Twelve specific pathogen-free male Sprague–Dawley (SD) rats were randomly divided into a control group and an HAPC group. For the HAPC rat model, HAPC rats were placed in a hypobaric hypoxic chamber (Guizhou Fenglei Oxygen Chamber, dyc-300; Guizhou Fenglei Oxygen Chamber Co., Ltd., Quizhou, China), simulating an altitude of 5,000 m, with the following conditions: atmospheric pressure, ∼50 kPa; oxygen concentration maintained at 10%; environmental temperature, 20°C–24°C; relative humidity, 40%–70%; 12-h/12-h light/dark cycle. The rats were kept under these conditions for 28 days. The control group rats were raised in specific pathogen–free-grade animal rooms, with the same temperature, humidity, lighting time, and feeding conditions as the HAPC group.

After anesthetizing all SD rats with an intraperitoneal injection of pentobarbital sodium, blood from the abdominal aorta was collected into vacuum heparin tubes. The whole blood in the anticoagulant tubes was used for routine blood tests, blood gas analysis, erythrocyte osmotic fragility (EOF) experiments, and peripheral blood RET detection. For plasma collection, the remaining anticoagulated whole blood was centrifuged at 4°C and 2,500 rpm for 10 min, and the supernatant plasma was aliquoted into 200-μL samples for subsequent enzyme-linked immunosorbent assay (ELISA) detections. For erythrocyte collection, after collecting the plasma, the layer with white blood cells was discarded from the anticoagulant tubes, and erythrocytes in the bottom layer were resuspended in 2 mL of 1× Tris-buffered saline (TBS), centrifuged at 4°C and 2,500 rpm/min for 5 min to remove the supernatant, washed twice, then the washed erythrocytes were kept for later use at 4°C.

After blood collection, the rats were euthanized by decapitation, and bilateral femurs were removed. Bone marrow was harvested by repeatedly flushing the marrow cavity with heparin/phosphate-buffered saline (PBS) using a 20-mL syringe and stored at 4°C in the dark for later examination.

TBS-washed erythrocytes (150 μL) were added to a 1.5-mL Eppendorf tube, and 1.5 mL of 1:5 diluted fixative (2.5% glutaraldehyde in 1 × PBS buffer) was slowly added along the tube wall to resuspend the cells. The cells were left to stand at 4°C for 5 min, then centrifuged at high speed (4°C, 12,000 rpm for 10 min). The supernatant was removed, and the sediment was retained. Then, 2.5% glutaraldehyde fixative was slowly added along the tube wall, and the sample was stored at 4°C.

Five hundred erythrocytes were counted and analyzed per sample. They were classified into two types based on morphology: normal or abnormal erythrocytes. The abnormal erythrocytes were further subdivided into acanthocytes (irregular serrated protrusions on the cell periphery), vesicular erythrocytes (irregular vesicular protrusions on the cell membrane), or severe abnormalities (both acanthocytic and vesicular erythrocyte changes in the same cell).

The washed erythrocytes were mixed with a strong radioimmunoprecipitation assay (RIPA) buffer (50 μL RIPA + 1 μL phenylmethylsulfonyl fluoride + 1 μL phosphatase inhibitor) at a 1:1 ratio for lysis. The mixture was lysed on ice for 30 min, vortexed for 5 s every 10 min, and then centrifuged at 4°C at 12,000 rpm for 10 min. The supernatant was collected, boiled with 2× loading buffer for 10 min, aliquoted, and stored at −80°C as samples for Western blotting.

Additionally, 2 mL of washed erythrocytes were treated with four freeze–thaw cycles with liquid nitrogen, followed by centrifugation at 4°C at 12,000 rpm for 10 min, and the supernatant was collected as the cytoplasmic extract of erythrocytes.

A total of 10 μL of TBS-washed erythrocytes were added to 990 μL of 1 × PBS. Subsequently, 10 μL of this diluted erythrocyte solution was transferred to a fresh tube containing 990 μL of 1 × PBS, gently mixed, and centrifuged at 300 × g for 5 min. The supernatant was discarded, and the cells were resuspended in 100 μL of annexin V binding buffer (1×). To this suspension, 2.5 μL of annexin V–fluorescein isothiocyanate (FITC) reagent and 2.5 μL of propidium iodine reagent (50 μg/mL) were added. The mixture was gently vortexed and incubated at room temperature in the dark for 20 min. After incubation, 900 μL of annexin V binding buffer (1×) was added, and the sample was mixed and filtered for analysis.

For the analysis of the proportion of erythrocytes in early apoptosis, the specific gating strategy is shown in Supplementary Figures S1H–J. FITC+/PerCP/Cy5.5− represents erythrocytes in early apoptosis, and FITC+/PerCP/Cy5.5+ represents erythrocytes in late apoptosis or necrosis.

One milliliter of BD Retic-Count™ reagent and 1 mL of PBS were added to 5 μL of whole blood and incubated at room temperature in the dark for 30 min. Then, 100 μL of the stained solution was added to 1 mL of PBS, and the sample was filtered to be ready for analysis. The specific gating strategy is shown in Supplementary Figures 1F–G, with Flow Jo v10.8.1 (https://www.flowjo.com/solutions/flowjo/downloads/previous-versions) used for analysis and the fluorescence intensity of RETs used as the indicator.

Bone marrow RETs were labeled with CD71-PE, leukocytes with CD45-FITC, and DNA with the nucleic acid dye DRAQ5 to establish a flow cytometry–based detection scheme for marrow micronuclei. Bone marrow cells were centrifuged at 300 × g for 5 min, and the supernatant was discarded. The cells were resuspended in 50 μL PBS, and 20 μL of the resuspension was added to 100 μL of the staining system (containing 2 μg CD71-PE, 2 μg CD45-FITC, and 2 μL FBS). After mixing well, the sample was left to stain in the dark at 4°C for 30 min. Following surface antigen staining, the bone marrow cells were transferred to a centrifuge tube containing 5 mL PBS for washing, and the tube was centrifuged at 300 × g for 5 min before the supernatant was discarded to remove free antibodies. The cell pellet was then resuspended in 1 mL of DRAQ5 staining solution (20 μmol/L), mixed well, and stained in the dark at 37°C for 30 min. After a further wash with 5 mL PBS and centrifugation (300 ×g, 5 min), the supernatant was discarded, and the cells were resuspended in 1,250 μL PBS, filtered, and transferred to a flow cytometry tube. Samples were stored in the dark at 4°C and analyzed by flow cytometry within 4 h (Zeng et al., 2020).

The specific gating strategy is shown in Supplementary Figures 1A–E, with Flow Jo v10.8.1 software used for analysis and the proportion of RETs used as the indicator. DRAQ5+/FITC + indicated RETs with micronuclei, DRAQ5−/FITC− indicated mature erythrocytes, and DRAQ5−/FITC + indicated RETs. The formula for calculating the RETs proportion (%) is (Q2 + Q3)/(Q1 + Q2 + Q3 + Q4) × 100, where Q1 represents red blood cells (RBCs) with micronuclei, Q2 represents RETs with micronuclei, Q3 represents RETs, and Q4 represents RBCs. The product numbers of all flow cytometry antibodies are presented in Supplementary Table S1.

Plasma-free hemoglobin, plasma 5′-nucleotidase (CD73), and extracellular adenosine were detected using ELISA with plasma samples, whereas S1P, intracellular adenosine, and 2,3-BPG were detected using erythrocyte cytoplasmic extraction. All ELISA kits were provided by Nanjing Camilo Biological Co., Ltd. (Nanjing, China), and the procedures were performed according to the manufacturer’s instructions. The product numbers of all ELISA kits are presented in Supplementary Table S1.

Adequate amounts of Parpart NaCl working solution were prepared at a Parpart NaCl solution/ddH₂O ratio of 1:9. Heparin-anticoagulated blood was prepared for later use: 5-mL Eppendorf tubes were obtained and numbered sequentially (1–17), and Parpart NaCl working solution and ddH₂O were added according to the Parpart NaCl dilution table, with a total volume of 4 mL (see Supplementary Table 2 for the Parpart NaCl dilution table). Next, 40 μL of heparin-anticoagulated blood was added to the above-described diluted Parpart NaCl solution, mixed immediately, and left at room temperature for 30 min. The tubes were then centrifuged at 300 × g for 5 min, and the supernatant was collected. The optical density at 540 nm was measured using a microplate reader, with the supernatant from the tube containing 8.5 g/L NaCl (tube 3) used as a blank and the tube containing 1 g/L NaCl (tube 17) used as a reference for 100% hemolysis, to calculate the percentage of hemolysis.

Protein extracts were separated using 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis and then transferred to a transfer membrane (Millipore, Burlington, MA, United States). After blocking with 5% bovine serum albumin for 1 h, the membrane was incubated with primary antibodies overnight at 4°C. The antibodies used were as follows: anti-adenosine A2b receptor [ADORA2B]/ADORA2B antibody (ab229671, 1:500; Abcam, Cambridge, United States), sphingosine kinase 1 (SPHK1) polyclonal antibody (10670-1-AP, 1:500; Proteintech, Wuhan, China), phospho-SPHK1 (Ser225) rabbit polyclonal antibody (19561-1-AP, 1:500; Proteintech), ENT1 polyclonal antibody (PA5-75507, 1:500; Invitrogen, Waltham, MA, United States), bisphosphoglycerate mutase (BPGM) polyclonal antibody (17173-1-AP, 1:500; Proteintech), anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH) antibody (ab181602, 1:10,000; Abcam), and beta-actin rabbit polyclonal antibody (20536-1-AP, 1:1,000; Proteintech). The membrane was then incubated with the secondary antibody, i.e., horseradish peroxidase–conjugated affinity-purified goat anti-rabbit IgG (H + L) (SA00001-2, 1:5,000; Proteintech) at room temperature for 1 h. Finally, protein bands were visualized using SuperKine ultra-sensitive ECL substrate (BMU102-CN/BMU103-CN; Abbkine, Wuhan, China), and quantified using ImageJ software (NIH, Bethesda, MD, United States).

Statistical analyses were performed using SPSS v25.0 (IBM Corp., Armonk, NY, United States). The normality of data distribution was assessed using the Kolmogorov–Smirnov and Shapiro–Wilk tests, and the homogeneity of variances was evaluated using the F test. Data that were normally distributed and had homogeneity of variance were analyzed using two-tailed independent-samples t-tests; data that were normally distributed but had heterogeneity of variance were analyzed using a tʹ test (data for the t and tʹ tests are presented as mean ± standard deviation). Data that were not normally distributed were analyzed using the Mann–Whitney U test (data are presented as P50 [P25, P75]). GraphPad Prism v9.5.0 (GraphPad Inc., La Jolla, CA, United States) was used for plotting graphs. p < 0.05 was considered statistically significant.

After placing the SD rats in a hypobaric hypoxic chamber for 28 days, blood was drawn from the abdominal aorta for routine blood tests and blood gas analyses. Rats in the HAPC group, which were exposed to chronic hypoxia, had significantly higher hemoglobin content, erythrocyte counts, and hematocrit values than rats in the control group (Table 1). The average hemoglobin content in the control group was 140.5 g/L, whereas the lowest in the HAPC group was 211 g/L, with an average of 237 g/L. The median erythrocyte count in the control group was 5.83 × 10¹²/L, compared with 9.41 × 10¹²/L in the HAPC group. The average hematocrit in the control group was 36.45% compared to 59.78% in the HAPC group. Additionally, arterial oxygen saturation and partial pressure of oxygen were significantly lower in the HAPC group than in the control group (p < 0.001). According to the Qinghai Criteria for chronic high-altitude disease, the HAPC model was thus successfully established in all rats.

A significant increase in peripheral blood erythrocytes was observed in the HAPC group, consistent with the mechanism behind HAPC, which is believed to involve an imbalance in erythrocyte apoptosis/proliferation (Wu and Du, 2013). Therefore, we further investigated the content of RETs in bone marrow and peripheral blood as well as the apoptosis of peripheral blood erythrocytes.

Short-term hypoxia caused by a rapid ascent into high altitudes can increase the incidence of erythrocyte morphological abnormalities (Gu et al., 2017). Therefore, we used electron microscopy to investigate changes in morphology and the number of abnormal-shaped erythrocytes in chronic HAPC rats (Figures 3A, B). The number of normal erythrocytes was significantly higher in the control than the HAPC group (p < 0.001), with abnormal-shaped erythrocytes primarily being echinocytes and acanthocytes. The number of abnormal-shaped erythrocytes was significantly higher in the HAPC than the control group (vesicular, p < 0.001; acanthocytes, p = 0.019; severe abnormalities, p = 0.003) (Table 2).

In rats with HAPC, both the number and morphology of erythrocytes underwent significant changes. Therefore, we further explored whether erythrocyte function was changed, with EOF and hemolysis being the main focus of our investigation. EOF is a measurement of the resistance of erythrocytes to hypotonic solutions and hemolysis, which primarily depends on the surface area/volume ratio in erythrocytes. Clinically, EOF can be applied to screening and evaluating certain hemolytic diseases, such as thalassemia, hereditary spherocytosis, and glucose-6-phosphate dehydrogenase deficiency (Emilse et al., 2018; Tatu and Sweatman, 2018). Routine blood tests showed that the average erythrocyte volume in HAPC rats had increased, leading to a relative decrease in the surface area/volume ratio of erythrocytes, which may increase osmotic fragility. Our subsequent EOF experiments further corroborated this theory. The 50% hemolysis rate for erythrocytes was 4.0–4.5 g/L NaCl in the control group and 4.5–5.0 g/L NaCl in the hypoxia group (Table 3). Long-term hypoxia led to a decrease in the resistance of erythrocytes to hypotonic solutions, increased osmotic fragility, reduced deformability, and decreased oxygen transport capacity in the HAPC group.

The increased EOF and reduced resistance to hemolysis in HAPC rats led us to investigate hemolysis further. Previous evidence indicates that, under hypoxic stimulation, most extracellular ATPs are derived from the lysis of erythrocytes, and increased ATP levels are closely related to plasma-free hemoglobin, which primarily reflects intravascular hemolysis (Ellsworth et al., 1995; Sikora et al., 2014; Grygorczyk and Orlov, 2017). ELISA showed that the plasma free hemoglobin concentration was significantly higher in the HAPC group (22.87 ± 3.59 μg/mL) than the control group (4.09 ± 0.61 μg/mL, p < 0.001) (Figure 3C).

Hypoxia can affect erythrocyte metabolism, such as adapting erythrocyte-mediated purinergic signaling to hypoxia (Sun et al., 2017), changing erythrocyte glycolysis (Rogers et al., 2009), and enhancing the oxidation of erythrocyte membrane lipids under hypoxia (Revin et al., 2019). Therefore, we continued to investigate the related metabolic changes in erythrocytes in HAPC.

In people who rapidly ascend to high altitudes, erythrocyte S1P level increases steeply. Moreover, a mouse model demonstrated elevated plasma CD73 and adenosine levels. The increased adenosine binds to erythrocyte ADORA2B, which further activates SPHK1-induced S1P production, promoting glycolysis to produce a large amount of 2,3-BPG, thereby inducing oxygen release to protect tissues from hypoxia (Liu et al., 2016; Sun et al., 2016). Moreover, in sickle cell disease, the binding of adenosine to ADORA2B activates SPHK1, thereby increasing S1P production (Sun et al., 2015). Therefore, we hypothesized that the lipid metabolism-related pathways in erythrocytes are also activated to alleviate tissue hypoxia under chronic hypoxic conditions in HAPC. A previous study by our group revealed that in humans and mice under chronic hypoxia, intracellular S1P in erythrocytes is continuously elevated (Qile et al., 2023). In this study, our results showed that plasma CD73 levels in the control group were significantly lower than in the HAPC group (p = 0.004) (Figure 4H). Hypoxia induces an increase in CD73, promoting the conversion of AMP to adenosine, which leads to elevated levels of free adenosine. The concentration of extracellular adenosine in erythrocytes was significantly higher in the HAPC group than the control group (p < 0.004) (Figure 4I). Adenosine binds to ADORA2B and activates the downstream SPHK1, and the phosphorylation of SPHK1 further promotes the expression of S1P. Our results revealed that ADORA2B, phosphorylated SPHK1, and S1P levels were all increased (p < 0.001, p = 0.086, p = 0.004, and p = 0.01, respectively) (Figures 4B, C, D, K), promoting the production of 2,3-BPG by elevating the expression of GAPDH in the glycolytic pathway. The expression levels of GAPDH and 2,3-BPG were significantly higher in the HAPC group than the control group (p < 0.001, p = 0.016) (Figure 4G, L).

Recently, Chen et al. (2023) discovered that erythrocyte equilibrative nucleoside transporter 1 (eENT1) can promote the uptake of extracellular adenosine into erythrocytes, which enhances purine metabolism to activate BPGM. This process leads to increased production of 2,3-BPG, alleviating the renal oxidative stress and damage induced by severe hypoxia. In our study, we also found that under long-term hypoxia, HAPC rats exhibit higher levels of ENT1 expression and intracellular adenosine in erythrocytes than did those in the control group (p = 0.001, p < 0.001) (Figures 4E, J). Additionally, we observed a significant increase in the expression of BPGM, the key enzyme in the Rapoport–Luebering shunt, a metabolic bypass of the glycolytic pathway, in the HAPC group (p < 0.004) (Figure 4F). This indicates that activation of this bypass induces a further increase in 2,3-BPG. Therefore, we believe that in HAPC rats, both the CD73/adenosine/S1P/2,3-BPG and eENT1/adenosine/BPGM/2,3-BPG metabolic pathways are activated in erythrocytes (Figure 4A). The activation of these pathways represents a metabolic adaptation of erythrocytes to hypoxia, ultimately leading to the production of 2,3-BPG, which, as a specific allosteric regulator in erythrocytes, reduces the affinity of hemoglobin for oxygen and promotes the release of oxygen.