Modulation of GSK-3 provides cellular and functional neuroprotection in the rd10 mouse model of retinitis pigmentosa

The rd10 mouse model of retinal neurodegeneration carries a homozygous phosphodiesterase 6b mutation (Pde6b ^rd10/rd10 ) on a C57BL/6 J background. Mice were kindly provided by Bo Chang from The Jackson Laboratory (Bar Harbor, ME, USA). Wild type (WT) control mice of the same background were also obtained from The Jackson Laboratory. All animals were housed and handled in accordance with the ARVO statement for the Use of Animals in Ophthalmic and Vision Research, European Union guidelines, and those of the local ethics committees of the CSIC and the Comunidad de Madrid. Mice were bred at the CIB core facilities on a 12/12-h light/dark cycle.

VP3.15 was synthesized in our laboratory as previously described [11]. This small molecule is a member of the iminothiadiazole family, the first group of substrate-competitive GSK-3 inhibitors described ([11]; see Additional file 1: Figure S1 for the structure data). The dose administrated was selected based on pharmacokinetic and IC₅₀ data ([11‐13] and data not shown). Mice received daily intraperitoneal injections of vehicle (10% DMSO, 0.9% NaCl) or VP3.15 (10 mg/kg) for the indicated period of time.

IP administration was selected based on the ability of VP3.15 to cross the brain blood barrier [13] and the necessity of performing repetitive administration without damaging the eye.

Mice underwent electroretinographic recordings (ERG) at different time points, using a previously described ERG protocol [14]. ERGs were performed by an observer blind to treatment. ERG signals were amplified and band filtered between 0.3 Hz and 1000 Hz (CP511 Preamplifier, Grass Instruments, Quincy, MA, USA) and digitized at 10 kHz using a PowerLab acquisition data card (AD Instruments Ltd., Oxfordshire, UK). Graphical representations of the signals recorded and the control light stimuli were generated using Scope v6.4 PowerLab software. ERG wave amplitudes were measured off-line and the results averaged.

Mice were euthanized at the indicated ages and their eyes enucleated. The right eye was processed for histological analyses and the left eye for retinal RNA extraction. For histological analyses, eyes were fixed for 1 h in 4% (w/v) paraformaldehyde in 0.1 M phosphate buffer (PB), pH 7.4, and then cryoprotected by incubation in increasing concentrations of sucrose (final concentration 50% (w/v) in PB). The eyes were then embedded in Tissue-Tek OCT (Sakura Finetec, Torrance, CA, USA) and frozen on isopropanol/dry ice. Cryostat equatorial sections (12 μm) were mounted on Superfrost® Plus slides (Thermo Scientific, Massachusetts, USA), dried at room temperature, and stored at − 20 °C. Before performing further analyses, sections were fixed in acetone for 10 min at − 20 °C and dried at 65 °C for 10 min. After rinsing in PBS and permeation with 1% (w/v) Triton X-100 in PBS, sections were blocked in BGT (2.5 g/L BSA, 100 mM glycine, 0.25% (w/v) Triton X-100 in PBS) for 1 h and then incubated overnight at 4 °C with primary antibodies (Table 1) diluted in BGT. Sections incubated in the absence of primary antibody were used as specificity controls. After rinsing in PBS and incubation with the appropriate secondary antibodies (Table 1), sections were stained with DAPI (4′,6-diamidino-2-phenylindole; Sigma-Aldrich Corp., St. Louis, MO, USA) and coverslipped with Fluoromont-G.

To assess preservation of the photoreceptor layer, we compared the thickness of the ONL (which primarily contains photoreceptors) with that of the corresponding INL (which contains bipolar, horizontal, and amacrine neurons and Müller glial cell bodies), and quantified the length of rod and cone outer segments (OS). Three sections per eye were analyzed: for each section, one image was captured for each of the 6 retinal zones (T1, T2, T3, T4, T5, and T6; Additional file 2: Figure S2). In each image, 3 measurements were recorded at random positions to obtain an average value per retinal zone per section. Measurements were performed using the “freehand line” and “measure” tools in Fiji software. The ONL thickness was normalized to that of the INL (not affected by the degeneration at this stage) to correct for possible inclinations of the sectioning plane.

Protein extraction was carried out by sonicating individual retinas in RIPA lysis buffer (containing 2 mM Na₃VO₄, 10 mM NaF, and 4 mM Na₄P₂O₇) and chilling the resulting solution on ice for 30 min. Protein (30 μg) from each sample was fractionated by electrophoresis on precast 10–12% (w/v) SDS-polyacrylamide gels (Criterion TGX, Bio-Rad, Munich, Germany), after which proteins were transferred to PVDF membranes using the Trans-Blot Turbo system (Bio-Rad). Blots were incubated overnight at 4 °C with primary antibodies (Table 1) diluted in TBS (Tris-buffered saline) containing 1% (w/v) Triton-X100, followed by incubation with the appropriate peroxidase-conjugated secondary antibody (Table 1). Proteins were visualized using the Pierce® ECL Western Blotting Substrate (ThermoFisher Scientific, Waltham, MA, USA), and quantified using ChemiDoc™ Touch Imaging System (Bio-Rad).

Total RNA from individual retinas was extracted using TRIzol Reagent, and 2.5 μg of RNA was typically reverse transcribed using the Superscript III Kit and random primers (all from ThermoFisher Scientific). Quantitative PCR (qPCR) was performed with the ABI Prism 7900HT Sequence Detection System using TaqMan Universal PCR Master Mix, no-AmpErase UNG, and Taqman assays (listed below) for detection (all from Thermo Fisher). The relative change in gene expression was calculated using the 2^ΔCt method, normalizing to expression levels of the Tbp (TATA-binding protein) gene. The primer-probe sets used are listed in Table 2.

Pathological neuroinflammation is a hallmark of neurodegeneration in many retinal dystrophies [2‐5]. We analyzed temporal alterations in pro-inflammatory gene expression in the rd10 retina between P14 (before the appearance of morphological signs of retinal degeneration) and P21 (at which stage the degenerative process has become clearly established) to identify the optimal period for therapeutic intervention. Quantitative PCR analysis of WT and rd10 retinas revealed increased expression of the pro-inflammatory markers α2m, Il1β and Tnfα in the degenerating retinas, beginning in the early stages of degeneration (P18) and increasing dramatically (by 25 to 50 fold) by P21 (Fig. 1). Reactive gliosis in response to retinal damage, as measured by upregulated Gfap expression, correlated with the increased expression of inflammatory markers. Likewise, the expression of the microglia genes Cd68 and Cd11b also showed a 4–5 fold increase (Fig. 1).

We next determined expression levels of GSK-3β, the most abundant GSK-3 isoform in the CNS [6], during the same degenerative period as above. We measured the expression of total GSK-3β and of its inactive Ser9-phosphorylated form (GSK-3β^Ser9) in rd10 mouse retinas. No significant differences in GSK-3β RNA levels were observed between rd10 and WT retinas (Fig. 2a). Furthermore, despite the morphological differences caused by the degenerative process, we observed similarly broad distribution patterns for GSK-3β protein in both P21 rd10 and WT retinas (Fig. 2b).

GSK-3β is an unusual kinase in that it is constitutively activated but is inhibited upon stimulation of its regulatory signaling pathways (reviewed in [7]). Activation of the PI3K/Akt pathway induces phosphorylation of GSK-3β at Ser9, and subsequent downregulation of GSK-3β activity [15]. During early retinal degeneration (P16 and P18), no differences in Ser⁴⁷³-phosphorylated Akt (pAkt^Ser473) or Ser⁹-phosphorylated GSK-3β (pGSK-3β^Ser9) levels were observed between WT and rd10 retinas (data not shown). Conversely, at P19 and P21, at which point the degenerative process is well established, as evidenced by decreased expression of the photoreceptor protein recoverin, pAkt^Ser473 and pGSK-3β^Ser9 levels were significantly higher in rd10 versus WT retinas (Fig. 3a and b). These observations in retinal extracts were in agreement with the immunostaining pattern observed at P21, which showed an increment in pAkt^Ser473 and pGSK-3β^Ser9 in the rd10 retina (Fig. 3c). This inhibitory phosphorylation of GSK-3β, which coincided with activation of the pro-survival Akt signaling pathway, may represent an intrinsic neuroprotective response to the genetic damage caused in the retina. To test this hypothesis, we sought to potentiate this intrinsic response by pharmacologically inhibiting GSK-3 activity.

In light of our findings concerning proinflammatory gene expression and GSK-3β modulation, we selected the P15–P21 interval to assess the effects of short-term treatment with VP3.15, a GSK-3 inhibitor. VP3.15 inhibitory activity on GSK-3 was confirmed by determining its effect on the β-catenin levels in the microglia cell line N9 (Additional file 3: Figure S3). rd10 littermates received daily intraperitoneal injections of either vehicle or VP3.15 (10 mg/kg). One eye from each animal was analyzed for gene expression and the other for retinal morphology. VP3.15 treatment significantly decreased the expression of the proinflammatory genes Il1β and α2m. A similar trend was observed for Tnfα expression (Fig. 4a), and secretion by rd10 retinal explants (Additional file 7: Figure S7b), although these failed to reach statistical significance. The microglia genes Cd11b and Iba1 were also decreased upon VP3.15 treatment, while the activated microglia marker Cd68 showed a decline trend (Fig. 4b). In parallel, VP3.15 significantly reduced the expression of Gfap, an effect that correlates with a reduced GFAP staining depicted by the radial process of Müller glial cells and by the most inner retinal layer formed by Müller end feet and astrocytes (Fig. 4c and d). Conversely, a trend towards increased gene expression of rod photoreceptor-specific rhodopsin was observed in VP3.15-treated rd10 mice (Fig. 4e). This observation correlated with the maintenance of ONL thickness and OS length, both of which were better preserved in VP3.15-treated than in the vehicle-treated rd10 mice (Fig. 4f). Moreover, while rhodopsin expression was specifically localized in the photoreceptor outer segment (OS) in WT and VP3.15-treated rd10 retinas, vehicle-treated rd10 retinas displayed shortening of the OS and mislocalization of rhodopsin in the ONL (Fig. 4f).

Based on the promising results obtained following short-term GSK-3 inhibition, we investigated whether a longer treatment period preserved visual function in the rd10 mouse model. rd10 littermates received daily injections of either vehicle or VP3.15 from P15 to P46 inclusive. Visual function was assessed weekly by ERG between P25 and P46. Compared with vehicle-treated counterparts, VP3.15-treated rd10 mice displayed better-defined ERG waves of greater amplitude (Fig. 5 and Additional file 4: Figure S4). The amplitude of ERG waves (a-mixed, b-mixed, b-photopic, and oscillatory potential [OP]) was significantly higher in VP3.15-treated than in vehicle-treated mice, indicating a prolongation of both rod- and cone-mediated vision function by VP3.15 (Fig. 5).

Histological evaluation of retinas at the end point of the study (P47, after the last ERG) revealed a greater relative ONL thickness in VP3.15- versus vehicle-treated retinas despite the advanced degeneration stage (Fig. 6a and b). The photoreceptor preservation effect of VP3.15 was more evident at mid-degeneration stages (P33; Additional file 5: Figure S5). Moreover, specific immunostaining showed better preservation of photoreceptor structure in VP3.15-treated retinas (Fig. 6a). Rhodopsin immunostaining in rod outer segments confirmed the partial preservation of rod outer segments in VP3.15-treated retinas. By contrast, rod outer segments were barely present in vehicle-treated retinas. Similarly, M-L opsin immunostaining revealed longer cone outer segments in VP3.15- versus vehicle-treated rd10 retinas (Fig. 6a and b).

Moreover, no toxic effect of VP3.15 administration was observed in other neuronal populations of the retina, particularly rod-bipolar and ganglion cells (Additional file 6: Figure S6).

Taken together, these results demonstrate a neuroprotective effect of VP3.15 on neuroinflammation, retinal structure, and visual function. Our findings thus support the potential of VP3.15 as a novel therapy for retinal dystrophies.