Albino homozygous P23H line-3 rats were obtained from Dr. Matthew LaVail (UCSF School of Medicine). Normal Sprague-Dawley (SD) rats, provided by Harlan Laboratories (Barcelona, Spain), were used as age-matched controls. All animals were bred in a colony at the University of Alicante and maintained under controlled conditions of humidity (60%), temperature (23°C), and photoperiod (LD 12:12). Rats were fed with dry food and water ad libitum. All the procedures were made following the specifications approved by the Ethics Committee of the University of Alicante (UA-2013-07-22). Animal handling was carried out in agreement with current regulations on the care and use of laboratory animals (EU Directive 2010/63/EU, NIH-guidelines, and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research).

Tauroursodeoxicholic acid was obtained from Calbiochem (Calbiochem, Merck Millipore, Darmstadt, Germany) and dissolved in phosphate-buffered saline (PBS, pH 7.4) by using an ultrasonic bath to avoid bubble formation. Treatment was performed as previously described by our group (Fernández-Sánchez et al., 2011; Noailles et al., 2014). Briefly, P23H line-3 animals were divided into two groups: TUDCA-treated and vehicle-treated animals. The TUDCA-treated group (n = 12) was administered with TUDCA (500 mg/Kg, i.p.) once a week from postnatal day (P)21 to P120. Vehicle-treated P23H rats (n = 15) were administered with a vehicle at the same time points. SD animals (n = 11) were administered with the vehicle. At the end of the treatment, all animals were sacrificed, and tissue was processed.

Animals were sacrificed between 10:00 a.m. and 12:00 p.m. by administering a lethal dose of sodium pentobarbital (200 mg/kg). Prior to eye enucleation, a suture was sewn in the dorsal limbus to label axis orientation. After enucleation, an incision was made at the sclerocorneal limbus to allow better penetration of the fixative solution. The eyecups were immersed in paraformaldehyde 4% (w/v) and PBS for 1 h. After that, eyes were washed three times in PBS and cryoprotected with increasing concentration of sucrose solutions: 15% (1 h), 20% (1 h), and 30% (overnight). The posterior pole of the eye was isolated by removing the cornea, lens, and vitreous body, and processed for nasal-temporal retinal cross-sections or wholemounts. In most cases, whenever possible, one retina from each animal was prepared as a whole mount and the other was used for sections. For whole mount preparations, retinas were carefully isolated from the rest of the eyecup by using a fine brush. Four partial incisions were made along the retinal axis to divide the retinas into superior, inferior, nasal, and temporal quadrants, and to better flatten them. Retinal wholemounts were finally mounted with the ganglion-cell-side up. For sections, the eyecups were embedded in optical coherence tomography (OCT) and frozen in liquid nitrogen. Sixteen-micrometer-thick cross-sections were obtained using a Leica CM 1900 cryostat (Leica Microsystems, Wetzlar, Germany). Sections were mounted on slides (Superfrost Plus; Menzel GmbH and Co., KG, Braunschweig, Germany) and air-dried. Prior to further processing, the slides were washed in PBS and treated for 1 h using blocking solution (10% normal donkey serum in PBS plus 0.5% TritonX-100).

Retinas from all groups were processed in parallel to objectively compare them. The cell-specificity of all antibodies and lectins used in this word has been well characterized by our group and others and has been used extensively in other studies. Staining of retinal wholemounts and sections was performed following the standard protocols previously described (Fernández-Sánchez et al., 2015a). Briefly, retinal tissue was washed three times in PBS and then incubated in 0.02% sodium borohydride (163314; Panreac, Barcelona, Spain) in PBS (5 min, RT) to increase permeability to the antibodies. Sections and whole mount retinas were incubated with a combination of primary antibodies at an optimal concentration (Table 1) at 4°C overnight or for 3 days, respectively. Then, sections or whole mount retinas were washed three times in PBS and incubated for 1 h or overnight, respectively, with the corresponding Alexa fluor-conjugated antibody or secondary antibody and Griffonia simplicifolia IB₄ (1:100, I21411; Isolectin GS-IB₄, Invitrogen, Carlsbad, CA, United States) at the optimal concentration. The secondary antibodies used in this work were Alexa Fluor 488-conjugated anti-rabbit IgG, Alexa Fluor 555-conjugated anti-mouse IgG, and Alexa Fluor 555-conjugated anti-rabbit IgG made in donkey, and all of them were obtained from Molecular Probes (Eugene, OR, United States). After washing, the preparations were mounted in Citifluor (Citifluor Ltd, London, United Kingdom) and coverslipped for further observation under a laser-scanning confocal microscopy Leica TCS SP2 (Leica Microsystems, Wetzlar, Germany).

The retinal vascular network was visualized using histochemistry of reduced nicotinamide adenine dinucleotide phosphate diaphorase (NADPH-d) as previously described by our group (Fernández-Sánchez et al., 2018). In brief, whole mount retinas were incubated in PBS containing 1 mg/ml NADPH (Sigma, St. Quentin Fallavier, France), 0.1 mg/ml nitroblue tetrazolium (NBT) (Sigma), and 1% (v/v) Triton X-100 for 2 h at 37°C. After washing three times in PBS, retinas were flat-mounted in Citifluor mounting medium (Citifluor Ltd.) with ganglion-cell-side up. Deep vascular plexus was drawn using a camera lucida coupled to a Leica DMR microscope (Leica Microsystems, Wetzlar, Germany). Drawings were digitalized and morphometric analysis was performed using ImageJ software (National Institutes of Health, Bethesda, MD, United States) and AngioTool software (0.6a version, National Cancer Institute, Bethesda, MD, United States) (Zudaire et al., 2011). Prior to analysis, Adobe Photoshop software (Adobe Systems, Inc., San Jose, CA, United States) was used to equally adjust brightness and contrast on the images. ImageJ software was used to measure the relative capillary density of the deep capillary plexus (DCP) with the “measure area” function, and the total number of capillary loops was studied using the “analyze particles” command. Angio-tool software (0.6a version, National Cancer Institute, Bethesda, MD, United States) was used to study in detail the vascular network complexity. Specifically, vessel junction density (which indicates the vascular branching per unit area), total vessel length, and mean lacunarity were analyzed. Lacunarity measurements describe the non-uniformity of the vascular pattern. It offers information about the size of lacunas or gaps present in the images (Gould et al., 2011; Zudaire et al., 2011). Thus, greater lacunarity means greater gaps and vascular degeneration. All data were normalized to the retinal area.

Astrocyte quantification was done as previously described by our group (Fernández-Sánchez et al., 2015a). Briefly, astrocytes were quantified in whole mount retinas immunolabelled against the glial fibrillary acidic protein (GFAP) and nuclear marker TO-PRO-3. Lectin GS-IB₄ was used to visualize retinal blood vessels. Quantifications were done in 12 representative areas measuring 0.227 mm² and homogeneously distributed in the superior, inferior, nasal, and temporal quadrants from retinal center to the periphery. Astrocyte cell bodies were manually quantified in each region in two confocal images, and averaged values were used for each area. Only GFAP-positive cells with a well-defined nucleus were included in the counting. At least three rats were used per data point.

Prism software (GraphPad Software; San Diego, CA, United States) was used to perform statistical analysis. One- or two-way ANOVA tests were applied to evaluate the effects of TUDCA on morphological parameters of the vascular network and astrocyte numbers. Normal distributions and homogeneity of variance were found for all the analyzed categories. p < 0.05 was considered statistically significant. When the level of significance was 0.05 or less, post-hoc pairwise comparisons using Tukey’s test were carried out. Data were plotted representing the mean ± standard error of the mean (Osipova et al., 2018).

Retinal vasculature in rodents is distributed in three laminar plexuses: the superficial capillary plexus (SCP) at the level of the ganglion cell layer (GCL), the DCP at the level of the outer plexiform layer (OPL), and the intermediate capillary plexus (ICP) which connects the other two, running at the edge between the inner nuclear (INL) and inner plexiform (IPL) layers. Figure 1 shows the state of photoreceptor cell death and vascular network degeneration at each of the retinal capillary plexuses in P120 rats from the different experimental groups. Retinal sections from SD rats showed 10–12 rows of photoreceptors at the outer nuclear layer (ONL) and normal photoreceptor morphology (Figure 1A). In those animals, NADPH-d staining revealed the typical density and distribution of capillary loops in the DCP (Figure 1D), well-defined arteries and veins at the SCP (Figure 1J), and a less dense but well-defined and continuous capillary network at the ICP (Figure 1G, arrowheads). In P23H rat retinas, an advanced degeneration of the ONL occurs at that age, where only a couple of photoreceptor cell rows remain and there are evident morphological alterations in both rods and cones (Figure 1B). Besides the photoreceptor degeneration, we also observed a reduction in the capillary content in the DCP with evident loss of capillary loops (Figure 1E). Nevertheless, the intermediate and superficial vascular plexuses showed their typical structure in these rats (Figures 1H,K, respectively). When TUDCA treatment was administered, it was able to diminish the loss of photoreceptor cells and preserve their morphology. Hence, cones and rods present longer outer segments and normal terminal axons (Figure 1C). In addition, it protected vascular degeneration, and treated animals displayed a better-preserved capillary network. The DCP of TUDCA-treated P23H rats (Figure 1F) showed more vessels and more capillary loops compared to vehicle-treated P23H rats. No differences were found in the ICP and SCP (Figures 1I,L, respectively).

To assess in more detail the effects of TUDCA on capillary plexus, blood vessels at the DCP throughout the entire retina were drawn and digitalized, and DCP vessel density and complexity were analyzed. Figure 2 shows the drawing of the entire retinal DCP of P120 rats from the three experimental groups. The area occupied by closed capillary loops was colored according to the size of the loops whose areas ranged from 200 to 10,000 μm². The density of the DCP in SD retinas (Figure 2A) and closed capillary loops of different sizes (Figure 2B) showed a typical and homogeneous distribution through the entire retina where small size loops were numerous (Figures 2C,K). TUDCA-treated P23H retinas showed less deteriorated DCP (Figure 2G) compared to vehicle-treated P23H retinas, in which a remarkable reduction of capillary density, affecting more extensively the superior and temporal quadrants, was detected (Figure 2D). This plexus degeneration was also revealed by the decrease in closed capillary loops density by 78% (Figures 2E,J), which predominantly affects the smallest capillary loops (Figures 2F,K). In these animals, TUDCA treatment resulted in a less deteriorated DCP in which closed capillary loops, even the smallest ones, were more abundant (Figures 2H–K) than in the vehicle-treated P23H retinas (Figures 2E,F).

Additional quantitative analysis of the DCP showed a decrease of 28% in the relative capillary surface area (vascular area/retinal area) of P23H compared to SD retinas (Figure 3A). TUDCA treatment was able to partially avoid this loss of DCP vessels. The capillary surface of DCP in the TUDCA-treated group was significantly higher (37%) than in vehicle-administered P23H rats (Figure 3A). No significant differences were observed between TUDCA-treated P23H retinas and SD retinas. In addition to the capillary density, the mean lacunarity was also calculated as a measure of the vascular network complexity. Although no significant differences were found between groups, lacunarity values were relatively higher in P23H retinas than in TUDCA-treated retinas and SD retinas (Figure 3B). Greater lacunarity indicates the presence of more gaps or lacunas. Thus, this tendency suggests a reduction in vascular complexity of dystrophic P23H retinas and a protective effect of TUDCA. The reduction in vascular complexity of dystrophic animals was ratified by a significant decrease in vessel junction density in P23H retinas as compared to SD retinas, indicating a reduction in vascular branching complexity (Figure 3C). Vessel junction density in TUDCA-treated P23H retinas was significantly higher than that measured in vehicle-treated dystrophic animals (Figure 3C), confirming the protective effects of TUDCA. Similar effects were observed in the percentage of total vessel length as observed in how vehicle-treated P23H rats showed lower values than TUDCA-treated animals. In addition, both groups presented a reduced vessel length compared to SD rats (Figure 3D). These results indicate that DCP of P23H rats has a more irregular vascular pattern with larger gaps and, overall, greater vascular degeneration that can be partially prevented by TUDCA administration.

To assess the effect of TUDCA treatment in the glial response of P23H rat retinas, the morphology and density of astrocytes were studied. As shown in Figure 4, astrocyte cell bodies were located at the ganglion cell layer and homogeneously distributed throughout the whole retina in all groups studied (Figures 4A–F; arrows). In all the animals, astrocytes displayed their typical flattened morphology with radial and longitudinal processes extending from the cell body to other astrocytes or blood vessels (Figure 4D). In P23H retinas, astrocytes presented gliotic features with hypertrophy of astrocytic processes (Figure 4E) compared to SD retinas (Figure 4D). In contrast, TUDCA-treated retinas (Figure 4F) showed milder gliosis, and the hypertrophy of the astrocytic processes was less evident than in vehicle-treated P23H retinas, suggesting that TUDCA partially prevents strong astrocyte gliosis.

Astrocyte quantification revealed a small, non-significant, increase in astrocyte density of P23H retinas that was exacerbated in TUDCA-treated retinas, in which astrocyte density was significantly higher than in vehicle-treated P23H and SD retinas (Figure 4G). Quantification of astrocytes in the central, medial, and peripheral retina showed a global TUDCA-induced increase in astrocyte density (Figure 4H), with greater differences in the peripheral retina. GFAP immunoreactive spots corresponding to end-feet of Müller cells were easily recognizable in TUDCA-treated and vehicle-treated P23H retinas (Figures 4B,C), suggesting Müller cell gliosis in those animals.

To better analyze the effects of TUDCA on retinal gliosis, we performed immunostaining against two intermediate filaments, GFAP and vimentin, in retinal cross-sections from the three experimental groups (Figures 5A–F). In SD retinas, GFAP immunostaining was only present in astrocytes (Figures 5A,B, green), while vimentin immunolabels both astrocytes and Müller cells (Figure 5B, red). In P23H rat retinas, vimentin immunostaining was similar to that observed in SD retinas, independent of the treatment (Figure 5D). However, Müller cells were highly positive to GFAP immunolabeling in both TUDCA-treated and vehicle-treated P23H rats, indicating extensively gliosis of Müller cells in dystrophic animals (Figures 5C,F, green) even if no differences in the number of Müller cells were detected between all groups (data not shown). GFAP staining was observed at the ONL level in the cross-sections of the TUDCA group since photoreceptors are preserved and Müller cell processes reach the external limiting membrane (Figures 5E,F, arrowheads). The loss of photoreceptor rows in vehicle-treated P23H rats resulted in less appreciable Müller cell apical processes at the ONL.

To assess TUDCA effects on the outer Müller cell processes of dystrophic rats, whole mount retinas from vehicle- and TUDCA-treated animals were immunolabeled with antibodies against GFAP, and confocal images were taken at the ONL (where the outer end of Müller processes is located) (Figures 5G–J). At this level, clusters of hypertrophied processes of Müller glia forming ring-shaped firework-like structures were numerous and widely distributed through the central and medial parts of the superior, inferior, nasal, and temporal quadrants of the retina of P120 vehicle-treated P23H rats (Figures 5G,I). In this animal model, the photoreceptor loss causes the outer end of Müller cells to laterally fall with the remnants of the ONL, developing these structures (García-Ayuso et al., 2013; Fernández-Sánchez et al., 2015a). Clusters of Müller cell processes were numerous and widely distributed through the central and medial parts of the superior, inferior, nasal, and temporal quadrants of the retina of P120 vehicle-treated P23H rats (Figures 5G,I). TUDCA decreased the presence of firework-like structures which were only visible in the medial part of the temporal and superior quadrants of the retina of treated animals (Figures 5H,J). The lack of these structures is due to the preservation of photoreceptors at the ONL.