Phospholipidomics reveals differences in glycerophosphoserine ...

Biochimica et Biophysica Acta 1828 (2013) 317–326

Contents lists available at SciVerse ScienceDirect

Biochimica et Biophysica Acta journal homepage: www.elsevier.com/locate/bbamem

Phospholipidomics reveals differences in glycerophosphoserine profiles of hypothermically stored red blood cells and microvesicles Beatriz Bicalho a, Jelena L. Holovati, a, b, Jason P. Acker a, b,⁎ a b

Canadian Blood Services, Research and Development, Edmonton, Alberta, Canada University of Alberta, Laboratory Medicine and Pathology, Edmonton, Alberta, Canada

a r t i c l e

i n f o

Article history: Received 1 September 2012 Received in revised form 12 October 2012 Accepted 25 October 2012 Available online 1 November 2012 Keywords: Stored RBC Microvesicles Phospholipids Orbitrap Shotgun mass spectrometry

a b s t r a c t During their normal in vivo life cycle erythrocytes (red blood cells, RBCs) undergo biochemical changes leading to membrane microvesiculation and shedding. RBC microvesiculation also occurs in vitro under conditions of blood bank storage, so microvesicles (MVs) accumulate in the storage (preservation) medium over storage time. Considerable effort has been put into gaining a mechanistic understanding of the RBC microvesiculation process, as this is crucial to better understand RBC biology in disease and in health. Additionally, MVs accumulated in stored RBCs have been implicated in transfusion adverse inflammatory reactions, with chloroform extractable compounds, thus lipophilic, known to trigger the effect. However, because thin layer chromatography resolution of RBC and MV lipids has always enabled one to conclude high compositional similarities, in depth analysis of MV lipids has not been extensively pursued. Here we present an orbitrap mass spectrometry (MS) approach to compare the phospholipid composition of RBCs and MVs from leukoreduced, hypothermically (2–6 °C) stored RBC units. We used shotgun MS analysis and electrospray ionization (ESI) intra-source separation, and demonstrated high similarity of compositional profiles, except for glycerophosphoserines (PS). Contrasting abundances of PS 38:4 and PS 38:1 characterized MV and RBC profiles and suggested that storage-associated microvesiculation possibly involves shedding of specific membrane rafts. This finding indicates that phospholipidomics could likely contribute to a better understanding of the RBC microvesiculation process. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Storage lesion, a term now in use for almost half century [1], describes the physical and chemical changes observed in red blood cells (RBCs) stored for transfusion, and has been consistently applied to refer to metabolic depletion, membrane loss of asymmetry and membrane shedding through microvesiculation [2]. This latter aspect of the RBC storage lesion, i.e. microvesiculation, has been a meeting point for a variety of research areas (e.g. biopreservation, aging, diseases, signaling) seeking insight into RBC lipid–protein physiology [3–5]. As for being fit into a liquid-like structure concept including long-range and short-range orders dynamically displayed in various lengths and time scales [6], the RBC membrane is still complex to current understanding, especially with reference to the microvesiculation process. In spite of other intrinsic factors [7], spectrin (RBC skeleton lattice protein) has been implicated in sustaining RBC membrane asymmetry through selective interactions with glycerophosphoserines (PS)

⁎ Corresponding author at: Canadian Blood Services, Research and Development, 8249 114 Street, Edmonton, Alberta, Canada T6G 2R8. Fax: + 1 780 702 8621. E-mail address: [email protected] (J.P. Acker). 0005-2736/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.bbamem.2012.10.026

[8,9]. Accordingly, structural damage to spectrin (e.g. oxidation), capable of compromising PS anchoring sites, is postulated to promote diffusion of PS to the outer membrane [10,11]. Once PS is exposed, the membrane becomes thrombogenic [12] and pro-inflammatory [13]. Increased PS exposure has been observed in membrane-derived microvesicles (MVs) from fresh [14,15] and hypothermically-stored [16] RBC preparations. In the RBC itself, membrane asymmetry is not significantly altered after microvesiculation [17], even during hypothermic storage [15]. For this reason, RBC-derived MVs have been postulated to be involved in transfusion related adverse reactions [18,19]. Moreover, accumulation of lipids with platelet activating factor (PAF)-like activity in the preservation medium (citrate–phosphate– dextrose–adenine) of whole blood and packed RBC units have been reported [19,20] and postulated to be relevant to transfusion related acute lung injury (TRALI) [21]. Their preliminary identification as lyso-glycerophosphocholines (LPC) [22] has been argued since accumulation of LPCs was shown to be blood plasma dependent [23] and hence unable to explain the PAF-like activity in the supernatant of stored, leukoreduced, RBC units [24]. More recently, a few pro-inflammatory oxidation products of arachidonic acid, mainly 5-hydroxyeicosatetraenoic acid (5-HETE), were reported to accumulate in the supernatant fraction of leukoreduced RBCs [25]. The PAF-like activity of leukoreduced RBC supernatants was shown to be

318

B. Bicalho et al. / Biochimica et Biophysica Acta 1828 (2013) 317–326


significantly diminished after 10,000 ×g centrifugation, indicating that the bioactive compounds were linked to membrane MVs [26]. Apart from the contributions cited above, lipid analysis of RBC MVs has been conducted using lipid group separation on thin layer chromatography prior to spectrophotometric determinations of cholesterol-to-phospholipid ratios [27–30], and these studies have repeatedly demonstrated that RBCs and derived MVs are compositionally similar. Interestingly, RBCs and derived MVs have never been compared from the perspective of phospholipid species distribution, despite it being acknowledged that membrane protein sorting [31] occurs during the onset of RBC microvesiculation. From a clinical point of view, recognizing membrane compositional patterns of transfusional RBCs and related MVs is important for establishing measurable membrane compositional parameters in reference to product quality that may contribute to assessing preservation strategies and broadening current understanding of the physical changes displayed by RBCs during storage. This present study was undertaken with the aim of comparing the phospholipid composition of blood bank hypothermically (2–6 °C) stored, leukoreduced, SAGM (saline–adenine–glucose–manitol) preserved RBCs and derived MVs. A non-targeted phospholipidomics approach was used, relying on a Fourier transform orbitrap mass spectrometer, whose level of performance with respect to mass resolution and accuracy has been tutorially described recently [32]. It was apparent that compositional differences in the membrane distribution of PSs exist between RBCs and MVs. 2. Materials and methods 2.1. Chemicals Organic solvents were HPLC grade. Chloroform was purchased from EMD Chemicals Inc. (Darmstadt, Germany), methanol and 2-propanol purchased from Sigma-Aldrich (St. Louis, MO, USA). Phosphate buffered saline (PBS 1×, without calcium and magnesium) was from Mediatech Inc. (Manassas, VA, USA). Phosphoric acid (H3PO4) was from EMD Chemicals Inc. and sodium carbonate (Na2CO3) was from Sigma-Aldrich. Aqueous solutions were prepared with deionized water. Dimyristoylglycerophosphoethanolamine (PE 28:0), used as an internal standard, was purchased from Avanti Polar Lipids (Alabaster, AL, USA). Standard ion calibration solutions (Pierce® LTQ) for electrospray ionization in positive and negative modes were purchased from Thermo Scientific (Rockford, IL, USA). 2.2. Red blood cell Whole blood (450 mL ± 10%) was collected from healthy volunteer donors into polyvinyl chloride blood collection packs containing 70 mL ± 10% citrate–phosphate–dextrose anticoagulant (Macopharma, Mouvaux, France) at Canadian Blood Services (Vancouver, BC, Canada). Whole blood units were rapidly placed on butanediol cooling plates (Fresenius CompoCool) and processed for RBC extraction within 24 h of post-collection. After separation of buffy coat (47.5 mL± 4%) and plasma (275 mL±7%) by centrifugation at 22 °C (3500 rpm, HBB-6 rotor, Sorval RC3BP, ThermoScientific), RBCs were suspended in 110 mL of SAGM additive solution and leukoreduced by filtration at room temperature. Resulting RBC units (284±5%) were hypothermically stored (2–6 °C) under standard blood bank conditions. Six RBC units stored for 53±4 days were

319

analyzed, as outdated units were expected to be enriched in MVs. Donors were females and males of diversified serotypes.

2.3. Separation of RBCs and microvesicles SAGM preserved RBCs (average hematocrit = 64% ± 2.9%) were centrifuged at 3200 ×g for 30 min, at 4 °C, in 50 mL polypropylene conical tubes. Supernatants resulting from this first step centrifugation were transferred to new 50 mL polypropylene conical tubes and centrifuged (3200 ×g, 30 min, 4 °C) for acellular debris clean-up. Cleaned-up supernatants were subjected to ultracentrifugation (50,000 ×g, 60 min, 4 °C; rotor JA-25.50, Beckman Coulter) and MVs obtained as bright red pellets. These were collected in phosphate buffered saline (PBS 1 ×), volume normalized to 0.75 mL, and stored at − 20 °C until lipid extraction for analysis. RBCs resulting from the first step centrifugation were sampled (500 μL) and washed (2 ×, 2500 ×g, 5 min, 4 °C) with 500 μL cold (4 °C) PBS 1 × and then stored at − 20 °C until lipid extraction for analysis.

2.4. Lipid extraction Sample aliquots of PBS-washed RBCs (100 μL) and PBS-suspended MVs (100 μL) were transferred to 10 mL borosilicate disposable glass tubes (100 × 10 mm i.d.) containing 2 μg (20 μL of 0.1 mg/mL) of PE 28:0 (in CHCl3–MeOH, 2:1, v/v) as an internal reference. Samples were acidified with 1 mL 5 mM aqueous H3PO4, and vortex-mixed (30 s) with 1 mL 2-propanol and 3 mL CHCl3. Then, 0.5 mL 25 mM aqueous Na2CO3 was added and vortex-mixed (30 s), bringing the final extraction medium to pH 7 (pH indicator strips, EMD Chemicals, Gibbstown, USA). Tubes were centrifuged (2200 ×g, 10 min, 4 °C) and resulting bottom phases transferred to 4 mL borosilicate glass vials (45 × 12 mm i.d.), which were heated (55 °C) for solvent evaporation under gentle nitrogen stream. Solvent-free extracts were redissolved in 0.3 mL MeOH–CHCl3–20 mM aqueous ammonium acetate (4:1:1, v/v/v) and refrigerated (− 20 °C) until analysis.

2.5. Shotgun mass spectrometry Lipid extracts were infused (3 μL/min) into an Ion Max electron spray ionization (ESI) source mounted on a LTQ Orbitrap XL (Thermo Scientific, San Jose, CA, USA). The orbitrap mass analyzer was calibrated with standard ion calibration solutions (Pierce LTQ, Thermo Scientific) and compound dependent instrument parameters were tuned using a 10 μg/mL PE 28:0 solution (MeOH–CHCl3–20 mM aqueous ammonium acetate, 4:1:1, v/v/v) either for electrospray positive (ESI+) or electrospray negative (ESI −) ionization modes. Acquisition parameters were as follows: source temperature (ESI + and ESI −) = 350 °C, source voltage (ESI + and ESI −) = 3.5 kV, capillary voltage = 13 V (ESI +) and − 14 V (ESI−), tube lens voltage = 90 V (ESI+) and − 130 V (ESI −), and sheath gas flow = 20 (arbitrary units, ESI + and ESI −). Full scan acquisitions lasting 1 min/sample were performed in the m/z 500–900 range, using a nominal resolving power of 60,000 FWHM (full-width at half-maximum peak height at m/z 400). Product ion scan acquisitions in the range of m/z 50–850 were performed for fatty acid analysis in ESI − by applying a collision energy of 35 (arbitrary units) for higher energy collision dissociation (HCD). Product ions were recorded for 1 min/sample.

Fig. 1. Representative shotgun (1 min/run) mass scan (m/z 600–900) profiles of red blood cells (RBC) and associated microvesicle (MV) lipids acquired with an orbitrap instrument using electrospray ionization (ESI) in positive (ESI+) and negative (ESI−) modes. Dimyristoyl-glycerophosphoethanolamine (PE 28:0) was spiked into samples before lipid extraction as an internal reference. Mass resolution is given as R, and accuracy values in ppm. Choline-containing phospholipids were detected as positively charged species in ESI+ whereas serine and inositol-containing phospholipids were detected as negatively charged in ESI−. Ethanolamine-containing phospholipids were detected in both ionization modes.

320


2.6. Data processing Recorded data were analyzed using Xcalibur 2.1 software (Thermo Scientific). Fourier transform full scan mass spectra resulting from 0.1 to 1.0 min acquisitions were analyzed for the presence of compounds containing at maximum 50 carbon ( 12C), 100 hydrogen ( 1H), 15 oxygen ( 16O), 2 nitrogen ( 14N) and 2 phosphorous ( 31P) atoms. The spectra, converted into lists relating m/z, peak intensity, and the most accurate mass-compatible elemental composition, were aligned using Excel tools. The peak intensities were normalized to the internal standard and the characterized phospholipids were assessed for relative abundance. 2.7. Statistical analysis All data were expressed as mean ± standard deviation (SD) from 6 independent RBC units and were subjected to two-tailed, unequal variance Student's test. A probability (p) value b 0.05 was considered statistically significant. 3. Results 3.1. Untargeted phospholipidome profiles Just prior to MV pelleting, the MV containing RBC supernatants were assessed for particle size using dynamic light scattering. Mixed populations of particle sizes were observed but none exceeding 280 nm in diameter (results not shown). Acquisition of ESI + and ESI − full scan (m/z 500–900) mass spectrometry (MS) lipidome profiles of outdated RBCs and associated MVs was straightforward. Representative samples of the spectra obtained are shown in Fig. 1. Choline-containing phospholipids [glycerophosphocholines (PC) and sphyngomyelins (SM)] were analyzed as positively charged in ESI +, while serine-containing species (PS) were analyzed as negatively charged in ESI −. Ethanolamine-containing phospholipids were detected in either ionization mode (Fig. 1, inserts: PE 28:0), although analyzed with superior sensitivity in ESI −. As indicated by the IS peak (Fig. 1, inserts), achieved resolution (R, given as full width at half peak maximum) and mass accuracy (given in parts per million, ppm) enabled the expected elemental composition assignments (C33H67O8NP [M + H] +; C33H65O8NP [M − H] −) for the correct

identification of PE 28:0 in both ionization modes (mean accuracy ± SD, n = 12, ESI +: − 1.0 ± 0.4 ppm; ESI −: 2.2 ± 0.5 ppm). Throughout the lipidome profiles, R ranged within 50,000–55,000 and accuracy within limits of ± 3.0 ppm. 3.2. Main elemental composition categories Acquired m/z peaks were grouped in elemental composition categories classified by the number of N, O and P atoms present in the most accurate molecular formula calculated (XCalibur software). They were then assessed for relative abundance (%) in the context of signal intensity. The following combinations of N, O and P were consistently observed: NO7P, NO8P, N2O6P, NO9P, NO10P, O13P; and these were correlated to the following phospholipid head groups: in ESI+, NO7P = PC sn-1 ethers (PC-o), NO8P = PC, and N2O6P = SM; in ESI −, NO7P = PE plasmalogens (PE-p), NO8P = PE, NO9P = oxidized PE (PEox), NO10P = PS, and O13P = glycerophosphatidylinositol (PI). These are illustrated in Fig. 2, with chemical formula assigned on the horizontal axis referring to charged molecular ions {[M + H] +, [M + H] −} in which carbon (C) and hydrogen (H) atoms subscripts “x” and “y” stand for odd numbers respectively, and “w” and “z” subscripts stand for even numbers, respectively. Within assigned categories, C and H numbers varied according to the length and degree of unsaturation (number of double bonds) of the phospholipidassociated acyl chains. 3.3. ESI + acquired phospholipids Overall, PCs (CwHyNO8P, Fig. 2) and PC-os (CwHyNO7P, Fig. 2) were assigned with − 2.2 ± 2.0 ppm and −3.4 ± 0.4 ppm accuracy, respectively. Their relative compositional profiles in MVs and RBCs are presented in Figs. 3 and 4. PC and PC-o species ranging from 32 to 40 total acyl chain C atoms bearing up to 8 double bonds were determined. PCs were similarly distributed in RBCs and MVs (Fig. 3). The species PC 34:1 (m/z 760.6) and PC 34:2 (m/z 758.6) were relatively predominant, and individually accounted for 20% of total cholinecontaining phospholipids in RBCs and MVs. PC-o species of 34 total acyl chain C atoms were also predominant in the PC-o profile of MVs and RBCs (Fig. 4). The distribution of a few PC-o species (PC-o 36:2, PC-o 38:2, PC-o 40:3) towards a higher abundance in RBCs relative to MVs can be observed in Fig. 4, although this is hard to

Fig. 2. Relative distribution of FTMS signal intensity per elemental composition category subsequently assigned to a phospholipid group in red blood cells (RBC) and microvesicles (MV). Chemical formula in horizontal axis refers to charged molecule ions {[M + H]+, [M + H]−} in which carbon (C) and hydrogen (H) subscripts “x” and “y” stand for an odd number, while subscripts “w” and “z” stand for an even number. Correspondingly assigned phospholipid groups are shown in parenthesis; (PC-o) glycerophosphocholine sn-1 ethers, (PC) glycerophosphocholine, (SM) sphingomyelin, (PE-p) glycerophosphoethanolamine plasmalogens, (PE) glycerophosphoethanolamine, (PE-ox) glycerophosphoethanolamine oxidized, (PS) glycerophosphoserine, and (PI) glycerophosphoinositol. Values are mean + SD from 6 different RBC units.


321

Fig. 3. Glycerophosphocholine (PC) species in red blood cells (RBC) and related microvesicles (MV) plotted by their relative abundance within the PC group. Values are mean + SD (n = 6 different RBC units).

interpret due to the low abundance of total PC-os in the general compositional context (PC-os = 1% of PCs, Fig. 2). SMs (CxHzN2O6P) were assigned with an error of −1.1 ± 0.8 ppm, assuming all the species were of the dihydroxy C18:1 sphingoid base type, which are known to predominate in human blood [33]. SMs in the range of 14–24 acyl chain C atoms were determined (Fig. 5), with the major species SM 16:0 (m/z 703.6), SM 24:0 (m/z 815.7) and SM 24:1 (m/z 813.7) accounting altogether for 20% of the total signal intensity attributed to choline-containing phospholipids in RBCs and MVs. 3.4. ESI − acquired phospholipids PE (CxHyNO8P, Fig. 2) and PE-p (CxHyNO7P, Fig. 2) species were assigned with accuracies of −1.5 ±0.5 ppm and −2.0 ± 0.6 ppm, respectively. In contrast to the ether counterpart PC-os, PE-ps accounted for 30% of total ESI− assigned phospholipids in either RBCs or MVs, and were as abundant as 46% PEs in RBCs and as much as 80% of PEs in MVs (Fig. 2). The observed disproportions resulted from the differences in signal intensity displayed by total PEs in the ESI− profiles of

MVs and RBCs (Fig. 2). Illustrated in Fig. 6 is the relative distribution of the assigned PE species in MVs and RBCs, which ranged from 34 to 40 total acyl chain C atoms and were determined containing up to total 6 unsaturations (double bonds). PE 34:1 (m/z 716.5) predominated over other PE species both in RBCs and MVs (Fig. 6), although relatively, these were even more abundant in RBCs (35% of PEs) than in MVs (25% of PEs). PE 34:1 accounted greatly for the observed differences in total RBC and MV PE intensities reported in Fig. 2. Apart from this observation, polyunsaturated (> 2 double bonds) PEs were, in general, slightly enriched in MVs relative to the RBCs (Fig. 6). Interestingly, among very low abundant species, which contributed to a maximum of 0.2% of the total signal intensity displayed by the PE group profile, odd-acyl chain PEs (PE 35:1, m/z 730.5; PE 35:2, m/z 728.5; PE 37:5, m/z 750.5; PE 37:6, m/z 748.5, Fig. 6 — inserts) were mostly enriched in RBCs. Determined PE-p species are presented in Fig. 7. These ranged from 36 to 40 total acyl chain C atoms, were predominantly polyunsaturated and similarly distributed in RBCs and MVs. Compounds grouped in the elemental composition category CxHyNO9P (Fig. 2) were assigned with an error of − 2.8 ± 0.4 ppm

Fig. 4. Glycerophosphocholine sn-1 ether (PC-o) species in red blood cells (RBC) and microvesicles (MV) plotted by their relative abundance within the PC-o group. Values are mean + SD (n = 6 different RBC units).

322


Fig. 5. Sphingomyelin (SM) species in red blood cells (RBC) and microvesicles (MV) plotted by their relative abundance within the SM group. Values are mean + SD (n = 6 different RBC units).

and attributed as oxidized PE species (PE-ox). These accounted for 2% (1.7% in MVs, 1.9% in RBCs) of the assigned ESI − acquired phospholipids (Fig. 2) and varied in the range of 34–44 total acyl chain C atoms (Fig. 8). Lower molecular weight PE-ox ranging from 34 to 36 total acyl chain C atoms [PE-ox 34:0 (m/z 732.5), PE-ox 34:1 (m/z 730.5), PE-ox 36:1 (m/z 758.5)] were enriched in the RBCs whereas PE-ox of higher molecular weight (>42 total acyl chain Cs) were enriched in the MVs. PSs (CwHyNO10P, Fig. 2) ranging from 36 to 44 total acyl chain C atoms (Fig. 9) were assigned with an error of 1.7 ± 0.7 ppm. In MVs, 23 ± 5.0% of the total assigned ESI − signal intensity was due to PSs, whereas in RBCs the same assigned PSs accounted for only 3 ± 1.2% of the attributed ESI − signal intensity (Fig. 2). As shown in Fig. 9, the compositional profile of MV PSs is characterized by the predominance of PS 38:4 (m/z 810.5) over PS 40:6 (m/z 834.5), PS 40:5 (m/z 836.5) and PS 40:4 (m/z 838.5), whereas in the RBCs, PS 38:1 (m/z 816.5) predominated over PS 38:4, PS 40:1 (m/z 844.6) and PS 40:0

(m/z 846.6). Differences in the relative abundance of PS 38:4, PS 38:1 and PS 40:6 in RBCs and MVs were statistically significant (p b 0.001). Insight into the acyl chain composition of the predominant PSs in MVs was gained through analysis of their product ions. As shown in Fig. 10, PS 38:4 contained a stearoyl chain (C18:0; m/z 283) and an eicosatetranoyl (arachidonoyl) chain (C20:4; m/z 303) respectively linked to the sn-1 (indicative fragment ions: m/z 419, 437) and sn-2 (indicative fragment ions: m/z 439) positions of the PS backbone. Similarly, Fig. 11 shows that PS 40:6 was predominantly formed by a stearoyl chain (m/z 283, C18:0) at position sn-1 (m/z 419, 437) and a docosahexanoyl chain (m/z 3237, C22:6) at position sn-2 (m/z 463) of the PS backbone. For the remaining MV PSs (Fig. 9), which were present in relatively low abundance, the product ion spectra obtained through shotgun MS were not as clear, and thus inconclusive. PIs (CxHzO13P, Fig. 2) were minor sample constituents accounting for only 1±0.6% of assigned ESI− acquired phospholipids in MVs and for less than 0.1% (0.02±0.01%) of assigned ESI− acquired phospholipids

Fig. 6. Glycerophosphoethanolamine (PE) species in red blood cells (RBC) and microvesicles (MV) plotted by their relative abundance within the PE group. Values are mean + SD (n = 6 different RBC units).


323

Fig. 7. Glycerophosphoethanolamine plasmalogen (PE-p) species in red blood cells (RBC) and related microvesicles (MV) plotted by their relative abundance within the PE-p group. Values are mean + SD (n = 6 different RBC units).

in RBCs. These amounts were substantially due to PI 38:4 (m/z 885.5), which was the only PI species detected in the RBCs and the predominant one (0.9%) in the MVs, in which trace amounts (0.02–0.05% of ESI− assigned species) of PI 38:5 (m/z 883.5), PI 36:2 (m/z 861.5) and PI 36:3 (m/z 859.5) were also determined. Since PIs are polyhydroxylated and expectedly low abundant phospholipids, these results suggest differences in extraction efficiency.

4. Discussion The analysis was performed relying on the hypothesis that RBCderived MVs constituted the predominant population of MVs in the hypothermically stored RBC units used in this study. The phospholipidomics approach used in this paper has not been previously applied to characterize stored RBCs produced under blood bank conditions. A similar analytical procedure was applied to the characterization of RBCs collected and treated outside the context of component processing and blood bank storage [34]. The phospholipid profile described by Leidl

et al. [34] is representative of a fresh (drawn>washed>analyzed) RBC preparation and is referred to in the following text as “fresh RBCs”. Non-targeted ESI-FTMS phospholipidome profiles of leukoreduced, SAGM preserved, hypothermically stored (53 ± 4 days) RBCs and associated MVs were compared in order to shed light onto the membrane compositional features of outdated blood bank-stored RBCs. On the basis of intrasource separation (Fig. 1), high mass resolution and high accuracy, we observed great similarity between RBCs and MVs in terms of their relative compositional abundance of PCs and SMs (Figs. 2, 3, 5). Observed PC and SM profiles closely resembled the PC and SM profiles previously reported for fresh RBCs [34]. The microvesiculation phenomenon apparently did not disturb PC and SM RBC membrane organization. Regarding PC-os (Fig. 4), profile differences were observed for the distribution of PC o-36:2, PC o-38:2, and PC o-40:3, which were enriched in RBCs. Further studies are required to better understand this finding. An interesting aspect of PC-os is the basic structure relationship they hold with PAF (1-O-palmityl-2acetyl-glycerophosphocholine), which might explain their relative downregulation in the biosystem. Nevertheless, presently characterized

Fig. 8. Oxidized glycerophosphoethanolamine (PE-ox) species in red blood cells (RBC) and related microvesicles (MV) plotted by their relative abundance within the PE-ox group. Values are mean + SD (n = 6 different RBC units).

324


Fig. 9. Glycerophosphoserine (PS) species in red blood cells (RBC) and related microvesicles (MV) plotted by their relative abundance within the PS group. Values are mean + SD (n = 6 different RBC units). *pb 0.05, **p b 0.01, ***p b 0.001.

PC-os should not per se be expected to elicit any PAF-like response in so far as their sn-2 position remains occupied by a long chain acyl group [35]. Apart from the assigned highly unsaturated PC-o species (PC o-36:7, PC o-36:8, PC o-38:8), remaining PC-os have been previously described in fresh RBCs [34]. The PE profile of stored RBCs (Fig. 6) resembles the PE profile of fresh RBC [34], although PE 34:1 is relatively enriched in the stored cells. PE 34:1 is also abundant in the MVs, though these are relatively enriched in polyunsaturated PEs. Conjugated double bonds imprint antioxidant capacity and also account for the rheological properties of the erythrocyte membrane [36]. The highly polyunsaturated PE-ps (Fig. 7) were also relatively enriched in the MVs, as in the groupphospholipidome profiles (Fig. 2), PE-ps presented a close to unit

abundance relationship with PEs in the MVs, whereas in RBCs this ratio dropped to half. PE-ps have been largely acknowledged for their biologically relevant antioxidant capacity [37]. On the other hand, when polyunsaturated phospholipids exert an antioxidant function, they result in the formation of oxidized polyunsaturated phospholipids: a potential source of pro-inflammatory lipid mediators [38,39]. PE-p 38:4, likely a combination of p-18:0/20:4 acyl groups, and therefore a source of arachidonic acid, was the most abundant PE-p in the stored RBCs and MVs, as well as in previously analyzed fresh RBCs [34]. The majority of assigned PE-oxs were polyunsaturated (Fig. 8) and also enriched in the MVs. Their elemental composition (CxHyNO9P) suggests two most likely interpretations: their correspondence to PE-p peroxides and/or to PE epoxides. Co-existence of both

Fig. 10. Product ion spectrum of PS 38:4 presenting evidence for the presence of a sn-1 stearoyl (C18:0) and a sn-2 eicosatetranoyl (C20:4, arachidonoyl) acyl moieties.


325

Fig. 11. Product ion spectrum of PS 40:6 presenting evidence for the presence of a sn-1 stearoyl (C18:0) and a sn-2 docosahexanoyl (C22:6) acyl moieties.

possibilities is actually reasonable in view of the acyl chain length and degree of unsaturation of the PE-oxs (Fig. 8). Their relative abundances, however, did not suffice to yielding conclusive product ion spectra in the shotgun MS approach. Further work will address complementary structural information. That oxidized phospholipids were observed was not surprising, since oxidative stress has been consistently observed to increase with storage duration in hypothermically (2–6 °C) stored RBCs [40]. The most intriguing outcome of our analysis is probably the observed differences in PS distribution in the stored RBCs and MVs (Fig. 9). The PS profile observed for MVs, in which the arachidonoyl-PS (PS 38:4, Fig. 10) predominates, is in perfect agreement with the PS profile of fresh RBCs [34]; in contrast the PS profile observed for stored RBCs, in which PS 38:1 predominates, disagrees with the profile of fresh RBCs, as PS 38:1 was found to be among the least abundant PS species in the latter [34]. Microvesiculation has interfered in the RBC PS composition to an extent not observed for the other phospholipids analyzed. Our results indicate an apparent preferential membrane shedding of polyunsaturated PSs to RBC MVs. On that basis we suggest that polyunsaturated PSs would be likely enriched in membrane rafts rather than evenly scattered in the inner layer. The involvement of lipid rafts in RBC microvesiculation has already been considered [41] since stomatin has been shown to accumulate in RBC MVs during hypothermic storage [12]. Therefore, phospholipidomics could likely contribute to a better understanding of the RBC microvesiculation process. Studies on RBC biopreservation could also benefit from phospholipidomics since in this field close control of the RBC membrane composition is required.

5. Conclusion This paper reported compositional differences between stored RBCs and associated MVs in terms of PSs, which suggested the onset of lipid bilayer compositional remodeling during microvesiculation, possibly in consequence of shedding membrane rafts enriched in polyunsaturated PSs. It was also shown that in stored, outdated

RBCs the PS profile no longer resembled the PS profile of fresh RBCs. In general, MVs were shown to be relatively enriched in polyunsaturated phospholipids. Future work will address i) whether observed PS profiles are generalizable, ii) how they change during the time course of hypothermic storage and iii) how they are influenced by different manufacturing methods used in modern blood banks. Acknowledgements The authors would like to thank volunteer blood donors, without whom this research would not have been possible. Dr. Richard Fahlman (Institute of Biomolecular Design, University of Alberta) allowed access to the orbitrap instrumentation, and Dr. Jason Dyck (Cardiovascular Research Group, University of Alberta) allowed access to the ultracentrifuge, so authors would like to express their gratitude. The authors also thank Dr. Geraldine Walsh, CBS Scientific Writer, for assistance with manuscript editing. This work has been funded by CIHR/CBS/NSERC. References [1] S. Rimon, F. Brok, A. Rimon, Changes in erythrocyte lipids during storage, Clin. Chim. Acta 12 (1965) 22–26. [2] E. Bennett-Guerrero, T.H. Veldman, A. Doctor, M.J. Telen, T.L. Ortel, T.S. Reid, M.A. Mulherin, H. Zhu, R.D. Buck, R.M. Califf, T.J. McMahon, Evolution of adverse changes in stored RBC, Proc. Natl. Acad. Sci. 104 (2007) 17063–17068. [3] N. Lion, D. Crettaz, O. Rubin, J.-D. Tissot, Stored red blood cells: a changing universe wainting for its map(s), J. Proteomics 73 (2010) 374–385. [4] E.M. Pasini, H.U. Lutz, M. Mann, A.W. Thomas, Red blood cell (RBC) membrane proteomics — part I: proteomics and RBC physiology, J. Proteomics 73 (2010) 403–420. [5] A.G. Kriebardis, M.H. Antonelou, K.E. Stamoulis, E. Economou-Petersen, L.H. Margaritis, I.S. Papassideri, RBC-derived vesicles during storage: ultrastructure, protein composition, oxidation, and signalling components, Transfusion 48 (2008) 1943–1953. [6] K. Jacobson, O.G. Mouritsen, R.G.W. Anderson, Lipid rafts: at a crossroad between cell biology and physics, Nat. Cell Biol. 9 (2007) 7–14. [7] M. Bitbol, C. Dempsey, A. Watts, P.F. Devaux, Weak interaction of spectrin with phosphatidylcholine–phosphatidylserine multilayers: a 2H and 31P NMR study, FEBS Lett. 244 (1989) 217–222.

326


[8] X. An, X. Guo, H. Sum, J. Morrow, W. Gratzer, N. Mohandas, Phosphatidylserine binding sites in erythroid spectrin: location and implications for membrane stability, Biochemistry 43 (2004) 310–315. [9] J.M. Thompson, R.E. Ellis, E.M. Green, C.P. Winlove, P.G. Petrov, Spectrin maintains the lateral order in phosphatidylserine monolayers, Chem. Phys. Lipids 151 (2008) 66–68. [10] D. Bratosin, J. Mazurier, J.P. Tissier, J. Estaquier, J.J. Huart, J.C. Ameisen, D. Aminoff, J. Montreuil, Cellular and molecular mechanisms of senescent erythrocyte phagocytosis by macrophages. A review, Biochimie 80 (1998) 173–195. [11] M. Girasole, G. Pompeo, A. Cricenti, G. Longo, G. Boumis, A. Belleli, S. Amiconi, The how, when, and why of the aging signals appearing on the human erythrocyte membrane: an atomic force microscopy study of surface roughness, Nanomedicine 6 (2010) 760–768. [12] U. Salzer, R. Zhu, M. Luten, H. Isobe, V. Pastushenko, T. Perkmann, P. Hinterdorfer, G.J.C.G.M. Bosman, Vesicles generated during storage of red cells are rich in the lipid raft marker stomatin, Transfusion 48 (2008) 451–462. [13] C.P.M. Reutelingsperger, W.L. van Heerde, Annexin V, the regulator of phosphatidylserine-catalyzed inflammation and coagulation during apoptosis, Cell. Mol. Life Sci. 53 (1997) 527–532. [14] S.-M. Chung, O.-N. Bae, K.-M. Lim, J.-Y. Noh, M.-Y. Lee, Y.-S. Jung, J.-H. Chung, Lysophosphatidic acid induces thrombogenic activity through phosphatidylserine exposure and procoagulant microvesicle generation in human erythrocyte, Arterioscler. Thromb. Vasc. Biol. 27 (2007) 414–421. [15] K. De Jong, Z. Beleznay, P. Ott, Phospholipid asymmetry in red blood cells and spectrin-free vesicles during prolonged storage, Biochim. Biophys. Acta 1281 (1996) 101–110. [16] J.L. Holovati, K.A. Wong, J.M. Webster, J.P. Acker, The effects of cryopreservation on red blood cell microvesiculation, phosphatidylserine externalization, and CD47 expression, Transfusion 48 (2008) 1658–1668. [17] S. Scott, S.A. Pendlebury, C. Green, Lipid organization in erythrocyte membrane microvesicles, Biochem. J. 224 (1984) 285–290. [18] D.B. Kim-Sapiro, J. Lee, M.T. Gladwin, Storage lesion: role of red blood cell breakdown, Transfusion 51 (2011) 844–851. [19] C.C. Silliman, N.F. Voelkel, J.D. Allard, D.J. Elzi, R.M. Tuder, J.L. Johnson, D.R. Ambruso, Plasma and lipids from stored packed red blood cells cause acute lung injury in an animal model, J. Clin. Invest. 101 (1998) 1458–1467. [20] C.C. Silliman, C.A. Johnson, K.L. Clay, G.W. Thurman, D.R. Ambruso, Compounds biologically similar to platelet activating factor are present in stored blood components, Lipids 28 (1993) 415–418. [21] C.C. Silliman, A.J. Paterson, W.O. Dickey, D.F. Stroneck, M.A. Popovsky, S.A. Caldwell, D.R. Ambruso, The association of biologically active lipids with the development of transfusion-related acute lung injury: a retrospective study, Transfusion 37 (1997) 719–726. [22] C.C. Silliman, K.L. Clay, G.W. Thurman, C.A. Johnson, D.R. Ambruso, Partial characterization of lipids that develop during routine storage of blood and prime the neutrophil NADPH oxidase, J. Lab. Clin. Med. 124 (1994) 684–694. [23] A.P.J. Vlaar, W. Kulik, R. Niewland, C.P. Peters, A.T.J. Tool, R. van Bruggen, N.P. Juffermans, D. de Korte, Accumulation of bioactive lipids during storage of blood products is not cell but plasma derived and temperature dependent, Transfusion 51 (2011) 2358–2366.

[24] A.P.J. Vlaar, J.J. Hofstra, M. Levi, W. Kulik, R. Niewland, A.T.J. Tool, M.J. Schultz, D. de Korte, N.P. Juffermans, Supernatant of aged erythrocytes causes lung inflammation and coagulopathy in a “two-hit” in vivo syngenic transfusion model, Anesthesiology 113 (2010) 92–103. [25] C.C. Silliman, E.E. Moore, M.R. Kelher, S.Y. Khan, L. Gellar, D.J. Elzi, Identification of lipids that accumulate during the routine storage of prestorage leukoreduced red blood cells and cause acute lung injury, Transfusion 51 (2011) 2549–2554. [26] L.J. Cardo, D. Wilder, J. Salata, Neutrophil priming, caused by cell membranes and microvesicles in packed red blood cell units, is abrogated by leukocyte depletion at collection, Transfus. Apher. Sci. 38 (2008) 117–125. [27] C.J. Draper, T.J. Greenwalt, U.J. Dumaswala, Biochemical and structural changes in RBCs stored with different plasticizers: the role of hexanol, Transfusion 42 (2002) 830–835. [28] P.F.H. Frank, E.M. Bevers, B.H. Lubin, P. Comfurius, D.T.-Y. Chiu, J.A.F. Op den Kamp, R.F.A. Zwaal, L.L.M. van Deenen, B. Roelofsen, Uncoupling of the membrane skeleton from the lipid bilayer, the cause of the accelerated phospholipid flip-flop leading to an enhanced procoagulant activity of sickled cells, J. Clin. Invest. 75 (1985) 183–190. [29] D. Allan, P. Thomas, A.R. Limbrick, The isolation and characterization of 60 nm vesicles (‘nanovesicles’) produced during ionophore A23187-induced budding of human erythrocytes, Biochem. J. 188 (1980) 881–887. [30] D. Allan, R.H. Michell, Accumulation of 1,2-diacylglycerol in the plasma membrane may lead to echinocyte transformation of erythrocytes, Nature 258 (1975) 348–349. [31] G.J.C.G.M. Bosman, E. Lasonder, M. Luten, B. Roerdinkholder-Stoelwinder, V.M.J. Novotny, H. Bos, W.J. De Grip, The proteome of red cell membranes and vesicles during storage of blood bank conditions, Transfusion 48 (2008) 827–835. [32] M. Scigelova, M. Hornshaw, A. Giannakopulos, A. Makarov, Fourier transform mass spectrometry, Mol. Cell. Proteomics 10 (2011) 1–19. [33] S.M. Hammad, J.S. Pierce, F. Soodavar, K.J. Smith, M.M. Al Gadban, B. Rembiesa, R.L. Klein, Y.A. Hannun, J. Bielawski, A. Bielawska, Blood sphingolipidomics in healthy humans: impact of sample collection methodology, J. Lipid Res. 51 (2010) 3074–3087. [34] K. Leidl, G. Liebisch, D. Richter, G. Schmitz, Mass spectrometric analysis of lipid species of human circulating blood cells, Biochim. Biophys. Acta 1781 (2008) 655–664. [35] D.J. Hanahan, Platelet activating factor: a biologically active phosphoglyceride, Annu. Rev. Biochem. 55 (1986) 483–509. [36] C. Popp-Snijders, J.A. Schouten, J. Van Der Meer, E.A. Van Der Veen, Fatty fish-induced changes in membrane lipid composition and viscosity of human erythrocyte suspensions, Scand. J. Clin. Lab. Invest. 46 (1986) 253–258. [37] B. Engelmann, Plasmalogens: targets for oxidants and major lipophilic antioxidants, Biochem. Soc. Trans. 32 (2004) 147–150. [38] V.J. Hammond, V.B. O'Donnell, Esterified eicosanoids: generation, characterization and function, Biochim. Biophys. Acta 1818 (2012) 2403–2412. [39] T. Nakamura, D.L. Bratton, R.C. Murphy, Analysis of epoxyeicosatrienoic and monohydroxyeicosatetraenoic acids esterified to phospholipids in human red blood cells by electrospray tandem mass spectrometry, J. Mass Spectrom. 32 (1997) 888–896. [40] A. D'Alessandro, G.M. D'Amici, S. Vaglio, L. Zolla, Time-course investigation of SAGM-stored leukoreduced-filtered red blood cell concentrates: from metabolism to proteomics, Haematologica 97 (2012) 107–115. [41] G.J.C.G.M. Bosman, E. Lasonder, Y.A.M. Groenen-Döpp, F.L.A. Willekens, J.M. Werre, The proteome of erythrocyte-derived microparticles from plasma: new clues for erythrocyte aging and vesiculation, J. Proteomics (2012), http://dx.doi.org/ 10.1016/j.jprot.2012.05.0.