PerspectiveDevelopmental Biology

From Genetic Association to Genetic Switch

See allHide authors and affiliations

Science  19 Dec 2008:
Vol. 322, Issue 5909, pp. 1803-1804
DOI: 10.1126/science.1169216

Deciphering the sequence of the human genome and the subsequent cataloging of common human DNA sequence variation marked a paradigm shift in human genetics. These resources, together with advances in cost-effective genotyping technologies, enabled the design of genome-wide association studies for the unbiased discovery of commonly occurring DNA sequence variations called single-nucleotide polymorphisms (SNPs) that are preferentially associated with a disease or other clinical trait (1). Although genome-wide association studies have uncovered disease-associated SNPs, identifying actual disease-causing variants—and gaining deep insights into how those variants generate the underlying molecular pathophysiology—have so far yielded only modest results. This has led to criticisms of the genome-wide association approach for investigating the etiologies of common diseases (2). However, this assessment may be premature. On page 1839 of this issue, Sankaran et al. (3) show how genome-wide association findings can lead to a detailed understanding of disease mechanisms and be used to ascertain novel therapeutic targets.

Previous genome-wide association studies conducted in independent populations have identified SNPs in three chromosomal loci that are associated with varying expression levels of human fetal hemoglobin (HbF) (4-6). HbF is a clinically important quantitative trait because elevated concentrations reduce the severity of sickle cell disease and β-thalassemia, disorders caused by different mutations in the human β-globin gene (7). Normally, HbF predominates in the fetus but declines to very low amounts postnatally due to repression and activation of the γ-globin and β-globin genes, respectively. The “switch” that controls these reciprocal changes in globin gene expression has been intensively investigated, but the molecular basis of this developmental process remains largely unknown. As reported by Sankaran et al., functional studies motivated by recent HbF genome-wide association findings have provided a major breakthrough in understanding the hemoglobin switching problem.

One of the SNPs associated with elevated HbF expression is found in an intron (noncoding region) of the BCL11A gene on human chromosome 2 (see the figure). BCL11A encodes a protein that represses transcription in the B lymphoid lineage (8).

Sankaran et al. hypothesized that BCL11A might repress expression of the γ-globin gene, with expression or activity of this repressor correlating inversely with HbF production both during normal development and in individuals of different genotypes at the BCL11A locus. They first determined that BCL11A is expressed as two long isoforms, encoded by alternatively spliced messenger RNAs (mRNAs), in primary adult erythroblasts. By contrast, only shorter variants of BCL11A are found in human embryonic erythroleukemia cells and in primary human fetal liver cells, both of which express high amounts of HbF. Moreover, the genotype at the BCL11A SNP that affects HbF production influences expression of mRNAs encoding the long isoforms in lymphoblastoid cell lines: High expression of BCL11AmRNA corresponds to homozygosity for the allele associated with low HbF production; low mRNA expression corresponds to homozygosity for the allele associated with high HbF production; and SNP heterozygotes express intermediate amounts of mRNA (see the figure). If the association between the BCL11A SNP and the expression level of this gene in lymphoblastoid cell lines also applies to erythroid cells, then the causative SNP either has a non-tissue-specific effect on transcription, or it acts at the posttranscriptional level, an issue that remains to be resolved.

From SNP to mechanism and potential therapy.

Hemoglobin genes on human chromosome 11 are differentially expressed in the embryo, fetus, and adult. A SNP in the BCL11A gene is associated with varying amounts of fetal hemoglobin in human populations. Because there is an inverse correlation between BCL11A and fetal hemoglobin expression, inhibiting BCL11A is a potential therapy for adult hemoglobinopathies. LCR HSS: locus control region hypersensitive sites. The asterisk indicates a DNA region that, when deleted, is associated with increased fetal hemoglobin production.

Although the results are consistent with the long isoforms of BCL11A functioning to suppress HbF production in a dose-dependent manner, they do not distinguish between a direct effect on γ-globin gene expression and an indirect effect on the kinetics of erythropoiesis that might cause cells containing HbF to be overproduced (9). Additional findings of Sankaran et al. argue strongly in favor of the former model. For example, BCL11A was found to interact with multiple components of the nucleosome remodeling and histone deacetylase (NuRD) complex, which functions in transcriptional repression (10), as well as with GATA-1 and FOG-1, the major activating transcription factor and cofactor, respectively, of the erythroid lineage (11). Further, RNA interference specific for BCL11A in primary erythroid progenitors increased γ-globin mRNA expression and HbF production without inducing a generalized effect on cellular differentiation or altering expression of known globin gene transcription factors. Finally, chromatin immunoprecipitation revealed that BCL11A binds to multiple regions within the human β-globin locus that control silencing of the γ-globin gene (7).

Thus, BCL11A plays an important role in fetal-to-adult hemoglobin switching during normal erythropoiesis, and its expression in adult erythroid cells affects the amount of HbF produced. This raises questions about what controls the stage-specific shift between the long and short BCL11A isoforms, and the molecular mechanism by which BCL11A represses γ-globin gene expression. It is also not clear which genetic variant in BCL11A sets the expression level of this gene, nor how it functions in this capacity. Higher-resolution studies of BCL11A chromatin occupancy, functional characterization of the putative cis-regulatory elements containing these binding sites, and sequencing of the entire region of BCL11A that is in linkage disequilibrium with the SNP used to discover its relevance to HbF expression are needed to address these questions.

The findings of Sankaran et al. also have potential consequences for developing new therapies to treat sickle cell disease and other hemoglobinopathies. The inverse correlation between BCL11A and HbF expression, combined with the known ameliorative effect of HbF on the pathophysiology of sickle cell disease and β-thalassemia, suggests that inhibition of BCL11A expression or function could be an effective treatment for these disorders. The study also illustrates that, when experiments are appropriately designed, the initial findings of genome-wide association studies can be successfully followed up at a functional level.


Navigate This Article