Capillary Affinity Electrophoresis for the Screening of Post

Capillary Affinity Electrophoresis for the Screening of Post-Translational Modification of Proteins with Carbohydrates Kazuki Nakajima, Yasuo Oda, Mitsuhiro Kinoshita, and Kazuaki Kakehi* Faculty of Pharmaceutical Sciences, Kinki University, Kowakae 3-4-1, Higashi-osaka 577-8502, Japan Received August 27, 2002

Glycosylation is one of the most important post-translational events for proteins, affecting their functions in health and disease, and plays significant roles in various information traffics for intracellular and intercellular biological events (Hancock, W. S. J. Proteome Res. 2002, 1, 297). We have attempted to obtain the information on the numbers and amounts of carbohydrate chains. Interaction between carbohydrate chains and proteins that recognize them is a target to understand the biological roles of glycosylation. To date, there have been a few strategies for simultaneous analysis of the interactions between complex mixtures of carbohydrates and proteins. Here, we report an approach to categorize carbohydrate chains using a few glycoprotein samples as models for the studies on the analysis of post-translational modification of proteins with carbohydrates. A combination of some specific lectins was used as carbohydrate-binding proteins. The method is based on high-resolution separation of fluorescent-labeled carbohydrates by capillary electrophoresis with laser-induced fluorescent detection in the presence of carbohydrate-binding proteins at different concentrations. The present technique affords (1) simultaneous determination of carbohydrate chains, (2) binding specificity of the constituent carbohydrate chains to specific proteins, and (3) kinetic data such as the association constant of each carbohydrate. We found that the lectins employed in the present study could discriminate subtle difference in linkages and resolved the carbohydrate mixtures. The results will be useful, for example, to understand the biological events expressed with carbohydrate changes on the cell surface. Keywords: capillary affinity electrophoresis • glycoprotein • lectin

Introduction Modification of proteins and lipids with carbohydrates plays an important role in modulation of their structures and functions. In the extracellular environment, carbohydrates exert effects on cellular recognition in infection, cancer, and immune response, but details of the specific mechanisms still remain as unsolved targets. Therefore, analysis of the carbohydrate moiety is of primary importance for understanding the role of carbohydrate chains. We have developed a method for the analysis of oligosaccharides derived from glycoproteins using capillary electrophoresis after labeling with fluorescent compounds.1 This is especially important for the evaluation of recombinant biopharmaceuticals because the products obtained from different cell lines or in different culture conditions sometimes contain a wide variety of carbohydrate chains. The development of new technologies including capillary electrophoresis (CE), two- and three-dimensional chromatographic techniques, and fluorophore-assisted carbohydrate electrophoresis in slab gel format has increased the ease of oligosaccharide analysis.2-5 A combination of CE and laser-induced fluorescence (LIF) detection is a powerful strategy for ultrahigh sensitive determination of * To whom correspondence should be addressed. Tel: +81-6-6721-2332 (ext 3822). Fax: +81-6-6721-2353. E-mail: [email protected]. 10.1021/pr020009v CCC: $25.00

 2003 American Chemical Society

carbohydrate chains in glycoprotein samples. A fluorogenic reagent, 8-aminopyrene-1,3,6-trisulfonate (APTS), was developed for derivatization and analysis of reducing sugars using capillary electrophoresis with LIF detection.6 We applied the technique to the analysis of carbohydrate chains derived from R1-acid glycoprotein (AGP) glycoforms isolated by capillary electrophoresis.7 Carbohydrate chains thus analyzed should be categorized according to their functions (e.g., binding characteristics to a specific protein). The interaction between proteins capable of binding glycoconjugates and carbohydrates is quite stimulating because carbohydrates are the mediators for the transmission of biological information. We analyzed carbohydrate chains of AGP after separation of AGP molecular species based on the constituent carbohydrate chains by affinity chromatography using an immobilized lectin column, and found that each fraction showed characteristic abundance of carbohydrate chains.8,9 Many methods have also been developed for the analysis of the interaction between carbohydrates and lectins. Most of them such as surface plasmon resonance,10 fluorescence polarization,11 and time-resolved fluorometry12 are based on the interaction between one protein and one carbohydrate. However, glycoproteins are usually composed of a complex mixture of carbohydrate chains, and a method is often required for simultaneous evaluation of the contributions of each Journal of Proteome Research 2003, 2, 81-88

81

Published on Web 12/04/2002

research articles

Figure 1. Fluorescent labeling of carbohydrates with 8-aminopyrene-1,3,6-trisulfonate (APTS).

carbohydrate on the binding for accurate understanding of biological roles of carbohydrate chains. Frontal affinity chromatography originally developed by Kasai13 using an immobilized lectin column has been online coupled to mass spectrometry (FAC/MS) and applied to the analysis of carbohydrate-protein interaction.14,15 FAC/MS allows the estimation of binding constants of individual carbohydrate in mixtures. In FAC/MS, an affinity column was prepared by immobilizing a lectin. A sample consisting of a mixture of carbohydrates was continuously infused through the column. The weak binding component was eluted earlier, and the high affinity component was eluted later. By monitoring the eluate by electrospray mass spectrometry, the dissociation constant (KD) of the constituent carbohydrates in the mixture could be simultaneously determined. However, frontal affinity chromatography is a method for the determination of interaction between the analyte and the “immobilized” receptor molecules. In some cases, determination of the interactions in “solution” state becomes necessary.16 Capillary affinity electrophoresis (CAE) allows the measurement of molecular interaction between carbohydrates and proteins in solution state when the electrophoretic mobility of the conjugate is different from those of carbohydrates and lectins. Shimura and Kasai reviewed capillary affinity electrophoresis as a sensitive tool for the study of molecular interactions,17 and they showed kinetic studies on the binding reaction using single oligosaccharide and a lectin. Taga et al. reported the simultaneous determination of the association constants of a mixture of simple oligosaccharides to a lectin by capillary electrophoresis.18 They used some typical disaccharides and isomaltooligosaccharides (i.e., R1,6-glucose oligomers) as models after labeling with 8-amino-naphthalene-1,3,6-trisulfonate or 1-phenyl-3-methyl-5-pyrazolone. Hong et al. employed a lectin, concanavalin A (Con A), as the model lectin and examined the interactions with oligosaccharides derived from ribonuclease B and fetuin.19 In the present paper, we propose a method for classification of a complex mixture of carbohydrate chains. A mixture of carbohydrates was previously labeled with a fluorescent tag and analyzed by capillary electrophoresis in the presence of a set of selected lectins. We could classify the carbohydrate chains on the basis of the migration behaviors of each carbohydrate chain in the presence of the lectins. Shimura and Kasai recommended the use of carbohydrates labeled with a fluorescent tag having electric charge.20 The charge becomes a driving force to migrate the carbohydrates with or without the binding protein. In the present study, the reducing ends of the carbohydrates were modified with APTS by reductive amination, as shown in Figure 1. Mono- and oligosaccharides derived from glycoprotein samples labeled with APTS showed excellent resolutions using a chemically modified capillary within 10 min.21 Negative 82

Journal of Proteome Research • Vol. 2, No. 1, 2003

Nakajima et al.

charges due to sulfonic acid residues of the aminopyrene are favorable for the studies on the binding reactions.17 We selected four lectins: Con A for mannose (Man), wheat germ agglutinin (WGA) for N-acetylglucosamine (GlcNAc), Tulipa gesneriana agglutinin (TGA) for complex-type carbohydrates, and Phaseolus vulgaris agglutinin (PHA-E4) for bisecting GlcNAc residue. As the initial study of screening of carbohydrate chains in glycoproteins, we proposed a method to categorize asialocarbohydrate chains and showed that a combination of these lectins at different concentrations makes it possible to discriminate subtle variations in carbohydrate structures and classify carbohydrate chains with improved efficiency.

Experimental Section Materials. Samples of R1-acid glycoprotein (AGP) and fetuin were obtained from Sigma-Aldrich Japan (Minato-ku, Tokyo, Japan). Chicken ovomucoid was purified from hen egg white according to the method reported previously.22 Concanavalin A (Con A), wheat germ agglutinin (WGA), and P. vulgaris agglutinin (PHA-E4) were obtained from Seikagaku Kogyo (Nihon-bashi, Tokyo, Japan). A mixture of chitooligosaccharides (oligomers of N-acetylglucosamine, GlcNAc oligomers) was also from Seikagaku Kogyo. T. gesneriana agglutinin (TGA) was isolated and purified from the bulbs of tulip according to the method reported previously.23 Peptide-N4-(acetyl-β-D-glucosaminyl)asparagine amidase (N-glycosidase F) was from Roche Molecular Biochemicals (Minato-ku, Tokyo, Japan). Highly purified 8-aminopyrene-1,3,6-trisulfonate (APTS) was obtained from Beckman-Coulter (Fullerton, CA). All other samples and reagents were of the highest grade commercially available or of HPLC grade. All aqueous solutions were prepared using water purified with a Milli-Q purification system (Millipore, Bedford, MA). Preparation of a Mixture of Fluorescent-Labeled Carbohydrate Chains from Glycoprotein Samples. We prepared the fluorescent-labeled carbohydrate chains (glycan library) from fetuin, AGP, and ovomucoid as shown below. Release of carbohydrate chains from glycoprotein was performed according to the method reported previously.9,24 Briefly, a sample of glycoprotein (1 mg) was dissolved in 20 mM phosphate buffer (pH 7.0, 40 µL). N-Glycosidase F (5 mU, 5 µL) was added, and the solution was incubated for 24 h at 37 °C. The solution was kept for 5 min on a boiling water bath and centrifuged for 10000g for 10 min. The supernatant containing the oligosaccharides was evaporated to dryness by a centrifugal vacuum evaporator (SpeedVac, Savant, Farmingdale, NY). The residue was dissolved in 2 M aqueous acetic acid (50 µL), and the mixture was kept at 80 °C for 3 h to remove sialic acids from the oligosaccharides.25 The residue was dissolved in 15% aqueous acetic acid (5 µL) containing APTS at the concentration of 100 mM. A freshly prepared solution of 1 M NaBH3CN in tetrahydrofuran (5 µL) was added to the mixture. The mixture was overlaid with mineral oil (100 µL, nD 1.4670, d 0.838; available from Aldrich) to prevent evaporation of the reaction solvent. This procedure to prevent evaporation of the solvent was important to keep the yield constant.7 The mixture was kept for 90 min at 55 °C. Water (200 µL) was added to the mixture, and the fluorescent yellowish aqueous phase was collected. The aqueous layer was applied on a column of Sephadex G-25 (1 cm, 50 cm length) equilibrated with water. The earlier eluted fluorescent fractions were collected and evaporated to dryness. The residue was dissolved in water (100

Post-Translational Modification of Proteins with Carbohydrates

research articles Table 1. List of the Asialocarbohydrate Chains Derived from Fetuina

Figure 2. Principle for categorization of carbohydrate chains.

µL), and a portion (20 µL) was used for capillary affinity electrophoresis. Removal of the reagent (APTS) from the reaction mixture was necessary to keep the carbohydrate derivatives at -20 °C without decomposition. Capillary Affinity Electrophoresis. Capillary affinity electrophoresis was performed using a P/ACE MDQ glycoprotein system (Beckman Coulter) equipped with an argon-laser induced fluorescence detection system. Detection was performed by installing a 520-nm filter for emission with a 488nm argon-laser for excitation. Separation was performed using an eCAP N-CHO coated capillary (10 cm effective length, 30 cm total length), 50 µm i.d., Beckman Coulter). A capillary coated with dimethylpolysiloxane (DB-1) of the same size (GL Science Co., Nishi-Shinjuku, Shinjuku, Tokyo, Japan) was also used. The detection window of the capillary was made at 10 cm from the outlet of the capillary. Separation was performed at 25 °C throughout the work. The detailed analytical conditions are given below and in the figure legends. Injections were performed automatically in the pressure method (0.5 psi, 5 s). Data were collected and analyzed with a standard 32 Karat software (Version 4.0, Beckman Coulter) on Windows 2000. In the present study, 100 mM Tris-acetate buffer (pH 7.4) was used as the electrolyte throughout the work. Prior to capillary affinity electrophoresis, fluorescent-labeled carbohydrates with APTS were analyzed by capillary electrophoresis as described above. Then, the same electrolyte containing a lectin at the specified concentration (see Figure 8) was filled in the capillary. Before each run, the capillary was washed with the running buffer for 1 min, then with the same buffer including a lectin for 1 min. Because the apparatus can handle the sample solution in a 96well plate, we can automatically perform a series of binding reactions.

Results Theory. Figure 2 illustrates the principle of the present technique for classification of carbohydrate chains. A mixture of fluorescent-labeled carbohydrate chains was analyzed by capillary electrophoresis in the electrolyte that does not contain a lectin and resolved as shown in Figure 2a. In the following step, the sample was analyzed in the presence of a lectin whose specificity is well established. When the lectin recognizes a carbohydrate (peak A in Figure 2), the peak is observed later due to the equilibrium formation between conjugate form and free form. On the contrary, the carbohy-

a The abbreviations used for the structures are as follows: Gal, galactose; GlcNAc, N-acetylglucosamine; Man, mannose.

drate (peak C) does not show affinity to the lectin, and is observed at the same migration time as in the absence of the lectin (Figure 2a). The carbohydrate (peak B) does not show higher affinity to the lectin than peak A and is observed slightly later. Thus, the migration order of the carbohydrate chains changes as shown in Figure 2B. The similar procedure is repeated using a different lectin whose specificity is also well studied. We will be able to categorize all carbohydrate chains by repeating the procedure with selecting an appropriate set of lectins. Selection of Glycoproteins. In the present study, we employed glycan libraries derived from fetuin, AGP, and chicken ovomucoid as model. Fetuin contains di- and triantennary oligosaccharides. A portion of the Galβ1-4GlcNAc branches is changed to Galβ1-3GlcNAc (Table 1).26 Carbohydrate chains of AGP are di-, tri-, and tetraantennary oligosaccharides, and some of the tri- and tetra-antennary oligosaccharides are substituted with fucose (Table 2).27 Chicken egg ovomucoid contains a quite complex mixture of oligosaccharides including hybrid-type oligosaccharides, and some of the oligosaccharides are substituted with bisecting GlcNAc residue.28 Typical oligosaccharides found in ovomucoid are shown in Table 3. A small oligosaccharide (OI) composed of core N-glycan structure is also present in chicken ovomucoid. Selection of Lectins. We examined various lectins having the following carbohydrate-binding specificity: (1) mannose-binding proteins for high-mannose type oligosaccharides; (2) lectins recognizing N-acetylglucosamine (GlcNAc) or its oligomers; (3) lectins recognizing galactose (Gal) or Galβ1-4/3GlcNAc; (4) lectins recognizing fucose. It is important to choose an appropriate set of lectins for observation of clear interaction with carbohydrates. For such a purpose, the lectins should be stable and show highly specific binding toward specific carbohydrate chains. In the present study, we chose Con A, WGA, TGA, and PHA-E4 according to their specificities toward the carbohydrate chains. It should be noticed that these lectins are relatively stable during analysis and storage prior to the analysis. Furthermore, these lectins show affinities to carbohydrates used in the present study at 104-107 M-1 order. Classification of Carbohydrate Chains Derived from Fetuin. Fetuin contains one di- and two triantennary carbohydrate chains as shown in Table 2.26 One of the triantennary carbohydrate chains (AII) has three Galβ1-4GlcNAc residues. AnJournal of Proteome Research • Vol. 2, No. 1, 2003 83

research articles Table 2. List of the Asialocarbohydrate Chains Derived from AGPa

a The abbreviations used for the structures are as follows: Gal, galactose; GlcNAc, N-acetylglucosamine; Man, mannose; Fuc, fucose.

other triantennary carbohydrate chain (FII) has one Galβ13GlcNAc branch. APTS derivatization of asialocarbohydrate chains allowed separation of these carbohydrate chains. The resolution of triantennary carbohydrate chains (AII and FII) was incomplete in the absence of a lectin as shown in Figure 3, as reported previously.1 Wheat Germ Agglutinin (WGA). The major triantennary carbohydrate chain (AII) was obviously observed at later migration times (around 8.9 min) at 12.0 µM of WGA than in the absence of WGA. Retardation of the migration time of another tri-antennary carbohydrate (FII), that has a Galβ13GlcNAc branch, was not obvious. FII was observed at 7.5 min at 12.0 µM of WGA, and AII and FII were completely separated. T. gesneriana Agglutinin (TGA). The bulbs of tulip contain two carbohydrate binding proteins.23,29 One of the lectins binds yeast cells (i.e., mannose-specific).29 The other lectin (TGA, this lectin) binds mouse erythrocytes specifically. The binding is inhibited by porcine thyroglobulin, although the binding specificity to carbohydrate chains has not yet been fully determined.23,30 Interestingly, TGA showed different affinities to AII and FII. The carbohydrate chain including a Galβ13GlcNAc branch showed obvious retardation of migration times, and the migration order of AII and FII was reversed at 2.0 µM of TGA. Difference in affinities to AII and FII was more clearly observed at 12.0 µM of TGA, and both triantennary carbohydrate chains were completely resolved. Classification of Carbohydrate Chains Derived from r1Acid Glycoprotein (AGP). AGP contains di-, tri-, and tetraantennary carbohydrate chains as shown in Table 2. Some of the 84


Nakajima et al. Table 3. List of the Carbohydrate Chains Derived from Chicken Ovomucoida

a The abbreviations used for the structures are as follows: Gal, galactose; GlcNAc, N-acetylglucosamine; Man, mannose; Fuc, fucose.

tri- and tetraantennary carbohydrate chains contain a fucose residue attached to the nonreducing Galβ1-4GlcNAc residue. We analyzed a mixture of asialocarbohydrate chains obtained from AGP and show the results in Figure 4. Identification of each peak was reported previously.9 The diantennary carbohydrate chain (AI, see Table 2) was observed at the earliest migration times (5.2 min). The tri- (AII) and tetraantennary (AIV) carbohydrate chains were observed at 6.2 and 7.2 min, respectively. The peaks observed at 6.5 and 7.5 min were those of tri- (AIII) and tetraantennary (AV) carbohydrate chains that contain a fucose residue, respectively. WGA. Addition of WGA in the electrolyte showed interesting migrations. At 6.0 µM of WGA concentration, AII and AIII fused into a single peak at 8.2 min, and AIV and AV were also observed as a single peak at 9.3 min. Finally, the migration orders of a set of AII and AIII and another set of AIV and AV were reversed at 12.0 µM of WGA. These data indicated that the fucose residue attached to the branches of tri- and tetraantennary carbohydrate chains decreased the binding with WGA. Concanavalin A. Addition of Con A to the electrolyte caused changes of the peak intensity of the diantennary carbohydrate chain (AI). Peak intensity of the di-antennary carbohydrate chain, which was observed at 5.2 min in the absence of the lectin, was gradually decreased at higher concentrations than 0.20 µM of Con A. On the contrary, Con A did not show obvious affinity to tri- and tetra-antennary carbohydrate chains at these concentrations. These results were well comparable to the data reported previously.9


research articles

Figure 3. Capillary affinity electrophoresis of carbohydrate chains derived from fetuin. Running buffer: 100 mM Tris-acetate buffer (pH 7.4) containing 0.5% poly(ethylene glycol) (PEG70000) and lectins. Capillary, eCAP N-CHO coated capillary (30 cm length (effective length, 10 cm); 50 µm i.d.). Applied potential: 10 kV. Injection: pressure method (0.5 psi for 5 s). Fluorescent detection at 520 nm excited with an argon-laser with a 488 nm filter. The accurate structures of the oligosaccharides are shown in Table 1.

Figure 4. Capillary affinity electrophoresis of carbohydrate chains derived from R1-acid glycoprotein. The analytical conditions were the same as those described in Figure 3. The accurate structures of the oligosaccharides are shown in Table 2.

TGA. Addition of TGA showed interesting effects on the migration of oligosaccharides from AGP. Both triantennary carbohydrate chains (AII and AIII) were quite sensitive to this lectin. At 2.0 µM concentration of TGA, the group of AII and AIII moved to 7.0 min. Finally, both AII and AIII were fused to a broad and single peak and observed the latest (around 9.3 min) at 12.0 µM. On the contrary, affinities of TGA toward diantennary (AI) and tetraantennary (AIV and AV) carbohydrate chains were not obvious, and similar patterns of migration were observed. Classification of Carbohydrate Chains Derived from Chicken Ovomucoid. Ovomucoid from chicken egg white has five glycosylation sites attaching carbohydrates that constitute 2025% of the glycoprotein.31 In chicken ovomucoid, more than 20 carbohydrate chains were reported. Some carbohydrate chains contain the bisecting GlcNAc residue. Some typical oligosaccharides found in ovomucoid are shown in Table 3. Separation of the oligosaccharides is shown in Figure 5. Although we could not confirm all the carbohydrate chains due to the complexity, classification of these carbohydrate chains is a good model for evaluation of the present technique. In the absence of the lectin, several broad peaks were observed from 4 to 8 min. The small peak observed the earliest was due

to the core pentasaccharide (OI) that does not contain the bisecting GlcNAc residue.32 Phaseolus vulgaris Agglutinin (PHA-E4). We found that addition of P. vulgaris agglutinin (PHA-E4) showed obvious effect on migrations of oligosaccharides. At 0.8 µM of PHA-E4, some groups of peaks were clearly observed at later migration times. Finally, oligosaccharides that do not contain the bisecting GlcNAc residue remained as observed in the absence of the lectin. On the contrary, oligosaccharides that contain the bisecting GlcNAc were extensively retarded and observed at later migration times (see broad peaks around 7-8 min at 6.0 µM of PHA-E4). WGA. We also examined the effect of WGA on the migration of the carbohydrate chains. Although the migration times of the peak of core pentasaccharide (OI, the earliest peak) were not changed at any concentrations of WGA, other peaks were gradually observed later with increase of WGA concentrations. Con A. The results observed in the electrolyte containing Con A supported the observations for PHA-E4 and WGA as shown in Figure 5. Even at low concentrations of Con A, disappeared the peak corresponding to core pentasaccharide (OI) as observed in the analysis of diantennary carbohydrate chain derived from AGP (see Figure 4). These results indicated that Journal of Proteome Research • Vol. 2, No. 1, 2003 85

research articles

Nakajima et al.

Figure 5. Capillary affinity electrophoresis of carbohydrate chains derived from ovomucoid. The analytical conditions were the same as those described in Figure 3. The accurate structures of some typical oligosaccharides are shown in Table 3.

Figure 6. Capillary affinity electrophoresis of a mixture of N-acetylglucosamine oligomers in the presence of WGA in the electrolyte. R is due to the reagent (APTS). The numbers with the peaks indicate tri (3)-, tetra (4)-, penta (5)- and hexa (6)-saccahride, respectively. An aqueous solution (20 µL) of the mixture (2 µg) of fluorescent-labeled N-acetylglucosamine oligomers was used as the sample solution. WGA was dissolved at the concentrations of (a) 0, (b) 0.20, (c) 0.80, and (d) 3.0 µM. Other analytical conditions were the same as in Figure 3.

the oligosaccharides containing the exposed mannose residues were observed at earlier migration times, and disappeared in the presence of Con A. Due to the complexity of the carbohydrate chains in ovomucoid, we could not assign the carbohydrate chains. However, we observed that the carbohydrate chains of chicken ovomucoid migrated in different manner based on their structural characteristics in the presence of the lectins. Interaction between WGA and GlcNAc Oligomers. It is important to determine the stoichiometry of the binding reaction as well as categorization of carbohydrate chains for understanding the biological significance of carbohydrate chains attached to glycoproteins. In the present study, we show the interactions between WGA and a mixture of GlcNAc oligomers after derivatization with APTS as model, because kinetics and mechanism of the binding of this lectin have been well studied.33,34 As shown in Figure 6, the mixture of oligomers of GlcNAc showed interesting change in migrations in the presence of WGA at various concentrations. Trisaccharide (3 in Figure 6) showed weak affinity to WGA 86


even at 3.0 µM of the lectin concentration. Tetrasaccharide (4) began to move at the slower velocity in the presence of 0.80 µM of WGA than that observed in the absence of WGA. Pentasaccharide (5) began to move at the slower velocity in the presence of 0.20 µM of WGA concentration. At higher concentrations of WGA, its migration velocities became obviously smaller than those of tetrasaccharide. Larger oligosaccharides than pentasaccharides showed similar behaviors, and the velocities became more clearly decreased even at very low concentrations of WGA. These data indicated that the larger oligomers showed higher affinities to WGA. Similar results observed above were reported using nuclear magnetic resonance, analytical ultracentrifugation, and isothermal titration microcalorimetry.34 Taga et al. reported a method for the calculation of association constants (Ka) of oligosaccharides to a lectin18 using a few disaccharides as model and derived the following equation t2 1 1 1 1 1 + ) t - t1 t1 t2 - t1 Ka [P] t2 - t1

(1)


research articles

Figure 8. Strategy for classification of carbohydrate chains from biological samples.

t1)-1

[P]-1

Figure 7. Plots of (t vs for tri-, tetra-, and pentasaccharides of GlcNAc: closed circle, pentasaccharide; closed triangle, tetrasaccharide; closed square, trisaccharide.

where t is the migration time of the ligand (APTS derivatives of GlcNAc oligomers in this case) in the presence of a protein (WGA in this case), t1 is the migration time of the ligand in the absence of the protein, and t2 is approximated as the migration time of the ligand at the concentration of the protein at the beginning of a plateau in the relationship between migration time35 and concentration of the protein.35 Thus, eq 1 is considered as eq 2 A 1 1 ) +B t - t1 Ka [P]

(2)

where A and B are the constants. As indicated by Taga et al., we can easily obtain the association constants (Ka’s) by plotting the relationship between (t - t1)-1 and [P]-1.18 It should be noticed that we do not have to determine the concentrations of the ligand (i.e., APTS-oligosaccharides) as indicated in eqs 1 or 2. This is quite important in the studies on the binding stoichiometry of a complex mixture of carbohydrate chains derived from biological samples, because it is difficult to determine the accurate concentration of each carbohydrate in a complex mixture of carbohydrates derived from biological samples. Using the data in Figure 6, we plotted the relationship between (t - t1)-1 and [P]-1. The results are shown in Figure 7. Tri-, tetra-, and pentasaccharides showed good linear relationships with the concentrations of WGA, and their binding constants were 0.56 × 106, 1.05 × 106, and 2.54 × 106 M-1, respectively. The results obtained by the present technique were well comparable to those reported by Dam and Brewer.33 Asansio et al. reported that the chitin-binding motif of WGA bound GlcNAc oligomers in a multivalent fashion.34 But we calculated the binding as 1:1 stoichiometry in the present work, and further mathematical consideration is required for the binding in multivalent fashion.

Discussion We propose a strategy for classification of carbohydrate chains derived from glycoproteins as shown in Figure 8. In the present paper, we selected four lectins, Con A, WGA, PHA-E4, and TGA, for classifying the carbohydrates. Con A recognizes diantennary carbohydrate chains. High-mannose and hybrid-type oligosaccharides are also recognized by this lectin. WGA shows obvious effect on the migration of di-, tri-, and tetraantennary oligosaccharides. However, tri- and tetraantennary oligosaccharides that contain a fucose residue

changes their migration order with those of the respective oligosaccharides that do not contain the fucose residue in the presence of WGA. TGA is quite useful to distinguish triantennary carbohydrate chains. PHA-E4 is applied to the recognition of the carbohydrate chains having the bisecting GlcNAc residue. In the present study, we found that lectins showed binding with carbohydrates in two different manners. Binding between Con A and the core pentasaccharide in ovomucoid (OI, see Figure 5) caused disappearance of the peak. In other cases, binding between lectins and carbohydrates resulted in retardation of the migration times. At present, we do not understand the reason such differences in binding were observed. Although further kinetic studies are necessary, reaction rate in the binding should be considered. We would like to emphasize that the present method is applicable to determine affinity constant (Ka) in the binding between carbohydrates and lectins simultaneously. Because we need not measure the concentrations of the fluorescent ligand (i.e., carbohydrates in this case) as shown in the binding between chito-oligosaccharides (Figure 6), the present method is especially useful for the measurement of kinetics of a complex mixture of carbohydrates derived from biological samples. We can easily select appropriate concentrations of the lectins in the running buffer, because the lectins have specific binding constants for some selected carbohydrates as shown in Figure 8. An appropriate combination of these lectins successfully allowed categorization of carbohydrate chains. Total analysis time required for one glycoprotein sample was within 2 h. Total amount of a glycoprotein sample was 2 µg (50 pmol), for example, for the analysis of AGP. Although we focused the present study on the discrimination of basic structural characteristics of carbohydrate chains (i.e., asialocarbohydrate chains), development of the method for characterization of sialic acid-containing carbohydrates will be the next target. In a recent paper,36 Feizi et al. proposed a microarray method for the determination of protein-carbohydrate interactions. More than 30 purified oligosaccharides were immobilized on a nitrocellulose membrane, and the interactions with specific proteins were examined. Although the microarray technique has several advantages, we should notice that preparation of purified oligosaccharides for immobilization is generally quite difficult. Furthermore, it is difficult to express or knock out specific oligosaccharides, because numerous glycosyltransferases and glycosidases are involved in their biosynthesis. Another approach described here can use a mixture of carbohydrate chains from biological sources. At present, we need at least a 5-µL sample solution (pmol order) for injection Journal of Proteome Research • Vol. 2, No. 1, 2003 87

research articles due to the restriction of the injection device, although the actual amount required for the present analysis is only amol or fmol levels as the injected amount. It is necessary to miniaturize the injection and analytical devices to detect carbohydrateprotein interaction at the cell level. In the present paper, we employed APTS as a fluorescent labeling reagent for carbohydrates. We have to evaluate the labeling efficiency of the present technique to improve the sensitivity. However, the present strategy will be quite useful for the understanding post-translational modification of proteins with carbohydrate chains, and will lead to a research such as carbohydrate-deficiency glycoprotein syndrome. Capillary affinity electrophoresis using a multichannel-microchip apparatus will allow high-throughput screening of carbohydrate chains in biological samples and lead to glycome analysis next in the proteome era.

References (1) Kakehi, K.; Funakubo, T.; Suzuki, S.; Oda, Y.; Kitada, Y. J. Chromatogr. 1999, 863, 205-218. (2) Kakehi, K.; Kinoshita, M.; Nakano, M. Biomed. Chromatogr. 2002, 16, 103-115. (3) Rice, K. G.; Takahashi, N.; Namiki, Y.; Tran, A. D.; Lisi, P. J.; Lee, Y. C. Anal. Biochem 1992, 206, 278-287. (4) Jackson, P. Biochem. Soc. Trans. 1993, 21, 121-125. (5) Mechref, Y.; Novotny, M. V. Chem. Rev. 2002, 102, 321-369. (6) Evangelista, R. A.; Liu, M. S.; Chen, F. T. A. Anal. Chem. 1995, 67, 2239-2245. (7) Sei, K.; Nakano, M.; Kinoshita, M.; Masuko, T.; Kakehi, K. J. Chromatogr. A. 2002, 958, 273-281. (8) Kinoshita, M.; Murakami, E.; Oda, Y.; Funakubo, T.; Kawakami, D.; Kakehi, K.; Kawasaki, N.; Morimoto, K.; Hayakawa, T. J. Chromatogr. A 2000, 866, 261-271. (9) Kakehi, K.; Kinoshita, M.; Kawakami, D.; Tanaka, J.; Sei, K.; Endo, K.; Oda, Y.; Iwaki, M.; Masuko, T. Anal. Chem. 2001, 73, 26402647. (10) Shinohara, Y.; Hasegawa, Y.; Kaku, H.; Shibuya, N. Glycobiology 1997, 7, 1201-1208. (11) Oda, Y.; Kinoshita, M.; Nakayama, K.; Kakehi, K. Biol. Pharm. Bull. 1998, 21, 1215-1217. (12) Lee, Y. C.; Kawasaki, N.; Lee, R. T.; Suzuki, N. Glycobiology 1998, 8, 849-856.

88


Nakajima et al. (13) Kasai, K.; Oda, Y.; Nishikata, M.; Ishii, S. J. Chromatogr. Biomed. Appl. 1986, 376, 33-47. (14) Schriemer, D. C.; Bundle, D. R.; Li, L.; Hindsgaul, O. Angew. Chem., Int. Ed. 1998, 37, 3383-3387. (15) Zhang, B.; Palcic, M. M.; Mo, H.; Goldstein, I. J.; Hindsgaul, O. Glycobiology 2001, 11, 141-147. (16) Kakehi, K.; Oda, Y.; Kinoshita, M. Anal. Biochem. 2001, 297, 111116. (17) Shimura, K.; Kasai, K. Anal. Biochem. 1997, 251, 1-16. (18) Taga, A.; Uegaki, K.; Yabusako, Y.; Kitano, A.; Honda, S. J. Chromatogr. A. 1999, 837, 221-229. (19) Hong, M.; Cassely, A.; Mechref, Y.; Novotny, M. V. J. Chromatogr. B 2001, 752, 207-216. (20) Shimura, K.; Kasai, K. Anal. Biochem. 1995, 227, 186-194. (21) Chen, F. T.; Evangelista, R. A. Anal. Biochem. 1995, 230, 273280. (22) Waheed, A.; Salahuddin, A. Biochem. J. 1972, 128, 49p. (23) Oda, Y.; Minami, K.; Ichida, S.; Aonuma, S. Eur. J. Biochem. 1987, 165, 297-302. (24) Ma, S.; Nashabeh, W. Anal. Chem. 1999, 71, 5185-5192. (25) Morimoto, N.; Nakano, M.; Kinoshita, M.; Kawabata, A.; Morita, M.; Oda, Y.; Kuroda, R.; Kakehi, K. Anal. Chem. 2001, 73, 54225428. (26) Green, E. D.; Adelt, G.; Baenziger, J. U.; Wilson, S.; Halbeek, H. V. J. Biol. Chem. 1988, 263, 18253-18268. (27) Stubbs, H. J.; Shia, M. A.; Rice, K. G. Anal. Biochem. 1997, 247, 357-365. (28) Takahashi, N.; Matsuda, T.; Shikami, K.; Shimada, I.; Arata, Y.; Nakamura, Y. Glycoconjugate J. 1993, 10, 425-434. (29) Oda, Y.; Minami, K. Eur. J. Biochem. 1986, 159, 239-245. (30) Oda, Y.; Tatsumi, Y.; Aonuma, S. Chem. Pharm. Bull. 1991, 39, 3350-3352. (31) Kato, I.; Schrode, J.; Kohr, W. J.; Laskowski, M. J. Biochemistry 1987, 26, 193-201. (32) Oda, Y.; Nakayama, K.; Abdul-Rahman, B.; Kinoshita, M.; Hashimoto, O.; Kawasaki, N.; Hayakawa, T.; Kakehi, K.; Tomiya, N.; Lee, Y. C. J. Biol. Chem. 2000, 275, 26772-26779. (33) Dam, T. K.; Brewer, C. F. Chem. Rev. 2002, 102, 387-429. (34) Asensio, J. L.; Canada, F. J.; Siebert, H. C.; Laynez, J.; Poveda, A.; Nieto, P. M.; Soedjanaamadja, U.; Gabius, H. J.; Barbero, J. J. Chem. Biol. 2000, 7, 529-543. (35) Abatangelo, G.; Weigel, P. H. Redefining Hyaluronan; Elsevier B.V.: Amsterdam, 2000. (36) Fukui, S.; Feizi, T.; Galstian, C.; Lawson, A. M.; Chai, W. Nature Biotechnol. 2002, 20, 1011-1017.

PR020009V

Capillary Affinity Electrophoresis for the Screening of Post

Recommend Documents