Genome-wide association studies of hematologic phenotypes: a window into human hematopoiesis

Abstract

The study of human hematopoiesis is often limited by the inability to manipulate this process in vivo and differences that exist between humans and commonly employed model organisms. However, human genetics provides a way to gain insight into natural variation in a variety of hematologic phenotypes and creates an opportunity to better understand hematopoiesis. In this review, we discuss how genome-wide association studies are revealing common genetic variation that is associated with hematologic traits and diseases. We discuss how the resulting insight from these studies promises to increase our understanding of human hematopoiesis and outline the challenges that lay ahead in this field.

Introduction

Hematopoiesis is among the best-understood paradigms for cell differentiation [1]. Studies in this field have traditionally been driven by the use of powerful reverse genetic approaches in a variety of model systems. However, there are important differences between clinically relevant aspects of hematopoiesis in humans and that of commonly used model organisms, including mice. For example, hemoglobin gene regulation is divergent between humans and mice [2,3]. Additionally, mutations in humans that result in specific hematologic phenotypes may not be faithfully recapitulated in mice [4-7]. Unless appropriate studies in humans are undertaken, such important differences may not be uncovered. However, the analysis of human hematopoiesis is hampered by the inability to perform in vivo manipulations, as can be readily accomplished in model organisms. Here we discuss how studies of common genetic variation in hematologic traits and diseases provide a window with which to explore aspects of human hematopoiesis in vivo. When coupled with studies in model systems, these approaches offer an opportunity to reveal new regulatory processes critical for the maintenance of the blood system.

The hematopoietic system is composed of three major cell types: red blood cells (erythrocytes) that provide oxygen carrying capacity to the circulation, platelets that are critical for hemostasis through blood clot formation, and white blood cells (primarily neutrophils, monocytes, and lymphocytes) that maintain host defense to infections [1]. Each day, a healthy human produces over 10¹¹ new red blood cells (erythrocytes) and platelets, as well as 10¹⁰ neutrophils. If this output is not maintained, the individual is at risk for a number of hematologic disorders. Because of this rapid turnover in circulating blood cells, the hematopoietic system is poised to continually replenish these lineages through the proliferation, differentiation, and maturation of immature progenitor populations that are the progeny of self-renewing stem cells (Figure 1) [1].

Figure 1.

A simplified model of the hierarchy of hematopoiesis illustrates how GWAS of hematologic phenotypes can provide insight into the differentiation of various lineages. The diagram illustrates multipotent progenitors, including the long-term hematopoietic stem cells (LT-HSC), the short term HSC (ST-HSC), the common lymphoid progenitor (CLP), the common myeloid progenitor (CMP), the granulocyte-macrophage progenitor (GMP), and the megakaryocyte-erythroid progenitor (MEP). Committed progenitors leading to the various lineages are also shown. Phenotypes examined in the GWAS studies discussed in this review are shown at the bottom.

Over the past decade, the search for the genetic basis of complex traits and disease has been enabled through the use of genome-wide association studies (GWAS) [8]. In such studies, hundreds of thousands to millions of polymorphic markers across the genome are assessed to enable searches for statistical association with either a disease-state or with a quantitative trait [9,10]. Information about the linkage disequilibrium pattern for various polymorphisms has allowed comprehensive searches for common variants in the genome associated with a disease or trait of interest. Here we review such studies that have been performed for various hematologic parameters and diseases and discuss how these findings are enabling new insight into human hematopoiesis.

The genetics of erythroid phenotypes

The earliest GWAS studies of hematologic phenotypes were focused on fetal hemoglobin (HbF) levels [11,12]. The interest in this trait stems from the ability of increased HbF to ameliorate the clinical symptoms of the two major disorders of hemoglobin, sickle cell disease (SCD) and β-thalassemia [13]. These GWAS studies revealed three major loci associated with HbF levels, which collectively explained nearly half of the variation in this trait – a contrast to the majority of other traits and diseases examined by GWAS [8,14]. The loci identified included the β-globin gene cluster on chromosome 11, as well as variants on chromosome 6 in a region between the genes HBS1L and MYB and variants within the gene BCL11A on chromosome 2 [14,15]. The elucidation of HbF-associated genetic variants in the BCL11A gene led to work showing that the product of this gene is a major transcriptional regulator of the physiologic fetal-to-adult hemoglobin switch and HbF silencing in humans [16]. More recent studies have shown that ablation of Bcl11a leads to substantial reactivation of HbF expression and amelioration of SCD phenotypes in mouse models [17], supporting the findings from studies of HbF-associated genetic variants in SCD patients [15]. Other studies have suggested that the MYB gene may have an important role in the regulation of HbF levels [18,19], but the exact mechanisms underlying these observations and how precisely this relates to the HbF-associated variants on chromosome 6 remains unclear [20]. It should be noted that the variants in the HBS1L-MYB intergenic region are pleiotropically associated with several other hematologic traits (including WBC and platelet counts), albeit to a far lesser extent than their effect on HbF levels [21]. These studies have revealed promising therapeutic targets for HbF induction in patients with β-thalassemia and SCD. Indeed, GWAS of clinical severity in β-thalassemia patient populations found that the HbF-associated loci are significantly associated with and may explain the majority of variation in the clinical course [22,23]. It should be noted that all the GWAS of HbF levels have been performed on adults and while the three loci that were uncovered may have important roles in silencing of HbF, other common genetic variants may influence the developmental fetal-to-adult globin switch.

Since the initial studies on HbF levels, further studies have examined a variety of other clinically-relevant erythroid parameters including hemoglobin levels, hematocrit (the percentage of the blood volume occupied by erythrocytes), red blood cell count, and red cell size (mean corpuscular volume, MCV) [24-32]. In these other erythrocyte phenotype studies, the loci reaching genome-wide significance collectively explain < 10% and often only a few percent of the total variation in those traits. These studies have implicated > 20 loci in erythrocyte traits. However, with a few exceptions, the underlying molecular mechanisms remain unclear. The BCL11A and HBS1L-MYB loci have been implicated in the regulation of erythrocyte size and this may be related to the same molecular mechanisms that lead to variation in HbF levels [24]. The TMPRSS6 locus has been strongly associated with a variety of erythroid traits, most notably hemoglobin levels, in a number of different GWAS [24-27]. Interestingly, rare mutations in the TMPRSS6 gene, which encodes a transmembrane serine protease expressed in the liver that is necessary for the regulation of iron absorption, are associated with iron deficiency states [33]. This is an excellent example of how GWAS can highlight informative biology. In addition, some of these variants may be important modifiers of common forms of anemia, though this remains to be formally tested [34].

The mechanisms underlying the loci implicated in the erythrocyte trait GWAS have been largely unexplored. However, using clues from these studies and by performing mechanistic follow up using both human and mouse erythroid cells, it was found that the CCND3 gene on chromosome 6 (encoding cyclin D3) plays a critical role as a regulator of the number of divisions that occur during terminal erythropoiesis, thereby regulating erythrocyte size and number [35]. In addition to providing important insight into how normal erythropoiesis proceeds, these findings may have implications for understanding pathologic states where erythropoiesis is disrupted that are often characterized by changes in erythrocyte size. Moreover, these findings could also have important implications for ongoing work to recapitulate the process of erythropoiesis ex vivo for potential use in transfusions [36].

Common genetic variation in platelet phenotypes

Platelets are critical for the formation of blot clots and are continually produced from multinuclear precursors in the bone marrow known as megakaryocytes (Figure 1). In clinical practice, both platelet count and size (mean platelet volume, MPV) are measured routinely. Given the association of platelets with the formation of blood clots, there has been great interest in examining these traits, which can predispose individuals to cardiovascular events, such as myocardial infarctions [21,37,38]. A number of GWAS have examined both platelet count and MPV in humans [25,31,32,37,39-41]. A large meta-analysis of these GWAS, involving > 66,000 individuals, has recently been reported and has identified 68 loci associated with platelet phenotypes, including 53 newly discovered loci [37]. This study highlights the value of performing large-scale meta-analyses to uncover new loci associated with particular phenotypes. While the effect sizes of the majority of uncovered variants are small (generally < 1% of the variation), many of the genes in these loci appear critical for aspects of megakaryopoiesis and platelet formation [37]. However, while genomic approaches have suggested that many of these genes are highly expressed in megakaryocytes and studies in zebrafish and Drosophila models of hematopoiesis support an important role for these genes, the precise molecular mechanisms underlying the association between platelet phenotypes and the variants in these loci remain unclear [37]. Some studies have begun to explore the relationship between specific platelet-associated variants and their effect on the expression of specific genes, such as PIK3CG [42] – however, even in this well-studied case, the mechanism by which the PIK3CG gene influences platelet formation remains unclear.

The landscape of common genetic variation in WBC counts and diseases

WBCs are critical for the defense against foreign microorganisms and compose both the innate and acquired immune systems [1]. In addition to their role in normal immunity, the development of these cells often goes awry in human conditions including autoimmunity and hematologic malignancies. As with other hematologic phenotypes, WBC phenotypes have also been the subject of GWAS [25,28,31,32,43-49]. In contrast to erythrocytes and platelets, WBCs constitute a heterogeneous population of cells. The majority of circulating WBCs are neutrophils (50-70% of WBCs in healthy individuals), which form an early defense against invading microorganisms. It is therefore not surprising that there is a significant overlap between loci that were implicated in variation of both the total WBC and neutrophil counts [21,46-48]. For example the PSMD3-CSF3 locus on chromosome 17 has been associated with both WBC and neutrophil counts in several studies. This association is interesting, since a strong candidate for these traits is the CSF3 gene, which encodes the granulocyte colony stimulating factor (G-CSF) that is used therapeutically to increase neutrophil output. Further work is needed to examine whether these variants may influence G-CSF levels in humans and whether this may also affect the response to G-CSF therapy [50]. Studies have also suggested the role of loci associated with monocyte levels [28,46-49], including the ITGA4 gene locus on chromosome 2 encoding the α4 integrin subunit that is necessary for monocyte migration [51]. Given their role in common allergic diseases and asthma, there has also been interest in understanding genetic variants that affect the levels of eosinophils and basophils. Some interesting candidate associations have been uncovered, including variants at the GATA2 and ERG loci [48]. Interestingly, several of the loci associated with eosinophil count were also associated with allergic asthma [43], supporting the recently uncovered relationship between eosinophils and asthma [52].

While some promising associations for neutrophil, monocyte, eosinophil, and basophil counts have been reported, the genetic studies of lymphocyte counts have yielded far less success [46-48]. This may be attributable to the fact that most lymphocyte counts do not differentiate between different subsets that require strikingly different molecular programs for their development [1]. Indeed, a study of CD4:CD8 ratios of T lymphocytes has yielded some promising association signals [44], suggesting that such approaches on a larger scale may be beneficial. In several instances however, studies of autoimmune diseases have revealed important regulators of particular lymphocyte subsets [53,54]. In addition, studies of B lymphocyte malignancies including chronic lymphocytic leukemia (CLL) and acute lymphoblastic leukemia (ALL) have revealed loci that contain genes known to be critical for normal B lymphocyte development [55-59]. These findings collectively suggest that common variation in the normal development of these lymphocyte populations may confer susceptibility to human disease and hence molecular studies of such processes in humans may be vital to develop better therapeutic approaches for these disorders.

Challenges and future directions

While the past few years has witnessed an impressive output of data on common genetic variation in human hematopoiesis, four major challenges arise from this work. (1) As illustrated by the example on platelet phenotypes, larger meta-analyses of GWAS can uncover new aspects of biology that are not easily appreciated [37]. Although many of the loci uncovered from such studies can have rather small effect sizes, the importance of the genes involved in these loci may be critical for understanding important aspects of hematopoiesis [35]. It is important to remember that the genetic contribution to variation seen in GWAS at any locus provides only a minimal estimate of the biological contribution and major perturbations of these genes can show dramatic phenotypes. (2) While the loci that have been uncovered from GWAS studies provide a “parts list” of potential regions associated with a particular trait, the genes underlying these effects remain to be uncovered. Occasionally, nearest gene approaches are used to implicate a specific gene in a trait, but more unbiased and rigorous genomic approaches should be applied to properly nominate candidate genes in these loci and uncover causal variants [10,60]. If such mapping approaches are coupled to other orthogonal genomic datasets, such as cell-type specific expression or functional datasets, insight into likely candidate genes can be gained [10,37]. (3) While broader analyses will be useful in annotating the genes that are involved in the various traits discussed above, detailed molecular and mechanistic studies are needed to better understand how these genes play a role in various aspects of hematopoiesis. Even for well-studied cases [16,35], further work is needed to better understand the mechanisms involved. (4) Finally, it will be important to integrate the information from common variant studies with rare mutation analyses and studies of genes involved in hematologic disorders. Rare variant studies of hematologic traits are just now being reported and will likely have improved power to detect new variants in the ensuing years [61]. As illustrated by the studies of hematologic malignancies and autoimmune conditions, important lessons can be learned about susceptibility to these diseases by studying aspects of normal human hematopoiesis [53,56]. As new studies surmount these challenges in the coming years, our understanding of human hematopoiesis will undoubtedly improve significantly.