<<

Systemic L- Administration Partially Protects Against NMDA, But Not Kainate-Induced Degeneration of Retinal Ganglion Cells, and Reduces Visual Discrimination Deficits in Adult Rats

Christian K. Vonuerk* MichaelR. Kreutz* Evan B. Dreyer,-f and Bernhard A. Sabel*

Purpose. Kynurenic (KYNA), an endogenous metabolite, is an N-methyl-D- aspartate (NMDA) antagonist active at the -binding site of the NMDA-receptor com- plex. The authors investigated whether systemic administration of a biochemical precursor of KYNA, L-kynurenine (L-Kyn), could block NMDA- or (KA)-induced cell death in adult rat retinal ganglion cells (RGCs) and protect NMDA-treated animals from lesion- induced visual deficits. Methods. Rats were injected with 20-nmol NMDA or 5-nmol KA intraocularly. To quantify the number of surviving RGCs, the retrograde tracer horseradish-peroxidase was injected into the superior colliculus contralateral to the lesioned eye. Surviving RGCs were counted on wholemounted retinae in a centroperipheral gradient, as well as in the four quadrants, using a computer-assisted image analysis system. Results. The NMDA-injections resulted in an approximately 82% RGC loss in the adult rat retina compared with control retinae and a cell loss of approximately 50% in KA-treated retinae. Pretreatment with L-Kyn significantly reduced NMDA-induced RGC degeneration to values of approximately 60%, but KA toxicity was not significantly affected by L-Kyn pretreat- ment. Intraocular injections of NMDA resulted in an impairment of visual discrimination behavior, which partially recovered within a period of approximately 3 weeks. However, when treated systemically with L-Kyn, brightness discrimination was significantly improved as com- pared with NMDA-treated rats. Conclusions. These findings show that systemic administration of L-Kyn in adult rats can block NMDA-induced retinal ganglion cell death in vivo and preserves brightness discrimination performance. Invest Ophthalmol Vis Sci. 1996; 37:2382-2392.

•Secondary cell loss due to damage after stroke central nervous system. Excitatory amino , such or trauma is not only the result of mechanical tissue as glutamate, have been suggested to act as a major damage but also due to excitotoxic insults. Activation for most classes of in the of excitatory receptors has been postulated retina.6"8 Lucas and Newhouse9 first established that to play a key role in stroke, degenerative disorders, glutamate is a potent toxin for neurons in the inner epilepsy, ischemia, and hypoglycemia.1"5 retina. In the past decade, various 10 The retina is the most accessible portion of the subtypes have been reported in retinal cell types, and specific glutamate receptor were shown to induce distinct lesions. Moreover, it was shown that From the * Institute, of Medical Psychology, Medical Faculty, Olto-von-Cumcke- the mammalian retina is highly susceptible to N- UniversUy of Magdeburg, Magdeburg, Cennany; and the f Department of methyl-D-aspartate (NMDA) toxicity.""14 The NMDA- Ophthalmology, Massachusetts Eye and Ear Infirmary and Harvard Medical School, boston, Massachusetts. induced neuronal degeneration was found to be dose Supported by a grant from the Deutsche Forschungsgemeinschaft (Sa 433/3-1). dependent and mainly restricted to the innermost reti- Submitted for publication February 20, 1996; revised July S, 1996; accepted July 9, 1996. nal layers, whereas more distal layers were less af- Proprietary interest category: N. fected.13 This pattern closely resembles the cell loss Reprint requests: H. A. Sabel, Institute of Medical Psychology, Medical Faculty, Otto- von-Cuericke University, Uipuger Sir. 44, 39120 Magdeburg, Germany. that is observed in several retinal disease states such

Investigative Ophthalmology & Vislal Sci. c:, November 1996, Vol. 37. No. 12 2382 Copyright © Association for Resea ch in ' on and Ophthalmology

Downloaded from iovs.arvojournals.org on 09/28/2021 Systemic Htynurenine Protects Against 2383

as ischemia14 and glaucoma.lllD Furthermore, it was Kyn.22 This is of pharmacologic interest because it has proposed that retinal glutamate neurotoxicity is been shown that KYNA is able to block excitotoxin caused predominantly by overactivation of NMDA re- lesions created in vivo during development.20 ceptors, pointing to their crucial role in retinal degen- erative disorders.16 Retinal lesions induced by exposure to kainic acid MATERIALS AND METHODS 1718 (KA) were first characterized in the chicken. Kai- Treatments nic acid is a powerful excitotoxin when injected intra- ocularly and causes substantial cell death in the inner All animal experiments were performed in accor- nuclear layer while essentially sparing photoreceptor dance with the standards of the Association for Re- and ganglion cells.13'819 Hence, retinal KA excitotoxi- search in Vision and Ophthalmology. Hooded rats city is associated with a marked decrease in (300 to 350 g; PVG-Mol, Moellegard, Denmark) were acetyltransferase activity, indicating that housed two animals per cage (wooden bedding) on a amacrine cells are the most vulnerable cell type to a 12:12 hour, dark-light cycle, food and water available KA insult.20 However, because the conclusions that ad libitum. Animals underwent surgery under general retinal ganglion cell (RGCs) are less susceptible to KA anesthesia using or 7% intraperitoneal toxicity were based on the estimation of cell size and chloralhydrate injection. Intraocular injections were not on retrograde labeling of RGC, these data have performed with a heat-pulled glass pipette connected to be treated with caution. Furthermore, prelabeled to a microsyringe (model 105; Drummond Microdis- RGC in vitro are susceptible to KA neurotoxicity by a penser, Life Sciences International GmbH, Frankfurt, mechanism that seems to involve NMDA receptors Germany). All intraocular injections were made and the presence of an endogenous glutamate recep- through the dorsal limbus, each over a period of ap- tor .21 proximately 1 minute. The NMDA and KA were dis- solved in phosphate-buffered saline and administrated Efforts to develop new pharmacologic treatment at a dose of 20 nmol in 2 /J,\ for each eye. The L-Kyn have focused on the development of glutamate recep- sulfate (Sigma-Aldrich Chemie GmbH, Diesenho- tor antagonists to prevent neuronal cell loss and de- fen, Germany) was dissolved in saline and then ad- generation. Further progress has partly been ham- justed to pH 7.4. All animals were injected with three pered by the lack of adequate in vivo models, where single doses of L-Kyn (100 mg/kg): 120 minutes be- the functional consequences of drug interventions can fore, 30 minutes before, and 30 minutes after NMDA be tested under appropriate conditions. The laminar (n = 8) or KA (n = 8) application. Control subjects organization of retinal neurons and their well-studied {n = 5) received three single injections of saline intra- synaptic connections provide excellent preconditions peritoneally and 2 pA phosphate-buffered saline intra- for studies on functional consequences of excitotoxi- ocularly. In some animals, the effect of L-Kyn on RGC city. number also was examined without subsequent excito- In the present study, we investigated the effect of toxin application. NMDA and KA excitotoxicity on retrogradely labeled ganglion cells in retinal whole mount preparations. In addition, we studied whether a pretreatment with Assessment of Retrograde Horseradish- systemic L-kynurenine (L-Kyn) attenuates NMDA-in- Peroxidase Transport duced excitotoxicity in the retina and whether such a To determine the number ofsundving RGCs, 48 hours neuroprotective effect has functional consequences as after intraocular injection, anesthetized rats were shown by visual discrimination performance. L-Kyn is placed in a stereotaxic frame (Stoelting, Woodale, IL). an endogenous metabolite of tryptophan A portion of the skull was removed, exposing the neo- that passes the blood-brain barrier and is metabolized cortex overlying the tectum on the side contralateral subsequently to (KYNA), which, in to the lesioned eye. Seven intracerebral injections of turn, acts as a competitive antagonist at the glycine 0.7 fi\ horseradish peroxidase each (HRP; Boehringer binding site of the NMDA receptor complex.22 The Mannheim, Oberkochen, Germany; 30% [wt/vol] precursor L-Kyn is transported selectively across the HRP dissolved in 2% [vol/vol] dimethyl sulfoxide) blood-brain barrier by the large neutral amino acid were made targeting all layers of the contralateral su- carrier system,23 and the subsequent synthesis of KYNA perior colliculus. Each injection was made over a dura- is catalyzed by kynurenine aminotransferase. This en- tion of 2 minutes. Forty-eight hours after HRP applica- zyme has been found in both rat and human brain24'23 tion, the rats were processed to obtain retinal whole and is thought to be localized in glial cells within the mounts. central nervous system.2<3"2S Extracellular KYNA levels The protocol for the preparation of retinal whole in the central nervous system can be increased dramat- mounts was similar to that first described by Perry ically by peripheral administration of its precursor, L- and Linden.30 Briefly, rats were given a lethal dose of

Downloaded from iovs.arvojournals.org on 09/28/2021 2384 Investigative Ophthalmology & Visual Science, November 1996, Vol. 37, No. 12

chloral hydrate followed by transcardial perfusion with ter et al.33 Briefly, rats learned to discriminate between 0.9% saline for 2 to 3 minutes, followed by 1.25% two visual stimuli (i.e., patterns of gray surface versus paraformaldehyde in 0.1-M phosphate-buffered saline vertical flickering bars of comparable illumination). and 1.25% glutaraldehyde for 1 to 2 minutes. An ori- The images were presented on a television screen at entation mark was placed in the dorsal margin of the the rear of a two-choice maze. The sequence of stimu- injected eye with a hot needle. The eyes were removed lus presentation on each side of the maze was deter- and placed in 1.25% paraformaldehyde in 0.1 M phos- mined by pseudo-random series. Rats were water de- phate-buffered saline for subsequent processing. The prived (85% to 90% of normal body weight) and retina then was dissected in 1.25% paraformaldehyde trained over a 3-week period to perform at a level of at 4°C, and four radial cuts were made into the retina at least 90% correct choices on three consecutive test to facilitate flattening of the tissue for placement on sessions over a 3-week period. After being trained, rats a glass slide. To determine shrinkage, the outline of were given bilateral intraocular NMDA-injection with the retina was immediately drawn thereafter at low or without L-Kyn pretreatment. Behavioral assessment magnification through a dissecting microscope con- started after a 72-hour recovery period. A trial was nected to a drawing tube before being coverslipped. initiated by placing a dark-adapted rat into the start This permitted the evaluation of retinal shrinkage dur- box. After 5 to 10 seconds, the first (opaque) of the ing the staining and mounting process. The histologic two guillotine doors was opened. At this time, the rats reaction for HRP was carried out with freely floating could see the two different stimuli screens at the rear retinae according to the protocol described by Hanker end of the goal box. After another 2 seconds, die et al31 as modified by Perry and Linden.30 Thereafter, second (translucent) guillotine door was opened, and the retinae were washed in water, transferred onto the rats were allowed to run toward the screens. A a gelatinized slide, dehydrated, and embedded with correct choice (vertical flickering bars) was rewarded DePeX (SERVA Feinbiochemica GmbH, Heidelberg, by a few drops of water. When a rat chose the wrong Germany). To evaluate the correct placement of the side (gray screen), it had no access to water. Rats were tectal HRP injection, the brain was removed after fur- given 30 trials a day for 21 days. ther perfusion for 5 to 10 minutes, postfixed for sev- eral hours, and transferred to 30% sucrose, and as Statistical Analysis soon as equilibrium was reached, coronal sections were cut on a freezing microtome. The tissue then Statistical analyses were carried out using the SPSS was reacted for HRP as described elsewhere.30 Frozen Software package in conjunction with a personal com- sections were used to assess the placement of HRP puter. All data were expressed in average cells per injections into the superior colliculus. square millimeter and analyzed by analysis of variance and appropriate group comparisons. The HRP-positive cells were counted in the whole mounted and stained retinae by using a computer- based image analysis system (SIS, Muenster, Ger- RESULTS many). Cell counting of the retina was performed at 48 sampling points, and cell density was estimated us- Retinal Ganglion Cell Counts ing a standardized protocol described previously by Representative photomicrographs of HRP-labeled Sautter and Sabel.32 Briefly, the retinae were drawn at RGCs in retinal whole mounts after the different treat- low magnification under a light microscope, and an ments are shown in Figure 1. The average estimates overlay of concentric circles centered on the optic of the number of retrogradely labeled RGC in control nerve heads was added. These circles had radii of 25%, retinae were 115,500 (1650 ± 27 cells/mm2; n = 5), 40%, 55%, 70%, and 85% of the mean radius of each which is in accordance with previously published retina. Thereafter, each circle was subdivided into data.32'34'35 A density gradient was observed peaking at equal sections containing 4, 8, or 12 sampling points. 40% of the main radius and decreasing significantly The sampling points were determined by starting at (P< 0.05, analysis of variance, Student's Hest) toward the dorsal cut of the flat-mount and then proceeding the periphery and toward the optic nerve head (Figs. along each circle in clockwise direction. These sam- 2, 3). The intraocular injection of 20-nmol NMDA pling points then were located on the retina, and the resulted in a severe loss of HRP-labeled cells. Almost cells then were counted using the image-analysis sys- 80% to 85% of all RGCs were affected (Figs. 1, 2). tem in a rectangular ocular grid of 200 jum X 200 //.m After NMDA treatment, an average density of 299 ± 2 at each sampling point. 9 cells/mm (n = 8) was counted corresponding to an estimated 21,000 RGCs ( equal 18.2% of control) Behavioral Test for Brightness Discrimination per retina. The NMDA-induced RGC loss was similar The apparatus used for the assessment of brightness in all quadrants, but the relative number of surviving discrimination has been described previously by Saut- cells decreased with eccentricity from the optic nerve

Downloaded from iovs.arvojournals.org on 09/28/2021 Systemic L-Kynurenine Protects Against Excitotoxicity 2385

i " •• • r * - • • .

a . . D

•* .. *>• •• * •••.-. > • -•' ••.

*. .* • v r ' * •« 4 • " «

#

*

FIGURE 1. Representative photomicrographs taken from the ventral part of flat-mounted retinae of various treatment groups. Photographs on the left panel are taken from N-methyl- D-aspartate (NMDA) (a) and kainic acid (KA)-treated retinae (c). The right panel shows retinae from animals treated with NMDA + L-kynurenine (L-Kyn) (b) and KA + L-Kyn (d). (e) The corresponding part of a control retina (bar = 50 \im). (f) The horseradish peroxi- dase injection side in the tecttim is depicted (bar = 1 mm).

Downloaded from iovs.arvojournals.org on 09/28/2021 2386 Investigative Ophthalmology 8c Visual Science, November 1996, Vol. 37, No. 12

RGC/mm2

itmmt

MM-17W

14W15W \tn-im tempcii UM-UM

MMW mm

ventral N=5 Control

dorsal dorsal D RGC/mnr

IOOO low

900 »99 mm •mm

temporal nasal temporal naial

\---x 3MJW m-m

IM iw TBntru N-8 s«Hnei.p. + NMDAU N-* L-Kyn Lp.+ NMDA Lo.

donal dorsal

ins vm

1400 15W

nasal temporal nasal temporal

tm-m «m--m

venti.il

N-7 L-Kyn Lp. + KA i.o.

FIGURE 2. Average density of retrogradely labeled retinal ganglion cells using a false-gray scale in a reconstructed model of the retina. Areas reflect average cell counts from exacdy the same locations of retinae from different animals at the same treatment group. (A) Control retinae (n = 5) and the bar graph of all treaunent groups expressed as average cell numbers per square millimeter over the whole retina are shown. (B) The N-methyl-D- aspartate (NMDA) {n = 8) and NMDA + L-kynurenine (L-Kyn) (n = 8) treated retinae are shown. (C) The kainic acid (KA) {n = 8) and KA + L-Kyn (n = 7) lesioned retinae are shown. Note the different scales for cell density in the control and KA lesioned (A,C) compared with the NMDA group (B).

Downloaded from iovs.arvojournals.org on 09/28/2021 m Control BNMDA DNMDA+L-Kyn EKA HKA+L-Kyn

40 55 70 85 % of mean radius m Control BNMDA DNMDA+L-Kyn HKA BKA+L-Kyn

dorsal nasal ventral temporal

FIGURE 3. (A) Average density of retrogradely labeled retinal ganglion cells (RCCs) per square millimeter shown as a function of the centroperipheral gradient. The mean radius was measured from the optic nerve head to the ora serrata, and 25% of the radius reflects average cell density in areas between optic nerve head and 25% of the mean radius. Subse- quently, bars at 40% include areas measured between 25% and 40%. Highest density appears at 40% and is constantly decreasing to the periphery (*P < 0.05; **P < 0.001; analysis of variance, two-tailed Rest). (B) Average density of retrograde labeled RGCs per square milli- meter in different retinal quadrants after various treatments. Ganglion cell density appears to be lower in the dorsal quadrant, whereas in other quadrants, the cell density is similar. Highly significant protection was seen after L-kynurenine (L-Kyn) pretreatment in the N- methyl-D-aspartate (NMDA) lesioned group in all four quadrants. The L-Kyn pretreatment did not significantly prevented RGC loss, seen after kainic acid induced lesion of the adult rat retina (*P < 0.05; **P < 0.001; analysis of variance, two-tailed /-test).

Downloaded from iovs.arvojournals.org on 09/28/2021 2388 Investigative Ophthalmology & Visual Science, November 1996, Vol. 37, No. 12

head (22.3% at 25% eccentricity and 15.6% at 85% observed in the L-Kyn + NMDA treated group over eccentricity of the mean radius; P < 0.01, two-tailed the 20-day period. The L-Kyn pretreated animals /-test), thus indicating that RGCs in the peripheral showed a consistently better performance at all time retina are more susceptible to NMDA toxicity. points after the lesion. Systemic administration of L-Kyn reduced NMDA- induced cell death significantly in all sectors of the rat DISCUSSION retina (Figs. 2, 3). Almost twice as many cells survived NMDA treatment when animals were treated with L- Kynurenic acid is an L-tryptophan metabolite and is Kyn. An average cell density of 559 ± 11 cells/mm2 believed to be die only known endogenous antagonist (n = 8) was counted in those retinae corresponding of excitatory amino acid receptors in the central ner- to a total estimate of 39,130 cells/retina (i.e., 66% cell vous system.36 In this study, we could show that sys- loss). The protective effect of L-Kyn was considerably temic administration of its precursor, L-Kyn, offers smaller in die temporal retina. Only 10.4% more cells protective effects after in vivo retinal NMDA toxicity survived after combined NMDA + L-Kyn treatment in in adult rats. After L-Kyn pretreatment, twice as many comparison to NMDA control retinae in die temporal RGCs could be labeled retrogradely by HRP. In a sub- region, whereas 15% to 20% more cells survived in sequent behavioral test, it was shown that this protec- the ventral, dorsal, and nasal region of the retina (P tion on the morphologic level is accompanied by im- < 0.01, two-tailed /-test). Thus, it appears that the proved brightness discrimination behavior, suggesting number of cells rescued by L-Kyn increases with dis- a causal link between the number of surviving RGCs tance to the optic nerve head (50% more cells survived and performance in this behavioral task. In accor- near the optic nerve head compared to 129% in die dance with previous results, NMDA induces severe periphery). RGC degeneration (approximately 80% cell loss),"" 13 Furthermore, intraocular injection of 5-nmol KA which was followed up by a partial recovery from 12 resulted in a significant RGC loss of 50% (833 ± 11 visual deficits within 2 to 3 weeks after the lesion. cells/mm2, corresponding to a total estimate of 58,300 Computer-assisted image analysis indicates a regional cells/retina; n = 8), but KA + L-Kyn treatment had heterogeneity of NMDA toxicity with cells in the reti- no significant effect on cell survival (907 ±11 cells/ nal periphery being more suscepdble to the excito- mm2; n = 7). However, some minor, but not signifi- toxic lesion. Moreover, we also could show that KA in cant, protection after KA + L-Kyn was found in the the dose used is a powerful toxin for RGC (approxi- ventral sector. The percentage of KA-induced cell mately 50% cell loss) and that KA-induced excitotoxi- death was independent of eccentricity or retinal quad- city only can be attenuated to a minor extent by L- rant. Control injections with systemic L-Kyn pretreat- Kyn pretreatment. ment or saline intraocularly did not affect RGC num- Different types of excitatory amino acid-iono- ber (data not shown). tropic receptors have been characterized in the retina, and it is known that more than one subtype of gluta- Behavior mate receptor is present on RGCs.s' Although dieir To balance daily variations in animal behavior, daily individual contribution to retinal function still re- visual performance scores (expressed as percentage mains unclear, it has been known for a long time that of correct choices) were pooled in 3-day blocks. After retinal neurons are highly susceptible to excitotoxins NMDA injecdons, all animals showed an initial impair- such as NMDA and KA.20ss A bulk of data suggests ment in die visual discrimination task (Fig. 4). As pre- that NMDA toxicity is mediated via an uncontrolled viously observed,12 functional recovery from NMDA- influx of N+, K+, and Ca2+ ions, leading, in turn, to induced loss was found to occur within the first 2 neuronal swelling and secondary degeneration.39'40 weeks after the lesion. The percentage of correct Also in the retina, the release of Ca2+ from intracellu- choices in the NMDA lesioned group (n = 7) was lar stores has been discussed as a putative mode of significantly lower when compared with control ani- NMDA action under bodi normal and pathologic con- mals (P < 0.001, analysis of variance, /-test; n = 7). ditions.10 Furdiermore, retinal NMDA toxicity is char- Visual performance dropped from more than 90% acterized by a substantial cell loss in the RGC layer correct choices before NMDA injection within 1 week and inner nuclear layer, a pattern of degeneration to less than 60% and subsequently recovered to almost that resembles closely those found in a variety of reti- 75% within the following 12 days. Systemic administra- nal disease states. Therefore, it has been proposed tion of L-Kyn significantly attenuated die NMDA-in- that the predominant form of glutamate excitotoxicity duced deficits at postoperative days 3 to 5 (Welch t- in the retina is mediated via NMDA receptors. Our test; P < 0.01), 6 to 11 (Welch /-test; P < 0.01), and data extend previous observations on the conse- 12 to 20 (Welch /-test; P < 0.05). A restitution of quences of in vivo NMDA toxicity on die number of discrimination performance toward control levels was retrogradely labeled RGCs. We found that the majority

Downloaded from iovs.arvojournals.org on 09/28/2021 Systemic L-Kynurenine Protects Against Excitotoxicity 2389

Visual performance in a two-choice brightness discrimination task

Control N=7

NMDA + L-Kyn N=7

NMDA N=7

3.5 9-11 12-14 15-17 18-20 days after surgery FIGURE 4. Behavioral data showing recovery of brightness discrimination performance after intraocular N-methyl-D-aspartate (NMDA) (20 nmol) injection. The test scores were sampled and averaged over 3 days to minimize daily variations. Therefore, discrimination behavior is depicted as the mean ± standard error of the mean of 3 days' blocks. Visual function in this test was impaired by NMDA lesion, but improved partly within 3 weeks. The visual performance after injury still is significantly different from control animals. The L-kynuren- ine pretreatment had a remarkable effect on discrimination behavior of NMDA-lesioned animals (*P < 0.05; **P < 0.001; analysis of variance, two-tailed «-test). Rats made fewer errors immediately after excitotoxicity and performed at a higher level than did NMDA control subjects over the course of the experiment.

of RGCs are susceptible to NMDA-induced cell death. subregions, it is rather unlikely that a simple concen- Only 15% of the total population survived the intraoc- tration gradient due to the injection procedure can ular NMDA injection, a figure that is in accordance account for this result. with previous results.12'3 Because also a higher dose It has been reported that KA-induced neuronal of 100 nmol12 did not further decrease the number loss in vivo only affects cells in the inner nuclear of surviving RGC, it seems reasonable to argue that layer.18 Our results'12 and other studies19'43"'16 clearly RGCs exhibit different susceptibilities to the excito- indicate that in vivo KA also can induce substantial toxin. This is surprising, given the finding that virtually RGC cell death in adult rats. To unambiguously iden- all RGCs express the NMDA Rl and NMDA R2 sub- tify RGCs, we decided to use retrograde HRP tracing, units.41 Therefore, it can be expected that other fac- which precludes the possibility that displaced ama- tors that determine the ion flow through the NMDA crine cells in the RGC layer are counted when estimat- ionophore (i.e., different subunit composition of the ing RGC numbers. Although the issue of KA-induced native ) might account for this result. Interest- RGC cell death could be resolved by retrograde label- ingly, we also found a peripheral gradient of NMDA ing of RGCs, the mechanism of KA toxicity in these but not KA-induced degeneration, also pointing to a cells is not yet understood completely. It has been regional heterogeneity of RGCs. Further evidence for suggested that NMDA receptors play a permissive role regional differences in NMDA excitotoxicity was pro- because NMDA antagonists prevent KA neurotoxicity vided by the observation that the neuroprotective ef- in rat RGCs in vitro.21 However, the involvement of fect of L-Kyn increased with retinal eccentricity. Be- NMDA receptors in KA-induced cell death has been cause KA neurotoxicity equally affected all retinal controversial because contradictory evidence was pre-

Downloaded from iovs.arvojournals.org on 09/28/2021 2390 Investigative Ophthalmology & Visual Science, November 1996, Vol. 37, No. 12

sented by the observation that MK-801, a noncompeti- levels of discrimination behavior over the 3-week pe- tive NMDA antagonist, did not prevent KA-induced riod of the experiment. This extent of protection on neuronal loss in the rat hippocampal GA3-CA4 re- the behavioral level is surprising considering that die gion, whereas it was effective in other regions.47 Fur- total number of rescued cells is modest. However, thermore, MK-801 and also competitive NMDA antag- taken in account is the fact that the increase in cell onists did not protect neurons in the KA-injected stria- number after L-Kyn administration is large compared tum'18 while preserving cells from neurotoxicity in the with that of NMDA (twice the number of RGC) and 49 olfactory cortex. Moreover, it has been assumed that that the visual task used in this study is rather simple. apart from the activation of NMDA receptors, a KA- Suggestive for a causal relation between cell number induced membrane depolarization and subsequent and behavioral performance is the finding that L-Kyn- 40 giutamate release is part of its excitotoxic action. treated animals immediately after the lesion made This would imply that depolarization of both bipolar fewer errors in die task and performed consistently cells and RGC with a subsequent activation of NMDA better than did the NMDA group. This substantiates receptors located on RGC is a prerequisite for RGC the hypothesis that increasing the levels of the endoge- degeneration after intraocular KA toxicity. The evi- nous competitive NMDA antagonist KYNA by pharma- dence for such a sequelae of events in die retina, how- cologic means has some promise for the protection ever, is rather sparse. Furthermore, in contrast to its against excitotoxic insults. effects after NMDA toxicity, the systemic administra- tion of L-Kyn only had a minor effect on retinal KA Key Words toxicity. Because L-Kyn is metabolized to KYNA,n0 a competidve antagonist at the allosteric glycine site of brightness discrimination, excitotoxicity, giutamate, kainic the NMDA receptor51 (also see below), it has to be acid, L-kynurenine, N-methyl-D-aspartate, retina, retinal ganglion cells expected that this compound also attenuates KA toxic- ity in vivo if the NMDA receptor plays a permissive Acknowledgments role during this excitotoxic event. Kynurenic acid is a considerably weaker antagonist at the non-NMDA The authors thank C. Dastig, M. Marunde, and B. Kracht receptors than at the glycine binding site of the NMDA for their technical assistance. receptor. With respect to the dose of L-Kyn used, this might explain its inefficacy in the prevendon of KA References toxicity. 1. Choi DW. Giutamate neurotoxicity and diseases of the nervous system. . 1988; 1:623-634. The allosteric glycine site of the NMDA receptor 2. Dingledine R, Myers SJ, Nicholas RA. Molecular biol- can be blocked with low micromolar concentrations ogy of mammalian amino acid receptors. FASEB J. of KYNA, whereas 20- to 30-fold higher concentrations 1990;4:2636-2645. are required to block the NMDA recognidon site.D2'D3 3. Faden AI, Demediuk P, Panter SS, Vink R. The role Therefore, it is possible that KYNA concentrations that of excitatory amino acids and NMDA receptors in block NMDA excitotoxicity can be achieved in vivo.3'1''''3 traumatic brain injury. Science. 1989;244:798-800. In this study, we administrated systemically 100 mg/ 4. Lipton SA, Rosenberg PA. Excitatory amino acids as kg L-Kyn, a dose that has been shown previously to a final common pathway for neurologic disorders. N evaluate striatal extracellular levels of KYNA 37-fold EnglJMed. 1994; 330:613-622. 22 5. Mosinger JL, Price MT, Bai HY, et al. Blockade of after intraperitoneal injections. This corresponds to both NMDA and non-NMDA receptors is required for a range where KYNA has been shown to block the optimal protection against ischemic neuronal degen- glycine site of the NMDA receptor. Therefore, it is eration in the in vivo adult mammalian retina. Exp likely that the neuroprotective effect of L-Kyn ob- Neurol. 1991; 113:10-17. served in this study is mediated by a KYNA-induced 6. Massey SC. Cell types using giutamate as a neurotrans- block of the glycine allosteric site of the NMDA recep- mitter in the vertebrate retina. In: Osborne E, Chader tor complex. G, eds. Progress in Retinal Research. VolIX. Oxford: Per- The most remarkable finding of the present study gamon Press; 1990:399-425. was the observation diat die neuroprotective effect of 7. Miller RF, Slaughter MM. Excitatory amino acids re- L-Kyn in the retina was correlated with improved visual ceptors of the retina: diversity of subtypes and conduc- tance mechanisms. Trends Neurosci. 1986;9:211-218. discrimination behavior. Only few pharmacologic 8. OlneyJW. Glutamate-induced retinal degeneration in studies so far emphasize die functional consequences neonatal mice: Electron microscopy of the acutely of neuroprotection on the cellular level after excito- evoking lesion. JNeuropalholExp Neurol. 1969;28:455- 56 toxicity. Our finding of a positive relation between 474. behavioral performance and the number of surviving 9. Lucas DR, Newhouse JP. The toxic effect of sodium cells is therefore particularly encouraging. Moreover, L-glutamate on the inner layers of the retina. Arch only die L-Kyn-treated rats achieved almost control Ophlhalmol. 1957;58:193-204.

Downloaded from iovs.arvojournals.org on 09/28/2021 Systemic L-Kynurenine Protects Against Excitotoxicity 2391

10. Lei SZ, Zhang D, Abele AE, Lipton SA. Blockade of 27. Turski WA, Nakamura M, Todd WP, et al. Identifica- NMDA receptor-mediated mobilization of intracellu- tion and quantification of kynurenic acid in human lar Ca2+ prevents neurotoxicity. Brain Res. 1992;598: brain tissue. Brain Res. 1989;545:164-169. 196-202. 28. Turski WA, Gramsbergen JBP, Traitler H, Schwarcz 11. Dreyer EB, Pan ZH, Storm S, Lipton SA. Greater sensi- R. Rat brain slices produce and liberate kynurenic tivity of larger retinal ganglion cells to NMDA-medi- acid upon exposure to L-kynurenine. J Neurochem. ated cell death. Neumreport. 1994;5:629-631. 1989;52:5:1629-1636. 12. Sabel BA, Sautter J, Stoehr T, Siliprandi R. A behav- 29. Jannson R. Glutamate neurotoxicity in the developing ioral model of excitotoxicity: retinal degeneration, rat cochlea is antagonized by kynurenic acid and MK- loss of vision, and subsequent recovery after intraocu- 801. Brain Res. 1992;590:201-206. lar NMDA-administration in adult rats. Exj) Brain Res. 30. Perry VH, Linden R. Evidence for dendritic competi- 1995; 106:93-105. tion in the developing retina. Nature. 1982; 297:683- 13. Siliprandi R, Canella R, Carmignoto G, et al. NMDA- 685. induced neurotoxicity in the adult rat retina. Vis Neu- 31. Hanker JS, Yates PE, Metz CB, Rustioni A. A new spe- rosci. 1992; 8:567-573. cific, sensitive and non carcinogenic reagent for the 14. Sucher NJ, Lei SZ, Lipton SE. channel antag- demonstration of horseradish peroxidase. Histochem [. onists attenuate NMDA receptor-mediated neurotox- 1977;9:789-792. icity of retinal ganglion cells in culture. Brain Res. 32. Sautter J, Sabel BA. Recovery of brightness discrimina- 199l;551:l-2:297-302. tion in adult rats despite progressive loss of retro- 15. Dreyer EB, Vorwerk CK. The glaucomas. In: Adelman gradely labeled retinal ganglion cells after controlled G, Smith BH, eds. Encyclopedia ofNeurosci. Amsterdam: optic nerve crush. Eur J Neurosci. 1993;5:680-690. Elsevier; 1996; in press. 33. Sautter J, Schwartz M, Duvdevani R, Sabel BA. GM1 16. Hahn JS, Aizenman E, Lipton SA. Central mammalian ganglioside treatment reduces visual deficits after neurons normally resistant to glutamate toxicity are graded crush of the rat optic nerve. Brain Res. 1991; made sensitive by elevated extracellular Ca'^: toxicity 565:23-33. is blocked by the N-methyl-D-aspartate antagonist MK 34. Dreher B, Potts RA, Bennett MR. Evidence that the 801. ProcNallAcad Sci USA. 1988; 85:6556-6560. early postnatal reduction in the number of rat retinal 17. Ingham CA, Morgan IG. Dose-dependent effects of ganglion cells is due to a wave of ganglion cell death. intravitreal kainic acid on specific cell types in chicken Neurosci Lett. 1983;36:255-259. retina. Neuwscience. 1983;9:165-181. 35. Schober W, Gruschka H. Zur Projektion der einzelnen 18. Schwarcz R, Coyle JT. Kainic acid neurotoxic effects Ganglienzellklassen der Retina der AJbinoratte. Eine after intraocular injection. Invest Ophlhalmol Vis Sci. Studie mit Meerrettich-Peroxidase. Z mikrosk-anat 1977;16:141-148. Forschung. 1978; 92:2:283-297. 19. Redburn DA, Donoso JA, Mitchell CK, et al. Kainic 36. Stone TW, ConnickJH. Quinolic acid and other kyn- acid-induced denervation supersensitivity of nicotinic, urenines in the central nervous system. Neuwscience. cholinergic receptors in ganglion cells of the rat ret- 1985; 15:597-617. ina. ExpEyeRes. 1984; 38:449-461. 37. Hughes TE, Hermans-Borgmeyer I, Heinemann S. 20. Hampton CK, Garcia C, Redburn DA. Localization of Differential expression of glutamate receptor genes kainic acid-sensitive cells in the mammalian retina. / (GluRl-5) in the rat retina. Visual Neurosci. 1992; NeurosciRes. 1981;6:99-111. 8:49-55. 21. Sucher NJ, Aizenman E, Lipton SE. NMDA antago- 38. Sisk DR, Kuwabara T. Histologic changes in the inner nists prevent kainate neurotoxicity in rat retinal gan- retina of albino rats following intravitreal injection of glion cells in vitro. / Neurosci. 1991; 11:966-971. monosodium L-glutamate. Graefe's Arch Clin Exp Oph- 22. Swartz KJ, During MJ, Freese A, Beal MF. Cerebral thalmol. 1985;223:250-258. synthesis and release of kynurenic acid: An endoge- 39. Choi DW, Maulucci Gedde M, Kriegstein AR. Gluta- nous antagonist of excitatory amino acid receptors. J mate neurotoxicity in cortical cell culture. J Neurosci. Neurosci. 1990; 10:2965-2973. 1987;7:357-368. 23. Fukui S, Schwarcz R, Rapoport SI, et al. Blood-Brain 40. Choi DW, Koh J, Peters S. Pharmacology of glutamate barrier transport of : Implications for neurotoxicity in cortical cell culture: Attenuation by brain synthesis and metabolism. J Neurochem. 1991; NMDA antagonists. J Neurosci. 1988;8:185-196. 56:2005-2017. 41. Brandstatter JH, Hartveit E, Sassoe-Pognetto M, Was- 24. Okuno E, Nakamura M, Schwarcz R. Two kynurenine sle H. Expression of NMDA and high-affinity kainate aminotransferases in human brain. Brain Res. 1991; receptor subunit mRNAs in the adult rat retina. EurJ 542:307-312. Neurosci. 1994; 6:1100-1112. 25. Okuno E, Schmidt W, Parks DA, Nakamura M, 42. Vorwerk CK, Kreutz MR, Sabel BA. Systemic L-Kyn- Schwarcz R. Measurement of rat brain kynurenine urenine protects against NMDA but not Kainate in- aminotransferase at physiological kynurenine concen- duced lesions in the adult rat retina. Soc Neurosci Abs. trations. J Neurochem. 1991; 57:2:533-540. 20:1994:121.1. 26. MinatogawaY, Noguchi T, Kido R. Kynurenine pyro- 43. Ferrari R, Biral G. Pattern of the retinal afferent distri- vate in rat brain. J Neurochem. 1974;23: bution in the rat superior colliculus after intravitreal 271-272. kainic acid injection. Med Sci Res. 1989; 17:49-50.

Downloaded from iovs.arvojournals.org on 09/28/2021 2392 Investigative Ophthalmology & Visual Science, November 1996, Vol. 37, No. 12

44. Gomez-Ramos P, Estrada C, Perez-Rico C. Choline 51. Heyes MP, Quearry BJ. Quantification of kynurenic acetyltransferase deplation in the rat retina after intra- acid in cerebrospinal fluid: Effects of systemic and ocular injection of neurotoxins. J Neurochem. 1985;44: central L-kynurenine administration./Chromatogr Bio- 993-995. medAppl. 1990;530:108-115. 45. Goto M, Inomata N, Ono H, Saito KI, Fukuda H. 52. Foster AC, Willis CL, Tridgett R. Protection against Changes of electroretinogram and neurochemical as- N-methyl-D-aspartate receptor-mediated neuronal de- pects of GABAergic neurons of retina after intraocular generation in rat brain by 7-chlorokynurenate and 3- injection of kainic acid in rats. Brain Res. 1981; amino-l-hydroxypryrolid-2-one, antagonists at the al- 211:305-314. losteric site for glycine. Eur J Neurosci. 1988; 2:270- 46. Sattayasai J, Rogers LJ, Ehrlich D. Sequential treat- 277. ment with lower doses of kainic acid alters sensitivity 53. Kessler M, Terramani T, Lynch G, Baudry M. A glycine of retinal cell types. Neurosci Lett. 1985;54:277-281. site associated with N-methyl-D- recep- 47. Rogers BC, Tilson HA. Kainate-induced functional tors: Characterization and identification of a new class deficits are not blocked by MK-801. Neurosci Lett. of antagonists. J Neurochem. 1989;52:1319-1332. 1990; 109:335-340. 54. Foster AC, Vezzani A, French ED, Schwarcz R. Kyn- urenic acid blocks neurotoxicity and seizures induced 48. Foster AC, Gill R, Woodruff GN. Neuroprotective ef- in rats by the related brain metabolite quinolic acid. fects of MK-801 in vivo: selectivity and evidence for Neurosci Lett. 1984;48:273-278. delayed degeneration mediated by NMDA receptor 55. Nozaki K, Beal MF. Neuroprotective effects of L-kyn- activation. J Neurosci. 1988;8:4745-4754. urenine on hypoxia-ischemia and NMDA lesions in 49. Guarnieri T, Virgili M, Villani L, et al. Pharmacologi- neonatal rats. / Cereb Blood Flow Melab. 1992; 12:400- cal manipulation of the NMDA receptor differentially 407. protects from systemic kainic acid neuropadiology: 56. Smith DH, Okiyama K, Thomas MJ, Mclntosh TK. Evaluation immunohistochemistry. Restor Neurol Neu- Effects of the excitatory amino acid receptor antago- rosci. 1993; 5:327-335. nists Kynurenate and Indole-2-carboxylic acid on be- 50. Speciale C, Schwarcz R. Uptake of kynurenine into havioral and neurochemical outcome following exper- rat brain slices./Neurochem. 1990;54:156-163. imental brain injury. J Neurosci. 1993; 13:5383-5392.

Downloaded from iovs.arvojournals.org on 09/28/2021