The Anti-Diabetic Drug Metformin Suppresses Pathological Retinal Angiogenesis via Blocking the mTORC1 Signaling Pathway in Mice (Metformin Suppresses Pathological Angiogenesis)

Abstract Purpose Metformin, a biguanide antihyperglycemic drug, can exert various beneficial effects in addition to its glucose-lowering effect. The effects of metformin are mainly mediated by AMP-activated protein kinase (AMPK)-dependent pathway. AMPK activation interferes with the action of the mammalian target of rapamycin complex 1 (mTORC1), and blockade of mTORC1 pathway suppresses pathological retinal angiogenesis. Therefore, in this study, we examined the effects of metformin on pathological angiogenesis and mTORC1 activity in the retinas of mice with oxygen-induced retinopathy (OIR). Methods OIR was induced by exposing the mice to 80% oxygen from postnatal day (P) 7 to P10. The OIR mice were treated with metformin, rapamycin (an inhibitor of mTORC1), or the vehicle from P10 to P12 or P14. The formation of neovascular tufts, revascularization in the central avascular areas, expression of vascular endothelial growth factor (VEGF) and VEGF receptor (VEGFR) 2, and phosphorylated ribosomal protein S6 (pS6), a downstream indicator of mTORC1 activity, were evaluated at P10, P13, or P15. Results Neovascular tufts and vascular growth in the central avascular areas were observed in the retinas of P15 OIR mice. The formation of neovascular tufts, but not the revascularization in the central avascular areas, was attenuated by metformin administration from P10 to P14. Metformin had no significant inhibitory effect on the expression of VEGF and VEGFR2, but it reduced the pS6 immunoreactivity in vascular cells at the sites of angiogenesis. Rapamycin completely blocked the phosphorylation of ribosomal protein S6 and markedly reduced the formation of neovascular tufts. Conclusions These results suggest that metformin partially suppresses the formation of neovascular tufts on the retinal surface by blocking the mTORC1 signaling pathway. Metformin may exert beneficial effects against the progression of ocular diseases in which abnormal angiogenesis is associated with the pathogenesis.


Introduction
Metformin, a biguanide antihyperglycemic drug, is widely used to treat type 2 diabetes mellitus.7][8][9] Furthermore, metformin suppresses the proliferation and migration of vascular endothelial cells 4, 10,11 and pathological retinal angiogenesis. 4,12][7] The effects of metformin are mainly mediated by AMPactivated protein kinase (AMPK)-dependent pathways, although the involvement of AMPK-independent mechanisms has also been suggested. 13AMPK activation interferes with the action of the mammalian target of rapamycin complex 1 (mTORC1), a protein that plays a critical role in controlling the transcription and growth, proliferation, and migration of cells. 14mTORC1 acts downstream of the vascular endothelial growth factor receptor (VEGFR) pathway in proliferating endothelial cells, facilitating retinal angiogenesis under both physiological and pathological conditions. 15,16Metformin reportedly decreases the expression of the VEGF mRNA and VEGFR2 protein 11,12 and exerts anti-angiogenic effects. 4,12However, the possible role of mTORC1 signaling pathway in establishing anti-angiogenic effects of metformin treatment remains unclear.Therefore, the purpose of the present study was to examine whether metformin suppresses retinal angiogenesis by inhibiting mTORC1 signaling.

Animals
Adult male and female Institute of Cancer Research (ICR) mice were purchased from Charles River Breeding Laboratories (Tokyo, Japan).After the acclimation period, the males and females were placed together for mating.Each pregnant mouse was placed in a separate cage and the day of birth (postnatal day [P] 0) was determined by daily inspection.The animals were provided ad libitum access to standard diet (Oriental Yeast, Tokyo, Japan) and tap water.
All animal procedures adhered to the guidelines of the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Institutional Animal Care and Use Committee of Kitasato University (approval number: T12-3).

Oxygen-induced retinopathy (OIR) mouse model
OIR mouse model was established according to the protocol described in our previous report. 16Briefly, P7 mice with their nursing mother were placed in hyperoxic conditions (80% oxygen) in an oxygen-regulated chamber for 3 days to produce retinal vasoobliteration, and at P10, the mice were returned to room air (normoxic conditions) to induce ischemic retinal neovascularization.Oxygen concentrations of the chamber were continuously monitored with a sensor placed inside the chamber and regulated by an oxygen controller (ProOx110; Biospherix, Redfield, NY, USA).Typical pathological changes, the formation of central avascular zones and neovascular tufts, in the OIR mouse were observed under our experimental conditions. 16

Drug treatments
To assess the effects of metformin and the inhibitor of mTORC1 rapamycin on neovascularization and mTORC1 activity, metformin (100 or 200 mg/kg), rapamycin (10 mg/ kg), or vehicle (saline) was subcutaneously administered once daily from P10 to P14 (Figure 1(A)).To examine the VEGF mRNA distribution in the retinal surface and distribution of VEGFR2 protein, metformin (200 mg/kg) or saline was subcutaneously administered once daily from P10 to P12 or P14 (Figure 1(B)).Metformin hydrochloride (Tocris Bioscience, Bristol, UK) was dissolved in saline (1 mg/mL and 2 mg/mL).Rapamycin (LC Laboratories, Woburn, MA, USA) was dissolved in dimethyl sulfoxide at a concentration of 25 mg/mL and diluted with 5% Tween 80, 5% polyethylene glycol 400, and 4% ethanol (1 mg/mL).In our preliminary study, administration of saline-and the Tween 80/ polyethylene glycol 400/ethanol mixtures had no significant effect on retinal blood vessels in mice.Metformin and rapamycin were subcutaneously administered at a volume of 10 lL/g body weight.The doses of metformin were selected based on previous studies. 6,17,18or immunohistochemistry, in situ hybridization, and western blot analysis, the eyes were collected at P10, P13, and P15.
For cross-sectional staining, the harvested eyes were kept in the same fixative (1% PFA in PBS) for 1 h at 4 � C, equilibrated overnight at 4 � C in 30% sucrose in PBS, and embedded in optimal cutting temperature (OCT) compound (Sakura Finetek, Torrance, CA, USA).The frozen tissue blocks were cut into 16-lm-thick sections (using a cryostat) and dried on glass slides.After removing the OCT compound by rinsing, the sections were blocked with 5% NGS in PBS-T for 0.5-1 h and then incubated overnight with rat monoclonal anti-PECAM-1 antibody (diluted 1:500 in 5% NGS in PBS-T, Thermo Fisher Scientific) and anti-VEGFR-2 antibody (diluted 1:500 in 5% NGS in PBS-T; #2479; rabbit monoclonal; 55B11; Cell Signaling Technology) at room temperature, to examine the distribution of VEGFR-2 in the retina.Following incubation with the primary antibodies, the sections were rinsed four times with PBS-T for 10 min each and incubated for 4 h with species-specific secondary antibodies conjugated with FITC or Cy3 (diluted 1:400 in 5% NGS in PBS-T; Jackson ImmunoResearch Laboratories).Next, the sections were rinsed five times with PBS-T for 10 min each and mounted with Vectashield containing 4 0 ,6diamidino-2-phenylindole (H-1200; Vector Laboratories).
Images were captured using a fluorescence microscope (BZ-9000; Keyence, Osaka, Japan).The neovascular tufts and vascularized areas were evaluated as previously reported. 16Briefly, neovascular tufts were manually outlined in five regions, and the area of the neovascular tufts relative to that of the selected region was calculated.The vascularized and total retinal surface areas were measured, and the vascularized area was determined as a percentage of the total retinal surface area.

Western blotting
Western blotting analysis was performed as previously described. 20The mice were perfused transcardially with icecold PBS under deep anesthesia with sodium pentobarbital (Nacalai Tesque).Subsequently, the eyes were enucleated and the retinas were isolated and homogenized in ice-cold RIPA buffer containing a protease inhibitor cocktail (Nacalai Tesque).The cell debris was pelleted at 12,000�g for 5 min at 4 � C, and the supernatant was collected and stored at −80 � C until use.
To assess the phosphorylation of S6 protein in the retina, protein samples (5 mg) were separated using 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred onto a polyvinylidene fluoride membrane.The membrane was blocked with 2% bovine serum albumin (BSA) in Tris-buffered saline containing 0.5% Tween 20 (TBS-T) for 1 h at room temperature, followed by overnight incubation with rabbit monoclonal anti-pS6 antibody (1:2000 in 2% BSA in TBS-T; Cell Signaling Technology) at 4 � C. Horseradish peroxidase-conjugated anti-rabbit IgG antibody (1:20,000 in 2% BSA in TBS-T; Nacalai Tesque) was used as the secondary antibody, followed by incubation for 1 h at room temperature.The immunoreactive bands were developed with the Immunostar V R LD detection system (Wako, Osaka, Japan) or the Clarity TM Western ECL Substrate (Bio-Rad) and examined with LAS-4000 mini luminescent image analyzer (Fujifilm, Tokyo, Japan).Band intensities were determined using the Multi Gauge software (Fujifilm).The membranes were then stripped and re-probed with rabbit monoclonal anti-S6 antibody (1:1000 in 2% BSA in TBS-T; #2217, 5G10; Cell Signaling Technology) to assess total S6 protein levels.Rabbit monoclonal anti-b actin antibody (1:5000 in 2% BSA in TBS-T; #4970, 13E5; Cell Signaling Technology) was used as the loading control.

In situ hybridization of whole-mount retinas
Our preliminary results indicated that revascularization begins on P13 in mice with OIR. 16Therefore, the VEGF mRNA distribution in the retinal surface was determined on P13.After completing whole-mount in situ hybridization protocol, the retinas were stained with anti-type IV collagen antibody (1:2000 in 5% NGS in PBS-T; #LB-1403, Cosmo Bio, Tokyo, Japan) to visualize the vascular network, as previously described. 15

Statistical analysis
All statistical analyses were performed using Prism 5.0 (GraphPad Software, Inc., San Diego, CA, USA).All values are expressed as mean ± standard error.Statistical comparisons between two groups of data and multiple groups of data were performed using the Student's t-test and analysis of variance, followed by a Tukey-Kramer post-test, respectively.A p value < 0.05 indicated statistical significance.

Results
We examined the effects of metformin on the retinal vasculature of OIR mice.The schedule of the metformin injections is shown in Figure 1(A).In P10 mice, immediately after the completion of hyperoxia exposure, avascular areas were observed in the retinas (Figure 2(Aa,Aa')).In the retinas of P15 OIR mice treated with saline for 5 days, many neovascular tufts formed at the border between the vascularized and central avascular areas, and revascularization occurred in the central avascular areas (Figure 2(Ab,Ab')).However, the administration of metformin (100 and 200 mg/ kg/day) for 5 days markedly reduced the formation of neovascular tufts without affecting revascularization in the avascular areas (Figure 2(Ac,Ac',Ad,Ad')).Quantitative data indicated that metformin diminished neovascular tufts but had no effect on the size of the vascularized area (Figure 2(B)).
Metformin reportedly decreased the VEGF mRNA and VEGFR2 expression in previous studies. 11,12In situ hybridization analysis revealed that the VEGF mRNA was faintly expressed in the central avascular areas of P10 OIR mice (Figure 3(Aa)), but thereafter it was enhanced, irrespective of metformin treatment (Figure 3(Ab,Ac)).Immunohistochemical staining revealed that VEGFR2 was expressed in cells located in the ganglion cell layer and inner nuclear layer of P10 OIR mice (Figure 3(Ba,Bb)).After 5 days, the VEGFR2 expression was enhanced in the endothelial cells of the capillaries and neovascular tufts on the retinal surface (Figure 3(Bc,Bd)).Thus, the altered distribution of VEGFR2 may be involved in the abnormal angiogenesis in OIR mice.However, there was no significant difference in the VEGFR2 expression in the vessels extending toward the deep retina and deep capillary plexus between saline-and metformintreated OIR mice, except for the extent of the neovascular tufts (Figure 3(Be,Bf)).
Metformin activates AMPK, which negatively regulates mTORC1 activation. 14,21To investigate whether metformin prevents abnormal vascular growth by decreasing mTORC1 activity, we evaluated the status of mTORC1 activity by determining the distribution and content of pS6.Strong pS6 immunoreactivity was observed in the neovascular tufts of saline-treated P15 OIR mice (Figure 4(Aa,Ab)).In addition, many pS6-positive cells were observed in the retinal parenchyma (Figure 4(Aa,Ab)).However, pS6 immunoreactivity was not observed in quiescent endothelial cells in the capillaries (Figure 4(Aa,Ab)).Notably, pS6 immunoreactivity was also not observed in a subset of the neovascular tufts (Figure 4(Aa,Ab)).In metformin-treated OIR mice, pS6 immunoreactivity associated with neovascular tufts was considerably lower than that in saline-treated OIR mice (Figure 4(Ac,Ad)).Metformin had no significant effect on pS6 immunoreactivity in the retinal parenchyma (Figure 4(Ac,Ad)), whereas rapamycin completely abolished pS6 immunoreactivity (Figure 4(Ae,Af)).Rapamycin markedly reduced the formation of neovascular tufts (Figure 4(Af)).
Western blotting analysis of whole retinal lysates revealed that the phosphorylation of S6 protein was unaffected by metformin but completely blocked by rapamycin (Figure 4(B)).
We next addressed the question of whether metformin inhibits the mTORC1 signaling pathway before the initiation of endothelial cell degeneration.However, becuase the heterogeneity of neovascular tufts made quantitative evaluation of pS6 difficult, we used P4 normal mice.Strong pS6 immunoreactivity was detected in proliferating endothelial cells at the front of the developing vasculature and non-vascular cells in the parenchyma of P4 normal mouse retinas (Supplementary Figure 1(Aa-c)).The pS6 immunoreactivity associated with vascular cells selectively decreased 6 h after metformin administration without affecting the vascular density (Supplementary Figure 1(Aa-c,Ae-g,B)).The colocalization of pS6 with PECAM-1-positive endothelial cells was lower in metformin-treated mice than that in saline-treated mice (Supplementary Figure 1(Ad,Ah,C)).These results suggested that mTORC1 inhibition by metformin occurred prior to endothelial cell degeneration.

Discussion
The present study demonstrates that metformin prevents the formation of neovascular tufts, but not revascularization in the central avascular areas, in the retinas of OIR mice.Metformin reduced pS6 immunoreactivity, a measure of mTORC1 activity, in the neovascular tufts of OIR mice and proliferating endothelial cells at the front of the developing vasculature.The mTORC1 signaling pathway, which is associated with the proliferation of endothelial cells, is partially activated in a VEGF-dependent manner. 15,16Therefore, the interruption of the VEGF-mTORC1 signaling pathway, at least in part, could contribute to the anti-angiogenic effect of metformin in the retina.
The increased level of the VEGF is involved in the vascular pathology of OIR. 22Previous reports have shown that metformin suppresses VEGF production in cultured vascular endothelial cells, 23,24 though it has no significant effect on the VEGF expression in OIR mouse retinas. 12Consistently, the current study showed that the VEGF mRNA expression increased in central avascular areas 3 days after the completion of hyperoxia exposure, irrespective of metformin treatment.Regarding the receptors for the VEGF, metformin decreased VEGFR2 expression levels by increasing its degradation through the ubiquitin-proteasome system. 12By contrast, our immunohistochemical study revealed that the distribution pattern of the VEGFR2 immunoreactivity in the retina was almost the same in saline-and metformin-treated mice, except for the extent of the neovascular tufts.Although the reasons for this inconsistency are unclear, our results suggest that metformin can suppress retinal pathological angiogenesis by interrupting the signaling pathways downstream of the VEGFR2 rather than by decreasing the expression levels of VEGF and/or VEGFR2.
VEGFR2 activation stimulates the phosphatidylinositol-3 kinase/protein kinase B/mTOR signaling pathway, which subsequently facilitates endothelial cell proliferation. 25A previous study on OIR mice reported an increased mTORC1 activity (indicated by pS6 immunoreactivity) in some neovascular tufts, and the inhibition of mTORC1 significantly attenuated the formation of neovascular tufts. 16he activation of AMPK suppresses the mTORC1 signaling pathway. 14Thus, we hypothesized that the ability of metformin to prevent abnormal angiogenesis may be mediated by the inhibition of the mTORC1 signaling pathway.This hypothesis was supported by the observations that pS6 immunoreactivity was no longer found in the neovascular tufts of metformin-treated OIR mice, although it was still detected in several blood vessels and non-vascular cells in the parenchyma.By contrast, rapamycin completely abolished pS6 immunoreactivity in the retina.Western blotting analysis of the whole retinal lysates revealed that rapamycin almost completely blocked the phosphorylation of S6, whereas metformin failed to stop S6 phosphorylation.Therefore, the reduction in pS6 in the neovascular tufts was possibly undetectable by western blotting.
The VEGF-mTORC1 signaling pathway in endothelial cells contributes to the development of retinal vasculature in mice. 15During the first postnatal week, endothelial cells located at the vascular front exhibited pS6 immunoreactivity, which was markedly diminished by rapamycin.To evaluate mTORC1 activity before the initiation of endothelial cell degeneration, we examined pS6 immunoreactivity 6 h after metformin administration in normal P4 mice.At this time point, the vascular area density did not differ between salineand metformin-treated mice.However, metformin decreased pS6 immunoreactivity in endothelial cells at the vascular front but not in non-vascular cells in the retinal parenchyma.These results suggested that metformin decreased mTORC1 activity in endothelial cells before the initiation of endothelial cell degeneration.Thus, the inhibitory effect of metformin on the mTORC1 signaling pathway may contribute to the prevention of abnormal retinal angiogenesis.
Metformin reduced mTORC1 activity in proliferating endothelial cells in growing blood vessels, although it did not prevent revascularization in the central avascular areas of OIR mice.We previously found that the inhibitory effects of mTORC1 inhibitors on revascularization in central avascular areas were weaker than those on the formation of neovascular tufts. 16mTORC1 inhibitors exhibited a more potent inhibitory effect on proliferating endothelial cells than on quiescent endothelial cells.Therefore, the importance of the mTORC1dependent mechanism is likely to increase with the proliferation of endothelial cells.Apparently, the inhibitory effect of metformin on mTORC1 was weaker than that of rapamycin.Thus, the inhibitory effects of metformin on revascularization in the central avascular areas may be undetectable in OIR mice.However, we cannot exclude the possibility that the prevention of neovascular tuft formation by metformin is due to its inhibitory effect on an mTORC1-independent mechanism.Further studies are needed to elucidate the mechanisms through which metformin suppresses the formation of neovascular tufts in mice with OIR.
In summary, we found that metformin partially suppressed retinal angiogenesis by blocking the mTORC1 signaling pathway.Ocular neovascularization is associated with the pathogenesis of several retinal diseases, including retinopathy of prematurity and age-related macular degeneration.Thus, metformin may exert beneficial effects on these retinal diseases as well.

Figure 1 .
Figure 1.Schedule for injections with metformin and rapamycin.A: To assess the effects of metformin and rapamycin on neovascularization and mTORC1 activity, metformin (100 or 200 mg/kg/day), rapamycin (10 mg/kg/day), or vehicle (saline) was subcutaneously administered once daily from P10 to P14.Eyes were collected at P10 and P15 for analyses.B: To examine the VEGF mRNA distribution in the retinal surface and distribution of VEGFR2 protein, metformin (200 mg/kg/day) or saline was subcutaneously administered once daily from P10 to P12 or P14.Eyes were collected on P10 and P13 or P15 for analyses.

Figure 2 .
Figure 2. Effect of metformin on retinal neovascularization in mice with OIR.A: Representative images of retinal flat mounts stained for PECAM-1 in P10 mice exposed to hyperoxia for 3 days (pretreatment, a) and P15 OIR mice treated with either saline (b), metformin (100 mg/kg/day, c), or metformin (200 mg/kg/day, d).Higher-magnification images of the box in panels (a-d) are shown in the lower panels (a'-d').Asterisk and arrowheads indicate central avascular area and neovascular tufts, respectively.The formation of neovascular tufts was markedly reduced in metformin-treated OIR mice compared to saline-treated OIR mice.Scale bars: 1 mm in a (applies to b-d), 500 mm in a' (applies to b'-d').B: Bar graphs show the size of vascularized area (left) and neovascular tufts (right) in the retina.Each column with a vertical bar represents the mean ± SE from 6 to 7 animals.� p < 0.05 versus P10.† p < 0.05 versus Saline.

Figure 3 .
Figure 3. Distribution of VEGF mRNA and VEGFR2 protein in the retinas of OIR mice.(A) Representative images of in situ hybridization for VEGF (blue) in retinas of P10 mice exposed to hyperoxia for 3 days (pretreatment, a) or P13 OIR mice treated with either saline (b) or metformin (200 mg/kg/day, c).Immunostaining for type IV collagen (red) was performed to visualize the vascular network.Strong expression of VEGF mRNA was detected in central avascular area in P13 OIR mice (black asterisks).Scale bar: 150 mm in a (applies to b-c).(B) Representative images of retinal cross-sections stained for VEGFR2 (red) and PECAM-1 (green) in P10 mice exposed to hyperoxia for 3 days (pretreatment, a,b) and P15 OIR mice treated with either saline (c,d) or metformin (200 mg/kg/day, e,f).Arrowheads indicate neovascular tufts.Scale bar: 50 mm in a (applies to b-f).GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer.

Figure 4 .
Figure 4. Effect of metformin on the mTORC1 activity in OIR mice.(A) Representative images of retinal flat mounts stained for PECAM-1 and phosphorylated S6 protein (pS6) in P15 OIR mice treated with either saline (a,b), metformin (200 mg/kg/day, c,d), or rapamycin (10 mg/kg/day, e,f) from P10 to P14. White and pink arrowheads indicate pS6-positive and -negative neovascular tufts, respectively.Like metformin, rapamycin decreased the formation of neovascular tufts.Scale bar: 100 mm in a (applies to b-f).(B) Western blot analysis of S6 phosphorylation after treatment with metformin or rapamycin.Each column with a vertical bar represents the mean ± SE from 5 animals.� p < 0.05 versus Saline.