Macrophage-derived interleukin-6 is necessary and sufficient for choroidal angiogenesis

Neovascular age-related macular degeneration (nAMD) commonly causes vision loss from aberrant angiogenesis, termed choroidal neovascularization (CNV). Interleukin-6 (IL6) is a pro-inflammatory and pro-angiogenic cytokine that is correlated with AMD progression and nAMD activity. We hypothesize that anti-IL6 therapy is a potential nAMD therapeutic. We found that IL6 levels were increased after laser injury and expressed by macrophages. Il6-deficiency decreased laser-induced CNV area and exogenous IL6 addition increased choroidal sprouting angiogenesis. Il6-null mice demonstrated equally increased macrophage numbers as wildtype mice. At steady state, IL6R expression was detected on peripheral blood and ocular monocytes. After laser injury, the number of IL6R+Ly6C+ monocytes in blood and IL6R+ macrophages in the eye were increased. In human choroid, macrophages expressed IL6, IL6R, and IL6ST. Furthermore, IL6R+ macrophages displayed a transcriptional profile consistent with STAT3 (signal transducer and activator of transcription 3) activation and angiogenesis. Our data show that IL6 is both necessary and sufficient for choroidal angiogenesis. Macrophage-derived IL6 may stimulate choroidal angiogenesis via classical activation of IL6R+ macrophages, which then stimulate angiogenesis. Targeting IL6 or the IL6R could be an effective adjunctive therapy for treatment-resistant nAMD patients.


Results
Since IL6 is a secreted molecule, we used RNAscope to identify IL6-producing cells. In 10-12 week-old female wildtype and Il6 −/− mice, we performed laser injury and harvested eyes on Day 3, the peak of macrophage recruitment 11 . We found no IL6 expression in Il6 −/− mice, confirming the validity of the RNAscope probe (Fig. 1a,c). In wildtype mice, IL6 expression was only detectable at the laser injury site (black arrow, Fig. 1b). In three independent wildtype mice, IL6 + cells (pink stain, green and yellow arrows) were found in the inflammatory lesion (Fig. 1d,e,h). Next, we performed immunohistochemical staining of serial sections (4 µ m sections, 2-4 sections apart) for IBA1 and F4/80 from Wildtype #2 (Fig. 1e) to identify if IL6 + cells were macrophages. We found both IBA1 + and F4/80 + staining at near identical locations to IL6 + cells ( Fig. 1e-g, colored arrows). In order to quantitatively confirm these results, we performed IL6 ELISA on posterior eye cups (retina and choroid-RPE-sclera complex). IL6 levels were increased 1.15-fold (p < 0.001) on Day 3 after laser injury (Fig. 1i). These data demonstrate that macrophages produce IL6 at the site of laser injury.
IL6 is increased after laser injury ( Fig. 1), and reduced by propranolol 3,4 . We next investigated whether IL6 is necessary for propranolol-driven CNV blockade. We subjected 10-12 week old female Il6 −/− mice to laser injury. Female mice were used because we previously demonstrated that male mice do not demonstrate propranololinduced CNV blockade 6 . Mice received daily intraperitoneal vehicle (PBS) or propranolol (20 mg/kg) injections for 14 days. In the context of IL6-deficiency, propranolol had no effect upon CNV area (Fig. 2a-c). Next, we investigated if IL6 deficiency affects CNV compared to wildtype mice. We subjected male and female 10-12 week old wildtype and Il6 −/− mice to laser injury and quantified CNV area on Day 14. IL6-deficient mice displayed a 42% reduction in CNV area (p < 0.01, Fig. 2d-f) with no sex-dependent effects ( Fig S1). These findings show that IL6 is necessary for both propranolol-driven CNV blockade and CNV pathogenesis.
IL6 is capable of directly stimulating angiogenesis outside the eye 10 . To test this effect in the choroid, we performed ex vivo choroidal sprouting assays in the presence of vehicle ( Fig. 3a-d) or exogenous IL6 ( Fig. 3e-h). Direct addition of IL6 at 10 and 30 ng/ml increased choroidal angiogenesis area by 1.2-fold (p < 0.05 for both) on Day 6, and 1.3-fold (p < 0.001 for both) on Day 7 (Fig. 3i). These data demonstrate that IL6 is sufficient to stimulate choroidal angiogenesis.
IL6 classically binds to cell surface IL6 receptor (IL6R) leading to intracellular signaling transduction via the gp130 coreceptor 15 . The IL6R is known to be expressed in leukocytes 15 , but has not been investigated in the eye. We used multi-parameter flow cytometry to identify IL6R expression. Wildtype 10-12 week old female mice were subjected to laser treatment, and eyes were harvested on Day 3. Mononuclear phagocytes were identified as CD45 + CD11b + Lin − identically to Fig. 4A,B. CD64 + Cx3cr1 + cells were gated forward (Fig. 5a), and microglia were defined as CD45 dim Cx3cr1 high while all other CD64 + Cx3cr1 + cells (Boolean gate) were delineated as ocular macrophages (Fig. 5b). Dendritic cells were identified from CD64 − Cx3cr1 − cells as MHCII + CD11c + (Fig. 5c). The non-DC population from Fig. 5c was gated forward (Boolean gate) and CD45 high Cx3cr1 + cells were defined as monocytes (Fig. 5d). The major populations of ocular IL6R + cells were monocytes and macrophages. At steady state, the number of IL6R + monocytes were greater than all other groups (Fig. 5e,i, p < 0.001). After laser injury, IL6R + monocytes were unaffected and were significantly more than all other groups except for macrophages (p < 0.001). The number of IL6R + macrophages were increased 9.3-fold (p < 0.01) by laser treatment, and were significantly greater than microglia and dendritic cell counts (Fig. 5i, p < 0.01). Few IL6R + microglia or dendritic cells were detected at steady state or after laser injury ( Fig. 5g-i). These findings show that monocytes and macrophages are the major populations of ocular IL6R + cells after laser injury.
In order to confirm our ocular monocyte findings, we performed multi-parameter flow cytometry on peripheral blood from unlasered and lasered mice. Wildtype 10-12 week old female mice were subjected to laser and peripheral blood was obtained on Day 3. B cells (CD19) and T cells (CD4, CD8) were identified from CD45 + cells (Fig. 6a). CD45 + CD19 − CD4 − CD8 − cells were gated forward and NK cells were delineated as NK1.1 + (Fig. 6b). Neutrophils were characterized as Ly6G + SSC med and eosinophils were defined as Ly6G − SSC high from  (Fig. 6c). Non-granulocytes were gated forward and identified as Ly6C − CD115 + or Ly6C + CD115 + monocytes (Fig. 6d). We found no change in the total number of B cells, T cells, NK cells, eosinophils, neutrophils, Ly6C + monocytes, or Ly6C − monocytes after laser injury ( Fig S2). B cells, T cells, NK cells, eosinophils, and neutrophils were < 5% IL6R + , found at low numbers, and were unchanged by laser ( Fig. 6e-i, 6l). At steady state, significantly more IL6R + Ly6C − (p < 0.01) and IL6R + Ly6C + (p < 0.001) monocytes were detected compared to all other groups with no difference between Ly6C − and Ly6C + monocytes ( Fig. 6j-l). After laser www.nature.com/scientificreports/ injury, IL6R + Ly6C − monocytes were unchanged and remained significantly more than all other groups (p < 0.001, Fig. 6j,l). Alternatively, the number of IL6R + Ly6C + monocytes increased with laser treatment (p < 0.001), and were significantly greater than all other groups (p < 0.001, Fig. 6k-l)). Since the total number of Ly6C + monocytes was unchanged with laser, these data suggest that laser stimulated existing Ly6C + blood monocytes to express   www.nature.com/scientificreports/ the IL6R. These data confirm that monocytes are IL6R + cells and constitute the overwhelming majority of IL6R + blood cells in the context of a laser-injury model. In order to investigate IL6 and IL6R expressing cells in humans, we re-analyzed a recently published singlecell RNA-seq data set from human RPE-choroid samples 16 . We integrated the data from all 7 patients, performed cell clustering with Seurat v3 17,18 , and visualized the clusters using the uniform manifold approximation and projection (UMAP) technique (Fig. 7a). We identified 5 subsets of IBA1(AIF1) + CD68 + macrophages, 3 subtypes of CD3E + or CD2 + T cells, 2 populations of CD79A + IGJ + B cells, and CPA3 + KIT + mast cells (Fig. 7a,b). IL6 expression was primarily detected in macrophages, and a few B cells (Fig. 7c). The IL6R was expressed predominantly by macrophage subsets Mac-C, Mac-D, and Mac-E (Fig. 7d). The IL6 coreceptor gp130 (IL6ST) demonstrated expression in all cell types (Fig. 7e).

Discussion
In this report, we used the choroid sprouting and laser-induced CNV models to investigate the IL6 pathway during choroidal angiogenesis. At steady state, no IL6 was detectable in the choroid, while monocytes expressed the IL6R (Fig. 8a). After laser injury, macrophages are recruited to the laser injury site, produce IL6, and express the IL6R (Fig. 8b). IL6 expression is both necessary and sufficient for choroidal angiogenesis. The likely mechanism by which IL6 stimulates angiogenesis is by classical activation of IL6R + macrophages to indirectly stimulate angiogenesis (Fig. 8d). Our data demonstrate that macrophages produce IL6 in the laser-induced CNV model (Fig. 1). Because IL6 is a secreted peptide, we used RNAscope to identify intracellular Il6 mRNA in order to delineate the cellular source www.nature.com/scientificreports/ of IL6. Our findings are in agreement with prior immunofluorescent staining of IL6 protein using CD11b 26 and Cx3cr1-GFP 27 co-labeling in the choroid. In addition, murine bone marrow-derived monocytes 28 and human alveolar macrophages 29 express and secrete IL6 protein. Therefore, our data add a new methodology to the literature supporting macrophages as a key source of IL6 production in the eye. Since IL6 is pro-inflammatory and pro-angiogenic cytokine 10 , we investigated the function of IL6 during choroidal angiogenesis. A prior group showed that male Il6 −/− mice display a 30% reduction in laser-induced CNV area 19 . Similarly, we demonstrate that Il6 −/− mice have a 42% reduction in CNV area in both male and female mice, with no sex-specific effects (Figs. 2 & S1). Additionally, we find that IL6 is sufficient to stimulate angiogenesis in the choroidal sprouting assay (Fig. 3). These data are similar to prior results that IL6 can stimulate angiogenesis in the aortic ring assay 10 , and promote endothelial cell line motility in the scratch wound assay 30 . Furthermore, human studies show that intraocular IL6 levels are associated with nAMD activity 31 , and systemic IL6 levels correlate with progression to advanced AMD 32 . Thus, our data show that IL6 is necessary and sufficient for choroidal angiogenesis in mice, and human studies support a pathogenic role for IL6 in nAMD.
The mechanism by which IL6 stimulates angiogenesis is unclear. IL6 uses both classical and trans-activation to exert its signaling effects 15 . In the classical pathway, IL6 binds to cell surface IL6R and signals intracellularly via gp130. Our data support a mechanism where macrophage-derived IL6 classically activates IL6R + macrophages, leading to macrophage-driven angiogenesis (Fig. 8d). IL6 was detected in mouse ( Fig. 1) and human (Fig. 7) macrophages. After laser injury in mice, IL6R + Ly6C + monocytes are increased (Fig. 6), and IL6R + macrophages are recruited to the eye (Fig. 5). In support of this mechanism, multiple studies have shown that macrophages, and specifically Ly6C + classical monocyte-derived macrophages, stimulate angiogenesis during laser-induced CNV 11,12,33 . Additionally, human choroidal macrophages expressed both IL6R and IL6ST (Fig. 7), demonstrating their ability to respond to the IL6 stimulus. Finally, IL6R + human macrophages demonstrated a transcriptional profile consistent with STAT3 activation, angiogenesis, and monocyte chemotaxis ( Fig. 7f-g). These data support a mechanism where macrophage-derived IL6 stimulates IL6R + macrophages to drive angiogenesis.
Our data do not exclude an additional mechanism where macrophage-derived IL6 classically signals to endothelial cells to promote angiogenesis (Fig. 8d). We did not investigate IL6R or co-receptor expression in endothelial cells. However, prior reports have shown IL6R expression on primary and cultured endothelial cells 10 . Furthermore, IL6 addition to aortic ring assays 10 or the choroidal sprouting assay (Fig. 3) stimulates angiogenesis. This effect could be directly on endothelial cells, but both aortic and choroidal tissue contain macrophages. These data suggest that macrophage-derived IL6 could signal classically through IL6R + endothelial cells to increase angiogenesis.
Trans-signaling is a third potential mechanism of IL6-driven angiogenesis (Fig. 8c). Our data show that monocytes are the major IL6R + cells in the eye and blood (Fig. 5-6). In human choroid, we did not detect monocytes, but macrophages were also the major IL6R + cell type (Fig. 7). IL6 may bind monocyte/macrophage-derived soluble IL6R and stimulate angiogenesis via trans-activation of endothelial cell gp130/IL6ST. In support of this model, exogenous IL6 and soluble IL6R stimulate angiogenesis via trans-signaling and downstream activation of STAT3 and Ca 2+ /calmodulin-dependent protein kinase IIδ(CaMKIIδ ) in endothelial cells 30 . Furthermore, conditional STAT3 and CaMKIIδ endothelial cell knockout mice demonstrate decreased retinal vascular development 30 . These data support a third possible mechanism of IL6-driven angiogenesis. Future studies using conditional IL6R knockout mice will be necessary to further elucidate the mechanism of IL6-driven angiogenesis.
In summary, we demonstrate that IL6 is up-regulated by laser injury and expressed by macrophages at the laser injury site. Additionally, IL6 is necessary and sufficient for choroidal angiogenesis. These findings support IL6 and the IL6R as possible therapeutic targets in nAMD. Finally, IL6 stimulates choroidal angiogenesis through one of three possible mechanisms: IL6 classical activation of macrophages to indirectly stimulate angiogenesis,

Methods
Animals. Breeding pairs of wildtype (C57BL6/J; #000664), and Il6 −/− (B6.129S2-Il6 tm1Kopf /J; 002650) were obtained from Jackson Labs (Bar Harbor, ME). Wildtype and Il6 −/− animals used in this study were first-or second-generation crosses of parental mice. One complete litter from each breeding pair was genotyped to confirm the correct genotype and the absence of the RD8 allele (Crb1). Genotyping services were performed by Transnetyx (Cordova, TN). All experiments were performed on 10-12 week old mice. Mice were housed in a specific pathogen free, barrier facility and maintained on a 12 h light/dark cycle. All experiments were conducted in accordance with the ARRIVE guidelines and were approved by the Northwestern University Institutional Animal Care and Use Committee.
Laser-induced CNV. Male and female 10-12 week-old mice were treated as previously described 6,11 . Briefly, mice were anesthetized with a ketamine/xylazine (Akorn, Lake Forest, IL) cocktail. Pain control and hydration were achieved with a 1 mg/kg subcutaneous injection of Meloxicam (Henry Schein Animal Health, Melville, NY). Eyes were anesthetized, dilated, and a cover slip was coupled to the cornea with Gonak (Akorn) for slit lamp microscopy and laser. Four (immunofluorescence) or eight (flow cytometry, immunohistochemistry and ELISA; to increase inflammatory cell numbers) focal burns (75 μm, 110 mW, 100 ms) were administered in each eye using a 532 nm argon ophthalmic laser (IRIDEX, Mountain View, CA) via a slit lamp delivery system (Zeiss, Oberkochen, Germany). Immunohistochemistry. Eyes were paraffin embedded identically to RNAscope. Slides were deparaffinized on an automated platform (Leica Autostainer XL, Leica Microsystems, Buffalo Grove, IL), then processed for antigen retrieval (Sodium Citrate pH 6 for 10 min at 110 °C). Slides were incubated with primary antibodies shown in Table 1 (Iba1, 1:1000; F4/80, 1:500) in PBS overnight at 4 °C in a humidified chamber. After washing with PBS, slides were incubated with the appropriate secondary antibody via an automated system (Biocare Intellipath, Pacheco, CA). Red chromogen development was completed with a Warp Red Chromogen kit (Biocare). Positive (brain and spleen) and no primary controls were used to confirm activity. All slides were counter stained with hematoxylin. Choroidal sprouting assay. Assays were performed as previously described 6 . Briefly, enucleated eyes from Immunofluorescence. Eyes were treated as previously described 6,11  Flow cytometry of whole eyes. Experiments were performed as previously described 13 . Briefly, female mice (unlasered control and Day 3 post laser) were sacrificed, and enucleated eyes were placed into HBSS. Animals were not perfused as we previously demonstrated no change in macrophage numbers at steady state or after laser injury with or without systemic perfusion 13 . Eyes were cleaned of optic nerve, extraocular muscles, orbital tissue, and conjunctiva. Whole mouse eyes including cornea, sclera, iris, ciliary body, lens, vitreous, retina, RPE, and choroid, were minced into small pieces. Eye pieces were further mechanically and chemically digested before passing through a 40 μm filter to obtain a single cell suspension. Cell suspensions were stained for live cells and washed. Cell suspensions were blocked and stained with cell surface antibodies found in Table 1. Both eyes were pooled from one mouse to determine cells per mouse, using counts beads. Samples were run on a modified LSRII (BD Biosciences, San Jose, CA) and analyzed using FlowJo v10. Please see our prior publication for all fluorescence minus one controls 13 .

ELISA. Enucleated eyes from control and
Flow cytometry of peripheral blood. Experiments were performed as previously described 11 on unlasered control and Day 3 post laser. Briefly, blood from freshly sacrificed animals was obtained via cardiac puncture. Samples were placed in EDTA tubes (Sarstedt, Numbrecht, Germany) to prevent clotting. Cells were blocked and stained by an antibody cocktail (Table 1). Stained cells were fixed and red blood cells were lysed using FACSLyse (BD Biosciences). After washing, samples were resuspended in MACS buffer (Miltenyi Biotec, Auburn, CA) and run on a modified LSRII with subsequent analysis using FlowJo v10.
Bioinformatics. Gene expression data (.tsv files) from human choroidal samples were downloaded from the GEO database (GSE135922). Data was imported into Seurat v3 17,18 . The FindIntegrationAnchors followed by the IntegrateData functions (dims 1:50) were used to integrate the data into one data set and perform batch corrections. The data were rescaled (ScaleData function), and principal component analysis (PCA) was performed www.nature.com/scientificreports/ (RunPCA, npcs = 50). The Elbow Plot technique was used to identify 19 significant principal components (PCs). Cells were clustered using FindNeighbors (dims = 1:19) followed by FindClusters (resolution = 0.4). The RunU-MAP function was used to visualize the cell clusters. Differential expression and cell identification were performed using FindAllMarkers (min.pct = 0.25, logfc.threshold = log2). The FeaturePlot and DotPlot functions were used to visualize gene expression. FindMarkers (min.pct = 0.25, logfc.threshold = log2) was used to generate a list of differentially expressed genes between Mac-C, Mac-D, Mac-E and all other immune cells. Gene ontology (GO) enrichment analysis was performed on up-regulated genes independently using a fold change cut-off = > 2.0, and adjusted p-value < 0.001. GOrilla was used for GO enrichment 34,35 , using a background of genes expressed only in choroidal leukocytes. GO terms were visualized using ggplot that met a number of genes (b) > 2, enrichment > 5, and false discovery rate (FDR) q-value < 0.05.
Statistical Analysis. Statistical analysis was performed in Prism. IL6 levels and CNV area in Il6 null mice (PBS vs Propranolol) were compared by Student's unpaired t-test. CNV area in wildtype vs in Il6 null mice were compared by Welch's t-test due to unequal variances between groups. Choroidal sprouting area analysis was performed using Two-Way ANOVA followed by Dunnett's multiple comparisons test. Flow cytometry comparisons of macrophage numbers were made using the Brown-Forsythe and Welch ANOVA followed by Dunnett's T3 multiple comparisons test due to unequal variances between unlasered and lasered mice. IL6R expression in eyes and blood was compared using Two-Way ANOVA followed by Tukey's multiple comparisons test.

Data availability
The datasets used and analyzed for mouse studies are available from the corresponding author on reasonable request. All human data analyzed during this study are included in this published article and its supplementary information files.