Experimental Autoimmune Uveoretinitis (EAU)-Related Tissue Damage and Angiogenesis Is Reduced in CCL2^−/−CX₃CR1^gfp/gfp Mice

Abstract

Purpose.: To investigate the roles of the CCL2-CCR2 and CX₃CL1-CX₃CR1 pathways in experimental autoimmune uveoretinitis (EAU)-mediated retinal tissue damage and angiogenesis.

Methods.: The C57BL/6J wild-type (WT) and CCL2^−/−CX₃CR1^gfp/gfp (double knockout [DKO]) mice were immunized with IRBP_1-20. Retinal inflammation and tissue damage were evaluated clinically and histologically at different days postimmunization (p.i.). Retinal neovascular membranes were evaluated by confocal microscopy of retinal flat mounts, and immune cell infiltration by flow cytometry.

Results.: At day 25 p.i., DKO mice had lower clinical and histological scores and fewer CD45^highCD11b⁺ infiltrating cells compared with WT mice. The F4/80⁺ macrophages constitute 40% and 21% and CD11b⁺Gr-1⁺Ly6G⁺ neutrophils constitute 10% and 22% of retinal infiltrating cells in WT and DKO mice, respectively. At the late stages of EAU (day 60–90 p.i.), DKO and WT mice had similar levels of inflammatory score. However, less structural damage and reduced angiogenesis were detected in DKO mice. Neutrophils were rarely detected in the inflamed retina in both WT and DKO mice. Macrophages and myeloid-derived suppressor cells (MDSCs) accounted for 8% and 3% in DKO EAU retina, and 19% and 10% in WT EAU retina; 71% of infiltrating cells were T/B-lymphocytes in DKO EAU retina and 50% in WT EAU retina.

Conclusions.: Experimental autoimmune uveoretinitis–mediated retinal tissue damage and angiogenesis is reduced in CCL2^−/−CX₃CR1^gfp/gfp mice. Retinal inflammation is dominated by neutrophils at the acute stage and lymphocytes at the chronic stage in these mice. Our results suggest that CCR2⁺ and CX₃CR1⁺ monocytes are both involved in tissue damage and angiogenesis in EAU.

Introduction

Chemokines and chemokine receptors are essential for homing of immune cells under patho-physiological conditions. Chemokines are classified into two functional groups: homeostatic chemokines that are constitutively produced in certain tissues and are responsible for basal leukocyte migration, and inflammatory chemokines that are produced under pathological conditions and are essential for migration of immune cells into the site of inflammation. In the retina, chemokine CX₃CL1 (also known as fractalkine in humans) is constitutively produced by retinal neurons¹ and its receptor CX₃CR1 is exclusively expressed by microglia.² The CX₃CL1-CX₃CR1 pathway is known to negatively regulate microglial activation under patho-physiological conditions.² The CX₃CL1 also is an important cytokine for homing of CX₃CR1⁺ monocytes to tissues.³ Chemokine CCL2 is a typical inflammatory cytokine. Many retinal cells, including retinal pigment epithelia (RPE) and microglia can produce large amounts of CCL2 under inflammatory conditions.^4,5 The CCL2 is crucial for the recruitment of CCR2⁺ monocytes to site of inflammation. 

The role of the CX₃CL1-CX₃CR1 and CCL2-CCR2 pathways in retinal health and disease has been studied extensively in the past decade. Although the CX₃CL1-CX₃CR1 pathway critically controls microglial activation, deletion of CX₃CR1 does not affect the recruitment and distribution of retinal microglia under normal physiological conditions.⁶ We have shown that the retinal microglia can be replaced by bone marrow–derived myeloid cells within 6 months after whole-body irradiation,⁷ and the recruitment of these cells is CCL2-CCR2 but not CX₃CL1-CX₃CR1 pathway dependent.⁸ However, under aging conditions, it has been reported that CX₃CR1 KO mice⁹ and CCL2 KO or CCR2 KO^10,11 mice develop atrophic lesions akin to human AMD. Although the early onset of retinal lesions in the CCL2/CX₃CR1 double knockout (DKO) mice reported by Tuo et al.¹² has been shown to be related to rd8 mutation in the crumbs1 gene,¹³ we observed age- and light-dependent development of localized retinal atrophies in the CCL2^−/−CX₃CR1^gfp/gfp mice that do not carry the rd8 mutation,¹⁴ suggesting that CCL2-CCR2 and CX₃CL1-CX₃CR1 pathways are involved in retinal defence under chronic stress conditions. 

The role of the CCL2-CCR2 and CX₃CL1-CX₃CR1 pathways in retinal autoimmune response has been investigated in experimental autoimmune uveoretinitis (EAU), a model for human noninfectious posterior uveitis.¹⁵ Although macrophage recruitment is impaired in CCL2 KO or CCR2 KO EAU mice, the severity of inflammation remains unchanged.¹⁶ Further studies suggest that retinal inflammation in CCL2 KO or CCR2 KO EAU mice^16,17 is dominated by neutrophils. The deletion of CX₃CR1 does not affect EAU,¹⁸ although one study reported increased disease severity in CX₃CR1 KO EAU mice.¹⁹ Furthermore, a recent study by London et al.²⁰ suggests that CX₃CR1^hi infiltrating macrophages are involved in the resolution of inflammation, and the CCL2-CCR2 pathway is required for the recruitment of myeloid-derived suppressor cells (MDSCs).²⁰ These data suggest that both CCL2-CCR2 and CX₃CL1-CX₃CR1 pathways may be involved in the late stages of inflammation in EAU. 

We have shown that EAU in wild-type (WT) C57BL/6J mice does not spontaneously resolve, and chronic inflammation persists for more than 4 months.¹⁷ We also have shown that chronic inflammation induces retinal angiogenesis,¹⁷ which is driven by CCL2 ligation.¹⁷ The roles of the CCL2-CCR2 and CX₃CL1-CX₃CR1 pathways in EAU, in particular, during the chronic stage warrant further investigation. 

In this study, we show that the severity of inflammation is reduced in CCL2^−/−CX₃CR1^gfp/gfp mice during the acute stage but not the chronic stage. Despite similar levels of inflammation at the late stages of EAU, retinal structural damage and angiogenesis are reduced in the CCL2^−/−CX₃CR1^gfp/gfp mice. Retinal inflammation in CCL2^−/−CX₃CR1^gfp/gfp mice is dominated by neutrophils at the acute stage and lymphocytes at the late stages of EAU. 

Methods

Animals

All mice used in this study were 2 to 3 months old at the time of EAU induction. The C57BL/6J mice were used as WT controls. The CCL2^−/−CX₃CR1^gfp/gfp (DKO) mice were generated using CCL2^−/− mice (B6.129S-Ccl^tm1R°l/J; Jackson Laboratory, Bar Harbor, ME, USA) and CX₃CR1^gfp/gfp mice²¹; described previously,¹⁴ DNA sequencing confirmed that the mice do not carry the rd8 mutation in the crumbs1 gene.¹⁴ All mice were maintained in the Biological Research Unit (BRU) at Queen's University Belfast, United Kingdom. All in vivo procedures were conducted under the regulation of the UK Home Office Animals (Scientific Procedures) Act 1986, and the study was in compliance with the ARVO Statement for the Use of Animals in Ophthalmology and Vision Research. 

Experimental Autoimmune Uveoretinitis Induction

Experimental autoimmune uveoretinitis was induced in mice using a protocol described previously.^17,22 Briefly, mice were immunized with IRBP_1-20 (GPTHLFQPSLVLDMAKVLLD, 10 mg/mL; GL Biochem, Shanghai, China) emulsified 1:1 in complete Freund's adjuvant (DIFCO Laboratories, Detroit, MI, USA) with an additional 2.5 mg/mL Mycobacterium tuberculosis H37Ra (DIFCO Laboratories); 50 μL emulsion was injected subcutaneously into each thigh (500 μg peptide per mouse). An additional 1 μg Bordetella pertussis toxin (Tocris Bioscience, Bristol, UK) was administered intraperitoneally immediately after peptide injection. 

Clinical Fundus Examination

Retinal inflammation was assessed clinically by the topical endoscopic fundus imaging system and scored using the criteria described previously (Supplementary Table S1).²³ Inflammation and structure damage were scored separately. Mouse pupils were dilated with 1% atropine sulphate and 2.5% phenylephrine hydrochloride (Chauvin, Essex, UK). The animals were then anesthetized using isoflurane (Merial Animal Health, Ltd., Essex, UK). Fundus images were captured by a Nikon D90 camera (Nikon UK Ltd., Surrey, UK) via an endoscope and saved in TIF format. 

Histopathological Examination

Mouse eyes were enucleated on days 25 and 90 post immunization (p.i.) and fixed in 2.5% (wt/vol) glutaraldehyde (Agar Scientific Ltd., Stansted, UK) for at least 24 hours. Eyes were then embedded in paraffin and processed for hematoxylin and eosin (H&E) staining. For each eye, four sections from four different layers were graded. The infiltrating and structural scores of retinal inflammation were graded using the criteria described previously.²⁴ 

Flow Cytometry

Preparation of Cells From Mice Blood and Tissues.

Mice were killed by CO₂ inhalation and blood collected by cardiac puncture; 100 μL blood was used for FACS staining. The spleens were homogenized and passed through a 100-μm cell strainer (BD Labware, Oxford, UK) to obtain single-cell suspension. Red blood cells were removed with lysis buffer and 2 × 10⁵ splenocytes were used for FACS staining. The eyes were enucleated, and the retinas were dissected in cold PBS. The retinas were then treated with 1 mg/mL collagenase I (Sigma-Aldrich, Dorset, UK) at 37°C for 30 minutes. The single-cell suspension was washed and filtered through a 100-μm cell strainer (BD Labware). The cell suspension from one retina was used for each FACS staining. 

FACS Staining and Acquisition.

After blocking the Fcγ receptor, the cells were incubated with 50 μL fluorochrome-conjugated antibody cocktail (Table 1) for 40 minutes on ice. The samples were washed with FACS buffer, re-suspended in 200 μL FACS buffer, and processed for FACS analysis using the BD FACSCantoII (BD Biosciences, Oxford, UK). The data were subsequently analyzed by FlowJo Software (TreeStar, Inc., Ashland, OR, USA). To quantify the cell number in the retinas, all samples were run at a constant flow speed. The cell number (N) is calculated by N = T/t × n. T is the time required to run 200 μL FACS buffer, t is the time used for each sample, and n is the cell number acquired. 

Table 1

 

Fluorochrome-Conjugated Antibodies Used in Flow Cytometry

Target Molecule	Conjugated Fluorochrome	Origin	Clone	Dilution	Company
CD11c	APC	Hamster	HL3	1:100	BD Biosciences
B220	APC	Rat	RA3-6B2	1:100	BD Biosciences
CD8a	APC-Cy7	Rat	53-6.7	1:100	BD Biosciences
Ly6G	APC-Cy7	Rat	1A8	1:100	BD Biosciences
CD4	Pacific blue	Rat	RM4-5	1:100	BD Biosciences
F4/80	PE	Rat	CI:A3-1	1:20	Serotech
Gr-1	PE	Rat	RB6-8C5	1:100	BD Biosciences
CD11b	PE-Cy7	Rat	M1/70	1:100	BD Biosciences
CD45	PerCP	Rat	30-F11	1:100	BD Biosciences

Real-Time RT-PCR

The total RNAs of retinal samples were extracted by RNeasy Mini Kit (Qiagen, West Sussex, UK) according to the manufacturer's instructions; 1 μg total RNA was used for reverse transcription using the SuperScript II Reverse Transcriptase kit (Invitrogen, Paisley, UK) according to the manufacturer's instructions. Murine mRNA expression levels were quantified by real-time PCR using the LightCycler 480 system with SYBR Green I Master (Roche Diagnostics GmbH, Mannheim, Germany). The primers used are listed in Table 2; 18S was used as a housing keeping gene. Gene fold changes were calculated by dividing the normalized values of EAU/DKO nonimmunized samples by normalized values of WT nonimmunized samples. 

Table 2

 

Primers Used in Real-Time RT-PCR

Targets	Annealing Temperature, °C	Sequences, 5′–3′
18s	58	Forward	AGGGGAGAGCGGGTAAGAGA
		Reverse	GGACAGGACTAGGCGGAACA
Arg-1	61	Forward	TTATCGGAGCGCCTTTCTCAA
		Reverse	TGGTCTCTCACGTCATACTCTGT
Ccl2	58	Forward	AGGTCCCTGTCATGCTTCTG
		Reverse	TCTGGACCCATTCCTTCTTG
Ccl5	56	Forward	ACTCCCTGCTGCTTTGCCTAC
		Reverse	GAGGTTCCTTCGAGTGACA
Cx3cl1	59	Forward	ACGAAATGCGAAATCATGTGC
		Reverse	CTGTGTCGTCTCCAGGACAA
Cxcl2	58	Forward	AAGTTTGCCTTGACCCTGAA
		Reverse	AGGCACATCAGGTACGATCC
Cxcl10	58	Forward	GGATGGCTGTCCTAGCTCTG
		Reverse	ATAACCCCTTGGGAAGATGG
Il1b	58	Forward	TCCTTGTGCAAGTGTCTGAAGC
		Reverse	ATGAGTGATACTGCCTGCCTGA
NOS2	58	Forward	GGCAAACCCAAGGTCTACGTT
		Reverse	TCGCTCAAGTTCAGCTTGGT
Tnfa	58	Forward	GCCTCTTCTCATTCCTGCTT
		Reverse	CTCCTCCACTTGGTGGTTTG
Vegfa	58	Forward	CCCACGTCAGAGAGCAACAT
		Reverse	TTTCTTGCGCTTTCGTTTTT

Retinal Flat Mount Staining

Mouse eyes were fixed with 2% paraformaldehyde (PFA; Agar Scientific, Ltd., Cambridge, UK) at room temperature for 2 hours. The retinas were dissected as previously described,²⁵ and permeabilized with 0.3% Triton X-100 (Sigma-Aldrich) in PBS for 6 hours. The samples were then incubated with anti-collagen IV antibody (rabbit anti-mouse; AbD Serotech, Kidlington, UK) at 4°C overnight. After washing with PBS, the samples were incubated with donkey anti-rabbit Alexa Fluor 594 for 3 hours. Samples were examined by confocal microscopy (Eclipse TE200-U; Nikon UK Ltd.). The NIS Element (Nikon) software was used for image analysis. 

Statistics

Experimental autoimmune uveoretinitis clinical and histological scores were compared by Mann-Whitney U test. Angiogenesis data of WT and DKO mice were analyzed using an unpaired two-tailed Student's t-test. The immune cell subsets, and immune-related gene expression in WT and DKO mice at different stages of EAU were analyzed by two-way ANOVA. All data were presented as mean ± SEM. Probability values of less than 0.05 were considered as statistically significant. 

Results

Retinal Inflammation in CCL2^−/−CX₃CR1^gfp/gfp EAU Mice

Nonimmunized WT and DKO mice presented no fundus abnormalities (Figs. 1AI, 1AII). At day 15 to 25 p.i., all mice developed retinal inflammation, characterized by swelling of the optic disc (papillitis), engorged blood vessels with severe perivascular cuffing (vasculitis), and multiple whitish retinal infiltrates (retinitis) (Figs. 1AIII, 1AIV). Clinical scores of retinal inflammation (papillitis, vasculitis, and infiltrates) and structural damage (retinal atrophy, the number and size of retinal lesions)²³ were slightly but statistically insignificantly lower in DKO EAU mice compared with those in WT EAU mice (inflammation score: WT versus DKO: 1.6 ± 0.2 vs. 1.4 ± 0.1, mean ± SEM, n = 10; structural damage score: WT versus DKO: 0.9 ± 0.2 vs. 0.6 ± 0.2, mean ± SEM, n = 10). At day 25 p.i., however, both inflammation score (Fig. 1B) and structural damage score (Fig. 1C) were significantly lower in DKO mice compared with those in WT mice. By day 60 and 90 p.i., inflammation remained visible in all mice (Fig. 1AV–VIII). Severe retinal atrophy and scar-like lesions were frequently observed in WT (Fig. 1AVII) but not in DKO EAU mice (Fig. 1AVIII). Double knockout EAU mice presented mild vascular cuffing and multiple small whitish infiltrates at day 60 and 90 p.i. (Figs. 1AVI, 1AVIII). Although the clinical inflammation score was similar between WT and DKO mice at days 60 and 90 p.i. (Fig. 1B), the clinical structural damage score was significantly lower in DKO mice compared with that in WT mice (Fig. 1C). 

Figure 1

 

Clinical evaluation of retinal inflammation at different stages of EAU. EAU was induced in WT and DKO mice. (A) Fundus examinations were performed in control nonimmunized mice and EAU mice at days 25, 60, and 90 p.i. Control nonimmunized WT (I) and DKO (II) mice; Day 25 p.i. WT (III) and DKO (IV) mice; Day 60 p.i. WT (V) and DKO (VI) mice; Day 80 p.i. WT (VII) and DKO (VIII) mice. (B) Clinical inflammation score (papillitis, vasculitis, and infiltrates) in WT and DKO EAU mice at different stages of EAU. (C) Clinical structural damage score (retinal atrophy, the number and size of retinal lesions) in WT and DKO mice at different stages of EAU. Data were presented as mean ± SEM in (B, C). *P < 0.05, **P < 0.01, Mann-Whitney U test.

Histological examination revealed a considerable number of infiltrating cells in the vitreous cavity and retina in day 25 p.i. WT mice (Fig. 2A). Severe vasculitis (arrow, Fig. 2A) and retinal folds were frequently observed in WT EAU mice (arrowhead, Fig. 2A). Fewer infiltrating cells and mild vasculitis were observed in DKO EAU mice (Fig. 2B). Both the infiltration score and structural damage score were significantly lower in DKO mice compared with those in WT mice (Fig. 2C). At day 90 p.i., photoreceptor outer segment was largely absent in WT EAU mice (Fig. 2D). Granuloma and large scars also were frequently observed (Fig. 2D). In some cases, severe degeneration of the retina and the total loss of photoreceptors were observed (Fig. 2E). In DKO EAU mice, few infiltrating cells were observed in the retina (Fig. 2F), and the retinal layers remained intact (Fig. 2F). Large granulomas and scar lesions were rarely observed (Fig. 2F), although photoreceptor outer segment damage was observed in some mice (Fig. 2G). The structural damage score was significantly lower in DKO mice compared with WT EAU mice (Fig. 2H), although the infiltrative score between WT and DKO EAU mice did not significantly differ (Fig. 2H). 

Figure 2

 

Histological evaluation of retinal inflammation in WT and DKO mice. Mouse eyes were fixed in 2.5% (wt/vol) glutaraldehyde, embedded in paraffin and processed for H&E staining. (A) Retina from a day 25 p.i. WT EAU mouse showing infiltrating cells in the vitreous, vasculitis (arrow), and retinal folds (arrowhead). (B) Retina from a day 25 p.i. DKO EAU mouse showing fewer infiltrating cells and a mild vasculitis (arrow). (C) Histology scores of WT and DKO EAU mice at day 25 p.i. (D, E) Retinal sections from day 90 p.i. WT EAU mice. (F, G) Retinal sections from day 90 p.i. DKO EAU mice. (H) Histology scores of WT and DKO EAU mice at day 90 p.i. Data were presented as mean ± SEM in (C, H). n = 5. *P < 0.05, Mann-Whitney U test.

Previous studies have shown that the severity of EAU was not affected in CCL2 KO¹⁶ or CX₃CR1 KO¹⁸ mice, and this was further confirmed in this study (Supplementary Fig. S1). Our results suggest that the combined deletion of CCL2 and CX₃CR1 protects the retina from EAU-mediated inflammatory damage. 

Retinal Angiogenesis in Late Stages of EAU in WT and CCL2^−/−CX₃CR1^gfp/gfp Mice

Confocal microscopy of retinal flat mounts from day 90 p.i. EAU mice showed that the neovascular membrane was significantly reduced in DKO mice (17,144 ± 11,094 μm²) compared with that in WT mice (243,163 ± 77,600 μm²) (Figs. 3A, 3B). Neovascular membrane is often associated with RPE detachment resulting from anastomosis.¹⁷ The number of pigmented areas observed in DKO EAU mice is significantly lower compared with WT mice (WT: 8.4 ± 2.0 versus DKO: 0.8 ± 0.6; Figs. 3C, 3D). When comparing the CCK2 KO and CX₃CR1 KO mice with the DKO mice, the chronic EAU-induced retinal angiogenesis was reduced by 93% in DKO EAU mice, 75% in CCK2 KO mice, and 45% in CX₃CR1 KO mice (Supplementary Fig. S2). 

Figure 3

 

Retinal neovascular membrane in day 90 p.i. EAU mice. Retinas from day 90 p.i. EAU mice were fixed in 2% PFA, stained for collagen IV, and examined by confocal microscopy. (A) Representative images showing collagen IV staining in day 90 p.i. WT and DKO mouse retina. (B) Histograms showing the area of neovascular membrane in WT and DKO EAU retina. (C) Representative phase contrast images showing the pigmented patches in day 90 p.i. WT and DKO EAU mouse retina. (D) Histograms show the number of pigmented areas in the retina. Data were presented as mean ± SEM in (B, D). n = 5. *P < 0.05, **P < 0.01, unpaired Student's t-test.

Kinetics of Circulating Immune Cells During Different Stages of EAU in WT and CCL2^−/−CX₃CR1^gfp/gfp EAU Mice

Previously we have shown that there is no difference in different subsets of immune cells in the blood and spleen between WT and DKO mice at different ages.¹⁷ In this study, we also observed no difference between nonimmunized WT and DKO mice (Figs. 4A, 4C). 

Figure 4

 

Flow cytometric analysis of immune cells in the blood and spleen of WT and DKO mice. Leukocytes from blood and spleen of WT and DKO mice were analyzed by flow cytometry for the presence of various immune cells. (A) Bar graph showing the percentages of various immune cells in the blood in nonimmunized WT and DKO mice. (B) The percentage of different subsets of immune cells in the blood at different stages of EAU in WT and DKO mice. (C) Bar graph showing the percentages of various immune cells in the splenic in nonimmunized WT and DKO mice. (D) the percentage of different subsets of immune cells in the spleen at different stages of EAU. Data were presented as mean ± SEM. n = 3. *P < 0.05, **P < 0.01 compared with nonimmunized mice. +P < 0.05 , ++P < 0.01 compared with WT mice of the same time point.

After EAU induction, WT mice experienced increased proportions of CD11b⁺ cells, Ly6G⁺ neutrophils, and F4/80⁺ macrophages in the blood at day 25 p.i. (Fig. 4B). By day 90 p.i., the proportion of Ly6G⁺ neutrophils had returned to basal levels, although CD11b⁺ and F4/80⁺ cells were higher than those in basal levels (i.e., nonimmunized mice) (Fig. 4B). Interestingly, the proportion of CD8⁺ cells decreased significantly in EAU mice compared with nonimmunized controls (Fig. 4B). The proportion of CD4⁺ cells and CD11b⁺Gr-1⁺Ly6G⁻ (i.e., MDSCs) remained unchanged in the blood at day 25 p.i. (Fig. 4B). By day 90 p.i., the proportion of CD4⁺ cells reduced and MDSCs increased compared with that in day 25 p.i. (Fig. 4B). The dynamic change of different subsets of blood leukocytes in DKO EAU mice was similar to that in WT mice apart from F4/80⁺ macrophages, which only slightly increased at day 25 p.i. in DKO mice (Fig. 4B). 

The dynamic change of a different subset of immune cells in the spleen during EAU is similar to that in the blood in WT and DKO mice (Fig. 4D). Double knockout mice had significantly lower levels of F4/80⁺ macrophages compared with WT mice at day 90 p.i. (Fig. 4D). 

Retinal Immune Cell Infiltration in WT and CCL2^−/−CX₃CR1^gfp/gfp EAU Mice

The CD11b⁺CD45^int cells are known to be resident microglial cells and the CD45^hi cells are known to be infiltrating leukocytes.²⁶ Flow cytometry analysis confirmed that the CD11b⁺CD45^int microglial cells (Fig. 5A) were negative for Ly6G, Gr-1, CD4, and CD8 (Fig. 5B, G1 cells), and the CD45^hi cells (Fig. 5A) include Ly6G⁺, Gr-1⁺, CD4⁺, and CD8⁺ cells (Fig. 5B, G2 cells). In nonimmunized animals, the percentage and the absolute number of retinal microglia (CD45^intCD11b⁺) was comparable between WT (0.19% ± 0.04%, 5030 ± 1044 cells/retina) and DKO (0.20% ± 0.02%, 4392 ± 1229 cells/retina) retinas (Fig. 5C). The number of microglia increased significantly in EAU retinas at days 25 and 90 p.i. (Fig. 5C), suggesting in situ proliferation of retinal microglia during inflammation.²⁷ Many more microglial cells were detected in WT EAU retinas compared with DKO EAU retinas (Fig. 5C). The proportion of CD45^hi cells was negligible in retinas from nonimmunized WT and DKO mice, but massively increased during EAU. Wild-type EAU retinas contained 8.4 times more infiltrating cells than DKO retinas at day 25 p.i. and 1.3 times more at day 90 p.i. (Fig. 5D). 

Figure 5

 

Microglia and infiltrating cells in the inflamed retina at different stages of EAU in WT and DKO mice. Retinas from nonimmunized, days 25 and 90 p.i. WT and DKO mice were collected. Single-cell suspension was prepared (see Methods section) and stained for CD45, CD11b, Gr-1, Ly6G, CD4, and CD8. Samples were analyzed by flow cytometry. (A) Dot-plot showing CD11b⁺CD45^int (G1) and CD45^hi (G2) cells. (B) G1 (CD11b⁺CD45^int) cells were Ly6G⁻, Gr-1⁻, CD4⁻, and CD8⁻ and were considered as microglia. The G2 (CD45^hi) cells were heterogeneous, including Ly6G⁺, Gr-1⁺, CD4⁺, and CD8⁺ cells, and were considered as infiltrating leukocytes. (C) The percentage and number of retinal microglia cells in different stages of EAU in WT and DKO mice. (D) The percentage and number of retinal infiltrating cells in different stages of EAU in WT and DKO mice. Data are presented as mean ± SEM. n = 3. *P < 0.05 compared with WT mice of the same time point, unpaired Student's t-test.

Phenotype analysis showed that both myeloid-derived cells, including CD11b⁺F4/80⁺ macrophages (Fig. 6A), CD11b⁺Gr-1⁺Ly6G⁺ neutrophils (Fig. 6B), CD11b⁺Gr-1⁺Ly6G⁻ MDSCs (Fig. 6B), CD11c⁺ dendritic cells (DCs; Fig. 6C), and lymphoid cells, including CD4⁺ and CD8⁺ T cells (Fig. 6D) and B220⁺ B cells (Fig. 6E) were present in both WT and DKO EAU retinas. Lymphoid:myeloid ratio was 0.66 at day 25 p.i., and 0.96 at day 90 p.i. in WT mice. In DKO mice, the ratio was 0.92 and 2.42 at days 25 and 90 p.i. respectively (Fig. 6F). Since the total number of infiltrating cells was lower in DKO mice, the results suggest that the recruitment of myeloid-derived cells to the inflamed retina is impaired in DKO mice, particularly at day 90 p.i. 

Figure 6

 

Phenotype of retinal infiltrating cells in WT and DKO EAU mice. Retinas from days 25 and 90 p.i. WT and DKO mice were collected. Single-cell suspension was prepared (see Methods section) and stained for different cell surface markers and analyzed by flow cytometry. (A) Dot-plot flow cytometry data showing the expression of CD11b and F4/80 in CD45^hi retinal infiltrating cells at days 25 and 90 p.i. (B) Gr-1 and Ly6G expression in CD45^hi retinal infiltrating cells at days 25 and 90 p.i. (C) CD11c⁺ cells in CD45^hi retinal infiltrating cells at days 25 and 90 p.i. (D) CD4 and CD8 expression in CD45^hi retinal infiltrating cells at days 25 and 90 p.i. (E) B220 and CD4 expression in CD45^hi retinal infiltrating cells at days 25 and 90 p.i. (F) Lymphocyte:myeloid cell ratio at different days of EAU in WT and DKO mice. Data are presented as mean ± SEM. n = 3. Unpaired Student's t-test. (H–K) Pie charts showing the percentages of macrophages (F4/80⁺), neutrophils (CD11b⁺Gr-1⁺Ly6G⁺), DCs (CD11c⁺), MDSCs (CD11b⁺Gr-1⁺Ly6G⁻), B cells (B220⁺), and CD4⁺ and CD8⁺ cells in WT and DKO EAU retinas at days 25 p.i. (H, I) and day 90 p.i. (J, K). Data in pie charts are the means of three mice.

At day 25 p.i., F4/80⁺ macrophages and CD11b⁺Gr-1⁺Ly6G⁺ neutrophils constitute 41.7% ± 2.2 % and 10.2% ± 1.5 % (Fig. 6H) of retinal infiltrating cells in WT mice and 21.1% ± 2.6% and 22.4% ± 2.6% in DKO mice (Fig. 6I), suggesting reduced macrophage but increased neutrophil infiltration in DKO mice. There is no significant difference between WT and DKO mice in other subsets of infiltrating cells, including CD4⁺, CD8⁺, B220⁺, CD11c⁺, and CD11b⁺Gr-1⁺Ly6G⁻ cells (Figs. 6 H, 6I). 

By day 90 p.i., neutrophils account for only 1% of infiltrating cells in both WT and DKO mice (Figs. 6J, 6K). Macrophages, DCs, and MDSCs are the main myeloid lineage cells in the inflamed retina (Figs. 6J, 6K). The percentage of CD11c⁺ DCs was similar between WT and DKO mice (Figs. 6J, 6K); however, significantly fewer macrophages (7.5% ± 1.0% vs. 19.7% ± 2.6%, P < 0.05) and MDSCs (3.4% ± 0.5% vs. 10.2% ± 0.5%, P < 0.001) were detected in DKO retina compared with WT retina (Figs. 6J, 6K). In contrast, the population of CD4⁺ and CD8⁺ T cells was increased in DKO EAU retina compared with WT EAU retina (Figs. 6J, 6K). Our results suggest that the recruitment of F4/80⁺ macrophages and CD11b⁺Gr-1⁺Ly6G⁻ MDSCs is impaired in CCL2^−/−CX₃CR1^gfp/gfp EAU mice. 

Inflammatory Gene Expression in Retinas

To further understand the retinal immune response in DKO EAU mice, we examined the gene expression of various chemokines and cytokines in the retinas of WT and DKO mice at different stages of EAU by real-time RT-PCR. 

Following EAU induction, the expression of two CC chemokines (Ccl2 and Ccl5) was massively increased (>200-fold) at day 25 p.i., and the expression remained at high levels (>150-fold) at days 60 and 90 p.i. in WT retinas (Fig. 7A). Double knockout mouse retinas also express high levels of CCL5 mRNA, although it was slightly lower than WT mice at day 25 p.i. The CCL2 mRNA was not detected in DKO mouse retina (Fig. 7A). The expression of CXC chemokines (Cxcl12 and Cxcl10) was markedly upregulated after EAU induction in WT mice, but only mildly increased in DKO mice (Fig. 7A). The expression of Cx3cl1 was slightly increased (less than 5-fold) in the retinas in both WT and DKO mice during EAU. 

Figure 7

 

Inflammatory gene expression in EAU retina in WT and DKO mice. Total RNA of WT and DKO mouse retina was extracted and cDNA was synthesized for real-time PCR. (A) Messenger RNA expression of chemokines Ccl2, Ccl5, Cxcl2, Cxcl10, and Cx3cl1 genes in WT and DKO mice at different stages of EAU. (B) Messenger RNA expression of Tnfa, Il1b, iNOS, Vegfa, and Arg-1 genes at different stages of EAU in WT and DKO mice. The results shown are gene fold change compared with WT nonimmunized control retina. Data are presented as mean ± SEM. n = 6. *P < 0.05, **P < 0.01 compared with DKO mouse retina of the same time point. Unpaired Student's t-test.

In WT EAU retina, the expressions of Tnfa, Il1b, and inducible nitric oxide synthase (iNOS or NOS2) increased significantly at day 25 p.i. and declined during chronic stages (days 60 and 80 p.i.; Fig. 7B). Although the expression of Tnfa was also increased in DKO EAU retina, the expression level was lower than that in WT EAU retina at each time point. The expression of iNOS and Il1b between DKO and WT retinas did not differ at any time point (Fig. 7B). The expression of Vegfa and Arg-1 in WT retinas was increased during the chronic stages of EAU, but only a mild increment of Vegfa was observed in DKO retinas. At day 90 p.i., DKO retinas expressed a significantly lower level of Arg-1 compared with WT retinas (Fig. 7B). 

Discussion

Previous studies have shown that the deletion of CCL2, CCR2, or CX₃CR1 does not affect the severity of EAU.^16,18 In this study, we found that the combined deletion of CCL2 and CX₃CR1 resulted in reduced retinal inflammation at the acute but not the chronic stage of EAU. Although the severity of chronic inflammation was comparable to WT mice, retinal structural damage and angiogenesis were markedly reduced in DKO mice, which appears to be related to impaired recruitment of macrophages and MDSCs to the inflamed retina. 

After immunization, DKO and WT mice had similar kinetic profiles of circulating immune cells (except F4/80⁺ macrophages, which were lower in DKO mice), suggesting that these mice had comparable levels of systemic immune response to IRBP immunization. The F4/80 is expressed predominately by tissue macrophages, although low levels of F4/80 also may be expressed by circulating monocytes (precursor of macrophages).²⁸ During inflammation, monocytes are recruited to the inflamed tissue and differentiate into F4/80⁺ macrophages. The CCL2-CCR2 and CX₃CL1-CX₃CR1 pathways play critical roles in monocyte trafficking to site of inflammation. It has been shown that inflammatory macrophages may not die locally, rather they migrate back to the circulation and end in secondary lymphoid organs where they are cleared by unknown mechanisms.²⁹ The reduced F4/80⁺ macrophages in the blood and spleen of DKO EAU mice may result from decreased tissue macrophage supply. 

At the acute stages of EAU, the ratio of lymphocytes:myeloid cells was slightly higher in DKO mice compared with WT mice. However, DKO retinas had more neutrophils and fewer macrophages compared with WT retinas (Fig. 6), suggesting neutrophil-dominated retinal inflammation at this stage. A previous study showed that retinal inflammation in CCL2 KO EAU mice was mediated predominately by neutrophils.¹⁶ However, unlike the DKO mice, CCL2 KO EAU mice had similar levels of retinal inflammation and tissue damage compared with WT mice¹⁶ (Supplementary Fig. S1). Tumor necrosis factor-α and nitric oxide are known to be the main mediators responsible for tissue damage at the acute stage of EAU.^30,31 Infiltrating macrophages, in particular M1 subset, are the main source of TNF-α during acute inflammation. Other cells, such as neutrophils and CD4⁺ T cells, also can produce this cytokine, but their ability is relatively low.³² Tumor necrosis factor-α expression was significantly lower in DKO mouse retina compared with that in WT mice. In the DKO mice, the trafficking of both CCR2⁺ and CX₃CR1⁺ monocytes is impaired, whereas deletion of CCL2 affects only CCR2⁺ monocyte trafficking. Our results suggest that CX₃CR1⁺ monocyte-derived macrophages may contribute, at least in part, to the severity of inflammation and tissue damage in EAU. 

At the chronic stages of EAU (i.e., day 60–90 p.i.), even though the severity of inflammation did not differ between WT and DKO mice, retinal structural damage, and in particular angiogenesis, is massively reduced in DKO mice. This may be partially explained by less inflammatory damage at the acute stage of EAU; however, the type of infiltrating immune cells may be more important. T cells (CD4⁺ and CD8⁺), B cells, macrophages, DCs, and MDSCs, but not neutrophils, were detected in significant amounts in the inflamed retina in both WT and DKO mice at day 90 p.i. This suggests that neutrophils may play an insignificant role in chronic EAU. Accumulation of CD8⁺ memory T cells has been observed in chronic EAU and they are known to express inhibitory receptors, such as PD-1, that can limit inflammation.³³ However, the amount of retinal infiltrating CD8⁺ T cells was similar between WT and DKO mice in chronic EAU. The MDSCs and F4/80⁺ macrophages constitute 10% and 19% of retinal infiltrating cells in WT mice but only 3% and 8% in DKO mice at day 90 p.i. CD4⁺ T cells constitute 51% of retinal infiltrating cells in DKO EAU mice. The MDSCs play an important role in suppressing inflammation³⁴ and the number is increased in late stages of EAU.³⁵ The reduced MDSCs in DKO mice may explain the sustained CD4⁺ T-cell infiltration and persistent inflammation in the mice. Whereas the reduced macrophage infiltration may account for less tissue damage and angiogenesis (see below discussion). Our results may support the role of the CCL2-CCR2 pathway in MDSC trafficking, reported previously by Sawanobori and colleagues.³⁶ The involvement of the CX₃CL1-CX₃CR1 pathway in MDSC trafficking remains to be elucidated. 

Tissue structural damage and angiogenesis at the late stages of inflammation is related to postinflammation tissue repair and remodeling. Despite comparable levels of chronic inflammation for more than 2 months (i.e., from day 25 to 90 p.i.) in WT and DKO mice, retinal structural damage and angiogenesis were significantly reduced in DKO mice. The expression levels of VEGF in DKO EAU retinas were only slightly lower than those in WT retina at days 60 and 90 p.i., suggesting that other factors might be involved in the chronic inflammation-mediated retinal angiogenesis. Macrophages, in particular M2-type wound-healing cells, are known to play an important role in tissue repair/remodeling. Previously we showed that Arginase-1⁺ macrophages are increased in late stages (i.e., angiogenic stage) of EAU.¹⁷ Whether the wound-healing macrophages originated from CCR2⁺ or CX₃CR1⁺ monocytes or both remains unknown. In this study, we further found that retinal angiogenesis was reduced by 75% in CCL2 KO EAU mice and 45% in CX₃CR1 KO EAU mice. The deletion of both CCL2 and CX₃CR1 resulted in a 93% reduction in EAU-induced retinal angiogenesis. Our result suggests that the CX₃CR1⁺ macrophages may work together with the CCR2⁺ macrophages in postinflammation retinal repair and remodeling, although CCR2⁺ macrophages may play a dominating role. 

In summary, we show in this study that the deletion of CCL2 and CX₃CR1 in mice results in neutrophil-dominated acute inflammation and T-lymphocyte–dominated chronic inflammation in the EAU model. Neutrophil-dominated retinal inflammation is less destructive compared with classic macrophage-dominated inflammation, and postinflammation tissue repair and wound healing is impaired in T-cell–dominated chronic inflammation. Our study highlights the role of both CCR2⁺ and CX₃CR1⁺ macrophages in retinal tissue damage and angiogenesis in EAU. 

Acknowledgments

The authors thank Aisling Lynch for helping with English expression. 

Supported by the National Eye Research Centre (SCAID061) and Fight for Sight (1361/2). 

Disclosure: J. Zhao, None; M. Chen, None; H. Xu, None