Delayed regression of laser‐induced choroidal neovascularization in TNFα‐null mice

We investigated the effects of lacking TNFα on the development and regression of Argon‐laser‐induced choroidal neovascularization (CNV) in mice. We lasered ocular fundus for induction of CNV in both wild‐type (WT) and TNFα‐null (KO) mice. Fluorescence angiography was performed to examine the size of CNV lesions. Gene expression pattern of wound healing‐related components was examined. The effects of exogenous TNFα on apoptosis of human retinal microvascular endothelial cells (HRMECs) and on the tube‐like structure of the cells were investigated in vitro. The results showed that Argon‐laser irradiation‐induced CNV was significantly larger in KO mice than WT mice on Day 21, but not at other timepoints. Lacking TNFα increased neutrophil population in the lesion. The distribution of cleaved caspase3‐labelled apoptotic cells was more frequently observed in the laser‐irradiated tissue in a WT mouse as compared with a KO mouse. Exogenous TNFα induced apoptosis of HRMECs and accelerated regression of tube‐like structure of HRMECs in cell culture. Taken together, TNFα gene knockout delays the regression of laser‐induced CNV in mice. The mechanism underlying the phenotype might include the augmentation of neutrophil population in the treated tissue and attenuation of vascular endothelial cell apoptosis.

In vivo research that targets AMD, macular neovascularization is usually modelled by choroidal neovascularization (CNV) induced by Argon laser irradiation in mice. Inflammation with neutrophils and macrophages plays a crucial role in the development of CNV. 4 Immune complex deposition and complement activation under the condition of tissue inflammation were proposed as important mediators of macular neovascularization or CNV growth. [5][6][7] We reported that the loss of Smad3, the major transforming growth factor β (TGFβ) signalling transmitter, suppressed the growth of argon laserinduced CNV in association with inhibition of local tissue inflammation in mice. 8 Infiltration of macrophages was detected in a surgically removed CNV tissue, and the macrophages were labelled for not only VEGF but also one of the major pro-inflammatory cytokines, tumour necrosis factor α (TNFα), suggesting that TNFα also contributes to the pathophysiology of CNV in AMD. 9 Blocking TNFα activity by administrating a neutralizing antibody exhibits therapeutic potential in inflammatory fibrotic diseases in various organs, for example, lung or joints in mice. 10,11 These reports strongly suggest that partial attenuation of TNFα signalling slowdowns the process of inflammatory fibrosis in pathological conditions, for example, rheumatoid arthritis. 12,13 Local administration of anti-TNFα antibody inhibits the development of laser-induced CNV in mice. [14][15][16] In human patients, Shi et al. reported that a TNFα-trapping peptide, etanercept, or a TNFα antibody reduced laser-induced CNV. 17 Systemic or intracameral infliximab, an anti-TNFα monoclonal antibody, was reported to be effective against age-related AMD 15,16 but did not exhibit an additional effect in patients treated with anti-VEGF antibody. 18 TNFα exerts a variety of biological activities by signalling via its two receptors, TNFR-1 and TNFR-2. Jasielska et al. reported that lacking TNFR-1 did not suppress, or even promotes, the development of laser-induced CNV and that deletion of TNFR-2 gene suppresses its growth. 19 In other tissue of the body, blocking TNFα is a powerful therapeutic option in the suppression of immune-related inflammatory disorders.
On the contrary, complete inhibition of TNFα signalling by ligand gene knockout reportedly paradoxically enhances the inflammatory process in a variety of disease models depending on tissues in animals. 20,21 It was also reported that TNFα accelerates the resolution of tissue fibrosis in a mouse lung. 22 TNFα in lymphocytes suppress tissue fibrosis in a mouse kidney. 23 We previously showed that the loss of TNFα promotes excess inflammation and resultant tissue scarring in corneal stroma during healing following an alkali burn in mice. 24 These reports suggest that partial neutralization of TNFα ligand does not recapitulate the phenotype by total loss of TNFα by gene knockout.
In the current study, we investigated the effects of the loss of TNFα on the development and regression of argon-laser-induced CNV in mice in order to better understand the roles of this growth factor in CNV pathobiology. These reports above mentioned promoted us to hypothesize that complete deletion of TNFα could lead to excess or could not affect the growth of CNV following laser irradiation in mice. The main outcome indicated that lacking TNFα gene does not suppress the growth of CNV but delays the regression of the CNV tissue in mice.

| Experimental argon-laser-induced CNV model in mice
Argon laser irradiation-induced CNV is the established mouse model for human AMD, although a mouse does not have macula. We irradiated ocular fundus for induction of CNV by using Argon laser in both male WT (n = 25) and male KO (n = 25) mice according to the procedure, we previously reported with a minor modification. 8  FITC: PBS = 1:1 weight ratio) was introduced into the cardiac cavity according to our previous publication. 8 The mice were killed, and the eyes were enucleated. 8 CNV in a flat-mounted choroidal specimen was then observed under Apotome2 fluorescent microscope (Zeiss, Germany) after removing the cornea, lens and sclera. 8 The hyperfluorescent tissues of CNV lesions (referred to as the size of CNV) were measured, and the total area of CNV lesions per eye was determined as their sum. 8 A constant threshold in pixels (corresponding to threshold fluorescence) was used to quantify neovascularization.
The total hyperfluorescent area of CNV was measured by using the software of WinROOF (Mitani, Japan). 8 In brief, FITC-visualized CNV was shown in RGB colour in the software of WinROOF and the area of FITC staining (not the whole area, but just the vessels) was determined. A two-sample Student's t-test with unequal variance was used for statistical analyses of the quantitative CNV flat-mount data.
Experiments were performed according to the previous publications by the authors and others. 8,24

| Gene expression analysis
Uninjured mice and the mice on Days 1, 3, 5, 7 and 9 days after the laser irradiation were sacrificed and the eye was enucleated. Ten WT and 10 KO mice (20 eyes at each timepoint in each genotype of mice) were used at each day-point. Chorio-sclera complex tissue was isolated by removal of the cornea, lens and retina because choroidal tissue and sclera tightly adhered to each other. We consider mRNA derived from sclera does not significantly affect the data because cellular components in the sclera was minimal. The  All the samples were fixed in 4% paraformaldehyde for 10 min and processed for immunohistochemistry with a rat monoclonal anti-LY6B.2 antibody. Alloantigen Antibody [a marker for a neutrophil leucocyte, 1:100 dilution in phosphate-buffered saline (PBS), Bio-Rad Laboratories] as previously reported. 8 It was reported that apoptotic cell death was observed following the peak of the growth of laser-induced CNV in mice. 2

| mRNA Expression of TGFβ1 or VEGF-A in macrophages of neutrophils
Mouse macrophages were obtained from the peritoneal cavity using a glycogen stimulation method, as previously reported by us. 24 In brief, 5% sterilized oyster glycogen (Sigma-Aldrich) in phosphatebuffered saline (PBS, 1 ml) was injected into the peritoneal space of either a WT (n = 5) or KO (n = 5) mouse aged 6-8 weeks. After 4 days, the peritoneal cavity was irrigated with Eagle's medium to harvest macrophages. Approximately 90% of the cells obtained by this method were positive for F4/80 as previously reported by us. 24 The cells were allowed to adhere to 60 mm culture dishes for 6 h, and then, nonadherent cells were washed out with PBS. RNA extracted from the adherent cells (macrophages) were processed TaqMan realtime RT-PCR for mRNA of TGFβ1 or VEGF-A. Three specimens were prepared for each condition. Five dishes were prepared for each culture condition.
Mouse neutrophils were harvested as follows. 24 5% (w/v) sterilized casein (Sigma-Aldrich) in phosphate-buffered saline (PBS, 2 ml) was injected into the peritoneal space of either a WT (n = 5) or KO (n = 5) mouse aged 6-8 weeks. After 3 h, the peritoneal cavity was irrigated with Eagle's medium to harvest neutrophils. Previous publications reported 26 that approximately over 90% of the cells obtained were neutrophils, which was also confirmed Giemsa staining (data not shown). The cells were allowed to adhere to 60 mm culture dishes for 6 h, and then nonadherent cells were washed out with PBS. RNA extracted from the adherent cells (neutrophils) were processed TaqMan real-time RT-PCR for mRNA of VEGF-A. Five specimens were prepared for each condition. Data at each time point were statistically analysed using the Mann-Whitney U test.

| Culture of a human retinal microvascular endothelial cell (HRMEC)
Frozen HRMECs (Commercially available primary cultured cells

| Effects of exogenous TNFα on the tube-like structure of HRMECs in vitro
HRMECs (1 × 10 6 cells/25 cm 2 ) were cultured in a T75 culture bottle for 24 h. At 80% confluency, the cells were washed in PBS and then further incubated for 30 min in a serum-free medium containing 2 μM Calcein AM. The cells were then trypsinized and were collected by a centrifugal separator (1000 × g, 5 min). Cultrex® Reduced Growth Factor Basement Membrane Extract (10 μl, Trevigen®) was placed in wells of a 96-well plate and allowed to set for 30 min in a CO 2 incubator. Cells (2 × 10 4 cells in 70 μl/well) were seeded and incubated for 10 h for tube-like structure formation. Then, TNFα at the final concentrations of 1, 5 or 10 ng/ml was administered, and the morphology of the tube-like structure was observed until 10 h under time-lapse microscopy. Four wells were prepared for each culture condition. The total length of the tube-like structure was measured as pixels in each condition at time points by an observation under microscopy (Keyence, BZ-X700) and statistically analysed by using the Mann-Whitney U-test.

| Experimental argon-laser-induced CNV model
CNV in a flat-mounted choroidal specimen was then observed under Apotome2 fluorescent microscope (Zeiss) after removal of the cornea, lens and sclera. As previously reported, laser irradiation successfully induced CNV (Figure 1a). 8 The hyper-fluorescent tissues (not the whole area, but just the vessels) of CNV lesions (referred to as the size of CNV) were measured, and the total area of CNV lesions per eye was determined as their sum as described above. 8 (Figure 1b,c). On Day 56, CNV was no longer observed in either genotype of mice (data not shown).

| Growth factor expression in macrophages and in neutrophils
Expression of VEGF-A was un unaffected by the loss of TNFα in cultured macrophages. That was, in turn, suppressed in a KO neutrophil as compared with each of their WT counterparts (Figure 4).

| Apoptosis in CNV tissue and TNFα-treated HRMEC culture
Immuno

| Effects of exogenous TNFα on tube-like structure of HRMECs in vitro
The effect of TNFα addition on the tube-like structure of HRMECs was observed. We previously reported human umbilical vein endothelial cells on the fibroblast feeder layer did not form the tube-like structure in the TNFα-containing medium. 20 Here, we administered

| DISCUSS ION
The size of the CNV was significantly larger in a KO mouse as compared with a WT mouse at week 3, that is, during the regression period after the peak of the size of the CNV. The finding suggests either that the loss of TNFα either promotes the growth of new vessels or that it inhibits the regression of CNV. Although it is to be noted that the loss of an inflammatory growth factor does not suppress the growth of CNV, our original hypothesis was TNFα deletion might promote its growth.
We previously reported that TGFβ/Smad-dependent infiltration of macrophages is critical to the development of laserinduced CNV in mice. 8   almost disappeared in the laser-irradiated tissue in our model of a WT mouse, as previously reported. 27 In an injured tissue, infiltration of neutrophils in the early phase of the tissue response to injury is followed by the invasion of macrophages, both of which are capable of secretion of angiogenic factors. 28 Neutrophils and macrophages contribute to the development of neovascular tissue in various settings. [29][30][31][32][33][34][35] It was reported that neutrophil depletion induced by an anti-murine neutrophil-antibody injection reduced CNV formation in mice. 36 Thus, the promotion of growth of CNV by the loss of TNFα could be attributable to neutrophil infiltration. However, the mechanism of more marked infiltration of neutrophils in the KO tissue is to be clarified. Inflammatory cells are considered to disappear from the local tissue, presumably by the mechanism of apoptosis. TNFα, as a pro-inflammatory cytokine, exerts its multiple biological activities by signalling via its two receptors, TNFR-1 and TNFR-2. 37 Gene knockout of TNFR-2 reportedly suppressed the growth of laser-induced CNV in association with less macrophage infiltration to the local tissue in mice, while TNFR-1-null mice exhibited a larger CNV as compared with a WT mouse. 19 Murray et al. reported that TNFα induces apoptosis in neutrophils mainly via TRFR-1. 38 This is consistent with a report that only TNF-R1 contains a cytoplasmic death domain and may directly induce apoptosis. 39 Ugan et al. reported that anti-TNFα antibody administration blocks neutrophil apoptosis in patients with ankylosing spondylitis. 40 Based on the findings in these reports by investigators, we consider that the increased population In the present study, we evaluated mRNA expression level of angiogenic factors, for example, IL-6 and VEGF-A and TGFb1, and a major factor involved in macrophage infiltration, MCP-1.
MMP2/9 are both involved in neovascularization development and growth. 41,42 Neutrophil reportedly expresses MMP9 that is a promoting factor for neovascularization. 43 The expression of MMP2  after treatment. 48 We, therefore, examined the degree of the formation of fibrovascular tissue in the area of CNV in a 35 days sample by using immune-detection of αSMA and collagen type I, the two major fibrosis markers, and did not observe the difference of the protein expression pattern of these components between WT and KO mice. We previously reported that TNFα-null eye cornea showed more severe inflammatory fibrosis during healing after an alkali exposure, 24 which differs from the current CNV model. The reason for discrepancy is to be investigated, but the severity of tissue damage F I G U R E 6 Effect of exogenous tumour necrosis factor α (TNFα) on the vessel-like structure developed by human retinal microvascular endothelial cells (HRMECs). Cultrex RGF BME (10 μl) was placed in wells of a 96-well plate and allowed to set for 30 min in a CO2 incubator. HRMEC cells (2 × 10 4 cells in 70 μl/well) were seeded and incubated for 10 h for induction of tube-like structure. The cultures were incubated for 10 h in the presence or absence of TNFα (1, 5, 10 ng/ ml). in these two injury models could affect the degree of tissue fibrosis in the later phase of healing.

CO N FLI C O F I NTE R E S T S
None for all authors.

FU N D I N G I N FO R M ATI O N
This study was supported by the competitive research grant from Bayer Japan (Grant #; BASJ20160403005).

DATA AVA I L A B I L I T Y S TAT E M E N T
Data available on request from the authors.