Metformin inhibits ALK1-mediated angiogenesis via activation of AMPK

Abstract

Anti-VEGF therapy has been proven to be effective in the treatment of pathological angiogenesis. However, therapy resistance often occurs, leading to development of alternative approaches. The present study examines if AMPK negatively regulates ALK1-mediated signaling events and associated angiogenesis. Thus, we treated human umbilical vein endothelial cells with metformin as well as other pharmacological AMPK activators and showed that activation of AMPK inhibited Smad1/5 phosphorylation and tube formation induced by BMP9. This event was mimicked by expression of the active mutant of AMPKα1 and prevented by the dominant negative AMPKα1. Metformin inhibition of BMP9 signaling is possibly mediated by upregulation of Smurf1, leading to degradation of ALK1. Furthermore, metformin suppressed BMP9-induced angiogenesis in mouse matrigel plug. In addition, laser photocoagulation was employed to evaluate the effect of metformin. The data revealed that metformin significantly reduced choroidal neovascularization to a level comparable to LDN212854, an ALK1 specific inhibitor. In conjunction, metformin diminished expression of ALK1 in endothelium of the lesion area. Collectively, our study for the first time demonstrates that AMPK inhibits ALK1 and associated angiogenesis/neovascularization. This may offer us a new avenue for the treatment of related diseases using clinically used pharmacological AMPK activators like metformin in combination with other strategies to enhance the treatment efficacy or in the case of anti-VEGF resistance.

Keywords: ALK1; AMPK; tumor angiogenesis.

Figure 1. Inhibition of phosphorylated-Smad1/5 and tube formation by metformin

(A) Dose-dependent effect. HUVECs were treated with metformin (Met) at varying doses, followed by BMP9. (B) Time-dependent effect. HUVECs were treated with metformin for different time (hours), followed by BMP9. Cell extracts (20 μg) were blotted with antibodies, as indicated in A and B. (C) Inhibition of tube formation. HUVECs were seeded on Matrigel and treated with or without BMP9 (10 ng/mL) and/or metformin (10 mM) for 7 hours. The cells were stained with Calcein AM and photos taken under fluorescent microscopy. Representative images are shown. Scale bar: 100 μm. (D) Quantitative analysis. Tube formation was calculated by lengths using ImageJ program and plotted. Analysis was performed from duplicate experiments, from which 5 visual areas were taken. Significance was tested by one way ANOVA (average pixels ± SD, n=10). ** p<0.01.

Figure 2. Inhibition of BMP9-induced phosphorylation of Smad1/5 and tube formation by different AMPK activators

HUVECs were treated with metformin (Met), AICAR (AI), or A769662 (A76), followed by BMP9. Western blot was performed with antibodies, as indicated (A). Tube formation (B) and statistical test (AMP activators-treated group vs BMP9 group) (C) were performed. Graph represents averages ± SD (n=10), * p<0.05, ** p<0.01, *** p<0.0001. Scale bar: 100 μm.

Figure 3. Effects of AMPK mutants on Smad1/5 phosphorylation and tube formation

(A)-(B) HUVECs were infected with adenovirus encoding active mutant of AMPKα1 (AMPK-CA) or GFP in different volumes (μl) for 2 days and then treated with BMP9. (A) Cell extracts were blotted with antibodies, as indicated. (B) Tube formation assay. HUVECs were infected with adenovirus for AMPK-CA or GFP for 2 days, and tube formation was performed, as described in Figure 1. Statistical analysis of tube formation was performed. Graph represents averages ± SD (n=10). **p<0.01. (C)-(D) HUVECs were infected with adenovirus expressing the dominant negative mutant of AMPKα1 (AMPK-DN) or GFP. (C) The cells were treated with or without metformin, followed by BMP9 and cell extracts blotted with antibodies, as indicated. (D) Tube formation was assayed on HUVECs infected with AMPK-DN or GFP adenovirus and treated with or without BMP9 (10 ng/mL) and/or metformin (10 mM) for 6 hours. Graph represents averages ± SD of tube length (n=10). Two way ANOVA was used to test statistical significance. **p<0.01, *** p<0.0001.

Figure 4. Inhibition of active mutant of ALK1 by AMPK

(A) HUVECs were infected with ALK1AAD or GFP adenovirus for 2 days and treated with metformin (10 mM) for 2 to 24 hours. (B) HUVECs were infected with adenovirus expressing an active mutant of ALK1 (ALK1AAD) or GFP adenovirus for 2 days and then treated with metformin (Met), berberine (25 μM), AICAR (AI), or and A769662 (A76) overnight. (C) Adenovirus expressing ALK1AAD was co-infected with adenovirus expressing AMPK-CA, AMPK-DN, or GFP. Western blot was performed with antibodies, as indicated in A-C. (D) HUVCs were infected with ALKAAD adenovirus and tube formation was conducted with or without metformin (10 mM) for 6 hours and graph plotted. Adenovirus expressing GFP served as a control. Statistical analysis of tube formation in C was performed as described in Figure 1. Graph represents averages ± SD (n=5). *p<0.05.

Figure 5. Metformin inhibits angiogenesis on mouse Matrigel plug

Matrigel was subcutaneously injected into the back of mice in the presence or absence of BMP9 (10 ng/mL). The mice were administered intraperitoneally with metformin (150 mg/kg/day) and/or PBS or 7 days. Matrigel plugs were then excised, sectioned and analyzed. (A) Gross anatomy of freshly removed Matrigel plug. (B) Slides were incubated with Dy594-lectin and images taken under fluorescent microscope. Scale bar: 100 μm. (C) The dense capillary structures were analyzed by nodes using ImageJ program and plotted. Graph represents averages of five different sights ± SD (n=5). Significance was tested by one way ANOVA, ** p<0.01.

Figure 6. The effect of metformin on laser-induced choroidal neovascularization (CNV) in mice

C57BL/6J mice were used for laser-induced CNV model. Four lesions were induced using the laser photocoagulation after anesthesia with ketamine hydrochloride. (A) Animals were equally divided into two groups (8 mice/group), treated with PBS or metformin (150 mg/kg/day) i.p. once daily from the day prior to laser irradiation (Day-1) to Day 6. For the evaluation of CNV size, at Day 7, mice were deeply anesthetized and perfused with FITC-lectin. The mice were sacrificed and the eyes were enucleated and fixed in 4% paraformaldehyde. RPE-choroid tissue was flat-mounted and observed under fluorescein microscope. Size of CNV was measured by μm² and vascular density in CNV lesion by ImageJ program. Significance was tested by student t test (average pixels ± SEM, n=7). *p<0.05, ** p<0.01. (B) Animals were injected i.p. with PBS or LDN-212854 (6 mg/kg, twice daily) and processed as described for A. CNV size and vascular density in CNV lesions were determined. Scale bar: 200 μm.

Figure 7. The effect of Metformin on ALK1 expression in Laser-induced CNV

Lesion was created as described in Figure 6A, except that eyes were harvested at day 5. The cryosectioned slides were incubated with antibodies against ALK1 and CD31, an endothelium marker, and DAPI Extensive staining of CD31 (red) and ALK1 (green) is seen in control eyes from animals treated with PBS (A and B). Many CD31 positive cells are also positive for ALK1 staining, indicating they are endothelial cells in the new blood vessels that expressed ALK1 (C, arrowheads). A DIC (differential interference contrast) image was superimposed to the fluorescent images to provide structural information (D). In contrast, The CD31 and ALK1 staining is much less in the eyes from metformin treated animals (E and F), and almost no co-localization was present (G and H). (I) Co-stained signal of CD31 and ALK1 was compared in the presence and absence of metformin and plotted. Graph represents averages ± SD from three independent experiments. ***p<0.001. Scale bars: 50 μm. Sc: Sclera; Ch: Choroid; RPE: Retinal pigment epithelium; ONL: outer nuclear layer.