Congenital dyserythropoietic anemia type II in human patients is not due to mutations in the erythroid anion exchanger 1

Type II congenital dyserythropoietic anemia (CDA-II or HEMPAS) is an autosomal recessive disorder, representing the most frequent form of congenital dyserythropoiesis.

Recently, the molecular basis of the retsina (ret) phenotype that resulted in zebrafish from mutations in the gene that encodes the erythroid anion exchanger 1 (AE1, also called SLC4A1) was reported.¹  The high number of binucleated erythroblasts, the presence of “double membranes,” and the reduction in posttranslational glycosylation of AE1, observed in the ret fish, are all reminiscent of the human CDA-II. Since the gene responsible for CDA-II in humans has not yet been identified, it is highly relevant to ask whether it could be AE1.

In first approximation, this does not seem likely for several reasons. First, as the authors themselves note, in a majority of families with CDA-II that have been subjected to linkage analysis, the disease maps to 20q11.2,²  whereas AE1 maps to 17q21-q22. Second, complete inactivation of AE1 in mice³  and in cattle⁴  causes severe hemolytic anemia, but not the CDA-II phenotype. Third, human erythroid AE1 absence has been found to cause severe hereditary spherocytosis (HS) but, again, no CDA-II.⁵ 

In order to test whether AE1 mutations might be responsible for a subset of patients with CDA-II we have used the resources of an international registry⁶  in which 108 patients are enrolled. In 14 of these patients, belonging to 7 unrelated families, the disease was not linked to 20q11.2 (Figure 1A).⁷  Of these 14 patients, 6 had a particularly severe form of CDA-II (they were transfusion dependent from the age of 1 year); the remaining 8 patients had “typical” CDA-II. In 5 families we were able to exclude linkage to chromosome 17. In addition, we proceeded in all 14 patients to quantitate erythrocyte AE1 by means of sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).^8,9  Whereas in patients with HS the amount of AE1 was significantly reduced (73.90 ± 2.42% of controls, P < .0001), in patients with CDA-II the amount of AE1 was virtually normal (96.50 ± 2.07% of controls, P = .783). In terms of the possible pathophysiologic significance of AE1 in CDA-II, while all these patients had the underglycosylation of AE1 that is characteristic of this condition, we did not find any difference in AE1 levels between patients with severe CDA-II versus “typical” CDA-II (96.83 ± 2.48% versus 96.25 ± 1.83%, respectively; P = .621) (Figure 1B).¹⁰ 

Based on genetic analysis and on biochemical findings, we suggest that in at least the vast majority of cases human CDA-II is not due to AE1 mutations. It is intriguing that, in zebrafish, AE1, through its protein 4.1R-binding domain, was found to affect the poles of the mitotic spindle specifically in erythroid cells; as a result, AE1 deficiency ultimately affects the completion of chromosome segregation in these cells, thus mimicking certain morphologic features of CDA-II. Therefore, in the rare patients in whom linkage to AE1 has not been ruled out, it will be important to look specifically for mutations within this particular domain of human AE1. At any rate, since these cases are a minority, there must be in humans at least one other gene involved in this pathway.

We thank the families for their enthusiastic cooperation and hospitality, and Prof Narla Mohandas for stimulating discussions.