Platelets (PLTs) deteriorate over time when stored within blood banks through a biological process known as PLT storage lesion (PSL). Here, we describe the refinement of the biochemical model of PLT metabolism, iAT-PLT-636, and its application to describe and investigate changes in metabolism during PLT storage. Changes in extracellular acetate and citrate were measured in buffy coat and apheresis PLT units over 10 days of storage in the PLT additive solution T-Sol. Metabolic network analysis of these data was performed alongside our prior metabolomics data to describe the metabolism of fresh (days 1–3), intermediate (days 4–6), and expired (days 7–10) PLTs. Changes in metabolism were studied by comparing metabolic model flux predictions of iAT-PLT-636 between stages and between collection methods. Extracellular acetate and glucose contribute most to central carbon metabolism in PLTs. The anticoagulant citrate is metabolized in apheresis-stored PLTs and is converted into aconitate and, to a lesser degree, malate. The consumption of nutrients changes during storage and reflects altered PLT activation profiles following their collection. Irrespective of the collection method, a slowdown in oxidative phosphorylation takes place, consistent with mitochondrial dysfunction during PSL. Finally, the main contributors to intracellular ammonium and NADPH are highlighted. Future optimization of flux through these pathways provides opportunities to address intracellular pH changes and reactive oxygen species, which are both of importance to PSL. The metabolic models provide descriptions of PLT metabolism at steady state and represent a platform for future PLT metabolic research.

Introduction

Platelets (PLTs) are blood cells that perform a vital role in wound hemostasis by clogging damaged blood vessels [1] and they also play a critical part in inflammation processes [2]. Some blood disorders, cancers, and cancer treatments can lead to abnormally low PLT counts. This is frequently corrected by carrying out blood transfusion of PLT concentrates which are collected and stored at blood banks.

The investigation of PLTs within a blood bank allows the understanding of PLT biology to be advanced in a rigorously defined and clinically relevant setting. When PLTs are stored over time, they sustain loss of function and change in morphology, a condition referred to as PLT storage lesions (PSLs), which mirrors cellular changes that occur during PLT activation and apoptosis [35]. The storage lesion results from a complex interplay of PLT metabolism and PLT physiology during storage [3] and is known to increase the risk of post-transfusion complications [6]. Studies of PLTs under storage conditions have been performed since the late 1950s. They have shown that the shelf-life of PLTs can be extended by (i) using plastic containers that are permeable to oxygen and CO2 [7]; (ii) inclusion of sodium chloride in storage solutions [8]; (iii) pH regulation [9]; (iv) inclusion of acetate as a buffering agent and a fuel for PLTs [10]; (v) inclusion of citrate as an anticoagulant [10]; and (vi) keeping glucose consumption and lactate production low [11]. Based on these findings, several PLT storage media (termed PLT additive solutions (PASs)) have been developed during the last few decades and subsequently adopted in the blood bank setting in order to improve PLT quality and storage time.

Blood banking guidelines in the U.S.A. dictate that PLTs collected via the apheresis method can be stored for up to 5 days. In Europe, PLTs collected via the buffy coat method can be stored for up to 7 days. The two main factors limiting the PLT shelf-life to such a short time period are the risk of pathogen infection and PSL [4]. Prolonging the storage time could lead to significant cost savings for blood banks around the world [12], and by slowing down the PSL, the likelihood of adverse clinical outcomes could be reduced as prolonged storage is inversely correlated with enhanced survival in vivo [13]. To improve on the state of the art, a detailed biochemical understanding of PLT metabolism during storage is required.

Genome-scale metabolic models (GEMs) integrate sequence, biochemical, and enzymatic data in a structured manner, which enables quantitative prediction of phenotypic behavior [14]. GEMs are increasingly being used to study human metabolism (reviewed in refs [15,16]). A PLT reconstruction termed iAT-PLT-636 [17] has been constructed that forms a basis for data-driven PLT metabolic network modeling and analysis [18]. The iAT-PLT-636 reconstruction of PLT metabolism is a manually curated cell-scale biochemical network based on 33 proteomic studies and over 350 peer-reviewed articles [17]. We recently published comprehensive data sets describing PLT metabolism under storage conditions within a controlled blood bank environment [19,20]. The data include concentration values of multiple intracellular and extracellular metabolites at eight different time points as well as several blood quality control parameters and activation markers, for PLTs collected under both apheresis (AP-PC) and buffy coat (BC-PC) conditions and stored in PASs.

Here, we describe the advancement of these studies to comprehensively describe PLT metabolism during storage in a blood banking environment. We addressed gaps in our previous metabolomics study by measuring acetate and citrate consumption in PLTs over time, performed 13C tracer analysis of uniformly labeled citrate, and analyzed the PLT proteome at different time points of storage using shotgun proteomic analysis. Alongside a literature review of PSL, this allowed us to update iAT-PLT-636 to better capture PSL metabolism. We then applied iAT-PLT-636 to describe metabolic shifts previously shown to occur over time, in apheresis- and buffy coat-derived PLTs. The models describe how changes in the extracellular medium are propagated through intracellular metabolic pathways. We investigated how the PLT nutrients, i.e. acetate, glucose, glutamine, and more, contribute to the differences observed in PLT metabolism during storage from the two separate preparation methods. The results summarize and provide the most accurate biochemical description of PLT metabolism during storage in PASs to date and represent a foundation for the optimization of future PAS and investigation of PLT metabolism ex vivo.

Materials and methods

Metabolite measurements

The metabolomics data sets used in the present study were obtained in two separate storage experiments previously performed by our group (Supplementary Table S1) and consist of extracellular metabolite concentrations and blood gas and hematological parameters (Supplementary Table S2). The collection of these data sets are described in refs [19,20]. In both experiments, the PLT concentrates (PCs) were stored in plastic containers (PL 2410, Fenwal) containing ∼70% T-sol storage solution (Baxter; sodium chloride 115 mM, sodium acetate 30 mM, and sodium citrate dihydrate 10 mM at pH 7.2) and 30% plasma.

The first data set consists of eight PCs collected via the buffy coat method (BC-PCs). Each PC was produced by collecting whole blood into citrate phosphate dextrose (CPD), then separating and pooling BCs from five healthy donors, as described in ref. [20]. The second data set consists of measurements of eight PCs collected by AP-PCs into CPD, from healthy donors, described in detail in ref. [19].

At each time point, the PLT samples were drawn from the PCs; immediately after sampling, a blood gas analyzer (abl90) was used to measure glucose and lactate concentrations. The intra- and extracellular fractions were then separated by centrifugation as described previously [19]. The fractions were aliquoted and stored at −80°C prior to LC–MS measurements. An additional sample was drawn from the PCs at each time point and mitochondrial activity was estimated by measuring the accumulation of the membrane-permeant dye JC-1 in the PLT mitochondria using flow cytometry. Here, we have expanded the analysis by measuring the extracellular concentration of acetate and citrate in the extracellular fractions.

Enzymatic assay kits were used to measure the extracellular concentrations of acetate (ACETAK) and citrate (K-CITR), according to the manufacturer's instruction (Megazyme International Ireland Ltd, Wicklow, Ireland) using a microplate reader (Spectramax M3, Molecular Devices, Sunnyvale, CA). Statistical analysis was carried out by a two-tailed paired t-test where applicable.

To account for citrate metabolism, an apheresis unit was incubated in T-Sol containing uniformly labeled 13C citrate (CLM-9021-PK, Cambridge Isotope Laboratories) and compared with an untreated apheresis unit from the same donor. Unit sampling and subsequent metabolomics analysis were performed in a similar manner as previously described to determine qualitative profiles of intracellular and extracellular metabolites [21,22]. Measured intensities were corrected for natural abundance of 13C-isotopes using the software Isocor [23].

Proteomic analysis

Three PC units were prepared by the buffy coat method from a pool of five healthy blood donors each, of mixed age and sex at the Blood bank, Landspitali University Hospital, Iceland, and stored as previously reported [20]. Three 4 ml samples were collected on days 1, 3, and 6 of storage. At the end of the experiment, pathogen analysis at Landspitali University Hospital was negative. Protein isolation, peptide cleavage, and purification are summarized in Supplementary File 1.

Two different methods of peptide separation and mass spectrometry analysis were performed. For the first PC unit, peptide separation was achieved by one-dimensional reversed phase-liquid chromatography using nanoACQUITY UPLC (Waters Corporation, U.S.A.) with a 75 µm × 25 cm C18 HSS T3 column. The experiments were performed using a 1.5 h gradient (5–40% acetonitrile over 90 min) at a flow rate of 300 nl/min. Mass spectrometry analysis was performed using SYNAPT G2Si HDMS (Waters, Manchester, U.K.) operating in a data-independent manner coupled with ion mobility. All samples were measured in triplicates. Protein identification and label-free quantification were performed using Progenesis QI for proteomics (NonLinear Dynamics, U.K.). For the second two PC units, the peptides were injected and washed on a trap column (Acclaim PepMap 100 C18, 100 µM × 2 cm, Thermo Scientific), then separated on a second column (Easy-Spray PepMap RSLC C18 column, 75 µM × 50 cm, Thermo Scientific) using the Ultimate 3000 RSLCnano system (Thermo Scientific) The mobile phases consisted of buffer A (0.1% formic acid) and buffer B (80% acetonitrile in 0.08% formic acid); the gradient went from 2% to 98% buffer B in 152 min, followed by 18 min with 2% solvent B with a flow rate of 300 nl/min. The samples were transferred to a mass spectrometer via an Easy-Spray source with temperature set at 50°C and a source voltage of 2.0 kV. The mass spectrometry analysis was performed using LTQ Orbitrap Velos Pro (Thermo Scientific) operating in a data-dependent acquisition mode. Protein identification and label-free quantifications were carried out using MaxQuant [24].

For the semi-quantitative analysis, only proteins were used that were detected with both methods with at least one unique peptide and a non-zero normalized intensity measured at at least one time point in all three PC units. To allow for a semi-quantitative comparison from both proteomic methods, protein abundances measured within each PC were first z-score transformed. Principal component analysis (PCA) was carried out with the statistical software R version 3.4.3; hierarchical clustering analysis and heat map generation were carried out using MetaboAnalyst 3.0 [25]. The Reactome knowledgebase [26] was used to sort the detected proteins by biological pathways.

A mathematical model of PLT metabolism

Based on the results obtained from the metabolite and proteomic analysis, iAT-PLT-636 [17] was refined by adding reactions necessary for a functional malate-aspartate shuttle, mitochondrial acetate transport and acetyl-coA synthetase, pyrimidine biosynthesis, diffusion of ammonia through the mitochondrial membrane, transport of and exchange of D-lactate, and production and transport of mitochondrial phosphoenolpyruvate. The reaction catalyzed by the cytosolic form of aconitase is split into two reactions, allowing the formation of the reaction's intermediate cis-aconitate. Aconitate plasma membrane transport and exchange reactions were added to account for the observed aconitate secretion.

More precisely, to address gaps in the iAT-PLT-636 model, 16 metabolic reactions (KGMALtm, ASPGLUm, CITtam, PEPtam, Lac_D_Transport, ACSm, PPAm, PPCKm, ADK1m, RPE, CBPS, ASPCT, DHORTS, DHORD_NAD, ORPT, and OMPDC) were added, all of which have been detected in proteomic experiments. Also added to the model was the electroneutral diffusion of acetate across the mitochondrial membrane (ACt2m) on the grounds that the unionized form of acetic acid can diffuse over lipid membranes. Furthermore, four reactions (EX_k(e), ASNS1, ASNtN1, NH4D, sink_orn[c], and EX_LacD_e) were added to resolve infeasibility of the model (Supplementary Table S3). Thirteen reactions were removed from the iAT-PLT-636 model either due to a lack of evidence for the reactions (ACCOAtm, MALSO3tm, MALTSULtm, MALtm, and LDH_D) and/or to eliminate internal loops (FBP, SUCOASm, UGLT, PFK26, GALT, PPDOx, PPDOy, and SO4t4_2) (Supplementary Table S4). Finally, the reaction ACONT was replaced by ACONT1 and ACONT1, and aconitate transport (acontT2) and exchange (EX_acon(e)) reactions were added to allow for cytosolic aconitate production (Supplementary Tables S3 and S4).

Model constraints derived from metabolomics data

Metabolomics data were used to define the boundary conditions of the models (i.e. their constraints) as follows: First, the storage time was divided into three stages representative of fresh (days 1–3), intermediate (days 4–6), and expired (days 7–10) PCs, hereafter referred to as Stages 1, 2, and 3, respectively.

For each of the three storage stages and each extracellular metabolite k included in the model, a line was fit through the corresponding data points (Figure 1 and Supplementary Table S5). The slope of the line, normalized to the average PLT count in each storage bag, defined as the exchange flux vk of the metabolite, with the positive slope denoting secretion and the negative slope denoting consumption (Figure 1).

Extracellular concentrations of acetate, citrate, glucose, and glutamine, and percentage of cells with polarized mitochondria at different time points during PLT storage.

Figure 1.
Extracellular concentrations of acetate, citrate, glucose, and glutamine, and percentage of cells with polarized mitochondria at different time points during PLT storage.

Eight BC-PCs are shown in (A) and eight AP-PCs in (B). The storage time is divided into three stages: Stage 1 (days 1–3), Stage 2 (days 4–6), and Stage 3 (days 7–10). For extracellular metabolites, at each stage, a line is fit through the measured concentration values. The sign and magnitude of the slopes of the fitted lines define the uptake/secretion rates of the metabolites within iAT-PLT-636.

Figure 1.
Extracellular concentrations of acetate, citrate, glucose, and glutamine, and percentage of cells with polarized mitochondria at different time points during PLT storage.

Eight BC-PCs are shown in (A) and eight AP-PCs in (B). The storage time is divided into three stages: Stage 1 (days 1–3), Stage 2 (days 4–6), and Stage 3 (days 7–10). For extracellular metabolites, at each stage, a line is fit through the measured concentration values. The sign and magnitude of the slopes of the fitted lines define the uptake/secretion rates of the metabolites within iAT-PLT-636.

In case this results in an infeasible model, i.e. no set of fluxes exist which fulfill all model constraints, the exchange fluxes are adjusted slightly to obtain a feasible model. The adjustment of exchange fluxes is determined by solving the following optimization problem [27]:

 
formula
1
 
formula
2
 
formula
3
 
formula
4
 
formula
5
 
formula
6

The set of exchange reactions to be relaxed is denoted by Rx and the set Rn represents all other reactions. The objective (1) is to minimize the total adjustment to exchange fluxes by adjusting the amount of upper (pj) and lower bound relaxations (nj) and network fluxes (v) while fulfilling the steady-state mass balance constraints (2). Constraints (3) enable relaxation of the exchange reactions. Constraints (4) prevent the uptake of metabolites that should only be secreted and constraints (5) are the usual flux balance analysis upper and lower bounds. The algorithm was implemented in Matlab (Mathworks, Natick, MA, U.S.A.) using the CVX modeling language (CVX Research 2012). The Gurobi solver (Gurobi Optimization 2017) was used to solve the optimization problem. Exchange flux alterations made to the models as a result of the optimization algorithm can be found in Supplementary Table S6.

Model analysis

All model computations and analysis were done in the Matlab environment (Mathworks, Natick, MA, U.S.A.) using the COBRA toolbox [28]. Graphical representations of reaction pathways were obtained with the Escher tool [29]. Flux balance analysis (FBA) [30] was used to determine the maximal amount of ATP produced by the models described above. Random sampling [31] was then used to obtain estimates of flux distributions in the models. ATP generation was constrained to 90% of the maximum ATP production prior to sampling. The flux distributions are affected both by the network topology and the intervals containing the exchange fluxes. Average fluxes obtained by random sampling of the allowed flux space in individual reactions under different conditions were compared.

Results

Acetate is the main PLT fuel during PLT storage

In the present study, we set out to capture and explain key features of PSL metabolism using cell-scale metabolic modeling. Extracellular metabolomics data descriptive of changes in metabolite concentration over time from our previous studies were used and formed the basis for the analysis in the present study [19,20]. When present in the storage media, acetate is the primary fuel of PLTs [9]. Acetate was not part of the previous metabolomics screening and to address this, we measured changes in extracellular acetate concentrations. In addition, to support previous citrate metabolomics measurements, we measured the extracellular concentration of citrate using an enzyme-coupled assay. These measurements were performed on the same PLT samples as the previously published metabolomics assays [19,20].

The rate of acetate uptake remained fairly constant over the first 7 days of storage in both AP-PCs (0.92 ± 0.12 mM/day) and BC-PCs (1.04 ± 0.07 mM/day) and decreased most between days 7 and 10. A significant decrease in acetate concentration was measured between days 1 and 7 in both AP-PC (P = 9.3 × 10−7) and BC-PC (P = 3.0 × 10−5). Conversely, a significant change in extracellular citrate was measured only in the AP-PCs between days 1 and 10 (P = 0.0059), but not in BC-PCs (P = 0.55) (Figure 1).

The measured uptake rate of citrate in the AP-PCs was considerably higher than previously reported [32,33]; we therefore confirmed citrate metabolism further by adding uniformly labeled 13C labeled citrate to a single AP-PC unit. Label enrichment was detected in intracellular citrate and malate, and extracellular citrate and aconitate (Figure 2). The enrichment of extracellular aconitate mirrored that of the extracellular citrate as opposed to the intracellular pool. This indicated that medium-derived citrate is converted into aconitate without mixing with the total intracellular citrate pool, possibly due to compartmentation. To account for this in the model, the reaction converting cytosolic citrate into isocitrate (ACONT) was split into its two separate reactions, one that reversibly converts citrate into aconitate (ACONT1) and another that reversibly converts aconitate into isocitrate (ACONT2). Only a small label enrichment was detected in intracellular malate and only as the fully labeled isotopologue. The relative contribution increased from 2% at day 0 to 7% at day 10 of storage, indicating that citrate is not metabolized via the TCA (tricarboxylic acid) to a significant extent.

Isotopologue distribution of 13C-enriched citrate and malate following the addition of UL-13C6 citrate to an AP-PC unit.

Figure 2.
Isotopologue distribution of 13C-enriched citrate and malate following the addition of UL-13C6 citrate to an AP-PC unit.

Label enrichment was detected in four metabolites in AP-PC containing uniformly labeled 13C citrate. Except for a small amount of the M+5 in the extracellular citrate, probably due to impurities of the added labeled compound, the only isotopologues detected were the naturally occurring M+0 and M+1, as well as the fully labeled metabolites. (A) The extracellular citrate label enrichment remained stable during storage with 45.2 ± 0.1% of the total citrate pool fully labeled. (B) Intracellular citrate was enriched with 16.2 ± 0.1% at day 0 on the 13C6 isotopologue, the enrichment of which increased linearly (R2 = 0.87) with time and reached 29.2 ± 0.1% at day 10. (C) The extracellular aconitate pool was immediately highly enriched at day 0 of storage, with label enrichment comparable to extracellular citrate throughout storage, indicating that the majority of extracellular aconitate originates from extracellular citrate. (D) Label enrichment was detected in relatively low amounts (<10%) in intracellular malate, and its relative abundance increases over the storage time. The lack of partially labeled isotopologues indicates that the citrate taken up from the medium is not broken down via the TCA cycle. Abbreviations: MDV, mass distribution vector; M+n, isotopologue with n carbon atoms labeled with 13C.

Figure 2.
Isotopologue distribution of 13C-enriched citrate and malate following the addition of UL-13C6 citrate to an AP-PC unit.

Label enrichment was detected in four metabolites in AP-PC containing uniformly labeled 13C citrate. Except for a small amount of the M+5 in the extracellular citrate, probably due to impurities of the added labeled compound, the only isotopologues detected were the naturally occurring M+0 and M+1, as well as the fully labeled metabolites. (A) The extracellular citrate label enrichment remained stable during storage with 45.2 ± 0.1% of the total citrate pool fully labeled. (B) Intracellular citrate was enriched with 16.2 ± 0.1% at day 0 on the 13C6 isotopologue, the enrichment of which increased linearly (R2 = 0.87) with time and reached 29.2 ± 0.1% at day 10. (C) The extracellular aconitate pool was immediately highly enriched at day 0 of storage, with label enrichment comparable to extracellular citrate throughout storage, indicating that the majority of extracellular aconitate originates from extracellular citrate. (D) Label enrichment was detected in relatively low amounts (<10%) in intracellular malate, and its relative abundance increases over the storage time. The lack of partially labeled isotopologues indicates that the citrate taken up from the medium is not broken down via the TCA cycle. Abbreviations: MDV, mass distribution vector; M+n, isotopologue with n carbon atoms labeled with 13C.

Intracellular metabolic protein abundance changes during storage

To account for a change in the abundance of metabolic proteins during storage and their effect on PSL metabolism, we carried out shotgun proteomics analysis of three BC-PC units at three different time points. The time points reflected the three-phase metabolic decay of PSL [19,20]. The measurements were made using two different methods whose analysis was then combined as described in Materials and Methods. With the first method (qTOF), a total of 1663 proteins with at least one unique peptide were identified. The second method (Orbitrap) identified 2118 proteins with at least one unique peptide. A total of 882 proteins were identified with both methods. Pathway analysis identified 332 metabolic proteins of which 160 were detected with both methods, on which a semi-quantitative analysis on the protein abundances was performed. PCA revealed a distinct separation of the samples into three groups based on storage time (Figure 3). The majority of the proteins contributing to the PCA separation decreased in abundance with storage time (Table 1). A similar result was obtained by unsupervised hierarchical clustering (Figure 3). The results from the proteomic analysis by both methods can be found in Supplementary File 2. Using these results supplemented by a literature review of the PLT proteome, 16 reactions were added to the model, including reactions necessary for a functional malate–aspartate shuttle, acetate metabolism, and pyrimidine biosynthesis (Supplementary Table S3).

PCA and hierarchical clustering analysis of metabolic proteins at days 1, 3, and 6 of storage.

Figure 3.
PCA and hierarchical clustering analysis of metabolic proteins at days 1, 3, and 6 of storage.

(A) PCA of the metabolic proteins identified by both proteomics methods (n = 160); the numbers in parenthesis represent the total percentage of variance explained by the principal components. The three biological replicates at each time point are represented by the same color. (B) Hierarchical clustering and a heatmap of the metabolic proteins. Only the 25 proteins contributing the most to the clustering are shown. The majority of the proteins decrease in abundance during the 6 days of storage. The heat map and hierarchical clustering are based on Euclidean distance and Ward's method.

Figure 3.
PCA and hierarchical clustering analysis of metabolic proteins at days 1, 3, and 6 of storage.

(A) PCA of the metabolic proteins identified by both proteomics methods (n = 160); the numbers in parenthesis represent the total percentage of variance explained by the principal components. The three biological replicates at each time point are represented by the same color. (B) Hierarchical clustering and a heatmap of the metabolic proteins. Only the 25 proteins contributing the most to the clustering are shown. The majority of the proteins decrease in abundance during the 6 days of storage. The heat map and hierarchical clustering are based on Euclidean distance and Ward's method.

Table 1
The top 25 metabolic proteins influencing the PCA separation

The proteins either increased in abundance during storage (arrow up), or decreased (arrow down) in all three PC units.

Gene ID Change in abundance Name 
TALDO1 ↓ Transaldolase 
SLC25A3 ↓ Phosphate carrier protein, mitochondrial 
PIK3CG ↑ Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit gamma isoform 
SLC25A5 ↓ ADP/ATP translocase 2 
RDH11 ↓ Retinol dehydrogenase 11 
RENBP ↓ N-acylglucosamine 2-epimerase 
ENO1 ↓ Alpha-enolase 
GPD2 ↓ Glycerol-3-phosphate dehydrogenase, mitochondrial 
IDH2 ↓ Isocitrate dehydrogenase [NADP], mitochondrial 
GRHPR ↓ Glyoxylate reductase/hydroxypyruvate reductase 
LDHA ↓ l-lactate dehydrogenase A chain 
PGM1 ↓ Phosphoglucomutase-1 
UGP2 ↓ UTP-glucose-1-phosphate uridylyltransferase 
ETFA ↑ Electron transfer flavoprotein subunit alpha, mitochondrial 
GSTO1 ↓ Glutathione S-transferase omega-1 
GLO1 ↓ Lactoylglutathione lyase 
PYGL ↓ Glycogen phosphorylase, liver form 
PGLS ↓ 6-Phosphogluconolactonase 
HADH ↓ Hydroxyacyl-coenzyme A dehydrogenase, mitochondrial 
DECR1 ↓ 2,4-Dienoyl-CoA reductase, mitochondrial 
PSAP ↓ Prosaposin 
PGK1 ↓ Phosphoglycerate kinase 1 
ATP5L ↓ ATP synthase subunit g 
FDPS ↓ Farnesyl pyrophosphate synthase 
PRDX3 ↓ Thioredoxin-dependent peroxide reductase, mitochondrial 
Gene ID Change in abundance Name 
TALDO1 ↓ Transaldolase 
SLC25A3 ↓ Phosphate carrier protein, mitochondrial 
PIK3CG ↑ Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit gamma isoform 
SLC25A5 ↓ ADP/ATP translocase 2 
RDH11 ↓ Retinol dehydrogenase 11 
RENBP ↓ N-acylglucosamine 2-epimerase 
ENO1 ↓ Alpha-enolase 
GPD2 ↓ Glycerol-3-phosphate dehydrogenase, mitochondrial 
IDH2 ↓ Isocitrate dehydrogenase [NADP], mitochondrial 
GRHPR ↓ Glyoxylate reductase/hydroxypyruvate reductase 
LDHA ↓ l-lactate dehydrogenase A chain 
PGM1 ↓ Phosphoglucomutase-1 
UGP2 ↓ UTP-glucose-1-phosphate uridylyltransferase 
ETFA ↑ Electron transfer flavoprotein subunit alpha, mitochondrial 
GSTO1 ↓ Glutathione S-transferase omega-1 
GLO1 ↓ Lactoylglutathione lyase 
PYGL ↓ Glycogen phosphorylase, liver form 
PGLS ↓ 6-Phosphogluconolactonase 
HADH ↓ Hydroxyacyl-coenzyme A dehydrogenase, mitochondrial 
DECR1 ↓ 2,4-Dienoyl-CoA reductase, mitochondrial 
PSAP ↓ Prosaposin 
PGK1 ↓ Phosphoglycerate kinase 1 
ATP5L ↓ ATP synthase subunit g 
FDPS ↓ Farnesyl pyrophosphate synthase 
PRDX3 ↓ Thioredoxin-dependent peroxide reductase, mitochondrial 

PLT metabolism was modeled as a three-phase metabolic decay using iAT-PLT-636

To make the model more descriptive of PSL metabolism, we refined and updated iAT-PLT-636 based on the results obtained from the 13C tracer and proteomic analysis, as well as through a literature review (Supplementary Tables S3 and S4). The refined model was used to analyze the PSL metabolomics data that we sectioned into three groups, representative of days 1–3 (Stage 1), days 4–6 (Stage 2), and days 7–10 (Stage 3) of storage. The measured changes in extracellular metabolite concentrations within each stage were then used as boundaries to define the maximum and minimum allowed flux of metabolites through the metabolic pathways (Figure 1). The extracellular fluxes together with the stoichiometric coefficients of the network reconstruction therefore implicitly define fluxes in metabolic pathways. Three models descriptive of steady-state metabolism in PLTs during Stages 1, 2, and 3 were created. Flux analysis of these models allowed the estimation of the most likely flux of metabolites through metabolic pathways and its comparison between all stages. In the following, analysis in terms of ‘fluxes’ refers to predictions made by the PLT model, specifically referring to the arithmetic means of the randomly sampled flux values.

To visualize the results, flux values were mapped onto graphical representations of the PLT biochemical network. Supplementary Figure S1 shows the average flux distributions calculated for the glycolytic and pentose phosphate pathways during Stage 1 in BC-PCs. The average flux values for each reaction are listed in Supplementary File 3. Selected metabolic features are described further below.

Buffy coat-derived PLT metabolism is captured in iAT-PLT-636

The primary role of PLT metabolism is to provide energy in the form of ATP necessary to maintain homeostasis. To investigate the ATP production potential, we compared the metabolism between the stages of BC-PCs by FBA. This analysis predicted a roughly 1.5-fold increase in maximal ATP turnover rate between Stages 1 and 2 with a decline in the ATP turnover rate during Stage 3 (Table 2). The contribution toward ATP production from glycolysis and the oxidative phosphorylation was constant during the first two stages with 11–13% of ATP produced via glycolysis and the remaining 87–89% in the mitochondria. In Stage 3, however, the relative contribution of glycolysis increased as the flux through oxidative metabolism decreased (Table 3 and Figure 4A,B). Previous experiments have estimated that 15–30% of ATP turnover derives from glycolysis with the remainder from oxidative phosphorylation [3437]. In accordance with energy production, the oxygen consumption rate was predicted to increase between the first two stages and decrease in Stage 3, with a parallel change in CO2 production (Table 2). To reveal further how specific nutrients contribute to these storage metabolic phenotypes, we investigated the main contributors to PLT metabolism during storage.

The major energy-producing pathways in PLTs all have a higher flux during days 4–6 of storage.

Figure 4.
The major energy-producing pathways in PLTs all have a higher flux during days 4–6 of storage.

A schematic overview of predicted flux changes in PLTs during storage. The storage time is divided into three stages: Stage 1 (days 1–3), Stage 2 (days 4–6), and Stage 3 (days 7–10). The figure depicts the difference in fluxes in consecutive stages in mmol/day/1012 PLTs. The flux difference is defined as Vkb − Vka, with a < b, where Vkn is flux through reaction k in mmol/day/1012 PLTs and b is the stage. Four extracellular metabolites are shown in the figure (glucose, lactate, acetate, and glutamine), and the flux distribution is, however, determined by 18–19 extracellular metabolites used to constrain the model (Supplementary Table S5). Metabolites are represented by orange dots and reactions by lines and curves connecting the dots. (A) Flux changes between the first two stages in BC-PCs, and all but two reactions shown have a higher flux value in Stage 2 compared with Stage 1. The largest difference is found in the reaction of the TCA cycle. (B) Flux changes between Stages 2 and 3 in BC-PCs. During Stage 3, there is a substantial decrease in acetate consumption and TCA cycle activity, which is compensated with higher flux through glycolysis. (C) Flux changes between the first two stages in AP-PCs. Flux through energy-producing pathways remains relatively stable between Stages 1 and 2. Flux through the PDH reaction is predicted to increase substantially, whereas citrate uptake (not shown) decreases. (D) Flux changes between Stages 2 and 3 in AP-PCs. A large decrease in glycolysis in Stage 3 is due to glucose depletion in the media. A decrease in mitochondrial energy production at Stage 3 is consistent with the results from BC-PC. Abbreviations: PDH, pyruvate dehydrogenase.

Figure 4.
The major energy-producing pathways in PLTs all have a higher flux during days 4–6 of storage.

A schematic overview of predicted flux changes in PLTs during storage. The storage time is divided into three stages: Stage 1 (days 1–3), Stage 2 (days 4–6), and Stage 3 (days 7–10). The figure depicts the difference in fluxes in consecutive stages in mmol/day/1012 PLTs. The flux difference is defined as Vkb − Vka, with a < b, where Vkn is flux through reaction k in mmol/day/1012 PLTs and b is the stage. Four extracellular metabolites are shown in the figure (glucose, lactate, acetate, and glutamine), and the flux distribution is, however, determined by 18–19 extracellular metabolites used to constrain the model (Supplementary Table S5). Metabolites are represented by orange dots and reactions by lines and curves connecting the dots. (A) Flux changes between the first two stages in BC-PCs, and all but two reactions shown have a higher flux value in Stage 2 compared with Stage 1. The largest difference is found in the reaction of the TCA cycle. (B) Flux changes between Stages 2 and 3 in BC-PCs. During Stage 3, there is a substantial decrease in acetate consumption and TCA cycle activity, which is compensated with higher flux through glycolysis. (C) Flux changes between the first two stages in AP-PCs. Flux through energy-producing pathways remains relatively stable between Stages 1 and 2. Flux through the PDH reaction is predicted to increase substantially, whereas citrate uptake (not shown) decreases. (D) Flux changes between Stages 2 and 3 in AP-PCs. A large decrease in glycolysis in Stage 3 is due to glucose depletion in the media. A decrease in mitochondrial energy production at Stage 3 is consistent with the results from BC-PC. Abbreviations: PDH, pyruvate dehydrogenase.

Table 2
Measured and predicted values of uptake/secretion rates measured in mmol/day/1012 PLTs during Stage 1 (days 1–3), Stage 2 (days 4–6), and Stage 3 (days 7–10) of storage

Positive values represent metabolite secretion and negative values uptake. Literature-sourced experimental values vary considerably on account of various experimental conditions and are, in some cases, quite different from the conditions used in the present study. The data are expressed as mean ± standard error of the mean.

Metabolite Condition Stage 1 Stage 2 Stage 3 Literature values References 
Glucose BC-PC −0.46 ± 0.39 −0.62 ± 0.38 −0.70 ± 0.20 −(0.16–1.3) [9,10,35,38,42,43,6063
AP-PC −0.64 ± 0.24 −0.65 ± 0.28 −0.32 ± 0.12 
Lactate BC-PC 0.90 ± 0.19 1.07 ± 0.24 1.27 ± 0.27 0.3–2.8 [9,10,35,38,42,43,60,61,63
AP-PC 1.19 ± 0.24 1.11 ± 0.37 0.57 ± 0.23 
Acetate BC-PC −0.95 ± 0.95 −1.20 ± 0.81 −0.64 ± 0.64 −(0.5–2.8) [9,32,38,42,64
AP-PC −0.82 ± 0.43 −0.80 ± 0.54 −0.42 ± 0.38 
Glutamine BC-PC −0.027 ± 0.29 −0.038 ± 0.029 −0.029 ± 0.012 −(0.1) [48
AP-PC −0.0072 ± 0.019 −0.0086 ± 0.0097 −0.0054 ± 0.0057 
Citrate BC-PC −0.0075 ± 0.42 0.034 ± 0.41 −0.019 ± 0.28 −(0–0.007) [32,33
AP-PC −0.041 ± 0.26 0.0097 ± 0.27 −0.034 ± 0.18 
O21 BC-PC −2.00 ± 0.01 −2.91 ± 0.08 −1.77 ± 0.01 −(0.76–1.96) [9,10,33,35,38,61,64
AP-PC −2.09 ± 0.03 −2.17 ± 0.04 −1.22 ± 0.01 
CO21 BC-PC 1.94 ± 0.01 2.72 ± 0.05 1.63 ± 0.01 NA  
AP-PC 2.07 ± 0.03 2.08 ± 0.02 1.23 ± 0.01 
Ammonia1 BC-PC 0.080 ± 0.0007 0.100 ± 0.002 0.080 ± 0.002 0.04–0.15 [33,65,66
AP-PC 0.026 ± 0.0004 0.023 ± 0.0008 0.019 ± 0.0003 
Maximum ATP (FBA) BC-PC 7.33 10.75 7.31 NA  
AP-PC 8.27 8.60 4.85 
Net ATP2 BC-PC 7.00 ± 0.02 9.76 ± 0.14 7.08 ± 0.03 NA  
AP-PC 7.83 ± 0.06 8.05 ± 0.07 4.70 ± 0.02 
Metabolite Condition Stage 1 Stage 2 Stage 3 Literature values References 
Glucose BC-PC −0.46 ± 0.39 −0.62 ± 0.38 −0.70 ± 0.20 −(0.16–1.3) [9,10,35,38,42,43,6063
AP-PC −0.64 ± 0.24 −0.65 ± 0.28 −0.32 ± 0.12 
Lactate BC-PC 0.90 ± 0.19 1.07 ± 0.24 1.27 ± 0.27 0.3–2.8 [9,10,35,38,42,43,60,61,63
AP-PC 1.19 ± 0.24 1.11 ± 0.37 0.57 ± 0.23 
Acetate BC-PC −0.95 ± 0.95 −1.20 ± 0.81 −0.64 ± 0.64 −(0.5–2.8) [9,32,38,42,64
AP-PC −0.82 ± 0.43 −0.80 ± 0.54 −0.42 ± 0.38 
Glutamine BC-PC −0.027 ± 0.29 −0.038 ± 0.029 −0.029 ± 0.012 −(0.1) [48
AP-PC −0.0072 ± 0.019 −0.0086 ± 0.0097 −0.0054 ± 0.0057 
Citrate BC-PC −0.0075 ± 0.42 0.034 ± 0.41 −0.019 ± 0.28 −(0–0.007) [32,33
AP-PC −0.041 ± 0.26 0.0097 ± 0.27 −0.034 ± 0.18 
O21 BC-PC −2.00 ± 0.01 −2.91 ± 0.08 −1.77 ± 0.01 −(0.76–1.96) [9,10,33,35,38,61,64
AP-PC −2.09 ± 0.03 −2.17 ± 0.04 −1.22 ± 0.01 
CO21 BC-PC 1.94 ± 0.01 2.72 ± 0.05 1.63 ± 0.01 NA  
AP-PC 2.07 ± 0.03 2.08 ± 0.02 1.23 ± 0.01 
Ammonia1 BC-PC 0.080 ± 0.0007 0.100 ± 0.002 0.080 ± 0.002 0.04–0.15 [33,65,66
AP-PC 0.026 ± 0.0004 0.023 ± 0.0008 0.019 ± 0.0003 
Maximum ATP (FBA) BC-PC 7.33 10.75 7.31 NA  
AP-PC 8.27 8.60 4.85 
Net ATP2 BC-PC 7.00 ± 0.02 9.76 ± 0.14 7.08 ± 0.03 NA  
AP-PC 7.83 ± 0.06 8.05 ± 0.07 4.70 ± 0.02 
*

Model predictions.

**

See the Supplementary Text for the definition of net ATP.

Table 3
Partition of energy productions during Stage 1 (days 1–3), Stage 2 (days 4–6), and Stage 3 (days 7–10) of storage, as predicted by the models reported as the percentage of ATP produced in mitochondria

The remaining ATP is generated in the cytosol via glycolysis. Three different methods to estimate the percentage of ATP produced in mitochondria are compared: flux balance analysis (FBA), net ATP, and P/O ratio. See the Supplementary Text for definitions.

Method Condition Stage 1 Stage 2 Stage 3 Literature values References 
FBA (%) BC-PC 87.2 88.7 80.9  NA NA 
AP-PC 83.8 84.7 85.8 
Net ATP (%) BC-PC 87.1 ± 0.5 88.7 ± 2.1 80.8 ± 0.6  NA NA 
AP-PC 83.7 ± 1.1 84.6 ± 1.3 85.8 ± 0.6 
P/O (%) BC-PC 92.4 ± 1.0 93.7 ± 3.9 88.4 ± 1.1 70–80 [3437
AP-PC 90.6 ± 2.2 91.5 ± 2.6 92.2 ± 1.2 
Method Condition Stage 1 Stage 2 Stage 3 Literature values References 
FBA (%) BC-PC 87.2 88.7 80.9  NA NA 
AP-PC 83.8 84.7 85.8 
Net ATP (%) BC-PC 87.1 ± 0.5 88.7 ± 2.1 80.8 ± 0.6  NA NA 
AP-PC 83.7 ± 1.1 84.6 ± 1.3 85.8 ± 0.6 
P/O (%) BC-PC 92.4 ± 1.0 93.7 ± 3.9 88.4 ± 1.1 70–80 [3437
AP-PC 90.6 ± 2.2 91.5 ± 2.6 92.2 ± 1.2 

Decreased acetate oxidation and increased glycolysis, pentose phosphate pathway, and polyol pathway flux accompany PSL

During storage in acetate-containing media, ATP is primarily produced from the catabolism of acetate and glucose, with a minor contribution from glutamine and citrate. Accordingly, in the BC-PC model, citrate and glutamine maximally accounted for 5.4–10.3% of O2 consumption combined (Table 4). Acetate, however, contributed most to O2 consumption. Nearly all consumed acetate (∼99%) was predicted to be converted into acetyl-coA via the acetyl-coA synthetase reaction in the mitochondria where it was predicted to be fully oxidized to CO2. Acetate oxidation was predicted to contribute to 83% of total ATP production during Stage 1, 73% during Stage 2, and 58% during Stage 3. Negligible amounts of acetate were predicted to be incorporated into fatty acids, consistent with low amounts of acetate incorporation into PLTs measured in 14C tracer studies [9,38].

Table 4
The maximal contribution of selected metabolites to oxygen consumption during Stage 1 (days 1–3), Stage 2 (days 4–6), and Stage 3 (days 7–10) of storage

The maximal contribution to oxygen consumption is defined such that the given metabolite is fully oxidized to CO2 and all oxygen consumption is mitochondrial. The data are expressed as mean ± standard deviation.

Metabolite Condition Stage 1 Stage 2 Stage 3 Literature values References 
Acetate (%) BC-PC 95.0 ± 1.3 82.6 ± 10.0 72.0 ± 0.9 85–106 [34,38
AP-PC 77.5 ± 2.8 73.7 ± 3.4 65.8 ± 0.4 
Glucose (%) BC-PC 1.1 ± 2.4 9.7 ± 31.8 17.3 ± 4.2 4.7–8.8 [37
AP-PC 16.6 ± 10.6 22.5 ± 12.9 26.5 ± 1.9 
Glutamine (%) BC-PC 5.4 ± 0.03 5.2 ± 0.04 6.6 ± 0.03 10 [48
AP-PC 1.40 ± 0.01 1.60 ± 0.01 1.8 ± 0.01 
Citrate (%) BC-PC 0.03 ± 0.0002 3.7 ± 0.02 NA  
AP-PC 9.84 ± 0.12 15.9 ± 0.03 
Metabolite Condition Stage 1 Stage 2 Stage 3 Literature values References 
Acetate (%) BC-PC 95.0 ± 1.3 82.6 ± 10.0 72.0 ± 0.9 85–106 [34,38
AP-PC 77.5 ± 2.8 73.7 ± 3.4 65.8 ± 0.4 
Glucose (%) BC-PC 1.1 ± 2.4 9.7 ± 31.8 17.3 ± 4.2 4.7–8.8 [37
AP-PC 16.6 ± 10.6 22.5 ± 12.9 26.5 ± 1.9 
Glutamine (%) BC-PC 5.4 ± 0.03 5.2 ± 0.04 6.6 ± 0.03 10 [48
AP-PC 1.40 ± 0.01 1.60 ± 0.01 1.8 ± 0.01 
Citrate (%) BC-PC 0.03 ± 0.0002 3.7 ± 0.02 NA  
AP-PC 9.84 ± 0.12 15.9 ± 0.03 

Glucose was the second largest source of energy during storage and the only source for ATP production in the cytosol. The majority of glucose (86–98%) was converted into lactate via aerobic glycolysis with 2–11% of glucose-6-phosphate predicted to enter the pentose phosphate pathway (PPP) (Table 5). To allow comparison with classical 14C studies [9,34,39,40], we calculated the proportion of glucose carbons converted into CO2 (Table 5). Assuming that all acetyl-coA formed from pyruvate was fully oxidized, the percentage of glucose-derived CO2 was predicted to be: 0.8% for Stage 1, 7.7% for Stage 2, and 7.3% for Stage 3. Notably, the increase in glucose-derived CO2 between Stages 1 and 2 was not only due to an increase in flux through the pyruvate dehydrogenase reaction and consequent oxidation in mitochondria, but also due to increased flux through the PPP. Analysis of model reaction flux showed that the increase through the PPP was driven by glucose entering the polyol pathway. Specifically, during Stage 2, 44% of glucose taken up by the PLTs was predicted to be funneled through aldose reductase (AR), creating an increased demand for NADPH. Metabolic network analysis thus identified this reaction accounting for the increase in flux into the PPP in Stage 2 (Table 5).

Table 5
Predicted glucose utilization during Stage 1 (days 1–3), Stage 2 (days 4–6), and Stage 3 (days 7–10) of storage

The proportion of total glucose-derived carbons oxidized to CO2 (glucose oxidized). The proportion of glucose entering the pentose PPP. The percentage of glucose carbons metabolized via aerobic glycolysis. The percentage of glucose carbons entering the mitochondria via pyruvate dehydrogenase (PDH). The percentage of glucose entering the polyol pathway (Polyol). The data are expressed as mean ± standard deviation.

 Condition Stage 1 Stage 2 Stage 3 Literature values References 
Glucose oxidized (%) BC-PC 0.8 ± 0.6 7.7 ± 6.7 7.3 ± 1.7 1–12 [9,34
AP-PC 9.1 ± 3.2 12.6 ± 3.8 16.7 ± 0.5 
PPP (%) BC-PC 1.9 ± 1.3 11.3 ± 8.8 2.0 ± 1.7 1–20 [9,39,40
AP-PC 4.8 ± 3.7 5.6 ± 4.5 2.8 ± 2.3 
Aerobic glycolysis (%) BC-PC 98.0 ± 0.3 86.1 ± 1.5 90.8 ± 0.2 NA  
AP-PC 92.8 ± 0.6 85.3 ± 0.7 87.7 ± 0.3 
PDH flux (%) BC-PC 0.5 ± 0.2 5.8 ± 1.5 6.9 ± 0.3 NA  
AP-PC 5.8 ± 0.7 11.7 ± 0.8 12.0 ± 0.4 
Polyol (%) BC-PC 12.5 ± 7.9 44.0 ± 22.8 8.9 ± 5.5 NA  
AP-PC 20.7 ± 12.4 23.8 ± 14.1 12.2 ± 7.7 
 Condition Stage 1 Stage 2 Stage 3 Literature values References 
Glucose oxidized (%) BC-PC 0.8 ± 0.6 7.7 ± 6.7 7.3 ± 1.7 1–12 [9,34
AP-PC 9.1 ± 3.2 12.6 ± 3.8 16.7 ± 0.5 
PPP (%) BC-PC 1.9 ± 1.3 11.3 ± 8.8 2.0 ± 1.7 1–20 [9,39,40
AP-PC 4.8 ± 3.7 5.6 ± 4.5 2.8 ± 2.3 
Aerobic glycolysis (%) BC-PC 98.0 ± 0.3 86.1 ± 1.5 90.8 ± 0.2 NA  
AP-PC 92.8 ± 0.6 85.3 ± 0.7 87.7 ± 0.3 
PDH flux (%) BC-PC 0.5 ± 0.2 5.8 ± 1.5 6.9 ± 0.3 NA  
AP-PC 5.8 ± 0.7 11.7 ± 0.8 12.0 ± 0.4 
Polyol (%) BC-PC 12.5 ± 7.9 44.0 ± 22.8 8.9 ± 5.5 NA  
AP-PC 20.7 ± 12.4 23.8 ± 14.1 12.2 ± 7.7 

Apheresis-derived PCs have different metabolic storage profiles than buffy coat PCs

A different metabolic profile was observed in AP-PCs compared with BC-PCs. Comparison of the models indicated that during the first two stages, glycolysis and pyruvate oxidation were more active in the AP-PCs, with less energy deriving from acetate oxidation (Tables 24). The increase in energy production between Stages 1 and 2 was comparatively small in AP-PCs compared with BC-PCs (Figure 4). The rate of glucose-derived pyruvate oxidation was predicted to increase substantially between the first two stages in both conditions (Table 5). However, between Stages 2 and 3, the energy production decreased in both conditions; but this decrease was more substantial in the AP-PCs, with the ATP turnover rate predicted to decrease by more than 40%. The acetate uptake rate decreased by roughly 50% between Stages 2 and 3 in both conditions. This decrease had a larger impact on the AP-PCs, as glucose was completely depleted in or before Stage 3. Glycolysis is therefore unable to compensate for the decrease in acetate consumption, resulting in a lower ATP turnover rate in AP-PC PLTs.

Glutamine and citrate can both compensate for lower pyruvate oxidation by generating acetyl-CoA via the TCA cycle. The measured glutamine uptake rate was substantially lower in AP-PCs compared with BC-PCs, resulting in a predicted lower rate of ammonia accumulation. The citrate uptake was, however, considerably higher in AP-PCs when compared with BC-PCs, in Stages 1 and 3 (Table 2).

Discussion

Recent studies have shown a correlation between PLT age and transfusion outcome [41]. Metabolism is one important contributor to PSL and can be manipulated to enhance the quality of PLTs during storage through targeted changes of PLT storage media. Employing metabolomics and standard PLT quality control screening, we previously monitored over 150 parameters in BC-PCs and AP-PCs. Here, we addressed gaps in this data set to more accurately define PLT metabolism under storage conditions. The majority proportion of PLT energy is derived from oxidative metabolism, with acetate being the preferred fuel when present in PASs [34]. Given the high concentration of acetate in T-Sol, a measure of acetate uptake was imperative for investigating and modeling PLT energy metabolism. Acetate was consumed over the duration of storage in both BC-PCs and AP-PCs at a rate comparable to previously reported values [42,43] (Figure 1 and Table 2). A comparison of buffy coat and apheresis acetate consumption rates shows that individual variation does not markedly influence the ability to metabolize acetate (Figure 1).

The data from our previous study indicated that citrate might contribute significantly to PLT central carbon metabolism [19]. The measured change in citrate concentration over time was, however, not statistically significant, possibly due to low precision of our LC–MS measurements. We aimed to accurately account for major carbon-contributing nutrients within iAT-PLT-636 and therefore remeasured the citrate during storage, which showed a significant drop in citrate concentration during storage in AP-PCs only (Table 1). 13C tracer analysis, of citrate in T-Sol with uniformly labeled 13C citrate, confirmed its metabolism in apheresis-derived PLTs. The lack of intracellular citrate and malate isotopologues, other than uniformly labeled suggest that negligible citrate is metabolized within the mitochondria, but rather metabolized to malate within the cytosol during storage. This is consistent with reduced citrate contribution to CO2 production in PLTs in the presence of acetate [32]. Combined with the acetate and citrate measurements reported here, our previous metabolomics data quantitatively define the extracellular metabolic changes of PSL. To better understand how changes in extracellular nutrients combine and contribute to and maintain intracellular energy homeostasis in PLTs during storage, we adopted a metabolic systems biology approach. It was necessary to modify iAT-PLT-636 for a more accurate interpretation of PLT metabolism during storage. These modifications allowed improved modeling of the main PLT nutrients, glucose, acetate, and glutamine/glutamate oxidation within iAT-PLT-636, on account of the addition of reactions relating to the aspartate–malate shuttle, acetate metabolism, ammonia, and more (Supplementary Tables S3 and S4). A literature review highlighted the ambiguous and variable experimental conditions from which PLT metabolic knowledge during storage originates, which is apparent from the broad range of metabolic fluxes during storage reported in the literature (Table 2). The modified model of PLT metabolism during storage represents a framework in which to define and interpret PLT metabolism.

BC-PC PLTs exhibit discrete metabolic phenotypes that represent fresh, intermediate, and expired PLTs that correlate with activation markers including GpIIb/P-selectin, sP-selectin, and sCD40L [19,20]. For comparative purposes, we modeled PSL as a three-phase decay. Systems analysis of buffy coat and apheresis PLT metabolic data using iAT-PLT-636 predicted that the requirement to produce ATP was different and reflected changes in measured nutrient utilization over time. Not surprisingly, glucose and acetate uptake, a measure of glycolytic and TCA cycle activity, respectively, within iAT-PLT-636, were responsible for the largest quantitative differences in intracellular metabolic pathway flux in PLTs irrespective of the collection procedure (Tables 2 and 4). Both models predicted a near full oxidation of acetate and decreased acetate consumption during storage in both AP-PC and BC-PC, which is reflected in the decreased contribution of acetate to total oxygen consumption between stages. This result is in good agreement with a recently published study [44] where the capacity for acetate incorporation into the acetyl-CoA was shown to decrease considerably during storage.

Most of the ATP generation in PLTs takes place in the mitochondria [18]. The metabolic models of PLT metabolism are consistent with previous reports on PSL metabolism. Prediction and literature values are compared in Tables 2, 4 and 5. Overall, model predictions correspond with previous literature values although some variability is to be expected given different PLT preparation methods, storage conditions, and experimental set-up. In previous reports, ATP production by glycolysis was assumed to be stoichiometric with lactate secretion, resulting in the overestimation of mitochondrial ATP production as a constant P/O value (moles ATP produced per mole oxygen atoms reduced by the electron transport chain) was used and the cost of acetyl-coA synthesis from acetate was not factored into calculations [3437]. Here, we have determined how the uptake of metabolic fuels is propagated during storage over time with better pathway resolution than previously reported. Our results define PSL metabolic changes at the systems level, the functional consequences of which are activated PLTs that represent lesser quality transfusion products [6]. It must be noted that the lack of lipid measurement and therefore lack of constraints to lipid metabolism in the models presents a limitation to the present study. Human PLTs do oxidize lipids [45]; however, with a high concentration of acetate in the media, lipid oxidation has been reported to account for only a negligible proportion of the total oxygen consumption [34].

The proteomic analysis allowed us to investigate whether the changes in metabolic activities are dictated by the changes in the abundances of metabolic proteins. With our analysis, we concluded that most of the metabolic enzymes either decrease in abundance or remain stable over the duration of the storage time. Furthermore, no obvious increase or decrease in protein abundances was observed within distinct metabolic pathways. This led us to the conclusion that global changes in metabolic protein abundances are not a major driving force behind the observed changes in the PLT metabolism during storage. This is in good agreement with Rijkers et al. [46] where only 21 out of the 2501 quantified proteins changed significantly during 16 days of storage measured in three BC-PCs, of which 18 decreased in abundance over time.

iAT-PLT-636 can be utilized as a hypothesis generator. Acetate addition to PLT storage counteracts lactate-induced acidification of PLT media during storage [9], and although lowering acetate concentration influences ATP maintenance, ATP has been found to be only weakly correlated with PLT viability [47]. Analysis of metabolic data with iAT-PLT-636 highlights alternative routes that may be pursued to counteract PSL. Flux alterations in glycolysis and the TCA cycle influenced the oxidation of glutamine and citrate differently, based on the collection method. In particular, citrate was more actively metabolized in AP-PCs. Citrate oxidation was allowed in our models to take into account previous reports [32], but was low compared with other oxidative fuels. The models suggest that glutamine and, to a lesser degree, citrate can contribute to ATP production and may serve as alternative ATP sources, with their efficacy being dependent on the PLT production method and media composition (e.g. inclusion of acetate).

In addition to ATP, the models account for ammonium production that influences intracellular pH. Ammonium is primarily derived from glutamine in PLTs, and model prediction of potential maximal glutamine oxidation to CO2 was comparable to literature values [48]. The models predicted ammonia accumulation during storage with a rate of 0.019–0.10 mmol/day/1012 PLTS (Table 2). Ammonia originates mostly from glutaminolysis (63.5–80.8%) and purine degradation (14.3–33.9%), with the remainder coming from asparagine metabolism (1.0–3.7%) (Supplementary Table S7). Ammonia secretion is predicted to be higher in BC-PC in all stages, on account of the higher uptake rate of glutamine.

Generation of the reducing equivalent NADPH was predicted to take place primarily in the cytosol. The percentage of total glucose shunted into the PPP during storage in T-Sol was 1.9–11.3 and contributed to 18–30% of the total cellular NADPH turnover (Supplementary Table S8). PPP activity in PLTs has previously been estimated from labeled carbon experiments and estimates vary considerably, ranging from 1% of the glucose uptake [39,40] to 20% [9]. The largest contribution to NADPH turnover is predicted to be the cytosolic isocitrate dehydrogenase reaction that accounted for 55–67% of the total NADPH turnover with its mitochondrial counterpart contributing to 6–11% of the NADPH turnover.

Changes in the NADPH turnover rate during storage were found to correlate to flux through the polyol pathway. The models predict a considerable percentage of glucose to be utilized by the polyol pathway (9–44%) (Table 5). Under euglycemic conditions, ∼3% of glucose is utilized by the polyol pathway in human erythrocytes [49]. In the rabbit lens, this ratio has been demonstrated to reach ∼30% under hyperglycemic conditions [50]. In human PLTs, the activity and abundance of AR, the enzyme catalyzing the first and rate-limiting step of the polyol pathway, increase on activation with collagen [51] and by activating glycoprotein VI with a specific monoclonal antibody [52]. Furthermore, it has been demonstrated that AR plays an important role in transducing collagen signaling in PLTs [51] and is involved in oxidative stress-associated activation of p38 mitogen-activated protein kinase (MAPK p38α) that has been strongly associated with PSL [53,54]. In light of these results, our modeling analysis suggests that flux through the polyol pathway increases during PSL in connection with enhanced glycolysis on account of mitochondrial dysfunction and suppressed ability to generate ATP.

Targeted media manipulation focused on these metabolic pathways and aimed at alternating carbon, ammonia, and NAPDH precursors may provide opportunities to manipulate intracellular pH and counteract oxidative stress. With prolonged storage, PLTs indeed show signs of mitochondrial dysfunction (Figure 1) [55] that are coupled to formation of the mitochondrial permeability transition pore and programmed cell death [56]. Disruption of oxidative metabolism increases the rate of glycolysis to compensate for a decrease in ATP turnover [57]. The decrease in oxidative metabolism in long-term stored PLTs has previously been demonstrated [58]. We have described the shifts in metabolic pathway utilization that take place in buffy coat and apheresis PLTs during storage through cell-scale modeling. We conclude that future PAS should focus on manipulation of biophysical properties such as temperature, pH, and ionic strength to suppress mechanisms that result in activation and apoptosis. Multiple PASs have been formulated with these parameters in mind [59]; however, few have managed to optimize both biochemical and biophysical properties simultaneously. The models that we have described here represent steps towards this goal.

Abbreviations

     
  • AP-PC

    apheresis-derived platelets concentrate

  •  
  • AR

    aldose reductase

  •  
  • BC-PC

    buffy coat-derived platelet concentrate

  •  
  • CPD

    citrate phosphate dextrose

  •  
  • FBA

    flux balance analysis

  •  
  • GEM

    genome-scale metabolic model

  •  
  • MDV

    mass distribution vector

  •  
  • PAS

    platelet additive solution

  •  
  • PC

    platelet concentrate

  •  
  • PCA

    principal component analysis

  •  
  • PLTs

    platelets

  •  
  • PPP

    pentose phosphate pathway

  •  
  • PSL

    platelet storage lesion

  •  
  • TCA

    tricarboxylic acid

Author Contribution

F.J. performed the experiments, analyzed the data, and wrote the paper. G.P., S.G., B.P., and O.E.S. designed the experiments and contributed to writing the paper. S.G. analyzed the data, designed the experiment, and wrote the paper. O.R. designed the experiments and wrote the paper.

Funding

This work was supported by the University of Iceland PhD grant, RANNIS [grant no. 130591-051 and 163888-0611], Landspitali The University Hospital of Iceland Research fund [grant no. A-2018-058], and European Research Council [grant proposal no. 232816].

Acknowledgments

The authors thank the staff at the blood bank and at the Center for System Biology at the University of Iceland along with Dr Joanne Connolly and Douglas Lamont for technical support.

Competing Interests

B.O.P. is the co-founder of Sinopia Biosciences.

References

References
1
Zucker
,
M.B.
and
Nachmias
,
V.T.
(
1985
)
Platelet activation
.
Arterioscler. Thromb. Vasc. Biol.
5
,
2
18
2
Gawaz
,
M.
,
Langer
,
H.
and
May
,
A.E.
(
2005
)
Platelets in inflammation and atherogenesis
.
J. Clin. Invest.
115
,
3378
3384
3
Seghatchian
,
J.
and
Krailadsiri
,
P.
(
1997
)
The platelet storage lesion
.
Transfus. Med. Rev.
11
,
130
144
4
Devine
,
D.V.
and
Serrano
,
K.
(
2010
)
The platelet storage lesion
.
Clin. Lab. Med.
30
,
475
487
5
Mittal
,
K.
and
Kaur
,
R.
(
2015
)
Platelet storage lesion: an update
.
Asian J. Transfus. Sci.
9
,
1
3
6
Sahler
,
J.
,
Grimshaw
,
K.
,
Spinelli
,
S.L.
,
Refaai
,
M.A.
,
Phipps
,
R.P.
and
Blumberg
,
N.
(
2011
)
Platelet storage and transfusions: new concerns associated with an old therapy
.
Drug Discovery Today: Dis. Mech.
8
,
e9
e14
7
Murphy
,
S.
and
Gardner
,
F.H.
(
1975
)
Platelet storage at 22°C: role of gas transport across plastic containers in maintenance of viability
.
Blood
46
,
209
218
PMID:
[PubMed]
8
Gulliksson
,
H.
,
Sallander
,
S.
,
Pedajas
,
I.
,
Christenson
,
M.
and
Wiechel
,
B.
(
1992
)
Storage of platelets in additive solutions: a new method for storage using sodium chloride solution
.
Transfusion
32
,
435
440
9
Murphy
,
S.
(
1995
)
The oxidation of exogenously added organic anions by platelets facilitates maintenance of pH during their storage for transfusion at 22 degrees C
.
Blood
85
,
1929
1935
PMID:
[PubMed]
10
Gulliksson
,
H.
(
1993
)
Storage of platelets in additive solutions: the effect of citrate and acetate in in vitro studies
.
Transfusion
33
,
301
303
11
Gulliksson
,
H.
(
2003
)
Defining the optimal storage conditions for the long-term storage of platelets
.
Transfus. Med. Rev.
17
,
209
215
12
McCullough
,
J.
,
Goldfinger
,
D.
,
Gorlin
,
J.
,
Riley
,
W.J.
,
Sandhu
,
H.
,
Stowell
,
C.
et al. 
(
2015
)
Cost implications of implementation of pathogen-inactivated platelets
.
Transfusion
55
,
2312
2320
13
Zimring
,
J.C.
,
Slichter
,
S.
,
Odem-Davis
,
K.
,
Felcyn
,
J.R.
,
Kapp
,
L.M.
,
Bell
,
L.N.
et al. 
(
2016
)
Metabolites in stored platelets associated with platelet recoveries and survivals
.
Transfusion
56
,
1974
1983
14
O'Brien
,
E.J.
,
Monk
,
J.M.
and
Palsson
,
B.O.
(
2015
)
Using genome-scale models to predict biological capabilities
.
Cell
161
,
971
987
15
Agren,
R.
,
Mardinoglu,
A.
,
Asplund,
A.
,
Kampf,
C.
,
Uhlen,
M.
and
Nielsen,
J
. (
2014
)
Identification of anticancer drugs for hepatocellular carcinoma through personalized genome-scale metabolic modeling
.
Mol. Syst. Biol.
10
,
721
16
Bordbar
,
A.
and
Palsson
,
B.O.
(
2012
)
Using the reconstructed genome-scale human metabolic network to study physiology and pathology
.
J. Intern. Med.
271
,
131
141
17
Thomas
,
A.
,
Rahmanian
,
S.
,
Bordbar
,
A.
,
Palsson
,
B.
and
Jamshidi
,
N.
(
2014
)
Network reconstruction of platelet metabolism identifies metabolic signature for aspirin resistance
.
Sci. Rep.
4
,
3925
PMID:
[PubMed]
18
Bordbar
,
A.
,
Yurkovich
,
J.T.
,
Paglia
,
G.
,
Rolfsson
,
O.
,
Sigurjónsson
,
Ó.E.
and
Palsson
,
B.O.
(
2017
)
Elucidating dynamic metabolic physiology through network integration of quantitative time-course metabolomics
.
Sci. Rep.
7
,
46249
19
Paglia
,
G.
,
Sigurjonsson
,
Ó.E.
,
Rolfsson
,
Ó.
,
Valgeirsdottir
,
S.
,
Hansen
,
M.B.
,
Brynjólfsson
,
S.
et al. 
(
2014
)
Comprehensive metabolomic study of platelets reveals the expression of discrete metabolic phenotypes during storage
.
Transfusion
54
,
2911
2923
20
Paglia
,
G.
,
Sigurjónsson
,
Ó.E.
,
Rolfsson
,
O.
,
Hansen
,
M.B.
,
Brynjólfsson
,
S.
,
Gudmundsson
,
S.
et al. 
(
2015
)
Metabolomic analysis of platelets during storage: a comparison between apheresis- and buffy coat-derived platelet concentrates
.
Transfusion
55
,
301
313
21
Rolfsson
,
Ó.
,
Sigurjonsson
,
Ó.E.
,
Magnusdottir
,
M.
,
Johannsson
,
F.
,
Paglia
,
G.
,
Guðmundsson
,
S.
et al. 
(
2017
)
Metabolomics comparison of red cells stored in four additive solutions reveals differences in citrate anticoagulant permeability and metabolism
.
Vox Sang.
112
,
326
335
22
Rolfsson,
Ó.
,
Johannsson,
F.
,
Magnusdottir,
M.
,
Paglia,
G.
,
Sigurjonsson,
Ó.E
,
Bordbar,
A.
et al. 
. (
2017
)
Mannose and fructose metabolism in red blood cells during cold storage in SAGM
.
Transfusion
57
,
2665
2676
23
Millard
,
P.
,
Letisse
,
F.
,
Sokol
,
S.
and
Portais
,
J.-C.
(
2012
)
Isocor: correcting MS data in isotope labeling experiments
.
Bioinformatics
28
,
1294
1296
24
Cox,
J.
and
Mann,
M
.
Maxquant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification
.
Nat. Biotechnol.
2008
;
26
:
1367
PMID:
[PubMed]
25
Xia
,
J.
,
Sinelnikov
,
I.V.
,
Han
,
B.
and
Wishart
,
D.S.
(
2015
)
Metaboanalyst 3.0 — making metabolomics more meaningful
.
Nucleic Acids Res.
43
,
W251
W2W7
26
Fabregat
,
A.
,
Sidiropoulos
,
K.
,
Garapati
,
P.
,
Gillespie
,
M.
,
Hausmann
,
K.
,
Haw
,
R.
et al. 
(
2016
)
The reactome pathway knowledgebase
.
Nucleic Acids Res.
44
,
D481
D487
27
Halldorsson
,
S.
,
Rohatgi
,
N.
,
Magnusdottir
,
M.
,
Choudhary
,
K.S.
,
Gudjonsson
,
T.
,
Knutsen
,
E.
et al. 
(
2017
)
Metabolic re-wiring of isogenic breast epithelial cell lines following epithelial to mesenchymal transition
.
Cancer Lett.
396
,
117
129
28
Schellenberger
,
J.
,
Que
,
R.
,
Fleming
,
R.M.
,
Thiele
,
I.
,
Orth
,
J.D.
,
Feist
,
A.M.
et al. 
(
2011
)
Quantitative prediction of cellular metabolism with constraint-based models: the COBRA Toolbox v2.0
.
Nat. Protoc.
6
,
1290
1307
29
King
,
Z.A.
,
Dräger
,
A.
,
Ebrahim
,
A.
,
Sonnenschein
,
N.
,
Lewis
,
N.E.
,
Palsson
,
B.O.
et al. 
(
2015
)
Escher: a web application for building, sharing, and embedding data-rich visualizations of biological pathways
.
PLoS Comput. Biol.
11
,
e1004321
30
Orth
,
J.D.
,
Thiele
,
I.
and
Palsson
,
B.O.
(
2010
)
What is flux balance analysis?
Nat. Biotechnol.
28
,
245
248
31
Schellenberger
,
J.
and
Palsson
,
B.O.
(
2009
)
Use of randomized sampling for analysis of metabolic networks
.
J. Biol. Chem.
284
,
5457
5461
32
Cartledge
,
S.
,
Candy
,
D.J.
and
Hawker
,
R.J.
(
1997
)
Citrate metabolism by human platelets
.
Transfus. Med.
7
,
211
215
33
Niu
,
X.
,
Whisson
,
M.E.
and
Guppy
,
M.
(
1997
)
Types and sources of fuels for platelets in a medium containing minimal added fuels and a low carryover plasma
.
Br. J. Haematol.
97
,
908
916
34
Guppy
,
M.
,
Whisson
,
M.E.
,
Sabaratnam
,
R.
,
Withers
,
P.
and
Brand
,
K.
(
1990
)
Alternative fuels for platelet storage: a metabolic study
.
Vox Sang.
59
,
146
152
35
Kilkson
,
H.
,
Holme
,
S.
and
Murphy
,
S.
(
1984
)
Platelet metabolism during storage of platelet concentrates at 22°C
.
Blood
64
,
406
414
PMID:
[PubMed]
36
Niu
,
X.
,
Arthur
,
P.
,
Abas
,
L.
,
Whisson
,
M.
and
Guppy
,
M.
(
1996
)
Carbohydrate metabolism in human platelets in a low glucose medium under aerobic conditions
.
Biochim. Biophys. Acta
1291
,
97
106
37
Guppy
,
M.
,
Abas
,
L.
,
Neylon
,
C.
,
Whisson
,
M.E.
,
Whitham
,
S.
,
Pethick
,
D.W.
et al. 
(
1997
)
Fuel choices by human platelets in human plasma
.
Eur. J. Biochem.
244
,
161
167
PMID:
[PubMed]
38
Shimizu
,
T.
and
Murphy
,
S.
(
1993
)
Roles of acetate and phosphate in the successful storage of platelet concentrates prepared with an acetate-containing additive solution
.
Transfusion
33
,
304
310
39
Murphy
,
S.
and
Gardner
,
F.H.
(
1971
)
Platelet storage at 22°C; metabolic, morphologic, and functional studies
.
J. Clin. Invest.
50
,
370
377
40
Baker
,
J.M.
,
Candy
,
D.J.
and
Hawker
,
R.J.
(
2001
)
Influences of pH on human platelet metabolism
.
Platelets
12
,
333
342
41
Kreuger
,
A.L.
,
Caram-Deelder
,
C.
,
Jacobse
,
J.
,
Kerkhoffs
,
J.-L.
,
van der Bom
,
J.G.
and
Middelburg
,
R.A.
(
2017
)
Effect of storage time of platelet products on clinical outcomes after transfusion: a systematic review and meta-analyses
.
Vox Sang.
112
,
291
300
42
Johnson
,
L.
,
Schubert
,
P.
,
Tan
,
S.
,
Devine
,
D.V.
and
Marks
,
D.C.
(
2016
)
Extended storage and glucose exhaustion are associated with apoptotic changes in platelets stored in additive solution
.
Transfusion
56
,
360
368
43
van der Meer
,
P.F.
,
Kerkhoffs
,
J.-L.
,
Curvers
,
J.
,
Scharenberg
,
J.
,
de Korte
,
D.
,
Brand
,
A.
et al. 
(
2010
)
In vitro comparison of platelet storage in plasma and in four platelet additive solutions, and the effect of pathogen reduction: a proposal for an in vitro rating system
.
Vox Sang.
98
,
517
524
44
Sims
,
C.
,
Salliant
,
N.
,
Worth
,
A.J.
,
Parry
,
R.
,
Mesaros
,
C.
,
Blair
,
I.A.
et al. 
(
2017
)
Metabolic tracing analysis reveals substrate-specific metabolic deficits in platelet storage lesion
.
Transfusion
57
,
2683
2689
45
Slatter
,
D.A.
,
Aldrovandi
,
M.
,
O'Connor
,
A.
,
Allen
,
S.M.
,
Brasher
,
C.J.
,
Murphy
,
R.C.
et al. 
(
2016
)
Mapping the human platelet lipidome reveals cytosolic phospholipase A2 as a regulator of mitochondrial bioenergetics during activation
.
Cell Metab.
23
,
930
944
46
Rijkers
,
M.
,
van den Eshof
,
B.L.
,
van der Meer
,
P.F.
,
van Alphen
,
F.P.J.
,
de Korte
,
D.
,
Leebeek
,
F.W.G.
et al. 
(
2017
)
Monitoring storage induced changes in the platelet proteome employing label free quantitative mass spectrometry
.
Sci. Rep.
7
,
105
47
Holme
,
S.
(
1998
)
Storage and quality assessment of platelets
.
Vox Sang.
74
,
207
216
48
Murphy
,
S.
,
Munoz
,
S.
,
Parry-Billings
,
M.
and
Newsholme
,
E.
(
1992
)
Amino acid metabolism during platelet storage for transfusion
.
Br. J. Haematol.
81
,
585
590
49
Morrison
,
A.D.
,
Clements
,
R.S.
,
Travis
,
S.B.
,
Oski
,
F.
and
Winegrad
,
A.I.
(
1970
)
Glucose utilization by the polyol pathway in human erythrocytes
.
Biochem. Biophys. Res. Commun.
40
,
199
205
50
González
,
R.G.
,
Barnett
,
P.
,
Aguayo
,
J.
,
Cheng
,
H.-M.
and
Chylack
,
L.T.
(
1984
)
Direct measurement of polyol pathway activity in the ocular lens
.
Diabetes
33
,
196
199
51
Tang
,
W.H.
,
Stitham
,
J.
,
Gleim
,
S.
,
Di Febbo
,
C.
,
Porreca
,
E.
,
Fava
,
C.
et al. 
(
2011
)
Glucose and collagen regulate human platelet activity through aldose reductase induction of thromboxane
.
J. Clin. Invest.
121
,
4462
4476
52
Schulz
,
C.
,
Leuschen
,
N.V.
,
Fröhlich
,
T.
,
Lorenz
,
M.
,
Pfeiler
,
S.
,
Gleissner
,
C.A.
et al. 
(
2010
)
Identification of novel downstream targets of platelet glycoprotein VI activation by differential proteome analysis: implications for thrombus formation
.
Blood
115
,
4102
4110
53
Skripchenko,
A.
,
Awatefe
,
H.
,
Thompson-Montgomery
,
D.
,
Myrup
,
A.
,
Turgeon
,
A.
and
Wagner
,
S.J.
(
2013
)
An inhibition of p38 mitogen activated protein kinase delays the platelet storage lesion
.
PLoS ONE
8
,
e70732
54
Canault
,
M.
,
Duerschmied
,
D.
,
Brill
,
A.
,
Stefanini
,
L.
,
Schatzberg
,
D.
,
Cifuni
,
S.M.
et al. 
(
2010
)
p38 mitogen-activated protein kinase activation during platelet storage: consequences for platelet recovery and hemostatic function in vivo
.
Blood
115
,
1835
1842
55
Albanyan
,
A.-M.
,
Harrison
,
P.
and
Murphy
,
M.F.
(
2009
)
Markers of platelet activation and apoptosis during storage of apheresis- and buffy coat–derived platelet concentrates for 7 days
.
Transfusion
49
,
108
117
56
Leytin
,
V.
,
Allen
,
D.J.
,
Mutlu
,
A.
,
Gyulkhandanyan
,
A.V.
,
Mykhaylov
,
S.
and
Freedman
,
J.
(
2009
)
Mitochondrial control of platelet apoptosis: effect of cyclosporin A, an inhibitor of the mitochondrial permeability transition pore
.
Lab. Invest.
89
,
374
384
57
Guppy
,
M.
,
Abas
,
L.
,
Arthur
,
P.G.
and
Whisson
,
M.E.
(
1995
)
The Pasteur effect in human platelets: implications for storage and metabolic control
.
Br. J. Haematol.
91
,
752
757
58
Ravi,
S.
,
Chacko,
B.
,
Kramer,
P.A.
,
Sawada,
H.
,
Johnson,
M.S
,
Zhi,
D.
et al. 
. (
2015
)
Defining the effects of storage on platelet bioenergetics: the role of increased proton leak
.
Biochim. Biophys. Acta
.
1852
,
2525
2534
59
Gulliksson
,
H.
(
2014
)
Platelet storage media
.
Vox Sang.
107
,
205
212
60
Zhang
,
J.G.
,
Carter
,
C.J.
,
Culibrk
,
B.
,
Devine
,
D.V.
,
Levin
,
E.
,
Scammell
,
K.
et al. 
(
2008
)
Buffy-coat platelet variables and metabolism during storage in additive solutions or plasma
.
Transfusion
48
,
847
856
61
Holme
,
S.
,
Heaton
,
W.A.
and
Courtright
,
M.
(
1987
)
Platelet storage lesion in second-generation containers: correlation with platelet ATP levels
.
Vox Sang.
53
,
214
220
62
Klinger
,
M.H.
,
Josch
,
M.
and
Klüter
,
H.
(
1996
)
Platelets stored in a glucose-free additive solution or in autologous plasma — an ultrastructural and morphometric evaluation
.
Vox Sang.
71
,
13
20
63
Hornsey
,
V.S.
,
McColl
,
K.
,
Drummond
,
O.
,
McMillan
,
L.
,
Morrison
,
A.
,
Morrison
,
L.
et al. 
(
2006
)
Extended storage of platelets in SSP+ platelet additive solution
.
Vox Sang.
91
,
41
46
64
Murphy
,
S.
,
Shimizu
,
T.
and
Miripol
,
J.
(
1995
)
Platelet storage for transfusion in synthetic media: further optimization of ingredients and definition of their roles
.
Blood
86
,
3951
3960
PMID:
[PubMed]
65
Edenbrandt
,
C.M.
and
Murphy
,
S.
(
1990
)
Adenine and guanine nucleotide metabolism during platelet storage at 22°C
.
Blood
76
,
1884
1892
PMID:
[PubMed]
66
Ukrainski
,
C.T.
,
Goldfinger
,
D.
,
Pomerance
,
J.J.
,
Lee
,
H.C.
,
Farber
,
S.
and
Sanchez
,
R.
(
1981
)
Ammonia accumulation in platelet concentrates during storage
.
Transfusion
21
,
113
117