LC-MS-based serum metabolomic analysis reveals dysregulation of phosphatidylcholines in esophageal squamous cell carcinoma

Journal of Proteomics xxx (2015) xxx–xxx

JPROT-02154; No of Pages 7

Contents lists available at ScienceDirect

Journal of Proteomics

j ourna l homepage: www.e lsev ie r .com/ locate / jp rot

LC–MS-based serum metabolomic analysis reveals dysregulation ofphosphatidylcholines in esophageal squamous cell carcinoma☆

Sartaj Ahmad Mir a,b, Pavithra Rajagopalan a,c, Ankit P. Jain a,c, Aafaque Ahmad Khan a,c, Keshava.K. Datta a,c,Sonali V. Mohan a, Syed Salman Lateef d, Nandini Sahasrabuddhe a, B.L. Somani a, T.S. Keshava Prasad a,Aditi Chatterjee a, K.V. Veerendra Kumar e, M. VijayaKumar e, Rekha V. Kumar f,Seetaramanjaneyulu Gundimeda d, Akhilesh Pandey g,h,i,j, Harsha Gowda a,⁎a Institute of Bioinformatics, International Technology Park, Bangalore 560066, Indiab Manipal University, Manipal 576104, Indiac School of Biotechnology, KIIT University, Bhubaneswar 751024, Indiad Agilent Technologies, Bangalore 560048, Indiae Department of Surgery, Kidwai Memorial Institute of Oncology, Bangalore 560029, Indiaf Department of Pathology, Kidwai Memorial Institute of Oncology, Bangalore 560029, Indiag McKusick-Nathans Institute of Genetic Medicine, Johns Hopkins University School of Medicine, Baltimore, MD 21205, USAh Department of Biological Chemistry, Johns Hopkins University School of Medicine, Baltimore, MD 21205, USAi Department of Pathology, Johns Hopkins University School of Medicine, Baltimore, MD 21205, USAj Department of Oncology, Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA

☆ This article is part of a Special Issue entitled: Proteom⁎ Corresponding author at: Institute of Bioinformatic

Building, International Tech Park, Bangalore 560 066, IndiE-mail address: [email protected] (H. Gowd

http://dx.doi.org/10.1016/j.jprot.2015.05.0131874-3919/© 2015 Published by Elsevier B.V.

Please cite this article as: S.A. Mir, et al., LC–Msquamous cell carcinoma, J Prot (2015), http

a b s t r a c t

Article history:Received 16 February 2015Received in revised form 10 April 2015Accepted 3 May 2015Available online xxxx

Keywords:CancerCholinesPCsLyso PCsSerum metabolomeMetabolism

Esophageal squamous cell carcinoma (ESCC) is one of themost aggressive cancers with poor prognosis. Here, wecarried out liquid chromatography–quadrupole time-of-flight mass spectrometry (LC–Q–TOF-MS)-baseduntargeted metabolomic analysis of ESCC serum samples. Statistical analysis resulted in the identification of652 significantly dysregulated molecular features in serum from ESCC patients as compared to the healthy sub-jects. Phosphatidylcholines were identified as a major class of dysregulated metabolites in this study suggestingpotential perturbation of phosphocholinemetabolism in ESCC. By using a targetedMS/MS approach both in pos-itive and negativemode, we were able to characterize and confirm the structure of sevenmetabolites. Our studydescribes a quantitative LC–MS approach for characterizing dysregulated lipid metabolism in ESCC.

Biological significanceAlteredmetabolism is a hallmark of cancer. We carried out (LC–MS)-based untargeted metabolomic profiling ofserum from esophageal squamous cell carcinoma (ESCC) patients to characterize dysregulated metabolites.Phosphatidylcholine metabolism was found to be significantly altered in ESCC. Our study illustrates the use ofmass spectrometry-based metabolomic analysis to characterize molecular alterations associated with ESCC.This article is part of a Special Issue entitled: Proteomics in India.

© 2015 Published by Elsevier B.V.

1. Introduction

Metabolomics is a powerful tool to investigate metabolomic pertur-bations in cancers [1,2]. Metabolomics has been used for identificationof potential biomarkers in breast [3], colorectal [4], pancreatic [5], liver[6–8], ovarian [9], kidney [10] and prostrate [11] cancers. Perturbationin the levels of bile acids, histidine and inosine were found in gas chro-matography–mass spectrometry (GC–MS) and liquid chromatography–

ics in India.s, Unit 1, 7th Floor, Discoverera.a).

S-based serummetabolomic://dx.doi.org/10.1016/j.jprot.2

mass spectrometry (LC–MS)-based studies of hepatocellular cancer [8].Altered levels of metabolites such as acylcarnitines, quinolinate,4 hydroxybenzoate, and gentisate have been reported in kidney cancer[12–14]. Several potential biomarkers have been reported in ESCC bycarrying out serummetabolomics [15].

Several complementary analytical platforms such as NMR, GC–MSand LC–MS are available for analysis of the metabolome [16]. LC–MS isa suitable and popular platform for in-depth investigation of themetab-olome owing to its high-throughput nature, sensitivity and relative easeof sample preparation [17,18]. There are reports pertaining to the me-tabolome of upper gastrointestinal tract cancers using nuclearmagneticresonance (NMR) spectroscopy or gas chromatography–mass spec-trometry (GC–MS) [19–22]. While both of these techniques have their

analysis reveals dysregulation of phosphatidylcholines in esophageal015.05.013

2 S.A. Mir et al. / Journal of Proteomics xxx (2015) xxx–xxx

advantages, they are not very conducive to structural elucidation of cer-tain metabolites. For example, NMR analysis is often inconclusive in theanalysis of species containing long chain fatty acid moieties while polarmolecules are difficult to analyze by GC–MS technique without a priorderivatization step.

We carried out untargeted global metabolomic profiling of serumfrom ESCC patients and compared them to age and sex matched con-trols. Metabolites that showed altered levels were identified and furthercharacterized by targeted fragmentation in positive and negative ionmode. Large scale validation of these metabolites might prove usefulin identifying novel blood-based biomarkers of ESCC.

2. Materials & methods

2.1. Sample collection

All blood samples were collected after obtaining approval from theinstitutional review board at the KidwaiMemorial Institute of Oncology,Bangalore, India. Forty blood samples were collected from patients whounderwent curative surgery for the removal of tumor and had histolog-ically confirmed esophageal squamous cell carcinoma (ESCC). PediatricESCC caseswere not included in this study. All the sampleswere collect-ed from pre-operative and treatment naïve patients. Healthy subjectswith no prior health conditions such as diabetes and cardiovascular dis-ease were selected as controls. Blood was collected from patients andcontrol individuals after obtaining informed consent. The sample detailsare provided in Supplementary Table 1. The blood was allowed to clotfor 30 min followed by centrifugation at 2500 rpm for 10 min to collectserum fractions. Subsequently, the serum samples were stored at−80 °C until further analysis.

2.2. Metabolite extraction

Metabolite extraction was carried out from 40 ESCC and 10 con-trol serum samples by adding 400 μl of methanol to 25 μl of serumfollowed by overnight incubation at −20 °C. The metabolite extractwas centrifuged at 13,000 rpm for 15 min at room temperature.The supernatant was collected, dried and stored at −20 °C untilLC–MS analysis.

2.3. LC–MS analysis

The metabolite extracts were reconstituted in 500 μl of 50% metha-nol and each sample was analyzed in triplicate on 6550 iFunnel Q–TOF LC–MS (Agilent Technologies, Santa Clara, CA, USA) equippedwith Dual AJS ESI. The metabolites were separated on 1260 infinityHPLC system (Agilent Technologies, Santa Clara, CA, USA) by injecting5 μl of the extract on a Polaris (150 × 2 mm, 3 μ; Agilent Technologies,Santa Clara, CA, USA) column.Afinal concentration of 5mMammoniumacetate was added to both solvent A (methanol: water: acetic acid75:24:1, v/v/v) and solvent B (methanol: acetic acid 99:1, v/v).The me-tabolites were resolved on the column by increasing gradient of solventB from 10% to 100% over 15 min. The gradient was held at 100% B for20 min before returning to 10% for re-equilibration for 5 min. Nitrogenwas used as the nebulizing gas. Dual Automatic Jet Stream (AJS)Electrospray Ionization (ESI) source was kept at a voltage (VCap) of3500 V in both positive and negative ion mode. The fragmentor voltagewasmaintained at 175 V for both ion polarities. The drying gas temper-ature was 200 °C, drying gas flow rate was 14 L/min and nebulizer pres-sure was 35 psi. The sheath gas temperature was set at 350 °C with aflow rate of 10 L/min.

2.4. Metabolite identification and statistical analysis

Raw data were acquired by using MassHunter acquisition software(Agilent Technologies, Santa Clara, CA, USA) in an untargeted mode.

Please cite this article as: S.A. Mir, et al., LC–MS-based serummetabolomicsquamous cell carcinoma, J Prot (2015), http://dx.doi.org/10.1016/j.jprot.2

The data pre-processingwas carried out using themolecular feature ex-tractor (MFE), an in-built algorithmof theMassHunter Qualitative anal-ysis software. The processed data was then used to generate a list ofunique molecular features with high mass accuracy (b5 ppm). The listof features generated inMFEwas exported toMass Profiler Professional(MPP) for interpretation and statistical analysis in the form of com-pound exchange files (CEF). The CEF files were grouped according tothe conditions and statistical analysis was carried out to find out thestatistically significant dysregulated molecular features in ESCC ascompared to the control serum samples. Supervised principal compo-nent analysis (PCA) was performed to demonstrate the variance ofmetabolomic phenotypes within the two conditions and across all sam-ples. Statistical evaluation of the data was performed by Welch's un-paired t-test for the two conditions. A cut-off value of p b 0.01 wasconsidered statistically significant and Benjamini and Hochberg falsediscovery rate was set to 5% for testing corrections [23]. A log trans-formed fold change of ≥5 andp-value ≤0.01were the parameters select-ed to identify metabolites that showed altered levels between the twoconditions. These significantly dysregulatedmetaboliteswere identifiedby matching accurate mass to personal compound database and library(PCDL; Agilent Technologies). For those features which did not get an-notated with PCDL alone, manual identifications were carried out bymatching accurate mass in HMDB [24], LipidMAPS [25] and METLIN[26] databases.

2.5. Targeted LC–MS/MS analysis

The list of significantly dysregulated molecular features wasexported from MPP as an inclusion list for targeted MS/MS analysis.The target list contained information about measured mass and reten-tion time for all the molecular features. MS/MS analysis was carriedout in both positive and negative modes. MS/MS spectra were acquiredat collision energies of 22 and 18 in positive and negative modes, re-spectively. Q–TOF was operated in extended dynamic range with highresolution filtermode and spectra were acquired at a rate of one spectraper second. Mass range was set at 100–1200 m/z for MS and 25–1200m/z for MS/MS data. Other parameters such as drying gas temper-ature, drying gas flow, nebulizer pressure, capillary and fragmentorvoltage were kept the same as that of MS only acquisition as describedearlier.

3. Results

3.1. Comparative metabolite profiling of serum from ESCC patients andcontrols

LC–MS analysis resulted in identification of 652 statistically significantdysregulated molecular features in ESCC as compared to healthy controls(Supplementary Table 2). PCA analysis of LC–MS results was carried outto demonstrate the variance between metabolomic phenotypes of ESCCand control samples (Supplementary Fig. 1). Several lipids such asglycerophosphocholines (PC), glycerophosphoethanolamines (PE),sphingomyelins, ceramides, acyl carnitines, glycerophosphoserines andfree fatty acidswere identified in serum. Among glycerophosphocholines,monoacyl glycerophosphocholines or lysophosphatidylcholines (LysoPCs) and diacyl glycerophosphocholines were identified. Some PCs andPEs were found to have the same nominal mass and thus, MS/MS exper-iments were necessitated to identify the molecules unambiguously. Bothprotonated and metallated ions of PCs were identified. Although[M+Na]+ species were generally predominant, occasionally [M+K]+

species were also observed. Sodiation of a PC was identified by the pres-ence of an ion at m/z 147 in the MS/MS spectra. Potassium addition wasidentified by the presence of an ion at m/z 163 in the MS/MS spectra.PCs were identified by screening LC–MS/MS profile for fragment ion atm/z 184.07. Lyso PCs and diacyl PCs fragmented in a similar way, exceptthat some lyso PCs also gave an additional ion atm/z 104.10. No structural

analysis reveals dysregulation of phosphatidylcholines in esophageal015.05.013

O

O O

O

OO

N+P

HO H -

A) PC (18:2/ 0:0)

O

O O

O

OO

N+P

HO H -

B) PC (20:4/ 0:0)

O O

OO

OO

N+P

O H -

C) PC (18:1/ 18:2)

O

3S.A. Mir et al. / Journal of Proteomics xxx (2015) xxx–xxx

information apart from the head group information could be obtainedfrom the MS/MS in positive mode analysis. Thus, negative mode analysiswas performed to identify the associated acyl chains in each lipid struc-ture. In the negative mode, PCs were detected in the form of their acetateadducts. Upon collision induced dissociation (CID), these adducts resultedin the neutral loss of 74 Da corresponding tomethyl acetate from the [M–H+CH3COO−] precursor ion [27]. Chain lengths were confirmed by thepresence of signature ions such as those with m/z 255, 253 and 281.Thus, neutral loss of 74 Da and fragment ions related to acyl chainswere the predominant features in the MS/MS spectra in negative mode.In addition to these, neutral loss of CO2 from the acyl chain [RCOO−]was also observed. Accurate mass-based metabolite database search re-sulted in large number of possible identifications. Hence, we adoptedthe approach of fragment mass filtration and manual inspection of indi-vidual MS/MS spectra with reproducible retention times for confidentidentification of metabolites.

D) PC (18:1/ 18:1)

O O

O

OO

N+P

O H -

O

O

Fig. 1. Chemical structures of four phosphatidylcholines that showed altered levels in ESCCpatient sera.

3.2. Phosphatidylocholines as a major class of dysregulated lipids in ESCC

Dysregulation of lipid metabolism has previously been described inthe context of various cancers. Out of 652 dysregulated molecular fea-tures, fragment mass scanning and manual inspection of data revealedthat 101 metabolites were phosphatidylcholines. Literature search wascarried out to find possible role of identified metabolites in cancer. Apie chart representation of metabolite class and their potential involve-ment in cancer is provided in Supplementary Fig. 2.By using targetedMS/MS in positive and negative mode, structures could be elucidatedfor seven metabolites (Table 1), chemical structures for four of whichare provided in Fig. 1. These seven metabolites were identified as PCswith different fatty acyl chain lengths. Out of the seven PCs, two weremonoacylated species. The levels of PC with m/z 544.34 were increasedand PCwithm/z 520.34were decreased in ESCC as compared to healthycontrols. Two peaks were observed in the extracted ion chromatogramof m/z 544.34 (Fig. 2) when only one was expected. We analyzed thesample in negativemode to characterize the second peak. LC chromato-gram in the negativemode revealed only a single peak corresponding topeak 1 in the positivemode. Peak 2may be due to a sodiated species of alower analog. This was confirmed by MS/MS analysis of peak 2 in posi-tive mode, revealing m/z 522.34 as an intense peak along with m/z544.34, thus confirming sodiation.MS/MS spectra of the acetate adductsof the two PCs, m/z 544. 34 andm/z 520.34 showed ion at m/z 303 cor-responding to arachidonoyl (20:4) chain and ion at m/z 279 corre-sponding to linoleoyl (18:2) chain, respectively in negative mode(Supplementary Fig. 3). Ion at m/z 544 has a longer chain (20:4) lengththan that of ion at m/z 520 (18:2). However, due to increased polarityby additional double bond, ion at m/z 544 eluted earlier.

Among the other five dysregulated metabolites, levels of two wereincreased and the levels of other three were decreased. Diacyl PCswith m/z 758.57 and 786.60 were increased, whereas, m/z 784.58 wasdecreased. Of the other twodecreasedmetabolites,m/z 774.56 has a hy-droxylated fatty acyl chain and m/z 770.60 has an ether-linked alkylchain. Both positive and negative mode analyses were used for thestructural elucidation of these molecules. A representative depictionfor the same is provided (Fig. 3).

Table 1List of dysregulated molecular features identified in ESCC.

S.No Observed mass Retention time (min) Fold change

1 520.3406 13.5 −11.52 544.341 13.4 11.43 758.5698 22.9 8.14 784.5849 23.1 −9.15 786.6005 23.4 4.96 770.6051 24.0 −9.47 774.5646 19.9 −8.7

Please cite this article as: S.A. Mir, et al., LC–MS-based serummetabolomicsquamous cell carcinoma, J Prot (2015), http://dx.doi.org/10.1016/j.jprot.2

A single diagnostic peak at m/z 279 was observed in the negativemode MS/MS spectra for the acetate adduct of ion at m/z 770.60. Thiscan be correlated to the presence of a C18:2 acyl chain. Signature ionsat m/z 255, 253 or 227 corresponding to the second acyl chain in thestructure were not present in theMS/MS spectra but a very low intensi-ty peak at m/z 265 was present (Fig. 4A). The presence of m/z 265 sug-gests two possibilities. The first one being that this species could be anether-linked alkenyl chain with one or more double bonds. But vinylicspecies are very stable and give dominant peaks, ruling out this possibil-ity. Second is that the observed species could be an ether-linked alkylchain with only one double bond. In this case, an ion at m/z 267would be expected but alkyl chains lose two hydrogen atoms to becomemore stable vinylic ions on collision induced dissociation [28]. Thiswould explain the low intense ion at m/z 265. Exact mass measure-ments with on-the-fly calibration confirmed the presence of an alkyl/acyl chain combination in the structure. Thus, the structure of ion atm/z 770.56 was deduced to be a PC (O-18:1/18:2).

MS/MS spectra of the acetate adduct of ion at m/z 774.56 in negativemode yielded three informative ions at m/z 255, 277 and 295 (Fig. 4B).Ion atm/z 255 can be correlated to the presence of a palmitoyl chain. Ionat m/z 295 is not a regular signature ion for any acyl chain, but could bedue to amodification. Thismodification can be attributed to a hydroxyl-ated C18:2 acyl chain. This could also be due to the presence of an oddnumbered fatty acyl chain, C19:1. The ambiguity was cleared with theexact mass measurement of 774.5646 in positive mode against a theo-retical mass of 774.5648 for PC with hydroxylated acyl chain.

Though isobaric phospholipids contain different acyl chains, thetotal number of carbons and the degree of unsaturation remains thesame, making the identification of these species challenging. MS/MS

(Log2) Structure Exact mass Formula

PC (18:2/0:0) 520.3403 C26H51NO7PPC (20:4/0:0) 544.3403 C28H51NO7PPC (16:0/18:2) 758.5699 C42H81NO8PPC (18:1/18:2) 784.5856 C44H83NO8PPC(18:1/18:1) 786.6012 C44H85NO8PPC(O-18:1/18:2) 770.6063 C44H85NO7PPC(16:0/h18:2) 774.5648 C42H81NO9P

analysis reveals dysregulation of phosphatidylcholines in esophageal015.05.013

RT (min)

0

20 40

50

100

Rel

ativ

e In

tens

ity

602.34

Peak 1

Peak 2

B)

C)

0

20 40

50

100

Rel

ativ

e In

tens

ity0

20 40

50

100

Rel

ativ

e In

tens

ity

A)

Fig. 2. Isobars of ion 544.34. (A) LC–MSprofile ofmetabolite extract from ESCC sample. (B) Extracted ion chromatogramof ion 544.34 in positivemode. (C) Extracted ion chromatogramofion 602.34 (acetate adduct of 544.34) in negative mode.

4 S.A. Mir et al. / Journal of Proteomics xxx (2015) xxx–xxx

spectra of closely eluting isobaric PCs were indistinguishable as ion atm/z 184.07 was the only dominant ion in the positive mode analysis.In contrast, negative mode analysis gave structural information thatallowed identification of isobaric masses. Acetate adduct ion 844.60

742.54

Inte

nsi

ty

m/z

279.23

816.57

255.23

480.31168.04

671.46

x10 5

0.0

0.4

0.8

1.2

1.6

2.0

900800700600500400300200100

255.23

279.23

BA)

Fig. 3. Structure of ion atm/z 758.57 with fragmentation from positive mode and negative modcorresponding protonated ion 758.57.

Please cite this article as: S.A. Mir, et al., LC–MS-based serummetabolomicsquamous cell carcinoma, J Prot (2015), http://dx.doi.org/10.1016/j.jprot.2

has more than three closely eluting isobaric masses. MS/MS spectra ofion at m/z 844.60 in negative mode revealed two clean MS/MS spec-tra for two isobaric metabolites. However, MS/MS spectra of thethird isomer had spectral peaks arising from unidentified isobars.

Inte

nsi

ty

m/z

184.07

x10 7

758.57

125.00

496.34

575.500.0

0.4

0.8

1.2

1.6

2.0

900800700600500400300200100

O

O

O O

O

OO

N+

P

OH -

184.07

)

e. (A) MS/MS spectra of acetate adduct ion 816.57 in negative mode. (B)MS/MS spectra of

analysis reveals dysregulation of phosphatidylcholines in esophageal015.05.013

PC (O-18:1 / 18:2)

O O

O

OO

N+P

O -

O

828.61

663.95

506.23

367.75

279.32

279.32

265.39

265.39

754.58

920.14119.05

1000900800700600500400300200100

m/z

1.6

2.0

1.2

0.8

0.4

0.0

x103

Inte

nsi

ty

A.

Inte

nsi

ty900800700600500400300200100

2.5

2.0

1.5

1.0

0.5

0.0

758.53

480.31

611.42

687.46

295.23 255.23

277.22

171.10

832.57

PC (16:0 / h18:2)

x105

B.

m/z

295.23

255.23

O

O

OOP

OO

OOH

-

O

N+

Fig. 4.MS/MS spectra of acetate adducts in negative mode. (A) MS/MS spectra of ion 828.61, acetate adduct of ion 770.60. (B) MS/MS spectra of ion 832.56, acetate adduct of ion 774.56.

5S.A. Mir et al. / Journal of Proteomics xxx (2015) xxx–xxx

Along with PC 18:2/18:0 (Fig. 5A) and PC 18:1/18:1 (Fig. 5B), PC20:2/16:0 (Fig. 5C) was also observed along with signature ionsfrom other isobaric PC species. Overall, our data demonstrates thatPC levels are significantly altered in ESCC cases as compared to thehealthy control subjects suggesting a dysregulation of cholinemetabolism.

4. Discussion

Lipids are pivotal for membrane structure and signal transduction tosupport proliferation of cancer cells. There is also accumulating evi-dence of altered lipid metabolism, particularly de novo lipogenesis incancer [29–32]. Several cancer metabolomic studies have identifiedthe dysregulation of choline and its metabolites with potential implica-tions in cancer prognosis [33–35]. Phosphocholine levels have been re-ported to be increased in ovarian and breast cancer [36,37]. Increasedcholine kinase activity has been previously attributed to the increasedphosphorylation of cholines in colon cancer [38]. It has been reportedthat most of the endogenous phosphocholine moieties are used forthe synthesis of PCs [39]. In addition to choline metabolism, fatty acidmetabolism also plays an important role in lipid synthesis. Increased ex-pression of several enzymes involved in the metabolism of lipids suchas fatty acid synthase (FASN), acetyl-CoA carboxylase and ATP citratelyase (ACL) have been reported in cancers [31,40]. Differential levelsof PCs have been reported in other serum metabolomic studies onESCC patients apart from other metabolites [15,41]. The altered levelsof PCs observed in our study indicates underlying aberration of cholineand phosphocholine metabolism in ESCC. This is in agreement with ob-servations reported by previous studies. The significantly dysregulatedmetabolites found in this study are associated with functions such ascell signaling, energy storage, maintenance of membrane integrity andstability according to human metabolome database (HMDB) functionalclassification.Modulation of signalingpathways are associatedwith car-cinogenesis and PCs play an important role in membrane mediated sig-naling. Dysregulation of PCs observed in this study suggests aberrant

Please cite this article as: S.A. Mir, et al., LC–MS-based serummetabolomicsquamous cell carcinoma, J Prot (2015), http://dx.doi.org/10.1016/j.jprot.2

signaling in ESCC. Although altered levels of several PCs were observedin earlier studies also, structural elucidation remains challenging due tothe presence of many isobaric and isomeric lipid species. We adopted atandemmass spectrometry approach in positive and negative modes ofionization to confirm the structure of seven PCs. Positive mode MS/MSgenerated headgroup information and negative mode MS/MS wascarried out to identify associated acyl chains. Thus, the analyticalstrategy applied here can be used for confident identification ofphosohatidylcholines and other lipid species. This approach is crucialin untargeted metabolomics studies where unambiguous identificationis required for hundreds of metabolites.

Human serum contains a wide variety of lipids including phos-phatidic acids, phosphatidylinositols, phosphatidylserines andphosphatidylglycerols along with PCs and PEs. Nitrogen containingPCs dominate the lipid profile because of their higher ionization effi-ciency. When isobaric masses from other lipid classes co-elute withPCs, the MS/MS spectra are dominated by the ion of m/z 184.07and in most cases, no spectral contribution from the co-eluting spe-cies is observed. In addition to head groups, hydrophobic chains inthe structure also contribute to ionization efficiency. Ester groupsin the acyl chains can be easily protonated as compared to theether linked alkyl or alkenyl chains. Hence, in a lipid mixture,ether lipids are subjected to ion suppression and acylated lipidsand PCs can be detected more readily compared to other classesof lipids. Therefore, separate methods need to be developed forthe identification of the ether lipids from a mixture. Plasmalogensare an important class of lipids that protect cells from free radicalattack and oxidative damage [42]. Alkyl/acyl phosphocholinelipids, such as the metabolite with m/z of 770.60 in the currentstudy, serve as a precursor to plasmalogens. Their downregulationmay lead to decreased levels of plasmalogen moieties. Overall, bycombing untargeted LC–MS approach with targeted MS/MS of sig-nificant molecular features, we demonstrate the differences inserum metabolome of ESCC patients as compared to healthysubjects.

analysis reveals dysregulation of phosphatidylcholines in esophageal015.05.013

C)

770.56

844.60255.23307.25

481.31589.80

900800700600500400300200100

5.0

4.0

3.0

2.0

1.0

0.0

x103

Inte

nsi

ty

m/z

279.23

307.25

O

O

OO

N+

P-O

O

255.23

O

O

PC (20:2/16:0)

900800700600500400300200100

1.5

1.0

0.0

0.5

770.57

844.60

279.23

168.04 508.34 725.52

A)

x105

283.26

Inte

nsi

ty

PC (18:2/18:0)

m/z

283.26

279.23

O

O

O

OO

N+

P-

O

O

O5.0

4.0

3.0

2.0

1.0

0.0

770.56

844.60281.25

383.28392.10

794.57

B)

x103

m/z900800700600500400300200100

Inte

nsi

ty

281.25

O

O

OO

N+

P-O

O

281.25

O

O

PC (18:1/18:1)

Fig. 5.MS/MS spectra of isobaric masses of ion 844 (acetate adduct of 786.60) in negative mode. (A)MS/MS spectra of ion 844.60 revealing PC18:2/18:0. (B) MS/MS spectra of ion 844.60revealing PC 18:1/18:1 (C) MS/MS spectra of ion 844.60 revealing PC 16:0/20:2.

6 S.A. Mir et al. / Journal of Proteomics xxx (2015) xxx–xxx

5. Conclusions

In the current study, we identified 652 significantly dysregulatedmolecular features from ESCC serummetabolome using liquid chroma-tography time-of-flight mass spectrometry approach. Our results pro-vide novel insights into the dysregulation of phosphatidylcholines andassociated lipid metabolism in ESCC as compared to healthy subjects.Large scale validation of these dysregulated metabolites might proveuseful for identification of blood based biomarkers of ESCC.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.jprot.2015.05.013.

Conflict of interest

The authors declare that they have no conflicts of interest.

Acknowledgments

We thank the Department of Biotechnology (DBT), Government ofIndia for research support to the Institute of Bioinformatics, Bangalore(BT/01/COE/08/05).We thank Agilent Technologies for access to instru-mentation and software. We thank the “Infosys Foundation” for re-search support to the Institute of Bioinformatics. SAM and KKD arerecipients of Senior Research Fellowship awards fromUniversity GrantsCommission (UGC), Government of India. PR is a recipient of Senior Re-search Fellowship from Council of Scientific and Industrial Research(CSIR), Government of India. AAK is a recipient of Senior Research Fel-lowship from Indian Council of Medical Research (ICMR), Government

Please cite this article as: S.A. Mir, et al., LC–MS-based serummetabolomicsquamous cell carcinoma, J Prot (2015), http://dx.doi.org/10.1016/j.jprot.2

of India. Dr. Harsha Gowda is a Wellcome Trust/DBT India AllianceEarly Career Fellow.

References

[1] D. Hanahan, R.A. Weinberg, Hallmarks of cancer: the next generation, Cell 144(2011) 646–674.

[2] M. Jain, R. Nilsson, S. Sharma, N. Madhusudhan, T. Kitami, A.L. Souza, et al., Metabo-lite profiling identifies a key role for glycine in rapid cancer cell proliferation, Science336 (2012) 1040–1044.

[3] J. Budczies, S.F. Brockmoller, B.M. Muller, D.K. Barupal, C. Richter-Ehrenstein, A.Kleine-Tebbe, et al., Comparative metabolomics of estrogen receptor positive andestrogen receptor negative breast cancer: alterations in glutamine and beta-alanine metabolism, J. Proteome 94C (2013) 279–288.

[4] Y. Qiu, G. Cai, M. Su, T. Chen, X. Zheng, Y. Xu, et al., Serum metabolite profiling ofhuman colorectal cancer using GC–TOFMS and UPLC–QTOFMS, J. Proteome Res. 8(2009) 4844–4850.

[5] S. Urayama, W. Zou, K. Brooks, V. Tolstikov, Comprehensive mass spectrometrybased metabolic profiling of blood plasma reveals potent discriminatory classifiersof pancreatic cancer, Rapid Commun. Mass Spectrom. 24 (2010) 613–620.

[6] J.F. Xiao, R.S. Varghese, B. Zhou, M.R. Nezami Ranjbar, Y. Zhao, T.H. Tsai, et al., LC–MSbased serum metabolomics for identification of hepatocellular carcinoma bio-markers in Egyptian cohort, J. Proteome Res. 11 (2012) 5914–5923.

[7] B.Wang, D. Chen, Y. Chen, Z. Hu, M. Cao, Q. Xie, et al., Metabonomic profiles discrim-inate hepatocellular carcinoma from liver cirrhosis by ultraperformance liquid chro-matography–mass spectrometry, J. Proteome Res. 11 (2012) 1217–1227.

[8] T. Chen, G. Xie, X. Wang, J. Fan, Y. Qiu, X. Zheng, et al., Serum and urine metaboliteprofiling reveals potential biomarkers of human hepatocellular carcinoma, Mol. Cell.Proteomics 10 (2011) (M110 004945).

[9] T. Zhang, X. Wu, C. Ke, M. Yin, Z. Li, L. Fan, et al., Identification of potential bio-markers for ovarian cancer by urinary metabolomic profiling, J. Proteome Res. 12(2013) 505–512.

[10] L. Lin, Z. Huang, Y. Gao, X. Yan, J. Xing, W. Hang, LC–MS based serum metabonomicanalysis for renal cell carcinoma diagnosis, staging, and biomarker discovery, J. Pro-teome Res. 10 (2011) 1396–1405.

analysis reveals dysregulation of phosphatidylcholines in esophageal015.05.013

7S.A. Mir et al. / Journal of Proteomics xxx (2015) xxx–xxx

[11] A. Sreekumar, L.M. Poisson, T.M. Rajendiran, A.P. Khan, Q. Cao, J. Yu, et al.,Metabolomic profiles delineate potential role for sarcosine in prostate cancer pro-gression, Nature 457 (2009) 910–914.

[12] T. Kind, V. Tolstikov, O. Fiehn, R.H.Weiss, A comprehensive urinarymetabolomic ap-proach for identifying kidney cancer, Anal. Biochem. 363 (2007) 185–195.

[13] K. Kim, S.L. Taylor, S. Ganti, L. Guo, M.V. Osier, R.H. Weiss, Urine metabolomic anal-ysis identifies potential biomarkers and pathogenic pathways in kidney cancer,OMICS 15 (2011) 293–303.

[14] S. Ganti, S.L. Taylor, K. Kim, C.L. Hoppel, L. Guo, J. Yang, et al., Urinary acylcarnitinesare altered in human kidney cancer, Int. J. Cancer 130 (2012) 2791–2800.

[15] J. Xu, Y. Chen, R. Zhang, Y. Song, J. Cao, N. Bi, et al., Global and targetedmetabolomicsof esophageal squamous cell carcinoma discovers potential diagnostic and thera-peutic biomarkers, Mol. Cell. Proteomics 12 (2013) 1306–1318.

[16] A. Zhang, H. Sun, P. Wang, Y. Han, X. Wang, Modern analytical techniques in meta-bolomics analysis, Analyst 137 (2012) 293–300.

[17] H.G. Gika, G.A. Theodoridis, R.S. Plumb, I.D. Wilson, Current practice of liquid chro-matography–mass spectrometry in metabolomics and metabonomics, J. Pharm.Biomed. Anal. 87 (2014) 12–25.

[18] T.O. Metz, Q. Zhang, J.S. Page, Y. Shen, S.J. Callister, J.M. Jacobs, et al., The future of liq-uid chromatography–mass spectrometry (LC–MS) in metabolic profiling andmetabolomic studies for biomarker discovery, Biomark. Med 1 (2007) 159–185.

[19] L.Wang, J. Chen, L. Chen, P. Deng, Q. Bu, P. Xiang, et al., 1H-NMR basedmetabonomicprofiling of human esophageal cancer tissue, Mol. Cancer 12 (2013) 25.

[20] X. Zhang, L. Xu, J. Shen, B. Cao, T. Cheng, T. Zhao, et al., Metabolic signatures ofesophageal cancer: NMR-based metabolomics and UHPLC-based focusedmetabolo-mics of blood serum, Biochim. Biophys. Acta 2013 (1832) 1207–1216.

[21] H. Wu, R. Xue, C. Lu, C. Deng, T. Liu, H. Zeng, et al., Metabolomic study for diagnosticmodel of oesophageal cancer using gas chromatography/mass spectrometry, J.Chromatogr. B Anal. Technol. Biomed. Life Sci. 877 (2009) 3111–3117.

[22] A. Ikeda, S. Nishiumi, M. Shinohara, T. Yoshie, N. Hatano, T. Okuno, et al., Serummetabolomics as a novel diagnostic approach for gastrointestinal cancer, Biomed.Chromatogr. 26 (2012) 548–558.

[23] T.R. Sana, D.B. Gordon, S.M. Fischer, S.E. Tichy, N. Kitagawa, C. Lai, et al., Global massspectrometry based metabolomics profiling of erythrocytes infected with Plasmodi-um falciparum, PLoS One 8 (2013) e60840.

[24] D.S. Wishart, T. Jewison, A.C. Guo, M. Wilson, C. Knox, Y. Liu, et al., HMDB 3.0—thehuman metabolome database in 2013, Nucleic Acids Res. 41 (2013) D801–D807.

[25] E. Fahy, S. Subramaniam, R.C. Murphy, M. Nishijima, C.R. Raetz, T. Shimizu, et al., Up-date of the LIPID MAPS comprehensive classification system for lipids, J. Lipid Res.50 (Suppl.) (2009) S9–S14.

[26] R. Tautenhahn, K. Cho, W. Uritboonthai, Z. Zhu, G.J. Patti, G. Siuzdak, An acceleratedworkflow for untargeted metabolomics using the METLIN database, Nat. Biotechnol.30 (2012) 826–828.

Please cite this article as: S.A. Mir, et al., LC–MS-based serummetabolomicsquamous cell carcinoma, J Prot (2015), http://dx.doi.org/10.1016/j.jprot.2

[27] D. Pacetti, E. Boselli, H.W. Hulan, N.G. Frega, High performance liquid chromatogra-phy–tandem mass spectrometry of phospholipid molecular species in eggs fromhens fed diets enriched in seal blubber oil, J. Chromatogr. A 1097 (2005) 66–73.

[28] F.F. Hsu, J. Turk, A.K. Thukkani, M.C. Messner, K.R. Wildsmith, D.A. Ford, Characteri-zation of alkylacyl, alk-1-enylacyl and lyso subclasses of glycerophosphocholine bytandem quadrupole mass spectrometry with electrospray ionization, J. MassSpectrom. 38 (2003) 752–763.

[29] F.P. Kuhajda, K. Jenner, F.D. Wood, R.A. Hennigar, L.B. Jacobs, J.D. Dick, et al., Fattyacid synthesis: a potential selective target for antineoplastic therapy, Proc. Natl.Acad. Sci. U. S. A. 91 (1994) 6379–6383.

[30] J.V. Swinnen, K. Brusselmans, G. Verhoeven, Increased lipogenesis in cancer cells:new players, novel targets, Curr. Opin. Clin. Nutr. Metab. Care 9 (2006) 358–365.

[31] J.A. Menendez, R. Lupu, Fatty acid synthase and the lipogenic phenotype in cancerpathogenesis, Nat. Rev. Cancer 7 (2007) 763–777.

[32] F. Zhang, G. Du, Dysregulated lipid metabolism in cancer, World J. Biol. Chem. 3(2012) 167–174.

[33] E. Rysman, K. Brusselmans, K. Scheys, L. Timmermans, R. Derua, S. Munck, et al., Denovo lipogenesis protects cancer cells from free radicals and chemotherapeutics bypromoting membrane lipid saturation, Cancer Res. 70 (2010) 8117–8126.

[34] E. Iorio, A. Ricci, M. Bagnoli, M.E. Pisanu, G. Castellano, M. Di Vito, et al., Activation ofphosphatidylcholine cycle enzymes in human epithelial ovarian cancer cells, CancerRes. 70 (2010) 2126–2135.

[35] A. Ramirez de Molina, R. Gutierrez, M.A. Ramos, J.M. Silva, J. Silva, F. Bonilla, et al.,Increased choline kinase activity in human breast carcinomas: clinical evidence fora potential novel antitumor strategy, Oncogene 21 (2002) 4317–4322.

[36] G. Eliyahu, T. Kreizman, H. Degani, Phosphocholine as a biomarker of breast cancer:molecular and biochemical studies, Int. J. Cancer 120 (2007) 1721–1730.

[37] E. Iorio, D. Mezzanzanica, P. Alberti, F. Spadaro, C. Ramoni, S. D'Ascenzo, et al., Alter-ations of choline phospholipid metabolism in ovarian tumor progression, CancerRes. 65 (2005) 9369–9376.

[38] K. Nakagami, T. Uchida, S. Ohwada, Y. Koibuchi, Y. Suda, T. Sekine, et al., Increasedcholine kinase activity and elevated phosphocholine levels in human colon cancer,Jpn. J. Cancer Res. 90 (1999) 419–424.

[39] R. Katz-Brull, D. Seger, D. Rivenson-Segal, E. Rushkin, H. Degani, Metabolomicmarkers of breast cancer: enhanced choline metabolism and reduced choline-ether-phospholipid synthesis, Cancer Res. 62 (2002) 1966–1970.

[40] D.A. Tennant, R.V. Duran, E. Gottlieb, Targeting metabolomic transformation for can-cer therapy, Nat. Rev. Cancer 10 (2010) 267–277.

[41] R. Liu, Y. Peng, X. Li, Y. Wang, E. Pan, W. Guo, et al., Identification of plasmametabolomic profiling for diagnosis of esophageal squamous-cell carcinoma usingan UPLC/TOF/MS platform, Int. J. Mol. Sci. 14 (2013) 8899–8911.

[42] P. Brites, H.R. Waterham, R.J. Wanders, Functions and biosynthesis of plasmalogensin health and disease, Biochim. Biophys. Acta 1636 (2004) 219–231.

analysis reveals dysregulation of phosphatidylcholines in esophageal015.05.013