INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

TRANSGENIC MANIPULATIONS in mice are increasingly used to probe genetic and physiological aspects of human cardiovascular physiology. However, the extremely small size (~0.1 g) and extremely fast rates (~600 beats/min) of murine hearts under physiological conditions have made noninvasive in vivo studies of murine cardiac metabolism, function, and morphology difficult.

The prime myocardial high-energy phosphates, ATP and phosphocreatine (PCr), can only be noninvasively quantified in the beating heart by ³¹P magnetic resonance spectroscopy (MRS), the sibling technology to magnetic resonance imaging (MRI). Image-guided ³¹P MRS studies in humans demonstrate that the cardiac PCr-to-ATP (PCr/ATP) ratio in the normal human heart under physiological conditions is 1.8-2.0 (12, 15, 17, 23) and that this ratio is altered transiently during stress-induced ischemia (23, 24) and chronically with heart failure (11, 12, 18). Prior ³¹P MRS studies in mouse hearts have been conducted on isolated, perfused preparations (20, 21). Despite the important metabolic insights generated from ³¹P MRS studies of isolated mouse hearts, the terminal preparation mandates acute, end-point studies under conditions that are not truly physiological. Chronic, repeated studies of cardiac energy metabolism in mice that are not reliant on parallel studies require a noninvasive approach akin to that used in humans. Regional and global left ventricular systolic function in humans can be assessed by a number of techniques, including MRI. The normal human left ventricular ejection fraction is typically found to be 60-70% (2, 8). In mice, hemodynamics are often measured invasively (13), whereas noninvasive measures of cardiac morphology and function are typically derived by echocardiography or MRI. High-resolution MRI has been reported and validated as an accurate means for determining murine myocardial mass (9, 19). MRI has also been used to assess murine cardiac function, but prior reports have been conducted at subphysiological heart rates (9, 19).

The aim of this study was to develop an integrated, noninvasive means of studying in vivo murine cardiac metabolism, morphology, and function under physiological conditions by adapting and modifying noninvasive cardiac MRI with image-guided ³¹P MRS techniques used in humans to mice. This approach would offer novel, important, and repetitive data to an existing field that otherwise evaluates murine cardiac physiology with separate, invasive techniques and that typically relies on parallel studies for multiple end points. The combination of fundamental energetic, functional, and anatomic cardiac information promises to be an important in vivo tool for assessing the physiological importance of transgenic manipulations in mice. In addition to developing this important physiological tool in mice, the technique is used to test the previously proposed hypothesis that cardiac high-energy phosphates scale with mammalian body size (7) by comparing novel measures in mice, the smallest species studied to date, with those previously reported in humans.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animal handling and general approach for MRI/MRS examinations. All procedures and protocols were reviewed and approved by the Institutional Animal Care and Use Committee of The Johns Hopkins University.

High-resolution ¹H MRI of murine hearts. High-resolution images were obtained to confirm position, define the regions of metabolic interest, or quantify ventricular function. After the cradle containing the mouse was positioned in the magnet, and the position of the spectroscopy coil relative to the heart ascertained by fast scout images, an ECG-gated, spin-echo transverse image for correlation with spectroscopic data was obtained [echo time (TE) = 11 ms, recycle time (TR) = ~500 ms, slice thickness (ST) = 1.6 mm, field of view (FOV) = 32 mm, matrix size = 256 × 128 (zero-filled to 256 × 256), numerical aperture = 2-8, total acquisition time = 2-8 min]. Diffusion gradients were applied symmetrically on both sides of the refocusing pulse to crush the blood signal and minimize flow artifacts. In some mice, a complete set of multislice short-axis images (ST = 1.2 mm, no gap between slices) for end diastole and end systole was acquired. Each slice was acquired exactly at the same time point in the R-R interval by waiting one R-R interval between slices. Left ventricular volumes at end diastole and end systole were determined using the software package NIH Image version 1.52 for a Macintosh computer from these multislice images (matrix size 256 × 256). The left ventricular ejection fraction was calculated from the relative difference in end-diastolic and end-systolic cavity volumes.

Spatially localized ³¹P MR murine cardiac spectroscopy. Spatially localized ³¹P NMR spectra were acquired after optimization of the magnetic field homogeneity using the ¹H coil to shim on a thick slice containing the heart. A one-dimensional chemical shift imaging (1D-CSI) sequence was used with a field of view of 32 mm, 32 phase-encode steps in the direction perpendicular to the plane of the coil, and a recycle delay of 2 s with 32 averages per phase encode step for a total acquisition time of ~34 min. With this protocol, well-resolved spectra from 1-mm slices parallel to the coil are obtained, and each slice of the spectroscopic data was spatially correlated with the high-resolution image to assign the anatomic origin. Studies in phantoms with these coils demonstrated that the image intensity across the region of interest is uniform, there are no noticeable "bleed artifacts" between the 32 phase-encoded 1D-CSI slices, and that when hard radio frequency (RF) pulses are used, higher flip-angle pulses provide better RF excitation at deeper slices (typical cardiac depth in mice) at the expense of overflipping and saturating spins in the immediate vicinity of the spectroscopy coil (superficial chest muscle depth in mice). Additional MRI studies with the ³¹P surface coil temporarily retuned to the ¹H frequency demonstrate that the signal intensity contribution of the chest "side" wall to anterior cardiac slices detected by this ³¹P coil centered over the mouse heart is a relatively small fraction (10%) of the total signal (Fig. 1). Other studies demonstrated that ¹H images obtained with standard imaging parameters over several minutes (Fig. 2A) were comparable to ¹H images acquired over 34 min with spectroscopy type parameters (Fig. 2B). These indicate that there are no artifacts present in the long spectroscopy-type acquisitions not present in the high-quality standard 5-min acquisition and demonstrate that respiratory motion and heartbeat influence 34-min spectroscopy acquisitions and their localization to a similar extent as the more rapid 2- to 5-min imaging sequences.

View larger version (85K): [in this window] [in a new window]   Fig. 1.   Transverse 1H magnetic resonance (MR) images of a mouse thorax through the heart at end diastole obtained using an electrocardiogram (ECG)-gated spin-echo sequence acquired with the 1H imaging coil (A) and with the 31P spectroscopy coil temporarily retuned to the 1H frequency. Thin white lines (B) annotate a nominal 1-mm slice interrogated by 31P with a one-dimensional chemical shift imaging (1D-CSI) sequence. "Side" chest walls contribute very little signal to the typical cardiac region-of-interest studied with spatially localized 31P NMR and these coils.

View larger version (79K): [in this window] [in a new window]   Fig. 2.   Transverse images of a mouse thorax at end diastole obtained using an ECG-gated spin-echo sequence acquired over 5 min with conventional imaging parameters (A) and another acquired over 34 min with spectroscopy-type parameters (B). These show that there are no artifacts present in the long spectroscopy-type acquisitions not present in the shorter conventional imaging acquisitions, demonstrating that respiratory motion and heartbeat influence spectroscopy acquisitions and their localization to a similar extent as the more rapid imaging sequences.

Calculation of myocardial [ADP] and G_~ATP. Calculations of intracellular [ADP] and the free energy of ATP hydrolysis (G_~ATP) were based on the following established relationships
where K_eq is 1.66 × 10⁹ (mol/l)¹ and [free creatine] is the difference between [total creatine] and [creatine phosphate]. The change in free energy state due to ATP hydrolysis
where G^o (30.5 kJ/mol) is the value of G_~ATP under standard conditions of molarity, temperature, and pH (10), R is the gas constant (8.3 J/mol K), and T is absolute temperature (in K). In mice, [total creatine] = 30 mmol/l, [ATP] = 9.6 mmol/l (20), and intracellular pH = 7.2, based on ³¹P NMR measures in perfused mouse hearts (21). [P_i] is estimated at  2 mmol/l, based on the inorganic phosphate peak relative to that of ATP, when P_i could be unambiguously identified. In humans, [total creatine] = 43.3 mmol/l (16), [ATP] = 12.08 mmol/l [= 5.8 mmol/g wet wt/ 0.48 ml intracellular water/g wet wt (4)], intracellular pH = 7.2, and Pi estimated at 2 mmol/l.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A representative high-resolution anatomic image and spatially localized ³¹P MR spectra acquired in the same examination at a physiological heart rate are shown in Fig. 3. The ³¹P MR spectrum arising primarily from the heart appears similar to that previously observed in other species in vivo, including humans, with nearly twice as much PCr as ATP (3, 12, 15, 17, 23). We measured the cardiac PCr/ATP ratios in six mice by using a low-flip angle, minimally saturated approach and in four other mice by using a high-flip angle (180° at the center of the coil, ~90° at the depth of the myocardium), partially saturated approach. The former has the advantage of directly providing accurate cardiac PCr/ATP ratios without correcting for saturation effects. The latter has the advantage of exciting deeper cardiac regions with trivial superficial skeletal muscle contamination but does require partial saturation correction. Maximum saturation factors were calculated from the literature for the latter series from a 90° pulse and the T₁ values of PCr (5.3 s) and [-P] of ATP (2.7 s) from porcine and human myocardium at similar field strengths (14, 15). The mean uncorrected cardiac PCr/ATP ratio was 1.50 ± 0.11 (means ± SD) for this latter, partially saturated series, and the saturation corrected PCr/ATP ratio was 2.02 ± 0.15. The murine cardiac PCr/ATP ratio in the low-flip angle experiments was 2.02 ± 0.24. Thus both approaches generated identical findings. Mean intracellular pH and Mg²⁺ were calculated in 9 of the 10 mice from the ³¹P MR spectra and were 7.17 ± 0.06 and 0.58 ± 0.27 mM, respectively.

View larger version (49K): [in this window] [in a new window]   Fig. 3.   Transverse image of a mouse thorax through the heart at end diastole obtained using an ECG-gated spin-echo sequence (gating trigger 90 ms after R wave) with recycle time (TR) = ~500 ms, echo time (TE) = 11 ms, numerical aperture (NA) = 8, field of view (FOV) = 32 mm, slice thickness (ST) = 1.6 mm, matrix size = 256 × 128 (zero-filled to 256 × 256 during processing), and total acquisition time ~8 min. Selected 31P spectral slices containing the myocardium (A) and the edge of a bulb containing phenylphosphonic acid, adjacent to the chest (B) from a 1D-CSI data set [FOV = 32 mm, 32 phase-encoded steps, NA = 32, pw =20 µs (180° flip angle at the center of the coil) acquired in ~34 min] are shown. C: actual ECG tracing of this mouse during study with a physiological R-R interval of ~113 ms or a heart rate of ~530 beats /min.

Figure 4 shows representative multislice, spin-echo cardiac MRI from which morphological and functional data were collected. The high-resolution images, as shown in Fig. 1, have a high contrast between the myocardial wall and intracavitary blood and allow determination of myocardial structures at multiple levels and at multiple views. The mean left ventricular ejection fraction, calculated from multislice, end-diastolic, and end-systolic images, was 65.1 ± 7.0% (means ± SD).

View larger version (60K): [in this window] [in a new window]   Fig. 4.   Representative multislice, spin-echo murine cardiac magnetic resonance images at end diastole. Left ventricular long-axis view (top) and 6 contiguous short-axis slices (bottom) from base (left) to apex (right). Imaging parameters are ST = 1.2 mm (no gap between slices), TR ~600 ms, TE = 11 ms, FOV = 32 mm, matrix 256 × 256, yielding in-plane pixel resolution of 0.125 mm, and no. of acquisitions = 2. All slices were obtained at exactly the same time point in the cardiac cycle by waiting one R-R interval between slices.

Table 1 summarizes the in vivo bioenergetic and functional findings in mice and indicates that the murine cardiac PCr/ATP ratio and left ventricular ejection fraction at physiological heart rates are both similar to those reported in humans (2, 8, 12, 15, 17, 23). Here, we measure in vivo murine myocardial PCr/ATP ratios of 2.0 ± 0.2 and calculate from them murine cardiac [ADP] as ~50 µmol/l and G_~ATP as ~60.2 kJ/mol. In vivo human cardiac energetic data are cited from studies that utilized the best reported ³¹P NMR spatial localization (smallest volume elements), potentially the least noncardiac contamination (15) or those that corrected for partial saturation and blood contamination effects (4, 17). Older studies reporting lower human cardiac PCr/ATP ratios did not correct for these confounding variables and/or interrogated larger volume elements with greater potential noncardiac contamination. On the basis of the most precise recent reports of in vivo high-energy phosphates in normal human hearts by us (11, 23) and others (15, 17) (PCr/ATP 1.8-2.0), we estimate human cardiac [ADP] as ~90 µmol/l and G_~ATP as ~59.1 kJ/mol.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Here we describe the first noninvasive, physiological, in vivo measures of myocardial high-energy phosphates and left ventricular contractile function in intact mice from image-guided ³¹P MRS and MRI. The approach adapted MR techniques used to noninvasively study cardiac energetics and function in humans to mice. Combined MRS/MRI offers novel, important, and repetitive data to an existing field that otherwise evaluates murine cardiac physiology with separate techniques and that typically relies on parallel studies for multiple end points. The combination of fundamental energetic, functional, and anatomic cardiac information promises to be an important in vivo tool for assessing the physiological importance of transgenic manipulations in mice.

Murine cardiac energetics were previously investigated with ³¹P NMR spectroscopy of isolated, perfused hearts. Our observations of in vivo murine cardiac PCr/ATP ratios ~2 are generally consistent with some, but not all, prior reports in excised mouse hearts (20, 21). Although studies of isolated, perfused animal hearts can provide important insights, they lack adrenergic stimulation, do not operate under physiological loading conditions, exhibit reduced contractility, and manifest markedly reduced heart rates compared with physiological, in vivo conditions. In addition, the "normal" cardiac PCr/ATP ratio reported in isolated, perfused wild-type mouse hearts varies by nearly 60% between studies (20, 21), and this variability may be due to differences in substrate availability and perfusion conditions. The terminal nature of isolated heart studies also precludes repeated measures in the same mouse. This can be a very important consideration in transgenic lines where the numbers of mice are often limited. The noninvasive nature of the current approach allows repeated integrated measures in a single mouse over time as adaptations to transgenic manipulations occur. For all of these reasons, including the variability and dependence of the PCr/ATP on experimental conditions in perfused hearts, the nonphysiological limitations of those preparations, and the terminal nature of isolated heart studies, the current noninvasive in vivo ³¹P MR techniques offer some clear advantages over prior approaches.

Prior MRI reports characterized wild-type murine left ventricular systolic function and calculated ejection fractions similar to those reported here (9, 19). Some of these reports (9, 19) have validated cardiac MRI as a precise tool for assessing ventricular mass in mice. The current approach builds on prior work and extends it in at least two ways. First, it demonstrates that high-resolution images can be generated in the same exam as cardiac metabolic information by ³¹P MRS. Second, it demonstrates that measures of murine ventricular function can be obtained by MRI at physiological heart rates. These observations, at the highest murine heart rates yet reported, confirm that the global left ventricular ejection fraction is similar in mice to that in normal humans.

In addition to developing this important physiological tool in mice, the technique was used to test the previously proposed hypothesis that cardiac high-energy phosphates scale with mammalian body size (7) by comparing novel measures in mice, the smallest species studied to date, with those previously reported in humans. Among mammalian species there are marked differences in heart rates, and hence energy demands, with smaller animals exhibiting faster heart rates than larger animals. It has been proposed that the creatine kinase energy metabolites (PCr/ATP) and the free energy of ATP hydrolysis (G_~ATP) scale with mammalian body size (7). Higher cardiac PCr/ATP ratios, lower calculated [ADP], and higher G_~ATP have been reported in the hearts of smaller animals (rats and rabbits, 0.3-3.0 kg/wt) than those in larger animals (dogs and humans, 10-80 kg/wt) (7). Such alometric bioenergetic scaling was proposed to affect metabolic efficiency and to constrain endotherm body size. On the basis of prior estimates from invasive measures in rats and rabbits, the prior literature predicts that mice (~0.03 kg/wt) would have cardiac PCr/ATP ratios of 2.6-2.7, [ADP] of ~11 µmol/l, and G_~ATP of ~67 kJ/mol, which are much different from those of humans. However, despite the 2,000-fold difference in body mass and the 10-fold difference in heart rates, we directly observe that mice have a similar in vivo cardiac PCr/ATP ratio to that of humans and a calculated [ADP] that is not one-seventh but one-half of that of humans and a G_~ATP that is only 1 kJ/mol higher, rather than 6-7 kJ/mol higher. These first direct, image-guided, bioenergetic observations in the smallest mammalian hearts studied in vivo to date do not provide evidence for sufficient metabolic scaling of creatine kinase energy metabolites to constrain body size. In addition, because skeletal muscle has a higher PCr/ATP ratio than cardiac muscle and because there may have been some small amount of skeletal muscle contamination of our murine cardiac spectra, these murine cardiac PCr/ATP ratios may be somewhat high, rather than artificially low, making the differences in [ADP] and G_~ATP between these species even smaller than we calculate and any metabolic scaling even less than the upper limit noted here. Prior evidence for metabolic scaling was based on cardiac measures in intubated, open-chest rats and rabbits (7). The current measures are not only more powerful because they investigate a species whose body weights are an order of magnitude smaller, but also because they more accurately reflect in vivo physiology because they were obtained noninvasively at physiological heart rates in lightly sedated, intact animals.

It is tempting to speculate about why the cardiac PCr/ATP ratio is ~2 in mice and humans and independent of mammalian body size. First, to the extent that PCr is an energy buffer and that the PCr/ATP ratio reflects the G_~ATP (10), this value may represent an optimal set point for endotherm chemical energy release. Second, the constancy of the PCr/ATP ratio may relate to metabolic control. Most prior in vivo studies within a species demonstrate that a two- to threefold increase in heart rate over the physiological range does not change the cardiac PCr/ATP ratio in normal hearts (1, 26). Those observations have been interpreted to indicate that in vivo cardiac energy metabolism is not classically regulated by [ADP] over the submaximal, physiological range in mammals. It is also tempting to speculate that a relatively constant cardiac PCr/ATP ratio and [ADP] across species with 10-fold differences in heart rate is consistent with the hypothesis that ADP does not control cardiac energy turnover under typical physiological conditions across mammalian species. Finally, the consistency of the PCr/ATP ratio and specifically of [ADP] across species may have implications for diastolic function. Studies in isolated muscle fibers indicate that ADP directly affects cross-bridge cycling and slows the velocity of actin filament sliding on cardiac myosin (25). Alterations in myofilament kinetics with [ADP] can even occur at high ATP concentrations indicating a separate mechanism affecting diastolic tone (25). Recent evidence from isolated hearts shows that increases in free [ADP] correlate closely with increases in left ventricular diastolic pressure and that failure to maintain a low [ADP] impairs diastolic function (22). To the extent that [ADP] may independently affect diastolic pressure, the current observations of a relatively constant [ADP] across species are consistent with the observed conservation of left and right ventricular diastolic pressures across species as well. These observations are consistent with prior conclusions about energy regulation and cardiac [ADP] from previous studies, but compelling new evidence to address these issues would require interventions to alter workload, metabolic control, or diastolic tone, and such experiments were not conducted and are beyond the scope of the current endeavors.

In summary, it is now possible with MRS and MRI to noninvasively study in vivo myocardial bioenergetics, morphology, and contractile function in mice under physiological conditions. The noninvasive nature of the technique will enable repeated studies in a single mouse over time. Despite the marked differences in size and heart rate, measures of relative cardiac high-energy phosphate metabolites and global contractile function are similar in normal, wild-type mice and normal humans. Transgenic murine lines with specific genes overexpressed or deleted are increasingly generated with the hope that they can provide models of human cardiovascular disease. Although these observations do not confirm a degree of metabolic scaling with body size in line with prior predictions, they do suggest that mice can serve, at least at this level, as a model for human cardiovascular physiology and demonstrate the feasibility of using these unique tools to noninvasively probe the pathophysiological impact of transgenic manipulations on cardiac metabolism and function in mice.