TY - JOUR
AU - Leroy,, Yves
AB - Abstract We have developed a method involving the formation of heptafluorobutyrate derivatives of O-methyl-glycosides liberated from glycoproteins and glycolipids following methanolysis. The stable derivatives of the most common monosaccharides of these glycoconjugates (Ara, Rha, Xyl, Fuc, Gal, Man, Glc, GlcNAc, GalNAc, Neu5Ac, KDN) can be separated and quantitatively and reproducibly determined with a high degree of sensitivity level (down to 25 pmol) in the presence of lysine as an internal standard. The GlcNAc residue bound to Asn in N-glycans is quantitatively recovered as two peaks. The latter were easily distinguished from the other GlcNAc residues of N-glycans, thus allowing a considerable improvement of the data on structure of N-glycans obtained from a single carbohydrate analysis. The most common contaminants present in buffers commonly used for the isolation of soluble or membrane-bound glycoproteins (SDS, Triton X-100, DOC, TRIS, glycine, and polyacrylamide or salts, as well as monosaccharide constituents of proteoglycans or degradation products of nucleic acids) do not interfere with these determinations. A carbohydrate analysis of glycoproteins isolated from a SDS/PAGE gel or from PDVF membranes can be performed on microgram amounts without significant interferences. Since fatty acid methyl esters and sphingosine derivatives are separated from the monosaccharide peaks, the complete composition of gangliosides can be achieved in a single step starting from less than 1 µg of the initial compound purified by preparative Silicagel TLC. Using electron impact ionization mass spectrometry, reporter ions for the different classes of O-methyl-glycosides (pentoses, deoxyhexoses, hexoses, hexosamines, uronic acids, sialic acid, and KDN) allow the identification of these compounds in very complex mixtures. The mass of each compound can be determined in the chemical ionization mode and detection of positive or negative ions. This method presents a considerable improvement compared to those using TMS derivatives. Indeed the heptafluorobutyrate derivatives are stable, and acylation of amino groups is complete. Moreover, there is no interference with contaminants and the separation between fatty acid methyl-esters and O-methyl glycosides is achieved. GLC, GC, MS, mass spectrometry, monosaccharide Introduction The interest over the carbohydrate moieties of glycoconjugates has been increasing because of their postulated functions in cell adhesion and recognition mechanisms and their importance in signal transduction and as markers of differentiation and carcinogenesis. The determination of the monosaccharide composition is a key step in choosing the strategy for glycan analysis. The most common method is the capillary column GLC analysis of the trimethylsilyl derivatives (TMS) obtained from O-methyl-glycosides after methanolysis induced cleavage of glycosidic bonds (for review, see Chaplin, 1994). This cleavage procedure, when used in controlled conditions (and especially the use of anhydrous methanol/HCl reagent; Zanetta et al., 1972), allows the almost quantitative liberation of the glycoprotein and glycolipid monosaccharides as stable O-methyl-glycosides. However, an intermediate step of re-acetylation of these compounds is necessary before the formation of the TMS derivatives in the classical procedure (for review, see Chaplin, 1994). Unfortunately, this re-acetylation step is time consuming, and the reaction yield is particularly difficult to control. As a result, the molar ratio of N-acetyl-hexosamines and sialic acid fluctuates from an experiment to the other, rendering monosaccharide molar composition of a glycan often problematic. One way to circumvent such a difficulty is the formation of volatile derivatives of O-methyl-glycosides using a strong acylating agent (acid anhydride). In a previous paper (Zanetta et al., 1972), the most common constituents of glycoproteins and glycolipids (and all constituents of glycolipids) could be determined quantitatively as trifluoroacetate derivatives (TFA) using a separation by classical GLC. However, this method lacked the sensitivity required for applications to small amounts of material (less than 100 µg of glycoprotein). The reason was that the use of capillary columns was prevented by the presence of an excess of trifluoroacetic anhydride which destroyed the liquid phase. Because of the volatile nature of the TFA derivatives, the excess of anhydride could not be eliminated without the complete loss of the most volatile compounds. We hypothesized that one way to solve this problem was to produce less volatile derivatives using an acylation with anhydrides of higher mass (pentafluoropropionic or heptafluorobutyric anhydrides). This article demonstrates that quantitative monosaccharide analyses of glycoproteins and glycolipids can be achieved on capillary columns using the heptafluorobutyrate derivatives of the O-methyl-glycosides. Results and discussion Choice of the acylating agent Standard monosaccharide mixtures were submitted to methanolysis (see Materials and methods), followed by acylation with pentafluoropropionic or heptafluorobutyric anhydrides, respectively. The derivatives were injected into the SE-30 column, and the area of the different peaks was quantified. These areas were compared to those of the same compounds injected after the complete evaporation of the acylation mixture under a stream of nitrogen and resolubilization in dried acetonitrile. Pentafluoropropionate derivatives of fucose, arabinose, and xylose and to a lesser extent the hexose derivatives showed a significant decrease during evaporation (a maximum of 20% loss for the most volatile compounds). In contrast, identical areas were found for all heptafluorobutyrate derivatives, even working with less than 1 nmol of each monosaccharide in the reaction vessel. This adequate volatility was verified after injection on capillary columns, the surface of the peaks being constant after evaporation of the solvent in the Ross injector. Conditions of acylation The acylation reaction with HFBAA was performed in acetonitrile because this solvent was able to solubilize the O-methyl-glycosides formed during methanolysis and because it formed a single phase with HBFAA (in contrast with less polar solvents such as dichloromethane, chloroform, heptane, methyl or ethyl acetate). Kinetics of acylation of the different O-methyl-glycosides were followed either at room temperature or at 100°C or 150°C, analysing the areas of the different peaks of standard mixtures. When experiments were performed at room temperature, the acylation was complete only after 24 h, the acylation being slower for monosaccharides possessing amino groups (hexosamines and sialic acid). When acylation was performed at 100°C, the acylation was maximal after 30 min; the relative molar responses did not vary for longer times of heating. At 150°C, the acylation was complete within 5 min (this procedure being recommended since the reaction takes place as a reflux, also acylating the material remaining on the vessel wall). Because acetonitrile showed some trailing on classical columns, we tried to solubilize the evaporated HFB derivatives in other solvents, i.e., heptane or dichloromethane. Heptane provoked an important loss of most compounds (especially hexosamines and sialic acid, but also pentoses and hexoses) at all initial concentrations of monosaccharides because HFB derivatives were not soluble. A significant loss of the same compounds was observed with dichloromethane when the initial concentration of the monosaccharide was more than 5 nmol in the vessels. In contrast, acetonitrile completely dissolved the different compounds and appeared suitable for capillary column analysis. Stability of the heptafluorobutyrate derivatives After acylation, the HFB-derivatives kept at room temperature in the closed reaction vial were analyzed at different periods of time. No variations were observed during several months, as already observed for the trifluoroacetate derivatives (Zanetta et al., 1972). The same type of experiment was also performed on the HFB derivatives stored for different periods of times in dried acetonitrile after the complete evaporation of the acylation mixture. No variations were observed when samples were stored for 2 days in the closed vial with the exception of the derivatives corresponding to the GlcNAc residue involved in the N-glycosidic bond (see below). Repetitive (up to 10 times) opening of the vials for time to time injections did not modify significantly the relative molar responses of the different compounds with the exception of the previous derivatives and, specifically, the derivatives of the furanic forms of all compounds (Fuc, Gal, GalNAc, GlcA, and GalA) and the derivatives of sialic acid and KDN. However, the initial response of these compounds could be restored, adding 25 µl of HFBAA and heating for 5 min at 100°C. This indicated that the HFB derivatives presented a strong stability in neutral or acidic conditions. Because of this stability, the cleaning of the reaction vessels before reuse should first involve an immersion in an alkaline solution (1 M NaOH for 10 min). Tubes were cleaned thereafter by heating for 1 h at 80°C in 50% sulfuric acid. Choice of GLC separations From previous observations (Zanetta et al., 1972, 1973), it appeared that fluorinated compounds strongly interact with fluorosilicone liquid phases and are poorly adsorbed on classical silicone phases, in such a way that they are eluted from the columns at temperature lower than their boiling points. In contrast, compounds rich in methylene groups strongly interact with classical silicone phases. In fact, despite their high mass, HFB of O-methyl-glycosides were eluted before fatty acid methyl-esters (FAMEs). The chosen chromatographic conditions were: injector and detector temperature 260 °C; helium carrier gas 0.8 bar; temperature program: 100°C to 140°C at 1.2°C/min, followed by 4°C/min until 240°C (when samples contained only monosaccharide derivatives, the temperature program was stopped after the elution of the neuraminic acid peak). Using these conditions, almost all HFB of the isomers of O-methyl-glycosides (Xyl, Ara, Rha, Fuc, Gal, Man, Glc, GlcNAc, GalNAc, NeuAc, and KDN) were separated from each other (Figure 1, Table I). The single overlapping was observed for the -mannose isomer and a furanic form of GalNAc. Furthermore, FAMEs were eluted at higher retention time than the derivatives of NeuAc. The quality of the separations remained stable during at least six months of intense use (about 600 injections), indicating that the derivatives did not destroyed the columns. However, when HFB derivatives are chromatographed on a column on which other derivatives were analyzed (acetates or TMS derivatives), the separation of the second isomer of Glc and the third of Gal are only partially achieved. This partial overlapping disappeared after two to three runs of HFB derivatives. Fig. 1. Open in new tabDownload slide Gas chromatography of the heptafluorobutyrates of O-methyl glycosides on a capillary column. (a) GC chromatogram of a mixture of standard monosaccharides (Fuc, Gal, Man, Glc, GlcNAc, GalNAc, Neu5Ac) and fatty acids (C16, C18, C20, C22 as major compounds) submitted to methanolysis then acylation by HFBAA. Note that all constituents of N-glycans are separated the one from the other and completely separated from FAME. The β-isomer of Glc is only partially resolved from a galactose isomer due to the injection of HFB derivatives immediately after a chromatography of TMS derivatives. Inset: separation of the different isomers of O-methyl glycosides of pentoses and deoxy-hexoses (Ara, Rha, Xyl, and Rib). The lines inside the chromatogram represent the temperature program. (b) GC chromatogram of glycoproteins found in wasp's net. Note the complete separation of Ara, Rha, Fuc, the presence of two peaks (GN1) corresponding to the first GlcNAc residue of N-glycans. This chromatogram revealed the presence of small amounts of glucuronic acid (peak in-between the major peaks of Gal and Man, and the two peaks migrating between the two first peaks of GalNAc; Rt = 25.43, 31.95, and 32.39 min, respectively). (c) GC chromatogram of the mucins of the eggs of Pleurodeles walti. This materiel contained high amounts of KDN, perfectly separated from all other compounds. The carbohydrate composition deduced from this analysis performed starting from 10 mg of mucin was identical (less than 1% variation) to a previous analysis performed on 1 mg mucin using the TFA derivatives (Zanetta et al., 1972). Because these material were injected onto a column perturbed by the injection of acetylated compounds, the separation of the Glc and Gal isomers was not optimal. Fig. 1. Open in new tabDownload slide Gas chromatography of the heptafluorobutyrates of O-methyl glycosides on a capillary column. (a) GC chromatogram of a mixture of standard monosaccharides (Fuc, Gal, Man, Glc, GlcNAc, GalNAc, Neu5Ac) and fatty acids (C16, C18, C20, C22 as major compounds) submitted to methanolysis then acylation by HFBAA. Note that all constituents of N-glycans are separated the one from the other and completely separated from FAME. The β-isomer of Glc is only partially resolved from a galactose isomer due to the injection of HFB derivatives immediately after a chromatography of TMS derivatives. Inset: separation of the different isomers of O-methyl glycosides of pentoses and deoxy-hexoses (Ara, Rha, Xyl, and Rib). The lines inside the chromatogram represent the temperature program. (b) GC chromatogram of glycoproteins found in wasp's net. Note the complete separation of Ara, Rha, Fuc, the presence of two peaks (GN1) corresponding to the first GlcNAc residue of N-glycans. This chromatogram revealed the presence of small amounts of glucuronic acid (peak in-between the major peaks of Gal and Man, and the two peaks migrating between the two first peaks of GalNAc; Rt = 25.43, 31.95, and 32.39 min, respectively). (c) GC chromatogram of the mucins of the eggs of Pleurodeles walti. This materiel contained high amounts of KDN, perfectly separated from all other compounds. The carbohydrate composition deduced from this analysis performed starting from 10 mg of mucin was identical (less than 1% variation) to a previous analysis performed on 1 mg mucin using the TFA derivatives (Zanetta et al., 1972). Because these material were injected onto a column perturbed by the injection of acetylated compounds, the separation of the Glc and Gal isomers was not optimal. Open in new tabDownload slide Open in new tabDownload slide Choice of an appropriate internal standard Classical techniques used polyols as internal standard (such as meso-inositol or mannitol). In fact mannitol could be released by reductive -elimination procedure from glycans with an O-linked mannose residue (Chiba et al., 1997), whereas meso-inositol is an essential constituent of GPI-anchors. Moreover, in the present system, meso-inositol showed trailing and interfered with the second Gal isomer. Because of these possible interferences, we preferred amino acids, the carboxyl group being methyl-esterified during methanolysis and the amino group acylated with HFBAA. Throughout the different amino acids presenting stable heptafluorobutyrate derivatives (Zanetta and Vincendon, 1973), lysine (Rt = 38.65) showed a chromatographic behavior compatible with the separation of the classical monosaccharides of glycoproteins and of glycolipids and did not interfere with any monosaccharides. Because of the presence of two minor impurities in commercially available lysine, it was necessary to recrystallize it from water. Subsequent studies of the possible contaminations of glycolipid and glycoprotein samples (see below) indicated that lysine was a suitable internal standard. Determinations of the relative molar responses The relative molar responses (RMR) of the different HFB derivatives were first determined on equimolar mixtures of different monosaccharides submitted to methanolysis in the standard conditions, followed by acylation with HFBAA and analyzed on the two types of GLC columns. Because of the presence of significant amounts of impurities in commercially available standards, these RMR were only approximate. Corrections due to the presence of the impurities eliminated some divergences observed between the different compounds of the same mass. These RMR were compared to those obtained from the analysis of oligosaccharides, reduced oligosaccharides or glycoasparagines, the structure and purity has been previously determined by methylation and NMR analysis. The final data from these determinations are presented in Table II: (1) taking into accounts all isomers of the different monosaccharides (RMR); (2) taking into account the major isomer (RMRmp), since the proportions of the different isomers of each monosaccharides were constant (with the exception of the derivative of sialic acid and that of galactose). Therefore, the quantitation of an analysis could be performed into two steps: (1) verification that the proportions of the different isomers are correct (a point that indicated the absence of contaminant in the major peaks); (2) calculation of the area of each monosaccharide based on the RMR of the major peak (RMRmp; Table II). The late form of calculation could not be applied systematically to the NeuNAc derivatives because theses compound are eluted, depending on the sample, either as a single peak or as two peaks. In the latter case, the major anomer represented always 88.17%, a point that remains to be explained. Similarly, the integration of the two peaks corresponding to the derivatives of the major product issued from the GlcNAc residue involved in the N-glycosidic bond, i.e., glucosamine (Maes et al., 1999), was needed, since these proportions were varying with time, due to a slow change in the proportions of the α and β isomers from a methanol/HCl to an acetonitrile medium. The case of the response of Gal will be discussed below. Open in new tabDownload slide Open in new tabDownload slide The analysis of the different glycoasparagines and of glycoproteins of known monosaccharide compositions indicated a deficit of GlcNAc, when compared to the oligosaccharides with a single or two GlcNAc residues. However, two major extraneous peaks were detected for glycoproteins containing N-glycans at Rt = 20.17 and 22.29 min, respectively (see Figures 1c, 2). A third extraneous peak was observed for glycoasparagines (Rt = 11.09 min), that corresponded to the di-methyl-ester N-heptafluorobutyrate of aspartic acid by EI/MS (357) and CI/MS (M + 18 = 375). Since this compound was a little bit too volatile (Zanetta and Vincendon, 1973), it could not be exactly quantified using the evaporation needed for capillary column analysis. Without evaporation, and taking into account a RMR of 0.500 for this derivative of Asp, it could be calculated that it corresponded to one residue of Asp per glycoasparagine. This indicated that the N-glycosidic bond was almost quantitatively cleaved (Maes et al., 1999). The nature of the products liberated from GlcNAc-Asn by methanolysis will be reported in detail elsewhere (Maes et al., 1999). From the analysis of different N-glycoproteins and glycoasparagines, a RMR of 0.750 was assigned to the sum of these two compounds. Therefore, for the analysis of glycoproteins containing N-glycans, the ratio of the different monosaccharides was calculated relative to these constituents representative of the GlcNAc residue involved in the N-glycosidic bond. Open in new tabDownload slide Open in new tabDownload slide Applications to glycoprotein and glycolipids Complete composition of gangliosides. HFB derivatives of O-methyl-glycosides were completely separated from FAMEs and from the HFB derivatives of sphingosines on the capillary column. The C14:0 FAME was eluted later than the NeuAc HFB derivative. The different FAME were fully separated from each and are eluted before 240°C (Figure 1a). Because the derivatives of sphingosines presented a high mass, they were eluted in between fatty acids (Figure 2). Therefore, it was possible to determine simultaneously the carbohydrate, sphingosine and fatty acid composition of gangliosides without interference with each other. The compositions of the GM1 and G5b gangliosides are given in Table III. The determination of the nature of the fatty acid methyl esters was obtained by comparison with standard FAMEs and GC/MS analysis. The nature of the heptafluorobutyrate derivatives of sphingosines was given by GC/MS analysis (Figure 2). Using EI ionization, the different bases gave characteristic fragments at M = 459 for the C18:1, M = 487 for the C20:1, M = 489 for the di-hydro C20:0 and M = 543 for the C24:1 derivatives of the sphingosine bases. Additional information was obtained by chemical ionization and detection of the positive ions with major fragments corresponding to [M + 18 - 214]+ of M = 691, M = 719, M = 721, and M = 775 for the previous compounds. It should be noted that in G5b, two different C18:1 sphingosines and three C20:1 sphingosines of the same mass are observed (Figure 2), corresponding to a partial degradation during the methanolysis step (Pons et al., unpublished data). The elution of the C:24 derivative implied a prolongation of the temperature up to 260°C. These determinations were in perfect agreement with the data obtained by the MALDI-TOF analysis of these compounds. Attempts were also performed using the detection of negative ions formed during chemical ionization, a detection that increased the sensitivity of detection of heptaflurobutyrate derivatives but did not allow the detection of FAME. The ions obtained using this detection system for the HFB derivatives of O-methyl glycosides and of sphingosines were either [M]_- or [M - 20]_-, the loss of mass of 20 corresponding to the loss of HF. (Figure 3) Fig. 2. Open in new tabDownload slide Identification of FAMEs ([M+ NH4]+) and sphingosine bases ([M − 214 + NH4]+) of G5b ganglioside by GC/MS in the chemical ionization mode (Gas chromatography was performed with a temperature gradient of 10°C/min from 90°C to 240°C. Note the presence of two peaks corresponding to the C18:1 sphingosine and three peaks corresponding to the C20:1 sphingosine. Derivatives of monosaccharides are eluted in the first part of the chromatogram (not shown). Fig. 2. Open in new tabDownload slide Identification of FAMEs ([M+ NH4]+) and sphingosine bases ([M − 214 + NH4]+) of G5b ganglioside by GC/MS in the chemical ionization mode (Gas chromatography was performed with a temperature gradient of 10°C/min from 90°C to 240°C. Note the presence of two peaks corresponding to the C18:1 sphingosine and three peaks corresponding to the C20:1 sphingosine. Derivatives of monosaccharides are eluted in the first part of the chromatogram (not shown). The RMR of FAMEs and sphingosines calculated from these experiments (Table IV) approached the theoretical RMR calculated from the number of carbon and hydrogen atoms of the molecules. It should be stressed that except for the quantity of glucose, the molar ratios for these two standard gangliosides were exact with a precision less than 1%. Monosaccharide composition of glycoproteins. The GLC profiles of the monosaccharides constituents of glycoproteins are shown in Figures 1b, 1c, 2 and 3, and examples of carbohydrate compositions are presented in Table V. Since the HFB derivative of the major compounds resulting from the glycosylamine of GlcNAc involved in the N-glycosidic bond (glucosamine; Maes et al., 1999) was also quantified, the calculation of the molar composition of N-glycans was performed relative to these compounds. A series of glycoasparagines isolated from the microsomal fraction of the adult rat brain were analyzed, starting from 1 µg amount of each glycan. The expression of the different results relative to the GlcNAc residue involved in the N-glycosidic bond allowed obtaining the composition of the different oligomannosides without ambiguities. But, because of the relative instability of the glucosamine derivative, the samples had to be analyzed within 48 h after the evaporation of the acylation mixture. As mentioned above, the peak corresponding to the di-methyl ester HFB of Asp, resulting from the Asn involved in the N-glycosidic bond of glycoasparagines, could not be determined quantitatively since this compound was partially lost during the evaporation of the acylation mixture and during the drying step in the Ross injector. Nevertheless, the presence of this peak was indicative that the initial compound was a glycoasparagine and not a glycan attached to a larger peptide. Open in new tabDownload slide Open in new tabDownload slide Open in new tabDownload slide Open in new tabDownload slide The method was applied to a series of reduced oligosaccharides with defined structures liberated by the reductive -elimination procedure. The molar ratios obtained for all these compounds approached the theoretical value at less than 1%, and sometimes less than 0.1% (Table V). Surprisingly, in analysis of O-glycans (reduced or not) in which the Gal residue is bound to GalNAc or GalNAc-OH, the first furanic isomer of Gal (Rt = 20.04) showed a significant relative increased compared to the other isomers, its area being approximately the double of that obtained for standard of galactose or for complex type N-glycans. This suggested that during the cleavage of this particular glycosidic bond, the formation of the O-methyl glycoside of this furanic form of Gal is favored, a point that remains to be explained. Analysis of the possible contaminants issued from glycoprotein isolation procedures. Since the methods of glycoprotein purification could introduce intrinsic contaminants, a special care was taken to this point. Consequently, deoxycholate, sodium dodecyl sulfate, Triton X-100, glycine, and Tris were submitted to the complete procedure of methanolysis/HFBAA and analyzed by GLC. A single peak corresponding to deoxycholate was eluted at the end of the temperature gradient. The peaks of the Tris and glycine derivatives were eluted close to the injection peak and were in large part lost during the evaporation of the acylating mixture because of their volatility. Triton X-100 is not significantly degraded under the methanolysis conditions and gave three small peaks without interferences with the monosaccharide derivatives. In contrast, SDS gave peaks corresponding to compounds with 10, 12, 14, and 16 carbon atoms, the major peak of the C:12 component interfering with the isomer of mannose. Although the relative proportions of the two Man isomers could not be determined, the molar ratio of Man could be determined based on the α-Man area (93.05%). Consequently, these compounds used for the purification of membrane-bound glycoproteins did not interfere significantly with the analysis. Fig. 3. Open in new tabDownload slide Chromatogram of the monosaccharide composition of delipized synaptosomal plasma membranes isolated from adult rat brain. Due to the small amount of GalNAc, it was suggested that O-glycans were minor glycans in these membranes. Note the presence of the peaks corresponding to the GlcNAc residue involved in the N-glycosidic bond (GN1). No traces of glucuronic acid were found in this material, suggesting that the HNK-1 (glucuronic acid 3-sulfate) epitope is not present on these purified plasma membranes. Although the membrane fraction was submitted to a drastic delipidation procedure, FAMEs were still detected in high amount, but interestingly, the FAME composition was not the same as that of the lipids extract from the same fraction (Figure 1a). This suggested that an important quantity of proteins or glycoproteins of these membranes are acylated by fatty acids. Note that after several injections, the separation between the Glc and Gal isomers is almost complete. Fig. 3. Open in new tabDownload slide Chromatogram of the monosaccharide composition of delipized synaptosomal plasma membranes isolated from adult rat brain. Due to the small amount of GalNAc, it was suggested that O-glycans were minor glycans in these membranes. Note the presence of the peaks corresponding to the GlcNAc residue involved in the N-glycosidic bond (GN1). No traces of glucuronic acid were found in this material, suggesting that the HNK-1 (glucuronic acid 3-sulfate) epitope is not present on these purified plasma membranes. Although the membrane fraction was submitted to a drastic delipidation procedure, FAMEs were still detected in high amount, but interestingly, the FAME composition was not the same as that of the lipids extract from the same fraction (Figure 1a). This suggested that an important quantity of proteins or glycoproteins of these membranes are acylated by fatty acids. Note that after several injections, the separation between the Glc and Gal isomers is almost complete. The sensitivity of the method theoretically allowed the analysis of very small amounts of glycoproteins isolated by electrophoresis. Therefore, it was of interest to examine if polyacrylamide could interfere with the monosaccharide determination. For this, a small piece of the polyacrylamide gel (made in Tris-glycine buffer containing 0.1% SDS) was submitted to methanolysis followed by acylation. The complete mixture gave only the peaks corresponding to SDS. In order to test the possibility of performing a carbohydrate composition directly on acrylamide gels, two glycoproteins (Thy-1 and CD24 isolated as concanavalin A-binding glycoproteins in young rat cerebellum) were isolated by SDS-PAGE. The gel was stained with Coomassie Brilliant Blue R (solvents were of analytical grade for this staining), the stained bands were dissected with a lancet and transferred to the methanolysis vessel, and the gels were dehydrated using repetitive washings in distilled methanol (six times during 10 min). The gels were submitted to methanolysis and after discarding the gels, the samples were evaporated under nitrogen then analyzed after acylation. Although the molar ratios of the different monosaccharides were identical to the values obtained for the same glycoproteins recovered from the gel by electroelution, the yield of the procedure was at least 10 times less than expected. This was due to the absence of penetration of the methanolysis reagent into the shrunken polyacrylamide gel (CBB still remained at the center of the gel after 20 h of methanolysis). Therefore, although the molar composition of the monosaccharides could be obtained directly on a polyacrylamide gel, the poor efficiency of the methanolysis procedure did not allow the determination of the carbohydrate content of a glycoprotein. This difficulty could be easily circumvented by blotting the glycoproteins on PDVF membranes. After extensive washing with water, the membrane was stained with Ponceau red, and then washed again with water. The stained band was cut, and repetitively washed with anhydrous methanol in the methanolysis vessel. After methanolysis, the filter was discarded (after washing with methanol) and the evaporated sample was acylated. In these conditions, the methanolysis procedure allows the quantitative liberation of Omethyl glycosides. No interference occurred if the PDVF membranes were repetitively washed in methanol. The technique could not be extended to nitrocellulose filters because of the complete solubilization of nitrocellulose in methanol and in acetonitrile. Because the carbohydrate analysis could be performed on crude glycoprotein preparation or tissue extracts, we also examined the possible contaminations induced by different carbohydrate containing compounds. Consequently, RNA, DNA, and a whole mixture of proteoglycans isolated from adult rat brain in the absence of detergent (Normand et al., 1988), were submitted to methanolysis in standard conditions followed by derivatization with HFBAA. Only ribose isomers interfered with the determination of fucose (Table I, Figure 1), but the presence of ribose in a sample could be easily detected by the inversion of the proportions of the two fuco-pyranose peaks. The small amount of derivatives of uronic acids liberated from proteoglycans during methanolysis (Table I) did not interfere with the separation of other constituents (Table I). In contrast, a significant amount of these compounds was liberated from branched glycans, in such a way that the presence of specific peaks was indicative of the presence of uronic acids in the glycans (Figure 1b). DNA methanolysis products resulted in a black suspension upon acylation, but again, none of the peaks interfered with the determination of the classical glycoprotein monosaccharides. Sensitivity of the technique and nanoanalysis The method allowed the quantitative determination of the carbohydrate composition of glycoproteins and glycolipids at the level of the sensitivity of the FID detection, i.e., at 25-50 pmol of each monosaccharide injected onto the GLC column, for optimal quantitations. In principle, since the totality of the sample can be applied to the GLC column, analyses could performed on samples containing such amounts of material. However, at this level, dramatic cautions had to be taken with the purity of the reagents and with the cleaning of the reaction vials, even with new vessels. In our conditions, new vials contained a significant amount of impurities that formed volatile derivatives eluted in the area of FAME, whereas xylose and glucose were detected systematically. Special sequentially redistilled reagents, absence of dust in the laboratory, and special cleaning of the vessels are needed to perform quantitative determinations starting from picomoles of each glycoprotein or glycolipid monosaccharide. Coupling with mass spectrometry Although the reproducibility of the GC separation allowed to assign the nature of the compound according to the retention times of the isomers (all isomers of Ara, Rha, Rib, Xyl, Fuc, Gal, Man, Glc, GlcNAc, GalNAc, KDN, Neu5Ac are separated, and separated from glucitol, mannitol, inositol, N-acetyl-glucosaminitol, N-acetyl-glucosaminitol, and uronic acids; from the fatty acid methyl esters larger than C14:0; from sphingosine bases; and from major contaminants found during glycoprotein and glycolipid isolation), the absolute identification of each compounds in samples with an unknown composition needs mass spectrometry. The HFB derivatives of the O-methyl glycosides of monosaccharides presented a relatively high mass (978 for hexoses, 977 for hexosamines, 1275 for sialic acid). Chemical ionization (NH3 gas) MS analysis of the compounds performed with detection of positive ions, giving the pseudo-molecular ion [M + NH4]+, was a very secure identification of the families of compounds (Table VI), provided that the mass limit of the apparatus was 2000 amu. Unfortunately, most of the routine GC/MS systems are out of this range. Open in new tabDownload slide Open in new tabDownload slide Fig. 4. Open in new tabDownload slide Reporter ion analysis in the EI mode of a mixture of pentoses, deoxy-hexoses, hexoses, uronic acids, N-acetyl-hexosamines, and sialic acid submitted to methanolysis then acylation with HFBAA. Note that specific ions allowed identifying the different families of compounds. Fig. 4. Open in new tabDownload slide Reporter ion analysis in the EI mode of a mixture of pentoses, deoxy-hexoses, hexoses, uronic acids, N-acetyl-hexosamines, and sialic acid submitted to methanolysis then acylation with HFBAA. Note that specific ions allowed identifying the different families of compounds. In contrast with the TMS derivatives, HFB derivatives of O-methyl glycosides of sugars gave secure information under the electron impact mode, allowing the identification of fragments specific for families of compounds with a routine apparatus with a mass limit of 1000 amu. Because of the presence of characteristic peaks due to the HFB part (M = 69 (CF3+), 119 (CF2-CF3+), and 169 (CF2-CF2-CF3+)), it could be ascertained that the compound was an heptafluorobutyrate derivative and contained hydroxyl or amino groups. Furthermore, using the EI/MS mode, specific ions were reproducibly obtained for each family of compounds (Table VI), independent on the quantity of material. At the present status, the nature of pentoses, deoxyhexoses, hexoses, hexosamines, uronic acids, sialic acid and KDN, aspartic acid, the internal standard lysine, fatty acid methyl esters, and sphingosines can be identified using a selective research of characteristic ions by chromatogram reconstitution. (Figures 2, 4). The presence of specific ions for the furanic forms of the O-methyl glycosides of Fuc, Gal, GalNAc, and uronic acids (legend of Table VI) reinforced the identification of the different compounds. Because of the presence of a large number of fluorine atoms in each derivative, we tested the chemical ionization mode (NH3 gas) with a detection of negative ions using the Automass II device (mass limit 1000 amu). The different compounds can be detected by intense ions corresponding to [M]°− ion and/or [M−20] °− ion, the latter issued from an association resonance capture of an electron with the elimination of a HF molecule. Although this detection did not allow obtaining signals for FAME, it appeared to be an extremely sensitive detection technique (more than 20 times that of the positive ion detection), not only for the HFB derivatives of O-methyl glycosides, but also for those of sphingosines and of hydroxylated fatty acid methyl esters. Conclusions and perspectives The method described here for the determination of the monosaccharide composition of glycoprotein and glycolipids provides important advantages compared to those previously described. The major features are simplicity of the technique, the stability of the derivatives, and the quasi-absence of interferences by classical contaminants. The quantitation of the hexosamine and sialic acid content is by far more reliable than that measured by the classical TMS method, since there is no requirement for re-acetylation. Although the molar response of the HFB derivatives is 50% lower than that of the TMS derivatives, the stability of the HFB derivatives allows one to inject (if necessary), after concentrating under a stream of nitrogen, the whole sample on the column. The quantitation available for the GlcNAc residue involved in the N-glycosidic bond as a separated compound provides an essential information on the structure of N-glycans. As demonstrated elsewhere (Maes et al., 1999), the N-glycosidic bond is cleaved with 96% yield under our methanolysis conditions. Several compounds are produced corresponding essentially to the α and β isomers of glucosamine that are clearly separated from the O-methyl glycosides of glucosamine, which are derived from O-linked GlcNAc. Although these compounds can be detected on chromatograms of TMS derivatives, their quantitation remains impossible because of uncontrolled degradation within the GC injector and interference with α-Glc; Maes et al., 1999). Because of the absence of large interferences stemming from buffers and polyacrylamide, the method allows the quantitative determination of the glycoprotein monosaccharide composition with high sensitivity. Indeed, less than 1 µg of glycoprotein purified from acrylamide or on PDVF membranes is required. As observed recently in our laboratory, the exact carbohydrate composition can be obtained directly from a glycoprotein dissolved in phosphate-buffered saline without interference. For glycoproteins and glycolipids, not only is the carbohydrate composition accessible but one can also obtain that of the fatty acids methyl esters and that of sphingosines. Specific ions in the EI/MS mode and the molecular mass in the CI/MS modes can be reproducibly obtained, thus nailing down the nature of the monosaccharides, and that of other constituents of glycoconjugates such as FAMEs and sphingosine bases. Works in the field of mass spectrometry are now in progress to identify specific ions allowing the identification of each isomer of the heptafluorobutyrate derivatives of all O-methyl glycosides. Because of the reproducibility of the spectra, independent on the quantity of material loaded of the GC/MS device, it is expected that computer storage of the data of these spectra can allow a clear and rapid identification of the different compounds using routine GC/MS systems. Materials and methods Chemicals Standard monosaccharides and sodium deoxycholate were from Sigma Chemical Co. (St. Louis, MO); heptafluorobutyric and pentafluoropropionic anhydrides were from Fluka (Buchs, Switzerland). SE-30 liquid phase and the Gas Chrom Q solid phase were purchased from Applied Science Inc. (State College, PA). The capillary columns (25 m 25QC3/BP1, 0.5 µm film phase) were from SGE France SARL (Villeneuve St. Georges, France). Purified calcinated calcium chloride was from Prolabo (Paris, France) and HPLC grade acetonitrile, TRIS, and acrylamide were from SDS (Peypin, France). Gangliosides GM1 was purified from the rat cerebellum (Zanetta et al., 1980), and ganglioside G5b was a generous gift of Prof. G. Tettamanti. Reduced oligosaccharides were kindly provided by Dr. E. Maes (Maes et al., unpublished observations). Fatty acid methyl esters were prepared from the synaptosomal plasma membranes of adult rat brain (Breckenridge et al., 1972). For use as an internal standard, lysine hydrochloride (Sigma) was recrystallized from a 1 mM HCl solution in water. Methanolysis Samples containing the internal standard (0.2-2 µg lysine) were lyophilized in conical heavy walled Pyrex tubes (2.0 ml) with Teflon-lined screw caps. 0.25-0.5 ml of the methanolysis reagent was added, and the closed vessels were left for 20 h at 80°C. The methanolysis reagent was obtained by dissolving anhydrous gaseous HCl (up to 0.5 M) at -50°C in anhydrous methanol previously redistilled on magnesium turnings (Zanetta et al., 1972). Gaseous HCl was prepared by the dropwise addition of concentrated sulfuric acid on crystallized sodium chloride. Acylation After methanolysis, samples were evaporated to dryness under a light stream of nitrogen in a ventilated hood, followed by the addition of 200 µl acetonitrile and 25 µl of heptafluorobutyric anhydride (HFBAA) with a pipette with plastic tips. The closed vessels were heated for 30 min at 100°C (or better for 5 min at 150°C in order to have a reflux reaction that derivatizes the traces of compounds on the vessel wall) in a sand bath. After cooling at room temperature, the samples can be stored for months at room temperature in the closed vessel without need of reacylation. When analysis had to be performed, the samples were evaporated in a light stream of nitrogen in a ventilated hood in order to eliminate the excess of reagent and the heptafluorobutyric acid (HFBA) formed during the acylation, then taken up in the appropriate volume of acetonitrile; this acetonitrile was stored in a closed vessel in the presence of calcinated calcium chloride in order to eliminate traces of water. An aliquot of the acetonitrile solution of the heptafluorobutyrate derivatives (HFB) was introduced in the Ross injector of the GC apparatus. Although we did not observed changes in the RMR of the different derivatives after storage for 2 days at room temperature of the samples in anhydrous acetonitrile, we do not recommend storing the derivatives in acetonitrile, but in the acylation mixture. Gas chromatography For preliminary studies of the volatility and stability of the HFBF derivatives, analyses were performed on a 3 m long column packed with 3% SE-30 (Applied Science Inc., State College, PA) on Gas chrom Q (Applied Science Inc.). Injector and detector temperature was 260°C; and the temperature program was 2°C/min from 100 to 240°C at a helium flow rate of 20 ml/min. For analytical purposes, analyses were performed on a Shimadzu GC-14A gas chromatograph equipped with a Ross injector and a 25 m long capillary column (25QC3/BP1; 0.5 µm film phase; SGE France SARL; Villeneuve St. Georges (France). Injector and flame ionization detector temperatures were 260°C and the temperature program was 1.2°C/min between 100 and 140°C, followed by 4°C/min from 140°C to 240°C then maintaining this temperature for 10 min. The carrier gas (helium) pressure was 0.8 bar. The second part of the temperature program is only partially involved in the separation of monosaccharide derivatives (except KDN and sialic acid), but can reveal the presence of additional compounds, especially fatty acids or sphingosines present in glycolipids and also in glycoproteins (GPI anchors, myristylation, etc.). Mass spectrometry For GC/MS analysis, the GLC separation was performed on a Carlo Erba GC 8000 gas chromatograph equipped with a 25 m 0.32 mm CP-Sil5 CB Low bleed/MS capillary column, 0.25 µm film phase (Chrompack France, Les Ullis, France). The temperature of the Ross injector was 260°C and the samples were analyzed using the following temperature program: 90°C for 3 min then 5°C/min until 240°C. The column was coupled to a Finnigan Automass II mass spectrometer or, for mass larger than 1000, to a Riber 10-10H mass spectrometer (mass detection limit 2000). The analyses were performed either in the electron impact mode (ionization energy 70 eV; source temperature 150°C) or in the chemical ionization mode in the presence of ammonia (ionization energy 150 eV, source temperature of 100°C). The detection was performed for positive ions or for negative ions, in separated experiments, the latter allowing the quite specific detection of heptafluorobutyrate derivatives with a higher sensitivity. Acknowledgements We thank Drs. J.-C.Michalski, G.Strecker, J.Lemoine, E.Maes, O.Kol, and Mrs. A.Copin for providing oligosaccharides and glycoproteins of known monosaccharide compositions and Mrs. C.Alonso for help in iconography. Abbreviations Abbreviations Ara arabinose Rha rhamnose Xyl xylose Fuc fucose Rib ribose Gal galactose Man mannose GlcNAc glucosamine GalNac N-acetyl-galactosamine ManNAc N-acetylmannosamine GalNAc-OH reduced GalNAc GlcNAc-OH reduced GlcNAc GlcA glucuronic acid GalA galacturonic acid NeuAc N-acetyl neuraminic acid KDN 3-deoxy-d-glycero-d-galacto-nonulosonic acid Meso meso-inositol Manni mannitol Asp aspartic acid Lys lysine HFB heptafluorobutyrate HFBAA heptafluorobutyric anhydride TFA trifluoroacetate EI electron impact CI chemical ionization MS mass spectrometry References Breckenridge W.C. , Gombos G. , Morgan I.G. . The lipid composition of adult rat brain synaptosomal plasma membranes , Biochim. Biophys. Acta , 1972 , vol. 266 (pg. 695 - 707 ) Google Scholar Crossref Search ADS PubMed WorldCat Chaplin M.F. . Chaplin M.F. , Kennedy J.F. . Monosaccharides , Carbohydrate Analysis. A Practical Approach , 1994 Oxford Oxford Universtity Press (pg. 1 - 41 ) Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Chiba A. , Matsumura K. , Yamada H. , Inazu T. , Shimizu T. , Kunusoki S. , Kanazawa I. , Kobata A. , Endo T. . Structures of sialylated O-linked oligosaccharides of bovine peripheral nerve alpha-dystroglycan. The role of a novel O-mannosyl-type oligosaccharide in the binding of alpha dystroglycan with laminin , J. Biol. Chem. , 1997 , vol. 272 (pg. 2156 - 2162 ) Google Scholar Crossref Search ADS PubMed WorldCat Maes A. , Strecker G. , Timmerman P. , Leroy Y. , Zanetta J.-P. . Quantitative cleavage of the N-glycosidic bond in the normal conditions of methanolysis used for the analysis of glycoprotein monosaccharides , Anal. Biochem. , 1999 inpress OpenURL Placeholder Text WorldCat Normand G. , Kuchler S. , Meyer A. , Vincendon G. , Zanetta J.-P. . Isolation and immunohistochemical localization of a chondroitin sulfate proteoglycan from adult rat brain , J. Neurochem. , 1988 , vol. 51 (pg. 665 - 676 ) Google Scholar Crossref Search ADS PubMed WorldCat Zanetta J.-P. , Vincendon G. . Gas-liquid chromatography of the N(O)-heptafluorobutyrates of the isoamyl esters of amino acids. I. Separation and quantitative determination of the constituent amino acids of proteins , J. Chromatogr. , 1973 , vol. 76 (pg. 91 - 99 ) Google Scholar Crossref Search ADS PubMed WorldCat Zanetta J.-P. , Breckenridge W.C. , Vincendon G. . Analysis of monosaccharides by gas-liquid chromatography of the O-methyl glycosides as their trifluoroacetate derivatives Application to glycoproteins and glycolipids , J. Chromatogr. , 1972 , vol. 69 (pg. 291 - 304 ) Google Scholar Crossref Search ADS PubMed WorldCat Zanetta J.-P. , Vitiello F. , Vincendon G. . Gangliosides from rat cerebellum: demonstration of a considerable heterogeneity using a new solvent for thin-layer chromatography , Lipids , 1980 , vol. 15 (pg. 1055 - 1061 ) Google Scholar Crossref Search ADS WorldCat © 1999 Oxford University Press
TI - Gas-liquid chromatography of the heptafluorobutyrate derivatives of the O-methyl-glycosides on capillary columns: a method for the quantitative determination of the monosaccharide composition of glycoproteins and glycolipids
JO - Glycobiology
DO - 10.1093/glycob/9.3.255
DA - 1999-03-01
UR - https://www.deepdyve.com/lp/oxford-university-press/gas-liquid-chromatography-of-the-heptafluorobutyrate-derivatives-of-NBBpg015An
SP - 255
EP - 266
VL - 9
IS - 3
DP - DeepDyve
ER -