Erythropoietin and its Carbamylated Derivative Prevent the Development of Experimental Diabetic Autonomic Neuropathy in STZ-Induced Diabetic NOD-SCID Mice

Abstract

Autonomic neuropathy is a significant diabetic complication resulting in increased morbidity and mortality. Studies of autopsied diabetic patients and several rodent models demonstrate that the neuropathologic hallmark of diabetic sympathetic autonomic neuropathy in prevertebral ganglia is the occurrence of synaptic pathology resulting in distinctive dystrophic neurites (“neuritic dystrophy”). Our prior studies show that neuritic dystrophy is reversed by exogenous IGF-I administration without altering the metabolic severity of diabetes, i.e. functioning as a neurotrophic substance. The description of erythropoietin (EPO) synergy with IGF-I function and the recent discovery of EPO’s multifaceted neuroprotective role suggested it might substitute for IGF-I in treatment of diabetic autonomic neuropathy. Our current studies demonstrate EPO receptor (EPO-R) mRNA in a cDNA set prepared from NGF-maintained rat sympathetic neuron cultures which decreased with NGF deprivation, a result which demonstrates clearly that sympathetic neurons express EPO-R, a result confirmed by immunohistochemistry. Treatment of STZ-diabetic NOD-SCID mice have demonstrated a dramatic preventative effect of EPO and carbamylated EPO (CEPO, which is neuroprotective but not hematopoietic) on the development of neuritic dystrophy. Neither EPO nor CEPO had a demonstrable effect on the metabolic severity of diabetes. Our results coupled with reported salutary effects of EPO on postural hypotension in a few clinical studies of EPO-treated anemic diabetic and non-diabetic patients may reflect a primary neurotrophic effect of EPO on the sympathetic autonomic nervous system, rather than a primary hematopoietic effect. These findings may represent a major clinical advance since EPO has been widely and safely used in anemic patients due to a variety of clinical conditions.

INTRODUCTION

Autonomic neuropathy is a significant clinical complication of diabetes which disturbs cardiovascular, alimentary and genitourinary function and results in increased patient morbidity and mortality (Ewing et al., 1980; Hosking et al., 1978; Rundles, 1945; Sampson et al., 1990; Vinik et al., 2003). Several series of autopsied diabetic patients (Duchen et al., 1980; Schmidt et al., 1993; Schmidt and Plurad, 1986) have established the reproducible development of markedly enlarged dystrophic axons and nerve terminals in diabetic prevertebral superior mesenteric (SMG) and celiac sympathetic ganglia (CG) in the absence of substantial loss of principal sympathetic neurons, a pattern similar to sympathetic ganglionic pathology which develops in aged patients (Schmidt et al., 1993).

The regular occurrence of degenerating, regenerating, and pathologically distinctive dystrophic axons and, to a lesser degree abnormal dendrites, in the absence of neuron loss has also been demonstrated in prevertebral sympathetic ganglia of streptozotocin (STZ)- and genetically-diabetic rodents, closely corresponding to human disease [reviewed in (Schmidt, 2002)]. Our previous studies have shown the striking improvement in the severity of diabetic autonomic neuropathy in rats treated with exogenous rhIGF-I in the absence of an effect on the severity of hyperglycemia (Schmidt et al., 1999), a result thought to reflect a neurotrophic role for IGF-I. The demonstration of endogenous IGF-I deficiency in the serum and sympathetic ganglia of diabetic rats (Schmidt et al., unpublished data), the known function of IGF-I as a sympathetic neurotrophic substance in vitro (Recio-Pinto et al., 1986) and differences in the development of sympathetic ganglionic dystrophy in types I and II diabetic rat and mouse models, i.e., animals deficient in or with increased levels of circulating IGF-I, respectively (Schmidt et al., 2004) suggest that loss of a neurotrophic effect of IGF-I might underlie the development of diabetic autonomic neuropathy.

Although the administration of IGF-I in a variety of human diseases has been accomplished, there has been concern that IGF-I may promote the development or progression of malignancies (Clark, 2004). As a result, substances with IGF-I like effects lacking its side effects have been sought. Interestingly, it has been noted that within the nervous system astrocytes respond to IGF-I by synthesizing EPO (Masuda et al., 1997). Further, a synergy has been observed between EPO and IGF-I (Digicaylioglu et al., 2004). These observations raised the question of whether EPO might substitute for IGF-I in treatment of diabetic neuropathy.

It is known that EPO receptors are located on peripheral dorsal root ganglia neurons, axons and Schwann cells and activate the PI3K/Akt signaling pathway, using receptors and early pathway intermediates distinct from IGF-I. Although EPO does not directly activate IGF-I or insulin receptors, EPO receptor activation results in stimulation of the PI-3Kinase/Akt signaling pathway which it shares with IGF-I and insulin signaling pathways.

Initially discovered as a mediator of erythropoiesis, for some time EPO has been recognized to have salutary effects on a variety of animal models of neurodegenerative processes including ischemic brain damage (Zhang et al., 2006), experimental allergic encephalomyelitis (Savino et al., 2006) and amyotrophic lateral sclerosis (Koh et al., 2007). Similarly, EPO is protective of peripheral nervous system insults (Hoke, 2006) including acrylamide and cisplatin toxic neuropathies (Bianchi et al., 2007; Keswani et al., 2004a; Melli et al., 2006), HIV sensory neuropathy (Keswani et al., 2004b) and, significantly, experimental diabetic somatic neuropathy (Bianchi et al., 2004; Tam et al., 2006). Therefore, to identify a possible role of EPO in the treatment of diabetic autonomic neuropathy, in this study we have demonstrated the presence of sympathetic neuronal EPO receptors (EPO-R) and examined the effect of exogenous administration of rhEPO on the frequency of neuritic dystrophy in our experimental mouse model of diabetic sympathetic autonomic neuropathy. Since EPO treatment of patients without anemia may possibly produce side effects of erythrocytosis or effect tumor growth, we have also examined the effect of the carbamylated derivative of EPO (CEPO) which has been shown to possess tissue protective activities but no erythropoietic potency (Leist et al., 2004; Montero et al., 2007; Savino et al., 2006).

MATERIALS AND METHODS

Animals

Male Non-Obese Diabetic-Severe Combined Immune Deficient (NOD-SCID) mice were purchased from the Jackson Laboratory (Bar Harbor, ME) and were kept in pathogen-free conditions at Washington University. NOD-SCID mice are the result of breeding of the SCID mutation to the NOD background for many generations, such that the NOD-SCID mouse is genetically identical to the NOD mouse save for the absence of DNA-dependent protein kinase, a DNA repair enzyme (Blunt et al., 1996) resulting in loss of B and T cell function. Our previous studies (Schmidt et al., 2003) showed that NOD-SCID mice treated with streptozotocin rapidly become severely diabetic and develop dramatic autonomic neuropathy within 3–5 weeks. All animals were housed and cared for in accordance with the guidelines of the Washington University Committee for the Humane Care of Laboratory Animals and with National Institutes of Health guidelines on laboratory animal welfare. All mice were allowed standard mouse chow and water ad libitum and maintained on a 12/12 hour light/dark cycle. A few male 4 month old Sprague-Dawley (Charles River Laboratories, Wilmington, Massachusetts) were also used in this study. Rats were fed Purina rodent chow ad libitum and were housed in small groups with a 0700–1900 light cycle.

Induction of Diabetes and Treatment Protocol

Mice were made diabetic by i.p. injection of freshly made streptozotocin (STZ [Sigma, St. Louis, MO], 50 mg/kg in citrate-saline buffer, pH 4.5) on four consecutive days under ketamine/xylazine anesthesia. Control animals received a comparable volume of citrate-saline buffer. The day following the last injection of STZ mice were bled and significantly hyperglycemic animals (plasma glucose >250 mg%) were considered diabetic. The effect of treatment with EPO and CEPO was examined in two separate experiments. Treatment with recombinant human erythropoietin (rhEPO, Dragon Pharmaceuticals, Vancouver, Canada) or carbamylated EPO (rhCEPO, Warren Pharmaceuticals) at a dose of 10 micrograms/kg (i.e., equivalent to ~1000IU/kg of EPO) in saline × 3 injections/week for 4 weeks was begun the day after the last STZ injection. Control and diabetic animals not receiving EPO or CEPO received injections of saline.

Tissue Preparation

After 4 weeks of treatment animals were anesthetized with ketamine/xylazine and perfused with 50 ml of heparinized saline followed by 100–200 ml of 3% glutaraldehyde in 0.1M phosphate buffer, pH 7.3, containing 0.45 mM Ca+2. The superior mesenteric-celiac ganglia (SMG-CG) were dissected as a single block, cleaned of extraneous tissue while maintaining the superior mesenteric artery with the ganglionic block, and fixation continued overnight at 4°C in the same buffer. Tissue samples were postfixed in phosphate-buffered 2% OsO₄, dehydrated in graded concentrations of ethanol and embedded in EMbed-812 (Electron Microscopy Sciences, Hatfield, PA) with propylene oxide as an intermediary solvent. One micron thick plastic sections were examined by light microscopy after staining with toluidine blue. Ultrathin sections of individual SMG-CG were cut onto formvar coated slot grids, which permits visualization of entire ganglionic cross sections. Tissues were subsequently stained with uranyl acetate and lead citrate and examined with a JEOL 1200 electron microscope.

Quantitative Histologic Methods

Dystrophic elements are typically intimately related to neuronal perikarya and, therefore, we routinely express their frequency as the ratio of numbers of lesions to nucleated neuronal cell bodies. This method, used in our previous studies (Schmidt et al., 2003), substantively reduced the variance in assessments of intraganglionic lesion frequency. In addition, its simplicity permits the quantitative ultrastructural examination of relatively large numbers of ganglia. In our current animal studies an entire cross section of the SMG-CG was scanned at 12,000X magnification and the number of dystrophic neurites and synapses was determined by an investigator blinded to the identity of individual animals. Dystrophic neurites consist of swollen axons, synapses, dendritic spines or dendrites containing a variety of organelles including tubulovesicular aggregates, admixed normal and degenerating subcellular organelles, multivesicular and dense bodies, neurofilaments, and pure aggregates of minute mitochondria. The number of nucleated neurons (range: 50–200 neurons examined in each ganglionic cross section) was then determined by recounting at 6000X magnification. The frequency of ganglionic neuritic dystrophy was expressed as the ratio of number of dystrophic neurites to the number of nucleated neurons in the same cross section.

PCR Demonstration of EPO-R in Sympathetic Ganglionic Neuron Cultures in Vitro

Primary rat sympathetic ganglia cultures (25,000 neurons/dish) were maintained in the presence of NGF for 6 d and then deprived of NGF for 24 hours as described previously (Estus et al., 1994). Total RNA was extracted from NGF maintained and NGF-deprived cultured neurons using TRI reagent (Molecular Research Center) following the manufacturer’s instructions. Two hundred nanograms of total RNA were reverse transcribed and amplified using Titan One Step RT-PCR kit (Roche). The primer sequences for amplification of erythropoietin receptor were 5′-CTA TGG CTG TTG CAA CGC GA-3′ (forward) and 5′-CCG AGG GCA CAG GAG CTT AG-3′ (reverse). RT reactions were performed at 55°C for 45 min. PCR conditions were denaturing at 94°C for 3 min followed by 30 cycles of denaturing at 94°C for 30 sec, annealing at 56°C for 45 sec and extension at 68°C for 1 min, and ended by 7 min extension at 68°C. The PCR products were analyzed by 1.5% agarose gel electrophoresis.

Immunohistochemistry

Rats were fixed by perfusion with freshly made 4% buffered paraformaldehyde at 0–4°C and processed routinely for paraffin embedding. Paraffin embedded sections 5–8 microns thick were deparaffinized in xylene and rehydrated in a series of ethanol dilutions. Sections were preincubated for 20 min at room temperature in phosphate-buffered saline containing 2% BSA and 0.3% Triton X-100. Rabbit anti-EPO receptor antibody (1:100-1:200, Santa Cruz, sc-697) was next added and the slides incubated overnight at 4°C, washed and a secondary biotinylated goat anti-rabbit IgG (1:500) was applied, washed and, in some experiments, followed by tyramide signal amplification using successive incubation with streptavidin HRP and cyanine-3 tyramide (Perkin-Elmer Life Science Products, Boston, MA).

Statistical Analysis

All results are expressed as means + SEM. Analysis of variance (ANOVA) was performed using the SAS general linear models procedure.

RESULTS

EPO Receptors are Present on Rat Sympathetic Neurons

To determine if EPO receptors are expressed on sympathetic neurons, we asked if the EPO receptor mRNA was expressed in sympathetic cultures enriched in neurons. A previously extensively examined and validated set of cDNAs from NGF-maintained (lane 1, Fig 1) and NGF-deprived rat sympathetic neurons (lane 2, Fig 1) were examined by semi-quantitative RT-PCR using a 402 bp product corresponding to EPO receptor. Consistent with a neuronal localization, 24 hours of NGF-deprivation in culture, which kills most of the neurons but not non-neuronal cells, greatly reduced the amount of receptor in the cultures (lane 2, Fig 1).

Figure 1.

Effect of NGF-deprivation on EPO-R in Rat Sympathetic Ganglionic Neurons in Vitro A validated set of cDNAs prepared from NGF-maintained (lane 1) and 24 hour NGF-deprived rat sympathetic neurons (lane 2) were examined by semi-quantitative RT-PCR using a 402 bp product corresponding to the EPO receptor. Consistent with a neuronal localization, 24 hours of NGF-deprivation in culture, which kills most of the neurons but not non-neuronal cells, greatly reduced the amount of EPO receptor in the cultures.

Immunolocalization of EPO receptors was performed on the sympathetic ganglia of control and diabetic rats. We found strong perikaryal staining in control rat SCG (Fig 2A) and celiac ganglia (Fig 2B) and those of diabetic rats (data not shown), confirming our results using cultured sympathetic neurons.

Figure 2.

Immunolocalization of EPO-Receptor to Sympathetic Neurons

EPO-R is present in sympathetic neuronal cell bodies (arrows) in rat SCG (A) and celiac (B) sympathetic ganglia. (Original Magnification 400X)

Metabolic and Hematologic Parameters in STZ-treated NOD-SCID Mice

NOD-SCID mice became diabetic (blood glucose readings of >250 mg%) within a few days of induction of diabetes (data not shown) and were markedly hyperglycemic and weighed less at the time of sacrifice (Tables 1,2). Treatment with EPO (Table 1) or CEPO (Table 2) did not significantly affect the body weight or degree of hyperglycemia in diabetic mice. Age-matched saline, EPO or CEPO treated control NOD-SCID mice were consistently normoglycemic.

TABLE 1.

Effects of EPO on diabetic NOD-SCID mice

Rx	n	SMG-CG	Glucose	Weight	Hematocrit
		Neuritic dystrophy	(mg%)	(grams)	(%)
		(#lesions/#neurons)
DM + EPO	8	0.30 ± 0.07*	>600	25.1 ± 0.8	64 ± 1.8
DM + SALINE	7	0.97 ± 0.14	>600	23.9 ± 1.0	61 ± 0.4
C + EPO	4	0.06 ± 0.04# (3)	100 ± 5	27 ± 1.2#	72 ± 3.8#,^
C + SALINE	5	0.20 ± 0.03# (3)	125 ± 5	29.4 ± 1.2#	55 ± 1.1#

Legend: C= control, DM=diabetic Values = mean + SEM of n mice.

^*=p≤0.001,

^#=p≤0.01 vs. saline treated diabetic;

^{^}=p≤0.01 vs. saline treated control

TABLE 2.

Effects of CEPO on diabetic NOD-SCID mice

Rx	n	SMG-CG	Glucose	Weight	Hematocrit
		Neuritic dystrophy	(mg%)	(grams)	(%)
		(#lesions/#neurons)
DM + CEPO	11	0.34 ± 0.09*	563 ±14	22.1 ± 0.2	50.6 ± 1.8@
DM + SALINE	6	1.09 ± 0.08	526 ± 26	24.4 ± 0.7	48.0 ± 1.0 (5)
C + CEPO	3	0.27 ± 0.07*	115 ± 5	28.7 ± 0.7#	43.6 ± 0.3
C + SALINE	7	0.26 ± 0.05*	128 ± 5	29.9 ± 0.6*	44.0 ± 1.4

Legend: C= control, DM=diabetic Values = mean + SEM of n mice.

^*=p≤0.001,

^#=p≤0.01, vs. saline treated diabetic;

^@=p≤0.05 vs. saline treated control

The hematocrit of diabetic animals was mildly increased compared to controls in both experiments which may reflect a mild degree of dehydration on the day of sacrifice. EPO increased the hematocrit of treated controls compared to saline treated controls but had little effect on diabetic mice (Table 1). As expected, CEPO failed to have a hematologic effect on controls or diabetics (Table 2).

Neuropathology of SMG-CG

Examination of 1 micron-thick plastic sections of SMG-CG in 4 week saline treated diabetics in both experiments showed a well preserved complement of principal sympathetic neurons (Fig 3) surrounded by neuropil composed of an admixture of axons and dendritic elements. There was no evidence of active neuronal degeneration (specifically, no apoptosis), nodules of Nageotte (tombstones of prior neuron loss) or chromatolysis. None of the diabetic ganglia contained an inflammatory infiltrate or an association of individual lymphocytes or macrophages with neuronal perikarya.

Figure 3.

Light microscopic appearance of one-micron thick toluidine blue stained plastic sections of the SMG-CG of 4 week diabetic NOD-SCID mouse (A,B) in comparison to saline treated controls (C) and EPO-treated diabetics (Fig D). Dystrophic neurites, which range from swollen processes with pale cytoplasm (arrows, Fig 3A) to those staining strongly with toluidine blue (arrows, Fig 3B), distort the contours of otherwise normal appearing neuronal perikarya in NOD-SCID diabetic mice. (original magnification: A-D-300X)

Large swollen neurites were prominent in light microscopic examination of one-micron thick toluidine-blue stained plastic sections (arrows, Figs 3A,B) of the SMG-CG of NOD-SCID mice diabetic for 4 weeks in comparison to saline treated controls (Fig 3C) and EPO-treated diabetics (Fig 3D) or CEPO-treated diabetics (not shown), although dystrophic neurites were found in all groups of treated and untreated diabetics and controls, differing only in number (Tables 1,2). Dystrophic neurites were typically located immediately adjacent to neuronal cell bodies within their satellite cell sheaths (Figs 3A,B), which resulted in the displacement and distortion of perikaryal contours of targeted neurons, as well as within the ganglionic neuropil. Dystrophic neurites ranged from swollen processes with pale cytoplasm (arrows, Fig 3A) to those staining strongly with toluidine blue (arrows, Fig 3B) and corresponding ultrastructurally to collections of mitochondria.

Ultrastructural examination confirmed the light microscopic appearance, demonstrating that swollen dystrophic elements were numerous in SMG-CG of saline-treated diabetic NOD-SCID mice and less frequent in EPO or CEPO treated diabetics and treated and untreated non-diabetic age-matched controls. Dystrophic elements exhibited a variety of ultrastructural patterns based on differences in their content of subcellular organelles (Fig 4), as we have previously described (Schmidt et al., 2003). The most typical appearance (Figs 4A,B) consisted of neuritic swellings containing large numbers of mitochondria which were tightly aggregated, usually without a significant amount of intervening axoplasm, and were significantly smaller than mitochondria within adjacent neuronal perikarya (Fig 4C). Less frequently, neurites contained mixed collections of organelles (mitochondria, autophagosomes, neurofilaments and multivesicular bodies, Fig 4D). Others were composed of numerous tubulovesicular elements ranging from small numbers within a pale unstructured cytoplasm (arrow, Fig 4E) to compact aggregates (arrow, Fig 4F). Occasional elements were identified as dendrites by their content of ribosomes, lipopigment or as postsynaptic elements (arrows, Figs 4G,H). Dystrophic neurites were typically completely enclosed within the cytoplasm of Schwann or satellite cells and were often separated from adjacent perikarya by interposed satellite cell processes. In many cases it was difficult to confidently identify dystrophic elements as either axons or dendrites and, thus, we have referred to dystrophic processes simply as involving neuritic elements and the process as neuritic dystrophy. Neuritic dystrophy in the current experiments is comparable to that we have previously described in spontaneously diabetic NOD mice, STZ-treated NOD-SCID mice and various STZ-treated mouse strains (Schmidt et al., 2003).

Figure 4.

Ultrastructural appearance of neuritic dystrophy in NOD-SCID mouse SMG-CG.

A–C) Markedly enlarged neurites containing nearly pure collections of mitochondria represent the major category of neuritic dystrophy. The accumulated mitochondria in neuritic processes (arrows, C) are typically smaller than those of adjacent perikarya (arrowhead, C). (saline-treated diabetic, original magnification: A,B-3000X; C-25,000X)

D) Dystrophic neurites may also contain a variety of admixed organelles including mitochondria, tubulovesicular elements, dense bodies and neurofilaments (saline treated control, original magnification: D-25,000X)

E,F) A common ultrastructural appearance of dystrophic neurites consists of dilatations containing tubulovesicular elements which may be delicate or coarse. The contours of the sympathetic neuron cell body adjacent to the large dystrophic element (E) are distorted but the neuron appears otherwise normal. (saline-treated diabetic, original magnification: E-4000X; F-15,000X)

G,H) Synaptic specializations (arrow, G), seen at higher magnification (H), are occasionally found upon dystrophic neurites. ((saline-treated diabetic, original magnification: G-20,000X; H-60,000X)

Since dystrophic neurites were present in all groups of mice, differing only in numbers, it was necessary to apply an ultrastructural quantitative method to accurately compare their relative numbers. The numbers of dystrophic neurites were counted and expressed as a ratio (numbers of dystrophic elements/numbers of nucleated neurons). This analysis established that dystrophic neurites were increased 4-5-fold in 4 week diabetic NOD mouse SMG-CG compared to age-matched non-diabetic siblings (Table 1). The frequency of neuritic dystrophy in EPO (Table 1) and CEPO (Table 2) treated diabetic mice was not significantly different from saline treated controls.

DISCUSSION

The results of the current studies demonstrate a clear effect of EPO and CEPO given in a preventative paradigm on the development of experimental murine diabetic autonomic neuropathy. Our studies are consistent with recent studies which have shown that EPO produces a salutary effect on altered mechanical and thermal nociception, biochemistry and electrophysiology in diabetic rat somatic nerves (Bianchi et al., 2004; Roesler et al., 2004). Similarly, CEPO has demonstrated neuroprotective physiologic and structural effects in diabetic sensory neuropathy (Bianchi et al., 2004; Leist et al., 2004). EPO has been proposed to be synergistic with IGF-I in the activation of the PI3K/Akt pathway, which is diminished in somatic and vagus nerves of rats with STZ-diabetes (Cai and Helke, 2003; Huang et al., 2005). Interestingly, at least within the brain, IGF-I directly stimulates EPO production (Masuda et al., 1997) so that EPO/IGF-1 synergy may exist physiologically in the nervous system. It is known that EPO and IGF-I downstream signaling pathways are also shared with insulin. Topical insulin application to diabetic rat sciatic nerve has been shown to have salutary neurotrophic effects on diabetic somatic neuropathy (Singhal et al., 1997) in the absence of improvement of systemic glucose levels. However, although treatment with insulin at doses resulting in partial blood glucose normalization do correct neuroaxonal dystrophy in STZ-rat diabetic mesenteric nerves (Schmidt et al., 1983), subglycemic systemic doses of insulin used to mimic the transient hypoglycemic effect of IGF-I injection do not (Schmidt et al., 1999)

Neuroprotection by EPO is mediated through a heteroreceptor complex comprising both the EPO receptor and a common β-receptor subunit, also known as CD131 (Brines et al., 2004; reviewed by Brines and Cerami, 2005). Activation of EPO receptors on neurons and Schwann cells triggers neuroprotective pathways involving the PI3K/Akt cascade (Siren et al., 2001; Lipton, 2004) and protects from excitotoxic, apoptotic and oxidative stress (Brines et al., 2000). In response to experimental axonal injury, nitric oxide is thought to stimulate periaxonal Schwann cells to release EPO which binds to neuronal EPO receptors and prevents axonal degeneration (Keswani et al., 2004a). Our experiments clearly establish that prevertebral sympathetic neurons also contain EPO-receptors and preliminary studies (Schmidt et al., unpublished results) show an EPO dose-dependent increase in phosphoAkt/total Akt ratio of sympathetic neurons in culture. A peptide sequence in EPO has been shown to induce differentiation and prevent cell death in neuroblastoma cell lines in the absence of an effect on erythropoietic cell lines or mouse primary spleen cells (Campana et al., 1998).

EPO has been used in the treatment of chemotherapy-associated and other refractory anemias. Anemia in diabetes [reviewed in (McGill and Bell, 2006)] may be multifactorial but is particularly common in diabetic patients with kidney disease even before the demonstrable loss of glomerular filtration capacity (Craig et al., 2005; McGill and Bell, 2006). One potential cause of anemia in diabetes is EPO deficiency or ineffectiveness (Bosman et al., 2001, 2002). A number of investigators have described the association of autonomic neuropathy in diabetic patients with anemia (Cotroneo et al., 2000; McGill and Bell, 2006; Ricerca et al., 1999; Saito et al., 2007; Spallone et al., 2004; Winkler et al., 1999). Since denervation of the kidney or use of beta adrenergic blocking agents are known to decrease kidney EPO production in experimental animals (Beynon, 1977; Fink and Fisher, 1976, it has been thought that loss of autonomic innervation of the kidney secondarily resulted in decreased EPO production. Serum EPO levels are decreased in type 1 diabetic patients with postural hypotension in comparison to age- and duration-matched type 1 diabetics free of complications as well as non-anemic, non-diabetic controls and patients with iron deficiency anemia (Winkler et al., 1999). A blunted EPO response to severe anemia has also been described as a result of autonomic neuropathy in studies of patients with both types 1 and 2 diabetes (Cotroneo et al., 2000; Ricerca et al., 1999; Saito et al., 2007; Spallone et al., 2004).

An effect of exogenous EPO on clinical autonomic neuropathy has been observed (Hoeldtke and Streeten, 1993). In one study, 6–10 weeks of EPO treatment of 8 patients with symptomatic orthostatic hypotension, anemia and decreased red cell mass in the clinical settings of type 1 diabetes, primary autonomic failure or sympathotonic orthostatic hypotension was reported to improve both postural hypotension and anemia (Hoeldtke and Streeten, 1993). Similarly, a 7 month trial of EPO in the treatment of postural hypotension in anemic type 1 diabetic patients also showed improvement in both anemia and postural hypotension (Winkler et al., 2001). Salutary effects of EPO on postural hypotension and anemia have also been described in patients with familial amyloidotic polyneuropathy (Ando et al., 1996), primary autonomic failure (Biaggioni et al., 1994; Perera et al., 1995) and multiple system atrophy with sympathetic failure (Winkler et al., 2002).

The mechanism of EPO’s salutary effect on autonomic neuropathy is unknown and may be multifactorial. Although initially the effect of EPO was thought to reflect an increase in red cell mass or direct effects on the vasculature in the absence of hypervolemia (Hoeldtke and Streeten, 1993), it is not clear that uncomplicated anemia produces orthostatic hypotension. Indeed, one report describes improvement of severe orthostatic hypotension in the absence of improved anemia (Kawakami et al., 2003). Increased levels of serum norepinephrine, binding of nitric oxide to an increased hemoglobin concentration with resultant loss of NO-induced vasodilatation, change in viscosity of blood, increased vascular sensitivity to angiotensin II or other direct effects on smooth muscle cells have been proposed to explain EPO-induced improvement in postural hypotension (Rao and Stamler, 2002; Winkler et al., 2001). However, based on the results reported in our current investigation and these clinical studies, it is possible that the reported clinical effects of EPO on red cell mass and autonomic dysfunction may reflect separate erythropoietic and neurotrophic effects. In future experiments, it will be possible to separate an effect of increased red cell mass from a neurotrophic effect in patients with autonomic neuropathy uncomplicated by anemia by treatment with CEPO or another non-erythropoietic EPO derivative.

Although experience with EPO is extensive with apparent safety (Sowade et al., 1998), there are reported side effects of EPO administration including accelerated hypertension and risk of thrombosis (Drueke et al., 2006; Rao and Stamler, 2002; Singh et al., 2006; Spivak, 2001). Recently, caution in the use of EPO has been proposed in light of the suggestion that it may exert a possible effect on tumor growth, particularly at higher doses (Crawford, 2007; Henke et al., 2006; Steensma and Loprinzi, 2005), prompting the FDA to issue a warning for the use of erythropoiesis stimulating agents in oncology patients. Additionally, the prothrombotic effects of EPO in the setting of injury are well known. A recent study has demonstrated neuroprotection from cisplatin-induced peripheral neuropathy in the absence of an effect on tumor growth (Bianchi et al., 2007). Although these issues are currently unresolved, it may prove that agents such as carbamylated EPO may well provide the salutary effects of EPO without triggering adverse effects such as pathological thrombosis or promoting growth of tumors.