Red Blood Cell Urate levels are linked to hemolysis in vitro and post-transfusion as a function of donor sex, population and genetic polymorphisms in SLC2A9 and ABCG2

Abstract

Background:

Storage of packed red blood cells (RBCs) for transfusion leads to biochemical and morphological changes, increasing hemolysis risk. Urate levels in blood bags at donation contribute to the molecular heterogeneity and hemolytic propensity of stored RBCs. However, studies to date have been underpowered to investigate at scale the contribution of donor demographics and genetics to the heterogeneity in urate levels across donations.

Study Design and Methods:

Urate levels were measured in 13,091 RBC units from the REDS study. Characteristics tested included hemolysis parameters (spontaneous, osmotic, oxidative) at storage end and post-transfusion hemoglobin (Hb) increments in recipients. Donor demographics, urate levels, and genetic variants were analyzed for associations with these outcomes.

Results:

Elevated urate levels were linked to male sex, older age, high BMI, and Asian descent. Units with high urate levels exhibited increased spontaneous and osmotic hemolysis, while oxidative hemolysis was unaffected. Genetic variants in SLC2A9 (V282I) and ABCG2 (Q141K) were strongly associated with elevated urate, particularly in Asian donors. Post-transfusion analyses revealed that units from female donors carrying these variants were associated with reduced Hb increments, with up to a 31% reduction in efficacy. This effect was not observed in male donors.

Discussion:

RBC urate levels and genetic traits significantly impact storage quality and transfusion outcomes. These findings highlight the importance of donor molecular characteristics for optimizing transfusion strategies. Moreover, genetic and metabolic insights may inform donor recruitment efforts, providing health feedback to volunteers while ensuring effective transfusion products.

INTRODUCTION

Red blood cells (RBCs) are central regulators of systems metabolism, first and foremost by scavenging oxygen in the lungs to distribute to peripheral tissues. Unlike the other 5 trillion cells in the human body, RBC lose mitochondria during the maturation process, hence RBC metabolism does not leverage O₂ as a final acceptor of electrons.¹ On the other hand, RBCs exclusively rely on glycolysis to synthesize high-energy phosphate compounds like adenosine triphosphate (ATP), whose loss results in the untimely removal of RBCs from circulation.¹ As they circulate through the human body, RBCs also communicate with various organs by releasing and importing small molecule metabolites through over 77 transporters.² For example, in response to hypoxia, RBCs release ATP which directly or indirectly (through breakdown to ADP or adenosine via extracellular ectonucleotidases such as CD73) triggers purinergic receptors to promote vasodilation and adaptation to low oxygen environments (e.g., at high altitude).¹ Circulating adenosine is highly bioactive through stimulation of adenosine receptors: for example, adenosine concentrations above 14 μM can impact cardiovascular activity and even cause arrhythmia. As such, adenosine is rapidly metabolized by adenosine deaminase to inosine and hypoxanthine, which is in turn further oxidized to xanthine and, ultimately, to urate – a process that not only prevents salvage of deaminated purines, but also concomitantly generates hydrogen peroxide. ATP breakdown and oxidation to urate, through the concerted activity of redox-sensitive deaminases (e.g., AMP deaminase 3 in RBCs³) and plasma NADPH-dependent xanthine oxidase, also contributes to the accumulation of circulating urate.¹

Beyond RBCs, dysfunctional mitochondrial metabolism, such as in the case of ischemic or hemorrhagic hypoxia, also contributes to ATP breakdown to hypoxanthine and urate accumulation, a process that manifests uricemia not only in the context of acute stress, but also chronic conditions such as in non-communicable diseases such as obesity and diabetes. Accumulation of urate crystals is an etiological contributor to gout, a condition more prevalent in older males with elevated body mass index.⁴

Every year, 5–6 million Americans donate blood, enabling life-saving transfusions, a medical intervention ranking second in frequency only to vaccination in the US. RBC components derived from donated blood can be stored for up to 42 days under refrigerated conditions, a process that – while logistically indispensable – comes at the cost of the progressive accumulation of biochemical and morphological alterations to the stored erythrocytes, a phenomenon collectively referred to as the “storage lesion”⁵. Over the last two decades, ongoing debate has questioned whether storage duration ultimately results in increased post-transfusion complications or less effective products. Since then, a wealth of studies have documented that metabolism and the hemolytic propensity of stored RBCs are significantly impacted by storage duration as a function of storage additives, donor genetics, sex, age, ethnicity, BMI, and dietary or other exposures.⁶ Following reassuring evidence from prospective randomized clinical trials on the age of blood, at the net of the caveats noted,⁷ it is now generally accepted that the chronological age of blood, i.e., time elapsed since donation, is intrinsically different from the molecular age of the unit.⁸ Indeed, while the metabolic storage lesion can be mathematically modeled based on key metabolic markers,⁹ the onset, progression and ultimate severity of storage-induced metabolic changes varies from donor to donor. As such, molecules like kynurenine – best predictors of osmotic fragility as gleaned via machine learning approaches – do not change at all as a function of storage duration, rather they correlate with donor BMI, age and genetic polymorphisms in regions coding for solute carrier transporters for large amino acids such as LAT1 (SLC7A5).¹⁰ This observation does not necessarily imply that all storage lesion(s) are irrelevant, but rather that combinations of pre-storage and storage lesion-associated factors may compound to influence the quality of transfused RBCs. Such is the case for example of glycolysis and ATP,¹¹ or L-carnitine pools,¹² both pathways involved in RBC energy and redox homeostasis which are depleted as a function of storage duration, but are also significantly affected by donor genetic and non-genetic factors (e.g., sex, age, diet or habits – such as smoking, consumption of caffeinated or alcoholic beverages¹³).

Small scale studies (n=3 and n=8, respectively) have suggested that urate is one key metabolite that varies as a function of donor sex and storage duration^14,15. While the process of urate synthesis is per se associated with the generation of pro-oxidant hydrogen peroxide, urate is an antioxidant whose supplementation has been proposed as an intervention to improve storage quality.^16,17 Building upon these preliminary observations, here we leveraged the Recipient Epidemiology and Donor Evaluation ^,Study (REDS) RBC Omics, by investigating urate levels in 13,091 end of storage packed RBCs from as many blood donor volunteers enrolled in 4 different blood centers across the United States.

METHODS

The methods in this section are consistent with previous reports from our group, adapted to focus on urate measurements and related elaborations.^10–12,18 While consistent with the literature, these methods are extensively reported in Supplementary Methods Extended and summarized below.

Donor recruitment in the REDS RBC Omics study:

Index donors

End of storage (day 42) RBC samples were obtained from leukocyte-filtered blood units donated by 13,091 consenting volunteer donors (henceforth referred to as the “Index cohort”), who were enrolled into the Recipient Epidemiology and Donor evaluation Study (REDS) RBC Omics (https://biolincc.nhlbi.nih.gov/studies/reds_iii/), as previously described and detailed in Supplementary Methods Extended. These end of storage packed RBC units were also tested for hemolysis parameters, including spontaneous hemolysis (n=12,753), or hemolysis following oxidative or osmotic insults - n=10,476 and 12,799, respectively),¹⁹ as detailed in Supplementary Methods.²⁰

Urate measurement:

Metabolomic measurements were performed as described previously and detailed in the Supplementary Methods Extended. In brief, small molecule metabolites from end of storage blood samples from 13,091 index donors were separated via ultra-high-pressure liquid chromatography (UHPLC) followed by high resolution mass spectrometry (Vanquish, Q Exactive – Thermo Fisher, San Jose, CA, USA), with targeted post hoc analysis via Maven (Princeton) for urate peak assignment based on accurate intact mass, retention time against a chemically pure standard (U2875 – Sigma Aldrich) and area integration.

Recalled donors:

Upon testing for hemolysis parameters, index donors who ranked below the 5^th and above the 95^th percentile (extremes) for any of the hemolysis parameters were contacted again and invited to donate a second unit of pRBCs (henceforth referred to as the “recalled cohort” – n=643). Samples were sterilely collected from these units at storage days 10, 23 and 42, for testing of hemolysis and mass spectrometry-based high-throughput metabolomics²¹, proteomics, lipidomics and trace-element analysis, as part of the REDS-IV-P program, as detailed in Supplementary Methods extended. In total, 1,929 serial samples were tested (n=643 units, each at storage days 10, 23 and 42).

Metabolite Quantitative Trait Loci (mQTL) analysis

The workflow for the mQTL analysis of urate is consistent with previously described methods for kynurenine,¹⁰ L-carnitine and acyl-carnitines,¹² ATP and hypoxanthine, glycolytic metabolites¹¹ and 2,3-bisphosphoglycerate¹⁸ studies in this same cohort and metabolomics studies in 250 donors from a pilot study.²² Details of the genotyping and imputation of RBC Omics study participants are consistent with published protocols²³ and are detailed in Supplementary Methods Extended. Briefly, genotyping was performed using a Transfusion Medicine microarray (879,000 Single Nucleotide Polymorphisms - SNPs) and the data are available in dbGAP accession number phs001955.v1.p1. Imputation was performed using 811,782 SNPs that passed quality control. After phasing using Shape-IT,²⁴ imputation was performed using Impute2²⁵ with the 1000 Genomes Project phase 3²⁵ all-ancestry reference haplotypes. We used the R package SNPRelate²⁶ to calculate principal components (PCs) of observed genotypes to infer genetic ancestry. We performed association analyses for urate using an additive SNP model in the R package SAIGEgds²⁷ on the 13,091 study participants who had both metabolomics data and imputation data on serial samples from stored RBC components that passed respective quality control procedures. We adjusted for relevant covariates for this trait, such as sex, age (continuous), frequency of blood donation in the last two years (continuous), blood donor center, and ten genetic PCs. Genome-wide statistical significance was defined for both SNPs and small insertion-deletions (indels) at a p-value threshold of 5×10⁻⁸. We only considered variants with a minimum minor allele frequency of 1% and a minimum imputation quality score of 0.80. The ENSEMBL variant effect predictor was used to annotate all genome-wide significant SNPs. This annotation includes information on the type of variant (e.g., exonic vs intronic), distance to closest gene, predicted functional impact (e.g., missense), tissue-specific gene expression, and other information.

Linking Urate mQTL to post-transfusion hemoglobin increments:

Genome-wide significant SNPs associated with urate levels in the index donors were next compared to hemoglobin increments following RBC transfusions in recipients by interrogating the REDS vein-to-vein database as described^28,29 and here briefly summarized. We interrogated the BioLINCC database to access public use data from the NHLBI REDS-III cohort.³⁰ The database includes blood donor, component manufacturing, and patient data collected at 12 academic and community hospitals from four geographically diverse regions in the United States (Connecticut, Pennsylvania, Wisconsin, and California) for the 4-year period from January 1, 2013 to December 31, 2016. Allele frequencies for SNPs associated with urate levels from the REDS-III Index donors³¹ (as a function of biological sex) were linked to hemoglobin increments in adult patients who received a single RBC unit during one or more transfusion episodes between January 1, 2013 and December 30, 2016. The outcome measure of interest was a change in hemoglobin (ΔHb; g/dL) following a single RBC unit transfusion episode. This outcome was defined as the difference between the post-transfusion and pre-transfusion levels, adjusted for biological and technical co-factors, including: donor characteristics (age, sex, BMI, hemoglobin), component apheresis, irradiation, storage age, storage solution, and recipient characteristics (age, sex, BMI, hemoglobin). For pre-transfusion hemoglobin, the value used was the most proximal hemoglobin measurements prior to RBC transfusion, but at most 24 hours prior to transfusion. Furthermore, we excluded transfusion episodes where the pre-transfusion hemoglobin was greater than 9.5 g/dL, and the hemoglobin increment may be confounded by hemorrhage events. For post-transfusion hemoglobin, the laboratory measure nearest to 24-hours post-transfusion, but between 12- and 36-hours following transfusion was used. Multivariable linear regression assessed associations between urate-linked SNPs from the donor and changes in post-transfusion hemoglobin levels in the recipient. Two-sided p-values less than 0.05 were considered to be statistically significant. Analyses were performed using Stata Version 14.1, StataCorp, College Station, TX. Further details are provided in Supplementary Methods Extended.

Data analysis Statistical analyses included hierarchical clustering analysis (HCA), linear discriminant analysis (LDA), Loess smoothing for line plot analyses, and correlation analyses, which were performed using both MetaboAnalyst 5.0³² and in-house developed code in R (4.2.3 2023–03-15).

RESULTS

Urate levels are higher in older, male donors with elevated BMI and of Asian descent

Urate levels were measured in packed RBC samples frozen at storage day 42 from 13,091 REDS RBC Omics index donations (Figure 1.A – all raw data and elaboration are reported in Supplementary Table 1). Urate levels followed a Gaussian distribution, though a subset of donors tailed towards the highest end of urate (2–4 fold change higher than the normalized values for this cohort). Linear discriminant analysis (LDA) of the index donor cohort indicates clustering for donors across urate quartiles, suggesting that urate levels are sufficient to discriminate amongst donors. To expand on this observation, we next focused on donors with the highest or lowest (n = 500 each) urate levels. High vs low (red vs blue) donors by urate levels showed enrichment of male, older (p < 0.5e-9– two-tailed T-test), higher BMI (p < 0.6e-29) donors of Asian descent in the high urate group, while donors of African and Hispanic descent were over-represented in the low urate sub-group (Figure 1.D). More specifically, urate levels were observed to increase generally with age with a plateau between approximately 30 – 50 years old (Figure 1.E). Male donors showed higher urate levels across the whole age range tested in this study (16–87 years – Figure 1.F), while urate levels were the highest in morbidly obese donors (BMI > 40 kg/m²) and lowest in donors with BMI < 25 kg/m² (Figure 1.G). In this cohort, units from donors of genetically inferred East Asian descent (EAS) were characterized by the highest levels of urate, while donors of inferred Hispanic (Mexico and Central America – HISP2) or African (AFR) descent had the lowest levels of urate (Figure 1.H).

Figure 1 – Impact of donor age, sex, BMI and ethnicity on end of storage urate levels in 13,091 end of storage packed RBCs from the REDS RBC Omics Index cohort.

Distribution of urate levels at storage day 42 in REDS RBC Omics (n=13,091 A). In B, linear discriminant analysis (LDA) of the index donor cohort indicates clustering for donors across urate quartiles. In C, donors with the highest or lowest (n=500 each) urate levels were bucketed into two distinct groups. High vs low (red vs blue) donors by urate levels showed a marker enrichment of male, older, higher BMI donors of Asian descent in the high urate group, while donors of African and Hispanic descent were over-represented in the low urate sub-group (D). In E, urate levels increased with age. Male donors showed higher urate levels across the whole age range tested in this study (16–87 years - F). Lowest BMI donors had significant lower levels of urate (G). Donors of East Asian descent were the ones with the highest urate levels, while donors of Hispanic descent (Mexico and Central America – HISP2) had the lowest levels of urate (H).

Higher urate levels are associated with higher spontaneous and osmotic hemolysis, especially in units from male donors

In the REDS RBC Omics index cohort, day 42 stored RBCs from donors with the highest urate levels (n = 500) had significantly higher rates of storage and osmotic hemolysis (p < 0.00001 – unpaired two-tailed T-test) and higher rates of oxidative hemolysis (albeit not significantly – p = 0.31) (Figure 2.A–B). A subset of the index donors ranking below the 5^th or above the 95^th percentile for all hemolytic parameters donated a second unit of blood, which was again stored for 42 days prior to measurements of hemolysis parameters. In this “recalled donor cohort” (n=643), storage and osmotic hemolysis were again confirmed to be higher in the units with the highest levels of urate compared to the low urate group (p < 0.01 and < 0.001 for storage and osmotic hemolysis, respectively) (Figure 2.B). Both in the index and recalled donor cohorts (Figure 2.C–D), highest osmotic fragility was specifically observed in blood units with the highest levels of urate who were obtained from male donors, suggesting that these phenotypes are mostly driven by sex dimorphism. These results are in part explained by the observation that donors with the highest levels of urate in the recalled cohort, when compared to the lowest (n=75 per group), manifest dramatic changes in metabolism such as higher reduced glutathione and lower oxidized glutathione (Supplementary Figure 1), consistent with increased resistance to oxidative hemolysis. We note the clear sex bias in this comparison with the recalled donors with high urate (n=75) being 100% male and low urate (n=75) being 100% female.

Figure 2 – Urate levels were associated with significantly higher storage and osmotic hemolysis.

In the REDS RBC Omics index cohort, day 42 stored RBCs from donors with the highest urate levels (n=500) had higher storage and osmotic hemolysis (p<0.00001), lower (albeit not significantly) oxidative hemolysis (A-B). A subset of the index donors ranking below the 5^th or above the 95^th percentile for all hemolytic parameters were invited to donate a second unit of blood, which was again stored for 42 days prior to measurements of hemolysis parameters. In this “Recalled” donor cohort (n=643), storage and osmotic hemolysis were again confirmed to be higher (p<0.01 and <0.001 for storage and osmotic hemolysis, respectively) in the units with the highest levels of urate compared to the low urate group (B). In C and D, highest osmotic fragility corresponded to higher levels of urate in units from male donors, suggesting that these phenotypes are mostly driven by sex dimorphism.

RBC urate levels decline during storage and are reproducible across multiple donations

In the Recalled donor population (n=643), urate levels were measured at storage day 10, 23 and 42 (Figure 1A), and we observed a progressive decline as a function of storage duration (Figure 3.B), consistent with the literature.^9,33–37 These declines were observed in both units from male and female donors, despite the latter units showing significantly higher urate levels (p < 0.00001) throughout storage (Figure 3.C). Since both index and recalled blood units were tested at storage day 42, we correlated end of storage levels of urate in the 643 units from index donors who were also recruited in the recalled cohort (Figure 3.D), which demonstrated a highly significant (p = 1.26e-61) degree of reproducibility of end of storage urate levels across two independent donations (6–12 months apart) (Spearman rho = 0.59 – Figure 3.E).

Figure 3 – Urate levels decrease over storage and end of storage urate levels are reproducible across multiple donations from the same donor.

In the Recalled donor population (n=643), urate levels were measured at storage day 10, 23 and 42, showing a progressive decline as a function of storage duration (B). These declines were observed in both units from male and female donors, despite the latter units showing significantly higher levels (p<0.00001) throughout the storage duration (C). Since both index and recalled blood units were tested at storage day 42, we correlate end of storage levels of urate in the 643 units from index donors who were also recruited in the recalled cohort (D), whowing a significant (p=1.26 e-61) degree of reproducibility of end of storage urate levels across two independent donations (6–12 months apart) from the same 643 donors (Spearman rho=0.59 - E).

Quantitative Trait Loci (QTL) analysis associates end of storage levels of urate to polymorphisms in the regions coding for SLC2A9 and ABCG2

Given the high degree of reproducibility of urate levels within the same donor across repeated donation (Figure 3.E), and the impact of inferred genetic ancestry on end of storage urate levels across all index donors, especially those with extremely high vs low urate levels (Figure 1), we hypothesized that genetic factors may contribute to the inter-donor heterogeneity of urate levels at donation and as a function of storage. To test this hypothesis, we performed metabolite Quantitative Trait Loci (mQTL) analyses, linking end of storage urate levels in 13,091 blood units to 879,000 SNPs that were assayed via a Precision Transfusion Medicine array in this cohort.³⁸ Genome-wide significant associations were observed at 1,139 SNPs and small indels across three independent regions (Figure 4.A), and no evidence of association inflation was observed (Figure 4.B). Two of these regions were on chromosome 4 containing genes solute carrier 2A9 (SLC2A9) and ATP-binding cassette group 2 (ABCG2) and one was on chromosome 3 near gene 1,4-alpha-glucan branching enzyme 1 (GBE1) The SNP with the lowest p-value (aka the lead SNP) at gene SLC2A9 was rs16890979, a missense mutation converting a valine 282 to isoleucine (V282I;– p= 3.72e-70 - Figure 4.C). The lead variant for ABCG2, rs45499402 (p=1.3e-17), was intronic (Figure 4.D). One intronic SNP in the coding region for GBE1 was observed at genome-wide significance, rs76758100 (p = 2.4e-8, Supplementary Figure 2). Of note, significant associations were also observed for other relevant SNPs coding for missense variants of SLC2A9 (rs2276961 – G25R – p= 7.20e-19) and ABCG2 (rs2231142 – Q141K – p=1.64e-17). On the other hand, while showing a structure for an association signal, the region coding for SLC22A12 (URAT1) did not pass genome-wide significance (Supplementary Figure 2).

Figure 4 – Quantitative Trait Loci (QTL) analysis associates end of storage levels of urate to polymorphisms in the regions coding for SLC2A9 and ABCG2.

In A, Manhattan plot for urate reveals single nucleotide polymorphisms (SNPs) on the coding regions for the solute carrier 2 A 9 (SLC2A9 – rs16890979) and ATP-binding cassette group 2 (ABCG2) transporter (rs4599402) as the top two loci associated with heterogeneous urate levels in day 42 packed RBCs (A). QQ plot (B) and locus zoom for each SNP (C-D, respectively).

These associations were observed across genetic ancestries, though they were the strongest in the largest population investigated here, i.e., donors of inferred European descent, followed by donors of inferred Asian and Hispanic descent (Supplementary Figure 3).

Extremes in urate levels, SLC2A9 and ABCG2 alleles are more frequent in donors of Asian descent

The rs16890979 SLC2A9 (missense mutation V282I) was strongly associated with urate levels, NAD-breakdown product nicotinamide, bile acids, ferritin and BMI in index donors, and was most prevalent in donors of genetically inferred Asian descent, (Figure 5.A–B; Supplementary Table 1). The intronic rs45599402 ABCG2 SNP was also linked to urate levels, nicotinamide and bile acids (taurocholate), also most prevalent in donors of genetically inferred Asian descent (Figure 5.C–D). Altogether, these analyses expand upon the results presented in Figure 1 and offer a key to interpret inter-donor heterogeneity in urate levels not just as a function of donor demographics, but also genetic background.

Figure 5 – Metabolic correlates to SLC2A9 and ABCG2 alleles in REDS RBC Omics index cohort (n=13,091).

The rs16890979 SLC2A9 (missense mutation V282I) was strongly associated with urate levels, NAD-breakdown product nicotinamide, bile acids, ferritin and BMI in index donors, and was most prevalent in donors of Asian descent (A-B). The intronic rs45599402 ABCG2 SNP was also linked to urate levels, nicotinamide and bile acids (taurocholate), also most prevalent in donors of Asian descent (C-D).

SLC2A9 V282I alleles are associated with multi-omics alterations linked to protein quality control, complement and coagulation and elevated hemolysis

By leveraging multi-omics workflows (Figure 6), including proteomics analyses (Supplementary Figure 4; Supplementary Table 1) on top of metabolomics, we expanded the association of urate-linked genetic traits to omics data in the recalled donor cohort as a function of storage duration. This analysis identified the top 50 variables linked to r16890979 alleles (Figure 6A). The top metabolic correlates to urate (adjusted by donor sex, BMI and age) within both index and recalled donor RBCs was creatinine (Figure 6B). Significantly elevated proteins in blood units from donors with high urate were enriched for pathways involved in liposome and ribosome machinery, while pathways involved in autophagy and protein quality control via heat shock proteins, ubiquitination and proteasomal degradation were enriched in blood units with low urate (Figure 6D). Of note, elevated urate was also associated with kynurenine levels – especially in index donor samples (Figure 6.B), consistent with the association of both metabolites with male sex, donor age, BMI and osmotic fragility, perhaps driven by complement and coagulation cascades - whose components are elevated in blood units with elevated urate (Figure 6.C–D) and, as recently reported, kynurenine¹⁰. Assessment of the impact of SLC2A9 top SNPs (rs16890979 and rs3733591) and ABCG2 (rs138409370) showed a significant (p < 1e-3) association of the former (both SNPs) with elevated osmotic fragility and storage hemolysis in heterozygous or homozygous males or homozygous females, consistent with a role of sex dimorphism and genetics in end of storage urate levels and hemolytic propensity (Supplementary Figure 5; Supplementary Table 1).

Figure 6 – SLC2A9 V282I alleles are associated with multi-omics alterations linked to protein quality control, complement and coagulation.

In A, heat map of the top 50 variables (mostly proteins and metabolites) linked to r16890979 alleles in 643 REDS RBC omics donors at storage day 10, 23 and 42. In B and C, volcano plot view (x axis indicates Spearman rho, y axis -log10 of p-values) of metabolic and multi-omics (metabolomics, proteomics, lipidomics) correlates to SLC2A9 alleles. In D, pathway analysis informed by results in A-C.

Blood units from female donors carrying SLC2A9 and ABCG2 SNPs are linked to significantly lower post-transfusion hemoglobin increments

To determine the translational relevance of the findings reported above, we interrogated the REDS-III vein-to-vein database to determine whether the genetic traits associated with urate levels also influenced post-transfusion outcomes of these RBC units. Specifically, we identified donors carrying SLC2A9 and ABCG2 relevant alleles, and investigated whether carrying 2, 1 or zero copies of those alleles was associated with altered hemoglobin increments at 24h upon transfusion in single unit recipients of packed RBCs from these donors. Interestingly, in keeping with a sex dimorphism in urate levels, a dose effect in the drop of hemoglobin increments (i.e., up to 31% and 25% loss of potency) was observed when units from female donors – but not males - carrying two alleles were transfused (Figure 7; Supplementary Table 1).

Figure 7 –

Blood units from female donors carrying SLC2A9 and ABCG2 SNPs are linked to a significant drop in post-transfusion hemoglobin increments in single unit transfusion recipients in the Kaiser Permanente vein-to-vein database. Adjusted delta hemoglobin (g/dl median and ranges are shown in the violin plot and tables) for donors as a function of SNP allele counts (0, 1, 2). Number of donors (“count” column in the tables) are reported for each sub-group.

DISCUSSION

In the present study, we focused on RBC urate levels as a function of donor demographics and genetics in over 13 thousand volunteers from the REDS RBC Omics study. We found that RBC urate levels are highest in older male donors, and donors with BMI higher than 40 kg/m². These findings are consistent with established literature on sex dimorphism in urate metabolism, generally non-pathologic, despite hyperuricemia most commonly found in males, who are also more likely to develop associated comorbidities such as kidney stones, cardiovascular diseases, and gout.³⁹ Indeed, the global prevalence of gout in 2020 was 3.26 times higher in males than in females and increased with age.⁴⁰ Multiple studies have documented a strong association between obesity and an increased risk of hyperuricemia and gout,^4,41 since weight is a determinant of serum urate levels and weight loss can mitigate the risk of developing gout.^4,42,43

Previous genome wide association studies (GWAS) in the literature have identified associations between circulating urate levels and common SNPs in regions coding for the solute carrier transporter SLC2A9 (also known as glucose transporter GLUT9)^43–46 and the high-capacity kidney urate transporter ATP-binding cassette group 2 (ABCG2)^43,47,48. By focusing on RBC storage biology, we combined metabolomics analysis with demographics and genomics data to investigate biological and genetic factors underpinning heterogeneity in end of storage RBC urate levels across donors. We report data on urate level changes during storage and reproducibility across multiple donations from the same 643 donors over a 6 to 12 month period, and ultimately identify genetic factors that are linked in this cohort to urate levels in stored blood, especially in donors of East Asian and Hispanic descent. Urate levels – or genetic traits associated with them - were also associated with higher hemolysis in vitro (spontaneous and after osmotic stress, but not oxidative stress) and transfusion efficacy – as gleaned by measuring hemoglobin increment through a vein-to-vein database - in thousands of critically ill recipients of single unit transfusion events of blood units donated by REDS donors. These data are consistent with reported minor allele frequencies for the ABCG2 rs2231142 SNP, which encodes the missense mutation Q141K. While here observed to be slightly less significantly linked to urate levels than the intronic SNP, this risk allele is common among several populations ranging from 3% in populations of African ancestry, 11% in those of European ancestry, and up to 31% in East Asian populations.⁴⁹ Sex dimorphism in urate responses is also, at least in part, explained by the established role of estrogen in the downregulation of SLC2A9 at the post-transcriptional level through ER-β induced proteasomal activation.⁵⁰ SLC2A9 knockout in mice of both biological sexes are associated with severe hyperuricemia, consistent with a role of this transporter in circulating urate homeostasis,⁵¹ while the missense V282I SLC2A9 mutation has already been reported in association with circulating urate,⁵² but not in RBCs. The other most common genetic contributor to hyperuricemia in GWAS studies, SLC22A12 or URAT1, was not identified in this study, suggesting that subjects carrying polymorphisms in the regions coding for this gene may not be sufficiently healthy to donate blood. Indeed, the frequency of the minor allele variant for URAT1 was lower in the REDS population compared to the general population.

By focusing on RBC biology, in light of the translational relevance to RBC hemolytic disorders and transfusion medicine, the findings reported here identify several elements of novelty. First, the data presented in this study could inform development of novel storage strategies by consideration of the relevance of urate biology to the optimization of blood transfusion protocols. Indeed, while urate is a potent antioxidant and its supplementation to current storage additives may be beneficial to mitigating the storage lesion¹⁶, the chemical reactions that generates urate in vivo in the blood donor prior to donation concomitantly produces hydrogen peroxide, thus revealing a higher basal level of oxidant stress in RBCs from donors with high urate at the time of donation^14,53, but perhaps also an increased activation of basal antioxidant pathways early on during storage¹⁵. Given the altruistic nature of blood donation, molecular testing of blood units may also serve a secondary public purpose as an indicator of population health through molecular epidemiology studies. While blood donors have historically been regarded as a uniquely healthy population (healthy donor effect), our findings add to the growing list of molecular epidemiology data on the blood donor population, suggesting that presently healthy volunteers may carry genetic factors and traits associated with genetic ancestries that, combined with relevant biological determinants of specific phenotypes (e.g., male sex, older age, higher BMI for urate in this study) may predispose donors to the onset of morbidities that could be detrimental to their health, other than prevent them from donating in the future. Therefore, our findings could inform donor recruitment and retention campaigns – rather than deferring current donors from donation, given the ongoing and aggravating global crisis with blood inventories - by providing feedback to the altruistic volunteers on the molecular make up of donated blood for traits relevant to their health status (e.g., urate biology and relevant associated genetic traits).