Quantitative and Wide-Ranging Profiling of Phospholipids in Human

Anal. Chem. 2010, 82, 9858–9864

Quantitative and Wide-Ranging Profiling of Phospholipids in Human Plasma by Two-dimensional Liquid Chromatography/Mass Spectrometry Yoshiaki Sato,†,‡ Tatsuji Nakamura,†,‡ Ken Aoshima,†,‡ and Yoshiya Oda*,†,‡ Eisai Co., Ltd., 5-1-3 Tokodai, Tsukuba, Ibaraki 300-2635, Japan, and Core Research for Evolutional Science and Technology, Japan Science and Technology, Saitama 332-0012, Japan Normal-phase or reverse-phase liquid chromatography has been used in phospholipidomics for lipid separation prior to mass spectrometry analysis. However, separation using a single separation mode is often inadequate, as high-abundance phospholipids can mask large numbers of low-abundance lipids of interest. In order to detect and quantify low-abundance phospholipids, we present a novel two-dimensional (2D) approach for sensitive and quantitative global analysis of phospholipids. The methodology monitors individual glycerolipids and phospholipids through the use of a new quantitative normal-phase, solidphase extraction procedure, followed by molecular characterization and relative quantification using an ion-trap Orbitrap equipped with a reverse-phase liquid chromatograph, with data processing by MS++ software. The CV (%) of the peak area of each lipid standard was less than 15% with this extraction method. When the method was applied to a liver sample, we could detect more phosphatidylserine (PS) compared to the previous method. Finally, our developed method was applied to Alzheimer’s disease (AD) plasma samples. Several hundred peaks were detected from a 60 µL plasma sample. A partialleast-squares discriminant analysis (PLS-DA) plot using peak area ratio gave a unique group of PLS scores which could distinguish plasma samples of Alzheimer’s disease (AD) patients from those of age-matched healthy controls. Lipids have multiple roles, functioning as membrane components, energy stores and cell-structural components, and acting as regulators of complex signaling pathways, as well as being ligands and mediators of various protein-protein interactions and cell-cell interactions. Therefore, defects in lipid metabolism play a significant role in numerous human diseases, including obesity, diabetes, atherosclerosis, Alzheimer’s disease, Parkinson’s disease, cancer, and cardiovascular diseases.1-5 * To whom correspondence should be addressed. Phone: +1-978-837-4926. Fax: +1- 978-689-0543. E-mail: [email protected]. † Eisai Co., Ltd. ‡ Japan Science and Technology. (1) Wenk, M. R. Nat. Rev. Drug Discov 2005, 4, 594–610. (2) Watson, A. D. J. Lipid Res. 2006, 47, 2101–2111. (3) Steinberg, D. J. Lipid Res. 2005, 46, 179–190.

9858

Analytical Chemistry, Vol. 82, No. 23, December 1, 2010

Glycerophospholipids consist of a polar headgroup with a phosphate moiety and two fatty acids that are attached to the glycerol backbone. Dozens of structural variants are therefore possible within each class of lipids since the headgroup can be combined with a large pool of fatty acids that vary in both chain length and degree of saturation. This yields a multitude of similarly constructed molecules having highly diverse physical properties, which allows the living cell to regulate the intrinsic heterogeneity of membranes in a dynamic fashion.6 Quantitative characterization of individual lipids as well as of lipid classes is conventionally performed by a combination of various analytical technologies, including high-performance liquid chromatography, thin-layer chromatography, gas chromatography, and mass spectrometry. Mass spectrometric analysis of lipids in lipidomics can be divided into two complementary approaches, which employ either direct injection into the mass spectrometer (shotgun lipidomics)7,8 or a liquid chromatography (LC)9,10 or gas chromatography (GC)11,12 separation process prior to mass spectrometric analysis. The advantages of the shotgun approach are speed, robustness and capacity for automation. However, in the analysis of complex lipids, shotgun lipidomics is highly susceptible to strong ion suppression/enhancement effects resulting from competition by analytes for charge during the ionization process. Another problem is coelution of low-abundance analytes with high-abundance molecules. Moreover, the presence of many isomers with a narrow range of mass and isotopic ions makes quantitative measurement of PLs difficult. Therefore, LC separation followed by MS detection for species identification is very important in the analysis of lipids extracted from complex biological matrices. In lipidomics, information regarding retention time in LC is important to enable separate identification of isobaric (4) Hu, C.; van der Heijden, R.; Wang, M.; van der Greef, J.; Hankemeier, T.; Xu, G. J. Chromatogr., B 2009, 877, 2836–2846. (5) Fernandis, A. Z.; Wenk, M. R. J. Chromatogr., B 2009, 877, 2830–2835. (6) Ekroos, K.; Chernushevich, I. V.; Simons, K.; Shevchenko, A. Anal. Chem. 2002, 74, 941–949. (7) Han, X.; Gross, R. W. Mass Spectrom. Rev. 2005, 24, 367–412. (8) Han, X.; Yang, J.; Cheng, H.; Yang, K.; Abendschein, D. R.; Gross, R. W. Biochemistry 2005, 44, 16684–16694. (9) Homan, R.; Anderson, M. K. J. Chromatogr., B 1998, 708, 21–26. (10) Kim, H. Y.; Wang, T. C.; Ma, Y. C. Anal. Chem. 1994, 66, 3977–3982. (11) Dickens, B. F.; Ramesha, C. S.; Thompson, G. A., Jr Anal. Biochem. 1982, 127, 37–48. (12) Gaskell, S. J.; Brooks, C. J. J. Chromatogr. 1977, 142, 469–480. 10.1021/ac102211r  2010 American Chemical Society Published on Web 11/09/2010

molecular species,13-15 and the separation of low-abundance lipid species from high-abundance lipid species by LC also prevents the occurrence of ion suppression in MS. The situation is similar to that in the case of proteomics, where, in order to identify as many peptides as possible, high-abundance proteins, such as albumin, are removed prior to analysis.16,17 In lipidomics, one of the approaches to remove the high-abundance species is twodimensional (2D) separation prior to MS analysis. Although some groups have reported 2D separation methods prior to MS analysis of PLs,18,19 there is so far no quantitative, comprehensive, highthroughput method using 2D separation prior to MS analysis that is applicable to many PLs classes. Here, we describe a novel and reliable 2D-LC/MS approach for global, sensitive and quantitative analysis of phospholipids, and show that is suitable for detection and quantification of low-abundance phospholipids. Further, to demonstrate the utility of this method, we have applied it to biological samples from Alzheimer’s disease patients, and show that it can identify a disease-specific group of biomarkers. EXPERIMENTAL SECTION Chemicals and Reagents. Most of the reagents used in the experiments were of analytical grade and were purchased from Wako Pure Chemicals Co. (Osaka, Japan). All phospholipid standards were purchased from Avanti Polar Lipids (Alabaster, AL). Ethylenediaminetetraacetic acid (EDTA) and 28% aqueous ammonia were purchased from Sigma-Aldrich(Dorset, UK). Deionized water was obtained from a Milli-Q water system (Millipore, Milford, MA). Sample Collection. Male Sprague-Dawley (SD) rats were purchased from Charles River Laboratories Japan (Yokohama, Japan). They were anaesthetized and the liver was collected after perfusion by liver puncture. Animals were treated humanely and the experiment received prior ethical approval in accordance with company policy. Human plasma samples of age-matched controls and subjects diagnosed with AD and mild cognitive impairment (MCI) were purchased from PrecisionMed Inc., San Diego, CA and stored at -80 °C until sample preparation. Standard Sample Preparation. Standard stock solutions (1 mg/mL) were prepared in methanol. To prepare working sample solutions for LC/MS, individual metabolite stock solutions were mixed to a final concentration at 100 ng/mL, and a 20 µL aliquot was injected into the LC for the evaluation of chromatographic separation. Sample Preparation for LC/MS Analysis. The deep-frozen plasma samples were thawed on ice and 20 mg rat liver samples were homogenized in 200 µL sodium phosphate buffer containing 0.01% butylated hydroxytoluene (BTH). 60 µL of plasma or rat liver homogenate was spiked with 200 µL of internal standards (IS) solution (100 ng/mL of PC (12:0-12:0), PE (12:0-12:0), PG (13) Houjou, T.; Yamatani, K.; Imagawa, M.; Shimizu, T.; Taguchi, R. Rapid Commun. Mass Spectrom. 2005, 19, 654–666. (14) Taguchi, R.; Nishijima, M.; Shimizu, T. Methods Enzymol. 2007, 432, 185– 211. (15) Ikeda, K.; Oike, Y.; Shimizu, T.; Taguchi, R. J. Chromatogr., B 2009, 877, 2639–2647. (16) Ahmed, N.; Barker, G.; Oliva, K.; Garfin, D.; Talmadge, K.; Georgiou, H.; Quinn, M.; Rice, G. Proteomics 2003, 3, 1980–1987. (17) Bjorhall, K.; Miliotis, T.; Davidsson, P. Proteomics 2005, 5, 307–317. (18) Ogiso, H.; Suzuki, T.; Taguchi, R. Anal. Biochem. 2008, 375, 124–131. (19) Johanson, R. A.; Berry, G. T. Methods Mol. Biol. 2009, 579, 189–200.

(14:0-14:0), and PA (14:0-14:0) in methanol). The solution was extracted by means of the optimized Bligh and Dyer method. Briefly, 2 mL of CHCl3/methanol (3:1) and 0.5 mL of 18% NaCl aqueous solution were added to the samples, and mixed well. The solution was centrifuged for 10 min at 3,500 rpm. The lower organic phase was transferred to a new tube and extracted again with CHCl3. The combined organic solution was evaporated to dryness under a nitrogen gas stream at 40 °C, the residue was reconstituted in CHCl3 and the solution was used as a sample for fractionation of lipid classes. A 140 mg bed of PL-WAX (Agilent Technologies, Palo Alto, CA) packed in a pasteur pipet, was sequentially washed with water, 1 M HCl, water, 0.1 M NaOH, water, methanol, acetic acid, methanol, and chloroform. The sample was applied to the PL-WAX column and stepwise elution was begun with chloroform (3 × 1 mL) to wash out neutral lipids. Among the bound lipids, phosphatidylcholine and lyso-phosphatidylcholine were eluted with chloroform/methanol (95: 5; 3 × 1 mL), followed by the elution of phosphatidylethanolamine with chloroform/methanol (1: 1) (3 × 1 mL). The acidic lipids (e.g., PA, PS, PG, PI, and cardiolipin [CL]) were then eluted with chloroform/methanol/ 28% aqueous ammonia/acetic acid (50: 25: 1.17: 0.35) (3 × 1 mL). Each eluate was dried under nitrogen gas and the fractionated samples were stored at -30 °C until use. When required, each fraction was redissolved in 100 µL of methanol for LC/ESI-MS analysis. Comprehensive Analysis of Phospholipid with LC/MS. Phospholipids were analyzed with a Shimadzu 20AD system with a SIL-20AC autosampler, a CTO-20A column oven, and an LTQ Orbitrap mass spectrometer (ThermoFisher, San Jose, CA) with an ESI probe. For each run, a total of 20 µL of sample was injected onto a 2.1 i.d. × 150 mm Capcell Pack C18 column (Shiseido, Tokyo, Japan), prewashed with 2 mM EDTA solution, at a flow rate of 200 µL/min, with a total run time of 80 min. The gradient used consisted of solvent A (H2O: acetonitrile: methanol ) 4:4:2 containing 0.1% formic acid and 0.028% aqueous ammonia) and solvent B (isopropanol: methanol ) 8: 2 containing 0.1% formic acid and 0.028% aqueous ammonia), starting at 5% B, ramping to 60% B over 10 min, ramping to 80% B over 40 min, ramping to 100% B over 5 min, holding for 10 min, returning to 5% B in 0.1 min, and then holding for 15 min. The mass spectrometer was operated in the negative ion mode. The spray voltage was set at -4500 kV. A cycle of one full FT scan mass spectrum (400-2000 m/z, resolution of 30 000) followed by three data-dependent MS/MS and MS/MS/MS acquired in the linear ion trap with normalized collision energy (setting of 35%) was repeated continuously throughout each step of the multidimensional separation. Application of mass spectrometer scan functions and HPLC solvent gradients were controlled by the XCalibur data system. To assess the recovery and reproducibility of PL using this method, 60 µL of plasma was spiked with 40 pmol of PC (12: 0-12:0), PE (12:0-12:0), PG (14:0-14:0), and PA (14:0-14:0) in methanol, extracted as described above and fractionated using PL-WAX. Each fraction was redissolved in 100 µL of methanol and the recovery (%) of added PLs and reproducibility of some endogenous plasma PLs were analyzed with LC/ESI-MS on the same day (n ) 3). Analytical Chemistry, Vol. 82, No. 23, December 1, 2010

9859

Figure 1. Extracted LC/MS chromatograms of LPA, PA, and PS separated on Luna C18 column (A) and Capcell Pack C18 prewashed with EDTA solution (B).

Data Analysis and Statistical Analysis. The mass spectrometric data were acquired using Xcalibur, and the initial metabolomics profiling was performed using in-house-developed Mass++ data analysis software (http://groups.google. com/group/massplusplus) to obtain a peak list, align retention times and obtain peak areas normalized with IS. The obtained data matrix of patients’ plasma was used for statistical analysis, including multivariate analysis. The p-values and fold changes of all peaks were compared between groups using the R software package (http://www.r-project.org/). Partial leastsquares discriminant analysis (PLS-DA) was performed with SIMCAP, v11 (Umetrics, Kinnelon, NJ) using the 30 most important features to identify similarities and differences between sample groups based on the lipid profiles and to find the most important principal components. Peak area subtraction between ADs, MCIs and age-matched controls was done from mass chromatograms obtained with XCalibur for peaks selected by multivariate analysis as showing marked differences. Oneway analysis of variance (ANOVA) was performed to evaluate group differences. A difference of p < 0.05 was regarded as significant. RESULTS AND DISCUSSION Extraction and Fractionation of Phospholipids. To extract PLs from biological samples, Bligh and Dyer’s method20 or modifications of it have been used for many years. It is suggested that the normal Bligh and Dyer method is not suitable for extracting acidic PLs, especially LPA and phosphatidylinositol phosphates, due to poor recovery. To improve the recovery of those compounds, an acidic aqueous phase has been used in some laboratories.21,22 On the other hand, Ishida et al. noted that LPC and LPA are generated as hydrolytic artifacts under acidic conditions. So, there is no consensus (20) Bligh, E. G.; Dyer, W. J. Can. J. Biochem. Physiol. 1959, 37, 911–917. (21) Ogiso, H.; Taguchi, R. Anal. Chem. 2008, 80, 9226–9232. (22) Shan, L.; Jaffe, K.; Li, S.; Davis, L. J. Chromatogr., B 2008, 864, 22–28.

9860

Analytical Chemistry, Vol. 82, No. 23, December 1, 2010

as to an appropriate extraction method for acidic PLs.23 In this context, we examined the effects of various additives on recovery and reproducibility. Finally, we found that the use of high-salt aqueous solution or acidic solution improved the recovery, especially of LPA (Supporting Information (SI) 1). Considering the possibility that LPA and LPC might be formed under acidic conditions, we employed the high salt aqueous solution as the water phase in further studies. Depending on the lipids of interest, normal-phase liquid chromatography (NPLC) or reverse-phase liquid chromatography (RPLC) is usually coupled with an MS detector. NPLC and RPLC columns provide characteristic lipid separations. For instance, the separation of PLs on a NPLC column is mainly achieved based on the difference in the polar headgroup of the PLs, whereas separation on a RPLC column relies on the difference of chain length or of the number of double bonds of fatty acids (i.e., essentially on the lipophilicity). As noted already, high-abundance phospholipids can mask many lowabundance lipids of interest, so lipid class separation after extraction can be useful for isolation of targeted lipids. For example, Johanson et al. reported that a rapid separation step using a small column of a strong cation exchange (SCX) gel can be utilized to adsorb or capture cationic lipids from lipid extracts and provide improved MALDI-TOF MS signals of acidic phospholipid.19 However, lipid class separation alone is insufficient. Sommer et al. used initial fractionation on NPLC for class separation, followed by RPLC-MS to fully characterize individual lipids.24 However, class separation using NPLC is time-consuming and incomplete. In addition they did not address the measurement of low-abundance PLs, such as PA and PS. Ogiso et al. proposed DEAE cellulose purification for the analysis of large numbers of phospholipids, especially acidic phospholipids (PLs), prior to ODS-LC/MS analysis.18 However, they did not mention the reproducibility of quantitative measurement. Moreover, using their method, PEs emerge in the same fraction as PC, and (23) Ishida, M.; Imagawa, M.; Shimizu, T.; Taguchi, R. J. Mass Spectrom. Soc. Jpn. 2005, 53, 217–226. (24) Sommer, U.; Herscovitz, H.; Welty, F. K.; Costello, C. E. J. Lipid Res. 2006, 47, 804–814.

Table 1. Molecular Species of PS and PA Detected by Means of 2D-Phospholipidomics in Fraction Four of Rat Liver Samples molecular species PS PS PS PS PS PS PS PS PS PS PS PS PS PS PS PS PS PS PS PS PS PS PA PA PA PA PA PA PA PA PA PA PA PA PA PA PA PA PA PA PA PA PA

34:2 36:5 36:4 36:3 36:2 36:1 38:6 38:5 38:4 38:4 38:3 38:2 38:1 40:7 40:5 40:5 40:4 40:4 40:2 40:1 42:4 42:2 32:0 32:1 32:2 32:3 34:0 34:1 34:2 34:3 34:4 34:4 36:1 36:2 36:2 36:3 36:4 38:1 38:2 38:3 38:4 38:5 40:1

retention time (min)

observed m/z

theoritical m/z

∆(ppm)a

relative Areas

CV (%)b

28.2 26.1 27.7 29.4 31.6 34.6 27.1 28.0 31.0 34.9 32.7 35.2 38.9 27.3 31.4 32.8 34.0 35.3 38.9 43.4 37.3 43.1 32.0 29.0 26.8 24.7 36.5 32.4 29.7 27.1 25.1 26.4 37.0 33.0 33.7 29.8 29.4 41.4 37.6 34.9 33.1 29.5 46.0

758.4983 780.4837 782.4977 784.5145 786.5298 788.5457 806.4987 808.5143 810.5292 810.5273 812.5448 814.5604 816.5771 832.5143 836.5439 836.5441 838.5602 838.5605 842.5924 844.6084 866.5923 870.6238 647.4657 645.4502 643.4349 641.4201 675.4982 673.4814 671.4661 669.4504 667.4358 667.4358 701.5140 699.4975 699.4976 697.4813 695.4673 729.5451 727.5293 725.5129 723.4978 721.4817 757.5762

758.4972 780.4815 782.4972 784.5128 786.5285 788.5442 806.4972 808.5128 810.5285 810.5285 812.5441 814.5598 816.5755 832.5128 836.5441 836.5441 838.5598 838.5598 842.5911 844.6068 866.5911 870.6224 647.4652 645.4495 643.4339 641.4182 675.4965 673.4808 671.4652 669.4495 667.4338 667.4338 701.5121 699.4965 699.4965 697.4808 695.4651 729.5434 727.5278 725.5121 723.4964 721.4808 757.5747

1.5 2.8 0.7 2.1 1.7 2.0 1.9 1.8 0.9 -1.4 0.8 0.7 2.0 1.8 -0.3 0.0 0.5 0.9 1.6 2.0 1.4 1.6 0.8 1.1 1.6 3.0 2.5 0.9 1.4 1.3 2.9 2.9 2.7 1.5 1.6 0.7 3.1 2.3 2.1 1.1 1.9 1.3 2.0

0.54 0.49 6.12 0.37 8.56 8.84 4.85 3.60 59.77 0.47 2.98 0.59 0.76 1.28 3.21 0.62 2.82 0.27 0.47 0.53 0.10 0.16 0.10 0.17 0.19 0.02 0.06 0.26 3.63 0.42 0.05 0.05 1.87 2.97 0.30 1.01 3.48 0.10 0.11 0.61 15.48 1.85 0.05

0.7 4.7 0.4 1.4 0.8 0.4 0.6 1.7 0.6 0.6 3.2 1.5 4.7 1.7 0.3 1.1 0.1 4.0 3.8 10.1 1.2 10.4 5.4 0.2 4.6 7.8 3.0 1.7 0.2 3.5 5.9 2.7 0.1 8.4 4.3 1.9 0.9 2.2 0.6 0.0 1.0 1.0 0.6

previous method

i.d.c i.d.

i.d. i.d. i.d.

i.d.

a Difference between theoretical m/z and observed m/z. b Reproducibility of the peak area (n ) 3 studies). c i.d.: this molecular species were already identified in rat liver with the previous method.

Figure 2. Selective ion monitoring of some PS species from rat liver.

PCs and isotopic ions may interfere with the quantification of PEs on MS analysis. Therefore, aiming at comprehensive and quantitative

measurement of multiple PLs classes via 2D separation prior to analysis, we attempted to develop a procedure for 2D-phospholipidomics that would separate high-abundance neutral lipids and cationic PLs, such as TAGs and PCs, from low-abundance acidic PLs with high throughput. Anionic extraction cartridges are expected to be suitable for the separation of acidic compounds from neutral and basic compounds, so we screened several cation exchange cartridges, including DEAE Sephadex A-25 (GE Healthcare, Piscataway, NJ), Oasis WAX (Waters, UK) and PL-WAX, using PLs standards. We found that the PL-WAX gave the best and most reproducible separation with methanol/CHCl3 ) 5: 95 for the elution of PCs after the removal of neutral lipid, methanol/CHCl3 ) 50: 50 for the elution of PE, and methanol/CHCl3/acetic acid/28% aqueous ammonia ) 25: 50: 0.35: 1.17 for the elution of acidic PLs. Some of the acidic PL classes, such PA and PS classes, tend to elute as quite broad peaks under the reverse-phase LC conditions generally used in lipidomics studies.18 So, before the analysis of each fraction obtained from biological samples, we optimized the LC conditions using standard samples. Attempts have previously been made to improve the peak tailing. For example, Ogiso et al. developed reverse-phase LC conditions that reduce peak tailing of both PA and PS, by using a starting mobile Analytical Chemistry, Vol. 82, No. 23, December 1, 2010

9861

Table 2. Statistical Results of 2D-LC/MS Analyzed by R Software, Showing the Most Significant Features/ Components That Can Be Used for Class Discrimination fold change peak no. fr. no R.t. (min) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 a

2 2 2 2 2 2 2 2 2 2 2 2 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4 4

15.2 15.2 15.1 27.3 17.3 36.9 13.9 14.8 15.3 15.2 17.3 17.1 23.0 14.7 25.8 17.3 17.1 15.1 21.2 27.9 27.2 28.0 33.2 62.4 36.1 35.7 30.0 33.4 30.0 33.1 30.5

m/z 1603.0 1083.7 1627.0 738.5 1087.7 744.6 538.3 564.3 1107.7 1151.7 506.3 1155.7 548.4 996.6 1339.7 999.6 478.3 1515.0 629.4 835.5 909.5 927.5 947.5 1061.8 1303.8 1329.8 1723.1 1729.1 1744.1 1750.1 1771.1

control vs MCI control vs AD 0.19* 0.27**b 0.34* 0.19** 0.29* 0.36* 0.32* 0.42** 0.42* 0.39* 0.55* 0.47* 0.36** 0.38** 0.86 0.52** 0.49** 0.51* 0.48** 0.26** 0.43** 0.34** 0.57** 0.56** 0.62 0.64 0.35* 0.37** 0.42** 0.40** 0.58*

0.19*a 0.29** 0.32* 0.32** 0.34* 0.36* 0.37* 0.38** 0.43* 0.43* 0.45* 0.49* 0.32** 0.45** 0.46* 0.48** 0.45** 0.40** 0.45** 0.47* 0.46** 0.49* 0.48***c 0.46** 0.31** 0.33** 0.40** 0.37** 0.48** 0.40** 0.33**