Laser photocoagulation induces focal RPE/BM injury, cell death and DNA damage

Laser-induced CNV model is an accelerated model resembling in part pathogenic processes for neovascular AMD, by which laser photocoagulation ruptures RPE/BM in animals, mimics acute RPE dysfunction and cell loss found in human ageing eyes [22,23]. We wished to initially determine the extent of damage to RPE that may elicit microglia or myeloid cell responses to drive pathological angiogenesis. Immunostaining for the tight junction protein, Claudin, on whole-mounts of normal RPE/choroid showed typical pentagonal or hexagonal RPE cell morphology (Figure 1A). The breakdown of RPE (arrow, Figure 1B(i)) was observed after laser burn and an autofluorescent circular injury of BM [24] was also revealed (arrowhead). Confocal reflection microscopy demonstrates tissue fragments and cellular debris within the lesion (Figure 1B(ii)). In addition, TUNEL⁺ cells with damaged DNA were observed close to the lesion site immediately after laser (Figure 1B(iii)), whilst normal RPE/choroid from age-matched mice (6 to 8-week old) contained no apoptotic cells by TUNEL staining. There were no TUNEL⁺ cells found 2-7 days post laser (data not shown), demonstrating that laser induced cell apoptosis does not persist.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Early recruitment of retina- and blood-derived macrophages to site of RPE/BM injury.

(A) Normal RPE shows typical pentagonal or hexagonal cell morphology examined on RPE/choroid whole-mounts using Claudin immunostaining and confocal microscopy. (B) Laser induction results in (i) disruption of RPE (arrow) and BM (arrowhead); (ii) tissue fragmentation; and (iii) TUNEL-positive cells with DNA damage contiguous as observed with confocal microscopic analysis of RPE on RPE/choroidal whole-mounts. (C) Flow cytometric analysis of pooled samples from three eyes shows CD45⁺F4/80⁺ cell infiltration in RPE/choroid and retina between 1–14 days post laser (mean ± SEM, n=2-4 for each time point). (D) Iba1 immunohistochemistry analysis shows local recruitment and morphological change of Iba1⁺ microglia/macrophages at site of injury at 6 hours (i), day 1 (ii), day 2 (iii), day 4 (iv), day 7 (v) and day 14 (vi) after laser treatment. Small image inset in (D, ii) shows ramified microglia in normal retina. (E) Top and side view of Z-stacks of confocal images of retina whole-mounts collected from CX3CR1^gfp/+ mice immediately after laser induction or 2 days after. (F) CCR2 immunohistochemistry of RPE/choroid or retina from CX3CR1^gfp/+ mice demonstrates accumulation of CX3CR1 ^hiCCR2^- (microglia-derived, arrow) and CX3CR1 ^loCCR2⁺ (monocyte-derived, arrowhead) macrophages at lesion site on both RPE and outer retina aspects. (G) QRT-PCR analysis of laser-induced Ccr2 mRNA expression in RPE/choroid and retina tissues over time. mRNA levels were normalised against a house keeping gene Gapdh and data are presented as fold change relative to control (mean ± SEM, n=3-6 for each time point). (H) Local Ccl2 and Ccr2 upregulation at the RPE/choroid lesions (day 2). ^*P<0.05 vs. control; ^**P<0.01 vs. control; ^# P<0.05 between groups.

https://doi.org/10.1371/journal.pone.0072935.g001

Retinal microglia and blood monocytes rapidly accumulate and are activated at lesion sites prior to onset of angiogenesis

Given the rapid RPE damage and localised disruption in outer retinal barrier observed above, we wished to assess the temporal relationship between the formation of neovascular membrane and myeloid cell recruitment. Angiogenesis and neovascular membrane formation was assessed via immunostaining RPE/choroid whole-mounts with IB4 (Figure S1), which demonstrates that onset of choroidal angiogenesis begins on day 4 post laser, while formation of CNV membrane is established by day 7 and further developed between 7–14 days, prior to involution and consistent with previous reports [25]. The kinetics of macrophage recruitment to both RPE/choroid and retina was assessed via quantitative flow cytometric analysis of cell infiltrate. Overall, there is a significant increase of CD45⁺F4/80⁺ cells in RPE/choroidal tissue between day 1 and 7 as compared to control tissue (Figure 1C). The peak of infiltrate occurs on day 2 (490% increase vs. control) but cell numbers return to basal levels by day 14. In retina, elevated CD45⁺F4/80⁺ cell numbers were also observed after 2-4 days, with the maximal infiltrate on day 4 (120% increase) which is both later and less pronounced than the change in RPE/choroid. These results demonstrate that accumulation of macrophages to the site of injury is transient and early following initial laser damage and importantly occurs prior to the onset of choroidal angiogenesis.

Macrophage morphology indicates cell activation and differentiation [26]. To further examine local macrophage recruitment and activation, RPE/choroidal tissue was collected at different time points post laser and immunostained with Iba1, a marker for both microglia and monocytes (Figure 1D). Iba1⁺ staining confirms the transient pattern of cell accumulation between day 1 and 14, and demonstrates that these cells are localised within or adjacent to the site of RPE/BM injury. Furthermore, the recruited Iba1⁺ cells display different morphology. At day 1 post-laser (Figure 1D(ii)), ramified cells with small soma and long processes (white arrow) similar in appearance to retinal microglia (inset), as well as relatively round (yellow arrow) or amoeboid Iba1⁺ cells with enlarged cell body and shortened dendrites (red arrowhead) are all present. However, after 2 days (Figure 1D(iii)), fewer ramified microglia at lesions and a greater number of amoeboid cells are detected, which infers microglial activation [26].

As there is evidence of both activated microglia as well as potential infiltrating cells, we wished to confirm the involvement of microglia in the CNV progression. We first took advantage of the CX3CR1^gfp/+ transgenic mouse in which resident microglia are fluorescently labelled. Confocal images (top and side views of Z-stacks) of retinas from CX3CR1^gfp/+ mice were examined 2 days post laser, and clearly demonstrate migration of microglia from the inner retina toward the lesion site following laser application (Figure 1E). Secondly, utilising the distinct chemokine receptor expression differences between microglia and inflammatory monocytes, CCR2^-CX3CR1^hi for the former and CCR2 ⁺ CX3CR1^lo for the latter [27,28], we noted that in RPE/choroidal and retinal whole-mounts from CX3CR1^gfp/+ mice day 4 post laser both CCR2^- microglia-derived (arrow) and CCR2⁺ monocyte-derived (arrowhead) cells were present (Figure 1F). In support of CCR2⁺ cell infiltration we documented an increase in Ccr2 mRNA in RPE/choroid (between day 1 and 4) and retina (day 4) (Figure 1G). Microglia do not express CCR2 [28], therefore the temporally enhanced Ccr2 mRNA expression is likely due to the accumulation of blood-derived CCR2⁺ monocytes in response to CCL2 [25]. Furthermore, dissection of CNV lesions from whole RPE/choroid tissue on day 2 reveals greater increases in both Ccl2 and Ccr2 mRNA levels compared to peripheral RPE/choroid tissue (Figure 1H), supporting a focal chemokine response.

Early and lesional-restricted inflammatory gene expression demonstrates a predominant Arg-1 expressing myeloid cell infiltrate

Given the data above and the recognised immune activation and inflammation associated with development of angiogenesis related to AMD [3,8], we sought to characterize the phenotype of the accumulating macrophages and explore how this contributes to CNV development. Initially we examined mRNA expression to test whether a segregated M1 or M2 phenotype was present in the whole RPE/choroid and retinal tissues between 3 hours and 14 days post laser. In whole RPE/choroid, Nos2 mRNA remained largely at basal level (Figure S2A). The expression of other M1 pro-inflammatory genes including Il1β and Tnfα were only slightly upregulated (<3 fold) in RPE/choroid during the early phase of CNV development (up to 7 days, Figure S2A) and we did not find significant changes of expression for Il18, Il4 and Il10 (data not shown). Notably, an M2 signature, Arg-1 mRNA expression was increased after 6 hours, and further increased between day 1 and 4 (70-200 fold), and then returned to basal levels during the period 7-14 days (Figure 2A). Expression of another M2 gene Ym1 demonstrated a similar pattern as Arg-1, although the induction fold was less (<8 fold, Figure 2A). Within retina, only Tnfα, showed any significant increase in gene expression during the early phase of CNV development (Figure S2A). In addition, mRNA level for Cd200 receptor (Cd200r), a myeloid-specific inhibitory receptor [29], was increased after 4-7 days in RPE/choroidal or retinal tissue (Figure S2A), which is generally later than other genes we tested.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Early and local inflammation-associated gene expression is mainly from accumulating macrophages.

(A) QRT-PCR analysis of time-dependent Arg-1 and Ym1 mRNA expression in RPE/choroid and retina tissues collected at indicated time points post laser induction. (B) RPE/choroid-lesional gene expression of Nos2, Tnfα, Il1β, Arg-1 and Ym1 from isolated lesions at day 2 confirms localised inflammatory gene signature. (C) Cellular gene expression of Arg-1 and Ym1 on day 2 using CD11b MACS-isolated cells pooled from 4 RPE/choroidal or retinal tissues demonstrates early and local Arg-1 and Ym1 expression largely produced by myeloid population. Gapdh served as a normalising control in (A–C). Data are presented as mean ± SEM, n=3-6 for each time point. ^*P<0.05 vs. control; ^**P<0.01 vs. control; ^# P<0.05 vs. non-lesion tissue on laser-treated eyes. ND, not detected.

https://doi.org/10.1371/journal.pone.0072935.g002

When lesions were isolated from the whole RPE/choroid on day 2 post laser the signature of Arg-1 was enhanced (Figure 2B) and to a lesser extent Tnfα, Il1β, and Ym1 transcripts were increased. To determine if the Arg-1 signature resided in infiltrating myeloid cells, we isolated CD11b⁺ cells from pooled single-cell suspensions of RPE/choroid or retina. In normal retina, there was negligible signal in either CD11b^- or CD11b⁺ (microglia) cells (Figure 2C). Conversely two days post laser, mRNA expression for Arg-1, Ym1, Tnfα and Il1β was largely restricted to CD11b⁺ cells from RPE/choroid (Figures 2C and S2B). The expression of Arg-1 by early accumulating myeloid cells was further confirmed by conventional RT-PCR (Figure S3A) and immunostaining showing cytoplasmic expression of Arg-1 in lesional CD11b⁺ cells on RPE/choroid whole-mounts collected on day 3 (Figure S3B).

Accumulating microglia/myeloid cells supply the earliest enhancement of VEGF expression within the lesion

Whilst we have described a specific macrophage phenotype prior to angiogenesis and although VEGF expression by macrophages [30–32], as well as other cells including RPE and neovascular endothelial cells [30,33] are described in experimental CNV, the chronological order of VEGF expression in these different cell types during CNV development or what initiates the drive toward pathological angiogenesis following injury is unknown.

In this study, a transient VEGF expression by accumulating macrophages during early stages (days 1-4) of the laser-CNV model was detected using immunostaining and confocal microscopy (Figure S4), as described in primates [30]. Within the normal tissue, resting microglia distributed sparsely in the sub-retinal space have no significant VEGF immuno-reactivity (Figure S4A). Six hours after laser induction, no significant Iba1⁺ cells and VEGF upregulation were observed within the lesion that contains abundant damaged cells and debris. By day 1 the majority of VEGF immuno-reactivity was clearly co-localised to recruited microglia or myeloid cells (Figure 3B). This is further confirmed when analysing the fluorescence plot profiles which shows that the VEGF signal peaks in Iba1⁺ cells and only low levels in RPE adjacent to the lesion on day 1 (Figure S5), the time when myeloid cells accumulate. This indicates that the earliest increase in VEGF within the lesion is produced by recruited microglia/myeloid cells. By day 2, the majority of lesional Iba1⁺ cells were co-localised with enhanced VEGF expression (Figure 3C). At this time point increased VEGF expression by the surrounding non-myeloid cells was also observed. VEGF expression in Iba1⁺ cells was maintained until day 4 (Figure 3D). Between day 7 and 14 post laser, lesional VEGF immuno-reactivity subsided, coinciding with the reduced number of VEGF⁺ macrophages (Figure 3E,F).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Lesional VEGF expression in laser CNV model.

Iba1 and VEGF dual-staining and confocal microscopy on RPE/choroid whole-mounts display transient VEGF expression at lesion site. (A) 6 hours post laser, no increase of Iba1⁺ cells or VEGF immuno-reactivity was found at laser lesions. (B) On day 1, recruited ramified microglia and amoeboid macrophages express VEGF (also see Figures S4 and S5). By day 2 (C) and 4 (D), greater VEGF immuno-reactivity are found in accumulating macrophages, as well as in Iba1^- cells. Between day 7 (E) and 14 (F), lesional VEGF expression lessened, in parallel with reduced macrophage number. Images are representatives of 12-24 lesions for each time point.

https://doi.org/10.1371/journal.pone.0072935.g003

Recruited myeloid cells are largely non-proliferating and their accumulation precedes recruitment of proliferating non-myeloid cells

To interrogate whether in situ myeloid cell expansion at the lesion site contributes to increased number of myeloid cells or to endothelial proliferation we co-immunostained Iba1 with Ki67. As shown in Figure 4A, neither myeloid nor proliferating cells were detected at lesions 6 hours post laser. On day 1 (Figure 4B), accumulating microglia/monocytes showed no proliferative capacity and there were no Ki67⁺ cells observed. A number of Ki67⁺ cells were seen at lesions on days 2 and 4, although largely not Iba1-positive (Figure 4C,D). Between day 7 and 14, Ki67⁺ cell expression subsided (Figure 4E,F). The data demonstrate that cell proliferation does not contribute to myeloid cell numbers observed and the accumulation of lesional myeloid cells occurs prior to notable non-myeloid proliferation or morphological evidence of angiogenesis. We did not pursue whether Ki67⁺ cells were endothelial or endothelial progenitors [33].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Microglia/monocyte accumulation precedes appearance of non-myeloid cell proliferation.

Iba1 and Ki67 dual-staining on RPE/choroid whole-mounts during the time course of CNV development. (A) Six hours post laser induction, neither macrophages nor cell proliferation are detected at lesions. (B) On day 1, accumulating Iba1⁺ cells are Ki67-negative and no proliferating cells are observed. By day 2 (C) and 4 (D), majority of macrophages are not co-localised with Ki67 immuno-reactivity, while abundant Iba1^-Ki67⁺ cells are seen at lesions. Between day 7 (E) and 14 (F), both Iba1⁺ and Ki67⁺ cell numbers subside. Images are representative of at least 12 lesions for each time point. Smaller images display separate stains for Iba1 and Ki67, respectively; bigger images are two channels merged.

https://doi.org/10.1371/journal.pone.0072935.g004

Microglia recruitment and VEGF expression persist after depletion of CCR2⁺ monocyte infiltrate

Our data thus far demonstrate that laser-induced injury at RPE/BM causes rapid macrophage recruitment from retina (CCR2^- microglia) and blood (CCR2⁺ monocytes), leading to early and local Arg-1 and VEGF expression, prior to formation of neovascular membranes. To further determine the role and extent of blood monocytes, we depleted CCR2⁺ cells with an anti-CCR2 (MC-21) mAb, which selectively depletes Ly6C ^hiCCR2⁺ monocytes from peripheral blood [34–36]. Following a daily i.p. injection regimen of MC-21, blood Ly6C ^hiCCR2⁺ monocytes were depleted by 92% on day 2 post laser (Figure 5A). Depletion resulted in suppression of Ccr2 mRNA induction (Figure 5B), loss of CCR2 immuno-positive cells (Figure S6), and reduced Arg-1 mRNA expression by 90% (Figure 5C) in RPE/choroid. However, migration of microglia to the lesions was not affected by CCR2⁺ monocyte depletion (Figure 5D). Loss of Ccr2 and Arg-1 signature also corresponded to reduced Iba1 immuno-reactivity at the lesion site (Figure 5E,F). The 40% reduction observed represents the percentage of infiltrating monocytes among the total number of myeloid cell infiltrate. Similarly lesional VEGF immuno-reactivity was downregulated by MC-21-mediated CCR2 monocyte depletion in laser-treated animals (Figure 5E,G). However, microglia at site of injury in CCR2-depleted mice remained VEGF expressing (Figure 5E and inset). Although microglia recruited to the lesion site remain functional in generating VEGF, MC-21 administration suppressed angiogenesis by 39% on day 7 post laser induction (Figure 5H,I), confirming that depletion of CCR2⁺ monocyte infiltration suppresses angiogenic development.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Systemic depletion of CCR2⁺ monocytes abolishes Arg-1 signature without altering microglia migration, and partly suppresses laser-induced CNV formation.

(A) Flow cytometric analysis of peripheral blood from mice following depletion of Ly6C ^hiCCR2⁺ monocyte subset using the MC-21 anti-CCR2 mAb, compared with rat IgG2b isotype control (mean ± SEM, n=3). Assessment of Ccr2 (B) or Arg-1 (C) mRNA expression normalised against Gapdh in RPE/choroid day 2 post laser in MC-21 treated mice by QRT-PCR (mean ± SEM, n=4-6). (D) Side view of Z-stacks of confocal images shows migration of retinal microglia (Iba1 staining) in MC-21 treated animals. Confocal images (E) and semi-quantitative mean fluorescence intensity (MFI) analysis using ImageJ software for Iba1⁺ cell accumulation (F) and whole lesional VEGF expression (G) on RPE/choroid whole-mounts taken on day 2 post laser (mean ± SEM, n≥12/group). Magnification image in (E, inset) shows macrophages co-stained for VEGF in MC-21 injected mice. (H and I) CNV formation on day 7 post laser was assessed by IB4 staining on RPE/choroidal whole-mounts and images were captured by confocal microscopy and analysed using a Volocity® 3D Image analysis software. Representative 3D reconstructions of neovascular complex (H) and quantitative data (I) show reduced CNV volume by 39% with MC-21 treatment compared with control (n=24/group). ^*P<0.05; ^**P<0.01; ^***P<0.001.

https://doi.org/10.1371/journal.pone.0072935.g005

Myeloid cell uptake of damaged RPE or its derivatives conditions toward Arg-1 phenotype and VEGF production

Iba1 immunostaining demonstrates macrophage surface ruffling on cells within the lesion (Figure S7A, arrow), a membrane change associated with cell phagocytic activity [37]. Furthermore, after laser application, fragmented RPE pigments were observed within the lesion and on both RPE/choroid and outer retina aspects. Despite abundance of RPE pigments we could elucidate pigment-engulfing macrophages at the lesion site from the retina side (Figure S7B), where brightfield images were colour-processed from black(pigment)/white(retina) to green(pigment)/black(retina) and then merged with Iba-1 staining (red). A pHrodo Red-Dextran endocytosis assay based on ex vivo retina culture also confirmed endocytic activity of lesional macrophages (Figure S7C). Based on this data, and to test whether uptake of damaged RPE/derivatives by lesional macrophages induced the phenotype and function described above, we adapted an in vitro phagocytosis assay in which heat-induced necrotic mouse RPEs (B6-RPE07) were added to bone marrow-derived macrophages (BMMΦs). Engulfment of dead RPEs and debris by macrophages was observed after 60 minutes (Figure 6A) and demonstrated dose-dependent induced BMMΦ Arg-1 mRNA expression (Figure 6B). Nos2 mRNA expression in macrophages conditioned by necrotic RPE or cell lysate is significantly less than Arg-1, which is generally consistent with our in vivo finding. In addition, macrophage VEGF expression at both transcript and protein levels were significantly enhanced by treatment with necrotic RPE or lysate (Figure 6C–E). To confirm our in vivo finding that microglia-derived macrophages also express VEGF, we analysed Vegf gene expression in a human microglia cell line (C13-NJ) treated with necrotic human RPE cells (ARPE-19). As expected, Vegf mRNA level in the human microglia cells was enhanced by addition of dead RPE cells (Figure 6F).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Macrophage uptake of necrotic RPE or its derivatives altering their inflammatory and proangiogenic phenotype.

(A) Necrotic RPE was produced from CFDA-labelled B6-RPE07 cells (green) by heating at 95°C for 15 minutes, while RPE lysate was generated by sonication for 5 minutes on ice. Necrotic RPE cells or lysate were then added to Violet Tracer-labelled BMMΦs at different ratios (RPE to BMMΦ). After 1 hour incubation, the cells were washed, fixed and observed by confocal microscopy. Top and side view of confocal images shows engulfment of dead RPEs/debris by BMMΦs. BMMΦs were collected after 24 hours of incubation with necrotic RPE or cell lysate and RNA was extracted for QRT-PCR analysis to determine Arg-1 and Nos2 (B), and Vegf gene expression (C). 18s rRNA was used as a normalising control. (D) Confocal images and MFI analysis of VEGF immunocytochemistry on BMMΦs treated with necrotic RPE. (E) In some of experiments, after 1 hour of incubation of BMMΦs with necrotic RPE, un-engulfed dead RPE cells and its derivatives were removed and BMMΦs were further cultured for 24 hours and cell culture supernatants collected for determination of VEGF concentration using ELISA. (F) Human microglia cells (C13-NJ cell line) were treated with heat-induced necrotic ARPE19 cells at different ratios for 24 hours and then examined for Vegf mRNA expression. Data are presented as mean ± SEM, n≥3. ^*P<0.05; ^** P<0.01; ^*** P<0.001 Arg-1 vs. Nos2 (B), or vs. control. NS, not significant.

https://doi.org/10.1371/journal.pone.0072935.g006

We then tested whether oxidative stress-induced RPE apoptosis could result in similar macrophage responses as necrotic RPE. H₂O₂ treatment of B6-RPE07 cells induces apoptosis as analysed by annexin V/7AAD staining (Figure S8A), and the in vitro phagocytosis assay demonstrates engulfment of apoptotic RPE/debris by BMMΦs within 60 minutes (Figure S8B). When BMMΦs were subsequently isolated using CD11b microbeads from the phagocytic assay co-culture after 24 hours of incubation (>95% purity analysed by flow cytometry), apoptotic RPEs induced a dose-dependent upregulation of macrophage Arg-1 as well as increase in Nos2 mRNA from BMMΦs (Figure S8C), and enhanced Vegf expression (Figure S8D).

Employing other necrotic cell sources including retinal cells (excluding RPE), retinal endothelial cells, B-cell hybridoma (YTS156) and bone marrow neutrophils did not confer the robust Arg-1 phenotype in BMMΦs we observed with necrotic RPE feeding (Figure 7). For example, whilst uptake of necrotic retinal endothelial cells promoted macrophage Nos2 expression there was no upregulation of Vegf or Arg-1 (Figure 7A–D). Furthermore, co-culture with necrotic BMMΦs did not elicit a macrophage response. The in vitro data support the contention that early macrophage Arg-1 phenotype and proangiogenic response (VEGF expression and secretion) is induced by damaged RPE.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. Gene expression of BMMΦs in response to different necrotic cell types.

Mice BMMΦs were treated with necrotic cells generated from B6-RPE07 cells (RPE), whole retinal tissues (Ret), retinal vascular endothelial cells (EC), YTS156 hybridoma (B cell) or bone marrow neutrophils for 24 hours. Cellular RNA was isolated and gene expression of Arg-1 (A), Vegf (B), Nos2 (C) and Il1β (D) was assessed by QRT-PCR (mean ± SEM, n≥3). 18s rRNA was utilised as a normalising control and results presented as fold change of control.

https://doi.org/10.1371/journal.pone.0072935.g007

Mice and CNV induction

C57BL/6J mice (no Rd8 mutation detected) were obtained from the Jackson Laboratory and breeding colonies were established within the Animal Services Unit at University of Bristol, UK. Mice were kept in the animal house facilities of the University of Bristol, according to the Home Office Regulations. CX3CR1^gfp/+ mice (C57BL/6 background, Jackson Laboratory), in which one allele of Cx3cr1 gene is replaced by Egfp gene [63], were bred and maintained in the Biological Resource Unit of Queen’s University, Belfast. Treatment of the animals conformed to the Association for Research in Vision and Ophthalmology (ARVO) statement. The experiments were conducted following protocols of the Home Office Project Licence 30/2745, approved by the University of Bristol Ethical Review Group.

CNV was induced by laser photocoagulation in mice aged 6-8 weeks. In brief, the pupils of animals were dilated using topical 1% tropicamide and mice anesthetized by intraperitoneal (i.p.) injection of 200 µl of Vetelar and Rompun mixed with sterile water in the ratio 0.6:1:84. Four laser spots were delivered to the posterior retina using an OculightSlx Krypton Red Laser system (power 200 mW, duration 75 ms, spot size 75 µm), unless in macrophage enrichment experiments where 8 laser burns were induced to obtain larger cell numbers.

Antibodies

APC-conjugated rat anti-mouse CD45 (30-F11,1: 100), Alexa Fluor 488 or biotin-conjugated rat anti-mouse CD11b (M1/70, 1:100), and fluorochrome-isotype antibodies were purchased from BD Biosciences. Alexa Fluor 647-labelled rat anti-mouse/rat Ki-67 (SolA15, 1:100) was from eBioscience. Rabbit polyclonal anti-Iba1 (1:500) was obtained from Wako Chemicals. Rabbit polyclonal anti-Arg-1 (1:50) was from Santa Cruz Biotechnology. Chicken polyclonal anti-VEGF (1:100), DyLight 650-goat anti-chicken IgY (1:400) and goat polyclonal anti-CCR2 (1:100) were from Abcam. Rabbit polyclonal anti-C9 (1:500) was generated at Professor B. P. Morgan’s laboratory in Cardiff University. Rabbit polyclonal anti-Claudin (1:100), Alexa Fluor 488 or 568-conjugated goat anti-rabbit IgG, Alexa Fluor 488-conjugated rabbit anti-goat IgG (1:400), and PE-labelled rat anti-mouse F4/80 (BM8, 1:100) were obtained from Life Technologies. DL557-donkey anti-goat IgG (1:200) was from R&D Systems.

RPE/Choroidal and retinal whole-mounts and confocal microscopy

Eyes were enucleated and fixed in 2% (wt/vol) paraformaldehyde (PFA). After dissection of retina and RPE/choroid, tissues were blocked and permeabilized in 5% BSA, 5% goat or donkey (for CCR2 staining only) serum with 0.3% Triton X-100 in PBS for 2 hours, followed by incubation with primary antibodies in 1% BSA with 0.15% Triton X-100 at 4°C overnight or for 2 hours (for Arg-1 staining). After thorough washing, samples were further incubated with secondary antibodies at RT for 3 hours in the dark. Tissues were washed and flat-mounted in Vectashield antifade medium (Vector Laboratories) and examined using a Leica TCS-SP2-AOBS confocal laser scanning microscope. Isotype antibodies or negative controls with primary antibody omitted did not show significant fluorescence signal. For evaluation of neovascular membrane on RPE/choroid, fixed tissues were blocked with 1% BSA and 0.5% Triton X-100, followed by incubation with biotin-conjugated isolectin B4 (IB4, Sigma-Aldrich) in PBS containing 1% Triton X-100, 0.1mM CaCl₂ and 0.1mM MgCl₂ overnight. Samples were washed before incubation with Rhodamine Red-X-labelled streptavidin (Jackson ImmunoResearch Laboratories) for 2 hours, and whole-mounts were prepared for confocal microscopy. The volume of choroidal neovascular complex was measured using a series of Z-stack images (from the surface to the deepest focal plane) using the Volocity® Image Analysis Software 6.0. TMR Red-dUTP TUNEL (Roche) staining of DNA breaks was performed on RPE/choroidal whole-mounts according to previous reports [64].

Quantitative RT-PCR (QRT-PCR)

Total RNA from tissues or cell cultures was isolated using TRIzol reagent (Life Technologies). One µg of total RNA was treated with RQ1 RNase-free DNase before reverse-transcription using the ImProm-II^TM Reverse Transcription System (Promega). cDNA was amplified using the Power SYBR® Green PCR Master Mix Reagent (Applied Biosystems) on a StepOne™ Real-Time PCR System (Applied Biosystems). Primer sequences for mouse Gapdh, 18s rRNA, Arg-1, Tnfα and Vegf were described previously [12,65]. Primers for other mouse or human genes were designed using the PrimerQuest online tool (http://eu.idtdna.com/Scitools/Applications/Primerquest/): mouse Ccr2, forward 5’-AAT GAG AAG AAG AGG CAC AGG GCT, reverse 5’-ATG GCC TGG TCT AAG TGC TTG TCA; mouse Ym1, forward 5’-TCA CAG GTC TGG CAA TTC TTC TG, reverse 5’-TGC ATT CCA GCA AAG GCA TA; mouse Cd200r, forward 5’-TGT GGC TGG GTC AAG TTG TAC TGA, reverse 5’-AAG CAG CAG AGC AGA GCC TTT GTA; mouse Nos2, forward 5’- TGG TCC GCA AGA GAG TGC T, reverse 5’- CCT CAT TGG CCA GCT GCT T; mouse Il1β, forward 5’-GCC CAT CCT CTG TGA CTC AT, reverse 5’-AGG CCA CAG GTA TTT TGT CG; human 18s rRNA, forward 5'-CAC GGA CAG GAT TGA CAG ATT, reverse 5'-GCC AGAGTCTCGTTCGTTATC; human Vegf, forward 5'-TGG TGT CTT CAC TGG ATG TAT TT, reverse 5'-AGT CTC TCA TCT CCT CCT CTT C.

CCR2⁺ monocyte depletion

MC-21(rat anti-mouse CCR2) or isotype control (IgG2b) was administered i.p. daily at 20 µg/dose per mouse, starting one day before laser induction to day 3 post laser injury (d-1, d0, d1, d2 and d3).

Flow cytometry

Enucleated eyes were prepared under a dissection microscope, and retina and RPE/choroid were processed separately. Retinal infiltrating cells were isolated by physical disruption of dissected retinas before passing cell suspensions through a 40-µm cell strainer using a syringe plunger, washed and resuspended in staining buffer (balanced salt solution with 0.1% BSA and 0.08% sodium azide). For isolation of RPE/choroidal cells, tissue was digested in Hank’s Balance Salt Solution (PAA) containing 0.5 mg/ml collagenase D, and 750 U/ml DNase I (Sigma-Aldrich). After 20 minutes at 37^oC, an additional 0.5 mg/ml collagenase D and 750 U/ml DNase I were added. The mixture was then incubated for an additional 10 minutes at 37^oC, before being passed through a cell strainer and resuspended as for retina. Single-cell suspensions from pooled retinal or RPE/choroidal samples from three eyes were incubated with anti-CD16/32 (2.4G2) mAb for 10 minutes at 4^oC before incubation with fluorochrome-conjugated mAbs against cell surface markers including CD45 and F4/80 or isotype antibodies at 4^oC for 20 minutes. To determine MC-21 depletion efficacy, whole blood collected from the tail vein was incubated, with MC-21 or isotype control (Rat IgG2b) for 1 hour on ice. Cells were then washed three times with ice-cold PBS, before second incubation for 30 minutes on ice with biotinylated mouse anti-rat IgG2b mAb. Cells were then blocked with 10% rat serum, before incubation with streptavidin-APC, and directly conjugated mAbs against Ly6C, F4/80, CD11b and CD45 surface markers. Cell suspensions were acquired with a LSR-II flow cytometer (BD Cytometry Systems) and analysis performed using FlowJo software version 7.0 (Treestar).

CD11b⁺ cell enrichment by magnetic-activated cell sorting (MACS)

Microglia and myeloid cell populations were isolated by enriching CD11b⁺ cells from single-cell suspensions of RPE/choroid or retina tissues pooled from four eyes, using anti-CD11b microbeads (Miltenyi Biotec) according to the manufacturer’s instruction. Both CD11b^- and CD11b⁺ cell fractions were collected and prepared for RNA isolation and QRT-PCR analysis.

Generation of necrotic or apoptotic cells

Necrotic RPE cells (RPEs) were generated from a mouse RPE cell line B6-RPE07 [66] or a human RPE cell line ARPE19 (American Type Culture Collection) by heating a known number of cells for 15 minutes at 95°C, and cell death was confirmed by trypan blue exclusion. Apoptotic RPEs were induced by incubation in serum-free medium containing different concentrations of hydrogen peroxide (H₂O₂, 0.5-2 mM) for 24 hours at 37 ^oC [67]. Cell apoptosis was confirmed by microscopic observation of nuclear condensation (DAPI staining) and flow cytometric analysis using annexin V-APC/7AAD staining (BD Biosciences).

For isolation of mouse retinal vascular endothelial cells, 1 ml of pooled retinal single-cell suspension from 14 retinas was incubated with rat anti-CD31 (2 µg, BD Biosciences) in PBS with 0.5% BSA at RT for 15 minutes. After wash, samples were then incubated with Dynabeads Sheep anti-rat IgG (Life Technologies) for 20 minutes, followed by separation of the vessel-enriched fraction using a magnetic separator according to manufacturer’s instruction (Life Technologies). Neutrophils were isolated from mouse bone marrow based on negative-selection using a Neutrophil Isolation Kit (Miltenyi Biotec). Necrotic cells prepared from the retinal endothelial cells, neutrophils, B cells (YTS156 hybridoma, a gift from Dr Awen Gallimore, Cardiff University) and whole retinal cells were generated as above.

Mouse bone marrow-derived macrophages, human microglia cell line and treatment

Mouse primary bone marrow-derived macrophages (BMMΦs) were generated from C57BL/6J mice as reported in our previous studies [29]. A human microglia cell line C13-NJ was a kind gift from Prof. A. Randall (University of Bristol) and was cultured and maintained as described previously [68]. Cells were plated at a concentration of 1×10⁶ cells/well in 24-well plates and incubated for 3 hours to allow cell attachment. Cell culture supernatant was replaced with serum-free DMEM medium containing either necrotic or apoptotic cells, for an additional 24 hours incubation. After wash, total RNA from mouse BMMΦs or human microglia treated with necrotic cells was harvested in TRIzol for QRT-PCR analysis. As for analysis of BMMΦs treated with apoptotic RPEs, to avoid contamination from any remaining live RPE cells in co-culture of BMMΦs and apoptotic cells, CD11b⁺ cells were MACS isolated and purity of the macrophages was confirmed by F4/80-PE staining and flow cytometry before subsequent gene expression analysis.

In vitro phagocytosis

B6-RPE07 cells were labelled with Vybrant CFDA SE Cell Tracer (Life Technologies) according to manufacturer’s instruction, before generation of necrotic or apoptotic cell suspensions, which were then added to CellTrace Violet-labelled BMMΦs grown on cell culture slides. After 1 hour incubation, cells were washed in PBS to remove un-internalised RPEs and debris, followed by fixation with 4% PFA. Slides were mounted using antifade medium for confocal microscopy.

VEGF ELISA

To measure VEGF secretion from mouse BMMΦs treated with necrotic RPEs, we removed un-internalised necrotic RPE/debris and supernatant after 60 minutes of incubation by washing 3 times with fresh culture medium to deprive any VEGF released from dead RPEs. After a further 24 hour-incubation in serum-free medium, supernatant from the macrophage culture was harvested and stored at -80°C until analysis for VEGF concentrations by a sandwich ELISA kit (R&D Systems).

Statistics

Results are presented as means ± SEM. Data were analysed using Student’s two-tailed t test as described previously [18,19]. Differences between groups were considered significant at P<0.05.