BH4 Increases nNOS Activity and Preserves Left Ventricular Function in Diabetes

Abstract

Rationale

In diabetic patients, heart failure with predominant left ventricular (LV) diastolic dysfunction is a common complication for which there is no effective treatment. Oxidation of the nitric oxide synthase (NOS) co-factor tetrahydrobiopterin (BH4) and dysfunctional NOS activity have been implicated in the pathogenesis of the diabetic vascular and cardiomyopathic phenotype.

Objective

Using mice models and human myocardial samples, we evaluated whether and by which mechanism increasing myocardial BH4 availability prevented or reversed LV dysfunction induced by diabetes.

Methods And Results

In contrast to the vascular endothelium, BH4 levels, superoxide production and NOS activity (by liquid chromatography) did not differ in the LV myocardium of diabetic mice or in atrial tissue from diabetic patients. Nevertheless, the impairment in both cardiomyocyte relaxation and [Ca²⁺]_i decay and in vivo LV function (echocardiography and tissue Doppler) that developed in wild type mice (WT) 12 weeks post-DM induction (streptozotocin, 42-45mg/kg) was prevented in mice with elevated myocardial BH4 content secondary to overexpression of GTP-cyclohydrolase 1 (mGCH1-Tg) and reversed in WT mice receiving oral BH4 supplementation from the 12^th to the 18^th week after DM induction. The protective effect of BH4 was abolished by CRISPR/Cas9-mediated knockout of neuronal NOS (nNOS) in mGCH1-Tg. In HEK cells, S-nitrosoglutathione led to a PKG-dependent increase in plasmalemmal density of the insulin-independent glucose transporter, GLUT-1. In cardiomyocytes, mGCH1 overexpression induced a NO/sGC/PKG-dependent increase in glucose uptake via GLUT-1, which was instrumental in preserving mitochondrial creatine kinase activity, oxygen consumption rate, LV energetics (by ³¹P MRS) and myocardial function.

Conclusions

We uncovered a novel mechanism whereby myocardial BH4 prevents and reverses LV diastolic and systolic dysfunction associated with diabetes via a nNOS-mediated increase in non-insulin dependent myocardial glucose uptake and utilization. These findings highlight the potential of GCH1/BH4-based therapeutics in human diabetic cardiomyopathy.

Nonstandard Abbreviations and Acronyms

BH4

Tetrahydrobiopterin

CK

Creatine Kinase

GCH1

GTP Cyclohydrolase 1

L-NAME

N-nitro-L-arginine methyl ester

³¹P MRS

³¹Phosphorous Magnetic Resonance Spectroscopy

nNOS

Neuronal Nitric Oxide Synthase

OCR

Oxygen Consumption Rate

PDK4

Pyruvate Dehydrogenase Kinase 4

PKG

Protein Kinase G

PPAR

Peroxisome Proliferator-Activated Receptor

SR

Sarcoplasmic Reticulum

UCP

Uncoupling protein

VDAC

Voltage-dependent anion channel

Introduction

Diabetes mellitus (DM) is a major cause of death and disability and a large economic burden on healthcare systems across the world ¹. Globally, 1 in 12 deaths in adults has been attributed to DM and its complications ²; amongst which, the proportion of heart failure (HF) cases is substantial both in type I and type II DM and persisting after adjustment for differences in coronary artery disease or other relevant risk factors ^{3, 4}. Together with post-mortem findings demonstrating left ventricular (LV) dysfunction in diabetic patients in the absence of coronary artery disease or hypertension ⁵, epidemiological data suggest that DM may in itself give rise to a specific cardiomyopathy characterized by predominant LV diastolic dysfunction, leading to heart failure with preserved or mildly impaired LV ejection fraction (LVEF) ⁶.

A number of factors, including mitochondrial dysfunction, oxidative stress, impaired calcium handling, dysfunctional NOS activity and remodeling of the extracellular matrix have been advocated in the pathogenesis of diabetic cardiomyopathy ⁶, however, a unifying mechanism upstream of the observed LV functional changes is still lacking.

Constitutive NO production regulates LV compliance and relaxation through its action on myofilament Ca²⁺ sensitivity and intracellular Ca²⁺ handling ⁷. Under physiological conditions, tetrahydrobiopterin (BH4) is a limiting factor in myocardial NO synthesis by the “neuronal” NOS isoform (nNOS) ⁸. Increasing cardiomyocyte BH4 content by myocardial-specific overexpression of the first enzyme involved in its synthesis, GTP cyclohydrolase 1 (GCH1), enhances nNOS activity and hastens the rate of intracellular Ca²⁺ reuptake and myocardial relaxation in healthy mice by increasing the phospholamban (PLB) phosphorylated fraction ^{8, 9}. In the presence of DM, increasing BH4 content by myocardial GCH1 overexpression or inhibition of GCH1 protein degradation has been shown to attenuate the increase in myocardial superoxide production and maintain nNOS in its dimeric form ¹⁰. Impaired NO signaling, due to BH4 oxidation and dysfunctional eNOS activity accounts for the endothelial dysfunction reported in diabetic patients and animal models ^11–13. Similar changes in the myocardium would be expected to lead to LV diastolic dysfunction and increased oxidative stress ^14–16; however, whether dysfunctional NOS activity and altered nitroso-redox balance are key factors in the pathogenesis of diabetic cardiomyopathy remains to be established. Likewise, the extent to which endothelial dysfunction induced by DM contributes to the cardiomyopathic phenotype is unclear.

Here we show that increasing myocardial BH4 and nNOS activity by transgenic overexpression of GCH1 does not preserve endothelial-mediated vasodilatation but prevents LV dysfunction in diabetic mice—not by averting NOS dysfunction, maintaining PLB phosphorylation, or reducing oxidative stress—but by preserving myocardial energetics via a nNOS-mediated increase in glucose uptake through the insulin-independent transporter, GLUT-1. Importantly, oral BH4 supplementation is able to reverse the cardiomyopathic phenotype in diabetic wild type (WT) mice, indicating that GCH1/BH4-based therapeutics may be used to treat as well as prevent diabetic cardiomyopathy.

Methods

Human samples

Samples of the right atrial appendage were collected from patients undergoing on-pump cardiac surgery for coronary revascularization and stored at -80°C. Investigations were approved by the Research Ethics Committee; all patients gave informed written consent.

Diabetes induction

DM was induced by low doses (42-45mg/kg) of STZ dissolved in citrate buffer and injected intra-peritoneal daily for 5 consecutive days. Control mice were injected in parallel with buffer only. Mice with glucose levels < 15mM after 2 weeks of STZ injection were excluded. Studies were carried out and data analyzed with the operator blind of the genotype or treatment allocation.

BH4 Supplementation

WT mice were allocated to normal chow (Teklad global 16% protein diet, Harlan Laboratories) or BH4-supplemented chow (200 mg/kg/day for 6 weeks) beginning at week 12 post-STZ. Data were collected and analyzed with the operator blind of treatment allocation. Randomization was performed by cage.

Representative images were selected as reflecting either the mean or the median (in case of non-normal data) of their respective data series.

Results

DM-induced LV dysfunction is prevented by myocardial GCH1 overexpression and reversed by oral BH4 supplementation

Streptozotocin (STZ) decreased plasma insulin levels and body weight, and increased plasma glucose similarly in wild type (WT) and mGCH1-Tg at 4 and 12 weeks post-DM induction (Online Table I). As expected, DM caused significant endothelial dysfunction at 4 and 12 weeks post-induction (Online Fig. IA; Fig. 1). At the latter time point, aortas from both diabetic WT and mGCH1-Tg displayed an enhanced contractile response to phenylephrine and impaired vasodilatation in response to acetylcholine or the peptide activator of the protease-activated receptor-2, SLIGRL, compared with sham-injected non-diabetic littermates (Fig. 1B-D), whereas endothelial-independent vasodilatation in response to the NO donor, sodium nitroprusside (SNP), was preserved in all groups (Fig. 1E). Pre-incubation of aortic rings with the NOS inhibitor, L-NAME (100 μmol/L), abolished all differences between diabetic and non-diabetic mice (Fig. 1F, G).

Figure 1. Endothelial function is impaired in the presence of DM in both WT and mGCH1-Tg mice (Tg).

A. Overview of the assessment of vascular function in diabetic (DM) and normoglycemic mice. B. Dose-dependent vasoconstrictor response to phenylephrine (PE) and C. vasodilator response to acetylcholine (ACh) in aortic rings from DM or normoglycemic WT and Tg (n; WT=8, WT DM=7, Tg =8, Tg DM= 9 aortas) at 12 weeks after DM induction or sham-injection. D. Endothelial-dependent (protease-activated receptor-2 activating peptide, SLIGRL) and (E. endothelial-independent (sodium nitroprusside, SNP) vasodilator responses in aortic rings from diabetic or normoglycemic WT and Tg (n= 6 aortas per group). F. PE and (G. ACh-mediated responses in the presence of NOS inhibition (L-NAME, 100μM; n, WT=5, WT DM=6,Tg=7, Tg DM=7 aortas) Normally distributed data (B, D, E, F and G) are shown as means ± SEM and were compared using 2-way ANOVA with Bonferroni’s correction. In panel C, non-normal data (D’Agostino-Pearson test, P=0.032) are show as median and IQR and were compared using Kruskal–Wallis one-way ANOVA, followed by the Dunn test.

*P < 0.05, **P < 0.01 vs. normoglycemic mice from either genotype.

LV function was assessed in vivo by echocardiography and tissue Doppler in all groups at 4 and 12 weeks after the STZ or sham-injections (Fig 2, Online Fig. V and Online Table II).

Figure 2. Myocardial GCH1 overexpression and oral BH4 supplementation prevent and reverse the impairment in LV function after DM induction, respectively, via a nNOS-dependent mechanism.

A. Representative LV Tissue Doppler and B. M-Mode traces in WT, Tg and in Tg mice lacking nNOS (Tg KO) at 4 and 12 weeks post-DM induction or sham-injection (n=12 mice per group). C. Average E/E’ ratio and D. average LV ejection fraction (LVEF) in the 6 groups over time.

Data (all normally distributed) are shown as means ± SEM and were compared using 3-way ANOVA with Bonferroni’s correction. For the E/E’ ratio: ***p<0.001, WT DM vs WT at 12 weeks and ****p<0.0001, Tg KO DM vs Tg KO at 12 weeks. P = 0.0004 for the interaction between genotype and diabetes. For LVEF: *p<0.05, WT DM vs WT at 12 weeks and **p<0.01, Tg KO DM vs Tg KO at 12 weeks. P = 0.0085 for the interaction between genotype and diabetes. Comparisons between diabetes-time and genotype-time are included in Online Table II. Scatterplots for these data are shown in Online Figure V. E. Representative LV Tissue Doppler and F. M-Mode traces in DM WT mice fed standard or BH4-supplemented chow from week 12 to week 18 post-DM induction and non-DM controls (n=12 mice per group). G. Average E/E’ ratio and H. LVEF in the 3 groups over time. Data (all normally distributed) are shown as means ± SEM and were analysed by 2-way ANOVA with Bonferroni’s correction.

***p<0.001 WT vs WT DM, *p<0.05 WT DM BH4 vs WT DM. Comparisons between week 12 and 18 are included in Online Table III. Scatterplots for these data are shown in Online Figure VI.

At 4 weeks post-DM induction, LV diastolic and systolic function was preserved (Fig. 2A-D and Online Table II) despite the development of endothelial dysfunction (Online Fig. IA).

At 12 weeks post-DM induction, non-diabetic groups showed similar function as at 4 weeks. However, LV diastolic function was significantly impaired in diabetic WT, as indicated by a higher ratio between the peak early mitral filling velocity (E) and the tissue Doppler-derived peak early diastolic velocity at the mitral annulus (E’) (P<0.001 vs WT non-DM controls and diabetic mGCH1-Tg mice, Fig. 2C). Echocardiographic examination showed no differences in LV end-diastolic volume (EDV) between genotypes in the presence or absence of DM whereas LV end-systolic volume (ESV) increased in diabetic WT (Online Table II), leading to a significant reduction in LVEF and fractional shortening when compared with both WT non diabetic controls and diabetic mGCH1-Tg mice (Fig. 2D and Online Table II). By contrast, LV diastolic and systolic function were unaltered in mGCH1-Tg mice at 12 weeks post-DM induction (Fig. 2A-D and Online Table II). These findings were mirrored by concordant changes in the myocardial performance index (MPI, Online Table II), confirming that impairment in this heart rate/arterial pressure-independent measurement of overall LV function was prevented in diabetic mGCH1-Tg.

LV mass was not different between groups at all time-points (Online Table II). Heart rate was lower in mGCH1-Tg compared with their WT littermates and did not change significantly after DM induction in either genotype (Online Table II).

We have previously shown that GCH1 overexpression and raised BH4 content significantly augment myocardial nNOS activity ⁸. However, BH4 is also an antioxidant molecule and a co-factor for the formation of biogenic amines and serotonin ¹⁷. To evaluate to which extent nNOS-derived NO was responsible for preserving LV function in the presence of DM, we performed these experiments in mGCH1-Tg in which nNOS was knocked out (Online Fig. IIIG) using CRISPR-Cas9-mediated gene editing. As shown in Figure 2A-D and Online Table II, the protective effect of myocardial GCH1 overexpression was lost in diabetic mice lacking nNOS, consistent with an essential role of this NOS isoform in mediating the cardioprotective effects of BH4 in the presence of DM.

We then tested whether LV function in diabetic WT mice could be restored by supplementing their diet with BH4 (200 mg/kg/day for 6 weeks, beginning at week 12 post-DM induction).

At 12 weeks and before BH4 was introduced in the protocol, both groups of WT DM mice showed reduced LV diastolic and systolic function (Figure 2E-H, Online Fig. VI and Online Table III)

After 18 weeks, DM was associated with a significant impairment in LV diastolic and systolic function in WT fed normal chow, as indicated by a lower LVEF and a higher E/E’ ratio and MPI (vs. non-diabetic WT); by contrast, LV diastolic and systolic dysfunction was completely reversed in diabetic WT receiving BH4-supplementation (Fig. 2E-H, Online Table III). BH4 supplementation increased cardiac BH4 levels (9.4 ± 1.3 vs 5.8 ± 0.8 pmol/mg protein in WT group. P<0.01, n= 6-8 per group), and NOS activity (% citrulline conversion: 0.6 ± 0.11 vs 0.2 ± 0.05. P<0.01, n= 6-8 per group).

Cardiomyocyte and [Ca²⁺]_i handling dysfunction induced by DM is prevented by mGCH1 overexpression and reversed by oral BH4 supplementation

To establish whether the changes in LV function observed in WT diabetic mice in vivo reflect altered cardiomyocyte dynamics and Ca²⁺ handling, we undertook these measurements in field-stimulated LV myocytes (3Hz, 35±0.5°C) (Figure 3A-F).

Figure 3. LV myocyte dysfunction in diabetic mice is prevented by GCH1 overexpression.

A. Representative [Ca²⁺]_i transients at 4 and 12 weeks post-DM induction or sham-injection in WT and mGCH1-Tg (Tg). B. Average amplitude of the [Ca²⁺]_i transient (Fura 2 ratio) and C. time constant of [Ca²⁺]_i decay (Tau1) in the 4 groups. D. Representative unloaded cell shortening and re-lengthening traces at 4 and 12 weeks post-DM induction or sham-injection in WT and mGCH1-Tg (Tg). E. Average % cell shortening and F. relaxation velocity in the 4 groups. Normally distributed data are shown as means ± SEM in B, E and F and were compared using 2-way ANOVA with Bonferroni’s correction.

Non-normally distributed data in panel C (D’Agostino-Pearson test, P=0.028) are shown as median and IQR and were compared using Kruskal-Wallis one-way ANOVA, followed by the Dunn test.

*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. (n), number of cells.

At 4 weeks after DM induction, cardiomyocyte relaxation and [Ca²⁺]_i transient characteristics were unaltered in both genotypes (Fig. 3A-F). As reported previously⁸, there were significant genotype differences in the rate of myocyte relaxation and [Ca²⁺]_i reuptake in non-diabetic mice, which were preserved in the presence of DM. By contrast, at 12 weeks post-DM, both relaxation velocity and the rate of decay of the [Ca²⁺]_i transient were slower in diabetic WT myocytes compared to non-diabetic controls, whereas myocardial overexpression of GCH1 prevented the adverse effect of DM on both parameters (Fig. 3 C, F). Both at 4 and 12 weeks post-DM induction, cell shortening was significantly higher in Tg compared with WT, but the amplitude of the [Ca²⁺]_i transient did not differ between groups (Fig. 3 B, E).

Oral BH4 supplementation for 6 weeks beginning at week 12 post-STZ injection reversed the effect of DM on both relaxation velocity and rate of [Ca²⁺]_i decay in isolated LV myocytes but did not affect cell shortening or the amplitude of the [Ca²⁺]_i transient significantly (Fig. 4A-E).

Figure 4. LV myocyte dysfunction in diabetic mice is reversed by oral BH4 supplementation.

A. Representative [Ca²⁺]_i transient and cell shortening traces from normoglycemic WT at 18 weeks post sham injection and DM WT fed standard or BH4-supplemented chow from 12 to18 weeks post-DM induction. Average data for B. relaxation velocity, C. time constant of [Ca²⁺]_i decay (Tau1), D. cell shortening and E. amplitude of the [Ca²⁺]_i transient. Normally distributed data are shown as means ± SEM in D and E, and were compared using 1-way ANOVA with Bonferroni’s correction. Non-normally distributed data in B. and C. (D’Agostino-Pearson test, P = 0.032 and 0.049, respectively,) are shown as median and IQR and were compared using Kruskal–Wallis one-way ANOVA, followed by the Dunn test. *P < 0.05; **P < 0.01; ****P < 0.0001.

(n), number of cells.

As reported previously⁸, phospholamban (PLB) protein content was lower in the LV myocardium of mGCH1-Tg, which also showed a significantly higher PLB Ser¹⁶ phosphorylated fraction compared to WT (Fig. 5A). There were no significant differences in the PLB Thr¹⁷ phosphorylated fraction or in SERCA2A between genotypes (Fig. 5A-B).

Figure 5. Diabetes does not affect SERCA and PLB protein level but increases LV hydroxyproline content in both genotypes.

A. and B. show representative immunoblots and quantification of total and phosphorylated Serine¹⁶ PLB, SERCA2A and GAPDH and total and phosphorylated Threonine¹⁷ PLB.

C. LV Hydroxyproline content and D. LV collagen deposition quantified by polarized light microscopy in Tg and WT at 12 weeks after DM induction or sham-injection. Normally distributed data in A, C and D are shown as means ± SEM and were compared using 2-way ANOVA with Bonferroni’s correction. Non-normally distributed data in B (D’Agostino-Pearson test, P = 0.0007) are shown as median and IQR and were compared using Kruskal–Wallis one-way ANOVA, followed by the Dunn’ test.

*P < 0.05, **P < 0.01, ***P < 0.001.

(n), number of hearts.

DM did not affect the LV content or phosphorylation status of any of these proteins in either genotype but significantly increased the LV content of hydroxyproline in both WT and mGCH1-Tg and led to a comparable non-significant increase in collagen staining in both genotypes (Fig. 5C-D).

LV dysfunction in DM is independent of BH4 oxidation and dysfunctional NOS activity

In the aortic endothelium, BH4 and superoxide production in sham-injected mGCH1-Tg did not differ significantly from WT. DM induction lowered the ratio between BH4 and its oxidized products to a similar extent in both genotypes and increased superoxide production (Fig. 6A-B). By contrast, in the LV myocardium of non-diabetic GCH1-Tg, BH4 content and NOS activity were significantly higher compared with WT (Fig. 6C-D), in the absence of differences in the protein content of NOS isoforms between genotypes (Online Fig. IB). Surprisingly, none of these parameters was altered at 12 weeks after DM induction in either genotype. Total and reduced myocardial glutathione was significantly elevated in mGCH1-Tg hearts compared to WT but, again, both measurements were unaltered by DM (Fig. 6E). In line with these findings, neither total nor NOS-derived (i.e., L-NAME-inhibitable) myocardial superoxide production was raised in diabetic mice from either genotype (Fig. 6F, G).

Figure 6. In contrast to the aortic endothelium, LV superoxide production, BH4 oxidation and NOS activity are unchanged at 12 weeks after DM induction.

A. Aortic BH4 level and the ratio between BH4 and its oxidized products (n=15-20 per group) and B. superoxide production detected by lucigenin-enhanced chemiluminescence (LEC; n=12-21 per group) in WT and Tg mice at 12 weeks post-DM induction or sham injection. C. Left ventricular (LV) BH4 level and the ratio between BH4 and its oxidized products in the LV myocardium of WT and Tg (n=9-15 hearts per group) at 12 weeks post-DM induction or sham injection. D. LV NOS activity (n=7 hearts per group) and E. myocardial levels of reduced and oxidised glutathione in WT and Tg (n=10 hearts per group) at 12 weeks post-DM induction or sham injection. F. LV NOS-derived O₂^- production by lucigenin-enhanced chemiluminescence (n=11-13 hearts per group, left panel). The inactive isomer D-NAME was used as a control; G. LV superoxide levels were also determined by HPLC (n=9-11 hearts per group). Normally distributed data in A, B and G are shown as means ± SEM and were compared using 2-way ANOVA with Bonferroni’s correction.

Non-normally distributed data in D, E and F (D’Agostino-Pearson test, P = 0.011, 0.002 and 0.005, respectively) were compared using Kruskal–Wallis one-way ANOVA, followed by the Dunn test.

*P < 0.05, **P < 0.01, ***P < 0.001.

(n) number of aortas or hearts.

As already mentioned, oral BH4 supplementation led to a significant increase in BH4 content and total NOS activity in the myocardium of diabetic WT but, even at 18 weeks post-DM induction, myocardial BH4 level (5.6 ± 0.6 vs 5.8 ± 0.8 pmol/mg protein in WT group) and NOS activity were unchanged in WT DM mice (% citrulline conversion, 0.3 ± 0.08 vs 0.2 ± 0.05 in WT group), suggesting that, in contrast to the vascular endothelium, myocardial BH4 oxidation and NOS dysfunction are not an early hallmark of DM nor are they required to induce the cardiomyopathic phenotype.

To establish whether these unexpected findings were also pertinent to the myocardium of diabetic patients, we measured myocardial biopterins in samples of the right atrial appendage from 17 diabetic patients and 19 matched non-diabetic controls undergoing coronary revascularization (Online Table IV). LV ejection fraction was significantly lower in diabetic patients compared with their matched non-diabetic controls; nevertheless myocardial BH4 content and the ratio between BH4 and its oxidized products were similar between groups and so was total and NOS-derived superoxide production (Online Figure II), indicating that, in agreement with our findings in diabetic mice, oxidant stress is not increased and NOS activity is not uncoupled in the myocardium of diabetic patients.

mGCH1 overexpression preserves myocardial energetics in DM by increasing myocardial glucose uptake and utilization via a NO/sGC/PKG-dependent mechanism

In the presence of DM, myocardial glucose transport and glycolysis are compromised and fatty acids (FA) become the exclusive source of ATP generation leading to an increase in oxygen consumption and a reduction in cardiac efficiency ¹⁸. Accordingly, the myocardial expression of PPARα, which promotes FA uptake and utilization, was significantly elevated in diabetic WT, but not in mGCH1-Tg, compared with sham-injected controls (Fig. 7A). Similarly, the LV content of the mitochondrial uncoupling protein-3 (UCP3), a downstream target of PPARα ¹⁹, was 50% higher in diabetic WT at 4 weeks post–DM induction (Online Fig. IIIA) in keeping with an early switch to FA metabolism. However, the increase in myocardial UCP3 was much greater in WT by 12 weeks post-DM induction (Fig. 7B) as was that of pyruvate dehydrogenase kinase 4 (PDK4; Online Fig. IIIB), implying that myocardial metabolism was increasingly compromised as the duration of DM increased.

Figure 7. mGCH1 overexpression prevents the decline in myocardial energetics in diabetic mice.

A. LV gene expression of the peroxisome proliferator-activated receptor a (PPARα), B. protein level of mitochondrial uncoupling protein-3 (UCP3) and C. activity of the mitochondrial (Mito) and D. cytoplasmic creatine kinase (CK) in WT and Tg mice at 12 weeks after DM induction or sham-injection. E. LV protein level of the mitochondrial voltage-dependent anion channel (VDAC) and F. citrate synthase (CS) activity in the 4 groups. G. Representative LV traces from ³¹P-magnetic resonance spectroscopy in isolated perfused hearts from diabetic WT and Tg mice, (left panel) and the average PCr/ATP ratio (right panel) in WT and Tg hearts at 12 weeks after DM induction or sham-injection.

Data (all normally distributed) are shown as means ± SEM and were analysed using 2-way ANOVA, with Bonferroni’s correction. *P < 0.05, **P < 0.01, ***P < 0.001.

(n), number of hearts isolations in A-C and myocytes in D-G.

Diabetic WT hearts also displayed a reduction in the activity of mitochondrial creatine kinase (CK) (Fig. 7C), indicating reduced capacity for high-energy phosphate shuttling out of the mitochondria. There was no difference in cytoplasmic CK activity (Fig. 7D) or in markers of mitochondrial cell density, as evaluated by citrate synthase activity and VDAC expression (Fig. 7E, F). Despite similar levels of plasma glucose and insulin in both genotypes (Online Table I), PPARα expression, UCP3 protein level and mitochondrial CK activity were unaltered in the heart of diabetic mGCH1-Tg (Fig. 7A-C).

To evaluate whether myocardial energetics differed between genotypes, we performed ³¹P magnetic resonance spectroscopy in isolated perfused hearts. In sham-injected mice, there was no difference in the LV phosphocreatine-to-ATP ratio (PCr/ATP) between genotypes (Fig. 7G). As observed in diabetic patients ^{20, 21}, the PCr/ATP ratio was significantly reduced in the LV myocardium of diabetic WT mice but was unchanged in diabetic mGCH1-Tg (Fig. 7G), suggesting that myocardial glucose uptake or utilization may be less affected by DM in mGCH-Tg. Indeed, myocardial 2-deoxyglucose uptake was significantly higher in LV myocytes from diabetic mGCH1-Tg compared with diabetic WT (Fig 8A). Selective inhibition of the insulin-independent glucose transporter, GLUT-1, with STF-31 (10 μmol/L) significantly reduced 2-deoxyglucose uptake in diabetic mGCH1-Tg and abolished differences between genotypes (Fig 8A).

Figure 8. mGCH1 overexpression preserves myocardial function at 12 weeks post-DM induction by increasing GLUT-1-dependent myocardial glucose uptake and utilization.

A. ³H-Deoxy-glucose uptake in isolated LV myocytes from DM WT or Tg under control conditions and in the presence of NOS, GLUT-1, sGC or PKG inhibition with L-NAME, STF-31, ODQ or Rp-8-pCPT-PET cGMP, respectively. B. Oxygen consumption rate (OCR) in isolated LV myocytes from DM or sham-injected mice in the presence of 5 mmol/L glucose or C. in isolated LV myocytes from normoglycemic mice in the presence of GLUT-1 inhibitor, STF-31 (10 μmol/L). Average data for D. relaxation velocity, E. time constant of [Ca²⁺]_i decay (Tau1), F. cell shortening and G. the amplitude of the [Ca²⁺]_i transient in LV myocytes from diabetic WT and Tg in the presence or absence of GLUT-1 inhibition (STF-31,10 μmol/L) in the 4 groups. Data (all normally distributed) are shown as means ± SEM and were analysed using 1-way ANOVA in A, and a 2-way ANOVA with Bonferroni’s correction in B-G.

*P < 0.05, **P < 0.01, ***P < 0.001, ****P<0.0001.

(n), number of hearts.

As shown in Online Fig. III, GLUT-1 expression and protein content were significantly higher in the myocardium of mGCH1-Tg mice whereas the insulin-dependent glucose transporter, GLUT-4, did not differ between genotypes. To evaluate the relative impact of myocardial GLUT-1 content versus dynamic GLUT-1 regulation by NO, sGC and PKG in GCH-Tg myocytes, we compared 2-deoxyglucose uptake in the presence of inhibitors of NOS (L-NAME, 1 mmol/L), sGC (ODQ (10 μmol/L), or PKG (Rp-8-pCPT-PET-cGMPS (10 μmol/L). All agents significantly inhibited glucose transport only in mGCH1-Tg and abolished the differences between genotypes (Fig 8A).

We determined whether higher myocardial glucose transport in mGCH1-Tg mice was accompanied by increased glucose oxidation by measuring OCR in intact isolated LV myocytes from diabetic and sham-injected mice from both genotypes (Fig. 8B). OCR was significantly higher in LV myocytes from mGCH1-Tg, irrespective of diabetic status. Pre-incubation of LV myocytes with the GLUT-1 inhibitor, STF-31 (10 μmol/L), significantly decreased OCR both in WT and mGCH1-Tg myocytes and abolished the difference between genotypes (Fig. 8C). OCR was also measured after blocking NOS (L-NAME, 1 mmol/L), or PKG activity (Rp-8-pCPT-PET-cGMPS, 10 μmol/L) and, again, under these conditions, OCR no longer differed between genotypes (Online Fig. IV), indicating that mGCH1 overexpression dynamically increases myocardial glucose uptake via the GLUT-1 transporter in a NO/PKG-dependent manner. In agreement with these findings, incubation with the NO donor S-nitrosoglutathione (GSNO, 1mmol/L) was associated with a 2-fold increase in the density of GLUT-1 on the cell surface of HEK-293 cells when compared to control cells. Inhibition of PKG significantly reduced GLUT-1 mobilization under these conditions (Online Fig. IV).

The functional impact of genotype differences in glucose uptake via GLUT1 was evaluated in LV myocytes from diabetic WT and mGCH1-Tg. As shown in Figure 8 D-G, GLUT-1 inhibition with STF-31 had no effect in myocytes from diabetic WT mice but abolished the advantage conferred by mGCH1 overexpression in the presence of DM by slowing the rate of [Ca²⁺]i decay and reducing both relaxation velocity and cell shortening in diabetic mGCH1-Tg, again without affecting the amplitude of the [Ca²⁺]i transient.

Discussion

We have shown that increasing myocardial BH4 levels can prevent or reverse the LV dysfunction induced by DM and uncovered a novel mechanism by which BH4 exerts its beneficial effects in the myocardium via nNOS (see Graphical abstract). In contrast to previous reports where prevention of diabetic vascular endothelial dysfunction by Tie2-driven GCH1 overexpression was mediated by preservation of eNOS coupled activity ¹³, we demonstrate that an increase in insulin-independent myocardial glucose uptake accounts for the maintenance of myocardial relaxation in diabetic mGCH1-Tg in the absence of NOS uncoupling.

DM-induced disruption of myocardial nitroso-redox balance has been reported in association with cardiac dysfunction and proposed as an important determinant of the diabetic cardiomyopathy phenotype^{10, 22}. Other studies, however, have shown no difference or a reduction in myocardial superoxide generation in diabetic mice ^{23, 24}. We did not observe a significant difference in myocardial total superoxide production, glutathione or BH4 oxidation, and NOS activity in the myocardium of diabetic mice with LV dysfunction. Similar findings were obtained in samples of the right atrial myocardium from diabetic patients with reduced LV ejection fraction undergoing myocardial revascularization.

Although additional beneficial effects of mGCH1/BH4 overexpression on myocardial oxidative stress may occur in more advanced stages of cardiomyopathy, as reported by Wu et al ¹⁰ and a subtle localized increase in mitochondrial reactive oxygen species cannot be categorically excluded by our experiments, our findings demonstrate that LV dysfunction associated with DM can occur in the absence of reduced NOS activity or PLB phosphorylation, and that augmenting myocardial BH4 content prevents diabetic cardiomyopathy and increases glucose transport and utilization, in the absence of BH4 oxidation.

Prevention of the diabetic myocardial phenotype by myocardial GCH1 overexpression was abolished by nNOS gene deletion or GLUT-1 inhibition, indicating that BH4 exerts its protective effect via a nNOS-mediated increase in glucose availability and utilization. A dynamic NO/PKG-dependent translocation of the insulin-independent glucose transporter, GLUT-1, to the sarcolemmal membrane may underpin enhanced glucose transport in the presence of increased levels of myocardial BH4, in line with previous reports showing that NO induces GLUT-1 membrane translocation and a PKG-dependent increase in insulin-independent glucose uptake in rat ovarian cells ²⁵. Our data indicate that higher glucose uptake via GLUT-1 in cardiomyocytes from diabetic mGCH-Tg is mediated by NO via cGMP-PKG signaling, as it is abolished by both sGC and PKG inhibition. Stimulation of myocardial PKG signaling by kinase oxidation ²⁶ seems unlikely as, at difference with the aorta, we did not observe an increase in oxidative markers or superoxide production in diabetic hearts of either genotype.

DM lowered LV mitochondrial CK activity and PCr/ATP ratio, as assessed by ³¹P cardiac magnetic resonance spectroscopy, in WT but not in mGCH1-Tg, where elevation of PPARα and UCP3 was also prevented. Both proteins have been found to be raised in rodent models of type 1 and type 2 DM²⁷ where they support FA utilization, enhance the FA inhibition of glucose oxidation, and promote oxygen wasting for non-contractile purposes ²⁸. Impairment in myocardial energetics is increasingly regarded as an important player in the pathogenesis of diabetic cardiomyopathy and other conditions leading to LV dysfunction ^29–31. Limitations to effective energy supply to the heart, can adversely impact the ATP-dependent Ca²⁺ reuptake during each cardiac cycle and impact diastolic function³². Indeed, impaired respiratory capacity precedes the development of LV dysfunction in type 1 DM ²⁴ and a reduced PCr/ATP ratio, proportional to the degree of LV diastolic dysfunction, has also been reported in patients with type 1 or type 2 DM ^{21, 33, 34}.

Although insulin and glycemic levels did not differ between 4 and 12 weeks of DM induction, impairment of LV function in WT mice was only observed at 12 weeks in association with significantly higher myocardial levels of UCP3 and PDK4 suggesting that DM caused a time-dependent substrate switch from glucose oxidation toward lactate production and increased fatty acid metabolism leading to LV dysfunction. These findings are in keeping with the lack of effect of acute hyperglycemia on human myocardial function and the development of insulin resistance and reduced ability to oxidize fatty acids observed in type 1 DM over time ^{35, 36}.

Our study has revealed important differences between the vascular endothelium and the myocardium in the response to hyperglycemia. Whereas LV dysfunction did not develop until 12 weeks following diabetes induction, endothelial dysfunction was already evident at 4 weeks. As reported previously ¹³, endothelial dysfunction was associated with increased superoxide production, significant BH4 oxidation, and impaired NO signaling. These findings highlight the relative vulnerability of the endothelium to hyperglycemia but also indicate that the development of significant endothelial dysfunction, as measured in the aorta, is not sufficient to induce the cardiomyopathic phenotype in diabetic mice in the early stages of the disease. We have previously reported that myocardial spillover of biopterins in mGCH1-Tg is minimal or absent and that BH4 levels are not increased in the heart non-myocyte cellular component ⁸, accordingly, there was no evidence of a protective effect on the vascular endothelium of diabetic mGCH1-Tg nor was there evidence of reduced myocardial fibrosis, despite which mGCH1 overexpression was still able to prevent the development of LV dysfunction.

Taken together, our findings open the possibility that BH4 supplementation may have beneficial effects in patients with diabetic cardiomyopathy. In a previous study in patients with ischemic heart disease, administration of a synthetic formulation of the active 6R-isomer of BH4, sapropterin, 2-6 weeks before coronary artery bypass surgery failed to enhance vascular NOS activity or improve endothelial-mediated vasodilatation ³⁷. Our current findings indicate that NOS re-coupling may not be the best surrogate end-point for gauging the efficacy of BH4 supplementation, at least in myocardial disease states. In contrast with findings in human vessels ³⁷, BH4 content (both in absolute terms and relative to its oxidized products) increased in a dose-dependent manner in myocardial samples collected from patients treated with sapropterin (unpublished results), confirming that oral BH4 supplementation increases myocardial BH4 content in humans.

Limitations

We did not investigate the mechanisms by which oral BH4 supplementation improved LV function in diabetic WT mice. In contrast with myocardial-specific GCH1 overexpression, oral BH4 supplementation is also known to improve endothelial function and prevent inflammation ^{38, 39}; to which extent these additional effects contributed to recovering LV function in WT diabetic mice remains to be established. Investigation of the effect of BH4 supplementation in diabetic mGCH-Tg lacking nNOS may provide important information on the contribution of the extra-myocardial additional effect of BH4 of the diabetic cardiomyopathic phenotype.

STZ injection results in loss of pancreatic β-cell activity, leading to hyperglycemia secondary to insulin deficiency that resembles human type 1 DM. Whilst type II DM is more common, patients with type 1 DM have a high risk of developing HF that is dependent on glycemic control and associated with higher fatality ^{3, 40, 41} even in the absence of factors, such as hypertension and obesity, which may confound the pathophysiology of LV dysfunction associated with type II DM. To this end, rodent models STZ-induced DM are well suited to evaluate the toxic effects of hyperglycemia and impaired glucose utilization in the myocardium. Male rodents are more susceptible to the diabetogenic action of STZ than females; for this reason our study was conducted in male mice. Whilst this is an important limitation, there is no evidence indicating that development of heart failure requiring hospitalization in patients with type I DM is different between men and women ³. Although off-target adverse effects of STZ delivered as a high-dose bolus have been reported, these can be minimised by the administration of low-dose STZ delivered by the multiple low-dose STZ injections protocol, used in our paper ⁴².

³¹P MRS experiments showing a reduction in PCr/ATP ratio in diabetic WT hearts but not in Tg were conducted in the absence of FA supplementation, raising the possibility that provision of this source of energy may have attenuated the impact of DM on myocardial energetics. Nevertheless, significant genotype differences were observed in hearts exposed to the same conditions and similar changes in myocardial PCr/ATP ratio have been reported in vivo in patients with type 1 or type 2 DM ^{21, 33, 34}.

Finally, since at difference with the in vivo data, measurements in isolated myocytes could not be obtained sequentially in the same mice, we have not compared these data over time but between groups at 4 and 12 weeks. We believe this is the appropriate way of comparing data obtained using this study design, as in an experiment of long duration variations due to equipment refurbishment and different batches of mice may affect longitudinal measurements but not cross-sectional comparisons. It should be noted that changes in unloaded cell shortening brought about by DM were not associated with differences in the amplitude of the [Ca²⁺]i transient, in keeping with data from human myocardial biopsies showing depressed cardiac myofilament function and Ca²⁺ responsiveness in the presence of diabetes ⁴³.

Conclusions

Our work provides original insights into the management and prevention of early metabolic triggers of diabetic cardiomyopathy and uncovers novel targets for the re-purposing of BH4-based therapeutics.

Novelty and Significance.

What Is Known?

• Nitric oxide (NO) regulates cardiac contractility and relaxation.

• Myocardial neuronal NO synthase (nNOS) activity and relaxation can be enhanced by increasing the intracellular level of the NOS co-factor, tetrahydrobiopterin (BH4).

• Diabetes mellitus (DM) reduces NO synthesis in the vascular endothelium by oxidizing BH4.

What New Information Does This Article Contribute?

• Cardiac dysfunction associated with DM can occur in the absence of BH4 oxidation, increased superoxide production, or reduced NOS activity both in humans and in mice.

• Nevertheless, the diabetic cardiomyopathic phenotype is prevented by raising BH4 content in cardiomyocytes through the overexpression of GCHI and reversed by oral administration of BH4.

• BH4-mediated preservation of myocardial function and energetics is abolished by nNOS gene deletion or GLUT-1 inhibition.

Summary.

The exact cause(s) that triggers diastolic dysfunction in diabetic patients are still unknown. In this work we have shown that diabetic mice develop diastolic dysfunction despite having preserved levels of BH4 and NO availability suggesting that NOS dysfunction and increase ROS are not the initial triggers of cardiac dysfunction.

By increasing intracellular BH4/NO levels in cardiomyocytes we prevented or reversed cardiac dysfunction, suggesting they may play an important role during the progression of the disease. BH4 exerts its protective effect on the diabetic myocardium via a nNOS -mediated increase in insulin independent glucose uptake and utilization. Our findings suggest that BH4 supplementation may both prevent or ameliorate LV dysfunction in patients with DM.

Data Availability

The authors declare that all data and methods supporting the findings of this study are available in the Online Data Supplement or from the corresponding authors on reasonable request. Please see expanded methods and the Major Resources Table in the Supplemental Materials.