Erianin inhibits high glucose-induced retinal angiogenesis via blocking ERK1/2-regulated HIF-1α-VEGF/VEGFR2 signaling pathway

Erianin is a natural compound found in Dendrobium chrysotoxum Lindl. Diabetic retinopathy (DR) is a serious and common microvascular complication of diabetes. This study aims to investigate the inhibitory mechanism of erianin on retinal neoangiogenesis and its contribution to the amelioration of DR. Erianin blocked high glucose (HG)-induced tube formation and migration in choroid-retinal endothelial RF/6A cells. Erianin inhibited HG-induced vascular endothelial growth factor (VEGF) expression, hypoxia-inducible factor 1-alpha (HIF-1α) translocation into nucleus and ERK1/2 activation in RF/6A and microglia BV-2 cells. MEK1/2 inhibitor U0126 blocked HG-induced HIF-1α and ERK1/2 activation in both above two cells. In addition, erianin abrogated VEGF-induced angiogenesis in vitro and in vivo, and also inhibited VEGF-induced activation of VEGF receptor 2 (VEGFR2) and its downstream cRaf-MEK1/2-ERK1/2 and PI3K-AKT signaling pathways in RF/6A cells. Furthermore, erianin reduced the increased retinal vessels, VEGF expression and microglia activation in streptozotocin (STZ)-induced hyperglycemic and oxygen-induced retinopathy (OIR) mice. In conclusion, our results demonstrate that erianin inhibits retinal neoangiogenesis by abrogating HG-induced VEGF expression by blocking ERK1/2-mediated HIF-1α activation in retinal endothelial and microglial cells, and further suppressing VEGF-induced activation of VEGFR2 and its downstream signals in retinal endothelial cells.

The non-migrated cells were wiped off with a cotton swab, and the migrated cells were stained by crystal violet. Images were taken under a microscope. (D) The migrated cells were quantified by manual counting, and presented as percentage of control (n = 3). a Control, b HG (25 mM), c HG + erianin (1 nM), d HG + erianin (5 nM), e HG + erianin (10 nM), f HG + erianin (25 nM), g HG + erianin (50 nM). (E) RF/6A cells were incubated with different concentrations of erianin (0-50 nM) for 48 h. Cell viability were determined by MTT assay (n = 5). Data = Means ± SEM. **P < 0.01, ***P < 0.001 vs. control; ## P < 0.01, ### P < 0.001 vs. HG. the inhibitors coming from natural products for retinal neovascularization including curcumin, isoliquiritigenin, genistein, and all those natural products may be used as the leading compound for designing more efficacious and safe drugs for the treatment of ocular neovascular diseases including DR 18 . Here, we investigated the attenuation of erianin on DR by inhibiting retinal angiogenesis in vitro and in vivo, along with the elucidation of the signaling pathways that may be involved.

Erianin inhibited HG-induced tube formation and migration in RF/6A cells. Hyperglycemia
produced during the development of diabetes will cause damage and promote angiogenesis in retinal endothelial cells 19,20 . Firstly, we observed the effects of erianin on high glucose (HG)-induced angiogenesis in retinal endothelial cells, and the monkey choroidal retinal vascular endothelial (RF/6A) cell line was used in this study. As compared with osmotic mannitol (25 mM) control group, the tube formation (Fig. 1A,B) and cell migration (Fig. 1C,D) were both enhanced in HG (25 mM)-treated RF/6A cells. However, erianin abrogated HG-induced both tube formation (Fig. 1A,B) and cell migration (Fig. 1C,D) in RF/6A cells in the concentration-dependent manner. In addition, erianin (1-50 nM) alone had no obvious cytotoxicity on RF/6A cells after incubated with cells for 48 h (Fig. 1E). The results imply that the inhibition of erianin on HG-induced tube formation and migration in RF/6A cells was not due to its cytotoxicity. We also observed the inhibitory effect of bevacizumab (Avastin) The quantitative densitometric analysis of VEGF and actin, and the results were presented as percentage of control (n = 3). (D) Erianin abrogated HG-induced HIF-1α translocation from cytoplasm into nucleus. Representative blots for HIF-1α in cytoplasm or nucleus, and the results represent six independent experiments. (E) The quantitative densitometric analysis of HIF-1α , lamin B1, and actin, and the results were presented as percentage of control (n = 6). Data = Means ± SEM. *P < 0.05, **P < 0.01 vs. control; # P < 0.05, ## P < 0.01, ### P < 0.001 vs. HG. on HG-induced RF/6A cell tube formation in vitro. The results ( Supplementary Fig. 2) showed that bevacizumab (0.5-10 mg/ml) inhibited HG-induced tube formation in RF/6A cells in the concentration-dependent manner, and the inhibition of erianin (50 nM) was better than 5 mg/ml bevacizumab, but weaker than 10 mg/ml bevacizumab. The molecular weight of bevacizumab is about 149 KDa. After conversion, 5 mg/ml bevacizumab is about 33.6 μ M bevacizumab. So, we can see that the inhibition of erianin on HG-induced tube formation in vitro is better than bevacizumab.

Erianin abrogated retinal neovascularization in STZ-induced hyperglycemic mice. Staining
with CD31 is generally used to indicate the presence of vessels. Fig. 7A is the picture of the whole mounting of CD31-stained retinas, while Fig. 7B is the partial enlarged picture of CD31-stained retinas. From Fig. 7B, we can see that there were more CD31-stained vessels in STZ-induced hyperglycemic mice ( Fig. 7B-b) than in normal control mice ( Fig. 7B-a), whereas erianin (1, 10 mg/kg) reduced the increased retinal vessels in STZ-induced hyperglycemic mice (Fig. 7B-c,d). After counting the number of retinal vessels (Fig. 7C), we found that retinal vessels were increased in STZ-induced hyperglycemic mice, but erianin (1, 10 mg/kg) reduced such increase. Figure 7D showed the morphological changes of retinas in mice. There were more vessels appeared in GCL (ganglion cell layer), INL (inner nuclear layer) and OPL (outer nuclear layer) in STZ-induced hyperglycemic mice than in normal control mice ( Fig. 7D-a,b). After erianin (1, 10 mg/kg) treatment, the increased vessels were reduced ( Fig. 7D-c,d). Next results showed that VEGF content in serum and vitreous cavity were both obviously increased in STZ-induced hyperglycemic mice, however erianin (1, 10 mg/kg) reduced the increased VEGF content in both serum and vitreous cavity (Fig. 7E,F). In addition, erianin (1, 10 mg/kg) also reduced the increased mRNA expression of VEGF in retinas in STZ-induced hyperglycemic mice (Fig. 7G). However, erianin had no effect on blood glucose concentration and body weight in STZ-induced hyperglycemic mice (Supplementary inhibited HG-induced ERK1/2 activation in BV-2 cells. Representative blots for phospho-ERK1/2 and ERK1/2, and the results represent five independent experiments. (F) The quantitative densitometric analysis of phospho-ERK1/2 and ERK1/2, and the results were presented as percentage of control (n = 5). (G) U0126 inhibited HG-induced HIF-1α translocation from cytoplasm into nucleus in BV-2 cells. Representative blots for HIF-1α in cytoplasm or nucleus, and the results represent five independent experiments. (H) The quantitative densitometric analysis of HIF-1α , lamin B1, and actin, and the results were presented as percentage of control (n = 5). Data = Means ± SEM. **P < 0.01, ***P < 0.001 vs. control; # P < 0.05, ## P < 0.01, ### P < 0.001 vs. HG. Fig. 5). Ionized calcium-binding adapter molecule 1 (Iba1) is an often used biomarker for microglia 21 . Next, we observed the expression of Iba1 in retinas from STZ-induced hyperglycemic mice by using Iba1 immunofluorescence. As shown in Fig. 7H, the number of Iba1-positive microglia was increased in OPL, the inner plexiform layer (IPL) and GCL in STZ-induced hyperglycemic mice than in normal control mice. Moreover, such increase was reduced in erianin (1, 10 mg/kg)-treated mice.
Erianin attenuated retinal neovascularization in OIR mice. The above results implied that erianin-induced alterations in retinal vessels can be achieved without a reduction of hyperglycemia in STZ-induced hyperglycemic mice. To further confirm whether the amelioration of erianin on DR occurred via the direct effects on the retina but was not due to the reduced blood glucose, nondiabetic OIR mice were used, which is a model of ischemia-induced retinopathy. Erainin was administered intraperitoneally once a day for 5 days from P12 in OIR mice. Figure 8A is the picture of the whole mounting of CD31-stained retinas at P17, while Fig. 8B is the partial enlarged picture of CD31-stained retinas. The results showed that erianin (20 mg/kg)    (Fig. 8A,B). After calculating the vasoobliterative area at the central area of retina, we found that erianin (20 mg/kg) reduced the increased vasoobliterative area in retinas from OIR mice (Fig. 8C). After counting the number of vessels and calculating the neovascularization area based on the formed capillary hemangioma at the peripheral area of retina, we found that the number of vessels and the neovascularization area were both increased in OIR mice, but erianin (20 mg/kg) reduced such increase (Fig. 8D,E). In addition, the retinal vascular permeability assay showed that erianin (20 mg/kg) reduced retinal vascular leakage as compared with vehicle-treated OIR mice (Fig. 8F). Further results showed that erianin (20 mg/kg) reduced the increased mRNA (Fig. 8G) and protein expression of VEGF (Fig. 8H,I) in retinas from OIR mice. As shown in Fig. 8H,I, erianin (20 mg/kg) also reduced the protein expression of Iba1 in retinas from OIR mice. Discussion A typical feature of diabetes is hyperglycemia, which will cause damage to endothelial cells and induce angiogenesis through promoting the expression of a variety of pro-angiogenic factors including VEGF (also known as VEGFA) 19,20,22 . Thus, high content of blood glucose is considered to play an important role in regulating the development of DR. In this study, retinal endothelial RF/6A cells were used to imitate the pathogenesis of DR under the condition of high content of glucose. Similar to other previous reports, our results showed that HG induced angiogenic responses in RF/6A cells including tube formation and cell migration 23,24 . However, erianin blocked HG-induced tube formation and migration in RF/6A cells in the concentration-dependent manner (Fig. 1). In addition, erianin also reduced the increased retinal vessels in STZ-induced hyperglycemic mice (Fig. 7A-D). All these results evidenced the abrogation of erianin on retinal angiogenesis induced by hyperglycemia during the process of diabetes. Moreover, erianin also reduced the increased number of vessels and neovascularization area in the peripheral area of retinas and attenuated retinal vascular leakage in OIR mice (Fig. 8A-F). These results implied that the inhibition of erianin on retinal neoangiogenesis was not dependent on the reduction of blood glucose.
A variety of reports demonstrated that VEGF content was elevated in serum or vitreous cavity prior to the appearance of discernible vascular lesions, and there were positive correlations between VEGF and blood glucose levels [25][26][27] . The source of elevated VEGF content in DR is still not fully elucidated, but may include the autocrine mechanism in retinal endothelial cells 23,24,28 , and the paracrine mechanisms in retinal pigment epithelial (RPE) cells or macroglia cells [29][30][31] . Endothelial cell-derived VEGF has recently shown to be an essential autocrine mechanism for the increased angiogenic activity in endothelial cells 32 . Microglial cells have been reported to be activated and play important roles in regulating retinal neuroinflammation during the development of DR 33,34 . However, whether there is microglia-derived paracrine VEGF mechanism under hyperglycemic condition remains unknown. HIF-1α is reported to be a well-known transcription factor for regulating the expression of VEGF 35 . In this study, we found that erianin reduced the HG-induced increased VEGF mRNA and protein expression, and also abrogated HG-induced HIF-1α transcriptional activation in both retinal endothelial RF/6A cells (Fig. 2) and microglia BV-2 cells (Fig. 3) in vitro. Erianin also reduced the elevated VEGF contents in both serum and vitreous cavity in STZ-induced hyperglycemic mice in vivo (Fig. 7E,F). In addition, erianin further reduced the increased retinal VEGF expression in STZ-induced hyperglycemic mice (Fig. 7G) and OIR mice (Fig. 8G,I). Furthermore, erianin also reduced the increased microglia activation in STZ-induced hyperglycemic mice (Fig. 7H) and OIR mice (Fig. 8H,I). All these above results evidenced that erianin reduced the increased VEGF expression during the development of DR, and such phenomena is related with its inhibition on HIF-1α -mediated autocrine and paracrine VEGF expression in both endothelial and microglial cells.
ERK1/2 is a subfamily member of MAPKs, which plays a critical role in regulating cell survival, proliferation and differentiation 36 . Previous reports have demonstrated that ERK1/2 played critical roles in regulating HIF-1α activation and VEGF expression during tumor development [37][38][39] . ERK1/2 was not required for hypoxia-induced HIF-1α and VEGF expression in RPE cells 40 . However, whether ERK1/2 will play critical roles in regulating HG-induced HIF-1α activation and VEGF expression in retinal endothelial cells or microglial cells remains unknown. Our present study showed that erianin inhibited HG-induced ERK1/2 phosphorylation in both RF/6A and BV-2 cells just like U0126 (Fig. 4A,B,E,F), which is a well-known MEK1/2 inhibitor and can effectively block ERK1/2 activation. Furthermore, U0126 and erainin both abrogated HG-induced HIF-1α transcriptional activation in RF/6A and BV-2 cells (Figs 2D,E, 3D,E and 4C,D,G,H). The above results indicate that erianin-induced inhibition on ERK1/2 activation contributes to its blockage on HIF-1α transcriptional activation and subsequent VEGF expression in retinal endothelial and microglial cells.
Further, the inhibition of erianin on VEGF-induced angiogenesis in vitro and in vivo was observed. The results showed that erianin abrogated VEGF-induced cell growth, migration and tube formation in retinal endothelial cells in vitro, and inhibited VEGF-induced neoangiogenesis in matrigels in vivo (Fig. 5). In addition, erianin alone had no obvious cytotoxicity on RF/6A cells (Fig. 1E). VEGF binds and activates two tyrosine kinase receptors: VEGFR1 and VEGFR2, of which VEGFR2 plays critical roles in the development of various pathological angiogenesis including DR by regulating proliferation and migration of retinal endothelial cells through distinct signal pathways 22,41 . Both cRaf-MEK1/2-ERK1/2 and PI3K-AKT-mTOR-P70S6kinase are the downstream signaling cascades of VEGFR2, and play critical roles in regulating VEGF-induced angiogenic responses in endothelial cells 42,43 . Our results demonstrated that erianin blocked VEGFR2 phosphorylation induced by VEGF; and it also abrogated the VEGF-induced activation of c-Raf-MEK1/2-ERK1/2 and PI3K-AKT-mTOR-P70S6kinase signaling cascades in retinal endothelial RF/6A cells (Fig. 6). All those results demonstrated that erianin blocked the activation of VEGF-stimulated VEGFR2 and its downstream c-Raf-MEK1/2-ERK1/2 and PI3K-AKT signaling cascades, which will contribute to its anti-angiogenic activity in vivo and in vitro.
In summary, our results demonstrate that natural compound erianin inhibits HG-induced retinal angiogenesis by blocking HG-induced VEGF expression through ERK1/2-mediated HIF-1α transcriptional activation in both retinal endothelial and microglial cells, and further abrogates VEGF-induced retinal angiogenesis by blocking the activation of VEGFR2 and its downstream cRaf-MEK1/2-ERK1/2 and PI3K-AKT signaling cascades. The present study indicates the potential application of erianin for DR treatment, and also provides an explanation for the eye-improving capacity of medicinal Dendrobium (Shi-Hu) for centuries.
Cell migration assay. RF/6A cells were starved in RPMI1640 containing 0.1 BSA% for 4 h, and then incubated with or without erianin for 15 min. Cells were seeded into the upper chamber of the transwells pre-coated with gelatin, and then VEGF (10 ng/ml) or HG (25 mM) was added into the lower chamber. The non-migrated cells were wiped off with a cotton swab after 8 h, and the migrated cells were stained by crystal violet. Images were pictured under the inverted microscope, and the migrated cells were counted in a blind manner.
Protein extraction. Nuclear and cytoplasmic proteins in cells were isolated as described in NE-PER nuclear and cytoplasmic extraction kit. Other cellular proteins were extracted by using a whole cell protein extraction kit. Retinas were homogenized in ice-cold lysis buffer containing 50 mM Tris, pH7.5, 150 mM NaCl, 1 mM EDTA, 20 mM NaF, 0.5% NP-40, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride, 10 μ g/ml aprotinin, 10 μ g/ml leupeptin, and 10 μ g/ml pepstatin A. After centrifugation, protein concentration of the resulting supernatants was determined. The protein concentration in each sample was normalized to the equal protein concentration.
Western-blot analysis. Protein samples were separated by SDS-PAGE and blots were probed with an appropriate combination of primary and secondary antibodies. The antibody-reactive bands were identified by enhanced chemiluminescence kits. The grey densities of the protein bands were normalized using β -actin or lamin B1 density as an internal control, respectively. RNA isolation and cDNA synthesis. Cellular and retinal total RNA were isolated by using an RNeasy Plus Mini kit and Trizol reagents according to the manufacturer's instructions. The RNA content was determined by measuring the optical density at 260 nm, and cDNA was synthesized by using PrimeScript RT Master Mix kit.
Real-time PCR assay. Real-time PCR was performed by using SYBR Premix Ex Taq kit, and the relative expression of target genes was normalized to GAPDH in RF/6A cells or actin in BV-2 cells and in mice retinas, analyzed by the 2 −ΔΔCt method and given as ratio compared with the control. The primer sequences used in this study are shown in Supplementary Table. Cell viability assay. RF/6A cells (2 × 10 4 cells/well) were seeded into 96-well plates. After attachment, cells were pretreated with different concentrations of erianin for 15 min, and then incubated with or without VEGF (10 ng/ml) for 48 h. Cell viability was determined by 3-(4,5-dimethylthiazol-2-yl) 2,5-diphenyltetrazolium bromide (MTT). Optical density was measured at 570 nm with 630 nm as the reference and cell viability was normalized as the percentage of control.
In vivo matrigel assay. Matrigel containing 10 ng/ml VEGF, heparin (30 U/ml) and erianin (10, 50 nM) in liquid form at 4 °C was injected into the abdominal subcutaneous tissue along the peritoneal mid-line of C57BL/6 mice (n = 6). After 7 days, the mice were sacrificed and the matrigel plugs were taken out and pictured. The matrigel plugs were also prepared for immunohistochemistry. Immunohistochemistry. Matrigel plugs were fixed in 4% paraformaldehyde and embedded in paraffin.
The fixed sections (5 μ m) were stained with CD31 and the whole experimental procedure was described in our previous published paper 15 .

STZ-induced hyperglycemic mice.
Thirty mice were administered intraperitoneally (i.p.) with 55 mg/kg STZ for 5 consecutive days, while the other ten mice were injected with physiological saline and served as control animals. The concentration of serum glucose was measured 7 days after the last injection; the glucose concentration of 28 mice was > 16.5 mmol/L in this study, and those mice were considered as diabetic mice and randomly divided into 3 groups. At 3 months after the injection of STZ, the mice were administered intraperitoneally with erianin (1 or 10 mg/kg per day) for 2 consecutive months. All mice were sacrificed at 5 months after the injection of STZ.
Retinal CD31 immunofluorescence staining. The procedure of retinal CD31 immunofluorescence staining was performed as described in our previous published papers 15,44,45 . Retinal vessels were pictured under the fluorescence microscopy and the quantity of vessels was counted as described previously 15,44,45 . The quantification of the vasoobliterative area at the central area of retina and the neovascularization area based on the formed capillary hemangioma at the peripheral area of retina from OIR mice was measured by using cellSens Dimension 1.6 (IX81, Olympus, Japan).
Histopathological observation. Retinas isolated from mice were fixed in 4% paraformaldehyde solution and embedded in paraffin, and then sectioned (5 μ m), stained with haematoxylin and eosin (H&E), and observed under the microscopy. ELISA assay. The whole blood and vitreous cavity suspended in PBS was centrifuged at 5000 g, 4 °C for 10 min. The serum and supernatant of vitreous cavity were collected for further ELISA assay according to the manufacturer's instruction.
Immunofluorescence staining of Iba1in retinas. Paraffin-embedded sections of retinas (5 μ m) were deparaffinized in xylene, and rehydrated in an ethanol gradient with distilled water. After endogenous peroxidase activity was quenched, retinas were incubated with 5% bovine serum albumin to minimize nonspecific binding. After rinsing three times (5 min each), retinas were incubated with Iba1 antibody at 4 °C overnight, and further incubated with Alexa fluor 488 goat anti-rabbit IgG (H + L) antibody at room temperature for 1 h after rinsing three times again. After rinsing three times again, retinas were incubated with DAPI for 10 min. Images were captured under an inverted microscope (IX81, Olympus, Japan). OIR mice. On postnatal day 12 (P12), after the C57BL/6 mice were exposed to 75 ± 2% oxygen for 5 days (P7-P12), the mice were returned to the normal room air and randomly divided into two groups: vehicle-treated OIR mice and erianin-treated (20 mg/kg/day) OIR mice. The normal control mice were kept under normal room conditions from birth until postnatal day 17 (P17). The mice were injected intraperitoneally with erianin (20 mg/kg per day) for 5 days. On P17, the mice were anesthetized and sacrificed.

Measurement of retinal vascular leakage using Evans Blue Assay. Mice were injected with 2%
Evans blue (10 μ l/g, i.p.) in PBS. After 2 h, blood was extracted through the left ventricle; and the mice were perfused with PBS to completely remove the Evans blue dye in blood vessels. Retinas were carefully dissected and the weight was determined after thoroughly drying. Next, the retinas were incubated in 120 μ l formamide for 18 h at 70 °C to extract the Evans blue dye. The extract was centrifuged at 10,000 g twice for 1 h at 4 °C. Absorbance of the supernatant was measured with a spectrophotometer at 620 nm. The concentration of Evans blue dye in extracts was calculated using a standard curve of Evans blue in formamide and normalized to the dried retinal weight.
Statistical analysis. Data were expressed as means ± standard error of the mean (SEM). The significance of differences between groups was evaluated by one-way ANOVA with LSD post hoc test; and P < 0.05 was considered as indicating statistically significant differences.