Putting a finger on the switch

Abstract

The transition from fetal to adult β-like globin expression is a key step in the maturation of the red blood cell lineage. Two new studies show that the KLF1 zinc finger protein uses direct and indirect means to regulate the final switch from fetal to adult globin expression and that monoallelic loss of KLF1 expression leads to persistence of fetal hemoglobin.

The study of red blood cell biology has provided fertile ground for classic cellular, biochemical, genetic and molecular studies. Deciphering how developmental changes in hemoglobin-chain expression are controlled has been of particular interest¹, driven not only by curiosity about basic mechanisms but also by the desire to alleviate morbidity associated with hemoglobinopathies, the most common single-gene disorders worldwide². On pages 801 and 742 of this issue^3,4, two new studies demonstrate a key role for the erythroid Krüppel-like factor KLF1 (also called EKLF) in regulating the developmental switch between fetal and adult hemoglobin expression. The studies also uncover a regulatory relationship between KLF1 and a second transcription factor, BCL11A, which helps explain how KLF1 orchestrates this switch (Fig. 1).

Figure 1.

KLF1 regulates globin switching. During embryonic and fetal development or in KLF1-haploinsufficient adults (left), KLF1 levels are low, resulting in low levels of adult β-globin and BCL11A and high levels of γ-globin. In adults with two functional copies of KLF1 (right), increased expression of KLF1 in definitive red blood cells promotes high levels of adult β-globin and BCL11A expression, which in turn represses γ-globin expression.

Multiple threads

Studies of individuals with thalassemia have provided insights into the mechanisms underlying developmental changes in globin gene expression. Among these are a subset of individuals whose clinical symptoms of β-thalassemia (deficient adult β-globin expression) are at least partly alleviated by elevated expression of fetal γ-globin, a β-like globin whose high level of expression at the fetal stage is normally attenuated shortly after birth; such individuals show a hereditary persistence of fetal hemoglobin (HPFH).

Unraveling of the mechanism of β-like globin switching was aided by the identification of KLF1, a zinc-finger protein with the proper credentials to function as a “switching factor”^5,6. KLF1 preferentially binds to a critical promoter element (CACCC) of the gene encoding adult β-globin (HBB), and its ablation in mice causes embryonic lethality due to profound β-thalassemia. Studies in Klf1-null mice support a role for this protein in switching, as embryonic βh1-globin and transgenic human fetal γ-globin genes are not downregulated properly in these mice. Humans with mutations in the CACCC element of HBB have β-thalassemia, and many also have greatly elevated γ-globin expression levels. Although later studies have illuminated how KLF1 plays a positive role in modifying chromatin and activating transcription at the adult β-globin gene, its role in attenuating embryonic and/or fetal β-like globin genes has been less apparent. The paucity of individuals or even murine models with hematologic disease caused by mutations in KLF1 has deepened this quandary.

Exceptions to this have occurred quite recently. The human In(Lu) phenotype, in which expression of the Lutheran blood-group antigen is markedly downregulated in red blood cells, was mapped to mutations in KLF1 that generate a nonfunctional allele⁷. Also, the mouse severe anemia mutant Nan was mapped to a missense mutation in Klf1 that affects the second zinc finger, resulting in a protein with altered specificity⁸. In Nan mice, the expression output of red blood cells is distorted, causing dramatic phenotypic changes. Notably, β-like globin switching is radically affected by the Nan mutation, as expression of embryonic βh1-globin persists even in adult Nan/+ mice.

Given the variability in human γ-globin persistence, efforts have been aimed at mapping modifier loci in individuals with HPFH who do not have probable causative mutations within the β-globin locus on chromosome 11. Such studies have identified SNPs at two loci associated with high γ-globin levels⁹: HBS1L-MYB on chromosome 6 and BCL11A on chromosome 2. Further studies in mouse and human cells have firmly established that BCL11A plays a critical role as a repressor of γ-globin expression in adult red blood cells and is thus another key “switching factor”¹⁰. However, these loci do not fully account for the known variation in γ-globin levels. Moreover, a link between the two important switching factors KLF1 and BCL11A has not been apparent.

Common ties

The new studies by Borg et al.³ and Zhou et al.⁴ link HPFH, KLF1 and BCL11A together in a unified model. In one study, Borg et al.³ performed linkage analysis to determine the cause of HPFH in a large Maltese family. After defining a candidate region on chromosome 19, they identified a heterozygous truncating mutation in KLF1 that segregated with the HPFH phenotype in this family. Next, they examined samples from affected individuals and KLF1 knockdown cells and found that KLF1 levels were directly proportional to BCL11A levels and inversely proportional to γ-globin levels. In addition, they showed that human KLF1 binds to the BCL11A gene in vivo. Interestingly, they could not detect a truncated KLF1 protein in the human samples, suggesting that the HPFH phenotype results from KLF1 haploinsufficiency. This model is supported by experiments showing that the wild-type condition (increased endogenous BCL11A expression and lowered γ-globin expression) is re-established by exogenous expression of full-length KLF1 in the HPFH samples.

In an independent study, Zhou et al.⁴ generated a Klf1 knockdown mouse model, which yielded the expected increase in (transgenic) γ-globin expression. Consistent with the findings of Borg et al.³, they found that Bcl11a levels were dramatically downregulated in the knockdown mice. They also determined that mouse Klf1 binds an upstream Bcl11a element in vivo and that shRNA-mediated knockdown of KLF1 in human cells causes a decrease in BCL11A levels and an increase in γ-globin levels, similar to the changes seen in the mouse knockdown model.

The results from these two studies support three conclusions. First, HPFH can result from KLF1 haploinsufficiency. Second, KLF1 indirectly regulates γ-globin expression by directly regulating BCL11A. Third, subtle variation in KLF1 levels can lead to key changes in gene expression output. Zhou et al.⁴ noted that levels of Klf1 are increased approximately threefold in mouse adult definitive erythroid cells compared to embryonic primitive erythroid cells and suggested that this contributes to the developmental changes in β-like globin expression. Together, these results suggest that KLF1 has a critical role in regulating globin switching—both in activating β-globin and in repressing γ-globin—and tie the two major globin switching proteins in an interactive model (Fig. 1).

More to come?

These studies raise a number of intriguing questions. First, Townes and colleagues previously showed (and now further support in their present study) that tagged Klf1 interacts directly in vivo with the mouse embryonic β-like globin gene promoter only in primitive cells and then switches to bind the adult β-globin promoter only in definitive cells¹¹. However, Klf1 is not necessary for expression of embryonic or fetal globin, so does its interaction with the embryonic/fetal promoter simply play a role in establishing its most optimal transcriptional activity before the final switch? How does the eightfold lower in vitro affinity of KLF1 for the γ- versus β-CACCC element¹² play into this scenario? Might it enable a quick release when the β-site becomes available? Also, if KLF1 knockdown is used to increase γ-globin expression in a therapeutic setting, as suggested by both studies, will that compromise β-globin expression, or is that actually a desirable outcome in individuals with sickle cell anemia but not β-thalassemia? Finally, do changes in KLF1 modification status, which are known to affect subsequent protein interactions, impinge on its role in switching? How does the Nan mutation fit into these schemes? Clearly there are more chapters to be completed in the switching story.