AMP-activated protein kinase (AMPK) signaling in endothelial cells is essential for angiogenesis in response to hypoxic stress

AMP-activated protein kinase (AMPK) is a stress-activated protein kinase that is regulated by hypoxia and other cellular stresses that result in diminished cellular ATP levels. Here, we investigated whether AMPK signaling in endothelial cells has a role in regulating angiogenesis. Hypoxia induced the activating phosphorylation of AMPK in human umbilical vein endothelial cells (HUVECs), and AMPK activation was required for the maintenance of pro-angiogenic Akt signaling under these conditions. Suppression of AMPK signaling inhibited both HUVEC migration to VEGF and in vitro differentiation into tube-like structures in hypoxic, but not normoxic cultures. Dominant-negative AMPK also inhibited in vivo angiogenesis in Matrigel plugs that were implanted subcutaneously in mice. These data identify AMPK signaling as a new regulator of angiogenesis that is specifically required for endothelial cell migration and differentiation under conditions of hypoxia. As such, endothelial AMPK signaling may be a critical determinant of blood vessel recruitment to tissues that are subjected to ischemic stress.


Introduction
Angiogenesis is central feature of normal embryonic and post-natal development, and this process plays a critical role in the neovascularization that is associated with tumor growth and occlusive vascular diseases (1)(2)(3). A large body of evidence has shown that vascular angiogenic growth factors promote angiogenesis through activation of MAP kinase (4) and AMP-activated protein kinase (AMPK 1 ) is a metabolite-sensing protein kinase that shares amino acid sequence homology with yeast SNF1 (6,7). In myocytes, AMPK is activated by increases in the AMP:ATP ratio, which is brought about by hypoxia/anoxia (8)(9)(10), vigorous exercise and muscle contraction (11)(12)(13)(14) or pressure-overload hypertrophy (15).
Endothelial nitric oxide synthase (eNOS) residue 1177 (in human) is a substrate for both Akt (27,28) and AMPK (29). Phosphorylation of eNOS at this residue leads to enzyme activation and nitric oxide (NO) production. AMPK is reported to phosphorylate eNOS in cardiac myocytes under hypoxic conditions (29), but it is not clear whether NO production by endothelial cells is regulated by this reaction. Recently, AMPK was detected in human umbilical vein endothelial cells (HUVECs) (30). The pharmacological stimulator of AMPK, 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside (AICAR), protects HUVECs from apoptosis under hyperglycemic conditions (31). Interestingly, AICAR was found to reverse the inhibitory effect of hyperglycemia on mitogen-stimulated Akt phosphorylation.
In this study we analyzed whether AMPK signaling plays a role in angiogenesis.
Specifically, we tested whether AMPK is essential for endothelial cell migration, differentiation into capillary-like structures, and NO production under both normoxic and hypoxic conditions. We also tested whether AMPK signaling is required for angiogenesis in by guest on March 17, 2020 http://www.jbc.org/ Downloaded from 5 Matrigel plugs that were subcutaneously implanted in mice. The results of this study suggest that crosstalk between AMPK and Akt is essential for angiogenesis under conditions of hypoxic stress, but dispensable for angiogenic cellular responses in normoxic endothelial cells.
by guest on March 17, 2020 http://www.jbc.org/ Downloaded from 8 antibodies at a 1:1000 dilution followed by the secondary antibody conjugated with horseradish peroxidase (HRP) at a 1:5000 dilution. ECL-PLUS Western Blotting Detection kit (Amersham Pharmacia Biotech, Piscataway, New Jersey) was used for detection.

Migration assay
Migration assays were performed as described previously using a modified Boyden chamber (Neuroprobe, Cabin John, Maryland) (36). HUVECs infected with the indicated adenoviral vectors were serum-starved overnight in serum-free media (EBM). Transduced cells were suspended with 0.05% trypsin-0.53 mM EDTA (Invitrogen, Carlsbad, California), washed with phosphate-buffered saline, and resuspended in EBM. EBM containing VEGF 150 ng/ml was put into the wells of the lower chamber. Polycarbonate filters with 8-µm pores (PVPF; OSMONICS, Kent, Washington) coated with 0.5% gelatin were used to separate the lower chamber and 1 X 10 5 suspended cells in the upper chamber. For some experiments, the indicated concentrations of SNP were added to both chambers. The chambers were then incubated for 8 hours at 37ºC under normoxic or hypoxic conditions. Filters were carefully removed and the cells attached to the upper side were removed by wiping. Cells migrating through the filter (cells on the lower side of filter) were fixed with methanol for 15 minutes, stained with Giemsa stain solution (Sigma) and five random microscopic fields per well were quantified. Each experiment was performed in duplicate and three separate experiments were performed. 9

Tube formation assay
The formation of vascular-like structures by HUVECs on growth factor-reduced Matrigel (Becton Dickinson) was performed as previously described (37). Twenty-four-well culture plates were coated with Matrigel according to the manufacturer's instructions. The indicated adenovirus-transduced HUVECs were seeded on coated plates at 5X10 4 cells/well in EBM containing VEGF 50 ng/ml and incubated at 37ºC for 18 hours under normoxic or hypoxic conditions. In some wells, SNP (1 µM -1 mM) was also added. Tube formation was observed using an inverted phase contrast microscope (Nikon, Tokyo, Japan). Images were captured with a video graphic system (DEI-750 CE Digital Output Camera, Optronics, Goleta, California). The degree of tube formation was quantified by measuring the length of tubes in 3 randomly chosen low-power fields (X100) from each well using the National Institutes of Health (NIH) Image Program. Each experiment was repeated for 4 times.

cGMP assay
Intracellular cGMP concentration was measured using cGMP enzyme immunoassay system purchased from Amersham Pharmacia (Piscataway, New Jersey) according to the manufacturer's instruction. After adenovirus infection, cells were incubated overnight in low serum medium to reduce the effect of mitogens. Then VEGF (50 ng/ml) was added to the cultures in the absence or presence of NO synthase inhibitor, L-NAME (1 mg/ml). Cells were then incubated under 10 hypoxic or normoxic conditions for 6 hours prior to the cGMP measurement. Each experiment was performed in 4 independent samples, and 2 sets of separate experiments were performed.

Mouse angiogenesis assay
The formation of new vessels in vivo was evaluated by Matrigel plug assay as described previously (38,39).

AMPK activation by hypoxia in endothelial cells
To test whether AMPK is activated in endothelial cells under conditions of hypoxia, HUVEC cultures were incubated in a hypoxia chamber for different lengths of time and Western immunoblot analyses were performed on lysates using an antibody that is specific for AMPK phosphorylation at Thr172 of the α subunit. A basal level of AMPK phosphorylation was detected in cell cultured under normoxic conditions, and hypoxia produced a time-dependent increase in AMPK phosphorylation (Fig. 1A). A representative Western blot of this time course is shown (Fig.1A). Thr172 phosphorylation leads to AMPK activation, and the induction of with dominant-negative AMPK also inhibited the hypoxia-induced phosphorylation of endogenous AMPK (Fig. 1A). Hypoxia did not affect the total level of α subunit expression in immunoblot analyses using a pan-α-AMPK antibody (Fig. 1A), nor did it affect the level of ACC protein expression (data not shown 2 ).
Hypoxia led to a modest reduction in basal Akt phosphorylation at Ser 473, which is required for kinase activity (40,41) (Fig.1A). Transduction with dominant-negative AMPK further suppressed Akt phosphorylation, inhibiting Akt phosphorylation at every time point examined (Fig. 1C). The relative magnitude of the inhibition by dominant-negative AMPK increased as the length of exposure to hypoxia increased.

Regulation of eNOS Ser1177 phosphorylation and NO production by AMPK
The production of endothelial NO is important for vascular homeostasis and angiogenesis (42-46). VEGF stimulation promotes Akt signaling which, in turn, produces NO through the activating phosphorylation of eNOS at residue 1177 (27). Because AMPK is also reported to phosphorylate eNOS at residue 1177 in cardiac myocytes (29), the contribution of AMPK to VEGF-stimulated eNOS phosphorylation was analyzed in HUVEC under normoxic and hypoxic conditions ( Fig. 2A). In our assays, VEGF did not affect the level of AMPK expression on phosphorylation (data not shown 2 ). However, transduction with dominant-negative AMPK effectively blocked eNOS phosphorylation under conditions of hypoxia, whereas it had little effect in normoxic HUVEC cultures (Fig. 2B). While basal levels of AMPK in normoxic endothelial cells has little effect on the status of eNOS 1177 phosphorylation, AMPK signaling was essential for ACC phosphorylation under these conditions ( Fig. 2A). A vector expressing dominant-negative Akt was used as a positive control in these experiments because it is known to inhibit VEGF-stimulated eNOS phosphorylation in normoxic endothelial cell cultures (27). In contrast to dominant-negative AMPK, transduction with dominant-negative Akt significantly inhibited eNOS phosphorylation at Ser1177 in normoxic cultures. Transduction with dominant-negative Akt also inhibited eNOS phosphorylation in hypoxic cultures. However, it had no effect on ACC phosphorylation under any condition.
Recently, it has been shown that eNOS activity is negatively regulated through phosphorylation of Thr-495 (29, 47,48). In normoxic cultures, VEGF stimulation led to a decrease in Thr-495 phosphorylation ( Fig. 2 A, C), consistent with a previous report (47). The reduction in Thr-495 phosphorylation by VEGF was reversed by transduction with dominant-negative Akt, while dominant-negative AMPK had no effect. Similar trends in Thr-495 phosphorylation were also seen in hypoxic cultures, but the overall magnitude of the VEGF effect was less than that seen in normoxic cultures. Furthermore, dominant-negative AMPK had no detectable effect on decline in Thr-495 under conditions of hypoxia. The production of NO by eNOS was measured in intact HUVEC cultures by monitoring the accumulation of cGMP (39). Under the conditions of these assays, VEGF-induced cGMP accumulation was similar between HUVEC cultures incubated under normoxic and hypoxic conditions (Fig. 3). Under both conditions, cGMP accumulation was largely inhibited by the eNOS inhibitor L-NAME. Transduction with dominant-negative AMPK significantly inhibited NO output in hypoxic cultures. There was a trend toward lower NO output in normoxic cultures transduced with dominant-negative AMPK, but this was not statistically significant (P = 0.1). In contrast, transduction with dominant-negative Akt significantly inhibited NO output in both hypoxic and normoxic HUVEC cultures. Transduction with dominant-negative Akt or dominant-negative AMPK did not affect basal NO production in the absence of VEGF in either hypoxic or normoxic cultures 2 .

AMPK signaling is essential for endothelial cell migration under hypoxic conditions
To evaluate the role of AMPK signaling on endothelial cell migration, HUVEC were transduced with Ad-dnAMPK and tested for their ability to migrate toward VEGF (150 ng/ml) in a Boyden chamber apparatus. Under hypoxic conditions, transduction with dominant-negative AMPK significantly suppressed cell migration relative to cells transduced with Ad-GFP control (Fig. 4). No differences in cell migration were detected between Ad-GFP-treated and non-transduced cells 2 . Furthermore, the reduction in cell migration did not appear to be due to decreased cell viability. Under these cell culture conditions dominant-negative AMPK-transduced cells did not display detectable increases in the frequencies of TUNEL-or pyknotic nuclei-positive cells, nor was there a detectable decrease in mitochondrial function (data not shown 2 ). Addition of 10µM sodium nitroprusside (SNP), an NO donor, partially reversed the inhibitory action of dominant-negative AMPK on HUVEC migration to VEGF in the hypoxic cultures (Fig. 4). Similar effects were also observed at 1 and 100 µM SNP (data not shown 2 ). However, co-transfection with an adenoviral vector expressing a constitutively-active form of Akt was more effective than the NO donor at reversing the inhibition by dominant-negative AMPK in hypoxic cultures (P<0.05 relative to dominant-negative AMPK plus SNP, and not significant relative to control). Transduction with dominant-negative AMPK had no effect on HUVEC migration to VEGF under normoxic conditions. In contrast, transduction with dominant-negative Akt suppressed migration under normoxic conditions, consistent with previous findings (49), and also exhibited this activity in hypoxic cultures.

AMPK regulates endothelial cell tube formation under hypoxic conditions
To examine the role of AMPK signaling on endothelial cell differentiation into vascular structures in vitro, HUVECs were plated on matrigel in the presence of VEGF after transduction with GFP, dominant-negative AMPK or dominant-negative Akt, and tube formation was assessed under normoxic and hypoxic conditions (Fig. 5A). Tube structure length was quantified after 18 hours using NIH imaging software (Fig. 5B). Transduction with dominant-negative AMPK suppressed tube formation to a similar extent as dominant-negative Akt under hypoxic conditions. The inhibition of tube formation by dominant-negative AMPK was partially reversed by the inclusion of 10 µM SNP in the culture media. Similar effects were observed at 1 and 100 µM SNP (data not shown 2 ). Co-transfection with an adenoviral vector expressing constitutively-active Akt was more effective than the NO donor in reversing the inhibition of tube formation by dominant-negative AMPK (P<0.05 relative to dominant-negative AMPK plus SNP, and not significant relative to control). In contrast, transduction with dominant-negative Akt, but not dominant-negative AMPK, suppressed on tube formation in normoxic cultures (Fig.   5).

AMPK regulates capillary formation in Matrigel plugs in vivo
To assess the role of AMPK signaling on angiogenesis in an in vivo model, matrigel plugs containing adenoviral vectors were subcutaneously implanted on the abdomen of mice. In this assay the matrigel plug serves as a reservoir for the adenoviral vector, and endothelial cells that infiltrate the plug become transduced and express transgene (39). Two weeks after implantation, plugs were harvested and subjected to histochemical analysis using Masson's trichrome staining to identify capillary structures and immunohistochemical analysis using the endothelial cell marker CD31. Matrigel plugs formulated with Ad-dnAMPK showed fewer A key finding of this study is that AMPK signaling has little or no effect on endothelial cell migration, tube formation or NO production when cultures are exposed to normoxia. These angiogenesis in vivo (37,56,57), the data presented here provide the first direct evidence showing that Akt signaling is essential for neovascularization in an animal model.
Several lines of evidence suggest that crosstalk between AMPK and Akt signaling is an integral component of the pro-angiogenic cellular response to hypoxic stress (Fig. 7). It was found that inhibition of AMPK signaling suppresses the activating phosphorylation of Akt at Ser 473 in hypoxic endothelial cells. The observation that the inhibition of endothelial cell migration and differentiation by dominant-negative AMPK can be reversed by co-transduction with constitutively-active Akt is also consistent with an AMPK-Akt crosstalk mechanism. AMPK-Akt crosstalk may also participate in the regulation of eNOS activation by phosphorylation, because transduction with either dominant-negative Akt or dominant-negative AMPK effectively suppressed eNOS Ser1177 phosphorylation and NO output from hypoxic endothelial cells.
Collectively, these data suggest that eNOS regulation is complex in hypoxic endothelial cells and depends upon the relative phosphorylation status of eNOS at Ser1177 and Thr-495, the relative activities of Akt and AMPK, and the length of time that cells are exposed to low oxygen tension which causes relatively slow changes in the activities of both of these signaling steps. Under the conditions of our assays, the direct phosphorylation of eNOS by AMPK appeared to contribute minimally to eNOS phosphorylation or endothelial NO output (Fig. 7). However, direct AMPK-mediated eNOS phosphorylation at Ser1177 may become significant when cultures are constitutively-active Akt was more effective than the NO donor at reversing the inhibitory effects of dominant-negative AMPK. These data suggest that while NO has a stimulatory effect on angiogenic cellular processes, AMPK signaling can regulate angiogenesis independent of NO production (Fig. 7). Similar conclusions have recently been drawn from analyses of the NO-independent regulation of angiogenesis by glycogen synthase kinase-3β (39).

23
In summary, we report that the activation of a pro-angiogenic phenotype in endothelial cells exposed to hypoxia is a new stress-related function for AMPK signaling. The finding that dominant-negative AMPK inhibited the vascularization of implanted Matrigel plugs suggests the physiological importance of AMPK in angiogenesis in vivo. Stimuli that may activate endothelial AMPK in vivo include decreases in oxygen tension that are associated with tumor growth or occlusive vessel diseases (29,59). In contrast, angiogenesis associated with normal tissue growth and organ enlargement during development can be subject to regulatory controls that function independently of tissue oxygen gradients (56,60). Collectively, these data suggest endothelial AMPK signaling may be an essential feature of the angiogenic response to ischemic stress, while having minimal effects on angiogenic endothelial cell responses during development or normal post-natal tissue growth. Therefore, endothelial AMPK may represent a pharmacological target for the selective inhibition of angiogenesis associated with pathological tissue growth.