Regulation of erythropoiesis

INTRODUCTION — The production of erythrocytes is a tightly regulated process. During steady state hematopoiesis, approximately 10¹⁰ red blood cells are produced per hour in the bone marrow to maintain the hemoglobin level within fairly narrow limits. Production can be rapidly increased in the setting of ongoing blood loss or hemolysis.

This topic review will discuss the elements that underlie erythropoiesis. A general discussion of hematopoiesis and stem cell function is presented separately. (See "Overview of hematopoietic stem cells".)

OVERVIEW OF ERYTHROPOIESIS — Erythropoiesis begins with the differentiation of multipotent hematopoietic stem cells (HSC) into the most primitive erythroid progenitors (figure 1). In the classical hematopoietic hierarchy, HSC progeny can differentiate into a common myeloid progenitor (CMP), which in turn differentiates into the common megakaryocyte-erythroid progenitor cell (MEP).

Studies in humans and mouse models using transplantation and single-cell lineage tracing have shown that the MEPs may derive directly from HSCs [1,2]. MEPs give rise to the erythroid progenitor cells, which subsequently follow a differentiation program that culminates in the emergence of mature erythrocytes. (See 'Erythroid progenitor cells' below.)

The existence of an MEP explains why iron deficiency is associated not only with anemia but frequently with thrombocytosis as well. Iron metabolism is an evolutionarily highly conserved process. Data from mouse models show that an iron-deficient diet or knockout of the gene that is altered in iron-refractory iron deficiency anemia (IRIDA), TMPRSS6, have thrombocytosis [3]. The mechanism was found to involve increased flux of progenitor cells through MEPs, with evidence for attenuation of extracellular signal-regulated kinase (ERK) phosphorylation. This mechanism illustrates an example of extracellular influences on progenitor cell fate.

●This process is driven by successive combinations of transcription factors that dictate the expression of adhesion and hematopoietic growth factor receptors (HGFRs). (See 'Transcription factors' below.)

●Adhesion receptors play an important role in the localization and release of maturing cells from specific niches in bone marrow. (See 'Stroma' below.)

●Hematopoietic growth factors (HGFs), such as interleukin 3 (IL-3), granulocyte-macrophage colony-stimulating factor (GM-CSF), and stem cell factor (SCF, also called Kit ligand or Steel factor), are important for the amplification of progenitor cells. (See 'Growth factors' below.)

Erythropoietin (EPO) is a growth factor essential for the amplification and terminal differentiation of erythroid progenitors and precursors. Information concerning the control of EPO expression by hypoxia has provided insight into the regulation of erythropoiesis. (See 'Hypoxia and EPO expression' below and "Molecular pathogenesis of congenital erythrocytoses and polycythemia vera", section on 'Oxygen sensor'.)

ERYTHROID PROGENITOR CELLS — The erythroid progenitor compartment of the bone marrow is invisible upon inspection by light microscopy. These committed single lineage progenitors are derived from the stochastic differentiation of bipotential or multipotential progenitors, which emerged from a tiny population of stem cells (figure 1) [4]. The two sets of erythroid progenitors described below, burst-forming units of the erythroid lineage and colony-forming units of the erythrocyte lineage (BFU-E and CFU-E), cannot be identified by specific morphological features, although they have been purified and analyzed using flow cytometric methods [5]. Evidence suggests that all hematopoietic progenitors or stem cells resemble lymphoblasts [6-8].

Erythroid burst-forming unit — In humans, the most primitive single lineage committed erythroid progenitor is the erythroid burst-forming unit (BFU-E). These cellular clusters are so named because they have the following characteristics in semisolid in vitro cultures:

●The colonies have a burst-like morphology.

●In response to the combination of EPO and one of SCF, IL-3, or GM-CSF, the progeny of the first few cellular divisions are motile and form subpopulations of erythroid colony-forming units (CFU-E) [9]. Each of these units subsequently forms a large colony of proerythroblasts, which mature into erythroblasts and a few enucleated erythrocytes.

The entire process requires approximately two weeks in vitro.

Erythroid colony-forming unit — Bone marrow also contains the more mature erythroid colony-forming unit (CFU-E) that, under the influence of EPO, form small colonies of erythroblasts in seven days of culture.

PRECURSORS AND MATURE CELLS — The erythroid precursor or erythroblast pool represents about one-third of the marrow cell population in the normal child (above the age of three) and the adult. Proerythroblasts are the earliest recognizable forms (picture 1). These cells divide and mature through basophilic, polychromatic, and orthochromatic normoblast cells to form the reticulocyte and finally to the circulating mature erythrocyte; this process involves a reduction in cell size, nuclear condensation and extrusion, and hemoglobin accumulation (figure 2).

●On average, each proerythroblast can form approximately eight reticulocytes. The mean transit time from proerythroblast to the emergence of the reticulocyte into the circulation is approximately five days [10].

●In the circulation, the reticulocyte takes approximately one day to become the mature erythrocyte (figure 3).

●The lifespan in the circulation of the mature erythrocyte is approximately four months. (See "Red blood cell survival: Normal values and measurement", section on 'Historical methods for determining RBC lifespan'.)

Acute anemia — In acute anemia, the length of this transit time may decrease to as little as one or two days because of skipped divisions [11]. In this setting, the red cells that emerge are macrocytic; they may also bear surface i antigen because they have had insufficient time to convert i antigen to I antigen [12]. They have other fetal characteristics such as increased fetal hemoglobin (Hb F); this results from an increase in differentiation from burst-forming units of the erythroid lineage (BFU-E), which have higher Hb F expression than colony-forming units of the erythroid lineage (CFU-E). The cells also contain excessive burdens of the debris that normally accumulates during cell assembly [13,14]. As a result, stress erythropoiesis is associated with circulating Pappenheimer bodies (iron granules), basophilic stippling (ribosomes), Heinz bodies (hemoglobin inclusions), and Howell-Jolly bodies (nuclear remnants) (picture 2A-B).

STROMA — The bone marrow stroma consists of fibroblastoid cells, endothelial cells, and macrophages that comprise the hematopoietic microenvironment [15,16]. Interaction between receptors on erythrocyte precursors and elements of the stromal matrix are important determinants of erythroid maturation. As an example, murine erythroleukemia cells bind specifically to the extracellular matrix protein fibronectin [17]; induction of differentiation of these cells subsequently leads to loss of adhesion. Binding to fibronectin involves very late antigen (VLA)-5, an erythroid integrin that is expressed on erythroid cells and is lost at the reticulocyte stage of development. VLA-5 binds to a specific peptide in the fibronectin molecule, Arg-Gly-Asp-Ser (RGDS) [18].

Normal burst-forming units of the erythroid lineage (BFU-E) bind preferentially to a stromal fibroblast cell strain [19]; after binding, these cells proliferate and differentiate in the presence of EPO alone [19]. This interaction is blocked by antibodies specific for the cell binding domain of fibronectin, which contains the RGDS sequence [19].

Macrophages — Interactions between developing erythroid cells and macrophages are also important [20-25]. This is classically exemplified by the "erythroblastic island", composed of developing erythroblasts surrounding a central macrophage [25]. These macrophages make physical contacts with developing erythroblasts, enabling signaling and the transfer of growth factors and nutrients (eg, iron) to these cells [26].

Deoxyribonuclease II from these macrophages may be required for the process through which mature red cells lose their nuclei [27]. In addition, Rac GTPases play an essential role in enucleation through activation of the formin mDia2, which nucleates unbranched actin filaments leading to formation of the contractile actin ring [28].

TRANSCRIPTION FACTORS — Much has been learned about the essential roles of several transcription factors via the study of their activation in leukemic translocations and the effect of gene disruption in embryonic stem cells [29]. Important factors that most likely act at the level of hematopoietic stem cells include TAL1/SCL, LMO2/RBTN2, and GATA2, while GATA1, FOG1 and EKLF are more important for erythropoiesis (figure 1).

TAL1/SCL — T-cell acute lymphocytic leukemia 1 (TAL1; also called stem cell leukemia [SCL]) transcription factor is expressed in leukemias (biphenotypic [lymphoid/myeloid] and T cell) [30,31], primitive hematopoietic progenitors, and more mature erythroid, mast, megakaryocyte, and endothelial cells [32,33]. Targeted disruption of this gene in mice leads to death in utero from the absence of blood formation. This finding, together with lack of in vitro myeloid colony formation [34,35], and data indicating that TAL1 contribute to angiogenesis [36], suggest a very early role for TAL1, most likely at the pluripotent or myeloid-erythroid stem cell level. Interestingly, conditional gene targeting in adult mice shows that TAL1 is not required for engraftment and self-renewal of adult stem cells, but is required for differentiation of erythroid and megakaryocyte lineages [37]. Overexpression of this factor in erythroid or bipotent cell lines produces an increase in erythroid differentiation [38,39], but there are no data on overexpression in adult hematopoietic stem cells.

LMO2/RBTN2 — Another transcription factor implicated in T cell acute lymphoblastic leukemia is the LIM-only domain nuclear protein 2 (LMO2; also called rhombotin 2 [RBTN2]) [40,41]. Mice that lack this factor die in utero and have the same bloodless phenotype as animals that lack TAL1 [42]. Although LMO2 does not show sequence-specific DNA binding, immunoprecipitates in erythroid cells reveal that LMO2 exists in a complex with TAL1 [43,44], thereby suggesting a physiologic interaction in vivo.

GATA2 — GATA2 is expressed in the regions of Xenopus and Zebrafish embryos that are fated to become hematopoietic and is highly expressed in progenitor cells [33,45-47]. Overexpression of GATA2 in chicken erythroid progenitors leads to proliferation at the expense of differentiation [48].

On the other hand, targeted disruption of the GATA2 gene leads to reduced primitive hematopoiesis in the yolk sac and embryonic death by day 10 to 11 [49]. Definitive hematopoiesis in liver and bone marrow is profoundly reduced with loss of virtually all lineages, and in vitro differentiation data reveal a marked deficiency of SCF-responsive definitive erythroid and mast cell colonies and reduced macrophage colonies. This finding suggests that GATA2 serves as a regulator of genes that control hematopoietic growth factor responsiveness or proliferation of stem and/or early progenitor cells.

The importance of GATA2 to hematopoiesis and lymphopoiesis has been brought into sharp focus by the discovery that heterozygous mutations in GATA2 are the cause of a broad spectrum of previously described syndromes including monocytopenia and non-tuberculous mycobacterial infections, usually Mycobacterium avium complex, (monoMAC syndrome); dendritic cell, monocyte, B and NK lymphoid deficiency; familial myelodysplastic syndrome (MDS) and acute myeloid leukemia; and Emberger syndrome (lymphedema and MDS). In addition, susceptibility to human papilloma virus, chronic neutropenia, aplastic anemia, and pulmonary alveolar proteinosis have also been described [50]. (See "Mendelian susceptibility to mycobacterial diseases: Specific defects", section on 'GATA2 deficiency (MonoMAC syndrome)' and "NK cell deficiency syndromes: Clinical manifestations and diagnosis", section on 'Autosomal dominant GATA2 deficiency' and "Congenital neutropenia", section on 'GATA2 deficiency/MonoMAC syndrome'.)

The finding of markedly decreased expression of GATA2 mRNA in CD34+ cells in patients with aplastic anemia suggests that an aberrant expression of this transcription factor may have been an early description of this disease [51].

GATA1 — GATA1 expression is limited to multipotent progenitors and to erythroid, megakaryocyte, mast, and eosinophil lineages [52-55]. Analysis of chimeric mice injected at the blastocyst stage with GATA1^-/- embryonic stem cells reveals a failure of such cells to contribute to erythrocyte development; however, they can contribute to other hematopoietic lineages and tissues [56,57]. Differentiation assays in vitro show proerythroblast arrest and apoptosis of these cells [58-60].

Mouse embryos lacking GATA1 die because of severe anemia secondary to arrested maturation of primitive erythroid cells, while GATA1 loss in adult mice results in a condition resembling pure red cell aplasia in humans [61].

Clinically, a constitutional pathogenic variant in GATA1 that interferes with the interaction with its essential cofactor FOG1 (friend of GATA1) is associated with familial dyserythropoietic anemia, defective megakaryocyte maturation, and macrothrombocytopenia [62]. Infants with trisomy 21 and transient myeloproliferative disorder (TMD) or acute megakaryoblastic leukemia (AMKL) have acquired mutations in the 5' region of GATA1 that lead to a N-terminal truncated GATA1 protein [63,64]. These variants are almost always present at birth, and therefore acquired in utero, and have also been noted in approximately 10 percent of neonates with trisomy 21 and no evidence of TMD or AMKL [65]. It remains to be seen whether such infants will go on to develop AMKL. (See "Overview of causes of anemia in children due to decreased red blood cell production", section on 'Congenital dyserythropoietic anemia' and "Down syndrome: Clinical features and diagnosis", section on 'Hematologic disorders' and "Megakaryocyte biology and platelet production", section on 'GATA1 and NF-E2 transcription factors'.)

Germline mutations in GATA1 that also lead to loss of exon 2 and truncation of the N-terminus of GATA1 were discovered in a small proportion of males with Diamond-Blackfan anemia. These mutations lead to synthesis of a short form of GATA1 that is impaired in its ability to drive erythroid differentiation [66]. Whether the more common ribosomal protein gene mutations that account for approximately 60 percent of cases are related in some way to GATA1 deficiency was an open question, but compelling evidence from a 2014 in vitro study demonstrates that GATA1 translation is impaired in cells from Diamond-Blackfan anemia patients with different ribosomal protein gene mutations because ribosomes are limiting [67]. Ribosome deficiency adversely affects the translation of complex structured mRNAs that have low initiation rates [68]. GATA1 has a complex 5' untranslated region (UTR) that predicts poor translation initiation rates, and such mRNAs are more sensitive to ribosome deficiency than mRNAs with high initiation rates [69]. (See "Overview of causes of anemia in children due to decreased red blood cell production", section on 'Diamond-Blackfan anemia'.)

Whether precursor stem cells commit to erythroid or myeloid lineages may depend in part upon interactions between the GATA proteins (both GATA1 and GATA2) and PU.1 (Spi-1), a transcription factor essential for the development of myeloid cells. PU.1 appears to interact with a zinc finger domain of the GATA proteins via a helix-turn-helix wing [70]; this interaction appears to inhibit PU.1 activation of critical myeloid genes, as well as GATA transactivating functions, thereby suggesting that these effects have a role in the commitment of cells to either the erythroid or myeloid lineage.

Krϋppel-like factor 1 — The Krϋppel-like factor 1 (KLF1) transcription factor is a master regulator of terminal erythroid differentiation, controlling expression of many key erythroid pathways and structures, including cell division, the cell membrane and cytoskeleton, iron metabolism, and heme and globin synthesis. KLF1 was found to bind to the beta-globin promoter [71]; it controls globin expression and some aspects of erythroid development. In addition to activating beta-globin expression, KLF1 plays a role in silencing gamma-globin expression, probably by regulating BCL11A and its interaction with Sox6 at the level of the gamma-globin gene [72-74]. (See "Fetal hemoglobin (Hb F) in health and disease", section on 'Biology of Hb F' and "Structure and function of normal hemoglobins".)

Heterozygous variants in KLF1 are characterized by hereditary persistence of fetal hemoglobin (HPFH) [72] and/or congenital dyserythropoietic anemia (CDE) type IV [75]. However, most monoallelic KLF1 mutations give rise to benign phenotypes, including significant increases in levels of Hb F and Hb A2, which may ameliorate the clinical severity of beta thalassemia [76]. (See "Fetal hemoglobin (Hb F) in health and disease", section on 'Hereditary persistence of fetal hemoglobin (HPFH)'.)

In contrast, compound heterozygous variants in the KLF1 gene have been associated with an unusual hereditary persistence of embryonic globin synthesis, as well as a wide variety of hematologic abnormalities including a thalassemic phenotype with hypochromia, microcytosis, target cells, fragments and anisopoikilocytosis; a picture suggestive of chronic non-spherocytic hemolytic anemia with schistocytes, fragmented cells and acanthocytes, and pyruvate kinase deficiency; and hydrops fetalis in a neonate with compound heterozygosity for null mutations in KLF1 [77-79].

GROWTH FACTORS — The regulation of the proliferation and maturation of erythroid progenitors depends upon interaction with a number of growth factors. The availability of pure recombinant growth factors, enrichment of target progenitor cells, use of defined "serum-free" culture conditions, and the targeted disruption of factors and their receptors have provided insights into the role of these factors during hematopoiesis.

Erythropoietin — Erythropoietin (EPO) is essential for the terminal maturation of erythroid cells [80]. Its major effect appears to be at the level of the CFU-E during adult erythropoiesis; recombinant preparations are as effective as the natural hormone [81-84]. (See "Treatment of anemia in nondialysis chronic kidney disease" and "Hyporesponse to erythropoiesis-stimulating agents (ESAs) in chronic kidney disease".)

CFU-E do not survive in vitro in the absence of EPO. Since the majority of CFU-E are cycling, their survival in the presence of EPO may be tightly linked to their proliferation and differentiation to mature erythrocytes.

EPO also acts upon a subset of presumptive mature burst-forming units of the erythroid lineage (BFU-E), which requires EPO for survival and terminal maturation. A second subset of BFU-E, presumably less mature, survive EPO deprivation if other hematopoietic growth factors such as IL-3 or GM-CSF are present [85].

Since EPO is crucial for the terminal differentiation of erythroid progenitors, mice with homozygous null mutations of the EPO or EPO receptor (EPOR) genes form BFU-E and CFU-E normally but these BFU-E and CFU-E fail to differentiate into mature erythrocytes [86]. Both the EPO^-/- and EPOR^-/- mutations are embryonically lethal due to failure of definitive (fetal liver) erythropoiesis. However, yolk sac erythropoiesis is only partially impaired, indicating the existence of a population of EPO-independent primary erythropoietic precursors.

Insight into the molecular mechanism by which EPO prevents apoptosis in erythroid cells comes from studies of the signal transducer and activator of transcription 5 (STAT 5) proteins. STAT5A^-/- STAT5B^-/- mouse embryos are severely anemic during fetal development; the erythroid progenitors in fetal liver are reduced in number and show increased apoptosis [87]. This result was explained by the finding that STAT5 mediates the early induction of the anti-apoptotic gene BCLXL through direct binding to its promoter. Although adult animals were thought to be hematologically normal, about one-half have chronic anemia with splenomegaly, due to an increase in early erythroid precursors; erythropoiesis is ineffective because of an increase in apoptosis in early normoblasts showing reduced levels of BCLXL [88]. In contrast, dominant gain-of-function mutations in the EPOR due to truncation of the cytoplasmic portion of the receptor lead to familial erythrocytosis (type 1) [89] because of elimination of the negative regulatory domain of the receptor that binds an inhibitory phosphatase SH-PTP-1 [90].

Insight into biased signaling downstream of the EPOR comes from study of a recessively inherited R150Q mutation of EPO [91]. This mutation was found in a transfusion-dependent child thought initially to have Diamond-Blackfan anemia who unexpectedly did not experience improvement after a matched related hematopoietic stem cell transplant. The mutation results in alterations in the kinetics of binding of EPO to the EPOR, with an increased on-rate and a markedly increased off-rate. This results in suboptimal JAK2 phosphorylation, which is essential for STAT5 activation. Perhaps surprisingly, STAT5 activation by JAK2 showed equivalent phosphorylation in response to wild type and mutant EPO, whereas STAT1 and STAT3 activation were both reduced, indicating a bias in the signal output from the mutant receptor.

Stem cell factor — Although stem cell factor (SCF) alone has no colony-forming ability, it has marked synergistic effects on BFU-E cultured in the presence of EPO [92-95]. SCF is crucial for the normal development of CFU-E, since mice that lack SCF (Steel mutants) or its receptor KIT (White spotting or W mutants) are severely anemic and have a reduction in fetal liver CFU-E [96]. Studies of cell lines that express both KIT and EPOR (HCD57 cells) or are transduced with cDNAs for these two receptors (32D cells) demonstrate that SCF supports the proliferation of these cells only if the EPOR is also present [97].

In vitro, SCF induces tyrosine phosphorylation of the EPOR and KIT associates with the cytoplasmic domain of the tyrosine phosphorylated EPOR. Whether such an interaction between KIT and the EPOR occurs during normal erythropoiesis is not known.

IL-3 and GM-CSF — Interleukin 3 (IL-3) and GM-CSF receptors share a common beta chain. Both cytokines enhance erythropoietin-dependent in vitro erythropoiesis by primary hematopoietic progenitors and factor-dependent cells. The beta chain interacts functionally and physically with the EPOR, thereby providing an explanation for the synergistic effects of IL-3 and GM-CSF on EPO-driven erythropoiesis [98].

Insulin and insulin-like growth factor 1 — Factors distinct from the classical colony-stimulating factors may positively regulate erythropoiesis, either directly or indirectly. Limiting dilution studies of highly purified CFU-E in serum-free culture show that insulin and insulin-like growth factor 1 (IGF-1) act directly on these cells in the presence of EPO [99]. By comparison, earlier murine studies of unfractionated cells found that CFU-E respond to IGF-1 or insulin in the absence of EPO [100].

TGF-beta gene family

●Gene family members – This gene family comprises several TGF-beta and bone morphogenic proteins (BMPs), growth and differentiation factors (GDFs), and activin and inhibin. Their receptors comprise two members of the serine/threonine kinase gene family. Activin is the factor produced by vegetal cells during blastogenesis that induces animal ectodermal cells to form primary mesoderm [101]. Activin enhances both BFU-E and CFU-E colony formation.

●Effect on erythropoiesis – This protein dimer (activin plus inhibin), also known as follicle stimulating hormone-releasing protein, appears to have a lineage-specific effect on erythropoiesis that is indirect, since removal of monocytes and/or T lymphocytes abrogates its activity [102].

Murine data suggest that TGF family members may play an intriguing regulatory role during erythropoiesis [103]. Under normal circumstances, GDF11, produced by developing erythroid progenitor cells, inhibits the differentiation of more mature erythroid precursors.

●Activin and GDF11 ligand traps (luspatercept) – Activin and GDF11 ligand traps consist of the extracellular domain of the activin receptor 2A or B linked to an immunoglobulin Fc domain [104]. One possible explanation of the mechanism of action is that luspatercept and other ligand traps alleviate the anemia by removing the GDF11-mediated inhibition of erythroid differentiation. However, SMAD2/3 signaling also decreases the availability of GATA1. Ligand traps may also act by alleviating the sequestration of GATA1 by HSP40, thereby increasing its availability [105].

●Uses of luspatercept – In thalassemia and myelodysplastic syndromes (MDS), there is markedly ineffective erythropoiesis, with increased production of GDF11 and consequent decreased erythroid differentiation.

Luspatercept is approved for transfusion-dependent thalassemia and some forms of MDS [106]. Clinical uses are discussed separately:

•Thalassemia – (See "Management of thalassemia", section on 'Luspatercept for transfusion-dependent beta thalassemia'.)

•MDS – (See "Treatment of lower-risk myelodysplastic syndromes (MDS)", section on 'Luspatercept' and "Sideroblastic anemias: Diagnosis and management", section on 'MDS/MPN variants'.)

•Congenital sideroblastic anemias – (See "Sideroblastic anemias: Diagnosis and management", section on 'Luspatercept for congenital sideroblastic anemias with ineffective erythropoiesis'.)

MicroRNAs — MicroRNAs (miRNAs) are 18 to 25 nucleotide non-protein encoding RNAs that play important roles in the regulation of post transcriptional gene expression by incorporation into an RNA-induced silencing complex (RISC) that contains the Argonaute proteins. The miRNA strand guides RISC typically to complementary sequences in the 3’ untranslated region of target mRNA by Watson Crick base-pairing and leads to either mRNA degradation or translational repression.

There is strong evidence that miRNAs are important for erythropoiesis. Many miRNAs are expressed in developing erythroid cells [107,108]; most appear to decline in expression level as precursors mature, suggesting that they might regulate the timing of genes important for the synthesis of late erythroid proteins. For example, miR-191 is downregulated, and overexpression in mouse fetal liver cells results in a block in enucleation, a late event in erythropoiesis [109]. Other miRNAs that decrease in expression during differentiation act at an earlier stage of erythropoiesis. Both miR-221 and miR-222 are encoded on a common pre-miRNA; they target and repress KIT expression, so overexpression impairs the proliferation of erythroid precursors [110].

Other miRNAs, such as miR-144 and miR-451, increase in expression level during erythropoiesis; both are co-expressed in a bi-cistronic pri-miRNA that is activated by GATA1 [111]. Enforced expression of these two miRs enhances erythropoiesis while depletion of miR-451 has the opposite effect [107,112]. GATA2 is a target of miR-451 in zebrafish [113], and since down-regulation of GATA2 is required for normal erythropoiesis, this might explain the effect of miR-451 depletion. Other experiments suggest that the miR-144/451 effect is more complex, and different mouse models show different phenotypes ranging from an erythroblast block [114] to a hemolytic anemia with an increase in reticulocytes [115].

HYPOXIA AND EPO EXPRESSION — EPO has long been recognized as the physiologic regulator of red cell production [116-119]. It is produced in the kidney and the fetal liver in response to hypoxia or exposure to cobalt chloride.

Hypoxia-inducible factor and the response to hypoxia — Specialized interstitial cells in the inner cortex and outer medulla of the kidney respond to hypoxia by producing and secreting EPO. Downstream events from the oxygen sensor involved in activation of EPO gene expression require stabilization and activation of specific transcription factors [120]. One of these is the heterodimeric transcription factor complex hypoxia-inducible factor (HIF-1), comprising both HIF-1-alpha and HIF-1-beta subunits (the dimer is called HIF-1) [120-123].

While HIF-1-beta is ubiquitously expressed independent of oxygen tension, HIF-1-alpha is only detectable under hypoxic conditions (figure 4). Above a critical oxygen concentration HIF-1-alpha is ubiquitinated and rapidly degraded in proteosomes. This oxygen sensor in mammalian cells appears to be a proline hydroxylase (prolyl hydroxylase domain proteins PHD1, PHD2, and PHD3) that adds a hydroxyl moiety in an oxygen- and iron-dependent manner to P-564 of HIF-1-alpha; P-564 lies within a highly conserved region of HIF-1-alpha that binds the von Hippel-Lindau (VHL) protein. When P-564 hydroxylated HIF-1-alpha binds VHL, an ubiquitin E3 ligase complex is activated, leading to ubiquitination and subsequent destruction of HIF-1-alpha [120,124-130]. A second related protein, HIF-2-alpha, is also subject to oxygen dependent prolyl hydroxylation, and also dimerizes with HIF-1-beta to form HIF-2. HIF-1-alpha is expressed in all nucleated cells, while HIF-2-alpha expression is restricted to vascular endothelial cells, renal interstitial cells, hepatocytes, cardiomyocytes and astrocytes [120].

A homozygous R200W mutation of the VHL gene leads to reduction in the interaction of VHL with hydroxylated HIF, accumulation of HIF, and excessive erythrocyte production in Chuvash polycythemia (familial erythrocytosis, type 2) [131]. A different set of dominantly inherited mutations is found in the von Hippel-Lindau syndrome, with loss of the second normal allele in the tumors that characterize this disease. (See "Molecular biology and pathogenesis of von Hippel-Lindau disease" and "Diagnostic approach to the patient with erythrocytosis/polycythemia", section on 'Secondary polycythemia' and "Molecular pathogenesis of congenital erythrocytoses and polycythemia vera", section on 'Chuvash erythrocytosis'.)

An autosomal dominant form of erythrocytosis occurs with mutations the EGLN1 gene that encodes PHD2 (familial erythrocytosis, type 3) [132]. These mutations may have dominant negative effects because PHD2^+/- heterozygous mice do not have erythrocytosis.

Additionally, mutations in the EPAS1 gene that encodes HIF-2-alpha can also cause familial erythrocytosis (type 4) [133]. These mutations interfere with the interaction of HIF-2-alpha with PHD2, thereby reducing hydroxylation. Mice with targeted disruption of the EPAS1 gene are anemic, emphasizing the importance of HIF-2-alpha in EPO regulation [134].

HIF-1 activity is induced in a variety of cell types in response to hypoxia or cobalt [135]; this finding suggests that, while HIF-1 is important in the activation of EPO expression, it also plays a role in the activation of other hypoxia-inducible genes, including angiogenesis and glycolysis [121,136-138], and may play a role in the regulation of more than 2 percent of all human genes [139]. HIF-1-alpha may also bind to and negatively regulate the hepcidin promoter, leading to decreased levels of hepcidin [140]. Through coordinate downregulation of hepcidin and upregulation of ferroportin and erythropoietin, the VHL pathway effectively mobilizes iron in support of increased red cell production. (See "Regulation of iron balance", section on 'Hepcidin'.)

The activity of HIF-1 is essential for survival. Mice that lack HIF-1 die at midgestation, while heterozygotes, although developmentally normal and similar to normal mice in normoxic conditions, fail to respond adequately to chronic hypoxia [141]. (See "Overview of angiogenesis inhibitors", section on 'Effect of hypoxia and cytokines'.)

HIF-1 binds to an enhancer sequence located approximately 130 base pairs downstream from the poly-A addition signal of the EPO gene. This enhancer segment renders other promoter-reporter gene constructs hypoxia-responsive with typical inductions in the range of 4- to 15-fold; this degree of stimulation is significantly less than the 50- to 100-fold induction observed with the chromosomal EPO gene in Hep 3B cells [142].

Other studies have identified a 53 base pair sequence in the EPO promoter with important regulatory characteristics. This sequence confers oxygen-sensitivity (6- to 10-fold inducibility) to a luciferase reporter gene [143]. The combination of both the enhancer and promoter sequences results in cooperative (50-fold) inducibility of transcription of EPO in response to hypoxia, enhancements approaching that observed in vivo.

Hepatocyte nuclear factor 4 and p300 — Because both promoter and enhancer sequence elements include the consensus hexanucleotide nuclear hormone receptor response elements, various orphan members of this gene family have been examined for binding to the EPO promoter and enhancer and for their presence in transcription complexes isolated from Hep 3B nuclear extracts. One of these, hepatocyte nuclear factor 4 (HNF-4) is present in these extracts and plays a critical role in hypoxia-induced activation of EPO gene expression in Hep 3B cells [144]. HIF-1a and -1b also bind to a large protein p300 (homologous to the cyclic AMP response element (CREB) binding protein [CBP]) that acts as a general transcriptional activator for a number of genes [145].

The net effect is that HIF-1a, which accumulates in response to hypoxia, combines with its partner HIF-1b and with HNF-4 and p300 to activate EPO transcription. The role of HIF-2 in this process is under active investigation [146]. (See "Molecular pathogenesis of congenital erythrocytoses and polycythemia vera", section on 'EPAS1 mutations'.)

STRESS AND ERYTHROPOIESIS — The level of oxyhemoglobin and the rate of delivery of oxygen to the tissues are the fundamental regulators of erythropoiesis (figure 5) [147]. In species that package hemoglobin in erythrocytes, erythropoietin mediates the response to oxygen demand; it performs this function by interacting with specific receptors on the surface of mature committed erythroid BFU-E and CFU-E (erythroid burst-forming units and colony-forming units, respectively) [148-152]. As noted above, hypoxia induces a transcriptional increase in expression of the EPO gene via HIF-1a and HIF-1b [121,122].

BFU-E mature to single CFU-E that divide, and, under the influence of lower concentrations of erythropoietin, form single, relatively small colonies of proerythroblasts in about one week in vitro. The bone marrow of an adult mouse contains about 500 CFU-E per 10⁵ nucleated cells. In response to anemic stress, as in hemorrhage or hemolysis, almost the entire burden of accelerated reticulocyte production is borne by the rapid erythropoietin-dependent influx from the progenitor compartment into the proerythroblast pool, resulting in an expanded proerythroblast pool [153].

Anemic stress produces little or no increase in the mitotic rate of recognizable erythroid precursors [153]. Instead, the following changes are seen:

●The late BFU-E and CFU-E proliferate (because of EPO binding to its receptors) and differentiate to proerythroblasts and beyond. In normal murine marrow, the CFU-E frequency increases from about 500 per 10⁵ nucleated cells to 1000 to 2000 CFU-E per 10⁵ nucleated cells following experimental hemorrhage or hemolysis. By comparison, hypertransfused mice, in which erythropoiesis is completely suppressed, show an 80 to 90 percent reduction in number of CFU-E.

●The orderly progression from immature BFU-E through CFU-E to proerythroblast is interrupted. High erythropoietin levels permit or induce differentiation of progenitors to proerythroblasts. In the rhesus monkey, this premature terminal differentiation may account for the marked increase in fetal hemoglobin content and F cells observed during stress erythropoiesis [154]. In contrast, progenitors in humans have a lesser ability to generate erythroblasts capable of synthesizing large quantities of fetal hemoglobin [155]. As a result, the accumulation of F cells in peripheral blood in response to anemia is relatively small [156]. (See "Fetal hemoglobin (Hb F) in health and disease", section on 'Secondary increases in Hb F'.)

Red cells produced under such situations are usually macrocytic and carry the i antigen. However, these two characteristics may relate to a short transit time through the marrow in response to erythropoietin rather than an intrinsic characteristic of fetal hemoglobin-containing cells themselves. Fetal hemoglobin-containing cells (F cells) are also present in very small numbers in the blood of normal individuals, but these cells are not macrocytic and do not bear i antigen. F cell progenitors can be easily detected in the bone marrow of patients with various dyspoieses [157,158] and even in normal marrows [159].

In mice, stress erythropoiesis appears to occur largely in the spleen, where "stress" BFU-E respond to EPO, SCF, and bone morphogenic protein 4 (BMP4), which signals via the Smad5 pathway; hypoxia increases both BMP expression and the BMP4/Smad5 signaling response [160]. This contrasts with "steady state" bone marrow BFU-E that require only SCF, IL-3/GM-CSF and become responsive to EPO as they mature to CFU-E and proerythroblasts. There is also evidence that Hedgehog signaling replenishes splenic stress BFU-E from the bone marrow BFU-E pool [161].

Glucocorticoids play an important role in the erythropoietic response to stress and are also the principal treatment for Diamond-Blackfan anemia (DBA). (See "Overview of causes of anemia in children due to decreased red blood cell production", section on 'Diamond-Blackfan anemia'.)

Mouse and human studies have shown that glucocorticoids have a greater effect on BFU-E than CFU-E proliferation [162,163]. A major insight into the mechanism of this effect comes from single-cell transcriptome profiling of purified mouse fetal liver erythroid progenitors [164]. These studies show that glucocorticoids enhance erythropoiesis by slowing the transit of early through late BFU-E, without affecting the rate of the cell cycle or asymmetric self-renewal divisions. Rather, the slowed rate of progression allows an increase in the number of cell divisions, thereby increasing output into the CFU-E pool. The mechanism by which this occurs is through prolonged expression of genes such as GATA2 that antagonize expression of genes such as GATA1 that drive terminal differentiation.

SUMMARY

●Production of red blood cells – During steady-state hematopoiesis, approximately 10¹⁰ red blood cells are produced per hour in the bone marrow to maintain the hemoglobin level within fairly narrow limits. Production can be rapidly increased in the setting of ongoing blood loss or hemolysis. Erythropoiesis begins with the differentiation of a small pool of multipotent stem cells into primitive erythroid progenitors. These develop into erythroid precursors, culminating in the emergence of mature erythrocytes. (See 'Erythroid progenitor cells' above and 'Precursors and mature cells' above.)

A general discussion of hematopoiesis and hematopoietic stem cell function is presented separately. (See "Overview of hematopoietic stem cells".)

●Transcription factors – The erythroid maturation process is driven by transcription factors as well as hematopoietic growth factors and their receptors (figure 1 and figure 2). (See 'Transcription factors' above and 'Growth factors' above.)

●Erythropoietin – Erythropoietin (EPO) is the physiologic regulator of red cell production, produced in the kidney and the fetal liver in response to hypoxia due to anemia or other causes (figure 4 and figure 5). (See 'Erythropoietin' above and 'Hypoxia and EPO expression' above.)

ACKNOWLEDGMENT — UpToDate gratefully acknowledges Stanley L Schrier, MD (deceased), who contributed as Section Editor on earlier versions of this topic review and was a founding Editor-in-Chief for UpToDate in Hematology.