Macrophages Are Essential for the Early Wound Healing Response and the Formation of a Fibrovascular Scar

Animals

C57BL/6 control mice and lysozyme M Cre, R26-stop^fl/fl-EYFP, Itgam-Dtr, VMD2-Cre, Stat6^−/−, and l10^−/− mice were obtained from the Jackson Laboratory (Bar Harbor, ME).^9–15 VMD2-Cre mice, which express Cre in the RPE postnatally, and lysozyme M Cre mice were crossed with R26-stop^fl/fl-EYFP mice for cell-lineage tracing experiments in RPE cells and myeloid cells, respectively, and with Vegfa^fl/fl mice to generate mice that are null for VEGF-A in the RPE and the myeloid lineage.¹⁶ Efficient Cre-mediated excision of the floxed VEGF-A allele in myeloid cells with lysozyme M Cre mice has been demonstrated.^17,18 Cre expression in RPE cells that are homozygous for the floxed VEGF-A was assessed by immunolabeling for Cre using a monoclonal mouse anti-Cre antibody (clone 2D8; EMD Millipore, Billerica, MA). The generation of Vegfa^lacZKI/WT, Vegfa^fl/fl, Flk1^lacZKI/WT, and Flt1^lacZKI/WT mice has been described previously.^19–21 To ablate macrophages in Itgam-Dtr transgenic mice or lysozyme M Cre Dtr^fl/fl mice, 20 ng/g body weight diphtheria toxin (List Biological Laboratories, Campbell, CA) was injected intraperitoneally at 1 day before laser induction and 1 day after laser injury.

Immunolabeling

For all immunolabeling experiments, eyes of at least five mice for each time point or genotype studied were used and consistent results were obtained. Representative immunolabeling images are shown. Eyes were enucleated and fixed in 4% paraformaldehyde. For choroidal flat mounts, eyes were permeabilized in 0.5% Triton X and subsequently blocked with serum in which the secondary antibodies were raised. For frozen sections, eyes were treated with 30% sucrose and subsequently embedded in optimal cutting temperature cutting compound and cut at 7-μm thickness. Tissue permeabilization was performed with 0.25% Triton X, and blocking was performed with blocking serum in which the secondary antibody was raised. Primary antibodies used were: rat anti-mouse CD31 (MEC13.3; BD Pharmingen, San Diego, CA), rabbit anti-mouse β-galactosidase (A11132; Life Technologies–Invitrogen, Carlsbad, CA), mouse monoclonal SMA-Cy3 conjugate (clone 1A4; Sigma-Aldrich, St. Louis, MO), rabbit anti-mouse NG2 (AB5320; EMD Millipore), goat anti-mouse arginase 1 (sc-18354; Santa Cruz Biotechnology, Santa Cruz, CA), rat anti-mouse F4/80 (CL8940AP; Cedarlane Laboratories, Burlington, ON), rat anti-mouse F4/80–Alexa Fluor 647 conjugate (clone C1:A3-1; BioLegend, San Diego, CA), rabbit anti–phospho-histone H3 (Ser10) (06-570; EMD Millipore), goat anti-mouse IL-1β (AF-401NA; R&D Systems, Minneapolis, MN), rabbit anti-mouse GFP (A11132; Life Technologies–Invitrogen), mouse anti-Cre (clone 2D8; EMD Millipore), and rat anti-mouse CD144 (550563, BD Pharmingen). The antibody to YM1 was generously provided by Dr. Shioko Kimura (NIH, Bethesda, MD). Secondary antibodies used were Alexa Fluor 488, Alexa Fluor 555, or Alexa Fluor 647 (Life Technologies–Invitrogen). Colabeling experiments were combined with single-labeling experiments and with experiments omitting either the primary or the secondary antibodies, to distinguish immunolabeling from autofluorescence. DAPI (Life Technologies–Invitrogen) was used for staining of nuclei. Staining for β-galactosidase (β-gal) activity was performed as described previously.²²

Microarray Analyses and Semiquantitative RT-PCR

Adult C57BL/6 control mice were used for experiments in which both eyes were exposed to laser photocoagulation, with 14 laser spots per eye. For each mouse, after removal of the retina, RPE–choroid tissue of both eyes was pooled and used for RNA isolation. Groups of mice (n = 3) were exposed to laser photocoagulation, and RNA was isolated from their RPE–choroid tissues separately at 0, 68, and 116 hours after laser injury. RPE–choroid tissues were lysed in RNeasy lysis buffer (Qiagen, Valencia, CA) using a Qiagen TissueLyser II disruption system at 30 Hz for 5 minutes. RNA isolation was performed with an RNeasy Plus kit with DNase treatment (Qiagen). Affymetrix mouse 430 2.0 microarray chips (Affymetrix, Santa Clara, CA) were used for expression profiling experiments, and analysis was performed with the associated dCHIP software version 2010.01. Data were analyzed for differentially expressed candidate genes of interest, whose significance was assessed by determining overexpression of the protein and the expression pattern in eyes of mice. Microarray experiments were performed in triplicate for each group of mice.

RPE–choroid tissue lysates of age- and gender-matched wild-type, Il10^−/−, and Stat6^−/− mice and of mice that either express or do not express diphtheria toxin receptor (DTR) in the myeloid lineage were used for semiquantitative RT-PCR experiments (n = 3 mice per group). cDNA was obtained using 100 ng total RNA with a Transcriptor first-strand cDNA synthesis kit (Roche Applied Science, Branchburg, NJ) using hexamer primers. Gene expression was quantified by semiquantitative RT-PCR using a LightCycler 480 system with the LightCycler 480 SYBR Green I master mix (Roche Applied Science, Indianapolis, IN). Primers used for mouse 36B4 as normalization control were m36B4-Forward 5′-TCACTGTGCCAGCTCAGAAC-3′ and m36B4-Reverse 5′-AATTTCAATGGTGCCTCTGG-3′. Primers used for arginase 1 (Arg1), CCL17, IL1RA (IL-1R antagonist), inducible nitric oxide synthase (iNOS), F4/80, and CD68 were as described previously.^23–26 Concentrations were determined using a standard dilution curve. Experiments for all samples were performed in triplicate, and reverse transcriptase was omitted for negative controls.

Western Blotting

RPE–choroid tissue lysates were used for Western blotting experiments. Freshly dissected posterior eye poles (RPE–choroid tissue after removal of the lens, retina, and anterior eye along the ora serrata) were lysed in NP40 lysis buffer (Life Technologies–Invitrogen) with 1 mmol/L phenylmethylsulfonyl fluoride and protease inhibitor cocktail (cOmplete; Roche Applied Science) using Qiagen TissueLyser II. After centrifugation, the supernatant was used for Western blotting. Total protein was determined by the Bradford method (Bio-Rad Laboratories, Hercules, CA), and equal amounts of protein (50 μg) were loaded onto NuPage 4% to 12% Bis-Tris gels (Life Technologies–Invitrogen) and blotted to nitrocellulose membranes. Equal protein loading was assessed using an anti-tubulin antibody. For detection of Arg1 protein levels, a goat anti-mouse Arg1 antibody (sc-18354; Santa Cruz Biotechnology) was used. Horseradish peroxidase–conjugated secondary antibodies were used, and chemiluminescence signal was determined with SuperSignal West Pico chemiluminescent substrate (Pierce; Thermo Fisher Scientific, Rockford, IL).

Experimental CNV Model

Eyes of age- and gender-matched mice were exposed to laser photocoagulation for induction of experimental CNV after eyes had been dilated with 1% tropicamide and the mice had been anesthetized with 75 mg/kg ketamine and 7.5 mg/kg xylazine. Laser photocoagulation was performed using a 532-nm laser (Visulas 532S; Carl Zeiss Meditec, Dublin, Ireland) attached to a slit lamp, and a coverslip was applied to the cornea to view the retina. Lesions were induced using a power of 200 mW, a spot size of 50 μm, and a duration of 100 ms. The lesions were located at the 2, 6, and 10 o'clock meridians centered on the optic nerve and two to three disc diameters away from the optic nerve. Laser-induced disruption of Bruch's membrane was identified by the appearance of a bubble at the site of photocoagulation. Laser spots that did not result in the formation of a bubble were excluded from the studies. Eyes were fixed and examined at various time points after laser treatment.

Quantitation of the Size of CNV Lesions

At 10 days after laser injury, the size of CNV lesions was measured in choroidal flat mounts from lysozyme M Cre Vegfa^fl/fl and lysozyme M Cre control mice of the same age and gender (n = 10 per group). The eyes were enucleated and fixed in 4% paraformaldehyde. The anterior segment, lens, and retina were removed, and the RPE–choroid tissue (posterior eye) was used for immunolabeling with anti-CD31 antibodies to detect blood vessels (Alexa Fluor 488 secondary antibody) and with SMA-Cy3 antibodies. After immunolabeling, four relaxing radial incisions were made in the RPE–choroid tissue, and the tissue was flat-mounted using ProLong Gold mounting medium (Invitrogen). Choroidal flat mounts were analyzed by epifluorescence microscopy using a Zeiss microscope (Carl Zeiss Microscopy, Jena, Germany); images were obtained with a 10× objective. CNV lesions were identified by smooth-muscle actin–positive (SMA⁺) laser spots, and neovascular area (CD31⁺ staining) per CNV lesion was determined using Zeiss AxioVision software version 4.8.2. Average CNV size was determined for each mouse, and differences between mutant and control mice were assessed by Student's t-test.

Macrophages Infiltrate the Site of Laser Injury before the Formation of a Myofibroblastic Scaffold and Subsequent Neovascularization

Laser injury to the RPE–choroid interface was induced in wild-type mice, and choroidal flat mounts were used at various time points for immunolabeling experiments. F4/80⁺ macrophages infiltrated the site of injury within the first 24 to 40 hours (Figure 1A). Subsequently, adjacent to the site of laser injury and in close contact with infiltrating macrophages, RPE cells exhibited strong expression of smooth muscle actin (SMA) at approximately 48 hours after injury (Figure 1B), a cellular response to injury and an indicator of an epithelial–mesenchymal transition (EMT)-like change. Distinct from the SMA⁺ RPE cells, stellate-appearing myofibroblastic cells with strong expression of SMA were seen in CNV lesions at approximately 60 hours after injury (Figure 1C), whereas macrophage accumulation increased to become the predominant cell type in CNV lesions at day 3 after injury (Figure 1D). At approximately day 4, endothelial cells formed a CD31⁺ neovascular network at the wound site, corresponding to a choroidal neovascularization (CNV) lesion (Figure 1, E–H). In fully formed CNV lesions, the SMA⁺ myofibroblastic scaffold covered the entire neovascular area (Figure 1, I–L), and neovascularization did not extend beyond this fibrovascular scaffold, which thus served to demarcate the area of neovascularization. Notably, SMA expression in RPE cells adjacent to the site of laser injury remained high even several months after the laser injury (Figure 1M).

Active Proliferation of Epithelial Cells, Macrophages, and Fibroblastic Cells in Evolving Fibrovascular Lesions

Next, we determined which cell populations undergo active proliferation in evolving fibrovascular lesions after laser injury. Immunolabeling for the proliferation marker phospho-histone H3 (Ser10) showed that some F4/80⁺ macrophages undergo proliferation in CNV lesions (Figure 1, N and O). Myofibroblastic stellate-like cells exhibited active proliferation with a strong signal for the proliferation marker phospho-histone H3 (Ser10) as well (Figure 1, N and P). Furthermore, some RPE cells adjacent to the injury site exhibited positive staining for this proliferation marker (Figure 1, N and Q). Although RPE cells are normally a quiescent cell population in the eye, these data show that laser injury can activate RPE cells and stimulate their proliferation. However, this RPE proliferation did not allow for reepithelialization of the injury site. Even 4 months after laser injury, the site of RPE cell ablation and Bruch's membrane disruption was not covered by RPE cells (Figure 1M). This is not surprising, given that laser injury disrupts the underlying basement membrane of RPE cells, which is required for epithelial cell migration.

Cell-Lineage Tracing Experiments after Laser-Induced Wounding

The time-course experiments showed that macrophages are the dominant cell type present in evolving CNV lesions after laser injury at day 3, whereas a neovascular network forms at day 4. Some studies have suggested that myeloid progenitor cells can transdifferentiate into endothelial cells and become part of neovessels.^4,5 To test whether this occurs in laser-induced CNV lesions, mice were generated that express an EYFP reporter in all cells derived from the myeloid lineage by crossing lysozyme M Cre mice with R26-stop^fl/fl-EYFP mice. CNV laser injury experiments in these mice showed that macrophages align in close proximity to CD144⁺ (VE-cadherin) vascular endothelial cells in formed CNV lesions, but they do not incorporate into neovessels or transdifferentiate into endothelial cells, because no colocalization of EYFP and the vascular endothelial-specific marker CD144 was observed (Figure 2, A and B). Similarly, colabeling experiments for the macrophage marker F4/80 with the vascular endothelial marker CD144 confirmed these findings (Figure 2C).

The time-course experiments showed that RPE cells adjacent to the injury site exhibit strong expression of SMA, and subsequently stellate SMA⁺ cells form a fibroblastic scaffold that eventually demarcates and covers the entire CNV lesion. To test whether these SMA⁺ cells are derived from an EMT-like process from SMA⁺ RPE cells, cell lineage-tracing experiments for RPE cells in CNV lesion were performed, by generating mice that express an EYFP reporter in cells derived from RPE cells. For this purpose, Vmd2-Cre mice, which express Cre recombinase in RPE cells, were crossed with R26-stop^fl/fl-EYFP mice, and these mice were used for CNV laser-injury experiments. EYFP expression was observed in RPE cells but not in the stellate SMA⁺ cells, indicating that these cells are not derived from an EMT-like transition from activated RPE cells (Figure 2D).

The Myofibroblastic Scaffold in CNV Lesions Consists of NG2⁺SMA⁺ Cells

The cell-lineage tracing experiments suggest that the myofibroblastic scaffold originates from cells of the underlying choroid, such as pericyte-like cells. Immunolabeling for NG2, which is commonly expressed by pericytes, showed that the fibroblastic scaffold at the center of CNV lesions consists of cells that are positive for SMA and NG2, but the adjacent RPE cells express only SMA at high levels (Figure 2, E–G). Notably, NG2⁺SMA⁺ cells were observed in CNV lesions before infiltration of the site with endothelial cells (Figure 2E). In fully formed CNV lesions, NG2⁺SMA⁺ cells were seen as part of the myofibroblastic scaffold and in a perivascular arrangement surrounding neovessels, such as would be expected for pericytes (Figure 2, G–K).

Macrophages Promote the Formation of the Fibroblastic Scaffold in Early CNV Lesions

The observation that macrophage infiltration precedes the formation of this SMA⁺NG2⁺ myofibroblastic scaffold and that RPE cells exhibit induced SMA expression in close proximity to macrophages suggests that these macrophages promote these changes in evolving fibrovascular CNV lesions. To test this hypothesis, macrophages were temporally ablated in this model, using Itgam-Dtr transgenic mice. In these mice, the diphtheria toxin receptor, DTR [alias heparin-binding EGF-like growth factor (HBEGF)], is expressed under the control of the ITGAM promoter, which is expressed in macrophages, and injection of diphtheria toxin results in reversible ablation of macrophages in these mice.¹⁵ Consistent with these observations, ablation of macrophages in these transgenic mice resulted in inhibition of SMA expression in RPE cells after laser injury and delayed formation of the SMA⁺NG2⁺ myofibroblastic scaffold (Figure 2, L–N). Similar observations were made using lysozyme M Cre Dtr^fl/fl mice instead of Itgam-Dtr transgenic mice, in which DTR is expressed in all myeloid cells, further confirming that macrophages promote formation of the myofibroblastic scaffold during the early wound healing response (data not shown). Thus, the observed early changes in CNV lesions require infiltrating myeloid cells.

Macrophages in CNV Lesions Undergo Alternative Activation Independently of IL-4/STAT6 or IL-10 Signaling

Global gene-expression profiling experiments with eyes of wild-type mice at 0, 3, and 5 days after laser injury showed that infiltrating macrophages are alternatively activated (Table 1 and Supplemental Tables S1 and S2), with strong induction of markers for alternative activation of macrophages (M2-type macrophages), including Arg1 (Arg1) and YM1, which peaked at day 3 after laser injury (Table 1). These data were confirmed by semiquantitative RT-PCR and by Western blotting of choroidal tissue lysates after laser injury (Figures 3A and 4). IL-1β expression was also significantly up-regulated at 3 days after laser injury, but decreased at day 5 (Table 1). Increased IL-1β in CNV lesions was confirmed by immunolabeling (Figure 5H). Notably, the NG2⁺SMA⁺ myofibroblastic scaffold in laser lesions stained strongly for IL-1β, whereas no strong staining was seen in wound macrophages (Figure 5H). Immunolabeling for Arg1, YM1, and F4/80 confirmed that most macrophages that accumulate at sites of laser injury are alternatively activated M2-like macrophages, whereas nonlesional choroidal F4/80⁺ cells were Arg1⁻YM1⁻ (Figure 3, B–D).

Alternative activation of macrophages has been reported in acute skin wound healing experiments, in which alternative activation occurs independent of IL-4/STAT6 signaling, which is a major pathway mediating alternative activation of macrophages, as is the IL-10 signaling pathway.²⁷ As in acute wound healing models in the skin, alternative activation of macrophages occurs in laser-induced CNV independent of IL-10 and the IL-4/STAT6 pathway; in STAT6 and IL-10 null mice, Arg1⁺YM1⁺ macrophages accumulated in CNV lesions (Figure 3, E–H). In mutant mice in which macrophages expressed DTR and that were treated with diphtheria toxin to ablate macrophages, Western blotting of laser-treated choroidal tissue lysates confirmed that the increase of Arg1 protein levels is due to increased expression in activated infiltrating macrophages in a process that does not require STAT6 or IL-10 signaling (Figure 3H).

To characterize the infiltrating macrophage population after laser injury and to quantitate the results obtained from the gene expression profiling and immunolabeling experiments, semiquantitative RT-PCR for macrophage markers was performed using RNA isolated from choroidal tissue lysates 3 days after laser injury at baseline. A significant increase in the macrophage population in choroidal tissue lysates after laser injury was observed in control and IL-10 and STAT6 null mice, with strongly increased expression of the macrophage markers F4/80 and CD68 (Figure 4, A and B). Expression of the M1-type marker iNOS was down-regulated (Figure 4C), whereas the M2-type markers Arg1, CCL17, and IL1RA were significantly up-regulated (Figure 4, D–F). Thus, immunolabeling, Western blotting, and semiquantitative RT-PCR experiments all clearly demonstrate that the infiltrating macrophages during the early wound healing response after laser injury are M2-type macrophages.

Conditional inactivation of PPAR-γ or IKK-β (NF-κB signaling) in the myeloid lineage, which have been implicated in the formation of M2-type macrophages, did not prevent alternative activation of macrophages in laser-induced CNV lesions as well (data not shown). These findings suggest that either additional signaling pathways are involved in promoting alternative activation of macrophages in response to laser injury or that compensation mechanisms among these pathways are present, and that targeting a single pathway can be compensated for by other pathways involved in this process.

RPE- or Macrophage-Derived VEGF-A Is Not Required for Neovessel Formation in CNV

It has been proposed that infiltrating macrophages express VEGF-A and promote neovascularization in CNV models and in neovascular AMD.⁷ In another study, however, no VEGF-A expression was observed in macrophages in neovascular AMD lesions in patients, although VEGF-A expression was observed in RPE cells.⁸

To assess whether macrophages or RPE cells express VEGF-A in CNV lesions, VEGF-A reporter mice (Vegfa^lacZKI/WT mice) were used, in which an NLS-lacZ allele was introduced into the 3′UTR of the VEGF-A gene.¹⁹ In these mice, nuclear β-gal actosidase (β-gal) staining reflects VEGF-A expression at a cellular resolution and allows precise characterization of the cellular origin of VEGF-A expression. In quiescent RPE–choroid tissues, these Vegfa^lacZKI/WT mice exhibited strong expression of VEGF-A in adult RPE cells; however no β-gal staining was seen in choroidal cells (Figure 5, A and B). The choroidal vessels did not express VEGF-A, but exhibited strong expression for both VEGF-A receptors Flt1 and Flk1, as demonstrated with β-gal staining of eyes of Flt1^lacZKI/WT mice and Flk1^lacZKI/WT mice (Figure 5, C and D).

Laser-induced CNV in Vegfa^lacZKI/WT mice revealed that RPE cells maintain strong expression of VEGF-A in CNV lesions (β-gal⁺), whereas no β-gal expression was seen in endothelial cells of forming vessels (Figure 5E). Importantly, lesional F4/80⁺ macrophages exhibited no expression of β-gal (Figure 5, F and G), and targeted ablation of VEGF-A in macrophages did not prevent CNV lesion formation, providing additional evidence that macrophages may promote neovascularization independently of VEGF-A in CNV (Figure 5, K and L). Furthermore, fully formed lesional neovessels could be seen in mice that lacked VEGF-A expression in the RPE (Figure 5, I and J).