De-identified human RBCs, procured from Rhode Island Blood Center by the study sponsor Hemanext, were conventionally acquired into CP2D solution after written informed consent. The units were then processed (including leukofiltration) and banked in conventional RBC storage bags using AS-3 additive solution (Haemonetics, Braintree MA). Each bag contained about 300 mL of RBCs. Two RBC units, selected based only on mutual ABO- and Rh-compatibility, were pooled, then split into three “subunits” (∼200 mL each) intended to become conventional (Control), hypoxic/Hemanext (HN-Std), and hypoxic/isocapnic (HN + CO₂) RBCs. HN + CO₂ RBC preparation was intended to preserve CO₂ at levels matching those in Control RBCs during 6 weeks of refrigerated (4 °C) storage. In total, n = 8 pooled RBC superunits (from 16 standard units) from healthy adult human donors were used. No data were collected on donor sex, age or other demographic variables for the purposes of the study. In the case of five of the superunits, initial production of the subunits was performed in Durham, NC, and in the remaining three superunits, production was performed in Lexington, MA followed by overnight shipping to Durham, NC. In both cases, collected whole blood was processed within 24 h.

Hemanext ONE® hypoxic blood storage bags were provided by Hemanext Inc. (Lexington, MA). Standard Hemanext oxygen reduction was performed by following the Hemanext ONE instructions and using a Hemanext oxygen reduction bag (ORB). Meanwhile, the CO₂-augmented (hereafter also referred to as CO₂-preserved, HN + CO₂) subunit was prepared by continuously supplying 5% CO₂/balance nitrogen) gas mix in the space between RBC-containing bag and outer gas barrier bag and was otherwise handled identically to the standard Hemanext subunit. In order to achieve the target SO₂ of 10% or lower, an estimate was made of the deoxygenation rate by calculating the rate of change in SO₂ at an early time point (5–10 min) and a later time point (60–90 min) of the Hemanext deoxygenation process. The RBC concentrates from the ORBs were then transferred into respective (HN-Std and HN + CO₂) Hemanext storage bags (PVC bag enclosed by gas barrier bag with sachet of O₂/CO₂ sorbent placed between the inner and outer bag), and placed in a 4 °C cold room. The Control subunit was exposed to no gas, stored in a conventional PVC RBC bag exposed to room air, and otherwise handled identically.

On each study day [on Days 3 ± 1 (one of the 7 assay sets was run on both day 2 and day 4; one of the seven on Day 2 only; and the other five on Day 3), 7–8 (2 of 8 on Day 8), Day 14, Day 21 and 42–45 (one of 8 on day 45) after acquisition] a small aliquot was removed anaerobically from each subunit bag. For simplicity, we refer hereafter to these timepoint ranges as Days 3, 7, 14, 21, and 42. (The averaged value of assay results from Days 2 and 4 (both from the same source) was imputed as a Day 3 value in one case, and a Day 45 value was imputed as Day 42). Other than brief, airtight sampling at these intervals, the units were stored conventionally at 4 °C throughout the 42-day maximal storage duration.

At each study timepoint, subunit RBC aliquots (12–17 mL) were withdrawn from the storage bag after gentle resuspension of the RBCs. Subaliquots were set aside before (“subunit suspension”) and after separation by centrifugation at 2700 RPM at 4 °C for 3 min (yielding a “subunit supernatant’ and a ‘pellet’). The RBC suspension samples were immediately transported to a cooximeter/blood gas analyzer (GEM Premier 5000, Werfen, Spain) operated by Duke University Hospital Clinical Laboratories (see below). Remainder subunit suspension samples were lysed hypotonically as described (Kirby et al., 2021), and assayed for hemoglobin (Hb) and ATP content (Kirby et al., 2021). RBCs were resuspended in Krebs-Henseleit buffer at 1% Hct and kept on ice in preparation for gas exposure (“tonometry”). This was repeated for each sample from each subunit on each day of the serial assays.

After completion of co-oximetry, the RBC suspension was transported in a syringe on ice for processing utilizing a Hemox Analyzer (TCS Scientific). Following manufacturer guidelines, 20 µL of bovine serum albumin 20% (BSA-20) and 10 µL of anti-foaming agent (TCS) were added to a microtube containing 5.0 mL of Hemox buffer solution (composition: 30 mM N-Tris (hydroxymethyl)methyl-2-aminoethanesulfonic acid (TES), 130 mM NaCl, and 5 mM KCl; pH 7.40 ± 0.01). Finally, 50 µL of RCC (red cell concentrate) from each sample was pipetted into the corresponding microtube, and capped microtubes were refrigerated at 4 °C until ODC analysis. This process was repeated for aliquoted samples from each subunit (Control, HN-Std, and HN + CO₂).

Using the Hemox Analyzer, 3 mL of the blood-buffer solution was added to the sampling cuvette and warmed until the temperature was steady at 37 °C. While incubating, the samples were fully oxygenated using medical air (21% O₂) until the Hemox Analyzer oxygen sensor demonstrated a steady partial pressure of oxygen (PO₂). Once the blood-buffer suspension was fully oxygenated and stable at 37 °C, the PO₂ was manually set to 148 mmHg to represent the calculated (predicted) partial pressure of O₂ (PO₂) in room air given the atmospheric pressure and elevation of our laboratory site in Durham, NC. The sample was then progressively de-oxygenated by switching the gas supply from air to nitrogen (N₂) gas. Utilizing the OEC3 Software (TSC Scientific), real-time data on PO₂ and SO₂ (percent hemoglobin saturation by O_2, measured directly by dual wavelength spectrophotometry) during the deoxygenation process was obtained. Once each sample was adequately deoxygenated (PO₂ < 1 mmHg), the resulting oxygen-dissociation curve (ODC) was displayed and stored, and the P₅₀ was calculated by the OEC3 software. Hemox utilizes spectrophotometry to compare absorption of oxyhemoglobin and deoxyhemoglobin, with ratios calculated by OEC3 software to generate ODCs. The sample cuvette was then washed with de-ionized water three times before loading the next RBC-buffer sample for processing. This analytical method helps ascribe differences in ODCs to differences in intracellular factors, because a constant temperature and pH are maintained throughout the oxygenation and deoxygenation process.

In order to determine O₂-sensitive RBC ATP export, the RBC samples (in Krebs buffer, pH 7.40) at 1% hematocrit were subaliquotted and placed in a rotating-bulb tonometer at 37 °C. To establish hypoxic or normoxic conditions in the rotating tonometer, a gas blender was used to establish “hypoxia” (94% N₂, 1% O₂, 5% CO₂) and “normoxia” (74% N₂, 21% O₂, 5% CO₂) conditions at 200 mL/min gas flow (in total to two parallel tonometers) over thin-film RBC suspensions (not “bubbled through”). 1.5 mL of each sample was added via a port on the tonometer before re-sealing, and was then exposed gently to gas for 8 min, sufficient for reaching steady-state changes in oxygenation, as previously described (Kirby et al., 2012). Sample exposures were performed in duplicate.

Following gas exposure, the post-tonometry sample suspension was centrifuged at 2,700 RPM at 4 °C for 3 min. The resulting supernatant was removed for exported ATP and free Hb assays. In the event that Hb and ATP assays could not be performed on the same day as the gas exposure, the corresponding samples were stored at −20 °C. We have previously determined that the values for both analytes (which are acellular) do not change with 1 or 2 freeze-thaw cycles. Subunit suspension samples (10 μL) were diluted in two steps (total 1000-fold dilution) into deionized water, then vortexed to produce a lysate for assays of total intracellular Hb and ATP. The Hb value is used as the “denominator” in RBC lysis calculations post tonometry.

Free (supernatant) and intracellular Hb values for each sample were obtained using a spectroscopic method via a microplate reader as described (Zhu et al., 2011). For ATP assays, the luciferin-luciferase assay was used, and values were extrapolated against an ATP standard curve generated daily in PBS (as described) (Kirby et al., 2021).

Red blood cells (10 µL) were mixed with 200 µL methanol and 10 µL 10 μg/mL ¹³C₃-2,3-BPG. The resulting supernatant samples were subjected to complete dryness under nitrogen, then resuspended into 80 µL 80% methanol solution before instrument injection. Samples were analyzed with the Sciex QTrap 6500+ system (Framingham, MA) with Waters Acquity I-class plus UPLC. Software Analyst 1.7.3 was used for data acquisition. The LC separation was performed on a Waters Acquity UPLC Phenyl-Hexyl column (2.1 × 100 mm, 1.7 μm) with mobile phase A (20 mM ammonium acetate with 0.02% ammonium hydroxide in water) and mobile phase B (acetonitrile). The flow rate was 0.45 mL/min. The linear gradient was as follows: 0–1 min, 0%B; 1–2.9 min, 50%B; 3–4.5 min, 95%B; 4.6–5.1 min, 0%B. The autosampler was set at 10 °C and the column was kept at 40 °C. The injection volume was 5 μL. Mass spectra were acquired under negative electrospray ionization (ESI) with the ion spray voltage of −4500 V. The source temperature was 450 °C. The curtain gas, ion source gas 1, and ion source gas 2 were 35, 50, and 70 psi, respectively. Multiple reaction monitoring (MRM) was used for quantitation as shown below: 2,3-BPG (target): m/z 265.1 → m/z 79.0; 13C₃-2,3-BPG (internal standard): m/z 268.1 → m/z 79.0.

RBC samples were transported immediately on ice to the co-oximetry laboratory. Utilizing a GEM Premier 5000 arterial blood gas analyzer (Werfen), ∼150 µL of sample was aspirated from each syringe for analysis before re-capping the syringe and returning it to ice. Co-oximetry results generated by the GEM Premier 5000 analyzer included concentration of total hemoglobin (tHb), oxyhemoglobin (oxyHb), carbonmonoxyhemoglobin (COHb), methemoglobin (metHb), and oxyhemoglobin saturation (SO₂). Additional results obtained include pH, PCO₂, and PO₂. All were measured at 37 °C.

All values are reported as individual data and the mean ± SEM unless otherwise noted. In the case of missing values, there was no imputation. Testing across assay exposure (performed in normoxia or hypoxia), RBC “treatment” status, and over time was performed using two-way repeated measures ANOVA mixed-effects model with appropriate post hoc sample testing corrected for multiplicity (Tukey’s). Statistics were performed using GraphPad Prism 10. p < 0.05 (*) was considered statistically significant unless otherwise noted. ** indicates p < 0.01; *** indicates p < 0.001; ****p < 0.0001.

Blood gas and cooximetric assays of 4 parent superunit pools (3 subsets each) of standard HN-Std and HN + CO₂ RBCs confirmed that their actual SO₂ values were near 10% (Supplementary Figure S1), in line with the projected SO₂ target at 10% based on a rate calculation made from SO₂ change during the process. Figures 1A,B, shows that while PO₂ levels in the Control (conventional) RBC subunit gradually increased over storage time, the PO₂ in both HN-Std and HN + CO₂ RBC subunits were initially and remained much lower. PO₂ differences were significant from Day 3 to Day 42 (p values ranged from <0.05 to <0.0001). The SO₂ data were similar to the PO₂ data: significantly lower in the HN-Std and HN + CO₂ RBC groups vs. the Control group (p < 0.0001) at every sampling time point during RBC storage (Figures 1C,D). Notably, SO2 values in the HN-Std and HN + CO₂ RBCs were near 10% on Day 3, indicating stability of the initial deoxygenation effect. When comparing the two (HN + CO₂ vs. HN-Std) Hemanext RBC subunits, there were generally no statistically significant differences on any sampling day for either PO₂ or SO₂. PCO₂ levels in the CO₂-preserved RBC subunit exceeded those in the standard Hemanext (p < 0.0001) RBCs. PCO₂ values in the HN + CO₂ subunits were modestly but significantly lower than in the Control subunits on Day 3 (p < 0.01) and Day 7 (p < 0.05), then matched those in the conventional subunit across all sampling time points from Day 14 to Day 42 of RBC storage (not significant (n.s.); Figures 1E,F). Hypoxic RBC preparation and storage, with or without CO₂ augmentation, did not generate substantial excess methemoglobin (oxidized Hb; Supplementary Figure S2). pH values (not shown) in nearly all samples at all time points were reported as “< 6.93,” i.e., outside of the GEM Premier 5000 analyzer’s clinical reportable range for pH (6.93–7.72). The only exceptions were that in <5% of cases of PO₂ measurements, instrument errors were reported; these error reports were evenly distributed with respect to RBC condition and storage time.

We observed changes in the ODCs (Supplementary Figure S3) and P₅₀ (Figure 2) throughout the storage period that varied across the three RBC storage conditions. ODCs were shifted rightward (Supplementary Figure S3) and the P₅₀ (Figure 2) was elevated (i.e., O₂ affinity was decreased) in the HN-Std group compared to the HN + CO₂ and Control groups except at Day 42. By mixed-effects analysis (Figure 2), storage condition-wise differences in P₅₀ were statistically significant overall, and by post hoc testing at Days 7 and 14. By Day 42 there was no difference in mean P₅₀ (p = 0.876). The higher mean P₅₀ in the HN-Std RBCs over at least the first 3 weeks of storage implies a higher capacity for oxygen dissociation (offloading). These findings demonstrate that during standard hypoxic RBC storage (HN-Std), a physiologic P₅₀ is approached, but this is not the case in CO₂-preserved/hypoxic (HN + CO₂) or Control RBC subunits (in both of which P₅₀ is below that in normal healthy blood).

Intra-RBC ATP values declined between 21 and 42 days of storage (Figure 3). Standard hypoxic (HN-Std) storage tended to depress RBC ATP from Day 3 to Day 7, then to increase RBC ATP after Day 7, peaking at Day 21. Conversely, in HN + CO₂ RBCs, intra-RBC ATP was stable from Day 3 to Day 7, then declined progressively through Day 42 (Figure 3). The changes in the intra-RBC ATP from Day 7 to Day 21 (a storage period during which RBCs are frequently transfused) therefore moved in opposite directions when comparing HN-Std (rising RBC ATP) and HN + CO₂ (falling RBC ATP) RBC subunits (Figure 3). By Day 42, ATP in all 3 RBC types was markedly depressed, despite HN-Std still showing significant higher ATP level compared to the HN + CO₂ subunit. Notably, we assayed for intra-RBC ATP in RBCs just after they had been exposed to normoxic (Figures 3A,B) or hypoxic (Figures 3C,D) gas for the ATP-export studies. The fact that RBC ATP values matched closely at each timepoint and under all conditions provides reassurance that the tonometry exposures did not substantially alter intra-RBC ATP by driving a large fraction into the extracellular space (note that extracellular ATP concentrations were >1000-fold lower than intra-RBC ATP).

2,3-BPG levels on Days 3 and 7 of RBC storage were greater (p < 0.05) in HN-Std RBCs than in HN + CO₂ or Control RBCs. By Day 21, 2,3-BPG was markedly depressed in all RBC types (Figures 3E,F). 2,3-BPG levels were stable in, and did not differ between, HN + CO₂ and Control RBCs during storage.

Changes in RBC ATP export as a function of time, HN ± CO₂ storage conditions, and normoxic or hypoxic assay conditions are displayed in Figure 4. Conventional RBC storage progressively depressed the ability of RBCs to export ATP in both normoxia and hypoxia (Figure 4), as we previously reported (Zhu et al., 2011). The effect of time was significant overall across the RBC preparation/storage conditions (p < 0.05 for hypoxia and p = 0.0598 for normoxia). The trend toward slightly greater RBC ATP export by HN + CO₂ (CO₂-augmented Hemanext) RBCs, and the downward trend in ATP export early during storage from HN-Std RBCs, reached statistical significance only on Day 7 in both normoxia (Figure 4A,B) and hypoxia (Figure 4C,D). In Control RBCs, intra-RBC ATP (which declined over the 6 weeks of storage) correlated inversely (p < 0.0001) with SO₂, which rose over the 6 weeks of storage (Supplementary Figures S4A,B). By contrast, the decline in RBC ATP export did not correlate significantly with SO₂ over time in Control RBCs (Supplementary Figures S4C,D). ATP export was significantly greater when assayed in hypoxia than in normoxia for conventionally stored RBCs, as previously reported (Zhu et al., 2011). ATP export was also significantly greater when assayed in hypoxia than in normoxia for HN-Std RBCs, but the difference did not reach statistical significance for HN + CO₂ RBCs (Supplementary Figure S5).

Absolute Hb concentrations and calculated hemolysis values in the supernatants of all three actual RBC storage types, assayed serially over RBC storage time (Supplementary Figure S6) under varying O₂ and CO₂ conditions, did not differ significantly overall by mixed-effects analysis, nor were there significant differences by post hoc Tukey’s t-test at any timepoint.

RBC lysis was measured at the end of tonometric exposure to normoxic or hypoxic gas and expressed as a function of time and the O₂/CO₂ conditions during RBC storage (Figure 5). The effect of time was significant or nearly so (p < 0.05 for normoxia, and p = 0.0549 for hypoxia), and the changes were biphasic (U-shaped curve) under all three conditions, with an initial decline (visible for all 3 condition sets on Days 7 and 14). However, except on Day 3 for normoxia there were no significant differences at any timepoint by post hoc Tukey’s t-test. A fraction of the extracellular ATP we measured in tonometers could have resulted, at least in part, from this RBC lysis. When exported ATP in normoxia was expressed as its ratio to % RBC lysis, values for this ratio followed an inverted U-shaped curve and were greatest around Day 14 for HN-Std and HN + CO₂ RBCs, but fell monotonically for Control RBCs (Supplementary Figure S7). These trends suggest that the extent of hemolysis is only one determinant of extracellular ATP.