Copper Transporter ATP7A Protects Against Endothelial Dysfunction in Type 1 Diabetic Mice by Regulating Extracellular Superoxide Dismutase

Heterozygous blotchy ATP7A mutant (ATP7A^mut) mice (21) back-crossed to the C57BL/6J background for 10 generations, SOD3^−/− mice on C57BL/6J background (22), and heterozygous transgenic mice overexpressing ATP7A on C57BL/6J background were weaned at 4 weeks of age and maintained on regular chow for 3 months. C57BL/6J mice were purchased from Jackson Laboratory (Bar Harbor, ME). ATP7A^mut mice carrying the X-linked blotchy ATP7A mutation have a splice site mutation introducing a new stop codon at amino acid residue 794 and show impaired copper transport function, but survive to more than 6 months of age (20). ATP7A transgenic mice that overexpress human ATP7A from the composite β-actin promoter (CAG) were generated as previously described (23). The University of Illinois at Chicago Animal Care and Use Committee approved the protocol for animal use.

Mice were studied between 8 and 12 weeks of age. Diabetes was induced by STZ (100 mg/kg body weight in 0.05 mol/L citrate buffer, pH.4.5, intraperitoneally) on two consecutive days after overnight fasting, and control mice were injected with citrate buffer, as previously described (24). Thirty days after STZ treatment, blood glucose was elevated significantly from 165 mg/dL in control mice to 552 mg/dL in diabetic mice. Regarding mouse body weight, it was not significantly changed at 30 days after STZ treatment as compared with those with vehicle treatment (28.7 ± 3.1 vs. 27.6 ± 3.2; NS), whereas it was significantly decreased at 60 days (24.5 ± 1.51 vs. 28.83 ± 0.75 in control; P < 0.05). The STZ mice were not treated with insulin in this study. Mice became hyperglycemic at 7–10 days after STZ injection for ∼20 days and were killed for experiments at day 30. All reagents were purchased from Sigma-Aldrich (St. Louis, MO), except when specified. For a genetic model of T1DM, we used Ins2^Akita diabetic mice on C57BL/6J background (Jackson Laboratory) which is a well-known animal model of T1DM with an autosomal dominant mutation in the Ins2 gene (25).

Adenovirus expressing human SOD3 (Ad.SOD3) and human SOD3 lacking HBD (Ad.SOD3-ΔHBD) were from the adenovirus core at University of Iowa (Iowa City, IA) (11). Ad.SOD3, Ad.SOD3-ΔHBD, and Lac Z (0.25 mL of 1 × 10¹² particles/mL in 3% sucrose in PBS) were injected intravenously. Three days after viral injection, mice were killed and vascular tissue and plasma were collected for further experiments. Plasma SOD3 proteins were identified by zymography (11).

Tissue harvesting and SOD activity assay were performed as described previously (21). Con A–Sepharose chromatography was used to isolate SOD3 from vessels of diabetic and control mice.

Total tissue and cell lysate were used for Western blot analysis as previously described (26).

VSMCs were isolated from male Sprague-Dawley rat thoracic aortas by enzymatic digestion. Cells were grown in Dulbecco's modified Eagle's medium (DMEM), containing 10% bovine serum and 4.5 g/L glucose, as previously described (27). Before stimulation with insulin, cells were starved for 24 h with serum-free media.

To examine the effect of insulin on intact diabetic vessels in the absence of neurohormonal adaptations and changes in blood pressure, we used organoid cultures, as previously described (28). In brief, after dissection of adventitial tissue, aortic segments were immersed in six-well dishes that contained DMEM and antibiotics (penicillin 100 units/mL, streptomycin 100 mg/L) and was supplemented with 0.1% calf serum. The vessel segments were then maintained in a tissue culture incubator at 37°C and exposed to various experimental conditions.

Control littermate, SOD3^−/−, ATP7A^mut, and ATP7A transgenic mice were killed by CO₂ inhalation. Vascular O₂^•− production was determined using lucigenin-enhanced chemiluminescence as described previously (21).

Isometric tension of mesenteric resistance arteries and mouse aorta were measured using wire myograph as described previously (29).

Data are presented as mean ± SE. Data were compared between groups of cells and animals by Student t test when one comparison was performed or by ANOVA for multiple comparisons. When significance was indicated by ANOVA, the Tukey-Kramer post hoc test was used to specify between group differences. Values of *P < 0.05, **P < 0.01, and ***P < 0.001 were considered statistically significant.

T1DM mice were created by STZ administration, as described in 1research design and methods. Thirty days after the STZ injection, blood glucose was elevated to 552.5 ± 25.15 mg/dL in diabetic mice vs. 165.2 ± 11.71 mg/dL in the C57BL/6J controls, whereas body weight was not significantly changed without insulin (28.7 ± 3.1 vs. 27.6 ± 3.2; NS) (Supplementary Fig. 1). Plasma insulin levels were significantly lowered in STZ-induced diabetes compared with controls.

We examined SOD activity and expression in aortas of STZ-induced diabetic mice. Figure 1A shows that SOD3 activity was significantly (P < 0.05) decreased in diabetic mice, as compared with control mice, whereas protein levels of SOD3 were significantly increased (Fig. 1B). In contrast, SOD1 activity and protein levels were not altered in diabetic vessels. Thus, the specific activity of SOD3, as determined by the ratio of activity to protein, was markedly (P < 0.001) decreased in diabetic vessels (i.e., increased “inactive” SOD3 protein), whereas that of SOD1 was not changed (Fig. 1C). Consistent results were observed in genetic T1DM Ins2^Akita mice (Supplementary Fig. 2A), indicating that the decrease in vascular SOD3 specific activity in STZ-induced diabetic mice is not due to a toxic effect of STZ but due to the diabetic condition. Of note, a decrease in SOD3 specific activity in aorta from T1DM mice was associated with a marked increase in O₂^•− production (Fig. 1D) and nitrotyrosine staining, an indicator of peroxynitrite (ONOO⁻) (data not shown).

Since the activity of SOD1 and SOD3 is dependent on the catalytic copper cofactor (5,17), we examined whether decreased SOD3 specific activity in diabetic vessels is due to deficiency of copper. Figure 1E shows that copper addition restored the decreased specific activity of SOD3 purified from diabetic vessels. In contrast, the specific activities of SOD3 purified from control vessels or that of SOD1 from either diabetic or control vessels were not affected by copper addition. These results suggest that copper loading to SOD3 is selectively impaired in diabetic vessels, whereas SOD3 enzyme from control vessels or SOD1 enzyme from either diabetic or control vessels is fully metallated.

Because increased O₂^•− production can alter endothelial function (5), we examined endothelium-dependent vasorelaxation in diabetic and control mice (Fig. 2). For this purpose, we used mesenteric arteries (∼200 µm in diameter) with the wire myograph approach, which contributes to blood pressure and tissue perfusion (30) as well as aortic segments. Acetylcholine (ACh)-induced endothelium-dependent vasorelaxation was significantly impaired in resistance arteries of diabetic mice compared with control mice (maximum relaxation 68 ± 2 vs. 92 ± 2%, respectively, P < 0.001), which was rescued by the SOD mimetic tempol (Fig. 2A). Of note, sodium nitroprusside (SNP)–induced endothelium-independent vasorelaxation was not different between diabetic and control mice. A similar response was observed in mouse aorta from diabetic mice (Supplementary Fig. 3). These results suggest that impaired endothelium-dependent vasorelaxation in diabetic mice is largely due to increased O₂^•− levels.

Since specific activity of SOD3, which is anchored to endothelial surfaces through HBD, is significantly decreased in diabetic vessels, we next examined the effects of adenoviral-mediated gene transfer of SOD3 (Ad.SOD3) and SOD3 lacking HBD (Ad.SOD3-∆HBD) on endothelial dysfunction in diabetic mice. Ad.SOD3 or Ad.SOD3-∆HBD were injected into diabetic mice intravenously, and the enzymatic activity of SOD3 or SOD3-∆HBD in plasma was confirmed by in-gel zymography (Fig. 2C), as previously reported (11). Interestingly, a more intense band was detected for Ad.SOD3-∆HBD than for Ad.SOD3, because SOD3 binds to vascular tissues, whereas SOD3-∆HBD circulates without binding to vascular tissues. Figure 2B shows that ACh-induced endothelium-dependent vasorelaxation was significantly improved after gene transfer of SOD3, but not SOD3-∆HBD, in diabetic mice (maximal relaxation 69 ± 3 vs. 57.4 ± 2.4%, respectively). Further, SNP-induced relaxation in diabetic mesenteric arteries was not altered by gene transfer of either SOD3 or SOD3-∆HBD. These results suggest that increased O₂^•− production in diabetic vessels is in part due to decreased SOD3 activity.

Because impaired SOD3 specific activity in diabetic vessels is due to copper deficiency, we next examined the expression of copper transport proteins in diabetic and control mice. Figure 3 shows that protein expression of ATP7A, but not Atox1, was decreased (P < 0.001) in diabetic aorta compared with control aorta. In contrast, protein expression for CCS, a copper chaperone for SOD1 in the cytoplasm (17), and COX17, a copper chaperone for cytochrome c oxidase in the mitochondria (18), was not changed in diabetic vessels. A similar response was also observed in mesenteric arteries from diabetic mice (Supplementary Fig. 4). Of note, Ins2^Akita mice, a genetic model of T1DM, exhibited results similar to STZ-induced diabetic mice (Supplementary Fig. 2B), indicating that the decrease in ATP7A expression in STZ-induced diabetic mice is not due to a toxic effect of STZ, but due to the diabetic condition. Thus, diabetic mice show a significant decrease in the expression of vascular copper transporter ATP7A.

To address the mechanism by which ATP7A protein expression is decreased in T1DM vessels, we next examined the role of hyperglycemia or hypoinsulinemia, which are characteristics of T1DM, using cultured VSMCs. In this cell type, ATP7A delivers copper to SOD3 at the secretory pathway, including the trans-Golgi network, and then SOD3 is secreted to the extracellular space (5). Figure 4A shows that insulin treatment (10 nmol/L) for 12 h significantly increased ATP7A protein expression (P < 0.01), but not another copper transport protein (Atox1), in VSMCs, whereas high glucose had no effects (data not shown). Insulin treatment also rescued the T1DM-induced decrease in SOD3 activity (Fig. 4B) and ATP7A protein expression (Supplementary Fig. 5) without affecting SOD1 activity in the organoid culture. These results suggest that hypoinsulinemia in T1DM may contribute to a decrease in ATP7A protein expression, resulting in decreased SOD3 specific activity.

Given that ATP7A plays a critical role in delivering cofactor copper to SOD3 for its full activation, we hypothesized that impaired endothelium-dependent vasorelaxation and SOD3 activity in diabetic mice might be due to a reduction of ATP7A expression. To address this question, we used ATP7A-overexpressing transgenic mice (23) and found that ATP7A protein expression was significantly increased by three- to fourfold in ATP7A transgenic compared with wild-type (WT) mice (Fig. 5A). Figure 5C and D shows that the decrease in SOD3 activity and specific activity in diabetic WT mice aorta was significantly improved in diabetic ATP7A transgenic mice. In parallel, diabetes-induced enhanced O₂^•− production was significantly decreased in diabetic ATP7A transgenic mice compared with WT mice (Fig. 5B). ACh-induced endothelium-dependent relaxation was significantly improved in resistance arteries of diabetic ATP7A transgenic mice compared with diabetic WT mice (maximal relaxation 74.8 ± 3.9 vs. 60.1 ± 3%, respectively) (Fig. 6A), whereas endothelium-independent relaxation to SNP was not changed between the two groups (Fig. 6B). These findings suggest that decreased ATP7A expression in diabetic vessels contributes to decreased SOD3 activity, thereby enhancing O₂^•− production and endothelial dysfunction.

To examine the role of endogenous ATP7A and SOD3 in diabetic vessels, we used mice lacking SOD3 (SOD3^−/− mice) or ATP7A dysfunctional mutant mice (ATP7A^mut mice) (20). Endothelium-dependent relaxation to ACh was markedly (P < 0.001) impaired in resistance arteries of diabetic SOD3^−/− and ATP7A^mut mice compared with diabetic WT mice, which was rescued by addition of the SOD mimetic tempol (Fig. 7A). In contrast, endothelium-independent relaxation to SNP was not changed between the two groups (Fig. 7A). Vascular O₂^•− production in T1DM was significantly increased in SOD3^−/− and ATP7A^mut mice to a greater extent than in diabetic WT mice (Fig. 7B). These findings further suggest that endogenous ATP7A and SOD3 play an important role in protecting endothelial function by regulating vascular O₂^•−.

The current study demonstrates a protective role for the Cu transporter ATP7A in T1DM-induced endothelial dysfunction by regulating SOD3 activity and vascular O₂^•− levels (Fig. 8). We found the following. 1) Specific activity of SOD3, but not SOD1, is significantly decreased in diabetic vessels, which is associated with increased O₂^•− production compared with control vessels. 2) In ex vivo experiments, addition of copper partially rescues decreased SOD3 specific activity in diabetic vessels. 3) Functionally, endothelium-dependent relaxation is impaired in mesenteric arteries of T1DM, which is rescued by SOD mimetic tempol or gene transfer of SOD3. 4) Mechanistically, ATP7A expression in T1DM aorta is dramatically decreased whereas other copper transport proteins (Atox1, CCS, and COX17) are not changed. 5) These effects may be due to hypoinsulinemia in T1DM mice, since insulin treatment, but not high glucose, increases ATP7A expression in VSMCs and restores SOD3 activity in organoid culture of T1DM vessels. 6) Transgenic mice overexpressing ATP7A exhibit restored T1DM-induced impaired endothelial function and SOD3 activity and increased vascular O₂^•− production. 7) The SOD3^−/− T1DM or ATP7A^mut T1DM mice show augmented endothelial dysfunction and increased O₂^•− production versus T1DM mice. Thus, restoring ATP7A-SOD3 function is an important therapeutic strategy for oxidant stress–dependent cardiovascular and metabolic diseases.

Previous studies have reported either increased or decreased total SOD activity (31–33) as well as either unaltered or decreased SOD1 expression in vascular tissues from diabetic animals (34–36) or in vascular progenitor cells from diabetic patients (37). However, the role of endogenous SOD3 in endothelial dysfunction in T1DM has not been investigated. In the current study, we provide compelling evidence that the specific activity of SOD3 is markedly decreased, and SOD3 protein expression is increased in aortas from STZ-induced diabetic mice, whereas activity or expression of SOD1 is not altered. This result is consistently observed in the genetic model of Ins2^Akita+/− T1DM mice. Our findings are divergent with a previous report showing that glycated SOD3 is increased in the serum (14), and tissue-bound SOD3 is decreased in diabetic patients or animals compared with nondiabetic controls (9,35), which is likely due to the reduction of heparin affinity of SOD3 via nonenzymatic glycation without changing the enzymatic activity (14,38). Given that pediatric patients show a significant decrease in plasma SOD3 compared with control subjects (39), it is conceivable that tissue-bound SOD3 levels in diabetes may be regulated by aging. Indeed, the current study has used young mice (3–4 months of age) to demonstrate that the tissue SOD3 level is increased in T1DM mice aorta. Alternatively, since copper loading to SOD3 in diabetic vessels is impaired, as discussed below, increased SOD3 protein levels in diabetic vessels (aortas and mesenteric arteries) may be caused by the accumulation of immature SOD3, as copper deficiency increases the ceruloplasmin accumulation in a pre-Golgi compartment in hepatocytes (40). Taken together, these results suggest that changes of SOD3 protein levels do not reflect its activity and that T1DM mice exhibit decreased SOD3 activity without altering SOD1 activity, which may contribute to increased vascular O₂^•− production.

The functional significance of decreased SOD3 activity in diabetic vessels is demonstrated by the finding that endothelium-dependent relaxation in mesenteric arteries is impaired in T1DM mice, which is rescued by gene transfer of SOD3 as well as the SOD mimetic tempol. By contrast, endothelium-independent vasorelaxation is not affected in diabetic mice, suggesting that inhibition of ACh-induced vasodilation in T1DM is likely attributed to a decrease in endothelial NO bioavailability, which is supported by increased ONOO⁻ formation, assessed by nitrotyrosine staining. Consistent with our results, previous studies reported that gene transfer of SOD3 rescues the endothelial function in other pathological conditions, such as pulmonary hypoxia, hypertension, and aging (5), and in different vascular beds (13). In the current study, we also found that SOD3-deficient T1DM mice show augmented impaired endothelium-dependent vasodilation and O₂^•− production versus T1DM mice, supporting the protective role of endogenous SOD3 in endothelial function. Our current study also shows that gene transfer of SOD3 lacking HBD does not rescue impaired endothelium-dependent relaxation in diabetic vessels. This result strongly supports the notion that binding of SOD3 to the endothelial surface and extracellular matrix of vascular tissue via HBD is essential for protecting against inactivation of endothelium-derived NO by O₂^•− during diffusion to vascular smooth muscle to induce vasorelaxation (5,8,11). Taken together, SOD3 plays an important role in protecting endothelial function by scavenging extracellular O₂^•− in the diabetic vessel wall.

Mechanistically, we found that the decreased specific activity of SOD3 purified from T1DM aorta is partially restored by Cu addition, suggesting that Cu loading to SOD3 is impaired in diabetic vessels. It has been shown that full activation of SOD3 requires copper transporter ATP7A and copper chaperone Atox1, which are involved in copper delivery to SOD3 (20,21,29,41), whereas SOD1 gets copper through copper chaperone CCS (17). In the current study, we provide the first evidence that ATP7A protein expression, but not Atox1, CCS, or COX17 (copper chaperone for cytochrome c oxidase), is selectively and significantly decreased in STZ-induced T1DM vessels. Consistent results are observed in genetically induced (i.e., Ins2^Akita) T1DM mice. Thus, these findings may explain why the activity of SOD3, but not SOD1, is decreased in T1DM vessels. Furthermore, our data imply that copper transport systems coupled to distinct copper enzymes can be differently regulated in response to T1DM, as reported for hypoxia (42). To address the mechanism by which ATP7A protein expression is decreased in T1DM vessels, we next examined the role of hyperglycemia or hypoinsulinemia, which are characteristics of T1DM, in VSMCs. Here we show that insulin treatment of VSMCs increases ATP7A expression without affecting the Atox1 level, whereas high glucose alone has no effects. Consistent with this, Hardman et al. (43) reported that insulin regulates ATP7A expression in human placental Jeg-3 cells. Furthermore, we found that insulin treatment directly restores the T1DM-induced decrease in SOD3 activity and ATP7A protein expression without affecting SOD1 activity in the organoid culture of T1DM vessels, which can exclude the possibility of a neurohormonal effect of insulin. Thus, these results indicate that hypoinsulinemia in T1DM may contribute to the decrease in ATP7A protein expression in VSMCs, thus reducing SOD3 activity in the blood vessels. Our preliminary study found that insulin increases ATP7A protein stability in a phosphatidylinositol (PI) 3-kinase/Akt-dependent manner. Given that the insulin–PI 3-kinase/Akt pathway has been shown to be impaired in vascular tissue from T2DM mice (44,45), it is tempting to speculate that a decrease in the insulin–PI 3-kinase/Akt pathway in T1DM and T2DM may contribute to the downregulation of ATP7A expression in VSMCs. This may result in a decrease in SOD3 activity and subsequent overproduction of O₂^•−, thereby inducing endothelial dysfunction. This issue should be investigated more in detail in future studies.

In this study, we have demonstrated the functional role of ATP7A in regulating SOD3 activity and endothelial function in diabetic vessels using transgenic mice overexpressing ATP7A or ATP7A^mut mice. Here we show that T1DM-induced endothelial dysfunction and decrease of SOD3 activity are rescued in transgenic mice overexpressing ATP7A, whereas SOD3-deficient T1DM mice or ATP7A^mut T1DM mice accelerate endothelial dysfunction and vascular O₂^•− production compared with T1DM mice. Consistent with this, we previously reported that angiotensin II–induced hypertension and endothelial dysfunction are further augmented in ATP7A^mut mice (21). Importantly, ATP7A plays a role not only in providing copper to some secretory cuproenzymes but also in regulating intracellular levels of copper by exporting copper (46). It has been shown that ATP7A-overexpressing transgenic mice alter tissue copper homeostasis (23), and that tissue copper levels are increased in STZ-induced diabetes, which is rescued by insulin treatment (47). Thus, it is tempting to speculate that increased copper levels in diabetes may be caused by decreased ATP7A expression and subsequent inhibition of copper export. Furthermore, due to its toxicity, increased copper levels may contribute to the pathogenesis of diabetic vascular complications. Indeed, copper chelation therapy has been shown to mitigate various pathogenic states of diabetes, such as left ventricular hypertrophy in diabetic patients (48), diabetic neuropathy (49), and diabetic nephropathy (50). Taken together, it is likely that the copper transporter protein ATP7A plays an important role in preventing diabetes-induced endothelial dysfunction by regulating SOD3 activity as well as intracellular copper homeostasis.

In summary, the current study provides direct evidence for the protective role of the ATP7A-SOD3 pathway in endothelial function by reducing extracellular O₂^•− levels and increasing bioavailability of NO in T1DM (Fig. 8). These findings provide novel insight into ATP7A as a potential therapeutic target for the treatment of oxidative stress–dependent cardiovascular and metabolic diseases such as T1DM.

This research was supported by National Institutes of Health (NIH) R01-HL-70187; Department of Veterans Affairs Merit Review Grant I01BX001232 (to T.F.); NIH R01-HL-077524, HL-077524-S1, and R21-HL-112293 (to M.U.-F.); and American Heart Association 12SDG12060100 (to N.U.) and 11POST5740006 (to V.S.).

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

V.S. designed the study, performed experiments, analyzed data, and wrote the manuscript. N.U. and J.O. helped to make diabetic mice. R.D.M. conducted mouse husbandry and genotyping. R.M.L. and J.F.B.M. developed the ATP7A transgenic mice. M.U.-F. and T.F. designed the overall study, analyzed data, and wrote, reviewed, and edited the manuscript. T.F. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Parts of this study were presented in abstract form at Experimental Biology 2012, San Diego, California, 21–25 April 2012.

The authors thank Dr. Maria Linder (California State University, Fullerton, CA) for sending the ATP7A transgenic mice, and Dr. Ayako Makino (University of Illinois at Chicago) for helpful discussions.