Regulation of GABA content by glucose in the chick retina

Experimental Eye Research 115 (2013) 206e215

Contents lists available at ScienceDirect

Experimental Eye Research journal homepage: www.elsevier.com/locate/yexer

Regulation of GABA content by glucose in the chick retina Vivian Sayuri Miya-Coreixas a, Raquel Maggesissi Santos a, b, Raul Carpi Santos a, Patrícia Franca Gardino b, Karin Calaza a, * a

Laboratory of Retinal Neurobiology, Department of Neurobiology and Neurosciences Program, Biology Institute, Fluminense Federal University, Outeiro São João Batista s/n., 24020-141 Niterói, Rio de Janeiro, Brazil b Laboratory of Retinal Neurobiology, Institute of Biophysics Carlos Chagas Filho, Health Sciences Center, Federal University of Rio de Janeiro, Carlos Chagas Filho Avenue 373, 21941-902 Rio de Janeiro, Rio de Janeiro, Brazil

a r t i c l e i n f o

a b s t r a c t

Article history: Received 16 April 2013 Accepted in revised form 25 July 2013 Available online 3 August 2013

Some visual information is processed in the retina by g-aminobutyric acid (GABA) signaling. Once retinal degeneration and visual impairment caused by diabetic retinopathy (DR) are affecting an increasing number of people worldwide, and the disease is characterized by hyper- and hypoglycemic events, the authors aimed to investigate how retinal GABA cell content is affected by variations in glucose availability. Using the ex vivo chick retinas exposed to different glucose concentrations, we observed that amacrine cells from both inner nuclear layer (INL) and ganglion cell layer (GCL) as well as their processes in the inner plexiform layer (IPL) released GABA through GABA transporter-1 (GAT-1) after 30 min of glucose deprivation. Extending this insult to 60 min triggered a permanent loss of GABA-positive amacrine cells, caused swelling of IPL and cell death. High glucose (35 mM) for 30 min induced an increment in GABA immunolabeling in both outer and inner retina. Further, glucose deprivation effects could not be reverted by basal glucose levels and high glucose did not prevent GABA loss upon a glucose deprivation insult. Therefore, GABA cell content is differently affected by short-term variations in glucose availability. While high glucose modulates outer and inner GABAergic circuits, glucose deprivation affects mainly the inner retina. Also, consecutive alteration in glucose supply was not able to rescue basal GABA content. Therefore, glucose oscillations interfering with GABAergic retinal functioning during early stages of retinopathies should be further investigated. Ó 2013 Elsevier Ltd. All rights reserved.

Keywords: glucose deprivation high glucose GABA chick retina

1. Introduction The retina is one of the major energy consuming tissues within the body, and its energetic source for normal functioning comes mainly from glucose metabolism. This necessity for glucose is evidenced by alteration in electroretinogram (ERG) responses and

Abbreviations: g-aminobutyric acid, GABA; diabetic retinopathy, DR; electroretinogram, ERG; embryonic day n, En; GABA immunoreactivity, GABA-IR; GABA transporter-1, GAT-1; ganglion cell layer, GCL; inner nuclear layer, INL; inner plexiform layer, IPL; outer nuclear layer, ONL; oscillatory potentials, OPs; outer plexiform layer, OPL; lactate dehydrogenase, LDH; sublamina, S; 1-[2-[[(Diphenylmethylene)imino]oxy]ethyl]-1,2,5,6-tetrahydro-3-pyridinecarboxylic acid hydrochloride, NNC-711. * Corresponding author. Departamento de Neurobiologia, Instituto de Biologia, Universidade Federal Fluminense, Outeiro São João Batista s/n., Campus do Valonguinho, Centro, CEP 24020-141 Niterói, Rio de Janeiro, Brazil. Tel.: þ55 2126292269. E-mail addresses: [email protected] (V.S. Miya-Coreixas), rmaggesissi@ yahoo.com.br (R. Maggesissi Santos), [email protected] (R. Carpi Santos), [email protected] (P.F. Gardino), [email protected], [email protected] (K. Calaza). 0014-4835/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.exer.2013.07.026

altered neurotransmitter release as described in ischemic and hypoglycemic conditions (Khan et al., 2011; Mukaida et al., 2004; Napper et al., 2001; Neal et al., 1994; Skrandies and Heinrich, 1992; Umino et al., 2006). Among the alterations observed after metabolic stress, reductions in GABA neuronal content could be noticed concomitant with a higher extracellular GABA availability and Müller cell localization (Napper et al., 2001; Osborne and Herrera, 1994; Zeevalk and Nicklas, 1991). GABA is the main inhibitory neurotransmitter in the mature retina and GABAergic inhibitory inputs constitute an important modulatory pathway for the processing of visual information (Barnstable, 1993; Massey and Redburn, 1987; Tachibana and Kaneko, 1988; Yang, 2004; Yazulla, 1986). On the other way, chronic elevated glucose availability to the retina, experimented by diabetic patients, induces a pathological condition known as DR. Although DR has long been recognized as a microvascular disease, increasing amount of evidence suggests that retinal neuronal cells are affected by high glucose availability prior to the development of vascular lesions. For instance, abnormalities in oscillatory potentials (OPs) can be detected in ERG as

V.S. Miya-Coreixas et al. / Experimental Eye Research 115 (2013) 206e215

early as 2 days following the onset of diabetes in rats (Phipps et al., 2004), and long before the DR diagnosis in humans (Lieth et al., 2000). Interestingly, it is speculated that early changes in amacrine cells could promote or contribute to these OPs abnormalities (Fletcher et al., 2007). Moreover, it has been reported that GABA levels were elevated in Müller cells of rat retina following 12 weeks of diabetes (Ishikawa et al., 1996), and it has also been suggested that many aspects of GABA signaling could be changed in in vivo hyperglycemia, e.g. GABA release, uptake, synthesis and degradation (Ishikawa et al., 1996; Kajiura et al., 1994; Vilchis and Salceda, 1996). Therefore, modifications in GABAergic signaling are observed in the context of DR whose pathophysiology is characterized by significant alterations in glucose supply to the retina. Moreover, recurrent hypoglycemic episodes are considered a worrying condition for insulin-treated diabetic patients’ health, including retinal and visual deficits (Frier, 1993; Khan et al., 2011) and there is no evidence whether short-term and consecutive variations in glucose offering to the retina can modify GABAergic signaling. Although in vivo studies have evidenced modifications in retinal GABAergic signaling (Ishikawa et al., 1996; Kajiura et al., 1994; Vilchis and Salceda, 1996), they do not rule out the possibility of a vascular component. In order to isolate and evaluate the direct effect of glucose onto GABA cell content without a possible vascular interference, we used the ex vivo avascular chick retina as a model of study. Thus, this study aimed to analyze the short-term effects of (i) different glucose concentrations and (ii) glucose oscillations onto retinal GABA content. 2. Material and methods 2.1. Animals White Leghorn chicks (post-hatched day 0e5) were housed in an adequate brooder after hatching with access to food and water ad libitum. They were maintained in a 12 h light/dark cycle. The procedures for the use of animals were approved by the Commission of Animal Care of Fluminense Federal University (CEUA/PROPPi) protocol number 73/2011. 2.2. Experimental procedure Chicks were killed and retinal pieces were obtained as previously described (Calaza et al., 2001; Guimaraes-Souza and Calaza, 2012; Guimaraes-Souza et al., 2011; Maggesissi et al., 2009). Modified Ringer’s solutions with different glucose concentrations (120 mM NaCl, 3 mM KCl, 1 mM NaH2PO4, 30 mM NaHCO3, 1 mM CaCl2, 1 mM MgCl2.6H2O and 0 mM, 5.6 mM, 10 mM or 35 mM glucose) were used. Prior to the experiments, these solutions were balanced with 95% O2/5% CO2 (Praxair Technology Inc., White Martins, Rio de Janeiro, Brazil) for 15 min, and then its pH was adjusted to 7.2e7.4. Each retinal piece was immersed in modified Ringer’s solution of interest and constantly perfused with 95% O2/ 5% CO2 throughout the experimental period at 37  C. In some experiments, Ringer’s solution contained different concentrations of mannitol, combined with different glucose concentrations. In some cases, retinal pieces were pre-incubated in the presence or absence of GAT-1 inhibitor, 1-[2-[[(Diphenylmethylene)imino] oxy]ethyl]-1,2,5,6-tetrahydro-3-pyridinecarboxylic acid hydrochloride (NNC-711, 100 mM) obtained from SigmaeAldrich (St. Louis, MO, USA) for 10 min. Then, they were submitted or not to glucose deprivation for an additional period (30 or 60 min) still in the presence of NNC-711.

207

2.3. Tissue processing After the experiment, retinal pieces were fixed by immersion in 4% paraformaldehyde in 0.16 M sodium phosphate buffer (pH 7.2) for 2 h. Retinas were submitted to sucrose gradient (15% and 30%) overnight for cryoprotection. Then, radial retinal sections (10 mm) were obtained in cryostat (CM1850, Leica Microsystems, Wetzlar, Germany). Control and treated retinal sections, from the same experiment, were collected in the same slide, and submitted to the same histological procedure. In some cases, retinal pieces were not submitted to incubation in Ringer’s solution. The animals were killed and retinal pieces were promptly fixed in paraformaldehyde, as described above. These non-incubated retinas were used in some experiments as the control condition. 2.4. Immunohistochemistry Sections were incubated with 5% bovine serum albumin (BSA, SigmaeAldrich, St. Louis, MO, USA) for 1 h, and then with 1:7000 polyclonal rabbit anti-GABA antibody (SigmaeAldrich, St. Louis, MO, USA) or 1:100 polyclonal rabbit anti-cleaved caspase-3 antibody (Asp 175) (Cell Signaling Technology, Danvers, MA, USA) overnight. Thereafter, sections were incubated with 1:200 biotinylated goat anti-rabbit IgG antibody (Vector Laboratories, Burlingame, CA, USA) for 2 h followed by avidin-biotin complex (1:50, Vectastain Elite, Vector Laboratories, Burlingame, CA, USA) for 90 min, and reacted with 0.05% 3,3-diaminobenzidine (Sigmae Aldrich, St. Louis, MO, USA) and 0.01% hydrogen peroxide for 10 min. Each incubation step previously mentioned was followed by washes with phosphate buffered saline (PBS, pH 7.4). BSA, primary and secondary antibodies and avidin-biotin complex were diluted in PBS plus 0.25% Triton X-100. The absence of primary antibody produced no specific labeling. 2.5. Cell quantification We applied a similar procedure as previously described (Maggesissi et al., 2009). Retinal sections were analyzed under light microscopy conjugated with differential interference contrast (DIC) (DM2500, 40 objective, Leica Microsystems, Wetzlar, Germany). The representative photomicrographs were obtained with DFC310 FX camera (Leica Microsystems, Wetzlar, Germany) under light microscopy conjugated with DIC (DM2500, 40 objective, Leica Microsystems, Wetzlar, Germany). 2.6. Optical densitometry We applied a similar procedure as previously described (Calaza et al., 2001; Guimaraes-Souza and Calaza, 2012; Guimaraes-Souza et al., 2011). The intensity of GABA immunoreactivity (GABA-IR) in both IPL and OPL was analyzed using digitized images captured with DFC310 FX camera (Leica Microsystems, Wetzlar, Germany) under light microscopy conjugated with DIC (DM2500, 40 objective, Leica Microsystems, Wetzlar, Germany). The digitized images were processed with Image J software (version 1.38, NIH, USA). Using the grid tool, the IPL was equally divided into five portions corresponding to sublaminas (S) 1-5 and the optical density (in arbitrary units, a.u.) of an area with the same size within each sublamina was measured. A similar procedure was applied to the OPL with no division. Densitometric analysis was performed using 8-bit gray scale images from retinal sections immunolabeled for GABA.

208

V.S. Miya-Coreixas et al. / Experimental Eye Research 115 (2013) 206e215

2.7. Morphometric analysis We measured the thickness of IPL in retinal sections stained by Nissl technique (0.25% cresyl violet) as a measure of the swelling process. Digitized images were captured with SV Micro camera (Sound Vision, USA) under light microscopy (Axioskop, 40 objective, Zeiss, Germany), and processed with Image J software (version 1.38, NIH, USA). Using a line tool, the length (mm) of the inner- to the outermost limits of the IPL was measured. The spatial resolution of the system was 7.0 pixels/mm.

(35 mM ¼ 5.6 mM glucose þ 29.4 mM mannitol), there was no modification in the number of GABA-positive cells, evidencing that the effects of 35 mM glucose were really induced by the augment in glucose concentration and not by a change in osmolarity. Meanwhile, retinas exposed to 0 mM glucose plus 5.6 mM mannitol still showed a reduction in the number of GABA-positive amacrine cells, indicating that the lack of glucose was the true trigger of reduced GABA content. However, the presence of mannitol during no glucose abolished the effect previously described for GCL cells, suggesting that these GABAergic cells are susceptible to osmotic variations.

2.8. Cellular viability assay Lactate dehydrogenase (LDH) activity was measured with a colorimetric assay (CytoTox 96Ò Non-Radioactive Cytotoxicity Assay, Promega Biotecnologia do Brasil Ltda., Rio de Janeiro, RJ, Brazil) as a measure of cell death. Briefly, after the experimental procedure, extracellular medium was collected and retinal pieces were lysed (with 0.9% Triton X-100). Then, both intracellular and extracellular samples were processed according to the manufacture’s guide. Absorbance was recorded at 490 nm in a microplate reader (iMark, BioRad, Philadelphia, USA). Data are represented as LDH released (extracellular)/total (extracellular þ intracellular). 2.9. Statistical analysis All data presented herein obtained from histological analysis were sampled as follows. At least five microscope fields were randomly chosen per retinal section (40 mm apart from each other), and at least three distinct retinal sections from the same retinal piece were used. Thus, the minimum of fifteen microscope fields per retinal piece, corresponding to one experimental group, were sampled. Also, at least three different animals were used for each experimental condition. All data present in the graphs are represented as mean  standard error of the mean (S.E.M.), whether in percentage of control or not. When comparing two groups, data were submitted to statistical analysis using Student t-test. And, when three or more groups were compared, One-way ANOVA followed by Bonferroni post hoc test was applied. Statistical significance was considered when p < 0.05 (GraphPad Prism, version 5.0, GraphPad Software, Inc., La Jolla, USA). 3. Results 3.1. GABA content in retinal cells is modified by glucose availability Retinal pieces were exposed to 0 mM, 5.6 mM or 35 mM glucose for 30 min. As 5.6 mM glucose is widely used as basal glucose concentration, it was referred herein as the control condition. Retinas incubated without glucose medium showed an overall reduction in GABA immunolabeling when compared to control (Fig. 1A). Indeed, GABA-positive amacrine and GCL cell populations were numerically diminished to approximately 60% and 75%, respectively (Fig. 1B and D). In addition, there was a significant change in the intensity of GABA-IR in sublamina (S) 4 and S5 of the IPL (Fig. 1F). However, the number of GABAergic horizontal cells and GABA-IR in the OPL did not change (Fig. 1C and E). 35 mM glucose, instead, clearly enhanced GABA immunolabeling in both OPL and IPL (S2 and S4) (Fig. 1E and F). Also, there was a statistically significant increase in the number of GABA-positive amacrine cells (Fig. 1B) and horizontal cells (Fig. 1C), but not in GCL cells (Fig. 1D). In order to exclude osmotic effects, mannitol was used as the osmotic control (Table 1). When 5.6 mM glucose was present

3.2. Glucose deprivation for 30 min does not induce death of GABApositive cells As shown in Fig. 1, glucose deprivation altered endogenous GABA content from the mature chick retina. However, it was not clear whether this effect was a physiological response of retinal cells to the lack of glucose, or represented a leakage of GABA because of cellular death triggered by the insult. Thus, we analyzed IPL thickness as a measure of retinal swelling, which could be an indicative of retinal damage. Retinas without glucose showed a slightly but not statistically significant augmented IPL thickness in relation to 5.6 mM glucose (control) retinas (0 mM: 56.84  2.23 mm, n ¼ 3; 5.6 mM: 51.49  2.74 mm, n ¼ 3; p > 0.05). Accordingly, we did not observe differences in retinal cell viability between these same groups when assessing LDH activity (Fig. 2B). Thus, the results obtained suggested that glucose deprivation for 30 min does not trigger retinal cell death when compared to the control condition containing 5.6 mM glucose. This led us to the hypothesis that the lack of glucose could modulate retinal GABA content, as shown in Fig. 1 and Table 1, in a manner independent of cell death. Interestingly, the blood glucose concentration of the chick embryo has been shown to be around 5 mM (Larger et al., 2004) whilst in the post-hatched, it is around 15 mM (Bowes et al., 1989). Therefore, since post-hatched chick blood glucose concentration is nearly 3-fold higher than the control condition used in the prior experiments, we supposed that 5.6 mM glucose may not be adequate to the chick retina and perhaps it may have had some effects on GABA content, leading us to a misinterpretation of the results obtained. In order to evaluate this issue, we augmented glucose concentration in Ringer’s solution about 2-fold (10 mM glucose), and analyzed IPL thickness and LDH activity in comparison to control condition containing 5.6 mM glucose after a 30 min incubation period. Insofar we have not found difference in IPL thickness from retinas incubated in 10 mM glucose solution when comparing to retinas exposed to 5.6 mM glucose (5.6 mM: 51.49  2.74 mm, n ¼ 3; 10 mM: 46.76  0.76 mm, n ¼ 3; p > 0.05), we have found a small but significant reduction of LDH-released activity in retinas exposed to 10 mM glucose (Fig. 2B). This result, thus, indicated that retinas incubated in a 5.6 mM glucose-containing solution for 30 min could exhibit some evidence of cell death, although in a small proportion, in relation to a 10 mM glucose condition. Consequently, to investigate if the observed cell death included some GABA-positive cells, we analyzed the number of GABApositive cells (amacrine cells, horizontal cells and GCL cells) and GABA-IR optical density in the IPL from 5.6 mM glucose-, 10 mM glucose-incubated and non-incubated retinas. Non-incubated retinas were used herein as the control condition since we consider it to be a standard for endogenous retinal GABA expression. Photomicrographs from retinal sections immunolabeled for GABA clearly show that endogenous GABA content is quite similar among the groups studied (Fig. 2A). Also, Fig. 2CeF shows that the number of GABA-positive amacrine, horizontal and GCL cells as well as optical density of GABA-IR in IPL were not different between 5.6 mM and

V.S. Miya-Coreixas et al. / Experimental Eye Research 115 (2013) 206e215

209

Fig. 1. GABA content in retinal cells is modified by glucose availability. (A) Photomicrographs of chick retinal sections immunolabeled for GABA after exposure to 0 mM, 5.6 mM (control) and 35 mM glucose for 30 min (BeD). Graphs show the percentual number of GABA-positive (B) amacrine cells (0 mM: 61.10  10.93%, n ¼ 5; 35 mM: 124.00  5.65%, n ¼ 6), (C) horizontal cells (0 mM: 89.84  8.77%, n ¼ 3; 35 mM: 246.50  19.99%, n ¼ 3) and (D) GCL cells (0 mM: 73.84  6.45%, n ¼ 3; 35 mM: 117.60  4.80%, n ¼ 4) from retinas exposed to the experimental conditions previously described in relation to 5.6 mM glucose (control) group. The mean number of cells counted/section in control condition corresponded to: amacrine cells ¼ 1437.00  176.40; horizontal cells ¼ 167.40  3.18; GCL cells ¼ 409.80  28.48. (E, F) Optical density of GABA-IR in the OPL (n ¼ 3) and in the IPL (S15, n ¼ 4) were measured in treated groups in relation to 5.6 mM glucose (control) group. Control retinas showed mean optical density (a.u.) in the OPL equal to 0.07  0.003. And the mean optical density in the IPL from control retinas was equal to: S1 ¼ 0.87  0.05; S2 ¼ 0.70  0.09; S3 ¼ 0.89  0.04; S4 ¼ 0.84  0.07 and S5 ¼ 0.96  0.07. *p < 0.05, **p < 0.01 and ***p < 0.001 in relation to control group. All retinal layers are identified in the figure as follows, and these abbreviations are applied to all other photomicrographs presented: ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer and S, sublamina. Scale bar ¼ 20 mm.

Amacrine cells

Horizontal cells

GCL cells

10 mM glucose conditions. Therefore, these data evidence that incubation of retinas in a 5.6 mM glucose solution for 30 min does not induce GABA-positive cell loss neither GABA release from IPL regardless of promoting cell death in relation to a 10 mM glucose solution.

100.00 76.41  2.96a

100.00 96.93  7.75

100.00 94.69  1.28

3.3. Glucose deprivation induces GABA release in a transporterdependent manner

97.40  3.71

90.15  8.83

101.2  0.61

Since our group has previously described that glutamatergic agonists and nitric oxide can modulate GABA release through GAT-1 in mature chick retinas (Calaza et al., 2001; Guimaraes-Souza and Calaza, 2012; Guimaraes-Souza et al., 2011; Maggesissi et al., 2009),

Table 1 Effects of glucose in the number of GABA-positive retinal cells in the presence of mannitol.

5.6 mM glucose (control) 0 mM glucose (0 mM glucose þ 5.6 mM mannitol) 35 mM glucose (5.6 mM glucose þ 29.4 mM mannitol) a

p < 0.01 in relation to control.

210

V.S. Miya-Coreixas et al. / Experimental Eye Research 115 (2013) 206e215

Fig. 2. Glucose deprivation for 30 min does not induce death of GABA-positive cells. (A) Photomicrographs of chick retinal sections immunolabeled for GABA from non-incubated retinas (control) or after exposure to 5.6 mM or 10 mM glucose for 30 min. (B) Graph shows the percentage of LDH released/total from retinas exposed to 0 mM and 10 mM glucose in relation to 5.6 mM glucose (control) condition after 30 min of incubation (0 mM: 103.90  3.87%, n ¼ 3; 10 mM: 79.93  5.87%, n ¼ 3). (CeE) Graphs show the percentual number of GABA-positive (C) amacrine cells (5.6 mM: 86.32  2.69%, n ¼ 3; 10 mM: 90.47  4.32%, n ¼ 3), (D) horizontal cells (5.6 mM: 57.08  12.00%, n ¼ 3; 10 mM: 83.40  21.94%, n ¼ 3) and (E) GCL cells (5.6 mM: 105.90  4.69%, n ¼ 3; 10 mM: 100.30  7.43%, n ¼ 3) from retinas exposed to 5.6 mM or 10 mM glucose in relation to non-incubated (control) group. The mean number of cells counted/section in control condition corresponded to: amacrine cells ¼ 803.60  39.05; horizontal cells ¼ 131.70  32.92; GCL cells ¼ 140.80  29.47. (F) Optical density of GABA-IR in the IPL (S1-5, n ¼ 3) measured in treated groups in relation to non-incubated (control) group. Control retinas showed mean optical density (a.u.) in the IPL equal to: S1 ¼ 1.15  0.06; S2 ¼ 1.13  0.06; S3 ¼ 1.17  0.06; S4 ¼ 1.18  0.04 and S5 ¼ 1.23  0.05. *p < 0.05 in relation to control (5.6 mM glucose or no incubation) group. Scale bar ¼ 20 mm.

we decided to investigate if a transporter-mediated process of GABA release could be involved in these glucose deprivation effects. Then, we incubated retinal pieces with GAT-1 inhibitor, NNC-711 (100 mM) for 10 min prior to glucose deprivation event, which was induced for 30 min still in the presence of NNC-711. Indeed, NNC-711 prevented the reduction in the number of GABA-positive amacrine and GCL cells induced by 0 mM glucose (Fig. 3). Thus, these results suggest that the diminished GABA content either caused by no glucose (in amacrine cells, Fig. 3B) or osmotic unbalance (in GCL cells, Fig. 3C) were due to an increased GABA release through GAT-1 activity. 3.4. Prolonged glucose deprivation exposure triggers retinal cell death Since we had observed a tendency of glucose deprivation to increase IPL thickness in relation to control condition containing

5.6 mM glucose, we were interested to investigate if a longer exposure (60 min) of retinal cells to this condition could induce damage or if GABAergic cells were resilient to this condition. However, GABA immunolabeling of retinas exposed to 5.6 mM glucose for 60 min was strongly different from the one observed in non-incubated retinas (Fig. 4A). Not only GABA immunolabeling was notorious- and qualitatively diminished as well as the number of GABA-positive cells was significantly reduced (Fig. 4AeD). Considering that 10 mM glucose solution was previously shown to effectively maintain retinal GABA content similar to non-incubated retinas for 30 min (Fig. 2), we wondered how retinas would respond to 60 min incubation in a 10 mM glucose solution. As expected, 10 mM glucose was also effective in maintaining retinal GABA content similar to non-incubated retinas after 60 min of incubation (Fig. 4AeD). Therefore, this was the glucose concentration chosen to be used in the control group for studies within a 60 min period.

V.S. Miya-Coreixas et al. / Experimental Eye Research 115 (2013) 206e215

211

Fig. 3. GAT-1 mediates GABA release induced by 30 min of glucose deprivation or osmotic unbalance. (A) Photomicrographs of chick retinal sections immunolabeled for GABA after exposure to 5.6 mM glucose (control), or glucose deprivation in the presence or absence of GAT-1 inhibitor (NNC-711, 100 mM). (B, C) Graphs show the percentual number of GABApositive (B) amacrine cells (0 mM: 56.67  8.02%, n ¼ 5; 0 mM þ NNC-711: 99.70  6.71%, n ¼ 4) and (C) GCL cells (0 mM: 70.99  5.87%, n ¼ 5; 0 mM þ NNC-711: 111.5  7.85%, n ¼ 4) after the treatments previously described in relation to 5.6 mM glucose (control) group. The mean number of cells counted/section in control condition corresponded to: amacrine cells ¼ 922.60  50.15; GCL cells ¼ 251.60  11.93. **p < 0.01 and ***p < 0.001 in relation to control group; ###p < 0.001 in relation to 0 mM glucose group. Scale bar ¼ 20 mm.

Further, we noticed that 60 min of glucose deprivation induced a robust increase in both IPL thickness (Fig. 4E) and extracellular LDH activity (Fig. 4F). Additionally, we did not find an augment in the number of cleaved caspase-3-positive cells (data not shown). Thus, these results suggest that a lack of glucose for a longer period induces retinal swelling and caspase-independent cell death. Finally, when retinal GABA content was analyzed after 60 min of glucose deprivation in comparison to 10 mM glucose control group, we observed an intense decrease in GABA immunolabeling mainly in the INL cells and in the IPL (Fig. 5A). The number of GABApositive amacrine cells was reduced to approximately 30% of control (Fig. 5B) while the number of GABA-positive GCL cells remained similar to one found after 30 min of glucose deprivation (compare Figs. 5C and 3C) and after 60 min of incubation in a 5.6 mM glucose condition (compare Figs. 5C and 4D). Pre-treatment of retinal pieces with NNC-711 had no effect on preventing the reduction in the number of GABA-positive cells after 60 min of 0 mM glucose, suggesting that the release of GABA to the extracellular medium was probably caused by cell death.

pieces to glucose deprivation for 30 min and, then, readily added glucose (10 mM) for additional 30 min. However, we observed that restoring glucose availability to the retina was not able to rescue GABA cell content (Fig. 6AeD). The restoration of glucose levels was indeed more detrimental to GABAergic signaling than 30 min of 0 mM glucose alone, because we could see a 75% decrease in GABApositive amacrine cell number and that all five sublaminas of IPL had a diminished GABA content (compare Figs. 1 and 6). Furthermore, in order to evaluate if a hyperglycemic-mimetic condition would prevent the effects of a hypoglycemic-like event, retinal pieces were exposed to 35 mM glucose for 30 min and then exposed to no glucose. GABA immunolabeling in amacrine cells and in IPL were diminished beyond control (Fig. 6AeD), indicating that a glucose deprivation event has a pronounced effect in counteracting the augment in GABA content promoted by 35 mM glucose (as seen in Fig. 1). However, the exposure to high glucose (35 mM) partially prevented retinal GABA release that would have been evoked by a glucose deprivation insult. Actually, in S5 of the IPL, 35 mM glucose pre-incubation maintained GABA-IR similar to control (Fig. 6D).

3.5. Modifications in GABA content induced by low and high glucose concentrations cannot be reversed

4. Discussion

In an attempt to mimic glucose oscillations, we investigated if consecutive modifications in glucose availability to the retina would rescue normal GABA cell content. We submitted retinal

In this study, mature chick retinas exposed to 5.6 mM glucose for 60 min showed an overall reduction of GABA immunolabeling when compared to non-incubated retinas. Interestingly, those

212

V.S. Miya-Coreixas et al. / Experimental Eye Research 115 (2013) 206e215

Fig. 4. Glucose deprivation for 60 min induces retinal swelling and retinal cell death. (A) Photomicrographs of chick retinal sections immunolabeled for GABA from non-incubated (control) retinas or after exposure to 5.6 mM glucose or 10 mM glucose conditions. (BeD) Graphs show the percentual number of GABA-positive (B) amacrine cells (5.6 mM: 56.99  2.79%, n ¼ 4; 10 mM: 89.07  4.17%, n ¼ 4), (C) horizontal cells (5.6 mM: 30.48  7.56%, n ¼ 4; 10 mM: 78.81  23.73%, n ¼ 4) and (D) GCL cells (5.6 mM: 78.86  8.12%, n ¼ 4; 10 mM: 89.27  1.28%, n ¼ 4) from retinas exposed to the experimental conditions previously described in relation to non-incubated (control) retinas. The mean number of cells counted/section in control condition corresponded to: amacrine cells ¼ 792.00  11.04; horizontal cells ¼ 117.70  18.07; GCL cells ¼ 157.40  15.52. (E) IPL layer thickness (mm) was measured as a morphological parameter of retinal swelling. The graph shows that retinal exposure to 0 mM glucose for 60 min significantly increases IPL thickness (81.75  5.21 mm; n ¼ 3) in relation to control 10 mM glucose (40.30  2.63 mm, n ¼ 3). (F) Graph shows the percentage of LDH released/total from retinas exposed to 0 mM glucose for 60 min in relation to control 10 mM glucose (217.00  27.60%; n ¼ 4). *p < 0.05 in relation to control group (no incubation or 10 mM glucose); ***p < 0.001 in relation to no incubation (control) group; ###p < 0.001 in relation to 10 mM glucose group. Scale bar ¼ 20 mm.

retinas resembled the ones submitted to 30 min of glucose deprivation, suggesting that 5.6 mM glucose usually present in experimental media induces a glucose deprivation-like event to mature chick retina. In agreement with that, the chick blood glucose concentration is around 15 mM as described elsewhere (Bowes et al., 1989), 2e3-fold higher than in rats (5.6e10 mM) (Johnson et al., 2013). It is interesting to mention that the blood glucose concentration of chicken embryo is similar to the rat’s (5.5 mM at embryonic day 8, E8) (Larger et al., 2004) and increases with development (8.7 mM at E17) (Larger et al., 2004; Yoshiyama et al., 2005). Also, differently from mammalian retinas, avian retinas are considered avascular. Avian do not have an intraretinian vascular bed associated with the inner retina, but they have a structure called pecten oculi instead. The pecten oculi projects itself to the vitreous body, and among its functions is the nutritional support of the retina since it is characterized to have blood-retinal-barrier properties (Brach, 1977; Gerhardt et al., 1996; Liebner et al., 1997; Mann, 1924). In fact, nutrients provided by the pecten oculi

become available to the vitreous body and, then, they reach the inner retinal cells by diffusion. Therefore, we believe that although chick blood glucose concentration can range around 15 mM, probably this is not exactly the glucose concentration that effectively reaches the retina. Even though we have not find differences in GABA immunolabeling among 5.6 mM and 10 mM glucose groups after 30 min of incubation (Fig. 2), we found 5.6 mM glucose-incubated retinas to have slightly fewer GABA-positive amacrine cells than nonincubated retinas. Then, we found that 10 mM glucose for 60 min could also maintain GABA immunolabeling very similarly to the pattern of non-incubated retinas. Thus, based on these evidences, we suggest 10 mM as a more appropriate glucose concentration for physiologic solutions used in ex vivo control conditions in studies concerning the mature chick retina. Embryonic chicken retina showed significant GABA release once subjected to glucose deprivation or chemical hypoglycemia (Zeevalk and Nicklas, 1991, 2000). Accordingly, in this paper,

V.S. Miya-Coreixas et al. / Experimental Eye Research 115 (2013) 206e215

213

Fig. 5. Glucose deprivation for 60 min induces a notorious reduction in retinal GABA content, which does not depend on GAT-1 activity. (A) Photomicrographs of chick retinal sections immunolabeled for GABA after exposure to 0 mM, 10 mM glucose (control), or 0 mM glucose þ NNC-711, for 60 min. (BeC) Graphs show the percentual number of GABApositive (B) amacrine cells (0 mM: 27.11  7.30%, n ¼ 6; 0 mM þ NNC-711: 49.88  12.47%, n ¼ 4) and (C) GCL cells (0 mM: 76.09  7.29%, n ¼ 7; 0 mM þ NNC-711: 72.97  9.81, n ¼ 5) from retinas exposed to the experimental conditions previously described in relation to 10 mM glucose (control) group. The mean number of cells counted/section in control condition corresponded to: amacrine cells ¼ 836.20  49.98; GCL cells ¼ 245.70  11.33. *p < 0.05 and ***p < 0.001 in relation to control group. Scale bar ¼ 20 mm.

mature chick retinas incubated in 0 mM glucose for 30 min displayed a GABA immunolabeling reduction, consequent of an increased GABA release through GAT-1. However, after 60 min without glucose, NNC-711 could no longer block GABA release probably because of cell death, as evidenced by LDH release. Therefore, glucose deprivation seems to promote changes in GABA cell content in the mature chick retina before triggering cell death. It is worth noting that the post-hatched stage is more suitable for the study of the mature retina because retinal cells have accomplished neurogenesis, cell death and differentiation periods (Mey and Thanos, 2000; Prada et al., 1991a, 1991b). Our results also showed that short period of normoglycemiclike conditions are not able to restore GABAergic retinal content after glucose deprivation. Moreover, we verified that data obtained from GABA-positive cell counting (Fig. 6B) was very similar to the one achieved with GABA-IR intensity analysis (Fig. 6C), evidencing that GABA-IR optical density analysis may be further used to measure GABA immunolabeling in retinal tissue. Our data showed that glucose deprivation for 30 min significantly reduced GABA immunolabeling in S4 and S5 of the IPL. However, even with a posterior restoration of basal glucose level, a reduction of GABA immunolabeling was still observed in all IPL sublaminas. Interestingly, the reduction of GABA immunolabeling observed in S5 promoted by 30 min of glucose deprivation could be partially prevented by incubation with high glucose (35 mM) for 30 min. As the neurons that extend to S4 and S5 are involved in ON retinal circuit (Famiglietti and Kolb, 1976), our data suggest that glucose deprivation may promote early changes in ON retinal circuit.

We showed that chick retinas exposure to high glucose presented an overall increase in GABA immunolabeling. Notably, this is the opposite effect of glucose deprivation. Therefore, one may argue that GABA synthesis was raised due to the increase of glucose availability, once glucose may be an indirect substrate for the GABA synthesis (Dingledine and Hassel, 2006), or that high glucose could promote an increase in GABA uptake, but this issue awaits further studies. Comparing to 5.6 mM glucose, high glucose (35 mM) promoted an increase in GABA immunolabeling in S2 and S4 of the IPL, which have been reported to display a preferential GABA release evoked by glutamate receptor agonists in retinas of different species (Calaza et al., 2001, 2006; Osborne and Herrera, 1994; Osborne et al., 1995). Therefore, glutamate signaling could be less active in high glucose conditions, inducing a reduction of GABA release and/ or an increase of GABA uptake. Some alterations of retinal GABAergic system have been reported in STZ-induced diabetic model (Ishikawa et al., 1996; Kajiura et al., 1994; Vilchis and Salceda, 1996). According to Kajiura and coworkers (Kajiura et al., 1994), increases in GABA content began 1 week after occurrence of diabetes mellitus and reached a peak at 2 months, maintaining up to 5 months. Additionally, the latencies of OPs of ERG were prolonged beginning 1 month after the occurrence of diabetes. This suggests that increased GABA signaling may be related to the abnormality of OPs. However, these in vivo studies cannot clearly demonstrate whether neural GABA changes are derived from vascular damage induced by hyperglycemia or from hyperglycemia directly. Our data indicate that neurons of avascular chick retina suffer a similar GABA labeling alteration after 30 min of

214

V.S. Miya-Coreixas et al. / Experimental Eye Research 115 (2013) 206e215

Fig. 6. Oscillations in glucose availability to the retina do not rescue normal GABA cell content. (A) Photomicrographs of chick retinal sections immunolabeled for GABA after different protocols of alternating glucose concentration: 10 mM glucose (control), 0 mM/10 mM (0 mM glucose solution for 30 min þ 10 mM solution for additional 30 min) and 35 mM/0 mM (35 mM glucose solution for 30 min þ 0 mM solution for additional 30 min). (B, C) Respectively, graphs show the percentual (B) number of GABA-positive amacrine cells (0 mM/10 mM ¼ 27.78  2.61%, n ¼ 3; 35 mM/0 mM ¼ 46.98  1.56, n ¼ 3) and (C) optical density of GABA-IR in amacrine cells (0 mM/10 mM ¼ 19.99  3.72%, n ¼ 4; 35 mM/ 0 mM ¼ 60.43  1.98, n ¼ 3) after retinal treatment as previously described in relation to 10 mM glucose (control) group. The mean number of amacrine cells counted in control and the mean optical density in the same condition corresponded to 1104  69.90 and 0.50  0.13, respectively. (D) Optical density of GABA-IR in IPL (S1-5, n ¼ 3) measured in treated groups in relation to 10 mM glucose (control) group. Control retinas showed mean optical density in the IPL equal to: S1 ¼ 0.68  0.20; S2 ¼ 0.64  0.17; S3 ¼ 0.74  0.21; S4 ¼ 0.80  0.27 and S5 ¼ 0.79  0.27. **p < 0.01 and ***p < 0.001 in relation to 10 mM glucose (control) condition; ##p < 0.01 and ###p < 0.001 when comparing 0 mM/10 mMe 35 mM/0 mM groups. Scale bar ¼ 20 mm.

high glucose exposure, suggesting that retinal neurons are directly and early affected by this condition. These early changes in GABA metabolism can account for the pathogenesis of the abnormal OPs in diabetes, as detected in diabetic SpragueeDawley rats (Phipps et al., 2004). Additionally, some authors consider that early glial/ neuronal abnormalities promoted by hyperglycemia could trigger late DR vascular lesions (Fletcher et al., 2007; Han et al., 2004; Lieth et al., 2000). The knowledge that hypoglycemia and hyperglycemia severely impair diabetic patients’ vision (Frier, 1993; Khan et al.,

2011; Umino et al., 2006) calls the attention for the need of studies that help to elucidate DR pathogenesis. The present study contributes to this body of evidence and sheds light on the necessity to further investigate the early effects of glycemic disturbs. Role of the founding source This work was supported by grants from CNPq, CAPES, PROPPiUFF, FAPERJ, INNT/INCT/CNPq and PRONEX/MCT. VSMC, RMS and

V.S. Miya-Coreixas et al. / Experimental Eye Research 115 (2013) 206e215

RCS were recipients of graduate student fellowships from CAPES. PFG and KCC are research fellows from CNPq. The sponsors had no involvement in the study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the article for publication. Author contributions Conceived and designed the experiments: VSMC, RMS, RCS, PFG and KCC. Performed the experiments: VSMC, RMS and RCS. Analyzed the data: VSMC, RMS, RCS, KCC and PFG. Contributed with reagents/materials/analysis tools: PFG and KCC. Participated in the article preparation: VSMC, RMS, RCS, PFG and KCC. Acknowledgments The authors thank MSc Elisa Maria Guimarães-Souza for helpful suggestions and manuscript revision. We also thank Dr Ana Lúcia Marques Ventura from Laboratory of Neurochemistry, Biology Institute, UFF, who provided the anti-cleaved caspase-3 antibody. We also thank Sarah Rodrigues for technical assistance. We state that none of this material has been submitted elsewhere. References Barnstable, C.J., 1993. Glutamate and GABA in retinal circuitry. Curr. Opin. Neurobiol. 3, 520e525. Bowes, V.A., Julian, R.J., Stirtzinger, T., 1989. Comparison of serum biochemical profiles of male broilers with female broilers and White Leghorn chickens. Can. J. Vet. Res. 53, 7e11. Brach, V., 1977. The functional significance of the avian pecten: a review. The Condor 79, 321e327. Calaza, K.C., de Mello, F.G., Gardino, P.F., 2001. GABA release induced by aspartatemediated activation of NMDA receptors is modulated by dopamine in a selective subpopulation of amacrine cells. J. Neurocytol. 30, 181e193. Calaza, K.C., Hokoc, J.N., Gardino, P.F., 2006. GABAergic circuitry in the opossum retina: a GABA release induced by L-aspartate. Exp. Brain Res. 172, 322e 330. Dingledine, R., Hassel, B., 2006. Glutamate. In: Siegel, G.J., Albers, R.W., Brady, S.T., Price, D.L. (Eds.), Basic Neurochemistry e Molecular, Cellular and Medical Aspects, seventh ed. Elsevier Academic Press, San Diego, CA, pp. 267e290. Famiglietti Jr., E.V., Kolb, H., 1976. Structural basis for ON-and OFF-center responses in retinal ganglion cells. Science 194, 193e195. Fletcher, E.L., Phipps, J.A., Ward, M.M., Puthussery, T., Wilkinson-Berka, J.L., 2007. Neuronal and glial cell abnormality as predictors of progression of diabetic retinopathy. Curr. Pharm. Des. 13, 2699e2712. Frier, B.M., 1993. Management of insulin-dependent diabetes: hypoglycaemia and the eye. Eye 7 (Pt 2), 293e297. Gerhardt, H., Liebner, S., Wolburg, H., 1996. The pecten oculi of the chicken as a new in vivo model of the blood-brain barrier. Cell Tissue Res. 285, 91e100. Guimaraes-Souza, E.M., Calaza, K.C., 2012. Selective activation of group III metabotropic glutamate receptor subtypes produces different patterns of gammaaminobutyric acid immunoreactivity and glutamate release in the retina. J. Neurosci. Res. 90, 2349e2361. Guimaraes-Souza, E.M., Gardino, P.F., De Mello, F.G., Calaza, K.C., 2011. A calciumdependent glutamate release induced by metabotropic glutamate receptors I/II promotes GABA efflux from amacrine cells via a transporter-mediated process. Neuroscience 179, 23e31. Han, Y., Bearse Jr., M.A., Schneck, M.E., Barez, S., Jacobsen, C.H., Adams, A.J., 2004. Multifocal electroretinogram delays predict sites of subsequent diabetic retinopathy. Invest. Ophthalmol. Vis. Sci. 45, 948e954. Ishikawa, A., Ishiguro, S., Tamai, M., 1996. Changes in GABA metabolism in streptozotocin-induced diabetic rat retinas. Curr. Eye Res. 15, 63e71.

215

Johnson, L.E., Larsen, M., Perez, M.T., 2013. Retinal adaptation to changing glycemic levels in a rat model of type 2 diabetes. PLoS One 8, e55456. Kajiura, Y., Inoue, M., Yamamoto, M., Okada, Y., 1994. Alteration of gammaaminobutyric acid in streptozotocin-induced diabetic rat retina. Nippon Ganka Gakkai Zasshi 98, 449e452. Khan, M.I., Barlow, R.B., Weinstock, R.S., 2011. Acute hypoglycemia decreases central retinal function in the human eye. Vis. Res. 51, 1623e1626. Larger, E., Marre, M., Corvol, P., Gasc, J.M., 2004. Hyperglycemia-induced defects in angiogenesis in the chicken chorioallantoic membrane model. Diabetes 53, 752e761. Liebner, S., Gerhardt, H., Wolburg, H., 1997. Maturation of the blood-retina barrier in the developing pecten oculi of the chicken. Brain Res. Dev. Brain Res. 100, 205e 219. Lieth, E., Gardner, T.W., Barber, A.J., Antonetti, D.A., Penn State Retina Research, G, 2000. Retinal neurodegeneration: early pathology in diabetes. Clin. Experiment. Ophthalmol. 28, 3e8. Maggesissi, R.S., Gardino, P.F., Guimaraes-Souza, E.M., Paes-de-Carvalho, R., Silva, R.B., Calaza, K.C., 2009. Modulation of GABA release by nitric oxide in the chick retina: different effects of nitric oxide depending on the cell population. Vis. Res. 49, 2494e2502. Mann, I.C., 1924. The function of the pecten. Br. J. Ophthalmol. 8, 209e226. Massey, S.C., Redburn, D.A., 1987. Transmitter circuits in the vertebrate retina. Prog. Neurobiol. 28, 55e96. Mey, J., Thanos, S., 2000. Development of the visual system of the chick. I. Cell differentiation and histogenesis. Brain Res. Brain Res. Rev. 32, 343e379. Mukaida, Y., Machida, S., Masuda, T., Tazawa, Y., 2004. Correlation of retinal function with retinal histopathology following ischemia-reperfusion in rat eyes. Curr. Eye Res. 28, 381e389. Napper, G.A., Pianta, M.J., Kalloniatis, M., 2001. Localization of amino acid neurotransmitters following in vitro ischemia and anoxia in the rat retina. Vis. Neurosci. 18, 413e427. Neal, M.J., Cunningham, J.R., Hutson, P.H., Hogg, J., 1994. Effects of ischaemia on neurotransmitter release from the isolated retina. J. Neurochem. 62, 1025e 1033. Osborne, N.N., Herrera, A.J., 1994. The effect of experimental ischaemia and excitatory amino acid agonists on the GABA and serotonin immunoreactivities in the rabbit retina. Neuroscience 59, 1071e1081. Osborne, N.N., Larsen, A., Barnett, N.L., 1995. Influence of excitatory amino acids and ischemia on rat retinal choline acetyltransferase-containing cells. Invest. Ophthalmol. Vis. Sci. 36, 1692e1700. Phipps, J.A., Fletcher, E.L., Vingrys, A.J., 2004. Paired-flash identification of rod and cone dysfunction in the diabetic rat. Invest. Ophthalmol. Vis. Sci. 45, 4592e 4600. Prada, C., Puga, J., Perez-Mendez, L., Lopez, Ramirez, G., 1991a. Spatial and temporal patterns of neurogenesis in the chick retina. Eur. J. Neurosci. 3, 1187. Prada, C., Puga, J., Perez-Mendez, L., Lopez, R., Ramirez, G., 1991b. Spatial and temporal patterns of neurogenesis in the chick retina. Eur. J. Neurosci. 3, 559e 569. Skrandies, W., Heinrich, H., 1992. Differential effects of mild hypoglycemia on proximal and distal retinal structures in man as revealed by electroretinography. Neurosci. Lett. 134, 165e168. Tachibana, M., Kaneko, A., 1988. Retinal bipolar cells receive negative feedback input from GABAergic amacrine cells. Vis. Neurosci. 1, 297. Umino, Y., Everhart, D., Solessio, E., Cusato, K., Pan, J.C., Nguyen, T.H., Brown, E.T., Hafler, R., Frio, B.A., Knox, B.E., Engbretson, G.A., Haeri, M., Cui, L., Glenn, A.S., Charron, M.J., Barlow, R.B., 2006. Hypoglycemia leads to age-related loss of vision. Proc. Natl. Acad. Sci. U. S. A. 103, 19541e19545. Vilchis, C., Salceda, R., 1996. Effect of diabetes on levels and uptake of putative amino acid neurotransmitters in rat retina and retinal pigment epithelium. Neurochem. Res. 21, 1167e1171. Yang, X.L., 2004. Characterization of receptors for glutamate and GABA in retinal neurons. Prog. Neurobiol. 73, 127e150. Yazulla, S., 1986. GABAergic mechanisms in the retina. Prog. Retin. Res. 5, 1e52. Yoshiyama, Y., Sugiyama, T., Kanke, M., 2005. Experimental diabetes model in chick embryos treated with streptozotocin. Biol. Pharm. Bull. 28, 1986e 1988. Zeevalk, G.D., Nicklas, W.J., 1991. Mechanisms underlying initiation of excitotoxicity associated with metabolic inhibition. J. Pharmacol. Exp. Ther. 257, 870e878. Zeevalk, G.D., Nicklas, W.J., 2000. Lactate prevents the alterations in tissue amino acids, decline in ATP, and cell damage due to aglycemia in retina. J. Neurochem. 75, 1027e1034.