Neuropilin-1 mediates myeloid cell chemoattraction and influences retinal neuroimmune crosstalk

Immunological activity in the CNS is largely dependent on an innate immune response and is heightened in diseases, such as diabetic retinopathy, multiple sclerosis, amyotrophic lateral sclerosis, and Alzheimer’s disease. The molecular dynamics governing immune cell recruitment to sites of injury and disease in the CNS during sterile inflammation remain poorly defined. Here, we identified a subset of mononuclear phagocytes (MPs) that responds to local chemotactic cues that are conserved among central neurons, vessels, and immune cells. Patients suffering from late-stage proliferative diabetic retinopathy (PDR) had elevated vitreous semaphorin 3A (SEMA3A). Using a murine model, we found that SEMA3A acts as a potent attractant for neuropilin-1–positive (NRP-1–positive) MPs. These proangiogenic MPs were selectively recruited to sites of pathological neovascularization in response to locally produced SEMA3A as well as VEGF. NRP-1–positive MPs were essential for disease progression, as NRP-1–deficient MPs failed to enter the retina in a murine model of oxygen-induced retinopathy (OIR), a proxy for PDR. OIR mice with NRP-1–deficient MPs exhibited decreased vascular degeneration and diminished pathological preretinal neovascularization. Intravitreal administration of a NRP-1–derived trap effectively mimicked the therapeutic benefits observed in mice lacking NRP-1–expressing MPs. Our findings indicate that NRP-1 is an obligate receptor for MP chemotaxis, bridging neural ischemia to an innate immune response in neovascular retinal disease.

The CNS had long been considered an immune-privileged system, yet it is now clear that the brain, retina, and spinal cord are subjected to complex immune surveillance (1, 2). This is evident in the retina, in which an intensified, largely microglial/macrophage-based immune response is associated with the progression of several sight-threatening diseases, such as diabetic retinopathy (DR) (3–5), age-related macular degeneration (AMD) (6–8), and retinopathy of prematurity (ROP) (9, 10). Together, these retinal diseases account for the principal causes of loss of sight in industrialized countries (6, 11, 12).

A common etiology for blinding vasoproliferative retinopathies such as DR and ROP is the initial degeneration of retinal microvasculature, followed by ischemic stress on the neuroretina. This triggers a second phase of deregulated and destructive angiogenesis (13). Given this sequence of events and prominent clinical features, the most widely used, current local ocular therapeutic interventions directly target pathological neovascularization yet present a number of nondesirable off-target effects. For example, front-line treatments, such as laser photocoagulation, which ablate ischemic retinal tissue with the aim of reducing production of angiogenic factors, compromise visual field and provoke scotomas. Moreover, given the neurosupportive properties of VEGF, a growing number of neurological and developmental side effects are being attributed to long-term intravitreal anti-VEGFs regimens (4, 11, 14–16). Overcoming these therapeutic limitations or exploring novel pharmacological avenues is therefore required to ameliorate the safety profiles of current interventions. In this regard, there is increasing evidence for the breakdown of the neurovascular unit in proliferative retinopathies (4, 17, 18). In particular, retinal ganglion cells and neurons of the inner nuclear layer that are in intimate contact with degenerating vasculature in DR and ROP (Supplemental Figure 1; supplemental material available online with this article; doi:10.1172/JCI76492DS1) are subjected to various forms of cellular stress, including ER stress (19), oxidative stress (13), inflammatory stress (18–21), and increased microglial activity (22–25). Whether stressed or damaged central retinal neurons themselves have the propensity to modulate the immune response to reestablish local homeostasis, clear debris (2), or segregate irreparably damaged tissue (17) is unknown and could provide therapeutic insight for designing targeted inhibitors of sterile neuroinflammation.

While a role for mononuclear phagocytes (MPs), such as microglia/macrophages, has been established in proliferative retinopathies (22–25), the local tissue-specific molecular cues and mechanisms of chemoattraction by which these cells are called to partake in or respond to neuroretinal injury remain largely ill defined. We therefore addressed this question with the widely used mouse model of oxygen-induced retinopathy (OIR) that mimics the destructive vasoproliferative phase of ocular neovascular diseases, such as DR (26). A candidate receptor that could bridge neuronal stress to an immune response is neuropilin-1 (NRP-1). This single-pass transmembrane receptor, with a large 860–amino acid extracellular domain, has the unconventional ability to bind at least 2 structurally unrelated ligands (VEGF₁₆₅ and semaphorin 3A [SEMA3A]) via distinct sites (27–29). While predominantly recognized for its role in neuronal or vascular guidance and as a coreceptor for VEGF₁₆₅ (30), it has also been described on microglia during vascular anastomosis (31) and recently described in tumor-associated macrophages with roles in vessel maturation (32) and proangiogenic properties (33). Importantly, the physiological role of myeloid-resident NRP-1 remains unclear, particularly in the context of the CNS. We therefore sought to determine the function of myeloid-resident NRP-1 in the pathogenesis of proliferative retinopathies.

Here, we demonstrate that, via production of SEMA3A and VEGF, the ischemic neural retina has the inherent capacity to modulate the innate immune response by summoning circulating MPs to sites of vascular injury. We provide evidence in both human and animal studies for the critical role of myeloid-resident NRP-1 in the chemotaxis and accretion of MPs in proliferative retinopathies. Our findings further suggest that current anti-VEGF therapies may in part be effective by reducing destructive ocular inflammation.

NRP-1 identifies a population of MPs that are mobilized secondary to vascular injury. To determine whether MPs such as microglia or macrophages partake in the vascular pathogenesis associated with proliferative retinopathies, we first carried out FACS analysis on whole mouse retinas to elucidate the kinetics of macrophage/microglial accumulation throughout the evolution of OIR (Figure 1A) (75% oxygen from P7–P12 to induce vaso-obliteration and room air until P17 to attain maximal preretinal neovascularization) (refs. 26, 33, and Figure 1, B, E, and H). Our data revealed significantly higher numbers of retinal macrophage/microglial cells (Gr-1^–, F4/80⁺, CD11b⁺ cells) (gating scheme in Supplemental Figure 2) in OIR at all time points analyzed, including a 36% increase during the vaso-obliterative phase at P10 (P = 0.0004) (Figure 1C), a 63% rise during the neovascular phase at P14 (P < 0.0001) (Figure 1F), and a 172% surge during maximal neovascularization at P17 (P = 0.0006) (Figure 1I). These findings agree with the prevailing notion of microglial involvement in retinal neovascularization (22, 23, 34).

Figure 1

NRP-1 identifies a population of MPs that are mobilized secondary to vascular injury. (A) Schematic depiction of the mouse model of OIR. Representative FACS plots of Gr-1^–/F4/80⁺/CD11b⁺ cells (microglia) in retinas collected at (B) P10, (E) P14, and (H) P17 from WT OIR and normoxic (N) control mice. (C, F, and I) The number of retinal microglia was significantly increased in OIR at all points analyzed; n = 7–8 (normoxic and OIR) (total of 28–32 retinas per condition; each “n” comprises 4 retinas). (D, G, and J) A proportional increase in the number of NRP-1⁺ microglia was observed in OIR retinas; n = 3–5 (normoxic and OIR) (total of 12–20 retinas per condition; each “n” comprises 4 retinas). (K and L) To investigate the role of MP-resident NRP-1, we generated LysM-Cre Nrp1^fl/fl mice, which have significantly compromised Nrp1 expression in retinal microglia; n = 3 (WT), n = 4 (LysM-Cre Nrp1^fl/fl) (total of 12–16 retinas per condition). (M and O) FACS analysis at P10 and P14 during the proliferative phase of OIR reveals that MP-resident NRP-1 is essential for MP infiltration into the ischemic retina, (N and P) as LysM-Cre Nrp1^fl/fl mice did not show an increase in numbers of Gr-1^–/F4/80⁺/CD11b⁺ cells in OIR at these time points. (Q and R) At P17, MPs infiltrate independent of NRP-1; n = 7–8 (normoxic), n = 7–9 (OIR) (total of 28–36 retinas per condition; each “n” comprises 4 retinas). (S) MP accumulation in the retinas over the course of OIR in WT and LysM-Cre Nrp1^fl/fl mice. Data is expressed as fold change relative to control ± SEM. *P < 0.05, **P < 0.001, ***P < 0.0001.

Importantly, at each time point investigated, we observed a proportional increase in NRP-1⁺ MPs in OIR, with a rise of 37% at P10 (P = 0.0240) (Figure 1D), 61% at P14 (P = 0.0196) (Figure 1G), and 155% at P17 (P = 0.0058) (Figure 1J), suggesting that this subpopulation of NRP-1⁺ MPs was being recruited to the neuroretina during the progression of the disease. For all OIR experiments, weights of mouse pups were recorded (Supplemental Figure 3) to ascertain adequate metabolic health (35).

In order to establish the role of MP-resident NRP-1 in retinopathy, we generated a myeloid-specific knockout of Nrp1 by intercrossing Nrp1-floxed mice with LysM-Cre mice (36), yielding LysM-Cre Nrp1^fl/fl progeny. The resulting mice showed an approximately 80% decrease in NRP-1 expression in retinal MPs when compared with that in LysM-Cre Nrp1^+/+ littermate controls (P = 0.0004) (Figure 1, K and L). Of note, mice tested negative for the rd8 mutation of the Crb1 gene (37). LysM-Cre Nrp1^fl/fl mice did not show any difference in body weight, size, or open-field activity when compared with littermates throughout the period of experimentation (from P1–P17) (data not shown) and had similar numbers of resident retinal microglia (Supplemental Figure 4). Remarkably, deletion of NRP-1 on myeloid cells fully abrogated the entry of macrophages/microglia at P10 and P14 OIR (Figure 1, M–P), revealing the critical role for this receptor in MP chemotaxis during the early stages of ischemic retinal injury. At P17, following maximal pathological neovascularization, MP infiltration occurred largely independent of NPR-1 (Figure 1, Q–S). Consistent with a potential microglial identity, the NRP-1–expressing Gr-1^–/F4/80⁺/CD11b⁺ cells identified above expressed high levels of CX3CR1 and intermediate/low levels of CD45 (Figure 2A). A comparison with spleen-derived cells is provided in Supplemental Figure 5. As expected, in LysM-Cre Nrp1^fl/fl retinas, CX3CR1^hi/CD45^lo MPs were devoid of NRP-1 (Figure 2B).

Figure 2

Retinal NRP-1⁺ MPs express microglial markers. Representative FACS plots from P14 OIR retinas depicting that (A) 95% (in WT) and (B) 94.7% (in LysM-Cre Nrp1^fl/fl) of Gr-1^–/F4/80⁺/CD11b⁺ cells express high levels of CX3CR1 and intermediate/low levels of CD45 consistent with a microglial phenotype. CX3CR1^hi and CD45^lo cells (A) express NRP-1 in WT retinas and (B) do not express NRP-1 in retinas from LysM-Cre Nrp1^fl/fl mice.

NRP-1⁺ myeloid cells localize to sites of pathological neovascularization in the retina. Given the pronounced influx of NRP-1⁺ macrophage/microglia during OIR, we next sought to determine where these cells localize during the progression of disease. Immunofluorescence on retinal flat mounts revealed that NRP-1⁺ macrophage/microglia (colabeled with IBA1 and NRP-1) were intimately associated with nascent pathological tufts at P14 of OIR (Figure 3, A–C) as well as mature tufts at P17 of OIR (Figure 4, A–C). White arrows in Figure 3B and Figure 4B point to NRP-1⁺ MPs associated with preretinal tufts. NRP-1 was also expressed by endothelial cells on the endothelium of neovascular tufts as previously reported (21). Consistent with data presented in Figure 1, LysM-Cre Nrp1^fl/fl mice had lower numbers of macrophage/microglia and less pronounced neovascularization (see below for full quantification) (Figure 3, D–F, and Figure 4, D–F).

Figure 3

NRP-1⁺ myeloid cells localize to sites of pathological neovascularization in the retina. Confocal images of isolectin B4– (vessel and microglia stain) and NRP-1–stained retinal flat mounts at P14 with budding neovascular tufts in (A) WT and (D) LysM-Cre Nrp1^fl/fl mice. (B) High-magnification images reveal colocalization of NRP-1⁺ microglia (IBA1) with nascent tufts. (D–F) LysM-Cre Nrp1^fl/fl mice had fewer MPs and fewer developed tufts. (C and F) Observations where confirmed by 3D reconstruction. White arrows in A point to nascent sprouting tufts. The white arrow in B points to NRP-1⁺ MPs associated with growing tufts. For all immunohistochemistry images, representative images of 3 independent experiments are shown. Scale bars: 100 μm (A and D); 50 μm (B and E).

Figure 4

NRP-1⁺ myeloid cells are associated with pathological neovascular tufts. Confocal images of isolectin B4– (vessel and microglia stain) and NRP-1–stained retinal flat mounts at P17 during maximal neovascularization in (A–C) WT and (D–F) LysM-Cre Nrp1^fl/fl mice. In WT retinas, NRP-1⁺ microglia (IBA1) associate with mature tufts. (C and F) Images were reconstructed in 3D. White arrows in A and D point to mature tufts. White arrows in B point to NRP-1⁺ MPs associated with mature tufts. (D–F) LysM-Cre Nrp1^fl/fl mice had fewer MPs and less tufting. For all immunohistochemistry images, representative images of 3 independent experiments are shown. Scale bars: 100 μm (A and D); 50 μm (B and E).

SEMA3A is elevated in the vitreous of patients suffering from active proliferative DR. To establish the clinical relevance of our findings on the obligate role of NRP-1 in MP chemotaxis in retinopathy, we investigated the concentrations of SEMA3A directly in the vitreous of patients suffering from active proliferative DR (PDR). Seventeen samples of undiluted vitreous were obtained from patients suffering from active PDR, and seventeen samples were obtained from control patients with nonvascular pathology. Detailed characteristics of patients are included in Table 1. Control patients presented with nonvascular pathology and showed signs of non-diabetes–related retinal damage, such as tractional tension on vasculature (Figure 5, A and B, white arrows) secondary to fibrotic tissue and macular bulging (Figure 5C). In contrast, all retinas from patients with PDR showed signs of disc (Figure 5D) or preretinal neovascularization (Figure 5F), with highly permeable microvessels (leakage of fluorescent dye) (Figure 5, D and G), microaneurysms (Figure 5, D–G), and fibrous scar tissue, indicative of advanced retinopathy (Figure 5G). In addition, patients showed some evidence of macular edema due to compromised vascular barrier function, including cystoid formation due to focal coalescence of extravasated fluid (Figure 5H).

Figure 5

The NRP-1 ligand SEMA3A is induced in patients suffering from PDR. Angiographies, funduscopies, spectral-domain optical coherence tomography (SD-OCT), and 3D retinal maps obtained from patients selected for the study. Control patients had nonvascular pathologies and were compared with patients with PDR. Control patients with epiretinal membrane shows signs of non-diabetes–related retinal damage, such as (A and B) tractional tension on vasculature (arrow) secondary to (C) fibrotic tissue (white arrow) and macular bulging (angiography and 3D map). Retinas from patients with PDR have (E) neovascularization (high-magnification image) with (D) highly permeable microvessels, as evidenced by leakage of fluorescent dye (high-magnification image), (F) microaneurysms (arrows in high-magnification image), and (G) fibrous scar tissue (arrow), indicative of advanced retinopathy. (H) Patients with PDR show some evidence of macular edema, including cystoid formation (white arrowhead) due to focal coalescence of extravasated fluid. (I) Vitreous humor analyzed by ELISA shows 5-fold increased levels of SEMA3A protein in patients with PDR. Horizontal bars represent mean concentration of SEMA3A, and dots represent concentrations of individual samples; n = 17 for controls and 17 for patients with PDR; *P < 0.05. (J) Western blot analysis of equal volumes of vitreous corroborates the increase in SEMA3A (~125 KDa and 95 KDa) in patients with PDR with respect to controls. Scale bars: 1,500 μm (A–G); 300 μm (C and H).

Table 1

Characteristics of patients having undergone vitreous biopsy

Consistent with a role in PDR, ELISA-based detection of SEMA3A revealed a 5-fold higher concentration of the protein in the vitreous humors of patients with PDR when compared with that in the vitreous of control patients (P = 0.0132) (Figure 5I). Results were confirmed by Western blot analysis on equal volumes of vitreous, where SEMA3A (125- and 95-kDa isoforms) (38, 39) was elevated in patients with PDR (Figure 5J). Thus, upregulation of SEMA3A in the vitreous is induced in diabetic ocular pathology.

NRP-1 ligands are induced in the retinal ganglion cell layer during OIR. To obtain an accurate kinetic profile of expression of the 2 prominent ligands of NRP-1 in proliferative retinopathy, we investigated levels of Sema3a and Vegf message in the mouse model of OIR. Real-time quantitative PCR (RT-qPCR) on whole retinas revealed that Sema3a was robustly induced in OIR both during the hyperoxic (vasodegenerative) phase at P10 and the ischemic/neovascular stage from P12 to P17 (Figure 6A). The observed induction occurred in both WT and LysM-Cre Nrp1^fl/fl retinas. Conversely, as expected, Vegf transcripts rose exclusively in the ischemic phase of OIR from P12 to P17 (Figure 6B). Importantly, Vegf was significantly less induced in LysM-Cre Nrp1^fl/fl retinas when compared with WT retinas (minimally increased at P12; P = 0.0451) and approximately 55% lower at P14 when compared WT OIR (P = 0.0003) (Figure 6B), indicative of a healthier retina.

Figure 6

Ligands of NRP-1 are induced in the retinal ganglion cell layer during OIR. (A and B) Retinas from WT and LysM-Cre Nrp1^fl/fl mice under normoxic conditions or in OIR were collected between P10 and P17 and analyzed by RT-qPCR. (A) Sema3a mRNA expression was induced throughout OIR in both WT and LysM-Cre Nrp1^fl/fl retinas, (B) while Vegf was significantly less induced in LysM-Cre Nrp1^fl/fl retinas compared with WT retinas. Data are expressed as a fold change relative to respective normoxic controls for each time point ± SEM; n = 4–7; *P < 0.05, **P < 0.01, ***P < 0.001. (C) Laser capture microdissection was performed on P14 mice, with care being taken to select avascular retinal zones in OIR. RT-qPCR on laser capture microdissection of retinal layers in control and OIR avascular zones showed an induction in both (D) Sema3a and (E) Vegf mRNA in the ganglion cell layer (GCL) in OIR retinas compared with normoxic retinas. (E) Vegf was also induced in the inner nuclear layer (INL) of OIR retinas. ONL, outer nuclear layer. Data are expressed as a fold change relative to normoxic GCL ± SEM. Scale bars: 500 μm (C; right); 300 μm (C; left).

We next performed laser capture microdissection followed by RT-qPCR on retinal layers in avascular zones to pinpoint the source of Sema3a and Vegf message in OIR (Figure 4C). Both Sema3a and Vegf were robustly induced in the ganglion cell layer, with Vegf also increasing in the inner nuclear layer (Figure 6, D and E). Thus, the source of both ligands is geographically consistent with the localization of retinal MPs (Figures 3 and 4).

MPs do not proliferate in the retina after vascular injury. In order to determine whether the noted rise in NRP-1⁺ MPs was due to an influx from systemic circulation or an increase in MP proliferation within the retina, we investigated local retinal proliferation of these cells. Mice were systemically injected with BrdU at P13 (24 hours prior to sacrifice), and FACS analysis was carried out on retinas (Figure 7A) and spleens (Figure 7B). Within the retinas, Gr-1^–/F4/80⁺/CD11b⁺ MPs did not show significant proliferation (P = 0.4708). Considerably more proliferation was observed in spleens. No significant difference was observed between normoxia and OIR (Figure 7C). The lack of proliferation of MPs in the retinas suggests that the noted accretion of NRP-1⁺ MPs during retinopathy has a systemic or bone marrow–derived origin, as previously elegantly demonstrated for proangiogenic macrophages (40, 41).

Figure 7

NRP-1⁺ MPs do not proliferate in the retina after vascular injury. Representative FACS histograms of Gr-1^–/F4/80⁺/CD11b⁺ cells obtained from (A) retinas and (B) spleens collected at P14 from WT OIR and normoxic control mice injected with BrdU at P13. (C) The number of BrdU⁺ cells was considerably higher in spleens but did not change significantly between OIR and normoxic mice; n = 4 (normoxic), n = 4 (OIR) (total of 16 retinas per condition; each “n” comprises 4 retinas). Data are expressed as a percentage of BrdU⁺ Gr-1^–/F4/80⁺/CD11b⁺ cells ± SEM.

SEMA3A and VEGF₁₆₅ mobilize MPs via NRP-1. In light of the requirement of NRP-1 for myeloid cell mobilization to sites of vascular lesion (Figure 1) as well as the induction of the principal ligands of NRP-1 in retinopathy (Figures 5 and 6) and the likely systemic origin of these cells (Figure 7), we sought to determine the propensity of these cues to provoke chemotaxis of MPs. Primary macrophage cultures were isolated from WT mice and subjected to a Transwell Boyden chamber migration assay. Both SEMA3A (100 ng/ml) (P < 0.0001) and VEGF₁₆₅ (50 ng/ml) (P = 0.0027) provoked macrophage chemotaxis to similar magnitudes as positive control MCP-1 (100 ng/ml) (P < 0.0001) (Figure 8, A and B). These data were validated by demonstrating that Y-27632, a selective inhibitor Rho-associated coiled coil–forming protein serine/threonine kinase (ROCK), abolished their chemotactic properties. ROCK is downstream of NRP-1 signaling (42) and is known to mediate monocyte migration (43). VEGF migration was partially yet not significantly diminished, suggesting a contribution from alternate receptors such as VEGFR1 as recently reported (33). Consistent with a role for NRP-1 in SEMA3A- and VEGF-mediated chemotaxis, macrophages from LysM-Cre Nrp1^fl/fl mice were uniquely responsive to MCP-1 and not mobilized by SEMA3A or VEGF (Figure 8C).

Figure 8

SEMA3A and VEGF are chemoattractive toward macrophages via NRP-1. Primary macrophages were isolated from WT or LysM-Cre Nrp1^fl/fl mice and subjected to a Transwell migration assay, with vehicle, MCP-1 (100 ng/ml), SEMA3A (100 ng/ml), or VEGF (50 ng/ml) added to the lower chamber. (A) Representative images of migrated cells stained with DAPI are shown. (B) SEMA3A and VEGF promoted macrophage migration to similar extents as the positive control MCP-1. To ascertain that SEMA3A and VEGF were stimulating macrophage chemotaxis, cells were pretreated with the selective ROCK inhibitor Y-27632 (100 μg/ml), which abolished chemotaxis. (C) Macrophages from LysM-Cre Nrp1^fl/fl mice were unresponsive to SEMA3A or VEGF but responsive to MCP-1. Data are expressed as a fold change relative to control (nontreated cells); horizontal bars represent mean value of the fold change, and dots represent individual fold changes; n = 6–22; **P < 0.01, ***P < 0.001. Scale bars: 100 μm.

NRP-1⁺ macrophages potentiate microvascular sprouting ex vivo. To investigate the impact of NRP-1 expressing macrophages on microvascular angiogenesis, we isolated choroid tissue from either LysM-Cre Nrp1^+/+ mice or LysM-Cre Nrp1^fl/fl mice and grew them in Matrigel to assess microvascular sprouting. Choroids from LysM-Cre Nrp1^fl/fl mice sprouted approximately 20% fewer microvessels when compared with ones from LysM-Cre Nrp1^+/+ mice (P = 0.018) (Figure 9A). To investigate the role of NRP-1⁺ macrophages in promoting microvascular sprouting, we used clodronate liposomes to eliminate endogenous macrophages from the isolated choroid tissues. In explants from both LysM-Cre Nrp1^fl/fl and LysM-Cre Nrp1^+/+ mice, PBS containing liposomes (i.e., vehicle control) had no affect on vascular sprouting, but clodronate liposomes reduced microvascular sprouting by approximately 60% (P = 0.0114 for LysM-Cre Nrp1^+/+ choroid and P = 0.0007 for LysM-Cre Nrp1^fl/fl choroid) (Figure 9, B–E). To verify whether NRP-1⁺ macrophages have a propensity to promote angiogenesis, we extracted peritoneal macrophages from LysM-Cre Nrp1^+/+ or LysM-Cre Nrp1^fl/fl mice and introduced them into choroid explant cultures that had been treated previously with clodronate liposomes and washed. LysM-Cre Nrp1^+/+ macrophages robustly potentiated microvascular sprouting by 50% to 100% when compared with macrophages from LysM-Cre Nrp1^fl/fl mice (P = 0.0068 for LysM-Cre Nrp1^+/+ choroid and P = 0.0491 for LysM-Cre Nrp1^fl/fl choroid) (Figure 9, D and E), independent of the genotype of the choroidal explant.

Figure 9

NRP-1⁺ macrophages promote microvascular growth in ex vivo choroid explants. (A) Quantification and representative images of choroid explants isolated from LysM-Cre Nrp1^+/+ and LysM-Cre Nrp1^fl/fl mice (n = 6; P = 0.018). (B and C) Representative images of choroid explants from (B) LysM-Cre Nrp1^+/+ and (C) LysM-Cre Nrp1^fl/fl mice following clodronate liposome treatment and subsequent addition of exogenous macrophages (Ma). PBS-Lipo, EGM-2 medium with PBS-filled liposome; Clo-Lipo, EGM-2 medium with dichloromethylenediphosphonic acid disodium salt–filled liposome. (D and E) Quantification of choroidal microvascular sprouting from (D) LysM-Cre Nrp1^+/+ and (E) LysM-Cre Nrp1^fl/fl mice depicted in B and C. PBS-Liposome, EGM-2 medium with PBS-filled liposome; Clodronate-Liposome, EGM-2 medium with dichloromethylenediphosphonic acid disodium salt–filled liposome. (n = 6, *P < 0.05, **P < 0.01, ***P < 0.001). Scale bars: 500 μm.

Deficiency in myeloid-resident NRP-1 reduces vascular degeneration and pathological neovascularization in retinopathy. Given the obligate role of NRP-1 in MP infiltration during the early stages of OIR (Figure 1), we next sought to determine the affect of myeloid cell–specific ablation of NRP-1 on the progression of disease. Upon exit from 75% O₂ at P12, LysM-Cre Nrp1^fl/fl mice showed significantly lower levels of retinal vaso-obliteration when compared with those of WT (P = 0.0011) and LysM-Cre Nrp1^+l+ (P < 0.0001) controls (Figure 10, A and B). This may be attributed to lower levels of IL-1β present in the retinas of LysM-Cre Nrp1^fl/fl mice (Supplemental Figure 6). Importantly, at P17, when pathological neovascularization peaks (26), deletion of myeloid-resident NRP-1 profoundly reduced avascular areas (~35% when compared with WT [P < 0.0001] and ~30% compared with LysM-Cre Nrp1^+l+ mice [P = 0.0008]) (Figure 10, C and D). In turn, we observed significant reductions in destructive preretinal neovascularization associated with ischemic retinopathy (~36% when compared with WT [P = 0.0008] and ~34% compared with LysM-Cre Nrp1^+l+ mice [P = 0.0013]) (Figure 10, E and F).

Figure 10

Deficiency in myeloid-resident NRP-1 reduces vascular degeneration and pathological neovascularization in retinopathy. WT, LysM-Cre/Nrp1^+/+, and LysM-Cre Nrp1^fl/fl mice were subjected to OIR, and retinas were collected at P12 and P17, flat mounted, and stained with isolectin B4. LysM-Cre Nrp1^fl/fl mice had (A and B) less vaso-obliteration at P12 and (C and D) reduced avascular areas and (E and F) preretinal neovascularization at P17 compared with both control WT or control LysM-Cre Nrp1^+/+ mice. Results are expressed as percentage of avascular or neovascular area versus the whole retinal area. Horizontal bars represent mean value of percentage, and dots represent individual values; n = 5–19; **P < 0.01, ***P < 0.001. Scale bars: 250 μm (B, D, and F).

Therapeutic intravitreal administration of soluble NRP-1 reduces MP infiltration and pathological neovascularization in retinopathy. To determine the translational potential of our findings, we next used a soluble recombinant mouse NRP-1 (rmNRP-1) (Figure 11A) as a trap to sequester OIR-induced ligands of NRP-1. A single intravitreal injection of rmNRP-1 at P12 lead to a 30% reduction at P14 (P = 0.0282) in the number of microglia present in retinas subjected to OIR (Figure 11B). This finding attests to the potency of soluble NRP-1 (1 μl of 500 μg/ml) to compromise microglial mobilization. Intravitreal administration of soluble NRP-1 provoked a significant approximate 40% decrease in pathological preretinal angiogenesis when compared with that of vehicle-injected controls (P = 0.0025) (Figure 11, C and D). Together, these data suggest that neutralization of ligands of NRP-1 is an effective strategy to reduce destructive neovascularization in retinopathy.

Figure 11

Therapeutic intravitreal administration of soluble NRP-1 reduces MP infiltration and pathological neovascularization in retinopathy. (A) WT mice were subjected to OIR and injected intravitreally at P12 with soluble rmNRP-1 as a trap to sequester OIR-induced ligands of NRP-1. (B) At P14, FACS analysis revealed a decrease of over 30% in the number of retinal MPs in rmNRP-1–injected retinas. Data are expressed as a fold change relative to control (vehicle-injected retinas) ± SEM; n = 3–4 (total of 12–16 retinas per condition; each “n” comprises 4 retinas); *P < 0.05. (C and D) Treatment with rmNRP-1 efficiently decreased pathological neovascularization at P17 when compared with vehicle-injected eyes. Results are expressed as percentage of neovascular area versus the whole retinal area. Horizontal bars represent mean value of percentage, and dots represent individual values; n = 11, **P < 0.01. Scale bars: 250 μm.

Although the contribution of MPs to ocular neovascular diseases such as DR has been established for over a decade, knowledge of the mechanisms by which they are locally recruited to areas of tissue damage and partake in the disease process remains limited. Here, we demonstrate that stressed retinal neurons (19, 21) and neural tissue have the inherent ability to modulate the local innate immune response via unconventional chemotactic agents, such as the classical repulsive guidance cue SEMA3A (21) and vasoattractive and neuroattractive VEGF (Figure 12). We provide evidence that these cues act as potent chemoattractants for NRP-1⁺ MPs during the early stages of ischemic injury in the retina and contribute to vascular pathology in retinopathy. This neuroimmune paradigm consolidates the stress-induced neurovascular response to injury with an innate immune response. Either genetic ablation of NRP-1 specifically in myeloid cells (LysM-Cre Nrp1^fl/fl) or therapeutic administration of a NRP-1–derived trap limits MP recruitment, reduces vascular degeneration, and decreases destructive pathological neovascularization.

Figure 12

Schematic depiction of findings. The schematic illustrates that during ischemic retinopathies, such as diabetes, avascular zones of the retina, ischemic neurons, and neural tissue produce ligands of NRP-1 (SEMA3A and VEGF), which in turn act as potent chemoattractive agents for proangiogenic microglia. The NRP-1⁺ microglia then partake in the pathogenesis of proliferative retinopathy. RGC, retinal ganglion cell.

The association of chronic low-grade inflammation with the progression of DR (5) and the link between neonatal infections and ROP (9, 44) underscores the potential therapeutic worth of antiinflammatory drugs in treatment of proliferative retinopathies (45), yet toxicity remains a concern especially in pediatric populations. This is observed with prolonged use of intravitreal glucocorticoids (widely used for diabetic macular edema), which leads to increased cataract formation and elevated intraocular pressure (46). Hence, in order to improve safety profiles of current antiinflammatory treatment paradigms, it will be necessary to provide a honed intervention that specifically targets damaging sterile neuroinflammation.

The identification of the SEMA3A/NRP-1 axis in chemoattraction of destructive MPs in retinopathy may provide such a target. In contrast to VEGF, which is essential for retinal homeostasis, expression of SEMA3A in healthy mature retinas is limited (20). However, in diabetes and OIR, retinal ganglion neurons, which are in intimate proximity with retinal vessels, generate significant amounts of this cue (20, 21). In addition to a role in local MP chemotaxis, SEMA3A also directly affects microvessels (47) and may directly induce endothelial apoptosis and promote cytoskeleton remodeling (47–50). Specifically in the retina, SEMA3A hinders physiological vascular regeneration in OIR (21) and provokes pathological compromise of barrier function in DR (20). Hence, therapeutic neutralization of SEMA3A may provide a multifaceted advantage over current intravitreally delivered compounds and may have limited toxicity, given that its expression is typically restricted to embryogenesis (51). Our data also raise the possibility that, beyond direct influence on angiogenesis and barrier function (52), the mode of action of current anti-VEGF therapies may also in part involve limiting MP-induced lesions to the neuroretina.

Together, the findings presented in this study establish a basis for future investigations on the role of NRP-1⁺ MPs and their ligands in CNS disorders characterized by excessive neuroinflammation. Understanding the signals that influence neuroimmune interplay may provide valuable therapeutic insight for the design of selective antiinflammatory treatments to counter destructive neuroinflammation.

View Supplemental data

P. Sapieha holds a Canada Research Chair in Retinal Cell Biology and The Alcon Research Institute Young Investigator Award. This work was supported by operating grants to P. Sapieha from the Canadian Institutes of Health Research (221478), the Canadian Diabetes Association (OG-3-14-4544-PS), and Natural Sciences and Engineering Research Council of Canada (418637). Additional support was obtained from the Fondation Hopital Maisonneuve-Rosemont, Réseau en Recherche en Santé de la Vision du Quebec, and the Fond en Recherche en Optalmologie de l’UdM. A. Stahl is supported by Deutsche Forschungsgemeinschaft (STA 1102/5-1). F. Beaudoin is supported by the Whitearn Foundation.

Address correspondence to: Przemyslaw (Mike) Sapieha, Hopital Maisonneuve-Rosemont Research Centre, 5415 Assomption Boulevard, Montreal, Quebec, H1T 2M4, Canada. Phone: 514.252.3400, ext. 7711; E-mail: mike.sapieha@umontreal.ca.

Conflict of interest: The University of Montreal, Maisonneuve-Rosemont Hospital, and Przemyslaw Sapieha have filed a patent pertaining to certain results presented in this paper.

Reference information: J Clin Invest. 2014;124(11):4807–4822. doi:10.1172/JCI76492.

1. Lampron A, Elali A, Rivest S. Innate immunity in the CNS: redefining the relationship between the CNS and Its environment. Neuron. 2013;78(2):214–232.
   View this article via: PubMed CrossRef Google Scholar
2. Ousman SS, Kubes P. Immune surveillance in the central nervous system. Nat Neurosci. 2012;15(8):1096–1101.
   View this article via: PubMed CrossRef Google Scholar
3. Adamis AP, Berman AJ. Immunological mechanisms in the pathogenesis of diabetic retinopathy. Semin Immunopathol. 2008;30(2):65–84.
   View this article via: PubMed CrossRef Google Scholar
4. Antonetti DA, Klein R, Gardner TW. Diabetic retinopathy. N Engl J Med. 2012;366(13):1227–1239.
   View this article via: PubMed CrossRef Google Scholar
5. Joussen AM, et al. A central role for inflammation in the pathogenesis of diabetic retinopathy. FASEB J. 2004;18(12):1450–1452.
   View this article via: PubMed Google Scholar
6. Ambati J, Fowler BJ. Mechanisms of age-related macular degeneration. Neuron. 2012;75(1):26–39.
   View this article via: PubMed CrossRef Google Scholar
7. Sennlaub F, et al. CCR2(+) monocytes infiltrate atrophic lesions in age-related macular disease and mediate photoreceptor degeneration in experimental subretinal inflammation in Cx3cr1 deficient mice. EMBO Mol Med. 2013;5(11):1775–1793.
   View this article via: PubMed CrossRef Google Scholar
8. Combadiere C, et al. CX3CR1-dependent subretinal microglia cell accumulation is associated with cardinal features of age-related macular degeneration. J Clin Invest. 2007;117(10):2920–2928.
   View this article via: JCI PubMed CrossRef Google Scholar
9. Dammann O. Inflammation and retinopathy of prematurity. Acta Paediatr. 2010;99(7):975–977.
   View this article via: PubMed CrossRef Google Scholar
10. Tremblay S, et al. Systemic inflammation perturbs developmental retinal angiogenesis and neuroretinal function. Invest Ophthalmol Vis Sci. 2013;54(13):8125–8139.
    View this article via: PubMed CrossRef Google Scholar
11. Hartnett ME, Penn JS. Mechanisms and management of retinopathy of prematurity. N Engl J Med. 2012;367(26):2515–2526.
    View this article via: PubMed CrossRef Google Scholar
12. Kempen JH, et al. The prevalence of diabetic retinopathy among adults in the United States. Arch Ophthalmol. 2004;122(4):552–563.
    View this article via: PubMed Google Scholar
13. Sapieha P, et al. Proliferative retinopathies: angiogenesis that blinds. Int J Biochem Cell Biol. 2010;42(1):5–12.
    View this article via: PubMed CrossRef Google Scholar
14. Robinson GS, et al. Nonvascular role for VEGF: VEGFR-1, 2 activity is critical for neural retinal development. FASEB J. 2001;15(7):1215–1217.
    View this article via: PubMed Google Scholar
15. Saint-Geniez M, et al. Endogenous VEGF is required for visual function: evidence for a survival role on muller cells and photoreceptors. PLoS One. 2008;3(11):e3554.
    View this article via: PubMed CrossRef Google Scholar
16. Hellström A, Smith LE, Dammann O. Retinopathy of prematurity. Lancet. 2013;382(9902):1445–1457.
    View this article via: PubMed CrossRef Google Scholar
17. Sapieha P. Eyeing central neurons in vascular growth and reparative angiogenesis. Blood. 2012;120(11):2182–2194.
    View this article via: PubMed CrossRef Google Scholar
18. Kern TS, Barber AJ. Retinal ganglion cells in diabetes. J Physiol. 2008;586(pt 18):4401–4408.
    View this article via: PubMed Google Scholar
19. Binet F, et al. Neuronal ER stress impedes myeloid-cell-induced vascular regeneration through IRE1α degradation of netrin-1. Cell Metab. 2013;17(3):353–371.
    View this article via: PubMed CrossRef Google Scholar
20. Cerani A, et al. Neuron-derived semaphorin 3A is an early inducer of vascular permeability in diabetic retinopathy via neuropilin-1. Cell Metab. 2013;18(4):505–518.
    View this article via: PubMed CrossRef Google Scholar
21. Joyal JS, et al. Ischemic neurons prevent vascular regeneration of neural tissue by secreting semaphorin 3A. Blood. 2011;117(22):6024–6035.
    View this article via: PubMed CrossRef Google Scholar
22. Checchin D, Sennlaub F, Levavasseur E, Leduc M, Chemtob S. Potential role of microglia in retinal blood vessel formation. Invest Ophthalmol Vis Sci. 2006;47(8):3595–3602.
    View this article via: PubMed CrossRef Google Scholar
23. Connor KM, et al. Increased dietary intake of omega-3-polyunsaturated fatty acids reduces pathological retinal angiogenesis. Nat Med. 2007;13(7):868–873.
    View this article via: PubMed CrossRef Google Scholar
24. Sapieha P, et al. 5-Lipoxygenase metabolite 4-HDHA is a mediator of the antiangiogenic effect of ω-3 polyunsaturated fatty acids. Sci Transl Med. 2011;3(69):69ra12.
    View this article via: PubMed Google Scholar
25. Stahl A, et al. Short communication: PPAR gamma mediates a direct antiangiogenic effect of omega 3-PUFAs in proliferative retinopathy. Circ Res. 2010;107(4):495–500.
    View this article via: PubMed CrossRef Google Scholar
26. Smith LE, et al. Oxygen-induced retinopathy in the mouse. Invest Ophthalmol Vis Sci. 1994;35(1):101–111.
    View this article via: PubMed Google Scholar
27. Lee P, Goishi K, Davidson AJ, Mannix R, Zon L, Klagsbrun M. Neuropilin-1 is required for vascular development and is a mediator of VEGF-dependent angiogenesis in zebrafish. Proc Natl Acad Sci U S A. 2002;99(16):10470–10475.
    View this article via: PubMed CrossRef Google Scholar
28. Gluzman-Poltorak Z, Cohen T, Shibuya M, Neufeld G. Vascular endothelial growth factor receptor-1 and neuropilin-2 form complexes. J Biol Chem. 2001;276(22):18688–18694.
    View this article via: PubMed CrossRef Google Scholar
29. Mamluk R, Gechtman Z, Kutcher ME, Gasiunas N, Gallagher J, Klagsbrun M. Neuropilin-1 binds vascular endothelial growth factor 165, placenta growth factor-2, and heparin via its b1b2 domain. J Biol Chem. 2002;277(27):24818–24825.
    View this article via: PubMed CrossRef Google Scholar
30. Soker S, Takashima S, Miao HQ, Neufeld G, Klagsbrun M. Neuropilin-1 is expressed by endothelial and tumor cells as an isoform-specific receptor for vascular endothelial growth factor. Cell. 1998;92(6):735–745.
    View this article via: PubMed CrossRef Google Scholar
31. Fantin A, et al. Tissue macrophages act as cellular chaperones for vascular anastomosis downstream of VEGF-mediated endothelial tip cell induction. Blood. 2010;116(5):829–840.
    View this article via: PubMed CrossRef Google Scholar
32. Carrer A, et al. Neuropilin-1 identifies a subset of bone marrow Gr1- monocytes that can induce tumor vessel normalization and inhibit tumor growth. Cancer Res. 2012;72(24):6371–6381.
    View this article via: PubMed CrossRef Google Scholar
33. Casazza A, et al. Impeding macrophage entry into hypoxic tumor areas by Sema3A/Nrp1 signaling blockade inhibits angiogenesis and restores antitumor immunity. Cancer Cell. 2013;24(6):695–709.
    View this article via: PubMed CrossRef Google Scholar
34. Ritter MR, Banin E, Moreno SK, Aguilar E, Dorrell MI, Friedlander M. Myeloid progenitors differentiate into microglia and promote vascular repair in a model of ischemic retinopathy. J Clin Invest. 2006;116(12):3266–3276.
    View this article via: JCI PubMed CrossRef Google Scholar
35. Stahl A, et al. Postnatal weight gain modifies severity and functional outcome of oxygen-induced proliferative retinopathy. Am J Pathol. 2010;177(6):2715–2723.
    View this article via: PubMed CrossRef Google Scholar
36. Clausen BE, Burkhardt C, Reith W, Renkawitz R, Forster I. Conditional gene targeting in macrophages and granulocytes using LysMcre mice. Transgenic Res. 1999;8(4):265–277.
    View this article via: PubMed CrossRef Google Scholar
37. Mattapallil MJ, et al. The Rd8 mutation of the Crb1 gene is present in vendor lines of C57BL/6N mice and embryonic stem cells, and confounds ocular induced mutant phenotypes. Invest Ophthalmol Vis Sci. 2012;53(6):2921–2927.
    View this article via: PubMed CrossRef Google Scholar
38. Klebanov O, Nitzan A, Raz D, Barzilai A, Solomon AS. Upregulation of Semaphorin 3A and the associated biochemical and cellular events in a rat model of retinal detachment. Graefes Arch Clin Exp Ophthalmol. 2009;247(1):73–86.
    View this article via: PubMed CrossRef Google Scholar
39. Koppel AM, Raper JA. Collapsin-1 covalently dimerizes, and dimerization is necessary for collapsing activity. J Biol Chem. 1998;273(25):15708–15713.
    View this article via: PubMed CrossRef Google Scholar
40. Kubota Y, et al. M-CSF inhibition selectively targets pathological angiogenesis and lymphangiogenesis. J Exp Med. 2009;206(5):1089–1102.
    View this article via: PubMed CrossRef Google Scholar
41. Okuno Y, Nakamura-Ishizu A, Kishi K, Suda T, Kubota Y. Bone marrow-derived cells serve as proangiogenic macrophages but not endothelial cells in wound healing. Blood. 2011;117(19):5264–5272.
    View this article via: PubMed CrossRef Google Scholar
42. Neufeld G, Kessler O. The semaphorins: versatile regulators of tumour progression and tumour angiogenesis. Nat Rev Cancer. 2008;8(8):632–645.
    View this article via: PubMed CrossRef Google Scholar
43. Worthylake RA, Burridge K. RhoA and ROCK promote migration by limiting membrane protrusions. J Biol Chem. 2003;278(15):13578–13584.
    View this article via: PubMed CrossRef Google Scholar
44. Dammann O, et al. Immaturity, perinatal inflammation, and retinopathy of prematurity: a multi-hit hypothesis. Early Hum Dev. 2009;85(5):325–329.
    View this article via: PubMed CrossRef Google Scholar
45. Kastelan S, Tomic M, Gverovic Antunica A, Salopek Rabatic J, Ljubic S. Inflammation and pharmacological treatment in diabetic retinopathy. Mediators Inflamm. 2013;2013:213130.
    View this article via: PubMed Google Scholar
46. Silva PS, Cavallerano JD, Sun JK, Aiello LM, Aiello LP. Effect of systemic medications on onset and progression of diabetic retinopathy. Nat Rev Endocrinol. 2010;6(9):494–508.
    View this article via: PubMed CrossRef Google Scholar
47. Miao HQ, Soker S, Feiner L, Alonso JL, Raper JA, Klagsbrun M. Neuropilin-1 mediates collapsin-1/semaphorin III inhibition of endothelial cell motility: functional competition of collapsin-1 and vascular endothelial growth factor-165. J Cell Biol. 1999;146(1):233–242.
    View this article via: PubMed CrossRef Google Scholar
48. Klagsbrun M, Eichmann A. A role for axon guidance receptors and ligands in blood vessel development and tumor angiogenesis. Cytokine Growth Factor Rev. 2005;16(4):535–548.
    View this article via: PubMed Google Scholar
49. Guttmann-Raviv N, Shraga-Heled N, Varshavsky A, Guimaraes-Sternberg C, Kessler O, Neufeld G. Semaphorin-3A and semaphorin-3F work together to repel endothelial cells and to inhibit their survival by induction of apoptosis. J Biol Chem. 2007;282(36):26294–26305.
    View this article via: PubMed CrossRef Google Scholar
50. Neufeld G, Sabag AD, Rabinovicz N, Kessler O. Semaphorins in angiogenesis and tumor progression. Cold Spring Harb Perspect Med. 2012;2(1):a006718.
    View this article via: PubMed Google Scholar
51. Bussolino F, Valdembri D, Caccavari F, Serini G. Semaphoring vascular morphogenesis. Endothelium. 2006;13(2):81–91.
    View this article via: PubMed CrossRef Google Scholar
52. Yancopoulos GD. Clinical application of therapies targeting VEGF. Cell. 2010;143(1):13–16.
    View this article via: PubMed CrossRef Google Scholar
53. Miloudi K, Dejda A, Binet F, Lapalme E, Cerani A, Sapieha P. Assessment of vascular regeneration in the CNS using the mouse retina. J Vis Exp. 2014;(88):e51351.
    View this article via: PubMed Google Scholar
54. Sapieha P, et al. Retinopathy of prematurity: understanding ischemic retinal vasculopathies at an extreme of life. J Clin Invest. 2010;120(9):3022–3032.
    View this article via: JCI PubMed CrossRef Google Scholar
55. Stahl A, et al. The mouse retina as an angiogenesis model. Invest Ophthalmol Vis Sci. 2010;51(6):2813–2826.
    View this article via: PubMed CrossRef Google Scholar
56. Stahl A, et al. Computer-aided quantification of retinal neovascularization. Angiogenesis. 2009;12(3):297–301.
    View this article via: PubMed CrossRef Google Scholar
57. Shao Z, et al. Choroid sprouting assay: an ex vivo model of microvascular angiogenesis. PLoS One. 2013;8(7):e69552.
    View this article via: PubMed CrossRef Google Scholar
58. van Rooijen N, van Kesteren-Hendrikx E. “In vivo” depletion of macrophages by liposome-mediated “suicide”. Methods Enzymol. 2003;373:3–16.
    View this article via: PubMed Google Scholar