METHODS

Adapted MS/MSALL Shotgun Lipidomics Approach for Analysisof Cardiolipin Molecular Species

Fei Gao1 · Justice McDaniel1 · Emily Y. Chen1 · Hannah E. Rockwell1 · Cindy Nguyen1 ·Matthew D. Lynes2 · Yu-Hua Tseng2,3 · Rangaprasad Sarangarajan1 · Niven R. Narain1 ·Michael A. Kiebish1

Received: 14 June 2017 / Revised: 3 October 2017 / Accepted: 11 October 2017© 2018 AOCS

Abstract Cardiolipin (Ptd2Gro) is a complex, doublycharged phospholipid located in the inner mitochondrialmembrane where it plays an essential role in regulating bio-energetics. Abnormalities in Ptd2Gro content or composi-tion have been associated with mitochondrial dysfunctionin a variety of disease states. Here, we report the develop-ment of an adapted high-resolution data-independent acqui-sition (DIA) MS/MSALL shotgun lipidomic method toenhance the accuracy and reproducibility of Ptd2Gromolecular species quantitation from biological samples.Utilizing the doubly charged molecular ions and the isoto-pic pattern with negative mode electrospray ionization massspectrometry (ESI-MS) using an adapted MS/MSALL

approach, we profiled more than 150 individual Ptd2Grospecies, including monolysocardiolipin (MLPtd2Gro). Themethod described in this study demonstrated high reproduc-ibility, sensitivity, and throughput with a wide dynamicrange. This high-resolution MS/MSALL shotgun lipidomicsapproach could be extended to screening aberrations ofPtd2Gro metabolism involved in mitochondrial dysfunctionin various pathological conditions and diseases.

Keywords Cardiolipin � Data-independent acquisition �Mass spectrometry � MS/MSALL � Shotgun lipidomics

Lipids (2018) 53: 133–142.

AbbreviationsALCAT acyl-CoA:lysocardiolipin acyltransferaseARA arachidonic acidBAT brown adipose tissueCDP-DAG cytidine diphosphate-diacylglycerolCE collision energyChoGpl choline glycerophospholipidCLS cardiolipin synthaseCMP cytidine monophosphateCUR curtain gasCV coefficient of variationDDA data-dependent acquisitionDHA docosahexaenoic acidDIA data-independent acquisitionDP declustering potentialESI electrospray ionizationEtnGpl ethanolamine glycerophospholipidFA fatty acidHPLC high-performance liquid chromatographyLNA linoleic acidLLOQ the lower limit of quantificationMLCLAT monolysocardiolipin acyltransferaseMLPtd2Gro monolysocardiolipinMS mass spectrometryPLA2 phospholipase A2Ptd2Gro cardiolipinPtdGro phosphatidylglycerolPtdIns phosphatidylinositolPtdOH phosphatidic acid

Electronic supplementary material The online version of this article(doi:10.1002/lipd.12004) contains supplementary material, which isavailable to authorized users.

* Michael A. Kiebishmichael.kiebish@berghealth.com

1 BERG, LLC, 500 Old Connecticut Path, Bldg B, 3rd Floor,Framingham, MA 01701, USA

2 Section on Integrative Physiology and Metabolism, JoslinDiabetes Center, Harvard Medical School, Boston, MA02215, USA

3 Harvard Stem Cell Institute, Harvard University, Cambridge, MA02138, USA

Lipids (2018) 53: 133–142DOI 10.1002/lipd.12004

Lipids (2018) 53: 133–142

PtdSer phosphatidylserineQQQ triple quadrupoleqTOF hybrid quadrupole high-resolution time-of-

flightSD standard deviationTAZ tafazzinTemp temperatureULOQ the upper limit of quantificationWAT white adipose tissue

Introduction

In mammalian cells, cardiolipin (Ptd2Gro) is found almostexclusively in the inner mitochondrial membrane where itplays an essential role in mitochondrial bioenergetics(Chicco & Sparagna, 2007; Claypool & Koehler, 2012; Para-dies, Paradies, De Benedictis, Ruggiero, & Petrosillo, 2014).In eukaryotes, Ptd2Gro is biosynthesized from phosphatidyl-glycerol (PtdGro) and cytidine diphosphate-diacylglycerol(CDP-DAG) by Ptd2Gro synthase on the inner face of theinner mitochondrial membrane (Chicco & Sparagna, 2007).Nascent Ptd2Gro is remodeled by a deacylation–reacylationprocess, which cleaves acyl chains from Ptd2Gro, generatingmonolysocardiolipin (MLPtd2Gro) and, via reacylation fromdonor chains (Fig. 1), allows the enrichment of certain acylchains, especially linoleic acid (18:2), on the backbone ofPtd2Gro in the majority of mammalian tissues (Hoch, 1992).Additional remodeling pathways of Ptd2Gro involve the acylCoA-independent transfer of 18:2 acyl chains directly fromcholine glycerophospholipids (ChoGpl) or ethanolamine gly-cerophospholipids (EtnGpl) via tafazzin (TAZ) (Xu, Kelley,Blanck, & Schlame, 2003).The molecular structure and composition of Ptd2Gro

enhance the functional activity of mitochondrial respiratorychain complexes by maintaining the balance of biophysicaland charge properties of the mitochondrial inner membrane(Paradies et al., 2014; Schlame, Rua, & Greenberg, 2000).In addition to its role in mitochondrial bioenergetics,Ptd2Gro also plays an important regulatory role in cyto-chrome C release, which triggers downstream signalingevents in apoptosis (McMillin & Dowhan, 2002). Abnor-malities in Ptd2Gro composition have been associated withmitochondrial dysfunction in multiple tissues in a variety ofpathological conditions, including Barth syndrome(Kiebish et al., 2013), ischemia (Lesnefsky, Slabe, Stoll,Minkler, & Hoppel, 2001), thyroid neoplasia (Zhang et al.,2016), traumatic brain injury (Ji et al., 2012), and braintumors (Kiebish, Han, Cheng, Chuang, & Seyfried, 2008).Quantitative analysis of Ptd2Gro, particularly at the molec-

ular species level, is largely based on electrospray ionizationmass spectrometry (ESI-MS), either coupled with normal-phase high-performance liquid chromatography (HPLC)

(Sparagna, Johnson, McCune, Moore, & Murphy, 2005;Valianpour, Wanders, Barth, Overmars, & van Gennip,2002), or by direct infusion, that is, shotgun lipidomics (Han,Yang, Yang, Cheng, & Gross, 2006). Under negative-modeESI, cardiolipin species are easily formed as doubly chargedions with a specific isotope distribution ([M – 2H]2−,[M – 2H]2− + 0.5, and [M – 2H]2− + 1), which could beused for the quantification and identification of Ptd2Gro spe-cies (Han et al., 2006). However, under analysis with triplequadrupole (QQQ) mass spectrometers, the lower resolutionmass to charge ratio (m/z) of the doubly charged ion and itsisotopic peaks require deconvolution to provide accuratequantification (Han et al., 2006). Furthermore, traditionalQQQ MS-based lipidomics requires preselection of specificions for analysis of diverse lipids (Han et al., 2006; Han &Gross, 2005). Recently, the strategy of data-independentacquisition (DIA) has been developed to overcome this short-coming by recording all the fragments in isolation windowswith various m/z widths (Gillet et al., 2012). MS/MSALL isone variant of DIA, which utilizes a hybrid quadrupole high-resolution time-of-flight (qTOF) with a 1 m/z isolation win-dow for profiling lipids in biological samples (Gao et al.,2017; Simons et al., 2012). We have developed an adaptedhigh-resolution MS/MSALL shotgun lipidomics methodspecifically for Ptd2Gro molecular species with an analysistime of 6 min. The isolation window and collisionenergy were adjusted to accommodate the isotopic peak[M – 2H]2− + 0.5. The adapted Ptd2Gro MS/MSALL shotgunlipidomics method exhibited a lower coefficient of variation(CV) for Ptd2Gro molecular species than that of the conven-tional MS/MSALL method for other negative lipids.Utilizing the high-resolution MS/MSALL shotgun lipido-

mics approach described in this study, the characteristicPtd2Gro molecular species in lipid extracts of mouse heart,kidney, liver, muscle, pancreas, brown adipose tissue (BAT),white adipose tissue (WAT), and brain were quantified. Thehigh sensitivity and wider dynamic range for quantificationis valuable for studying alterations in Ptd2Gro metabolism ina variety of metabolically diverse biological samples.

Materials and Methods

Mice

All animal procedures were approved by the InstitutionalAnimal Use and Care Committee at Joslin Diabetes Center.C57BL/6 J male mice (Stock no. 000664) were purchasedfrom The Jackson Laboratory and used at 12 weeks of age.All mice were housed in the Joslin Diabetes Center AnimalCare Facility. At 25 weeks of age, mice were sacrificed anddissected. Tissues were snap-frozen in liquid nitrogen.

134 Lipids

Lipids (2018) 53: 133–142

Materials

Ptd2Gro 14:0–14:0–14:0–14:0 standard, heart Ptd2Gro,egg PtdOH, and egg PtdGro standards were purchasedfrom Avanti Polar Lipids (Alabaster, AL, USA). All sol-vents were HPLC or LC/MS grade and were purchasedfrom Fisher Scientific (Waltham, MA, USA) or VWRInternational (Radnor, PA, USA). Mouse tissues were pro-vided by Joslin Diabetes Center and were frozen beforeextraction. All animal procedures were approved by theInstitutional Animal Use and Care Committee at JoslinDiabetes Center.

Sample Preparation and Extraction

Twenty milligrams of mouse tissue samples were trans-ferred to an OMNI bead tube with 0.8 mL 10× diluted PBSand was homogenized for 2 min at 4 �C using the OMNICryo Bead Ruptor 24 (OMNI International, Inc., Kenne-saw, GA, USA). Protein concentration was measured usingBCA assay, and 1 mg of protein was used for extractionwith Ptd2Gro 14:0–14:0–14:0–14:0 added as the internalstandard (IS). Lipid extraction was performed using a mod-ified Bligh and Dyer method as previously described (Gao

et al., 2017). Extraction was automated using a customizedsequence on a Hamilton Robotics STARlet system(Hamilton, Reno, NV, USA) to meet the high-throughputneeds. Lipid extracts were dried under N2 and reconstitutedin chloroform/methanol (1:1, by vol.). Samples wereflushed with N2 and stored at −20 �C.To study the dynamic range of Ptd2Gro under the

ESI-MS, the synthetic Ptd2Gro standard, Ptd2Gro 14:0–14:0–14:0–14:0, was spiked at different serial concentra-tions into 1 mg protein extract from mouse hearthomogenate to generate the calibration curves. For assayreproducibility, five aliquots of 1 mg heart homogenatewere extracted with IS and analyzed using two differentmethods – one for negative lipids, including phosphati-dylserine (PtdSer), phosphatidylglycerol (PtdGro), phos-phatidylinositol (PtdIns), and phosphatidic acid (PtdOH),and the other for the Ptd2Gro species. The coefficient ofvariation (CV) was calculated for comparison.To assess carry-over, aliquots of 10 nmol each for heart

Ptd2Gro, egg PtdOH, and egg PtdGro extracts with theappropriate mixture of internal standards cocktail were pre-pared in 0.4 mL chloroform/methanol (1:1, v/v) separatelyand analyzed using the described cardiolipin mass spec-trometry method.

CLS

TAZ

CDP-DAG

CMP

EtnGpl, ChoGplFFA

PLA2

MLCLAT/ALCAT

Acyl-CoA CoA-SH

Nascent Ptd2GroPtdGro

Remodeled Ptd2GroMLPtd2Gro

OH

P P

OH

OHP

OH

P P

OH

HO

PP

PLA2

Fig. 1 Cardiolipin remodeling in mammalian cells. ALCAT, acyl-CoA:lysocardiolipin acyltransferase; CDP-DAG, cytidine diphosphate-diacylgly-cerol; Ptd2Gro, cardiolipin; CLS, cardiolipin synthase; CMP, cytidine monophosphate; MLPtd2Gro, monolysocardiolipin; MLCLAT, monolysocar-diolipin acyltransferase; ChoGpl, choline glycerophospholipids; EtnGpl, ethanolamine glycerophospholipids; PtdGro, phosphatidylglycerol; PLA2,phospholipase A2; TAZ, tafazzin

135Lipids

Lipids (2018) 53: 133–142

Direct Infusion Quadrupole Time of Flight (qTOF)MS/MSALL Mass Spectrometry

Concentrated samples were diluted 50× in isopropanol/metha-nol/acetonitrile/H2O (3:3:3:1, by vol) with 2 mM ammoniumacetate, and 50 μL diluted lipid extract was automaticallyloaded and directly delivered to the ESI source using anEkspert microLC 200 system (SCIEX, Framingham, MA,USA) with a flow rate of 6 μL/min on a customized sampleloop. The mass spectrometry experiments were carried out ona TripleTOF 5600+ (SCIEX). ESI source parameters were setat nebulizing gases GS1 at 10, GS2 at 10, curtain gas at15, ion spray voltage at −4500 V, declustering potential at100 V, and temperature at 300 �C. The atmospheric pressurechemical ionization probe was connected to a calibrant pumpthat delivered the mass calibration solution for MS andMS/MS. The MS/MSALL data acquisition was controlled byAnalyst® software (SCIEX). For the conventional negativelipid MS/MSALL analysis (Gao et al., 2017; Simons et al.,2012), all precursors were selected in the Q1 quadrupole at1 Da isolation window in the MS range 200–1200, in whichthe precursor ions were equally spread in the isolation win-dows (Fig. 2). The collision energy for negative lipidsMS/MSALL was −30 � 15 V. For the Ptd2Gro species, theMS range was shortened to 400–1000 with the isolation win-dow shifting to fit the doubly charged isotopologue Ptd2Grospecies (Fig. 2). In addition, the optimal collision energy forPtd2Gro species was −20 � 10 V. The identification ofPtd2Gro species was based on the high-resolution molecularweight, the [M – 2H]2− + 0.5 isotope, and the specific fattyacyl chain fragments. The quantification was performed bycomparing the peak area of molecular species isotope(M + 0.5) to that of the IS within a linear dynamic rangewith the isotopic corrections (Han et al., 2006) and normal-ized to 1 mg protein across different tissues.

Results

Data-Independent Acquisition (DIA) MS/MSALL forPtd2Gro Molecular Species

Under negative ESI, Ptd2Gro molecular species demonstratea different ionization compared to other negatively chargedphospholipids through the formation of a doubly chargedmolecular ion. This doubly charged molecular ion demon-strated an isotopic pattern with [M – 2H]2−, [M – 2H]2− + 0.5, and [M – 2H]2− + 1 peaks (Fig. 2d). Using a previ-ously assigned strategy for shotgun lipidomics (Han et al.,2006), the diagnostic [M – 2H]2− + 0.5 peak was exploitedfor the identification and quantification of Ptd2Gro using thehigh-resolution DIA MS/MSALL shotgun lipidomics strat-egy. Applying an adapted MS/MSALL approach for Ptd2Grocompared to the conventional analysis method for negative

phospholipids (Fig. 2a), the Ptd2Gro precursors wereselected at a shifted 1 – m/z wide isolation window in Q1 inorder to accommodate the doubly charged [M – 2H]2− + 0.5peak with a narrow scan range 400–1000 m/z (Fig. 2b). Inaddition, the collision energy (CE) for Ptd2Gro was lowerthan those of the other negative phospholipids in order topreserve both doubly charged molecular ions and fragments.This process is illustrated in Fig. 2c in which the precursorions are isolated in a 1 – m/z wide window in Q1, fragmen-ted in Q2 using ramping CE, and the molecular ions andfragment ions were all analyzed by the high-resolution TOFMS analyzer at a high scan speed.

Optimization of High-Resolution MS/MSALL ShotgunLipidomics for Ptd2Gro Molecular Species

Ptd2Gro species were isolated in the 1 – m/z isolation windowby Q1, and then dissociated in the collision cell (Fig. 2). CE isa key factor that influences the isotopic peaks and the frag-ments of the Ptd2Gro species and requires optimization. Onenanomolar Ptd2Gro 14:0–14:0–14:0–14:0 solution wasinfused into the mass spectrometer and the doubly chargedmolecular ions were fragmented under a ramping CE rangingfrom −100 to 0 V (Fig. 3a). Them/z 619.9 of the isotopic peak[M – 2H]2− + 0.5 and the m/z 227.2 of the fatty acid fragmentpeak were monitored during the infusion. The optimal CErange for the [M – 2H]2− + 0.5 peak was between −20 and−5 V. We chose a ramping CE as −20� 10 V as our final CEfor the Ptd2Gro species to produce both stable [M – 2H]2− + 0.5 and the diagnostic fatty acid fragment (Fig. 3b).Quantification using the isotopic peak [M – 2H]2− + 0.5under lower-resolution mass spectrometers was reportedpreviously (Han et al., 2006; Kiebish et al., 2013). To deter-mine the linear dynamic range and sensitivity of the high-resolution DIA method for Ptd2Gro species quantification, aserial dilution of Ptd2Gro 14:0–14:0–14:0–14:0 in one ali-quot of 1 mg mouse heart homogenate was analyzed andthe calibration curve ranging from 0.04 to 400 nM wasestablished with an R2 regression value of 0.9993 (Fig. 3c).

Specificity of High-Resolution MS/MSALL ShotgunLipidomics for Ptd2Gro Molecular Species

To determine the specificity of the adapted cardiolipinMS/MSALL method, we examined potential interference ofother phospholipids standards, with Ptd2Gro analysis, namelyheart Ptd2Gro versus egg PtdOH or egg PtdGro. We chosePtdOH and PtdGro extracts because both of phospholipidsionize in the negative mode and share a similar mass rangethat partially overlaps with that of Ptd2Gro. Analysis of eggPtdOH and egg PtdGro stocks revealed <3% interference forthe analysis of cardiolipin, which could be associated withbackground (Fig. S1, Supporting Information).

136 Lipids

Lipids (2018) 53: 133–142

Assessment of Variation for the Ptd2Gro MS/MSALL

Method for Quantifying Molecular Species

To assess the reproducibility of Ptd2Gro adaptedMS/MSALL method, five aliquots of 1 mg mouse heart

extract were analyzed using the method for the conven-tional negative lipid analysis as well as the adapted methodfor Ptd2Gro analysis (as described in the methods section).The CV of the top 20 Ptd2Gro species in the mouse hearttissue using the novel Ptd2Gro method was less than 10%,

Fig. 2 Data-independent acquisition MS/MSALL schemes for common negative lipids and cardiolipin. (a) MS/MSALL for common negativelipids. All precursors are selected in the Q1 at a 1 – m/z wide isolation window across 200–1200 m/z. Each isolation window is selected toaccommodate singly charged negative lipid species with a higher collision energy. (b) MS/MSALL for Ptd2Gro. The Ptd2Gro precursors areselected in Q1 at a 1 – m/z wide isolation window across 400–1000 m/z with low collision energy and each single adjusted isolated window tofit the doubly charged Ptd2Gro species and isotopic peaks (see inset). (c) ESI-QTOF-MS/MSALL workflow. (d) The molecular structure andisotopic distribution of Ptd2Gro 18:2–18:2–18:2–18:2 under negative ESI conditions. CE, collision energy; ESI, electrospray ionization;PtdOH, phosphatidic acid; PtdGro, phosphatidylglycerol; PtdIns, phosphatidylinositol; PtdSer, phosphatidylserine; Q, quadrupole; TOF, timeof flight mass spectrometer

137Lipids

Lipids (2018) 53: 133–142

whereas that of most Ptd2Gro using the method for othernegative lipids was more than 10% (Fig. 4), most likelybecause the isotopic peaks are near the edge of the isolationwindow and these ions become less stable than the ions inthe middle of the window. The quantitative value of cardio-lipin species also demonstrated minimal variation in repli-cates from the adapted MS/MSALL Ptd2Gro workflow

compared to negative MS/MSALL analysis (Fig. 4b). Thelower variation in the new adapted cardiolipin method com-pared to the standard negative MS/MSALL method may inpart be due to the lower collision energy creating more sta-ble ions for the 0.5 isotopologue compared to the higherCE in the negative MS/MSALL method, which might bemore unstable, creating more variability.

Fig. 3 Optimization of high-resolution shotgun lipidomics for cardiolipin. (a) Plot of intensity of the isotopic [M – 2H]2− + 0.5 peak (m/z 619.9257)(upper panel) and of the fatty acid fragment m/z 227.2023 (lower panel) of Ptd2Gro 14:0–14:0–14:0–14:0 against varying rolling collision energy (CE).(b) MS/MS spectra and fragments of Ptd2Gro 14:0–14:0–14:0–14:0 at CE = −20 V. (c) The peak area of the isotopic ion [M – 2H]2− + 0.5(619.4! 619.9257) against the concentration of Ptd2Gro 14:0–14:0–14:0–14:0 yielded a linear curve (R2 = 0.9993) when spiked in mouse tissue. Theinsert table shows the related quantitative parameters. The spectra were acquired using the MS/MSALL for cardiolipin under the following conditions:GS1 = 10, GS2 = 10, CUR = 15, Temp = 300 �C, and DP = 100 V. CE, collision energy; Ptd2Gro, cardiolipin; CUR, curtain gas; DP, declusteringpotential; FA, fatty acid; GS, nitrogen gas flow; LLOQ, the lower limit of quantification; ULOQ, the upper limit of quantification; Temp, temperature

138 Lipids

Lipids (2018) 53: 133–142

Analysis of Ptd2Gro Molecular Species from DifferentMouse Tissues

We applied the adapted MS/MSALL DIA Ptd2Gro shotgunlipidomics method to quantify the Ptd2Gro molecular speciesfrom various mouse tissues, including heart, kidney, liver,muscle, pancreas, brown adipose tissue (BAT), white adi-pose tissue (WAT), and brain (Fig. 5). The molecular struc-ture of Ptd2Gro species was identified by the [M – 2H]2− +0.5 isotopic peaks and the MS/MS spectra, as illustrated inFig. 6 where endogenous Ptd2Gro 18:2–18:2–18:2–18:2 and

Ptd2Gro 18:2–18:2–18:2–22:6 in the mouse heart were con-firmed. Across the profile of Ptd2Gro molecular species, twoseparate clusters of MLPtd2Gro (around m/z 600) andPtd2Gro (between m/z 650 and 850) were clearly present.Among all tissues, the heart contained the highest content ofPtd2Gro per milligram of protein, followed by kidney andliver (Fig. 5a). The dominant Ptd2Gro species in all tissues(Table S1), except brain (Table S2), were 18:2–18:2–18:2–18:2 and 18:2–18:2–18:2–18:1, as previously reported(Han et al., 2006) (Fig. 5b). For mouse brain tissue, themajor Ptd2Gro species were enriched with oleic acid(FA 18:1) and polyunsaturated fatty acids (PUFA), such asarachidonic acid (ARA) and docosahexaenoic acid (DHA).

Discussion

Ptd2Gro is a polyglycerol phospholipid composed of twophosphatidyl moieties and four fatty acyl chains, and playsan essential role in mitochondrial function (Paradies et al.,2014; Ren, Phoon, & Schlame, 2014). Disruption ofPtd2Gro content or composition in mammalian cells hasbeen related to the initiation of mitochondrial dysfunctionin various disease states (Chicco & Sparagna, 2007; Clay-pool & Koehler, 2012). The complexity and low abundanceof Ptd2Gro in most tissues poses a bioanalytical challengeto accurately quantify Ptd2Gro at the molecular specieslevel. Under ESI conditions, Ptd2Gro forms as a doublynegatively charged ion with a distinctive isotopic distribu-tion, the characteristics of which could be exploited forPtd2Gro identification and quantification (Han et al., 2006).However, most previous studies were performed with lowerresolution mass spectrometers, such as QQQ, and the isoto-pic peaks can overlap and require deconvolution for accu-rate quantitation (Han et al., 2006). In this study, weapplied an adapted high-resolution qTOF mass spectrome-try along with the DIA MS/MSALL strategy to accuratelyprofile the Ptd2Gro molecular species from different mousetissues. The isotopic peaks were clearly identified and usedfor subsequent quantification. The MS/MSALL isolationwindow, the collision energy, and other parameters wereoptimized to provide a high-throughput, sensitive methodfor specifically quantifying Ptd2Gro molecular species fromvarious biological samples.DIA is a strategy of molecular analysis in which all ions

within a selected m/z window are fragmented and analyzed(Gillet et al., 2012). DIA methods have been widely used forproteomics (Distler et al., 2014), metabolomics (Tsugawaet al., 2015), and lipidomics (Simons et al., 2012). Com-pared with traditional data-dependent acquisition (DDA)methods, DIA overcomes the disadvantages of irreproduc-ibility and undersampling issues of DDA (Gillet et al., 2012)

Fig. 4 Assessment of the two different methods-MS/MSALL for nega-tive lipids and MS/MSALL for cardiolipin. (a) The coefficient of varia-tion (CV) and (b) The concentration profile of the top 20 Ptd2Grospecies from mouse heart tissue. Mass spectrometry conditions:GS1 = 10, GS2 = 10, CUR = 15, Temp = 300 �C, DP = 100 V,CE = −35 � 15 V, and −20 � 10 V for the negative lipid MS/MSALL

and the Ptd2Gro MS/MSALL methods, respectively. Data areexpressed as mean � SD (n = 5)

139Lipids

Lipids (2018) 53: 133–142

by covering all fragments of all molecules within the sample.MS/MSALL is a DIA method developed specifically for lipi-domics (Gao et al., 2017; Simons et al., 2012). TheMS/MSALL approach uses 1 m/z isolation window to selectthe molecular ions, which not only collects all the fragmentsfrom the precursor ions but also maintains the connectionsbetween the precursors and fragments. For Ptd2Gro analysis,we adjusted the conventional method by shifting the windowedges and lowering the collision energy to accommodate the[M – 2H]2− + 0.5 isotopic ion to achieve more stable signalsand lower variance between signals.

More than 150 Ptd2Gro molecular species, includingMLPtd2Gro and Ptd2Gro, were quantified from eight dif-ferent mouse tissues. Among the tissues investigated,heart provided the highest content of Ptd2Gro, with thedominant species of Ptd2Gro being 18:2–18:2–18:2–18:2.Our findings are in line with previous results analyzed byQQQ (Han et al., 2006). In addition to heart, liver, pan-creas, kidney, and muscle, we also determined Ptd2Gromolecular species in BAT and WAT. The Ptd2Gro contentof BAT is more than twice that of WAT, indicating higheramounts of mitochondria and the high metabolic activity

Fig. 5 Cardiolipin species profiles from different mouse tissues. (a) The total Ptd2Gro and (b) the Ptd2Gro molecular species profile from heart,kidney, liver, muscle, pancreas, brown adipose tissue, white adipose tissue demonstrates the predominant Ptd2Gro species present is18.2–18.2–18.2–18.2, whereas brain Ptd2Gro species are vastly different as previously described (Bayir et al., 2007; Cheng et al., 2008; Kiebishet al., 2008). The spectra were acquired using the MS/MSALL for Ptd2Gro method under the following conditions: GS1 = 10, GS2 = 10,CUR = 15, Temp = 300 �C, DP = 100 V, and CE = −20 � 10 V. Data are expressed as mean � SD (n = 3)

140 Lipids

Lipids (2018) 53: 133–142

of BAT, which helps regulate energy expenditure by ther-mogenesis (Cannon & Nedergaard, 2004; Cypess et al.,2009). The Ptd2Gro profile from brain tissue is quite differ-ent from other tissues and contains more oleic acid (18:1)and poly-unsaturated fatty acids (PUFA) such as ARA and

DHA. The Ptd2Gro species we detected from the brain areconsistent with previously reported studies (Bayir et al.,2007; Cheng et al., 2008; Kiebish et al., 2008).In summary, we developed an adapted high-resolution

DIA MS/MSALL approach to accurately quantify and

Fig. 6 MS/MS spectra and fragmentation pattern for Ptd2Gro species identified from mouse heart. (a) Species Ptd2Gro 18:2–18:2–18:2–18:2 and (b) 18:2–18:2–18:2–22:6 were identified from mouse heart. Insets: The enlarged isotopic distribution of [M – 2H]2− of Ptd2Grospecies. The spectra were obtained using the MS/MSALL for Ptd2Gro under the following conditions: GS1 = 10, GS2 = 10, CUR = 15,Temp = 300 �C, DP = 100 V, and CE = −20 � 10 V. DHA, docosahexaenoic acid; LNA, linoleic acid

141Lipids

Lipids (2018) 53: 133–142

profile Ptd2Gro molecular species from biological samples.The specialized chemical properties of the doubly chargedisotopic molecular ions under the negative ESI mode wereused to accurately analyze a multitude of individualPtd2Gro molecular species, including MLPtd2Gro species.The method is reproducible and sensitive with limits ofquantification as low as 40 pM. The workflow presentedhere offers an efficient and accurate approach for the quan-titation of Ptd2Gro molecular species in biological samplesand could be extended for screening abnormalities ofPtd2Gro metabolism in various disease states.

Acknowledgements This work was supported in part by USNational Institutes of Health (NIH) grants R01DK077097 (to Y.-H.T.), a research grant from the American Diabetes Foundation (ADA7-12-BS-191, to Y.-H.T.), and funding from the Harvard Stem CellInstitute (to Y.-H.T.). M.D.L was supported by NIH fellowships(T32DK007260 and F32DK102320). This study was also funded byBERG, LLC. The authors would like to thank K. Wilkinson and K.Thapa for scientific writing assistance.

Conflict of interest F.G., J.M., E.C., H.R., C.N., R.S., N.N., andM.K. are current employees of BERG, LLC, and have stock options.N.N. and R.S. are cofounders of BERG, LLC.

References

Bayir, H., Tyurin, V. A., Tyurina, Y. Y., Viner, R., Ritov, V.,Amoscato, A. A., … Zhao, Q. (2007) Selective early cardiolipinperoxidation after traumatic brain injury: An oxidative lipidomicsanalysis. Annals of Neurology, 62:154–169.

Cannon, B., & Nedergaard, J. (2004) Brown adipose tissue: Functionand physiological significance. Physiological Reviews, 84:277–359.

Cheng, H., Mancuso, D. J., Jiang, X., Guan, S., Yang, J., Yang, K.,… Sun, G. (2008) Shotgun lipidomics reveals the temporallydependent, highly diversified cardiolipin profile in the mammalianbrain: Temporally coordinated postnatal diversification of cardioli-pin molecular species with neuronal remodeling. Biochemistry, 47:5869–5880.

Chicco, A. J., & Sparagna, G. C. (2007) Role of cardiolipin alterationsin mitochondrial dysfunction and disease. American Journal ofPhysiology. Cell Physiology, 292:C33–C44.

Claypool, S. M., & Koehler, C. M. (2012) The complexity of cardioli-pin in health and disease. Trends in Biochemical Sciences, 37:32–41.

Cypess, A. M., Lehman, S., Williams, G., Tal, I., Rodman, D.,Goldfine, A. B., … Kuo, F. C. (2009) Identification and importanceof brown adipose tissue in adult humans. New England Journal ofMedicine, 360:1509–1517.

Distler, U., Kuharev, J., Navarro, P., Levin, Y., Schild, H., &Tenzer, S. (2014) Drift time-specific collision energies enabledeep-coverage data-independent acquisition proteomics. NatureMethods, 11:167–170.

Gao, F., McDaniel, J., Chen, E. Y., Rockwell, H. E., Drolet, J.,Vishnudas, V. K., … Tolstikov, V. (2017) Dynamic and temporalassessment of human dried blood spot MS/MS(ALL) shotgun lipi-domics analysis. Nutrition & Metabolism, 14:28.

Gillet, L. C., Navarro, P., Tate, S., Röst, H., Selevsek, N., Reiter, L.,… Bonner, R. (2012) Targeted data extraction of the MS/MS spec-tra generated by data-independent acquisition: A new concept for

consistent and accurate proteome analysis. Molecular & CellularProteomics, 11:O111.016717.

Han, X., & Gross, R. W. (2005) Shotgun lipidomics: Electrosprayionization mass spectrometric analysis and quantitation of cellularlipidomes directly from crude extracts of biological samples. MassSpectrometry Reviews, 24:367–412.

Han, X., Yang, K., Yang, J., Cheng, H., & Gross, R. W. (2006) Shot-gun lipidomics of cardiolipin molecular species in lipid extracts ofbiological samples. Journal of Lipid Research, 47:864–879.

Hoch, F. L. (1992) Cardiolipins and biomembrane function. Biochi-mica et Biophysica Acta, 1113:71–133.

Ji, J., Kline, A. E., Amoscato, A., Samhan-Arias, A. K.,Sparvero, L. J., Tyurin, V. A., … Tyurina, Y. Y. (2012) Lipidomicsidentifies cardiolipin oxidation as a mitochondrial target for redoxtherapy of brain injury. Nature Neuroscience, 15:1407–1413.

Kiebish, M. A., Han, X., Cheng, H., Chuang, J. H., & Seyfried, T. N.(2008) Cardiolipin and electron transport chain abnormalities inmouse brain tumor mitochondria: Lipidomic evidence supportingthe Warburg theory of cancer. Journal of Lipid Research, 49:2545–2556.

Kiebish, M. A., Yang, K., Liu, X., Mancuso, D. J., Guan, S.,Zhao, Z., … Sims, H. F. (2013) Dysfunctional cardiac mitochon-drial bioenergetic, lipidomic, and signaling in a murine model ofBarth syndrome. Journal of Lipid Research, 54:1312–1325.

Lesnefsky, E. J., Slabe, T. J., Stoll, M. S., Minkler, P. E., &Hoppel, C. L. (2001) Myocardial ischemia selectively depletes car-diolipin in rabbit heart subsarcolemmal mitochondria. AmericanJournal of Physiology Heart and Circulatory Physiology, 280:H2770–H2778.

McMillin, J. B., & Dowhan, W. (2002) Cardiolipin and apoptosis.Biochimica et Biophysica Acta, 1585:97–107.

Paradies, G., Paradies, V., De Benedictis, V., Ruggiero, F. M., &Petrosillo, G. (2014) Functional role of cardiolipin in mitochondrialbioenergetics. Biochimica et Biophysica Acta, 1837:408–417.

Ren, M., Phoon, C. K., & Schlame, M. (2014) Metabolism and func-tion of mitochondrial cardiolipin. Progress in Lipid Research,55:1–16.

Schlame, M., Rua, D., & Greenberg, M. L. (2000) The biosynthesisand functional role of cardiolipin. Progress in Lipid Research, 39:257–288.

Simons, B., Kauhanen, D., Sylvänne, T., Tarasov, K.,Duchoslav, E., & Ekroos, K. (2012) Shotgun lipidomics by sequen-tial precursor ion fragmentation on a hybrid Quadrupoletime-of-flight mass spectrometer. Metabolites, 2:195–213.

Sparagna, G. C., Johnson, C. A., McCune, S. A., Moore, R. L., &Murphy, R. C. (2005) Quantitation of cardiolipin molecular speciesin spontaneously hypertensive heart failure rats using electrosprayionization mass spectrometry. Journal of Lipid Research, 46:1196–1204.

Tsugawa, H., Cajka, T., Kind, T., Ma, Y., Higgins, B., Ikeda, K., …Kanazawa, M. (2015) MS-DIAL: Data-independent MS/MS decon-volution for comprehensive metabolome analysis. Nature Methods,12:523–526.

Valianpour, F., Wanders, R. J., Barth, P. G., Overmars, H., & vanGennip, A. H. (2002) Quantitative and compositional study of car-diolipin in platelets by electrospray ionization mass spectrometry:Application for the identification of Barth syndrome patients. Clini-cal Chemistry, 48:1390–1397.

Xu, Y., Kelley, R. I., Blanck, T. J. J., & Schlame, M. (2003) Remo-deling of cardiolipin by phospholipid transacylation. Journal ofBiological Chemistry, 278:51380–51385.

Zhang, J., Yu, W., Ryu, S. W., Lin, J., Buentello, G., Tibshirani, R.,…Suliburk, J. (2016) Cardiolipins are biomarkers of mitochondria-richthyroid oncocytic tumors. Cancer Research, 76:6588–6597.

142 Lipids

Lipids (2018) 53: 133–142