Monocyte-Derived Macrophages Are Necessary for Beta-Adrenergic Receptor-Driven Choroidal Neovascularization Inhibition

Abstract

Purpose: Beta-adrenergic receptor (AR) antagonists, like propranolol, inhibit angiogenesis in multiple ocular conditions through an unknown mechanism. We previously showed that propranolol reduces choroidal neovascularization (CNV) by decreasing interleukin-6 levels. Since macrophages are one of the central producers of interleukin-6, we examined whether macrophages are required for propranolol-driven inhibition of choroidal angiogenesis.

Methods: We tested the anti-angiogenic properties of propranolol in the choroidal sprouting assay and the laser-induced CNV model. Bone marrow-derived monocytes (BMDMs) were added to the choroidal sprouting assay and Ccr2^−/− mice were subjected to laser-induced CNV. Multi-parameter flow cytometry was performed to characterize the ocular mononuclear phagocyte populations after laser injury and during propranolol treatment.

Results: Propranolol reduced choroidal angiogenesis by 41% (P < 0.001) in the choroidal sprouting assay. Similarly, propranolol decreased laser-induced CNV by 50% (P < 0.05) in female mice, with no change in males. BMDMs increased choroidal sprouting by 146% (P < 0.0001), and this effect was ablated by propranolol. Beta-AR inhibition had no effect upon laser-induced CNV area in female Ccr2^−/− mice. MHCII⁺ and MHCII⁻ macrophages increased 20-fold following laser treatment in wildtype mice as compared to untreated mice, and this effect was completely attenuated in lasered Ccr2^−/− mice. Moreover, propranolol increased the numbers of MHCII⁺ and MHCII⁻ macrophages by 1.9 (P = 0.07) and 3.1 (P < 0.05) fold in lasered female mice with no change in macrophage numbers in males.

Conclusions: Our data suggest that propranolol inhibits angiogenesis through recruitment of monocyte-derived macrophages in female mice only. These data show the anti-angiogenic nature of beta-AR blocker-recruited monocyte-derived macrophages in CNV.

Over the last 10 years, the use of beta-adrenergic receptor (beta-AR) blockers has expanded to include multiple anti-angiogenic indications within ophthalmology. Leaute-Labeze et al. first reported the tumor regressive properties of the non-specific beta-AR blocker propranolol in severe hemangioma of infancy.¹ Additionally, systemic² and topical³ propranolol regresses pre-threshold retinopathy of prematurity. Furthermore, patients on aqueous suppressants (including beta-AR blockers such as timolol) demonstrate improved visual acuity and reduced retinal thickness compared to controls in the comparison of age-related macular degeneration (AMD) treatments trial.⁴ Finally, adjuvant topical dorzolamide-timolol improves retinal thickness in patients with neovascular AMD on maximal anti-vascular endothelial growth factor (VEGF) therapy.^5,6 

Despite the clinical efficacy of beta-AR antagonists in angiogenic eye diseases such as neovascular AMD, its mechanism of action is poorly understood. Isolated hemangioma cells demonstrate reduced VEGF protein levels after culture with propranolol,⁷ but immunohistochemical analysis revealed no change in VEGF or VEGF receptor expression of surgically excised hemangiomas,⁸ suggesting alternative mechanisms. We previously showed that propranolol inhibits laser-induced choroidal neovascularization (CNV) by 50%.⁹ Additionally, reduced CNV area correlates with a decrease in the ocular level of the proinflammatory cytokine, interleukin-6, while the levels of VEGF remain unchanged.¹⁰ Since macrophages are known to be large producers of interleukin-6, we examined whether propranolol-driven CNV inhibition is mediated by macrophages. Macrophage depletion has already been shown to significantly reduce laser-induced CNV area,^11,12 and the beta2-AR is necessary for particulate matter-triggered interleukin-6 expression in lung macrophages.¹³ 

In this report, we investigated the mechanism of propranolol-mediated suppression of choroidal angiogenesis. Propranolol reduced the laser-induced CNV area in female but not male mice. Additionally, propranolol failed to inhibit laser-induced CNV in female C-C chemokine receptor type 2 (CCR2) knockout mice. Finally, propranolol increased macrophage recruitment to laser lesions in female but not male mice. These data suggest that beta-AR antagonist-driven recruitment of monocyte-derived macrophages are necessary for inhibition of CNV and that macrophage recruitment explains the sex-specific effects of beta-AR inhibitors. 

Methods

Mice

All experiments were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Northwestern University Institutional Animal Care and Use Committee. C57BL/6J wildtype (#664) and Ccr2^−/− (#4999) mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA). Mice were bred in-house and maintained in a pathogen-free barrier facility within Northwestern University's Center for Comparative Medicine. Any animals received directly from The Jackson Laboratory were housed for at least 2 weeks before use. All experiments were performed at 10 to 12 weeks of age. Genotypes of all mice, including the absence of the Rd8 mutation, were confirmed by services provided by Transnetyx, Inc. (Cordova, TN, USA). 

Choroidal Sprouting Assay

Eyes were carefully enucleated and placed in ice-cold PBS. Eyes were dissected in EGM2-MV supplemented (hydrocortisone was omitted) medium (CC3202; Lonza, Walkersville, MD, USA) into posterior eye cups containing sclera, RPE, and choroid complex. The peripheral choroid was separated from the central choroid, cut into 0.5 mm × 0.5 mm pieces, and placed into growth factor reduced Matrigel (#356231; Corning, Bedford, MA, USA) in a 48-well plate on ice. The Matrigel was solidified by incubating at 37°C for 10 minutes. EGM2-MV medium was changed every 2 days. Sunitinib (PZ0012) and propranolol (P0884) were purchased from Sigma (St. Louis, MO, USA). Sunitinib was added on Day 0, and propranolol was added on either Day 0 or Day 2. Pictures were taken on a Nikon Ti2 Widefield microscope (Buffalo Grove, IL, USA) using a 4× objective and Nikon NIS Elements software. Images were analyzed with the Nikon Elements General Analysis. Images were preprocessed with edge detection and segmented with thresholding. The largest area was measured for each image. The central choroidal tissue area was subtracted from the total angiogenesis area. 

Bone Marrow-Derived Monocyte (BMDM) Isolation

Mice were killed and hind limbs were removed while maintaining integrity of the hip joint. Legs were kept in DMEM (#10-013-CV; Corning, Manassas, VA, USA) on ice until removal of muscles. The femur and tibia were cut proximal to the joint before flushing bone marrow with DMEM applied via a syringe fitted with a 21G needle. Flushed bone marrow was pushed through a 40 μm filter (#352340; Falcon, Durham, NC, USA) with a syringe plunger, and the filter was rinsed with additional DMEM. Cells were spun at 350g for 15 minutes and then passed through a second 40 μm filter. Monocytes were isolated from whole bone marrow using a negative selection magnetic bead protocol from Miltenyi Biotec (#130-100-629; Auburn, CA, USA) where non-monocytes were antibody labeled and positively selected via a magnetic column, while the non-labeled monocyte-enriched fraction was isolated from the flow through (LS 130-042-401; Miltenyi Biotec). Isolated BMDMs were counted and added to the choroidal sprouting assay in EGM2-MV medium on Day 2. 

Laser-Induced CNV

Mice were anesthetized via intraperitoneal (IP) delivery of ketamine (90 mg/kg; Akorn, Lake Forest, IL, USA)/xylazine (12 mg/kg; Akorn) cocktail. Eyes were anesthetized with 1 drop of 0.5% tetracaine (Alcon, Fort Worth, TX, USA). Pupils were dilated with 1 drop of 2.5% phenylephrine (Akorn) and 0.5% tropicamide (Akorn). The whiskers were trimmed with scissors. A subcutaneous injection of meloxicam (1 mg/kg; Henry Schein Animal Health, Melville, NY, USA) was given for pain control. Mice were moved to the slit lamp stage and a cover slip was coupled to the cornea using Goniosoft (OCuSOFT, Rosenberg, TX, USA) to allow direct retinal visualization. Mouse eyes were treated with 4 (CNV area quantitation) or 8 (flow cytometry) focal laser burns (75 um, 110 mW, 100 msec) in each eye using an IRIDEX (Mountain View, CA, USA) 532 nm argon ophthalmic laser delivered via a Zeiss (Oberkochen, Germany) slit lamp. IP injections of propranolol (20 mg/kg) or PBS were performed daily during the course of experiments. 

Fluorescein Angiography (FA) Imaging

Mice were anesthetized and prepped as described above on Day 14 after laser-induced CNV. Fluorescein (0.03–0.05 ml, 100 mg/ml; Akorn) was injected into the tail vein. The fundus was imaged using the Micron3 fundus camera (Phoenix Technology Group, Pleasanton, CA, USA). After imaging, mice were given ophthalmic ointment (Akorn) and placed in a heated recovery cage. Area of leakage was assessed by a reviewer blinded to treatment and genotype by using the freehand tool to draw a perimeter around the area of leakage and using the ROI manager to calculate an area in pixels via Fiji (NIH, Bethesda, MD, USA). Grading of leakage was assessed by two blinded reviewers using the 1–4 grading scale, as previously described.¹⁴ Because we did not have early frames for every eye, only late phase angiograms were scored and grade 2 was not assessed as 2A and 2B.¹⁵ When the two blinded reviewers occasionally differed by 1 point, the scores were averaged. 

Immunofluorescence (IF) Imaging

Mice were killed directly after FA imaging, and enucleated eyes were placed into 1% paraformaldehyde (#15713-S; Electron Microscopy Sciences, Hatfield, PA, USA) for 1 hour at room temperature. Eyes were moved to PBS; conjunctiva and optic nerve remnants were removed. A limbal incision was carried 360° to remove the cornea and iris. The lens and retina were carefully teased apart and discarded leaving a posterior eye cup of RPE, choroid, and sclera. Cups were blocked overnight in Donkey Serum Blocking Buffer (DSBB): Tris buffered saline (#BP2471, Fisher, Fair Lawn, NJ, USA) containing 5% donkey serum (S30, Sigma), 2.5% bovine serum albumin (A2153, Sigma), and 0.5% Triton X-100 (X100; Sigma). Eye cups were then incubated with anti-ICAM-2 antibody (Table) in DSBB overnight at 4°C before washing 5 times with TBS-T (TBS with 0.5% Tween-20, #00777; Amresco, Solon, OH, USA). Secondary antibody staining was performed with an Alexa Fluor 488-conjugated anti-rat antibody (Table) overnight at 4°C before washing 5 times with TBS-T containing 0.5% Tween-20 (Amresco). Eye cups were flat mounted on HistoBond microscope slides (16004-406, VWR; Batavia, IL, USA) in Immu-Mount (#9990402; ThermoFisher, Carlsbad, CA, USA). Pictures were taken on a Nikon Ti2 widefield microscope using Nikon NIS Elements software. Area of neovascularization was assessed by a blinded reviewer using the freehand tool to draw a perimeter around the area of leakage and an ROI manager to calculate an area in pixels via Fiji. Next, pixel area was transformed to microns² by multiplying by a factor related to pixel to micron ratio (in our case, 0.5249). CNV rate was calculated by determining the ratio of CNV lesions detected on FA and IF per mouse over the number of bubbles formed during the laser process. 

Table

 

List of Antibodies Used for Flow Cytometry and Immunofluorescence

 

Flow Cytometry

Enucleated eyes were minced into small pieces and digested using a mixture of Liberase TL (0.1 mg/ml, #5401020001; Roche, Basel, Switzerland) and DNase I (0.2 mg/ml, #10104159001; Roche) in HBSS (14025-092; Fisher) at 37°C for 60 minutes with mechanical digestion at time 0, 30, and 60 minutes via a gentleMACS Octo Dissociator (130-095-937; Miltenyi Biotec) in C-tubes (130-096-334; Miltenyi). Cells were filtered through 40 μm nylon mesh in MACS Buffer (130-091-221; Miltenyi). Cells were spun at 350g for 10 minutes and washed once in MACS Buffer. Erythrocytes were lysed using PharmLyse (1 minute, #555899; BD Biosciences) and washed once as above. Cells were stained for dead cells using Aqua Live/Dead (15 minutes, Table) and washed twice. Cells were incubated with Fc-Block (20 minutes, Table), stained with fluorochrome-conjugated Abs (30 minutes, Table), washed twice, fixed with 2% PFA and washed twice. Data were acquired on an LSR II flow cytometer (BD Biosciences). Compensation matrices and data analysis were performed using FlowJo (Tree Star, Ashland, OR, USA). Approximately 3,000,000 singlet events were counted per mouse (2 eyes). First, singlet cells were identified and dead cells were excluded. CD45, a marker of leukocyte lineage, was used to identify leukocytes versus non-leukocytes. CD45⁻ non-leukocytes were gated on CD31 and CD140b to identify endothelial cells (EC) and pericytes, respectively. CD45⁺ leukocytes were gated on CD11b⁺ staining to identify mononuclear phagocytes and SiglecF, Ly6G, Nk1.1, B220, CD4, and CD8 negative staining to exclude eosinophils, neutrophils, NK cells, B cells, and T cells. Microglia have previously been shown to be CD45^dim and MHCII^low compared to infiltrating monocytes and macrophages.¹⁶ Therefore, we used CD45^dim versus CD45^high to identify microglia from infiltrating cells. Microglia were identified in the CD45^dim group as CD64⁺ and predominantly MHCII^low. In the CD45^high group, we identified CD64⁻MHCII⁻ monocytes, CD64⁻MHCII⁺ dendritic cells (DC), and two populations of macrophages: CD64⁺MHCII⁻ macrophages (MHCII⁻ Macs) and CD64⁺MHCII⁺ macrophages (MHCII⁺ Macs). Fluorescence minus one controls were used to establish positive staining and gates (Supplementary Fig. S1). 

Statistical Analysis

Statistical analysis was performed using GraphPad Prism 8 (San Diego, CA, USA). Choroidal explants were analyzed using 2-way ANOVA (time and treatment) followed by Dunnett's or Tukey's multiple comparisons test. We used Dunnett's test when we were comparing only to control, and Tukey's test when comparing among all groups. Laser-induced CNV lesion areas were averaged for each mouse and compared using the Mann-Whitney U test because the data were non-parametrically distributed. FA leakage grading was analyzed by Mann-Whitney on a per lesion basis. Flow cytometry data were analyzed as percent live cells for each cellular population per mouse. Comparisons between unlasered wildtype, lasered wildtype, and lasered Ccr2^−/− mice were made by 1-way ANOVA followed by Tukey's multiple comparisons test because there were three groups. Comparisons between PBS and propranolol were made using the Mann-Whitney test because the data were non-parametrically distributed. 

Results

The choroidal sprouting ex vivo model is a 3-dimensional tissue model of angiogenesis that includes endothelial cells, pericytes, retinal pigment epithelium (RPE), and tissue resident choroidal macrophages.¹⁷ We previously showed that propranolol reduces VEGF and IL-6 expression in primary mouse choroidal endothelial cells, pericytes, RPE, and microglia cell lines, all of which express beta-ARs.^9,10 To further test the anti-angiogenic properties of propranolol, we performed a dose-response analysis of propranolol in the choroidal sprouting assay. Angiogenic sprouts were first visible on Day 3, and representative images of vehicle- and propranolol-treated choroidal explants are shown in Figures 1A through 1F. We identified a dose-dependent reduction in choroidal angiogenesis from 14% to 41% with 50 nM to 5 μM propranolol (Fig. 1G, P < 0.001 for 5 μM). Propranolol was nearly as effective as the 56% inhibition obtained with the multi-receptor tyrosine kinase inhibitor sunitinib, which potently inhibits VEGF receptor 2 signaling (Fig. 1G; P < 0.0001). 

Figure 1

 

Propranolol inhibits choroidal angiogenesis in the ex vivo sprouting assay. Representative photos of vehicle (PBS, [A–C]) and propranolol 5 μM (D–F) treated choroidal explants on Day 3, Day 5, and Day 7. (G) Choroidal angiogenesis growth curve from Day 4–Day 7 (N = 6–12 per group, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 compared to control, color of symbol identifies comparison to control).

Our prior report demonstrated that propranolol inhibits laser-induced CNV by 50% in female mice using immunofluorescence (IF) staining.⁹ However, we did not investigate male mice or the effect size using fluorescein angiography (FA). Here, we performed laser-induced CNV in male and female mice undergoing daily intraperitoneal injection of vehicle (PBS) or propranolol (20 mg/kg) starting on the day of laser treatment. Propranolol reduced laser-induced CNV area by 50% by both FA and IF (Fig. 2H–2I; P = 0.01 for both). Propranolol had no effect upon FA leakage (Fig. 2G) or rate of CNV formation (Fig. 2J) in female mice. In contrast, propranolol had no effect upon CNV area by FA or IF (Fig. 2L–2M) in male mice. Interestingly, propranolol caused more Grade 3–4 lesions on FA in male mice (Fig. 2K; P < 0.01) without changing Grade 1 lesions or CNV rate (Fig. 2N). These data suggest that propranolol reduced the area of lesions without affecting their FA leakage characteristics in female mice. Alternatively, propranolol caused more lesions with leakage in male mice, but the size of the lesions was not significantly different. These data suggest a sex-specific response to beta-AR driven CNV inhibition, although the mechanism is unknown. 

Figure 2

 

Propranolol reduces laser-induced CNV area in female mice. Representative fundus photography, FA, and IF images from PBS- (A–C) and propranolol-treated (D–F) female mice on Day 14. CNV leakage grading on FA (G) was not changed in female mice (violin plot with solid line indicating median; dashed lines indicated quartiles; N = 48–49 lesions per group). CNV area on FA (H) and IF (I) was reduced by 50% in female mice with no change in CNV rate (J, N = 6–7 mice per group, *P < 0.05). CNV leakage grading on FA was increased by propranolol in male mice (K, **P < 0.01, N = 80–86 lesions per group). CNV area (L, M) and CNV rate (N) were unchanged in male mice (N = 10–11 mice per group).

We recently identified that intravitreal beta2-AR inhibition reduces CNV area and ocular IL-6 levels without affecting the expression of VEGF.¹⁰ Macrophages, which are known to produce high levels of IL-6 and express beta-ARs,¹³ were examined in the choroidal sprouting ex vivo model with an adaptation using bone marrow-derived macrophages (BMDMs). The addition of 10,000 and 20,000 BMDMs increased choroidal angiogenesis by 127% (P < 0.01) and 146% (P < 0.0001) on Day 7, respectively (Fig. 3J). To test the effect of propranolol on BMDM-stimulated choroidal angiogenesis, we added propranolol in conjunction with monocyte addition. We found that propranolol alone mildly reduced choroidal angiogenesis by 10% without BMDMs (P = NS; Fig. 3K). Alternatively, BMDMs stimulated angiogenesis by 118% (P < 0.001), and this effect was attenuated in the presence of propranolol (P = NS versus control, P < 0.05 versus BMDM; Fig. 3K). These data suggest that monocytes or monocyte-derived macrophages may regulate the effects of propranolol on choroidal angiogenesis. 

Figure 3

 

Propranolol blunts BMDM-stimulated choroidal angiogenesis in the ex vivo sprouting assay. Representative photos of PBS- (A–C), BMDM-treated (D–F), BMDM and propranolol-treated (1 μM, [G–I]) choroidal explants on Day 4, Day 5, and Day 7. The small cells around the explant are BMDMs (D, G), and are easily distinguishable from the microvascular angiogenesis of the explant. (J) Choroidal angiogenesis growth curve from Day 4–Day 7 (N = 6–12 per group, *P < 0.05, **P < 0.01, ****P < 0.0001 compared to control, color of symbol identifies comparison to control). (K) BMDM increase angiogenesis area on Day 7 (***P < 0.001), and this effect is inhibited by propranolol (*P < 0.05, N = 11–12 per group).

C-C motif chemokine receptor 2 (CCR2) is a receptor that is necessary for emigration of classical monocytes from the bone marrow into the circulation and for recruitment of monocytes to sites of tissue injury.¹⁸ We subjected Ccr2^−/− mice to laser-induced CNV and tested whether propranolol was capable of inhibiting CNV in female mice. Propranolol treatment had no effect on CNV area or CNV rate in female Ccr2^−/− mice (Figs. 4A–C). These data demonstrate that monocyte-derived macrophages are necessary for propranolol-driven CNV inhibition. 

Figure 4

 

Propranolol has no effect upon laser-induced CNV area in the context of CCR2 inhibition or Ccr2 deficiency. CNV area (A, B) and CNV rate (C) were unaffected by daily PBS or propranolol (20 mg/kg) in Ccr2^−/− female mice (N = 7 per group).

Since macrophages are known to be heterogenous, we examined the different populations of tissue macrophages via flow cytometric analysis on Day 3, the peak of macrophage recruitment.¹² Microglia were identified in the CD45^dim group as CD64⁺ and predominantly MHCII^low (Fig. 5). In the CD45^high group, we identified CD64^−-MHCII⁻ monocytes, CD64⁻MHCII⁺ dendritic cells (DC), and two populations of macrophages (Fig. 5): CD64⁺MHCII⁻ macrophages (MHCII⁻ Macs) and CD64⁺MHCII⁺ macrophages (MHCII⁺ Macs). 

Figure 5

 

Flow cytometry gating strategy. Flow cytometry gating strategy for unlasered (A) and lasered (B) mice. Singlets were identified and gated forward. Dead cells were excluded. Live cells were separated into CD45⁻ and CD45⁺ populations. Endothelial cells (EC) were delineated by CD31+ staining and pericytes were identified by CD140b+ staining in CD45− cells. Eosinophils, neutrophils, NK cells, T cells, and B cells were excluded by negative staining for SiglecF, Ly6G, NK1.1, CD4, CD8, and B220 (Dump gate). CD11b⁺Dump⁻ cells were separated into CD45^dim and CD45^high groups. Microglia were identified as CD64⁺MHCII^low in the CD45^dim group. Macrophages were detected as CD64⁺MHCII⁻ (MHCII⁻ Macs) and CD64⁺MHCII⁺ (MHCII+ Macs) in the CD45^high group. Monocytes were delineated as CD64⁻MHCII⁻, and dendritic cells (DC) were identified as CD64⁻MHCII⁺.

To determine which cellular populations were CCR2-dependent during laser-induced CNV, we compared female untreated and lasered wildtype (WT) mice with female lasered Ccr2^−/− mice using multi-parameter flow cytometry on Day 3. There was no change in the percent or numbers of pericytes, endothelial cells, or monocytes regardless of treatment or genotype (Figs. 6D, 6E, 7H). DC numbers increased with laser treatment by 9.9-fold (P < 0.05) in WT mice and were non-significantly increased 4-fold in Ccr2^−/− lasered mice (Fig. 7I). We detected 2.1- and 1.9-fold more microglia in lasered WT (P < 0.01) and Ccr2^−/− (P < 0.01) mice, respectively (Fig. 7D). MHCII⁻ and MHCII⁺ macrophages increased 26.2- (P < 0.05) and 18.3-fold (P < 0.05) after laser treatment, respectively, and this effect was completely ablated in lasered Ccr2^−/− mice (Figs. 7J–7K). These data suggest that MHCII⁻ and MHCII⁺ macrophages are the predominant CCR2-dependent tissue infiltrating cells during laser-induced CNV. 

Figure 6

 

Pericytes and endothelial cells are unaffected by laser or Ccr2-deficiency. Representative flow cytometry plots of CD45⁻ cells in WT unlasered (A), WT lasered (B), and Ccr2^−/− lasered (C) mice. Pericytes (D) and endothelial cells (E) were unaffected by laser treatment nor Ccr2-deficiency (N = 3–4 per group).

Figure 7

 

MHCII⁺ Macs and MHCII⁻ Macs are the primary CCR2-dependent tissue infiltrating cells. Representative flow cytometry plots of CD45^dim cells in WT unlasered (A), WT lasered (B), and Ccr2^−/− lasered (C) mice. (D) Microglia increased with laser treatment in WT and Ccr2^−/− mice compared to WT unlasered mice. Representative flow cytometry plots of CD45^high cells in WT unlasered (E), WT lasered (F), and Ccr2^−/− lasered (G) mice. (H) Ocular circulating monocytes were non-significantly increased by laser treatment in wildtype (WT) mice, but unchanged in Ccr2^−/− mice. (I) Dendritic cells (DC) increased with laser in WT mice and were non-significantly reduced by Ccr2-deficiency. MHCII⁺ (J) and MHCII⁻ (K) Macs dramatically increased with laser in WT mice and this effect was completely blocked in Ccr2^−/− mice (N = 3–4 per group *P < 0.05, **P < 0.01).

Next, we compared female, lasered WT mice treated with vehicle or propranolol on Day 3 to determine if propranolol affects macrophage recruitment to laser lesions. We found no change in pericytes or endothelial cells between PBS- and propranolol-treated mice (Figs. 8A, 8B). Monocytes, DC, and microglia numbers were non-significantly increased by 1.5-fold (P = 0.62), 2.1-fold (P = 0.12), and 1.2-fold (P = 0.10), respectively, with propranolol treatment (Figs. 8C–E). However, we detected 3.1-fold more MHCII⁻ macrophages (P < 0.05) and 1.9-fold more MHCII⁺ macrophages (P = 0.07) with propranolol treatment (Figs. 8F, 8G). 

Figure 8

 

 Propranolol increased MHCII⁻ Macs during laser-induced CNV in female mice. Propranolol had no effect upon pericytes (A), endothelium (B), monocytes (C), dendritic cells (D), or microglia (E). (F, G) Propranolol treatment increased MHCII⁺ Macs (P = 0.07) and MHCII⁻ Macs (P = 0.03) on Day 3 (N = 9–10 per group).

Finally, we compared male, lasered WT mice treated with vehicle or propranolol on Day 3 to determine if recruitment of MHCII⁻ and MHCII⁺ macrophages is responsible for the lack of CNV inhibition in male propranolol-treated mice. Similar to female mice, there was no significant differences in the numbers of pericytes, endothelial cells, microglia, DC, or monocytes between male vehicle and propranolol-treated mice (Figs. 9A–9E). In contrast to female mice, male propranolol-treated mice demonstrated no change in the numbers of MHCII⁻ nor MHCII⁺ macrophages (Figs. 9F, 9G). These data suggest that beta-AR antagonist-driven recruitment of MHCII⁻ and MHCII⁺ macrophages explains the sex-specific differences between male and female mice. 

Figure 9

 

 Propranolol has no effect upon ocular mononuclear phagocytes during laser-induced CNV in male mice. Propranolol had no effect upon pericytes (A), endothelium (B), monocytes (C), dendritic cells (D), microglia (E), MHCII⁺ Macs (F), or MHCII⁻ Macs (G, N = 7 per group).

Discussion

Beta-AR inhibitors are clinically used for many ophthalmic indications including severe hemangioma of infancy,¹ retinopathy of prematurity,^2,3 and persistent intraretinal or subretinal fluid in patients with neovascular AMD⁶ or retinal vein occlusion.¹⁹ In this report, we used two models of choroidal angiogenesis to study these effects further: the ex vivo choroidal sprouting assay and the laser-induced CNV model. We found that propranolol inhibits angiogenesis in both of these models (Figs. 1, 2). In the laser-induced CNV model, we treated the mice with propranolol directly before laser injury. Because laser-induced CNV does not develop for several days, this is both a prevention and treatment study. Since there is no change in CNV formation or Grade 1 FA leakage, the data do not suggest that propranolol prevented CNV formation. In epidemiologic studies beta-AR blockers have been shown to both increase²⁰ and decrease²¹ the incidence of nAMD diagnoses. Therefore, our data do not clarify this controversial subject. Instead, our data demonstrate that propranolol decreased CNV size without affecting leakage characteristics in female mice, suggesting a treatment effect. These data are supported by multiple clinical studies that showed improved retinal thickness with dorzolamide-timolol in conjunction with anti-VEGF treatment in nAMD patients.^5,6 

Despite their experimental and clinical efficacy, the mechanism by which beta-AR blockers inhibit angiogenesis is poorly understood. We previously demonstrated that intravitreal beta2-AR inhibition reduces laser-induced CNV and is correlated with decreased ocular IL-6 levels with no change in VEGF levels.¹⁰ Because IL-6 is produced by macrophages and monocyte-derived macrophages are necessary for laser-induced CNV,¹¹ we sought to determine if monocyte-derived macrophages are critical for the anti-angiogenic effects of beta-AR inhibition. Additionally, we showed that addition of BMDMs increased choroidal angiogenesis (Fig. 3J), and this effect was ablated by propranolol treatment (Fig. 3K). In confirmation of the critical role of monocyte-derived macrophages, we demonstrated that propranolol no longer decreased laser-induced CNV in Ccr2^−/− mice, presumably due to the lack of monocyte-derived macrophages (Fig. 4). These data illustrate that CCR2-dependent monocyte-derived macrophages are necessary for beta-AR driven CNV inhibition. 

Because loss of monocyte-derived macrophages reduces the laser-induced CNV area,¹¹ we tested whether propranolol inhibits recruitment of monocyte-derived macrophages to the laser injury. This mechanism of action is supported by a cardiac injury model where the beta2-AR is necessary for CCR2 expression and leukocyte infiltration to the injury site.²² We measured mononuclear phagocytes on Day 3 after laser injury, the peak of macrophage recruitment.¹² After laser treatment, MHCII⁻ and MHCII⁺ Macs are the primary CCR2-dependent monocyte-derived macrophage types recruited to the injury site (Fig. 7). Propranolol did not change the endothelial cell, pericyte, or monocyte numbers on Day 3, which was expected because the CNV has not formed on Day 3, and the monocytes detected are not tissue recruited but in the ocular circulation. Importantly, beta-AR blockade increased, rather than decreased, the number of MHCII⁻ and MHCII⁺ Macs after laser injury compared to vehicle control in female mice (Fig. 8). These data clearly display that reduction of monocyte-derived macrophage recruitment is not the mechanism by which beta-AR blockade inhibits CNV. 

Although Ccr2-deficiency causes reduced macrophage infiltration to laser lesions (Fig. 7) and decreased CNV area,¹¹ both pro-angiogenic and anti-angiogenic macrophages have been described. For example, Kelly et al.²³ reported that aged macrophages display a pro-angiogenic phenotype that is regulated by increased interleukin-10 expression. Alternatively, intravitreal injection of isolated splenic macrophages from young mice²³ or M1-polarized macrophages²⁴ are capable of inhibiting laser-induced CNV area. Furthermore, IL-10 knockout mice demonstrate decreased CNV area and increased macrophage numbers,²⁵ demonstrating three examples of anti-angiogenic ocular macrophages. Therefore, beta-AR blockade may cause either more anti-angiogenic macrophages to be recruited to the laser lesion or a shift in macrophage gene expression and phenotype toward an anti-angiogenic subtype. In support of this anti-angiogenic mechanism, beta-AR agonism increases Vegfa expression in bone marrow-derived macrophages.²⁶ Future studies will investigate if beta-AR blockade decreases the expression of pro-angiogenic genes, such as Vegfa, in monocyte-derived macrophages. 

It was surprising that male mice did not demonstrate reduced CNV area (Fig. 2) nor increased macrophage recruitment to laser injury (Fig. 9) during beta-AR blockade. Epidemiologic studies are variable, but several reports show a positive association between female sex and AMD risk.²⁷ Furthermore, the incidence of late stage AMD, including neovascularization and geographic atrophy, was 38% higher in women in a meta-analysis.²⁸ Although women have a longer life expectancy and are thus more likely to get age-related diseases, both studies remark upon the sex predilection being larger than expected for age differences. Additionally, there are sex-specific differences in macrophage function, including activation, phagocytic capacity, and IL-10 production.²⁹ Interestingly, IL-10 is a pro-angiogenic cytokine.²⁵ Beta-AR blockade may influence a sex-specific inflammatory component of the multifactorial disease pathogenesis of AMD, possibly including IL-10 production. 

In conclusion, propranolol inhibits choroidal angiogenesis in both the ex vivo choroidal sprouting assay and the in vivo laser-induced CNV model (Figs. 1, 2). Furthermore, monocyte-derived macrophages are critical for beta-AR blockade inhibition of laser-induced CNV (Figs. 3, 4). Finally, we displayed that propranolol increased macrophage tissue infiltration after laser injury only in female mice (Figs. 8, 9). Our data suggest a mechanism where beta-AR antagonism increases monocyte-derived macrophage recruitment to the laser lesion, which is necessary for reduction in laser-induced CNV, implicating the anti-angiogenic nature of beta-AR blocker-recruited monocyte-derived macrophages. 

Acknowledgments

The authors thank David Kirchenbuechler for development of the automated angiogenesis area measurement software for the choroidal sprouting assay. 

Supported by the Global Ophthalmology Awards Program (GOAP), a Bayer-sponsored initiative committed to supporting ophthalmic research across the world. HP was supported by NIH Grant AR064546, HL134375, AG049665, UH2AR067687, the United States-Israel Binational Science Foundation (2013247), and the Rheumatology Research Foundation (Agmt 05/06/14). HP was also supported by the Mabel Greene Myers Professor of Medicine and generous donations to the Rheumatology Precision Medicine Fund. 

Disclosure: S. Droho, None; C.M. Cuda, None; H. Perlman, None; J.A. Lavine, None