Identification and Characterization of Keyhole Limpet Hemocyanin N-Glycans Mediating Cross-reactivity with Schistosoma mansoni*

Keyhole limpet hemocyanin (KLH) of the mollusc Megathura crenulata is known to serologically cross-react with Schistosoma mansoni glycoconjugates in a carbohydrate-dependent manner. To elucidate the structural basis for this cross-reactivity, KLH glycans were released from tryptic glycopeptides and fluorescently labeled. Cross-reacting glycans were identified using a polyclonal antiserum reacting with soluble S. mansoni egg antigens, isolated by a three-dimensional fractionation scheme and analyzed by different mass spectrometric techniques as well as linkage analysis and exoglycosidase treatment. The results revealed that cross-reacting species comprise ∼4.5% of released glycans. They all represent novel types of N-glycans with a Fuc(α1-3)GalNAc(β1-4)[Fuc(α1-3)]GlcNAc motif, which is known to occur also in schistosomal glycoconjugates. The tetrasaccharide unit is attached to the 3-linked antenna of a trimannosyl core, which can be further decorated by galactosyl residues, a xylose residue in 2-position of the central mannose and/or a fucose at the innermost N-acetylglucosamine. This study provides for the first time detailed structural data on the KLH carbohydrate entities responsible for cross-reactivity with glycoconjugates from S. mansoni.

was a generous gift of M. J. Doenhoff, School of Biological Sciences, University of Wales, Bangor, UK. Anti-SEA antibodies, coupled to NHS-activated Sepharose, were provided by T. Lehr, Institute of Biochemistry, University of Giessen. The monoclonal antibody M2D3H was produced and kindly provided by Q. Bickle, London School of Hygiene and Tropical Medicine, London University, UK (25).
Isolation of Oligosaccharides-Tryptic glycopeptides were subjected to hydrazinolysis (26,27) or treated with PNGase F as described previously (21). The resulting mixture of products was applied to a reversephase cartridge (C18ec, Macherey und Nagel, Düren, Germany). Released oligosaccharides were recovered in the flow-through and were collected. For desalting, glycans were applied to a porous graphiticcarbon cartridge (Supelclean ENVI-Carb, Supelco, Bellefonte, PA). The cartridges were washed with water, and oligosaccharides were eluted with 25% (v/v) aqueous acetonitrile.
Obtained oligosaccharide fractions were applied to a column filled with 10 ml of A. aurantia lectin (AAL)-agarose, which had been equilibrated with phosphate-buffered saline (PBS, 6.5 mM KH 2 PO 4 , 0.15 M NaCl, pH 7.4). The column was eluted with 50 ml of PBS, 30 ml of PBS containing 1 mM fucose, and 30 ml of PBS containing 50 mM fucose. Fractions of 10 ml were collected and lyophilized.
Immunoaffinity Chromatography-Anti-SEA-Sepharose (1 ml) was packed into a column of 8-mm diameter and washed with TBS (25 mM Tris/HCl, pH 7.5, 100 mM NaCl). PA-oligosaccharides were applied to the column in 150 l of TBS, followed by 1-h incubation at room temperature. After washing the column with 10 ml of TBS, bound PA-oligosaccharides were eluted with 2 ml of 100 mM triethylamine, pH 11.5, 150 mM NaCl. Both flow-through and eluate were applied separately to a 25-mg porous graphitic-carbon cartridge (Thermoquest, Kleinostheim, Germany) for desalting. The cartridge was washed with 10 ml of water, and PA-oligosaccharides were eluted with 5 ml of 25% (v/v) aqueous acetonitrile. Samples were dried in a Speed-Vac concentrator.
Matrix-assisted Laser-desorption Ionization Time-of-flight Mass Spectrometry-MALDI-MS was performed either on a Vision 2000 instrument (ThermoFinnigan, Egelsbach, Germany) equipped with a UV nitrogen laser (337 nm) or on an Ultraflex TOF/TOF mass spectrometer (Bruker Daltonik GmbH, Bremen, Germany) equipped with a LIFT-MS/MS facility. The instruments were operated in the positiveion reflectron mode using 6-aza-2-thiothymine (Sigma) as matrix (20,21). Measurements with the Vision 2000 instrument allowed the determination of average masses only, whereas isotopic resolution was achieved with the Ultraflex mass spectrometer. Fragment ion analysis by tandem mass spectrometry (MS/MS) after laser-induced dissociation (LID) was performed as detailed earlier (21).
For high energy collision-induced dissociation (CID) experiments, a 5 mg/ml solution of 2,5-dihydroxybenzoic acid (Sigma) in 30% acetonitrile in 0.1% trifluoroacetic acid was prepared. About 1-2 l of glycan sample and 0.5 l of matrix solution were applied to an 800-m hydrophilic anchor of an AnchorChip TM MALDI sample plate (Bruker) and dried by a cold air stream resulting in a typical preparation with large heterogeneous crystals at the rim and a more homogeneous thin crystal layer in the center of the spot. From this central region, exclusively sodiated glycan ions were obtained and selected for further MS/MS experiments. The CID spectra were obtained by using an Ultraflex TOF/TOF II instrument (Bruker) using the LIFT device for selection and fragmentation of the sodiated glycan ions as described before (29,30). Acceleration voltage in the ion source was 8 kV, the Timed Ion Selector was set to 0.4% (relative to parent mass), and argon was used as collision gas (ϳ4 -6 ϫ 10 Ϫ6 mbar). Resulting fragments were further accelerated in a second source by 19 kV and analyzed by a two-stage gridless reflectron. In addition, a new all solid state laser system called SmartBeam TM was used allowing a 200-Hz acquisition of high quality spectra from all matrix and preparation types comparable to the quality obtainable with N 2 lasers. Typically, 200 shots were accumulated for the parent ion signal and 1000 -2000 shots for the fragments. Compass 1.1 consisting of FlexControl 2.4, and FlexAnalysis 2.4 was used as instrument control and processing software.
Nano-LC-Electrospray Ionization-Ion Trap-Mass Spectrometry-N-Glycans were separated on a nanoscale Amide-80 column (5 m, 80 Å, 75 m ϫ 100 mm, Tosohaas, Montgomeryville, PA) as outlined previously for native oligosaccharides (31). The system was directly coupled with an Esquire HCT ESI-IT-MS (Bruker) equipped with an online nanospray source operating in the positive-ion mode. For electrospray (900 -1200 V), capillaries (360 m out diameter, 20 m inner diameter with 10-m opening) from New Objective (Cambridge, MA) were used. The solvent was evaporated at 120°C with a nitrogen stream of 6 liters/ min. Ions from m/z 50 to m/z 2500 were registered.
Enzymatic and Chemical Degradation-PA-oligosaccharides were treated with ␣-fucosidase from bovine kidney (4 milliunits/l, Roche Diagnostics) directly on the MALDI-MS target (32). The enzyme was dialyzed for 2 h against 25 mM ammonium acetate, pH 5. After measurement of the educts by MALDI-MS, the dialyzed enzyme (1 l) was added to samples on the target, and spots were analyzed again by MALDI-MS after overnight incubation at 37°C. Digestions with ␤-galactosidase, ␣-mannosidase, or ␤-N-acetylhexosaminidase from jack beans as well as ␣-galactosidase from green coffee beans were performed with 1 l of the respective dialyzed enzymes in the same manner. For chemical defucosylation, dried samples were treated with 48% HF at 4°C overnight (modified from Ref. 33). HF was removed by a stream of nitrogen. The anomeric configuration of fucosyl and galactosyl residues was determined by chromium trioxide oxidation (34).
Monosaccharide Composition and Linkage Analysis-After hydrolysis in 4 N aqueous trifluoroacetic acid at 100°C for 4 h and labeling with anthranilic acid, monosaccharides were determined by HPLC, and fluorescence detection (34,35). For linkage analysis, PA-oligosaccharides were permethylated and hydrolyzed. Partially methylated alditol acetates obtained after sodium borohydride reduction and peracetylation were analyzed by capillary GLC/MS in the chemical ionization as well electron impact ionization mode using the instrumentation and microtechniques described elsewhere (36,37).

RESULTS
Isolation of KLH Oligosaccharides Sharing Carbohydrate Epitopes with S. mansoni Glycoconjugates-KLH exhibits, in part, oligosaccharide moieties, which are serologically cross-reacting with S. mansoni carbohydrate antigens. Using ELISA as assay system, KLH glycopeptides were recognized by murine S. mansoni infection sera, the monoclonal antibody M2D3H reacting with the Fuc(␣1-3)GalNAc-epitope (23) and polyclonal antibodies raised against SEA of S. mansoni (Fig. 1). In contrast to rabbit hyperimmune serum raised against KLH, antibody binding was mostly eliminated by mild HF treatment, which results in a release of fucose residues leaving the remaining carbohydrate chains largely intact (23). In agreement with previous data (22) it may be, therefore, concluded that cross-reacting oligosaccharides comprise a fucosylated carbohydrate epitope. Furthermore, this experiment established rabbit polyclonal anti-SEA serum as a suitable tool for the identification of cross-reactive KLH glycan species.
To isolate these carbohydrates in a preparative scale for structural studies, KLH glycopeptides were subjected to hydrazinolysis or treatment with PNGase F. Released oligosaccharides were fluorescently labeled by reductive amination with 2-aminopyridine and analyzed by MALDI-MS (Fig. 2).  In agreement with previous studies, mass spectrometry revealed a highly heterogeneous mixture of different glycans comprising at least 50 different compositional species with monosaccharide compositions of Hex 0 -9 HexNAc 2-4 dHex 0 -3 Pent 0 -1 , the deoxyhexose and pentose residues of which represented exclusively fucose (20) and xylose (see below), respectively (TABLE ONE). Previous studies (20, 21) focusing mainly on the major, non-cross-reactive sugar chains of KLH have already dem-onstrated that individual compositional species are frequently representing different isomeric and/or isobaric structures (see, for example, structures of Hex 3 HexNAc 2 , Hex 3 HexNAc 2 Fuc 1 , Hex 4 HexNAc 2 Fuc 1 , and Hex 5 HexNAc 3 compounds in TABLE ONE), demonstrating again the vast heterogeneity of KLH glycosylation. Intriguingly, minor species, comprising Ͻ5% of the total glycans, could be additionally registered in the higher mass range (see Fig. 2), which contained more than one fucose and thus

Compilation of total KLH-derived PA-glycans registered by MALDI-MS
Compositions are assigned in terms of hexose (H), N-acetylhexosamine (N), deoxyhexose (fucose; F) and pentose (xylose; X). Relative occurrence of individual compositional species is roughly estimated from the respective signal intensities. Published KLH glycan structures and corresponding references are included. Compositional species serological cross-reacting with S. mansoni glycoconjugates are given in bold and italics.
represented potential candidates for serological cross-reactivity with schistosomal glycans. To identify and verify such cross-reacting oligosaccharides, an aliquot of the total KLH-derived PA-glycans was subjected in analytical scale to immunoaffinity chromatography employing immobilized polyclonal anti-SEA antibodies. Subsequent analyses by MALDI-MS revealed the presence of 15 cross-reacting species with monosaccharide compositions of Hex 2-7 HexNAc 4 Fuc 2-3 Xyl 0 -1 (Fig. 3)  For isolation and fractionation of the cross-reacting carbohydrate species PA-glycans were separated by HPLC on an aminophase column (Fig. 4A). Resulting oligosaccharide fractions were analyzed by MALDI-MS, which demonstrated that all of them still comprised heterogeneous glycan mixtures (see, for example, Fig. 4, B-E). Obviously, efficient size fractionation of these oligosaccharides by aminophase HPLC could not be achieved possibly due to the great structural variability within one type of compositional species. Serologically cross-reacting oligosaccharides with more than one fucose constituent ( Fig. 3) occurred preponderantly in fractions 12-15 (Fig. 4, B-E). Careful inspection of the highly complex MALDI-MS spectrum of fraction 11 glycans revealed no evidence for the presence of difucosylated oligosaccharides species, whereas the eluate recovered after fraction 15 was found to contain trace amounts of trifucosylated glycans with six to seven hexoses and, in part, one xylose residue that were not further analyzed. The obtained MALDI-MS spectra further demonstrated that the respective oligosaccharides represented often only minor constituents of these fractions.
To enrich these fucosylated oligosaccharides, relevant HPLC fractions 12Ϫ15 (Fig. 4A) were subjected in preparative scale to lectin affinity chromatography using immobilized A. aurantia lectin (AAL). Unbound glycans appearing in the flow-through as well as weakly and tightly bound oligosaccharide species, sequentially eluted with 1 mM and 50 mM fucose in PBS, were separately collected and analyzed by MALDI-MS. As shown in Fig were separately subjected to preparative reverse-phase HPLC for further fractionation (Fig. 6). Resulting subfractions were again analyzed by MALDI-MS. The results revealed that some of them still represented mixtures of oligosaccharides. In part, species with identical molecular   DECEMBER 9, 2005 • VOLUME 280 • NUMBER 49 masses also occurred in different fractions (Fig. 6, A-D). Hence, it has to be concluded that these glycans represent isomeric or isobaric variants.

KLH Cross-reacting Carbohydrate Epitopes
To verify that the isolated glycan fractions comprised serologically cross-reacting oligosaccharide species, aliquots of fractions 12-3, 13-1, 13-2, 14-1, 14-2, 15-1, 15-2, and 15-3 were combined and subjected to immunoaffinity chromatography using immobilized polyclonal anti-SEA antibodies. Starting glycans, unbound material obtained in the flow-through, and oligosaccharides recovered in the immunoaffinity eluate were analyzed by MALDI-MS (see supplemental Fig. S1). The results revealed a specific binding of the glycans to the affinity matrix, thus clearly demonstrating that these oligosaccharides represented, in fact, cross-reacting species.
Structural Characterization of Difucosylated Serologically Cross-reacting KLH Glycans-Recovered major PA-oligosaccharide subfractions (see Fig. 6 , TABLE TWO, and supplemental Table S1) were subjected to MALDI-MS/MS(LID) (tandem mass spectrometry after laserinduced dissociation), linkage analysis, chromium trioxide oxidation, chemical defucosylation by HF treatment, or digestion with exoglycosidases in conjunction with MALDI-MS. Selected fractions were further studied by MALDI-MS/MS(CID) (tandem mass spectrometry after collision-induced dissociation) and nano-LC-ESI-IT-MS/MS. The results revealed a number of structural parameters that were common to all glycans studied. Linkage analysis, for example, demonstrated that all of them comprised terminal fucose, terminal galactose, 3-substituted Gal-NAc, and 3,4-disubstituted GlcNAc residues (supplemental Table S1). Terminal mannose was not detectable. In agreement with previous studies (23) that demonstrated a partial conversion of 3-substituted GalNAc and 3,4-disubstituted GlcNAc into terminal GalNAc and 4-substituted GlcNAc upon chemical defucosylation of total KLH glycopeptides, these findings are consistent with the presence of a Fuc(1- Glycan fractions comprising species with a pentose residue exhibited, in addition, terminal xylose and 2,3,6-trisubstituted mannose. The present xylose residue could be, therefore, assigned to C-2 of the central mannose of the pentasaccharide N-glycan core. With the exception of fraction 15-1 glycans, which expressed a 4,6-disubstituted GlcNAc residue, all oligosaccharides contained 4-substituted GlcNAc, which is assumed to originate from the PA-modified chitobiose cores of these molecules. Intriguingly, galactose was found exclusively in terminal positions. Whereas all glycan fractions exhibited a related pattern of 2,4and 3,6-disubstituted mannose residues, differences could be observed with respect to the presence of 2,6-disubstituted, 2-substituted, or 6-substituted mannosyl residues, thus reflecting subtle structural differences (supplemental Table S1). Hence, monosubstituted mannose occurred only in low abundance in agreement with a highly branched architecture of most of the glycans.
Sequential exoglycosidase digestion with jack bean ␤-galactosidase and ␣-mannosidase in combination with MALDI-MS revealed the release of one Gal and zero to two Man residues (supplemental Table  S1). The finding of only one Gal being liberated is remarkable, because this monosaccharide represented the sole terminal sugar constituent apart from fucose and xylose and has been shown to be ␤-linked by chromium trioxide oxidation (data not shown). Possibly, efficient release of these moieties is impaired by sterical hindrance and/or the type of glycosidic linkages some of which might be resistant to the enzyme employed. Treatment with ␣-galactosidase from green coffee beans, performed as a further control, did not result in any release of galactosyl residues.
The ␣-anomeric configuration of fucose was verified by chromium trioxide oxidation as well as by ␣-fucosidase treatment, although only one fucose was released by the enzyme in each case (supplemental Table  S1). This finding is in agreement with earlier studies that demonstrated  DECEMBER 9, 2005 • VOLUME 280 • NUMBER 49     (Fig. 6) provided additional structural information. In each case resulting mass spectra revealed the preferential loss of a HexNAc 2 Fuc 2 moiety as evidenced by B-type fragment ions with compositions of HexNAc 1 , HexNAc 1 Fuc 1 , HexNAc 2 Fuc 1 , and HexNAc 2 Fuc 2 leading to major Y-type fragments with compositions of Hex 4 -6 HexNAc 2 Xyl 0 -1 PA (Fig. 7 and TABLE TWO). In addition, series of minor fragments were registered reflecting the sequential loss of four to six hexoses and one xylose, if present, as well as one HexNAc residue yielding a final HexNAcPA moiety. Hence, the obtained MALDI-MS/MS(LID) spectra confirmed unambiguously the presence of the aforementioned terminal difucosylated GalNAc(1-4)GlcNAc unit. In all cases, intense signals at m/z 503 (HexNAc 2 PA) were detected, whereas signals at m/z 446 (HexNAc 1 Fuc 1 PA) were not registered. Therefore, it can be concluded that this set of cross-reactive glycans lacked core fucosylation. Isomeric compounds obtained, for example, in the case of Hex 6 HexNAc 4 Fuc 2 species in fractions 15-1 to 15-3 (TABLE TWO) further revealed, in part, changes in fragment ion intensities that might reflect differences in structure (data not shown). Due to the lack of appropriate PA-oligosaccharide standards, however, this information could not be used for structural assignments.

Nano-ESI-IT-MS/MS
To allocate the identified difucosylated GalNAc(1-4)GlcNAc unit within the glycan molecule, Hex 4 HexNAc 4 Fuc 2 PA and Hex 5 Hex-NAc 4 Fuc 2 PA species were subjected to nano-LC-ESI-IT-MS n . The results revealed again a preferential loss of the terminal HexNAc 2 Fuc 2 unit in the MS 2 spectrum resulting in an Y 4␣ type of fragment ion (see, for example, Fig. 8 and TABLE TWO). Analysis of this fragment ion by an additional isolation and fragmentation cycle revealed in the corresponding MS 3 spectrum two pairs of ions (B 5 Y 4␣ /C 5 Y 4␣ and B 5 Y 3␣ / C 5 Y 3␣ ). Intriguingly, C 5 Y 4␣ ions obtained from precursors with four (Hex 4 HexNAc 4 Fuc 2 PA, not shown) or five (Hex 5 HexNAc 4 Fuc 2 PA; Fig.  8B) hexose units did not yield further ring fragmentation. In agreement with previous studies (39), it can, therefore, be concluded that the reducing hexose is still substituted at C3 in this fragment. In contrast, related C 5 Y 3␣ ions, comprising three (Hex 4 HexNAc 4 Fuc 2 PA) or four (Hex 5 HexNAc 4 Fuc 2 PA) hexose units, allowed further ring fragmentation in accordance with a C6 substitution of the reducing hexose. From these results it can be concluded that the difucosylated GalNAc(1-  (see TABLE  TWO and supplemental Fig. S2A). Only the D-ion led to 3,5 A 5 (m/z 421.2) and 0,4 A 5 (m/z 407.2) fragmentation due to ring cleavage of the core mannose (40,41), thus demonstrating that this mannose carried two additional hexose moieties linked to its C6 position (cf . TABLE  ONE). This finding is corroborated by the observed Y 4␣ and Y 3␣ ions, which verified that the 3-linked antenna comprised only one hexose residue. Furthermore, evidence for an unsubstituted chitobiose core has been provided by an Y 2␣ fragment ion (m/z 525.2) with a composition of HexNAc 2 PA. In conclusion, MALDI-MS/MS(CID) data are in perfect agreement with the assignment made by ESI-ion trap-MS n described above.
By the same line of reasoning Hex 5 HexNAc 4 Fuc 2 PA species obtained in fractions 13-2 and 14-1 (Fig. 6) were shown to represent two different structural isomers (TABLE ONE) carrying either two (fraction 13-2 glycans) or three (14-1) hexose residues at C6 of the branching mannose as evidenced by the respective D-ions and corresponding 3,5 A 5 and 0,4 A 5 ring fragmentation (TABLE TWO and supplemental Fig. S2, B and C). The latter structural isomer has been also identified by ESI-ion trap-MS n . The observed Y 2␣ fragment ion (HexNAc 2 PA) verified again an unsubstituted chitobiose core of these oligosaccharides. Hex 6 HexNAc 4 Fuc 2 PA species obtained in fractions 15-1 (Fig. 6) represented a novel type of N-glycan carrying a galactosyl residue linked to C6 of the penultimate GlcNAc of the chitobiose core. This unusual structural feature could be verified by methylation analysis (see supplemental Table S1) and, in particular, by MALDI-MS/MS(CID) due to (a) the observed mass difference between the registered B 6 (m/z 1896.7) and B 5 (m/z 1531.5) fragment ions amounting 365 mass units in agreement with a loss of a Hex 1 HexNAc 1 unit (Fig. 9A and TABLE TWO), (b)  were registered together with a corresponding 3,5 A 5 (m/z 421.2) and 0,4 A 5 (m/z 407.2) ring fragmentation. In addition, the mass difference between the Y 4␣ (1497.5; Hex 6 HexNAc 2 PA) and the Y 3␣ fragment (1173.4; Hex 4 HexNAc 2 PA) corresponded to two hexose units. Altogether, the obtained mass spectrum clearly demonstrated that this particular isomer carried two hexoses at C6 of the branching mannose, one hexose bound to the 3-linked mannose of the pentasaccharide core, and one hexose attached to the chitobiose unit (TABLE ONE).
Hex 6 HexNAc 4 Fuc 2 PA species obtained in fraction 14-2 furnished similar results as fraction 15-3 compounds. Because both glycans exhibited closely related retention times in reversed-phase HPLC (Fig. 6), it is assumed that they represent the same type of oligosaccharide species probably as a result of incomplete separation during size fractionation (Fig. 4A). In contrast, fraction 15-2 Hex 6 HexNAc 4 Fuc 2 PA glycans clearly differed from fraction 15-1 and 15-3 oligosaccharides by their retention times in reverse-phase HPLC, but led to similar MALDI-MS/ MS(CID) data with regard to the observed D-ion and subsequent ring fragmentation (TABLE TWO and Table S1) that xylose is linked to the central mannose of the pentasaccharide core. Based on these data, it could be concluded that these glycans represented the xylosylated variants of the Hex 5 HexNAc 4 Fuc 2 PA isomers discussed above. Respective glycans obtained in fractions 12-3 and 13-1 could not be analyzed due to low amounts (30).
Analysis of Trifucosylated Serologically Cross-reacting KLH Glycans-Due to small amounts, trifucosylated cross-reactive glycans (Fig. 3) were only analyzed by MALDI-MS/MS(LID) (TABLE THREE and Fig. 10). In this context, it turned out, however, that the applied mass spectrometric techniques required in most cases a preceding fractionation of the sugar chains: when starting from the total mixture of cross-reacting glycans (Fig.  3) successful and efficient mass spectrometric isolation of the respective precursor ions was often impaired by the small mass differences between related compositional species, thus preventing the isolation of pure parent ions. In the case of Hex 3 HexNAc 4 Fuc 3 PA, Hex 4 HexNAc 4 Fuc 3 PA, and Hex 5 HexNAc 4 Fuc 3 PA species recovered after lectin (AAL) affinity separation of aminophase HPLC fractions 12, 13, and 14 ( Fig. 4) in the 50 mM Fuc eluates (Fig. 5) (Fig. 10A). The fragment ion spectra of Hex 5 Hex-NAc 4 Fuc 3 Xyl 1 PA and Hex 6 HexNAc 4 Fuc 3 PA species were generated directly from the total immunoaffinity eluate (Fig. 3). Intriguingly, both of them furnished in addition to the Y-type fragment ions with m/z 446.1, m/z 649.3, and m/z 811.3 a signal at m/z 608 corresponding to a fragment composition of Hex 1 HexNAc 1 Fuc 1 PA (Fig. 10B). This finding is in agreement with earlier studies that demonstrated that the core-linked fucosyl residues of KLH-glycans may be, in part, further capped by galactosyl residues (21). Based on the sensitivity of the trifucosylated components toward PNGase F (42), the core fucose can be assigned to C6 of the innermost GlcNAc.
Structural Conclusions-Obtained analytical data allowed a number of important structural conclusions regarding the architecture of KLH N-glycans cross-reacting with S. mansoni glycoconjugates (see TABLE     corresponding glycan fractions studied by linkage analyses often represented mixtures of different compounds, a heterogeneous substitution pattern of the involved mannosyl residues was obtained. Therefore, allocation of terminal Gal units remained mostly ambiguous (TABLE ONE). In the case of fraction 15-1-derived Hex 6 HexNAc 4 Fuc 2 PA species, however, mass spectrometric analyses allowed in conjunction with linkage data a precise assignment of all additional Gal residues present (TABLE ONE). The proposed structure is further corroborated by the results obtained by sequential exoglycosidase treatment revealing the release of one ␤-linked galactose and one ␣-linked mannosyl residue (see supplemental Table S1). Hence, these glycans represent a novel type of N-glycans comprising an unusual substitution of the chitobiose unit by an additional Gal residue. Remaining Hex 6 HexNAc 4 Fuc 2 PA species obtained in fraction 15-2 and 15-3 represented exclusively the isomer II type of glycans. Based on linkage data (i.e. occurrence of 6-substituted mannose as well as 2,4-and 3,6-disubstituted mannosyl residues) and exoglycosidase results (release of one ␤-linked galactose and two ␣-linked mannosyl residues) fraction 15-2 glycans are concluded to exhibit a linear trisaccharide unit at C6 of the central mannose of the pentasaccharide core. In contrast, fraction 15-3 species comprised exclusively 2,4-, 2,6-, or 3,6-disubstituted mannose. Exoglycosidase treatment led to the release of one ␤-linked galactose, whereas subsequent liberation of mannosyl residues could not be achieved. Therefore, a highly branched structure can be proposed in this case, the outer galactose residues of which, however, could not be unambiguously assigned (TABLE  ONE). In conclusion, this study provides for the first time a detailed characterization of the KLH N-glycans responsible for the known cross-reactivity with S. mansoni.

DISCUSSION
The aim of this study was the characterization of glycans from keyhole limpet hemocyanin (KLH) sharing serologically cross-reacting carbohydrate motifs with glycoconjugates from S. mansoni. To this end oligosaccharides of KLH were released by hydrazinolysis or PNGase F and tagged with 2-aminopyridine. Cross-reacting glycans were identified by immunoaffinity chromatography using immobilized polyclonal antibodies directed against soluble egg antigens of S. mansoni. Respective oligosaccharides were isolated by sequential aminophase HPLC, lectin affinity chromatography, and reverse-phase HPLC and characterized in particular by different mass spectrometric techniques as well as linkage analysis and exoglycosidase treatment. The results revealed that all cross-reacting species represent novel types of N-glycans (cf. corresponding structures in TABLE ONE) all of which comprise a terminal Fuc(␣1-3)GalNAc(␤1-4)[Fuc(␣1-3)]GlcNAc(␤1-epitope, which has been reported to occur in glycosphingolipids and glycoprotein glycans from S. mansoni (38,(43)(44)(45). This finding is in agreement with recent observations demonstrating that the interaction of glycolipids from different S. mansoni life-cycle stages with monoclonal antibodies recognizing Fuc(␣1 Analysis of the carbohydrate structure of KLH proved to be an extremely difficult task due to the vast microheterogeneity of its glycans resulting in the expression of a great variety of different isomeric and/or isobaric structures. This feature has been already recognized in the case of the high mannose-type glycans and paucimannosidic sugar chains, which represent major carbohydrate substituents of KLH (20). Our present study shows that structural microheterogeneity is even more pronounced in the case of the numerous minor constituents present. According to our experience, the only way to analyze these glycans is to focus on defined structural or biological features expressed by distinct subpopulations of KLH sugar chains. In this study we have, therefore, concentrated exclusively on glycans serologically cross-reacting with S. mansoni glycoconjugates. In this context it should be emphasized, however, that the structural assignments made would not have been possible without prior separation of these oligosaccharides by different HPLC methods. Without fractionation, MALDI-MS/MS(LID or CID) analyses showed in most cases contaminations by neighboring signals due to an inefficient isolation of pure parent ions. Furthermore, the quality of CID spectra is known to strongly depend on the complexity of the sample (30).
We have demonstrated previously (23) that a Fuc(␣1-3)GalNAc motif is implicated in the cross-reactivity of S. mansoni infection sera with KLH. In that study, however, only an unfractionated mixture of total KLH glycopeptides has been employed. In the present contribution, the relevant carbohydrate chains have been isolated, individually characterized, and identified as N-glycans. Due to the structural heterogeneity mentioned above, however, it was not possible to obtain homogeneous oligosaccharide fractions, although a three-dimensional fractionation scheme has been employed. Furthermore, the oligosaccharides of interest represented minor compounds and could be isolated only in small amounts. As judged from the peak intensities of the MALDI spectra, cross-reacting difucosylated glycans, comprising six compositional species, represented ϳ3% of total PNGase F-released carbohydrates, whereas trifucosylated oligosaccharides, encompassing nine different species, comprise only 1.5%. Therefore, it was not possible to fully establish the structures for all cross-reacting oligosaccharides under study. The common feature of these glycans, however, could be unambiguously identified as being composed of a Man 3 GlcNAc 2 pentasaccharide core the 3-linked mannose of which is further substituted at C2 by the Fuc(␣1-3)GalNAc(␤1-4)[Fuc(␣1-3)]GlcNAc(␤1-moiety. As evidenced by minor cross-reacting species with compositions of Hex 2 HexNAc 4 Fuc 3 (Fig. 3) truncated variants lacking the 6-linked Man might also occur in small amounts. In most cases, however, the pentasaccharide core is further decorated by one to three galactosyl residues. In this context, a novel type of N-glycan could be identified carrying galactose at C6 of the penultimate GlcNAc residue of the chitobiose unit. Further modifications of the core structure include the substitution of the central mannose at C2 by xylose and/or the attachment of an additional Fuc residue to C6 of the innermost GlcNAc, which may be, in turn, further substituted by a galactosyl residue in agreement with previous data (21). In conclusion, the cross-reacting KLH carbohydrates represent a distinct class of glycoprotein-glycans.
In previous studies, galactosyl residues have been shown to occur in different positions in KLH glycans, including Gal(␤1-6)Man, Gal(␤1-4)GlcNAc, Gal(␤1-3)GlcNAc, and GlcNAc(␤1-2)[Gal(␤1-6)]Man (20) as well as Gal(␤1-4)Gal and Gal(␤1-4)Fuc linkages (21). In the present contribution, linkage analysis data suggest that galactose may also occur in Gal(␤1-2)Man and Gal(␤1-4)Man linkages, leading, in part, to 2,4-disubstituted and further 2,6-disubstituted mannosyl residues. Because these analyses were mostly performed on mixtures of glycans or different isomers, assignment of the exact linkages positions of galactosyl residues remains ambiguous in most cases. Likewise, recorded MALDI-MS/MS spectra did not yield sufficient ring cleavage ions to substantiate defined linkage positions of the galactose units. Possibly due to the small amounts of glycans available for these studies, MS/MS spectra obtained under CID conditions were dominated by glycosidic B-/Y-ions and displayed only less intense cross-ring cleavage ions (30). It has to be pointed out, however, that the present galactosyl residues are obviously not crucial for antibody binding.
Using peanut agglutinin and anti-Gal(␤1-3)GalNAc IgM antibodies, Wirguin et al. (8) demonstrated that KLH contains Gal(␤1-3)GalNAcbearing oligosaccharides. Furthermore, immunization with KLH-induced antibodies cross-reacting with the Thomson-Friedenreich (T) antigen. This finding has been later interpreted as an indication of O-glycosylation of KLH (47). To our knowledge, however, O-glycans have neither been isolated from KLH nor characterized yet. Insofar this point is important, because it has been suggested that these Gal(␤1-3)GalNAc-epitopes are ␤-linked in KLH (8), thus ruling out a conventional core 1, i.e. Gal(␤1-3)GalNAc(␣1-O)Ser/Thr, type of O-glycosylation. To address this question, we have subjected KLH glycopeptides to mild hydrazinolysis conditions, allowing the simultaneous release of both N-linked and O-linked sugar chains (27). Subsequent analysis of the resulting PA-glycans by immunoaffinity chromatography and MALDI-MS disclosed neither further carbohydrate chains nor additional cross-reactive oligosaccharide species (data not shown). Therefore, it may be concluded that the cross-reacting carbohydrate epitope is expressed exclusively by the type of N-glycans described in this study.
Cross-reactive KLH glycans represent, in terms of quantity, minor constituents of this glycoprotein. Nevertheless, KLH acts as an efficient inhibitor of S. mansoni infection serum-mediated recognition of schistosomal glycoconjugates and may be also used for serodiagnosis of schistosomiasis. One explanation for this phenomenon might reside in the giant size and heterogeneity of this molecule. KLH consists of two isoforms, KLH1 and KLH2, each of which is composed of several subunits of ϳ400 kDa. Whereas KLH1 occurs as a cylindrical didecamer, KLH2 forms a mixture of didecamers and tubular multidecamers (2,3,48). Each didecamer has a molecular mass of roughly 8,000,000 Da. Because each KLH1 and KLH2 subunit comprises eight and four potential N-glycosylation sites, respectively (EMBL data base: AJ698339 and AJ 698340), sufficient copies of cross-reactive oligosaccharides may exist within each KLH molecule, while representing only minor carbohydrate constituents.
In addition to the Fuc(␣1-3)GalNAc(␤1-4)[Fuc(␣1-3)]Glc-NAc(␤1-epitope, analyzed glycans contained, in part, a xylose residue. This monosaccharide constituent has already been detected in our previous study (21). N-Glycans comprising (␤1-2)-linked xylose are also present in hemocyanins of Helix pomatia (49,50) and Lymnea stagnalis (51). In contrast to KLH, however, xylosylated glycans have been reported to represent major carbohydrate constituents of these molecules. Intriguingly, N-glycans with a xylosylated trimannosyl core are also constituents of glycoproteins from S. mansoni eggs and cercariae (44,52). Hence, xylose (1-2)-linked to the central ␤-mannose represents a second carbohydrate epitope shared by KLH and schistosomal glycoproteins. Insofar, this is remarkable because plant glycoproteinglycans carrying such (␤1-2)-linked xylose residues have been shown to be immunogenic in animals, in particular in conjunction with (␣1-3)linked fucosyl substituents at the innermost GlcNAc of the chitobiose unit (53)(54)(55). The question remains open, however, as to whether xylosylated glycan species of KLH may also raise antibodies directed against this sugar moiety. Nevertheless, the detection of xylosylated glycans in KLH highlights again the multifarious glycosylation capacities of Megathura crenulata and demonstrates that this organism can be considered a rich reservoir for many, in part, novel glycosyltransferases.
As stated in the introduction, KLH is frequently used as an immunogenic protein carrier for the generation of glycoconjugate vaccines (see, for example Refs. 56 and 57). Because KLH itself carries a number of potentially immunogenic carbohydrate epitopes, resulting carbohy-drate-specific antibody responses have to be carefully analyzed. In this context, it has been demonstrated, for example, that sera from mice immunized with native KLH or KLH treated with glutaraldehyde crossreacted with several other carbohydrate antigens like lipoarabinomannan from Mycobacterium tuberculosis, glucuronoxylomannan from Cryptococcus neoformans, Lewis Y tetrasaccharide, or lipopolysaccharide from Escherichia coli serotype O55:B5 (58). Likewise, a strong reaction of sera from Trichinella patients with KLH has been described (59), which allows the use of KLH for the diagnosis of both trichinellosis and schistosomiasis and leads to the need of differential diagnosis in areas where both diseases occur (60). The structural basis for this pronounced cross-reactivity of Trichinella spiralis infection sera with KLH is not known. The Fuc(␣1-3)GalNAc(␤1-4)[Fuc(␣1-3)]GlcNAc(␤1-epitope described in this study, however, may provide the basis for the development of neoglycoproteins, which might be useful as highly specific reagents for serodiagnostic purposes.