Effects of Storage Time on Glycolysis in Donated Human Blood Units

Background: Donated blood is typically stored before transfusions. During storage, the metabolism of red blood cells changes, possibly causing storage lesions. The changes are storage time dependent and exhibit donor-specific variations. It is necessary to uncover and characterize the responsible molecular mechanisms accounting for such biochemical changes, qualitatively and quantitatively; Study Design and Methods: Based on the integration of metabolic time series data, kinetic models, and a stoichiometric model of the glycolytic pathway, a customized inference method was developed and used to quantify the dynamic changes in glycolytic fluxes during the storage of donated blood units. The method provides a proof of principle for the feasibility of inferences regarding flux characteristics from metabolomics data; Results: Several glycolytic reaction steps change substantially during storage time and vary among different fluxes and donors. The quantification of these storage time effects, which are possibly irreversible, allows for predictions of the transfusion outcome of individual blood units; Conclusion: The improved mechanistic understanding of blood storage, obtained from this computational study, may aid the identification of blood units that age quickly or more slowly during storage, and may ultimately improve transfusion management in clinics.

Research donors were screened by health history questionnaire and vital signs and provided consent to donate whole blood. The metabolomics data from these donations consist of measurements of serially sampled units of leukoreduced ADSOL red blood cells (RBCs) over 42 days of refrigerated storage.
Specifically, the data came from twelve units donated by nine volunteers (please refer to Table S1 for demography). Among these, six volunteers each donated one unit (group1, donors: X1, X2, X3, X4, X5, X6); three additional volunteers donated two units with several months between donations (group2 and group2_match, donors: X850, X867, X1145). After leukoreduction and removal of platelet-rich plasma, the residual RBC pellet was mixed with ADSOL additive solution, and the packed RBC unit was stored at 2-6°C for up to 42 days. At selected time points, RBC bags were gently but thoroughly mixed, and 1 mL samples were aseptically removed, added to labeled cryovials, snap frozen on liquid nitrogen, and stored at −80°C. Samples from each study were stored until all time points were collected and then analyzed with gas chromatography/mass spectrometry and liquid chromatography/tandem mass spectrometry. For the former analysis, samples were dried, derivatized using bistrimethyl-silyl-triflouroacetamide, and run on a Thermo-Finnigan Trace DSQ fast-scanning single-quadrupole mass spectrometer (Waltham, MA) using electron impact ionization. The LC/MS/MS used a Waters ACQUITY UPLC (Milford, MA) and a Thermo-Finnigan LTQ mass spectrometer, consisting of an electrospray ionization source and linear ion-trap mass analyzer. Specific compounds were identified by comparison to library entries of purified standards or recurrent unknown entities. The peak areas for each named metabolite were log-transformed and normalized to Bradford protein content.
The ADSOL medium contained adenine, dextrose, and other ingredients. The possible direct regulations of reaction kinetics by these ingredients were already accounted for by the kinetic formalisms we used. The indirect effects of these molecules, by means of affecting secondary metabolites, are accounted for by the metabolomics data and their usage in the kinetic models and the stoichiometric model. V_PEP_PYR_PK; V_PYR_LAC_LDH; V_RU5P_R5P_R5PI; V_RU5P_X5P_X5PI; V_X5PE4P_F6PGA3P_TK2; V_X5PR5P_GA3PS7P_TK1]. The format of the flux names starts with the capital letter V, then substrate, product, and the catalyzing enzyme. These four parts within a flux name are separated by an underline symbol. For abbreviations of enzymes, please refer to the legend of Deriving fluxes from metabolomics data.
The dynamics of some fluxes can be directly derived from the metabolomics data with a stoichiometric model and the method of dynamic flux estimation (DFE); in other words, they do not require the use of a kinetic model.

Kinetic formulations for individual biochemical reactions.
The following kinetic models surveyed from the literature were used in this study to compute fluxes under normal physiological conditions at discrete time points. To determine the enzymatic activity of the HK reaction, we used a kinetic formulation from the literature [2], namely:  For the enzymatic activity of the ALD reaction, we used a kinetic model proposed by [3]  For the rate of the PK reaction, we used the following kinetic formulation [5]:

Donation batch
Differences among batches increase during the first two weeks of storage, and then decrease until the end of storage. Donor variations are much smaller than batch variations, especially during weeks 2 to 4 of storage.

Reaction dependency
Storage time effects are reaction dependent, and this reaction dependency is consistent among donors.
ALD, HK, and PK

Kinetic model
Quantification of storage effects on a specific flux is consistent for all donors in regard to different kinetic models. PFK Figure S1. Dynamics of metabolite levels for donors in group2 and group2_match. After conversion to absolute concentrations, the data were calibrated according to the signal intensity of carbon atoms in consideration of 5-6% instrument variability and 13-18% total process variability. Shown here are the dynamic levels of metabolites for donors in group2_match and group2. Symbols are experimental data points, while lines represent the interpolated data set. The X-axis shows storage time (unit: days), while the Y-axis represents absolute metabolite concentrations (unit: µM). Figure S2. Dynamics of metabolite levels for donors in group1. After conversion to absolute concentrations, the data were calibrated according to the signal intensity of carbon atoms in consideration of 5-6% instrument variability and 13-18% total process variability. Shown here are dynamic levels of metabolites for the donors in group1. Symbols are experimental data points, while lines represent the interpolated data set. The X-axis shows storage time (unit: days), while the Y-axis represents absolute metabolite concentrations (unit: µM). For easier comparison with Figure S1, the plots are arranged in the same manner. Some metabolite measurements were not available.