Erythropoiesis is a tightly regulated system, but the details are still not fully elucidated. Much remains to be learned from uncovering the molecular basis of many congenital and acquired mutations that disrupt the control of erythropoiesis. Two competing hypothesis have been proposed to explain the differentiation of the hematopoietic progenitors cells toward erythroid lineage.
DETERMINISTIC MODEL OF LINEAGE COMMITMENT
According to the deterministic model of lineage commitment, specific extracellular signals, such as cytokines, play an instructive role in lineage specification.
Multipotential progenitors (Chap. 18) and erythroid unipotential progenitors, the BFU-E, require stem cell factor, interleukin-3, granulocyte-macrophage colony-stimulating factor, and/or thrombopoietin for growth and survival (Fig. 32–4).
A. Cytokine influence on hematopoiesis. CFU-GEMM, colony-forming unit–growing granulocyte, erythrocyte, megakaryocyte, and macrophage precursors; G-CSF, granulocyte colony-stimulating factor; IL3, interleukin-3; SCF, stem cell factor; TPO, thrombopoietin. B. Regulation of erythropoiesis by hypoxia. HIF-1 hypoxia inducible factor-1; VEGF, vascular endothelial growth factor 1.
STOCHASTIC MODEL OF ERYTHROID DIFFERENTIATION
In contrast, the stochastic model proposes that spontaneous formation of a set of transcription factors, independently of extrinsic signals, mediates lineage commitment. These transcription factors activate a unique set of genes for a particular lineage and repress the action of alternative transcription factors and cytokines play only a permissive role. Most of evidence based on gene targeting studies and in vitro culture studies support the stochastic model of differentiation. Several transcription factors, such as GATA-1, FOG1, erythroid Kruppel-like factor (EKLF), PU.1, and SCL/TAL1 (stem cell leukemia/T-cell acute lymphoblastic leukemia 1 factor) have been characterized that are involved in erythroid differentiation.
The GATA family of zinc-finger transcription factors was first identified as the nuclear factors that bind to the GATA sequence in the enhancer region of the globin genes.49,50 GATA-1 protein is expressed during erythroid differentiation, with highest expression in CFU-Es and pronormoblasts. GATA-1 promotes erythroid differentiation by activating several erythroid-specific genes and represses transcription of Kit receptor and GATA-2. GATA-1 deficient mice die at embryonic day 10.5 with severe anemia from maturation arrest at the stage of pronormoblasts.51 In vitro, GATA-1 null the embryonic stem cells fail to mature beyond pronormoblast and undergo apoptosis. GATA-1 and its cofactor CBP are essential for the formation of an erythroid-specific histone acetylation pattern of histones at the active globin genes and the β-globin locus control region.52 GATA-1, along with EPO, induces expression of the antiapoptotic protein Bcl-xL53 and interacts with multiple proteins, including FOG-1 and PU.1,54 and FOG-1 acts as a cofactor for GATA-1.55 GATA-1 interaction with PU.1 appears to counteract erythropoiesis by inducing differentiation of pluripotent stem cell to myeloid and B lymphopoiesis and inhibition of erythropoiesis.54,56,57 Whereas PU.1 absence appears to be required for completion of terminal erythroid differentiation, low levels of PU.1 expression are essential for fetal erythropoiesis and for proper augmentation of adult erythropoiesis at times of stress.58
FOG-1, a member of friend of GATA family of zinc finger proteins, act as cofactors for GATA-1. It was first identified in a yeast two-hybrid screen for GATA-1 interacting proteins.55 It binds to the amino zinc finger of GATA-1. FOG-1−/− mice die during embryonic days 10.5 to 11.5 from severe anemia with arrest in erythroid maturation at a stage similar to that observed in the GATA-1− mice.59 FOG-1 physically interacts with GATA-1 to augment or inhibit its transcriptional activity depending on the promoter context.
GATA-2 was initially cloned as a GATA motif-binding factor is present in all erythroid cells; and, targeted deletion of GATA-2 resulted in embryonic lethality at day 10.5 from ablation of blood cell development.60 GATA-1 and GATA-2 directly regulate GATA-2 transcription in a reciprocal fashion during erythroid differentiation.61,62 GATA-2 autoregulates its transcription by binding to its own regulatory elements in the promoter region. This autoregulation is abolished by the displacement of GATA-2 by GATA-1 (GATA-2/GATA-1 switch), an interaction facilitated by FOG-1.63 Chromatin immunoprecipitation studies indicate that FOG-1 facilitates occupancy by GATA-1 at selected cis-regulatory chromatin elements. Double knockout of GATA-1 and GATA-2 results in embryonic lethality with complete absence of primitive erythropoiesis.64 The severity of this phenotype compared to either single GATA-1 or GATA-2 knockout suggests overlapping functions of these two transcription factors in primitive erythropoiesis.
EKLF is a zinc finger protein identified by subtractive hybridization of the mRNA of erythroid cells with common messages in a myeloid cell line.65 It interacts with CACCC sequence in the β-globin promoter, where it modifies chromatin structure permitting β-globin gene transcription. EKLF-deficient mice die at embryonic day 14.5 to 15 from severe anemia from defective definitive erythropoiesis.66 There is a marked decrease in β-globin mRNA and protein levels in EKLF-deficient erythroid cells. Large amounts of iron accumulate in the reticuloendothelial system of EKLF-deficient mice, consistent with an ineffective erythropoiesis.
STEM CELL LEUKEMIA/T-CELL ACUTE LYMPHOBLASTIC LEUKEMIA 1
SCL/TAL1 is a member of basic helix-loop-helix transcription factors essential for maturation of the erythroid and megakaryocytic lineages.67 Knockout of SCL/TAL1 leads to failure of hematopoiesis.68 Selective rescue of SCL/TAL1 null embryonic stem cells under the control of stem cell enhancer revealed differentiation blocks in erythroid and megakaryocytic maturation.69 Conditional knockout studies have revealed that erythroid and megakaryocytic precursors do not develop in the marrow of mice upon deletion of SCL/TAL1.70 Heterodimerization of SCL with other transcription factors, such as E2A, is a prerequisite for its functions.71
BCL11A, a transcription factor initially identified in lymphoid cells, regulates erythroid differentiation, especially in switching from fetal to adult hemoglobin.71 Fetal hemoglobin (HbF) levels decline after birth and are then replaced by adult hemoglobin A. The molecular mechanisms responsible for this switch are not completely known. Genome-wide association findings have provided a major breakthrough in understanding this phenomenon.72 There is an inverse correlation between BCL11A and HbF expression in erythroid cells. BCL11A occupies several discrete sites in the β-globin gene cluster and likely plays an important role in hemoglobin switching during erythroid differentiation.
GROWTH ARREST-SPECIFIC 6 PROTEIN
Growth arrest-specific 6 (Gas6) protein is a secreted vitamin K–dependent protein that interacts with cell membranes and leads to intracellular signaling (via its receptor tyrosine kinases). Gas6 receptors are expressed in hematopoietic tissue, megakaryocytes, myelomonocytic precursors, and marrow stromal cells. Gas6 amplifies the erythropoietic response to EPO using a mouse model of Gas6 knockout.73 Gas6 is known to downregulate the expression of inflammatory cytokines such as tumor necrosis factor-α by macrophages.74
Figure 32-5 outlines the interrogation of the molecular mechanisms that regulate lineage-specific differentiation and commitment reveals the existence of separate megakaryocytic/erythroid progenitors versus both myeloid and lymphoid lineages.75
Schematic outline of emerging hierarchy of hematopoiesis outlining a separate progenitor for erythroid/megakaryocytic lineage (MEP). RUNX-1, Scl1/Tal1, Lmo1, MllTel, PU-1, GATA-1, GATA-2, gf-1,C/EBPα, FOG1, EKLF, Ikaros, E2A, EBF, PAX-5, BCL 11A, NOTCH, TCF-1 are transcription factors. CLP, common lymphoid progenitor; CMP, common myeloid progenitor; GMP, granulocyte/macrophage progenitor; LT-HSC, long-term hematopoietic stem cell; MEP, megakaryocyte/erythroid progenitor; RBCs, red blood cells; ST-HSC, short-term hematopoietic stem cell. (Reproduced with permission from Orkin SH1, Zon LI: Hematopoiesis: an evolving paradigm for stem cell biology, Cell. 2008 Feb 22;132(4):631-644.)
ERYTHROPOIETIN, OXYGEN SENSING, AND HYPOXIA-INDUCIBLE FACTOR
The principal hormone regulating erythropoiesis is EPO, which is produced principally in the kidney.7 Erythroid progenitors express their own EPO.76 Different levels of kidney-produced EPO are optimal for various stages of erythroid maturation.77 Purification of EPO provided a partial protein sequence that led to cloning of the gene and permitted mass production of the recombinant protein.78 EPO and its recombinant form are heavily glycosylated α-globulins with a molecular mass of 34,000 daltons and a specific activity of approximately 200,000 IU/mg.79,80 Sixty percent of the molecular weight of the recombinant protein is contributed by amino acids; the remaining 40 percent is composed of carbohydrate. Using molecular probes for EPO, mRNA enabled the localization of the synthesis of EPO to renal cortical interstitial cells81,82 of endothelial or fibroblastic lineage. The cells appear to function in an all-or-none fashion, with the overall production of mRNA dependent on the number of cells activated.83
Certain 5′ sequences located 6000 to 12,000 bp upstream also affect EPO gene transcription.84 These sequences are not hypoxia sensitive but appear necessary for tissue and cellular specificity.84 Hepatic production is contributed primarily by hepatocytes but is a much less important source than is the kidney.85 During fetal life, however, hepatic EPO production is of major importance for red cell production (Chap. 7).86,87 EPO production is regulated exclusively at the level of its transcription by hypoxia at the transcription level. The transcriptional activation of the EPO gene is controlled by a specific sequence located in the 3′ flanking region termed hypoxia-responsive element.88,89,90 The core of the enhancer is constituted by the sequence CACGTGCT and mutations in this core sequence abolish hypoxia responsiveness.
EPO is not stored; it is secreted immediately.81,82,83 Circulating recombinant EPO and presumably native EPO have a half-life (T1/2) of 4 to 12 hours, with a volume of distribution slightly larger than that of the plasma volume.91 EPO is degraded after it binds to EPOR (see “Erythropoietin Receptor” below).92
Interaction of EPO with its receptor EPOR results in (1) stimulation of erythroid cell division, (2) erythroid differentiation by induction of erythroid-specific protein expression, and (3) prevention of erythroid progenitor apoptosis.93 Earlier models of this interaction were based on the ligand (EPO)-induced homodimerization of EPOR. In reality, EPOR is a preformed homodimer that undergoes a major conformational change upon binding,94 which initiates the EPO-specific erythroid signal transduction cascade (Fig. 32–6). The cytoplasmic portion of EPOR contains a positive regulatory domain that interacts with Janus kinase 2 (JAK2).95 Immediately after EPO binding, JAK2 cross-phosphorylates the EPOR itself, and other proteins such as STAT5 (signal transducer and activator of transcription 5), thus initiating a cascade of erythroid-specific signaling.96 JAK2/STAT5 signaling plays an essential role in EPO–EPOR–mediated regulation of erythropoiesis (see Fig. 32–6).97 Deficiency of EPO–EPOR is lethal by abrogating fetal liver erythropoiesis (but not the “primitive” yolk sac erythropoiesis). However, in these EPO or EPOR knockout mice, differentiation of pluripotential stem cells to BFU-E occurs, but not the subsequent erythroid differentiation. This occurrence demonstrates the crucial role of EPO in terminal erythroid maturation.35,98,99 The C-terminal cytoplasmic portion of EPOR also possesses a domain essential for prevention of apoptosis (see Fig. 32–6) by inducing expression of Bcl-xL via phosphoinositide 3′-kinase (PI3K).53 However, the cytoplasmic portion of EPOR also contains a negative regulatory domain100 that interacts with hematopoietic cell phosphatase (HCP, also known as SHP1) and down-modulates signal transduction.101 Once recruited by EPOR tyrosine (Y)429, HCP attaches to the cytoplasmic EPOR domain and dephosphorylates JAK2. Inactivation of the HCP binding site leads to prolonged phosphorylation of JAK2/STAT5.101,102 CIS3 (also known as SOCS3), another negative regulator of erythropoiesis, binds to the cytoplasmic portion of the EPOR Y401 and suppresses EPO-dependent JAK2/STAT5 signaling.103,104 Thus, deletion of the distal C-terminal cytoplasmic portion of EPOR results in a truncated EPOR, abolishes negative regulatory elements, and results in increased proliferation of erythroid progenitor cells. Gain-of-function mutations resulting from deletion of the negative regulatory domain of the EPOR gene (Chap. 57) have been demonstrated in a small proportion of individuals with primary familial and congenital polycythemia, but are rarely found in erythroleukemia105; however, the rearranged EPOR has also been identified in a subtype of high-risk B-progenitor acute lymphoblastic leukemia.106
Outline of erythropoietin–erythropoietin receptor (EPO–EPOR) signaling. Activation of Janus kinase 2 (JAK2) and signal transducer and activator of transcription 5 (STAT5) represents erythropoiesis-promoting signals. Interaction of CIS, a signal transduction protein that downregulates activity of erythropoietin receptor, and hematopoietic cell phosphatase (HCP) inhibit erythropoiesis. Phosphatidylinositide 3′-kinase (PI3K) activation of Bcl-xL inhibits apoptosis of erythroid progenitors.
Because the activation signal after EPO binding to its receptor is rapidly downregulated and EPO briskly disappears after binding to EPOR, EPO–EPOR internalization is one mechanism of downregulation of EPO signaling.92 After EPO binds to the receptor, EPO–EPOR complexes are ubiquinated, rapidly internalized, and targeted for degradation. This process involves two proteolytic systems, the proteosomes that remove part of the intracellular domain of EPOR at the cell surface and the lysosomes that degrade the EPO–EPOR complex in the cytoplasm.107
Another incompletely understood mechanism of erythropoiesis regulation is the presence of several EPOR isoforms, some of which may have an inhibitory function on erythropoiesis.108,109,110
Nonerythroid Effect of Erythropoietin Signaling
Soon after the erythroid effects of recombinant EPO were described, nonerythroid effects were identified.111 Some of these effects are beneficial, including roles in neural, cardiovascular, and retinal tissues, and in immune function and in tissue repair. It has been claimed that the hormone also exerts beneficial effects on athletic performance and improved neurocognition, but these are not convincingly substantiated. The effects of EPO in nonerythroid tissues are the result of EPO binding to EPOR, and, as in erythroid cells, the EPO–EPOR interaction initiates a signal transduction process that regulates the survival, growth and differentiation of the involved tissue.112 EPO and EPOR play a physiologic role in many nonerythroid cells including endothelial cells,113 megakaryocytes, and cells of the brain, heart, uterus, breast, and testis. However, in some tissues (e.g., brain, heart, and kidney) the signaling mechanism may be different because EPO can interact with EPOR and CD131 heterodimers.114,115
Detrimental EPO effects include a poorly understood increased cancer mortality,95,116,117 increased blood pressure, and thrombosis.114
Under normal conditions, EPO production is mediated by decreased oxygen saturation of hemoglobin, that is, hypoxemia.77 Hypoxia is an important factor in development, energy metabolism, vasculogenesis, iron metabolism, tumor promotion and is the principal regulator of erythropoiesis. The response to hypoxia is controlled by transcriptional factors termed hypoxia-inducible factors (HIFs).118,119 Adaptive physiologic responses to hypoxia serve to (1) increase O2 delivery to cells, (2) allow cells to survive under reduced O2 by activating glycolysis, and (3) reduce the formation of reactive oxygen species.120 HIFs are heterodimeric transcription factors composed of a highly-regulated α subunit and a constitutively expressed β subunit that belongs to the basic helix-loop-helix containing the PER-ARNT-SIM (PAS)-domain family of transcription factors. The first HIF to be discovered, HIF-1, is induced in hypoxic cells and binds to a cis-acting nucleotide sequence of hypoxia-controlled genes referred to as the hypoxia-responsive element, first identified in the 3′-flanking region of the human EPO gene.121 Two other HIF homologues, HIF-2 and HIF-3, have been identified. HIF-2 has more limited tissue expression than HIF-1 but it is the principal regulator of EPO expression.118,119 Many hypoxia-inducible genes are directly regulated by HIF-1. Approximately 3 percent of all genes expressed in endothelial tissue are HIF-1 regulated.122 The half-life of HIF-1α in the cell is minutes under normoxic conditions. HIF-1 and HIF-2 α subunits are rapidly degraded by the von Hippel-Lindau (VHL) protein–ubiquitin–proteasome pathway.123 The targeting and subsequent polyubiquitination of HIF α subunits requires VHL, iron, O2, and proline hydroxylase activity, and this complex constitutes the oxygen sensor (Fig. 32-7).124,125
Schematic outline of regulation of HIF-2 and HIF-2α subunits by hypoxic and nonhypoxic pathways. HIF, hypoxia-inducible factor; HSP90, heat shock protein 90; PHDs, proline hydroxylases; p300 and CBP, cofactors of hypoxia response transcription with HIF-1; RACK1, receptor of activated protein kinase C; ub, ubiquitin residues; VHL, von Hippel-Lindau protein.
This degradation of HIFs α subunits is initiated by a posttranslational hydroxylation event at residue proline 564 (P564) that is mediated by one of several iron-containing proline hydroxylases (PHDs). The hydroxylation of HIFs α subunits facilitates binding to the VHL protein and subsequent ubiquitination and proteasomal degradation. Osteosarcoma protein 9 (OS-9) binds to both HIF-1α and PHD2 and is required for efficient prolyl hydroxylation.126 Under hypoxic conditions, HIF-1 and HIF-2 α proteins are not degraded and are translocated to the cell nucleus where they dimerize with HIF-β to form the HIF heterodimer that activates transcription through binding to specific hypoxia-responsive elements on target genes. Another regulatory step involves O2-dependent asparaginyl-hydroxylation of asparagine (N) 803 in HIF-1α that requires the enzyme HIF-3, also known as FIH-1 (factor inhibiting HIF-1). Hydroxylation of N803 during normoxia blocks the binding of transcription factors p300 and CBP to HIF-1, resulting in inhibition of HIF-1–mediated gene transcription. Under hypoxic conditions, HIF α subunits are not hydroxylated. The unmodified protein escapes VHL-binding, ubiquitination, and degradation (see Fig. 32–7). When N803 of HIF-1α is not asparaginyl-hydroxylated, p300 and CBP can bind to the HIF-1 heterodimer, allowing transcriptional activation of HIF-1 target genes.
HIF-2 Transcription Factor
HIF-1α and HIF-2α exhibit a high degree sequence homology but have differing mRNA expression patterns: HIF-1α is expressed ubiquitously, whereas HIF-2α expression is restricted to certain tissues.118,127 The kidney is the main site of EPO production (i.e., renal interstitial cells), and HIF-2 and, to a lesser degree, HIF-1 are the principal regulator of EPO transcription in the kidney.118,127 In other tissues, such as brain128 and liver87 (which generates approximately 15 percent of circulating EPO), EPO gene transcription is HIF-2–dependent.127 The discovery of an iron-responsive element in the 5′ untranslated region of HIF-2α reveals a novel regulatory link between iron availability and HIF-2α expression129 that may also influence control of erythropoiesis. The importance of HIF-2α in regulation of EPO gene was demonstrated by a gain-of-function HIF-2α mutation causing erythrocytosis.130
Hypoxia-Independent Regulation of Hypoxia-Inducible Factor
While O2-dependent regulation of the HIF-1α subunit is mediated by prolyl hydroxylases, VHL protein, and the proteasomal complex, hypoxia-independent regulation of HIF-1α has been uncovered. This novel mechanism involves the receptor of activated protein kinase C (RACK1) as a HIF-1α–interacting protein that promotes prolyl hydroxylase/VHL-independent proteasomal degradation of HIF-1α. RACK1 competes with heat shock protein 90 (HSP90) for binding to the PAS-A domain of HIF-1α. HIF-1α degradation is abolished by loss-of-function RACK1. RACK1 binds to the proteasomal subunit, elongin-C, and promotes ubiquitination of HIF-1α (see Fig. 32–7). Therefore, RACK1 and HSP90 are the essential components of an O2/PHD/VHL-independent mechanism for regulating HIF-1α.131
The rapid degradation of HIF is complex and tightly regulated, and mutations affecting the genes that encode the regulatory factors may underlie some of the unexplained congenital polycythemias.
This complex (Chaps. 34 and 57) constitutes the oxygen sensor (see Fig. 32–7).124,125,132
INSULIN-LIKE GROWTH FACTOR-1, RENIN–ANGIOTENSIN SYSTEM, AND HEMATOPOIESIS
Although in vitro studies of erythropoiesis have provided crucial information about the regulation of erythropoiesis, many experiments were performed in the presence of serum and serum-component proteins capable of stimulating and inhibiting erythropoiesis.133,134 Using serum-free conditions, insulin-like growth factor-1 (IGF-1) can partially substitute for EPO in BFU-E cultures. Furthermore, anephric, nonanemic patients with no detectable EPO have elevated levels of IGF-1.135