The Analysis of Lipids other than Fatty Acids

Sections H and I.  GC of Sphingolipids and of Total Lipid Extracts

H.  Sphingolipids

1.  Preliminaries to GC separation

The basic lipid moiety of a sphingolipid is a ceramide, consisting of a long-chain(sphingoid base) linked via an amide bond to a fatty acid. Though free ceramides can occur in small amounts in tissues, they usually form part of complex lipids containing sugar moieties, as in glycosphingolipids, or a phosphorus group, as in sphingomyelin and certain phosphonolipids. One approach to the analysis of sphingolipids by GC therefore consists of preparation of the ceramides by suitable procedures for conversion to nonpolar derivatives such as the TMS ethers. Intact sphingolipids can be analysed by GC only with difficulty, and this is one area where HPLC has particular advantages; the compounds can be analysed in native form or after conversion to benzoyl derivatives, which can be detected with some sensitivity by UV spectrophotometry [168]. The analysis of long-chain bases is described in Chapter 10.

It is a relatively simple matter to hydrolyse sphingomyelin and ceramide aminoethylphosphonate to ceramides with the enzyme phospholipase C from Clostridium welchii, and practical details of the hydrolysis procedure and of methods of purifying the product are given in Section C.1. Hydrolysis of sphingomyelin can also be accomplished chemically by reaction with hydrofluoric acid [760].

Unfortunately, it is much less easy to prepare ceramides from glycosphingolipids. Compounds containing dihydroxy bases can be converted to ceramides by a chemical procedure devised by Carter et al. in 1961 [150], in which the glycosidic ring is opened with periodate and the resulting product reduced with sodium borohydride before being hydrolysed under mild acidic conditions. Derivatives of trihydroxy bases cannot be converted to ceramides by this procedure, as the bases would be cleaved across the vicinal diol group by the peroxide oxidation step. Although many enzymes have been described that will cleave bonds between the ceramide and carbohydrate residues, none appear to be readily obtainable with a sufficiently high specific activity for structural studies of ceramides.

The nature of the fatty acids and long-chain bases in sphingolipids is described in Chapter 2, but it is perhaps worth reiterating briefly that the fatty acids are commonly saturated and monoenoic components (up to C26), and they may also have a free hydroxyl group in position 2; di- and trihydroxy bases occur, varying in degree of unsaturation, but frequently with a trans-double bond in position 4. Ceramides derived from these therefore have variable numbers of free hydroxyl groups in different regions of the molecule, i.e. two to four in total (two or three in the base and zero or one in the fatty acid). Some preliminary separation prior to analysis by GC or other means can then often be helpful. Karlsson and Pascher have published a valuable paper on TLC of ceramides [467], and layers of silica gel or better of silica gel containing diol-complexing agents, such as sodium tetraborate (Na2B4O7·10H2O) or sodium meta-arsenite (NaAsO2), can be used to effect separations that depend on the number and configuration of the hydroxyl groups present. Layers are prepared by incorporating 1% (w/v) of the salt into the water used to prepare the slurry of adsorbent. In addition, ceramides having a trans-double bond in position 4 of the long-chain base are separable on silica gel impregnated with sodium borate (saturated compounds migrate ahead of unsaturated), although the reason for the effect is not understood [624]. The long-chain base and fatty acid constituents may each contain up to two double bonds, so TLC with silica gel impregnated with silver nitrate (5%, w/w) can be used to good effect.

From these studies, Karlsson and Pascher were able to recommend a programme for the preliminary fractionation of natural mixtures by TLC, although they recognise that, as the components of natural ceramides vary greatly in chain length, bands may tend to spread more than is the case with pure model compounds. Four groups of ceramide can be separated on the arsenite layers with chloroform-methanol (95:5, v/v) as the solvent system; dihydroxy base-normal acid, dihydroxy base-hydroxy acid and similar species containing trihydroxy bases, as illustrated in Figure 8.12 (plate A). Derivatives of dihydroxy bases isolated in this manner can then be separated on borate-impregnated layers using the same solvent mixture for development, according to whether they contain long-chain bases with trans-double bonds in position 4 or not (Fig. 8.12, plate B). Finally all the components isolated thus far can be further separated into groups, according to the combined numbers of cis-double bonds (trans-double bonds have comparatively little effect) in the component fatty acids and in the long-chain bases on silver nitrate-impregnated layers. Better results are obtained in this instance if the free hydroxyl groups are first acetylated with acetic anhydride and pyridine (see Chapter 4); chloroform-benzene-acetone (80:20:10, by volume) was chosen as a suitable solvent system for the development, but it would now be considered safer to replace the benzene with toluene.

Figure 8.12. Schematic separation of ceramides by TLC on layers of silica gel G, impregnated with 2% sodium arsenite (plate A) and 2% sodium borate (plate B), using a solvent system of chloroform-methanol (95:5, v/v) for development [467].

Abbreviations: na, normal fatty acid; ha, hydroxy fatty acid; db, dihydroxy base; tb, trihydroxy base; sat, saturated.

TLC chromatogram of ceramides

The preliminary separation of ceramides is undoubtedly an area where HPLC could make an effective contribution, assuming that enterprising analysts are prepared to devote time to the problem.

For GC analysis, ceramides have generally been converted to TMS ether derivatives, but BDMS ethers [649] and cyclic boronates [279] have also been prepared.


2.  Gas chromatography-mass spectrometry

Mass spectrometry of intact glycosphingolipids by direct probe insertion and soft ionisation procedures has become one of the major means of structure determination for these compounds [460,896,974]. Information is thereby obtained on the nature of both the lipid and carbohydrate moieties, although the presence of different molecular species hampers the interpretation of the spectra greatly. Detailed discussion of this aspect is beyond the scope of this book.

The molecular weights of most glycosphingolipids are too high to permit fractionation by GC, but monoglycosylceramides (cerebrosides) have been successfully analysed by the technique in the form of the TMS ether derivatives [347,350,405,633,681,788,892]. Packed columns containing nonpolar phases have been employed for the purpose under conditions similar to those required for the separation of intact triacylglycerols (see Section B.1), so species are resolved by molecular weight only and not by degree of unsaturation. As an example, a separation of glucocerebrosides from the spleen in Gaucher's disease is illustrated in Figure 8.13 [681]. A glass column (1 m) of 1.5% OV-1TM on Chromosorb WTM (100-120 mesh) was maintained at 320°C. In this instance, there was essentially only one long-chain base (sphingosine), so the separation depends simply on the nature of the fatty acid constituents. Components were identified by chemical ionisation MS with isobutane as the reagent gas, as this technique gives a detectable quasi-molecular ion, together with other ions indicative of the molecular weight ([MH−90]+ and [MH+73]+), while ions revealing the molecular size of the carbohydrate, ceramide, long-chain base and fatty acid moieties are also prominent. Naturally occurring ceramides in spleen were similarly subjected to GC-MS as the TMS ethers in this study. While electron-impact MS can give valuable data, ions in the high mass range are less easily detected [347,405,788].

GC chromatogram of the TMS ether derivatives of glucocerebrosides

Figure 8.13. GC separation of the TMS ether derivatives of glucocerebrosides from the spleen of a patient with Gaucher's disease [681]. A glass column (1 m) packed with 1.5% OV-1TM on Chromosorb WTM (80-100 mesh) was maintained at 320°C. Only the fatty acid component is identified as the long-chain base in each fraction was sphingosine. (Reproduced by kind permission of the authors and of Chemistry and Physics of Lipids, and redrawn from the original paper.)


It is perhaps rather surprising, but intact ceramide aminoethylphosphonate from shellfish has been successfully subjected to GC separation in the form of the TMS ether derivative [601]. Presumably the phosphonate bond is much more stable than a phosphate at elevated temperatures.

Much more work has been done on GC of sphingolipids after conversion to simpler ceramides and thence to appropriate volatile derivatives, and these are eluted from GC columns under similar conditions to diacylglycerols (see Section C of this chapter). As ceramide standards are not readily available, and considerable variation in molecular structure can occur, identification of peaks can present enormous difficulty unless the analyst has access to GC-MS. Much of the early work in this area consisted of systematic studies with model ceramides, that provided a solid foundation for subsequent analyses of natural samples. The chromatographic conditions employed in the pioneering studies in 1969 were a glass column (1.2 m × 3 mm i.d.), containing 2% OV-1TM on GasChrom QTM (60-80 mesh), maintained at temperatures of 300 to 320°C, with helium as the carrier gas [786]; similar general conditions have been employed by most others who subsequently worked on related topics. As discussed before for triacylglycerols, separation is solely on the basis of molecular weight. The detector response was found to vary with such factors, as column temperature, the flow-rate of the carrier gas and the molecular weight of the ceramide derivative, so careful standardisation of conditions and calibration was necessary for quantitative work [789]. In order to ascribe meaningful relative retention times to ceramide derivatives, they were assigned "triglyceride carbon unit" (TGCU) values according to the retention times of saturated triacylglycerol standards. Some representative values of this kind are listed for reference in Table 8.1.

Table 8.1. Triglyceride carbon units (TGCU) for the TMS ether derivatives of sphingosine and sphinganine ceramides [789]
Fatty acidnormal2-hydroxynormal2-hydroxy
16:0 37.4 38.1 37.7 38.0
18: 0 39.5 40.0 39.7 40.0
18:1 39.2   39.1  
19:0 40.5   40.7  
20:0 41.6 42.0 41.7 41.9
22:0 43.6 43.9 43.7 43.9
22:1 43.3      
23:0 44.6   44.7  
24:0 45.7 45.8 45.8 45.8
24:1 45.5      
25:0 46.7   46.7  


The electron-impact mass spectrum of the 1,3-di-O-TMS ether derivative of N-stearoyl-sphingosine is illustrated in Figure 8.14 [786,789]. A molecular ion (M = 709) is not apparent, but there are ions at [M−15]+ (m/z = 694), [M−90]+ (m/z = 619) and [M−103]+ (m/z = 606), formed by elimination of a methyl, trimethylsilanol and the terminal methylene plus trimethylsilanol groups, respectively, and these serve to indicate the molecular weight. Loss of the fatty acid residue produces characteristic ions for [M−(b+1)]+ at m/z = 426 and [M−(b+1+90)]+ at m/z = 336 among others. Cleavage between carbons 2 and 3 of the molecule gives a prominent ion, diagnostic for the long-chain base, at m/z = 311. In addition, ions at m/z = 471, equivalent to [M−(a-73)]+, and m/z = 398, equivalent to [M−a]+, are of great value in structure assignment. Minor differences only are seen when the long-chain base is saturated. Analogous spectra are seen when the ceramide contains a 2-hydroxy fatty acid, although there are also certain distinctive fragmentations [351,789], and this is also true of ceramides containing trihydroxy bases, such as phytosphingosine [345].

mass spectrum of 1,3-di-O-trimethylsilyl-N-stearoyl sphingosine

Figure 8.14. The mass spectrum of 1,3-di-O-trimethylsilyl-N-stearoyl sphingosine [789]. (Reproduced by kind permission of the authors and of Chemistry and Physics of Lipids, and redrawn from the original paper.)

Because of the difficulty of preparing ceramides from cerebrosides, relatively little work has been done on GC separations of such compounds. Nonetheless, cerebrosides from bovine [346] and mouse [350] brain were converted to ceramides by the chemical procedure described in the previous section, and these were separated into hydroxy and nonhydroxy acid-containing substituents and then by degree of unsaturation as the acetates, before hydrolysis and conversion to the TMS ethers for analysis by GC-MS. The GC trace obtained from a fraction containing sphingosine in combination with saturated 2-hydroxy fatty acids is illustrated in Figure 8.15. The chromatographic trace is relatively simple enabling mass spectrometric identification of each peak, and similar uncomplicated traces were obtained from the remaining TLC fractions.

GC chromatogram of the TMS ethers of a ceramide fraction

Figure 8.15. GC separation of the TMS ethers of a ceramide fraction containing sphingosine and non-hydroxy fatty acids from brain cerebrosides [346]. A glass column (1.7 m × 3.5 mm), packed with 2.4% OV-1TM on Gas Chrom QTM (100-120 mesh), was maintained at 300°C. Only the fatty acid component is identified. (Reproduced by kind permission of the authors and of the European Journal of Biochemistry, and redrawn from the original paper.)


More attention has been given to the separation of molecular species of ceramides derived from sphingomyelin by GC-MS. For example, native ceramides and sphingomyelins from human plasma [787,789] and bovine intestines [126], and ceramide 2-N-methylaminoethylphosphonate from the shellfish Turbo cornutus [601,603] have been analysed as the TMS ethers. In addition, ceramides from sphingomyelins of human plasma [649] and bovine brain [278,279] have been examined as the BDMS and cyclic boronate derivatives, respectively, since these have improved mass spectral fragmentation characteristics. In this last work, glass WCOT columns were used resulting in greatly improved resolution. Little work has been done in general in this area in recent years because of the dominance of HPLC methodology, yet it ought to be possible to devise an analytical scheme to rival this but involving separation of molecular fractions, as stable BDMS or boronate derivatives, by adsorption chromatography (TLC or HPLC) followed by GC-MS on modern WCOT columns of fused silica with thermally stable polar phases.


I.  Determination of Lipid Profiles by Gas Chromatography

The ultimate objective of all analysts is to obtain as much information as possible on the lipid composition of a particular sample in the shortest possible time. Ideally this would mean obtaining a quantitative profile of the lipid classes present in a tissue together with the fatty acid composition of each in a single chromatographic step. This ideal goal cannot yet be attained, but a great deal has been accomplished by separating natural lipid mixtures according to molecular weight by high-temperature GC, principally in the laboratory of Kuksis in Canada but with a significant contribution from Mares in Czechoslovakia, who have both reviewed the topic [507,516,590]. In brief, the lipids are first digested with phospholipase C which converts phosphatidylcholine, lysophosphatidylcholine and sphingomyelin to diacylglycerols, monoacylglycerols and ceramide, respectively. The hydrolysis products are converted to the TMS ether (or related) derivatives, while the cholesterol and free fatty acids also react to form TMS ether and ester derivatives, respectively. Tridecanoin is added as an internal standard for quantification purposes, and the mixture is subjected to GC separation over a large temperature range so that as many as possible of the components are separated. The absolute amounts of the various lipid classes are easily determined, while the proportions of the molecular species give an indication of the chain-length distributions of the fatty acid constituents.

Although such methodology could in theory be applied to any tissue, most work has been done on human plasma lipids and related body fluids such as lymph, as rapid screening procedures here can lead to the diagnosis of disorders of lipid metabolism and can assist in monitoring the effects of clinical therapy. For these purposes, a partially resolved lipid profile may often contain sufficient information, provided that enough data are available on normal and diseased states to enable significant comparisons to be made. In historical terms, the development of the methodology has followed closely behind that for the separation of intact triacylglycerols since these inevitably must be resolved if the procedure is to be of value. Thus, in the pioneering paper in 1967 [513], a short packed column containing a nonpolar methylsilicone phase was employed, whereas nowadays WCOT columns of fused silica would be preferred. Improvements in computerised data handling have contributed greatly to applications involving routine screening.

Those aspects of quantification in high temperature GC described in Section B, including column preparation, injection technique and calibration, are equally apposite here and need not be duplicated. It is, however, worth noting that calibrations should be carried out with standards similar in composition and overall concentration to the corresponding lipids in the tissue under study. For example if this is plasma, the tridecanoin used as an internal standard should be made up to a concentration of 100 μg per mL of chloroform, and 100-200 μL of this is added to 0.25 to 0.5 mL of plasma during extraction of the lipid with chloroform/methanol (see Chapter 2). The lipid extract is then incubated with the phospholipase C of Clostridium welchii, as described in Section C of this chapter, and the products are isolated for analysis. Alternatively [525] -

Whole plasma (0.1 to 1 mL) with ethylenediaminetetraacetate (EDTA) added as an anticoagulant is digested in a stoppered tube with phospholipase C (2 to 4 units) in Tris buffer (17.5 mM; pH 7.3; 4 mL) with 1% calcium chloride (1.3 mL) and diethyl ether (1 mL) for two hours at 30°C with shaking. The reaction is stopped by adding 0.1 M hydrochloric acid (0.2 mL), and the mixture is extracted with chloroform-methanol (2:1, v/v; 10 mL) containing tridecanoin (0.1 to 0.25 mg). After brief centrifugation to separate the layers, the chloroform layer is removed from the bottom of the tube by Pasteur pipette and is taken to dryness in a stream of nitrogen.

The lipids are immediately converted to TMS or BDMS ether derivatives (see Chapter 4 for details), the latter now often being preferred because of their greater chemical stability and good MS fragmentation properties. In some of the early work [529], acetate derivatives were employed, but free carboxylic acid groups are not rendered inert by this means.

With the short packed columns used initially for this work, the resolution obtainable was limited, although adequate for many purposes, as shown in Figure 8.16 [520]. Derivatives of the free acids (two peaks) are the first significant components to emerge, followed in turn by cholesterol, the internal standard tridecanoin, and the diacylglycerol and ceramide derivatives, which run together (six peaks); cholesterol esters (three peaks) are then eluted before the triacylglycerols (five peaks). The relative proportions of ceramides and diacylglycerols were determined using a correction factor, based on the area of the first peak in the group which contains palmitoylsphingosine only, and this gave a reasonable approximation to the true result. The method was found to give accurate results over a range of lipid concentrations, and was capable of a high degree of automation in terms of sample injection and computerised data handling [525]. In a direct comparison of data obtained in this manner with that from automated colorimetric (enzymatic) methods, the GC technique gave results that were more accurate and reproducible, i.e. with a within-day standard deviation of 2.3 mg% for cholesterol and 3.5 mg% for triacylglycerols [520]. The colorimetric methods tended to overestimate these lipids because of interfering chromogens in the plasma extracts. Similarly in comparisons of total phospholipids, phosphatidylcholine, lysophosphatidylcholine and sphingomyelin between results obtained by the GC method and by an alternative TLC separation and phosphorus assay, agreement was excellent for all parameters except for the lysophosphatidylcholine, where variable amounts of endogenous monoacylglycerols in plasma caused errors [521]. The value of this GC procedure in clinical diagnosis has been confirmed by a number of studies of which those cited are representative only [518-520,528,650,651,908].

GC chromatogram of a total lipid profile

Figure 8.16. GC trace of the total lipid profile, following phospholipase C digestion and preparation of the TMS ether derivatives, of plasma from a normolipemic subject [520]. A stainless-steel column (50 cm × 2 mm i.d.), packed with 3% OV-1TM on Gas Chrom QTM (100-120 mesh), was temperature-programmed from 175 to 350°C at 4°C/min with nitrogen as the carrier gas at a flow-rate of 40 mL/min. Tridecanoin is the internal standard. Peak "A" is the TMS ether derivative of palmitoylsphingosine. (Reproduced by kind permission of the authors and of the Journal of Chromatography, and redrawn from the original paper.)


Apart from the limited resolution of this procedure, it was not possible to assay small amounts of endogenous diacylglycerols and monoacylglycerols in plasma as these were masked by similar lipids derived from phospholipase C digestion of the phospholipids. A comparable automated procedure was therefore developed by the Czechoslovakian group in which the phospholipase C digestion step was omitted and only the simple lipids were assayed [595,596]. Here also the methodology has proved its worth in clinical trials [848,849] and in comparisons with alternative methods [593].

WCOT columns of glass or fused silica have so far only been used to obtain plasma lipid profiles to a limited degree [518,528,544,647], but greatly enhanced resolution is obtained and they will surely be employed more often in the future. Figure 8.17 illustrates the separation that can be achieved with TMS ether derivatives of plasma lipids (digested with phospholipase C) on a WCOT column of fused silica coated with an apolar stationary phase, temperature-programmed to 340°C, and with on-column injection [647]. While there is some separation of lower molecular weight components by degree of unsaturation, the chief feature is improved resolution among lipid classes; many more fractions are obtained and for example, free acid and monoacylglycerol derivatives are clearly seen, better resolution of ceramide and diacylglycerols is apparent and there is no overlap of the peaks for cholesterol esters and triacylglycerols. Quantification was comparable to that with packed columns, and indeed the accuracy of the determination of the ratio of phosphatidylcholine to sphingomyelin, an important parameter in clinical research, was certainly greatly superior. In addition, identification of the peaks by mass spectrometry is greatly facilitated when WCOT columns of fused silica are utilised. Unfortunately, on-column injection cannot yet be automated, but one can hope this will come in time.

GC chromatogram of derivatized plasma lipids

Figure 8.17. High-temperature GC of derivatized plasma lipids on a WCOT column (8 m × 0.3 mm) of fused silica, coated with SE-54TM, and temperature-programmed to a maximum of 340°C [647]. Abbreviations: A, TMS ester derivatives of free acids; B, TMS ethers of monoacylglycerols (derived from lysophosphatidylcholine); C, TMS ether of cholesterol; D, tridecanoin (internal standard); E, TMS ether of 16:0-sphingosine (derived from sphingomyelin); F, TMS ethers of diacylglycerols (derived from phosphatidylcholine) and ceramides; G, more TMS ethers of ceramides; H, cholesterol esters; J, triacylglycerols. (Reproduced by kind permission of the authors and of the Journal of Biochemical and Biophysical Methods, and redrawn from the original paper.)


Among other applications of methodology of this kind, the products of hydrolysis of triacylglycerols with pancreatic lipase [628] and of re-esterification of glycerol with long-chain fatty acids [218] have been analysed.



The following abbreviations are employed at various points in the text of these chapters:

amu, atomic mass units; BDMS, tert-butyldimethylsilyl; BHT, 2,6-di-tert-butyl-p-cresol; CI, chemical ionisation; DNP, dinitrophenyl; ECL, equivalent chain length; ECN, equivalent carbon number; EI, electron-impact ionisation; FCL, fractional chain length; GC, gas chromatography; GLC, gas-liquid chromatography; HPLC, high-performance liquid chromatography; IR, infrared; MS, mass spectrometry; NMR, nuclear magnetic resonance; PAF, platelet-activating factor; ODS, octadecylsilyl; TLC, thin-layer chromatography; TMS, trimethylsilyl; UV, ultraviolet.


This document is part of the book Gas Chromatography and Lipids by William W. Christie and published in 1989 by P.J. Barnes & Associates (The Oily Press Ltd), who retain the copyright.

Return to Part 1 of this Chapter

Return to Part 2 of this Chapter

Updated July 8, 2011