Adrb2 Expression in Ocular-Infiltrating Macrophages Is Necessary for Interleukin-6 Expression and Choroidal Neovascularization

Abstract

Purpose: Effective therapies for treatment resistant neovascular age-related macular degeneration (nAMD) remain an unmet need. Beta-adrenergic receptor (AR) blockers can decrease laser-induced choroidal neovascularization (CNV) size in mice. We have shown that monocyte-derived macrophages (MDMs) and interleukin-6 (IL-6) are necessary for beta-AR blockers to inhibit CNV. However, the specific beta-AR and the mechanism of this pathway are not fully elucidated. We hypothesized that beta2-AR (Adrb2) signaling on MDMs increases IL-6 production and stimulates CNV.

Methods: Previously published single-cell RNA-sequencing data was reanalyzed to determine which mononuclear phagocytes express beta-ARs. Adrb2^flox/flox: Cx3cr1^CreER/+ mice (Adrb2^ΔMacs) or Adrb2^flox/flox (Adrb2^flox) controls were given tamoxifen injections at either four weeks before or at the time of laser-induced CNV to knockout Adrb2 in tissue resident or all macrophages, respectively. Mice underwent laser induced-CNV, and eyes were collected for choroidal wholemount immunofluorescence imaging to measure CNV area, multiparameter flow cytometry to analyze macrophage heterogeneity, and ELISAs to quantitate IL-6 levels.

Results: Adrb2 was the predominantly expressed beta-AR and was found on microglia, macrophages, and monocytes. Adrb2 deletion in tissue resident macrophages had no effect upon CNV area. Adrb2 deletion in all macrophages decreased CNV area by 1.4-fold. Adrb2^ΔMacs posterior eye cups demonstrated similar levels of pro-angiogenic CD11c⁺ macrophages compared to Adrb2^flox controls, but Ly6C^negCD11c^neg macrophages were significantly increased. IL-6 levels increased with laser in Adrb2^flox controls, but IL-6 levels in Adrb2^ΔMacs posterior eye cups were unchanged.

Conclusions: Beta2-AR deletion in ocular-infiltrating macrophages decreases laser-induced CNV area. Beta2-AR expression regulates IL-6 expression in monocyte-derived macrophages.

Age-related macular degeneration (AMD) is the leading cause of vision loss in the developed world. Although neovascular AMD (nAMD) is less prevalent than the dry form, it accounts for the majority of severe vision loss associated with the disease. The nAMD is characterized by choroidal neovascularization (CNV): the pathological growth of blood vessels into the subretinal space. The primary treatment for nAMD is currently anti-vascular endothelial growth factor (VEGF) therapy. Intravitreal anti-VEGF injections demonstrate significant efficacy in stabilizing or improving vision.^1,2 Despite these benefits, approximately 30% of patients do not demonstrate improved visual acuity with anti-VEGF therapy.³ These factors have prompted the need to identify and develop alternatives to anti-VEGF therapies. 

Beta-adrenergic receptor (AR) inhibitors (beta-blockers) were first recognized for their anti-angiogenic properties in the treatment of severe capillary hemangioma of infancy.⁴ This serendipitous finding has prompted investigation into the applications of beta-blockers for neovascular disease. Propranolol, a nonselective beta-blocker, inhibits endothelial cell tubulogenesis, proliferation, and migration in vitro.^5,6 These results were first studied in the setting of retinal disease using the oxygen-induced retinopathy mouse model. In this model, propranolol or genetic deletion of beta1-AR or beta2-AR reduces retinal neovascularization and leakage.^7–9 On the basis of these results, we and others have applied these findings to the laser-induced CNV mouse model of nAMD, where propranolol either systemically or locally inhibits CNV.^10–14 

The aforementioned basic science studies, relative safety of beta-blockers, and success in treating severe hemangioma of infancy have prompted research into the utility of beta-blockers for treating human retinal neovascular diseases. In retinopathy of prematurity (ROP), propranolol reduces the risk of ROP stage progression.^15–19 In diabetic retinopathy, beta-blocker treatment may be associated with a decreased likelihood of requiring laser therapy or anti-VEGF agents, but propranolol has no effect on neovascularization in proliferative diabetic retinopathy patients.^20,21 In nAMD, neoadjuvant topical dorzolamide-timolol reduces persistent retinal fluid in treatment-resistant nAMD patients receiving maximal anti-VEGF therapy.^22,23 The mechanism is unknown but may include the effects of beta-blockers, reduced aqueous humor production causing increased retention of anti-VEGF agents, or the effects of carbonic anhydrase inhibitors. Nevertheless, these findings highlight the potential of beta-blockers as a therapeutic approach for managing neovascular retinal disease, with the most compelling basic and clinical evidence in the context of nAMD. 

In the laser-induced CNV mouse model, we have previously shown that propranolol and beta2-AR blockers inhibit CNV.^10,13 In retinal cell lines, we have shown that beta2-AR antagonists and agonists regulate interleukin-6 (IL-6) production.^10,13 Furthermore, we have shown that classical monocyte-derived macrophages (MDMs) and IL-6 are necessary for propranolol to inhibit CNV.^12,24 Finally, we showed that macrophages produce IL-6 and that IL-6 is necessary for CNV.²⁴ These prior results suggest that beta2-AR expression on MDMs regulates IL-6 production to drive CNV. To test this hypothesis, we first sought to genetically confirm that beta2-AR expression in MDMs is required for laser-induced CNV. We then explored how beta2-AR deletion impacts MDM heterogeneity. Finally, we measured IL-6 levels in mice lacking beta2-AR expression in MDMs. Our findings reveal that MDM-specific beta2-AR deletion reduced CNV, promoted the recruitment of potentially anti-angiogenic macrophages, and decreased IL-6 levels. 

Methods

Sex as a Biological Variable

All studies were carried out on female mice because we previously showed that propranolol (a beta-adrenergic receptor blocker) has no effect on laser-induced CNV area in male mice.¹² 

Animals

Adrb2 ^flox mice were a generous gift from Gerard Karsenty.²⁵ Cx3cr1^CreER (Strain 020940) and C57BL/6J (Strain 000664) mice were obtained from Jackson Labs (Bar Harbor, ME) and crossed to generate Adrb2^flox/flox: Cx3cr1^CreER/+ mice (Adrb2^ΔMacs) or Adrb2^flox/flox (Adrb2^flox) controls for experiments. Genotypes of all mice were confirmed by Transnetyx (Cordova, TN). All experiments were performed on 10- to 12-week-old female mice. Mice were housed and bred in a pathogen-free barrier facility in the Center for Comparative Medicine at Northwestern University (Chicago, IL, USA). 

Tamoxifen Administration

Tamoxifen (20 mg/mL, T5648; Sigma-Aldrich, St. Louis, MO, USA) was prepared by dissolving tamoxifen powder in corn oil (C8267; Sigma-Aldrich) with shaking at 37°C overnight. Tamoxifen solutions were stored in the dark at room temperature for less than one week. Adrb2^ΔMacs and Adrb2^flox controls were given two intraperitoneal tamoxifen (100 mg/kg body weight) injections separated by 48 hours at either six or 10 weeks of age to knockout Adrb2 in tissue resident or all monocytes/macrophages, respectively. 

Laser-Induced CNV

Female 10- to 12-week-old mice were treated as previously described.¹² Briefly, mice were anesthetized with a ketamine/xylazine (Akorn, Lake Forest, IL, USA) cocktail. A 1 mg/kg subcutaneous injection of meloxicam (Henry Schein Animal Health, Melville, NY, USA) was given for pain control and hydration. Eyes were anesthetized and 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 and ELISA; to increase inflammatory cell numbers) focal burns (75 µm, 120–180 mW, 100 ms) were administered to each eye using a 532 nm argon ophthalmic laser (IRIDEX, Mountain View, CA, USA) via a slit lamp delivery system (Zeiss, Oberkochen, Germany). Mice were given daily intraperitoneal PBS injections after laser to induce stress and increase endogenous (nor)epinephrine levels if indicated in the schematics. 

Choroidal Sprouting Assay

Wildtype female mice were euthanized for the ex vivo choroidal sprouting assay using peripheral choroid as previously described.^12,26,27 Bone marrow-derived monocytes were isolated from Adrb2^flox control and Adrb2^ΔMacs female mice as previously described using a negative selection magnetic bead protocol from Miltenyi Biotec (no. 130-100-629; Auburn, CA, USA). We added 20,000 monocytes per well in EGM2-MV medium on Day 2. Pictures were taken on a Nikon Ti2 widefield microscope (Buffalo Grove, IL, USA) using a 10x objective and Nikon NIS Elements software. Images were analyzed with the Nikon Elements General Analysis as previously described.^12,27 

Immunofluorescence

Mice were sacrificed on day 14 for immunofluorescence imaging and CNV quantification. Eyes were processed for choroidal whole mounts as previously described.¹² Briefly, enucleated eyes were dissected in Tris-buffered saline solution (TBS) to isolate the posterior eye cup of RPE, choroid, and sclera. Cups were fixed in 4% paraformaldehyde (no. 15,713-S; Electron Microscopy Sciences, Hatfield, PA, USA) for one hour at room temperature and then washed in TBS. Cups were then blocked overnight at 4°C in TBS with 5% donkey serum (S30; Sigma-Aldrich), 2.5% bovine serum albumin (A2153; Sigma-Aldrich), and 0.5% Triton X-100 (X100; Sigma-Aldrich). Primary incubations were performed overnight at 4°C (Table). Next, cups were washed with TBS-T (TBS with 0.5% Tween-20; 00777; Amresco, Solon, OH, USA) six times and incubated with secondary antibodies overnight at 4°C (Table). Cups were then washed with TBS-T and mounted on HistoBond microscope slides (16004–406, VWR, Batavia, IL) with Immu-Mount (9990402, Thermo Fisher Scientific, Waltham, MA, USA). Imaging was performed on a Nikon Ti2 Widefield using Nikon NIS Elements software (Nikon, Melville, NY, USA). Area was analyzed using ImageJ after investigators were masked to animal identities. 

Table.

 

Antibodies Used in Flow Cytometry and Immunofluorescence

 

Flow Cytometry

Mice were sacrificed on day 3 after laser. Enucleated eyes were placed into Hanks’ balanced salt solution. Eyes were cleaned of the cornea, iris, ciliary body, lens, optic nerve, and conjunctiva until only the posterior eye cup consisting of retina, RPE, choroid, and sclera remained. Posterior eye cups were digested into single cell suspensions and processed for flow cytometry as previously described²⁸ with a few modifications outlined below. Eyes from each mouse were combined to create one sample per mouse. Posterior eye cups were cut into four pieces and digested at 180 RPM, 37°C for one hour without additional mechanical dissociation. Eye pieces were passed through a 40 µm filter to obtain a single cell suspension, then stained for live cells, blocked, and stained with fluorochrome-conjugated antibodies (Table). Samples were run with count beads (01-1234-42; Thermo Fisher Scientific) on a BD FACS Symphony A5-Laser Analyzer (Becton Dickinson, Franklin Lakes, NJ, USA) at the Northwestern University RHLCCC Flow Cytometry Core Facility and analyzed using FlowJo version 10 (FlowJo, Ashland, OR, USA). 

Bone marrow was isolated as previously described.¹² After passing through a 40 µm filter, bone marrow was selected for monocytes as described above or untreated. Samples were then stained for live cells, blocked, and stained with fluorochrome-conjugated antibodies (Table). Samples were then run on a BD FACS Symphony A5-Laser Analyzer at the Northwestern University RHLCCC Flow Cytometry Core Facility and analyzed using FlowJo version 10. 

ELISA

Mice were sacrificed on day 3 after laser. Experiments were performed as previously described²⁴ with a few below modifications. Briefly, enucleated eyes were placed in cold PBS with protease inhibitor (87786; Thermo Fisher Scientific). Posterior eye cups were dissected in cold PBS and a 2 mm hole punch (21909-132; VWR, Batavia, IL, USA) centered around the optic nerve was used to obtain equal sized pieces of posterior eye cups between groups including all laser spots. Supernatants of homogenized samples were harvested for analysis via endpoint DuoSet ELISA for IL6 (DY406; R&D Systems, Minneapolis, MN, USA) using manufacturer's instructions. Plates were read (Bio-Rad, Hercules, CA, USA) and standard curves and quantification of samples were conducted using the simple linear regression function in Prism (Graphpad Software LLC). Results for each run were normalized to control wildtype mice. 

Bioinformatics

We analyzed our own prior scRNA-seq data (GSE239941 and GSE222094)^29,30 using Seurat v4.³¹ We used the same quality control metrics as prior reports.^29,30 Data were independently normalized, scaled, and then integrated using reciprocal principal component analysis.³² Clustering was performed with the standard workflow, including 23 principal components at a resolution of 0.4. After creating a mononuclear phagocyte subset, the data was again scaled, normalized, and clustered using 21 principal components and 0.4 resolution. Canonical markers and the DotPlot function were used to identify cell types. FeaturePlots were used to visualize Adrb1, Adrb2, and Adrb3 expression. Specific code is available on reasonable request. 

Statistics

Data normality was investigated using the Shapiro-Wilk test. Laser-induced CNV area was compared using Student's unpaired two-tailed t-test. Bone marrow enrichment and choroidal sprouting angiogenesis were analyzed by two-way ANOVA with Sidak's multiple comparisons test. Flow cytometry data was analyzed using Brown-Forsythe and Welch ANOVA followed by Dunnett's T3 multiple comparisons test. ELISA data was compared using one-way ANOVA followed by Tukey's multiple comparisons test. 

Study Approval

All procedures were approved by the Northwestern University Institutional Animal Care and Use Committee and adhered to the ARVO Statement for Animal Use in Ophthalmic and Vision Research. 

Data Availability

Raw scRNA-seq data was deposited on the GEO omnibus (GSE239941, GSE222094). All other data will be available on reasonable request to the corresponding author. 

Results

To investigate Adrb2 (beta2-AR) expressing cells, we re-analyzed our prior scRNA-seq data.^29,30 This dataset included cells from wildtype and Ccr2^−/− mouse eyes with and without laser injury,²⁹ and wildtype and Nr4a1^−/− mouse retina/choroid complex with and without laser injury.³⁰ We reintegrated and then reclustered mononuclear phagocytes to identify three Tmem119⁺P2ry12⁺ microglia clusters (Microglia-1, Microglia-2, and Microglia-3), Gpnmb⁺Cst7⁺ disease–associated microglia, Cd14⁺Lgals3⁺Spp1⁺Vegfa⁺ MDMs, 3 Aif1⁺C1qa⁺Cd68⁺Mrc1⁺Ms4a7⁺ macrophage populations (Mac-1, Mac-2, and Mac-3), Ccr2⁺Ly6c2⁺ classical monocytes (C-Mono), Spn⁺Ace⁺ nonclassical monocytes (NCM), four Flt3⁺ dendritic cell clusters (cDC-1, c-DC2, migDC, pDC), and Kit⁺Fcer1a⁺ mast cells (Figs. 1a, 1b). The data were well integrated, and MDMs were increased by laser in each genotype except for Ccr2^−/− eyes, which was expected since these mice lack classical monocytes and classical MDMs (Fig. 1c). Adrb3 was not detectable in any mononuclear phagocyte population. Adrb1 was expressed at low levels, mostly in Tmem119⁺ microglia (Fig. 1d). Adrb2 was expressed in Tmem119⁺ microglia, a subset of Mrc1⁺ macrophages, MDMs, and Ly6c2⁺ monocytes (Fig. 1d). These data suggest that Adrb2 is the predominant beta-adrenergic receptor expressed in mononuclear phagocytes with detectable expression on microglia, macrophages, monocytes, and MDMs. 

Figure 1.

 

Adrb2 is expressed in multiple macrophage subtypes. (A) UMAP of mononuclear phagocyte clusters. (B) Dot plot of cluster-defining genes. (C) UMAP split by experiment shows good integration between studies and increased MDMs in laser-treated eyes. Light red circles delineate the MDM group. (D) Feature plots of Adrb1, Adrb2, and cluster defining genes showing that Adrb2 is the most abundant betaAR and that it is expressed in multiple macrophage clusters.

We previously showed that both pan beta-AR blockade with propranolol and specific beta2-AR blockade reduced laser-induced CNV area in female mice only.^10,13 Furthermore, we also demonstrated that propranolol did not effectively reduce laser-induced CNV in Ccr2^−/− mice, suggesting the importance of the beta2-AR in classical monocytes and classical MDMs.¹² However, the exact Adrb2-expressing mononuclear phagocytes that are necessary for laser-induced CNV remain uninvestigated. We crossed Adrb2^flox/flox with Cx3cr1^CreER/CreER mice to generate Adrb2^flox/flox: Cx3cr1^CreER/+ mice (Adrb2^ΔMacs) and Adrb2^flox/flox (Adrb2^flox) controls. We administered tamoxifen to six-week-old Adrb2^ΔMacs and Adrb2^flox control female mice, then subjected them to laser-induced CNV at 10-12 weeks of age (Fig. 2a). Our prior studies using Cx3cr1^CreER/+: Rosa26^zsGreen mice demonstrated that four weeks after tamoxifen, >95% of microglia, and 40% to 80% of other macrophage populations are zsGreen⁺, whereas blood monocytes and MDMs are zsGreen^neg.³³ Thus this treatment regimen results in the deletion of Adrb2 in primarily microglia and other tissue resident macrophages without affecting monocytes and MDMs. Mice underwent daily intraperitoneal PBS after laser to induce stress and increase endogenous catecholamine levels. We found that Adrb2^ΔMacs demonstrated comparable CNV area to Adrb2^flox controls (Figs. 2b–d). These findings suggest that Adrb2 expression in microglia and ocular tissue resident macrophages is not necessary for CNV. 

Figure 2.

 

Adrb2 deletion in tissue resident macrophages has no effect upon CNV area. (A) Schematic representation of experimental plan. (B, C) Representative immunofluorescence images of choroidal wholemounts from Adrb2^flox/flox (Adrb2^flox) and Adrb2^flox/flox: Cx3cr1^CreER/+ mice (Adrb2^ΔMacs) mice. (D) Adrb2^ΔMacs mice have equivalent CNV area to controls (N = 11–14 mice per group, Student's unpaired two-tailed t-test).

On the basis of these data, we hypothesized that Adrb2 expression in MDMs is necessary for CNV. We tested this hypothesis by subjecting Adrb2^ΔMacs and Adrb2^flox control female mice to laser injury at 10 weeks of age with tamoxifen treatment two days before and at the time of laser injury (Fig. 3a). Our previous data using Cx3cr1^CreER/+: Rosa26^zsGreen mice showed that at one week after tamoxifen, 75% of monocytes and 80% to 100% of ocular macrophages were zsGreen⁺.³³ Therefore tamoxifen administration at this stage will delete Adrb2 in the majority of monocytes, MDMs, and ocular macrophages. Similar to above, mice underwent daily intraperitoneal PBS after laser to increase endogenous catecholamine levels. We found that Adrb2^ΔMacs displayed a 1.4-fold reduction in CNV area compared to Adrb2^flox controls (P < 0.05, Figs. 3b–d). Because Adrb2 deletion in tissue resident macrophages had no effect on CNV (Fig. 2), whereas Adrb2 deletion in all macrophages reduced CNV area (Fig. 3), these data suggest that deletion of Adrb2 in MDMs decreased CNV area. Next, we repeated the above study without daily intraperitoneal PBS injections (Fig. 3e). We found that without daily PBS injections, Adrb2^ΔMacs and Adrb2^flox control mice demonstrated similar CNV areas (Figs. 3f–h). 

Figure 3.

 

Adrb2 deletion in all macrophages decreases CNV area. (A) Schematic representation of experimental plan. (B, C) Representative immunofluorescence images of choroidal wholemounts from Adrb2^flox/flox (Adrb2^flox) and Adrb2^flox/flox: Cx3cr1^CreER/+ mice (Adrb2^ΔMacs) mice. (D) Adrb2^ΔMacs mice have decreased CNV area to controls (N = 14 mice per group, Student's unpaired two-tailed t-test, *P < 0.05). (E) Schematic representation of experimental plan without PBS injections. (F, G) Representative immunofluorescence images of choroidal wholemounts from Adrb2^flox and Adrb2^ΔMacs eyes. (H) Adrb2^ΔMacs mice have similar CNV area to controls (N = 10–13 mice per group, Student's unpaired two-tailed t-test).

To confirm that Adrb2^ΔMacs mice had a reduction in beta2-AR expression in macrophages, we performed immunofluorescence imaging on day 3 after laser, the peak of macrophage infiltration.³³ In Adrb2^flox eyes, we were able to observe co-expression of beta2-AR and IBA1 in the center of lesions (Fig. 4A, white arrows). Interestingly, it was recently reported that Ccr2⁺F4/80⁺ macrophages primarily occupy the center of CNV lesions in the two-hit laser CNV fibrosis model.³⁴ In Adrb2^ΔMacs eyes, little to no beta2-AR expression was detected in the center of the lesion (Fig. 4B). However, both genotypes displayed beta2-AR expression in the periphery of the lesion, which was negative for IBA1 staining, suggesting a potential stromal source of beta2-AR expression in CNV lesion. 

Figure 4.

 

Adrb2 deletion in all macrophages decreases the number of beta2-AR+ macrophages in CNV lesions from Adrb2^flox (A) and Adrb2^ΔMacs (B) mice. IBA1 (macrophage, green) and beta-AR (magenta) immunofluorescence imaging of choroidal wholemounts on Day 3 after laser injury. Arrows indicate IBA1 and beta2-AR co-localization. No double positive cells were observed in Adrb2^ΔMacs mice.

To test the effect of beta2-AR expressing monocytes/macrophages in a second model, we performed the ex vivo choroidal sprouting assay. Choroidal explants were dissected from female wildtype mice. On Day 2, bone marrow-derived monocytes were isolated from tamoxifen-treated Adrb2^ΔMacs and Adrb2^flox control female mice. Flow cytometry was used to confirm that bone marrow was enriched from 5.1% to 56.0% for classical monocytes after the enrichment process (P < 0.001, Supplementary Fig. S1). We added 20,000 monocytes per well and quantitated choroidal sprouting angiogenesis on Days 5-7. We found no significant difference in choroidal angiogenesis between Adrb2^ΔMacs and Adrb2^flox groups (Supplementary Fig. S2). These data suggest important differences between the ex vivo choroidal sprouting assay and the in vivo laser-induced CNV model. 

We next performed multi-parameter flow cytometry to explore how Adrb2 deletion in MDMs impacted macrophage heterogeneity. We subjected 10- to 12-week-old Adrb2^ΔMacs and Adrb2^flox mice to tamoxifen treatment two days before and at the time of laser injury. Eyes were collected on day 3 post-laser and posterior eye cups (retina and choroid-RPE-sclera complex) were isolated and processed for multi-parameter flow cytometry analysis. After multiplet exclusion, live cells were gated forward (Fig. 5a, top left). From live, single cells, CD45⁺ cells were identified (Fig. 5a, bottom left). We captured mononuclear phagocytes by CD11b⁺Lineage^neg staining, excluding B cells (B220), T cells (CD4, CD8), eosinophils (SiglecF), neutrophils (Ly6G), and NK cells (NK1.1) in the Lineage cocktail (Fig. 5a, middle top). CD64 staining was used to identify CD64⁺ macrophages (Fig. 5a, middle bottom). From CD64⁺ cells, microglia were identified as Cx3cr1^highCD45^dim (Fig. 5a, top right). CD64⁺ cells that were not identified as microglia were delineated into Ly6C⁺CD11c^neg, Ly6C⁺CD11c⁺, Ly6C^negCD11c⁺, or Ly6C^negCD11c^neg macrophages (Fig. 5a, bottom right). Microglia numbers were not increased by laser in Adrb2^flox controls, but had a 1.5-fold (P < 0.05) increase by laser in Adrb2^ΔMacs mice (Fig. 5b). We found no significant change in number of Ly6C⁺CD11c^neg macrophages by laser in either group (Fig. 5c). In contrast, laser injury increased Ly6C⁺CD11c⁺ macrophage numbers in Adrb2^flox controls by 4.7-fold (P < 0.01) and Adrb2^ΔMacs mice by 3.2-fold (P < 0.05), with no significant difference between laser groups (Fig. 5d). Similarly, Ly6C^negCD11c⁺ macrophage numbers were increased with laser by 2.4-fold (P < 0.01) in Adrb2^flox controls and 1.8-fold (P < 0.05) in Adrb2^ΔMacs mice with no significant difference between laser groups (Fig. 5e). Finally, the number of Ly6C^negCD11c^neg macrophages were both increased with laser in Adrb2^flox controls and Adrb2^ΔMacs mice (P < 0.05 for both, Fig. 5f). Interestingly, laser-treated Adrb2^ΔMacs posterior eye cups showed 1.4-fold greater numbers of Ly6C^negCD11c^neg macrophages compared to laser-treated Adrb2^flox posterior eye cups (P < 0.05, Fig. 5f). Since we previously showed that CD11c⁺ macrophages are pro-angiogenic,^29,30 these results suggest that macrophage-specific deletion of Adrb2 leads to the recruitment of more Ly6C^negCD11c^neg macrophages, which we hypothesize could be anti-angiogenic. 

Figure 5.

 

Adrb2 ^ΔMacs posterior eye cups show greater numbers of Ly6C^negCD11c^neg macrophages. (A) Flow cytometry gating strategy, see the results text for a full description. (B–F) Numbers of different macrophage populations in Adrb2^flox and Adrb2^ΔMacs posterior eye segments with and without laser (N = 7–9 mice per group, Brown-Forsythe and Welch ANOVA followed by Dunnett's T3 multiple comparisons test). *P < 0.05, **P < 0.01.

Because IL-6 is necessary for laser-induced CNV,^24,35 beta-AR blockers reduce IL-6 levels,¹³ and beta-AR blockers are ineffective at decreasing CNV area in IL-6 knockout mice,²⁴ we hypothesized that IL-6 is downstream of the beta2-AR and that IL-6 would be decreased in Adrb2^ΔMacs mice. To test this hypothesis, we measured IL-6 by ELISA in control wildtype, lasered wildtype, lasered Adrb2^flox, and lasered Adrb2^ΔMacs posterior eye cups on day 3 after laser injury. Adrb2^flox and Adrb2^ΔMacs mice received tamoxifen two days before and on the day of laser injury. Compared to control wildtype mice, IL-6 expression was increased by 1.2-fold in both lasered wildtype mice (P < 0.001) and Adrb2^flox posterior eye cups (P < 0.001, Fig. 6a). Adrb2^ΔMacs posterior eye cups were not significantly elevated compared to control (Fig. 6a). After combining the two equivalent groups, lasered wildtype and Adrb2^flox mice, into a single “laser” group, we found that IL-6 levels were elevated by 1.2-fold (P < 0.0001) in lasered posterior eye cups compared to control, whereas IL-6 levels were significantly decreased by 12.5% (P < 0.05) in Adrb2^ΔMacs posterior eye cups compared to the laser group (Fig. 6b). These results suggest that Adrb2 expression in macrophages is necessary for increased IL-6 levels in laser-induced CNV. 

Figure 6.

 

Adrb2 deletion in all macrophages reduces IL-6 levels. (A) IL-6 expression levels normalized to control from wildtype control, wildtype laser, Adrb2^flox laser, and Adrb2^ΔMacs laser posterior eye segments (N = 9–16 mice per group, one-way ANOVA followed by Tukey's multiple comparisons test). (B) Same data as a except analyzed with wildtype laser and Adrb2^flox laser combined since Adrb2^flox mice are on a C57BL/6J background. FC = fold change. *P < 0.05, ***P < 0.001, ****P < 0.0001. (C, D) Model of beta2-AR signaling during CNV in control and Adrb2^ΔMacs mice. DN, double negative (Ly6C^negCD11c^neg).

Discussion

In summary, we found that Adrb2 is expressed in multiple macrophage subtypes (Fig. 1), its deletion in tissue resident macrophages had no effect upon laser-induced CNV area (Fig. 2), whereas Adrb2 deletion in all macrophages decreased CNV area (Fig. 3) in female mice. Furthermore, Adrb2 deletion in all macrophages increased Ly6C^negCD11c^neg macrophages without affecting pro-angiogenic CD11c⁺ macrophages during laser CNV (Fig. 5). Finally, Adrb2 deletion in all macrophages prevented IL-6 up-regulation. Based upon these data and our prior results, we present a model where (nor)epinephrine binds the beta2-AR on MDMs (Fig. 6c). This leads to elevated IL-6 levels, which increases CNV size. Further, beta2-AR signaling on MDMs stimulates the recruitment of pro-angiogenic CD11c⁺ macrophages, which increase angiogenesis through VEGF-A, and potentially affects anti-angiogenic Ly6C^negCD11c^neg macrophages (Fig. 6c). In Adrb2^ΔMacs mice, deletion of the beta2-AR on MDMs has no effect on pro-angiogenic CD11c⁺ macrophages, but decreases IL-6 levels and increases anti-angiogenic Ly6C^negCD11c^neg macrophages, resulting in reduced angiogenesis and diminished CNV area (Fig. 6d). 

Our findings demonstrate that Adrb2 expression in MDMs, and not in microglia, plays a critical role in CNV. This observation aligns with previous research conducted by our group and others, and supports the growing body of evidence highlighting the role of infiltrating MDMs in pathological angiogenesis. For example, Ccr2^−/− mice, which lack classical MDMs, show reduced CNV size compared to wildtype mice.^33,36,37 Building upon these findings, our study reveals that Adrb2 is necessary for MDM-driven CNV, despite microglia exhibiting higher Adrb2 expression (Fig. 1). This discrepancy raises questions about the specific role of beta2-AR in microglia, which remains unknown, and warrants further investigation. 

Previous studies have highlighted the role of specific macrophage populations in regulating CNV, including the existence of anti-angiogenic ocular macrophages. For example, intravitreal injections of young splenic or M1-polarized macrophages reduces laser-induced CNV area.^38,39 Additionally, IL-10 knockout mice show decreased CNV area and increased macrophage numbers,⁴⁰ supporting the existence of anti-angiogenic macrophage subtypes. FurtherMORE, beta2-AR signaling has been recognized as a key modulator in myeloid cells, influencing macrophage recruitment and polarization.⁴¹ Therefore lack of beta2-AR signaling in MDMs shifts macrophage heterogeneity toward more Ly6C^negCD11c^neg macrophages, a population we propose to be anti-angiogenic (Fig. 5). In agreement, we previously showed that propranolol increased the number of MHCII^neg macrophages,¹² which could be the same macrophage subtype. Because classical monocytes express the IL-6 receptor,²⁴ it is possible that reduced IL-6 levels influence macrophage heterogeneity during laser-induced CNV. Future studies will investigate if beta2-AR signaling, IL-6 receptor signaling, or both pathways are responsible for greater number of Ly6C^negCD11c^neg macrophages. 

We found that deletion of the Adrb2 in MDMs decreased IL-6 levels. Because IL-6 knockout mice have reduced CNV area,^24,35 this is likely a key mechanism by which beta2-AR signaling promotes CNV. The mechanism by which beta2-AR activation increases IL-6 expression has been well investigated in cell lines such as cardiac fibroblasts. Activation of beta2-AR by its endogenous ligands, norepinephrine and/or epinephrine, increases cAMP levels, activates protein kinase A (PKA), phosphorylates transcription factors such as cAMP response element-binding protein (CREB), and enables CREB to drive Il6 transcription.⁴² In addition to the cAMP-PKA-CREB pathway, beta2-AR signaling can also activate the p38 mitogen-activated protein kinase pathway through a PKA-independent mechanism, which increases IL-6 expression.^43,44 These pathways have been corroborated in astrocytes, macrophages, and in vivo using identical mice and particulate matter-induced IL-6 production by alveolar macrophages.^45–47 By deleting Adrb2 in MDMs, these transcriptional mechanisms are likely disrupted, thus reducing IL-6 levels and CNV size. 

We previously showed that propranolol had no effect on CNV size in male wildtype mice.¹² Therefore we would expect that conditional deletion of Adrb2 in macrophages would also have no effect on CNV size in male mice. We hypothesize that because female mice have more MDM infiltration after laser compared to male mice³³ and the fact that we previously showed that MDMs are necessary for beta2-AR and propranolol dependent reduction in CNV size¹² that greater MDM numbers are the underlying reason for this sex-specific effect. Nevertheless, the fact that we did not confirm the absence of CNV reduction in male Adrb2^ΔMacs mice is a limitation of this study. 

There are several additional limitations to our study. First, the tamoxifen induction schemas we used in our Adrb2^flox/flox: Cx3cr1^CreER/+ mice resulted in the deletion of Adrb2 in either tissue-resident macrophages alone or in all ocular macrophages, including tissue-resident macrophages, monocytes, and MDMs. A more precise approach would have been to use Adrb2^flox/flox: Ccr2^CreER/+ mice, which could specifically target classical monocytes and MDMs, ensuring that Adrb2 deletion is restricted to these cell populations. Second, while we found an increase in recruitment of Ly6C^negCD11c^neg macrophages in Adrb2^ΔMacs mice, which also exhibited reduced CNV size compared to Adrb2^flox controls, we can only infer that they are anti-angiogenic based off of this observation. Dedicated studies are needed to confirm whether Ly6C^negCD11c^neg macrophages indeed have anti-angiogenic properties and to elucidate the relationship between this macrophage subset and IL-6 expression. Third, the potential translatability of these findings is limited by the fact that daily intraperitoneal PBS injections are required for reduced CNV in Adrb2^ΔMacs mice (Fig. 3). Evidence exists that surgical sympathetic nerve transection inhibits CNV size,⁴⁸ potentially because catecholamines facilitate VEGF-dependent angiogenesis.⁴⁹ Although similar data is absent in human patients, it is reassuring that adjuvant dorzolamide-timolol reduces intravitreal injection burden compared to placebo in patients with treatment-resistant neovascular AMD.²² Fourth, the potential translatability is also reduced by the fact that we could not replicate our laser CNV results in the choroidal sprouting assay (Supplementary Fig. S2). However, it should be noted that the choroidal sprouting assay is ex vivo and a worse model than the in vivo laser-induced CNV. We hypothesize the choroidal sprouting assay failed to replicate our laser CNV result because the choroidal microenvironment is necessary compared to Matrigel or adequate beta2-AR agonism was lacking. A second potential in vivo model is the Vldlr^−/− mouse, where spontaneous CNV is observed. However, we recently published that MDMs are dispensable in the Vldlr^−/− CNV model,⁵⁰ making it not suitable to test our hypothesis. Finally, to achieve a statistically significant difference in IL-6 levels between Adrb2^flox and Adrb2^ΔMacs posterior eye cups, it was necessary to combine lasered Adrb2^flox and lasered wildtype mice into a single control group. Nevertheless, we did find that IL-6 levels in lasered Adrb2^flox mice were increased compared to unlasered mice, whereas Adrb2^ΔMacs posterior eye cups showed no difference compared to unlasered mice. 

In summary, we demonstrated that while Adrb2 deletion in tissue-resident macrophages does not affect CNV, its deletion in all macrophages leads to a significant reduction in CNV, highlighting the critical role of beta2-AR signaling in MDM-mediated CNV. Additionally, in mice with Adrb2 deleted in all macrophages, we observed decreased IL-6 levels and an increase in Ly6C^negCD11c^neg macrophages. Future studies will focus on selective deletion of Adrb2 in all MDMs to better determine their specific contribution to CNV and further investigate the potential mechanisms by which Ly6C^negCD11c^neg macrophages regulate angiogenesis. 

Acknowledgments

Supported by NIH grant K08 EY030923, the Research to Prevent Blindness Medical Student Eye Research Fellowship, and an Unrestricted Departmental Grant from Research to Prevent Blindness. JAL was additionally supported by R01 EY034486 and the Research to Prevent Blindness Sybil B. Harrington Career Development Award for Macular Degeneration. Imaging work was performed at the Northwestern University Center for Advanced Microscopy supported by NCI P30 CA060553. Flow cytometry was performed at the Northwestern University - Flow Cytometry Core Facility supported by Cancer Center Support Grant (NCI P30 CA060553). No funding body had any role in the design of the study, collection, analysis, interpretation of data, or in writing the manuscript. 

Disclosure: J. Gong, None; K.S. Chan, None; A. Rajesh, None; S. Droho, None; J.A. Lavine, Line 6 Biotechnology (C), Genentech (C), Therini Bio (R)