Mass Spectrometric Identification of Aberrantly Glycosylated Human Apolipoprotein C-III Peptides in Urine from Schistosoma mansoni-infected Individuals*

Schistosomiasis is a parasitic infection caused by Schistosoma flatworms, prime examples of multicellular parasites that live in the mammalian host for many years. Glycoconjugates derived from the parasite have been shown to play an important role in many aspects of schistosomiasis, and some of them are present in the circulation of the host. The aim of this study was to identify novel glycoconjugates related to schistosomiasis in urine of Schistosoma mansoni-infected individuals using a combination of glycopeptide separation techniques and in-depth mass spectrometric analysis. Surprisingly, we characterized a heterogeneous population of novel aberrantly O-glycosylated peptides derived from the C terminus of human apolipoprotein C-III (apoC-III) in urine of S. mansoni-infected individuals that were not detected in urine of non-infected controls. The glycan composition of these glycopeptides is completely different from what has been described previously for apoC-III. Most importantly, they lack sialylation and display a high degree of fucosylation. This study exemplifies the potential of mass spectrometry for the identification and characterization of O-glycopeptides without prior knowledge of either the glycan or the peptide sequence. Furthermore, our results indicate for the first time that as a result of S. mansoni infection the glycosylation of a host protein is altered.


Schistosomiasis is a parasitic infection caused by Schistosoma flatworms, prime examples of multicellular parasites that live in the mammalian host for many years.
Glycoconjugates derived from the parasite have been shown to play an important role in many aspects of schistosomiasis, and some of them are present in the circulation of the host. The aim of this study was to identify novel glycoconjugates related to schistosomiasis in urine of Schistosoma mansoni-infected individuals using a combination of glycopeptide separation techniques and indepth mass spectrometric analysis. Surprisingly, we characterized a heterogeneous population of novel aberrantly O-glycosylated peptides derived from the C terminus of human apolipoprotein C-III (apoC-III) in urine of S. mansoni-infected individuals that were not detected in urine of non-infected controls. The glycan composition of these glycopeptides is completely different from what has been described previously for apoC-III. Most importantly, they lack sialylation and display a high degree of fucosylation. This study exemplifies the potential of mass spectrometry for the identification and characterization of O-glycopeptides without prior knowledge of either the glycan or the peptide sequence. Furthermore, our results indicate for the first time that as a result of S. mansoni infection the glycosylation of a host protein is altered. Molecular & Cellular Proteomics 9:667-681, 2010.
Schistosomiasis (also known as bilharzia) is one of the "neglected tropical diseases" affecting hundreds of million of people worldwide and is caused by infection with Schistosoma (1). Schistosomes have a complex life cycle, requiring adaptation for survival in fresh water as free-living forms and as parasites in snail intermediate hosts and vertebrate definitive hosts. Free-swimming cercariae are released from snails in water, penetrate the skin of the definitive host while shed-ding their tails, and transform into schistosomula. In the course of about 4 -6 weeks, the schistosomula migrate via the blood circulation and become adult male or female worms. In the case of Schistosoma mansoni, the paired male and female worms can live in the mesenteric venules for many years. The female worms deposit hundreds of eggs each day. Many of these are transferred to the intestine and excreted with the feces to eventually continue the life cycle, but a significant fraction is trapped in the liver of the host instead. Here, they provoke eosinophilic inflammatory and granulomatous reactions, which are progressively replaced by fibrotic deposits (2,3), damaging the overall function and integrity of the liver and thereby causing most of the morbidity associated with schistosomiasis.
During all developmental stages of the schistosome, a large variety of characteristic glycoconjugates are expressed (Ref. 4 and references cited therein), and a large number of the antibodies produced by infected subjects are directed against glycan epitopes of such schistosome glycoconjugates (5,6). Glycoproteins produced by the eggs play an important role in the modulation of the host's immune response and in the induction of the main pathology (7)(8)(9). Some secretory glycan and glycoconjugate antigens such as the worm gut-associated circulating anodic antigen and circulating cathodic antigen (CCA) 1 are released in the circulation of the host and form the basis for diagnosis of Schistosoma infection using a sandwich immunoassay with anti-carbohydrate monoclonal antibodies (10,11). Recently, the schistosome-specific multifucosylated glycan epitope recognized by a carbohydratespecific antibody that binds to egg glycoprotein antigens has been characterized (12). Interestingly, this antibody immunocaptured free oligosaccharides containing the same multifucosylated structural elements from urine of Schistosoma-infected individuals (13).
We hypothesized that other glycoconjugates specific for S. mansoni infection are present in the circulation. These might end up in the urine of infected individuals and could potentially serve as novel markers to monitor Schistosoma infection. To study this, we performed a comparative mass spectrometric analysis of urinary glycopeptides from Schistosoma-infected individuals and non-infected controls. Interestingly, we identified a set of aberrantly O-glycosylated, highly fucosylated peptides from human apolipoprotein C-III in urine from infected individuals but not in that from noninfected individuals. samples were analyzed (three infected and three non-infected). All serum samples were given a MS analysis number (non-infected:  11254, female, 33 years; 11255, male, 30 years; and 11251, male, 37  years; infected: 11250, female, 13 years, egg count of 8147 epg;  11249, female, 50 years, egg count of 287 epg; and 11252, male 11 years, egg count of 6080 epg).
Isolation of Urinary Peptides-Urines were centrifuged at 1500 ϫ g for 10 min at room temperature (RT), and the pellet was discarded. Three volumes of cold ethanol were then added to 1 volume of urine followed by gentle mixing, and urinary proteins were precipitated overnight at Ϫ20°C. The samples were subsequently spun for 45 min at 10,000 rpm, and the precipitated proteins were removed. The samples were then completely dried and stored at Ϫ20°C.
Strong Cation Exchange Chromatography (SCX)-Samples were resuspended in 500 l of Solvent A (10 mM KH 2 PO 4 (pH 2.9), 20% ACN). 100 l of each sample was injected on a PolySULFOETHYL A column (100 ϫ 2.1 mm, 3 m, 300 Å; PolyLC, Columbia, MD) at a flow rate of 0.2 ml/min using an Ä KTA TM Purifier (GE Healthcare) controlled by UNICORN software. After washing for 3.5 min with 100% Solvent A, peptides were eluted using a linear gradient from 30% Solvent B (500 mM KCl, 10 mM KH 2 PO 4 (pH 2.9), 20% ACN) to 100% Solvent B in 45 min. A total number of 16 fractions with a volume of 0.5 ml (2.5 min/fraction) were collected.
Hydrophilic Interaction Liquid Chromatography (HILIC)-Fractions 5 and 6 from five consecutive SCX fractionations from the same urine sample were pooled, lyophilized, and resuspended in 1 ml of Solvent C (50 mM ammonium formate (pH 4.4) containing 70% ACN). The sample was then loaded on a TSK-gel Amide-80 column (4.6-mm inner diameter ϫ 25 cm long; particle size, 5 m; Tosoh Bioscience, Stuttgart, Germany) at a flow rate of 0.4 ml/min using an Ä KTAPurifier controlled by UNICORN software. Peptides were eluted using a linear gradient of 12.5-50% Solvent D (50 mM ammonium formate) in 60 min. UV absorbance was measured at 215 nm. A total of 33 fractions with a volume of 1 ml (2.5 min/fraction) were collected, freeze-dried, and resuspended in 40 l of 0.1% TFA.
MALDI-TOF mass analyses were performed on an Ultraflex II timeof-flight mass spectrometer controlled by the FlexControl 3.0 software package (Bruker Daltonics). The MS acquisitions were performed in positive ion reflectron mode at a laser frequency of 50 Hz. The scanner m/z range was up to 5000, and the matrix suppression (deflection) mode was up to m/z 400. For the MS/MS analysis, precursors were accelerated and selected in a time ion gate after which fragments arising from metastable decay were further accelerated in the LIFT cell, and their m/z values were analyzed after passing the ion reflector.
Nano-LC ESI MS/MS-Nanoflow LC was performed on an Ultimate LC system (Dionex, Sunnyvale, CA). A volume of 10 l of sample was injected onto a C 18 PepMap TM 0.3 ϫ 5-mm trapping column (Dionex) and washed with 100% A (2% ACN in 0.1% formic acid in MQ water, v/v) at 20 l/min for 40 min. Following valve switching, peptides were separated on a C 18 PepMap 75-m ϫ 150-mm column (Dionex) at a constant flow of 200 nl/min. The peptide elution gradient was from 10 to 60% B (95% ACN in 0.1% formic acid in MQ water, v/v) over 50 min. The nanoflow LC system was coupled to an HCTultra IonTrap (Bruker Daltonics) using a nanoelectrospray ionization source. The spray voltage was set at 1.2 kV, and the temperature of the heated capillary was set to 165°C. Eluting peptides were analyzed using the data-dependent MS/MS mode over a 300 -1500 m/z range. The five most abundant ions in an MS spectrum were selected for MS/MS analysis by collision-induced dissociation using helium as the collision gas. Additionally, for MS 3 experiments, fragments of interest observed in an MS/MS spectrum were manually isolated and fragmented.
MALDI-TOF MS of Full-length Apolipoprotein C-III from Serum-Apolipoprotein C-III isoforms in serum were measured by MALDI-TOF MS according to Nelsestuen et al. (18). Briefly, serum (0.8 l) was diluted with MQ water:ACN:TFA (20 l, 95:5:0.1) and allowed to stand for 1 h at RT. The hydrophobic compounds were then extracted with a reverse phase C 18 ZipTip (Millipore) following the manufacturer's instructions. Following standard procedures, 1 l of the eluted sample in MQ water:ACN:TFA (25:75:0.1) was applied to the MALDI target along with sinapinic acid (1 l of saturated solution in MQ water:ACN: TFA, 50:50:0.1). Uniform crystallization was achieved by manual mixing of the sample with the pipette tip. The sample was dried and analyzed on an Ultraflex II MALDI-TOF mass spectrometer (Bruker Daltonics) operating in the linear positive ion mode. Two thousand laser shots were collected for each sample.
Trypsin Digestion-Five microliters of a 10% buffered aqueous solution of human apolipoprotein C-III (Sigma-Aldrich) was diluted with 10 l of 50 mM ammonium bicarbonate. Then, 0.5 l of 100 mM dithiothreitol was added, and samples were incubated for 30 min at 56°C. Subsequently, 5 l of 55 mM iodoacetamide was added, and samples were kept at RT for 20 min. Tryptic digestion was then performed by adding 5 g of trypsin (sequencing grade modified trypsin, Promega, Madison, WI) and incubating overnight at 37°C.

RESULTS
Analysis of Glycopeptides in Urine from S. mansoni-infected Individuals and Non-infected Controls-Urine samples collected from S. mansoni-infected individuals and non-in-fected controls were subjected to organic precipitation to deplete large proteins. Subsequently, samples were desalted on a reverse phase cartridge and fractionated by strong cation exchange chromatography. Following desalting, every fraction was analyzed using MALDI-TOF MS. Because we were specifically interested in the analysis of schistosomiasisrelated glycopeptides, we primarily focused on the higher m/z ranges. A representative MALDI-TOF mass spectrum from one SCX fraction from both non-infected and infected individuals is seen in Fig. 1. In the S. mansoni-infected individuals, we observed several signals between m/z 2500 and m/z 3500 that were not detected in the non-infected individuals. Furthermore, between several of the masses present in the individual spectra, mass differences corresponding to monosaccharides were evident, indicating the presence of a series of glycopeptides in these fractions. The fraction from the heavily infected individual was further analyzed with nano-LC-ion trap MS. The five most abundant ions in every MS spectrum were automatically selected for MS/MS, and spectra were searched for the presence of glycan-specific oxonium ions (m/z 366 ([Hex 1 -HexNAc 1 ϩ H] ϩ ) and 512 ([Fuc 1 -Hex 1 -HexNAc 1 ϩ H] ϩ ). An MS/MS spectrum of one of the glycoconjugates observed at m/z 1068.2 [M ϩ 3H] 3ϩ is given in Fig. 2. This peptide was observed at m/z 3202.0 in the MALDI-TOF mass spectrum from the SCX fraction from the heavily infected individual but not in the slightly and non-infected samples (Fig. 1). The MS/MS frag- In addition, sequential losses of glycosyl residues from the parent ion were observed. The strong signal at m/z 1500. The absence of large oxonium ions containing more than two HexNAc residues or more than one Hex element indicates a branched glycan structure. The HexNAc 1 -Hex 1 -HexNAc 1 (H 1 N 2 ) element is observed as a fragment at m/z 569.

Overview of aberrantly glycosylated human apolipoprotein C-III peptides identified in this study
Urinary peptides from S. mansoni-infected and non-infected individuals were separated using strong cation exchange chromatography. Fractions from urine of the infected individuals containing aberrantly glycosylated peptides from apoC-III were analyzed using LC-ion trap MS/MS, and the glycan composition and peptide backbone were assigned based on the fragment ions. Corresponding fractions from non-infected individuals were analyzed similarly. Sample numbers correspond to MS numbers assigned at the beginning of the study. The previously described glycan composition of the three human apoC-III glycoforms is also shown. y

Aberrant Apolipoprotein C-III Glycosylation
ative for the H 1 N 1 F 1 composition, probably representing the other arm of the branched structure. However, this ion may also result from fragmentation of the larger arm. After the loss of three Fuc residues, two HexNAc residues, and one Hex residue, the branched structure gives rise to the signal at m/z 1098.4 [M ϩ H] ϩ , which has a composition of N 2 H 1 -pep (where pep is the peptide moiety) and points toward a core 2 type O-glycosylation. Taken together, this CID MS/MS spectrum indicates that the composition of the glycan moiety is H 2 N 4 F 3 . The fragmentation data support the sequence Fuc 2 -(HexNAc 1 -Hex 1 -HexNAc 1 )-[Fuc 1 -(HexNAc 1 -Hex 1 )]-HexNAc 1 . The monosaccharide identity and linkage positions of the individual carbohydrate units cannot be derived from the MS/MS spectrum. Differentiation between GalNAc and Glc-NAc residues and linkage information are assigned on the basis of current knowledge of human O-glycosylation. We assume that a GalNAc is directly linked to the peptide moiety, whereas the other core HexNAc residue is a GlcNAc, and the hexose residues are assumed to be Gal, together forming the core 2 GlcNAc␤1-6Gal(␤1-3)GalNAc motif. In contrast, the outer HexNAc residues of this glycopeptide may represent GalNAc or GlcNAc residues.
Based on the glycan composition, the peptide backbone mass was deduced to be 1627.0 [M ϩ H] ϩ . We did not observe a substantial number of fragments that could be assigned to peptide backbone cleavages. This is a known phenomenon in ion trap CID MS/MS of glycopeptides (19 -21). Therefore, MS 3 experiments of the fragment at m/z 814.0 within the MS/MS spectrum were performed, but only a clear fragment at m/z 1097.5 was observed, and the quality of the resulting spectrum was insufficient to identify the peptide backbone (data not shown).

Identification of Aberrantly Glycosylated Human Apolipoprotein C-III Peptides in Urine of S. mansoni-infected
Individuals-Having identified a highly fucosylated glycopeptide in urine from a heavily infected individual in our comparative, global screen, we decided to perform a preparative purification of these urinary glycopeptides using 5 ml of urine. The SCX fractions containing the glycopeptides of interest were subsequently fractionated using HILIC HPLC, and fractions were analyzed by MALDI-TOF MS. Fig. 3A shows the spectrum from the fraction containing, among others, the above mentioned peptide at m/z 3201.6. The MALDI-TOF-TOF mass spectrum of this glycopeptide (Fig. 3B) showed a fragment at m/z 1626.7 [M ϩ H] ϩ . As described above, we predicted this to be the mass of the peptide backbone. Moreover, the fragment at m/z 1097.5 that was observed in the MS 3 analysis of the putative peptide backbone (see above) was also evident in this spectrum, indicating peptide backbone cleavages during MALDI-TOF-TOF fragmentation.
The MALDI-TOF mass spectrum revealed masses that potentially correspond to truncated peptides (missing Ala, AA, or VAA) carrying an identical glycan structure. MALDI-TOF-TOF analysis of the peptides at m/z 3130.6 and 2960.4 confirmed this prediction (Fig. 3B) because the mass of the fragments containing the peptide backbone shifted accordingly (from 1626.7 to 1555.6 and 1385.9, respectively). The glycopeptide at m/z 2611.3 [M ϩ H] ϩ was interpreted as having the same peptide backbone as the peptide at m/z 2960.4 [M ϩ H] ϩ lacking one Fuc 1 -HexNAc 1 element (Fig. 3B). The MALDI-TOF-TOF mass spectrum of the glycopeptide at m/z 3232.6 also revealed a fragment at m/z 1626.7. This indicates that the primary structure of the peptide backbone is the same as for the major species at m/z 3201.6, but the nature of the 31-Da mass difference remains elusive.
To identify the peptide backbone from the glycopeptides purified from urine from S. mansoni-infected individuals using SCX and HILIC HPLC, a high quality MALDI-TOF-TOF mass spectrum (Fig. 4A) from the most abundant glycopeptide (m/z 3201.6) was accumulated. This spectrum was used for manual interpretation, taking into account all the structural information obtained so far.
As mentioned above, the mass of the peptide backbone was predicted to be 1626.7 [M ϩ H] ϩ . We reasoned that the fragment at m/z 1097.5 was the result of a cleavage at a labile bond within the peptide backbone, such as N-terminal of a proline and/or C-terminal of an aspartic acid. The fragment at m/z 2672.9 represents the same peptide fragment carrying the full glycan structure. The ions at m/z 530.0 and m/z 1097.5 would form a pair of b* and y* ions (* indicating ions without the glycan moiety) arising from the fragmentation of one specific peptide bond. Similarly, a pair is formed by the ions at m/z 616.1 and 1011.3.
Assuming that the 1097.5 is a y n * ion N-terminal of a proline and that the ion at 871.4 is the y n Ϫ 2 * generates the tag Pro-Glu. Following this line of reasoning, the mass difference between 871.4 and 616.1 could correspond to RV/VR, but because of the ion at 755.2, the most plausible tag would then be Pro-Glu-Val-Arg with the aforementioned 755.2 ion corresponding to y n Ϫ 3 * Ϫ NH 3 . As discussed above, a set of glycopeptides corresponding to truncated peptides (lacking maximum AAV/VAA) was observed. Subsequently, BLAST searches using AAVXX(X)PEVR and PEVRXXX(X)VAA in the SchistoDB database (release 2.0; July 2008; 13,273 entries) were performed, but no peptide with full homology was identified. Therefore, partial matches were checked to see whether these could explain the generated MS data, but this approach also did not result in the identification of an S. mansoni peptide. As an alternative, BLAST searches in the human NCBInr database (released November 15, 2009; 225,299 entries) were performed, and surprisingly full homology with the C-terminal region of human apolipoprotein C-III (apoC-III; accession number GI:521205) was found using PE-VRXXXXVAA as the search query (E-value, 119). Moreover, the C-terminal peptide WDLDPEVRPTSAVAA of apoC-III has a theoretical mass of 1626.8 [M ϩ H] ϩ , nicely corresponding to what was predicted based on both the ion trap MS/MS data as well as the MALDI-TOF-TOF MS data. In addition, many of the theoretical y and b ions could be matched with this peptide sequence (Fig. 4A), and it also explains the MS/MS data of the truncated peptides (Fig. 3B).
Peptide WDLDPEVRPTSAVAA of apoC-III contains an Oglycosylation site at Thr 74 within full-length apoC-III, which normally carries a core 1 mucin type O-glycan structure containing one galactose, one N-acetylgalactosamine, and zero, one, or two sialic acids (22)(23)(24)(25). This heterogeneity in sialic acids is used in the nomenclature of the different human apoC-III isoforms, that is apoC-III 0 , apoC-III 1 , and apoC-III 2 , respectively. Although we cannot formally exclude the possibility that the aberrantly glycosylated apoC-III peptide is glycosylated at Ser 75 , we assume that it is also attached to Thr 74 .
To further corroborate our assignment of the peptide moiety, purified commercially available human apolipoprotein C-III was digested with trypsin and analyzed with MALDI-TOF-(TOF) MS. As expected, a heterogeneous population of glycopeptides corresponding to the apoC-III 0 , apoC-III 1 , and

Aberrant Apolipoprotein C-III Glycosylation
apoC-III 2 isoforms was evident (data not shown). The MALDI-TOF-TOF mass spectrum from the C-terminal tryptic glycopeptide DKFSEFWDLDPEVRPT 74 -(GalNAc 1 -Gal 1 )SAVAA from apoC-III 0 (Fig. 4B) showed a similar mode of peptide backbone fragmentation as observed by MALDI-TOF-TOF MS of the major glycopeptide purified from urine of S. mansoni-infected individuals (Fig. 4A). This is most apparent for the y ion series because both peptides have the same C terminus but obviously differ at their N terminus. Specifically, the y 11 ion is a prominent species in both spectra either with the glycan structure (m/z 1462.8 for apoC-III 0 and m/z 2672.9 for the urinary peptide) or without it (y 11 * at m/z 1097.5 in both spectra). Furthermore, the y 7 * , y 8 * Ϫ NH 3 , and y 9 * ions are clearly distinguishable in both spectra. Altogether, in urine of an S. mansoni-infected individual, we identified apoC-III-derived glycopeptides carrying a glycan structure that is totally different from that of normal human apoC-III.
Aberrantly Glycosylated Apolipoprotein C-III Peptides Are Exclusively Identified in S. mansoni-infected Individuals-As a next step, all the selected urines (four controls and six infected) were similarly analyzed. All LC-MS runs were manually inspected for the presence of highly fucosylated glycopep-tides containing the peptide backbone from the C terminus of apoC-III. Importantly, in none of the control samples were such glycopeptides identified (Table I). Fragmentation spectra corresponding to highly fucosylated apoC-III peptides were present in five of the six urines from S. mansoni-infected individuals. We used the information from the fragmentation patterns of the individual glycopeptides to predict the glycan composition and the peptide backbone (Table I). All these glycans differ drastically from the known glycan composition of human apoC-III. In total, they represent a very heterogeneous population of glycopeptides at both the peptide level and the glycan level. In one of the infected samples (22828), a mass corresponding to an aberrantly glycosylated peptide with an m/z and retention time similar to those of the one identified in sample 22830 was found, but it was not selected for MS/MS analysis and therefore not unambiguously identified (Table I). Purified glycopeptides from sample 10413 were also analyzed in the higher m/z range using MALDI-TOF MS. This revealed the presence of species containing one or two additional Hex 1 -HexNAc 1 -Fuc 1 units compared with the glycopeptide at m/z 3202.0 (Fig. 5). These units may represent additional Lewis X moieties (Gal␤134(Fuc␣133)GlcNAc␤1), Similarly, one putative extra Lewis X unit was noticed also for the truncated glycopeptide at m/z 3130.9 (lacking the Cterminal Ala).
We then analyzed samples containing the smaller peptides and glycan structures to gain more insight in the fucose positions. In one of the samples (22824), the apoC-III peptide WDLDPEVRPTSA containing the glycans H 2 N 2 F 2 and H 2 N 2 F 3 was found (Table I). In the MS/MS spectrum of the glycopep-tide containing two fucoses (m/z 1204.2 [M ϩ 2H] 2ϩ ), a singly charged ion corresponding to the peptide backbone with and without a HexNAc (m/z 1588.6 and 1385.5, respectively) was clearly present (Table I and  mentation of protonated ions, which may lead to misinterpretation of fragmentation spectra (26). Therefore, a fragmentation spectrum of the sodiated precursor [M ϩ H ϩ Na] 2ϩ at m/z 1288.2 (Fig. 6, bottom)  Analysis of Full-length ApoC-III in Serum from Infected and Non-infected Individuals-To study whether the aberrantly glycosylated apoC-III can be detected also in serum from S. mansoni-infected individuals, full-length apoC-III glycoforms in serum of infected and non-infected individuals were analyzed by MALDI-TOF MS (Fig. 7). A clear shift in the relative amount of non-, single, and double sialylated apoC-III (apoC-III 0 , apoC-III 1 , and apoC-III 2 ) in infected versus non-infected individuals was found. Most prominently, the ratio C-III 2 /C-III 0 is higher in infected than in non-infected individuals. However, in these analyses no full-length apoC-III glycoforms carrying fucosylated elements similar to those observed in the glycopeptides in the urine of infected individuals were identified. DISCUSSION Mass spectrometry has become an indispensable tool for the characterization of glycopeptides and the analysis of protein glycosylation. Using this technology, we identified and characterized a set of human apolipoprotein C-III peptides aberrantly glycosylated at the O-glycosylation site (Thr 74 ) in urine of S. mansoni-infected individuals.
Although many methods based on mass spectrometric characterization of protein glycosylation are available, they are primarily suited for the analysis of N-linked glycans (27). The identification and structural characterization of N-glycosylation are simplified by the fact that it is usually part of a consensus sequence and by the availability of broad spectrum enzymes for the release of N-glycans. It is apparent that the characterization of O-glycosylated peptides, at the level of the glycan structure as well as of the peptide backbone, is much more challenging, particularly when the protein under investigation is unknown (21, 28 -30). By combining different chromatographic techniques with various mass spectrometric analyses, we succeeded to characterize novel human apoC-III-derived O-glycosylated peptides. The glycan structures of these peptides are completely different from those described previously for apoC-III that consist of one galactose, one N-acetylgalactosamine, and zero, one, or two sialic acids (23,25). In contrast, non-sialylated but highly fucosylated core 2 O-glycosylated peptides were identified from the urine samples.
ApoC-III is a small (79-amino acid) protein synthesized primarily in the liver and to a lesser extent in the intestine. It is an essential constituent of circulating particles rich in triacylglycerol, very low density lipoproteins, and high density lipoprotein. ApoC-III inhibits the hydrolysis of triglyceride-rich particles by the lipoprotein lipase (31). As a result, overexpression of apoC-III is associated with hypertriglyceridemia (32,33) and atherosclerosis (34). Glycosylation on Thr 74 of apoC-III has been well described (23,35) but is not necessary for its secretion or lipid binding (36). Changes in the profile of the different apoC-III glycoforms analyzed with isoelectric focusing are used to identify O-glycan biosynthesis defects (37).
Currently, it is unclear what the relation is between aberrant apoC-III glycosylation, as described in this study, and S. mansoni infection. S. mansoni lacks a homologous apoC-III gene (38), and we found no homologous peptide in the S. mansoni sequence database using a BLAST search with the identified peptide sequence, indicating that the novel apoC-III glycoforms we found in S. mansoni infection urine are of human origin. It may be hypothesized, however, that the modified apoC-III glycosylation is due to the action of the schistosomal glycosylation machinery. Some of the multifucosylated glycan structures present on the C-terminal peptide of the schistosomiasis-correlated apoC-III glycoforms seem to contain Lewis X elements. Multiple Lewis X elements are commonly found in S. mansoni, for example as part of CCA (39), indicating that the glycosylation machinery of the schistosome might be capable of synthesizing the type of glycan we identified on apoC-III. Such an action would require that the host-derived apoC-III is taken up by the schistosome and processed by the action of schistosomal glycosidases and glycosyltransferases. Although this would be a highly speculative hypothesis, it is not unlikely that schistosomes take up host-derived apoC-III. S. mansoni flukes cannot synthesize sterols and free fatty acids, and they acquire these from their host. How this is accomplished is still enigmatic, but binding of low density lipoproteins to the schistosome tegument has been shown (40), and several (very) low density lipoproteinlike receptors have been identified within the genome of S. mansoni (38). It is noteworthy that human schistosomiasis is frequently accompanied by dyslipoproteinemia and influences the development of atherosclerosis (41)(42)(43)(44)(45). Besides being a nutrient source providing the parasite with cholesterol and other lipids, surface binding of lipoproteins to various Schistosoma life stages may inhibit binding of anti-schistosome antibodies (46,47) that may be part of the immune evasion process that underlies survival of the parasite.
Nevertheless, the fucosylated glycan elements identified on the apoC-III peptides in this study are not uncommon in other human glycoproteins (48 -50). Moreover, profound effects on liver function and architecture are caused by immunopathological reactions to schistosome eggs deposited in the liver of the infected individual. Therefore, it seems plausible that aberrant glycosylation of liver apoC-III can be caused by changes in the glycosylation activity in the liver due to schistosomiasis-induced reactions. This interpretation is supported by the fact that we observed changes in the profile of the common apoC-III glycoforms, resulting in higher levels of the doubly sialylated form and lower levels of the non-sialylated glycoform. We have not observed fulllength fucosylated apoC-III glycoforms, which might be the source of the presence of aberrantly urinary glycosylated peptides, in serum of infected individuals. This may be due to the relatively low levels of these forms in the serum or might have been the result of our methodology, which was developed to measure the common apoC-III glycoforms (18). Furthermore, the high level of heterogeneity in aberrant glycosylation, as observed on the urinary peptides, complicates straightforward identification of the full-length apoC-III equivalents.
Changes in the ratio of apoC-III glyco-isoforms have also been demonstrated in other diseases (51). Moreover, changes in the glycosylation of (serum) proteins in general have been observed in a variety of liver diseases (52). Interestingly, hyperfucosylation is one of the common alterations observed under these circumstances (53,54). Consequently, it might be that aberrant glycosylation as described in this study is also present in other liver disorders. Studies with a larger cohort, including not only urine samples of schistosomiasis patients with different infection levels but also of patients with other liver diseases, could give more insight in the potential of the glycopeptides identified in this study as novel morbidity markers of (certain) liver disorders or as specific and sensitive markers of Schistosoma infection.
In conclusion, this study identified novel aberrantly glycosylated apoC-III peptides in urine of S. mansoni-infected individuals. Future studies will have to demonstrate the specificity of this phenomenon for this parasitic infection and to unravel the role of both S. mansoni and human factors. This will provide new insights in the molecular changes that are associated with S. mansoni infection.