Background
The existence of diabetic cardiomyopathy, a myocardial disease associated with diabetes in the absence of coronary artery disease, hypertension or any other known cardiac disease, is supported by evidence accumulated from a large and expanding literature [
1‐
4]. Abnormalities in cardiac function and structure in diabetic subjects have been demonstrated in animal [
5‐
8] and human studies [
9‐
12]. The pathogenesis of this ventricular dysfunction remains unknown, but includes impaired metabolism, myocardial fibrosis, small vessel disease and autonomic neuropathy [
1‐
4].
Mice are increasingly used in diabetes research mainly because of the ability to create genetically engineered mice for investigation of the molecular mechanisms underlying cardiovascular complications [
13‐
19]. Chemical induction of insulin deficiency by a cytotoxic agent for pancreatic beta cells, streptozotocin (STZ), produces a well-characterized model of type 1 diabetes [
20]. This model allows more accurate timing of selected metabolic events and correlation with their pathophysiologic consequences. Assessment of cardiac function in mouse models of both type 1 and type 2 diabetes has relied on conventional echocardiography [
15‐
17,
19], invasive in vivo catheterization [
18,
19,
21], or ex vivo [
14,
22,
23] techniques with their associated restrictions upon accuracy or serial studies in timed experiments. The noninvasive magnetic resonance imaging (MRI) techniques provide a 3-dimensional representation of cardiac structure and function, and have proven to be a reliable and reproducible means of studying in vivo cardiac morphology and function in mice under physiological conditions [
24‐
28]. Only recently, the use of cardiac MRI has been reported for assessment of the cardiomyopathic changes in type 2 diabetic mice [
29].
The purpose of the present study was to use MRI techniques to examine the early onset of myocardial dysfunction in STZ-induced diabetic mice. These studies can then be applied to interventions that may alter or prevent the onset of this condition using the advantage of repetitive examinations over selected periods of time.
Methods
Mice
6–8 week old C57BL/6 female mice were obtained from Harlan (Indianapolis, IN) and housed at the MRI animal facility. All animals had free access to standard laboratory chow and water. Animal protocols were approved by the local Institutional Animal Care and Use Committee.
Induction of diabetes
Mice were injected intraperitoneally with 100 mg/kg of STZ (in 0.05 mol/l citrate buffer, pH 4.5, Sigma, St. Louis, MO) or vehicle once a day for two consecutive days. Tail blood glucose (TBG) levels were measured using a glucose oxidase test strip (Lifescan, Milpitas, CA). Animals were considered to be diabetic if their TBG is > 13 mmol/l. The blood glucose levels and body weight of all mice were monitored biweekly. When their TBG is > 22 mmol/l, they were given intermittent low dosages of long acting insulin glargine [
30] to maintain blood glucose at 19–25 mmol/l. Mitochondrial NADH oxidase activity was determined in ventricular tissues from diabetic mice, and MRI studies were performed on diabetic mice at 4 weeks. Age matched controls were included for comparisons.
Mouse preparation
For MRI studies, mice were anesthetized with inhaled isoflurane (1.5% at 1 L/min oxygen flow) via a nose cone. During the experiment, the mouse was positioned supine on a nonmagnetic warming pad to maintain constant body temperature throughout the MR study.
In vivo MRI
Experiments were performed on a 7 Tesla MR scanner (Bruker BioSpin, Ettlingen, Germany). For cardiac imaging the gradient echo sequence [
31] with
te = 4.0 ms,
tr = 100 ms, and a flip angle,
θ = 85°–90° was used for all experiments. The heart was initially positioned by measurement to the center of the magnet and then further set by observing the position using fast imaging with steady state precession (FISP) [
32] as the animal was moved. Once the heart was in the iso-center a gradient echo sequence with
te = 4.0 ms and
tr = 50 ms was used to acquire single slice images in all three (axial, sagittal and coronal) planes. The slice position for the cine was set using this pilot scan. For cine and volume imaging a field of view,
FOV = 2.56 × 2.56 cm was chosen and 256 each of phase encode and read steps were used to resolve the spatial distribution of excited spins. The image was made after zero filling in each direction to 512 × 512. Window functions were not used and some minor image quality enhancement may be achieved if window functions are used. For our purpose the signal contrast of the myocardium and "white" blood was sufficient to allow calculation of the relevant parameters.
Spectrometer triggering with combined respiratory and ECG triggers was used in all experiments. Thus, all MRI data are adjusted to accommodate for rate changes and the data are reflective of the ambient cardiac function and rate. The cardiac period, P, of mice used in these studies ranged from ~120–180 ms, and a respiratory rate of ~18–30 bpm. Trigger delays, td, were incremented at the spectrometer hardware in a chosen number of steps, nsteps, (nsteps = 10) according to td = (P-(P/nsteps))/P, where period is the time between the peaks of the R wave in the ECG trace. The images at each increment of the period were combined and displayed by cine in the mid-ventricular short axis slice. These provided data for assessment of left ventricular (LV) systolic and diastolic dynamics, and for calculations of fractional circumferential shortening and wall thickening after measuring the endocardial circumferential length and myocardial wall thickness at end diastole and end systole.
For volume calculations, a new set of images were collected with the slice position advanced by the slice thickness of 1.0 mm. In order to reduce the time that multi-slice experiments fit in to the repetition time (Tr) used for single slice experiments, the slice thickness was increased to 1.25 mm. Thus, in this case, the resolution in the slice shifting is compromised by 0.25 mm at most, and contributes an error of about 0.1% in total volume calculations compared with a 1 mm slice thickness. The total blood volume at end diastole and end systole was estimated by taking the sum of all cavity slice volumes assuming a uniform thickness of excitation across a chosen slice at the two trigger points.
MRI analysis
For LV blood volume measurements, endocardial borders were manually delineated. An independent observer was used whenever a questionable or indistinct image was encountered. These measurements rarely changed the summed data obtained from a given data acquisition. The end-diastolic volume (EDV) and end-systolic volume (ESV) were calculated as the sum of all cavity slice volumes at end diastole and end systole. For assessment of LV systolic and diastolic dynamics, the cavity slice volume was measured in all acquired images and was plotted against the time from onset of the QRS trigger, and a volume-time curve was established. Peak ejection rate and peak filling rate were calculated from the maximum slopes of the LV ejection and filling curve [
33].
Echocardiography
M-mode/Doppler echocardiography was performed using a 13 MHz 15L8 probe and an Acuson Sequoia 256 echocardiography system (Acuson, Mountain View, CA). The wall and cavity dimensions during diastole and systole were determined from long-axis views of the left ventricle [
34]. The mitral flow velocities were assessed by Doppler, and the E wave and A wave were measured to provide an estimate of diastolic function [
35].
Mitochondrial function
Frozen and thawed mitoplasts from hearts were diluted into 10 mmol/l 3-(N-morpholino)propane-sulfonic acid (MOPS), 25 mmol/l KCl at pH 7.4 to a final concentration of 50 μg/ml [
36]. NADH oxidase activity was measured as the rate of NADH disappearance at 340 nm upon addition 200 μM NADH [
36]. Citrate synthase activity was monitored by detection of 5,5'-dithiobis(2-nitrobenzoic acid) (DTNB) reactive CoASH (412 nm, ε = 13,600) upon addition of 0.1 mmol/l DTNB, 0.3 mmol/l acetyl CoA, and 0.5 mmol/l oxaloacetate to frozen and thawed mitoplasts (20 μg/ml) [
37]. All enzyme assays were performed at room temperature.
Statistical analysis
Data are expressed as mean ± SEM. An unpaired 2-tailed Student t test was used to make comparisons between control and diabetic groups. Significance was ascribed to P values < 0.05.
Discussion
It has been difficult in obtaining sequential analyses of early diabetic cardiomyopathy in humans, and even in small animal models. The cardiac phenotype of diabetic mice has been examined using in vivo and ex vivo techniques in both type 1 and type 2 diabetic mouse models [
14‐
19,
21‐
23]. In vivo evidence for cardiac dysfunction in diabetic mice is commonly obtained from echocardiography [
15‐
17] and/or invasive catheterization [
18,
21].
Transthoracic echocardiography is a noninvasive method for in vivo assessment of cardiac function. It is readily available and very useful for serial studies on age-dependent cardiac changes [
38,
39]. M-mode echocardiography provides indices of LV systolic function and calculated LV mass, while Doppler measurements of transmitral flow provide an index of diastolic function [
26,
40]. Systolic and diastolic dysfunction has been demonstrated in diabetic mice [
15‐
17]. However, these animals were either severely diabetic (blood glucose of > 33 mmol/l for 3 weeks) or chronically diabetic (12 weeks). In a study of 4-week STZ-diabetic mice, echocardiography did not reveal a significantly reduced systolic function [
19]. This may relate to the less sensitive techniques used for systolic function evaluation. Additionally, methods for measurement of cardiac mass and volume generally depend on use of linear measurements and calculation of mass and volume based on geometric assumptions that may or may not be accurate for that organ.
MRI provides accurate, reproducible, noninvasive 3-dimensional representations of cardiac structure and function. The MRI techniques have been adapted and validated to assess mouse cardiac function in vivo [
24‐
28]. Unlike echocardiography, the 3-dimensional MR images also provide an accurate analysis of morphological changes without requiring geometric assumptions [
26]. Our studies using this technique clearly demonstrated that STZ-diabetic mice had ventricular wall thinning, increased end-systolic diameter and volume, diminished ejection fraction, decreased circumferential shortening, and decreased peak ejection and filling rates. These changes are consistent with those described using other in vivo techniques at various stages in development of diabetic cardiomyopathy [
15‐
18,
21]. The MRI interrogation of myocardial performance appears to offer the sensitivity needed for the early detection of physiological events that correlate with the onset of metabolic and other pathophysiologic mechanisms that are etiologic in diabetic cardiomyopathy.
In preliminary studies, we examined one marker of the onset of mitochondrial dysfunction (NADH oxidase activity) and demonstrated this parameter was decreased by 50% at 4 weeks. This translates into a significant deficiency of mitochondrial electron transport and serves as one of several indicators of an impaired cardiac energy state in early diabetes. These data help confirm a metabolic basis for the observed impairment of cardiac function using MRI interrogation. Mitochondrial dysfunction has been reported in the diabetic heart [
41,
42]. Such reduced energy stores would lead to subsequent myocardial dysfunction. Since mitochondrial ATP synthesis is the major source of energy in cardiomyocytes, repetitive measurement of this parameter would be of value. Future work must establish precise molecular mechanisms by which diabetes leads to declines in mitochondrial function and determine whether these decrements are of sufficient magnitude to limit the supply of energy for cardiac function.
Recently, MR spectroscopy has been used in mice to examine myocardial energy metabolism [
27,
43].
31P spectra provide the only
noninvasive method to quantitate myocardial high-energy phosphates (ATP, phosphocreatine). This novel technique should prove complementary to standard MRI interrogation to study the contribution of altered energy metabolism to the development of diabetic cardiomyopathy. The major drawback at present is the design of appropriate small coils that will permit accurate localization of data to the myocardium separate from the underlying blood.
Despite its advantages, MRI has limitations. It is costly, time consuming, has blood flow and motion generated artifacts, and has limited availability. Application of sensitive, discriminate software to minimize manual determination of selected cardiac parameters such as cardiac borders in repetitive slices will decrease the time required for analysis. We have found that the time required for positioning and acquisition of the relevant data was significantly decreased with experience of the personnel.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
XY and DCK conceived the study, participated in its design and coordination, performed and analyzed the data with statistical analysis and drafted the manuscript. RAT participated in the design of the study, provided expertise and oversight in the MRI procedures and helped with manuscript preparation. YAT and AA carried out the MRI data acquisition and helped with data analysis and manuscript preparation. SH and MWG carried out the diabetes induction and mouse maintenance. EP and SC carried out the echocardiography and data collection, and helped with data analysis. SM performed the mitochondrial assay, and LIS helped with data interpretation and manuscript preparation. BEG supervised the mouse maintenance and assisted with data analysis. All authors read and approved the final manuscript.