Novel O-linked methylated glycan antigens decorate secreted immunodominant glycoproteins from the intestinal nematode Heligmosomoides polygyrus

Graphical abstract


Introduction
The prominence of glycan structures in the immune recognition of parasitic helminths has been known for nearly 70 years (Campbell, 1936). Indeed, anti-carbohydrate specificities have been found to dominate the host antibody response in many different helminth infections (Omer-Ali et al., 1986;Maizels et al., 1987;Eberl et al., 2001;Kariuki et al., 2008;Hewitson et al., 2011;Paschinger et al., 2012). However, the generation of anti-glycan antibodies occurs both in susceptible hosts lacking overt antiparasite immunity (Omer-Ali et al., 1986;Eberl et al., 2001;Kariuki et al., 2008), as well as in immunised animals resistant to infection (Vervelde et al., 2003;Kariuki et al., 2008). In some instances it is possible that glycan epitopes eliciting non-protective antibodies may even block potentially protective anti-protein responses (Dunne et al., 1987). As helminth molecules become better defined at the structural level, it is likely that the contrasting roles of specific glycans will become resolved.
Indeed, as the range and complexity of helminth-associated glycans become increasingly well-characterised, it is already clear that many specific glycans and carbohydrate motifs fulfil critical and important biological roles in the host-parasite relationship (Maizels and Hewitson, 2012;Prasanphanich et al., 2013). Most importantly, they can direct and modify the development of immunity to the benefit of the parasite (van Die and Cummings, 2010;Prasanphanich et al., 2013). This occurs through glycan binding to host pattern recognition receptors, particularly lectins http://dx.doi.org/10.1016/j.ijpara.2015.10.004 0020-7519/Ó 2015 The Authors. Published by Elsevier Ltd. on behalf of Australian Society for Parasitology Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). such as C-type lectin receptors (CLRs) van Vliet et al., 2005;Saunders et al., 2009;Meevissen et al., 2012;Klaver et al., 2013) and galectins (van den Berg et al., 2004;Breuilh et al., 2007;Burton et al., 2010), which are expressed by host innate cells such as dendritic cells (DC) and macrophages. CLR-triggered signalling pathways can both cooperate with and antagonise Toll-like receptor (TLR) signalling in helminth infection (van Liempt et al., 2007;Ritter et al., 2010;van Stijn et al., 2010a;Terrazas et al., 2013). Carbohydrate-specific interactions can further promote Th2 differentiation, as shown in the example of the schistosome x-1 glycoprotein which enters cells through glycan binding to the mannose receptor, and subsequently subverting DC gene expression (Everts et al., 2012).
A well-studied helminth model system is that of the mouse intestinal nematode Heligmosomoides polygyrus, which reproduces the chronic infection pattern of both human and veterinary parasites (Reynolds et al., 2012). Infected mice mount dominant antibody responses to two distinct glycosylated antigens, termed Glycan A and Glycan B (Hewitson et al., 2011). Glycan A is an O-linked sugar present in the secretory products of adult parasites, termed H. polygyrus excretory-secretory products (HES), that are highly immunomodulatory (Grainger et al., 2010;McSorley et al., 2012McSorley et al., , 2014. Glycan A is conjugated to abundantly secreted proteins including venom allergen/Ancylostoma secreted protein-like (VAL)-1 and -2, which are members of a large multi-gene CAP-domain family (Pfam00188) expressed in many phyla including nematodes, cestodes and chordates (Gibbs et al., 2008;Cantacessi et al., 2009;Chalmers and Hoffmann, 2012). The Glycan A epitope is also expressed on the surface of both tissue-stage larvae and adult parasites (Hewitson et al., 2011(Hewitson et al., , 2013. In contrast, Glycan B is present on a heterogeneous high molecular weight component that is highly abundant in parasite somatic tissues, as well as some glycoproteins such as those released from eggs in the intestinal lumen (Hewitson et al., 2011(Hewitson et al., , 2013. To assess the potential immunological properties of parasite glycans, both as targets of the host antibody response and as potential immunomodulators, we characterised the glycan structures within HES and investigated the structures of Glycan A and Glycan B through multiple approaches including antibody binding to glycan arrays, chemical deglycosylation and MS-based structural analysis. In addition, we analysed the glycosylation of a major glycoprotein component of HES, VAL-2 that bears Glycan A. These data reveal the range of novel structures from this helminth, including methylated fucose and hexose components that form antibody targets. Additionally, experiments with purified native VAL-2 reveal that, unlike total HES, this major glycoprotein (and by implication Glycan A) is unable to down-regulate allergic lung inflammation.

Parasite material and antibodies
Adult HES material was collected as described elsewhere (Johnston et al., 2015). Production, purification and antigen specificity of anti-Hp monoclonal antibodies and generation of secondary infection immune sera were as reported previously (Hewitson et al., 2011). Native VAL-2 and VAL-3 were affinitypurified from HES using specific monoclonal antibodies (mAbs 5-S2 and 5-S1, respectively) coupled to Sepharose beads (Hewitson et al., 2011), before dialysis into PBS. ES material from adult Nippostrongylus brasiliensis (NES) was prepared as previously reported (Holland et al., 2000). Endotoxin levels as measured by the Limulus amebocyte lysate (LAL, Pierce, USA) assay were generally very low and in the range of 0.01-0.1 U/lg of protein. Silver staining was carried out as previously described (Hewitson et al., 2008).

ELISA
Antibody reactivity to HES and hydrogen fluoride (HF)-treated HES was performed as described previously (Hewitson et al., 2011). For competition ELISA, plate bound HES was first incubated for 30 min at 37°C with 250 lg/ml of unlabelled mAb before addition of 10 lg/ml of biotin-labelled mAb for 2 h at 37°C. Antibodies were biotinylated with a 20-fold molar excess of biotin using EZ-link Sulfo-NHS Biotinylation kit (Thermo Fisher Scientific, USA). Bound biotinylated antibody was detected with 1/1000 streptavidin-horseradish peroxidase (Sigma-Aldrich, USA) and developed as described by Hewitson et al. (2011). 2.3. Release and labelling of N-glycans from HES and VAL proteins HES (150 lg), VAL-2 or VAL-3 (both 10 lg) were incubated with trypsin-coupled Sepharose for 16 h at 37°C with shaking. Peptides within the supernatant were then treated with PNGase F (2 U; Roche, Germany) for 24 h at 37°C as described previously (Borloo et al., 2013). The reaction mixture was then applied to a C 18 reverse-phase (RP) column (150 mg; Chromabond, Macherey-Nagel, Germany), with (glycan-containing) flow-through (2 ml of 10% acetonitrile; ACN) and wash (2 ml of water) combined, partially lyophilised to remove ACN, then applied to carbon columns (150 mg; Carbograph SPE, Grace, USA). Carbon columns were washed with 6 ml of water and the PNGase F-released glycan pool eluted with 3 ml of 25% ACN and 3 ml of 25% ACN/0.05% trifluoroacetic acid (TFA), then lyophilised. Remaining peptides and glycopeptides on the C 18 RP column were eluted with 4 ml of 30% ACN/0.1% TFA and 4 ml of 60% ACN/0.1% TFA then lyophilised. These were then resuspended in 0.1 M sodium acetate buffer pH 5 and treated with PNGase A (0.2 mU; Roche) for 24 h at 37°C. Liberated PNGase A-sensitive glycans and (glyco)-peptides were separated as above with C 18 RP and carbon columns. PNGase F and A-released glycan pools were resuspended in 50 ll of water, which was mixed with 50 ll of water with 2-aminobenzoic acid (anthranilic acid, AA) labelling mix (48 mg/ml 2-AA, 1 M 2-picoline-borane dissolved in 30% acetic acid/DMSO) followed by 2 h incubation at 65°C. Labelled glycans were then cleaned up using Biogel P-10 (BioRad, The Netherlands) in 75% ACN, washed with 80% ACN, then eluted with water.

b-Elimination and permethylation of O-glycans from HES and VAL proteins
Following PNGase F and A treatment, remaining (glyco)peptides were resuspended in 200 ll of 0.1 M NaOH/1 M NaBH 4 and incubated for 24 h at 40°C. Samples were neutralised on ice with 4 M acetic acid, lyophilised, then boric acid was removed by repeated evaporations (seven+) in 1% acetic acid in methanol. Released glycans were purified with C 18 RP and carbon columns as for N-glycans in Section 2.3. Samples were permethylated exactly as described (Borloo et al., 2013).

HF treatment of HES
HES (100 lg) was dialysed into 50 mM ammonium acetate pH 7.5 and lyophilised, then twice resuspended in water and re-lyophilised, before addition of 100 ll of HF (48% v/v;Sigma) for 48 h at 4°C as previously described (Haslam et al., 2000). Samples were then dried under nitrogen and washed twice with methanol, followed by b-elimination as described in Section 2.4.

Hydrazinolysis release of O-glycans from HES
O-Glycans were released from lyophilised HES (500 lg) using the Ludger Liberate Hydrazinolysis Glycan Release Kit (Ludger, United Kingdom) according to the manufacturer's instructions (6 h, 60°C). Sample acidification (0.1% TFA) during the release procedure prevents glycan peeling/degradation (Kozak et al., 2012). O-Glycans were then labelled with AA as described for N-glycans in Section 2.3.

MALDI-TOF(/TOF) MS and LC-MS analysis
Glycans were analysed as described (Smit et al., 2015) with an Ultraflex II MALDI-TOF mass spectrometer (Bruker Daltonics, Germany) in negative ion (AA labelled PNGase F and A-released N-glycans) or positive ion (permethylated b-elimination-released O-glycans) reflectron mode with 2,5-dihydroxy benzoic acid (DHB) (20 mg/ml in 20% ACN) as matrix. N-glycans were eluted directly onto target plates with matrix following Zip-Tip C 18 . MS spectra were annotated using Glyco-Peakfinder (http://www. glyco-peakfinder.org/). For LC-MS analysis, glycans were applied to a RP column (PepMap, 3 lm Â 75 lm Â 100 mm) using an Ultimate 3000 nano-LC system (both Dionex/LC Packings) at room temperature. The column was equilibrated with eluent A (0.1% formic acid/water) at flow rate 200 nl/min. Following sample injection, conditions were changed to 10% solvent B (0.1% formic acid/95% ACN) with a gradient to 60% B over 45 min, then isocratic elution for 10 min. The LC column was coupled to an Esquire HCT Ultra ESI-IT-MS (Bruker Daltonics) equipped with an online nanospray source in positive-ion mode. Conditions were as in Borloo et al. (2013). Ions from m/z 300-1800 and 140-2200 were registered in MS and MS/MS mode, respectively. MS/MS spectra were manually interpreted using Bruker Daltonics DataAnalysis software (Bruker Daltonics).

OVA airway allergy model and mice
Induction of airway inflammation was carried out in BALB/c female mice as previously described (McSorley et al., 2012 with HES (5 lg) or VAL-2 (2 lg) added as soluble proteins to a suspension of alum-precipitated ovalbumin (OVA), and injected i.p. in PBS on days 0 and 14. Heat inactivation of proteins was carried out at 100°C for 20 min. At 28, 29 and 30 days, mice were challenged with ovalbumin in the airways and at day 31 lung tissue and bronchioalveolar lavage (BAL) cells were analysed by flow cytometry for CD45.2 + CD11b + SiglecF + SS hi eosinophil infiltration as a marker of allergic inflammation (McSorley et al., 2012). All animal protocols adhered to the guidelines of the UK Home Office, complied with the Animals (Scientific Procedures) Act 1986, were approved by the University of Edinburgh Ethical Review Committee, and were performed under the authority of the UK Home Office Project Licence number 60/410.

Glycan array analysis indicates that Glycan A is a novel structure, whilst Glycan B is related to sulphated glycosaminoglycan molecules
Infection with H. polygyrus elicits an immunodominant antibody response against two glycan targets defined as Glycans A and B (Hewitson et al., 2011). Glycan A is an O-linked sugar coupled to multiple carrier proteins, most prominently members of the CAP superfamily of glycoproteins (Pfam00188) which includes several VAL antigens. Glycan B is present on high molecular weight species which migrate diffusely on SDS-PAGE as well as a $65-kDa molecule distinct from the VAL proteins (Hewitson et al., 2011).
Panels of mAbs directed against Glycans A and B were used to screen an extensive glycan array (Consortium for Functional Glycomics (CFG), v5.1) containing 610 native and synthetic sugars, predominantly related to mammalian glycosylation. Similar screening has previously been performed with polyclonal antisera against the related parasites Haemonchus contortus (van Stijn et al., 2010b) and Trichinella spiralis (Aranzamendi et al., 2011), as well as mAbs to tumour-associated epitopes (Noble et al., 2013;Chua et al., 2015). First, three IgM anti-Glycan A mAbs (mAbs 13.1, 3-42 and 3-55) were selected as showing the highest binding affinity to HES by ELISA from amongst 12 available antibodies (Supplementary Fig. S1). These mAbs, however, did not bind to any of the glycans on the array (data not shown). Anti-Glycan A mAbs also failed to react with ES material obtained from the closelyrelated rat nematode, N. brasiliensis (data not shown), suggesting that Glycan A is a novel structure specific to H. polygyrus.

Mass spectrometric characterisation of ES glycans
Previously, we showed that both Glycans A and B are predominantly conjugated through O-rather than N-linked glycosylation, and that antibodies to these glycans dominate the humoral response of infected mice (Hewitson et al., 2011). In preparing HES glycoproteins for MS analysis, therefore, we first liberated all N-linked sugars through sequential treatment with PNGase F and PNGase A, prior to reductive b-elimination to release O-linked structures. This procedure also offered the opportunity to characterise N-linked structures through MALDI-TOF-MS. We thereby identified typical N-glycan structures released both by PNGase F ( Fig. 2A) (often with core a1,6-linked fucose) and, to a very limited extent, by PNGase A (with additional core a1,3-linked fucose (not shown)). N-Glycan structures extended with multiple hexose and N-acetylhexosamine residues were detected, some of which appear to contain a phosphorylcholine (PC) substitution as previously suggested to be present in HES (Hewitson et al., 2011). Closer inspection of the MALDI-TOF-MS revealed several sets of peaks with mass differences of 14 Da, indicating the possible occurrence of methylated glycans. These sets of glycan-ions with a mass difference of 14 Da were confirmed by LC-MS analysis and fragmentation of selected parent ions (e.g. m/z 691.5 [M+2H] 2+ (F 1 H 3 N 3 ) and m/z 698.7 [M+2H] 2+ (meF 1 H 3 N 3 ); F, deoxyhexose/fucose; H, hexose; N,N-acetylhexosamine; meF, methylated fucose; Fig. 2B, C) reveals that this mass difference is due to the presence of a 160 Da monosaccharide in place of the core fucose (146 Da), consistent with an O-methylated fucose residue, as previously reported in some other nematode glycans (Khoo et al., 1991;Wohlschlager et al., 2014).

Mass spectrometric characterisation of O-glycans present in HES
We then analysed total HES O-glycans released by reductive belimination. Released O-glycans were permethylated followed by purification using liquid-liquid extraction to permit analysis by MALDI-TOF-MS and LC-MS/MS, which indicated that HES contains a relatively complex pattern of at least nine small O-linked glycans with two to five monosaccharide units (Fig. 3A) (Haslam et al., 2000). MALDI-TOF-MS ascertained that HF treatment caused the complete loss of the dominant deoxyhexose-containing peak at m/z 1086.6 (F 2 H 2 N 1 ), as well as the minor deoxyhexose-containing glycan species observed at m/z 953.5 (F 1 H 1 N 2 ), 841.5 (F 1 H 3 ), and 708.4 (F 1 H 1 N 1 ), with the emergence of m/z 779.5 (H 1 N 2 ), possibly derived from m/z 953.5 (F 1 H 1 N 2 ) ( Fig. 3B) (all ions observed as [M+Na] + ). In contrast, HF treatment left the dominant peaks of m/z 942.5 (H 3 N 1 ) and 738.4 (H 2 N 1 ), the latter with increased relative intensity due to defucosylation of F 1-2 H 2 N 1. Only a single, minor signal was left of a glycan species with a deoxyhexose residue (m/z 912.5 (F 1 H 2 N 1 ) after HF treatment. Possibly, this signal remains due to incomplete removal of both fucoses from the major F 2 H 2 N 1 glycan. These observations suggest that most deoxyhexoses in HES O-glycans are a1,3-linked fucose and not a1,4/6-linked fucose (Haslam et al., 2000), although it is not clear how sensitive a1,2-linked fucose, if present, would be to HF.
To investigate whether HF sensitive fucose residues form part of antigenic motifs in HES we also tested the immunological reactivity of HES following HF treatment. Binding of anti-Glycan A mAbs is actually enhanced following HF treatment (Fig. 3C), suggesting that the target epitope may be one of the HFresistant structures described. In contrast, binding of anti-Glycan B mAb 9.1.3 which targets sulphated LDN (Fig. 1) was completely ablated (Fig. 3D), consistent with the reported effects of HF to remove side chains from GAGs such as chondroitin sulphate (Olson et al., 1985). In addition, we tested binding of an antibody to PC, a common modification of nematode sugars (Houston and Harnett, 2004), using the mAb Bp-1 (Sutanto et al., 1985), and confirmed that whilst this epitope is present in HES, it is also completely removed by HF (Fig. 3E). The PCcontaining target of mAb Bp-1 and the sulphate-substituted LDN Glycan B motif are apparently not present or present in only undetectable relative amounts amongst the HES O-glycans detected in our MS analyses.

Structure and antigenicity of hydrazinolysis-released, nonpermethylated, HES O-glycans
To be able to further characterise the O-glycan target of anti-Glycan A mAb we next employed a recently described approach to determining antigenic glycans, in which they are isolated, labelled, fractionated and printed onto solid-phase arrays for probing with antibody reagents (van Diepen et al., 2012b. O-glycans liberated by b-elimination are unsuitable for this purpose as their reducing end is converted to a non-reactive alditol, and permethylation modifies the native structure to the extent it would likely prevent antibody binding. In addition, in view of the indications that methylated glycans are present amongst the HES N-glycans ( Fig. 2A-D), structural investigation of intact unmodified O-glycans allows the identification of methyl-substitutions that would be obscured by permethylation. O-glycans were therefore released from HES through hydrazinolysis, using conditions that favour O-glycan release, followed by N-acetylation, resulting in free glycans with an intact reducing end sugar. These were then labelled at the reducing sugar with AA, fractionated, and analysed by LC-MS/MS prior to printing on a solid support .
RP fractionation of the labelled hydrazinolysis-released O-glycan repertoire, followed by LC-MS/MS analysis of each fraction revealed a complex peak pattern with multiple glycan structures (Fig. 4A, Table 1). Together, this analysis of hydrazinolysis-released glycans revealed several key new insights. First, stoichiometric labelling at the reducing end combined with LC and fluorescence detection provides a more accurate quantification of different glycans than can be obtained from MALDI-TOF-MS, where ion suppression prevents reliable observation of the small di-and trisaccharide glycans. LC (Fig. 4A) and LC-MS/MS (Fig. 4B-D) reveal that the most abundant glycan structures released by hydrazinolysis are disaccharides, including H 1 N 1 -AA (fractions 14-17), N 2 -AA (fractions 17-19) and F 1 H 1 -AA (fractions 23-24).
Secondly, whilst there was good concordance between most abundant glycan species identified following b-elimination and hydrazinolysis (i.e. F 1 H 1 N 1 -AA, H 2 N 1 -AA, F 1 H 3 -AA, F 1 H 2 N 1 -AA, F 1 H 1 N 2 -AA; Table 1), the latter did not yield the H 3 N 1 -AA species (equivalent to b -elimination m/z 942.5 [M+Na] + ). Instead, fractions 15-16 contained a glycan with the putative structure of meH 1 H 2 -N 1 -AA, where meH represents a methylated hexose residue, based on LC-MS/MS fragment differences of 176 Da (hexose + 14) (Fig. 4B), confirming that similar to HES, N-glycans the O-glycans contain methylated species. We discount the possibility of this mass representing a hexuronic acid residue (U) as earlier analysis of permethylated b-elimination material detected H 3 N 1 -AA but no U 1 H 2 N 1 -AA structures (Fig. 3). Interestingly, several O-glycan species contain a hexose residue at the reducing end (Table 1). This is uncommon as O-glycans are normally based on HexNAc. Specific degradation during hydrazinolysis of Galb1-3GalNAc core types has been observed, leading to the formation of a reducing hexose residue in schistosome O-glycan samples ( Van Diepen et al., 2015). However, hexose-based glycans were observed also in the b-elimination-released O-glycan pool suggesting that degradation during hydrazinolysis is not their primary source.
We then probed the printed RP-LC fractions with both polyclonal and mAb reagents. Polyclonal antibodies were collected from mice rendered immune by sequential H. polygyrus infection, drug-induced clearance, and secondary challenge infection (McCoy et al., 2008;Hewitson et al., 2011), and were highly reactive to HES but not to the complex glycosylated egg antigens (SEA) from Schistosoma mansoni (data not shown). Both IgG (Fig. 4E) and IgM (data not shown) antibodies showed preferential reactivity with fraction 20 comprising putative methylhexosecontaining di-and trisaccharides; no signal was seen with naïve mouse sera (data not shown). Furthermore, we found some anti-Glycan A IgM antibodies (mAb 2-13, 2-62, 3-29 and 3-55) bound to fraction 20 ( Fig. 4F and data not shown), whilst a control IgM mAb did not (anti-DNP; Fig. 4G), indicating that methylated hexose is the target of anti-Glycan A mAb 2-13, 2-62, 3-29 and 3-55. A mAb directed against Glycan B (mAb 9.1.3) was unreactive to the array (data not shown).
Fragmentation analysis of the antibody-reactive fraction 20 by MS/MS revealed two principal ion species, m/z 591.2 (Fig. 4H) and m/z 681.6 ( Fig. 4I) [M+H] + . Both contain methylated hexoses and the latter corresponds to the m/z 738.6 [M+Na] + ion of the H 2 N 1 species observed in permethylated HF-treated HES which retained antibody binding (Fig. 3B). The meH 1 H 1 N 1 O-glycan consequently represents a strong candidate for the identity of Glycan A.

Marked heterogeneity in O-glycans on the VAL-2 antigen
To complement our analysis of total HES O-glycans, we also analysed glycosylation of a single protein component of HES, the Glycan A-bearing glycoprotein VAL-2, a member of the CAP domain protein family (Pfam00188). This was affinity purified from HES using a specific mAb, alongside the related VAL-3 protein that lacks Glycan A ( Fig. 5A; (Hewitson et al., 2011)). PNGase F-and PNGase A-treated VAL-2 glycopeptides from each protein were subjected to b-elimination and MALDI-TOF-MS. Despite VAL-2 and VAL-3 each containing a single predicted N-glycosylation site (N340GS and N143LS, respectively), in neither case did we detect PNGase F-or A-released N-glycans by MALDI-TOF-MS (data not shown). Permethylated O-glycans liberated from VAL-2 by b -elimination were readily detected (Fig. 5B), revealing similarly heterogeneous O-glycan modifications to those observed in total HES (Fig. 3A), with m/z 708.5 (F 1 H 1 N 1 ) particularly prominent and m/z 942.6 (H 3 N 1 ) also strongly represented. In all, 10 distinct structures were determined by LC-MS/MS, including one with m/z 779.6 (H 1 N 2 ) that was not observed in total HES. Other differences to total HES O-glycans included the complete absence from VAL-2 of the m/z 1086.6 (F 2 H 2 N 1 ) species, confirming this structure is not Glycan A, a conclusion already supported by the HF-sensitivity of this glycan (Fig. 3B). In contrast to VAL-2, except for signals with a very low signal-to-noise ratio representing H 2 N 1 (m/z 738.6, [M+Na] + ) and H 1 N 1 (m/z 512.3 [M+H] + ), no O-linked glycans were detected in VAL-3 (data not shown), which is consistent with VAL-3 having very few serine or threonine residues predicted to be Oglycosylated (Hewitson et al., 2011).

Glycan A is not responsible for the immunomodulatory effects of HES
HES is able to both inhibit in vitro DC activation in response to inflammatory stimuli (Segura et al., 2007;Massacand et al., 2009) and in vivo allergic lung inflammation following allergen sensitisation (McSorley et al., 2012). In both instances, these effects have been reported to be resistant to heat denaturation, suggesting a role for parasite glycans. To investigate whether (heat-stable) Glycan A was responsible for these effects, we initially determined whether native or heat-denatured Glycan A-bearing VAL-2 was able to inhibit LPS-dependent production of IL-12p70 by bonemarrow derived DCs (BMDCs). However, in contrast to an earlier report (Massacand et al., 2009) we found that the inhibitory effect of HES was completely ablated when HES was heat-treated, and that native Glycan A-bearing VAL-2 was similarly unable to modulate DC activation (data not shown). We did, however, find that in accordance with our previous publication (McSorley et al., 2012), heat-treated HES retained immunosuppressive activity in a murine allergy model. Thus, treatment of mice with HES during immunization with ovalbumin in alum adjuvant significantly inhibited eosinophilia in both BAL (Fig. 6A) and lung tissue (Fig. 6B), and this effect was replicated with heat-denatured HES. However, an equivalent treatment regime with purified VAL-2 (native or heat-denatured) failed to reduce eosinophil numbers. Together, this suggests that whilst parasite glycans may be able to inhibit alum-dependent sensitisation to allergic stimuli, this is not due to either Glycan A or indeed any of the other O-glycans linked to VAL-2.

Discussion
Glycans play a critically important part in the ability of parasites to interact with their chosen host, and in the manner in which the host immune response to infection unfolds (Nyame et al., 2004;Hokke et al., 2007;Harn et al., 2009;van Die and Cummings, 2010;Prasanphanich et al., 2013). The murine intestinal nematode H. polygyrus is a widely studied model of chronic helminth infection in which host immunity is profoundly down-modulated by a range of parasite secreted products, termed HES (Hewitson et al., 2009;Reynolds et al., 2012). We now show that the HES molecules include a wide range of glycan formations including several methylhexose and methylfucose conjugates that represent novel structures, and two glycosylated targets recognised by mAbs generated in infected mice. The host immune response to H. polygyrus is dominated by antibodies to glycan specificities in the secreted products, HES (Hewitson et al., 2011). Whilst class-switched IgG1 antibodies are essential mediators of immunity to gut nematodes (McCoy et al., 2008;Liu et al., 2010;Hewitson et al., 2015), the anti-glycan response (comprising both IgM and class-switched antibody isotypes) appears non-protective (Hewitson et al., 2011). Glycan immunogenicity in the absence of overt antiparasite effects is consistent with glycan ''gimmickry", whereby (potentially antigenic) helminth sugars actively manipulate the host immune response to prevent immunity (van Die and Cummings, 2010;Prasanphanich et al., 2013). Because of this, in characterising the overall glycan repertoire within HES, we also focussed on two main antigenic sugars in H. polygyrus infection, Glycans A and B.
The global glycan profile of HES reveals extensive and varied O-glycosylation, together with relatively restricted forms of N-glycosylation. Nevertheless, our studies provide evidence for methylation of both O-and N-linked glycans. One of the major sugars released from HES following b-elimination is a tri-substituted reducing HexNAc with three linked hexoses (H 3 N 1 following sample permethylation). However, this is not detected in the (non-permethylated) hydrazine-released glycan pool, and instead a HexNAc with two bound hexoses and a 176 Da molecular species is seen. Whilst this could represent either a methylated hexose or a hexuronic acid, we view the latter as unlikely since we failed to detect a permethylated hexuronic acid in the b-elimination material. Additionally, the presence of methylated fucoses in the enzymatically released N-glycans (as well as in the hydrazinolysis released O-glycans) provides further evidence for this modification Glycan methylation has previously been found in bacteria, fungi and plants (Staudacher, 2012), as well as both parasitic (Khoo et al., 1991) and free-living (Guérardel et al., 2001;Haslam and Dell, 2003;Cipollo et al., 2005) nematodes, although multiple methylation of a single monosaccharide as found in H. polygyrus has only previously been reported in the deep sea annelid worm Alvinella pompejana (Talmont and Fournet, 1991).
Our data suggest that Glycan A represents a methylated di-or trisaccharide where the methyl group is present on a hexose residue. Glycan A is an immunodominant O-linked glycan coupled to Table 1 Comparison of O-glycans released from Heligmosomoides polygyrus excretory-secretory products by hydrazinolysis and reductive b-elimination. O-glycans were liberated from Heligmosomoides polygyrus excretory-secretory products by hydrazinolysis as detailed in Section 2.6, labelled with anthranilic acid, and fractionated by reverse-phase liquid chromatography (RP-LC  a novel O-glycan array for screening fractions. Whilst this approach extends that taken previously with N-linked and O-linked glycans and glycolipids in schistosome-infected individuals (van Diepen et al., 2012a, we were concerned that hydrazinolysis and subsequent labelling of the reducing sugar might also destroy Olinked epitopes, which encompass core residues, rather than antennary structural motifs common for N-linked glycans or larger O-glycans observed in schistosomes. Nevertheless, anti-Glycan A mAb bound to a single RP fraction (F20), as did polyclonal antibodies from immune mice. This reactive fraction contains methylated di-and tri-saccharides. Interestingly, it was previously found that methylated hexoses are also immunodominant antibody targets following infection with the nematode Toxocara canis (Schabussova et al., 2007) Our work also provides a structural basis for antigenicity of Glycan B. This second immunogenic epitope is expressed on two components in HES, a heterodispersed high molecular weight antigen and a $65 kDa molecule, neither of which stained well with silver, suggestive of low protein content (Hewitson et al., 2011). Using a mammalian glycan array we now show that anti-Glycan B antibodies bind strongly to sulphated LDN motifs. Similar binding was also observed to LDN with a2,3-linked sialic acid, a monosaccharide generally absent from helminths (Johnston et al., 2009) and not found in this study. However, LDN-like structures were not observed following b-elimination or hydrazinolysis. This, coupled with the high molecular weight diffuse character of the major Glycan B-bearing antigen, suggests that the target is a GAG-containing polysaccharide of molecular mass greater than the detection range of the MS analyses (Yamada et al., 1999;Toyoda et al., 2000).
Heligmosomoides polygyrus glycans have previously been implicated in the immunosuppressive effects of this parasite, since heat denatured HES is able to prevent allergic lung inflammation (McSorley et al., 2012) and TLR-dependent DC activation (Massacand et al., 2009). Due to the novel nature of Glycan A, we tested whether it is an immunomodulatory structure by administering the naturally glycosylated protein VAL-2 to mice undergoing allergic airway sensitisation. However, our results showed that despite its immunodominance, Glycan A is not immunomodulatory in this setting, as neither native nor heat-denatured Glycan A-bearing VAL-2 reduces lung eosinophilia. The immunogenicity of Glycan A in the absence of overt immunosuppressive ability would be consistent with it fulfilling a diversionary role in generating ineffective antibodies. Alternative biochemical separation and affinity purification approaches are now being taken to identify the heat-stable immunomodulatory components in HES, which may or may not be carbohydrate in character.
Taken together, our results establish that H. polygyrus secretes a number of O-linked glycoconjugates, with marked glycan heterogeneity evident even on a single glycoprotein. Thus, if Glycan A is elaborated by the parasite as a decoy, this may have evolved to permit other glycans to function as modulators with minimal interference from host antibodies. Functional roles for these many other glycoconjugates are yet to be explored, although a number of potential interactions with the host can be envisaged. For example, glycans from other helminth species can, through their ability to ligate host lectin receptors such as C-type lectins (CTL) and galectins expressed by innate cells (e.g. DCs and macrophages), direct the host immune response (Prasanphanich et al., 2013). Thus, LeX-containing glycoconjugates similar to those produced by schisosomes can induce regulatory B cells and macrophages (Velupillai and Harn, 1994;Atochina et al., 2001Atochina et al., , 2008Terrazas et al., 2001), and delay transplant rejection (Dutta et al., 2010) as well as the development of insulin resistance (Bhargava et al., 2012). LeX containing schistosome glycans have also been shown to have a key role in Th2 induction (Okano et al., 1999), and are required for mannose receptor-mediated internalisation of the ribonuclease omega-1, a pre-requisite for its Th2 skewing ability (Everts et al., 2012). Although these precise structures are not expressed by H. polygyrus, similar interactions through one or more of the many host pathogen pattern recognition receptors may prove to be important in the ability of this parasite to establish in the mammalian host. n.s. n.s. *** *** Fig. 6. Venom allergen/Ancylostoma secreted protein-like antigen (VAL-2) does not inhibit allergic lung inflammation. Ovalbumin in alum adjuvant (OVA) airway inflammation was induced as detailed in Section 2.9 with Heligmosomoides polygyrus excretory-secretory products (HES), heat-denatured (hd) Heligmosomoides polygyrus excretory-secretory products, native VAL-2 or hdVAL-2 delivered during sensitisation. On day 31 after first sensitisation, bronchioalveolar lavage (A) and lung tissue (B) eosinophil numbers were determined by fluorescence-activated cell sorting. Values represent the mean of four mice per group ± S.E.M. Significance was determined by ANOVA compared with the ovalbumin in alum adjuvant group, or as indicated. * P < 0.05, ** P < 0.01, *** P < 0.001, not significant (n.s.; P > 0.05).