- Email: [email protected]
Common conditions for high-performance liquid chromatographic microdetermination of aldoses, hexosamines, and sialic acids in glycoproteins
ANALYTICAL
BIOCHEMISTRY
Common
142, 167-174 (1984)
Conditions for High-Performance Liquid Chromatographic Microdetermination of Aldoses, Hexosamines, and Sialic Acids in Glycoproteins SUSUMU HONDA' ANDSHIGEOSUZUKI
Faculty of Pharmaceutical Sciences, Kinki University, 3-4-l Kowakae, Higashi-Osaka, Japan Received April 19, 1984 All aldoses known to be present in annnal and plant glycoproteins were separated in partition mode on a column of a highly crosslinked cation-exchange resin (Shodex DC-613, H+ form) and sensitively monitored by photometric detection after postcolumn labeling with 2cyanoa&amide. Linearity for the sample amount ranging from 0.1 to 80 nmol was observed with high reproducibility. Hexosamines could be determined as their N-acetyl derivatives under the identical conditions with approximately one order of magnitude higher range of linearity. Sialic acid could be estimated as N-acylhexosamines produced by the action of N-acetylneuraminate pyruvate-lyase. The usefulness of these common conditions was demonstrated by analyzing component monosaccharides of some glycoprotein preparations. Q 1984kadcmic m inc. KEY WORDS: monosaccharide composition analysis of glycoproteins; high-performance liquid chromatography of aldoscs, hexosamines, and sialic acids; analytical application of Nacetylneuraminate pyruvate-lyase.
In structural studies of carbohydrate chains in glycoproteins, the analysis of monosaccharide composition is fundamentally important. The monosaccharide species hitherto reported to be present in glycoproteins include arabinose, xylose, galactose, glucose, mannose, fucose, rhamnose, ZV-acetylglucosamine, and iV-acetylgalactosamine, together with ZV-and O-acylated neuraminic acids. While the identification and determination of these saccharides have been achieved by various chromatographic techniques, high-performance liquid chromatography (HPLC) has proved most convenient because it is rapid and easily automated. There have been numerous papers on HPLC of monosaccharides, and we also have devised conditions for individual classes of monosaccharides, including aldoses (1,2), hexosamines (3,4), and sialic acids (5), as well as uranic acids (6) and alditols (7). However, there have been no general condi’ To whom correspondence should be addressed.
tions applicable to all three classes (aldoses, hexosamines, and sialic acids) of monosaccharides in glycoproteins. The present work has been undertaken to develop common conditions for these classes of monosaccharides. MATERIALS
AND METHODS
Chemicals. Special-grade (for HPLC, Cica brand) samples of 2-cyanoacetamide and acetonitrile were purchased from Kanto Chemicals (Nihonbashi, Chuo-ku, Tokyo). The samples of bovine submaxillary mucin (types I and II), fetal calf serum fetuin (type III), bovine serum acid glycoprotein (fraction IV), hen’s egg albumin (grade V), and human serum transferrin were obtained from Sigma Chemical Company (St. Louis, MO.). Clostridium perfringens neuraminidase (type IV) and Escherichia coli N-acetylneuraminate pyruvate-lyase were obtained from Sigma and Nakarai (Nijo-karasuma, Nakakyo-ku, Kyoto) Chemical Companies, respectively. 167
0003-2697184 $3.00 Copyrieht 0 1984 by Acafhic Press, Inc. All rights of reproduction in any form reserved
168
HONDA
AND SUZUKI
Other chemicals and standard carbohydrates were of the highest grade commercially available. Instrumentation. The following equipment was used: a Hitachi 638 high-performance liquid chromatograph for pumping eluant, a Rheodyne injector carrying a 20-~1 loop, a Riko CB-500-TF hot-water circulator for column temperature regulation, Atto SJ-2396 twin-piston pumps for postcolumn labeling, a Hitachi 034-0736 reaction bath, an Atto SF- 1205-A ultraviolet absorption monitor having an 8-411 quartz cell, and a built-in integrator and data processor. Shodex DC-613 resin (Hf form) was suspended in 2 M hydrochloric acid, and the slurry was heated for several hours at 8090°C. The resin was repeatedly washed with distilled water until the washing fluids became neutral and then packed in a stainless-steel column (4-mm i.d., 25-cm length) by the standard slurry method (pressure range, lOO150 kg./cm2). The column was first washed with distilled water, followed by aqueous acetonitrile with increasing acetonitrile concentrations at a flow rate of 0.3 ml/min. Finally it was equilibrated with aqueous 92 v/v% acetonitrile at 30°C at a flow rate of 0.60 ml/min. The eluate (how rate, 0.60 ml/min) was successively mixed with 0.50 M borate buffer (pH 8.5) and an aqueous 1% 2-cyanoacetamide solution at the same flow rate of 0.50 ml/min by using Y-shaped TPFP connectors, and the resultant effluent was passed through a TPFE reaction coil (0.5-mm i.d., 10-m length) set in an oven thermostated at 100 f 1‘C. After the reaction mixture was cooled by passing it through a l-m TPFE coil of the same bore size, the absorbance at 280 am was recorded as a function of elution time. Hydrolysis of glycoprotein samples. For the analysis of component aldoses, ca. 200 pg of a glycoprotein sample was dissolved in 200 ~1 of 2 M trifluoroacetic acid (TFA),’ and the solution was heated for 6 h on a boiling-water bath under a nitrogen atmo2 Abbreviation used: TFA, trifluoroacetic acid.
sphere. After the solution was cooled to room temperature, it was deionized by passing it through a column containing 1 ml each of Amberlite CG- 120 (H+ form) and CG-400 (CH&OOform), and the column was washed with water. The combined eluate and washing fluids were evaporated to dryness under reduced pressure, and the residue was dissolved in a small volume (30- 100 ~1) of aqueous 92 v/v% acetonitrile. For the analysis of hexosamines, ca. 200 pg of a glycoprotein sample was dissolved in 200 ~1 of 4 M hydrochloric acid and the solution was heated for 6 h on a boilingwater bath under a nitrogen atmosphere. Then the solution was evaporated to dryness, and the residue was treated with a mixture of 0.5 ml of a saturated aqueous solution of sodium bicarbonate and 0.1 ml of acetic anhydride overnight. The mixture was processed in a manner similar to that described for the TFA hydrolysate except for the use of fivefold volumes of Amberlite resins. For the analysis of sialic acids, ca. 200 pg of a glycoprotein sample was dissolved in 800 ~1 of 0.06 M phosphate buffer (pH 7.0) containing neuraminidase (0.5 U) and Nacetylneuraminate pyruvate-lyase (0.3 U), and the mixture was incubated for 1 h at 37’C. After the mixture was heated for 1 min at 100°C to inactivate the enzymes, it was processed in a manner similar to that described for the TFA hydrolysate. Analysis of the component monosaccharides of glycoproteins. A 20-~1 portion of an aqueous acetonitrile solution finally obtained after hydrolysis of a glycoprotein sample was injected into the HPLC column, and the monosaccharides were determined. The content of each monosaccharide was estimated from its peak response and that from the standard monosaccharide, obtained under identical conditions. RESULTS
AND
DISCUSSION
I. Analysis of Aldoses and Hexosamines Optimization of separation. Although component aldoses (1,2) and sialic acids (5)
CHROMATDGRAPHIC
ALDGSE, HEXOSAMINE,
in glycoproteins could be separated by anionexchange chromatography as borate complexes, they were difficult to analyze under isocratic conditions. Borate complexes of hexosamines and their N-acetyl derivatives were easily dissociated on columns giving extremely broad peaks (5), which were superimposed on the peaks of other saccharides. Cation-exchange chromatography was effective for separation of hexosamines, but it resolved neither aldoses nor sialic acids. Therefore, we examined separation by partition chromatography. Since silica-based stationary phases are not durable to repeated analyses, ion-exchange resins of porous polymer type were adopted as stationary phases in this study. Figure la shows the temperature
FIG. 1. Dependence of the separation of aldoses and N-acetylhexosamines derivable from glycoproteins on (a) column temperature and (b) solvent composition. Column: Shodex DC-613 (4-mm i.d., 25cm length, Hf form); column temperature: (a) 4, 15, 25, 30, and 40°C and (b) 30°C; eluant: (a) aqueous acetonitrile, 92 v/v%, and (b) 80, 85, 87.5, 90, and 92 v/v?&; flow rate: 0.60 ml/min; detection: absorption at 280 nm al& postcolumn labeling with 2-cyanoacetamide (for details see Materials and Methods). Rha = rhamnose, Xyl = xylose, Fuc = fucose, Ara = arabinose, Man = mannose, Glc = glucose, Gal = @lactose, GlcNAc = N-acetylglucosamine, GalNAc = N-acetylgalactosamine.
AND SIALIC ACID DETERMINATION
169
dependence of the separation of aldoses and N-acetylhexosamines derivable from glycoproteins, as obtained by using a column of Shodex DC-6 13, a highly crosslinked, sulfonated polystyrene resin which showed the best separation, with aqueous acetonitrile as eluant. Since the presence of metals as counterions caused resolution of anomers (8), due to ligand exchange making determination complicated, the column was used in the H+ form. It is indicated that all these saccharides were eluted faster as the column temperature rose, and that the resolution was sufficiently high at 30°C to resolve all pairs of aldoses except for those of fucose-arabinose and mannose-glucose. The R, value for the fucase-arabinose pair was gradually increased with elevation of column temperature to reach a maximum at 30°C which, however, did not exceed unity. The separation of the mannose-glucose pair was almost independent of column temperature. On the other hand, Fig. 1b shows the effect of solvent composition. Elution time was increased with increasing concentrations of acetonitrile, but the peaks of hexosamines were broadened with tailing and sensitivity fell remarkably when the acetonitrile concentration became higher than 92 v/v%. The relationship between resolution and acetonitrile concentration was similar except for the fucose-arabinose and glucose-galactose pairs. These pairs gave the highest R, values at ca. 88 v/v% acetonitrile, but this concentration was unfavorable for other pairs. Therefore, we used 92 v/v% acetonitrile at 30°C. A flow rate of 0.60 ml/min was the most appropriate. Figure 2 shows a chromatogram for an equimolar mixture of these saccharides, as obtained under these conditions. Although the resolution of the fucose-arabinose and mannose-glucose pairs was incomplete, all major aldoses (fucose, mannose, and galactose) and N-acetylhexosamines were completely separated from each other. The numbers of theoretical plates for aldoses and ZVacetylhexosamines under these conditions were 2000 and 800, respectively.
170
HONDA
AND SUZUKI
Rho
least 100 repetitions of monosaccharide analysis. Column life could be extended by regeneration, which was conveniently achieved by washing the column with dilute hydrochloric acid, followed by distilled water. Postcolumn labeling. 2Cyanoacetamide is a sensitive and noncorrosive reagent for the detection of aldoses and their derivatives. We have already applied the reactions with this reagent to the postcolumn labeling in highperformance ion-exchange chromatography of carbohydrate-borate complexes (l-7). In these analyses, monosaccharides could be monitored both photometrically and fluorometrically. In the present work, however, the reaction condition was slightly altered because the eluate contained a high concentration of acetonitrile. Optimization studies indicated that the use 0 36 l-k d0 of 0.50 M borate buffer having a pH value Elutlon tlm (mln) of 8.5 as the reagent buffer gave the highest FOG.2. Analysis of an equimolar mixture of aldoses response to all saccharides in photometric and hexosamines (as N-acetyl derivatives) derivable from detection. The other conditions need not be glycoproteins. Column: Shodex DC-613 (4-mm i.d., 25 cm length, H+ form); column temperature: 30°C; eluant: changed. On the basis of this result improved aqueous 92 v/v% acetonitrile; flow rate: 0.60 ml/min; conditions for postcolumn labeling were dedetection: absorption at 280 nm after postcolumn labeling vised, as described under Materials and with 2cyanoacetamide (for details see Materials and Methods. F’luorometric detection was not so Methods); sample scale: 5 nmol each. The abbreviations effective due to quenching with acetonitrile. are the same as those in Fig. 1. Sensitivity. Table 1 gives the sensitivity of Durability of the resin and column mainte- the present method to aldoses and N-acetylnance.The resin used in this work was stable, hexosamines. The lower limit of detection the elution profile being unchanged after at varied between 35 and 104 pmol at the TABLE 1 SENSITIVITY
AND REPRODUCIBILITY
Coefficient of variation (96)’ Saccharide
Lower limit of detection (pm00
Relative molar response
Linear range (nmol)
0.2 nmol” 4 nmol’
10 nmolb 40 nmol’
80 nmol” 800 nmol’
Rhamnose. Xylose Fucose Mannose Galactose IV-Acetylghtcosamine IV-Acet ylgalactosamine
35.3 40.0 41.7 94.3 104 415 388
0.80 0.78 0.97 0.69 I 0.84 1.00
0.1-80 0.1-80 0.1-80 0.2-200 0.2-200 4-800 4-800
1.6 1.8 2.1 3.3 3.5 4.0 4.9
0.5 0.6 1.0 1.4 2.3 3.9 3.7
2.0 1.6 2.3 2.6 2.6 3.6 3.3
DThe number of determinations was 10 in all cases. b For aldoses. ’ For N-acetylhexosamines.
CHROMATOGRAPHIC
ALDOSE,
HEXOSAMINE,
171
AND SIALIC ACID DETERMINATION
tein samples, but we recommend heating in 2 M TFA for 6 h at 100°C under a nitrogen atmosphere, on the basis of our time-course study of aldose liberation (e.g., Ref. (9)). Application to some commercial glycoprotein preparations demonstrated the accuracy of the present method including the hydrolysis process, as seen from Table 2, except that the mannose content of human transferrin was slightly lower than the reported value. The chromatograms are shown in Fig. 3. On the other hand, component hexosamines were quantitatively released by acid hydrolysis under more drastic conditions, i.e., heating in 4 M hydrochloric acid for 6 h at 100°C under a nitrogen atmosphere. However, hydrolysates must be re-N-acetylated, because the N-acetyl group was removed concurrently. Re-N-acetylation was conveniently achieved by treating hydrolysates with a mixture of acetic anhydride and a saturated aqueous solution of sodium bicarbonate. The recovery was 92% for both galactosamine and glucosamine at the lo-nmol level. All analytical data for hexosamines in the same glycoprotein samples as described for aldose analysis were in good agreement with the reported values, though the galactosamine content of bovine submaxillary mucin was rather lower than the reported value.
signal-to-noise ratio of 2. Table 1 also lists the relative molar responses of these saccharides, as referred to galactose. The values were in the range 0.7-l. Calibration curves. As seen from Table 1, linearity existed for 0.1-80 nmol of fastereluting aldoses (rhamnose, xylose, and fucase), but the linear range was rather higher (0.2-200 nmol) with slower-eluting aldoses (mannose and galactose) due to peak broadening. The linear range for N-acetylhexosamines was higher still (4-800 nmol) due to heavier tailing. Reproducibility. Table 1 indicates that the determination of faster-eluting aldoses was highly reproducible, the coefficient of variation (CV) being OS-2.3% over the whole range of linearity. The CV values for slowereiuting aldoses were slightly higher, especially at the lowest level. The reproducibility for N-acetylhexosamines was rather low as compared to that for aldoses, but was satisfactory for the analysis of monosaccharide composition. Application to the analysis of component aldoses and hexosamines in glycoproteins. Aldoses released by the acid hydrolysis of glycoproteins could be directly quantified by the present method. The conditions for hydrolysis may be varied according to glycopro-
TABLE 2 DETERMINATION OF ALLWSES AND HEXOSAMINES IN VARIOUS GLYCOPROTEINS Amount of monosaccharide (&me) Glycoprotein
Fucose
Bovine submaxillary mucin, type. II
9.53 (10, 10)
Fetal calf serum fetuin Hen’s egg albumin
0.27 -
Mannose
Galactose
Glucosamine
Galactosamine
2.07
15.2 (14, 10)
168 (170, IO)
69.2 (85, 10)
27.3 (30, II)
45.9 (46, Jl) 1.48
53
(56, 11)
6.74 (7. 11) -
24.0
(20, 12) Human serum transferrin
0.31
11.1
(18, 14 Note. The numbers in parentheses are reported values.
12.8
(1%14 8.62 (7.4, 11)
17.9
(20, II)
-
172
HONDA
,.0
20
400
AND SUZUKI
.
80
,.r.
100 Elutlon
0
20
40 0
80
1;0-
time (min)
FIG. 3. Analysis of the component aldoses and hexosamines (as N-acetyl derivatives) of commercial samples of (a) bovine submaxillary mucin, type II, (b) fetal calf serum fetuin, (c) hen’s egg albumin, and (d) human serum transferrin. The glycoprotein samples were hydrolyzed in 2 M TFA for 6 h (aldoses) and in 4 M hydrochloric acid for 6 h (hexosamines) on a boiling-water bath under a nitrogen atmosphere. The hydrolysates were directly (aldoses) or after being N-acetylated (hexosamines) analyzed by HPLC under the conditions described in Fig. 2. Each set of chromatograms was obtained from the same glycoprotein sample. The abbreviations are the same as those in Fig. I.
II. Analysis of Sialic Acids The C-3-C-4 bonds in sialic acids are readily cleaved by the action of N-acetylneuraminate pyruvate-lyase (EC 4.1.3.3) to give N-acylmannosamines and pyruvate. Since this enzymatic reaction is rapid and quantitative, sialic acids may be determined from the amounts of the resultant N-acylmannosamines ( 13). Resolution of N-acetyland N-glycolylmannosamines was complete under the conditions described for aldoses, as well as N-acetylglucosamine and N-acetylgalactosamine, as seen from Fig. 4. The sensitivities to N-acylmannosamines were ap proximately of the same magnitude as those of aldoses, and the relative molar response of the N-glycolyl derivative to the N-acetyl derivative was 1.02. Sialic acids are so labile in acids that they are gradually decomposed on heating. For this reason sialic acids in glycoconjugates are usually released with neuraminidase. The hydrolytic reaction is quantitative. One of the authors (Honda) and his colleagues have already established a one-pot procedure to
0
30 Elutlon
time
60 (mln)
90
FIG. 4. Separation of siahc acids (as N-acylmannosamines) in the presence of aldoses and hexosamines (as N-acetyl derivatives). NGNA = N-glycolylneuraminic acid, NANA = N-acetylneuraminic acid. Other abbreviations and analytical conditions are the same as those described in Fig. 1.
CHROMATOGRAPHIC
ALDOSE,
HEXOSAMINE,
ElUtlOn
tlm
173
AND SIALIC ACID DETERMINATION
(min)
FIG. 5. Analysis of the component sialic acids (as N-acylmannosamines) in commercial samples of (a) fetal calf serum fetuin, (b) human serum transfetrin, (c) bovine serum acid glycoprotein, and (d) bovine submaxillary mucin, type I. The glycoprotein samples were hydrolyzed and converted to N-acylmannosamines by the combined action of neuraminidase and N-acetylneuraminate pyruvate-lysase. The analytical conditions are the same as those described in Fig. 2. NGNA = N-glycolylneuraminic acid, NANA = N-acetylneuraminic acid.
transform bound sialic acids into N-acylmann&mines by combined use of neuraminidase (mucopolysaccharide N-acetylneuraminidase, EC 3.2.1.18) and N-acetylneuraminate pyruvate-lyase ( 13). Figure 5 shows the chromatograms of the products obtained by applying this procedure to some glycoproteins, and Table 3 summarizes the estimated values of sialic acid content. Both N-acetyl and N-glycolyl derivatives were observed in bovine serum acid glycoprotein and bovine submaxillary mucin, whereas only the ZV-acetyl derivative was found in fetal calf serum fetuin and human
transferrin. The peak at 68 min, observed for the product from bovine submaxillary mucin, type I, is presumably due to an N,Odiacetylneuraminic acid. The intense peak at 13 min from the same sample is ascribable to a contaminant ultraviolet-absorbing compound. The chromatogram (not shown) of the 2 M TFA and 4 M hydrochloric acid hydrolysates of this glycoprotein sample also gave intense peaks at the same elution time as that observed in Fig. 5d, although the other preparation (type II) of this glycoprotein gave no such large peaks in this region (Fig. 3b).
TABLE 3 DETERMINATION OF SIALIC ACIDS IN VARIOUS GLYCOPROTEINS Amount of sialic acid (&mg) Glycoprotein Fetal calf serum fetuin Human serum transferrin Bovine serum acid glycoprotein Bovine submaxillary mu&, type I
N-Acetylneuraminic 46.2 14.0 11.2 16.9
acid
&Glycolylneuraminic 0.0 0.0 11.6 12.5
acid
HONDA
174
AND SUZUKI
ACKNOWLEDGMENTS The authors thank Showa Denko for the generous git? of the Shodex resin. This work was supported by grants (58570946 and 58890012) from the Japanese Ministry of Education.
7. 8. 9.
REFERENCES 1. Honda, S., Takahashi, M., Kakehi, K., and Ganno, S. (1981) Anal. Biochem. 113, 130-138. 2. Honda, S., Takahashi, M., Nishimura, Y., Kakehi, K., and Ganno, S. (198 I) Anal. Biochem. 118, 162-167. 3. Honda, S., Konishi, T., Suzuki, S., Takahashi, M., Kakehi, K., and Ganno, S. ( 1983) Anal. Biochem. 134,483-488. 4. Honda, S., Konishi, T., Suzuki, S., and Kakehi, K. (1983) J. Chromatogr. 281, 340-344. 5. Honda, S. et al., unpublished results. 6. Honda, S., Suzuki, S., Takahashi, M., Kakehi, K.,
10. 11.
12. 13.
and Ganno, S. (1983) Anal. B&hem. 134, 3439. Honda, S., Takahashi, M., Shimada, S., Kakehi, K., and Ganno, S. (1983) Anal. B&hem. 128, 429437. Honda, S., Suzuki, S., and Kakehi, K. (1984) J. Chromatogr. 291, 3 17-325. Honda, S., Suzuki, S., Kishi, Y., and Kakehi, K. (1981) J. Chromatogr. Biomed. Appl. 226, 341350. Pigman, W. (1977) in The Glycoconjugates (Horowitz, M. I., and Pigman, W., eds.), Vol. 1, pp. 131-152, Academic Press, New York. Schwick, H. G., Heide, K., and Haupt, H. (1977) in The Glycoconjugates (Horowitz, M. I., and Pigman, W., eds.), Vol. 1, pp. 261-321, Academic Press, New York. Gottschalk, A. (1972) Glycoproteins, Their Composition, Structure and Function, Part B, pp. 73 1-76 1, Elsevier, Amsterdam/New York. Kakehi, K., Maeda, K., Teramae, M., Honda, S., and Takai, T. (1983) J. Chromatogr. Biomed. Appl. 272, l-8.
BIOCHEMISTRY
Common
142, 167-174 (1984)
Conditions for High-Performance Liquid Chromatographic Microdetermination of Aldoses, Hexosamines, and Sialic Acids in Glycoproteins SUSUMU HONDA' ANDSHIGEOSUZUKI
Faculty of Pharmaceutical Sciences, Kinki University, 3-4-l Kowakae, Higashi-Osaka, Japan Received April 19, 1984 All aldoses known to be present in annnal and plant glycoproteins were separated in partition mode on a column of a highly crosslinked cation-exchange resin (Shodex DC-613, H+ form) and sensitively monitored by photometric detection after postcolumn labeling with 2cyanoa&amide. Linearity for the sample amount ranging from 0.1 to 80 nmol was observed with high reproducibility. Hexosamines could be determined as their N-acetyl derivatives under the identical conditions with approximately one order of magnitude higher range of linearity. Sialic acid could be estimated as N-acylhexosamines produced by the action of N-acetylneuraminate pyruvate-lyase. The usefulness of these common conditions was demonstrated by analyzing component monosaccharides of some glycoprotein preparations. Q 1984kadcmic m inc. KEY WORDS: monosaccharide composition analysis of glycoproteins; high-performance liquid chromatography of aldoscs, hexosamines, and sialic acids; analytical application of Nacetylneuraminate pyruvate-lyase.
In structural studies of carbohydrate chains in glycoproteins, the analysis of monosaccharide composition is fundamentally important. The monosaccharide species hitherto reported to be present in glycoproteins include arabinose, xylose, galactose, glucose, mannose, fucose, rhamnose, ZV-acetylglucosamine, and iV-acetylgalactosamine, together with ZV-and O-acylated neuraminic acids. While the identification and determination of these saccharides have been achieved by various chromatographic techniques, high-performance liquid chromatography (HPLC) has proved most convenient because it is rapid and easily automated. There have been numerous papers on HPLC of monosaccharides, and we also have devised conditions for individual classes of monosaccharides, including aldoses (1,2), hexosamines (3,4), and sialic acids (5), as well as uranic acids (6) and alditols (7). However, there have been no general condi’ To whom correspondence should be addressed.
tions applicable to all three classes (aldoses, hexosamines, and sialic acids) of monosaccharides in glycoproteins. The present work has been undertaken to develop common conditions for these classes of monosaccharides. MATERIALS
AND METHODS
Chemicals. Special-grade (for HPLC, Cica brand) samples of 2-cyanoacetamide and acetonitrile were purchased from Kanto Chemicals (Nihonbashi, Chuo-ku, Tokyo). The samples of bovine submaxillary mucin (types I and II), fetal calf serum fetuin (type III), bovine serum acid glycoprotein (fraction IV), hen’s egg albumin (grade V), and human serum transferrin were obtained from Sigma Chemical Company (St. Louis, MO.). Clostridium perfringens neuraminidase (type IV) and Escherichia coli N-acetylneuraminate pyruvate-lyase were obtained from Sigma and Nakarai (Nijo-karasuma, Nakakyo-ku, Kyoto) Chemical Companies, respectively. 167
0003-2697184 $3.00 Copyrieht 0 1984 by Acafhic Press, Inc. All rights of reproduction in any form reserved
168
HONDA
AND SUZUKI
Other chemicals and standard carbohydrates were of the highest grade commercially available. Instrumentation. The following equipment was used: a Hitachi 638 high-performance liquid chromatograph for pumping eluant, a Rheodyne injector carrying a 20-~1 loop, a Riko CB-500-TF hot-water circulator for column temperature regulation, Atto SJ-2396 twin-piston pumps for postcolumn labeling, a Hitachi 034-0736 reaction bath, an Atto SF- 1205-A ultraviolet absorption monitor having an 8-411 quartz cell, and a built-in integrator and data processor. Shodex DC-613 resin (Hf form) was suspended in 2 M hydrochloric acid, and the slurry was heated for several hours at 8090°C. The resin was repeatedly washed with distilled water until the washing fluids became neutral and then packed in a stainless-steel column (4-mm i.d., 25-cm length) by the standard slurry method (pressure range, lOO150 kg./cm2). The column was first washed with distilled water, followed by aqueous acetonitrile with increasing acetonitrile concentrations at a flow rate of 0.3 ml/min. Finally it was equilibrated with aqueous 92 v/v% acetonitrile at 30°C at a flow rate of 0.60 ml/min. The eluate (how rate, 0.60 ml/min) was successively mixed with 0.50 M borate buffer (pH 8.5) and an aqueous 1% 2-cyanoacetamide solution at the same flow rate of 0.50 ml/min by using Y-shaped TPFP connectors, and the resultant effluent was passed through a TPFE reaction coil (0.5-mm i.d., 10-m length) set in an oven thermostated at 100 f 1‘C. After the reaction mixture was cooled by passing it through a l-m TPFE coil of the same bore size, the absorbance at 280 am was recorded as a function of elution time. Hydrolysis of glycoprotein samples. For the analysis of component aldoses, ca. 200 pg of a glycoprotein sample was dissolved in 200 ~1 of 2 M trifluoroacetic acid (TFA),’ and the solution was heated for 6 h on a boiling-water bath under a nitrogen atmo2 Abbreviation used: TFA, trifluoroacetic acid.
sphere. After the solution was cooled to room temperature, it was deionized by passing it through a column containing 1 ml each of Amberlite CG- 120 (H+ form) and CG-400 (CH&OOform), and the column was washed with water. The combined eluate and washing fluids were evaporated to dryness under reduced pressure, and the residue was dissolved in a small volume (30- 100 ~1) of aqueous 92 v/v% acetonitrile. For the analysis of hexosamines, ca. 200 pg of a glycoprotein sample was dissolved in 200 ~1 of 4 M hydrochloric acid and the solution was heated for 6 h on a boilingwater bath under a nitrogen atmosphere. Then the solution was evaporated to dryness, and the residue was treated with a mixture of 0.5 ml of a saturated aqueous solution of sodium bicarbonate and 0.1 ml of acetic anhydride overnight. The mixture was processed in a manner similar to that described for the TFA hydrolysate except for the use of fivefold volumes of Amberlite resins. For the analysis of sialic acids, ca. 200 pg of a glycoprotein sample was dissolved in 800 ~1 of 0.06 M phosphate buffer (pH 7.0) containing neuraminidase (0.5 U) and Nacetylneuraminate pyruvate-lyase (0.3 U), and the mixture was incubated for 1 h at 37’C. After the mixture was heated for 1 min at 100°C to inactivate the enzymes, it was processed in a manner similar to that described for the TFA hydrolysate. Analysis of the component monosaccharides of glycoproteins. A 20-~1 portion of an aqueous acetonitrile solution finally obtained after hydrolysis of a glycoprotein sample was injected into the HPLC column, and the monosaccharides were determined. The content of each monosaccharide was estimated from its peak response and that from the standard monosaccharide, obtained under identical conditions. RESULTS
AND
DISCUSSION
I. Analysis of Aldoses and Hexosamines Optimization of separation. Although component aldoses (1,2) and sialic acids (5)
CHROMATDGRAPHIC
ALDGSE, HEXOSAMINE,
in glycoproteins could be separated by anionexchange chromatography as borate complexes, they were difficult to analyze under isocratic conditions. Borate complexes of hexosamines and their N-acetyl derivatives were easily dissociated on columns giving extremely broad peaks (5), which were superimposed on the peaks of other saccharides. Cation-exchange chromatography was effective for separation of hexosamines, but it resolved neither aldoses nor sialic acids. Therefore, we examined separation by partition chromatography. Since silica-based stationary phases are not durable to repeated analyses, ion-exchange resins of porous polymer type were adopted as stationary phases in this study. Figure la shows the temperature
FIG. 1. Dependence of the separation of aldoses and N-acetylhexosamines derivable from glycoproteins on (a) column temperature and (b) solvent composition. Column: Shodex DC-613 (4-mm i.d., 25cm length, Hf form); column temperature: (a) 4, 15, 25, 30, and 40°C and (b) 30°C; eluant: (a) aqueous acetonitrile, 92 v/v%, and (b) 80, 85, 87.5, 90, and 92 v/v?&; flow rate: 0.60 ml/min; detection: absorption at 280 nm al& postcolumn labeling with 2-cyanoacetamide (for details see Materials and Methods). Rha = rhamnose, Xyl = xylose, Fuc = fucose, Ara = arabinose, Man = mannose, Glc = glucose, Gal = @lactose, GlcNAc = N-acetylglucosamine, GalNAc = N-acetylgalactosamine.
AND SIALIC ACID DETERMINATION
169
dependence of the separation of aldoses and N-acetylhexosamines derivable from glycoproteins, as obtained by using a column of Shodex DC-6 13, a highly crosslinked, sulfonated polystyrene resin which showed the best separation, with aqueous acetonitrile as eluant. Since the presence of metals as counterions caused resolution of anomers (8), due to ligand exchange making determination complicated, the column was used in the H+ form. It is indicated that all these saccharides were eluted faster as the column temperature rose, and that the resolution was sufficiently high at 30°C to resolve all pairs of aldoses except for those of fucose-arabinose and mannose-glucose. The R, value for the fucase-arabinose pair was gradually increased with elevation of column temperature to reach a maximum at 30°C which, however, did not exceed unity. The separation of the mannose-glucose pair was almost independent of column temperature. On the other hand, Fig. 1b shows the effect of solvent composition. Elution time was increased with increasing concentrations of acetonitrile, but the peaks of hexosamines were broadened with tailing and sensitivity fell remarkably when the acetonitrile concentration became higher than 92 v/v%. The relationship between resolution and acetonitrile concentration was similar except for the fucose-arabinose and glucose-galactose pairs. These pairs gave the highest R, values at ca. 88 v/v% acetonitrile, but this concentration was unfavorable for other pairs. Therefore, we used 92 v/v% acetonitrile at 30°C. A flow rate of 0.60 ml/min was the most appropriate. Figure 2 shows a chromatogram for an equimolar mixture of these saccharides, as obtained under these conditions. Although the resolution of the fucose-arabinose and mannose-glucose pairs was incomplete, all major aldoses (fucose, mannose, and galactose) and N-acetylhexosamines were completely separated from each other. The numbers of theoretical plates for aldoses and ZVacetylhexosamines under these conditions were 2000 and 800, respectively.
170
HONDA
AND SUZUKI
Rho
least 100 repetitions of monosaccharide analysis. Column life could be extended by regeneration, which was conveniently achieved by washing the column with dilute hydrochloric acid, followed by distilled water. Postcolumn labeling. 2Cyanoacetamide is a sensitive and noncorrosive reagent for the detection of aldoses and their derivatives. We have already applied the reactions with this reagent to the postcolumn labeling in highperformance ion-exchange chromatography of carbohydrate-borate complexes (l-7). In these analyses, monosaccharides could be monitored both photometrically and fluorometrically. In the present work, however, the reaction condition was slightly altered because the eluate contained a high concentration of acetonitrile. Optimization studies indicated that the use 0 36 l-k d0 of 0.50 M borate buffer having a pH value Elutlon tlm (mln) of 8.5 as the reagent buffer gave the highest FOG.2. Analysis of an equimolar mixture of aldoses response to all saccharides in photometric and hexosamines (as N-acetyl derivatives) derivable from detection. The other conditions need not be glycoproteins. Column: Shodex DC-613 (4-mm i.d., 25 cm length, H+ form); column temperature: 30°C; eluant: changed. On the basis of this result improved aqueous 92 v/v% acetonitrile; flow rate: 0.60 ml/min; conditions for postcolumn labeling were dedetection: absorption at 280 nm after postcolumn labeling vised, as described under Materials and with 2cyanoacetamide (for details see Materials and Methods. F’luorometric detection was not so Methods); sample scale: 5 nmol each. The abbreviations effective due to quenching with acetonitrile. are the same as those in Fig. 1. Sensitivity. Table 1 gives the sensitivity of Durability of the resin and column mainte- the present method to aldoses and N-acetylnance.The resin used in this work was stable, hexosamines. The lower limit of detection the elution profile being unchanged after at varied between 35 and 104 pmol at the TABLE 1 SENSITIVITY
AND REPRODUCIBILITY
Coefficient of variation (96)’ Saccharide
Lower limit of detection (pm00
Relative molar response
Linear range (nmol)
0.2 nmol” 4 nmol’
10 nmolb 40 nmol’
80 nmol” 800 nmol’
Rhamnose. Xylose Fucose Mannose Galactose IV-Acetylghtcosamine IV-Acet ylgalactosamine
35.3 40.0 41.7 94.3 104 415 388
0.80 0.78 0.97 0.69 I 0.84 1.00
0.1-80 0.1-80 0.1-80 0.2-200 0.2-200 4-800 4-800
1.6 1.8 2.1 3.3 3.5 4.0 4.9
0.5 0.6 1.0 1.4 2.3 3.9 3.7
2.0 1.6 2.3 2.6 2.6 3.6 3.3
DThe number of determinations was 10 in all cases. b For aldoses. ’ For N-acetylhexosamines.
CHROMATOGRAPHIC
ALDOSE,
HEXOSAMINE,
171
AND SIALIC ACID DETERMINATION
tein samples, but we recommend heating in 2 M TFA for 6 h at 100°C under a nitrogen atmosphere, on the basis of our time-course study of aldose liberation (e.g., Ref. (9)). Application to some commercial glycoprotein preparations demonstrated the accuracy of the present method including the hydrolysis process, as seen from Table 2, except that the mannose content of human transferrin was slightly lower than the reported value. The chromatograms are shown in Fig. 3. On the other hand, component hexosamines were quantitatively released by acid hydrolysis under more drastic conditions, i.e., heating in 4 M hydrochloric acid for 6 h at 100°C under a nitrogen atmosphere. However, hydrolysates must be re-N-acetylated, because the N-acetyl group was removed concurrently. Re-N-acetylation was conveniently achieved by treating hydrolysates with a mixture of acetic anhydride and a saturated aqueous solution of sodium bicarbonate. The recovery was 92% for both galactosamine and glucosamine at the lo-nmol level. All analytical data for hexosamines in the same glycoprotein samples as described for aldose analysis were in good agreement with the reported values, though the galactosamine content of bovine submaxillary mucin was rather lower than the reported value.
signal-to-noise ratio of 2. Table 1 also lists the relative molar responses of these saccharides, as referred to galactose. The values were in the range 0.7-l. Calibration curves. As seen from Table 1, linearity existed for 0.1-80 nmol of fastereluting aldoses (rhamnose, xylose, and fucase), but the linear range was rather higher (0.2-200 nmol) with slower-eluting aldoses (mannose and galactose) due to peak broadening. The linear range for N-acetylhexosamines was higher still (4-800 nmol) due to heavier tailing. Reproducibility. Table 1 indicates that the determination of faster-eluting aldoses was highly reproducible, the coefficient of variation (CV) being OS-2.3% over the whole range of linearity. The CV values for slowereiuting aldoses were slightly higher, especially at the lowest level. The reproducibility for N-acetylhexosamines was rather low as compared to that for aldoses, but was satisfactory for the analysis of monosaccharide composition. Application to the analysis of component aldoses and hexosamines in glycoproteins. Aldoses released by the acid hydrolysis of glycoproteins could be directly quantified by the present method. The conditions for hydrolysis may be varied according to glycopro-
TABLE 2 DETERMINATION OF ALLWSES AND HEXOSAMINES IN VARIOUS GLYCOPROTEINS Amount of monosaccharide (&me) Glycoprotein
Fucose
Bovine submaxillary mucin, type. II
9.53 (10, 10)
Fetal calf serum fetuin Hen’s egg albumin
0.27 -
Mannose
Galactose
Glucosamine
Galactosamine
2.07
15.2 (14, 10)
168 (170, IO)
69.2 (85, 10)
27.3 (30, II)
45.9 (46, Jl) 1.48
53
(56, 11)
6.74 (7. 11) -
24.0
(20, 12) Human serum transferrin
0.31
11.1
(18, 14 Note. The numbers in parentheses are reported values.
12.8
(1%14 8.62 (7.4, 11)
17.9
(20, II)
-
172
HONDA
,.0
20
400
AND SUZUKI
.
80
,.r.
100 Elutlon
0
20
40 0
80
1;0-
time (min)
FIG. 3. Analysis of the component aldoses and hexosamines (as N-acetyl derivatives) of commercial samples of (a) bovine submaxillary mucin, type II, (b) fetal calf serum fetuin, (c) hen’s egg albumin, and (d) human serum transferrin. The glycoprotein samples were hydrolyzed in 2 M TFA for 6 h (aldoses) and in 4 M hydrochloric acid for 6 h (hexosamines) on a boiling-water bath under a nitrogen atmosphere. The hydrolysates were directly (aldoses) or after being N-acetylated (hexosamines) analyzed by HPLC under the conditions described in Fig. 2. Each set of chromatograms was obtained from the same glycoprotein sample. The abbreviations are the same as those in Fig. I.
II. Analysis of Sialic Acids The C-3-C-4 bonds in sialic acids are readily cleaved by the action of N-acetylneuraminate pyruvate-lyase (EC 4.1.3.3) to give N-acylmannosamines and pyruvate. Since this enzymatic reaction is rapid and quantitative, sialic acids may be determined from the amounts of the resultant N-acylmannosamines ( 13). Resolution of N-acetyland N-glycolylmannosamines was complete under the conditions described for aldoses, as well as N-acetylglucosamine and N-acetylgalactosamine, as seen from Fig. 4. The sensitivities to N-acylmannosamines were ap proximately of the same magnitude as those of aldoses, and the relative molar response of the N-glycolyl derivative to the N-acetyl derivative was 1.02. Sialic acids are so labile in acids that they are gradually decomposed on heating. For this reason sialic acids in glycoconjugates are usually released with neuraminidase. The hydrolytic reaction is quantitative. One of the authors (Honda) and his colleagues have already established a one-pot procedure to
0
30 Elutlon
time
60 (mln)
90
FIG. 4. Separation of siahc acids (as N-acylmannosamines) in the presence of aldoses and hexosamines (as N-acetyl derivatives). NGNA = N-glycolylneuraminic acid, NANA = N-acetylneuraminic acid. Other abbreviations and analytical conditions are the same as those described in Fig. 1.
CHROMATOGRAPHIC
ALDOSE,
HEXOSAMINE,
ElUtlOn
tlm
173
AND SIALIC ACID DETERMINATION
(min)
FIG. 5. Analysis of the component sialic acids (as N-acylmannosamines) in commercial samples of (a) fetal calf serum fetuin, (b) human serum transfetrin, (c) bovine serum acid glycoprotein, and (d) bovine submaxillary mucin, type I. The glycoprotein samples were hydrolyzed and converted to N-acylmannosamines by the combined action of neuraminidase and N-acetylneuraminate pyruvate-lysase. The analytical conditions are the same as those described in Fig. 2. NGNA = N-glycolylneuraminic acid, NANA = N-acetylneuraminic acid.
transform bound sialic acids into N-acylmann&mines by combined use of neuraminidase (mucopolysaccharide N-acetylneuraminidase, EC 3.2.1.18) and N-acetylneuraminate pyruvate-lyase ( 13). Figure 5 shows the chromatograms of the products obtained by applying this procedure to some glycoproteins, and Table 3 summarizes the estimated values of sialic acid content. Both N-acetyl and N-glycolyl derivatives were observed in bovine serum acid glycoprotein and bovine submaxillary mucin, whereas only the ZV-acetyl derivative was found in fetal calf serum fetuin and human
transferrin. The peak at 68 min, observed for the product from bovine submaxillary mucin, type I, is presumably due to an N,Odiacetylneuraminic acid. The intense peak at 13 min from the same sample is ascribable to a contaminant ultraviolet-absorbing compound. The chromatogram (not shown) of the 2 M TFA and 4 M hydrochloric acid hydrolysates of this glycoprotein sample also gave intense peaks at the same elution time as that observed in Fig. 5d, although the other preparation (type II) of this glycoprotein gave no such large peaks in this region (Fig. 3b).
TABLE 3 DETERMINATION OF SIALIC ACIDS IN VARIOUS GLYCOPROTEINS Amount of sialic acid (&mg) Glycoprotein Fetal calf serum fetuin Human serum transferrin Bovine serum acid glycoprotein Bovine submaxillary mu&, type I
N-Acetylneuraminic 46.2 14.0 11.2 16.9
acid
&Glycolylneuraminic 0.0 0.0 11.6 12.5
acid
HONDA
174
AND SUZUKI
ACKNOWLEDGMENTS The authors thank Showa Denko for the generous git? of the Shodex resin. This work was supported by grants (58570946 and 58890012) from the Japanese Ministry of Education.
7. 8. 9.
REFERENCES 1. Honda, S., Takahashi, M., Kakehi, K., and Ganno, S. (1981) Anal. Biochem. 113, 130-138. 2. Honda, S., Takahashi, M., Nishimura, Y., Kakehi, K., and Ganno, S. (198 I) Anal. Biochem. 118, 162-167. 3. Honda, S., Konishi, T., Suzuki, S., Takahashi, M., Kakehi, K., and Ganno, S. ( 1983) Anal. Biochem. 134,483-488. 4. Honda, S., Konishi, T., Suzuki, S., and Kakehi, K. (1983) J. Chromatogr. 281, 340-344. 5. Honda, S. et al., unpublished results. 6. Honda, S., Suzuki, S., Takahashi, M., Kakehi, K.,
10. 11.
12. 13.
and Ganno, S. (1983) Anal. B&hem. 134, 3439. Honda, S., Takahashi, M., Shimada, S., Kakehi, K., and Ganno, S. (1983) Anal. B&hem. 128, 429437. Honda, S., Suzuki, S., and Kakehi, K. (1984) J. Chromatogr. 291, 3 17-325. Honda, S., Suzuki, S., Kishi, Y., and Kakehi, K. (1981) J. Chromatogr. Biomed. Appl. 226, 341350. Pigman, W. (1977) in The Glycoconjugates (Horowitz, M. I., and Pigman, W., eds.), Vol. 1, pp. 131-152, Academic Press, New York. Schwick, H. G., Heide, K., and Haupt, H. (1977) in The Glycoconjugates (Horowitz, M. I., and Pigman, W., eds.), Vol. 1, pp. 261-321, Academic Press, New York. Gottschalk, A. (1972) Glycoproteins, Their Composition, Structure and Function, Part B, pp. 73 1-76 1, Elsevier, Amsterdam/New York. Kakehi, K., Maeda, K., Teramae, M., Honda, S., and Takai, T. (1983) J. Chromatogr. Biomed. Appl. 272, l-8.