Depression of Retinal Glutamate Transporter Function Leads to Elevated Intravitreal Glutamate Levels and Ganglion Cell Death

Abstract

purpose. Elevated levels of extracellular glutamate have been implicated in the pathophysiology of neuronal loss in both central nervous system and ophthalmic disorders, including glaucoma. This increase in glutamate may result from a failure of glutamate transporters (molecules that ordinarily regulate extracellular glutamate; Excitatory Amino Acid Transporter; EAAT). Elevated glutamate levels can also lead to alterations in glutamate receptor expression. It was hypothesized that selective blockade of glutamate transporters would be toxic to retinal ganglion cells.

methods. Glutamate transporters were blocked either pharmacologically or with subtype-specific antisense oligonucleotides against EAAT1. Glutamate levels, transporter levels and ganglion cell survival were assayed.

results. Pharmacological inhibition of glutamate transporters with either an EAAT2 specific inhibitor or a nonspecific inhibitor of all the subtypes of transporters was toxic to ganglion cells. Treatment with oligonucleotides against the glutamate transporter EAAT1 decreased the levels of expression of the transporter, increased vitreal glutamate, and was toxic to ganglion cells.

conclusions. These results demonstrate that normal function of EAAT1 and EAAT2 is necessary for retinal ganglion cell survival and plays an important role in retinal excitotoxicity. Manipulation of retinal glutamate transporter expression may become a useful tool in understanding retinal neuronal loss.

Glutamate is the principal excitatory neurotransmitter in the mammalian central nervous system. ¹ Excessive levels of extracellular glutamate have been implicated in the pathogenesis of many neurologic and ophthalmic diseases, including stroke, trauma, epilepsy, dementia, and glaucoma. ^{1 2 3 4} Glutamate can damage neurons via an “excitotoxic” pathway, mediated primarily through the N-methyl-d-aspartate (NMDA) subtype of glutamate receptor. ¹ Excitotoxic neuronal loss is frequently associated with elevated levels of extracellular glutamate, which can contribute to additional neuronal loss. ^{5 6 7} Increased extracellular glutamate is assumed to result from the failure of glutamate transporters, the reversed operation of glutamate transporters, and the death of neurons with subsequent release of intracellular contents (containing approximately 10 M glutamate). ⁸ Under normal conditions, however, glutamate transporters, located in the plasma membrane of both neurons and glia, rapidly transport glutamate into the intracellular space and maintain physiological glutamate concentrations. ⁹ In the developing mammalian retina, for example, up to 50% of the retinal ganglion cells (RGCs) die by programmed cell death with no significant elevation in extracellular glutamate. ¹⁰ Synaptically released glutamate can also reach high local concentrations with no demonstrable toxicity. Consequently, if elevated extracellular glutamate is involved in neuronal loss, one must consider the possibility of a transporter abnormality. Glutamate transporter malfunction may contribute to the neuronal loss seen in amyotrophic lateral sclerosis, dementia, stroke, and glaucoma. ^{6 11 12 13 14}  

Five excitatory amino acid transporters have been identified. In the retina, EAAT1 (Excitatory Amino Acid Transporter; GLAST) is found in Müller cells and astrocytes. ¹⁵ EAAT2 (GLT-1) is localized to cones and two types of bipolar cells. ¹⁶ EAAT3 (EAAC1) is found on horizontal, amacrine, and ganglion cells and, rarely, on bipolar cells. ¹⁷ EAAT5 is localized to photoreceptors and bipolar cells ¹⁸ ; EAAT4 has not been found in retinal tissue. We have previously demonstrated a loss of EAAT1 in human glaucomatous tissue, consistent with the elevated glutamate found in glaucomatous vitreous. ¹¹  

To evaluate the role of retinal glutamate transporters we used both pharmacological agents and antisense technology to simulate transporter malfunction. Dihydrokainate (DHK) selectively blocks EAAT2 glutamate transport. l-trans-pyrrolidine-2,4 dicarboxylate (PDC) is a nonselective glutamate transport substrate inhibitor. ^{19 20 21 22 23 24 25 26} Both have well-documented neurotoxic effects. We further used antisense oligonucleotides (ODNs) against EAAT1. Antisense ODNs can selectively and specifically inhibit the expression of specific genes that regulate physiological functions or mediate various diseases. ⁵ ODNs make it possible to study the function of individual transporters. Work in other laboratories has established that animals treated with antisense to EAAT1 or EAAT2 demonstrate motor dysfunction and neuronal loss. ⁵  

Barnett and Pow have explored the effect of antisense ODNs against GLAST on the rat retina. ²⁷ They found that antisense ODNs suppressed the electroretinogram b-wave as early as 5 days after injection of antisense ODNs and were sustained for at least 20 days. They found little or no effect on ganglion cell survival. In the present study, we examined the effects of pharmacological inhibition of transporters and of antisense ODNs against EAAT1 in the rat retina, focusing our attention on ganglion cell survival. 

We previously demonstrated that the EAAT1 glutamate transporter is downregulated in human glaucoma and that this is associated with diminished levels of the NMDA receptor subunit, NMDAR1. ¹¹ We, therefore, also explored whether exogenously lowering levels of EAAT1 also led to the loss of NMDAR1. 

Methods

All procedures were in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Long–Evans rats were housed in a 12-hour light/dark cycle with water and food ad libitum. Anesthesia was attained with chloral hydrate (6 ml/kg body wt of a 7% solution), administered intraperitoneally. 

Intraocular injections were performed with a heat-pulled glass capillary connected to a microsyringe (2 μl; Drummond Microdispenser). The total volume injected at any one time was 2 μl. Injections were directed to the posterior pole of the eye to avoid any damage of the lens over a 30-second time period. Any animal with visible lens damage was euthanatized and not included further. 

DHK and PDC were injected intraocularly to a final concentration of 250 and 150 μM, respectively (six eyes each), assuming a vitreal volume of approximately 25 μl. Phosphorothioate ODNs (S-ODNs) were prepared according to the following sequences and were dissolved in sterile phosphate-buffered saline (pH 7.4) and injected at a concentration of 1 nM in 2 μl vehicle. After the initial injection, two subsequent injections of S-ODN were made 48 and 96 hours later. Intraocular injections of sense ODN and random ODN sequences were used as controls, and were injected in a similar fashion at identical concentrations. Sequences of ODNs were as follows ²⁸ : sense, 5′-GAA AGA TAA AAT ATG ACA AAA AGC AAC-3′; antisense, 5′-GTT GCT TTT TGT CAT ATT TTA TCT TTC-3′; and random, 5′-TGT CGT TTT GTT ATC TAT ATT CTT TCT-3′. 

Quantification of RGC Loss

Three days after the last intraocular injection, the fluorescent tracer hydroxystilbamidine methanesulfonate (Fluorogold, FG; Molecular Probes, Eugene, OR) was injected into the superior colliculus. Two days later (to allow the FG to be transported retrogradely to the RGCs’ somata) eyes were enucleated and the retina prepared as a flatmount. FG-stained RGCs were quantified from 12 different areas overlaying the entire retina. Three areas per retinal quadrant, aligned along the vertical and horizontal axes from the optic nerve, and 1/4, 1/2, and 3/4 the distance to the retinal edge, were counted. Labeled cells were thereby counted in 12 distinct areas of 62,500 μm each per retina. Images were obtained via a digital image system connected to a microscope equipped with appropriate epifluorescence illumination, coded, and analyzed in a masked fashion. Data are expressed as labeled RGCs/mm². 

Vitreous Collection and High-Performance Liquid Chromatographic Analysis

Animals were deeply anesthetized, and eyes were enucleated and placed in a sterile, dry petri dish. With a sterile scalpel blade, a 2-mm radial incision in the sclera was made from the optic nerve head toward the cornea. A sterile glass capillary was used to carefully remove approximately 10 μl of vitreous. The entire procedure, including enucleation, took approximately 3 minutes. Vitreous samples were immediately stored at −70°C until high-performance liquid chromatographic analyses for amino acid content was carried out. Amino acid analysis was carried out by the BioResource Center (Cornell University, Ithaca, NY). The samples were derivatized and separated on a 4.6 × 300 mm Nova Pack C18 column using a modified Pico-Tag buffer system. ²⁹ Ten eyes were included in each group. 

To evaluate retinal damage other than to ganglion cell layers, a separate cohort of eyes were enucleated and embedded in paraffin. Sections 5-μm-thick were cut from each block and stained with hematoxylin-eosin. The thickness of each retinal layer was evaluated from sections taken equidistant from the optic nerve. Retinal images were recorded with a digital camera connected to the microscope, and the thickness of the different retinal layers was measured using an image analysis system. EAAT1 had no effect on retinal layer thickness (not shown). 

Western Blot Analysis

Separate cohorts of ODN-treated eyes were used for protein analysis. Animals were euthanatized 16 hours after the last intraocular treatment and the eyes immediately enucleated. Seven retinas were used in each group. Retinal wholemounts were dissected within 2 to 3 minutes in Hanks’ balanced salt solution and shock-frozen at −80°C until analysis. Before protein extraction, wholemounts were homogenized in a solution of 20 mM Hepes, 1 mM EDTA, and 2 mM MgCl₂ at 4°C. Analyses of GLAST and actin expression were performed essentially as described in a previous study. ³⁰ Protease inhibitors were added to the homogenate, and the homogenate was diluted with 1:1 solubilizing buffer. Equal amounts of protein were loaded in each lane (200 μg), and proteins were separated using 10% sodium dodecyl sulfate–polyacrylamide, followed by transfer to Immobilon-P membranes. Anti-GLAST/EAAT1 antibodies were obtained from Jeffrey Rothstein (Johns Hopkins University) and directed against an amino-terminal peptide. ³¹ Immunoblots were probed with this anti-GLAST/EAAT1 antibody (diluted 1:5000) and an anti-actin antibody (diluted 1:5000; Sigma Chemical, St. Louis, MO). These immunoblots were visualized with enhanced chemiluminescence. In these immunoblots, there is a band that migrates to a molecular weight of approximately 42 kDa and is consistent with actin. There is also a larger diffuse molecular weight band of approximately 60 kDa, which is consistent with GLAST/EAAT1. There is a sharp band between these two. This band was not quantified because it is observed in gels in which the anti-GLAST antibody is not included (data not shown). Therefore, it is either due to the anti-actin antibody or the secondary antibody. The density of the immunoreactive bands was quantified using Image software (NIH). The data were compared by normalizing the GLAST to actin in each lane and then compared with that observed in uninjected controls. The data were also compared without normalization to actin; there was a significant effect (P < 0.05) of the antisense ODN on GLAST immunoreactivity. For immunocytochemistry, 5-μm-thick paraffin sections were deparaffinized in xylene, incubated in alcohol, and subjected to antigen unmasking as follows. Sections were rinsed in water and placed in 10 mM citrate buffer (pH 6.0) in a plastic coplin jar. The staining dishes were placed in the microwave (600 W; Sharp) with inverted lid on top and heated for 1.5 minutes at maximum power. The slides remained in the microwave for 10 minutes, with the power off, after which they were removed and left covered on the counter for 20 minutes. The slides were rinsed in 0.1 M Tris-buffered saline (TBS) buffer and processed for immunohistochemistry. The sections were then preincubated (20 minutes) with 1% bovine serum albumin diluted in TBS. Sections were incubated for 24 hours at 4°C, with affinity-purified antibodies against EAAT1 or NMDAR1 (Chemicon, Temecula, CA) at dilutions of 1:2000, 1:400, and 1:200, respectively. After primary antibody incubation, sections were rinsed (2 × 5 minutes) in TBS and incubated with biotinylated secondary antibodies at a dilution of 1:300 (1 hour room temperature). The sections were rinsed (2 × 5 minutes) in TBS with an ABC/AP (DAKO, Carpinteria, CA). After rinsing the sections (2 × 5 minutes) in TBS, they were incubated with the New Fuchsin, alkaline phosphatase substrate solution for 15 to 20 minutes The sections were then rinsed with distilled water and cover-slipped using Faramount (DAKO). Sections from control and experimental eyes were stained simultaneously, using a common preparation of reagents in masked fashion. 

Control sections were prepared omitting the primary antibody; no specific labeling was detected in any control section (data not shown). 

To quantify immunohistochemical staining, the following protocol was derived from previously published techniques. ^{32 33 34} Images were obtained via a digital image system (Image Pro Plus; Media Cybernetics, Del Mar, CA) connected to a microscope equipped with appropriate illumination, coded, and analyzed in a masked fashion. All images were recorded under identical illumination conditions. Using Adobe Photoshop (version 5.0.2; Adobe, San Jose, CA), all images were pasted into a single image. A region of unambiguous staining was identified and selected using the “magic wand” tool (tolerance set to 25). The “similar” command was used to highlight all stained regions in the composite figure simultaneously. These regions were then cut and pasted into a new image. The “invert” command was applied to the entire“ flattened” new composite (so that stained regions would be brighter than the background), and the intensity of each original retinal image was quantified with the “histogram” command. Six retinal sections, taken equidistant from the optic nerve, were used for each data point. 

One-way multivariate factorial ANOVA with Bonferroni correction was performed to compare amino acid levels between EAAT1 sense and antisense. ³⁵ Effects of DHK and PDC on RGC survival were evaluated by unpaired Student’s t-tests. EAAT1 immunoreactivity for retinas treated with sense and antisense ODNs was analyzed by ANOVA with Fisher’s least significant difference post-hoc procedure. Continuous data are expressed in terms of the mean and SD. Values for the Western blot analyses are the mean and SE. Reported probability values are two-tailed. The SAS program (version 6.12) was used for all statistical comparisons (SAS Institute, Cary, NC). 

Results

Pharmacologic Blockade of Glutamate Transporters

We first examined the effects of pharmacologic blockade of glutamate transporters in the retina with either PDC or DHK. DHK and PDC were injected daily into the vitreous of a rat eye, over a period of 4 days to induce prolonged transporter blockade. Both DHK and PDC were toxic to RGCs. Quantification of ganglion cell survival from retinal wholemounts revealed that 2367 ± 427 RGCs per mm² (mean ± SD) survived after an intraocular injection of Hanks’ solution (n = 12) compared with 1803 ± 247 RGCs per mm² after PDC treatment (P < 0.01) and 1888 ± 234 after DHK treatment (P = 0.02; n = 6 for both; Fig. 1 ). 

Intraocular ODN Administration

To investigate whether specific blockade of EAAT1 protein synthesis resulted in an increase in extracellular glutamate in the retina and excitotoxicity, antisense and sense ODNs were injected intraocularly. Antisense, sense, or random ODNs were injected intraocularly three times over a period of 7 days, at a dose of 1 nM per injection. Vitreous samples were collected to evaluate amino acid levels. Retinas were either retrogradely labeled to estimate the number of surviving RGCs or collected for protein analysis. A histologic analysis of the ODNs in the retina as well as their effect on retinal thickness was carried out. 

Twenty amino acids were evaluated in vitreal samples (Asp, Glu, Asn, Ser, Gln, Gly, His, Arg, Thr, Ala, Pro, Tyr, Val, Met, Cys, Ile, Leu, Phe, Trp, Lys). The vitreal glutamate concentration in eyes treated with sense ODNs to EAAT1 was 57.8 ± 24.8 μM (mean ± SD); antisense treated eyes had a vitreal glutamate concentration of 199.3 ± 167.4 mM (P = 0.01). Proline levels were also elevated after antisense treatment, from 27.9 ± 16.4 μM in sense-treated eyes to 74.9 ± 54.0 μM (P = 0.02). There was no statistical difference in amino acid levels between sense- and antisense-treated eyes for any other amino acids. 

Vertical sections of eyes treated with EAAT1 sense and antisense ODNs were immunostained with antisera against EAAT1. For controls the primary antibody was excluded; no specific staining was seen. 

EAAT1 immunoreactivity was present throughout the retina of sense treated animals. Müller cells were labeled in their entirety. Retinas which had been treated with antisense ODNs showed a reduction of EAAT1 immunoreactivity (Figs. 2A , 2B ). 

Western blot analysis was used to examine the effectiveness of antisense ODNs in reducing transporter expression, comparing sense- and antisense-treated retinas. The anti-EAAT1 antibody recognized a band at approximately 65 kDa. Figure 2C shows a representative Western blot indicating a clear downregulation of EAAT1 protein levels after antisense treatment. Sense ODNs had no effect on EAAT1 expression. Figure 2D summarizes the results of several independent experiments and demonstrates that the antisense ODNs caused the levels of GLAST immunoreactivity to decrease to slightly less than 60% of control (P < 0.01). 

In eyes treated with EAAT1-sense ODNs (n = 10) 2459 ± 499 RGCs survived, compared with 1871 ± 237 RGCs, which remained after EAAT1-antisense (n = 10) treatment (P < 0.01, Fig. 1 ). Sense treatment had no effect on ganglion cell survival; antisense to EAAT1 led to loss of ganglion cells. 

Histologic Evaluation of Retinas Treated with ODNs

To address the question of whether inhibition of EAAT1 can cause retinal damage other than to RGCs, we examined cross sections of control and sense- and antisense-treated eyes, stained with hematoxylin–eosin. EAAT1 antisense treatment had no significant effect. Although in many cases ODNs can be toxic to neurons on their own, we found no toxicity of sense or random ODNs in these experiments. 

Furthermore, we quantified the effect of antisense against EAAT1 on NMDAR1 levels. Downregulation of the EAAT1 transporter led to diminished levels of this subunit of the NMDA receptor, as shown both in immunohistochemical staining (Figs. 3A , 3B ) and quantified in Figure 3C . 

Discussion

The results presented here indicate a significant role for glutamate transporters in maintaining normal glutamate levels and in ensuring RGC viability. Interference with glutamate transport, either pharmacologically or with antisense ODNs directed against EAAT1, resulted in elevated vitreal glutamate levels, diminished ganglion cell viability, and, most intriguingly, a loss of the glutamate receptor subunit, NMDAR1, after EAAT1 disruption. 

Glutamate toxicity to RGCs has been well documented; it is, therefore, not surprising that an elevated level of vitreal glutamate, irrespective of the mechanism by which it is produced, leads to ganglion cell loss. We did not see the magnitude of ganglion cell loss that is attainable with intravitreal administration of exogenous glutamate or NMDA. ^{36 37} However, this may be either because of our inability to completely block the transporter under investigation or the ability of the remaining glutamate transporters to regulate extracellular glutamate levels. 

Previous work by Harada and coworkers has shown that EAAT1-deficient mice have loss of the b-wave of the electroretinogram and a reduction of oscillatory potentials. ³⁸ In addition, the retinas in these animals are more sensitive to ischemic damage. In contrast, EAAT2-deficient mice have normal electroretinograms; their response to retinal ischemia is closer to normal. 

Interference with separate glutamate transporters, because of their heterogeneous distribution, can selectively elevate glutamate in different parts of the retina. How this heterogeneous distribution regulates local glutamate concentrations is clearly significant. l-trans-PDC, an inhibitor with similar affinities for each of the subtypes of transporter, causes a loss of ganglion cells. It is possible that the effect of L-trans-PDC was due to reduced glutamate uptake, but it is also possible that L-trans-PDC was stimulating heteroexchange through the transporter, because this compound is also substrate for the transporters. The nonsubstrate inhibitor DHK also caused a loss of ganglion cells. DHK is selective for EAAT2 at the concentrations used and does not inhibit EAAT1 activity at concentrations over 3 mM, suggesting that the ganglion cell loss observed after administration of this compound is related to inhibition of EAAT2. Finally, we demonstrated that antisense ODNs directed against EAAT1 cause ganglion cell loss. Together, these data suggest that both EAAT1 and EAAT2 are important for limiting excitotoxicity in the ganglion cell layer. This is in accordance with prior work demonstrating the sensitivity of the inner retina to glutamate-mediated toxicity, which is not surprising because ganglion cells are the most sensitive to glutamate-mediated damage. ³⁹ Pharmacological intervention to regulate glutamate transporters becomes a viable approach for states in which glutamate transporters are either diminished or not functioning effectively. We have previously shown, for example, that glial-derived neurotrophic factor can lead to upregulation of EAAT1 ¹¹ and that epidermal growth factor can lead to increased expression of both EAAT1 and EAAT2 in astrocyte cultures. ³⁰ Therefore, growth factors may potentially correct states in which it is lacking or diminished. ¹¹ Viral vectors containing glutamate transporters could be used to augment levels as needed. In parallel to work done in other parts of the central nervous system, these results suggest that glutamate transporters have a critical role in RGC survival and retinal function. 

The reduction in NMDAR1 levels suggests that glutamate alterations can be associated with multiple consequences. First, and foremost, the loss of NMDAR1 receptors may reflect a loss of ganglion cells. Cebers and coworkers, in studies on cultured cerebellar granule cells, have demonstrated that pharmacological blockade of glutamate transporters leads to decreased levels of NMDAR1. ²⁰ When we administered antisense ODNs against EAAT1, we found a similar loss of NMDAR1 in vivo (Fig. 3) . This experiment indicates that the loss of NMDAR1 can be a direct consequence of the loss of EAAT1. It is possible that a neuron, when faced with elevated levels of extracellular glutamate, may attempt to compensate by lowering levels of glutamate receptors and, consequently, its sensitivity to excitotoxic damage. 

In summary, it is likely that EAAT1 and EAAT2 both are critical for the regulation of extracellular glutamate levels; interference with their function increases extracellular glutamate concentrations and adversely impacts ganglion cell survival. 

⁶  Present address: Department of Ophthalmology, Otto–von–Guericke–University, D–39120 Magdeburg, Germany.

 Supported by National Institutes of Health Grant R01 NS36465 (MBR); a Merit Grant from the Veteran’s Administration (EBD), and grants from Research to Prevent Blindness, Potts Foundations, Allergan Corporation, and the Jody Lynn Sack Memorial Fund (EBD). EBD is the recipient of a Research to Prevent Blindness Lew R. Wasserman Merit Award. CKV was supported by a Theodor–Leber–Stipend from the Basotherm Förderkreis, Germany, and from the Ernst and Berta Grimmke Stiftung, Germany.

 Submitted for publication February 14, 2000; revised May 24, 2000; accepted June 20, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

RGC survival after pharmacological or antisense blockade of glutamate transporters. DHK (250 μM) and PDC (150 μM) are toxic to RGCs in vivo. Antisense or sense treatment was as described in the text. RGCs were retrogradely labeled with Fluorogold from the superior colliculus. PDC (P < 0.01) and DHK (P = 0.02) significantly reduced ganglion cell survival. Sense treatment had no effect compared with control eyes injected with buffer. EAAT1 antisense treatment diminished ganglion cell survival when compared with sense treatment in a significant fashion (P < 0.001 and P < 0.01, respectively). Asterisks indicate significance.

Figure 2.

 

Effects of intraocular injection of antisense ODNs on the expression of the glutamate transporter EAAT1 in the retina. Retinal sections stained with antibody against EAAT1 after treatment with sense (A) and antisense (B) ODNs against EAAT1. The ganglion cell layer is uppermost. There is a clear loss of EAAT1 staining in antisense-treated retinas. (C) Representative Western blot showing EAAT1 and actin immunoreactivity. Retinas treated with EAAT1 sense or antisense ODNs. A 200-μg aliquot of total protein was loaded in each lane. (D) Summary of EAAT1 expression levels from several independent experiments. Data are expressed relative to the levels observed in control (untreated) retinas and are represented as mean ± SD of 7 independent experiments. Data were compared by ANOVA with Fisher’s least significant difference post-hoc analysis.* P < 0.01 compared with treatment with the sense oligonucleotide; **P < 0.001.

Figure 3.

 

Effects of intraocular injection of antisense ODNs on the expression of the glutamate receptor subunit NMDAR1 in the retina. Retinal sections stained with antibody against NMDAR1 after treatment with sense (A) and antisense (B) ODNs against EAAT1. The ganglion cell layer is uppermost. There is a clear loss of NMDAR1 staining in antisense-treated retinas. (C) Summary of NMDAR1 expression levels from several independent experiments. * Indicates P < 0.001 by Student’s t-test.