Ethyl glucuronide, a marker of alcohol consumption, correlates with metabolic markers of oxidant stress but not with hemolysis in stored red blood cells from healthy blood donors

Abstract

BACKGROUND:

Red blood cell (RBC) storage in the blood bank is associated with the progressive accumulation of oxidant stress. While the mature erythrocyte is well equipped to cope with such stress, recreative habits like alcohol consumption may further exacerbate the basal level of oxidant stress and contribute to the progress of the storage lesion.

STUDY DESIGN AND METHODS:

RBC levels of ethyl glucuronide, a marker of alcohol consumption, were measured via ultra-high-pressure liquid chromatography coupled with high-resolution mass spectrometry. Analyses were performed on 599 samples from the recalled donor population at Storage Days 10, 23, and 42 (n = 250), as part of the REDS-III RBC-Omics (Recipient Epidemiology Donor Evaluation Study III Red Blood Cell-Omics) study. This cohort consisted of the 5th and 95th percentile of donors with extreme hemolytic propensity out of the original cohort of 13,403 subjects enrolled in the REDS-III RBC Omics study. Ehtyl glucuronide levels were thus correlated to global metabolomics and lipidomics analyses and RBC hemolytic propensity.

RESULTS:

Ethyl glucuronide levels were positively associated with oxidant stress markers, including glutathione consumption and turnover, methionine oxidation, S-adenosylhomocysteine accumulation, purine oxidation, and transamination markers. Decreases in glycolysis and energy metabolism, the pentose phosphate pathway and ascorbate system were observed in those subjects with the highest levels of ethyl glucuronide, though hemolysis values were comparable between groups.

CONCLUSION:

Though preliminary, this study is suggestive that markers of alcohol consumption are associated with increases in oxidant stress and decreases in energy metabolism with no significant impact on hemolytic parameters in stored RBCs from healthy donor volunteers.

Red blood cells (RBCs) are the most abundant cells in the human body, accounting for approximately 84% of the total host cells in an adult.¹ During the maturation process, RBCs lose nuclei and organelles and are thus devoid of de novo protein synthesis capacity. In parallel, RBCs are loaded with hemoglobin (92% of the total cell protein content) and iron (approx. 66% of the total bodily iron is in RBCs), a combination that maximizes the erythrocyte capacity to transport and deliver oxygen in vivo.² While critical to systems physiology, the lifelong exposure to oxygen and oxidant stress that arises as a consequence results in RBCs being exposed to significant amounts of oxidant stress, which is aggravated by the incapacity to replace oxidatively damaged components with new proteins. Overall, these phenomena limit the life span of the circulating erythrocyte, which averages approximately 120 days. The lack of mitochondria in mature mammalian erythrocytes also makes them incapable to leverage the very oxygen they carry to fuel energy-generating pathways. Vice versa, RBCs have evolved a series of antioxidant systems to cope with such oxidant stress, including mechanisms to directly counteract reactive oxygen species or recycle oxidatively damaged components.³

The evolutive pressure that has positively selected for these strong antioxidant systems has serendipitously conferred to the mature erythrocyte the capacity to cope with the oxidant stress that arises during refrigerated storage under blood bank conditions for up to 42 days. Indeed, it is now commonly accepted that cold liquid storage promotes the accumulation of a series of oxidant stress-driven changes to the erythrocyte, alterations that are collectively referred to as the storage lesion.⁴ Some of these lesions are reversible, such as those driven by the acidification of the intracellular pH⁵ and cold storage temperatures.^6,7 Other alterations to energy and redox metabolism are only partially reversible, and in part preventable by the use of alkaline additives^8–11 when compared to current-generation storage solutions like SAGM^7,12–14; Additive Solutions 1,¹⁵ 3,^16,17 5¹⁸; and PAGGSM.^8,19

Energy and redox metabolism are deeply intertwined in the mature erythrocyte, where high-energy phosphate compounds like adenosine triphosphate (ATP) and 2,3-diphosphoglycerate (DPG) represent critical substrates for the modulation of critical RBC processes, such as deformability, by constraining phosphorylation events and membrane lipid homeostasis (e.g., phosphatidylserine exposure), and hemoglobin affinity for oxygen. As storage progresses, ATP and DPG are consumed, and hemoglobin oxygen affinity increases, along with the generation of reactive oxygen species.^7,20 While these events have been observed in all laboratory studies on the storage lesion to date, the rate and extent of the progression of the so-called storage lesion appears to be dependent on donor-to-donor variability in the capacity to cope with oxidant stress.^21–25 As such, we now appreciate that the metabolic age of a unit of blood is qualitatively different from its chronological age, that is, the amount of time elapsed since the day of donation.^26,27 Significant heterogeneity is indeed observed across donors with respect to storage quality markers such as hemolysis²⁸ and posttransfusion recovery in healthy autologous volunteers in humans²⁹ and rodent models of storage and transfusion.^29–32 While metabolic markers of the storage lesion have been identified,²¹ the rate at which these markers accumulate/are depleted varies on a donor-to-donor basis, such as in the case of hypoxanthine,^8,29,33 or lipid peroxidation products.³² In keeping with recent evidence from randomized clinical trials,^34–37 the age of stored blood may not represent a clinically relevant parameter, to the extent that the heterogeneity in donor biology impacts blood storability and the molecular signatures that accompany it. This heterogeneity can be driven by genetic polymorphisms that have been selected for by evolution owing to the pressure exerted by specific infections (e.g., malaria on sickle cell traits or glucose 6-phosphate dehydrogenase [G6PD] deficiency). Incidence of some of these phenotypes is extremely high in subsets of the healthy donor population; for example, G6PD deficiency, the most common enzymopathy in humans that affects approximately 400 million people, impacts approximately 10% of the African American donor population in the New York metropolitan area.³⁸ These numbers can be further explained by the need to match donor and recipient biologies, a necessity that further results in the higher incidence of transfusion of blood with certain traits that are overrepresented in the donor population carrying rare/minor blood group antigens and/or sickle cell traits.

Other factors contributing to interdonor heterogeneity in storability may result from donor lifestyles and habits that can potentially impact erythrocyte redox biology and, consequently, storability. Recently, it has been reported that other factors beyond donor age, sex, and ethnicity²⁵ could be associated with blood storability, such as interdonor variability in the levels of circulating antioxidants like uric acid³⁹ or smoking habits. The latter has an incidence of approximately 13% in the US donor population and was found to correlate with increases in the levels of oxidant stress markers (e.g., carboxyhemoglobin) at baseline and as a function of storage duration.^40–42

Notably, alcohol consumption has been previously associated with increases in circulating markers of oxidant stress,^43,44 though to the best of the authors’ knowledge no study has reported data on the impact of alcohol consumption on the metabolic phenotypes of stored RBCs from healthy donors. To address this gap of knowledge in the literature, in the present study we correlated the levels of metabolic markers of alcohol consumption to metabolomics data, mainly focusing on energy and redox pathways and their changes during storage in RBCs from healthy donor volunteers. To this end, given the unavailability of self-reported questionnaire data on this habit for the tested cohort, we measured the levels of ethyl glucuronide in 599 samples from the Recipient Epidemiology and Donor Evaluation Study III (REDS-III RBC Omics) recalled donor popultation. Ethyl glucuronide is a metabolite of ethanol, which is formed in the body by glucuronidation following exposure to ethanol, usually from drinking alcoholic bever-ages.⁴⁵ As such, ethyl glucuronide is a marker of recent alcohol consumption, as it is a stable, nonvolatile, water-soluble metabolite that can be detected for an extended time period (up to 80 hr) after alcohol is completely eliminated from the body. Ethyl glucuronide levels in the urine,^45,46 hair,^47,48 or blood⁴⁹ have been used forensically and clinically to monitor chronic alcohol use and abuse.

The REDS-III RBC Omics study has been extensively described in recent papers.⁵⁰ Briefly, this project involved four blood centers across the United States and enrolled 13,403 healthy donor volunteers.^25,51 These subjects were asked to donate a unit of blood, which was stored until the end of the shelf life before measuring the RBC susceptibility to hemolysis, either spontaneous or following oxidant or ormotic insults. Those donors showing extreme hemolytic phenotypes (5th and 95th percentiles) were asked to donate a second unit of blood, which was sampled at Storage Day 10, 23, and 42 for untargeted metabolomics analyses. Preliminary analyses of data from the REDS-III RBC Omics study revealed a significant impact of donor sex (and testosterone levels⁵²), age, and ethnicity on the hemolytic propensity of stored RBCs,^25,50 a variable that showed donor-dependency across multiple subsequent donations, especially with respect to osmotic hemolysis.⁵¹ On the other hand, stored RBC capacity to cope with oxidant insults at the end of storage—though reproducible in the same donor across multiple donations—showed a higher degree of variance across multiple donations. Preliminary analyses of the metabolomics data from a pilot study suggested that this phenomenon could be at least in part explained by metabolic markers of processing strategies (e.g., storage additives) or donor lifestyles. To this end, in the light of the well-established impact of alcohol consumption on RBC metabolism, here we hypothesized that direct measurements of metabolic markers of alcohol consumption—like ethyl glucuronide—in the REDS-III RBC Omics recalled donor population could discriminate RBCs with different capacity to cope with oxidant insults, like those arising from storage in the blood bank.

MATERIALS AND METHODS

REDS-III RBC-Omics study participants and samples

Donor selection and recruitment for the RBC-Omics study were previously described in detail and, as such, the description herein is redundant with prior reports in the literature.^25,51,53 However, for the sake of completeness, we report again that donors were enrolled at four participating REDS-III US blood centers. Overall, 97% (13,403) of the whole blood donations provided by 13,758 donors age 18+ who provided informed consent were evaluable for hemolysis parameters. Donors in the 5th and 95th percentile for hemolysis measurements were recalled and asked to donate a second unit of blood. These units (n = 250 for the earliest time point available) were sampled at Storage Day 10, 23 and 42 for a total of 599 samples, as described.²³ Blood collection, sample processing and other aspects of the screening and recall phases of the RBC-Omics Study, including results of pilot studies to optimize processing and storage of samples for metabolomics analyses, are detailed in prior publications.^50,51

Sample processing and metabolite extraction

An isotopically labeled internal standard mixture including a mix of ¹³C¹⁵N-labeled amino acid standards (2.5 μM) was prepared in methanol. A volume of 100 μL of frozen RBC aliquots was mixed with water and the mixture of isotopically labeled internal standards (1:1:1, υ/υ/υ). The samples were extracted with methanol (final concentration of 80% methanol). After incubation at −20°C for 1 hour, the supernatants were separated by centrifugation and stored at −80° C until analysis.⁵⁴ Samples were vortexed and insoluble material pelleted as described.^16,55

Ultra-high-pressure liquid chromatography–mass spectrometry metabolomics

Analyses were performed using a Vanquish ultra-high-pressure liquid chromatography (UHPLC) coupled online to a mass spectrometer (Q Exactive, Thermo Fisher). Samples were analyzed with a 3-minute isocratic condition⁵⁶ or a 5-, 9-, or 17-minute gradient as described.^16,30,57,58 For the analysis of ethyl glucuronide levels were determined, samples were analyzed with a 5-minute C18 gradient and mass spectrometry (MS) negative ion mode acquisition as described.^16,30,57,58 MS acquisition, data analysis, and elaboration were performed as described. Solvents were supplemented with 0.1% formic acid for positive-mode runs and 1 mM of ammonium acetate for negative-mode runs. MS acquisition, data analysis, and elaboration were performed as described.^16,56 Additional analyses, including untargeted analyses, were performed with computer software (Compound Discoverer 2.0, Thermo Fisher). Graphs and statistical analyses (repeated-measures analysis of variance [ANOVA] and correlation analyses) were prepared with computer software (Prism 8.0, GraphPad Software, Inc.). Pathway analyses and other statistical analyses were performed with online tools for metabolomics data analysis and interpretation (MetaboAnalyst 4.0⁵⁹ and the Kegg pathway–based interactome was generated with the OmicsNet tool.⁶⁰

RESULTS

Ethyl glucuronide levels across the REDS-III RBC Omics recalled donor population

Measurements of ethyl glucuronide in the REDS III RBC Omics recalled donor population were performed via UHPLC–MS and are extensively reported in tabulated form in Table S1, available as supporting information in the online version of this paper. This population was characterized by approximately 250 healthy donor volunteers enrolled in four different blood centers across the United States. These donors represent a subset of a larger cohort enrolled within the framework of the REDS-III RBC Omics study (n = 13,403). This larger cohort was originally enrolled to donate a unit of blood that was stored for up to 42 days, when RBC aliquots from the unit were tested for hemolytic propensity (either spontaneous or following oxidant or osmotic insults). Donors with extreme phenotypes (5th and 95th percentiles) were recalled and asked to donate a second unit of blood, which was sampled at Storage Days 10, 23, and 42 for untargeted metabolomics analyses, also including detection of ethyl glucuronide (Fig. 1A). In the present study, we thus delved into the untargeted MS data collected within the framework of the REDS-III RBC Omics study and confirmed the high-resolution detection of ethyl glucuronide (mass-to-charge ratio: 221.0646; <1 ppm error on the intact mass and isotopic pattern consistent with an eight-carbon atom molecule and spectra consistent with available databases Chemspider and the Human Metabolome DataBase; Fig. 1B). We thus confirmed the reproducibility of this measurement across 40 nonconsecutive technical mixes ran approximately every 15 samples during the analysis of the whole REDS-III RBC Omics recalled donor set of 599 samples: coefficients of variation (CV = standard deviation/mean) were calculated to be below 10%, consistent with a reliable quantitation as per clinical chemistry laboratory standards (accepted threshold, 20%⁶¹). Finally, we confirmed the absence of carryovers and false positives from solvents, tubing, or other artifacts in 15 nonconsecutive blank runs, which were queued at regular intervals during the processing of the large sample set (<2% of the average of tech mixes; Fig. 1C).

Fig. 1.

UHPLC-MS measurements of ethyl glucuronide in the REDS III RBC Omics recalled donor population. Healthy donor volunteers (n = 13,400) were enrolled to participate in the REDS III RBC Omics study (A). They donated a unit of blood that was stored for up to 42 days, when RBC aliquots from the unit were tested for hemolytic propensity (either spontaneous or following oxidant or osmotic insults). Donors with extreme phenotypes (5th and 95th percentiles; n = 250) were recalled and asked to donate a second unit of blood. Metabolomics analyses were performed on samples from the second unit at Storage Days 10, 23, and 42. Within the framework of this study, we performed UHPLC-MS–based measurements of ethyl glucuronide, a marker of alcohol consumption. Correct assignments were performed against standard libraries in freely available database (e.g., HMDB, Chemspider) and validated against high-resolution intact mass and isotopic patterns (B). Reproducibility of the measurement was confirmed over 40 technical mixes ran approximately 15 samples apart during processing of one of the 599 samples of the recalled donor study (C), with CVs (standard deviation/mean) lower of approximately 10%. Blank runs were conducted at regular intervals during the processing of the large sample set to confirm the absence of carryovers (<2% of the average of tech mixes) (C). In (D), ethyl glucuronide measurements in the whole population at Storage Days 10, 23, and 42. In (E), an overview of the distribution in ethyl glucuronide levels across the population at any given storage day. Extreme lowest and highest (n = 15 per group) ethyl glucuronide at Storage Day 10 (earliest time point available in this study) informed the selection of a subgroup of donors for further analysis. Ethyl glucuronide levels in this subgroup at Storage Days 10, 23, and 42 are shown in (F). [Color figure can be viewed at wileyonlinelibrary.com].

Upon validation of technical standards, we quantified relative levels of ethyl glucuronide in the whole population at Storage Days 10, 23, and 42 (Fig. 1D), noting a trend toward decrease as a function of storage progression. We thus plotted ethyl glucuronide distribution across the whole population at any given time point tested in this study (Fig. 1E). Extreme lowest and highest (n = 15 per group) ethyl glucuronide at Storage Day 10 (earliest time point available in this study; Fig. 1F) informed the selection of a subgroup of donors for further analyses, as described below.

Healthy donor volunteers with the highest and lowest levels of ethyl glucuronide are characterized by significantly different metabolic profiles

Subjects were sorted on the basis of ethyl glucuronide levels at Storage Days 10, 23, and 42. The metabolic profiles (Table S1, available as supporting information in the online version of this paper; also including the population demographics) of the subjects characterized by the 15 highest and lowest measurements at any given storage day were juxtaposed for further multivariate analyses, including partial least squares discriminant analysis (PLS-DA – Fig. 2A), hieararchical clustering analysis (HCA; Fig. 2B) and metabolic pathway enrichment analysis (Fig. 2C). PLS-DA was successful in separating samples on the basis of the metabolic phenotypes. A total of 90 samples (15 per group for low– and high–ethyl glucuronide subjects at Storage Days 10, 23, and 42) were clustered as a function of storage time (vector aligned with PC2, explaining 10.7% of total variance) and ethyl glucuronide levels (PC3: 7.8% of the total variance—red vs. blue; Fig. 2C). HCA of significant metabolites by repeated measure ANOVA (two factors: time and ethyl glucuronide levels) revealed a significant association between ethyl glucuronide levels and RBC metabolic markers of oxidant stress. A vectorial version of this panel is provided in Fig. S1, available as supporting information in the online version of this paper, and a tabulated list of these metaboites is included in Table S1, available as supporting information in the online version of this paper. The list included metabolites involved in methionine and glutathione metabolism, fatty acid metabolism and oxidation, energy metabolism and the pentose phosphate pathway (PPP), amino acid transamination and carboxylic acids – as further confirmed by pathway analyses via OmicsNet (Fig. 2C).

Fig. 2.

Multivariate analysis of metabolomics data from the 15 subjects showing the highest and lowest ethyl glucuronide levels at Storage Days 10, 23, and 42. In (A), PLS-DA separates samples on the basis of the metabolic phenotypes and clusters them as a function of storage time (vector aligned with PC2, explaining 10.7% of total variance) and ethyl glucuronide levels (PC3: 7.8% of the total variance, red vs. blue). In (B), hierarchical clustering of significant metabolites by repeated measure ANOVA (two factors: time and ethyl glucuronide levels) for the 15 subjects showing extremes in ethyl glucuronide levels (a vectorial version of this panel is provided in Fig. S1, available as supporting information in the online version of this paper). In (C), pathway analysis of significant metabolites by ANOVA. Briefly, metabolites significant by ANOVA were submitted to the free online tool Omicsnet to perform a pathway analysis against the KEGG database. The most significantly enriched pathways are noted as groups of nodes (independent metabolites) and edges (connecting lines) and the pathways are summarized according to the legend enclosed in panel (C). [Color figure can be viewed at wileyonlinelibrary.com].

High levels of ethyl glucuronide are associated with lower energy metabolism and higher levels of oxidant stress markers

Energy metabolism was significantly lower at earlier storage time points in those subjects with the highest levels of ethyl glucuronide (Fig. 3)—Specifically, significantly lower levels of glucose, glucose 6-phosphate, fructose bisphosphate, and lactate at all the storage days tested in this study (Fig. 3A). Significant differences were also observed in the levels of triose phosphate compounds phosphoglcyerate and phosphoenol-pyruvate at Storage Days 10 and 42 (Fig. 3A), as well as pentose phosphate compounds (ribose phosphate) and the PPP metabolite erythrose phosphate (Fig. 3B), suggestive of alteration of RBC antioxidant capacity in light of the role of the oxidative phase of this pathway in the generation of the reduced form of the cofactor nicotinamide adenine dinucleotide phosphate (NADPH). Consistent with increases in oxidant stress in the subjects showing the highest levels of ethyl glucuronide, we noted significant oxidation of triose sugars, resulting in the formation of methylglyoxal (Fig. 4A). However, the glutathione system, which is involved in methylglyoxal detoxification, is significantly and negatively impacted in this group (Fig. 4A). Indeed, while the levels of reduced glutathione are higher in this group (though significantly so only at Day 10), the detoxification products of the glyoxylate pathway—lactoyl-glutathione and lactaldehyde—were significantly higher in the low ethyl glucuronide subjects through the whole storage period (Fig. 3A). Higher levels of cysteine, glutamate, and glutamyl-cysteine in the RBCs from high ethyl glucuronide subjects is suggestive of higher glutathione synthesis, which was, however, also accompanied by significantly higher turnover through the ablated gamma-glutamyl cycle (5-oxoproline; Fig. 3A). Significantly lower levels of ascorbate and dehydroascorbate were noted in the high–ethyl glucuronide group as well, suggestive of a further impaired oxidized glutathione recycling capacity—on top of the observed decrease in PPP metabolism (Fig. 3A). The same group showed the highest levels of glutathionylated hydroxyperoxylipids, a pathway that detoxifies the end products of lipid oxidation (4-hydroxynonenal) in a glutathione and NADPH-dependent fashion (Fig. 3A).

Fig. 3.

Glycolysis (A) and the PPP (B) in RBCs at Storage Days 10, 23, and 42 in 15 subjects with low (blue) or high (red) levels of ethyl glucuronide in the REDS III RBC Omics recalled donor population. Asterisks indicate significant (ANOVA, Tukey multiple column comparison: * p < 0.05; ** p < 0.01 *** p < 0.001). [Color figure can be viewed at wileyonlinelibrary.com].

Fig. 4.

Glutathione, ascorbate, and glyoxylate metabolism (A) and purine oxidation (B) in RBCs at Storage Days 10, 23, and 42 in 15 subjects with low (blue) or high (red) levels of ethyl glucuronide in the REDS III RBC Omics recalled donor population. Asterisks indicate significant (ANOVA, Tukey multiple column comparison: * p < 0.05; ** p < 0.01 *** p < 0.001). [Color figure can be viewed at wileyonlinelibrary.com].

Glutamate does not only represent a critical intermediate in glutathione synthesis, but also participates in transamination reactions, which have been reported to occur in the mature erythrocyte.^56,62 This pathway is relevant owing to the well-established role of serum⁶³ and liver transaminases⁶⁴ as markers of alcohol consumption. Alanine and aspartate—but also serine and other transamination substrates and products—were all significantly higher in the high–ethyl glucuronide group (Fig. 3B). Carboxylic acid products of these transamination events are also substrates for cytosolic isoforms of Krebs cycle enzymes in the mature erythrocyte, enzymes that contribute to reducing equivalent homeostasis and are intertwined to salvage reactions to repair deaminated purines in the face of oxidant stress.²⁹ Higher levels of the ATP breakdown product adenosine monophosphate (AMP) in the high–ethyl glucuronide group were accompanied by significantly higher levels of purine oxidation products through the whole cascade, including hypoxanthine, xanthine, urate, hydroxyisourate, and allantoate (Fig. 3B). Increases in adenosine but not AMP are suggestive that purine oxidation in this group is fueled by AMP breakdown and oxidation of adenosine (adenosine deaminase, or direct chemical oxidation; Fig. 4A) rather than the activation of redox-sensitive RBC-specific AMP deaminase 3, as previously noted in other studies.²⁹ Notably, prior studies stemmed from the REDS III RBC Omics initiative have helped reappreciate the pioneering work by Lou and Clarke⁶⁵ on the linkage between adenosine metabolism and RBC antioxidant pathways—specifically, methionine-dependent protein damage repair mechanisms—within the framework of RBC storage.⁵⁴ Methionine can indeed serve as a direct scavenger of reactive oxygen species, leading to the formation of methionine sulfoxide, here higher in the subjects with high levels of ethyl glucuronide (Fig. 4A). The same subjects seemed to display a higher rate of S-adenosylmethionine synthesis and consumption in common to low–ethyl glucuronide subjects, resulting in extremely significantly higher levels of S-adenosylhomocysteine in the former group (Fig. 5A). Despite these observations, limited (fold changes <1.5) and nonsignificant increases in the levels of oxylipins were noted in the high–ethyl glucuronide group, which was instead characterized by lower levels of mono- (16:1) and polyunsaturated fatty acids (20:4, 20:5, 22:5, 22:6; Fig. 5B). On the other hand, increases in citrulline, ornithine, spermine (Fig. 5C), tryptophan, and kynurenine (Fig. 5D) are suggestive of measurable alterations of arginine and tryptophan metabolism in the RBCs from subjects with high ethyl glucuronide levels.

Fig. 5.

Methionine (A), fatty acid and oxylipin (B), arginine (C), and tryptophan metabolism (D) in RBCs at Storage Days 10, 23, and 42 in 15 subjects with low (blue) or high (red) levels of ethyl glucuronide in the REDS III RBC Omics recalled donor population. Asterisks indicate significant (ANOVA, Tukey multiple column comparison: * p < 0.05; ** p < 0.01 *** p < 0.001). [Color figure can be viewed at wileyonlinelibrary.com].

Metabolic correlates to ethyl glucuronide confirm a linkage with oxidant stress

To further expand on the analyses above, we performed correlation analyses (Spearman) between ethyl glucuronide levels and other metabolites in the whole population (Fig. 6A) or just the 15 subjects with extreme levels of ethyl glucuronide (Fig. 6B). The results are extensively reported in tabulated form in Table S1, available as supporting information in the online version of this paper. Notably, S-adenosylhomocysteine was the top positive correlate to ethyl glucuronide levels in both the extreme subjects (as noted above), but also in the total population (r² = 0.284, p < 0.0001; Fig. 6C), with lactaldehyde (Fig. 6D) ranking among the top negative correlates. This analysis further confirmed a linkage between the capacity to manage oxidant stress and ethyl glucuronide, with glucose 6-phosphate—a rate-limiting substrate for the PPP—and dehydroascorbate (vitamin C/glutathione axis) ranking among the top negative correlates to ethyl glucuronide (Fig. 6E). However, this analysis also pointed out further interesting observations, including a potential association between markers of alcohol consumption and indole/tryptophan metabolites of microbial origin, such as methyleneoxindole and picolinic acid (Fig. 6E). Of note, we have previously reported alterations in the levels of these metabolites in the context of oxidant stress associated with aging⁶⁶ and Down syndrome.⁶⁷ Notably, picolinic acid has been shown to inihibit the neurotoxicant but not the neuroexcitant effects of other tryptophan catabolites like quinolinic acid.⁶⁸

Fig. 6.

Correlation analysis of ethyl glucuronide levels and metabolomics data in the whole REDS III RBC Omics recalled donor population (A), in the subjects showing the highest and lowest ethyl glucuronide level at any storage day (B) and highlighted correlations (C-E). [Color figure can be viewed at wileyonlinelibrary.com].

Despite the metabolic observations detailed above, in the present study we did not observe a significant correlation between ethyl glucuronide levels and hemolytic parameters, either spontaneous or induced by oxidant or osmotic stress. Ethyl glucuronide levels did not correlate to hemolytic parameters either when considering all of the samples or just the 15 subjects with extremes in ethyl glucuronide levels (Fig. S2A and B, respectively, available as supporting information in the online version of this paper).

DISCUSSION

In the present study, we measured ethyl glucuronide in 599 samples from the REDS-III RBC Omics recalled donor population as a marker of alcohol consumption. This parameter was correlated to the metabolic phenotypes of RBCs stored for 10, 23, and 42 days and noted a significant association between oxidative stress markers and ethyl glucuronide levels. Specifically, we report that RBCs from donors with the highest levels of ethyl glucuronide were also characterized by higher glutathione synthesis and turnover into oxoproline, a metabolic dead end in the mature erythrocyte owing to the lack of the enzyme oxoprolinase.¹⁶ In parallel, we noted significant decreases in glucose metabolism, ATP levels, and lactate generation, consistent with a decreased flux through glycolysis. Technical caveats are acknowledged with respect to the measurement of redox-sensitive thiol-containing metabolites (e.g., reduced glutathione) and high-energy phosphate compounds, like ATP, whose oxidation or breakdown, respectively, may in part result from technical artifacts associated with sample collection, extraction, and processing. However, it is worth noting that samples were immediately snap frozen at the time of collection. In addition, processing (including MS analyses) remained consistent throughout the analysis of the whole sample set tested in this study. After acknowledging these technical caveats, the phenotypes described above could be explained in part by increased oxidation of triose sugars into aldehydes and their glutathionylated metabolites. Similarly, increased levels of glutathionylated lipid peroxidation end products (hydroxynonenal metabolites) were detected in the RBCs from healthy donors with the highest levels of ethyl glucuronide. Methionine oxidation was increased in this group as well, along with S-adenosylmethionine consumption and S-adenosyl-homocysteine accumulation, a marker of increased protein isoaspartyl damage repair.⁵⁴ Increased markers of purine oxidation and purine salvage were observed in the same group. Despite increases in markers of oxidant stress, decreases in the levels of PPP metabolites and vitamin C (ascorbate and its oxidized form) were noted in the subjects with higher levels of ethyl glucuronide, suggestive of decreased glutathione recycling. Finally, free fatty acids (especially monounsaturated and polyunsaturated) and oxylipins were lower in this group, which was instead characterized by a higher degree of transamination products (alanine, aspartate, serine, glutamate). Overall, this phenotype is consistent with a negative correlation between elevated levels of metabolic markers of alcohol consumption and a decrease in RBC markers of energy and antioxidant metabolism during storage in the blood bank.

This study has several limitations. First of all, given the unavailability of self-reported data on drinking habits, it is unclear whether and to what extent RBC levels of ethyl glucuronide represent a reliable marker of alcohol consumption. While prior work has tested the performance of these markers in hair, urine, and whole blood, the specificity and sensitivity of this marker with RBCs as a matrix remain to be prospectively assessed. Being a very sensitive marker of ethanol exposure, circulating levels of ethyl glucuronide may be confounded by exposure to alcohol-containing food or medication, or even mouthwash. In addition, interdonor variability in ethyl glucuronide levels may be explained by factors such as the time that has passed since the last consumption of alcoholic beverages, the dose consumed, or even the heterogeneity in alcohol metabolism due to genetics or other factors (e.g., exposure to dietary or xenometabolites that are metabolized by the same aldehyde dehydrogenase and cytochrome P450 enzymes). As such, we decided to focus our analyses on the 15 highest and lowest subjects based on ethyl glucuronide measurements, with the goal to characterize the most extreme phenotypes associated with ethyl glucuronide levels. Therefore, given the nature of this analysis, we warn the readers about the risk of potentially overgeneralizing the impact of our findings to the whole REDS-III RBC Omics recalled donor population. Follow-up studies will be necessary to test the relevance of alcohol consumption on blood storage quality in a prospective rather than retrospective fashion. Indeed, in the present study, ethyl glucuronide was identified from a reanalysis of the original raw data generated through high-resolution untargeted metabolomics analyses. As such, the specific hypothesis of a potential impact of alcohol consumption on stored RBC metabolism was formulated retrospectively and subject to the statistical limitations associated with potential confounders unaccounted for in this study. Factors like donor sex, ethnicity, age,²⁵ storage additives,²³ and, potentially, G6PD status⁶⁹ did not seem to play a role in the meta-analysis of the subset of subjects displaying extremes in ethyl glucuronide levels in this study. However, these subjects represent but a subgroup of the REDS-III Omics recalled donor population, a population that was already selected on the basis of extreme hemolytic propensity phenotypes as assessed in the first phase of the REDS-III study.⁵¹ Similar analyses will have to be performed with the auxilium of stable isotope-labeled internal standards in prospective studies testing the impact of alcohol consumption in a controlled setting, evaluating the performance of ethyl glucuronide as a marker of alcohol consumption (and potential linear or threshold-dependent dose responses). Despite the limitations mentioned above, the present study represents a provocative first attempt at dissecting the potential impact of alcohol consumption on stored RBC metabolism. While the levels of ethyl glucoronide here seem to correlate with decreased energy metabolism and increased markers of oxidant stress in several pathways, no significant correlation between ethyl glucuronide levels and hemolytic propensity of the stored RBCs was observed in this study. As such, it remains to be assessed whether increases in detectable ethyl glucuronide levels in RBCs and accompanying metabolic changes represent meaningful predictors of posttransfusion performances in vivo, such as the capacity of old RBCs not to hemolyze,²⁸ circulate,^29,31,32 deliver, and offload oxygen^70,71 upon transfusion, as preliminary data in humans and rodent models seem to suggest.

ABBREVIATIONS:

AMP

adenosine monophosphate

ANOVA

analysis of variance

ATP

adenosine triphosphate

DPG

2,3-diphosphoglycerate

G6PD

glucose 6-phosphate dehydrogenase

HCA

hieararchical clustering analysis

MS

mass spectrometry

NADPH

nicotinamide adenine dinucleotide phosphate

PLS-DA

partial least squares discriminant analysis

PPP

pentose phosphate pathway

REDS-III

Recipient Epidemiology and Donor Evaluation Study III

UHPLC

ultra-high-pressure liquid chromatography