SEARCH

SEARCH BY CITATION

Keywords:

  • lipid;
  • metabolism;
  • signaling;
  • mass spectrometry;
  • lipid droplets

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Lipidomics in Tissues
  5. MASS SPECTROMETRY IMAGING: Localizing Lipids In Situ
  6. Subcellular Lipidomics
  7. Other Potential Approaches for Plant Lipidomics in the Future
  8. Acknowledgements
  9. References

Mass spectrometry (MS) advances in recent years have revolutionized the biochemical analysis of lipids in plants, and made possible new theories about the structural diversity and functional complexity of lipids in plant cells. Approaches have been developed to profile the lipidome of plants with increasing chemical and spatial resolution. Here we highlight a variety of methods for lipidomics analysis at the tissue, cellular and subcellular levels. These procedures allow the simultaneous identification and quantification of hundreds of lipids species in tissue extracts by direct-infusion MS, localization of lipids in tissues and cells by laser desorption/ionization MS, and even profiling of lipids in individual subcellular compartments by direct-organelle MS. Applications of these approaches to achieve improved understanding of plant lipid metabolism, compartmentation and function are discussed.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Lipidomics in Tissues
  5. MASS SPECTROMETRY IMAGING: Localizing Lipids In Situ
  6. Subcellular Lipidomics
  7. Other Potential Approaches for Plant Lipidomics in the Future
  8. Acknowledgements
  9. References

The analysis of lipid composition in plant tissues has seen considerable developments in high-resolution lipidomics aimed at improving our knowledge of the metabolism and functional roles of plant lipids. The past decade has witnessed a significant investment in mass spectrometry (MS)-based instrumentation and techniques that, when applied to plant tissues, have built substantially upon an already strong foundation of lipid research. Complementary advances in genomics applications, including annotation of candidate genes involved in lipid metabolism (Mekhedov et al., 2000; Rhee et al., 2003; Li-Beisson et al., 2010), the availability of transgenic materials with altered lipid compositions (Lemieux et al., 1990; Wallis and Browse, 2002), and a wealth of developmental and stress-related gene expression data (Beisson et al., 2003), have revealed new experimental systems for which lipidomics approaches are ideally suited and can contribute new information. The central goal of lipidomics is to identify, quantify and now visualize all lipids in plant tissues (the lipidome). The lipidome of eukaryotes consists of thousands of lipids that are structurally and functionally diverse (Han and Gross, 2005a; Ejsing et al., 2009; Dennis et al., 2010). Visualizing the lipidome in plant cells and tissues using high-resolution techniques will ultimately lead to a more comprehensive and detailed molecular understanding of plant lipid function at the tissue, cellular and subcellular levels.

Lipidomics in Tissues

  1. Top of page
  2. Summary
  3. Introduction
  4. Lipidomics in Tissues
  5. MASS SPECTROMETRY IMAGING: Localizing Lipids In Situ
  6. Subcellular Lipidomics
  7. Other Potential Approaches for Plant Lipidomics in the Future
  8. Acknowledgements
  9. References

Conventional profiling of lipid compositions from plant tissues requires solvent-based methods for comprehensive/selective lipid extraction as well as prevention of enzymatic or non-enzymatic degradation or loss of molecular species (Christie, 1993). To accurately identify and quantify the lipid metabolites present in plant tissues, it is critical to minimize any trauma associated with extractions (Galliard, 1970). Most commonly, plant tissues are flash-frozen to stop all cellular processes, followed by incubation in isopropanol at high temperatures to minimize enzymatic activity and structural artifacts (Roughan et al., 1978) of highly active hydrolytic and oxidative enzymes (Galliard, 1975). Popular extraction methods based on those developed by Bligh and Dyer (1959) and Folch et al. (1957) rely on differences in metabolite polarity to selectively purify lipid components. Several research groups have compared and reviewed various lipid extraction techniques in terms of optimal extraction efficiency and minimal lipid degradation (Fishwick and Wright, 1977; Kolarovic and Fournier, 1986; Manirakiza et al., 2001; Li-Beisson et al., 2010). Most lipid extracts from plant tissues, especially non-seed tissues, comprise a structurally diverse set of lipids, with a large range of concentrations within and across lipid classes (Devaiah et al., 2006), although it may be necessary to alter solvent ratios to favor extraction of more polar lipids such as sphingolipids (Markham et al., 2006). Experimental conditions, including modes of sample preparation, are therefore key to accurate lipidomics analysis. Many of the recent high-resolution analytical techniques are designed to minimize any form of sample preparation that may adversely influence the analysis, and although advances in chromatographic separations play an important role in lipidomics analysis, we focus most of our attention here on the mass spectrometric aspects of lipidomics procedures.

Shotgun lipidomics: high-resolution direct-infusion mass spectrometry

The analysis of lipids directly from extracts or biological sources by direct-infusion MS (i.e. no chromatographic online separation) is termed ‘shotgun lipidomics’ (Han and Gross, 2005a). By exploiting the diverse physical and chemical properties of lipids, contemporary mass spectrometers can be set up to analyze hundreds of lipids almost simultaneously (Han and Gross, 2005a). Improvements in the mass accuracy, resolution and sensitivity of contemporary mass spectrometers (e.g. the Thermo-Fisher Orbitrap mass analyzer; Hu et al., 2005) has made it possible to identify many compounds by direct comparison of the detected mass-to-charge ratio (m/z) with the theoretical m/z in public databases such as the LipidMAPS initiative (Sud et al., 2007). Structural characterization of each lipid species by ion fragmentation (MS/MS) may improve the molecular identification (e.g. lipid class, acyl chain length and saturation state, chemical modifications, etc.) and also serves as a mean of quantification. The development of shotgun lipidomics has resulted in the development of three analytical platforms optimized for different applications but ultimately all leading to improved identification and quantification of lipid species.

The first direct-infusion MS method operates by scanning for a characteristic fragment of each lipid class using the precursor and neutral-loss scanning capabilities of a triple quadrupole (QQQ) mass spectrometer (Brügger et al., 1997; Welti et al., 2002; Welti and Wang, 2004). A triple quadrupole MS has a linear series of three quadrupoles: two identical quadrupoles that function as mass analyzers (Q1 and Q3) and a collision cell that is responsible for ion fragmentation (Q2) between Q1 and Q3. In a precursor ion scan, Q3 selects for a single m/z diagnostic product [e.g. m/z+184, a phosphatidylcholine (PC) head group] that is generated by fragmentation (Q2) of each precursor ion scanned in a particular m/z range (e.g. all lipids from m/z 50–1200). Neutral-loss scanning is a variation of the precursor ion scan in which Q3 is also scanned to produce a spectrum that only shows the precursor ions that have lost a particular fragment (e.g. m/z +141, all ions that when fragmented lose a phosphoethanolamine head group). This methodology has been particularly effective in resolving polar lipid classes within mammalian (Brügger et al., 1997) and plant (Welti and Wang, 2004) tissues. However, a large set of internal standards is recommended to account for potential quantitative differences resulting from energy- and mass-specific MS responses in precursor and neutral-loss modes (Welti et al., 2007). Internal standards that represent each lipid class and are suitable for analysis (i.e. do not occur endogenously) are becoming increasingly available commercially.

The second direct-infusion MS method utilizes tandem MS comprising product ion scans at each unit mass (Ejsing et al., 2009). Performance of a product ion scan, using a triple quadrupole instrument as described above, allows transmission of a single m/z through Q1 [e.g. Tri-18:1-TAG (triacylglycerol)] with all product ions scanned in Q3 [e.g. Di-18:1-DAG (diacylglycerol) fragment] being produced via fragmentation in Q2. This method essentially profiles the structural fragments of each molecule transmitted, and identifies the lipid class and acyl chain composition for each molecule independently by software analysis. An automated version of this platform with yeast cells quantified 250 molecular lipid species in 21 major classes (approximately 95% lipidome coverage), an improvement in sensitivity and coverage relative to other compared approaches (Ejsing et al., 2009). However, similar to precursor/neutral-loss scanning, inclusion of a large set of representative internal standards is required for quantification and identification purposes.

A third direct-infusion MS method for determining lipid compositions in complex extracts is multi-dimensional mass spectrometry-based shotgun lipidomics (MDMS-SL or 2D-MS), which combines neutral-loss scanning and precursor ion scans in a mass-ramp fashion (Yang et al., 2009; Han et al., 2011). Plots are generated from the lipid component fragments present (i.e. tandem m/z fragments), including glycerolipid backbones, head groups and aliphatic chains on the y axis versus the molecular ion m/z on the x axis (Han and Gross, 2005b). In MDMS, each set of cross peaks for the scanned fragments and the molecular ions may be used to determine the identity of all lipid species analyzed (Yang et al., 2009). For quantification purposes, applicable internal standards are required as described above to account for differences in fragmentation efficiency (Han et al., 2004). Abundant and fully resolved species are quantified by ratiometric comparison with an internal standard of the same lipid class. Species that are of low abundance, or not fully resolved, are quantified by one or more class-specific precursor ion scan, neutral-loss scan or both (Yang et al., 2009). Automated processing algorithms identify each molecular species from the 2D-MS data using an appropriate level of mass accuracy (Yang et al., 2009). This method has been used to profile the hepatic cellular lipidome with improved coverage, and also in a number of other studies on mammalian tissues (Han et al., 2004, 2008, 2011).

Application of shotgun-based lipidomics leads to new insights into plant lipid metabolism and signaling

Direct-infusion MS approaches now play a central role in experimental analysis of lipid metabolism and function. Analyzing the lipid compositions in Arabidopsis mutants has shed new light on the roles of many genes involved in lipid metabolism and homeostasis (Wallis and Browse, 2002; Li-Beisson et al., 2010), and this is illustrated here by several representative (but not exhaustive) examples.

Comparative quantitative profiles of polar glycerolipid species in Arabidopsis wild-type and phospholipase Dα1 (PLD) knockout mutants showed distinct lipid profiles suggesting that PLD is involved in membrane lipid degradation in seeds and contributes significantly to phosphatidic acid levels in roots, seeds and flowers, but not siliques and leaves (Devaiah et al., 2006). Also, profiles of lipid class alterations in trigalactosyldiacylglycerol (TGD3) mutants, which are disrupted in incorporation of endoplasmic reticulum-derived lipid precursors into thylakoid lipids, supported a role for TGD3 as a component of the ATPase lipid transport complex (Lu et al., 2007). Further, lipidomics analysis of an Arabidopsis mutant with a disruption in the homolog of a human lipodystrophy gene (CGI-58) showed marked increases in the triacylglycerol (TAG) content in leaves, suggesting a role for this plant protein in the regulation of neutral lipid accumulation and turnover in vegetative plant tissues (James et al., 2010).

Determining the detailed lipid compositions in oilseeds engineered for enhanced or altered lipid metabolism is likely to be important for designing optimal accumulation of various lipid products (Lu et al., 2011). Many of the lipids produced in transgenic oilseeds accumulate at lower concentrations than in the natural host from which the transgene was derived (Lu et al., 2011), and lipidomics profiling has helped to address these difficulties in several ways. Attempts to enhance the content of hydroxy fatty acids in Arabidopsis using Ricinus communis type 2 acyl coenzyme A: diacylglycerol acyltransferase (RcDGAT2) substantially modified the composition of the TAG pool (Burgal et al., 2008), as analyzed by shotgun lipidomics. In another study, a bottleneck in the biosynthesis of very long chain fatty acids was identified in lipidomic profiles of tobacco (Nicotiana tabacum) and linseed (Linum usitatissimum) plants over-expressing fatty acyl desaturases and elongases, in which immediate shuttling of newly synthesized Δ6 fatty acids into TAGs prevented their accumulation in acyl CoA pools that is required for further elongation into C20 fatty acids (Abbadi et al., 2004). Lipidomics profiling reveals global changes in lipid compositions, sometimes revealing novel consequences of gene expression and the accumulation of new products. For example, lipidomic profiles showed increased levels of polyunsaturated di-18:2 (dilinoleoyl) phosphatidylcholine (PC) and polyunsaturated di-18:2 (dilinoleoyl) phosphatidylethanolamine when PLD expression was suppressed in soybean seeds (Glycine max), which has implications for the role of PLD in altering the properties of edible and industrial soybean lecithin (Lee et al., 2011).

Results from high-resolution MS of Arabidopsis tissues have suggested that lipids play a much more substantial role in plant signaling and stress responses than previously appreciated (Welti and Wang, 2004). Global changes in lipid compositions in response to various plant stresses, including drought, salinity, freezing and nutrient deficiency, emphasized the role of PLD- and phosphatidic acid-mediated signaling in plants (Li et al., 2009). For example, profiling of PLDζ1 and PLDζ2 knockout mutants in Arabidopsis demonstrated their role in lipid remodeling in rosette leaves during phosphate starvation, with a decreased concentration of phospholipids and an increased concentration of galactolipids relative to wild-type plants (Li et al., 2006). Similarly, changes in the polar lipid species content, especially phosphatidic acid, during wound-induced metabolism (Zien et al., 2001), reduced nitrogen (Hong et al., 2009), phosphorous deprivation (Li et al., 2006) and freezing conditions (Li et al., 2008), suggests specialized roles for the various PLD isoforms. Analysis of the Arabidopsis lipin homologs, AtPAH1 and AtPAH2, showed that the products of these two genes were responsible for the eukaryotic pathway of galactolipid synthesis during lipid remodeling under phosphate starvation (Nakamura et al., 2009). In addition, adding front-end liquid chromatography to MS-based lipidomics showed global changes in the Arabidopsis glycerolipidome in response to light and temperature that should be considered when establishing experimental growth conditions in order to ensure reproducibility across laboratories (Burgos et al., 2011).

MASS SPECTROMETRY IMAGING: Localizing Lipids In Situ

  1. Top of page
  2. Summary
  3. Introduction
  4. Lipidomics in Tissues
  5. MASS SPECTROMETRY IMAGING: Localizing Lipids In Situ
  6. Subcellular Lipidomics
  7. Other Potential Approaches for Plant Lipidomics in the Future
  8. Acknowledgements
  9. References

Lipidomic profiles of tissue extracts have contributed substantially to elucidating the plant lipidome and its changes in association with physiological conditions. However, the spatial organization of lipids in tissues, cells and subcellular compartments is mostly lost as a result of chemical-based extraction of plant tissues. Characterization of the location of lipids in plant tissues and cells is at least as important to determining the complexity of the lipidome and its quantification. The cellular distribution of lipids can be partially resolved using optical imaging techniques such as confocal or electron microscopy with selected fluorescent dyes, antibodies and chemical modifications (Eggeling et al., 2009; Wessels et al., 2010), but visualizing the detailed composition of individual molecular species is limited using conventional microscopy approaches. The emerging field of mass spectrometry imaging (MSI) captures the spatial distribution of lipids in situ, with detailed compositional information at the tissue, cellular, and, with some instrument modifications, subcellular levels. Several spectrometers are available for MSI, which vary in terms of ionization, detection sensitivity, spatial and spectral resolution, and the type of information acquired. Some instruments require advanced sample preparation but overall enable a diversified approach towards analyzing the lipidome by MSI. Although there have been several comprehensive reviews of MSI (Chaurand et al., 2005; Dill et al., 2011; Harris et al., 2011; Lee et al., 2012), here we discuss methods for visualizing plant lipids at the cellular level – focusing on matrix-assisted laser desorption/ionization MS (MALDI-MS) and desorption electrospray ionization (DESI). New techniques allow the analysis of lipids at the subcellular level, such as direct organelle MS (DOMS) and live single-cell MS. Collectively, these techniques provide plant scientists with the opportunity to obtain a breadth of chemical composition information within a biologically relevant spatial context (see Table 1 for summary of methods discussed). Some alternative MS approaches for analyzing lipids in plant cells are also mentioned below, including secondary ion MS (SIMS) and ion mobility MS (IMMS).

Table 1.   Summary of selected techniques for profiling plant lipids
TechniqueAdvantagesDisadvantagesSpatial resolutionPlant lipids profiled to dateRepresentative references
Shotgun lipidomicsHigh-throughput analysis; detailed compositional analysis; well-established methodsLow spatial resolution; cellular localization lostTissuePhospholipids, glycolipids, sphingolipids, TAGs, sterols, othersWelti and Wang, 2004; Han and Gross, 2005a,b
MALDIIn situ localization; cellular resolutionDifficult sample preparation; lipid fragmentation in source; slow imaging at high resolutionTissue/cellular (10–50 μm)Cutilcular waxes, polyketides, chlorophyll, TAGs, fatty acids, PCMurphy et al., 2009; Fuchs et al., 2010;
DESILittle to no sample preparation; in situ localizationLow spatial resolution; penetration of plant cuticleTissue/cellular (100–250 μm)TAGs, fatty acidsTakáts et al., 2004; Wiseman et al., 2008
DOMSDirect sampling; organellar analysis; high sensitivityLow throughput; limited size of nanospray tips; clogging of tips in in situ analysisCellular/subcellular (<10 μm)TAGs, galactolipidsHorn and Chapman, 2011; Horn et al., 2011;
Video MSDirect sampling; cellular analysisLow throughputCellular/subcellular (<10 μm)IsoprenoidsTsuyama et al., 2008; Lorenzo Tejedor et al., 2009
SIMSVery high spatial resolutionVariable ionization efficiencies within matricesCellular/subcellular (400 nm–2 μm)PolyketidesPassarelli and Winograd, 2011; Heien et al., 2010

Matrix-assisted laser desorption/ionization

Matrix-assisted laser desorption/ionization MS is a versatile imaging platform that is particular amenable for sampling biological tissues. Although MALDI has traditionally been utilized to analyze larger biomolecules such as proteins and peptides, it is increasingly being applied to characterizing lipid extracts and visualizing lipids in situ. Tissues are imaged by preparing thin sections that can be coated with a suitable matrix that limits sample damage and promotes ionization. The choice of matrix (e.g. dihydroxybenzoic acid, cinnamic acid, etc.) and the method by which it is applied to the sample (e.g. sublimation) are important considerations as they may significantly affect ionization efficiency and the resulting compositional analysis. Samples coated with matrix are then ionized, as described in detail previously (Zenobi and Knochenmuss, 1998). Briefly, a laser ablates a region of tissue, and both matrix- and tissue-derived ions are directed towards a mass analyzer. The laser is rastered over the tissue sample, and mass spectra are collected at each location, producing a chemical map of the plant material. A resolution of 10–50 μm is typically used for laser-based imaging, with some reports of submicron MALDI imaging (Guenther et al., 2010). As ionization typically occurs under vacuum, sample preservation and preparation, in addition to the compatibility of the ionization matrix, are key factors for comprehensive analysis and compositional integrity. Quantification of lipid species in tissues and sample extracts by MALDI-MS is limited due to suppression of certain lipid classes [i.e. PC generally suppresses other phospholipids and TAGs (Emerson et al., 2010)]. Validation using conventional shotgun lipidomics, as well the inclusion of internal standards within or in addition to the matrix, may resolve suppression-influenced quantification.

Several studies have used MALDI-TOF MS to profile plant lipid extracts as an alternative to separation-based MS methods. These MALDI-TOF MS data also support standardization of the platform for future in situ analysis. An in-depth study on PC species extracted from soybean (Glycine max) and potato tubers (Solanum tuberosum) found that the individual isomeric PC species could be resolved by enzymatic digestion of PC species and analysis by MALDI-TOF in full-scan mode, and post-source decay of fragments generated during their flight in the mass spectrometer (Zabrouskov et al., 2001). Similar methodology was used successfully for analysis of the glycerophospholipid and glycolipid species of the plant-like green alga Chlamydomonas reinhardtii (Vieler et al., 2007). Lipid-soluble betacyanin pigments, utilized for food colorants and as antioxidants, were extracted from Amaranthus tricolor seedlings, Gomphrena globosa flowers and Hylocereus polyrhizus fruits, and structurally characterized by MALDI-quadrupole ion trap-TOF MS (Cai et al., 2006). Similar characterization was performed for chlorophyll and its derivatives extracted from spinach (Spinacia oleracea L.) leaves using MALDI-TOF MS (Suzuki et al., 2009). The TAG and fatty acid compositions of total lipid extracts from olive (Olea europaea) and pomegranate (Punica granatum) seeds have been standardized using MALDI-TOF as an alternative to fatty acid analysis by gas chromatography for quality assurance in industrial production of seed oils (Wiesman and Chapagain, 2010).

In terms of imaging lipids in situ, most analysis by MALDI-MS to date has focused on mammalian tissues in the context of genetic disorders of lipid metabolism (Murphy et al., 2009; Fuchs et al., 2010). Recently, however, MALDI-MS has been used for in situ analysis of plant lipids and several other types of metabolites (Kaspar et al., 2011). Surface lipids were imaged directly in Arabidopsis floral and leaf tissues by MALDI-MS (Cha et al., 2008; Jun et al., 2010; Vrkoslav et al., 2010). Elsewhere, MALDI-MS using colloidal graphite as a matrix for ionization was used to image the free fatty acids in strawberry seeds (Fragaria × ananassa) and apple tissues (Malus domestica) (Zhang et al., 2007). MSI in rice grains (Oryza sativa) identified PC molecular species (Zaima et al., 2010).

Infrared-laser desorption/ionization oTOF MS has been used to directly profile metabolite changes in defense responses in tobacco leaves (Nicotiana tabacum) against the oomycete Phytophthora nicotianae (Ibanez et al., 2010). These time-dependent profiling experiments identified a group of peroxidized oxylipins, which can act as stress signaling molecules, and their precursor γ-linolenic acid, in response to pathogen infection. This technique was also used to generate lipid profiles (i.e. TAGs, fatty acids, PCs, etc.) directly from pieces of green olive (Olea europaea), sesame seed (Sesamum indicum), sunflower seed (Helianthus annuus L.), white coconut flesh (Cocos nucifera L.), strawberry seeds (Fragaria x ananassa) , and a red rose leaf (Rosa sp. cv El toro), which demonstrates the diversity of plant tissues that can be quickly profiled (Dreisewerd et al., 2007). Atmospheric pressure infrared MALDI-MS was used to obtain profiles from various plant organs, including TAGs within sections of almond seeds (Prunus amygdalus) (Li et al., 2007). Laser ablation ESI (LAESI) was used to profile the 2D distribution (with approximately 350 μm lateral resolution and approximately 50 μm depth resolution) of metabolites within the polyketide kaempferol biosynthesis pathway in leaf tissue of the variegated zebra plant (Aphelandra squarrosa) (Nemes et al., 2008). Similar imaging analysis was performed to reconstruct 3D metabolite profiles of the leaves of A. squarrosa and Spathiphyllum lynise (peace lily), including the kaempferol pathway and chlorophyll pigments (Nemes et al., 2009). Localizing lipid metabolites directly in plant tissues provides a new basis for lipidomics at the cellular level, and will provide important spatial information for studies of lipid metabolism and function.

Desorption electrospray ionization (DESI)

Desorption electrospray ionization is an ambient ionization technique that combines features of electrospray and desorption ionization for direct tissue analysis (Harris et al., 2011). DESI ionizes molecules by generating pneumatically assisted charged droplets directed towards the tissue surface that, upon collision, give rise to secondary droplets from compounds on the tissue surface that can be analyzed by high-resolution MS (Takáts et al., 2004). DESI-MS is an attractive technique for MSI applications as it requires little to no sample preparation (no matrix required) compared with MALDI-MS, and samples are not introduced into high-vacuum conditions (Harris et al., 2011). A major limitation of DESI is that its mode of ionization results in poor spatial resolution: of the order of 100–250 μm, which is generally larger than MALDI (Dill et al., 2009).

Despite the ease of ionization, DESI applications focusing on lipids have been limited (Eberlin et al., 2011). The in situ distributions of phospholipids, sphingomyelin and sulfatides have been shown to distinguish disease states in rat, human and canine species (Wiseman et al., 2008; Dill et al., 2009). In plants, the majority of MSI studies employing DESI have focused on analysis of secondary metabolites. Application of DESI to plant tissue containing thick cuticles is difficult due to problems of penetration below the surface, resulting in insufficient signal intensity and stability (Thunig et al., 2011). The same study confronted this limitation by making tissue prints of plant material from leaves and petals of Hypericum perforatum (St John’s wort) and leaves of Datura stramonium (thorn apple), achieving increased signal intensity and sampling reproducibility for identification of a number of secondary metabolites (Thunig et al., 2011). Similar tissue prints on ordinary printer paper were used to image the distribution of the alkaloid malabaricone C in cross-sections of Myristica malabarica Lam. seeds (Ifa et al., 2011). Principal-component analysis of TAGs analyzed in common plant oils (e.g. olive (Olea europaea), safflower (Carthamus tinctorius L.), hazelnut (Corylus L.)) on glass slides by DESI demonstrated a high resolution of MS analysis with limited sample preparation (Gerbig and Takáts, 2010). Recently, glycosides were detected from Stevia rebaudiana leaves with additional limited spectral information possibly corresponding to fatty acids and other lipids (Jackson et al., 2009). A number of additional ambient sampling/ionization MS methods and applications that are currently being developed will probably be available for analysis of plant lipids in the near future (Harris et al., 2011; Liu et al., 2011).

Subcellular Lipidomics

  1. Top of page
  2. Summary
  3. Introduction
  4. Lipidomics in Tissues
  5. MASS SPECTROMETRY IMAGING: Localizing Lipids In Situ
  6. Subcellular Lipidomics
  7. Other Potential Approaches for Plant Lipidomics in the Future
  8. Acknowledgements
  9. References

Direct organelle mass spectrometry

Neither shotgun lipidomics of tissues, nor MALDI-MS imaging of plant specimens, provide information regarding lipid composition at the subcellular level. Recently, a new procedure was developed (DOMS) that profiles lipids at the organelle level (Horn et al., 2011). This technology is based on the robust methodology of direct-infusion lipidomics, but is performed on a miniature scale that is suitable for analyzing individual subcellular compartments. DOMS involves sampling organelles on a microscope stage using nano-electrospray capillary tips. The lipids are extracted from organelles within the capillary tip by microphase extraction, and identified and quantified by direct-infusion nanospray MS. Organelles may be isolated first by fractionation procedures, or even selected from intact cells. A multi-port nano-manipulation apparatus (Brown et al., 2010; Ledbetter et al., 2010; Horn et al., 2011) designed by Zyvex Inc (http://www.zyvex.com/). is interfaced with a conventional, inverted light microscope to visualize, manipulate and sample individual organelles. The nano-manipulator is equipped with six isolated, low-impedance electrical connections and two glass capillary attachments. These capillary attachments are designed to hold conventional glass nanospray emitters (1.2 mm outer diameter, approximately 1–5 μm tip opening) that are compatible with offline nanospray MS. A programmable injector controls gas flow to emitters to fill with liquids under negative pressure and dispense liquids at positive pressure. The x,y,z position of the emitters is controlled by a user-operated joystick with high-resolution motion that is capable of fine cellular and subcellular manipulations.

The utility of DOMS has been demonstrated using lipid droplets (LDs) from a variety of plant cells and tissue types. LDs were obtained from roots, leaves and seeds, and the detailed lipid composition was determined by nanospray MS (Horn et al., 2011). LDs from plant tissues (approximately 0.1–2 μm diameter) are enriched in TAGs and easily soluble in organic solvents suitable for nanospray MS analysis. LDs can be visualized selectively by epifluorescence microscopy using a neutral lipid-specific dye (i.e. BODIPY 493/503) to confirm organelle identity. Individual LDs are sampled by applying a slight negative pressure to solvent-loaded nanospray capillaries (including internal standards for quantification). Lipids are identified and quantified by nanospray MS, and the acyl chain composition is confirmed by MS/MS. TAG compositions determined from LD pools (approximately 20 LDs) using DOMS were in good agreement with direct-infusion ESI-MS analysis of TAGs from the same tissues, indicating that microsampling organelles by DOMS accurately reflects lipid compositions at an organellar resolution (Horn and Chapman, 2011; Horn et al., 2011).

New insights into lipid droplet composition using DOMS

Direct organelle MS was applied to directly visualize organellar lipid compositional differences between metabolic mutants in cotton (Gossypium hirsutum) seed tissues (Horn et al., 2011). Purified LDs were obtained from mature cotton embryos expressing a non-functional allele of the Brassica napusΔ12 fatty acid desaturase (Bnfad2). LDs from mutant seeds showed distinctly different lipid profiles compared with those from wild-type seeds (cv. Coker 312). There was a dramatic shift towards an increased oleic acid (18:1) acyl chain distribution within TAGs, apparently at the expense of linoleic acid (18:2), consistent with suppression of endogenous oleic acid desaturation by FAD2 (dominant-negative mutation). Although similar conclusions were reached using total seed lipid extracts analyzed by direct-infusion MS, DOMS facilitated the visualization of isolated LD and analysis of TAG molecules within a small population of purified LDs (10–25 LDs). Use of direct-infusion MS of tissue extracts or even other forms of MSI at the cellular level does not allow organelle level analysis, and this is important when sampling tissues of unknown composition. For example, DOMS was utilized to identify TAGs in isolated LDs (approximately 12) from leaves of wild-type and genetically modified mutant Arabidopsis plants (James et al., 2010). Although LDs are abundant in cells of seeds and overwhelm the total seed fatty acid composition of seed extracts, LDs are rare within leaf tissues, and their contribution to the total fatty acid composition of leaves is overshadowed by that of membrane lipids. In Arabidopsis mutants with a disruption in the homolog of a human lipodystrophy gene (CGI-58), the total number and distribution of neutral-lipid-stained particles was substantially increased, similar to the β-oxidation acx1/acx2 double mutants (Slocombe et al., 2009). DOMS was able to demonstrate directly that TAGs were indeed up-regulated and were localized in cytosolic LDs in these mutants. By DOMS analysis, it was also possible to suggest the likely metabolic source of these TAGs (James et al., 2010).

Importantly, DOMS also allows questions regarding the heterogeneity of organelle composition to be asked for the first time (Figure 1). In other words, does the lipid composition of organelles vary from one to another, or is it identical? DOMS has enough sensitivity to profile a single LD within nanoliters of solvent at the nanospray tip opening. Seven individual LDs were directly visualized and analyzed by DOMS from both wild-type- and Bnfad2 seeds. Their relative TAG compositions were quantified by addition of an internal standard to the extraction/infusion solvent. Both wild-type and mutant LDs showed substantial compositional heterogeneity within their respective genetic backgrounds. Despite the variability in TAG composition seen between individual droplets, when the overall composition of individual LDs was averaged, the TAG composition resembles that of standard cottonseed oil. Analysis of tissue extracts by direct-infusion MS or cellular imaging MS techniques in their current form lacks the resolution to demonstrate organelle-to-organelle heterogeneity. There are many reasons to characterize the lipid composition of organelles at high resolution (Horn and Chapman, 2011). Identifying lipid compositional heterogeneity may help to explain compartmentation limitations in metabolic engineering strategies, and determining the heterogeneity of organelle membranes may affect our current understanding of membrane structure and function. Certainly, localized changes in lipid signaling molecules at the subcellular level may be unmasked by DOMS that would otherwise go undetected when analyzing lipids in whole-tissue extracts, and it may be possible to answer questions regarding metabolic channeling at the subcellular level through the use of DOMS.

image

Figure 1.  Determination of subcellular lipid droplet heterogeneity by direct organelle mass spectrometry. (a) Bright-field image reproduced from Horn et al. (2011) and cartoon model of purified seed lipid droplets comprising a phospholipid monolayer (green) with a core of TAG molecules with different acyl chains [for example, blue represents linoleic acid (L, 18:2), yellow represents oleic acid (O, 18:1), and red represents palmitic acid (P, 16:0)]. The nanospray capillary tip can be seen near the center of the image and has a 1 μm opening. Scale bar = 5 μm. (b) Representative spectrum of a single wild-type cottonseed lipid droplet analyzed by DOMS, with TAG species labeled according to their acyl chain composition. Reproduced from Horn et al. (2011), permission requested. (c) Total fatty acid (top) and TAG molecular species (bottom) composition of single wild-type cottonseed lipid droplets analyzed by DOMS. Data re-plotted from Horn et al. (2011). (d) Structural model of a Tri-18:1-TAG molecule represented as a rectangular prism, where length (l) is the distance between the carbon at the omega end of the longest acyl chain and the glycerol backbone, height (h) is the length of the glycerol backbone, and width (w) is the thickness of atoms using van der Waals’ radii if modeled as hard spheres.

Download figure to PowerPoint

Another interesting concept that arises from DOMS analysis is measurement of the amount of total neutral lipid that is packaged into a single LD. Using the infusion flow rates for the nanospray capillary (approximately 100 nl min−1 for a 4 μm diameter tip opening), the analysis spray time (i.e. 1.1 min for a single wild-type LD of approximately 1 μm diameter), and the concentration of a spiked internal standard (i.e. 2 μm Tri-15:0-TAG), the estimated number of moles of 15:0-TAG for an overall MS analytical run is 200 fmol. By normalizing the TAG peak areas to the spiked standards peak area, with correction for isotopic overlap, the sum of the number of moles of TAGs for a single LD is estimated to range between 9.9 and 50.9 fmol, or approximately 6.0 × 109–3.1 × 1010 molecules of TAG in a single isolated cottonseed LD. Given the actual range of LD sizes of slightly more than 1 μm diameter, this finding fits well with a theoretical estimate of the number of TAG molecules in a LD, if a few assumptions are made about the space occupied by a single molecule of TAG. For a 1 μm diameter spherical LD, the internal volume is 0.52 μm3. Assuming that a TAG molecule (Figure 1d) approximates the shape of rectangular prism on a flattened plane, where the length (l) is the distance between the carbon at the omega end of the longest acyl chain and the glycerol backbone, the height (h) is the length of the glycerol backbone, and the width (w) is the thickness of atoms using van der Waals’ radii if the atoms are modeled as spheres, a Tri-18:1-TAG molecule with a length (l) of 1.97 nm, height (h) of 0.5 nm, and width (w) of 0.34 nm (van der Waals’ radius of carbon), has a size of approximately 0.39 nm3. Thus an estimation of the number of theoretical Tri-18:1-TAG molecules that could fit into a single LD with 1 μm diameter is 1.34 × 109, reasonably close to the number determined using DOMS. The marked sensitivity of MS makes lipidomics analysis at the subcellular level readily achievable.

Chloroplast galactolipid analysis

LD analysis by DOMS is of interest in its own right, but may DOMS be extended to other subcellular compartments? Here we demonstrate the determination of galactolipid composition in isolated chloroplasts of spinach leaves (Figure 2). Chloroplasts were visualized directly by both bright-field and epifluorescence microscopy, and five chloroplasts were selected, drawn into the capillary tip and analyzed by DOMS. Galactolipid molecular species and lipid-soluble pigments were easily resolved in these spectra, suggesting that chloroplast analysis at a single-organelle level may be possible with DOMS. Are all chloroplasts functionally similar and equally susceptible to similar stresses, and do these functional aspects of this organelle relate to the compositional differences among chloroplasts? While many such questions remain open, a new tool in the micro-analysis of cells (DOMS) will help to provide information about cellular compartments at a whole new scale of biochemistry. In fact, the expansion of DOMS to lipidomics analysis of other organelles is easy to envisage, but it is likely that this approach could also be modified for in-tip protein digestion and micro-proteomics analysis by nanospray MS. In this way, DOMS would allow analysis of protein compositions at an organelle level.

image

Figure 2.  Imaging subcellular chloroplast lipid composition by direct organelle mass spectrometry. (a) Bright-field image of purified chloroplasts from spinach leaves and a nanospray capillary with a tip opening of approximately 5 μm. Scale bar = 20 μm. (b) Epifluoresence image of purified autofluorescent chloroplasts from spinach leaves. Scale bar = 20 μm. (c) Representative mass spectrum of a set of five chloroplasts sampled by DOMS. Selected major galactolipid and lipophilic pigment species are labeled, including monogalactosyldiacylglycerol (MGDG) species with acyl chain compositions of 34:3 (18:3/16:0), 34:6 (18:3/16:3), 36:6 (18:3/18:3) and 36:3 (18:3/18:0), digalactosyldiacylglycerol (DGDG) species with acyl chain compositions of 36:6 (18:3/18:3) and 36:3 (18:3/18:0), and chlorophyll/pheophytin species.

Download figure to PowerPoint

Other isolated organelle membranes may be amenable to DOMS analysis, including mitochondria, plasma membranes, endoplasmic reticulum, etc. Care would need to be taken to ensure organelle purity and unequivocal membrane identification, as with all fractionation-based approaches. It may be that microfluidics devices could be developed that would couple subcellular fractionation and DOMS applications for certain sample types. Even if these developments are in the distant future, the ability to profile lipids in individual subcellular compartments substantially increases the resolution at which biochemical studies can be performed.

Single-cell video MS

Similar analytical principles to DOMS have been developed by Tsuyama et al. (2008, 2011) in a technique that they termed video MS and applied to analysis of compounds from a single plant cell (Lorenzo Tejedor et al., 2009). Gold-coated glass nanospray tips were used to select cellular extracts from leaf and leaf stalks of geranium (Pelargonium zonale) using a micro-manipulator connected to a syringe via tubing. Over 1000 MS peaks were detected from cellular extracts. Based on the m/z reported in these studies, most of these peaks appeared to correspond to small metabolites, including geranic acid, α-terpineol and caffeine. As the ionization solvent selected (acetonitrile containing 0.5% formic acid) is not typically used for analysis of plant membrane and storage lipids, it is likely that the lipid compounds may not have been efficiently solubilized or ionized before analysis, or were at concentrations far below the other more abundant metabolites. Nonetheless, direct probing into plant cells using manipulation devices compatible with MS analysis shows great promise for future (sub)cellular lipidomic analysis.

Other Potential Approaches for Plant Lipidomics in the Future

  1. Top of page
  2. Summary
  3. Introduction
  4. Lipidomics in Tissues
  5. MASS SPECTROMETRY IMAGING: Localizing Lipids In Situ
  6. Subcellular Lipidomics
  7. Other Potential Approaches for Plant Lipidomics in the Future
  8. Acknowledgements
  9. References

Secondary ion mass spectrometry

Secondary ion mass spectrometry (SIMS) is particular well-suited for visualizing lipid compositions at much higher average resolutions than other contemporary imaging techniques. SIMS is based on use of an accelerated primary ion beam directed at a sample that, upon impact with the surface, generates secondary ions that are analyzed by MS (Boxer et al., 2009). As the ion beam source is not limited by the diffraction of light, resolutions of approximately 100 nm are possible using specialized instruments (Boxer et al., 2009), with routine resolution of 400 nm to 1–2 μm being achievable (Brunelle and Laprévote, 2009). Although SIMS has been available for several decades, recent developments in ion sources have increased the upper mass limit of approximately 200 to >1000 m/z, which is necessary for imaging lipids (Johansson, 2006). Similar to other imaging techniques, SIMS is performed under vacuum, so sample preparation is important to the integrity of the analytes visualized (Boxer et al., 2009). SIMS depth profiling using C60 buckminsterfullerene primary ions has been used to generate 3D profiles of phosphocholine and cholesterol in freeze-dried oocytes (Fletcher et al., 2007), which is not possible using other imaging techniques at present. There are several other applications of TOF-SIMS (for review, see Passarelli and Winograd, 2011), for example in determining the distribution of phosphocholine and sphingolipid head groups in cultured mouse neurons (Yang et al., 2010) and that of glycerolipids in microbial mats (Thiel et al., 2007). Conveniently, newer SIMS instrumentation appears to preferentially visualize lipids from biological tissues, making it a potentially attractive technique for future plant lipid research (Johansson, 2006). There have been a limited number of SIMS applications in plant tissues to date possibly due the high cost of instrumentation. SIMS analysis of arsenic and silicon in rice showed that arsenic was sequestered in roots, which is important to know when developing strategies to reduce the arsenic concentration in rice (Moore et al., 2011). In another study, SIMS showed that polyketide flavonoids were present at increased concentrations in seed coats of peas (Pisum sativum) and Arabidopsis thaliana, with mutants showing variations in flavonoid content (Seyer et al., 2010). SIMS has also been recently used to directly detect and image lipid membranes in both synthetic membrane model systems and biological tissues at high resolution (Johansson, 2006). One particular problem with SIMS, as with other indiscriminate cellular/subcellular MS approaches, is correct identification of specific organelle membranes in images. This may be overcome if the organelle membrane morphology lends itself to identification (such as thylakoid membranes) or if specific biochemical markers can be incorporated into the analysis (such as chlorophyll a for chloroplasts), but this an area that requires further attention as the resolution of chemical mapping makes imaging smaller and smaller subcellular particles possible.

Ion mobility MS

Another under-utilized aspect of MS analysis that will substantially improve the resolution and ability to determine plant lipid compositions is inclusion of an ion mobility cell (see Kanu et al., 2008 for review). Ion mobility is the motion of gas-phase ions in a pressurized chamber, unlike in MS, where the separations take place under vacuum. Under low electric field conditions, the time for an ion to move through an ion mobility drift cell is related to its collision cross-sectional area (i.e. its physical structure) and interaction with the gas. This form of separation can be coupled with conventional MS to provide additional resolution and structural information about the analytes, especially for isobaric ions (Kanu et al., 2008). Differences in the drift times of lipids due to their structural characteristics, including the length of acyl chains, the number of double bonds, head group and degree of ionization, produce an additional dimension of molecular resolution (Jackson et al., 2008). Determination of the drift time properties of individual phospholipid molecules (including soybean PCs) by ion mobility coupled to a MALDI source showed a marked improvement in mass resolution when applied to complex phospholipid mixtures. The phospholipid drift time decreases relative to the degree of acyl chain unsaturation, independently of the head group and m/z, at an approximately linear rate (Jackson et al., 2008). As the unsaturation state results in bending of the acyl chains, ion mobility may provide additional resolution of two isobaric species whose double bonds result in different bending patterns. Phospholipid head groups that are larger (hence a larger cross-sectional area) tend to have longer drift times than phospholipids with smaller head groups, and this difference may be used to reconcile the m/z assignment in complex mixtures (Jackson et al., 2008). Inclusion of cations (e.g. H+, Na+ and Cs+) shifts the drift time m/z space for a set of compounds, and is necessary to uncover a number of additional compounds present in a mixture whose presence was masked using traditional MALDI TOF MS(Jackson et al., 2008). This additional dimension of analysis through ion mobility is an important development in visualizing the lipidomes of biological organisms. IM methodology has mostly been applied to mammalian tissue extracts (Trimpin et al., 2009; Dwivedi et al., 2010), but will be important for plant lipid research in the future.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Lipidomics in Tissues
  5. MASS SPECTROMETRY IMAGING: Localizing Lipids In Situ
  6. Subcellular Lipidomics
  7. Other Potential Approaches for Plant Lipidomics in the Future
  8. Acknowledgements
  9. References

Research on lipidomics in the authors’ laboratory is supported in part by grants from the US Department of Energy Office of Science (Biological and Environmental Research, DE-SC0000797 and Basic Energy Sciences, DE-FG02-05ER15647). We gratefully acknowledge the contributions of Dr Guido Verbeck’s laboratory (Department of Chemistry, University of North Texas, Denton, TX) in the development of direct-organelle mass spectrometry and associated applications.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Lipidomics in Tissues
  5. MASS SPECTROMETRY IMAGING: Localizing Lipids In Situ
  6. Subcellular Lipidomics
  7. Other Potential Approaches for Plant Lipidomics in the Future
  8. Acknowledgements
  9. References