Content uploaded by Nan Zhao
Author content
All content in this area was uploaded by Nan Zhao on Nov 04, 2015
Content may be subject to copyright.
1
SCIENTIFIC RepoRts | 5:15778 | DOI: 10.1038/srep15778
www.nature.com/scientificreports
Association analyses of large-scale
glycan microarray data reveal
novel host-specic substructures in
inuenza A virus binding glycans
Nan Zhao1,2, Brigitte E. Martin1, Chun-Kai Yang1, Feng Luo3 & Xiu-Feng Wan1,2
Inuenza A viruses can infect a wide variety of animal species and, occasionally, humans. Infection
occurs through the binding formed by viral surface glycoprotein hemagglutinin and certain types
of glycan receptors on host cell membranes. Studies have shown that the α2,3-linked sialic acid
motif (SA2,3Gal) in avian, equine, and canine species; the α2,6-linked sialic acid motif (SA2,6Gal) in
humans; and SA2,3Gal and SA2,6Gal in swine are responsible for the corresponding host tropisms.
However, more detailed and rened substructures that determine host tropisms are still not clear.
Thus, in this study, we applied association mining on a set of glycan microarray data for 211
inuenza viruses from ve host groups: humans, swine, canine, migratory waterfowl, and terrestrial
birds. The results suggest that besides Neu5Acα2–6Galβ, human-origin viruses could bind glycans
with Neu5Acα2–8Neu5Acα2–8Neu5Ac and Neu5Gcα2–6Galβ1–4GlcNAc substructures; Galβ and
GlcNAcβ terminal substructures, without sialic acid branches, were associated with the binding
of human-, swine-, and avian-origin viruses; sulfated Neu5Acα2–3 substructures were associated
with the binding of human- and swine-origin viruses. Finally, through three-dimensional structure
characterization, we revealed that the role of glycan chain shapes is more important than that of
torsion angles or of overall structural similarities in virus host tropisms.
Inuenza A viruses infect a wide range of hosts, such as humans, sea mammals, swine, bats, and avian,
equine, and canine species1–4. e carbohydrates or glycans in host cells serve as the receptors for inu-
enza viruses and are key to successful virus entry, the rst step in inuenza infection5. e structures of
these glycan receptors have been shown to be unique in animal hosts and even within dierent tissues
in the same host, and these unique glycan structures determine host and tissue tropisms of inuenza A
viruses.
In humans, glycans with α 2,6-linked sialic acid (SA2,6Gal) are detected more plentifully in the upper
respiratory tract than the lower respiratory tract. SA2,6Gal and SA2,3Gal (α 2,3-linked sialic acid) are
heterogeneously distributed in the human nasopharynx and bronchi. e expression of SA2,3Gal is
greater than that of SA2,6Gal in the respiratory tract of young children6,7. In avian species, SA2,3Gal and
SA2,6Gal are distributed within respiratory and intestinal tracts. Although SA2,3Gal are mostly found
in waterfowl, it is possible that SA2,3Gal and SA2,6Gal could be expressed very dierently in terrestrial
birds8–11. Swine express both SA2,3Gal and SA2,6Gal in the respiratory tract, but SA2,6Gal are abundant
in the upper trachea and bronchi, and SA2,3Gal are more abundant in the lower respiratory tract10,12–14.
e presence of both SA2,3Gal and SA2,6Gal (e.g. Neu5Acα 2–3Gal and Neu5Acα 2–6Gal) in swine
1Department of Basic Sciences, College of Veterinary Medicine, Mississippi State University, MS, USA. 2Institute
for Genomics, Biocomputing & Biotechnology, Mississippi State University, MS, USA. 3School of Computing,
Clemson University, Clemson, SC, USA. Correspondence and requests for materials should be addressed to
X.-F.W. (email: wan@cvm.msstate.edu)
Received: 23 June 2015
Accepted: 29 September 2015
Published: 28 October 2015
OPEN
www.nature.com/scientificreports/
2
SCIENTIFIC RepoRts | 5:15778 | DOI: 10.1038/srep15778
could allow these animals to be susceptible to avian-origin and human-origin inuenza A viruses; thus
swine have been proposed as “mixing vessels” for inuenza viruses12.
By presenting multiple glycans or glycoconjugates printed on a single slide, the glycan microarray
technique has oered high-throughput analyses of the glycan-binding prole of inuenza viruses15,16.
Glycan microarray has become a routine experimental tool for characterizing the receptor-binding pro-
les of inuenza viruses17. To date, > 500 inuenza virus–related glycan microarray data entries have
been deposited in the Consortium for Functional Glycomics (CFG) glycan microarray database18, and
this number is still increasing. Glycan microarray proling of inuenza virus pandemic strains has
shed light on the receptor-binding specicities of their hemagglutinin (HA). For example, such analyses
revealed that the 2009 inuenza A(H1N1)pdm09 pandemic virus bound to α 2,6-linked and to a large
range of α 2,3-linked sialyl sequences19–22. Moreover, glycan microarray analysis has been widely used
to study receptor recognition and host tropism of inuenza virus mutants23–28. In addition to providing
data on Neu5Acα 2–3Gal and Neu5Acα 2–6Gal, glycan microarray analysis also provided data on other
complicated glycan substructures. In one study, structural topology (i.e., two glycan chain shapes, one
cone-like and the other umbrella-like) was reported to be related to SA2,3Gal and SA2,6Gal during inu-
enza virus–receptor interactions29. On the other hand, during a glycan microarray screening, inuenza A
viruses were shown to bind receptors other than SA2,3Gal and SA2,6Gal, although such bindings have
not been conrmed by interventional experiments22,30. For example, inuenza A (H1N1) virus can bind
α 2,8-linked polysialyl sequences22. Nevertheless, it is still unclear what specic substructures or moieties
in host receptors determine inuenza virus host tropisms.
To better understand structural specicities for glycan binding, Cholleti et al.31 developed an algo-
rithm called GlycanMotifMiner, or GLYMMR, that is frequently used with subtree mining to identify
motifs for protein–glycan interactions for a given glycan microarray data entry. Porter et al.32 applied a
clustering algorithm to identify glycan substructures with high intensities in the glycan array data. More
recently, we developed a novel quantitative structure–activity relationship (QSAR) method to analyze
glycan array data; the method focuses on glycan substructure features by applying PLS regression and
selection functions to the glycan microarray data33. Another frequent glycan structure mining of inu-
enza virus data also detected sulfated glycan motifs increased viral infection34. However, none of the
above methods were designed for large-scale glycan microarray data analysis that integrates multiple
microarray data entries for a particular research interest. Particularly, statistic-based motif identica-
tion methods rely on pre-dened hypothesis and could not discover unexpected and infrequent ones.
Feature selection strategies for the regressions of glycan microarray data have not considered modeling
multiple microarrays. us, a computational method is needed to characterize glycan substructure motifs
by utilizing the information across multiple datasets, especially glycan microarray data across various
platforms, and this method must be able to tolerate the noises within and across glycan microarray data.
e relationship between host receptors (glycan substructures) and inuenza A viruses (e.g., viruses
with dierent host origins) can be naturally formulated as a computational problem of data integration
plus association rule mining. erefore, in this study, we rstly applied a PLA regression on individual
glycan microarray data entries as normalization and then used association rule analysis on extracted
glycan substructure features to identify motifs for inuenza virus host tropisms. In addition to SA2,3Gal
and SA2,6Gal, results showed that glycan substructures with SA2,8SA, non-sialic acid saccharides (Galβ
and GlcNAcβ terminal substructures), and sulfated SA2,3Gal could contribute to inuenza host tro-
pism dierently. Additional computational modeling demonstrated that, for trisaccharide substructures,
a shape angle formed by mass centers of three residues, instead of linkage torsion angles, may determine
the overall glycan chain shapes and thus distinguish glycans with SA2,3Gal from those with SA2,6Gal
or SA2,8SA. ese ndings may imply a more general property caused by glycan terminals than just by
sialic acid with dierent linkages during inuenza – glycan binding.
Methods
Figure1 shows a simplied owchart of the computational strategy we used, with glycan microarray data,
to identify host-specic glycan substructures. In brief, we collected and integrated glycan microarray
datasets (Fig.1A), dened and extracted substructures from glycans (Fig.1B), and applied association
rule mining to identify the inuenza viruses’ specic glycans and their substructures (Fig.1C).
Datasets. Collection of inuenza A virus–specic glycan microarray data. A computational script was
written to automatically retrieve glycan microarray datasets from CFG18 by using the keyword “inu-
enza.” A total of 542 glycan microarray entries were retrieved, of which 324 were excluded from the nal
dataset: 31 entries for inuenza B viruses, 182 for mutant viruses, 51 for mouse-adapted strains, 53 for
HA recombinant proteins, and 7 for microarrays with incomplete binding anity values. e remaining
218 entries were for inuenza A virus–specic glycan microarray datasets with complete binding anity
values. ese datasets, which were used for further analyses (Table1), consisted of inuenza A viruses of
human origin (n = 154), waterfowl origin (n = 17), terrestrial bird origin (n = 13), canine origin (n = 6),
and swine origin (n = 21). e metadata associated with these datasets, including CFG entry identica-
tion codes, investigators’ names, inuenza virus sample names, glycan array version, raw array binding
les, and host species, are listed in Table S1 in the supporting information.
www.nature.com/scientificreports/
3
SCIENTIFIC RepoRts | 5:15778 | DOI: 10.1038/srep15778
Integration of glycan microarray datasets. In the CFG database, the datasets were generated by using
11 versions of glycan microarrays, each of which had a dierent number of glycan entries. For example,
version 1 had 200 glycans, whereas version 5.1 had 611 glycans. However, most glycans present in earlier
glycan microarray version are present in later versions. To facilitate the data analyses across dierent
datasets, we merged all microarray versions into one with 936 unique glycans (Table S2) and generated
a single matrix for the 211 data entries (211 viruses × 936 glycans, Table S3). Glycan-binding anities
(i.e., uorescent signal values) in our dataset were assigned to corresponding elements in the matrix. e
elements for which there was no corresponding anity value among the 936 glycans were assigned a “not
available” value and excluded from the glycan substructure feature selection.
Figure 1. Flowchart of the computational analysis protocol. (A) Glycan microarray data collection for
inuenza virus bindings. (B) Glycan structure organizations and substructures’ feature extraction and
selection. (C) Association rule mining on viral host labeled substructure feature vectors. CFG, Consortium
for Functional Glycomics; HA, hemagglutinin; QSAR, quantitative structure–activity relationship.
www.nature.com/scientificreports/
4
SCIENTIFIC RepoRts | 5:15778 | DOI: 10.1038/srep15778
Glycan substructure feature extraction. Glycan substructures were dened as described else-
where33. Specically, mono-, di-, tri-, and tetrasaccharide substructures were extracted from 936 glycans
as features. ese extractions resulted in 249 monosaccharide, 738 disaccharide, 1,198 trisaccharide, and
1,477 tetrasaccharide substructures (Tables S4–S7). e uorescent signal value for the corresponding
glycan on the array was assigned as the binding anity for each individual substructure. Only uores-
cent signal values ≥ 2,000 were considered as eective numbers in regression, and those < 2,000 were
treated as background noise. Next, a partial least squares (PLS) regression and feature selection algorithm
(QSAR33) were adapted to select the features predominating glycan binding from an inuenza virus–
specic glycan microarray dataset (see details in Supplementary Information). is PLS regression was
performed four times for each single data entry from our 211 glycan microarrays by using four sets of
substructure feature denitions (mono-, di-, tri-, and tetrasaccharides). Each feature vector was nally
labeled according to the host origin of the inuenza A viruses used in the glycan microarray experiments
(i.e., human, swine, canine, waterfowl, or terrestrial bird [chicken, quail, and turkey] host).
Association rule mining for selected glycan substructures. We formulated the detection of
host-specic glycan substructures as an association-mining task (see more details in Supplementary
Information), where we let items
=,,,Iiii{}
n12
represent a set of items and let
=,,,Ttt t{}
m12
be a set of transactions forming a database. An association rule,
⇒XY
, where
,⊆XY I
, is usually
interpreted to mean that when the items in
X
exist, those in
Y
also occur at a certain condence level35.
Here, for our glycan microarray dataset, transactions T were the data derived from inuenza virus–spe-
cic glycan microarray entries, so m = 211; the substructure features X derived from glycans on the array
by previous PLS-β selection and the labeled features Y with host origin will form I. Given a rule
⇒XY
,
the condence is dened as
∪
(⇒)= ()/()Conf XY supp XYsupp X
, where sup (X) is the support of
item set
X
. e support was dened as the proportion of transactions in the dataset, which contains the
item set. Another measurement, li, is the ratio of the observed support and was dened as
∪
(⇒)= ()/( ()×())Lift XY supp XYsupp XsuppY
35. erefore, we expected to obtain interesting
association rules with high condences (≥ 80%), high lis (has a li value ≥ 136), and low supports
(≥ 0.005, infrequent but potentially interesting) to supply highly probable, unexpected, and infrequent
conclusions. We adapted the Apriori algorithm implemented in R37 to infer these host substructure–spe-
cic associations. Moreover, during the mining process, redundant rules were also removed by dening
super rules as redundancy. A super rule is a rule with the same or lower li value, where the le hand
side, X, contains more items than a previous rule, but still results in the same right hand side, Y. Last,
we kept only satised rules, which were ltered by leaving only those with terminal saccharides on the
substructure features.
Three-dimensional structural modeling and analysis. Structural characterization for terminal gly-
can saccharides. To understand the structural determinants for a specic glycan associated with certain
inuenza A virus, we compared the spatial relationship between six terminal trisaccharide features derived
from data mining. ese six features were (Neu5Acα 2–6)-(6Galβ 1–4)-4GlcNAc (PDB38 accession number:
Array
versions No. Inuenza
virus arrays
Host species
Human Swine Canine
Avian
Wat erfo w l Terrestrial
V1.0 7 5 0 0 2 0
V2.0 11 10 0 0 1 0
V2.1 4 4 0 0 0 0
V3.0 10 10 0 0 0 0
V3.1 21 2 1 0 8 10
V3.2 5 4 0 0 0 1
V4.0 33 31 2 0 0 0
V4.1 28 15 13 0 0 0
V4.2 20 9 3 0 6 2
V5.0 68 64 0 4 0 0
V5.1 4 0 2 2 0 0
Tot a l 211 154 21 6 17 13
Table 1. Glycan-binding microarray data collected for 211 wild-type inuenza A virus–specic glycan
microarray datasets with complete binding anity values.
www.nature.com/scientificreports/
5
SCIENTIFIC RepoRts | 5:15778 | DOI: 10.1038/srep15778
3UBN39), (Neu5Acα 2–8)–(8Neu5Acα 2–8)–8Neu5Ac (3HMY40), (Neu5Acα 2–3)–(3Galβ 1–4)-4GlcNAc
(3UBQ39), (Neu5Gcα 2–3)–(3Galβ 1–4)–4GlcNAc (4POT41), (Galβ 1–4)–(4GlcNAcβ 1–3)–3Gal (2XRS42),
and (Galβ 1–4)(Fucβ 1–3)–(3,4GlcNAcβ 1–3)–3Gal (1SL543). e following three geometric measure-
ments were calculated:
(1) e angle formed by the mass centers of three saccharides. We calculated the angle formed by the mass
centers of three saccharides as a measurement of the glycan chain’s turning shape.
(2) e root-mean-square deviation (RMSD). Given two glycan substructures, each containing the termi-
nal saccharide, we superimposed the corresponding atoms on the six-membered rings of the terminal
saccharides. From there, while keeping the terminal saccharides superimposed, the following two
values of RMSD were measured: RMSD2 and RMSD3. Using the standard formula of calculating
RMSD from two sets of six-membered ring atoms
v
and
w
:
()
(, )= ∑(− )+(− )+(− )
=
RM
SD vw vw vw
vw
n
i
nix ix iy iy iy iy
1
1
222
, where n = 6, RMSD2 was
calculated for the two saccharides linked directly to their respective terminal saccharides. If both
glycan substructures had a third saccharide, RMSD3 was then calculated for the third pair of
saccharides.
(3) φ and ψ torsion angles. We calculated the φ and ψ torsion angles for each linkage between two adja-
cent residues. Glycosidic torsions were dened by only heavy linkage atoms as
φ
= O5–C1–On–Cn
and ψ = C1–On–Cn–Cn-144. Accordingly, a trisaccharide substructure has two linkages with two sets
of torsions.
ree-dimensional structures for protein–glycan interactions. To demonstrate structural interactions
between inuenza viruses and glycan substructures at the molecular level, we used four HA protein
crystal structures. ese structures were from viruses with dierent host origins (A/California/04/09
H1N1, human origin, SA2,6Gal specic; A/swine/Iowa/15/1930 H1N1, swine origin, SA2,6Gal spe-
cic; A/Canine/Colorado/06 H3N8, canine origin, SA2,3Gal specic; A/Vietnam/1203/2004 H5N1,
avian origin, SA2,3Gal specic) and were obtained from PDB38 data entries, 3LZG45, 1RVT46, 4UO547,
and 2FK020, respectively. We docked these HA proteins with corresponding glycan substructures
(6SLN [α 2–6-sialyl-N-acetyllactosamine], analogous to Neu5Acα 2–6Galβ 1–4GlcNAc, and 3SLN [α
2–3-sialyl-N-acetyllactosamine], analogous to Neu5Acα 2–3Galβ 1–4GlcNAc) and highlighted con-
served and diverse regions on the receptor-binding pocket (amino acid residues 98, 133–138, 153, 183,
188–195, 221–228 by H3 numbering). Substructures, (Neu5Acα 2–8)–(8Neu5Acα 2–8)–8Neu5Ac and
(Galβ 1–4)–(4GlcNAcβ 1–3)–3Gal, were also simulated against bound 6SLN and 3SLN to be docked at the
receptor-binding pocket of the human-origin and the avian-origin viral HAs respectively. e HA-glycan
docking was conducted by following three steps: 1) an initial structure was obtained by superposing the
structures of HA and a glycan analog against a native HA 3D structure with glycan; 2) an energy min-
imization with 500 steps of conjugation and 500 additional steps of steepest descent was performed by
using the AMBER force eld48 at a GROMACS49 dynamic simulation process; and 3) the nal complex
structure was obtained aer an binding free energy repairing by using the FoldX soware50 at simulation
temperature of 298K and without hydrogen atoms.
Results
Inuenza virus–specic features derived from glycan microarray data by PLS regression and
feature selection. Certain saccharide residues are enriched at glycan substructures contributing to
inuenza virus binding. In the integrated dataset of glycan microarrays with 936 unique glycans, the
glycans with inuenza virus–binding anities mostly consist of sialic acids (Neu5Ac and Neu5Gc),
galactose (Gal), N-acetylgalactosamine (GalNAc), glucose (Glc), N-acetylglucosamine (GlcNAc), fucose
(Fuc), and mannose (Man). Table2 summarizes the number and percentage of glycan substructures that
have these saccharides by each one of the four substructure feature denitions on the glycan microarrays
(i.e., mono-, di-, tri-, and tetrasaccharides) and thus reects their existence according to the microarray
design. Table 3 lists the same distribution values of inuenza virus–specic substructures selected by
PLS regression and illustrates that only a small portion of glycan substructures (73/249 monosaccha-
rides, 230/738 disaccharides, 322/1,198 trisaccharides, and 320/1,477 tetrasaccharides) was determined
to contribute to a binding signal of ≥ 2,000 with inuenza viruses. All PLS-selected substructure features
are also summarized in Table S8.
A comparison of data in Table2 with that in Table3 shows that Neu5Ac, Neu5Gc, Gal, and GlcNAc
were more abundant in the glycan substructures contributing to inuenza virus binding. For example,
Neu5Ac appeared in 9.59% of the monosaccharides, 11.3% of the disaccharides, 18.0% of the trisaccha-
rides, and 31.9% of the tetrasaccharides when QSAR was used to select signicant glycan substructures
for inuenza virus binding (Table3), compared with 4.82%, 6.10%, 7.68%, and 10.9%, respectively, of all
the glycan substructures from microarrays (Table2). Similar dierences were observed for Neu5Gc, Gal,
and GlcNAc. ese ndings suggest that inuenza virus–specic glycan substructures are prone to have
these four saccharides. Nevertheless, when QSAR was used, glycan substructures with GalNAc, Glc, Fuc,
and Man were equally or less frequently correlated with inuenza virus binding than those on the glycan
www.nature.com/scientificreports/
6
SCIENTIFIC RepoRts | 5:15778 | DOI: 10.1038/srep15778
arrays (Tables2 and 3), which indicates a limited contribution to inuenza binding by substructures with
these saccharides.
Host-specic glycan substructures derived from association rule mining. To understand the
specic substructures associated with each inuenza A virus, we performed associate rule analyses across
211 inuenza virus–specic glycan microarray data. On the basis of their host origins, the 211 inuenza
A viruses were categorized into human (n = 154), canine (n = 6), swine (n = 21), waterfowl (n = 17),
terrestrial (i.e., chicken, quail, and turkeys, n= 13), and avian (waterfowl plus terrestrial birds, n = 30).
e association analysis results (summarized in Fig.2 and Table S9) illustrate the specic substructures
being associated with each of six host origins; these associations aid in our understanding of the key
substructures that determine inuenza host and tissue tropisms.
Neu5Acα2–8Neu5Acα2–8Neu5Ac and Neu5Gcα2–6Galβ1–4GlcNAc substructures contribute to the glycan
binding with human-origin inuenza A viruses. In addition to the reported α 2,6-linked sialic acid glycan
substructures (with a Neu5Acα 2–6 terminal), which were detected multiple times (28 rules in Table S9)
to be associated with human-origin inuenza A viruses (Fig.2), Neu5Acα 2–8Neu5Acα 2–8Neu5Ac (fre-
quency = 0.0299, condence = 1.00, li = 1.44) and Neu5Gcα 2–6Galβ 1–4GlcNAc (frequency = 0.0479,
condence = 1.00, li = 1.44) substructures were also found to be associated with human-origin inuenza
A viruses’ glycan binding (Fig.2 and Table S9). In Fig.3A,B, as case studies, two glycan microarray data
entries of human-origin viruses demonstrate the signicantly high binding anities to these substruc-
tures separately. Specically, virus A/OK/5342/2010 binds to glycan Neu5Acα 2–8Neu5Acα 2–8Neu5Ac
and A/Memphis/4/73 to glycan Neu5Gcα 2–6Galβ 1–4GlcNAc. Although Neu5Gc has not been reported
to be present in human respiratory track tissues, human-origin viruses may have the binding ability to
Residue
No. (%) glycan substructures
Monosaccharide
N = 249 Disaccharide
N = 738 Trisaccharide
N = 1,198 Tetrasaccharide
N = 1,477
Neu5Ac 12 (4.82) 45 (6.10) 92 (7.68) 162 (10.9)
Neu5Gc 4 (1.61) 10 (1.36) 16 (1.34) 18 (1.22)
Gal 44 (17.7) 333 (45.1) 811 (67.7) 1,064 (72.0)
GalNAc 35 (14.1) 166 (22.5) 329 (27.5) 400 (27.1)
Glc 31 (12.4) 109 (14.8) 175 (14.6) 201 (13.6)
GlcNAc 45 (18.1) 364 (49.3) 816 (68.1) 1,206 (81.7)
Fuc 6 (2.41) 43 (5.83) 167 (13.9) 310 (20.9)
Man 28 (11.2) 89 (12.1) 235 (19.6) 558 (37.8)
Table 2. Distribution of saccharide residues in the glycan substructures from all glycans on
the microarrays. Abbreviations: Fuc, fucose; Gal, galactose; GalNAc, N-acetylgalactosamine; Glc,
glucose; GlcNAc, N-acetylglucosamine; Man, mannose; Neu5Ac, N-acetylneuraminic acid; Neu5Gc, N-
glycolylneuraminic acid.
Residue
No. (%) glycan substructures
Monosaccharide
N = 73 (out of 249)
Disaccharide
N = 230 (out of
738)
Trisaccharide
N = 322 (out of
1,198)
Tetrasaccharide
N = 320 (out of
1,477)
Neu5Ac 7 (9.59) 26 (11.3) 58 (18.0) 102 (31.9)
Neu5Gc 1 (1.37) 3 (1.30) 5 (1.55) 6 (1.88)
Gal 13 (17.8) 107 (46.5) 241 (74.8) 264 (82.5)
GalNAc 10 (13.7) 54 (23.5) 84 (26.1) 76 (23.8)
Glc 4 (5.48) 13 (5.65) 26 (8.07) 32 (10.0)
GlcNAc 15 (20.5) 146 (63.5) 242 (75.2) 268 (83.8)
Fuc 0 (0.00) 7 (3.04) 24 (7.45) 41 (12.8)
Man 12 (16.4) 27 (11.7) 38 (11.8) 79 (24.7)
Table 3. Distribution of saccharide residues in the glycan substructures selected by using the
quantitative structure–activity relationship. Abbreviations: Fuc, fucose; Gal, galactose; GalNAc,
N-acetylgalactosamine; Glc, glucose; GlcNAc, N-acetylglucosamine; Man, mannose; Neu5Ac, N-
acetylneuraminic acid; Neu5Gc, N-glycolylneuraminic acid.
www.nature.com/scientificreports/
7
SCIENTIFIC RepoRts | 5:15778 | DOI: 10.1038/srep15778
α 2,6-linked Neu5Gc substructures on glycan microarrays. is observation also suggests that, in addi-
tion to Neu5Acα 2–6 terminal, other sialic acids with either α 2–6 or α 2–8 linkages may be recognized
by human-origin inuenza A viruses.
Galβ and GlcNAcβ terminal substructures, in addition to Sialic Acids, associated with the glycan bind-
ing of human-, swine-, and avian-origin inuenza A viruses. e α 2,3-linked and α 2,6-linked sialic
acid glycan substructures were identied as predominated glycan binding motifs of all types of inu-
enza A viruses (Fig.2). However, it was interesting to observe that glycan substructures with Galβ and
GlcNAcβ terminals were detected to be associated with human-, swine-, and avian-origin viruses. ese
two terminal saccharides are usually followed by β 1,3-, β 1,4-linked, and occasionally α 1,3-, α 1,3-linked
Gal, GlcNAc, or GalNAc (e.g., Galβ 1–4(Fucα 1–3)GlcNAcβ 1–3Gal with human-origin virus binding:
frequency = 0.0359, condence = 0.857, li = 1.23; GlcNAcα 1–4Galβ 1–4GlcNAc with swine-origin virus
binding: frequency = 0.012, condence = 1.00, li = 9.28; Galβ 1–4GlcNAcβ 1–3Gal with avian-origin
virus binding: frequency = 0.0419, condence = 1.00, li = 5.96) (Fig. 2 and Table S9). Moreover, these
Figure 2. Host-specic glycan substructures detected by association rule mining. Human-associated,
mammal (swine and canine)-associated, and avian (waterfowl and terrestrial bird)-associated terminal
substructures.
www.nature.com/scientificreports/
8
SCIENTIFIC RepoRts | 5:15778 | DOI: 10.1038/srep15778
Figure 3. Case studies: individual glycan microarrays of inuenza A viruses with interesting binding
motifs. (A) Human-origin virus A/OK/5342/2010 binds glycan Neu5Acα 2–8Neu5Acα 2–8Neu5Ac.
(B) Human-origin virus A/Memphis/4/73 binds glycan Neu5Gcα 2–6Galβ 1–4GlcNAc. (C) Human-origin
virus A/Oklahoma 309/06 shows binding ability to glycan with Galβ 1–4(Fucα 1–3)GlcNAcβ 1–3Gal
substructure. (D) Swine-origin virus A/swine/New Jersey/1/76 shows binding ability to glycans with
GlcNAcα 1–4Galβ 1–4GlcNAc substructure. (E) Waterfowl-origin virus A/Duck/HK/149/1977 shows binding
ability to glycans with Galβ 1–4GlcNAcβ 1–3Gal substructure. (F) Human-origin virus A/Aichi/2/68 binds to
sulfated glycan with α 2,3-linked sialic acid terminals. (G) Swine-origin virus A/Swine/Iowa/1876 binds to
sulfated glycan with α 2,3-linked sialic acid terminals.
www.nature.com/scientificreports/
9
SCIENTIFIC RepoRts | 5:15778 | DOI: 10.1038/srep15778
Galβ and GlcNAcβ terminal substructures could result in inuenza viruses binding independently, since
many glycans with either Galβ or GlcNAcβ terminals, out of the 936 unique glycans on microarrays, do
not contain any sialic acid saccharides (Table S2). And there are individual microarray data entries show-
ing inuenza viruses bind to Galβ or GlcNAcβ terminal glycans with no other branches at the same time.
For instance, in Fig.3C,E, human-origin virus A/Oklahoma 309/06, swine-origin virus A/Swine/New
Jersey/1/76, and waterfowl-origin virus A/Duck/HK/149/1977 all have relatively high binding signals on
their individual glycan microarrays. We then conclude that, without existing sialic acid saccharide resi-
dues, glycans having Galβ or GlcNAcβ terminal could serve as potential receptors for inuenza A virus.
Sulfation causes Neu5Acα2–3 substructures associated with the glycan binding of human- and swine-origin
inuenza A viruses. As with human-origin inuenza A viruses, not surprisingly, multiple substructures
(19 rules in Table S9) with Neu5Acα 2–6 were identied as being associated with the glycan binding of
swine-origin inuenza A viruses. In addition, the substructures with Neu5Acα 2–3 were also associated
with swine-origin inuenza A viruses (Fig.2). However, interestingly, the substructures with Neu5Acα 2–3
terminals are usually sulfated on the following saccharide residues, when they were identied to be
associated with either human- or swine-origin inuenza A viruses. For example, Neu5Acα 2–3(6OSO3)
Galβ 1–4GlcNAc was associated with human-origin (frequency = 0.0719, condence = 0.80, li = 1.51)
and swine-origin (frequency = 0.0179, condence = 1.00, li = 9.28) viruses separately (Fig. 2 and
Table S9). As shown in Fig. 3F,G, a human-origin virus A/Aichi/2/68 and a swine-origin A/Swine/
Iowa/1976 both have their highest binding anity to sulfated glycans with α 2,3-linked sialic acid termi-
nals (Neu5Acα 2–3Galβ 1–4(6OSO3)GlcNAc, and Neu5Acα 2–3(6OSO3)Galβ 1–4GlcNAc). is associa-
tion rule linking sulfated Neu5Acα 2–3 glycans and human-, swine-origin virus binding may support a
unique role of sulfation during human and swine adaptation of avian-origin inuenza A viruses.
Consensus among the inuenza virus–specic glycan substructures. To identify common fea-
tures from the substructures associated with dierent hosts, we compared the structural similarity among
them by calculating, as described in Methods, the angle formed by three mass centers of all residues for
trisaccharide substructures, RMSD2 and RMSD3, and φ and ψ torsion angles of linkages for the six rep-
resentative glycan substructures (Fig.4). In addition, superposition images of these glycan substructures
are shown in Fig.5.
3D structural characterization for glycan substructures with sialic acid terminals. As shown in Fig.4A,D,
we obtained four trisaccharide three-dimensional structures, of which one has SA2,6Gal terminal, one
has SA2,8SA terminal, and two have SA2,3Gal terminals. ree observations were made from the sub-
structures. First, the residue mass centers for the SA2,6Gal substructure formed acute angles (63.1°), for
the SA2,8SA substructure formed an angle of 91.1°, and the mass centers for both α 2,3-linked substruc-
tures formed obtuse angles (142.9° and 132.2°). is observation suggests that SA2,6Gal and SA2,3Gal
substructures are fundamentally dierent from each other on saccharide chain shapes and thus could
lead to virus host tropism, in which human inuenza viruses recognize glycans with SA2,6Gal termi-
nals, canine and avian viruses recognize glycans with SA2,3Gal terminals specically, but swine viruses
can recognize and bind to both shapes. Moreover, the α 2,8-linked polysialyl substructure with a right
angle shares a more similar turning shape to the one of SA2,6Gal and then may cause the human-origin
inuenza virus binding.
Second, the all-against-all RMSD values for these glycan substructures indicate that none of the sub-
structures with sialic acid terminals are similar on the basis of both RMSD2 and RMSD3 values, if we
dene similar saccharide structures by using RMSD2 smaller than 3 Å and RMSD3 smaller than 5 Å
(Table4, Fig.5). is nding shows that shape angles formed by residue mass centers are not the sole
factor for glycan structural diversity.
e third observation involves the linkage torsion angles of these four representative host-specic
trisaccharides (Table5). It is shown that, although most torsions of linkage 2 share similar values and
hence do not contribute much to virus host types, both φ and ψ angles of linkage 1 distribute vari-
ously and indicate the shape-forming roles of α 2,6, α 2,8 and α 2,3 linkages with terminal sialic acids.
In particular, the linkage 1 φ angle values of these four trisaccharides, combined with their shape angles
formed by residue mass centers, could shed some light on the relationship between glycan geometric
shapes and inuenza virus host types. On one hand, when trisaccharides with SA2,6Gal or SA2,8SA
have angles of acute shapes (Fig.4A,B), a positive φ angle (e.g., 71.32° of Neu5Acα 2–6Gal or 55.05° of
Neu5Acα 2–8Neu5Ac in Table 5) might be necessary to make the glycan associated with human host
type; however, the association might not be unique because the positive φ angle might also result in an
association with swine viruses. On the other hand, when trisaccharides with SA2,3Gal have angles of
obtuse shapes (Fig.4C,D), dierent terminal residues (i.e., Neu5Ac and Neu5Gc) form φ angles with
dierent values (e.g., a − 59.47° of Neu5Acα 2–3Gal and a 50.95° of Neu5Gcα 2–3Gal in Table5). is
observation illustrates that an obtuse angle of α 2,3-linked trisaccharides is sucient, but not necessary,
for a glycan to associate with non-human-origin viruses and that a positive φ torsion angle at linkage
1 may make the trisaccharides associated with canine- and avian-origin viruses only. Furthermore, all
four trisaccharides, except Neu5Acα 2–8Neu5Acα 2–8Neu5Ac, have linkage 1 ψ angles of similar negative
values and therefore do not show a clear relationship with virus host types.
www.nature.com/scientificreports/
10
SCIENTIFIC RepoRts | 5:15778 | DOI: 10.1038/srep15778
In summary, the structural characteristics of glycan trisaccharides with sialic acid terminals might
be associated with inuenza virus host tropisms. For example, it seems that the shape angle formed by
residue mass centers plus the linkage 1 φ torsion angle, not just torsion angles themselves, might suggest
certain glycan structural patterns associated with inuenza virus host tropism.
3D structural characterization for glycan substructures with a Gal terminal. For glycan trisaccharide
substructures with a Gal, we found two additional representative three-dimensional structures of glycans
Figure 4. ree-dimensional structures of host-specic glycan substructures. (A–D) Representative
structures of host-specic trisaccharide substructures with sialic acid terminals. e shape angles were
calculated by the mass centers of three residues. (E–F) Structures of trisaccharide substructures with Gal
(non-sialic acid) terminal saccharide.
Figure 5. Superposition of terminal saccharide residues and the root-mean-square deviations (RMSDs)
between the second (RMSD2) and the third (RMSD3) residues.
www.nature.com/scientificreports/
11
SCIENTIFIC RepoRts | 5:15778 | DOI: 10.1038/srep15778
that had the same terminal residues and linkages associated with viruses of dierent host-origins:
Galβ 1–4GlcNAcβ 1–3Gal and the fucosylated one Galβ 1–4(Fucα 1–3)GlcNAcβ 1–3Gal (Fig. 4E,F). e
Galβ 1–4GlcNAcβ 1–3Gal backbone of both substructures formed similar obtuse angles (156.9° and
146.1°), while the additional fucose residue formed a 54.9° angle with the Galβ 1–4GlcNAcβ 1 terminal.
is additional turning angle introduced by fucosylation may lead to the human-origin virus bind-
ing (Fig. 4F). Next, we measured RMSDs between the Galβ 1–4GlcNAcβ 1–3Gal substructure and the
ones with sialic acid terminals (Table 4 and Fig. 5C,D). Although Galβ 1–4GlcNAcβ 1–3Gal has simi-
lar RMSD2 values with Neu5Acα 2–6Galβ 1–4GlcNAc (2.78 Å) and with Neu5Acα 2–3Galβ 1–4GlcNAc
(2.99 Å), a smaller RMSD3 value (6,44 Å) with Neu5Acα 2–3Galβ 1–4GlcNAc showed its better struc-
tural similarity to the SA2,3Gal motif. us, this nding might suggest that the glycan substructure
Galβ 1–4GlcNAcβ 1–3Gal does not have to maintain a sialic acid terminal to share a structural similarity
with the avian–virus-binding motif (SA2,3Gal). Last, nevertheless, torsion angles of both linkages of
Galβ 1–4GlcNAcβ 1–3Gal have unique values comparing to those of sialic acid substructures and there-
fore do not support any relationships with various virus bindings.
Structural conservation of receptor binding pocket in inuenza A viruses. In Fig. 6A,B,
we show superposed HA receptor binding pockets of dierent inuenza viruses (human-origin with
swine-origin viruses and canine-origin with avian-origin viruses) interacting with 6SLN and 3SLN (anal-
ogous to glycan substructures). Human- and swine-origin HAs recognize glycans with α 2,6-linked sialic
acid terminals, and they share a very conserved receptor binding pocket, which diers by only four
amino acid residues (133A, 225, 227, and 189) for the dierent host type viruses (Fig. 6A). Similarly,
canine- and avian-origin HAs recognize glycans with α 2,3-linked sialic acid terminals, and they also
have a conserved receptor binding pocket, but with more diverse residues (133A is deleted on canine
HA, and residues dier at amino acids 135, 137, 221, 222, 224, 188, and 189) (Fig.6B).
In Fig.6C,D, we docked Neu5Acα 2–8Neu5Ac and Galβ 1–4GlcNAcβ 1–3Gal to the receptor binding
pocket of the human-origin HA (PDB 3LZG) and the avian-origin HA (PDB 2FK0) separately by using
a HA-glycan structural complex as the template (see Methods). Previous association results suggested
Glycan substructure names RMSD 2
value, Å RMSD 3
value, ÅSubstructure 1 Substructure 2
Neu5Acα 2-6Galβ 1-4GlcNAc Neu5Acα 2-8 Neu5Acα 2-8Neu5Ac 3.13 5.58
Neu5Acα 2-6Galβ 1-4GlcNAc Neu5Acα 2-3Galβ 1-4GlcNAc 3.32 7.15
Neu5Acα 2-6Galβ 1-4GlcNAc Neu5Gcα 2-3Galβ 1-4GlcNAc 2.86 7.26
Neu5Acα 2-6Galβ 1-4GlcNAc Galβ 1-4GlcNAcβ 1-3Gal 2.78 8.97
Neu5Acα 2-8 Neu5Acα 2-8Neu5Ac Neu5Acα 2-3Galβ 1-4GlcNAc 5.08 8.90
Neu5Acα 2-8 Neu5Acα 2-8Neu5Ac Neu5Gcα 2-3Galβ 1-4GlcNAc 3.15 4.86
Neu5Acα 2-8 Neu5Acα 2-8Neu5Ac Galβ 1-4GlcNAcβ 1-3Gal 3.19 7.83
Neu5Acα 2-3Galβ 1-4GlcNAc Neu5Gcα 2-3Galβ 1-4GlcNAc 3.29 8.01
Neu5Acα 2-3Galβ 1-4GlcNAc Galβ 1-4GlcNAcβ 1-3Gal 2.99 6.44
Neu5Gcα 2-3Galβ 1-4GlcNAc Galβ 1-4GlcNAcβ 1-3Gal 1.94 4.38
Table 4. All-against-all RMSD values of representative inuenza A host-specic glycan trisaccharide
substructures. Abbreviations: RMSD, root-mean-square deviation; RMSD2, the RMSD between the two six-
membered rings of the saccharides linked to the terminal saccharide; RMSD3, the RMSD between them of
the third pair of saccharides.
Glycan substructure name
Linkage 1 Linkage 2
φψφψ
Neu5Acα 2-6Galβ 1-4GlcNAc 71.32 − 151.25 − 94.84 82.75
Neu5Acα 2–8 Neu5Acα 2–8Neu5Ac 55.05 112.07 57.35 121.40
Neu5Acα 2–3Galβ 1–4GlcNAc − 59.47 − 126.54 − 81.61 124.23
Neu5Gcα 2–3Galβ 1–4GlcNAc 50.95 − 134.56 − 67.73 105.46
Galβ 1–4GlcNAcβ 1–3Gal − 96.61 95.29 − 46.87 − 139.85
Table 5. Torsion angles (φ and ψ) of linkage 1 and linkage 2 of representative inuenza A host-specic
glycan substructures.
www.nature.com/scientificreports/
12
SCIENTIFIC RepoRts | 5:15778 | DOI: 10.1038/srep15778
a relationship of Neu5Acα 2–8Neu5Ac and Galβ 1–4GlcNAcβ 1–3Gal with the binding for inuenza A
viruses; thus, their comparable binding poses are expected to occur at the virus HA binding pockets
(Fig.6C,D).
Discussion
e objective of this study was to characterize the host-specic glycan substructure responding to inu-
enza A virus infections. Glycan microarray data provide an opportunity to systematically study the
factors that determine virus–glycan binding. However, such analyses have several limitations. e rst
limitation is that glycan microarray data are not quantitative because values from batch to batch are
highly variable. e variability is caused by spot intensities dependent on immobilization eciency and
results in the misleading use of uorescence intensities to quantify binding anities51. e second lim-
itation is that the glycans on microarray do not represent all glycans or all substructures in the natural
hosts, and they are also distributed dierently from those in nature. e last limitation is that the number
of datasets for inuenza A viruses from viruses of dierent host origins are not equal. For example, we
have 155 datasets for human-origin inuenza A viruses but only 7 for canine-origin inuenza A viruses.
In this study, we expected association analysis to detect signicantly nonrandom, but possibly infre-
quent, substructure features contributing to inuenza A virus binding. To ensure better coverage of all
potential substructures, hierarchical clusters (mono-, di-, tri-, and tetra-) of substructure proles were
characterized and integrated into data mining, and our analyses focused on the terminal structures. To
minimize the potential noise across dierent datasets due to variations in glycan microarray versions and
experiments, we integrated the signicant substructures extracted from each individual dataset by PLS
regression. To identify the host-associated glycan substructures, we categorized 211 data entries into ve
categories (human, swine, canine, waterfowl, and terrestrial birds) and then formulated glycan substruc-
ture problems as a typical association mining problem, where we treated glycan substructure features
as products, virus host types as the only label of customs, and the glycan–virus binding signals in the
Figure 6. ree-dimensional structures of dierent hemagglutinin (HA) receptor binding pockets
interacting with various glycan substructures. (A) Human-origin (Protein Data Bank [PDB] entry 3LZG)
and swine-origin (PDB 1RVT) HA (recognizing α 2,6 sialic acid) superposed and bound to 6SLN (analogous
to Neu5Aca2–6Galb1–4GlcNAc). (B) Canine-origin (PDB 4UO5) and avian-origin (PDB 2FK0) HA
(recognizing α 2,3 sialic acid) superposed and bound to 3SLN (analogous to Neu5Aca2–3Galb1–4GlcNAc).
(C) A predicted docked structure of Neu5Acα 2–8 Neu5Acα 2–8Neu5Ac interacting with HA receptor
binding pocket (human-origin, PDB 3LZG). (D) A predicted docked structure of Galβ 1–4GlcNAcβ 1–3Gal
interacting with HA receptor binding pocket (avian-origin, PDB 2FK0).
www.nature.com/scientificreports/
13
SCIENTIFIC RepoRts | 5:15778 | DOI: 10.1038/srep15778
dataset as transactions. Comparing to other methods, either statistical or mining strategies, our formu-
lation of the problem benets the novel observations in this study in two following ways. First, aer the
PLS regressions on individual glycan microarray entries, the binding transection denition was used to
integrate all of them for a cross-array analysis, by which we overcame the challenges from the varying
numbers of glycans on dierent version of arrays. Second, the association mining strategy avoided par-
ticular hypothesis before analyses and were able to detect rare but potentially signicant rules.
We have not been able to use this method to identify the specic substructures for glycan bind-
ings when multiple terminal glycans are present. For example, glycans with dierent terminals (e.g.
sialic acid and Gal) were observed frequently, but they may both be important players during inuenza
virus binding because they could bind inuenza viruses simultaneously. To avoid this problem, in this
study, we ignored the associated substructures with branch linkages, because they may be extracted from
a glycan with other terminals and by themselves may not contribute to virus binding. To avoid such
false-positives, we included in the results only terminal substructures without branches. Moreover, four
substructure denitions (mono-, di-, tri-, tetrasaccharide) could lead to overlapped glycan features that
were associated with the same virus host. For example, in Fig.2, swine-associated disaccharides are all
subsets of the corresponding trisaccharides, which are subsets of corresponding tetrasaccharides. Similar
patterns could be observed with other host-origin categories (Table S9). To be consistent, we interpreted
these overlapped rules by ignoring subset features and by keeping only substructures with the highest
number of saccharide residues (see Supplementary Methods).
Our results show that (1) human-origin inuenza A viruses could bind glycans with
Neu5Acα 2–8Neu5Acα 2–8Neu5Ac and Neu5Gcα 2–6Galβ 1–4GlcNAc substructures; (2) Galβ and
GlcNAcβ terminal substructures, without any existing sialic acid terminals, are associated with the glycan
binding of human-, swine-, and avian-origin inuenza A viruses; (3) Sulfated Neu5Acα 2–3 substruc tures
are believed to be associated with the glycan binding of human- and swine-origin inuenza A viruses.
ese observations, on one hand, are consistent with previously reported results about various types of
host-origin inuenza A viruses5. On the other hand, we also identied other substructures: α 2,6-linked
Neu5Gc substructures, α 2,8-linked multiple sialic acids, substructures with a Gal and GlcNAc terminals,
and sulfated α 2,3-linked Neu5Ac, which contribute to dierent virus bindings. ese newly discovered
inuenza A binding moieties, particularly those with the non-sialic acidic saccharides (Gal, GlcNAc),
may suggest that it is the structural pattern of acidic acids, instead of just Neu5Ac, Neu5Gc themselves,
which are recognized by inuenza viruses of various host origins.
e potential glycan receptors with α 2,8-linked sialic acid were reported to be associated with
inuenza virus binding22, which supports our results with Neu5Acα 2–8Neu5Acα 2–8Neu5Ac for
human inuenza viruses. e relatively low 3D structural similarities between this substructures and
human-like α 2,6-linked sialic acid substructures (Table4) could imply a potentially novel binding mode
for Neu5Acα 2–8Neu5Acα 2–8Neu5Ac (Fig.6C). Similarly, it has been reported that glycans with Gal ter-
minals could play a role in some virus receptor binding52,53. Our association results detailed this conclu-
sion, especially for Galβ 1–4GlcNAcβ 1–3Gal substructure, by supplying similar structural characteristics
to substructures with sialic acid. Concerning the associations detected for sulfated α 2,3-linked Neu5Ac,
it was reported that sulfated glycan motifs might increase inuenza virus binding34. Our results further
suggest that this sulfation process may lead to a SA2,3Gal binding for a SA2,6Gal-binding virus. It is
worth mentioning that all these substructure motifs were infrequent substructures in association rules
(Table S9), indicating eectiveness of association mining in this study.
Our three-dimensional structure analysis of representative host-specic substructures showed that for
trisaccharides, the shape angle formed by mass centers of three residues could be the key feature that dis-
tinguishes α 2,6-linked, α 2,8-linked and α 2,3-linked glycans and their virus host tropisms (Fig.4A–D).
Although recent studies argued that the dierent torsion angles of residue linkages could be the reason
for their diverse chain shapes29,44, our torsion angle values calculated from the three-dimensional struc-
tures did not support a role for torsion angle in forming the overall trisaccharide chain shapes. Hence,
we argue that signicant host-specic patterns related to glycan shape may become evident if shape
angles are measured instead of exible torsion angles. In addition, for trisaccharides without sialic acid
terminals (e.g. Galβ 1–4GlcNAcβ 1–3Gal), neither torsion angles nor RMSD values could suggest any
host-specic patterns from our results. However, since we only found a few unique such glycans asso-
ciated with inuenza viruses, we considered them only as individual cases of virus binding without an
identiable structural feature for host tropism.
References
1. Webster, . G., Bean, W. J., Gorman, O. T., Chambers, T. M. & awaoa, Y. Evolution and ecology of inuenza A viruses.
Microbiol. ev. 56, 152–179 (1992).
2. Sehel, J. J. & Wiley, D. C. eceptor binding and membrane fusion in virus entry: the inuenza hemagglutinin. Annu. ev.
Biochem. 69, 531–569, doi: 10.1146/annurev.biochem.69.1.531 (2000).
3. Tong, S. et al. A distinct lineage of inuenza A virus from bats. Proc. Natl. Acad. Sci. USA 109, 4269–4274, doi: 10.1073/
pnas.1116200109 (2012).
4. Tong, S. et al. New world bats harbor diverse inuenza A viruses. PLoS Pathog. 9, e1003657,doi: 10.1371/journal.ppat.1003657
(2013).
5. de Graaf, M. & Fouchier, . A. ole of receptor binding specicity in inuenza A virus transmission and pathogenesis. EMBO
J. 33, 823–841, doi: 10.1002/embj.201387442 (2014).
www.nature.com/scientificreports/
14
SCIENTIFIC RepoRts | 5:15778 | DOI: 10.1038/srep15778
6. Nicholls, J. M., Bourne, A. J., Chen, H., Guan, Y. & Peiris, J. S. Sialic acid receptor detection in the human respiratory tract:
evidence for widespread distribution of potential binding sites for human and avian inuenza viruses. espir. es. 8, 73, doi:
10.1186/1465-9921-8-73 (2007).
7. Walther, T. et al. Glycomic analysis of human respiratory tract tissues and correlation with inuenza virus infection. PLoS Pathog.
9, e1003223, doi: 10.1371/journal.ppat.1003223 (2013).
8. Franca, M., Stallnecht, D. E. & Howerth, E. W. Expression and distribution of sialic acid inuenza virus receptors in wild birds.
Avian Pathol. 42, 60–71, doi: 10.1080/03079457.2012.759176 (2013).
9. Costa, T. et al. Distribution patterns of inuenza virus receptors and viral attachment patterns in the respiratory and intestinal
tracts of seven avian species. Vet. es. 43, 28, doi: 10.1186/1297-9716-43-28 (2012).
10. Nelli, . . et al. Comparative distribution of human and avian type sialic acid inuenza receptors in the pig. BMC Vet. es. 6,
4, doi: 10.1186/1746-6148-6-4 (2010).
11. Trebbien, ., Larsen, L. E. & Viu, B. M. Distribution of sialic acid receptors and inuenza A virus of avian and swine origin in
experimentally infected pigs. Virol. J. 8, 434, doi: 10.1186/1743-422X-8-434 (2011).
12. Scholtisse, C. Pigs as ‘mixing vessels’ for the creation of new pandemic inuenza A viruses. Med. Prin. Pract. 2, 65–71 (1990).
13. Ito, T. et al. Molecular basis for the generation in pigs of inuenza A viruses with pandemic potential. J. Virol. 72, 7367–7373
(1998).
14. Van Pouce, S. G., Nicholls, J. M., Nauwync, H. J. & Van eeth, . eplication of avian, human and swine inuenza viruses in
porcine respiratory explants and association with sialic acid distribution. Virol. J. 7, 38, doi: 10.1186/1743-422X-7-38 (2010).
15. Blixt, O. et al. Printed covalent glycan array for ligand proling of diverse glycan binding proteins. Proc. Natl. Acad. Sci. USA
101, 17033–17038, doi: 10.1073/pnas.0407902101 (2004).
16. Alvarez, . A. & Blixt, O. Identication of ligand specicities for glycan-binding proteins using glycan arrays. Methods Enzymol.
415, 292–310, doi: 10.1016/S0076-6879(06)15018-1 (2006).
17. Stevens, J., Blixt, O., Paulson, J. C. & Wilson, I. A. Glycan microarray technologies: tools to survey host specicity of inuenza
viruses. Nat. ev. Microbiol. 4, 857–864, doi: 10.1038/nrmicro1530 (2006).
18. aman, . et al. Advancing glycomics: implementation strategies at the consortium for functional glycomics. Glycobiology 16,
82–90, doi: 10.1093/glycob/cwj080 (2006).
19. Stevens, J. et al. Glycan microarray analysis of the hemagglutinins from modern and pandemic inuenza viruses reveals dierent
receptor specicities. J. Mol. Biol. 355, 1143–1155, doi: 10.1016/j.jmb.2005.11.002 (2006).
20. Stevens, J. et al. Structure and receptor specicity of the hemagglutinin from an H5N1 inuenza virus. Science 312, 404–410,
doi: 10.1126/science.1124513 (2006).
21. umari, . et al. eceptor binding specicity of recent human H3N2 inuenza viruses. Virol. J. 4, 42, doi: 10.1186/1743-422X-
4-42 (2007).
22. Childs, . A. et al. eceptor-binding specicity of pandemic inuenza A (H1N1) 2009 virus determined by carbohydrate
microarray. Nat. Biotechnol. 27, 797–799, doi: 10.1038/nbt0909-797 (2009).
23. Nobusawa, E., Ishihara, H., Morishita, T., Sato, . & Naajima, . Change in receptor-binding specicity of recent human
inuenza A viruses (H3N2): a single amino acid change in hemagglutinin altered its recognition of sialyloligosaccharides.
Virology 278, 587–596, doi: 10.1006/viro.2000.0679 (2000).
24. Liu, Y. et al. Altered receptor specicity and cell tropism of D222G hemagglutinin mutants isolated from fatal cases of pandemic
A(H1N1) 2009 inuenza virus. J. Virol. 84, 12069–12074, doi: 10.1128/JVI.01639-10 (2010).
25. Yang, Z. Y. et al. Immunization by avian H5 inuenza hemagglutinin mutants with altered receptor binding specicity. Science
317, 825–828, doi: 10.1126/science.1135165 (2007).
26. Belser, J. A. et al. Eect of D222G mutation in the hemagglutinin protein on receptor binding, pathogenesis and transmissibility
of the 2009 pandemic H1N1 inuenza virus. PLoS One 6, e25091, doi: 10.1371/journal.pone.0025091 (2011).
27. Puzelli, S. et al. Transmission of hemagglutinin D222G mutant strain of pandemic (H1N1) 2009 virus. Emerg. Infect. Dis. 16,
863–865, doi: 10.3201/eid1605.091815 (2010).
28. Yang, G. et al. Mutation tryptophan to leucine at position 222 of haemagglutinin could facilitate H3N2 inuenza A virus
infection in dogs. J. Gen. Virol. 94, 2599–2608 (2013).
29. Chandrasearan, A. et al. Glycan topology determines human adaptation of avian H5N1 virus hemagglutinin. Nat. Biotechnol.
26, 107–113, doi: 10.1038/nbt1375 (2008).
30. Stevens, J. et al. eceptor specicity of inuenza A H3N2 viruses isolated in mammalian cells and embryonated chicen eggs. J.
Virol. 84, 8287–8299, doi: 10.1128/JVI.00058-10 (2010).
31. Cholleti, S. . et al. Automated motif discovery from glycan array data. OMICS 16, 497–512, doi: 10.1089/omi.2012.0013 (2012).
32. Porter, A. et al. A motif-based analysis of glycan array data to determine the specicities of glycan-binding proteins. Glycobiology
20, 369–380, doi: 10.1093/glycob/cwp187 (2010).
33. Xuan, P., Zhang, Y., Tzeng, T. ., Wan, X. F. & Luo, F. A quantitative structure-activity relationship (QSA) study on glycan array
data to determine the specicities of glycan-binding proteins. Glycobiology 22, 552–560, doi: 10.1093/glycob/cwr163 (2012).
34. Ichimiya, T., Nishihara, S., Taase-Yoden, S., ida, H. & Aoi-inoshita, . Frequent glycan structure mining of inuenza virus
data revealed a sulfated glycan motif that increased viral infection. Bioinformatics 30, 706–711, doi: 10.1093/bioinformatics/
btt573 (2014).
35. Hand, D. J., Mannila, H. & Smyth, P. In Principles of data mining. Ch. 13, 254–267 (MIT press, 2001).
36. Geng, L. & Hamilton, H. J. Interestingness measures for data mining: A survey. ACM Comput. Surv. (CSU) 38, 9 (2006).
37. Borgelt, C. & ruse, . Induction of association rules: Apriori implementation. In Proceedings of the 15th conference on Compstat.
395–400 (Springer, 2002).
38. Berman, H. M. et al. e protein data ban. Nucleic Acids es. 28, 235–242 (2000).
39. Xu, ., McBride, ., Nycholat, C. M., Paulson, J. C. & Wilson, I. A. Structural characterization of the hemagglutinin receptor
specicity from the 2009 H1N1 inuenza pandemic. J. Virol. 86, 982–990, doi: 10.1128/JVI.06322-11 (2012).
40. Chen, C., Fu, Z., im, J. J., Barbieri, J. T. & Baldwin, M. . Gangliosides as high anity receptors for tetanus neurotoxin. J. Biol.
Chem. 284, 26569–26577, doi: 10.1074/jbc.M109.027391 (2009).
41. han, Z. M. et al. Crystallographic and glycan microarray analysis of human polyomavirus 9 VP1 identies N-glycolyl neuraminic
acid as a receptor candidate. J. Virol. 88, 6100–6111, doi: 10.1128/JVI.03455-13 (2014).
42. Holmner, A. et al. Crystal structures exploring the origins of the broader specicity of escherichia coli heat-labile enterotoxin
compared to cholera toxin. J. Mol. Biol. 406, 387–402, doi: 10.1016/j.jmb.2010.11.060 (2011).
43. Guo, Y. et al. Structural basis for distinct ligand-binding and targeting properties of the receptors DC-SIGN and DC-SIGN.
Nat. Struct. Mol. Biol. 11, 591–598, doi: 10.1038/nsmb784 (2004).
44. amiya, Y., Yagi-Utsumi, M., Yagi, H. & ato, . Structural and molecular basis of carbohydrate-protein interaction systems as
potential therapeutic targets. Curr. Pharm. Des. 17, 1672–1684 (2011).
45. Xu, . et al. Structural basis of preexisting immunity to the 2009 H1N1 pandemic inuenza virus. Science 328, 357–360, doi:
10.1126/science.1186430 (2010).
www.nature.com/scientificreports/
15
SCIENTIFIC RepoRts | 5:15778 | DOI: 10.1038/srep15778
46. Gamblin, S. J. et al. e structure and receptor binding properties of the 1918 inuenza hemagglutinin. Science 303, 1838–1842,
doi: 10.1126/science.1093155 (2004).
47. Collins, P. J. et al. ecent evolution of equine inuenza and the origin of canine inuenza. Proc. Natl. Acad. Sci. USA 111,
11175–11180, doi: 10.1073/pnas.1406606111 (2014).
48. Ponder, J. W. & Case, D. A. Force elds for protein simulations. Adv Protein Chem 66, 27–85 (2003).
49. Pron, S. et al. GOMACS 4.5: a high-throughput and highly parallel open source molecular simulation toolit. Bioinformatics
29, 845–854, doi:10.1093/bioinformatics/btt055 (2013).
50. Schymowitz, J. et al. e FoldX web server: an online force eld. Nucleic Acids es. 33, W382–388, doi: 10.1093/nar/gi387
(2005).
51. Liang, P. H., Wu, C. Y., Greenberg, W. A. & Wong, C. H. Glycan arrays: biological and medical applications. Curr. Opin. Chem.
Biol. 12, 86–92, doi: 10.1016/j.cbpa.2008.01.031 (2008).
52. Shen, S., Bryant, . D., Brown, S. M., andell, S. H. & Asoan, A. Terminal N-lined galactose is the primary receptor for adeno-
associated virus 9. J. Biol. Chem. 286, 13532–13540 (2011).
53. Upham, J. P., Picett, D., Irimura, T., Anders, E. M. & eading, P. C. Macrophage receptors for inuenza A virus: role of the
macrophage galactose-type lectin and mannose receptor in viral entry. J. Virol. 84, 3730–3737, doi: 10.1128/JVI.02148-09 (2010).
Acknowledgements
We thank Drs. Robert Woods and Chi-Ren Shyu for critical discussions. is study was supported by
grants 1R15AI107702 and P20GM103646 from National Institutes of Health.
Author Contributions
N.Z. and X.F.W. conceived the experiment design. N.Z. conducted the experiments. N.Z. and B.E.M.
performed the data collection. F.L. implemented the feature extraction program. N.Z. and C.K.Y. analyzed
the results. N.Z., C.K.Y. and X.F.W. wrote the manuscript.
Additional Information
Supplementary information accompanies this paper at http://www.nature.com/srep
Competing nancial interests: e authors declare no competing nancial interests.
How to cite this article: Zhao, N. et al. Association analyses of large-scale glycan microarray data
reveal novel host-specic substructures in inuenza A virus binding glycans. Sci. Rep. 5, 15778; doi:
10.1038/srep15778 (2015).
is work is licensed under a Creative Commons Attribution 4.0 International License. e
images or other third party material in this article are included in the article’s Creative Com-
mons license, unless indicated otherwise in the credit line; if the material is not included under the
Creative Commons license, users will need to obtain permission from the license holder to reproduce
the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/