Erythropoietin (EPO) is a glycopeptide hormone produced in the kidney in response to hypoxia. It is known for its role in red blood cell formation but has also, in recent years, been found in other locations like the central nervous system where its receptors are specifically expressed, suggesting an endogenous role in this location. EPO and its mimetics hold promise for clinical use in neuroprotection. As in studies of previous neuroprotective agents, in vitro and in vivo animal models have found that EPO has important neuroprotective actions in stroke, ischaemia and other neurological injury, mediated by inhibition of apoptosis and induction of angiogenesis. Although a randomised trial of 40 stroke patients found a non-significant neurological improvement with EPO, there are no large trials of EPO neuroprotection. Therefore the benefits of EPO recorded in animal models have not yet been replicated in human populations.
Larger trials of EPO in stroke and other neurological disease patients will need to be embarked upon. Practical issues like drug dosing, randomisation and point in therapeutic window at which intervention is received will need to be reviewed.
The above notwithstanding, EPO and its mimetics look likely to become clinically useful in neuroprotection in the near future.
Keywords – Erythropoietin, receptor, mimetics, neuroprotection, stroke.
Knowledge of erythropoietin (EPO) as a haematopoietic cytokine produced mainly in the kidney has, over the years, translated into widespread use of its genetically engineered form, producing clinical benefit in treatment of anaemia of diverse causes (Littlewood, 2004). Its role as the primary regulator of erythrocyte production is mediated through erythropoietin receptors (EpoR) found on target erythroid progenitor cells in the bone marrow. In recent years however, EPO and EPO receptors have been found to be expressed in the brain (Digicaylioglu, 1995) and developing human central nervous system (CNS) tissue (Juul, 1998), suggesting a role for the hormone in neural function.
Several animal studies have investigated the role of EPO in protection from diverse types of neural injury (such as ischaemia, inflammation and trauma) and suggest a neurotrophic and neuroprotective effect (Siren, 2001) possibly mediated through a number of mechanisms involving the EPO receptor. If such effect of EPO is demonstrated in the human CNS, it could potentially represent a major advancement in the way human neurological and neuropsychiatric conditions are treated.
A well-recognised clinical challenge with EPO therapy as it stands is the cost and inconvenience of long term parenteral administration of the hormone (Joliffe, 1995). This has led to efforts to develop molecules that may be structurally unrelated to EPO but have the ability to bind to the EPO receptor and potentially mimic the biological actions of the naturally-occurring hormone. These EPO mimetics can become clinically useful if they demonstrate similar biological potency to EPO.
This paper critically evaluates the literature on the potential neuroprotective actions of EPO, and by extension, its mimetics. It explores the role of EPO within the central nervous system and the possibility that the neuroprotective action of EPO and EPO mimetics can be exploited clinically.
What is EPO?
EPO is a 30.4kD glycoprotein hormone produced mainly in the kidney but also in the liver and spleen in response to low oxygen tension in the body. It stimulates the production of red blood cells in the bone marrow and has a serum concentration of about a hundredth of the average concentration of most other body hormones (Lappin, 2002).
EPO comprises 165 amino acids and four carbohydrate groups. An important feature of the structure of EPO is its possession of two disulphide bonds linking the cysteine amino acids at positions 6/161 and 29/33. The former is functionally more important as it acts as a tether, ensuring that the whole molecule is held in the correct shape for binding to the EPO receptor (Lappin, 2003). Any factor that compromises the shape of the molecule can result in inability of EPO to bind to its receptor.
Each molecule of EPO binds to the extracellular domain of two identical EpoR present on the surfaces of target cells (see figure 1) and initiates a series of intracellular processes that results in erythrocyte production. Each receptor has a cytokine binding domain comprising an N-terminal and a C-terminal domain. The peculiar structural characteristics of these domains place the EPO receptor in a distinctive class of receptors called the haematopoietin receptor superfamily, to which human growth hormone and prolactin receptors also belong (Krantz, 1991).
Apart from its role in erythrocyte formation, EPO also has antiapoptotic, anti-inflammatory and neurotrophic properties and is highly expressed in the brain during development and following neuropathological insults (Eid, 2002). Indeed, the brain is known to produce its own EPO, which is slightly smaller in size than the EPO produced by the kidney (Lappin, 2003). Although it is primarily associated with erythropoiesis, it has become widely accepted that the effects of EPO transcend erythropoiesis to include trophic effects on non-erythroid cells.
Further sections of this paper will review the non-erythropoietic functions of EPO and the locations of EpoR, focusing on the presence and possible roles of the hormone in the central nervous system.
Location and haematopoietic functions of EPO
There are two major approaches by which the presence of EpoR in any tissue can be determined. The first approach is immunohistochemistry (Juul, 1998). This process relies on ligand-receptor binding and uses either monoclonal or polyclonal antibodies to determine the location of specific proteins in tissues. The limitation of this method is the possibility that a non-specific interaction could occur between an anti-EPO receptor antibody and a cell surface protein that resembles the EPO receptor, such as human growth hormone and prolactin receptors. The second approach is in situ hybridization (ISH), a method that uses labelled DNA or RNA probes to detect and localise specific DNA or RNA sequences. It identifies EPO receptor mRNA cells and when used in tandem with immunohistochemistry, overcomes the limitation of non-specific binding as it confirms the presence of both EPO receptor mRNA and protein in the same cell (Lappin, 2003).
With respect to its haematopoietic functions, EPO is produced in the kidney in response to hypoxia through stimulation by Hypoxia Induced Factor (Graber, 1989). It circulates in plasma to the bone marrow where each EPO molecule binds to the extracellular domains of two EpoR triggering a conformational change. In this conformational change, Janus Kinase 2 (JAK2) molecules associated with the EpoR are brought into close proximity to one another and activated by transphosphorylation (Cheung, 2001). Activated JAK2 further phosphorylates other tyrosines in the cytoplasmic domain of the receptor, initiating an intracellular signaling cascade (involving signaling proteins like STAT5) that regulates expression of numerous target genes in the nucleus (including Bcl-x, the apoptosis inhibitor). This in turn controls cell survival, proliferation, and differentiation (Silva et al, 1995).
The activation of JAK2/STAT5/Bcl-x pathway by EPO is key to the differentiation of erythroid progenitor cells to form red blood cells but whether JAK2 is the sole direct signaling molecule downstream of EPO receptor required for biological activity may not be firmly established yet. Using mutational analysis of over 40 mutant EpoR, Pelletier and colleagues (2006) concluded that JAK2 is the only direct signaling molecule downstream of the receptor required for biological activity. Tsuji-Takayama and colleagues (2006) however studied EPO signaling in primitive erythroid (EryPs) and definitive erythroid (EryDs) cells using an embryonic stem-derived culture system and asserted the existence of a JAK2-independent pathway of EPO signaling that induces STAT5 activation.
The importance of the JAK2/STAT5/Bcl-x pathway in erythropoiesis is underlined by research findings suggesting that JAK2 deficiency can result in embryonic death due to a lack of erythropoiesis (Parganas et al, 1998). In addition, mice that were deficient in STAT5 were found to have anaemia of varying severity correlating with severe apoptosis in erythroblasts and reduced expression of the Bcl-x gene (Sokolovsky, 2001).
The haemopoietic functions of EPO are well established. The following next section of this paper will begin to explore its presence and relevance in the central nervous system.
The physiological role of EPO in the central nervous system
There are potentially three distinctive mechanisms of EPO production and function as it relates to the central nervous system. The first is an endocrine system, in which EPO produced in the kidney is secreted into plasma and circulates to various tissues, including CNS tissue, binding to EpoR in those tissues. The second system is a paracrine system, in which EPO is produced by some brain cells and binds EpoR on nearby brain cells. The third system is an autocrine system in which some brain cells produce their own EPO. It is thought that the paracrine and autocrine EPO systems of the CNS are independent of the endocrine EPO secreted by the kidney (Sasaki, 2003) and may be pointers to an endogenous role of EPO in the nervous system.
The findings of EPO and EpoR within the central nervous system have fuelled efforts to define their precise role within it. An important role that has been explored widely is neuroprotection. One of the earlier works on the potentially neuroprotective role of EPO was carried out by Morishita and colleagues (1997). They used immunohistochemical methods and reverse-transcription polymerase chain reaction assays to elicit the expression of EpoR in cultured hippocampal and cerebral cortical neurons of day-19 rat embryo. They also found that EPO prevented in-vitro glutamate-induced neuronal death in a dose-dependent fashion. For neuroprotection to be elicited, neurons had to be exposed to EPO for a while before exposure to glutamate-induced neuronal death.
Other neuroprotective roles of EPO have been elucidated in animal models. Recombinant human EPO (rhEPO) administered systemically prevented apoptotic neuronal loss and reduced mean infarct volume compared to controls in newborn rat hypoxic injury models (Kumral et al, 2003). Survival for up to 72 hours was recorded after subarachnoid haemorrhage (SAH) in rats that received intraperitoneal rhEPO compared to a mortality rate of 42.9% within 72 hours for control rats (Buemi et al, 2000). In general, neural injury is attenuated by EPO in animal models. In addition to neuroprotection, EPO- induced neuronal tissue repair has been demonstrated in rodent models. Bocker-Meffert et al (2002) compared the outgrowth of rat retinal explants under the influence of EPO and vascular endothelial growth factor and found both molecules to have significant neuroregenerative effects in ischaemic retinal conditions.
Some evidence for potential neuroprotection by EPO does exist in human-subject studies. Autopsy brains of neuropathologically normal subjects were compared to those with ischemic infarcts or hypoxic damage. It emerged that while EPO-EpoR immunoreactivity was mainly neuronal in normal brain, reactivity appeared in endothelium, microvessels and neuronal fibers in fresh infarcts. According to the authors, the pronounced up-regulation of EPO-EpoR association in human ischemic/hypoxic brains is a pointer to their role as an endogenous neuroprotective system (Siren et al, 2001).
Having established the presence of an endogenous EPO system in the CNS, it is important to examine the relative roles of the endogenous and systemic EPO systems in neuroprotection. Large glycosylated molecules are generally not able to cross the intact blood brain barrier (BBB) due to the negligible permeability of the brain capillary endothelial wall. The abundant expression of EpoR on brain capillaries could however provide a route for systemic EPO to enter the brain and mediate neuroprotection. Indeed, Brines et al (2000) showed that systemically administered rhEPO administered intraperitoneally to rats reduced global and focal ischaemia. The extent to which EPO crosses the intact human BBB is negligible however (Dame et al, 2001) and studies of the dynamics of EPO in cerebrospinal fluid (CSF) suggest that the systemically circulating hormone is less critical to human brain neuroprotection. Springborg and colleagues (2003) collected 83 corresponding serum and CSF samples from 18 patients with aneurysmal SAH and compared the concentrations of EPO with those of blood-derived markers of BBB function and proteins with well-known CNS synthesis. EPO showed dynamics similar to CNS-derived proteins indicating that EPO in the CSF of patients with aneurysmal SAH originates mainly from the CNS. This finding may have important implications for the potential systemic use of EPO in human neuroprotection.
The neuroprotective and neuroregenerative roles of EPO are essentially reactive processes that aim to mitigate the outcomes of neural injury. In more recent years, evidence has also begun to emerge of the role of EPO in CNS development. Yu et al (2002) provide evidence that EPO acts to stimulate neural progenitor cells and prevent apoptosis in the rat embryonic brain. In addition, EpoR are expressed throughout the human foetal CNS (Dame et al, 2000) and upregulated by conditions of hypoxia.
In summary, EPO and its receptors are widely expressed in animal and human CNS where it may have neuroprotective and neurotrophic functions. In clinical terms, a potential protective function of this hormone in CNS conditions like stroke would be a major medical scientific breakthrough. The next section explores this possibility.
Stroke incidence and existing treatments
Stroke affects about 114 people in every 100,000 population in the UK and about 75% of affected people are over 56years old. It is responsible for 10% of deaths in England and Wales (Macwalter, 1999).
The approach to treatment of stroke can be conceptualised as primary, secondary and acute treatments (Silver et al, 2005). Primary approaches include aspirin, statins and exercise. However aspirin fails to prevent stroke in men over 50yrs; platelet antiaggregants pose a risk of gastrointestinal bleeding and statins showed no benefit in stroke primary prevention (Shepherd, 1995). Secondary measures (in patients who have had a stroke) include antiaggregants, statins, antihypertensives and thrombin inhibitors. Antiaggregants show on average about 9% relative risk reduction in secondary stroke prevention; statins and antihypertensives about 32% reduction and thrombin inhibitors about 70% (Shepherd, 1995). Acute treatments are used in the immediate periods following a stroke and aim to prevent or limit neuronal damage. The agents in this group include tissue-type plasminogen activator (TPA), antiaggregants, and anticoagulants. The failure in many cases to demonstrate benefit in stroke outcomes with these drugs as well as the practical difficulties of using agents like agents TPA has led to exploration of the potential for improvement of clinical outcomes in stroke using neuroprotectants like EPO.
Wang and colleagues (2004) investigated the role of EPO in brain repair in mice following stroke. They induced embolic stroke, injected rhEPO 24 hours later, and studied angiogenesis, neurogenesis and neurological function in treated rats compared to controls. The diagrams below display the improvement in neurological function and angiogenesis in treated versus control rats.
In a double blind randomised safety and proof-of-concept trial, Ehrenreich and colleagues (2002) randomised 40 patients with MRI-confirmed ischaemic stroke and who were within 8hrs of symptom onset. RhEPO was compared to saline and the outcomes assessed were functional outcome at day 30 and evolution of infarct size. It emerged that CSF concentrations of EPO were 60-100 times greater in intervention than control patients suggesting that systemically administered EPO reached the CNS. EPO treatment was associated with a non-significant improvement in neurological outcomes at one month as well as greater reduction in infarct size. Longer-term data in a larger group of study patients would however have been more convincing.
CNS role and mechanisms of neuroprotection
Neuroprotection refers to the mechanisms used to protect against neuronal injury or degeneration in the CNS following acute disorders like stroke or chronic neurodegenerative diseases like Alzheimer’s disease (medicalnewstoday, 2006). Apoptosis is one important way by which the nervous system responds to severe injury like hypoxia and trauma but neuroprotection limits neural dysfunction or death after CNS injury and maintains viability of cellular processes and neural function In occlusive stroke, the concept of neuroprotection involves inhibition of a cascade of pathological molecular events occurring under ischaemia and leading to calcium influx, activation of free radical reactions and cell death (Walgren et al, 2004).
The physiological role of EPO in the CNS has been established in earlier sections as mainly that of neuroprotection and neurotrophism accomplished by EPO through anti-apoptotic mechanisms (Bernaudin et al, 2000).
As mentioned earlier, the upregulation of the JAK2/STAT5/bcl-x signaling pathway is key to anti-apoptosis and neuroprotection. Other mechanisms of neuroprotection however include increasing nitric oxide production (Genc et al, 2001), the modulation of Akt1 and induction of caspase 1, 3 and 8 (Chong et al, 2003).
The concept of neuroprotection has however been more a successful laboratory concept than a clinical breakthrough and the reasons for this are explored in the next section.
Use of EPO in neuroprotection
So can EPO and its mimetics prove to be clinically useful novel neuroprotective agents? It may be that the key question to answer is really this – ‘are there any characteristics of EPO and its mimetics that sufficiently distinguish them from previously failed neuroprotectives to offer any real hope that they will succeed where others have failed?’
Some factors count in favour of the possibility. First, EPO is clinically safe and less toxic than many earlier known neuroprotective agents (Hasselblatt et al, 2006). Secondly, it is thought to be selectively uptaken by capillaries in the BBB and therefore accessible to the CNS on systemic administration. In addition, the possibility that newer EPO mimetics will be smaller in size than EPO may improve the CNS bioavailability of EPO.
On the other hand, the factors that weigh against the potential clinical usefulness of EPO and its mimetics as novel neuroprotective agents include the fact that although animal studies have shown impressive neuroprotective results for EPO, those results have not been replicated by any studies in human subjects with stroke (Wahlgren et al, 2004). The reasons for this failure will essentially determine any real possibility of EPO overcoming the failure of its neuroprotective predecessors and proving a clinical breakthrough in neuroprotection.
Some of these reasons include the fact the failed trials of the neuroprotective agents had entry windows that went far beyond the windows of success seen in tests of neuroprotective agents in animals (Jonas et al, 2001). Other reasons proffered include the imperfect standardisation of animal study procedures; the use of animals with ‘clinical characteristics’ that do not nearly reflect those of actual stroke patients; small numbers and poor randomisation procedures in these human trials. Curry (2003) used allometric scaling of glutamate data to support a hypothesis that trials should be restricted to patients with progressing stroke and associated elevated glutamate levels, noting that patients without elevated glutamate might show adverse effects of drug treatment if normal levels of glutamate associated with neuroprotection are antagonized.
EPO and its mimetics hold an impressive promise for potential clinical use in neuroprotection. As in studies of previous neuroprotective agents, in vitro and in vivo animal models have found that EPO has important neuroprotective actions in stroke, ischaemia and other neurological injury, mediated by inhibition of apoptosis and induction of angiogenesis. Although a small randomised trial of stroke patients found a non-significant neurological improvement with high dose EPO, there are no large trials of EPO neuroprotection. Therefore the benefits of EPO recorded in animal models have not yet been replicated in human populations. The likelihood of this happening in the near future will be determined by a number of factors, favourable and unfavourable. Larger trials of EPO in stroke and other neurological disease patients will need to be embarked upon. Practical issues like drug dosing, randomisation and point in therapeutic window at which intervention is received will need to be reviewed.
These do not however take away from the fact that EPO is a promising weapon in the armament against neurological injury and will someday, along with its mimetics, prove useful novel neuroprotective agents.