METABOLOMICS ANALYSIS OF LIPID METABOLIZING ENZYME ACTIVITY

Abstract

Lipids exert key structural, metabolic, and signaling functions in cells. Lipid diversity found in cells and tissues is regulated principally by metabolic enzymes whose activity is modulated post-translationally to shape head group and fatty acyl composition of membrane lipids. Methodologies capable of monitoring in vivo changes in the lipidome are needed to assign substrate specificity of metabolic enzymes, which represents a key step towards understanding structure-function of lipids in living systems. The resulting lipid annotations also serve as important biomarkers for understanding mode of action for pharmacological agents targeting metabolic enzymes in cells and animal models. In this chapter, we describe a general metabolomics workflow to complement (chemo)proteomic efforts to modulate lipid pathways for basice science and translational applications.

1. INTRODUCTION

Development of potent and selective small-molecule inhibitors has been expedited by recent advancements in chemical biology and modern analytical chemistry that enable rational small molecule design and quantitative profiling of target inhibition [1, 2]. In vivo assessment of inhibitor efficacy has recently been accomplished with the introduction of activity-based protein profiling (ABPP) for parallel evaluation of compound potency and selectivity directly in native biological environments [3, 4]. While ABPP is a powerful tool for validation of target engagement in vivo, additional bioanalytical platforms are needed to fully elucidate mode of inhibition of lead pharmacological agents. Complimentary mass spectrometry metabolomics methods, for example, provides a means for understanding cellular changes in metabolism in response to pharmacological and genetic perturbation of proteins of interest [5, 6]. Metabolomics has proven to be an effective bioanalytical technique for the determination of enzyme inhibition through quantitative measurement of its related metabolites to establish biomarkers for probe and drug outcome [7].

Lipids constitute a relatively unexplored, yet important class of metabolites for measuring the therapeutic potential of enzyme targets and pathway perturbation following inhibition [8]. Many promising therapeutic targets are involved in regulation of lipid metabolism due to the central role lipids play in regulation of cell signaling [9, 10]. Cellular lipids possess intricate structural complexity resulting in a large number and vast diversity of structures through combinations of head groups and fatty acyl chains. Additionally, lipids are tightly regulated through metabolism and produced on-demand in response to cell activation, which results in dynamic changes in cellular lipid compositions and levels [11]. Liquid chromatography-mass spectrometry (LC-MS) analysis of lipids (i.e. lipidomics) offers a solution to analytical and technical challenges associated with functional studies of lipids[12]. Lipidomics can be adapted for different analytical objectives depending on the target lipid class, biological sample processing method, MS acquisition mode, and the pharmacological probe being tested [13]. As technical accessibility and needs will vary based on the studies, this chapter will focus primarily on the general application of lipidomics for measuring lipid changes by evaluating several sample preparation techniques, online resources, and post-acquisition data analysis software packages for interpreting data and drawing conclusions on cellular lipid metabolism in response to pharmacological or genetic perturbation.

2. OVERVIEW AND COMPARISON OF LIPID EXTRACTION TECHNIQUES

2.1. BROAD LIPID CLASS EXTRACTION (FOLCH METHOD, BLIGH and DYER METHOD)

The analysis of lipid species within tissues and cells requires robust methods that are quantitative, rapid, and efficient. Two of the most widely used lipid extraction techniques in the field of lipidomics are the Folch [14] and Bligh and Dyer methods [15]. Both methods utilize chloroform and methanol for the extraction; however, the ratios of the solvents differ. These methods are designed for total lipid extraction, which includes phospholipids as well as neutral lipids including diacylglycerols and triacylglycerols [16, 17].

The extraction method that was developed and published in 1957 by Jordi Folch is one of the most widely used lipid extraction methods that can be adapted for quantitative analyses. The “Folch method” uses three solvents that include chloroform, methanol, and water. The bi-phasic extraction system yields selective separation of lipids and non-lipid species into two phases (chloroform and methanol/water layers, respectively). The Folch method improved on its predecessor by allowing rapid extraction and isolation of non-lipid substances into the methanol/water layer whereas lipids were extracted into the chloroform layer. The method has been widely employed for lipid metabolic analysis of diverse tissue types (brain, liver, muscle) [14].

The Folch method calls for a 2:1 (v:v) chloroform:methanol ratio and a final 4:1 organic solvents: water ratio. The mixture of all three solvents generates two phases consisting of a top aqueous later and bottom organic layer. Analysis of the layers showed a top layer solvent composition of 3:48:47 chloroform:methanol:water and 86:14:1 for the bottom layer. One of the caveats for using this method is the volume of the organic phase relative to that of the sample used for the extraction. In analyzing various biological tissues that contain substantial amounts of lipids, this method requires an organic solvent mixture volume of at least 20 times higher than that of the sample. This particular stipulation was one of the main incentives for development of the “Bligh and Dyer” extraction method [15] (see below). Nevertheless, the Folch method remains a mainstream and widely accepted lipid extraction method with broad applications for lipidomics analysis [18–20].

The Bligh and Dyer method was reported in 1959 by E.G. Bligh and W.J. Dyer [15], and it was developed originally to investigate lipid compositions in fish. The primary challenges in advancing this field of lipid analysis was development of an extraction technique that could be adapted for rapid and quantitative analysis of lipids. The necessity of a rapid extraction method for fish lipid analysis is due to the relatively high composition of polyunsaturated fatty acids (PUFAs) in fish combined with possible elevated temperatures or exposure to an oxidative environment that could alter these labile lipids and thereby confound accurate lipid analysis.

The Bligh and Dyer method uses a reduced amount of solvent for lipid extraction. In brief, tissues are homogenized in a mixture of chloroform and methanol, and the mixed homogenate results in a monophasic solution. Next, water and chloroform are added until a biphasic system is produced. The ratio of chloroform and methanol used for the extraction is 1 part chloroform and 2 parts methanol. The ratios of the resulting solution before and after the dilution are 1:2:0.8, and 2:2:1.8. Therefore, the resulting ratio of the sample to the total solvent volume is 4:1 compared to 20:1 used for Folch extractions. The subsequent analysis of the total residual lipids left in the aqueous/methanol phase as well as those in the tissue residue showed that the total extracted lipids in chloroform accounts for approximately 94% of the total extractable lipids. Reports utilizing both Folch and Bligh and Dyer methods for lipid analysis have shown good extraction efficiencies compared to other techniques [13, 19, 21–23]

2.2. TARGETED LIPID EXTRACTION TECHNIQUES (HIP, BUTANOL, MTBE METHOD)

For certain applications, the presence of high-abundance lipid species can hinder LC-MS detection of lower abundance species when these lipids co-elute and higher abundance lipids suppress the ion signals of targeted lower abundance lipids. One potential solution is to modify extraction techniques in an effort to isolate lipid classes of interest and thereby enrich lower abundance lipids. Various targeted extraction techniques have been developed to optimize extraction of certain lipid classes based on their physicochemical properties and by partitioning into organic extraction layers with varying degrees of polarity.

Hara and colleagues published a method that employs a 3:2 (v:v) hexane/isopropanol (HIP) solvent system preferentially extract specific lipid classes [24]. HIP efficiently extracted triacylglycerols, cholesteryl esters, and free fatty acids while excluding phosphoinositides, lysophospholipids, and cholesterol sulfates. This lipid selection could be due to the greater hydrophobicity of hexane and therefore improved retention of select lipids in the organic phase as opposed to chloroform in alternative, more commonly-used systems. Additionally, it was shown by Guckert and colleagues that the HIP extraction method could preserve the formation of native nucleopore membranes from microbiota whereas the chloroform-based systems would solubilize the polycarbonate membranes, thereby losing its biological context [25]. The HIP method has been shown to be a viable alternative to the traditional chloroform biphasic systems while offering distinct advantages in sampling discrete lipid classes.

Butanol extractions have also been explored as an alternative to chloroform-based methods. Cham and colleagues showed that a 40:60 (v/v) butanol/di-isopropyl ether solvent system was an efficient means for extracting triglycerides, cholesterol, phospholipids, and fatty acids [26]. However, the method was shown to require large volumes of solvents, restricting the scale of extraction procedures. Löfgren and colleagues demonstration that a 3:1 (v/v) butanol/methanol mixture with 1% acetic acid (BUME) could improve extraction efficiency for microliter sample quantities while improving extraction of additional sphingolipid and diacylglycerol lipids [27].

Matyash and colleagues developed a methyl-tert-butyl ether (MTBE) system for high-throughput lipidomic analyses [22]. The MTBE method showed improved extraction efficiency potentially due to the ability for isolating the organic phase as the upper layer. A potential drawback compared with other chloroform methods is increased aqueous content in the organic phase, which could introduce contamination between phases. Recently it was shown that replacement of 1% acetic acid with LiCl produced comparable yields of select lipid classes while simultaneously reducing the risk of hydrolytic artifacts from use of an acidic extraction environment [28]. Additionally, the use of LiCl expanded the variety of biological samples amenable to this extraction method broadening lipidomic investigations of lipids in relevant biofluids, tissues, and cultures.

3. TOTAL LIPID EXTRACTION USING FOLCH AND BLIGH AND DYER METHODS

The methods for quantitative lipid extraction from biological samples mentioned above enables opportunities for studying lipids in biological systems and, specifically, how these metabolites are modulated in response to pharmacological and genetic manipulations. The extraction techniques described here (Folch, Bligh and Dyer) take advantage of biphasic solvent separation (Fig 1). The extraction steps will result in three distinct layers consisting of a bottom chloroform layer, a top aqueous layer, and a protein layer at the interface. The extraction steps include addition of butylated hydroxytoluene (BHT), which is an anti-oxidant that minimizes oxidation of lipids that contain unsaturated fatty acids. The Bligh and Dyer and Folch method are both capable of extracting a wide variety of lipid classes.

Figure 1. Two-phase solvent extraction workflow for lipid analyses.

Tissue sample lipid extraction protocol is described in a step by step manner. Tissues are dounced in water until the tissue is completely homogenized in the douncer (A). The samples are transferred to a dram vial with the appropriate organic solvent ratio composition (chloroform/methanol). Addition of aqueous homogenate to the organic mixture separates the chloroform and methanol into biphasic (two) layers (B). After centrifugation, biphasic layers are separated into three distinct layers (top aqueous/methanol layer, middle protein disk, and bottom chloroform layer) (C). The bottom chloroform layer is transferred to a clean dram vial (D). The aqueous layer is extracted with organic solvent once more, and the resulting two organic layers are combined. Combined organic layers are dried under nitrogen stream (E). Dried lipid extracts are re-suspended in 1:1 isopropanol/methanol and stored at −80°C until further analysis.

In general, the Folch method may be more suitable for extraction of saturated and long-chain glycerolipids and cholesterol esters compared with the Bligh and Dyer method. Additionally, extraction of lipids into the organic phase may be more technically straightforward due to the larger volume of solvents used in the Folch method (2:1 CHCl₃/MeOH vs. 1:2 CHCl₃/MeOH). However, the Bligh and Dyer method can accept modifications to the method (e.g. addition of acid/base, salts, etc.) because of the increased water content, providing improved solubility of more polar lipids that benefit from these changes. In summary, the Folch method may have some advantage for extracting highly hydrophobic lipids including triglycerides and cholesterol esters while the Bligh and Dyer method (specifically the acid-modified version) is generally applicable for all other lipid classes.

3.1. Equipment

1. Nitrogen gas stream for solvent evaporation

2. Sorvall ST 40R centrifuge

3. Thermo Scientific Q-Exactive Plus mass spectrometer (MS)

4. Dionex Ultimate 3000 RS UHPLC

5. Kinetex® 1.7 μm C18 100Å, LC column 100 × 2.1mm

6. Dounce homogenizer for cell and tissue lysis

7. 2 Dram vials for phase separation and extraction

8. Pasteur pipettes

3.2. Chemical/Solvents

1. HPLC grade H₂O or water (Catalog #: W5–4, Fisher Chemical)

2. HPLC grade CHCl₃ or chloroform (Catalog #: C607–4, Fisher Chemical)

3. HPLC grade CH₃OH or methanol (Catalog #: A454SK-4, Fisher Chemical)

4. 2,6-Di-tert-butyl-4-methylphenol (BHT) (Catalog #: 112990010, Acros Organics)

5. HPLC grade Isopropanol (Catalog #: A461–4, Fisher Chemical)

6. HPLC grade Acetonitrile (Catalog #: A955–4, Fisher Chemical)

7. Ammonium Formate (Catalog #: 14517–18, Alfa Aesar)

8. Formic Acid (Catalog #: 85178, Thermo Scientific)

9. HPLC grade Hexanes (Catalog #: H302–4, Fisher Chemical)

10. HPLC grade Acetone (Catalog #: A929–4, Fisher Chemical)

4. Protocol

4.1. Cell/Tissue sample extraction (Folch)

1. Prepare a 2:1 (v:v) chloroform/methanol solution with BHT (50 μg/mL). Make BHT stock in methanol and add appropriate amount of chloroform to make the 2:1 solution. The total volume of solvent is dependent on the total number of samples analyzed.

2. Prepare chosen deuterated lipid standards (diluted to 2 pmol/μL concentrations) in a 2:1 chloroform/methanol solution (containing BHT) in separate dram vials (1 mL per sample). The concentration of each standard should be ~10 pmol/3 mL final volume. We recommend making a 100X stock solution containing all desired standards so that a single transfer is performed to the extraction mix. Synthetic lipid standards chosen are dependent on the lipid classes evaluated. In general, a cocktail of diverse lipid classes is recommended to increase the coverage of lipidomics analysis.

   *For example, if you have 3 samples, you need to prepare 12 mL of 2:1 chloroform/methanol (3 samples + 1 extra). Next, you will add 120 μL of 100X lipid stock (containing lipid standards at 1000 pmol/3 mL concentration) to extraction mix.

   *Deuterated lipid standards can be purchased from Cayman Chemical (https://www.caymanchem.com/Home) and Avanti Lipids (https://avantilipids.com/).

3. Add 3 mL of extraction mix (containing lipid standards) from (2) to each 6 mL dram vial.

4. For cells, re-suspend samples in 1 mL HPLC grade water and keep cold. For tissues*, measure tissue weights by placing the tissues in liquid N₂, tare the scale with an empty microcentrifuge tube, weigh tissue, and quickly return the sample to liquid N₂. Prepare a 15 mL Dounce homogenizer (7 mL maximum volume) by first pre-cleaning the Dounce homogenizer in the following order: hexanes, acetone, and HPLC water. Adjust volume of cold HPLC water so that all tissues are normalized by weight to 10 mg/mL (e.g. 2 mL for 20 mg, and 1 mL for 10 mg of tissue so both samples are 10 mg/mL) and homogenize tissue immediately (Fig 1A).

   *samples should be kept frozen in liquid nitrogen until just before addition of extraction solvent.

5. Transfer 1 mL of (4) to each dram vial containing 3 mL of 2:1 chloroform/methanol solution.

6. Vortex samples to extract lipids and place on ice for 20 minutes (Fig 1B).

7. Centrifuge sample vials at 2,000 × g for 5 minutes.

8. You should observe phase separation at this stage and protein “disk” at the interface (Fig 1C).

9. Transfer the bottom organic layer using a Pasteur pipette into a new set of dram vials (Fig 1D).

10. Add an additional 1.6 mL of 2:1 chloroform/methanol solution (without lipid standards) to each original vial with the residual aqueous layer from the previous step and vortex.

11. Centrifuge vials at 2,000 × g for 5 minutes, re-extract the organic layer, and combine with the previous organic layer set aside in step 9.

12. Dry organic layers under nitrogen stream (Fig 1E) and re-suspend lipids in 120 μL 1:1 isopropanol/MeOH (Fig 1F). Store lipid samples at −80°C until ready for LC-MS analysis.

4.2. Cell sample extraction (Bligh and Dyer)

1. Prepare a 1:2 chloroform/methanol solution with BHT (50 μg/mL). Make BHT stock in MeOH and add appropriate amount of CHCl₃ to make 1:2 solution. The total volume of solvent is dependent on the total number of samples analyzed.

2. Prepare chosen deuterated lipid standards (diluted to 2 pmol/μL concentrations, Cayman Chemical and Avanti Lipids) in a 2:1 chloroform/methanol solution in separate dram vials (1 mL per sample). The concentration of each standard should be ~10 pmol/3 mL final volume. We recommend making a 100X stock solution containing all representative lipid standards so that a single transfer is performed to the extraction mixture.

   *For example, if you have 3 samples, you need to make 12 mL of 2:1 chloroform/methanol (3 samples + 1 extra). Next, you will add 120 μL of 100X lipid stock (containing lipid standards at 1000 pmol/3 mL concentration) to extraction mixture.

3. Add 3 mL of extraction mixture (containing lipid standards) from (2) to each 6 mL dram vial.

4. For cells, re-suspend samples in 1 mL HPLC grade water and keep cold. For tissues*, measure tissue weights by placing the tissues in liquid N2, tare the scale with an empty microcentrifuge tube, weigh tissue, and quickly return the sample to liquid N2. Prepare 15 mL Dounce homogenizer (7 mL maximum volume) by rinsing the sample in the following order: hexanes, acetone, and HPLC water 3 mL each. Adjust volume of cold HPLC water so that all tissues are normalized by weight to 10 mg/mL (e.g. 2 mL for 20 mg, and 1 mL for 10 mg of tissue so both samples are 10 mg/mL) and homogenize tissue immediately (Fig 1A).

   *samples should be kept frozen in liquid nitrogen until just before addition of extraction solvent.

5. Transfer 1 mL of (4) to each dram vial containing 3 mL of 1:2 chloroform/methanol solution.

6. Vortex samples to extract lipids and place on ice for 20 minutes (Fig 1B).

7. Centrifuge sample vials at 2,000 × g for 5 minutes.

8. You should observe phase separation at this stage and a protein “disk” at the interface (Fig 1C).

9. Transfer the bottom organic layer using a Pasteur pipette into a new set of dram vials (Fig 1D).

10. Add an additional 1.6 mL of 1:2 chloroform/methanol solution to each original vial with the residual aqueous layer from previous step and vortex.

11. Centrifuge vials at 2,000 × g for 5 minutes, re-extract the organic layer and combine with the previous organic layer set aside in step 9.

12. Dry organic layers under nitrogen stream (Fig 1E) and re-suspend lipids in 120 μL 1:1 isopropanol/methanol (Fig 1F). Store lipid samples at −80°C until ready for LC-MS analysis.

4.3. LC-MS/MS instrument operation and method development (Q-Exactive Plus Orbitrap MS)

1. Check that LC solvent reservoirs and gas cylinders are full before performing instrumental analyses.

2. Replace sweep plate and ion transfer tube in the MS instrument ion source with a set that was cleaned using the following protocol of 15 min sonication steps for each: 100% water, 50% water and 50% methanol, 100% isopropanol with 5% formic acid, and 100% dichloromethane, followed by thorough drying.

3. Perform a calibration of the mass spectrometer for both positive (Thermo Fisher Scientific, Catalog #: PI-88323) and negative ion modes (Thermo Fisher Scientific, Catalog #: PI-88324) before each analysis using noted respective calibration mixtures.

4. Take the lipid samples from frozen storage, thaw, and aliquot into clean amber vials with glass inserts. Place the sample vials in the autosampler ensuring that the vial used for “blank” washes of the LC column is filled with chosen solvent (isopropanol).

5. Perform evaluation of instrument performance for the LC separations (lipid standards; retention times, peak widths) and MS data (standard lipid extract; mass accuracy, mass resolution, lipid molecular ion abundance).

6. Perform LC separation of lipids using a reverse-phase gradient and an appropriate LC column (Kinetex® 10 cm length, 2.1 mm i.d., packed with 1.7 μm diameter 100 Å pore size C18 particles). The gradient conditions used to produce an example MS1 total ion chromatograms (TIC) and MS2 spectra in Figure 2 are as follows: 35-min reverse-phase LC gradient (Mobile phase A: 50% ACN, 50% H2O, 0.1% formic acid, 10 mM ammonium formate; Mobile phase B: 10% ACN, 88% isopropanol, 2% H2O, 0.1% formic acid, 10 mM ammonium formate) with the following steps: 0–4 min 65% to 40% A, 250 nL/min; 4–12 min 40% to 15% A, 250 nL/min; 12–21 min 15% to 0% A, 250 nL/min; 21–24 min 0% A, 250 nL/min; 24–24.1 min 0% to 100% A, 250 nL/min; 24.1–27 min 100% to 0% A, 250 nL/min; 27–30 min 0% to 100% A, 250 nL/min; 30–33 min 100% to 0% A, 250 nL/min; 33–35 min 0% to 65% A, 250 nL/min.

7. *Prepare the MS instrument method for Top10 data-dependent MS-MS/MS (ddMS2) data acquisition, in which 1 cycle consists of 1 Full MS scan followed by 10 MS2 scans of the most abundant ions recorded from that preceding MS1 scan (Fig 2A). A dynamic exclusion of 8 seconds was used to ensure that the instrument selects different abundant ions for MS2 scans thereby minimizing redundant analyses.

   *Prepare an MS instrument method for parallel reaction monitoring (PRM) analysis by first designing an inclusion list containing all necessary precursor ion adduct m/z values that will be selected for MS2. Accurate identifications and quantifications of desired lipid ions should be compared to spiked in lipid standards of similar chemical structure (Fig 2B).

8. Include in both MS methods acquisition of data from blank samples (LC column and plumbing washes) between each sample injection and ensure that necessary biological and technical replicates are included in the batch sequence.

Figure 2. Example LC-MS data acquisition from extracted cellular lipidomes.

Shown are examples of MS1 total ion chromatograms (TIC) and MS2 spectra acquired in positive and negative ion modes using the methods described in section 4.3. (A) Example MS1 TIC and MS2 spectra in both positive and negative mode collected from lipid samples following Bligh and Dyer extraction from HEK293T cells, resolved by reverse-phase UHPLC, and analyzed using a Top10 ddMS2 on a Q-Exactive Plus Orbitrap MS instrument. (B) Example MS1 chromatogram and MS2 spectra of selected deuterated lipid standard (1-stearoyl-2-arachidonyl-sn-glycerol-d8, Cayman Chemical Company) detected using parallel reaction monitoring conditions and parameters.

5. Data analysis

5.1. Bioinformatics and analytical software

Significant improvements in mass spectrometers such as the benchtop Orbitrap MS instruments permit rapid isolation and fragmentation of eluting ions from reverse-phase chromatography with high resolution, accurate mass to broaden analyses of biological molecules from proteins to small molecules. When lipid extracts are chromatographically resolved using reverse-phase ultra-HPLC (UHPLC) systems, varying lipid classes are eluted at distinct and characteristic retention times. The eluting lipid species at any given time are ionized by electrospray ionization and analyzed using a mass spectrometer. The basic operating procedure of online LC-MS instruments consists of LC separation complex mixture components, ionization of eluting molecules, measurement of molecular ion masses, isolation of selected molecular ions, fragmentation of those precursor ions with HCD (higher-energy collisional dissociation), and finally measurement of resulting fragment ions. The combination of LC retention time and accurate molecular precursor and fragment ion mass measurement at high resolving power enables identification of the precursor molecular ions. Commercial software, as well as freeware made available by the research community, can facilitate identification and quantitation of lipid species, which enables determination of lipid changes in different biological samples. For example, LipidMAPS (Lipid Metabolites and Pathway Strategy) was supported by the National Institute of Health (NIH) to develop databases and resources that can be used for lipid research using LC-MS techniques.

5.2. Broad lipidome analysis using LipidSearch^™

LipidSearch^™ from Thermo Fisher Scientific is an automated lipid identification and relative abundance quantitation platform for interpreting MS data acquired by Thermo Scientific brand mass spectrometers. It is highly encouraged to manually analyze raw MS data to ensure proper data acquisition and global lipid detection. The LC performance can be evaluated by measuring the LC retention times and peak widths and shape (symmetrical and not tailing) of internal standards or selected analytes that are commonly detected and comparing these values with those measured in standards-only analyses and/or other reference data. Evaluate the MS1 performance by measuring LC peak area abundance of reconstructed ion chromatograms for selected internal standard m/z ions. Confirm that the measured m/z values do not differ versus the expected m/z values by more than the instrument mass accuracy specification. Check the signal-to-noise levels in selected MS1 spectra across the LC gradient (data acquisition time) to ensure that there are no unexpected ions or high background signal intensity, which could indicate contaminants in the sample or LC-MS system. This evaluation ensures appropriate detection sensitivity and mass measurement accuracy, as well as provides a measure of sample and LC-MS system cleanliness (with blank injection data).

Evaluate the MS2 (MS/MS) performance by checking fragment ions of selected precursor ions (e.g. internal standards) to ensure satisfactory ion abundances and mass measurement accuracy based on reference data. For a data-dependent acquisition method, check that the maximum number of MS2 spectra (e.g. 10 MS2 spectra per data acquisition cycle in a “Top 10” method) were acquired specifically during the time period when many lipids are eluting, to ensure sampling of low abundance analytes. Check LC-MS data corresponding to blank injections (solvent or internal standards only) to determine if there is carryover of sample analytes.

However, to feasibly perform high-throughput discovery-based experiments, an automated approach that rapidly determines the structural identities and sample abundances of thousands of lipid species is necessary. LipidSearch^™ processes raw files (.raw extension) from lipid sample analyses and compares the recorded experimental MS spectra with their online database of expected fragmentation spectra of 50+ lipid class subtypes considering 10+ ionization adducts (e.g., [M+H]+, [M+NH4]+, [M-H]−). This database is a compilation of high-resolution experimental and in silico lipid fragmentation datasets, as well as validated structures from standards that can be used to make accurate lipid identifications with statistical confidence. By using LipidSearch^™ and other similarly automated programs, analysis time of complex datasets can be dramatically reduced to speed up research without sacrificing quality of data analysis workflows. In the following sections, we describe steps involved in making lipid identifications using LipidSearch^™ and how to interpret results from this bioinformatics workflow.

The first step in using LipidSearchTM is to identify lipid molecules by the measured masses of the precursor and fragment ions contained in MS1 and MS2 spectra, respectively. (Total ion current (TIC) chromatogram consisting of MS1 and MS2 scans, Fig 3A). Lipid identification is accomplished by the “search” component of the program, which requires selection of the target lipid MS database, with which the experimental MS data are compared, ensuring mass resolution and mass measurement accuracy tolerances appropriate to the MS data acquisition modes used. Additional parameters can be set based on the desired lipid species for analysis. These parameters include “Class” and “Ion”, or simply the lipid classes to consider as well as the possible adduct ions that could be formed dependent on the ionization mode and LC conditions. Once these conditions are set, LipidSearchTM will report a table of search results comprised of all lipids the algorithm was able to find from the selected raw data files with an associated grade of confidence in making corresponding lipid assignments. These grades range from A to D based on the presence of diagnostic ions unique to lipids by class followed by fatty acyl chain position. A grade of “A” means that the lipid class and fatty acid structures are completely annotated whereas a grade of “D” denotes that the lipid ID was made by other fragment ions typically shared across classes and subtypes and the best the program can give is a sum composition species from the precursor ion m/z. Each lipid identification will have associated with it a corresponding lipid fragmentation match score that compares the expected fragment ions for the chosen classes and ion adduct selections with the experimental MS2 spectra contained in the raw file (Fig 3B). Filtering to yield only high-grade lipid identifications from these results will give a cohort of species that can be reliably detected using the current LC and MS instruments and methods. Additionally, each lipid result will have the relative abundance of the MS1 peak used to produce the fragmentation spectra read by LipidSearchTM thus, making it an integrated lipid identification and semi-quantitation platform.

Figure 3. Workflow of untargeted lipidomics analysis using LipidSearch^™ software.

The Q-Exactive Plus mass spectrometer acquires MS1 spectra of lipid species from a given sample and selected MS1 peaks are fragmented to give MS2 fragment spectra (A). The fragment spectra are compared against in-silico database that has been curated to contain different classes of lipid species including phosphatidylcholine (PC), phosphatidylserine (PS), triacylglycerol (TGs), any other diverse lipid classes. The software shows overlap of experimental and in silico fragment spectra across different samples (B). LipidSearchTM can align MS1 spectra of the molecular ion corresponding to identified lipids across multiple samples based on the extent of shared fragmentation spectra and chromatographic elution times (C). Annotated lipid species can be viewed based on retention time and mass-to-charge ratio (m/z) associated with each species (D). The intensities of the lipid species that are detected can be used to measure changes between conditions using a box and whisker plot. In addition, intensity data for lipids can be exported for further sample analyses for each experiment (E).

Comparative analyses across samples can also be performed using LipidSearch^™ by using the alignment feature of the software. Aligning the lipid search results for each sample will compare the peak intensities of each identified lipid and can be used to observe changes in relative lipid abundance across the tested samples (Fig 3C). Many features exist in the software platform that allows for both global (Fig 3D) and lipid-specific comparisons (Fig 3E) across the tested conditions and how each condition perturbs the lipidome. However, true quantitative analyses of data should be performed by exporting raw data from LipidSearch^™ and use of matching isotopically labeled lipid standards for normalization to account for sample to sample and MS run to run variability. Nevertheless, for measuring a broad swath of the lipidome, an automated process that yields confident identifications and quantitation of relative abundances of a large number and variety of lipids will dramatically increase post-acquisition data analysis and reliable interpretation of results for testing hypotheses.

5.3. Targeted parallel reaction monitoring (PRM) using TraceFinder^™

For confident, quantitative analyses of lipid samples, a high-throughput analytical workflow process is required to account for the numerous variables prevalent in metabolomics-based experiments. With software packages such as TraceFinder^™ developed by ThermoFisher Scientific it is possible to design master method programs that perform reproducible analyses on batches of lipid samples, ensuring consistent performance of instrumental methods and similar conditions imposed on post-acquisition data processing. The power of TraceFinder^™ however, lies in its ability to integrate lipid species identification with sample-dependent abundance determination based on preset chromatography characteristics (retention time), precursor ion adduct masses (parent ion m/z scan that produces MS1 peaks), and diagnostic fragment ions (product ions in MS2 spectra) for targeted analyses. TraceFinder^™ provides the necessary features required to build a target ion compound database, expected fragment ion patterns, and peak processing to ensure confident identification of lipid species while simultaneously performing quality control on all quantitation conditions. These settings and how to utilize them for lipid analysis will all be discussed in the following sections pertaining to the pre-acquisition, acquisition, analysis, and data review modes of the TraceFinder^™ analytical workflow.

1. Method Setup

Prior to LC-MS analysis of lipid samples, steps must be taken to select the ions that the mass spectrometer will dedicate MS1/MS2 cycle time for detection. Targeted analysis, in comparison to unbiased data-dependent analysis, of selected lipids is necessary to maximize the likelihood of detecting lower abundance lipids that exist at sub-picomolar concentrations [29]. Additionally, isotopically labeled synthetic lipid standards corresponding to lipids of interest are spiked into lipid extracts for quantitative purposes. A narrow precursor ion m/z isolation window of 0.4 Da is used for the MS2 scans to increase targeted ion selectivity and restrict the inclusion of un-desired lipid ion signals [30]. In TraceFinder^™, the strategy for selecting specific lipids for detection and analysis involves the development of a compound database file (.cdb extension). In this database lipid species are distinguished based on their expected precursor ion m/z (mass of lipid +/− the atom adduct; molecular ion mass, e.g., [M+H]+, [M-H]−, [M+NH4]+) and chromatographic retention time (Fig 4A). Once the database has been compiled, it can be associated with any master method, which integrates instrumental acquisition and fragmentation conditions with chromatography gradients and flow rates. With a list of lipid targets, TraceFinderTM can be utilized for quantitative analyses using previously validated instrument methods.

Figure 4. Targeted lipidomics workflow by TraceFinderTM software.

Quantitative analysis of lipids begins with development of a compound database that contains the precursor ion adduct ion masses and retention times that will be connected to a master method containing all necessary instrumental and chromatography parameters (A). Lipid samples are analyzed by LC-MS-MS/MS via a batch sequence format with necessary blank washes and technical replicates (B). Lipid species are identified by TraceFinderTM software through detection of pre-determined MS2 fragment ions within matching precursor m/z MS1 peak scan events taken directly from raw data (C). Satisfactory detection of chosen lipids is observed as green flags in the data review tab of the software and may be used to ensure the reproducibility and quality of the instrumental analysis in real-time (D).

2. Data Acquisition

Lipid samples are analyzed in batch format using TraceFinder^™, subjecting all samples to the same master method and post-acquisition data processing for standardized multiple sample comparative investigations (Fig 4B). Another advantage of the batch format is the immediate performance feedback provided that reveals deviations in internal standards/quality control (QC) data, which allow distinguishing low- and high-quality analyses. Samples that yielded inferior QC data can be reinjected while the current batch sequence is underway. The real-time status pane displays information on the current total ion chromatogram and batch progress. Taken together, the acquisition mode of TraceFinder^™ allows researchers to check the performance and quality of mass spectrometry analyses throughout the data acquisition process, optimizing instrument time and resources.

3. Data Analysis and Review

TraceFinder^™ allows the user to stipulate the fragmentation data criteria required to make a positive identification of the lipid species contained in the compound database. Each lipid species can be validated based on the expected fragmentation ion profile, which often is distinctive of lipid class type, fatty acyl chain composition, and adduct form of the precursor ion (Fig 4C). Following lipid molecule identification (based on precursor and fragment ions), the abundance of respective lipid is determined by calculating the area of the reconstructed ion chromatogram peak corresponding to the precursor m/z value. Flags will be displayed showing the success (green) or failure (red) of detecting the desired compound in each sample offering another means of ensuring consistent quantitative analysis (Fig 4D). Quantitative analysis is performed by comparing with the peak areas of known quantities of internal lipid standards. The batch sample data may be exported as a spreadsheet containing all the relevant elution, precursor ion scan, and peak intensity and quality information. TraceFinder^™ offers users an all-inclusive analytical platform that integrates targeted lipid identification with abundance determination in a user-friendly program to quantitate virtually any class of lipids that is amenable to electrospray ionization (ESI).

6. Summary

As the sensitivity and robustness of bioanalytical technologies improves so too does our capability of developing potent, selective, and efficacious small-molecules designed to inhibit protein function. Many lipid-metabolizing enzymes have been associated with disease progression and thus, make promising therapeutic targets. However, past efforts devoted towards targeting these proteins have been hampered because effects from in vitro and in vivo blockade are not always directly correlated. The emergence of omics-based methods has opened new avenues for researchers to determine the basic science and translational potential of pharmacological agents directed at lipid metabolizing enzymes. Lipidomics is emerging as a critical tool for evaluating the functional impact of inactivating lipid metabolic pathways and the consequent effects on cell biology and mammalian physiology. Rapid lipid extractions combined with powerful mass spectrometry capabilities are driving new discoveries in the burgeoning field of lipidomics. The integration of sample preparation, LC-MS-MS/MS data acquisition, and data processing software into a seamless workflow improves efficiency of investigations, reproducibility of analysis, and exposure of samples to similar restrictions and conditions for analysis. Collectively, these features give researchers more control over variables being tested without sacrificing depth and breadth of analyses. The capabilities of lipidomics methods will continue to improve, which should aid in both laboratory and translational research.