Zwitterionic modifications of glycans, such as phosphorylcholine and phosphoethanolamine, are known from a range of prokaryotic and eukaryotic species and are recognized by mammalian antibodies and pentraxins; however, defined saccharide ligands modified with these zwitterionic moieties for high-throughput studies are lacking. In this study, we prepared and tested example mono- and disaccharides 6-substituted with either phosphorylcholine or phosphoethanolamine as bovine serum albumin neoglycoconjugates or printed in a microarray format for subsequent assessment of their binding to lectins, pentraxins, and antibodies. C-Reactive protein and anti-phosphorylcholine antibodies bound specifically to ligands with phosphorylcholine, but recognition by concanavalin A was abolished or decreased as compared with that to the corresponding nonzwitterionic compounds. Furthermore, in array format, the phosphorylcholine-modified ligands were recognized by IgG and IgM in sera of either non-infected or nematode-infected dogs and pigs. Thereby, these new compounds are defined ligands which allow the assessment of glycan-bound phosphorylcholine as a target of both the innate and adaptive immune systems in mammals.
Zwitterionic modifications of glycans, such as phosphorylcholine and phosphoethanolamine, are known from a range of prokaryotic and eukaryotic species and are recognized by mammalian antibodies and pentraxins; however, defined saccharide ligands modified with these zwitterionic moieties for high-throughput studies are lacking. In this study, we prepared and tested example mono- and disaccharides 6-substituted with either phosphorylcholine or phosphoethanolamine as bovine serum albumin neoglycoconjugates or printed in a microarray format for subsequent assessment of their binding to lectins, pentraxins, and antibodies. C-Reactive protein and anti-phosphorylcholine antibodies bound specifically to ligands with phosphorylcholine, but recognition by concanavalin A was abolished or decreased as compared with that to the corresponding nonzwitterionic compounds. Furthermore, in array format, the phosphorylcholine-modified ligands were recognized by IgG and IgM in sera of either non-infected or nematode-infected dogs and pigs. Thereby, these new compounds are defined ligands which allow the assessment of glycan-bound phosphorylcholine as a target of both the innate and adaptive immune systems in mammals.
Glycans present
in cell-surface
glycoprotein and glycolipids have been shown to play a major role
in the immune cross-talk between parasites and their hosts, leading
to immunomodulating effects.[1,2] Particular examples
are N- or lipid-linked glycans modified with phosphorylcholine (PC),
which are a conserved signature of nematodes,[3−6] a phylum with many parasitic species.
These may have adopted PC-modified glycans as a means of improving
their chances of survival in the host by modulating vertebrate immune
systems, possibly via interactions involving the Toll-like receptor
TLR4.[6,7] In another parasite, the immunodominant
Ag5 antigen of Echinococcus granulosus (a cestode)
carries PC residues on its biantennary N-glycans,[8] while recently, PC has been found on the N-glycans of glycoproteins
originating from moths and moth cell lines.[9] PC is also present on various fungal glycoconjugates such as N-glycans
of Penicillium nordicum, the peptidophosphogalactomannan
of Penicillium charlesii, and glycolipids of Aspergillus fumigatus or Acremonium sp.[10−13] Finally, PC is a modification not only of annelid (earthworm) glycolipids[14] but of the lipopolysaccharides of bacteria such
as Haemophilus influenzae and Pasteurella
multocida, pilin glycans of Neisseria meningitidis, or the teichoic acid of Streptococcus pneumoniae.[15−18] In Neisseria and Haemophilus species,
on–off switching of PC expression occurs depending on whether
the bacteria reside in the upper respiratory tract (where PC is advantageous
for adhesion) or in systemic sites (where PC may be recognized by
the immune system).[19,20] The nonmethylated form of PC,
phosphoethanolamine (PE), is also a modification, e.g., of lipopolysaccharide
from N. meningitidis, glycolipids of dipteran insects,
all eukaryotic glycosylphosphatidylinositol (GPI) anchors, and N-glycans
of various Penicillium species, the sexually transmitted
parasite Trichomonas vaginalis, and the N-glycans
of honeybee royal jelly, venom, and larvae.[12,21−26] Such PC/PE-modified structures are distinct from certain immunogenic
bacterial zwitterionic glycans, such as the S. pneumoniae serotype 1 polysaccharide (Sp1) with free amino and carboxyl functions
on different monosaccharide units.[27]In bacteria, PC and PE are found attached to different hydroxyls
of various hexose, heptose, or N-acetylhexosamine
residues depending on the species, whereas in eukaryotes, zwitterionic
modifications are typically on the 6-OH of either GlcNAc or Man.[28] Phosphorylcholine–carbohydrate conjugates
are efficient inducers of hapten-specific antibodies and are specific
epitopes of IgA myeloma proteins,[29] while
both PC and PE are ligands for human “short” pentraxins;
in particular, C-reactive protein, an acute phase protein often measured
as an inflammation marker in diagnostic tests, is known to recognize
phosphorylcholine and to mediate complement activation, while serum
amyloid P binds phosphoethanolamine.[30−32]Considering the
biological context, we have set out to provide
a series of reagents for detection and characterization of immunologically
relevant binding to these phosphodiester-substituted carbohydrates.
Herein, we describe the synthesis of the respective 6-O-PC- and 6-O-PE-substituted mannoside ligands as
well as the disaccharide ligand 6-O-PC-βGlcpNAc-(1→2)-α-Manp, which correspond
to part of the structures identified, e.g., in N-glycans of various
fungi or protists (PE or PC modifications of mannose residues) or
of free-living and parasitic nematodes [PC modification of GlcNAcβ1,2Man
motifs (see Figure A)]; these conjugates were either printed in a glycan microarray
format or converted into BSA neoglycoconjugates, prior to subsequent
testing with antibodies and pentraxins. Thereby, in comparison to
conjugates such as PC directly conjugated to BSA,[33] these novel reagents better mimic the interaction of components
of the innate and adaptive immune systems to glycans carrying zwitterionic
modifications than those based on the PC or PE alone. Therefore, these
zwitterionic glycosides were tested for binding to either lectins,
pentraxins, monoclonal antibodies, or antibodies in sera of infected
animals.
Figure 1
Example zwitterionic glycan structures and Western blotting using
neoglycoconjugates. (A) Structures of the three zwitterionic neoglycoconjugates
(PE-6Man, PC-6Man, and PC-6GlcNAcβ1,2Man) are shown in the Symbol
Nomenclature for Glycans format (see also symbols) alongside example
natural phosphoethanolamine- and phosphorylcholine-modified glycans
from T. vaginalis (PE-modified N-glycan), Penicillium (PE- and PC-modified N-glycans), Trypanosoma
cruzi [PE-modified glycosylphosphatidylinositol lipid (GIPL)], Acremonium sp. no. 413 [PC-modified glycosylphosphoinositolceramide
(GPIC)], nematodes, and cestodes (PC-modified biantennary N-glycan
and nematode glycosphingolipid). (B) Five BSA neoglycoconjugates (Man,
PC-6Man, PE-6Man, GlcNAcβ1,2Man, and PC-6GlcNAcβ1,2Man;
i.e., compounds 24, 25, and 27–29, respectively) as well as the native BSA were subjected to sodium
dodecyl sulfate–polyacrylamide gel electrophoresis followed
by Western blotting with either concanavalin A (ConA), C-reactive
protein (CRP), TEPC15, or serum amyloid P (SAP). Coomassie Blue staining
and MALDI-TOF MS data for these conjugates are shown in the Supporting Information (page S41).
Example zwitterionic glycan structures and Western blotting using
neoglycoconjugates. (A) Structures of the three zwitterionic neoglycoconjugates
(PE-6Man, PC-6Man, and PC-6GlcNAcβ1,2Man) are shown in the Symbol
Nomenclature for Glycans format (see also symbols) alongside example
natural phosphoethanolamine- and phosphorylcholine-modified glycans
from T. vaginalis (PE-modified N-glycan), Penicillium (PE- and PC-modified N-glycans), Trypanosoma
cruzi [PE-modified glycosylphosphatidylinositol lipid (GIPL)], Acremonium sp. no. 413 [PC-modified glycosylphosphoinositolceramide
(GPIC)], nematodes, and cestodes (PC-modified biantennary N-glycan
and nematode glycosphingolipid). (B) Five BSA neoglycoconjugates (Man,
PC-6Man, PE-6Man, GlcNAcβ1,2Man, and PC-6GlcNAcβ1,2Man;
i.e., compounds 24, 25, and 27–29, respectively) as well as the native BSA were subjected to sodium
dodecyl sulfate–polyacrylamide gel electrophoresis followed
by Western blotting with either concanavalin A (ConA), C-reactive
protein (CRP), TEPC15, or serum amyloid P (SAP). Coomassie Blue staining
and MALDI-TOF MS data for these conjugates are shown in the Supporting Information (page S41).
Results and Discussion
Preparation of the PC- and PE-Mannoside Ligands
and Conjugates
To test the interactions of proteins with
zwitterionically modified
saccharides, the initial focus was on mimicking PC/PE-Man motifs of
fungal and trichomonad glycans. These mannoside ligands were equipped
with a 2-(2-azidoethoxy)ethyl spacer group, suitable for coupling
to proteins and solid surfaces after formation of the corresponding
ω-amino group. The known[34] mannoside 1 was first deprotected via Zemplén transesterification
to give 2, followed by reduction of the azido group to
provide the nonphosphorylated glycoside 3 as the control
ligand in 96% yield (Scheme ). To address position 6 for selective phosphorylation, tetraol 2 was treated with tBDPSCl/imidazole in DMF
to afford the 6-O-silyl ether derivative 4 in 93% yield. Subsequent benzylation with benzyl bromide/NaH in
DMF gave the tri-O-benzyl derivative 5 in 80% yield followed by fluoride treatment to produce primary alcohol 6 in 87% yield. Several approaches have been reported for
the introduction of phosphorylcholine residues into carbohydrates,
capitalizing on phosphoramidite-based coupling reactions or acylation
with POCl3 followed by reaction with choline tosylate.[35−39] In addition, PC groups have been generated via reaction of a primary
alcohol with 2-chloro-2-oxo-1,3,2-dioxaphospholane followed by ring
opening with dry trimethylamine or via formation of a 2-bromoethyl
phosphate with subsequent nucleophilic substitution with aqueous Me3N.[40−43]
Scheme 1
Synthesis of Spacer Mannosides
Reagents and conditions:
(a)
0.1 M NaOMe, MeOH, room temperature (rt), 2 h, 89%; (b) 10% Pd-C,
H2, MeOH, 12 h, rt, 96% for 3, 10% Pd-C, H2, 0.06 M HCl in MeOH, 12 h, rt, 95% for 8; (c) tBDPSCl, imidazole, DMF, 93%; (d) BnBr, NaH, DMF, 12 h,
rt, 80%; (e) 1 M Bu4NF, THF, 12 h, rt, 87%; (f) POCl3, Et3N, 3 Å molecular sieves, DCM, then choline
tosylate, pyr, 12 h, rt, 75%; (g) (iPr)2NP(OBn)O(CH2)2NHBoc, 1H-tetrazole,
MeCN, 3 h, rt, then mCPBA, DCM, −80 °C
to rt, 1 h, 64%; (h) 10% Pd-C, MeOH, 12 h, rt, 92%.
Synthesis of Spacer Mannosides
Reagents and conditions:
(a)
0.1 M NaOMe, MeOH, room temperature (rt), 2 h, 89%; (b) 10% Pd-C,
H2, MeOH, 12 h, rt, 96% for 3, 10% Pd-C, H2, 0.06 M HCl in MeOH, 12 h, rt, 95% for 8; (c) tBDPSCl, imidazole, DMF, 93%; (d) BnBr, NaH, DMF, 12 h,
rt, 80%; (e) 1 M Bu4NF, THF, 12 h, rt, 87%; (f) POCl3, Et3N, 3 Å molecular sieves, DCM, then choline
tosylate, pyr, 12 h, rt, 75%; (g) (iPr)2NP(OBn)O(CH2)2NHBoc, 1H-tetrazole,
MeCN, 3 h, rt, then mCPBA, DCM, −80 °C
to rt, 1 h, 64%; (h) 10% Pd-C, MeOH, 12 h, rt, 92%.In this study, we employed activation of 6 with POCl3 followed by coupling with dry choline tosylate
in pyridine;
this proved to be straightforward and furnished the zwitterionic phosphodiester 7 in 75% yield. Hydrogenation of 7 had to be
performed under strongly acidic conditions to completely remove the
benzyl groups and to reduce the terminal azido group. By this approach,
the target ligand 8 was obtained in 95% yield.For the synthesis of the phosphoethanolamine (2-aminoethylphosphate)
derivative, Boc protection of the 2-aminoethyl group was required
to have access to the terminal amino spacer for conjugation applications,
without touching the aminoethylphosphate moiety.[44] The phosphodiester derivative 9 was obtained
from the primary alcohol 6 using coupling with benzyl-(2-tert-butoxycarbonylaminoethyl)-N,N-isopropyl-phosphorodiamidite
reagent followed by oxidation with mCPBA. Compound 9 was thus obtained as an ∼3:2 diastereoisomeric mixture
in 64% yield. Hydrogenation of 9 with 10% Pd-C furnished
the target spacer ligand 10 in 92% yield suitable for
ensuing conjugation reactions.
Synthesis of GlcNAc-(1→2)-Man
Disaccharide Ligands and
Conjugates
To test the binding of antibodies or pentraxins
to a common motif of nematode and cestode N-glycans, a PC-GlcNAcβ1,2Man
ligand was prepared. For the preparation of the underlying β-d-GlcpNAc-(1→2)-α-d-Manp disaccharide, the 2-O-acetyl-protected
mannosyl trichloroacetimidate donor 11 was prepared according
to the literature.[45] Reaction of 11 with 2-(2-azidoethoxy)-ethanol in the presence of catalytic
TMS-triflate and 4 Å molecular sieves in dichloromethane at −40
°C then gave an excellent yield (96%) of α-glycoside 12 (Scheme ).
Scheme 2
Synthesis of Disaccharide Ligands
Reagents and conditions: (a)
TMSOTf (0.05 equiv), DCM, 4 Å molecular sieves, −40 °C,
5 min, 96%; (b) 0.1 M NaOMe, MeOH, rt, 7 h, 84% for 13 or K2CO3, MeOH, rt, 17 h, 98% for 13, 85% for 17, 98% for 21; (c) TMSOTf (0.077
equiv), DCM, 4 Å molecular sieves, −40 °C for 6.5
h, 4 °C for 12 h, 53%; (d) 1 M LiOH, THF; (e) Ac2O,
DMAP, pyr, 15 h, rt, 75% (two steps for 16); (f) tBDPSCl, DIPEA, DMF, then Ac2O, DMAP, pyr, 2
h, rt, 80%; (g) AcOH, DMF, then 1 M TBAF, THF, 15 h, rt, 62%; (h)
2-cyanoethyl-N,N,N′,N′-tetraisopropylphosphorodiamidite,
1H-tetrazole, DCM, 4 Å molecular sieves, 30
min, then choline tosylate, 100 min, then mCBPA,
2 h, 49%; (i) 10% Pd-C, H2, THF/water, 1% AcOH, 2 daus,
rt, 64% for 22, 59% for 23 under microflow
conditions.
Synthesis of Disaccharide Ligands
Reagents and conditions: (a)
TMSOTf (0.05 equiv), DCM, 4 Å molecular sieves, −40 °C,
5 min, 96%; (b) 0.1 M NaOMe, MeOH, rt, 7 h, 84% for 13 or K2CO3, MeOH, rt, 17 h, 98% for 13, 85% for 17, 98% for 21; (c) TMSOTf (0.077
equiv), DCM, 4 Å molecular sieves, −40 °C for 6.5
h, 4 °C for 12 h, 53%; (d) 1 M LiOH, THF; (e) Ac2O,
DMAP, pyr, 15 h, rt, 75% (two steps for 16); (f) tBDPSCl, DIPEA, DMF, then Ac2O, DMAP, pyr, 2
h, rt, 80%; (g) AcOH, DMF, then 1 M TBAF, THF, 15 h, rt, 62%; (h)
2-cyanoethyl-N,N,N′,N′-tetraisopropylphosphorodiamidite,
1H-tetrazole, DCM, 4 Å molecular sieves, 30
min, then choline tosylate, 100 min, then mCBPA,
2 h, 49%; (i) 10% Pd-C, H2, THF/water, 1% AcOH, 2 daus,
rt, 64% for 22, 59% for 23 under microflow
conditions.Deprotection of the 2-O-acetyl group under Zemplén
conditions afforded the glycosyl acceptor 13 in 84% yield.
Better yields (98%) were obtained using K2CO3 as the base in methanol. Coupling of the mannoside acceptor 13 with known[46−48]N-Troc-protected trichloroacetimidate
donor 14 according to Schmidt[49] was carried out at −40 °C in the presence of TMS-triflate
as the promoter. The Troc group provided the neighboring group participation
to produce the β-linked disaccharide in 53% yield. The β-anomeric
configuration of the GlcN residue of the (1→2)-linked disaccharide 15 was confirmed by the value of the homonuclear coupling
constant J1′,2′ (8.3 Hz).
Next, the Troc group had to be exchanged for the N-acetyl group without affecting the spacer-terminating azido function.
Exchange of the protecting group[50,51] was achieved
in two steps by first removing all acetyl groups via alkaline hydrolysis
with LiOH followed by reacetylation under conventional conditions
to produce the N,O-acetylated disaccharide derivative 16 in 75% yield. The ester-linked acetates were subsequently
removed by treatment with K2CO3 in methanol
to give 17 in 85% yield, suitable for global deprotection
to give disaccharide 23 as well as introduction of the
6-O-phosphorylcholine moiety. For the regioselective
introduction of the 6-O-phosphorylcholine appendage,
the primary alcohol group was protected as a temporary tert-butyldiphenylsilyl ether followed by acetylation to afford 18 in 80% yield.Removal of the tBDPS
group was effected by treatment
with TBAF but was accompanied by minor 4′-O-acetyl migration to position 6′. Separation of the latter
regioisomer from 19 was achieved by chromatography and
eventually gave the pure primary alcohol 19 in 62% isolated
yield. In contrast to the PC-modified monosaccharide ligand 8, the introduction of the phosphorylcholine group using POCl3 and choline tosylate could not be accomplished. Therefore,
sequential coupling of the alcohol 19 was performed with
2-cyanoethyl-N,N,N′,N′-tetraisopropylphosphorodiamidite
in dichloromethane in the presence of 1H-tetrazole,
followed by reaction of the intermediate phosphomonoamidite with choline
tosylate; subsequent oxidation with mCPBA was successful
and gave the phosphodiester derivative 20 in 49% isolated
yield after alkaline cleavage of the cyanoethyl protecting group.
Removal of the protecting groups was effected by alkaline cleavage
of the 3′- and 4′-O-acetyl groups using
K2CO3 in MeOH to afford 21 in near
theoretical yield, followed by hydrogenolysis of the benzyl groups
and reduction of the terminal azide. Hydrogenation reactions in methanol
or ethanol solutions, however, resulted in the simultaneous formation
of reductive amination products, which could not be separated from 22. This side reaction has been noted in the literature.[52] Finally, the hydrogenation was carried out in
a THF/water mixed solvent containing 1% acetic acid. Purification
of the product was achieved via immobilization of 22 to
a cation-exchange resin, followed first by elution with water and
eventually by elution with aqueous ammonia to release the product
in 64% yield. As a control, the disaccharide ligand 23 was obtained in 59% yield by hydrogenation of 17 on
Pd-C under microflow conditions in a THF/water mixture containing
1% acetic acid.The structures of the PE- and PC-substituted
ligands were fully
confirmed by assignment of the NMR spectra, which showed 31P/13C heteronuclear spin coupling interactions of C-6
and C-5 of the mannoside derivatives 8 and 10 as well as for C-6 and C-5 of the GlcNAc unit in the disaccharide 22. Additional coupling interactions were also observed for
the methylene groups involved in the phosphodiester linkage.
Synthesis
of Neoglycoconjugates
Conjugation of ligands 3, 8, 10, 22, and 23 was achieved via in situ activation as
an isothiocyanate derivative by reaction with thiophosgene followed
by reaction with bovine serum albumin (Scheme ).[53] The neoglycoconjugates 24–29 were obtained by exhaustive dialysis. Acidic
cleavage of the Boc group was first elaborated for the model compound 10. Cleavage of the Boc group using aqueous 1.2 mM TFA at
room temperature led to a complete removal of the Boc group within
the 10 min reaction time. The reaction was continued for 15 h, and
TLC monitoring confirmed that the PE group was still intact. Milder
conditions (0.35 mM TFA, 30 min reaction time) were then chosen for
hydrolyzing the Boc group of the ligands in BSA conjugate 26 to give 27. The ligand:protein ratios of the glycoconjugates
were assessed by MALDI-TOF MS (see the Supporting Information, page S41) and gave 11.9 for 24, 13.1
for 25, 6.9 for 26, 7.2 for 28, and 3.4 for 29. Cleavage of the Boc group of conjugate 26 to give the final PE conjugate 27 was supported
by MALDI-TOF MS data, indicating a shift of the average molecular
mass from 70.2 to 70.0 kDa.
Scheme 3
Conjugation with Bovine Serum Albumin
(a) CSCl2, CHCl3, 0.1 M NaHCO3, 3 h, rt; (b) BSA, 0.3 M NaCl, 0.1
M NaHCO3; (c) 0.35 mM TFA, 30 min, rt.
Conjugation with Bovine Serum Albumin
(a) CSCl2, CHCl3, 0.1 M NaHCO3, 3 h, rt; (b) BSA, 0.3 M NaCl, 0.1
M NaHCO3; (c) 0.35 mM TFA, 30 min, rt.
Western Blotting and Microarray Experiments
The BSA
neoglycoconjugates were then tested for binding to a monoclonal antibody
(TEPC15) known for its ability to bind (i) Pneumococcus C-polysaccharide as well as various other phosphorylcholine-containing
glycoconjugates[29] and (ii) the human pentraxins
C-reactive protein and serum amyloid P,[30−32] which recognize phosphorylcholine
and/or phosphoethanolamine (Figure B). TEPC15 bound both PC-Man and PC-GlcNAcβ1,2Man,
but not the Man-, PE-Man-, and GlcNAcβ1,2Man-BSA conjugates.
On the other hand, C-reactive protein recognized best the PC-GlcNAcβ1,2Man
conjugate, but the PC-Man-BSA to a lesser extent, while serum amyloid
P only significantly bound PE-Man-BSA, while ConA did not bind the
PC/PE-modified conjugates.For the microarray experiments, concentrations
of compounds 3, 8, 10, 22, and 23 were first normalized by quantification
using HILIC/ELSD (see the Supporting Information, pages S13 and S14) to ensure that equimolar amounts of the
2-(2-aminoethoxy)ethyl saccharides were printed before assessing their
binding to plant lectins. Similarly to the blotting data, it could
be shown that the interaction of concanavalin A with Man was abolished
or that with GlcNAcβ1,2Man decreased when the Man or GlcNAc
was substituted with either PE or PC, respectively (Figure A). No binding to five other
tested lectins was observed [AAL, WGA, RCA-I, GSL-II, and LCA (data
not shown)].
Figure 2
Glycan array data for binding to concanavalin A, anti-phosphorylcholine
antibodies, and human C-reactive protein. (A) Array data for binding
of the biotinylated lectin concanavalin A (10 μg/mL; detection
with anti-biotin) to synthetic compounds 3 (Man), 8 (PC-Man), 10 (in deprotected form; PE-Man) 22 (PC-GlcNAcβ1,2Man), and 23 (GlcNAcβ1,2Man);
previously synthesized aminooxy conjugates of chitobiose (Chito-L)
and LacdiNAc (LDN-L) as well as 2-amino-N-(2-aminoethyl)-benzamide-labeled
blood group A (BGA-AEAB) were used as controls (40 fmol per spot).
ConA strongly recognizes the mannose compound; however, this interaction
is completely blocked by the zwitterionic modifications phosphoethanolamine
(PE) or phosphorylcholine (PC), whereas the level of binding to GlcNAcβ1,2Man
was reduced by ∼60% if the PC modification was present. No
binding to Con A was detected for chitobiose, LacdiNAc, or blood group
A; neither AAL, WGA, RCA-I, GSL-II, nor LCA recognized any of the
compounds (data not shown). (B) The mouse myeloma IgA anti-PC TEPC
15 antibody (detection with fluorescent anti-mouse IgA) binds specifically
to PC-modified compounds. (C) A similar binding pattern could be detected
for the anti-PC 6G3 antibody (detection with fluorescent anti-mouse
IgG, which is cross-reactive to rat antibodies), but the binding was
weaker as judged by the lower fluorescence intensity. (D) Human C-reactive
protein (detected with anti-CRP, followed by fluorescent anti-mouse
IgG) binds specifically and in a calcium-dependent manner, as shown
by comparison to incubation in the presence of 5 mM EDTA, to the PC-modified
synthetic compounds. The charts represent the uncorrected mean fluorescence
value with standard deviations (n = 10 spots) as
well as a negative control (spotting buffer). No unspecific binding
to anti-biotin, anti-mouse IgA Alexa Fluor 647, or anti-CRP antibody
was detected.
Glycan array data for binding to concanavalin A, anti-phosphorylcholine
antibodies, and human C-reactive protein. (A) Array data for binding
of the biotinylated lectin concanavalin A (10 μg/mL; detection
with anti-biotin) to synthetic compounds 3 (Man), 8 (PC-Man), 10 (in deprotected form; PE-Man) 22 (PC-GlcNAcβ1,2Man), and 23 (GlcNAcβ1,2Man);
previously synthesized aminooxy conjugates of chitobiose (Chito-L)
and LacdiNAc (LDN-L) as well as 2-amino-N-(2-aminoethyl)-benzamide-labeled
blood group A (BGA-AEAB) were used as controls (40 fmol per spot).
ConA strongly recognizes the mannose compound; however, this interaction
is completely blocked by the zwitterionic modifications phosphoethanolamine
(PE) or phosphorylcholine (PC), whereas the level of binding to GlcNAcβ1,2Man
was reduced by ∼60% if the PC modification was present. No
binding to Con A was detected for chitobiose, LacdiNAc, or blood group
A; neither AAL, WGA, RCA-I, GSL-II, nor LCA recognized any of the
compounds (data not shown). (B) The mouse myeloma IgA anti-PC TEPC
15 antibody (detection with fluorescent anti-mouse IgA) binds specifically
to PC-modified compounds. (C) A similar binding pattern could be detected
for the anti-PC 6G3 antibody (detection with fluorescent anti-mouse
IgG, which is cross-reactive to rat antibodies), but the binding was
weaker as judged by the lower fluorescence intensity. (D) Human C-reactive
protein (detected with anti-CRP, followed by fluorescent anti-mouse
IgG) binds specifically and in a calcium-dependent manner, as shown
by comparison to incubation in the presence of 5 mM EDTA, to the PC-modified
synthetic compounds. The charts represent the uncorrected mean fluorescence
value with standard deviations (n = 10 spots) as
well as a negative control (spotting buffer). No unspecific binding
to anti-biotin, anti-mouse IgA Alexa Fluor 647, or anti-CRP antibody
was detected.The microarrays were then also
tested with human C-reactive protein[30] as
well as two monoclonal anti-phosphorylcholine
antibodies: the aforementioned murine TEPC15[29] as well as the 6G3 derived from rats infected with Trichinella.[54] Although the absolute recorded intensities
differed widely, the pattern of binding for all three was identical
in that only PC-Man and PC-GlcNAcβ1,2Man were recognized in
a specific manner, whereby the binding to C-reactive protein as detected
by anti-CRP binding was, as expected, dependent on the presence of
Ca(II) ions; however, in this format, testing arrays with the combination
of serum amyloid P and anti-SAP showed nonspecific high background
values (data not shown).Next, the five immobilized 2-(2-aminoethoxy)ethyl
saccharides were
tested with sera from one control and two nematode-infected dogs (naturally
infected with Dirofilaria immitis microfilariae),
whereby both IgM and IgG reactivities were assessed (Figure A,B). As part of a previous
study, we concluded that dogs had “natural” antibodies
to chitobiose regardless of the infection status,[55] a result that appears to be reproducible. Here, we now
observe significant binding to especially PC-Man (an epitope found
in fungi) and, to a lesser extent, PC-GlcNAcβ1,2Man (an epitope
known from nematodes), even with the sera of the control dog. Interestingly,
when array experiments were performed with native N-glycans from Dirofilaria, it was the pool of glycans from which phosphorylcholine
had been removed (using hydrofluoric acid) to which binding of dog
antibodies was stronger than the native ones.[55] In contrast, one of two Ascaris-infected pigs and
two of three Oesophagostomum-infected pigs displayed
IgG and IgM reactivity with either of the phosphorylcholine-modified
compounds, with the highest fluorescence intensities recorded for
binding of IgM for one Ascaris-infected animal; interestingly,
IgG recognized only the PC-GlcNAcβ1,2Man compound, while IgM
bound both the PC-GlcNAcβ1,2Man and PC-Man conjugates (Figure C–F). In the
case of both control pigs, there is no detectable reactivity to phosphorylcholine.
Figure 3
Binding
of dog and pig IgG and IgM antibodies to synthetic glycans.
(A and B) The binding IgG and IgM antibodies in three different dog
sera (1:250 diluted; non-infected dog, Dirofilaria-infected dog I, and Dirofilaria-infected dog II)
were tested with respect to synthetic compounds. In all three dog
sera, IgG as well as IgM antibodies that strongly interact with PC-modified
compounds (8 and 22) are found; especially
sera from both dogs infected with D. immitis show
statistically significantly larger amounts of antibodies toward the
PC-Man compound but not the PC-GlcNAcMan. In all three sera, there
was no significant binding detected to mannose (Man; 3), phosphoethanolamine-mannose (PE-Man; 10), or GlcNAcβ1,2Man
(GlcNAcMan; 23). As already shown by Martini et al.,[55] dog antibodies bind the chitobiose (Chito-L)
and LacdiNAc (LDN-L) compounds. In this array also, the blood group
A antigen was included, which showed some interaction with dog IgG
as well as IgM antibodies. (C–F) Similar experiments were also
performed with sera of pigs infected with Ascaris suum or Oesophagostomum dentatum and show statistically
significant interactions of IgM and IgG from infected animals with
PC-GlcNAcβ1,2Man (22) and IgM also to PC-Man (8) as compared to non-infected animals.
Binding
of dog and pig IgG and IgM antibodies to synthetic glycans.
(A and B) The binding IgG and IgM antibodies in three different dog
sera (1:250 diluted; non-infected dog, Dirofilaria-infected dog I, and Dirofilaria-infected dog II)
were tested with respect to synthetic compounds. In all three dog
sera, IgG as well as IgM antibodies that strongly interact with PC-modified
compounds (8 and 22) are found; especially
sera from both dogs infected with D. immitis show
statistically significantly larger amounts of antibodies toward the
PC-Man compound but not the PC-GlcNAcMan. In all three sera, there
was no significant binding detected to mannose (Man; 3), phosphoethanolamine-mannose (PE-Man; 10), or GlcNAcβ1,2Man
(GlcNAcMan; 23). As already shown by Martini et al.,[55] dog antibodies bind the chitobiose (Chito-L)
and LacdiNAc (LDN-L) compounds. In this array also, the blood group
A antigen was included, which showed some interaction with dog IgG
as well as IgM antibodies. (C–F) Similar experiments were also
performed with sera of pigs infected with Ascaris suum or Oesophagostomum dentatum and show statistically
significant interactions of IgM and IgG from infected animals with
PC-GlcNAcβ1,2Man (22) and IgM also to PC-Man (8) as compared to non-infected animals.Although the differences between the two Dirofilaria-infected and one non-infected dog for binding to PC-Man are significant
(p ≤ 0.0005), the binding to PC-GlcNAcMan
for all three sera is similar; thus, it is possible that the anti-phosphorylcholine
antibodies in dog sera are “natural”, i.e., result from
exposure to environmental bacteria or fungi, and are not related to
the infection with the parasitic nematode. This could agree with the
concept that nematodes are in part “invisible” to the
immune system in some host/parasite systems,[56] despite being immunogenic in other cases.[57] It is also compatible with the aforementioned results with C-reactive
protein, as the interaction of this pentraxin with the PC on N-glycans
has been proposed not to result in activation of the complement cascade
despite binding to C1q, due to the inherent flexibility of N-glycans
as compared to bacterial polysaccharides.[58]
Conclusion
The synthesis of glycans modified with phosphorylcholine
and phosphoethanolamine is inherently challenging; there are few examples
of their chemical construction and none targeted at the production
of glycan arrays. Using a phosphoramidite-based approach or activation
with phosphorus oxychloride as previously employed to construct PC-Gal-based
glycodendrimers akin to annelid glycolipids or PE-heptosyl-modified
fragments of bacterial lipopolysaccharides, we have succeeded in synthesizing
zwitterionically substituted mono- and disaccharides. The ligands
had been equipped with spacer groups containing terminal amino groups
for ready conversion into BSA conjugates and glycan arrays. Biologically
significant is the fact that the zwitterionically modified Man- or
GlcNAcβ1,2Man-mimicking motifs of N- and lipid-linked glycans
of various lower eukaryotes were shown to interact with antibodies,
lectins, and pentraxins. Reagents of this type have potential value
as “molecular diagnostics” for parasite infections,
in vaccination trials to combat helminth infections, or as tools to
investigate the innate immune system or mechanisms of immunomodulation.
Methods
Chemical Syntheses
Descriptions of the syntheses of
the 2-(2-aminoethoxy)ethyl glycosides 3, 8, 10, 22, and 23 as well as
of the BSA conjugates 24, 25, and 27–29 are given in the Supporting Information.
Western Blotting
BSA conjugates
(24, 25, and 27–29)
were subjected to sodium
dodecyl sulfate–polyacrylamide gel electrophoresis and transferred
to a nitrocellulose membrane via semidry blotting. This membrane was
then incubated with blocking buffer [Tris-buffered saline containing
Tween 20; TTBS supplemented with 0.5% (w/v) BSA] for 1 h, prior to
incubation with either biotinylated concanavalin A (Vector), TEPC15
primary antibody (Sigma), human C-reactive protein (CRP; MPBio), or
human serum amyloid P (SAP; Merck) for 45 min. Thereafter, the membrane
was washed three times with TTBS before incubation with either alkaline
phosphatase-conjugated anti-biotin (Sigma), alkaline phosphatase-conjugated
anti-mouse IgA secondary antibody (Sigma), rabbit anti-CRP, or rabbit
anti-SAP for 45 min; in the case of the detection of pentraxins, alkaline
phosphatase- or peroxidase-conjugated anti-rabbit IgG was used. The
membrane was again washed three times with TTBS and then stained with
either SigmaFast 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium
or SigmaFast 3,3-diaminobenzidine in the dark. (All steps were performed
at room temperature; see the Supporting Information, page S42, for dilutions.)
Glycan Array-Based Lectin,
Pentraxin, and Antibody Screening
Solutions of 2-(2-aminoethoxy)ethyl
glycosides (3, 8, 10, 22, and 23;
∼1 mg mL–1) were quantified by HILIC (Merck
ZIC; 150 mm × 4.6 mm, 3.5 μm) using a Shimadzu HPLC-MS-ELSD
system and a gradient from 95% to 40% aqueous acetonitrile. ELSD peak
areas were used to quantify ligand solutions; for the calibration
curves, xylotriose, xylotetraose, and α-cyclodextrin were used
as reference materials (see the Supporting Information, page S42). Ligand solutions were then diluted with spotting
buffer [300 mM sodium phosphate (pH 7.5) and 0.005% Tween 20] to a
final concentration of 50 μM and then spotted (n = 10) by noncontact printing (Scienion Flexarrayer S1) onto NHS-derivatized
Nexterion H glass slides (Schott). After hybridization for 16 h, slides
were blocked [50 mM ethanolamine in 50 mM sodium borate (pH 9.0)]
for 1 h at room temperature and washed serially with TSM [20 mM Tris
(pH 7.4), 150 mM NaCl, 2 mM CaCl2, and 2 mM MgCl2] with 0.05% (w/v) Tween 20 (TSM wash buffer, TSMWB), TSM alone,
and water prior to drying and storage at −20 °C before
use.[59] The slides were incubated with dilutions
in TSMWB of either one control and two Dirofilaria-infected dog sera,[55]Ascaris- or Oesophagostomum-infected pig sera, biotinylated
Con A, or other lectins (Vector Laboratories), murine IgA monoclonal
TEPC15 (Sigma), monoclonal rat anti-phosphorylcholine antibody 6G3
(raised in Trichinella-infected rats; provided by
J. Appleton),[54] or natural human C-reactive
protein (MPBio) followed by the relevant secondary and/or tertiary
antibodies (see the Supporting Information for concentrations). After the first incubation, the slides were
washed serially with TSMWB (i.e., Tris-buffered saline supplemented
with CaCl2, MgCl2, and Tween), TSM, and water.
For detection of lectins or TEPC15, the directly labeled anti-biotin
FITC conjugate (Sigma), anti-mouse IgA FITC conjugate (Invitrogen),
or anti-mouse IgG AF 647 conjugate (Invitrogen) was added in binding
buffer and the slides were again washed as described above. For detection
of CRP or antibodies in sera, unconjugated murine anti-CRP (biotechne),
rabbit anti-dog, or rabbit anti-pig IgG/IgM (Sigma) was used; the
slides were washed again as described above and dried prior to adding
the tertiary antibody anti-mouse IgG AF 647 conjugate and anti-rabbit
AF 647 conjugate (Invitrogen). Slides were scanned with an Agilent
G2565CA Microarray Scanner [multiple photomultiplier tube (PMT) gain
values from 10 to 100%], and raw image files were analyzed by GenePix
7 software. The negative controls (spotting buffer or no primary reagent)
show fluorescence due to either the labels themselves or nonspecific
binding of the fluorescent secondary antibodies. As controls for lectin
and antibody binding, additional synthetic compounds were included
in the array: chitobiose linker (Chito-L), LacdiNAc linker (LDN-L),
and AEAB [2-amino-N-(2-aminoethyl)-benzamide]-labeled
blood group A tetrasaccharide (BGA-AEAB). The chitobiose and LacdiNAc
compounds were synthesized on an aminoxy-based linker as previously
described.[55,60] The fluorescence values (green
for FITC and red for Alexa Fluor 647) were used to calculate the mean
and standard deviation from all 10 spots; data were analyzed for significant
differences of the means by Welch’s t test
using the R progamming language for statistical computing, whereby
data were grouped by chemoenzymatically synthesized compounds and
test samples. Plots were visualized using the ggboxplot function of
the ggplot2 library.
Authors: Halina Mikolajek; Simon E Kolstoe; Valerie E Pye; Palma Mangione; Mark B Pepys; Stephen P Wood Journal: J Mol Recognit Date: 2011 Mar-Apr Impact factor: 2.137
Authors: N L Stults; Y C Lee; C A Hoppe; K Kawaguchi; S Kohda; I Takagahara; T Koishi; T Y Liu Journal: Anal Biochem Date: 1987-03 Impact factor: 3.365
Authors: Alba Hykollari; Barbara Eckmair; Josef Voglmeir; Chunsheng Jin; Shi Yan; Jorick Vanbeselaere; Ebrahim Razzazi-Fazeli; Iain B H Wilson; Katharina Paschinger Journal: Mol Cell Proteomics Date: 2015-10-29 Impact factor: 7.381
Authors: Rhiannon Stanton; Alba Hykollari; Barbara Eckmair; Daniel Malzl; Martin Dragosits; Dieter Palmberger; Ping Wang; Iain B H Wilson; Katharina Paschinger Journal: Biochim Biophys Acta Gen Subj Date: 2017-01-08 Impact factor: 4.117