- Made available online as an Accepted Preprint 5 December 2008
- Accepted Preprint first posted online on 5 December 2008
How to study lipidomes
- (Correspondence should be addressed to M Orešič Email: matej.oresic{at}vtt.fi)
Abstract
Lipidome is loosely defined as the entire spectrum of lipids in a biological system. Given the modern lipidomics platforms for the first time empower us with the ability to obtain a snapshot of complete cellular/organismal lipidome, many surprises and discoveries are likely awaiting us in the area of lipids as related to cellular/organismal physiology. The lipidomics approaches can be applied both as the phenotyping platform and for the hypothesis-driven research aiming to elucidate, e.g., a specific pathway or gene function. Modern lipidomics methods combine the latest mass spectrometry technology and bioinformatics methods with traditional methods such as for sample preparation and lipid extraction.
Introduction
The ‘omics’ revolution has stimulated the concept of molecular profiling in biological systems. Although having been an intensive area of research already in the 1960s, lipid research has recently gained prominence with the emergence of lipidomics (Han & Gross 2005, Orešič et al. 2008). Chemically, lipids are broadly defined as hydrophobic or amphiphilic small molecules that originate either entirely or in part from two distinct types of building blocks: ketoacyl and isoprene groups (Fahy et al. 2005). Structural diversity of lipid building blocks and many ways they can be combined lead to a vast number of possible molecular lipids in a given biological system. We estimated that this number is of the order of 200 000 (Yetukuri et al. 2008). This clearly demands specific solutions for lipidomic analysis, handling of lipidomics data, both at the level of data processing and lipid identification, as well as for the lipid pathway analysis (Yetukuri et al. 2007).
Recent advances in analytical techniques for metabolomics (Goodacre et al. 2004, Van der Greef et al. 2004) have empowered us with better tools for global characterization of lipids at the molecular level. Modern approaches for lipidomic analysis are dominated by mass spectrometry (MS; Want et al. 2005), commonly preceded by separation by liquid chromatography (LC; Sommer et al. 2006) or gas chromatography (GC; Jiye et al. 2005). The novel MS-based lipidomics methods afford the study of intact lipid molecular species from very small amounts of samples (Han & Gross 2005, Schwudke et al. 2006), e.g. from cells or tissue biopsies. Such methods, due to their wide dynamic range, enable quantitative or relative determination of compounds across a broad range of concentrations. Albeit still being far from complete coverage of lipids in biological systems, the progress of lipidomics affords increasing penetrance into global lipidomes as well as their characterization in physiological context.
Methods for lipidomics
This section provides an overview of different stages of a typical lipidomics platform, from sample preparation to bioinformatics. For in-depth discussions and specific methods, the reader is advised to refer to the recently published series of books on the topic (Brown 2007a,b,c).
Samples
The samples should be stored frozen at −80 °C or lower, multiple melting of the samples should be avoided and the samples should be handled in cold (e.g. at −20 °C) during the preparation to avoid lipid degradation mostly due to oxidation of polyunsaturated fatty acids. Cultured or isolated cells from animal or microbial sources are usually suspended in buffer solutions and determining the protein content of such a sample enables also quantification of lipids. After taking an aliquot for protein measurement, the rest of the sample can be spiked with internal standards and used to extract lipids. Proteins are precipitated simultaneously when using chloroform:methanol mixtures or, for example acetonitrile for lipid extraction. To break the cells, homogenization with glass beads is efficient. Ultrasonic bath has also been applied. However, as a potential drawback, this approach is to some extent equivalent to warming of the sample.
Tissue samples must be first homogenized, and if the fat content within the sample set is variable or if the sample amount is extremely small, determination of total protein concentration and addition of appropriated internal standards are essential steps for quantification. Homogenization on glass beads by using frozen tissues is sufficient for muscle tissue samples, for example. Proteins are precipitated most commonly by methanol during the extraction procedure with chloroform:methanol.
In a clinical setting, a serum is the most common sample type when studying lipids. Serum is normally first spiked with internal standards and due to the high albumin content, precipitation of proteins is essential. In addition to methanol or isopropanol, acid is sometimes used for removing proteins, but it may hydrolyse some lipids and can thus have an effect on the free fatty acid content. The type of anticoagulant used during blood drawing can affect the amount of lysophospholipids, the use of EDTA is mostly recommended.
Extraction of lipids
The extraction methods developed for the isolation of lipids containing both neutral and polar constituents at the end of the 1950s (Folch et al. 1957, Bligh & Dyer 1959) are still widely used, mostly in modified forms. The technique is based on a mixture of chloroform and methanol (2:1) including washing with salt solution or adjusting the amount of water to improve lipid extraction. To enhance the recovery of phospholipids, the proportion of methanol can be increased (von Thiele 1979, Holland et al. 2003). The extraction method is often highly dependent on the analytical method used. For example, for the determination of fatty acid composition, chloroform:methanol extract is evaporated and the residue is hydrolysed and methylated for GC analysis. On the other hand, avoiding extraction of triacylglycerols by using methanol only is beneficial if the target is to determine free fatty acids and sterols simultaneously with hydrophilic metabolites by using GC MS-based metabolomic profiling. In addition to specific solvent systems, solid phase extraction or chromatographic separation on silica thin-layer plates (Lutzke & Braughler 1990) or a silica column is used to fractionate lipid extracts before more detailed analysis. Targeted analyses of eicosanoids (Masoodi et al. 2008) or sphingolipids (Bodennec et al. 2000), for example, provide special and well-optimized extraction and fractionation systems.
Traditional lipid and fatty acid analyses
Routine kit-based spectrophotometric measurements for total and high-density lipoprotein cholesterol and further calculation of low-density cholesterol levels by Friedewald equation (Friedewald et al. 1972), as well as determining serum triacylglycerol levels are still reasonable and informative methods to describe general lipid balance in the body. Global lipidomics techniques gives data on the level of different lipid subclasses and the sum of cholesterol containing lipid species is not directly linked to the total amount of cholesterol. Correspondingly, the sum of triacylglycerols is not necessarily the same as the spectrophotometric result that indicates the amount of glycerol after hydrolysis of fatty acids. The basic reason for these inconsistencies is the large amount of different overlapping cholesteryl ester and triacylglycerol species. It is known that the length and unsaturation as well as lipid concentrations are affecting instrument responses of phospholipids (Koivusalo et al. 2001) in electrospray ionization-MS (ESI-MS).
GC analysis of fatty acid composition of total serum, phospholipids or cholesteryl esters is still a method of choice in respect of interpreting fatty acid metabolism (Seppänen-Laakso et al. 2002). The determination of fatty acid methyl ester is a highly reproducible and repeatable method and even very small changes may turn out both significant and important. Isomeric forms of fatty acids, like cis and trans, can be analysed by GC when using a proper capillary column. Fatty acids originating from the sn-1 or sn-2 positions in glycerophospholipids can be analysed after the treatment with phospholipases. The analysis of free fatty acids can be performed without derivatization or after trimethylsilylation, for example.
HPLC is a useful method for lipid extracts to separate different classes as groups, such as cholesteryl esters, triacylglycerols, free cholesterol, phosphatidylethanolamines, phosphatidylcholines, sphingomyelins and lysophospholipids (Fig. 1). Normal-phase techniques with evaporative light-scattering detection have proved to be the most successful for class separation (Homan & Anderson 1998, Seppänen-Laakso et al. 2001), although these methods are not usually coupled online with the MS detection. The reversed-phase technique, which enables partial separation of single molecular lipid species is still more routinely used in LC-MS (Yetukuri et al. 2007).
Summary of methods for lipid analysis. (A) Traditionally, serum cholesterol and triacylglycerol levels have been measured. (B) High performance liquid chromatographic separation using normal-phase techniques and evaporative light-scattering detection enables the separation of major lipid classes. (C and D) Isolation of phospholipid fraction (PL) is performed on silica thin-layer chromatography (TLC) and the fatty acid composition of total serum or its PL fraction is determined by capillary gas chromatography (GC). (E) Two-dimensional GC coupled to time-of-flight mass spectrometry (GC×GC-TOF) is used for small molecule metabolomics and (F–G) liquid chromatography-mass spectrometry (LC-MS) is commonly used in lipidomics. (H) Data mining and processing methods are needed to analyse the data.
Lipidomic analysis
Global lipidomics means mapping the entire spectrum of lipids and lipid molecular species in a biological system, in lipidome. It can be shotgun lipidomics if the lipid extract is introduced directly into the mass spectrometer (Han & Gross 2005, Schwudke et al. 2006) or LC-MS when the extract is introduced to MS after chromatographic separation (Houjou et al. 2005, Laaksonen et al. 2006). Targeted lipidomics, in turn, aims to quantify a single or multiple related lipids, e.g. from a specific class such as prostaglandins or triacylglycerols. The most common ionization technique in global lipidomics approaches is ESI, which is a soft ionization method that can be combined with several types of mass analysers such as triple quadrupole, ion trap, time-of-flight (TOF) and orbitrap. ESI is most often used in positive ion mode (ESI+) for global approaches, since ESI+ enables coverage of diverse classes such as major phospholipids, sphingolipids and acylglycerols. However, in particular when phospholipids are the focus, some classes such as phosphatidylinositol, phosphatidylserine and phosphatidic acid are not suitable for ESI+ analysis and the negative ion mode (ESI−) is advantageous. Additionally, ESI− analysis can provide information about carboxylate anions of the fatty acids in different phospholipids (Watson 2006).
Quantification of lipid molecular species is usually performed by using internal standards, often as a mixture covering most lipid classes. If available, labelled reference compounds would best correspond to the compounds in lipid mixtures. The quantification procedures including several practical aspects have been discussed recently in detail (Moore et al. 2007).
Identification of compounds in LC-MS can be a time-consuming issue. Self-made libraries based on reference substances are of high importance and should contain also information on MS/MS fragmentation obtained by tandem MS (LC-MS/MS). Precursor ion scanning and neutral loss scanning of polar head groups or fatty acids in phospholipids are widely used in detection and identification (Taguchi et al. 2005).
Following the raw data acquisition from the MS instrument, the data need to be processed to convert the raw signal into a list of lipid molecular concentrations for each sample measured (Katajamaa & Orešič 2007, Yetukuri et al. 2008). The specific data processing steps may vary depending on the instrument and the data acquisition mode. The analysis and interpretation of lipidomics data, following the data processing, typically consist of statistical (e.g. testing for differential expression, multivariate analysis such as clustering) and bioinformatics approaches (e.g. mapping to lipid pathways; Yetukuri et al. 2007, 2008).
Applications of lipidomics
In this section we briefly overview selected and illustrative applications of lipidomics, aiming to cover multiple strategies for lipidomics as well as multiple research areas and challenges. The lipidomics profiles from cells, tissues or body fluids provide a global snapshot of lipid concentrations in a particular biological sample, corresponding e.g. to a specific physiological state, time or intervention response. Due to untargeted and global nature of lipidomics, new important biological functions of cells and tissues can arise as derived from lipidomics data.
The lipotoxicity hypothesis states that the link between obesity and these metabolic complications may be related to organ-specific toxic effects induced by excess of nutrients in the form of reactive lipid species such as ceramides (Unger 2002). Limited adipose tissue (AT) expandability may be one mechanism leading to excess flux of fatty acids from AT to peripheral tissues, resulting in the synthesis of toxic lipids. In this context, we studied the PPARG(−/−) Lep(ob)/Lep(ob) (POKO mouse), which is characterized by limited AT expandability due to ablation of PPARG and a severe diabetic phenotype (Medina-Gomez et al. 2007). Using UPLC/MS-based global lipidomics approach, we indeed found that the pancreatic islets, skeletal muscle and liver in POKO mouse are characterized by elevated ceramides. These changes were accompanied by depletion of antioxidant phospholipids in islets and a distinct triacylgycerol profile in the liver (Medina-Gomez et al. 2007). Using new lipid pathway reconstruction methodology, we later demonstrated that the synthesis of ceramides in liver is due to increased de novo synthesis as well as activity of the salvage pathway (Yetukuri et al. 2007).
Schmitz et al. (2006) have applied the lipidomics approach with ESI-MS/MS techniques in studying structural and functional defects of ABC-transporters in cellular lipid trafficking. A higher APOA1 specific efflux was observed for monounsaturated PC species together with a decreased contribution of polyunsaturated PC species. Medium-chain sphingomyelin species (SM14:0 and SM16:1) were preferentially translocated to APOA1 in both fibroblasts and macrophages. Combining the analysis of bile acids and phytosterols to lipidomics would further help to understand also the formation of toxic bile.
Breast cancer is the most frequently diagnosed cancer among women, an estimated 200 000 new cases in 2006 in the USA. Lysophospholipids are known activators of proliferation and growth in breast cancer and have been suggested to be potential biomarkers in blood stream (Murph et al. 2007). Methanol extract of plasma contains lysophosphatidic acid (LPA) species that are determined by LC-MS/MS in negative ion mode. Lysophosphatidylcholines (LPCs) are then analysed in positive ion mode after different HPLC run. Already during blood sampling it is important to use EDTA or citrate to chelate calcium ions and decrease the ability of LPA-producing enzymes to actively form LPA during sample processing (Murph et al. 2007).
Quantitative determination of eicosanoids in biological samples is problematic. Practical advices have been given in Wang & DuBois (2007), for example. Masoodi et al. (2008) have been focused purely on LC/ESI-MS/MS-based targeted lipidomics in the analysis of bioactive lipid mediators, leucotrienes, resolvins, protectins and related hydroxy-fatty acids. They have reported the limits of detection and quantification to be 10–20 and 20–50 pg respectively. For prostanoid analysis, the corresponding levels were 0.5–50 and 2–100 pg respectively (Masoodi & Nicolaou 2006).
Autoxidation of cholesterol induced by a free-radical mechanism generates ‘oxycholesterols’, the major ones being 25-hydroxycholesterol and cholestantriol. Several of those components have been demonstrated to be cytotoxic, atherogenic, mutagenic and carcinogenic. An atmospheric pressure chemical ionization LC/MS technique has been used in their quantification (Manini et al. 1998). The level of 7-ketocholesterol in lyophilized beef was about the same as obtained by LC with u.v. detection (27.3 vs 24.7 μg/g), but the advantage of using MS techniques was that two other oxidation products could also be found and quantified.
The oxidation of phospholipids in biological membranes has been implicated in a variety of human diseases, such as atherosclerosis, ischaemia, carcinogenesis and Alzheimer's disease. Plasmalogen phospholipids have a vinyl ether substituent at the sn-1 position of the glycerol backbone and a polyunsaturated fatty acid in sn-2 position, which are susceptible to oxidative reactions that occur at cell membranes. The structure determinations of these oxidation products have been performed both by electron ionization GC/MS and LC-MS/MS (Khaselev & Murphy 2000, Zemski Berry & Murphy 2005). Accurate masses of phosphatidylcholine and LPCs oxidation products in human plasma (Adachi et al. 2006) have been determined by quadrupole TOF MS (Q-TOF).
Concluding remarks
Lipidomics is a relatively new addition to the ‘omics’ nomenclature. The lipidomics approaches can be applied both as the phenotyping platform, e.g. for biomarker screening and for the hypothesis-driven research aiming to elucidate, e.g. specific pathway or gene function. Given the modern lipidomics platforms for the first time empower us with the ability to obtain a snapshot of complete cellular/organismal lipidome, many surprises and discoveries are likely awaiting us in the area of lipids as related to cellular/organismal physiology.
However, the ‘old school’ lipidology may still be of great value. Since the lipids have been an intense area of research already in the 1960s, the traditional methods in particular for sample preparation and lipid extraction are a valuable resource also for today's lipidomics platforms. Modern lipidomics methods, utilizing the latest MS technology and bioinformatics methods, therefore still require in-depth understanding of traditional lipid biochemistry.
Declaration of interest
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.
Funding
This research did not receive any specific grant from any funding agency in the public, commercial or not-for-profit sector.
- Revision received 24 November 2008
- Accepted 5 December 2008
- © 2009 Society for Endocrinology
