Proteostasis and metabolic dysfunction characterize a subset of storage-induced senescent erythrocytes targeted for posttransfusion clearance

Abstract

Although refrigerated storage slows the metabolism of volunteer donor RBCs, which is essential in transfusion medicine, cellular aging still occurs throughout this in vitro process. Storage-induced microerythrocytes (SMEs) are morphologically altered senescent RBCs that accumulate during storage and are cleared from circulation following transfusion. However, the molecular and cellular alterations that trigger clearance of this RBC subset remain to be identified. Using a staining protocol that sorts long-stored SMEs (i.e., CFSEhi) and morphologically normal RBCs (CFSElo), these in vitro aged cells were characterized. Metabolomics analysis identified depletion of energy, lipid-repair, and antioxidant metabolites in CFSEhi RBCs. By redox proteomics, irreversible protein oxidation primarily affected CFSEhi RBCs. By proteomics, 96 proteins, mostly in the proteostasis family, had relocated to CFSEhi RBC membranes. CFSEhi RBCs exhibited decreased proteasome activity and deformability; increased phosphatidylserine exposure, osmotic fragility, and endothelial cell adherence; and were cleared from the circulation during human spleen perfusion ex vivo. Conversely, molecular, cellular, and circulatory properties of long-stored CFSElo RBCs resembled those of short-stored RBCs. CFSEhi RBCs are morphologically and metabolically altered, have irreversibly oxidized and membrane-relocated proteins, and exhibit decreased proteasome activity. In vitro aging during storage selectively alters metabolism and proteostasis in these storage-induced senescent RBCs targeted for clearance.

Keywords: Cell biology; Cellular senescence; Hematology; Proteomics; Ubiquitin-proteosome system.

Figure 1. CFSE^hi RBCs are storage-induced micro-erythrocytes that can be sorted by flow cytometry.

(A) Weekly quantification of storage-induced microerythrocytes (SMEs) (white squares) and CFSE^hi RBCs (green circles) in RBC concentrates stored in SAGM solution at 4°C for 42 days (mean ± SEM of 8 RBC concentrates). (B) Proportion of SMEs in CFSE-stained, long-stored, unsorted (LS unsorted) and flow-sorted CFSE^lo (LS CFSE^lo) and CFSE^hi (LS CFSE^hi) RBC subsets. Data are represented as individual points with mean ± SEM of 12 RBC concentrates. (C) Representative scanning electron microscopy images showing typical RBC morphology of CFSE-stained RBCs. Scale bar: 2 μm. In B, *P < 0.05, ****P < 0.0001 by a Friedman’s 1-way ANOVA followed by Dunn’s multiple comparison test (n = 12).

Figure 2. Metabolomics identifies subset-specific alterations in the metabolism of long-stored CFSE^hi RBCs during aging in vitro.

(A) Principal component analysis (PCA) of metabolomics data on flow-sorted short-stored CFSE^lo (SS-CFSE^lo, in red), long-stored CFSE^lo (LS-CFSE^lo, in light green), and long-stored CFSE^hi (LS-CFSE^hi, in dark green) RBCs. (B) Hierarchical clustering analysis of the 30 metabolites whose levels vary among the 3 groups by 1-way ANOVA followed by Tukey’s multiple comparison test. (C) Schematic distribution of the 28 metabolites whose levels vary significantly when comparing long-stored CFSE^lo and CFSE^hi RBC subsets. (D–F) Overview of glycolysis (D), lipid repair (E), and glutathione (F) pathways, highlighting key metabolites whose levels vary among the 3 groups. In C–F, the metabolites that show significant increases (red font) and decreases (blue font) in long-stored CFSE^hi RBCs (vs. long-stored CFSE^lo RBCs) are shown. Arrows represent a single metabolic step in the pathways, and dotted lines represent multiple steps. P values of 1-way ANOVA followed by a FDR correction are indicated under each graph, and θ represents a significant difference found by a positive Tukey’s post hoc test between groups. Abbreviation definitions are detailed in Supplemental data (Supplemental Methods, Metabolomics and redox-proteomics).

Figure 3. Redox proteomics identifies a subset-specific capacity to resist storage-induced oxidative stress during aging in vitro.

(A) PCA of reversible (“oxidation”) and irreversible (“Cys to DHA”) protein oxidation data obtained from flow-sorted short-stored CFSE^lo (SS-CFSE^lo, in red), long-stored CFSE^lo (LS-CFSE^lo, light green), and long-stored CFSE^hi (LS-CFSE^hi, in dark green) RBCs. (B) Hierarchical clustering analysis of the 79 proteins (of a total of 659 proteins analyzed, representing 11.7%) with significant reversible oxidation across the 3 groups by 1-way ANOVA followed by Tukey’s multiple comparison test. (C) Hierarchical clustering analysis of the 6 proteins (of a total of 176 proteins analyzed, representing 3.6%) with significant irreversible oxidation across the 3 groups by 1-way ANOVA followed by Tukey’s multiple comparison test.

Figure 4. Proteomics identifies subset-specific membrane relocation of specific proteins in long-stored CFSE^hi RBCs, notably affecting proteins in the proteostasis family.

(A) PCA of proteomics data obtained from membrane preparations isolated from flow-sorted short-stored CFSE^lo (SS-CFSE^lo, in red), long-stored CFSE^lo (LS-CFSE^lo, light green), and long-stored CFSE^hi (LS-CFSE^hi, dark green) RBCs. (B) Hierarchical clustering analysis of the 96 proteins that significantly differ between the 3 groups following Pearson’s clusterization and a FDR correction without data imputation (proteins detected in at least 70% of samples were considered). A z-score scale calculated from copy number/cell shows each protein level, while lack of detection is represented by a gray area. (C) Interaction network analysis for proteins increased (full filled circle) and decreased (white, transparent, inner circle) in membrane preparations of long-stored CFSE^hi RBCs, as compared with long-stored CFSE^lo RBCs. Gray lines represent physical interactions between proteins. This network was realized using Cytoscape StringApp 3.9. Colored dotted lines represent the main protein families identified (relative proportion within the 96 significant proteins), and colored circles represent functional groups within a family.

Figure 5. Proteasomal degradation capacity decreases during storage, especially in CFSE^hi RBCs.

(A) Weekly quantification, at the whole-population level, of the chymotrypsin-like (blue curve), trypsin-like (green curve), and caspase-like (red curve) proteasome activities during storage of RBC concentrates in SAGM solution for 42 days (mean ± SEM of 7 RBC concentrates). (B–D) Chymotrypsin-like (B), trypsin-like (C), and caspase-like (D) activities were measured on subsets for CFSE-stained long-stored unsorted (LS-unsorted) and flow-sorted CFSE^lo (LS-CFSE^lo) and CFSE^hi (LS-CFSE^hi), RBCs. In A, a 2-way ANOVA followed by Dunnett’s multiple comparison test was performed, and no significant difference was observed. In B–D, data are represented as individual points with mean ± SEM of 6 RBC concentrates and **P < 0.01 by Friedman’s 1-way ANOVA followed by Dunn’s multiple comparison test.

Figure 6. Storage lesions occurring during aging in vitro are concentrated in the long-stored CFSE^hi RBC subset; these are preferentially cleared during ex vivo human spleen perfusion.

(A) Retention rate by microsphiltration, (B) dynamic RBC adhesion to endothelial cells, (C) RBC surface PS exposure quantified by lactadherin staining, (D) osmotic fragility determined by measuring the NaCl concentration required to induce 50% hemolysis, (E) intracellular ATP levels, normalized to hemoglobin content, measured with CFSE-stained unsorted long-stored RBCs (LS-unsorted) RBCs and on flow-sorted LS-CFSE^lo and LS-CFSE^hi RBCs. In A–E, data are represented as individual points with mean ± SEM of ≥6 RBC concentrates). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 by Friedman’s 1-way ANOVA followed by Dunn’s multiple comparison test. In C, CTV-stained RBCs were analyzed to allow lactadherin-FITC staining. (F) Kinetics (mean ± SEM of 4 independent perfusions) of the normalized circulating concentrations of stained short-stored (SS), long-stored (LS), LS-CFSE^lo, and LS-CFSE^hi RBC subsets during ex vivo perfusion of human spleen. In F, a 2-way ANOVA followed by Šidák’s multiple comparison test was performed. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, and ^####P < 0.0001 by a 2-way ANOVA followed by Šidák’s multiple comparison test. Different symbols are used to show statistical significance: asterisks are used for comparisons of SS vs. LS and pound signs are used for comparisons of LS-CFSE^lo vs. LS-CFSE^hi. No statistically significant differences were seen when comparing SS-CFSE^lo vs. LS-CFSE^lo.

Figure 7. A proposed model to illustrate the main alterations affecting each RBC subset during storage.

The main results of the comparative molecular and cellular characterizations (black boxes) of long-stored CFSE^lo (light green, discocytes) and long-stored CFSE^hi (dark green, echinocytes III, spheroechinocytes, and spherocytes) RBCs. In long-stored CFSE^lo RBCs (left), functional energy and redox metabolism sustains effective lipid-repair and proteostasis functions, limiting protein aggregation, membrane relocation, and vesiculation. Functional molecular properties contribute to maintaining normal cellular properties (e.g., morphology, deformability, endothelial cell adhesion) and, thus, their ability to circulate. In long-stored CFSE^hi RBCs (right), decreased energy metabolism (e.g., ATP, PP, GAPD) is unable to fuel the redox system effectively, leading to decreased GSH and increased GSSG levels, culminating in dysfunction in lipid repair and proteostasis. The accumulation of irreversibly oxidized proteins could also lead to decreased proteasome function. The combined effect of oxidized protein accumulation and decreased proteasome degradation could fuel a negative feedback loop that further inhibits the redox system, thereby favoring production of toxic protein aggregates, protein relocation to the RBC membrane, and, ultimately, vesiculation. Vesiculation alters morphology, leading to decreased deformability, increased endothelial cell adhesion, and splenic retention. Mild (light blue boxes) and marked alterations are shown (pink boxes); black and red lines represent normal and negative interactions, respectively. PP, pentose phosphate, GADP, DL-GAPDH; GSH, reduced glutathione; GSSG, oxidized glutathione; Acyl-C, acyl-carnitines; Reversible/Irreversible Ox, reversible/irreversible oxidation.