Microglia-derived IL-1β promotes chemokine expression by Müller cells and RPE in focal retinal degeneration

Expression of retinal IL-1β in relation to chemokine up-regulation and macrophage infiltration following photo-oxidative damage

Retinal expression of Il-1β and genes associated with inflammasome assembly and activation (Casp1, Casp8, and Nlrp3) were assessed by qPCR following 24 h photo-oxidative damage (Fig. 1a). Il-1β was dramatically up-regulated after 24 h photo-oxidative damage, consistent with our prior reports [13], and in concert with expression of Casp1, Casp8, and Nlrp3 (P < 0.05, Fig. 1a). Immunoreactivity for IL-1β was not detected in dim-reared control retinas, but after photo-oxidative damage IL-1β immunoreactivity was evident on some ramified IBA1+ microglia/macrophages in the ONL and subretinal space (Fig. 1b-c, yellow). IBA1+ cells located in the choroid did not exhibit IL-1β immunoreactivity (Fig. 1c).

Comparison of Il-1β expression with changes in retinal Ccl2, Cxcl1 and Cxcl10 (Fig. 1d) shows a correlation between Il-1β and chemokine expression. Over the 24 h time-course of photo-oxidative damage (3, 6, 12, 17, and 24 h), Il-1β expression was markedly upregulated after 6 h, and increasing Il-1β expression was associated with an upregulation of Ccl2, Cxcl1, and Cxcl10, with all markers reaching peak expression at 24 h (P < 0.05; one-way ANOVA). Consistent with previous reports [28, 29] we also observed incursions of IBA1+ macrophages into the ONL and subretinal space by 24 h of photo-oxidative damage (P < 0.05, Fig. 1e-g).

Effect of IL-1β suppression on photoreceptor death, macrophage accumulation, and chemokine expression

Inhibition of IL-1β and its effect on photo-oxidative retinal damage was ascertained using both siRNA and antibody neutralisation approaches (Figs. 2 and 3). Intravitreal injection of the Il-1β-specific siRNA induced a 1.8 fold reduction in the expression of retinal Il-1β at 24 h photo-oxidative damage, compared to negative control siRNA (P < 0.05, Fig. 2a). Animals injected with Il-1β siRNA had ~60% fewer TUNEL+ photoreceptors 24 h post-exposure to photo-oxidative damage compared to controls (P < 0.05, Fig. 2b). In experiments where IL-1β was neutralised using the antibody, injected intravitreally prior to photo-oxidative damage, there was an 80% reduction in the number of TUNEL+ photoreceptors in the retina, compared to those injected with an isotype antibody control (P < 0.05, Fig. 2c-e). Counts of IBA1+ macrophages confirm that inhibition of IL-1β with either siRNA or neutralising antibody reduces the number of IBA1+ macrophages in the outer retina (ONL and subretinal space), compared with the respective controls (P < 0.05, Fig. 2f-h).

We then sought to determine the effect of IL-1β suppression on the retinal expression of chemokines Ccl2, Cxcl1, and Cxcl10 (Fig. 3a). In both modes of IL-1β inhibition, there was a significant reduction in the expression of Ccl2 and Cxcl1 compared to controls (P < 0.05, Fig. 3a). While Il-1β-specific siRNA did not modify expression of Cxcl10, antibody neutralisation of IL-1β was effective in reducing Cxcl10 expression (P < 0.05). By use of in situ hybridisation, we also confirmed that Ccl2 mRNA was present in vimentin-immunoreactive Müller cell processes after 24 h photo-oxidative damage (Fig. 3d-e; arrows), and that Ccl2 mRNA labelling was reduced in IL-1β-inhibited retinas compared to controls (Fig. 3b-c; arrows). Ccl2 mRNA was not detected in RPE cells (Fig. 3b-c), consistent with our previous findings [13, 28]. We detected Cxcl1 mRNA labelling in the INL (Fig. 3f; arrows) and RPE layer (Fig. 3 h; arrows) after photo-oxidative damage, which was reduced in retinas where IL-1β had been inhibited via neutralising antibody (Fig. 3 g, i). INL staining for Cxcl1 mRNA correlated with vimentin-immunoreactive Müller cells (Fig. 3 j-k; arrows), consistent with our previous report [13].

Effect of IL-1β stimulation on chemokine expression in Müller and RPE cell cultures

The capacity for IL-1β to stimulate chemokine up-regulation in Müller cells and RPE cells was assessed in MIO-M1 and ARPE-19 cell cultures, respectively (Fig. 4). Both cultures were incubated with 10 ng/mL IL-1β protein for 12 h, at which point we observed dramatic increases in the expression of Ccl2, Cxcl1 and Cxcl10 in MIO-M1 cells compared to unstimulated control wells (P < 0.05, Fig. 4a). To discount the possibility that the IL-1β-induced up-regulation of chemokines was a result of cell stress/death, we conducted an MTT assay on the cultures after IL-1β stimulation (Fig. 4b). We determined that both cultures exhibited no reduction in viability as a result of IL-1β stimulation compared to controls (P > 0.05). We also found that both MIO-M1 and ARPE-19 cells express Il-1r1 and Il-1rap receptor genes necessary for IL-1β signal transduction (Fig. 4c-d).

Changes in chemokine expression and macrophage infiltration following intravitreal delivery of IL-1β protein

Finally, we investigated the effect of IL-1β protein administered intravitreally on the expression of chemokines and the accumulation of macrophages over a 24 h period (Fig. 5). The data show a broad up-regulation of retinal Ccl2, Cxcl1, and Cxcl10 (Fig. 5a), up to 24 h post-injection compared to PBS-injected controls. This was particularly evident for Cxcl1, which by 6 h-post-injection increased ~22 fold (P < 0.05), although was somewhat reduced by 12 and 24 h. The localisation and number of IBA1+ macrophages at 24 h after the injection of IL-1β protein (Fig. 5b-f) demonstrated a significant increase in the total number of retinal macrophages present compared to the PBS-control group (P < 0.05; Fig. 5b). IBA1+ cells were observed predominantly in the GCL and optic nerve head (Fig. 5e-f; arrows), and comprised a population with a rounded (activated) rather than ramified (resting) morphology. PCR conducted on isolates of control rat retinal tissue confirmed the presence of Il-1r1 and Il-1rap genes (Fig. 5 g-h).

Animals and photo-oxidative damage

All experiments were conducted in accordance with the ARVO Statement for Use of Animals in Ophthalmic and Vision Research, and had approval from the Australian National University (ANU) Animal Experimentation Ethics Committee (Ethics ID: A2014/56). Adult Sprague–Dawley (SD) albino rats aged were used for all experiments. Animals were born and reared under dim light conditions (5 lux) prior to photo-oxidative damage. For the photo-oxidative damage paradigm, animals were placed into transparent Perspex open-top cages under a light source (COLD F2, 2x36W, IHF, Thorn Lighting, Australia) at 1000 lux for either 3, 6, 12, 17, or 24 h (hrs), with access to food and water ad libitum. Following light exposure, animals were immediately euthanized using an overdose of barbiturate via an intraperitoneal injection (Valabarb; Virbac, NSW, Australia). For each animal, eyes were processed for either cryosectioning or RNA extraction, according to protocols detailed in our previous publications [29].

Intravitreal injections

Intravitreal injections were performed as described in detail previously [47], wherein animals were anaesthetised using an intraperitoneal injection of ketamine (100 mg/kg; Troy Laboratories, NSW, Australia) and xylazil (12 mg/kg; Troy Laboratories). Injections consisted of either siRNA- or antibody-based IL-1β inhibitors, or IL-1β protein.

A neutralising antibody to IL-1β (Cat# AF-501-NA, R&D Systems, Minneapolis, MN) was administered intravitreally to rats immediately prior to photo-oxidative damage. A 3 μL solution containing either anti-IL-1β or an isotype control antibody was injected into individual animals, which equated to a delivery of 0.6 μg of antibody per eye. After intravitreal injections, the animals were immediately transferred to individual cages designed to allow light to enter unimpeded. Animals were exposed to photo-oxidative damage for 24 h, during which corneal hydration was maintained though application of a synthetic tear gel (GenTeal Gel; Novartis, NSW, Australia) until the animals awoke.

RNA-interference (RNAi) was conducted using Il-1β-specific siRNA (Cat# s127941; Thermo Fisher Scientific, Waltham, MA, USA), while a scrambled negative siRNA (Cat# 12935300, Stealth RNAi Med GC; Thermo Fisher Scientific) served as a control, which were encapsulated using a cationic liposome-based formulation (Invivofectamine 3.0 Reagent; Thermo Fisher Scientific) according to the manufacturer’s instructions. To purify and concentrate the siRNA formulation, the samples were centrifuged at 4000 g through an Amicon Ultra-4 Centrifugal Filter Unit (Merck Millipore, MA, USA). The final concentration of the encapsulated siRNA formulation was 1 μg/μl in endotoxin-free 0.1 M PBS. For injection, animals were anaesthetised in the same fashion as the antibody neutralisation series. 3 μl of either Il-1β or negative siRNA was then intravitreally delivered to both eyes of each animal, which equated to a final dosage of 3ug siRNA per eye. Animals were then exposed to 24 h photo-oxidative damage under the same parameters as the antibody neutralisation cohort.

IL-1β protein was administered to rats following the same intravitreal methodology as described in the inhibition experiments. Recombinant IL-1β protein (Cat# AF-501-RL-010, R&D Systems) was injected at final concentration of 10 ng per eye, as established in a previous study [48]; injections with only the PBS vehicle served as controls. Animals were euthanized at 3, 6, or 12 days post-injection, with eyes then processed as described in the previous section.

RPE and Müller cell cultures

Immortalised human cell lines MIO-M1 (Müller 1 Moorefields Institute of Ophthalmology; Dr A. Limb, Institute of Ophthalmology, University College London) and ARPE-19 (ATCC CRL-2302, American Tissue Culture Collection, Manassas, VA) were used to study responses to IL-1β stimulation. Cell lines were authenticated by CellBank, Australia. The cells were grown in Dulbecco's Modified Eagle Medium (DMEM; Life Technologies, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS; Life Technologies) and 3 mM L-glutamine (Life Technologies), in a humidified atmosphere consisting of 95% air and 5% CO₂ at 37 °C. The cells were passaged by trypsinization every 3 to 4 days.

The MIO-M1 and ARPE-19 lines were incubated in IL-1β protein to assess its effect on their expression of chemokines. Cells were first seeded to a density of 1 × 10⁵ cells on 24-well plates and left to recover for at least 48 h. 24 h prior to experimentation, the cells were placed in serum-deficient DMEM containing 1% FBS. IL-1β protein (Cat# 201-LB, R&D Systems) was then added to the culture medium at a concentrations of 10 ng/mL, as per previous literature [49, 50], and then incubated for 12 h. The cells were then harvested for either RNA extraction and PCR, or an MTT assay to verify cell viability. RNA was extracted from each sample well using a retinal RNA extraction protocol that we have established previously [28], with slight modifications for cell culture. The MTT assay was performed with a kit supplied by Roche (Cell Proliferation Kit I, Roche Applied Science, Penzberg, Germany) following the supplied instructions. Following IL-1β stimulation, MTT reagent was added to the sample wells and left to incubate for 4 h, after which a solution of 0.04 M HCL in isopropanol was then added to each well to dissolve the resulting formazan crystals. The absorbance for each sample was then read at 570 nm on a TECAN Infinite 200 PRO (TECAN Seestrasse, Männedorf, Switzerland), and quantified as a percentage relative to unstimulated culture samples.

TUNEL and immunohistochemistry

Retinal cryosections were stained for apoptotic cells using a terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL) kit (Roche Applied Science) and following our previous methodology [51, 52]. To quantify photoreceptor cell death, TUNEL+ cells in the ONL were counted throughout the full length of each section cut in the parasagittal plane (supero-inferior). For each animal, technical duplicates were counted, and these counts were averaged for each experimental group.

Polymerase chain reaction (PCR)

In preparation for quantitative real-time PCR (qPCR) and standard PCR, the RNA from retina or cell culture samples was synthesised into cDNA using a Tetro first-strand cDNA Synthesis Kit (Bioline Reagents, London, UK), as described in our previous investigation [13].

Standard PCR was conducted from cDNA synthesised from cells cultures or retinal homogenates, using primers specific to Il-1β receptor-related genes in human: Il-1r1 (F: 5’ ATCGTGATGAATGTGGCTGA 3’; R: 5’ TCTCATTAGCTGGGCTCACA 3’), Il-1rap (F: 5’ CGTTTCATCTCACCAGGACTC 3’; R: 5’ CCAAACCTCATTGCGAGAAT 3’), or rat: Il-1r1 (F: 5’ ACATTCGCAGCTGTCCTCTT 3’; R: 5’ TGTGCTCTTCAGCCACATTC 3’), Il-1rap (F: 5’ TCATACCGCCAAGGTACACA 3’; R: 5’ GGGCTCAGGACAACAATCAT 3’). The primers designed using the Primer3 web-based design program [54]; both transverse an intron splice site to avoid genomic amplification. The PCR was performed using MyTaq DNA Polymerase (Bioline) as per the manufacturer’s instructions, and the presence and specificity of the PCR product were inferred by gel electrophoresis.

In situ hybridisation

Ccl2 and Cxcl1 were cloned from PCR products (550-bp and 504-bp amplicons respectively) using cDNA synthesised from retinal RNA (as described above). Digoxigenin (DIG)-labelled riboprobes were then prepared as described in our previous publication [13]. In situ hybridisation on retinal cryosections was carried out according to our established methodology [55]; briefly, each riboprobe was hybridised to sections overnight at 55 °C, then was washed in decreasing concentrations of saline sodium citrate (pH 7.4) at 60 °C. The bound probe was visualised with either NBT/BCIP or HNPP/Fast-Red (Roche Applied Science).

Quantitative and statistical analysis

Graphing and statistical analysis for this study was performed using Prism 7 (GraphPad Software, CA, USA). Statistical analysis was conducted using a Students t-test for instances of single comparisons. For assessing trends or multiple comparisons over protracted time-courses, a Kruskal–Wallis one-way analysis of variance (ANOVA) with Dunn’s multiple comparison post-test was applied, as per our previous investigation [13]. A P value of < 0.05 was considered statistically significant.