltamivir carboxylate (10
µM). Suspensions of the inactivated viruses with Oseltamivir
carboxylate were overlaid onto the arrays and incubated at room
temperature for 1 h. Slides were subsequently washed by successive
rinses in PBS-0.05% Tween, PBS, and deionized water three times.
Bound viruses were detected using the following antibodies:
homemade rabbit anti-H1 antibody both for SOV California/07/
2009 and H1N1 Brisbane; anti-H3 antibody for H3N2 Brisbane
(NR3118, Biodefense and Emerging Infections Research Resources
Repository, National Institute of Allergy and Infectious Diseases,
MD, USA); anti-H5 (R-293s) (Sino Biological Inc. Beijing, CH)
antibody for H5N1 (RG14), and anti-flu B (Abcam, MA, USA)
for B/Lee/40. The slides were gently rocked at room temperature
for 60 min. After the repeating washing steps, binding was detected
by overlay with labeled secondary antibodies.
Acknowledgment. The authors thank Dr. Yih-Shyun E. Cheng
for influenza B/Lee/40 viruses, Mr. Yung-Fung Lin for his help
with NMR analysis of synthetic compounds, and Mrs. Jennifer Chu
for critically reading the manuscript. The authors also thank
Academia Sinica, Taiwan and National Science Council, Taiwan
for financial support (Grant # NSC 96-2321-B-001-025-MY2 to
C.-Y. Wu and C.-H Wong).
Supporting Information Available: Complete refs 2 and 12
experimental procedures and compound characterization data
(PDF). This material is available free of charge via the Internet
at http://pubs.acs.org.
(41) Liang, P. H.; Imamura, M.; Li, X.; Wu, D.; Fujio, M.; Guy, R. T.;
Wu, B. C.; Tsuji, M.; Wong, C. H. J. Am. Chem. Soc. 2008, 130,
12348–12354.
JA104657B
9
14856 J. AM. CHEM. SOC. VOL. 132, NO. 42, 2010