Differential Roles of Macrophages and Microglia in Subretinal Fibrosis Secondary to Neovascular Age-Related Macular Degeneration

Abstract

Purpose

To investigate the differential role of infiltrating CCR2⁺ macrophages and CX3CR1⁺ microglia in neovascular AMD (nAMD)-mediated subretinal fibrosis.

Methods

Subretinal fibrosis was induced using the two-stage laser protocol in C57BL/6J or CX3CR1^gfp/+ mice. The fibrotic lesion was detected using collagen-1 staining in retinal pigment epithelial /choroidal flatmounts. Infiltrating macrophages and microglial were identified using F4/80, CCR2, and CX3CR1 markers at one, three, six, and 10 days after the second laser. Circulating CCR2⁺ monocytes were depleted using the MC-21 antibody, whereas CX3CR1⁺ microglia were depleted using PLX5622. BV2 microglia were treated with TGF-β1 for 96 hours, and their profibrotic potential was examined by quantitative PCR and immunocytochemistry.

Results

Subretinal fibrosis lesions developed three days after the second laser, accompanied by persistent CCR2⁺F4/80⁺ macrophage and CX3CR1⁺ cell infiltration. Inflammation in the first three days after the second laser was dominated by filtrating CX3CR1⁺ cells, and the number increased until day (D) 10 post-second laser. Depletion of CCR2⁺ monocytes from D5-10 significantly reduced the vascular and fibrotic components of the lesion, while CX3CR1⁺ cell depletion reduced Isolectin B4⁺ but not collagen-1⁺ lesion size. Bone marrow–derived macrophages from D6 and D10 mice expressed significantly higher levels of α-smooth muscle actin (α-SMA) and collagen-1 compared to cells from D1 and D3. TGFβ1 treatment increased TMEM119, CX3CR1, IL1b and iNOS gene expression but did not affect Acta2 and Col1a1 gene expression in BV2 cells.

Conclusions

CCR2⁺ monocytes, but not CX3CR1⁺ microglia, critically contribute to the development of subretinal fibrosis in nAMD.

Age-related macular degeneration (AMD) is a worldwide condition leading to irreversible sight loss in the elderly. Neovascular AMD (nAMD) is characterized by the growth of new blood vessels, which can originate from the retinal vasculature or the choroid, causing leakage and hemorrhage and retinal pigment epithelial (RPE) and photoreceptor damage.^1,2 The standard treatment for nAMD is intravitreal injections of VEGF inhibitors. However, up to 71% of nAMD patients develop macular fibrosis after 10 years of disease duration with current treatment, highlighting the high prevalence of this condition.^3,4 Subretinal macular fibrosis is one of the main reasons for anti-VEGF resistance in nAMD, and the underlying mechanism is poorly defined. Currently, there is no medication to treat or prevent the conditions.

Subretinal fibrosis in nAMD is the conversion of the new blood vessels (e.g., choroidal neovascularization [CNV]) into fibrovascular membranes.^5,6 Subretinal inflammation, in particular, immune cell infiltration and complement activation, play an important role in the initiation and progression of subretinal fibrosis. Both macrophages and microglia are known to play critical roles in nAMD.^7–9 Phagocyte accumulation in the subretinal space has been observed in the normal aging retina^2,10,11 and models of AMD^12,13. When the retina is damaged, microglia and macrophages infiltrate the damaged area^14–17 to remove debris and initiate retinal repair. If the injury sustains or becomes chronic, infiltrating phagocytes may acquire a fibrovascular phenotype as seen in nAMD.¹⁸ The role of macrophage and microglia in AMD has been well documented previously,^8,15,16 although their roles in subretinal fibrosis remain elusive. Recently, we have shown that macrophages can directly transdifferentiate into myofibroblasts, key fibrotic players, through macrophage-to-myofibroblast transition (MMT) in nAMD-mediated subretinal fibrosis.¹⁸ Macrophages can also produce a range of inflammatory, proangiogenic, and profibrotic factors and promote the conversion of macular new blood vessels into fibro-vascular membranes.^18–21

The CX3CL1/CX3CR1 axis is critically involved in retinal homeostasis,²² and dysfunction of this signaling pathway has been linked to subretinal accumulation of microglia and macrophages during ageing and the development of AMD.¹¹ However, the role of microglia in the progression of subretinal fibrosis remains elusive. Here, we demonstrated that infiltrating macrophages (CCR2⁺ macrophages), not microglia, actively participate in the development and progression of subretinal fibrosis in a mouse model of a two-stage laser-induced subretinal fibrosis.

Methods

Animals

C57BL/6J wild-type and CX3CR1^gfp/+ mice aged between two to three months were used in this study. The mice were screened and confirmed negative for the Crb1 gene rd8 mutation.^23,24 All animals were housed and bred at the Biological Services Unit (Queen's University Belfast, UK) and exposed to 12-hour light/dark cycles with free access to food and water. All in vivo procedures were conducted under the regulation of the UK Home Office Animals (Scientific Procedures) Act 1986 and followed the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmology and Vision Research. Both males and females were included in this study.

Two-Stage Laser-Induced Subretinal Fibrosis Model

To induce subretinal fibrosis, as previously established by Little et al.,⁶ mice received four laser burns/eye at day (D7) using an HGM Medical Laser System (Salt Lake City, UT, USA). The settings for the laser were as follows: laser power, 250 mv; duration, 0.1 s; and spot size, 100 µm. Seven days later (D0), the mice received a second laser burn to the previously formed CNV lesion. To study the dynamic of subretinal fibrosis and cell infiltration, tissues were collected at D0, D1, D3, D6, and D10 after the second laser. Enucleated eyes were fixed in 2% paraformaldehyde for two hours and stored in autoclaved 1× PBS at 4°C or snap-frozen for further RNA extraction.

CX3CR1⁺ Microglia and CCR2⁺ Macrophage Depletion

We used PLX5622 (HY-114153; MedChem Express, Monmouth Junction, NJ, USA), a highly selective brain penetrant and colony stimulating factor 1 receptor inhibitor to deplete microglia in CX3CR1^gfp/+ mice. A 400 mg/mL stock solution of PLX5622 was prepared in DMSO (34943-M; Sigma-Aldrich, St. Louis, MO, USA). The stock solution was then dissolved in 0.5% (hydroxypropyl)methyl cellulose (09963-100G; Sigma-Aldrich) and 1% polysorbate 80 (59924-1KG-F; Sigma Aldrich) to make a 20 mg/mL working solution. PLX5622 20 mg/kg of body weight was administered to each mouse via gavage feeding daily, starting three days before the second laser until three days before the experiment endpoint. Control mice received DMSO in 0.5% (hydroxypropyl)methyl cellulose and 1% polysorbate 80. Fundus images were taken with a Micron IV system (Phoenix-Micron, Inc, Bend, OR, USA) at D3 after the first oral administration to visualize CX3CR1^gfp/+ retinal microglial cells. The depletion efficiency was also evaluated by quantification of CX3CR1^gfp/+ cells in retinal wholemount (three days after the last oral gavage).

Circulating monocytes were depleted using a CCR2-depleting antibody, MC-21 (rat anti-mouse CCR2).²⁵ The antibody (20 µg/mouse) was injected intraperitoneally daily for five days. Control mice received intraperitoneal injections of Rat IgG2b.κ (1 mg/mL; 556968; BD Biosciences, San Jose, CA, USA). Depletion was validated by flow cytometry of peripheral blood samples.

Immunostaining of RPE/Choroidal Flatmounts and Bone Marrow-Derived Macrophages

After four hours in permeabilization-blocking solution (1% triton with 10% fetal calf serum), the RPE/choroid complex was incubated with collagen-1 (fibrotic marker, rabbit anti-collagen 1, ab21286, Abcam, Cambridge, UK), isolectin B4 (B-1205; Vector Laboratories, Newark, NJ, USA), CD44 (BLR038, Abcam), CCR2 (Bs-23026R; Bioss, Woburn, MA, USA) or F4/80 (Ab6640; Abcam, Cambridge, UK) primary antibodies. Samples were then incubated for two hours with the following secondary antibodies: donkey anti-goat IgG AF594 (705-585-147), donkey anti-rabbit IgG-AF488 (711-545-152), donkey anti-rat IgG-AF647 (712-605-153), and streptavidin-AF647 (016-600-084) (all from Jackson ImmunoResearch, West Grove, PA, USA), and goat anti-rat IgG (Ab150157; Abcam, Cambridge, UK). RPE/choroidal wholemounts were counterstained using DAPI (D3571, molecular probes; ThermoFisher Scientific, Waltham, MA, USA). Fixed bone marrow–derived macrophages were stained with antibodies against F4/80 (ab6640, Abcam) and α-SMA (C6198-100UL, Merck, Darmstadt, Germany) followed by goat anti-rat IgG-AF488 (Ab150157, Abcam). Cells were counterstained for nuclei visualization using diluted Vectashield anti-fading medium with DAPI (H1200, Vector Laboratories, Newark, CA, USA). Images were taken with an inverted light microscope (Leica Dmi8; Leica Microsystems, Wetzlar, Germany) or confocal microscope (Leica TSC SP8 inverted microscope; Leica Microsystems).

Quantification of CX3CR1⁺ and F4/80⁺CCR2⁺ Infiltrating Immune Cells

Merged wholemount images were imported into ImageJ, and four circles with a diameter of 300 µm, 500 µm, 700 µm, and 900 µm were added to the ROI manager. Each circle was positioned around the center of each lesion. Infiltrating CX3CR1⁺ cells were manually counted in each circle. For subretinal infiltration, the lesion area was omitted from the quantification (i.e., only CX3CR1⁺ cells outside the 900 µm circle for fibrosis samples were counted). All quantification was blinded by labeling RPE/Choroid complexes with letters or numbers without study group information until the end of the analysis.

Flow Cytometry

Anticoagulated peripheral blood was incubated for 30 minutes, protected from light with CD11b-PE-Cy7 (25-0112-82; eBioscience, San Diego, CA, USA), Ly6C-APC (17-5932-82, eBioscience) and CCR2-PE (150610; Biolegend, San Diego, CA, USA) antibodies. Red blood cells were removed using Red Blood Cell Lysis buffer (420302; Biolegend). Samples were run using a BD FACS Canto II (BD Bioscience, Franklin Lakes, NJ, USA) and analyzed using FlowJo.

RNA Extraction and Reverse Transcription-Quantitative PCR

Total RNA was extracted from RPE/choroid/sclera and retina using the Maxwell RSC simplyRNA Tissue Kit (AS1340; Promega, Madison, WI, USA). Four RPE/choroidal samples were pooled for RNA extraction and quantitative PCR (qPCR) studies.

The extracted RNA was reverse transcribed following the Superscript II protocol (18064-014; Invitrogen, Carlsbad, CA, USA) and the obtained cDNA was used as the template for the qPCR reaction. All qPCR experiments were performed using a 384-well PCR plate and run using the LightCycler 480 384-well plate LHQ (Roche Diagnostics, Basel, Switzerland). The following proinflammatory and anti-inflammatory gene sequences were used (Table).

Table.

Primer Sequences of the Transgenes Used in This Study

Genes	Primer Sequence (5′ – 3′)
iNos	Forward	TCT TTG ACG CTC GGA ACT GTA GCA
	Reverse	ACC TGA TGT TGC CAT TGT TGG TGG
Il-6	Forward	ATC CAG TTG CCT TCT TGG GAC TGA
	Reverse	TAA GCC TCC GAC TTG TGA AGT GGT
Tnf-a	Forward	TCT CAT GCA CCA CCA TCA AGG ACT
	Reverse	ACC ACT CTC CCT TTG CAG AAC TCA
Cd206	Forward	TCA GCT ATT GGA CGC GAG GCA
	Reverse	TCC GGG TTG CAA GTT GCC GT
Arg-1	Forward	GGA ATC TGC ATG GGC AAC CTG TGT
	Reverse	AGG GTC TAC GTC TCG CAA GCC A
Ym-1	Forward	ACC CCT GCC TGT GTA CTC ACC T
	Reverse	CAC TGA ACG GGG CAG GTC CAA A
α-sma/Acta2	Forward	TGG CAC CAC TCT TTC TAT AAC G
	Reverse	GGT CAT TTT CTG CCG GTT GG
Vegfa	Forward	GCA CTG GAC CCT GGC TTT A
	Reverse	CTT GAT CAC TTG ATG GGA CTT CTG
Ccl2	Forward	AGG TCC CTG TCA TGC TTC TG
	Reverse	TCT GAA CCC ATT CCT TCT TG
Cx3cr1	Forward	GAG TAT GAC GAT TCT GCT GAG G
	Reverse	CAG ACC GAA CGT GAA GAC GAG
18S	Forward	AGG GGA GAG CGG GTA AGA GA
	Reverse	GGA CAG GAC TAG GCG GAA CA
Gapdh	Forward	TGG CAA AGT GGA GAT TGT TGC C
	Reverse	AAG ATG GTG ATG GGC TTC CCG
Tmem119	Forward	TGT CTC TGC TGC TAC TTG CG
	Reverse	TCA GGG AAC GAG GAT GGG TA
P2ry12	Forward	TTG CAC GGA TTC CCT ACA CC
	Reverse	ATT GGG GTC TCT TCG CTT GG
Col1a1	Forward	CTG GCG GTT CAG GTC CAA T
	Reverse	TTC CAG GCA ATC CAC GAG C
Fn1	Forward	GCC GTT AGA TGT GCA AGC TG
	Reverse	TGC TGA AGC TGA GAA CTA GGC
Cd68	Forward	TCT GAT CTT GCT AGG ACC GC
	Reverse	TCA TCG TGA AGG ATG GCA GG
Cd86	Forward	TGT TTC CGT GGA GAC GCA AG
	Reverse	TTG AGC CTT TGT AAA TGG GCA
Hif1a	Forward	TGA GTT CTG AAC GTC GAA AAG A
	Reverse	CTG TCT AGA CCA CCG GCA TC
Tgfb1	Forward	CAT CCA TGA CAT GAA CCG GC
	Reverse	GAA GTT GGC ATG GTA GCC CT
Pdgf	Forward	CTG TGT TCC TCT GCC CCT TT
	Reverse	ACA GTG AAG CCC AAC AGC TT
IL-1β	Forward	TCC TTG TGC AAG TGT CTG AAG C
	Reverse	ATG AGT GAT ACT GCC TGC CTG A
IL-10	Forward	TGC AGG ACT TTA AGG GTT ACT TGG
	Reverse	GGC CTT GTA GAC ACC TTG GTC
Hprt1	Forward	CAA ACT TTG CTT TCC CTG GT
	Reverse	TCT GGC CTG TAT CCA ACA CTT C
Aif1	Forward	GAG CCA AAG CAG GGA TTT GC
	Reverse	GCT TCA AGT TTG GAC GGC AG
Trem2	Forward	GTT TCA TCC TGT GGG TCA CCT
	Reverse	TGA GGA TCT GAA GTT GGT GCC
Emr1	Forward	TTC CTC GCC TGC TTC TTC TG
	Reverse	TAG CCA AAG GCA CAG AGG TG
Itgam	Forward	GCT CGA CAC CAT CGC ATC TA
	Reverse	AAG GGA CAC ACT GAC ACC TG

Bone Marrow–Derived Macrophage (BMDM) Culture and Polarization

Bone marrow cells were isolated from the femur and tibia of eight- to 12-week-old mice and cultured with Dulbecco's modified Eagle medium supplemented with 1% penicillin-streptavidin, 15% fetal calf serum, and 20% L929 conditioned media for five to seven days. BMDM phenotype was validated by flow cytometry using CD11b (552850; BD Biosciences) and F4/80 (48-01-80; eBioscience), and > 90% of cells were confirmed to be F4/80⁺CD11b⁺ in all experiments. BMDMs were polarized into M1 using 50 ng/mL LPS + 100 ng/mL IFN-γ (485-MI; R&D Systems, Minneapolis, MN, USA), and M2 with 20 ng/mL of IL-4 (404-ML; R&D Systems) for 24 hours.

Macrophage(Microglia)-to-Myofibroblast Transition (MMT)

Bone marrow cells were collected from the lasered mice on different days (D0, D2, D5, and D10) after the second laser and differentiated into BMDMs as detailed above. The potential of spontaneous MMT of BMDMs was evaluated by qPCR for myofibroblast-related genes including Acta2, Col1a1 and Fn. BV2 microglial cells were treated with culture medium supplemented with TGF-β1 (10 ng/mL) (7666-MB; R&D Systems) for 96 hours (media was replaced with fresh treatment media after 48 hours of culture). Cells were processed for qPCR or immunocytochemistry as detailed above.

Statistical Analysis

All the graphs and statistical analyses were generated using Prism 9 Software (GraphPad Software, San Diego, CA, USA). Outliers were excluded using the GraphPad Outlier Calculator Tool. For data not normally distributed, a Mann-Whitney test was used in experiments involving two groups, and a nonparametric Kruskal Wallis test (with a Dunn's test for multiple comparisons) was used for experiments involving three groups or more.

Results

The Dynamic Change of Subretinal Fibrotic Lesion Size in the Two-Stage Laser-Induced Mouse Model

As the traditional laser-induced CNV does not progress to fibrovascular membrane because of the regression and healing of the CNV after 14 to 21 days,²⁶ our group established a mouse model of two-stage laser-induced subretinal fibrosis, mimicking the clinical features of macular fibrosis in nAMD.⁶ The lesion size stabilizes from D10 to D30 after the second laser (Supplementary Data 1). We set D10 as the post-second laser endpoint in this study.

To understand how fibrosis develops from CNV into fibrovascular membrane, we looked at the dynamic of the fibrotic development between D0 to D10 post-second laser (Fig. 1A). The collagen-1⁺ fibrotic lesion size significantly increased three days after the second laser and maintained a similar size afterward (Fig. 1B, 1C).

Figure 1.

Subretinal fibrosis progression in the two-stage laser-induced mouse model. (A) Diagram showing study design. Subretinal fibrosis was induced in C57BL/6J mice using the two-stage laser-induced model. Eyes were collected at D0, D1, D3, D6, and D10 after the second laser and RPE/Choroid complexes were immunostained with collagen-1 (fibrotic marker), F4/80, and CCR2. Infiltration and accumulation of CX3CR1⁺ macrophages/microglia were studied using the CX3CR1^gfp/− mice. (B) Representative images of collagen-1 (red) staining of the fibrotic lesions in RPE/choroidal wholemounts at different times. Scale bar: 100 µm. (C) Bar/dot plot showing the fibrotic lesion size (mm²) at each time point. n = 39–50 lesions per group. (D) Representative images of collagen-1 (red), F4/80 (magenta), and CCR2 (green) staining of the fibrotic lesions in RPE/choroidal wholemounts at different times after the second laser. Eyes were counterstained with DAPI (blue) for nuclei visualization. Scale bar: 100 µm. (E) Bar/dot plot showing the percentage of F4/80⁺CCR2⁺ cells in the lesion. n = 6–8 lesions per group. (F) Bar/dot plot showing the number of F4/80⁺ cells infiltrating the lesion. n = 6–8 lesions per group. (G) Representative images showing CX3CR1⁺ cells (green – GFP) in the fibrotic lesions in RPE/choroidal wholemounts at different times after the second laser. Scale bar: 100 µm. (H) Representative images of a fibrotic lesion at D10 post-second laser showing the distribution of CX3CR1⁺ microglial cells (green) and F4/80⁺ macrophages (magenta). Samples were counterstained with DAPI (blue). Scale bars: 100 µm. (I) Visual representation of CX3CR1⁺ cell distribution in subretinal fibrotic lesion using circle area method. (J–M) Graph representations of the number of CX3CR1⁺ cells in the 300 µm circle (J), the 300–500 µm band (K), the 500–700 µm band (L) and the 700–900 µm band (M). n = 30–41 lesions per group. Mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 with Kruskal-Wallis test.

The Dynamic Change of Infiltrating Macrophage and Microglia in the Two-Stage Laser-Induced Mouse Model of Subretinal Fibrosis

Macrophages and microglia are known to be critically involved in retinal wound-healing response.^9,14,15 We investigated the dynamic change of F4/80⁺CCR2⁺ infiltrating macrophage and CX3CR1⁺ microglial (using CX3CR1^gfp/+ mice) (Figs. 1D, 1G). We detected very few CCR2⁺F4/80⁺ macrophages within the lesion at D0 and D1 post-second laser, but the number progressively increased from D3 to D10 (Figs. 1D, 1E). The CCR2⁺F4/80⁺ macrophages account for 20% of the F4/80⁺ population in the lesion at D6 and D10 (Fig. 1E). Most CCR2⁺ macrophages were detected inside the fibrotic lesion (Fig. 1D). The number of F4/80⁺ cells decreased by half on D10 compared to D6 (Fig. 1F), indicating a potential withdrawal of macrophages from the lesion.

To track the dynamics of CX3CR1⁺ microglia, we used CX3CR1^gfp/+ mice. Interestingly, we detected many microglia in the lesion site (Fig. 1G). From D0 to D1, CX3CR1⁺ microglia were contained within the lesion, and as the disease progressed to D6 and D10, microglia accumulated at the peripheral and outside the lesion (Figs. 1G, 1H). Co-staining of F4/80 in the CX3CR1^gfp/+ revealed the accumulation of CCR2⁺ infiltrating macrophages at the center of the lesion and CX3CR1^gfp/+ cells in the peripheral and outside the lesion at D10 post-second laser (Fig. 1H).

We further quantified the number of CX3CR1⁺ microglial cells in 300 µm, 500 µm, 700 µm, and 900 µm distance from the lesion center (Fig. 1I). At the center of the lesion (300 µm circle area), CX3CR1⁺ cells increased significantly on D6 post-second laser (Fig. 1J). Outside the fibrotic lesion (300 µm–500 µm), a significant number of CX3CR1⁺ microglia were detected at D0 to D3 and the number increased and reached a plateau phase by D10 (Fig. 1K). Very few CX3CR1⁺ microglia were observed on D0 and D1 in the 700 µm and 900 µm bands, but many cells were detected on D6 and D10 (Figs. 1L, 1M).

Altogether, these results suggest a two-phase dynamic of immune cell infiltration and accumulation in the development of fibrovascular membrane. The acute stage (D0 to D3 post- second laser) is dominated by CX3CR1⁺ microglial cells with minimal fibrosis progression. The chronic stage (D5 to D10 post-second laser) is characterized by the infiltration of CCR2⁺ macrophages at the center of the lesion and CX3CR1⁺ microglia in the peripheral and outside the lesion accompanied by progressive fibrotic lesion development.

The Effect of Microglial Depletion in Subretinal Fibrosis

To investigate the role of infiltrating microglia in the initiation and progression of subretinal fibrosis, we depleted microglia using PLX5622 in CX3CR1^gfp/+ mice. Three days of PLX5622 treatment (20 mg/kg of body weight, daily oral) resulted in 70% microglia depletion from the retina in normal mice (Supplementary Data 2). Mice received PLX5622 treatment three days before the second laser until D7 post-second laser (Fig. 2A). Fundus image and retinal flatmount examination confirmed a significant reduction of CX3CR1⁺ cells in the lasered retina in PLX5622-treated mice (Fig. 2B). Three days after the last oral gavage (D10), the number of CX3CR1⁺ cells in PLX5622-treated subretinal fibrosis mice was ∼50% of that in vehicle-treated subretinal fibrosis mice (Fig. 2C). Surprisingly, we did not detect any significant reduction of collagen 1⁺ fibrotic lesion size in the PLX5622-treated group compared to the vehicle-treated group (Figs. 2D, 2E). However, a significant reduction of the vascular lesion (identified by IB4 staining) was observed following PLX5622 treatment (Figs. 2F, 2G). Our results suggest that CX3CR1⁺ microglia do not play a major role in the development of subretinal fibrosis but may be involved in the progression of subretinal angiogenesis.

Figure 2.

The effect of CX3CR1⁺ microglial cell depletion in subretinal fibrosis. (A) Diagram showing the experimental design. CX3CR1^gfp/+ mice were administered with 20 mg/kg of body weight of PLX5622 or DMSO by oral gavage daily for 11 days starting three days before the second laser. Fundus images were taken on day 10 after the second laser, and the RPE/choroidal complex was immunostained with collagen-1 and Isolectin B4 and examined by confocal microscopy. (B) Fundus images of DMSO control mice and PLX5622-treated mice showing the lesions at D10 post-second laser and CX3CR1⁺ cells distribution (left panel) and retina wholemount images and their zoom-in images (right panel). (C) Bar graph showing the number of CX3CR1⁺ cells in control and PLX5622-treated mouse retina. Mean ± SEM, n = 9–10 eyes per group. ****P < 0.0001, Mann-Whitney test. (D) Representative images of subretinal collagen-1⁺ (red) fibrotic lesion in control and PLX5622-treated mice. Scale bar: 100 µm. (E) Bar graph showing the average size (mm²) of collagen-1⁺ lesions in different groups. Mean ± SEM, n = 34–39 lesions per group. Mann-Whitney Test. ns, nonsignificant. (F) Representative images showing Isolectin B4⁺ (Green) angiogenesis lesion in control and PLX5622-treated mice. Scale bar: 100 µm. (G) Bar graph showing the average size (mm²) of Isolectinb4⁺ lesions in different groups. Mean ± SEM, n = 34–39 lesions per group. *P < 0.05, Mann-Whitney test.

The Effect of CCR2⁺ Macrophage Depletion in Subretinal Fibrosis

To understand the role of infiltrating CCR2⁺ macrophages subretinal fibrosis, we depleted CCR2⁺ cells using a depleting antibody, MC-21 (Mack et al., 2005), either during the acute stage (from D1 to D5) or the chronic stage (from D5 to D9) of inflammation in the two-stage laser-induced model (Fig. 3A). Flow cytometry showed that the number of circulating CCR2⁺ monocytes reduced by 50 ∼ 65% after MC-21 treatment (Figs. 3B, 3C).

Figure 3.

The effect of CCR2⁺ macrophage depletion in subretinal fibrosis. (A) Diagram showing the study design of CCR2 depletion in subretinal fibrosis. CCR2⁺ monocytes were depleted with MC-21 antibody (20 µg/mouse, daily intraperitoneal [IP] injection) for six days (acute stage [AS]) or five days (chronic stage [CS]). Rat IgG was used as a control. Blood was collected to validate CCR2⁺ cell depletion after MC-21 treatment using flow cytometry. Subretinal fibrosis was induced in C57BL/6J wild-type mice using the two-stage laser protocol. Eyes were collected at D10 after the second laser. (B) Bar graph showing the population of CCR2⁺CD11b⁺ monocytes in CD45⁺ leukocytes at different times after the first MC-21 injection (D0) in the acute stage group assessed by flow cytometry. n = 5 mice. (C) Bar graph showing the population of CCR2⁺CD11b⁺ monocytes in CD45⁺ leukocytes on different days after the first MC-21 injection (D0) in the chronic stage group by flow cytometry. n = 5 mice. (D) Representative images showing collagen-1⁺ (red) fibrotic lesions in different groups at D10. Scale bar: 100 µm. (E) Bar graph showing collagen-1⁺ fibrotic lesion size (mm²) in different groups. n = 32–38 lesions per group. (F) Representative images showing Isolectin B4⁺ (green) angiogenesis lesion in different groups. Scale bar: 100 µm. (G) Bar graph showing the Isolectin B4⁺ lesion size (mm²) in different groups at D10. n = 32–38 lesions per group. All data were shown as mean ± SEM. *P < 0.05; **P < 0.01 with Kruskal-Wallis test.

Depleting CCR2⁺ macrophages during the chronic stage, but not the acute stage, significantly reduced the size of collagen 1⁺ fibrotic lesions (Figs. 3D, 3E) and isolectin B4⁺ vascular lesions (Figs. 3F, 3G). The results suggest a critical role of infiltrating CCR2⁺ macrophages in subretinal fibrosis at the chronic stage of inflammation.

Immune-Related Gene Expression in Different Stages of Subretinal Fibrosis

The phenotype and function of infiltrating macrophages and microglia are regulated by microenvironmental cues of the subretinal space. To understand the dynamic change of the microenvironment, we investigated the immune-related gene expression at different stages of subretinal fibrosis. Interestingly, we did not detect significant changes at D2 and D5 after the second laser in inflammatory and anti-inflammatory genes including iNos, Tnfa, Ccl2, Yam1, Il-10 and Cd206 (Fig. 4). The expression of iNos, Ccl2, and Yam1 was significantly increased on D10 after the second laser (Fig. 4). Surprisingly, we found that the expression of Vegfa expression was significantly reduced at D5 and D10 post-second laser treatment (Fig. 4). The coexpression of high levels of Ccl2, Inos and Ym-1 at the late stages of subretinal fibrosis highlights the complexity of the subretinal microenvironment.

Figure 4.

The dynamics of pro-/anti-inflammatory gene expression in subretinal lesions. Subretinal fibrosis was induced in C57BL/6J mice using the two-stage laser protocol and eyes were enucleated at D0, D2, D5, and D10 after the second laser. Total RNA from four pooled RPE/choroidal complexes were used for RT-qPCR. Mean ± SEM, n = 20-28 RPE/Choroid/Sclera. *P < 0.05; ****P < 0.0001 with Kruskal-Wallis test.

Macrophages From the Chronic Stages of the Disease are Pro-Fibrotic

MMT is known to contribute to organ fibrosis and has more recently been linked to subretinal fibrosis progression.¹⁸ Bone marrow cells were collected from the lasered mice on different days (D0, D2, D5, and D10) after the second laser and differentiated into BMDMs. We found that the expression of profibrotic markers Acta2, Col1a1, and Fn in BMDMs from D5 and D10 was significantly higher than those from D0 and D2 after the second laser (Fig. 5A). Immunostaining of α-SMA and collagen-1 showed that BMDMs from D0 and D2 mice did not express α-SMA while few cells expressed collagen-1 (Fig. 5B). In contrast, many α-SMA⁺Col-1⁺ cells were detected in BMDMs from D5 and D10 mice (Fig. 5B). The results suggest that macrophages from the chronic stages of the disease are pro-fibrotic and have a higher potential to undergo myofibroblast trans-differentiation.

Figure 5.

Macrophage phenotype in different stages of subretinal fibrosis. Subretinal fibrosis was induced in C57BL/6J mice using the two-stage laser protocol. Bone marrow cells from the lasered mice at different days (D0, D2, D5, and D10) after the second laser were differentiated into macrophages (BMDMs). The cells were processed for qPCR (A) or immunocytochemistry (B). (A) Gene expression levels of the fibrotic markers Acta2, Col1a1, and Fn of BMDMs from different days after the second laser treatment. Mean ± SEM, n = 3. *P < 0.05; **P < 0.01, ***P < 0.001 with Kruskal-Wallis test. (B) Representative immunofluorescence images of BMDMs stained with collagen-1 (green), αSMA (red), and DAPI (blue). Scale bar: 100 µm.

Microglia Do Not Undergo Myofibroblast Differentiation on TGFβ Stimulation

TGFβ is the master regulator of fibrosis through inducing EMT, EndoMT, and MMT. To further understand if microglia can transdifferentiate into myofibroblast in retinal fibrosis, we investigate the response of microglia (BV2 cells) to TGFβ1 treatment. Four days of TGFβ1 (10 ng/mL) treatment reduced the number of loosely attached round-shape cells and increased the number of flattened adherent cells (Fig. 6A). qPCR analysis showed that the expression of TREM119, Cx3cr1, and Itgam was significantly increased, and the expression of Trem2, Aif1 (Iba-1), Emr1 (F4/80) did not change (Fig. 6B), suggesting a shift to homeostatic microglial phenotype.

Figure 6.

The effect of TGFβ in microglial cells. Murine microglial cell line BV2 cells were treated with TGFβ1 (10 ng/mL) for 4 days, and the cells were collected for real-time qPCR analysis of different genes. (A) Phase-contrast images showing the morphology of BV2 cells in control and TGFβ1-treated groups. (B) The expression of microglial marker genes in control and TGFβ1-treated BV2 cells. (C) The expression of myofibroblast-related genes Acta2, Col1a1, and Fn in control and TGFβ1-treated BV2 cells. #, undetectable mRNA levels. (D) The expression of inflammation-related genes in control and TGFβ1-treated BV2 cells. Data in B-D were expressed as mean ± SD, n = 3. *P < 0.05 unpaired Student t-test.

The myofibroblast-related genes including Acta2 and Col1a1 were detected at low levels (Cq value > 33), and Fn was not detected in BV2 microglia (Cq > 40). TGFβ treatment did not affect their expression (Fig. 6C). The results suggest that microglia-to-myofibroblast transmission. Interestingly, TGFβ treatment increased the expression of iNos, Il1b, Ccl2, Tnfa, Arg-1, and Vegfa but decreased the expression of Cd206 (Fig. 6D), suggesting that TGFβ may regulate microglial activation and cytokine production.

Discussion

In this study, we investigated the dynamic progression of subretinal fibrosis in our two-stage laser subretinal fibrosis model to better understand the disease mechanism, particularly, the role of different types of immune cells in different stages of subretinal fibrosis. We found that the fibrotic lesion increased rapidly in the initial three days, and the size remained stable afterwards. Therefore we define the first three days as the acute stage (or fibrosis initiation stage) and from day 3 onward as the chronic stage (fibrotic lesion maturation stage, Fig. 7). We further found that inflammation in the acute stage was dominated by infiltrating CX3CR1⁺ cells. When the disease moved to the chronic stages (D6-10), many more cells (both CCR2⁺ cells and CX3CR1⁺ cells) infiltrated the lesion (Fig. 7). Interestingly, the CCR2⁺ macrophages were located within the fibrotic lesion, whereas the CX3CR1⁺ microglia were distributed around the lesion at the chronic stages. Surprisingly, depletion of microglia with PLX5622 did not affect collagen-1⁺ fibrotic lesion but reduced the Isolectin B4⁺ vascular lesion, whereas depletion of CCR2⁺ cells in the chronic stage, not the acute stage, reduced subretinal fibrosis and angiogenesis. Our results suggest that CCR2⁺ macrophages not microglia are critically involved in the maturation of subretinal fibrosis.

Figure 7.

The dynamic change of subretinal fibrosis and immune cell infiltration. After the second laser treatment, the initial three days are the subretinal fibrosis initiation stage. Infiltrating immune cells are dominated by CX3CR1 cells. The lesion size remains stable afterwards. Significant immune cell infiltration begins from day 3.

In our model, fibrosis arises from the pre-existing laser-induced CNV, mimicking closely macular subretinal fibrosis development in nAMD patients. The classical laser-induced CNV would be absorbed within a couple of weeks as part of the wound-healing response but was prevented by the second laser treatment.⁶ The second laser burn induces severe, persistent inflammation,^6,27 which converts the CNV into the fibro-vascular membrane. Inflammation in the fibrosis initiation stage is dominated by the CX3CR1⁺ cells followed by infiltrating CCR2 macrophages at D6-D10. Surprisingly, depleting microglia with PLX5622 did not affect subretinal fibrosis. There are three possible reasons to explain this. First, the majority of the CX3CR1⁺ cells at the center of the lesion in the acute stage may not be microglia. Although CX3CR1 is highly expressed in microglia, it is also expressed in other immune cells including a subset of monocytes, dendritic cells, and NK cells²⁸ and the CX3CR1-expressing cells are known to play an important role in tissue homeostasis.^28,29 These immune cells will be less affected by PLX5622 treatment. Second, in this model, microglia depletion was conducted in mice with laser-induced CNV starting three days before the second laser burn (i.e., four days after the first laser burn). Because of laser-induced retinal inflammation, microglia are activated and undergo proliferation, and it is difficult to deplete the proliferating microglia. Indeed, PLX5622 treatment only reduced retinal microglia by 50% (Fig. 2). This may partially explain the lack of effect on subretinal fibrosis. Third, the infiltrating microglia may have dural roles in subretinal fibrosis. Although infiltrating microglia at the acute stage may enhance inflammation and initiate fibrosis development, their presence at the chronic stage may clear debris, and reduce inflammation and fibrosis. A recent study has shown that microglia depletion or repopulation does not affect light-induced retinal degeneration,³⁰ highlighting the complexity of microglia in retinal health and diseases. Nonetheless, the treatment significantly reduced subretinal angiogenesis, suggesting that microglia are critically involved in retinal angiogenesis and their role in retinal fibrosis is limited.

The role of macrophages in organ fibrosis including retinal fibrosis is well-documented.^31–33 It has been proposed that infiltrating macrophages, not the tissue-resident macrophages are the main subsets involved in fibrosis progression.³³ We have shown that infiltrating macrophages can promote subretinal fibrosis through MMT,¹⁸ a process involving macrophage elastase (MMP-12).²⁷ However, whether macrophages are involved in the induction or maturation of subretinal fibrosis is unknown. A previous study in a carbon tetrachloride (CCl₄) injection-induced liver inflammation and fibrosis model showed that depleting macrophages during the early recruitment phase was beneficial and protective, whereas depleting macrophages at a later stage was detrimental.³⁴ Based on our observation of the disease course, we depleted CCR2⁺ cells either at the fibrosis initiation or maturation stage. Our results suggest that CCR2⁺ monocytes play a critical role in subretinal fibrosis maturation. Mechanistically, we found that bone marrow-derived macrophages from D6∼D10 mice had a pro-fibrotic phenotype and a higher potency to undergo MMT. Microglia are unable to transdifferentiate into myofibroblast even with TGF-β1 treatment. The mechanism by which the retinal laser burn induces phenotype change of bone marrow-derived macrophages remains elusive. When the retina is damaged, it releases inflammatory cytokines and chemokines that may reach the bone marrow through blood circulation. This retina-to-bone marrow axis of wound healing response was reported in our previous study,³⁵ although the molecular pathways remain to be elucidated.

In summary, we show in this study that CNV-mediated subretinal fibrosis follows a two-stage disease course, the fibrosis initiation stage and the maturation stage. Infiltrating CCR2⁺ macrophages are critically involved in fibrosis maturation and depleting CCR2 macrophages at this stage can reduce subretinal fibrosis. Microglia may contribute to retinal angiogenesis and their role in subretinal fibrosis is limited. Further studies will be needed to investigate the role of the CX3CR1⁺CCR2⁻ cells in converting CNV to fibrovascular membrane.