Erythropoietin protects the developing brain from hyperoxia-induced cell death and proteome changes.код для вставкиСкачать
Erythropoietin Protects the Developing Brain from Hyperoxia-Induced Cell Death and Proteome Changes Angela M. Kaindl, MD, PhD,1–3 Marco Sifringer, MSc,4 Andrea Koppelstaetter, MSc,5 Kerstin Genz, MD,4 Rebecca Loeber,4 Constanze Boerner,4 Janine Stuwe,5 Joachim Klose, MD, PhD,5 and Ursula Felderhoff-Mueser, MD, PhD4 Objective: Oxygen toxicity has been identified as a risk factor for adverse neurological outcome in survivors of preterm birth. In infant rodent brains, hyperoxia induces disseminated apoptotic neurodegeneration. Because a tissue-protective effect has been observed for recombinant erythropoietin (rEpo), widely used in neonatal medicine for its hematopoietic effect, we examined the effect of rEpo on hyperoxia-induced brain damage. Methods: Six-day-old C57Bl/6 mice or Wistar rats were exposed to hyperoxia (80% O2) or normoxia for 24 hours and received rEpo or normal saline injections intraperitoneally. The amount of degenerating cells in their brains was determined by DeOlmos cupric silver staining. Changes of their brain proteome were determined through two-dimensional electrophoresis and mass spectrometry. Western blot, enzyme activity assays and real-time polymerase chain reaction were performed for further analysis of candidate proteins. Results: Systemic treatment with 20,000 IE/kg rEpo significantly reduced hyperoxia-induced apoptosis and caspase-2, -3, and -8 activity in the brains of infant rodents. In parallel, rEpo inhibited most brain proteome changes observed in infant mice when hyperoxia was applied exclusively. Furthermore, brain proteome changes after a single systemic rEpo treatment point toward a number of mechanisms through which rEpo may generate its protective effect against oxygen toxicity. These include reduction of oxidative stress and restoration of hyperoxia-induced increased levels of proapoptotic factors, as well as decreased levels of neurotrophins. Interpretation: These findings are highly relevant from a clinical perspective because oxygen administration to neonates is often inevitable, and rEpo may serve as an adjunctive neuroprotective therapy. Ann Neurol 2008;64:523–534 Although the mortality of children with extremely low birth weight and gestational age has decreased markedly because of advances in perinatal care, long-term neurological morbidity continues to impose a major personal burden for affected individuals and their families, and constitutes a considerable socioeconomic problem.1,2 In many cases, however, there is neither an obvious clinical explanation nor a corollary on conventional ultrasound imaging studies for the neurological morbidity. Results of clinical studies suggest that oxygen toxicity may be associated with an adverse neurological outcome in survivors of preterm birth.3 We recently dem- onstrated that increased oxygen tension is a powerful trigger for widespread apoptotic cell death in the developing rodent brain.4 –7 The central nervous system (CNS) of preterm infants is particularly sensitive to free radical–mediated oxidative stress.8,9 These newborns not only exhibit developmental immaturity of their free radical defenses, but they are also inevitably exposed to relative hyperoxia compared with intrauterine hypoxic conditions and are more likely to encounter further situations of increased oxidative stress such as oxygen supplementation or systemic infections. Experimental evidence suggests that hyperoxia-induced apoptosis is associated with oxidative stress, up- Additional Supporting Information may be found in the online version of this article. From the 1Department of Pediatric Neurology, Charité–Universitätsmedizin Berlin, Campus Virchow-Klinikum, Berlin, Germany; 2 Institut National de la Santé et de la Recherche Médicale, U676; 3 Université Paris 7, Faculté de Médecine Denis Diderot, IFR02 and IFR25, Paris, France; 4Department of Neonatology, Charité–Universitätsmedizin Berlin; and 5Institute of Human Genetics, Charité– Universitätsmedizin, Berlin, Germany. Received April 16, 2008, and in revised form Jun 11. Accepted for publication Jun 20, 2008. Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/ana.21471 Potential conflict of interest: Nothing to report. Address correspondence to Dr Kaindl, Department of Pediatric Neurology, Charité, Campus Virchow-Klinikum, Universitätsmedizin Berlin, Augustenburger Platz 1, 13353 Berlin, Germany. E-mail: email@example.com © 2008 American Neurological Association Published by Wiley-Liss, Inc., through Wiley Subscription Services 523 regulation of proinflammatory cytokines, decreased expression of neurotrophins, decreased activation of neurotrophin-regulated pathways, and changes in brain proteins associated with cell maintenance and growth, neuronal migration/morphology, and synaptic activity.4,5,7,10 The timing of greatest vulnerability coincides with the peak of the brain growth spurt, which starts at midpregnancy in humans and extends well into the third postnatal year. In mice and rats, this developmental period occurs within the first 3 postnatal weeks.11 Although our findings and those of others caution the use of oxygen, its administration cannot always be avoided in neonatal intensive care regardless of the dangers this may bear for the developing brain. Thus, the search for adjunctive neuroprotective measures that can prevent or ameliorate the toxicity of oxygen for the developing brain is highly warranted. Erythropoietin (Epo) has a long track record of use in preterm infants to prevent anemia of prematurity and has been approved by the US Food and Drug Administration for this clinical use in its recombinant form (rEpo). Erythropoiesis was considered originally to be the sole physiological action of Epo. This premise was changed through the knowledge that Epo and its receptor are expressed in several organs including the CNS and the subsequent discovery of its neuroprotective properties.12,13 Homozygous Epo receptor gene knock-out mice have abnormal heart and brain development with extensive neuronal apoptosis.14 We hypothesized that systemic rEpo treatment can protect the immature brain from hyperoxia-induced damage. In this study, we demonstrate a neuroprotective effect of rEpo in hyperoxia-induced brain damage and identify molecular mechanisms potentially involved in this process. Materials and Methods Animal Experiments Animal experiments were performed according to institutional guidelines of the Charité–Universitätsmedizin Berlin (Berlin, Germany). Six-day-old C57Bl/6 mice (Charles River, Sulzfeld, Germany) or Wistar rats (Charité) were separated into four groups and treated as follows: (1) normoxia (21% O2) and normal saline intraperitoneal (IP) injections, (2) normoxia and 20,000 IE/kg rEpo (Recormon; Boehringer-La Roche, Grenzach, Germany) IP injections, (3) hyperoxia (80% O2)4,7 and normal saline IP injections, and (4) hyperoxia and 20,000 IE/kg rEpo IP injections. For the analysis of the antiapoptotic effect of rEpo with DeOlmos cupric silver staining, a fifth group of animals was treated with hyperoxia (24 hours) and 10,000 IE/kg rEpo IP injections. For hyperoxia or normoxia exposure, pups were kept with their dam. In all study groups, rEpo or normal saline was injected at the beginning of an oxygen/room air exposure. 524 Annals of Neurology Vol 64 No 5 November 2008 Tissue Preparation After treatment, mice were killed with chloral hydrate 1 gm/kg IP, perfused transcardially with normal saline solution, and decapitated. Brains were removed, the cerebellum discarded, the hemispheres divided sagittally into left and right halves, and snap-frozen in liquid nitrogen. For optimal reproducibility, brain protein extracts from each of the sexmatched sample pairs of mice were processed together throughout the two-dimensional electrophoresis (2-DE) procedure (n ⫽ 8 per group; female/male ratio, 1:1). For real-time polymerase chain reaction (PCR), postnatal day 6 (P6) Wistar rat pups (n ⫽ 6 – 8 per group) were subjected to treatments as described earlier and killed 2, 6, 12, 24, and 48 hours after treatment initiation. Animals were decapitated, the olfactory bulb and the cerebellum removed, and brain halves snap-frozen in liquid nitrogen. For histological analysis, P6 Wistar rat pups (n ⫽ 6 per group) were treated as specified earlier, anesthetized with chloral hydrate 1 gm/kg IP, transcardially perfused with heparinized 0.01 M phosphate-buffered saline at pH 7.4, and subsequently with 4% paraformaldehyde in cacodylate buffer at pH 7.4. After a postfixation time of 3 days at 4° C, brains were embedded in paraffin and processed for DeOlmos cupric silver staining. Histology and Quantification of Cell Death in Different Brain Regions The amount of degenerating cells was determined in DeOlmos cupric silver stained 70 m coronal brain sections of hyperoxia, normoxia, normal saline, and/or rEpo-treated infant rats, as described previously.15 Degenerating cells were identified by their distinct dark appearance caused by the silver impregnation. A total apoptotic score from cortex, caudate nucleus, nucleus accumbens, corpus callosum and adjacent white matter, thalamus, hippocampus, and hypothalamus was determined by means of a stereological dissector (8 –10 dissectors per brain region).16 Mean numerical densities of degenerating cells (cells/mm3) were determined. An unbiased counting frame (0.05 ⫻ 0.05 mm; dissector height, 0.07 or 0.012 mm) and a high aperture objective were used for the sampling. Protein Extraction Procedure and Two-Dimensional Gel Electrophoresis Total protein extracts were prepared (see Supplementary Methods) and separated by large-gel 2-DE as described previously in detail.17 Proteins were visualized in sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) gels by high-sensitivity silver staining.18 The gel format was 40 cm (isoelectric focusing) ⫻ 30 cm (SDS-PAGE) ⫻ 0.75 mm (width). Initially, 2-DE gels were evaluated visually by a trained observer on a light box (Biotec-Fischer, Reiskirchen, Germany). On a 2-DE gel, one protein may be represented by one spot or constitute a pattern of multiple spots (isospots) caused by cotranslational and/or posttranslational modifications of the primary protein product or by protein processing. Spot changes between treatment groups were considered with respect to presence/absence, quantitative variation, and altered mobility. Such modifications may re- sult from alterations of the isoelectric point, molecular weight, and/or conformation of a protein. The latter can cause a shift in the position of a spot on a 2-DE gel, and thus a change of spot intensity, that is, a decrease in the relative concentration of an unmodified protein. Only quantitative changes greater than 10% were considered. Mobility variants are spots that “move” to a different position in the 2-DE gel indicating a shift of isoelectric point, molecular weight, or both. Small differences between experimental groups may remain undetected on account of the limited dynamic range of silver staining. Protein spots found to be reproducibly altered in at least six of eight protein patterns of mice subjected to hyperoxia and/or treatment with rEpo when compared with those of control animals were further evaluated with the Proteomweaver imaging software version 22.214.171.124 (Definiens, Munich, Germany). (Chemicon) according to the manufacturer’s instructions. In short, proteins were derivatized, separated on SDSPAGE gels, and Western-blotted, and oxidatively damaged proteins were identified using an antibody specific to the dinitrophenylhydrazine-derivatized residues. Derivatization controls were performed for each sample. Blotted 2-DE gels were stained with high-sensitivity silver staining. Statistical Analyses Values are presented as mean ⫾ standard error of the mean. Comparisons among groups were made using one-way analysis of variance with Newman–Keul’s post hoc test or unpaired Student’s t test, as appropriate. Results In this study, we analyzed the effect of rEpo on hyperoxia-induced brain damage in infant rodents on P6. Protein Identification For protein identification by mass spectrometry (MS), approximately 800 g protein extract was separated on 1.5 mm diameter isoelectric focusing and 1.0 mm-thick SDS-PAGE gels. Resulting 2-DE gels were stained by an MS-compatible silver staining protocol, and protein spots of interest were excised from 2-DE gels for further MS analysis.19 (See Supplementary Methods for detail on protein identification through MS.) Sex Determination PCR detection of Y chromosome allows rapid and accurate identification of male DNA. Sex-determining region protein gene (Sry) is encoded by Y chromosome and well conserved among species.20 Genomic DNA was prepared and genotyped for Sry using the published method of Lambert and colleagues.21 Real-Time Polymerase Chain Reaction and Western Blots Real-time PCR was performed for brain-derived neurotrophic factor (BDNF), collapsing response mediator protein (Crmp) 2 and 4 messenger RNA (mRNA). Western blots were performed for caspase-2, -3, -8, BDNF, extracellular signal-regulated kinases 1 and 2 (ERK1/2), phospho-ERK1/2, Akt, phospho-Akt, Crmp2, Crmp4, and glyoxalase 1 (Glo1). (For detailed protocols, see the Supplementary Methods.) Measurements of Caspase-2 and -8 Enzyme Activity Caspase-2 and -8 enzyme activity was measured using fluorometric caspase-2 and -8 activity assay kits (caspase-2; Chemicon, Planegg-Munich, Germany; caspase-8; Sigma, St. Louis, MO) according to the manufacturer’s protocols (see Supplementary Methods). Protein Carbonyl Assay Brain protein extracts were extracted as described earlier from brains of mice subjected to normoxia and normal saline IP, hyperoxia and rEpo, or hyperoxia and normal saline IP injections. Protein samples (approximately 160 g) were separated by SDS-PAGE and assayed for protein carbonyls using OxyBlot Protein Oxidation Detection Kit Erythropoietin Decreases Hyperoxia-Induced Apoptosis in Infant Brains Exposure of P6 rodents to a high inspiratory oxygen concentration of 80% (hyperoxia) over a period of 24 hours increased the rate of cell death in the developing rat brain acutely as shown by DeOlmos silver staining (Fig 1). A significant decrease of apoptotic cell death was detected when rats were treated with rEpo 10,000 or 20,000 IE/kg IP before hyperoxia (see Fig 1). Brain Proteome Changes after Hyperoxia and Erythropoietin Treatment We have reported previously proteome changes in mice subjected to hyperoxia or normoxia at P6 without drug treatment.7 To identify proteome changes caused by hyperoxia in infant mouse brains (P6) in the presence of rEpo, we compared brain proteomes of hyperoxiaand rEpo-treated mice with those of control mice (normoxia, normal saline injections). Thereby, we detected reproducible qualitative differences in 25 protein spots on corresponding 2-DE gels (Fig 2). Changes in spot intensities usually varied in the range of 10 to 40% increase or decrease, and seven mobility variants were detected. MS enabled the identification of 24 discrete proteins (see Fig 2B; see Supplemental Table 1). Proteins altered after hyperoxia-exposure in the presence of rEpo have been associated with processes such as oxidative stress, apoptosis, as well as cell maintenance and growth. rEpo inhibited most brain proteome changes observed in infant C57BL/6 mice when hyperoxia was applied exclusively. Brain Proteome Changes after a Single Treatment with Erythropoietin Pursuing the question by what mechanisms rEpo ameliorates the apoptotic component of hyperoxia-induced brain damage, we compared brain proteomes of rEpotreated mice with those of untreated littermates (nor- Kaindl et al: rEpo and O2-Induced Brain Damage 525 Fig 1. Erythropoietin reduces hyperoxia-induced apoptosis in infant rodent brains. (A–H) Light micrographs showing DeOlmos cupric silver–stained cells in cortex (A–D) and thalamus laterodorsalis (E–H). Degenerating cells can be identified by their distinct dark appearance caused by the silver impregnation. In control animals, only few positive profiles were found (A, E), whereas massive apoptotic neurodegeneration was seen at 24 hours after initiation of hyperoxia (B, F). Concomitant recombinant erythropoietin (rEpo) treatment reduced the number of apoptotic cells after hyperoxia (D, H). Exclusive rEpo treatment (in normoxic condition) did not have a significant effect on cell death (C, G). (I) Effect of concomitant rEpo treatment on hyperoxia-induced apoptosis. Cumulative apoptotic scores (see Materials and Methods; means ⫾ standard error of the mean; n ⫽ 7–12 per group) are presented for rats exposed to hyperoxia with or without coadministration of rEpo, and for control animals kept at room air and injected with normal saline. rEpo diminished the apoptotic effect of hyperoxia. moxia, normal saline). We thereby detected reproducible qualitative differences in 37 protein spots (Fig 3). Changes in spot intensities usually varied in the range of 10 to 40% increase or decrease, and five mobility variants were detected. MS enabled the identification of 36 discrete proteins (see Fig 3B; see Supplemental Table 1). Proteins altered after rEpo treatment have been associated with processes such as oxidative stress, inflammation, cell maintenance and growth, and neuronal circuit formation. Erythropoietin Reduces Hyperoxia-Induced Production of Caspase-2, -3, and -8 To determine whether pretreatment with rEpo reduces the hyperoxia-induced increase of caspase-2, -3, and -8 protein levels and enzyme activities at P6 in parallel to the effect on cell death, we analyzed these treatments with hyperoxia and rEpo, hyperoxia and normal saline, and normoxia, respectively. Hyperoxia during infancy not only led to apoptosis but also increased the production of caspase-2, -3, and -8, as well as the enzyme activity of caspase-2 and -8 (Figs 4A–F). There were no changes in the protein levels of ␤-actin that served as an internal control. The protein levels of all three caspases increased within 12 hours and were still increased 48 hours after treatment initiation (see Figs 4A–D). Caspase-2 and -8 enzyme activities are increased similarly at 6 and 12 hours after treatment initiation, respectively (see Figs 4E, F). rEpo treatment reduced the hyperoxia-induced increase of caspase pro- 526 Annals of Neurology Vol 64 No 5 November 2008 tein levels and enzyme activities (see Figs 4A–F), whereas exclusive rEpo treatment did not affect them significantly (see Figs 4A–C, E, F). Erythropoietin Reduces Hyperoxia-Induced Oxidative Stress Epo treatment induces quantitative and qualitative changes of several oxidative stress-associated proteins such as peroxiredoxin isoforms, Glo1, and latexin (see Fig 3B; see Supplemental Table 1). To confirm whether the observed reduction of apoptotic cell death may be associated with a reduction of oxidative stress, we examined infant rodent brains after hyperoxia, normoxia, rEpo, and/or normal saline treatment for protein carbonyl levels (a general marker of oxidative stress). Hyperoxia leads to a significant increase of protein levels, as reported previously by our group,7 and concomitant rEpo-treatment reduces these levels significantly (Fig 5A). Derivatization controls were performed and did not show a signal (data not shown). Furthermore, proteins involved in oxidative stress and apoptosis changed on hyperoxia7 (14-3-3 isoforms, protein phosphatase 2A inhibitor 2I, phosphoprotein enriched in astrocytes, programmed cell death 5, and annexin A5) were not dysregulated when Epo was applied before exposure to high oxygen levels. Glo1, as part of a glutathione-dependent glyoxalase enzyme system that protects cells from advanced glycation end-product formation,22–25 was dysregulated on exposure to rEpo (see Fig 2; see Supplemental Table 1). Glo1 showed a change of electrophoretic mobility on cotreatment with hyperoxia and rEpo (see Fig 2B; see Supplemental Table 1). A single systemic rEpo treatment similarly induced a mobility variant of the same Glo1 isospot in the brains of infant mice (see Fig 3; see Supplemental Table 1). To further determine the regulation of Glo1 in this context, we analyzed protein levels through Western blotting. This analysis showed a reduction of the total amount of Glo1 protein after systemic rEpo treatment, but normal Glo1 levels when hyperoxia and rEpo were applied together (see Fig 5B). pAkt, ERK1/2, and pERK1/2 protein levels in the brain after hyperoxia and/or rEpo treatment. Hyperoxia triggered a significant reduction in BDNF mRNA and BDNF protein levels, and this decrease was ameliorated significantly through rEpo administration (see Figs 4G, H). Similarly, pAkt and pERK1/2 protein levels decreased after hyperoxia and were partially recovered through rEpo coadministration (see Figs 4I, J); Akt and ERK1/2 protein levels did not change significantly after hyperoxia or rEpo treatment, or both (data not shown). Exclusive rEpo treatment did not have a significant effect on BDNF, pAkt, and pErk1/2 levels when compared with control animals (see Figs 4G–J). Erythropoietin Partially Reconstitutes Downregulated Neurotrophic Factors after Hyperoxia To explore mechanisms involved in the neuroprotective effect of rEpo coadministration in the hyperoxia model and in view of the dysregulation of growth and energy metabolism-associated proteins in infant mouse brains after a single rEpo treatment (see Fig 2B; see Supplemental Table 1), we analyzed whether rEpo increases the decreased levels of BDNF mRNA and BDNF, Akt, Erythropoietin Does Not Change Hyperoxia-Induced Increase of Crmp Levels Rho GTPase downstream regulator Crmp2 and Crmp4 isoforms were dysregulated on hyperoxia, as reported previously.7 These proteome changes were blocked by a cotreatment with rEpo (see Fig 2; see Supplemental Table 1). A single systemic rEpo treat- Š Fig 2. Brain proteins altered acutely in hyperoxia- and erythropoietin (Epo)-treated infant mice. (A) Left-hemisphere brain proteins (without cerebellum and rhinencephalon) were resolved by two-dimensional electrophoresis (2-DE) according to isoelectric point (pI) in the first and molecular weight (Mw) in the second dimension. Protein spots were demonstrated by silver staining. A representative 2-DE protein pattern with a resolving power of about 6,000 discrete protein spots from a 7-day-old male mouse exposed to normoxia and saline intraperitoneal (IP) injections on postnatal day 6 (P6) is shown. Numbers mark protein spots that were reproducibly altered in response to hyperoxia and 20,000 IE/kg recombinant erythropoietin (rEpo) IP treatment when compared with controls kept at room air and injected with saline solution. (B) Representative 2-DE gel sections displaying altered brain proteins. Most of these proteins are associated with oxidative stress, apoptosis, growth and energy metabolism, and neuronal circuit formation. Altered spots are marked by arrows on their right side. (See Supplemental Table 1 for further information.) Glo1 ⫽ glyoxalase 1; Hspb2 ⫽ heat shock protein 2; Cbr1 ⫽ carbonyl reductase 1; ⫽ Psma7 proteasome subunit alpha type 7; Snev ⫽ nuclear matrix protein SNEV; Uqcrc1 ⫽ ubiquinolcytochrome c reductase complex core protein 1; Rplp2 ⫽ ribosomal protein, large P2; Sfpq ⫽ splicing factor proline/glutamine rich; Vdac3 ⫽ voltage-dependent anion channel 3; Gnb1 ⫽ guanine nucleotide-binding protein, ␤-1 subunit; Phb ⫽ prohibitin; Snx3 ⫽ sorting nexin 3; Gmps ⫽ guanine monophosphate synthetase; Gda ⫽ guanine deaminase; Ddah2 ⫽ dimethylargininase 2; Dctn3 ⫽ dynactin 3; Mlc ⫽ myosin regulatory light chain; Pcpn3 ⫽ procollagen (type III) N-endopeptidase; Smarcd1 ⫽ SWI/SNF-related, matrixassociated, actin-dependent regulator of chromatin, subfamily D, member 1; Pafahb ⫽ MNCb-1930 protein, plateletactivating factor acetylhydrolase IB subunit ␤; Nipsnap3a ⫽ Nipsnap homolog 3A; Upp ⫽ unnamed protein product; LOC241593 ⫽ hypothetical protein LOC241593. Kaindl et al: rEpo and O2-Induced Brain Damage 527 ment induced the increase in concentration of a single Crmp2 isospot (see Fig 3; see Supplemental Table 1). We further analyzed in what way a concomitant rEpo treatment changes total mRNA and protein levels of Crmp2 and Crmp4. Hyperoxia induces an increase of both Crmp2 and Crmp4 mRNA and protein levels when analyzed by real-time PCR or Western blot, respectively, in left brain halves of rats at different intervals up to 48 hours after initiation of hyperoxia (Fig 6). Concomitant or exclusive rEpo treatment did not significantly modulate the increase of Crmp2 and Crmp4 mRNA levels (see Fig 6). of 24 hours increased the rate of cell death in the developing rat brain acutely (see Fig 1). In previous studies, we demonstrated that hyperoxia-induced cell death displays morphological features of physiological apoptotic cell death, occurs in a disseminated fashion throughout the immature rat brain, is age dependent, and affects rodents most severely in the first week of life.4,6 A significant decrease of cell death was detected when rats were treated systemically with 10,000 or 20,000 IE/kg rEpo before an exposure to high oxygen levels (see Fig 1). This effect indicates possible therapeutic implications of rEpo in neonatal medicine as a preventive neuroprotective agent. A neuroprotective effect of rEpo has been reported in experimental rodent models of hypoxia-ischemia, excitotoxicity, and neonatal stroke.12 Moreover, an improvement of long-term neurological outcome of neonatal stroke/hypoxiaischemia has been reported.26,27 Still, tissue-protective properties of rEpo have been more extensively studied in adult models than in developing organisms, and the Discussion In this study, we report that a systemic treatment of neonatal rodents with rEpo provides neuroprotection against hyperoxia-induced apoptosis. Our results of proteome analysis point out potential molecular mechanisms that may entertain this tissue-protective effect of rEpo. Erythropoietin Decreases Hyperoxia-Induced Apoptotic Cell Death in Infant Brains An exposure of infant rodents (P6) to a highinspiratory oxygen concentration of 80% over a period 528 Annals of Neurology Vol 64 No 5 Š November 2008 Fig 3. Brain proteins altered acutely in erythropoietin (Epo)treated infant mice. (A) Representative two-dimensional electrophoresis (2-DE) protein pattern from a 7-day-old male mouse exposed to normoxia and saline intraperitoneal (IP) injections on postnatal day 6 (P6) is shown. Numbers mark protein spots that were reproducibly altered in response to a single 20,000 IE/kg recombinant erythropoietin (rEpo) IP injection when compared with littermates injected with saline solution. (For an explanation of the protein spots, refer to Supplemental Table 1.) (B) Representative 2-DE gel sections displaying altered brain proteins. Most of these proteins are associated with oxidative stress, apoptosis, growth and energy metabolism, and neuronal circuit formation. Altered spots are marked by arrows on their right side (see Supplemental Table 1 for further information). Some proteins such as Stmn1 and hnRNPA3 are represented by several isospots (isoforms) on 2-DE gels. Glo1 ⫽ glyoxalase 1; Lxn ⫽ latexin; Ppp3r1 ⫽ protein phosphatase 3 regulatory subunit B, ␣ isoform; Prdx ⫽ peroxiredoxin; Gmfb ⫽ glia maturation factor-␤; Ftl1 ⫽ ferritin light chain 1; Nmral1 ⫽ Nmra-like family domain containing 1; Rplp2 ⫽ ribosomal protein, large P2; Stmn1 ⫽ stathmin 1; hnRNP ⫽ heterogeneous nuclear ribonucleoprotein; hnRNPA3 ⫽ heterogeneous nuclear ribonucleoprotein A3; Hmg2A ⫽ highmobility group box 3; Eno2 ⫽ gamma enolase; P4ha1 ⫽ prolyl 4-hydroxylase subunit ␣-1; Ckm ⫽ creatine kinase, mitochondrial; Sdha ⫽ succinate dehydrogenase flavoprotein subunit, mitochondrial; hnRNPA0 ⫽ hnRNP A0 isoform 2; Cip29 ⫽ cytokine-induced protein 29kDa; hnRNPA1 ⫽ hnRNP A1, isoform b; Ef1b ⫽ elongation factor 1-␤ homolog; Etfb ⫽ electrontransfer flavoprotein, ␤ polypeptide; Fkbp1a ⫽ FK506 binding protein 1a; Snc ⫽ synuclein; Snca ⫽ synuclein-␣; Sncb ⫽ synuclein-␤; Sh3bgr ⫽ SH3 domain binding glutamic acid–rich protein like; Crmp2 ⫽ collapsin response mediator protein 2; Es1 ⫽ ES1 protein; Rbp ⫽ retinoblastoma-binding protein mRbAp48; Dnajc8 ⫽ DnaJ homolog subfamily C member 8 protein; Daz1 ⫽ DAZ-associated protein 1; Spag7 ⫽ spermassociated antigen 7; Isoc1 ⫽ isochorismatase domain containing 1; Upp ⫽ unnamed protein product. Fig 4. Erythropoietin (EPO) treatment partially prevents hyperoxia-induced increase of caspases and decrease of neurotrophins. (A–D) Increased protein levels of active caspase-2, -3, and -8, as detected by Western blot, was evident in total brain extracts 24 hours after the initiation of hyperoxia when compared with normoxic animals. These levels were decreased through systemic recombinant erythropoietin (rEpo) pretreatment (20,000 IE/kg). A representative Western blot series and results of densitometric Western blot quantification are shown. Values represent mean normalized ratios of the caspase-2, -3, and -8 bands to ␤-actin as an internal standard (n ⫽ 6). (E, F) Increased caspase-2 and -8 enzyme activity, as detected by corresponding activity assays, was evident in total brain extracts 12 to 24 hours after the initiation of hyperoxia when compared with normoxic animals. Systemic rEpo treatment decreased these levels (n ⫽ 3– 4). (G–J) Decreased brain-derived neurotrophic factor (BDNF) messenger RNA (mRNA) and BDNF, pAkt, and phosphoextracellular signal-regulated kinases 1 and 2 (pERK1/2) protein levels, as detected by real-time polymerase chain reaction (PCR) or Western blot, was evident in total brain extracts after the initiation of hyperoxia when compared with normoxic animals. These levels were increased through systemic rEpo treatment (20,000 IE/kg). Results of densitometric Western blot or real-time PCR quantification are shown (n ⫽ 5– 6). Student’s t test: comparison of hyperoxia with control (normoxia). *p ⬍ 0.1, **p ⬍ 0.01, ***p ⬍ 0.001, comparison of hyperoxia with hyperoxia ⫹ rEpo. ##p ⬍ 0.01, ###p ⬍ 0.001. Solid black lines represent hyperoxia; dashed lines represent hyperoxia plus EPO; gray lines represent normoxia plus EPO. Kaindl et al: rEpo and O2-Induced Brain Damage 529 precise mechanisms of Epo action in the immature CNS, which might differ from those in the adult CNS,28 have not been fully defined. Potential effects of rEpo include an induction of antiapoptotic signaling pathways,29 –31 a decrease of inflammation,31 and a decrease of excitotoxicity.32,33 Erythropoietin Modulates Hyperoxia-Induced Proteome Changes in Infant Brains Pursuing the question by what mechanisms rEpo ameliorates the apoptotic component of hyperoxia-induced brain damage, we compared brain proteomes in 2-DE gels of infant mice treated concomitantly with hyperoxia and rEpo with those of controls (normoxia, normal saline). Through this approach, we detected reproducible qualitative and quantitative differences in 24 proteins (see Fig 2; see Supplemental Table 1). We demonstrated previously that apoptotic cell death after hyperoxia is associated with a dysregulation of brain proteins in infant mice that have been associated with oxidative stress, apoptosis, cell maintenance and growth, and neuronal circuit formation.7 Consistent with our histological results, systemic rEpo treatment before the initiation of hyperoxia inhibited the occurrence of these brain protein changes. Only myosin regulatory light chain was dysregulated after hyperoxia regardless of a cotreatment with rEpo (see Fig 2B; see Supplemental Table 1). Potential Mechanisms of Erythropoietin Action in Amelioration of Hyperoxia-Induced Brain Damage DECREASE OF EFFECTOR CASPASE PRODUCTION. Fig 5. Erythropoietin (Epo) treatment reduces hyperoxiainduced oxidative stress. (A) Analysis of equal amounts of brain lysates from 24-hour hyperoxia-treated and control mice with and without recombinant erythropoietin (rEpo) treatment of 20,000IE/kg intraperitoneally (IP) after sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) shows increased levels of protein carbonyls after hyperoxia and a reduction of these levels through a cotreatment with rEpo (n ⫽ 4 for each condition). (B) Both combined hyperoxia and rEpo treatment and single rEpo treatment led to a dysregulation of glyoxalase 1 (Glo1) when total protein extracts were analyzed by two-dimensional electrophoresis (2-DE) coupled with mass spectrometry (MS; see Supplemental Table 1). Further analysis of Glo1 protein levels through Western blot analysis showed a reduction of total Glo1 protein levels 12, 24, and 48 hours after a systemic treatment with rEpo. This Glo1 decrease could not be detected when hyperoxia was administered simultaneously. Solid black line represents hyperoxia; dashed line represents hyperoxia plus EPO; gray line represents normoxia plus EPO. *** ⫽ p ⬍ 0.001. 530 Annals of Neurology Vol 64 No 5 November 2008 A systemic concomitant rEpo treatment blocked the dysregulation of apoptosis-associated brain proteins after hyperoxia (see Fig 2B; see Supplemental Table 1). To determine whether the decrease in hyperoxia-induced apoptotic cell death after rEpo treatment is associated with a decreased activation of intrinsic and extrinsic apoptosis/caspase pathways, we analyzed activated caspase-2, -3, and -8 protein levels, as well as caspase-2 and -8 enzyme activities, in rats subjected to hyperoxia and/or rEpo treatment at P6. Hyperoxia during infancy increased the production of caspase-2, -3, and -8 protein levels and enzyme activities, and this effect was reduced by a treatment with rEpo (see Figs 4A–F). In line with our data, rEpo has been demonstrated to prevent inflammatory cell demise through pathways that involve phosphatidylserine exposure, microglial activation, protein kinase B (Akt), and the regulation of caspases.34 ANTIOXIDANT DEFENSE. Proteins that can be linked to rEpo counter steering of an increased production of reactive oxygen species in the developing brain were modulated on combined rEpo and hyperoxia exposure, as well as on single rEpo Fig 6. Erythropoietin (Epo) does not change hyperoxia-induced increase of total collapsing response mediator protein (Crmp) messenger RNA (mRNA) and protein levels. Postnatal day 6 (P6) Wistar rats were subjected to hyperoxia (80%) or normoxia treatment and recombinant erythropoietin (rEpo) or saline IP injections. Densitometric Western blot quantification and real-time polymerase chain reaction (PCR) analysis demonstrate a strong, hyperoxia-induced increase of Crmp2 and Crmp4 protein (A) and mRNA (B) levels (n ⫽ 4 – 6). rEpo treatment led to a dysregulation of one Crmp2 isospot (see Fig 2); however, the total amount of Crmp2 and Crmp4 protein and mRNA does not change significantly (A, B) when rEpo is injected. *p ⬍ 0.1, ***p ⬍ 0.001 compared with control (normoxia), Student’s t test. Solid black lines represent hyperoxia; dashed lines represent hyperoxia plus EPO; gray lines represent normoxia plus EPO. treatment. We have reported increased levels of oxidized glutathione and protein carbonyls in the brains of infant rodents exposed to hyperoxia compared with those kept at room air, findings that are consistent with oxidative stress.4,7 Here, we demonstrate that protein carbonyl levels are decreased by concomitant systemic rEpo treatment (see Fig 5A). Our finding that rEpo decreases hyperoxia-induced oxidative stress is supported by reports of other groups: (1) rEpo reduces NO-mediated formation of free radicals or antagonizes their toxicity through, for example, an increase in the activity of antioxidant enzymes in neurons35; (2) rEpo decreases increased lipid peroxidation levels in fetal rat ischemia-reperfusion and hypoxic-ischemic brain injury models36,37; (3) rEpo increases glutathione peroxidase enzyme activity and decreases nitric oxide overproduc- Kaindl et al: rEpo and O2-Induced Brain Damage 531 tion in an hypoxic-ischemic model36; (4) rEpo stimulates glutathione peroxidase production in astrocyte cultures38; (5) rEpo protects microglia from oxidative stress–induced cell death39; and (6) rEpo improves the survivability of manganese superoxide dismutase double knock-out (Sod2⫺/⫺) mice astrocytes in vitro.40 Such properties may be relevant in the therapeutic prevention of hyperoxia injury to the developing brain of premature infants, in which antioxidant systems are immature. Brain proteins altered on systemic rEpo treatment indicate molecular mechanisms potentially involved in a reduction of oxidative stress. When rEpo was applied before hyperoxia treatment, we detected a dysregulation of Glo1, heat shock protein 2, and carbonyl reductase 1 isoproteins (see Fig 2B; see Supplemental Table 1). Exclusive rEpo treatment similarly led to a dysregulation of Glo1 and Prdx2, but also induced a change in the proteins latexin, protein phosphatase 3 regulatory subunit B, ␣ isoform, ferritin light chain, and glia maturation factor-␤ (see Fig 3; see Supplemental Table 1). Glo1 modulation may be part of rEpo antioxidant defense mechanism because its expression is partially regulated by oxidative stress,23 it is part of a glutathione-dependent glyoxalase enzyme system that protects cells from advanced glycation end-product formation,22–25 and it is overexpressed in several apoptosis-resistant tumor cells.41 Glo1 gene expression is also regulated by mitogen-activated protein kinases under stress conditions42 and by tumor necrosis factor␣.40 On the other hand, glycation end products can activate multiple signaling pathways including Erk1/2 mitogen-activated protein kinases, Rho GTPases, phosphoinositol-3 kinase, the Jak/Stat pathway, as well as downstream effectors.22,25 NEUROTROPHIN RESCUE. A potential mechanism of rEpo action in the amelioration of hyperoxia-induced damage of the immature brain is the recovery of decreased neurotrophic factors, because several proteins that were modulated by rEpo have been associated with cell maintenance and growth and may influence neurotrophins. Neurotrophins provide trophic support to developing neurons, and their withdrawal may lead to neuronal death.43 We detected a decreased synthesis of neurotrophic factors BDNF and GDNF, as well as reduced levels of the active phosphorylated forms of ERK1/2 and Akt (see Figs 4G–J). ERK1/2 and Akt are key players in two major survival-promoting pathways, the Mek-ERK1/2 and the phosphatidylinositol-3 kinase-Akt pathways, which are activated by tyrosine kinase receptors on binding of growth factors. Ras activation results from binding of growth factors to the respective receptors, and initiated signaling via the MEK and phosphatidylinositol-3 pathways. rEpo application partially counteracted the 532 Annals of Neurology Vol 64 No 5 November 2008 effects of hyperoxia on brain BDNF mRNA expression/protein levels and restored levels of phosphorylated ERK1/2 and Akt in the brain (see Figs 4G–J). These results indicate a cross talk between rEpomediated signaling and intracellular signaling mediated by neurotrophins, a mechanism that may account for its protective action. In this regard, rEpo has been shown to exert cytoprotective effects in a variety of experimental models.44 –50 rEpo inhibits apoptosis of cultured neurons deprived of growth factors or exposed to kainic acid, and of endothelial cells by activating MEK-ERK1/2 and phosphatidylinositol-3 kinase-Akt pathways.51 Also, rEpo is able to induce the expression of neuroprotective genes via nuclear factor-B.52 Can Erythropoietin Treatment Also Have Negative Effects on the Developing Brain? To assess the impact of erythropoietin on infant brain proteins, and thus distinguish this from effects caused by a concomitant rEpo and hyperoxia treatment, we assessed brain protein changes after a single systemic rEpo treatment of infant mice. The 36 brain proteins that were dysregulated after such treatment (see Fig 3) have been associated with oxidative stress, inflammation, cell maintenance and growth, and neuronal circuit formation (see Fig 3B; see Supplemental Table 1). Though this proteome modulation may be beneficial, our results raise the question whether rEpo may also exert negative effects on brain development. It remains to be elucidated, for example, whether the increased Crmp levels (see Fig 6) are beneficial (eg, regeneration of injured neurons) or detrimental (abnormal circuit development). This concern is supported by reports on neurotoxic effects of high-dose rEpo treatment in vivo and in vitro, as well as by the report of a possible amplification of diffuse axonal injury.44,51 Reassuringly, lower doses of rEpo (2,500 and 5,000 U/kg) did not show acute toxicity in neonatal rats or adverse longterm behavioral effects,51,53 and single clinical studies on preterm infants do not report any negative longterm neurodevelopmental effects of erythropoietic doses of rEpo when compared with placebo.54 In Europe, a clinical study initiated by the Swiss Neonatal Network and Follow-up group on the effect of greater doses of rEpo administered to prevent adverse neurological outcome is currently under way. Conclusions Our results demonstrate a protective effect of rEpo on hyperoxia-induced brain damage and suggest mechanisms involved in this process. Our findings are highly relevant from a clinical perspective because oxygen administration to neonates is often inevitable and rEpo is a candidate for adjunctive neuroprotective therapy. However, the experimental evidence presented here and data of others call for caution with an unrestricted use of rEpo in neonatal medicine before further studies on its effect on physiological brain development and also before randomized, controlled clinical trials. This work was supported by the German Ministry for Education and Research (BMBF) within the National Genome Research Network (NGFN) program (project no. FK2 01GR0442), the European Commission (Sixth Framework Program, contract no LSHMCT-2006-036534, www.NEOBRAIN.eu), the Sanitätsrat Dr Emil Alexander Huebner und Gemahlin-Stiftung (project no. T114/ 14880/2005/5m), a Rahel Hirsch scholarship from the Charité– Universitätsmedizin Berlin, and the Sonnenfeld-Stiftung Berlin. References 1. Taylor HG, Minich NM, Klein N, Hack M. Longitudinal outcomes of very low birth weight: neuropsychological findings. J Int Neuropsychol Soc 2004;10:149 –163. 2. Wood NS, Marlow N, Costeloe K, et al. Neurologic and developmental disability after extremely preterm birth. EPICure Study Group. N Engl J Med 2000;343:378 –384. 3. Collins MP, Lorenz JM, Jetton JR, Paneth N. Hypocapnia and other ventilation-related risk factors for cerebral palsy in low birth weight infants. Pediatr Res 2001;50:712–719. 4. Felderhoff-Mueser U, Bittigau P, Sifringer M, et al. Oxygen causes cell death in the developing brain. Neurobiol Dis 2004; 17:273–282. 5. Felderhoff-Mueser U, Sifringer M, Polley O, et al. Caspase-1processed interleukins in hyperoxia-induced cell death in the developing brain. Ann Neurol 2005;57:50 –59. 6. Hoehn T, Felderhoff-Mueser U, Maschewski K, et al. Hyperoxia causes inducible nitric oxide synthase-mediated cellular damage to the immature rat brain. Pediatr Res 2003;54: 179 –184. 7. Kaindl AM, Sifringer M, Zabel C, et al. Acute and long-term proteome changes induced by oxidative stress in the developing brain. Cell Death Differ 2006;13:1097–1109. 8. Gitto E, Reiter RJ, Karbownik M, et al. Causes of oxidative stress in the pre- and perinatal period. Biol Neonate 2002;81: 146 –157. 9. Buonocore G, Perrone S, Bracci R. Free radicals and brain damage in the newborn. Biol Neonate 2001;79:180 –186. 10. Gerstner B, DeSilva TM, Genz K, et al. Hyperoxia causes maturation-dependent cell death in the developing white matter. J Neurosci 2008;28:1236 –1245. 11. Dobbing J. The later growth of the brain and its vulnerability. Pediatrics 1974;53:2– 6. 12. Juul S, Felderhoff-Mueser U. Epo and other hematopoietic factors. Semin Fetal Neonatal Med 2007;12:250 –258. 13. Spandou E, Papadopoulou Z, Soubasi V, et al. Erythropoietin prevents long-term sensorimotor deficits and brain injury following neonatal hypoxia-ischemia in rats. Brain Res 2005;1045: 22–30. 14. Yu X, Lin CS, Costantini F, Noguchi CT. The human erythropoietin receptor gene rescues erythropoiesis and developmental defects in the erythropoietin receptor null mouse. Blood 2001;98:475– 477. 15. DeOlmos JS, Ingram WR. An improved cupric-silver method for impregnation of axonal and terminal degeneration. Brain Res 1971;33:523–529. 16. West MJ, Gundersen HJ. Unbiased stereological estimation of the number of neurons in the human hippocampus. J Comp Neurol 1990;296:1–22. 17. Klose J, Kobalz U. Two-dimensional electrophoresis of proteins: an updated protocol and implications for a functional analysis of the genome. Electrophoresis 1995;16:1034 –1059. 18. Klose J, Nock C, Herrmann M, et al. Genetic analysis of the mouse brain proteome. Nat Genet 2002;30:385–393. 19. Shevchenko A, Wilm M, Vorm O, Mann M. Mass spectrometric sequencing of proteins silver-stained polyacrylamide gels. Anal Chem 1996;68:850 – 858. 20. Gubbay J, Vivian N, Economou A, et al. Inverted repeat structure of the Sry locus in mice. Proc Natl Acad Sci U S A 1992; 89:7953–7957. 21. Lambert JF, Benoit BO, Colvin GA, et al. Quick sex determination of mouse fetuses. J Neurosci Methods 2000;95: 127–132. 22. Basta G, Lazzerini G, Massaro M, et al. Advanced glycation end products activate endothelium through signal-transduction receptor RAGE: a mechanism for amplification of inflammatory responses. Circulation 2002;105:816 – 822. 23. Thornalley PJ. Glyoxalase I—structure, function and a critical role in the enzymatic defence against glycation. Biochem Soc Trans 2003;31:1343–1348. 24. Thornalley PJ. Glutathione-dependent detoxification of alphaoxoaldehydes by the glyoxalase system: involvement in disease mechanisms and antiproliferative activity of glyoxalase I inhibitors. Chem Biol Interact 1998;111-112:137–151. 25. Yan SF, Ramasamy R, Naka Y, Schmidt AM. Glycation, inflammation, and RAGE: a scaffold for the macrovascular complications of diabetes and beyond. Circ Res 2003;93: 1159 –1169. 26. Chang YS, Mu D, Wendland M, et al. Erythropoietin improves functional and histological outcome in neonatal stroke. Pediatr Res 2005;58:106 –111. 27. Demers EJ, McPherson RJ, Juul SE. Erythropoietin protects dopaminergic neurons and improves neurobehavioral outcomes in juvenile rats after neonatal hypoxia-ischemia. Pediatr Res 2005;58:297–301. 28. Sola A, Wen TC, Hamrick SE, Ferriero DM. Potential for protection and repair following injury to the developing brain: a role for erythropoietin? Pediatr Res 2005;57:110R–117R. 29. Digicaylioglu M, Lipton SA. Erythropoietin-mediated neuroprotection involves cross-talk between Jak2 and NF-kappaB signalling cascades. Nature 2001;412:641– 647. 30. Matsushita H, Johnston MV, Lange MS, Wilson MA. Protective effect of erythropoietin in neonatal hypoxic ischemia in mice. Neuroreport 2003;14:1757–1761. 31. Sun Y, Zhou C, Polk P, et al. Mechanisms of erythropoietininduced brain protection in neonatal hypoxia-ischemia rat model. J Cereb Blood Flow Metab 2004;24:259 –270. 32. Kawakami M, Iwasaki S, Sato K, Takahashi M. Erythropoietin inhibits calcium-induced neurotransmitter release from clonal neuronal cells. Biochem Biophys Res Commun 2000;279: 293–297. 33. Keller M, Yang J, Griesmaier E, et al. Erythropoietin is neuroprotective against NMDA-receptor-mediated excitotoxic brain injury in newborn mice. Neurobiol Dis 2006;24:357–366. 34. Maiese K, Li F, Chong ZZ. Erythropoietin in the brain: can the promise to protect be fulfilled? Trends Pharmacol Sci 2004; 25:577–583. 35. Sakanaka M, Wen TC, Matsuda S, et al. In vivo evidence that erythropoietin protects neurons from ischemic damage. Proc Natl Acad Sci U S A 1998;95:4635– 4640. 36. Kumral A, Gonenc S, Acikgoz O, et al. Erythropoietin increases glutathione peroxidase enzyme activity and decreases lipid peroxidation levels in hypoxic-ischemic brain injury in neonatal rats. Biol Neonate 2005;87:15–18. 37. Solaroglu I, Solaroglu A, Kaptanoglu E, et al. Erythropoietin prevents ischemia-reperfusion from inducing oxidative damage in fetal rat brain. Childs Nerv Syst 2003;19:19 –22. Kaindl et al: rEpo and O2-Induced Brain Damage 533 38. Genc S, Akhisaroglu M, Kuralay F, Genc K. Erythropoietin restores glutathione peroxidase activity in 1-methyl-4-phenyl1,2,5,6-tetrahydropyridine-induced neurotoxicity in C57BL mice and stimulates murine astroglial glutathione peroxidase production in vitro. Neurosci Lett 2002;321:73–76. 39. Li F, Chong ZZ, Maiese K. Microglial integrity is maintained by erythropoietin through integration of Akt and its substrates of glycogen synthase kinase-3beta, beta-catenin, and nuclear factor-kappaB. Curr Neurovasc Res 2006;3:187–201. 40. Liu Q, Yu L, Gao J, et al. Cloning, tissue expression pattern and genomic organization of latexin, a human homologue of rat carboxypeptidase A inhibitor. Mol Biol Rep 2000;27:241–246. 41. Inoue Y, Tsujimoto Y, Kimura A. Expression of the glyoxalase I gene of Saccharomyces cerevisiae is regulated by high osmolarity glycerol mitogen-activated protein kinase pathway in osmotic stress response. J Biol Chem 1998;273:2977–2983. 42. Van Herreweghe F, Mao J, Chaplen FW, et al. Tumor necrosis factor-induced modulation of glyoxalase I activities through phosphorylation by PKA results in cell death and is accompanied by the formation of a specific methylglyoxal-derived AGE. Proc Natl Acad Sci U S A 2002;99:949 –954. 43. Huang EJ, Reichardt LF. Neurotrophins: roles in neuronal development and function. Annu Rev Neurosci 2001;24:677–736. 44. Adembri C, Massagrande A, Tani A, et al. Carbamylated erythropoietin is neuroprotective in an experimental model of traumatic brain injury. Crit Care Med 2008;36:975–978. 45. Dzietko M, Felderhoff-Mueser U, Sifringer M, et al. Erythropoietin protects the developing brain against N-methyl-Daspartate receptor antagonist neurotoxicity. Neurobiol Dis 2004;15:177–187. 534 Annals of Neurology Vol 64 No 5 November 2008 46. 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 U S A 2002;99:9450 –9455. 47. Grimm C, Wenzel A, Groszer M, et al. HIF-1-induced erythropoietin in the hypoxic retina protects against light-induced retinal degeneration. Nat Med 2002;8:718 –724. 48. Junk AK, Mammis A, Savitz SI, et al. Erythropoietin administration protects retinal neurons from acute ischemia-reperfusion injury. Proc Natl Acad Sci U S A 2002;99:10659 –10664. 49. Kumral A, Ozer E, Yilmaz O, et al. Neuroprotective effect of erythropoietin on hypoxic-ischemic brain injury in neonatal rats. Biol Neonate 2003;83:224 –228. 50. Li Y, Lu ZY, Ogle M, Wei L. Erythropoietin prevents blood brain barrier damage induced by focal cerebral ischemia in mice. Neurochem Res 2007;32:2132–2141. 51. Weber A, Dzietko M, Berns M, et al. Neuronal damage after moderate hypoxia and erythropoietin. Neurobiol Dis 2005;20: 594 – 600. 52. Tubbs RS, Shoja MM, Jamshidi M, Shokouhi G. Does the neuroprotective agent erythropoietin amplify diffuse axonal injury in its early stages? Med Hypotheses 2007;69:1385–1386. 53. McPherson RJ, Juul SE. Recent trends in erythropoietinmediated neuroprotection. Int J Dev Neurosci 2008;26:103–111. 54. Ohls RK, Ehrenkranz RA, Das A, et al. Neurodevelopmental outcome and growth at 18 to 22 months’ corrected age in extremely low birth weight infants treated with early erythropoietin and iron. Pediatrics 2004;114:1287–1291.