Isolation and Characterization of Major Glycoproteins of Pigeon Egg White

Ovotransferrin (POT), two ovalbumins (POA(hi) and POA(lo)), and ovomucoid (POM) were isolated from pigeon egg white (PEW). Unlike their chicken egg white counterparts, PEW glycoproteins contain terminal Galα1–4Gal, as evidenced by GS-I lectin (specific for terminal α-Gal), anti-P1(Galα1–4Galβ1–4GlcNAcβ1–3Galβ1–4Glcβ1–1Cer) monoclonal antibody, and P fimbriae on uropathogenic Escherichia coli(specific for Galα1–4Gal). Galα1–4Gal on PEW glycoproteins were found in N-glycans releasable by treatment with glycoamidase F. The respective contents of N-glycans in each glycoprotein were 3.5%, POT; 17%, POA(hi); and 31–37%, POM. POA(hi) has four N-glycosylation sites, in contrast to chicken ovalbumin, which has only one. High performance liquid chromatography analysis showed that N-glycans on POA(hi) were highly heterogeneous. Mass spectrometric analysis revealed that the major N-glycans were monosialylated tri-, tetra-, and penta-antennary oligosaccharides containing terminal Galα1–4Gal with or without bisecting N-acetylglucosamine. Oligosaccharide chains terminating in Galα1–4Gal are rare amongN-glycans from the mammals and avians that have been studied, and our finding is the first predominant presence of (Galα1–4Gal)-terminated N-glycans.

However, ovomucoids produced in pigeon (Columba livia) and turtle doves (Streptopelia resoria) possess high level of P 1 antigenic activity (21)(22)(23)(24). Pigeon ovomucoid (POM), 1 one of the major glycoproteins in pigeon egg white (PEW), was also found to bind some pathogenic microbes, e.g. uropathogenic E. coli (24 -26) and S. suis (12)(13)(14). Moreover, POM had been recently utilized for isolation of Shiga-like toxin type 1 (27). Although the presence of Gal␣1-4Gal on ovomucoids of turtle dove and pigeon has been evidenced and these glycoproteins can be utilized for the study of microbiology, complete carbohydrate structures of these ovomucoids are not yet known. Only a tentative structure had been proposed for the main oligosaccharide structure of turtle dove ovomucoid, which includes Gal␣1-4Gal sequence at the non-reducing terminus of the Nglycan (28, 29).
The presence of Gal␣1-4Gal on glycoproteins is rare among modern birds and mammals that have been studied. For example, no oligosaccharides from chicken, quail, or duck ovomucoid (30 -34) contain Gal␣1-4Gal structure. Whole chicken egg white (CEW) failed to inhibit hemagglutination of P 1 erythrocytes by P-fimbriated uropathogenic E. coli, whereas PEW inhibited it successfully (24). The presence and absence of Gal␣1-4Gal in various avian egg whites exemplifies the evolutionary diversity of glycoconjugates among birds. Investigating the relationships between phylogeny of birds and distribution of oligosaccharides with Gal␣1-4Gal should help us understand how and why this unique structural feature in the N-glycans has appeared during evolution.
In this study, we examined the major pigeon egg white glycoproteins for the presence of Gal␣1-4Gal, and discovered that not only ovomucoid, but also major glycoproteins including two ovalbumins and ovotransferrin, possess N-glycan with terminal Gal␣1-4Gal. Structures of major oligosaccharides from POA(hi) were determined by mass spectrometric analysis as novel type N-glycans containing Gal␣1-4Gal.
Isolation of Major Glycoproteins from Pigeon Egg White by HPLC-Lyophilized pigeon egg white (200 mg) was dissolved with 1.5 ml of distilled water and centrifuged to remove insoluble materials. Supernatant was filtered through a 0.45-m membrane and injected onto a C4 column for HPLC (2 mg for C4 column of 4.6 ϫ 250 mm, 20 mg for C4 column of 21.2 ϫ 250 mm). The mobile phase was A (0.05% trifluoroacetic acid) and B (90% CH 3 CN in H 2 O containing 0.05% trifluoroacetic acid). The elution (1 ml/min for 4.6-mm C4 column, 8 ml/min for 21.2-mm C4 column) was by a linear gradient of 10 -70% of B in A developed over 30 min, followed by isocratic elution for 10 min. Proteins eluted were detected by A 280 nm . Individual peaks were collected, dialyzed against water in the cold, and then lyophilized.
SDS-PAGE, GS-I Lectin Blotting, and Immunoblotting-Glycoproteins from pigeon egg white were heated at 100°C for 10 min with 3% SDS and 5% (v/v) 2-mercaptoethanol and separated on 12.5% polyacrylamide gels (36), then transferred to PVDF membrane in 25 mM Trisglycine (pH 8.3) in 15% methanol. After blocking with 3% BSA with 0.1% NaN 3 overnight, the membranes were washed three times with TBST and once with Lectin Buffer 1, then incubated with 1 ml of alkaline phosphatase-conjugated GS-I lectin (1 g/ml) for 1.5 h. After washing three times with TBST, 5 ml of Lectin Buffer 2 containing 100 g/ml nitro blue tetrazolium and 200 g/ml 5-bromo-4-chloro-3-indolyl phosphate was applied to the membranes to visualize the bound GS-I, then washed three times with H 2 O. For immunoblotting with anti-P 1 mAb or anti-(Gal␣1-3Gal) mAb (mouse IgM), the transferred membranes were blocked with 3% BSA containing 0.1% NaN 3 overnight, washed three times with TBST, and incubated with the first antibody for 1.5 h. The membranes were washed with TBST and incubated with the second antibody, alkaline phosphatase-conjugated goat anti-mouse IgM, for 1.5 h. After washing with TBST, the bound antibody was visualized as described above.
Hemagglutination Assay and Measurement of Inhibitory Activity-For bacterial or lectin agglutination assays, 3 l of a suspension of P-fimbriated E. coli strain J96 (ϳ1 ϫ 10 11 colony-forming units/ml in 5% methyl ␣-D-mannopyranoside in PBS) or 3 l of GS-I lectin (1 mg/ml in 10 mM phosphate, 0.15 M NaCl, 0.5 mM CaCl 2 , pH 7.4), 3 l of inhibitors of various concentrations or 30 mg/ml BSA in PBS, and then 1 l of a 5% (v/v) suspension of human P 1 -phenotype RBCs in PBS were mixed on microscope slides. The slides were rocked for 60 s, and agglutination of RBCs was observed microscopically. Inhibitors were tested at serial 2-fold dilutions to determine the dilution beyond which inhibition first dropped to below 100% ("last full inhibition") and where it was first completely absent ("first full agglutination"). For the anti-P 1 assays, 20 l of the anti-P 1 mAb and 20 l of inhibitors or 30 mg/ml BSA in PBS were mixed in test tubes and incubated for 15 min on ice. Then, 20 l of a 5% (v/v) suspension of human P 1 -phenotype RBCs were added and incubated for 30 min on ice. The cells were spun down to a pellet and gently resuspended by rocking the tube, then scored for presence or absence and degree of RBC agglutination. Serial 10-fold dilutions of each inhibitor were tested in the anti-P 1 assays to determine the last dilution still showing inhibition of hemagglutination. Protein Sequence Analysis-PEW glycoproteins were separated by SDS-PAGE and followed by low voltage electroblotting to PVDF membrane in 10 mM CAPS, pH 11, in 10% methanol, and subjected to sequencing. N-terminal sequence analysis was carried out by Edman degradation using a PerkinElmer Life Sciences/Applied Biosystems model 494 Procise ® protein sequencer. Sequence identity search was performed using SWISS-PROT data base (updated July 25, 1999) interfaced with the Wisconsin Package of the Genetics Computer Group. Primary accession numbers for the data base are as follows: chicken ovomucoid, P01005; chicken ovotransferrin, P02789; chicken ovalbumin, P01012.
Cyanogen Bromide Cleavage and Fragment Purification-Since the N termini of both forms of POA (53 and 49 kDa by SDS-PAGE) are blocked, these proteins were submitted to chemical cleavage by BrCN in 70% formic acid at 37°C for 4 h. Following the cleavage, the mixture was diluted 10-fold with distilled water and the solution was freezedried. A single fragment from each POA was isolated either by reverse phase HPLC on a Vydac C4 column eluted with a linear gradient of acetonitrile (from 8% to 48% developed over 16 min, 0.5 ml/min) in 0.1% trifluoroacetic acid at 40°C, and subjected to N-terminal amino acid sequencing.
Monosaccharide Composition Analysis-Glycoproteins (20 g of POM, 100 g each of POT, POA(hi), and POA(lo)) were hydrolyzed with 2 M trifluoroacetic acid to release neutral sugars or with 4 M HCl to release amino sugars (37). The released monosaccharides were analyzed with HPAEC-PAD using a CarboPac PA-1 column and isocratic elution with 18 mM NaOH. To release terminal ␣-linked galactose specifically, glycoproteins were incubated at 25°C overnight with coffee bean ␣-galactosidase (15 milliunits) in 100 mM citrate-phosphate buffer (pH 6.5), and the released galactose was measured with HPAEC-PAD under the same conditions as above. Sialic acids were released from glycoproteins with sialidase from A. ureafaciens (1 milliunit) by incubating in 50 mM sodium acetate (pH 5.6) at 37°C for 2 h, and analyzed with HPAEC-PAD using a CarboPac PA-1 column and isocratic condition with 100 mM sodium acetate and 100 mM NaOH.
GAF Treatment of PEW Glycoproteins-Glycoproteins (100 g) were dissolved with 25 l of 0.5% SDS and heated at 90°C for 3 min to denature. After the solutions were cooled to room temperature, 80 mM sodium phosphate buffer (pH 8.6), 1% (v/v) Nonidet P-40, and 1 unit of GAF were added to the heat-denatured glycoproteins. The reaction mixtures (100 l each) were incubated at 37°C for 16 h to complete de-N-glycosylation, and heated at 100°C for 5 min to inactivate GAF. For partial de-N-glycosylation on POA(hi), glycoproteins were incubated with 0.4 units of GAF for 5 min, 40 min, and 3 h, and heated at 100°C for 5 min.
To release N-glycans from POM completely, POM was heat-denatured with 0.5% SDS and 100 mM 2-mercaptoethanol at 90°C for 3 min, to which 1% (w/v) CHAPS was added instead of Nonidet P-40, and incubated with GAF at 37°C overnight. This reaction mixture was heat-denatured again at 90°C for 3 min, and incubated at 37°C overnight after adding CHAPS to 1% (w/v) and 1 unit of GAF again. This step had to be repeated three times to achieve complete deglycosylation of POM, as determined by SDS-PAGE. Detergents in the reaction mixture were removed by gel filtration with Extracti-Gel ® D detergent removing gel.
Matrix-assisted Laser Desorption/Ionization (MALDI-TOF) Mass Spectrometry-MALDI-TOF mass spectrometry was performed using a PerSeptive Biosystems STR Biospectrometry research station coupled with a Delayed Extraction laser-desorption mass spectrometer. An aliquot of each sample was diluted with 0.1% trifluoroacetic acid and analyzed using matrix of 3,5-dimethoxy-4-hydroxycinnamic acid (sinapinic acid). In some cases in which signals were weak, a portion of the sample was further purified using a C8 cartridge (1 ϫ 10 mm). After washing with 0.1% trifluoroacetic acid, protein was eluted with 90% acetonitrile in 0.1% trifluoroacetic acid and analyzed in the same matrix.
Isolation of Major Oligosaccharides from Pigeon Ovalbumin (High) by HPLC-Procedures for isolation of N-linked oligosaccharides, pyridylamine (PA)-derivatization, and HPLC analysis were based on the methods established by Tomiya et al. (38,39) and modified as follows. POA(hi) (100 mg) was digested with 1 mg each of trypsin and chymotrypsin in 1.5 ml of TBS (pH 7.3) containing 5 mM CaCl 2 at 37°C for 16 h, and then heated at 100°C for 10 min. The same amounts of trypsin and chymotrypsin were added, and the digestion and heat denaturation was repeated once more. Oligosaccharides were released from the resultant glycopeptides with 20 units of GAF in the same buffer at 37°C for 16 h. The residual peptides were digested to amino acids or very short peptides by incubating with 1 mg of Pronase in the same buffer at 55°C for 16 h, and the oligosaccharide fractions were separated by gel filtration on a Sephadex G-50 column (2.5 ϫ 98 cm) in water. For the PA derivatization by reductive amination, lyophilized oligosaccharide fractions were dissolved in 80 l of 2-aminopyridine solution (1 g/580 l in concentrated HCl, pH 6.8), and heated at 90°C for 15 min with heating block. Freshly prepared NaCNBH 3 solution (14 mg/8 l) was added into the reaction mixture, then heated at 90°C for 1 h. PA-oligosaccharides were fractionated by gel filtration on a Sephadex G-15 column (1.0 ϫ 40 cm, in 10 mM NH 4 HCO 3 ) and lyophilized. The PA-oligosaccharides were purified by HPLC with three different columns. In the first step, the PA-oligosaccharide mixture was separated on a TSKgel DEAE-5PW column (7.5 ϫ 75 mm) as described previously (39), and the neutral, monosialyl, disialyl, trisialyl, and tetrasialyl fractions were collected separately and lyophilized. In the second step, monosialylated oligosaccharides were separated on a YMC-Pak HRC-ODS-A column (6.0 ϫ 150 mm). Elution was performed at a flow rate of 1.0 ml/min at 55°C using Eluent A (0.005% trifluoroacetic acid) and Eluent B (0.5% 1-butanol in Eluent A). The column was equilibrated with a mixture of Eluents A and B (80:20, v/v), and after injection of a sample, the ratio of the eluents was increased linearly to A:B ϭ 50:50 in 60 min. Each peak was collected and lyophilized. In the third step, major peaks from ODS column were further separated on a TSKgel Amido-80 column (4.6 ϫ 250 mm) as described previously (38). In all the HPLC systems, PA-oligosaccharides were monitored by fluorescence using excitation and emission wavelengths of 310 and 400 nm, respectively.

Permethylation, Fast Atom Bombardment (FAB)-and Electrospray (ESI)-Mass Spectrometry (MS)
Analyses-PA-oligosaccharides were permethylated using the NaOH/dimethyl sulfoxide slurry method as described by Dell et al. (40). FAB-and ESI-MS analyses were performed on an Autospec orthogonal acceleration-TOF mass spectrometer (Micromass), fitted with a magnet bypass flight tube, and interchangeable FAB and ESI source assemblies. For FAB-MS experiments, the fitted cesium ion gun was operated at 26 kV and the source accelerating voltage at 8 kV. Operation in the magnet bypass mode allowed an acquisition of up to a mass range of m/z 7000 with the fitted TOF detector. Cesium iodide was used as external calibrant for both the magnet and TOF mass analyzers. The permethyl derivatives of the N-glycans were redissolved in CH 3 OH for loading onto the probe tip coated with glycerol:m-nitrobenzyl alcohol:trifluoroacetic acid (50:50:1, v/v/v) as matrix. Collision-induced dissociation (CID) FAB-MS/MS was performed by introducing argon gas to the collision cell to a reading of ϳ1.2 ϫ 10 Ϫ6 millibars on the TOF ion gauge, at a lab frame collision energy of 800 eV and a push-out frequency of 56 kHz for orthogonal sampling. A 1-s integration time per spectrum was chosen for the TOF analyzer with a 0.1-s interscan delay. Individual spectra were summed for data processing. For ESI-MS, the accelerating voltage was maintained at 4 kV, which allowed magnet scanning above the mass range of m/z 5000. Hall probe calibration was used. The permethyl derivatives were dissolved in CH 3 OH, and 10-l aliquots were injected through a Rheodyne loop into the mobile phase (methanol:water:acetic acid, 50: 50:1, v/v/v), delivered at a flow rate of 5 l/min into the ESI source by a syringe pump.
Linkage Analysis-For gas chromatography (GC)-electron impact (EI)-MS linkage analysis, partially methylated alditol acetates were prepared from the permethyl derivatives by hydrolysis (2 M trifluoroacetic acid, 121°C, 2 h), reduction (10 mg/ml NaBH 4 , 25°C, 2 h), and acetylation (acetic anhydride, 100°C, 1 h). GC-EI-MS was carried out using a Hewlett-Packard Gas Chromatograph 6890 connected to a Hewlett-Packard Mass Selective Detector 5973. Sample was dissolved in hexane prior to splitless injection into a HP-5MS fused silica capillary column (30 m ϫ 0.25 mm inner diameter, Hewlett-Packard). The column head pressure was maintained at around 8.2 p.s.i. to give a constant flow rate of 1 ml/min using helium as carrier gas. Initial oven temperature was held at 60°C for 1 min, increased to 90°C in 1 min, and then to 290°C in 25 min.
Exoglycosidase Digestion-PA-oligosaccharides were first digested with 50 milliunits of sialidase (from A. ureafaciens, Roche) in 25 l of 50 mM sodium acetate buffer, pH 5.0, for 24 h at 37°C. Subsequently, samples were redissolved in 25 l of 50 mM sodium citrate phosphate buffer, pH 4.5, and 5 l of ␤-galactosidase (20 milliunits, from bovine testes; Roche) was added. After 24 h at 37°C, 500 milliunits of N-acetyl-␤-D-hexosaminidase (from jack bean, Calbiochem) in 20 l of the same buffer, pH 5, was added and the incubation was left for another 24 h. The digestion products were desalted by passing through a Sep-Pak C18 cartridge (Waters), washed with water, and the PA-oligosaccharides were then eluted with 50% methanol in water.

Analysis of PEW Proteins on SDS-PAGE and Their
Identification by Partial Peptide Sequencing-The SDS-PAGE profile of PEW revealed four major bands, and overall was considerably different from the profile of chicken egg white (Fig. 1A). N-terminal amino acid sequencing of PEW proteins separated by SDS-PAGE indicated that the highest (APQKASVRWXTIS-SAEEKKXNNLRE) and the lowest (VEVDXSRYHXTTNVEG-REGL) molecular size proteins of these four were homologous to chicken ovotransferrin (COT) and ovomucoid (COM), respectively. The N termini of the middle two proteins were fractal to sequencing by Edman degradation. This may be attributable to N-terminal blockage as in the case of chicken ovalbumin (COA), which is acetylated at the N terminus. Internal peptide sequences of the higher (MERKRVKVYLPRM and MPFRVTE-QESKPVQM) and lower (MLYLGARGNTKAQIDKVVHFD A, profiles of proteins from pigeon and chicken egg whites by SDS-PAGE. Ten g of PEW and CEW were heat-denatured with sample buffer containing 3% SDS and 5% 2-mercaptoethanol and loaded to a 12.5% polyacrylamide gel as described under "Experimental Procedures." After electrophoresis, the gel was stained with CBB. B, detection of ␣-galactosyl residue on PEW by lectin blotting and immunoblotting. Five g of glycoproteins were separated by SDS-PAGE, transferred to PVDF membranes, and stained with GS-I lectin, anti-P 1 mAb, or anti-(Gal␣1-3Gal) mAb. Mouse laminin and its fragments (MLN) are positive controls for glycoproteins possessing Gal␣1-3Gal. and MGLGITDLFSSXADLSGISSV) molecular size proteins indicated high homologies to those of COA.
Detection of ␣-Galactoside on PEW Proteins by GS-I Lectin, Anti-P 1 mAb, and Anti-(Gal␣1-3Gal) mAb-The bands of PEW proteins separated by SDS-PAGE were transferred onto PVDF membrane and stained with alkaline phosphatase-conjugated GS-I lectin, which binds terminal ␣-Gal residues specifically. All four major glycoproteins were stained with GS-I lectin, as shown in Fig. 1B. POM showed especially intense bands by GS-I, whereas it showed only weak bands by staining with Coomassie Brilliant Blue (CBB). None of the glycoproteins from chicken egg white were stained with GS-I lectin (data not shown).
These pigeon glycoproteins were also stained positively with anti-P 1 mAb, specific for P 1 blood type (Fig. 1B). Mouse laminin, which contains Gal␣1-3Gal (42,43), was not stained with anti-P 1 mAb but was stained with GS-I lectin (Fig. 1B). On the other hand, the glycoproteins from PEW were stained with anti-(Gal␣1-3Gal) mAb only feebly, whereas mouse laminin was stained clearly (Fig. 1B). The four major PEW glycoproteins also did not react with human polyclonal antibody purified with (Gal␣1-3Gal)-Sepharose (tested with microplatecoating method). These data indicate that all four major glycoproteins from PEW contain an epitope most likely to be Gal␣1-4Gal, but not Gal␣1-3Gal.
Hemagglutination Assay-Human P 1 -erythrocytes were agglutinated by three different agglutinators: P-fimbriated E. coli strain J96 (PapG class I-and III-positive), GS-I lectin, and anti-P 1 antibody. All three are capable of binding to P 1 antigen, although their specificities are different. Hemagglutination by type I fimbriae on E. coli was eliminated in this assay by addition of excess methyl ␣-D-mannopyranoside. Table I shows the activities of whole PEW and individual PEW glycoproteins in inhibiting the hemagglutination. The inhibitory activities for agglutination mediated by E. coli or GS-I lectin are indicated as last full inhibition and first full agglutination as defined under "Experimental Procedures," and for agglutination mediated by anti-P 1 antibody are indicated as the final concentrations of inhibitors giving detectable inhibition. Varying degrees of inhibition of hemagglutination by E. coli strain J96, GS-I, and anti-P 1 antibody were observed with all four PEW glycoproteins as well as with whole PEW, whereas no inhibition was detected with 12.9 mg/ml BSA. The limited solubility in PBS of some inhibitors precluded accurate measurement of the concentration required for full inhibition. POM exhibited the highest inhibitory activity among PEW glycoproteins against agglutination by E. coli strain J96, while POA(hi) was a more potent inhibitor of agglutination mediated by GS-I and anti-P 1 antibody. ␣-Galactosidase-treated POM showed a significant reduction of the hemagglutination inhibitory activity, suggesting that ␣-galactosyl residues on the glycoprotein are responsible for the inhibition.
Analysis of Monosaccharide Composition and Release of Galactose by ␣-Galactosidase-The monosaccharide composition of HPLC-purified PEW glycoproteins are shown in Table II. All four glycoproteins contain GlcNAc, Man, Gal, and NeuAc, which are commonly found in N-glycans. No Fuc or GalNAc were detected. POM, POT, and POAs contain more Gal than their chicken counterparts, whose N-glycans mostly consist of Man and GlcNAc with some ␤-Gal and NeuAc (30,31,(45)(46)(47)(48)(49)(50). The ratios of GlcN:Man:Gal were similar among POM, POA(hi), and POA(lo), whereas POT had a lower content of Gal. POM had more NeuAc than others. ␣-Galactosidase released 15-18% of total Gal from POM and POAs, and 37.5% of total Gal from POT. These data provide evidence that the PEW glycoproteins are rich in terminal ␣-galactosides.
Release of Oligosaccharides Containing ␣-Galactoside from POT, POA(hi), POA(lo), and POM by GAF-Existence of Nglycans in the four major glycoproteins from PEW were probed by digestion with GAF. HPLC-separated POT, POA(hi), POA(lo), and POM were heat-denatured in the presence of 0.5% SDS and incubated with GAF overnight in the presence of 1% Nonidet P-40. Decrease in molecular size after GAF digestion (analyzed by SDS-PAGE) was apparent in all four glycoproteins (data not shown). The shifts in molecular sizes were smaller for POT than for POA(hi), POA(lo), and POM. The molecular sizes of the deglycosylated POA(hi) and POA(lo) were still different from each other. HPLC-purified POM showed at least three down-shifted bands on SDS-PAGE after digestion with GAF in the presence of 1% Nonidet P-40, which is suggestive of incomplete deglycosylation by GAF.
In another experiment, whole PEW was treated with GAF in the presence of 1% Nonidet P-40, separated by SDS-PAGE, transferred to PVDF membrane, and stained with CBB or GS-I lectin. POT, POA(hi), and POA(lo) treated with GAF showed shifts in molecular sizes, and the deglycosylated POT, POA(hi), and POA(lo) could no longer be stained with GS-I lectin (data not shown). This indicates oligosaccharides containing ␣-Gal on POT and POAs were completely released by GAF under the conditions used. The POM in the whole PEW treated with GAF still showed substantial staining with GS-I even after GAF digestion, possibly due to incomplete deglycosylation.
To achieve complete deglycosylation of POM, HPLC-purified POM was digested with GAF in the presence of 2% CHAPS instead of Nonidet P-40, and the GAF treatment was repeated three times. POM exhaustively digested with GAF in this way revealed only one band (31 kDa) by CBB, which was not stained by GS-I (data not shown). Thus, it was concluded that the 31-kDa band has lost all oligosaccharides containing ␣-galactoside by GAF digestion. The fact that two broad bands of native POM became a single band by exhaustive GAF digestion indicates that POM has a varied number of glycosylation sites, such as is found in the chicken counterpart (51,52). Table III shows the data of MALDI-TOF mass spectrometry for HPLC-separated POM, POT, and POA(hi) before and after GAF treatment. De-N-glycosylated POA(lo) could not be analyzed successfully, because of its lower solubility in water in the absence of detergents. Molecular masses of POT and POA(hi) before and after GAF-treatment (Table III) were in agreement with the molecular sizes indicated by the SDS-PAGE, whereas molecular masses of POM, from mass spectrometric analysis, both before and after GAF treatment, were much lower than those from the SDS-PAGE method. Intact POM showed two broad peaks by mass spectrometric analysis, which might represent two distinct states of glycosylation. Similarly, the three different molecular masses of the GAF-treated POM shown by MALDI-TOF mass spectrometry (Table III)  3.51 (%), and POA (hi); 16.5 (%). From the carbohydrate content of each glycoprotein in PEW, N-linked oligosaccharides in POM, POT, and POA(hi) were deduced to be 23-28, 7, and 70 mg, respectively, in 1 g of whole PEW proteins.
Determination of the Number of N-Glycosylation Sites on PEW Glycoproteins-To determine the number of N-linked glycosylation sites on POA(hi), the glycoprotein was partially de-N-glycosylated by using 0.4 units of GAF with different incubation times, and completely de-N-glycosylated by using 1 unit of GAF with overnight incubation as described under "Experimental Procedures." POA(hi) was sequentially deglycosylated with three intermediates bands on SDS-PAGE between native and completely deglycosylated proteins (data not shown). This result indicates that deglycosylation on POA(hi) progressed in four sequential steps; therefore, it was deduced that four N-    c Full inhibition was not achieved at the maximal inhibitor concentrations used in these assay. linked glycosylation sites are present in POA(hi). By the same method, three to four sites in POM, and only one site in POT were deduced (data not shown). On their chicken counterparts, N-glycosylation sites are reported to be four to five sites for chicken ovomucoid (51,52), one site each for chicken ovotransferrin (53), and chicken ovalbumin (54,55). No O-linked oligosaccharides have been detected on these chicken egg white glycoproteins. The varied number of glycosylation sites on chicken ovomucoid is attributable to partial N-glycosylation on Asn 175 (51,52). Similarly varied numbers of glycosylation sites are also found in POM (Table III). Isolation of Major Oligosaccharides from POA(hi) by HPLC-Since POA(hi) is the major glycoprotein in PEW and has a higher carbohydrate content than POT, we investigated the structure of major oligosaccharides from POA(hi). PA-derivatized oligosaccharides from POA(hi) were separated with anion exchange HPLC on a DEAE column as the first step (39). As shown in Fig. 3A, neutral, monosialyl, disialyl, and trisialyl oligosaccharides are separated based on their charges, and the molar ratio of their fractions was 11.6%, 69.6%, 17.3%, and 1.47%, respectively. The monosialylated fraction, being the major fraction, was further analyzed and separated with reverse phase HPLC on an ODS column. As shown in Fig. 3B, the monosialylated PA-oligosaccharides were separated into more than 10 peaks, suggesting that the oligosaccharides from POA(hi) are highly heterogeneous. The neutral and disialylated oligosaccharides from the DEAE column were also heterogeneous, according to their elution profiles on the ODS column (data not shown). The five monosialylated major fractions separated on the ODS column were designated as ms-5, ms-7, ms-8, ms-9, and ms-12, and further analyzed with normal phase HPLC on an Amido-80 column (Fig. 3C). The elution profiles showed that each fraction had a single major peak designated as ms-5-4, ms-7-1, ms-8 -2, ms-9 -2, and ms-12-2, respectively. Contents of these five major fractions on the Amido-80 column were 19.1%, 4.68%, 7.41%, 1.75%, and 7.01% of the total PA-oligosaccharides from POA(hi), respectively.
Structural Analysis of the PA-tagged Major Oligosaccharides from POA(hi)-The structures of each of the major peaks were deduced based on mass spectrometry analysis of the permethyl derivatives. Fraction ms-5-4 afforded a single [M ϩ H] ϩ molecular ion at m/z 4015 (Table IV) and two major non-reducing terminal oxonium type fragment ions at m/z 668 and 825, corresponding to Hex 2 HexNAc ϩ and NeuAc 1 Hex 1 HexNAc ϩ , respectively (low mass range; data not shown). This is in agreement with the assignment of a tetra-antennary complex type structure for ms-5-4, with three antenna being Hex-Hex-Hex-NAc and a fourth one as NeuAc-Hex-HexNAc (see Table IV). The other four fractions of lower abundance, ms-7, ms-8, ms-9, and ms-12, were analyzed directly after the ODS column, and not after further subfractionation on the Amido-80 column. Each yielded a major molecular ion signal and one or more minor signals (Table IV). To ascertain the origin of the fragment ions observed, the major molecular ion signals were subjected to CID-MS/MS (Fig. 4, A-C). As with ms-5-4, the [M ϩ H] ϩ molecular ion of ms-7 (m/z 4669, Fig. 4A) yielded the key daughter ions at m/z 825 and 668, corresponding to NeuAc-Hex-HexNAc ϩ and Hex-Hex-HexNAc ϩ , respectively. In support of these two non-reducing terminal structures are the ions    Table IV.
Notably, methylation analysis by GC-EI-MS demonstrated that all fractions contained terminal Gal, 4-linked Gal, and 6-linked Gal. The former two were consistent with the deduced terminal Gal␣1-4Gal epitope, whereas the last suggested that the single NeuAc residue was ␣2,6-linked to Gal. The major difference among the various fractions analyzed was found to be the Man residues detected. The presence of 2,4-and 2,6linked Man in ms-5, ms-8, and ms-9 is consistent with a tetraantennary structure whereas the presence of a 2,4,6-linked Man instead of a 2,6-Man in ms-7 supported a pentaantennary structure, and the presence of a 2-linked Man instead of a 2,6-linked Man in ms-12 is consistent with a triantennary structure. The latter triantennary structure, with a single antennae on the 6-arm, resulted in its later elution time than the others on an ODS column (56). In addition, the presence of a bisecting GlcNAc in fractions ms-8, ms-9, and ms-12 was indi-cated by the presence of 3,4,6-linked Man instead of 3,6-linked Man, which also contributed to their longer retention time on an ODS column than ms-5 and ms-7. In short, the structural analysis data are fully consistent with the HPLC elution time of each sample, which also suggested that the single monosialylated antennae is probably located on the same position for each N-glycan.
To further define the location of the monosialylated antennae, ms-5-4 and ms-7 were sequentially digested with neuraminidase from A. ureafaciens, ␤-galactosidase from bovine testes, and ␤-N-acetylhexosaminidase from jack bean. ESI-MS analysis of the permethyl derivatives of the digestion products (Fig. 5) showed that one NeuAc␣2-6Gal␤1-4GlcNAc␤1-antennae was completely removed to yield a predominant product with a molecular composition of one NeuAc, one Hex, and one HexNAc residues less than the undigested samples. The results also further confirmed that the other antenna were capped by an ␣-Gal residue, which rendered them resistant to ␤-galactosidase digestion. Upon removal of the sialylated antennae, linkage analysis of the digestion products showed that 6-linked Gal and 2,4-linked Man in both ms-5-4 and ms-7 have disappeared (data not shown). Since 2-linked Man was not detected, whereas 4-linked Man co-eluted with 4-linked Gal, it was concluded that the single sialylated antennae was located on the GlcNAc 2-linked to the Man on the 3-arm. This antennae (i.e. -GlcNAc1-2Man1-3Man-) is in fact commonly present in all tri-, tetra-, and pentaantennary structures. Due to the pres-

FIG. 4. CID-MS/MS of selected major molecular parent ions afforded by FAB-MS analysis of the permethyl derivatives of the PA-oligosaccharides from POA(hi) ms-7 (A), ms-8 (B), and ms-12 (C).
Major daughter ions correspond to A-type non-reducing end oxonium ions, accompanied by signals at 32 mass units lower resulting from further elimination of a CH 3 OH moiety as described by Dell et al. (40). The insets show the parent ions with an A-type fragment ion resulting from cleavage between the two GlcNAc of the chitobiose core. No fragment ion was detected within the mass range of m/z 1000 -3000.

FIG. 5. ESI-MS analysis of the exo-glycosidases treated permethyl derivatives of ms-5-4 (A) and ms-7 (B) from POA(hi).
The insets show the disodiated doubly charged molecular ions acquired with a separate scan. At high cone voltage used to acquire the MS data, several fragment ions due to sequential loss of non-reducing terminal residues from the singly charged sodiated molecular ion could also be detected (signals between m/z 2000 and 3000 in A). By virtue of mass difference, these could be differentiated from the genuine presence of other molecular ion signals corresponding to components with more antenna having trimmed off by the ␤-galactosidase and/or ␤-N-acetylhexosaminidase digestions. Thus, signals at m/z 3225 and 2571 in B correspond, respectively, to components with two and three antenna completely removed. These could originate from minor components with incomplete ␣-galactosylated Gal␤1-4GlcNAc␤-or GlcNAc1␤-on one or more antenna. ence of other minor components and the incompleteness of digestion resulting from the steric hindrance imposed by the bisecting GlcNAc, the location of the sialylated antennae could not be conclusively demonstrated for other fractions (ms-8, ms-9, and ms-10). DISCUSSION Although POM as well as the whole PEW manifested P 1 antigenic activity and inhibit adherence of uropathogenic E. coli (24 -26), it was not known whether the antigenicity of PEW is derived solely from POM or from other PEW glycoproteins also. We demonstrated that, in addition to POM, other major glycoproteins (POT, POA(hi), and POA(lo)) also contain terminal Gal␣1-4Gal, but none of them contain Gal␣1-3Gal as expressed most commonly in mammals. These data suggest that the synthetic mechanisms to produce Gal␣1-4Gal in pigeon is not limited to the N-glycan of POM, but is also operative for the other PEW glycoproteins. Since synthesis and glycosyl modification of egg white glycoproteins are mostly carried out in the oviduct, the Gal␣1-4Gal linkages may also be formed in this organ.
The presence of Gal␣1-4Gal on the PEW glycoproteins can explain the inhibitory activities of the glycoproteins for Pfimbriated E. coli. That POAs and POT, like POM, specifically inhibit binding of P fimbriae of uropathogenic E. coli suggest that they also can bind to some other pathogenic bacterial adhesins and enterotoxins, which recognize Gal␣1-4Gal. Different PEW glycoproteins showed somewhat different inhibitory potency against three different agglutinators (Table I). This may reflect the different amounts and the structures of the carbohydrates in the individual glycoproteins, and the different carbohydrate-binding specificities among the agglutinators.
We have demonstrated that the ␣-galactoside-containing oligosaccharides on POT, POA(hi), POA(lo), and POM exist only as N-glycans. The molecular masses of de-N-glycosylated polypeptide chains of POM, POT, and POA(hi) measured by MALDI-TOF-MS (Table III) were in the range of the masses of the peptide portions of chicken counterparts (as calculated from the peptide sequences), COM (20,098), COT (75,827), and COA (42,750), respectively, although the mass of POA(hi) is about 2 kDa higher than COA. The MS data also indicated that the differences of molecular sizes between the intact glycoproteins from PEW and CEW were mostly due to the glycosylation. In other words, PEW proteins contain more carbohydrates than their chicken counterparts.
Intriguingly, two POAs of distinctly different sizes were found in PEW, whereas only one species of ovalbumin (excluding heterogeneities due to phosphorylation and carbohydrate structures) is expressed in CEW. The difference in the molecular sizes between POA(hi) and POA(lo), about 5 kDa from MALDI-TOF-MS, is not attributable to glycosylation, because the molecular mass difference persists after deglycosylation (data not shown). During the isolation and sequence analyses of glycopeptides from POA(hi) and POA(lo), we have found they are homologous but not identical in their peptide sequences suggesting that these POAs are expressed from different but related genes (data not shown).
The heterogeneity of N-glycans from POA(hi) was evident by HPLC and subsequent MS analysis for their structures. The data indicated that the glycoform heterogeneity of POA(hi) resulted mainly from the presence of 1) a mixture of tri-, tetra-, and pentaantennary complex type structures; 2) bisecting GlcNAc on the ␤-Man among a subpopulation; and 3) different degrees of sialylation. The extensive heterogeneity makes complete characterization of all glycoforms a laborious and daunting task (57). The presence of tri-and tetra-antennary structures as the major N-glycans suggests that POA(hi) has complex population of N-glycans from those of turtle dove ovomucoid, whose proposed structures were pentaantennary (28, 29). In the following paper (57), we have studied 25 structures of PA-derivatized N-glycans of POT, POAs, and POM, and found that the exact structure proposed for the turtle dove ovomucoid is not present in the major PEW glycoproteins. MS analysis for the structures of major N-glycans from POA(hi) also suggested the presence of Gal␣1-4Gal and the absence of Gal␣1-3Gal in all the analyzed oligosaccharides as predicted by the data of immunostaining (Fig. 1B). However, the structural feature of the analyzed N-glycans from POA(hi), i.e. occupation of non-reducing termini by either ␣1,4-galactosides or ␣2,6-sialic acids, are quite similar to N-glycans containing Gal␣1-3Gal produced in mammals, e.g. N-glycan from bovine thyroglobulin (58). Formation of Gal␣1-3Gal on N-glycans expressed in mammalian cells is considered to reduce sialylation on non-reducing Gal␤1-4GlcNAc terminal in vivo (59). Gal␣1-4Gal␤1-4GlcNAc in pigeon can be speculated to have an effect on sialylation similar to that of Gal␣1-3Gal in mammals.
The presence of Gal␣1-4Gal or substances similar to P 1 antigen only in some avian species is phylogenetically interesting. It was reported that P 1 antigenic activities is absent in the blood of chicken, gander, turkey, quail, duck, and pheasant (22,60), and no ␣-galactoside is found in the glycans of ovomucoid from chicken, quail, and duck (30 -34). On the other hand, P 1 antigenic activities have been reported to be present in the blood and/or eggs of pigeon, turtle dove, budgerigar, and cockatiel (23,24,60). Interestingly, salivary gland mucin glycoproteins of Chinese swiftlets (genus Collocalia) contain O-linked glycans with Gal␣1-4Gal␤1-4Gal (61). These reports suggest that Gal␣1-4Gal is absent in the orders Galliformes (chicken, turkey, quail, and pheasant) and Anseriformes (duck and gander), but present in the orders Columbiformes (pigeon and turtle dove), Psittaciformes (budgerigar and cockatiel), and Apodiformes (swiftlet) (62), based on the limited information available. Galliformes and Anseriformes were regarded as sister taxa by morphological analyses (63), nuclear DNA-DNA hybridization analyses (64), and mitochondrial DNA sequence analyses (65), and this sister group was placed as branched group from most of the other groups of modern birds including Columbiformes, Psittaciformes, and Apodiformes. Therefore, it is possible that, in the course of evolution and diversification of the modern birds, some of the avian species acquired (e.g. ancestor of Columbiformes, Psittaciformes, and Apodiformes) or lost (e.g. ancestor of Galliformes and Anseriformes) the ability to express Gal␣1-4Gal, because of some selective pressures. Distribution of Gal␣1-4Gal among birds and its phylogenic relationships will be refined by further investigation into whether birds of other orders also produce glycoproteins with Gal␣1-4Gal.
The real biological significance of Gal␣1-4Gal in pigeon is unknown at this stage, although one possibility is defense against microbial infection. Our recent data indicated that pigeon serum and liver extracts also contain glycoproteins showing the presence of Gal␣1-4Gal, by SDS-PAGE and Western blot with GS-I and anti-P 1 mAb. This suggests that the putative function of Gal␣1-4Gal may not be limited to some specific organs in pigeon, but may be required for the entire body. Study of genetic control and enzymatic behavior of ␣-1,4galactosyltransferase, which is responsible for production of Gal␣1-4Gal in pigeon, would be one of the ways to understand the evolution of glycoproteins.
Furthermore, the demonstration that Gal␣1-4Gal sequence is abundantly present in pigeon egg white glycoproteins led us to utilize these unique features for biomedical applications. We have constructed microbeads bearing PEW glycoproteins, and they agglutinate the strains of E. coli that have P-fimbriae (data not included). The beads can be utilized as diagnostic reagents to determine bacteria or toxins with the specificity. In addition, Sepharose beads bearing these glycoproteins have been successfully used to isolate Shiga-like toxin, which recognized galabiose structure (66). The potential of PEW glycoproteins as biomedical tools will expand to use as a therapeutic agent to prevent or treat urinary tract infections due to E. coli.