key: cord-0758602-3ccjsth7
authors: Du, Xuan; Penalva, Rosana; Little, Karis; Kissenpfennig, Adrien; Chen, Mei; Xu, Heping
title: Deletion of Socs3 in LysM(+) cells and Cx3cr1 resulted in age-dependent development of retinal microgliopathy
date: 2021-02-18
journal: Mol Neurodegener
DOI: 10.1186/s13024-021-00432-9
sha: 1ebc763aa491fd3f91529246a6e1f3fbbdc3a7bf
doc_id: 758602
cord_uid: 3ccjsth7

BACKGROUND: We generated a mouse model of primary microglial dysfunction by deleting two negative immune regulatory genes, Cx3cr1 and Socs3 (in LysM(+) cells). This study aimed to understand how primary microglial dysfunction impacts retinal neurons during aging. METHODS: The LysMCre-Socs3(fl/fl)Cx3cr1(gfp/gfp) double knockout (DKO), LysMCre-Socs3(fl/fl), Cx3cr1(gfp/gfp) and Socs3(fl/fl) mice were maintained up to 12 months. Eyes were collected and processed for immunohistochemistry of IBA-1, cone arrestin, secretagogin, PKCα and GABA. Brain microglia from DKO and WT mice were stimulated with LPS + IFN-γ or IL-4. The expression of TNF-α, IL-1β, IL-6, iNOS, IL-12p40, IL-23p19, CCL2, CCL5, CXCL2, IL-10, CD206 and Arg1 were examined by qRT-PCR and protein production was measured by Luminex assay. Retinal explants from C57BL/6 J mice were co-cultured with microglia from DKO or WT mice for 24 h, after which the number of cone arrestin(+) cells in retinal flatmount were quantified. RESULTS: In 3–5 month old mice, the number of microglia in retinal ganglion cell layer (GCL) and inner plexiform layer (IPL) were comparable in all strains of mice. The DKO mice had a significantly higher number of microglia in the outer plexiform layer (OPL) but significantly lower numbers of cone arrestin(+), secretagogin(+) and GABA(+) cells compared to Socs3(fl/fl) and single KO mice. During aging, 57% of the DKO mice died before 12 months old. The 10–12 months old DKO mice had significantly higher numbers of microglia in GCL/IPL and OPL than age-matched Socs3(fl/fl) and single KO mice. The aged DKO mice developed retinal pigment epithelial (RPE) dysmorphology accompanied by subretinal microglial accumulation. The number of photoreceptors, bipolar cells (Secretagogin(+) or PKCα(+)) and GABA(+) amacrine cells was significantly lower in aged DKO mice compared to age-matched Socs3(fl/fl) and single KO mice. Microglia from DKO mice showed significantly higher levels of phagocytic activity and produced higher levels of TNF-α, IL-6, CCL2, CCL5, CXCL2 and CXCL10 compared to microglia from Socs3(fl/fl) mice. Co-culture of retinal explants with LPS + IFN-γ or IL-4 pre-treated DKO microglia significantly reduced cone photoreceptor survival. CONCLUSIONS: The LysMCre-Socs3(fl/fl)Cx3cr1(gfp/gfp) DKO mice displayed primary microglial dysfunction and developed age-related retinal microgliopathy characterized by aggragated microglial activation and multiple retinal neuronal and RPE degeneration. TRIAL REGISTRATION: Not applicable. The article does not contain any results from human participants. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s13024-021-00432-9.

Microglia, the resident innate immune cells of the central nervous system (CNS), which includes the retina, play an important role in safeguarding neurons from exogenous and endogenous insults [1] [2] [3] . Microglia can detect damage-associated molecular patterns (DAMPs) released by diseased neurons or pathogenassociated molecular patterns (PAMPs) from invading microbes with pattern recognition receptors (PRRs) such as toll-like receptors (TLRs), NOD-like receptors (NLRs) and RIG-like receptors (RLRs). PRR activation can lead to a cascade of signalling transduction and upregulation of a variety of inflammatory gene expression. The primary role of microglial activation is to eliminate the insults (e.g., dead cells or invading pathogens), which therefore is neuroprotective [4, 5] . Under certain conditions, active microglia can promote neuronal regeneration [6] . Uncontrolled microglial activation, however, is detrimental and contributes to various neurodegenerative diseases such as Alzheimer's disease [7, 8] , Parkinson's disease [9] , age-related macular degeneration (AMD) [10] [11] [12] , diabetic retinopathy [13, 14] , inherited retinal degeneration [15] and glaucoma [16] . Targeting microglial activation is considered to be an effective approach for the management of neurodegenerative diseases [17] [18] [19] . Until now, the attempts to control microglial activation with general immune suppression (e.g., with steroids or immune suppressive drugs) have achieved limited clinical success [20] [21] [22] . In most conditions, microglial activation is secondary to other pathological processes that damage the CNS or retina. Thus, activated microglia may both exacerbate pathology and aid tissue repair. To develop effective microglia-targeted therapy, it is important that we understand the molecular pathways specific to the neurotoxic effects of activated microglia.

Microglial activation is tightly controlled at multiple levels. Microglia express various receptors (e.g., CX3CR1, SIRPα and CD200R) that can sense inhibitory molecules from surrounding neurons (e.g., CX3CL1, CD47 and CD200 from ganglion cells and amacrine cells of the retina) [23] [24] [25] . This neuron-to-microglia communication prevents overt microglial activation and protects neurons [23, 24] . For example, when the CX3CL1-CX3CR1 pathway is disrupted, microglial activation and retinal degeneration is increased in diabetes [26] or acute stress conditions [27] . Microglial activation is also regulated at intracellular levels. When the PRRs or cytokine receptors of microglia are activated, the intracellular signalling transduction pathways are regulated by various immune checkpoint proteins. The Janus kinase (JAK) and signal transducer and activator of transcription protein (STAT) signalling pathway of the cytokine receptors are negatively regulated by the suppressor of cytokine signaling (SOCS) proteins [28, 29] . Studies have shown that deletion of Socs3 in LysM + cells (LysMCre-Socs3 fl/fl ) did not affect microglia in normal conditions, yet enhanced the inflammatory response in the CNS in experimental autoimmune encephalomyelitis and increased demyelination [30] . The inflammatory response in experimental autoimmune uveoretinitis was also exaggerated in LysMCre-Socs3 fl/fl mice accompanied by severe retinal degeneration and angiogenesis [31] . The LysMCre-Socs3 fl/fl mice also developed more severe diabetic retinopathy compared with WT mice [32] . These results suggest that SOCS3 and CX3CR1 are both critically involved in regulating microglial activation in disease conditions. During aging, oxidative stress accumulates in the retina, particularly the macula. As a result, a low-grade parainflammation presented as mild microglial activation, subretinal migration and complement activation is initiated [33] . This parainflammatory response, when controlled properly, is beneficial and the retina may age healthily. When the parainflammatory response is dysregulated, it may turn into chronic inflammation leading to retinal degeneration [11] . We hypothesized that the combined deletion of two negative immune regulators in microglia, CX3CR1 in the cell surface and SOCS3 in the cytosol, may subject the cells to uncontrolled activation in response to age-mediated low-levels of oxidative insults. In this study, we generated a Cx3cr1 and Socs3 double knockout (LysMCre-Socs3 fl/fl Cx3cr1 gfp/gfp ) mouse line (referred to as DKO mice in this paper), in which Cx3cr1 was deleted globally and Socs3 was deleted in LysM + cells using the Cre/LoxP technology (Additional file 1B). In the retina, these two genes were depleted only in innate immune cells, predominately microglia. We found that the DKO mice presented with significant microglial dysfunction, and with ageing, microglia were activated accompanied by substantial neuronal degenerationa typical example of primary microglial disorder-mediated pathology i.e., "microgliopathy" [34] .

Animals and study design C57BL/6 J wild type, Socs3 fl/fl [31] , LysMCre-Socs3 fl/fl [31] , Cx3cr1 gfp/gfp [27, 35] and LysMCre-Socs3 fl/fl Cx3cr1 gfp/gfp DKO mice (all in C57BL/6 J background) were used in the study. The Socs3 fl/fl mice served as WT control. The LysMCre-Socs3 fl/fl mice were obtained by crossbreeding Socs3 fl/fl mice with LysMCre mice [31] . The Cx3cr1 gfp/gfp mice were provided by Steffen Jung (Weizmann Institute of Science, Rehovot, Israel). The LysMCre-Socs3 fl/fl mice and Cx3cr1 gfp/gfp mice were used to generate heterozygous LysMCre +/− Socs3 fl/− Cx3cr1 +/gfp mice. Subsequent crossings between offspring mice allowed the generation of LysMCre-Socs3 fl/fl Cx3cr1 gfp/gfp DKO (Abbreviated as DKO in the following context) mice (Additional file 1). The genotype of the DKO mice was confirmed by PCR (Additional file 1C). Mouse strains used for this study were listed in additional file 2. All animal procedures were conducted in accordance with the Animals (Scientific Procedures) Act of the UK Home Office (1986) and were approved by the Animal Welfare and Ethics Review Body of Queen's University Belfast. All mice were housed at the Biological Research Unit at Queen's University Belfast in a 12-h light / dark cycle with free access to water and food.

Fundus images were obtained. Animals were anesthetized with an intraperitoneal injection of 60 mg/kg ketamine hydrochloride (Fort George Animal Centre, Southampton, UK) and 5 mg/kg xylazine (Pharmacia & Veterinary Products, Kiel, Germany). The pupils were dilated with a drop of tropicamide and phenylephrine (Chauvin, Essex, UK). Images were captured with a Nikon D90 camera via an endoscope [36, 37] or the Micron IV system (Phoenix Research Labs, Pleasanton, CA, USA). Retinal green fluorescent images in Cx3cr1 gfp/gfp and DKO mice were acquired using the Micron IV system. The illumination settings and the gain were consistent in each subject. Images were saved in TIFF format.

Ganzfeld ERG was performed using a Diagosys Espion System (Diagnosys Technologies, MA, USA) using the protocol described in our previous publication [38] . In brief, mice were dark-adapted overnight and anaesthetised with 60 mg/kg ketamine hydrochloride and 5 mg/ kg xylazine and the pupils dilated and moisturised as described above. For each animal, 8 light intensities (from 0.008 to 25 cd.s/m 2 ) were applied. A-wave, b-wave amplitudes and oscillatory potential (OPs, summed amplitude of wavelets 2-5) were measured using the Espion analysis software.

The eyes were collected and fixed with 2% paraformaldehyde (PFA, Agar Scientific Ltd. Cambridge, UK) for 2 h at room temperature. The anterior segment (cornea, iris and ciliary body, and lens) of the eye was removed. Retinal tissue was carefully removed from the eyecup. Retinal tissues and RPE/choroid/sclera eye-cups were processed for flatmount staining using the protocol described in our previous paper [36, 37] . The antibodies used in the study were detailed in Table 1 . All samples were examined by Dmi8 fluorescence microscopy (Leica Microsystems CMS, Mannheim, Germany). Images were analysed using ImageJ (National Institutes of Health, Bethesda, MD, USA).

Eyes were fixed as described above and processed for 6 μm-thick paraffin sections. Eye sections were dewaxed using xylene, dehydrated by ethanol and rehydrated by H 2 O. Sections were then processed for either H&E staining or immunohistochemistry. Antigen retrieval was 

Fluorescent images were obtained to quantify the number of individual retinal neurons and neuronal structure. Three sections around the optic disc were used for retinal neuronal investigations, with 3-5 eyes used in each group. Four images were taken from the mid-central part of the retina in each eye section. The settings of fluorescence microscope remained constant between images and images were analyzed using FIJI software (National Institutes of Health, Bethesda, MD). The following retinal cells were quantified and analysed using the methods established in our groups [38, 39] : 1) Rod and cone photoreceptor (cone arrestin + ) cell count; 2) Cone photoreceptor segment length; 3) Cone-bipolar (PKCα + ) cell number; 4) Rod-bipolar (secretagogin + ) cell number; and 5) GABA + amacrine cell number. Cell numbers were averaged and normalised. The quantification was confirmed by two independent researchers.

Five fluorescent images (40X objective lens) were taken from each RPE/choroidal flatmount. The number of microglial cells and the morphology of RPE in aged DKO mice were quantified. 6 eyes per group were used in this study. FIJI ImageJ plug-in Bio-voxxel software [40] was used to outline the shape of each RPE cell. For each RPE cell, the area and perimeter were measured using FIJI and the shape factor was calculated by the eq. 4π * Area/Perimeter 2 [41] . The shape factor value ranges from 0 to 1, with one indicating a perfect circular cell, and zero indicating an elongated cell. The shape factor of a perfect hexagonal cell (eg. a RPE cell) is around 0.9 [42] . The relationship between microglial cell number and the shape factor of the RPE was analyzed by linear regression using SPSS statistics software (IBM, Chicago, IL, USA). The quantification was confirmed by two independent researchers.

Microglia were cultured from 4 to 6 week old Socs3 fl/fl and DKO mice using a previously described protocol [43] . Briefly, smashed brain tissue suspensions were filtered through a 70 μm cell strainer (BD Falcon, BD Biosciences). Then the cell suspensions were centrifuged at 600 g for 8 min. The cell pellet was re-suspended in media containing 10% fetal calf serum (FCS), 20 ng/ml M-CSF (Bio-techne, Minneapolis, Minnesota, US), 1% penicillin/streptomycin, 1 mM glutamine in DMEM/F12 (all from Thermo Fisher Scientific, Waltham, MA, USA), and seeded in a 6-well plate (Thermo Fisher Scientific). Floating cells were removed 3 days later and media were changed every 3 days until cells reached 90% confluence. The phenotype of the cells was confirmed by their CD11b and IBA-1 expression (> 90%). The microglia were then treated with 1) M1, or pro-inflammatory stimuli by adding 100 ng/ml IFN-γ (Bio-Techne) and 50 ng/ ml LPS (Sigma-Aldrich) or 2) M2, or anti-inflammatory stimuli by adding 20 ng/ml IL-4 (Bio-Techne) overnight. The cells were collected for real-time RT-PCR for cytokine/chemokine gene expression and the supernatants were collected for Luminex assay.

Microglia from Socs3 fl/fl or DKO mice were seeded into a black-walled 96-well plate at the density of 1 × 10 5 / well. The phagocytosis assay was performed using 1) the pHrodo™ Red E. coli BioParticles™ and 2) Escherichia coli (K-12 strain) Alexa Fluor 594 BioParticles™ (both from Thermo Fisher Scientific) following the manufacturer's instructions. Culture medium with Bioparticles alone was served as background control. The pHrodo Red E.coli BioParticles do not show any fluorescence at neutral pH, but are fluorescence brightly in acidic environments (e.g., engulfed into lysosome after phagocytosis). The fluorescence intensity was measured at 0. Table 2 . β-actin was used as housekeeping control.

The cytokine & chemokine magnetic bead panel kit (LXSAMSM-10, R&D, USA) was designed to detect CCL2, CCL5, CXCL2, CXCL10, Granulocyte-macrophage colony-stimulating factor (GM-CSF), IL-1β, IL-6, IL-10, IL-12p70 and TNF-α in the supernatant after different treatments. All procedures were carried out according to the manufacturer's instructions. The signals were detected and data analyzed by the Bio-plex 200 System (BIO-RAD, Richmond, CA). Chemokine concentrations were calculated using StarStation software (Applied Cytometry Systems, UK) with a four parameter curve-fitting algorithm applied for standard curve calculation.

Retinal explants and microglia co-culture

The method for retinal explants and microglia coculture was adapted from a previously published paper [44] . Briefly, microglial cells from Socs3 fl/fl and DKO mice were seeded onto polycarbonate filters (Fisher scientific, UK) and cultured in DMEM/F12 for 24 h to allow firm attachment to the filters. Cells were then given M0/ M1/ M2 treatments for overnight. Cells were washed with PBS three times to remove stimuli. Retinas from C57BL/6 J mice were placed with the photoreceptor side in contact with microglia for 24 h at 37°C. Retinal explants were fixed and stained for cone arrestin.

Statistical analysis was performed in GraphPad Prism (GraphPad Software, San Diego, CA, USA) and SPSS. Student's t tests were performed for analysis and comparisons between two groups. Two-way analysis of variance 

General phenotype of the DKO mice

The DKO mice bred normally and litter size was similar to C57BL/6 WT and Socs3 fl/fl control mice. They were not prone to infections in our Specific-Pathogen-Free (SPF) standard housing facility. Some mice started showing signs of slower and limited movements at 7~8 months old. 57% of the DKO mice died before 12 months old and the majority of death occurred between 7 and 9 months of age (Additional file 3). Post-mortem investigations (n = 27 mice) showed that~30% of the mice had tumors in the liver, spleen, and intestinal system. No visible abnormality was observed in the brain, the lung or the heart. The majority of the DKO mice died before 15 months of age. Our results suggest that the DKO mice may have accelerated aging.

As the two immune regulatory genes, Cx3cr1 and Socs3, were depleted exclusively in microglia in the retina, we examined microglial cells under normal aging conditions. For this purpose, we compared the DKO mice with each single KO mice (Cx3cr1 gfp/gfp and LysMCre-Socs3 fl/fl ) and used the Socs3 fl/fl mice as controls. Due to the short lifespan of the DKO mice, "aged" mice were 10-12 months old and "young" mice were 3-5 months old in this study. Fundus images did not show any abnormality in 3 and 12 month-old Socs3 fl/fl mice, LysMCre-Socs3 fl/fl mice, and Cx3cr1 gfp/gfp mice, as well as 3 month-old DKO mice (Figs. 1a) . Patches of whitish lesions and multiple small whitish dots were observed in 75% of 12 month-old Arrowa patch of whitish lesion; arrowheadsmultiple whitish dots. b Fundus green fluorescence images were taken from 3-month or 12-month old Cx3cr1 gfp/gfp and DKO mice using Micron IV. Arrowspatches of GFP aggregations; arrowheadsperivascular macrophages. ODoptic disc DKO mice (arrow and arrowheads, Fig. 1a) . These lesions were not observed in age-matched WT (Socs3 fl/fl ) and single KO mice (i.e., LysMCre-Socs3 fl/fl , and Cx3cr1 gfp/gfp ).

Retinal microglia in the Cx3cr1 gfp/gfp and DKO mice were imaged in vivo due to their GFP expression. Dense GFP signals were observed in the optic disc (OD) in Cx3cr1 gfp/gfp and DKO mice (Fig. 1b) . GFP + microglia were evenly distributed throughout the fundus. Perivascular macrophages were also visible in some areas (Fig. 1b, arrowhead) . No obvious difference was observed between 3 month old Cx3cr1 gfp/gfp and DKO mice. The density of GFP + cells appeared to be increased in 12 month old DKO mice and patches of GFP aggregations were observed in these mice (Fig. 1b, arrows) .

Microglia were further quantified in retinal flatmounts in all groups. Microglia in Socs3 fl/fl and LysMCre-Socs3 fl/fl mice (which do not express GFP) were identified by IBA-1 staining. In 3 month old mice, IBA-1 + cells resided in the GCL and IPL (Fig. 2a, d) , and OPL (Fig. 2b, e) . The numbers of microglia in the GCL and IPL were comparable in all four groups; whereas DKO mice had higher number of microglia in the OPL than those in other strains (Fig. 2e) . In the aged mice, in addition to GCL, IPL and OPL, microglia were detectable in the outer retinal layers, including ONL + PRL, particularly in DKO mice (Fig. 2c, f) . Aged DKO mice had a significantly higher number of microglia than other age-matched strains in all layers studied (Fig. 2g) . The number of microglia in RPE flatmounts of aged DKO mice were significantly higher than those in other age-matched strains (Additional file 4A, B). Microglia in photoreceptor layer and RPE flatmounts of aged DKO mice also expressed Isolectin B4, indicative of microglial activation (Additional file 4C). Our results 100 μm. Two-way ANOVA, Tukey's test, ** P < 0.01, *** P < 0.001, **** P < 0.0001 (**** showed the difference between aged DKO and other groups in f and g). GCLganglion cell layer; IPLinner plexiform layer; OPLouter plexiform layer; ONLouter nuclear layer; PRLphotoreceptor inner/outer segment layer suggest that age-related microglial activation was significantly enhanced in the DKO mice compared with Socs3 fl/fl and single KO mice.

To understand if uncontrolled microglial activation led to retinal pathology in aged DKO mice, we conducted ERG in these mice. Our result show that a-wave, b-wave and the OPs were significantly reduced in aged DKO mice compared with those in young DKO mice (Additional file 5), suggesting a global reduction in retinal neuronal function. Therefore, we conducted further immunohistological investigations on retinal neurons.

In young mice, with the exception of a slightly decreased number of cone photoreceptors in DKO mice, the numbers of cells in the entire ONL, cone arrestin + photoreceptors, and rod photoreceptors were comparable in different strains of mice (Fig. 3a-e) . In aged mice, the numbers of cells in the ONL, cone arrestin + photoreceptors and rod photoreceptors were significantly decreased in DKO mice compared with single KO and Socs3 fl/fl mice (Fig. 3a-e) , accompanied by reduction of the length of cone photoreceptor segments ( Fig. 3f-g) . The number of secretagogin + rod-bipolar cells was lower in young DKO mice than age-matched single KO or Socs3 fl/fl controls and was further reduced in aged DKO mice (Fig. 4a-b) . The number of PKCα + conebipolar cells was comparable in young mice (Fig. 4c-d) . An age-related reduction was observed in Cx3cr1 gfp/gfp and DKO mice although the reduction was more significant in aged DKO mice (Fig. 4c-d) . significantly lower in young DKO mice (Fig. 4e-f ). The age-related reduction was observed in all strains of mice although the reduction was more profound in DKO mice ( Fig. 4e-f ).

Histological investigation did not show any structural abnormalities in eyes from young mice (Fig. 5a) . Localized RPE swelling (Fig. 5a , arrows) and vacuoles (Fig. 5a , asterisks) with thinner ONL in corresponding area were frequently observed in aged DKO mice. RPE damage in aged DKO mice was further confirmed in flatmount investigations and the lesions were always accompanied by microglial/macrophage infiltration (Fig. 5b) . Aged DKO mice had enlarged RPE cell size, increased perimeter and reduced cell density. Areas of total RPE damage with massive microglial infiltration were frequently observed in aged DKO mice (Fig. 5b) . The number of infiltrating microglia/macrophages negatively correlated with RPE shape factor in aged DKO (Fig. 5c) .

In line with retinal neuronal degeneration, we also observed increased GFAP expression in 10~12 month old DKO mice (Additional file 6), indicative of Müller gliosis.

Our results suggest that aged DKO mice suffered from significant neuroretinal degeneration and RPE damage.

To understand how Cx3cr1 and Socs3 depletion may affect microglial function, microglial cultures from the DKO and Socs3 fl/fl mice were subjected to phagocytosis assay using the Alexa Fluor™ 594 conjugate E. coli BioParticles™ and the pH-sensitive pHrodo™ Red E. coli BioParticles™ systems. In both assays, microglia from DKO mice had significantly higher levels of fluorescence intensities compared to cells from Socs3 fl/fl control mice (Fig. 6A-C) . Fluorescence microscopy further showed intense intracellular particles in DKO cells treated with Alexa Fluor™ 594 E. coli bioparticles (although uningested extracellular bioparticles were also visible) (Fig. 6B-2) . In cells treated with pH-sensitive bioparticles, fluorescent signals were inside the cells and DKO cells demonstrated brighter fluorescent signals than Socs3 fl/fl cells (Fig. 6C-2) . Our results suggest that the phagocytic activity of DKO microglia was significantly higher than that of Socs3 fl/fl cells. Interestingly, we also detected cone arrestin antigens inside GFP + cells in the outer layers of retina from aged DKO mice but not in other strains (Fig. 6D) , an indication of in situ uptake of damaged cone photoreceptors by microglia.

Real-time RT-PCR showed that microglia from DKO mice expressed significantly higher levels of C1qb, CD16 

. Scale bar: 25 μm. Mean ± SD, N = 3~5 mice, * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001; Two-way ANOVA followed by Tukey's test. GCLganglion cell layer; IPLinner plexiform layer; INLinner nuclear layer; OPLouter plexiform layer; ONLouter nuclear layer (Fcgr3), Trem2, CD36, and CD64 (Fcgr1) compared to microglia from Socs3 fl/fl mice (Fig. 7) . The expression of CD32 (Fcgr2b), CD47, and Anxa1 did not differ between DKO and SOCS3 fl/fl microglia (Fig. 7) . The higher phagocytic activity of DKO microglia may be related to higher levels of C1qb, Trem2, CD36, and Fc-gamma receptor expression by these cells.

Under normal culture conditions, naïve microglia (M0) from DKO mice expressed significantly higher levels of TNF-α and iNOS mRNA (Fig. 8a) and produced significantly higher amounts of TNF-α, IL-6, CCL2, CCL5, CXCL2 and CXCL10 proteins compared to their counterparts from Socs3 fl/fl mice (Fig. 8b) .

Following LPS + IFN-γ stimulation (M1), increased mRNA expressions of pro-inflammatory mediators such as TNF-α, iNOS, IL-1β, IL-6, IL-12p40, IL-23p19, CCL2, CCL5 and CXCL2 were observed in microglia from both Socs3 fl/fl and DKO mice compared to naïve M0 cells although the upregulations were generally more significant in DKO M1 cells compared to Socs3 fl/fl M1 cells (Fig. 8a) . M1 microglia from DKO mice produced and released significantly higher levels of TNF-α, IL-6, CCL2, CCL5, CXCL2, CXCL10 and IL-10 compared with M1 microglia from WT mice (Fig. 8b) . Interestingly, although the mRNA of IL-1β was significantly increased in M1 c Correlation between RPE shape factor and the number of infiltrating GFP cells. Statistics were applied by five images per mouse from six aged DKO mice microglia, this cytokine was below detectable levels in the supernatants (data not shown).

IL-4 treated microglia (M2) from DKO mice expressed significantly higher levels of IL10, CD206 and Arg1 mRNA compared to their counterpart from Socs3 fl/fl mice (Fig. 8a) . Interestingly, the productions of TNF-α, IL-6, CCL2, CCL5, CXCL2 and CXCL10 were also significantly higher in IL-4-treated DKO microglia than those in IL-4-treated Socs3 fl/fl microglia (Fig. 8b) .

Together, our data suggested that microglia with Socs3 and Cx3cr1 deficiency produce significantly higher basal levels of cytokines and chemokines, and they are also more sensitive to cytokine stimuli (such as LPS + IFNγ or IL-4).

To further understand if altered microglial activation contributes to retinal degeneration in aged DKO mice, retinal explants from C57BL/6 J mice were cultured alone or co-cultured with microglia from Socs3 fl/fl and DKO mice without and with M1 or M2 polarization ( Fig. 9a) , and the number of cone arrestin + cells was evaluated in retinal flatmounts. The addition of naïve (M0) microglia from Socs3 fl/fl or DKO mice did not affect the number of cone arrestin + cells compared to retinal explants alone (109.14 and 100.42% respectively, Fig. 9b, c) . M1-type microglia from both Socs3 fl/fl and DKO significantly reduced cone photoreceptor survival (73.67 and 64.87% respectively, Fig. 9b, c) . Interestingly, M2-type microglia from DKO but not Socs3 fl/fl mice also significantly reduced cone photoreceptor survival (57.48%, Fig. 9b, c) . Our results suggest that active microglia from DKO mice are toxic to photoreceptors regardless of how they are activated; whereas, microglia from Socs3 fl/fl mice are only toxic to photoreceptors when they are activated to a pro-inflammatory phenotype.

In this study, we showed that deletion of two immune negative regulators, Cx3cr1 and Socs3 (in LysM + cells) resulted in uncontrolled retinal microglial activation during aging, accompanied by a significant degeneration [45] . Deletion of Cx3cr1 and Socs3 in the DKO mice would therefore predominately affect retinal microglia. The degeneration of retinal neurons in the aged DKO mice was likely the consequence of primary microglial dysfunction. Therefore, retinal pathologies observed in this study fall into the disease category of "microgliopathay" [46] .

The DKO mice were not prone to any infection when housed in a standard SPF facility although the lifespan of the DKO mice was significantly shortened compared to WT or single KO mice. The DKO mice appeared to have slower and limited movements and higher incident of tumor development (evidence of accumulation of mutations). This suggests that the DKO mice might have accelerated aging. Therefore, this mouse model may be valuable for understanding the contribution of low-grade chronic inflammation (inflammaging) [47] to the aging process.

Retinal aging is accompanied by a low-grade of parainflammation characterized by mild microglial activation, subretinal migration and complement activation [33, 48] , and a dysregulated parainflammatory response contributes critically to age-related retinal degenerative diseases such as AMD [11] . Previous studies have shown that the mice with Cx3cr1 deficiency developed retinal degeneration by 1-2 years of age [49, 50] , and the pathology is known to be related to increased subretinal infiltration of CCR2 + phagocytes and the impairment in their removal [50] . In our study, significant microglial activation and retinal degeneration were only observed in DKO mice by the age of 10-12 months. Our results suggest that microglial activation is collectively regulated by neuronderived signals (e.g., CX3CL1) and intracellular immune inhibitors (e.g., SOCS3). The discrepancy in retinal pathology in Cx3cr1 KO mice between our study and that of Combadiere et al. may be related to different housing conditions.

We observed a number of functional alterations in microglia from the DKO mice. These include higher cytotoxicity to photoreceptors in retinal explants, significantly higher levels of phagocytic activities and more secretory proteins (TNF-α, IL-6, CCL2 CCL5, CXCL2, CXCL10 and IL-10). The secretory profile was similar to the senescence-associated secretory phenotype (SASP) of macrophages [51] with the exception of IL1β, which was not detected in microglial supernatants in our study. The JAK/STAT pathway is known to be involved in cell senescence and the phenotype of SASP [52] and JAK inhibition alleviated SASP phenotype and frailty in old age [52] . Previously, we have shown that the LysMCre-Socs3 fl/fl mice had higher levels of pSTAT3 in the retina [53] . Furthermore, a recent study has shown that CX3CR1-deficient microglia exhibit a premature aging transcriptome [54] . It is possible that microglia from the DKO mice may also undergo premature aging although this warrants further investigation. Fig. 7 The mRNA expression of phagocytosis related genes in microglia from Socs3 fl/fl and DKO mice. The mRNA was extracted from brain microglia cultured from Socs3 fl/fl and DKO mice. The expression of C1qb, CD16a (Fcgr3), Trem2, CD36, and CD64 (Fcgr1), CD32 (Fcgr2b), CD47, and Anxa1was evaluated by qRT-PCR. Mean ± SD, N = 3. ** P < 0.01; **** P < 0.001, unpaired Student t test SASP can have significant pathological effects on the aging retina. In our study, microglia from DKO mice produced significantly higher levels of cytokines and chemokines compared to their counterparts regardless of their activation states (Fig. 8a) ; however, only M1 and M2 (but not naïve M0) microglia were toxic to photoreceptors in retinal explants (Fig. 9b-c) . Compared to naïve (M0) microglia, M2 microglia from DKO mice produced higher levels of IL-6, CCL2 and CXCL2, suggesting that the three mediators might contribute to M2 microgliamediated photoreceptor death.

The high phagocytic activity of DKO microglia may be due to the increased expression of relevant receptors such as C1q, Trem2, FcγR and CD36 (Fig. 7) . The underlying mechanism related to the upregulation of these surface molecules remains unknown but may be related to uncontrolled activation of microglia in DKO mice. For example, C1q expression in microglia is known to be upregulated during brain injury [55] and infection [56] . Increased TREM2 expression was reported in amyloid plaque-associated microglia [57] . Altered expression of phagocytosis-related molecules and enhanced phagocytic activity appear to be an important part of primary microglial dysfunction in our model system.

A recent study has shown that microglial function is distinct in different anatomical locations during retinal homeostasis and degeneration [58] . We have shown that microglia in different retinal layers are regulated by diverse neurons and molecules [25] . Specifically, ganglion cell and amacrine cells work together to control microglial activation in the GCL/IPL layer; whereas microglia in the OPL is regulated by horizontal cells and photoreceptors assisted by bipolar cells [25] . In the present Fig. 8 The expression and production of cytokines and chemokines by microglia from Socs3 fl/fl and DKO mice. Brain microglia from Socs3 fl/fl and DKO mice were untreated (M0, naive microglia) or treated with LPS + IFN-γ (M1) or IL-4 (M2). The mRNA expression of TNF-α, IL-1β, IL-6, iNOS, IL-12p40, IL-23p19, CCL2, CCL5, CXCL2, IL-10, CD206 and Arg1 were examined by real-time RT-PCR (a). The protein levels of TNF-α, IL-6, CCL2, CCL5, CXCL2, CXCL10 and IL-10 in the supernatants were measured using the cytokine Luminex multiple assay (b). Mean ± SD, N = 3. * P < 0.05; ** P < 0.01; *** P < 0.001. Two-way ANOVA followed by Tukey's multiple comparisons. Experiments were repeated three times model of retinal microgliopathy, microglia in all retinal layers were activated and a wide range of retinal neurons and RPE cells were affected. The initial trigger of retinal microglial activation in the aging DKO mice would likely be oxidative stress. Previously, we detected ox-LDL and 2,4-Dinitrophenol (DNP) in all retinal layers of the aged mice [33] suggesting ubiquitous oxidative stress in the aging retina. Our results also suggest that that Cx3cr1 and Socs3 are involved in the parainflammatory response of microglia in all retinal layers.

The crosstalk between microglia and retinal neurons is important for maintaining retinal homeostasis. Active microglia release excessive amounts of proinflammatory cytokines and chemokines that kill retinal neurons. Dead neurons in turn, further stimulate microglial activation. The reduction in retinal neurons also weakens the inhibitory signals to microglia further escalating microglial activation. Apart from expressing the SASP, microglia in the DKO mice had enhanced phagocytic activities and the hyperactive engulfment (named as phagoptosis by [59] ) can also lead to unnecessary neuron loss [60] . Recent evidence indicates that phagoptosis contributes to neuronal loss in various CNS diseases [61] [62] [63] .

The damage to RPE cells in our model may be a consequence of excessive retinal neuronal degeneration. We observed a strong correlation between the number of Fig. 9 The effect of microglia from Socs3 fl/fl and DKO mice on photoreceptor survival in retinal explant cultures. Brain microglia were cultured from Socs3 fl/fl and DKO mice. Retinal explants were from Socs3 fl/fl mice. Retinal explants were co-cultured with different types of microglia for 24 h. The flatmounts of retinal explants were then stained for cone arrestin and imaged by Dmi8 fluorescence microscopy. a Schematic diagram showing study design. b Representative images showing cone arrestin + cells in retinal explants in different treatment groups. c Quantitative data of cone arrestin + cells in different groups. The number of cone arrestin + cells in different treatment groups was compared to retinal explants without microglia co-culture. Mean ± SD. N = 3~6 explants. * P < 0.05, ** P < 0.01, *** P < 0.001. Two-way ANOVA followed by Tukey's multiple comparisons. Scale bar: 50 μm subretinal microglia/macrophage and RPE dysmorphology ( Fig. 6B-C) . Damaged neurons are removed from the retina by subretinal phagocytes (including microglia), likely through the subretinal space -RPEchoroid route (a CD47-dependent pathway) [64] . Impaired elimination ability of these subretinal phagocytes is known to contribute to RPE damage and the development of agerelated macular degeneration [65] . Although we do not know if the mechanisms involved in subretinal phagocyte elimination is altered in the DKO mice, the continued retinal neuronal damage during aging may lead to accumulation of active phagocytes into the subretinal space, which may cause RPE damage [65] .

Age-induced retinal microglial activation in the LysMCre-Socs3 fl/fl Cx3cr1 gfp/gfp DKO mice is dysregulated. This uncontrolled low-grade microglial activation resulted in retinal microgliopathy characterized as extensive neuronal degeneration and RPE damage. In this model of microgliopathy, active microglia may damage neurons through SASP expression or phagoptosis although the underlying mechanisms remain to be investigated. This model offers a unique opportunity to study the mechanisms of microglial dysfunction-mediated retinal pathology and will also help to elucidate the detrimental role of inflammaging in age-related retinal degeneration.

Microglia are required for protection against lethal coronavirus encephalitis in mice

Microglia Are Indispensable for Synaptic Plasticity in the Spinal Dorsal Horn and Chronic Pain

Microglia in the aging retina

The microglia in healthy and diseased retina

Roles of microglia in brain development, tissue maintenance and repair

Central nervous system regeneration is driven by microglia necroptosis and repopulation

Microglia turnover with aging and in an Alzheimer's model via long-term in vivo single-cell imaging

Disease-related microglia heterogeneity in the hippocampus of Alzheimer's disease, dementia with Lewy bodies, and hippocampal sclerosis of aging

Regional microglia are transcriptionally distinct but similarly exacerbate neurodegeneration in a culture model of Parkinson's disease

Retinal microglia: just bystander or target for therapy?

Parainflammation, chronic inflammation, and age-related macular degeneration

Microglia in retinal degeneration

Microglial activation in human diabetic retinopathy

Microglia/macrophages migrate through retinal epithelium barrier by a transcellular route in diabetic retinopathy: role of PKCζ in the Goto Kakizaki rat model

Activated microglia in human retinitis pigmentosa, late-onset retinal degeneration, and age-related macular degeneration

Neurodegeneration severity can be predicted from early microglia alterations monitored in vivo in a mouse model of chronic glaucoma

Modulation of three key innate immune pathways for the most common retinal degenerative diseases

Kir6.1/K-ATP channel modulates microglia phenotypes: Implication in Parkinson's disease

Modulators of microglial activation and polarization after intracerebral haemorrhage

A randomized pilot study of systemic immunosuppression in the treatment of age-related macular degeneration with choroidal neovascularization

Clinical aspects of inflammation in Alzheimer's disease

Systemic immunosuppression and risk of age-related macular degeneration

Microglia activation in retinal degeneration

Immune regulation in the aging retina

Single-cell RNA sequencing study of retinal immune regulators identified CD47 and CD59a expression in photoreceptors-implications in subretinal immune regulation

CX3CR1 deficiency accelerates the development of retinopathy in a rodent model of type 1 diabetes

Paraquat-induced retinal degeneration is exaggerated in CX3CR1-deficient mice and is associated with increased retinal inflammation

SOCS proteins, cytokine signalling and immune regulation

SOCS3, a major regulator of infection and inflammation

Signal transducer and activator of transcription-axis in myeloid cells regulates neuroin fl ammation

Cytokine signaling protein 3 deficiency in myeloid cells promotes retinal degeneration and angiogenesis through Arginase-1 upregulation in experimental autoimmune Uveoretinitis

STAT3 activation in circulating myeloid-derived cells contributes to retinal microvascular dysfunction in diabetes

Para-inflammation in the aging retina

Nasu-Hakola disease and primary microglial dysfunction

Analysis of Fractalkine receptor CX3CR1 function by targeted deletion and Green fluorescent protein reporter gene insertion

Identification of novel dendritic cell populations in normal mouse retina

Age-dependent accumulation of lipofuscin in perivascular and subretinal microglia in experimental mice

IL-33 deficiency causes persistent inflammation and severe neurodegeneration in retinal detachment

Intravitreal injection of Normal saline induces retinal degeneration in the C57BL/6J mouse

Spatial coordination between cell and nuclear shape within micropatterned endothelial cells

Cultivation of human microvascular endothelial cells on topographical substrates to mimic the human corneal endothelium

Glucose transporter 1 critically controls microglial activation through facilitating glycolysis

Subretinal mononuclear phagocytes induce cone segment loss via IL-1β

Utility of LysM-cre and Cdh5-cre driver mice in retinal and brain research: an imaging study using tdTomato reporter mouse

Microglia and brain macrophages in the molecular age: from origin to neuropsychiatric disease

The integration of inflammaging in age-related diseases

Immune activation in retinal aging: a gene expression study

CX3CR1-dependent subretinal microglia cell accumulation is associated with cardinal features of age-related macular degeneration

CCR2 + monocytes infiltrate atrophic lesions in age-related macular disease and mediate photoreceptor degeneration in experimental subretinal inflammation in Cx3cr1 deficient mice

Cellular senescence: molecular mechanisms and pathogenicity

JAK inhibition alleviates the cellular senescence-associated secretory phenotype and frailty in old age

STAT3 activation in circulating monocytes contributes to Neovascular age-related macular degeneration

Cx3cr1-deficient microglia exhibit a premature aging transcriptome

Complement C1q is dramatically up-regulated in brain microglia in response to transient global cerebral ischemia

Increase of C1q biosynthesis in brain microglia and macrophages during lentivirus infection in the rhesus macaque is sensitive to antiretroviral treatment with 6-chloro-2′,3′-dideoxyguanosine

TREM2 is upregulated in amyloid plaque-associated microglia in aged APP23 transgenic mice

Microglial Function Is Distinct in Different Anatomical Locations during Retinal Homeostasis and Degeneration

Eaten alive! Cell death by primary phagocytosis: "Phagoptosis

Microglial phagocytosis of live neurons

Microglial phagocytosis of living photoreceptors contributes to inherited retinal degeneration

Activated microglia cause reversible apoptosis of pheochromocytoma cells, inducing their cell death by phagocytosis

Living Neurons with Tau Filaments Aberrantly Expose Phosphatidylserine and Are Phagocytosed by Microglia

Complement factor H inhibits CD47-mediated resolution of inflammation

On phagocytes and macular degeneration

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations

The authors thank the staff in Biological Research Unit at Queen's University Belfast for maintaining the animals, John Conlon for helping with the Luminex Assay and Daniel Xu for helping with English expression.

All procedures concerning animals were performed with permission of the local authority.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The online version contains supplementary material available at https://doi. org/10.1186/s13024-021-00432-9.Additional file 1. Genotype of the DKO mice. Genotyping PCRs were performed using primer combinations distinguishing the WT and knockout alleles for LysMCre, Socs3 and Cx3cr1 respectively. (A) Primers used for mouse genotyping. (B) Representative PCR images. For Socs3 fl/fl loci, DKO mice showed the band at 700 bp while wild type (WT) sample showed band at 500 bp and the heterozygous showed both bands. For the LysMCre loci, both DKO and heterozygous samples showed band at 700 bp while the WT at 350 bp. DKO mice showed a band of 1200 bp for Cx3cr1 allele, while WT showed band at 970 bp and the heterozygous showed both bands.Additional file 2. Mouse strain used for each experiment. 

All data generated or analysed during this study are included in this published article (and its additional files).Ethics approval and consent to participate Not applicable.

Not applicable.

The authors declare that they have no competing interests.Received: 5 August 2020 Accepted: 8 February 2021