Subtle thinning of retinal layers without overt vascular and inflammatory alterations in a rat model of prediabetes

Animals

Male Wistar rats aged 16 weeks (Charles River Laboratories, Lyon, France) were housed in certified facilities, in a temperature- and humidity-controlled environment under a 12 h:12 h light-dark cycle. Animals were provided with a standard rodent diet ad libitum and were randomly divided into two groups: rats that received 35% sucrose in drinking water for 9 weeks (high sucrose, HSu), as previously described [19] and the age-matched control group (water).

Evaluation of metabolic and biochemical parameters was performed as previously described [19,22]. All procedures involving animals were performed in agreement with the EU Directive 2010/63/EU for animal experiments, and in accordance with the Association for Research in Vision and Ophthalmology (ARVO) statement for the use of animals in ophthalmic and vision research.

Preparation of total retinal extracts and western blotting analysis

The preparation of the total retinal extracts and western blotting analysis was performed as previously described [12]. Primary antibodies used are listed in Table 1. After protein detection, the membranes were reprobed for β-actin immunoreactivity (1:5,000; Sigma, St. Louis, MO) as a loading control.

Terminal Transferase dUTP Nick End Labeling (TUNEL) staining.

Retinal cells undergoing apoptosis were identified by TUNEL assay in retinal cryosections using the DeadEnd Fluorimetric TUNEL system (Promega Corporation, WI) as previously described [24]. Cells were stained with DAPI to label the nuclei. The images were acquired with a fluorescence microscope (Axio HXP 120, Zeiss, Oberkochen, Germany).

Assessment of BRB permeability using Evans blue dye

Retinal blood vessel permeability was evaluated using Evans blue dye, following the procedure previously described [25,26]. Evans blue binds tightly to serum albumin, which crosses the BRB when its permeability is affected. Evans blue (100 mg/ml; Sigma) was dissolved in PBS (in mM: 137 NaCl, 2.7 KCl, 10 Na₂HPO₄, 1.8 KH₂PO₄; pH 7.4), sonicated for 20 min, and filtered (pore size, 0.2 μm).

HSu-treated rats and age-matched controls were anesthetized with an intraperitoneal injection of ketamine (50 mg/kg bodyweight; Imalgene 1000, Merial, Lyon, France) and xylazine (10 mg/kg bodyweight; Rompum®, Bayer, Leverkusen, Germany), and Evans blue dye (100 mg/kg bodyweight) was injected using a 30-gauge needle through the tail vein, previously warmed up using a heating lamp. Additional anesthesia (ketamine, 50 mg/kg bodyweight) was provided during the procedure, if needed. Animals were kept warm during the procedure. After the dye circulated for 120 min, the chest cavity was opened, and the animals were perfused via the left ventricle, with 1% paraformaldehyde (PFA) in 50 mM citrate buffer (pH 4.2, 37 °C; about 56 ml) to clear the dye from the vessels. Immediately following perfusion, the eyes were enucleated, and the retinas were carefully dissected under an operating microscope and rinsed in PBS. Retinas were thoroughly dried in a Speed-Vac (5 h, with heating), and the dry weight of each retina was measured. Albumin leakage into the retina was assessed via measurement of the extravasated Evans blue dye. This dye was extracted by incubating each retina in 110 μl formamide for 18 h at 70 °C with gentle shaking. The extract was ultracentrifuged (TL-100, Beckman, CA) at a speed of 200,000 ×g for 45 min, at 4 °C. The absorbance of supernatant (50 μl) was measured in a spectrophotometer (Synergy HT, BioTek, Winooski, VT) at 620 nm absorbance (maximum) and 740 nm absorbance (minimum). The concentration of the dye in the samples was calculated from a standard curve (100–5,000 ng/ml) of Evans blue dye in formamide and normalized to the dry retinal weight.

Optical coherence tomography

The structural changes in the retina were evaluated in vivo with SD-OCT (Phoenix Micron IV, Phoenix Research Labs, Pleasanton, CA). The animals were anesthetized with an intraperitoneal injection of ketamine (75 mg/kg bodyweight) and xylazine (10 mg/kg bodyweight). The data were collected from both eyes after corneal anesthesia with topical 4 mg/ml oxybuprocaine hydrochloride (Anestocil®, Laboratório Edol, Carnaxide, Portugal) and pupil dilation with topical 10 mg/ml tropicamide (Tropicil®, Laboratório Edol). As the OCT technique includes the contact of a lens with the animal eye, we used topical oxybuprocaine to obtain ocular analgesia, in addition to the animal anesthesia. This procedure was adopted to minimize pain and improve animal welfare. To our knowledge, this procedure does not affect the structural parameters assessed by OCT. The cornea was kept hydrated and optically cleared with hydroxypropyl methylcellulose (Methocel™ 2%, Dávi II – Farmacêutica S.A., Barcarena, Portugal) during the whole procedure. The SD-OCT system used was able to capture 10,000–20,000 A scans per second with an axial resolution of 2 µm and a transverse resolution of 4 µm. For each eye, 13 tomographic images, obtained from line scans, plus the respective fundus of the eye, were acquired. The first image was acquired centered on the optic nerve head, followed by six scans above and six scans beneath that reference. In this way, in total, a 1,700 μm high region in the retina was scanned.

Tomographic images were analyzed using the segmentation software InSight (Phoenix Research Labs). Four layers in the retina were considered: (1) the ganglion cell layer (GCL), the retinal nerve fiber layer (RNFL), and the inner plexiform layer (IPL), between the inner limiting membrane (ILM) and the interface IPL-inner nuclear layer (INL); (2) the INL, between the interface of the IPL and the INL and the outer plexiform layer (OPL); (3) the outer nuclear layer (ONL), between the OPL and the outer limiting membrane (OLM); and (4) the inner segments (IS) and the outer segments (OS), between the OLM and the RPE.

Preparation of retinal cryosections and immunohistochemistry

The preparation of the retinal cryosections (16 µm) and immunohistochemistry were performed as previously described [12]. Primary antibodies used in immunohistochemistry are listed in Table 1.

Image acquisition and analysis

Sections were observed with a fluorescence microscope (Axio HXP 120, Zeiss, Oberkochen, Germany), using an LD Plan-Neofluar 40x/ 0.6 Korr Ph2 M27 objective. For the analysis of neuronal, macroglial, and microglial cell markers, slides containing retinal slices from the control and HSu groups were blind coded and captured under the same conditions (four retinal sections per animal were used for quantification). In each image, the number of cells stained with 4',6-diamidino-2-phenylindole (DAPI; ganglion/displaced amacrine cells – DAPI-positive cells at the GCL), the number of Brn3a (RGCs) immunoreactive cells, or the number of cells immunoreactive to Iba1 (Iba1-positive cells) was counted, and the results were expressed per millimeters of the retina. Representative images were acquired using a laser scanning confocal microscope LSM 710 META (Zeiss).

RNA extraction and reverse transcription

Total RNA was isolated from the rat retinas using TRIzol reagent (Invitrogen, Life Technologies, Carlsbad, CA). RNA samples were dissolved in 16 µl of Milli-Q water, and the concentration and purity of the total RNA were determined using NanoDrop ND1000 (Thermo Scientific, Waltham, MA). Resulting RNA samples were treated with DNase I (DNase I; amplification grade, Invitrogen, Life Technologies) to eliminate possible contamination with genomic DNA. Total RNA (1 µg) was reversed transcribed according to the instructions provided by the manufacturer (NZYTech, Lisbon, Portugal). The resultant cDNA was treated with RNase-H for 20 min at 37 °C. For all cDNA samples, a 1:2 dilution was prepared, and the samples were stored at −20 °C until quantitative PCR (qPCR) analysis. Genomic DNA contamination was evaluated with conventional PCR for β-actin using intron-spanning primers (Table 2), as previously described [24].

Real-time quantitative PCR

qPCR was performed using StepOnePlus (Applied Biosystems, Foster City, CA) based on the real-time monitoring of fluorescent SYBR Green. The PCR conditions were as follows: iTaq™ Universal SYBR® Green Supermix (Bio-Rad, Hercules, CA), 200 nM primers (Table 2), and 2 µl of 1:2 dilution cDNA, in a total volume of 20 µl. Cycling conditions were a melting step at 95 °C for 15 s, annealing-elongation at 60 °C for 45 s, and extension at 72 °C for 30 s, with 40 cycles. A dissociation curve at the end of the PCR run was performed by ramping the temperature of the sample from 60 °C to 95 °C, while continuously collecting fluorescence data. Ct values were converted to “Relative quantification” using the 2^(-ΔΔCt) method previously described [27]. Hprt, Ywhaz, and rhodopsin were evaluated as housekeeping genes using NormFinder Add In for Microsoft Excel [28]. Because Hprt was the most stable gene in all conditions, it was used as the reference gene.

Enzyme-linked immunosorbent assay

Quantification of the interleukin-1β beta (IL-1β) and tumor necrosis factor (TNF) protein levels was performed with enzyme-linked immunosorbent assay (ELISA), as previously described [24]. The protein concentration of each sample was determined by the bicinchoninic acid protein assay according to the manufacturer’s instructions (Pierce Biotechnology, Waltham, MA). The cytokine concentration of each sample was normalized to the total protein concentration.

Statistical analysis

The results are presented as mean ± standard error of the mean (SEM). Statistical analysis was performed in Prism 5.03 Software for Windows (GraphPad Software, Inc.). The normality of the data was assessed with Shapiro-Wilk and Kolmogorov–Smirnov normality tests. Accordingly, data were analyzed with a non-parametric Mann–Whitney test. Differences with a p value of less than 0.05 were considerably statistically significant.

Evaluation of bodyweight and metabolic and biochemical parameters

The intake of 35% sucrose for 9 weeks did not affect the bodyweight of the HSu-treated rats when compared to that of the control animals (Table 3). Regarding the glycemic profile, there were no differences in plasma glucose levels between the HSu-treated rats in fasted states compared to the control animals. However, a statistically significant increase in plasma glucose levels in the postprandial state in the HSu rats was found. In addition, high sucrose consumption resulted in reduced glucose tolerance, as the recovery of glucose levels to basal levels was slower compared to the control animals, confirmed by the area under the curve (AUC) of the GTT (Table 3). Simultaneously, there was an impairment in insulin response after 9 weeks of sucrose intake, as confirmed by the insulin tolerance test (ITT). In this test, after the insulin injection, the reduction of blood glucose levels was slower in the HSu rats, revealing a reduced sensitivity to insulin. Moreover, the HOMA-IR (Homeostatic Model Assessment for Insulin Resistance) index, an additional measure of insulin resistance, was statistically significantly higher in the HSu rats compared with the controls.

Concerning the lipid profile, no changes were found in the total cholesterol levels between the two groups of animals, but the HSu-treated animals presented higher levels of triglycerides. Overall, these metabolic and biochemical data (Table 3) are in accordance with previous data obtained in the same prediabetic animal model [19,22]: IGT, fasting normoglycemia, hyperinsulinemia, insulin resistance, and hypertriglyceridemia.

Tight junction proteins and BRB permeability are not affected in the retinas of HSu-treated rats

Increased BRB permeability in diabetic animals is associated with changes in the content and distribution of tight junction proteins in retinal vessels [29,30]. Claudins, occludin, and zonula occludens-1 (ZO-1) are crucial determinants of BRB permeability [29]. BRB permeability was assessed using the Evans blue dye assay. Non-statistically significant changes were found in BRB permeability. Evans blue accumulation in the retina was not statistically significantly different between the two groups of animals (9.26±1.19 µg Evans blue/g retina versus 10.51±1.98 µg Evans blue/g retina, for HSu-treated and control animals, respectively; Figure 1A). The content of claudin-5, occludin, and ZO-1 in the retina was evaluated with western blotting. No statistically significant alterations were found in the HSu-treated animals compared with the controls (Figure 1B). These results indicate that at this early stage of insulin resistance the integrity of the BRB is not affected.

HSu diet induces retinal thinning

Diabetes is also characterized by structural changes in the neural retina. However, to our knowledge, it is unknown whether these alterations are already present in a prediabetic state. SD-OCT was used to analyze, noninvasively, the thickness of retinal layers in vivo (Figure 2A). After 9 weeks of the HSu diet, there was a subtle, but statistically significant, thinning of several retinal layers, particularly the RNFL, GCL, and IPL and the INL (Figure 2B).

It has been demonstrated that neuronal cells of the retina, namely, RGCs, are affected by diabetes [31-33]. In the HSu model, despite the thinning of the RNFL, GCL, and IPL, no changes in the number of cells (DAPI-positive cells) were observed in the GCL (Figure 2C). In addition, no apoptotic cells (terminal deoxynucleotidyl transferase (TdT) dUTP nick-End labeling (TUNEL)-positive cells) were detected in the retinas of the HSu-treated rats (Figure 2D). The impact of the HSu diet on RGCs was further studied with the evaluation of the Brn3a transcription factor that is expressed only in RGCs in the retina [34]. There was no alteration in the number of RGCs (Brn3a-positive cells), as well as in the Brn3a mRNA expression in HSu-treated animals compared to the controls (Figure 2E).

Undetected changes in neuronal and glial markers in the retinas of HSu-treated rats

Changes in the neural components of the retina have been described in experimental diabetes [3]. In the retinas of diabetic animals, several neuronal markers are altered [8,11,35]. However, because it is unknown whether these alterations are already present in a prediabetes state, the protein levels of neuronal marker TUJ-1 (Figure 3A) and synaptic marker synaptophysin (Figure 3B) were assessed with immunoblotting and immunohistochemistry. In the HSu model, no statistically significant changes were detected in the content and distribution of these neuronal markers (Figure 3A,B), as well as in the protein levels and distribution of the glial markers, vimentin (Figure 3C), and glial fibrillary acidic protein (GFAP; Figure 3D).

Undetected changes in proinflammatory mediators in the retina of HSu-treated rats

Increased levels of proinflammatory markers, such as inducible nitric oxide synthase isoform (iNOS) [29], TNF, and IL-1β [14,36-38], as well as activation of microglial cells [36,39], in the retina have been detected in diabetic animals and retinal cell cultures exposed to elevated glucose. The HSu diet did not statistically significantly change the transcript or protein levels of iNOS in the retina, compared with those of the age-matched controls (Figure 4A), as assessed with qPCR and western blotting, respectively. Similarly, no changes were detected in the mRNA expression and protein levels of IL-1β and TNF (Figure 4B), assessed with qPCR and ELISA, respectively.

The number of microglial cells remains unaltered in the retinas of HSu-treated rats

Microglial cell reactivity was assessed with immunohistochemistry, with retinal sections labeled for Iba1 (a general marker of microglial cells; red) and MHC-II (a marker of reactive microglia and macrophages; green). MHC-II immunoreactive cells were not detected in the retina (MHC-II-positive cells were detected only at the choroid and sclera levels; data not shown) in the control and HSu-treated animals (Figure 5A). Translocator protein (TSPO) has been used as a marker of microglia reactivity [40]. No changes were detected in the TSPO transcript levels in the retinas of the HSu-treated animals compared with those of the controls (Figure 5B). The number of Iba1-positive cells in the retinas of the HSu-treated animals was also not statistically significantly increased when compared with that of the controls. Nevertheless, there was a tendency (p=0.056) for an increase in the number of microglia in HSu retinas at the IPL and the GCL (Figure 5C).