We used 57 BALB/c (albino) and 70 C57BL/6J (pigmented) male mice aged 7 to 8 weeks. Experiments were conducted in compliance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Hebrew University animal ethics committee. Animals were maintained on a 12-hour dark/light cycle, with unlimited food and water. For electroretinography and optical coherence tomography (OCT) imaging, mice were anesthetized by intraperitoneal injections of ketamine (Parke–Davis, London, UK) in combination with the relaxing agent xylazine (VMD n.v./s.a., Arendonk, Belgium). 

We freshly prepared an SI solution just prior to IP injection from solid NaIO₃ (Sigma-Aldrich Corp., St. Louis, MO, USA) diluted in 0.9% sterile sodium chloride for injection. We injected 25, 50, or 100 mg/kg SI once intraperitoneally. Retinal function and structure was assessed before SI administration (baseline) and 1 day, 3 days, and 1, 2, 3 and 4 weeks postinjection. 

Visual acuity was measured using an optokinetic testing apparatus (OptoMotry; Cerebral Mechanics, Inc., Lethbridge, AB, Canada) by recording the tracking response (optokinetic reflex) to a rotating visual stimulus displayed on four LCD panels surrounding the mouse. Visual acuity was measured at 100% contrast. 

Full-field ERGs were recorded in anesthetized animals after overnight dark adaptation using a ganzfeld stimulator (ColorDome; Diagnosys LLC, Littleton, MA, USA) and a computerized system (Espion E²; Diagnosys LLC). We recorded ERGs inside a Faraday cage, under dim red lighting. Pupils were dilated with 1% tropicamide and 2.5% phenylephrine. Drops of 0.4% benoxinate HCl (Fisher Pharmaceuticals Ltd., Tel-Aviv, Israel) were topically administered for local anesthesia prior to applying gold-wire active electrodes on the corneas. A reference electrode was placed on the tongue and a needle ground electrode was placed intramuscularly in the hip area. Dark-adapted rod and mixed rod-cone responses as well as light-adapted (with 30 cd/m² white background light) 1- and 16-Hz cone responses to white flashes of increasing intensities were recorded. In the dark-adapted state, flash intensities ranging between 0.000006 and 10 cd*s/m² were used; in the light-adapted state, intensities ranged from 0.63 to 20 cd*s/m². Electroretinographic responses were filtered with a bandpass of 0.3 to 500 Hz and signal averaging was used. 

In vivo microscopy of the retina was performed using an OCT system (Spectralis; Heidelberg Engineering, Inc., Carlsbad, CA, USA). All images were obtained from the left eye. Retinal cross-sections were obtained along horizontal and vertical meridian, centered on the optic nerve. In addition, fundus autofluorescence (FAF) images were taken (excitation wavelength: 488 nm, emission wavelength >500 nm). 

Mice were euthanized under anesthesia by cervical dislocation and eyes were enucleated, fixed at room temperature in Davidson solution (glacial acetic acid, 95% alcohol, 16% paraformaldehyde, and distilled water in a ratio of 4:12:5:15, respectively) for 24 hours, and embedded in paraffin (Paraplast Plus; Leica Biosystems Nussloch GmbH, Nußloch, Germany). We cut 5-μm thick retinal sections in the nasotemporal plane through the optic nerve and the center of the cornea. For descriptive histology and quantitative analysis, sections were stained with hematoxylin and eosin. All observations and photography were performed using a fluorescent microscope (BX41; Olympus Corp., Tokyo, Japan) equipped with a digital camera (DP70; Olympus Corp.). Image processing and quantification were performed using a raster graphics editor (Photoshop CS3; Adobe Systems, Mountain View, CA, USA) and image analysis software (ImagePro Plus; Media Cybernetics, Inc., Rockville, MD, USA). 

In order to standardize the measurements, only sections crossing the centers of the optic nerve and the cornea were used. The thickness of the retinal layers was measured at 150 μm intervals nasally and temporally from the center of the optic nerve. 

Deparaffinized sections were incubated in a decloaking chamber (Biocare Medical, Pacheco, CA, USA) with 10 mM citrate buffer (pH 6.0) at 125°C, blocked with PBS solution containing 1% bovine serum albumin, 0.1% Triton X-100, and 10% normal donkey serum, and subsequently incubated for 1 hour with one of the following primary antibodies: anti-rhodopsin (mouse monoclonal, 1:100; Lab Vision Corp., Fremont, CA, USA); anti-blue–sensitive opsin (goat polyclonal, 1:75; Santa Cruz Biotechnology, Inc., Dallas, TX, USA); anti-red/green opsin (rabbit polyclonal, 1:100; Chemicon International, Inc., Billerica, MA, USA); anti-RPE65 (mouse monoclonal, 1:10; Novus Biological, Littleton, CO, USA); or anti–zonula occludens (ZO)-1 (rabbit polyclonal, 1:10; Zymed Laboratories, Inc., San Francisco, CA, USA). After washing in PBS, specimens were incubated for 1 hour at room temperature with one of the following secondary antibodies: DyLight 488 donkey anti-rabbit; DyLight 549 donkey anti-mouse; DyLight 649 donkey anti-mouse; or rhodamine Red-X-conjugated donkey anti-goat IgG (1:250; all from Jackson ImmunoResearch Laboratories, West Grove, PA, USA). Nuclei were counterstained with 4,6-diamidino-2-phenylindole–containing mounting medium (Vector Laboratories, Inc., Burlingame, CA, USA). To determine the specificity of the antigen–antibody reaction, corresponding negative controls with the secondary antibody alone were performed. 

A wide range of SI concentrations are used today by investigators for induction of retinal degeneration in different experimental animals. To optimize the dose of SI, we conducted preliminary experiments in which three groups of mice were treated with a single intraperitoneal injection of 25, 50, and 100 mg/kg of SI. All animals were then tested by ERG and evaluated histologically. The lowest concentration of 25 mg/kg SI showed inconsistent results that varied from near normal ERG responses and histology to severe impairment of retinal function and structure. By contrast, a dose of 100 mg/kg led to systemic toxicity with high mortality of experimental animals. At an intermediate dose of 50 mg/kg RPE and retinal injury seemed more consistent, and no significant systemic toxicity was observed, with no mortality and with the animals maintaining stable body weight (Supplementary Fig. S1). Therefore, this dose was deemed optimal and was used in the experiments described below. 

Animals were serially tested by OKT under controlled lighting conditions and at identical times of the day. In C57Bl/6J mice, baseline visual acuity was 0.332 ± 0.013 cyc/deg. Twenty four hours following a single injection of SI at a dose of 50 mg/kg, a dramatic decrease in visual acuity was observed, to 0.024 ± 0.008 cyc/deg. A very similar level was measured at 72 hours postinjection. Interestingly, some restoration of visual function was observed by the end of the first week (0.152 ± 0.038 cyc/deg). However, already by the end of the second week, visual acuity declined again and no response was detectable 3 and 4 weeks postinjection (Fig. 1). In albino BALB/c mice, spatial visual acuity was under the limit of detection already at baseline, even before exposure to SI, so this functional measure could not be followed in this strain (similar absent or severely reduced OKT responses in albino mice have been previously reported by others).^30–32 

Rod and mixed rod-cone function was assessed by full field ERG following overnight dark adaptation. At baseline, BALB/c mice were characterized by lower b- and a-wave amplitudes as compared with C57Bl/6J age-matched mice (e.g., at the highest dark-adapted stimulus intensity of 10 cd*s/m², mixed–rod cone b-wave amplitude was 436 ± 20 μV in BALB/c mice versus 609 ± 45 μV in C57Bl/6J and a-wave amplitude was 161 ± 8 μV versus 333 ± 30 μV; mean ± SEM, P < 0.00001, Fig. 2). Twenty-four hours following a single injection of 50 mg/kg of SI, rod-dominated ERG responses in BALB/c mice were reduced by more than 50% as compared to baseline. Dark-adapted responses in C57Bl/6J mice were better preserved. Moreover, b-wave amplitudes in C57Bl/6J mice were not statistically different from the baseline level at highest stimulus intensities. At this time point, cone-dominated responses in BALB/c were not different from the baseline at most stimulus intensities. Unexpectedly, C57Bl/6J mice showed significantly higher light-adapted 1-Hz b-wave amplitudes as compared with baseline (293 ± 35 μV versus 106 ± 15 μV, respectively, Fig. 2). However, 3 days following SI administration, rod- as well as cone-dominated ERG responses were significantly decreased in both strains as compared with baseline and with the 24-hour time point. Some restoration of dark-adapted responses was observed in BALB/c mice by 1 week postinjection while amplitudes in C57Bl/6J remained at reduced levels. In both strains, dark-adapted ERG in response to low stimulus intensities was completely abolished by 2 weeks postinjection and progressively decreased at higher stimulus intensities. At the 4-week time point, the a-wave was practically abolished, whereas a residual b-wave was recordable merely when the highest intensity stimuli were applied. Some restoration of cone-dominated ERG responses was also observed at 2 weeks postinjection in BALB/c and at 2 to 3 weeks in C57Bl/6J mice. However, at 4 weeks postinjection, light-adapted ERG responses were practically undetectable in both strains. In both albino and pigmented mice, prolonged implicit time of rod- and cone-mediated responses was observed following exposure to SI, more prominent in the C57Bl/6J strain (Fig. 3). 

In vivo visualization of the retina was obtained using a scanning laser ophthalmoscopy–based imaging system. Representative optical coherence tomography images of the central retina as well as FAF images of the fundus over time are summarized in Figure 4. Images are presented in grayscale, consisting of darker layers containing mainly nuclei, and of brighter layers consisting mainly of nerve fibers, plexiform layers, and other highly reflective sections.³³ 

At baseline, well-defined retinal layers are visible (Figs. 4A1, 4A2). As early as 24 hours postinjection, a slight altering of reflectivity and moderate swelling of the retina was observed (Figs. 4B1, 4B2). At 72 hours, OCT images demonstrate a blurring of the boundaries between layers and a continued increase in reflectivity, possibly as a result of worsening edema (Figs. 4C1, 4C2). This apparent change in the retina can be correlated with the sharp decrease in ERG signal at this time point (Fig. 2), accompanied by poor visual function as observed in the OKT testing (Fig. 1). At 1-week postinjection, the overall reflectivity of the retina is mostly restored to original levels while highly reflective matter accumulates beneath outer layers (Figs. 4D1, 4D2). This restoration can also be correlated with concurrent improvement in both dark-adapted ERG signals (Fig. 2) and OKT response (Fig. 1). Starting from the 1-week time point, gradual thinning of the outer layers is apparent on OCT, suggesting loss of photoreceptors over time (Figs. 4D1–G1, 4D2–G2). A similar sequence of events as visualized by OCT was also described in previous studies using IV sodium iodate at doses above 30 mg/kg in both mice and rat models.^16,28,34 

Fundus autofluorescence images display increasing irregularity over time, with formation of patchy hyper- and hypofluorescent spots. This may represent areas of photoreceptor debris accumulation and atrophy, respectively. These changes were more prominent in BALB/c mice (Figs. 4H1–N1) as compared with C57Bl/6J mice (Figs. 4H2–N2), possibly related to the melanin present in C57Bl/6J mice. Recently, the cumulative size of abnormal retinal FAF in SI-exposed albino rats was shown to correlate with the SI dose, and inversely correlate with ERG b-wave amplitudes.³⁵ In BALB/c mice, a few hyperfluorescent spots were already apparent at baseline (Fig. H1). This finding was described in previous studies and can possibly be explained by BALB/c susceptibility to photoreceptor-cell apoptosis induced by light, regardless of sodium iodate injection.^36–38 

The effects of 50 mg/kg SI on the retinal structure were assessed using descriptive histology as well as a morphometric approach. Representative images from histologic sections stained with hematoxylin and eosin are summarized in Figure 5. Baseline retinal histology (before SI administration) was characterized by well-organized and clearly defined retinal layers including the outer nuclear layer (ONL), photoreceptor inner segment (IS), outer segment (OS), and RPE monolayer (Figs. 5A, 5H). Twenty four hours after SI injection, swelling of OS and IS complicates differentiation of these sublayers. In albino, BALB/c retina, the RPE layer was already poorly distinguishable at this time point, and RPE cell nuclei largely disappeared (Fig. 5B). In the retina of C57Bl/6J mice, a decrease of melanin granules was observed (Fig. 5I). At 3 days, disorganization and loss of photoreceptors took place in both strains, accompanied by cellular infiltration in the IS/OS region (Figs. 5C, 5J). Local migration of melanin into OS was seen in C57Bl/6J retinas. At 1, 2, and 3 weeks worsening of the toxic injury caused increased thinning of the outer retina, (Figs. 5D–F, 5K–M). The number of photoreceptor nuclei in the ONL dropped continuously, with loss of the orderly linear structure of this layer and formation of hemi-rosettes and invaginations. Multiple areas of photoreceptor and RPE loss were often observed and accumulation of pigment conglomerates accompanied these changes in C57Bl/6J retinas. Retinal degeneration was most pronounced 4 weeks following SI injection (Figs. 5G, 5N). At this time point, the RPE layer was largely absent and most photoreceptors and their OS were lost, correlating with the functional loss as measured by electroretinography. 

To quantify these histologic observations, measurements of retinal layers thickness were performed in a consistent and controlled fashion. For standardization, only sections crossing the centers of the cornea and the optic nerve were used. In general, changes were similar in both strains. Results given in Figure 6 show that thickness of the outer layers, including ONL and inner and outer segments of the photoreceptors, thinned significantly by 1 week and continued to decrease later, reaching a level of nearly half of the baseline pretreatment thickness by 4 weeks (Figs. 4C–F). The most dramatic changes were observed in the RPE layer. We calculated linear density of nuclei in RPE cells and found dramatic loss of nuclei soon after SI injection (Figs. 6A, 6B). In both strains, at 4 weeks, only scarce RPE nuclei could be identified, usually in the periphery. Interestingly, at baseline, RPE nuclei density was higher in BALB/c than in C57Bl/6J eyes. Full measurements of additional retinal layers are provided in Supplementary Figure S2. 

For further characterization of retinal changes, immunohistochemical staining was performed. Representative images are shown in Figure 7, and additional data are provided in Supplementary Figure S3. 

As mentioned above, retinal RPE cells are considered a primary target of SI toxicity. As one measure of RPE status and function, immunostaining of ZO-1 protein, a marker of tight junctions playing a crucial role in the formation of the blood–retinal barrier and RPE integrity, was performed. Naïve retinas showed strong ZO-1 expression in the outer limiting membrane (OLM) and in the RPE monolayer, concentrated at the apical cellular side and at cell-cell junctions. Already 24 hours following a single IP injection of SI, dramatic changes in ZO-1 expression were observed, and no signal was detected in RPE cells (Figs. 7A1–D2). Later, at the 1-week time point, anti–ZO-1 staining showed local disruptions of the OLM that further increased over time. An additional RPE-specific protein that plays a role in vitamin A metabolism, RPE65, was also markedly affected following exposure to SI: in normal retina, RPE65 was specifically and strongly expressed in the RPE cells with better visualization in BALB/c albino retinas (Figs. 7E1–H1, 7E2–H2). Twenty four hours after SI injection, RPE65 expression was significantly reduced in both BALB/c and C57Bl/6J retinas. At 2 and 3 weeks, RPE65 was observed only in limited retinal areas, almost completely disappearing by the third week of experiments. This signifies loss of RPE function, correlating with RPE cell loss. 

At baseline, in both strains, rhodopsin and cone opsins were properly localized to the photoreceptors outer segments (Figs. 7I1, 7I2, 7M1, 7M2, and Supplementary Figure S3: C1, D1, E1, H1, J1, K1). In general, changes in rhodopsin expression were very similar between strains, showing steady decrease of rhodopsin content progressing together with disorganization and destruction of rod outer segments (Figs. 7J–L). Some similarity was also observed in expression of red/green opsin. The level of staining progressively decreased from 24 hours to 1 week after SI injection when red/green opsin was only sporadically expressed. Surprisingly, at the 2-week time point, level of red/green opsin staining increased in C57Bl/6J retinas and much more significantly in BALB/c retinas with protein translocation to the synaptic terminals (Supplementary Fig. 3: D5, J5 compared to D4, J4). At 4 weeks, red/green opsin staining has largely disappeared in both strains (Supplementary Fig. 3: D7, J7). Blue opsin expression in the IS/OS layer decreased in BALB/c retinas 24 hours after injection, whereas its level was relatively preserved in C57Bl/6J (Figs. 7N1, 7N2). However, blue opsin mislocalization to the synaptic regions was already evident in C57Bl/6J mice at this time (Fig. 7N2). At 1 week, mislocalization and expression of blue opsin was quite prominent in both stains. This pattern was also observed in BALB/c at a later time and remained in both strains up to 4 weeks (Figs. 7O1, 7P1, 7N2–P2). 

In this study, we characterized the effect of a single intraperitoneal injection of 50 mg/kg sodium iodate on young adult albino and pigmented mice. Our observations indicated that SI injection induced extensive RPE, retinal, and possibly choroidal injury as assessed by both morphology and functional studies and was evident in in vivo and ex vivo assays. We also observed that retinal damage fluctuated extensively during the first week postinjection, but subsequently deteriorated steadily culminating in practically complete loss of function by week 4. These results were consistent in both pigmented and nonpigmented mice strains, and should be considered when using the SI model. 

Interestingly, a bimodal pattern in the OKT results following SI injection was observed, with initial severe loss, some recovery at 1 week postinjection, and then irreversible deterioration. Two mechanisms of damage may be associated with such a pattern. One may involve the observed rapid disorganization of RPE cells with complete loss of ZO-1 expression assessed by histology and immunohistochemistry, indicating disruption of tight junction structures. This damage may affect the function of the outer blood–retinal barrier between the fenestrated choriocapillaris and the photoreceptor layer and its ability to regulate and maintain normal ionic and metabolic gradients required for proper vision.³⁹ Disintegration and dysfunction of RPE cells decreases absorption of fluid from the subretinal space and may lead to enhanced diffusion of proteins across the paracellular space from the choriocapillaris into the subretinal space.⁴⁰ Notably, similar findings were recently observed in an SI-induced model in which higher doses and the IV administration mode were used and were also observed when mRNA levels were assessed.¹⁵ Previous studies in rabbits exposed to SI, using horseradish peroxidase as a tracer in one study and assessing leakage of gadolinium diethylenetriamine pentaacetic acid using MRI in another, also showed that there is marked breakdown of the diffusion barrier normally maintained by the RPE.^19,41 Recent studies in cell cultures of ARPE-19 further show that exposure to sublethal doses of SI retard RPE cell migration and cause reduced expression of the cell junction protein ZO-1.⁴² Taken together, these processes probably underlie the rapid formation of retinal edema following SI administration as detected by OCT and histology. The formation and resolution of such edema caused by the early toxicity of SI may explain the observed initial OKT loss followed by partial recovery at 1 week. Other investigators also noted initial loss of OKT function with partial recovery following IV administration of SI, although with somewhat different kinetics.^43,44 

An additional mechanism of injury, possibly explaining the later and irreversible loss of visual acuity as assessed by OKT, may be the loss of RPE function (as exemplified by the progressive decrease in RPE65 expression), leading to secondary rod photoreceptor degeneration with impairment of visual function, decreasing rhodopsin expression and thinning of the ONL over time. This phenomenon was also observed in several recent studies and was supported by mRNA data.^15,16 As noted above, we demonstrated improvement of OKT results 1 week post-SI administration. This was paralleled by enhanced blue cone opsin production as assessed by immunohistochemistry. In contrast, production of the second cone opsin—red/green—was dramatically depressed at this time point and increased at 2 weeks postinjection. However, cell distribution of both opsins was abnormal, showing protein expression in the cell bodies and even in the synaptic regions. Of relevance, mislocalization at the synaptic terminals occurs before a general redistribution of opsin around the entire cell. This phenomena occurs prior to cell death in many types of photoreceptor degeneration.⁴⁵ 

Despite the overall similarity and correlation between the different functional and structural measures assessed, at early stages we observed some discrepancy between OKT and ERG results. Whereas a transient improvement in the OKT results was apparent 1 week post-SI injection, a consistent deterioration of electroretinographic responses was evident. These observations may represent the early inconsistent nature of the model already noted by others.^25,44 Based on these observations, it appears that the results of experiments using the SI model are best interpreted at least 1 week after its induction. 

Several different methods of SI administration were used in previous studies. This fact may account for some of the discrepancies in outcome evident between different studies. The most common administration method used was the intravenous route. However, while this method provides immediate absorption of injected SI into the systemic circulation, it requires additional use of anesthesia and involves a high rate of technical failure, especially in small or neonatal animals.⁴⁶ Therefore, we elected to use the intraperitoneal administration method. Because the pharmacokinetic profile of IP injection was anticipated to differ from the IV route due to liver first pass effect, we performed a dose ranging experiment and finally used the 50 mg/kg dose, which allowed for a consistent retinal effect without evident signs of systemic toxicity.⁴⁷ 

In this study, we tested both C57Bl/6J (pigmented) and BALB/c (albino) mouse strains. These strains differ in many aspects, including visual function and immune response to pathologic states.⁴⁸ These include differences of susceptibility and maturation of regulatory and helper T cells, susceptibility to tumor induction, and autoimmune diseases.⁴⁹ While C57Bl/6J were used in most recent iodate studies, BALB/c mice were less studied. However, the use of BALB/c mice is beneficial in the study of therapeutic interventions such as RPE cell transplantation because of the readily detected pigmented RPE. Additionally, BALB/c mice are used as the background for immune-deficient mouse strains favored for transplantations. Therefore, we elected to further test the SI model in BALB/c mice as well, in order to allow their use in future studies. 

In summary, we have performed a thorough characterization of the SI model for the first time using the IP administration route in albino and pigmented mice. Our observations demonstrate consistency with results obtained by others using the IV administration. We also conclude that the SI model can be applied to BALB/c mice and that assessment of the model is best performed after the first week of SI administration. 

Supported by grants from the Foundation Fighting Blindness, the Moxie Foundation, and by the Yedidut 1 Research Grant. This study was performed as part of the requirements toward a doctor of medicine degree for Guy Chowers at the Hebrew University-Hadassah Medical School, Jerusalem, Israel. 

Disclosure: G. Chowers, None; M. Cohen, None; D. Marks-Ohana, None; S. Stika, None; A. Eijzenberg, None; E. Banin, None; A. Obolensky, None