Cardiomyocyte GTP Cyclohydrolase 1 Protects the Heart Against Diabetic Cardiomyopathy

Diabetic cardiomyopathy increases the risk of heart failure and death. At present, there are no effective approaches to preventing its development in the clinic. Here we report that reduction of cardiac GTP cyclohydrolase 1 (GCH1) degradation by genetic and pharmacological approaches protects the heart against diabetic cardiomyopathy. Diabetic cardiomyopathy was induced in C57BL/6 wild-type mice and transgenic mice with cardiomyocyte-specific overexpression of GCH1 with streptozotocin, and control animals were given citrate buffer. We found that diabetes-induced degradation of cardiac GCH1 proteins contributed to adverse cardiac remodeling and dysfunction in C57BL/6 mice, concomitant with decreases in tetrahydrobiopterin, dimeric and phosphorylated neuronal nitric oxide synthase, sarcoplasmic reticulum Ca2+ handling proteins, intracellular [Ca2+]i, and sarcoplasmic reticulum Ca2+ content and increases in phosphorylated p-38 mitogen-activated protein kinase and superoxide production. Interestingly, GCH-1 overexpression abrogated these detrimental effects of diabetes. Furthermore, we found that MG 132, an inhibitor for 26S proteasome, preserved cardiac GCH1 proteins and ameliorated cardiac remodeling and dysfunction during diabetes. This study deepens our understanding of impaired cardiac function in diabetes, identifies GCH1 as a modulator of cardiac remodeling and function, and reveals a new therapeutic target for diabetic cardiomyopathy.


GCH1 is obligatory for cardiac function.
To determine whether GCH1 is important for cardiac function, we used the specific inhibitor of GCH1, 2,4-diamino-6-hydroxy-pyridine (DAHP) 34 , to treat C57BL/6 WT mice and measured blood pressure, in vivo left ventricular pressure, and in vivo and ex vivo cardiac function. After the animals were orally given DAHP for 4 weeks 34 , cardiac GCH1 activity was significantly decreased in WT + DAHP group compared with WT control (Fig. 1A). A pressure-tipped catheter was inserted into the right carotid artery and subsequently the chamber of the left ventricle (LV) to measure the pressure 35 . Diastolic and systolic blood pressure, in vivo LV systolic pressure and diastolic pressure were significantly elevated in WT + DAHP groups compared with WT controls (P < 0.05, n = 10 mice/group) ( Fig. 1B-E). The values of + dP/dt (an indicator of systolic function) and − dP/dt (an indicator of diastolic function) calculated from contraction and relaxation of the LV were significantly greater in WT + DAHP than WT control groups (Fig. 1F,G). Hypertension is well known to affect cardiac function 36,37 . To study the direct effect of GCH1 inhibition on the heart, we measured cardiac function in isolated Langendorff-perfused hearts, as described 38,39 . The LV diastolic pressure in Langendorff-perfused hearts was adjusted to the level of the in vivo LV pressure (5.8 ± 0.6 mmHg in WT control and 17.5 ± 1.9 mmHg in WT + DAHP group) (Fig. 1D,H). The ex vivo LV systolic pressure was significantly decreased in WT + DAHP group compared with WT control (P < 0.05) (Fig. 1I). The values of ± dP/dt were significantly smaller in WT + DAHP than WT control groups (Fig. 1J,K). Coronary flow rate was not altered by DAHP (Fig. 1L). These results suggest that cardiac GCH1 is necessary for cardiac function in the mouse.

Diabetes results in degradation of cardiac GCH1 proteins in WT mice. Streptozotocin (STZ)
destroys pancreatic β -cells and is often used to induce T1DM of insulin deficiency in experimental animals 40,41 . We used STZ to induce diabetes in C57BL/6 WT mice (WT STZ). Control animals were given citrate buffer (WT control). Fasting blood glucose, GCH1 mRNA, and GCH1 proteins were measured at baseline (0 weeks after administration of STZ or citrate buffer) and 2, 4, 8, and 12 weeks after intraperitoneal injection of STZ or citrate buffer, as previously shown 42 . There were no significant differences in blood glucose, GCH1 mRNA, and GCH1 proteins between WT STZ and WT control groups at baseline (Fig. 2). Administration of STZ induced a dramatic increase in blood glucose levels but did not change the levels of GCH1 mRNA from 2 to 12 weeks after injection of STZ. Interestingly, the expression of GCH1 proteins was significantly decreased compared with baseline levels (P < 0.05, n = 6 mice/group). Since GCH1 mRNA levels were not changed, decreased GCH1 proteins by diabetes were not due to reduction of synthetic GCH1 rather than increased degradation of GCH1 proteins. GCH1 transgene drives GCH1 protein expression in non-diabetic and diabetic myocardium. To study how transgenic GCH1 impacts cardiac GCH1 proteins in diabetics, the Tg mice and their WT littermates were made diabetic with STZ (Tg STZ and WT STZ) or given citrate buffer as control (Tg control and WT control). There were no significant differences in body weight and fasting blood glucose among 4 experimental groups at baseline and between Tg control and WT control groups from 2 to 12 weeks after injection of the vehicle (Fig. 3). Compared with WT control groups, body weight was decreased in WT STZ and Tg STZ groups 8 and 12 weeks after induction of diabetes (Fig. 3A), and blood glucose was significantly increased in WT STZ and Tg STZ groups from 2 to 12 weeks after induction of diabetes (P < 0.05, n = 10 mice/group) (Fig. 3B). There were no significant differences in body weight and fasting blood glucose between Tg STZ and WT STZ groups throughout the experiment (P > 0.05). Heart weight normalized to tibia length was greater in WT STZ or Tg STZ than WT control groups 12 weeks after induction of diabetes (Fig. 3C). Cardiomyocyte cross-sectional areas measured from Masson's trichrome-stained hearts were larger in WT STZ than WT control groups (Fig. 3D). There were no significant differences in myocyte cross-sectional area between Tg control and WT control or Tg STZ groups.
In the Tg mice, human GCH1 gene was transferred to mice 32 . We used Western blot to measure the expression of human and mouse GCH1 proteins in both diabetics and nondiabetics. In the Tg mice, both mouse and human GCH proteins were expressed in myocardium (Fig. 3E). Expression of human GCH1 proteins did not significantly alter mouse GCH1 proteins in the hearts of Tg control mice compared with WT control. Administration of STZ resulted in a significant decrease in the expression of GCH1 proteins in WT mice. Interestingly, transgenic overexpression of human GCH1 gene dramatically increased the expression of GCH1 proteins (mouse GCH1 proteins + human GCH1 proteins) not only in Tg control mice but also in Tg STZ mice compared with WT control (P < 0.05, n = 4 mice/group) (Fig. 3E,F). These results suggest that human GCH1 transgene drives GCH1 protein expression in both non-diabetic and diabetic myocardia.

GCH1 overexpression ameliorates diabetes-induced cardiac remodeling and dysfunction.
DCM is characterized by cardiac (both diastolic and later systolic) dysfunction that occurs independently of coronary artery disease or hypertension 43 . Non-invasive transthoracic echocardiography is often used to evaluate cardiac function in patients with DCM 3,4 . We used an echocardiography specific for evaluation of mouse hearts following induction of diabetes 32,44 . As shown in Fig. 4, anterior and posterior wall thickness of the LV at end diastole and end systole, fractional shortening (an indicator of systolic function), isovolumic relaxation time (IVRT), and mitral E/A ratio (an indicator of diastolic function) were comparable among 4 groups: WT control, WT STZ, Tg control, and Tg STZ at baseline (0 week after injection of STZ) (P > 0.05, n = 10 mice/group). STZ-induced diabetes decreased the thickness of the LV wall, fractional shortening, and mitral E/A ratio and increased in IVRT in WT mice 12 weeks after induction of diabetes. These deleterious effects of diabetes were significantly attenuated by GCH1 overexpression (P < 0.05 between Tg STZ and WT STZ groups). These results indicate that GCH1 overexpression reduces diabetes-induced adverse remodeling and dysfunction of the LV.

GCH1 overexpression improves intracellular Ca 2+ signaling impaired by diabetics.
Abnormalities in intracellular Ca 2+ signaling in cardiomyocytes are an important cause of DCM 18 . To study if GCH1 overexpression may reduce diabetes-induced impairments in intracellular Ca 2+ signaling, we first measured intracellular [Ca 2+ ] i in fura-2-loaded cardiomyocytes isolated from Tg and WT mice with or without diabetes 12 weeks after administration of STZ or vehicle (Fig. 5A). There were no significant differences in basal [Ca 2+ ] i among 4 groups (Fig. 5B). Electric stimulation at 0.5 Hz evoked a significant Ca 2+ transient in all cells. Compared with WT controls, changes in [Ca 2+ ] i (∆ [Ca 2+ ] i ) was significantly decreased, and time to half (T50) decay of the Ca 2+ transients was significantly prolonged in WT STZ groups (Fig. 5C,D). There were no significant differences in ∆ [Ca 2+ ] i and T50 decay between Tg controls and WT controls, Tg STZ and WT controls, or Tg STZ and Tg controls (all P > 0.05).
Secondly, we studied the effect of GCH1 overexpression on SR Ca 2+ content in diabetic cardiomyocytes. Caffeine was used to induce SR Ca 2+ release in cardiomyocytes isolated from Tg and WT mice with or without diabetes in the presence of 0 Na + and 0 Ca 2+ Tyrode buffer 17 . Application of 10 mM caffeine to fura-2-loaded myocytes elicited a significant [Ca 2+ ] i transient in all 4 groups of mice (Fig. 5E). The amplitude of the caffeine-induced [Ca 2+ ] i transient was significantly lower in WT SZT mice than WT control mice (P < 0.05, n = 50-55 cells in 5 mice/group) (Fig. 5F). Interestingly, GCH1 overexpression significantly elevated the amplitude of the caffeine-induced [Ca 2+ ] i transient in diabetes compared with WT STZ group (P < 0.05 between Tg STZ and WT STZ groups). There were no significant differences in the caffeine-induced [Ca 2+ ] i transient between Tg STZ and Tg control or WT control (P > 0.05).
Thirdly, we determined the effects of GCH1 overexpression on the expression of SR Ca 2+ handling proteins in diabetic hearts using Western blot analysis 44,45 . Compared with WT controls, GCH1 overexpression increased the ratios of RyR2/total phospholamban (T-PLB), SERCA2a/T-PLB, and phosphorylated phospholamban at

Effects of GCH1 overexpression on BH 4 and dimeric and phosphorylated NOS in diabetics.
Oxidation of BH 4 and dysregulation of NOS in diabetes are implicated in the pathogenesis of DCM 27 . We used high performance liquid chromatography (HPLC) to determine the concentrations of BH 4 and Western blot to analyze the dimers, monomers, and phosphorylation of all 3 isoforms of NOS in the hearts of diabetic and non-diabetic Tg or WT mice. As shown in Fig. 6, diabetes significantly decreased cardiac BH 4 concentrations, the ratio of nNOS dimers/monomers, phosphorylated nNOS (p-nNOS)/nNOS, eNOS dimers/monomers, and phosphorylated eNOS (p-eNOS)/eNOS and increased the expression of iNOS proteins in C57BL/6 WT mice compared with WT controls (P < 0.05 between WT STZ and WT controls, n = 4-5 hearts/group). GCH1 overexpression significantly increased cardiac BH 4 concentrations and the ratios of nNOS dimers/monomers and p-nNOS/nNOS compared with WT controls. Interestingly, there were no significant differences in the ratios of nNOS dimers/monomers and p-nNOS/nNOS between Tg STZ and WT control groups (P > 0.05). Different from the effects of GCH1 overexpression on nNOS, GCH1 overexpression did not significantly alter the ratio of iNOS/GAPDH (glyceraldehyde 3-phosphate dehydrogenase) and p-eNOS/eNOS compared with WT controls. However, increased ratio of iNOS/GAPDH by diabetes was significantly reduced by GCH1 overexpression. Diabetes-induced decrease in the ratios of eNOS dimers/monomers and p-eNOS/eNOS were not significantly changed by GCH1 overexpression (P > 0.05 between Tg STZ and WT STZ groups, n = 4 hearts/group). Coupled NOS generates NO, whereas uncoupled NOS produces O 2 •− 22 . Thus, we measured the levels of myocardial NO x (tissue NO and its metabolite products, nitrate and nitrite) and O 2 •− . The production of NO x and O 2 •− was

GCH1 overexpression decreases phosphorylated p38 MAPK in diabetics. Phosphorylated p38
(p-p38) MAPK is associated with cardiomyopathy 29,30 . We used Western blot to analyze the expression of p-p38 MAPK at tryptophan 180/tyrosine 182, p38 MAPK in Tg and WT mice with or without diabetes (Fig. 7). The ratio of p-p38 MAPK/p38 MAPK was significantly increased in WT STZ group compared with WT control (P < 0.05, n = 4 mice/group). The ratio of p-p38 MAPK/p38 MAPK was comparable between Tg control and WT control groups. Intriguingly, there were no significant differences in the ratio of p-p38 MAPK/p38 MAPK between Tg STZ and Tg control or WT control groups (P > 0.05). To study if GCH1 contributes to a decrease in the ratio of p-p-38/p38 MAPK in diabetic Tg mice, we used DAHP to treat diabetic and nondiabetic Tg mice for 4 weeks. The ratio of p-p38 MAPK/p38 MAPK was not significantly altered by DAHP in nondiabetic GCH1-Tg mice (P > 0.05 between Tg + DAHP and Tg control groups, n = 4 mice/group). However, the ratio of p-p38 MAPK/p38 MAPK was significantly increased in Tg STZ + DAHP group compared with Tg control, Tg + DAHP, or Tg STZ group (P < 0.05).

MG132 preserves GCH1 proteins and improves cardiac function in diabetic WT mice.
To study whether pharmacological approaches may preserve cardiac GCH1 proteins and diminish the severity of DCM, we used MG 132 to treat diabetic C57BL/6 mice for 8 weeks after 4 weeks of diabetes or nondiabetic C57BL/6 mice. The treatment of nondiabetic WT mice with MG 132 did not significantly alter the dimensions and function of the LV and ratio of GCH1/GAPDH (P > 0.05, n = 8-10 mice/group) (Fig. 8). Consistent with above results, diabetes caused decreases in wall thickness of the LV, fractional shortening, mitral E/A ratio, and ratio of GCH1/ GAPDH and increases in LV internal diameters and IVRT 12 weeks after administration of STZ. Interestingly, these detrimental effects of diabetes were blocked by MG 132.

Discussion
The results of the present study demonstrate that GCH1 is necessary for cardiac function in the mouse, however, cardiac GCH1 proteins are degraded in T1DM, leading to negative cardiac remodeling and dysfunction. Intriguingly, either cardiomyocyte-specific overexpression of GCH1 or inhibition of the 26S proteasome with MG 132 preserved cardiac GCH1 proteins and diminished diabetes-induced cardiac remodeling and dysfunction. GCH1-induced cardioprotection against DCM mainly involves the BH 4 /nNOS/SR Ca 2+ handling proteins signaling pathway and depression of p38 MAPK in T1DM.
DAHP is a selective inhibitor of GCH1 and is often used to evaluate the function of GCH1 34,46 . In the present study, the treatment of WT mice with DAHP markedly elevated blood pressure and systolic and diastolic pressure of the LV. A previous study showed that the treatment of mice with DAHP resulted in eNOS uncoupling in vascular tissue due to deficiency of BH 4 34 . Thus, DAHP-elicited increases in blood pressure and left ventricular pressure may be related to vascular endothelial dysfunction, that leads to hypertension 34 . In isolated Langendorff-perfused hearts, DAHP significantly decreased systolic pressure of the LV while the end-diastolic pressure was adjusted to the in vivo levels. Thus, GCH1 is necessary for cardiac function in the mouse.
The expression of cardiac GCH1 proteins was decreased by diabetes but the levels of GCH1 mRNA were not significantly altered. Therefore, diabetes-induced decreases in GCH1 proteins resulted from their degradation rather than a decrease in their biosynthesis. A previous study demonstrated that high glucose elevated the activity of 26S proteasome, leading to the degradation of vascular GCH1 proteins 20 . It is possible that decreased GCH1 proteins result from increased activity of 26S proteasome by diabetes. In the Tg mice, human GCH1 gene was used to increase GCH1 mRNA 32 . Interestingly, the total content of cardiac GCH1 proteins was more in the diabetic Tg mice than WT control mice. These results suggest that GCH1 transgene drives GCH1 protein expression in diabetic myocardium.
The present study kinetically monitored changes in the dimensions and function of the LV in WT and Tg mice following induction of diabetes. Diabetes resulted in significant decreases in the wall thickness and function of the LV and increases in LV volumes 12 weeks after induction of diabetes. GCH1 overexpression did not alter echocardiographic parameters and overall cardiovascular function of the mice, consistent with our previous report 32 . Interestingly, GCH1 overexpression reduced diabetes-induced remodeling and dysfunction of the LV. Moreover, GCH1 overexpression attenuated myocardial interstitial fibrosis (Fig. S9) and apoptotic cardiomyocytes (Fig. S10) induced by diabetes. Collectively, cardiomyocyte GCH1 may serve as a therapeutic target for DCM.
Our results indicated that GCH1 overexpression restored intracellular [Ca 2+ ] i and SR Ca 2+ content that were decreased by diabetes. We also found that GCH1 overexpression increased the expression of RyR2 and SERCA2a proteins in nondiabetics and prevented diabetes-induced decreases in these two SR Ca 2+ handling proteins. Thus, improved intracellular Ca 2+ signaling by GCH1 overexpression in diabetes may be mainly attributed to up-regualation of SR Ca 2+ handling proteins. In addition, GCH1 overexpression increased p-PLB at serine 16 in diabetics and nondiabetics. PLB is an inhibitory protein for SERCA2a, and its phosphorylation at serine 16 increases the affinity of SERCA2a for Ca 2+ 47 . Changes in p-PLB at serine 16 critically regulate the function of SERCA2a 48 . It is possible that increased p-PLB by GCH1 overexpression also contributes to improvements in intracellular Ca 2+ signaling.
We demonstrated that diabetes decreased dimeric and phosphorylated nNOS and eNOS without altering the expression of total nNOS and total eNOS but increased the expression of total iNOS in WT mice. Interestingly, GCH1 overexpression eliminated diabetes-induced decreases in dimeric and phosphorylated nNOS and increases in iNOS but did not alter diabetes-induced reduction in dimeric and phosphorylated eNOS. Previously, we and other investigators have shown that cardiomyocyte-specific overexpression of GCH1 results in elevation of BH 4 in cardiomyocytes rather than endothelial cells 19,33 . The selective effect of GCH1 overexpression on nNOS and iNOS may be related to the fact that nNOS and iNOS are mainly expressed in cardiomyocytes, whereas eNOS is predominantly expressed in vascular endothelial cells 49,50 . Within cardiomyocytes, nNOS is localized in the SR and is physically linked with RyR2, PLB, and SERCA2a 51 . This enables nNOS to exert a direct effect on these proteins by NO from coupled nNOS or O 2 •− from uncoupled nNOS. Caffeine-induced Ca 2+ release is mediated by RyR2 in cardiomyocytes 52 . In the present study, GCH1 overexpression restored decreased RyR2-mediated Ca 2+ release by diabetes, and the beneficial effects of GCH-1 overexpression on RyR2-mediated Ca 2+ release were abrogated by S-methyl-L-thiocitrulline (SMTC, a specific inhibitor of nNOS) (Fig. S8). Therefore, GCH1 overexpression may exert favorable effect on RyR2-mediated Ca 2+ release via nNOS in diabetes.
NO and O 2 •− are the important mediator of myocardial protection and damage, respectively 53,54 . Diabetes elevated myocardial NO and O 2 •− levels. Since nNOs and eNOS levels were decreased in diabetes, increased NO x derived from elevated expression of iNOS. Previous studies have shown that iNOS-derived NO is able to cause myocardial damage 55,56 . This may be due to reaction of NO to high levels of O 2 •− to form peroxynitrite in diabetes 57,58 . GCH1 overexpression decreased the expression of iNOS and O 2 •− levels in diabetes. Thus, GCH1 overexpression is able to suppress oxidative stress in diabetes.
In the present study, diabetes increased p-p38 MAPK, and the negative effect of diabetes on p38 MAPK was suppressed by GCH1 overexpression. Furthermore, the effect of GCH1 overexpression on p38 MAPK was inhibited by DAHP. These results suggest that GCH1 overexpression can exert an inhitory effects on activation of p38 MAPK in diabetes. A growing body of evidence indicates that p38 MAPK regulates the expression and/or function of RyR2 and SERCA2a through direct or indirect (PLB) mechanisms 29,[59][60][61][62] . We speculate that increased expression of RyR2 and SERCA2a proteins in diabetic Tg mice may be related to inhibiton of p-38 MAPK by GCH1.
We showed that MG 132 diminished the degradation of cardiac GCH1 proteins and reduced cardiac dysfunction in diabetes. MG 132 is a potent inhibitor for the 26S proteasome that is responsible for the degradation of GCH1 proteins 20,63 . Our present results suggest that inhibition of the 26S proteasome is a useful method to prevent the development of DCM. However, MG 132 may inhibit degradation of all classes of 26S proteasomal substrate rather than specific GCH1 proteins 64 . Recent progress on the study of proteasome reveals that the 26S proteasome can selectively recognize specific proteins for degradation through its 19S regulatory particle 65 . This property of the 26S proteasome suggests that there is a possibility to develop 26S proteasome inhibitors with specificity towards targeting proteins 66 . It is reasonably believed that developing novel 26S proteasome inhibitors specific for GCH1 proteins may be useful in the clinical treatment of DCM.
In summary, the present study indicates that degradation of cardiac GCH1 proteins contributes to the pathogenesis of DCM. Either cardiomyocyte-specific overexpression of GCH1 or inhibiton of the 26S proteasome with MG 132 protects the heart against DCM by elevating cardiac GCH1 proteins. The beneficial effects of GCH1 on diabetic hearts are associated with an improvement in intracellular Ca 2+ signaling as a result of increases in BH 4 bioavailability, dimeric and phosphorylated nNOS, and SR Ca 2+ handling proteins and decreases in p-p38 MAPK and O 2 •− production (Fig. S11). Our present study suggests that cardiomyocyte GCH1 is a potent therapeutic target for DCM, and developing novel 26S proteasome inhibitors with specificity towards cardiac GCH1 may be useful for the clinic treatment of DCM.

Methods
Animals. The Tg mice with cardiomyocyte-specific overexpression of human GCH-1 gene on a C57BL/6 background were generated, as described previously 32 . The Tg mice were identified by the presence of human GCH-1 gene using polymerase chain reaction (PCR) on tail-derived genomic DNA 32 . C57BL/6 WT littermates were used as control for the Tg mice. All protocols (Figs S1 and S2) were approved by the Animal Care and Use Committee of the Medical College of Wisconsin (Milwaukee, WI) and conformed to the Guide for the Care and Use of Laboratory Animals (Institute for Laboratory Animal Research, National Academy of Sciences, 8th edition, 2011). DAHP treatment of mouse. WT and Tg mice were orally given 50 mg/kg/day DAHP twice daily for 4 weeks (WT + DAHP and Tg + DAHP groups) using a plastic feeding tube or 1:1 dimethyl sulfoxide/NaCl (vehicle) as control (n = 10 mice/group) 34,46 . Determination of GCH1 activity. GCH1 enzyme activity was assessed as the conversion of GTP into 7,8-dihydroneopterin triphosphate, as described 19 .
Induction of diabetes. T1DM was induced in Tg and WT mice at 6-8 weeks of age by daily intraperitoneal injection of 50 mg/kg/day STZ freshly prepared in citrate buffer (pH 4.5) for 5 consecutive days (WT STZ and Tg STZ groups) 20 . Control animals were given equivalent amounts of citrate buffer (WT control and Tg control groups).
Administration of MG 132. Diabetic and nondiabetic C57BL/6 mice were injected intraperitoneally 10 μ g/ kg/day MG-132 for 8 weeks after 4 weeks of diabetes or equal amounts of vehicle as control 63 .

Measurement of blood pressure, LV hemodynamics, and blood glucose. A Millar Mikro-Tip
Pressure Transducer Catheter (Millar Instruments, Inc.) was inserted into the right carotid artery to monitor blood pressure and subsequently placed in the middle of the LV chamber to measure left ventricular systolic and diastolic pressure 35 . Fasting blood glucose of diabetic and nondiabetic mice was measured with a blood gas analyzer (ABL-725 Radiometer) 45 . Langendorff-perfused mouse hearts. Mouse hearts were mounted on a Langendorff apparatus and perfused retrogradely through the aorta at a constant pressure of 80 mmHg with Krebs-Henseleit buffer, as described 38,44 . The LV pressure signal, heart rate, and coronary flow rate were monitored, and + dP/dt (maximum rate of increase of left ventricular developed pressure) and − dP/dt (maximum rate of decrease of left ventricular developed pressure) were determined. PCR analysis. C57BL/6 mouse hearts were excised 0, 2, 4, 8, 12 weeks after administration of STZ, and the LV was homogenized at 4 °C for PCR analysis of mouse GCH-1 mRNA 33,44 . Transthoracic echocardiography. Echocardiography was performed with a VisualSonics Vevo 770 High-resolution Imaging System equipped with a 30 MHz transducer (Scanhead RMV 707), as described 39,45 . Left ventricular dimensions and ejection fraction were measured by two-dimension guided M-mode method. Pulsed Doppler waveforms recorded in the apical-4-chamber view were used for the measurements of the peak velocities of mitral E (early mitral inflow) and A (late mitral inflow) waves. Assay of biopterins. BH 4 and 7,8-dihydroneopterin were quantified in LV tissue homogenates by HPLC with electrochemical detection (ESA Biosciences CoulArray ® system Model 542), as described 68,69 . Authentic BH 4 and 7,8-dihydroneopterin solutions (10-100 nM) were used as standards and sample concentrations were normalized to protein content measured by the bicinchoninic acid protein assay.

Measurement of NO x and O 2
•− . Tissue NO and its metabolite products (nitrate and nitrite) in the supernatant, collectively known as NO x , were assayed using a NO chemiluminescence analyzer (Siever 280i NO Analyzer) 32,70 . Lucigenin, a compound that emits light upon interaction with O 2 •− , was used to quantify the O 2 •− production from myocardium, as described 71 . The data were presented in relative light units (RLUs) per mg protein.
Relative O 2 •− levels were expressed as percentages compared to WT controls.
Immunoblotting. The effects of diabetes and GCH-1 overexpression on the expression of GCH-1 proteins, NOS and phosphorylated NOS proteins, SR Ca 2+ handling proteins, and p38 MAPK were examined using standard Western blot techniques 44,70 . The normal function of nNOS and eNOS to produce NO requires dimerization of the enzyme 24,72 . To investigate nNOS and eNOS homodimer formation in the myocardium, non-boiled cellular lysate was resolved by 6% SDS-PAGE at 4 °C overnight 45,69 . The density of the Immunoreactive bands was analyzed with Image J (image acquisition and analysis software, NIH). In the Fig. 3, total GCH1 was the sum of human GCH1 band density and mouse GCH1 band density.
Statistical analysis. All data are expressed as mean ± S.E.M. Statistical analysis was performed with one-way ANOVA followed by Bonferroni post-hoc test for multiple comparisons of multiple group means or with Student's t test for comparisons between two group means. Repeated-measures ANOVA was used to compare the differences in body weight, blood glucose, heart rate, LV wall thickness, LV volumes, ejection fraction, mitral E/A ratio, GCH1 mRNA, GCH1 proteins, and ± dP/dt values at different time points. A value of P< 0.05 was considered as statistically significant.