journal of chromatography b - cuscholar.cu.edu.eg/sites/default/files/ahmed_kamal_2.pdfjournal of...

11
Journal of Chromatography B, 1009 (2016) 44–54 Contents lists available at ScienceDirect Journal of Chromatography B jou rn al hom epage: www.elsevier.com/locate/chromb Phospholipidomic identification of potential serum biomarkers in dengue fever, hepatitis B and hepatitis C using liquid chromatography-electrospray ionization-tandem mass spectrometry Alaa Khedr a,, Maha A. Hegazy b , Ahmed K. Kammoun a , Mostafa A. Shehata b a Department of Pharmaceutical Chemistry, Faculty of Pharmacy, King Abdulaziz University, P.O. Box 80260, Jeddah 21589, Saudi Arabia b Department of Analytical Chemistry, Faculty of Pharmacy, Cairo University, Cairo, Egypt a r t i c l e i n f o Article history: Received 2 October 2015 Received in revised form 2 December 2015 Accepted 7 December 2015 Available online 9 December 2015 Keywords: Phospholipids LC-ESI-MS/MS Differentiating biomarkers Dengue fever Hepatitis B Hepatitis C a b s t r a c t The serum phospholipid (PL) profiles of healthy volunteers (HE) and patients with recently diag- nosed dengue fever (DF), hepatitis B (HBV), and hepatitis C (HCV) were investigated using liquid chromatography-ion trap-mass spectrometry (LC-IT-MS) and liquid chromatography-triple quad-mass spectrometry (LC-TQ-MS). Major PLs, including lyso-phosphatidylcholins (LPCs), phosphatidylcholins (PCs), phosphatidylinositols (PIs), phosphatidylethanolamines (PEs) and phosphatidylserines (PSs), were characterized in human serum using LC-IT-MS. Thirty-five PLs were quantified using seven non- endogenous odd-carbon PL standards. An MS search protocol for the identification of PLs is described. The analytical method was optimized to achieve maximum recovery and detection. PLs were detected with minimal ionization suppression. The PLs species were characterized on the basis of (i) MS 2 peaks due to polar head, (ii) precursor ion or neutral loss scans, (iii) identification of fatty acid, (iv) identifica- tion of sn-1 and sn-2 fatty acid. The quantitation data were subjected to principal component analysis (PCA), and a significant difference was observed between the PL profiles of the investigated diseases and those of HE subjects. The significance of the changes in each lipid among the four groups was statis- tically assessed using one-way analysis of variance (ANOVA) followed by Bonferroni post hoc multiple comparison. The serum profiles of 28 PLs were determined to be significantly different and enabled the discrimination between HE individuals and the studied patients. Potentially dysregulated PLs were considered as differentiating biomarkers to diagnose DF, HBV, and HCV. © 2015 Elsevier B.V. All rights reserved. 1. Introduction Lipid analysis has attracted increasing interest because of the importance of lipids in medical, biological, industrial, and biotech- nological applications [1–3]. PLs play an important role in the biochemistry of all living cells. These hydrophobic molecules are the building blocks of the cellular membrane. Phospholipids per- form important biological functions, such as housing the proteins involved in cell signaling, intracellular adhesion, and cytoskele- ton support, and also serve as a precursor pool for biologically Abbreviations: DF, dengue fever; HBV, hepatitis B virus; HCV, hepatitis C virus; IT-MS, oon trap mass spectrometry; LC, liquid chromatography; LOD, limit of detec- tion; LOQ, limit of quantification; PLs, phospholipids; TQ-MS, triple quad mass spectrometry. Corresponding author. E-mail address: [email protected] (A. Khedr). active lipid mediators [4,5]. Plasma and/or serum profiling of PLs and lysophospholipids (LPLs) have been reported because these molecules are potentially clinically important biomarkers of certain human diseases, including breast cancer [6], diabetic nephropathy [7], osteoarthritis [8] and cardiovascular diseases [9]. Viral infec- tious diseases, including dengue fever (DF), HBV and HCV, infect millions of people worldwide. DF is an infectious tropical disease caused by the dengue virus, and in a small portion of cases, the dis- ease develops into the life-threatening dengue hemorrhagic fever [10]. The clinical diagnosis of DF is typically based on reported symptoms and physical examination [11], especially in endemic areas. However, in the earliest stages of DF, the disease cannot be differentiated from other viral infections [12]. Physiologically, a strong correlation between DF and lipids has been demonstrated [13,14]. van Gorp et al. assessed the plasma lipid profiles of DF patients and concluded that the most severe cases of DF were asso- ciated with significantly lower levels of cholesterol, high-density http://dx.doi.org/10.1016/j.jchromb.2015.12.011 1570-0232/© 2015 Elsevier B.V. All rights reserved.

Upload: lamcong

Post on 12-May-2018

214 views

Category:

Documents


1 download

TRANSCRIPT

Pdc

Aa

b

a

ARRAA

KPLDDHH

1

inbtfit

Its

h1

Journal of Chromatography B, 1009 (2016) 44–54

Contents lists available at ScienceDirect

Journal of Chromatography B

jou rn al hom epage: www.elsev ier .com/ locate /chromb

hospholipidomic identification of potential serum biomarkers inengue fever, hepatitis B and hepatitis C using liquidhromatography-electrospray ionization-tandem mass spectrometry

laa Khedr a,∗, Maha A. Hegazy b, Ahmed K. Kammoun a, Mostafa A. Shehata b

Department of Pharmaceutical Chemistry, Faculty of Pharmacy, King Abdulaziz University, P.O. Box 80260, Jeddah 21589, Saudi ArabiaDepartment of Analytical Chemistry, Faculty of Pharmacy, Cairo University, Cairo, Egypt

r t i c l e i n f o

rticle history:eceived 2 October 2015eceived in revised form 2 December 2015ccepted 7 December 2015vailable online 9 December 2015

eywords:hospholipidsC-ESI-MS/MSifferentiating biomarkersengue feverepatitis Bepatitis C

a b s t r a c t

The serum phospholipid (PL) profiles of healthy volunteers (HE) and patients with recently diag-nosed dengue fever (DF), hepatitis B (HBV), and hepatitis C (HCV) were investigated using liquidchromatography-ion trap-mass spectrometry (LC-IT-MS) and liquid chromatography-triple quad-massspectrometry (LC-TQ-MS). Major PLs, including lyso-phosphatidylcholins (LPCs), phosphatidylcholins(PCs), phosphatidylinositols (PIs), phosphatidylethanolamines (PEs) and phosphatidylserines (PSs), werecharacterized in human serum using LC-IT-MS. Thirty-five PLs were quantified using seven non-endogenous odd-carbon PL standards. An MS search protocol for the identification of PLs is described.The analytical method was optimized to achieve maximum recovery and detection. PLs were detectedwith minimal ionization suppression. The PLs species were characterized on the basis of (i) MS2 peaksdue to polar head, (ii) precursor ion or neutral loss scans, (iii) identification of fatty acid, (iv) identifica-tion of sn-1 and sn-2 fatty acid. The quantitation data were subjected to principal component analysis(PCA), and a significant difference was observed between the PL profiles of the investigated diseases and

those of HE subjects. The significance of the changes in each lipid among the four groups was statis-tically assessed using one-way analysis of variance (ANOVA) followed by Bonferroni post hoc multiplecomparison. The serum profiles of 28 PLs were determined to be significantly different and enabledthe discrimination between HE individuals and the studied patients. Potentially dysregulated PLs wereconsidered as differentiating biomarkers to diagnose DF, HBV, and HCV.

© 2015 Elsevier B.V. All rights reserved.

. Introduction

Lipid analysis has attracted increasing interest because of themportance of lipids in medical, biological, industrial, and biotech-ological applications [1–3]. PLs play an important role in theiochemistry of all living cells. These hydrophobic molecules arehe building blocks of the cellular membrane. Phospholipids per-

orm important biological functions, such as housing the proteinsnvolved in cell signaling, intracellular adhesion, and cytoskele-on support, and also serve as a precursor pool for biologically

Abbreviations: DF, dengue fever; HBV, hepatitis B virus; HCV, hepatitis C virus;T-MS, oon trap mass spectrometry; LC, liquid chromatography; LOD, limit of detec-ion; LOQ, limit of quantification; PLs, phospholipids; TQ-MS, triple quad masspectrometry.∗ Corresponding author.

E-mail address: [email protected] (A. Khedr).

ttp://dx.doi.org/10.1016/j.jchromb.2015.12.011570-0232/© 2015 Elsevier B.V. All rights reserved.

active lipid mediators [4,5]. Plasma and/or serum profiling of PLsand lysophospholipids (LPLs) have been reported because thesemolecules are potentially clinically important biomarkers of certainhuman diseases, including breast cancer [6], diabetic nephropathy[7], osteoarthritis [8] and cardiovascular diseases [9]. Viral infec-tious diseases, including dengue fever (DF), HBV and HCV, infectmillions of people worldwide. DF is an infectious tropical diseasecaused by the dengue virus, and in a small portion of cases, the dis-ease develops into the life-threatening dengue hemorrhagic fever[10]. The clinical diagnosis of DF is typically based on reportedsymptoms and physical examination [11], especially in endemicareas. However, in the earliest stages of DF, the disease cannotbe differentiated from other viral infections [12]. Physiologically,a strong correlation between DF and lipids has been demonstrated

[13,14]. van Gorp et al. assessed the plasma lipid profiles of DFpatients and concluded that the most severe cases of DF were asso-ciated with significantly lower levels of cholesterol, high-density

atogr

lalcrcdeHma[cP

PlPpoatr(cltstcsT(EbucuctsqstudfcePcmf

qadiTcdad

applied using two different mobile systems. For quantitative anal-ysis, the TQ–MS system was controlled by MassHunter software.

A. Khedr et al. / J. Chrom

ipoprotein, and low-density lipoprotein compared with mild casesnd healthy controls [14]. HBV and HCV are the leading causes ofiver diseases worldwide, and approximately 50–80% of infectedarriers become chronically infected, which may lead to liver cir-hosis and hepatocellular carcinoma [15,16]. Both viral infectionsause dramatic changes within infected hepatocytes, including theisruption of different aspects of lipid metabolism [17,18]. Chent al. performed serum lipid profiling of patients with chronicBV by ultra-fast liquid chromatography-ion trap-time of flight-ass spectrometry (LC-IT-TOF MS) and concluded that lipids were

bnormally regulated in patients compared with healthy controls17]. Additionally, the lipidomic profiling of HCV infected hepato-ytes was studied by Roe et al. and Qu et al., who observed alteredL metabolism in infected cells [18,19].

Several methods have been reported for the analysis of differentL classes. Thin-layer chromatography (TLC) and high-performance

iquid chromatography (HPLC) have been used to separate differentL classes [20–23]. Normal-phase (NP) HPLC with different mobilehases has been described by many researchers for the separationf PL classes based on head group polarity [24,25]. However, thepplicability of NP-HPLC is limited because the presence of water inhe solvent system can alter the affinity of PLs for silica in replicateuns, resulting in fluctuating retention times [26]. Reversed-phaseRP) HPLC can also be used to separate mixture of PLs [27,28]. Inontrast to NP, RP-HPLC allows the characteristic cataloguing ofipids according to their overall polarity and the fatty acid composi-ion in the sn-1 and sn-2 locations of the glycerol skeleton [8]. Masspectrometry (MS), and electro spray ionization- mass spectrome-ry (ESI-MS) in particular, is a very useful tool for the analysis of PLompositions, and this technique has strong identification ability,pecificity and sensitivity compared to other methods [24,29,30].he quantitative analysis of lipids by MS requires internal standardsIS) to control for the variable recovery from biological matrixes,SI sensitivity and other factors that can affect the ion yield. Sta-le isotope-labeled versions of each phospholipid are commerciallynavailable, and relying on such standards is impractical due to theomplexity of biological matrixes. Therefore, an alternative is tose an IS that has similar structural, ionization and fragmentationharacteristics to the categories of compounds under investiga-ion. Synthetic non-endogenous PLs are commonly used as internaltandards and for analyses of the extraction rate [17]. Shaner et al.uantitatively analyzed sphingolipids using triple quadrupole andingle quadrupole linear ion trap mass spectrometry and a cock-ail of internal standards containing a sphengoid base with anncommon chain length (C17) [31]. Accordingly, Castro-Perez et al.eveloped a validated method using five non-endogenous lipids

or the quantitative analysis of lipids (in terms of relative con-entration) in both positive and negative ESI modes [8]. Differentxtraction methods have been reported for the optimal recovery ofLs from different biological matrixes and reported varying per-entage recoveries [32,33]. Different ratios of dichloromethane,ethanol and chloroform have been used as the solvents of choice

or the optimal extraction recovery of PLs from plasma or serum.In this study, we describe a method for the characterization and

uantification of major PLs in human serum using LC-ESI-MS/MSnd a non-endogenous phospholipid mixture as an internal stan-ard. The quantification of major PLs and profiling of three viral

nfectious diseases, including DF, HBV and HCV, were investigated.he obtained data were statistically analyzed to confirm the signifi-ant and nonsignificant PL changes in 14 patients infected with eachisease compared with 14 healthy volunteers. The statistical valuesnd plots clearly indicate which PLs can be used as biomarkers to

ifferentiate between the three selected viral infections.

. B 1009 (2016) 44–54 45

2. Experimental

2.1. Chemicals and materials

Synthetic lipid standards were purchased from Avanti PolarLipids, Inc. (Alabaster, AL, USA), and the purity of eachwas >99.0%: 17:0-lysophosphatidylcholine (LPC), 17:0/17:0-phosphatidylocholine (PC), 15:0/15:0- phosphatidylethanolamine(PE), 17:0/17:0-PE, 17:0/17:0- phosphatidylserine (PS), 17:1-PSsodium salt, and 17:1-lysophosphatidylinositol (LPI) sodium salt.Distilled water was filtered through a Millipore Milli-Q system (Bil-lerica, MA, USA). All other materials were of analytical grade. Allsolvents were of HPLC grade and were used without further purifi-cation (Merck, Darmstadt, Germany). Stock standard solutionswere prepared in MeOH:CH2Cl2 (2:1, v/v) as follows: 0.2 mg/mLof LPI(17:0) and 1 mg/mL each of PC(17:0/17:0), PE(17:0/17:0)and PS(17:0/17:0). The internal standard mixture was prepared inMeOH:CH2Cl2, (2:1, v/v) and contained 1 mg/mL each of LPC(17:0),PE(15:0/15:0) and LPS(17:1). LPC(17:0) and LPS(17:1) were usedas internal standards for the determination of biogenic PC and PSspecies, respectively, while PE(15:0/15:0) was used as the IS forthe determination of biogenic PE and PI species. Six concentrationlevels were prepared using acetonitrile: methanol (1:1, v/v) as thesolvent. The concentration ranges used for calibration were as fol-lows: 10–60 ng/�L of PC (17:0/17:0) and PE(17:0/17:0), 5–30 ng/�Lof PS(17:0/17:0) and 20–120 ng/�L of LPI (17:0). The final concen-trations of ISs in the injected solutions were 40 ng/mL LPC(17:0)and PE(15:0/15:0) and 20 ng/mL LPS(17:0). The injection volumewas 5 �L. All of the standard solutions were stored in a freezer at−80 ◦C. Linear regression curves were generated by calculating thepeak area ratio of each PL and its corresponding IS and plotting thatvalue against the concentration (ng/�L). Serum samples were col-lected from 14 healthy subjects and 14 patients in each group: DF(early febrile phase), diagnosed HBV and diagnosed HCV. Male andfemale subjects aged between 20 and 45 years old were included.Samples were portioned into 4-mL sterile vacuumed blood collec-tion tubes. The serum samples were either analyzed immediatelyor kept at −80 ◦C until analysis. Volunteers diagnosed as hyper-lipidemic or who had been administered any drug were excluded.Serum samples from both healthy and infected individuals wereobtained with the help of King Abdulaziz University Hospital (Jed-dah, Saudi Arabia) with informed consent.

2.2. LC-ESI-MS–MS analysis

Two ESI-MS-MS systems from Agilent (Agilent Technology,Germany) were utilized. An Agilent 6320 Ion Trap mass spec-trometer (IT-MS) and Agilent 6420 Triple Quad mass spectrometer(TQ-MS) were used for the characterization and quantification,respectively, of PLs. Each MS system was connected to an HPLC-Agilent 1200 system equipped with an autosampler, a quaternarypump, and a column compartment (Palo Alto, CA, USA). Both sys-tems were equipped with ChemStation software (Rev. B.01.03SR2(204)). The IT–MS was controlled using 6300 Series Trap Con-trol version 6.2 Build No. 62.24 (Bruker Daltonik GmbH), and thegeneral MS adjustments were set as follows: capillary voltage,3500 V; nebulizer, 35 psi; drying gas,12 L/min; desolvation temper-ature, 350 ◦C; ion charge control (ICC) smart target, 150,000; andmax accumulation time, 150 millisecond (ms). The MS scan rangewas 50–1100 m/z. Auto-MSn positive and negative polarities were

The TQ–MS conditions utilized were a gas temperature of 330 ◦C,a gas flow of11 L/min, a nebulizer pressure of 40 psi, a capillary

46 A. Khedr et al. / J. Chromatogr. B 1009 (2016) 44–54

entifi

vtambUrf

PAsvgie0

maaottr

LPI(17:1)—into pooled healthy human serum .The concentra-tion ranges of each non-endogenous lipids were the same as thoseused in the calibration solutions. The lipids were extracted from

Fig. 1. Schematic search and id

oltage of 4000 V, a fragmenter voltage 135 V, and a cell accelera-or voltage of 7 V. The MS scan range was 450–1100 m/z. Positivend negative auto-scan modes were collected using two differentobile systems. The separation was performed on an Agilent Zor-

ax Eclipse Plus-C18 column (4.6 × 150 mm, 5 �m, Palo Alto, CA,SA) maintained at 45 ◦C. The screw-capped (PTFE/silicon) total

ecovery auto-sampler vials (1 mL, 12 × 32 mm) were purchasedrom Waters (Milford, MA, USA).

Negative auto-scan mode was applied to monitor the PIs, PEs,Ss, and corresponding LPLs. The mobile phase consisted of line: water and acetonitrile (90:10, v/v) containing 0.1% ammoniaolution (w/v), and line B: propane-2-ol and acetonitrile (80/20,/v). The pump was programmed to deliver 50% B for 2 min,radually increase to 60% B (40% A) from 2 to 15 min, furtherncrease to 100% B from 15 to 25 min, and then perform isocraticlution of 100% B for an additional 15 min. The flow rate was.5 mL/min.

Positive auto-scan mode was used to analyze PCs and LPCs. Theobile phase consisted of one liter of double-distilled water and

cetonitrile in a ratio 90:10 (v/v) mixed with 1 mL of 100% formiccid (line A) and one liter of solvent mix containing propane-2-land acetonitrile (80:20, v/v) (line B). The pump was programmedo deliver 50% B for 2 min, adjust to 100% B from 2 to 70 min, andhen perform isocratic elution for an additional 10 min at a flowate of 0.5 mL/min.

cation of the investigated PLs.

2.3. Lipid extraction

A volume of 100 �L of the serum was transferred to 15-mL cleanscrew-capped test tubes and then mixed with 3 mL of extractionsolvent (MeOH: CH2Cl2, 2:1, v/v) and 10 �L of IS solution mixturecontaining LPC (17:0), PE(15:0/15:0) and LPS(17:0) at concentra-tions of 100, 100 and 50 ng/�L, respectively. The mixed solutionwas sonicated for 15 min at 40 ◦C and centrifuged at 5100 rpm for20 min. The clear supernatant was transferred to 10-mL tubes, driedwith a gentle stream of nitrogen gas at room temperature, recon-stituted in 100 �L of MeOH:CH2Cl2 (2:1, v/v) and vortexed for 10 s.The solution was quantitatively transferred to a 1.0-mL clean total-recovery auto-sampler vial, dried with a gentle stream of nitrogengas at room temperature, and reconstituted in 25 �L of acetoni-trile:methanol (1:1,v/v) with the aid of 10 s of vortexing. A volumeof 5 �L was injected for LC-MS analysis.

2.4. Method optimization

The quality control (QC) samples were prepared by spik-ing a standard solution mix containing four non-endogenouslipids—PC(17:0/17:0), PE(17:0/17:0), PS(17:0/17:0), and

the QC samples as described in Section 2.3. Sample stability was

A. Khedr et al. / J. Chromatogr. B 1009 (2016) 44–54 47

F usingt

e0fettisnPewps

2

atst(

ig. 2. Base peak chromatograms (BPC) of PLs extracted from human serum collectedime points.

valuated using the spiked QC samples and analyzed on days, 7, and 14. Three QC samples were also analyzed after threereeze/thaw cycles. Intra-day accuracy and precision (n = 6) werevaluated by analyzing the QC samples on the same day, whilehe inter-day accuracy and precision were determined analyzinghe QC samples over six days. The precision was determinedn terms of the relative standard deviation (RSD). Three QCamples that were spiked with a solution containing the fouron-endogenous phospholipids—PC(17:0/17:0), PE(17:0/17:0),S(17:0/17:0), and LPI(17:1)—at three concentration levels werextracted and analyzed to estimate the % recovery. The % recoveryas assessed by matching the ratio between the area each com-

ound spiked in the QC sample and that of the standard workingolution.

.5. Statistical analysis

All quantitative data collected for the investigated groups werenalyzed using principal component analysis (PCA) to determine

he lipid profile relationships using Multivariate Statistical Packageoftware (MVSP, version 3.22, Kovach Computing Services, Pen-raeth, Isle of Anglesey, UK, 2013). One-way analysis of varianceANOVA) followed by the Bonferroni Post Hoc multiple compar-

positive (a) and negative (b) ESI-TQ-MS modes; the lipid classes eluted at clustered

isons [34] were also applied to identify significant differences ofeach lipid between groups. The univariate analysis was performedusing SPSS statistical software (version 13, SPSS, USA), and statis-tical significance was identified at p < 0.05.

3. Results and discussion

Human serum is a complex biofluid composed of a variety ofmetabolites, including many lipid molecular species. Determin-ing the relationship between the lipid profile and its biologicalimplications requires sensitive analytical methods to catalogue andquantify lipids into their respective molecular classes. RP-HPLC/MShas better selectivity and sensitivity for serum lipid profiling thanNP-HPLC-MS [17]. In the present study, IT-MS and TQ-MS enabledthe characterization and quantification of major five classes ofserum PLs.

3.1. Optimization of the LC-MS method

Biogenic PLs are known to exhibit ionization suppression inESI-MS, resulting in fluctuating signals and low responses [35].The matrix effect was tested by analyzing a non-endogenous PLstandard solution prepared in CHCl3:MeOH (1:1, v/v) and the

4 atogr

sshr2vvi

wCtMcPisrndsttdiapiartare

3

fiaasaootMwmMMsoTmsLdwm+o1T1m

8 A. Khedr et al. / J. Chrom

ame concentrations in extracted serum. The PLs-ESI-MS responsehowed a varying peak area due to ionization suppression atigh concentrations and large injection volumes. The peak areaesponses obtained from repetitive injection of 10 �L (6 times) of50, 500, and 1000 �L serum extract showed a relatively high RSDalue of 10.2, 16.4, and 18.8%, respectively. The use of a small serumolume (50–100 �L) and a small injection volume (1–5 �L) resultedn the collection of precise data (RSD ≤ 0.5%).

The recommended lipid extraction solvents reported recentlyere optimized [36–38]. The extraction procedures described by

astro-Perez et al. [8] indicates recovery values ranging from 74o 94% for 5 non-endogenous lipids. In this work, the use of

eOH:CH2Cl2 (2:1, v/v) as the extraction solvent resulted in per-ent recoveries of 89.4%, 83.2%, 91.5% and 91.2% for PC (17:0/17:0),E (17:0/17:0), PS (17:0/17:0) and LPI (17:0), respectively. The son-

cation and heating of the extraction mixture at 40 ◦C for 15 min, increw caped test tube, was necessary to obtain precise data. Theesults shown in Table 1 indicate that the average RSD of the fouron-endogenous PLs in serum at three time points (0, 7, 14 days)id not exceed 1.5%. The same result was obtained by analyzing theerum samples after three freeze/thaw cycles. These results illus-rated that all of the investigated PLs were stable for the duration ofhe analysis period of 14 days (stored at −80 ◦C). The linearity of theeveloped method was assessed using four non-endogenous lipids,

ncluding PC (17:0/17:0), PE (17:0/17:0), PS (17:0/17:0), LPI (17:0),t six different concentrations (Table 2). The calculated calibrationarameters, including the squared regression coefficient (r2), slope,

ntercept, concentration range (�g/mL), limit of detection (LOD)nd limit of quantification (LOQ), are listed in Table 2. These resultsevealed that the method was sufficiently sensitive and precise forhe quantitative analysis of the targeted PLs in human serum. Intra-nd inter-day precision values are presented in Table 1. The accu-acy was expressed as the percent relative error (%RE) and did notxceed 4% relative to the recovered concentration.

.2. Systematic identification of investigated PLs

Fig. 1 presents the procedure used for the search and identi-cation of PLs. For the feasible detection and separation of LPCsnd PCs, a mobile system was applied, which included formic acidnd positive MS mode, while a mobile system consisting ammoniaolution with negative MS mode was used to monitor PSs, PEs, PIsnd their lyso-forms. The phospholipid species were characterizedn the basis of (i) MS2 peaks due to polar head, (ii) precursor ionr neutral loss scans, (iii) identification of fatty acid, (iv) identifica-ion of sn-1 and sn-2 fatty acid. To detect all of the lyso-PLCs, the

S2 ion (m/z 184) corresponding to the phosphocholine polar headas extracted. Moving the computer mouse over this MS2 chro-atogram at different retention times revealed the correspondingS parent ion. This parent ion was then extracted to obtain both theS and MS2 spectra of that specific ion. One fragment from the MS2

pectra was selected and extracted. The average MS2 spectrum wasbtained along the elution time of the parent ion chromatogram.his MS2 spectrum showed the minor and major abundant frag-ents, which were used for further confirmation of the chemical

tructure of LPCs with the aid of chemDraw software and the onlineipidomics-Gateway database. Phosphatidylcholines seldom pro-uced a phosphocholine ion (m/z 184); however, this class of PLsas easily identified by searching for neutral loss fragments of/z 183 [M-phosphocholine + H]+ and/or m/z 59 [M-choline group

H]+. The -MS2 spectra of PIs were characterized by the presencef two abundant peaks in addition to the parent ion. A value of m/z

62 resulted from the subtraction of both abundant MS2 peaks.his value corresponds to the loss of inositol [(m/z 180)-H2O (m/z8)]. Searching the -MS2 chromatogram for the neutral loss of/z 87u allowed for the identification of PSs. The most abundant

. B 1009 (2016) 44–54

PS ion was characterized by the [M-serine (105)-H2O (18)]− ion.Phosphatidyl-ethanolamines are detectable in positive and nega-tive modes. However, the negative mode was preferable because itallowed for easy identification. Searching for m/z 43 u, which cor-responds to the aziridine head group, identified the PEs. The fattyacid moieties were identified from the negative-MS2 spectra, m/z[fatty acid–H]−.

Fig. 2 shows a representative, extracted base peak chro-matogram (BPC) of serum PLs detected using both positive andnegative MS modes. The positive MS2 spectrum of the parent ion atm/z 496.3 (Fig. 3a) was identified as LPC (16:0). The MS2 spectrumof this parent ion yielded two major abundant m/z peaks at 184and 478.5, corresponding to the phosphocholine head group and[M-H2O + H]+, respectively. Minor abundant positive ions at m/z419 and 258 corresponded to [M-(H2O+ (CH)3N) + H]+ and [M + H-RCH C O]+ (loss of a fatty acid, C16:0, in the form of a ketene),respectively. The positive MS2 spectrum of PC (36:2), as shown inFig. 3b, exhibited a [M+H]+ ion at m/z 786.7. The loss of a phos-phocholine cation was confirmed by the presence of m/z 603.5[M−184 + H]+. Further cleavage of C18:2 (sn-2) and C18:0 (sn-1)resulted in the production of m/z 506 [M-R2COO + H]+ and m/z 502[M-R1COO + H]+, respectively. The loss of fatty acid (C18:2), in formof ketene (sn-2), was confirmed by the abundant peak at m/z 524[M-C17H31C O +2H]+. R1 and R2 correspond to the types of fattyacid at glycerol carbons sn-1 and sn-2, respectively (R1 = C18:0, andR2 = C18:2).

The negative MS2 spectra of the parent ion at m/z 743(Fig. 3c)were identified as PE (36:2). The MS2 spectrum of this parent ionyielded two major peaks at m/z 480 and 279, corresponding to [M-(R2CH C O)-H]−, which was the result of the loss of a fatty acid(C18:2 in the form of a ketene) and the fatty acid C18:2, respectively.The loss of an aziridine anion was confirmed by the presence of m/z699 [M-43-H]−. As reported previously [39], the [M − RCH C O]−

fragment derived from the fragmentation of the carboxylate ion ofthe fatty acid at the sn-2 position is prominent, and the R1 and R2correspond to the types of fatty acid (R1 = C18:0, and R2 = C18:2).The negative MS2 spectrum of PI (38:4), which is shown in Fig. 3d,revealed a [M-H]−ion at m/z 886. The MS2 spectrum of this parention yielded two major peaks at m/z 581 and 419, correspondingto [M-(R2CH C O)-H]− (loss of a fatty acid C20:4 in the form of aketene) and [M- (162–R2COOH)-H]− (loss of an inositol fragmentand the fatty acid C20:4 of m/z 303), respectively. The loss of theinositol anion, m/z 162, was confirmed by the difference betweenthe two major peaks at m/z 581 and 419. The fatty acids C20:4 (sn-2) and C18:0 (sn-1) resulted in the production of m/z 303 and m/z283, respectively. The negative MS2 spectrum of PS (36:2), shownin Fig. 3e, exhibited a [M−H]−ion at m/z 786. The MS2 spectrumof this parent ion showed one abundant peak at m/z 699.5, corre-sponding to [M-88]− due to loss of the serine head group. The twominor peaks at m/z 417and 281 correspond to [M- (88 + RCOOH)-H]−, which is the result of the loss of a serine moiety and the fattyacid C18:1 (sn1) or C18:1 (sn2), respectively.

3.3. Dysregulated phospholipids and related pathways in viralinfectious diseases

The obtained data of 14 subjects from each of the healthy (HE),DF, HB, and HVC groups are listed in Table 3 as average values.These data were used to determine significant and nonsignifi-cant differences in each PL between HE and the viral-infectedcases and between each viral group. The quantified data of 35PLs in human serum samples from HE versus DF, HBV, and HCV

viral infections were statistically analyzed using principal compo-nent analysis (PCA). As seen in Fig. 4, the patients in each diseasegroup have distinctly different lipid profiles from those of healthycontrols, indicating that patients suffering from viral infectious

A. Khedr et al. / J. Chromatogr. B 1009 (2016) 44–54 49

Fig. 3. Representative MS2 spectra of the five PLs classes. (a) LPC, m/z 497[M+H]+, (b) PC, m/z 787[M+H]+, (c) PE, m/z 743[M−H]− , (d) PI, m/z 886 [M−H]− , (e) PS, m/z 786[M−H]− .

50 A. Khedr et al. / J. Chromatogr. B 1009 (2016) 44–54

Table 1The percentage recovery, precision, and percentage residual values obtained from the analysis of the spiked solution (n = 6).

Phospholipid % Recoverya Intra-day precision, % RSD Inter-day precision, % RSD Residual %

PC(17:0/17:0) 89.39 ± 2.05 0.77 1.02 −10.6PE(17:0/17:0) 83.18 ± 2.56 1.22 1.65 −16.8PS(17:0/17:0) 91.48 ± 4.61 0.45 0.99 −8.5LPI(17:1) 91.19 ± 2.07 0.90 1.37 −8.8

a Mean values of the three levels, including 100 ± 25% of the median concentration.

F thy co

dpiaAiotAa

3

itLcc

TR

ig. 4. PCA score plot of mean-centered data variables (n = 14 for each group). Heal

iseases have disrupted lipid metabolism. To determine whichhospholipid molecules were dysregulated in patients with viral

nfectious diseases and to identify the phospholipid biomarkersble to distinguish among the three studied diseases, one-wayNOVA followed by the Bonferroni post hoc multiple compar-

sons test [34] were applied to assess the statistical significancef each phospholipid between groups. The ANOVA results revealedhat 28 of the 35 quantified PLs were differentially regulated (PNOVA < 0.05) between groups, including 6 LPCs, 5 PCs, 9 PIs, 3 PEsnd 5 PSs.

.4. Lysophosphatidylcholines (LPCs)

LPCs are important signaling molecules involved in regulat-ng cellular proliferation and inflammation [40]. Table 3 indicates

hat six LPCs in DF serum—LPC (16:0), LPC (18:1), LPC(18:2),PC (18:0), LPC(20:4) and LPC (20:3)—and only two LPCs in HCVases—LPC(18:2) and LPC(20:4)—were significantly downregulatedompared with HE subjects. However, two LPCs in HBV—LPC (16:0)

able 2egression parameters of the non-endogenous PL standards (n = 3).

Phospholipid Intercept Slope Squared regression coeffici

PC(17:0/17:0) −0.0613 0.1186 0.9993

PE(17:0/17:0) 0.0232 0.0196 0.9995

PS(17:0/17:0) 0.0416 0.0012 0.9991

LPI(17:1) 0.0036 0.0472 0.9990

ntrol (upright triangle), HBV (inverted triangle), HCV (square) and DF (diamond).

and LPC (18:1)—were significantly upregulated in HBV comparedwith HE subjects. Based on these results, LPCs could have potentialas characteristic biomarkers to differentiate between HE subjectsand the three studied viral infections. This finding was in agreementwith the results of Khedr et al., who observed a dramatic decrease inesterified fatty acids in DF [41]. LPCs are the substrate of lysophos-pholipase enzymes, which hydrolyze 2-lysophosphatidylcholine inthe presence of water to produce glycerophosphocholine and liber-ate esterified fatty acids. Wu et al. observed the overexpression oflysophospholipase enzymes in one liver disease, the hepatocellularcarcinoma [42], which may represent the mechanism underlyingreduced serum LPC in DF and HCV diseases in our study. Table 3showed that six LPCs, including, LPC(16:0), LPC(18:1), LPC(18:2),LPC (18:0), LPC(20:4) and LPC(20:3), were significantly upregu-lated in HBV compared with DF. This difference could represent

a characteristic biomarker with the ability to differentiate betweenDF and HBV. Moreover, five LPCs, including, LPC(16:0), LPC(18:0),LPC(18:1), LPC(18:2) and LPC(20:3), were significantly upregulatedin HCV compared with DF. This difference may also be applied as

ent (r2) Range, (�g/mL) LOQ, (�g/mL) LOD, (�g/mL)

10.0–60.0 10.0 3.110.0–60.0 10.0 3.2

5.0–30.0 5.0 1.620.0–120.0 20.0 6.5

A.

Khedr

et al.

/ J.

Chromatogr.

B 1009

(2016) 44–54

51

Table 3List of the phospholipids quantified in sera of three viral infectious diseases and calculated P (0.05) values in comparison with each other.

Name m/z Ion Healthya DFa HBVa HCVa DF to HE HBV to HE HCV to HE HBV to DF HCV to DF HCV to HBV(�g/mL ± SD) (�g/mL ± SD) (�g/mL ± SD) (�g/mL ± SD) P value P value P value P value P value P value

LPC (16:0) 496.5 [M+H]+ 59.41 ± 8.6 38.23 ± 9.7 80.17 ± 16.2 71.03 ± 12.0 1.3E-04↓ 1.8E-04↑ 8.2E-02 9.3E-12↑ 1.3E-08↑ 3.0E-01LPC (18:2) 520.5 [M+H]+ 16.78 ± 3.9 7.91 ± 3.1 16.21 ± 3.3 11.69 ± 3.9 1.8E-07↓ 1.0E + 00 2.8E-03↓ 8.3E-07↑ 4.6E-02↑ 1.0E-02↓

LPC (18:1) 522.5 [M+H]+ 16.35 ± 4.6 7.44 ± 2.4 20.06 ± 3.2 18.00 ± 3.2 6.7E-08↓ 4.1E-02↑ 1.0E + 00 2.6E-12↑ 6.8E-10↑ 7.5E-01LPC (18:0) 524.5 [M+H]+ 36.77 ± 13.3 13.38 ± 4.9 45.57 ± 8.3 44.35 ± 7.6 6.7E-08↓ 5.4E-02 5.4E-02 8.1E-13↑ 8.1E-13↑ 1.0E + 00LPC (20:4) 544.5 [M+H]+ 5.62 ± 2.1 3.17 ± 0.7 5.03 ± 1.3 4.02 ± 1.4 4.4E-04↓ 1.0E + 00 4.1E-02↓ 1.2E-02↑ 8.5E-01 4.9E-01LPC (20:3) 546.5 [M+H]+ 3.94 ± 0.6 2.46 ± 1.0 4.36 ± 1.0 4.32 ± 1.1 1.5E-03↓ 1.0E + 00 1.0E + 00 3.6E-05↑ 4.9E-05↑ 1.0E + 00PC (32:0) 734.7 [M+H]+ 6.52 ± 2.7 7.16 ± 2.5 6.42 ± 1.0 6.14 ± 2.6 1.0E + 00 1.0E + 00 1.0E + 00 1.0E + 00 1.0E + 00 1.0E + 00PC (34:2) 758.7 [M+H]+ 82.97 ± 29.6 88.24 ± 20.6 88.60 ± 24.1 77.33 ± 23.2 1.0E + 00 1.0E + 00 1.0E + 00 1.0E + 00 1.0E + 00 1.0E + 00PC (34:1) 760.7 [M+H]+ 57.13 ± 25.0 67.57 ± 19.5 62.49 ± 14.9 55.68 ± 14.0 9.0E-01 1.0E + 00 1.0E + 00 1.0E + 00 6.1E-01 1.0E + 00PC (36:4) 782.7 [M+H]+ 34.41 ± 12.7 38.66 ± 8.8 21.27 ± 4.3 21.11 ± 4.2 1.0E + 00 7.0E-04↓ 5.9E-04↓ 6.6E-06↓ 5.5E-06↓ 1.0E + 00PC (36:3) 784.7 [M+H]+ 41.45 ± 7.9 44.50 ± 9.8 42.86 ± 8.6 41.28 ± 8.0 1.0E + 00 1.0E + 00 1.0E + 00 1.0E + 00 1.0E + 00 1.0E + 00PC (36:2) 786.7 [M+H]+ 55.06 ± 18.3 56.76 ± 16.4 56.12 ± 17.5 55.74 ± 13.9 1.0E + 00 1.0E + 00 1.0E + 00 1.0E + 00 1.0E + 00 1.0E + 00PC (36:1) 788.7 [M+H]+ 18.43 ± 5.8 17.10 ± 6.5 15.98 ± 4.0 17.81 ± 5.1 1.0E + 00 1.0E + 00 1.0E + 00 1.0E + 00 1.0E + 00 1.0E + 00PC (38:6) 806.7 [M+H]+ 23.50 ± 8.7 20.71 ± 6.8 12.94 ± 3.7 19.50 ± 6.3 1.0E + 00 6.5E-04↓ 7.1E-01 2.0E-02↓ 1.0E + 00 7.2E-02PC (38:5) 808.7 [M+H]+ 7.81 ± 2.2 8.67 ± 2.7 4.61 ± 1.1 4.56 ± 1.1 1.0E + 00 3.5E-04↓ 2.8E-04↓ 5.8E-06↓ 4.5E-06↓ 1.0E + 00PC (38:4) 810.7 [M+H]+ 26.29 ± 8.2 27.44 ± 8.2 16.29 ± 3.4 14.05 ± 2.7 1.0E + 00 5.1E-04↓ 1.9E-05↓ 9.8E-05↓ 3.4E-06↓ 1.0E + 00PC (38:3) 812.7 [M+H]+ 25.55 ± 7.6 26.95 ± 5.9 15.61 ± 5.1 13.93 ± 2.1 1.0E + 00 1.2E-05↓ 4.6E-07↓ 2.8E-06↓ 4.6E-07↓ 1.0E + 00LPI (16:0) 572.6 [M−H]− 35.34 ± 12.1 139.02 ± 33.5 82.10 ± 22.8 127.94 ± 30.7 4.6E-07↑ 1.1E-04↑ 4.6E-07↑ 2.8E-06↓ 1.0E + 00 1.5E-04↑

LPI (18:0) 599.3 [M−H]− 147.60 ± 29.2 148.67 ± 33.0 362.07 ± 92.3 163.91 ± 40.1 1.0E + 00 4.6E-07↑ 1.0E + 00 1.3E-14↑ 7.4E-01 2.8E-12↓

PI (34:2) 833.6 [M−H]− 165.84 ± 51.3 87.03 ± 11.7 92.10 ± 25.9 91.41 ± 21.6 4.6E-07↓ 4.6E-07↓ 4.6E-07↓ 1.0E + 00 1.0E + 00 1.0E + 00PI (34:1) 835.6 [M−H]− 138.57 ± 37.3 95.80 ± 31.4 106.90 ± 30.8 70.84 ± 19.9 3.1E-03↓ 5.0E-02 4.6E-07↓ 1.0E + 00 2.1E-01 1.8E-02↓

PI (36:4) 857.6 [M−H]− 198.88 ± 58.1 58.82 ± 18.4 103.41 ± 29.3 71.60 ± 21.2 4.6E-07↓ 4.6E-07↓ 4.6E-07↓ 9.7E-03↑ 1.0E + 00 1.3E-01PI (36:2) 861.6 [M−H]− 205.58 ± 48.1 75.64 ± 21.7 108.91 ± 31.7 81.28 ± 24.0 4.6E-07↓ 4.6E-07↓ 4.6E-07↓ 6.2E-02↑ 1.0E + 00 1.9E-01PI (36:1) 863.6 [M−H]− 70.33 ± 18.4 25.78 ± 7.5 67.98 ± 10.3 35.74 ± 9.9 4.6E-07↓ 1.0E + 00 4.6E-07↓ 1.4E-11↑ 2.2E-01 3.4E-08↓

PI (38:4) 885.6 [M−H]− 806.33 ± 261.1 319.29 ± 96.0 408.38 ± 109.8 250.65 ± 54.7 4.6E-07↓ 4.6E-07↓ 4.6E-07↓ 1.0E + 00 1.0E + 00 3.4E-08↓

PI (38:3) 887.6 [M−H]− 111.67 ± 36.7 21.28 ± 5.8 67.36 ± 17.9 38.40 ± 10.5 4.6E-07↓ 1.2E-05↓ 4.6E-07↓ 3.1E-06↑ 2.3E-01 4.3E-03↓

LPE (20:4) 500.4 [M−H]− 5.70 ± 1.8 5.69 ± 2.0 6.28 ± 1.4 5.17 ± 1.2 1.0E + 00 1.0E + 00 1.0E + 00 1.0E + 00 1.0E + 00 5.3E-01PE (36:2) 741.9 [M−H]− 8.55 ± 1.6 9.54 ± 2.7 7.39 ± 2.6 10.24 ± 2.8 1.0E + 00 1.0E + 00 4.7E-01 1.6E-01 1.0E + 00 2.3E-02↑

PE (38:5) 764.5 [M−H]− 17.02 ± 4.2 7.03 ± 2.1 7.05 ± 2.6 4.63 ± 1.3 4.6E-07↓ 4.6E-07↓ 4.6E-07↓ 1.0E + 00 1.6E-01 1.5E-01PE (38:4) 766.5 [M−H]− 9.48 ± 2.9 9.91 ± 2.2 6.18 ± 1.4 5.59 ± 1.5 1.0E + 00 2.1E-03↓ 4.6E-07↓ 7.1E-16↓ 1.1E-16↓ 1.0E + 00PS (36:3) 784.5 [M−H]− 22.99 ± 5.8 12.32 ± 4.2 11.55 ± 4.1 11.53 ± 3.1 4.6E-07↓ 4.6E-07↓ 4.6E-07↓ 1.0E + 00 1.0E + 00 1.0E + 00PS (36:2) 786.5 [M−H]− 6.50 ± 2.4 6.34 ± 1.2 2.80 ± 1.0 2.72 ± 0.7 1.0E + 00 7.1E-06↓ 7.1E-06↓ 2.3E-08↓ 1.6E-08↓ 1.0E + 00PS (36:1) 788.5 [M−H]− 14.69 ± 5.7 10.28 ± 3.7 9.54 ± 3.2 11.93 ± 3.6 4.6E-02↓ 1.2E-02↓ 5.3E-01 1.0E + 00 1.0E + 00 8.3E-01PS (36:0) 790.5 [M−H]− 40.48 ± 10.4 12.42 ± 3.4 13.14 ± 5.0 10.49 ± 3.3 4.6E-07↓ 4.6E-07↓ 4.6E-07↓ 1.0E + 00 1.0E + 00 1.0E + 00PS (38:5) 808.5 [M−H]− 32.79 ± 8.0 21.52 ± 7.8 9.81 ± 3.9 13.16 ± 3.9 9.5E-05↓ 4.6E-07↓ 4.6E-07↓ 4.9E-05↓ 5.2E-03↓ 9.8E-01

↑ Significant increase.↓ Significant decrease.

52 A. Khedr et al. / J. Chromatogr. B 1009 (2016) 44–54

ration

abd

3

[iac

Fig. 5. Difference between the average concent

biomarker to differentiate between these two viral infections. Inoth cases, HBV and HCV, the level of LPC(18:2) was significantlyownregulated.

.5. Phosphatidylcholines (PCs)

PCs are the major PL class in plasma according to van der Veen

43]. Table 3 indicates that five polyunsaturated PCs in HBV serum,ncluding PC(36:4), PC(38:6), PC(38:5), PC(38:4), and PC(38:3),nd the same PCs except PC(38:6) in HCV serum were signifi-antly downregulated compared with HE subjects. Based on this

s of the PLs that exhibited significant changes.

result, these PCs could be considered as biomarkers to differentiatebetween HE subjects and both HBV and HCV cases. PCs are mainlysynthesized through the catalyzed methylation of PEs by PE-N-methyltransferase [44]. Tessitore et al. observed that hepatocellularcarcinoma was associated by decreased PE-N-methyltransferaseactivity [45], which may be the mechanism underlying reducedserum PCs in HBV and HCV in our study. Table 3 shows thatfive polyunsaturated PCs, including PC(36:4), PC(38:3), PC(38:4),

PC(38:5), and PC(38:6), were significantly downregulated in HBVcompared with DF. This quantitative profile could represent a char-acteristic biomarker to distinguish DF from HBV. Moreover, four

A. Khedr et al. / J. Chromatogr. B 1009 (2016) 44–54 53

Table 4Matching the PLs changes as per our findings with the reported data.

Potential biomarkers Developed method Reported methodFold change of the mean concentration of PL

LPC (16:0) −1.56 (HE/DF) −1.91 (HE/DF) [50]LPC (18:2) −2.12 (HE/DF) −2.61 (HE/DF) [50]LPC(18:1) −2.19 (HE/DF) −1.71 (HE/DF) [50]LPC(18:0) −2.76 (HE/DF) −1.95 (HE/DF) [50]PC (36:4) −1.62 (HE/HBV) SDRa (HE/HBV) [17]

a

pPDtfit

3

dsPitPLTtpioaeltgvawfbLca

3

hss(iadeasict

PC (38:4) −1.61 (HE/HBV)

LPI (16:0) +3.62 (HCV/HE)

a SDR, Significant down regulated, no value found.

olyunsaturated PCs, including PC(38:3), PC(38:4), PC(38:5), andC(36:4), were significantly downregulated in HCV compared withF. This profile could also represent as a characteristic biomarker

o differentiate between these two viral infections. The PC pro-les in HBV and HCV were very similar and thus cannot be used

o differentiate between these two diseases.

.6. Phosphatidylinositols (PIs)

PIs, a major class of PLs, are involved in intracellular trans-uction [46]. Table 3 presents the significant downregulation ofeven PIs, including PI(34:1), PI(34:2), PI(36:1), PI (36:2), PI(36:4),I(38:3) and PI(38:4), and the significant upregulation of LPI(16:0)

n both DF and HCV compared with the sera of HE subjects. Addi-ionally, five PIs, including PI(34:2), PI(36:2), PI(36:4), PI(38:3) andI(38:4), were downregulated and two LPIs, including, LPI(16:0 andPI (18:0), were upregulated in HBV compared with HE subjects.hese results support the hypothesis that PI hydrolysis occurs inhese three viral infectious diseases due to the activation of thehospholipase A2 enzyme. This enzyme is responsible for hydrolyz-

ng the ester linkage between fatty acids and the hydroxyl groupf PLs and results in the generation of more free fatty acids, as wells lysophospholipids [47]. This finding was in agreement with Roet al. and Qu et al., who observed a significant increase in LPIs fol-owing HCV infection [18,19]. Generally, the quantitative profile ofhe PI class could represent a characteristic biomarker to distin-uish between HE subjects and patients suffering from these threeiral diseases. The results listed in Table 3 indicate that LPI(18:0)nd four PIs, including PI(36:2), PI(36:1), PI(36:4) and PI(38:3),ere significantly upregulated in HBV compared with DF. This dif-

erence could be used as a diagnostic biomarker to differentiateetween DF and HBV. Moreover, five PIs in HCV serum, includingPI(18:0), PI(34:1), PI(36:1), PI(38:3) and PI(38:4), were signifi-antly downregulated compared with HBV. This profile could bepplied to distinguish between these two viral infections.

.7. Phosphatidylethanolamines (PEs)

PEs are found in all living cells and constitute 25% of all PLs. Inuman physiology, they are particularly prevalent in nervous tis-ue, where they account for 45% of all phospholipids [48]. Table 3hows the significant downregulation of two PEs—PE(38:4) and PE38:5)—in HBV and HCV sera and the downregulation of PE(38:5)n DF serum compared with the healthy group. These results arettributed to the hydrolysis of PEs in these three viral infectiousiseases resulting from the activation of the phospholipase A2nzyme, as was the case for the PI class. This profile could be a char-cteristic biomarker with the ability to differentiate between HE

ubjects and these three viral diseases. Table 3 presents the signif-cant upregulation of PE(36:2) in HCV compared with HBV, whichould be a biomarkers differentiating these two diseases. Moreover,he significant downregulation of PE(38:4) in HBV and HCV com-

SDR (HE/HBV) [17]+2.13 (HCV/HE) [18]

pared with DF could be used to distinguish DF from the other twoviral diseases studied here.

3.8. Phosphatidylserines (PSs)

PSs constitute approximately 2–10% of total cellular lipids,depending on species and cell type [49]. Table 3 shows the sig-nificant downregulation of five PSs, including PS(36:0), PS(36:1),PS(36:2), PS(36:3) and PS(38:5) in HBV compared with HE subjects.Furthermore, the same PSs, except PS(36:1), were significantlydownregulated in HCV compared with the HE group. The previousresults were the result of the hydrolysis of PSs in the three studiedviral infectious diseases, which may attributed to the activation ofthe phospholipase A2 enzyme, as in the previous two classes. Thisdifference could be used as a biomarker to differentiate betweenHE subjects and these three viral diseases. Table 3 presents thesignificant downregulation of PS(36:2) and PS(38:5) in HBV andHCV sera compared with DF. This difference could be a character-istic biomarker able to differentiate between DF and the two otherstudied viral infections.

3.9. Differentiating biomarkers

To summarize the differentiating biomarkers, Fig. 5 was plottedto discriminate between the three viral infectious diseases and thehealthy cases and between each of the viral diseases. This figureconsists of a plot of the differences in average PL concentrationsthat exhibited significant changes between the studied groups.

Decreases in PI(38:4), PI(36:4), and PI(36:2) and an increasein LPI(16:0) were observed to have potential as differentiatingbiomarkers upon matching the three disease groups versus thehealthy subjects. Moreover, LPI(18:0) and LPI(16:0) could be useddiscrimination between DF and HBV. The dramatic increases in bothLPC(16:0) and LPC (18:0) and the decrease in PC(36:4) could poten-tially be applied to differentiate between HCV and DF. Ultimately,three novel PLs were identified as biomarkers with the ability to dif-ferentiate between HCV and HBV cases. Decreases in both LPI(18:0)and PI(38:4) and an increase in LPI(16:0) in HCV compared withHBV were observed.

Some of these PLs, have been reported as potential biomarkersusing plasma of DF, HBV, or HCV versus healthy subjects. A signifi-cant increase of four PLs, including LPC(16:0), LPC(18:2), LPC(18:1)and LPC(18:0) in DF cases have been reported by Cui et al. [50].Table 4 showed the reported fold changes of the mean concentra-tions of PLs in HE and viral infected cases and the measured valuesusing the developed method. Although, the number of reportedPLs that monitored as biomarker are few, the obtained data werecomparable.

4. Conclusions

Optimized LC-ESI-MS and extraction procedures were appliedfor the identification and quantification of PLs in human serum.

5 atogr

TeibdscrTeFwjdhTafcea

C

A

i1

R

[

[[[[

[[[

[[

[

[[[[[

[

[

[[

[

[

[[

[

[[[

[

[

[

[[

[

[[

[

4 A. Khedr et al. / J. Chrom

he minimization of ionization suppression and enhanced% recov-ry improved the method sensitivity and precision. The structuraldentification of potential phospholipids biomarkers was achievedy Ion trap-MSn following a described schematic MS-scan proce-ure. The statistical and principle component analyses of the majorerum PLs in HE subjects and the three groups of viral infectionases revealed distinctly different profiles that reflected the dis-upted lipid metabolism in patients with viral infectious diseases.he obtained results revealed that 19 PLs can be used as differ-ntiating biomarkers between healthy subjects and DF patients.urthermore, twenty-one PLs were significantly different in HBV,hile 20 PLs were able to differentiate HCV from the healthy sub-

ects. The obtained results revealed that 20 PLs could significantlyifferentiate between DF and HBV diseases and that 12 PLs couldave application as biomarkers to distinguish between DF and HCV.he PI class can be used as biomarkers to differentiate between HCVnd HBV because 6 PIs were significantly different in sera collectedrom patients with these two diseases. The PLs of the three viralases, that showed potential concentration changes, were consid-red as differentiating biomarkers. This method enabled fast andccurate diagnosis of the three clinically similar viral cases.

onflict of interest

The authors have declared no conflict of interest.

ppendix A. Supplementary data

Supplementary data associated with this article can be found,n the online version, at http://dx.doi.org/10.1016/j.jchromb.2015.2.011.

eferences

[1] E. Fahy, S. Subramaniam, R.C. Murphy, M. Nishijima, C.R. Raetz, T. Shimizu, F.Spener, G. van Meer, M.J. Wakelam, E.A. Dennis, J. Lipid Res. 50 (2009) S9–S14.

[2] O. Quehenberger, A.M. Armando, A.H. Brown, S.B. Milne, D.S. Myers, A.H.Merrill, S. Bandyopadhyay, K.N. Jones, S. Kelly, R.L. Shaner, J. Lipid Res. 51(2010) 3299–3305.

[3] C. Zhu, Q.L. Liang, P. Hu, Y.M. Wang, G.A. Luo, Talanta 85 (2011) 1711–1720.[4] M.J. Berridge, Nature 361 (1993) 315–325.[5] A.J. Marcus, D.P. Hajjar, J. Lipid Res. 34 (1993) 2017–2031.[6] G. Corona, J. Polesel, L. Fratino, G. Miolo, F. Rizzolio, D. Crivellari, R. Addobbati,

S. Cervo, G. Toffoli, J. Cell Physiol. 229 (2014) 898–902.[7] L.Q. Pang, Q.L. Liang, Y.M. Wang, L. Ping, G.A. Luo, J. Chromatogr. B Anal.

Technol. Biomed. Life Sci. 869 (2008) 118–125.

[8] J.M. Castro-Perez, J. Kamphorst, J. DeGroot, F. Lafeber, J. Goshawk, K. Yu, J.P.

Shockcor, R.J. Vreeken, T. Hankemeier, J. Proteome Res. 9 (2010) 2377–2389.[9] M.Z. Ashraf, N.S. Kar, E.A. Podrez, Int. J. Biochem. Cell Biol. 41 (2009)

1241–1244.10] A. Varatharaj, Neurol. India 58 (2010) 585–591.

[[[[

. B 1009 (2016) 44–54

11] J. Whitehorn, J. Farrar, Br. Med. Bull. 95 (2010) 161–173.12] S. Ranjit, N. Kissoon, Pediatr. Crit. Care Med. 12 (2011) 90–100.13] N.S. Heaton, G. Randall, Cell Host Microbe 8 (2010) 422–432.14] E.C. van Gorp, C. Suharti, A.T. Mairuhu, W.M. Dolmans, J. van Der Ven, P.N.

Demacker, J.W. van Der Meer, Clin. Infect. Dis. 34 (2002) 1150–1153.15] H.B. El-Serag, K.L. Rudolph, Gastroenterology 132 (2007) 2557–2576.16] C.W.G. The Global Burden of Hepatitis, J. Clin. Pharmacol. 44 (2004) 20–29.17] S. Chen, P. Yin, X. Zhao, W. Xing, C. Hu, L. Zhou, G. Xu, Electrophoresis 34

(2013) 2848–2856.18] B. Roe, E. Kensicki, R. Mohney, W.W. Hall, PLoS One 6 (2011) e23641.19] F. Qu, S.-J. Zheng, C.-S. Wu, Z.-X. Jia, J.-L. Zhang, Z.-P. Duan, Anal. Bioanal.

Chem. 406 (2014) 555–564.20] J.L. Little, M.F. Wempe, C.M. Buchanan, J. Chromatogr. B Anal. Technol.

Biomed. Life Sci. 833 (2006) 219–230.21] J.J. Myher, A. Kuksis, S. Pind, Lipids 24 (1989) 396–407.22] J. Pucsok, L. Kovacs, A. Zalka, R. Dobo, Clin. Biochem. 21 (1988) 81–85.23] J. Lee, H. Min, M. Moon, Anal. Bioanal. Chem. 400 (2011) 2953–2961.24] H.Y. Kim, T.C. Wang, Y.C. Ma, Anal. Chem. 66 (1994) 3977–3982.25] E.J. Lesnefsky, M.S. Stoll, P.E. Minkler, C.L. Hoppel, Anal. Biochem. 285 (2000)

246–254.26] E. Boselli, D. Pacetti, P. Lucci, N.G. Frega, J. Agric. Food Chem. 60 (2012)

3234–3245.27] N. Mazzella, J. Molinet, A.D. Syakti, A. Dodi, P. Doumenq, J. Artaud, J.C.

Bertrand, J. Lipid Res. 45 (2004) 1355–1363.28] E.A. Vernooij, J.F. Brouwers, D.J. Crommelin, J. Sep. Sci. 25 (2002) 285–289.29] M. Koivusalo, P. Haimi, L. Heikinheimo, R. Kostiainen, P. Somerharju, J. Lipid

Res. 42 (2001) 663–672.30] R. Taguchi, J. Hayakawa, Y. Takeuchi, M. Ishida, J. Mass Spectrom. 35 (2000)

953–966.31] R.L. Shaner, J.C. Allegood, H. Park, E. Wang, S. Kelly, C.A. Haynes, M.C. Sullards,

A.H. Merrill Jr., J. Lipid Res. 50 (2009) 1692–1707.32] M.A. Masood, N. Salem Jr., Lipids 43 (2008) 171–180.33] Y. Rabagny, W. Herrmann, J. Geisel, S. Kirsch, R. Obeid, Anal. Bioanal. Chem.

401 (2011) 891–899.34] S. Olejnik, J. Li, S. Supattathum, C.J. Huberty, J. Educ. Behav. Stat. 22 (1997)

389–406.35] D.V. Yaroshenko, L.A. Kartsova, J. Anal. Chem. 69 (2014) 311–317.36] J. Folch, M. Lees, G.H. Sloane Stanley, J. Biol. Chem. 226 (1957) 497–509.37] C. Hu, H. Kong, F. Qu, Y. Li, Z. Yu, P. Gao, S. Peng, G. Xu, Mol. Biosyst. 7 (2011)

3271–3279.38] C. Hu, J. van Dommelen, R. van der Heijden, G. Spijksma, T.H. Reijmers, M.

Wang, E. Slee, X. Lu, G. Xu, J. van der Greef, T. Hankemeier, J. Proteome Res. 7(2008) 4982–4991.

39] F.F. Hsu, J. Turk, J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 877 (2009)2673–2695.

40] M.N. Barber, S. Risis, C. Yang, P.J. Meikle, M. Staples, M.A. Febbraio, C.R. Bruce,PLoS One 7 (2012) e41456.

41] A. Khedr, M. Hegazy, A. Kamal, M.A. Shehata, J. Sep. Sci. 38 (2015) 316–324.42] J.M. Wu, Y. Xu, N.J. Skill, H. Sheng, Z. Zhao, M. Yu, R. Saxena, M.A. Maluccio,

Mol. Cancer 9 (2010) 71.43] J.N. van der Veen, S. Lingrell, D.E. Vance, J. Biol. Chem. 287 (2012)

23418–23426.44] D.E. Vance, C.J. Walkey, Z. Cui, Biochim. Biophys. Acta 1348 (1997) 142–150.45] L. Tessitore, B. Marengo, D.E. Vance, M. Papotti, A. Mussa, M.G. Daidone, A.

Costa, Oncology 65 (2003) 152–158.46] G. van Meer, A.I. de Kroon, J. Cell Sci. 124 (2011) 5–8.

47] A. Skoura, T. Hla, J. Lipid Res 50 (Suppl) (2009) S293–S298.48] J.E. Vance, G. Tasseva, Biochim. Biophys. Acta 1831 (2013) 543–554.49] A. Yamaji-Hasegawa, M. Tsujimoto, Biol. Pharm. Bull. 29 (2006) 1547–1553.50] L. Cui, Y.H. Lee, Y. Kumar, F. Xu, K. Lu, E.E. Ooi, S.R. Tannenbaum, C.N. Ong,

PLoS Negl. Trop. Dis. 7 (2013) e2373.