The Glycosylation of Rat Intestinal Muc2 Mucin Varies between Rat Strains and the Small and Large Intestine

The large glycosylated domains obtained from the rat intestinal mucin Muc2 were isolated from the large and small intestine of the inbred rat strains GOT-W and GOT-BW. The expression of the rat Muc2 in the large intestine was confirmed immunochemically and by Northern blotting. Released oligosaccharides were structurally characterized by gas chromatography-mass spectrometry (neutral and sialylated species) or by tandem mass spectrometry (sulfated species), and a total of 63 structures was assigned. The large intestinal oligosaccharides were found to be identical between the strains, while the small intestinal glycosylation differed. Until now, detailed structural analysis of oligosaccharides isolated from a single mucin core or mucin domain with different origin have not been performed, and the information of different mucin glycoforms has been limited to immunochemistry. Blood group A-determinants (GalNAcα1–3(Fucα1–2)Galβ1-, and structures related to the blood group Sda/Cad-related epitope NeuAc/NeuGcα1–3(GalNAcβ1–4)Galβ1-, were found in GOT-BW small intestine, and also in both large intestines. Blood group H-determinants and NeuAc/NeuGcα1–3Galβ1- were found in all samples. Core 1 (Galβ1–3GalNAcα1-), core 2 (Galβ1–3(GlcNAcβ1–6)GalNAcα1-), core 3 (GlcNAcβ1–3GalNAcα1-), and core 4 (GlcNAcβ1–3(GlcNAcβ1–6)GalNAcα1- were also found in all the samples. The large intestine were enriched in sulfated oligosaccharides and the small intestine contained higher amounts of sialylated species. Sulfation were found exclusively on C-6 of GlcNAc.

The family of highly glycosylated glycoproteins found at the mucosal surfaces are known as mucins (1). Characteristic for mucins is the high degree of O-linked glycosylation known as the mucin-type, where ␣-N-acetylgalactosamine is linked to serine or threonine of the protein backbone. These amino acids together with proline are mainly found in long and often repeated protein sequences, called mucin domains, that become a scaffold for the glycosylation. The mucins are largely responsible for the physical properties of mucus that serves as a lubricant for the mucosal surface and protects the underlying epithelium from mechanical and chemical stress. However, the identification of several different encoded mucins, and an enormous repertoire of possible mucin oligosaccharides indicates that the tasks for these glycoproteins may be more subtle than the macroscopic properties suggest.
Mucins produced in the intestine are mainly derived from the goblet cells. The major part (at least 80%) of the rat intestinal mucins, measured as protease-resistant mucin domains, can be recovered as an "insoluble" glycoprotein complex in as denaturating conditions as 6.0 M guanidinium chloride (2), providing that shear homogenization is avoided. Two highly glycosylated peptides, named glycopeptide A (gpA) 1 and B (gpB), 650 kDa and 335 kDa, respectively, were isolated by trypsin digestion of subunits from this insoluble mucin complex. A polyclonal antiserum raised against the deglycosylated gpA was used to isolate a cDNA clone (3). This sequence together with two other partial sequences (4 -6) have been shown to have homology within separate regions of the intestinal human mucin MUC2 (7), proposing that all three are parts of the rat Muc2 gene. The organization of rat Muc2 mucin has recently been further explored and compared with its human homologue (8). The Muc2 mucin has two large domains of sequences rich in serine, threonine, and proline residues, with a small cysteine-rich region in between. It was concluded that the gpB was the smaller domain at the N-terminal side, while gpA was the larger one at the C-terminal side. The two highly glycosylated mucin domains are flanked by less glycosylated cysteine-rich regions. The Nand C-terminal of the human MUC2 and rat Muc2 are believed to be responsible for oligomerization via intramolecular disulfide bridges. Recent biosynthetic studies in the colon cancer cell line LS 174T have shown that a disulfide-stabilized dimer of MUC2 is formed (9), similar to the initial biosynthesis of the von Willebrand factor.
Several reports elucidating the enormous diversity of mucin oligosaccharides do not provide any further information about possible heterogeneity due to the presence of different mucin subpopulations (for example, Refs. 10 and 11). The glycosylation of separate mucins or parts of mucins is rarely described, but there is a growing interest due to the fact that the presentation of oligosaccharides will most likely be dependent on the protein backbone (12). We have previously reported a detailed structural characterization of the oligosaccharides from gpA and gpB from an inbred white rat strain (GOT-W), known to express H-determinants but not A-determinants in the small intestine (2,8). It was concluded that the glycosylation of the two domains was similar. To further explore the glycosylation of a mucin domain from Muc2, we studied the glycosylation of the small intestinal gpA from a black-and-white inbred rat strain (GOT-BW). This strain has been shown to express blood group A positive glycosphingolipids in the small intestine (13). The analysis was also extended by the identification of oligosaccharides from gpA of rat large intestine of both rat strains, where both strains express blood group A positive glycosphingolipids (14). The question addressed was to what extent a single mucin core exhibit a tissue-dependent glycosylation and if it is influenced by the variation of the glycosylation between individuals. Detailed qualitative and quantitative information from GC-MS, FAB-MS, and MS/MS revealed that the Muc2-derived mucin domain gpA existed as different glycoforms in the small and large intestine. It was further concluded that the small intestinal glycosylation of this domain differed between the two inbred rat strains.

Isolation of Intestinal Mucin Glycopeptide A-
The mucin gpA was prepared from mucosal scrapings essentially as described (2) from GOT-W and GOT-BW rats small intestines (17 and 9 animals, respectively) and large intestines (7 and 5 animals, respectively). Shortly the procedure involved extracting insoluble material with 95% ethanol, followed by reduction and alkylation with dithiothreitol and iodoacetamide in 6.0 M guanidinium chloride. The samples were digested with RNase and DNase and two highly glycosylated domains (gpA and gpB) were purified on Sephacryl S-200 and S-500 after trypsin digestion.
Measurement of A-Transferase Enzyme Activity-The large and small intestine were removed and thoroughly washed with phosphatebuffered saline. A 40-cm piece of the small intestine was removed 40 cm from the distal end and mucosal scrapping was collected. Mucosal scrapping was also collected from the entire large intestine. The samples were homogenized with a Dounce homogenizer (tight piston, 25 strokes) in 35 ml of phosphate-buffered saline and centrifuged at 25,000 ϫ g for 20 min. The pellet was reconstituted in phosphatebuffered saline (2 ml for the small intestine and 1 ml for the large) and stored frozen at Ϫ20°C. For the enzyme assay, 20 g of LNF-1 (H type 1 pentasaccharide) in 10 l of cacodylate buffer, 0.50 M, with 1.5 M NaCl, pH 6.8, was used as substrate. After the addition of 10 l of ATP (100 mM), 10 l of MnCl 2 (150 mM) with 100 mM NaN 3 and 1.0% Triton X-100, 5 l of UDP-[ 14 C]GalNAc (0.12 Ci), and 65 l of the redissolved pellet, the mixture was incubated for 18 h at 37°C and the reaction stopped by heating to 90°C for 30 min. The reaction mixture was desalted on a column with a mixture of AG3-X4A and AG 50W-X8 (2 ml) and eluted with 5 ml of water, lyophilized, and redissolved in 100 l of TBS. An affinity column (3 mm ϫ 10 cm) was prepared as described (15) after washing concanavalin A-Sepharose for 1.5 h (6 ml/h) with Tris-HCl-buffered saline (50 mM) (TBS). The monoclonal anti-blood group A antibody A003 (Monocarb) (10 mg) was added at 20°C, followed by 3 h washing with TBS (6 ml/h). An aliquot (25 l) of the reaction mixture from above was applied to the affinity column at 37°C and fractions of 7 drops (approximately 250 l) were collected. After the addition of 3 ml of liquid scintillation mixture (Ready Safe, Beckman Instruments, Fullerton, CA) the radioactivity was measured by a Beckman LS6000TA liquid scintillator.
Detection of Rat Muc2 Expression-GpA (0.4 g) was coated into microtiter plates (Maxisorb, Nalge Nunc, Roskilde, Denmark) by slow evaporation of a water solution at 37°C for 12 h. The plates were further dried in an exicator for 2 h and treated with gaseous hydrogen fluoride in an HF-apparatus (Peptide Institute, Tokyo, Japan) for 18 h. The samples were recoated in their original wells by adding 100 l of phosphate-buffered saline and incubating for 24 h at 37°C. The strips were washed twice with 5 mM Tris-HCl buffer, pH 8.0, containing 0.15 M NaCl, 0.005% Tween 20, and 0.02% NaN 3 and once with 0.15 M NaCl followed by blocking with 0.1% bovine serum albumin and 6% sorbitol in 50 mM Tris-HCl buffer, pH 8.0, with 0.15 M NaCl, 90 M CaCl 2 , 4 M EDTA, and 0.02 M NaN 3 for 6 h at 37°C. The dissociation enhanced lanthanide fluoroimmunoassay was performed as described (16), using a rabbit polyclonal antibody against the deglycosylated gpA from GOT-W small intestine (␣-gpA, PH497) (3) or serum from unimmunized rabbit. As secondary antibody a goat anti-rabbit antiserum (Jackson Immunoresearh, West Grove, PA) labeled with europium was used.
Northern Blot Analysis-mRNA (from approximately 50 g of total RNA) from GOT-W large and small intestine were prepared from mucosal scrapping, electrophoresed, blotted, and probed with the VR-1A probe as described (3,8).
Release and Fractionation of Oligosaccharides-Oligosaccharides from gpA of the various preparations (4 -20 mg) were released by ␤-elimination in 0.05 KOH with 1.0 M NaBH 4 for 45 h at 45°C (1 ml/mg glycopeptide). The reactions were quenched by adding acetic acid, followed by desalting using an AG 50W-X8 column (1.5 ml of resin/ml reaction solution) eluted with water (5 ml/ml resin), and repeated treatment with acetic acid in methanol with subsequent evaporation. Oligosaccharides were applied to DE23 cellulose (Ac Ϫ form, 1-2 g), and neutral oligosaccharides were eluted with 50 -100 ml of water (containing 1% 1-butanol), followed by elution of the acidic ones with 1.0 M pyridinium acetate, pH 5.4, and the fractions were lyophilized. An aliquot of the acidic oligosaccharides, or alternatively an aliquot of the total fraction of mucin oligosaccharides (from 2 to 10 mg of glycoprotein), was applied to DEAE-Sephadex A-25 (Ac Ϫ form, 0.5 ml of packed resin/mg of glycopeptide). Neutral oligosaccharides (if present) were eluted with methanol (5 ml/ml resin), sialylated oligosaccharides were methyl esterified as described (17) by loading 200 l of dimethyl sulfoxide/MeI (5:1) per ml of DEAE-Sephadex resin and incubating for 5 min. This procedure was repeated 3 times, and the sialylated oligosaccharides were eluted with 5 ml of dry methanol/ml resin. Finally, the sulfated oligosaccharides were eluted with 1.0 M pyridinium acetate, pH 5.4. Solvents were removed by rotary evaporation and lyophilization. Sulfated oligosaccharides were desalted by a G-10 column (Pharmacia, Sweden) 16 m ϫ 400-mm eluted with water containing 1% 1-butanol. Sialylated species (from 2 to 10 mg of gpA) were converted from their methylesters into N-methylamides by stirring for 10 min in methanol containing 6 -12% methylamine.
Monosaccharide and Amino Acid Analysis-Analyses of neutral monosaccharides (1/100 -1/20 of either the neutral, sialylated, or sulfated species) were performed after acidic hydrolysis in 4.0 M trifluoroacetic acid followed by either conversion into alditol acetates (18), and analysis by GC, or by re-N-acetylation followed by analyzes using high performance anion exchange chromatography-pulsed amperometric detection (19). The amino acid composition of the glycopeptides was determined as described (20) with an Alpha Plus amino acid analyzer (Pharmacia).
Characterization of Neutral and Sialylated Oligosaccharides-The neutral oligosaccharides and the N-methylamide derivative of sialylated ones were permethylated with methyl iodide in a slurry of NaOH in dimethyl sulfoxide as described (21,22). Samples were analyzed by GC and GC-MS after dissolving in ethyl acetate (25-100 l) and 0.5-1 l were injected on-column at 70°C. Permethylated N,N-dimethyl amides of the sialylated species were purified on a Sephadex LH-20 column (7 ϫ 500 mm) eluted with methanol (0.25 ml/min), before being analyzed by GC and GC-MS. High-temperature capillary columns for GC were prepared as described (17) from fused silica capillaries (11-12 m ϫ 0.25 mm, inner diameter, HT-polyimide coated, Chrompack, Middelburg, The Netherlands) which were coated with 0.02-0.04 m of PS264 or SE-54 and cross-linked. Capillary GC was performed on a Hewlett-Packard 5890A gas chromatograph with hydrogen as carrier gas (0.7 bar, linear gas velocity of 114 cm/s at 70°C) including an oxygen trap (Oxypurge, Alltech) in the carrier gas line. The flame ionization detector was kept at 390 or 395°C. Sialylated oligosaccharides were analyzed by a temperature program from 70°C (1 min) to 200°C by 50°C/min and then by 10°C/min to 390°C (5 min). Neutral oligosaccharides were analyzed by a linear temperature program from 70°C (1 min) up to 395°C (5 min) at 10°C/min. For GC-MS, helium was used as carrier gas (0.2 bar, linear gas velocity of 75 cm/s at 70°C) and a Hewlett-Packard 5890A-II gas chromatograph working in a constant flow mode. The gas chromatograph was coupled to a JEOL SX-102A mass spectrometer (JEOL, Tokyo, Japan). The conditions for the mass spectrometer: interface temperature, 385°C; ion source temperature, 370°C; electron energy, 70 eV; trap current, 300 A; acceleration voltage, ϩ10 kV; mass range scanned, m/z 100 -1600; total cycle time, 1.4 -1.8 s; resolution, 1400 (m/⌬m 10% valley); pressure in the ion source region, 5 ϫ 10 Ϫ4 pascal.
Characterization of Sulfated Oligosaccharides-Sulfated species were peracetylated with 100 -400 l of pyridine and 50 -200 l of acetic acid anhydride-d 6 for 12 h in room temperature before analyzing the sample with negative ion FAB-MS and MS/MS. Peracetylated sulfated oligosaccharides were dissolved in 50 -100 l of methanol and 1-2 l was mixed with triethanolamine and ionized by fast atom bombardment for analyzing negative ions. The FAB-MS of the mixture of peracetylated sulfated mucin oligosaccharides was done on MS1 of a JEOL HX/HX 110A four sector tandem mass spectrometer scanning m/z from 100 to 3000 with a cyclic time of 37 s and a linear magnet scan. The ions were accelerated to 10 keV. MS/MS was performed as described (23) by selecting [M-H] Ϫ of the sulfated oligosaccharides as primary ions for CID. In the field free region between the mass spectrometers the primary ion was attenuated by approximately 70% in a collision cell (floating at Ϫ8 kV) with helium as collision gas and a collision energy of 2 keV. CID spectra were obtained from m/z 70 up to 30 atomic mass units below the precursor ions. The resolution was set to 1000 in both MS1 and MS2 and the daughter ions were detected with a JEOL MS-ADS II focal plane array detector.

RESULTS
Isolation of Glycopeptide A-By reduction and alkylation of mucosa from the large intestine of the GOT-W strain and the small and large intestine of the GOT-BW strain followed by nuclease and trypsin digestion and Sephacryl S-200 and S-500 gel filtration, two highly glycosylated mucin domains (gpA and gpB, respectively) were isolated (Fig. 1). Isolation of the two mucin domains from GOT-W small intestine and their relation to the rat Muc2 mucin have already been reported (2,8). The reactivity of the polyclonal antiserum (␣-gpA, reacting with rat Muc2) with the hydrogen fluoride deglycosylated gpA isolated from the various sources (Table I) it could be concluded that they all contain the same major peptide part. The lower reactivity of the large intestinal glycopeptides could reflect difficulties in purifying the significantly smaller amounts of material from this tissue.
So far, rat Muc2 has only been known to be expressed in the rat lung and small intestine (3)(4)(5)(6). To verify the observation that Muc2 is present in the large intestine, mRNA was isolated from GOT-W large intestine and Northern blot was performed using the VR-1A clone from rat Muc2 (3,8) as probe. A single band was detected with identical size (Ͼ12 kilobases) as for Muc2 in the GOT-W small intestine (Fig. 2). The amino acid analysis of GOT-W large intestinal gpA showed a high content of proline, serine, and threonine, with a similar distribution as the GOT-W small intestinal gpA. However, an increased level  of especially glycine, glutamine/glutamate, and lysine was found. This result together with the slightly lower reactivity of the gpA from large intestine with the ␣-gpA antiserum (Table I), indicated that there may be additional unidentified biomolecules within these samples. Alternatively, this is a reflection of the variation in glycosylation between the large and small intestine, causing a difference in the size of the tryptic gpA fragments isolated from different sources. The different Muc2 content of the gpA preparations from GOT-W and GOT-BW large intestine, measured as the reactivity with ␣-gpA antiserum, did not render in different glycosylation (described below). This indicated that glycosylation of the large intestinal gpA as presented here reflected the glycosylation of the rat Muc2 large mucin domain from this tissue, even though small amounts of other glycosylated molecules could be present within the samples.
Blood Group Reactivity of Glycopeptide A-From the reactivity with blood group specific antibodies with gpA from the small and large intestine of the GOT-W and GOT-BW rat strains it was concluded that the glycosylation was different both between the tissues and strains (Table I). Blood group A-determinants were found in all of the gpA except for the small intestinal gpA from GOT-W. This correlated with the previously analyzed expression of blood group A active glycosphingolipids (24). To confirm that the presence of blood group A epitopes was due to the action of an GalNAc␣1-3 transferase using blood group H epitopes as substrate, an in vitro enzyme assay was used. The LNF-1 oligosaccharide was used as substrate and UDP-[ 14 C]GalNAc as sugar nucleotide. After incubation with tissue homogenates the mixtures were analyzed using a blood group A affinity column, where oligosaccharides containing A-determinants are retarded (15). Indeed, the assay specifically excluded the GOT-W small intestine from having A-transferase activity, while the other samples all had an active enzyme, as shown by the retarded signal from [ 14 C]Gal-NAc-labeled oligosaccharides (Fig. 3).
Monosaccharide Analyzes of Neutral, Sialylated, and Sulfated Oligosaccharides-The oligosaccharides were released and fractionated into neutral, sialylated, and sulfated species. The number of GalNAcol in each of the neutral (not retarded), sialylated (esterifyable), and sulfated (high salt buffer eluted) fractions is a measure of the number of oligosaccharide chains recovered in each fraction. From Table II it can be concluded that while the distribution within the three oligosaccharide categories from the large intestine was very     similar between the strains, the glycosylation of small intestinal gpA differed both from each other, and from the large intestinal gpA. The same was also true when the general features of the monosaccharide composition were compared (Table III). It is thus proposed, that although the GOT-BW small intestine have the blood group A activity in common with the two large intestinal gpA, the degree of sulfation of the GOT-BW small intestinal gpA is lower (5% of the sugars) compared with that of the large intestine (46%). The sulfation of the GOT-W small intestinal glycopeptides were in between these (20% sulfated sugars). Neutral Oligosaccharides from Glycopeptide A-The monosaccharide composition indicated that the glycosylation of the large intestinal gpA from the strains were similar, while the small intestinal gpA were not. To substantiate this observation the neutral oligosaccharides were permethylated and analyzed by high-temperature capillary GC and GC-MS. The gas chromatograms of the four permethylated neutral samples are shown in Fig. 4, where the numbering system refers to structures assigned from GC-MS gathered in Table IV. The interpretation of the structures was based on the intense B i and Z i fragment ions, while the determination of the C-3 and C-6 branch of GalNAc was mainly based on the ␣-cleavage between C-4 and C-5 of the GalNAcol (2). The fragmentation is exemplified by the mass spectrum of the blood group A type structure 6.6 from GOT-BW small intestine (Fig. 5). The oxonium fragment ions (B i ) at m/z 189, m/z 260, m/z 638, and m/z 883 revealed a branched A-type tetrasaccharide sequence HexNAc-(Fuc-)Hex-HexNAc-. The fragment ion m/z 1087 show that this sequence was attached to C-3 of the GalNAcol moiety. The inductive Z i fragment ions m/z 521 and m/z 766 concluded the total sequence to HexNAc-(Fuc-)Hex-HexNAc-3(HexNAc-6)-HexNAcol. The fragment ion at m/z 693 was from a further fragmentation of the oxonium ion at m/z 883 by the loss of 1 fucose unit and one additional proton. Empirically this kind of fragmentation has been described to occur only for type 1 chains (Gal␤1-3GlcNAc␤1-) and not type 2 chains (Gal␤1-4GlcNAc␤1-) (25).
The results from GC and GC-MS imply that the glycosylation of all the gpAs was conducted by a similar initial glycosylation, and that the oligosaccharides became more diversified during the subsequent elongation. The core structures of the oligosaccharides were the same in all the samples (Table IV); core 1 (Gal␤1-3GalNAc␣1-), core 2 (Gal␤1-3(GlcNAc␤1-6)GalNAc␣1-), core 3 (GlcNAc␤1-3GalNAc␣1-), and core 4 (GlcNAc␤1-3(Glc-NAc␤1-6)GalNAc␣1-). A family of oligosaccharides (structure 3.4, 4.5, and 6.5, Table IV) was found exclusively in the large intestine. This family contained a series of oligosaccharides with a GlcNAc linked to Gal. GlcNAc was otherwise only found directly linked to the GalNAcol. The length of the oligosaccharides in all samples was estimated using the relative amount of GalNAcol. Overall, the oligosaccharides were relatively short in all four samples, with the average length ranging from 4.7 to 6.1 residues. The same estimations based on the relative amounts given in Table IV would be between 3.1 and 3.7 residues. This discrepancy could be explained by the presence of larger oligosaccharides not analyzable using GC-MS, but may also reflect the presence of peptides with unreleased oligosaccharides attached.
Comparing results from GC and GC-MS ( Fig. 4 and Table  IV), one can conclude that the large intestinal neutral oligosaccharides were similar, if not identical, between the strains, but differed from the oligosaccharides from the small intestinal gpA. The most obvious difference was that the oligosaccharides from gpA in the large intestine were dominated by blood group A type structure 4.3 (GalNAc-3(Fuc-2)Gal-3GalNAcol) while the small intestinal oligosaccharides, both in GOT-W and GOT-BW, were dominated by the cognate precursor 3.1 (Fuc-2Gal-3GalNAcol). In general, the blood group A activity, measured as the ratio of any blood group A type structure in Table IV divided by its blood group H type precursor, was higher in the two large intestines, than in the GOT-BW small intestine. The large  Table IV) from GOT-BW small intestine. Positive EI-spectrum from the average of 9 scans from GC-MS of a hexasaccharide with a blood group A-determinant at the C-3 branch of the GalNAcol corresponding to the component with a retention time of 30.7 min in the gas chromatogram (Fig. 4).  (Table V) assigned by GC-MS.
intestinal samples were also shown to contain more unidentified contaminants in the gas chromatogram, indicating some impurities from the preparation of these mucin glycopeptides.
Sialylated Oligosaccharides from Glycopeptide A-Mass chromatograms of the oxonium fragment ions at m/z 389 [NeuAc] ϩ and at m/z 419 [NeuGc] ϩ are shown in Fig. 6. The fragmentation of the permethylated N,N-dimethylamide derivatives of the sialylated oligosaccharides is similar to that of the neutral ones, with the addition of pronounced fragmentation of the molecular ion due to an ␣-cleavage of the N,N-dimethylamide side group of the sialic acids, giving an oxonium ion [M-CON(CH 3 )] ϩ (26). The fragmentation is exemplified by component 4.2 with the sequence NeuGc-(HexNAc-)Hex-3HexNAcol from GOT-W large intestine (Fig. 7). The [M-CON(CH 3 )] ϩ fragment ion at m/z 1088 showed that the oligosaccharide have the composition NeuGc, Hex, HexNAc, HexNAcol, in the ratio 1:1:1:1 and the indicative fragment ions at m/z 260 (terminal HexNAc), m/z 276 (terminal HexNAcol), and m/z 419 (terminal NeuGc), revealed a branched sequence. The fragment ion at m/z 1072 from an ␣-cleavage between C-4 and C-5 of the GalNAcol concluded the structure.
Sialic acid was found almost exclusively attached to galactose in all samples (Table V), even though gpA of the large intestines and in GOT-BW small intestine also contained disaccharide structures with sialic acid attached directly to C-6 of GalNAcol, indicating the presence of sialyltransferases with additional specificity in these locations.
Sulfated Oligosaccharides from Glycopeptide A-Sulfated oligosaccharides were analyzed as their perdeuteroacetylated derivative by negative FAB-MS, and in the case of high abundant pseudomolecular [M-H] Ϫ ions also by MS/MS (Table VI) Table V)    gpA have already been analyzed in detail by this technique (8) and the results are included as a comparison with the other gpA preparations. In the mass spectra obtained from the MS1 of the tandem mass spectrometer, [M-H] Ϫ ions were detected ranging from monosulfated trisaccharides up to octasaccharides (Fig. 8). The abundances given in Table VI of  Similar to the pig small intestine, the rat intestinal oligosaccharide sulfation was mainly directed to C-6 of the GlcNAc residue adjacent to the GalNAcol residue (23). A C-6 sulfation is also concluded from CID-MS/MS spectra of less intense [M-H] Ϫ ions, even though no complete sequence information could be obtained. The position of the sulfate group could be deduced from the presence of the fragment ion at m/z 139 (CHO-CHOSO 3 Ϫ ) typical for C-6 sulfation, and the absence of TABLE VI Composition and structural assignment of sulfated oligosaccharides from gpA of the two strains GOT-W and GOT-BW fragment ions arising from C-3 and C-4 sulfation (result not shown).

DISCUSSION
Previous investigations have established that rat Muc2 is expressed in both the small intestine and in the lungs (4 -6). Here, we presented evidence that this mucin is also present in the large intestine, using both immunochemical detection and identification of an mRNA transcript with the same size as rat small intestinal Muc2. The human MUC2 homologue is also expressed both in the large and small intestine (28). The slightly lower Muc2 reactivity of gpA from the large intestine (Table I) and the differences in amino acid composition (Fig. 2) suggested that there could be other unidentified material present in these preparations. In addition to the large mucin glycopeptide (gpA) in rat Muc2 the glycosylation of the smaller glycopeptide (gpB) (Fig. 1) has also been analyzed. The glycosylation of gpA and gpB from the small intestine of the GOT-W strain has been shown to be almost identical (2). This was also observed for the glycosylation of the corresponding gpA and gpB from the large intestine from GOT-W and GOT-BW. 2 This suggests that the contaminations observed in gpA from the large intestine was not affecting the results.
The glycosylation of the large mucin domain of the Muc2 in the rat intestine was demonstrated to be different in the large and small intestine, but also that there is a variation of the glycosylation in the small intestine between rat strains. Previous investigations of rat intestinal mucin oligosaccharides have revealed that both core 1, core 2, core 3, and core 4 oligosaccharides are present (2,29). These core structures were also found here. However, we found no obvious differences in the abundance of these core structures in the two rat strains and between the small and large intestine, indicating a similar regulation of the initial oligosaccharide biosynthesis.
The expression of blood group A-determinants in the GOT-W and GOT-BW strain large intestine as well as in the GOT-BW but not in GOT-W small intestine, have previously been noted for the glycosphingolipids (24). Here, the oligosaccharides from the large glycosylated tandem repeat of rat Muc2 mucin demonstrated the same feature shown both by GC-MS (Table II), and by using blood group A-directed monoclonal antibodies (Table I). The findings of blood group Adeterminant containing structures in GOT-BW small intestine, and GOT-BW and GOT-W large intestine correlated with the measured blood group A enzyme activity in homogenates from these tissues.
Among the sialylated oligosaccharides a family of structures (NeuAc/NeuGc␣2-(GalNAc1-)Gal1-3(R1-6)GalNAcol) was found, referred to as Sd a /Cad-like, due to its structural similarity with two closely related human blood group antigens. The structure NeuAc␣2-3(GalNAc␤1-4)Gal␤1-3(NeuAc␣2-6)GalNAcol has been isolated from erythrocytes of Cad-positive individuals (30) and is believed to be involved in the rare Cad-blood group antigen, while the more common Sd a -blood group (92% of Caucasian population are positive (31)), is shown Collision induced dissociation mass spectrum by 70% attenuation of the precursor ion by helium gas in the collision cell placed in the field free region between the tandemly arranged mass spectrometers. The isomeric trisaccharide sequences of the unsulfated C-3 branch of the GalNAcol were found to be a linear H-determinant Fuc-Gal-GlcNAc-and a branched A-determinant Fuc-(GalNAc-)Gal-by pronounced 1,5 X i -cleavages within the pyranose ring, and the cleavages of the glycosidic bond (Z i -cleavages). Fragmentation was occasionally followed by further loss of acetate-d 3 (mass of 62 Da) or acetyl-d 3 (mass of 46 Da).
to be expressed on both glycosphingolipids and glycoproteins. The Sd a -active structures share the trisaccharide epitope with the G M2 -glycosphingolipid (NeuAc␣2-3(GalNAc␤1-4)Gal␤1-), but the Sd a structure requires an extra N-acetylglucosamine in a ␤1-4 linkage (32). The biosynthetic pathways are believed to be similar for G M2 -glycosphingolipid and Sd a /Cad active structures, where GalNAc␤1-4 is added to the NeuAc␣2-3Gal␤1moiety. The blood group A expression coincided with the appearance of the rat intestinal Sd a /Cad-like structures (GOT-BW small intestine and both the GOT-W and the GOT-BW large intestine). Both are due to the activity of GalNAc transferases, but while the A-determinants were expressed both on the mucin and on the glycosphingolipids, the Sd a /Cad-like structures have no homologues on the glycosphingolipids. In contrast, the large intestinal glycosphingolipids express blood group B-antigens, while the mucin oligosaccharides do not (Table VII). Among the Muc2 oligosaccharides there was no evidence for difucosylated epitopes similar to the ALe x -and BLe xepitopes found on glycosphingolipids in the rat large intestine. The knowledge of a diversified glycosylation of the mucosal layer and the underlying epithelium should be important when investigating the interaction and colonization of microbes on mucosal surfaces.
While the small intestinal glycosylation may vary between strains, it is tempting to speculate about the preserved large intestinal glycosylation between the two strains. Intestinal dwelling microbes, preferentially found in the large intestine, are believed to digest mucin oligosaccharides by secreting various linkage specific exoglycosidases (35)(36)(37). The presence of A-determinants and Sd a /Cad-like structures should thus require a larger battery of exoglycosidases. Interestingly, the removal of the sialic acid has been suggested as an initial step for the bacterial digestion of the mucus, and the trisaccharide unit NeuAc␣2-3(GalNAc␤1-4)Gal␤1-has been shown to withstand the action of various sialidases (32). This is the first description of the glycosylation variation of a defined mucin domain, the large domain of rat Muc2. This illustrates the necessity of identifying a mucin not only by its apomucin, but also by its oligosaccharides. The identification of three glycoforms of the rat Muc2 mucin, presumably related to each other only by the regulation of glycosyltransferase activities, implies that this is a flexible system, capable of facing the demand of a non-static environment. That the change of the oligosaccharides will influence the molecular milieu in the intestine is undoubtful. An interesting question is whether the glycosylation also can alter the macroscopic behavior of a mucin by, for example, lectin-like interaction or carbohydratecarbohydrate interactions.