Regulation of erythropoietin production

Introduction

The hormone erythropoietin (Epo) is essential for red blood cell (RBC) production. The relationship between the O₂ content of the blood and erythropoiesis was first described by the French anatomist Francois-Gilbert Viault in 1890 (Viault, 1890), who observed a rise in RBC numbers on a journey to the highlands of Peru (Morococha, about 4500 m). Indeed, the specific stimulus for Epo expression is a fall in tissue O₂ pressure (). Epo production increases under hypoxic conditions in the kidneys and, in minor amounts, in distinct other organs such as the liver and the brain. This article summarises the present understanding of the control of Epo production.

Physiology of Epo

Epo is primarily expressed by hepatocytes during the fetal state. After birth, peritubular fibroblasts in the renal cortex become the main production site. Epo synthesis is regulated at the transcriptional level. Epo mRNA is also detectable in brain, liver, spleen, lung and testis, but these organs are not able to substitute for renal Epo in chronic kidney disease (CKD). Brain-derived Epo acts locally as a neuroprotective factor (see Noguchi et al. 2007). Epo is an acidic glycoprotein of about 30 kDa and comprises 165 amino acids and four glycans. Circulating Epo exhibits several glycosylation isoforms that differ in electrical charge and biological activity. Epo amounts are usually expressed in international units (IU), with one IU exerting the same erythropoiesis stimulating activity in rodents as 5 μmol cobaltous chloride (see Jelkmann, 2007).

Systemic Epo is an anti-apoptotic agent for erythrocytic progenitors, predominantly the colony-forming units-erythroid (CFU-Es). In response to Epo the CFU-Es proliferate and differentiate to generate cohorts of proerythroblasts and normoblasts (Fig. 1). The human haematopoietic Epo receptor (Epo-R) is a 484 amino acid glycoprotein of about 60 kDa, which belongs to the cytokine class I receptor family and forms homodimers. On the binding of Epo to the Epo-R dimer, cytoplasmic Janus kinases 2 (JAK2) catalyse the phosphorylation of tyrosine residues of the Epo-R and of various intracellular proteins (enzymes and transcription factors). Erythropoiesis is a slow-acting process. Following a rise in plasma Epo it takes 3–4 days before reticulocytosis becomes apparent.

Figure 1

Diagram of the feedback regulation of erythropoiesis

Lack of O₂ (hypoxia) is a stimulus for the synthesis of erythropoietin (Epo), primarily in the kidneys. Epo is a survival, proliferation and differention factor for the erythrocytic progenitors, particularly the colony-forming units-erythroid (CFU-Es). The O₂ capacity of the blood increases with the enhanced release of reticulocytes. The role of extra-renal sites (brain, skin) in the control of the renal Epo synthesis is still incompletely understood. BFU-E, burst-forming unit-erythroid.

Epo is essential for erythropoiesis. However, the action of Epo is augmented by several other hormones, namely testosterone, somatotropin and insulin-like growth factor 1. The higher RBC counts and haemoglobin concentrations [Hb] in men compared to women result from the stimulation of erythropoiesis by androgens and its inhibition by oestrogens.

A role has been proposed for Epo as a cytoprotective agent for several non-haematopoietic tissues, including the brain, the heart, blood vessels and the kidneys (Brines & Cerami, 2006). Some investigators believe that the effects of Epo on non-erythrocytic cells are mediated by a heterotrimeric receptor consisting of one Epo-R molecule in complex with a dimer of the common cytokine β receptor (βcR). The Epo-R/βc-R concept is mainly based on the observation that erythropoietically inactive Epo derivatives and analogues provide tissue protection in animal models (Leist et al. 2004). However, the physiological function of Epo outside the bone marrow of healthy creatures has been questioned based on molecular biology studies. First, transgene-rescued Epo-R-null mutant mice expressing Epo-R exclusively in the haematopoietic lineage develop normally and are fertile (Suzuki et al. 2002). Second, Epo-R protein is not generally detectable in cells of non-haematopoietic origin, if appropriate anti-Epo-R antibody is used for study (Sinclair et al. 2010).

Molecular mechanism of the hypoxia-induced Epo expression

Most of the present knowledge of the O₂-sensing mechanism in control of Epo production has been based on in vitro studies utilising human hepatoma cells (lines Hep3B and HepG2). Noteworthily, the mechanisms of the renal and the hepatic Epo expression differ. (i) Renal cells respond in an all-or-nothing fashion to hypoxia (Koury et al. 1989), whereas hepatoma cells respond in a graded way. (ii) The hypoxia-response elements (HREs) in control of the Epo gene are located upstream in the kidney (between 9.5 and 14 kb 5′ to Epo) but downstream in the liver (within 0.7 kb 3′ to Epo) according to studies in transgenic mice (Kochling et al. 1998). In both tissues Epo expression is under the control of distinct transcription factors. The Epo promoter is suppressed by GATA-2 in normoxia (Tsuchiya et al. 1997). GATA-2 levels decrease in hypoxia (Imagawa et al. 2003). More importantly, the Epo enhancer is activated by hypoxia-inducible transcription factors (HIFs). These are composed of an O₂-labile α-subunit (120 kDa; isoforms 1α, 2α or 3α) and a constitutive β-subunit (90–95 kDa). Although the prototype HIF-1 was discovered in studies of Epo (Wang & Semenza, 1995), later investigations have identified HIF-2 (also called EPAS1 for endothelial PAS domain protein 1) as the primary transcription factor inducing Epo expression (Warnecke et al. 2004; see Haase, 2010). HIF-2α is activated by the stress-responsive deacetylase Sirtuin 1 (Dioum et al. 2009). On using Cre-loxP recombination to ablate renal HIF-2α,Kapitsinou et al. (2010) have shown that hepatic HIF-2 takes over as the main regulator of the serum Epo level. Interestingly, hepatocyte-derived HIF-2 is also involved in the regulation of iron metabolism genes, supporting a role for HIF-2 in the coordination of Epo synthesis with iron homeostasis (Kapitsinou et al. 2010).

The HIFs cooperate with hepatocyte nuclear factor 4 (HNF-4) and the transcriptional co-activators p300 and cAMP response element-binding protein (CREB)-binding protein (CBP) (Bunn et al. 1998). Using short interfering RNAs (siRNAs) and chromatin immunoprecipitation (ChIP) analysis, Wang et al. (2010) studied in more detail the roles of p300, CBP and p160 steroid receptor coactivator (SRC) in Epo expression in hypoxic Hep3B cells. Knocking down p300 resulted in a great reduction of Epo expression, negated recruitment of RNA polymerase II to the gene's promoter, and eliminated hypoxia-stimulated acetylation at the promoter and recruitments of SRC-1 and SRC-3 to the enhancer. Knocking down CBP only slightly decreased the hypoxia-induced Epo transcription (Wang et al. 2010). The directly repeated (DR2) DNA element neighbouring the HIF-binding site (Blanchard et al. 1992) is probably responsible for the effects of thyroid hormone (Fandrey et al. 1994) and retinoic acid (Kambe et al. 2000), which stimulate Epo expression in a tissue-specific but O₂-independent manner.

The C-terminus of the HIF-α subunits comprises O₂-dependent degradation domains (O-DDD) that are prolyl hydroxylated (at Pro ^{and Pro in HIF-1α, and Pro and Pro in HIF-2α) in the presence of O}₂ (Epstein et al. 2001; Bruick & McKnight, 2001; Jaakkola et al. 2001; Ivan et al. 2001; Yu et al. 2001). The reaction is catalysed by specific Fe^{-containing prolyl-4-hydroxylases (PHD-1, -2 and -3), which transfer one O-atom to the proline and the other to 2-oxoglutarate yielding CO}₂ and succinate (see Bruegge et al. 2007). The prolyl hydroxylated HIF-α combines with von Hippel-Lindau tumour suppressor protein (VHL)/E3 ligase und promptly undergoes proteasomal degradation (Pugh et al. 1997; Huang et al. 1998; Maxwell et al. 1999). As PHD-2 and PHD-3 are themselves HIF-target genes, their expression increases and HIF-α levels decline during long-term hypoxic periods (del Peso et al. 2003; Marxsen et al. 2004; Aprelikova et al. 2004; Stiehl et al. 2006). This feedback regulation may explain the declining Epo production during chronic anaemia or prolonged stay at high altitude (see below). Furthermore, the transcriptional activity of the HIFs is suppressed by HIF-α asparaginyl hydroxylation (at Asn ^{in HIF-1α and Asn in HIF-2α), which prevents the binding of the transcriptional co-activator CBP/p300. This reaction is catalysed by the ‘factor inhibiting HIF-1’, FIH-1 (}Mahon et al. 2001; McNeill et al. 2002), another Fe^{-containing and 2-oxoglutarate-requiring dioxygenase. According to} in vitro assays, the K_m values of the three PHDs for O₂ are above the arterial (∼170 mmHg), whereas FIH-1 operates at a lower (∼60 mmHg) (Koivunen et al. 2004). Thus, it is likely that the HIF-α PHDs are the primary O₂ sensors in control of Epo production.

The discovery that Fe ^{is required for HIF-α degradation may provide an explanation for the increased plasma Epo level in patients treated with the iron chelator deferoxamine (}Kling et al. 1996). Furthermore, cobalt is thought to enhance Epo expression by replacing the essential Fe ^{in the HIF-α dioxygenases, which results in HIF-α stabilisation. 2-Oxoglutarate competitors (clinical jargon: ‘HIF stabilisers’) likewise inhibit the hydroxylation of HIF-α. HIF stabilisers have been shown to stimulate Epo production} in vitro and in mice (Hsieh et al. 2007). Chronic oral dosing of compound FG-2216 in rhesus macaques proved to increase erythropoiesis and to prevent anaemia induced by weekly phlebotomy (Hsieh et al. 2007). Epo mRNA and HIF-2α have been co-localised in renal fibroblasts of rats treated with the 2-oxoglutarate competitor FG-4497 (Paliege et al. 2010).

Systemic Epo response

The primary functions of Epo are (i) to keep RBC mass and [Hb] constant day by day, and (ii) to hasten RBC recovery after haemorrhage. Little Epo is needed to maintain the steady state in healthy persons. The basal plasma concentration of Epo ranges from 6 to 32 IU l ^{(about 10 mol l). The levels vary greatly between individuals, with the result that significant sex- or age-specific differences cannot be detected. Of note is the diurnal fluctuation, with a nadir in the morning. An acute loss of about 0.5 l blood does not induce a major rise in circulating Epo in men (}Miller et al. 1982), yet the plasma Epo concentration increases exponentially when [Hb] falls below ∼125 g l ^{in humans not suffering from renal disease or inflammation. The response is dynamic, with initially very high Epo values that drop towards the normal ones before [Hb] normalises. The mechanism of the rapid decrease is not fully understood, but it may in part be caused by lowered HIF-α levels during long-term hypoxia (}Stiehl et al. 2006). Furthermore, one has to remember that the plasma Epo level is not only dependent on the rate of Epo production, but also on its removal. In vitro studies have demonstrated that Epo is internalised and degraded by its target cells (Gross & Lodish, 2006). Accordingly, anaemic patients with bone marrow hypoplasia exibit extremely high plasma Epo levels (10,000 IU l ^{or more) compared with subjects suffering from haemolytic anaemia. Independently of changes in tissue oxygenation, the level of circulating Epo increases when the erythrocytic progenitors are arrested by the administration of chemotherapeutics (}Birgegard et al. 1989).

Because Epo production depends on the tissue , Epo expression is also activated when the arterial declines or when the O₂ affinity of the blood increases. On ascent to altitude, Epo levels reach peak values after 1–2 days and then fall to a new plateau at about twice that present at sea-level (Abbrecht & Littell, 1972). As noted above, HIF-α levels decline during long-term hypoxic periods. In addition, the decrease in Epo production at continued hypoxia may be associated with the decrease in O₂-affinity of the blood resulting from an increase in the intraerythrocytic concentration of 2,3-bisphosphoglycerate (Klausen, 1998). Nutritional factors, such as low protein intake, were excluded as a reason for the rapid fall in circulating Epo in a controlled high-altitude study in the Chilean Andes, which was performed on shift workers and Caucasian lowlanders (Gunga et al. 1996). Since Epo expression studies cannot be performed in humans and there is incomplete knowledge of the mechanisms in control of the degradation of the hormone, the question remains as to how much the increased number of Epo-consuming erythrocytic progenitors contribute to the decline of circulating Epo.

The location of the Epo-expressing cells in the kidney cortex is well suited for the regulated production of the hormone: there is a constant ratio of blood flow rate and O₂ consumption, and the O₂ extraction is small. The renal cortical is barely affected by changes in cardiac output and blood flow because the renal O₂ consumption decreases with the glomerular filtration rate. A renal flow reduction by 50% of normal was necessary to elicit at least some Epo formation in rats (Pagel et al. 1989). By comparison, the production of Epo increases much more during systemic hypoxia due to anaemia or hypoxaemia.

Role of angiotensin II

The signal to produce more erythrocytes following haemorrhage is apparently linked to the signals to retain salt and water by means of the renin–angiotensin system (Dunn et al. 2007). Angiotensin II (Ang II) is thought to stimulate erythropoiesis by two means. First, Ang II increases Epo production. Second, Ang II is a growth factor for the myeloid erythrocytic progenitors (Dunn et al. 2007; Vlahakos et al. 2010). The infusion of Ang II has been shown to raise the plasma Epo level to 17 IU l ⁽vs. 11 IU l ^{in controls) in healthy men (}Gossmann et al. 2001). Furthermore, Ang II treatment at blood pressure-increasing doses (1.3 μg min ^{for 6 h) was found to raise the Epo concentration in men subjected to a haemorrhage of 750 ml as a basic physiological stimulus (}Freudenthaler et al. 1999). The effect of Ang II on Epo production is prevented by Ang II type 1 receptor blockers (Freudenthaler et al. 1999; Gossmann et al. 2001). Transgenic mice harbouring the human renin and angiotensinogen genes exhibit erythrocytosis as well as hypertension (Kato et al. 2005). Bone marrow transplantation experiments have shown that Ang II-1a receptors on bone marrow-derived cells are dispensable for the Ang II-dependent erythrocytosis. Plasma Epo levels and renal Epo mRNA expression in the double transgenic mice were increased compared with those of the wild-type control, while the elevated plasma Epo levels were attenuated in the compound mice. Thus, the renin–Ang II system enhances erythropoiesis mainly through the Ang II-1a receptor of the Epo-producing renal cells. The question still needs to be answered whether Ang II stimulates Epo synthesis directly or indirectly by decreasing the renal O₂ supply/O₂ demand ratio (Dunn et al. 2007).

A feedback regulation of red cell mass and blood volume by means of Epo and Ang II seems to exist. Treatment of healthy men with recombinant human Epo (rhEpo) produces an increase in red cell mass. However, the increase in haematocrit (Hct) is accompanied by a decrease in plasma volume, which is probably due to a downregulation of the renin–angiotensin–aldosterone system and results in a constancy of the blood volume. Thus, Epo treatment in healthy humans raises [Hb] by two mechanisms: (i) an increase in red cell mass, and (ii) a decrease in plasma volume (Lundby et al. 2007).

Extra-renal sites affecting renal Epo production

The question arises as to how far renal Epo synthesis is influenced by extra-renal sites under hypoxic conditions (Fig. 1). One hypothesis suggests that the brain modulates O₂-dependent Epo expression in the kidney. Local hypoxia of the brain stem was associated with an increase in renal Epo production in experimental animals (von Wussow et al. 2005). Both astrocytes and neurons express Epo. Moreover, evidence suggests that glial cells contribute to circulating Epo following the induction of hypoxia (Weidemann et al. 2009). Renal (but not hepatic) Epo mRNA levels are suppressed in transgenic mice lacking VHL or both VHL and HIF-1α in astrocytes (Weidemann et al. 2009). The mechanism by which astrocytes influence renal Epo expression still needs to be explored. In addition, the O₂ supply to the skin has been implicated in the control of renal Epo expression (Boutin et al. 2008). According to this concept, an increased blood flow to the skin causes a reduction in the renal O₂ supply. Mice with an epidermal deletion of VHL have increased Epo synthesis and develop erythrocytosis (Boutin et al. 2008). However, the concept of the dermal control of renal Epo production is not generally accepted (Paus et al. 2009), amongst other things because blood flow is not a major parameter in renal Epo synthesis (Pagel et al. 1989). Also, dermal blood flow depends on body heat, which has not been shown to affect Epo production. Note, here, that renal nerve inputs appear to be less relevant for O₂-dependent Epo expression in the kidney (Eckardt et al. 1992). Frankly speaking, the evidence assigning extra-renal sites a major role in the control of renal Epo production is far from convincing.

Information on the various other humoral factors that have been implicated in the control of renal Epo production, such as adenosine and prostanoids, has been published elsewhere (Fisher, 2003).

Pathophysiology

Normochromic normocytic anaemia due to insufficient Epo synthesis develops in patients with CKD (‘renal anaemia’), systemic inflammations or malignancies. The relative lack of Epo in patients with anaemia associated with chronic disease has been related to the negative effects of the cytokines interleukin-1 (Il-1) and tumour necrosis factor-α on Epo expression (Jelkmann, 1998). In vitro, Il-1 inhibits HNF-4α mRNA formation and causes proteasome-dependent degradation of HNF-4α protein, thereby suppressing the hypoxic inducibility of the Epo enhancer (Krajewski et al. 2007). The anaemias of patients with CKD or cancer in combination with chemotherapy can be corrected by replacement therapy with rhEpo or its analogues (see Macdougall & Ashenden, 2009).

Erythrocytosis is due to persistent over-stimulation of erythropoiesis (see Hodges et al. 2007). Hct, RBC counts and [Hb] are abnormally high. Primary erythrocytosis is generally a myeloproliferative disorder. Secondary erythrocytosis is due to Epo over-production, most commonly caused by hypoxaemia. Theoretically, an increase in [Hb]– and thus in the O₂ capacity of the blood – should be beneficial for tissue oxygenation (Brauner & Wang, 1997). However, blood viscosity increases with Hct, which raises the cardiac afterload and impedes the flow in microvessels. Hence, the erythrocytosis in high-altitude residents can be considered a maladaptive reaction, bearing risks for thromboembolic events and mortality. There have been distinct genetic adaptations in the evolution of tolerance of humans to high altitude (Hochachka et al. 1998). Tibetans living at about 4000 m altitude have relatively low [Hb] (Winslow et al. 1989), while South-American high-altitude natives often suffer from erythrocytosis and chronic mountain sickness (Leon-Velarde et al. 1991). Cobalt exposure may contribute to the excessive erythrocytosis in high-altitude miners (Jefferson et al. 2002). Beall et al. (2010) have recently identified 31 EPAS1 (encoding HIF-2α) single-nucleotide polymorphisms that correlate with low [Hb] in Tibetans residing at high altitude. This pioneering work may be taken as a good example of how information from molecular research can improve understanding at the systemic physiological level.