INTRODUCTION

Metabolic dysfunction in diabetes mellitus results in several systemic changes in metabolite concentrations across multiple organs. Earlier studies in our laboratory have linked the receptor for advanced glycation end-products (RAGE) and its tail binding partner diaphanous 1 (DIAPH1) as key mediators of metabolic, signaling and functional changes in diabetic mice [1-3]. In this study, we probed for organ specific changes in metabolites of diabetic mice and the impact of DIAPH1 deletion in influencing metabolic changes in diabetic mice using proton magnetic resonance spectroscopy (¹H-MRS) and chemical-shift encoded (CSE) MRI.

METHODS

The animal experiments were conducted according to the procedures approved by the Institutional Animal Care and Ethical Committee (IACUC) of our institution. Three cohorts of mice were used consisting of male wild type (WT) C57BL/6 mice (Jackson Laboratory, ME) (n=5), diabetic WT mice (WT-DM) (n=3), and diabetic Diaph1 (DIAPH1 KO-DM) knockout mice (n=3, bred in-house). Type-1 diabetes in WT and Diaph1 KO mice was induced by destruction of the pancreatic cells by injection of streptozotocin (55mg/kg for 5 days) at the age of 6 weeks as published earlier[4]. The blood glucose was monitored for all mice, with the diabetic mice used in this study having a blood glucose greater than 250 mg/dl.

All MRI/MRS scans were performed on a 7T micro-MRI (Bruker Bio-Spin, USA) equipped with 750mT/m gradients, with a vendor supplied transmit-receive volume coil. For the MR-imaging, isoflurane anesthesia was maintained at 1.5-2% delivered in 1:1 ratio of oxygen to air mixture via a nose cone. The isoflurane level was adjusted such that a respiration rate of 25-30 /minute was achieved. All mice underwent single voxel ¹H-MRS and CSE-MRI of the liver and heart. MRS experiments used a PRESS sequence with TR=2000ms, TE=16ms, bandwidth=4000Hz, 256 averages with and without water suppression using the variable pulse power and optimized relaxation delays (VAPOR) module, total scan time was about 8 min. The voxel sizes used were 3x3x3 mm³ for the liver and 1x2x2 mm³ for the heart respectively. The liver and heart MRS were respiratory triggered. The voxel in the liver was placed away from the liver edges, and avoiding major blood vessels, while the voxel in the heart was rotated and placed along the interventricular myocardial septum (IVS). CSE-MRI used a non-commercial 3D multi-echo gradient recalled echo (GRE) sequence covering the torso of the mouse. The technique uses optimized echo spacing with a 3-point Dixon water-fat suppression technique with the following parameters: TR=10ms, TEs=2.6ms, 2.93ms, and 3.26ms, FA=10°, FOV=30x30x30 cm³ image matrix=128x128x64, in-plane resolution=2.3mm², scan time=4min.

The MRS data was post-processed and quantified in the time domain using jMRUI software [5]. Spectra were phase and frequency corrected and apodised using a Lorentzian line shape with a damping factor of 5Hz. For water suppressed spectra, the residual water peak was removed using singular value decomposition (SVD). Quantification of the spectra was done using the AMARES algorithm. The unsuppressed MR-spectra were used to determine the ratio of water and lipid peaks. For CSE-MRI image post-processing, the raw data at multiple TE’s were reconstructed using the IDEAL algorithm [6]. Separate water and fat images were reconstructed, and water and fat fraction maps were also computed. A non-parametric Kruskal-Wallis test was used to compare the cohorts, with a significance level set to 0.1.

RESULTS

Figure 1 shows the ¹H-MRS results in the liver. Figure 1(a)-(c) show representative liver spectra from WT, WT-DM and DIAPH1 KO-DM cohorts. Figure 1(d) shows the location of the voxel for MRS, and figure 1(e) summarizes the quantification in percent of metabolite in the three cohorts. In WT-DM, the liver shows considerably higher lipid levels. Compared to the WT mice, the methylene peaks are significantly higher in WT-DM and DIAPH1 KO-DM cohorts (P=0.063). The DIAPH1 KO-DM cohort shows considerable reduction in the diallylic peaks compared to WT and the WT-DM cohorts (P=0.01).

Figure 2 shows the ¹H-MRS results in the heart. Figure 2(a-c) show representative cardiac spectra from WT, WT-DM and DIAPH1 KO-DM cohorts. Figure 2(d) shows the location of the voxel in the IVS, and figure 2(e) shows the summary of the cardiac metabolite quantification among the cohorts. Of note, is the significant reduction in triglycerides in the DIAPH1 KO-DM mice compared to WT and WT-DM mice (P=0.058). The creatine levels in all diabetic mice were significantly lower compared to WT mice (P=0.071).

Figure 3 shows the results from the CSE-MRI. In the myocardium of the heart, there are considerable increases in the fat fraction from WT to the WT-DM cohort, which reduces in the DIAPH1 KO-DM mice cohort. The fat fraction in the liver is the highest in the WT-DM cohort, followed by the DIAPH1 KO-DM mice, and is the least in the WT mice.

DISCUSSION AND CONCLUSION

This study shows significant differences in metabolite levels between WT, WT-DM and DIAPH1 KO-DM mice in the liver and heart using ¹H-MRS and CSE-MRI techniques. The metabolite levels and water-fat distribution in the three cohorts suggest that DIAPH1 KO-DM mice, in spite of being diabetic, experience a protective effect owing to DIAPH1 deletion.

Acknowledgements

This study was supported by NIH grants PO1 HL146367, RO1 DK109675, R01 AR067156, and R01 AR068966.