Choroidal neovascularization is inhibited via an intraocular decrease of inflammatory cells in mice lacking complement component C3

In early age-related macular degeneration (AMD), complement component C3 can be observed in drusen, which is the accumulation of material beneath the retinal pigment epithelium. The complement pathways, via the activation of C3, can upregulate the expression of cytokines and their receptors and the recruitment of inflammatory leukocytes, both of which play an important role in the development of choroidal neovascularization (CNV) in exudative AMD. Laser-induced CNV lesions were found to be significantly smaller in C3−/− mice than in wild-type mice. By using flow cytometry, we demonstrated that the proportions of intraocular granulocytes, CD11b+F4/80+Ly6Chi and CD11b+F4/80+Ly6Clo cells, were lower in C3−/− mice than in wild-type mice as early as day 1 after laser injury, and the proportions of granulocytes and three macrophage/monocyte subsets were significantly lower on day 3. In contrast, C3−/− mice had more granulocytes and CD11b+F4/80+Ly6Chi cells in peripheral blood than wild-type mice after injury. Further, the expression levels of Vegfa164 were upregulated in intraocular Ly6Chi macrophages/monocytes of C3−/− mice, but not as much as in wild-type mice. Collectively, our data demonstrate that despite a more pronounced induction of systemic inflammation, inhibition of complement factor C3 suppresses CNV by decreasing the recruitment of inflammatory cells to the lesion.

drusen of patients with AMD 6 . The alternative pathway is activated by spontaneous hydrolysis of C3. The anaphylatoxins produced by C3 hydrolysis, such as C3a and C3b, and their downstream factors, including C5a, are important chemoattractants that can trigger the recruitment of neutrophils and macrophages 7,8 . In a mouse CNV model, both C3a and C5a have been suggested to contribute to angiogenesis and upregulate the expression of vascular endothelial growth factor (VEGF) 9 . The alternative pathway via C3b anaphylatoxin can also lead to the formation of the membrane attack complex (MAC) C5b-9 8 . It has been demonstrated that inhibition of MAC formation results in the inhibition of CNV in mice 10 .
The involvement of inflammatory cells, including macrophages, in the pathogenesis of exudative AMD has been reported in histologic studies of CNV 11,12 . In experimental models, macrophages and granulocytes have also been found to infiltrate laser-induced CNV lesions 13,14 . Many studies have suggested that macrophage depletion correlates with reduced CNV responses 14,15 , although others have demonstrated that an intraocular injection of macrophages reduces CNV size 16,17 . These contradictory findings suggest that the effect of macrophages on CNV is complex and may depend on the age of the mice used and the age and subtypes of macrophages.
The bone marrow and peripheral blood are the primary sources from which monocytes are mobilized to enter into tissue sites after injury 18 . There are three subsets of macrophages (Ly6C hi , Ly6C int , and Ly6C lo ) that are derived from circulating inflammatory monocytes in mouse models of chronic disease 19 . Ly6C hi macrophages/monocytes, which are similar to M1 macrophages, migrate to injured tissues and produce pro-inflammatory cytokines and chemokines in a mouse model of arteriosclerosis, chronic heart failure, and chronic kidney failure [20][21][22] . In contrast, Ly6C lo macrophages/monocytes that differentiate into the M2 macrophage subtype promote wound healing in myocardium, inflamed skeletal muscle, and brain tissues [23][24][25] . Ly6C hi monocytes have been demonstrated to leave the bone marrow and enter the circulating blood via CC-chemokine receptor 2-mediated migration 26 . In the steady state, Ly6C hi macrophages/ monocytes differentiate into Ly6C lo macrophages/monocytes in the circulation 27 . Ly6C lo macrophages/ monocytes are recruited into tissue via the chemokine receptor CX3CR1 28 and are considered to become tissue-resident macrophages; however, recent studies suggest tissue-resident macrophages originate from the yolk sac or fetal liver progenitors and self-renew in situ [29][30][31] . Ly6C int macrophages/monocytes are thought to be a phenotype that the Ly6C hi subtype adopts (under steady-state conditions) before they form the Ly6C lo subtype 32 . Moreover, their functions depend on the context in which they are found 32,33 .
To test directly the involvement of C3 in the recruitment of inflammatory cells in the development of CNV, we generated laser-induced CNV in wild-type and C3-deficient (C3 −/− ) mice. First, we compared lesion size between wild-type and C3 −/− mice. Second, we used flow cytometry to evaluate the time course of changes in the proportions of granulocytes and macrophage subsets in the posterior segment of the eye and the peripheral blood in both types of mice after laser photocoagulation. Lastly, we examined the expression of Vegfa164 and Vegfr1 in intraocular Ly6C hi macrophages/monocytes with or without laser treatment.

Results
We used C3 −/− mice and their wild-type littermates to investigate the role of C3 in CNV. C3 −/− mice and wild-type mice were all Rd1 and Rd8 negative and had similar retinal morphology in the absence of laser treatment. At the age of 6 months, no drusen, retinal lesions, or immune cell infiltration were detected in mice of either genotype (data not shown).
Male mice 7 to 8 weeks of age were used for the laser-induced CNV mouse model. At 7 days after laser injury, C3 −/− mice had lesions that were 35% smaller than those in wild-type mice (P < 0.05; Fig. 1A-C).
In wild-type mice, the proportion of granulocytes recruited into the posterior segment of the eye among the total number of cells increased greatly from day 1 after laser injury and peaked on day 3. In contrast, the increase in the proportion of granulocytes in C3 −/− mice was moderate; it peaked at 1 day after laser injury and decreased thereafter, and was significantly lower on days 1 (43.8 ± 11.0%) and 3 (16.5 ± 3.9%) than in wild-type mice ( Fig. 2A,B).
The three subsets of intraocular macrophages/monocytes (Ly6C hi , Ly6C int , and Ly6C lo ) peaked at 3 days after laser injury in wild-type mice. Compared with wild-type mice, the increase in Ly6C hi and Ly6C lo macrophages/monocytes was suppressed from day 1 after injury in C3 −/− mice (to 52.0 ± 14.9% and 20.5 ± 11.5% of wild-type values, respectively), and the three macrophage/monocyte subsets were significantly lower on day 3 after injury relative to those in wild-type mice (8.0 ± 1.3%, 23.1 ± 4.1%, and 25.5 ± 0.7% of wild-type values, for Ly6C hi , Ly6C int , and Ly6C lo , respectively; P < 0.05 for all comparisons) (Fig. 3A,B).
Considering that circulating inflammatory cells migrate into the lesion, we next asked whether there were any changes in the inflammatory cells circulating in peripheral blood. There was no significant difference in the number of granulocytes and macrophages/monocytes between wild-type and C3 −/− mice without treatment (Figs 4B and 5B). Although granulocytes and Ly6C hi macrophages/monocytes circulating in peripheral blood in wild-type mice subsequently increased until day 7 after laser treatment, significantly higher percentages of granulocytes were observed in C3 −/− mice than in wild-type mice at days 1 (7.3 ± 2.0-fold compared with wild-type values), 3 (3.2 ± 0.6-fold compared with wild-type values), and 7 after laser injury (2.4 ± 0.2-fold compared with wild-type values; P < 0.05 for all comparisons) (Fig. 4B). With regard to Ly6C hi macrophages/monocytes, C3 −/− mice also had more Ly6C hi cells in the peripheral blood than wild-type mice at days 1 (8.9 ± 0.3-fold compared with wild-type values), 3 (4.9 ± 0.7-fold compared with wild-type values), and 7 after laser injury (3.6 ± 0.1-fold compared with wild-type values; P < 0.05 for all comparisons) (Fig. 5B). There was no significant difference in the proportions of Ly6C lo and Ly6C int macrophages/monocytes in the blood between the two groups during CNV formation (Fig. 5). Moreover, there was no significant change in the percentages of splenic and myeloid inflammatory cells between both groups (data not shown).
This seemingly discrepant increase in inflammatory cells in the blood and decrease in cells infiltrating the eye led us to investigate Vegf and one of its receptors, Vegfr1, which is expressed on monocytes and functions as a chemoattractant receptor. To investigate Vegf and Vegfr1 expression in intraocular inflammatory cells in both groups before and after laser injury, we performed real-time quantitative reverse transcriptase-polymerase chain reaction (RT-PCR). Vegfa164 expression was barely detected in intraocular Ly6C hi or Ly6C lo macrophages/monocytes in both groups without laser treatment. In contrast, Vegfa164 expression levels were upregulated in intraocular Ly6C hi cells after laser injury; however, the upregulated expression in C3 −/− mice (4.5 ± 1.7-fold compared with the Ly6C lo values of wild-type mice) was significantly lower than in wild-type mice (11.6 ± 2.7-fold compared with the Ly6C lo values of wild-type mice) (Fig. 6A). There was no significant difference in the expression levels of Vegfr1 in intraocular Ly6C hi cells between both groups before or after laser injury. Vegfr1 expression was not detected in intraocular Ly6C lo cells of both groups with or without laser treatment (Fig. 6B).

Discussion
Prior studies have demonstrated that the activation of complement component C3 in drusen can induce chronic inflammation in the RPE and play an important role in angiogenesis in AMD 3,5,6 . Recent studies have also described the importance of C3 and the C3a receptor in leukocyte recruitment to the choroid after laser treatment in a laser-induced CNV model 9,34 . In contrast, another group reported that C3 -/mice have increased neovascularization 35 . In this study, we investigated the effects of C3 on CNV lesion size and the involvement of C3 deficiency in inflammatory cell recruitment, such as granulocytes and macrophage/monocyte subsets, during the formation of CNV.
The result that CNV lesion size was significantly decreased in C3 −/− mice was consistent with a previous report 34 , and several groups have reported that maximal neutrophil infiltration into the choroid occurs 1 day after laser injury 9,36 . However, others have reported that neutrophil infiltration is maximal at 3 days after laser injury in wild-type mice 13 . All studies concur that macrophage infiltration peaks on day 3 after laser injury 9,36,37 . In our study, granulocytes and all three macrophage subsets showed maximal infiltration into the posterior segment of the eyes of wild-type mice 3 days after laser treatment. However, in C3 −/− mice, the proportions of intraocular granulocytes and Ly6C hi macrophages were maximal at 1 day after laser injury, whereas Ly6C int macrophages were maximal at 3 days and the proportion of Ly6C lo macrophages increased until at least 7 days. In sharp contrast to the larger proportion of inflammatory cells in the peripheral blood of C3 −/− mice, immune cell infiltration of the eyes was markedly suppressed by C3 deficiency. It is likely that tissue migration and the recruitment of immune cells were inhibited because of the lack of C3a and C5a anaphylatoxins. It is also plausible that when the number of inflammatory cells infiltrating the eyes is increased, a feedback loop may occur and reduce the number of inflammatory cells in the peripheral blood 38 . Our results also suggest that this feedback mechanism might be abrogated in C3 −/− mice, probably due to the deletion of C3 and subsequent loss of anaphylatoxin production.
Complement C3 is produced in the RPE/choroid and upregulated in the CNV mouse model 34,39 . C3a and C5a anaphylatoxins have been shown to regulate tissue infiltration of macrophages and monocytes, and control VEGF production and secretion from RPE cells. RPE cells also contribute to the VEGF-VEGFR1 axis. It is likely that macrophage/monocyte recruitment into the eye and the upregulation of VEGF expression are suppressed in C3 −/− mice because of the absence of C3a and C5a. Moreover, the inhibition of laser-induced CNV in C3 −/− mice is likely to be due to the reduction of VEGF expression in the eye. However, further studies are still required to identify the role of C3 in the production of VEGF by RPE cells and to clarify whether locally or systematically produced C3 plays a pivotal role.
Blockade of the complement pathway is a new strategy for the treatment of AMD. Many new drugs are currently being studied in clinical trials. POT-4 (Potentia Pharmaceuticals, Louisville, KY) is a peptide capable of binding to human complement factor C3 and preventing its activation, and a phase I clinical trial of POT-4 for exudative AMD has been completed 40 . Our findings provide additional evidence that an ocular C3-targeting agent is likely to be effective in the treatment of exudative AMD. In conclusion, our data show that inhibition of intraocular complement factor C3 might suppress CNV formation via reduction of macrophage/granulocyte infiltration and Vegfa164 expression. Thus, a C3 inhibitor might be useful in the treatment of AMD. Importantly, our results also support the pro-angiogenic nature of Ly6C hi macrophages/monocytes in CNV.

Methods
Ethics statement. All animal experiments were performed in accordance with the guidelines of the University of Tokyo and the Association for Research in Vision and Ophthalmology. All experimental protocols were approved by the department's animal experimentation committee of the University of Tokyo.
Laser-induced CNV. CNV lesions were induced by laser photocoagulation. Laser radiation was delivered between the major retinal vessels and at equal distances from the optic disc with a diode laser (DC-3000 ® ; NIDEK, Osaka, Japan) and a slit lamp delivery system (SL-7F; Topcon, Tokyo, Japan). Through a glass cover used as a contact lens, we delivered laser light (200 mW intensity, 170 μ m size, 0.02 s duration) to 4 spots per eye for the lesion size study and to 12 spots per eye for the flow cytometry studies. The rupture of Bruch's membrane for each lesion was evidenced by bubble formation at the time of laser exposure 44,45 . Each experimental group consisted of 6 or 10 mice. For the analysis of CNV area, 6 mice per group were subjected to laser treatment and used for the analysis. For flow cytometric analysis, cells isolated from 6 mice were used for a single FACS analysis and cells from 10 mice were used for a single RPE flatmounts. At 7 days after laser injury, the mice were perfused transcardially with phosphate-buffered saline (PBS) and then with FITC-conjugated concanavalin A (20 μ g/mL in PBS; Vector Laboratories, Burlingame, CA) to label the CNVs. The eyes were removed and the posterior segment of the eyeball was prepared as a flatmount. RPE flatmounts were observed using a fluorescence microscope (Keyence Corporation, Tokyo, Japan). Measurements of CNV lesion area were performed using ImageJ software (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsbweb.nih.gov/ij/index.html). The outline of the CNV was drawn around the perimeter of the lesion and then total lesion area was measured 46 . The average CNV area obtained from 4 lesions in each eye was used to give a single value for analysis. Flow cytometry. To isolate cells from eyes, retinas and RPE/choroids were harvested from 12 eyes (6 mice) of each group and combined in collagenase (1 mg/mL; Wako, Osaka, Japan) and dispase (1 mg/mL; Invitrogen, Waltham, MA, USA) in PBS. After 10 min incubation at 37 °C, the cells were passed through a 40-μ m nylon mesh and then centrifuged at 700 × g for 5 min at 4 °C. The supernatant was discarded, and cells resuspended with PBS were kept on ice for flow cytometry. Peripheral blood was collected from the heart with 1000 U/mL heparin (Novo-Heparin ® ; Mochida, Tokyo, Japan). After the spleens and femurs were harvested, the cells were isolated from them as described previously 47 . After single cell suspensions were prepared as described previously 48    Representative flow cytometry plots of CD11b + F4/80 + Ly6C hi , CD11b + F4/80 + Ly6C int , and CD11b + F4/80 + Ly6C lo cells in peripheral blood from WT and C3 −/− mice. A no-stain control was used to exclude cells with spontaneous fluorescence. There was no significant difference in the proportions of macrophages/monocytes between WT and C3 −/− mice without treatment. Although CD11b + F4/80 + Ly6C hi cells circulating in peripheral blood in WT mice increased until day 7 after laser treatment, significantly higher proportions of CD11b + F4/80 + Ly6C hi cells were observed in C3 −/− mice than in WT mice at 1, 3, and 7 days after laser injury. There was no significant difference in the proportions of CD11b + F4/80 + Ly6C lo and CD11b + F4/80 + Ly6C int cells in blood between both groups during CNV formation. Hi, int, and lo correspond to CD11b + F4/80 + Ly6C hi , CD11b + F4/80 + Ly6C int , and CD11b + F4/80 + Ly6C lo cells, respectively. (B) Ratios of CD11b + F4/80 + Ly6C hi (region hi in A), CD11b + F4/80 + Ly6C int (region int in A), and CD11b + F4/80 + Ly6C lo (region lo in A) cells to total number of live cells. *P < 0.05 compared with the same subtype of WT mice. Six mice were used to give a single value; n = 3 in each experiment.
Scientific RepoRts | 5:15702 | DOi: 10.1038/srep15702 Statistical analysis. Unpaired t tests were performed to compare the differences between the groups. P values less than 0.05 were considered statistically significant. Figure 6. Fold changes in Vegfa164 and Vegfr1 expression levels before and after laser. Ly6C hi and Ly6C lo cells were sorted from the eyes of WT and C3 −/− mice without or with laser treatment (3 days after laser injury) by using flow cytometry. Real-time RT-PCR was performed. Vegfa164 expression was not detected in intraocular Ly6C hi or Ly6C lo macrophages/monocytes in both groups without laser treatment. In contrast, Vegfa164 expression levels were upregulated in intraocular Ly6C hi cells of C3 −/− mice after laser injury, but not as much as in those of WT mice. There was no significant difference in the expression levels of Vegfr1 in intraocular Ly6C hi cells between both groups before or after laser injury. Vegfr1 expression was not detected in intraocular Ly6C lo cells of both groups with or without laser treatment. *P < 0.05 compared with WT mice. Ten mice were used to give a single value; n = 3 in each experiment.