Small interfering RNA-mediated suppression of Ccl2 in Müller cells attenuates microglial recruitment and photoreceptor death following retinal degeneration

The study was approved by the Animal Experimentation Ethics Committee (AEEC) of the Australian National University (R.BSB.05.10). All experiments conducted were in accordance with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research.

Adult Sprague–Dawley (SD) rats aged between postnatal days 160 and 190 were used for the experiments. The rats were born and reared in dim cyclic light conditions with an ambient level of approximately 5 lux, until the commencement of bright-light exposure.

RNAi was achieved using a cocktail of two commercially available modified siRNAs, specific for Ccl2 (Stealth siRNA; #RSS302703 and #RSS302704; Invitrogen Inc., Carlsbad, CA USA). A scrambled siRNA equivalent, not homologous to any known gene, served as a negative control, and was conjugated to an Alexa 555 fluorophore to assess the uptake of siRNA in the retina (#14750–100; Invitrogen Inc.). Before administration, siRNAs were encapsulated using a cationic liposome-based formulation (Invivofectamine; #1377–901; Invitrogen Inc.) in accordance with the manufacturer’s instructions. The final concentration of each encapsulated siRNA formulation was 0.33 μg/μl in a 5% glucose solution.

For intravitreal injections, animals were anaesthetized with an intraperitoneal injection containing 30 mg ketamine (100 mg/ml; Troy Laboratories, NSW, Australia) and 3 mg xylazil (20 mg/ml; Parnell, NSW, Australia). A drop of 1% atropine (Chauvin Pharmaceuticals, London, England) was applied to the ocular surface to produce mydriasis, and the injection site was then swabbed with 5% povidone iodine (Betadine; Faulding Pharmaceuticals, SA, Australia). Intravitreal injection was then performed as described previously [59]; 3 μL of either positive or negative siRNA complex, corresponding to 1 μg of siRNA, was injected into both eyes of each animal. For an additional control, 3 μLof transfection agent only was also injected into both eyes of some animals. After injection, neomycin ointment 5 mg/g (Amacin; Jurox, NSW, Australia) was applied to the injection site to prevent infection.

After intravitreal injections, the animals were immediately transferred to individual cages designed to allow light to enter unimpeded. BCL exposure was achieved using an 18 W fluorescent light source (Cool White; TFC, Taipei, Taiwan) positioned above the cages, , which was run from 11.00 to 24.00 hours, and kept at an intensity of approximately 1000 lux at the cage floor. Corneal hydration was maintained by application of a synthetic tear gel (GenTeal Gel; Novartis, NSW, Australia) during BCL, until the animals awoke. Animals were exposed to BCL for 24 hours before tissue collection.

Animals were killed using an overdose (60 mg/kg bodyweight) of barbiturate (Valabarb; Virbac, Australia) given as intraperitoneal injection, then retinal tissue was obtained from each treatment group for analysis. Eyes from some animals were marked at the superior surface for orientation, then enucleated and processed for sectioning on a cryostat, while the retina from others was excised through a corneal incision and prepared for RNA extraction.

Eyes for sectioning were immediately immersion-fixed in 4% paraformaldehyde in 0.1 M PBS (pH 7.3) for 3 hours at room temperature, then processed as previously described [56], and sectioned at 16 μm on a cryostat. Retinas for RNA extraction were immediately immersed in chilled solution (RNAlater; #7024; Ambion Inc., Austin, TX, USA), then stored in accordance with the manufacturer’s instructions. The RNA was then extracted from each sample and analyzed as previously described [55, 60].

First-strand cDNA synthesis was performed as described previously [55]. Gene amplification was measured using commercially available hydrolysis probes (TaqMan®; Applied Biosystems, Foster City, CA, USA) (Table 1). The hydrolysis probes were used in accordance with a previously described quantitative (q)PCR protocol [55]. The fold change was analyzed using the ^ΔΔC_q method, with expression of the target gene normalized relative to the expression of two reference genes: glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and β-actin. Amplification specificity was assessed using gel electrophoresis.

To investigate localization of Ccl2 mRNA transcripts in the retina following RNAi, a riboprobe to Ccl2 was generated for in situ hybridization on frozen sections of retinal tissue. Synthesis of the Ccl2 riboprobe and in situ hybridization were performed as described previously [55, 61]. The Ccl2 riboprobe was hybridized overnight at 55°C, and then washed in saline sodium citrate (pH 7.4) at 60°C. After hybridization, some sections were further stained using immunohistochemistry (see below).

Following BCL, terminal dUTP nick end labeling (TUNEL) was used to quantify photoreceptor apoptosis in cryostat sections for each treatment group, using a previously published protocol [62]. Counts of TUNEL-positive cells in the outer nuclear layer (ONL) were carried out along the full length of retinal sections cut in the parasagittal plane (superio-inferior), within the vertical meridian. The total count from each retina is the average of four sections at comparable locations.

Frozen sections from each treatment group were used for immunohistochemical analysis, using the primary antibodies listed in Table 2. Immunohistochemistry was performed as previously described [55]. Immunofluorescence was viewed using a laser scanning microscope (Carl Zeiss, Jena, Germany), and acquired using PASCAL software (version 4.0; Carl Zeiss). Images were prepared for publication using Adobe Photoshop software.

Monocyte/microglia counts were performed on sections immunolabeled jointly with the markers ED1 and ionized calcium binding adaptor (IBA)1. Numbers of ED1+/IBA + and ED1−/IBA1+ nuclei were assessed long the full length of retinal sections cut in the parasagittal plane (supero-inferior) within the vertical meridian. Counts were made of all ED1+/IBA + monocytes throughout the retina, including the retinal vasculature, ONL, and choroidal vasculature. Counts of /ED1−/IBA1+ parenchymal microglia encompassed those in the outer plexiform layer (OPL) and ONL/subretinal space (but not the resting population in the inner plexiform layer; IPL), as microglia recruit to these areas when activated during retinal degeneration [9, 47, 63]. The total counts of ED1+/IBA + and ED1−/IBA1+ nuclei from each retina was the average of four sections at comparable locations.

Ccl2 expression following RNAi in Müller cells was assessed on frozen sections after either immunohistochemistry or in situ hybridization for Ccl2 (as described above). In Ccl2-immunolabeled sections, the number of Ccl2-immunoreactive Müller cell processes was assessed. In sections used for in situ hybridization, counts were made of Ccl2-expressing Müller cell bodies. Both sets of counts were conducted across the full length of retinal sections cut in the parasagittal plane (supero-inferior) within the vertical meridian; the total count was the average of four sections at comparable locations.

Statistical analysis for each experiment was performed using one-way ANOVA with Tukey’s multiple comparison post hoc test. For each analysis, P < 0.05 was considered significant.

To assess the efficacy of the siRNA transfection protocol, animals reared in dim light conditions were injected intravitreally with siRNA tagged with Alexa 555 to determine the cellular uptake of siRNA in the retina (Figure 1). At 24 hours after injection of the fluorophore-tagged siRNA, fluorescence for siRNA was visible deep within the retinal cellular layers, including transfection in the ganglion cell layer (GCL), inner nuclear layer (INL), and ONL (Figure 1B). Control animals who had not been injected with the fluorophore-tagged siRNA had no comparative fluorescence (Figure 1A). Using fluorescent markers, the transfected siRNA also showed colocalization with vimentin-immunoreactive Müller cell processes within the INL (Figure 1G-I; arrows), ONL, and outer limiting membrane (Figure 1C-E; arrows).

Retinal expression of Ccl2 following Ccl2 siRNA injection was assessed using qPCR (Figure 2). In animals injected with Ccl2 siRNA, expression of Ccl2 decreased significantly to 29.3% (P < 0.05; ANOVA/Tukey’s test) of that in retinas injected with Invivofectamine after 24 hours of BCL. Expression of Ccl2 in retinas injected with scrambled siRNA did not change appreciably, remaining at 95.4% if of that of the Invivofectamine-only controls (95.4%, P > 0.05; ANOVA/Tukey’s test).

Localization of Ccl2 mRNA and protein following 24 hours of BCL was assessed in retinas using in situ hybridization (Figure 3) and immunoreactivity (IR) (Figure 4) respectively. The distribution of both Ccl2 mRNA and protein following BCL showed strong colocalization for, respectively, vimentin-immunoreactive (Figure 3E-G, arrows) and S100B-immunoreactive (Figure 4E-G, arrows) Müller cell processes, consistent with our previous investigation [55]. After Ccl2 siRNA injection, the number of Müller cells expressing Ccl2 mRNA decreased significantly to 12.6 per retina, compared with 47.4 per retina in Invivofectamine-only controls (P < 0.05, ANOVA-Tukey’s test) (Figure 3). Scrambled siRNA-injected retinas showed no significant change in the number of Ccl2-expressing Müller cells, compared with Invivofectamine-only controls (50.1 per retina, P > 0.05; ANOVA/Tukey’s test). IR for Ccl2 protein (Figure 4 histogram) showed a significant reduction in the number of Ccl2-IR Müller cell processes in retinas injected with Ccl2 siRNA (11.6 per retina, P < 0.05; ANOVA/Tukey’s test) compared with retinas treated with Invivofectamine only or with scrambled siRNA (45.1 and 45.2 per retina, respectively).

The recruitment of monocytes/microglia in the retina following siRNA injections was assessed using immunolabeling for IBA1 and ED1 markers (Figures 5 and 6), which identify both monocytes (ED1+/IBA1+) and ramified microglia (ED1−/IBA1+) [64, 65]. After BCL exposure, ED1+/IBA1+ nuclei (Figure 5A-C) were recruited to the retinal and choroidal vasculature (as described previously [55]). These nuclei were reduced significantly in total counts to 33.1 per retina in the Ccl2 siRNA group (P < 0.05; ANOVA-Tukey’s), compared with 57.1 and 55.6 per retina in the Invivofectamine-only and the scrambled siRNA groups respectively (Figure 5G). The numbers of recruited ED1+/IBA1+ cells in different locations (retinal vasculature, choroid) are shown in Figure 5H. These counts show a significant reduction in the recruitment of ED1+/IBA1+ nuclei to both the choroid and retinal vasculature in Ccl2 siRNA treated animals, compared with controls (P < 0.05; ANOVA-Tukey’s). ED1−/IBA1+ nuclei were recruited to the ONL and outer plexiform layer (OPL) after 24 hours BCL (Figure 6A-C). Following Ccl2 siRNA injection, the total number of these nuclei was found to decrease significantly to 34.6 per retina (P < 0.05; ANOVA-Tukey’s) in comparison to those injected with either Invivofectamine only or scrambled siRNA (67.3 and 64.3 per retina respectively, Figure 6G). The recruitment of ED1−/IBA1+ nuclei to both locations (OPL, and ONL/subretinal space) was reduced in siRNA-injected animals, compared with controls (P < 0.05; ANOVA/Tukey’s test, Figure 6B).

There was no significant change in the number of TUNEL-positive photoreceptors (Figure 7A-E,) seen throughout the retina following BCL between animals intravitreally injected with either Invivofectamine only or scrambled siRNA (263.8 and 250.1 per retina, P > 0.05; ANOVA/Tukey’s test). However, in animals injected with Ccl2 siRNA, a marked decrease in the number of TUNEL-positive photoreceptors (to 85.1 per retina) was seen after BCL compared with both the Invivofectamine-only group and the scrambled siRNA control group (P < 0.05; ANOVA/Tukey’s test). In conjunction, expression of the apoptosis-related gene Jun (activator protein-1) [66] following BCL (Figure 7B) was markedly reduced in animals injected with Ccl2 siRNA, compared with both control groups (P < 0.05; ANOVA/Tukey’s test).

Exposure to bright continuous light in rats has been used to model retinal degeneration for over 40 years [83, 84]. Several lines of evidence also indicate that light damage is a useful model of AMD [56, 85‐87]. This model, like the established laser-induced model of neovascular AMD, uses an acute damaging stimulus to evoke site-specific AMD-like retinal degeneration. Although the rat retina lacks a macula and fovea centralis, it includes an homologous feature, the area centralis, in superiotemporal retina [88‐90]. Previous studies have identified the focal degeneration of photoreceptors and RPE cells and associated changes to the blood–retinal barrier as being localized to the area centralis, thus mimicking many of the histopathological aspects of advanced ‘dry’ AMD [56, 85‐87].

Recruitment of monocytes/microglia has been associated with the progression and severity of AMD pathology for many years [2, 12‐15], while several investigations have shown that microglial attenuation reduces lesion size in the laser-induced model of neovascular ‘wet’ AMD [91‐93]. Retinas from human donors show increased expression for Ccl2 in all forms of AMD [94], while increased levels of Ccl2 protein have been detected in aqueous humor samples taken from patients in advanced stages of ‘wet’ and ‘dry’ AMD [95, 96]. Increased Cc2 expression has also been described in the retinas of aged (20-month-old) mice, compared with young (3-month-old) mice [97]. Moreover, studies in experimental laser-induced choroidal neovascularization (CNV) have shown that ablation of either Ccl2 or the receptor Ccr2 inhibits the infiltration of monocytes/microglia and reduces lesion size following CNV [98, 99]. Conversely, it has been previously suggested that aging Ccl2⁻/⁻ Ccr2⁻/⁻ mice develop AMD-like retinal degeneration [100, 101], indicating that a degree of Ccl2 signaling is also required for homeostasis, although the AMD-like phenotype in the knockout has been questioned [98].

siRNA-mediated gene therapy is considered to have therapeutic potential in knocking down deleterious genes in various human pathologies (reviewed in [102, 103]). Our investigation is the first to show that monocyte recruitment, and in turn photoreceptor death, may be modified in the retina by siRNA-mediated suppression of Ccl2 in vivo in the CNS. Previous studies in AMD have shown that intravitreally injected siRNA targeting vascular endothelia growth factor ameliorates retinal degeneration in experimental CNV [104, 105], and has also been the basis for several clinical trials [106]. However, unlike the current investigation, these early studies used ‘naked’ unmodified siRNA molecules, which are now known to produce non-specific effects via Toll-like receptor 3 signaling in the retina [106]. Nevertheless, modulation of Ccl2 expression using appropriately targeted RNAi may provide a powerful means to control excessive microglial recruitment and activation in retinal dystrophies such as AMD.