Real-time cardiac metabolism assessed with hyperpolarized [1-¹³C]acetate in a large-animal model

Abstract

Dissolution-Dynamic Nuclear Polarization (dissolution-DNP) for Magnetic Resonance (MR) Spectroscopic Imaging has recently emerged as a novel technique for non invasive studies of the metabolic fate of biomolecules in vivo. Since acetate is the most abundant extra- and intra-cellular short-chain fatty acid, we focused on [1-¹³C]acetate as a promising candidate for a chemical probe to study myocardial metabolism of beating heart.

Dissolution-DNP procedure of Na[1-¹³C]acetate for in vivo cardiac applications with 3T MR scanner was optimized in pigs during bolus injection of doses up to 3 mmoles. The Na[1-¹³C]acetate formulation was characterized by a liquid-state polarization of 14.2% and T_1Eff in vivo of 17.6 ± 1.7 s. In vivo Na[1-¹³C]acetate kinetic displayed a bimodal shape: [1-¹³C]acetyl carnitine (AcC) was detected in a slice covering the cardiac volume, and the signal of ¹³C-acetate and ¹³C-AcC was modeled using the total Area Under the Curve (AUC) for kinetic analysis. A good correlation was found between the ratio AUC(AcC)/AUC(acetate) and the apparent kinetic constant of metabolic conversion k_AcC/r₁ from [1-¹³C]acetate to [1-¹³C]AcC. Our study proved the feasibility and limitations of administration of large doses of hyperpolarized [1-¹³C]acetate with dissolution DNP to study by MR spectroscopy the myocardial conversion of [1-¹³C]acetate in [1-¹³C]acetyl-carnitine generated by acetyltransferase in healthy pigs.

1. Introduction

The dissolution-Dynamic Nuclear Polarization (dissolution-DNP) technique [1] of specific metabolic substrates was increasingly exploited for several metabolic studies in vivo by Magnetic Resonance Spectroscopy (MRS) [2–4]. The clinical relevance of its use was recently suggested in the first FDA-Investigational New Drug Study of Hyperpolarized Carbon-13 Metabolic MRI [5,6]. To date, [1-¹³C]pyruvate represents the reference molecule for hyperpolarized ¹³C Magnetic Resonance Spectroscopic Imaging (MRSI) [2, 3] due to its specific properties for dissolution-DNP experiments and its pivotal role in cell metabolism. In particular, the DNP formulation of [1-¹³C]pyruvic acid provides high ¹³C density, a long T₁ relaxation time and excellent self-glassing properties [2, 7].

Besides C-1 and C-2 labeled pyruvate [2, 8–11], several ¹³C-enriched metabolic precursors were studied with dissolution-DNP technique [2, 12–14]. Acetate, a short chain fatty acid (SCFA), was suggested as a novel hyperpolarized substrate for use in DNP enhanced MRSI, to define the pathways of SCFA cardiac metabolism in real time. In fact, fatty acids (FA) are the main myocardial energy substrate [15, 16] of normal heart.

Acetate is taken up by cardiomyocytes and readily forms acetyl coenzyme-A (CoA), which is then transported into mitochondria after conjugation with carnitine by the enzyme carnitine acetyl-transferase [10, 17], and finally oxidized in the tricarboxylic acid (TCA) cycle.

FA and glucose metabolism are competitive processes for the fuel supply to the heart that depends on aerobic vs anaerobic microenvironment, expression of enzymes and transporters or substrate concentration [16]. Hyperpolarized [1-¹³C]acetate could be used to exploit with MR scanner the role of SCFA as a competitor of glucose- and lactate-derived pyruvate in order to provide structure and function information of the heart related to myocardial metabolism under resting and stress conditions. As previously described, [1-¹¹C]acetate was already used as a Positron Emission Tomography (PET) tracer of TCA cycle activity and its efficacy as a marker of myocardial oxidative capacity was clinically proven with this technique [18]. However, the efficacy of hyperpolarized [1-¹³C]acetate as MRSI marker of cardiac substrate selection is still not well proven in a model of human heart.

Even if the dissolution-DNP of [1-¹³C]acetic acid should be the most straightforward approach for the hyperpolarization of acetate, it was not performed due to the inadequate glassy properties of this molecule [19]. Nevertheless, the dissolution-DNP of [1-¹³C]acetic acid could be efficiently carried out with proper neutralization and buffering.

To date, tris(hydroxymethyl)aminomethane (Tris) [1-¹³C]acetate and sodium (Na) [1-¹³C]acetate formulations were proposed for dissolution-DNP applications [20–27], while experiments in vivo were conducted only in small size animals. We previously tested hyperpolarized Tris [1-¹³C]acetate for MRSI in pigs [28]. In the present study, we focused on the optimization of Na[1-¹³C]acetate dissolution-DNP formulation and the implementation of a protocol for the hyperpolarization of large volumes to be injected in healthy pigs during real-time cardiac MRS.

2 Results and Discussion

2.1 Dissolution-DNP measurements

MRS and imaging experiments with hyperpolarized tracers require a dose scale up proportional to the model body weight in order to increase the sensitivity. This represented a challenge in our experimental setting since Na[1-¹³C]acetate displays a known tendency to crystallize in a dose-dependent manner. We obtained a maximum ¹³C concentration of 7.3 M for Na[1-¹³C]acetate that is suitable for studies in large animals. The stability of the compound was most likely favored by glycerol supplementation, which improved the glassing properties of the mixture.

A dose escalation was performed following the same protocol previously described by us [29]. The DNP performance at solid-state of either small or large doses was compared considering the buildup time constant (τ) and the normalized plateau value (Π̄). We did not measure the solid-state polarization percentage of the Na[1-¹³C]acetate formulation, due to the critical assessment of the thermal polarization. However, a direct comparison of the polarization enhancement of the samples could be performed through the Π̄ value, which was proportional to the maximum solid-state polarization achievable with dissolution-DNP. The τ constant was representative of the polarization dynamics and in particular of the speed of the polarization.

As a reference, we estimated the solid-state polarization of the large dose of [1-¹³C]acetate from the one previously measured for [1-¹³C]pyruvate. We obtained a solid-state polarization of approximately 20% by taking into account the Π value and the different amount of ¹³C in each sample.

In addition, the liquid-state polarization percentage (%LSP) and T₁ variations were studied by switching to a larger dose. The %LSP and T₁ decreased respectively from 24 ± 2 % to 14.2 ± 4.0 % and from 58 ± 3 s to 44 ± 4 s from the small (n = 6) to the large dose (n = 8), as shown in Fig. 1. As shown in Figs. 1a and 1b, the solid-state values (τ and Π̄) were comparable for the small and the large doses, while the resulting T₁ and %LSP were significantly different after dissolution (Figs. 1c and 1d). The dissolution was the most critical step moving from the small to the large dose, using the Hypersense dissolution-DNP system. Similar results were also obtained by our group in a previous study with hyperpolarized [1-¹³C]pyruvate [29]. Two main factors could contribute to decrease the liquid state polarization for the large dose compared to the small dose: a) the slower and more critical dissolution process; b) the paramagnetic-induced relaxation given by the higher OX063 final concentration (0.22 ± 0.05 mM for the small dose and 0.56 ± 0.09 mM for the large dose).

Fig. 1.

a) solid state build-up time constants (τ); b) normalized plateau values (Π̄ ); c) T₁ relaxation time; d) liquid state polarization percentage (%LSP), obtained for the hyperpolarized Na[1-¹³C]acetate small and large volumes (mean ± SD). T₁ values were estimated at 1.05 T/ 40 °C from a mono-exponential fitting of the decay curve of the hyperpolarized signal [29].

The latter issue was a consequence of the different scale-up performed at solid- and liquid-state switching to larger dose. In fact, a 10-fold increase of the sample volume (from 60 up to 600 μl) was carried out to perform the solid-state polarization, whereas the dissolution volume was increased by only 4 times (from 5 up to 21 ml). These approaches were performed due to both the limited capacity of the heating circuit of the polarizer and the need to maintain the final ¹³C concentration as high as possible for in vivo injection.

Hence, we investigated how the %LSP and T₁ were affected by different OX063 concentrations (Fig. 2). These measurements were performed at 1.05 T and 40 °C by dissolution-DNP of Na[1-¹³C]acetate samples with different ¹³C enrichments (20%, 50% and 100%): the dissolution volume was chosen according to each sample in order to maintain a constant ¹³C final concentration. The effect produced by different amounts of trityl radical on the liquid-state T₁ (Fig. 2) was quantified resulting in an increased relaxation rate R₁ proportional to the OX063 concentration (estimated relaxivity (r₁) of OX063 on the C-1 of acetate = 9.4 M⁻¹ s⁻¹, Fig. 2a). A relaxometry study at low fields reported a value for r₁ = 4.24 M⁻¹ s⁻¹ at 0.5 T and 37 °C for 80 mM pyruvate solution [30, 31]; the different field strength and in particular the presence of Gd³⁺-chelate in the solution could explain the higher value measured by us.

Fig 2.

a) the liquid-state relaxation rate (R₁) for the hyperpolarized Na[1-¹³C]acetate samples is reported with respect to the OX063 final concentration, after dissolution. The relaxivity (r₁) of OX063 on the C-1 of acetate was calculated from the linear fit of R₁ vs OX063 data (R² = 0.97). The samples at 21 and 25 mM OX063 were dissolved using the same dissolution volume (21 ml); b) the liquid-state polarization percentage (%LSP) for the hyperpolarized Na[1-¹³C]acetate samples is shown with respect to the OX063 final concentration, after dissolution. R₁ values were estimated at 1.05 T/40 °C from the hyperpolarized signal decay, as reported in Fig. 1. 13C% indicates the data obtained after dissolution of Na[1-¹³C]acetate samples with different ¹³C enrichment.

Moreover, the trend of %LSP, estimated for the same dataset, parallels with the trend detected for the liquid-state T₁, showing a decreased %LSP with increasing OX063 concentration (Fig. 2b). The same dataset allowed to evaluate the solid-state hyperpolarization performance of Na[1-¹³C]acetate at different ¹³C concentration (using different ¹³C enrichment). Therefore, as a control, we compared the normalized solid-state polarization values (Π̄) and we found that Π̄ values were not significantly affected by ¹³C concentration in the range 1.5–7.3 M [p > 0.05, p-values determined by two-tailed Student’s t-test; n = 4]. Our findings were consistent with previous data obtained by other groups with different formulations, i.e., [1-¹³C]pyruvate/DMSO/OX063, ¹³C-acetate/ H₂O:ethanol/TEMPO [7, 22]. According to these results, the observed %LSP variation was not correlated to previous differences at solid-state.

For a separate dataset, we performed the dissolution-DNP of Na[1-¹³C]acetate samples with different OX063 concentrations (21 and 25 mM) using the same dissolution volume (21 ml). As shown in Fig.2, the R₁ values obtained in this case were consistent with the linear fit previously reported. According to these data, it is conceivable that the paramagnetic-induced relaxation was the main factor leading to the reduction of the liquid-state polarization for the larger samples. Otherwise, the dissolution process (in terms of different volumes) had only a minor effect in our experimental set-up.

A primary role of electron paramagnetic agents in the relaxation mechanism was also confirmed by NMR experiments performed for [1-¹³C]acetate at 9.4 T, where a significant reduction of the T₁ relaxation time was found for increasing OX063 concentrations. In particular, a value of r₁ = 3.9 M⁻¹ s⁻¹ was found at 9.4 T and 25 °C for the C-1 relaxivity of 20%-enriched Na[1-¹³C]acetate samples at different OX063 final concentrations.

2.2 In vivo experiments and MRS

Large animals are often the model of choice in translational medicine, since they better mimic aspects of human anatomy, physiology and especially metabolism, compared to mice and rats [32]. In particular, pigs are a well-known and characterized model of human cardiovascular system.

The injection of hyperpolarized [1-¹³C]acetate in pigs required the preparation and optimization of a large dose in order to increase both the myocardial bioavailability of [1-¹³C]acetate and the maximum Signal-to-Noise Ratio (SNR) during the MRS session. The dynamic conversion of [1-¹³C]acetate into its metabolite [1-¹³C]acetyl-carnitine ([1-¹³C]AcC) was detectable in swine heart after the intravenous injection of 150 mM of Na[1-¹³C]acetate.

The ¹³C spectroscopic signals of Na[1-¹³C]acetate and of its metabolite [1-¹³C]AcC were detected using a 40-mm axial slice selected through the heart of the animal (Fig. 3a). The [1-¹³C]AcC peak was identified at -8.6 ppm (276 Hz) with respect to Na[1-¹³C]acetate in the ¹³C spectrum (Fig. 3b, c).

Fig. 3.

a) Anatomical localization for MRS studies: a slice was selected through the heart of the pig (thickness = 40 mm); b) dynamic evolution over time of [1-¹³C]acetate and [1-¹³C]AcC spectroscopic signals detected after bolus injection of hyperpolarized 150 mM Na[1-¹³C]acetate; c) the spectra of [1-¹³C]acetate and [1-¹³C]AcC have been summed in the 20–36 s time interval.

The SNR is a critical aspect for hyperpolarized [1-¹³C]acetate in pigs. As shown in Fig. 3, the amplitude of the spectroscopic signal of [1-¹³C]AcC was typically 2 orders of magnitude lower than [1-¹³C]acetate (at the maximum value). The spectra at the maximum signals of [1-¹³C]acetate and [1-¹³C]AcC were selected and the SNR was estimated; a typical ratio of 1:25 was found for the AcC/acetate SNR. Low levels of ¹³C-AcC were also reported by other groups in small animals [21, 25], where the efficacy dose (mmol ¹³C/Kg body weight) was 1.5 to 5.5 times higher (0.123 to 0.44 mmol/Kg) compared to the maximum value administered in this study (0.08 mmol/Kg). Even if the rise of the injected dose could be beneficial to improve the SNR in vivo, the volume of the loading cap and the maximum achievable ¹³C concentration in this formulation limited the larger dose to a maximum of 3 mmol.

The use of different coil configurations, such as a transmit/receive circular coil [33] or a transmit-only birdcage and a receive-only circular coil [34], could improve the SNR of a maximum factor of two or three in the anterior left ventricular wall segments, which are closest to the circular coil (i.e. anterior and antero-lateral segments). However, the gain in sensitivity is still insufficient to allow a CSI mapping of ¹³C-AcC.

The spectroscopic signal of [1-¹³C]acetyl CoA and of other metabolites (i.e., citrate) [27] could not be detected with this experimental approach due to the very low abundance of the converted byproducts and to insufficient spectroscopic resolution of 3T scanner. A different strategy using a single 90° flip-angle pulse was previously applied for the detection of [1-¹³C]acetyl CoA signal in small animals [21]. The lack of [1-¹³C]acetyl CoA could also be attributed to the role of this molecule as an intermediate stage of the ¹³C-acetate metabolic pathway [15, 17]. The metabolic curves, which describe the evolution over time of the spectroscopic signals of Na[1-¹³C]acetate and [1-¹³C]AcC, were extracted (Fig. 4). The typical rise times recorded in the experiments (n = 4) were 9 ± 1 s for [1-¹³C]acetate and 24 ± 3 s for [1-¹³C]AcC.

Fig. 4.

Panel 1: a) graphic representation of the metabolic path of acetate in myocardial cells, where CAT stands for the enzyme Carnitine Acetyl-Transferase, CS is Citrate Synthase and CPT I is the Carnitine PalmitoylTransferase I; FFA stands for Free Fatty Acids, β-oxid for β oxidation and CoASH for the nonesterified form of CoA; b) typical dynamic metabolic curves of [1-¹³C]acetate and [1-¹³C]AcC obtained at 3 T from the spectroscopic signals after bolus injection of hyperpolarized 150mM Na[1-¹³C]acetate in pigs; Panel 2) Blood and myocardial ¹¹C-acetate tissue tracer activity data (symbols) along with model fits (solid line), from Sciacca et al. [35] by courtesy of the Editor.

The dynamic data-set of n = 4 pigs was used to test the kinetics of the injected acetate. As is known, the evolution over time of the spectroscopic signal is determined by both the metabolic conversion and the magnetization loss due to the hyperpolarization decay and to RF excitations. We found a bimodal shape of the [1-¹³C]acetate signal as reported in Fig. 5, underlying a different behavior of acetate compared to similar cardiac studies with hyperpolarized [1-¹³C]pyruvate [11]. In order to obtain the dynamic parameters in vivo, such as the effective T₁ (T_1Eff ), the modeling of the [1-¹³C]acetate curve was performed using a γ-variate [11, 36] (eq.1) and a mono-exponential fitting. The [1-¹³C]acetate T_1Eff was estimated from the mono-exponential fitting of the final part of the metabolic curve (Fig. 5): a value of T_1Eff = 17.6 ± 1.7 s was found (corrected for the TR and flip-angle). The fitting parameters of the γ-variate function, describing the shape of the curve, are reported using box-plots in Fig. 5b. The Relative Standard Deviation (%RSD) of the parameters was calculated in order to provide an index of statistical reliability of the estimates: the %RSD for the T_1Eff was 9.7% and 25% and 12.5% for the γ-variate function parameters a and b, respectively. Taking into account the physiological variability of the animals, our results indicated a good inter-study reproducibility.

Fig. 5.

a) Diagrammatic representation of the γ-variate and mono-exponential fitting of myocardial [1-¹³C]acetate metabolic curves; b) box-plots of the fitting parameters a and b of the γ-variate function used to model the initial part of the [1-¹³C]acetate dynamic curve; c) values of the fitting parameters (mean ± SD).

Positron Emission Tomography (PET) studies in humans were previously conducted using ¹¹C-acetate to evaluate oxygen consumption and myocardial perfusion [18, 35, 37, 38]. The typical time frame of these studies was of the order of 20–30 min, for which a bimodal shape of the time-activity curve of ¹¹C-acetate was found, underlying a compartmentalization of the tracer [18, 37]. However, in a few studies [35, 38, 39] early myocardial uptake of ¹¹C-acetate (60–180 s after i.v. injection) was also investigated and a multi-compartment approach was proposed to analyze the kinetic uptake by myocardial tissue, using a two-compartment model to fit the first 3 min of the time-activity curve [35]. In the present study, we investigated the same time-frame using a supra-physiological concentration of hyperpolarized ¹³C-acetate ( 3–5 times more than the plasmatic concentration in pigs [40]). Despite comparable dynamic profiles, we detected the metabolized fraction of acetate converted into [1-¹³C]AcC (Fig. 4).

According to similar findings reported using PET with ¹¹C-acetate [35, 38, 39], we suggest that the bi-phasic shape of the [1-¹³C]acetate curve was likely due to a compartmentalization of the tracer, where the first phase describes the bolus dynamics and the second phase describes a predominant contribution from myocardial uptake of [1-¹³C]acetate.

In order to better investigate the [1-¹³C]acetate and the [1-¹³C]AcC kinetics, we applied a model based on the total Area Under the Curve (AUC) [25, 41] to the dynamic dataset (considering the second-phase of the acetate curve), which was compared to a standard two-site exchange model [11, see Additional Material].

A good correlation (R² = 0.9974, n = 4) was found between the ratio of the AUC([1-¹³C]AcC)/AUC([1-¹³C]acetate) and the apparent kinetic constant of metabolic conversion from [1-¹³C]acetate into [1-¹³C]AcC, divided by the [1-¹³C]AcC relaxation rate r₁ (Fig. 6). These results suggested that the spectroscopic signal of ¹³C–acetate, detected in the second phase of the metabolic curve, mainly derives from intracellular acetate with a minor contribution from the blood pool. To best of our knowledge, the slice-selective approach applied in our study did not allow the distinction between ¹³C spectroscopic signals from the blood and myocardial tissue. This is a limitation for the full comprehension of the metabolic fate of hyperpolarized ¹³C -acetate.

Fig. 6.

Correlation between the AUC([1-¹³C]AcC)/AUC([1-¹³C]acetate) ratio and the apparent kinetic constant of metabolic conversion k_AcC/r₁ from [1-¹³C]acetate to [1-¹³C]AcC, where r₁ is the [1-¹³C]AcC relaxation rate; the obtained values are reported in the Table.

As a future perspective, the use of spectrally-spatially resolved acquisition sequences, able to detect the temporal evolution of the spectroscopic signal in the different regions of the heart, could clarify the in vivo ¹³C-acetate kinetics, allowing to apply a multi-compartmental model [42,43] or a semi-quantitative model-free approach based on the AUC.

Indeed, the analysis of [1-¹³C]acetate kinetics and [1-¹³C]AcC signal could provide insights into cardiac SCFA metabolism. The comparison of signals from hyperpolarized [1-¹³C]acetate and hyperpolarized [1-¹³C]pyruvate could represent a novel approach to investigate in real time the myocardial substrate selection in vivo.

Our protocol required the implementation of a few changes in the hardware configuration; namely caps to hold large doses and an upgraded design of the dissolution path, to improve the dissolution of larger volumes. With regard to this, the recently developed “clinical” polarizer [5] could certainly contribute to overcome the hardware limitations, due to the possibility to simultaneously polarize up to 4 sterile samples with increased volume (> 500 μl).

3. Conclusions

We demonstrated that the real-time MRS assessment of cardiac metabolism is feasible in pigs using 3T scanner and injectable dose up to 600 μL of Na[1-¹³C]acetate formulation. We found reproducible metabolic profiles displaying a bimodal pattern of the kinetic data. Our findings highlight a compartmentalization of the tracer, in accord with PET studies performed with ¹¹C-acetate. We obtained a semi-quantitative estimation of the acetate kinetic of conversion into AcC using MRS and dissolution-DNP. The limitation of our study arises mainly from the low SNR, given by the use of a volume coil and a limited amount of ¹³C ( 5–6 times lower in pigs) compared to small animal studies. This is still a critical issue to perform CSI experiments in pigs. However, the reported results render Na[1-¹³C]acetate a potential candidate for cardiac ¹³C MRS with large-dose dissolution-DNP for diagnostic use.

4. Experimental

4.1 Sample preparation for dissolution DNP

[1-¹³C]acetate samples were prepared by dissolving Na[1-¹³C]acetate (Cambridge Isotope Laboratories Inc., Andover, MA, USA), trityl radical (OX063, Oxford Instr. Ltd, Abingdon, UK) and Gd³⁺-chelate (Dotarem, Guebert, Roissy CdG Cedex, France) in a 60:40 w:w ultrapure (mQ, Millipore, Billerica, MA, USA) water/glycerol mixture. The mixture was sonicated at 60 °C until dissolution. The final concentration values were Na[¹³C-acetate] = 7.3 M, [OX063] = 25 mM, [Gd³⁺-chelate] = 1.4 mM; Na[1-¹³C]acetate small doses (up to 60 μL) and large doses (up to 600 μL) were prepared as previously described and subsequently hyperpolarized. Samples with different ¹³C isotopic enrichment (20%, 50%, 100%, corresponding respectively to 1.5 M, 3.6 M and 7.3 M ¹³C concentration in the sample) and fixed OX063 concentration (25 mM) were also prepared. In a separate session of the study, samples with 7.3 M Na[¹³C-acetate] and different OX063 concentration (21 and 25 mM) were formulated.

4.2 Dissolution-DNP measurements

DNP was performed using a commercial system (Hypersense®, Oxford Instruments plc) slightly modified with an optimized design of the nozzle and using caps suitable for holding up to 800 μL, to allow the polarization and rapid dissolution of increased sample volumes. The samples were polarized at 3.35 T and 1.4 K for about 50 min. The maximum signal, normalized by the number of ¹³C spins in the sample, (Π̄), and the build-up time constants (τ), were estimated by fitting the buildup curves in Matlab (The Mathworks, Inc., Natick, MA, USA).

The dissolution of the hyperpolarized Na[1-¹³C]acetate samples was performed using different volumes of buffer (40 mM Trizma® Pre-set crystals pH 7.6, Sigma-Aldrich [St. Louis, MI, USA], 0.27 mM EDTA, Sigma-Aldrich, and 50 mM NaCl), according to the sample amount. The final volumes were 5 mL for the small dose (up to 60 μL) and 21 mL for the large dose (up to 600 μL). In the latter case 10 mL were placed in the solvent container and 11 mL were added in the external receiving vessel, as described in our previous work [29]. The final pH was 7.6 ± 0.2 and temperature 37 ± 2 °C. The large dose was close to isotonic.

1 mL of the hyperpolarized solution was transferred to a 1.05 T NMR analyzer (Minispec MQ, Bruker BioSpin GmbH, Germany) for measurement of liquid-state polarization (%LSP) and T₁, as described in [29]. The intensity of the hyperpolarized signal was measured at 40 °C, every 5 s for 200 s, using a 5° flip-angle. The T₁ was obtained from the mono-exponential fit of the hyperpolarization decay curve, corrected for the flip-angle and TR [44]; the fitting was performed in Matlab.

Dissolution-DNP of the Na[1-¹³C]acetate samples with different ¹³C enrichments was performed to study the %LSP and T₁ dependence on OX063 concentration (n = 4): a constant ¹³C final concentration was provided by selecting an appropriate dissolution volume for each sample. The relaxivity value (r₁) was estimated at 1.05 T and 40 °C from the linear fit of the relaxation rate (R₁), measured at each OX063 concentration value. R₁ was calculated as T₁⁻¹, measured at 1.05 T and 40 °C as previously reported.

C-1 relaxivity was also determined at 9.4 T (VnmrS, Agilent Technologies, Santa Clara, CA, USA) for 20% enriched Na[1-¹³C]acetate samples with increased trityl radical concentration (0.44 mM, 2.5 mM and 5 mM respectively, ¹³C-acetate final concentration 90 mM) in a 40 mM Trizma® (Sigma-Aldrich), 0.27 mM EDTA (Sigma-Aldrich) buffer. T₁ was determined from a mono-exponential fit using an Inversion Recovery (IR) sequence; r₁ was thus estimated at 9.4 T and 25 °C from the linear fit of the relaxation rate (R₁) as a function of the OX063 concentration. A Shigemi NMR tube was employed for the measurements at higher temperatures to minimize any convection effects.

4.3 Animal protocol

We used four male healthy farm pigs (38–40 kg, bw). The animals were first sedated with a cocktail of tiletamine hydrochloride and zolazepam hydrochloride (8 mg/kg i.m.) and pre-medicated with atropine sulfate (0.1 mg/kg) [11]. A polyethylene catheter was inserted in the marginal ear vein for saline and drug infusion. Each experiment was performed after overnight fasting in lightly sedated animals with continuous infusion of propofol 1% (1 mg/kg/h) at spontaneous breathing. A total of 600 μL of Na[1-¹³C]acetate sample was polarized and dissolved in 21 mL of buffer solution, using the same approach previously reported for the dissolution of large doses of [1-¹³C]pyruvate [29]. The hyperpolarized [1-¹³C]acetate solution (150 mM acetate/20ml 0.08 mmol/Kg) was manually injected into the right ear vein over 10 s. Proton MRI and MRS were performed using a GE Excite HDX clinical 3T scanner (GE Healthcare, USA) and a ¹³C quadrature birdcage coil (Rapid Biomedical, Rimpar, Germany) while monitoring heart rate, rhythm and mean arterial pressure using an MR-compatible system (SA Instruments Inc., Stony Brook, NY, USA). The experimental protocol was approved by the Animal Care Committee of the Italian Ministry of Health and was in accordance with the Italian law (DL-116, Jan. 27, 1992), which is in compliance with the National Institutes of Health publication Guide for the Care and Use of Laboratory Animals.

The ¹³C spectroscopic signals, detected in a 40 mm thick axial slice placed over the heart of the animal, were acquired from the start of the injection, every 2 s for 120 s, using a (slice-selective) pulse-and-acquire sequence. A soft pulse excitation (bandwidth 2200 Hz, 2048 pts, 10° flip-angle), with the center frequency adjusted to the acetate frequency, was employed for the acquisition. An anatomical reference image was acquired before the injection of the hyperpolarized compound using a cardiac-gated breath-held SSFP sequence (FOV = 35 cm, FA = 45°, TE/TR = 1.71ms/3.849ms).

4.4 Post-processing

Peaks relative to Na[1-¹³C]acetate and [1-¹³C]acetyl-carnitine ([1-¹³C]AcC) were identified using the AMARES algorithm as implemented in the jMRUI software package 3.0 [45]. Initial conditions for the peak frequencies and line widths were provided by means of prior knowledge (Gaussian line shape, line width = 0.0–20.0 Hz, [1-¹³C]acetate frequency = 0.0 Hz, [1-¹³C]AcC = 276 Hz).

Metabolic curves were produced by plotting the Na[1-¹³C]acetate and [1-¹³C]AcC signals as a function of time, in order to display the evolution over time of the signal of different metabolites. Fitting of the acetate curves was carried out in Matlab, using a mono-exponential and a γ-variate function, according to the following equation [36]:

$$S(t)=k{(t-t_0)}^a{e}^{(t-t_0)/b}$$ Eq.1

where a,b,k are fitting parameters describing the shape of the curve. t₀ is the arrival time of the bolus and was set to 0 during the fitting process, assuming that the start of acquisition coincided with the arrival of the hyperpolarized compound to the region of interest in the myocardium. The fitting using the γ-variate was performed after correction for the T_1Eff decay, estimated from the mono-exponential fitting.

A two-site exchange model based on modified Bloch equations was used to fit the dynamic metabolic curves of [1-¹³C]acetate and [1-¹³C]AcC for the estimation of the apparent kinetic constant of metabolic conversion from [1-¹³C]acetate to [1-¹³C]AcC. An alternative method based on the total AUC of the [1-¹³C]acetate and [1-¹³C]AcC [41] was also applied and compared to the two-compartments model. The analysis was performed in Matlab, details of the two methods are reported in the Additional Material section.

Abbreviations

dissolution-DNP

dissolution Dynamic Nuclear Polarization

SCFA

short-chain-fatty-acids

FA

fatty-acids

CoA

coenzyme A

AcC

acetyl-carnitine

TCA

tricarboxylic acid

EDTA

Ethylenediaminetetraacetic acid

SSFP

single slice free precession

DMSO

dimethyl-sulfoxide

%LSP

liquid-state polarization percentage

Tris

Tris(hydroxymethyl)aminomethane