Comprehensive Approach to Structural and ... - ACS Publications

Jan 19, 2008 - ... School of Advanced Life Science, and Frontier Research Center for ..... commercially available from Sumitomo Bakelite in the near f...
0 downloads 0 Views 598KB Size
Download PDF
Anal. Chem. 2008, 80, 1094-1101

Comprehensive Approach to Structural and Functional Glycomics Based on Chemoselective Glycoblotting and Sequential Tag Conversion Jun-ichi Furukawa,† Yasuro Shinohara,† Hiromitsu Kuramoto,†,‡ Yoshiaki Miura,† Hideyuki Shimaoka,†,‡ Masaki Kurogochi,† Mika Nakano,†,§ and Shin-Ichiro Nishimura*,†

Laboratory of Advanced Chemical Biology, Graduate School of Advanced Life Science, and Frontier Research Center for Post-Genome Science and Technology, Hokkaido University, Sapporo 001-0021, Japan, Bio Product Development Project Team, Sumitomo Bakelite Co. Ltd., Tokyo 140-0002, Japan, and Discovery Research Laboratories, Shionogi & Co. Ltd., Osaka 553-0002, Japan

Changes in protein glycosylation profoundly affect protein function. To understand these effects of altered protein glycosylation, we urgently need high-throughput technologies to analyze glycan expression and glycan-protein interactions. Methods are not available for amplification of glycans; therefore, highly efficient sample preparation is a major issue. Here we present a novel strategy that allows flexible and sequential incorporation of various functional tags into oligosaccharides derived from biological samples in a practical manner. When combined with a chemoselective glycoblotting platform, our analysis enables us to complete sample preparation (from serum to released, purified, methyl-esterified, and labeled glycans) in 8 h from multiple serum samples (up to 96 samples) using a 96-well microplate format and a standard de-N-glycosylation protocol that requires reductive alkylation and tryptic digestion prior to PNGase F digestion to ensure maximal de-N-glycosylation efficiency. Using this technique, we quantitatively detected more than 120 glycans on human carcinoembryonic antigens for the first time. This approach was further developed to include a streamlined method of purification, chromatographic fractionation, and immobilization onto a solid support for interaction analysis. Since our approach enables rapid, flexible, and highly efficient tag conversion, it will contribute greatly to a variety of glycomic studies. The importance of protein glycosylation is becoming widely realized from a growing number of embryonic lethal phenotypes seen in knockout mice with defects in glycoconjugate assembly or processing1 and also from the involvement of glycosylation in each and every aspect of tumor progression, from cellular proliferation to angiogenesis and metastasis.2 Hence, there is a * To whom correspondence should be addressed. Phone: +81 11 706 9043. Fax: +81 11 706 9042. E-mail: [email protected]. † Hokkaido University. ‡ Sumitomo Bakelite Co. Ltd. § Shionogi & Co. Ltd. (1) Yamashita, T.; Wada, R.; Sasaki, T.; Deng, C.; Bierfreund, U.; Sandhoff, K.; Proia, R. L. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 9142-9147. (2) Fuster, M. M.; Esko, J. D. Nat. Rev. Cancer 2005, 5, 526-542.

1094 Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

need for a large-scale structural and functional analysis of protein glycosylation (glycomics). Glycans, the carbohydrate portion of glycoconjugates such as glycoproteins, lack strong UV absorption and their physicochemical properties do not vary widely. Therefore, they are often derivatized (typically at the reducing end of the oligosaccharide chains) to introduce a chromophore or a fluorophore; this increases detection while simultaneously enhancing the solutes’ hydrophobicity to enable separation. Derivatization also increases the relatively low response of mass spectrometric analyses to oligosaccharides. In addition, oligosaccharides must often be immobilized onto a solid support for the study of carbohydratebinding proteins. The derivatization reagent employed usually differs depending on the purpose of the study and the analytical technique chosen, but in response to these diverse demands, efforts have been directed to making the derivatization reagent multifunctional.3-6 Although multifunctional reagents are useful in that they can introduce multiple functions into oligosaccharides, once the functional group has fulfilled its intended purpose, it has essentially no role and is even potentially disadvantageous. For instance, lectins can possess a hydrophobic cavity adjacent to the carbohydrate binding site, which could artificially enhance the binding affinity upon derivatization of the oligosaccharide with a hydrophobic tag.7 It would be ideal if the introduced tag could be flexibly replaced depending on the purpose of the study. Recently, we reported that carbohydrates can be rapidly and efficiently purified by employing glycan-specific chemical ligation onto aminooxy-functionalized polymers, which we termed “glycoblotting”.8,9 In this technique, oligosaccharides in a crude mixture can be selectively (3) Hsu, J.; Chang, S. J.; Franz, A. H. J. Am. Soc. Mass Spectrom. 2006, 17, 194-204. (4) Xia, B.; Kawar, Z. S.; Ju, T.; Alvarez, R. A.; Sachdev, G. P.; Cummings, R. D. Nat. Methods 2005, 2, 845-850. (5) Rothenberg, B. E.; Hayes, B. K.; Toomre, D.; Manzi, A. E.; Varki, A. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 11939-11943. (6) Shinohara, Y.; Sota, H.; Gotoh, M.; Hasebe, M.; Tosu, M.; Nakao, J.; Hasegawa, Y.; Shiga, M. Anal. Chem. 1996, 68, 2573-2579. (7) Loris, R.; Hamelryck, T.; Bouckaert, J.; Wyns, L. Biochim. Biophys. Acta 1998, 1383, 9-36. (8) Nishimura, S.-I.; Niikura, K.; Kurogochi, M.; Matsushita, T.; Fumoto, M.; Hinou, H.; Kamitani, R.; Nakagawa, H.; Deguchi, K.; Miura, N.; Monde, K.; Kondo, H. Angew. Chem., Int Ed. 2004, 44, 91-96. 10.1021/ac702124d CCC: $40.75

© 2008 American Chemical Society Published on Web 01/19/2008

Figure 1. Elucidation of four different types of tag conversions. (a) Schematic representation of imine exchanges. (b) Structures of stable isotope-coded oligosaccharide derivatization reagents. (c) Stability of a hydrazone (AcWR-labeled chitotriose, AcWR-GN3) and an oxime (aoWRlabeled chitotriose, aoWR-GN3). AcWR-GN3, and aoWR-GN3 were heated to various temperatures (40, 60, or 90 °C) under various pH conditions (pH 3-6) for 1 h. The y-axis is the fraction of compound that remained intact following the incubation. (d) Efficiencies of the various types of tag exchange. Either aoWR(H)- or AcWRh(H)-GN3 was incubated with a 10-fold molar excess of aoWR(D) or AcWRh(D) under various pH conditions at 90 °C for 1 h. The y-axis represents yield. Quantification was performed by MALDI-TOF MS.

conjugated onto glycoblotting beads via oxime bond formation between the aminooxy group displayed on the bead and the aldehyde/ketone group at the reducing terminus of all oligosaccharides. Furthermore, we found that oligosaccharides covalently bound to the polymer could be simultaneously released and derivatized in the presence of a large excess of O-substituted aminooxy reagents through transoximization;10-12 the oligosaccharides were released in the form of oxime derivatives of the O-substituted aminooxy compound that had been added.13 Despite the novel observation that transoximization could be targeted to the hemiacetal group of an oligosaccharide and may enable flexible and versatile tag conversion, the exchange yield in the previous study was too low (∼25%) for practical applications. In this study, we evaluated the usefulness of tag conversion using hydrazones as well as oximes to improve the conversion efficiency, since it has previously been reported that hydrazones are more labile to the imine exchange than oximes.10 After finding that tag conversion from hydrazone to oxime proceeds quite efficiently, we prepared novel high-density hydrazide-type glycob(9) Niikura, K.; Kamitani, R.; Kurogochi, M.; Uematsu, R.; Shinohara, Y.; Nakagawa, H.; Deguchi, K.; Monde, K.; Kondo, H.; Nishimura, S.-I. Chem. Eur. J. 2005, 11, 3825-3834. (10) Cousins, G. R. L.; Poulsen, S. A.; Sanders, J. K. M. Chem. Commun. 1999, 16, 1575-1576. (11) Polyakov, V. A.; Nelen, M. I.; Nazarpack-Kandlousy, N.; Ryabov, A. D.; Eliseev, A. V. J. Phys. Org. Chem. 1999, 12, 357-363. (12) Goral, V.; Nelen, M. I.; Eliseev, A. V.; Lehn, J. M. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 1347-1352. (13) Shimaoka, H.; Kuramoto, H.; Furukawa, J.-i.; Miura, Y.; Kurogochi, M.; Kita, Y.; Hinou, H.; Shinohara, Y.; Nishimura, S.-I. Chem. Eur. J. 2007, 13, 16641673.

lotting material and established a generally applicable technology platform for high-throughput oligosaccharide purification/derivatization from crude biological mixtures. The power of the glycoblotting and subsequent tag conversion was demonstrated by glycomic analysis of human serum and other model glycoproteins, including carcinoembryonic antigens. Furthermore, we developed a method of multiple tag conversions that enables streamlined purification, chromatographic fractionation, and immobilization onto a solid support. EXPERIMENTAL SECTION Method for Quantitating pH- and Temperature-Dependent Hydrolysis and Tag Conversion in Solution. Tri-N-acetylchitotriose (GN3) was derivatized with NR-((aminooxy)acetyl)tryptophanylarginine methyl ester (aoWR) and NR-acetyltryptophanylarginine hydrazine (AcWRh) as described (see Figure 1)14 and purified by RP-HPLC. For the examination of the hydrolytic stability, the labeled GN3 (0.1 mM, 10 µL) was dissolved in citrate buffer (25 mM, pH 3.0-6.0, 10 µL) and was incubated at 90 °C for 1 h. For the examination of the efficiency of various tag conversions, a solution of reagent for imine exchange (2 mM, 5 µL) was further added into the labeled GN3 solution as described above and was incubated at 90 °C for 1 h. Both starting material and product upon hydrolysis or tag conversion were quantified by comparing relative signal strength of a known concentration of internal standard, GN3 labeled with deuterated reagents. The (14) Shinohara, Y.; Furukawa, J.-i.; Niikura, K.; Miura, N.; Nishimura, S.-I. Anal. Chem. 2004, 76, 6989-6997.

Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

1095

Figure 2. Preparation of a novel hydrazide-functionalized glycoblotting polymer (BlotGlyco H). (a) Preparation of methacrylamide monomer (3). Reagents and conditions: (a) CHCl3, 0 °C, 16 h; b, EDC, CHCl3, 0 °C, 16 h. (b) Preparation of hydrazide-functionalized polymer (5). Reagents and conditions: (c) AIBN, CHCl3, H2O, 60 °C, 16 h; (d) hydrazine monohydrate, room temperature, 2 h. (e) SEM view of polymer particles.

labeled tetra-N-acetylchitotetraose (GN4) with AcWRh or aoWR was prepared in a similar fashion and used as internal standard for the quantitative evaluation of hydrazine-hydrazone and oxime-oxime tag conversions. Enzymatic Deglycosylation by Peptide N-Glycosidase F (PNGase F). Serum was collected from a healhy subject with the permission of the Commission of Bioethics. Glycoproteins (human R1-acid glycoprotein (AGP); bovine fetuin (Sigma-Aldrich, St. Louis, MO). and carcinoembryonic antigens (CEAs) purified from liver metastasis of colon carcinoma (Cortex Biochem, San Leandro, CA) or pleural and ascites fluids (Acris Antibodies, Hiddenhausen Germany) and the human sera (5 µL) were deglycosylated using PNGase F (Hoffman La Roche Chemicals, Penzberg, Germany) as previously described.15 Preparation of a Novel Hydrazide-Functionalized Glycoblotting Polymer (BlotGlyco H). N-[2-[2-(Methylacryloylaminoethoxy)ethoxy]ethyl]-2-methacrylamide (Structure 3 in Figure 2). To a solution of the methacrylamide derivative 2 (40 g, 0.18 mol), monomethyl succinate (29 g, 0.22 mol) in chloroform (350 mL) and EDC (42 g, 0.22 mol) in chloroform (500 mL) were added dropwise for 30 min, and the mixture was stirred under a nitrogen atmosphere for 16 h. After the reaction was completed, as determined by thin-layer chromatography, aqueous sodium bicarbonate was added, and the solution was extracted with chloroform. The separated organic layer was washed with brine, (15) Kita, Y.; Miura, Y.; Furukawa, J.-i.; Nakano, M.; Shinohara, Y.; Ohno, M.; Takimoto, A.; Nishimura, S.-I. Mol. Cell. Proteomics 2007, 6, 1437-1445.

1096 Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

dried over anhydrous magnesium sulfate, and evaporated to give the methyl ester derivative 3 (40 g, 65%). The spectroscopy values were 1H NMR (500 MHz, CDCl3) δ 1.97 (s, 3H), 2.49 (t, J ) 6.9 Hz, 2H), 2.67 (t, J ) 6.9 Hz, 2H), 3.45-3.64 (m, 12H), 3.69 (s, 3H), 5.34 (s, 1H), 5.71 (s, 1H), and 7.28 (s, 1H); MALDI-TOF MS, calculated for C15H26N2NaO6 [M + Na]+: 353.17, found. m/z 352.93 [M + Na]+. Preparation of Methyl Ester Polymer (4). PVA (2.1 g) was dissolved in 800 mL of distilled water in a flask. Solutions of monomer 3 (40 g, 0.12 mol), ethylene glycol dimethacrylate (1.26 g, 6.4 mmol), and R,R′-azobisisobutylonitrile (1.88 g, 11.4 mmol) in chloroform (30 mL) were added dropwise into the PVA solution. The dispersion mixture was stirred under an argon atmosphere at 60 °C for 16 h. The polymer particles were collected by centrifugation and harshly washed with methanol and distilled water to remove unreacted materials and PVA, followed by lyophilization to give the methyl ester polymer 4 (4.0 g, 9.7%). Synthesis of Hydrazide Functionalized Polymer (5). The methyl ester polymer particles 4 (4.0 g) were suspended in hydrazine monohydrate (40 mL), and the mixture was shaken at room temperature for 2 h. The reaction mixture was filtered, and the resulting polymer particles were washed with methanol, 1 M HCl, and distilled water and lyophilized to give the hydrazide polymer 5 (4 g) in quantitative yield. The morphological features of the polymer particles were observed by scanning electron microscopy (SEM). The dry polymer particles were coated with Pt/Pd (300 Å in thickness). SEM images were obtained using a JEOL JSM7401F (FE-SEM) at 1.0 kV/10 µA. The concentration of the active hydrazide functional group in the polymer was estimated using the 2,4,6-trinitrobenzenesulfonate method.16 BlotGlyco H will be commercially available from Sumitomo Bakelite in the near future. General Procedure for Glycoblotting with BlotGlyco H. A suspension of BlotGlyco H in water (500 µL, 5 mg) was transferred to one well of a MultiScreen Solvinert (Millipore). An aliquot of the digested mixture (typically 20 µL) and 180 µL of acetonitrile containing 2% acetic acid were added to the well containing the polymer particles, and the plate was incubated at 80 °C for 45 min. The well was successively washed twice with 200 µL of 2 M guanidine hydrochloride in 16.6 mM ammonium bicarbonate, water, and 1% triethylamine in methanol. Then, 100 µL of 10% acetic anhydride in methanol was added, and the plate was incubated at room temperature for 30 min to cap unreacted hydrazide, followed by washes with water and methanol. Solidphase methyl esterification was performed as described.17 The polymer-surface blotted glycans were recovered upon adding the aminooxy-containing compound aoWR or the hydrazine-containing compound anthraniloyl hydrazine (Ah). Twenty microliters of aoWR or Ah aqueous solution (20-50 mM) and 180 µL of 2% acetic acid in acetonitrile were added to the above polymer, and mixture was incubated at 80 °C for 45 min. Recovery of the labeled oligosaccharides was performed by adding 100 µL of water and collecting the eluent. MALDI-TOF MS. An aliquot (1 µL) of the recovered glycans was directly mixed with 9 µL of 2,5-dihydroxybenzoic acid (DHB; 10 mg/mL in 30% acetonitrile) and subjected to MALDI-TOF MS analysis using an Ultraflex time-of-flight mass spectrometer I and (16) Qi, X. Y.; Keyhani, N. O.; Lee, Y. C. Anal. Biochem. 1988, 175, 139-144. (17) Miura, Y.; Shinohara, Y.; Furukawa, J.-i.; Nagahori, N.; Nishimura, S.-I. Chem. Eur. J. 2007, 13, 4797-4804.

II (Bruker Daltonics) controlled by the FlexControl 2.0 software package. All of the spectra were obtained using a reflectron mode with an acceleration voltage of 25 kV, a reflector voltage of 26.3 kV, and a pulsed ion extraction of 160 ns in the positive ion mode. The spectra were the results of signal averaging of ∼1000 laser shots. All peaks were picked by FlexAnalysis 2.0 using SNAP algorithm that fits isotopic patterns to the matching experimental data. Estimation of N-linked-type oligosaccharide structures was obtained by input of peak masses into the GlycoMod Tool (http:// www.expasy.ch/cgibin/glycomod_form.html) and GlycoSuite (https://tmat.proteomesystems.com/glycosuite/). Multiple Tag Conversions for Solid-Phase Molecular Interaction Analysis. Oligosaccharides of AGP were blotted onto Blotglyco H as described and were recovered as fluorescent oligosaccharides labeled with Ah. Ah-derivatized oligosaccharides were subjected to normal-phase HPLC according to the procedure described.18 Each oligosaccharide’s fraction was collected, added to 10 µL of 0.2 mM biotin hydrazide, and then incubated for 1 h at 90 °C. Excess biotin hydrazide was removed using a simple solid-phase extraction (MassPREP HILIC µElution plate, Waters). Interactions of purified, fractionated, and biotinylated oligosaccharides from AGP were monitored by surface plasmon resonance (SPR) using a Biacore 2000 (Biacore AB, Uppsala, Sweden) as previously described.19 RESULTS Evaluation of the Efficiencies of Various Types of Tag Conversion. The efficiencies of four different types of imine exchanges (Figure 1a) were evaluated quantitatively using MALDITOF MS with stable isotope-coded labeling reagents. We recently developed aoWR(H) and aoWR(D), a deuterated (d3-methyl) analogue of aoWR(H), (Figure 1b) as aminooxy-functionalized labeling reagents that allow highly sensitive and quantitative analysis by MALDI-TOF.20 To enable quantitative evaluation of hydrazones as well, we synthesized novel stable isotope-coded hydrazide-type reagents by incorporating the deuterated acetyl group into WRh14 (Figure 1b). Hydrazones and oximes are kinetically fairly inert under neutral conditions because the mesomeric effects that lead to their high stability decrease the electrophilicity of the imine and slow down hydrolysis and transimination.21 It is important to balance transimination with hydrolysis for efficient tag exchange, since both hydrolysis and transimination reach high rates under acidic conditions (at pH 4 or below) or at high temperatures11. Therefore, we first examined the stability of the hydrazone and oxime when held at various temperatures under various pH conditions for 1 h. Chitotriose (GN3) labeled with either aoWR(H) or AcWRh(H) was used as model compound. At 90 °C, the hydrazone [AcWRh(H)GN3] was already hydrolyzed ∼50% at pH 6 and was almost completely hydrolyzed at pH values lower than 5 (Figure 1c, left). AcWRh(H)-GN3 was stable at pHs higher than 5 at 60 °C and higher than 4 at 40 °C. The oxime [aoWR(H)-GN3] was also (18) Naka, R.; Kamoda, S.; Ishizuka, A.; Kinoshita, M.; Kakehi, K. J. Proteome Res. 2006, 5, 88-97. (19) Shinohara, Y.; Sota, H.; Kim, F.; Shimizu, M.; Gotoh, M.; Tosu, M.; Hasegawa, Y. J. Biochem. 1995, 117, 1076-1082. (20) Uematsu, R.; Furukawa, J.-i.; Nakagawa, H.; Shinohara, Y.; Deguchi, K.; Monde, K.; Nishimura, S.-I. Mol. Cell. Proteomics 2005, 4, 1977-1989. (21) Carey, F. A.; Sunberg, R. J. Advances Organic Chemistry Part A, 3rd ed.; Plenum Press: New York, 1990; p 451.

hydrolyzed in a pH-dependent manner, but it was much more stable than the hydrazone. Even at 90 °C, GN3-aoWR(H) was stable at pH 5, and ∼80% of aoWR(H)-GN3 survived at pH 4. At 40 or 60 °C, aoWR(H)-GN3 was stable throughout the pH range studied (Figure 1c, right). Once we had clarified the pH- and temperature-dependent stabilities of the hydrazone and oxime, we quantified the efficiencies of the various types of tag exchange. aoWR(H)-GN3 or AcWRh(H)-GN3 was incubated with a 10-fold molar excess of either aoWR(D) or AcWRh(D), under various pH conditions at 90 °C for 1 h. When aoWR(H)-GN3 was incubated with aoWR(D), we observed a yield of ∼20% newly formed oxime [aoWR(D)-GN3] at pHs ranging from 3 to 5 (Figure 1d, upper left). Since the consumption of the starting material [aoWR(H)-GN3] was accelerated in the presence of aoWR(D) (compare with Figure 1c), the newly formed oxime were considered to be generated mainly through a transoximization reaction. When aoWR(H)-GN3 was incubated with AcWRh(H), the ratio of unreacted starting material [aoWR(H)-GN3] was not affected by the presence of AcWRh(H), and newly formed hydrazone [AcWRh(H)-GN3] was scarcely observed at any pH tested (Figure 1d, upper right). In contrast, when AcWRh(H)-GN3 was incubated with aoWR(H), AcWRh(H)-GN3 disappeared at pH 5 or below, and aoWR(H)GN3 was produced in nearly quantitative yields (Figure 1d, bottom left). Consumption of the starting material [AcWRh(H)-GN3] was accelerated in the presence of aoWR(H), indicating that transimination from hydrazone to oxime contributed to the formation of aoWR(H)-GN3. The observed high tag-exchange efficiency may be in part attributable to the relabeling of the regenerated reducing sugar upon hydrolysis, considering that hydrazone is labile to acid hydrolysis at high temperature. When AcWRh(H)-GN3 was incubated with AcWRh(D), the yield of newly formed hydrazone [AcWRh(D)-GN3] increased as the pH value increased and was ∼40% at pH 6 (Figure 1d, bottom right). Based on the results described above, we conclude that the efficiencies of four different types of tag conversions differ substantially and that of hydrazone to oxime proceeds almost quantitatively in only 1 h under the conditions employed. Preparation of a Novel Hydrazide-Functionalized Blotting Polymer, BlotGlyco H. In our previous reports, we demonstrated the feasibility of using an aminooxy-functionalized polymer (BlotGlyco) as a blotting material.13 One limitation was that the recovery yield of bound oligosaccharides from BlotGlyco was rather low (∼25%) when trapped oligosaccharides were recovered via transoximization. In the current study, having shown the superiority of hydrazone over oxime in terms of the tag conversion efficiency, we designed and synthesized a hydrazide-type glycoblotting polymer (BlotGlyco H, Figure 2a, b). The particle size of BlotGlyco H was 10-70 µm (average ∼40 µm) in diameter (Figure 2c). We measured the concentration of active hydrazide groups (which we defined as reactivity to 2,4,6-trinitrobenzenesulfonate) on the particles to be ∼2.1 µmol/mg, indicating that we had successfully produced a high-density hydrazide-functionalized polymer. One-Pot Solid-Phase Glycoblotting and Probing Using BlotGlyco H for Quantitative N-Glycan Profiling. The feasibility of combining BlotGlyco H and subsequent N-glycan recovery upon incubation with an O-substituted aminooxy reagent was first Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

1097

Figure 3. Schematic representation showing the protocol used for glycan sample preparation. Starting from complex biological samples (e.g., serum), N-glycans are enzymatically released and blotted onto BlotGlyco H. After thorough washing to remove nonspecifically bound substances, the covalently bound glycans are subjected to on-bead methyl esterification to stabilize sialic acids and are recovered in the form of oxime derivatives of the O-substituted aminooxy compound that had been added.

Figure 4. MALDI-TOF spectra showing quantitative N-glycan profiles of bovine fetuin (a), human R1-acid glycoprotein (b), and human whole serum glycoproteins (c). Glycan samples of the model glycoproteins and serum were prepared according to the protocol illustrated in Figure 3 (see detailed description in Experimental Section). The amount of sample used for (a), (b), and (c) was 190 µg, 116 µg, and 5 µL, respectively. In (a) and (b), estimated structures are shown. Putative composition of the numbered oligosaccharide signals in (c) are summarized in SI Table 1.

evaluated using bovine fetuin, AGP, and human serum glycoproteins as model samples. Following standardized enzymatic de-Nglycosylation15 of each model glycoprotein sample, the N-glycans were blotted onto BlotGlyco H. After sequential washing, unreacted polymer surface hydrazide groups were acetyl capped, and solid-phase methyl esterification of sialic acid residues17 to render sialylated oligosaccharides chemically equivalent to neutraloligosaccharides was performed. Finally, blotted and processed oligosaccharides were recovered as derivatives of aoWR(H) by heating the polymer to 80 °C for 1 h in the presence of 20 mM aoWR(H) (Figure 3). An aliquot of the recovered fraction was 1098

Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

directly mixed with DHB and was analyzed by MALDI-TOF MS in positive reflector mode. The glycomic profiles of the model glycoproteins agreed well, both quantitatively and qualitatively, with previous reports15,22,23 (Figure 4a-c), and N-glycan signals from human serum are summarized in SI Table 1. Thus, this novel hydrazide-type glycoblotting polymer and tag conversion from hydrazone to oxime functioned as expected. Note that no desialylated products were observed in the spectra for fetuin and AGP, indicating that removal of sialic acid residue(s) is negligible during the whole processes. The reproducibility and accuracy of the described technique were investigated by analysis of human serum whle N-glycans prepared from normal human serum in four separate experimental trials (including glycoblotting, on-bead processing, labeling transimination with aoWR, and MALDI-TOF analysis), and the relative abundance of each oligosaccharide was compared. Low coefficients of variance (1% (SI Figure 1). The overall efficiency including blotting, subsequent processing, and recovery via transimination was 72.5 ( 5.0% (N ) 3). As alternatives to incubation with aminooxy-functionalized compounds (i.e., aoWR), we investigated conditions needed to recover blotted glycans from BlotGlyco H by incubating with hydrazide-functionalized compounds and by acid hydrolysis. Tag exchange from hydrazone to hydrazone was the second most efficient of the four types examined (Figure 1d). By employing a relatively higher concentration of hydrazide compound than aminooxy compound, recovery was improved. For instance, a comparable yield (∼70%) for simultaneous recovery and probing by a hydrazide compound (i.e., AcWRh, benzylhydrazide) was achieved when we used a concentration (∼50 mM) 2.5-fold higher than that of aoWR (∼20 mM). Additionally, the blotted oligosaccharides were quantitatively recovered as unlabeled reducing sugars (92.5% ( 5.5%, N ) 5) by incubating them in 0.1% TFA at 60 °C for 15 min without significant loss of sialic acid(s) (data not shown). Elucidation of the N-Glycosylation Profile of Human Carcinoembryonic Antigen. The described method was further applied to N-glycosylation analysis of CEA, also known as carcinoembryonic antigen-related cell adhesion molecule 5 (CEACAM5) or CD66e. CEA is an oncofetal cell-surface glycoprotein that serves as an important tumor marker for colorectal (22) Ku ¨ ster, B.; Hunter, A. P.; Wheeler, S. F.; Dwek, R. A.; Harvey, D. J. Electrophoresis 1998, 19, 1950-1959. (23) Morelle, W.; Flahaut, C.; Michalski, J. C.; Louvet, A.; Mathurin, P.; Klein, A. Glycobiology 2006, 16, 281-293.

Figure 5. MALDI-TOF spectra showing quantitative N-glycan profiles of human CEAs purified from a liver metastasis of colon carcinoma (a) and pleural and ascites fluids (b and d). The amount of CEAs used for (a), (b), and (d) was 6.6 µg, 6.6 µg, and 100 ng, respectively. Putative composition and relative abundance of the numbered oligosaccharide signals are summarized in SI Table 2. When oligosaccharides from purified CEA obtained from pleural and ascites fluids were digested with R-L-fucosidase, new signals corresponding to glycans A-Y (see structures in SI Table 3) were observed (c).

and some other carcinomas.24,25 CEA contains 28 potential Nglycosylation sites, and ∼50-60% of its 180-kDa molecular mass is attributable to carbohydrates.26 Due to the limited sample availability and its extremely complicated glycosylation profile, however, the structures of the carbohydrate moieties of CEA have only partially been elucidated.27-29 Pioneering work on CEA revealed that CEA purified from liver metastases of primary colon cancer contains ∼25 N-glycans of various types, including highmannose-type oligosaccharides, complex type with or without a core fucose, and bisecting GlcNAc residues.27 Further detailed chemical analyses of the N-glycan structures are only available for CEACAM1 from rat liver30 and from human granulocytes.31 The former provided ∼20 and the latter 35 putative N-glycan (24) Goldenberg, D. M.; Neville, A. M.; Carter, A. C.; Go, V. L.; Holyoke, E. D.; Isselbacher, K. J.; Schein, P. S.; Schwartz, M. J. Cancer Res. Clin. Oncol. 1981, 101, 239-242. (25) Gold, P.; Freedman, S. O. J. Exp. Med. 1965, 121, 439-462. (26) Hammarstrom, S.; Engvall, E.; Johansson, B. G.; Svensson, S.; Sundblad, G.; Goldstein, I. J. Proc. Natl. Acad. Sci. U.S.A. 1975, 72, 1528-1532. (27) Yamashita, K.; Totani, K.; Kuroki, M.; Matsuoka, Y.; Ueda, I.; Kobata, A. Cancer Res. 1987, 47, 3451-3459. (28) Nichols, E. J.; Kannagi, R.; Hakomori, S. I.; Krantz, M. J.; Fuks, A. J. Immunol. 1985, 135, 1911-1913. (29) Chandrasekaran, E. V.; Davila, M.; Nixon, D. W.; Goldfarb, M.; Mendicino, J. J. Biol. Chem. 1983, 258, 7213-7222.

structures following sialidase treatment. When we used our method to analyze the N-glycan profiles of human CEAs (∼6 µg) purified from liver metastases of colon carcinoma and pleural and ascites fluids, we observed very complicated mass spectra, comprising as many as 127 signals (Figure 5a and b). All the numbered signals in Figure 5 were confirmed to be aoWR(H) derivatives by comparing m/z differences when labeling was performed with aoWR(D). Putative compositions and the relative abundance of each oligosaccharide’s signal are summarized in SI Table 2. That oligosaccharides with masses greater than ∼1000 Da exhibited similar signal strengths, irrespective of structure, when examined by MALDI-TOF MS was first communicated by Naven and Harvey,32 and was also confirmed in this study, as shown in SI Figure 2. Of the 127 N-glycans, 52 were not annotated in the database of GlycoSuite. The unannotated signals typically contain either highly branched or polylactosamine structures with multiple fucose residues. All the signals assigned to be fucosylated glycans (SI Table 3) either disappeared or were markedly (30) Kannicht, C.; Lucka, L.; Nuck, R.; Reutter, W.; Gohlke, M. Glycobiology 1999, 9, 897-906. (31) Lucka, L.; Fernando, M.; Grunow, D.; Kannicht, C.; Horst, A. K.; Nollau, P.; Wagener, C. Glycobiology 2005, 15, 87-100. (32) Naven, T. J.; Harvey, D. J. Rapid Commun. Mass Spectrom. 1996, 10, 13611366.

Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

1099

Figure 7. Sequential multiple tag conversion. (a) Normal-phase chromatographic separation of Ah-labeled glycans purified from AGP. Fr. 1-3 correspond to di-, tri-, and tetrasialylated glycans. (b) Sensorgrams showing interactions of the biotinylated glycans of Fr. 1 (left), Fr. 2 (middle), and Fr. 3 (right) with four lectins (Sambucus sieboldiana lectin (SSA), Maackia amurensis lectin (MAM), Ricinus communis agglutinin 120 (RCA120), and concanavalin A (Con A)).

Figure 6. Schematic diagram showing the sequential multiple tag conversion for the construction of glycan array having fractionated naturally derived glycans as contents.

decreased upon R-L-fucosidase digestion (Figure 5c). When the starting amount of CEA was reduced to 100 ng, we still detected more than 80 glycans (Figure 5d). Multiple Sequential Tag Conversions for Purification, Fractionation, and Solid-Phase Interaction Analysis. The described technique was further applied to a solid-phase molecular interaction analysis (Figure 6). AGP was chosen as a model. Oligosaccharides of AGP were blotted onto BlotGlyco H and processed as described, except that the solid-phase methyl esterification step was skipped. Finally, the blotted oligosaccharides were recovered as Ah derivatives. The structure and fluorescence properties of Ah are similar to those of 2-aminobenzamide, which is a fluorescent dye often used for the analysis of oligosaccharides in liquid chromatography.33 Use of hydrazone for the chromatographic separation would be advantageous over oxime since it has been reported that cyclic adducts with β-configuration are produced predominantly in reactions of oligosaccharides with N-substituted hydrazine derivative while both cyclic and acyclic products are generated in reactions of oligosac(33) Bigge, J. C.; Patel, T. P.; Bruce, J. A.; Goulding, P. N.; Charles, S. M.; Parekh, R. B. Anal. Biochem. 1995, 230, 229-238.

1100 Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

charides with O-substituted aminooxy derivatives.6,34,35 Ah-derivatized oligosaccharides were separated and fractionated by normalphase HPLC according to the number of sialic acid residue(s) (Figure 7a). Each fraction was then incubated with a large excess of biotin hydrazide to promote the conversion of Ah derivatives to biotin derivatives. We confirmed the elimination of Ahderivatized oligosaccharides and the appearance of biotin-derivatized oligosaccharides on MALDI-TOF MS spectra. Following the removal of excess reagent in a one-step solid-phase extraction, each biotinylated oligosaccharide’s fraction was introduced onto a surface with preimmobilized streptavidin, and the interactions with various lectins were monitored by SPR. The specific interaction depended on the fraction used (Figure 7b), demonstrating that the multiple tag exchange functions as intended and can be applied to multiple purposes (e.g., purification, separation, and immobilization). This approach may provide a powerful technology platform to construct glycan arrays where naturally derived glycans are used as array contents. DISCUSSION In this study, we examined various imine-exchange reactions targeted to the hemiacetal group of an oligosaccharide and succeeded, for the first time, in establishing a novel methodology that allows an efficient, simple, and sequential tag incorporation/ conversion in a practical manner. That arbitrary functions can be easily and flexibly incorporated depending on the research purpose is a great advantage of the described technique. Although derivatization has become an indispensable technique for structural and functional analyses of oligosaccharides because it allows very sensitive detection,36-38 high-resolution separation,33 or im(34) Ernholt, B. V.; Thomsen, I. B.; Lohse, A.; Plesner, I. W.; Jensen, K. B.; Hazell, R. G.; Liang, X.; Jakobsen, A.; Bols, M. Chem. Eur. J. 2000, 6, 278-287. (35) Lee, M. R.; Shin, I. Org. Lett. 2005, 7, 4269-4272. (36) Yoshino, K.; Takao, T.; Murata, H.; Shimonishi, Y. Anal. Chem. 1995, 67, 4028-4031.

mobilization on a solid support,39 no method has previously been available to flexibly exchange the incorporated tag depending on the research purposes, and this limitation makes functional analysis of naturally derived oligosaccharides difficult. Methods to regenerate free oligosaccharides from labeled glycans have been reported for pyridylamino-labeled40 and Fmoc-labeled glycans.41 However, the former requires multistep reactions and purifications, and the yield is poor (∼30%). Although the conversion efficiency of the latter is fairly good (∼75%), it requires tedious liquid-liquid extractions with diethyl ether for purification, so that application to naturally derived oligosaccharides may be limited. The evaluated tag conversion based on imine exchange was further integrated with the glycoblotting platform, to realize high throughput and multiplexing glycomics study. A novel highdensity hydrazide-type blotting polymer (BlotGlyco H) was prepared for this purpose. Although hydrazide resins are also commercially available (e.g., Affigel Hydrazide from Bio-Rad and UltraLink Hydrazide Gel from Pierce), the concentration of hydrazide groups of these resins is typically 10-20 µmol/mL. The protocol described in this study works well only when BlotGlyco H, whose concentration of hydrazide groups is ∼2.2 µmol/ mg (equivalent to ∼460 µmol/mL), was used as hydrazide resin. This may be attributable to the difference in the density of hydrazide groups on the resin, since efficiency of hydrazone formation is highly dependent on the high-density hydrazide residue. To rapidly profile multiple glycomic samples, such as serum biomarker discovery, the methodology must be reproducible and high throughput. Purification of oligosaccharides from the biological samples is often a stumbling block since several chromatographic steps are typically required to obtain a clean sample for analysis.42 Immobilized lectin affinity chromatography is obviously an efficient technique to rapidly fractionate oligosaccharides from a crude mixture.43 However, no lectin exists to exhaustively recover oligosaccharides regardless of the glycan structure. On the contrary, chemoselective ligation targeted at the reducing terminal8,44 enables rapid and exhaustive recovery of gross glycans from the sample. The chemoselective glycoblotting platform established in this study was demonstrated to be perfectly (37) Kameyama, A.; Kaneda, Y.; Yamanaka, H.; Yoshimine, H.; Narimatsu, H.; Shinohara, Y. Anal. Chem. 2004, 76, 4537-4542. (38) Dell, A.; Morris, H. R. Science 2001, 291, 2351-2356. (39) Vila-Perello´, M.; Gutie´rrez Gallego, R.; Andreu, D. Chembiochem. 2005, 6, 1831-1838. (40) Takahashi, C.; Nakakita, S.; Hase, S. J. Biochem. 2003, 134, 51-55. (41) Kamoda, S.; Nakano, M.; Ishikawa, R.; Suzuki, S.; Kakehi, K. J. Proteome Res. 2005, 4, 146-152. (42) Pilobello, K. T.; Mahal, L. K. Curr. Opin. Chem. Biol. 2007, 11, 300-305. (43) Xiong, L.; Andrews, D.; Regnier, F. J. Proteome Res. 2003, 2, 618-625. (44) Lohse, A.; Martins, R.; Jørgensen, M. R.; Hindsgaul, O. Angew. Chem., Int. Ed. 2006, 45, 4167-4172. (45) Kyselova, Z.; Mechref, Y.; Al Bataineh, M. M.; Dobrolecki, L. E.; Hickey, R. J.; Vinson, J.; Sweeney, C. J.; Novotny, M. V. J. Proteome Res. 2007, 6, 1822-1832. (46) Abbreviations: GN3, tri-N-acetylchitotriose; GN4, tetra-N-acetylchitotetraose; AcWRh, NR-acetyl-tryptophanylarginine hydrazine; aoWR, NR-((aminooxy)acetyl)tryptophanylarginine methyl ester; Ah, anthraniloyl hydrazine; AGP, R1-acid glycoprotein; CEA, carcinoembryonic antigen; SPR, surface plasmon resonance.

compatible with the recently standardized de-N-glycosylation protocol, in which serum was subjected to reductive alkylation in the presence of a surfactant, followed by tryptic digestion prior to PNGase F treatment for maximal deglycosylation efficiency.15 This is of particular importance to ensure the maximum de-Nglycosylation and to accumulate/comparative glycomic data. Our analysis allowed quantitatively detecting 49 N-glycans (both neutral and sialylated oligosaccharides) in human serum glycoproteins of a normal subject. This number is apparently higher than those detected in the recent publications utilizing sophisticated mass spectrometry23,45. Purified and labeled glycans can be prepared in only 8 h from multiple serum samples (up to 96 samples), using the standard de-N-glycosylation protocol in 96-well microplate format. This throughput is astonishingly rapid compared with existing techniques, which require at least a few days for sample preparation and are not amenable to large-scale parallel processing. We have demonstrated the usefulness of the described glycomic analysis by elucidating the very complicated N-glycosylation profiles of CEAs. More than 120 glycans on CEAs were quantitatively detected for the first time. This success may be attributable to the use of the standard deglycosylation protocol, a highly sensitive derivatization reagent, and highly efficient purification. It may be worth mentioning that glycans modified with more than four fucoses were not observed by Naka et al., who recently analyzed the N-glycome of cell membrane fractions from four cancer cell lines (e.g., A549, ACHN, U937, MKN45 cells) and identified 80 distinct N-glycans using a sophisticated sequential HPLC fractionation procedure followed by MALDI-TOF analysis.43 These results may indicate either that hyperfucosylated N-glycans are unique to particular cancers or that our method is extremely sensitive. Since the major N-glycans of CEA were readily detected in 100 ng of sample, our method may be useful for analysis of clinically derived CEA samples. We recently proposed and demonstrated the feasibility of glycomic profiling in developing a focused approach for glycoproteomics (glycoforms-focused reverse proteomics/genomics).20 Trials to discover novel disease-specific glycan markers based on this concept are currently in progress in our laboratory, where the described high-throughput technique is widely used. ACKNOWLEDGMENT We thank Satomi Kudo and Kazue Okada for their expert technical assistance. This work was supported by SENTAN, JST. This work was also supported by Special Coordination Funds for Promoting Science and Technology of the Ministry of Education, Culture, Sports, Science and Technology, the Japanese Government. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review November 15, 2007.

October

16,

2007.

Accepted

AC702124D

Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

1101