Adenosine A2A Receptor: A New Neuroprotective Target in Light-Induced Retinal Degeneration

Continuous illumination induces the degeneration of photoreceptors. This animal model of light-induced retinal degeneration resembles many characteristics of human degenerative diseases of the outer retina, such as age-related macular degeneration. This work aimed to evaluate the potential neuroprotective effect of the modulation of adenosine A2A receptor in the model of light-induced retinal degeneration. Sprague-Dawley rats were intravitreally injected in the right eye with either CGS 21680, an adenosine A2A receptor agonist, or SCH 58261, an adenosine A2A receptor antagonist. Contralateral eyes were injected with respective vehicles as control. Then, rats were subjected to continuous illumination (12,000 lux) for 24 h. Retinas were processed by glial fibrillary acidic protein (GFAP) immunohistochemistry, terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) technique, Western blotting (WB), and quantitative reverse transcription-polymerase chain reaction (qRT-PCR). Another group of rats was subjected to functional studies by electroretinography. Animals treated with CGS21680 showed a significant increase of apoptotic nuclei in the outer nuclear layer and a significant increase of GFAP immunoreactive area of the retinas but did not alter WB nor electroretinography results. qRT-PCR showed that CGS 21680 significantly increased the expression of interleukin-1β. On the opposite, SCH 58261 significantly decreased apoptotic nuclei in the outer nuclear layer and GFAP immunoreactive area of the retinas. It also significantly decreased GFAP and activated caspase-3 levels as measured by WB and preserved retinal function, as treated eyes showed significantly greater amplitudes of a- and b-waves and oscillatory potentials. qRT-PCR revealed that SCH 58261 significantly decreased the expression of tumor necrosis factor-α. These results show that the blockade of the A2A receptor before the start of the pathogenic process is neuroprotective, as it prevents light-induced retinal damage. The use of A2A receptor antagonists deserves to be evaluated in retinal degenerative diseases.

Continuous illumination induces the degeneration of photoreceptors. This animal model of light-induced retinal degeneration resembles many characteristics of human degenerative diseases of the outer retina, such as age-related macular degeneration. This work aimed to evaluate the potential neuroprotective effect of the modulation of adenosine A2A receptor in the model of light-induced retinal degeneration. Sprague-Dawley rats were intravitreally injected in the right eye with either CGS 21680, an adenosine A2A receptor agonist, or SCH 58261, an adenosine A2A receptor antagonist. Contralateral eyes were injected with respective vehicles as control. Then, rats were subjected to continuous illumination (12,000 lux) for 24 h. Retinas were processed by glial fibrillary acidic protein (GFAP) immunohistochemistry, terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) technique, Western blotting (WB), and quantitative reverse transcriptionpolymerase chain reaction (qRT-PCR). Another group of rats was subjected to functional studies by electroretinography. Animals treated with CGS21680 showed a significant increase of apoptotic nuclei in the outer nuclear layer and a significant increase of GFAP immunoreactive area of the retinas but did not alter WB nor electroretinography results. qRT-PCR showed that CGS 21680 significantly increased the expression of interleukin-1β. On the opposite, SCH 58261 significantly decreased apoptotic nuclei in the outer nuclear layer and GFAP immunoreactive area of the retinas. It also significantly decreased GFAP and activated caspase-3 levels as measured by WB and preserved retinal function, as treated eyes showed significantly greater amplitudes of a-and b-waves and oscillatory potentials. qRT-PCR revealed that SCH 58261 significantly decreased the expression of tumor necrosis factor-α. These results show that the blockade of the A2A receptor before the start of the pathogenic process is neuroprotective, as it prevents lightinduced retinal damage. The use of A2A receptor antagonists deserves to be evaluated in retinal degenerative diseases.

INTRODUCTION
In recent years, the modulation of adenosine receptors has become a possible neuroprotective strategy to treat a wide range of insults and degenerative diseases of the central nervous system (CNS) (Stone et al., 2009). It has been reported that A1 receptor agonists are neuroprotective in animal models of inflammatory, hypoxic, epileptic, and degenerative diseases of the CNS (Boeck et al., 2005;Rosim et al., 2011;Gori and Girardi, 2013), whereas the use of A2A receptor agonists and/or antagonists has been useful against the neurotoxicity of 6-hydroxydopamine (Nobre et al., 2010), spinal cord injury (Paterniti et al., 2011), convulsions induced by pilocarpine (Rosim et al., 2011), memory dysfunction (Kaster et al., 2015), and degenerative diseases such as Alzheimer disease ) and Parkinson's disease (Gyoneva et al., 2014;Jenner, 2014). A broader review of the neuroprotective role of A2A receptor modulation may be found in Cunha (2016).
The release of adenosine is an important component of the response to ischemic/hypoxic insults of the retina Li et al., 1999), probably through the production of hyperemia that protects neurons from glutamate toxicity (Ostwald et al., 1997). The neuroprotective role of adenosine after ischemic retinal injury could be mediated by A1 (A1R) and/or A2 receptors (Housley et al., 2009).
Adenosine A1 and A2A receptors (A2AR) crosstalk with interleukin (IL)-6 and regulate the production of brain-derived neurotrophic factors (Perígolo-Vicente et al., 2014). The A2AR has been described as a transactivator of many other signaling families, including cannabinoids and neurotrophins (Tebano et al., 2010). The effect of adenosine receptors on other neurotrophins as glial cell line-derived neurotrophic factor and vascular endothelial growth factor seems to be responsible for the neuroprotective role of adenosine (Leibovich et al., 2002;Gomes et al., 2009). In the CNS and in the retina, it has been proposed that A2AR also plays a role in the microglial response to neuronal degenerative diseases (Santiago et al., 2014). The neuroprotective role of adenosine has been reviewed in different models of retinal degeneration, including ischemic retinopathy and diabetes (Dos Santos Rodrigues et al., 2015). Using a model of light-induced retinal degeneration (LIRD), which resembles many of the characteristics of age-related macular degeneration and retinitis pigmentosa (RP), we demonstrated that cyclopentyladenosine, an A1R agonist, protects the photoreceptors from light-induced damage, whereas dipropylcyclopentylxanthine, an A1R antagonist, has the opposite action (Soliño et al., 2018). Besides, the neuroprotective role of A2AR antagonists, such as KW6002 and SCH58261, has been demonstrated against the retinal damage induced by ischemia-reperfusion (Madeira et al., 2016a;Boia et al., 2017). Therefore, the present study aimed to investigate if the modulation of A2AR before the exposure to continuous illumination (CI) is able to prevent retinal damage in the model of LIRD as well.

Light-Induced Retinal Degeneration Procedure
One hour after intravitreal injections, rats were continuously illuminated for 24 h at 12,000 lux as previously described (Soliño et al., 2018). Groups of 3-5 rats were simultaneously placed in an open white acrylic box of 60 cm × 60 cm x 60 cm with 12 halogen lamps (12 V, 50 W each) located on top. Lighting level and temperature (21°C) were monitored. Then, the retinas were obtained around 2 p.m. and processed for glial fibrillary acidic protein (GFAP) immunohistochemistry (IHC) (n = 4), terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) technique (n = 4), Western blotting (WB) (n = 5), or quantitative reverse transcription-polymerase chain reaction (qRT-PCR) (n = 5). In every case, the numbers indicate the number of rats per drug treatment. A separate group of five rats per drug treatment was used for electroretinography (ERG) studies. After performing a basal ERG, rats were subjected to the intravitreal injections of drugs and exposed to CI. A week later, a second ERG was performed. All animals received food and water ad libitum. Eyes containing the retinas were fixed by immersion in a 4% paraformaldehyde solution for 24 h. After cryoprotection in a 30% sucrose solution, the eyes were embedded in gelatine, blocks were frozen, and sections were obtained using a Lauda Leitz cryostat. Sections (thickness: 20 µm) were mounted on gelatine-coated glass slides and processed by IHC or TUNEL techniques.

Terminal Deoxynucleotidyl Transferase dUTP Nick End Labeling Assay
Cryostat sections were processed using the ApopTag Peroxidase In Situ kit (Chemicon Int, CA, United States, S701), following the instructions described in Soliño et al. (2018). Briefly, sections were incubated with terminal deoxynucleotidyl transferase (Chemicon Int, CA, United States, Cat 90,418) (1 h at 37°C) followed by the anti-digoxigenin conjugate (Chemicon Int, CA, United States, Cat 90,420) (30 min at RT).
The reaction was developed using the diaminobenzidine/nickel intensification procedure, and sections were counterstained with eosin.

Image Analysis of Terminal Deoxynucleotidyl Transferase dUTP Nick End Labeling and Glial Fibrillary Acidic Protein Immunostained Sections
Six retinal sections of both eyes from each experimental group were analyzed (CGS21680, n = 4; SCH58261, n = 4). Anatomically matched areas of retina among animals were selected, and images were taken using a Zeiss Axiophot microscope attached to a video camera (Olympus Q5) under the same light conditions (Soliño et al., 2018).
The following parameters were measured, blind to treatment, on 8-bit images, using the Fiji software (NIH, Research Services Branch, National Institutes of Mental Health, Bethesda, MD, United States): GFAP-positive area: Images of drug-treated and control retinas were randomly selected. The immunoreactive area of the whole section was thresholded. The region of interest (ROI) was the retinal surface between the two limiting membranes where Müller cells extend their processes. The GFAP-positive area was calculated as the percentage of the ROI immunostained by GFAP.
TUNEL positive nuclei/1,000 µm 2 : Images of drug-treated and control retinas were randomly selected and thresholded. As for ROI, frames of 1,000 µm 2 were randomly determined on the outer nuclear layer (ONL) of treated and control retinas. The analyzed particles function of Fiji was used (Grishagin, 2015), and the TUNEL positive nuclei/1,000 µm 2 ratio was then obtained for each ROI.

Electroretinography
After overnight dark adaptation, rats were anesthesized under dim red illumination with ketamine and xylazine, as was mentioned earlier. An ophthalmic solution containing 5% phenylephrine hydrochloride and 0.5% tropicamide (Fotorretin, Laboratorios Poen, Argentina) was used to dilate the pupils. Recordings were made from both eyes simultaneously (Soliño et al., 2018). Scotopic ERG: 20 responses to flashes of white light (1 ms, 1 Hz) from a photic stimulator (light-emitting diodes) set at maximum brightness were recorded with an Akonic BIO-PC electroretinograph (Argentina). The registered response was amplified and filtered (1.5-Hz low-pass filter, 500-Hz highpass filter, notch activated).
Oscillatory potentials (OP): The same photic stimulator was used with filters of high (300 Hz) and low (100 Hz) frequencies.
The amplitudes of the OPs were estimated following described methodology (Severns et al., 1994).
The a-and b-waves and OP were measured 40 times, and the values from each eye were averaged. The resultant mean values were used to obtain the group means of a-and b-waves and OP amplitudes ± standard deviation.

RNA Isolation and Quantitative Reverse Transcription-Polymerase Chain Reaction
The retinas were processed as detailed in Soliño et al. (2018). Briefly, tissues were homogenized with TRIzol (Invitrogen, Madrid, Spain), and RNA was isolated with RNeasy Mini kit (Qiagen, Germantown, MD, United States). Three micrograms of total RNA were treated with 0.5-µl DNAseI (Invitrogen) and reverse-transcribed into first-strand copy DNA using random primers and the SuperScript III kit (Invitrogen). Reverse transcriptase was omitted in control reactions. The resulting copy DNA was mixed with SYBR Green PCR master mix (Invitrogen) for qRT-PCR using 0.3 µM forward and reverse oligonucleotide primers (see Table 1). Quantitative measures  were performed using a 7,300 Real-Time PCR System (Applied Biosystems, Carlsbad, CA, USA). Cycling conditions were an initial denaturation at 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. At the end, a dissociation curve was implemented from 60 to 95°C to validate amplicon specificity. Gene expression was calculated using relative quantification by interpolation into a standard curve. All values were divided by the expression of the housekeeping gene 18S.

Statistical Analysis
D'Agostino, KS, Shapiro-Wilk, and F tests confirmed that the data of image analysis of GFAP, IHC, and TUNEL studies of CGS21680-treated rats (n = 4) and SCH 58261-treated rats (n = 4) display a Gaussian distribution. Then, all the data of this study [GFAP-IHC (n = 4), TUNEL (n = 4), WB (n = 5), qRT-PCR (n = 5), and ERG (n = 5)] were analyzed using unpaired Student's t-test (GraphPad Software, San Diego, CA, United States). In every case, n is the number of animals per drug treatment. Values are expressed as mean ± standard deviation. Differences were considered significant when p < 0.05.
The sample size was calculated based on data published by Soliño et al. (2018). In that study, the number of apoptotic cells, as quantified by TUNEL analysis, was 4.25 positive nuclei/1,000 µm 2 in animals subjected to LIRD and was reduced to 1.45 when subjects were treated with N6-cyclopentyladenosine, with a standard deviation of 0.74. Free software (http://biomath.info/ power/ttest.htm) was used to calculate the sample size. Power was set as 80% for an alpha of 5%, resulting in less than six animals per group to reach a significant improvement of the variable with an unpaired t-test.

Effects on Photoreceptor Apoptosis and Gliosis
In the sections stained with the TUNEL technique, a greater density of positive nuclei was observed in the ONL of the retina of CGS21680-treated eyes (9.78 ± 2.752 vs. 5.974 ± 0.3612, p < 0.05, n = 4) ( Figures 1A,B,E and Supplementary Figure S3).

Apoptosis and Glial Reactivity After Administration of CGS21680
No significant differences were seen in protein expression of GFAP and activated caspase-3 between the retinas of eyes treated with CGS21680 and CTL (Figure 2 and Supplementary Figure S4).

Electroretinography After Administration of CGS21680
The eyes treated with CGS21680 did not show significant differences in any of the studied parameters between treated eyes 7 days post-illumination and control eyes 7 days postillumination: a-wave (p = 0.27), b-wave (p = 0.19), and oscillatory potentials (p = 0.22, Table 2 and Figure 3).

Effects of CGS21680 on Gene Expression (Quantitative Reverse Transcription-Polymerase Chain Reaction)
To investigate the mechanism of action of the A2AR agonist, CGS21680, we studied the expression of genes involved in cell damage and inflammation. qRT-PCRs of the retinas were performed after the treatment with CGS21680, followed by 24 h of CI. Cytokine IL-1β expression increased significantly (1.278 ± 0.9059 vs. 0.5573 ± 0.3511; p < 0.05) ( Figure 4A), whereas inducible nitric oxide synthase (iNOS) messenger RNA (mRNA) and TNF-α mRNA did not change significantly ( Figures 4B,D). The astroglial marker, GFAP mRNA, did not change after treatment with CGS21680 ( Figure 4C).

Effects on Photoreceptor Apoptosis and Gliosis
TUNEL staining showed that the number of positive nuclei decreased in SCH58261-treated eyes, indicating a lower number of apoptotic photoreceptors (8.354 ± 1.701 vs. 4.247 ± 2.056, p < 0.05) ( Figures 5A,B,E and Supplementary Figure S3).

Effects of SCH58261 on Apoptosis and Glial Reactivity by Western Blot
Changes in the levels of activated caspase-3 and GFAP are in agreement with previous TUNEL and IHC results. A significant decrease of GFAP levels in the treated eyes was found (0.5477 ± 0.09308 vs. 1 ± 0.06348, p < 0.01) (Figures 6A,C and Supplementary Figure S4). Also, a statistically significant decrease in activated caspase-3 levels in the retinas of SCH58261-treated eyes was confirmed (0.7853 ± 0.1611 vs. 1 ± 0.030220, p < 0.05) ( Figures 6B,D and Supplementary Figure S4). Effects of SCH58261 on Retinal Function Determined by Electroretinography SCH58261-treated eyes showed a greater response of the photoreceptors as larger a-wave was recorded compared with CTL eyes (p < 0.05) (Figures 7A,B; Table 3). Also, the function of the inner retina was protected, as the amplitude of the b-wave and OP was significantly larger in SCH58261-treated eyes compared with CTL eyes (p < 0.05 in both cases) (Figures 7C-E; Table 3).

Effects of SCH58261 on Gene Expression (Quantitative Reverse Transcription-Polymerase Chain Reaction)
Similarly, to investigate the neuroprotective mechanism of the A2AR antagonist SCH58261, we studied the expression of genes involved in cell damage and inflammation. qRT-PCRs of the retinas were performed after the treatment with SCH58261, followed by 24 h of CI. TNF-α decreased significantly in SCH58261-treated retinas (1.089 ± 0.1431 vs. 1.271 ± 0.2668,

DISCUSSION
Results presented herein indicated that an A2AR agonist (CGS21680) exacerbated the damage induced by light exposure to the retina. CGS21680-treated eyes presented higher densities of apoptotic nuclei in ONL and higher glial reactivity, evidenced by an increase of the GFAP immunoreactive area. However, the WB study did not confirm these data, and functional studies using ERG showed no differences between treated eyes and their CTL. This phenomenon could be explained because CI produced such functional damage that it cannot be worsened by an A2AR agonist. Also, it may be speculated that the techniques used to detect apoptosis (TUNEL) and GFAP immunohistochemistry have greater sensitivity to detect damage at the cellular level than the WB of the full retinal tissue. Alternatively, it must be mentioned that there are strong pharmacological arguments in brain tissue questioning the selectivity of CGS21680, as it could also act as an agonist for the A1 receptor (Zhang et al., 1994;Cunha et al., 1996;Cunha et al., 1999). This could also be the reason underlying the lack of effect of this drug.
On the opposite, we found that a lower A2AR activity due to an antagonist intravitreal injection was protective to the retina, as SCH58261-treated retinas showed smaller amounts of apoptotic nuclei in ONL and lower glial reactivity. Both data were confirmed by WB because SCH58261-treated retinas had lower levels of activated caspase-3 and GFAP. Finally, SCH 58261 also protected the retinal function, as all ERG parameters were preserved.
It was demonstrated that A2AR inhibitors provide important protective mechanisms in the CNS, as low concentrations of adenosine activate A1R and inhibit the release of excitatory amino acids, but higher concentrations of adenosine activate A2AR and block A1R through a receptor-receptor allosteric trans-inhibition (Ciruela et al., 2006).
The role of adenosine in the inflammatory response to retinal injury is well known, and it was reported that SCH58261 protects from photoreceptor loss because it prevents the upregulation of proinflammatory mediators and the alterations of the complement system in microglial cells (Madeira et al., 2018). Also, KW6002, another A2AR antagonist, reduced the inflammatory microglial response and protected the retina from ischemic injury and reperfusion (Boia et al., 2017). Caffeine, an unspecific A2AR antagonist, was neuroprotective, as it lowered intraocular pressure and reduced the activation of microglia and the inflammatory cytokines IL-1β and TNF-α in a mouse glaucoma model (Madeira et al., 2016b). To investigate the neuroprotective mechanism of the A2AR antagonist, we studied the expression of genes involved in inflammation and cell damage by qRT-PCR, and we found that SCH58261 lowered the levels of TNF-α expression, supporting the idea that it reduces the upregulation of inflammatory microglial mediators. This lower inflammatory milieu could favor the survival of the photoreceptors and could allow the conservation of the function.
The observed neuroprotective effect of A2AR antagonism in LIRD is aligned with the neuroprotective effect of chronic caffeine intake (Cunha and Agostinho, 2010) and with caffeine reduction of apoptosis in oxygen-induced retinopathy (Zhang et al., 2017).
On the contrary, using two different models of perinatal brain injury, CGS21680 produced a partial increase of some microglial cytokines such as IL-1β and TNF-α, as well as an increase of iNOS (Colella et al., 2018). In agreement with this, in our model of CI, CGS21680 produced a similar phenomenon, as we also detected an increase of the cytokine IL-1β. Surprisingly, no significant effects were observed in iNOS and TNF-α mRNA expression. Herein, we detected lower levels of Müller glia activation in SCH58261-treated eyes, suggesting that modulation by A2AR may be involved in the inflammatory reaction. This finding is in line with the observation that the injection of antagonist SCH 442416 reverses the changes in the expression of channels and transporters in Müller cells responsible for maintaining retinal homeostasis (Yang et al., 2015).
Although the role of A2AR antagonist on microglial inflammation seems to be the main mechanism involved in the protection of the retina, we postulate that a direct neuroprotective effect on the photoreceptors themselves cannot be ruled out, as A2 receptors are localized in rabbit and mouse outer retinas (Blazynski, 1990;Blazynski and Perez, 1991), and A2AR was also found in the photoreceptors of different species (McIntosh and Blazynski, 1994;Rey and Burnside, 1999;Stella et al., 2002;Li et al., 2014). Further studies are needed to confirm this hypothesis.
Finally, we studied the changes in the expression of the A2A receptor. CGS 21680 produced no significant changes (Supplementary Figure S1), but SCH 58261 produced a significant reduction in the expression of the receptor (Supplementary Figure S2). This would increase the antagonistic effects of SCH 58261 by diminishing the available receptors and may also explain the protective effect of the drug.
In summary, this study shows that the blockade of A2A receptors before exposure to continuous light prevents retina damage and preserves retinal function by lowering inflammation, glial reactivity, and apoptosis in ONL. These results allow us to postulate that the modulation of A2AR activity may be a strategy that deserves to be evaluated in degenerative pathologies of the retina.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material; further inquiries can be directed to the corresponding author.   2018/20020170100493BA). JL-C was the director of the lab in BA, where animals were illuminated, and IHC, WB, and ERGs were carried out. This research was funded in part by a grant (PI19/ 01805) from the Instituto de Salud Carlos III, co-funded by European Regional Development Fund (ERDF) "A way to build Europe". IML was supported by Miguel Servet contracts (CP15/00198 and CPII20/00029) from the Instituto de Salud Carlos III, co-funded by European Social fund (ESF) "Investing in your future" and by the Fundación Rioja Salud (FRS) given to AM. AM was the director of the Spanish lab where qRT-PCRs were carried out.