Oxidative Stress Promotes Proliferation and Dedifferentiation of Retina Glial Cells in Vitro

Journal of Neuroscience Research 87:964–977 (2009)

Oxidative Stress Promotes Proliferation and Dedifferentiation of Retina Glial Cells In Vitro

Carolina E. Abrahan, M. Fernanda Insua, Luis E. Politi, O. Lorena German, and Nora P. Rotstein* Instituto de Investigaciones Bioquı´micas de Bahı´a Blanca, Universidad Nacional del Sur—CONICET, Bahı´a Blanca, Buenos Aires, Argentina

Oxidative damage is involved in triggering neuronal Harada et al., 2002), which results in their apoptosis in death in several retinal neurodegenerative diseases. The the rat retina (Tso et al., 1994). Further evidence points recent finding of stem cells in the retina suggests that to the involvement of oxidative stress in photoreceptor both preventing neuronal death and replacing lost neu- apoptosis: antioxidants ameliorate light-induced retinal rons might be useful strategies for treating these dis- degeneration, and an early and sustained increase in in- eases. We have previously shown that oxidative stress tracellular reactive oxygen species accompanies photore- induces apoptosis in cultured retinal neurons. We now ceptor apoptosis in vitro (Carmody et al., 1999). More- investigated the response of Mu¨ ller cells, proposed as over, oxidative stress participates in photoreceptor death retina stem cells, to this damage. Treatment of glial cell in retinal degenerative diseases and in animal models of cultures prepared from rat retinas with the oxidant para- retinitis pigmentosa (Sun and Nathans, 2001; Lohr et al., quat (PQ) did not induce glial cell apoptosis. Instead, 2006). PQ promoted their rapid dedifferentiation and prolifera- Because photoreceptor apoptosis is a hallmark of tion. PQ decreased expression of a marker of differenti- retinal neurodegenerative diseases (Chang et al., 1993; ated glial cells, simultaneously increasing the expres- Portera-Cailliau et al., 1994), uncovering ways to pre- sion of smooth muscle actin, shown to increase with vent this apoptosis has been the most sought-after strat- glial dedifferentiation, the levels of cell-cycle markers, egy for treating these diseases. However, finding sources and the number of glial cells in the cultures. In addition, for replacing dead neurons might provide a powerful, al- glial cells protected neurons in coculture from apoptosis ternative therapeutic strategy. Mu¨ller glial cells might induced by PQ and H2O2. In pure neuronal cultures, PQ contribute to both strategies; they are the major glial cell induced apoptosis of photoreceptors and amacrine type in the retina and play multiple roles in this tissue, neurons, simultaneously decreasing the percentage of such as recycling of neurotransmitters and provision of neurons preserving mitochondrial membrane potential; trophic and metabolic support (Poitry-Yamate et al., coculturing neurons with glial cells completely pre- 1995; Derouiche, 1996; Poitry et al., 2000). Mu¨ller cells vented PQ-induced apoptosis and preserved mitochon- are crucial for photoreceptor survival; many trophic fac- drial potential in both neuronal types. These results tors that prevent photoreceptor death seem to act by demonstrate that oxidative damage activated different responses in Mu¨ ller glial cells; they rapidly dedifferentiated and enhanced their proliferation, concurrently preventing neuronal apoptosis. Glial cells might not only Abbreviations: BrdU, bromodeoxyuridine; CRALBP, cellular retinalde- preserve neuronal survival but also activate their cell hyde-binding protein; DMEM, Dulbecco’s modified Eagle’s medium; cycle in order to provide a pool of new progenitor cells FCS, fetal calf serum; GFAP, glial fibrillary acidic protein; PCNA, prolif- that might eventually be manipulated to preserve retinal erating cell nuclear antigen; PQ, paraquat; SMA, smooth muscle actin. functionality. VC 2008 Wiley-Liss, Inc. Contract grant sponsor: National Research Council of Argentina (CONICET); Contract grant sponsor: FONCyT; Contract grant sponsor: Key words: apoptosis; cell survival; photoreceptor; Universidad Nacional del Sur (to N.R. and L.P.). proliferation; oxidative stress Present address of Fernanda Insua: Instituto Valenciano de Infertilidad, Plaza de la Policıá Local 3, Valencia, Spain. *Correspondence to: Nora P. Rotstein, INIBIBB, CC857, B8000FWB, Because of its relatively high oxygen consumption Bahıá Blanca, Buenos Aires, Argentina. and its constant exposure to light, the retina is extremely E-mail: [email protected] prone to oxidative damage. Photooxidative mechanisms Received 24 July 2008; Accepted 4 August 2008 significantly contribute to photoreceptor death because Published online 14 October 2008 in Wiley InterScience (www. of excessive light exposure (Sun and Nathans, 2001; interscience.wiley.com). DOI: 10.1002/jnr.21903

' 2008 Wiley-Liss, Inc. Oxidative Stress and Glial Proliferation 965 activating glial cells, which in turn release trophic factors gentamycin, 4,6-diamidino-2-phenylindole (DAPI), paraquat to rescue photoreceptors (Harada et al., 2000, 2002; dichloride (methyl viologen, 1,10-dimethyl-4,40-bipyridinium Wahlin et al., 2000, 2001). Apoptosis of Mu¨ller cells dichloride, PQ), the polyclonal antibody against glial fibrillary leads to the apoptotic death of photoreceptors (Dubois- acidic protein (GFAP), fluorescein-conjugated secondary anti- Dauphin et al., 2000). bodies, paraformaldehyde, monoclonal antisyntaxin clone The demonstration that glial cells reenter the cell HPC-1 syntaxin, and anti–smooth muscle actin were from cycle on neurotoxic stimuli of the chick retina and then Sigma (St. Louis, MO). Monoclonal antibodies against prolif- start expressing neuronal transcription factors (Fischer erating cell nuclear antigen (PCNA), secondary antibody, goat and Reh, 2001, 2003) suggests that retina glial cells antimouse IgG-HRP, and goat antirabbit IgG-HRP were might also play a role in the replacement of dying neu- from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Poly- rons. To accomplish this role, injuries to the retina clonal antibody against GFAP was from Dako (Glostrup, should activate a regenerative response in Mu¨ller cells, Denmark). Secondary antibody, Alexa 488–conjugated goat which as a first step requires activation of the cell cycle antimouse, MitoTracker Red CMXRos, Alexa Fluor 488– in these cells in order to provide an adequate number of conjugated annexin V, propidium iodide (PI), terminal deoxy- undifferentiated progenitor cells. These progenitors nucleotidyl transferase, recombinant 5-bromo-20-deoxyuridine might eventually differentiate into neuronal precursors. 50-triphosphate (BrdU), and terminal deoxynucleotidyl trans- We have investigated the specific reaction of retina ferase (TdT) buffer were from Molecular Probes, Invitrogen Mu¨ller glial cells to oxidative damage induced by para- (Argentina). Tyramine was from NEN Life Science Products, quat (PQ) and hydrogen peroxide (H2O2). PQ is an and ABC reagents were from Vector Laboratories (Burlin- anion superoxide generator (Yang and Sun, 1998; Li and game, CA). Hydrogen peroxide (H2O2) 30% was from Baker, Sun, 1999), and its intravitreous injection leads to oxida- Argentina. Anti-BrdU (G3G4) was from Developmental Stud- tive damage and cell death in retinal outer and inner nu- ies Hybridoma Bank (developed under the auspices of the clear layer (Cingolani et al., 2006). H2O2, by itself a re- NICHD and maintained by the University of Iowa, Depart- active oxygen species, has been extensively used to ment of Biological Sciences, Iowa City, IA). Monoclonal induce oxidative damage in neuronal and neuroglial co- Rho4D2 antibody and polyclonal antibody against cellular ret- cultures (Desagher et al., 1996; Hoyt et al, 1997; Liddell inaldehyde-binding protein were generous gifts from Dr. R. et al., 2006). We have previously shown that treatment Molday (University of British Columbia, Canada) and Dr. J. of pure retina neuronal cultures with either PQ or Saari (Washington University, St. Louis, MO), respectively. H2O2 induces the apoptosis of photoreceptor cells (Rot- All other reagents used were analytical grade. stein et al., 2003; Chucair et al., 2007). We have now dissected the response of pure, isolated glial cells to oxidative damage and analyzed the effect of these cells on Retinal Cultures neurons in coculture when subjected to this damage. Glial Cell Cultures and Neuroglial Cocultures. Our findings strongly suggest that one of the first Long-term cultures of Mu¨ller glial cells were prepared from responses of Mu¨ller cells to oxidative stress is their rapid rat retinas following protocols previously described (Politi dedifferentiation and reentry into the cell cycle. In addi- et al., 1996, 2001a; Insua et al., 2008). Briefly, retinas dis- tion, when confronted with the same oxidative injury sected from 2-day-old rats were chemically and mechanically that induces neuronal apoptosis, glial cells not only dissociated with trypsin [250 lL of 0.25% trypsin in 6 mL of remained viable but also prevented the apoptosis of pho- calcium and magnesium-free Hanks (CMF) for 7.5 min]. toreceptor and amacrine cells in coculture. These results Digestion was stopped by using 0.25% trypsin inhibitor, and show that glial cells in vitro react to injury by rapidly the cells were resuspended in DMEM containing 10% FCS activating different pathways that might converge in an and seeded at a density of 2.0 3 106 cells per dish in 35-mm attempt to support retinal functionality. diameter plastic dishes with no previous treatment. To obtain purified cultures of Mu¨ller cells, the culture medium was rou- tinely replaced every 2 days until almost all retinal neurons MATERIALS AND METHODS had detached from the glial cells (fewer than 5% of neuronal cells still remained attached). Under these conditions, the cells, Materials rather than being evenly distributed on the dish surface, One- and 2-day-old albino Wistar rats bred in our own tended to grow in patches that became confluent after 14– colony were used in all the experiments. All proceedings con- 16 days; the cells were then used for the different experimen- cerning animal use were done in accordance with the guide- tal procedures. lines published in the NIH Guide for the Care and Use of Labo- To obtain cocultures of Mu¨ller cells and retinal neurons, ratory Animals. Plastic 35-mm-diameter culture dishes a cell suspension was prepared as described above, resuspended (NUNC) were purchased from GBO (Argentina). Fetal calf in a chemically defined medium (Politi et al., 1996; Rotstein serum (FCS) was from Centro de Virologıá Animal (CEVAN, et al, 1996), seeded on 35-mm diameter dishes without previ- Buenos Aires, Argentina). Dulbecco’s modified Eagle’s me- ous treatment, and incubated for 24 hr. The culture medium dium (DMEM; Gibco) was purchased from Life Technologies was then gently replaced by DMEM containing 10% FCS, (Grand Island, NY). Trypsin, trypsin inhibitor, transferrin, thus allowing neurons to remain attached to and develop on hydrocortisone, putrescine, insulin, polyornithine, selenium, the glial cells.

Journal of Neuroscience Research 966 Abrahan et al.

Pure Retinal Neuronal Cultures. Pure retinal neu- Photoreceptors have a small, round cell body (3–5 lm) ronal cultures were obtained following previously established with a single neurite at one end, which usually ends in a con- procedures (Politi et al., 1996; Rotstein et al, 1996). In brief, spicuous synaptic ‘‘spherule.’’ Sometimes they display a con- after dissection and dissociation of 2-day-old rat retinas, the necting cilium at the opposite end, but they fail to develop cells were resuspended in a chemically defined medium sup- their characteristic outer segments. Opsin is diffusely distrib- plemented with insulin and lacking FCS and photoreceptor uted over the cell body, which is usually darker than that of trophic factors (Politi et al., 1996; Rotstein et al, 1996). About amacrine neurons. To be identified as photoreceptors, the cells 1–1.2 3 106 cells/dish were seeded on 35-mm diameter had to display at least three of these criteria. Amacrine neu- dishes that had been sequentially pretreated with polyornithine rons are larger than photoreceptors (7–20 lm) and have mul- (0.1 mg/mL) and with schwannoma-conditioned medium tiple neurites. Almost all of them show HPC-1 immunoreac- (Adler, 1982) for 8 and 12 hr, respectively. Cultures were tivity starting in early stages of development, which is retained incubated at 368C in a humidified atmosphere of 5% CO2. even after undergoing degenerative changes that alter their morphological appearance (Politi et al., 1996). Induction of Oxidative Damage To analyze the effects of oxidative stress, two oxidants Evaluation of Glial Cell Proliferation were used, paraquat (PQ) and H2O2. For PQ treatment of Different strategies were used to establish whether treat- pure glial cultures, 2 days after becoming confluent these cul- ment with PQ affected the proliferation of glial cells. Two tures were carefully washed twice with the chemically defined markers of cell-cycle progression were evaluated: bromodeox- medium used for neuronal cultures and then cultured for yuridine (BrdU) uptake and the number of cells expressing 24 hr with this medium. They were then treated with PQ, PCNA. Two days after reaching confluence, glial cell cultures using either a 48-lM concentration over 24 hr or a 240-lM were treated with 240 lM PQ for 4 hr. BrdU, an S-phase concentration for 4 hr. marker, was added to the culture medium (at a 10 lM final Glial cells in coculture were incubated for 5 days in concentration) 1 hr after PQ treatment, and cells were fixed DMEM with 10% FCS in order to allow glial cells to de- 4 hr later. Cells were then treated with 1N HCl for 30 min velop. The culture medium was then removed, and the cocul- for DNA denaturation and neutralized with 0.1M boric acid. tures were incubated with the chemically defined medium for BrdU uptake was then determined using a monoclonal anti- another 24 hr. The effect of PQ on both cocultures and pure body against BrdU. neuronal cultures was analyzed as previously described (Rot- The number of cells expressing PCNA, a G1-S cyclin stein et al., 2003), treating 5-day cultures with a 48-lM con- used as a marker for proliferating cells, was detected by immu- centration of PQ for 24 hr. nocytochemical methods using a mouse monoclonal anti- Treatment of 5-day cocultures and pure neuronal cul- PCNA antibody. tures with H2O2 was performed as previously described (Chu- The effect of PQ on the amount of glial cells in the cul- cair et al, 2007), with minor modifications. The cultures were tures was also determined. After reaching confluence, the 10% treated with 100 lMH2O2 for 30 min, then washed and FCS–containing medium was discarded, and glial cell cultures incubated with chemically defined medium for another 3 hr were incubated in the chemically defined medium used for before fixation. neuronal cultures for 24 hr. They were then treated either with 240 lM PQ for 4 hr or with 48 lM PQ for 24 hr. After Immunocytochemical Methods fixation, the number of glial cells in each culture condition was established by counting their nuclei, labeled with DAPI, a Cultures were fixed for at least 1 hr with 2% parafor- fluorescent dye that binds to DNA. Briefly, cells were perme- maldehyde in phosphate-buffered saline (PBS), followed by ated with Triton X-100 in PBS, washed with PBS, and incu- permeation with Triton X-100 (0.1%) for 15 min. In every bated with DAPI for 20 min. culture condition, Mu¨ller glial cells were identified by their morphology and by immunocytochemistry using the polyclonal antibody against cellular retinaldehyde-binding protein Protein Extraction and Western Blotting (CRALBP). The changes in the expression of CRALBP, smooth Neuronal cell types, both in pure neuronal cultures and muscle actin (SMA), and GFAP after PQ treatment were eval- in coculture with glial cells, were identified by their morphol- uated by Western blot. Glial cultures were treated with either ogy using phase-contrast microscopy and by immunocyto- 240 lM PQ for 4 hr or 48 lM PQ for 24 hr. Cells were chemistry using the monoclonal antibodies syntaxin (HPC-1) then rinsed with PBS, collected in lysis buffer [3 mM KCl, 50 and Rho4D2, which selectively react with amacrine and pho- mM Tris HCl (pH 7.4), 150 mM NaCl, 1 mM EDTA, 1% toreceptor neurons, respectively (Barnstable, 1980; Hicks and Tween-20, and 1% NP-40] containing a protease inhibitor Barnstable, 1987; Kljavin et al., 1994), as previously described mixture, and lysed in ice for 20 min. Proteins were separated (Rotstein et al., 1998). Alexa 488–conjugated goat antimouse by one-dimensional SDS-PAGE (Laemmli, 1970). Briefly, was used as the secondary antibody. Tyramide signal amplifi- samples were mixed with 63 Laemmli sample buffer [250 cation was occasionally used to improve visualization, follow- mM Tris-HCl (pH 6.8), 8% SDS, 40% glycerol, 20% 2-mer- ing the procedure described by the manufacturers. Controls captoethanol, and 0.02% bromophenol blue] and heated for for immunocytochemistry were done by omitting either the 5 min at 958C, and proteins (either 40 or 50 lg/sample) were primary or the secondary antibody. subjected to electrophoresis on 10% SDS-polyacrylamide

Journal of Neuroscience Research Oxidative Stress and Glial Proliferation 967 minigels and then transferred to Immobilon P (PVDF) mem- Evaluation of Mitochondrial Membrane Potential branes. The membranes were immersed in PBS buffer con- To assess preservation of mitochondrial membrane taining 5% skim milk for 2 hr at room temperature or potential, cultures were incubated for 20 min before fixation 8 overnight at 4 C, to block nonspecific binding. Anti- with the fluorescent probe MitoTracker Red (0.1 lg/mL), CRALBP, -GFAP, -SMA, and -actin antibodies were allowed which labels mitochondria retaining their membrane potential 8 to react with the membrane overnight at 4 C or 2 hr at room with a bright red fluorescence. temperature. The membranes were then washed twice for 15 min and once for 10 min with Tris-buffered saline with Statistical Analysis 0.1% Tween 20 (TBS-T) and incubated with horseradish per- For cytochemical studies, 10 fields per sample, randomly oxidase–conjugated goat antimouse or antirabbit secondary chosen, were analyzed in each case. Each value represents the antibodies in TBS-T for 1 hr at room temperature. Mem- average 6 SD of at least three experiments, with 3–4 dishes branes were washed 3 times with TBS-T and then visualized for each condition. Statistical significance was determined by using an enhanced chemiluminescent technique (ECL), the Student two-tailed t test. according to the manufacturer’s instructions. Images were obtained by scanning at 600 dpi and printing at 300 dpi. RESULTS Bands were quantified using ImageJ software (freely available at http:/rsb.info.nih.gov/ij; not shown). Effect of Paraquat on Glial Cell Survival In pure glial cultures, Mu¨ller glial cells grew and developed normally for more than 30 days, adopting a Evaluation of Apoptosis flat, irregular morphology, with a size ranging from 30 Apoptosis was determined by three methods: evaluation to 300 lm (Fig. 1A). These cells had large, usually oval of nuclei integrity, terminal deoxynucleotide transferase dUTP nuclei (Fig. 1C) and showed intense perinuclear expres- nick-end labeling method (TUNEL) staining, and annexin V/ sion of CRALBP (Fig. 1B), a marker of differentiated propidium iodide (PI) labeling. Mu¨ller glial cells (Rich et al., 1995; Kennedy et al., Nuclei integrity was evaluated after staining cell nuclei 2003); this expression increased with time of develop- with DAPI, as described above. Glial and neuronal cells were ment in culture, as the cells differentiated as mature glial considered apoptotic when they showed either fragmented or cells (Insua et al., 2008). condensed (pycnotic) nuclei. The number of apoptotic photo- Retinal neurons in culture are very sensitive to receptors or amacrine cells was counted in cultures double- oxidative damage; their treatment with PQ induces apo- labeled with DAPI and either Rho4D2 or HPC-1 in order to ptosis of both photoreceptors and amacrine neurons unambiguously identify cells as either photoreceptors or ama- (Rotstein et al., 2003). To evaluate whether Mu¨ller glial crine neurons, respectively, and thus to establish the total cells in culture were similarly affected by oxidative dam- number of each cell type. The percentage of apoptotic photo- age, pure glial cultures were treated either with 48 lM receptors or amacrine neurons was then calculated. PQ for 24 hr or with 240 lM PQ for 4 hr. Apoptosis For TUNEL staining, cultures treated with either H2O2 of neuronal cells induced by oxidative stress is associated or PQ were fixed on day 5 and day 6, respectively, with 2% with the presence of many annexin V/PI-labeled neu- paraformaldehyde for 15 min and then stored in 70% ethanol rons and with loss of their mitochondrial membrane for 48 hr at 2208C. Before labeling, cells were washed twice potential (Rotstein et al., 2003; Chucair et al, 2007). In with PBS for 5 min each at room temperature. Samples were contrast, almost no annexin V– and PI-labeled glial cells preincubated with 13 TdT buffer for 15 min and then incu- were found in control glial cells after 20 days in culture bated with the TdT reaction mixture (0.05 mM BrdUTP, 0.3 (Fig. 2A–C), and PQ treatment did not increase their U/lL TdT in TdT buffer) at 378C in a humidified atmos- number (Fig. 2D–F). Glial cells preserved their mito- phere for 1 hr. The reaction was stopped by a 15-min incuba- chondrial membrane potential; many brilliantly labeled tion with stop buffer [300 mM NaCl, 30 mM sodium citrate mitochondria were visible in these cells in control con- (pH 7.4)] at room temperature. Negative controls were pre- ditions (Fig. 2G–I) and even after PQ treatment (Fig. pared by omitting TdT. The presence of BrdU was deter- 2J–L). In addition, PQ did not affect the integrity of mined with an anti-BrdU monoclonal antibody according to a nuclei of glial cells (Fig. 2I,L). Almost no TUNEL- standard immunocytochemical technique. labeled glial cells were detected in these cultures, with For annexin V/PI staining, the incubation medium was or without PQ (Fig. 2M–P). As a whole, these results removed, and dishes were washed twice with ice-cold PBS. showed that glial cells preserved their viability and were Cells were then incubated with a 1:4 dilution of annexin V far more resistant than retinal neurons in vitro to a simi- conjugate in annexin-binding buffer [10 mM HEPES, 140 lar oxidative injury. mM NaCl, and 2.5 mM CaCl2 (pH 7.4)] at room temperature in the dark for 15 min. PI was added immediately after, Effect of Oxidative Stress on Dedifferentiation and cells were incubated for another 15 min in the same con- of Glial Cells ditions. Cells were then washed twice in cold PBS, fixed in Although oxidative stress did not affect the viability 1% PF in annexin-binding buffer for 1 hr, and washed in cold of glial cells, it did affect glial cell differentiation. In con- PBS. Labeling with annexin V, PI, or both was then analyzed trol conditions, most glial cells were already differenti- by fluorescence microscopy. ated and expressed CRALBP in confluent cultures (Insua

Journal of Neuroscience Research 968 Abrahan et al.

Fig. 1. Expression of CRALBP in glial cells in culture. Phase (A) Fig. 2. Preservation of viability of glial cells after paraquat treatment. and fluorescent (B, C) micrographs of 15-day pure Mu¨ller glial cells Phase (A, D, G, J, M, O) and fluorescent micrographs showing showing perinuclear labeling with CRALBP (B) and nuclei labeled annexin V (B, E) and propidium iodide (PI; C, F) labeling, mito- with DAPI (C). Scale bar 5 30 lm. chondria labeled with MitoTracker (H, K), nuclei labeled with DAPI (I, L), and TUNEL labeling (N, P) in 20-day cultures of Mu¨l- ler glial cells treated without (A–C, G–I, M, N) or with (D–F, J–L, l l O, P)48 M paraquat (PQ) for 24 hr. Note that in either condition, et al., 2008; Fig. 3A–C). Treatment with 48 M PQ for glial cells showed brightly labeled mitochondria and intact nuclei, 24 hr markedly decreased CRALBP expression, which and almost no glial cells showed annexin V, PI, or TUNEL staining was reduced in most cells and completely disappeared in (B, C, E, F, M–P, arrows). Small, bright nuclei in I, L and some (Fig. 3D–F). Analysis of CRALBP expression by TUNEL-labeled small nuclei in N, P (arrows) belong to neuronal Western blot indicated that it was slightly diminished cells still remaining in the glial cultures. Scale bar 5 15 lm. with 24 lM PQ [Fig. 3G(1,2)] and distinctly decreased with 48 lM PQ for 24 hr [Fig. 3G(1,3)]. Dedifferentiation of glial cells was formerly linked cultures [Fig. 3H(1)] and was significantly increased after to the expression of abnormal markers such as SMA and a 24-hr treatment with 48 lM PQ [Fig. 3H(2)]. Con- gradual loss of the expression of GFAP (Guidry et al., versely, PQ treatment decreased GFAP expression, com- 1996; McGillem et al., 1998). Western blots showed pared with that in control, untreated cultures [Fig. that expression of SMA was very low in control glial 3I(1,2)]. Hence, oxidative stress induced changes in pro-

Journal of Neuroscience Research Oxidative Stress and Glial Proliferation 969

Fig. 3. Effect of PQ on expression of CRALBP, SMA, and GFAP in determined by Western blot. After PQ treatment, cells were lysed, glial cells. Confluent cultures of pure Mu¨ller glial cells were treated and total protein extracts (50 lg/lane in G,40lg/lane in H, I) were without [A–C, G(1), H(1), I(1)] or with [D–F, G(3), H(2), I(2)] 48 analyzed using anti-CRALBP, -SMA and -GFAP. After stripping, lM PQ or with 24 lMPQ(G2) for 24 hr. Phase (A, D) and fluo- the loading of equal amounts of protein was verified by reprobing rescent micrographs (B, C, E, F) show glial cells labeled with an the same blots with antiactin antibody. Western blots are representa- antibody against CRALBP (B, E) and their nuclei labeled with tive of three independent experiments with similar results. Note that DAPI (C, F). Scale bar 5 20 lm. Expression of CRALBP [G(1–3)], 48 lM PQ markedly decreased the expression of CRALBP [B, E, smooth muscle actin [SMA; H(1, 2)], GFAP [I(1, 2)], and actin were G(1,3)] and GFAP (I), whereas it increased that of SMA (H). tein expression in glial cells similar to those described in that oxidative stress rapidly enhanced the reentry to and these cells when they undergo dedifferentiation. progression of Mu¨ller cells in the cell cycle.

Glial Cells Protected Retinal Neurons in Effect of Oxidative Stress on Glial Cell Coculture from PQ-Induced Oxidative Stress Proliferation PQ induced the apoptosis of amacrine and photo- The decrease in differentiation was accompanied receptor cells in pure neuronal cultures (Rotstein et al., by an increase in the ability of glial cells to proliferate. 2003). By day 6, photoreceptor degeneration had already Both PQ treatments (240 lMfor4hror48lM for started because of a lack of trophic factors (Rotstein 24 hr) rapidly stimulated the uptake of BrdU, a parame- et al., 1997), as evidenced by the presence of several ter of cell cycle progression (Fig. 4C,D, arrowheads); in TUNEL-positive cells in pure neuronal cultures (Fig. control cultures about 10% of glial cells took up BrdU, 6A,B). Oxidative stress increased neuronal apoptosis, as whereas this percentage increased about 3 times in PQ- demonstrated by the higher number of TUNEL-positive treated cultures (Fig. 5A). PQ also increased PCNA cells in PQ-treated cultures (Fig. 6C,D). To evaluate expression, another parameter of cell-cycle progression; whether glial cells protected neurons from oxidative the percentage of glial cells expressing PCNA doubled stress–induced apoptosis, 5-day neuroglial cocultures after PQ treatment (Fig. 5B), increasing from less than were treated with 48 lM PQ for 24 hr. Glial cells effec- 4% in controls to 10% in glial cells treated with PQ. tively prevented PQ-induced neuronal apoptosis, as very The increases in markers of cell-cycle progression few TUNEL-stained cells were observed in their pres- were consistent with an increase in the number of glial ence after PQ treatment (Fig. 6G,H). This protection cells in PQ-treated cultures (Fig. 5C). With PQ treat- was already visible in 6-day control cocultures, where ment, glial cells grew and divided faster than controls. few apoptotic neurons were observed (Fig. 6E,F). Although all cultures initially had a similar number of We then evaluated whether glial cells protected cells, PQ-treated cultures showed 37.5% 6 12.4% more photoreceptors and amacrine neurons from oxidative cells than did controls. Together, these results suggest stress to a similar extent. As mentioned above, in 6-day

Journal of Neuroscience Research 970 Abrahan et al.

Fig. 4. Effect of PQ treatment on BrdU uptake in glial cells. Glial cells were cultured until 2 days after reaching confluence in FCS- containing media, which was then replaced with chemically defined media. Cultures were treated without (A, C, E) or with (B, D, F) 240 lM PQ for 4 hr. One hour after the addition of PQ, 10 lM BrdU was added to the cultures. Phase (A, B) and fluorescent photomicrographs showing BrdU uptake (C, D, arrowheads), determined with an anti-BrdU monoclonal antibody and their nuclei labeled with DAPI (E, F, arrowheads). Scale bar 5 30 lm. pure neuronal cultures photoreceptor degeneration had already started; hence, both apoptotic and viable photoreceptors were observed in controls [Fig. 7A–C (arrow- Fig. 5. Effect of PQ on cell cycle and proliferation of Mu¨ller glial cells. heads and thin arrows, respectively), M]. PQ increased Glial cells were incubated in FCS-supplemented media until 2 days after photoreceptor apoptosis (Fig. 7D–F, arrowheads), from reaching confluence. The culture medium was then replaced by a about 50% in controls to almost 75% of total photore- chemically defined medium, and cells were treated without (2PQ) or ceptors (Fig. 7M). Glial cells efficiently protected photo- with (1PQ) 240 lM PQ for 4 hr. A: Percentage of glial cells showing receptors; most of them were viable in control cocul- BrdU uptake. B: Percentage of glial cells expressing PCNA, determined tures (Fig. 7G–I, thin arrows) and were unaffected by by immunocytochemistry using a monoclonal antibody. C: Amount of PQ treatment (Fig. 7J–L). Only 8% of the photorecep- glial cells/dish. In A and B, the percentage of glial cells showing BrdU uptake and PCNA expression was calculated relative to the total amount tors showed apoptotic nuclei in both control and PQ- of glial cells in the culture. Each value represents the mean 6 SD of at treated cocultures (Fig. 7M). least three separate experiments, with 3–4 dishes per experiment Most amacrine neurons were viable in 6-day pure (*statistically significant difference at P < 0.05 compared with control). neuronal cultures (Fig. 8A–C, thin arrows). PQ induced their apoptosis (Fig. 8D–F, arrowheads), increasing the percentage of apoptotic amacrine neurons from less than glial cells protected them from oxidative stress, as they 10% in controls to more than 35% in PQ-treated neuro- were unaffected by PQ [Fig. 8J–L (thin arrows), M]. nal cultures (Fig. 8M). Nearly all amacrine neurons were Protection by glial cells involved the preservation viable in untreated cocultures (Fig. 8G–I, thin arrows); of mitochondrial membrane potential. Oxidative stress

Journal of Neuroscience Research Oxidative Stress and Glial Proliferation 971

Glial Cells Protected Retinal Neurons in Coculture from H2O2-Induced Oxidative Stress We then evaluated whether glial cells had a similar neuroprotective effect on treatment with another pro- oxidant, H2O2. TUNEL staining conﬁrmed previous results (Chucair et al., 2007): few TUNEL-positive neurons were observed in control, pure neuronal cultures (Fig. 10A,B), and H2O2 increased their number (Fig. 10C,D). Glial cells were not only unaffected by H2O2 but also almost completely prevented neuronal apoptosis; very few TUNEL-positive neurons were found in either control (Fig. 10E,F) or H2O2-treated (Fig. 10G,H) cocultures.

DISCUSSION Our work shows that glial cells in culture activated different responses when exposed to oxidative stress. They rapidly dedifferentiated, reentered the cell cycle, and increased proliferation. Glial cells were better suited than retinal neurons to withstand oxidative stress; they not only survived the oxidative injury but also protected neurons, preventing their apoptosis. Hence, glial cells might carry out several roles in their interaction with neurons exposed to oxidative damage, preserving neurons already present in the retina and providing a source of actively dividing precursors. In the amphibian eye, which is capable of regener- ating the retina from pigmented epithelium, regeneration is a two-step process: an initial epithelial cell dedifferentiation and extensive proliferation to give rise to retinal progenitors, followed by differentiation of these progenitors to obtain different neuronal types (Moshiri et al., 2004). Hence, dedifferentiation and proliferation are the first steps to be accomplished in a regeneration process. Fig. 6. Protective effect of glial cells on neuronal apoptosis induced Our results reveal that one of the first responses of Mu¨l- by oxidative stress. Phase (A, C, E, G) and fluorescent (B, D, F, H) ler glial cells to oxidative stress was to activate their de- photomicrographs of pure retinal neurons (A–D) and neuroglial cocultures (E–H) incubated for 5 days in chemically defined media or differentiation. This was evidenced by the rapid decrease in FCS-containing media, respectively. Cocultures were then in the expression of CRALBP and GFAP and by the switched to the chemically defined media used for purified neurons, parallel increase in SMA expression. These changes are and cells were treated without (A, B, E, F) or with (C, D, G, H) consistent with those occurring with dedifferentiation in 48 lM PQ for 24 hr. Apoptotic cells were identified by TUNEL cultured porcine Mu¨ller cells (Guidry et al., 1996). SMA labeling (right column). Scale bar 5 10 lm. levels are high in rat brain astrocytes in early stages of postnatal development and decrease with maturation; similar changes in SMA expression occur during astro- cyte development in culture (Buniatian et al., 1999). induced the loss of mitochondrial membrane potential in Low GFAP expression might be associated with the pro- both photoreceptors and amacrine neurons in 4-day liferation of Mu¨ller cells; this expression increases during pure neuronal cultures (Rotstein et al., 2003). Similarly, development in vivo and in vitro (Buniatian et al., in 6-day neuronal cultures, the percentage of neurons 1999), and it has been reported that those glial cells that preserving mitochondrial membrane potential decreased reenter the cell cycle after neurotoxic damage fail to from almost 90% to about 60% of amacrine neurons and increase GFAP expression (Fischer and Reh, 2003). from nearly 50% to 27% of photoreceptors in controls Dedifferentiation of glial cells was parallel to an and PQ-treated cultures, respectively (Fig. 9). Glial cells increase in proliferation; oxidative stress increased the almost completely blocked mitochondrial depolarization mitogenic potential of confluent Mu¨ller cells, inducing in these neurons; the percentage of amacrine and photo- their reentry into the cell cycle and thus augmenting the receptor neurons retaining their mitochondrial mem- pool of available glial cells. Damage to fish and chick brane potential in neuroglial cocultures was almost the retinas promotes Mu¨ller cell proliferation (Braisted et al., same with or without PQ treatment (Fig. 9). 1994; Wu et al., 2003), and a small subpopulation of

Journal of Neuroscience Research 972 Abrahan et al.

Fig. 7. Glial cell protection on photoreceptor apoptosis induced by (G–I, thin arrows), and PQ treatment did not affect this viability oxidative stress. Phase (left column) and fluorescent (middle and right (J–L, thin arrows). M: Percentage of apoptotic photoreceptors in cul- columns) micrographs showing pure retinal neurons (A–F) and neu- tures with (1PQ) or without (2PQ) PQ was determined relative to roglial cocultures (G–L), treated as described in Figure 6, without the total amount of photoreceptors/dish by analyzing nuclei integrity (A–C, G–I) or with (D–F, J–L) PQ. After fixation, photoreceptors with DAPI. Scale bar 5 10 lm. Values represent average 6 SD of at were visualized with the specific monoclonal antibody Rho4D2 least three separate experiments with 3–4 dishes for each culture con- (middle column), and nuclei were stained with DAPI (right column). dition. Notice the distinct reduction in photoreceptor apoptosis in By day 6, pure neuronal cultures showed both viable and apoptotic cocultures treated or not treated with PQ compared with pure neu- photoreceptors (thin arrows and arrowheads, respectively, in A–C), ronal cultures (*statistically significant difference at P < 0.01 com- and the addition of PQ augmented this apoptosis (D–F, arrowheads). pared with neuronal cultures). In contrast, most photoreceptors were viable in 6-day cocultures

Journal of Neuroscience Research Oxidative Stress and Glial Proliferation 973

Fig. 8. Glial cell protection against apoptosis of amacrine cells M: Percentage of apoptotic amacrine neurons in cultures with induced by oxidative stress. Phase (A, D, G, J) and fluorescent (B, (1PQ) or without (2PQ) PQ was determined relative to the total C, E, F, H, I, K, L) micrographs showing pure retinal neurons (A– amount of amacrine neurons/dish. Scale bar 5 10 lm. Values repre- F) and neuroglial cocultures (G–L), treated as described in Figure 7, sent average 6 SD of at least three separate experiments, with 3–4 without (A–C, G–I) or with (D–F, J–L) PQ. Amacrine neurons dishes for each culture condition. Although in the absence of PQ were visualized with the specific monoclonal antibody HPC-1 treatment the percentage of apoptotic amacrine neurons was similar (middle column), and nuclei were stained with DAPI (right column). in pure neuronal cultures and cocultures, a significant reduction in By day 6, most amacrine cells were viable (A–C, thin arrows) in amacrine cell apoptosis was observed in cocultures compared with pure neuronal cultures, and PQ markedly increased their apoptosis neuronal cultures after PQ treatment (*statistically significant differ- (D–F, arrowheads). In contrast, 6-day cocultures had mostly viable ence at P < 0.01 compared with neuronal cultures). amacrine neurons, even after PQ treatment (G–L, arrowheads). these cells goes on to express neuronal markers (Fischer retinal neurons (Das et al., 2006). Even subtoxic doses of and Reh, 2001; Bernardos et al., 2007) or to differenti- glutamate induce the reentry of Mu¨ller cells into the cell ate as cone photoreceptors (Raymond et al., 2006). In cycle and the generation of retinal neurons in adult mice the rodent retina, neurotoxic damage activates the prolif- retinas (Takeda et al., 2008). Hence, our results showing eration and expression of progenitor markers in Mu¨ller that Mu¨ller cells had the capacity to rapidly dedifferenti- cells (Ooto et al., 2004), which differentiate into specific ate and reenter the cell cycle are consistent with previ-

Journal of Neuroscience Research 974 Abrahan et al.

Fig. 10. Protective effect of glial cells on neuronal apoptosis induced by H2O2. Phase (A, C, E, G) and fluorescent (B, D, F, H) micrographs of retinal neurons cultured alone (A–D) or cocultured with glial cells (E–H) were treated on day 5 without (A, B, E, F) or with (C, D, G, H) 100 lMH2O2 for 30 min. The media were then replaced with fresh chemically defined media, and cells were fixed 3 hr Fig. 9. Effect of glial cells on preservation of mitochondrial mem- later and processed for TUNEL staining (B, D, F, H). Pure neuronal brane potential in neurons after PQ-induced oxidative stress. Retinal cultures showed few TUNEL-labeled neurons, and treatment with neurons were cultured alone or cocultured with glial cells as previ- H2O2 markedly increased their number (B, D, arrowheads). In con- ously described. The percentage of photoreceptors and amacrine neu- trast, almost no TUNEL-labeled neurons were visible in neuroglial rons preserving their mitochondrial membrane potential, relative to cocultures, and no increase in their number was observed after H2O2 the total amount of each neuronal type in the culture, was deter- treatment (F, H, arrowheads). Scale bar 5 25 lm. mined using MitoTracker Red. Values represent the average 6 SD of at least three separate experiments, with 3–4 dishes for each culture condition (*statistically significant difference at P < 0.05 compared with control). bFGF, and CNTF in the chick retina (Fischer et al., 2004), a similar autocrine release of GDNF and other trophic factors might occur upon oxidative stress to pro- ous data supporting their role as potential stem cells of mote proliferation in glial cells. the retina. Signals inducing Mu¨ller cells to reenter the Glial cell survival was not affected by oxidative cell cycle still remain to be identified. We have recently stress, at least at an oxidant concentration that triggered shown that exogenous addition of GDNF, bFGF, or in- neuronal death (Rotstein et al., 2003; Chucair et al., sulin rapidly induces Mu¨ller cells to reenter the cell cycle 2007). Brain glial cells are more resistant than neurons to (Insua et al., 2008). Because cultured Mu¨ller glial cells oxidative damage (Bolanos et al., 1995; Watts et al., produce bFGF and GDNF (Lin et al. 1994; Harada 2005), apparently because of their efficient detoxification et al., 2000; Hisaoka et al., 2007) and neurotoxic dam- pathways, such as the glutathione pathway, with a rele- age increases the levels of trophic factors such as IGF-II, vant role in protecting glial cells from this damage (Ben-

Journal of Neuroscience Research Oxidative Stress and Glial Proliferation 975

Yoseph et al., 1996). In the rat retina, glutathione is indicate that Mu¨ller cells have the potential to enhance concentrated in Mu¨ller glial cells, which transfer it to their proliferation and protect neurons after exposure to neurons under ischemic conditions (Schutte and Werner, oxidative stressors. 1998). Hence, glial cells might have detoxification In conclusion, our results provide evidence that mechanisms that grant them enhanced resistance to oxi- Mu¨ller cells might participate in different compensatory dative damage and that might play a role in protecting responses after oxidative injury of the retina. In addition neurons from this damage as well. to preventing neuronal apoptosis, they might also be The existence of a protective effect was clearly evi- involved in attempts to repair neuronal damage. Dedif- dent in the almost complete prevention of neuronal apo- ferentiation and activation of their proliferation would ptosis in neuroglial cocultures on treatment with PQ or provide a pool of progenitors that might either supply H2O2. The presence of glial cells virtually blocked the more glial cells, essential for granting trophic support to marked increase in apoptosis of photoreceptors and ama- avoid neuronal death, or for eventually differentiating crine cells, which is triggered by these oxidants in pure into neuronal progenitors, compensate for neuronal neuronal cultures. This protection was associated with losses and consequently maintain retina functionality. the preservation of mitochondrial membrane potential in both neuronal types. Active molecular trafficking exists between glial cells and neurons. In addition to providing ACKNOWLEDGMENTS efficient detoxification mechanisms, glial cells provide L.P. and N.R. are CONICET Research Career neurons with several trophic factors that might activate members; O.L.G. and C.E.A. are CONICET posdoc- survival pathways upstream of mitochondrial permeabili- toral and doctoral fellows, respectively; M.F.I. had a zation. Mu¨ller glial cells can protect bipolar cells in cul- CONICET Research fellowship during this work. The ture and prevent the neurotoxic-induced death of gan- authors are grateful to Dr. A. Suburo (Universidad Aus- glion cells (Wexler and Berkovich, 1998; Kawasaki tral, Buenos Aires, Argentina) and Dr. D. Hicks et al., 2000). They have been proposed to indirectly reg- (INSERM, Strassbourg, France) for generously providing ulate the survival of photoreceptors promoted by trophic SMA and GFAP antibodies, respectively, and to Beatriz factors during degeneration (Wahlin et al., 2000, 2001; de los Santos and Horacio De Genaro for their excellent Harada et al., 2002). Among several trophic factors, glial technical assistance. cells can provide docosahexaenoic acid (DHA) to photoreceptors in coculture (Politi et al, 2001a). DHA shields photoreceptors from oxidative stress–induced apoptosis REFERENCES by preserving mitochondrial potential through activation of the extracellular signal–regulated kinase (ERK) path- Adler R. 1982. Regulation of neurite growth in purified retina neuronal way (Rotstein et al., 1997, Politi et al., 2001a,b; German cultures: effects of PNPF, a substratum-bound, neurite-promoting factor. J Neurosci Res 8:165–177. et al., 2006). Mu¨ller cells might then protect almost all Barnstable CJ. 1980. Monoclonal antibodies which recognize different neuronal types in the retina from diverse injuries, either cell types in the rat retina. Nature 286:231–235. directly providing the multiple trophic factors required Ben-Yoseph O, Boxer PA, Ross BD. 1996. Assessment of the role of the for neuronal survival or, in a more general fashion, acti- glutathione and pentose phosphate pathways in the protection of pri- vating survival pathways in glial cells that might then mary cerebrocortical cultures from oxidative stress. J Neurochem 66: release signals promoting neuronal survival. 2329–2337. It must be borne in mind that these results repre- Bernardos RL, Barthel LK, Meyers JR, Raymond PA. 2007. Late-stage sent the responses of Mu¨ller glial cells in vitro to oxida- neuronal progenitors in the retina are radial Mu¨ller glia that function as tive damage, responses that may differ from those of retinal stem cells. J Neurosci 27:7028–7040. their in vivo counterparts. The retina is a highly struc- Bolanos JP, Heales SJ, Land JM, Clark JB. 1995. Effect of peroxynitrite on the mitochondrial respiratory chain: differential susceptibility of neu- tured system, with specific interactions and contacts rones and astrocytes in primary culture. J Neurochem 64:1965–1972. between its different cell types. This specificity is lost in Braisted JE, Essman TF, Raymond PA. 1994. Selective regeneration of vitro; the number and type of interactions and the prox- photoreceptors in goldfish retina. Development 122:1427–1438. imity between glial and neuronal cells are greatly dis- Buniatian GH, Gebhardt R, Mecke D, Traub P, Wiesinger H. 1999. turbed, leading to profound modifications in the cross Common myofibroblastic features of newborn rat astrocytes and cir- talk between these cell types. In addition, the change in rhotic rat liver stellate cells in early cultures and in vivo. Neurochem the surroundings of glial cells from tissue in vivo to a Int 35:317–327. culture medium undoubtedly affects many of their fea- Carmody RJ, McGowan AJ, Cotter TG. 1999. Reactive oxygen species tures. These changes might explain the differences in the as mediators of photoreceptor apoptosis in vitro. Exp Cell Res 248: protein profile of Mu¨ller glia in vitro and in vivo 520–530. (Hauck et al., 2003), in the responses to ischemia of cul- Chang GQ, Hao Y, Wong F. 1993. Apoptosis: final common pathway of photoreceptor death in rd, rds, and rhodopsin mutant mice. Neuron tured astrocytes and of those in hippocampal slices 11:595–605. (Duffy and MacVicar, 1996), and in the susceptibility to Chucair AJ, Rotstein NP, Sangiovanni JP, During A, Chew EP, Politi excitatory amino acids between neuroglial cocultures LE. 2007. Lutein and zeaxanthin protect photoreceptors from apoptosis and retinal explants (Heidinger et al., 1998). Even induced by oxidative stress. Relation with docosahexaenoic acid. Invest acknowledging these differences, our results nevertheless Ophthalmol Vis Sci 48:5168–5177.

Journal of Neuroscience Research 976 Abrahan et al.

Cingolani C, Rogers B, Lu L, Kachi S, Shen J, Campochiaro PA. 2006. Kawasaki A, Otori Y, Barnstable CJ. 2000. Mu¨ller cell protection of rat Retinal degeneration from oxidative damage. Free Rad Biol Med retinal ganglion cells from glutamate and nitric oxide neurotoxicity. 40:660–669. Invest Ophthalmol Vis Sci 41:3444–3450. Das AV, Mallya KB, Zhao X, Ahmad F, Bhattacharya S, Thoreson WB, Kennedy BN, Li C, Ortego J, Coca-Prado M, Sarthy VP, Crabb JW. Hegde GV, Ahmad I. 2006. Neural stem cell properties of Mu¨ller glia 2003. CRALBP transcriptional regulation in ciliary epithelial, retinal in the mammalian retina: regulation by Notch and Wnt signaling. Dev Mu¨ller and retinal pigment epithelial cells. Exp Eye Res 76:257–260. Biol 299:283–302. Kljavin IJ, Lagenaur C, Bixby JL, Reh TA. 1994. Cell adhesion mole- Derouiche A. 1996. Possible role of the Mu¨ller cell uptake and metabo- cules regulating neurite growth from amacrine and rod photoreceptor lism of glutamate in the mammalian outer retina. Vis Res 36:3875– cells. J Neurosci 14:5035–5049. 3878. Li X, Sun AY. 1999. Paraquat induced activation of transcription factor Desagher S, Glowinski J, Premont J. 1996. Astrocytes protect neurons AP-1 and apoptosis in PC12 cells. J Neural Transm 106:1–21. from hydrogen peroxide toxicity. J Neurosci 16:2553–2562. Liddell JR, Hoepken HH, Crack PJ, Robinson SR, Dringen R. 2006. Glutathione peroxidase 1 and glutathione are required to protect mouse Dubois-Dauphin M, Poitry-Yamate C, de Bilbao F, Julliard AK, Jourdan astrocytes from iron-mediated hydrogen peroxide toxicity. J Neurosci F, Donati G. 2000. Early postnatal Mu¨ller cell death leads to retinal but Res 84:578–586. not optic nerve degeneration in NSE-Hu-Bcl-2 transgenic mice. Neu- rosci 95:9–21. Lin LF, Zhang TJ, Collins F, Armes LG. 1994. Purification and initial characterization of rat B49 glial cell line-derived neurotrophic factor. Duffy S, MacVicar BA. 1996. In vitro ischemia promotes calcium influx J Neurochem 63:758–768. and intracellular calcium release in hippocampal astrocytes. J Neurosci Lohr HR, Kuntchithapautham K, Sharma AK, Rohrer B. 2006. Multi- 16:71–81. ple, parallel cellular suicide mechanisms participate in photoreceptor cell Fischer AJ, Reh TA. 2001. Mu¨ller glia are a potential source of neural death. Exp Eye Res 83:380–389. regeneration in the postnatal chicken retina. Nat Neurosci 4:247–252. McGillem GS, Guidry C, Dacheux RF. 1998. Antigenic changes of rab- Fischer AJ, Reh TA. 2003. Potential of Mu¨ller glia to become neuro- bit retinal Mu¨ller cells in culture. Invest Ophthalmol Vis Sci. 39:1453– genic retinal progenitor cells. Glia 43:70–76. 1461. Fischer AJ, Omar G, Eubanks J, McGuire CR, Dierks BD, Reh TA. Moshiri A, Close J, Reh TA. 2004. Retina stem cells and regeneration. 2004. Different aspects of gliosis in retinal Mu¨ller glia can be induced Int J Dev Biol 48:1003–1014. by CNTF, insulin, and FGF2 in the absence of damage. Mol Vis Ooto S, Akagi T, Kageyama R Akita J, Mandai M, Honda Y, Takahashi 10:973–986. M. 2004. Potential for neural regeneration after neurotoxic injury in German OL, Insua MF, Gentili C, Rotstein NP, Politi LE. 2006. Doco- the adult mammalian retina. Proc Natl Acad Sci U S A 101:13654– sahexaenoic acid prevents apoptosis of retina photoreceptors by activat- 13659. ing the ERK/MAPK pathway. J Neurochem 98:1507–1520. Poitry S, Poitry-Yamate C, Ueberfeld J, MacLeish PR, Tsacopoulos M. Guidry C. 1996. Isolation and characterization of porcine Mu¨ller cells. 2000. Mechanisms of glutamate metabolic signaling in retinal glial Myofibroblastic dedifferentiation in culture. Invest Ophthalmol Vis Sci (Mu¨ller) cells. J Neurosci 20:1809–1821. 37:740–752. Poitry-Yamate CL, Poitry S, Tsacopoulos M. 1995. Lactate released by Harada T, Harada C, Nakayama N, Okushama S, Yoshida K, Kohsaka S, Mu¨ller glial cells is metabolized by photoreceptors from mammalian Matsuda H, Wada K. 2000. Modification of glial-neuronal cell interac- retina. J Neurosci 15:5179–5191. tions prevents photoreceptor apoptosis during light-induced retinal Politi LE, Bouzat C, de los Santos EB, Barrantes FJ. 1996. Heterologous degeneration. Neuron 26:533–41. retinal cultured neurons and cell adhesion molecules induce clustering Harada T, Harada C, Kohsaka S, Wada E, Yoshida K, Ohno S, Mamada of acetylcholine receptors and polynucleation in mouse muscle BC3H- H, Tanaka K, Parada L, Wada K. 2002. Microglia–Mu¨ller glia cell inter- 1 clonal cell line. J Neurosci Res 43:639–651. actions control neurotrophic factor production during light-induced Politi LE, Rotstein NP, Carri N. 2001a. Effects of docosahexaenoic acid retinal degeneration. J Neurosci 22:9228–9236. on retinal development: cellular and molecular aspects. Lipids 36:927– 935. Hauck SM, Suppmann S, Ueffing M. 2003. Proteomic profiling of pri- Politi LE, Rotstein NP, Carri NG. 2001b. Effects of GDNF on neuro- mary retinal Mu¨ller glia cells reveals a shift in expression patterns upon blast proliferation and photoreceptor survival: additive protection with adaptation to in vitro conditions. Glia 44:251–263. docosahexaenoic acid. Invest Ophthalmol. Vis Sci 42:3008–3015. Heidinger V, Dreyfus H, Sahel J, Christen Y, Hicks D. 1998. Excito- Portera-Cailliau C, Sung CH, Nathans J, Adler R. 1994. Apoptotic pho- toxic damage of retinal glial cells depends upon normal neuron-glial toreceptor cell death in mouse models of retinitis pigmentosa. Proc interactions. Glia 23:146–155. Natl Acad Sci U S A 91:974–978. Hicks D, Barnstable CJ. 1987. Different rhodopsin monoclonal antibodies Raymond PA, Barthel LK, Bernardos LK, Bernardos RL, Perkowski JJ. reveal different binding patterns on developing and adult rat retina. 2006. Molecular characterization of retinal stem cells and their niches in J Histochem Cytochem 35:1317–1328. adult zebrafish. BMC Dev Biol 6:36. Hisaoka K, Takebayashi M, Tsuchioka M, Maeda N, Nakata Y, Yama- Rich KA, Figueroa SL, Zhan Y, Blanks JC. 1995. Effects of Mu¨ller cell waki S. 2007. Antidepressants increase glial cell line-derived neurotro- disruption on mouse photoreceptor cell development. Exp Eye Res phic factor production through monoamine-independent activation of 61:235–248. protein tyrosine kinase and extracellular signal-regulated kinase in glial Rotstein NP, Aveldanõ MI, Barrantes FJ, Roccamo AM, Politi LE. cells. J Pharmacol Exp Ther 321:148–157. 1997. Apoptosis of photoreceptors in cultures of rat retinal cells. Pro- Hoyt KR, Galaher AJ, Hastings TG, Reynolds IJ. 1997. Characteristics tective effect of docosahexaenoic acid. J Neurochem 69:504–513. of hydrogen peroxide toxicity in cultured rat forebrain neurons. Neuro- Rotstein NP, Aveldanõ MI., Barrantes FJ, Politi LE. 1996. Docosahexae- chem Res 22:333–340. noic acid is required for the survival of rat retinal photoreceptors in Insua MF, Simon MV, Garelli A, De los Santos EB, Rotstein NP, Politi vitro. J Neurochem 66:1851–1859. LE. 2008. Trophic factors and neuronal interactions regulate the cell Rotstein NP, Politi LE, Aveldanõ MI. 1998. Docosahexaenoic acid pro- cycle and pax6 expression in Mu¨ller stem cells. J Neurosci Res 86: motes differentiation of developing photoreceptors in culture. Invest 1459–1471. Ophthalmol Vis Sci 39:2750–2758.

Journal of Neuroscience Research Oxidative Stress and Glial Proliferation 977

Rotstein NP, Politi LE, German OL, Girotti R. 2003. Protective effect and other cells of the inner retina, but not photoreceptors. Invest Oph- of docosahexaenoic acid on oxidative stress-induced apoptosis of retina thalmol Vis Sci 41:927–936. photoreceptors. Invest Ophthalmol Vis Sci 44:2252–2259. Wahlin KJ, Adler R, Zack DJ, Campochiaro PA. 2001. Neurotrophic Schutte M, Werner P. 1998. Redistribution of glutathione in the ische- signaling in normal and degenerating rodent retinas. Exp Eye Res mic rat retina. Neurosci Lett 246:53–56. 73:693–701. Sun H, Nathans J. 2001. ABCR, the ATP-binding cassette transporter Watts LT, Rathinam ML, Schenker S, Henderson GI. 2005. Astrocytes responsible for Stargardt macular dystrophy, is an efﬁcient target of all- protect neurons from ethanol-induced oxidative stress and apoptotic trans-retinal-mediated photooxidative damage in vitro. J Biol Chem death. J Neurosci Res 80:655–666. 276:11766–11774. Wexler EM, Berkovich O. 1998. Role of the low-afﬁnity NGF receptor Takeda M, Takamiya A, Jiao J-W, Cho K-S, Trevino SG, Matsuda T, (p75) in survival of retinal bipolar cells. Vis Neurosci 15:211–218. Chen DF. 2008. Aminoadipate induces progenitor cell properties of Mu¨ller glia in adult mice. Invest Ophthalmol Vis Sci 49:1142–1150. Wu LY, Li M, Hinton DR, Guo L, Jiang S, Wang JT, Zeng A, Xie JB, Tso MO, Zhang C, Abler AS, Chang CJ, Wong F, Chang GQ, Lam TT. Snead M, Shuler C, Maxson RE Jr, Liu YH. 2003. Microphthalmia 1994. Apoptosis leads to photoreceptor degeneration in inherited retinal resulting from MSX2-induced apoptosis in the optic vesicle. Invest dystrophy of RCS rats. Invest Ophthalmol Vis Sci 35:2693–2699. Ophthalmol Vis Sci 44:2404–2412. Wahlin KJ, Campochiaro PA, Zack DJ, Adler R. 2000. Neurotrophic Yang W, Sun AY. 1998. Paraquat-induced free radical reaction in mouse factors cause activation of intracellular signaling pathways in Mu¨ller cells brain microsomes. Neurochem Res 23:47–53.

Journal of Neuroscience Research