Hyperpolarized ketone body metabolism in the rat heart

The aim of this work was to investigate the use of 13C‐labelled acetoacetate and β‐hydroxybutyrate as novel hyperpolarized substrates in the study of cardiac metabolism. [1‐13C]Acetoacetate was synthesized by catalysed hydrolysis, and both it and [1‐13C]β‐hydroxybutyrate were hyperpolarized by dissolution dynamic nuclear polarization (DNP). Their metabolism was studied in isolated, perfused rat hearts. Hyperpolarized [1‐13C]acetoacetate metabolism was also studied in the in vivo rat heart in the fed and fasted states. Hyperpolarization of [1‐13C]acetoacetate and [1‐13C]β‐hydroxybutyrate provided liquid state polarizations of 8 ± 2% and 3 ± 1%, respectively. The hyperpolarized T 1 values for the two substrates were 28 ± 3 s (acetoacetate) and 20 ± 1 s (β‐hydroxybutyrate). Multiple downstream metabolites were observed within the perfused heart, including acetylcarnitine, citrate and glutamate. In the in vivo heart, an increase in acetylcarnitine production from acetoacetate was observed in the fed state, as well as a potential reduction in glutamate. In this work, methods for the generation of hyperpolarized [1‐13C]acetoacetate and [1‐13C]β‐hydroxybutyrate were investigated, and their metabolism was assessed in both isolated, perfused rat hearts and in the in vivo rat heart. These preliminary investigations show that DNP can be used as an effective in vivo probe of ketone body metabolism in the heart.


| METHODS
Ex vivo experiments were performed using a Bruker Avance 11.7-T vertical bore magnetic resonance imaging (MRI) system (Bruker Biospin GmbH, Ettlingen, Germany). In vivo experiments were performed using an Agilent 7-T MRI system (Agilent, Santa Clara, CA, USA). DNP was performed using a HyperSense hyperpolarizer (Oxford Instruments, Abingdon, Oxfordshire, UK) for ex vivo experiments, or a previously described prototype hyperpolarizer 6 (GE Healthcare, Amersham, Buckinghamshire, UK) for in vivo experiments. All animal investigations conformed to Home Office Guidance on the Operation of the Animals (Scientific Procedures) Act (HMSO) of 1986, to institutional guidelines and were approved by the University of Oxford Animal Ethics Review Committee. All compounds were obtained from Sigma Aldrich (Gillingham, Dorset, UK) unless otherwise specified.

| Sample preparation
It was not possible to directly obtain free acetoacetic acid (or its sodium salt) owing to its chemical instability and comparatively short half-life (approximately 2 h as an acid, several days as a base) at room temperature prior to spontaneous decarboxylation into acetone and carbon dioxide.
Immediately prior to hyperpolarization, a stock solution was formulated by mixing 280 mg of the generated sodium [1- 13  Gd-DOTA (Gd, gadolinium; DOTA, 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid; Dotarem, Guerbet, France) prior to vortexing and pipetting into a sample cup for rapid freezing. DNP was then performed at~94 GHz and 100 mW for 60 min. 6 Dissolution was carried out using 6 mL of heated and pressurized deionized water (~180°C, 10 bar), resulting in a final liquid state concentration of 40mM for both substrates. As a result of reagent and instrument availability, separate batches were synthesized for in vivo, ex vivo and phantom experiments. Deionized water was used to ensure that the compounds injected were kept under basic conditions, minimizing the spontaneous rate of decarboxylation occurring during the initial high-temperature part of the dissolution process, which may otherwise lead to erroneously high recordings of bicarbonate originating from chemically evolved CO 2 dissolved in buffer. The spontaneous rate of decarboxylation of acetoacetate is expected to obey an Arrhenius relation whose spectrophotometrically measured 23 apparent first-order rate constant is given by where k a ≈ 29 × 10 −6 s −1 , k i ≈ 0.5 × 10 −6 s −1 and K a ≈ 2.84 × 10 −4 which, at 37°C and pH 7, is~5.5 × 10 -7 s -1 . Given that the rate of decarboxylation is expected to approximately double for each 10°C rise, 24 we therefore only expect the spontaneous decarboxylation of acetoacetate to arise from exposure to temperatures in excess of~150°C.

| Polarization and T 1 measurement
The liquid state polarization and T 1 relaxation times were measured at 11.7 T. Two millilitres of hyperpolarized liquid were injected into a 20-mm nuclear magnetic resonance (NMR) tube within a dual-tuned 13

| Isolated heart perfusion
Hearts from male Wistar rats (body weight,~300 g; Harlan, Bicester, Oxfordshire, UK) were prepared for perfusion in the Langendorff mode, as described previously. 14 Briefly, rats were terminally anaesthetized with 140 mg/kg pentobarbitone. Hearts were then excised and rapidly washed in ice-cold Krebs-Henseleit (KH) buffer, followed by dissection to reveal the aorta. The aorta was cannulated, tied off with monofilament 3/0 (0.3 mm in diameter) silk suture (Pearsalls, Taunton, Somerset, UK) and the heart was perfused in the Langendorff mode at constant pressure (85 mmHg/11.3 kPa, corresponding to an approximate flow rate of 15 mL/min) using KH buffer at a temperature of 37°C and oxygenated with 95% O 2 /5% CO 2 gas. [27][28][29][30] A polyethylene tube was inserted into the left ventricle and through the apex of the heart in order to drain Thebesian flow. A polypropylene balloon connected to a pressure transducer and a PowerLab system (AD Instruments, Abingdon, Oxfordshire, UK) was then inserted into the left ventricle to monitor contractile function, and the heart was subsequently placed inside a 20-mm NMR test tube, which was inserted into the bore of the 11.7-T magnet described above.

| Assessment of ketone body metabolism in the perfused heart
Hearts from male Wistar rats (n = 3 per substrate, i.e. six rats in total) were initially perfused with KH buffer containing 10mM glucose. For the assessment of ketone body metabolism using hyperpolarized 13

| Assessment of ketone body metabolism in the in vivo heart
Male Wistar rats (n = 4, distinct from the ex vivo study animals; body weight,~300 g) were scanned in both the fed and subsequently overnight fasted metabolic states. To induce a fasted metabolic state, rats were kept in a 12-h light/dark cycle and fasted for at least 12 h prior to scanning.
A fed metabolic state was induced by scanning under the same conditions without overnight fasting, and with access to 'standard chow' ad libitum.
Anaesthesia was induced by 2.5-3% isoflurane in oxygen and nitrous oxide (2 L/min oxygen/isoflurane, 0.2 L/min nitrous oxide). Anaesthesia was maintained by means of 2% isoflurane delivered to, and scavenged from, a nose cone during the experiment. A tail vein catheter was placed for intravenous injection of hyperpolarized [1-13 C]acetoacetate. Animals were then placed in a home-built animal handling system. 33 Body temperature was maintained using air heating, and a two-lead electrocardiogram (ECG) for cardiac gating was obtained using leads placed subcutaneously into the upper forelimbs. Blood ketone and glucose concentrations were measured via a commercially available hand-held meter (Accu Chek, Bayer AG, Berlin, Germany) following injection of acetoacetate in a separate cohort of rats (n = 4). 1 H images were acquired using a 72 mm-inner-diameter quadrature birdcage transmit/receive coil (Rapid Biomedical GmbH, Rimpar, Germany). 13 C data were acquired using a 20 mm-diameter 13 C butterfly surface coil for signal transmission and reception. The surface coil was positioned on the anterior chest wall, and a thermally polarized 5 M 13 C urea phantom was used to calibrate the 13 C transmitter power before the 13 C scans. A volume covering a 40-mm slab including the heart was employed for shimming using a three-dimensional gradient echo automated shim routine 34 to reduce the proton linewidth to <1 ppm across the heart. Immediately prior to the injection of the hyperpolarized sample into the rat, an ECGgated 13 C pulse-acquire sequence (nominal TR = 1 s; hard pulse of 15 μs; flip angle, 10°; sweep width, 6000 Hz; 2048 complex points; 60 measurements) was started, and 2 mL of hyperpolarized [1-13 C]sodium acetoacetate solution was injected over 10 s into the rat via the previously placed tail vein catheter. As documented previously, variation in the R-R interval of the anaesthetized rat yielded a slight variation in TR spacing (on the order of 1-10%) which was accurately measured using custom-built hardware. 35

| 13 C data analysis
All acquired 13 C data were summed over a 60-s period from the first spectrum in which hyperpolarized [1-13 C]acetoacetate or [1-13 C]βhydroxybutyrate could be clearly observed, and the summed spectra were fitted using the AMARES algorithm in the jMRUI software package.
The AMARES algorithm was adequately constrained by prior knowledge of the spectral location of the numerous closely spaced resonances to enable their resolution. Fitted metabolite amplitudes were subsequently normalized to the maximum [1-13 C]acetoacetate/[1-13 C]βhydroxybutyrate amplitude to remove any contribution from polarization and injection timing differences between experiments. In addition, for kinetic analyses, a sliding window of length 5 (~5 s) was applied to the reconstructed data to improve the reconstructed SNR. Further information on spectral fitting is provided in Supporting Information.

| Statistical methods
All data are given as mean ± standard error. Mean differences between fed and fasted states were assessed using an unpaired, unequal variance t-test.

| Metabolism of hyperpolarized ketone bodies within the isolated heart
Interconversion between the ketone bodies could be seen for both substrates, whereas infusion of hyperpolarized acetoacetate also allowed for the observation of metabolism into the Krebs cycle intermediates citrate and glutamate, together with the acetyl-CoA buffering compound, acetylcarnitine. The infusion of ketone bodies did not significantly change the left ventricular developed pressure (LVDP), heart rate (HR) or the rate-pressure product (RPP) within the time course of the NMR experiment (shown in Supporting Information Figure S4). For 60 s prior to infusion, the mean ± standard deviation (SD) LVDP, HR Figure S9), and are believed to originate spontaneously during the initial high-temperature phase of the dissolution process used for in vivo experiments on prototype hardware. In separate phantom experiments at approximately 37°C, the apparent T 1 value of the resonance at 182 ppm was seen to be the same as that of [1-13 C] acetoacetate (~30 s), indicating that it is in rapid exchange with acetoacetate ( Figure 3). In vivo, this resonance decays with a different time course, with a prolongation consistent with metabolic production, and has a different relative amplitude to phantom experiments (in which the total integrated impurity to acetoacetate ratio had a maximum value of~0.01), suggesting that the Krebs cycle product [5-13 C]glutamate is observable in addition to this impurity. If glutamate was not produced, but this resonance entirely reflected that of the impurity, we would expect it to have the same kinetic behaviour as the injected [1-13 C]acetoacetate. In phantom experiments, both peaks reach their maximum amplitude at the same time and in the same spectra. In vivo, however, the time to peak of 'glutamate' occurs 2-4 s after the maximum acetoacetate signal, which is consistent with metabolic production.

| DISCUSSION
Ketone body metabolism is altered in diseases affecting the heart, including diabetes and diabetic cardiomyopathy. For example, the sodium-glucose co-transporter-2 inhibitor and novel anti-diabetic drug, Empaglifozin, has been shown to dramatically reduce mortality and morbidity from cardiovascular causes in a large cohort study of patients with type II diabetes mellitus, 40 and is believed to be associated with alterations in ketone body metabolism. 41 A better understanding of the metabolic derangements seen in these diseases would provide a new perspective on potential treatments. This work has demonstrated the potential to hyperpolarize the ketone bodies acetoacetate and β-hydroxybutyrate to a level sufficient to allow the observation of their metabolism in both the isolated perfused rat heart and, in the case of acetoacetate, the in vivo organ. The higher polarization level and longer T 1 value achieved with acetoacetate enabled the observation of its rapid uptake, interconversion with β-hydroxybutyrate and downstream metabolism into the Krebs cycle, as summarized in Figure 5. However, the spectral position of β-hydroxybutyrate, in addition to the lower polarization level and shorter T 1 , meant that the reliable detection of the resonances of citrate and glutamate were not possible.
In the in vivo experiments, we observed significant differences in the metabolism of acetoacetate to acetylcarnitine, 'glutamate' and bicarbonate between the fed and fasted states (Figure 3). The apparent increased rate of acetylcarnitine production following feeding is consistent with the reported role of acetylcarnitine as a store of acetyl moieties should they be abundant in a post-prandial state, into which ketone oxidation is directed. 42 In the fasted state, 'glutamate' levels were higher, which is consistent with an increased flux of ketone bodies into the TCA cycle during fasting. However, further work is required on the synthesis methodology to prevent/remove the impurity that overlaps the glutamate resonance to confirm this finding, as we note that, owing to the presence of the impurity, the relative ratios of other metabolites to 'glutamate' are quantitatively dissimilar to those obtained through other methods. 15 The ratio of β-hydroxybutyrate to acetoacetate was apparently unchanged between the fasted and fed states, reflecting the unchanged appearance rate of exchange into β-hydroxybutyrate; this may reflect on the constancy of the mitochondrial redox state during the relatively short overnight period of fasting undertaken in this experiment. The appearance of 13 C-bicarbonate following the injection of [1-13 C]acetoacetate was observed in both the perfused heart and in vivo. This is presumed to be caused by the spontaneous decarboxylation of acetoacetate to acetone and carbon dioxide. An alternative pathway is the in vivo decarboxylation of acetoacetate catalysed by acetoacetate decarboxylase, which is present within blood. 43,44 The time-to-peak intensity for this resonance was the same as that of acetoacetate (A) (B) (C) FIGURE 3 Representative kinetic time courses following the injection of hyperpolarized [1-13 C]acetoacetate after application of a sliding window averaging scheme, obtained following infusion of the agent into (a) a plastic phantom and in the (b) fasted and (c) fed states in vivo. Nonacetoacetate peaks have been multiplied by a factor of ten. The dotted line present shows the point of maximum acetoacetate signal. The difference in the time-to-peak between injected acetoacetate and the glutamate + impurity peak is 0 s in the phantom experiment, rising to 2-4 s in vivo. We believe that this delay indicates that additional glutamate is produced underneath the impurity resonance. AcAc, acetoacetate; AcAc-H, acetoacetate hydrate; BHB, β-hydroxybutyrate FIGURE 4 Metabolite ratios for β-hydroxybutyrate (β-HB), acetylcarnitine, glutamate and bicarbonate following the administration of hyperpolarized [1-13 C]acetoacetate to the in vivo rat heart. Statistically significant differences (p < 0.05) were observed in 13 C-labelled acetylcarnitine, glutamate + unknown, [3-13 C]acetoacetate and bicarbonate between fed and fasted states itself, which indicates that production is rapid. Given that a delay of several seconds was observed between peak acetoacetate and peak acetylcarnitine and β-hydroxybutyrate, this observation is likely to be consistent with a mechanism not involving transport across the plasma membrane.
Similar to Wang et al., 17  Interestingly, we observed substantial 13 C-bicarbonate production almost exclusively in the fed state, but the in vivo significance of this result is not fully understood, although we note that this result is consistent with a difference in label exchange rates based on equilibrium positions, together with the observation in rodents that breath acetone is significantly increased by fasting or ketogenic diets. 45 The spontaneous rate of decarboxylation would be expected to be independent of metabolic conditions, and proportional only to the (fixed) quantity of injected labelled compound. We therefore propose that increased bicarbonate production may reflect enzymatic blood decarboxylation, which is expected to be exceptionally rapid. 46 In future studies, as well as mechanistically elucidating these changes, we will investigate differences in ketone body metabolism between baseline and diseased states (e.g. in models of type II diabetes), as well as during high cardiac workload. It may also be possible to monitor noninvasively pathological changes in the mitochondrial redox state as indexed by the ratio between β-hydroxybutyrate and acetoacetate.
In this proof-of-principle study, we observed 13 C label exchange representing the metabolism of hyperpolarized [1-13 C]acetoacetate and [1- 13 C]β-hydroxybutyrate, whose labelling positions result in reasonable liquid state polarization and T 1 values. However, the in vivo linewidth (0.5 ppm) of these probes following injection makes it challenging to distinguish some of the downstream products of [1-13 C]acetoacetate and, FIGURE 5 Summary of the metabolic pathways together with label positions believed to be probed by both hyperpolarized [1-13 C]acetoacetate and [1-13 C]β-hydroxybutyrate in both the in vivo and ex vivo rat heart as presented in this work. Compared with the fasted state, acetylcarnitine production was increased, glutamate decreased and bicarbonate increased by feeding as such, other labelling positions may offer experimental advantages. In particular, [3-13 C]acetoacetate has a chemical shift of 210 ppm, upfield of the carbonyl resonances, although we note that flux through to citrate may introduce label exchange into the [1-13 C]position. In addition, the chemical instability of acetoacetate mandates the use of in-house synthesis methods, which we have not fully optimized, leading to the production of two prominent impurity resonances that we believe are acetate and acetoacetate hydrate. Owing to the chemical instability of acetoacetate and the necessarily increasing time delay between batch synthesis and each individual experiment, despite our best efforts, it is likely that the exact impurity profile differs between each experiment. Nevertheless, we have demonstrated the regulation of ketone body metabolism in the rat heart, and have shown that the in vivo regulation of ketone body metabolism can be rapidly measured via dissolution DNP using a simple, accessible synthesis route. It should be noted that the chemical instability of the probe may make future translation to human studies challenging, although, if the compound is stored below 0°C, the rate of spontaneous decarboxylation is believed to be low, in part owing to its mechanism (reproduced in Supporting Information). Future work will continue to optimize this technique to improve the purity and yield of the synthetic route, and increase polarization, and also further explore ketone body metabolism in the pathophysiological heart with hyperpolarized MR, which is of critical importance in numerous disease states.

| CONCLUSIONS
We report on the use of 13 C-labelled acetoacetate and β-hydroxybutyrate as novel hyperpolarized substrates in the study of cardiac metabolism.
Methods for the generation of hyperpolarized [1-13 C]acetoacetate and [1-13 C]β-hydroxybutyrate were described, and their metabolism was investigated in both isolated, perfused rat hearts and in the in vivo rat heart.