In Vivo Activation of AMP-Activated Protein Kinase Attenuates Diabetes-Enhanced Degradation of GTP Cyclohydrolase I

A full description of materials, animals, and methods used, including cell culture, gene transfection to HUVECs, measurement of biopterins, detections of NO and reactive oxygen species (ROS), streptozotocin (STZ)-induced diabetes in mice, 26S proteasome activity assay, immunoprecipitation and Western blot analysis, reverse-transcription PCR for GTPCH I, and organ chamber, can be found in the online-only data supplement available at http://diabetes.diabetesjournals.org/cgi/content/full/db09-0267.

Ad-GFP, a replication-defective adenoviral vector expressing green fluorescence protein (GFP), was used as control. An adenoviral vector expressing a dominant-negative mutant of AMPK (AMPK-DN) was constructed from AMPK bearing a mutation altering lysine 45 to arginine (K45R), as described previously (23). To generate an adenoviral vector expressing a constitutively active mutant of AMPK (AMPK-CA), a rat cDNA encoding residues 1–312 of AMPK and bearing a mutation of threonine 172 to aspartic acid (T172D) was subcloned into a shuttle vector (pShuttle CMV [cytomegalovirus]).

A low-dose STZ-induction regimen was used to induce pancreatic islet cell destruction and persistent hyperglycemia, as previously described (24) by the Animal Models of Diabetic Complications Consortium (http://www.amdcc.org). Hyperglycemia was defined as a random blood glucose level of >300 mg/dl for >2 weeks after injection.

The levels of BH4 and total biopterins were determined by high-performance liquid chromatography, as previously described with some modification (25).

The levels of O₂^–. production in cultured cells were detected by using dihydroethidium (DHE) fluorescence assay. NO release in cultured cells was detected by using the DAF fluorescent probe.

The 26S proteasome function was measured, as described previously (26).

Reverse-transcription PCR was performed according to the manufacturer's protocol (ThermoScript, RT-PCR System, Invitrogen). PCR primer pairs were as described previously (27).

Organ chamber study was performed, as described previously (28).

Data are reported as means ± SE. All data were analyzed by one- or two-way ANOVA followed by multiple t tests, and P < 0.05 was considered statistically significant.

Data published by our group and others demonstrate that AMPK activation exerts beneficial effects by increasing NO bioactivity via an increase in the phosphorylation of eNOS and/or an increase in the anti-oxidant potential of endothelial cells (29,–31). Therefore, it was of interest to investigate the effects of hyperglycemia/diabetes on AMPK activity. As the phosphorylation of Thr172 of AMPKα is required for AMPK activity, and because AMPK-Thr172 is closely related to AMPK activity, we first measured the levels of phosphorylated AMPK-Thr172 in mouse aortas isolated from mice with STZ-induced diabetes or nondiabetic mice. As depicted in Fig. 1,A, the levels of phospho-AMPK (Thr172) were significantly reduced in diabetic mouse aortas relative to nondiabetic mice. In contrast, the level of AMPKα was not different in diabetic and nondiabetic mice (Fig. 1,A). Consistent with our earlier report (9), the levels of GTPCH I were significantly reduced in diabetes relative to the control group (Fig. 1,B). In contrast, the levels of eNOS were not altered in diabetic mice or metformin-treated mice (Fig. 1 B).

Metformin, one of most widely used antidiabetic drugs, markedly reduces cardiovascular incidences and improves endothelial function in patients with diabetes (32). Our earlier studies reported that metformin increases NO bioactivity by activating AMPK (29). We sought to determine whether metformin alters the diabetes-induced reduction of GTPCH I levels. Administration of metformin (300 mg · kg⁻¹ · day⁻¹) for 4 consecutive weeks in the drinking water, which did not alter the blood glucose levels in control or diabetic mice (P > 0.05, data not shown), significantly increased the levels of both AMPK-Thr172 and GTPCH I in aortas isolated from both diabetic and nondiabetic mice (Fig. 1,A and B). Consistently, meformin attenuated the diabetes-induced reduction of both biopterins and BH4 in vivo (Fig. 1 C).

We next asked whether metformin reverses the diabetes-induced endothelial dysfunction in vivo. Endothelium-dependent relaxation was assayed in aortas isolated from metformin-treated diabetic mice. Compared with nondiabetic mice, the acetylcholine (ACh)-induced endothelium-dependent relaxation in diabetes was markedly reduced (Fig. 1,D). Further, metformin alone did not affect the ACh-induced endothelium-dependent relaxation in control mice (data not shown) but abolished the diabetes-induced impairment of endothelium-dependent relaxation (Fig. 1 D). In contrast, sodium nitroprusside (SNP)-induced endothelium-independent relaxation was not altered by diabetes or metformin treatment (data not shown).

l-sepiapterin, a precursor of BH4, is converted to BH4 via a salvage pathway (33). We next investigated whether supplementation of l-sepiapterin improved endothelial dysfunction ex vivo. Aorta isolated from diabetic mice were incubated with l-sepiapterin (10 μmol/l) for 1 h in an organ bath. As shown in Fig. 1 D, l-sepiapterin, which did not alter the endothelium-dependent relaxation in control mice (data not shown), restored the ACh-induced maximal relaxation (81.6 ± 8.9% vs. 47.8 ± 9.6%, P < 0.05) in diabetic mice, implying that diabetes-induced endothelial dysfunction was likely to be because of BH4 deficiency.

As metformin treatment increased the amount of GTPCH I and BH4 in diabetic aortas, we reasoned that metformin improves endothelial function through the inhibition of diabetes-induced GTPCH I degradation. To test this hypothesis, cultured HUVECs were exposed to normal glucose (5 mmol/l), high glucose (30 mmol/l d-glucose), or hyperosmotic control (5 mmol/l d-glucose plus 25 mmol/l l-glucose) in the presence or absence of AICAR (2 mmol/l), an AMPK activator. High glucose did not alter AMPK and eNOS expressions (Fig. 2,A). However, consistent with the reduction of both total biopterins and BH4 levels, the levels of AMPK-Thr172 were markedly reduced in high-glucose–exposed HUVECs (Fig. 2,A and B). Consistent with our previous report, neither total biopterins nor BH4 levels were altered in the hyperosmatic control group, compared to HUVECs in normal glucose (Fig. 2,B). As expected, AICAR treatment increased the level of AMPK phosphorylation at Thr172 in HUVECs exposed to either normal glucose or high glucose without altering AMPK and eNOS expressions (Fig. 2 A).

We next determined the effects of high glucose on GTPCH I. As expected, the level of GTPCH I was significantly reduced in high-glucose–exposed cells compared with control cells (Fig. 2,A). Consistently, both total biopterins and BH4 levels were lower in high-glucose–treated cells than those in controls. AICAR treatment increased the levels of both total biopterins and BH4 in HUVECs exposed to normal glucose or high glucose (Fig. 2 B).

As high glucose reduced the levels of both AMPK-Thr172 and GTPCH I in HUVECs, we next asked whether genetic inhibition of AMPK would cause a reduction in GTPCH I levels. Toward this end, AMPK was suppressed by adenoviral overexpression of AMPK-DN. Overexpression of AMPK-DN but not GFP significantly reduced the levels of GTPCH I (Fig. 2,C), total biopterins (Fig. 2,D), and BH4 (Fig. 2 D) in HUVECs.

To establish the essential role of AMPK in metformin-enhanced GTPCH I protein levels, metformin was administered in HUVECs infected with AMPK-DN or GFP. As shown in Fig. 2 C and D, metformin increased GTPCH I, total biopterins, and BH4 levels in GFP-infected HUVECs but not in cells infected with AMPK-DN, implying that AMPK is required for the metformin-induced inhibition of GTPCH I reduction in HUVECs.

We next determined the effects of genetic activation of AMPK on GTPCH I, biopterins, and BH4 by over expressing a constitutively active AMPK (AMPK-CA). As described in Fig. 3,A and B, overexpression of AMPK-CA but not GFP abolished the high-glucose–induced reduction of GTPCH I (Fig. 3,A), biopterins (Fig. 3,B), and BH4 levels (Fig. 3 B).

We next determined if AMPK activation altered eNOS-derived NO/ROS production in response to high glucose. As expected, high glucose significantly increased ROS while it suppressed NO production in HUVEC (Fig. 3 C). As expected, AICAR significantly increased NO release but decreased ROS, implying that increased GTPCH I and consequent BH4 increase might normalize the function of eNOS in HUVEC.

Increased levels of GTPCH I could be because of the transcriptional activation of the GTPCH I gene or attributable to the decrease of GTPCH I protein degradation. We first determined whether AMPK activation altered GTPCH I mRNA levels in HUVECs. As shown in Fig. 4,A, treatment of HUVECs with AICAR (2 mmol/l) for 24 h did not alter the levels of GTPCH I mRNA in HUVECs exposed to normal glucose or high glucose. In addition, neither cycloheximide nor actinomycin D altered the levels of GTPCH I protein caused by AICAR or metformin (Fig. 4,B). Neither cycloheximide nor actinomycin D altered the high-glucose–induced reduction of GTPCH I levels in HUVECs (Fig. 4 B). Taken together, our data suggest that the increase of GTPCH I protein levels caused by AMPK activation is independent of transcriptional activation of the GTPCH I gene.

The levels of GTPCH I protein are controlled by both GTPCH I synthesis and GTPCH I degradation. Our earlier study demonstrated that high glucose activates the 26S proteasome, resulting in the enhanced degradation of GTPCH I in endothelial cells (9). Co-administration of MG132 (0.5 μmol/l, 4 h), a potent proteasome inhibitor, abolished the high-glucose–induced reduction of GTPCH I levels (Fig. 5 A), confirming that the alteration of GTPCH I levels was because of the activation of the 26S proteasome.

We next investigated the effects of AMPK activation on 26S proteasome activity in HUVECs. As shown in Fig. 5 B, high glucose exposure significantly increased 26S proteasome activity, whereas AICAR treatment significantly reduced 26S proteasome activity in cells exposed to normal or high glucose.

We next determined whether genetic activation of AMPK with AMPK-CA would alter 26S proteasome activity in HUVECs. Infection of AMPK-CA but not GFP reduced 26S proteasome activity in control cells (Fig. 5 C). In addition, AMPK-CA but not GFP abolished high-glucose–enhanced 26S proteasome activity in HUVECs. These results suggest that AMPK functions as a physiological suppressor for 26S proteasome-mediated GTPCH I degradation in HUVECs.

Next, we examined whether activation of AMPK affects the ubiquitination of GTPCH I in high-glucose–treated HUVECs. HUVEC were incubated with high glucose for 24 h with or without AICAR. After the incubation, GTPCH I was immunoprecipitated by the specific antibody against GTPCH I. GTPCH I ubiquitination was identified in Western blotting by using the antibody against ubiquitinated protein. As shown in Fig. 5,D, AICAR alone significantly increased the level of ubiquitination GTPCH I in HUVEC. Conversely, high glucose significantly decreased the ubiquitination GTPCH I in HUVEC (Fig. 5,D). Interestingly, co-administration of high glucose with AICAR abolished the effects of high glucose on the levels of ubiquitinated GTPCH I in HUVEC (Fig. 5 D).

We next determined the effects of AMPK depletion on 26S proteasome activity and the levels of GTPCH I, biopterins, and BH4 in isolated mice aortas. We used AMPKα1 or -α2 knockout mice. As shown in Fig. 6 A, relative to genetic control mice, C57BL6, GTPCH I protein levels were markedly lower in AMPKα2^−/− but not in AMPKα1^−/− mice aortas, implying that AMPKα2 subunit plays a key role in the regulation of GTPCH I protein expression. MG132 treatment (0.5 μmol/l, 4 h), which only slightly increased GTPCH I levels in control and AMPKα1^−/− mice, significantly increased GTPCH I levels in the aortas of AMPKα2^−/− mice.

The deficiency of GTPCH I contributes to the lack of BH4 and eNOS uncoupling in endothelial cells (9,34). We determined the levels of both total biopterins and BH4 in wild-type (WT) and AMPKα2^−/− mice aortas. As shown in Fig. 6 B, both total biopterins and BH4 levels were lower in AMPKα2^−/− mice aortas than those in WT mice aortas (P < 0.05), but injection of AMPKα2^−/− mice with MG132 (5 mg · kg⁻¹ · day⁻¹ for 2 days, intraperitoneally) reversed the effects on both total biopterins and BH4 levels in aortas.

We further detected proteasome activity in AMPKα2^−/− mice aortas. As shown in Fig. 6 C, the 26S proteasome activity was greater in AMPKα2^−/− mice than that in WT mice. Metformin treatment suppressed 26S proteasome activity in WT mice, but not in AMPKα2^−/− mice, implying that AMPKα2 was required for metformin's effects on 26S proteasome in vivo.

To establish a cause-effect relationship between proteasome-dependent GTPCH I degradation and impaired endothelium-dependent relaxation in AMPKα2^−/− mice, we injected AMPKα2^−/− mice with MG132 (5 mg · kg⁻¹ · day⁻¹ for 2 days, intraperitoneally) or l-sepiapterin (10 mg · kg⁻¹ · day⁻¹ for 7 days, intraperitoneally). After the treatments, mouse aortas were isolated and assayed for endothelium-dependent and endothelium-independent relaxation. As shown in Fig. 7,A, the maximal relaxation (Emax) induced by ACh is 83.63 ± 7.60% in WT aortas. However, Emax of ACh is only 60.9 ± 5.5 in AMPKα2^−/− aortas (P < 0.05 vs. WT). MG132 did not alter the ACh-induced endothelium-dependent relaxation in WT mice but improved the relaxation in AMPKα2^−/− mice. As shown in Fig. 7 B, l-sepiapterin also mitigated the decrease of ACh-induced Emax in AMPKα2^−/− mice (73.82 ± 9.50% vs. AMPKα2^−/−, P < 0.05). SNP-induced, endothelium-independent relaxations in aortas were comparable among three groups (data not shown). Taken together, these results support the idea that the impaired endothelial function of AMPKα2^−/− mice is because of proteasome-dependent GTPCH I degradation.

The major finding of this study is that AMPK inhibition in diabetes or genetic depletion of AMPK leads to abnormal 26S proteasome activity, resulting in the accelerated degradation of GTPCH I, a key enzyme in the de novo synthesis of BH4. The regulation of GTPCH I function can be achieved at both the mRNA and protein levels (stability, activity, and modification) (27). Here, we found that AMPK regulates GTPCH I protein stability in endothelial cells by inhibiting the 26S proteasome. This conclusion is supported by several findings: First, inhibition of the proteasome by MG132 produced suppressive effects similar to AMPK activation on high-glucose–triggered GTPCH I degradation. Second, inhibition of AMPK by high glucose or knockout of AMPKα2 increased 26S proteasome activity. Third, overexpression of AMPK-CA or AICAR abolished high-glucose–induced upregulation of the 26S proteasome activity. Consistent with our result, Moreno et al. and Viana et al. have reported that two AMPK activators, AICAR and A769662, inhibit 26S proteasome activity in WT mouse embryonic fibroblast cells and human skin fibroblast cells through AMPK-dependent and independent mechanisms (22,35). Fourth, earlier studies by our group and others have demonstrated that GTPCH I protein is ubiquitinated and degraded by the 26S proteasome system under conditions of hyperglycemia or 4-hydroxy-2-nonenal (9,34). Finally, deletion of AMPKα2 increased the ubiquitination of GTPCH I in the presence of MG132 in HUVECs and mice aortas. However, how AMPK activation suppresses 26S proteasomes remains largely unknown and warrants further investigation.

Metformin, which was shown recently to exert its therapeutic effect in diabetes by activating AMPK (36), is known to improve vascular functions and to dramatically reduce cardiovascular end points and mortality for diabetic patients in large-scale clinical trials (32). Increasing evidence suggests that AMPK activation leads to a series of beneficial metabolic consequences, including augmentation of fatty acid oxidation and glucose uptake in skeletal muscle, as well as inhibition of glucose production, lipogenesis, and cholesterol synthesis in the liver. In addition to producing beneficial metabolic effects, AMPK may regulate endothelial function. There are abundant data to suggest that AMPK is critical for eNOS activation by VEGF, shear stress, estradio, adiponectin, and antidiabetic drugs, including metformin (37). AMPK-dependent eNOS activation is generally ascribed to increased phosphorylation of eNOS at Ser1177 by AMPK, increased association of HSP90 with eNOS, or both. As BH4 is an essential cofactor for the NO synthetic activity of eNOS and GTPCH I deficiency uncouples eNOS (producing superoxide anions instead of NO), AMPK-mediated inhibition of GTPCH I degradation might be essential for maintaining the physiological functions of eNOS. Thus, this study is the first to uncover a novel mechanism (Fig. 7 C) by which AMPK activation prevents or ameliorates vascular endothelial function by suppressing proteasome-dependent GTPCH I degradation. Notably, a recent study of obese rats (38) demonstrated that decreased AMPK activity in aortic endothelium is closely associated with impaired endothelium-dependent relaxation, as well as an increased number of apoptotic endothelial cells, raising the possibility that a decline in AMPK activity contributes to endothelial dysfunction (37). AMPK could be an important target in treating cardiovascular complications in diabetes.

J.X. is supported by a postdoctoral fellowship from the American Heart Association. This work was supported by National Institutes of Health grants (HL079584, HL074399, HL080499, HL089920, and HL096032), as well as research awards from the American Diabetes Association, the Juvenile Diabetes Research Foundation, the Oklahoma Center for Advancement of Science and Technology, and the Travis Endowed Chair in Endocrinology, University of Oklahoma Health Sciences Center. M.H.Z. is a recipient of the National Established Investigator Award of the American Heart Association.

No potential conflicts of interest relevant to this article were reported.

We thank Sima Asfa, Melissa Guzman, and Junhua Zhang for their excellent technical support.