necrosis and apoptosis after retinal ischemia: involvement...

8
Necrosis and Apoptosis after Retinal Ischemia: Involvement of NMDA-Mediated Excitotoxicity and p53 Choun-Ki Joo, 1 Jun-Sub Choi, 1 Hyuk W. Ko 2 Kye-yoon Park, 2 Seonghyang Sohn, 5 Myung H. Chun, 4 Young J. Oh, 5 and ByoungJ. Gwag 2 PURPOSE. Accumulated evidence has shown that apoptosis and necrosis contribute to neuronal death after ischemia. The present study was performed to study the temporal and spatial patterns of neuronal necrosis and apoptosis after ischemia in retina and to outline mechanisms underlying necrosis and apoptosis. METHODS. Retinal ischemia was induced by increasing intraocular pressure to a range of 160 mm Hg to 180 mm Hg for 90 minutes in adult rats. The patterns of neuronal cell death were determined using light and electron microscopy and were visualized by TdT-dUTP nick-end labeling (TUNEL). The mRNA expression profile of p53 was examined using reverse transcription-polymerase chain reaction (RT-PCR) and in situ hybridization histochemistry. Immunohistochemistry was performed using anti-p53, anti-microtubule associated protein-2, and anti-glial fibrillary acidic protein antibod- ies. RESULTS. Within 4 hours after ischemia, neurons in the inner nuclear cell layer (INL) and ganglion cell layer (GCL) underwent marked necrosis, made apparent by swelling of the cell body and mitochondria, early fenestration of the plasma membrane, and irregularly scattered condensation of nuclear chromatin. After 3 days, the INL and GCL neurons showed further degeneration through apoptosis marked by cell body shrinkage, aggregation, and condensation of nuclear chromatin. Apoptotic neurons were also observed sparsely in the outer nuclear cell layer. Intravitreal injections of MK-801 prevented early neuronal degeneration after ischemia. Of note, mRNA and protein levels of p53, the tumor suppressor gene known to induce apoptosis, were increased in the retinal areas undergoing apoptosis 1 to 3 days after ischemic injury. CONCLUSIONS. Ischemia produces the ./V-methyl-D-aspartate-mediated necrosis and slowly evolving apoptosis of neurons in the retina. The latter may depend on the expression of the p53 proapop- tosis gene. {Invest Ophthalmol Vis Sci. 1999;40:713-720) G lutamate, an excitatory neurotransmitter, becomes toxic through overactivation of receptors sensitive to W-methyl-D-aspartate (NMDA), a-amino-3-hydroxy-5- methyl-4-isoxazolepropionic acid (AMPA), and kainate. 1 ' 2 Intra- cellular accumulation of Ca 2+ is known to cause glutamate- mediated neuronal degeneration. 3 Glutamate neurotoxicity is marked by swelling of the cell body and mitochondria, early cell membrane lysis, and the irregularly scattered heterochro- matin condensation characteristic of necrosis. 4 " 6 In support of this, neuronal death produced by NMDA, AMPA, or kainate is not attenuated by cycloheximide or neurotrophic factors that attenuate apoptosis in various neurons in the same culture system. 4 ' 7 ' 8 Extensive evidence indicates that hypoxic isch- From the 'Departments of Ophthalmology and 4 Anatomy, Catho- lic University Medical College, Seoul; the 2 Department of Pharmacol- ogy and ^Laboratory of Cell Biology, Institute for Medical Science, Ajou University School of Medicine, Suwon; and the department of Biol- ogy, Yonsei University College of Science, Seoul, Korea. Supported by Korean Ministry of Health and Welfare Grant HMP- 96-M-2-1042. Submitted for publication May 7, 1998; revised September 29, 1998; accepted November 4, 1998. Proprietary interest category: N. Reprint requests: Byoung Joo Gwag, Department of Pharmacol- ogy, Ajou University School of Medicine, San5, Wonchondong, Paldalku, Suwon, Kyungkido 442-749, Korea. emia produces neuronal necrosis predominantly through mechanisms involving activation of glutamate receptors. 9 "" However, nuclear chromatin condensation, prominent internu- cleosomal DNA fragmentation, and dependence on protein synthesis also have been observed in various models of brain ischemia, 12 " 15 which suggests that neuronal necrosis and apo- ptosis may occur in parallel after ischemic injuries in vivo and in vitro. 16 " 18 Molecular mechanisms of apoptosis are undergoing exten- sive investigation. Specifically, the p53 tumor suppressor gene has been recognized as a proapoptosis molecule in many types of cells. Mice deficient in p53 are highly resistant to apoptosis induced by various insults, including ionizing irradiation, 19 ' 20 whereas overexpression of p53 induces apoptosis in myeloid leukemia or melanoma cells. 21 ' 22 In the nervous system, brain ischemia or kainate-induced seizure increases p53 expression in the hippocampal and cortical neurons programed to die. 23 " 25 Recentfindingsprovide compelling evidence for p53- dependent neuronal death that transgenic mice deficient in p53 are resistant to neuronal injuries such as y-irradiation, seizure, ischemia, or adrenalectomy. 26 " 29 The retina has been widely used in investigation of mech- anisms of neuronal degeneration after ischemia, excitotoxicity, and axotomy. 2 ' 22 ' 30 " 32 In particular, elevation of intraocular pressure in the retina is known to cause feasible and adjustable ischemic neuronal death. 33 In this model, mixed patterns of Investigative Ophthalmology & Visual Science, March 1999, Vol. 40, No. 3 Copyright © Association for Research in Vision and Ophthalmology 713

Upload: others

Post on 17-May-2020

5 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Necrosis and Apoptosis after Retinal Ischemia: Involvement ...web.yonsei.ac.kr/neurolab/published/17.pdf · Necrosis and Apoptosis after Retinal Ischemia: Involvement of NMDA-Mediated

Necrosis and Apoptosis after Retinal Ischemia:Involvement of NMDA-Mediated Excitotoxicity and p53

Choun-Ki Joo,1 Jun-Sub Choi,1 Hyuk W. Ko2 Kye-yoon Park,2 Seonghyang Sohn,5

Myung H. Chun,4 Young J. Oh,5 and ByoungJ. Gwag2

PURPOSE. Accumulated evidence has shown that apoptosis and necrosis contribute to neuronaldeath after ischemia. The present study was performed to study the temporal and spatial patternsof neuronal necrosis and apoptosis after ischemia in retina and to outline mechanisms underlyingnecrosis and apoptosis.

METHODS. Retinal ischemia was induced by increasing intraocular pressure to a range of 160 mm Hgto 180 mm Hg for 90 minutes in adult rats. The patterns of neuronal cell death were determinedusing light and electron microscopy and were visualized by TdT-dUTP nick-end labeling (TUNEL).The mRNA expression profile of p53 was examined using reverse transcription-polymerase chainreaction (RT-PCR) and in situ hybridization histochemistry. Immunohistochemistry was performedusing anti-p53, anti-microtubule associated protein-2, and anti-glial fibrillary acidic protein antibod-ies.

RESULTS. Within 4 hours after ischemia, neurons in the inner nuclear cell layer (INL) and ganglioncell layer (GCL) underwent marked necrosis, made apparent by swelling of the cell body andmitochondria, early fenestration of the plasma membrane, and irregularly scattered condensation ofnuclear chromatin. After 3 days, the INL and GCL neurons showed further degeneration throughapoptosis marked by cell body shrinkage, aggregation, and condensation of nuclear chromatin.Apoptotic neurons were also observed sparsely in the outer nuclear cell layer. Intravitreal injectionsof MK-801 prevented early neuronal degeneration after ischemia. Of note, mRNA and protein levelsof p53, the tumor suppressor gene known to induce apoptosis, were increased in the retinal areasundergoing apoptosis 1 to 3 days after ischemic injury.

CONCLUSIONS. Ischemia produces the ./V-methyl-D-aspartate-mediated necrosis and slowly evolvingapoptosis of neurons in the retina. The latter may depend on the expression of the p53 proapop-tosis gene. {Invest Ophthalmol Vis Sci. 1999;40:713-720)

Glutamate, an excitatory neurotransmitter, becomestoxic through overactivation of receptors sensitive toW-methyl-D-aspartate (NMDA), a-amino-3-hydroxy-5-

methyl-4-isoxazolepropionic acid (AMPA), and kainate.1'2 Intra-cellular accumulation of Ca2+ is known to cause glutamate-mediated neuronal degeneration.3 Glutamate neurotoxicity ismarked by swelling of the cell body and mitochondria, earlycell membrane lysis, and the irregularly scattered heterochro-matin condensation characteristic of necrosis.4"6 In support ofthis, neuronal death produced by NMDA, AMPA, or kainate isnot attenuated by cycloheximide or neurotrophic factors thatattenuate apoptosis in various neurons in the same culturesystem.4'7'8 Extensive evidence indicates that hypoxic isch-

From the 'Departments of Ophthalmology and 4Anatomy, Catho-lic University Medical College, Seoul; the 2Department of Pharmacol-ogy and ^Laboratory of Cell Biology, Institute for Medical Science, AjouUniversity School of Medicine, Suwon; and the department of Biol-ogy, Yonsei University College of Science, Seoul, Korea.

Supported by Korean Ministry of Health and Welfare Grant HMP-96-M-2-1042.

Submitted for publication May 7, 1998; revised September 29,1998; accepted November 4, 1998.

Proprietary interest category: N.Reprint requests: Byoung Joo Gwag, Department of Pharmacol-

ogy, Ajou University School of Medicine, San5, Wonchondong,Paldalku, Suwon, Kyungkido 442-749, Korea.

emia produces neuronal necrosis predominantly throughmechanisms involving activation of glutamate receptors.9""However, nuclear chromatin condensation, prominent internu-cleosomal DNA fragmentation, and dependence on proteinsynthesis also have been observed in various models of brainischemia,12"15 which suggests that neuronal necrosis and apo-ptosis may occur in parallel after ischemic injuries in vivo andin vitro.16"18

Molecular mechanisms of apoptosis are undergoing exten-sive investigation. Specifically, the p53 tumor suppressor genehas been recognized as a proapoptosis molecule in many typesof cells. Mice deficient in p53 are highly resistant to apoptosisinduced by various insults, including ionizing irradiation,19'20

whereas overexpression of p53 induces apoptosis in myeloidleukemia or melanoma cells.21'22 In the nervous system, brainischemia or kainate-induced seizure increases p53 expressionin the hippocampal and cortical neurons programed todie.23"25 Recent findings provide compelling evidence for p53-dependent neuronal death that transgenic mice deficient inp53 are resistant to neuronal injuries such as y-irradiation,seizure, ischemia, or adrenalectomy.26"29

The retina has been widely used in investigation of mech-anisms of neuronal degeneration after ischemia, excitotoxicity,and axotomy.2'22'30"32 In particular, elevation of intraocularpressure in the retina is known to cause feasible and adjustableischemic neuronal death.33 In this model, mixed patterns of

Investigative Ophthalmology & Visual Science, March 1999, Vol. 40, No. 3Copyright © Association for Research in Vision and Ophthalmology 713

Page 2: Necrosis and Apoptosis after Retinal Ischemia: Involvement ...web.yonsei.ac.kr/neurolab/published/17.pdf · Necrosis and Apoptosis after Retinal Ischemia: Involvement of NMDA-Mediated

714 Joo et al. IOVS, March 1999, Vol. 40, No. 3

ischemic necrosis and apoptosis appeared in the retina. Thepresent study was performed to investigate temporal and spa-tial degeneration patterns of retinal neurons exposed to isch-emia. In addition, the expression profiles of p53 mRNA andprotein and NMDA-mediated excitotoxicity were also analyzedto correlate with cell death patterns.34

MATERIALS AND METHODS

Induction of Retinal IschemiaRetinal ischemia was induced as previously described withminor modifications.33 In brief, adult male Sprague-Dawleyrats (180-220 g) were anesthetized intraperitoneally with 400mg/kg chloral hydrate. A 30!^-gauge needle connected to apneumatic pressure device was inserted into the anteriorchamber to increase the intraocular pressure (IOP) to a rangeof 160 mm Hg to 180 mm Hg for 30 to 90 minutes. Retinalischemia was monitored externally by absence of red reflex ofthe fundus, hardness of the eyeball, and the immediate returnof the red fundus reflex after removal of the needle from theanterior chamber. In experiments to examine NMDA-depen-dent retinal degeneration after ischemia, 10 /xl 5 mM MK-801,a selective NMDA antagonist, or vehicle was injected into theeye through a 10-jul Hamilton syringe inserted into the vitreouschamber 15 minutes before ischemic injury.

Animals were euthanatized 4 hours or 1, 3, 7, or 14 daysafter reperfusion. Eyes were removed, immersed in 10% for-malin for hematoxylin-eosin staining (n = 10 for condition),or fixed and processed for electron microscope study afterremoving the cornea and lens (n = 6 for condition). Anadditional group of eyes (n = 8-10 for each) were removedand immediately processed to isolate total RNA for reversetranscription-polymerase chain reaction (RT-PCR). For exper-iments with in situ hybridization histochemistry, eyes (n — 6for condition) were quickly frozen on dry ice after the superiorlimbus was marked. All animals were used in accordance withthe ARVO Statement for the Use of Animals in Ophthalmic andVision Research.

Quantitative Analysis of Retinal LesionEnucleated eyes were fixed in 2% paraformaldehyde plus 2%glutaraldehyde in phosphate-buffered saline (PBS) and post-fixed in 1% osmium tetroxide. Blocks of retina were embeddedin epoxy resin, cut at 1-jam thickness, and stained with 1%toluidine blue. The ischemic injury was manifested by thedecrease in retinal thickness. To determine neuronal survivor-ship in the retinal layers, the number of viable neurons wascounted in 100 X 25-ju,m squares (five squares randomly cho-sen in each rat; three rats per experimental condition) under-lying GCL or INL in retinal sections stained with toluidine blue.

Statistical analysis between groups was performed usinganalysis of variance and the Student-Newman-Keuls test.

Transmission Electron MicroscopyTransmission electron microscopy was performed as de-scribed.6 In brief, retinal tissues were fixed in Karnovsky'sfixative solution (2% paraformaldehyde, 2% glutaraldehyde, 2mM calcium chloride, and 100 mM cacodylate buffer [pH 7.4]).The tissues were washed with cacodylate buffer, postfixed in1% osmium tetroxide, and dehydrated in a series of gradedethanol. The tissues were then embedded in Araldite resin

(Taab Laboratory Equipment, Alder Mastron, UK). Ultrathinsections were mounted on grids, stained with uranyl acetateand lead citrate, and examined with an electron microscope(model 902A; Carl Zeiss, Oberkochen, Germany).

Reverse Transcriptase-Polymerase ChainReactionThe retina was removed from each eye immediately, 4 hours,1 day, and 3 days after elevation of IOP to 160 mm Hg to 180mm Hg for 90 minutes. Total RNA was isolated by a modifiedguanidine isothiocyanate-acid phenol method (Ultraspec-II;Biotecx Laboratories, Houston, TX). To prepare single-strandedcDNA, 5 ju,g total RNA was incubated in a reaction mixturecontaining oligo(dT)12"15; 1 mM each deoxyadenosine triphos-phate, deoxycytidine triphosphate, deoxyquanosine triphos-phate, and deoxythymidine triphosphate; 1 U//u,l RNAsin; and0.6 U/jid AMV reverse transcriptase at 42°C for 90 minutes. ThePCR reaction was performed using 1/100 of synthesized single-stranded cDNA as a template and specific sets of oligonucleo-tide primers for each gene. The annealing temperature wasadjusted to 58°C for p53 and 59°C for actin. The sequences ofprimer sets for each gene and the size of products are asfollows: p53, forward primer 5'-GCCATCTACAAGAAGTCA-CAGC-3' and reverse primer 5'-GATGATGGTAAGGATAG-GTCGG-3' (expected size, 362 bp); actin, forward primer5'-GATGATATCGCCGCGCTCGTCGTCGAC-3' and reverseprimer 5'-AGCCAGGTCCAGACGCAGGATGGCATG-3' (ex-pected size, 538-bp). After polymerization, the reaction mix-ture was subjected to electrophoresis on a 2% agarose gel.

In Situ HybridizationIn situ hybridization experiments using [35S] labeled cRNAprobes were performed as described previously.35

Probe SynthesisThe p53 cDNA was prepared by an RT-PCR reaction of totalRNA extracted from retinal tissues using the primers described.The sequence analysis of amplified cDNA verified a 362-bpproduct corresponding to part of exons 5 and 6 of rat p53. ThePCR product was inserted in the EcoRV site of pGEM-5zf(+).The antisense probes for p53 were labeled with [35S]uridinetriphosphate (UTP), using T7 polymerase.

Tissue Section and PrehybridizationWhole eyes were horizontally sectioned on a cryostat at—23°C, 10 jam-thick sections were thaw-mounted onto slidespretreated with gelatin, and sections were stored at —80°C.Sections were dried at room temperature, fixed in 3% parafor-maldehyde in 0.1 M PBS (pH 7.4) for 5 minutes at roomtemperature, and then acetylated for 10 minutes in 0.1 Mtriethanolamine (pH 8.0) with 0.25% acetic anhydride. Thesections were then dehydrated through a series of gradedethanol (70%, 80%, 95%, and 100%) and air dried.

Hybridization and AutoradiographySections were then hybridized with [35S]UTP-labeled cRNAprobes in a reaction buffer containing 40% deionized form-amide, 2X SSC, 1 mg/ml tRNA, 1 mg/ml sperm DNA, 10%dextran sulfate, 10 mM dithiothreitol, and IX Denhardt's so-lution at 50°C for 4 hours. During the hybridization, sectionswere covered with parafilm. Immediately after the hybridiza-

Page 3: Necrosis and Apoptosis after Retinal Ischemia: Involvement ...web.yonsei.ac.kr/neurolab/published/17.pdf · Necrosis and Apoptosis after Retinal Ischemia: Involvement of NMDA-Mediated

fOVS. March 1999. Vol. 40. No. Ischemic Neuronal Necrosis and Apoptosis 715

times after ischemia

FIGURE 1. Temporal pattern of degeneration caused by retinal ischemia. The retina was exposed to ischemic injury by increasing the IOP to arange of 160 mm Hg to 180 mm Hg for 90 minutes. (A) Bright-field photomicrographs of retinal sections stained with hematoxylin-eosin (a, b)or TUNEL (c, d) 1 day after a sham operation (a, c) or retinal ischemia (b, d). Note that the ischemic injury produced dark, shrunken, condensedneurons (arrows). (B) Bright-field photomicrographs of retinal sections immiinolabeled with anti-MAP-2 (a, b , arrows indicate MAP-2-positiveneurons) or anti-GFAP (c, d, arrows indicate GFAP-positive glia) 3 days after a sham operation (a, c) or retinal ischemia (b, d). (C) Quantitativeanalysis of retinal damage after ischemia. The number of viable neurons in a 100 X 25-/xm square overlying the GCL or the INL was analyzed afterstaining with toluidine blue (mean ± SEM; n = 15 fields randomly chosen from three rats per condition), scaled to the mean number of viableneurons in the corresponding retinal area after a sham operation (set at 100). 'Significant difference from relevant control (CTRL) at P < 0.05, usinganalysis of variance and the Student-Newman-Keuls test. IPL, inner plexiform layer; OPL, outer plexiform layer; ONL, outer nuclear layer. Bar, (A,B) 50 juLm.

tion reaction, sections were treated with 50% deionized form-amide at 52°C for 25 minutes and 100 ng/ml RNase at 37°C for

30 minutes. Sections were then dehydrated through a graded

ethanol series and exposed to film (XAR-5; Eastman Kodak,Rochester, NY) for 5 to 10 days. To obtain cellular resolution,sections were dipped in nuclear track emulsion (NTB-3; East-

Page 4: Necrosis and Apoptosis after Retinal Ischemia: Involvement ...web.yonsei.ac.kr/neurolab/published/17.pdf · Necrosis and Apoptosis after Retinal Ischemia: Involvement of NMDA-Mediated

716 Joo et al.

INL

IOVS, March 1999, Vol. 40, No. 3

INL

GCL ONL ONL

FIGURE 2. Occurrence of early neuronal necrosis and latent apoprosis after 90 minutes of retina ischemia.Electron photomicrographs of retinal sections prepared 4 hours (left), 1 day {middle), and 3 days (tight)after a 90-minute ischemic injury. Note that neurons in the INL and the GCL 4 hours after ischemia showednecrotic morphology marked by swelling of the cell body and mitochondria (mi), disintegration of theplasma membrane (pm), and scattering of nuclear chromatin with relatively conserved nuclear membrane(tun). Within 1 day after ischemia, some neurons in the inner nuclear layer (INL) and outer nuclear layer(ONL) underwent apoptosis (*) with cell body shrinkage, aggregation, and condensation of nuclearchromatin. The INL showed no further degeneration 3 days later, whereas a few neurons in the ONLshowed apoptosis-like degeneration. Bar, 5 jLLm.

man Kodak) diluted 1:1 with distilled water, developed 6 to 8weeks later, and connterstained -with hematoxylin-eosin. Us-ing this procedure, hybridization with sense cRNA probes orantisense cRNA probes after pretreatment of sections withRNase A revealed the level of background signal (data notshown). The levels of mRNA were analyzed on film autoradio-grams using a computerized image analysis procedure (Image-Quant 3.22; Molecular Dynamics, Sunnyvale, CA). Optical den-sity values were measured in a portion of retina near the opticnerve and then subtracted from background signal in the vit-reous body.

Immunohistochemistry and TdT-dXJTP Nick-EndLabelingThe retina was fixed in 2% paraformaldehyde for 2 hours,washed in PBS, dehydrated in alcohol, and embedded in par-affin. The paraffin-embedded retina was sectioned on a mic-rotome at room temperature at a thickness of 5 jam, deparaf-finized in xylene for 10 minutes, and rehydrated in a gradedalcohol series. For irnmunocytochemistry using antibodies spe-cific for microtubule associated protein-2 (anti-MAP-2) or glialfibrillary acidic protein (anti-GFAP), 40 /nm-thick vibratomesections were prepared. The retinal sections were incubated in2% H2O2 for 5 minutes and then in 20% normal horse serumcontaining 0.05% Triton X-100 for 1 hour. The sections werereacted with anti-p53 at room temperature for 2 hours or

anti-MAP-2 or anti-GFAP at 4°C for 10 to 12 hours. The sectionswere then incubated in biotinylated goat anti-mouse IgG for 10minutes and ABC reagents (Vectastain; Vector, Burlingame,CA) for 10 minutes and then reacted with 3-amino-9-ethylcaba-zole and 0.01% H2O2 (HRP/AEC; Laboratory Vision, Fremont,CA). The immunolabeled sections were counter-stained withhematoxylin (for p53) or observed under a microscope with adifferential interference contrast filter (MAP-2 and GFAP).

For TdT-dUTP nick-end labeling (TUNEL) staining, theparaffin-embedded sections were treated with 20 fig/ml pro-teinase K and incubated in 3% H2O2 for 5 minutes. As previ-ously reported,18 DNA was end-labeled at 3' using biotinylateddUTP and terminal deoxyuridyltransferase in a reaction buffercontaining 30 mM Tris, 140 mM sodium cacodylate, and 1 mMcobalt chloride at 37°C for 2 hours. The signal was visualizedwith an ABC staining kit (TACS apoptotic DNA laddering kit,Trevigen, Gaithersburg, MD) using diaminobenzidine as a sub-strate.

RESULTS

Temporal Degeneration of Retinal Neurons afterRetinal IschemiaElevation of the intraocular pressure to 160 mm Hg to 180 mmHg for 40 to 100 minutes resulted in widespread neuronal

Page 5: Necrosis and Apoptosis after Retinal Ischemia: Involvement ...web.yonsei.ac.kr/neurolab/published/17.pdf · Necrosis and Apoptosis after Retinal Ischemia: Involvement of NMDA-Mediated

IOVS, March 1999, Vol. 40, No. 3 Ischemic Neuronal Necrosis and Apoptosis 717

OvP

100

80 -coo

CD

2 60CDQ.O

E 40JZCO

20 -

0

I

I

6CTRL MK MK(multi)

GCL

i4CTRL MK MK(multi)

INLFIGURE 3. The neuroprotective effect of MK-801 against retinal ischemia. Rats underwent 90 minutes ofischemia 15 minutes after intravitreal injections of 10 ju-1 saline (CTRL) or 5 mM MK-801 (MK). MK-801-treated rats also received additional injections of 5 mM MK-801 24 and 48 hours after ischemia (MK-[multi]). Neuronal death in the GCL and INL was analyzed 4, 24, and 72 hours later by counting viableneurons in 100 X 25-ixm square (mean ± SEM; n = 15 fields randomly chosen from three rats percondition), as described in Figure 1. 'Significant difference from relevant control in viable neurons (P <0.05; analysis of variance and Student-Newman-Keuls test).

degeneration throughout the retina over the 1 to 3 days afterischemia. For the present study, the 90-minute retinal ischemiaparadigm was chosen to produce reproducible moderate neu-ronal injury manifested by a decrease in retinal thickness.Neurons in the GCL and INL showed pyknotic nuclei and DNAfragmentation revealed by the TUNEL stain within 1 day afterischemia (Fig. 1A). This neuronal injury was confirmed by amarked decrease in MAP-2 immunoreactivity throughovit thecell bodies and processes in the INL, inner plexiform layer, andGCL 3 days later, whereas glial cells immunoreactive to GFAPwere not decreased in number, but rather increased in inten-sity (Fig. IB). Neurons in the INL and GCL underwent rapiddegeneration by 30% to 40% within 4 hours after retinal isch-emia (Fig. 1C). The number of degenerating neurons wasfurther increased in the retinal layers. Three days after isch-emia, surviving neurons in the GCL and INL were approxi-mately 40% in the sham-operation control (Fig. 1C). The retinalthickness gradually decreased over 7 days after ischemia. Atthis point, the INL and GCL neurons were intermingled but didnot show further degeneration.

Necrosis and Apoptosis of Retinal Neurons afterIschemia

Electron microscopy showed that most of the degeneratingneurons in the INL and GCL 4 hours after retinal ischemiaaccompanied swelling of the cell body and mitochondria, a

marked collapse of the plasma membrane with preservation ofthe nuclear membrane, and irregular scattering of nuclearchromatin (Fig. 2). All these effects are typically seen in thecourse of necrosis.16'22 The necrotic neurons were rarely de-tectable 1 day after ischemia. Instead, neurons in the INL andGCL showed apoptosis-like morphologic changes character-ized by aggregation and condensation of nuclear chromatinand cell body shrinkage 1 day later. Neurons in the outernuclear layer also underwent apoptotic degeneration graduallyover 1 to 7 days after ischemia, although it was sparselydistributed.

Effect of MK-801 against Ischemic RetinalDegeneration

Experiments were performed to study the possibility that theNMDA receptor antagonist MK-801 would protect against thenecrosis component of neuronal death in ischemia.18 Theintravitreal injections of MK-801 substantially attenuated thenecrosis of neurons appearing in the GCL and INL 4 hours afterretinal ischemia (Fig. 3). The neuroprotective effect of MK-801lasted over the next 1 to 3 days, but neurons in die GCL andINL continued to undergo extensive degeneration over time.Repeated injections of MK-801 eveiy 24 hours after ischemiadid not show additional protection. This implies that the de-layed neuronal apoptosis in retina occurs in an NMDA-indepen-dent fashion.

Page 6: Necrosis and Apoptosis after Retinal Ischemia: Involvement ...web.yonsei.ac.kr/neurolab/published/17.pdf · Necrosis and Apoptosis after Retinal Ischemia: Involvement of NMDA-Mediated

718 Joo et al.

Upregulation of p53 mRNA and Protein

The expression pattern of mRNA encoding p53 or actin wasanalyzed at 4 hours, and 1, 3, and 7 days after retinal ischemia.RT-PCR experiments showed that levels of actin mRNA in theretina remained unchanged at any time tested after ischemia.Retinal ischemia did not change the mRNA expression of p53over the next 4 hours. The levels of p53 mRNA were signifi-cantly increased 1 day after ischemia (Fig. 4). Thereafter, theischemia-induced expression of p53 mRNA seemed to declinebut was still evident 3 days later. The temporal induction ofp53 mRNA expression after ischemia was confirmed by in situhybridization histochemistry using [35S]-labeled antisensecRNA probes (Figs. 4B, 4C, 4D, 4E, 4F). p53 mRNA expressionin the retina returned to the control level within 7 days afterischemia. In emulsion-dipped sections after hybridization withp53 cRNA probe, an increase in the p53 mRNA hybridizationsignal was observed exclusively in neurons overlying the GCLand INL 1 day after ischemia (Fig. 5B). Immunoreactivity to p53also increased in the GCL and INL neurons 1 day after ischemia(Fig. 5D), which suggests that retinal ischemia increases levelsof p53 protein and mRNA.

IOVS, March 1999, Vol. 40, No. 3

0 4 24 72 (hr)

DISCUSSION

The present study shows that retinal neurons undergo twomorphologically distinct patterns of degeneration after isch-emic injuries. One evolves rapidly in the INL and GCL within 4hours, marked by prominent necrosis and prevented by block-ing NMDA receptors. The second phase of neuronal death wasobserved through all retinal layers over the next 1 to 3 days.The slowly evolving death was characterized by shrinkage ofthe cell body, aggregated condensation of nuclear chromatin,and expression of p53 primarily in the GCL and INL.

Excitotoxicity has been well documented as a primarycause of neuronal death in ischemia in the retina and thebrain.23 Glutamate antagonists protect retinal neurons againstischemic insult in vivo and in vitro.30'32'36'37 Neurons treatedwith excitotoxins show the typical morphology of necrosisaccompanied by cell body swelling, early disruption of plasmamembrane, and irregularly scattered condensation of nuclearchromatin.4'6'7 The overall characteristics of excitotoxicitywere observed in the course of the rapidly evolving necrosis ofthe GCL and INL neurons after retinal ischemia. In addition, theadministration of MK-801 substantially prevented the necroticdegeneration of retinal neurons induced by ischemia. Thepresent findings support the possibility that retinal ischemiaproduces necrosis of the GCL and INL neurons in part throughthe activation of NMDA receptors.

Although necrosis is a predominant form of neuronatdeath in ischemia, accumulating evidence supports apoptosisas an additional pattern of death. For example, internucleoso-mal DNA fragmentation, chromatin condensation, and cyclo-heximide-sensitive death have been observed in degeneratingareas after ischemic insults in brain and spinal cord.38 Mixedpatterns of necrosis and apoptosis have been shown in GCLand INL neurons after retinal ischemia.33 We also found thatretinal ischemia produced slowly evolving neuronal apoptosissubsequent to the rapid onset of necrosis in the INL and GCLneurons. In addition, apoptotic neurons were sparsely andcontinually observed in the outer nuclear layer at late timepoints after ischemia.

300 •,

CTRL 4 hr 1 d 3d

times after ischemia

FIGURE 4. Temporal induction of p53 mRNA expression after retinalischemia. (A) Total RNA was examined from the retina 4, 24, and 72hours after 90 minutes of retinal ischemia or sham operation. Levels ofmRNA encoding p53 and actin were analyzed using RT-PCR adjusted to28 cycles. Note that levels of p53 mRNA were increased 24 and 72hours after the ischemia. (B, C, D, E, F) In situ hybridization histo-chemistry showing the expression of p53 mRNA in the rat eye 4 hours(C), and 1 (D), 3 (E), and 7 (F) days after 90 minutes of ischemia. Thep53 hybridization signals were taken from autoradiogram films ex-posed for 5 days. Compared with levels in sham-operation control (B),levels of p53 mRNA were markedly increased in the retina 1 to 3 daysafter the ischemia. (G) Quantification of p53 mRNA levels. The densityof the p53 mRNA hybridization signals was measured in retina(mean ± SEM; n = 3-6 retinal sections chosen from two to three ratsper group), scaled to the mean value of the retinal p53 mRNA signalsfrom sham-operation control groups (100%). 'Significant differencefrom sham-operation control (CTRL; P < 0.05; analysis of variance andStudent-Newman-Keuls test). R, retina; C, cornea. Bar, 2 mm.

Even though the two different types of death, apoptosisand necrosis, both contribute to ischemic injury, the underly-ing mechanisms remain to be delineated. Recently, neurons

Page 7: Necrosis and Apoptosis after Retinal Ischemia: Involvement ...web.yonsei.ac.kr/neurolab/published/17.pdf · Necrosis and Apoptosis after Retinal Ischemia: Involvement of NMDA-Mediated

IOVS, March 1999, Vol. 40, No. 3 Ischemic Neuronal Necrosis and Apoptosis 719

FIGURE 5. Concurrent induction of p53 mRNA and protein afterretinal ischemia. Retinal sections prepared 1 day after a sham operation(A, B) or 90 minutes of ischemia (C, D) were hybridized with a[J5S]-labeled antisense cRNA probe for p53 (A, C) or immunolabeledwith an antl-p53 antibody (B, D). (A, C) Dark-field photomicrographsof the emulsion-dipped sections showing ischemia-induced increase inp53 mRNA hybridization signal in retina (arrows). (B, D) Bright-fieldphotomicrographs of retinal sections showing ischemia-induced in-crease in p53 immunoreactivity (arrows). Bar, 50 jum.

deprived of oxygen and glucose have been shown to undergoslowly evolving and cycloheximide-sensitive apoptosis whenexcitotoxic necrosis is blocked,18 indicating that ischemic ap-optosis may occur through nonexcitotoxic mechanisms. Thisis supported by the present results showing that the intravitrealinjections of MK-801 attenuate the early neuronal necrosiswithout interfering with the slowly evolving neuronal death inretina after ischemia. Parallel occurrence of excitotoxicity andapoptosis has been observed in cortical areas after focal cere-bral ischemia.161'7 Expression of c-Jun, Bax, and cyclooxygen-ase-2 is increased in degenerating neurons after brain ischemia,which can contribute to the process of neuronal apopto-sis. 39"' i l Mitochondria] damage, accumulation of reactive oxy-gen species, and activation of proteases such as caspases havebeen proposed as key transduction pathways leading to neu-ronal apoptosis in ischemia."'2'43 Although these molecular andcellular events are thought to play a role in the process ofischemic neuronal degeneration in the retina, further study isneeded to determine whether they are specifically required forexecution of apoptosis, not excitotoxic necrosis, after retinalischemia.

The p53 tumor suppressor gene is upregulated in neuronsvulnerable to brain ischemia or adrenalectomy.23'24'44 In thepresent study, retinal ischemia did not alter expression of p53within 4 hours when necrosis was the dominant form ofneuronal death in the retina. In fact, levels of p53 mRNA andprotein were substantially increased in the INL and GCL areasundergoing widespread neuronal apoptosis over the 1 to 3

days after retinal ischemia. Increase in p53 expression does notnecessarily mean causality in the delayed neuronal apoptosis,but the expression pattern in the apoptosis zone probablycontributes to the progress of neuronal apoptosis in retinalischemia. In support of this, neuronal loss in ischemia in theretina and brain was reduced in transgenic mice deficient inp53-26'45 Neuronal death dependent on p53 has been observedin the hippocampus after seizure.28 These findings suggest thatp53 plays a dynamic role in the process of apoptotic neuronaldegeneration after brain injuries such as ischemia or seizure.

The overexpression of p53 has been shown to downregu-late Bcl-2 expression and to upregulate Bax expression invitro.46 The p53-depenclent increase in Bax protein was alsoobserved in the degenerating cortical and hippocampal neu-rons.47 The transgenic expression of Bcl-2 reduces the size ofischemic brain lesion in mice and rats.'iw'iy It is conceivablethat the Bcl-2 family is coupled to downstream events in thep53-mediated apoptosis pathway after ischemia.

The notion that necrosis is the sole pattern of ischemicneuronal death has been challenged by a flood of evidenceshowing apoptosis after ischemic insults. Our findings suggestthat retinal ischemia causes the fulminant neuronal necrosis ofthe GCL and INL neurons within several hours, in part throughthe activation of NMDA receptors. In contrast, apoptosis ap-pears as the latent component of neuronal death in the sameretinal areas. Cooperative strategy to counteract apoptosis(e.g., modulation of p53 expression) and excitotoxic necrosis(e.g., block of NMDA neurotoxicity) should be contemplatedfor efficient protection of neurons against ischemia.

Acknowledgments

The authors thank Ji-Eun Kim for help with preparation of the manu-script.

References

1. Choi DW. Glutamate neurotoxicity and diseases of the nervoussystem. Neuron. 1988;]:623-634.

2. Olney JW. The toxic effects of glutamate and related compoundsin the retina and the brain. Retina. 1982;2:341-359-

3. Choi DW. Ionic dependence of glutamate neurotoxicity. J Neuro-sci, 1987;7:369-379.

4. Gwag BJ, KohJY, DemaroJ, Ying HS, Jacquiti MF, Choi DW. Slowlytriggered excitotoxicity occurs by necrosis in cortical cultures.Neuroscience. 1997 ;77:393 - 401.

5. KerrJFK, Harmon BV. The molecular basis of cell death. In: TomeiLD, Cape FO, eds. Apoptosis. New York: Cold Spring HarborLaboratory 1991:5-29.

6. Sohn S, Kim EY, Gwag BJ. Glutamate neurotoxicity in mousecortical neurons: atypical necrosis with DNA ladders and chroma-tin condensation. Neurosci Lett. 1998;240:l-4.

7. Dessi F. Charriaut-Marlangue C, Khrestchatisky M. Ben-Ari Y. Glu-tamate-induced neuronal death is not a programmed cell death incerebellar cultures./Neurocbem. 1993;6O:1953-1955.

8. Koh JY, Cotman CW. Programmed cell death: its possible contri-bution to neurotoxicity mediated by calcium channel antagonists.Brain Res. 1992;587:233-240.

9. Goldberg MP, Weiss JH, Pham PC, Choi DW. Ar-methyl-i>aspartatereceptors mediate hypoxic neuronal injury in cortical culture./ Pharmacol Exp Tber. 1987;243:784 -791.

10. Sheardown MJ, Nielsen EO, Hansen AJ, Jacobsen P, Honore T.2,3-Dihydroxy-6-nitro-7-sulfamoyl-benzo(F)quinoza]ine: a neuro-protectant for cerebral ischemia. Science. 1990;247:571-574.

11. Simon RP, Swan JH, Griffith T, Meldrum BS. Blockade of W-methyl-i>aspartate receptors may protect against ischemic damage in thebrain. Science. 1984;226;850-852.

Page 8: Necrosis and Apoptosis after Retinal Ischemia: Involvement ...web.yonsei.ac.kr/neurolab/published/17.pdf · Necrosis and Apoptosis after Retinal Ischemia: Involvement of NMDA-Mediated

720 Joo et al. IOVS, March 1999, Vol. 40, No. 3

12. Heron A, Pollard H, Dessi F, et al. Regional variability in DNAfragmentation after global ischemia evidenced by combined histo-logical and gel electrophoresis observations in rat brain. ] Neuro-chem. 1993:61:1973-1976.

13- Linnik MD, Zobrist RH, Hatfield MD. Evidence supporting a rolefor programmed cell death in focal cerebral ischemia in rats.Stroke. 1993;24:2002-2009.

14. MacManus JP, Buchan AM, Hill IE, Rasquinha I, Preston E. Globalischemia can cause DNA fragmentation indicative of apoptosis inrat brain. Neurosci Lett. 1993;l64:89-92.

15. Shigeno T, Yamasaki Y, Kato G, et al. Reduction of delayed neu-ronal death by inhibition of protein synthesis. Neurosci Lett. 1990;120:117-119.

16. Charriaut-Marlangue C, Margaill I, Represa A, Popovici T, PlotkineM, Ben-Ari Y. Apoptosis and necrosis after reversible focalischemia: an in situ DNA fragmentation analysis. / Cereb BloodFlowMetab. 1996;l6:186-194.

17. Du C, Hu R, Csernansky CA, Liu XZ, Hsu CY, Choi DW. Additiveneuroprotective effects of dextrophan and cycloheximide in ratssubjected to transient focal cerebral ischemia. Brain Res. 1996;718:233-236.

18. Gwag BJ, Lobner D, Koh JY, Wie MB, Choi DW. Blockade ofglutamate receptors unmasks neuronal apoptosis after oxygen-glucose deprivation in vitro. Neuroscience. 1995;68:6l5-6l9.

19- Clarke AR, Purdie CA, Harrison DJ, et al. Thymocyte apoptosisinduced by p53-dependent and independent pathways. Nature.1993:362:849-852.

20. Lowe SW, Schmitt EM, Smith SW, Osborne BA, Jacks T. p53 isrequired for radiation-induced apoptosis in mouse thymocytes.Nature. 1993:362:847-849.

21. Cirielli C, Riccioni T, Yang C, et al. Adenovirus-mediated genetransfer of wild-type p53 results in melanoma cell apoptosis invitro and in vivo. IntJ Cancer. 1995:63:673-679.

22. Yonish-Rouach E, Resnitzky D, Lotem J, Sachs L, Kimchi A, OrenM. Wild-type p53 induces apoptosis of myeloid leukaemic cellsthat is inhibited by interleukin-6. Nature. 1991:352:345-347.

23. Chopp M, Li Y, Zhang ZG, Freytag SO. p53 expression in brainafter middle cerebral artery occlusion in the rat. Biochem BiopbysRes Commun. 1992;182:1201-1207.

24. Li Y, Chopp M, Zhang ZG, Zaloga C, Niewenhuis L, Gautam S. p53-immunoreactive protein and p53 mRNA expression after transient mid-dle cerebral artery occlusion in rats. Stroke. 1994;25:849 - 856.

25. Sakhi S, Bruce A, Sun N, Tocco G, Baudry M, Schreiber SS. p53induction is associated with neuronal damage in the central ner-vous system. Proc Natl Acad Sci USA. 1994;91:7525-7529.

26. Crumrine RC, Thomas AL, Morgan PF. Attenuation of p53 expres-sion protects against focal ischemic damage in transgenic mice.J Cereb Blood Flow Metab. 1994;l4:887-891.

27. Enokido Y, Araki T, Tanaka K, Aizawa S, Hatanaka H. Involvementof p53 in DNA strand break-induced apoptosis in postmitotic CNSneurons. Eur J Neurosci. 1996;8:1812-1821.

28. Morrison RS, Wenzel HJ, Kinoshita Y, Robbins CA, Donehower LA,Schwartzkroin PA. Loss of the p53 tumor suppressor gene protectsneurons from kainate-induced cell death. / Neurosci. 1996; 16:1337-1345.

29. Sakhi S, Gilmore W, Tran ND, Schreiber SS. p53-deficient mice areprotected against adrenalectomy-induced apoptosis. Neuroreport.1996;8:233-235.

30. El-Asrar AM, Morse PH, Maimone D, Torczynski E, Reder AT.MK-801 protects retinal neurons from hypoxia and the toxicity ofglutamate and aspartate. Invest Ophthalmol Vis Sci. 1992;33:3436-3438.

31. Mey J, Thanos S. Intravitreal injections of neurotrophic factorssupport the survival of axotomized retinal ganglion cells in adultrats in vivo. Brain Res. 1993;602:304-317.

32. Yoon YH, Marmor MF. Dextromethorphan protects retinaagainst ischemic injury in vivo. Arch Ophthalmol. 1989; 107:409-411.

33. Buchi ER. Cell Death in the rat retina after a pressure-inducedischemia-reperfusion insult: an electron microscopic study, I: Gan-glion cell layer and inner nuclear layer. Exp Eye Res. 1992;55:605-613.

34. Joo CK, Park KY, Park MS, Gwag BJ, Choun MH. Occurrence ofneuronal necrosis and apoptosis following the retinal ischemia:induction of p53 and bcl-2 mRNA. Soc Neurosci Abstr. 1996;26:464.15.

35. Gwag BJ, Sessler FM, Waterhouse BD, Springer JE. Regulation ofnerve growth factor mRNA in the hippocampal formation effectsof A -̂methyl-D-aspartate receptor activation. Exp Neurol. 1993;121:160-171.

36. Sucher NJ, Aizenman E, Lipton SA. ./V-methyl-D-aspartate antago-nists prevent kainate neurotoxicity in rat retinal ganglion cells invitro. / Neurosci. 1991; 11:966 -971.

37. Romano C, Price MT, Almli T, Olney JW. Excitotoxic neurodegen-eration induced by deprivation of oxygen and glucose in isolatedretina. Invest Ophthalmol Vis Sci. 1998;39:4l6-423.

38. Choi DW. Ischemia-induced neuronal apoptosis. Curr Opin Neu-robiol. 1996;6:667-672.

39. Chen J, Zhu RL, Nakayama M, et al. Expression of the apoptosis-effector gene, Bax, is up-regulated in vulnerable hippocampalCA1 neurons following global ischemia. JNeurochem. 1996;67:64-71.

40. McGahan L, Hakim AM, Robertson GS. Hippocampal myc and p53expression following transient global ischemia. Mol Brain Res.1998;56:133-145.

41. Nakayama M, Uchimura K, Zhu RL, et al. Cyclooxygenase-2 inhi-bition prevents delayed death of CA1 hippocampal neurons fol-lowing global ischemia. Proc Natl Acad Sci. USA. 1998;95:10954 -10959.

42. Gottron FJ, Ying HS, Choi DW. Caspase inhibition selectivelyreduces the apoptotic component of oxygen-glucose deprivation-induced cortical neuronal cell death. Mol Cell Neurosci. 1997;9:159-169.

43. Hara H, Friedlander RM, Gagliardini V, et al. Inhibition of interleu-kin 1/3 converting enzyme family proteases reduces ischemic andexcitotoxic neuronal damage. Proc Natl Acad Sci USA. 1997;94:2007-2012.

44. Schreiber SS, Sakhi S, Dugich-Djodjevic MM, Nichols NR. Tu-mor suppressor p53 induction and DNA damage in hippocam-pal granule cells after adrenalectomy. Exp Neurol. 1994; 130:368-376.

45. Rosenbaum DM, Rosenbaum PS, Aggarwal A, Kessler JA. Trans-genic mice deficient in the p53 protein are resistant to transientretinal ischemia. Soc Neurosci Abstr. 1996;26:464.13.

46. Miyashita T, Krajewski S, Krajewska M, et al. Tumor suppressorp53 is a regulator of bcl-2 and bax gene expression in vitro and invivo. Oncogene. 1994;9:1799-1805.

47. Xiang H, Kinoshita Y, Knudson MC, Korsmeyer SJ, SchwartzkroinPA, Morrison RS. Bax involvement in p53-mediated neuronal cellDeath./Neurosci. In press.

48. Lawrence MS, McLaughlin JR, Sun GH, et al. Herpes simplex viralvectors expressing Bcl-2 are neuroprotective when delivered aftera stroke. / Cereb Blood Flow Metab. 1997r;17:~'40-7'44.

49. Martinou JC, Dubois-Dauphin M, Staple JK, et al. Overexpressionof BCL-2 in transgenic mice protects neurons from naturally oc-curring cell death and experimental ischemia. Neuron. 1994; 13:1017-1030.