Methylation of protein aspartates and deamidated asparagines as a function of blood bank storage and oxidative stress in human red blood cells

Abstract

Background: Being devoid of de novo protein synthesis capacity, red blood cells (RBCs) have evolved to recycle oxidatively damaged proteins via mechanisms that involve methylation of dehydrated and deamidated aspartate and asparagine residues. Here we hypothesize that such mechanisms are relevant to routine storage in the blood bank.

Study design and methods: Within the framework of the REDS-III RBC-Omics (Recipient Epidemiology Donor Evaluation Study III Red Blood Cell-Omics) study, packed RBC units (n = 599) were stored under blood bank conditions for 10, 23, and 42 days and profiled for oxidative hemolysis and time-dependent metabolic dysregulation of the trans-sulfuration pathway.

Results: In these units, methionine consumption positively correlated with storage age and oxidative hemolysis. Mechanistic studies show that this phenomenon is favored by oxidative stress or hyperoxic storage (sulfur dioxide >95%), and prevented by hypoxia or methyltransferase inhibition. Through a combination of proteomics approaches and ¹³ C-methionine tracing, we observed oxidation-induced increases in both Asn deamidation to Asp and formation of methyl-Asp on key structural proteins and enzymes, including Band 3, hemoglobin, ankyrin, 4.1, spectrin beta, aldolase, glyceraldehyde 3-phosphate dehydrogenase, biphosphoglycerate mutase, lactate dehydrogenase and catalase. Methylated regions tended to map proximal to the active site (e.g., N316 of glyceraldehyde 3-phosphate dehydrogenase) and/or residues interacting with the N-terminal cytosolic domain of Band 3.

Conclusion: While methylation of basic amino acid residues serves as an epigenetic modification in nucleated cells, protein methylation at carboxylate side chains and deamidated asparagines is a nonepigenetic posttranslational sensor of oxidative stress and refrigerated storage in anucleated human RBCs.

Figure 1 –. The rate of methionine consumption correlates with oxidative hemolysis in the REDS III Omics study.

In A, ~13.8 thousand donors were enrolled within the framework of the REDS III-Omics study. A single unit of blood was donated, leukofiltered and stored at 4°C for 42 days (standard blood bank conditions). At day 42, ~8,500 eligible units were tested for oxidative hemolysis (AAPH assay) and extreme hemolyzers (<5 and >95% percentile) recalled to donate a second unit. Subsequent units were processed as in phase I, though sampling occurred at storage days 10, 23 and 42 for a second measurement of oxidative hemolysis and targeted quantitation of methionine via standard UHPLC-MS metabolomics approaches. Methionine levels decreased significantly over the storage duration (n = 599; p <0.00001 ANOVA - B). In particular, methionine measurements in extreme 20 high vs low hemolzyers in the recalled cohort showed that methionine levels were the highest in high oxidative hemolyzers at storage day 10, but not in subsequent time points (C). Total methionine consumption (ΔMethionine at day 42 – day 10) was significantly higher in high oxidative hemolyzers (D). An overview of methionine metabolism is shown in E. * p < 0.05; ** p < 0.01 T-test.

Figure 2 –. Highlights of proteins with high-confidence assignment of D-methylation or N-deamidation (→methylation) in human RBCs

(UniProt abbreviations are listed – A; full list is provided in Supplementary Table 2, along with peptide sequences). In B, an overview of the cytosolic domain of band 3, from residues N-term (1) through residue 380 (SLC4A1_HUMAN; P02730). Stars highlight methylated aspartate and glutamate residues (single letter code D and E, respectively – in blue) as well as deamidated and deamidated → methylated residues (N – in green) on the N-term cytosolic domain of band 3. Rectangles also show the sequences on the N-term of band 3 that have been previously identified^– as essential mediators of the interaction between this membrane protein and structural proteins and glycolytic enzymes.

Figure 3 –. State of the art proteomics characterizes extensive methylation of N-term of Band 3.

Two representative spectra with extensive coverage of fragment ions (including those covering disambiguation of methylated residues D6/7 or D38) are provided in A and B. Additional spectra are shown in Supplementary Figure 3 and all peptides with Mascot ion scores and Δscores are reported in Supplementary Table 2. In C, an overview of the crystal structure of band 3 (N-term cytosolic domain, based on pdb: 1hyn). In D, the very N-term (1-16 residues of band 3 are highlighted (based on pdb: 3btb). D-methyl and N-deamidated residues are highlighted in red and green, respectively.

Figure 4 –. Methylation of isoaspartate and deamidated asparagine residues in human hemoglobin alpha and beta,

as mapped on their sequence (A), highlighted in the deoxyhemoglobin structure (based on pdb: 1a3n – B) and shown in a representative MS/MS spectrum (methylation of D22 – C). A top and side view of deoxyhemoglobin, along with the N-term of band 3 that has been shown to interact with the deoxyhemoglobin tetramer is shown in D. D-methyl and N-deamidated residues are highlighted in red and green, respectively.

Figure 5 –. Methylation of isoaspartate and deamidated asparagine residues in human glyceraldehyde 3-phosphate dehydrogenase (GAPDH),

as highlighted in a 3D structural model of the human ternary complex. (A-B) Surface representation of the holo GAPDH tetramer (PDB accession 1WNC) with coenzyme (red) and substrate (yellow). (C) Modified Asn residues (green) and oxidized residues (blue). Asn316 is within 3 Å of the coenzyme. The substrate was modeled based on a superposition with the bacterial ternary complex (PDB accession 1DC4). (D) Representative MS/MS spectrum (methylation of deamidated N155 in the active site of GAPDH) and the protein sequence.

Figure 6 –. Storage of leukocyte-filtered packed RBCs under normoxic, hyperoxic (SO₂>95%) or hypoxic conditions (SO₂ = 20, 10, 5 or <3 % - A) generated weekly samples for (B) metabolomics analysis of methionine, S-adenosylmethionine, and S-adenosylhomocysteine.

Continuous lines indicate interpolation by third degree polynomial of medians values for weekly measurements in 4 biological replicates (+ ranges – dotted lines). Y-axis values are intensity in arbitrary units. Color code for each condition is indicated in A: normoxia = blue; hyperoxia (SO2>95%) = purple; Hypoxia: SO2 = 20% (yellow); 10% (green); 5% (orange); <3% (red).

Figure 7 –. [¹³C₅,¹⁵N]methionine tracing experiments (A) were performed in fresh human RBCs, stored at 4°C for 0, 1 and 16h under normoxic or hypoxic condtions, in presence or absence of the methyltransferase inhibitor adenosine dialdehyde.

Additionally, RBCs were exposed to H₂O₂ (0.1% and 0.5%) under normoxic or hypoxic conditions for 1 to 16 h (time course). Hypoxia decreased methionine uptake (B) and consumption (C), respectively, while an opposite effect was observed in response to H₂O₂. Inhibition of methyltransferases decreased methionine consumption and labeled S-adenosylmethionine (SAM) consumption (D). Labeled SAM/SAH ratios were increased in response to the inhibition of methyltransferases, while these ratios decreased in normoxic RBCs exposed to H₂O₂ (E). Asterisks in panels B-E indicate significance by ANOVA with Tukey post-hoc test for multiple column comparisons (* p< 0.05; ** p < 0.01; *** p < 0.001). Relative quantitation of a representative deamidated (F) and total deamidated → methylated peptides (G) in the N-term of band 3 in RBCs stored under normoxic, hyperoxic (SO₂>95%) or hypoxic conditions (SO₂<5%) at storage day 1 and 42. In H, fold changes of total ¹³C-methyl-D labeled peptides (peak areas normalized to normoxic RBCs) in normoxic RBCs (+/− methyltransferase inhibitor – dark/light blue, respectively), hydrogen peroxide (0.5% - green) and hypoxia (red) upon 16h incubation with [¹³C₅,¹⁵N]methionine (1 mM). In I, a summary of the model proposed here.