¹³C-Labeled Diethyl Ketoglutarate Derivatives as Hyperpolarized Probes of 2-Ketoglutarate Dehydrogenase Activity

Abstract

The TCA cycle is a central metabolic pathway for energy production and biosynthesis. A major control point of metabolic flux through the cycle is the decarboxylation of 2-ketoglutarate by the TCA cycle enzyme 2-ketoglutarate dehydrogenase (2-KGDH). In this project, we developed ¹³C labeled 2-ketoglutarate derivatives to monitor 2-KGDH activity in vivo. ¹³C NMR analysis of liver extracts revealed that uniformly ¹³C labeled 2-ketogutarate, in its cell permeable ester form, was rapidly taken up and hydrolyzed in liver and underwent extensive metabolism to produce labeled glutamate, succinate, lactate and other metabolites. Diethyl [1,2-¹³C₂]-2-ketoglutarate was successfully polarized by dynamic nuclear polarization and within seconds after injection into rats, the probe produced hyperpolarized [¹³C]bicarbonate in the liver reflecting flux through the TCA cycle. These experiments demonstrate that this tracer offers the possibility of directly monitoring flux through 2-KGDH in vivo.

Graphical Abstract

The 2-ketoglutarate dehydrogenase catalysed decarboxylation of 2-ketoglutarate is an important control point of metabolic flux in the TCA cycle. Here, we demonstrate that hyperpolarized diethyl [1,2-¹³C₂]-2-ketoglutarate is rapidly taken up by the liver and hydrolyzed to the free acid, which enters the TCA cycle and undergoes decarboxylation to produce hyperpolarized bicarbonate. These data indicate that this hyperpolarized tracer could be used to monitor flux through 2-ketoglutarate dehydrogenase in vivo.

The mitochondrial tricarboxylic acid (TCA) cycle is a fundamentally important metabolic pathway in all aerobic organisms.^[1] A dysfunctional TCA cycle has been implicated in a number of diseases including cancer, diabetes, metabolic syndrome, and fatty liver disease. ^[2] A major control point of metabolic flux through the TCA cycle is the decarboxylation of 2-ketoglutarate to produce CO₂ and succinyl-CoA catalyzed by 2-ketoglutarate dehydrogenase (2-KGDH). The activity of this enzyme is regulated by both the energy and redox state of the cell.^[3] Yet, there is no clinically translatable imaging method for the real time monitoring of this enzyme in vivo. While the combination of ¹³C-enriched tracers and NMR spectroscopy can provide valuable information about metabolic fluxes through various pathways intersecting in the TCA cycle,^[4] the inherent low sensitivity of ¹³C NMR limits such studies to spectroscopic measurements that require lengthy signal averaging. Hyperpolarized (HP) ¹³C magnetic resonance imaging (MRI) and spectroscopy (MRS) overcomes this problem by dramatically increasing the sensitivity of ¹³C (over 5 orders of magnitude) thereby allowing detection of individual tissue metabolites in vivo in real time.^[5] The TCA cycle is fed by acetyl-CoA derived from carbohydrates (via pyruvate dehydrogenase, PDH), fatty acids (via β-oxidation) and ketones. Production of HP-bicarbonate from HP-[1-¹³C]pyruvate is often used as a measure of TCA cycle activity^[6] but this is problematical for a two reasons; 1) PDH is not part of the TCA cycle itself so other sources of acetyl-CoA can contribute to TCA flux without involving pyruvate.^[7] Second, HP-[1-¹³C]pyruvate can produce HP-[¹³C]bicarbonate by other pathways.^[6b] HP-[2-¹³C]pyruvate, [1-¹³C]acetate and [3-¹³C]acetoacetate have also been used to assess TCA cycle activity because the ¹³C label at C1 of acetyl-CoA enters the TCA cycle and appears in metabolites such as [1‐¹³C]citrate and [5-¹³C]-2-ketoglutarate/[5-¹³C]glutamate.^[8] Nevertheless, these probes, like [1-¹³C]pyruvate, enter the cycle as only one source of acetyl-CoA^{[8e, 9]} so may not fully reflect TCA cycle flux. In addition, the detection of some downstream metabolites can be ambiguous due to overlapping peaks (e.g. [1-¹³C]acetate, [5-¹³C]glutamate and [5-¹³C]-2-ketoglutarate all resonate at around 182 ppm).^{[8b, 8d, 8e]}

To avoid these limitations, our goal was to develop a ¹³C-enriched probe that reports directly and unambiguously TCA cycle flux regardless of the source of acetyl-CoA. 2-Ketoglutarate satisfies this requirement. However, 2-ketoglutarate is a dianion at physiological pH and cannot be transported by monocarboxylate transporters (MCTs). Slow uptake of 2-ketoglutarate into human fibroblast cells was shown to occur slowly by unmediated diffusion.^{[10], [11]} However, cancer cells seem to take up ketoglutarate more rapidly and HP-[1-¹³C]-2-ketoglutarate was successfully used in vivo to monitor mutant isocitrate dehydrogenase activity in glioma via the production of HP-[1-¹³C]-2-hydroxyglutarate.^[12] One approach for unlimited diffusion of molecules across cell membranes is to use uncharged esters of carboxylic acids for delivery followed by rapid hydrolysis by intracellular esterases to quickly liberate the free acids.^[13] While this approach has been successfully used for intracellular delivery of HP-[1-¹³C]pyruvate, it remains unknown whether HP dicarboxylate derivatives can be delivered to cells this way.^[14]

In this work, we used dissolution dynamic nuclear polarization (DNP) and ¹³C NMR as tools to follow the metabolic fate of ¹³C enriched metabolites in vivo. A number of factors influenced the tracer design. First, the HP spin state decays to thermodynamic equilibrium by spin lattice relaxation (T₁) so long T₁ values are essential. The 2-keto-carboxyl functional group present in both pyruvate and ketoglutarate (Figure S2), is of special interest because of its great biochemical significance and the exceptionally long T₁ values of these ¹³C spins.^[15] Second, for in vivo HP-¹³C MRS studies, extensive homonuclear coupling networks in the downstream products are undesirable because ¹³C-¹³C coupling spins result in reduced resonance intensities and hence lower signal-to-noise ratio. The ¹³C chemical shift of bicarbonate is well-separated from other carbonyl carbon frequencies and can conveniently be used to follow biochemical processes that involve decarboxylation. Thus, for our purpose the logical place for the ¹³C label would be at the C1 carbon as this is lost as CO₂ in the 2-ketoglutarate dehydrogenase catalyzed reaction. However, it occurred to us that additional labeling at C2 would allow us to monitor not only the decarboxylation, but also the metabolism beyond the ketoglutarate dehydrogenase step. For this reason, the doubly labeled [1,2-¹³C₂]-2-ketoglutarate was targeted for HP-¹³C studies while the uniformly labeled derivative was selected for non-HP steady-state ¹³C isotopomer experiments. Here, we report the synthesis of these labeled 2-ketoglutaric acid ester derivatives, evaluate the in vivo fate of labeled ketoglutarate, and demonstrate that decarboxylation of 2-ketoglutarate can be detected in vivo by ¹³C MRS after the injection of HP-diethyl [1,2-¹³C₂]-2-ketoglutarate.

Both [1,2-¹³C₂]- and [U-¹³C₅]-2-ketoglutaric acid were synthesized in reasonable yields starting from commercially available labeled oxalic and succinic acid by Claisen condensation, a frequently used synthetic method for 2-ketocarboxylic acid esters, including 2-ketoglutarate derivatives.^[16] The labeled 2-ketoglutaric acids were converted to the ethyl ester derivatives by esterification with acetyl chloride in ethanol (Scheme S1).

Metabolism of diethyl [U-¹³C₅]-2-ketoglutarate was first examined under steady-state perfusion experiments in isolated fasted mouse livers (n=2). Healthy mouse livers were isolated and perfused with 2 mM diethyl [U-¹³C₅]-2-ketoglutarate plus 0.2 mM unlabeled octanoate to provide excess acetyl-CoA units for 30 minutes with continuous oxygen consumption monitoring. The livers were freeze-clamped and extracted using perchloric acid. A ¹³C NMR spectrum of one liver extract (Figure 1 and Figure S3) showed the presence of glutamate, succinate and lactate among other metabolites.

Figure 1.

The ¹³C NMR spectrum of isolated mouse liver extract after perfusion with 2 mM diethyl [U-¹³C₅]-2-ketoglutarate. The scheme above the spectra shows how the ¹³C atoms (red) of [U-¹³C₅]-2-ketoglutarate move through the cycle.

The presence of labeled lactate demonstrates that some [U-¹³C₅]-2-ketoglutarate was converted to succinate and other 4-carbon TCA cycle intermediates which in turn were converted to labeled pyruvate either via PEPCK and pyruvate kinase (PK) (oxaloacetate → phosphoenolpyruvate → pyruvate) or the malic enzyme (ME), (malate → pyruvate) (Figure S1). Together, these pathways are referred to as pyruvate cycling pathways.^[17] Glutamate labeling however can arise by direct conversion of [U-¹³C₅]-2-ketoglutarate to [U-¹³C₅]glutamate by glutamate dehydrogenase or by transamination but other isotopomers of glutamate can also be formed by reentry of labeled pyruvate arising from pyruvate recycling back into the TCA cycle.

The ¹³C-¹³C homonuclear coupling patterns in glutamate, succinate, and lactate provide insights into the pathways involved in metabolism of [U-¹³C₅]-2-ketoglutarate.^[18] The C3 resonance of [U-¹³C₅] glutamate appears as a triplet given that J_C2–C3 and J_C3–C4 are nearly identical. Any [U-¹³C₅]-2-ketoglutarate remaining in the TCA cycle would lose the C1 carbon as CO₂ at the 2-KGDH step while the four carbon intermediates pass through the cycle to produce [1,2,3-¹³C₃]-2-ketoglutarate upon condensation with unlabeled acetyl-CoA. This gives rise to the distinct doublet in the C3 resonance of [1,2,3-¹³C₃]glutamate. The C2 resonance of uniformly labeled succinate appears as non-first order multiplet at 34.8 ppm.^[19] In a similar way, useful information about the metabolic fate of three carbon intermediates can be garnered from the multiplet pattern in lactate. Here, the methyl resonance at around 21 ppm appeared largely as a doublet while the C2 resonance at 69.0–69.6 ppm consisted of a doublet-of-doublets characteristic of [U-¹³C₃]lactate and a doublet characteristic of [2,3-¹³C₂]lactate. The latter isotopomer can only be produced by cycling of [U-¹³C₃]pyruvate from [U-¹³C₄]oxaloacetate back into oxaloacetate via pyruvate carboxylase, backward scrambling through malate and fumarate before leaving the cycle once again via PEPCK plus PK or ME to produce [2,3-¹³C₂]pyruvate. This labeling pattern is a characteristic of tissues with active gluconeogenic activity.^[6b]

While the perfusion experiments clearly demonstrate that the tracer is taken up and metabolized in the TCA cycle within 30 minutes, it does not indicate whether the compound is metabolized on a time scale relevant for HP-¹³C studies (typically a HP ¹³C signal is only detectable for ~2–3 minutes).^[15] To examine this question further, [U-¹³C₅]-2-ketoglutarate diethylester was administered to healthy, fasted Wistar rats (n=2) by tail vein injection (4 mL, 40 mM). After 3 minutes, the animals were sacrificed, the livers quickly isolated, freeze-clamped, and extracted for NMR analysis. The ¹³C NMR spectrum of one such liver extract is shown in Figure 2 and Figure S4.

Figure 2.

¹³C NMR spectrum of a liver tissue extract after infusion of diethyl [U-¹³C₅]-2-ketoglutarate for only 3 minutes.

Here again, the ¹³C-¹³C coupling patterns in glutamate, succinate, and glucose clearly show that the ester was rapidly taken up by liver, hydrolyzed to [U-¹³C₅]-2-ketoglutarate by liver esterases, and metabolized through multiple metabolic steps. The multiplet pattern observed in glutamate indicates the probe entered the TCA cycle and underwent at least two turns within 3 minutes. The observation of labeled glucose is exciting because this suggests one could potentially also image gluconeogenesis with this probe.

Next, samples of diethyl [1,2-¹³C₂]-2-ketoglutarate were polarized in DMSO/glycerol using a HyperSense DNP polarizer and OX063 trityl radical as the polarizing agent (Figure 3). The achieved ¹³C polarization levels (6% for C1 and 5% for C2) were low likely due to the limited solubility of the ester in water. T₁ values of 33 s and 21 s were measured for the C1 and C2 carbons, respectively, at 9.4 T. Polarizations were also performed using a SPINlab clinical polarizer in combination with a 3 T clinical scanner. At this field, the T₁ values were slightly longer (39 s and 23 s, respectively) likely reflecting a relaxation contribution due to chemical shift anisotropy in this probe.

Figure 3.

¹³C NMR spectrum of HP-diethyl [1,2-¹³C₂]-2-ketoglutarate recorded in water at pH 7 and the decay of ¹³C magnetization at 9.4 T (insert).

It is worth noting that this compound exists as an equilibrium mixture of the 2-keto and the hydrate (2-gem-diol).forms. This phenomenon is well-documented for 2-ketoacids and their esters.^[20] The interconversion of the two species occurs with a half time of a few seconds, slow on the NMR timescale but rapid with respect to metabolism.^[20a]

For in vivo studies, diethyl [1,2-¹³C₂]-2-ketoglutarate was polarized using the SPINlab clinical polarizer and administered via tail vein injection (4 mL, 40 mM) to fasted rats (n=3). Dynamic ¹³C spectra were collected at 3 T using a ¹³C surface coil placed over the liver. Within a few seconds, the signal of HP-[¹³C]bicarbonate appeared at 161 ppm and reached maximum intensity ~18 s after injection and about 6 s after the C1 signal of 2-ketoglutarate (~171 ppm) had reached maximum intensity (Figure 5). The C2 resonance of [1,2-¹³C₂]-2-ketoglutarate was also observed as a weak signal near 205 ppm (Figure S5). The very low intensity of the C2 signal is surprising and may be due to a combination of the shorter T₁ value and the hydration equilibrium. No peaks corresponding to the diester were observed. While the absence of diester peaks in the in vivo spectra may indicate that the ester cleavage is very facile, it does not rule out the presence of monoesterified intermediates. The C1 chemical shift of the C5 monoesterified [1,2-¹³C₂]-2-ketoglutarate is likely very similar to that of the C1 carbon of [1,2-¹³C₂]-2-ketoglutarate and may overlap with it. There were two minor peaks detected in the 165 – 170 ppm region, however, these could not be assigned to TCA cycle intermediates unequivocally. It should also be pointed out that the [1,2-¹³C₂] labeling pattern may not offer significant advantages over the [1-¹³C] derivative for metabolic studies because no downstream metabolites containing the C2 carbon were identified in the in vivo HP-spectra. However, the 1,2-labeled derivative is synthetically more easily available than [1-¹³C]-2-ketoglutarate. As expected, no evidence of metabolism of the unesterified [1,2-¹³C₂]-2-ketoglutaratic acid was observed in isolated perfused livers (Figure S6). This observation is in agreement with previous literature reports and highlights the usefulness of esterification to facilitate cellular uptake of hyperpolarized probes.

In conclusion, we have synthesized diethyl [1,2-¹³C₂]-2-ketoglutarate and diethyl [U-¹³C₅]-2-ketoglutarate. The metabolism of the uniformly labeled diester derivative was tested in isolated, perfused livers over 30 minutes and in vivo on a 3 min timescale. High-resolution ¹³C NMR spectra of liver extracts demonstrated that the tracer was rapidly taken up and hydrolyzed in liver and underwent extensive metabolism to produce labeled glutamate, succinate, glucose and other intermediates of the TCA cycle within 3 min. Diethyl [1,2-¹³C₂]-2-ketoglutarate was polarized using two commercially available polarizers under standard DNP conditions and was found to have reasonably long T₁ values (>30 s at 3 T). Successful in vivo detection of HP-[¹³C]bicarbonate within seconds after injection of HP-diethyl [1,2-¹³C₂]-2-ketoglutarate in rat liver indicates that this tracer could be used to monitor flux through 2-ketoglutarate dehydrogenase, an important control point in the TCA cycle. Evaluation of HP-¹³C labeled ketoglutarate esters in various disease models is underway and will be published separately.

Figure 4.

Detection of HP-[¹³C]bicarbonate at 3 T in the liver of a fasted male Wistar rat after the injection of HP-[1,2-¹³C₂]-2-ketoglutarate (40 mM, 4 mL). Top: a stackplot of HP-¹³C spectra in the liver. Bottom: Time course of ¹³C signal intensities (○, 2-ketoglutatare C1; , bicarbonate).