High-performance liquid chromatography of long-chain neutral glycosphingolipids and gangliosides

498

Biochimica et Biophysics Acta, 7 12 (1982) 498-504 Elsevier Biomedical Press

BBA 51176

HIGH-PE~O~NCE GLYCOSP~N~LIPIDS WILLIAM

M.F. LEE, MARY ALICE WESTRICK

Cancer Research Institute (Received

LIQUID CHRO~TOG~P~ AND GANGLIOSIDES

February

and Department

OF LONG-CHAIN NEUT~L

and BRUCE A. MACHER *

of Medicine, University of California, San Francisco, CA 94143 (U.S.A.)

lOth, 1982)

Key words: ~anglios~de, ~~cosphingo~ipi~

HPLC;

Perb~nzoy~atio~; i@osidase

We describe a method for analyzing the perbenzoyl derivatives of both neutral glycosphingolipids and gangliosides with a single high-performance liquid chromatography system. Use of this system, combined with endo- and/or exoglycosidase treatment of glycosphingolipids, provides a sensitive method for obtaining structural info~ation on these ~rn~~ds. This system has two advantages over previously published chromatography procedures: (i) it uses a commercially available column, and (ii) this single column can be used to analyze gangliosides and their neutral glycosphingolipid products generated by neuraminidase treatment. With this method, we have studied 24 different glycosphingolipids, containing one to ten sugars and one or two sialic acid residues, and have demons~at~ its usefulness in evaluating the gangliosides present in human leukocytes.

Introduction

of commercially available, prepacked Zipax columns for HPLC analysis of perbenzoylated, short-chain neutral glycosphingolipids [3]. We have successfully used the latter chromatographic system, in combination with glycosidase hydrolysis of glycosp~ngoIipids, to elucidate the structure of neutral compounds containing less than five sugars isolated from various types of normal and malignant human leukocytes [4,5]. We now describe the application of a modification of this chromatographic system and technique for the structural analysis of gangliosides and longer-chain neutral glycosphingolipids.

The structural analysis of glycosphingolipids has tradionally involved the use of thin-layer and gas-liquid chromatography, mass spectrometry and glycosidase hydrolysis [I]. The app~cation of these techniques requires milligram quantities of the purified glycosphingolipid being studied. More recently, quantitative high-performance liquid chromatograpbic (HPLC) methods for the analysis of small amounts of neutral ~ycosp~ngolipids and ganglioside have been described [2,3]. Bremer et al. [2] demonstrated that the perbenzoylated derivatives of monosialogangliosides (GM4, GM3, GM2 and GMl) could be separated on a chromatography column custom-packed with LiChrospher Si 4000. Ullman and McCluer [3] described the use

Materials and Methods Clycosphingolipids

GMl, GM2 and GDla were purchased from Supelco, Inc. (Bellefonte, PA), GM4 was the gift of Dr. Alan Yates (Ohio State University, Columbus, OH); G9 [6] and V13NeuAc nornLcOse,Cer were the gift of Dr. Reiji Kannagi

* To whom correspondence should be-addressed. Abbreviations: the common trivial names and abbreviations for neutral glycosphingolipids and gangliosides are given in Tables I and II, along with their structures.

~5-2760/82/0~0-0~/$02.75

0 1982 Elsevier Biomedical

Press

499

(University of Washington, Seattle, WA); H-2 and H-3 glycosphingolipids were the gift of Drs. Michiko and Minoru Fukuda (La Jolla Cancer Research Foundation, La Jolla, CA); IV3GalnLcOse,Cer, GalCer, GaOse,Cer and I 3S03GalCer were the gift of Dr. Bader Siddiqui (Veterans Administration Hospital, University of California, San Francisco; CA); IV3GalNAcGbOse,Cer (Forssman antigen) was the gift of Dr. S.S.J. Sung (Rockefeller University, New York, NY). GlcCer, LacCer, LcOse,Cer, nLcOse,Cer (paragloboside), GM3 and IV3NeuAc-nLcOse,Cer (sialoparagloboside) were purified from human leukocytes as previously described [7,8]. GbOse,Cer, GbOse,Cer (globoside) were purified from human erythrocytes, and GgOse,Cer (asialo GM2) and GgOse,Cer (asialo GMl) were prepared by neuraminidase treatment of GM2 and GMl, respectively. Human leukocyte neutral glycosphingolipids, [4,5] and gangliosides [4] were obtained as previously described. [ 3H]GM, and [ 3H]LacCer were prepared by the method described by Schwarzmann [9] using NaB3H, (New England Nuclear, Boston, MA; spec. act. 280 mCi/mmol). The 3H-labeled glycosphingolipids were purified by preparative thin-layer chromatography as previously described [7].

Perbenzoylation of glycosphingolipids Neutral glycosphingolipids were perbenzoylated as previously described [3]. Gangliosides were perbenzoylated by the same method as used by Ullman and McCluer [3] for neutral glycosphingolipids. In this procedure, gangliosides were dissolved in 0.5 ml of freshly prepared 10% benzoyl chloride in pyridine, and the reaction mixture was incubated at 37°C. We determined the minimum time for completion of this reaction for gangliosides by stopping the reaction after a variable incubation period. The reaction mixture was evaporated under a stream of nitrogen, and the perbenzoylated ganglioside dissolved in 3 ml of hexane. This was washed four times with methanol/O.2 M aqueous sodium carbonate (80: 20, v/v) and once with methanol/water (80: 20, v/v). The hexane was evaporated under nitrogen, and the residue suspended in chloroform for HPLC.

High-performance liquid chromatography All chromatographic studies were carried out using a Du Pont 850 HPLC system (E.I. Du Pont de Nemours and Co., Inc., Wilmington, DE) with an SP4100 Analyzer (Spectra-Physics, Santa Clara, CA). Perbenzoylated glycosphingolipids were detected by ultraviolet absorbance at 230 nm using a Hitachi 100-10 Spectrophotometer with a continuous flow-cell. HPLC of both perbenzoylated neutral glycosphingolipids and perbenzoylated gangliosides was performed using a 2.1 mm X 100 cm Zipax (pellicular silica) column (E.I. Du Pont de Nemours and Co., Inc., Wilmington, DE). Elution of neutral compounds containing less than six sugars was carried out with linear gradient A (l-23% dioxane in hexane over 30 min at a flow rate of 2ml/min at an approximate pressure of 1000 lbs/in&); elution of perbenzoylated gangliosides and/or long-chain neutral glycosphingolipids was usually performed with linear gradient B (7-318 dioxane in hexane over 33 min at a flow rate of 2 ml/min), although occasionally gradient A was used for the shorter gangliosides. Note that both gradients A and B have a slope of 0.73% dioxane/hexane per n-tin and differ only in their starting and finishing concentrations. Glycosidase hydrolysis of gangliosides Gangliosides were treated with Clostridium perfringens neuraminidase with or without detergent as previously described [8]. Gangliosides and neutral glycosphingolipids were treated with Escherichia freundii endo-&galactosidase (a gift from Drs. Michiko and Minoru Fukuda, La Jolla Cancer Research Foundation, La Jolla, CA) as previously described [lo]. Results Perbenzoylation of gangliosides The time course of perbenzoylation at 37°C was tested for several standard gangliosides (GM3, GM 1, GD 1a and IV3NeuAc-nLcOse,Cer). Reaction times of 8 h or less resulted in incomplete perbenzoylation of gangliosides and the appearance of multiple peaks on HPLC due to partially derivatized compounds. Incubation for 1224 h resulted in complete perbenzoylation of all gangliosides tested, as evidenced by a single peak

500

TABLE

I

RETENTION

TIMES

OF PERBENZOYLATED

NEUTRAL

GLYCOSPHINGOLIPIDS

The retention time given is the median of at least three runs; range is -C 1 min. Gradient A, l-23& dioxane/hexane, flow rate of 2 ml/min. Gradient B, 7-318 dioxane/hexane over 33 min, at a flow rate of 2 ml/min. R=GalPl-4GlcBl+ 1Cer. Neutral

glycosphingolipid

Structures

over 30 min. at a In the structures,

Retention Gradient A

GbOsesCer GgOse,Cer LcOse,Cer

(asialo GM2)

GbOse,Cer GgOse,Cer nLcOse,Cer

(globoside) (asialo GMl) (paragloboside)

IV3Gal-nLcOse,Cer IV3GalNAc-GbOse4Cer

5.5 5.5

GalPl - 4Glcfi 1 - 1Cer Gala1 -4Gal/?l+ 1Cer

10.0 9.0

Galnl -4R GalNAcj31 - 4R GlcNAcP 1 - 3R

14.0 15.5 18.0

6.0 7.5 10.0

18.5 20.5 22.5

10.5 12.5 14.5

25.0 25.0

17.0 17.0

GalNAcPl - 3Galal-4R GalPI -3GalNAc$l-4R Gal,&1 -4GlcNAc/31-3R Gale1 -3GalPl-4GlcNAcbl-3R GalNAcal - 3GalNAcbl3Gal/31-

(Forssman)

Gradient B

GlcPl1Cer Gal/31 - 1Cer

GlcCer GalCer LacCer GaOse,Cer

time

4R

H-2

Fucal-2Gal~l-4GlcNAc~l-3Gal/3l-4GlcNAc~1+

3R

23.0

H-3

Fuccvl -2Gal,f% -4GlcNAc,01-6 Fuccul - 2Gal/31+ 4GlcNAc$l+

3R

28.0

TABLE

3

Gal/?1 -4GlcNAcPl+

II

RETENTION

TIMES

OF PERBENZOYLATED

GANGLIOSIDES

The retention time given is the median of at least three runs; range is * 1 min. Gradient A, l-23% flow rate of 2 ml/min. Gradient B, 7-31% dioxane/hexane over 33 min, at a flow rate R=j31-4Glc/3+lCer. Gangliosides

Structures

GM4 GM3 GM2 GM1 GDla IIINeuAc-LcOse,Cer IV3NeuAc-nLcOse,Cer V13NeuAc nor-nLcOs%Cer G9

dioxane/hexane of 2 ml/min.

Retention

NeuAccu23GalPl+ 1Cer NeuAca2 - 3GalR GalNAcPl - 4Gal(3 c 2 aNeuAc)R GalPI - 3GalNAcbl+ 4Gal(3 + 2 cyNeuAc)R NeuAca2-3Gal~l-3GalNAc/31-4Gal(3+2aNeuAc)R

NeuAcaZ-

over 30 min, at a In the structures,

NeuAca - GlcNAcPI - 3GalR NeuAccv2 - 3GalPl4GlcNAcPl-+ 3GalR 3Gal~l-+4GlcNAc/31-+ 3Gal/31-4GlcNAcPl3GalR

L-Fuctul-2Gal/3l-4GlcNAc~l-6 NeuAca23Gal~l-4GlcNAc/31~

SO;

I ‘SO,-GalCer a See Ref. 7. The retention times given are based upon analysis example) and should be considered tentative.

Gala1 -4GlcNAc~l3

of mixture

of gangliosides

- 3Gal/31-

3GalR 1Cer

which contain

time

Gradient A

Gradient B

14.5 17.5 20.5 24.0 27.5

6.5 9.5 12.5 16.0 ‘19.5

21.5 27.0 -

13.5 a 19.0 25.5

-

29.0

18.0

10.0

this compound

(see Fig. 3 for

501

on the chromatogram and the achievement of the maximum area under the elution peak (data not shown). Recovery of material derivatized under these conditions was determined by following the recovery of [3H]GM3 and [3H]LacCer. Recovery of ‘H-GM, and 3H-LacCer after derivatization was 92 and 94%, respectively. When the HPLC column eluate was monitored for tritium, the recovery of 3H-GM, and 3H-LacCer was 82 and 85%, respectively. Thus, the overall recovery of 3H-GM3 and 3H-LacCer was 75 and 80% respectively. These values are comparable to those (85%

GM4

G0lC.r

GgOre4Cr

GMI

for GM,) reported by Bremer et al. [2] using the LiChrospher HPLC method. Our procedure apparently gave complete perbenzoylation of all the glycosphingolipids listed in Tables I and II. HPLC of perbenroylated gangliosides and neutral glycosphingolipids using the Zipax column Two solvent gradients were utilized to elute the various perbenzoylated glycosphingolipids, and the retention times of these compounds for each gradient are listed in Tables I and II. Short-chain (l-5 sugars) neutral and acidic glycosphingolipids are eluted within a reasonable period of time by gradient A (l-238 dioxane in hexane over 30 min); longer-chain compounds are more efficiently eluted by gradient B (7-258 dioxane in hexane over 30 min). The difference in retention times for a compound eluted with these two gradients is about 8 min. All glycosphingolipids with less than three sugars elute near the solvent front with gradient B, and are poorly resolved. However, they have a retention time of 5-10 min with gradient A and are well resolved. The use of gradient B, which starts at 7% dioxane in hexane, for the elution of longer-chain compounds saves time and solvent, and does effect resolution. Figs. 1 and 2 show the HPLC chromatograms

GDlo

L 5

IO

RETENTION

15

TIME

20

25

,

30

RETENTION

(minutes)

Fig. 1. High-performance liquid chromatography of perbenzoylated standard gangliosides and their corresponding neutral glycosphingolipids. Gangliosides and the neutral glycosphingolipids resulting from neuraminidase treatment were perbenzoylated and chromatographed as described in Materials and Methods. Structures of the glycosphingolipids are given in Tables I and II.

TIME

(minutes)

Fig. 2. High-performance liquid chromatography of perbenzoylated long-chain gangliosides and neutral glycosphingohpids. Compounds were perbenzoylated as chromatographed as described in Materials and Methods. Perbenzoylated glycosphingolipids shown are: (A) IV’Gal-nLcOse,Cer and IV3GalNAcGbOse,Cer (Forssman antigen); (B) H-2 glycosphingolipid; (C) H-3 glycosphingolipid; (D) IV3NeuAc nor-nLcOs%Cer; (E) G9 glycosphingolipid. Structures of these compounds are given in Tables I and II.

502

for 16 of the glycosphingolipids tested. Fig. 1 illustrates the separation of compounds with six sugars or less using gradient A. This figure demonstrates that the retention time is dependent on carbohydrate chain length and charge, and that good separation is obtained when glycosphingolipids of the same structural family are analysed. Thus, GgOse,Cer is resolved from GgOse,Cer and GM3, GM2, GM1 and GDla are separated. When members of different glycosphingolipid families are analyzed, overlap of retention times occurs. For example, GD 1a and IV3NeuAc-nLOse,Cer (Spg) coelute. Fig. 2 shows the separation of longer-chain glycosphingolipids (including blood group active compounds, with and without branched carbohydrate moieties) using gradient B. Separation of compounds with different carbohydrate chain lengths is achieved, but overlap is seen for the two neutral pentaosylceramides. It should be noted that sulfated glycosphingolipids can be chromatographed with this HPLC system (Table II), and that glycosphingolipids containing hydroxylated fatty acids have retention times slightly longer (approx. 1 min) than the equivalent compounds with nonhydroxylated fatty acids (data not shown). Glycosidase treatment of gangliosides Because glycosphingolipids with different structures are not always resolved from one another by HPLC, we have used glycosidases to facilitate the identification of these compounds. This approach is illustrated for standard compounds in Fig. 1 and for an unknown mixture of gangliosides isolated from acute leukemia cells in Fig. 3. Each chromatogram in Fig. 1 shows the resolution of a parent ganglioside (e.g., GM4) and its neuraminidasegenerated product (e.g., GalCer). The usefulness of this procedure for structural identification is illustrated in the bottom two chromatograms of Fig. 1. The retention times of the gangliosides GD 1a and IV 3NeuAc-nLcOse,Cer (Spg) are nearly identical. However, removal of sialic acid residues from these two gangliosides by neuraminidase produces products which are resolved by HPLC. In addition, GDla can be hydrolyzed sequentially to GM1 (neuraminidase without detergent) and GgOse,Cer (neuraminidase with detergent), whereas IV3NeuAc-nLcOse,Cer gives the same

c %

2

0

P a

:-

:o RETENTION

;

:o TIME

2-5

3;

i

(minutes)

Fig. 3. High-performance liquid chromatography of perbenzoylated gangliosides from human acute leukemia cells before and after neuraminidase treatment. Upper panel: chromatogram of perbenzoylated intact gangliosides which shows peaks corresponding to perbenzoylated (A) GM3, (B) III NeuAcLcOse,Cer and (C) IV3NeuAc-nLcOse,Cer. Lower panel: chromatogram of perbenzoylated neutral glycosphingolipids derived from the gangliosides by neuraminidase treatment. It shows peaks corresponding to perbenzoylated (a) LacCer, (b) LcOse,Cer and (c) nLcOse,Cer.

product with either treatment. Compounds with a lacta/neolacto structure, such as IV3NeuAcnLcOse,Cer, are specifically hydrolyzed to GlcCer (or LcOse,Cer for longer-chain linear compounds) and an oligosaccharide by E. freundii endo+galactosidase treatment [lo], and can be useful for distinguishing glycosphingolipids with similar elution times. Fig. 3 demonstrates the combined application of neuraminidase treatment and HPLC for the tentative identification of gangliosides isolated from human acute leukemia cells. The upper panel is the HPLC chromatogram of the gangliosides preparation, and the lower panel is the chromatogram of neuraminidase hydrolysis product. Compounds A, B and C have retention times which corresponds to the gangliosides GM3, III NeuAcLcOse,Cer (a ganglioside thus far identified only in human leukocytes [S]) and IV3NeuAcnLcOse,Cer (Spg); however, other structural assignments are possible (see Tables I and II). A small peak at 10 min probably corresponds to the major neutral glycosphingolipid of these cells, LacCer. Following neuraminidase treatment of the gangliosides and perbenzoylation of the neutral glycosphingolipid products, three compounds (a, b

503

and c) are seen by HPLC. The retention times of these compounds corresponds to those of the expected products LacCer, LcOse,Cer and nLcOse,Cer. These products can be studied further by hydrolysis with other exoglycosidases or endo-P-galactosidase to provide a greater degree of confidence for the structural assignments [4]. Discussion The HPLC analysis of perbenzoylated shortchain neutral glycosphingolipids and gangliosides has been described by Ullman and McCluer [3] and Bremer et al. [2], respectively. The conditions described for the perbenzoylation of these two classes of glycosphingolipid were different, and separation of the derivatized compounds was achieved using columns containing different chroGangliosides were sepmatographic material: arated on a column custom-packed with LiChrospher Si 4000 [2], whereas neutral compounds were separated on a commercially available Zipax column [3]. In this report, we show that a single derivatization procedure can prepare both classes of glycosphingolipid for chromatography, and that the commercially available Zipax column can be used to separate short- and long-chain neutral glycosphingolipids, and gangliosides. We show that the previously described derivatization procedure of Ullman and McCluer [3] can be used to perbenzoylate mono- and disialylated gangliosides, and neutral glycosphingolipids with as many as ten sugars. After perbenzoylation, all of the compounds were conveniently analyzed with a single column by adjusting the solvent gradient to suit the glycosphingolipids being studied. Using the Zipax column, we studied the elution characteristics of 24 different glycosphingolipids. An examination of the retention times of shortchain (one to four sugars) neutral glycosphingolipids reveals some interesting structure-retention time relationships. Comparison of the triaosylceramides shows that the compounds with an amino sugar have a longer retention time than those with only neutral sugars, and that the retention times have the following order: LcOse,Cer > GgOse,Cer > GbOse,Cer. This elution order is also maintained when the corresponding tetraosylceramides are analyzed: nLcOse,Cer > GgOse,Cer

> GbOse,Cer. Even when these compounds are sialylated, this order is maintained: IV3NeuAcnLcOse,Cer > GMl. In fact, it takes two sialic acid residues on a GgOse,Cer structure (GDla) to give the same retention time as IV3NeuAcnLcOse,Cer. The effect of two other structural features upon the retention time can be seen from the data in Tables I and II: (i) the effect of branching, and (ii) the addition of negatively charged groups. In general, a 3-4 min increase in retention time is seen with the addition of a single sugar to the glycosphingolipid structure. Thus, one would predict a 9- 12 min difference in retention time between H-2 and H-3, and between VI 3NeuAc-nLcOs% Cer and G9. The actual difference is only 3-5 min, indicating that carbohydrate chain branching results in a smaller change in the retention time per additional sugar residue than linear structures. The addition of a sialic acid also causes an increase in the retention time. This increase is inversely related to the chain length of the neutral core. For example, the difference in the retention for GalCer and GM4 is 9 min, whereas the retention time difference for nLcOse,Cer and IV3NeuAcnLcOse,Cer is only 5 min. Two other polar groups that are found among the glycosphingolipids which we studied are the sulfate group and the hydroxy fatty acids. Sulfated GalCer was retained on the Zipax column about 12 min longer than GalCer, and the presence of a hydroxy fatty acid in GbOse,Cer increased the retention time by 1 min. The advantages of using HPLC to analyze glycosphingolipids are the sensitivity of the technique (requiring only microgram amounts of material) and the ability to study mixtures of compounds (see Fig. 3 for example). Its disadvantages are that (i) it is a ‘comparative’ analytical technique, requiring the availability of standards of known structure, and (ii) compounds with different structures may cochromatograph and not be resolved. The addition of specific glycosidase hydrolysis of glycosphingolipids to HPLC of these compounds aids in the resolution of the latter problem. Using this combined approach, one can make preliminary structural assignments with greater confidence than by using HPLC alone. However, when compared to a single HPLC analysis [2,3], the combined approach requires a greater amount of the material

504

for analysis. Nevertheless, the HPLC-glycosidase method that we describe provides a rapid and reasonably reliable means for identifying small amounts of glycosphingolipids. This advantage plus the commercial availability of pre-packed Zipax columns, which can resolve both neutral glycosphingolipids and gangliosides, makes this a convenient system for glycosphingolipid analysis. Acknowledgements

W.M.F.L. and M.A.W. are recipients of postdoctoral research fellowships from the National Institutes of Health/National Cancer Institute. This work was supported in part by Cancer Research Funds from the University of California, San Francisco.

References

1 Esselman,W.J.,

Laine, R.A. and Sweeley, C.C. (1972) Methods Enzymol. 28, 140- 156 2 Bremer, E.G., Gross, S.K. and McCluer, R.H. (1979) J. Lipid Res. 20, 1028-1035 3 Ullman, M.D. and McCIuer, R.H. (1978) J. Lipid Res. 19, 910-913 4 Lee, W.M.F., Westrick, M.A., Klock, J.C. and Macher, B.A. (1982) Biochim. Biophys. Acta 711, 166-175 5 Lee, W.M.F., Westrick, M.A. and Macher, B.A. (1982) J. Biol. Chem., in the press 6 Watanabe, K., Powell, M. and Hakomori, Chem. 253, 8962-8967 7 Macher, B.A. and Klock, 2092-2096

S. (1978) J. Biol.

J.C. (1980) J. Biol. Chem.

255,

8 Macher, B.A., Klock, J.C., Fukuda, M.N. and Fukuda, M. (1981) J. Biol. Chem. 256, 1968-1974 9 Schwarzmann, G. (1978) Biochim. Biophys. Acta 529, 106ll4 IO Fukuda, M.N., Watanabe, K. and Hakomori, S.-I. (1978) J. Biol. Chem. 253, 6814-6819