Retinal compensatory changes after light damage in albino mice

Animal handling

Two-month-old female BALB/cAnNHsd albino mice from Harlan (Barcelona, Spain) were used. All animals were treated according to our institutional guidelines, the European Union regulations, and the Association for Research in Vision and Ophthalmology Guidelines for the use of animals in research.

Before light exposure, the animals were reared in cages containing four animals, fed ad libitum, and kept in a 12h:12h light-dark cycle under controlled temperature (22–24 °C) and humidity in the viverium of the University of Murcia. Light intensity within the cages oscillated between 5 and 50 lx.

Animals were studied at 1 (n=8), 3 (n=8), 7 (n=16), 14 (n=12), 30 (n=13), 90 (n=10), or 180 (n=10) days ALE. The ERG responses obtained from groups processed 30 days or earlier ALE were compared to their baseline responses recorded before light exposure. The ERG responses from groups processed 90 or 180 days ALE were compared to those obtained from age-matched control animals to take into account the effect of age on retinal function.

For all manipulations and experimental procedures (except light exposure), the animals were anesthetized with an intraperitoneal injection of a mixture of ketamine (70 mg/kg Ketolar®, Pfizer, Alcobendas, Madrid, Spain) and xylazine (10 mg/kg Rompun®, Bayer, Kiel, Germany) in 0.1 ml of saline.

Light exposure

Before light exposure, left pupil mydriasis was induced by topical application of a drop of 1% atropine (Colirio de atropina 1%®; Alcon S.A., Barcelona, Spain). The right eye was not dilated to be used for comparison.

Because the severity of retinal phototoxicity in rodents depends on the time of the day when exposure starts (circadian rhythm) and on the previous light exposure (rearing conditions and dark adaptation period) [3,26-28], the exposure always began between 10 and 12 AM and after 12 h of dark adaptation.

Animals were individually exposed to 24 h of continuous fluorescent cold white light. Fluorescent bulbs were situated in the ceiling right above the animal cages, which were transparent. Light intensity within the cages was 3,000±100 lx. Mydriasis was inspected 12 h after the initiation of light exposure, and when necessary, a second drop of atropine was instilled in the left eye at this time.

Electroretinography

ERG recordings were performed as previously described [10]. Briefly, animals were dark adapted overnight before ERG recordings, and their manipulation was performed under dim red light (λ>600 nm). Mice were anesthetized and bilateral pupil mydriasis was induced by topical application of a drop of 1% tropicamide (Colirio de tropicamida 1%®; Alcon, S.A., El Masnou, Barcelona, Spain) in both eyes; the animals’ body temperature was kept at 37 °C by placing them on top of a water heat pad (TP500 T/Pump; Gaymar Industries, Orchard Park, NY). For light stimulation, a Ganzfeld dome and multiple reflections of light generated by light-emitting diodes were used. For high-intensity illuminations, a single light-emitting diode placed close (1 mm) to the eye was used. Light intensity was calibrated by a dual-biosignal generator device specifically adapted for ERG responses. ERG recordings were made using Burian-Allen bipolar electrodes shaped as a corneal contact lens (Hansen Labs, Coralville, IA). The electrical signals generated in the retina were amplified (×1,000) and filtered (band pass from 1 Hz to 1,000 Hz) with a commercial amplifier (Digitimer Ltd., Letchworth Garden City, UK). The recorded signals were digitized (Power Lab; ADInstruments Pty. Ltd., Chalgrove, UK) and displayed on a personal computer. ERG recordings were taken simultaneously from both eyes. Light stimuli were calibrated before each experiment to assure identical recording parameters for both eyes. The retina was stimulated using light intensities ranging between 10⁻⁵ and 10⁻⁴ cd·s·m⁻² for the STR, 10⁻⁴ and 10⁻² cd·s·m⁻² for the response mediated by rods, and between 10⁻² and 10² cd·s·m⁻² for the mixed response (response mediated by rods and cones). The animals were then light adapted for 5 min and a maximum intensity stimulus (10² cd·s·m⁻²) was used to obtain the photopic response (mediated by cones). For each light stimulus, a series of ERG responses was averaged and the interval between light flashes was adjusted to allow response recovery between stimuli. After each session, tobramycin ointment (Tobrex®; Alcon, S.A., El Masnou, Barcelona, Spain) was applied on both corneas to prevent corneal desiccation during recovery from anesthesia. Recording analysis was performed using the normalization criteria established by the International Society for Clinical Electrophysiology of Vision.

Horseradish peroxidase injection: vascular leakage

Four control animals and various experimental animals that were processed 1 (n=4), 3 (n=4), 7 (n=4), 14 (n=4), and 30 (n=4) days ALE received, 10 min before processing, an injection in the inferior cava vein of 0.1 ml of horseradish peroxidase (100 μg/μl diluted in saline; HRP; type I, MW 44 kDa, Sigma P 8125, Steimheim, Germany).

Fluoro-Gold application: retinal ganglion cell tracing

In four control animals and various experimental animals that were processed 3 (n=8) and 6 months (n=4) ALE, Fluoro-Gold (FG) was applied to both superior colliculi 7 days before processing, following the techniques already described [29,30]. Briefly, animals were anesthetized, the mid-brain was exposed and, after removing the pia mater overlying the superior colliculi, a piece of sponge (Spongostan Film, Ferronsan, Denmark) soaked in 3% FG diluted in 10% dimethyl sulfoxide-saline (Sigma Aldrich, Madrid, Spain) was applied over the surface of both superior colliculi. FG is incorporated to the axon terminals and retrogradely transported to the retina where it accumulates in the RGC somas.

Tissue processing

Animals were sacrificed with an intraperitoneal lethal dose (0.4–0.5 ml) of 20% sodium pentobarbital (Dolethal®; Vetoquinol S.A., Lure, France). The eyes were enucleated and processed to obtain either retinal whole mounts or cross sections.

Whole mounts

Horseradish peroxidase–injected animals. Eyes were enucleated and immersed for 1 h in 4% paraformaldehyde in 0.1M PBS (pH 7.2–7.4). Later, the retinas were dissected as whole mounts by making four radial cuts in the superior, inferior, temporal, and nasal retina. To maintain their orientation, the deepest cut was made in the superior retina. The retinas were then postfixed for 1 h in the same fixative solution, rinsed in PBS, reacted for HRP demonstration using a modified Hanker-Yates reaction [31] to visualize the retinal vasculature [5,6,18,20,32], rinsed, and mounted on gelatin-coated slides, vitreal side up, with a small quantity of a solution of 50% glycerol in PBS.

Fluoro-Gold-traced animals. Animals were first perfused through the left ventricle with saline and subsequently with 4% paraformaldehyde at 4 °C. The eyes were enucleated and processed in the same manner as the eyes of the HRP-injected animals. The whole mounts were immediately mounted as before, and observed and photographed with a fluorescence microscope (Axioscop 2 Plus; Zeiss, Jena, Germany).

Cross sections

Retinal cross sections were obtained from control (n=5) and experimental animals processed 1 (n=5), 3 (n=8), 7 (n=5), 14 (n=4), 30 (n=7), 90 (n=6), or 180 (n=5) days ALE. After perfusion, as above, the eyes were enucleated and the superior portion of the sclera and the superior rectus muscle marked with china ink to maintain their orientation. The cornea and lens were removed and the eyecups were postfixed for 48 h in the same fixative, dehydrated through alcohols and 1-butanol, and embedded in paraffin. Three-micron-thick cross sections were obtained in a rotational microtome (Microm HM-340-E; Microm Laborgerate GmbH, Walldorf, Germany) and mounted on slides coated with 0.01% poly-L-lysine (Sigma-Aldrich, St. Louis, MO). Only the sections containing the optic nerve head were kept and stained with hematoxylin and eosin, or processed for TdT-mediated dUTP nick-end labeling (TUNEL) or immunohistofluorescence.

Hematoxylin and eosin staining

A series of sections (4–6 per animal) were deparaffinized, rehydrated, stained with Hansen’s hematoxylin and eosin, and mounted with DePex (BDH Laboratory Supplies, Poole, UK).

TdT-mediated dUTP nick-end labeling

Some sections (4–6 per animal) were used to detect apoptotic nuclei using the TUNEL method. TUNEL assay was performed according to the manufacturer’s protocol (FragEL™ DNA Fragmentation Detection Kit, Qiagen, Merck Bio, Nottingham, UK) with slight modifications as follows: Biotin-labeled DNA was detected by 2 h incubation at room temperature with avidin- tetra-methyl-rhodamine isothiocyanate (TRITC; Sigma-Aldrich, Madrid, Spain) diluted 1:500 in PBS containing 0.1% Triton. After washing, slides were mounted with antifading medium containing 4',6-diamidino-2-phenylindole (DAPI; VectaShield Mounting Medium with DAPI, Vector, Atom, Alicante, Spain) to counterstain all retinal nuclei.

Immunohistofluorescence

Additional sections (4–6 per animal) were washed in PBS, immersed in citrate buffer (pH 6.0), and heated for 10 min in a microwave for antigen retrieval. They were then cooled at room temperature for 20 min, washed in PBS, and processed for immunohistofluorescence. In brief, the sections were incubated overnight at 4 °C with either mouse anti–protein kinase Cα (PKCα; 1:250; Santa Cruz Biotechnology, Heidelberg, Germany) or mouse anti-synaptophysin (1:250; Sigma-Aldrich, Madrid, Spain) diluted in blocking buffer (0.5% TritonX-100 and 10% normal goat serum in PBS). Secondary detection was performed using Alexa Fluor-488 goat anti-mouse or Alexa Fluor-568 goat anti-mouse (Molecular Probes, Invitrogen, Barcelona, Spain), each diluted 1:500 in PBS–0.5% Triton. Finally, all sections were thoroughly washed in PBS and mounted with a small amount of antifading medium (Vectashield® Mounting Medium; Vector; Atom, Alicante, Spain). Negative controls were obtained by the suppression of primary antibodies.

Image analysis

Retinas were observed and photographed with a light/fluorescence microscope (Axioscop 2 Plus; Zeiss) equipped with different fluorescence filters, a digital high-resolution camera (ProgRes^TM c10; Jenoptic, Jena, Germany), and a motorized stage (ProScan^TM H128; Prior Scientific Instruments Ltd., Cambridge, UK) connected to an image analysis system (Image-Pro Plus 5.1 for Windows®; Media Cybernetics, Silver Spring, MD) with an automatic frame-grabber device (Scope-Pro 5.0 for Windows®; Media Cybernetics). Individual frames were taken from areas of interest in some cross sections. Photomontages of whole mounted retinas or ocular cross sections were also obtained. These images were used for qualitative or quantitative analysis, and in some cases (see below) were automatically analyzed.

Hematoxylin and eosin–stained cross sections were qualitatively analyzed. In addition, the thickness of the outer nuclear layer (ONL) was quantified in these sections by counting the number of nuclei rows at eight equidistant retinal locations (four dorsal and four ventral), taking the optic nerve as reference (location 0). This analysis was performed in the right and left retinas of three control and three experimental animals processed at different times ALE. Three retinal cross sections were quantified per retina.

Using two different fluorescence filters, TUNEL and DAPI signals were acquired from the same representative regions of the peripapillary retina. Gross numbers of TUNEL-labeled nuclei were qualitatively analyzed in these photographs. To obtain composites of these photographs, TUNEL images were coupled using the “dissolve” option of the Adobe® Photoshop® (Adobe Systems Inc., San Jose, CA) CS3 software. This option gives an artificial signal, and thus the TUNEL red fluorescence is enhanced.

Automated quantification of Fluoro-Gold-traced retinal ganglion cells in whole mounts

Photomontages of FG-traced retinal whole mounts were generally composed of 140 individual rectangular images (0.627 mm² each). In each frame, RGCs were automatically counted to obtain a total number of RGCs per retina using previously published methods [33]. Briefly, we used Image-Pro® Plus (version 5.1 for Windows™; Media Cybernetics Inc., Bethesda, MD) macro language to apply a sequence of filters and transformations to each image to clarify cell limits and to separate individual cells for automatic cell counting. Each frame was also divided into 64 rectangular areas in which the density of FG-traced RGCs was calculated, color-coded, and graphically represented to obtain an isodensity map for each retina, using methods that have been previously described [33]. Briefly, we developed a specific subroutine using Image-Pro® Plus macro language in which every retinal frame was divided into 64 equally-sized rectangular areas of interest (AOI). In each AOI, the RGC number was obtained and the cell density was calculated. RGC densities per AOI were exported to a spreadsheet (Microsoft Office Excel 2003; Microsoft Corporation, Redmond, WA) and finally, data were represented as filled contour plots using a graphic software (Sigmaplot® 9.0; Systat Software Inc., Richmond, CA).

Automatic measurement of fluorescence in cross sections

Measurement of fluorescence was performed in the retinal cross sections processed for immunohistofluorescence. All images were acquired using a fluorescence microscope (20×; ProgRes™ c10; Jenoptic) with a motorized stage (ProScan™ H128; Prior Scientific Instruments Ltd.) connected to an image analysis system (Image-Pro Plus 5.1 for Windows®; Media Cybernetics) with an automatic frame-grabber device (Scope-Pro 5.0 for Windows®; Media Cybernetics). Only one photomontage of one whole cross section spanning the optic nerve was automatically analyzed per animal and time point using a specific subroutine developed for this purpose. In brief, we used the Image-Pro Plus 5.1 for Windows® macro language to apply a sequence of actions and filters to obtain the fluorescent area, maximum and minimum intensity values, average, and standard deviation.

In the first step, the image was converted to 16-bit grayscale, where 0 represents black color (no fluorescence) and 65.535 (2^16) represents white (maximum fluorescence). Then, the contour of the section was drawn on the whole-mount image, and a new image containing the selection and a clear background was created. In the second step, a line was drawn manually in the border between the inner nuclear layer (INL) and the outer plexiform layer (OPL) to separate the external retina (up to the OPL) and internal retina (down to the INL). The contrast of the external retina was inverted, maintaining the contrast of the inner retina untouched to enable a separate layer study. In a third step, 20 equidistant points were automatically dotted along the retinal section (reference point 0 in the optic nerve, 10 dorsal, and 10 ventral). These dots occupied the center of 20 round areas of interest, in which fluorescence density was measured automatically, both in the outer and inner retina. Finally, measurements were exported to a Microsoft Office Excel 2003, (Microsoft Corporation, Redmond, WA) worksheet.

Statistical analyses

Numeric data were fed into Microsoft® Office Excel 2003 worksheets and exported and analyzed using SigmaStat® 3.1 for Windows® (Systat Software, Inc., Richmond, CA). Descriptive statistics were calculated for each group and different tests were used for comparison between groups. Differences were considered significant when p≤0.05.

Electroretinographic analysis

To measure the ERG waves and the implicit times, the International Society for Clinical Electrophysiology of Vision criteria were used. The scotopic threshold response (STR) was analyzed for each stimulus; the positive STR (pSTR) was measured from baseline to the “hill” of the positive deflection, approximately 110 ms from the flash onset, and the negative STR (nSTR) was measured from baseline to first “valley” after the pSTR, approximately 220 ms from the flash onset. The paired t test was used to compare the ERG responses in the same group of animals before and after light exposure. The Mann–Whitney and Kruskal–Wallis tests were used to compare the ERG parameters between groups of animals.

Analysis of the retinal ganglion cell population

The Mann–Whitney, Kruskal–Wallis, and Student t test were used to compare the numbers of RGCs between groups of animals.

Fluorescence intensity analysis for protein kinase C alpha and synaptophysin

The Kruskal–Wallis test was used to compare fluorescence intensity between groups of animals.

Photoreceptor loss after light exposure

Phototoxicity-induced vascular leakage

We have documented recently that in the rat, light exposure induces vascular leakage in an arciform area located in the dorsal retina. Vascular leakage occurs during the first week and first month ALE in albino and pigmented rats, respectively [5,6]. Importantly, this arciform area of vascular leakage is where the highest sensitivity to light was observed. To document whether phototoxicity also caused vascular leakage in mice, HRP was intravenously injected in control (Figure 5A) and photoexposed mice (Figure 5B-D) and the retinas processed for HRP. These data demonstrated that in mice, there does not appear to be vascular leakage after light exposure, at least with the experimental conditions and times ALE studied here.

Retinal function: electroretinographic responses

The baseline ERG responses obtained before light exposure were normal and similar in both eyes in all the animals studied. The same animals were recorded prior and ALE in the groups of animals processed for up to 30 days ALE, while age-matched animals were used as control animals in the groups processed 90 and 180 days ALE to account for any possible aging effects. The scotopic ERG responses of the photoexposed animals varied depending on the survival interval, the wave, and the stimuli, but were similar in both right and left (dilated) eyes (Figure 6 and Figure 7). On the other hand, the photopic responses were diminished but varied greatly between animals and thus were not conclusive (data not shown).

The STR is the ERG wave that reflects the functionality of the inner retina (Figure 6, three first rows; Figure 7A,C). One day ALE, the pSTR and nSTR were reduced to approximately 35% of the values obtained before light exposure (baseline), and this reduction was statistically significant (paired t test, p<0.05; Figure 7C). Three days ALE, the pSTR and nSTR had recovered compared to 1 day ALE, but only to approximately 65% and 78% of their baseline values, respectively (Figure 7A,C). The absolute values of pSTR were significantly smaller than the baseline records (paired t test, p<0.05; Figure 6C), while the values of the nSTR were significantly smaller for only one of the three stimuli studied (the −4.02 log light unit stimulus; Figure 7A). Seven days ALE, the recovery of the pSTR and nSTR continued, reaching approximately 80% and 90% of the baseline values, respectively (Figure 7C). This reduction was still significant (paired t test, p<0.05) for the pSTR for all the stimuli studied, but for the nSTR this was only significant for the −4.12 log light unit stimulus (Figure 7A). From 14 until 180 days ALE, the STR waves were not significantly different from control recordings (baseline or age-matched controls; Student’s t test, p>0.05) or from animals processed at earlier times ALE (Kruskal–Wallis, p>0.05; Figure 7A,C).

The a- and b- waves of the ERG reflect the functionality of the photoreceptors and the bipolar and/or Müller cells, respectively (Figure 6, five last rows; Figure 7B,D). One day ALE, there was a reduction of the a- and b-waves (paired t test, p<0.05) to approximately 40% of baseline values (Figure 7D). Three days ALE, both waves recovered compared with the previous time point, but still were significantly lower than the baseline values (70%, paired t test, p<0.05; Figure 7D). At seven days ALE, their amplitudes continued to be significantly reduced (paired t test, p<0.05), although they had recovered to 65% (a-wave) and 90%, (b-wave) of their baseline amplitudes (Figure 7B,D). Fourteen days ALE, the a- and b-waves were reduced to approximately 65% and 80% of their baseline values. This reduction was significant compared to baseline (paired t test, p<0.05), but not compared to 3 or 7 days ALE in the case of the a-wave or to 7 days ALE in the case of the b-wave (Kruskal–Wallis, p>0.05; Figure 7D). Thirty days ALE, the amplitudes of the a- and b-waves were still reduced to approximately to 65% and 80%, and these reductions were statistically significant compared to controls (paired t test, p<0.05), although not significantly different from those observed 14 days ALE (Kruskal–Wallis, p>0.05; Figure 7D). Ninety and 180 days ALE, the a-wave was still maintained at 70% of control values (paired t test, p<0.05 at 90 days, p<0.001 at 180 days), while the b-wave had recovered to 90% of controls. This value was not different from that observed from 7 to 30 days ALE (Figure 7D). At these time points, the reduction of the b-wave was significantly smaller only for the most intense flash stimulus (Student’s t test, p≤0.05; Figure 7B).

In conclusion, our data indicate that phototoxicity causes a transient functional impairment of the inner retina, because after a significant decrease during the first 7 days ALE, the STR recovers, regaining its baseline values at 14 days. Photoreceptor functionality is permanently affected, as the a-wave impairment lasts until 180 days ALE, although it showed some recovery during the period of the study. Bipolar and/or Müller cells do not regain full functionality because, even though the b-wave improves almost to control values, at 90 and 180 days ALE the response to the brightest stimulus is still affected (Figure 6 and Figure 7).

Compensatory changes: protein kinase C α and synaptophysin expression levels after light exposure

Because our functional data indicated a partial recovery of the a-wave, an almost complete recovery of the b-wave in the first 3 months ALE, and a complete recovery of the STR waves in the first 14 days, we speculated that there could be compensatory retinal changes to account for the functional recovery. To assess this possibility, we studied the expression pattern and levels of two proteins: PKCα, a protein kinase associated with bipolar cells, and synaptophysin, an integral membrane protein expressed in the presynaptic vesicles [34].

In control retinas, PKCα is expressed in the inner plexiform layer (IPL) and the INL (Figure 8A). Quantification of the intensity of fluorescence per area (Figure 8K) showed that this expression is around 3.6±0.3 (×10⁴) and 2.9±0.3 (±10⁴) arbitrary units for the INL and IPL, respectively. Synaptophysin is expressed in the OPL and IPL (Figure 8D), with a fluorescence intensity of 1.3±0.2 (×10⁴) and 1.6±0.2 (×10⁴) arbitrary units, respectively (Figure 8L).

One day ALE, the PKCα signal decreases in both layers below control values (INL: 0.9±0.09 [×10⁴]; IPL: 1.6±0.13 [×10⁴] arbitrary units; Figure 8B,G,K). From this time forward, PKCα expression recovers gradually, and by 90 days ALE in the IPL and 180 days ALE in the INL, reaches values similar to those found in control animals (Kruskal–Wallis, p>0.05). In the IPL, the PKCα signal at 180 days is significantly higher than in control retinas (5.7±0.49 [×10⁴] arbitrary units, Kruskal–Wallis, p<0.05; Figure 8C,H,K).

The behavior of synaptophysin follows a similar trend. One day ALE, its signal diminishes below control values in the IPL (0.36±0.04 [×10⁴]) and OPL (0.66± 0.08 [×10⁴]) (Kruskal–Wallis, p<0.05; Figure 8E,I,L). In the OPL its signal is kept at the same lower levels up to day 14 ALE, and thereafter its expression augments until reaching normal values at day 180 ALE (Kruskal–Wallis, p>0.05; Figure 8L). In the IPL, however, after reaching its lowest level at day 14 ALE, the synaptophysin signal continuously increases and by day 180 ALE, its value is significantly higher than in control animals (3.9±0.3 [×10⁴] arbitrary units, Kruskal–Wallis, p<0.05; Figure 8F,J,L).

Retinal ganglion cell population and spatial distribution

In albino rats, phototoxicity induces a delayed death of RGCs, which affects the whole retinal surface, although there are specific sectors devoid of RGCs. To analyze whether there is also RGC death in albino mice, we automatically quantified FG-labeled RGCs and assessed their spatial distribution. The mean number±standard deviation (SD) of FG-traced RGCs in control animals was 42,477±3,953 cells (Figure 9 and Figure 10A). RGCs were distributed in a nonhomogeneous way. An area of maximal RGC density was observed above the optic nerve horizontally from the nasal to the temporal retina (Figure 10B), as described in normal mice [33,35].

The population of RGCs decreased slightly, with lower numbers present at 180 than at 90 days ALE. However, this diminution was not significant (Mann–Whitney, p>0.05; Figure 9 and Figure 10C,E). In addition, the distribution of the FG-traced RGCs in light-exposed mice was similar to the distribution observed in control retinas (Figure 10D,F). In conclusion, although the number of FG-traced RGCs decreased with time, there was not a significant RGC loss during the first 180 days ALE.