Genetic complexity in sickle cell disease

Sickle cell disease (SCD) was the first human monogenic disorder to be characterized at the molecular level (1). It results from the substitution of glutamic acid by valine at position 6 of the β-chain of hemoglobin. The clinical manifestations of SCD arise from the tendency of sickle hemoglobin (known as HbS or α₂β^S₂) to polymerize at reduced oxygen tensions and deform red cells into the characteristic rigid sickle cell shape. Such inflexible red cells cannot pass through the microcirculation efficiently, and this results in anemia (due to destruction of the red cells) and intermittent vasoocclusion causing tissue damage and pain (2). Although all patients with homozygous SCD have exactly the same molecular defect, there is considerable clinical variation, ranging from death in early childhood (3) to a normal life span with few complications (4). Genetic modifiers of SCD include α-thalassemia (5), and it has been known for many years that patients with increased levels of fetal Hb (HbF or α2γ2) often tend to have a relatively mild clinical course (2) because HbF reduces the tendency of HbS to polymerise within the red cell. Increased HbF may result from rare deletions within the β-globin gene cluster or from point mutations in the promoters of the fetal γ-globin genes (hereditary persistence of fetal hemoglobin, HPFH), but additional loci are known to increase HbF levels in adult life. Identifying such loci has been a painstaking task, but a combination of genome-wide analysis within a large kindred (6) and within twin pairs (7) has identified two quantitative trait loci (QTL) with major influences on fetal hemoglobin levels in adults. In this issue of PNAS, Lettre et al. (8) now show that a significant proportion of the variation in HbF levels and the frequency of painful crises in patients with SCD is accounted for by five common single-nucleotide polymorphisms (SNPs) at these loci (Fig. 1).

Fig. 1.

Locations of the five SNPs that act as QTL on the level of fetal hemoglobin in adults. They may act directly on the expression of the γ-globin genes or affect the process of erythropoiesis to increase the proportion of F cell production.

HbF is the predominant type of Hb in fetal life. After birth, the γ-gene is switched down and the β-gene is switched on so that adults mainly produce HbA (α₂β₂). After this developmental switch, low levels of HbF are still produced, and this is distributed heterogeneously, some red cells (F cells) expressing more than others (9). The level of HbF (and the associated proportion of F cells) is a highly heritable trait (10), and within a normal population the distribution of HbF is very skewed. Most normal individuals produce <0.6% HbF distributed among 1–7% F cells, but a small proportion (≈2%) produce up to 5% HbF and 25% F cells. These individuals are said to have heterocellular hereditary persistence of fetal Hb (hHPFH). Genetic studies extending back over 25 years have investigated families that include normal, nonanemic individuals with hHPFH and individuals with SCD. In these families, the patients who have SCD tend to have higher levels of HbF than usually seen in such patients and a less severe clinical course.

For many years it has been known that hHPFH may be associated with an SNP (rs7482144 or Xmn-^Gγ) in cis to the β-globin (HBB) locus on chromosome 11 (11). Normal individuals with the T allele have a barely detectable increased level of HbF (12), but this allele confers a greater tendency to increase the level of HbF and F cells under conditions of erythroid stress (e.g., SCD and β-thalassemia) in which there is accelerated erythropoiesis. More recently, it has been shown that SNPs at two loci unlinked to the β-locus also have significant effects on the levels of HbF. The first locus on chromosome 6 contains five genes (including the hematopoietic transcription factor MYB) with the most significantly associated SNPs lying between a gene (HBS1L) of unknown function and the oncogene MYB (6). The second locus lies on chromosome 2 with associated SNPs lying within the oncogene BCL11A (7, 13). Together, the pattern of inheritance of common variants at these three QTL explain 44% of the total variance in HbF and F cells in normal, nonanemic populations (7). Clearly, common and rare variants at other QTL influencing the expression HbF remain to be discovered.

By studying three independent cohorts of patients with SCD, the report from Lettre et al. (8), together with that from Uda et al. (13), has shown that the same three loci (HBB, HBS1L-MYB, and BCL11A) that control HbF and F cells in normal individuals are also important determinants of HbF in patients with SCD. QTL at the three loci that modify the levels of HbF in SCD explain a remarkably large proportion (20%) of the variance, compared with the combined effects of common variants at QTL associated with complex diseases (14).

To identify the mechanisms by which common variants ultimately influence phenotype involves not only identifying the gene(s) whose expression they alter but also determining how they alter expression and the effects of this on cell fate. To date, there have been few successes in this area of genetic research. The single-nucleotide change (rs7482144 or Xmn-^Gγ) in the promoter of the γ-globin gene could be explained by a direct effect on γ-globin gene expression, but despite extensive analysis of this SNP, it is still unclear whether this is the functional SNP or a haplotypic marker. The SNP that correlates most strongly with HbF expression in patients with SCD (rs4671393) lies in the intron of an oncogene, BCL11A, that is expressed in erythroid precursors and hence is a good candidate for the effector of this variant but requires functional analysis. The three SNPs (rs28384513, rs9399137, and rs4895441) that lie in the intergenic region between HBS1L and MYB have independent effects on HbF variance in SCD. Although the role of HBS1L is unknown, MYB is known to play an important role in normal erythropoiesis and again is a good candidate as the gene responsible for the effect of the variant.

Although the key genes associated with these QTL are not yet known, two plausible mechanisms of action can be proposed. The first is that one or more of these variants directly affect γ-globin gene expression, increasing the amount of HbF produced per cell. An alternative is that they alter the kinetics of erythropoiesis, mimicking some aspect of stress erythropoiesis and resulting in accelerated differentiation (15). Because globin gene expression in erythroid cells is asynchronous during maturation, this may result in the release from the bone marrow of red cells expressing more fetal Hb than normal. Increasing the number of F cells in this way will indirectly lead to an increase in the amount of HbF. Understanding the mechanism by which these QTL increase HbF will hopefully illuminate the complex interacting pathways that modify SCD, perhaps leading to new insights and ultimately new therapeutic approaches.

A second aim of QTL analysis is to improve the prediction of clinical outcome. Although the steady-state levels of HbF correlate with clinical severity in SCD, the relationship is not simple. Some patients with high levels of HbF have variable and severe disease, whereas some patients with very low levels of HbF may run a mild clinical course (16). It was interesting that, in retrospect, in these cohorts, the five SNPs in combination appeared to provide additional information (beyond the simple baseline measurement of HbF) on the frequency of painful episodes associated with sickling. This, in turn, may predict morbidity and mortality in SCD (17), but it should be remembered that these are important but preliminary studies. The environment is also known to be an important modifier, and it seems likely that there will be many other QTL to find.

As so often in human genetics, disorders of hemoglobin provide important models for establishing new principles. Sickle cell disease was initially considered to be a simple condition with a monotonous genotype. However, we now know that to fully understand the pathophysiology and phenotypes of sickle cell disease, as the study of Lettre et al. (8) shows, it now has to be considered as a complex multigenic disorder.