Iron is an essential element for almost every organism on earth. Iron can donate and accept electrons from various substrates due to its unique oxidation-reduction properties, making it an important cofactor in mammalian cells. Despite its abundance in the Earth’s crust, high oxygen availability in the atmosphere leads to iron oxidation and the formation of poorly soluble ferric iron. As a consequence, iron-dependent organisms like mammals have evolved complex mechanisms to conserve and recycle iron, preventing its loss and enabling its enhanced absorption during a deficit or in times of increased demand. The physiological status of iron in biological systems can be delineated into three categories: iron coordinated by protein side chains, iron complexed within the porphyrin ring of heme, and iron within iron-sulfur clusters. Outside of these contexts, iron displays promiscuous reactivity that can damage cells and tissues poorly equipped for handling what is termed ‘labile iron.’ Extensive coordination is also required in multicellular organisms for movement of iron between organs, in and out of cells, and between compartments within cells to effectively avoid inadequate iron supply that may limit systemic functioning and excess iron that may be toxic to cells and tissues. Thus, dysregulated iron homeostasis can manifest as total body iron deficit (iron deficiency) or excess (iron overload), as well as iron maldistribution among tissues in which individual organs may be iron-deficient while other iron-overloaded. Such iron disorders may be caused by genetic lesions that directly impair iron regulation or conditions that impact iron regulation indirectly.

Although all cells in mammalian systems require small amounts of iron, systemic iron homeostasis is mainly regulated by the specific compartments involved in iron absorption, transport, storage and recycling, and high-level utilization. These include duodenal enterocytes; serum transferrin-bound iron; hepatocytes and macrophages in the liver and spleen; and erythroid precursors in the bone marrow, respectively. The largest of these compartments is the erythron, comprising the majority of total body iron in adult humans, mainly contained within hemoglobin inside erythroblasts and ultimately red blood cells (RBCs). Said another way, erythropoiesis, even at steady state, consumes most of the transferrin bound iron in circulation. Furthermore, stimulation of increased erythropoiesis (e.g. via exogenous erythropoietin [EPO], bleeding, phlebotomy, or hypoxia) requires enhanced iron availability to keep pace with acutely increased hemoglobin synthesis within erythroblasts in the bone marrow. A large body of observational work amassed in the last >50 y suggested the presence of a direct ‘erythroid regulator’ of iron metabolism to enable this crosstalk. The last few decades have provided a series of substantial advances, enabling significant progress in our understanding of iron metabolism regulation; normal, stress, and ineffective erythropoiesis; and the crosstalk between them. In addition, the development of novel tools, both experimental methods and models of disease, has furthered and provided additional opportunities to better understand these pathways, robustly confirming and extending previously held assumptions and conjectures. This article aims to bring together our collective understanding of what has been learned and developed to study iron metabolism and erythropoiesis in health and disease.

Erythropoiesis is a continuous process required for making new RBCs in order to replace the senescent RBCs lost at the end of their lifecycle. To determine how many RBCs are required, the bone marrow relies on the kidney in which interstitial fibroblasts in the renal medulla sense hypoxia, leading to the hypoxia-inducible factor 2 (HIF-2) mediated production and excretion of EPO. EPO binding to EPO receptor on erythroid precursors in the bone marrow in turn induces their survival, cell division, and differentiation to ultimately enucleate, producing reticulocytes that mature to RBCs in the circulation. Hemoglobin synthesis in the developing erythroblasts requires iron-containing heme.

The first wave of erythropoiesis starts as early as embryonic day 7.5 (E7.5) in mouse (Palis et al., 1999; Palis and Yoder, 2001; Baron et al., 2013) and 3 wk in human (Palis and Yoder, 2001; Dzierzak and Philipsen, 2013) in the yolk sac and produces large nucleated erythroid cells from hemangioblasts (Dzierzak and Philipsen, 2013; Palis, 2008; Baron et al., 2013). Due to its transient nature, it is termed ‘primitive’ erythropoiesis to distinguish it from ‘definitive’ erythropoiesis that gradually replaces primitive erythropoiesis in the fetal liver and bone marrow, persisting thereafter throughout life. Definitive erythropoiesis produces erythroid progenitor cells from hematopoietic stem cells (HSCs) that seed and differentiate within the fetal liver by ~E14.5 in the mouse (Palis et al., 1999; Palis and Yoder, 2001; Baron et al., 2013) and 7–8 wk in the human embryo (Palis and Yoder, 2001; Dzierzak and Philipsen, 2013). After birth, sites of erythropoiesis transition from liver to spleen and eventually bone marrow and enable robust RBC production needed to reach and sustain steady-state adult erythropoiesis levels by age 7 wk in the mouse (Chen et al., 2021).

Definitive erythropoiesis is a multi-step, complex process that can be divided into three maturational stages, namely early-stage erythropoiesis, terminal erythroid differentiation, and reticulocyte maturation (Figure 1). Early-stage erythropoiesis consists of two erythroid progenitor stages, burst-forming unit-erythroid (BFU-E) and colony-forming unit-erythroid (CFU-E). BFU-E and CFU-E were initially defined by their ability to form distinct types of colonies of erythroid cells in semisolid media (Iscove and Sieber, 1975; Gregory and Eaves, 1977). With the development of flow cytometry technology in conjunction with the identification of surface markers, both murine and human BFU-E and CFU-E cells can now be identified and isolated by fluorescence-activated cell sorting (FACS) for subsequent cellular and molecular studies (Flygare et al., 2011; Li et al., 2014; Zhang et al., 2021). Furthermore, a recent study showed that human erythroid progenitor (EP) populations can be further subdivided into four subpopulations with EP1 representing predominantly BFU-E with EP2, EP3, and EP4 representing increasingly mature CFU-E populations with reduced proliferative responsiveness to stem cell factor (SCF), indicating heterogeneity of previously identified CFU-E population (Yan et al., 2021).

Terminal erythroid differentiation is a process by which proerythroblasts (Pro) differentiate sequentially to basophilic (Baso), polychromatic (Poly), and orthochromatic (Ortho) erythroblasts that expel their nuclei to become reticulocytes (Figure 1). During terminal erythroid differentiation, several notable changes occur, enabling one to selectively identify erythroblasts at distinct developmental stages based on morphological features. These changes include decreasing cell size, hemoglobin accumulation, chromatin condensation, and enucleation. Previously, Giemsa staining and morphological assessment by light microscopy were the only means for identifying at these terminally differentiating erythroblast stages. Currently, isolating and purifying both murine and human erythroblasts at distinct stages can be accomplished using surface markers and FACS analysis (Chen et al., 2009; Liu et al., 2013; Hu et al., 2013). These novel methods for isolating erythroid lineage cells at each stage have not only enabled the study of normal and disordered erythropoiesis in a stage-specific manner (Liu et al., 2013; Hu et al., 2013) but have also facilitated global molecular characterization of erythropoiesis. Various global omics analyses have been performed using purified erythroid cell populations (Li et al., 2014; An et al., 2014; Edwards et al., 2016; Schulz et al., 2019). Transcriptome analyses of human and murine erythroblasts reveal significant stage- and species-specific differences across stages of terminal erythroid differentiation (An et al., 2014). The epigenetic landscape of human erythropoiesis reveals that erythroid cells exhibit chromatin accessibility patterns distinct from other cell types. It also reveals stage-specific patterns of gene regulation (Schulz et al., 2019). Global omics analyses have also been performed on primitive erythroblasts during embryogenesis that revealed molecular similarities and differences between primitive and definitive erythropoiesis (Nemkov et al., 2022). The omic databases generated during the last decade not only contribute novel understanding of normal erythropoiesis regulation but also provide insight into disease pathophysiology and serve as rich resources for future studies.

Enucleation is a distinctive feature of mammalian erythropoiesis. Given the physiological significance of enucleation in generating highly deformable RBCs for effective oxygen delivery, understanding the mechanistic basis of enucleation has been an active area of investigation. Cytoskeleton proteins such as actin (Takano-Ohmuro et al., 1996; Ji et al., 2008; Watanabe et al., 2013; Ubukawa et al., 2020; Liu et al., 2021), tubulin (Chasis et al., 1989; Wang et al., 2012), myosin (Takano-Ohmuro et al., 1996; Ubukawa et al., 2012), tropomodulin (Sui et al., 2014), and chromatin condensation and lipid rafts (Ji et al., 2010; Konstantinidis et al., 2012; Malik et al., 2017; Zhao et al., 2019; Wang et al., 2021; Jeffery et al., 2021) are essential for normal enucleation. Although the function of actin and tubulin in enucleation suggests that erythroblast enucleation is a form of asymmetric cytokinesis, evidence documents that vesicle trafficking rather than cytokinesis is required for enucleation (Keerthivasan et al., 2010). This conclusion is supported by a recent finding that vesicle formation regulated by ERK/MAPK pathway mediates human erythroblast enucleation (An et al., 2021). Furthermore, although some evidence suggests that actin forms a contractile ring to help expel the nucleus (Ji et al., 2008), other data demonstrates that a dedicated cytoskeletal assembly in the cytoplasm, the ‘enucleosome,’ located contralateral to the site of enucleation, that is, at the rear of the nucleus, is most likely the driver of nuclear expulsion (Nowak et al., 2017). These findings indicate that despite relatively extensive studies on enucleation, the relevant mechanisms remain incompletely understood, somewhat controversial, and would benefit from further investigation.

Reticulocyte maturation is the final step of erythropoiesis. During this process, a series of major changes occur. These include membrane surface area loss via membrane vesiculation (Waugh et al., 1997), organelle (mitochondria and ribosome) clearance via autophagy (Gronowicz et al., 1984; Kundu et al., 2008; Zhang et al., 2009b), and membrane skeleton reorganization (Chasis et al., 1989; Liu et al., 2010; Liu et al., 2011). These changes together lead to fully functional mature RBCs with maximum hemoglobin carrying capacity and flexible yet stable membranes.

Erythropoiesis is tightly regulated by multiple soluble factors. To determine how many RBCs are required, the bone marrow relies on the kidney in which interstitial fibroblasts in the renal medulla sense hypoxia, leading to the HIF-2-mediated production and excretion of EPO. Very recent preliminary data using single-cell RNA and transposase-accessible chromatic (ATAC) sequencing to molecularly identify EPO-producing cells indicate that a distinct population of renal stromal cells, termed Norn cells, are the main source of EPO production in mice and humans (Kragesteen et al., 2023). EPO binding to EPO receptor (EPOR) on erythroid precursors in the bone marrow in turn induces their survival, cell division, and differentiation to ultimately produce RBCs. EPO and EPOR are indispensable for definitive erythropoiesis. Deletion of EPO or EPOR leads to embryonic lethality at approximately E13 due to severe anemia associated with defects in definitive erythropoiesis in mice (Wu et al., 1995). Other cytokines, growth factors, and hormones such as stem cell factor, interleukin-3, insulin-like growth factor 1, and glucocorticoids, although not essential for erythropoiesis, also promote proliferation of erythroid progenitors (Sonoda et al., 1988; Sonoda et al., 1994; Kolbus et al., 2003; Perry et al., 2007; Flygare et al., 2011). At the transcriptional level, red cell development is regulated by multiple transcription factors (Dzierzak and Philipsen, 2013; Andrieu-Soler and Soler, 2022), two of which, GATA1 and KLF1, are considered master regulators, indispensable for normal erythropoiesis (Pevny et al., 1995; Siatecka and Bieker, 2011).

Similar to EPO, iron is also essential for erythropoiesis. Hemoglobin synthesis in the developing erythroblasts requires iron-containing heme. While EPO allows survival of erythroid progenitors and early-stage erythroblasts by activating transcription of anti-apoptotic genes via the EPO/EPOR/JAK2/STAT5 signaling pathway (Witthuhn et al., 1993), iron modulates EPO responsiveness (Khalil et al., 2018) and is also essential for differentiation of early- to late-stage erythroblasts, during which time iron incorporation into the protoporphyrin ring is required as the last step in heme synthesis. EPO can thus be regarded as a ‘driver’ of erythropoiesis, while iron acts as a ‘modulator’ and also serves as the fuel for the production of RBCs. In addition, recent data demonstrates that iron may be involved in the modulation of EPO responsiveness via the regulatory function of monoferric transferrin.

Hepcidin expression is predominantly regulated by iron, inflammation, and erythropoiesis (Figure 4A). First, HAMP transcription in hepatocytes is upregulated by iron loading and suppressed by iron deficiency as well as expanded or ineffective erythropoiesis. The regulation of hepcidin by iron is incompletely understood. Murine models suggest hepatocytes sense local and systemic iron status by binding bone morphogenetic proteins (BMPs), primarily BMP6, BMP2, and/or their heterodimers via BMP receptors, and transferrin-bound iron via TFR1 and TFR2, leading to signaling to induce hepcidin expression (Camaschella et al., 2020; Figure 4B). Second, hepatocyte hepcidin expression is mediated indirectly by iron in response to iron-induced BMP production by liver sinusoidal endothelial cells (LSECs) (Enns et al., 2013). Recent evidence demonstrates that BMP6 expression in primary LSEC ex vivo is induced in response to iron only when co-cultured with primary hepatocytes or supernatants from primary hepatocyte cultures (Colucci et al., 2022). Additional recent effort provides evidence for a minor functional role of LSEC TFR1-mediated iron uptake and BMP6 induction in iron limited conditions and TFR1-independent iron-mediated regulation of LSEC BMP6 expression in iron-rich conditions (Fisher et al., 2022). Although how BMP expression is induced in LSEC is not fully understood, the BMP pathway is critical for hepcidin the regulation by iron (Truksa et al., 2006; Babitt et al., 2007). Specifically, BMP6 and BMP2 binding hepatocellular BMP receptor triggers phosphorylation and signaling via SMAD1/5/8 that, coupled with SMAD4, translocate to the nucleus to induce HAMP expression (Figure 4B). Third, several hepatocellular surface molecules modulate HAMP activation in response to iron status, enabling hepatocytes to directly sense iron via expression of TFR1, TFR2, and HFE. To briefly delineate, HFE association with TFR1 under low iron conditions is displaced when TFR1 binds monoferric or diferric transferrin (Feder et al., 1998; Bennett et al., 2000; Giannetti and Björkman, 2004; Lebrón et al., 1998). Although a mechanistic understanding of how TFR2 contributes to hepcidin regulation remains unclear, we surmise that as serum iron concentration increases, increased transferrin:TFR2 binding induces TFR2 membrane stabilization (Johnson and Enns, 2004; Robb and Wessling-Resnick, 2004) and possibly HFE binding to TFR2. This HFE:TFR2 complex interacts with hemojuvelin (HJV), the iron-specific BMP co-receptor, and potentiates the BMP signaling pathway to HAMP expression (Figure 4B). Furthermore, the specific mechanism by which HJV regulates hepcidin is incompletely understood. For example, recent evidence demonstrates that HJV interaction with neogenin, a ubiquitously expressed transmembrane protein, is required for hepcidin regulation (Enns et al., 2021). Finally, the pathway is negatively regulated by the transmembrane serine protease matriptase 2 (i.e. TMPRSS6), which by binding HJV, BMP receptor, and/or HFE decreases signaling to HAMP expression (Enns et al., 2020). Thus, both TFR2 and HFE:TFR1 complex function as the main iron sensors (Schmidt et al., 2008; Robb and Wessling-Resnick, 2004) and communicate systemic iron status to modify hepatocyte hepcidin production and secretion with multiple co-factors modulating this signal.

Interestingly, hepatocyte TFR1 also influences systemic iron homeostasis by interacting with the hemochromatosis protein HFE to regulate hepcidin production (Fillebeen et al., 2019; Xiao et al., 2023). Although loss of hepatic Tfrc is not associated with grossly altered iron metabolism, hepatocyte-selective Tfrc knockout mice show predisposition to anemia making their unchanged hepcidin levels inappropriately high relative to serum and liver iron concentrations and ERFE levels. In addition, ablation of hepatocyte Tfrc does not modify the iron phenotype in Hfe knockout mice. Lack of Tfrc also ameliorates hepcidin deficiency and liver iron loading.

Hemoglobin synthesis in erythroblasts requires large amount of iron, providing a strong rationale for erythropoiesis-mediated regulation of iron availability. For instance, stimulated erythropoiesis (e.g. in response to bleeding, repeated or large volume phlebotomy, hypoxia, or administration of exogenous EPO) leads to increased iron absorption, and the last few decades have provided a more robust mechanistic understanding of how iron availability is regulated by erythropoiesis. For the purposes of this article, we will discuss several aspects of this crosstalk, including how iron is taken up and chaperoned in erythroblasts, how erythropoiesis modulates iron metabolism directly and indirectly, and how iron metabolism itself impacts erythropoiesis.

Although there is a broad range of anemia-causing mechanisms, we will focus in this review on causes related to iron metabolism. In addition to iron deficiency anemia, these foremost include anemia of chronic inflammation (ACI). In addition, iron refractory iron deficiency anemia (IRIDA) is an interesting albeit rare form of IDA discussed here. In a subsequent section below, we will also discuss iron metabolism in the context of ineffective erythropoiesis in iron loading anemias, namely β-thalassemias and MDS.

Ineffective erythropoiesis can be defined as the diminished production of enucleated RBCs despite an increase in the number of erythroid precursors (Figure 6). When the quantity and/or the ability of enucleated RBCs to transport oxygen declines below a certain level, patients require RBC transfusions for survival (Kattamis et al., 2022; Ginzburg and Rivella, 2011). Ineffective erythropoiesis has been a subject of intensive investigation in β-thalassemia, a disease in which ineffective erythropoiesis manifests with an expanded number of erythroid precursors and their reduced ability to differentiate, an increase in erythroblast death in these terminally differentiated cells, and reduced survival of enucleated RBCs (Ginzburg and Rivella, 2011). The proposed underlying mechanism leading to ineffective erythropoiesis in β-thalassemia is the relative excess of alpha-globin chains (EACs) (Kattamis et al., 2022; Ginzburg and Rivella, 2011). EACs alter erythropoiesis in at least two ways: in complex with free heme molecules, they form hemichromes, which are the main source of oxidative stress and cell death (Kattamis et al., 2022). In addition, EACs reduce the stability of GATA1, the main erythroid transcription factor in erythroblasts, interfering with their survival and maturation (Arlet et al., 2014). Furthermore, the abnormal and reduced number of RBCs in circulation lead to hypoxia-mediated EPO production, which in turn exacerbates further the expanded number of immature erythroid precursors (Libani et al., 2008).

Expansion in the number of erythroid precursors leads to increased ERFE production and consequent hepcidin suppression (Kautz et al., 2014). Hypoxia also plays a role by increasing expression of genes responsible for iron absorption in the duodenum (Anderson et al., 2013). Although in severely affected patients organ iron overload develops because of frequent RBC transfusions, in individuals with β-thalassemia who are not regularly transfused, the mechanisms responsible for increased iron absorption (i.e. insufficiently elevated hepcidin expression), along with chronic hemolysis, lead to progressive tissue iron deposition and toxicity, requiring even in these cases the use of iron chelators to prevent significant morbidity and mortality (Kattamis et al., 2022; Musallam et al., 2021).

Novel drugs and genetic approaches are now being translated to improve the quality of life in β-thalassemia (and other) patients or even cure them. Although a full description of these therapeutics is beyond the scope of the current review, these drugs can be broadly classified based on their mechanism of action as shown in Table 1. Among these, only gene therapy provides a curative approach. However, the risks remain high and alternative therapeutic options are welcome for those patients who are ineligible for cure. Drugs that act on iron metabolism (e.g. hepcidin-mimetics and transferrin) could also improve RBC quality and survival by limiting erythroid iron intake and hemichromes formation, as shown in mouse models of β-thalassemia (Li et al., 2010; Guo et al., 2013). However, even these drugs may fail to do so in β-thalassemia patients, their use could reduce iron absorption in combination with iron chelators if the patient is already iron overloaded. In addition, they could prevent iron from being accumulated in combination with drugs that improve RBC quality and production (e.g. luspatercept, mitapivat, and TFR2 inhibitors).

Beyond the inherited forms of iron-loading anemias, that is, β-thalassemia, MDS is an acquired form of ineffective erythropoiesis associated with iron overload. MDS is a heterogeneous group of bone marrow stem cell disorders; several subtypes of MDS are characterized by ineffective erythropoiesis, leading to blood cytopenias and increased incidence of transformation to acute myeloid leukemia (Haferlach, 2019). The majority of MDS patients have a long median survival (e.g. 10 y), with 30–50% requiring only regular RBC transfusions to alleviate anemia and its associated symptoms (Dayyani et al., 2010; Kröger, 2019). RBC transfusions, however, are the main cause of progressive iron overload and consequent end-organ damage in transfusion-dependent MDS patients (Oliva et al., 2010). However, the risk–benefit ratio of treating iron overload in MDS patients remains controversial. Furthermore, RBC transfusion-dependence and iron overload correlate strongly with decreased survival in MDS patients (Malcovati et al., 2005; Malcovati et al., 2006; Garcia-Manero et al., 2008; Malcovati et al., 2011). In vitro experiments demonstrate that excess iron inhibits erythroid lineage differentiation in both murine and human hematopoietic progenitors, exhibiting dysplastic changes with increased intracellular reactive oxygen species (ROS), decreased expression of anti-apoptotic genes, and DNA damage, triggering apoptosis, and worsening disease in MDS (Pan et al., 1999; Camaschella et al., 2007; Fibach and Rachmilewitz, 2012; Taoka et al., 2012; Hartmann et al., 2013).

The TELESTO trial was designed and executed to test whether iron chelation provides clinical benefit in iron-overloaded lower risk MDS patients (Angelucci et al., 2020). The results demonstrate prolonged event-free survival in deferasirox-treated MDS patients without clear improvement in hemoglobin, reduction of RBC transfusion burden, or effect on overall survival. We hypothesize, based on evidence from mutant monoferric mice (see above), that the lack of effect of deferasirox on erythropoiesis is a consequence of the specific iron chelator selected. Deferasirox and most other commercially available iron chelators have a higher iron binding affinity relative to transferrin (Sohn et al., 2008); only deferiprone iron-chelating effect enables iron transfer from parenchymal cells to increased transferrin saturation (Schmidt et al., 2015; Casu et al., 2016a), potentially modulating monoferric transferrin concentrations. Furthermore, increasing transferrin saturation itself results in increased hepcidin expression in the liver (Schmidt et al., 2015; Casu et al., 2016b), in turn preventing further iron absorption and recycling, trapping iron within macrophages (Ganz, 2005), and decreasing iron availability for erythropoiesis to potentially ameliorate ineffective erythropoiesis in MDS. A direct beneficial effect of DFP on erythropoiesis in MDS has yet to be demonstrated. Recent data demonstrates partial reversal of ineffective erythropoiesis in addition to iron overload, normalizing erythroblast iron trafficking and restoring EPO responsiveness in deferiprone-treated mouse model of MDS, NUP98-HOXD13 transgenic mice (An et al., 2022).

In addition, a specific subtype of low-risk MDS, namely MDS with ringed sideroblasts, occurs in 20% of all MDS patients. Splicing factor 3 B subunit 1 (SF3B1) mutations in hematopoietic stem and progenitor cells is a hallmark of this disease, inducing aberrant splicing of genes involved in heme biosynthesis and mitochondrial iron transport, leading to the abnormal deposition of iron in erythroblasts, and resulting in dysfunctional hemoglobin synthesis and formation of ringed sideroblasts (Visconte et al., 2015; Dolatshad et al., 2015; Dolatshad et al., 2016; Shiozawa et al., 2018; Clough et al., 2022). Although preclinical data in mouse models predicted a therapeutic effect of splicing inhibition, a recent phase I clinical trial did not yield significant clinical improvement (Lee et al., 2016; Steensma et al., 2021). Growing evidence suggests that the heme-regulated EIF2AK1 kinase pathway affects erythropoiesis in health and disease. For example, EIF2AK1 effector DDIT3 is overexpressed in MDS hematopoietic stem and progenitor cells (Berastegui et al., 2021), DDIT3 overexpression delays erythroid differentiation in CD34-positive cells from MDS patients, and EIF2AK1 inhibition increases expression of both mitochondrial heme biosynthesis enzymes and iron transporters, reversing SF3B1 mutation-induced arrest of erythroid differentiation in vitro (Adema et al., 2022).

Lastly, novel therapy has recently been US Food and Drug Administration (FDA) and European Medicines Agency (EMA)-approved for very low- to intermediate-risk MDS with ringed sideroblasts. Luspatercept is a modified activin receptor IIB ligand trap, a member of the transforming growth factor-β (TGF-β) superfamily. While this agent was studied in multiple different MDS subgroups of patients in a phase II trial, its efficacy was most robust in patients with MDS with ringed sideroblasts (Platzbecker et al., 2017) and led to the phase III MEDALIST trial, a double-blind, placebo-controlled, multicenter study in transfusion requiring patients with MDS with ringed sideroblasts (Fenaux et al., 2018). In the MEDALIST trial, luspatercept led to transfusion independence for >8 wk in 45% of enrolled subjects and median duration of response was 6 mo (Farrukh et al., 2022). To consider how MDS with ringed sideroblasts is unique, prior evidence demonstrates that these patients exhibit iron overload prior to the initiation of RBC transfusion (Gattermann, 2005) likely as a consequence of especially suppressed hepcidin relative to other MDS subtypes (Gu et al., 2013; Santini et al., 2011) and HFE gene polymorphisms that predispose to iron overload are detected in up to 21% of MDS with ringed sideroblasts, significantly higher than in other MDS subtypes (Nearman et al., 2007; Valent et al., 2008). A more complete understanding of how the formation of ringed sideroblasts in MDS contributes to worsening ineffective erythropoiesis is currently lacking. Taken together, these recent findings provide supporting evidence for dysregulated iron trafficking in the pathophysiology of ineffective erythropoiesis in MDS.

Polycythemia vera (PV), one of the chronic myeloproliferative neoplasms, is a clonal hematopoietic stem cell disorder driven by EPO hypersensitive signaling via the JAK2-STAT5 pathway, resulting in excess proliferation of erythroid precursors (Rampal et al., 2014; Levine et al., 2005; Baxter et al., 2005; Kralovics et al., 2005; James et al., 2005; Lu et al., 2008; Akada et al., 2010). The vast majority of PV patients are the result of acquired JAK2 mutations in their stem cells, namely JAK2V617F on exon 14 (95% of PV patients) or mutations in exon 12 of the JAK2 gene (2–3% of PV patients) (Pardanani et al., 2007; Scott et al., 2007). PV patients are frequently iron deficient at the time of diagnosis (Gianelli et al., 2008; Thiele et al., 2001; Kwapisz et al., 2009), and this is further exacerbated by therapeutic phlebotomies administered with the goal of maintaining hematocrit below 45% to decrease thrombotic risk (Marchioli et al., 2013). Repeated phlebotomies may in part dampen erythropoiesis by inducing iron deficiency but also potentially contribute to PV-associated systemic symptoms due to the depletion of iron stores in non-hematopoietic tissues (Pratt and Khan, 2016). Recent analysis of PV patients treated with ruxolitinib, a JAK1/2 inhibitor, corroborates the role of iron deficiency in the manifestations of this disease as symptom improvement with ruxolitinib is at least partly attributable to reversal of systemic iron deficiency (Verstovsek et al., 2017).

We previously demonstrate that PV, compared to secondary forms of erythrocytosis, is associated with relative suppression of hepcidin, potentially due to more expanded erythropoiesis and iron depletion (Ginzburg et al., 2018). In addition, PV patients experience erythrocytosis despite a more profound iron deficiency relative to healthy blood donors (Feola et al., 2019; Stetka et al., 2023). This is evidenced by significantly lower MCVs, serum iron and ferritin concentrations, and transferrin saturation; this systemic iron deficiency in PV patients does not resolve despite elevated ERFE with consequent hepcidin suppression. Hepcidin suppression would be expected to result in enhanced intestinal iron absorption and mobilization of intracellular recycled and stored iron, leading to cellular iron efflux, more circulating iron, and recovery from systemic iron deficiency. However, recovery from iron deficiency does not occur in PV, where a low hepcidin state is insufficient to replenish iron stores, implying dysregulated iron homeostasis.

We hypothesize that relative hepcidin suppression without recovery from iron deficiency in PV may result from the combined effects of concurrent inflammation, insufficiently elevated ERFE with insufficiently suppressed hepcidin, and/or aberrant hypoxia signaling in the intestine preventing recovery from iron deficiency (Nemeth and Ganz, 2014; Frise et al., 2016; Shah et al., 2009). A recent report suggests that ERFE exerts a relatively diminished effect on hepcidin regulation relative to that of inflammation in PV patients (Bennett et al., 2023). In this study, Erfe deletion in Jak2^V617F mice did not alter hepcidin levels or disease severity, resulting in stably elevated hematocrits and RBC counts. Using a human hepatocyte cell line, HepG2 cells, the authors further explored the hypothesis that PV-associated inflammation leads to hepcidin upregulation, demonstrating that the increased hepcidin expression was induced by plasma from PV patients but not plasma from normal controls and was normalized by blocking IL-6 binding to its receptor (Bennett et al., 2023). These findings suggest that inflammatory cytokines in PV may be crucial to disordered iron utilization. In addition, persistent erythropoiesis despite iron deficiency in PV may also occur as a consequence of aberrant erythropoiesis that is insensitive to iron deficiency, preventing physiological mechanisms that normally coordinate iron supply with erythropoietic output (Khalil et al., 2017; Khalil et al., 2018; Feola et al., 2019). An excellent review on dysregulated iron metabolism in PV was recently published (Ginzburg et al., 2018).

To elucidate briefly, suboptimally suppressed hepcidin prevents recovery from iron deficiency, enables absorption of iron to maintain pathologically enhanced erythropoiesis, and provides a rationale for maximizing this finding for therapeutic purposes in PV. Said another way, although we do not yet understand the pathophysiological mechanism that enables persistent erythropoiesis in PV despite iron deficiency, we anticipate that using hepcidin mimetics to further suppress iron absorption and recycling may prevent erythropoiesis in PV, redistributing iron to non-hematopoietic cells and possibly reversing iron deficiency associated symptoms in PV patients.

Lastly, significant advances in the translation of hepcidin mimetics in PV are worth noting. Our understanding about hepcidin’s mechanism of action predicts that hepcidin elevation would be expected to sequester recycled and stored iron and prevent iron absorption, resulting in reduced iron availability for erythropoiesis and replenishing iron stores within liver and splenic macrophages, thus aiding in recovery from systemic iron deficiency (Casu et al., 2018, Aschemeyer et al., 2018; Ginzburg et al., 2018; Ginzburg, 2019; Figure 7). Preclinical studies demonstrated proof of principle for this approach using minihepcidins, engineered peptides with the necessary functional ferroportin binding domain (Preza et al., 2011), resulting in a significant dose-dependent decrease in RBC count, hematocrit, and splenomegaly in Jak2^V617F mice, a well-established mouse model of PV (Casu et al., 2016b). In addition, minihepcidin results in increased iron in the splenic red pulp of Jak2^V617F mice, consistent with sequestration of recycled iron. More recently, another hepcidin mimetic agent—antisense oligonucleotide targeting TMPRSS6, leading to the downregulation of TMPRSS6 gene product that prevents the degradation of HJV, yielding an increase in endogenous hepcidin expression in the liver—used in Jak2^V617F mice also resulted in decreased RBC counts and hematocrit levels as well as suppression of bone marrow erythroblast numbers (Casu et al., 2021). Similar findings were also recently demonstrated using a parenteral synthetic hepcidin (Taranath et al., 2021) and an orally bioavailable ferroportin inhibitor (Stetka et al., 2023).

Most recently, preliminary results from phase II clinical trials evaluating the safety and efficacy of hepcidin mimetic rusfertide (PTG-300) in phlebotomy-requiring PV patients demonstrate a virtual elimination of phlebotomy requirements, control of RBC count, increase in systemic iron stores, and a potential decrease in systemic symptoms (Hoffman, 2021; Hoffman et al., 2022; Ginzburg et al., 2021). A dramatic reduction in phlebotomy requirements, with 84% of PV subjects achieving phlebotomy-independence, was observed in the first 28 wk of treatment, and hematocrit control was sustained for up to 2 y on study drug. Several other hepcidin-inducing agents are either enrolling PV patients to a phase II clinical trial (NCT05143957) or in planning stages, and the global, multicenter, randomized, placebo-controlled phase III trial (NCT05210790) is currently underway (Verstovsek et al., 2021) to further clarify the potential role of rusfertide in the management of patients with PV.

Erythropoiesis occurs at the erythroblastic island (EBI) that is composed of a central macrophage surrounded by developing erythroid cells (Bessis, 1958) and granulocyte progenitors (Romano et al., 2022). The functional role of EBI was first suggested by Mohandas and colleague, who showed that in hyper-transfused rats, the numbers of EBI in the bone marrow were significantly decreased (Mohandas and Prenant, 1978). The importance of the central macrophage in supporting normal erythropoiesis was further supported by the abnormal macrophage differentiation in EMP-null (Wei et al., 2019; Soni et al., 2006), KLF1-null (Mukherjee et al., 2021; Porcu et al., 2011), and other mouse models (Chow et al., 2013; Sadahira et al., 1995; Kawane et al., 2001; Mankelow et al., 2004), leading to significantly impaired erythropoiesis and anemia. Furthermore, the depletion of macrophages with either clodronate liposomes or CD169-diptheria toxin leading to impaired erythropoiesis provide direct evidence that macrophages play critical roles in supporting erythropoiesis in vivo, particularly during stress erythropoiesis (Chow et al., 2013; Ramos et al., 2013). In vitro studies showed that macrophages promoted erythroblast proliferation/survival (Rhodes et al., 2008; Lopez-Yrigoyen et al., 2019; Perron-Deshaies et al., 2020). It has been reported that fetal liver macrophages can efficiently engulf extruded nuclei in a phosphotidylserine-dependent manner (Yoshida et al., 2005) and that failing to degrade the engulfed DNA by macrophages due to lack of DNAase II led to severe anemia and embryonic death (Sadahira et al., 1995). Notably, in both Jak2^V617F (PV mouse model) and Hbb^3/+ (transfusion independent β-thalassemia mouse model) mice, macrophage depletion normalized erythropoiesis (Ramos et al., 2013). Together, these findings indicate that macrophages play important roles in supporting normal erythropoiesis and can be targeted to at least partly ameliorate the disordered erythropoiesis in PV and β-thalassemia. However, due to the inability to identify and isolate EBI macrophages for cellular and molecular studies, the mechanisms by which EBI macrophages support normal erythropoiesis or contribute to disordered erythropoiesis have not been fully interrogated. We recently discovered that EBI macrophages are characterized by the expression of EPOR (Li et al., 2019; Zhang et al., 2021) and that EPO enhanced the ability of macrophages to form EBIs with erythroblasts both in vitro and in vivo (Li et al., 2019). Supporting the functional role of EPO/EPOR in EBI macrophages, others also documented that Epo/EpoR signaling in macrophages is required for stress erythropoiesis in the spleen (Chen et al., 2020). In addition, fine characterization of EBIs indicates neutrophil precursors specifically associated with BM EBI macrophages, suggesting that erythro-(myelo)-blastic islands are a site for terminal granulopoiesis and erythropoiesis (Romano et al., 2022). Finally, the relative proportion of granulocytes within EBIs increases during inflammatory conditions and decreases during stress erythropoiesis, suggestive of a functional plasticity of the central macrophage within the EBIs (Romano et al., 2022).

To develop a comprehensive characterization of the EBI macrophages at the molecular level and gain insights into the mechanisms by which they support erythropoiesis, we performed RNA-seq analyses on the sorted bone marrow F4/80⁺EpoR⁺ and F4/80⁺EpoR^- macrophages (Li et al., 2019). Bioinformatics analyses revealed that the expression levels of Vcam1 and CD169 known to be involved in macrophage–erythroblast interaction (Chow et al., 2013; Sadahira et al., 1995) were significantly higher in F480⁺EpoR⁺ than in F480⁺EpoR^- macrophages. Similarly, the expression levels of Mertk required for pyrenocyte engulfment (Toda et al., 2014), and DNase2α (DNAse2) critical for DNA degradation of the engulfed nuclei (Kawane et al., 2001), were also significantly higher in the F480⁺EpoR⁺ macrophages. Intriguingly, key molecules involved in iron recycling such as phosphotidylserine receptor Tim4, heme oxygenase-1, iron exporter ferroportin, and iron transporter transferrin are also abundantly expressed in bone marrow F4/80⁺EpoR⁺ macrophages. In addition, insulin growth factor 1, one of the known erythropoiesis-promoting cytokines, is expressed in EBI macrophages but not non-EBI macrophages. These findings provide support for the long-standing expectation that EBI macrophages are unique in providing essential elements to enable differentiation of the surrounding erythroblasts. Further evidence is forthcoming regarding the mechanistic nature of the support EBI macrophages provide to differentiating erythroblasts.

To study erythropoiesis, it is important to identify and isolate erythroid lineage cells at distinct stages of differentiation. During the past decade, considerable progress has been made, and methods for analyzing and isolating murine and human erythroblasts at distinct developmental stages have been developed. Here, we summarize these methods.

Taken together, a great deal is currently known about the physiology of erythropoiesis and iron metabolism as well as the pathophysiology of diseases in which these biological systems are dysregulated. Despite this, significant unknowns remain, both regarding the mechanisms of normal function and their disordered regulation in disease; we thus include here an incomplete list to guide the next generation of targeted inquiry along these lines: