Adalimumab Reduces Photoreceptor Cell Death in A Mouse Model of Retinal Degeneration

Growing evidence suggests that inflammation is involved in the progression of retinitis pigmentosa (RP) both in patients and in animal models. The aim of this study was to investigate the effect of Adalimumab, a monoclonal anti-TNFα antibody, on retinal degeneration in a murine model of human autosomal recessive RP, the rd10 mice at postnatal day (P) 18. In our housing conditions, rd10 retinas were seriously damaged at P18. Adalimumab reduced photoreceptor cell death, as determined by scoring the number of TUNEL-positive cells. In addition, nuclear poly (ADP) ribose (PAR) content, an indirect measure of PAR polymerase (PARP) activity, was also reduced after treatment. The blockade of TNFα ameliorated reactive gliosis, as visualized by decreased GFAP and IBA1 immunolabelling (Müller cell and microglial markers, respectively) and decreased up-regulation of TNFα gene expression. Adalimumab also improved antioxidant response by restoring total antioxidant capacity and superoxide dismutase activity. Finally, we observed that Adalimumab normalized energetic and metabolic pattern in rd10 mouse retinas. Our study suggests that the TNFα blockade could be a successful therapeutic approach to increase photoreceptor survival during the progression of RP. Further studies are needed to characterize its effect along the progression of the disease.

Tumor necrosis factor alpha (TNFα ) is a pleiotropic cytokine essential for the induction and maintenance of the inflammatory immune responses. It is a well characterized mediator of cellular activities including proliferation, survival, differentiation, inflammation and cell death. In the retina, it is likely secreted from activated macrophages, astrocytes, microglia and Müller glial cells. TNFα can trigger several well-characterized death-promoting (caspase-dependent and caspase-independent cell death) and survival-promoting pathways, depending upon the predominating signalling pathway in the particular cell type 13 . In the eye, TNFα appears to have a role in the pathogenesis of inflammatory diseases such as uveitis 14 , as well as in retinal degenerations such as diabetic retinopathy, age-related macular degeneration and, recently, in RP 11,[15][16][17] .
Several anti-TNFα agents (Infliximab, Adalimumab, Etarnecept, Golimumab and Certolizumab pegol) have been developed and approved for clinical use in inflammatory diseases such as rheumatoid arthritis, psoriasis and ankylosing spondylitis 18 . These anti-TNFα agents are antibodies against TNFα or TNFα receptor. In Ophthalmology, they are widely used as an alternative to traditional immunosuppressive treatments in non-infectious uveitis. More recently anti-TNFα agents have been used for retinal diseases such as neovascular age-related macular degeneration, diabetic macular edema and retinal vein occlusions [19][20][21] .
To further explore the in vivo potential benefits of blocking TNFα we adopted the rd10 mouse, a model of human autosomal recessive RP. This mouse carries a mutation on the β subunit of the cGMP PDE6 gene (Pde6β) that produces retinal degeneration 2 .
In this study, we first analysed the early stages of retinal degeneration and the profile of TNFα expression in rd10 mice under our housing conditions. We observed a peak of TNFα gene expression and photoreceptor cell death at postnatal day (P) 18. Therefore, we studied the effect of Adalimumab, a monoclonal antibody against TNFα , on the progression of the retinal degeneration at this age. We observed that Adalimumab prevented TNFα upregulation, reduced photoreceptor cell death, slowed microglial and Müller cell activation, improved antioxidant response and ameliorated the energetic and metabolic dysfunction at P18. Based on these results we suggest that anti-TNFα therapies could be promising treatments to improve photoreceptor cell survival in humans with RP. However, further studies are needed to investigate the molecular mechanisms involved in the protective effect of anti-TNFα agents and their long-term effect.

Results
Temporal progression of retinal degeneration in rd10 mice. Light exposure accelerates progression of retinal degeneration in many animal models of RP 22,23 . Hence we first investigated the time course of retinal degeneration in rd10 mice from P13 to P20 stabulated under light intensities between 98 and 131 lux. Sytox Green nuclear staining showed a decline of the number of photoreceptor cell nuclei rows in the outer nuclear layer (ONL) in rd10 mice. As expected, we did not find a significant difference between control and rd10 mice at P13. But the number of cell nuclei rows in the ONL dropped abruptly from eleven to five from P15 to P20 (Fig. 1A). Degeneration was evident at P18 (Fig. 1D).
Then, we assessed photoreceptor cell death by TUNEL staining which detects DNA breaks in both apoptotic and necrotic nuclei. From P15 to P20 there was a significant increase of TUNEL-positive cells in rd10 mice, reaching a peak at P18 (12.5 ± 2.4 TUNEL-positive cells/normalized ONL area (five-fold increase between P15 and P18), Mann-Whitney U test, p < 0.0001) (Fig. 1B,D, and Supplementary  Table S1).
It has been previously described that TNFα is upregulated in the retina of rd10 mice previous to photoreceptor cell death peak 11,24 . Therefore, we analysed gene expression of TNFα during the time course of retinal degeneration in our housing conditions. As shown in Fig. 1C, TNFα was upregulated from P13 to P20 with the highest expression at P18 (6.1 ± 1.2 relative expression (six-fold increase over control retinas), Mann-Whitney U test, p = 0.0002) concomitantly with the peak of TUNEL-positive cells. At P18 a significant increase of TNFα protein level was also observed in the retina of rd10 mice (mean = 65.5 pg/mg protein, confidence interval [25.1-170.9], generalized least squares model, p = 0.0034) compared to control retinas (mean = 17.1 pg/mg protein, confidence interval [15.8-18.5]).
In view of these results, P18 was chosen as the end point for the ensuing analysis of the potential protective effect of Adalimumab.

Adalimumab reduced photoreceptor cell loss and PARP activation. To determine whether
Adalimumab treatment protects against photoreceptor cell death, we quantified the normalized ONL thickness ratio and the number of rows of nuclei in the ONL at P18. Normalized ONL thickness decreased slower in Adalimumab-treated rd10 retinas (reduction of 20% over control retinas) than in vehicle-treated rd10 retinas (reduction of 40% over control retinas) ( Fig. 2A and Supplementary Table  S1). A mixed linear model analysis also revealed that Adalimumab significantly reduced (p = 0.047) Relative TNFα gene expression from P13 to P20 in rd10 mice. Values are the mean ± SEM of, at least, five retinas per group. Values that are significantly different are indicated by asterisks *p < 0.05, **p < 0.001, ***p < 0.0001 (Mann-Whitney U test). (d) Photoreceptor cell death was visualized by TUNEL (red) in Sytox Green-counterstained in retinal sections from control mice at P18 and from rd10 mice at P15, P18 and P20. photoreceptor cell loss (7.6 ± 0.4 number of rows of nuclei in the ONL) compared to vehicle-treated rd10 retinas (5.9 ± 0.3 number of rows of nuclei in the ONL) ( Fig. 2B and Supplementary Table S1).
We previously described that Infliximab, other monoclonal anti-TNFα antibody, significantly reduced the number of TUNEL-positive cells in Zaprinast-treated explants of porcine retina 25 . In the current study we observed that intraperitoneal administration of Adalimumab significantly decreased (∼ 78%, p = 0.048) the number of TUNEL-positive cells in the ONL at P18 (2.7 ± 0.4 TUNEL-positive cells) compared to vehicle-treated rd10 retinas (12.5 ± 2.4 TUNEL-positive cells) (Fig. 3A,C, and Supplementary  Table S1). Therefore the blockade of TNFα was preventing photoreceptor cell loss in rd10 mice. However, the mechanisms of TNFα -induced cell death remained unclear.
Paquet-Durand et al. (2007) demonstrated non-apoptotic mechanisms involved in photoreceptors cell death such as the activation of PARP which led to necroptosis in a similar mouse model of RP 26 . To assess whether TNFα inhibition prevented PARP activation, we visualized the content of PAR polymers (an indirect marker for PARP activity) and its retinal location at P18. As shown in Fig. 3B, an intense and nuclear PAR staining was found in photoreceptor cells of vehicle-treated rd10 retinas. In contrast, Adalimumab substantially decreased (up to 90%, p < 0.001) nuclear PAR staining in photoreceptor cells (2.1 ± 0.6 PAR-positive cells), compared to vehicle-treated rd10 retinas (25.1 ± 5.5 PAR-positive cells) (Fig. 3C). Taking together these results suggested that Adalimumab could slow down photoreceptor cell death through PARP pathway inhibition in rd10 mice. Further studies will be designed to test this hypothesis.
Adalimumab reduced reactive gliosis. Chronic microglia activation is associated with various neurodegenerative diseases including RP. We investigated whether Adalimumab treatment reduced microglial activation in rd10 mouse retinas. Microglial cells were identified by specific labelling with Iba1, a microglia/macrophage-specific calcium-binding protein. The level of migration from the inner to the outer layers of retina was employed as an indicator of microglial activation. Iba1 immunostaining revealed that microglia was located in all layers, especially in outer layers in vehicle-treated rd10 retinas. Conversely, after intraperitoneal administration of Adalimumab, microglia was preferentially located in the inner layers of the retina. A few microglial cells were observed in the outer plexiform layer but not in the ONL in Adalimumab-treated rd10 retinas (Fig. 4A). To quantify microglial behaviour, we performed a weighted analysis where values close to 1 indicate a high rate of microglial migration to outer layers. Statistical analysis confirmed that microglial activation was reduced (p = 0.006) in Adalimumab-treated rd10 retinas (0.28 ± 0.03) compared to vehicle-treated rd10 retinas (0.43 ± 0.02) at P18 (Fig. 4C and Supplementary Table S1).
Gliosis commonly involves upregulation of the glial fibrillary acidic protein (GFAP), an intermediate filament protein, in Müller glial cells. We also studied whether the blockade of TNFα with Adalimumab ameliorates glial activation in rd10 mice. Figure 4B shows that GFAP was upregulated in vehicle-treated rd10 retinas (p < 0.001) compared to control retinas. Adalimumab tended to decrease (up to 50%) (p = 0.27) GFAP-positive immunolabelling compared to vehicle-treated rd10 retinas at P18 (Fig. 4C).
After evaluating the effect of Adalimumab in prevention of reactive gliosis, we analysed whether Adalimumab affected gene expression of the inflammatory mediators TNFα and IL-6 by real time PCR at P18. After Adalimumab treatment, TNFα expression was significantly reduced (p = 0.006) ( Fig. 4C and Supplementary Table S1). At P18 we did not observe changes in IL-6 expression neither in vehicle-treated rd10 retinas nor in Adalimumab-treated rd10 retinas (data not shown). Adalimumab ameliorated energetic and metabolic dysfunction. It has been described a coordinated downregulation of several metabolic pathways at the onset of retinal degeneration, affecting amino acid biosynthesis, glycerol phospholipid metabolism and oxidative phosphorylation in 14-d-old Rh1P37H retinas, a Drosophila model expressing the equivalent of the most common RP-linked mutation, RhoP23H 27 . First, we assessed whether rd10 mice presented a deregulated expression of enzymes of energy metabolism and mitochondrial structure and dynamics at P18 by quantitative reverse phase protein microarrays (RPPM) 28 using validated antibodies (Supplementary Figure S1). We analysed the expression of proteins of glycolysis (GAPDH, LDHA), electron transport and oxidative phosphorylation (OXPHOS) (NDUFS3, SDHB, CORE 2, COXII and β -F1-ATPase), mitochondrial structure (HSP60) and dynamics (OPA1), β -oxidation (HADHA), the pentose phosphate pathway (G6PDH), the antioxidant response (SOD1, SOD2 and catalase) and the expression of β -actin as reference protein (Supplementary Figure S1).
At the onset of RP, the expression of most of these markers was significantly downregulated in retinas of rd10 mice leading to a global energy depletion that might contribute to cell death (Table 1). No significant differences in β -actin were found between control (0.64 ± 0.05 arbitrary units (a.u.)/ng total protein) and rd10 retinas (0.71 ± 0.06 a.u./ng total protein). Hence we observed an energy depletion and mitochondrial dysfunction during photoreceptor cell death. As shown in Table 1, intraperitoneal administration of Adalimumab significantly ameliorated this phenotype in retinas from rd10 mice at P18. Adalimumab improved antioxidant response. It has been reported that inflammatory processes and oxidative stress are closely linked in several pathologies. The contribution of oxidative damage to cell death in rd10 mouse retinas has been previously described [29][30][31] . In particular, anti-TNFα treatments have been demonstrated to have a positive effect on improving antioxidant response and reducing the  production of reactive oxygen species (ROS) in several pathological situations 32 . Anti-TNFα antibodies decrease cytokine concentrations, thus directly influencing ROS/RNS production by inflammatory cells or indirectly modulate ROS-stimulated signalling pathways activity. Therefore we evaluated whether Adalimumab improved the antioxidant response in retinas from rd10 mice. For this purpose, we measured the total antioxidant capacity (TAC), enzymatic activity of superoxide dismutase (SOD) and catalase (CAT) which play a major role in the first line of antioxidant defence. The TAC was reduced (p = 0.031) in vehicle-treated rd10 retinas (92.6 ± 7.8 nmol of Trolox/mg protein) compared to control retinas (119 ± 5.7 nmol of Trolox/mg protein). Adalimumab significantly increased TAC to 204.5 ± 11.6 nmol of Trolox/mg protein (p < 0.001) ( Fig. 5A and Supplementary Table  S1).

Discussion
TNFα inhibitors have been widely used for the treatment of different autoimmune disease, including ocular inflammatory disorders, with promising results [33][34][35] . Nevertheless, to our knowledge, we present the first study that evaluates the effect of TNFα inhibitors on the progression of the retinal degeneration in a murine model of RP. The results obtained in the present study revealed that intraperitoneal administration of Adalimumab, a recombinant human monoclonal antibody against TNFα , reduced retinal degeneration decreasing photoreceptor cell death and reactive gliosis at early stages of RP in rd10 mice. We used the rd10 mouse model of human autosomal recessive RP. This mutant mouse mimics the disease progression seen in some human conditions and offers an amenable therapeutic window. In general, rod photoreceptors start to degenerate between P16 and P20, with a maximum cell death around P21 and P25. By P60, rods are undetectable and only cones remain 2,36 . Under our housing conditions, we observed that photoreceptor degeneration started around P15, with maximum TUNEL-positive cells around P18. This time point of maximum cell death agreed with results recently described by Arango-Gonzalez et al. (2014) 37 . In our study light intensity was higher (98-131 lux) than light intensity described by other authors (around 60 lux) 38,39 . Therefore, the early onset of photoreceptor degeneration could be explained by light conditions.
In the present study, we demonstrated that intraperitoneal injections of Adalimumab displayed a neuroprotective effect on photoreceptors survival at the onset of their death in rd10 mice. The administration of 3 mg/kg Adalimumab from P9 to P17 contributed to preserve retinal structure reducing the loss of photoreceptors at P18. Further, Adalimumab strongly reduced the number of TUNEL-positive cells in the ONL. Based on several studies using different animal models of RP, it has been suggested that apoptosis is playing a minor role in retinal degeneration 37 . However, non-apoptotic mechanisms seem to prevail as major contributors to cell death including accumulation of cGMP and activation of histone deacetylase (HDAC), PARP (PAR accumulation) and calpains 40,41 .
On the other hand, TNFα is considered both apoptotic and necroptotic inducer mainly through its binding to the receptor TNFR1. TNFα binds to TNFR1 leading to the recruitment of multiple proteins that can trigger different downstream pathways as activation of the transcription factor NF-κ B, MAPKs (mitogen-activated protein kinases), caspase 8 promoting cell survival, apoptosis or necroptosis, etc 42 . The two most extensively studied ways of necroptosis are initiated by TNFR1 and by PARP pathway. PARP pathway is initiated by overactivation of the DNA repair enzyme PARP-1, resulting in the massive synthesis of PAR and, eventually in necroptosis 43 . An evident controversial exists concerning to the link between TNFα and PARP pathway during necroptosis. Previously, PARP pathway has been suggested as a key element of TNFα -mediated necroptosis 44 . But other authors propose that TNFα -induced necroptosis and PARP pathway represent distinct and independent routes to necroptosis 45,46 . In our previous study we found that TNFα inhibition did not prevent PAR accumulation but also increased it supporting the idea that PARP pathway was independent of TNFα -associated pathways in cultures of porcine retina exposed to Zaprinast 25 . In the current study we confirmed that there was an excessive nuclear PAR content in vehicle-treated rd10 mouse retinas that was effectively prevented by Adalimumab treatment. Therefore our results suggest that PARP activation is involved in TNFα -mediated photoreceptor cell death in rd10 mice. As described above, PARP activation could be consequence of excessive cGMP accumulation that in turns would activate HDAC. PARP activation could be also consequent of the inflammatory process. We should confirm whether reduced inflammation could be the primary responsible for down regulation of PARP activity or could be secondary to reduced cell death. Thus, the mechanistic interpretation of these results is complicated and deserves further studies.
Upregulation of TNFα and activation of microglial cells are early events involved in photoreceptor death in murine models of RP 11,47 . We corroborated microglial activation and migration to outer retina in rd10 mice. However, we did not visualize Iba1-positive cells in the ONL in retinas from rd10 mice treated with Adalimumab. Thus, inhibition of TNFα reduced excessive microglial activation and migration of microglial cells to the outer layers. Microglial activation is a hallmark of neuroinflammation. It is associated with various neurodegenerative diseases including retinal degenerations 10,48 . Some therapeutic approaches are designed to inhibit microglial activation. For instance, minocycline attenuated photoreceptor cell death in different animal models of retinal degeneration 49,50 and improved vision in patients with diabetic macular edema 51 . Around 80 clinical trials are in progress to investigate the effect of minocycline in several pathologies including retinal degenerations (ClinicalTrials.gov, Identifiers: NCT02140164 and NCT01468844).
Recent studies showed that Adalimumab decreased glial cell activation (overexpression of GFAP) and preserved retinal organization in organotypic cultures of porcine neuroretina exposed to TNFα 52 . Activated Müller cells can release antioxidants, growth factors and cytokines, including TNFα , contributing to retinal regeneration or degeneration. Müller cells are activated in models of RP resulting in overexpression of GFAP, activation of ERK (extracellular signal-regulated kinase), translocation of Müller cell bodies to the outer retina and thickening of their processes 25,53 . In the present study, Adalimumab tended to ameliorate GFAP overexpression in rd10 retinas.
Adalimumab specifically blocks the interaction of soluble TNFα , and to a lesser extent, transmembrane TNFα with its receptors, inhibiting downstream cellular signalling mechanisms 21 . We found that blocking TNFα tends to normalize the increased TNFα production, as described in other pathological situations 54 . TNFα blockade prevented the activation of signalling pathways involving the activation of p38 MAPK or NF-κ β 55,56 . MAPKs such as ERK, JNK (c-Jun N-terminal kinase) and p38, are involved in many inflammatory and degenerative processes including retinal cell death 53 . Thus, we speculate that Adalimumab could attenuate the inflammatory process by interfering downstream signalling of TNFα , at the level of p38 MAPK or ERK activation. Further studies are needed to confirm this hypothesis.
Inflammation and oxidative stress are closely linked processes in several pathological situations. It has been widely suggested that oxidative stress contributes to the pathogenesis of RP in animal models and patients [29][30][31] . For instance, we previously found that RP patients present a reduced antioxidant response 9 . In the current study we assessed whether down regulation of inflammation (TNFα and reactive gliosis) induced by Adalimumab also influenced antioxidant response. Adalimumab improved antioxidant defence machinery in rd10 mouse retinas. This treatment improved the total antioxidant capacity and total SOD activity probably by increasing SOD1 content in the retina of the rd10 mouse. It has been shown that SOD1 deficient mice are more sensitive to the damaging effects derived of oxidative insults 57,58 . Thus, SOD1 seems to be a key defender against oxidative damage in retina. It has been previously shown that Adalimumab can influence the antioxidant and oxidant status in other pathological situations such as psoriasis, rheumatoid arthritis, ankylosing spondylitis or uveitis 32,59,60 . For instance, Adalimumab is capable of upregulating SOD activity and downregulating catalase activity in mesenchymal stem cells from skin of patients with psoriasis 59 and to restore glutathione (GSH) content and glutathione peroxidase activity in eyes of a uveitis experimental model 60 .
The retina is an organ with one of the highest energy demands per tissue weight due to the photoreceptor cell metabolism. Glucose is metabolized through two major metabolic pathways: (1) glycolysis to obtain energy (ATP), reducing power (NADH) and pyruvate and (2) the pentose phosphate pathway (PPP) to obtain reducing power (NADPH) and ribose for nucleic acid biosynthesis. Under aerobic conditions, pyruvate supplies ATP, precursors of certain amino acids as well as NADH through the citric acid cycle or Krebs cycle (TCA cycle). The NADH generated by the TCA cycle is fed into the mitochondrial oxidative phosphorylation (OXPHOS) pathway. During OXPHOS, electrons are transferred from electron donors to electron acceptors such as oxygen in redox reactions releasing energy and finally ATP. Photoreceptors contain a dense concentration of mitochondria in their inner segments that provides the ATP for the phototransduction processes mainly through the OXPHOS pathway 61,62 . Part of the oxygen consumed in OXPHOS is transformed to reactive oxygen species which are converted to H 2 O by antioxidant enzymes. However, a deficient antioxidant defence results in oxidative damage that initiates a cascade of events resulting in mitochondrial dysfunction and cell death 63 . In this study we corroborated the downregulation of OXPHOS pathway observed at the onset of photoreceptor degeneration in a Drosophila model expressing the equivalent of the most common RP-linked mutation, RhoP23H 27 . We also observed significant alterations in other energetic pathways as glycolysis, pentose phosphate pathway and β -oxidation indicating global energy depletion in rd10 retinas at P18. We also detected mitochondrial alterations in proteins related to its dynamics and structure supporting the results reported by other authors 63,64 . Adalimumab ameliorated energetic and metabolic dysfunction in rd10 mice.
It is tempting to speculate that Adalimumab or other anti-TNFα agents could be a promising therapy for RP and other retinal degenerations. However, further studies are needed to investigate in detail the molecular mechanisms involved in the neuroprotective effect of Adalimumab and its long-term effect.

Methods
Animals and treatment. rd10 mice were used as human model of autosomal recessive retinitis pigmentosa. Wild-type C57Bl6 mice with the same genetic background as rd10 mice were used as control. Mice were kept under a 12 hours light/dark cycle, humidity and temperature controlled and with food and water supplied ad libitum. All cages were placed on the lower shelf of an IVC rack with light illuminance of 115 ± 7 lux (95% CI: 98-131). Mice were housed in the Animal Facility of Unitat Central d'Investigació (UCIM) of Valencia University. This study was carried out in accordance with guidelines from the European Union Guidelines for the Care (European Union Directive (2010/63/EU) and the Use of Laboratory Animals. All animal procedures and protocols were approved (A1361179906177) and monitored by the Committee of Ethics in Research of the Faculty of Medicine, University of Valencia.
To determine the profile of retinal degeneration at early stages in our housing conditions, untreated rd10 and C57Bl6 mice were euthanized at P13, P15, P18 and P20. The eyes were rapidly removed and processed as described below.
To evaluate the effect of Adalimumab, each rd10 mouse received one intraperitoneal injection of Adalimumab (Humira, Abbot Laboratories) saline solution at 3 mg/kg every three days starting at P9 and until P17. The dose of 3 mg/kg of Adalimumab was chosen based on previous studies with murine models of rheumatoid arthritis and uveitis 60,65,66 . No apparent side effects (e.g. opportunistic infections) were detected in animals treated with Adalimumab.
Control mice received the same volume of saline (vehicle) at the same time. At P18, treated and untreated rd10 mice and C57Bl6 mice were sacrificed. The eyes were rapidly removed and processed as described below. Tissue processing. To obtain retinal sections, the eyes were rapidly removed and fixed in 4% filtered paraformaldehyde for two hours at room temperature and cryoprotected in a sucrose gradient Scientific RepoRts | 5:11764 | DOi: 10.1038/srep11764 (15-20-30%). Eyes were frozen embedded in OCT and 10 μ m sections were cut in a cryostat (Leica CM1900, Nussloch, Germany).
For biochemical determinations and gene expression analysis, retinas were isolated, placed immediately into the appropriate buffer and stored at − 80 °C.

Retinal immunohistochemistry and TUNEL assay. Immunofluorescent staining procedures
were performed in 10 μ m cryosections. Sections were post-fixed in 4% filtered paraformaldehyde (Sigma-Aldrich, Madrid, Spain) in 0.1 M phosphate buffer pH 7.4 for 15 minutes at room temperature. Sections were pre-treated with citrate buffer pH 6.0 for epitope retrieval and incubated in blocking solution containing 5% normal goat serum, 1% bovine serum albumin and 0.25% Triton X-100 for one hour. Afterwards, they were incubated with primary antibody anti-PAR (1:200, Enzo Life Science, Madrid, Spain) as an indirect marker for PARP activity, anti-Iba1 (1:300, Wako Pure Chemical Industries Ltd., Osaka, Japan) and anti-GFAP (1:400, Sigma-Aldrich, Madrid, Spain) overnight at 4 °C. Sections were then incubated with the fluorescence-conjugated secondary antibody Alexa Fluor 647, 555 or 488 (Invitrogen, Life Technologies, Madrid, Spain) for one hour at room temperature. After labelling, the sections were mounted in Fluoromount-G (Southern Biotechnology, Birmingham, AL, USA) and observed under a fluorescence or confocal microscope.
To evaluate cell death, the terminal deoxynucleotidyl transferase dUTP nick and labelling (TUNEL) assay was used as previously described 67 . Corrected fluorescence of GFAP was quantified as previously described 25 . TUNEL and PAR-positive cells, microglial migration index and the corrected fluorescence of GFAP were quantified from four non-adjacent sections of at least five retinas for each experimental group.

Microscopy and quantification.
Isolation of total RNA and cDNA synthesis. Total RNA was isolated from frozen retina using RNeasy mini kit (Qiagen, Hilden, Germany) following the manufacturer's protocol. Then, cDNA was synthesized starting from 1 μ g of RNA by reverse transcription using the GeneAmp Gold RNA PCR Reagent kit (Applied Biosystems, Carlsbad, CA, USA) following manufacturer's instructions. Evaluation of antioxidant response. Retinas were homogenized in 5 mM phosphate buffer pH 7, 0.9% NaCl, 0.1% glucose, centrifuged at 10,000 × g (total antioxidant capacity (TAC) and catalase (CAT) activity) or 1,500 × g (superoxide dismutase (SOD) activity) for 15 minutes at 4 °C. Supernatants were used to determine TAC, CAT and SOD activities with commercial kits following Manufacturer´s Instructions (Cayman Chemical, Ann Arbor, MI, USA). Protein concentrations were measured by the bicinchoninic acid (BCA) protein assay. For total SOD activity assay one unit of SOD was defined as the amount of enzyme needed to exhibit 50% dismutation of the superoxide radical. For CAT activity assay Scientific RepoRts | 5:11764 | DOi: 10.1038/srep11764 one unit was defined as the amount of enzyme that will cause the formation of 1.0 nmol of formaldehyde per minute at 25 °C. SOD and CAT activities were expressed as U/mg protein. Retinal TAC levels were expressed as nmol/mg protein.
Retinal content of SOD1 (cytosolic), SOD2 (mitochondrial) and CAT enzymes were measured as described below 28 .

Reverse Phase Protein Microarrays (RPPM).
Thirty-six retinas were prepared for printing onto RPPM. Frozen retinas were grinded with liquid nitrogen and the tissue powder was extracted with 90 μ l of 50 mM Tris/HCl pH 8, containing 100 mM NaCl, 1 mM DTT, 1% (v/v) Triton X100, 0.1% SDS, 0.4 mM EDTA and a cocktail of protease (Roche) and phosphatase (Sigma) inhibitors. After protein extraction, samples were centrifuged at 15,000 g for 30 min at 4 °C. The protein concentration in the supernatants was determined with the Bradford reagent (Bio-Rad Protein Assay) using BSA as standard. For printing and processing, samples were diluted in PBS to a final protein concentration of 0.75 μ g/μ l. Serially diluted protein extracts (0-1 μ g/μ l) derived from HCT-116 and OVCAR8 carcinoma cells were also prepared to assess printing quality and the linear response of protein recognition by the antibodies used 28 . A solution of BSA (1 μ g/μ l) was also prepared for printing as internal negative control. Approximately, 1 nl volume of each sample was spotted in duplicate onto nitrocellulose-coated glass slides (FAST Slides, Scheleicher & Schuell BioScience, Inc.) using a BioOdyssey Calligrapher MiniArrayer printer (Bio-Rad Laboratories, Inc.) equipped with a solid pin (MCP310S) at constant humidity and temperature 28 . After printing, arrays were allowed to dry and further blocked in PBS-T containing 5% skimmed milk. After, the arrays were incubated overnight at 4 °C with the indicated concentrations of the following highly specific primary monoclonal antibodies (mAbs): β -F1-ATPase (0,4 μ g/ml) and anti-Hsp60 (4 ug/ml) 28 ; anti-LDHA (4 μ g/ml) 68  and anti-G6PDH (1/500; Thermo Scientific). After incubation, the arrays were processed and revealed as previously described 28 . The mean fluorescent intensity of the spots was quantified using FIJI software (N.I.H., USA) and converted into arbitrary units of expressed protein/ng of total protein in the tissue extract using the expression obtained in the linear plot of the HCT116 cell line as standard (see Supplementary Figure S1).

Statistical analyses.
Statistical analysis was performed using R software (version 3.1.2). Comparisons between different age categories were performed using Mann-Whitney U. A generalized least squares model was fitted to the log-transformed TNFα protein values to assess differences between the two experimental groups and also the effect of Adalimumab (ADA). Linear models were used to assess the association between ADA and the two experimental groups with the different response variables. To account for the non-independence of observations in the case of repeated measures per sample, a random intercept was added to the linear models with the variable sample as random factor. P values for the mixed linear models were computed using Satterthwaite approximation to degrees of freedom. Estimations for each variable in a linear mixed model are interpreted similarly than estimations from an ordinary linear model, that is, in an additive way. Negative values indicated a negative effect of being rd10 or having received Adalimumab. And positive values indicated a positive effect. Some response variables were logarithmically transformed to avoid negative values for the estimated control values. Result tables from the models are shown in Supplementary Table S1. Reverse Phase Protein Microarrays were analysed by One-way ANOVA and Newman-Keuls post-test. A p-value < 0.05 was considered statistically significant.