Tumor Necrosis Factor-α Mediates Photoreceptor Death in a Rodent Model of Retinal Detachment

Abstract

Purpose.: Photoreceptor degeneration is a major cause of visual loss in various retinal diseases, including retinal detachment (RD) and neovascular AMD, but the underlying mechanisms remain elusive. In this study, the role of TNFα in RD-induced photoreceptor degeneration was investigated.

Methods.: RD was induced by subretinal injection of hyaluronic acid. Photoreceptor degeneration was assessed by counting the number of apoptotic cells with TdT-dUTP terminal nick-end labeling (TUNEL) 3 days after RD and measurement of the outer nuclear layer (ONL) thickness 7 days after RD. As the target of anti-inflammatory treatment, the expression of TNFα, with or without dexamethasone (DEX) was examined in rats by real-time PCR. To understand the role of TNFα in photoreceptor degeneration, RD was induced in mice deficient in TNFα or its receptors (TNFR1, TNFR2, and TNFR1 and -2), or in wild-type (WT) mice by using a functionally blocking antibody to TNFα. CD11b⁺ cells in the outer plexiform layer (OPL) and subretinal space were counted by immunohistochemistry (IHC).

Results.: Treatment with DEX (P = 0.001) significantly suppressed RD-induced photoreceptor degeneration and the expression of TNFα. RD-induced photoreceptor degeneration was significantly suppressed with specific blockade of TNFα (P = 0.032), in mice deficient for TNFα (P < 0.001), TNFR2 (P = 0.001), or TNFR1 and -2 (P < 0.001). However, lack of TNFR1 did not protect against RD-induced photoreceptor degeneration (P = 0.060). Müller cell activation was unchanged in WT and TNFα^−/− mice. Recruitment of CD11b⁺ monocytes was significantly lower in the TNFα^−/− mice compared to WT mice (P = 0.002).

Conclusions.: TNFα plays a critical role in RD-induced photoreceptor degeneration. This pathway may become an important target in the prevention of RD-induced photoreceptor degeneration.

Photoreceptors are vulnerable in several retinal disorders, including macular degeneration, ¹ retinal detachment (RD), ^{2 –4} diabetic retinopathy, ⁵ retinopathy of prematurity, ⁶ and retinitis pigmentosa. ⁷ In these pathologic conditions, photoreceptors undergo apoptosis. ^{2,5 –7} Therefore, new insights about the mechanisms that underlie photoreceptor degeneration in the ocular diseases would be of clinical interest and could lead to new neuroprotective treatments. Previously, we used the rodent model of RD to clarify the mechanism of RD-induced photoreceptor degeneration. We found that RD-induced photoreceptor apoptosis went through a caspase-dependent ^8,9 or caspase-independent pathway. ¹⁰ Furthermore, we found that monocytes recruited through the upregulation of monocyte chemoattractant protein (MCP)-1 in Müller glial cells play a neurodestructive role in photoreceptor degeneration. ^11,12  

Tumor necrosis factor (TNF)-α is synthesized, mainly in monocytes, as a 26-kDa precursor ¹³ that is cleaved proteolytically and secreted as a 17-kDa protein. ¹⁴ TNFα acts via either the low-affinity TNF receptor (TNFR1) or high-affinity TNF receptor (TNFR2). ¹⁵ TNFα is upregulated in several neurodegenerative disorders including multiple sclerosis, Parkinson's disease, and Alzheimer's disease and suppression of TNFα has demonstrated therapeutic effects. ¹⁶ In ophthalmic disorders, the vitreous samples from patients with RD contain significantly higher levels of TNFα than samples from patients with other retinal conditions, such as macular hole or idiopathic premacular fibrosis. ^17,18 However, the role of RD-induced elevated TNFα on photoreceptor degeneration remains unclear. 

Recently, TNFα-suppressing monoclonal antibodies such as infliximab have been successfully used to treat patients with inflammatory ocular disease, including Behçet's disease, ¹⁹ diffuse subretinal fibrosis (DSF) syndrome, ²⁰ posterior scleritis, ²¹ retinal vascular tumors, ²² and neovascular age-related macular degeneration. ²³ Thus, if TNFα plays a neurodestructive role in RD-induced photoreceptor degeneration, anti-TNFα treatment may be a good candidate for neuroprotective treatment in retinal diseases. In this study, we induced RD in mice deficient in TNF, TNFR1, and TNFR2 and investigated the role of the TNFα pathway on RD-induced photoreceptor apoptosis. 

Materials and Methods

Animals

All animal procedures were performed in accordance with the ARVO Statement for the use of Animals in Ophthalmic and Vision Research and the National Institute of Health Guidance for the Care and Use of Laboratory Animals. The protocol was approved by the Animal Care Committee of the Massachusetts Eye and Ear Infirmary and by the Ethics Committee for Animal Experiments of Tohoku University Graduate School of Medicine. 

Adult male Brown-Norway rats, TNFα-deficient mice (TNFα^−/−), TNF receptor 1 and -2 double-deficient mice (TNFR^−/−; B6.129SF2J background, 20–25 g; Jackson Laboratory, Bar Harbor, ME), TNFR1-deficient mice (TNFR1^−/−, C57BL6 background; Jackson Laboratory), TNFR2-deficient mice (TNFR2^−/−, C57BL6 background; Jackson Laboratory), and age- and sex-matched B6.129SF2J mice or C57BL6 mice (wild-type [WT]) were housed in covered cages. Rats and mice were fed with standard rodent diet ad libitum and kept on a 12-hour light (250 lux)–dark cycle. 

Surgical Induction of RD

RD was induced in rats and mice, as previously described. ^11,12,24 Briefly, anesthesia was preformed with a mixture of xylazine hydrochloride (mice, 10 mg/kg; rats, 20 mg/kg) and ketamine hydrochloride (100 mg/kg). The pupils were dilated and a sclerotomy was created approximately 1 mm posterior to the limbus with a 30-gauge needle. A Glaser subretinal injector (20-gauge shaft with a 32-gauge tip; BD Biosciences, San Diego, CA) connected to a syringe filled with sodium hyaluronate (HealonGV; Pharmacia and Upjohn Co., Kalamazoo, MI) was then introduced into the vitreous cavity. A retinotomy was created in the peripheral retina with the tip of the subretinal injector, and 2 μL of sodium hyaluronate was slowly injected into the subretinal space (SRS), causing detachment of one half of the retina. One hour before RD, dexamethasone (DEX 1 mg/kg; Sigma-Aldrich, St. Louis, MO) or vehicle was injected intraperitoneally (IP). DEX was first dissolved in 100% ethanol alcohol to 1 mg/mL and then diluted in Dulbecco's phosphated-buffered saline (DPBS) to 0.5 mg/mL. Fifty percent ethanol alcohol in DPBS was used as a vehicle control. To block the TNFα in rat RD, the blocking antibody for TNFα was used (goat anti-rat TNFα, 0.1 μg/μL; R&D Systems, Minneapolis, MN), and goat normal IgG (NGS, 0.1 μg/μL, azide free; R&D Systems) was the control. TNFα blocking antibody 1 μL was injected subretinally RD with a syringe (Hamilton, Reno, NV) equipped with a 32-gauge needle by introducing the tip of needle through the sclerotomy into the SRS and then injecting solution over 3 minutes. RDs were created only in the right eye of each animal, with the left eye serving as the control. 

The activities of caspase-8 were measured according to the manufacturer's instruction in a commercially available kit (cat. no. APT171; Millipore, Billerica, MA). The retinal samples (150 μg) were harvested 1 to 3 days after the induction of RD, and these were treated with the vehicle, control IgG (1 μg/mL, 705-035-003; Jackson ImmunoResearch Products, West Grove, PA), and the blocking antibody for TNFα (1 μg/mL, goat IgG, cat. no. AB410-NA; R&D Systems) in mice (8 weeks). 

Quantification of TUNEL⁺ Photoreceptor In Vivo

To assess photoreceptor cell loss and apoptosis quantitatively, we used two methods as previously reported: measurement of the outer nuclear layer (ONL) thickness stained by hematoxylin-eosin (H-E) at 7 days after RD ¹¹ and cell counting with TdT-dUTP terminal nick-end labeling (TUNEL; ApopTag Fluorescein In Situ Apoptosis detection kit S7110; Chemicon International, Inc., Temecula, CA) at 3 days after RD. ¹¹ The number of TUNEL⁺ cells was counted in a masked fashion. The area of the ONL was measured in the captured images (OpenLab software; Improvision Inc., Lexington, MA), and the cell count per square millimeter calculated. 

RNA Extraction and RT-PCR

Total RNA extraction and quantitative RT-PCR was performed as previously reported. ^11,12,24 Briefly, total RNA was extracted (RNA Purification System; Invitrogen, Carlsbad, CA) and 3 μm of total RNA was subjected to RT (SuperScript III First-Strand Synthesis System; Invitrogen). First strand cDNAs were amplified using a real-time PCR thermal cycler (ABI7700; Applied Biosystems, Inc. [ABI], Foster City, CA) with a PCR core kit (SYBER Green; Applied Biosystems). The primer sets used in this study were as follows; IL-1β forward- TCAGGAAGGCAGTGTCACTCATTG and reverse- ACACACTAGCAGGTCGTCATCATC; TNF-α forward- CCCAGACCCTCACACTCAGATCAT and reverse- GCAGCCTTGTCCCTTGAAGAGAA; and MCP-1 forward- ATGCAGGTCTCTGTCACGCTTCTG, reverse- GACACCTGCTGCTGGTGATTCTCTT, all of which are described in another publication. ¹² For relative comparison of each gene, we analyzed the Ct value of the real-time PCR data with the ^ΔΔCt method, according to the manufacturer's instructions (ABI). To normalize the amount of sample cDNA added to each reaction, the Ct value of the endogenous control (18rRNA) was subtracted from the Ct value of each target gene. 

Adult Mouse Retinal Primary Cultures

Adult primary retinal cultures were prepared as described elsewhere. ¹¹ Briefly, isolated neural retinas were incubated at 37°C for 20 minutes in a CO₂ incubator in a digestion solution containing papain (10 U/mL; Worthington, Lakewood, NJ). Cell density was adjusted to 4.0 × 10⁵ cells each well with neuronal cell growth medium (Neurobasal A, containing B27 supplement, NBA/B27; Invitrogen) and 1 μg/mL insulin, 2 mM L-glutamate, and 12 μg/mL gentamicin. One hour later, TNFα was added to culture medium to reach final concentrations (0.001, 0.01, and 0.1 ng/mL) and cells were incubated for further 24 hours. To assess the viability of photoreceptors, we performed immunocytochemistry (ICC) with rabbit anti-recoverin antibody (1:500 dilution, AB5585; Chemicon), as published. ¹¹ The number of recoverin⁺ photoreceptors was counted at 10 random fields per well with a fluorescence microscope (×20 objective) equipped with an imaging system, and the number of recoverin⁺ cells was counted in a blind fashion with ImageJ software (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html). Values are given as the mean ± SEM of counts in four replicate wells. 

Immunohistochemistry

Immunohistochemistry was performed as previously reported. ^11,12,24 Briefly, 10-μm retinal sections through the optic nerve head were prepared and subjected to reaction with primary antibodies against phosphorylated ERK (pERK, 1:200; Cell Signaling Technology Inc., Beverly, MA), c-Fos (1:200; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), or CD11b (1:200; Serotec Inc., Raleigh, NC). Retinal sections incubated with a buffer without the primary antibodies were used as negative controls. Fluorescence-conjugated secondary antibodies, including goat anti-mouse immunoglobulin G (IgG) and anti-rabbit IgG and anti-rat IgG conjugated to Alexa Fluor 488 (Molecular Probes, Eugene, OR) were used. Retinal sections were mounted with antifade medium (Vectashield; Vector Laboratories, Burlingame, CA) containing 4′,6 diamidino-2-phenylindole (DAPI) to reveal the nuclear structure. Immunolabeled cells were counted in photomicrographs of the center of the detached retinal area by using a microscope equipped with fluorescence illumination (DMRXA; Leica, Bannockburn, IL) and were analyzed (OpenLab software, ver. 2.2.5; Improvision). The pERK⁺, c-Fos⁺, and CD11b⁺ cells (at ×200 magnification) were counted in a masked fashion. 

Statistical Analysis

The statistical significance of the RT-PCR result was determined with the Mann-Whitney U test. The data from the TUNEL and in vitro survival assays were analyzed with the Scheffé post hoc test (StatView 4.11J software for Macintosh; Abacus Concepts, Inc., Berkeley, CA). The significance level was set at P < 0.05. The data represent the mean ± SD except for culture results. 

Results

Anti-inflammatory Treatment with DEX Suppress the Photoreceptor Death and the Upregulation of TNFα Expression after RD

Previously, we showed that the expression of IL-1β, TNFα, and MCP-1 increased significantly after RD in rats. ¹² In the present study, we first investigated the anti-inflammatory effect of the IP-DEX (1 mg/kg) on RD-induced photoreceptor degeneration in rats. This concentration of DEX was chosen because it resulted in a neuroprotective effect on photoreceptors after photodynamic therapy (PDT) in our previous studies. ²⁵ IP-DEX suppressed the RD-induced upregulation of IL-1β expression (P = 0.004; Fig. 1A) and TNFα expression (P = 0.006; Fig. 1B) 1 hour after RD (n = 6), and MCP-1 also showed a trend toward suppression (P = 0.055; Fig. 1C). We next investigated whether the IP-DEX anti-inflammatory effect resulted in neuroprotection for the RD-induced photoreceptor degeneration. IP-DEX significantly suppressed the RD-induced photoreceptor degeneration (n = 9, P = 0.0005; Figs. 1D–F). To examine the role of IL-1β and TNFα on photoreceptor degeneration, we injected anti-IL-1β or anti-TNFα neutralizing antibody subretinally and counted TUNEL⁺ cells in the ONL 3 days after RD in rats. We found that anti-TNFα antibody significantly decreased the number of TUNEL⁺ photoreceptors (P = 0.004), but anti-IL-1β antibody did not (P = 0.109) compared with the effect of normal goat IgG (Fig. 2). These data demonstrate that anti-inflammatory treatment with DEX and anti-TNFα blocking antibody has a significant neuroprotective effect on the RD-induced photoreceptor degeneration. 

Figure 1.

 

DEX suppresses the expression of cytokines and RD-induced TUNEL⁺ photoreceptors in rats. (A–C) Quantitative real-time PCR data for IL-1β (A), TNFα (B), and MCP-1 (C) mRNA 1 hour after RD (n = 6). (D–E) Representative photography of detached retina labeled by TNUEL with (E) or without (D) DEX Quantification of TUNEL⁺ photoreceptors 72 hours after RD with (+) or without (–) DEX (F). Green: TUNEL; blue: DAPI nuclear staining. *P < 0.5, treated versus untreated condition (n = 6 each). GLC, ganglion cell layer; IPL, inner plexiform layer.

Figure 2.

 

A TNFα-blocking antibody prevents RD-induced photoreceptor loss 72 hours after RD in rats. Data are the number of TUNEL⁺ photoreceptors 72 hours after RD with subretinal administration of α-IL-1β antibody and α-TNFα antibody. NGS, normal goat serum.

Cytotoxic Effect of TNFα on Cultured Photoreceptors

To investigate the roles of the increased expression of TNFα after RD, we administered TNFα on cultured photoreceptors from adult WT mice in mixed primary retinal cell culture. Photoreceptors were identified by ICC with an antibody to recoverin, a commonly used cellular marker for photoreceptors in vitro. ¹¹ The number of recoverin⁺ photoreceptors declined progressively with increasing TNFα concentration (Fig. 3), and 0.1 ng/mL of TNFα had significant cytotoxic effects (Fig. 3C). 

Figure 3.

 

Cytotoxicity of TNFα for adult cultivated photoreceptors. (A, B) Recoverin⁺ photoreceptors in culture with (B) or without (A) TNFα. Red: recoverin⁺ photoreceptors; blue: DAPI nuclear staining. Scale bar, 100 μm. (C) Dose–response effects of TNFα cytotoxicity on recoverin⁺ photoreceptors. *P < 0.05 compared with controls without TNFα.

Genetic Deletion of TNFα and Its Receptor Prevents RD-Induced Photoreceptor Degeneration

TNFα acts via two known receptors, TNFR1 and -2. To investigate the contribution of TNFα and the receptors to the pathophysiological events described earlier, we induced RD in TNFα or TNFR1 and -2 double-knockout mice (TNFR^−/−). Photoreceptor degeneration was quantified by measuring the ONL thickness ^11,24 7 days after RD. In the absence of RD, the general appearance of the retina and the thickness of the ONL were similar in TNFα ^−/−, TNFR ^−/−, and WT mice (Fig. 4). Seven days after induction of RD, the thickness of the ONL decreased significantly in the WT mice (P = 0.001, n = 10, Fig. 4). In contrast, after RD in TNFα ^−/− and TNFR ^−/− mice, the thickness of the ONL remained unchanged from baseline (n = 10 respectively, Fig. 4). TUNEL at 72 hours after RD showed a greater number of TUNEL⁺ photoreceptors in the WT mice (2517 ± 210 cells/mm²) than in the TNFα ^−/− and TNFR ^−/− mice (P < 0.001); with no TUNEL⁺ cells in the untreated retinas (Fig. 5). These data suggest that RD-induced upregulation of TNFα has a cytotoxic effect on RD-induced photoreceptor degeneration via its receptors in vivo. 

Figure 4.

 

Deletion of the Tnfa and Tnfr genes prevents ONL degeneration after RD in mice. (A) H-E-stained sections through the retinas of WT mice, TNFα, TNFR^−/−, or TNFR1 and -2 with or without RD. (B) Quantification of the ONL thick ness with (+) or without (–) RD in WT, TNFα^−/−, and TNFR^−/− mice. *P < 0.05 compared with WT controls without RD (n = 10 each).

Figure 5.

 

Cytotoxic effect of TNFα on RD-induced photoreceptor degeneration. (A) Representative photography with TUNEL 72 hours after RD in WT, TNFα^−/−, and TNFR^−/− mice. Green: TUNEL; blue: DAPI nuclear staining. (B) Quantification of TUNEL⁺ cells in WT, TNFα^−/−, and TNFR^−/− mice. *P < 0.05 compared with WT mice with RD (n = 8 each). (C) Quantification of TUNEL⁺ cells in WT, TNFR1^−/−, and TNFR2^−/− mice. *P < 0.05 compared with WT mice with RD (n = 8 each). GLC, ganglion cell layer; IPL, inner plexiform layer.

One to 3 days after injury, caspase-8 was significantly active in the detached retina. The blocking antibody for TNFα significantly prevented the activation of caspase-8 (Fig. 6). 

Figure 6.

 

Blocking TNFα suppresses RD-induced caspase-8 activation. The data show caspase-8 activation on day 1, declining by day 3 after treatment with anti- TNFα after RD (n = 4). NS, not significant.

RD-Induced Photoreceptor Degeneration Is Mediated Via TNFR2

To further delineate the separate contribution of TNFR1 and -2 to the RD-induced photoreceptor degeneration, we induced RD on TNFR1 or -2 single-knockout mice. In the WT mice, the number of TUNEL⁺ photoreceptors 72 hours after RD was 2608 ± 262 cells/mm² (Fig. 5C) and was similar in the TNFR1^−/− mice. In contrast, the number of TUNEL⁺ photoreceptors in the TNFR2^−/− mice was significantly less than that in the WT mice (P < 0.0001, Fig. 5C). Thus, RD-induced photoreceptor degeneration appeared to be mediated by TNFα acting via TNFR2. 

Effect of TNFα on the RD-Induced Müller Cell Activation

We previously reported that RD activates the intracellular MAPK/c-Fos signaling pathway in the Müller cell immediately after insult and that the response is critical for retinal gliosis and pathogenesis of photoreceptor degeneration. ^11,24 The genetic ablation of GFAP and vimentin leads to reduced activation of pERK and c-Fos. ²⁴ To determine whether RD-induced upregulation of TNFα is an upstream event of MAPK/c-Fos activation, we compared the number of pERK⁺ or c-Fos⁺ cells in the inner nuclear layer (INL) in the WT mice and TNFα ^−/− mice, before and after RD. Surprisingly, the number of pERK⁺ or c-Fos⁺ cells in the TNFα ^−/− mice was not significantly different from that in the WT mice (Fig. 7). These data demonstrate that the deletion of TNFα is not related to the retinal glial activation associated with RD. 

Figure 7.

 

Glial activation in the WT mice and TNFα^−/− mice in the detached retina. (A) Representative photography of IHC with phosphorylated ERK (top) and c-Fos (bottom) in the WT and TNFα^−/− mice. (B, C) The quantitative data of pERK⁺ (B) and c-Fos⁺ (C) cells in the INL 72 hours after RD. NS, not significant. GLC, ganglion cell layer; IPL, inner plexiform layer.

TNFα Contributes to Monocyte Recruitment in the Detached Retina

We have shown that monocytes recruited after RD mediates their cytotoxic effect on photoreceptors through oxidative stress. ¹¹ To examine the role of TNFα in monocyte recruitment, we investigated whether deletion of the TNFα gene would alter monocyte recruitment after RD. After RD in WT mice, monocytes were recruited to the outer plexiform layer (OPL) and the SRS (Fig. 8). This recruitment was strongly suppressed in TNFα-deficient mice and these data suggest that TNFα plays a critical role in recruiting monocytes to the OPL and SRS after RD. 

Figure 8.

 

Monocyte recruitment is significantly reduced in TNFα^−/− mice. (A, B) Representative photography of IHC with CD 11b (red), recruited macrophage and resident microglia in the WT and TNFα^−/− mice. (B) Data show the number of CD11b⁺ cells in the OPL (B) and SRS (C) 72 hours after RD in the WT and TNFα^−/− mice.

Discussion

The increased expression of TNFα in the human vitreous in several retinal disorders suggests that TNFα contributes to the pathophysiology of these diseases ^17,18 ; however, under certain conditions, TNFα plays not only a cytotoxic role but also a neuroprotective one in damaged retinal neurons. Using a mouse model of RD, we showed for the first time that TNFα is a critical mediator of RD-induced photoreceptor death. Acute blockade of TNFα with a functionally blocking antibody or deletion of TNFα or its receptor gene in genetically altered mice almost completely eliminated RD-induced photoreceptor degeneration. We further showed that the cytotoxic effect of TNFα on photoreceptors is mediated through TNFR2 but not -1. These data suggest that anti-TNFα treatment has potential as a neuroprotective therapy for photoreceptor degeneration. 

We have previously demonstrated increased expression of IL-1β, TNFα, and MCP-1 in detached retina 1 hour after RD. ¹² MCP-1 had a cytotoxic effect on RD-induced photoreceptor degeneration through recruited monocytes, ¹¹ but the roles of IL-1β and TNFα remained unclear. To further explore the role of these cytokines, we administered DEX as an anti-inflammatory treatment to examine its effect on RD-induced photoreceptor degeneration. Interestingly, DEX significantly suppresses the expression of IL-1β and TNFα and reduced photoreceptor degeneration. Generally, IL-1β and TNFα are multifunctional proinflammatory cytokines with effects dependent on the timing and dosage. ^26,27 In this study, only acute blockade of TNFα, but not IL-1β, with specific blocking antibody suppressed the RD-induced photoreceptor degeneration. TNFα also has multifunctional roles in the neuronal homeostasis and neuropathology. ²⁸ Previously, we have shown that the expression of TNFα after RD is biphasic (peaking at 1 and 6 hours after RD) ¹² and that the source of TNFα is primarily via recruited monocytes and resident microglia and to a lesser extent retinal neurons of all types. ¹² Up to now, TNFα has been shown to be a critical mediator for the cytotoxic roles of neurons in various neurodegenerative diseases, including multiple sclerosis, Parkinson's disease, Alzheimer's disease, ¹⁶ and glaucoma. ²⁹ On the other hand, TNFα has a neuroprotective role against neuronal damage ³⁰ including retinal ganglion cell death after axotomy by suppressing the potassium channel via channel phosphorylation. ³¹ In this model of RD, we demonstrated the cytotoxic roles of TNFα on RD-induced photoreceptor degeneration. These data suggest that anti-TNFα or blockade of TNFα receptors may have beneficial effects in the treatment of ocular diseases associated with RD. 

The in vitro cytotoxic effect of TNFα on photoreceptors was detectable at a concentration as low as 0.1 ng/mL (Fig. 3). This cytotoxic concentration of TNFα was very similar to the concentration of TNFα (0.095 ng/mL) in the rodent eye after RD. ¹² In the vitreous sample of human eyes with RD, the concentration of TNFα was reported to be 2 to 22.4 pg/mL ¹⁸ and 2.52 to 32.26 pg/mL. ³² In contrast, the subretinal administration of 500 ng/μL of TNFα had no effect on photoreceptor apoptosis 24 hours after RD. ¹² Thus, the action of TNFα appears to be influenced by the concentration of TNFα, and the lower concentration (physiological dose) of TNFα results in a neurodestructive effect on RD-induced photoreceptor degeneration. 

Our data show that RD-induced photoreceptor degeneration depends on the TNFR2, but not the TNFR1, receptor. Generally, in neurons, TNFR1 has an intracellular death domain and its activation elicits caspase pathways that lead to neuronal cell death. TNFR2, on the other hand, activates the Akt signaling pathway and promotes cell survival. ³³ While the role of TNFR2 activation was opposite that in the visual systems, this finding suggests that the role of TNFR2 depends on the tissue. Microglia express both TNFR1 and -2, whereas oligodendrocytes and astrocytes primarily express TNFR1, ^34,35 and recruited macrophages express TNFR2. ³⁶ Critical roles of TNFR2 over TNFR1 on the neuronal cell responses had been reported in the axotomized facial motor nucleus through cytotoxic lymphocyte recruitment, ³⁷ and the suppression of oxidative stress in cultivated microglia, ³⁸ as well as the oligodendrocyte regeneration/proliferation and nerve remyelination in demyelinating diseases. ³⁹ We found that RD induced the caspase-8 activation, and the blocking antibody for TNFα suppressed the caspase-8 activation. The data suggest that TNFα activates caspase-8, which is downstream of the TNFα receptor. Thus, the effects of TNFα through its TNF receptors depend on the predominance of the receptor type expression for TNFR1 or -2 in each of the cell types. 

This study clearly showed that TNFα activated caspase-8 in the detached mouse retina. On the other hand, there have been some studies in which RD-induced Fas was found to be activated in rats. ^40,41 Interestingly, TNFα enhanced the Fas-mediated apoptosis of T cells in the eye. ⁴² Thus, TNFα may be associated with the FAS pathway in the process of RD-induced photoreceptor degeneration. 

The blockade of TNFα had no effect on RD-induced glial activation; however, the recruitment of monocytes was significantly suppressed. We previously reported that recruited monocytes played a critical role in RD-induced photoreceptor degeneration. ¹¹ Moreover, in primary retinal cell cultures, TNFα had a cytotoxic effect on the photoreceptors without macrophage recruitment. Taken together, the neurotoxic roles of TNFα may exert a direct effect on photoreceptors as well as an indirect effect via the recruitment of monocytes. 

In conclusion, we demonstrate that TNFα upregulation after RD plays a critical role in photoreceptor degeneration. The neurotoxic effects of TNFα on photoreceptors are mediated through its chemotactic properties, which lead to monocyte recruitment and monocyte-generated oxidative stress and possibly by a direct effect of TNFα on the photoreceptors. Blockade of TNFα and/or its receptors may provide new therapeutic avenues to treat photoreceptor degeneration in the setting of RD and of other retinal disorders that share common features. 

Footnotes

 Supported by an Alcon Research Award (JWM), a Bausch & Lomb Vitreoretinal Fellowship (TN), Grants-in-Aid 21659395 and 22689045 from the Ministry of Education, Science, and Technology of Japan (TN), the Uehara Memorial Research Foundation, the Takeda Research Foundation and Imai Glaucoma Research Foundation (TN), and National Eye Institute Grant EY014104 (MEEI Core Grant).

Footnotes

 Disclosure: T. Nakazawa, None; M. Kayama, None; M. Ryu, None; H. Kunikata, None; R. Watanabe, None; M. Yasuda, None; J. Kinugawa, None; D. Vavvas, None; J.W. Miller, None

The authors thank Norman Michaud and Sreedevi Mallemadugula for technical assistance.