Measuring changes in substrate utilization in the myocardium in response to fasting using hyperpolarized [1-13C]butyrate and [1-13C]pyruvate

Cardiac dysfunction is often associated with a shift in substrate preference for ATP production. Hyperpolarized (HP) 13C magnetic resonance spectroscopy (MRS) has the unique ability to detect real-time metabolic changes in vivo due to its high sensitivity and specificity. Here a protocol using HP [1-13C]pyruvate and [1-13C]butyrate is used to measure carbohydrate versus fatty acid metabolism in vivo. Metabolic changes in fed and fasted Sprague Dawley rats (n = 36) were studied at 9.4 T after tail vein injections. Pyruvate and butyrate competed for acetyl-CoA production, as evidenced by significant changes in [13C]bicarbonate (−48%), [1-13C]acetylcarnitine (+113%), and [5-13C]glutamate (−63%), following fasting. Butyrate uptake was unaffected by fasting, as indicated by [1-13C]butyrylcarnitine. Mitochondrial pseudoketogenesis facilitated the labeling of the ketone bodies [1-13C]acetoacetate and [1-13C]β-hydroxybutyryate, without evidence of true ketogenesis. HP [1-13C]acetoacetate was increased in fasting (250%) but decreased during pyruvate co-injection (−82%). Combining HP 13C technology and co-administration of separate imaging agents enables noninvasive and simultaneous monitoring of both fatty acid and carbohydrate oxidation. This protocol illustrates a novel method for assessing metabolic flux through different enzymatic pathways simultaneously and enables mechanistic studies of the changing myocardial energetics often associated with disease.

Scientific RepoRts | 6:25573 | DOI: 10.1038/srep25573 new techniques that can measure substrate competition will serve an important purpose in establishing which model of myocardial metabolism is most accurate.
Plasma substrate concentrations can vary dramatically depending on the physiological state of the body and as a result the myocardium modulates its substrate selection in order to maintain high levels of ATP 8 . For instance, fasting leads to an increase of circulating free fatty acids (FFA) 9 , an acceleration of lipid oxidation and a reduction of glycolysis in peripheral tissues 10 , as well as an elevation of ketone body concentrations, which eventually inhibit carbohydrate and fatty acid oxidation 11,12 . Acetyl-CoA is the metabolite formed at the crossroads between lipid and carbohydrate metabolism and it is located at the entry of the tricarboxylic acid (TCA) cycle (Fig. 1). It is produced from fatty acids via β -oxidation and from carbohydrates through the glycolytic pathway, via the intramitochondrial pyruvate dehydrogenase (PDH) complex. Substrate selection is regulated by many mechanisms, including the intramitochondrial acetyl-CoA to CoA ratio. An increased ratio results in the inhibition of PDH activity whereas a decreased ratio activates PDH 13 . When acetyl groups are abundant they can be stored as acetylcarnitine, a reaction catalyzed by carnitine acetyl transferase (CAT) 8 .
Recent developments in hyperpolarized 13 C magnetic resonance (MR) allow the study of real-time metabolism of 13 C labeled pyruvate, providing a direct measurement of the flux through PDH by the formation of CO 2 and bicarbonate [14][15][16] . Hyperpolarized 13 C pyruvate alone, has been used in promising pre-clinical investigations of myocardial metabolism of the failing heart [17][18][19][20] . The effect of increased fatty acid availability on the metabolism of hyperpolarized pyruvate has been studied in perfused rat heart 14,21 , while hyperpolarized [1-13 C]butyrate was recently proposed as a substrate to study cardiac fatty acid metabolism 22,23 . Metabolism of hyperpolarized [1-13 C] butyrate provides an abundance of metabolic information about its exchange with butyrylCoA via the formation of [1-13 C]butyrylcarnitine, TCA cycle turnover via the formation of [5-13 C]citrate, the 13 C transfer rate between α -ketoglutarate and glutamate through the evolution of the [5-13 C]glutamate signal, the enzymatic activity of CAT via the formation of [1-13 C]acetylcarnitine, and a process referred to as pseudoketogenenesis 24,25 , via the formation of [1-13 C]β -hydroxybutyrate (BHB) and [1-13 C] The considerable advantage of the hyperpolarized 13 C MR method is that substrate selection is monitored by the appearance of specific metabolites produced by pathways leading to acetyl-CoA production and its subsequent entry into the TCA cycle proper. The aim of this study was to measure the change in substrate utilization, in vivo and in real-time, in response to an induced shift in metabolism using the co-administration of hyperpolarized [1-13 C]pyruvate and [1-13 C]butyrate.   (Fig. 2a). The measured metabolite ratios between total lactate, bicarbonate and alanine relative to pyruvate were 0.049 ± 0.010, 0.027 ± 0.012 and 0.027 ± 0.009 respectively (Fig. 3a-c). The detection of both bicarbonate and CO 2 in the fed animals allowed us to determine the myocardial pH. A stable value of 7.3 ± 0.1 was observed and demonstrated that the injections did not significantly disturb pH during the experiment (Supplemental Fig. S1).
Butyrate metabolism was similarly informative. The C1 of butyrate has a relatively long longitudinal relaxation time (T 1 ) compared to longer chain fatty acids and its resonance signal does not overlap with that of [5-13 C]glutamate like [1-13 C]acetate does (Fig. 2b). The following butyrate-derived metabolites were detected in vivo: [5- Table 1). The resonance observed at 176.5 ppm was assigned to that of [1-13 C]butyrylcarnitine based on high resolution 13 C MR of a mixture containing both 13 C labeled acetylcarnitine and butyrylcarnitine (Supplemental Fig. S2). Butyrate β -oxidation produces two units of acetyl-CoA, only one of which is labeled and hyperpolarized. As a fraction of the parent butyrate signal, acetylcarnitine (Fig. 4a) was approximately twice the intensity of the glutamate (Fig. 4b) peaks and four times the intensity of the butyrylcarnitine (Fig. 4c) and acetoacetate peaks (Fig. 4d) and showed nearly identical intensities between the fed and fasted state. The normalized fractions of acetylcarnitine, glutamate, butyrylcarnitine, and acetoacetate relative to butyrate were 0.008 ± 0.001, 0.004 ± 0.001, 0.0012 ± 0.0004, and 0.0012 ± 0.0002 respectively in the fed state (Fig. 4).  and alanine (c) relative to pyruvate where hyperpolarized [1-13 C]pyruvate was either injected separately or co-injected with hyperpolarized [1-13 C]butyrate, in both fed and fasted animals. *P = 0.02 in co-injection compared with single injection groups, § P = 0.03 in fasted groups compared with fed groups, *** P = 0.001 in fasted groups vs. fed groups.

Metabolite
Chemical shift (ppm) [ Substrate competition was clearly evident upon co-injection of the agents. Hyperpolarized 13 C MRS clearly show the time evolution of both hyperpolarized substrates (Fig. 2c). Although there was some spectral overlap, most metabolites could be detected in the co-injection experiments (Fig. 2d). The competition presented by butyrate results in a decrease in the bicarbonate signal, and a significant increase in the lactate to pyruvate ratio. From the perspective of butyrate metabolism, the co-injection of pyruvate produced multiple signs of competition in the fed state. The acetylcarnitine to butyrate ratio ( Fig. 4b), increased upon co-injection, but did not reach statistical significance. The [5-13 C]glutamate to butyrate ratio ( Fig. 4b) decreased significantly upon introduction of pyruvate (0.41 ± 0.06 to 0.08 ± 0.01). Additionally, the co-injection of pyruvate nearly quenched the appearance of acetoacetate derived from butyrate,(0.0012 ± 0.0002 to 0.0002 ± 0.0002) (Fig. 4d).
The fractional contribution of the downstream metabolites to the total detected signal was not significantly different when fasting or in the case of co-injection (Supplemental Fig. S3).

Carbohydrate and Fatty Acid Metabolism in the Fasted State.
Fasting led to significant changes in both the pyruvate and butyrate control experiments and in the substrate selection protocol. Alanine as a fraction of pyruvate signal was largely constant across all injection conditions (Fig. 3c). After an overnight fast, the ratio of bicarbonate to pyruvate reduced significantly in both the single and co-injection groups, and a significant increase in the lactate to pyruvate ratios could be observed (Fig. 3a).
In the butyrate control, a significant increase of the ketone body acetoacetate was observed in both the single and co-injection experiment after fasting (Fig. 4d). The ratio of butyrylcarnitine to butyrate was stable in the fed relative to the fasted cases (Fig. 4c).
Co-injection of pyruvate and butyrate in the fasted state resulted in changes to the downstream metabolites of both metabolic probes compared to the fed animals. The lactate signal rose significantly between the fed and fasted states during co-injection, while the alanine and bicarbonate signals did not show significant changes. From the perspective of butyrate metabolism in the fasted state, the co-injection of pyruvate decreased the glutamate to butyrate ratio (Fig. 4b) and also that of acetoacetate (Fig. 4d). Fasting did not induce a change in the acetylcarnitine to butyrate ratio when comparing the butyrate protocol to the competition protocol (Fig. 4a). Additionally, fasting results in an apparent restoration of the acetoacetate signal even in the presence of pyruvate compared with fed animals (Fig. 4d).

Discussion
During the switch to a fasting profile, ketone concentrations climb in the plasma, as does the availability of fatty acids. Therefore, in addition to a change in enzyme expression in the myocardium, there is a different profile of In vivo metabolite ratios of 13 C labeled acetylcarnitine (a), glutamate (b), butyrylcarnitine (c), and acetoacetate (d) relative to their hyperpolarized metabolic precursor butyrate. Animals were exposed to a single injection of butyrate, or a co-injection of butyrate and pyruvate. *P = 0.01 in co-injection compared with single injection groups, **P = 0.006 in co-injection compared with single injection groups, § P = 0.05 in fasted groups compared with fed groups, ‡ P = 0.09 in fasted groups vs. fed groups. Butyrylcarnitine cannot be directly observed during co-injection due to spectral overlap.
circulating substrates available to the heart. The circulating glucose concentration in the fed animals was measured as 9.0 ± 0.7 mM and in the fasted animals it was 4.5 ± 1.0 mM. These changes in substrate availability will affect the signals derived from the hyperpolarized experiments.
Pyruvate metabolism largely recapitulated previous studies and is summarized in supplemental information. Like pyruvate, butyrate metabolism is exquisitely sensitive to the nutritional state of the animal. The 13 C labeling of butyrylcarnitine was stable in both the fed and fasted cases (Fig. 4c), indicating that butyrate transport, reflected by 13 C labeling of butyrylcarnitine, in the myocardium was not affected by fasting. The acetoacetate resonances were more intense in fasted animals, a direct result of increased ketone body concentrations 10 and the pseudoketogenic effect 27 . The 13 C labeling observed in β -hydroxybutyrate and acetoacetate is a result of the exchange between acetoacetyl-CoA and free, unlabeled acetoacetate in the mitochondria, as opposed to true synthesis from two acetyl-CoA precursor units. These ketone bodies were also detected in a perfused heart model injected with hyperpolarized [1-13 C]butyrate 22 , leading us to believe that they are not an artifact of circulation from the liver. True ketogenesis would result in the 13 C labeling of [3-13 C]acetoacetate (Fig. 1), a resonance (~209 ppm) which was never observed. In previous studies using hyperpolarized acetate, which is not β -oxidized, ketone body labeling in the muscle was entirely absent 26,28-32 , supporting our hypothesis of negligible 13 C label flux from acetyl-CoA to the ketone bodies in the heart. Acetoacetyl-CoA formation is therefore necessary for observation of β -hydroxybutyrate (Fig. 5a).
Interpretation of the glutamate signal intensities in carbon-13 NMR is subject to a long history of research on the glutamate-oxaloacetate transaminase (GOT) and its interplay with the malate-aspartate shuttle (MAS). Glutamate is 13 C labeled by exchange with α -ketoglutarate, a pathway alternatively referred to as V X 33 or F1 34 . The glutamate to acetylcarnitine ratio was lower in the fasted compared to fed state while the acetylcarnitine to butyrate ratio remained unchanged, indicating a decrease in glutamate labeling (Fig. 5a). Based on comparisons relative to the butyrate signal, acetylcarnitine was not significantly changed by fasting (Fig. 4a) and the fluxes maintaining the acetylcarnitine pool acting as storage for excess mitochondrial acetyl-CoA do not seem to be affected on the time scale of the experiment 35 . The glutamate pool size remains unchanged for up to a 48 h fast 8 . Therefore it is likely that the decrease in glutamate signal is not a result of a change in glutamate pool size. Substrate competition from circulating, unlabeled nutrients or a change in the equilibration between α -ketoglutarate and glutamate could both be mechanisms underlying the decreased glutamate signal. Competition between butyrate and pyruvate in the fed animals, compared with a single injection of butyrate or pyruvate, caused a reduction in the observed [ 13 C]bicarbonate signal, an increase in the [1-13 C]lactate signal, a reduction in the [5-13 C]glutamate signal, a nearly absent [1-13 C]acetoacetate signal, and an increase in the [1-13 C] acetylcarnitine signal. All of these observations fit into a detailed narrative of substrate selection between pyruvate and butyrate in the myocardium (Fig. 5b).
The presence of medium or short chain fatty acids can severely restrict the entry of pyruvate into the TCA cycle as well as increase the lactate pool size in perfused hearts 14 . These results are largely duplicated in vivo, with the caveat that the pyruvate and butyrate co-injection did not take place over a period long enough to establish a metabolic steady state. High concentrations of pyruvate can partially overcome PDH inhibition by high fatty acid concentrations 36 . Hence, the production of [ 13 C]bicarbonate was only halved as opposed to abolished when butyrate was also present, as observed in perfused rat heart 22 , and the lactate signal changed only marginally. Nevertheless these results indicate an almost instantaneous adaptation of myocardial metabolism to substrate availability.
These changes were also a direct reflection of butyrate uptake in the myocardium. The increase in [1-13 C] acetylcarnitine and the decrease of the [5-13 C]glutamate and [1-13 C]acetoacetate signals indicate competition from pyruvate for the limited number of free CoA units in the mitochondria (Fig. 5a). Acetylcarnitine serves as a buffer for extra acetyl units that exceed the capacity of the TCA cycle for oxidation and this circulation of acetyl groups is a fast, well equilibrated process [37][38][39] . When pyruvate was co-injected, free CoA was rapidly reacted to acetyl-CoA, and the extra acetyl units derived from butyrate oxidation were stored by reaction with carnitine, resulting in the increased observed acetylcarnitine. The lowering of the [5-13 C]glutamate signal can be accounted for by the same competitive phenomena; [1-13 C]pyruvate cannot produce labeling of glutamate in the myocardium. However, identification of the mechanism(s) underlying the decreased glutamate signal upon co-injection will require further studies due to the considerations regarding glutamate-α -ketoglutarate exchange.
The mitochondrial pyruvate transporter preferentially antiports acetoacetate over any ion except hydroxide 40,41 . Upon co-injection of pyruvate, we hypothesize that the consequent lack of free, unlabeled acetoacetate available for exchange short-circuits the process of pseudoketogenesis. Further experiments, for example blocking the mitochondrial pyruvate transporter, would be required to confirm this hypothesis, but this explanation seems straightforward given the previous work 40 .
In the case of fasting, co-injection resulted in a similar set of results to the fed control, but with some notable differences. Similar to the results with pyruvate only, fasting quenched the appearance of [ 13 C]bicarbonate. Compared to the fed state, the sum of the products of hyperpolarized [1-13 C]butyrate metabolism trended lower (Supplemental Fig. S3). But, the acetylcarnitine signal was the only driver of this difference. Fasting caused a significant increase in the signal of acetoacetate. Given what is known about the increase in concentration of ketone bodies in fasting, the nominal increase in acetoacetate signal is most likely due to restored free acetoacetate availability and increased pseudoketogenesis (Fig. 5d). Further experiments varying the dose of pyruvate should serve to confirm this explanation. The near equivalence of the glutamate signal between the fed and fasted co-injection depends upon both TCA cycle turnover and V x . Interpretation of the data would demand other independent measures of both pathways.
In summary, two separate agents, hyperpolarized [1-13 C]pyruvate and hyperpolarized [1-13 C]butyrate, were used as controls for a new protocol that measures substrate selection by the simultaneous injection of both agents. The substrate competition experiment has been shown capable of measuring carbohydrate versus fatty acid utilization in the in vivo heart, as evidenced by statistically significant changes in the [ 13 C]bicarbonate signal (evidence of fatty acid competition with pyruvate) and in glutamate and acetoacetate signals (evidence of pyruvate displacement of fatty acids for the production of acetyl-CoA). These experiments demonstrate that the chemical selectivity inherent to magnetic resonance, when paired with the sensitivity enhancement of hyperpolarization, can be harnessed to produce a protocol that assesses substrate selection in real time, in vivo, in the functioning heart.
No changes were observed between fed and fasted states in vivo in a recent study using hyperpolarized [1-13 C] butyrate 22 whereas many significant differences were detected in the present study. This can likely be attributed to increased spectral resolution at the higher magnetic field strength of 9.4 T used here. The clinical potential of the use of 13 C hyperpolarized magnetic resonance techniques to probe in vivo metabolism has been recently discussed and the first trials in humans for application in cancer monitoring and diagnosis has already been completed 42 . Applications in cardiology also seem at least as promising, and are expected to affect both preclinical and clinical practice 43 . The protocol suggested here should augment the usefulness of hyperpolarized imaging further.
Biological significance. Pyruvate has been suggested as a treatment for heart failure, where its administration has been shown to increase contractility. Mechanistically, this increase in function was attributed to increased transport of Ca 2+ ions 44 . However, as aforementioned, pyruvate produces more equivalents of ATP per mole of O 2 consumed than any other substrate 6 . Due to the superior sensitivity and time resolution (1 s or even lower) available with hyperpolarized MR, it is possible to observe kinetic processes in vivo that were previously amenable to study only in cell culture models. If the mitochondrial pyruvate carrier indeed antiports acetoacetate and other competing substrates 40,41 , this would alter myocardial substrate selection for acetyl-CoA production. We propose that the beneficial effects of pyruvate administration are not only mediated by Ca 2+ efflux but also by optimization of the myocardial efficiency as promoted by the export of substrates that compete with pyruvate for oxidation. This hypothesis is amenable to testing in the perfused heart and suggests concurrent treatment with pyruvate and PDH kinase inhibitors should maximally increase energy availability in the myocardium, perhaps as an acute treatment for myocardial infarction or failure.
Scientific RepoRts | 6:25573 | DOI: 10.1038/srep25573 Study limitations. Hyperpolarization is limited by the 13 C longitudinal relaxation time (T 1 ) inherent to the molecules used for the studies 45 . While glucose would be a more obvious choice for this protocol, the additional enzymatic steps associated with glycolysis along with the short 13 C T 1 in all carbon positions of the glucose molecule greatly reduce the signal intensity of the metabolites relevant for the study of substrate competition. Our choice of [1-13 C]pyruvate, which has a T 1 at 9.4 T of ~45 s, implies that we are not measuring glycolytic metabolism, except indirectly through exchange with lactate. Lactate could also be used as a substitute for pyruvate, but the T 1 of [1-13 C]lactate is ~25 s in solution, considerably shorter than pyruvate. A similar rationale led us to choose butyrate as the surrogate for fatty acid oxidation. Butyrate is a short chain fatty acid that does not have to be solubilized with albumin, a known source of T 1 relaxation 21 . Binding of a compound, including any long chain fatty acid tested, to albumin is sufficient to destroy the hyperpolarized signal (data not shown). Other water soluble fatty acids like [1-13 C]octanoate could have been used 46 , but the increased number of unlabeled carbons subject to β -oxidation would have further halved the signals associated with fatty acid utilization without providing any information on the activity of fatty acid transporters like carnitine palmitoyl transferase I (CPT-I) 13 . The use of [1-13 C]acetate would ensure a 100% transfer of labeled carbons, and the detection of the TCA cycle intermediate citrate, but it does not follow β -oxidation and the resonance of [1-13 C]acetate overlaps with that of [5-13 C]glutamate 26,30 . The lack of facilitated transport of butyrate as butyrylcarnitine by CPT-I is the primary drawback of butyrate as a molecular imaging agent, as CPT-I is a primary point of regulation for fatty acid oxidation 13 . Despite its limitations, the use of the short chain fatty acid butyrate still provides unprecedented data about substrate competition and β -oxidation, as evidenced by the data presented here. Although CPT-I is bypassed and no information can be obtained related to the long chain fatty acid transport process itself, the insights gained are still expected to be diagnostic of myocardial dysfunction.
Some objections have been raised to the high concentrations of the exogenous agents used for these experiments, with claims that bolus administration can produce flux through pathways due to thermodynamic effects. However, the short T 1 of the hyperpolarized carbons works in the favor of the experiment. Before pool sizes can change noticeably from the challenge of the injected substrates, the experiment is over. From a safety perspective, high concentrations of injected pyruvate have been shown safe in the first human hyperpolarized imaging experiments 42 with injected doses up to 0.10 mmol/kg. We have no reason to believe butyrate would not have a similar safety profile.

Conclusions
The combination of HP pyruvate and butyrate provides unprecedented insight into the selection of carbohydrates and fatty acids for oxidation in the myocardium in real-time, and enables a noninvasive and simultaneous measurement of different enzymatic pathways. The co-injections of hyperpolarized metabolic fuels was sensitive to a simple metabolic perturbation and is thus promising for mechanistic studies of myocardial energetics. The results shown here lead us to hypothesize that injection of a bolus of exogenous pyruvate might increase cardiac function not only due to Ca 2+ modulation but also due to the export of substrates less efficient than pyruvate for ATP production.

Methods
Animals. All studies were performed between 10 am and 2 pm, and all animal experiments were conducted according to federal ethical guidelines. The study was approved by the EPFL Animal Care and Use Committee. Wild-type male Sprague Dawley rats (n = 36, 281 ± 26 g) were anesthetized with 1.5% isoflurane in oxygen. Catheters were placed into the femoral vein for substrate delivery and in the artery to monitor the blood pressure used for cardiac triggering, as previously described 26,30,47 . The respiration rate, cardiac rhythm, and temperature were monitored and maintained stable, and were not affected by the injections. Plasma glucose concentrations were quantified with the glucose oxidase method using two multi-assay analyzers (GW7 Micro-Stat, Analox Instruments, London, UK). Animals were exposed to an injection of hyperpolarized butyrate (n = 12), pyruvate (n = 10) or both (n = 14) and each group was divided to be in a fed or fasted state (Table 2). Six different groups were studied and each animal was injected once. The animals were fasted overnight.
Sample preparation and hyperpolarization protocol. Sodium [1-13 C]butyrate and sodium [1-13 C] pyruvate (Sigma-Aldrich, Buchs, Switzerland), were hyperpolarized for 2.5 h in a 5 T custom-designed DNP polarizer 48,49 . Substrates were prepared separately using TEMPO nitroxyl radicals as previously described 26,29 , and were combined in the polarizer to yield identical concentrations of both 13 C labeled pyruvate and butyrate. Using an automated process 50 the sample was rapidly dissolved and transferred within 2 s following dissolution to a separator/infusion pump, containing 0.6 mL of phosphate buffered saline and heparin, located inside the bore of the MR scanner. Subsequently, 800 μ L of a 70 mM 13 C-labeled hyperpolarized substrate solution was automatically infused into the animal, yielding an average injected substrate dose of 0.19 mmol/kg. Typical polarization levels were 13 ± 2%.
In vivo magnetic resonance imaging and spectroscopy. Measurements were carried out in a horizontal bore 9.4 T magnet (Magnex Scientific, Oxford, UK) with a Direct Drive spectrometer (Agilent, Palo Alto, CA, USA). A custom-made radiofrequency (RF) hybrid probe, consisting of a 10 mm diameter proton surface coil and a pair of 10 mm diameter 13 C surface coils in quadrature mode, was positioned over the chest of the rat for transmission and reception. A hollow glass sphere with a 3 mm inner diameter (Wilmad-LabGlass, NJ, USA) was filled with an aqueous 1 M [1-13 C]glucose solution and used to adjust the radiofrequency (RF) excitation pulse power and set the reference frequency. Acquisition of gradient echo proton images confirmed the correct placement of the coil and was used to determine the voxel used for shimming. Cinematographic images (field of view = 40 × 40 mm 2 ; matrix size: 256 × 256; repetition time T R = 140 ms; echo time T E = 4.5 ms; number of averages = 8; number of frames: 14; slice thickness: 1 mm) were acquired to confirm and set the timing of the cardiac trigger in the end-diastolic phase. The cardiac trigger was typically sent 50 or 60 ms after the maximum observed blood pressure. Cardiac triggered and respiratory gated shimming was performed using the FAST(EST)MAP gradient shimming routine 51 to reduce the localized proton line width in a myocardial voxel of 4 × 5 × 5 mm 3 to 20-30 Hz acquired with stimulated echo localized spectroscopy (STEAM), resulting in a non-localized proton line width of 80-120 Hz. The MR console was triggered to start acquisition at the beginning of the automated injection process. Series of single pulse acquisitions were sequentially recorded using 30° adiabatic RF excitation pulses (BIR-4) 52 , with 1 H decoupling using WALTZ 53 . Free induction decays were acquired with 4129 complex data points over a 20 kHz bandwidth. All acquisitions were cardiac triggered and respiratory gated, resulting in a nominal T R between 3 and 3.5 s. The adiabatic pulse offset and power were calibrated to ensure that the RF excitation angle θ = 30° for all observed metabolites in the entire tissue of interest, and was sufficient to excite the entire range of observable resonances from 100 to 220 ppm. Following in vivo data acquisition, a 200 μ L liquid sample was extracted from the separator/infusion pump to retrospectively determine the precise injected substrate concentrations. Data analysis. The 13 C MR spectra were summed and analyzed with Bayes (Washington University, St. Louis, MO, USA) as previously described and metabolite resonance areas were calculated 26,47 . All metabolite areas were normalized to their hyperpolarized injected precursor to determine specific metabolite ratios.
Statistical Methods. Statistics on the metabolite ratios were computed via two-way unbalanced ANOVA, considering the factors of "nutritional state" and "injected substrates". In case the ANOVA analysis reported significant interaction between the two factors, specific instances were further analyzed via two-tailed student t-tests for unpaired data with equal variance. All data were expressed as mean ± standard error of the mean (SEM).