Paraquat-Induced Retinal Degeneration Is Exaggerated in CX3CR1-Deficient Mice and Is Associated with Increased Retinal Inflammation

Abstract

Purpose.: To investigate the role of the Fractalkine receptor CX3CR1 pathway in oxidative insults–mediated retinal degeneration and immune activation.

Methods.: A prooxidant, paraquat (0.75 μM) was injected into the vitreous of C57BL/6J, CX3CR1^gpf/+, and CX3CR1^gfp/gfp mice. Retinal lesions were investigated clinically by topic endoscopic fundus imaging and fluorescence angiography, and pathologically by light- and electron microscopy. Retinal immune gene expression was determined by real-time RT-PCR. Microglial activation and immune cell infiltration were examined by confocal microscopy of retinal flatmounts.

Results.: Intravitreal injection of paraquat (0.75 μM) resulted in acute retinal capillary nonperfusion within 2 days, which improved from 4 days to 4 weeks postinjection (p.i.). Panretinal degeneration was observed at 4 days p.i. and progressed further at 4 weeks p.i. In the absence of CX3CR1, retinal degeneration was exaggerated and was accompanied by increased TNF-α, iNOS, IL-1β, Ccl2, and Casp-1 gene expression. Confocal microscopy of retinal flatmounts revealed microglial activation and CD44⁺MHC-II⁺ monocyte and GR1⁺ neutrophil infiltration in paraquat-injected eyes. The number of activated microglia and infiltrating leukocytes was significantly higher in CX3CR1^gfp/gfp mice than in CX3CR1^gfp/+ mice.

Conclusions.: Our results suggest that the CX3CR1 signaling pathway may play an important role in controlling retinal inflammation under oxidative and ischemia/reperfusion conditions. In the absence of CX3CR1, uncontrolled retinal inflammation results in exaggerated retinal degeneration.

Introduction

Immune response in neuronal tissue is a “double-edged sword” representing a fine balance between protective antipathogen responses and detrimental neurocytotoxic effects. Neuronal inflammation is, therefore, tightly regulated at multiple levels, including neuron-to-microglia, microglia-to-astrocyte, and astrocyte-to-neuron interactions to ensure beneficial effects of the response. Microglia are the main innate immune cells safeguarding the neuronal tissue. Like other innate immune cells, microglia express pattern-recognition receptors (PRRs), including toll-like receptors (TLRs) ^1,2 and NOD-like receptors (NLRs), ³ which can bind various pathogen-associated molecular patterns (PAMPs), leading to microglial activation. In addition to PRRs, various other molecules are also involved in microglial activation. Fractalkine (CX3CL1) is constitutively produced by neurons in the CNS, ^4,5 including the neural retina, ^6–8 whereas its receptor CX3CR1 is exclusively expressed by microglial cells. ^5,8,9 CX3CL1–CX3CR1 signaling is one of the main pathways involved in neuron–microglia interactions and plays a crucial role in regulating microglial activation. ^5,9,10 Under physiological conditions, the CX3CL1–CX3CR1 pathway is known to suppress microglial activation. ⁹ However, under pathologic conditions, the role of the CX3CL1–CX3CR1 pathway in neuronal inflammation is controversial. 

Early work by Cardona et al. ⁹ has shown that CX3CR1 is important in controlling microglial neurotoxicity. In the absence of CX3CR1, microglial responses to stimuli (lipopolysaccharide and damaged neuronal tissues) are dysregulated, leading to neurotoxicity. ⁹ The neuroprotective role of the CX3CL1–CX3CR1 pathway has also been observed in other neurodegenerative and neuroinflammatory diseases. ^11,12 In addition, in vitro studies have shown that both soluble and membrane-bound fractalkines attenuate lipopolysaccharide–induced microglial activation, and fractalkine suppresses microglial activation through the PI3K pathway. ¹³ In contrast to these observations, in the models of CNS ischemia/reperfusion, lesion development was protected in both CX3CR1-deficient mice ¹⁴ and CX3CL1-deficient mice, ¹⁵ and the protection was related to reduced IL-1β and TNF-α expression and decreased leukocyte infiltration in those mice. ¹⁴ Furthermore, CX3CL1–CX3CR1 deletion promotes recovery after spinal cord injury by limiting the recruitment and activation of Ly6C^lo/iNOS⁺ macrophages. ¹⁶  

In the retina, CX3CR1 is not required for the distribution and recruitment of microglia under normal physiological conditions. ⁸ In a radiation-induced retinal parainflammation model, although Fractalkine (CX3CL1) expression is temporarily upregulated, the CX3CR1 pathway is not involved in the recruitment of bone marrow–derived myeloid cells. ¹⁷ Under inflammatory conditions (e.g., uveitis), CX3CL1 expression is increased. ⁶ However, CX3CR1 deficiency does not appear to affect the severity of retinal inflammation in a model of experimental autoimmune uveoretinitis, ¹⁸ although a mild increase in disease severity was observed in another study. ¹⁹ During aging, CX3CR1 deficiency results in increased microglial activation and subretinal migration and retinal degeneration. ²⁰ The role of the CX3CL1–CX3CR1 pathway in retinal immunity remains to be fully elucidated. 

In this study, we investigated the role of CX3CR1 in oxidative damage–mediated retinal degeneration in the paraquat injection model previously reported by Cingolani and colleagues. ²¹ We show that paraquat-mediated retinal degeneration is accompanied by an acute retinal inflammation characterized by increased inflammatory gene expression (e.g., TNF-α and iNOS), microglial activation, and the recruitment of circulating neutrophils and monocytes. In the absence of CX3CR1, the inflammatory response was exaggerated and was related to severe retinal degeneration. 

Materials and Methods

Animals

Adult (8–12 weeks) C57BL/6J, CX3CR1^gfp/+, CX3CR1^gfp/gfp (C57BL/6J background) ²² mice were bred and maintained in standard animal housing rooms and exposed to 12-hour light/12-hour dark cycle. All procedures were conducted under the regulations of the United Kingdom Home Office Animals (Scientific Procedures) Act 1986, and were in compliance with the Association for Research in Vision and Ophthalmology (ARVO) statement for the use of animals in ophthalmic and vision research. 

DNA Sequencing

DNA sequencing was used to confirm that the rd8 allele mutation in the Crb1 gene was not present in the wild-type (WT) C57BL/6J and CX3CR1^gfp/gfp mice. Ear punch samples were collected, genomic DNA was extracted, and the region of the Crb1 gene containing the rd8 mutation was amplified by PCR using the following primers: Forward: GCACAATAGAGATTGGAGGC; Reverse: TGTCTACATCCACCTCACAG. The PCR products were purified using an extraction kit (GeneJet, Fermentas; Thermo Fisher Scientific UK Ltd., Loughborough, UK). The sequencing reaction was carried out using a sequencing kit (BigDye; Applied Biosystems/Life Technologies Ltd., Paisley, UK) in the Core Technology Unit at Queen's University Belfast. DNA samples from the rd8^−/− mice (courtesy of Ulrich Luhmann, Institute of Ophthalmology, University College of London) were used as positive controls. Our results confirmed that neither our WT nor CX3CR1^gfp/gfp mice carry the Crb1 rd8 mutation. 

Intravitreal Injection

Intravitreal injections were conducted under a surgical microscope. Mice were anesthetized with an intraperitoneal injection of ketamine hydrochloride (60 mg/kg; Fort George Animal Centre, Southampton, UK) and xylazine (5 mg/kg; Pharmacia & Veterinary Products, Kiel, Germany). Pupils were dilated using 1% tropicamide and 2.5% phenylephrine (both from Chauvin, Essex, UK). A puncture was made at the pars plana using a 30-gauge needle. A blunt-tip 33-gauge needle (Hamilton Bonaduz AG, Bonaduz, Switzerland) was inserted through the puncture hole with a 45° injection angle into the vitreous. One microliter paraquat (0.75 μM; Sigma-Aldrich, Poole, UK) was injected in each eye ²¹ using a repeating dispenser (PB-600‐1; Hamilton Bonaduz). PBS was used as vehicle controls. 

Clinical Investigations

On different days (24 hours, 4 days, and 4 weeks) after intravitreal injection, mice were anesthetized and pupils were dilated for clinical investigations, including fundus imaging and fluorescein angiography. Fundus images were taken using the topic endoscopic fundus imaging (TEFI) system as described previously. ²³ Fluorescein angiography was conducted using a home-built scanning laser ophthalmoscope using the technique detailed in our previous publications. ^24,25 In brief, 100 μL of 0.1% sodium fluorescein (Sigma-Aldrich) was injected through the tail vein. A 488-nm excitation and 520-nm filter barrier were used to collect fluorescent images. 

Retinal Flatmount Preparation

Mouse eyes were collected and fixed in 2% paraformaldehyde (Agar Scientific Ltd., Cambridge, UK) for 2 hours, then replaced with PBS and stored at 4°C. Retinal tissues were dissected carefully under a dissecting microscope. ^26,27 Retinal tissues were permeabilized with 0.3% Triton X-100 for 2 hours, and then incubated with primary antibodies at 4°C overnight. The primary antibodies used in the study were rat anti-mouse CD44 (IM7, a marker for activated leukocytes), rat anti-mouse GR1 (RB6‐8C5, a marker for both monocytes and granulocytes), biotin anti-mouse I-A/I-E (M5/114, for mouse Major Histocompatibility Complex class II [MHC-II]), biotin anti-mouse CD49 (DX5, a mouse NK cell marker) (1:50; all from BD Biosciences, Oxford, UK), or rabbit anti-mouse collagen IV (1:50; AbD Serotec Ltd., Oxford, UK). After three washes, samples were incubated with goat anti-rat R-PE (AbD Serotec), streptavidin APC (BD Biosciences), or donkey anti-rabbit Alexa Fluor 594 (1:200; Invitrogen Molecular Probes, Paisley, UK) at room temperature in the dark for 2 hours. Retinal tissues were flatmounted on glass slides using mounting medium (Vectashield; Vector Laboratory Ltd., Peterborough, UK). All samples were examined using a laser scanning microscope (LSM510 META microscope; Carl Zeiss, Jena, Germany). 

Real-Time RT-PCR

Two days after paraquat injection, mouse eyes were collected and retinal tissues were dissected and stored at −80°C for total RNA extraction. 

Total RNA was extracted using a commercial kit (RNeasy Mini Kit; Qiagen Ltd., Crawley, UK) following manufacturer's instructions. cDNA was synthesized from the same amount of total RNA using reserve transcriptase and random primers (Superscript II; Life Technologies). Real-time RT-PCR was performed from the same amount of cDNA with a real-time PCR platform (LightCycler 480 System, using SYBR Green I master; Roche Diagnostics GmbH, Mannheim, Germany). GAPDH was used as a housekeeping gene. The relative expression levels of mRNA were calculated using the standard curve method, and normalized with the housekeeping gene. Primers used in this study are listed in the Table. 

Table. 

 

Primer Sequences for Real-Time RT-PCR

Gene	Sequences
C4	Forward: ACCCCCTAAATAACCTGG
	Reverse: CCTCATGTATCCTTTTTGGA
IL-1b	Forward: TCCTTGTGCAAGTGTCTGAAGC
	Reverse: ATGAGTGATACTGCCTGCCTGA
iNOS	Forward: GGCAAACCCAAGGTCTACGTT
	Reverse: TCGCTCAAGTTCAGCTTGGT
TNF-a	Forward: GCCTCTTCTCATTCCTGCTT
	Reverse: CTCCTCCACTTGGTGGTTTG
CCL2	Forward: AGGTCCCTGTCATGCTTCTG
	Reverse: TCTGGACCCATTCCTTCTTG
Casp-1	Forward: CACAGCTCTGGAGATGGTGA
	Reverse: TCTTTCAAGCTTGGGCACTT
GAPDH	Forward: ACTTTGTCAAGCTCATTTCC
	Reverse: TGCAGCGAACTTTATTGATG

Histology

Mouse eyes were collected at 4 days and 4 weeks after paraquat injection (n ≥ 4 eyes at each time point) and fixed in 2.5% (w/v) glutaraldehyde (Fisher Chemicals, Loughborough, UK) for at least 24 hours. Samples were embedded in paraffin and processed for standard hematoxylin and eosin (H&E) staining. 

For transmission electron microscopy, mouse eyes were collected and fixed with 2.5% glutaraldehyde for 48 hours. The anterior segment of the eye, including cornea, iris, lens, and vitreous were removed. The eyecup was cut into small pieces and then postfixed in osmium tetroxide, dehydrated in ethanol, and embedded in epon resin. Ultrathin sections were stained with uranyl acetate and lead citrate, and examined with a transmission electron microscope (Phillips CM 10; Eindhoven, The Netherlands). 

Results

Increased Retinal Damage in CX3CR1^gfp/gfp Mice following Paraquat Injection

Twenty-four hours after intravitreal injection of paraquat (0.75 μM), the mouse fundus lost its normal pink color and appeared pale in both WT (Fig. 1A) and CX3CR1^gfp/gfp mice (Fig. 1B). Some vessel segments were obscured by swollen retinal tissue (black arrow, Figs. 1A, 1B). Patches of whitish lesions were observed in CX3CR1^gfp/gfp mice (arrowheads, Fig. 1B) but not in WT mice (Fig. 1A). By day 4 postinjection (p.i.) the fundus began to regain its pink color (Figs. 1C, 1D). Whitish lesions in CX3CR1^gfp/gfp mice were more severe at this stage (arrowheads, Fig. 1D). By 4 weeks p.i., the mouse fundus had regained normal appearance in WT mice (Fig. 1E). In CX3CR1^gfp/gfp mice, although the fundus had regained its pink color, whitish lesions remained (arrowheads, Fig. 1F) and some vessels, particularly retinal arterioles, had lost their red color and had become ghost vessels (white arrows, Fig. 1F). PBS-treated mice (both WT [Fig. 1G] and CX3CR1^gfp/gfp) did not show any fundus abnormalities when compared with noninjected controls (Fig. 1H). 

Figure 1. 

 

Retinal damage following paraquat injection. C57BL/6J mice or CX3CR1^gfp/gfp mice were injected intravitreally with 0.75 μM (1 μL) paraquat. Control animals were injected with 1 μL PBS. On different days p.i., retinal lesions were examined by TEFI (A–H), or fluorescein angiography (I–N). (A, C, E) Fundus images from paraquat-treated WT mice at 24 hours (A), 4 days (C), and 4 weeks (E) p.i. (B, D, F) Fundus images from paraquat-treated CX3CR1^gfp/gfp mice at 24 hours (B), 4 days (D), and 4 weeks (F) p.i. (G) Fundus image from a PBS-treated mouse 24 hours p.i. (H) Fundus image from a normal noninjected WT mouse. OD, optic disc. (I, K, M) Retinal fluorescein angiographic images from paraquat-treated WT mice at 24 hours (I), 4 days (K), and 4 weeks (M) p.i. (J, L, N) Retinal fluorescein angiographic images from paraquat-treated CX3CR1^gfp/gfp mice at 24 hours (J), 4 days (L), and 4 weeks (N) p.i.

Fluorescein angiography showed that 24 hours after paraquat injection there was a lack of capillary perfusion as well as increased background fluorescence in the fundus of WT (Fig. 1I) and CX3CR1^gfp/gfp mice (Fig. 1J). Four days p.i., retinal capillaries were partially visible by fluorescein angiography (Figs. 1K, 1L), and by 4 weeks p.i., retinal circulation had returned to normal in both WT (Fig. 1M) and CX3CR1^gfp/gfp mice (Fig. 1N). 

Retinal Degeneration Is Exaggerated in CX3CR1^gfp/gfp Mice following Paraquat Injection

H&E staining showed that at 4 days p.i., the neuroretina of paraquat-treated mice was thinner than that of PBS-treated mice, and the thickness continued to decrease at 4 weeks p.i. (Figs. 2A, 2B). All retinal layers, including the inner plexiform layer (IPL), inner nuclear layer (INL), and outer nuclear layer (ONL) were affected (Fig. 2A). The number of photoreceptor nuclei was reduced in all retinal areas in both WT and CX3CR1^gfp/gfp mice following paraquat injection (Fig. 2C), suggesting panretinal degeneration. It is interesting to note that the reductions in neuroretinal thickness (Fig. 2B) and number of photoreceptor nuclei (Fig. 2C) were greater in CX3CR1^gfp/gfp mice than in WT mice (Figs. 2B, 2C). 

Figure 2. 

 

Histology of retinal tissue. C57BL/6J mice or CX3CR1^gfp/gfp mice were injected intravitreally with 0.75 μM paraquat. On day 4 or 4 weeks p.i., eyes were collected for H&E staining. Control eyes were collected from PBS-treated mice at 4 days p.i. (A) Representative images from PBS-injected control eyes, paraquat-injected eyes from WT and CX3CR1^gfp/gfp mice at day 4 or 4 weeks p.i. GL, ganglion layer; Ch, choroid. (B) Retinal thickness measured at the equator area from different groups of mice. n ≥ 12. *P < 0.05, unpaired Student's t-test. (C) The number of photoreceptor nuclei in different areas of retina from different groups of mice. *P < 0.05; **P < 0.01 compared with PBS-treated mice of the same strain at the same location. n ≥ 12, ANOVA (Kruskal–Wallis test).

In addition to the reduction of retinal thickness, discrete cell nuclei were detected at the photoreceptor inner and outer segments as well as in the subretinal space in paraquat-treated mice (Fig. 2A, top panel). Occasionally, small vacuoles were observed in retinal pigment epithelial (RPE) cells in WT mice 4 weeks after treatment (asterisk, Fig. 2A, top panel). However, in CX3CR1^gfp/gfp mice, many large vacuoles were observed in RPE cells at 4 days p.i. (asterisks, Fig. 2A, bottom panel). The patches of RPE vacuolation may be the cellular basis of whitish lesions detected in fundus macroscopic investigation (Figs. 1B, 1D, 1F). By 4 weeks p.i., most of the vacuolated RPE cells disappeared; instead, areas of pigmented scar-like changes were observed (arrows, Fig. 2A, bottom panel). Choroidal atrophy was observed in both WT and CX3CR1^gfp/gfp mice 4 weeks following paraquat treatment (Fig. 2A). 

Transmission electron microscopy (TEM) on day 4 p.i. revealed abundant membrane destruction and loss of electron density in the cytosol in the retinal ganglion layer in WT (Fig. 3A, asterisks) and CX3CR1^gfp/gfp mice (Fig. 3E). Membrane destruction was also observed in the IPL. Vacuolated changes were observed in the ONL in WT (Fig. 3B, arrows) and CX3CR1^gfp/gfp mice (Fig. 3F). Fewer photoreceptor nuclei were observed in CX3CR1^gfp/gfp mice (Fig. 3F) compared with WT mice (Fig. 3B). Photoreceptor cell nuclei were detected in the photoreceptor outer segment layer and at the subretinal space in WT mice (Fig. 3C). Small vacuoles were observed in RPE cells in paraquat-treated WT mice (arrowheads, Figs. 3C, 3D). Occasionally, giant melanophagosomes were detected in RPE cells (Fig. 3D). Similar to H&E staining, large RPE vacuoles were observed in paraquat-treated CX3CR1^gfp/gfp mice by TEM (Fig. 3G). In the area of large RPE vacuoles, photoreceptor inner and outer segments were absent (Fig. 3G). Occasionally, condensed melanin granules along with other electron-dense materials were detected in RPE cells in CX3CR1^gfp/gfp mice (Fig. 3H). 

Figure 3. 

 

Transmission electron micrographs of mouse retina. C57BL/6J mice or CX3CR1^gfp/gfp mice were injected intravitreally with 0.75 μM paraquat. Four days later, eyes were collected for TEM. (A–D) TEM images from paraquat-treated WT mice showing membrane disruption in retinal ganglion cell and inner plexiform layers (A), photoreceptor damage (B), photoreceptor nuclei (Nu) in photoreceptor outer segment and subretinal space (C) and giant melanophagosome in RPE cells (D). (E–H) TEM images from paraquat-treated CX3CR1^gfp/gfp mice showing membrane disruption in retinal ganglion cell and inner plexiform layer (E), vacuoles in photoreceptors (F), large vacuoles in RPE cells (G), and condensed melanin granules inside RPE cells (H). Nu, nucleus.

Retinal Immune Gene Expression following Paraquat Injection

To understand how the retinal immune system responds to acute oxidative insult-mediated tissue damage, we first examined immune gene expression in retinal tissues. Two days after paraquat injection, the expression of TNF-α, iNOS, and Ccl2 genes was significantly increased in the neuroretina of WT mice (Fig. 4). Other genes, including IL-1β, Casp-1, and complement C4 genes were not affected as compared with control eyes (Fig. 4). In CX3CR1^gfp/gfp mice, paraquat treatment also increased the expression of TNF-α, iNOS, and Ccl2 genes but to a greater extent than in WT mice (Fig. 4). In addition, IL-1β and Casp-1 were significantly increased by paraquat treatment in CX3CR1^gfp/gfp mice, whereas the complement C4 gene was not affected (Fig. 4). 

Figure 4. 

 

Retinal immune activation following paraquat injection. Retinal immune gene expression 2 days after paraquat injection. n = 6, †P < 0.05, ††P < 0.01 compared to control WT mice; *P < 0.05, **P < 0.01 compared with paraquat-injected WT mice. Unpaired Student's t-test.

Increased Microglial Activation in Paraquat-Treated CX3CR1^gfp/gfp Mouse

Microglial cells are the main resident immune cells in the retina. To understand how microglial cells respond to paraquat-mediated retinal damage, we took advantage of CX3CR1^gfp/+ mice, in which one allele of the CX3CR1 gene is replaced with Gfp and myeloid-derived cells, including microglial cells, express functional CX3CR1, and myeloid cell function is not affected compared with WT mice. ²² PBS injection did not induce significant morphologic changes in microglial cells in either inner (Fig. 5A) or outer (Fig. 5B) retinal layers. Two days after paraquat injection, microglial cells displayed large cell bodies and shorter dendrites in both inner (Figs. 5C, 5E) and outer (Figs. 5D, 5F) retinal layers. Significantly more activated microglial cells were observed in paraquat-treated retinae of CX3CR1^gfp/gfp mice (Figs. 5E, 5F) compared with those of CX3CR1^gfp/+ mice (Figs. 5C, 5D) in both the inner retinal vascular layer (Figs. 5C, 5E, 5I) and outer vascular layer (Figs. 5D, 5F, 5I). Between 2 and 4 days postparaquat injection, a significant population of microglial cells displayed multiple vacuoles (arrowheads, Figs. 5G, 5H), an indication of phagocytic activity. However, this phenomenon was more profound in paraquat-treated CX3CR1^gfp/gfp mice (Fig. 5H) compared with the counterpart CX3CR1^gfp/+ mice (Fig. 5G). 

Figure 5. 

 

Microglial activation in CX3CR1^gfp/gfp mice and CX3CR1^gfp/+ mice. Mice were injected intravitreally with PBS or paraquat. At different times, eyes were collected and retinal flatmounts were prepared and stained for collagen IV. (A, B) Confocal microscopic images of retinal flatmounts from CX3CR1^gfp/+ mice treated with PBS at 2 days p.i. (C, D) Confocal images from the inner (C) and outer (D) retina of a CX3CR1^gfp/+ mouse 2 days after paraquat injection. (E, F) Confocal images from the inner (E) and outer (F) retina of a CX3CR1^gfp/gfp mouse 2 days after paraquat injection. (G, H) Confocal image from the outer retina of a CX3CR1^gfp/+ mouse (G) or CX3CR1^gfp/gfp mouse (H) 4 days after paraquat injection. Arrows indicate cytoplasm vacuoles. (I) The number of GPF⁺ microglial cells in the inner and outer retinal vascular layer in different strains of mice 2 days after paraquat injection. *P < 0.05; **P < 0.01, n = 6. Tukey's Multiple Comparison Test.

Enhanced Immune Cell Infiltration in CX3CR1^gfp/gfp Mouse Retina following Paraquat Injection

CD44 is known to be expressed by a variety of activated leukocytes. ^28,29 Immunostaining of retinal flatmounts from paraquat-treated CX3CR1^gfp/+ mice revealed a substantial number of CD44⁺ infiltrating leukocytes, particularly at the ganglion cell layer (Fig. 6A). Infiltrating cells are small and round, have no dendrites, and can be easily differentiated from activated microglial cells (Fig. 6A). The majority of CD44⁺ infiltrating leukocytes are negative or weakly positive for CX3CR1 (Fig. 6A), and a small population of CD44⁺ cells also express MHC-II (Fig. 6A). The lack of CX3CR1 in the majority of CD44⁺ cells suggests that they are likely to be infiltrating neutrophils or B/T lymphocytes. Further staining using GR1 antibody confirmed that GR1⁺CX3CR1⁻ (neutrophils) and GR1⁺CX3CR1⁺ (monocytes) cells were present in paraquat-treated retinae (Fig. 6B). CD49b⁺ NK cells were not detected in the retina (Fig. 6B). CD44⁺ infiltrating leukocytes were not detected in PBS-treated retinae (data not shown). 

Figure 6. 

 

Leukocyte infiltration in paraquat-treated CX3CR1^gfp/+ and CX3CR1^gfp/gfp mouse retina. Two days after paraquat injection, mouse eyes were collected and retinal flatmounts were immunostained for CD44-PE and MHC-II-APC (A, C), or CD49-APC and GR1-PE (B, D) and examined by confocal microscopy. (A, B) Confocal images from paraquat-treated CX3CR1^gfp/+ mice. (C, D) Confocal images from paraquat-treated CX3CR1^gfp/gfp mice. (E) The number of CD44⁺, MHC-II⁺, and GR1⁺ cells in paraquat-treated CX3CR1^gfp/+ and CX3CR1^gfp/gfp mouse retina. *P < 0.05; **P < 0.01 compared with CX3CR1^gfp/+ mice, n = 6 eyes. Unpaired Student's t-test.

CD44⁺ infiltrating leukocytes, including CD44⁺MHC-II⁺, CD44⁺MHC-II⁻ (Fig. 6C), and GR1⁺CX3CR1⁺ (Fig. 6D) cells were also detected in paraquat-treated CX3CR1^gfp/gfp mice. The infiltrating cells were predominantly located at the inner retinal layer, and were rarely detected at the outer retinal layer. Significantly more infiltrating leukocytes were detected in CX3CR1^gfp/gfp mice (Figs. 6C, 6D, 6E) as compared with CX3CR1^gfp/+ mice (Figs. 6A, 6B, 6E). 

Discussion

Paraquat is known to cause acute oxidative insults and induce neurodegeneration. ^30,31 Previous studies have shown that intravitreal injection of 0.75 μM paraquat induces retinal degeneration via oxidative injury. ^21,32 In this study, we show that paraquat-induced retinal degeneration is exaggerated in mice with fractalkine receptor CX3CR1 deficiency (i.e., CX3CR1^gfp/gfp mice). We further show that the enhanced retinal degeneration is related to increased immune activation in CX3CR1^gfp/gfp mice. Our results suggest that the CX3CL1–CX3CR1 pathway may have neuroprotective roles in the paraquat-induced retinal degeneration model. 

Twenty-four hours after paraquat injection, the fundus had a pale appearance in all mice. Fluorescein angiography revealed a lack of retinal capillary perfusion (Figs. 1I, 1J), suggesting retinal ischemia. The choroid may also suffer from ischemia at this time and choroidal capillary nonperfusion may contribute in part to the pale appearance of the fundus. The thickness of the choroid reduced significantly at 4 weeks after paraquat injection (Fig. 2A), suggesting choroidal atrophy. Oxidative insults and ischemia/reperfusion may both contribute to tissue damage in this model; however, definition of the relative contributions of each type of insult to retinal degeneration remains elusive. 

All layers of the retina were affected by paraquat treatment, and retinal degeneration was more severe in CX3CR1^gfp/gfp mice than that in WT mice. Although we were not able to quantify the levels of inner retinal damage, significantly fewer photoreceptors remained in paraquat-treated CX3CR1^gfp/gfp mice than in WT mice (Fig. 2C). In addition, CX3CR1^gfp/gfp mice also had more severe RPE damage than did WT mice. 

Microglia, the resident retinal innate immune cells, provide the first line of defense against tissue insults. Following paraquat injection, microglial cells were activated and appeared to have phagocytic activities (Fig. 5). In a model of paraquat-induced dopaminergic cell degeneration, microglial activation is known to be the priming event. ³³ In the absence of microglia, paraquat-mediated dopaminergic neurotoxicity is attenuated. ³⁴ A previous in vitro study has shown that paraquat does not directly activate microglia. ³⁵ Microglial activation is, therefore, a consequence of paraquat-mediated retinal insults, and may be involved in clearing dead cells and damaged molecules (Figs. 5G, 5H). In addition to activation (e.g., morphologic change from resting state to active state), the number of microglial-like cells also increased significantly, which was accompanied by infiltration of circulating leukocytes. We have shothat bone marrow–derived monocytes can infiltrate the retina and differentiate into cells with microglial phenotype. ^17,36 Previous studies have also shown that microglia in the CNS ^37,38 and the retina ³⁹ can proliferate in situ under various pathologic conditions. The increased number of microglial-like cells in paraquat-treated mice could be attributed to immune cell infiltration and in situ proliferation. 

Although the physiologic role of the retinal innate immune response may be to remove dead cells and damaged molecules and maintain homeostasis, uncontrolled immune activation may be detrimental. The CX3CL1–CX3CR1 pathway is known to play an important role in regulating microglial activation in the CNS. ^5,9,10 In the retina, CX3CR1 appears to have a redundant role in microglial homeostasis, ⁸ radiation-mediated para-inflammation, ¹⁷ as well as in uveoretinitis. ¹⁸ In this paraquat-induced retinal degeneration model, however, retinal inflammation, including microglial activation and leukocyte infiltration, is massively increased in the absence of functional CX3CR1. We propose that in paraquat-treated WT mice, retinal degeneration is predominantly mediated by oxidative and ischemia/reperfusion insults. Retinal immune response (e.g., microglial activation and recruitment of circulating leukocytes) is tightly controlled and may have a limited contribution to retinal degeneration. In paraquat-treated CX3CR1-deficient mice, oxidative and ischemia/reperfusion insults are also important (similar contributions compared with WT mice) to the initial retinal damage. The lack of the CX3CR1 pathway results in uncontrolled microglial activation and excessive production of proinflammatory cytokines (e.g., TNF-α, IL-1β) and chemokines (e.g., CCL2), which, in turn, recruit more circulating immune cells and may also induce microglial proliferation. The exaggerated retinal immune response following oxidative and ischemia/reperfusion insults may further induce severe retinal degeneration. 

Previous studies have shown that the CX3CR1 deficiency is protective in ischemia/reperfusion–mediated brain injury ^14,15 through reducing the recruitment of Ly6C^lo/iNOS⁺ macrophages, ¹⁶ which contrasts with our observations in this study. It is possible that the role of the CX3CR1 pathway in neural damage/protection is insult/inflammation–dependent. When tissue damage is caused predominantly by infiltrating macrophages and active T cells such as in uveoretinitis, CX3CR1⁺ microglia may have redundant roles. Under oxidative insult conditions, where microglial activation constitutes the main component of immune response (instead of infiltrating CX3CR1⁺ macrophages), a functional CX3CR1 pathway is important to limit inflammation-mediated tissue damage. Although CD44⁺ infiltrating leukocytes were detected in paraquat-treated WT mice, the number was low (∼100 cells/retina, Fig. 6E), and inflammation may be dominated by activated microglial cells in this condition. 

In summary, we show in this study, that CX3CR1 plays an important role in regulating the retinal immune response to ischemic and oxidative insults. In the absence of CX3CR1, retinal inflammation is exaggerated and is related to increased retinal degeneration under ischemic and oxidative conditions. 

Acknowledgments

The authors thank Ulrich Luhmann (Institute of Ophthalmology, University College of London) for providing ear clip samples from rd8^−/− mice for DNA sequencing study and David Simpson for help with English expression. 

Footnotes

 Supported by Fight for Sight Grant 1362.

Footnotes

 Disclosure: M. Chen, None; C. Luo, None; R. Penalva, None; H. Xu, None