Page 200 - 2020_08-Haematologica-web

D. Stefanoni et al.
Results
Fresh red blood cells from macaques and humans differ metabolically
Metabolomic analyses were performed on leukocyte- filtered, fresh (day 0) RBC from healthy human volun- teers (n=21) and RM (n=20) (Figure 1A; Online Supplementary Table S1). Significant differences between species were determined by partial least squares-discrim- inant analysis, t-test-informed hierarchical clustering, and volcano plots (Figure 1B-D, respectively). Significant changes were noted in the levels of purines (e.g., hypox- anthine, urate), arginine and sulfur metabolites (e.g., glu- tathionylcysteine, glutathione, phytochelatins), car- nitines, and xenometabolites (e.g., caffeine) in fresh RBC from the two species (Figure 1D). In the light of these changes, correlation analyses (Spearman) were performed across all metabolites to define the level of metabolic linkage in RBC from either species (Figure 1E). Identifying correlates in one species (e.g., macaques), disruptions of such correlations, and generation of novel correlations indicate metabolic rewiring (Figure 1F, from left to right). These measurements can then be used to subtract the cor- relations observed for every pair of metabolites in each species, resulting in a differential heat map (Figure 1F, rightmost panel). This map highlights pathways that are preserved (Figure 1G, leftmost panel), and those that undergo metabolic rewiring in RBC from these species (>30% Δr between species, P<0.05), such as those involv- ing glutathione homeostasis, sulfur, purine, carboxylic acid, and arginine metabolism (Figure 1G).
Metabolic tracing experiments with [1,2,3-13C ]glucose 3
in fresh human or macaque red blood cells
In light of metabolic differences in glutathione home- ostasis observed at steady state, we hypothesized that species-specific alterations of glucose fluxes affect the NADPH-generating pentose phosphate pathway (PPP). NADPH is required to preserve RBC redox homeostasis by favoring reduction of glutathione and other reversibly oxidized thiols. To test this hypothesis, leukoreduced RBC lysates from RM (n=20) and humans (n=21) were incubated with [1,2,3-13C3]glucose for 1 h at 37oC (Figure 2). This allows comparisons of 13C incorporation through glycolysis (+3 isotopologues) and the PPP (+2) in RM and human RBC at steady state (Figure 2; Online Supplementary Figure S1), as described previously.9,33 Importantly, RM and human RBC showed similar rates of glucose consumption, along with comparable levels of the glycolytic products pyruvate and lactate, in the absence of any apparent preference in glycolysis/PPP flux- es (Online Supplementary Figure S1). However, early steps of glycolysis showed significantly different rates between these species, with higher levels of 13C3-glucose 6-phos- phate (and hexose phosphate isobars) in human RBC and higher levels of 13C3-fructose bisphophate in RM (Figure 2). RM RBC showed significantly higher levels of 13C accumulation in intermediates of the glyoxylate pathway (e.g., 13C3-methylglyoxal and 13C3- and 13C5-lactoyl-glu- tathione) and late PPP products (e.g., 13C2-ribose), but lower levels of labeled glutathione and purines (e.g., 13C2- AMP) (Figure 2). Further focusing on reducing equiva- lents,34 there were species-specific preferences in sub- strates used to generate carboxylic acids, with 13C3-malate preferred in RM and 13C2-malate in humans (Figure 2).
Interspecies comparison of the red blood cell metabolome during refrigerated storage
Metabolomic analyses were performed on 574 samples of stored RBC and supernatants from RM and humans (Figure 3A; Online Supplementary Table S1). Partial least squares-discriminant analysis of RBC data showed signifi- cant species- and time-dependent clustering of these sam- ples across principal component 1 (28.8% of total variance) and principal component 2 (18.6%), respectively (Figure 3B). The Venn diagram in Figure 3C, D shows the number of significant metabolites by repeated measures two-way analysis of variance and related pathway analyses. Hierarchical clustering analyses further highlighted signifi- cant storage- and species-dependent differences (Figure 3E; a vectorial version of this figure, including metabolite names, is provided in Online Supplementary Figure S2).
Targeted metabolomic analyses were accompanied by untargeted metabolomic analyses (Online Supplementary Figure S3A-D). Interestingly, these analyses expanded on the targeted metabolomic data by highlighting species-spe- cific changes in levels of xenometabolites stored RBC (Online Supplementary Figure S3E) derived from personal habits (e.g., cotinine from smoking in 2 of 21 donors), chemical exposure (e.g., aniline, nitrosopiperidine), thera- peutic drugs (e.g., acetaminophen, in 2 subjects), and diet (e.g., caffeine and theophylline in humans; methyl-histi- dine, phloionic acid, lupinine, gallocatechin, azelaic acid, and asarone in RM).
Although limited by the relatively small numbers of male and female human and RM donors evaluated, a pre- liminary breakdown by sex identified a significant impact in RM, especially regarding carboxylic acid, arginine, fatty acid, and purine metabolism (Online Supplementary Figure S4A-D).
Species-specific differences in metabolic phenotypes of stored red blood cells
Significant differences were observed in RBC levels of sulfur-containing metabolites involved in one-carbon and glutathione metabolism (Figure 4), including taurine, S- adenosylmethionine (SAM), cysteine, cystathionine, and glutathione [both reduced (GSH) and oxidized (GSSG)]; all were higher in RM than human RBC throughout storage, except GSH. Increased glutathione pools and activation of the gamma-glutamylcycle, ascorbate metabolism, and glu- taminolysis were observed in RM, as compared to human, RBC (Figure 4). These observations were not accompanied by significantly different levels of PPP intermediates (except for higher levels of the non-oxidative phase PPP metabolite sedoheptulose phosphate in humans). However, in contrast to what was observed in fresh RBC, stored RM RBC showed higher levels of intracellular glu- cose in supernatants and cells (Figure 4), despite compara- ble levels of intracellular and supernatant levels of lactate (i.e., ~10% increase in RM, P<0.05) (Online Supplementary Table S1). Of note, human RBC showed lower levels of 2,3- diphosphoglycerate, but higher levels of ATP, during the first 2 weeks of storage (Figure 4).
Expanding on these observations, a deeper focus on purine metabolism revealed significantly higher levels of all purines and purine-containing metabolites in RM RBC. However, dramatic species-specific changes in purine oxi- dation metabolites were observed (Figure 5A), with RM RBC showing significantly higher levels of hypoxanthine (accumulating during storage in both species) and human
2176
haematologica | 2020; 105(8)