William W. Christie

William W. Christie

HPLC SEPARATION of PHOSPHOLIPIDS


Abstract: The principles involved in the selection of solvents and the use of counter ions, together with the choice of column, for the separation of phospholipids by high-performance liquid chromatography (HPLC) are described.


1.  Introduction

In recent years, HPLC methods for separation of lipid classes have gained in popularity, largely at the expense of TLC. The equipment required for the latter is relatively simple and inexpensive and excellent resolution can be achieved, but disadvantages are that large amounts of silica dust are generated (lipid laboratories used to be recognisable by the grey patina that covered every surface), and trace contaminants, including silica and fluorescent dyes, can be introduced into samples. HPLC, in contrast, can deliver clean fractions for further analysis.

One of the most widely used applications of HPLC has been for the analysis of phospholipid classes. Over the years, the methodology has developed to suit the detection techniques available. Thus, in the early days the only convenient detectors were UV spectrophotometers operated at low wavelengths, so it was necessary to find combinations of solvents for the mobile phase that were transparent in the appropriate regions of the spectrum. Modern evaporative light-scattering detectors and mass spectrometric detection have no such limitations. Of course, silica gel was the only stationary phase available initially.

In this short review, I will discuss the properties of some of the mobile phases that have been used to separate phospholipid classes on silica and other stationary phases, the need to use ionic species in the mobile phases, the column types available, a nd lastly some of the problems and opportunities that remain. I do not intend to give recommended "recipes". For a more detailed discussion of specific methods, readers should consult my book (written jointly with Xianlin Han) Lipid Analysis (4th edition) (Oily Press, Bridgwater) or a substantial review article on the topic [1,2].


2. The Basic Solvent Systems

In spite of the detection difficulties, two basic solvent systems were developed for phospholipid separations that still find almost universal application today, i.e. hexane-propan-2-ol-water and acetonitrile-water (sometimes with added methanol) mixtures. The use of the latter was first described in 1976, i.e. acetonitrile-methanol-water (61:21:4 by volume) [3]. In this system, phosphatidylethanolamine (PE) elutes before phosphatidylcholine (PC) and then sphingomyelin, and indeed all the choline-containing phospholipids are well resolved. A further special virtue is that the acidic lipids, such as phosphatidylinositol (PI) and phosphatidylserine (PS) elute ahead of PE. For example, by adding a little sulfuric acid to the mixture (not now recommended from a practical standpoint), phosphatidic acid (PA), cardiolipin, PI and PS were clearly separated from each other before PE emerged.

Mobile phases based on hexane-propan-2-ol-water have also been used in many laboratories since their introduction in 1977 [4]. Again, PE elutes before PC, but the other choline-containing lipids, such as sphingomyelin and PC, tend to be less well resolved. The acidic lipids, such as PI, PS and PA, are separated from each other, but in this instance they emerge between PE and PC. By adding further solvents to the basic mixture or by using gradients, it has been possible to improve the resolution of the choline-containing lipids especially. This system has proved easier to adapt to simultaneous separation of simple lipids and glycolipids than has that based on acetonitrile.

These results are summarised in Table 1. It is evident that the selectivities of the solvents used in the mobile phase can exert a marked effect on the separation of individual phospholipids, and in particular it can change the order of elution of specific components. You will note that both of these solvent systems were in essence first described thirty years ago. A few alternatives have of course been reported in the interval, but the advent of light-scattering detectors has opened up many new opportunities, since the primary limitation is now the volatility of the solvent. There is a host of solvents - ethers, alcohols, ketones, aromatic, and halogen or other hetero-atom containing compounds that should be tried. Appreciable potential must remain to change the selectivity of separations yet further and to enhance the opportunities for isolation of specific phospholipid components. I have neither time nor opportunity to do this, but I hope someone will.

Table 1. The order of elution of phospholipids in mobile phases based on acetonitrile and propan-2-ol.*
Acetonitrile-based Propan-2-ol-based
phosphatidic acid cardiolipin
cardiolipin phosphatidylethanolamine
phosphatidylinositol phosphatidylinositol
phosphatidylserine phosphatidylserine
phosphatidylethanolamine phosphatidic acid
phosphatidylcholine phosphatidylcholine
sphingomyelin sphingomyelin
lysophosphatidylcholine lysophosphatidylcholine
* Note there may be some modification to the order given (especially of cardiolipin), depending on the nature of other solvents and any ionic species in the mobile phase.


3.  Counter Ions

Phospholipids are ionic molecules and require a counter ion in solution. If they are obtained from tissues by a conventional chloroform-methanol extraction, including a wash with sodium chloride solution, sodium will be the predominant counter ion. However, special precautions may be necessary to ensure that small amounts of other ions do not remain, since different ionic forms of the acidic phospholipids may have different mobilities on chromatography. One practical method of avoiding difficulties during HPLC is to add counter ions or acids to the mobile phase. Sulfuric and phosphoric acids have often been used, but apart from dissolving HPLC equipment they will bring about complete destruction of any plasmalogens present. Most analysts now add organic buffers, for example with triethylamine and acetic or formic acids, which are relatively innocuous, and appeared to ensure well-shaped peaks with difficult analytes such as PI. In spite of the presence of ionic species in the mobile phase, it is still possible to use evaporative light-scattering detection (not with inorganic ions). This was also possible with mobile phases containing ammonia [5], when such addition again affected the selectivity of the separation. On the other hand, column life may greatly reduced because silica dissolves at a high pH. Once more, there must be scope to test a range of ionic species, both organic and inorganic, at controlled pH values for their suitability as counter ions. There may also be opportunities to change the selectivity of the mobile phase to effect specific separations, especially of the acidic phospholipids.

In spite of all we have learned, I am not yet convinced that we have found an ideal combination of solvent and counter ions, especially for the resolution of acidic phospholipids such as PS.


4.  The Column and Stationary Phase

For a great part of the published work on phospholipid separations, silica gel has been the stationary phase employed. It is a porous solid in which the surface area is inversely related to the size of the pores in the particles. The adsorptive properties are dependent on hydroxyl or silanol groups attached to the surface. In addition, water of hydration is present in a strongly bound layer and then in one or more layers that are more loosely held on the surface. Different brands of silica gel can vary greatly in their properties, and this may in part be due to the content of extraneous metal ions. During prolonged use, the adsorptivity of a column can diminish as polar impurities from lipid extracts accumulate at active sites. While this activity can be restored by careful washing procedures, it is better to add appropriate ionic species to mobile phases for phospholipid analysis as a preventative measure.

The degree of hydration is especially important, and can change markedly during elution, with consequences on the separations unless some water is present in the mobile phase. This is especially important in the analysis of less polar lipids.

Experienced analysts gain a "feel" for the properties of silica gel that is not easy to express in scientific terms, and they can then find ways to circumvent the problems associated with the heterogeneous nature of the surface. Silica gel is relatively inexpensive and robust, and it can give excellent results. It will undoubtedly continue to be used, especially in preparative applications. However, it would obviously be preferable to have a uniform chemically defined surface on a stationary phase. To this end, several phases of a polar nature have been manufactured, consisting of substituent groups, such as diol, nitrile, nitro, methyl- or phenylcyano, and phenylsulphonate, bonded chemically via a short alkyl bridge or spacer to a silanol support. The nature of the bond between the support and the organic group is important in that it determines the stability of the phase to changes in the pH of the mobile phase. Partial polymerization and cross-linking of the silanols are used to further stabilize them, although some residual silanol groups inevitably exert some influence on separations. In practise, bonded phases equilibrate much more rapidly with the mobile phase, especially in gradient applications, and they tend to give much more reproducible separations.

cartoonThe size of the particles is obviously important and 5 m silica gel has become the standard for most analytical purposes, although shorter columns packed with 3 m particles are being used increasingly. Column dimensions and physical characteristics are important in determining the absolute quality of resolution attainable, but they do not affect the selectivity of separation so I do want to dwell on them at length. In brief, I know of no study that compares some of the commercial cartridge systems with more conventional columns for phospholipid separations, although the former are claimed by manufacturers to give better resolution in general. Reducing the internal column diameter of columns from 4.6 mm to 3.2 mm should enable solvent consumption to be reduced by half, and increase sensitivity by giving sharper peaks, as well as improving resolution. Even narrower bore columns (1 to 2 mm i.d.) are often preferred, especially for connection to mass spectrometry interfaces. Again, there appear to be no systematic comparison studies. Columns fabricated from glass-lined stainless steel are available, and might be helpful in reducing tailing with ionic phospholipids; they should certainly be tried.

Until relatively recently, the bonded phase closest to silica gel in its properties was a "diol" phase, consisting of 1,2-dihydroxypropyl moieties linked covalently to silica gel, and these have been widely used in HPLC of phospholipids. On the other hand, it is now apparent that PVA SilTM columns (YMC Corp., Japan), which have polymerized polyvinyl alcohol bond to the silica surface, are an excellent replacement for silica and are better than diol columns [6]. The whole surface is covered and deactivated so that the mobile phase and analytes interact with a uniform layer of hydroxyl groups only. Components elute with less polar solvents than is required with silica.

A cyanopropyl-bonded phase with an acetonitrile-based mobile phase gave an interesting separation of acidic phospholipids, which eluted in the order phosphatidylglycerol, PI, cardiolipin and phosphatidylserine, before PE, PC and so forth emerged [8]. Again, the mobile phase was of much lower polarity than would have been required for silica gel. Cyanopropyl columns can also give excellent resolution of simple lipids, but I have observed that such columns from different manufacturers can vary greatly in their properties. There are many other bonded stationary phases that are available from commercial sources and do not appear to have been tried for phospholipid separations. For example, there are several alternative phases with phenylcyano, carboxymethyl and nitrophenyl residues.

Some excellent separations have been achieved with acidic and basic stationary phases. For example, A spectacular change in selectivity is observed with an aminopropyl-bonded phase [7]. In this instance PC and the other choline-containing phospholipids are eluted ahead of phosphatidylethanolamine (PE), rather than the reverse, as is usually the case, and under relatively mild conditions. Although it is now known that acidic lipids are recovered only with difficulty from such columns, this is not a problem in many micro-preparative applications and could be put to good use in many circumstances.

A further type of bonded-phase to have been used in phospholipid separations has benzene-sulfonic acid as the functional group [9]. In this instance, excellent resolution of each of the ethanolamine and choline-containing lipids was achieved. As the stationary phase is itself ionic, it is possible that the need for counter ions in the mobile phase is lessened. A zwitterionic stationary phase (Merck ZICTM-HILIC) has recently been utilized to give remarkable separations of a wide range of phospholipid classes, including the acid PS and PI, and even their plasmalogen forms [10].


5.  Conclusions

It should be recognised is not always necessary to have a perfect analytical system that resolves every single lipid class. For example, if the analyst is concerned with the properties of a specific phospholipid class, it may simply be necessary to optimise the elution scheme so that the compound of interest is isolated in a relatively pure state; resolution of other components can be ignored. A wider knowledge of factors controlling the separations is then invaluable. However, an ideal analytical system should give sharp well-resolved peaks for all the main phospholipids in tissue extracts, especially the acidic and choline-containing components, and it should be stable and reproducible for months in continuous use. In addition, it should be adaptable to the simultaneous analysis of the simple lipids and of glycolipids, such as those found in plant tissues. The signs are that we may be getting there.


References

  1. Christie, W.W. and Han, X. Lipid Analysis - Isolation, Separation, Identification and Lipidomic Analysis (4th edition), 446 pages (Oily Press, Bridgwater, U.K.) (2010) - www.pjbarnes.co.uk/op/la4.htm.
  2. Christie, W.W. Separation of phospholipid classes by high-performance liquid chromatography. In: Advances in Lipid Methodology - Three, pp. 77-107 (ed. W.W. Christie, Oily Press, Dundee) (1996).
  3. Jungalwala, F.B., Evans, J.E. and McCluer, R.H. High-performance liquid-chromatography of phosphatidylcholine and sphingomyelin with detection in the range of 200 nm. Biochem. J., 155, 55-60 (1976).
  4. Hax, W.M.A. and Geurts van Kessel, W.S.M. High-performance liquid chromatographic separation and photometric detection of phospholipids. J. Chromatogr., 142, 735-741 (1977).
  5. Abidi, S.L. High-performance liquid chromatography of phosphatidic acids and related polar lipids. J. Chromatogr., 587, 193-203 (1991).
  6. Christie, W.W., Gill, S., Nordbck, J., Itabashi, Y., Sanda, S. and Slabas, A.R. New procedures for rapid screening of leaf lipid components from Arabidopsis. Phytochemical Anal., 9, 53-57 (1998).
  7. Carunchio, V., Nicoletti, I., Frezza, L. and Sinibaldi, M. High-performance liquid chromatography of phospholipids on chemically bonded silica gel. Annali di Chimica, 74, 331-339 (1984).
  8. Samet, J.M., Friedman, M. and Henke, D.C. High-performance liquid chromatography separation of phospholipid classes and arachidonic acid on cyanopropyl columns. Anal. Biochem., 182, 32-36 (1989).
  9. Gross, R.W. and Sobel, B.E. Isocratic high-performance liquid chromatography separation of phosphoglycerides and lysophosphoglycerides. J. Chromatogr., 197, 79-85 (1980).
  10. Rezanka, T. Siristova, L., Matoulkova, D. and Sigler, K. Hydrophilic interaction liquid chromatography: ESI-MS/MS of plasmalogen phospholipids from Pectinatus bacterium. Lipids, 46, 765-780 (2011); DOI: 10.1007/s11745-011-3556-y).

This article has been updated appreciably from two earlier papers (now amalgamated) by the author that first appeared in Lipid Technology (Christie, W.W. Lipid Technology, 5, 121-123 (1993); Christie, W.W. Lipid Technology, 6, 17-19 (1994)) (by kind permission of P.J. Barnes & Associates (The Oily Press Ltd), who retain the copyright to the original articles).


W. Christie

James Hutton Institute (and Mylnefield Lipid Analysis), Invergowrie, Dundee (DD2 5DA), Scotland.

Author Updated: August 4th, 2011 Credits/disclaimer AOCS