Effect of Intravitreal Administration of Somatostatin and sst₂ Analogs on AMPA-Induced Neurotoxicity in Rat Retina

Abstract

purpose. The aim of the present study was to use an in vivo model of retinal excitotoxicity to investigate the neuroprotective effect of somatostatin (SRIF)-ergic agents.

methods. Adult Sprague–Dawley rats (weight range, 250–300 g) intravitreally received (RS)-α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid hydrobromide (AMPA; 21, 42, 84 nmol/eye) or PBS (50 mM). Time-dependent responses were examined in animals that received AMPA (42 nmol/eye). Animals received AMPA (42 nmol) alone or in combination with SRIF (10⁻⁵, 10⁻⁴ M) or the sst-selective ligands lanreotide (sst₂,10⁻⁵,10⁻⁴ M), L-779976 (sst_2,10⁻⁶,10⁻⁵, 10⁻⁴ M), L-797591 (sst₁,10⁻⁴ M), and L-803087 (sst₄,10⁻⁴ M). Immunohistochemistry and TUNEL studies were used to examine retinal cell loss and protection. Immunochemistry, Western blot analysis, and radioimmunoassay assessed the viability of sst_2A receptors and SRIF levels, respectively, in control and AMPA-treated tissue.

results. AMPA (42 nmol) treatment resulted in total and major loss of ChAT and bNOS immunoreactivity, respectively, 24 hours after its administration. This loss was sustained up to 30 days for ChAT- and 8 days for bNOS-expressing amacrine cells. SRIF and the sst₂ receptors were not affected by AMPA. SRIF and the sst₂ analogs protected the retina from the AMPA insult in a dose-dependent manner, whereas activation of the sst₁ and sst₄ subtypes had no effect. TUNEL staining confirmed AMPA-induced retinal ischemia and L-779976 neuroprotection.

conclusions. These results demonstrate for the first time that SRIF and the sst₂ analogs, administered intravitreally, protect the retina from excitotoxicity. Further studies are essential to ascertain the therapeutic relevance of these results.

Ischemia is the underlying cause of many ocular diseases that lead to blindness. A large volume of evidence has implicated glutamate excitotoxicity as a leading player in retinal ischemia. ¹ Glutamate is the leading neurotransmitter in the retina. It is released from photoreceptors, bipolar cells, and ganglion cells in the vertical signal pathway and plays a major role in vision transduction by activating different receptor subtypes belonging to the ionotropic (N-methyl-d-aspartate [NMDA], kainite, and (RS)-α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid hydrobromide [AMPA] receptors) and metabotropic receptor families. ^{2 3}  

Overstimulation of glutamate receptors leads to retinal toxicity. The NMDA receptor family has been widely studied, and evidence from in vitro and in vivo studies has suggested a leading role for NMDA receptors in retinal excitotoxicity. ^{1 4 5} Intravitreal injection of NMDA into the adult rat eye caused primarily ganglion but also amacrine cell loss, whereas agents that act as antagonists have protected these retinal neurons. ^{6 7} It is accepted that the activation of NMDA receptors, the increase in calcium ions, and the subsequent activation of nitric oxide radical formation is the underlying mechanism involved in excitotoxicity. ¹ The non-NMDA receptors kainate and AMPA have also been implicated in retinal ischemic insult. ^{8 9 10} Kainate induced cultured chick amacrine-like cell damage, and this was shown to be mediated by AMPA. ¹¹ Intraocular injections of kainate caused changes in GABA and choline acetyltransferase (ChAT)-containing cells, ^{12 13} whereas kainate/AMPA receptor blockade was effective in retinal neuroprotection. ^{1 14}  

Many strategies have been used to develop therapeutic agents for the successful treatment of ischemia-induced retinopathies and the prevention of blindness. However, the lack of success so far in attaining this goal makes more evident the complexity of the systems involved in the pathophysiology of these diseases and, therefore, their therapeutics. Recently, we focused on the use of the neuropeptide SRIF as a putative neuroprotective agent in retinal ischemia. 

SRIF ¹⁵ is found in the retina, ¹⁶ where it activates SRIF receptors, ¹⁷ ssts, found in retinal neurons and the retinal pigment epithelium (RPE; for a review, see Thermos ¹⁸ ). Specifically, sst_2A receptors are localized in rod bipolar, amacrine, and photoreceptor neurons in the retina, ^{19 20 21 22 23 24} and sst_2B receptors are localized in photoreceptors and the RPE. ^{22 24} In addition, sst₁ and sst₄ receptors are present in amacrine (SRIF and TH-containing) ²⁵ and ganglion cells, ^{26 27} respectively. 

SRIF and its analog, sst₂, inhibit ischemia-induced neovascularization in a mouse model of oxygen-induced retinopathy ^{28 29} and depict neuroprotective effects against different paradigms of neurotoxicity in the central nervous system, such as NMDA, kainate-induced neurotoxicity, ^{30 31} and middle cerebral artery. ³² We used an in vitro model ³³ of chemical ischemia in the retina to assess the neuroprotective actions of SRIF. ³⁴ Chemical ischemia involves the blockade of oxidative phosphorylation and glycolysis and is believed to be useful in the understanding of the early events underlying the pathophysiology of ischemia. In this model, SRIF analogs selective for the sst₂ subtype protected the retina from the ischemic insult. ³⁴ In a subsequent study using transgenic mice, Catalani et al. ³⁵ showed that increased levels of the sst₂ subtype protected the retina from ischemia (hypoxia and iodoacetic acid). This study is in agreement with our previous data. ³⁴ It indirectly showed the involvement of the sst₂ subtype and recommended that sst₂ receptor agonists may be useful in retinal diseases. 

The aim of the present investigation was to examine SRIF actions in vivo. To this end, an in vivo model of excitotoxicity was used, and the assessment of the ability of SRIF and sst₂ analogs to protect the retina, when administered intravitreally, was attempted. AMPA was chosen as the excitotoxic agent, and initial studies focused on the experimental conditions necessary for AMPA-induced retinal cell loss. 

Methods

Animals

Adult male and female Sprague–Dawley rats (weight range, 250–350 g) were housed at room temperature two to three animals per cage and had free access to food and water. A 12-hour light/12-hour dark cycle was maintained. Euthanatization was performed with ether inhalation. All procedures were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and in compliance with Greek national laws (Animal Act, P.D. 160/91). 

AMPA Excitotoxicity and Neuroprotection

The animals were anesthetized with intramuscular injections of xylazine (14.5 mg/kg) and ketamine (200 mg/kg) and were placed in a stereotaxic instrument to stabilize the head and facilitate intravitreous administration. Injections were made with 27- or 30-gauge needles connected to a Hamilton syringe adapted to a minipump. The tip of the needle was inserted behind the sclera–cornea border into the vitreous humor. Each intravitreal injection was performed with a flow rate of 1 μL/min for 5 minutes. 

One eye received phosphate-buffered saline (50 mM K₂HPO₄/NaH₂PO₄, 0.9% NaCl, pH 7.4) as control, and the other eye received AMPA (Tocris) 21, 42, or 84 nmol/eye, respectively, diluted in PBS 50 mM. Rats administered AMPA (42 nmol/eye) were humanely killed at different times (1, 2, 4, 8, 16, and 30 days) after treatment. Neuroprotection experiments involved the injection of AMPA (42 nmol/eye) alone and in combination with SRIF (10⁻⁵, 10⁻⁴M) or the sst-selective ligand lanreotide (sst₂ 10⁻⁵, 10⁻⁴ M), L-779976 (sst₂ 10⁻⁶, 10⁻⁵, 10⁻⁴ M), L-797591 (sst₁ 10⁻⁴ M), or L-803087 (sst₄ 10⁻⁴ M). Control retinas received PBS (50 mM). 

Immunohistochemical Studies

Tissue Preparation.

After each treatment, the eyes were removed and the eyecups were fixed by immersion in 4% paraformaldehyde in 0.1 M phosphate buffer for 1 hour at 4°C. After fixation, eyecups were rinsed in phosphate buffer and were incubated in 30% sucrose overnight at 4°C for cryoprotection. Tissues were frozen in isopentane at −45°C for 1 minute and kept at −80°C until further use. Eyecups were sectioned vertically at 10-μm thickness using a cryostat, thaw mounted on slides (Superfrost; Erie Scientific, Portsmouth, NH), and stored at −20°C. Slices were cut near the optic nerve every 100 μm. Nine slices were put on every slide. 

Immunohistochemistry.

A mouse monoclonal antibody raised against ChAT (1:100; Biotrend, Cologne, Germany) and a rabbit polyclonal antibody raised against bNOS (1:1000; Sigma, St. Louis, MO) were used as markers for, respectively, acetylcholine and neuronal nitric oxide synthase containing retinal cells. Mouse monoclonal antibodies for protein kinase C (PKC; 1:50; Leinco Technologies, St. Louis, MO) and for microtubule associating protein (MAP)-1 (1:100; Sigma) were used as markers for rod bipolar cells and ganglion cells, respectively. Rabbit polyclonal antibodies for tyrosine hydroxylase (TH) (1:1000; Chemicon, Temecula, CA) and for recoverin (0.044 mg/mL; gift from Eleonora N. Grigoryan) were used as markers for dopamine-containing amacrine cells and photoreceptors/cone bipolar cells, respectively. A mouse monoclonal antibody raised against SRIF (2 μg/mL; Abcam, Cambridge, UK) and a rabbit polyclonal antibody for the SRIF receptor subtype sst_2A (1:1000; Gramsch Laboratories, Schwabhausen, Germany) were also used for the detection of SRIFergic- and sst2A-expressing retinal cells, respectively. 

Cryostat sections were incubated in 0.1 M Tris-HCl buffer, pH 7.4, containing 3.3% normal goat serum for 30 minutes, washed in 0.1 M Tris-buffered saline (TBS), and incubated with the primary antibodies in 0.1 M TBS containing 0.3% Triton X-100 and 0.5% normal goat serum overnight at room temperature. The sections were washed again and incubated for 1 hour with the appropriate secondary antibody, Alexa-Fluor 488 goat anti-mouse IgG (H+L; 1:400; Molecular Probes, Eugene, OR) for the monoclonal antibodies or Alexa-Fluor 546 goat anti-rabbit IgG (H+L; 1:400; Molecular Probes) for the polyclonal antibodies. Finally, the sections were washed and coverslipped with mounting medium with DAPI (Vector Laboratories, Burlingame, CA). Negative controls were included by omitting the primary antibody. 

TUNEL Staining.

To determine cell loss, enzymatic in situ labeling of apoptosis-induced DNA strand breaks was performed on tissue sections obtained from retinas 1 day after AMPA treatment. The terminal deoxynucleotidyl transferase (TDT)-mediated TMR-dUTP nick-end labeling (TUNEL) assay (Roche, Germany) was used. 

Microscopy.

Light microscopy images were taken with a camera (Axioskop with Plan-Neofluar ×40/0.75; Carl Zeiss, Oberkochen, Germany). Optic sections were taken with a z-axis resolution of 1.1 μm through the immunolabeled cells. Light and contrast adjustments of images were processed with the use of commercial software (Photoshop ver. 7.0; Adobe Systems, San Jose, CA). 

Quantification

For the quantification of bNOS-expressing retinal cells, each section of control or treated retina was studied by means of a ×40/0 lens (Planapo; Hewlett-Packard, Palo Alto, CA). The total number of bNOS-immunoreactive neurons in each retinal section was counted. As mentioned in Tissue Preparation, for the immunohistochemistry studies, all slices used were cut near the optic nerve. Nine slices were put on every slide. bNOS-expressing cells were counted on all nine sections per slide. Therefore, the maximum number of sections that could be examined was n (number of retinas) × 9. Given that some slices were not appropriate for cell counting (e.g., because of tissue folding or damage during cutting), the actual number of sections used for the quantification was equal to or less than n × 9. 

The numbers of sections used were as follows: control groups: 1-day group, 41 sections were used from 5 different retinas (n = 5); 8-day group, 20 sections were used from 3 different retinas (n = 3), 16-day group, 27 sections were used from 3 different retinas (n = 3), 30-day group, 27 sections were used from 3 different retinas (n = 3); AMPA-treated groups: 1-day group, 26 sections were used from 4 different retinas (n = 4); 8-day group, 23 sections were used from 3 different retinas (n = 3); 16-day group, 20 sections were used from 3 different retinas (n = 3); 30-day group, 25 sections were used from 3 different retinas (n = 3). Twenty-seven sections from 3 different retinas (n = 3) were used for the AMPA+L-779976–treated group (neuroprotection studies). 

Radioimmunoassay

SRIF levels were examined in retinas that were removed from eyes treated with PBS (50 mM; n = 6) or AMPA (42 nmol/eye; n = 7) 1 day after treatment. Samples were prepared according to the method described by Mastrodimou and Thermos. ³⁶ In short, retinas were centrifuged for 15 minutes at 12,000 rpm and 4°C. Acetic acid (2 N) was added to the precipitant, and the mixture was boiled for 10 minutes, homogenized with a sonicator, and stored at −80°C for 24 to 48 hours. Subsequently, the mixture was centrifuged for 20 minutes at 13,000 rpm, and the supernatant was lyophilized and kept at −80°C. The assay mixture contained the SRIF antiserum (kindly provided by Günther Sperk) at a final dilution of 1:3000, [¹²⁵I]-Tyr ¹¹ -SRIF (20,000 cpm), and SRIF standards (0–400 pg/100 μL) or samples in a final assay volume of 0.5 mL, according to Sperk and Windmann. ³⁷ Incubation was carried out at 4°C for 24 hours. Separation of bound and free peptide was achieved by the addition of 900-μL aliquots of a mixture containing 2.5% charcoal and 0.25% dextran T70 and subsequent centrifugation. 

Western Blotting

Retinal Membrane Preparation.

Sprague–Dawley rats were anesthetized, and retinas were removed and homogenized (Ultra-Turrax homogenizer; IKA Works, Staufen, Germany) in Tris-HCl, pH 7.4. The homogenate was centrifuged at 1000g for 10 minutes, and the supernatant was aspirated and stored. The pellet was resuspended, rehomogenized, and centrifuged as described. The two supernatants were combined and centrifuged at 11,000g for 20 minutes. The pellet was resuspended in Tris buffer and centrifuged at 27,000g for 10 minutes. Finally, the pellet was resuspended in lysis buffer (10 mM Tris-HCl, pH 7.4, 250 mM sucrose, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1% Triton, 2 μg/mL leupeptin, 2 μg/mL aprotinin). All steps of this procedure were performed at 4°C. The protein content was determined according to Bradford, ³⁸ and aliquots of each sample containing equal amounts of protein (32 μg) were subjected to 10% SDS-PAGE and immunoblotted onto nitrocellulose membranes. Blots were incubated with anti-sst_2A (1:2500) overnight at 4°C. For adsorption controls, sst_2A antigen at concentrations of 10⁻⁴ M were incubated with the antisera for 2 hours at room temperature. Blots were developed using peroxidase-conjugated secondary antibodies and enhanced chemiluminescence. These experiments were performed two to four times from the same or different preparations of retinas. 

Retinal Protein Samples from Treated Animals.

Animals were humanely killed, and retinas were removed and homogenized (Ultra-Turrax homogenizer; IKA Works) in lysis buffer. Because of the scarcity of the tissues, the whole membrane preparation was used. Retinal samples (60 μg) were subjected to 10% SDS-PAGE and immunoblotted onto nitrocellulose membranes. Blots were incubated with a rabbit polyclonal antibody against SRIF receptor subtype sst_2A (1:2500; Gramsch Laboratories) overnight at 4°C. Blots were developed using peroxidase-conjugated secondary antibodies and enhanced chemiluminescence. These experiments were performed two to four times using protein samples from the same or different preparations of retinas. 

Statistical Analysis

Analysis of data was performed using a one-way ANOVA with post hoc analysis (Dunnett; GraphPad Prism 2.01; GraphPad Software, San Diego, CA). 

Results

Effect of AMPA on ChAT Immunoreactivity: Dose- and Time-Response Experiments

ChAT immunoreactivity was observed in amacrine cells located in the inner nuclear layer (INL) and the ganglion cell layer (GCL) and in processes in the inner plexiform layer of PBS-treated retinas (Fig. 1A) . Intravitreal administration of the excitatory amino acid AMPA (21, 42, 84 nmol/eye) caused a dose-dependent loss of ChAT expression. The intravitreal injection of AMPA (21 nmol/eye) caused a partial loss of ChAT immunoreactivity because processes were still visible 1 day after injection. AMPA (42 nmol/eye) caused a complete loss of ChAT expression, whereas a higher concentration of 84 nmol/eye gave similar results (Fig. 1) . ChAT immunoreactivity loss was sustained for 8, 16, or 30 days after AMPA administration (Fig. 2) . The effect of AMPA on ChAT immunoreactivity was also examined 2 and 4 days after injection, and similar results were obtained (data not shown). 

Effect of AMPA on PKC, TH, bNOS, Recoverin, and MAP^1A Immunoreactivity

To examine the effect of the AMPA treatment (42 nmol/eye) on other retinal cell types, immunoreactivity studies were performed using antibodies against retinal markers that recognize photoreceptors and rod and cone bipolar, amacrine, and ganglion cells. Figure 3depicts the PKC and MAP^1A immunoreactivities in retinal tissues isolated 1 day after AMPA treatment. PKC immunoreactivity was not different in the two groups of retinas (n = 2), even though the staining pattern in Figure 3Bmight have suggested more dense staining in rod bipolar cells of the AMPA-treated retinas. Similarly, MAP^1A immunoreactivity was rich in the processes and some cell bodies of ganglion cells (Figs. 3C 3D) , but no differences were observed between control and AMPA-treated retinas (n = 3). TH immunoreactivity in amacrine cells was not affected by the AMPA treatment (n = 3; Figs. 4A 4B ). Recoverin immunoreactivity was evident in the outer nuclear layer (ONL) and the INL in control and AMPA-treated retinas (n = 2; Figs. 4C 4D ). Immunohistochemical studies using the same antibodies were performed on tissues obtained 8, 16, and 30 days after AMPA treatment. Results were similar to those described for the 1-day groups (data not shown). 

bNOS immunoreactivity studies showed that NOS-containing neurons were decreased 1 day after AMPA administration. However, the signal returned to control levels 8 days after AMPA injection. To avoid the repetition of immunoreactivity staining patterns, bNOS-containing amacrine cells were counted in PBS- and AMPA-treated retinal sections, and the results are shown in Figure 5 . 

Effect of AMPA on SRIF and sst_2A Levels

The most significant changes in ChAT and bNOS immunoreactivities were observed as early as 1 day after AMPA (42 nmol/eye) treatment. Therefore, we examined whether SRIF levels and the expression of the sst₂ receptor, the most predominant receptor subtype in the retina, were affected in the same paradigm. SRIF levels were measured by radioimmunoassay. No statistically significant changes were observed between the control and the AMPA-treated retinas (Fig. 6) . The presence of the sst₂ receptor in the retina was examined by immunoreactivity and Western blot methodologies. Immunoreactivity differences with sst_2A were not apparent between the PBS- and AMPA-treated retinas (Figs. 7A 7B) . To analyze further the retinal samples under study, Western blot analysis was performed. Initially, we performed control studies to examine the presence of the receptor in retinal membranes. A protein (molecular mass, approximately 77 kDa) was detected in the retina that was not visible in the presence of the sst₂ antigen (Fig. 7C) . Therefore, we concluded that the 77-kDa protein represented the sst₂ receptor subtype. 

Western blot analysis using crude protein preparations of PBS- and AMPA-treated retinas depicted a protein of apparent molecular weight of approximately 72 kDa (Fig. 7D) . We presumed this protein to be the equivalent of the 77-kDa protein in the control blots. Actin immunoblots were also examined in the two samples to substantiate that equal protein concentrations were applied, and no differences were evident between the PBS- and AMPA-treated retinas. 

Effect of Somatostatin on AMPA-Induced Loss of ChAT Immunoreactivity

To examine whether SRIF could protect the retina from the AMPA-induced loss of ChAT immunoreactivity, SRIF was coinfused with AMPA (42 nmol/eye). ChAT immunoreactivity was assayed 1 day after the treatment. SRIF protected the cholinergic neurons in a dose-dependent manner (10⁻⁵, 10⁻⁴ M; Fig. 8 ), but it did not afford full protection. 

Effect of the Selective Agonists Lanreotide (sst_2/5) and L-779976 (sst₂) on AMPA-Induced Loss of ChAT and bNOS Immunoreactivity

The putative neuroprotective action of selective analogs of SRIF receptors was also studied (Fig. 9) . Lanreotide (10⁻⁵ M) coinfused with AMPA (42 nmol/eye) mimicked the effects of SRIF at 10⁻⁴ M (Fig. 9C) . ChAT immunoreactivity completely recovered at the higher lanreotide concentration of 10⁻⁴ M (Fig. 9D) . The sst₂-selective ligand L-779976 was also examined. As shown in Figure 10 , L-779976 protected the retina in a dose-dependent manner, with full recovery at the highest dose used (10⁻⁴M). bNOS immunoreactivity was also examined in retinas treated with L-779976 (10⁻⁴ M). The viable cells observed in the presence of L-779976 were comparable to the control samples, suggesting full recovery (Fig. 10F) . 

TUNEL staining was used to assess retinal cell death by AMPA (42 nmol/eye) and protection by the L-779976 analog 1 day after treatment. The results, shown in Figure 11 , confirm the excitotoxic effect of AMPA and the protection afforded by L-779976 (10⁻⁴ M). 

Effect of the sst₁- and sst₄-Selective Analogs on AMPA-Induced Loss of ChAT Immunoreactivity

Receptor subtypes sst₁ and sst₄ have also been detected in amacrine and ganglion cells, respectively, of rat retina. ²⁶ Therefore, the putative neuroprotective effect of selective sst₁ and sst₄ analogs was examined. The AMPA-induced loss of ChAT immunoreactivity was not reversed by the presence of L-797591 (sst₁ selective) or L-803087 (sst₄ selective) at the concentration of 10⁻⁴ M, a concentration that was effective for the sst₂ ligands (Fig. 12) . 

Discussion

The main finding of this study is that SRIF and its analogs, lanreotide and L-779976, provide neuroprotection against excitatory insults in vivo. Injected intravitreally, these agents reduce the retinal damage induced by the excitatory amino acid AMPA. 

Glutamate, the major neurotransmitter in the retina, is responsible for the transfer of visual information from the retina to the brain. It is released by photoreceptors, bipolar cells, and ganglion cells and acts on ionotropic and metabotropic glutamate receptors found in retinal neurons. ^{2 39 40} Excess glutamate release and overactivation of its ionotropic receptors is believed to be associated with toxicity and cell death. Therefore, the use of glutamate and other excitatory amino acids with high affinity for the ionotropic receptor family, such as NMDA, kainate, and AMPA, have been used as models of excitotoxicity. These studies have been instrumental in the evaluation of the events leading to toxicity caused by ischemic insults because excitotoxicity is believed to be responsible for the ischemia-related cell death. ¹  

In the present study, we used AMPA and examined its effects when given intravitreally in the retina. Subsequently, we studied whether the neuropeptide SRIF could reverse the toxic insult. AMPA receptors belong to the non-NMDA family of ionotropic glutamate receptors. They are composed of four subunits, GluR1 to GluR4, that are differentially expressed in retina. ^{41 42} AMPA receptors lacking GluR2 have an increased permeability to intracellular calcium ions, making the neurons that express these receptors more vulnerable. ^{3 43 44}  

The present data showed that AMPA differentially influenced retinal cell viability. It eliminated, in a dose-dependent manner, ChAT immunoreactivity 24 hours after its injection into the rat eye. This loss persisted 30 days after the AMPA intravitreal injection. Similarly, it reduced the number of bNOS-containing amacrine cells. However, the signal recovered 8 days after injection, in contrast to what was observed for the ChAT immunoreactivity signal. No changes were observed in the retinal markers used to label photoreceptors or cone and rod bipolar, TH-containing amacrine and ganglion cells. 

The degree of loss observed in the different cell types is directly coupled to the presence of the AMPA receptor subtype. mRNA and protein studies support the localization of AMPA receptors in all retinal cell types of different animal species, ^{2 40 45} but a differential localization of the AMPA receptor subunits has been reported. ⁴⁵ In agreement with the present findings, ChAT immunoreactivity was shown to be obliterated after intravitreal kainic ¹³ and quisqualic acid injection. ⁴⁶ The presence and function of AMPA receptors on cholinergic neurons was also shown by release studies. Selective blockade of AMPA receptors in rabbit retina inhibited the light-evoked increase in acetylcholine release, whereas the subunits GluR2 and GluR3 were found to be localized on ChAT immunoreactive processes. ⁴⁷ A drastic reduction in ChAT immunoreactivity was also observed after exposure of the retinas to kainate and in a model of high intraocular pressure. The authors suggested that in paradigms in which the ischemia is of sufficient severity, kainate receptors associated with cholinergic neurons are overstimulated, leading to membrane disruption and enzyme (ChAT) release. ¹³ This may reflect the changes observed in the present paradigm in which AMPA receptors are activated, leading to ChAT immunoreactivity loss. 

An earlier study, examining the effect of excitotoxicity in the rabbit retina, indicated that NADPH-diaphorase (marker for NOS) amacrine cells contained NMDA and kainic acid receptors. ⁴⁸ In addition, in turtle retina, NOS inhibitors antagonized the kainic acid increase in cGMP-like immunoreactivity, suggesting that the activation of kainic acid receptors in NOS-containing neurons stimulate NO release. ⁴⁹ To our knowledge, there are no immunohistochemical data to support the localization of AMPA receptors on NOS-containing amacrine cells in the retina. However, these data pertaining to the use of kainic acid (also an agonist of AMPA receptors) indirectly support the presence of AMPA receptors and their implication in the loss of bNOS immunoreactivity in the present study. In contrast to ChAT immunoreactivity, bNOS immunoreactivity reappeared 8 days after AMPA treatment. The morphology of the bNOS-expressing cells was the same as in the control samples (data not shown). 

In an extensive immunocytochemical study, the differential effects of ischemia/reperfusion on different amacrine cell subtypes was examined. Loss of immunoreactivity was observed in amacrine cells expressing Glyt1 (glycinergic), substance P, and CR (subpopulation of cholinergic cells) at 2 hours after ischemia, but the immunoreactivity recovered to a large extent between 4 and 12 hours after ischemia. ChAT immunoreactivity in this paradigm was reduced with no subsequent recovery. ⁵⁰ This latter study and the present results support differential changes in amacrine cells as a result of ischemia/reperfusion and excitatory amino acid receptor activation, respectively. The transient loss of immunoreactivity may be the result of the inhibition of protein synthesis and its subsequent restoration, whereas the ChAT immunoreactivity loss may be a result of the stimulation of protease activity (caspases) and long-term effects. ⁵⁰ Ischemia-reperfusion injury was shown to activate different caspases depending on the neuronal phenotype in the retina, with caspases 2 and 3 acting in parallel in amacrine neurons. ⁵¹  

In agreement with the present data, intravitreal infusion of quisqualic acid did not affect photoreceptors, bipolar cells, or ganglion cells, and only a small loss of TH-containing amacrine cells from the central retina was reported. ⁴⁶ Furthermore, GluR1–4 expression was not found in photoreceptors, ⁴⁵ suggesting that AMPA administration should have no effect on these retinal neurons. Further evidence in the literature supports the lack of effect of AMPA on rod and cone bipolar and ganglion cells. Electrophysiological studies have shown 2-amino-4-phosphonobutyric acid (APB)-sensitive functional responses on the activation of rod and cone ON-bipolar cells. ⁵² These bipolar cells are activated by glutamate primarily through the APB-type metabotropic receptor and involve cGMP signaling. ⁵³ However, recent colocalization immunohistochemical studies supported the presence of AMPA-type glutamate receptors (GluR2) in PKC-containing cells (rod and cone ON-bipolar) in the rat retina, ⁵⁴ whereas single-cell RT-PCR showed the presence of GluR1/GluR2 in rod and cone bipolar cells. ⁴⁵ As stated, the presence of GluR2 renders the receptor impermeable to calcium ions; thus, its presence in neurons should spare them from toxicity. Kainate and NMDA receptors are expressed in ganglion cells, ^{55 56} and GluR4 mRNA has also been detected. ⁴⁵ The relative survival of photoreceptors, bipolar cells, and ganglion cells after AMPA exposure may be attributed to the absence or low levels of the receptor, as mentioned, or to their ability to deal with sustained depolarization and ionic fluctuation. It was suggested that the large cytosolic volume of ganglion cells might buffer the accumulation of intracellular calcium and thus make these cell less vulnerable to excitotoxicity. ⁴⁶  

The findings thus far suggested that AMPA is a good model of retinal cell loss (subtypes of amacrine cells) and support its use for the testing of agents as neuroprotectants of retinal neurons against toxicity. To this end, the subsequent studies investigated whether SRIF and specific sst analogs could be useful in the reversal of the actions of AMPA. SRIF is released from a subclass of amacrine cells in the retina ^{16 36} and activates its receptors, sst₁ to sst₅, found in different retinal neurons. ²⁰ The sst₂ subtype appears to be the predominant one in vertebrate retina. 

AMPA did not have any effect on SRIF levels, as measured by radioimmunoassay (Fig. 6) . This suggests that AMPA receptors are not present in SRIF-containing amacrine neurons. In addition, the intravitreal injection of AMPA did not influence the sst₂ receptors, as shown by immunohistochemical and Western blot analysis studies (Fig. 7) . The presence of viable sst₂ receptors enabled us to proceed with the investigation of whether SRIF itself and the specific sst₂ analogs could protect the retina from AMPA excitotoxicity. As shown by ChAT immunoreactivity, intravitreal administration of SRIF was able to partially reverse the damage imposed by AMPA on the retina (Fig. 8) . The lack of full recovery of the signal may be attributed to peptide degradation. However, similar results were obtained in the presence or absence of protease inhibitors (data not shown). Thus, other factors (e.g., absorption) may influence SRIF action. The sst₂-selective ligands lanreotide and L-779976 were able to reverse the AMPA damage and bring ChAT immunoreactivity to control levels (Figs. 9 10) . 

TUNEL staining was in agreement with the ChAT immunoreactivity data, showing increased apoptosis only in the AMPA-treated tissue and providing further support for the neuroprotective actions of the sst₂ analog (Fig. 11) . However, TUNEL labeling was observed in all retinal layers. This may suggest that AMPA receptor expression, which has not been detected in some retinal cells to date, may actually exist or that AMPA activation of caspases may be lead to cell death in other retinal cell layers. ^{57 58} The extensive TUNEL labeling, however, does not correlate with the lack of change of expression in retinal markers other that ChAT. This is in partial agreement with a recent study in the mouse that used iodoacetic acid and hypoxia to develop a retinal ischemia model. ³⁵ TUNEL labeling was observed in the three retinal layers—ONL, INL, and GCL—1 hour after exposure of the retina to the ischemia mixture, yet markers such as cone arrestin (marker for cone photoreceptors) were not affected up to 6 hours (maximum exposure) of ischemic treatment. In the same study, it was observed that different retinal markers were differentially affected according to the time of exposure to the ischemia mixture. ChAT immunoreactivity was completely lost as early as 1 hour and remained low, TH immunoreactivity was decreased but was still evident after 6 hours of ischemia, and TUNEL labeling increased as a function of time. ³⁵ In a model of chemical ischemia (iodoacetic acid and sodium cyanide), rod bipolar, ChAT, TH, and bNOS-expressing amacrine cells were mostly affected. However, TUNEL labeling was observed predominantly in the ONL and INL layers. ³⁴  

Retinal ischemia models involving hypoxia and hypoglycemia lead to the release of glutamate and the activation of NMDA and non-NMDA receptors and to subsequent toxicity and cell death. ^{1 34 35} A lack of concordance between TUNEL labeling and retinal marker expression was also observed in a study that characterized quisqualic acid and NMDA-induced excitotoxicity in chick retina. In this study, DNA fragmentation was observed in rod bipolar cells in which no detectable loss of PKC immunoreactivity was detected. The authors suggested that some cells damaged by the excitatory amino acid, though labeled by the TUNEL technique, were either not committed to apoptosis or were rescued by DNA repair. ⁴⁶ The lack of concordance between TUNEL labeling and retinal marker proteins and between protein and retinal marker transcripts ⁵⁹ has been observed in more paradigms than can be cited in this study. Future investigations focusing on the elucidation of the mechanisms involved in these processes will be important in providing specific answers to the questions posed. 

The present data support that the neuroprotective effect of SRIF is mediated by its activation of sst₂ receptors, because sst₁- and sst₄-selective ligands had no effect on drug concentrations that afforded protection by lanreotide and L-779976 (Fig. 12) . These results are in agreement with previous data from our laboratory that showed sst₂-selective ligands to be neuroprotective in a chemical model of retinal ischemia. ³⁴ Octreotide (sst_2/5 ligand) was also shown to protect the guinea pig retina from ischemia reperfusion injury. ⁶⁰ A more recent study ³⁵ complemented these data by showing that the overexpression of sst₂ receptors in mice lacking the sst₁ subtype prevented the retinal ischemia produced by iodoacetic acid and hypoxia. 

The mechanisms by which SRIFergic ligands act as neuroprotectants are still under investigation. SRIF is known to inhibit voltage-gated calcium channels ⁶¹ and neuronal calcium currents, the latter through a mechanism involving a cGMP-dependent protein kinase. ⁶² cGMP was also important in SRIF-protective actions against NMDA-induced neuronal death in cortical cultures. ³⁰ In a recent study, we showed that NO/peroxynitrite and cGMP are important mediators in the protection of rat retina from chemical ischemia and that a NO/sGC/cGMP signaling pathway is involved in the neuroprotective effects bestowed on the retina by the sst₂ SRIF ligands in the same model. ⁶³ Studies are in progress to examine whether such a mechanism is involved in SRIF protection of the retina against the AMPA insult. 

The importance of the present data lies in the fact that SRIF, particularly the two ligands tested (lanreotide and L-779976), afforded neuroprotection against excitotoxicity when administered intravitreally in vivo. Lanreotide is known to have a high affinity for sst₂ and sst₅ receptors. ⁶⁴ Until recently, the sst₅ subtype was not localized in the retina; thus, we assumed that the neuroprotective effects of lanreotide were restricted to its activation of the sst₂ subtype. Nevertheless, a recent study supported the presence of the sst₅ receptor in dopaminergic and cholinergic amacrine cells, ⁶⁵ and the possible contribution of this receptor to lanreotide’s effects cannot be excluded. L-779976 is a selective sst₂ ligand. ⁶⁶ However, one may criticize the concentrations of lanreotide and L-779976 (10⁻⁵M, 10⁻⁴ M) that were shown to be efficacious as neuroprotectants as high and possibly nonspecific. Similar concentrations were shown to be efficacious in the chemical ischemia model ³⁴ and in an experimental paradigm of choroidal neovascularization. ⁶⁷ Although these cited data do not justify the protocol, we can support the present findings for two reasons. First, we cannot conjecture about the actual drug concentration reaching the retina (possibly small) because we have no means of measuring drug levels after intravitreal injection. Second, the lack of neuroprotective effects of the sst₁- and sst₄-selective ligands ⁶⁶ and the absence of the sst₃ receptor in vertebrate retinas ¹⁸ suggest that lanreotide and L-779976 mediate their neuroprotective effects through activation of the sst₂ and possibly the sst₅ receptors. 

Many investigations in the literature have focused on the study of the use of SRIF analogs in the therapeutics of retinal diseases. Proliferative diabetic retinopathy (PDR) is a major cause of visual loss. Its pathophysiology involves increased vascularization or angiogenesis. SRIF has been shown to have antiangiogenic properties ^{28 29} and possibly to be useful in the treatment of PDR. A recent study reported that SRIF-28 is attenuated in the vitreous fluid of patients with PDR, adding further support to the significance of SRIF analogs in the treatment of this disease. ⁶⁸ A similar deficit of SRIF was detected in the vitreous of patients with diabetic macular edema. ⁶⁹ SRIF analogs have been used in a small number of ocular clinical studies. Recently, in a case study, octreotide (100 μg) was administered subcutaneously three times a day for 8 months to a 52-year-old woman with cystoid macular edema (CME). ⁷⁰ This treatment resulted in improvement of the patient’s visual acuity and partial resolution of CME. In addition, lanreotide (Somatuline; Ipsen, Paris, France) was administered subcutaneously twice a month for 6 months to patients with age-related maculopathy and was found to stabilize the patients’ visual acuity. ⁷¹  

The present in vivo data support the pharmacologic and possible therapeutic significance of the intravitreal administration of lanreotide and L-779976. Further studies are essential to examine whether the intravitreal injection of these analogs is devoid of pharmacokinetic problems associated with systemic administration, such as proper drug absorption and distribution. Octreotide acetate was injected into the vitreous cavity of kitten eyes, the long-acting octreotide-LAR was injected into the mid-vitreous of rabbit eyes, and the pharmacokinetics and toxicity of the ligands were examined. ⁷² This study concluded that intravitreal doses up to 1.1 mg octreotide-LAR could provide sustained high concentrations of the drug that would render it useful as a potential treatment of proliferative eye diseases. Another study evaluated the ocular toxicity of intravitreous octreotide and came to a similar conclusion as to the safety of octreotide at a concentration of less than 1 mg. ⁷³ More recently, a SRIF–camptothecin conjugate was examined for its ocular toxicity and efficacy after intravitreal administration in rabbits. It was found to be safe at the concentration of 10⁻⁵ M or less and to be efficacious in the treatment of age-related macular degeneration. ⁶⁷  

In conclusion, the present study offers new in vivo evidence suggesting that the SRIFergic ligands lanreotide and L-779676 protect the retina from excitatory amino acid damage. The pharmacologic profile of these agents renders them promising therapeutic agents in retinal diseases whose pathophysiologies involve ischemic and excitotoxic insult. There is still a great need for new, effective drugs for the treatment of retinopathies. Therefore, further investigations to substantiate the proper pharmacokinetics and the lack of toxicity of these agents in ocular tissues may render them useful in the treatment of retinopathies. 

 Supported by the European Social Fund and National Resources, Program PENED (KT), with the support of Novartis Pharma AG.

 Submitted for publication December 21, 2007; revised February 7, 2008; accepted May 9, 2008.

 Disclosure: F. Kiagiadaki, None; K. Thermos, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Dose-dependent effect of AMPA treatment on ChAT immunoreactivity in the retina. ChAT immunoreactivity in control rat retina (A) and in retina that received 21 (B), 42 (C), and 84 (D) nmol AMPA 1 day after injection. Partial loss of ChAT I-R was observed with 21 nmol/eye (only some processes of the cholinergic neurons appear), whereas the larger doses caused the complete loss of ChAT I-R. Scale bar, 50 μm.

Figure 2.

 

Time-dependent effect of AMPA treatment on ChAT immunoreactivity in the retina. ChAT immunoreactivity in rat retina that intravitreally received PBS (50 mM, control tissue; A) or AMPA (42 nmol/eye) 1 (B), 8 (C), 16 (D), and 30 (E) days after treatment. The complete loss of choline acetyltransferase expression from the first until the thirtieth day after treatment was observed. Arrows: nonspecific binding of the secondary antibody to blood vessels. Scale bar, 50 μm.

Figure 3.

 

Effect of AMPA treatment on PKC and MAP expression in the retina. PKC immunoreactivity in control (A) and AMPA (42 nmol/eye)–treated tissue (B), MAP^1A immunoreactivity in control (C), and AMPA-treated tissue 1 day after injection (D). Scale bar, 50 μm.

Figure 4.

 

Effect of AMPA treatment on TH and recoverin expression in the retina. TH immunoreactivity in control (A) and AMPA (42 nmol/eye)–treated tissue and (B) recoverin immunoreactivity in control (C) and AMPA-treated tissue (D) 1 day after injection. Scale bar, 50 μm.

Figure 5.

 

Effect of AMPA on bNOS immunoreactivity. bNOS-containing amacrine cells were counted in PBS and AMPA (42 nmol/eye)–treated retinal sections at 1, 8, 16, and 30 days. NOS-containing neurons were decreased 1 day after AMPA administration. However, the signal returned to control levels 8 days after AMPA injection. Results are presented as mean ± SEM.

Figure 6.

 

Effect of AMPA on SRIF levels. SRIF levels were measured by radioimmunoassay. No statistically significant changes were observed between the control and the AMPA (42 nmol/eye)–treated retinas 1 day after injection.

Figure 7.

 

Effect of AMPA on sst_2A expression levels. sst_2A Immunoreactivity in PBS-treated and AMPA (42 nmol/eye)–treated retina 1 day after injection (A, B). sst_2A Immunoreactivity is located in somata, axons, and processes of bipolar cells, and no differences between control and excitotoxic tissue were observed. Scale bar, 50 μm. Western blot studies on control retinal membranes (C). A protein of approximate molecular mass of 77 kDa was detected in the retina that was not visible in the presence of the sst₂ antigen. No differences in the expression of the receptor protein (72 kDa) were observed in the PBS- and AMPA-treated retinas (D).

Figure 8.

 

Effect of somatostatin on the AMPA-dependent loss of ChAT immunoreactivity in the retina. ChAT immunoreactivity in PBS-treated retina (A) and in retina that received AMPA (42 nmol) alone (B) or in combination with SRIF (10⁻⁵ M; C,10⁻⁴ M; D) 1 day after injection. Partial protection was afforded by SRIF 10⁻⁴ M. Scale bar, 50 μm.

Figure 9.

 

Effect of lanreotide (sst_2/5 analog) on the AMPA-dependent loss of ChAT immunoreactivity in the retina. ChAT immunoreactivity in PBS-treated retina (A) and in retina that received AMPA (42 nmol) alone (B) or in combination with lanreotide (10⁻⁵ M; C, 10⁻⁴ M; D) 1 day after injection. Lanreotide protected the retina in a dose-dependent manner. Scale bar, 50 μm.

Figure 10.

 

Effect of L-779976 (sst₂ analog) on the AMPA-dependent loss of ChAT immunoreactivity in the retina. ChAT immunoreactivity in PBS-treated retina (A) and in retina that received AMPA (42 nmol) alone (B) or in combination with L-779976 (10⁻⁶ M; C, 10⁻⁵ M; D, 10⁻⁴ M; E) 1 day after injection. L-779976 protected the retina in a dose-dependent manner. Scale bar, 50 μm. (F) Effect of AMPA (42 nmol) alone or in combination with L-779976 (10⁻⁴ M) on the number of bNOS expressing retinal cells. The AMPA effect is statistically different when compared to the control (*P < 0.05).

Figure 11.

 

Effect of L-779976 (sst₂ analog) on AMPA-induced apoptosis. TUNEL staining used to assess retinal cell death by AMPA and protection by the L-779976 analog 1 day after injection. AMPA afforded excitotoxic effects (B) compared with control (A). L-779976 (10⁻⁴ M) protected the retina from AMPA-induced apoptosis (C). Scale bar, 50 μm.

Figure 12.

 

Effect of L-797591 (sst₁ analog) and L-803087 (sst₄ analog) ligands on the AMPA-dependent loss of ChAT immunoreactivity in the retina. ChAT immunoreactivity in PBS-treated retina (A) and in retina that received AMPA (42 nmol) alone (B) or in combination with L-797591 (10⁻⁴ M; C) or L-803087 (10⁻⁴ M; D) 1 day after injection. Activation of sst₁ and sst₄ receptors had no effect on the AMPA-induced loss of ChAT expression. Scale bar, 50 μm.

The authors thank Anna Vasilaki and Miltiadis Tsilimbaris for the constructive discussions on this work, Eleni Renieri for excellent technical assistance, and Merck for kindly providing the selective sst L-analogs.