Glycobiology Advance Access originally published online on April 15, 2005
Glycobiology 2005 15(8):747-775; doi:10.1093/glycob/cwi061
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© Published by Oxford University Press 2005.
Altered O-glycosylation and sulfation of airway mucins associated with cystic fibrosis
2 Department of Biochemistry and Molecular Biology, University of Oklahoma Health Sciences Center, Oklahoma City, OK 73104; 3 The Oklahoma Center for Medical Glycobiology, University of Oklahoma Health Sciences Center, Oklahoma City, OK 73104; 4 College of Medicine, University of Oklahoma Health Sciences Center, Oklahoma City, OK 73104; 5 College of Pharmacy, University of Oklahoma Health Sciences Center, Oklahoma City, OK 73190; 6 Pediatric Pulmonary and Cystic Fibrosis Center, University of Oklahoma Health Sciences Center, Oklahoma City, OK 73104; and 7 Department of Pediatrics, University of Oklahoma Health Sciences Center, Oklahoma City, OK 73104
1 To whom correspondence should be addressed; e-mail: richard-cummings{at}ouhsc.edu
Received on October 13, 2004; revised on March 31, 2005; accepted on April 1, 2005
 |
Abstract |
|---|
 Top Abstract IntroductionResultsDiscussionMaterials and MethodsAcknowledgmentsReferences |
|---|
Cystic fibrosis (CF) is the most lethal genetic disorder in Caucasians and is characterized by the production of excessive amounts of viscous mucus secretions in the airways of patients, leading to airway obstruction, chronic bacterial infections, and respiratory failure. Previous studies indicate that CF-derived airway mucins are glycosylated and sulfated differently compared with mucins from nondiseased (ND) individuals. To address unresolved questions about mucin glycosylation and sulfation, we examined O-glycan structures in mucins purified from mucus secretions of two CF donors versus two ND donors. All mucins contained galactose (Gal), N-acetylglucosamine (GlcNAc), N-acetylgalactosamine (GalNAc), fucose (Fuc), and sialic acid (Neu5Ac). However, CF mucins had higher sugar content and more O-glycans compared with ND mucins. Both ND and CF mucins contained GlcNAc-6-sulfate (GlcNAc-6-Sul), Gal-6-Sul, and Gal-3-Sul, but CF mucins had higher amounts of the 6-sulfated species. O-glycans were released from CF and ND mucins and derivatized with 2-aminobenzamide (2-AB), separated by ion exchange chromatography, and quantified by fluorescence. There was nearly a two-fold increase in sulfation and sialylation in CF compared with ND mucin. High performance liquid chromatography (HPLC) profiles of glycans showed differences between the two CF samples compared with the two ND samples. Glycan compositions were defined by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS). Unexpectedly, 260 compositional types of O-glycans were identified, and CF mucins contained a higher proportion of sialylated and sulfated O-glycans compared with ND mucins. These profound structural differences in mucin glycosylation in CF patients may contribute to inflammatory responses and increased pathogenesis by Pseudomonas aeruginosa.
Key words: cystic fibrosis / airway mucin / sulfation / O-glycans / MALDI-TOF-MS
|
Introduction |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and MethodsAcknowledgmentsReferences |
|---|
Cystic fibrosis (CF) is an autosomal genetic disease resulting in the accumulation of mucus in exocrine organs. The disease results from mutations in the gene encoding the CF transmembrane conductance regulator (CFTR) on chromosome 7, which is a member of the adenosine triphosphate (ATP) binding cassette (ABC) family of ion transporters (Riordan et al., 1989
; Welsh et al., 2001). The disease is common, and 1 in 25 individuals of northern European extraction carry a mutation in CFTR (Gibson et al., 2003). A characteristic of CF is lung disease, resulting from excessive mucin production and accumulation and thickening of the mucus plaques and plugs (Boat and Cheng, 1989; Rose et al., 2000; Boucher, 2004). The accumulated mucus promotes bacterial adhesion and colonization, primarily by Pseudomonas aeruginosa, resulting in lung pathology and obstruction and eventual death owing to pneumonia (Welsh et al., 2001; Gibson et al., 2003). The estimated mean survival age of affected individuals as of 2001 was only 33.4 years (Gibson et al., 2003).
The submucosal bronchial glands and surface goblet cells secrete high molecular weight airway mucins encoded by numerous mucin (MUC) genes that contain abundant Ser/Thr-linked O-glycans (Lamblin et al., 1991, 2001), with sugar components representing over one half of the weight of the mucins (Lamblin et al., 1991, 2001; Thornton et al., 1991). At least 14 different mucin genes are known (Gendler and Spicer, 1995; Shankar et al., 1997; Moniaux et al., 1999, 2001), although eight MUC genes appear to be expressed at the mRNA level in respiratory tracts termed MUC1-8. However, only two of these mucins, MUC5AC, which is produced by goblet cells in the tracheobronchial surface epithelium (Hovenberg et al., 1996), and MUC5B, which is secreted by the submucosal glands (Wickstrom et al., 1998), appear to be major gel-forming mucins in both normal and physiological secretions in the airway (Kirkham et al., 2002; Henke et al., 2004).
Many studies have focused on the mucus-derived O-glycans in CF patients rather than mucins from nondiseased (ND) donors, because mucins are much more abundant and more easily isolated from CF patients (Carnoy et al., 1993; Devaraj et al., 1994; Scharfman et al., 1996; Ramphal and Arora, 2001). Several early comparative studies indicated that airway mucins from CF patients are more sialylated and sulfated than those from ND donors (Roussel et al., 1975; Boat et al., 1976; Lamblin et al., 1977; Chace et al., 1983, 1985). Moreover, the levels of neutral hexose and sialic acid in the CF secretions appear to increase with increasing severity of the disease (Chace et al., 1983). Recent studies on several dozen CF patients documented increased levels of sulfate and sialic acid in high molecular weight airway mucins from CF patients compared with mucins from individuals with chronic bronchitis (Davril et al., 1999).
The accumulated data indicate that CF-derived airway mucins have multiple types of O-glycans containing galactose (Gal), N-acetylglucosamine (GlcNAc), N-acetylgalactosamine (GalNAc), fucose (Fuc), sialic acid [N-acetylneuraminic acid (NeuAc)], and sulfate (van Halbeek et al., 1982; Lamblin et al., 1984a, 1991; Breg et al., 1987; Mawhinney et al., 1987, 1992a,b; Sangadala et al., 1992, 1993; Lo-Guidice et al., 1994; Chance and Mawhinney, 1996; Thomsson et al., 1998; Morelle et al., 2001). In a recent study on O-glycan structures of airway mucin from a single CF patient, 60 different O-glycans were identified (Thomsson et al., 1998), but it has been expected that airway mucins may contain many more O-glycans not yet identified (Lamblin et al., 1991).
Many questions remain about airway mucins from CF patients compared with mucins from ND donors, and more information is needed in regard to detailed and quantitative comparisons of O-glycan structures, an estimate of the total numbers of possible O-glycans and the types and quantities of sulfated O-glycans. To address these issues, we have taken the approach of releasing O-glycans from purified airway mucins by nonreductive ß elimination and derivatization with the fluorescent label 2-aminobenzamide (2-AB) by reductive amination to allow quantitative determinations of glycan species. We compared the O-glycan profiles from airway mucin preparations from two CF individuals and mucins isolated from mucus of two ND individuals. Our results show that airway mucins from CF patients are more highly O-glycosylated with a predicted higher density of O-glycans, and the O-glycans differ in many ways from mucins derived from ND individuals including the degree of sialylation and sulfation and the types of sulfation. Unexpectedly, airway mucins contain at least 260 different O-glycans distinguishable by mass spectrometry. The differences in O-glycan density and modification in CF compared with ND individuals could be important in the pathogenesis of the disease and potential interactions of the mucins with P. aeruginosa.
|
Results |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and MethodsAcknowledgmentsReferences |
|---|
Isolation and quantitation of mucins from CF and ND donors
Airway mucins were purified by established procedures using mucus from two CF patients (CF1 and CF2) and aspirates from two ND donors (ND1 and ND2), as described in Material and Methods. The purified mucins were extensively dialyzed against water, lyophilized, and weighed. Previous studies have shown that this highly purified mucin contains little if any contaminants, such as salts or small molecules, and that the dry weight is similar to the weight predicted from amino acid composition, sugar, and sulfate composition (Chace et al., 1989
). Mucins (1.00 mg) from ND1 and CF1 donors were dissolved in water, and the absorbance at 280 nm per mg for ND1 mucin was 0.353 and for the CF1 mucin, it was 0.158. Thus, as defined by both absorbance and dry weight, CF mucin contains significantly more sugar and less protein per mg protein than the ND mucins, which is consistent with previous studies (Chace et al., 1989) and with the analyses described below showing higher sugar content for the CF mucin.
Equal amounts by weight of each mucin from the four donors were examined by sodium dodecyl sulfate/polyacrylamide gel electrophoresis (SDS–PAGE) and stained with periodic acid/Schiff reagent (Zacharius et al., 1969), which is specific for carbohydrate (Figure 1). Only high molecular weight mucins were observed, and the CF mucin samples contained higher amounts of stainable sugar, consistent with the evidence below that these mucins are more highly glycosylated. The two ND samples resembled each other in mobility and showed slight differences in overall mobility in comparison with the two CF samples. No staining was observed with Coomassie Brilliant Blue, indicating that the high molecular weight mucins have no detectable low molecular weight contaminants.
|
Carbohydrate compositional analysis
Monosaccharide analyses were conducted on CF1 and ND1 mucins by acid hydrolysis, followed by Dionex high performance anion exchange chromatography (HPAEC). The only major monosaccharides identified were GalNAc, GlcNAc, Gal, Fuc, and Neu5Ac; no significant amounts of Man were observed, consistent with the presence of a large amount of O-glycans (Figure 2 and Table I). The results indicate that the overall carbohydrate composition of the mucins is generally similar, consistent with previous studies (Chace et al., 1985, 1989), although the CF mucin contains more total carbohydrate per mg mucin. Overall, the average carbohydrate content by dry weight of ND1 mucin was 
59%, whereas that of CF1 mucin was 76%. This is in the range of that determined previously for carbohydrate in mucins from respiratory mucus (Chace et al., 1983, 1989). The increase in sugar in CF mucin is partly due to an increased number of O-glycans, because the amount of GalNAc in CF mucin was increased nearly 58% over that in ND mucin. As described below, this interpretation is also consistent with the increased amount of 2-AB-labeled O-glycans recovered from the CF mucins compared with the ND mucins. Although the general monosaccharide composition CF versus ND mucins is relatively similar, mass spectrometric analyses described below show that there are many structural differences in O-glycans between the CF versus ND mucins.
|
|
Identification of sulfated monosaccharides
Previous studies have shown that airway mucins contain sulfated sugars and suggested that there are differences in the amounts of sulfated sugars in CF mucin compared with ND mucin, including differences in GlcNAc-6-sulfate (GlcNAc-6-Sul), Gal-3-Sul, and possibly Gal-4-Sul (Mawhinney et al., 1992a,b; Degroote et al., 1999, 2003; Lamblin et al., 2001). To address questions regarding these sulfated sugars, we utilized a partial hydrolysis strategy, in which sugar sulfates are liberated during hydrolysis and identified by Dionex HPAEC, as outlined in Materials and methods. A sample chromatographic profile is shown in Figure 3, and the results tabulated in Table I. The major sulfated monosaccharides in both CF and ND mucins were identified as GlcNAc-6-Sul (+GlcNH2-6-Sul), Gal-6-Sul, and Gal-3-Sul, with GlcNAc-6-Sul being the most abundant in both CF and ND mucins. CF mucins contained nearly twice as much GlcNAc-6-Sul and over twice as much Gal-6-Sul as ND mucin. These sulfation changes were specific, because the amount of Gal-3-Sul was only slightly increased in CF compared with that of ND mucin. It should be noted that because we used a partial acid hydrolysis strategy to define sulfated sugar content, the absolute amounts of sulfated sugars in the mucins are undoubtedly higher than that shown in Table I. However, the different recovery of sulfated sugars from equal amounts of the mucin samples can be used to document the relative difference in sulfated sugar content. These quantitative differences in amounts of sulfated sugars are confirmed by ion exchange chromatography of desialylated O-glycans, as discussed below.
|
Nonreductive ß elimination of O-glycans from CF and ND mucins and quantitation of O-glycans species by ion exchange column chromatography
To characterize the total O-glycan profile and quantify each glycan pool and individual species, we developed a general strategy to liberate and analyze the O-glycans from mucin as depicted in Figure 4. In this approach, equal amounts of dry weight and purified mucin were treated with basic ammonia to liberate reducing O-glycans by ß elimination of the O-glycosylated Ser and Thr residues. The released glycosylamine derivatives were converted to reducing sugars by boric acid treatment, as developed by Novotny’s laboratory, Bloomington, IN (Huang et al., 2001). This method generates little if any undesirable by-product, such as fragmentation or products resulting from "peeling" reactions in base. As control for this, we independently generated released glycans from bovine fetuin, as described by Huang et al. (2001), and compared the glycans released by this procedure with those released by PNGase F (N-glycanase). Analysis of the released glycans by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) showed that either method released the same N-glycans in precisely the same proportion without discernible differences, except that the ammonia also released the O-glycans (data not shown). Thus, the results indicate that ammonia release does not result in appreciable degradation of glycans.
|
We then fluorescently tagged all released glycans by reductive amination and derivatization with 2-AB, thus providing a quantitative fluorescent tag to the reducing end of each O-glycan through a secondary amine linkage. The O-glycans were separated by ion exchange into neutral and charged (sialylated, sulfated, or sialylated and sulfated) fractions by anion exchange chromatography on QAE-Sephadex, as described in Materials and methods. Fluorescence monitoring allowed quantitation of each glycan pool. In all cases below, we started the analyses with equal amounts of dry weight mucin, and all samples were compared in similar volumes. Thus, differences in fluorescence intensity or glycan amounts seen below in different data sets reflect differences in the initial amounts of glycans in the starting material. Unexpectedly, we recovered approximately twice as much 2-AB-labeled O-glycans from both CF1 and CF2, compared with the amounts recovered from ND1 and ND2, as discussed below, indicating an increased amount of O-glycans in CF mucins.
All four mucin samples were analyzed by this approach. Approximately 54% of the O-glycans from both ND1 and ND2 were neutral (ND1 is 55.1% and ND2 is 54.2%), whereas only 37% of the O-glycans from the CF1 and CF2 were neutral (CF1 is 38.8% and CF2 is 37.1%). The charge distribution of the anionic glycans was defined by a stepwise elution on QAE-Sephadex, in which 1-, 2-, 3-, and 4-charged anionic species are separated (Cummings et al., 1983). While 25% of the ND1 and ND2 anionic glycans had –1 charge, nearly 40% of the CF1 and CF2 anionic glycans had –1 charge. The next most abundant anionic glycan for all mucins was –2, and this was recovered in nearly similar amounts for all mucins. The –3 and –4 anionic species were relatively minor for all mucin glycans. To define whether the anionic nature of the glycans was due to sialic acid and/or sulfate, the anionic glycans A-ND1 and A-CF1 were treated with neuraminidase and reanalyzed on QAE-Sephadex. A majority of anionic glycans was converted to neutral species (NA-ND1 and NA-CF1) by desialylation (Figure 5). The desialylation was quantitative, and the bulk of the residual anionic material was sulfated, as defined below in mass spectrometric analyses. Thus, the CF mucin contains more highly sialylated and charged species than ND mucin, and the sialylated O-glycans that are not sulfated account for 26% of ND1 mucin O-glycans and 35% of the CF1 O-glycans. The amounts of glycans that are sulfated after desialylation were 18.7% for ND1 and 26.2% for CF1. These percentages should be corrected upward for the nearly doubling of total 2-AB-labeled glycans in CF compared with ND mucin, as discussed below. Thus, there is an approximate two-fold increase in sulfated and sialylated O-glycans in CF mucin compared with ND mucin. The increased amount of sulfated and sialylated glycans in the CF mucins is consistent with increased sugar and sulfated sugar content of CF mucins compared with ND mucins, as shown above.
|
Glycosep-C column chromatography comparison of O-glycans from CF and ND mucins
Because the released O-glycans from all four mucin samples were fluorescently labeled, we quantified the O-glycan amounts in mucins samples and compared their overall profiles by chromatography on Glycosep-C. This is a weak anion exchange column employing a polymeric stationary phase with amine functionality that allows the resolution of neutral, sialylated, sulfated, or phosphorylated glycans. Although the chromatographic profiles of all the mucin samples are somewhat similar except for differences in intensity, the profiles are different in several ways, suggesting some common structural differences between CF versus ND samples. Upon chromatography on Glycosep-C, the O-glycans from ND1 and ND2 mucins had relatively similar profiles to each other (Figure 6A), whereas the O-glycans from both CF1 and CF2 mucins showed distinct profiles from the ND1 and ND2 mucins (Figure 6B). Again, the CF samples exhibited nearly twice the overall fluorescence compared with the ND samples, indicating a near doubling of total O-glycans in the CF samples compared with the ND samples. These results are consistent with the ion exchange chromatography results in Figure 5, showing increased anionic character of the CF-derived compared with the ND-derived O-glycans.
|
Analysis of neutral O-glycans by MALDI-TOF-MS
To define the compositions of each glycan pool, we analyzed samples by MALDI-TOF-MS. Because a variety of analytical approaches indicated that the profiles of O-glycans in CF1 and CF2 were very similar, we focused our analyses on one of the samples, CF1, which was available in greater quantities. By a similar rationale, we chose to analyze ND1. To aid in approximating the relative amounts of different glycan species between samples in MALDI-TOF-MS, we employed an internal standard. For neutral glycans, this standard was lacto-N-neotetraose (LNnT—Galß1-4GlcNAcß1-3Galß1-4Glc), whereas the 2-anthranilic acid (2-AA) derivative of LNnT was used for anionic glycans. LNnT was chosen because its mass and the masses of its 2-AB and 2-AA derivatives are unique and distinct from any glycan mass observed in the mucin samples. To each sample to be analyzed by MALDI-TOF-MS, we added 20 pmol of the internal standard per µL of sample before spotting. We observed that the counts from the internal standards were consistently reproducible in many different analyses, thus indicating that it can be used as an internal standard to approximate the relative abundance of the O-glycans by using the scale given in Table II footnotes. We were also concerned that MALDI-TOF-MS analysis could cause partial fragmentation of 2-AB-labeled glycans. As controls for this, we performed MALDI-TOF-MS on commercial, purified sialyl Lewis x (sLex) tetrasaccharide and sulfo-Lewis A trisaccharide, both labeled with 2-AB. These analyses were under negative mode in the presence of ammonium citrate (20 mM) to protect against the degradation of sialic acid, the same conditions used our studies on 2-AB-labeled mucin O-glycans. We observed no loss of Fuc, sialic acid, or sulfate, because no defucosylated, desialylated, or desulfated fragments were found (data not shown). Thus, these results indicate that the spectrum of 2-AB-labeled glycans observed below is not a result of analytical artifacts introduced by either the ammonia-based elimination or the MALDI-TOF-MS.
|
The neutral glycans (N-ND1 and N-CF1) and the anionic glycans (A-ND1 and A-CF1) were analyzed by MALDI-TOF-MS, and representative complete spectra are shown in Figures 7A and 8A, respectively. Figure 7A shows the entire unannotated spectra for both N-ND1 and N-CF1, indicating the internal standard LNnT. Similarly, Figure 8A shows the unannotated spectra for both A-ND1 and A-CF1, indicating the internal standard LNnT-AA. The overall spectra show many differences, but because of the spectral complexity, the spectra are divided into separate panels shown in Figures 7B–E and 8B–D. Notable glycans in Figures 7B–E and 8B–D are identified by mass, but not all peaks are annotated to allow the figures to be readable. As described below, much of the data is presented in Tables.
|
|
|
|
|
Compositional analyses in Table I revealed that each mucin sample contained a single sialic acid, identified as Neu5Ac, a single hexose, identified as Gal, a single deoxyhexose, identified as Fuc, and HexNAc (GlcNAc or GalNAc). Mucin-type O-glycans are linked via GalNAc to peptide (Roussel et al., 1975; Boat et al., 1976), thus generating R-GalNAc-AB upon derivatization. In our analyses, we separated the neutral (N-ND1 and N-CF1) and anionic O-glycans (A-ND1 and A-CF1) and the neuraminidase-treated O-glycans, after separating them into neutral (NA-ND1 and NA-CF1) or residually anionic species (AA-ND1 and AA-CF1), as shown in Figure 4.
Surprisingly, 260 different 2-AB labeled glycans were compositionally identified in MALDI-TOF-MS when the results from analyses of both ND and CF mucins are combined. But the samples probably contain even more compositionally distinct glycan species that are at the borderline of detection, which we generally considered to be >1000 counts on the MALDI-TOF-MS spectra (Table II footnote). To aid in the analysis of the data, we tabulated the O-glycans as those in common and in approximately similar amounts between ND1 and CF1, or as not in common or in different amounts between the samples. Twenty-seven different neutral O-glycans were identified and found to be in common and in similar quantities to both mucin preparations (Table II). By contrast, 16 O-glycans exhibited differences in expression levels between the two samples (Table III). The observed m/z of the identified neutral O-glycan derivatives ranged from 504.82 to 1898.64.
A common O-glycan core structure in mammalian mucins is the core 1 O-glycan structure (Galß1-3GalNAc
1-Ser/Thr), and this has been reported previously in human airway mucins (Lamblin et al., 1984a,b,c; Breg et al., 1988; Mawhinney et al., 1992a,b; Thomsson et al., 1998, 2002; Degroote et al., 1999; Scanlin and Glick, 1999; Morelle et al., 2001; Rhim et al., 2001). The simplest glycan we observed in the neutral samples had the composition [Gal1GalNAc-AB+H]+ (obs. m/z 504.82), consistent with that predicted for the core 1 O-glycan. Another common O-glycan structure in mammalian mucins is the core 2 O-glycan (Galß1-3(Galß1-4GlcNAcß1-6)GalNAc1-Ser/Thr), which has been previously reported in human airway mucins (Lamblin et al., 1984a,b,c; Breg et al., 1988; Mawhinney et al., 1992a,b; Thomsson et al., 1998, 2002; Degroote et al., 1999; Scanlin and Glick, 1999; Morelle et al., 2001; Rhim et al., 2001). O-Glycans with a composition consistent with the core 2 O-glycan [Gal2HexNAc1GalNAc–AB+H]+ or [Gal2HexNAc1GalNAc–AB+Na]+ (obs. m/z 869.78 and 891.91) were found for both mucin samples.
A considerable number of the common O-glycans in the N-ND1 and N-CF1 samples (Table III) also contained the composition R-HexNAc2GalNAc-2AB and R-HexNAc3GalNAc-2AB, indicating the possible presence of core 3 and core 4 O-glycans, which have the core structures of GlcNAcß1-3GalNAc1-Ser/Thr and GlcNAcß1-3(GlcNAcß1-6)GalNAc1-Ser/Thr, respectively. O-Glycans with these core structures are known to occur in airway mucins (Lamblin et al., 1984b; Breg et al., 1988; Mawhinney et al., 1992a; Lo-Guidice et al., 1997; Degroote et al., 2003). Another important feature of many of the neutral O-glycans in Table III is the presence of Fuc residues. A majority of the O-glycan species contained from one to three Fuc residues.
The sizes of the neutral O-glycans exhibiting differences in expression between ND1 and CF1 ranged from obs. m/z 1054.12 to 1794.13 (Table III). A major glycan recovered in N-CF1, but not in N-ND1, was [Fuc1Gal1HexNAc2 GalNAc–AB+H]+ (obs. m/z 1056.14). Several other Fuc-containing glycans were also found in N-CF1 but not in N-ND1 mucins, such as [Fuc2Gal1HexNAc2GalNAc–AB+H]+ (obs. m/z 1202.28), [Fuc1Gal2HexNAc3GalNAc–AB+H]+ (obs. m/z 1444.48), and [Fuc2Gal2HexNAc4GalNAc-AB+Na]+ (obs. m/z 1794.13). By contrast, N-ND1 mucins contained several nonfucosylated O-glycans not found in N-CF1 mucin, for example, Gal5HexNAc1GalNAc-R, Gal4HexNAc2GalNAc-R, and Gal4HexNAc3GalNAc-R. The results tend to show increased content of Fuc residues in neutral O-glycans in CF1 compared with ND1. Several O-glycans with unusual compositions, such as [Gal4HexNAc2GalNAc–AB+Na]+ (obs. m/z 1419.63) and [Gal5HexNAc1GalNAc–AB+Na]+ (obs. m/z 1378.4), were identified in the neutral O-glycans in N-CF1, but not in N-ND1, and their identities, along with other unusual O-glycan compositions, are discussed in a section below on unusual O-glycan compositions.
Analysis of anionic O-glycans by MALDI-TOF-MS
The anionic O-glycans bound by QAE-Sephadex from both ND1 and CF1 contained 169 O-glycans in common and in similar amounts (Table IV). Of these, 97 were sulfated (and some of these were sialylated), and the remaining 72 glycans were sialylated only. Thus, most anionic O-glycan species in common between A-ND1 and A-CF1 are sulfated. The presence of sulfate was verified in analyses below of desialylated samples. Interestingly, 48 anionic O-glycans were found to be differentially expressed in the two mucin samples, and of these, 23 were sulfated (and some of the sulfated glycans were also sialylated), whereas the remainder was sialylated but not sulfated (Table V).
The anionic species in common (Table IV) ranged in size from obs. m/z 582.24 to 3285.51, with the largest O-glycan having the composition [Sul1Neu5Ac1Fuc4Gal6HexNAc5GalNAc-H]– (obs. m/z 3285.51), which represents one of the largest O-glycan compositions yet identified in human airway mucins (Thomsson et al., 1998). Most of the anionic O-glycans in common between ND1 and CF1 contained a single sulfate residue or a single Neu5Ac residue, or one of each. This is consistent with the tabulation following anion exchange chromatography in Figure 5, showing a predominance of –1 and –2-charged species in the anionic pools from both ND1 and ND2 and in CF1 and CF2. We did not observe O-glycans containing more than one sulfate residue at this resolution. QAE-Sephadex chromatography (Figure 5) showed that some O-glycans contained –3 and –4 charges, indicating the possible presence of multiple sialylated and multiple sulfated O-glycans. Studies are in progress to prepare more of these species for separate analyses by MALDI-TOF-MS.
Most of the sialylated samples contained a single Neu5Ac residue, but 31 glycans in common between the A-ND1 and A-CF1 were found to contain two sialic acid residues (Table IV). Interestingly, none of the O-glycans identified at this level of sensitivity contained two sialic acid residues along with a sulfate residue. Most high molecular species contained from one to six Fuc residues. The most highly fucosylated sample was [Sul1Fuc6Gal6HexNAc5GalNAc–AB-H]– (obs. m/z 3285.51) (Table IV). Thus, both ND and CF mucins contain in common many high molecular weight sulfated, sialylated, and fucosylated O-glycans.
However, 48 anionic O-glycans were differentially expressed in A-CF1 versus A-ND1 mucins (Table V). Most of the sialylated samples contained a single Neu5Ac residue, but six glycans contained two sialic acid residues (Table V). Of these 48 O-glycans, 23 were sulfated, but 16 of these were sulfated but not sialylated, and only seven were both sialylated and sulfated. Again, none of the O-glycans containing two sialic acid residues also contained sulfate, and no O-glycans were found at this level of analysis to contain more than one sulfate residue. There was an increase in O-glycans from CF mucin with compositions typical of core 1- and core 2-type O-glycans, such as [Neu5Ac1Gal1GalNAc–AB–H]– (obs. m/z 793.59) and [Neu5Ac1Gal2HexNAc1GalNAc–AB–H]– (obs. m/z 1159.0), as well as increased levels of [Neu5Ac1GalNAc–AB–H]– (obs. m/z 631.29) (Table V), which is a composition corresponding to the sialyl-Tn antigen, NeuAc2-6GalNAc1-Ser/Thr. In general, many of the anionic O-glycans found to be more highly expressed in CF compared with ND mucin were fucosylated and contained from one to six Fuc residues. Although the highly fucosylated O-glycan [Sul1Fuc6Gal6HexNAc5GalNAc–AB–H]– (obs. m/z 3139.6) containing sulfate but not sialic acid was found to be expressed at similar levels between A-CF1 and A-ND1, A-CF1 contained higher amounts of highly fucosylated O-glycan(s) containing sialic acid but not sulfate, for example, [Neu5Ac1Fuc6Gal4HexNAc3GalNAc–AB/Neu5Ac2Fuc4Gal4HexNAc3GalNAc–AB–H]– (obs. m/z 2766.48) (Table V).
Analysis of neuraminidase-treated and desialylated O-glycans by MALDI-TOF-MS
The results above established that the anionic O-glycans from both CF and ND samples were sialylated and/or sulfated. To examine more closely the underlying O-glycan of the sialylated and sulfated species and confirm the presence of Neu5Ac, the entire anionic samples (A-ND1 and A-CF1) were treated with neuraminidase and then reapplied to QAE-Sephadex to separate the resultant neutral species (NA-ND1 and NA-CF1) from the residually anionic species (AA-ND1 and AA-CF1) (Figure 4). Upon treatment with neuraminidase, 26.2% of the A-ND1 O-glycans were recovered as neutral species (NA-ND1), which represented 26.2/44.9 (x 100) = 58% of the starting anionic material. Thus, 26.2% of the O-glycans in the starting material ND1 are sialylated but not sulfated. By contrast, treatment of A-CF1 with neuraminidase resulted in the production of 35.1% as neutral species (NA-CF1), which represented 35.1/61.3 (x 100) = 57% of the starting anionic material. Thus, 35.1% of the O-glycans in the starting material CF1 are sialylated but not sulfated. The residual anionic species following neuraminidase treatment (AA-ND1 and AA-CF1) represent 18.7 and 26.2%, respectively, of the total O-glycans from the two mucin samples, demonstrating a higher degree of sulfation of the O-glycans in CF mucin compared with ND mucin.
To further analyze the O-glycan composition and confirm the removal of sialic acid and the residual nature of the sulfate residues, we analyzed the NA-ND1, NA-CF1, AA-ND1, and AA-CF1 O-glycans by MALDI-TOF-MS. The results are presented in Tables VI–IX. The compilation of those neutral O-glycans in NA-ND1 and NA-CF1 found to be expressed at common levels between the two mucin samples are shown in Table VI. None of the O-glycans in NA-ND1 and NA-CF1 contained either Neu5Ac or sulfate as expected. Most of the O-glycans were highly fucosylated and contained from one to five Fuc residues, and the largest O-glycan had an obs. m/z 2542.42. As might be predicted, all of the major glycans in the NA-ND1 and NA-CF1 were found as sialylated compounds in the MALDI-TOF-MS spectra of the original A-ND1 and A-CF1 material (Table IV), thus providing further confirmation to the effectiveness of neuraminidase in quantitatively removing sialic acid from the glycans and the stability of the glycans to the manipulation.
|
|
|
|
The compilation of the neutral O-glycans recovered following desialylation by neuraminidase in NA-ND1 and NA-CF1 and differentially expressed between the two mucin samples is summarized in Table VII. Again, these asialo compounds were observed as sialylated species in the A-ND1 and A-CF1 samples. For example, [Neu5Ac1Gal2HexNAc1GalNAc–AB–H]– was more abundant in A-CF1 than in A-ND1 (Table V), and the desialylated form of that glycan [Gal2HexNAc1GalNAc–AB+H]+ was recovered and found to be more abundant in NA-CF1 than in NA-ND1 (Table VII). Similarly, the Fuc-containing anionic O-glycans [Neu5Ac1Fuc1Gal2HexNAc1GalNAc–AB–H]– (obs. m/z 1305.10) and [Neu5Ac2Gal2HexNAc1GalNAc–AB–H]– (obs. m/z 1450.57) were more abundant in A-CF1 compared with A-ND1 (Table V), and following neuraminidase treatment, the desialylated, neutral derivative [Fuc1Gal2HexNAc1GalNAc–AB+H]+ (obs. m/z 1015.89) was recovered and was more abundant in NA-CF1 compared with NA-ND1 (Table VII).
The anionic O-glycans AA-ND1 and AA-CF1 recovered following desialylation by neuraminidase and found to be expressed similarly in the different glycan samples are summarized in Table VIII. All of the recovered O-glycans in AA-ND1 and AA-CF1 were monosulfated, as predicted by the results from anion exchange chromatography on QAE-Sephadex and the compositional analyses in Table IV. Nearly all of the sulfated and nonsialylated O-glycans identified in the A-ND1 and A-CF1 fractions in Table IV were recovered as AA-ND1 and AA-CF1, as summarized in Table VII. The results support the presence of a high number of sulfated and nonsialylated O-glycans in common between CF1 and ND1.
Thirty-one different sulfated O-glycans were identified as anionic species following desialylation by neuraminidase in the AA-ND1 and AA-CF1 and found to be differentially expressed between the two mucins and mostly found in the AA-CF1 material (Table IX). In particular, a relatively large amount of [Sul1Gal2HexNAc1GalNAc-AB-H]– (obs. m/z 947.31) was recovered in AA-CF1 compared with AA-ND1 (Table IX). This is similar to the higher amounts of the sulfated and fucosylated form of this underlying Gal2HexNAc1GalNAc-AB O-glycan, which corresponds in composition to a core 2-type O-glycan in all CF1 samples.
Unusual O-glycan compositions
We identified several glycans with unusual compositions not commonly found in mucins. These include [Gal2GalNAc–AB+H]+ (obs. m/z 666.60) and the possibly related compounds containing one or two Fuc residues, as in [Fuc1Gal2GalNAc–AB+H]+ (obs. m/z 812.50) and its [Na]+ form (obs. m/z 834.85) and [Fuc2Gal2GalNAc–AB+Na]+ (obs. m/z 981.09) (Table II). [Gal3GalNAc–AB+Na]+ (obs. m/z 852.78) was also identified (Table VII) as well as the possibly sulfated, sialylated, and/or fucosylated derivatives [Sul1Gal3GalNAc–AB–H]– (obs. m/z 906.89) (Table V), [Neu5Ac1Gal3GalNAc–AB–H]– (obs. m/z 1117.27) (Table IV), [Neu5Ac2Gal3GalNAc–AB–H]– (obs. m/z 1409.41) (Table IV), [Sul1Neu5Ac1Fuc1Gal3GalNAc–AB–H]– (obs. m/z 1346.45) (Table V), and [Sul1Neu5Ac1Fuc3Gal3GalNAc–AB–H]– (obs. m/z 1636.43) (Table V). Some of these compounds were enriched in the CF compared with the ND mucin. Another unusual set of sulfated glycans includes [Sul1Gal5GalNAc–AB–H]– (obs. m/z 1230.09) (Tables V and IX) and [Sul1Gal4GalNAc–AB–H]– (obs. m/z 1067.58) (Table V). These two species were found in CF1, but not in NF1, and may represent new O-glycan core structures rich in Gal. Several O-glycans with unusual core structures and compositions relating to those we observed have been seen previously in human mucins, although not in airway mucins. These include Galß1-6GalNAcitol and Galß1-6(Galß1-3)GalNAcitol identified in human gastric mucin (Slomiany et al., 1984a,b), the digalactosylated O-glycan Gal-Galß1-3GalNAcitol found in human MUC1 derived from MCF7 cells (Backstrom et al., 2003), the -galactosylated trisaccharide NeuAc2-6(Gal1-3)GalNAcitol (van Halbeek et al., 1994), and a digalactosylated O-glycan Galß1-3(Galß1-6)GalNAcitol, which was reported in O-glycans in the human blood fluke Schistosoma mansoni (Huang et al., 2001). Studies are in progress to precisely define the structures and sequences of these potentially unusual O-glycans, which we identified here in airway mucins.
|
Discussion |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and MethodsAcknowledgmentsReferences |
|---|
Our study reveals that the O-glycans in airway mucins purified from patients with CF and from ND individuals contain at least 260 compositionally different O-glycans. The discovery of such a large number of different O-glycans is in line with earlier predictions that mucins may contain several hundred different carbohydrate chains based on ion exchange chromatography and high performance liquid chromatography (HPLC) of oligosaccharide alditols generated by reductive ß elimination (Lamblin et al., 1992
). Furthermore, the results demonstrate that the O-glycans from CF patients differ significantly from airway mucins of ND individuals in the predicted numbers of O-glycan chains per mg protein, O-glycan composition, and the degree and type of sulfation. Along with higher amounts of monosaccharides, CF mucin also contains higher amounts of GlcNAc-6-Sul and Gal-6-Sul compared with ND mucin. A majority (54%) of the O-glycans in ND mucin is neutral, whereas a majority (63%) of the O-glycans from CF donors is anionic and contains both sialylated and/or sulfated species. Overall our results are largely in agreement with previous studies on selected glycan fractions and significantly extend our understanding of the important differences in O-glycosylation between CF and ND mucins.
Many excellent studies have defined the structures of some O-glycans in airway mucins from both CF and ND individuals or individuals having chronic pulmonary disease, and have shown that the mucins contain a variety of sulfated, sialylated, and fucosylated O-glycans, generally of the core 1 and core 2 types, but also including core 3 and core 4 types (van Halbeek et al., 1982; Lamblin et al., 1984c, 1991; Breg et al., 1987, 1988; Mawhinney et al., 1987, 1992a,b; van Kuik et al., 1991; Sangadala et al., 1992, 1993; Lo-Guidice et al., 1994; Chance and Mawhinney, 1996; Thomsson et al., 1998; Morelle et al., 2001; Degroote et al., 2003). Importantly, most if not all of the mucin O-glycans structures identified in these previous studies, including the 60 species recently identified by mass spectrometry (Thomsson et al., 1998), correspond to glycan compositions identified in our study, thus confirming and extending these previous results. Just a few examples of this identity will be discussed. One of the large-sized sulfated and fucosylated O-glycans in airway mucin is a core 4-based O-glycan having Scheme 1 (Degroote et al., 2003), which corresponds in composition to [Sul1Fuc2Gal3HexNAc3GalNAc–AB–H]– (obs. m/z 1808.54) (Table IV). A core 2-based O-glycan in airway mucin has Scheme 2 (Klein et al., 1988), which corresponds in composition with [Fuc1Gal3HexNAc2GalNAc–AB+Na]+ (obs. m/z 1403.51) (Table II). Finally, a sialylated and sulfated core 4-based O-glycans in CF mucin has Scheme 3 (Mawhinney et al., 1992b), which corresponds in composition with [Sul1Neu5Ac1Fuc1Gal2HexNAc2GalNAc–AB–H]– (obs. m/z 1588.53) (Table IV). Thus, the results of our study fit in well with the overall analyses by many other laboratories using different techniques.
|
|
|
Most studies on glycan structure and composition have been performed on proteolytic fractions (e.g., trypsin or pronase treated) derived from intact mucins rather than intact high molecular weight mucins (e.g., Breg et al., 1988; van Kuik et al., 1991; Lo-Guidice et al., 1994; Thomsson et al., 1998; Degroote et al., 2003). Other approaches have involved analysis of endo-glycosidase fractions from whole mucin preparation. For example, a recent study utilized Fast atom bombardment-mass spectrometry (FAB-MS) characterization of fragments released by endo-ß-galactosidase treatments of total O-glycans to demonstrate the presence of sLex in all mucin preparation from eight CF patients, whereas sLex was found in fragments from only three of eight patients suffering from chronic bronchitis (Morelle et al., 2001).
In our approach, we attempted to identify as many intact O-glycans species recoverable from high molecular intact mucin and exploit mass spectrometry and derivatization chemistry to allow quantitation of glycan species and estimate their diversity. To this end, we used fluorescent-labeling with 2-AB and the inclusion of internal standard glycans to allow an overall comparison and quantitation of all glycan species. Our approach is dependent on the recent discovery of chemical release of O-glycans as glycosylamines (1-aminoglycans) by ß elimination from Ser/Thr residues by treatment with ammonia and subsequent conversion to reducing glycans through boric acid treatment (Huang et al., 2001). Our study is one of the first to exploit this release strategy of nonreductive ß elimination of purified airway mucin to allow the derivatization of released glycans with 2-AB and their quantification by fluorescence measurements. A recent elegant study on mucin type O-glycan (Robbe et al., 2003) employed an approach related to ours, except that the nonreductively released glycans were labeled by reductive amination with different fluorophores for analysis by gel electrophoresis. An advantage of both such approaches is that O-glycans are easily released and quantitatively derivatized and the glycans can be precisely quantified by fluorescence measurements. In addition, the derivatization of O-glycans with 2-AB converts the O-glycans from a form that does not bind reverse phase resins, for example, C18, to a derivative that does bind, thereby aiding in purification of O-glycans and reduction of background in MS analyses.
The presence of sulfated sugars in airway mucins has been previously noted as well as an overall increase in sulfation in mucins from CF patients (Boat et al., 1976; Chace et al., 1983; Cheng et al., 1989; Scharfman et al., 1996). It appears that mucin sulfation may be elevated both in CF patients and in individuals suffering from bronchitis with the degree of sulfation relating to the severity of the infection (Davril et al., 1999). Many previous studies have reported total sulfate content of mucins (Boat et al., 1976; Chace et al., 1983; Carnoy et al., 1993; Davril et al., 1999), but only a few studies have identified individual sugar sulfates, such as GlcNAc-6-Sul, Gal-3-Sul, Gal-4-Sul, and Gal-6-Sul (Lamblin et al., 1991; Mawhinney et al., 1992a,b; Sangadala et al., 1993; Lo-Guidice et al., 1994; Thomsson et al., 1998; Degroote et al., 2003), and there have been uncertainties about the sugar sulfate species present in airway mucins. In addition, there has been no relative determination of individual sugar sulfate species in CF versus ND mucin.
Our results show that mucins from both ND and CF donors contain GlcNAc-6-Sul, Gal-6-Sul, and Gal-3-Sul, but that CF mucins contain significantly more of the 6-sulfated sugars compared with Gal-3-Sul, which is only slightly increased in CF compared with ND mucin. One advantage of the approach we have taken is the use of intact mucin, rather than analysis of proteolytic fractions of mucins or selected sulfated glycans (Mawhinney et al., 1987, 1992a,b Lo-Guidice et al., 1994; Chance and Mawhinney, 1996). All of these studies employing analysis of selected fractions of material reflect the difficulty in working with high molecular weight mucin samples and the problems in chemical quantification of sulfated sugars. Our approach employing limited hydrolysis and the identification of recovered sulfated sugars by Dionex HPAEC using whole mucin samples may be useful in future studies to define sulfated sugar content as a function of disease and pathology. This approach is an extension of the methods developed to analyze radioactive sugar sulfates in L-selectin ligands (Hemmerich et al., 1994, 1995; Bistrup et al., 1999). Many sulfate esters of monosaccharides are relatively stable to acid treatments that can cleave glycosidic linkages and thus provides for the identification and relative measurement of sulfated sugar content of glycoproteins.
It has been estimated that there could be up to 30 O-glycans per 100 amino acids in human respiratory mucins based on their Ser/Thr content (Lamblin et al., 1991), but little is known about the actual extent of O-glycosylation in normal human respiratory mucin versus mucin from CF patients. It has been predicted that only about 60% of the Ser/Thr residues within the consensus mucin tandem repeats of human M