Superoxide measurement as a novel probe of red blood cell storage quality

INTRODUCTION

Standard refrigerator storage causes metabolic, protein, and lipid changes in human red blood cells (RBCs). Depending on processing and donor variability, the resulting changes can negatively impact post-transfusion viability of the RBCs¹. Donor blood processing is a current topic of research directed toward optimizing methods to assess post-transfusion effectiveness^2,3, while research on donor variability focuses on defining the contributions of donor-specific characteristics, e.g., the exposome, diet, drugs, age, sex, and genetics^4,5. These factors are now known to be cumulative contributors to changes in the molecular, biochemical and physiological quality of RBCs intended for transfusion. Oxidative stress is a major modifier of normal RBC biology. Within the ex vivo setting of refrigerator storage, cumulative oxidative damage may increase because of ineffective or inadequate protective and repair processes⁶. A current hypothesis suggests that, rather than just chronological age, donor RBC differences may accelerate aberrant changes in proteins, lipids, and metabolism that are known to decrease RBC quality⁷. Therefore, new methods for assessing factors that are more determinative of donor blood quality are a logical step toward optimizing blood storage methodology.

One unit of packed RBCs for transfusion contains approximately 60 grams of Hb and 210 mg of iron⁸. Iron must be coordinated within each globin chain’s heme to facilitate efficient O₂ transport through allosteric transitioning between oxyHb (HbFe²⁺O₂) and deoxyHb (HbFe²⁺). Within RBCs, HbFe²⁺O₂ is involved in a continuous cycle of spontaneous one-electron oxidations that generates superoxide (O₂ ^{• −}) and metHb (HbFe³⁺) (Figure 1A). Autoxidation of erythrocytic Hb proceeds at different rates for the α- and β-chains, and is influenced by Hb concentration, pH, and allosteric effectors⁹. To our knowledge, autoxidation is the only experimentally defined source of O₂ ^{• −} in stored RBCs. Under conditions of homeostasis, NADH-dependent cytochrome b5 reductase (CB5R, also termed metHb reductase or diaphorase 1) efficiently reduces HbFe³⁺ to HbFe²⁺ to maintain low levels of oxidized Hb, absent congenital CB5R deficiencies¹⁰. In addition to autoxidation, refrigerator-stored RBCs are primed for oxidative stress due to impaired generation of antioxidant enzyme cofactors NADH and NADPH through glycolysis and the pentose phosphate pathway (PPP), respectively (Figure 1A)^11,12.

Figure 1.

Key biochemical pathways in erythrocytes and the use of a molecular probe for detecting superoxide by electron paramagnetic resonance (EPR)

A. Glycolysis and the pentose phosphate pathway (PPP) generate the reducing cofactors, NADH and NADPH, respectively. NADPH serves detoxifying systems based on glutathione (GSH) and thioredoxins (Trx). During O₂ transport, a small fraction of oxyhemoglobin (HbFe²⁺·O₂) transforms into methemoglobin (HbFe³⁺) with concomitant generation of superoxide (O₂ ^{• −}), which is membrane-impermeant and thus retained in the RBC. Regeneration of ferrous hemoglobin (HbFe²⁺) by CB5R requires NADH. Reactive O₂ ^{− −} is transformed by superoxide dismutase 1 (SOD1) into hydrogen peroxide (H₂O₂), which is, in turn, detoxified by multiple enzymes, i.e., catalase (Cat), peroxiredoxins (Prx), and glutathione peroxidase (GPx). The cyclic hydroxylamine, CMH¹⁵, is membrane-permeant¹⁶ and thus readily enters RBCs, where it is rapidly oxidized by O₂ ^{− −} to form the nitroxide, CM•, which is also membrane-permeant¹⁷. B. CMH is EPR-silent, whereas the product of its oxidation, CM•, is EPR active and exhibits a characteristic 3-line EPR spectrum. The height, h, of the center line (peak) is used for quantitation in this study.

Murine blood storage offers the advantage of intra-strain genetic homogeneity in proof-of-concept testing. To date, PTR experiments in multiple mouse strains demonstrate metabolic correlates to both fresh and stored RBC^13,14. Here we hypothesize that X-band electron paramagnetic resonance (EPR) spectroscopy and cyclic hydroxylamine-based spin probing of murine RBCs known to store well (C57BL/6J) and poorly (FVB) under refrigeration can identify the generation of O₂^{• −} as a predictive biomarker of oxidative stress. We further suggest that O₂^{• −} kinetics during storage are consistent with reported PTR in these mice^13,14.

MATERIALS AND METHODS

General

Chemicals were purchased from commercial vendors and used without further purification. Xanthine oxidase and superoxide dismutase from bovine erythrocytes were from Sigma-Aldrich (St. Louis, MO, USA). High-resolution mass spectra were acquired on an electrospray ionization spectrometer (AccuTOF-CS, JEOL, Peabody, MA, USA). ¹H-NMR spectra were recorded in deuterium oxide on a 400-MHz spectrometer (400 MR, Varian, Palo Alto, CA, USA). EPR spectra were recorded on an X-band spectrometer (EMXnano, Bruker Corp, Billerica, MA, USA). EPR samples were contained in 50-μL borosilicate capillary micropipettes (Drummond Scientific Company, Broomall, PA, USA) whose ends were closed with sealing clay.

Synthesis of 1-hydroxy-3-methoxycarbonyl-2,2,5,5-tetra-methylpyrrolidine hydrochloride (CMH HCl salt): 1-Benzyloxy-3-methoxycarbonyl-2,2,5,5-tetramethyl-pyrrolidine (0.226 g, 0.776 mmol), prepared from the known compound, 3-methoxycarbonyl-2,2,5,5-tetra-methylpyrrolidinyl-1-oxyl¹⁸, through copper-catalyzed O-benzylation¹⁹, was dissolved in ethanol (10 mL) in a hydrogenation bottle, to which 5% palladium on activated carbon (143 mg) was added. The mixture was agitated overnight under 27 psi H₂ in a hydrogenation apparatus (Parr Instrument Company, Moline, IL, USA). Thereafter, thin-layer chromatography showed complete consumption of starting material. The reaction mixture was filtered through Celite and the filtrate was acidified with concentrated HCl (72 μL). The solvent was removed by rotary evaporation to leave a clear oil which, upon drying under vacuum, crystallized to give a quantitative yield of CMH•HCl. The compound was stored at −20°C under dry argon. ¹H-NMR (δ): 1.40–1.68 (m, 12H), 2.27–2.50 (m, 2H), 3.41–3.46 (m, 1H), 3.79 (s, 3H). HRMS: [M+H]⁺ C₁₀H₂₀NO₃ requires 202.1443, observed 202.1436.

Measurement of CMH oxidation by EPR spectroscopy

All samples for EPR measurement contained 1 mM CMH. Other reagents, when used, were at the following concentrations: xanthine oxidase (XO), 0.03 U/mL; superoxide dismutase (SOD), 5 μM; hypoxanthine (HX), 0.5 mM; hydrogen peroxide (H₂O₂), 1 mM. Cell-free measurements were made in phosphate-buffered saline (PBS) containing 1 mM diethylenetriaminepentaacetic acid (DTPA). For measurements on RBCs, each sample was made by mixing 2 μL of 50 mM CMH•HCl stock solution with 98 μL of RBC suspension (~40% hematocrit) in Citrate Phosphate Dextrose Adenine (CPDA-1, Sigma-Aldrich, St. Louis, MO, USA). Time zero (t=0) is the time at which CMH is mixed into the sample. The amplitude (h) of the center peak of the CM• nitroxide EPR spectrum was converted into concentration (C in μM) by using the calibration equation C=(h–0.03308)/0.06461, determined using a series of standard samples.

Murine blood collection and storage

Mouse blood collections were approved by the University of Maryland, Baltimore Institutional Animal Care and Use Committee (IACUC) (protocol # 0520008) and two different strains (C57BL/6J or B6 for short, Jackson Laboratory, Barr Harbor, ME and FVB, Charles River, Worcester, MA, USA) were studied. Blood was collected via cardiac puncture into citrate phosphate dextrose from n=15 mice/strain, pooled, centrifuged, followed by removal of buffy coat and plasma. CPDA-1 (12.5% of the total volume) was added to RBCs (hematocrits 70%). Aliquots were stored in 1.7 mL sterile Eppendorf tubes in a refrigerator at 4–6°C. Measurements were made as four technical replicates from the same pooled RBC tube on day 1 (immediately after processing) and then on days 2, 4, and 8 after processing.

Data analysis and presentation

Least-squares curve fitting and statistical analysis were performed using OriginPro software (OriginLab Corp., Northampton, MA, USA). Results are presented as mean ± SD. Difference between the two strains on the same day was assessed by t-test; within-strain differences over multiple days of storage were assessed through ANOVA with Scheffé post hoc test.

RESULTS

CMH oxidation by superoxide

The time course of the reaction of CMH with O₂ ^{• −} is shown in Figure 2A (red symbols and lines, label “XO”). Oxidation of hypoxanthine (HX) by xanthine oxidase (XO) generated O₂ ^{• −}. CMH, which is EPR-silent, was oxidized by O₂ ^{• −} to CM•, a stable nitroxide with a 3-line EPR spectrum. The spectral peak height was converted to CM• concentration and plotted against time. As oxidation proceeded, CM• accumulated and its concentration rose steadily. Addition of superoxide dismutase (SOD) inactivated O₂^{• −} efficiently, and blocked the conversion of CMH to CM• (Figure 2A, green symbols and lines, label “XO+SOD”) (p=3.53×10⁻⁵). Because SOD converts O₂ ^{• −} to H₂O₂, the XO+SOD result implies that H₂O₂ did not oxidize CMH to CM•. This is confirmed by the experiment shown in Figure 2B, where 1 mM H₂O₂ (blue symbols and lines) caused negligible CMH oxidation compared to control (gray symbols and lines, label “C”) (p=0.144).

Figure 2.

Reaction of CMH with superoxide and hydrogen peroxide

CMH, which is EPR-silent, reacts with superoxide, generated during the oxidation of hypoxanthine (HX) by xanthine oxidase (XO), to yield CM•, a nitroxide that has a 3-line EPR spectrum. The amplitude of the center spectral peak is converted to CM• concentration and plotted against time. Quadruplicate data sets are plotted with different symbols (○△⋄▼) with corresponding least-squares fitted lines (dashed, solid, dotted, short-dashed). The bold solid line is the average of all four least-squares lines in each data set. A. Red symbols and lines (labeled “XO”): Time courses of the reaction in replicate samples containing 1 mM CMH, 0.03 U/mL XO, 0.5 mM HX and 1 mM DTPA in PBS. Green symbols and lines (labeled “XO+SOD”): Same samples with 5 μM superoxide dismutase (SOD) added. Insets show the first and last EPR spectra acquired in a typical run; all spectra are on the same scale. B. Blue symbols and lines (labeled “H₂O₂”): Reaction time course in samples containing 1 mM CMH, 1 mM H₂O₂, and 1 mM DTPA in PBS. Gray symbols and lines (labeled “C”): Control samples containing only CMH and DTPA in PBS.

Superoxide generation in red blood cells from FVB and B6 mice during storage

Superoxide measurements using CMH were performed on RBCs of both mouse strains immediately after collection and processing (on Day 1), and on Days 2, 4, and 8 after storage. The results are shown in Figure 3. Across all days in storage, CMH oxidation occurred more rapidly in RBCs from FVB mice than in RBCs from B6 mice (compare slopes of linear least-squares fits, Figure 3A and 3B; p<2.00×10⁻³), implying a higher rate of superoxide production in FVB RBCs. The oxidation rate is a characteristic of each strain and did not vary significantly with days in storage (Figure 3B; B6: p>0.414; FVB: p>0.802). A higher rate of superoxide production could increase the steady-state level of superoxide, which would be reflected in higher initial [CM•], i.e., larger values of for the y-intercepts of least-squares lines fit to the data. Indeed, the y-intercepts are consistently higher for FVB RBCs (Figure 3C; p ≤ 0.0209).

Figure 3.

Superoxide production in FVB and B6 RBCs after different times in storage

A. Comparing oxidation of CMH to CM• in suspension of RBCs after various times in storage; results for FVB are in red and results for B6 are in blue. Quadruplicate data sets are plotted with different symbols (○△⋄▼) with corresponding least-squares fitted lines (dashed, solid, dotted, short-dashed). The bold solid line is the average of all four least-squares lines in each data set. Gray symbols and lines (labeled “C”) are control measurements in CPDA-1. B. The rates of CMH oxidation in FVB and B6 RBCs on various days of storage. The average rate for each strain (the slope of the averaged least-squares lines) is plotted vs day in storage. Comparing FVB and B6 shows significant difference on all days (p≤ 2.00×10⁻³, *). Within-strain comparisons across all days show no significant difference (FVB: p>0.802; B6: p>0.414). C. Initial concentrations of CM• formed in FVB and B6 RBCs on various days of storage. Average [CM•] for each strain (the y-intercept of the averaged least-squares lines) is plotted vs day in storage. Comparing FVB and B6 shows significant difference on all days (p<0.0209,*). Within-strain comparisons across all days showed little difference (FVB: p>0.600; B6: p>0.268 except for days 1 vs 2, p=0.0139, and 1 vs 4, p=0.0661).

DISCUSSION

Refrigerator storage and human donor characteristics alter metabolic processes that induce RBC injury^6,11. From a translational research perspective, murine strains are useful models for blood storage and transfusion experiments. Several studies have demonstrated that FVB mice produce RBCs with a poor storage phenotype, based on PTR^13,14,20. Current regulatory standards require processed human RBCs meet a 24-hour PTR criterion not less than 75% of the infused RBC mass³. RBCs of FVB mice have essentially normal lifespans in vivo^{13, 20,21}; however, storage reduces PTR by as much as 80%¹³. Further, PTR of stored donor FVB mouse RBCs correlates with metabolic profiles that are consistent with cellular oxidative stress, particularly lipid peroxidation¹³. The reactive oxygen species (ROS) responsible for lipid peroxidation are presumed to originate from O₂ ^{• −}; however, their identity remain unknown. Because of the poor-storing RBC phenotype, the present study was designed to evaluate CMH oxidation to CM• in the RBCs from donor FVB and B6 murine strains on days 1, 2, 4 and 8 of refrigerator storage. This proof-of-concept study used X-band EPR measurements to define the potential utility of using CMH oxidation as a probe of O₂ ^{• −} generation in RBC storage units. We show that CMH does not react with H₂O₂, suggesting specificity for reaction with O₂ ^{• −}. SOD and catalase are the primary sequential systems in RBCs for enzymatic dismutation of O₂ ^{• −} to H₂O₂, followed by conversion of H₂O₂ to molecular oxygen and water. Therefore, CMH is oxidized during a physiological excess of O₂ ^{• −}. In this study, FVB mice produced greater amounts of O₂ ^{• −} at the pre-storage measurement and these levels were maintained throughout the duration of storage. Here, stored FVB mouse RBCs showed significantly enhanced rates of CMH to CM• oxidation starting from day 1 of ex vivo storage compared to RBCs from B6 mice. Interestingly, in this study a 3-fold greater mean rate of CMH to CM• oxidation was consistently observed in FVB versus B6 RBCs on days 1, 2, 4 and 8 of storage. Further, on each day of measurement, the mean [CM•] was consistently ~9 μM higher in RBCs from FVB mice at the time of CMH addition.

A potential limitation of this study is that a leukoreduction step was not used in our blood processing method. Therefore, residual white cells or platelets could have produced their own O₂ ^{• −} and thus contributed toward CMH oxidation independent of RBCs. Nonetheless, RBCs from both murine strains were processed the same way; therefore, extraneous cellular O₂ ^{• −} production would likely be similar. Further, we do not know if the source of O₂ ^{• −} is its excess production or impaired dismutation in FVB mouse RBCs maintained under refrigerator storage, or if O₂ ^{• −} is the primary contributor to the poor storage phenotype. Additional preclinical and clinical experiments are critical to understanding the importance of rapid and accurate measurements of ROS in the assessment of donor RBC quality.

CONCLUSIONS

We demonstrate that a cell-permeant molecular probe (CMH) that is oxidized by O₂ ^{• −} to generate a stable nitroxide (CM•), in conjunction with a compact X-band EPR spectrometer, can quantify superoxide generation in RBCs and differentiate murine strains whose RBCs store well or poorly in the refrigerator. Currently, blood storage is guided only by process control (e.g., collection particulars, temperature of storage, outdate, etc.); transfused RBCs are not subjected to any quality control metrics relevant to PTRs. As such, the described method has the potential to identify RBCs that are likely to store poorly through measurements at time of collection and prior to storage. Applying this method to testing collected and/or stored RBCs has the practical potential to improve clinical outcomes by avoiding transfusion of units that do not meet FDA criteria. This method also has the potential of revealing novel determinants of RBC quality that can inform improvements in blood storage technology.