Congenital Anemia Phenotypes due to KLF1 mutations

To the Editors,

We read with interest the article published online in your journal by Belgemen-Ozer et al., entitled “A very rare CDA-type IV in a patient with a novel mutation in the KLF1 Gene”. This is indeed another interesting case of severe congenital anemia due to compound heterozygosity for KLF1 mutations. It is a valuable addition to the literature and highlights the clinical heterogeneity of presentation of such cases. Since KLF1 mutations are quite prevalent in some ethic groups [1], it is becoming increasingly important to screen for them as part of the workup of cases of unexplained direct anti-globin test (DAT)-negative congenital hemolytic anemia.

Although we agree the reported case has some features of congenital dyserythropoietic anemia (CDA) with a high number of circulating nucleated red blood cells (n-RBCs) in the blood, the original case described by Arnaud et al. was a dominant disease with a specific mutation in a DNA-contacting residue within zinc finger 2 of KLF1 (i.e. p.E325K) (OMIM 613673) [2]. The second allele of KLF1 was wild type in that case. Subsequent reported additional cases of dominant CDA-IV have the identical heterozygous mutation and a dominant phenotype characterized by hemolytic anemia and extensive circulating (n-RBCs) [3, 4]. Recently, two groups have independently solved the puzzle as to why the p.E325K mutation leads to a dominant disease whereas all other KLF1 gene mutations described so far lead to a recessive disease. Both groups demonstrated KLF1-E325K confers aberrant DNA-binding specificity to the mutant KLF1 protein that results in ectopic binding across the genome of erythroid progenitor cells and ectopic expression of genes [5–7]. Thus, this mutant KLF1 protein has a ‘neomorphic’ function, which explains why it acts in a dominant fashion.

This situation has been described before in the context of a diagnosis of NSHA [8, 9], but this recent case is important because the phenotype is more severe with more pronounced elevation of n-RBCs resembling CDA. Interestingly, the paper by Viprakasit described two cases with the same p.R301H mutation. One was present in the context of additional thalassemia mutations so it was not easy to resolve the contribution of the R301H mutation to the phenotype. We believe Belgemen-Ozer et al. have slightly misinterpreted the conclusions made by Viprakasit et al. with respect to the proposed mechanism of hemolysis. Although pyruvate kinase (PK) levels are often low in these patients, and some of the abnormally shaped cells resembled ‘prickle cells’ as seen in genetic PK deficiency, this is because KLF1 directly binds to the PKLR gene promoter to activate its expression. Viprakasit did not conclude the disease is PK deficiency. Rather, it somewhat resembles PK deficiency. There are a large number of critical KLF1 target genes, which encode transmembrane proteins, cytoskeletal proteins and enzymes. Loss of expression of many of these likely contributes to the RBC fragility in these cases [10, 11]. We think the case presented by Belgemen-Ozer is a severe form of DAT-negative hemolysis with high n-RBCs, so it resembles CDA. In truth, there is likely as spectrum of disease phenotypes from mild NSHA to CDA. There could also be failed expression of critical KLF1 target genes that are directly involved in enucleation in these patients rather than a simple response to hemolysis.

The case reported by Belgemen-Ozer et al. proves the pR310H mutation is pathogenic. This residue is within the zinc finger domain but is not involved in direct co-ordination of the zinc ion or involved directly in DNA-binding. It is highly conserved in all members of the KLF family [12], unlike neighbouring amino acids, so it is likely to be functionally important. The case of Belgemen-Ozer et al. is less than severe than hydrops fetalis cases due to complete loss of KLF1 function [13], so the R301H mutation is a hypomorphic mutation rather than a complete loss of function mutation. There are many other missense mutations in DNA-binding domain of KLF1 ({summarized in [12]}. While many are likely to be hypomorphic mutations and lead to a similar phenotype to that described by Belgemen-Ozer and others [8, 9], some may result in neomorphic functions and therefore different phenotypes. This is most likely to be true for missense mutations that involve amino acids involved in direct interactions with DNA bases such as p.E325K and p.R328H/L. We recommend re-sequencing of KLF1 should be undertaken in cases that present with a variety of phenotypes including apparent PK deficiency, HPFH, CDA, NSHA or even milder versions of chronic hemolysis, as it is difficult to be confident of genotype-phenotype correlations for many of missense mutations within the zinc finger domain.