Anovel endogenous erythropoietin mediated pathway prevents axonal degeneration.код для вставкиСкачать
A Novel Endogenous Erythropoietin Mediated Pathway Prevents Axonal Degeneration Sanjay C. Keswani, MRCP,1 Ulas Buldanlioglu, MS,1 Angela Fischer, MS,1 Nicole Reed, BSc,1 Michelle Polley, BSc,1 Hong Liang,1 Chunhua Zhou,1 Christelene Jack, MS,1 Gerhard J. Leitz, MD,3 and Ahmet Hoke, MD, PhD1,2 Clinically relevant peripheral neuropathies (such as diabetic and human immunodeficiency virus sensory neuropathies) are characterized by distal axonal degeneration, rather than neuronal death. Here, we describe a novel, endogenous pathway that prevents axonal degeneration. We show that in response to axonal injury, periaxonal Schwann cells release erythropoietin (EPO), which via EPO receptor binding on neurons, prevents axonal degeneration. We demonstrate that the relevant axonal injury signal that stimulates EPO production from surrounding glial cells is nitric oxide. In addition, we show that this endogenous pathway can be therapeutically exploited by administering exogenous EPO. In an animal model of distal axonopathy, systemic EPO administration prevents axonal degeneration, and this is associated with a reduction in limb weakness and neuropathic pain behavior. Our in vivo and in vitro data suggest that EPO prevents axonal degeneration and therefore may be therapeutically useful in a wide variety of human neurological diseases characterized by axonopathy. Ann Neurol 2004;56:815– 826 Peripheral neuropathies are common and cause significant morbidity. Most peripheral neuropathies, including diabetic and human immunodeficiency virus (HIV)–associated neuropathy, are “dying-back” axonopathies, characterized by degeneration of the most distal portions of axons, with centripetal progression.1,2 Although most published in vitro studies of neurotoxicity and neuroprotection in the peripheral nervous system have focused on neuronal apoptosis as the sole outcome measure, neuronal death, in contrast with distal axonal loss, is not a prominent pathological feature of most human peripheral neuropathies. Furthermore, the signaling pathways mediating axonal degeneration are distinguishable from those mediating neuronal apoptosis.3–7 Thus, in considering whether a particular “neuroprotective” agent may have therapeutic relevance to human peripheral neuropathies (and to other neurological diseases where axonopathy is prominent), it is important to discover if it robustly prevents axonal degeneration, independent of neuronal death.8 The glycoprotein, erythropoietin (EPO), and its cognate receptor, EPOR, are expressed both in the central nervous system (CNS) and the peripheral nervous system (PNS). In the CNS, endogenous EPO production by astrocytes is thought to mediate the phenomenon of ischemic preconditioning.9 In this study, we examine the interaction between sensory axons and Schwann cells, the major glial cells of the PNS, and demonstrate a novel, endogenous “axonoprotective” pathway mediated by Schwann cell–derived EPO. We show that axonal injury from a variety of causes stimulates adjacent Schwann cells to produce EPO, which via EPOR binding on neurons, prevents axonal degeneration. We demonstrate that nitric oxide (NO) is a relevant “axonal injury factor” that stimulates neighboring Schwann cells to produce EPO. Finally, we show that this endogenous axonoprotective pathway can be exploited for therapeutic purposes. In a well-characterized rodent model of distal sensorimotor axonal polyneuropathy, we demonstrated that systemic EPO administration ameliorates axonal degeneration, limb weakness, and neuropathic pain behavior. Our data suggest that recombinant EPO may be therapeutically useful in peripheral neuropathies and other neurodegenerative dis- From the Departments of 1Neurology and 2Neuroscience, The Johns Hopkins Hospital, Baltimore, MD; and 3Ortho Biotech Products LP, Raritan, NJ. Address correspondence to Dr Keswani, Department of Neurology, Johns Hopkins University, 600 N. Wolfe Street, Path 627, Baltimore, MD 21287. E-mail: firstname.lastname@example.org Received Received Jul 1, 2004, and in revised form Aug 13. Accepted for publication Aug 13, 2004. Published online Oct 6, 2004 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/ana.20285 © 2004 American Neurological Association Published by Wiley-Liss, Inc., through Wiley Subscription Services 815 eases where dying-back axonal degeneration is a characteristic feature. Materials and Methods Pharmacological Agents Anti–EPOR antibody and anti–glial fibrillary acidic protein (GFAP) antibody were obtained from Santa Cruz Biotechnology (Santa Cruz, CA); anti–EPO antibody, FITCconjugated ␣-bungarotoxin, TRIM (1-2, trifluoromethylphenylimidazole), L-NAME, ddC, and acrylamide from SigmaAldrich (St Louis, MO); anti–␤-III tubulin antibody from Promega (Madison, WI); recombinant HIV-1 gp120-MN (⬎95% pure) from ImmunoDiagnostics (Woburn, MA); SNAP and NOR-3 from Calbiochem (San Diego, CA); anti–PGP 9.5 antibody from Biogenesis (Poole, UK). Recombinant EPO was kindly provided by Ortho-Biotech Pharmaceuticals (Raritan, NJ). Tissue culture supplies were obtained from Invitrogen (Carlsbad, CA) unless noted otherwise. Preparation of Pure Schwann Cell Cultures Schwann cells were prepared from 1-day-old Sprague-Dawley rat pups and purified by a modified Brocke’s method.10 Two days before coculture experiments, purified Schwann cells were dissociated using brief trypsinization and plated onto poly-L-lysine and rat-tail collagen-coated (Collaborative Biomedical Products, Bedford, MA) glass coverslips in 24-well tissue culture plates at a density of 10,000 cells per well. Preparation of Dissociated Dorsal Root Ganglion Cocultures Dissociated primary dorsal root ganglion (DRG) neuronal cell cultures were prepared as described by Eldridge and colleagues.11 In brief, the DRGs from Day 15 embryos were dissected and dissociated with 0.25% trypsin in L-15 medium. Dissociated cells (13,000 cells/well) then were directly plated onto glass coverslips already bearing Schwann cells. The cultures were maintained in Neurobasal medium containing 1% fetal bovine serum (FBS; HyClone, Logan, UT) and glial cell line-derived neurotrophic factor (10ng/ml). Measurement of Axonal Degeneration After 24 hours of incubation of dissociated DRG cocultures with the agents of interest, the cells were fixed and then immunostained with anti–␤-III tubulin antibody. Total axonal length for a minimum of 10 neurons per coverslip per condition was measured using an image analysis system. Each experimental condition was done in triplicate wells and repeated three to six times. The results from each set of experiments were averaged and counted as n ⫽ 1 for statistical analysis. Statistical significance was determined in Statview (Macintosh version 5.0.1) using analysis of variance (ANOVA) with correction for multiple comparisons (the critical ␣ level set at p ⫽ 0.005). The same method was used for the statistical analysis of other data (see below). Erythropoietin and Erythropoietin Receptor Immunocytochemistry in Dorsal Root Ganglion Cultures To demonstrate the presence of EPO and EPOR in DRG neurons and Schwann cells, we performed triple immunoflu- 816 Annals of Neurology Vol 56 No 6 December 2004 orescent labeling in fixed DRG cultures with either anti– EPO (1:100) or anti–EPOR (1:100) antibodies, and with antibodies against ␤-III tubulin (neuronal marker) at 1 to 1,000 and against GFAP (Schwann cell marker) at 20g/ml. Omission of one of these primary antibodies or its replacement with nonspecific IgG yielded no labeling. To demonstrate whether there was any change in EPO or EPOR immunostaining in the DRG cultures after the addition of the agent of interest, we performed the above-described triple immunostaining in DRG cultures 6 hours after the addition of the agent or vehicle control. Each experiment was performed in triplicate wells and repeated twice. Erythropoietin and Erythropoietin Receptor Immunocytochemistry in Adult Rat Dorsal Root Ganglion Sections Adult male Sprague-Dawley rats (n ⫽ 4) were perfused with 4% paraformaldehyde, and the L4 and L5 DRGs were harvested. After further fixation in 4% paraformaldehyde overnight, the tissue were transferred to 30% sucrose and sectioned at 7m on a cryostat. Immunostaining for EPO and EPOR with double labeling for ␤-III tubulin or GFAP were done as described above. Western Blotting for Erythropoietin Protein concentrations of the samples of interest were determined by the Bradford assay, and the samples boiled in sodium dodecyl sulfate sample buffer for 5 minutes. An equal amount of protein from each sample (20g) was loaded into the lanes of an sodium dodecyl sulfate polyacrylamide gel electrophoresis 12% gel. After electrophoresis, the proteins were electrotransferred onto nitrocellulose membrane. The blot then was probed with anti–EPO antibody (1:100). Immunoreactive bands were visualized using enhanced chemiluminescence according to the manufacturer’s protocol (Amersham, Buckinghamshire, UK). Experiments were performed at least three times. Erythropoietin Enzyme-Linked Immunosorbent Assay This was performed using the human EPO enzyme-linked immunosorbent assay (ELISA) kit (R&D, Minneapolis, MN) according to the manufacturer’s protocol. Each experimental condition was performed in triplicate wells and repeated three times. In addition, the supernatant from each well was analyzed by ELISA in a triplicate fashion. The results from each set of experiments were averaged and counted as n ⫽ 1 for statistical analysis. Statistical analysis was done using ANOVA as described above. Reverse Transcription Polymerase Chain Reaction for Erythropoietin and Erythropoietin Receptor Total RNA from DRG and proximal sciatic nerve was isolated at 4 hours (n ⫽ 4) and 24 hours (n ⫽ 4) after sciatic axotomy, using standard methods. Real-time RT-PCR was performed using SYBR green kits and standard protocols on the Opticon real-time PCR machine (MJ Research, Waltham, MA). The primers for EPO were AGTCGCGTTCTGGAGAGGTA (forward) and TGCAGAAAGTATCCGCTGTG (reverse) with a Tm of 65.5°C. The primers for EPOR were CCCAAGTTTGAGAGCAAAGC (forward) and GCGTCCAGGAGCACTACTTC (reverse) with a Tm of 64°C. In each animal, the amount of EPO or EPOR mRNA on the injured side was expressed as a percentage of the contralateral side, which served as an intrinsic control. Statistical analysis was done using ANOVA as described above. Detection of Intraneuronal Nitric Oxide Production This was performed using the fluorometric cell-associated NOS detection system (Sigma), according to the manufacturer’s protocol. This kit is a specific assay for the measurement of free NO and NO synthase (NOS) activity in living cells and comprises a reaction mix containing the cofactor NADPH, arginine, and the fluorescent NO indicator DAF2A.12 DAF-2A (nonfluorescent) enters cells and is hydrolyzed by cytosolic esterases to DAF-2 trapped inside the cells. DAF-2, in turn, reacts with NO produced by NOS to form DAF-2T (triazolo-fluorescein), which is green and highly fluorescent (DAF-2T). To determine the cellular location of DAF-2T, we then fixed the cocultures and immunostained them with anti–␤-III tubulin (a neuronal marker) antibody, secondarily labeled with a Texas Red–conjugated antibody. Erythropoietin Small Interfering RNA Experiments Downregulation of erythropoietin expression in Schwann cells was accomplished using RNA interference. Two target sequences in the rat erythropoietin gene (GenBank accession number NM 017001) were identified using the small interfering RNA (siRNA) target finder and design tool at Ambion’s Web site (http://www.ambion.com/techlib/misc/ siRNA_finder.html). These were labeled R1 (AACTTCTACGCTTGGAAAAGA) and R2 (AAAAGAATGAAGGTGGAAGAA). Appropriate siRNAs were generated with the Silencer siRNA Construction kit (catalogue no. 1620) and labeled with Silencer siRNA Labeling kit (Cy3, catalogue no. 1632), both from Ambion (Austin, TX). Various concentrations of the R1 and R2 were used to find the optimum construct and concentration to downregulate the erythropoietin gene expression in Schwann cells. Of the two siRNA constructs R2 achieved the best Schwann cell transfection rate at 0.5M and was consequently used. Pure Schwann cell cultures were transfected with Lipofectamine 2000 (Invitrogen, Carlsbad, CA) for 24 hours and the cells were washed before plating of DRG neurons. Efficiency of siRNA transfection was determined by counting the number of neurons and Schwann cells labeled with Cy3-labeled siRNA constructs. No neurons were Cy-3 labeled, but greater than 90% of Schwann cells were Cy-3 labeled. Acrylamide Axonopathy Model Acrylamide neuropathy was induced in Sprague-Dawley rats according to published reports.13,14 The experimental procedure was approved by the Animal Care and Use Committee of the Johns Hopkins University. All procedures were conducted in 6- to 8-week-old female Sprague-Dawley rats weighing 70 to 80gm at the beginning of the study. Neuropathy was induced by administration of acrylamide (electrophoresis grade; Sigma-Aldrich) in their drinking water at a concentration of 400ppm for 2 weeks. One group of animals (n ⫽ 10) received daily intraperitoneal (IP) administration of recombinant human erythropoietin (rhEPO) at 2,500IU/kg/ day for 3 weeks. The second group (n ⫽ 10) received daily vehicle saline intraperitoneal injections for the same duration. Another group of animals (n ⫽ 5) did not receive any acrylamide or rhEPO and served as normal controls. Behavioral testing was done according to standard protocols. Motor strength was examined by grip strength measurements as detailed by Crofton and colleagues,15 and sensory testing was done by examining for mechanical hypo/hyperalgesia using von Frey filaments and the method outlined by Dixon and colleagues.16 At the end of the study, animals were perfused with 4% paraformaldehyde and tissues were obtained for further analysis. Skin innervation was assessed by PGP 9.5 immunohistochemistry according to published protocols.13 Changes in motor innervation were assessed by examining the innervated neuromuscular junction density in the intrinsic foot muscles using ␣-bungaratoxin binding.17 Statistical analysis was done using ANOVA as described above. Results In Vitro and In Vivo Erythropoietin and Erythropoietin Receptor Expression in Dorsal Root Ganglion Neurons and Schwann Cells Immunostaining of dissociated DRG neuron Schwann cell cocultures showed that both neurons and Schwann cells expressed EPO (Fig 1A), whereas neurons predominantly expressed EPOR (see Fig 1B). Of interest, as shown in Figure 1B, neuronal EPOR was localized on axons as well as perikarya. A similar pattern of EPO and EPOR immunostaining was observed in DRG sections harvested from adult rats (see Fig 1C, D), EPOR immunostaining again being particularly intense in DRG neurons compared with Schwann cells. Neighboring Schwann Cells Produce Erythropoietin in Response to Axonal Injury Axonal injury, in the absence of neuronal death, was induced in multiple ways. In one injury paradigm, the axons of well-established, dissociated DRG cultures were transected and 4 hours later were fixed and triple labeled for EPO, GFAP (Schwann cell marker), and ␤-III tubulin (neuronal marker). As demonstrated in Figure 2A, EPO immunostaining was markedly increased in Schwann cells that were in close proximity to the transected axons as compared with those adjacent to uninjured axons. An in vivo axotomy model also was established. Four hours after unilateral sciatic nerve transection in adult rats, EPO mRNA levels in both the proximal sciatic nerve and lumbar DRG harvested from the same side as the transection, were fourto fivefold higher by semiquantitative real-time RTPCR than those harvested from the nontransected side (see Fig 2B). By 24 hours, EPO mRNA levels were less than at 4 hours but were still more than twofold higher compared with the control side. Of interest, mRNA levels for EPOR, the receptor for EPO, were increased in lumbar DRG (but not in sciatic nerve) harvested Keswani et al: EPO Prevents Axonal Degeneration 817 Fig 1. In vitro and vivo erythropoietin (EPO) and EPO receptor (EPOR) expression by dorsal root ganglion (DRG) neurons and Schwann cells. (A) Triple immunoflouresent labeling of dissociated DRG neuron Schwann cell cocultures shows that EPO immunostaining (red) is present in both neurons (␤III tubulin labeled) and Schwann cells (GFAP labeled). (B) EPOR is present in axons and cell bodies of DRG neurons in the cocultures (C) Immunostaining of adult rat DRG sections shows that EPO (red) is expressed by DRG neurons and perineuronal Schwann cells (arrows). (D) In contrast, similar to the in vitro staining, EPOR immunoreactivity (red) is mainly in DRG neurons, rather than Schwann cells. Scale bars ⫽ 50m. from the transected side, being 3.5- and 2.5-fold higher than the control side at 4 and 24 hours, respectively. To explore other axonal injury paradigms, we exposed dissociated DRG cultures to neurotoxins at doses known to reproducibly cause “dying-back” axonal degeneration but not neuronal death. The neuro- 818 Annals of Neurology Vol 56 No 6 December 2004 toxins chosen were gp120, the HIV envelope glycoprotein, and ddC, an antiretroviral agent known to cause peripheral neuropathy. The neurotoxicity profiles of these agents have been well characterized in our culture system.18,19 After 24 hours of exposure to gp120 (1pg/ml) or ddC (0.1M), EPO immunostaining in perineuronal Schwann cells was greatly increased com- Fig 2. Neighboring Schwann cells (SC) produce erythropoietin (EPO) in response to axonal injury. (A) Axotomy in dorsal root ganglion (DRG) neuron Schwann cell cocultures results in a marked increase at 4 hours in the expression of EPO (red) in Schwann (GFAP labeled) cells neighboring transected axons (␤III tubulin labeled as highlighted by arrows). Scale bar ⫽ 100m. (B) Semiquantitative real-time polymerase chain reaction for rat EPO shows a robust increase in EPO mRNA levels in both the Lumbar DRG and the proximal sciatic nerve at 4 and 24 hours after axotomy at the midsciatic level. Accompanying this, there is a more modest increase in EPOR mRNA in the lumbar DRG after sciatic nerve transection (*p ⬍ 0.05 compared to control). (C) After the application of doses of ddC (0.1M) or gp120 (1pg/ml) known to cause modest axonal degeneration (and no neuronal death), expression of EPO (red) is greatly increased in periaxonal Schwann cells. Cultures were fixed 24 hours after toxin application. Scale bar ⫽ 100m. (D) Western blotting of supernatants of DRG cocultures exposed for 24 hours to ddC (0.1M), gp120 (1pg/ml) or vehicle control shows evidence of EPO release in neurotoxin-treated cultures but not in vehicle control–treated cultures. In contrast, there is no evidence of EPO release in pure Schwann cell cultures treated with ddC and gp120. EPO-R ⫽ EPO receptor. Keswani et al: EPO Prevents Axonal Degeneration 819 pared with vehicle control–treated cultures (see Fig 2C). To investigate whether there was extracellular EPO release, we analyzed the supernatants of gp120, ddC, and vehicle control–treated DRG cultures by Western blotting for the presence of EPO. As shown in Figure 2D, EPO was increased in the supernatants of neurotoxin-treated cultures compared with that of vehicle control–treated cultures. In contrast, when gp120 or ddC was applied to pure Schwann cell cultures, no EPO induction was noted by EPO immunostaining or Western blotting of supernatants. These findings coupled with our observations that, after neurotoxin exposure or axonal transection, only those Schwann cells in intimate contact with axons had increased EPO immunostaining suggest that “sick axons” are needed in close proximity to Schwann cells for EPO induction in those Schwann cells to occur. Correlating with the observed increase in DRG EPOR mRNA after in vivo axotomy, we noted an increase in the intensity of neuronal EPOR immunostaining in DRG cultures after exposure to the neurotoxins, ddC, or gp120 (data not shown). Axonal Injury Stimulates Schwann Cell Production of Erythropoietin via Nitric Oxide We next attempted to determine the nature of the “axonal injury factor” that triggers EPO production by neighboring Schwann cells. We noted that axonal injury caused by a wide variety of stimuli, including exposure to gp120, ddC, and acrylamide (a known axonal toxin), all resulted in a robust increase in intraneuronal NO production, as detected by fluorescent DAF-2 staining (Fig 3A). Furthermore, NO donors, such as SNAP and NOR-3, stimulated pure Schwann cell cultures to produce EPO mRNA, as early as 30 minutes after exposure (see Fig 3B). This was mirrored by an increase in intracellular EPO production by Western blotting (see Fig 3C) and a large increase in EPO content in the supernatants of these cultures, as measured by ELISA (see Fig 3D). To evaluate how critical nitric oxide was in the induction of EPO production by axonal injury, we administered L-NAME, a broad NOS inhibitor, with gp120 (at the previously used dose known to cause axonal degeneration and no death) to the DRG cultures and measured the EPO content in the supernatants after 24 hours. As shown in Figure 3E, the addition of L -NAME obliterated the ability of gp120 to induce EPO release from the mixed cultures. Moreover, the application of TRIM, a specific neuronal NOS (nNOS) inhibitor, completely prevented the induction of EPO mRNA by gp120 in these cultures, suggesting that NO generated by nNOS was responsible for triggering EPO production by surrounding glial cells (see Fig 3F). In our dissociated DRG cultures, immunostaining for nNOS only occurred in neurons, in contrast with iNOS staining, which was present in both neurons and Schwann cells (data not shown). 820 Annals of Neurology Vol 56 No 6 December 2004 Schwann Cell–Derived Erythropoietin Prevents Axonal Degeneration To assess the relevance of this endogenous EPO production, we incubated dissociated DRG cultures with antibodies known to bind EPO (anti-EPO) or block its interaction with EPOR (anti-EPOR) along with gp120, ddC, or vehicle control.9 As before, doses of gp120 and ddC that we knew caused only modest axonal degeneration (and no neuronal death) were chosen. Figure 4A shows the results of these experiments. When anti-EPO or anti-EPOR was coadministered with gp120 or ddC, the neurotoxicity of these agents was markedly exacerbated, as judged by more extensive axonal degeneration. There was no associated increase in neuronal death, as judged by ethidium homodimer staining (data not shown), suggesting that the observed increase in axonal degeneration was not caused by neuronal death. Administration of the EPO/EPOR antagonists by themselves did not cause any neurotoxicity. To investigate the importance of EPO production by Schwann cells in the cocultures, we first transfected the Schwann cell monolayer on the coverslips with antiEPO siRNA using a cationic lipid vector and then washed it before the addition of dissociated DRG neurons. Transfection with fluorescently labeled (Cy3tagged) anti-EPO siRNA confirmed that greater than 90% of Schwann cells in the cultures were transfected as indicated by the proportion of GFAP-positive cells that were Cy3 labeled. Of note, there was no evidence of any neuronal transfection in the mixed cultures (ie, no Cy3 labeling of ␤-III tubulin–positive cells). After the addition of DRG neurons, these transfected cultures then were exposed to gp120, ddC, or vehicle control for 24 hours, before being fixed and triple labeled for EPO, ␤-III tubulin, and GFAP. Figure 4B shows representative confocal microscope slides of these immunostained cultures. It can be observed that in the siRNAtransfected cultures, gp120 or ddC failed to induce EPO immunostaining in perineuronal Schwann cells, which had been observed in a robust manner in similarly treated nontransfected cultures. The selectivity of the effect of the anti-EPO siRNA construct (R2) was supported by the finding that another EPO siRNA construct (R1) did not suppress EPO production in Schwann cells. In the R2-transfected cultures, there was marked exacerbation of axonal degeneration by gp120 and ddC (see Fig 4C). Indeed, normally nonneurotoxic doses of these agents could be rendered neurotoxic by anti-EPO siRNA transfection of Schwann cells in the culture. As shown in Figure 4C, anti-EPO siRNA transfection alone did not cause significant neurotoxicity. Application of an Neuronal NOS Inhibitor Augments gp120-Induced Axonal Degeneration Similar to the effect of endogenous EPO antagonism in the DRG cocultures, the application of TRIM, a spe- Fig 3. Axonal injury stimulates Schwann cell production of erythropoietin (EPO) via nitric oxide. (A) Exposure of dorsal root ganglion (DRG) cocultures to agents causing axonal degeneration, including gp120 (1pg/ml) and acrylamide (1mM), induces NO production (green) at 6 hours in ␤-III tubulin–labeled neurons (red), as assayed by DAF-2T fluorescence. Scale bar ⫽ 100m. (B) Nitric oxide (NO) donors, SNAP (10M) and NOR-3 (100nM), induce increased EPO mRNA levels (three- to fourfold at 1 hour, p ⬍ 0.05) in pure Schwann cell cultures. (C) Cell lysates of pure Schwann cell cultures treated for 6 hours with SNAP (10M) or NOR-3 (100nM) have increased EPO protein by Western blotting, compared with those treated with vehicle control. (D) Supernatants of pure Schwann cell cultures treated for 24 hours with SNAP (10M) or NOR-3 (100nM) have markedly increased EPO content by ELISA, compared with vehicle control treatment (*p ⬍ 0.05). (E) L-NAME (100M) coadministration prevents gp120-induced EPO release from DRG neuron Schwann cell cocultures, as measured by EPO ELISA (*p ⬍ 0.05). (F) TRIM (100M) coadministration abrogates gp120-induced (18-fold) increase in Schwann cell EPO mRNA in DRG neuron Schwann cell cocultures (*p ⬍ 0.05). cific nNOS inhibitor, resulted in markedly increased axonal degeneration induced by gp120 (1pg/ml; see Fig 4D). The application of TRIM by itself did not cause any axonal degeneration. As in previous experiments, TRIM coadministration with gp120 did not result in increased neuronal death. These findings, in combination with our previous observations (see Fig 3), suggest the importance of nNOS in the endogenous EPO “axonoprotective” response to axonal injury. Systemic (exogenous) EPO Administration Prevents Axonal Degeneration in an Animal Model of Peripheral Neuropathy We then assessed whether systemic administration of recombinant EPO could prevent axonal degeneration in an animal model of distal axonopathy. In this wellestablished model, oral acrylamide administration to Sprague-Dawley rats results in severe dying-back degeneration of both sensory and motor fibers, in the absence Keswani et al: EPO Prevents Axonal Degeneration 821 Fig 4. Endogenous erythropoietin (EPO) production by Schwann cells is axonoprotective. (A) The application of antibodies against EPO or EPO receptor (EPOR) increases the sensitivity of DRG axons to ddC- (0.1M) and gp120- (1pg/ml) induced degeneration. (*p ⬍ 0.05). (B) Upregulation of EPO expression in Schwann cells in response to ddC or gp120 can be prevented by prior transfection of Schwann cells with small interfering RNA (siRNA) against EPO. Note that in siRNA-treated cultures the neurons have very short axons. Scale bar ⫽ 50m. (C) Downregulation of endogenous erythropoietin expression in Schwann cells within DRG cocultures by EPO siRNA increases axonal degeneration induced by ddC and gp120 (*p ⬍ 0.05). (D) Similarly, coadministration of TRIM (100M) to DRG cocultures increases gp120-induced axonal degeneration (*p ⬍ 0.05). of significant neuronal death.14,20 Affected rats characteristically have distal limb weakness and an ataxic gait.13,21 As shown in Figure 5A and quantified in Figure 5C, acrylamide-treated rats given EPO had significantly less sensory axonal degeneration as indicated by greater cutaneous innervation (increased epidermal nerve fiber density) on PGP 9.5 immunohistochemistry, compared with those given placebo. This correlated with decreased mechanical hyperalgesia on von Frey filament testing (see Fig 5B). Furthermore, EPO-treated rats had significantly less motor axonal degeneration as demon- 822 Annals of Neurology Vol 56 No 6 December 2004 strated by higher innervated neuromuscular junction density in the intrinsic foot muscles using ␣-bungaratoxin binding (see Fig 5C). This correlated with greater grip strength (see Fig 5D). Discussion Progressive dying-back degeneration of the distal regions of long axons, rather than neuronal loss, is the predominant pathological change in the most common peripheral neuropathies afflicting humans, such as diabetic sensorimotor polyneuropathy, HIV-associated Fig 5. Systemic erythropoietin (EPO) prevents axonal degeneration in an animal model of distal axonopathy. (A) Representative PGP 9.5–immunostained skin biopsies from rat paw showing increased cutaneous innervation in EPO versus vehicle-treated animals. Scale bar ⫽ 100m. (B) EPO administration prevents mechanical hyperalgesia induced by acrylamide at Day 21, as indicated by increased paw threshold response to von Frey filament (p ⬍ 0.05). Scale bars ⫽ 100m. (C) EPO application prevents acrylamide-induced distal motor axonal degeneration as indicated by greater neuromuscular junction (NMJ) density in foot intrinsic muscles, as compared with vehicle-treated animals. EPO also prevents acrylamide-induced distal sensory axonal degeneration as indicated by increased intraepidermal nerve fiber (ENF) density in treated versus nontreated animals (*p ⬍ 0.05). (D) Systemic EPO administration ameliorates loss of grip strength induced by acrylamide (p ⬍ 0.05). sensory neuropathy, and toxic neuropathies.1,2,22–24 Furthermore, progressive axonal loss is observed in multiple sclerosis and is now thought to highly correlate with disability.25,26 Consequently, agents with axonoprotective properties may be very helpful therapeutically. However, often only the antiapoptotic properties of putative neuroprotective agents are evaluated, with little or no attention paid to whether axonal degeneration can be prevented. It does not necessarily follow that an agent that prevents neuronal apoptosis will prevent axonal degeneration, because it is now well recognized that the two processes may exploit different signaling pathways.3–7 The glycoprotein EPO is a very promising neuroprotective agent, whose antiapoptotic properties have been thoroughly evaluated by several investigators. The ad- ministration of EPO prevents central nervous system neurons from death caused by a variety of insults, including hypoxia, hypoglycemia, glutamate toxicity, growth factor deprivation, and free radical injury.9,27–30 Recently, Campana and Myers demonstrated that EPO administration also prevented apoptosis of DRG sensory neurons.31 However, the ability of EPO to prevent axonal degeneration has as yet been unexplored. Furthermore, although EPO and EPOR are known to be expressed in both the CNS and PNS,32–39 their functional relevance is largely unknown, except for their integral role in mediating the phenomenon of ischemic preconditioning of the brain. In keeping with a recent study,33 we show that EPO is expressed by DRG neurons and Schwann cells both in vivo and in our in vitro cultures. In addition, we Keswani et al: EPO Prevents Axonal Degeneration 823 demonstrate that axons, in addition to neuronal cell bodies, robustly stain for EPOR. Schwann cells in contrast did not have appreciable EPOR staining in vivo and in vitro cultures. After transection of sensory axons, we noted that Schwann cells in contact with transected axons markedly increased their expression of EPO. In contrast, Schwann cells that were more remote from the transected axons and those Schwann cells in contact with nontransected axons did not upregulate their EPO production. After unilateral sciatic nerve transection in adult rats, EPO mRNA levels increased four to fivefold in both the sciatic nerve as well as the lumbar DRG harvested from the cut side as compared with the contralateral noncut side. Because sciatic nerve does not contain neuronal mRNA (there are no neuronal cell bodies in peripheral nerve), it is likely that the increased EPO mRNA production occurred in Schwann cells rather than in neurons. This correlates with a study by Campana et al33, which showed that Schwann cells in peripheral nerve have increased EPO immunostaining after in vivo axotomy. Of interest, in our study, EPOR mRNA levels were increased in the DRG from the cut side, suggesting an increase in neuronal EPOR expression. Correlating with this, we did observe an increase in intensity of neuronal EPOR immunostaining after axotomy in our in vitro cultures. Other axonal injury paradigms were used to investigate the universality of the endogenous EPO response. Dissociated DRG cocultures were exposed to neurotoxins at doses known to reproducibly cause dying-back axonal degeneration but not neuronal death in our culture system.18,19 Exposure to these agents resulted in increased EPO production and release by periaxonal Schwann cells. In contrast, when the agents were applied to pure Schwann cell cultures, no EPO induction was noted. Our findings thus far suggested that only Schwann cells in intimate contact with injured axons increased their EPO production. We then attempted to elucidate the identity of the neuronal/axonal “injury factor” that stimulated EPO production by neighboring Schwann cells. We screened several promising candidates, including ␤-neuregulin-1 and insulin growth factor–1 (IGF-1), without success. Finally, we discovered that nitric oxide may be the relevant signaling molecule, on the basis of the following observations. All the agents that we noted had caused dying-back axonal degeneration in our cultures, including gp120, ddC, and acrylamide, increased neuronal intracellular NO production. This observation correlates with previous studies showing that nNOS gene expression is significantly increased at 4 hours in ipsilateral DRG samples after sciatic nerve injury in a rat tourniquet model.40 We also noted that NO donors, such as SNAP and NOR-3, increased EPO mRNA levels in pure Schwann cell cultures as early as 30 minutes 824 Annals of Neurology Vol 56 No 6 December 2004 after administration, with a three- to fourfold increase noted at 1 hour. This correlated with an increase in EPO levels in cell lysates and supernatants of pure Schwann cell cultures exposed to NO donors as compared with vehicle control. Coadministration of L-NAME, a nonspecific NOS inhibitor, almost completely obliterated the ability of gp120 to induce EPO release into the supernatants of DRG cultures. Moreover, TRIM, a specific nNOS inhibitor,41 completely prevented the 18-fold induction of EPO mRNA by gp120 in these cultures, suggesting that NO generated by nNOS was responsible for triggering EPO production by surrounding glial cells. In our dissociated DRG cultures, immunostaining for nNOS occurred only in neurons, in contrast with iNOS staining which was present in both neurons and Schwann cells (data not shown). What is the relevance of this Schwann cell–derived EPO? When EPO gene silencing in Schwann cells was performed by transfection with anti-EPO siRNA, DRG axons were noted to be far more vulnerable to degeneration by ddC and gp120. This was associated with a lack of EPO induction in periaxonal Schwann cells by ddC and gp120 in the transfected cultures. The “axonoprotective” efficacy of endogenous EPO was further suggested by similarly increased axonal degeneration by ddC and gp120 when antagonist antibodies to EPO or to EPOR were coadministered. No associated increase in neuronal death was observed by ethidium homodimer staining (which would detect both apoptotic and necrotic death), suggesting that the increased axonal degeneration by EPO/EPOR antagonism was not caused by neuronal death. Figure 6 summarizes our model of endogenous axonoprotection by Schwann cell–derived EPO. The dual role of NO for neurotoxicity and neuroprotection has been commented on in the literature.42,43 Although NO-mediated neurotoxicity has been explored by several groups over the years, NOmediated neuroprotection is poorly understood. In our study, nNOS inhibition exacerbates the axonal degeneration induced by gp120. This correlates with prevention by TRIM of Schwann cell–derived EPO production in response to axonal injury. Of some relevance to this discussion is a study by Keilhoff and colleagues, which showed that nNOS knockout mice had worsened axonal degeneration following sciatic nerve transaction compared with wild-type mice.44 We show that this endogenous axonoprotective pathway can be therapeutically exploited by systemic administration of EPO in a well-characterized animal model of peripheral axonal degeneration, namely, the rat acrylamide toxicity model.13,14,20,22 EPO administration significantly ameliorated sensory and motor axonal degeneration, with associated reduction in neuropathic pain behavior and improved grip strength. We Fig 6. Schematic diagram of erythropoietin (EPO)–mediated intrinsic axonoprotective pathway, Based on our data, we hypothesize the following: Axonal injury (1) induces nitric oxide production (2) within neurons. This neuron-derived nitric oxide (NO) stimulates EPO production (3) by neighboring Schwann cells. This Schwann cell–derived EPO results in activation of an “axonoprotective” pathway (4) via EPOR ligation on neurons. feel that this animal model of dying-back axonopathy is far more relevant to most human peripheral neuropathies, compared with the frequently used “models” of peripheral neuropathy that comprise experimental crush injury or transection of peripheral nerves. Our findings suggest that recombinant EPO may be therapeutically useful in peripheral neuropathies where a dying-back axonopathy is a characteristic feature. These neuropathies, which include HIV-associated sensory neuropathy and diabetic neuropathy, have high prevalence, cause a great deal of morbidity, and currently have no pathogenesis-directed therapies. EPO is a particularly attractive candidate for clinical use as a neuroprotective agent, because it has already been used for several years to treat the anemia associated with renal failure and hematological malignancy and is known to be safe and effective with few side effects.45 This work was supported by grants from the NIH (National Institutes of Health, NS43991, NS46262, A.H.; NS47972, S.K.), and the R. W. Johnson Pharmaceutical Research Institute (A.H.). References 1. Sidenius P. The axonopathy of diabetic neuropathy. Diabetes 1982;31:356 –363. 2. Pardo CA, McArthur JC, Griffin JW. HIV neuropathy: insights in the pathology of HIV peripheral nerve disease. J Peripher Nerv Syst 2001;6:21–27. 3. Raff MC, Whitmore AV, Finn JT. Axonal self-destruction and neurodegeneration. Science 2002;296:868 – 871. 4. Zhai Q, Wang J, Kim A, et al. Involvement of the ubiquitinproteasome system in the early stages of wallerian degeneration. Neuron 2003;39:217–225. 5. Ehlers MD. Deconstructing the axon: Wallerian degeneration and the ubiquitin-proteasome system. Trends Neurosci 2004; 27:3– 6. 6. Glass JD, Culver DG, Levey AI, Nash NR. Very early activation of m-calpain in peripheral nerve during Wallerian degeneration. J Neurol Sci 2002;196:9 –20. 7. Korhonen L, Lindholm D. The ubiquitin proteasome system in synaptic and axonal degeneration: a new twist to an old cycle. J Cell Biol 2004;165:27–30. 8. Coleman MP, Perry VH. Axon pathology in neurological disease: a neglected therapeutic target. Trends Neurosci 2002; 25:532–537. 9. Ruscher K, Freyer D, Karsch M, et al. Erythropoietin is a paracrine mediator of ischemic tolerance in the brain: evidence from an in vitro model. J Neurosci 2002;22:10291–10301. 10. Brockes JP, Fields KL, Raff MC. Studies on cultured rat Schwann cells. I. Establishment of purified populations from cultures of peripheral nerve. Brain Res 1979;165:105–118. 11. Eldridge CF, Bunge MB, Bunge RP, Wood PM. Differentiation of axon-related Schwann cells in vitro. I. Ascorbic acid regulates basal lamina assembly and myelin formation. J Cell Biol 1987;105:1023–1034. 12. Kojima H, Nakatsubo N, Kikuchi K, et al. Detection and imaging of nitric oxide with novel fluorescent indicators: diaminofluoresceins. Anal Chem 1998;70:2446 –2453. 13. Ko MH, Chen WP, Lin-Shiau SY, Hsieh ST. Age-dependent acrylamide neurotoxicity in mice: morphology, physiology, and function. Exp Neurol 1999;158:37– 46. 14. Fullerton PM, Barnes JM. Peripheral neuropathy in rats produced by acrylamide. Br J Ind Med 1966;23:210 –221. 15. Crofton KM, Padilla S, Tilson HA, et al. The impact of dose rate on the neurotoxicity of acrylamide: the interaction of administered dose, target tissue concentrations, tissue damage, and functional effects. Toxicol Appl Pharmacol 1996;139:163–176. 16. Dixon WJ. Efficient analysis of experimental observations. Annu Rev Pharmacol Toxicol 1980;20:441– 462. 17. Ma J, Smith BP, Smith TL, et al. Juvenile and adult rat neuromuscular junctions: density, distribution, and morphology. Muscle Nerve 2002;26:804 – 809. 18. Keswani SC, Chander B, Hasan C, et al. FK506 is neuroprotective in a model of antiretroviral toxic neuropathy. Ann Neurol 2003;53:57– 64. 19. Keswani SC, Polley M, Pardo CA, et al. Chemokine receptors on Schwann cells mediate HIV-1 gp120 neurotoxicity in primary sensory neurons. Ann Neurol 2003;54:287–296. 20. LoPachin RM Jr, Lehning EJ. Acrylamide-induced distal axon degeneration: a proposed mechanism of action. Neurotoxicology 1994;15:247–259. Keswani et al: EPO Prevents Axonal Degeneration 825 21. DeGrandchamp RL, Lowndes HE. Early degeneration and sprouting at the rat neuromuscular junction following acrylamide administration. Neuropathol Appl Neurobiol 1990;16: 239 –254. 22. Gold BG, Griffin JW, Price DL. Slow axonal transport in acrylamide neuropathy: different abnormalities produced by singledose and continuous administration. J Neurosci 1985;5: 1755–1768. 23. LoPachin RM. Redefining toxic distal axonopathies. Toxicol Lett 2000;112–113:23–33. 24. Keswani SC, Pardo CA, Cherry CL, et al. HIV-associated sensory neuropathies. AIDS 2002;16:2105–2117. 25. Bjartmar C, Trapp BD. Axonal and neuronal degeneration in multiple sclerosis: mechanisms and functional consequences. Curr Opin Neurol 2001;14:271–278. 26. Bjartmar C, Wujek JR, Trapp BD. Axonal loss in the pathology of MS: consequences for understanding the progressive phase of the disease. J Neurol Sci 2003;206:165–171. 27. Digicaylioglu M, Lipton SA. Erythropoietin-mediated neuroprotection involves cross-talk between Jak2 and NF-kappaB signalling cascades. Nature 2001;412:641– 647. 28. Siren AL, Fratelli M, Brines M, et al. Erythropoietin prevents neuronal apoptosis after cerebral ischemia and metabolic stress. Proc Natl Acad Sci USA 2001;98:4044 – 4049. 29. Chong ZZ, Kang JQ, Maiese K. Erythropoietin is a novel vascular protectant through activation of Akt1 and mitochondrial modulation of cysteine proteases. Circulation 2002;106:2973–2979. 30. Gorio A, Gokmen N, Erbayraktar S, et al. Recombinant human erythropoietin counteracts secondary injury and markedly enhances neurological recovery from experimental spinal cord trauma. Proc Natl Acad Sci USA 2002;99:9450 –9455. 31. Campana WM, Myers RR. Exogenous erythropoietin protects against dorsal root ganglion apoptosis and pain following peripheral nerve injury. Eur J Neurosci 2003;18:1497–1506 32. Baciu I, Oprisiu C, Derevenco P, et al. The brain and other sites of erythropoietin production. Rom J Physiol 2000;37:3–14 33. Campana WM, Myers RR. Erythropoietin and erythropoietin receptors in the peripheral nervous system: changes after nerve injury. FASEB J 2001;15:1804 –1806 826 Annals of Neurology Vol 56 No 6 December 2004 34. Dame C, Bartmann P, Wolber E, et al. Erythropoietin gene expression in different areas of the developing human central nervous system. Brain Res Dev Brain Res 2000;125:69 –74. 35. Dame C, Juul SE, Christensen RD. The biology of erythropoietin in the central nervous system and its neurotrophic and neuroprotective potential. Biol Neonate 2001;79:228 –235. 36. Juul SE, Anderson DK, Li Y, Christensen RD. Erythropoietin and erythropoietin receptor in the developing human central nervous system. Pediatr Res 1998;43:40 – 49. 37. Juul SE, Yachnis AT, Rojiani AM, Christensen RD. Immunohistochemical localization of erythropoietin and its receptor in the developing human brain. Pediatr Dev Pathol 1999;2: 148 –158. 38. Li Y, Juul SE, Morris-Wiman JA, et al. Erythropoietin receptors are expressed in the central nervous system of mid-trimester human fetuses. Pediatr Res 1996;40:376 –380. 39. Masuda S, Nagao M, Takahata K, et al. Functional erythropoietin receptor of the cells with neural characteristics. Comparison with receptor properties of erythroid cells. J Biol Chem 1993;268:11208 –11216. 40. Mizusawa I, Abe S, Kanno K, et al. Expression of cytokines, neurotrophins, neurotrophin receptors and NOS mRNA in dorsal root ganglion of a rat tourniquet model. Leg Med (Tokyo) 2003;5:S271–S274. 41. Haga KK, Gregory LJ, Hicks CA, et al. The neuronal nitric oxide synthase inhibitor, TRIM, as a neuroprotective agent: effects in models of cerebral ischaemia using histological and magnetic resonance imaging techniques. Brain Res 2003;993: 42–53. 42. Bolanos JP, Garcia-Nogales P, Almeida A. Provoking neuroprotection by peroxynitrite. Curr Pharm Des 2004;10:867– 877. 43. Wiggins AK, Shen PJ, Gundlach AL. Neuronal-NOS adaptor protein expression after spreading depression: implications for NO production and ischemic tolerance. J Neurochem 2003;87: 1368 –1380. 44. Keilhoff G, Fansa H, Wolf G. Differences in peripheral nerve degeneration/regeneration between wild-type and neuronal nitric oxide synthase knockout mice. J Neurosci Res 2002;68: 432– 441. 45. Goldman SA, Nedergaard M. Erythropoietin strikes a new cord. Nat Med 2002;8:785–787.