Apolipoprotein E2 and E3, but Not E4, Promote Retinal Pathologic Neovascularization

Abstract

Purpose: To determine the relationship between the different isoforms of apolipoprotein E (ApoE) and retinal neovascularization.

Methods: The concentrations of ApoE and VEGF in vitreous humor samples with either a macular hole (MH), or diabetic macular edema (DME), or proliferative diabetic retinopathy (PDR) with or without intravitreal injection of bevacizumab (IVB) were measured by ELISA. The effects of each isoform of ApoE on human retinal microvascular endothelial cells (HRMECs) in culture or on the retina of oxygen-induced retinopathy (OIR) mice were investigated.

Results: The concentrations of ApoE and VEGF were significantly higher in the vitreous humor of patients with PDR and DME than in patients with an MH. There was a significant positive correlation between the concentrations of ApoE and VEGF in vitreous humor of patients. In vitro assays showed that ApoE2 and ApoE3, but not ApoE4, promoted the VEGF-induced cell proliferation and migration. In vivo assays showed that intravitreal injections of ApoE2 and ApoE3 increased the number and area of nodes in the retina of OIR mice. Moreover, ApoE was expressed in the vascular endothelial cell in both normal and OIR retinas, but their expression levels were different at postnatal day (P) 12 and P17.

Conclusions: These results demonstrate that ApoE2 and ApoE3, but not ApoE4, have proangiogenic effects, and the increased expression of ApoE in the vitreous humor of patients with PDR and DME indicates that ApoE2 and ApoE3 are involved in the development of retinal neovascularization in eyes.

Diabetic retinopathy (DR) is the most frequent and consequential complication of diabetes mellitus and is the main cause of reduced vision in the working age population. The number of patients with diabetes will increase by 70% in developing countries and by 20% in industrialized countries by 2030.¹ Therefore, the number of patients with DR will no doubt increase, and finding an effective treatment for DR is essential. 

Diabetic retinopathy is caused by damage of the smaller retinal vessels, and it can progress through three stages: simple retinopathy, preproliferative DR (PPDR), and proliferative retinopathy (PDR). In the simple retinopathy stage, capillary aneurysms, ecchymoses, and hard exudate are seen in the fundus by ophthalmoscopy. In the PPDR stage, some of the capillaries are clogged, which results in ischemia of the retina, although most patients do not recognize this because the decrease of visual function is not so severe. In the PDR stage, retinal neovascularization develops to compensate for the retinal ischemia, and some of the vessels run into the vitreous cavity and can rupture and cause vitreous hemorrhages. In addition, proliferative membranes can form and cause retinal detachment by traction on the retina. The vitreous hemorrhage and retinal detachment cause eye floaters and severe reduction of vision.² In patients with DR associated with diabetic mellitus, the level of low-density lipoprotein (LDL) is very high because of abnormal lipid metabolism.³ Molecular mechanisms underlying the pathogenesis of PDR are still not completely understood; therefore, there is yet no efficient treatment for its management. 

The Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications cohort by Lyons' group⁴ has shown the link between retinopathy in type 1 diabetes and average lipoprotein particle diameters and the distribution of subclasses of high-density lipoprotein (HDL), LDL, and very low-density lipoprotein (VLDL). Moreover, they found that the circulating immune complexes containing oxidized-LDL in that cohort predicted severe DR.⁵ It also was reported that the intraretinal deposits of apoB-100 (the main protein component of LDL and VLDL), oxidized-LDL, and oxidized-LDL immune complexes were completely absent from healthy human retinas, but were present in diabetic retinas even before the onset of clinical retinopathy, and were associated with DR severity.^6,7 Furthermore, modified lipids have been consistently observed to promote oxidative and endoplasmic reticulum stress, inflammation, and death toward cultured human retinal cells, including capillary endothelial cells, pericytes, RPE cells, and Müller cells.^7–14 Recently, Yu's group reported that diabetes confers susceptibility to retinal injury imposed by intravitreal injection of modified LDL.¹⁵ These findings suggest that extravasated, modified plasma lipoproteins contribute to the propagation of DR. 

Apolipoprotein E (ApoE) is a 34-kDa glycoprotein and exists in three common isoforms: ApoE2, ApoE3, and ApoE4.¹⁶ The ApoEs are produced mainly in the liver and serve as ligands for the LDL receptor and its family members. The ApoEs are involved in lipid metabolism.^17,18 Apolipoprotein E is also expressed in the photoreceptor outer segments, the retinal ganglion layer, and Bruch's membrane.^16,19–22 It is also secreted by RPE cells, and is synthesized by Müller glial cells and secreted into the vitreous.²³ 

Apolipoprotein E2 transgenic targeted replacement mice have an accumulation and deposits of lipids in the RPE, which are typical features of AMD in humans.¹⁹ In ApoE4 targeted replacement transgenic mice, the laser-induced choroidal neovascularization was more extensive than in transgenic ApoE3 targeted replacement mice.²⁴ It has been reported that the VEGF level is significantly and positively correlated with the ApoE levels in the retina and choroid of murine eyes and also in the human brain.^24–27 Because VEGF has been shown to be associated with neovascularization, these findings suggest that ApoE may be related to retinal neovascularization. However, the effects of the ApoEs (e.g., ApoE2, ApoE3, and ApoE4) on retinal neovascularization have not been fully investigated. Thus, the aim of this study was to evaluate the effects of the ApoEs on retinal pathologic neovascularization focusing on the differences among the ApoE2, ApoE3, and ApoE4 isoforms. 

Materials and Methods

Animals

Eight-week-old male and female C57BL/6J mice (Japan SLC, Hamamatsu, Japan) were used. The mice weighed 20 to 30 g and were kept under 12/12-hour light/dark cycle. They had access to food and water ad libitum. All of the experimental procedures were approved by the Animal Experimental Committee of the Gifu Pharmaceutical University, and they were performed in accordance with the ARVO Statement on the Use of Animals in Ophthalmic and Vision Research. 

Preparation of Vitreous Samples

The clinical procedures conformed to the tenets of the Declaration of the Helsinki. After explaining the purpose and procedures of the study, informed consent was obtained from all patients. The study was approved by the Ethics Committee of the Gifu Pharmaceutical University and the Gifu University Graduate School of Medicine (application number 24-84). Vitreous humor was collected from 40 eyes with a macular hole (MH), 8 eyes with diabetic macular edema (DME), and 39 eyes with PDR just before pars plana vitrectomy. For the eyes with PDR, 15 had an intravitreal injection of bevacizumab (IVB+) and 24 eyes did not have an IVB (IVB−) before collection of vitreous samples. Demographics of patients (age, sex, LDL level, and HDL level) are listed in Supplementary Table S1. There were significant differences in age between the PDR with IVB and the eyes in the other groups (Supplementary Table S1). These differences were probably because younger patients with PDR tended to receive IVB treatment. There was a significant difference in HDL level between PDR (IVB−) group and DME group. There was no significant correlation between ApoE and LDL or HDL (data not shown). 

Enzyme-Linked Immunosorbent Assay

Samples of the vitreous humor were diluted by 10, 50, and 100 times. The concentration of ApoE in the vitreous humor of the patients was measured with Human Apolipoprotein E/ApoE Quantikine ELISA Kit (DAPE00; R&D Systems, Inc., Boston, MA, USA), and the VEGF with the Human VEGF Quantikine ELISA Kit (DVE00; R&D Systems, Inc.). The procedures were performed according to the manufacturer's protocol. In these assays, assay diluent, standard, control, or sample solution were added to each well that had polyclonal anti-ApoE antibodies coated on the bottom. The vitreous samples were incubated at room temperature for 3 hours on a microplate shaker. The wells were washed four times, and then horseradish peroxidase–conjugated monoclonal anti-ApoE antibody solution was added and incubated with shaking at room temperature for 1 hour. After washing, the substrate solution was added and incubated at room temperature for 30 minutes. Then, a stop solution was added to each well, and the absorbance was measured at 450 nm and 540 nm (as wave length correction) using VARIOSKAN FLASH (Thermo Fisher Scientific, Waltham, MA, USA) within 30 minutes of adding the stop solution. 

Cell Cultures

Primary human retinal microvascular endothelial cells (HRMECs) obtained from DS Pharma Biomedical (Osaka, Japan) were cultured in Cell Systems Corporation (CSC) complete recombinant medium with 10% fetal bovine serum (FBS) but without hormones, antibiotics, and phenol red (DS Pharma Biomedical). The medium contained culture boost growth factor (DS Pharma Biomedical), attachment factor (DS Pharma Biomedical), 100 U/mL penicillin (Meiji Seika, Tokyo, Japan), and 100 μg/mL streptomycin (Meiji Seika). The cultures were grown at 37°C in a humidified atmosphere with 5.0% CO_2. The cells were released by trypsinization and passaged every 3 to 4 days. Human retinal microvascular endothelial cells from passages 3 to 8 were used for the experiments. 

Cell Proliferation Assay

Cell proliferation was assayed as described in detail with some modification.²⁸ In brief, HRMECs were seeded at a density of 2.0 × 10³ cells per well in 96-well plates and incubated at 37°C in a humidified atmosphere of 5.0% CO₂ for 24 hours. After replacing the culture medium by CSC medium (DS Pharma Biomedical) with 10% FBS but without growth factor, the cells were further incubated for 24 hours. Apolipoprotein E2, ApoE3, or ApoE4 (Perotech, Inc., Rocky Hill, NJ, USA) was added to a final concentration of 0.03 to 3 μM and incubated for 1 hour before adding VEGF₁₆₅ (R&D Systems, Inc.) at a final concentration of 10 ng/mL. This VEGF concentration is the best for proliferation assay using HRMECs.^28–32 After 24 hours of incubation with VEGF₁₆₅, the medium was replaced by fresh CSC medium with 10% FBS to remove the ApoEs and VEGF. The number of living cells was determined by the WST-8 assay (Cell Counting Kit-8, Dojindo Kagaku, Kumamoto, Japan). The absorbance at 450 and 650 nm was measured with a microplate reader (Varioskan Flash 2.4; Thermo Fisher Scientific, Waltham, MA, USA). In this assay, WST-8 is turned orange by the dehydrogenase in living cells. The change in the differences in the absorption at 450 and 650 nm at 3 hours was taken to be the number of living cells. 

Migration Assays

Cell migration assays were performed as described earlier with some modification.²⁸ In brief, HRMECs were seeded at a density of 4.0 × 10⁴ cells per well in 12-well plates and incubated at 37°C in a humidified atmosphere of 5.0% CO₂ for 24 hours. After replacing the culture medium by CSC medium with 1% FBS and incubation for 6 hours, the HRMEC monolayer at the center of the well was scraped with a 1.0 × 10³ μL micro-tip. To remove the floating cells, the culture medium was replaced by CSC medium with 1% FBS again. Then, ApoE2, ApoE3, or ApoE4 was added to a final concentration 0.03 to 3 μM and incubated for 1 hour before adding VEGF₁₆₅ at a final concentration of 10 ng/mL. After adding the reagents, the cells were incubated at 37°C and 5.0% CO₂ for 24 hours. To determine the degree of migration, four photographs covering an area of 3.6 mm² each well were taken with a charge-coupled device camera (Olympus, Tokyo, Japan). The photographs were taken before and after the cell migration. The number of cells that migrated into the scraped area was counted. 

Oxygen-Induced Retinopathy (OIR) Model

Mice with OIR were created as described in detail with some modification.²⁸ Neonatal mice and their mothers were transferred to a custom-built chamber on postnatal day (P)7 and exposed to an atmosphere of 75% O₂ for 5 days with the oxygen level controlled by an oxygen controller (PRO-OX 110; Rerning Bioinstruments Co., Redfield, SD, USA). The pups were returned to room air on P12, and then human ApoE was injected intravitreally at a final concentration of 3 μM, which exerted promotive effects against cell proliferation and migration in the in vitro study. On P17, mice were deeply anesthetized with an intraperitoneal injection of 10 mg/kg sodium pentobarbital (Nembutal; Sumitomo-Dainippon Pharmaceutical Co. Ltd., Osaka, Japan), and 1 mL PBS containing 20.0 mg/mL FITC (MW 2000 kDa; Sigma-Aldrich Corp., St. Louis, MO, USA) was injected into the left ventricle. Then, the eyes were enucleated and fixed in 4% paraformaldehyde (PFA) for 8 hours. The retinas were isolated and flat-mounted with Fluoromount (Diagnostic Bio Systems, Pleasanton, CA, USA) on microscope slides. 

Imaging and Quantification of Retinal Neovascularization

Photographs of the entire flat-mounted retina were taken with the MetaMorph Microscopy Automation and Image Analysis Systems (Universal Imaging Corp., Downingtown, PA, USA). To evaluate the degree of pathologic neovascularization, the number and area of the nodes in the retina were determined with the Angiogenesis Tube Formation module of the MetaMorph Microscopy Automation and Image Analysis Systems (Universal Imaging Corp.). A node is a region containing connected blobs and is determined by the thickness and maximum width of the vessels. 

Immunohistochemistry

Eyes of OIR mice were enucleated on P17, and fixed in 4% PFA at 4°C for 24 hours. After fixation, the eyes were placed in optimum cutting temperature compound (Sakura Finetechnical Co., Ltd., Tokyo, Japan), frozen with liquid nitrogen, and stored at −80°C. For the analysis, transverse 10-μm serial sections were cut on a cryostat and mounted on microscope slides (MAS COAT; Matsunami Glass Ind. Ltd., Osaka, Japan). The sections were washed twice in 0.01 M PBS for 10 minutes each, followed by incubation in mouse-on-mouse IgG blocking reagent (Vector Laboratories, Burlingame, CA, USA) for 1 hour. Mouse anti-CD31 antibody (1:500 dilution; AnaSpec, Inc., San Jose, CA, USA) and rabbit anti-ApoE antibody (1:100 dilution; Bioworld Technology, St. Louis Park, MN, USA) were used. Anti-CD31 antibody was applied to the sections overnight at 4°C. The sections were blocked with 5% goat serum in PBS, and anti-ApoE antibody was applied overnight at 4°C. The immunoreactivity was made visible by incubating the sections with goat anti-mouse antibody conjugated with Alexa 546 (1:2000; Thermo Fisher Scientific) as the secondary antibody for 1 hour at room temperature. Secondary goat anti-rabbit antibody conjugated with Alexa 488 (1:2000; Thermo Fisher Scientific) was applied for 1 hour at room temperature. Finally, Hoechst 33342 (1:2000) was applied to the samples for 30 minutes. The fluorescent images were obtained by the Metamorph system. The tissue that was incubated without primary antibody and followed by incubation with secondary antibodies and detection reagents was treated as a negative control. 

Western Blotting

Eyes of normal and OIR mice were enucleated on P17, and the retinas isolated and quickly frozen in liquid nitrogen. To extract the proteins, the retinas were homogenized in RIPA buffer with cocktails 2 and 3 protease and phosphatase inhibitors (Sigma-Aldrich Corp.). Lysates of the retina were centrifuged at 12,000g for 20 minutes. The protein concentration of the supernatant was measured using BSA and a bicinchoninic acid protein assay kit (Pierce Chemical, Rockford, IL, USA). Five micrograms of the samples of normal and OIR retinas were dissolved in 20% 2-mercaptoethanol (Nacalai Tesque, Inc., Kyoto, Japan) for the Western blot analysis. The solution containing the samples was subjected to electrophoresis on 5% to 20% SDS-PAGE. Protein bands were electroblotted onto polyvinylidene difluoride (PVDF) membranes (Immobilon-P; Millipore Corporation, Billerica, MA, USA), and the membranes were blocked for 30 minutes at room temperature with Block One-P (Nacalai Tesque, Inc.). The membranes were then incubated overnight at 4°C with rabbit anti-ApoE primary polyclonal antibody (1:1000 dilution) and mouse anti-β-actin antibody (1:10000 dilution; Sigma-Aldrich Corp.) as a loading control. Goat anti-rabbit horseradish peroxidase–conjugated antibody (1:2000; Thermo Fisher Scientific) was used as a secondary antibody. The bands were made visible by Immuno Star LD (Wako Pure Chemical, Osaka, Japan), and Lumino Imaging Analyzer (LAS-4000 Mini; Fuji Film, Tokyo, Japan) and quantified. 

Statistical Analyses

Data are presented as the means ± SEMs. Statistical comparisons were performed by Student's t-tests, Dunnett's multiple comparison tests, or Mann-Whitney U tests with Bonferroni correction using statistical software SPSS (version 16.0J; IBM SPSS Statistics, IBM Corporation, Chicago, IL, USA). A significant difference between groups was set at P < 0.05. 

Results

Concentration of ApoE and VEGF in Vitreous Humor of Patients With DME, PDR, and an MH

We determined the concentrations of ApoE and VEGF in the vitreous humor of patients with DME, PDR, and an MH with ELISA kits. The expressions of both ApoE and VEGF were increased by approximately 3-fold in the patients with DME and PDR compared with that in eyes with an MH (Figs. 1A, 1B). In eyes with IVB, the concentration of VEGF was significantly lower (Fig. 1B). There was a positive and significant correlation between the concentrations of ApoE and VEGF in eyes of all patients (ρ = 0.530, P < 0.01; Spearman's rank correlation coefficient). The relationship between ApoE and VEGF can be described by the equation, y = 0.3052x + 2.7132, in which y is the logarithm level of ApoE in ng/mL and x is the logarithm level of VEGF in pg/mL. 

Figure 1

 

Concentrations of ApoE and VEGF in the vitreous humor of patients with an MH, DME, and PDR. (A, B) Concentrations of ApoE and VEGF in vitreous humor of patients with MH, DME, and PDR. The PDR samples for eyes that were treated with or without IVB were analyzed separately. (C) Relationship between VEGF and ApoE in eyes of all patients. The relationship between ApoE and VEGF can be described by the equation, y = 0.3052x + 2.7132 in which y is the logarithm level of ApoE in ng/mL and x is the logarithm level of VEGF in pg/mL. Data are shown as the means ± SEMs (MH, n = 40; DME, n = 8; PDR [IVB−], n = 24; PDR [IVB+], n = 15). ^*P < 0.008, ^**P < 0.002 versus MH, ^#P < 0.008 versus DME, ^$$P < 0.002, N.S. versus PDR (IVB−) (Mann-Whitney U test with Bonferroni correction).

Effect of ApoE on VEGF-induced Cell Proliferation and Migration

We determined the effect of ApoE on VEGF-induced cell proliferation and migration in HRMECs in culture. Both ApoE2 and ApoE3 increased the VEGF-induced proliferation of HRMECs (Figs. 2A, 2B). Apolipoprotein E2 and ApoE3 also significantly enhanced the VEGF-induced cell migration (Figs. 3A, 3B). On the other hand, ApoE4 tended to decrease VEGF-induced cell proliferation (Fig. 2C), and had no effect on the VEGF-induced cell migration (Fig. 3C). 

Figure 2

 

Effects of ApoE on VEGF-induced cell proliferation. The effects of ApoE2 (A), ApoE3 (B), and ApoE4 (C) on VEGF-induced proliferation of HRMECs are shown. Data are the means ± SEMs (n = 6). ^**P < 0.01 versus control (Student's t-test); ^**P < 0.01 versus control group (Student's t-test); ^##P < 0.01 versus vehicle-treated group (Dunnett's test); ^$P < 0.05, ^$$P < 0.01 versus control group (Dunnett's test).

Figure 3

 

Effects of ApoE on VEGF-induced cell migration. The effects of ApoE2 (A), ApoE3 (B), and ApoE4 (C) on the VEGF-induced migration of HRMECs. Representative images are shown in (A[a]), (B[a]), and (C[a]). Scale bar represents 100 μm. The values are the means ± SEM (n = 6). ^**P < 0.01 versus control (Student's t-test); ^**P < 0.01 versus control group (Student's t-tests); ^#P < 0.05 versus vehicle-treated group (Dunnett's test); ^$$P < 0.01 versus control group (Dunnett's test).

Effect of ApoE on Retinal Neovascularization in OIR Mice

To investigate the effects of ApoE on retinal neovascularization, the number and size of the areas of the nodes were determined in OIR and control mice. The number and sizes were both significantly increased in the ApoE2-treated group compared with the vehicle-treated group (Figs. 4A, 4C). In addition, the number and area of nodes were also significantly increased in the ApoE3-treated group (Figs. 4B, 4D). 

Figure 4

 

Effects of ApoE on retinal neovascularization in OIR mice. Photomicrographs of flat-mounted retinas obtained on P17 from vehicle- and ApoE2-treated groups are shown in (A[a]) and (A[c]), respectively. Magnified images are shown in (A[c]) and (A[g]) and the analyzed images are shown in (A[b]), (A[d]), (A[f]), and (A[h]). The flat-mounted retinas obtained on P17 in vehicle- and ApoE3-treated groups are shown in (B[a]) and (B[c]), respectively. The magnified images are shown in (B[c]) and (B[g]) and the analyzed images are shown in (B[b]), (B[d]), (B[f]), and (B[h]). Greenish areas in the analyzed images represent the node regions that are indices of pathologic neovascularization. The quantified results of the number and size of the area of the nodes are shown in (C) and (D). Data are the means ± SEMs ([C], n = 5; [D], n = 8). ^*P < 0.05 versus vehicle-treated group (Student's t-tests). Scale bars: 250 and 500 μm.

Expression and Localization of ApoE and CD31 in Retinas of OIR Mice

Apolipoprotein E and CD31, a marker for endothelial cells, were colocalized in the vascular endothelial cells in both normal and OIR retinas (Fig. 5A). The expression of ApoE in the OIR retina was upregulated at P12 but was downregulated at P17 (Fig. 5B). 

Figure 5

 

Expression and localization of ApoE in retinas of normal and OIR mice. (A) Photomicrographs of immunostained retinas of normal and OIR mice group are shown. Areas indicated by white arrows are magnified at the bottom right. Hoechst33342, cyan; ApoE, green; CD31, red. (B) Western blots of retina of ApoE on normal and OIR retinas. Western band images and the quantified results are shown: (a, b) at P12; (c, d) at P14; (e, f) at P17. Data are the means ± SEM (P12, n = 6; P14, n = 6; P17, normal, n = 8; OIR, n = 9). ^*P < 0.05 versus normal group (Student's t-tests). Scale bar represents 50 μm.

Discussion

Our results showed that the expressions of ApoE and VEGF were upregulated in the vitreous of eyes with DME and PDR, but not eyes with an MH, and there was a positive and significant correlation between the levels of ApoE and VEGF in eyes of all patients. In addition, we confirmed that ApoE2 and ApoE3 had proangiogenic effects in vitro and in vivo, and the ApoEs were partially expressed in vascular endothelial cell retina of normal and OIR mice. Interestingly, the expression of the retinal ApoEs was increased at P12 just after returning the pups to normal oxygen levels and was decreased at P17 compared with that in normal mice. 

Human ApoEs have three major isoforms, ApoE2, ApoE3, and ApoE4, that are encoded by ε2, ε3, and ε4 alleles. These isoforms have different amino acid residues that determine the sites of the receptor-binding domain, positions 112/158; ApoE2 at Cys/Cys, ApoE3 at Cys/Arg, and ApoE4 at Arg/Arg. These polymorphisms control the receptor and lipid-binding affinities of the three isoforms of ApoE, and they differ for different for cellular responses.^33,34 Apolipoprotein E binds to the LDL receptor family members, including the LDL receptor, LDL receptor-related protein 1 (LRP1), VLDL receptor, and ApoE receptor 2 (ApoER2, LRP8). The expression of LDLR, VLDLR, and LRP1 in endothelial cells has been reported by many researchers.^35–40 Moreover, we confirmed the expression of ApoER2 in HRMECs (Supplementary Fig. S1). These findings suggest that there is a possibility that the interaction of ApoE and each receptor is included in proangiogenic effects. 

Assay by ELISA showed that amount of ApoE remained at high concentrations (Fig. 1). Moreover, in an in vitro assay, ApoE2 and ApoE3 promoted cell proliferation and migration even without VEGF. These findings indicate that ApoE has a proangiogenic mechanism not only dependent but also independent of VEGF. 

It was recently reported that ApoE3 stimulates endothelial nitric oxide synthase (eNOS) and endothelial cell migration by ApoER2 and kinase/Akt on PI3.⁴¹ On the other hand, ApoE4 antagonizes ApoE3/ApoER2 binding and does not stimulate eNOS or endothelial cell migration.⁴¹ In addition, intracellular NO production was increased in human umbilical vein endothelial cells by cell-derived human ApoE2 and also in EA.hy926 human endothelial cells by cell-derived human ApoE2 and ApoE3.^42,43 Because eNOS plays an important role in angiogenesis,^44–47 the proangiogenic effects of ApoE2 and ApoE3 may act by activating eNOS. We performed the cell proliferation assay with NOS antagonist N omega-nitro-L-arginine methyl ester (L-NAME). As same as data in Figure 1, ApoE2 and ApoE3 increased the cell proliferation and L-NAME pretreatment inhibited it (Supplementary Fig. S2). Moreover, Ulrich et al.⁴¹ demonstrated that L-NAME inhibited the ApoE3-induced cell migration. These data indicate that eNOS is partly involved in ApoE-induced cell proliferation. 

Apolipoprotein E interacts with HDL, whereas ApoE2 and ApoE3 bind with HDL more than ApoE4.^48,49 It was recently reported that HDL interacts with the scavenger receptor, SR-B1, which then activates the PI3K pathway and HIF-1α/VEGF pathway, resulting in hypoxia-induced angiogenesis.^50,51 Thus, under hypoxic conditions, HDL enhances the expressions of HIF-1α, VEGF, and VEGFR2. Indeed, the expressions of VEGF and BFGF, both potent proangiogenic factors,^52–54 were also stronger in eyes of human ApoE2 transferred replacement transgenic mice than in control C57BL/6J mice.⁵⁵ Furthermore, the retinal VEGF level is lower in human ApoE4 transgenic mice than human ApoE3 transgenic mice,^24,27 and ApoE3 treatment increases the level of VEGF in the RPE cell culture medium.⁵⁶ These findings indicate that ApoE2 and ApoE3, but not ApoE4, have the potential of increasing the level of ocular VEGF. Earlier, Antes et al.²⁴ reported that the laser-induced choroidal neovascularization was more exacerbated in ApoE4 transgenic mice than in ApoE3 transgenic mice. However, neonatal ApoE4 transgenic mice (P4–P7) have low levels of ApoE and an increase in the vascular density, buds, and branching compared with ApoE3 transgenic mice.²⁷ These findings are present because ApoE4 may have protective effects, and its expression is increased to protect the retina from laser-induced choroidal neovascularization. Indeed, ApoE4 blocked the ApoE3-induced NOS activity, cell migration, and phosphorylation of Akt. Moreover, the anti-VEGF treatment outcome is better in neovascular AMD patients who have ApoE ε4 allele compared with those who have the ApoE ε2 allele. It has been reported that HDL activates the PI3K/Akt/NO pathway to promote the differentiation of peripheral blood monocytes into endothelial progenitor cells.⁵⁷ Based on these findings, the proangiogenic effects of ApoE2 and ApoE3 may control eNOS activation by binding to HDL; however, it is known that the HDL level of humans is different from that of mice: human HDL is much lower. Therefore, the hypothesis including HDL might be not applicable. 

Another possibility as the mechanism is heparin binding. In the study by Yamauchi et al.,⁵⁸ they demonstrated that ApoE4-heparin binding was greater than ApoE2- and ApoE3-heparin binding in both lipid-free and lipidated states. There is a possibility that ApoE2 and ApoE3 have poor bonding to heparan sulfate proteoglycan (HPSG) and great binding to cell surface receptor, and then these exhibited additive effects with VEGF. On the other hand, ApoE4 has great binding to HSPG (including free HSPG) competing with VEGF and poor binding to cell surface receptor or binds different receptor from ApoE2 and ApoE3 binding receptor. However, this hypothesis is completely conjecture, and therefore further study will be needed. 

Under normal conditions, the expression of ApoE in the murine retina begins to increase on P7.²² In contrast, our results showed that the expression of “retinal” ApoE in OIR mice was higher at P12, the day of replacement to normoxia (Day 0 [D0]) and lower at P17 (D5) compared with that of mouse retinas reared under normal conditions (Fig. 5B). In addition, the results of ELISA showed that the concentration of ApoE in the “vitreous humor” of DME and PDR patients was higher than in patients with an MH (Fig. 1A). These findings suggest that under hypoxic and inflammatory conditions such as OIR, DME, and PDR, ApoE is upregulated in the retina and might be released into the vitreous humor. 

In the present study, we demonstrated promotive effects of ApoE2 and ApoE3 on angiogenesis. These findings suggest that there is a possibility that retinal angiogenesis in the eyes of patients who have ε2 allele or ε3 allele might be promoted by ApoE, which remains a high concentration in the eyes even given anti-VEGF treatment. On the other hand, there is a possibility that patients who have ε4 allele obtain the maximum benefits of anti-VEGF treatment. Indeed, it has been reported that in wet-type AMD therapy, prognosis of patients with ε4 allele is better than that of patients with the ε2 allele.⁵⁹ However, there are some differences in ApoE expression, receptor expression, and lipid metabolism between humans and mice, and between human ApoE targeted replacement mice used in a previous report.^{24,27,60–63} Therefore, further studies focusing on alleles or isoforms of patients are needed to reveal the role of ApoE in DR completely. 

In conclusion, ApoE2 and ApoE3, but not ApoE4, have proangiogenic effects, which is in keeping with the increased expression of ApoE in the vitreous of patients with PDR and DME. These findings indicate that ApoE2 and ApoE3 most likely play important roles in the pathogenesis for PDR, which are characterized by retinal neovascularization. 

Acknowledgments

The authors thank the patients and families who participated in this study. 

Disclosure: T. Masuda, None; M. Shimazawa, None; Y. Hashimoto, None; A. Kojima, None; S. Nakamura, None; S. Suemori, None; K. Mochizuki, None; H. Kawakami, None; K. Kawase, None; H. Hara, None