Relative Quantitation of Unsaturated Phosphatidylcholines Using 193 nm Ultraviolet Photodissociation Parallel Reaction Monitoring Mass Spectrometry

Abstract

Structural characterization of glycerophospholipids beyond the fatty acid level has become a major endeavor in lipidomics, presenting an opportunity to advance the understanding of the intricate relationship between lipid metabolism and disease state. Distinguishing subtle lipid structural features, however, remains a major challenge for high-throughput workflows that implement traditional tandem mass spectrometry (MS/MS) techniques, stunting the molecular depth of quantitative strategies. Here, reversed phase liquid chromatography is coupled to parallel reaction mass spectrometry utilizing the double bond localization capabilities of ultraviolet photodissociation (UVPD) mass spectrometry to produce double bond isomer specific responses that are leveraged for relative quantitation. The strategy provides lipidomic characterization at the double bond level for phosphatidylcholine phospholipids from biological extracts. In addition to quantifying monounsaturated lipids, quantitation of phospholipids incorporating isomeric polyunsaturated fatty acids is also achieved. Using this technique, phosphatidylcholine isomer ratios are compared across human normal and tumor breast tissue to reveal significant structural alterations related to disease state.

Graphical Abstract

INTRODUCTION

Glycerophospholipids are a diverse class of biomolecules that engage in an array of cellular roles. Apart from comprising the cellular membrane, these lipids influence cellular functions by modulating membrane protein structure, compartmentalizing organelles, and serving as signaling molecules.^1–4 Consequently, lipid composition is carefully regulated to ensure proper cellular function. Dysregulation in lipid profiles has been associated with several diseases such as cancer, diabetes, neurodegeneration and cardiovascular disease.^5–7 Accordingly, deciphering lipid structures has become imperative to understanding the relationship between disease state and aberrant lipid profiles, promising to inform disease biomarkers, potential therapeutic targets, and disease mechanisms.^8–10 Thus, the development of deep lipidomic strategies that provide both fine structural detail and quantitation at the lipidomic scale has become a major endeavor to unravel how disease and lipid composition are intertwined.

Major advances in lipidomic studies have largely been catalyzed by mass spectrometry (MS) and tandem mass spectrometry (MS/MS) based strategies. High sensitivity, resolution, and throughput afforded by modern MS technologies have enabled the routine quantitation and characterization of glycerophospholipids to provide insight into the structure and function of lipids.^4,8–14 MS-based strategies largely fall into direct introduction (“shotgun”), separation, or imaging workflows, with each approach featuring unique advantages such as rapid analyses, enhanced molecular depth, or spatial resolution, respectively, which together offer tremendous versatility for a range of applications.^4,8–14 From traditional collision-based MS/MS strategies, several crucial structural features can be discerned for glycerophospholipids, including the identity of the glycerophospholipid class as denoted by the phospho-headgroup esterified at position sn-3 of the glycerol backbone as well as the linkages and total number of carbons and double bonds in the acyl chains linked at positions sn-1 and sn-2 of the glycerol backbone.¹⁵ Although these features are informative, structural characterization at this level is typically incomplete, failing to reveal subtle yet important structural features such as double bond position and stereospecific numbering (sn) stereochemistry.¹⁵ Achieving deeper levels of characterization requires the use of alternative approaches and/or derivatization strategies to surpass the characterization achieved alone through conventional collisional activation methods.^8–10,15 In recent years, surpassing these analytical hurdles has become a major goal in structural lipidomic analyses, with each successful development revealing more structural features and underlining the complexity of lipids in cell physiology and disease.

To more comprehensively uncover the structures of glycerophospholipids, several advanced MS techniques have been introduced. Most prominently, several approaches have focused on localizing double bonds (C=C) along the acyl chains as sites of unsaturation influence protein–lipid binding and membrane properties, such as fluidity, packing density, and oxidation chemistry.^16–18 Application of charge switching derivatization,¹⁹ epoxidation with meta-chloroperoxybenzoic acid (m-CPBA),^20,21 or ion/ion reactions^22–25 prior to collisional dissociation as well as the implementation of alternative ion activation techniques, such as electron induced dissociation (EID) of metal adducted lipids^26,27 and metastable atom dissociation (MAD),²⁸ have demonstrated great success in localizing double bonds along the fatty acyl chains. In addition to double bond localization, several approaches even reveal sn stereochemistry. Electron impact excitation of ions from organics (EIEIO),^29,30 radical directed dissociation (RDD)^31,32 induced by bicarbonate adduction and subsequent collision induced dissociation (CID),^33,34 Paternò Büchi derivatization of C=C followed by CID (PB-MS/MS),^35–40 and ozone induced dissociation (OzID) in tandem with CID (CID/OzID² and (CID/OzID)²)^41–45 are advanced MS strategies that in addition to double bond localization, also inform the sn stereochemistry of glycerophospholipids. Notably, combining these strategies with imaging and separation workflows has revealed that these structural features vary relative to disease state. For example, resolving double bond or sn isomers via matrix assisted laser desorption ionization (MALDI) coupled to OzID, PB-MS/MS, or EID has shown that alterations in the relative abundances in glycerophospholipid isomers spatially correlate with tumorous tissue.^27,46–48 Moreover, results obtained from desorption electrospray ionization (DESI) coupled to m-CPBA²⁰ derivatization and liquid microjunction surface sampling probe (LMJ-SSP) combined with PB-MS/MS⁴⁹ have suggested that tumorous tissue may be differentiated by aberrant ratios of double bond isomers. Although highly informative for ascertaining the relationship between lipid profile and disease state, imaging methods are often limited in lipidomic depth when compared to separation-based workflows that provide a greater dynamic range for deeper profiling.

Even though many of the aforementioned strategies have been integrated into LC-based applications,^{21,30,37,39,40,44,45} few have been thoroughly validated as quantitative strategies sufficiently robust to compare isomer compositions across diseased tissues in a high-throughput manner. Among the strategies that have undergone rigorous evaluation, PB-MS/MS coupled to hydrophilic interaction chromatography (HILIC) revealed that significant changes in the ratios of glycerophospholipid double bond isomers occur in human breast cancer tissue and type 2 diabetes plasma samples.³⁷ Examination of unsaturated glycerophospholipids from cancerous and healthy human sera revealed similar outcomes via m-CPBA epoxidation coupled to reversed phase LC (RPLC).²¹ As this emerging field of global structural lipidomic analyses continues to mature, additional strategies that address the challenges of quantitative lipidomics are needed.

Another activation method that has allowed detailed characterization of lipids is ultraviolet photodissociation (UVPD).⁵⁰ For double bond localization, 193 nm UVPD features high versatility and applicability, characterizing fatty acids,^51,52 sphingolipids,⁵³ and various glycerophospholipid classes^54–59 via the production of diagnostic products generated by the predictable cleavage of bonds adjacent to the double bonds present in fatty acyl chains. Pairs of diagnostic product ions exhibit a distinctive mass difference of 24 Da that enable the facile identification of lipid double bond isomers. Furthermore, sn isomer resolution may also be achieved for glycerophospholipids through 193 nm UVPD by coupling the technique to collisional dissociation or metal adduction.^56,59 Moreover, the versatility of 193 nm UVPD allows it to be incorporated into different workflows. Both imaging and LC workflows that implement UVPD for lipid analyses have been developed to spatially correlate double bond isomer ratios with tissue histology^52,55 and to examine the double bond structure of bacterial extracts,^58,60 respectively. Most notably, DESI-UVPD studies have revealed that double bond isomer ratios vary significantly between healthy and cancerous tissues.^52,55

These results have motivated the development of a global quantitative identification workflow that would appropriately trade-off the histological specificity of imaging MS for the high throughput afforded by LC-MS to differentiate glycerophospholipids from healthy or cancerous origins. Leveraging the complementary separation of double bond isomers via RPLC and double bond isomer identification via UVPD in conjunction with quantitative parallel reaction monitoring (PRM), we have herein developed a RPLC-PRM-UVPD method, evaluated its quantitative merits, and demonstrate its applicability for the quantitative evaluation of phosphatidylcholine (PC) double bond isomers. Characterization of the double bond isomer composition of healthy and cancerous breast tissues via this technique uncovered significant differences between the tissue types, empowering the use of UVPD for global identification of unsaturated lipids from biological tissue.

EXPERIMENTAL SECTION

Materials.

PC standards, bovine liver extract, and egg PC extract were purchased from Avanti Polar Lipids (Alabaster, AL) and used without further purification. The structures and molecular weights of all PC standards are displayed in Table S1. HPLC grade water (H₂O), methanol (MeOH), and acetonitrile (ACN) were purchased from EMD Millipore (Billerica, MA). HPLC grade isopropanol (IPA), LC/MS grade formic acid, and ammonium formate were purchased from Thermo Fisher Scientific (Waltham, MA). Total lipid extracts from human tissue were prepared by performing a Bligh and Dyer lipid extraction⁶¹ on tissue samples procured through the Cooperative Human Tissue Network. All samples were diluted to working concentrations using MeOH prior to LC-MS analysis.

Liquid Chromatography and Mass Spectrometry.

The LC-MS system was composed of a Dionex Ultimate 3000 microflow liquid chromatography system (Sunnyvale, CA) coupled to a Thermo Fisher Orbitrap Fusion Lumos mass spectrometer (San Jose, CA) via a heated-ESI source. The mass spectrometer was modified with a Coherent Excistar XS excimer laser (Santa Clara, CA) to enable 193 nm UVPD in the high-pressure linear ion trap, as previously described.⁶² ESI source settings were set with an electrospray voltage of 5.5 kV, sheath gas at 25 (arbitrary units), auxiliary gas at 0 (arbitrary units), capillary temperature at 300 °C, and vaporizer temperature at 90 °C.

Separations were performed at a constant flow rate of 260 μL/min using a Waters CSH C18 column (2.1 × 100 mm, 1.7 μm particle size) (Milford, MA) heated to 50 °C using a modified version of a previously described gradient⁵⁸ using 60:40 ACN/H₂O (10 mM ammonium formate, 0.1% formic acid) as mobile phase A (MPA) and 90:10 IPA/ACN (10 mM ammonium formate, 0.1% formic acid) as mobile phase B (MPB). The 30 min gradient initiates at 10% MPB and ramps to 45% MPB over 1 min, starting at 2 min after injection, followed by a second step to 60% MPB over 22 min, before initiating a 5 min wash step at 95% MPB. The system is subsequently equilibrated to starting conditions for 10 min.

For data-dependent acquisition experiments, survey MS1 spectra were collected using the Orbitrap mass analyzer with a resolution setting of 30k at m/z 200 in the negative ion mode. MS1 spectra were collected over the m/z 400–1600 range, with the AGC target set to 1E6, and a max injection time (MIT) of 100 ms. The instrument was operated in top-speed mode with a cycle time of 3 s. The monoisotopic peak selection filter (MIPS, set to small molecule) and intensity threshold filter (minimum set to 5E5) were used as precursor selection filters. MS2 precursors were isolated by the quadrupole with a 1 m/z isolation width, with AGC set to 5E4 and a MIT of 100 ms, prior to HCD at 25 NCE. MS2 spectra were collected with a resolution setting of 7.5k at m/z 200.

For targeted acquisition experiments, MS2 spectra for m/z values included in targeted mass lists were collected during the duration of the LC-MS experiment for HCD analyses or collected at specified elution windows for UVPD analyses. All targeted MS2 spectra were collected at a resolution setting of 15k at m/z 200. Targeted UVPD MS2 spectra were collected by activating quadrupole-isolated precursor ions in the high-pressure linear ion trap using 15 laser pulses set to 3 mJ each (AGC = 5E5, MIT = 400 ms). Targeted HCD MS2 spectra were collected by activating quadrupole isolated precursor ions via HCD set to 25 NCE (AGC = 5E4, MIT = 100 ms) in the negative ion mode. Targeted UVPD MS2 spectra were collected in full-profile mode using the positive ion mode.

Data Analysis.

LipidCreator⁶³ and Skyline^64,65 were used to generate isolation and HCD transition lists for targeted experiments of formate-adducted [M+COO]⁻ PC precursors and to process RAW files generated by targeted HCD LC-MS experiments. For the analysis of targeted UVPD data, transition lists used in Skyline were modified to monitor the protonated [M + H] PC precursors and to include neutral loss fragments associated with the constituent acyl chains (Table S2). All data imported into Skyline was manually inspected to confirm assignments. Area under the curve (AUC) values were exported from Skyline for statistical analysis using R base language. Data dependent experiments were manually interpreted with the aid of ChemDraw (PerkinElmer).

Nomenclature.

Lipid shorthand notation, as described by Liebisch et al.,¹⁵ was implemented to describe PC structure. In brief, lipids may be described at the sum composition level by the total number of carbons in the acyl chains followed by “:” and the number of degrees of unsaturation (e.g., PC 36:2). If the identity of the constituent acyl chains is known, each acyl chain is described by its respective number of carbons and degrees of unsaturation, separated by “/” or “_” if the sn stereochemistry is known or unknown, respectively. The double bond positions are indicated in parentheses following the degrees of unsaturation and are noted with either Z, E, or Δ depending on whether the double bond geometry is cis, trans, or unknown, respectively. For example, PC 18:0/18:2(9Z,12Z) represents a PC with a fatty acid (FA) with 18 carbons and 0 degrees of unsaturation at position sn-1 and a FA with 18 carbons and 2 degrees of unsaturation at position sn-2, with a cis double bond at carbon 9 and carbon 12.

RESULTS AND DISCUSSION

Relative Quantitation of PC Double Bond Isomers by PRM-UVPD.

UVPD of unsaturated glycerophospholipids induces cleavage of C–C bonds adjacent to the site of unsaturation, resulting in diagnostic fragment ions that localize double bonds.⁵⁰ These fragments feature a diagnostic mass difference of 24 Da that are explained by the dissociation mechanism proposed by Ryan et al. for unsaturated sphingolipids involving cis 1,2 elimination reactions at the C–C bonds adjacent to the double bond as demonstrated in Scheme S1.⁵³ Furthermore, the masses of these diagnostic fragments are easily predicted as the losses of neutral masses are dependent on the double bond position along the fatty acyl chain. Figure 1A,B demonstrates the diagnostic neutral mass losses, −114.14/−138.14 Da and −86.11/−110.11 Da, associated with double bond positions for protonated isomeric PC standards, PC 18:1(9Z)/18:1(9Z) and PC 18:1(11Z)/18:1(11Z), respectively. UVPD mass spectra confirming the presence of the predicted fragments for each respective isomer are shown in Figure S1 and affirm the possibility of de novo double bond localization by matching specific pairs of neutral losses with a corresponding double bond position. Although double bond localization via UVPD has been demonstrated for multiple lipid classes,^{51,53,54,57,58} the quantitative merits of UVPD for the double bond isomers in global lipidomic applications have not yet been evaluated.

Figure 1.

PRM-MS of PC double bond isomers utilizing UVPD generated transitions. (A) Structure of PC 18:1(9Z)/18:1(9Z) mapped with the expected double bond specific transitions. (B) Structure of PC 18:1(11Z)/18:1(11Z) mapped with the expected double bond specific transitions. (C) XIC of m/z 786.60 for a 50/50 mixture of PC 18:1(9Z)/18:1(9Z) and PC 18:1(11Z)/18:1(11Z). (D) Chromatogram displaying the intensities of the PC 18:1(9Z)/18:1(9Z) double bond diagnostic transitions (blue and purple traces) shown in (A) and the intensities of the PC 18:1(11Z)/18:1(11Z) double bond diagnostic transitions (orange and red traces) shown in (B) for a 50/50 mixture of the two isomers in a targeted UVPD (15 pulses; 3 mJ per pulse) method.

An LC-UVPD-MS method was developed to enable both quantitative analysis and differentiation of double bond isomers. Reversed phase LC is capable of separating double bond isomers^39,44,45 but cannot alone identify the location of the double bonds, a shortcoming that may be redressed through integration with UVPD. To explore this opportunity, a 1:1 mixture of PC isomers, PC 18:1(9Z)/18:1(9Z) and PC 18:1(11Z)/18:1(11Z), was subjected to RPLC (50 pmol injection of each isomer), resulting in chromatographic separation of the two isomers but not in baseline resolution. Figure 1C displays the extracted ion chromatogram (XIC) for the precursor, m/z 786.60, corresponding to [M + H]⁺ for both isomers, displaying two overlapping distributions with retention times of 19.7 and 20.1 min (based on peak centroids). Although sufficient chromatographic resolution (~1.0) is achieved to confirm the presence of both isomers, two issues hamper effective analysis by RPLC alone. First, RPLC cannot confirm the order of elution of the two isomers, and second, the interfering signals for the two isomers in the chromatogram inhibit integration of the area under the curve for each species, a step necessary for quantitative analysis. Circumventing these issues, a parallel reaction monitoring (PRM) workflow was developed wherein specific precursor-to-fragment ion transitions, instead of the precursor m/z, are integrated to quantitate the isomers. This approach entails a targeted UVPD method to monitor the PC isomers throughout the elution period, producing double bond diagnostic transitions that effectively distinguish double bond isomers in both retention time and m/z space. When applied to the same mixture of PC isomers and targeting the precursor of m/z 786.60 continuously with UVPD (15 pulses at 3 mJ each), this strategy yields isomer-specific responses (Figure 1D), allowing assignment of the peak at 19.7 min as PC 18:1(11Z)/18:1(11Z) and at 20.1 min as PC 18:1(9Z)/18:1(9Z), while also enabling integration of isomer-specific curves, and consequently, relative quantitation. Although mixed acyl chain double bond isomers would be more representative of biologically relevant PC lipids, the PC 18:1/18:1 double bond isomers herein utilized, nonetheless, aptly illustrate the PRM strategy employed to evaluate mixtures of PC double bond isomers. Chromatographic separation and PRM-UVPD of mixed acyl chain double bond isomers of PC 16:0/18:1 are demonstrated in Figure S2. Although a mixture of the three isomers (PC 16:0/18:1(9Z), PC 16:0/18:1(11Z), PC 16:0/18:1(6Z)) produces a broad peak that would impede analysis by liquid chromatography alone (Figure S2D), UVPD-PRM reveals three distinct distributions (Figure S2E) produced by the unique transitions corresponding to each double bond isomer (Figure S2A–C).

In previous UVPD analyses of unsaturated FAs, the rich UVPD mass spectra resulted in “satellite” ions that confounded quantitation.^51,52 These fragments are produced from cleavages of C–C bonds within the acyl chains but remote from the C=C bond, resulting in product masses that are equal to those of diagnostic fragments from isomeric counterparts. Exemplifying this phenomenon, UVPD of PC 18:1(9Z)/18:1(9Z) results in a similar satellite ion of m/z 700.49 (Figure S3A and B) that is isomeric to the double bond diagnostic fragment for PC 18:1(11Z)/18:1(11Z) and would normally stymie analysis. However, chromatographic resolution is sufficient that the “satellite” ion from PC 18:1(9Z)/18:1(9Z) can be distinguished from the actual diagnostic fragment for PC 18:1(11Z)/18:1(11Z), as demonstrated in Figure S3C.

Accordingly, this PRM-UVPD strategy for relative quantitation of double bond isomers was assessed using mixtures of PC 18:1(9Z)/18:1(9Z) and PC 18:1(11Z)/18:1(11Z) prepared at various concentration ratios. Each mixture contained a fixed total concentration of 20 μM PC but a variable ratio of PC 18:1(9Z)/18:1(9Z) to PC 18:1(11Z)/18:1(11Z) from 19:1 to 1:19. PRM-UVPD was implemented to track the intensity of double bond diagnostic transitions during the elution period (Figure 2A); the resulting responses were integrated and the ratio for the area under the curve (AUC) for PC 18:1(9Z)/18:1(9Z) to the AUC of PC 18:1(11Z)/18:1(11Z) were plotted against the ratio of PC 18:1(9Z)/18:1(9Z) concentration to PC 18:1(11Z)/18:1(11Z) concentration (Figure 2B). The resulting calibration curve features excellent linearity (R² = 0.9995), demonstrating a highly linear correlation between the AUC ratio and isomer concentration ratio. Limits of identification for the two isomers were calculated based on the respective AUCs to be 0.60 and 1.12 pmol for PC 18:1(9Z)/18:1(9Z) and PC 18:1(11Z)/18:1(11Z), respectively. Moreover, the slope of the regression curve (1.1379) approximates unity, indicating low preferential ionization or dissociation of either isomer, a feature that enables relative quantitation of the analytes without necessitating a calibration curve for each pair of isomers compared. These features also indicate that any interferences from satellite ions at isomer concentration ratios where AUCs may interfere with one another, such as the 1:1 mixture in Figure 2A and Figure 1D, are quantitatively negligible and do not affect the quantitative merits of this strategy.

Figure 2.

PRM-UVPD of isomeric PC standards, PC 18:1(9Z)/18:1(9Z) and PC 18:1(11Z)/18:1(11Z), combined at various ratios. (A) Relative summed intensities of PC 18:1(9Z)/18:1(9Z) double bond transitions (blue trace, m/z 786.60 → m/z 672.46 and m/z 786.60 → m/z 648.46) and relative summed intensities of PC 18:1(11Z)/18:1(11Z) double bond transitions (red trace, m/z 786.60 → m/z 700.49 and m/z 786.60 → m/z 676.46) at multiple 9Z:11Z ratios. (B) Plot of the concentration ratio of PC 18:1(9Z)/18:1(9Z): PC 18:1(11Z)/18:1(11Z) versus the AUC ratio for double bond transitions of PC 18:1(9Z)/18:1(9Z): PC 18:1(11Z)/18:1(11Z). Inset shows an expanded view of the concentration ratios (0–1.0). Error bars represent one standard deviation, n = 5.

The limit of linearity for this method was approximated by generating mixtures with a fixed PC 18:1(9Z)/18:1(9Z) concentration of 1 μM, with increasing amounts of PC 18:1(11Z)/18:1(11Z) from 1 μM to 100 μM, analyzing each mixture with PRM-UVPD. Plotting the AUC ratio for PC 18:1(11Z)/18:1(11Z) to PC 18:1(9Z)/18:1(9Z) versus concentration ratio resulted in good linearity in the concentration range explored (R² = 0.9988), demonstrating the effectiveness of this approach for relative quantitation over 2 orders of magnitude (Figure S4).

Quantitation of PC isomers in biological samples.

To adequately characterize double bond isomers in complex biological samples, a workflow utilizing two MS/MS methods, HCD and UVPD, was implemented as illustrated in Figure 3 and prototyped using egg PC extract, a relatively simple and commercially available biological extract. In this strategy, an initial PC target list is created and applied to analyze the extract in a PRM-HCD workflow using the negative ionization mode to determine PC lipid composition at the FA level and select elution time windows for subsequent targeted methods. For egg PC extract, FA methyl ester analysis by gas chromatography is supplied by the manufacturer, providing sufficient information to compile an initial target list. Using LipidCreator,⁶³ a precursor target list of all possible formate-adducted PC lipids, [M + HCOO]⁻, that may contain any combination of FAs reported by the manufacturer (FA 16:0, FA 16:1, FA 18:0, FA 18:1, FA 18:2, FA 20:3, FA 20:4) was compiled for a targeted analysis in the negative mode via RPLC-PRM-HCD and applied to analyze egg PC lipids. The resulting HCD mass spectra were imported into Skyline,^64,65 monitoring the fatty acid and the (CH₃+HCOO) loss transitions to confirm FA compositions and headgroup identity, respectively, for each of the PC lipids included in the target list. The resulting elution profile for each possible PC lipid was manually validated to exclude any PC lipids that could not be confidently assigned and to define elution time windows for any PC lipids that were confidently detected. An example is shown in Figure S5A, B, and C where PC 18:0_18:1, m/z 832, was included in the target list, confidently identified by monitoring HCD transitions m/z 832.61 → m/z 281.25 to confirm the presence of FA 18:1, m/z 832.61 → m/z 283.26 to confirm the presence of FA 18:0, and m/z 832.61 → m/z 772.61 to confirm the PC headgroup, using the resulting elution profile to define retention time windows. A list of all 17 unsaturated egg PC species characterized at the fatty acyl chain level by this approach is included in Table S3.

Figure 3.

Schematic representation of the PRM based strategy implemented to analyze double bond isomer abundances in biological samples. After sample extraction, an initial targeted HCD method is developed and applied to the sample of interest. The collected PRM-HCD data is analyzed to determine the presence or absence of the targeted lipids. If present, the elution time is determined for subsequent targeted analyses. After refinement of the HCD results, a new target list is compiled and integrated into a targeted UVPD method designed to target all identified lipids with potential double bond isomerism. Results are surveyed to identify isomers present in the sample and used to refine the target list to only encompass lipids that feature multiple double bond isomers. The refined list is integrated into the UVPD method for subsequent replicate analyses and sample comparisons. The transitions and m/z values shown in this workflow are illustrated for PC 18:0_18:1(9Δ) and PC 18:0_18:1(11Δ).

Following RPLC-PRM-HCD analysis, the target list is refined by excluding any PC lipids not detected or not expected to feature double bond isomerism, such as saturated lipids. Retention time windows are added to the list, and each precursor m/z is modified for positive mode analysis of protonated [M + H]⁺ precursor targets for PRM-UVPD analysis, instead of the [M + HCOO]⁻ anions used for PRM-HCD. UVPD mass spectra for each targeted PC are manually interpreted to identify double bond isomers, and the corresponding double bond diagnostic neutral loss transitions (Table S2) are amended into the transition list in Skyline to track the abundances of the respective double bond isomers as demonstrated in Figure S5D–F for PC 18:0_18:1, shown to encompass PC 18:0_18:1(9Δ) and PC 18:0_18:1(11Δ) at the double bond level. Probing beyond the fatty acyl chain level, a total of 25 distinct PC lipids were identified at the double bond level (Table S4) for egg PC. After initial RPLC-PRM-UVPD analysis, the m/z list is refined to target PC lipids featuring double bond isomerism, producing a final target list apt for replicate analysis and relative quantitation of PC double bond isomers in biological samples. PC double bond isomers in egg PC extract, ones exhibiting double bond isomerism for both FA 18:1 (9Δ and 11Δ) and FA 16:1 (7Δ and 9Δ), were quantified by determining the ratio of the major double bond isomer to the minor double bond isomer, as shown in Figure S6. Technical replicate analyses featured good reproducibility by this approach (CV < 15%, n = 5) for all ratios reported for egg PC, except for PC 18:0_18:1 (CV = 21%) for which the 9Δ:11Δ ratio of 148 was beyond the limit of linearity established in Figure S4.

Relative quantitation of double bond isomers in bovine liver total lipid extract, a biological sample featuring higher complexity, was performed using the same workflow. For this sample, the initial target list was compiled through manual analysis of a data dependent survey HCD run (negative ionization mode) to identify PC lipids present in the total extract, while all subsequent steps were performed as described for egg PC. Exemplifying the high level of double bond isomerism present for these lipids, Figure 4 shows the resulting analysis for PC 16:0_18:1 from bovine liver for which three distinct double bond isomers were identified at various ratios. Ratios between the most abundant PC double bond isomer and minor isomer are displayed to emulate quantitative lipidomic strategies that incorporate ratios of the analyte to a spiked internal standard correct for sample-to-sample variations. Highlighting the effectiveness of PRM-UVPD to probe deep into the lipidomic profile, Figure 4A displays the XIC of PC 16:0_18:1 (precursor m/z 760.59) for which double bond isomerism is not evident using traditional HCD for MS/MS (Figure S7). Analysis of UVPD mass spectra collected in a PRM workflow across different regions of the elution profile elucidated double bonds at position 11Δ, 9Δ, and 10Δ (Figure 4B), revealing the underlying heterogeneity in double bond locations along FA 18:1. Tracking the intensity of the diagnostic neutral losses for each position (Table S2) allows generation of isomer specific traces (Figure 4C) that can be used to quantify each of the minor isomers relative to the major isomer, PC 16:0_18:1(9Δ), with high reproducibility (Figure 4D).

Figure 4.

PRM-UVPD of PC 16:0_18:1 in bovine liver extract. (A) XIC of precursor m/z 760.59, assigned as PC 16:0_18:1 via PRM-HCD (Figure S7). (B) Expanded view of m/z 615–695 range for spectra corresponding to different regions of the elution profile shown in (A). Diagnostic fragments for double bonds at positions 11Δ, 9Δ, and 10Δ along the 18:1 fatty acid are identified and labeled. Diagnostic neutral losses associated with each isomer are listed in Table S2. (C) Relative summed intensities of the double bond transitions of the three PC 16:0_18:1 isomers identified in (B). (D) Double bond isomer ratios for the major PC 16:0_18:1 double bond isomer relative to each of the minor isomers, as determined by the AUC ratios produced by PRM-UVPD. Error bars represent one standard deviation, n = 5.

Because uncommon fatty acid isomers were identified at low abundances, such as PC 16:0_18:1(10Δ) in Figure 4, a validation method was developed to ensure that these identifications are not false identifications resulting from the complex UVPD mass spectra. To develop this validation method, a 1:1 mixture of PC 18:1(9Z)/18:1(9Z) and PC 18:1(11Z)/18:1(11Z) featuring the common 18:1 double bond isomers was evaluated using the UVPD-PRM approach. This mixture should not contain PC species with any minor 18:1 fatty acid isomers, such as PC 18:1(10Δ)/18:1(10Δ), PC 18:1(8Δ)/18:1(8Δ), and PC 18:1(12Δ)/18:1(12Δ), and thus serves as a “blank” from which artifactual noise that may lead to false identifications can be quantified and used to establish an LOD to serve as a cutoff for confident identifications. From the UVPD-PRM data obtained for this mixture, AUCs for diagnostic ions for PC 18:1(9Z)/18:1(9Z) and PC 18:1(11Z)/18:1(11Z) were quantified. Additionally, AUCs for noise peaks matching diagnostic fragments from PC 18:1(10Δ)/18:1(10Δ), PC 18:1(8Δ)/18:1(8Δ), and PC 18:1(12Δ)/18:1(12Δ) were also quantified as a percent of the real diagnostic fragments for PC 18:1(9Z)/18:1(9Z) and PC 18:1(11Z)/18:1(11Z). The LOD for each isomer was quantified conventionally as the average percent of the noise peaks plus 3-times the standard deviation of the noise peak signal (n = 5). These LODs are tabulated in Table S5. Similar LODs were established for minor 16:1 double bond isomers using PC 16:1(9Z)/16:1(9Z), as reported in Table S5 (n = 5). To confirm the presence of the minor PC 16:0_18:1(10Δ) in Figure 4, the percent signal for the diagnostic 10Δ fragments relative to the abundance of the major diagnostic fragments for the 9Δ and 11Δ isomers was quantified for the two distributions observed for PC 16:0_18:1(10Δ) in Figure 4C by only considering diagnostic fragment abundances within the integration windows shown in Figure S8. The distribution at 19.4 min coelutes with PC 16:0_18:1(9Δ) and had an AUC percent of 5.6 ± 0.1% which is below the LOD cutoff of 5.7% and indicates that this peak is an artifact from the major isomer. The distribution at 20.0 min is chromatographically separated from the major isomers and features an AUC percent that is significantly above the LOD cutoff at 28.9 ± 2.0%, allowing the confident assignment of this distribution as PC 16:0_18:1(10Δ). Accordingly, the isomer ratio calculated in Figure 4D only considers the latter distribution for relative quantitation. These results align with previous reports of 18:1(10Δ) fatty acids in bovine liver extracts.^37,54

Importantly, the bovine liver extract also displayed double bond heterogeneity in polyunsaturated fatty acids, for which relative quantitation has not been previously described. During analysis via RPLC-PRM-HCD, two distributions were detected for PC 18:0_20:3 at 21.4 and 22.6 min (Figure 5A). Averaging UVPD mass spectra across the two distributions revealed that double bond isomerism of FA 20:3 was responsible for the two distributions observed; PC 18:0_20:3(8Δ, 11Δ, 14Δ) eluted at 21.4 min, whereas PC 18:0_20:3(5Δ, 8Δ, 11Δ) eluted at 22.6 min (Figure 5B,C). While the complexity of the fragmentation patterns of polyunsaturated FAs coupled with the lack of polyunsaturated PC standards precludes the ability to validate a UVPD-based quantitative strategy, the polyunsaturated double bond isomers detected are baseline resolved via RPLC, an aspect that allows for quantitation through alternate means. Additionally, the significant chromatographic separation renders any confounding ions, such as the “satellite ions” discussed for monounsaturated isomers, a nonissue. Leveraging the RPLC-PRM-HCD data collected as part of the described strategy, relative quantitation of the polyunsaturated PC double bond isomers was performed by monitoring the intensities of the FA transitions produced by HCD. Validating this strategy, mixtures of isomeric PC standards PC 18:1(9Z)/18:1(9Z) and PC 18:1(6Z)/18:1(6Z), that are baseline resolved by RPLC (Figure S9A), were prepared at concentration ratios of PC 18:1(9Z)/18:1(9Z) to PC 18:1(6Z)/18:1(6Z) of 1:99 to 99:1 and subjected to RPLC-PRM-HCD. The AUC ratio for the HCD transition m/z 786.60 → m/z 281.25 for the PC 18:1(9Z)/18:1(9Z) peak (20.1 min) to the PC 18:1(6Z)/18:1(6Z) peak (21.2 min), corresponding to the intensity of the FA 18:1 fragment from the formate-adducted precursor, was plotted against the concentration ratio of PC 18:1(9Z)/18:1(9Z) to PC 18:1(6Z)/18:1(6Z), displaying a highly linear correlation (R² = 0.9991) in Figure S9B. Owing to the linear correlation between isomer AUC ratio to isomer concentration ratio for PC 18:1(6Z)/18:1(6Z) and PC 18:1(9Z)/18:1(9Z) isomers through the PRM-HCD approach in Figure S9, no corrections were applied to quantify polyunsaturated isomers in the bovine liver extract. Although a deviation from unity is observed for the calibration curve in Figure S9B, this is attributed to ion source saturation occurring at the high sample concentrations used to test the limits of linearity for the PRM-HCD approach. AUC ratio of PC 18:1(9Z)/18:1(9Z) to PC 18:1(6Z)/18:1(6Z) for the 1:1 isomer mixture approximates unity (1.02 ± 0.01). Collecting data points over a narrower range (76:1 to 1:76) and using a lower injection amount (1 picomol total PC) results in a slope closer to unity (Figure S9C). Applying the same PRM-HCD approach to polyunsaturated FAs, the AUC for the FA transitions of PC 18:0_20:3 (m/z 856.60 → m/z 283.26, and m/z 856.60 → m/z 305.25) were used to quantify PC 18:0_20:3(8Δ, 11Δ, 14Δ) and PC 18:0_20:3(5Δ, 8Δ, 11Δ). The process was repeated for all polyunsaturated isomers identified in the bovine liver extract, using UVPD to inform double bond isomerism, and RPLC-PRM-HCD for relative quantitation; results are reported in Figure S10. As with RPLC-PRM-UVPD of monounsaturated PC double bond isomers, the UVPD-guided RPLC-PRM-HCD approach proved reproducible (CV < 10%, n = 3) and well-suited for relative quantitation of isomers in biological extracts. Although fragment ions of higher abundances adjacent to the diagnostic pairs were not utilized in this study for characterization purposes, the presence of these fragments presents an interesting opportunity to increase sensitivity of UVPD analyses in future studies by coupling spectral matching methodologies to the techniques presented here. Speculated to occur via photoexcitation of C=C bonds to excited diradical states followed by concomitant allylic cleavages, vinylic cleavages, hydrogen migration, and formation of radical species,⁵⁶ these ions are not well understood and were excluded in this study in favor of the more common and well-evaluated Δ 24 diagnostic pairs used for de novo characterization

Figure 5.

(A) PRM-HCD of PC 18:0_20:3 in bovine liver extract. The green trace represents the transition for the loss of the formate adduct and a methyl group (−60 Da), diagnostic of the PC headgroup, whereas the red and blue transitions confirm the presence of the 18:0 and 20:3 acyl chains, respectively. Two peaks are detected at 21.4 and 22.6 min. (B) UVPD mass spectrum corresponding to the elution profile at 21.4 min for PC 18:0_20:3, characterizing the double bond structure of the lipid as PC 18:0_20:3(8Δ,11Δ,14Δ). (C) UVPD spectrum corresponding to the elution profile at 22.6 min for PC 18:0_20:3, characterizing the double bond structure of the lipid as PC 18:0_20:3(5Δ,8Δ,11Δ). Selected precursor ions are denoted with a star. Fragment maps are labeled with the predicted double bond diagnostic fragments in parts A and B.

PC Double Bond Isomers in Normal and Tumorous Breast Tissue.

Previous analyses comparing phospholipid and FA compositions across normal and tumorous tissue have revealed significant disease-driven alterations at the double bond level.^{21,37,38,46–49,51,52,55} Aberrations in the abundance of 18:1(9Δ) and 18:1(11Δ) free FAs and phospholipids have been shown to distinguish normal and cancerous breast tissue³⁷ and, more recently, leveraged to classify human breast cancer cell subtypes by considering phospholipid sn-stereochemistry as well.³⁸ Notably, recent comparison of tumorous human breast tissue via DESI-UVPD imaging elucidated the effects of hormone receptor status on free FA double bond positions.⁵² By the examination of a cohort of tissues, it was determined that progesterone receptor positive tissue featured a statistically significant increase in the abundance of FA 18:1(11Δ) relative to FA 18:1(9Δ), likely resulting from alterations in fatty acid synthase and desaturase enzyme function.^52,66,67 To expand beyond this initial work and emphasize the analysis of FAs esterified to PC lipids, RPLC-PRM-UVPD was applied to explore the effects of disease state on isomer composition in normal and tumorous human breast tissue.

Double bond isomer abundances for three normal human breast tissue extracts and three human breast tumor extracts were quantified in technical triplicate. Tumor samples with the same hormone receptor statuses (progesterone receptor negative, estrogen receptor negative, human epidermal growth factor receptor 2 negative) were selected to draw comparisons using tissue of the same subtype. Results for the PC isomers are summarized in Figure 6. Each of the monounsaturated PC lipids noted to be significantly distinct between tumor and normal samples in Figure 6A and that feature uncommon fatty acids (18:1(10Δ), 18:1(12Δ), 18:1(8Δ), 16:1(6Δ)) were subjected to the percent AUC validation approach described above to exclude any assignments that may occur as false identifications from UVPD artifacts. Only lipids that were above the LOD cutoff in both sample sets were included in Figure 6B–E. A table with the percent AUCs in the normal and tumor samples is included as Table S6. Incredible heterogeneity was observed at the double bond level. Multiple monounsaturated FAs including FA 16:1, FA 17:1, and FA 18:1 displayed various double bond isomers, revealing a total of five different double bond positions for FA 18:1 and FA 16:1 across the PCs identified. Highlighting the high degree of double bond isomerism, Figures S11 and S12 display the profiles of the diagnostic transitions for all four double bond isomers identified for PC 16:0_18:1 and all five double bond isomers identified for PC 16:0_16:1, respectively. The quantitation strategy employed also identified polyunsaturated isomers for PCs with FA 18:3, FA 20:3, and FA 22:5, allowing unprecedented characterization of the modulation of polyunsaturated phospholipid structure related to tumorigenesis. Across the PC isomers quantified, 18 PC ratios were determined to be statistically different for normal and tumor samples with a significance value less than 0.05, according to Welch’s t test at the 95% confidence level. Moreover, results reported here manifest trends that generally align with those achieved via HILIC-PB analysis of PC lipids incorporating FA 18:1(9Δ) and FA 18:1(11Δ).³⁷ Three of the four PC lipids (i.e., PC 16:0_18:1, PC 18:1_18:1, PC 18:0_18:1) found to display significant differences between normal and breast cancer tissue through both HILIC-PB analysis³⁷ and the present RPLC-PRM-UVPD method also reported the same trends in 18:1(9Δ):18:1(11Δ) ratio (Figure 6C). The only discrepancy was observed for PC 18:1_18:2 which gave a higher 18:1(9Δ):18:1(11Δ) ratio via RPLC-PRM-UVPD, and a lower 18:1(9Δ):18:1(11Δ) ratio via PB-HILIC for the tumor samples relative to normal tissue.³⁷ Regardless, the overall agreement between the two strategies substantiates global structural lipidomics as a powerful tool capable of revealing trends in disease-induced metabolic shifts that would otherwise remain invisible.

Figure 6.

(A) Comparison of PC double bond isomers in human normal breast tissue and human breast tumor tissue. Three tumor and three normal samples were each analyzed in technical triplicates (total n = 9 tumor analyses and n = 9 normal analyses). Significance values were determined at the 95% confidence interval by Welch’s t test. AUC ratios are shown on a log-scaled y-axis to display the range of isomer ratios identified. (Parts B–D) Select PC isomer ratios from (A) with significance value <0.05 and above the LOD cutoff, grouped by fatty acid isomerism to highlight differences in structure between normal and tumor samples. (B) Isomer ratios for PC species containing FA 16:1. (C) Isomer ratios for PC species containing FA 18:1(9Δ) and 18:1(11Δ). (D) Isomer ratios for PC species containing FA 18:1, where the minor isomer presents the double bond at position 10Δ, or 8Δ. (E) Isomer ratios for PC species containing polyunsaturated FAs.

Surpassing the capabilities of previous studies,³⁷ significant differences correlating with disease state were observed for PCs constituting additional 18:1 isomers (10Δ, 12Δ, 8Δ) as well as 16:1, 18:3, and 20:3 acyl chain double bond isomers (Figure 6B,D,E). Specifically, 18:1(10Δ) trended lower in the tumor tissue samples relative to the major 18:1(9Δ) isomer, whereas 18:1(8Δ) trended higher. Additionally, 16:1(9Δ) trended higher in the normal tissue samples compared to the minor 16:1(6Δ) and 16:1(7Δ) FAs, and polyunsaturated fatty acids 20:3 and 18:3 presented greater abundances of double bonds closer to the methyl end of the acyl chain for the tumor samples. Notably, these results highlight the importance in quantifying polyunsaturated isomers that are typically omitted from analogous quantitative phospholipid double bond-focused studies due to complex fragmentation, arduous structural assignment, and challenging quantitation based on double bond diagnostic ions alone. These differences align with previous reports indicating an increase in fatty acid desaturase 2 (FADS2) activity resulting in an increase of 16:1(6Δ) and 18:1(8Δ) fatty acids in cancer cells.^68,69 Analysis of a larger cohort of samples and additional subtypes by the same strategy should enlighten a more robust understanding of lipid metabolism and may reveal potential diagnostic and prognostic biomarkers. Importantly, all data analysis was performed with widely available tools including Skyline, a freely available software package broadly used across the quantitative mass spectrometry community, to facilitate widespread adoption of the methodology for this goal.

Challenges and Limitations.

Even though RPLC-PRM-UVPD significantly enhances the depth of lipidomic analyses, there were certain instances where assignments could not be made unambiguously. One such example is illustrated for PC 32:1 (Figure S13), which is mostly composed of PC 16:0_16:1 but also presents a minor abundance of PC 14:0_18:1 with insufficient chromatographic resolution to distinguish whether the observed diagnostic neutral losses are due to PC 14:0_18:1(9Δ) and PC 14:0_18:1(11Δ) or PC 16:0_16:1(7Δ) and PC 16:0_16:1(9Δ). In such cases, only ratios for the major species are reported (PC 16:0_16:1(7Δ) and PC 16:0_16:1(9Δ)) to represent the distribution at a particular retention time, since the resulting ratio for the analogous minor FA isomers (PC 14:0_18:1(9Δ) and PC 14:0_18:1(11Δ)) would be exactly the same. This may be a possible source of uncommon fatty acid assignments, such as 16:1(8Δ), 16:1(10Δ), or 18:1(12Δ) noted in Figures S11 and S12. Although 16:1(8Δ) may be rationalized through FADS2 desaturation of 14:0 followed by elongation,⁶⁹ probable explanations for 16:1(10Δ), or 18:1(12Δ) are not present in the literature, even though these assignments were confidently made here through identification and validation of the diagnostic signals. Further validation of these assignments may require systematic evaluation or confirmation through orthogonal methods. This issue, however, does not detract from the ability to detect changes in abundances for a particular distribution across different samples. Such ambiguities could be resolved through incorporating ion/ion reactions in an MSⁿ workflow to probe individual fatty acids released from the intact lipids, as described in previous studies.^23,24

Other confounding instances involve lipids that contain two monounsaturated FAs, such as PC 18:1_18:1, where the identification of diagnostic neutral losses for double bonds at positions 9Δ and 11Δ would suggest the presence of PC 18:1(9Δ)_18:1(9Δ), PC 18:1(11Δ)_18:1(11Δ), and PC 18:1(9Δ)_18:1(11Δ), while the latter would be impossible to distinguish from the two former isomers with the chromatographic resolution achieved here (example shown in Figure S14). A practical assumption for resolving any resulting ambiguity is that only PC 18:1(9Δ)_18:1(9Δ), PC 18:1(11Δ)_18:1(11Δ) are present, because ultimately, it is the overall ratio of 18:1(9Δ) to 18:1(11Δ) that will be used to compare and classify biological samples. Additionally, the sum intensity for the diagnostic neutral losses of PC 18:1(9Δ)_18:1(11Δ) would simply be equal to the sum intensity of both the PC 18:1(9Δ)_18:1(9Δ) and PC 18:1(11Δ)_18:1(11Δ), which is largely uninformative. Another potential source of ambiguity are acyl chain isomers. However, these isomeric species are easily resolved due to the two-tiered characterization utilizing both HCD and UVPD coupled with significant chromatographic resolution. Examples are shown in Figures S15 and S16, for monounsaturated (PC 36:2) and polyunsaturated lipids (PC 38:5), respectively. The final scenario in which ambiguities may arise is in the case of PC lipids that feature a mixture of monounsaturated and polyunsaturated FAs, where the rich fragmentation of the polyunsaturated acyl chain would interfere with double bond localization for the monounsaturated FA. For example, UVPD of PC 18:1_18:2 results in the identification of neutral losses that correspond to double bonds at positions 9Δ and 11Δ along FA 18:1 (Figure S17C,D), however, UVPD also results in fragmentation of FA 18:2 (Figure S17A,B) and produces an abundant neutral loss of −110.11 Da that interferes with the diagnostic neutral loss of 110.11 Da for the double bond at position 11Δ. Previous studies implementing UVPD for relative quantitation have noted that utilizing only one instead of both diagnostic signals for a particular double bond position offers a practical solution to circumvent the negative effects of interfering signals in quantitation.^51,52 Applying the same approach for the analysis of PC 18:1_18:2, the effects of the interfering signal from the 18:2 acyl chain could be mitigated by quantifying 18:1(9Δ) and 18:1(11Δ) acyl chain isomers using the ratio of the 9Δ diagnostic fragment m/z 670.44 and 11Δ diagnostic fragment m/z 698.48, while excluding the 11Δ diagnostic fragment m/z 674.48, affected by interference from 18:2 dissociation, and the analogous 9Δ diagnostic fragment, m/z 646.45. Additionally, PC lipids featuring two distinct fatty acyl chains may also present sn regioisomerism, introducing an additional level of structural complexity. Although previous RPLC-MS analyses of PC regioisomers demonstrated some separation of these isomeric lipids,^56,70,71 the degree of separation achieved with the chromatographic method used in the present study for UVPD-PRM is not sufficient to affect the PRM strategy (Figure S18). UVPD of isomeric standards PC 16:0/18:1(9Z) and PC 18:1(9Z)/16:0 produce the same fragment ions at comparable intensities (Figure S18A,B), and the minor retention time shift observed for the two isomers does not confound their assignment (Figure S18C). UVPD-PRM analyses of 25 pmol injections of PC 16:0/18:1(9Z), PC 18:1(9Z)/16:0, and a 1:1 mixture of the two isomers resulted in no significant differences in the respective AUCs, indicating that the entire population of double bond isomers can be interrogated, regardless of sn isomerism. Simultaneous distinction and quantitation of sn and double bond isomers will be the subject of further studies. We envision that the UVPD-PRM approach could be implemented for quantitative sn characterization, albeit requiring metal adduction and an MS³ workflow as reported previously for identification of sn positions of phospholipids in a nonquantitative method.⁵⁶ In regards to cis/trans isomers, quantitative identification of cis/trans configurations is not anticipated to be readily achieved by the UVPD-PRM method because cis/trans isomerization is prevalent upon photoexcitation.⁷² Despite certain limitations, the capabilities of the workflow presented here for the relative quantitation of double bond PC isomers outweigh its limitations and present a significant approach enabled to probe deep into lipidomic profile of complex samples, reaching structural characterization at the double bond level for biological extracts.

Considering the charge-agnostic nature of photodissociation, UVPD generates double bond diagnostic product ions for phospholipids in both the positive and negative ionization modes;^54,58 however, PC requires adduction with a counterion for negative mode ionization. In the mobile phases implemented for RPLC, PC readily ionizes as [M+HCOO]⁻ which fragments to yield abundant [M-15]⁻ products upon UVPD, resulting in signal dilution for the double bond diagnostic fragments which originate both directly from the [M +HCOO]⁻ precursor and via secondary dissociation from the [M-15]⁻ product, thus decreasing sensitivity (Figure S19). Hence, UVPD was only applied in the positive ion mode for the analysis of PC isomers in the present study but would merit further exploration in the negative ion mode for glycerophospholipids that preferentially ionize as anions. Such phospholipid classes, including cardiolipin, phosphatidylglycerols, phosphatidic acids, phosphatidylethanolamines, phosphatidylserines, and phosphoinositides, have been characterized at the double bond level in UVPD workflows^57,58 and will be the subject of future studies. Overall, the PRM-UVPD approach presented offers a compelling strategy for relative quantitation of protonated PC double bond isomers that motivated exploration of the method for quantitation of PC isomers in complex biological extracts.

CONCLUSIONS

As a result of remarkable advances in structural lipidomics, significant evidence correlating FA double bond isomerism with disease state has been accumulated, introducing an additional facet to explore disease-induced metabolic shifts, identify additional therapeutic targets, or uncover previously hidden biomarkers. Aiming to extend this frontier, a high-throughput strategy leveraging the strengths of both HCD and UVPD to provide acyl chain composition and double bond level characterization, in synergy with separation of double bond isomers via RPLC, was developed and applied for relative quantitation of PC isomers in a PRM workflow. The unique merits of this approach include the unprecedented ability to quantify double bond isomerism at polyunsaturated FAs and the use of widely available and robust software packages (LipidCreator and Skyline) to evaluate data, an important consideration for facilitating the widespread adoption of this methodology. To underscore both the robustness of the PRM-based strategy and the general suitability of double-bond level structural lipidomics, PC isomers from human normal and tumor tissue were quantified to reveal disease-related shifts in double bond structure of the 18:1 acyl chain that were in agreement with previous reports, while also unveiling new trends for 16:1 and polyunsaturated FA isomers. Although PCs were the only glycerophospholipids examined herein, UVPD has been applied to characterize the most complex glycerophospholipids, such as cardiolipins, and even to elucidate glycerophospholipid sn stereochemistry,⁵⁶ a feature reportedly associated with disease state.^38,46 The strategy presented here lays the foundation for the development and application of high-throughput quantitative UVPD workflows to study additional glycerophospholipid classes or to probe other challenging structural features including regiochemistry, either coupled to LC or other MS-UVPD compatible separation techniques such as capillary zone electrophoresis⁷³ or ion mobility methods. The versatility of the current workflow should be amenable for the quantitation of lipids endogenous to human sera or to serotype bacterial strains presenting differing double bond structure.⁵⁸ Although the current incarnation of the UVPD-based strategy faces limitations in sensitivity, as do other MS/MS methods aimed at double-bond level characterization, additional enhancement through the integration of fractionation, lipid derivatization, or auxiliary separation techniques, such as FAIMS or TIMS ion mobility, are envisioned. Overall, UVPD-based PRM strategies present an auspicious opportunity to unravel the relationship between subtle changes in lipid structure and biological function.