The Localization of Endogenous Zinc and the in vitro Effect of Exogenous Zinc on the GABA Immunoreactivity and Formation of Reactive Oxygen Species in the Retina*

ISSN 0306-3623/98 $19.00 1 .00 PII S0306-3623(97)00358-3 All rights reserved

Gen. Pharmac. Vol. 30, No. 3, pp. 297–303, 1998 Copyright  1998 Elsevier Science Inc. Printed in the USA.

The Localization of Endogenous Zinc and the in vitro Effect of Exogenous Zinc on the GABA Immunoreactivity and Formation of Reactive Oxygen Species in the Retina* Marta Ugarte and Neville N. Osborne† Nuffield Laboratory of Ophthalmology, University of Oxford, Walton Street, Oxford OX2 6AW, United Kingdom ABSTRACT. 1. Endogenous zinc is localized mainly in the retinal photoreceptors and retinal pigment epithelial cells in the mammalian retina. No other types of retinal neurons contain large amounts of zinc. 2. Low concentrations of exogenous zinc, like the N-methyl-D-aspartate (NMDA) antagonist MK801, counteract the NMDA-induced changes in the g-aminobutyric acid (GABA) immunoreactivity in the rabbit retina. However, greater concentrations of zinc exacerbate the effects of NMDA and ischemic-like insults (lack of glucose and oxygen) on GABA immunoreactivity. The data suggest that low concentrations of zinc are neuroprotective, but higher concentrations of zinc have a negative effect. 3. When low concentrations of zinc are present during ischemic-like insults to the retina, the GABA immunoreactivity is localized to the Mu¨ller cells, suggesting that the metabolism of GABA in the Mu¨ller glial cells is prevented. 4. Ascorbate/iron-induced generation of reactive oxygen species (ROS) in the retina is prevented by deferoxamine but not by zinc. High concentrations of zinc potentiate the ascorbate/iron induced formation of ROS. gen pharmac 30;3:297–303, 1998.  1998 Elsevier Science Inc. KEY WORDS. Zinc, mammalian retina, GABA, reactive oxygen species, in vitro studies, NMDA, ischemia INTRODUCTION 21

Zinc ions (Zn ) exist in high amounts in the brain and retina (Eckert, 1979; Frederickson 1989). Most studies have been carried out on brain tissues, where Zn21 is postulated to have a number of functions (Frederickson, 1989). Experimental evidence exists to suggest that Zn21 can have positive or negative effects, depending on the local concentration (Harrison and Gibbons, 1994). Low amounts of Zn21 can act as a protective agent (Khulusi et al., 1986; Peters et al., 1987); but when the concentration reaches a certain level, it can be detrimental and exacerbate damage caused by excitotoxicity (Choi et al., 1988; Koh and Choi, 1988), which is one of the mechanisms in ischemic neuronal damage. Administration of Zn21 for therapeutic purposes must therefore be carried out with caution. Recent studies suggest that Zn21 can serve as a messenger between cells (Koh et al., 1996). It is coreleased with glutamate from some neurons (Assaf and Chung, 1984; Howell et al., 1984) and can modulate synaptic neurotransmission by depressing the response of Nmethyl-d-aspartate (NMDA) receptors (Christine and Choi, 1990; Peters et al., 1987) and block certain voltage-gated calcium channels (Winegar and Lansman, 1990). However, during excessive depolarization, as occurs in ischemia, the “released” Zn21 reaches high levels and can participate in neurotoxicity (Koh et al., 1996). In this case, the zinc passes into the postsynaptic neuron through the “open” NMDA receptor voltage-gated calcium channels. Therefore, *Part of this work was presented at a meeting on ‘‘Basic mechanisms related to stress, ischemia and neuroprotection in the retina’’ held in Coimbra, Portugal (11–12 April 1997) and organized by A. P. Carvalho and N. N. Osborne. †To whom correspondence should be addressed. Received 1 May 1997; accepted 13 July 1997.

neurons possessing NMDA receptors that are situated postsynaptic to glutamatergic cells containing Zn21 are particularly susceptible during ischemia (Koh et al., 1996). Retinal ganglion cells are known to possess NMDA receptors (Massey, 1990) and during ischemia, these cells are destroyed (Buchi, 1992; Selles-Navarro et al., 1994, 1996). Because ganglion cells appear to be the most susceptible neurons in the retina in, for example, glaucoma (Nickells, 1996), the possibility therefore exists that, if Zn21 is released from presynaptic neurons (amacrine and bipolar cells), it may participate in the destructive process. One aim of the present study was therefore to test this idea and process the retina for the localization of endogenous Zn21. Another aim was to examine whether exogenous Zn21 can offer protection to retinal neurons. Previous studies have shown that exposure of the isolated rabbit retina to certain glutamate agonists (NMDA or kainate), ischemic-like insults or metabolic poison (KCN) caused a characteristic change in the nature of g-aminobutyric acid (GABA) immunoreactivity (Osborne and Herrera, 1994). The changes in GABA immunoreactivity can be blocked by MK801 (selective NMDA receptor antagonist) and CNQX (selective antagonist at kainate receptors) (Osborne and Herrera, 1994). The effect of different concentrations of Zn21 on GABA immunoreactivity in either the presence or the absence of NMDA was therefore studied to gain an understanding of the neuroprotective role of Zn21. Exogenous Zn21 can allosterically reduce the effect of glutamate at NMDA receptors and so act as a neuroprotective agent. Excessive influx of calcium and production of reactive oxygen species (ROS) are known to contribute to cell destruction (Bondy and LeBel, 1990; Michikawa et al., 1994; Verity et al., 1995), so other possible modes of neuroprotection by Zn21 also must be considered. In the present

298 study, we specifically examined the effect of Zn21 on ROS generated in vitro by ascorbate and iron.

M. Ugarte and N. N. Osborne bital and 140 mM sodium acetate buffer (pH 10) for 5 min. After a brief wash in saline, TSQ-Zn21 fluorescence was observed by using fluorescence microscopy.

MATERIALS AND METHODS

Histochemical localization of chelatable zinc in the rat retina The method used was that described by Danscher (1996), which is to localize Zn21 by formation of zinc-sulfide complexes. Wistar rats were anesthetized (pentobarbitone 45 mg/kg intraperitoneally) and sodium sulfide (200 mg/kg) in phosphate-buffer saline was injected into the vena cava. Four minutes after the injection, animals were transcardially perfused, initially with saline and then with a mixture of 1% glutaraldehyde and 2% paraformaldehyde in 0.1 M phosphate buffer. The eyes were removed, postfixed in the same fixative for 1 hr and transferred into a 30% sucrose in 0.1 M phosphate buffer. Retinas were dissected from the eyes, and frozen cryostat (10 mm) sections were placed on gelatinized glass slides. The sections were air dried (30–60 min), coated with 0.5% gelatin and developed in a silver lactate developer [60 ml protecting colloid (1 kg of gum arabic in 2 l of deionized water), 10 ml citrate buffer (25.5 g citric acid monohydrate, 23.5 g sodium citrate in 100 ml of distilled water), 15 ml reducing agent (0.85 g hydroquinone in 15 ml of distilled water) and 15 ml of a silver ion supply (0.11 g of silver lactate in 15 ml of distilled water)] for 60 min at 268C. The sections were then rinsed in running tap water (408C) for 20 min and cover-slip mounted in DPX. The sections were unstained if: (1) rats had not been subjected to the sulfide ions, (2) 1 g/kg body weight of the chelating agent diethyldithiocarbamate was injected intraperitoneally 1 hr before the intravenous sulfide administration; and (3) sections containing zinc sulfide crystals were treated with 0.1 M HCl, for 10 min, prior to development.

Fluorescence procedure for the localization of chelatable Zn21 in the rabbit retina The method was that described by Fredrickson et al. (1987) to localize Zn21 binding to 6-methoxy-8-p-toluene sulfonamide quinoline (TSQ). Freshly dissected retinas (New Zealand albino rabbits) were frozen in dry ice, and 10-mm-thick cryostat sections were mounted on gelatinized glass slides. Sections were immersed in a solution of TSQ (4.5 mM, Molecular Probes, Holland) in 140 mM sodium bar-

FIGURE 1. Localization of chelatable Zn21 (by formation of TSQZn21 fluorescent complexes) in the isolated rabbit retina. Zn21 is associated with the inner segments (IS) of the photoreceptors (arrows). No TSQ-Zn21 fluorescence is associated with other parts of the retina. Scale bar520 mm.

Effect of Zn21 on GABA immunoreactivity Anesthetized rabbits were killed with an overdose of pentobarbital and the retinas rapidly dissected. The retinas were cut into pieces and placed in Mg21-free Locke’s buffer (157 Na1, 5.6 K1, 2.3 Ca21, 164 Cl2, 3.6 HCO32, 5 HEPES and 5.6 glucose in mmoles per liter) that had been equilibrated with a gas mixture (95% O2 and 5% CO2) maintained at 378C with a pH of 7.4. In some cases, modified Locke’s solution was used; this modified solution lacked glucose and was equilibrated with 95% N2 and 5% CO2. Retinal pieces were incubated for 45 min at 378C in 6 ml Locke’s solution or modified Locke’s solution either alone or containing NMDA, MK-801 or Zn21 or all three. The tissues were then fixed and processed for the localization of GABA as described elsewhere (Osborne and Herrera, 1994).

Effect of Zn21 on ROS The assay is based on that described by Lebel and Bondy (1990). This assay requires the use of DCFH-DA (29,79-dichlorodihydrofluorescein diacetate from Molecular Probes, Holland), which is a stable, nonfluorescent molecule that readily crosses cell membranes and is hydrolyzed by intracellular esterases to nonfluorescent 29,79dichloro-fluorescin (DCFH). DCFH is then rapidly oxidized in the presence of reactive oxygen species to highly fluorescent DCF (29,79-dichlorofluorescein). The retinas from newborn chicks were dissociated by an enzymatic procedure that consisted of incubating the retinas in L-15 medium (Gibco/Life Technologies) containing 4 mg/ml dispase (2 retinas/2 ml) for 10 min at 308C. Repetitive trituration was carried out to dissociate the tissue mechanically. Cellular debris and undissociated tissues were removed until a homogeneous suspension was obtained. The cells were washed with HBSS (Gibco/Life Technolo-

FIGURE 2. Localization of chelatable Zn21 (by formation of Zn21sulfide complexes) in the intact rat retina. Zn21 is associated with the inner segments (IS) of photoreceptors, retinal pigment epithelial (RPE) cells, processes in the inner and outer plexiform layers (small arrows), cell bodies in the ganglion cell layer (arrow heads) and specific terminal processes between the inner plexiform and nuclear layers (double arrows). Some light staining is also associated with perikarya in the inner and outer nuclear layers and a few fibers traversing the retina, which could correspond to Mu¨ller cells (long arrows). Scale bar520 mm.

Zinc and ROS in the Retina

299

FIGURE 3. The effect of different concentrations of ZnSO4 (1–500 mM) on GABA immunoreactivity in the isolated rabbit retina. (A) The normal distribution of GABA. The responses to ZnSO4 (1, 100, 500 mM) itself are shown in (B–D), respectively. The effect of ischemic-like (lack of glucose and oxygen) insults on GABA immunoreactivity can be seen in (E), and the influences of NMDA plus ZnSO4 on these changes are illustrated in (F) ZnSO4, 1 mM; (G) ZnSO4, 100 mM and (H) ZnSO4, 500 mM. The NMDA (100 mM) response is shown in (I) and the influences of ZnSO4 are shown in (J) ZnSO4, 1 mM; (K) ZnSO4, 100 mM and (L) ZnSO4, 500 mM. In the control tissues (A), GABA immunoreactivity is associated with some perikarya in the inner nuclear and ganglion cell layers and with the compact band of processes in the inner plexiform layer. Normal GABA immunoreactivity is unaffected when the tissues are exposed to ZnSO4 at concentrations ranging from 1 to 100 mM (B and C, respectively). However, ZnSO4, 500 mM, results in a reduction in the number of GABA-containing perikarya and three distinct GABA-immunoreactive bands in the inner plexiform layer (IPL) (D). Ischemic-like insults and NMDA (100 mM) resulted in a decrease in the number of GABA-positive perikarya and a banding pattern in the IPL (E and I, respectively). The ischemia-like response was similar to that occurring with high concentrations of ZnSO4, causing the appearance of three immunoreactive bands in the IPL (E). Exposure of the tissues to NMDA (100 mM) caused the appearance of 3–4 bands in the IPL. ZnSO4 (1 mM) caused a redistribution of GABA and its association with Mu¨ller cells in the tissues subjected to ischemic-like insults (F) but had no effect on the NMDA-induced changes (J). The ischemia-like and NMDA responses were partly blocked or unaffected by 100 mM ZnSO4 (G and K, respectively); however, they were exacerbated by 500 mM ZnSO4 (H and L, respectively). Scale bars530 mm.

gies) medium and centrifuged. Cells were then suspended in fresh HBSS (10 ml for 14–18 retinas) containing DCFH-DA (20 mM) for 30 min at 308C. After it was loaded with DCFH-DA, the cellular suspension was centrifuged at 30003g for 3 min at 48C. The pellet was resuspended in fresh HBSS and divided into 2 ml samples (1 retina/ml) in cuvettes and incubated at 308C for 60 min. Various substances [e.g., ascorbate (100 mM) plus FeSO4 (5 mM) with or without deferoxamine and/or ZnSO4] were added as required, the pH was adjusted to 7.4 and the cuvettes were transferred to a Perkin Elmer LS50B spectrofluorimeter. The cuvettes were maintained at 308C and the fluorescence (with excitation and emission wavelengths of 488 nm and 525 nm, respectively, and band widths of 5 nm) was measured at intervals of 15, 45 and 60 min. The ROS formed was quantified by constructing a standard curve and using various concentrations of pure DCF (Sigma).

ments of the photoreceptors can be seen to give an intense positive reaction. No staining is associated with other parts of the retina. Figure 2 shows the histochemical localization of Zn21, as zinc-sulfide, in the rat retina. This is a more sensitive method than the fluorescent-TSQ procedure for localizing zinc (Danscher, 1996). Intense staining for Zn21 is associated not only with the inner segments of the photoreceptors, but also with retinal pigment epithelial cells. Positive staining is also clearly associated with processes in the inner and outer plexiform layers, cell bodies in the ganglion cell layer and specific terminal process just between the inner plexiform and nuclear layers. Some light staining is also associated with perikarya in the inner and outer nuclear layers. Zinc-positive staining is also associated with a few fibers that traverse the retina, suggesting an association with Mu¨ller cells.

Effect of Zn21 on GABA immunoreactivity RESULTS

Localization of chelatable zinc Figure 1 shows a section through the rabbit retina demonstrating the fluorescence localization of TSQ-Zn21 products. The inner seg-

Rabbit retinal tissues incubated in Locke’s solution revealed a pattern of GABA immunoreactivity identical with that observed when tissues are fixed directly after dissection (result not shown). In these cases, GABA immunoreactivity is associated with some perikarya in

300

M. Ugarte and N. N. Osborne

FIGURE 4. The effect of NMDA (100 mM) on GABA immunoreactivity in the isolated rabbit retina and the influence of MK801 (2 mM). Exposure of tissues to 100 mM NMDA caused the appearance of 3–4 immunoreactive bands in the inner plexiform layer and a nearly complete absence of GABA-containing perikarya in the inner nuclear and ganglion cell layers (see also Fig. 3E). (B) The NMDA response was nullified by MK801 (2 mM). Scale bars530 mm.

the inner nuclear and ganglion cell layers and forms an almost homogeneous band of immunoreactivity in the inner plexiform layer (Fig. 3A). ZnSO4 at concentrations of 1–100 mM did not have an effect on the normal “staining” for GABA immunoreactivity (Fig. 3B and C). However, at a concentration of 500 mM, Zn21 induced a change in the nature of GABA immunoreactivity that was not dissimilar from that which occurs when the tissue is exposed to modified Locke’s solution to simulate an ischemic-like effect (Fig. 3D). In these cases, GABA-containing parikarya were very few in number and a “banding” pattern of immunoreactivity appeared in the inner plexiform layer. NMDA (100 mM) induced the clearest change in the GABA immunoreactivity, revealing the appearance of 3–4 bands of immunoreactivity in the inner plexiform layer (Figs. 3A and 4A). This NMDA effect on the change in the normal GABA immunoreactivity was completely counteracted by MK-801 (Fig. 4B). The changes caused by NMDA and modified Locke’s solution to the normal GABA immunoreactivity were also partly blocked by ZnSO4 at a concentration of 100 mM (Fig. 3G and L, respectively). When the concentration of ZnSO4 was as low as 1 mM, it had no effect on the NMDA-induced changes to GABA immunoreactivity (Fig. 3J) but, for the retinas in modified Locke’s solution (ischemiclike insult), it had a rather interesting effect on the localization of GABA. In this case, GABA was found to be redistributed and associated with the Mu¨ller cells (Fig. 3F). ZnSO4 at high concentrations (500 mM) had an exacerbating effect on the changes in GABA immunoreactivity induced by NMDA or modified Locke’s solution (Fig. 3H and L, respectively).

Effect of Zn21 on ROS FeSO4 (5 mM) together with ascorbate (100 mM) stimulate the formation of ROS in chick retinal homogenates (Fig. 5). When deferoxamine is present to chelate Fe31, the stimulation of ROS formation is prevented. Addition of ZnSO4 at low concentrations to FeSO4 /ascorbate appeared to have no effect on the formation of ROS. Higher concentrations of ZnSO4 (100–500 mM), in contrast, enhanced the formation of ROS. DISCUSSION Evidence from studies on the brain suggests that zinc-containing neurons are a subclass of glutamate cells located primarily in the telencephalon (Frederickson and Moncrieff 1994; Slomianka, 1992).

FIGURE 5. The effect of ZnSO4 (10–500 mM) on the formation of reactive oxygen species (ROS) induced by oxidative stress (5 mM FeSO4 and 100 mM ascorbic acid) in chick retinal homogenates. The production of ROS (mean6SEM, n57 experiments per bar) was prevented by the addition of the Fe31 chelator deferoxamine (100 mM). The presence of ZnSO4 (10–100 mM) had no effect on this formation, whereas, ZnSO4 (500 mM) increased it. The amount of ROS formed was scaled to the mean value in samples without oxidative stimulation, controls (5100). *P50.01 (Student t-test).

Glutamate is utilised as a transmitter in the retina by a variety of cell types, including the photoreceptors, bipolar cells, amacrine and ganglion cells (Daw et al., 1989; Massey, 1990; Wu, 1994), and various types of glutamate receptor have been located to virtually every cell type in the retina (Peng et al., 1995). In the present study, we find no evidence that zinc is specifically located in high amounts to a single class of retinal neurons (see Figs. 1 and 2). Two procedures were used in an attempt to localize zinc. To locate the zinc-sulfide complex, it is necessary to perfuse the anesthetized animal with a solution of sulfide, and this was performed in the rat, where the retina has a blood supply. The rabbit retina lacks a blood supply; so, in this case, the localization of zinc in the dissected tissues was instead detected by a less-sensitive method including the use of a fluorescence dye (TSQ) that forms a complex with zinc. Both methods suggest that the photoreceptors stain positively for zinc with the greatest concentration in the inner segments. Hirayama (1990) also concluded that zinc was primarily associated with the photoreceptors in the mammalian retina but that it was mainly found in the outer segments. It is difficult to understand why we find the greatest localization of zinc in the inner segments and not outer segments of the photoreceptors but this could relate to the functional state of the

Zinc and ROS in the Retina

301

FIGURE 6. GABA–glutamine shuttle between GABAergic amacrine neurons and Mu¨ller cells in the rabbit retina. GABA is formed within the GABAergic amacrine neuron by the action of glutamic acid decarboxylase (GAD), which decarboxylates glutamate. Glutamate is formed in GABAergic amacrine and Mu¨ller cells by the enzyme aspartate amino transferase (AAT), which catalyzes the transamination between glutamate/a-ketoglutarate and aspartate/oxaloacetate. The enzyme glutamate dehydrogenase (GDH) can also catalyze the formation of glutamate from a-ketoglutarate and NH41. These two enzymes (AAT and GDH) catalyze reversible reactions and are found in GABAergic amacrine cells and Mu¨ller’s cells. After its release, GABA is taken up by Mu¨ller cells through a highaffinity mechanism. Once within the Mu¨ller cells, it returns to the tricarboxylic acid (TCA) cycle through the action of the enzymes GABA transaminase (GABA-T) and succinate semialdehyde decarboxylase (SSAD). The constituents of GABA in the TCA cycle are converted into a-ketoglutarate and glutamate. Glutamate will be amidated to form glutamine through the action of glutamine synthetase (GS). Glutamine is free to return to the nerve ending of the GABAergic amacrine cell, where it is reconverted into GABA. [Modified from Robin and Kalloniatis (1992).]

rats in the different studies. The report by Wu et al. (1993) may be of significance in this context. These authors studied the distribution of zinc in the dark-adapted larval tiger salamander retina and found a dense band of staining in the outer nuclear layer involving the cell bodies of both rods and cones, with the heaviest staining near the base (close to the synaptic terminals) of each cell. The intensity decreased gradually toward the distal end of the inner segments. It seems possible that the light and dark adaptation processes could include redistribution of Zn21 within the photoreceptors. Photoreceptor function depends on an interaction with RPE cells (Nash and Osborne, 1996). Zinc is known to be associated with RPE cells, where it influences the activity of various enzymes (e.g., metallothioneins) (Cousins, 1985; Prasad, 1991a, 1991b; Tate et al., 1995), and may also act as an antioxidant (Bray and Bettger, 1990). It is also known that, when zinc is administered by mouth, it is partly taken up by RPE cells (Newsome et al., 1992), and zinc deficiency causes night blindness (Prasad, 1985). In the present study, we provide histochemical evidence to confirm that zinc is indeed present in high concentrations in the RPE cells (Fig. 2). Moreover, because we used albino rats, the zinc detected is most probably not associated with melanin (Potts and Au, 1976) in the RPE but rather with zincbinding proteins (e.g., metallothioneins) or phagosomes derived

from photoreceptors or both. It has been suggested that metallothioneins serve as an active site for the storage and release of zinc (Oliver et al., 1992). Zinc has also been suggested to be incorporated into the disc membranes in the photoreceptors (McCormick, 1985). Although the more sensitive autometallographic technique (to localize zinc-sulfide) showed zinc to be located in the plexiform layers (more intensily in the inner plexiform layer), cell-like structures in the ganglion cell layer, borders between the inner plexiform and nuclear layers, some Mu¨ller cells and perikarya in the nuclear layers of the rat retina, it is clear that zinc is not specifically associated with a distinct type of retinal neuron. The cell-like structures in the ganglion cell layer that contain zinc may be ganglion cells, but the resolution does not allow for a definite conclusion. The zinc staining in the inner plexiform layer and the heavier stain in the outer border of this area may be interpreted as suggesting that presynaptic neurons to ganglion cells contain zinc and that some of the zinc from these cells has been translocated into the “cells” in the ganglion cell layer. The perfusion of the rat with sulfide solution may have resulted in an excessive amount of glutamate being released from retinal neurons, as would occur in ischemia. One source of insight into this may be to carry out the perfusion at very low temperatures, administering an NMDA antagonist before perfusion and

302

FIGURE 7. The effect of 10 mM vigabatrin, the GABA transaminase inhibitor, on GABA immunoreactivity in the rabbit retina subjected to lack of glucose and oxygen (ischemic-like insults). Ischemic-like insults resulted in a reduction of GABA-containing perikarya and the appearance of three immunoreactive bands in the inner plexiform layer (A) (see also Fig. 3E). Vigabatrin caused the redistribution of the inhibitory neurotransmitter GABA in the tissues exposed to ischemic-like conditions; vigabatrin appears associated with the processes and cell bodies of Mu¨ller cells, owing to the inhibition of the metabolism of GABA in these cells. Scale bars530 mm. adapting animals to dark and light before localizing zinc. For the present, however, it must be concluded that there is a lack of evidence to suggest that zinc is present in high amounts in specific retinal neurons. Evidence from a variety of studies shows that zinc can act as a neuroprotective agent [for review, see Harrison and Gibbons, (1994)]. For example, a zinc-binding site is associated with the NMDA receptor, which on activation reduces the influx of calcium (Christine and Choi, 1990; Westbrook and Mayer, 1987; Wong and Kemp, 1991). Kikuchi et al. (1995) demonstrated that the neuroprotective effect of zinc against glutamate neurotoxicity in cultured retinal neurons is produced by its selective blockade of the NMDA receptor. In the present study, we confirm that this can occur. Retinal GABA neurons are known to have NMDA receptors (Cunningham and Neal, 1985; Huba and Hoffman, 1991; Osborne and Herrera, 1994) that, on activation, cause a release of GABA (Neal et al., 1994; O’Malley and Masland, 1989). The release of GABA can be measured biochemically (Neal et al., 1994) or monitored by “staining” for the localization of GABA (Osborne and Herrera, 1994; Osborne et al., 1995a, 1995b). In this study, we show that the NMDA-induced change in GABA immunoreactivity is prevented by the NMDA antagonist MK-801 (Fig. 4) and that low or moderate amounts of zinc have the same effect (Fig. 3). However, greater amounts of zinc have the opposite effect, exacerbating the “release” or change in the nature of the GABA “staining” caused by NMDA or inducing an effect directly on normal GABA immunoreactivity (Fig. 3). This suggests that low or moderate amounts of zinc can act as a neuroprotective agent by acting at the NMDA receptor but, when the zinc reaches a critical level, it is no longer beneficial but induces an insult to the retina. Such a conclusion has also been reached from studies conducted by other authors (Choi et al., 1988; Koh and Choi, 1988; Yokohama et al., 1986). Dozens of zinc-sensitive proteins and enzymes have been identified to date (Coleman, 1992; Prasad, 1991; Vallee and Auld, 1990, 1995). In the present study, we postulate that zinc may interact with a protein/enzyme that prevents the metabolism of GABA in Mu¨ller cells (Fig. 6). This conclusion is reached from the ischemic-like experiments in which the presence of low concentrations of zinc causes a redistribution of GABA immunoreactivity. Ischemic-like insults cause a “release” of GABA and subsequent change in the nature of GABA immunoreactivity (see Fig. 3). The

M. Ugarte and N. N. Osborne “released” GABA is taken up by Mu¨ller cells and metabolized rapidly to glutamine, which is then transported back to the neurons (Fig. 6) (Robin and Kalloniatis, 1992). However, when low amounts of zinc are present (Fig. 3), GABA immunoreactivity is found to be associated with Mu¨ller cells. An identical distribution of GABA occurs immediately after retinal ischemia when no energy is available to metabolize GABA to glutamine (Barnett and Osborne, 1995) or when the enzyme GABA transaminase is inhibited by vigabatrin (Neal et al., 1989) (Fig. 7). Thus it would appear that zinc can prevent the metabolism of GABA in glial cells and influence the GABA-glutamine shuttle between neurons and glial cells. This would be potentially harmful in, for example, ischemia. There are a number of common mechanisms in necrotic cell death induced by toxicity due to NMDA, ischemia, and so forth, including the formation of ROS, which are destructive when present at an elevated level (Bondy and LeBel, 1990). In this study, we provide clear evidence that any neuroprotective action of zinc is not due to a mechanism that reduces formation of ROS. The contrary seems to be the case when zinc is present at high levels. In this case, the zinc can stimulate the formation of destructive ROS. The conclusion from the present series of experiments is that low amounts of zinc can not only protect against NMDA toxicity, but also have a negative effect in that zinc disrupts the GABA–glutamine shuttle between neurons and glial cells. When the level of zinc is high, the effect is detrimental, stimulating the formation of ROS and release of GABA. Endogenous zinc is mainly localized in retinal photoreceptors and its associated RPE cells with none of the other retinal neurons containing a large amount of zinc. We are grateful to the Human Capital and Mobility Programme of the European Community for financial support.

References

Assaf S. Y. and Chung S. H. (1984) Release of endogenous Zn21 from brain tissue during activity. Nature 308, 734–736. Barnett N. L. and Osborne N. N. (1995) Redistribution of GABA immunoreactivity following central retinal artery occlusion. Brain Res. 677, 337– 340. Bondy S. C. and LeBel C. P. (1993) The relationship between excitotoxicity and oxidative stress in the central nervous system. Free Radical Biol. Med. 14, 633–642. Bray T. M. and Bettger W. J. (1990) The physiological role of zinc as an antioxidant. Free Radical Biol. Med. 8, 281–291. Buchi E. R. (1992) Cell death in the rat retina after a pressure-induced ischemia-reperfusion insult: an electron microscopic study I: ganglion cell layer and inner nuclear layer. Exp. Eye Res. 55, 605–613. Choi D. W., Yokoyama M. and Koh J. (1988) Zinc neurotoxicity in cortical cell culture. Neuroscience. 24, 67–79. Christine C. W. and Choi D. W. (1990) Effect of zinc on NMDA receptormediated channel currents in cortical neurons. J. Neurosci. 10, 108–116. Coleman J. E. (1992) Zinc proteins: enzymes, storage proteins, transcription factors, and replication proteins. Annu. Rev. Biochem. 61, 897–946. Cousins R. J. (1985) Absorption, transport, and hepatic metabolism of copper and zinc: special reference to metallothionein and ceruloplasmin. Physiol. Rev. 65, 238–309. Cunningham J. R. and Neal M. J. (1985) Effect of excitatory amino acids and analogues on [3H]acetylcholine release from amacrine cells of the rabbit retina. J. Physiol. (Lond.) 366, 47–62. Danscher, G. (1996) The autometallographic zinc-sulphide method: a new approach involving in vivo creation of nanometer-sized zinc crystal lattices in zinc-enriched synaptic and secretory vesicles. Histochem. J. 28, 361–373. Daw N. W., Brunken W. J. and Parkinson D. (1989) The function of synaptic transmitters in the retina. Annu. Rev. Neurosci. 12, 205–225. Eckert C. D. (1979) A comparative study of the concentrations of Ca, Fe, Zn and Mn in ocular tissue. (Abstract.) Fed. Proc. 38, 392. Frederickson C. J. (1989) Neurobiology of zinc and zinc-containing neurons. Int. Rev. Neurobiol. 31, 145–238.

Zinc and ROS in the Retina Frederickson C. J. and Moncrieff D. W. (1994) Zinc-containing neurons. Biol. Signals 3(3), 127–139. Frederickson C. J., Kasarskis E. J., Ringo D. and Frederickson R. E. (1987) A quinoline fluorescence method for visualizing and assaying the histochemically reactive zinc (bouton zinc) in the brain. J. Neurosci. Methods. 20(2), 91–103. Harrison N. L. and Gibbons S. J. (1994) Zn21: an endogenous modulator of ligand- and voltage-gated ion channels. Neuropharmacology 33, 935–952. Hirayama Y. (1990) Histochemical localization of zinc and copper in rat ocular tissues. Acta Histochem. 89, 107–111. Howell G. A., Welch M. G. and Frederickson C. J. (1984) Stimulationinduced uptake and release of zinc in hippocampal slices. Nature 308, 736–738. Huba R. and Hofmann H. D. (1991) Transmitter-gated currents of GABAergic amacrine-like cells in chick retinal cultures. Visual Neurosci. 6, 303–314. Khulusi S. S., Brown M. W. and Wright D. M. (1986) Zinc and paired-pulse potentiation in the hippocampus. Brain Res. 363, 152–155. Kikuchi M., Kashii S., Honda Y, Ujihara H., Sasa M., Tamura Y. and Akaike A. (1995) Protective action of zinc against glutamate neurotoxicity in cultured retinal neurons. Invest. Ophthalmol. Visual Sci. 36, 2048–2053. Koh J. Y. and Choi D. W. (1988) Zinc alters excitatory amino acid neurotoxicity on cortical neurons. J. Neurosci. 8, 2164–2171. Koh J. Y., Suh S. W., Gwag B. J., He Y. Y., Hsu C. Y. and Choi D. W. (1996) The role of zinc in selective neuronal death after transient global cerebral ischemia. Science 272, 1013–1016. LeBel C. P. and Bondy S. C. (1990) Sensitive and rapid quantitation of oxygen reactive species formation in rat synaptosomes. Neurochem. Intern. 17, 435–440. Massey S. C. (1990) Cell types using glutamate as a neurotransmitter in the vertebrate retina. In Progress in Retinal Research (Edited by Osborne N. N. and Chader G. J.), Vol. 9, pp. 339–426. Pergamon Press, Oxford. McCormick L. D. (1985) Bound trace element content of bovine retinal disk membranes as determined by particle-induced x-ray emission. Biophys. J. 47, 381–385. Michikawa M., Lim K. T., McLarnon J. G. and Kim S. U. (1994) Oxygen radical-induced neurotoxicity in spinal cord neuron cultures. J. Neurosci. Res. 37, 62–70. Nash M. S. and Osborne N. N. (1996) Cell surface receptors associated with the retinal pigment epithelium: the adenylate cyclase and phospholipase C signal transduction pathways. In Progress in Retinal Research (Edited by Osborne N. N. and Chader G. J.), Vol. 15, pp. 501–546. Pergamon Press, Oxford. Neal M. J., Cunningham J. R., Shah M. A. and Yazulla S. (1989) Immunocytochemical evidence that vigabatrin in rats causes GABA accumulation in glial cells of the retina. Neurosci. Lett. 98, 29–32. Neal M. J., Cunningham J. R., Hutson P. H. and Hogg J. (1994) Effects of ischaemia on neurotransmitter release from the isolated retina. J. Neurochem. 62, 1025–1033. Newsome D. A., Oliver P. D., Deupree D. M., Miceli M. V. and Diamond J. G. (1992) Zinc uptake by primate retinal pigment epithelium and choroid. Curr. Eye. Res. 11, 213–217. Nickells R. W. (1996) Retinal ganglion cell death in glaucoma: the how, the why, and the maybe. J. Glaucoma 5, 345–356. Oliver P. D., Tate D. J. Jr. and Newsome D. A. (1992) Metallothionein in human retinal pigment epithelial cells: expression, induction and zinc uptake. Curr. Eye Res. 11, 183–188. O’Malley D. M. and Masland R. H. (1989) Co-release of acetylcholine and gamma-aminobutyric acid by a retinal neuron. Proc. Natl. Acad. Sci. USA 86, 3414–3418. Osborne N. N. and Herrera A. J. (1994) The effect of experimental is-

303 chaemia and excitatory amino acid agonists on the GABA and serotonin immunoreactivities in the rabbit retina. Neuroscience 59, 1071–1081. Osborne N. N., Larsen A. and Barnett N. L. (1995a) Influence of excitatory amino acids and ischaemia on rat retinal choline acetyltransferase-containing cells. Invest. Ophthalmol Visual Sci. 36, 1279–1288. Osborne N. N., Wood J. and Mu¨ller A. (1995b) The influence of experimental ischemia on protein kinase C and the GABAergic system in the rabbit retina. Neuropharmacology 34, 1279–1288. Peng Y. W., Blackstone C. D., Huganir R. L. and Yau K. W. (1995) Distribution of glutamate receptor subtypes in the vertebrate retina. Neuroscience 66, 483–497. Peters S., Koh J. and Choi D. W. (1987) Zinc selectively blocks the action of N-methyl-d-aspartate on cortical neurons. Science 236, 589–593. Potts A. M. and Au P. C. (1976) The affinity of melanin for inorganic ions. Exp. Eye Res. 22, 487–491. Prasad A. S. (1985) Laboratory diagnosis of zinc deficiency. J. Am. Coll. Nutr. 4, 591–598. Prasad A. S. (1991a) Role of zinc in human health. Bol. Asoc. Med. P. R. 83, 558–560. Prasad A. S. (1991b) Discovery of human zinc deficiency and studies in an experimental human model. Am. J. Clin. Nutr. 53, 403–412. Robin L. N. and Kalloniatis M. (1992) Interrelationship between retinal ischaemic damage and turnover and metabolism of putative amino acid neurotransmitters, glutamate and GABA. Doc. Ophthalmol. 80, 273– 300. Selles-Navarro I., Villegas-Pe´rez M. D., Salvador M. and Vidal-Sanz M. (1994) Short- and long-term effects of ischaemia-induced retinal ganglion cell death: a quantificative study. Abstract No. 123, JERMOV Meeting, Montpellier. Selles-Navarro I., Villegas-Perez M. P., Salvador-Silva M., Ruiz-Gomez J. M. and Vidal-Sanz M. (1996) Retinal ganglion cell death after different transient periods of pressure-induced ischemia and survival intervals: a quantitative in vivo study. Invest. Opthalmol. Visual Sci. 37, 2002–2014. Slomianka L. (1992) Neurons of origin of zinc-containing pathways and the distribution of zinc-containing boutons in the hippocampal region of the rat. Neuroscience 48, 325–352. Tate D. J., Miceli M. V., Newsome D. A., Alcock N. W. and Oliver P. D. (1995) Influence of zinc on selected cellular functions of cultured human retinal pigment epithelium. Curr. Eye Res. 14, 897–903. Vallee B. L. and Auld D. S. (1990) Zinc coordination, function, and structure of zinc enzymes and other proteins. Biochemistry 29, 5647–5659. Vallee B. L. and Auld D. S. (1995) Zinc metallochemistry in biochemistry. Exper. Suppl. 73, 259–277. Verity M. A., Bredesen D. E. and Sarafian T. (1995) Role of reactive oxygen species (ROS) in neuronal degeneration: modulation by protooncogene expression. Ann. N. Y. Acad. Sci. 765, 340. Westbrook G. L. and Mayer M. L. (1987) Micromolar concentrations of Zn21 antagonize NMDA and GABA responses of hippocampal neurons. Nature 328, 640–643. Winegar B. D. and Lansman J. B. (1990) Voltage-dependent block by zinc of single calcium channels in mouse myotubes. J. Physiol. (Lond.) 425, 563–578. Wong E. H. and Kemp J. A. (1991) Sites for antagonism on the N-methyld-aspartate receptor channel complex. Annu. Rev. Pharmac. Toxicol. 31, 401–425. Wu S. M. (1994) Synaptic transmission in the outer retina. Annu. Rev. Physiol. 56, 141–168. Wu S. M., Qiao X., Noebels J. L. and Yang X. L. (1993) Localization and modulatory actions of zinc in vertebrate retina. Vision Res. 33, 2611– 2616. Yokoyama M., Koh J. and Choi D.W. (1986) Brief exposure to zinc is toxic to cortical neurons. Neurosci. Lett. 71, 351–355.