Application of inductively coupled plasma mass spectrometry to phospholipid analysis

Abstract

Phospholipids are the main constituents of membranes in all types of prokaryotic and eukariotic cells. Due to their complexity and heterogeneity in biological samples, qualitative and quantitative analyses of membrane phospholipids in cellular extracts represent major analytical challenges, mainly due to suitable and sensitive detection methods. The inductively coupled plasma mass spectrometer (ICP-MS) is a suitable detector for selective determination of phospholipids as they all contain phosphorus. Phospholipids are extractable with organic solvents, therefore liquid chromatography with an organic mobile phase was used for separation of different lipid species. Solvent load to the plasma was reduced by splitting the mobile phase prior to reaching the nebulizer, by chilling the spray chamber to −5 °C and by optimisation of carrier gas flow for maximum condensation of organic vapours. Despite desolvation, oxygen was added to prevent carbon deposition on interface cones. To reduce polyatomic interferences at m/z ratio 31 (e.g.³¹CH₃O⁺) and to improve detection limits, helium was used as a collision gas. The achieved absolute detection limits were between 0.21 and 1.2 ng of phosphorus and were superior to those obtained by an evaporative light scattering detector, which provides an alternative detection system for lipid analysis. The usefulness of the developed method was demonstrated by analysis of lipid extracts from the yeast Saccharomyces cerevisiae.

---

Introduction

Phospholipids (PL) are complex lipids, which contain as their backbone glycerol esterified in the sn-3 position with a phosphate residue. The glycerol-3-phosphate is esterified at its sn-1 and sn-2 positions with fatty acids (typically C₁₆ to C₁₈ carbon atoms), and at its phosphoryl group to an alcohol X, to form various phospholipids as shown in Fig. 1.¹ Some of the alcohol residues are neutral (e.g. inositol) or positively charged at cellular pH (ethanolamine, choline), leading to negatively charged or neutral phospholipids due to the negatively charged phosphate residue. The amphipathic character of the PL molecules forces them spontaneously into bilayers in aqueous systems. This tendency to form bilayers is the basis of all cellular membranes in nature.

Fig. 1 Chemical structures of most common phospholipids and their abbreviations* (R₁ and R₂ are different fatty acids groups).¹

Phospholipid classes are characterized by the alcohol residue esterified to the phosphate group. The basic PL structure is phosphatidic acid, or 1,2 diacylglycerol-3-phosphate. For example, phosphatidylcholine (PtdCho) describes a phospholipid harboring choline (N,N,N-trimethylaminoethanol-2) esterified to the phosphate residue. The diversity of phospholipid molecular species is brought about by the diversity of fatty acids esterified to the glycerol backbone. Yeast, for instance, produces some 20 different molecular species of phosphatidylcholines.

Pure synthetic phospholipids are chemically defined and they can be designated by their systematic name. For example, abbreviation “DOPC” stands for 1,2-dioleoyl-phosphatidylcholine.²

A major goal of research on phospholipids is to understand the significance of these compounds in the functioning of cellular membranes. The majority of studies are based on HPLC analysis of lipid extracts. Since phospholipids are soluble only in organic solvents, normal phase liquid chromatography is the most frequently used separation method.³ The columns are packed either with silica-, amino- or with diol-type stationary phases and all have in common, that they separate only different chemical classes of phospholipids, rather than molecular species differing in the fatty acid composition.⁴ Mobile phases commonly employed are usually mixtures of different solvents, such as methanol, acetone, hexane, chloroform, tetrahydrofurane, 2-propanol or acetonitrile. For better separation, gradient elution should be applied, starting from a weak (nonpolar) mobile phase to a strong (more polar) mobile phase. However, reverse phase liquid chromatography can also be used for their separation.^5,6

Due to the absence of intrinsic molecular properties and the heterogeneity of the substance classes, detection of phospholipids is a major analytical problem. Different detection systems have been described for phospholipid analysis.⁷ The refractive index detection suffers from poor detection limits and is only useful for simple mixtures, since gradient elution cannot be applied due to a baseline drift.⁸ Reported detection limits are approximately 20 ng of phosphorus.⁹ Because many solvents used for liquid chromatography are non-transparent in the UV range (200–210 nm, useful for detection of phospholipids) these detectors have serious constraints with respect to the selection of mobile phases.¹⁰ To overcome lack of chromophores in the phospholipid molecules, post-column derivatisation was applied by Rastegar et al.⁶ They derivatised the sample with Naproxen and derivatives were subjected to HPLC with UV absorption measurements at 230 nm. The reported detection limit was 0.3 ng (expressed as phosphorus). With the introduction of the evaporative light scattering detector, analysis of all lipid species became easier.^8,11,12 However, this detector suffers from a limited linear range, with 400 ng of phosphorus as a lowest sample amount giving linear response.^13,14 Lower sample quantities can be detected (10 ng of phosphorus), however the calibration response is not linear.¹⁵ Another possibility for detection of phospholipids is hyphenation of HPLC to mass spectrometry, giving better selectivity towards the compounds of interest.¹⁶ With electrospray ionisation mass spectrometry a quantification limit for phosphatidylserine was reported to be 0.05 ng of phosphorus.¹⁷ Electrospray ionisation used together with tandem mass spectrometry can be used to obtain molecular information about the phospholipids without prior chromatographic separation.^18–20

It is a common property of all described detectors that their responses are depending on the molecular structure of species being analysed and, therefore, standard compounds are required for their quantification. To overcome this problem, an inductively coupled plasma mass spectrometer (ICP-MS) can be used as a detection technique, since its response is theoretically dependant only on the element in question. The hyphenation of ICP-MS to LC for phospholipid analysis was first described by Axelsson et al.²¹ The authors have reported successful normal phase HPLC separation and ICP-MS detection of three standard compounds belonging to PC, PE and PG phospholipid classes. However, authors did not provide any details on analysis procedure and they did not test this method on real complex biological samples. Additionally they used a low flow membrane desolvation unit for removal of interfering organic solvents, which is still not widely accessible in laboratories. It should be noted, that the determination of phosphorus and its compounds by an ICP-MS is not an easy task because phosphorus has a high ionisation potential and consequently is poorly ionised in the plasma. Additionally, it suffers from polyatomic interferences at m/z ratio 31 when low mass resolution instruments are used. Therefore not only ³¹P⁺ ions are measured, but also polyatomic ions such as ¹²C¹H₃¹⁶O⁺.^22,23 As a result, poorer detection limits are achieved when carbon compounds are in the sample matrix. Jiang and Houk²⁴ reported a decrease of the mass spectrometer response with increasing concentration of organic modifiers in mobile phase of the LC-ICP-MS system when polyphosphates and adenosine phosphates were analysed.

In this work we present a detailed study of processes occurring during the determination of phospholipids by a quadrupole ICP-MS instrument equipped with a conventional double pass spray chamber for desolvation. Additionally, we are demonstrating its use for the characterization and detection of phospholipids from yeast lipid extracts by applying a modified chromatographic separation published by Sas et al.³

Experimental

Reagents and sample preparation

Acetone, hexane and triethylamine all of p.a. purity were purchased from Fluka (Buchs, Switzerland). Methanol, acetic acid (96%), chloroform and MgCl₂ all of p.a. purity were purchased from Merck (Darmstadt, Germany). Chemically defined phospholipids were purchased from Avanti (Alabaster, AL, USA): 1,2-dioleoyl-phosphatidic acid monosodium salt (C₃₉H₇₂O₈PNa, DOPA), 1,2-dioleoyl-phosphatidylcholine (C₄₄H₈₄NO₈P, DOPC), 1,2-dioleoyl-phosphatidylethanolamine (C₄₁H₇₈NO₈P, DOPE), 1,2-dioleoyl-phosphatidylglycerol sodium salt (C₄₂H₇₈O₁₀PNa, DOPG), 1,2-dioleoyl-phosphatidylserine sodium salt (C₄₂H₇₇NO₁₀PNa, DOPS) and phosphatidylinositol sodium salt isolated from bovine liver (C₄₇H₈₂O₁₃PNa, PI).

The standard mixture of six phospholipids was prepared by diluting each phospholipid standard in a chloroform/methanol mixture (2/1, v/v). The concentrations of each expressed as phosphorus were as follows: 3.3 mg l⁻¹ DOPA, 2.9 mg l⁻¹ DOPG, 2.7 mg l⁻¹ PI, 3.1 mg l⁻¹ DOPE, 3.0 mg l⁻¹ DOPS and 2.9 mg l⁻¹ DOPC. Approximately 1 g of wild-type yeast cells (Saccharomyces cerevisiae W303, MATα, leu2, ura3, his3, ade2, trp1) were resuspended in 10 ml of deionised water and cells were disrupted in a glass bead homogeniser (B. Braun Melsungen, Germany). Then 80 ml of chloroform/methanol mixture (2/1, v/v) was added and the suspension was stirred for 30 min at room temperature, additionally 20 ml of a MgCl₂ solution (0.034%) was added for phase separation and it was again stirred for 30 min. After centrifugation for 5 min, the organic phase was transferred to a round bottom flask and evaporated to dryness. Lipids were dissolved in 2 ml of chloroform/methanol mixture (2/1, v/v) and the solution was transferred to a HPLC vial. Solutions were stored at −20 °C.

Chromatographic system

HPLC separations were carried out using an Agilent 1100 chromatographic system (Waldbronn, Germany) equipped with a thermostated autosampler (variable injection loop 0–100 µl), YMC-Pack Diol-120 column (250 × 4.6 mm, 5 µm) (Kyoto, Japan) maintained at 50 °C and a flow rate of 0.6 ml min⁻¹. The composition of mobile phase A was acetone/hexane/acetic acid/triethlyamine—900/70/14/2 (v/v) and the composition of mobile phase B was methanol/hexane/acetic acid/triethlyamine—900/70/14/2 (v/v). The following gradient elution program was used: 95% of A at 0 min, 82% of A at 40 min, 55% of A at 42 min, 40% of A at 44 min, 40% of A at 49 min, 95% of A at 49.5 min and 95% of A at 58 min.

Detection system

The 0.6 ml min⁻¹ flow from the HPLC column was split to approximately 130 µl min⁻¹ before reaching the 7500c ICP-MS (Agilent, Waldbronn, D) equipped with a PFA microconcentric nebulizer, a Scott double pass spray chamber and an octopole reaction cell (ORC). Since constant intake of sample to nebulizer was desired, self-aspiration was minimised by a 69 cm long capillary (0.127 mm id) mounted between splitter and nebulizer. According to the desired flow, appropriate resistance of splitter drain was achieved by a 16 cm long capillary (0.127 mm id). To prevent deposition of carbon on the interface cones, an optional gas (20% oxygen in argon) was applied through a T-piece connecting spray chamber and torch (narrow injection tube—1.5 mm). Since added oxygen promotes corrosion of interface cones, a platinum sampler cone was used. The skimmer cone was made from nickel. The detection was carried out by recording m/z ratio 31 at scan rate of 0.3 s per point.

The system was optimised by pumping mobile phase A containing 2 mg l⁻¹ of phosphorus as a DOPE. During the tuning procedure, such conditions were chosen that the signal at m/z 31 was as high as possible and that the mass peak at m/z 31 and neighbouring peaks m/z 30 and m/z 32 were clearly separated. Further optimisation of detector response was performed in flow injection analysis mode with mobile phase A as a carrier solution and by injection of 2 µl of 90 mg l⁻¹ of phosphorus as a DOPE diluted in mobile phase A. To evaluate results, signal to background ratios (SB) were calculated as a quotient between height of phosphorus signal and height of background signal. After having optimal conditions for detection of phosphorus, helium was added as a collision gas to reduce the background signal.

For the detection of the phospholipids the following optimised conditions were used: Plasma gas 15 l min⁻¹, auxiliary gas 1.0 l min⁻¹, carrier gas 0.50 l min⁻¹, optional gas flow rate 24% (of carrier gas flow rate), rf power 1600 W (reflected power ≤10 W), ORC gas (helium) 4.0 ml min⁻¹, spray chamber temperature −5 °C and sample depth (torch-interface distance) 10 mm. All chromatograms were smoothed before integration.

Results and discussion

Optimisation of detector response

The detection of phospholipids by ICP-MS suffers from two problems. The first one is the stability of the plasma in the presence of organic solvents. Since phospholipids are not soluble in aqueous solutions, organic solvents have to be used for sample extraction and chromatographic separation. Despite a modified sample introduction system (cooled spray chamber and torch with narrow injector tube), intake rates should be low, therefore splitting of mobile phase before reaching the detector was utilised. Additionally, oxygen has to be added to carrier gas to avoid carbon deposition on interface cones. The second problem is the formation of carbon-based polyatomic ions in the plasma with m/z ratio 31 (³¹CH₃O⁺).^22,23 Their formation is unavoidable, since organic solvents are in the mobile phase. Therefore, the optimisation procedure should at the same time maximise the phosphorus signal and minimise the intensity of the polyatomic ions, to avoid high background in the chromatogram.

The first parameter we optimised was the rf power. We tested powers from 1200 to 1600 W by recording flow injection signal. The experiment was performed at three different sample depths (6, 9 and 12 mm) at constant carrier gas flow rates of 0.68 l min⁻¹ and without any make-up gas flow. As expected, when highest rf power was used, SB ratio was highest at all three tested sample depths, since high temperature is required to decompose organic matrix and to improve ionisation of phosphorus. According to this results rf power of 1600 W was chosen for further optimisation and routine measurements.

In order to find optimum carrier and make-up gas flows, several combinations of these gas flows were tested by flow injection technique at 1600 W of rf power. The whole experiment was conducted at three different sample depths (6, 9 and 12 mm). Carrier gas flows were tested in the range from 0.2 to 0.70 l min⁻¹ at four different make-up gas flows (0, 0.10, 0.15 and 0.20 l min⁻¹). The signal to background ratios are not improved by the addition of a make-up gas. The usage of a make-up gas only shifts the optimal carrier gas flow rate to lower values. That means the signal to background ratio is only depending on total flow of gases through the spray chamber and the most important process in spray chamber is condensation of organic vapours. When too high total gas flows are applied, the residence time of the aerosol in the spray chamber is shorter, the amount of condensed organic matter is lower and therefore the background derived from organic vapours in the form of carbon based polyatomic ions is increased. However, when too low flow rates are applied, condensation is efficient, but transport of aerosol particles is compromised and SB ratios are decreased. This fact was also confirmed by observation of the colour of the plasma and the required amount of oxygen to prevent carbon deposition. At higher carrier or make-up gas flow rates the plasma was greener and more oxygen was needed. On the basis of these results it was decided not to use any make-up gas.

The optimum sampling point of ions in the plasma is of course influenced by the gas flows and sample depth. Therefore both parameters were optimised, while no make-up gas flow was used. The tested carrier gas flow rates were in the range from 0.35 to 0.55 l min⁻¹ and sampling depths were in the range from 5 to 11 mm. The results are presented in Fig. 2 as a two-dimensional plot, where colour intensity is representing SB ratio. The highest SB ratios were observed when carrier gas flow rate of 0.50 l min⁻¹ was applied and sample depths were from 5 to 9 mm. These conditions are therefore giving highest sensitivity for detection of phospholipids. However, there is an empirical rule that the intensive green coloured zone on the front end of the plasma should end before reaching the tip of the sampler cone.²⁵ This requires higher amounts of oxygen and means shorter life time of the cones. As a compromise between good sensitivity and system robustness, a carrier gas flow rate of 0.50 l min⁻¹ and sample depth of 10 mm were chosen for routine work.

Fig. 2 Signal to background ratios (SB) at different carrier gas flow rates (x-axis) and different sample depths (y-axis) (1600 W, no make-up gas).

Despite optimal conditions for the desolvation process in the spray chamber, the background signal was still high and was negatively influencing the limits of detection. Another negative consequence was noticeable drift of baseline in the chromatogram due to change of mobile phase composition during chromatography. This problem was solved with helium as a collision gas. We explored helium flow rates through the ORC in the range from 0 to 6 ml min⁻¹. At each tested helium flow rate, the flow injection signal was recorded and SB ratios were calculated (Fig. 3). As can be seen, helium improved SB ratio at flow rates between 4 and 6 ml min⁻¹. For the practical applications a flow rate of 4 ml min⁻¹ was applied, giving five-times higher SB ratio than without using any collision gas. At higher helium flow rates, suppression of both phosphorus and background signal was too strong and the noise of the signal was hindering improved SB ratios. At the same time, background was efficiently reduced for about two orders of magnitude, minimising the problem of baseline drift during the chromatography.

Fig. 3 Phosphorus signal (left y-axis, full line), background signal (left y-axis, dotted line) and signal to background ratio (SB) (right y-axis, full bold line) at different helium flow rates through the octopole reaction cell.

Chromatographic separation

Successful chromatographic separation of six standard phospholipids was achieved by utilisation of modified conditions described by Sas et al.³ A typical chromatogram is presented in Fig. 4. Almost all six phospholipid standards were base-line separated. However, when a yeast lipid extract was analysed (Fig. 5), a splitting of the largest signal at ∼36–39 min belonging to PC class was observed. All attempts (changing the gradient) to improve the chromatographic resolution of these two signals were unsuccessful. Only the higher signal at the low retention time side matched the retention time of DOPC. There are two possible explanations. The first could be that we separated various species of PC containing fatty acids with different chain length. The second possibility is that the compound eluting at ∼38 min (marked with a ?, Fig. 5) corresponds to another class of yeast phospholipids that have until now not been identified. In that case a better separation is essential for accurate quantification and more standard compounds should be tested for identification via matching of retention times. Another possible solution would be application of tandem molecular mass spectrometry for identification of the unknown signal. Besides the split signal of PC, eight chromatographic signals were obtained in the chromatogram of the yeast lipid extract. Four of them were identified as PA, PI, PE, PS, and PC, but four peaks were unclassified. The signals at retention time ∼5 and ∼57 min are related to the mobile phase. The signal at ∼5 min is coming from the solvent, since its intensity is correlating with the amount of injected solvent. This signal might derive from some polyatomic ions in the solvent mixture or it is belonging to some impurities in one of the solvents. The origin of the peak at ∼57 min could also be attributed to the mobile phase. Besides these two signals deriving from the chromatographic system, we clearly found two unknown compounds that did not match the retention times of any of the phospholipid standards available to us. While X₁ (Fig. 5) is a non-identified phosphorous containing compound, we can only speculate about the identity of “X₂”. Its retention time is similar to the retention time of orthophosphate. Therefore, we cannot affirm that it belongs to phospholipids, in spite of the fact, that the presence of highly polar orthophosphate in the chloroform/methanol mixture is very unlikely.

Fig. 4 Separation of six chemically defined phospholipids in standard mixture on YMC Pack Diol-120 column (250 × 4.6 mm, 5 µm) with ICP-MS detection of phosphorus at m/z ratio 31 (5 µl injected, 0.6 ml min⁻¹, each peak corresponds ∼15 ng of phosphorus).

Fig. 5 Separation of phospholipid classes in yeast lipid extract on YMC Pack Diol-120 column (250 × 4.6 mm, 5 µm) with ICP-MS detection of phosphorus at m/z ratio 31 (2 µl injected, 0.6 ml min⁻¹).

Quantification

Different volumes (0.5–20 µl) of standard mixtures of the six phospholipids were injected for construction of calibration curves and for the determination of the detection limits. A good linearity with correlation coefficients (R) equal or above 0.9997 was obtained for all tested compounds and it was generally covering range from ∼1.5 to ∼60 ng of phosphorus (Table 1). The exceptions were compounds DOPA and DOPG which had linear response only up to ∼15 ng of phosphorus due to poor chromatographic separation at higher injected amounts, and compounds DOPE and DOPS, with linear range starting at 3.0 ng of phosphorus. Limits of detection were estimated as peak height equalling three times the baseline noise and are also presented in the Table 1. As expected, limits of detection were lowest for those compounds having well defined peak shapes (DOPA 0.36 ng and DOPG 0.21 ng of phosphorus) and highest for compounds having broad peaks (DOPE 1.2 ng and DOPS 1.2 ng of phosphorus). Table 1 additionally contains data on reproducibility of peak areas at the lowest point of the calibration curve. Compared to performances of other detectors (especially evaporative light scattering detector) that are routinely used for phospholipids analysis, an ICP-MS gives superior detection limits. However, it cannot be compared to molecular mass spectrometry, since it does not give any structural information.

Table 1 Calibration parameters (expressed as a mass of phosphorus)

Compound	Retention time/min	Calibration curve	R	Linear range/ng	LOD/ng	Reproducibility^a (%)
a At lowest point of calibration curve.
DOPA	6.7	A = 18000 × m − 830	0.9998	1.6–16	0.36	±6
DOPG	7.8	A = 23400 × m − 5650	0.9997	1.4–14	0.21	±5
PI	14.2	A = 25000 × m − 850	0.9999	1.4–55	0.54	±7
DOPE	18.3	A = 21000 × m − 10400	0.9999	3.0–61	1.2	±7
DOPS	28.1	A = 16500 × m − 18600	0.9998	3.0–59	1.2	±16
DOPC	35.9	A = 19500 × m − 230	0.9999	1.5–59	0.50	±14

---

The usefulness of the developed method for analysis of phospholipids was demonstrated on a complex lipid extract from yeast as presented in Fig. 5. Each identified compound was quantified by using calibration curves given in Table 1. The results are presented in Table 2 as peak areas and as calculated masses and concentrations of each identified compound in the sample extract. It should be noted, that all masses and concentrations are expressed as phosphorus and that peak coeluting with chemical class of PC was integrated and considered as belonging to PC class. According to the extraction method, calculating the concentration of each class of phospholipids in yeast is impossible, since many steps in the extraction procedure are not quantitative. If such data were required, then the extraction procedure should be modified or an internal standard should be added before the extraction. Since properties of yeast sample are defined by its phospholipid composition, relative amounts of identified compounds were calculated and are given as percentages in Table 2.

Table 2 Peak areas, calculated masses and concentrations, relative amounts of identified compounds and semi-quantitatively determined relative amounts of all phospholipids in yeast lipid extract (all values are expressed as phosphorus)

Class	Peak area/10^3 units	Mass/ng	Concentration/mg l^−1	Relative amounts^a (%)	Semi-quantiative relative amounts^b (%)
a Relative amounts of identified compounds.b Relative amounts determined by semi-quantitative procedure.
PA	65.5	3.7	0.74	1.6	1.2
PI	783	31	6.3	13	15
PE	1440	69	14	29	27
PS	150	10	2.0	4.3	2.8
PC	2380	120	24	51	44
X1	400	—	—	—	7.6
X2	45.6	—	—	—	0.9

---

Theoretically, the response of an ICP-MS is element depending, meaning it is the same for all compounds regardless to their structure. The determined response factors, which are part of calibration curves in Table 1, have values from 16500 to 25000 peak area units per ng of phosphorus. Deviations from theory are expected, since we used a gradient elution program, giving a different matrix composition for each compound resulting in different nebulisation efficiencies. Moreover, the changing matrix composition might also change the ionisation efficiency of phosphorus in the plasma. Therefore simplification of quantification procedure by using a calibration curve based only on one compound should be used with critical evaluation of obtained results and it already belongs in the field of semi-quantitative analysis. In cases, when we are interested only in obtaining approximate ratios between classes of phospholipids in the sample, only peak areas without any calibration can be used with an awareness of possible errors, of course. To show the usefulness of such a quick semi-quantitative analysis, a yeast lipid extract was also treated in that way. Peak areas of all peaks found in the chromatogram were summed, their relative amounts were calculated and are also presented in Table 2. Compared to literature data² one can clearly say that this semi-quantitative approach gives good agreement.

Acknowledgements

Support by the Ministry of Education, Science and Sports of the Republic of Slovenia for a travel grant to M.K. and the Austrian Science Fund FWF (SFB Biomembranes F706) to S.D.K. is gratefully acknowledged.

References

1. D. Voet and J. G. Voet, Biochemistry, Wiley, 1995, pp. 280–282.
2. Phospholipids Handbook, ed. G. Cevc, Marcel Dekker, 1993, ch. 1 and 2, pp 23–38.
3. B. Sas, E. Peys and M. Helsen, J. Chromatogr. A, 1999, 864, 179–182.
4. R. M. Sheeley, W. J. Hurst, D. M. Sheeley and R. A. Martin, Jr, J. Liq. Chromatogr., 1987, 10, 3173–3181.
5. G. M. Patton, J. M. Fasulo and S. J. Robins, J. Lipid Res., 1982, 23, 190–196.
6. A. Rastegar, A. Pelletier, G. Duportail, L. Freysz and C. Leray, J. Chromatogr. A, 1990, 518, 157–165.
7. S. L. Abidi and T. L. Mounts, J. Chromatogr. B, 1995, 671, 281–297.
8. J. L. Robinson and R. Macrae, J. Chromatogr. A, 1984, 303, 386–390.
9. M. Grit, D. J. A. Crommelin and J. Lang, J. Chromatogr. A, 1991, 585, 239–246.
10. P. Van Der Meeren, J. Vanderdeelen, M. Huys and L. Baert, J. Chromatogr. A, 1988, 447, 436–442.
11. A. Stolyhwo, H. Colin, M. Martin and G. Guiochon, J. Chromatogr. A, 1984, 288, 253–275.
12. T. Seppänen-Laakso, I. Laakso, H. Vanhanen, K. Kiviranta, T. Lehtimäki and R. Hiltunen, J. Chromatogr., B, 2001, 754, 437–445.
13. F. D. Conforti, C. H. Harris and J. T. Rinehart, J. Chromatogr., 1993, 645, 83–88.
14. W. S. Letter, J. Liq. Chromatogr., 1992, 15, 253–266.
15. B. J. Stith, J. Hall, P. Ayres, L. Waggoner, J. D. Moore and W. A. Shaw, J. Lipid Res., 2000, 41, 1448–1454.
16. H. Kim and N. Salem, Jr., Anal. Chem., 1987, 59, 722–726.
17. E. Hvattum, Å. Larsen, S. Uran, P. M. Michelsen and T. Skotland, J. Chromatogr. B, 1998, 716, 47–56.
18. B. Brügger, G. Erben, R. Sandhoff, F. T. Wieland and W. D. Lehmann, Proc. Natl. Acad. Sci. USA, 1997, 94, 2339–2344.
19. R. Schneiter, B. Brügger, R. Sandhoff, G. Zellnig, A. Leber, M. Lampl, K. Athenstaedt, C. Hrastnik, S. Eder, G. Daum, F. Paltauf, F. T. Wieland and S. D. Kohlwein, J. Cell Biol., 1999, 146, 741–754.
20. M. Koivusalo, P. Haimi, L. Heikinheimo, R. Kostiainen and P. Somerharju, J. Lipid Res., 2001, 42, 663–671.
21. B.-O. Axelsson, M. Jörnten-Karlsson, P. Michelsen and F. Abou-Shakra, Rapid Commun. Mass Spectrom., 2001, 15, 375–385.
22. T. W. May and R. H. Wiedmeyer, At. Spectrosc., 1998, 19, 150–155.
23. D. R. Bandura, V. I. Baranov and S. D. Tanner, Anal. Chem., 2002, 74, 1497–1502.
24. S. Jiang and R. S. Houk, Spectrochim. Acta, 1988, 43B, 405–411.
25. D. Potter and E. McCurdy, Agilent ICP-MS Journal, 2002, 11, 2.

---