Lack of the nucleoside transporter ENT1 results in the Augustine-null blood type and ectopic mineralization

The first paper referring to the At^a blood group antigen, which can be responsible for severe hemolytic transfusion reactions¹  and mild hemolytic disease of the newborn,²  dates back to the 1960s. Applewhaite et al³  identified an antibody with a novel specificity in the serum of Mrs Augustine, when the red cells of her third child gave a positive direct antiglobulin reaction at birth. Her alloantibody, abbreviated as anti-At^a after her name, reacted with >6600 blood donors tested at that time, indicating that At^a was a high-frequency antigen. Since then, many other At(a−) individuals have been identified,^4,5  usually after they (or one of their family members) produced an anti-At^a. All anti-At^a producers reported thus far were of African ancestry like Mrs Augustine.

Twenty years ago, the red cells of a French woman of European ancestry were found to be negative for the At^a antigen (G.D., unpublished data, 1994; supplemental Methods available on the Blood Web site). Stimulated by pregnancy, this woman had developed a potent alloantibody directed against an antigen that was present on all tested red cells but those of her siblings (Figure 1A). However, her alloantibody was not anti-At^a because it reacted with At(a−) red cells also. Although its specificity remained unsolved at that time, G.D. suggested that this woman might have the Augustine-null phenotype, in which her red cells lack all the antigens belonging to the yet-to-be discovered Augustine blood group system, including the At^a antigen. Encouraged by our recent success in elucidating the molecular bases of 3 other long-sought-after blood group antigens,^6-8  we decided to reinvestigate the case of this French woman.

The 1994 study was performed within the ethics guidelines of the International Blood Group Reference Laboratory (Bristol, United Kingdom) and the study performed since 2012 was approved by the Ethical Committee of the French National Institute of Blood Transfusion (Paris, France).

Red cells of blood donors were incubated with the proband’s serum in low-ionic strength solution. Membranes of sensitized red cells were prepared by hypotonic lysis and then solubilized in a triple detergent buffer. Immune complexes were isolated using immobilized protein A/G, resolved by polyacrylamide gel electrophoresis, and studied by silver staining before mass spectrometry-based protein identification. Details can be found in supplemental Methods.

Red cell membrane extracts were probed with a rabbit affinity-purified anti-ENT1 (Applied Biological Materials, Richmond, BC, Canada) and the mouse monoclonal anti-ABCG2 BXP21 (Santa Cruz Biotechnology, Dallas, TX). Details can be found in supplemental Methods.

The primers used to amplify and sequence SLC29A1 are described in supplemental Table 1. Details can be found in supplemental Methods.

To identify the potential protein targeted by the alloantibody of the aforementioned French proband, the immune complex that is formed with the red cells of random blood donors was purified and analyzed by mass spectrometry as previously described.⁶  Three tryptic peptides encoded by SLC29A1 were unambiguously identified (LEGPGEQETK, WLPSLVLAR, and YYQQLK). To understand why the proband developed an alloantibody to the product of SLC29A1, we first analyzed this gene in the proband and controls. Sequencing of SLC29A1 revealed that the proband was homozygous for a G-to-C transversion at the +1 position of intron 6-7 (c.589+1G>C in NM_004955; Figure 1B-C). This splice mutation was absent from all sequenced controls and public databases (National Center for Biotechnology Information Single Nucleotide Polymorphism Database, National Heart, Lung, and Blood Institute Grand Opportunity Exome Sequencing Project, and 1000 Genomes Project). In contrast, c.589+1G>C was found in the heterozygous state in both parents of the proband and in the homozygous state in her siblings (Figure 1B). A genealogy search revealed that the parents were first cousins once removed (Figure 1A), which provides an explanation for the sharing of this rare mutation.

SLC29A1 encodes the equilibrative nucleoside transporter 1 (ENT1; also called SLC29A1), which enables facilitated diffusion of hydrophilic nucleosides across the plasma membrane.⁹  Of note, SLC29A1 is highly expressed in early erythroid cells (supplemental Figure 1), and ENT1 is particularly abundant in the red cell membrane.¹⁰  From a therapeutic point of view, ENT1 plays an important role in the cellular uptake of nucleoside analog drugs, such as ribavirin, which is widely used in the treatment of hepatitis C.^11,12  To determine the effect of the c.589+1G>C mutation, we set up the analysis of ENT1 expression by immunoblotting red cell membrane extracts (supplemental Figure 2). As shown in Figure 1D, we detected no ENT1 protein in the proband and hence concluded that c.589+1G>C defines a null allele of SLC29A1, explaining how the proband could have developed an alloantibody to ENT1.

As homozygotes for a null allele of SLC29A1, the proband and her siblings are the first reported individuals lacking the complete expression of ENT1. They showed no gross developmental abnormities and had a normal lifestyle, at least until age 45 to 50 years, when they were contacted for this study. However, strikingly, all experienced frequent attacks of pseudogout since they were 18 to 20 years old (mainly in the hands but also in the feet, elbow, and knee), which was not the case for their parents and children. Hand x-rays indicated multiple small calcifications around the joints (Figure 1E; supplemental Figure 3). Ectopic calcification/mineralization was also observed in the hips, pubic symphysis, and lumbar discs of the proband at age 47 (supplemental Figures 4 and 5). Serum levels of urea, uric acid, creatinine, and electrolytes were in the normal range (supplemental Table 2). Although it is virtually impossible to prove that ENT1 deficiency is solely responsible for the ectopic mineralization observed in the proband and her siblings, recent data with Slc29a1^−/− mice fully support this hypothesis.^13,14  The precise role of ENT1 in bone homeostasis remains to be established in mice and humans, but it is most likely linked to its ability to transport adenosine, which has emerged as a key regulator of bone metabolism.¹⁵  In fact, other genes of the adenosine pathway have been found to be responsible for calcification disorders.^16,17 

As mentioned above, it was suggested that the proband did not express the protein carrying the At^a antigen. We therefore analyzed ENT1/SLC29A1 in At(a−) people. In line with the reactivity of the proband’s antibody with At(a−) red cells, ENT1 was found to be equally present in At(a+) and At(a−) red cells (Figure 2A). Sequencing of SLC29A1 in 5 At(a−) individuals revealed that all were homozygous for a G-to-A transition in exon 12 (c.1171G>A in NM_004955; Figure 2B) encoding the amino acid change Glu391Lys. Glutamine 391 is located in the fifth extracellular loop of the ENT1 transporter¹⁸  (Figure 2C), and the nonconservative substitution p.Glu391Lys likely disrupts the epitope of anti-At^a. Nevertheless, p.Glu391Lys was shown not to alter the activity of the ENT1 transporter¹⁹  so that those with At(a−), in contrast to the 3 siblings reported above, have normal ENT1 function.

Interestingly, c.1171G>A in SLC29A1 corresponds to the minor allele of rs45458701 that has only been detected in people of African ancestry (supplemental Table 3) like the At(a−) blood type. In the United States, the NHLBI GO Exome Sequencing Project found 49 carriers of the A allele of rs45458701 by sequencing 2203 African Americans and none of 4300 European Americans. As a Taq^α1 restriction site is absent in the A allele of rs45458701 (Figure 2B), we developed a polymerase chain reaction-restriction fragment length polymorphism assay to detect it (supplemental Figure 6). We thus analyzed 8 reference anti-At^a sera, including Mrs Augustine’s serum, as well as the family of a British patient who was recently identified as At(a−) (Figure 2D-E). In all these cases (of African ancestry), we also found the At(a−) blood type associated with homozygosity for the A allele of rs45458701.

In conclusion, we provide evidence that allelic variations in SLC29A1 underlie a new blood group system. Typing rs45458701 (c.1171G>A in SLC29A1) could be implemented in blood group genotyping platforms to identify people with the At(a−) blood type. Outside transfusion medicine, we recommend looking for SCL29A1 loss-of-function mutations (or alternatively for the Augustine-null blood type) in patients with periarticular calcifications of unknown cause.

The authors thank the proband and her siblings for generously contributing time and material to this study. The authors also thank J. Poole, P.-Y. Le Pennec, M. Nagler, A. Brownell, D. Gien, P. Martin, G. Nicolas, M. Bonny, and M. Le Gall for their help and assistance.

Mass spectrometry analysis was funded by the Vermont Genetics Network (National Institutes of Health National Institute of General Medical Sciences/IDeA Networks of Biomedical Research Excellence grant 8P20GM103449) (to B.A.B.).

Contribution: G.D., B.A.B., V.H., C.S., S.G., and L.A. performed experiments and analyzed data; G.D., B.A.B., L.M., H.H., E.L., and T.P. contributed essential material; G.D., B.A.B, J.-P.C., and T.P. discussed the results and commented on the manuscript; L.A. designed and supervised the study, made the figures, and wrote the manuscript; and all authors edited the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Lionel Arnaud, National Institute of Blood Transfusion, 6 rue Alexandre Cabanel, 75015 Paris, France; e-mail: lionel.arnaud.contact@gmail.com.