BLOOD COMPONENTS Comprehensive metabolomic study of platelets reveals the expression of discrete metabolic phenotypes during storage Giuseppe Paglia,1 Ólafur E. Sigurjónsson,2,3 Óttar Rolfsson,1 Soley Valgeirsdottir,1 Morten Bagge Hansen,4 Sigurður Brynjólfsson,1 Sveinn Gudmundsson,2 and Bernhard O. Palsson1
BACKGROUND: Platelet (PLT) concentrates are routinely stored for 5 to 7 days. During storage they exhibit what has been termed PLT storage lesion (PSL), which is evident by a loss of hemostatic function when transfused into patients. The overall goal of this study was to obtain a comprehensive data set describing PLT metabolism during storage. STUDY DESIGN AND METHODS: The experimental approach adopted to achieve this goal combined a series of standard assays to monitor the quality of stored PLTs and a deep-coverage metabolomics study using liquid chromatography coupled with mass spectrometry performed on both the extracellular and the intracellular environments. During storage we measured 174 different variables in 6 PLT units, collected by apheresis. Samples were collected at eight different time points resulting in a data set containing more than 8000 measurements. RESULTS: Stored PLTs did not undergo a monotonic decay, but experienced systematic changes in metabolism reflected in three discrete metabolic phenotypes: The first (Days 0-3) was associated with active glycolysis, pentose phosphate pathway, and glutathione metabolism and down regulation of tricarboxylic acid (TCA) cycle. The second (Days 4-6) was associated with a more active TCA cycle as well as increased purine metabolism. A third metabolic phenotype of less clinical relevance (Days 7-10) was associated with a faster decay of cellular metabolism. CONCLUSION: PSL is not associated with a linear decay of metabolism, but rather with successive metabolic shifts. These findings may give new insight into the mechanisms underlying PSL and encourage the deployment of systems biology methods to PSL.
P
latelet concentrates (PCs) may be used in many clinical situations, including treatment of hemorrhages, inborn or acquired defects of platelet (PLT) function, or as prophylaxis in patients with thrombocytopenia after chemotherapy.1 In recent decades PC transfusion has significantly improved with methods such as leukoreduction for removing contaminating white blood cells (WBCs), improved PLT additive solutions (PASs), and the practice of using gas-permeable storage containers.2-4 Nevertheless, during storage, PLTs develop a condition known as PLT storage lesion (PSL). In addition to the PSL, risk of bacterial contamination during storage has limited the use of PLTs beyond 5 days of storage. With the additional requirement of bacterial surveillance or pathogen inactivation, many countries allow PLT transfusion with up to 7 days of storage.5,6 Considering the important role of PLT transfusion in modern medicine, an improvement in quality of the
ABBREVIATIONS: HCA = hierarchical cluster analysis; LC = liquid chromatography; MS = mass spectrometry; PC(s) = platelet concentrate(s); PCA = principal component analysis; PPP = pentose phosphate pathway; PSL = platelet storage lesion; sP-selectin = soluble P-selectin; TCA = tricarboxylic acid. From the 1Center for Systems Biology, University of Iceland; 2 The Blood Bank, Landspitali-University Hospital; and the 3 School of Science and Engineering, Reykjavik University, Reykjavik, Iceland; and the 4Department of Clinical Immunology, Rigshospitalet, Copenhagen University Hospital, Copenhagen, Denmark. Address reprint requests to: Ólafur E. Sigurjónsson, The Blood Bank, Landspitali University Hospital, Snorrabraut 60, 105 Reykjavik, Iceland; e-mail: [email protected]. This work was supported by the European Research Council Grant Proposal No. 232816. Received for publication November 14, 2013; revision received February 22, 2014, and accepted February 25, 2014. doi: 10.1111/trf.12710 © 2014 AABB TRANSFUSION 2014;54:2911-2923. Volume 54, November 2014 TRANSFUSION
2911
PAGLIA ET AL.
PLTs at the end of storage or an extension of the present expire criteria could have an important impact on modern blood banking.7 PSL is thought to be the result of a combination of several factors, including the methods of collection, processing, storage, and manipulation after collection. This lesion causes a combination of changes in PLT morphology and metabolism, physiologic activation, and release of PLT granules,8 resulting in deteriorated PLT function during storage.9,10 A model of linear decay over storage time has traditionally been suggested due to evidence of a constant consumption of glucose and a constant release of lactate. Accumulation of lactate causes metabolic acidosis and the low pH, in turn, inhibits PLT function and leads to inferior PLT quality upon transfusion.11-13 A recent retrospective study confirmed the association of the quality of PCs with time of storage by comparing the effect of transfused PLTs stored for 4 and 5 days, suggesting that a restriction of storage time from 5 to 4 days may result in an improvement of the clinical outcomes.14 PLT quality is routinely assessed by swirling, PLT number, lactate accumulation, pH, PLT surface markers, and functional assays such as in vitro aggregometry assays. The use of integrated omics and systems biology approaches may become a recognized strategy in transfusion medicine in the future.15-23 The development of new technologies facilitates the collection of more comprehensive data sets that may enable a better understanding of the metabolism of blood cells during storage. The understanding of PLT metabolism during storage has two major goals: 1) illuminating the profound mechanism behind the storage lesions and 2) discovering pathways and biomarkers to better define the quality of PLTs during storage. A better understanding of PLT metabolism is one of the prerequisites for increasing the quality and function of PLTs as well as decreasing adverse events in patients after transfusion.24 The totality of all metabolites at any given moment, known as the metabolome, is a chemical readout of the health of the cell, as changes in the concentrations of specific groups of metabolites are sensitive and specific to pathologically and physiologically relevant factors, such as genetic variations and diet.25,26 The metabolome comprises the endometabolome (all intracellular metabolites) and the exometabolome (all metabolites released into the extracellular fluid).27 Here we present the first comprehensive metabolomics study on stored PLTs. The strategy adopted combines liquid chromatography (LC)-mass spectrometry (MS)28,29 with several assays to define distinct stages of stored PLTs. The exo- and endometabolome were obtained during 10 days of storage, resulting in the most comprehensive metabolomics data set for stored PLTs yet published. 2912
TRANSFUSION Volume 54, November 2014
MATERIALS AND METHODS Sample collection PCs were harvested from healthy blood donors by apheresis using an apheresis separator instrument (Amicus 4R580, Fenwal, Zurich, Switzerland) at the Blood Bank, Landspitali-University Hospital, Iceland. In total, 6 units of apheresis PLTs were obtained from six donors of mixed age, sex, and blood types using an apheresis kit (Amicus, Fenwal). PLTs were stored in T-SOL (Fenwal) in a plastic container (PL 2410, Fenwal). Sampling was done on Day 0 (day of collection by apheresis) and subsequently on Days 1, 3, 4, 5, 6, 7, and 10. On Day 0 sampling was carried out 2 hours after collection. Units were stored in a shaking incubator set at 22°C and were discarded after completion of sampling on Day 10. All units were checked for bacterial contamination on Day 10. The National Bioethics Committee of Iceland and the Icelandic Data Protection Authority approved the study.
Blood bank quality controls Immediately after sample collection, a blood gas analyzer (ABL90 FLEX, Radiometer Medical ApS, Copenhagen, Denmark) was used for determination of pH, pO2, and pCO2; concentration of total hemoglobin; and concentrations of K+, Na+, Cl–, glucose, and lactate in the PLT media. PLT concentration, mean PLT volume, plateletcrit, PLT distribution width, and WBC count was determined using an automated hematology analyzer (CELL-DYN Ruby, Abbott Diagnostics, Abbott Park, IL). Flow cytometry (FACSCalibur, Becton Dickinson, La Jolla, CA) was used to determine expression of CD41, CD42b, CD62P, CD63, annexin V (all from Becton Dickinson), and JC1 to indicate the level of mitochondrial polarization (Life Technologies, Boston, MA). Lactate dehydrogenase (LDH) activity in the PC medium was assessed by an LDH assay kit (ab102526, Abcam, Cambridge, UK). The soluble CD40 ligand concentration was determined using an enzyme-linked immunosorbent assay (ELISA) kit (DCDL40, Quantikine, R&D Systems, Minneapolis, MN). Soluble P-selectin (sP-selectin) concentration was determined via ELISA (BBE6, R&D Systems, Abingdon, UK). Aliquots of 0.25 mL were used to determine the adenosine triphosphate (ATP) concentration employing a luminescent cell viability assay (CellTiter-Glo, Promega, Madison, WI) using a microplate reader (Spectramax M3, Molecular Devices, Sunnyvale, CA). For more detailed methods refer to Appendix S1 (available as supporting information in the online version of this paper).
Metabolomics Sample preparation Supernatant and cells were separated by centrifugation (1600 × g, 22°C, 5 min) of 0.5 mL of PC. Cell pellets were
METABOLIC PROPERTIES OF STORED PLTs
washed twice by adding 1 mL of phosphate-buffered saline, and after centrifugation (1600 × g, 22°C, 5 min), the supernatant was discharged. Thirty microliters of labeled internal standard mixture (Appendix S1) and 1 mL of cold (−20°C) methanol: water (7:3) were added to the cell pellets. Cell lysis was achieved by performing two consecutive freeze-and-thaw steps. Samples were centrifuged (15,000 × g, 4°C, 15 min) and supernatant was transferred to a new tube. One milliliter of cold (−20°C) methanol:water (7:3) was added to the pellets. Samples were mixed vigorously for 1 minute and after centrifugation (15,000 × g, 4°C, 15 min) the supernatant was added to the precedent. Samples were dried using a vacuum concentrator, reconstituted in 300 μL H2O:ACN (50:50), and filtered by centrifugation using a Pierce protein 96-well precipitation plate (2000 × g, 4°C, 15 min). Two-hundred microliters of medium sample was processed by adding 30 μL of labeled internal standard mixture (Appendix S1) and 0.5 mL of MeOH. Samples were mixed vigorously and centrifuged (15,000 × g, 4°C, 15 min). Supernatant was transferred into a new tube and dried using a vacuum concentrator. Samples were reconstituted in 600 μL of H2O:ACN (50:50) and filtered by centrifugation using a Pierce protein 96-well precipitation plate (2000 × g, 4°C, 15 min). Pooled QC samples were prepared for both extracellular and intracellular analysis by pooling 20 μL for each processed sample. Spiked pooled QC samples were prepared by spiking pooled QC samples with a mix containing 17 standard compounds (Appendix S1).
Ultraperformance LC-MS Metabolomic analysis was performed using a modified version of a previously reported method.28,29 Ultraperformance liquid chromatography (Acquity, Waters, Manchester, UK) was coupled with a quadrupole time-offlight mass spectrometer (Synapt G2, Waters). Chromatographic separation was achieved by hydrophilic interaction liquid chromatography (HILIC) using an Acquity amide column, 1.7 μm (2.1 × 150 mm; Waters). All intracellular and extracellular samples were analyzed three times: once in positive ionization mode and twice in negative ionization mode using acidic and basic chromatographic conditions. In positive and in negative acidic conditions, Mobile Phase A was 100% ACN and B was 100% H2O, both containing 0.1% formic acid. The following elution gradient was used: 0 minutes 99% A, 7 minutes 30% A, 7.1 minutes 99% A, and 10 minutes 99% A. In negative mode basic conditions, Mobile Phase A contained 10 mmol/L ACN:sodium bicarbonate (95:5) and Mobile Phase B contained 10 mmol/L ACN:sodium bicarbonate (5:95). The following elution gradient was used: 0 minutes 99% A, 6 minutes 30% A, 6.5 minutes 99% A, and 10 minutes 99% A. In all conditions, the flow rate was
0.4 mL/min, the column temperature was 45°C, and the injection volume was 3.5 μL. In all conditions, the mass spectrometer operated using a capillary voltage of 1.5 kV, and the sampling cone and the extraction cone were of 30 and 5 V, respectively. The cone and the desolvation gas flow were 50 and 800 L/ hr, respectively, while the source and desolvation gas temperatures were 120 and 500°C, respectively. MS spectra were acquired in centroid mode from m/z 50 to 1000 using scan time of 0.3 seconds. Leucine enkephalin (2 ng/μL) was used as lock mass (m/z 556.2771 and 554.2615 in positive and negative experiments, respectively). A typical analytical run consisted of pooled QC samples to equilibrate the system, calibrators, samples, and spiked pooled QC samples.29 Stability of the LC-MS system during analysis was validated by analyzing spiked pooled QC samples every 10 samples. Spiked pooled QC samples had a relative standard deviation (RSD%) for the retention times never higher than 2. Intra- and interbatch variation of signal intensities provided a RSD% never higher than 13 (Table S6, available as supporting information in the online version of this paper).
Data processing The identification of unexpected metabolites was achieved by integration, alignment, and conversion of MS data points into exact mass retention time pairs (MarkerLynx, v4.1, Waters). The identity of the unexpected metabolites was established by verifying peak retention time, accurate mass measurements, and tandem MS against our in-house database and/or online databases, including HMDB30 and METLIN.31 TargetLynx (v4.1, Waters) was used to integrate chromatograms of targeted metabolites. Ion chromatograms were extracted using a 0.02-mDa window centered on the expected m/z for each targeted compound. Each metabolite was normalized using a proper isotopically labeled internal standard as described in Table S4 (available as supporting information in the online version of this paper). The ratio metabolites/internal standard was then used for deriving concentration using external calibration with reference standards.
Statistical analysis Normalization was performed to minimize differences between two different experiments (Experiment 1 [Exp1], Units 1-3; Experiment 2 [Exp2], Units 4-6). Normalized Exp2 (Exp2″) was obtained by applying the following formula to each variable:
Exp2′′ = (Exp2 − average Exp2) + average Exp1. Concentrations of extracellular measurements were computed to obtain consumption and secretion rates (ΔC/Δt) during the storage. Volume 54, November 2014 TRANSFUSION
2913
2914
TRANSFUSION Volume 54, November 2014
Day 7 263 ± 3 6.78 ± 0.08 4.1 ± 0.2§ 4.3 ± 0.8 16.9 ± 1.0 0.05 ± 0.06 1170 ± 126 41.7 ± 7.7 23.4 ± 6.9 Day 6 267 ± 3 6.80 ± 0.06 4.1 ± 0.2§ 4.4 ± 1.2 17.0 ± 0.9 0.06 ± 0.05 1180 ± 240 40.8 ± 7.6 24.9 ± 4.7 Day 5 272 ± 3 6.82 ± 0.06 4.0 ± 0.1§ 4.3 ± 0.7 17.1 ± 0.8 0.05 ± 0.05 1145 ± 127 40.3 ± 7.1 24.9 ± 5.6 Day 4 276 ± 3 6.85 ± 0.04 4.4 ± 0.5§ 4.7 ± 1.5 17.1 ± 0.8 0.09 ± 0.1 1220 ± 191 41.3 ± 6.9 25.4 ± 5.4 Day 3 281 ± 3 6.86 ± 0.03 4.1 ± 0.2§ 5.5 ± 1.7 17.3 ± 0.7 0.10 ± 0.14 1368 ± 250 43.6 ± 10.2 24.6 ± 6.2
TABLE 1. Selected blood bank PLT QC variables*
Unless otherwise noted, results are presented as mean ± standard error (n = 6). Percentage of the total number of PLTs that satisfy the conditions specified under “variable.” n = 5. n = 3.
Sample collection was designed and optimized considering the maximum sample volume that was possible to collect from each unit without affecting the storage of PLTs. The total samples collected in all experiments were on average approximately 10% of the initial weight (Table 1). Blood bank QCs were selected to define the quality of stored PLTs, applying analyses and methods published elsewhere.34 The quality of the PLTs used in the experiments was in conformance with other publications.35-37 The percentage of PLTs expressing the activation markers CD62P increased in the first 4 days, probably due to activation triggered by the apheresis process, and then stayed constant until Day 7, with an additional increase in
* † ‡ §
Sampling and blood bank QCs
Day 1 286 ± 3 6.91 ± 0.05 4.2 ± 0.2‡ 5.0 ± 0.7 17.5 ± 0.8 0.18 ± 0.26 1253 ± 123 45.8 ± 10.2 19.5 ± 7.7
The overall goal of this work was to obtain a better and more comprehensive understanding of PLT metabolism during storage conditions. The experimental approach designed to reach this goal involved a four-step procedure: 1) sampling, 2) blood bank quality control (QC), 3) metabolomics, and 4) data analysis. We monitored 174 different variables for each of the six stored PLT containers at each of the eight time points, resulting in a data set containing more than 8000 measurements. To date, this represents the most comprehensive data set describing stored PLTs.
Day 0 291 ± 2 6.96 ± 0.08 4.9 ± 0.9 4.9 ± 0.5 18.2 ± 0.8 0.94 ± 0.78 1026 ± 162 36.9 ± 11.5 6.9 ± 1.6
RESULTS
Variable† PC volume/bag (mL) pH (22°C) Mean PLT volume (fL) Plateletcrit (mL/L) PLT distribution width (10 GSD) WBC concentration (×109 cells/L) PLT concentration (×109 cells/L) CD41+; CD62P+ (%)† CD41+; CD63+ (%)†
Pathway analysis was performed using metaboanalyst.32 This approach combines pathway enrichment and pathway topology analysis. Every variable having an HMDB ID was included in the analysis using the Homo sapiens KEGG library. The pathway enrichment analysis method was global test and the pathway topology analysis used relative betweenness centrality. The fold change, normalized to Day 0, was calculated for each variable monitored and one way analysis of variance (ANOVA) test was used to find measurements presenting a significant change with the time of storage (p < 0.05). Principal component analysis (PCA; EZinfoUmetrics) was performed on 104 measurements (p < 0.05 one-way ANOVA test). Before PCA, data were scaled (unit variance scaling). Selected metabolites (based on pathway analysis and ANOVA test < 0.05) were analyzed via hierarchical cluster analysis (HCA).33 Pearson correlation was used on both column (time points) and row (metabolites). Both column and row were normalized by subtracting the mean and data were log-transformed before clustering. The average of each time points was used for this analysis.
Day 10 258 ± 3 6.81 ± 0.09 4.0 ± 0.1§ 4.5 ± 0.7§ 17.6 ± 0.5§ 0.15 ± 0.32 910 ± 204 48.0 ± 9.1 27.3 ± 6.1
PAGLIA ET AL.
METABOLIC PROPERTIES OF STORED PLTs
the last days of storage. The activation marker CD63 had a similar profile, confirming activation of PLTs in the first 4 days of storage (Table 1). PLT count increased during the first 4 days. This profile might reflect a previously described artifact of PLT aggregation with subsequent resolution of the aggregates resulting in apparent increase in the number of PLTs.37,38 Stored PLTs may also demonstrate an increase in PLT count arising from PLT division in the concentrate as has been reported previously.39 Bacterial contamination was not detected in any of the units on Day 10.
Metabolomics A metabolomics approach, based on ultraperformance LC–hydrophilic interaction LC–quadrupole time-offlight–MS strategy, was applied to both extracellular and intracellular environments. This strategy includes a targeted approach of selected polar metabolites, but it also has the ability to identify unexpected metabolites (Tables S4 and S5).28,29 Targeted metabolites were selected based on preliminary investigations on PCs.28 Metabolomic analysis identified 49 metabolites in the exometabolome and 96 metabolites in the endometabolome (Tables S4 and S5). During storage, a continuous exchange of metabolites occurred between the extracellular and intracellular environments. PLTs are expected to use plasma metabolites to sustain their metabolism. Indeed, glucose and glutamine from an extracellular source were the main metabolic fuels, showing the largest decreases in the exometabolome. Nevertheless, the most significant compositional change in the storage solution involved the accumulation of several metabolites over time, such as lactate, malate, succinate, hypoxanthine, inosine, and xanthine (Fig. 1). The consumption and secretion rate of most of these metabolites was not constant over time, but exhibited a transition point. For instance, in the first day of storage approximately 1 mmol/L glucose was consumed and approximately 2 mmol/L lactate was secreted. From Day 1 to Day 6, the mean glucose consumption dropped to reach approximately 0.6 mmol/L/day, whereas lactate release was measured as approximately 1.2 mmol/L/day (Fig. 1). Other extracellular metabolites, however, including succinate, hypoxanthine, malate, citrate, xanthine, and inosine, showed a transition point after 3 days of storage. Similar to exometabolomic profiles, intracellular metabolites showed changes in concentration that were not linear with time. In addition, some of the intracellular profiles were strongly associated with the extracellular profiles, such as succinate, malate, and citrate, showing a transition point after 3 days of storage. Other metabolites,
such as sphingosine 1-phosphate were almost completely depleted after 3 days of storage (Figs. 1 and 2).
Statistical analysis Pathway analysis Pathway analysis32 was used to investigate which metabolic pathways were most affected during storage of PLTs in commercial PAS. This analysis indicated that purine metabolism, glycerophospholipids and glutathione metabolism, pentose phosphate pathway (PPP), and amino sugars and nucleotide sugar metabolism, as well as glycolysis, were the most affected pathways in the endometabolome during storage (Fig. 3A). Altered purine metabolism was also revealed by pathway analysis of the exometabolome together with identified changes in glyoxylate and dicarboxylate metabolism, tricarboxylic acid (TCA) cycle and glycine, and serine and threonine metabolism (Fig. 3b). All the results obtained by pathway analysis are reported in Tables S7 and S8 (available as supporting information in the online version of this paper). Profiles of extra- and intracellular metabolites related to these pathways are shown in Figs. S1 to S5 (available as supporting information in the online version of this paper).
Multivariate data analysis A total of 104 variables showed a significant change over storage time (fold change, p < 0.05). Systematic changes in the data set were assessed using PCA.40,41 PCA clearly defined a trend in PLT metabolism during storage (Fig. 4). The first principal component (PC1) accounted for 41% of the total variance in the data set. It mainly demonstrated how metabolic concentrations changed over time. In fact, variables such as external lactate, external glucose, and internal glutamine have the highest contribution in the PC1 due to significant changes in concentration levels over the time (Table S9, available as supporting information in the online version of this paper). The second principal component (15% of total variance) represented the variability in the data set during the first 3 days of storage and highlights a metabolic transition. In the PC2 most of the variables with high contribution are intracellular measurements such as lysine, arginine, UMP, malate, xanthine, and mitochondrial activity. However, external measurements such as %CD41+CD63+, hypoxanthine, and K+ (Table S9) also influence this metabolic transition in PC2. Thus, a “short-term metabolic” phenotype was observed from Day 0 to Day 3 (short-term-stored PLTs) and a “medium-term metabolic” phenotype from Day 4 to Day 6 (medium-term-stored PLTs). A third phenotypic group, from Day 7 to Day 10, was also observed (longterm-stored PLTs). This metabolic phenotype may represent nonreversible changes (Fig. 4). Volume 54, November 2014 TRANSFUSION
2915
Fig. 1. Time profiles of selected extracellular concentrations. The update rates were computed from the concentration profiles as ΔC/Δt (see Materials and Methods): extracellular glucose (p < 4.3 × 10−16), extracellular lactic acid (p < 1.4 × 10−7), extracellular succinic acid (p < 0.007), extracellular malic acid (p < 7.9 × 10−10), extracellular hypoxanthine (p < 0.006), extracellular citric acid (p < 0.088), extracellular citric acid (Day 0-Day 3; p < 0.011), extracellular glutamine (p < 0.0001), extracellular xanthine (p < 4.5 × 10−6), and extracellular inosine (p < 0.039). Bars represent SDs. ANOVA test was used to calculate p values (see Materials and Methods).
PAGLIA ET AL.
2916
TRANSFUSION Volume 54, November 2014
METABOLIC PROPERTIES OF STORED PLTs
Fig. 2. Time profiles of selected intracellular concentrations. The polarization of the mitochondrial membrane was measured as described under Materials and Methods. Fold change was normalized to Day 0, and p value was calculated using ANOVA as described under Materials and Methods. Intracellular glucose 6-phosphate (Glucose 6-P; p < 1.2 × 10−9), intracellular sphingosine 1-phosphate (Sphingosine 1-P; p < 6.7 × 10−7), intracellular lactic acid (p < 0.00001), intracellular ATP (p < 0.003), mitochondrial activity JC-1 (p < 4.2 × 10−6), intracellular citric acid (p < 9.7 × 10−7), intracellular malic acid (p < 0.00005), intracellular succinic acid (p < 0.001), intracellular hypoxanthine (p < 1.52 × 10−8), intracellular AMP (p < 1.0 × 10−15), intracellular ADP (p < 0.0005), and intracellular IMP (p < 0.008). Bars represent SDs.
The contribution of each bag to this succession of expressed metabolic phenotypes was very similar for five of the six bags analyzed. Bag 3 presented a trend that did not completely overlay the others, suggesting that interindividual differences might play a role in defining PLTs’ metabolic phenotype during storage (Fig. 4C).
HCA Based on pathway analysis results, metabolites involved in purine metabolism, TCA, lipid metabolism, glycolysis and PPP, glutathione metabolism, and routine QC measurements were analyzed by HCA. The results obtained were consistent with PCA, highlighting a metabolic shift
after 4 days of storage (Fig. 5A). This analysis also highlighted four main clusters associated with the three metabolic phenotypes (Fig. 5A). The first cluster highlights metabolites that demonstrated a fast decrease in concentration after the first metabolic shift. The second cluster represents those metabolites that had a rapid decrease in concentration after the second metabolic shift. In the third cluster we found metabolites that increased their concentration in short-term-stored PLTs. The fourth cluster represents the activation markers that correlated well with some external metabolites (Fig. 5B). In particular, we identified a significant correlation of external malic acid and external hypoxanthine with sP-selectin (Fig. 5B). Volume 54, November 2014 TRANSFUSION
2917
PAGLIA ET AL.
Fig. 3. Pathway analysis. The pathway analysis combines results from pathway enrichment analysis with the pathway topology analysis (see Materials and Methods). Pathway enrichment analysis refers to quantitative enrichment analysis using the compound concentration values. It has the potential to identify consistent changes among compounds involved in the same biologic pathway. The pathway toplogy analysis uses node centrality measures to estimate node importance (number of shortest paths going through the node; see Materials and Methods). Results are shown by plotting at the ordinated p value (−log p) calculated from the enrichment analysis and at the axis of the pathway impact value calculated from pathway topology analysis. (A) Pathway analysis of the endometabolome; (B) pathway analysis of the exometabolome.
DISCUSSION It is known that PLTs undergo a process of decay during storage known as PSL. Lactate accumulates in the storage solution and causes metabolic acidosis and a decrease in pH, contributing to inferior PLT quality.11-13 The data reported here support the theory that PLTs do not undergo a monotonic decay, but rather experience different metabolic phenotypes during storage and that these phenotypic changes may potentially be blood donor dependent. PCA showed the existence of metabolic shifts defining three metabolic phenotypes during storage (Fig. 4). In fact, both extracellular and intracellular profiles were not linear with time and suggested the existence of a strong transition point in PLT metabolism after 4 days of storage. Most of these metabolites are involved in purine metabolism, glycolysis, PPP, glutathione metabolism, glycerophospholipid metabolism, and the TCA cycle, as shown by pathway analysis (Figs. 1-3). These results suggest that the PSL is associated with a series of metabolic changes leading to a different regulation of the energy metabolism. HCA was consistent with PCA, confirming the metabolic shift occurring on Day 3 (Fig. 5A). It also revealed 2918
TRANSFUSION Volume 54, November 2014
four main clusters, which represent the main metabolites and pathways involved in the different metabolic phenotypes (Fig. 5A). Glycolysis, PPP, and glutathione metabolism are strongly associated with short-term-stored PLTs, while purine metabolism and TCA cycle are associated with medium-term-stored PLTs (Fig. 5A). The fourth cluster also highlights a strong correlation between activation marker and secretion of some metabolites such as malate and hypoxanthine (Fig. 5B). Short-term-stored PLTs (Days 0-3) experienced complete conversion of glucose to lactate. Every mole of glucose consumed resulted in 2 moles of lactate secreted into the medium. There is ample evidence that nonoxidative metabolism of glucose is not due to low levels of O2 (pO2 < 2 mmHg) since PLTs are stored in modern gas-permeable storage containers.42-44 Thus, pyruvate was presumably not oxidized in the TCA cycle for ATP production through oxidative phosphorylation. Other substrates can feed into the TCA cycle coupling oxidative phosphorylation with ATP production,45-48 but high secretion of succinate in the first 4 days seems to exclude a normal activity of the TCA cycle (Fig. 1). Indeed, succinate secretion is a known marker for loss of mitochondrial function,49 which is also suggested by the mitochondrial
METABOLIC PROPERTIES OF STORED PLTs
Fig. 4. PCA assumes that the n-dimensional data can be reduced to a linear combination of principal components that best explains the variance in the data. The principal components are rank ordered by the variability that they represent in the data set, with the first principal component accounting for the greatest variability in the data and so on.40,41 (A) PCA performed on all samples; (B) every time point (in days) represents the mean from six bags; (C) every time point (in days) is connected by line to show the specific trend of each bag.
membrane potential profile showing a decrease of approximately 20% in the number of polarized cells in short-term-stored PLT (Fig. 2). In PLTs, intracellular ATP is located in the cytoplasm (metabolic ATP) and in the dense granules (stored ATP).50 The ATP metabolic pool is mainly associated just with anaerobic oxidation of glucose. In short-term-stored PLTs (Days 0-3), the ATP metabolic pool is mainly associated just with anaerobic oxidation of glucose. The down regulation of aerobic oxidation and consequent metabolic stress should then result in increased levels of ADP and AMP degradation products. In fact, inosine, hypoxanthine, and xanthine accumulated in the storage solution (Figs. 1, 2, and 6). The release of malate into solution in the
first 4 days of storage is consistent with its intracellular profile (Figs. 1 and 2). Malate might be generated by citrate, which was present in the storage solution. PLTs consumed citrate, even if the fold change during 10-day storage was not significant (ANOVA p = 0.088). During the first 4 days, there was a significant consumption of citrate (ANOVA p = 0.011). In fact, intracellular citrate accumulated during the first 4 days of storage, and it exhibited a profile similar to malate. These observations suggest that short-term-stored PLTs were able to produce malate from citrate, but not to consume it. Consequently, malate appears to be released into the storage solution (Fig. 6). Short-term-stored PLTs consumed 40% of available glucose and secreted several metabolites, generating, in the first 4 days of storage, a microenvironment that was significantly different from its initial composition. Moreover, accumulation of lactate and malate caused a decrease in pH in the storage solution and potentially also inside the cells. This internal change of pH might affect the activity of some important enzymes such as phosphofructokinase.51 This change in composition and in pH is most likely the root cause of the metabolic shift detected at 4 days of storage (Figs. 4-6), which, according to our results, corresponds to a fundamentally different state of the PLT’s metabolic network. Medium-term PLTs (4-6 days of storage) still converted glucose almost completely to lactate (glucose consumption approx. 0.6 mmol/L/day, lactate secretion approx. 1.2 mmol/L/day), but TCA cycle and oxidative phosphorylation became more active as suggested by changes in extracellular and intracellular levels of succinate. Indeed, the secretion rate of succinate dropped after Day 3, as did its intracellular level, suggesting its consumption by the TCA cycle. Increased mitochondrial membrane potential was consistent with mitochondrial metabolism being more active (Figs. 1, 2, and 6). Since the vast majority of pyruvate coming from glycolysis seems to be converted into lactate, the TCA cycle had to use other substrates. In fact, medium-term-stored PLTs stopped secreting malate, indicating its use in the TCA cycle and the production of ATP by oxidative phosphorylation (Fig. 6). ATP can also be generated by reutilization of hypoxanthine. Short-term-stored PLTs showed decreasing intracellular levels of AMP, IMP, ADP, and ATP and increasing levels of extracellular inosine, hypoxanthine, and xanthine (Figs. 1, 2, and 6). Medium-term-stored PLTs demonstrated an opposite trend, suggesting that the purine metabolism recycled hypoxanthine to produce additional ATP (Figs. 1, 2, and 6). Hypoxanthine is converted to xanthine by xanthine oxidoreductase (irreversible reaction). Once converted into xanthine, it cannot be recycled. In fact, inosine, hypoxanthine, and xanthine were released by short-term-stored PLTs (0-3 days), but Volume 54, November 2014 TRANSFUSION
2919
PAGLIA ET AL.
Fig. 5. HCA and correlation. (A) HCA provides a measure of closeness of metabolites and groups (time points). (B) Section of fourth cluster of the HCA and Pearson correlation of selected metabolites. Malic acid external versus sP-selectin, r = 0.85, p = 3.3 × 10–14; hypoxanhine external versus sP-selectin, r = 0.90, p = 1.6 × 10–18.
the secretion rate slowed down from Day 4 to Day 6 (Figs. 1, 2, and 6), suggesting that hypoxanthine was not converted into xanthine anymore, but was instead being recycled through the purine salvage pathway. In summary, the proposed metabolic phenotype for short-term-stored PLTs involved the down regulation of the TCA cycle and the accumulation of ATP degradation products. The metabolic signature of this phenotype is the accumulation of succinate, hypoxanthine, malate, xanthine, and inosine in the storage solution (Figs. 1 and 6). The development of the short-term-stored PLTs’ metabolic phenotype might be due to the Crabtree effect. In fact, in excess of glucose, cross-inhibition of oxidative phosphorylation might occur to regulate the equilibrium between the cytosolic and the mitochondrial oxidative 2920
TRANSFUSION Volume 54, November 2014
pathways.46,52 The activation of PLTs during the first day of storage as a consequence of collection and storage medium could also play an important role in this process (Table 1). In fact, we observed a significant correlation (p < 0.0001) of activation markers with secretion of hypoxanthine and malate in the storage solutions (Figs. 5A and 5B). Medium-term-stored PLTs instead had moderate activity in TCA cycle and in metabolism, which in part compensated for the depletion of intracellular ATP. This compensation was able to sustain PLT metabolism until Day 6, when PLTs started to experience a faster decay leading to cell lysis and the accumulation of several metabolites in the storage medium, defining the last metabolic phenotypes (long-term-stored PLTs). This
METABOLIC PROPERTIES OF STORED PLTs
different PLTs’ metabolic activity. The interindividual differences that might play a role in defining PLT metabolic phenotypes during storage will also need clinical investigation and will require a higher number of samples. ACKNOWLEDGMENTS The authors thank Marketa Foley, Kristine Wichuk, Manuela Magnusdottir, Steinunn Thorlacius, and the staff at the blood bank for technical support. The authors especially thank Sirus Palsson, Sisse Ostrowski, Marc Abrams, and Hans Gulliksson for their input. GP designed and performed the experiments, analyzed the data, and wrote the paper; OES designed and performed the experiments and wrote the
Fig. 6. Proposed metabolic phenotypes for short-term- and medium-term-stored PLTs. Blue lines represent active metabolic pathways in both phenotypes; black lines
paper; OR wrote the paper and performed ATP analysis; SV performed the MS experiments; MBH designed experiments
represent down regulated metabolic pathways; red lines represent metabolic pathway characteristic of short-term-stored PLTs; and green lines represent meta-
and wrote the paper; SB designed the experiments; SG designed the experiments and wrote the paper; and BOP designed the
bolic pathway characteristic of medium-term-stored PLTs.
experiments and wrote the paper.
metabolic phenotype may represent nonreversible changes and is of less clinical relevance than the first two since PLTs are never transfused beyond 7 days of storage within the present regulatory framework (Fig. 4). Even though data interpretation and the proposed metabolic phenotypes were mainly focused on purine metabolism and the TCA cycle, several pathways are likely involved in the development of the storage lesions and their metabolic phenotype. For instance, glycerophospholipid metabolism might play an important role. (Figs. 3 and S5). Moreover, it is interesting to note that the intracellular pool of sphingosine 1-phosphate (Fig. 2) was completely depleted after 4 days of storage. Recent findings suggest that this metabolite is involved in the control of thrombopoiesis.53 Together, these findings may expand the understanding of the underlying mechanisms behind the storage lesions and may help identify biomarkers that will better enhance the quality of PLTs during storage. The presented metabolic changes are representative for the storage of PLTs collected with particular apheresis equipment and stored in a specific PAS and thus may not represent the changes of buffy coat PLTs collected from whole blood, for instance. Studies of metabolic changes of PLTs processed from buffy coats are in progress. Clinical studies evaluating the quality of short- and medium-term-stored PLTs used in transfusion will be necessary to understand the clinical relevance of
CONFLICT OF INTEREST The authors have disclosed no conflicts of interest.
REFERENCES 1. Michelson AD. Platelets. San Diego (CA): Academic Press; 2002. 2. Hogman CF. Storage of blood components. Curr Opin Hematol 1999;6:427-31. 3. Vasconcelos E, Figueiredo AC, Seghatchian J. Quality of platelet concentrates derived by platelet rich plasma, buffy coat and Apheresis. Transfus Apher Sci 2003;29: 13-6. 4. Dijkstra-Tiekstra MJ, Pietersz RN, Hendriks EC, et al. In vivo PLT increments after transfusions of WBC-reduced PLT concentrates stored for up to 7 days. Transfusion 2004;44:330-6. 5. Hess JR. Red cell changes during storage. Transfus Apher Sci 2010;43:51-9. 6. Bashir S, Cookson P, Wiltshire M, et al. Pathogen inactivation of platelets using ultraviolet C light: effect on in vitro function and recovery and survival of platelets. Transfusion 2013;53:990-1000. 7. Goodnough LT, Shander A, Brecher ME. Transfusion medicine: looking to the future. Lancet 2003;361:161-9. 8. Shrivastava M. The platelet storage lesion. Transfus Apher Sci 2009;41:105-13. Volume 54, November 2014 TRANSFUSION
2921
PAGLIA ET AL.
9. Kocazeybek B, Arabaci U, Akdur H, et al. Prospective evaluation of platelets prepared by single and random methods during 5 days of storage: aspects related to quality and quantity. Transfus Apher Sci 2002;26: 29-34. 10. Akay OM, Gunduz E, Basyigit H, et al. Platelet function testing during 5-day storage of single and random donor plateletpheresis. Transfus Apher Sci 2007;36:285-9. 11. Holme S. Storage and quality assessment of platelets. Vox Sang 1998;74(Suppl 2):207-16. 12. Gyongyossy-Issa MI. Glucose in platelet additive solutions: to add or not to add? Transfus Apher Sci 2011; 44:283-95. 13. Gulliksson H. Defining the optimal storage conditions for the long-term storage of platelets. Transfus Med Rev 2003; 17:209-15. 14. Heuft HG, Goudeva L, Krauter J, et al. Effects of platelet concentrate storage time reduction in patients after blood stem cell transplantation. Vox Sang 2013;105: 18-27. 15. Thiele T, Iuga C, Janetzky S, et al. Early storage lesions in apheresis platelets are induced by the activation of the integrin alphaIIbbeta(3) and focal adhesion signaling
inflammation and clinical outcomes. Thromb Res 2011; 127:287-91. 25. Holmes E, Loo RL, Stamler J, et al. Human metabolic phenotype diversity and its association with diet and blood pressure. Nature 2008;453:396-400. 26. Illig T, Gieger C, Zhai G, et al. A genome-wide perspective of genetic variation in human metabolism. Nat Genet 2010;42:137-41. 27. Fiehn O. Combining genomics, metabolome analysis, and biochemical modelling to understand metabolic networks. Comp Funct Genomics 2001;2:155-68. 28. Paglia G, Magnusdottir M, Thorlacius S, et al. Intracellular metabolite profiling of platelets: evaluation of extraction processes and chromatographic strategies. J Chromatogr B Analyt Technol Biomed Life Sci 2012;898:111-20. 29. Paglia G, Hrafnsdottir S, Magnusdottir M, et al. Monitoring metabolites consumption and secretion in cultured cells using ultra-performance liquid chromatography quadrupole-time of flight mass spectrometry (UPLC-QToF-MS). Anal Bioanal Chem 2012;402:1183-98. 30. Wishart DS, Jewison T, Guo AC, et al. HMDB 3.0–The Human Metabolome Database in 2013. Nucleic Acids Res 2013;41 (Database issue):D801-7.
pathways. J Proteomics 2012;76 Spec No.:297-315. 16. Paglia G, Palsson BO, Sigurjonsson OE. Systems biology of
31. Smith CA, O’Maille G, Want EJ, et al. METLIN: a metabolite mass spectral database. Ther Drug Monit 2005;27:747-
stored blood cells: can it help to extend the expiration date? J Proteomics 2012;76 Spec No.:163-7. 17. Gevi F, D’Alessandro A, Rinalducci S, et al. Alterations of
51. 32. Xia J, Mandal R, Sinelnikov IV, et al. MetaboAnalyst 2.0— a comprehensive server for metabolomic data analysis.
red blood cell metabolome during cold liquid storage of erythrocyte concentrates in CPD-SAGM. J Proteomics
Nucleic Acids Res 2012;40 (Web Server issue):W127-33. 33. Reich M, Liefeld T, Gould J, et al. GenePattern 2.0. Nat
2012;76 Spec No.:168-80. 18. Pienimaeki-Roemer A, Ruebsaamen K, Boettcher A, et al.
Genet 2006;38:500-1. 34. Ostrowski SR, Bochsen L, Salado-Jimena JA, et al. In vitro
Stored platelets alter glycerophospholipid and sphingolipid species, which are differentially transferred to newly released extracellular vesicles. Transfusion 2013;53:612-
cell quality of buffy coat platelets in additive solution treated with pathogen reduction technology. Transfusion 2010;50:2210-9.
26. 19. Antonelou MH, Tzounakas VL, Velentzas AD, et al. Effects
35. Yockey C, Murphy S, Eggers L, et al. Evaluation of the Amicus Separator in the collection of apheresis platelets.
20.
21.
22.
23.
24.
2922
of pre-storage leukoreduction on stored red blood cells signaling: a time-course evaluation from shape to proteome. J Proteomics 2012;76 Spec No.:220-38. Parguina AF, Garcia A. Platelet proteomics in transfusion medicine: a reality with a challenging but promising future. Blood Transfus 2012;10(Suppl 2):s113-4. Zolla L, D’Alessandro A. Shaking hands with the future through omics application in transfusion medicine and clinical biochemistry. Blood Transfus 2012;10(Suppl 2):s1-3. Schubert P, Devine DV. De novo protein synthesis in mature platelets: a consideration for transfusion medicine. Vox Sang 2010;99:112-22. D’Alessandro A, Gevi F, Zolla L. Red blood cell metabolism under prolonged anaerobic storage. Mol Biosyst 2013;9:1196-209. Refaai MA, Phipps RP, Spinelli SL, et al. Platelet transfusions: impact on hemostasis, thrombosis,
TRANSFUSION Volume 54, November 2014
Transfusion 1998;38:848-54. 36. Singh H, Chaudhary R, Ray V. Platelet indices as quality markers of platelet concentrates during storage. Clin Lab Haematol 2003;25:307-10. 37. Macher S, Sipurzynski-Budrass S, Rosskopf K, et al. Function and activation state of platelets in vitro depend on apheresis modality. Vox Sang 2010;99:332-40. 38. Hagberg IA, Akkok CA, Lyberg T, et al. Apheresis-induced platelet activation:comparison of three types of cell separators. Transfusion 2000;40:182-92. 39. Schwertz H, Koster S, Kahr WH, et al. Anucleate platelets generate progeny. Blood 2010;115:3801-9. 40. Trygg J, Holmes E, Lundstedt T. Chemometrics in metabonomics. J Proteome Res 2007;6:469-79. 41. Edward JJ. A user’s guide to principal components. New York: John Wiley & Sons; 1991. 42. Niu X, Arthur P, Abas L, et al. Carbohydrate metabolism in human platelets in a low glucose medium under
METABOLIC PROPERTIES OF STORED PLTs
aerobic conditions. Biochim Biophys Acta 1996;1291:97106. 43. Guppy M, Abas L, Arthur PG, et al. The Pasteur effect in human platelets: implications for storage and metabolic control. Br J Haematol 1995;91:752-7. 44. Kilkson H, Holme S, Murphy S. Platelet metabolism during storage of platelet concentrates at 22 degrees C. Blood 1984;64:406-14. 45. Cartledge S, Candy DJ, Hawker RJ. Citrate metabolism by human platelets. Transfus Med 1997;7:211-5. 46. Baker JM, Candy DJ, Hawker RJ. Influences of pH on human platelet metabolism. Platelets 2001;12:333-42. 47. Whisson ME, Nakhoul A, Howman P, et al. Quantitative study of starving platelets in a minimal medium: maintenance by acetate or plasma but not by glucose. Transfus Med 1993;3:103-13. 48. Saunders C, Rowe G, Wilkins K, et al. Impact of glucose and acetate on the characteristics of the platelet storage lesion in platelets suspended in additive solutions with minimal plasma. Vox Sang 2013;105:1-10. 49. King A, Selak MA, Gottlieb E. Succinate dehydrogenase and fumarate hydratase: linking mitochondrial dysfunction and cancer. Oncogene 2006;25:4675-82. 50. Diab YA, Thomas A, Luban NL, et al. Acquired cytochrome C oxidase impairment in apheresis platelets during storage: a possible mechanism for depletion of metabolic adenosine triphosphate. Transfusion 2012;52: 1024-30. 51. Nishino T, Yachie-Kinoshita A, Hirayama A, et al. Dynamic simulation and metabolome analysis of longterm erythrocyte storage in adenine-guanosine solution. PloS One 2013;8:e71060. 52. Crabtree HG. Observations on the carbohydrate metabolism of tumours. Biochem J 1929;23:536-45. 53. Zhang L, Urtz N, Gaertner F, et al. Sphingosine kinase 2 (Sphk2) regulates platelet biogenesis by providing intracellular sphingosine 1-phosphate (S1P). Blood 2013;122:
Appendix S1. Blood bank quality controls. Fig. S1. Profiles of selected intra- and extracellular concentrations of metabolites involved in purine metabolism. Fold change was normalized to Day 0. Bars represent standard deviations. Fig. S2. Profiles of selected intra- and extracellular concentrations of metabolites involved in TCA cycle. Fold change was normalized to Day 0. Bars represent standard deviations. Fig. S3. Profiles of selected intra- and extracellular concentrations of metabolites involved in glutathione metabolism. Fold change was normalized to Day 0. Bars represent standard deviations. Fig. S4. Profiles of selected intra- and extracellular concentrations of metabolites involved in glycolysis and pentose phosphate pathway. Fold change was normalized to Day 0. Bars represent standard deviations. Fig. S5. Profiles of selected intra- and extracellular concentrations of metabolites involved in glycerophospholipid metabolism. Fold change was normalized to Day 0. Bars represent standard deviations. Table S1. Summary of sample preparation for flow cytometry analysis Table S2. In vitro measurements of apheresis platelet quality in PC samples stored for a period of 10 days. Unless otherwise noted, results are presented as mean ± standard deviation (n = 6) Table S3. Markers of platelet activation and metabolism in apheresis platelets stored for a period of 10 days. Results are presented as mean ± standard deviation (n = 6) Table S4. Targeted metabolites identified in the platelets metabolome Table S5. Identified unexpected metabolites Table S6. Metabolomics quality controls Table S7. Pathway analysis of the endo-metabolome Table S8. Pathway analysis of the exo-metabolome Table S9. Variable contribution to PCA
791-802.
SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher’s Web site:
Volume 54, November 2014 TRANSFUSION
2923
BACKGROUND: Platelet (PLT) concentrates are routinely stored for 5 to 7 days. During storage they exhibit what has been termed PLT storage lesion (PSL), which is evident by a loss of hemostatic function when transfused into patients. The overall goal of this study was to obtain a comprehensive data set describing PLT metabolism during storage. STUDY DESIGN AND METHODS: The experimental approach adopted to achieve this goal combined a series of standard assays to monitor the quality of stored PLTs and a deep-coverage metabolomics study using liquid chromatography coupled with mass spectrometry performed on both the extracellular and the intracellular environments. During storage we measured 174 different variables in 6 PLT units, collected by apheresis. Samples were collected at eight different time points resulting in a data set containing more than 8000 measurements. RESULTS: Stored PLTs did not undergo a monotonic decay, but experienced systematic changes in metabolism reflected in three discrete metabolic phenotypes: The first (Days 0-3) was associated with active glycolysis, pentose phosphate pathway, and glutathione metabolism and down regulation of tricarboxylic acid (TCA) cycle. The second (Days 4-6) was associated with a more active TCA cycle as well as increased purine metabolism. A third metabolic phenotype of less clinical relevance (Days 7-10) was associated with a faster decay of cellular metabolism. CONCLUSION: PSL is not associated with a linear decay of metabolism, but rather with successive metabolic shifts. These findings may give new insight into the mechanisms underlying PSL and encourage the deployment of systems biology methods to PSL.
P
latelet concentrates (PCs) may be used in many clinical situations, including treatment of hemorrhages, inborn or acquired defects of platelet (PLT) function, or as prophylaxis in patients with thrombocytopenia after chemotherapy.1 In recent decades PC transfusion has significantly improved with methods such as leukoreduction for removing contaminating white blood cells (WBCs), improved PLT additive solutions (PASs), and the practice of using gas-permeable storage containers.2-4 Nevertheless, during storage, PLTs develop a condition known as PLT storage lesion (PSL). In addition to the PSL, risk of bacterial contamination during storage has limited the use of PLTs beyond 5 days of storage. With the additional requirement of bacterial surveillance or pathogen inactivation, many countries allow PLT transfusion with up to 7 days of storage.5,6 Considering the important role of PLT transfusion in modern medicine, an improvement in quality of the
ABBREVIATIONS: HCA = hierarchical cluster analysis; LC = liquid chromatography; MS = mass spectrometry; PC(s) = platelet concentrate(s); PCA = principal component analysis; PPP = pentose phosphate pathway; PSL = platelet storage lesion; sP-selectin = soluble P-selectin; TCA = tricarboxylic acid. From the 1Center for Systems Biology, University of Iceland; 2 The Blood Bank, Landspitali-University Hospital; and the 3 School of Science and Engineering, Reykjavik University, Reykjavik, Iceland; and the 4Department of Clinical Immunology, Rigshospitalet, Copenhagen University Hospital, Copenhagen, Denmark. Address reprint requests to: Ólafur E. Sigurjónsson, The Blood Bank, Landspitali University Hospital, Snorrabraut 60, 105 Reykjavik, Iceland; e-mail: [email protected]. This work was supported by the European Research Council Grant Proposal No. 232816. Received for publication November 14, 2013; revision received February 22, 2014, and accepted February 25, 2014. doi: 10.1111/trf.12710 © 2014 AABB TRANSFUSION 2014;54:2911-2923. Volume 54, November 2014 TRANSFUSION
2911
PAGLIA ET AL.
PLTs at the end of storage or an extension of the present expire criteria could have an important impact on modern blood banking.7 PSL is thought to be the result of a combination of several factors, including the methods of collection, processing, storage, and manipulation after collection. This lesion causes a combination of changes in PLT morphology and metabolism, physiologic activation, and release of PLT granules,8 resulting in deteriorated PLT function during storage.9,10 A model of linear decay over storage time has traditionally been suggested due to evidence of a constant consumption of glucose and a constant release of lactate. Accumulation of lactate causes metabolic acidosis and the low pH, in turn, inhibits PLT function and leads to inferior PLT quality upon transfusion.11-13 A recent retrospective study confirmed the association of the quality of PCs with time of storage by comparing the effect of transfused PLTs stored for 4 and 5 days, suggesting that a restriction of storage time from 5 to 4 days may result in an improvement of the clinical outcomes.14 PLT quality is routinely assessed by swirling, PLT number, lactate accumulation, pH, PLT surface markers, and functional assays such as in vitro aggregometry assays. The use of integrated omics and systems biology approaches may become a recognized strategy in transfusion medicine in the future.15-23 The development of new technologies facilitates the collection of more comprehensive data sets that may enable a better understanding of the metabolism of blood cells during storage. The understanding of PLT metabolism during storage has two major goals: 1) illuminating the profound mechanism behind the storage lesions and 2) discovering pathways and biomarkers to better define the quality of PLTs during storage. A better understanding of PLT metabolism is one of the prerequisites for increasing the quality and function of PLTs as well as decreasing adverse events in patients after transfusion.24 The totality of all metabolites at any given moment, known as the metabolome, is a chemical readout of the health of the cell, as changes in the concentrations of specific groups of metabolites are sensitive and specific to pathologically and physiologically relevant factors, such as genetic variations and diet.25,26 The metabolome comprises the endometabolome (all intracellular metabolites) and the exometabolome (all metabolites released into the extracellular fluid).27 Here we present the first comprehensive metabolomics study on stored PLTs. The strategy adopted combines liquid chromatography (LC)-mass spectrometry (MS)28,29 with several assays to define distinct stages of stored PLTs. The exo- and endometabolome were obtained during 10 days of storage, resulting in the most comprehensive metabolomics data set for stored PLTs yet published. 2912
TRANSFUSION Volume 54, November 2014
MATERIALS AND METHODS Sample collection PCs were harvested from healthy blood donors by apheresis using an apheresis separator instrument (Amicus 4R580, Fenwal, Zurich, Switzerland) at the Blood Bank, Landspitali-University Hospital, Iceland. In total, 6 units of apheresis PLTs were obtained from six donors of mixed age, sex, and blood types using an apheresis kit (Amicus, Fenwal). PLTs were stored in T-SOL (Fenwal) in a plastic container (PL 2410, Fenwal). Sampling was done on Day 0 (day of collection by apheresis) and subsequently on Days 1, 3, 4, 5, 6, 7, and 10. On Day 0 sampling was carried out 2 hours after collection. Units were stored in a shaking incubator set at 22°C and were discarded after completion of sampling on Day 10. All units were checked for bacterial contamination on Day 10. The National Bioethics Committee of Iceland and the Icelandic Data Protection Authority approved the study.
Blood bank quality controls Immediately after sample collection, a blood gas analyzer (ABL90 FLEX, Radiometer Medical ApS, Copenhagen, Denmark) was used for determination of pH, pO2, and pCO2; concentration of total hemoglobin; and concentrations of K+, Na+, Cl–, glucose, and lactate in the PLT media. PLT concentration, mean PLT volume, plateletcrit, PLT distribution width, and WBC count was determined using an automated hematology analyzer (CELL-DYN Ruby, Abbott Diagnostics, Abbott Park, IL). Flow cytometry (FACSCalibur, Becton Dickinson, La Jolla, CA) was used to determine expression of CD41, CD42b, CD62P, CD63, annexin V (all from Becton Dickinson), and JC1 to indicate the level of mitochondrial polarization (Life Technologies, Boston, MA). Lactate dehydrogenase (LDH) activity in the PC medium was assessed by an LDH assay kit (ab102526, Abcam, Cambridge, UK). The soluble CD40 ligand concentration was determined using an enzyme-linked immunosorbent assay (ELISA) kit (DCDL40, Quantikine, R&D Systems, Minneapolis, MN). Soluble P-selectin (sP-selectin) concentration was determined via ELISA (BBE6, R&D Systems, Abingdon, UK). Aliquots of 0.25 mL were used to determine the adenosine triphosphate (ATP) concentration employing a luminescent cell viability assay (CellTiter-Glo, Promega, Madison, WI) using a microplate reader (Spectramax M3, Molecular Devices, Sunnyvale, CA). For more detailed methods refer to Appendix S1 (available as supporting information in the online version of this paper).
Metabolomics Sample preparation Supernatant and cells were separated by centrifugation (1600 × g, 22°C, 5 min) of 0.5 mL of PC. Cell pellets were
METABOLIC PROPERTIES OF STORED PLTs
washed twice by adding 1 mL of phosphate-buffered saline, and after centrifugation (1600 × g, 22°C, 5 min), the supernatant was discharged. Thirty microliters of labeled internal standard mixture (Appendix S1) and 1 mL of cold (−20°C) methanol: water (7:3) were added to the cell pellets. Cell lysis was achieved by performing two consecutive freeze-and-thaw steps. Samples were centrifuged (15,000 × g, 4°C, 15 min) and supernatant was transferred to a new tube. One milliliter of cold (−20°C) methanol:water (7:3) was added to the pellets. Samples were mixed vigorously for 1 minute and after centrifugation (15,000 × g, 4°C, 15 min) the supernatant was added to the precedent. Samples were dried using a vacuum concentrator, reconstituted in 300 μL H2O:ACN (50:50), and filtered by centrifugation using a Pierce protein 96-well precipitation plate (2000 × g, 4°C, 15 min). Two-hundred microliters of medium sample was processed by adding 30 μL of labeled internal standard mixture (Appendix S1) and 0.5 mL of MeOH. Samples were mixed vigorously and centrifuged (15,000 × g, 4°C, 15 min). Supernatant was transferred into a new tube and dried using a vacuum concentrator. Samples were reconstituted in 600 μL of H2O:ACN (50:50) and filtered by centrifugation using a Pierce protein 96-well precipitation plate (2000 × g, 4°C, 15 min). Pooled QC samples were prepared for both extracellular and intracellular analysis by pooling 20 μL for each processed sample. Spiked pooled QC samples were prepared by spiking pooled QC samples with a mix containing 17 standard compounds (Appendix S1).
Ultraperformance LC-MS Metabolomic analysis was performed using a modified version of a previously reported method.28,29 Ultraperformance liquid chromatography (Acquity, Waters, Manchester, UK) was coupled with a quadrupole time-offlight mass spectrometer (Synapt G2, Waters). Chromatographic separation was achieved by hydrophilic interaction liquid chromatography (HILIC) using an Acquity amide column, 1.7 μm (2.1 × 150 mm; Waters). All intracellular and extracellular samples were analyzed three times: once in positive ionization mode and twice in negative ionization mode using acidic and basic chromatographic conditions. In positive and in negative acidic conditions, Mobile Phase A was 100% ACN and B was 100% H2O, both containing 0.1% formic acid. The following elution gradient was used: 0 minutes 99% A, 7 minutes 30% A, 7.1 minutes 99% A, and 10 minutes 99% A. In negative mode basic conditions, Mobile Phase A contained 10 mmol/L ACN:sodium bicarbonate (95:5) and Mobile Phase B contained 10 mmol/L ACN:sodium bicarbonate (5:95). The following elution gradient was used: 0 minutes 99% A, 6 minutes 30% A, 6.5 minutes 99% A, and 10 minutes 99% A. In all conditions, the flow rate was
0.4 mL/min, the column temperature was 45°C, and the injection volume was 3.5 μL. In all conditions, the mass spectrometer operated using a capillary voltage of 1.5 kV, and the sampling cone and the extraction cone were of 30 and 5 V, respectively. The cone and the desolvation gas flow were 50 and 800 L/ hr, respectively, while the source and desolvation gas temperatures were 120 and 500°C, respectively. MS spectra were acquired in centroid mode from m/z 50 to 1000 using scan time of 0.3 seconds. Leucine enkephalin (2 ng/μL) was used as lock mass (m/z 556.2771 and 554.2615 in positive and negative experiments, respectively). A typical analytical run consisted of pooled QC samples to equilibrate the system, calibrators, samples, and spiked pooled QC samples.29 Stability of the LC-MS system during analysis was validated by analyzing spiked pooled QC samples every 10 samples. Spiked pooled QC samples had a relative standard deviation (RSD%) for the retention times never higher than 2. Intra- and interbatch variation of signal intensities provided a RSD% never higher than 13 (Table S6, available as supporting information in the online version of this paper).
Data processing The identification of unexpected metabolites was achieved by integration, alignment, and conversion of MS data points into exact mass retention time pairs (MarkerLynx, v4.1, Waters). The identity of the unexpected metabolites was established by verifying peak retention time, accurate mass measurements, and tandem MS against our in-house database and/or online databases, including HMDB30 and METLIN.31 TargetLynx (v4.1, Waters) was used to integrate chromatograms of targeted metabolites. Ion chromatograms were extracted using a 0.02-mDa window centered on the expected m/z for each targeted compound. Each metabolite was normalized using a proper isotopically labeled internal standard as described in Table S4 (available as supporting information in the online version of this paper). The ratio metabolites/internal standard was then used for deriving concentration using external calibration with reference standards.
Statistical analysis Normalization was performed to minimize differences between two different experiments (Experiment 1 [Exp1], Units 1-3; Experiment 2 [Exp2], Units 4-6). Normalized Exp2 (Exp2″) was obtained by applying the following formula to each variable:
Exp2′′ = (Exp2 − average Exp2) + average Exp1. Concentrations of extracellular measurements were computed to obtain consumption and secretion rates (ΔC/Δt) during the storage. Volume 54, November 2014 TRANSFUSION
2913
2914
TRANSFUSION Volume 54, November 2014
Day 7 263 ± 3 6.78 ± 0.08 4.1 ± 0.2§ 4.3 ± 0.8 16.9 ± 1.0 0.05 ± 0.06 1170 ± 126 41.7 ± 7.7 23.4 ± 6.9 Day 6 267 ± 3 6.80 ± 0.06 4.1 ± 0.2§ 4.4 ± 1.2 17.0 ± 0.9 0.06 ± 0.05 1180 ± 240 40.8 ± 7.6 24.9 ± 4.7 Day 5 272 ± 3 6.82 ± 0.06 4.0 ± 0.1§ 4.3 ± 0.7 17.1 ± 0.8 0.05 ± 0.05 1145 ± 127 40.3 ± 7.1 24.9 ± 5.6 Day 4 276 ± 3 6.85 ± 0.04 4.4 ± 0.5§ 4.7 ± 1.5 17.1 ± 0.8 0.09 ± 0.1 1220 ± 191 41.3 ± 6.9 25.4 ± 5.4 Day 3 281 ± 3 6.86 ± 0.03 4.1 ± 0.2§ 5.5 ± 1.7 17.3 ± 0.7 0.10 ± 0.14 1368 ± 250 43.6 ± 10.2 24.6 ± 6.2
TABLE 1. Selected blood bank PLT QC variables*
Unless otherwise noted, results are presented as mean ± standard error (n = 6). Percentage of the total number of PLTs that satisfy the conditions specified under “variable.” n = 5. n = 3.
Sample collection was designed and optimized considering the maximum sample volume that was possible to collect from each unit without affecting the storage of PLTs. The total samples collected in all experiments were on average approximately 10% of the initial weight (Table 1). Blood bank QCs were selected to define the quality of stored PLTs, applying analyses and methods published elsewhere.34 The quality of the PLTs used in the experiments was in conformance with other publications.35-37 The percentage of PLTs expressing the activation markers CD62P increased in the first 4 days, probably due to activation triggered by the apheresis process, and then stayed constant until Day 7, with an additional increase in
* † ‡ §
Sampling and blood bank QCs
Day 1 286 ± 3 6.91 ± 0.05 4.2 ± 0.2‡ 5.0 ± 0.7 17.5 ± 0.8 0.18 ± 0.26 1253 ± 123 45.8 ± 10.2 19.5 ± 7.7
The overall goal of this work was to obtain a better and more comprehensive understanding of PLT metabolism during storage conditions. The experimental approach designed to reach this goal involved a four-step procedure: 1) sampling, 2) blood bank quality control (QC), 3) metabolomics, and 4) data analysis. We monitored 174 different variables for each of the six stored PLT containers at each of the eight time points, resulting in a data set containing more than 8000 measurements. To date, this represents the most comprehensive data set describing stored PLTs.
Day 0 291 ± 2 6.96 ± 0.08 4.9 ± 0.9 4.9 ± 0.5 18.2 ± 0.8 0.94 ± 0.78 1026 ± 162 36.9 ± 11.5 6.9 ± 1.6
RESULTS
Variable† PC volume/bag (mL) pH (22°C) Mean PLT volume (fL) Plateletcrit (mL/L) PLT distribution width (10 GSD) WBC concentration (×109 cells/L) PLT concentration (×109 cells/L) CD41+; CD62P+ (%)† CD41+; CD63+ (%)†
Pathway analysis was performed using metaboanalyst.32 This approach combines pathway enrichment and pathway topology analysis. Every variable having an HMDB ID was included in the analysis using the Homo sapiens KEGG library. The pathway enrichment analysis method was global test and the pathway topology analysis used relative betweenness centrality. The fold change, normalized to Day 0, was calculated for each variable monitored and one way analysis of variance (ANOVA) test was used to find measurements presenting a significant change with the time of storage (p < 0.05). Principal component analysis (PCA; EZinfoUmetrics) was performed on 104 measurements (p < 0.05 one-way ANOVA test). Before PCA, data were scaled (unit variance scaling). Selected metabolites (based on pathway analysis and ANOVA test < 0.05) were analyzed via hierarchical cluster analysis (HCA).33 Pearson correlation was used on both column (time points) and row (metabolites). Both column and row were normalized by subtracting the mean and data were log-transformed before clustering. The average of each time points was used for this analysis.
Day 10 258 ± 3 6.81 ± 0.09 4.0 ± 0.1§ 4.5 ± 0.7§ 17.6 ± 0.5§ 0.15 ± 0.32 910 ± 204 48.0 ± 9.1 27.3 ± 6.1
PAGLIA ET AL.
METABOLIC PROPERTIES OF STORED PLTs
the last days of storage. The activation marker CD63 had a similar profile, confirming activation of PLTs in the first 4 days of storage (Table 1). PLT count increased during the first 4 days. This profile might reflect a previously described artifact of PLT aggregation with subsequent resolution of the aggregates resulting in apparent increase in the number of PLTs.37,38 Stored PLTs may also demonstrate an increase in PLT count arising from PLT division in the concentrate as has been reported previously.39 Bacterial contamination was not detected in any of the units on Day 10.
Metabolomics A metabolomics approach, based on ultraperformance LC–hydrophilic interaction LC–quadrupole time-offlight–MS strategy, was applied to both extracellular and intracellular environments. This strategy includes a targeted approach of selected polar metabolites, but it also has the ability to identify unexpected metabolites (Tables S4 and S5).28,29 Targeted metabolites were selected based on preliminary investigations on PCs.28 Metabolomic analysis identified 49 metabolites in the exometabolome and 96 metabolites in the endometabolome (Tables S4 and S5). During storage, a continuous exchange of metabolites occurred between the extracellular and intracellular environments. PLTs are expected to use plasma metabolites to sustain their metabolism. Indeed, glucose and glutamine from an extracellular source were the main metabolic fuels, showing the largest decreases in the exometabolome. Nevertheless, the most significant compositional change in the storage solution involved the accumulation of several metabolites over time, such as lactate, malate, succinate, hypoxanthine, inosine, and xanthine (Fig. 1). The consumption and secretion rate of most of these metabolites was not constant over time, but exhibited a transition point. For instance, in the first day of storage approximately 1 mmol/L glucose was consumed and approximately 2 mmol/L lactate was secreted. From Day 1 to Day 6, the mean glucose consumption dropped to reach approximately 0.6 mmol/L/day, whereas lactate release was measured as approximately 1.2 mmol/L/day (Fig. 1). Other extracellular metabolites, however, including succinate, hypoxanthine, malate, citrate, xanthine, and inosine, showed a transition point after 3 days of storage. Similar to exometabolomic profiles, intracellular metabolites showed changes in concentration that were not linear with time. In addition, some of the intracellular profiles were strongly associated with the extracellular profiles, such as succinate, malate, and citrate, showing a transition point after 3 days of storage. Other metabolites,
such as sphingosine 1-phosphate were almost completely depleted after 3 days of storage (Figs. 1 and 2).
Statistical analysis Pathway analysis Pathway analysis32 was used to investigate which metabolic pathways were most affected during storage of PLTs in commercial PAS. This analysis indicated that purine metabolism, glycerophospholipids and glutathione metabolism, pentose phosphate pathway (PPP), and amino sugars and nucleotide sugar metabolism, as well as glycolysis, were the most affected pathways in the endometabolome during storage (Fig. 3A). Altered purine metabolism was also revealed by pathway analysis of the exometabolome together with identified changes in glyoxylate and dicarboxylate metabolism, tricarboxylic acid (TCA) cycle and glycine, and serine and threonine metabolism (Fig. 3b). All the results obtained by pathway analysis are reported in Tables S7 and S8 (available as supporting information in the online version of this paper). Profiles of extra- and intracellular metabolites related to these pathways are shown in Figs. S1 to S5 (available as supporting information in the online version of this paper).
Multivariate data analysis A total of 104 variables showed a significant change over storage time (fold change, p < 0.05). Systematic changes in the data set were assessed using PCA.40,41 PCA clearly defined a trend in PLT metabolism during storage (Fig. 4). The first principal component (PC1) accounted for 41% of the total variance in the data set. It mainly demonstrated how metabolic concentrations changed over time. In fact, variables such as external lactate, external glucose, and internal glutamine have the highest contribution in the PC1 due to significant changes in concentration levels over the time (Table S9, available as supporting information in the online version of this paper). The second principal component (15% of total variance) represented the variability in the data set during the first 3 days of storage and highlights a metabolic transition. In the PC2 most of the variables with high contribution are intracellular measurements such as lysine, arginine, UMP, malate, xanthine, and mitochondrial activity. However, external measurements such as %CD41+CD63+, hypoxanthine, and K+ (Table S9) also influence this metabolic transition in PC2. Thus, a “short-term metabolic” phenotype was observed from Day 0 to Day 3 (short-term-stored PLTs) and a “medium-term metabolic” phenotype from Day 4 to Day 6 (medium-term-stored PLTs). A third phenotypic group, from Day 7 to Day 10, was also observed (longterm-stored PLTs). This metabolic phenotype may represent nonreversible changes (Fig. 4). Volume 54, November 2014 TRANSFUSION
2915
Fig. 1. Time profiles of selected extracellular concentrations. The update rates were computed from the concentration profiles as ΔC/Δt (see Materials and Methods): extracellular glucose (p < 4.3 × 10−16), extracellular lactic acid (p < 1.4 × 10−7), extracellular succinic acid (p < 0.007), extracellular malic acid (p < 7.9 × 10−10), extracellular hypoxanthine (p < 0.006), extracellular citric acid (p < 0.088), extracellular citric acid (Day 0-Day 3; p < 0.011), extracellular glutamine (p < 0.0001), extracellular xanthine (p < 4.5 × 10−6), and extracellular inosine (p < 0.039). Bars represent SDs. ANOVA test was used to calculate p values (see Materials and Methods).
PAGLIA ET AL.
2916
TRANSFUSION Volume 54, November 2014
METABOLIC PROPERTIES OF STORED PLTs
Fig. 2. Time profiles of selected intracellular concentrations. The polarization of the mitochondrial membrane was measured as described under Materials and Methods. Fold change was normalized to Day 0, and p value was calculated using ANOVA as described under Materials and Methods. Intracellular glucose 6-phosphate (Glucose 6-P; p < 1.2 × 10−9), intracellular sphingosine 1-phosphate (Sphingosine 1-P; p < 6.7 × 10−7), intracellular lactic acid (p < 0.00001), intracellular ATP (p < 0.003), mitochondrial activity JC-1 (p < 4.2 × 10−6), intracellular citric acid (p < 9.7 × 10−7), intracellular malic acid (p < 0.00005), intracellular succinic acid (p < 0.001), intracellular hypoxanthine (p < 1.52 × 10−8), intracellular AMP (p < 1.0 × 10−15), intracellular ADP (p < 0.0005), and intracellular IMP (p < 0.008). Bars represent SDs.
The contribution of each bag to this succession of expressed metabolic phenotypes was very similar for five of the six bags analyzed. Bag 3 presented a trend that did not completely overlay the others, suggesting that interindividual differences might play a role in defining PLTs’ metabolic phenotype during storage (Fig. 4C).
HCA Based on pathway analysis results, metabolites involved in purine metabolism, TCA, lipid metabolism, glycolysis and PPP, glutathione metabolism, and routine QC measurements were analyzed by HCA. The results obtained were consistent with PCA, highlighting a metabolic shift
after 4 days of storage (Fig. 5A). This analysis also highlighted four main clusters associated with the three metabolic phenotypes (Fig. 5A). The first cluster highlights metabolites that demonstrated a fast decrease in concentration after the first metabolic shift. The second cluster represents those metabolites that had a rapid decrease in concentration after the second metabolic shift. In the third cluster we found metabolites that increased their concentration in short-term-stored PLTs. The fourth cluster represents the activation markers that correlated well with some external metabolites (Fig. 5B). In particular, we identified a significant correlation of external malic acid and external hypoxanthine with sP-selectin (Fig. 5B). Volume 54, November 2014 TRANSFUSION
2917
PAGLIA ET AL.
Fig. 3. Pathway analysis. The pathway analysis combines results from pathway enrichment analysis with the pathway topology analysis (see Materials and Methods). Pathway enrichment analysis refers to quantitative enrichment analysis using the compound concentration values. It has the potential to identify consistent changes among compounds involved in the same biologic pathway. The pathway toplogy analysis uses node centrality measures to estimate node importance (number of shortest paths going through the node; see Materials and Methods). Results are shown by plotting at the ordinated p value (−log p) calculated from the enrichment analysis and at the axis of the pathway impact value calculated from pathway topology analysis. (A) Pathway analysis of the endometabolome; (B) pathway analysis of the exometabolome.
DISCUSSION It is known that PLTs undergo a process of decay during storage known as PSL. Lactate accumulates in the storage solution and causes metabolic acidosis and a decrease in pH, contributing to inferior PLT quality.11-13 The data reported here support the theory that PLTs do not undergo a monotonic decay, but rather experience different metabolic phenotypes during storage and that these phenotypic changes may potentially be blood donor dependent. PCA showed the existence of metabolic shifts defining three metabolic phenotypes during storage (Fig. 4). In fact, both extracellular and intracellular profiles were not linear with time and suggested the existence of a strong transition point in PLT metabolism after 4 days of storage. Most of these metabolites are involved in purine metabolism, glycolysis, PPP, glutathione metabolism, glycerophospholipid metabolism, and the TCA cycle, as shown by pathway analysis (Figs. 1-3). These results suggest that the PSL is associated with a series of metabolic changes leading to a different regulation of the energy metabolism. HCA was consistent with PCA, confirming the metabolic shift occurring on Day 3 (Fig. 5A). It also revealed 2918
TRANSFUSION Volume 54, November 2014
four main clusters, which represent the main metabolites and pathways involved in the different metabolic phenotypes (Fig. 5A). Glycolysis, PPP, and glutathione metabolism are strongly associated with short-term-stored PLTs, while purine metabolism and TCA cycle are associated with medium-term-stored PLTs (Fig. 5A). The fourth cluster also highlights a strong correlation between activation marker and secretion of some metabolites such as malate and hypoxanthine (Fig. 5B). Short-term-stored PLTs (Days 0-3) experienced complete conversion of glucose to lactate. Every mole of glucose consumed resulted in 2 moles of lactate secreted into the medium. There is ample evidence that nonoxidative metabolism of glucose is not due to low levels of O2 (pO2 < 2 mmHg) since PLTs are stored in modern gas-permeable storage containers.42-44 Thus, pyruvate was presumably not oxidized in the TCA cycle for ATP production through oxidative phosphorylation. Other substrates can feed into the TCA cycle coupling oxidative phosphorylation with ATP production,45-48 but high secretion of succinate in the first 4 days seems to exclude a normal activity of the TCA cycle (Fig. 1). Indeed, succinate secretion is a known marker for loss of mitochondrial function,49 which is also suggested by the mitochondrial
METABOLIC PROPERTIES OF STORED PLTs
Fig. 4. PCA assumes that the n-dimensional data can be reduced to a linear combination of principal components that best explains the variance in the data. The principal components are rank ordered by the variability that they represent in the data set, with the first principal component accounting for the greatest variability in the data and so on.40,41 (A) PCA performed on all samples; (B) every time point (in days) represents the mean from six bags; (C) every time point (in days) is connected by line to show the specific trend of each bag.
membrane potential profile showing a decrease of approximately 20% in the number of polarized cells in short-term-stored PLT (Fig. 2). In PLTs, intracellular ATP is located in the cytoplasm (metabolic ATP) and in the dense granules (stored ATP).50 The ATP metabolic pool is mainly associated just with anaerobic oxidation of glucose. In short-term-stored PLTs (Days 0-3), the ATP metabolic pool is mainly associated just with anaerobic oxidation of glucose. The down regulation of aerobic oxidation and consequent metabolic stress should then result in increased levels of ADP and AMP degradation products. In fact, inosine, hypoxanthine, and xanthine accumulated in the storage solution (Figs. 1, 2, and 6). The release of malate into solution in the
first 4 days of storage is consistent with its intracellular profile (Figs. 1 and 2). Malate might be generated by citrate, which was present in the storage solution. PLTs consumed citrate, even if the fold change during 10-day storage was not significant (ANOVA p = 0.088). During the first 4 days, there was a significant consumption of citrate (ANOVA p = 0.011). In fact, intracellular citrate accumulated during the first 4 days of storage, and it exhibited a profile similar to malate. These observations suggest that short-term-stored PLTs were able to produce malate from citrate, but not to consume it. Consequently, malate appears to be released into the storage solution (Fig. 6). Short-term-stored PLTs consumed 40% of available glucose and secreted several metabolites, generating, in the first 4 days of storage, a microenvironment that was significantly different from its initial composition. Moreover, accumulation of lactate and malate caused a decrease in pH in the storage solution and potentially also inside the cells. This internal change of pH might affect the activity of some important enzymes such as phosphofructokinase.51 This change in composition and in pH is most likely the root cause of the metabolic shift detected at 4 days of storage (Figs. 4-6), which, according to our results, corresponds to a fundamentally different state of the PLT’s metabolic network. Medium-term PLTs (4-6 days of storage) still converted glucose almost completely to lactate (glucose consumption approx. 0.6 mmol/L/day, lactate secretion approx. 1.2 mmol/L/day), but TCA cycle and oxidative phosphorylation became more active as suggested by changes in extracellular and intracellular levels of succinate. Indeed, the secretion rate of succinate dropped after Day 3, as did its intracellular level, suggesting its consumption by the TCA cycle. Increased mitochondrial membrane potential was consistent with mitochondrial metabolism being more active (Figs. 1, 2, and 6). Since the vast majority of pyruvate coming from glycolysis seems to be converted into lactate, the TCA cycle had to use other substrates. In fact, medium-term-stored PLTs stopped secreting malate, indicating its use in the TCA cycle and the production of ATP by oxidative phosphorylation (Fig. 6). ATP can also be generated by reutilization of hypoxanthine. Short-term-stored PLTs showed decreasing intracellular levels of AMP, IMP, ADP, and ATP and increasing levels of extracellular inosine, hypoxanthine, and xanthine (Figs. 1, 2, and 6). Medium-term-stored PLTs demonstrated an opposite trend, suggesting that the purine metabolism recycled hypoxanthine to produce additional ATP (Figs. 1, 2, and 6). Hypoxanthine is converted to xanthine by xanthine oxidoreductase (irreversible reaction). Once converted into xanthine, it cannot be recycled. In fact, inosine, hypoxanthine, and xanthine were released by short-term-stored PLTs (0-3 days), but Volume 54, November 2014 TRANSFUSION
2919
PAGLIA ET AL.
Fig. 5. HCA and correlation. (A) HCA provides a measure of closeness of metabolites and groups (time points). (B) Section of fourth cluster of the HCA and Pearson correlation of selected metabolites. Malic acid external versus sP-selectin, r = 0.85, p = 3.3 × 10–14; hypoxanhine external versus sP-selectin, r = 0.90, p = 1.6 × 10–18.
the secretion rate slowed down from Day 4 to Day 6 (Figs. 1, 2, and 6), suggesting that hypoxanthine was not converted into xanthine anymore, but was instead being recycled through the purine salvage pathway. In summary, the proposed metabolic phenotype for short-term-stored PLTs involved the down regulation of the TCA cycle and the accumulation of ATP degradation products. The metabolic signature of this phenotype is the accumulation of succinate, hypoxanthine, malate, xanthine, and inosine in the storage solution (Figs. 1 and 6). The development of the short-term-stored PLTs’ metabolic phenotype might be due to the Crabtree effect. In fact, in excess of glucose, cross-inhibition of oxidative phosphorylation might occur to regulate the equilibrium between the cytosolic and the mitochondrial oxidative 2920
TRANSFUSION Volume 54, November 2014
pathways.46,52 The activation of PLTs during the first day of storage as a consequence of collection and storage medium could also play an important role in this process (Table 1). In fact, we observed a significant correlation (p < 0.0001) of activation markers with secretion of hypoxanthine and malate in the storage solutions (Figs. 5A and 5B). Medium-term-stored PLTs instead had moderate activity in TCA cycle and in metabolism, which in part compensated for the depletion of intracellular ATP. This compensation was able to sustain PLT metabolism until Day 6, when PLTs started to experience a faster decay leading to cell lysis and the accumulation of several metabolites in the storage medium, defining the last metabolic phenotypes (long-term-stored PLTs). This
METABOLIC PROPERTIES OF STORED PLTs
different PLTs’ metabolic activity. The interindividual differences that might play a role in defining PLT metabolic phenotypes during storage will also need clinical investigation and will require a higher number of samples. ACKNOWLEDGMENTS The authors thank Marketa Foley, Kristine Wichuk, Manuela Magnusdottir, Steinunn Thorlacius, and the staff at the blood bank for technical support. The authors especially thank Sirus Palsson, Sisse Ostrowski, Marc Abrams, and Hans Gulliksson for their input. GP designed and performed the experiments, analyzed the data, and wrote the paper; OES designed and performed the experiments and wrote the
Fig. 6. Proposed metabolic phenotypes for short-term- and medium-term-stored PLTs. Blue lines represent active metabolic pathways in both phenotypes; black lines
paper; OR wrote the paper and performed ATP analysis; SV performed the MS experiments; MBH designed experiments
represent down regulated metabolic pathways; red lines represent metabolic pathway characteristic of short-term-stored PLTs; and green lines represent meta-
and wrote the paper; SB designed the experiments; SG designed the experiments and wrote the paper; and BOP designed the
bolic pathway characteristic of medium-term-stored PLTs.
experiments and wrote the paper.
metabolic phenotype may represent nonreversible changes and is of less clinical relevance than the first two since PLTs are never transfused beyond 7 days of storage within the present regulatory framework (Fig. 4). Even though data interpretation and the proposed metabolic phenotypes were mainly focused on purine metabolism and the TCA cycle, several pathways are likely involved in the development of the storage lesions and their metabolic phenotype. For instance, glycerophospholipid metabolism might play an important role. (Figs. 3 and S5). Moreover, it is interesting to note that the intracellular pool of sphingosine 1-phosphate (Fig. 2) was completely depleted after 4 days of storage. Recent findings suggest that this metabolite is involved in the control of thrombopoiesis.53 Together, these findings may expand the understanding of the underlying mechanisms behind the storage lesions and may help identify biomarkers that will better enhance the quality of PLTs during storage. The presented metabolic changes are representative for the storage of PLTs collected with particular apheresis equipment and stored in a specific PAS and thus may not represent the changes of buffy coat PLTs collected from whole blood, for instance. Studies of metabolic changes of PLTs processed from buffy coats are in progress. Clinical studies evaluating the quality of short- and medium-term-stored PLTs used in transfusion will be necessary to understand the clinical relevance of
CONFLICT OF INTEREST The authors have disclosed no conflicts of interest.
REFERENCES 1. Michelson AD. Platelets. San Diego (CA): Academic Press; 2002. 2. Hogman CF. Storage of blood components. Curr Opin Hematol 1999;6:427-31. 3. Vasconcelos E, Figueiredo AC, Seghatchian J. Quality of platelet concentrates derived by platelet rich plasma, buffy coat and Apheresis. Transfus Apher Sci 2003;29: 13-6. 4. Dijkstra-Tiekstra MJ, Pietersz RN, Hendriks EC, et al. In vivo PLT increments after transfusions of WBC-reduced PLT concentrates stored for up to 7 days. Transfusion 2004;44:330-6. 5. Hess JR. Red cell changes during storage. Transfus Apher Sci 2010;43:51-9. 6. Bashir S, Cookson P, Wiltshire M, et al. Pathogen inactivation of platelets using ultraviolet C light: effect on in vitro function and recovery and survival of platelets. Transfusion 2013;53:990-1000. 7. Goodnough LT, Shander A, Brecher ME. Transfusion medicine: looking to the future. Lancet 2003;361:161-9. 8. Shrivastava M. The platelet storage lesion. Transfus Apher Sci 2009;41:105-13. Volume 54, November 2014 TRANSFUSION
2921
PAGLIA ET AL.
9. Kocazeybek B, Arabaci U, Akdur H, et al. Prospective evaluation of platelets prepared by single and random methods during 5 days of storage: aspects related to quality and quantity. Transfus Apher Sci 2002;26: 29-34. 10. Akay OM, Gunduz E, Basyigit H, et al. Platelet function testing during 5-day storage of single and random donor plateletpheresis. Transfus Apher Sci 2007;36:285-9. 11. Holme S. Storage and quality assessment of platelets. Vox Sang 1998;74(Suppl 2):207-16. 12. Gyongyossy-Issa MI. Glucose in platelet additive solutions: to add or not to add? Transfus Apher Sci 2011; 44:283-95. 13. Gulliksson H. Defining the optimal storage conditions for the long-term storage of platelets. Transfus Med Rev 2003; 17:209-15. 14. Heuft HG, Goudeva L, Krauter J, et al. Effects of platelet concentrate storage time reduction in patients after blood stem cell transplantation. Vox Sang 2013;105: 18-27. 15. Thiele T, Iuga C, Janetzky S, et al. Early storage lesions in apheresis platelets are induced by the activation of the integrin alphaIIbbeta(3) and focal adhesion signaling
inflammation and clinical outcomes. Thromb Res 2011; 127:287-91. 25. Holmes E, Loo RL, Stamler J, et al. Human metabolic phenotype diversity and its association with diet and blood pressure. Nature 2008;453:396-400. 26. Illig T, Gieger C, Zhai G, et al. A genome-wide perspective of genetic variation in human metabolism. Nat Genet 2010;42:137-41. 27. Fiehn O. Combining genomics, metabolome analysis, and biochemical modelling to understand metabolic networks. Comp Funct Genomics 2001;2:155-68. 28. Paglia G, Magnusdottir M, Thorlacius S, et al. Intracellular metabolite profiling of platelets: evaluation of extraction processes and chromatographic strategies. J Chromatogr B Analyt Technol Biomed Life Sci 2012;898:111-20. 29. Paglia G, Hrafnsdottir S, Magnusdottir M, et al. Monitoring metabolites consumption and secretion in cultured cells using ultra-performance liquid chromatography quadrupole-time of flight mass spectrometry (UPLC-QToF-MS). Anal Bioanal Chem 2012;402:1183-98. 30. Wishart DS, Jewison T, Guo AC, et al. HMDB 3.0–The Human Metabolome Database in 2013. Nucleic Acids Res 2013;41 (Database issue):D801-7.
pathways. J Proteomics 2012;76 Spec No.:297-315. 16. Paglia G, Palsson BO, Sigurjonsson OE. Systems biology of
31. Smith CA, O’Maille G, Want EJ, et al. METLIN: a metabolite mass spectral database. Ther Drug Monit 2005;27:747-
stored blood cells: can it help to extend the expiration date? J Proteomics 2012;76 Spec No.:163-7. 17. Gevi F, D’Alessandro A, Rinalducci S, et al. Alterations of
51. 32. Xia J, Mandal R, Sinelnikov IV, et al. MetaboAnalyst 2.0— a comprehensive server for metabolomic data analysis.
red blood cell metabolome during cold liquid storage of erythrocyte concentrates in CPD-SAGM. J Proteomics
Nucleic Acids Res 2012;40 (Web Server issue):W127-33. 33. Reich M, Liefeld T, Gould J, et al. GenePattern 2.0. Nat
2012;76 Spec No.:168-80. 18. Pienimaeki-Roemer A, Ruebsaamen K, Boettcher A, et al.
Genet 2006;38:500-1. 34. Ostrowski SR, Bochsen L, Salado-Jimena JA, et al. In vitro
Stored platelets alter glycerophospholipid and sphingolipid species, which are differentially transferred to newly released extracellular vesicles. Transfusion 2013;53:612-
cell quality of buffy coat platelets in additive solution treated with pathogen reduction technology. Transfusion 2010;50:2210-9.
26. 19. Antonelou MH, Tzounakas VL, Velentzas AD, et al. Effects
35. Yockey C, Murphy S, Eggers L, et al. Evaluation of the Amicus Separator in the collection of apheresis platelets.
20.
21.
22.
23.
24.
2922
of pre-storage leukoreduction on stored red blood cells signaling: a time-course evaluation from shape to proteome. J Proteomics 2012;76 Spec No.:220-38. Parguina AF, Garcia A. Platelet proteomics in transfusion medicine: a reality with a challenging but promising future. Blood Transfus 2012;10(Suppl 2):s113-4. Zolla L, D’Alessandro A. Shaking hands with the future through omics application in transfusion medicine and clinical biochemistry. Blood Transfus 2012;10(Suppl 2):s1-3. Schubert P, Devine DV. De novo protein synthesis in mature platelets: a consideration for transfusion medicine. Vox Sang 2010;99:112-22. D’Alessandro A, Gevi F, Zolla L. Red blood cell metabolism under prolonged anaerobic storage. Mol Biosyst 2013;9:1196-209. Refaai MA, Phipps RP, Spinelli SL, et al. Platelet transfusions: impact on hemostasis, thrombosis,
TRANSFUSION Volume 54, November 2014
Transfusion 1998;38:848-54. 36. Singh H, Chaudhary R, Ray V. Platelet indices as quality markers of platelet concentrates during storage. Clin Lab Haematol 2003;25:307-10. 37. Macher S, Sipurzynski-Budrass S, Rosskopf K, et al. Function and activation state of platelets in vitro depend on apheresis modality. Vox Sang 2010;99:332-40. 38. Hagberg IA, Akkok CA, Lyberg T, et al. Apheresis-induced platelet activation:comparison of three types of cell separators. Transfusion 2000;40:182-92. 39. Schwertz H, Koster S, Kahr WH, et al. Anucleate platelets generate progeny. Blood 2010;115:3801-9. 40. Trygg J, Holmes E, Lundstedt T. Chemometrics in metabonomics. J Proteome Res 2007;6:469-79. 41. Edward JJ. A user’s guide to principal components. New York: John Wiley & Sons; 1991. 42. Niu X, Arthur P, Abas L, et al. Carbohydrate metabolism in human platelets in a low glucose medium under
METABOLIC PROPERTIES OF STORED PLTs
aerobic conditions. Biochim Biophys Acta 1996;1291:97106. 43. Guppy M, Abas L, Arthur PG, et al. The Pasteur effect in human platelets: implications for storage and metabolic control. Br J Haematol 1995;91:752-7. 44. Kilkson H, Holme S, Murphy S. Platelet metabolism during storage of platelet concentrates at 22 degrees C. Blood 1984;64:406-14. 45. Cartledge S, Candy DJ, Hawker RJ. Citrate metabolism by human platelets. Transfus Med 1997;7:211-5. 46. Baker JM, Candy DJ, Hawker RJ. Influences of pH on human platelet metabolism. Platelets 2001;12:333-42. 47. Whisson ME, Nakhoul A, Howman P, et al. Quantitative study of starving platelets in a minimal medium: maintenance by acetate or plasma but not by glucose. Transfus Med 1993;3:103-13. 48. Saunders C, Rowe G, Wilkins K, et al. Impact of glucose and acetate on the characteristics of the platelet storage lesion in platelets suspended in additive solutions with minimal plasma. Vox Sang 2013;105:1-10. 49. King A, Selak MA, Gottlieb E. Succinate dehydrogenase and fumarate hydratase: linking mitochondrial dysfunction and cancer. Oncogene 2006;25:4675-82. 50. Diab YA, Thomas A, Luban NL, et al. Acquired cytochrome C oxidase impairment in apheresis platelets during storage: a possible mechanism for depletion of metabolic adenosine triphosphate. Transfusion 2012;52: 1024-30. 51. Nishino T, Yachie-Kinoshita A, Hirayama A, et al. Dynamic simulation and metabolome analysis of longterm erythrocyte storage in adenine-guanosine solution. PloS One 2013;8:e71060. 52. Crabtree HG. Observations on the carbohydrate metabolism of tumours. Biochem J 1929;23:536-45. 53. Zhang L, Urtz N, Gaertner F, et al. Sphingosine kinase 2 (Sphk2) regulates platelet biogenesis by providing intracellular sphingosine 1-phosphate (S1P). Blood 2013;122:
Appendix S1. Blood bank quality controls. Fig. S1. Profiles of selected intra- and extracellular concentrations of metabolites involved in purine metabolism. Fold change was normalized to Day 0. Bars represent standard deviations. Fig. S2. Profiles of selected intra- and extracellular concentrations of metabolites involved in TCA cycle. Fold change was normalized to Day 0. Bars represent standard deviations. Fig. S3. Profiles of selected intra- and extracellular concentrations of metabolites involved in glutathione metabolism. Fold change was normalized to Day 0. Bars represent standard deviations. Fig. S4. Profiles of selected intra- and extracellular concentrations of metabolites involved in glycolysis and pentose phosphate pathway. Fold change was normalized to Day 0. Bars represent standard deviations. Fig. S5. Profiles of selected intra- and extracellular concentrations of metabolites involved in glycerophospholipid metabolism. Fold change was normalized to Day 0. Bars represent standard deviations. Table S1. Summary of sample preparation for flow cytometry analysis Table S2. In vitro measurements of apheresis platelet quality in PC samples stored for a period of 10 days. Unless otherwise noted, results are presented as mean ± standard deviation (n = 6) Table S3. Markers of platelet activation and metabolism in apheresis platelets stored for a period of 10 days. Results are presented as mean ± standard deviation (n = 6) Table S4. Targeted metabolites identified in the platelets metabolome Table S5. Identified unexpected metabolites Table S6. Metabolomics quality controls Table S7. Pathway analysis of the endo-metabolome Table S8. Pathway analysis of the exo-metabolome Table S9. Variable contribution to PCA
791-802.
SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher’s Web site:
Volume 54, November 2014 TRANSFUSION
2923
