Toll-Like Receptor 3 Activation Initiates Photoreceptor Cell Death In Vivo and In Vitro

Abstract

Purpose: Accumulating evidence has demonstrated that excessive immunoreaction plays a prominent role in the pathogenesis of dry AMD. Toll-like receptor 3 (TLR3) can be activated by double-stranded (ds)RNA in retinal pigment epithelia and trigger an innate immunity-mediated inflammatory response. However, its role in photoreceptor cells, the effectors of AMD geographic atrophy, remains unclear.

Methods: The expression of TLR3 was examined in mouse retina and in a murine photoreceptor cell line (661W). Retinal structure, function, and cell death in the polyinosine-polycytidylic acid (poly I:C)–treated retina were investigated by optical coherence tomography, electroretinography (ERG), and immunostaining. Cytokine and chemokine expression as well as cell death were measured in poly I:C–exposed 661W cells and explant retinas. By comparing the RNA sequencing (seq) data of 661W cells and murine retina, we comprehensively investigated the contribution of photoreceptor in poly I:C–induced retinal immune response.

Results: Toll-like receptor 3 was highly expressed in the inner segment of the photoreceptor and in 661W cells. We found poly I:C induced significant retinal structural damages and impairment of ERG responses. Focal ERG demonstrated that injected and parainjected zones were functionally damaged by poly I:C. In addition, poly I:C acted on cultured photoreceptor cells directly and evoked an inflammatory response that exhibited similarities with the immune response in mouse retina. Moreover, TLR3 activation initiated cell death in murine photoreceptor cells in vivo and in vitro. Additionally, poly I:C initiated immune response in explant retinas.

Conclusions: We deciphered the TLR3-mediated inflammatory response in photoreceptor cells. Our findings suggested TLR3-mediated inflammatory response in photoreceptor cells may play an important role in dry AMD, offering new insights of potential treatments targeting photoreceptor immunity.

Age-related macular degeneration is the leading cause of senile blindness worldwide. In patients with dry-type AMD, the appearance of drusen represents early-to-intermediate stage and the increase of drusen number and size significantly enhances the risk of advanced disease. At the late stage of dry AMD, geographic atrophy (GA) occurs due to photoreceptor and RPE cell death.¹ Accumulating dsRNA can be found in GA retina,^2,3 which has been identified as an important activation signal for innate immune response. 

A number of studies have demonstrated that innate immunity plays a key role in the pathogenesis of AMD.^4,5 On the other hand, pattern recognition receptors (PRRs) have been recognized as a primitive part involved in the innate immunity system.⁶ Among them, toll-like receptors (TLRs) can recognize extracellular pathogens or endogenic molecules, trigger the cytokine synthesis and secretion, and activate host defense programs, which are overarching for innate immune responses.^7,8 

Among the 11 TLRs, toll-like receptor 3 (TLR3) has been identified as a strong risk factor for the development of GA, but not choroidal neovascularization in patients with AMD by genetic association studies.^9,10 Toll-like receptor 3 is capable of recognizing viral components, such as double-stranded (ds)RNA and polyinosine-polycytidylic acid (poly I:C).¹¹ Moreover, dsRNA and poly I:C can activate TLR3 either in vivo or in vitro, which thereafter triggers production of proinflammatory cytokines and type I IFNs through MyD88- or TRIF-dependent signaling pathways.^7,8 In immune cells, TLR3 prompts secretion of IFN-β; monocyte chemotactic protein (MCP)-1 (chemokine [C-C motif] ligand 2 [CCL2]); and nuclear factor kappa-light-chain-enhancer of activated B cells, and activates interferon regulatory factor (IRF)-3.¹² Ligands of TLR3, including dsRNA and endogenous mRNA, are the components of drusen accumulating in the RPE of human eyes with GA.^13–15 Toll-like receptor 3 is highly expressed in human RPE cells and its activation mediates an inflammatory response.^16,17 Previous studies have demonstrated that poly I:C treatment urges secretion of IFNs, IL-6, IL-8, and MCP-1 in RPE cells, which plays a key role in retinal immune defense.^18,19 

Provided that photoreceptors are most adjacent to RPE in the retina, the aim of this study was to determine whether immune defense occurs in photoreceptors other than RPE. To the best of our knowledge, it has not been thoroughly investigated so far. Few studies have demonstrated that photoreceptor cell damage occurred in poly I:C–treated murine retina.^20,21 However, whether this damage was secondary to RPE damage, or poly I:C acted on photoreceptors directly has not been clearly deciphered. In this study, we examine the inflammation evoked by TLR3 ligand in murine retina as well as murine-derived cone photoreceptor cells. We provide in vivo and in vitro evidences that photoreceptor-derived innate immunity is alternatively pivotal in a dry-type AMD-mimetic model with dsRNA treatment, offering new insights of potential treatments targeting photoreceptor immunity for the disease. 

Materials and Methods

Animals and Subretinal Injection

We bought 1-month-old C57BL/6J mice from SLA CCAS (Shanghai, China) and maintained them on a 12-hour day/night cycle with free access to food and water. All the animal experiments were performed in adherence to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Mice were anesthetized with an intraperitoneal injection of ketamine 80 mg/kg body weight and xylazine 16 mg/kg. Preliminary experiments tested the injection dose and 1.25 μL poly I:C (1 mg/mL in PBS; Sigma-Aldrich Corp., St. Louis MO, USA) caused consistent damages to murine retina. After pupil dilation with 0.5% tropicamide, 1.25 μL poly I:C (Sigma-Aldrich Corp.) was injected into the subretinal space of the right eye using a 36-gauge beveled needle (Hamilton, Reno, NV, USA). Another control group was injected with PBS. The creation of retinal detachment of more than one-half of the retina was confirmed by microscope. 

Fundus Photograph and Optical Coherence Tomography (OCT)

Mice were anesthetized and the pupils were dilated as described above. Retinal images and OCT were used to monitor retinal damage using a retinal microscope (Micron IV; Phoenix Research Labs, Pleasanton, CA, USA) following the manual instructions. Vendor's image acquisition software was used to generate bright field images and OCT scans. 

Electroretinography

Full-field ERGs were performed using the same equipment and methods as previously described.²² Briefly, after 4 hours of dark adaptation, mice were anesthetized and the pupils were dilated. To record the ERGs, an Ag/AgCl wire loop electrode was placed over the cornea. A reference electrode was placed subcutaneously at the cheek and a ground electrode was inserted into the tail. Scotopic ERG was recorded at 0.4 log scot cd/m² intensity with a 30-second interstimulus interval. After 10 minutes of light adaptation, photopic ERG was measured at 1.4 log scot cd/m² in the presence of 30 cd/m² background light. 

Focal ERGs were performed with a retinal microscope (Phoenix Research Labs) as mentioned above following manual instruments. Routine procedures were carried out as described above. Then, responses to focal light stimuli were measured at luminances ranging from 5.9 scot cd/m⁻² with five sweeps at 20-second intervals within a 0.75-mm stimulus spot. 

TUNEL Staining

Tissue sections were prepared and a TUNEL kit (Promega Corp., Madison, WI, USA) was applied to detect the apoptotic cells following the manufacturer's instructions. Briefly, frozen sections of mouse retina were washed in PBS and then treated with 0.5% Triton X-100 in PBS. After being washed with PBS, retina samples were incubated in equilibration buffer containing terminal deoxynucleotidyl transferase, recombinant enzyme for 1 hour and the reaction was suspended. Images were acquired on a laser scanning confocal fluorescent microscope (Zeiss, Gottingen, Germany). Statistical analysis of apoptotic cells were performed at the poly I:C–injected and –parainjected zones. 

Immunohistochemistry

Mouse retinas were dissected and fixed in ice-cold 4% polyformaldehyde, then frozen in embedding resin (Cryomatrix; Thermo Fisher Scientific, Kalamazoo, MI, USA). Sections (12-μm thick) were prepared along the horizontal meridian. The sections were washed in PBS, blocked, and incubated with rabbit anti-TLR3 (1:400; Sigma-Aldrich Corp.), mouse anti-rhodopsin (1:10,000; Sigma-Aldrich Corp.), rat anti–glial fibrillary acidic protein (GFAP, 1:200; Sigma-Aldrich Corp.) and anti–ionized calcium-binding adapter molecule 1 (IBA1, 1:200; Wako, Osaka, Japan) followed by secondary labeling. 

Cell Culture and Poly I:C Transfection

Murine photoreceptor–derived 661W cells were donated by Muayyad R. Al-Ubaidi (Department of Cell Biology, University of Oklahoma Health Sciences Center, Oklahoma City, OK, USA). We maintained 661W cells in Dulbecco's modified Eagle's medium (DMEM; Life Technologies Corp., Beijing, China) containing 10% fetal bovine serum (Gibco, Thermo Fisher Scientific, Grand Island, NY, USA) and 100 U/mL penicillin-streptomycin (Thermo Fisher Scientific). We transfected poly I:C or poly (2′-deoxyinosinic-2′-deoxycytidylic acid) sodium salt (poly dI:dC, 1 mg/mL in PBS, Sigma-Aldrich Corp.) into starved 661W cells using a transfection reagent (Lipofectamine 2000; Invitrogen, Carlsbad, CA, USA) to a final concentration of 10 μg/mL. Another group of 661W cells was treated with PBS as control. 

Quantitative RT-PCR

Total RNAs were extracted using a commercial reagent (TRIzol; Life Technologies Corp., Carlsbad, CA, USA), and then reverse-transcribed into cDNA using Moloney murine leukemia virus reverse transcriptase (Promega Corp.). The samples of cDNA were used for semiquantitative PCR or quantitative PCR in a real-time PCR system (7500 Fast Real-Time PCR System; Roche, Mannheim, Germany) using a master mix (FastStart Universal SYBR Green Master [ROX]; Roche). Primer sequences specific to the genes of interest are listed in Supplemental Table S1. Amplification of each sample was performed in triplicate for each gene, and the expression levels were normalized to the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) using the delta-delta Ct method. 

RNA Sequencing

Total RNAs were collected from mouse retinas or 661W cells treated with PBS, poly dI:dC or poly I:C for 48 or 12 hours, respectively, using a commercial extraction kit (RNeasy Mini Kit; QIAGEN, Hilden, Germany). Library construction was performed by Biomarker Technologies (Beijing, China). The high throughput sequencing (HTSeq) package was used to gain aligned reads to the gene annotation. The differential expression analysis for poly I:C or poly dI:dC and PBS group comparisons were performed by exact test with mouse retina and 661W cells. We used R package for dot plot and heat map generation. 

Mitochondrial Membrane Potential Assay (Δψm)

We analyzed Δψm using a commercial assay kit (MitoProbe JC-1; Life Technologies Corp., Grand Island, NY, USA), according to the manufacturer's instructions. Briefly, cells were cultured on cover slips in 24-well plates and treated with poly I:C or poly dI:dC for 18 hours, followed by JC-1 staining. 

Annexin-V/PI Double Staining

The plasma membrane changes were analyzed by double staining with a commercial staining kit (Annexin V-FLUOS; Roche Molecular Systems, Inc.) and propidium iodide (PI) according to the manufacturer's instructions. Cells were harvested 18 hours after poly dI:dC or poly I:C treatment and stained with a commercial stain (annexin V-FLUOS/PI; Roche) in binding buffer for 15 minutes, and thereafter analyzed using a flow cytometer (BD Accuri C5; BD Biosciences, Parkland, MI, USA). 

Explants

Retinas were dissected from a mouse on postnatal day 30 based on the protocol described previously²³ with slight modifications. Briefly, retinas were dissected in Dulbecco's PBS (Life Technologies Corp.) and transferred to membrane filters (Nucleopore Track-Etch; Whatman plc, Maidstone, UK). Four cuts were made to flatten out evenly on the membrane. Explants were cultured in medium containing 2 mM L-glutamine and 10% FBS in a 1:1 mix of DMEM and F12 (Life Technologies Corp.) with penicillin and streptomycin. Transfection reagent (FuGene; Promega Corp.) was used for poly I:C or poly dI:dC delivery to retinal cells in a final concentration of 20 μg/mL. Another group was treated with PBS as control. 

Statistical Analysis

Data were presented as mean ± SEM. The graphical pictures and statistical significance of findings were determined via ANOVA (unpaired t-test) using statistical software (GraphPad Prism 5; GraphPad Software, Inc., La Jolla, CA, USA). Values of P < 0.05 were considered statistically significant. 

Results

TLR3 Expression in Photoreceptor

Expression of TLR3 was analyzed using immunostaining and RT-PCR in retinal photoreceptor and in a murine cone photoreceptor cell line, 661W. Toll-like receptor 3 expressed in the inner segment of photoreceptor in mouse retina, with rhodopsin expression indicating the outer segment (Fig. 1A). Moreover, TLR3 expression was also observed in retinal ganglion cells (Fig. 1B). In cultured 661W cells, four different types of TLRs were expressed, including TLR2, TLR3, TLR4, and TLR5. Among them, TLR3 was the most highly expressed one (Fig. 1C). Additionally, TLR3 expression was further detected by Western blotting (Fig. 1D). 

Figure 1

 

Expression of TLR3 in photoreceptor. (A) Toll-like receptor 3 expresses in the inner segment (IS) of the photoreceptor with rhodopsin expression indicating the outer segment (OS). Green: TLR3. Red: rhodopsin. Bar: 10 μm. (B) Expression of TLR3 in different retinal layers. Green: TLR3. Red: DAPI. Bar: 25 μm. (C) Semiquantitative RT-PCR and (D) western blotting analyze the expression of TLRs in the photoreceptor cell line, 661W. Both samples (R1 and R2) of 661W cells express TLR3.

Poly I:C Treatment Induced Structural and Functional Damage of Mouse Retina

We injected TLR3 ligand poly I:C into the mouse subretinal space (Fig. 2A). Subsequently, fundus photography and OCT were carried out to determine the structural damage of mouse retina. After 1 week of poly I:C stimulation, RPE cells were almost completely destroyed around the injection site (Fig. 2B). Compared with the noninjected fellow eye, the RPE and photoreceptor cell layers were seriously damaged in the poly I:C–treated eye (Fig. 2C). After 2 weeks or 1 month postinjection, the structural damage progressed around the injection site (Supplemental Figs. S1A, S1B). 

Figure 2

 

Structural and functional damage of mouse retina induced by poly I:C. (A) Graphical workflow of experiments in mouse model. (B) Color fundus photographs and (C) OCT images of poly I:C–injected eye shows serious damage compared with the nontreated fellow eye. Asterisks and arrowheads indicate damage points around injection sites. (D) Responses of ERG of poly I:C–treated eye display severely decline compared with noninjected fellow eyes. (E) Box plots show significant difference of ERG responses in poly I:C– and PBS-injected controls. Black line in the box indicates the median. Each dot represents an individual measurement. ** P < 0.01. *** P < 0.001. ns, not significant, unpaired t-test. Scotopic and photopic b-waves represent rod and cone signal transduction, respectively. Standard a- and b-waves represent the combined function of rod and cone. (F) Focal ERG reveals reduction of scotopic a-/b-wave responses at poly I:C–injected and parainjected zones compared with PBS-treated eyes. Two records are taken in each poly I:C–injected-eye, one from poly I:C–injected and the other one from the parainjected zones. Control focal ERGs are taken at random position of PBS-injected eyes. Box plot creates from absolute value of a-/b-wave amplitudes for an n = 7 sample. Black line in the box represents the median and the whiskers indicate the variability of the data. Each dot represents an individual measurement. ** P < 0.01. *** P < 0.001.

To further analyze the damage of retinal function in murine retina caused by poly I:C, we conducted an ERG analysis. Scotopic b-wave amplitude of poly I:C–treated eyes was decreased to 25.9% of the nontreated fellow eyes (P < 0.005, Figs. 2D, 2E), indicating damage of rod signal transduction. Additionally, both a- wave and b-wave amplitudes of maximal combined response of poly I:C–treated eyes were significantly reduced (P < 0.001, Fig. 2E). Moreover, photopic b-wave amplitude showed a slight decrease in the poly I:C–treated eye compared with PBS treatment. There were significant differences in amplitudes of scotopic a- and b-waves by focal ERG between poly I:C– and PBS-treated eyes. However, no significant differences in scotopic responses were found in the poly I:C eyes between injected and parainjected zones (Fig. 2F, Supplemental Fig. S1C). These results demonstrated that photoreceptor cells, especially rod photoreceptor cells, were severely damaged after poly I:C exposure. 

Photoreceptor Cell Death and Glia Cell Activation Induced by Poly I:C in Mouse Retina

To further reinforce the idea that TLR3 ligand may lead to photoreceptor cell death, TUNEL assay was performed. Cells that were TUNEL positive were observed in poly I:C–treated retina (Fig. 3A). Furthermore, the number of TUNEL-positive cells was significantly higher in poly I:C–treated retinas compared with controls (P < 0.001, Fig. 3B). Most of the TUNEL-positive cells were located in the poly I:C–injected zone (Fig. 3B). These findings suggested that poly I:C treatment resulted in photoreceptor cell death in mouse retina. 

Figure 3

 

TLR3 ligand induced photoreceptor death and glial cell activation. (A) Staining with TUNEL of mouse retina after poly I:C injection for 2 weeks. Red rectangle indicates the poly I:C–injected zone and the area outside the rectangle is the poly I:C–parainjected zone. Arrows indicate amplified sites. Bar: 150 μm. (B) Statistical analysis of apoptotic cells (TUNEL positive cells) in (A). Nine sections from three retinas were randomly selected for the calculation (n = 9). ** P < 0.01. *** P < 0.001, unpaired t-test. (C) Immunostaining of Müller glia cell marker, GFAP and (D) microglia cells marker IBA1 after poly I:C injection for 7 days. Bar: 20 μm.

Additionally, we also examined the activation of immune cells in poly I:C–injected retina around the injected-zone, such as Müller glia and microglia cells (Figs. 3C, 3D). The expression of GFAP expression representing Müller glia, was increased after poly I:C injection compared with PBS control (Fig. 3C). Obviously, the number of microglia cells (IBA1-positive cells) was also raised in poly I:C–treated retina (Fig. 3D). In summary, Müller glia and microglia cells were activated by poly I:C treatment in mouse retina. 

Poly I:C Initiated Inflammatory Response in Photoreceptor Cells

To assess the effect of dsRNA on immune-related gene expression, we performed RNA sequencing (RNAseq) on cultured photoreceptor cell treated with synthetic dsRNA poly I:C, synthetic dsDNA poly dI:dC, and PBS (Supplemental Table S2). Differential expression genes were determined and changes considered significantly affected were 2-fold or greater (fold change [FC] ≥2) with a false discovery rate (FDR) ≤0.05 from the PBS control. Consequently, we found that the expression of 225 genes in poly I:C–treated group were significantly changed. However, only 14 genes expression were significantly altered in poly dI:dC–treated cells. 

To validate the RNAseq data, some altered genes were picked out for quantitative RT-PCR detection including TLR3 and genes encoding well-characterized inflammatory factors (Figs. 4A, 4B). The level of mRNA of TLR3, chemokine (C-X-C motif) ligand (CXCL) 10, and CCL5 increased after poly I:C transfection at different time points (Fig. 4C). However, there were no significant differences between poly I:C– and poly dI:dC–treated cells at all tested time points on the expression of CCL20 (Fig. 4C). 

Figure 4

 

Inflammatory response induced by TLR3 ligand in 661W cells. Changes in gene expression levels (x-axes, log2 FPKM) are compared for poly dI:dC (A) or poly I:C (B) versus PBS control (y-axes, log2 FC). True (significantly changed) is defined as FC ≥2 with an FDR ≤0.05. (C) Expression of mRNA levels of indicated genes (genes labeled blue in [A, B]) at different time points relative to GAPDH. Data are reported as mean value ± SEM (n = 3). * P < 0.05. ** P < 0.01. *** P < 0.001. (D) Heat map present FC of gene expression in poly dI:dC– and poly I:C–treated 661W cells relative to that of PBS control, in genes of cytokine–cytokine interaction pathway.

In the cytokine–cytokine receptor interaction pathway, 23 genes were highly upregulated in poly I:C–treated 661W cells relative to PBS control (Fig. 4D). This upregulation was not detected in poly dI:dC, indicating that the activation of cytokine–cytokine receptor interaction pathway was specifically affected by poly I:C. Collectively, these findings suggested that poly I:C, instead of poly dI:dC, activated TLR3 and evoked inflammatory response in photoreceptor cells. 

Poly I:C Evoked Similar Inflammatory Response in Cultured Photoreceptor Cells and in Murine Retina

To address the contribution of photoreceptor cells in retinal immunity, retina samples were subjected to RNAseq as well, after PBS-, poly dI:dC– or poly I:C–treated for 48 hours (Supplemental Table S3). We compared gene expression in poly dI:dC– or poly I:C–treated retina with that of PBS control. The number of genes affected in poly I:C–treated retina was much larger than in poly dI:dC–treated retina (Table). 

Table

 

List of RNAseq Experiments and Number of Genes With Altered Expression

 

When comparing the changed gene sets between retina and 661W cells within the group of poly I:C versus PBS, three significantly enriched Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were related with immune response, including the cytokine-cytokine receptors interaction, toll-like receptor signaling pathways and chemokine signaling pathways (Fig. 5A). Additionally, two downstream pathways of immune process were significantly enriched as well, such as the Janus kinase (JAK)–signal transducer and activator of transcription (STAT) signaling pathway and MAPK signaling pathway. We next explored the gene expression changes in toll-like receptor signaling pathway, from which some key signaling factors were discovered upregulated in both poly I:C–treated retina and 661W cells (Fig. 5B). Furthermore, all the overlapped genes affected by poly I:C in retina and 661W cells could be clustered into three groups related to innate immunity according to the functions of their encoded proteins (Fig. 5C). Together, these findings suggested that immune response induced by poly I:C in 661W cells is similar with that in mouse retina, indicating that photoreceptor cells could be the primary cell initiating retinal immune response. 

Figure 5

 

Comparison of immune response in mouse retina and 661W cells. (A) Significantly enriched KEGG pathways in retina and 661W cells. Numbers of altered-gene in different KEGG pathways are shown at the end of the diagram. (B) Differentially expressed genes of TLR signaling pathway in poly I:C–treated retina and 661W cells. (C) Heat map presents FCs of gene expression in poly I:C–treated retina or 661W cells relative to that of PBS.

TLR3 Activation-Initiated Photoreceptor Cell Apoptosis

To probe the effect of TLR3 activation on photoreceptor cells, we also specifically analyzed the expression of genes involved in cell apoptosis, including positive regulation of apoptotic process (GO:0043065) and apoptotic process (GO:0006915). Expression changes were seen in many known apoptosis-related components, such as the caspase3 activator, caspase12 and DNA damage signal factor, ataxia-telangiectasia mutated (Fig. 6A). In addition, TNFs and IFN-induced proteins (IFIT2 and IFI205) were also upregulated. These findings indicated that TLR3 activation might promote photoreceptor cell apoptosis. 

Figure 6

 

Activation of TLR3 induced photoreceptor cell apoptosis. (A) Heat map presents FCs of gene expression in poly dI:dC– and poly I:C–treated cells relative to that of PBS control, in subsets of gene involved in cell apoptosis. (B) Cells treated for 18 hours were stained with JC-1 dye, an indicator of mitochondrial membrane potential. The number of apoptotic cells (green) is increased in poly I:C–treated group compared with the poly dI:dC–treated group. Bar: 50 μm. (C) Apoptosis of 661W cells was analyzed with annexin V/PI staining after poly dI:dC and poly I:C treatment for 18 hours. In the poly I:C–treated group, the number of apoptotic cells (annexin V single positive, green) is much more than in the poly dI:dC–treated group. Bar: 100 μm. (D) Flow cytometry analyzed the annexin V/PI staining. (E) Cell counts comparison of apoptotic cells (annexin V single positive).

To find out whether photoreceptor cell went apoptosis after TLR3 activation, mitochondrial membrane potential assay and annexin V/PI double staining were conducted. Consequently, less red fluorescence and enhanced green fluorescence were observed in poly I:C–treated cells, indicating that cells were undergoing apoptosis (Fig. 6B). The proportion of annexin V single–positive cells in the poly I:C–treated group was much higher than poly dI:dC–treated group analyzed with flow cytometry (Figs. 6C–E), indicating that poly I:C treatment induced early apoptosis in 661W cells. Taken together, these results suggested that TLR3 activation initiated photoreceptor cell apoptosis. 

TLR3 Activation in Explant Retinas

In order to eliminate interference of RPE cells, neural retinas were dissected and treated with poly I:C. A significant increase of TLR3 expression was found. Moreover, the expression of IL-33 and CCL5 was also significantly higher in poly I:C–treated explants than poly dI:dC controls (P < 0.05, Fig. 7). However, no significant differences were found in CCL20 expression, which is similar with the result found in cultured 661W cell. 

Figure 7

 

Inflammatory response in explant retina induced by poly I:C. (A–C) Significant increase TLR3, IL-33, and CCL5 expression are induced by poly I:C treatment. (D) No significant differences are found in CCL20 expression. Data are reported as mean value ± standard error of the mean (n = 3). * P < 0.05.

Discussion

In present study, we aimed to explore the TLR3-mediated inflammatory response in photoreceptor cells. It has been confirmed that TLR3 expresses in the inner segment of photoreceptor in murine retina, and highly expresses in cultured photoreceptor cells. Additionally, TLR3 ligand can activate TLR3 and trigger inflammatory response in murine cone photoreceptor cells, which have shown great similarity with that in murine retina. Finally, our results demonstrate that TLR3 activation can induce photoreceptor cell death in vivo and in vitro. 

We performed ERGs to evaluate the functional damage caused by subretinal injection of poly I:C. Scotopic/photopic ERG pattern reflects signal transduction initiated from rods or cones, respectively. In vivo data have shown that it can induce great damage in scotopic responses and a slight decrease in photopic responses (Fig. 2E). What's the reason for the difference between scotopic and photopic responses? There are at least two possible causes. In murine retina, the number of cones is much fewer than rods. Only 3% of the photoreceptor cells are cones and 97% of them are rods.²⁴ Moreover, signals are amplified during signal transduction between photoreceptors and secondary neurons. Thus, the decrease of photopic responses is not comparable as scotopic responses and both rods and cones are damaged in poly I:C–injected murine eyes. 

Photoreceptor cells are the most prevalent cells in the retina; however, the role of photoreceptor cells in retinal inflammation is barely known. Photoreceptor cells are the major source of oxidative stress in retinas of diabetic mice and induction of proinflammatory proteins does not occur in retina without photoreceptor cells, indicating the predominant role of photoreceptor cells in immune response in retina.²⁵ Here we use poly I:C to mimic dsRNA production in AMD to investigate TLR3-mediated inflammation in cultured photoreceptor cells. Consequently, the hallmark of TLR3 activation, cytokine production has been detected (Fig. 4). Thus, we provide some evidence that photoreceptor cells can respond to dsRNA directly and initiate inflammatory response. 

Even though some signaling factors have been reported, extensive exploration is still needed to detail the molecular network in the TLR3 signaling pathway, as it relates to retinal innate immunity. For the first time, we have examined TLR3-mediated inflammatory response in photoreceptor cells. Consistent with findings in RPE cells, increased expression of proinflammatory cytokines IL-6 and CCL2 have been detected.¹⁸ Meanwhile, signaling factor myeloid differentiation primary response gene 88 (MyD88) in the TLR-link signaling pathway—and cytokines and chemokines such as CXCL10, CCL5, IL-18, CCL4, IFN-α, and IFN-β—found upregulated in poly I:C–treated photoreceptor cells.²⁶ In line with previous studies, MyD88 plays an important role in retinal degeneration by regulating expression of CCL2, CCL4, CCL7, CXCL10 and CCL5.²⁷ In double-stranded RNA accumulated RPE cells of GA, MyD88 activation contributes to RPE degeneration and potentially vision loss through proinflammatory IL-18 regulation.³ The upregulation of MyD88 and its downstream genes suggests that dsRNA might induce inflammation through the MyD88-dependent signaling pathway in murine photoreceptor cells. There is still the possibility that other receptors or sensors might respond to poly I:C. In further study, TLR3 knockout mice or TLR3/dsRNA complex inhibitors could be used to thoroughly understand the signaling pathway evoked by poly I:C. 

It has been reported that STAT3^28,29 and c-Jun N-terminal kinases (JNK3)³⁰ are key factors in TLR3 activation in retinal cells. We found significant activation of JAK-STAT signaling pathway, but not upregulation of JNK3 in poly I:C–treated retina and 661W cells. Typically, TLR3 uses TIR-domain–containing adapter-inducing interferon-β (TRIF) to activate IRF3,³¹ and IRF3 activation has been observed in dsRNA-treated mice retina.^15,20 In our study, upregulation of IRF5 and IFR7 has been found. Our findings will be important for determining the function of dsRNA in the pathogenesis of retinal degeneration and illustrating the network of dsRNA induced innate immunity. 

In murine retina and cultured photoreceptor cells, a group of genes involved in immune response were found upregulated after poly I:C treatment compared with PBS control (Fig. 5C). Among them, eight genes are IFN-induced genes (IFIH1, IFIH3, IFIH3b, IFI44, GBP2, IFIT1, GM12185, and IIGP1),^32–34 which corresponds to the increased expression of IFNs (Fig. 4D). Other genes participated in transcriptional regulation of immune factors are detected as well, such as FCGR1, ATF3, PHF11, and SP110.^35–38 In a related vein, gene clusters that are important factors in cytokine response and IFN production (e.g., NLRC5, IFI44, and ZBP1) are found upregulated in both retina and cultured photoreceptor cells.^39–42 On the other hand, the expression of class II major histocompatibility complex–associated gene CD74 was also increased, which might attenuate a TLR3-triggered innate inflammatory response.⁴³ For the first time, these immune-related genes are reported to be expressed in photoreceptor cells. 

The ligand of TLR3 can cause retinal cell apoptosis or necrosis in vivo^20,21 and in vitro.^29,44,45 Photoreceptor cell death has been observed in a poly I:C–injected mouse model; however, whether photoreceptor cells respond to dsRNA directly or are affected by RPE dysfunction remains unexplained. Here we have shown that poly I:C treatment induced photoreceptor cell apoptosis, although the underlying mechanisms have not been fully elucidated (Fig. 6). Additionally, we also found rod dysfunction earlier than cone damage, which is consistent with the situation in AMD patients, where rod photoreceptor cells die before cones in macular degeneration.⁴⁶ Along with TLR3 activation and photoreceptor cell death, glia cell activation has also been observed, which might be regulated by IL-33.^3,15,47 These findings suggest that photoreceptors might be a primary target rather than a secondary effect of RPE dysfunction in AMD pathology and further studies are needed to clarify the function of photoreceptor in cell death and inflammation in AMD. 

In this study, we used the murine photoreceptor cell line because the 661W cells have been well characterized in gene expression⁴⁸ and have been demonstrated to be suitable for scientific research.^49,50 Additionally, primary photoreceptor cells are technically difficult to isolate and culture. In combination with photoreceptor cells and the mouse model, our results have demonstrated that photoreceptor cells can respond directly to dsRNA and evoke inflammatory response, which causes photoreceptor cell apoptosis. Our findings may help to illustrate the function of photoreceptor cells in retinal immune response and the contribution of photoreceptor cells in the pathology of AMD. 

Acknowledgments

The authors thank for Dong-Qing Li's contribution to this work and Fei-Fei Cheng and De-Fu Chen's technical assistance on data analysis. 

Supported by the National Natural Science Foundation of China (81371059, 81522014); the Zhejiang Provincial Natural Science Foundation of China (LR13H120001, Q14B020018, LQ17H120005); the National Key Basic Research Program (2013CB967502); the Zhejiang Provincial Key Research and Development Program (2015C03029); the Zhejiang Provincial Key Research and Development Program (2015C03029); the NHFPC Grant-in-Aid for Medical and Health Science (WJK-ZJ-1417); the Wenzhou Science and Technology Innovation Team Project (C20150004); and institutional grants (KYQD141108, YNKT201503). 

Disclosure: M.-L. Gao, None; K.-C. Wu, None; W.-L. Deng, None; X.-L. Lei, None; L. Xiang, None; G.-H. Zhou, None; C.-Y. Feng, None; X.-W. Cheng, None; C.-J. Zhang, None; F. Gu, None; R.-H. Wu, None; Z.-B. Jin, None