Wild-type C57BL/6 (B6) mice (6-7 weeks old) were obtained from the National Laboratory Animal Center, Taiwan, and kept in the animal facilities at Chang Gung University. Animal care and experiments were approved by and performed in accordance with the Institutional Animal Care and Use Committee of Chang Gung University, Taiwan.

According to a previous study (12), the aEAU animal model was established on the day 0 by subcutaneous immunization of an emulsion containing 100 μl of human IRBP_1-20 peptide (2 μg/μl) in 100 μl of CFA (Sigma-Aldrich), distributed over 4 spots on the tail base to flank. A single dose (100 ng) of intraperitoneal injection of PTX was used as a costimulatory adjuvant. Age- and sex-matched B6 mice that did not undergo immunization were used as disease-free controls. aEAU mice were sacrificed on the 12th or 28th day after disease induction. The eyes and spleen were harvested, and TLR expression and immune responses were analyzed.

Nylon wool-purified splenic T cells were isolated from day-12 aEAU mice and cultured in the presence of IRBP_1-20 (10 μg/mL) for 2 days (14). The tEAU animal model was then established by intraperitoneal injection of IRBP-specific T cells (5×10⁶/mouse) into naïve 6-week-old B6 mice.

The method was performed according to previous studies with modifications (32–34). After enucleation of infant mice (10-14 days old) or adult EAU mice, the eyeballs were cut into halves along the circumferential line to create a ciliary body-free posterior eyecup. The retinal layer was removed from the eyecup, which was pretreated with trypsin, and RPE tissue was scraped (Supplementary Figure 1) and minced into a single-cell suspension in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, Thermo Fisher Scientific) containing 10% (v/v) fetal bovine serum (FBS) (Gibco).

Primary RPE cells were cultured in DMEM supplemented with 10% FBS, 100 U/mL penicillin (Gibco), and 100 µg/mL streptomycin (Gibco). Splenocytes and splenic T cells were cultured in minimum essential medium α (α-MEM) (Gibco) containing 10% FBS, 55 nM β-mercaptoethanol (Gibco), 100 U/mL penicillin, and 100 µg/mL streptomycin. ID8 cells were maintained in DMEM with 3% (v/v) FBS and 1× ITS (Sigma-Aldrich, Merck) containing 5 μg/mL insulin and transferrin and 5 ng/mL sodium selenite. The IRBP_1-20 (GPTHLFQPSLVLDMAKVLLD) peptide was synthesized by the Peptide Synthesis Core Facility of Academia Sinica, Taiwan.

Total RNA was extracted from cultured primary RPE cells or RPE cells that were freshly isolated from EAU mice by the TRIzol (Invitrogen, Thermo Fisher Scientific) method. Reverse transcription and real-time quantitative PCR were performed by an iScript cDNA Synthesis Kit and iQ SYBR Green Supermix with a CFX Connect Real-Time PCR Detection System (Bio-Rad Laboratories). A total of 10 targets were analyzed, and the primer sets used were published elsewhere (35–37). The production of the proinflammatory cytokine IL-6 by RPE cells was analyzed by a Ready-Set-Go! ELISA kit (eBioscience, Thermo Fisher Scientific). Specific signaling protein levels were identified by MILLIPLEX^® multiplex assays for Luminex (Millipore, Merck) according to the manufacturer’s instructions.

Primary RPE cells were seeded at a density of 7×10⁴ cells/well in ECIS 8W10E culture dishes. When the cell confluence reached 100% (≃16 hours), recombinant mouse IL-17 (PeproTech) or R848 was added to the RPE cultures. The level of impedance at low frequency (400 Hz) was monitored by an ECIS system (Applied BioPhysics), and the effects on barrier function during the entire culture period were determined according to the manufacturer’s instructions.

TLR7-specific shRNA (5’-GCCCTTTACCTGGATGGAAAC-3’) and scramble shRNA were subcloned into the pAAV-IRES-GFP plasmid (Stratagene) AAV vector and were named pAAV-shTLR7 and pAAV-scramble, respectively (Supplementary Figure 2). Recombinant AAV-8 virus production was performed with the AAV helper system by the Institute of Biomedical Sciences of Academia Sinica, Taiwan. Titers of rAAV-TLR7-shRNA (rAAV.shTLR7) and rAAV-scramble-shRNA (rAAV.sc) were determined by RT-qPCR analysis for green florescence protein (GFP) by calculating the viral genome copy number.

The TLR7 agonist R848 (4 μg/mouse) (Resiquimod; InvivoGen) and rAAV-TLR7-shRNA (4×10¹⁰ vg/mouse) were subretinally injected into both eyes of tEAU mice on day 0. tEAU mice treated with balanced salt solution (BSS) or rAAV-scramble shRNA served as controls. After 12 days of treatment, clinical disease severity was examined by fundoscopy as described previously (38, 39). Treated tEAU mice were sacrificed on the 28th day after disease induction. The eyes and spleen were harvested, and gene expression, histopathology and the immune response were analyzed in vitro.

From the 5th day to 28th day after disease induction, disease severity scoring was performed by fundoscopy every 3 weekdays; the final evaluation was conducted on the 28th day. After anesthetizing and dilating the pupil with proparacaine 0.5% and tropicamide 1% (ALCON), the clinical score of uveitis in each eye was graded based on the degree of cell infiltration as published elsewhere (38, 39). The severity score was graded from 0 to 4 as follows: 0, no change; 0.5 to 3, few focal lesions to a pattern of linear lesions, increasing vasculitis, neovascularization, retinal hemorrhage, and papilledema; and 4, large retinal detachment and retinal atrophy.

The eyeballs were collected and fixed in 3.7% formaldehyde on the 28th day after tEAU induction. The fixed tissues were embedded in paraffin, sectioned at a thickness of 5 μm, and stained with hematoxylin and eosin (H&E) (Sigma-Aldrich).

Splenocytes were obtained from untreated tEAU mice, and samples containing 10⁵ cells were labeled with anti-CD4-PerCP (BD Biosciences) and anti-CD25-APC (eBioscience) to examine surface marker expression. Intracellular Foxp3 analysis was performed using anti-Foxp3-PE (eBioscience) and the Foxp3/transcription factor staining buffer set (eBioscience) according to the manufacturer’s instructions. Flow cytometric analysis was performed by an Attune NxT flow cytometer (Thermo Fisher Scientific), and the data were analyzed by FlowJo software (Clever.ly, BD Biosciences).

Splenocytes were isolated from the treated and untreated tEAU mice and cultured in the presence of IRBP_1-20 (10 μg/mL) at a density of 4×10⁶ cells/well in 24-well plates for 4 days. The levels of the cytokines IFN-γ, IL-17, IL-10 and IL-6 were analyzed using Ready-Set-Go! ELISA kits (eBioscience, Thermo Fisher Scientific).

Nonparametric analyses, including the Mann-Whitney U test or Wilcoxon signed rank test, were performed by statistical tools in GraphPad Prism software. A statistically significant difference was defined as one with a p-value less than 0.05.

TLRs seem to be expressed on RPE cells, so we hypothesized that TLR signaling is associated with the development of uveitis. First, primary RPE cells were freshly isolated from EAU mice, and TLR expression was analyzed. To ensure the purity of RPE cells, the expression of RPE65 was examined in freshly obtained RPE cells. The real-time RT-PCR results show that rpe65 expression was significantly increased in freshly isolated RPE cells compared with the retina, which is the tissue adjacent to the retinal pigment epithelium (Figure 1A). The presence of RPE65 in these RPE cells was also confirmed by immunofluorescence analysis (Figure 1B).

Next, we examined which TLRs were involved in the development of EAU, and the expression levels of different TLRs were analyzed in RPE cells from EAU mice. Primary RPE cells were obtained from aEAU and tEAU animals and analyzed. Figure 1C shows significantly increased levels of tlr7 expression in both aEAU and tEAU mice. However, RPE cells from only aEAU animals exhibited upregulated tlr4 expression levels, indicating that the adjuvant in the aEAU induction protocol induced tlr4 expression. This result suggests that the induced expression of tlr7 but not tlr4 contributes directly to the induction or development of EAU.

To clarify whether TLR7 contributes to the development of autoimmune uveitis, the TLR7 agonist R848 was used to identify the effects of TLR7 on a mouse model of tEAU. On the day of tEAU induction, 6-week-old B6 mice received an intraperitoneal injection of IRBP-specific T cells (5×10⁶/mouse) and subretinal injection of R848 (4 µg/mouse). Healthy C57BL/6 mice without EAU induction were administered a subretinal injection of R848 and served as disease-free controls (data not shown). As shown in Figure 2, tEAU mice treated subretinally with R848, exhibited significantly higher tlr7 gene expression in RPE cells than tEAU mice that received BSS treatment (Figure 2A). Figure 2B shows the disease severity curves of the mice receiving R848 or BSS as controls, and the severity score was determined by fundoscopy during the period of disease induction and development (38, 39). In tEAU mice that received R848 treatment, disease peaked at days 8-13 after induction, and severe disease lasted until the day of sacrifice. In contrast, in the absence of R848 treatment, disease peaked on day 19 in mice that received BSS treatment. The peak score in R848-treated mice was approximately 30% higher than that in BSS-treated control mice. Figure 2C summarizes the disease severity distribution in tEAU animals. Overall, 37.5% of BSS-treated tEAU mice developed mild disease (score ≤ 1), while the percentage was 5.8% in the R848-treated group. Moreover, 17.6% of R848-treated mice exhibited very severe disease (score=4). Fundoscopy (Figure 2D) and histological analysis (Figure 2E) showed many more infiltrated cells in the ocular fundus, along with neovascularization, disordered structures and significant morphological changes, in the R848-treated group, indicating that R848 exacerbated ocular inflammatory cell infiltration and impaired retinal tissue structure.

Treatment with R848 exacerbated ocular inflammation. To examine whether R848 affected antigen-specific immunity, the production of pathogenic cytokines associated with uveitis, IFN-γ and IL-17, and IL-6 by splenocytes in response to IRBP in R848-treated tEAU mice were compared with those from BSS-treated tEAU mice. The levels of the cytokines IFN-γ, IL-17 and IL-6 were relatively increased in R848-treated tEAU mice (Figure 3A). The IRBP-specific IL-17 level was significantly elevated. In addition, the gene expression of tgf-β was upregulated in RPE cells from the tEAU+R848 group compared to the BSS-treated group (Figure 3B). To examine the direct effects of R848 on RPE cells, we treated primary RPE cells with different doses of R848 and analyzed the production of IL-6, which is a proinflammatory cytokine that is produced in response to specific pathogen-associated molecular patterns (PAMPs). R848 (30 and 300 ng/mL) could significantly induce RPE cell production of IL-6. Moreover, IL-6 production was significantly increased when RPE cells were stimulated with both R848 and recombinant IL-17, revealing a synergistic effect (Figure 3C). In contrast, treatment with the TLR7 antagonist hydroxychloroquine sulfate (HCQ) (4000 ng/mL) significantly reduced IL-6 production. Although the addition of IL-17 enhanced IL-6 levels, it failed to restore the suppressive effect of HCQ on IL-6 production (Supplementary Figure 3).

Because IL-17 can impair the expression of tight junction proteins in ARPE-19 cells (40), we next determined whether IL-17 combined with TLR7 signaling to affect the barrier function of RPE cells. Impedance was measured to quantify the integrity of the epithelial barrier by ECIS analysis. Figure 4A shows an instant decrease in impedance in primary RPE cells after treatment with recombinant IL-17, indicating an impairment of barrier function. Statistical analysis showed a significant difference between the IL-17-treated and untreated groups, with an approximately 10% decrease from 6 to 24 hours after treatment (Figure 4B). Likewise, R848 treatment also resulted in dysfunction of the epithelial barrier in primary RPE cells, and Figure 4C shows that R848 induced a dose-dependent reduction from 3 ng/mL to 30 ng/mL. Treatment with 30 ng/mL R848 reduced the impedance to 90% (Figure 4D). Moreover, under the stimulation of IL-17 or R848, the expression of tight junction-associated genes was down-regulated. In particular, treatment of IL-17 significantly reduced the expression of claudin1 (Figure 4E). These results indicated that the TLR7 agonist R848 exacerbated disease by increasing inflammatory cytokine production and impairing barrier function. This result suggested that the increase in tlr7 gene expression was associated with the exacerbation of uveitis.

RNA interference is widely used to manipulate gene expression and is considered a powerful therapeutic tool. To investigate whether tlr7 gene expression is directly associates with the induction of uveitis and serves as a potential therapeutic target, a recombinant AAV (rAAV) delivering an shRNA against TLR7 was used. Before performing the treatment, we first identified the appropriate serotype of rAAV with robust efficiency in primary RPE cells. rAAV-8 had the highest infection rate according to GFP expression, followed by rAAV-9 (Figure 5A). Next, rAAV.shTLR7 was constructed, and its inhibitory effect on TLR7 expression was examined. Figure 5B shows that infection with rAAV.shTLR7 dose- and time-dependently inhibited tlr7 expression in primary RPE cells at a viral load of 2 or 5×10¹⁰ vg/mL at 24 or 48 hours in vitro. To assess the ability of rAAV.shTLR7 to downregulate TLR7 levels, RPE cells were isolated from tEAU mice that received rAAV.shTLR7, and TLR7 expression was compared with that in rAAV.sc-treated tEAU animals. The mice treated with rAAV.shTLR7 showed lower tlr7 gene expression in the RPE layers than rAAV.sc-treated mice (Figure 5C).

To evaluate whether rAAV.shTLR7 could treat autoimmune uveitis, on the day of tEAU induction, the mice were treated with rAAV (4×10¹⁰ vg/mouse) by subretinal injection, and the severity score was monitored during disease development (Figure 6A). After the incidence of the highest score between these two groups was analyzed, an almost 25% reduction in the disease score in mice treated with rAAV.shTLR7 compared with mice treated with rAAV.sc was observed (Figure 6B). No mice with severe disease (score≥3) were observed in the rAAV.shTLR7-treated group, while 40% were present in the rAAV.sc-treated group (Figure 6B). Histological analysis and fundoscopy examination also revealed disease lesions with improved pathological morphology and lower cell infiltration in ocular after being treated with rAAV.shTLR7 (Figures 6C, D). In addition, compared with mice in the rAAV.sc group, mice treated with rAAV.shTLR7 exhibited reduced production of the pathogenic uveitis cytokines IFN-γ and IL-17 by splenocytes in response to IRBP (Figure 6E), although no significant difference was found in the production of IL-10 (Figure 6F) or Treg frequency (Figure 6G). Similar results were found in aEAU animal model. The aEAU mice treated with rAAV.shTLR7 showed lower tlr7 gene expression in the RPE layers and disease severity (Supplementary Figures 4A, B). Histological analysis and fundoscopy examination indicated improved pathology after being treated with rAAV.shTLR7 (Supplementary Figures 4C, D). Reduction of pathogenic cytokines IFN-γ and IL-17 were observed in splenocytes in response to IRBP in rAAV.shTLR7 group (Supplementary Figure 4E), No significant difference in Treg frequency and/or IRBP-specific Treg response was found in these groups of mice (data not shown). Collectively, the above results in aEAU and tEAU models suggested that TLR7-mediated responses were involved in the underlying mechanism contributing to inflammation and disease development.