Anal. Chem. 1996, 68, 2573-2579
Bifunctional Labeling Reagent for Oligosaccharides To Incorporate Both Chromophore and Biotin Groups Yasuro Shinohara,* Hiroyuki Sota, Masanori Gotoh, Masahisa Hasebe, Mariko Tosu, Junji Nakao, and Yukio Hasegawa
Department of R&D, Pharmacia Biotech K.K., 5-37 Kami-Ohsaki 4-Chome, Shinagawa-ku, Tokyo 141, Japan Masanobu Shiga
Dojindo Laboratories, Tabaru 2025-5 Mashiki-machi, Kamimashiki-gun, Kumamoto 861-22, Japan
We have developed a convenient and effective method for biotinylation of oligosaccharides at their reducing ends. A novel biotin hydrazide having a phenyl group produced the biotin adduct of N-acetyllactosamine (LacNAc) by simple incubation at 90 °C for 1 h. Although the biotin adduct was obtained as a mixture of several stereoisomers, one of the isomers, cyclic β-glycoside, became predominant upon letting the reaction mixture stand in a weakly acidic state (pH 3.5). This conversion may be very advantageous for functional analysis of oligosaccharides because natural N-linked oligosaccharides exist in the cyclic β form. The limit of detection of labeled LacNAc in reversed-phase chromatography was 330 fmol and showed good linearity in the range from 330 fmol to 261 pmol. When this procedure was applied to complex type and high mannose type N-linked oligosaccharides, the labeled oligosaccharides were easily detected and separated by reversed-phase, gel filtration, and anion exchange chromatographies. Furthermore, these labeled oligosaccharides were able to be immobilized onto the solid phase using avidin-biotin technology and were stable enough to allow the binding assay to be performed repeatedly and under the conditions for in situ exoglycosidase digestion. These results suggest that this derivatization technique might be useful for both separation and functional analysis of oligosaccharides. Many kinds of lectins and related molecules have been found in a variety of species. The importance of oligosaccharides in typical multicellular organism events such as development, differentiation, and morphogenesis is becoming widely recognized.1 To clarify the functional roles of these molecules, it is necessary to evaluate their sugar binding properties and the mechanisms involved. Many assay systems to analyze the interactions between lectins and oligosaccharides have already been developed, including affinity chromatography,2 microdialysis,3 microcalorimetry,4 (1) Varki, A. Glycobiology 1993, 3, 97-130. (2) Kobata, A. Eur. J. Biochem. 1992, 209, 483-501. (3) Pinckard, R. N. Handbook of Experimental Immunology; Blackwell: Oxford, 1978; Vol. 1, Chapter 17. (4) Bains, G.; Lee, R. T.; Lee, Y. C.; Freire, E. Biochemistry 1992, 31, 1262412628. S0003-2700(96)00004-2 CCC: $12.00
© 1996 American Chemical Society
electrophoresis,5 and NMR.6 However, because these methods usually detect the equilibrium of interactions, information about the on- and off-rate kinetics remains scarce. Recent findings have revealed that the interactions between oligosaccharides and carbohydrate-recognizing molecules are quite complicated, and those interactions need to be analyzed in more detail, rather than simply observing binding or not. Several important aspects of the interactions have been pointed out, including the clustering effect,7 the importance of transient interactions,8 and the effect of blood flow.9 For the functional analysis of an oligosaccharide in the binding assay, development of a more sophisticated analytical method seems to be of premier importance. Recent progress in biospecific interaction analysis using a biosensor based on surface plasmon resonance enables analysis of interactions in real-time in a dynamic flow system,10 which seems to be appropriate for the evaluation of oligosaccharidemediated interactions. We have already shown that the surface plasmon resonance detection coupled with biotin-derivatized oligosaccharides has several advantages for characterizing the sugar-binding specificities of lectins.11,12 In this system, biotinylated oligosaccharides are immobilized on the sensor surface, and the interaction with lectin can be monitored in a microflow system. This method is superior to the other previous methods because the interaction can be monitored kinetically and with high sensitivity. However, when applying this procedure to oligosaccharides derived from natural sources, frequent problems that must be overcome are the low quantities of material and extensive heterogeneity in their structures. Therefore, an advanced strategy is required for the purification of oligosaccharides. Introducing a biotinyl group to the reducing end of an oligosaccharide enables preparation of neoglycoproteins13 and an (5) Honda, S.; Taga, A.; Suzuki, K.; Suzuki, S.; Kakehi, K. J. Chromatogr. 1992, 597, 377-382. (6) Jordan, F.; Basset, E.; Redwood, W. R. Biochem. Biophys. Res. Commun. 1977, 75, 1015-1021. (7) Lee, Y. C. FASEB J. 1992, 6, 3193-3200. (8) van der Merwe, P. A.; Barclay A. N. Trends Biochem. Sci. 1994, 19, 354358. (9) Alon, R.; Hammer, D. A.; Springer, T. A. Nature 1995, 374, 539-542. (10) Granzow, R.; Reed, R. Bio/Technology 1992, 10, 390-393. (11) Shinohara, Y.; Kim, F.; Shimizu M.; Goto, M.; Tosu, M.; Hasegawa, Y. Eur. J. Biochem. 1994, 223, 189-194. (12) Shinohara, Y.; Sota, H.; Kim, F.; Shimizu M.; Gotoh, M.; Tosu, M.; Hasegawa, Y. J. Biochem. 1995, 117, 1076-1082. (13) Yet, M.-G.; Yan, S.-C. B.; Wold, F. FASEB J. 1988, 2, 22-31.
Analytical Chemistry, Vol. 68, No. 15, August 1, 1996 2573
oligosaccharide-immobilized solid phase for the binding assay.14 With this objective, several biotinylation procedures have already been reported.15,16 However, because most of these procedures are designed to incorporate only a biotinyl group, they are hardly applicable for purification use because of low sensitivity. For this purpose, the labeling reagent should be bifunctional; it must possess both a biotinyl group and some chromophore. From this viewpoint, biotinyldiaminopyridine (BAP)17 seems to be the only choice. However, BAP requires a high reagent concentration, and the derivatization is not quantitative. Furthermore, the conjugation of BAP with an oligosaccharide is based on reductive amination, which leads to fixation of the reducing terminal residue in an acyclic form. Formation of nonbiological determinants by ringopening events may affect the biological activity of the glycan moiety. Here, we have synthesized a novel biotin hydrazide derivative having ultraviolet absorbance, and we studied the labeling conditions and the chromatographic behavior of the labeled oligosaccharides. By carefully investigating the reaction conditions, we found that the cyclic β-glycoside can be selectively obtained by this derivatization, as is found in the N-glycosidic linkage between proteins and oligosaccharides. Finally, we evaluated the feasibility of the labeled oligosaccharides for the binding assay with the biosensor. EXPERIMENTAL SECTION Materials. Sambucus sieboldiana lectin (SSA), Maackia amurensis lectin (MAM), and Ricinus communis agglutinin-120 (RCA120) were purchased from Honen Seiyu (Tokyo). N-Acetyllactosamine (LacNAc) was obtained from Toronto Research Chemicals (Downsview, Canada), and 6′- and 3′-sialyl LacNAc (6′- and 3′-SLN) were purchased from Oxford GlycoSystems (Abingdon, U.K.). The abbreviations and structures of the other oligosaccharides used in this study are shown in Chart 1. All the oligosaccharides shown in Chart 1 were purchased from Oxford GlycoSystems. BIAcore sensor chip SA-5, surfactant P20, and chemical activation reagents were obtained from Pharmacia Biosensor AB: 100 mM Nhydroxysuccinimide (NHS) in water, 400 mM N-ethyl-N′-(3dimethylaminopropyl)carbodiimide hydrochloride (EDC) in water, and 1 M ethanolamine hydrochloride adjusted to pH 8.5 using NaOH. The HBS buffer comprised 10 mM HEPES, pH 7.4, 150 mM NaCl, and 0.05% BIAcore surfactant P20 in distilled water. Arthrobacter ureafaciens sialidase was purchased from Boehringer Mannheim (Mannheim, Germany). Synthesis, Purification, and Characterization of 4-(Biotinamido)phenylacetylhydrazide (BPH, Figure 1). Thionyl chloride (6.69 mL, 93.3 mmol) was dissolved in iced methanol (50 mL). After the mixture was stirred for 30 min, p-aminophenylacetic acid (1, 3.0 g, 19.9 mmol) was added. The reaction mixture was stirred for 30 min at 0 °C, and then for 20 h at room temperature. The mixture was evaporated to dryness, and the residue was dissolved in methanol. Crystallization from ether yielded 3.7 g (yields, 91.8%) of 4-aminophenylacetic acid methyl ester hydrochloride (2). Biotin (3 g, 12.3 mmol) was suspended in DMF (80 mL), and TEA (3.4 mL, 24.4 mmol) and isobutyl chloroformate (2.1 mL, 16.4 mmol) were added and stirred for (14) Shao, M.-C. Anal. Biochem. 1992, 205, 77-82. (15) Hase, S. J. Biochem. 1992, 112, 266-268. (16) Manger, I. D.; Wong, S. Y. C.; Rademacher, T. W.; Dwek, R. A. Biochemistry 1992, 31, 10733-10740. (17) Toomre, D.; Varki, A. Glycobiology 1994, 4, 653-663.
2574
Analytical Chemistry, Vol. 68, No. 15, August 1, 1996
Chart 1. Structures and Abbreviations of Oligosaccharides Used in This Study
30 min at 0 °C. Then, 2 (2.3 g, 12.1 mmol) was added, and the mixture was stirred for 16 h at room temperature. The mixture was evaporated, and the residue was fractionated on a column of silica gel (silica gel 60, Merck) to obtain 4-(biotinamido)phenylacetic acid methyl ester (3, 3.2 g, 66.5%). Compound 3 (2.0 g, 5.11 mmol) was suspended in hydrazine monohydrate (64 mL, 1.32 mol) and the suspension stirred for 24 h at room temperature. Crystallization from ether yielded 1.59 g (79.5% yield) of the titled compound: UV λmax 252 nm ( 16 000 in 30% acetonitrile); positiveion fast atom bombardment mass spectrometry m/z 392 (M + H)+. BPH Tagging of Oligosaccharides (Figure 1). LacNAc was used as a model oligosaccharide to determine the optimal conditions for the reaction with BPH. LacNAc (1.3 pmol-10 nmol, in 10 µL) in water and was incubated with BPH (3-83 nmol, in 10 µL) in 30% acetonitrile at 90 °C for 1 h. The reaction mixture was cooled in an ice bath. Twenty microliters of 50 mM formate buffer (pH 3.5) was added to the reaction vial and allowed to stand at 4 °C. The reaction mixture was directly injected into reversedphase HPLC to purify the BPH adduct for NMR analysis. Purified BPH-labeled LacNAc (∼500 µg in D2O) was examined with a 400MHz NMR. A one-dimensional spectrum was recorded at 27 °C
pH 9.0, containing 20% acetonitrile) and solvent B (500 mM ammonium acetate buffer, pH 8.0, containing 20% acetonitrile). After injection, solvent A was passed through the column for 10 min, after which a 25-min linear gradient up to a 50:50 ratio of A:B was applied. For gel filtration chromatography, two linked Superdex peptide columns (Pharmacia Biotech, 3.2 mm × 300 mm) were used isocratically with 20% acetonitrile at 50 °C at a flow rate of 0.4 mL/min. Interaction Analysis with a Biosensor Based on SPR. A biosensor, BIAcore (Pharmacia Biosensor AB, Uppsala, Sweden), which is based on surface plasmon resonance, was used to measure the biomolecular interactions. Immobilization of the BPH-labeled oligosaccharide was carried out according to the procedure described in a previous report.12 The BPH-labeled oligosaccharide (1-10 pmol) was introduced onto a streptavidin preimmobilized sensor surface (sensor chip SA-5, Pharmacia Biosensor AB). MAM and SSA were purified by Superdex 200 (Pharmacia Biotech) using HBS buffer as solvent. The lectin concentrations were determined with use of a micro-BCA protein assay kit (Pierce, Rockford). MAM and SSA solutions (75 and 15 µg/mL in HBS buffer, respectively) were injected across the surface at a flow rate of 10 µL/min. The interaction was monitored at 25 °C as the change in the SPR response. After 3 min of monitoring, the HBS buffer was introduced onto the sensor chip in place of the lectin solution to start the dissociation. Regeneration was accomplished by washing away the surface-bound lectin with 50 mM H3PO4. The equilibrium response (in RU) of lectins at the injected lectin concentration (Req) was obtained from the SPR signal binding data and calculated using the BIA evaluation software 2.1 (Pharmacia Biosensor AB).12 The association rate for the binding between the biotinyl glycan and lectins can be expressed by the following equation: Figure 1. Structure and synthesis of BPH and the chemistry of adduct formation with oligosaccharides.
and referenced to DHO as 4.80 ppm. Partial assignments of the spectral resonances of derivatives were made by comparison with reference compounds.18,19 Detection and Separation of Oligosaccharide-BPH Adducts. All chromatographic detections were performed at 252 nm. Reversed-phase chromatography was conducted on a TSKgel ODS 80 column (Tosoh, 4.6 mm × 250 mm). Separation of BPHlabeled oligosaccharides was performed at 40 °C at a flow rate of 0.5 mL/min. Elution was performed both isocratically and linearly changed using solvent A (70 mM phosphate buffer, pH 6.8, containing 10% acetonitrile) and solvent B (70 mM phosphate buffer, pH 6.8, containing 30% acetonitrile). For the analysis of BPH-labeled LacNAc, elution was performed isocratically with 20% B. Analysis of BPH adducts of N-linked oligosaccharides was performed in a 30-min linear gradient from 100% A to 80% A, followed by a 30-min linear gradient from 80% A to 50% A. Anion exchange chromatography was performed on a MonoQ column (Pharmacia Biotech, 1.6 mm × 50 mm). Separation of BPHlabeled oligosaccharides was performed at ambient temperature at a flow rate of 0.05 mL/min. Elution was performed by a linear gradient using solvent A (aqueous ammonia diluted with water, (18) Perkins, S. J.; Johnson, L. N.; Phillips, D. C.; Dwek, R. A. Carbohydr. Res. 1977, 59, 19-34. (19) Ikura, M.; Hikichi, K. Org. Magn. Reson. 1982, 20, 266-273.
dR/dt ) -(kassocC + kdissoc)R + kassocCRmax
(1)
where kassoc is the association rate constant and kdissoc is the dissociation rate constant, R is the amount of bound lectin measured by the SPR response (RU) at time t, Rmax is the maximum binding capacity (in RU) of lectins, and C is the constant concentration of lectin injected. The Req as well as ks was analyzed by fitting the association phase to the integrated form of eq 1:
Rt ) Req[1 - exp(-kst)]
(2)
with nonlinear least-squares analysis,20 where Req ) CkassocRmax/ (Ckassoc + kdissoc) and ks ) Ckassoc + kdissoc. In Situ Enzymatic Digestion of BPH-Labeled Oligosaccharide on the Sensor Surface. 3′-SLN was used as a model oligosaccharide. 3′-SLN was BPH-labeled as described above and immobilized onto the surface. A. ureafaciens sialidase (2.5 munits/ µL, 10 µL) was introduced onto the surface, and the flow rate was set at zero. After the injection of sialidase, the surface was washed with 50 mM phosphoric acid at regular intervals to chase the time course of enzyme digestion by measuring the interaction with MAM and RCA120. The concentrations used were 100 µg/ mL for MAM and 1 µg/mL for RCA120. (20) O’Shannessy, D. J.; Brigham-Burke, M.; Soneson, K. K.; Hansley, P.; Brooks, I., Anal. Biochem. 1993, 212, 457-468.
Analytical Chemistry, Vol. 68, No. 15, August 1, 1996
2575
RESULTS AND DISCUSSION BPH Tagging of Oligosaccharides and Tautomerization Procedure. Oligosaccharides were efficiently coupled with BPH at the reducing terminus by simple condensation with BPH at 90 °C. When LacNAc was used as a model sugar and was reacted with different concentrations of BPH, the peak area of BPH adducts reached a plateau if more than a 2-fold molar excess of BPH was used. However, there were several new peaks detected after the reaction, which might be stereoisomers of the products (Figure 2, panel A-a), acyclic Schiff base types, and cyclic hydrazino types. This result was unfavorable because of the inconvenience of purifying the BPH adducts in the oligosaccharide mixture by reversed-phase chromatography and the uncertainty of the results of binding analysis with lectins due to the structural heterogeneity in the reducing end of the oligosaccharide. To overcome these problems, we further studied the effect of the pH on the formation of hydrazone during reaction and storage. Although the reaction of sugars with substituted hydrazine is thought to proceed through nucleophilic attack of the hydrazine at the hemiacetal carbon atom of a protonated form of the sugar, our previous results indicated that an acidic catalyst was not necessary under the conditions we employed.12 In this case, the reaction yield was even lower when the reaction was carried out under acidic conditions (pH 4.0), while the proportion of each stereoisomer was significantly different (Figure 2, Panel A-b). However, after storing the reacted mixture under acidic conditions by adding formate buffer (pH 3.5) at 4 °C for 12 h, one peak became predominant, while almost no change was detected when the mixture was stored under neutral conditions (Figure 2, panel A-c,d). In the 1H-NMR analysis of the peak, the resonances of the anomeric protons of the GlcNAc moiety of LacNAc at 5.24 ppm (R) and 4.51 ppm (β) were lost after the reaction, and a single anomeric resonance was observed at 4.22 ppm (d, 1H, J1,2 ) 9.53 Hz, βH1 proton of GlcNAc moiety of the product). These results indicate that the other isomers besides β-glycoside were tautomerized to β-glycoside under acidic conditions. On the other hand, the reverse tautomerization was not detected when the β-glycoside was allowed to stand under neutral conditions. Kinetic analysis of the tautomerization showed that the peak areas of the cyclic β-glycoside increased gradually for the first 3 h during storage, and then the peak area reached a plateau (Figure 2, panel B). The other peaks except cyclic β-glycoside disappeared rapidly after the addition of formate buffer for the first 5 h. Upon prolonging the storage time beyond 5 h, all the peaks including β-glycoside showed a slight decrease. Since the peak area of BPH showed a slight increase at this period, BPH adducts were gradually decomposed in acidic conditions to give BPH and LacNAc. To minimize the other peaks except β-glycoside, the storage should be continued for 5 h. The overall yield to produce β-glycoside from LacNAc was estimated to be ∼70%. To avoid undesirable decomposition of the β-glycoside in acidic conditions, it is recommended to neutralize the reaction mixture or purify the biotin adduct by chromatography. The cyclic β-glycoside was fairly stable under neutral conditions. The effect of pH on the tautomerism of saccharide hydrazones during storage has been less studied, while many derivatization procedures using hydrazone chemistry have been reported.21-23 (21) Zhang, R.-E.; Cao, Y.-L.; Hearn, M, W. Anal. Biochem. 1991, 195, 160167.
2576 Analytical Chemistry, Vol. 68, No. 15, August 1, 1996
Figure 2. Effect of pH on the formation of hydrazone during reaction and storage (A), and time course of the tautomerization of BPHlabeled N-acetyllactosamine (LacNAc) (B). (A) LacNAc was incubated with a 2-fold molar excess of BPH for 1 h at 90 °C in water (a) and in 50 mM formate buffer (pH 4.0) (b). To evaluate the effect of pH on the formation of hydrazone during storage, a portion of (a) was mixed with an equal volume of water or 50 mM formate buffer (pH 3.5). Both were allowed to stand at 4 °C for 12 h and were then analyzed again (c,d). Separation was done by reversed-phase chromatography. Peak assignments: 1-3, BPH-adducts; 4, BPH. (B) LacNAc was incubated with a 4-fold molar excess of BPH at 90 °C for 1 h. Formate buffer (pH 3.5) was added to the reaction vial, and an aliquot of the reaction mixture was injected into the reversed-phase column to monitor the peak area of cyclic β-glycoside (b), the other stereoisomers (O), and BPH (0).
Figure 3. Chromatograms showing the separation of BPH-labeled oligosaccharides by reversed-phase chromatography: (a) glucose oligomers, (b) high-mannose type oligosaccharides (M5-M9), and (c) complex type asialooligosaccharides.
The only exception is the tautomerism study of monosaccharide hydrazone in solution performed by Williams.24 According to that study, the derived saccharide hydrazine after hydrazinolysis first formed mainly as the syn acyclic tautomer, and then slow tautomerization occurred to give glycosylhydrazines, and the tautomerization was rapid under slightly acidic conditions (pH 6.0) compared with slightly basic or neutral conditions. These results are in good agreement with our present findings. Considering the fact that native N-linked oligosaccharides exist as the β-N-glycoside form, this derivatization and tautomerization procedure seems to have a great advantage for measuring the sugar-binding specificities of lectins and characterizing the function of oligosaccharides. Calibration Graph for BPH-Labeled LacNAc on ReversedPhase Chromatography. LacNAc was diluted over a range of from 1.3 pmol to 1 nmol; 3 nmol of BPH was added to each vial, and derivatization and tautomerization were carried out. Onefourth of each reaction mixture was then injected into the reversed-phase column, and the peak area of BPH-labeled LacNAc was calculated. The response was linear over the range from 330 fmol to 261 pmol (r ) 0.9985). These results also indicate that a few picomoles of an oligosaccharide is enough for the derivatization and detection by chromatography. Separation of BPH-Labeled Oligosaccharides by ReversedPhase, Gel Filtration, and Anion Exchange Chromatographies. The chromatographic behaviors of BPH-derivatized oligosaccharides were evaluated using reversed-phase, gel filtration, and anion exchange chromatographies. Chromatograms showing the separation of oligosaccharides by reversed-phase chromatography are shown in Figure 3. Glucose oligomers were well separated up to 22-mer in 60 min. Some typical N-linked oligosaccharides, including the high-mannose type and complex (22) Zhang, R.; Zhang, Z.; Liu, G.; Hidaka, Y.; Shimonishi, Y. J. Chromatogr. 1993, 646, 45-52. (23) Muramoto, K.; Yamauchi, F.; Kamiya, H. Biosci. Biotechnol. Biochem. 1994, 58, 1013-1017. (24) Williams, J. M. Carbohydr. Res. 1983, 117, 89-94.
type oligosaccharides, were also well separated. The BPH-labeled 3′-sialyl-N-acetyllactosamine (3′-SLN) and 6′-sialyl-N-acetyllactosamine (6′-SLN) were also well separated under this condition, with modification of LacNAc by sialic acid at the 6′ position decreasing retention time. Since no peak was detected at the eluting position for BPH-LacNAc in each case, our previous finding that the derivatization procedure was applicable to sialylated oligosaccharides without any removal of sialic acid was also true for BPH. Gel filtration chromatography was carried out using two tandemly linked Superdex peptide columns. As shown in Figure 4A, up to 14-mer glucose oligomers were well separated in 90 min. Mono-, di-, and trisialylated N-linked oligosaccharides were well separated by MonoQ according to their charges (Figure 4B). This desirable chromatographic behavior demonstrates the feasibility of BPH tagging for use in chromatographic purification. Stability and in Situ Enzymatic Digestion of BPH-Labeled Oligosaccharides on the Sensor Surface of the Biosensor. The BPH-labeled oligosaccharides were successfully immobilized onto the streptavidin preimmobilized surface of BIAcore. To evaluate the stability of BPH-labeled oligosaccharides on the sensor surface, the effects of acid contact were evaluated. For successive measurements of the interaction using the same sensor surface, it is necessary to establish the appropriate regeneration conditions. We have already shown that 50 mM phosphoric acid gave a satisfactory results for this purpose. In our previous study, a slight decrease in the response was detected when the interaction analysis was carried out successively (
