Male pigmented transgenic rats, heterozygous for the P23H rhodopsin mutation, were bred from a cross between transgenic homozygous P23H Line 1 and normal pigmented Long Evans (LE) rats. The visual performance of the animals was monitored by means of an optomotor test given on a monthly basis (n = 8) and full field ERG recordings at P18, P21, P29, P58, P90 and P180 (n = 4). SD-OCT was performed at P130. Four normal LE rats crossed with Sprague-Dawley (SD) were used as wild-type controls at age P29. Transgenic rats were obtained from Dr. M. LaVail (UCSF), and bred in a colony at the University of Utah and at the University of Zaragoza, Spain, and maintained under a 12-hour light/dark cycle (light cycle illumination varied from 7 to 30 lux, depending on the front-to-back position within the respective cages). SD rats were obtained from Harlan Laboratories (Barcelona, Spain) and LE rats from Charles River Laboratories (Barcelona, Spain). They were housed and handled with the authorization and supervision of the Institutional Animal Care and Use Committees at both Universities. All procedures were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.

To assess visual parameters, 8 rats were measured at P30, P60, P90, P120, P150 and P180, whereas control wild-type rats were evaluated at P30. The evaluation was carried out using an OptoMotry© system (CerebralMechanics, Lethbridge, Alberta, Canada) (Prusky et al., 2004; Douglas et al., 2005). The device consists of four screens positioned around a square testing chamber. An unrestrained rat is placed on a platform in the center of the square. A virtual cylinder consisting of a sine wave grating is drawn in 3D coordinate space and rotates around the animal, while it is recorded with a video camera. The spatial frequency of the grating is maintained at the viewing position by recentering the cylinder on the rat head. The cylinder is rotated at a constant speed (12°/sec). If the grating is seen by the animal, it tracks the stimulus with reflexive head and neck movements. Spatial frequency thresholds can be measured by systematically increasing the spatial frequency of the grating at 100% contrast until animals no longer track it. This threshold is considered to be maximum visual acuity. A contrast sensitivity curve was generated by identifying the minimum contrast that generates tracking, over a range of spatial frequencies.

Correlation between visual acuity and age was assessed using Pearson’s coefficient. Differences in contrast sensitivity between ages were evaluated using a Mann Whitney U test.

Full field ERGs were recorded at P18, P21, P29, P58, P90 and P180, with four animals being studied at each time point. Four normal LE rats crossed with SD were used as wild-type controls at age P29. Following overnight dark adaptation, animals were prepared for recording under dim red light. After being anesthetized with a mixture of ketamine (150 mg/kg i.p.) and xylazine (10 mg/kg i.p.), the animals’ head was secured with a stereotaxic head holder and their body temperature was monitored with a rectal thermometer and maintained homoeothermic at 38°C. Pupils were dilated using equal parts of topical phenylephrine (2.5%) and tropicamide (1%). Topical anesthesia with 0.75% bupivacaine was used to prevent any corneal reflexes, and a drop of 0.9% saline was applied regularly to the cornea to prevent dehydration and to allow electrical contact with the recording electrode (gold wire loop). A 25-gauge platinum needle inserted under the scalp between the eyes served as the reference electrode. Amplification (at 1–1000 Hz bandpass, without notch filtering), stimulus presentation and data acquisition were provided by a UTAS-3000 system from LKC Technologies (Gaithersburg, MD).

For comparison with histopathological findings, six eyes from three P23H pigmented rats at P130 and four control rats were studied using a Spectralis HRA-OCT device. Rats were anesthetized by i.p. injection of a mixture of xylazine (10 mg/kg) and ketamine (90 mg/kg), and their pupils were dilated with tropicamide (Tropicamida^®, Alcon labs, Barcelona, Spain). In order to improve the image acquisition quality, the rats wore a PMMA afocal contact lens, purchased from Cantoor Nissel (Market Place Brackley, Northamptonshire, UK). Tear substitutive drops were continuously added to improve transparence (Systane^®, Alcon Labs, Barcelona, Spain). For examination, the animal was anesthetized using the same anesthetic mixture used for ERG recording, and placed on a platform mounted in the chin rest of the imaging device, with the eye in front of the system, so that the optic nerve could be easily visualized. A heat mat was used to keep the animals warm during the acquisition process.

Combined confocal scanning laser ophthalmoscopy (cSLO) and SD-OCT imaging was performed using a commercially available device (Spectralis HRA OCT system, Spectralis Eye Explorer 5.6b, Heidelberg Engineering, Heidelberg Germany). To adapt the system to the rat eye, a commercially available +25D lens was fit to the system (Heidelberg lens, HE 50744), and the length of the reference pathway was manually adapted according to the manufacturer’s instructions. The cSLO system used was an HRA2, using solid source for 488 and 514 nm and an infrared diode laser for 785 and 815 wavelengths. The protocol was always centered on the optic nerve. To evaluate the different thicknesses, one researcher (IP) manually changed the segmentation lines after acquisition. Values were obtained for total retinal thickness, Outer Nuclear Layer (ONL) + Retinal Pigment Epithelium (RPE) thickness, ONL thickness and Inner Plexiform Layer (IPL) thickness. Peripheral areas were used to assess intraindividual differences between superior and inferior retina thickness and temporal and nasal values. The mean values of all square areas, except those surrounding the optic nerve, were compared to evaluate differences between groups using a t test. These mean values were provided by the protocol used for the examination (Spectralis posterior pole protocol) using the optic nerve as the fixation position.

From each animal, a total of 5 single thickness profiles of the posterior map analysis were separately measured: crossing the optic nerve (section 31/61), middle upper (section 45/61) and upper (section 61/61), and middle lower (section 15/61) and lower (section 1/61). Thickness in each profile was measured at positions every 0.5 mm and compared between groups.

To acquire FAF images, fluorescence was excited using laser diodes at 488 nm and recorded between 500 and 700 nm. Fluorescein angiography was performed injecting 0.1 ml of 10% sodium fluorescein intraperitoneally.

Animals were euthanized with CO₂ at P21, P40, P58, P90, P120 and P180, and their eyes were then enucleated. The eyecups were fixed in 4% paraformaldehyde in 0.1 M PBS at pH 7.4 for 1 h, and then washed in 0.1 M PBS before being cryoprotected in 10% sucrose for 1 h, 20% sucrose for 1 h and 30% sucrose overnight at 4°C. The next day, they were embedded in OCT medium and 10-µm thick retinal sections were cut on a cryostat in a horizontal plane and mounted on glass slides. Sections were treated for immunostaining as previously described (Cuenca et al., 2004), using the primary antibodies shown in Table 1. Slides were then mounted in watermount (Vector Labs) and coverslipped for viewing by means of confocal microscopy (Leica TCS SP2 Confocal System). Pinholes measured 77 µm and the widths of optical sections were 0.9 µm. Final images were obtained from the projections of 4–7 single frames. To control for non-specific staining, some sections were stained omitting the primary antibody.

For comparing the OCT and immunochemistry findings, four control and six P23H rat sections were measured at P120 to correlate the thickness of the different layers. The measurements for the histology sections were taken in the central area, close to the optic nerve in 8 retinal sections. The OCT measurements were made in the central retina with the values obtained previously described. ImageJ software (National Institutes of Health, Bethesda, MD, USA) was used for the morphometric analysis of confocal images. Results were analyzed by GraphPad Prism (GraphPad Software, Inc. La Jolla, CA 92037 USA). After checking for normal distribution using the Kolmogorov Smirnov test, a two-tailed Student’s t-test was performed to compare the OCT with immunohistochemistry measurements. P values equal to or less than 0.01 were considered to be statistically significant.

Visual acuity for control rats was 0.54 cyc/deg. A progressive visual acuity loss with age was observed in P23H (Figure 1A), with values of 0.500 cyc/deg at P30, 0.459 cyc/deg at P60, 0.382 cyc/deg at P90, 0.285 cyc/deg at P120, 0.253 cyc/deg at P150 and 0.182 cyc/deg at P180. This decline in visual acuity was well fit by a linear regression model, with a Pearson R² value of 0.9903. The same trend was observed for the contrast sensitivity curve (Figure 1B). Peaks of 54.56 (P30), 50.30 (P60), 35.82 (P90), 22.04 (P120), 10.69 (P150) and 2.98 (P180) were obtained for a spatial frequency of 0.089 cycles/deg. The control rat value for this spatial frequency was 56.15.

Both the visual acuity and contrast sensitivity parameters began to decline at 30 days of age, and continued this trend throughout all the time points studied until the end of the experiment, when some degree of visual acuity still remained.

Scotopic a-waves were already reduced in amplitude at P18, achieving maximal values of 223 ± 27 µV (Figure 2A). Instead of increasing from P18 to P29, as would be expected (Dembinska et al., 2002) a-wave amplitudes reached a maximum of 206 ± 11 µV at age P29. This value amounted to only 34% of the corresponding value in wild-type LExSD rats aged P29 (604 ± 45 µV). Despite being affected at early time points, a-waves from pigmented P23H rats could still be elicited up to P180, the latest time points studied (Figure 2A).

In addition to being reduced in amplitude, the mixed scotopic a-waves in P23H rats had faster leading edges than in wild-type LExSD rats, indicating greater sensitivity. Figure 2B shows an example at 29 days of age, when the a-wave amplitude in P23H rats was only 29% of that in wild-type rats. Here, a-wave traces have been normalized against their respective maximal a-wave amplitude. As such, the resulting b-wave in P23H rats appears relatively higher than in wild-type rats because in the former the b-wave amplitude is relatively better preserved than the a-wave (see the trend of the respective ascending b-wave traces on the far right of Figure 2B).

At all ages, b-waves were less affected by the P23H rhodopsin transgene than were a-waves. While maximal b-wave amplitudes in both rod- and cone-driven ERG responses increased from P18 to P29, only cone-driven b-wave responses reached developmental maturity in pigmented P23H rats (Figures 2C–F). Although double flash-isolated rod-driven responses were already affected at P29, with a- and b-waves peaking at 203 and 564 µV vs. 610 and 998 µV in wild-type rats, these responses were still elicited at P180 (a- and b-wave maximal amplitudes of 60 and 211 µV). The rod-driven contribution to mixed scotopic b-waves was slightly lower in P23H (70%) as compared to LExSD (79%) rats, but remained steady at 58% by P180 (Figure 2D). Double-flash isolated scotopic cone b-waves (Figure 2E) and photopic single flash b-waves (Figure 2F) were comparable to those recorded in wild-type LExSD rats at P29. After P29, cone-driven single flash responses underwent slow, protracted deterioration (Figure 2F). By P180, maximal photopic b-waves still reached half of the amplitudes recorded in wild-type rats. Flicker ERGs (Figure 2G) confirmed the findings obtained with single flash cone-driven responses. Flicker amplitudes were comparable between P23H and wild-type rats at P29 (271 ± 30 µV vs. 306 ± 22 µV, respectively). Flicker fusions were also comparable at this age (42 ± 6 Hz and 42 ± 3 Hz in P23H and wild-type LExSD rats, respectively). By P180, flicker fusions in P23H rats were reduced in a statistically significant manner (31 ± 4 Hz compared to 42 ± 3 Hz in wild-type rats), as were maximal amplitude values (147 ± 19 µV in P23H rats vs. 306 ± 22 µV in control rats).

An indication of inner retina activity can be obtained by measuring the amplitude of oscillatory potentials (OP; Figure 2H). As with cone-driven b-waves, OP amplitudes increased from P18 until P29, at which point these values were comparable to those in age-matched wild-type rats. After P29, OP amplitudes decreased gradually to 16 ± 4% of the OP amplitudes in wild-type rats.

In summary, in P23H rats, only cone-driven b-waves (and mixed scotopic OPs) reached similar maximum amplitude values as in wild-type rats (Figure 2H). While mixed scotopic a-waves were the most severely affected ERG component, all components underwent a similar rapid reduction in maximal amplitude from P29 to P58, followed by a mild protracted drop from P90 to P180.

In all animals, a basic examination using en face cSLO imaging was performed to identify any retinal abnormalities. The examination included red-free (513 nm), infrared (830 nm) and FAF (492 nm) imaging. Fluorescence angiography was used to analyze retinal vessels (Figure 3).

Almost all the rats analyzed showed features of the Bergmeister papilla, a remnant of the developmental vascular vitreous. No other abnormalities were seen in the red-free and infrared examination (Figures 4A,B).

Control rats showed no autofluorescence in their fundus (Figure 3A). Change in FAF in the P23H rats was noted as sparse autofluorescent dots (Figure 3B). No relevant changes were observed in the angiography pattern (Figures 3C–F). Major retinal vessels emerged from the optic nerve in a radial fashion, and retinal vessels ramified and narrowed into a sparse capillary network. Both great vessels and the capillary network in P23H pigmented rat showed no permeability changes and no lost areas of capillary circuitry at this age.

Structurally, the sections showed changes between control and P23H rats. Control rats exhibited a normal OCT profile, with hyporeflectivity at the nuclear layers and hyperreflectivity in the plexiform layers and external retina, including outer segments, RPE and choroid. In normal rats, ganglion cells and their fibers are distributed together at the internal retina, limited by the internal limiting membrane (ILM); vessels were shown transversally cut with high reflectivity in their walls, distributed between the ganglion cells and the retinal nerve fiber layer. A hyperreflective IPL was followed by a hyporeflective INL, a high reflectivity OPL and a low reflectivity ONL. The four hyperreflectivity layers could be clearly observed at the external retina (Figures 4A, 5A; Spaide and Curcio, 2011). In transgenic rats, the ONL appeared clearly diminished in all the retinal profiles (Figures 4B, 5C). The RPE was disrupted and also diminished in thickness. Although the OPL was easy to find, the ELM showed gaps in its continuity and was more difficult to recognize (Figures 4B, 5C). The hyperreflective lines of the external retina were less evident in the P23H pigmented rat. The thickness of the IPL appeared almost normal. Figure 5 shows the retinal profiles obtained by OCT in control (Figure 5A) and P23H (Figure 5C) rats and its comparison with immunohistochemistry staining (Figures 5B,D). We applied both techniques in order to compare the data acquired in P23H and control rats at P120 (Figure 5E). The usefulness of OCT measurements for evaluating the progression of retinal disease is clear, albeit the data comparing their values with histology measurements are limited. Figure 6 shows an example of thickness maps of control and P23H rats. Mean peripheral areas were statistically higher in control vs. P23H rats. No differences were found between superior and inferior areas and nasal and temporal thicknesses in rats from the same group with any of the techniques.

All the values obtained in the retinal thickness profiles and maps of transgenic rats were significantly lower than those of control rats, with the exception of IPL thickness (Figure 5E). Total thickness retinal values were diminished in transgenic rats and the reduction was evident in the segmented layers (total retina thickness, ONL + RPE, and ONL). The main decrease was observed for ONL measurements. In addition, we found no differences between the data obtained with OCT and immunohistochemistry in the inner retina, while the outer retina showed differences between both techniques, with higher OCT (p < 0.001) than immunochemistry measurement values (Figure 5E).

Anatomical studies were performed to compare them with the visual acuity, ERG and OCT image findings. For correlation with functional findings, we focused on cells involved in the generation of a- and b-waves, such as photoreceptors and their corresponding bipolar cells, respectively, and in the plexiform layers to check the synaptic relationship between the retinal cells.

OCT is an important tool for retinal evaluation in all kinds of ophthalmological diseases. The fast acquisition and the high number of scans have made SD-OCT a valuable tool in ophthalmic research. The resolution of cross sectional images obtained by third generation OCT has been described as being similar to low power micrographs obtained from light microscopy (Huber et al., 2009). In our study, retinal thicknesses presented a good correlation with histological findings, and OCT proved able to assess the loss of retinal thickness with degeneration, as it showed a clear reduction in the ONL layer, as would be expected in a rod-loss disease. Other authors have shown a good correlation between retinal thickness measured by SD-OCT (Spectralis OCT) in wild-type mice and transgenic animals (rd1, Rho−/–, RPE65−/–) and histological findings (Huber et al., 2009; Jiao et al., 2013; Berger et al., 2014). Their results improved depending on the fixation protocol; thicknesses showed a better correlation when they used a direct embedding method as opposed to overnight fixation. Our previously described protocol used fixation and cryoprotection overnight, with increasing sucrose concentration. Using our standard protocol, the values obtained by immunohistochemistry are significantly lower than those measured in vivo during OCT exploration. However, the decreasing trend at the ONL level was clear with both techniques, and both methods exhibited an almost perfect correlation between them in terms of these measures. These findings were not supported by the IPL measurements, despite the fact that the IPL was the best preserved layer with both methods. The main reason for this finding is probably the difficulty in establishing a good limit for the layer, not only with OCT, but also histochemistry. The IPL is the latest layer to be affected, and its measurements remain close in value to those obtained in normal rats. In the immunohistochemistry examination, the S5 stratus was sometimes difficult to establish; OCT limits are clear at these ages, but it may be difficult to establish them in older rats (data not shown), due to cellular and vascular changes during retinal remodeling.

FAF findings are consistent with other retinal degenerative diseases, where photoreceptor debris and the accumulation of lipid degradation products are shown as hyperfluorescent dots. The findings resembled those found in RPE65−/– mice, but not so much as our own results for albino P23H rats (data not shown). Albino rodents are more susceptible to light-induced damage and show faster accumulation of lipid degradation products. P23H albino homozygous rats have an appearance resembling that of pigmented heterozygous rats at 2 months of age, with no changes at P21 (unpublished data). Albino wild-type mice BALB/c showed hyperfluorescent dots at early ages (Huber et al., 2009). Changes in FAF related to A2E accumulation has been described in other animal models of retinal degeneration (Charbel Issa et al., 2013).

Although is well known that retinal vessels change with degeneration (Villegas-Pérez et al., 1998; Fernández-Sánchez et al., 2012), Fluorescence angiography failed to show these changes at the age studied (4 months of age). All retinal vessels conserved their morphology, with no changes in their walls or any other signs or staining.

The visual function of retinal degenerative animal models has been measured in previous studies using an OptoMotry system (Umino et al., 2006; McGill et al., 2007; Zhou et al., 2009; Lu et al., 2010). However, this study is the first to describe the spatial frequency and contrast sensitivity thresholds in the P23H-1 rat. Our results in control pigmented rats are similar to those reported by other studies (Douglas et al., 2005). Visual acuity reached 0.540 cyc/deg. The contrast sensitivity curve displayed a typical inverted-U shape. Both visual acuity and contrast sensitivity decreased with age in P23H-1 rats. At P30, visual performance was close to that found in control rats, and the results were consistent with those observed with ERG. As in control rats, the cone driven a-wave had reached maturity, but by this age, the rod driven b-wave was already reduced. Both functional tests demonstrate a gradual decrease in visual acuity, contrast sensitivity and ERG responses from the P30 age point to early in the P240 interval, showing a progressive photoreceptor deterioration that affects first the rods and then the cones. This agrees with the progressive loss of photoreceptors, which first affects the rods, and then both the rods and cones.

The comparison of the present ERG results with previously published results from similar tests applied to heterozygous P23H line 1 rats, bred on an albino background (Machida et al., 2000), supports the hypothesis that albinism exacerbates the P23H-related retinal degeneration. While mixed scotopic maximal a-wave and b-wave amplitudes were initially better preserved in pigmented animals (at ages P29 and P28 in pigmented and albino rats, respectively), they underwent a similar decline up to the last time points tested (P180 and P203 in pigmented and albino rats, respectively). However, maximal photopic b-wave amplitudes did have higher proportions as compared to the respective wild types at early and late time points. These comparative findings support the suggestion that cone function might be more affected than rod function over time by the lack of pigmentation. Finally, the findings that the a-wave leading edge is faster in pigmented P23H than in wild types, as reported in albino P23H rats (Machida et al., 2000), suggests that pigmentation has no effect on rod phototransduction.

At the anatomical level, heterozygous P23H line 1 rats bred on an albino background showed a progressive loss of rod photoreceptors over several months. By P28, the ONL had lost approximately 40% of its cells (Machida et al., 2000). At this age, we observed rhodopsin mislocated along the remaining rod photoreceptor cells, which is a regular trait in RP diseases (Cuenca et al., 2004, 2005; Barhoum et al., 2008; Martinez-Navarrete et al., 2011). By P180, the ONL had fewer than two rows of cells remaining, and they were essentially cones. Our results indicate that pigmented P23H rats at age P21 had a similar number of photoreceptor rows as wild-type LE rats, while around 12 rows of cells remained at P40. This number then fell to 6 rows at P58, with a further progressive decline until P180, when only 2 layers of photoreceptors were left, consisting mainly of cone photoreceptor cells. As in other RP models reported by different group (Cuenca et al., 2014), the loss of rod photoreceptors leads to morphological changes in other retinal cells. Cone photoreceptor cells undergo changes affecting their length, the position of their cell bodies in the ONL, as well as changes in their outer segments, which become shorter and swollen than in control retinas.

The loss of photoreceptor cells is reflected as a reduction of synaptic contacts between photoreceptors and their postsynaptic bipolar and horizontal cells. Changes in bipolar cells were slower and more subtle than those reported in albino P23H rats (Cuenca et al., 2004): they became apparent only around P60, while this set of changes in bipolar cells appeared around P40 in the albino rat. The degenerative trends observed in the horizontal cell distribution and organization in P23H-L1 pigmented rats at P90 is almost achieved at P40 in the albino rat (Cuenca et al., 2004). All these observations confirm that the degenerative process that took place in the inner retina is closely related to the rod degeneration rate in each line, and should be considered in the design of RP therapies.

The good correlation shown between OCT and histology results is the important key to our work. The possibility of evaluating the retina using non-invasive techniques reduces the number of animals that need to be sacrificed and still provides good data about the progression of the disease and the integrity of the retinal layers. Retinal thickness is not the only variable to provide valuable results; the autofluorescent findings also contribute information about the integrity of the RPE layer.

Pigmented RP rodent models have clear benefits over the albino ones. As in other models, albinism itself exacerbates the retinal degenerative processes. The changes seen in the inner retina are slower that those reported for the albino background in terms of the rate of decline in ERG responsiveness. Preservation of the retinal circuitry is an important factor in the effectiveness of potential preventive treatments. In addition, pigmented rats perform better on behavioral tests aimed at assessing vision. For these reasons, a pigmented background might be optimal when relying on P23H transgenic rats to develop and study experimental therapies, such as cell-based therapy.

The finding that cone function is unaffected by P29 provides us with the opportunity to study approaches aimed at the full functional prevention of blindness, using therapeutic approaches in the pigmented heterozygous P23H line 1 rat, providing that it is bred on a pigmented background. It should be noted that in our study, we were unable to detect a functional correlate of the subtle changes in cone cell body distribution at P21. Whether this morphological change has any impact on retinal function remains to be clarified.