Red blood cell phenotype fidelity following glycerol cryopreservation optimized for research purposes

Intact red blood cells (RBCs) are required for phenotypic analyses. In order to allow separation (time and location) between subject encounter and sample analysis, we developed a research-specific RBC cryopreservation protocol and assessed its impact on data fidelity for key biochemical and physiological assays. RBCs drawn from healthy volunteers were aliquotted for immediate analysis or following glycerol-based cryopreservation, thawing, and deglycerolization. RBC phenotype was assessed by (1) scanning electron microscopy (SEM) imaging and standard morphometric RBC indices, (2) osmotic fragility, (3) deformability, (4) endothelial adhesion, (5) oxygen (O2) affinity, (6) ability to regulate hypoxic vasodilation, (7) nitric oxide (NO) content, (8) metabolomic phenotyping (at steady state, tracing with [1,2,3-13C3]glucose ± oxidative challenge with superoxide thermal source; SOTS-1), as well as in vivo quantification (following human to mouse RBC xenotransfusion) of (9) blood oxygenation content mapping and flow dynamics (velocity and adhesion). Our revised glycerolization protocol (40% v/v final) resulted in >98.5% RBC recovery following freezing (-80°C) and thawing (37°C), with no difference compared to the standard reported method (40% w/v final). Full deglycerolization (>99.9% glycerol removal) of 40% v/v final samples resulted in total cumulative lysis of ~8%, compared to ~12–15% with the standard method. The post cryopreservation/deglycerolization RBC phenotype was indistinguishable from that for fresh RBCs with regard to physical RBC parameters (morphology, volume, and density), osmotic fragility, deformability, endothelial adhesivity, O2 affinity, vasoregulation, metabolomics, and flow dynamics. These results indicate that RBC cryopreservation/deglycerolization in 40% v/v glycerol final does not significantly impact RBC phenotype (compared to fresh cells).


Introduction
Red blood cell (RBC) cryopreservation was described almost seventy years ago [1] and subsequently explored as means to prolong RBC storage prior to transfusion [2,3]. Of note, RBC cryopreservation has never been optimized to satisfy research requirements, which differ from those for clinical use.
The main challenge in cryopreservation (i.e., the use of very low temperature to preserve intact cells [4]) is not endurance of extremely low temperature, but rather that of an intermediate temperature zone (approx. between -10 to -60˚C) to which cells are exposed twice-during cooling and warming [5]. This intermediate temperature zone may damage cells due to imbalanced transmembrane water transport, a temperature-dependent process [6]. If water is not given enough time to exit, injury occurs from intracellular ice formation (IIF), leading to mechanical shearing [6,7]. However, if water exits too quickly, injury may result from cell shrinkage and related changes in intracellular solute concentrations ("solution effects" injury) [8,9]. Between these two ends of the spectrum, optimal cooling and warming rates exist that minimize freezing injury [10,11].
Adoption of glycerol-based RBC cryopreservation has been limited due to cost (money and time) of the deglycerolization process [28][29][30]. To prevent hypo-osmotic hemolysis during CPA removal, manual washing techniques have been developed using aqueous sugar [31] or salt [28]. Additionally, an automated system has been developed for clinical use [14], in which: (1) residual glycerol concentration is below threshold for hemolysis (of endogenous RBCs)  [29], (2) final wash hemolysis is <1%, and (3) 24h post-transfusion RBC recovery is >75%, meeting current regulatory requirements [32]; however, total RBC loss is typically~15% [20]. This may be of clinical benefit, by culling damaged/senescent cells prior to transfusion. However, for research the goal is to obtain a RBC population identical to that circulating in subjects. Moreover, the most vulnerable cells are likely of greatest interest. RBC assessment following cryopreservation has focused on basic in vitro analyses (hemolysis [2,3,21], morphologic [2,33], rheologic [34], osmotic [2,33,34], surface marker [35], and biochemical [2,3,33]) and on in vivo survival [3,21,36], rather than on physiologic function (O 2 transport, NO processing, vasoreactivity, and metabolism). Given new understanding of the pivotal role played by RBCs in vasoregulation [37], such analyses are relevant as well. Since such phenotypic analyses require intact cells, research has been limited by subject availability/ proximity. To enable separation between subject and sample analysis, we identified need for research-optimized RBC cryopreservation, with simplified processing and maximized fidelity of recovered cells to in vivo phenotype. To achieve these goals, we developed a novel glycerolization procedure, minimizing front-end processing time. Additionally, we optimized a deglycerolization protocol, assessing sample osmolality and CPA removal at each step, thereby minimizing RBC lysis. Finally, to assess cryopreservation impact, we compared the phenotypic profile of cryopreserved/deglycerolized RBCs to that of fresh RBCs.

Blood sampling and preparation of RBCs
This project was approved by the Washington University and Duke IRBs and written informed consent was obtained from all participants. Venous samples (EDTA or heparinized) were drawn from healthy volunteers. Unless otherwise stated, whole blood was centrifuged (1,500-2,000g, 10min, 4˚C); plasma supernatant and buffy coat were removed prior to RBC preparation. All subsequent steps were carried out at RT, given that glycerol flux into RBCs is temperature dependent [38].

Adapted glycerolization
We then modified the cryopreservation process to shorten processing and to allow collection of smaller sample volumes, more appropriate for research analysis (Fig 1A). Initial collection and centrifugation were unchanged from the standard protocol. Briefly, RBCs (300μl) were aliquoted from the bottom of the RBC pellet into cryopreservation tubes containing 300μl buffer and 300μl buffer/glycerol stock (60% v/v glycerol in starting buffer solution). Immediately thereafter, samples were mixed gently by repeated inversion, after which additional PBS-glycerol (900μl, 60% v/v glycerol in starting buffer solution) was added and the samples were again inverted (glycerol final 40% v/v; see S1 Extended Methods). Samples were then placed into a Coolcell freezing container. Matched fresh RBC samples were processed (washed 3X in PBS) for immediate analysis. Diagram of the standard glycerolization protocol and our adapted method, designed to minimize sample processing time and allow collection/processing/storage of smaller RBC aliquots. In the standard glycerolization method, RBC volumes were added dropwise resulting in a final solution that was 40% w/v glycerol; in the adapted method, RBCs were added as a bolus, resulting in a final 40% v/v glycerol solution. All centrifugation was performed at 1,500g, 5min, at 4˚C. (B) Whisker-plot comparing RBC lysis resulting from the addition of buffer/glycerol (60% v/v glycerol buffer stock added to samples) at different temperatures to pRBCs in a one step (bolus addition; 300μl pRBC to 600μl buffer/glycerol stock) or two step (300μl pRBCs to 300μl buffer to 300μl buffer/glycerol stock = step one, followed by addition of further 900μl buffer/glycerol stock = step two) process. Note, both procedures resulted in a 40% v/v glycerol final. Additionally in this experiment, no samples were frozen. Glycerol solutions were added and then samples immediately centrifuged and lysis measured. (C) Whisker-plot comparing the order of addition of pRBC, buffer, and buffer/glycerol (60% v/v glycerol buffer stock added to samples) in the two step glycerolization procedure, on the amount of lysis following sample freezing at

Adapted deglycerolization
We optimized the above method to minimize RBC lysis by extending the dilution sequence (reducing osmolality less rapidly). The process required~1hr (Fig 2A) (see S1 Extended Methods).

RBC lysis
To determine the extent of RBC lysis, we assayed Hb concentration ([Hb]) spectrophotometrically (Biotek μQuant, Winooski, VT or FLUOstar Omega, BMG Labtech, Cary, NC) in both the post-centrifugation supernatant (numerator) and the original RBC-glycerol suspension (denominator), accounting volumes of each, and multiplying by 100 to index hemolysis to the starting total [Hb].
[Hb] assays and calculation of lysis percentage were repeated at each stage following cryopreservation, at various points during deglycerolization (Fig 2A) and successive washes. An alternative lysis calculation is outlined in S1 Extended Methods.
[Hb] values from supernatants and RBC lysates did not differ whether measured by simple spectrophotometric or Drabkin's assays, using the following formula: Heme mM ð Þ ¼ ðsample absorbance @ 540 nMÞ X ðdilution factorÞ ðextinction coefficient @ 540 nMÞ X ðpathlength correctionÞ where the extinction coefficient was 11mM -1 cm -1 (this extinction coefficient is for the product of the reaction between Drabkins reagent and hemoglobin, i.e., cyanomethemoglobin) and the pathlength correction was 0.8 for a 300μl volume in a 96-well plate.
-80˚C and thawing for 2 minutes at 37˚C. Order of addition of components is highly important. (D) Whisker-plot comparing RBC lysis following sample freezing and thawing, for the standard (40% w/v glycerol final) and our adapted (40% v/v glycerol final) glycerolization procedure. Mean RBC lysis was not different between the two methods; however, our adapted glycerolization procedure was 40% more time efficient (n = 9 matched samples from 3 individual blood donors). https://doi.org/10.1371/journal.pone.0209201.g001

Sample osmolality
Osmolality was determined by freezing point depression (Osmette A, Precision Systems Inc, Natick, MA).

Scanning electron microscopy
See S1 Extended Methods for details; briefly, RBCs were fixed to poly-L-lysine coated coverslips, dehydrated in ethanol, dried (Leica EM CPD 300, Vienna, Austria), mounted on aluminum stubs and sputter coated with iridium (Leica ACE 600, Vienna, Austria), then loaded into a FE-SEM (Zeiss Merlin, Oberkochen, Germany) and imaged (3KeV, 300pA) using the Everhart Thornley secondary electron detector. Cell morphology was quantitated using a standard method [40]. Briefly, three separate images of cells (~50-200 cells per image), either fresh, glycerolized, thawed, or deglycerolized, were counted and categorized according to shape, into one of six classes; (a) smooth disc (1.0), (b) crenated disc (0.8), (c) crenated discoid (0.6), (d) crenated spheroid (0.4), (e) crenated sphere (0.2), (f) smooth sphere (0). The percentage of each cell type was multiplied by the factor assigned to each cell type (in parentheses above), and the sum of these products was added together to calculate the morphological index (Im).

RBC adhesion (ex vivo)
RBC adhesivity to endothelium, in a custom variable height plexiglass flow chamber, was determined as described [41][42][43][44] (see S1 Extended Methods). Labeled RBCs (fluorescent dye PKH26) were introduced to the chamber (1.5ml/min), then allowed to dwell for 5-10min. The cell number at each location (height) was recorded. Then, 5-10min of fluidic flow was conducted with PBS at rates calculated to produce shear stress ranging 1-10dynes/cm 2 and the number of adherent cells at each location was re-counted. Shear stress and percent adhesion were calculated at each height.

O 2 affinity
O 2 -hemoglobin dissociation curves (ODC) were determined in vitro (Hemox-Analyzer, TSC Scientific, New Hope, PA). The p50 (pO 2 at which HbSO 2 is 50%) was extrapolated; the Hill co-efficient was also calculated from these data (TCS Hemox Data Aquisition System, TCS Scientific Corp, New Hope, PA).
Samples were prepared for UHPLC-MS metabolomics, as described [50] (also, see S1 Extended Methods). The Q Exactive mass spectrometer (Thermo, Bremen, Germany) was operated in Full MS mode (2μscans) at 70,000 resolution in the 60-900m/z range, 4kV spray voltage, 15 sheath gas and 5 auxiliary gas, in positive and negative ion modes (separate runs). Metabolite assignments and isotopologue distributions were determined using the software Maven (Princeton, NJ) [54] and assignments confirmed against 700 standards (IROATech, Sigma Aldrich, St. Louis, MO).

In vivo analysis of RBC adhesion, RBC velocity, and Hb saturation
Nude mice (n = 16; Foxn1 nu ; Jackson Labs, strain 002019) were anesthetized (isoflurane) and fitted with a dorsal window chamber, as described [56]. After recovery (1-3d), mice were anesthetized (isoflurane), placed upon a temperature-controlled platform and inspected to verify intact microcirculation. PKH-26 labeled RBCs (10μL/g body weight) were infused over 10sec and label movement was tracked by intravital microscopy (5X, Zeiss Axioobserver Z1, Zeiss Zen software). Videos (transmission and red fluorescent channel) and still images of the microvasculature were captured for each subject and analyzed offline for RBC adhesion, RBC velocity, and Hb saturation (ImageJ, NIH)(MatLab R2017b, Mathworks) [57].
Adhesion was measured by time-averaging a~60sec sequence, yielding a composite image in which adherent cells appeared as distinct fluorescent entities, while moving cells blurr. Individual adherent RBCs were then hand-counted. Adhesion values for each subject were calculated and presented as percent adherent cells relative to total cells present in a single frame.
To measure RBC velocity, five venules of various sizes were randomly selected from each subject's video. Individual RBCs were selected from these venules, then tracked through 5-10 frames of movement. Because individual RBC velocities through different venules is non-uniform, two cells were selected from a particular venule and their velocities through a shared segment of that venule were averaged. Venule diameter was measured perpendicular to the midpoint of the travel path. Each two-cell velocity average was then normalized to the inverse square of the radius of the venule segment through which the cells passed. This provided five normalized RBC velocities per subject, which were then averaged to yield a single (n = 1) average radius 2 -normalized RBC velocity per subject.
To measure Hb saturation, still images of each subject were taken at a range of wavelengths and processed (Matlab) to generate a Hb saturation heat map of the region's venules [56]. Five segments corresponding to those used for the velocity analysis were selected for each subject. The Hb saturation values of these segments were averaged to give a single final averaged Hb saturation percentage value for each subject.

Statistical analyses
Results are presented as mean ± standard error of the mean (SEM). Column statistics were performed to determine data distribution normality. Where appropriate, groups were compared by t-test (Student's or Wilcoxon rank) or repeated measures analysis of variance (RMA-NOVA; Bonferroni's posthoc test) (PRISM, GraphPad Inc.; La Jolla, CA).

Standard morphological imaging (SEM) and measurement (CBC)
SEM images were acquired at each stage of sample assessment (Fig 3A and S1 Fig). Fresh RBCs demonstrated the characteristic bi-concave disc shape, with minimal morphological variance. Following glycerolization, the majority of cells appeared spherocytic, as reflected in the low morphological index (Im = 40.6 ± 2.1). After thawing, RBCs were observed to be embedded in a highly structured cryopreservation matrix, with almost all cells demonstrating a crenated sphere morphology (Im = 20.1 ± 0.1). Following full deglycerolization, cell morphology was indistinguishable from that for fresh RBCs, with no significant difference in morphological index (Im = 97.1 ± 1.1 vs 99.1 ± 0.2, respectively; paired t-test p = 0.26).
Deformability. Deformability of deglycerolized and fresh RBCs was identical across a physiologically relevant range of shear stress (Fig 3D). The shear stress at which RBCs demonstrated 50% elongation (SS 50 ) was did not differ between fresh and deglycerolized RBCs (following prolonged buffer immersion) (Fig 3E).

RBC-based hypoxic vasodilatory response.
In an organ-chamber bioassay of endothelium-intact rabbit aortic rings held at low O 2 tension (~1% O 2 ), matched fresh and deglycerolized RBCs demonstrated similar vasodilation (Fig 4E). Single representative traces were plotted to visually demonstrate the relaxation response (Fig 4D).
RBC NO content. Photolysis-chemiluminescence was used to measure RBC hemoglobin NO content in matched: fresh, frozen (without glycerol) and cryopreserved RBCs (without/ with glycerol removal). Total Hb NO content was not significantly different between any of Illustration of the characteristic biconcave disc morphology of fresh RBCs. Following glycerol addition, the majority of RBCs become swollen and spherocytic. After thawing, RBCs remain encapsulated in a glycerol matrix. Following deglycerolization RBCs return to their original morphology. (B) NaCl induced osmotic lysis in matched fresh, deglycerolized, and deglycerolized RBCs exposed to a prolonged buffer incubation (n = 5 or 10 individual donors, shown on graph). NaCl induced osmotic lysis increased immediately following deglycerolization, but returned to the original fresh phenotype following prolonged incubation (1 hour) of RBCs in the final deglycerolization buffer. (C) Half maximal effective concentration (EC 50 ) increased immediately following deglycerolization, in comparison to fresh RBC samples (p < 0.05). However, prolonged incubation (1 hour) of RBCs in the final deglycerolization buffer at 37˚C restored the original fresh RBC phenotype (n = 5 or 10 individual donors, shown on graph). (D) Shear induced maximal RBC deformability (Elongation Index maximum; EI max ) was not different between matched fresh and deglycerolized RBCs (exposed to the prolonged buffer immersion) (n = ±6 individual donors). (E) The shear stress (Pa) at which RBC demonstrated 50% elongation (SS 50 ) was also not different between matched fresh and deglycerolized RBCs (exposed to the prolonged buffer immersion) (n = 6 individual donors). (F) RBC adhesivity to endothelium was determined ex vivo. A trend was observed toward a decrease in RBC adhesivity in the deglycerolized RBCs, across all sheer stresses > 1 dyne/cm 2 , that did not reach statistical significance.

Biochemical analysis
Metabolomics-steady state phenotypes and metabolic fluxes in response to oxidant stress. UHPLC-MS metabolomics were performed to compare phenotypes of fresh vs cryopreserved and deglycerolized RBCs (Fig 5A-5H). Analyses of both RBCs and media indicated similar glucose consumption rates (Fig 5D and 5F) and total lactate generation (Fig 5E  and 5G) in both groups over a 6h time course. Oxidative loading (SOTS-1) was employed to unveil more subtle metabolic differences between fresh and deglycerolized RBCs, by extending the dynamic range for metabolite flux. Supplementation to the media of [1,2,3-13 C 3 ]glucose was used to determine the rate of lactate isotopologue enrichment via glycolysis (M+3) or the pentose phosphate pathway (M+2). Results indicated that fresh RBCs were characterized by significantly higher levels of supernatant-but not intracellular-M+2 lactate (Fig 5H). Conversely, deglycerolized RBCs were characterized by faster glycolysis than fresh RBCs (Fig 5H). Even though lower lactate M+2 export in deglycerolized RBCs following SOTS-1 was observed, intracellular glycolysis to pentose phosphate pathway ratios were comparable between fresh and deglycerolized RBCs following treatment with SOTS-1 (Fig 5H). Deglycerolized RBCs were also characterized by lower levels of oxidized glutathione (GSSG) and higher reduced to oxidized glutathione ratios (GSH/GSSG) in comparison to fresh RBCs treated with SOTS-1, suggestive of decreased oxidative damage upon oxidative loading, in comparison to SOTS-1-treated fresh RBCs (not shown).
Considering the minor differences noted in (above) tracing exeperiments, to further compare the metabolic phenotypes of fresh and cryopreserved RBCs, we performed unsupervised analyses of metabolomics data via PLS-DA ( Fig 6A) and hierarchical clustering. Dendrograms for time course data from control vs DG +/-SOTS-1 are shown in Fig 6B. The impact of time and oxidative loading (SOTS-1) informed sample clustering along principal components 1 and 2 (explaining 21.9 and 17.4% of the total variance, respectively). Both PLS-DA and hierarchical clustering analyses confirmed that control and DG groups fell into identical clusters at the tested time points, both in the absence and presence of SOTS (Fig 6A and 6B). Results are further shown in the form of heat maps (Fig 6C and S2 Fig), visually indicating a time course and SOTS-1 dependency of the metabolic phenotypes observed both in control and DG groups. The only notable exception was glycerol-phosphate, a product of glycerol metabolism, observed to be significantly higher in DG RBCs at 0h time point and declining to 0h control levels just after 6h.
In vivo RBC analysis RBC response to transfusion in nude mice (mapped hemoglobin O 2 saturation, RBC velocity, and RBC adhesion). Post-transfusion function of human RBCs was assessed in a nude mouse model (Fig 7A), with comparison made between fresh or deglycerolized RBCs. Labeled human RBCs were transfused into healthy nude mice in volumes mimicking a 2-unit transfusion in an adult human. As described [58,59], control fresh healthy human RBCs did not alter blood hemoglobin O 2 saturation maps. There was also no effect on (transfused) RBC velocity. Adhesion of control transfused human RBCs in vivo was low, and adhesion of transfused DG RBCs did not differ significantly.
We found that the cryopreservation protocol did not alter mapped hemoglobin O 2 saturation in a dorsal skinfold window chamber model either qualitatively or in quantitative Hb saturation analysis (Fig 7B and 7C), nor was RBC velocity altered ( Fig 7D). There was a modest but statistically insignificant effect of cryopreservation on in vivo adhesivity of transfused RBCs (Fig 7E), as compared to that of fresh human RBCs from the same donors.

Discussion
Here, we report a RBC cryopreservation protocol optimized to satisfy requirements particular to RBC research. This protocol improved glycerolization efficiency (~40% time reduction to achieve 40% v/v, glycerol final) and was equivalent to the standard method [28,38,39] in terms of RBC recovery following freezing (-80˚C) and thawing (37˚C). Of note, following full glycerol removal (>99.9%), our modified procedure reduced total cumulative lysis by~35%, in comparison to the standard method. These improvements minimize the burden of initial sample processing, render cryopreservation protocols more feasible for clinical research staff, improve RBC recovery following cryopreservation and enable analyses of vulnerable RBC subpopulations that might otherwise be lost. Our modified methodology resulted in a post-cryopreservation/deglycerolization RBC phenotype that was indistinguishable from that of fresh RBCs, with regard to physical RBC parameters (morphology, volume, and density), osmotic fragility, deformability, endothelial adhesivity, O 2 affinity, vasoregulation, metabolomics, and flow dynamics.
In the process of inter-laboratory method validation, we noted that sample handling during both the glycerolization and deglycerolization procedures influenced RBC lysis amount. Others [34] have also reported that RBC sample processing contributes more to cryostored RBC quality than does storage duration. Specificially, we determined that when adding packed RBCs to glycerolizing solutions, immediate gentle sample inversion (to fully suspend cells) was necessary to prevent RBC clumping and greater post-thaw lysis. Similarly, throughout the deglycerolization procedure, lysis was increased by excessive agitation or tube plucking/tapping to accelerate RBC re-suspension. Ultimately, simple rotation of the tube on its axis (to mix the sample) proved optimal. We did not address the effect of storage duration on RBC recovery or phenotype fidelity, as all samples were analyzed and measured within a month of freezing.
Our cryopreservation protocol was specifically developed for maximal recovery of human RBCs. It should be noted that plasma osmolality [60][61][62] as well as RBC size and volume (and subsequently surface area to volume ratios) [63][64][65][66][67][68] differ widely across species commonly used in RBC research, all of which may contribute to differing tolerance to osmotic and mechanical stress. It is thus likely that species-specific adjustments to this cryopreservation protocol may be needed to attain similar results. Additionally, even with human RBCs, we still observed 8% lysis following cryopreservation. We speculate that the cells lost may represent cells or cell sub-population(s) least resilient to osmotic or mechanical stress, suggesting possible need for additional method validation prior to study of subjects with such pathologies. With this in mind, while we focused on glycerol as our CPA due to its lack of toxicity and extensive literature on its clinical use, other cryoprotectants are available (both cell-penetrating and non-cell penetrating) that might better address this issue.
The finding that there was a small degree of baseline adhesion of control transfused human RBCs is in agreement with prior published findings [58,59]. It is possible that the slightly larger (EMP) glycolytic pathway (generating isotopologue M+3) or through the pentose phosphate pathway (PPP), generating isotopologue M+2 of lactate, (C) owing to the release of the first carbon atom of heavy glucose in the form of 13 CO 2 at the oxidative phase reactions of the PPP. (F) Schematic overview of glucose uptake and consumption in RBC supernatants and cells, in fresh, or deglycerolized RBCs, either untreated, or treated with the superoxide generator (SOTS-1), (D) demonstrating in both RBCs and media similar glucose consumption rates. (E) Total lactate, i.e., sum of lactate isotopologues (unlabeled M+0, M+3 generated via glycolysis, or M+2 generated via the PPP), in fresh and DG RBCs, either untreated or treated with the superoxide generator (SOTS-1), demonstrating in both RBCs and media similar lactate accumulation. (G and H) Schematic overview of expected isotopologues and relative ratios of lactate, from glycolysis (EMP, M+3 isotopologue) or PPP (M+2) and relative ratios in fresh and DG RBCs, either untreated, or treated with the supoxide generator SOTS-1. (H) Fresh RBCs were characterized by significantly higher levels of supernatant-but not intracellular-M+2 lactate. Conversely, deglycerolized RBCs were characterized by faster glycolysis than fresh RBCs. Lower lactate M+2 export in deglycerolized RBCs following SOTS-1 was observed, while intracellular glycolysis to pentose phosphate pathway ratios were comparable between fresh and deglycerolized RBCs following treatment with SOTS-1. size of human red cells may lead to pseudostatic appearance in mouse capillaries, where some fraction of the transfused RBCs might require a slightly longer transit time. In addition, there may be infrequent, random adhesion of even murine RBCs to murine (or human to human) endothelium in vivo. The adapted cryopreservation technique did not increase RBC adhesivity.
Unsupervised analyses of metabolomics data confirmed similarity of metabolic phenotypes of fresh and deglycerolized RBCs, both at rest and under oxidant loading conditions, affirming the conclusion that our research-optimized procedure for RBC cryopreservation and deglycerolization in 40% v/v glycerol final does not significantly impact RBC phenotype (compared to fresh cells).