Cytoprotection by endogenous zinc in the vertebrate retina


  • Ivan Anastassov,

    1. Department of Biological Sciences, Hunter College and the Graduate Center, CUNY, New York, New York, USA
    2. Marine Biological Laboratory, Woods Hole, Massachusetts, USA
    Current affiliation:
    1. Department of Biochemistry and Molecular Biology, Baylor College of Medicine, Houston, USA
    Search for more papers by this author
  • Harris Ripps,

    1. Marine Biological Laboratory, Woods Hole, Massachusetts, USA
    2. Departments of Ophthalmology and Visual Sciences, Anatomy and Cell Biology, and Physiology and Biophysics, University of Illinois College of Medicine, Chicago, Illinois, USA
    Search for more papers by this author
  • Richard L. Chappell

    Corresponding author
    1. Department of Biological Sciences, Hunter College and the Graduate Center, CUNY, New York, New York, USA
    2. Marine Biological Laboratory, Woods Hole, Massachusetts, USA
    3. Center for Vision Research, Brown Institute for Brain Science, Providence, Rhode Island, USA
    • Address correspondence and reprint requests to Dr Richard L. Chappell, Marine Biological Laboratory, 7 MBL Street, Woods Hole, MA 02543, USA. E-mail:

    Search for more papers by this author


Our recent studies have shown that endogenous zinc, co-released with glutamate from the synaptic terminals of vertebrate retinal photoreceptors, provides a feedback mechanism that reduces calcium entry and the concomitant vesicular release of glutamate. We hypothesized that zinc feedback may serve to protect the retina from glutamate excitotoxicity, and conducted in vivo experiments on the retina of the skate (Raja erinacea) to determine the effects of removing endogenous zinc by chelation. These studies showed that removal of zinc by injecting the zinc chelator histidine results in inner retinal damage similar to that induced by the glutamate receptor agonist kainic acid. In contrast, when an equimolar quantity of zinc followed the injection of histidine, the retinal cells were unaffected. Our results are a good indication that zinc, co-released with glutamate by photoreceptors, provides an auto-feedback system that plays an important cytoprotective role in the retina.


Recent studies show that endogenous photoreceptor zinc, co-released with glutamate, provides feedback that reduces calcium entry at photoreceptor terminals, thereby regulating vesicular transmitter release. The experiments reported here provide evidence that the in vivo removal of endogenous zinc by chelation results in inner retinal glutamate toxicity. Thus, zinc auto-feedback likely plays a cytoprotective role important to retinal health and disease.

Abbreviations used



ethidium homodimer III


ganglion cell layer


inner nuclear layer


inner plexiform layer

A unique feature of the vertebrate photoreceptor is that, in darkness, a sustained inward cation current holds the cell in a depolarized state. Consequently, the photoreceptor continually releases its neurotransmitter, glutamate, into the synaptic cleft. It has long been known that excessive levels of glutamate are toxic to retinal neurons (Reif-Lehrer et al. 1975), and as shown by Rothman (1984), over-activation of glutamate receptors results in the death of neurons in culture. These findings were extended to the intact isolated retina and other nerve centers in the CNS by Olney and co-workers (cf. Olney 1982, 1994; Romano et al. 1998). In a related series of experiments, we attempted to determine whether free zinc (Zn2+), packaged within the synaptic vesicles of the photoreceptor terminal and co-released with glutamate (Redenti and Chappell 2004, 2005; Redenti et al. 2007; Chappell et al. 2008), can serve a neuromodulatory role. This notion was based on results by Wu and co-workers (Wu et al. 1993), showing decreased calcium entry into photoreceptor terminals when exogenous zinc was applied to the salamander retina preparation. We have since confirmed these results, and in addition have shown that using chelators to remove endogenous zinc leads to a marked increase in both calcium entry (Anastassov et al. 2013) and in the photoreceptor's dark current (Chappell et al. 2008). These findings indicate that a reduced zinc concentration results in a concomitant increase in the discharge of glutamate, and led us to suggest that if endogenous zinc can suppress transmitter release, it may serve to protect the retina from the toxic effects of glutamate. In addition, we wished to determine how cell death evolves in the intact retina when challenged with kainate (cf. Olney et al.1974) an analog of glutamate and a potent activator of a subset of highly sensitive ionotropic glutamate receptors even at extremely low concentrations (Shen et al. 2004). To examine these issues, it was important to perform in vivo experiments on dark-adapted retinas in which kainate is introduced into the vitreal chamber adjacent to the retina, and also to study the pathological changes that occur when zinc is chelated and glutamate is continuously released.

Materials and methods

All surgical and animal handling procedures were conducted in accordance with ARRIVE guidelines.


Adult skates (Raja erinacea), members of the family of cartilaginous elasmobranchs that includes sharks and rays, were used in these experiments. Prior to injections, the fish were anesthetized with 0.02% tricaine methane sulfonate (Argent Chemical, Redmond, WA), and a local anesthetic (2% lidocaine) was applied to the cornea. Injections into the vitreal space were performed using a 33 gage (6.35 mm) Hamilton micro-syringe. Drugs were dissolved in skate Ringer solution (in mM): NaCl (250), KCl (6), CaCl2 (4), Urea (360), D-glucose (10), NaHCO3 (20), MgCl2 (4), NaH2PO4 (0.2), HEPES (5), pH 7.6; and injected in one (test) eye. The contralateral eye was either not injected or received vehicle only. Injections were made posterior to the lens, along the horizontal midline of the eye and approximately over the tapetal area of the retina. The total volume (control or test) per injection was 10 μL. To maximize vesicular release from photoreceptors, fish were kept in the dark for the entire experiment. In short: pre-injection, the animals were dark-adapted for 8–12 h; following adaptation, an injection was performed under dim red illumination every 2 h (total of 5 injections over 10 h); finally, the animals were left to recover in darkness for another 12 or 24 h post-injection. Tissue was collected after the recovery stage. Drugs and chemicals were purchased from Sigma-Aldrich, St. Louis, MO, USA (kainate, histidine) and VWR, Radnor, PA, USA (Ringer components).

Detection of Necrosis

Eyes were enucleated under dim red light, the cornea and lens removed, and the vitreous drained with cotton wicks. To visualize necrosis, control and drug-treated eyes from the same animal were simultaneously incubated with the nucleic acid probe ethidium homodimer III (EthD-III), which is impermeant to healthy cells with uncompromised membranes, but stains cells whose membranes have lost their integrity. To minimize the chance of false positives resulting from significant cell death as a result of tissue removal, the EthD-III was applied to the eye cup immediately after enucleation and the incubation was done on ice. Control and treatment eyes from the same animal were removed and incubated with the probe simultaneously. The EthD-III was dissolved in proprietary Binding Buffer, as per the manufacturer's instructions (PromoCell, Heidelberg, Germany). After staining, the eye cup was fixed with 2% paraformaldehyde in the dark, on ice for 1 h. Eye cups were cryoprotected by incubation in 30% sucrose in phosphate-buffered saline at 4°C overnight, and protected from light to avoid bleaching of the fluorescent signal. Cryoprotected tissue was cut into pieces, embedded in O.C.T. (Tissue Tek, Torrence, CA) and flash frozen. Blocks of frozen tissue were stored at −80°C and sectioned at −25°C on a Leica CM 1850 cryostat in 14–18 μM sections. To prevent photobleaching of the necrosis signal, we used Vectashield mounting media (Vector Labs, Burlingame, CA, USA) that already contains 4′,6-diamidino-2-phenylindole (DAPI) at a concentration of 1.5 μg/mL. As per manufacturer's instructions, 1 drop of ~ 25 μL from the provided dispenser was used to cover each section on every slide. Slides were examined with LSM 710 and LSM 780 Zeiss confocal microscopes (Carl Zeiss Microscopy GmbH, Jena, Germany). Images were collected and analyzed using Zeiss ZEN imaging software.

Data analysis and statistical methods

The ratio of EthD-II and DAPI fluorescence intensity was computed from raw images of retinal sections. A transect line was drawn across each retinal layer and the fluorescence ratio for each pixel on the line was computed. The highest and lowest 5% of values were excluded because these were likely to represent 0 in either numerator or denominator, indicating non-specific or debris staining not associated with cell bodies. 2-factor anova was conducted with Prism 6.0 (GraphPad Software, Inc., La Jolla, CA, USA) to compare ratios between treatments and across layers. Post hoc comparisons were made with paired, Bonferroni-corrected t-tests for comparisons across layers within treatment groups, and unpaired, corrected t-tests for comparisons of the same layers in different treatment groups. All p values reported are two-tailed.


Eyes were enucleated under dim red light, the cornea and lens removed and the vitreous drained. For histological observation, the retina was left attached to the choroid and cartilaginous sclera to protect it from damage and aid in subsequent sectioning. The resulting eyecup was immediately fixed with 2% paraformaldehyde/2% Glutaraldehyde in 0.1 M Cacodylate buffer with 3% sucrose and 4.5 mM CaCl2 followed by post-fixation with 1% OsO4. Dehydration was done with ethanol steps (30–100%) and rotation in propylene oxide. Tissue was infiltrated with a 1 : 1 mixture of propylene oxide and EMbed 812 (Electron Microscopy Sciences, PA, USA) followed by fresh EMbed 812. Polymerization was done at 60°C for 24–48 h. Blocks of embedded tissue were trimmed by hand and subsequently sectioned (0.5–1.0 μm) on a Reichart Jung Ultracut E microtome. Sections were stained with 1% Methylene Blue/1% Azure II/1% Sodium borate solution for 20 s on a gently heated hotplate. Sections were mounted and sealed and images were taken with a Zeiss AxioImager Z2 light microscope (Carl Zeiss Microscopy GmbH) running AxioVision software.


In this brief report, we present representative images of the results obtained from the study of eyes injected with Ringer (control), 20 mM histidine, 2 mM kainate, or with 10 mM histidine + 10 mM zinc. Eyes were prepared either for necrosis analysis (early onset cell death) or histological study (to reveal the gross cellular damage that follows).

Figure 1(a and b) shows results obtained following intraocular injections of Ringer in the eye (control) of an experimental animal. There is very limited co-staining with DAPI and EthD-III (Fig. 1a). A graphical representation of the peak DAPI (blue) and EthD-III (red) signals across a large section of the retina is shown in Fig. 1(b). Note the relatively small degree of necrosis, that is cells showing EthD-III signals.

Figure 1.

Kainate and histidine toxicity evidenced by necrosis in the skate retina. (a) Cross-section of skate retina from a control eye receiving an injection of Ringer solution. DAPI staining of cell nuclei (blue) identifies cell bodies of the outer nuclear layer (ONL), inner nuclear layer (INL), and ganglion cell layer (GCL). Overlay of the image (red) identifying necrotic cells with ethidium homodimer III (EthD-III) shows that relatively few cells are necrotic. (b) The DAPI and EthD-III fluorescence signals measured across the entire GCL for the Ringer control condition in (a) are overlaid and displayed graphically. (c) Similar staining of a retinal cross-section from an experimental eye injected with 2 mM kainate reveals pronounced EthD-III staining in the GCL. (d) The DAPI and EthD-III fluorescence signals measured across the entire GCL for tissue from the 2 mM kainate treated eye; peaks in fluorescence of blue and red traces represent nuclei stained with DAPI and EthD-III, respectively. (e) Skate retina cross-section from experimental eye which received 20 mM histidine injection. EthD-III staining indicates that there is significant cell death by necrosis in the GCL. (f) The DAPI and EthD-III fluorescence signals measured across the entire GCL for tissue from the 20 mM histidine treated eye. As before, peaks in fluorescence of blue and red traces in (f) represent nuclei stained with DAPI and EthD-III, respectively. Scale bars = 50 μm.

In contrast, injections of 2 mM kainate in the contralateral eye resulted in widespread necrosis of cells in the ganglion cell layer (GCL, Fig. 1c and d). Surprisingly, little necrosis was seen in the inner nuclear layer (INL) and outer nuclear layer. Results similar to those obtained with kainate were seen when the retina was injected with 20 mM histidine. As shown in Fig. 1(e) and (f), significant degree of necrosis was observed after chelation of endogenous zinc by the histidine injection. Statistical analyses of these experimental results, illustrated by the bar graphs in Fig. 2(a) and (b), confirm the findings that both kainate and histidine induce a significantly greater degree of necrosis (2-way anova, Bonferroni post hoc test, p < 0.0001, n = 9 for Ringer/kainate, n = 10 for Ringer/histidine) in the GCL when compared with the Ringer-injected eye. Cells in the INL and outer nuclear layer showed little to no necrosis following injection of either drug.

Figure 2.

Statistical analysis of necrotic cell death in the retinal nuclear layers. The Ringer control is compared with eyes injected with kainate or histidine. (a) Kainate results based on data from Fig. 1(c) and (d). Cells in the ganglion cell layer (GCL) are significantly more necrotic after injection of 2 mM kainate when compared with contralateral Ringer control (two-way anova, Bonferroni post hoc test, ***p < 0.0001, n = 9; for Ringer: mean ± SEM = 1.541 ± 0.237, for kainate: mean ± SEM = 2.742 ± 0.368). (b) Cells in the GCL are significantly more necrotic as a result of 20 mM histidine when compared with contralateral Ringer control (two-way anova, Bonferroni post hoc test, ****p < 0.0001, n = 10; for Ringer: mean ± SEM = 0.299 ± 0.035, for histidine: mean ± SEM = 0.523 ± 0.039).

Figure 3(a) shows the normal retinal structure of a Ringer-injected eye. There are no signs of damage; all nuclear and plexiform layers appear normal, and the cells are structurally intact. In contrast, Fig. 3(b) illustrates the abnormal cellular structure, seen throughout the inner retina, that results from exposure to 2 mM kainate. Cell bodies in the INL and GCL are swollen, and processes in the inner plexiform layer (IPL) appear to have lost their protoplasmic content, giving the IPL a ‘sponge-like’ appearance.

Figure 3.

Drug-induced morphological changes in the skate retina. (a) A retinal section from a control eye injected with Ringer alone. Note the absence of cell swelling or other abnormalities in the inner nuclear layer (INL), IPL and ganglion cell layer (GCL). (b) In vivo intraocular injections of the glutamate receptor agonist kainate (2 mM) result in cellular changes typical of glutamate toxicity. There is pronounced swelling of cells of the INL and GCL and the structural integrity of their cellular membranes appears compromised. The inner plexiform layer exhibits the typical ‘sponge-like’ appearance and loss of cell material associated with apoptosis. (c) Intraocular injections of 20 mM histidine, a membrane-impermeable chelator of zinc, result in histological damage throughout the retina. Cells of the INL and GCL are swollen and normal morphology is lost. The IPL has the characteristic ‘sponge-like’ appearance associated with the loss of cell cytoplasm. Note also that the photoreceptor outer segments and cell bodies are contorted and appear less dense, but this effect was not seen in every animal. (d) When both zinc (10 mM) and then histidine (10 mM) were injected separately into the experimental eye, the glutamate-like toxicity found following injection of histidine alone was not observed. In tissue from that eye, the nuclear layers and plexiform layers are clearly visible and appear normal. OS, outer segments; IS, inner segments; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer. Scale bars = 50 μM.

We have shown that the zinc released from photoreceptor terminals feeds back to reduce calcium entry, and thereby inhibits exocytosis of glutamate-containing vesicles. On this view, chelation of endogenous zinc could lead to an increased release of glutamate, and result in cytotoxic effects similar to those seen with kainate. As shown in Fig. 3(c), zinc chelation by 20 mM histidine (injections were done as before, i.e. 10 μL given every 2 h over a period of 10 h; tissue collection after recovery period) resulted in gross morphological changes in the retina. Cells in the INL and GCL were swollen, and the IPL was severely affected. There is a significant loss of cellular material as evidenced by the sparse staining of the tissue, and the IPL has the same ‘sponge-like’ appearance as observed with kainate. Moreover, the inner limiting membrane has been disrupted, and there is a marked loss of ganglion cells. Note that photoreceptor structure also shows signs of damage, although this was not observed in all animals.

Results illustrated in Fig. 3(d) show the cytoprotective effects of exogenous zinc, when introduced in equimolar amounts 2–3 min following each of the five injections of the zinc chelator histidine (10 mM each, every 2 h). In these circumstances, no appreciable retinal damage was observed, confirming that it is the removal of endogenous zinc by histidine, and not histidine itself, that caused the cytotoxic effects observed in Fig. 1(e) and 3(c).


Excitotoxicity in the CNS is a process considered to lead to neuronal cell death through the over-activation of glutamate receptors (Dong et al. 2009). It is understood that ionotropic glutamate receptors of the NMDA and non-NMDA type (AMPA and kainate) are the predominant mediators of excitotoxicity (Sattler and Tymianski 2001), but there is some evidence to suggest that metabotropic glutamate receptors may also be involved (McDonald et al. 1993). Similarly, glutamate excitotoxicity in the vertebrate retina is a well-known phenomenon, described over the years in a number of studies (Olney 1982, 1994; Izumi et al. 1995, 2002; Chen et al. 1998). In the retina, NMDA, AMPA, and kainate have proven to be potent agents of excitotoxicity, where treatment of the isolated tissue ex vivo with those compounds quickly leads to damage characterized by cell swelling, pyknosis, and spongy appearance of the inner plexiform layer (Olney et al. 1974; Olney 1994).

In the experiments described here, it was necessary to modify and adapt some of the above methodologies. The goal was to chelate endogenous zinc in vivo and monitor the effects of zinc removal on retinas exposed to the unregulated release of glutamate from photoreceptors in darkness. To achieve this, we used the skate (Raja erinacea), and an important consideration before starting the zinc chelation experiments was the need to confirm the viability of the skate as a model system for excitotoxicity. To that end, intraocular injections of kainate were performed and the tissue examined microscopically. The morphological changes observed following this protocol included pronounced tissue damage and cell swelling and a very characteristic spongy appearance of the inner plexiform layer. Our results confirmed that the skate retina undergoes excitotoxic events very similar to the ones observed by other groups in the retinas of chick, rat and mouse (Olney et al. 1986; Romano et al. 1998).

Severe insult to neuronal tissues usually results in necrosis and/or apoptosis, but the time course of these indices of cell death seems to vary widely depending on the type of neuronal tissue and the species of animal used. Based on the numerous studies by Olney and co-workers (cited above), it is likely that in the retina the majority of affected cells are undergoing necrosis within the first 24 h after exposure to cytotoxic levels of glutamate. This is followed by more extensive destructive changes over a broad extent of the retina, characteristic of apoptotic cell death (Gwag et al. 1997; Joo et al. 1999; Ientile et al. 2001). Looking quantitatively at the early effects of kainate on cell death of the various cell groups within the skate retina, we found substantial evidence of necrosis in the GCL with very little necrosis observed elsewhere. It is likely that apoptosis will follow these early necrotic events, as previously reported for goldfish retina (Villani et al. 1995, 1997).

We had suggested that negative feedback control of glutamate release by zinc serves not only a neuromodulatory, but also a cytoprotective role in the vertebrate retina (Chappell et al. 2008). To test this hypothesis, we performed injections of histidine, a membrane-impermeable chelator of zinc, into the eye of the dark-adapted skate. This approach allowed us to continuously chelate zinc in vivo to learn its effect while the animal was kept in the dark, when glutamate release from photoreceptors is maximal. If zinc does indeed regulate glutamate release, then over time, with chelation of zinc, we would expect to see the retina suffer from the effects of excessive glutamate exposure, that is excitotoxicity. Injections of histidine (20 mM) did indeed result in morphological changes in the tissue that very strongly resembled what was observed with kainate injections, and in fact sometimes exceeded the effects of kainate. As was the case for kainate injections, early necrotic changes were observed almost exclusively in the GCL. At later times, tissue swelling, loss of cell material and spongy appearance of the IPL similar to kainate treatment was observed histologically.

Finally, in control experiments, we found that separate injections of identical concentrations of the chelator histidine and zinc result in no significant damage to the tissue. This demonstrates the specificity of action of histidine in that it does not appear to act directly on the tissue, but rather that its effect is accomplished by the chelation of zinc and the subsequent unregulated glutamate release. Interestingly, exogenous zinc has been shown to have a modulatory effect on bipolar and amacrine cells (Han and Yang 1999; Luo et al. 2002; Zhang et al. 2002). In light of our experiments, these findings do not exclude the possibility of Zn2+ diffusion from the outer to the inner retina. To our knowledge there is as yet no evidence for zinc and glutamate co-release from the bipolar cell terminal, perhaps because of the differences in the nature of transmitter release from photoreceptors and bipolar cells.

It should be noted that despite the millimolar concentrations listed in the figures, the actual concentration of drugs reaching the tissue was likely at least 20 to 50 times less, because of the large volume of the skate eye, the small volume of treatment injections, and the highly viscous vitreous humor. Effectively, we believe that actual concentrations of kainate and histidine were very likely in the micromolar range. Based on histological observations in experiments like those shown in Fig. 3, we were expecting a large number of cells to exhibit signs of necrosis throughout the retina. To our surprise, relatively little evidence for cell necrosis was observed in the outer retina. On the other hand, significant necrosis was observed within the GCL. This finding is consistent, however, with earlier reports that excitotoxic damage to the inner retina is generally what is first observed in other animal models of glutamate toxicity, that is in avian and murine retinas (Romano et al. 1998).

A study by Kikuchi et al. (1995) demonstrated the protective action of zinc against the neurotoxic effects of glutamate on retinal neurons in culture. In an alternative approach, Hyun et al. (2001) showed that zinc depletion, induced by treatment with a membrane permeable chelator, rendered cultured human retinal pigment epithelial cells highly vulnerable to cell death from UV radiation or exposure to hydrogen peroxide. Furthermore, zinc deficiency has serious consequences, and the resultant pathology has been well documented in a wide variety of tissues (reviewed in Prasad 2013). The results reported here suggest that zinc, co-released endogenously with glutamate from vertebrate photoreceptor terminals, plays an important role in protecting the retina from the excitotoxic damage that results from unregulated tonic release of glutamate in darkness, that is when photoreceptors are maximally depolarized. Thus, zinc deficiency may have subtle but serious consequences for the health of the inner retina.

Acknowledgements and Conflict of interest disclosure

We thank Dr Robyn Crook for her help with statistical analysis and with preparation of the necrosis data. We thank Marjeta Argjir for help with tissue sectioning. These studies were supported by grants from the National Science Foundation (1026531 & 1214162: RC) and NCRR/NIH (RR003037: RC). The authors have no disclosures or conflicts to be reported.