Labile Photo-Induced Free Radical in α-Ketoglutaric Acid: a Universal Endogenous Polarizing Agent for In Vivo Hyperpolarized ¹³C Magnetic Resonance

Introduction

Metabolic imaging using hyperpolarized ¹³C magnetic resonance spectroscopic imaging (HP ¹³C MRSI) is a unique modality providing real-time in vivo information about cellular metabolism in a non-invasive and non-radiative manner.¹ Most studies have been performed following the injection of HP [1-¹³C]pyruvate, the only substrate that has been translated to humans to date.² Pyruvate not only plays a central role in intermediary metabolism, but it can also be readily polarized by dynamic nuclear polarization (DNP) when in its neat acid form and it has relatively long, and field-insensitive liquid-state ¹³C longitudinal relaxation times, on the order of 1 min for both the C1 and C2 carbons. Nevertheless, because of the rapid loss of hyperpolarization, the technology currently available for clinical applications requires tight synchronization of the polarization and injection of a HP [1-¹³C]pyruvate bolus, making it a challenging procedure.

The polarization procedure can be decoupled from injection and ¹³C MR signal acquisition by replacing the persistent radical used as a polarizing agent by a labile radical that is photo-generated in frozen neat pyruvic acid.³ This radical is not only quenched upon dissolution of the frozen pyruvic acid in water, removing the need for a filtration process and associated quality control required for clinical applications when synthetic radicals are used,⁴ but it can also be quenched by simply warming the frozen pyruvic acid to above 190 K. The latter can be done while maintaining the enhanced ¹³C spin polarization built up via DNP, allowing storage and transport of the polarized ¹³C-pyruvic acid.⁵

Although widely used, HP [1-¹³C]pyruvate yields metabolic data that can be difficult to interpret because of the supraphysiological dose that is injected intravenously. Recently it has been shown that high exogenous pyruvate concentrations inhibit lactate dehydrogenase (LDH) activity in cells,⁶ and that highly concentrated bolus injections of pyruvate in vivo inhibit or saturate pyruvate dehydrogenase (PDH) in the rat liver.⁷ The two alternative ¹³C-substrates that can be injected at higher concentration to probe carbohydrate metabolism without substantially increasing their plasma level are lactate and glucose, due to their high inherent blood concentrations. HP [1-¹³C]lactate has already been used in several preclinical applications,⁸ and shows promise for clinical studies. However, it has so far never been injected and observed at doses low enough to maintain physiological plasma levels. HP [U-²H, U-¹³C]glucose has been used for in vivo preclinical studies at physiological concentrations,⁹ but is a challenging substrate because of the short ¹³C T₁s, on the order of 12 s in vitro and 7 s in vivo. ¹³C-labelled glucose has been hyperpolarized using a photo-generated radical, although the radical precursor used is an exogenous synthetic molecule (trimethylpyruvic acid) with an unknown safety profile for in vivo applications, especially as a large concentration (≥0.7 M) needs to be admixed with glucose for this purpose.¹⁰ Pyruvate could also be used to co-polarize glucose, as was done with butyrate,¹¹ or acetate,¹² but even if not ¹³C-labeled, it would still interfere with the metabolism being investigated. Here we propose the use of a radical photo-generated in the endogenous molecule alpha-ketoglutaric acid (α-KG) as a universal endogenous polarizing agent for HP ¹³C MR. We demonstrate its efficiency by using it to prepare a HP [1-¹³C]lactate solution that is injected at a physiological dose in rats to probe hepatic metabolism in vivo.

Results and Discussion

α-KG exhibits an extraordinarily high radical conversion rate under ultraviolet-visible (UV/Vis) light irradiation at 77 K when admixed with an aqueous lactic acid solution. The radical concentration deduced from the X-band electron spin resonance (ESR) spectrum recorded at 77 K after 100 s of irradiation was 26±2 mM in a frozen 80:20 lactic acid:water (w:w) solution (6 μL vitrified droplet or “bead”) containing a starting concentration of 75 mM of α-KG, corresponding to a radical yield of 35 % (Figure 1 a). When the starting concentration of α-KG was increased to 300 mM, a radical concentration of 150±10 mM was reached after 300 s, corresponding to a 50 % conversion rate. Such a free radical concentration is however significantly greater than the optimal concentration required for DNP, and for the preparation of the samples used for HP ¹³C experiments the irradiation time for 300 mM α-KG was shortened to 30 s in order to photo-generate a radical concentration of 50±5 mM. The observed linewidth of the X-band ESR spectrum was 6.57 mT (Figure 1 b).

To identify photo-generated byproducts, photo-irradiated and non-photo-irradiated frozen 80:20 [1-¹³C]lactic acid:water (w:w) samples, containing 300 mM [1-¹³C]α-KG, were dissolved in water and analyzed by liquid-state ¹³C MR spectroscopy. As expected from analogous experiments performed previously with photo-irradiated frozen keto-acids,^{3, 13} the UV/Vis light induces the decarboxylation of α-KG. ¹³CO₂ was the only detectable breakdown product in the ¹³C MR spectrum of the photo-irradiated sample. Although succinic acid is the other expected byproduct, its ¹³C MR signal could not be resolved because it is unlabeled and moreover its ¹³C resonance will be obscured by the large [1-¹³C]lactic acid peak. However, liquid chromatography-mass spectrometry (LC-MS) analysis of a photo-irradiated sample of 300 mM α-KG in water confirmed the presence of succinic acid (see supplementary information). The concentration of ¹³CO₂ estimated from the ¹³C MR spectrum of a 300 mM [1-¹³C]α-KG solution photo-irradiated for 30 s was 50±10 mM (Figure 2), which corresponds to the maximum solubility of carbon dioxide in water at room temperature and is equivalent, as expected, to the radical concentration determined by X-band ESR at 77 K. In HP ¹³C MR experiments, the concentration of radical breakdown products is further diluted by a factor of 200±30 post dissolution, resulting in a maximum expected concentration of 250 μM for a sample containing 50 mM of photo-induced free radicals.

The performance of the radical photo-induced in α-KG as a polarizing agent for DNP was assessed by measuring the maximum solid-state ¹³C polarization at 7 T and 1.35 K as a function of the microwave frequency in a photo-irradiated frozen 80:20 [1-¹³C]lactic acid:water (w:w) sample containing 300 mM α-KG (Figure 3). Modulation of the microwave irradiation frequency increased the solid-state ¹³C polarization at the optimum polarizing microwave frequency by a factor 5. The ¹³C polarization build-up curve at the optimal microwave frequency (196.778 GHz), with a microwave output power of 25 mW, followed a mono-exponential function with a time constant of 50±5 min (Figure 4 a). The maximum solid-state ¹³C polarization after 3 h of microwave irradiation was 40±10 %.

The HP ¹³C MR [1-¹³C]lactate signal measured at 30 s post dissolution was compared to the thermally polarized ¹³C MR signal and gave a liquid-state ¹³C polarization of 19±2 % (n=5). (Figure 4 b). The room-temperature longitudinal ¹³C relaxation time constant (T₁) at 14.1 T was 36±2 s. Assuming that the ¹³C polarization during the 30 s transfer time decreased with a low-field ¹³C T₁ of 50 s, the calculated ¹³C polarization at the start of the dissolution process was estimated to be 35±5 %. The dissolution is driven by high pressure helium gas and, as a result, the sample degasses during the dissolution process and the residual ¹³CO₂ concentration was undetectable.

A series of ¹³C spectra were recorded using a ¹³C surface coil placed on top of the rat liver region every 2 s following the intravenous injection (tail vein) of a 1 mL bolus of a 42±3 mM [1-¹³C]lactate solution (Figure 5). In addition to the resonance of the injected [1-¹³C]lactate, the following downstream metabolites could be detected (n=2): [1-¹³C]pyruvate (170.7 ppm) and [1-¹³C]pyruvate hydrate (179 ppm), [1-¹³C]alanine (176.7 ppm), [1-¹³C]aspartate (175.2 ppm), [1-¹³C]malate (180.6 ppm), and [¹³C]bicarbonate (160.8 ppm). [2-¹³C]malate may have been present however it would be obscured by the shoulder of the lactate peak at 181.7 ppm. The calculated lactate/pyruvate and alanine/pyruvate ratios were 15±2 and 0.8±0.2 respectively.

Although other keto acids have been proposed previously as free radical precursors for in vivo HP ¹³C MR,^{11, 13, 14} the exceptionally high radical yield together with the excellent safety profile of α-KG makes this molecule attractive for translation to humans. The molecule indeed has been shown to be beneficial when injected i.v. at doses of up to 200 mmol h⁻¹ in patients undergoing heart surgery.¹⁵ The high radical yield allows the starting concentration of α-KG in the sample to be very low; in this study the concentration of α-KG in the 1 mL bolus was less than 1.5 mM. In addition, the high yield decreases the UV irradiation time required to reach the target radical concentration. The slow cell uptake of α-KG¹⁶ means that within the short time frame of a HP ¹³C MR experiment, it is not expected to alter the metabolic response to the molecule of interest, in this case lactic acid. Moreover, α-KG and its breakdown products (CO₂ and succinate) are endogenous molecules which would not be of any concern for injection into humans, especially at doses lower than 5 μmol kg⁻¹ for α-KG and 1 μmol kg⁻¹ for succinic acid and CO₂, as was the case in this study.

The X-band ESR line width of the free radical photo-induced in α-KG is approximately 11 % narrower than that in photo-irradiated pyruvic acid, although it is not as narrow as for some exogenous photo-irradiated keto acids.^{10, 13} This parameter, however, is less relevant for DNP at high fields, especially at 7 T as demonstrated in the present study, where microwave frequency modulation can be used to compensate for a wider microwave spectrum.¹⁷ The unusually large ¹³C polarization enhancement factor of 5 induced by microwave frequency modulation also demonstrates that other properties of the free radical photo-induced in α-KG, such as its spin-lattice relaxation rate, are favorable for DNP at 7 T. The 40±10 % ¹³C polarization at the time of dissolution determined from direct solid-state ¹³C MR measurements, and corroborated by back-calculation from the liquid-state ¹³C polarization measured in vitro, is comparable with the 50 % ¹³C polarization obtained in [1-¹³C]lactic acid with trityl radicals at 5 T and 0.8 K.^8e Assuming a linear dependence of the ¹³C polarization on B₀/T,¹⁸ the two results are indeed equivalent.

The large ¹³C polarization allowed measurement of the real-time metabolism in vivo of a low dose of HP [1-¹³C]lactate (0.12 mmol kg⁻¹), which gives a plasma concentration of approximately 1.8±0.4 mM in a 350 g rat.¹⁹ This represents an increase in the plasma concentration of less than a factor of two as compared to basal plasma concentration, which is on the order of 2 mM in rats.^8a Therefore, the signal intensity ratio between the metabolic product [1-¹³C]pyruvate and the injected [1-¹³C]lactate should be close to the total (labelled and non-labelled) pyruvate/lactate ratio if label exchange is sufficiently rapid. The lactate-to-pyruvate ratio of 15±2 observed in our experiments is similar to a ratio of 8–15 measured in the perfused rat liver.²⁰ The low alanine/pyruvate ratio observed here implies that a large fraction of the labeled pyruvate is used for mitochondrial metabolism.

α-KG was not detected in the in vivo experiments, and it is also unlikely that α-KG has any influence on the outcome of the HP ¹³C measurements despite the fact that the estimated plasma concentration immediately after injection (60 μM) is 2.5-fold larger than the normal physiological concentration.²¹ Given the exceptional yield of the α-KG radical, its concentration could be reduced by up to a factor 4 if needed, bringing it closer to physiological levels. The concentration of succinic acid, a radical breakdown product, in the injected lactate solution was estimated to be less than 3 μM, compared to a normal plasma concentration of 17 μM.²¹

Conclusion

We have demonstrated that α-KG can be used as a free-radical precursor for the preparation of HP ¹³C-labelled molecules, which can be used to probe metabolism in vivo. The high levels of polarization achieved in [1-¹³C]lactate allow injection at physiological concentrations. The high radical yield and safety profile of α-KG and its recombination products following UV/Vis irradiation offer the possibility of translating this method to human studies. An unexplored advantage in the use of α-KG is that the radical is labile and can be quenched in the solid state, opening up the opportunity for polarization storage and transport, as has been demonstrated with other such non-persistent radicals.^{5, 22}

Conflict of interest

Arnaud Comment was employed by General Electric Medical Systems Inc. at the time of manuscript preparation and submission.