Defining the Interaction of Human Soluble Lectin ZG16p and Mycobacterial Phosphatidylinositol Mannosides

Article abstract of DOI:10.1002/cbic.201500103

ZG16p is a soluble mammalian lectin that interacts with mannose and heparan sulfate. Here we describe detailed analysis of the interaction of human ZG16p with mycobacterial phosphatidylinositol mannosides (PIMs) by glycan microarray and NMR. Pathogen-rela

Full text of DOI:10.1002/cbic.201500103

DOI: 10.1002/cbic.201500103
Full Papers
Defining the Interaction of Human Soluble Lectin ZG16p
and Mycobacterial Phosphatidylinositol Mannosides
Shinya Hanashima,[a] Sebastian Gçtze,[b, c] Yan Liu,[d] Akemi Ikeda,[a] Kyoko Kojima-Aikawa,[e]
Naoyuki Taniguchi,[f] Daniel Varón Silva,[b, c] Ten Feizi,[d] Peter H. Seeberger,[b, c] and
Yoshiki Yamaguchi*[a]
ZG16p is a soluble mammalian lectin that interacts with man-
nose and heparan sulfate. Here we describe detailed analysis
of the interaction of human ZG16p with mycobacterial phos-
phatidylinositol mannosides (PIMs) by glycan microarray and
NMR. Pathogen-related glycan microarray analysis identified
phosphatidylinositol mono- and di-mannosides (PIM1 and
PIM2) as novel ligand candidates of ZG16p. Saturation transfer
difference (STD) NMR and transferred NOE experiments with
chemically synthesized PIM glycans indicate that PIMs prefer-
entially interact with ZG16p by using the mannose residues.
The binding site of PIM was identified by chemical-shift pertur-
bation experiments with uniformly 15N-labeled ZG16p. NMR
results with docking simulations suggest a binding mode of
ZG16p and PIM glycan; this will help to elucidate the physio-
logical role of ZG16p.
Introduction
ZG16p is a soluble protein that was initially identified in rat
pancreas where it is associated with the zymogen granule
membrane.[1] The protein was recently detected in human
colon, small intestine, and serum, as well as in the pancreas.[2]
ZG16p plays a role in packaging pancreatic enzymes into zym-
ogen granules and separating them from constitutively secret-
ed proteins.[3] ZG16p has been proposed as a primary binding
partner of glycosaminoglycans (GAGs) in pancreatic granules.[4]
The ZG16p amino acid sequence has homology with the
carbohydrate-recognition domain (CRD) of jacalin, a jackfruit
lectin.[4–5] A recent X-ray crystallographic analysis revealed that
human ZG16p has a jacalin-related b-prism fold, similar to that
in Banlec, a mannose-binding lectin in bananas.[5] Glycan mi-
croarray screening has demonstrated that ZG16p has specifici-
ty for glycans consisting of mannose, including mannan and
Ser/Thr-linked O-mannose.[2,5–6] Asp151 is a key mannose-bind-
ing residue, as mutation of Asp151 to Asn abolished glycan
binding.[2] Interestingly, ZG16p also binds GAGs, especially hep-
arin and heparin sulfate.[4] Our recent crystallographic and
NMR studies indicated that mannose and GAG binding to
ZG16p occurs with distinct binding modes:[6] mannose uses
a shallow binding site made from three loops (the GG loop
(between b1 and b2 strands, Gly31–Gly35), the recognition
loop (between b7 and b8, Lys102–Tyr104), and the binding
loop (between b11 and b12, Ser146–Leu149)), whereas sulfated
oligosaccharides bind to a positively charged surface consist-
ing of a cluster of basic amino acid residues.
[a] Dr. S. Hanashima,+ A. Ikeda, Dr. Y. Yamaguchi
Structural Glycobiology Team
RIKEN-Max Planck Joint Research Center for Systems Chemical Biology
RIKEN Global Research Cluster
Wako, Saitama 351-0198 (Japan)
E-mail: yyoshiki@riken.jp
[b] Dr. S. Gçtze,+ Dr. D. Varón Silva, Prof. P. H. Seeberger
Department of Biomolecular Systems
Max Planck Institute of Colloids and Interfaces
Am Mühlenberg 1, 14424 Potsdam (Germany)
[c] Dr. S. Gçtze,+ Dr. D. Varón Silva, Prof. P. H. Seeberger
Institute of Chemistry and Biochemistry, Freie Universität Berlin
Arnimallee 22, 14195 Berlin (Germany)
Most of the mannose-binding animal C-type lectins, includ-
ing dendritic cell-specific intercellular adhesion molecule-3-
grabbing non-integrin (DC-SIGN), the mannose receptor,
Dectin-2, macrophage inducible C-type lectin (Mincle), and
Langerin, are involved in host immunity through recognition
of the mannans of pathogenic bacteria.[7]
[d] Dr. Y. Liu,+ Prof. T. Feizi
Glycosciences Laboratory, Department of Medicine
Imperial College London
Du Cane Road, London W12 0NN (UK)
[e] Prof. K. Kojima-Aikawa
The Glycoscience Institute, Ochanomizu University
Otsuka, Bunkyo-ku, Tokyo 112-8610 (Japan)
Most of these lectins are signaling molecules with trans-
membrane domains, although some, also involved in host im-
munity, are secreted. Mannose-binding lectin (MBL) is a liver-
derived serum protein that has a role in the innate immune re-
sponse by binding to the surface glycans of a wide range of
pathogens.[7a,11] Proteins of the regenerating islet-derived (Reg)
family are secreted proteins containing a C-type lectin-like
[f] Prof. N. Taniguchi
Disease Glycomics Team
RIKEN-Max Planck Joint Research Center for Systems Chemical Biology
RIKEN Global Research Cluster
Wako, Saitama 351-0198 (Japan)
[+] These authors contributed equally to this work.
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.2.201500103.
ChemBioChem 2015, 16, 1502 – 1511
1502
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
domain, and play a role in pancreatic function and associated
diseases.[12]
Similar to the C-type lectins and Regs, ZG16p binds patho-
genic fungi Candida and Malassezia in a mannose-dependent
manner.[2] Therefore, it is possible that human ZG16p is in-
volved in the gastrointestinal immune system through binding
target glycans of pathogens. In order to gain insights into the
structure–function relationships of ZG16p in pathogen recogni-
tion, we determined the glycan-binding specificity of ZG16p
by using a pathogen-related glycan microarray. ZG16p binds
to phosphatidyl inositol mannosides (PIM1 and PIM2,
Scheme 1), major cell-wall components of some pathogenic
Scheme 1. Chemical structures of PIM1 and PIM2 and synthesized PIM gly-
cans 1 and 2.
Figure 1. Microarray analysis of a small pathogen-related array of glycan
probes (lipid-tagged probes or polysaccharides). The fluorescence intensities
are shown for A) wild-type ZG16p, B) the D151N mutant, and C) plant lectin
Con A (included for comparison). Glycan probes are detailed in Table 1. Each
probe was printed in duplicate at two levels (light and dark gray bars; 1–8:
2 and 5 fmol per spot; 9–20: 0.03 and 0.1 ng).
bacteria including Mycobacterium tuberculosis.[16] We also eluci-
dated details of the binding mode of human ZG16p with PIM
glycans by using NMR experiments and docking simulations.
The findings raise the possibility that human ZG16p is involved
in mucosal defense against bacteria through recognitions of
short PIM glycans.
Bacterial infections and host defense mechanisms must be
studied from various aspects. The huge diversity of bacterial
glycans and the presence of numerous uncharacterized host
lectins preclude rapid progress in this research area. The bind-
ing preference of human ZG16p lectin is unique in that it in-
volves both mannose and sulfated glycosaminoglycans. How-
ever the role of ZG16p has not yet been fully characterized in
terms of sugar binding. Here we combine glycan array screen-
ing, synthetic chemistry and structural biology approaches to
elucidate the possible role and mechanism of ZG16p in patho-
gen recognition.
study[6] and served as controls here. Wild-type ZG16p gave
strong binding signals with PIM1 and PIM2 (arrayed as a mix-
ture), with similar intensities to those for the positive controls
1 and 2 (Man-Thr-DH and Man-Ser-DH; Figure 1A). Interesting-
ly, the intensity for hexamannoside PIM6 was much lower.
PIM2 and PIM6 are the two most abundant PIM classes in My-
cobacterium bovis, M. tuberculosis H37Rv, and Mycobacterium
smegmatis 607 (Figure S1 in the Supporting Information).[18]
ZG16p preferentially bound PIM1 and PIM2 over PIM6. DC-
SIGN, through its interaction with PIMs, is considered a key
molecule during infection with M. tuberculosis.[19] In contrasts
to ZG16p, DC-SIGN preferentially binds PIM5 and PIM6 over
shorter PIMs.[20] In contrast, the D151N mutant of ZG16p[2]
showed little or no bindings to these probes (Figure 1B), in ac-
cordance with a previous report on loss of mannose binding.[2]
Lipoarabinomannan (LAM) and lipomannan (LM), the major
glycolipids in the cell wall of all Mycobacterium species, elicited
little or no binding signal with ZG16p. LAM and LM from M. tu-
berculosis (but not LAM from M. smegmatis; capped by phos-
phatidyl inositols) were well bound by plant lectin concanava-
lin A (Con A), which was included as a positive control (Fig-
ure 1C). No significant binding was observed for ZG16p to
M. tuberculosis cord factor trehalose-6,6’-dimycolate (TDM), sul-
folipids, or arabinogalactans, nor to the fungal-derived glucan
Results and Discussion
Pathogen glycan-focused microarray
We speculated that ZG16p plays a role in the host immune de-
fense by binding cell-wall glycans of pathogenic bacteria,
hence we performed a pathogen-related carbohydrate micro-
array analysis (Figure 1).[17] The array (Table 1) comprised
a small set of glycoconjugates, lipid-linked glycans, and poly-
saccharides derived from mycobacteria and fungal pathogens,
in addition to mannose-containing mammalian neoglycolipids
(Probes 1–4, Table 1) that were tested with ZG16p in a previous
ChemBioChem 2015, 16, 1502 – 1511
www.chembiochem.org
1503
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
commenced from the myo-inosi-
tol building block 3.[22] To pre-
pare 1, a temporary para-me-
thoxybenzyl (PMB) ether was
placed at C2 of 3, followed by
benzylation and acidic cleavage
to furnish glycosyl acceptor 4 in
58% yield over three steps. Gly-
cosylation of 4 with phosphate
5[23] at À408C in toluene exclu-
sively formed the a-linked pseu-
dodisaccharide 6 in 84% yield
(1JC1–H1 =175 Hz). In situ isomeri-
zation of the allyl ether generat-
ed iridium hydride,[24] and hy-
drolysis of the corresponding
enol ether resulted in an alcohol
function at C1 of inositol. Phos-
phonylation with the mixed an-
hydride of pivalic acid and H-
phosphonate 7[25] followed by
oxidation provided the triethyl-
ammonium salt 8 in 73% yield
over three steps from 6. Subject-
ing 8 to deacetylation and sub-
sequent final hydrogenolysis
over palladium in methanol
Table 1. Glycan probes in the microarray analysis.
No. Probe[a]
Structure
1
2
Man-Thr-DH
Man-Ser-DH
Mana-Thr-DH[b]
Mana-Ser-DH
3
Man7(D1)GN2-AO
4
5
Man9GN2-AO
M. tuberculosis TDM
purified trehalose dimycolate (TDM) from M. tuberculosis, strain
H37Rv
(BEI number NR-14844)
6
7
TDB
trehalose-6,6-dibehenate (TDB), a synthetic analogue of TDM (Sigma)
purified PIM 1 and 2 from M. tuberculosis, strain H37Rv
(BEI number NR-14846)
M. tuberculosis PIMs 1 and 2
8
9
M. tuberculosis PIM 6
M. tuberculosis MME
purified PIM 6 from M. tuberculosis, Strain H37Rv
(BEI number NR-14847)
purified mycolic acid methyl esters from M. tuberculosis, Strain H37Rv
(BEI number NR-14854)
10 M. tuberculosis TDM
as for probe 5
11 M. tuberculosis Sulfolipid-1
purified sulfolipid-1 from M. tuberculosis, Strain H37Rv
(BEI number NR-14845)
12 M. tuberculosis PIMs 1 and 2
13 M. tuberculosis PIM 6
14 M. tuberculosis LAM
as for probe 7
as for probe 8
yielded
1 in excellent yield
purified lipoarabinomannan (LAM) from M. tuberculosis, strain H37Rv
(BEI number NR-14848)
(97%) over two steps.
15 M. smegmatis LAM
16 M. tuberculosis LM
purified LAM from M. smegmatis (BEI number NR-14849)
purified lipomannan (LM) from M. tuberculosis, strain H37Rv
(BEI number NR-14850)
purified arabinogalactan from M. tuberculosis, strain H37Rv
(BEI number NR-14852)
The synthesis of 2 continued
with double glycosylation of 3
with 5 under conditions similar
to those used in the synthesis of
6. Pseudotrisaccharide 9 was ob-
tained in 62% yield, and the a-
configuration of both anomeric
17 M. tuberculosis arabinogalactan
18 pullulan from Pullularia pullulans
19 curdlan from Alcaligenes faecalis
20 pustulan from Umbilicaria papullo- b1-6 glucose polysaccharide (CalBiochem)
mixed-linked a1-4,a1-6 glucose polysaccharide (Megazyme)
b1-3 glucose polysaccharide (in 50 mm NaOH) (Megazyme)
sa
linkages was confirmed (1JC1–H1
=
[a] Probes 1–8 were printed at 2 and 5 fmol per spot; the rest were at 0.03 and 0.1 ng per spot. The neoglycoli-
pids (positions 1–4) are from the collection assembled in the course of research in the Glycosciences Laborato-
ry at Imperial College. [b] DH, amino lipid 1,2-dihexadecyl-sn-glycero-3-phosphoethanolamine (DHPE). [c] AO,
aminooxy-functionalized DHPE.
173 and 176 Hz). Exposure of 9
to PdCl2 in CH2Cl2/methanol se-
lectively removed the allyl ether
(64% yield). Final phosphoryla-
tion of the corresponding alco-
hol 10 with 7 formed phosphate
polysaccharides. The glycan array data suggest that ZG16p has
a preference for short a-mannose-related glycans, including
PIM1 and 2, over more-complex mannose-containing glycans.
11, which was deacetylated and submitted to hydrogenolysis
to provide 2 in 49% yield over four steps.
STD-NMR analysis of the interaction of PIM glycans with
ZG16p
Chemical synthesis of PIM1 and PIM2 glycans 1 and 2
To investigate the binding mode of ZG16p to PIM1 and PIM2
by NMR, we prepared phosphoglycans 1 and 2 by reported
procedures with some modifications (Scheme 2).[21] Instead of
a phosphodiester bearing a diacylglycerol moiety or functional-
ity that enables covalent attachment to surfaces or beads,
PIM1 glycan 1 and PIM2 glycan 2 structures used in our experi-
ments had a monoester of phosphoric acid at the C1-position
of myo-inositol to ensure solubility. The syntheses of 1 and 2
In order to understand the interactions of ZG16p with PIM gly-
cans, saturation-transfer difference (STD)-NMR spectra were re-
corded (Figure 2). In each case, there was a 100-fold excess of
ligand over protein in the NMR buffer (PBS, pH 7.4, 99% D2O).
Under these conditions, PIM1 glycan 1 apparently exhibited
a potent STD-NMR signal, whereas glycerol (asterisks in Fig-
ure 2A and C), which could not be fully removed during the
purification process of ZG16p, nearly disappeared. This clearly
ChemBioChem 2015, 16, 1502 – 1511
www.chembiochem.org
1504
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Scheme 2. Synthesis of PIM glycans 1 and 2. Reagents and conditions: a) PMBCl (1 equiv), NaH, DMF, À208C, 41% (69% based on recovered starting materi-
al); b) BnBr, NaH, DMF, 08C, 90%; c) CHCl3/TFA (9:1), 93%; d) 5 (1.3 equiv), TMSOTf, toluene, À408C, 84%; e) i: [Ir(COD)(PPh2Me)2]PF6, H2, THF; ii: HCl(aq), 92%;
f) i: 7, PivCl, pyridine; ii: I2, H2O, 79%; g) NaH, MeOH, 99%; h) 10% Pd/C, H2, MeOH, 98%; i) 5 (2.6 equiv), TMSOTf, toluene, À408C, 62%; j) PdCl2, CH2Cl2/MeOH
(1:1), 64%; k) i: 7, PivCl, pyridine; ii: I2, H2O, 68%; l) NaH, MeOH, 78%; m) 10% Pd/C, H2, MeOH, 92%.
indicates that the main constituent in the STD-NMR spectrum
shown in Figure 2B reflects the saturation transfer effect from
ZG16p. Likewise, PIM2 glycan 2 exhibited potent STD-NMR sig-
nals, thus confirming binding to ZG16p (Figure 2C and D).
The relative STD effects (STD%) suggest that the most pro-
nounced interactions between PIMs and ZG16p were at pro-
tons at C3–C6 of the mannose residue (Figure 2E). In the case
of the monomannosylated PIM1 glycan 1, the binding epitope
is mainly the pyranose ring at C3ÀC6, because the protons
showed a potent saturation effect (70–100%), whereas the ino-
sitol moiety showed weaker saturation (<70%). Although the
data obtained under these conditions must be considered
qualitative, this observation is consistent with the crystal struc-
ture of ManOMe–ZG16p complex (PDB ID: 3VZF), where C4–C6
of mannose interact with the protein.[6] In the case of PIM2
glycan 2, protons at M4 and M’4 exhibited similar saturation
effects. Although partial signal overlapping at positions 2, 3,
and 5 of two mannoses prevents precise epitope mapping, the
data imply that ZG16p interacts with PIM2 at the mannose res-
idues.
trum of PIM1 glycan 1 (4.6-fold excess) in the presence of
ZG16p provided the key inter-residual correlation between
Man-H1 and Ino-H2, and intra-residual correlations in negative
NOE (Figure 3A). In contrast, no correlation was observed
when using the D151N ZG16p mutant (Figure 3B). These data
suggest that the identified NOEs are transferred (TR)-NOEs
originating from ZG16p-bound state. The atomic distance of
inter-residue Man-H1–Ino-H2 was determined as 2.2 Š from rel-
ative intensity of the signal.
1
The H,1H NOESY spectrum of PIM2 glycan 2 (4.6-fold excess)
with ZG16p provided inter-residual correlations in Man-H1–
Ino-H2, Man’-H1–Ino-H6, and Man’-H1–InoH1 in negative NOE
(Figure 3C). In contrast, only the trace NOE correlations were
identified in the control spectrum (ZG16p-D151N; Figure 3D).
The atomic distances of the inter-residue protons, ManH1-
InoH2, Man’H1-InoH6, and Man’H1-InoH1, were determined at
2.3, 2.2, and 2.8 Š, respectively. We observed TR-NOE signals,
thus also suggesting selective binding of PIMs to ZG16p. The
binding site was further analyzed in titration experiments.
Chemical shift perturbation experiments of ZG16p with
phosphoglycans PIM1 and PIM2
TR-NOE analysis of the PIM1 and 2 glycans bound with
ZG16p
The interaction site(s) of the ligands on ZG16p was determined
by chemical shift perturbation experiments and 1H,15N HSQC
spectra. To achieve this, we prepared uniformly 15N-labeled
ZG16p ([15N]ZG16p) for 1H,15N HSQC, and 13C,15N doubly labeled
ZG16p ([13C,15N]ZG16p) for sequential signal assignments. The
In order to investigate the interactions of PIM glycans, we col-
lected 1H,1H NOESY spectra (Figure 3). In order to minimize
spin diffusion, the appropriate mixing time was determined by
1
the NOE build-up curve (Figure S2). The 2D H,1H NOESY spec-
ChemBioChem 2015, 16, 1502 – 1511
www.chembiochem.org
1505
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
The results of the titration with PIM1 glycan 1 are depicted
1
in Figure S3, and the weighted H,15N chemical-shift changes
(Ddavg) of ZG16p upon addition of a 20-fold excess of 1 are in
Figure 5A. The backbone amide signals of Lys36, Arg37, and
Gly147 showed large chemical-shift changes (Ddavg >0.06),
whereas those of Gly35, Ser146, and Leu149 were moderate
(0.04<Ddavg <0.06). Small chemical-shift changes (0.025<
Ddavg <0.04) were observed for Glu29, Tyr30, Gly31, Ser32,
Gly33, Gly34, Arg37, Asp82, Asn129, Ile142, Arg145, Asp151,
and Ala152. During the titration, the Leu149 signal broadened
at 20-fold ligand excess. In addition, the signal from Arg37 was
strongly affected (Ddavg, 0.06). The KD of PIM1 glycan 1 was
5.0 mm (Figure S4A).
The backbone amide signals perturbed on titration with
PIM2 glycan 2 were very similar to those with PIM1 (Figure 5).
The Ddavg of [15N]ZG16p upon addition of a 20-fold excess of 2
(Figure 5B) revealed large chemical-shift changes for Gly35,
Lys36, Arg37, and Gly147 (Ddavg >0.09). Moderate chemical-
shift changes (0.06<Ddavg <0.09) were observed for Gly33,
Gly34, Ser146, Asp151, and Ala152, and small changes (0.03<
Ddavg <0.06) were seen for Glu29, Tyr30, Gly31, Ser32, Ile142,
Val62, Val76, Asp82, and Arg145. The perturbed signals ap-
peared to broaden at an earlier stage in the titration compared
with PIM1 glycan 1. For example, the signals for Gly35 and
Gly147 were broad in the presence of a 20-fold excess of PIM2
glycan 2, whereas the corresponding signals remained sharp in
the presence of a 20-fold excess of PIM1 glycan 1. Additionally,
the Leu149 signal broadened before the 20-fold excess of
ligand was reached. These results might reflect a chemical ex-
change process induced by PIM2 glycan 2, similar to the NMR
time scale. The titration yielded a KD for PIM2 glycan 2 of
3.0 mm (Figure S4B). The limited differences in perturbed back-
bone NH signals between PIM1 and PIM2 glycans 1 and 2 sup-
port both glycans interacting at the same binding site.
To clarify the interaction between ZG16p and PIMs, the
amino acid residues that showed chemical-shift perturbations
were identified onto the crystal structure[5] (Figure 6; heat map
color scale reflects the averaged chemical-shift changes in the
PIM2 titration). The mapping clearly indicates that most of the
perturbed residues are at the surface of the protein, except for
Asp82 and Ile142. The ligand-binding site is nearly identical to
that for mannose-specific plant lectins of the Jacalin-related
lectin family with a Greek-key motif.[30] They are in two seg-
ments, the GG loop (Gly29–Arg37) and the binding loop
(Ser146–Ala152). The two loops form a shallow ligand-binding
site, with Gly33, Gly34, Gly35, and Gly147 below, Arg37,
Arg145, and Lys36 on one side, and Asp151 on the other;
these can easily accommodate a mannose residue.
Figure 2. STD-NMR binding epitope of PIM1 glycan 1 and PIM2 glycan 2 to
ZG16p: A) 1H NMR off-resonance spectrum, and B) STD-NMR spectrum of
PIM1 glycan 1; C) 1H NMR off-resonance spectrum, and D) STD-NMR spec-
trum of PIM2 glycan 2. E) The binding epitopes of PIM1 and PIM2 were de-
termined by relative STD (STD%). In PIM2 glycan 2, STD effects were as-
signed as identical for overlapping signals at M2/M’2, M3/M’3, and M5/M’5.
Asterisks indicate signals from residual glycerol (in the protein purification
buffer). The STD-NMR spectra was collected at 58C for samples in PBS
(pH 7.4, 99% D2O). I: inositol, M: mannose
assignment of the backbone amide signals of ZG16p was ach-
ieved for [13C,15N]ZG16p in 2D and 3D NMR experiments
(Table S1).[29]
Model of ZG16p-PIM1 complex
Human ZG16p has a higher affinity for PIM1 glycan 1 and PIM2
glycan 2 (KD =5.0 and 3.0 mm, respectively) than a-methylman-
noside (ManOMe, KD =15 mm, Figure S4C), even though the
contact-site epitopes are almost identical. This implies that ad-
PIM1 glycan 1 and PIM2 glycan 2 were titrated into a
[15N]ZG16p solution, and the signal perturbations were record-
ed (Figure 4). The results suggest that binding is a fast ex-
change process, because each set of specific signals featured
a gradual chemical-shift change under the titration conditions.
ditional residues (not identified by H,15N HSQC measurements)
1
might be involved in phosphoglycan binding. Attempts to co-
ChemBioChem 2015, 16, 1502 – 1511
www.chembiochem.org
1506
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
epitope determined by STD-
NMR. The PIM1 mannose-bind-
ing mode in the model is consis-
tent with the ZG16p–ManOMe
X-ray crystal structure (Fig-
ure S5).[6] However, importantly,
the model suggests that another
amino acid residue (Tyr104) in
the recognition loop and the
side chain of Ser148 in binding
loop interact with the inositol
moiety. Unfortunately, the NMR
signals from Tyr104 and Ser148
were broadened in the 1H,15N
HSQC spectrum, possibly due to
chemical exchange. Although
there is no direct evidence for
interactions with Tyr104 and
Figure 3. NOESY spectra of A) PIM1 glycan 1 with ZG16p, B) PIM1 glycan 1 with ZG16p-D151N, C) PIM2 glycan 2
with ZG16p, and D) PIM1 glycan 2 with ZG16p-D151N. The spectra were collected with a mixing time of 250 ms
at 108C. Gray: positive signals; black: negative signal; dotted lines: chemical shifts of Man-H1 (A and B; PIM1), or
Man-H1 and Man’-H1 (C and D; PIM2). Residual HDO signals are observed at ~5.0 ppm.
Ser148, it is very likely that these amino acids are responsible
for the tighter binding of PIM1 and PIM2 glycans in compari-
son to ManOMe. In line with this, our structural analysis shows
that the hydroxy group of Tyr104 exhibits a water-mediated in-
teraction with Man-O-Ser, thereby assisting ligand binding.[6]
The chemical-shift perturbation experiments for PIM2 glycan
2 suggest that the binding mode is similar to that for PIM1
glycan 1. However, the STD-NMR epitope analysis indicated
the contribution of two mannoses. One possible explanation is
that PIM2 has two independent binding modes that use the
identical binding site on the protein (Figure S6).
Naturally occurring PIMs have a hydrophobic phosphatidyl
group at position O1 of inositol; this anchors it to the cell sur-
face of mycobacteria and is crucial for interactions with differ-
ent mammalian proteins. For example, mouse CD1d, a known
receptor of PIMs, has hydrophobic grooves where acyl chains
of glycolipids bind.[32] ZG16p lacks these structural motifs, and
evidently binding of this protein involves the sugar and per-
haps the phosphate moiety of PIMs.
Figure 4. 1H,15N HSQC spectra of uniformly 15N-labeled ZG16p in titration
with PIM2 glycan 2. Black: signals with no ligand; red: signals in the pres-
ence of 5 equiv of 2; green: signals in the presence of 20 equiv of 2; blue
arrows: directions of the chemical-shift changes.
crystallize ZG16p with either PIM1 glycan 1 or PIM2 glycan 2
were unsuccessful. Therefore a PIM–ZG16p complex model
based on the crystal structure of glycerol–ZG16p (PDB ID:
3APA)[5] was constructed. The docking simulation was per-
formed with the Glide[31] program of the software package
Maestro. The resulting models were ranked based on the data
obtained by STD and chemical shift perturbation experiments,
and a feasible model was thus identified (Figure 7). This was
validated by the results from the TR-NOE experiments. The dis-
tance between H1 of PIM1 mannose to InoH2 was 2.2 Š in the
TR-NOE experiments, and 2.3 Š in the docking simulations.
In the model, the interaction of PIM1 mannose with ZG16p
is mediated by intermolecular hydrogen bonds to the back-
bones of Gly35, Gly147, Ser148, and Leu149, and to the side
chain of Asp151 (GG loop and binding loop; Figure 7). In PIM1,
hydroxy groups at mannose positions 4 and 6 are highly in-
volved in the hydrogen bonds. This is consistent with our
chemical shift perturbation experiments and with the binding
Conclusion
We demonstrate that ZG16p preferentially interacts with Myco-
bacterium glycolipids PIM1 and PIM2 over PIM6 and other my-
cobacterial glyco-components. STD-NMR studies reveal the in-
teraction of this human lectin with the C3–C6 moiety of the
mannose residue. NMR perturbation experiments demonstrate
that the phosphoglycans of PIMs interact with the GG loop
(Gly31–Lys36) and the binding loop (Ser146-Asp151) of ZG16p,
and that their dissociation constants are three to five-fold
lower than those of ManOMe. Docking simulations combined
with NMR data implicate Tyr104 (in the recognition loop) in an
interaction with the inositol moiety of the PIMs. ZG16p might
play a role in the mucosal immune response by associating
with exogenous short PIMs of pathogenic bacteria, including
M. tuberculosis, and with endogenous glycosaminoglycans at
independent binding sites. Further work is required to eluci-
date the physiological function of this lectin.
ChemBioChem 2015, 16, 1502 – 1511
www.chembiochem.org
1507
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
1
Figure 5. The weighted H,15N chemical-shift changes (Ddavg) of the backbone amide of [15N]ZG16p upon binding with a 20-fold excess of A) PIM1 glycan 1 or
*
*
*
B) PIM2 glycan 2. : proline; : undetermined signals; : signal could not be analyzed because of perturbation effects.
KCl (3 mm)). After washing with PBS, the protein was eluted in PBS
Experimental Section
containing imidazole (500 mm). The His6-MBP tag was removed by
digestion with TEV protease at 48C for 12 h. The digested proteins
were passed through a Ni-Sepharose column, and final purification
was performed by size-exclusion chromatography (HiLoad 16/60
Superdex 75 pg; GE Healthcare) or cation exchange chromatogra-
phy (TOYOPEARL SP-650M or Giga Cap S-650M; Tosoh, Tokyo,
Japan). The buffer of the purified proteins were replaced with PBS
including D2O (99 or 10%, v/v); the pH was adjusted to 7.4 for
ligand experiments by STD-NMR and TR-NOESY, and to pH 6.5 for
1H,15N HSQC and backbone amide focusing experiments.
Expression and preparation of ZG16p: Recombinant ZG16p pro-
tein, uniformly 15N-labeled ([15N]ZG16p), and uniformly 13C/15N-la-
beled ([13C,15N]ZG16p) were prepared in the pCold-MBP (maltose-
binding protein) vector, according to a previously reported proce-
dure with slight modifications.[5] DNA fragments encoding human
ZG16p (residues 21–159 for crystallization, 21–167 for NMR; the
core lectin domain) were subcloned into pCold-MBP[35] for the pro-
duction of recombinant proteins. For the microarray analysis, gluta-
thione-S-transferase (GST)-tagged ZG16p protein (21–167) and its
mutant GST-ZG16p-D151N were prepared in the pCold-GST vector,
which was modified from pCold vector (Takara Bio Inc„ Japan). The
plasmid constructs were transformed into Escherichia coli
BL21(DE3) codon plus (Stratagene), and the cells were grown at
378C in either LB medium, M9 with [15N]NH4Cl (ISOTEC, Kürten-
Herweg, Germany; for [15N]ZG16p), or Spectra 9 (Cambridge Iso-
tope Laboratories, Tewksbury, MA; for [13C,15N]ZG16p). After induc-
tion with isopropyl b-d-thiogalactoside (0.1 mm), the cells were
cultured at 158C for 24 h, then harvested, resuspended, and soni-
cated in Tris-HCl (50 mm, pH 8.0) containing NaCl (50 mm), and
Bugbuster (Novagen/Merck Millipore). After centrifugation, the su-
pernatants containing His6-MBP-fused protein was collected and
applied to a Ni Sepharose column (GE Healthcare) equilibrated
with PBS (Na2HPO4 (8 mm, pH 7.4), KH2PO4 (1 mm), NaCl (137 mm),
Glycan microarray analysis: The microarrays comprised four man-
nose-containing neoglycolipids (NGLs), nine mycobacterial com-
pounds for antigen preparations (Biodefense and Emerging Infec-
tions Research Resources Repository; http://www.beiresources.org/
), and three fungal-derived glucan polysaccharides (sample infor-
mation in Table 1). The arrays were generated robotically using
a non-contact arrayer on nitrocellulose coated microarray slides.[17e]
All the probes were arrayed with carrier lipids (phosphatidylcholine
and cholesterol; Sigma–Aldrich) as previously described,[17e] except
for M. tuberculosis arabinogalactan and the glucan polysaccharides
(positions 17–20); these were arrayed without lipid carriers. The
lipid-linked probes (positions 1–8) were arrayed at 2 and 5 fmol
per spot, and these samples were quantified based on primulin
ChemBioChem 2015, 16, 1502 – 1511
www.chembiochem.org
1508
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Chemical synthesis of PIM1 and PIM2 glycans: Synthesis and
NMR spectra are in the Supporting Information.
General procedures for NMR experiments: NMR spectra were re-
corded with 500, 600, 700, and 800 MHz spectrometers (Bruker).
Protein solutions in PBS (sodium phosphate (10 mm, pH 6.5 or 7.4)
with NaCl (150 mm); in 99 or 10% (v/v) D2O) were used for NMR
1
experiments. H NMR chemical shifts were calibrated with an exte-
rior standard chemical shift (4,4-dimethyl-4-silapentane-1-sulfonic
acid (DSS), set to 0 ppm). 13C and 15N chemical shifts were calibrat-
ed by indirect referencing based on IUPAC-IUB recommended X/1H
resonance ratios (0.251449530 (13C/1H); 0.10132911 (15N/1H)).[37]
NMR data were processed with XWIN-NMR (ver. 3.5, Bruker) and
TopSpin (ver. 3.1, Bruker). The spectra were analyzed with SPARKY 3
(T. D. Goddard and D. G. Kneller, University of California, San Fran-
cisco) and displayed with XWIN-PLOT (ver. 3.5, Bruker).
STD-NMR: 1D STD-NMR experiments were performed in
a
600 MHz spectrometer with a TXI probe. The protein signal at À0.5
or À1 ppm was saturated with a 50 ms Gaussian pulse train with
60 times (on-resonance); reference spectra were obtained with irra-
diation at 40 ppm (off-resonance). The on-resonance and off-reso-
nance spectra were collected in an interleaved manner, and accu-
mulated into two different data sets. Water suppression was ach-
ieved with a WATERGATE pulse sequence (3–9–19 pulse train). In
STD-NMR experiments, 64 scans with three repetition loops were
required to obtain a good signal-to-noise ratio, and protein signals
were partially suppressed by using a 3–10 ms spin-lock pulse. In
binding epitope analyses, the signal intensities were evaluated
from (IoffÀIon)/Ioff, where Ion is the on-resonance signal intensity and
Ioff is off-resonance signal intensity. The values were normalized
within each glycan structure (highest value assigned 100%). Unla-
beled ZG16p (50 mm) in PBS (pH 7.4, 99% D2O) was used for STD-
NMR experiments, and 100-fold excess of ligand (PIM1 glycan 1 or
PIM2 glycan 2) was titrated into the protein solution.
Figure 6. Mapping surface residues in the PIM2 glycan interaction on the
crystal structure of human ZG16p (PDB ID; 3APA).[5] Ile142 (shift Ddavg 0.04)
is inside the protein. The signal from Ile149 (green) was broadened upon
PIM2 binding.
TR-NOESY: 2D 1H,1H NOESY spectra were collected in a 700 MHz
spectrometer equipped with TCI probe (probe temperature, 108C).
The residual HDO signal was suppressed by using WATERGATE
pulse sequence with a 3–9–19 pulse train. NOESY spectra of PIM1
glycan 1 (300 mm) in PBS (500 mL, 99% D2O, pH 7.4) were collected
in the presence of ZG16p (65 mm) or ZG16p-D151N (60 mm). The
data were collected as 1024 (F2)”256 (F1) data points with 32
scans (mixing times: 150, 250, 350, 500, and 700 ms). NOESY spec-
tra for PIM2 glycan 2 (300 mm) in PBS (250 mL, 99% D2O, pH 7.4)
were collected in the presence of ZG16p (65 mm) or ZG16p-D151N
(60 mm) by using micro-cells (Shigemi, Tokyo, Japan). The data
were collected as 2048 (F2)”256 (F1) data points (32 or 64 scans;
mixing times, 150, 250, 350, and 500 ms).
Figure 7. Binding model of PIM1 glycan to human ZG16p, created with
docking simulations. The purple dot lines indicate potential hydrogen bond
(3.0 Š).
1H,15N HSQC titration experiment: 1H,15N HSQC spectra were ob-
tained in a 500 MHz spectrometer equipped with a cryo-TXI probe
(probe temperature 258C). The spectra were collected with 1024
(F2)”256 (F1) data matrix points with either four or eight scans.
Appropriate molar equivalents of PIM1 glycan 1 or PIM2 glycan 2
(20 mm in PBS with 10% D2O, pH 6.5) were added to [15N]ZG16p
staining of lipid tags.[17e] The bacterial-and fungal-derived glycocon-
jugates and polysaccharides (positions 9–20) were arrayed at 0.03
and 0.1 ng per spot (dry weight of the samples received from com-
mercial sources; see Table 1). Microarray analysis with GST-tagged
ZG16p (GST-ZG16p and GST-ZG16p-D151N) were performed as de-
scribed previously.[17e,36] In brief, microarray slides were blocked at
room temperature for 60 min with bovine serum albumin (BSA;
3% (w/v) in PBS). GST-ZG16p and GST-ZG16p-D151N were overlaid
(100 mgmLÀ1 as neat, supplemented with 1% BSA) and incubated
for 90 min, followed by rabbit anti-GST antibody Z-5 (1:200; Santa
Cruz Biotechnology), and then biotinylated anti-rabbit IgG (1:200;
Sigma–Aldrich). Biotinylated concanavalin A (Con A; Vector Labora-
1
(0.1 mm) in PBS (pH 6.5, 10% D2O), and submitted to H,15N HSQC
1
analysis. Weighted averages of H and 15N chemical-shift changes
(Ddavg) were calculated from Equation (1):
2
2 1=2
ð1Þ
Ddavg ¼ ½ðDdHÞ þð0:2  DdNÞ Š
where DdH and DdN are the observed chemical-shift changes
tories, Peterborough, UK) was analyzed at 0.5 and 15 mgmLÀ1
.
1
(ppm) of H and 15N. The backbone amide signals of [13C,15N]ZG16p
Binding was detected with Alexa Fluor-647-labeled streptavidin
(1 mgmLÀ1; Molecular Probes/Life Technologies).
(0.2 mm) were assigned sequentially via analysis of 3D HNCA,
HN(CO)CA, HNCACB, and CBCA(CO)NH spectra obtained from an
ChemBioChem 2015, 16, 1502 – 1511
www.chembiochem.org
1509
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[5] M. Kanagawa, T. Satoh, A. Ikeda, Y. Nakano, H. Yagi, K. Kato, K. Kojima-
Aikawa, Y. Yamaguchi, Biochem. Biophys. Res. Commun. 2011, 404, 201–
205.
800 MHz spectrometer equipped with a cryo-TCI probe at 258C
(Table S1).
Modeling of PIM1-ZG16p complex: The docking of PIM1 glycan
1 to ZG16p was performed in the software package Glide 5.8[31] of
the Maestro suite (ver. 9.3.5, Schrçdinger, LLC, New York). Prior to
docking, ZG16p (PDB ID; 3APA[5]) was prepared using Protein Prep-
aration Wizard by adding hydrogen, assigning bond orders, and
optimizing bond lengths, bond angles, and torsion angles. The
minimization was performed with the force field OPLS-2005. The
ligand glycan was prepared with LigPrep and minimized using the
force field OPLS-2005. The receptor glide was generated based on
the glycerol in the crystal data. The docking study was performed
using a standard precision (SP) Glide docking with default parame-
ters. For PIM1 glycan 1, several ligand poses were provided, and
the model with the highest docking score (À4.0) agreed best with
the NMR data. All possible ligand poses were manually evaluated
[6] M. Kanagawa, Y. Liu, S. Hanashima, A. Ikeda, W. Chai, Y. Nakano, K.
Kojima-Aikawa, T. Feizi, Y. Yamaguchi, J. Biol. Chem. 2014, 289, 16954–
16965.
[7] a) K. Drickamer, Curr. Opin. Struct. Biol. 1999, 9, 585–590; b) C. G. Figdor,
Y. van Kooyk, G. J. Adema, Nat. Rev. Immunol. 2002, 2, 77–84; c) T. B. H.
Geijtenbeek, S. I. Gringhuis, Nat. Rev. Immunol. 2009, 9, 465–479.
[8] a) C. A. Wells, J. A. Salvage-Jones, X. Li, K. Hitchens, S. Butcher, R. Z.
Murray, A. G. Beckhouse, Y.- L. S. Lo, S. Manzanero, C. Cobbold, K.
Schroder, B. Ma, S. Orr, L. Stewart, D. Lebus, P. Sobieszczuk, D. A. Hume,
J. Stow, H. Blanchard, R. B. Ashman, J. Immunol. 2008, 180, 7404–7413;
b) S. Yamasaki, M. Matsumoto, O. Takeuchi, T. Matsuzawa, E. Ishikawa,
M. Sakuma, H. Tateno, J. Uno, J. Hirabayashi, Y. Mikami, K. Takeda, S.
Akira, T. Saito, Proc. Natl. Acad. Sci. USA 2009, 106, 1897–1902; c) E. Ishi-
kawa, T. Ishikawa, Y. S. Morita, K. Toyonaga, H. Yamada, O. Takeuchi, T.
Kinoshita, S. Akira, Y. Yoshikai, S. Yamasaki, J. Exp. Med. 2009, 206,
2879–2888.
1
based on the STD-NMR data and the H,15N HSQC titration results.
[9] L. de Witte, A. Nabatov, M. Pion, D. Fluitsma, M. A. W. P. de Jong, T. de
Gruijl, V. Piguet, Y. van Kooyk, T. B. Geijtenbeek, Nat. Med. 2007, 13,
367–371.
[10] a) H. Tateno, K. Ohnishi, R. Yabe, N. Hayatsu, T. Sato, M. Takeya, H. Nari-
matsu, J. Hirabayashi, J. Biol. Chem. 2010, 285, 6390–6400; b) H. Fein-
berg, M. E. Taylor, N. Razi, R. McBride, Y. A. Knirel, S. A. Graham, K. Drick-
amer, W. I. Weis, J. Mol. Biol. 2011, 405, 1027–1039.
[11] a) R. A. Childs, K. Drickamer, T. Kawasaki, S. Thiel, T. Mizuochi, T. Feizi,
Biochem. J. 1989, 262, 131–138; b) K. Takahashi, R. A. B. Ezekowitz, Clin.
Infect. Dis. 2005, 41, S440-S444.
[12] E. Laurine, X. Manival, C. Montgelard, C. Bideau, J.-L. BergØ-Lefranc, M.
Érard, J.-M. Verdier, Biochim. Biophys. Acta Gene Struct. Expression 2005,
1727, 177–187.
[13] a) H. L. Cash, C. V. Whitham, C. L. Behrendt, L. V. Hooper, Science 2006,
313, 1126–1130; b) R. E. Lehotzky, C. L. Partch, S. Mukherjee, H. L. Cash,
W. E. Goldman, K. H. Gardner, L. V. Hooper, Proc. Natl. Acad. Sci. USA
2010, 107, 7722–7727.
Acknowledgements
This work was supported in part by funding from RIKEN-Max
Planck Joint Research Center, the UK Research Council Basic Tech-
nology Initiative Glycoarrays (GRS/79268) and Translational
Grant (EP/G037604/1), the Wellcome Trust biomedical resource
grants WT093378MA and WT099197MA (T.F.), Grant-in-Aid for
Young Scientists (B) (24710257) and for Scientific Research (C)
(25460054) from the Japan Society for the Promotion of Science
(S.H., Y.Y.), and by a SUNBOR GRANT from the Suntory Founda-
tion of Life Sciences (S.H.). We acknowledge BEI Resources (NIAID,
NIH) for the Mycobacterial glycoconjugates. We thank members
of the Glycosciences Laboratory for their collaboration in the
establishment of the neoglycolipid-based microarray system. We
thank Dr. Yutaka Muto (RIKEN Yokohama) for his help in acquir-
ing and processing 3D NMR data. We thank Dr. Yukishige Ito
(RIKEN, ERATO) and Prof. Osamu Kanie (ERATO, Tokai University)
for enabling us to use the cryoprobe-equipped NMR spectrometer.
We thank Dr. Mayumi Kanagawa (RIKEN) for her efforts in trying
to co-crystallize ZG16p and PIMs. We acknowledge Masaki Kato
(RIKEN) and RIKEN Integrated Cluster of Clusters (RICC) facility for
the trial of MD simulations. We thank Noriko Tanaka (RIKEN) for
secretarial assistance and Dr. Yu-Hsuan Tsai and Dr. Ivan Viloti-
jevic for valuable discussions regarding the synthesis of PIM1 and
2. We thank Drs. Iria Uhia and Brian D. Robertson (Imperial Col-
lege London) for specialist opinions on ZG16p-mycobacteria in-
teractions.
[14] J. Iovanna, J. M. Frigerio, N. Dusetti, F. Ramare, P. Raibaud, J. C. Dagorn,
Pancreas 1993, 8, 597–601.
[15] M.-R. Ho, Y.-C. Lou, S.-Y. Wei, S.-C. Luo, W.-C. Lin, P.-C. Lyu, C. Chen, J.
Mol. Biol. 2010, 402, 682–695.
[16] a) M. E. Guerin, J. Kordulµkovµ, P. M. Alzari, P. J. Brennan, M. Jackson, J.
Biol. Chem. 2010, 285, 33577–33583; b) A. K. Mishra, N. N. Driessen, B. J.
Appelmelk, G. S. Besra, FEMS Microbiol. Rev. 2011, 35, 1126–1157.
[17] a) T. Feizi, F. Fazio, W. Chai, C.-H. Wong, Curr. Opin. Struct. Biol. 2003, 13,
637–645; b) J. L. de Paz, T. Horlacher, P. H. Seeberger, Methods Enzymol.
2006, 415, 269–292; c) I. Shin, S. Park, M.-r. Lee, Chem. Eur. J. 2005, 11,
2894–2901; d) Y. Liu, T. Feizi, M. A. Campanero-Rhodes, R. A. Childs, Y.
Zhang, B. Mulloy, P. G. Evans, H. M. I. Osborn, D. Otto, P. R. Crocker, W.
Chai, Chem. Biol. 2007, 14, 847–859; e) Y. Liu, R. A. Childs, A. S. Palma,
M. A. Campanero-Rhodes, M. S. Stoll, W. Chai, T. Feizi, Methods Mol. Biol.
2012, 808, 117–136.
[18] M. Gilleron, V. F. J. Quesniaux, G. Puzo, J. Biol. Chem. 2003, 278, 29880–
29889.
[19] N. N. Driessen, R. Ummels, J. J. Maaskant, S. S. Gurcha, G. S. Besra, G. D.
Ainge, D. S. Larsen, G. F. Painter, C. M. J. E. Vandenbroucke-Grauls, J.
Geurtsen, B. J. Appelmelk, Infect. Immun. 2009, 77, 4538–4547.
[20] S. Boonyarattanakalin, X. Liu, M. Michieletti, B. Lepenies, P. H. Seeberger,
J. Am. Chem. Soc. 2008, 130, 16791–16799.
[21] a) P. S. Patil, S.-C. Hung, Chem. Eur. J. 2009, 15, 1091–1094; b) A. Ali,
M. R. Wenk, M. J. Lear, Tetrahedron Lett. 2009, 50, 5664–5666; c) B. S.
Dyer, J. D. Jones, G. D. Ainge, M. Denis, D. S. Larsen, G. F. Painter, J. Org.
Chem. 2007, 72, 3282–3288; d) G. D. Ainge, J. Hudson, D. S. Larsen, G. F.
Painter, G. S. Gill, J. L. Harper, Bioorg. Med. Chem. 2006, 14, 5632–5642;
e) A. Stadelmaier, R. R. Schmidt, Carbohydr. Res. 2003, 338, 2557–2569;
f) C. J. J. Elie, C. E. Dreef, R. Verduyn, G. A. Van Der Marel, J. H. Van Boom,
Tetrahedron 1989, 45, 3477–3486.
Keywords: carbohydrate microarrays · chemical synthesis ·
lectins · microarrays · NMR spectroscopy · phosphatidyl inositol
mannoside
[1] U. Cronshagen, P. Voland, H. F. Kern, Eur. J. Cell Biol. 1994, 65, 366–377.
[2] H. Tateno, R. Yabe, T. Sato, A. Shibazaki, T. Shikanai, T. Gonoi, H. Nari-
matsu, J. Hirabayashi, Glycobiology 2012, 22, 210–220.
[22] X. Liu, B. L. Stocker, P. H. Seeberger, J. Am. Chem. Soc. 2006, 128, 3638–
3648.
[23] A. Ravidà, X. Liu, L. Kovacs, P. H. Seeberger, Org. Lett. 2006, 8, 1815–
1818.
[3] R. Kleene, H. Dartsch, H.-F. Kern, Eur. J. Cell Biol. 1999, 78, 79–90.
[4] K. Kumazawa-Inoue, T. Mimura, S. Hosokawa-Tamiya, Y. Nakano, N.
Dohmae, A. Kinoshita-Toyoda, H. Toyoda, K. Kojima-Aikawa, Glycobiology
2012, 22, 258–266.
ChemBioChem 2015, 16, 1502 – 1511
www.chembiochem.org
1510
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[24] J. J. Oltvoort, C. A. A. Van Boeckel, J. H. De Koning, J. H. Van Boom, Syn-
thesis 1981, 305–308.
[32] D. M. Zajonc, G. D. Ainge, G. F. Painter, W. B. Severn, I. A. Wilson, J. Im-
munol. 2006, 177, 4577–4583.
[25] L. Knerr, X. Pannecoucke, G. Schmitt, B. Luu, Tetrahedron Lett. 1996, 37,
5123–5126.
[33] K. Schmidt, M. Schrader, H.-F. Kern, R. Kleene, J. Biol. Chem. 2001, 276,
14315–14323.
[26] a) M. Mayer, B. Meyer, J. Am. Chem. Soc. 2001, 123, 6108–6117; b) M.
Mayer, B. Meyer, Angew. Chem. Int. Ed. 1999, 38, 1784–1788; Angew.
Chem. 1999, 111, 1902–1906; c) B. Meyer, T. Peters, Angew. Chem. Int.
Ed. 2003, 42, 864–890; Angew. Chem. 2003, 115, 890–918.
[27] a) C. A. Lepre, J. M. Moore, J. W. Peng, Chem. Rev. 2004, 104, 3641–
3676; b) A. Ardµ, P. Blasco, D. Varón Silva, V. Schubert, S. AndrØ, M.
Bruix, F. J. CaÇada, H.-J. Gabius, C. Unverzagt, J. JimØnez-Barbero, J. Am.
Chem. Soc. 2013, 135, 2667–2675.
[34] K. Hase, K. Kawano, T. Nochi, G. S. Pontes, S. Fukuda, M. Ebisawa, K. Ka-
dokura, T. Tobe, Y. Fujimura, S. Kawano, A. Yabashi, S. Waguri, G. Nakato,
S. Kimura, T. Murakami, M. Iimura, K. Hamura, S.-I. Fukuoka, A. W. Lowe,
K. Itoh, H. Kiyono, H. Ohno, Nature 2009, 462, 226–230.
[35] a) T. Satoh, Y. Chen, D. Hu, S. Hanashima, K. Yamamoto, Y. Yamaguchi,
Mol. Cell 2010, 40, 905–916; b) K. Hayashi, C. Kojima, J. Biomol. NMR
2010, 48, 147–155.
[36] A. S. Palma, T. Feizi, Y. Zhang, M. S. Stoll, A. M. Lawson, E. Díaz-Rodrí-
guez, M. A. Campanero-Rhodes, J. Costa, S. Gordon, G. D. Brown, W.
Chai, J. Biol. Chem. 2006, 281, 5771–5779.
[28] C. P. J. Glaudemans, L. Lerner, G. D. Daves, Jr., P. Kovac, R. Venable, A.
Bax, Biochemistry 1990, 29, 10906–10911.
[29] A. S. Edison, F. Abildgaard, W. M. Westler, E. S. Mooberry, J. L. Markley,
Methods Enzymol. 1994, 239, 3–79.
[37] D. S. Wishart, C. G. Bigam, J. Yao, F. Abildgaard, H. J. Dyson, E. Oldfield,
J. L. Markley, B. D. Sykes, J. Biomol. NMR 1995, 6, 135–140.
[30] a) Y. Bourne, V. Zamboni, A. Barre, W. J. Peumans, E. J. M. Van Damme, P.
RougØ, Structure 1999, 7, 1473–1482; b) Y. Bourne, V. Roig-Zamboni, A.
Barre, W. J. Peumans, C. H. Astoul, E. J. M. Van Damme, P. RougØ, J. Biol.
Chem. 2004, 279, 527–533.
[31] R. A. Friesner, J. L. Banks, R. B. Murphy, T. A. Halgren, J. J. Klicic, D. T.
Mainz, M. P. Repasky, E. H. Knoll, M. Shelley, J. K. Perry, D. E. Shaw, P.
Francis, P. S. Shenkin, J. Med. Chem. 2004, 47, 1739–1749.
Manuscript received: March 13, 2015
Accepted article published: April 28, 2015
Final article published: June 11, 2015
ChemBioChem 2015, 16, 1502 – 1511
www.chembiochem.org
1511
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim