Rapid ¹³C Hyperpolarization of the TCA-Cycle Intermediate α-Ketoglutarate via SABRE-SHEATH

Abstract

α-ketoglutarate is a key biomolecule involved in a number of metabolic pathways—most notably the TCA cycle. Abnormal α-ketoglutarate metabolism has also been linked with cancer. Here, isotopic labeling was employed to synthesize [1-¹³C,5-¹²C,D₄]α-ketoglutarate with the future goal of utilizing its [1-¹³C]-hyperpolarized state for real-time metabolic imaging of α-ketoglutarate analyte and its downstream metabolites in vivo. The Signal Amplification By Reversible Exchange in Shield Enables Alignment Transfer to Heteronuclei (SABRE-SHEATH) hyperpolarization technique was used to create 9.7% [1-¹³C] polarization in 1 minute in this isotopologue. The efficient ¹³C hyperpolarization, which utilizes parahydrogen as the source of nuclear spin order, is also supported by favorable relaxation dynamics at 0.4 μT field (the optimal polarization-transfer field): The exponential ¹³C polarization buildup constant T_b is 11.0±0.4 s, versus the ¹³C polarization decay constant T₁ of 18.5±0.7 s. An even higher ¹³C polarization value of 17.3% was achieved using natural-abundance α-ketoglutarate disodium salt, with overall similar relaxation dynamics at 0.4 μT field, indicating that substrate deuteration leads only to a slight increase (~1.2-fold) in the relaxation rates for ¹³C nuclei separated by three chemical bonds. Instead, the gain in polarization (natural abundance versus [1-¹³C]-labeled) is rationalized through the smaller heat capacity of the “spin bath” comprising available ¹³C spins that must be hyperpolarized by the same number of parahydrogen present in each sample—in line with previous ¹⁵N SABRE-SHEATH studies. Remarkably, the C-2 carbon was not hyperpolarized in both α-ketoglutarate isotopologues studied; this observation is in sharp contrast with previously reported SABRE-SHEATH pyruvate studies, indicating that the catalyst-binding dynamics of C-2 in α-ketoglutarate differ from that in pyruvate. We also demonstrate that ¹³C spectroscopic characterization of α-ketoglutarate and pyruvate analytes can be performed at natural ¹³C abundance with estimated detection limit of 80 micromolar concentration×*%P_13C. All in all, the fundamental studies reported here enable a wide range of research communities with a new hyperpolarized contrast agent potentially useful for metabolic imaging of brain function, cancer, and other metabolically challenged diseases.

Graphical Abstract

The detection sensitivity of NMR techniques is directly proportional to the degree of nuclear spin alignment of the samples with the applied static magnetic field, i.e., the nuclear spin polarization (P).¹ Because equilibrium P is very low (≤10⁻⁵) even in high-field MRI scanners (e.g., 3 T), MRI is regarded as a low-sensitivity technique in the context of real-time in vivo metabolic studies.² The advent of non-equilibrium approaches to enhance P far beyond thermal equilibrium all the way to the order unity allows boosting NMR sensitivity by 4+ orders of magnitude (also termed hyperpolarization), thereby enabling in vivo tracking of relatively dilute metabolites (0.1–10 mM) such as [1-¹³C]pyruvate,³ [1-¹³C]lactate,⁴ [¹³C]bicarbonate⁵ and other hyperpolarized (HP) ¹³C injectable biocompatible molecules.⁶ The ¹³C carboxylic nucleus offers long T₁ (on the order of 1 minute, therefore enabling in vivo tracking for up to several minutes.⁷ One hyperpolarization technology, dissolution Dynamic Nuclear Polarization (d-DNP),⁸ is already under investigation in 29 clinical trials (www.clinicaltrials.gov) to identify efficacy of HP ¹³C MRI for detecting aberrant metabolism of HP [1-¹³C]pyruvate in diseases.⁹ Despite the success of d-DNP in research settings to be able to provide safe, injectable HP ¹³C contrast agents for clinical utilization, this technology has a number of limitations including high device cost (>$2M) and slow production speed (≥1h for clinical device) that potentially pose barrier to translation for widespread clinical use.¹⁰ Lower cost and higher throughput hyperpolarization techniques are desirable to address these challenges.^11–12

Another alternative group of HP techniques that has been developed over the years relies of the parahydrogen (p-H₂) as a source of hyperpolarization.¹³ Historically, parahydrogen induced polarization (PHIP)^13–14 relying on pairwise p-H₂ addition was developed first, and translated to in vivo application approximately 15 years after the initial in vitro demonstration.¹⁵ Because this approach relies on the use of a PHIP catalyst required for p-H₂ pairwise addition, it needs to be removed prior to potential clinical use. Moreover, hydrogenation reactions can be performed directly in aqueous media^16–17 to facilitate biocompatible contrast agent preparation. Alternatively, hydrogenation in organic solvents followed by reconstitution into biocompatible aqueous buffer media has also been demonstrated.^18–20 A wide range of biocompatible HP molecular probes has been developed and translated in vivo^{21–22 23–24} using this technique including most prominently HP [1-¹³C]pyruvate²⁵ via a PHIP variant involving side arm hydrogenation (PHIP-SAH). The key advantages of the PHIP-SAH approach are fast hyperpolarization times (<1 min), and low costs of HP hardware. The key limitation of this method is the requirement of chemical modification of the to-be-hyperpolarized substrate through hydrogenation reactions, as well as the need for additional chemical de-esterification of the side arm after ¹³C nuclei have been hyperpolarized.^{12, 26–27}

Signal Amplification By Reversible Exchange²⁸ in SHield Enables Alignment Transfer to Heteronuclei (SABRE-SHEATH)²⁹ is an alternative technique to create HP [1-¹³C]pyruvate.^30–31 This method relies on simultaneous exchange of p-H₂ and a to-be-hyperpolarized substrate (e.g., [1-¹³C]pyruvate) with a metal complex (e.g., Ir-IMes hexacoordinate complex, e.g., 3b in Scheme 1) to enable spontaneous polarization transfer from p-H₂-derived hydrides to spin-spin-coupled heteronuclei (e.g., ¹³C of [1-¹³C]pyruvate^30–31). SABRE-SHEATH was first demonstrated for ¹⁵N,²⁹ and then quickly expanded to ¹³C,^{32 19}F,^{33 31}P,³⁴ and other nuclei. In 2019, Duckett and co-workers demonstrated the feasibility of SABRE-SHEATH hyperpolarization of [1-¹³C]pyruvate using DMSO as a co-ligand, but ¹³C polarization levels were limited to 1.85%.^{30, 35} The addition of DMSO leads to the formation of SABRE-active complex and 3b (Scheme 1); note that axial positions are not exchanging on the time scale of the SABRE-SHEATH experiment, and therefore, only complex 3b undergoes effectively reversible exchange with the HP substrate. The key advantages of the SABRE-SHEATH hyperpolarization approach are the low cost of HP hardware (<$20k) and rapid (~1 min) hyperpolarization process.³⁶ Moreover, this HP technique does not modify the to-be-HP substrate.^37–38 Since HP [1-¹³C]pyruvate is the leading HP contrast agent, the reader is reminded that d-DNP performs hyperpolarization of [1-¹³C]pyruvic acid, which is chemically modified to sodium [1-¹³C]pyruvate via a neutralization reaction. The key limitation of SABRE is that it is most efficient in alcohol solutions, e.g., in methanol, which has poor bio-compatibility.¹² Pioneering efforts to produce aqueous solutions of HP compounds via SABRE have improved this issue but so far have led to marked (by an order of magnitude or more) P decrease.^39–44 Development of bio-compatible formulations for agents hyperpolarized via SABRE technology remains a key unaddressed translational need, which is outside the scope of this manuscript. The reader is also directed to several recent reviews for detailed descriptions of leading HP technologies.^{9, 12, 26–27, 45–46}

Scheme 1.

Formation of [IrCl(H)₂(DMSO)₂(IMes)] (2) and [Ir(H)₂(η2-substrate)(DMSO)(IMes)] (3) complexes following activation of [IrIMes(COD)Cl] (1) pre-catalyst. Catalyst 1 was prepared previously⁵⁰ according to Cowley et al.⁵¹ The species 1, 2, 3a and 3b are as indicated by Iali et al. for pyruvate variants.³⁰ R = CH₃ for pyruvate and R = CH₂-CH₂-COO⁻ for α-KG.

We and others have recently demonstrated a series of innovations (including the use of water as an additional potential co-ligand,^47–48 and temperature cycling^{37, 49}) demonstrating that catalyst-bound [1-¹³C]pyruvate ¹³C polarization (P_13C) may reach order-unity values,⁴⁷ and ¹³C bulk P_13C of up to 13% can be obtained.

The work presented here aims to expand efficient SABRE-SHEATH hyperpolarization to other biologically relevant ketocarboxylic acids, exemplified by α-ketoglutarate. α-ketoglutarate is a central intermediate of the tricarboxylic acid cycle and other important metabolic pathways.⁵² Moreover, HP [1-¹³C]α-ketoglutarate ([1-¹³C]α-KG) prepared by d-DNP has been successfully employed to delineate altered metabolism in the isocitrate dehydrogenase 1 (IDH1) gene mutant. Therefore, HP [1-¹³C]α-KG can inform on IDH1 status via detection of the metabolic product HP [1-¹³C]2-hydroxyglutarate ([1-¹³C]2HG), which is missing in wild-type cells.⁵³ IDH1 mutants are potentially druggable.⁵³ Accordingly, HP ¹³C MRI with HP [1-¹³C]α-KG can inform on IDH1 mutation, and thus, has the potential to guide cancer patient treatment.⁵³

A pioneering study employing HP [1-¹³C]α-ketoglutarate showed that the HP resonance of metabolic product ([1-¹³C]2HG may overlap with the natural-abundance HP resonance from the ¹³C-5 carbon of [1-¹³C]α-KG. To mitigate this potential translational challenge of spectral overlap, Miura and co-workers have recently synthesized [1-¹³C,5-¹²C]α-KG and employed this HP isotopologue for detecting HP [1-¹³C]2HG in IDH1 mutants without natural-abundance HP ¹³C-5 background signal.⁵⁴ Taglang and co-workers have also demonstrated that late-stage deuteration of HP carboxylic acids can extend carboxylate ¹³C T₁ by up to a factor of 4.⁵⁵ That work indicated that ¹³C HP metabolic probes or in vivo analytes could generally benefit from deuteration to improve the lifetime of the ¹³C HP state of the molecular probe.⁵⁵

Inspired by the previous works, we have synthesized [1-¹³C,5-¹²C,D₄]α-KG, Scheme 2. Here, we have hyperpolarized two α-KG isotopologues to study their utility for SABRE-SHEATH: natural-abundance α-KG and newly developed [1-¹³C,5-¹²C,D₄]α-KG depicted in Scheme 1.

Scheme 2.

Overall synthetic schematic of [1-¹³C,5-¹²C,D₄]α-KG. See Supporting Information (SI) for details of experimental procedure and characterization details.

MATERIALS AND METHODS

The standard pre-catalyst used for SABRE, [IrCl(COD)(IMes)](IMes=1,3-bis(2,4,6-trimethylphenyl)imidazol-2-ylidene; COD=cyclooctadiene)],^{51, 56} catalyst 1 (6 mM) was activated with p-H₂ bubbling in the presence of ~5 mM α-KG, 20 mM DMSO and 0.5 M H₂O in CD₃OD (Figure 1, top), corresponding to our previously optimized protocol for preparation of HP [1-¹³C]pyruvate (P_13C=13%).⁴⁷ We used a low α-KG concentration (5 mM) because of the low α-KG solubility in CD₃OD. Fast activation (<5 mins) leads to the formation of species 2, 3a and 3b in accord with the notation introduced by Duckett and co-workers (in the context of pyruvate SABRE).^30–31 Complex 3b is the primary SABRE-active species as discussed above. For natural-abundance substrate studies, we have also performed a comparative study with pyruvate.

Figure 1.

a) Schematic of SABRE-SHEATH hyperpolarization process and relevant polarization transfer path of [1-¹³C,5-¹²C,D₄]α-KG. b) ¹³C spectrum of HP [1-¹³C,5-¹²C,D₄]α-KG; carbon positions are labeled by blue numerals. c) Representative stacked variable-temperature ¹³C spectra of 5 mM [1-¹³C,5-¹²C,D₄]α-KG demonstrating the interplay between HP complex 3b and the “free” peak as a function of temperature during the SABRE-SHEATH hyperpolarization process; d) corresponding ¹³C spectrum of thermally polarized neat [1-¹³C]acetic acid employed as signal reference for computation of signal enhancements. e) Build-up and decay of total ¹³C polarization of ¹³C-1 (i.e., integrating over all bound and free resonances) in [1-¹³C,5-¹²C,D₄]α-KG at B_T=0.42 μT; f) corresponding ¹³C-1 T₁ relaxation curves at the Earth’s and the clinically-relevant 1.4 T field of the benchtop spectrometer. Total ¹³C polarization of ¹³C-1 in [1-¹³C,5-¹²C,D₄]α-KG as a function of temperature (g) and magnetic transfer field (h). All experiments are performed at 1.4 T using a SpinSolve NMR spectrometer in CD₃OD at T_T=+10 °C (unless otherwise noted).

¹³C SABRE-SHEATH experiments were performed similarly to the recently reported study for [1-¹³C]pyruvate⁴⁷ by bubbling >98.5% p-H₂⁵⁷ at a flow rate of 70 standard cubic centimeters per minute (sccm) at 8 atm p-H₂ partial pressure for 2 min. Parahydrogen was temporarily stored (1–4 days) in aluminum cylinders using the storage and distribution system described in detail in the SI. ¹³C signal detection was performed using a clinically relevant 1.4 T field. The thermally polarized sample of neat (~17.5 M) [1-¹³C]acetic acid (Figure 1d) was employed as a signal reference to compute ¹³C polarization enhancement and P_13C (see SI).

RESULTS AND DISCUSSION

Simultaneous exchange of p-H₂ and α-KG leads to hyperpolarization of the 1-¹³C site detected in both isotopologues, Figs. 1,2(a,b). The temperature sweep (Figs. 1c,g and Fig. 2f) reveals that at cold temperatures (e.g., −12 °C), HP species 3b dominates the spectral intensity. The HP free (i.e., not catalyst bound) is negligible compared to the HP 3b resonance because the substrate exchange is too slow on the time scale of the experiment.⁴⁷ SABRE-SHEATH polarization at elevated temperature leads to the rise of the resonances from HP free species with an optimal polarization transfer temperature T_T of 6–10 °C, Figs. 1g,2f, for both isotopologues studied. At this temperature, the exchange rate of the substrate is likely matched to the size of the spin-spin coupling between the parahydrogen-derived hydrides and the ¹³C-1 nucleus. A field sweep revealed a P_13C maximum at B_T of ~0.42 μT (Figs. 1f, 2g).

Figure 2.

a) Schematic of SABRE-SHEATH hyperpolarization process and relevant polarization transfer path of natural-abundance α-KG; carbon positions are labeled by blue numerals. b) Representative HP ¹³C spectrum of 5.6 mM natural-abundance α-KG obtained by performing SABRE-SHEATH at +10 °C in CD₃OD at 1.4 T; c) corresponding ¹³C spectrum of thermally polarized neat [1-¹³C]acetic acid; d) Total (bound+free) ¹³C polarization build-up and decay at B_T=0.42 μT and T_T=+10 °C; e) Total (bound+free) ¹³C polarization decay at the Earth’s field and 1.4 T. Total ¹³C polarization of ¹³C-1 in natural abundance α-KG as a function of temperature (f) and magnetic transfer field (g). All experiments are performed at 1.4 T using a SpinSolve NMR spectrometer in CD₃OD.

The high P_13C level allows detecting dilute biomolecules with high SNR at natural ¹³C abundance (ca. 1.1%) even using a benchtop NMR spectrometer, which is remarkable as it substantially expands the capability of the demonstrated approach to potentially screen a wide range of new metabolic analytes that could be amenable to NMR hyperpolarization. Indeed, Figure 2b shows a spectrum of HP α-KG with a corresponding signal reference spectrum (neat [1-¹³C]acetic acid) shown in Figure 2c. We estimate the detection limit of 5 μM at the reported polarization level or ~80 μM*%polarization of ¹³C.

Just like in Figure 1, the microtesla relaxation dynamics of the ¹³C-1 carbon of natural abundance isotopologue (Figure 2d) shows that the total P_13C (bound+free) build-up time (T_b=12.8±0.5 s) is substantially shorter that the corresponding T₁ value of 22.4±0.8 s, indicating that the P_13C build-up rate substantially exceeds the rate of spin polarization destruction via T₁ relaxation (note the bound 3b and free species likely have different intrinsic ¹³C T₁ values (mostly because the catalyst itself is a relaxation agent), but since their exchange rate is substantially faster than 1/T₁, then one should observe “the same” average ¹³C T₁ for both lines that would be weighted by the mole fractions of each site under some set of conditions—hence, the current simplified analysis of our relaxation data). These favorable relaxation dynamics fundamentally enable relatively high P_13C levels. Indeed, P_13C of up to 17.3% was observed for natural-abundance α-KG (Fig. 2b). Somewhat lower maximum P_13C values were observed for [1-¹³C,5-¹²C,D₄]α-KG (e.g., Fig. 1b) even though the relaxation dynamics at B_T = 0.42 μT is effectively similar in the labeled isotopologue (T_b=11.0±0.4 s and T₁=18.5±0.8 s respectively, Fig. 1e). Instead, we explain the P_13C increase in the naturally abundant isotopologue in terms of the smaller heat capacity of the “spin bath” comprising available ¹³C spins that must be hyperpolarized by the same number of p-H₂ spins.⁵⁸ Indeed, corresponding ¹⁵N SABRE-SHEATH studies with isotopically enriched metronidazole and natural abundance metronidazole revealed substantially greater P_15N in the natural abundance compound^59–60 versus fully labeled analogue.^61–62 Future work on improving p-H₂ access to the polarization-transfer complex may potentially yield much better P_13C levels in HP [1-¹³C,5-¹²C,D₄]α-KG.

The observation of similar ¹³C-1 T_b and T₁ values measured at 0.42 μT of the two α-KG isotopologues is also remarkable in the context of exchangeable substrate deuteration. Deuterium is a quadrupolar nucleus that may potentially act as a source of relaxation for the nearby spins due to enhanced scalar relaxation of the second kind.^{31, 63–64}. Therefore, potentially deleterious effects of deuterium on ¹³C-1 T₁ (and by extension, on the maximum attainable P_13C-1 due to the increased rate of spin destruction) may be expected in a manner similar to that observed for ¹⁵N T₁ relaxation arising from ¹⁵N-¹⁴N two-bond interactions (where ¹⁴N spins act as quadrupolar relaxation sinks).⁶² We rationalize the above observation by the negligible strength of the relevant interactions over three (¹³C-C-C-D) chemical bonds. Indeed, the corresponding ¹⁵N-C-C-¹⁴N three-bond interactions have also been shown to be too weak to have a measurable effect of microtesla ¹⁵N relaxation in SABRE-SHEATH.⁶⁵

This observation of only a minor effect of γ-position deuteration on ¹³C-1 T₁ relaxation is important in the context of attaining high ¹³C-1 polarization levels because it implies that such substrate deuteration may cause no undesirable rapid 1-¹³C-1 depolarization in microtesla magnetic fields (and by extension, may not interfere with achieving high P_13C-1 values because the spin destruction rate remains lower than the polarization build-up rate). On the other hand, substrate deuteration will likely have a positive impact on extending 1-¹³C T₁ in α-KG at clinically relevant magnetic fields (particularly in the absence of the SABRE catalyst).⁵⁵ Of note, ¹³C-1 T₁ values (e.g., 54.4±2.3 s at 1.4 T and 39.6±1.1 s at the Earth’s field, Figure 1f) are likely limited by the interplay of the Iridium-catalyst induced- and exchange-induced T₁ relaxation processes. This notion is supported by the substantially longer ¹³C-1 T₁ values even at the higher (and more clinically relevant) 3 T field and in aqueous media (with no SABRE catalyst): e.g., 97.1±0.4 s for [1-¹³C,5-¹²C,D₄]α-KG—note that the non-deuterated isotopologue yields substantially lower ¹³C-1 T₁ at 3 T of 63.0±0.8 s, clearly demonstrating the potential clinical translation benefit of substrate deuteration for HP metabolic ¹³C studies.

We also note that ¹³C T₁ values at the Earth’s field and higher fields are substantially greater than those at the sub-microtesla magnetic fields employed here for polarization build-up as part of the (static-field) SABRE-SHEATH process. Therefore, other polarization-transfer approaches operating at higher magnetic fields (including but not limited to LIGHT-SABRE,⁶⁶ SLIC-SABRE,⁶⁷ QUASR-SABRE-SHEATH,⁶⁸ pulsed SABRE-SHEATH,⁶⁹ alt SABRE-SHEATH,⁷⁰ etc.^71–72) may potentially perform polarization build-up with overall lower ¹³C T₁ losses and thus may yield overall higher P_13C values—future work is certainly warranted in this direction.

Remarkably, no resonance from the HP ¹³C-2 carbon was detected from either isotopologue (expected between 200 and 210 ppm, see e.g. the spectrum in Figure 2b). This observation is somewhat unexpected, because the spin-spin couplings between parahydrogen-derived hydrides and the ¹³C-2 site are overall expected to be similar to those of the ¹³C-1 site, Figure 2a. Moreover, no HP ¹³C-5 resonance was observed either, suggesting that catalyst coordination with the other carboxyl group does not take place. Prompted by this unexpected lack of ¹³C-2 hyperpolarization in both α-KG isotopologues, we have performed corresponding studies with natural abundance pyruvate at similar (8.6 mM) pyruvate concentration and otherwise the same sample preparation and experimental conditions, Figure 3.

Figure 3.

a) Schematic of SABRE-SHEATH hyperpolarization process and relevant polarization transfer path of natural-abundance sodium pyruvate; carbon positions are labeled by blue numerals. b) Representative HP ¹³C spectrum of 8.6 mM natural-abundance sodium pyruvate obtained by performing SABRE-SHEATH at +10 °C in CD₃OD at 1.4 T; c) corresponding ¹³C spectrum of thermally polarized neat [1-¹³C]acetic acid; d) Total (bound+free) ¹³C polarization build-up and decay at B_T=0.42 μT and T_T=+10 °C; e) Total (bound+free) ¹³C polarization decay at the Earth’s field and 1.4 T. Total ¹³C polarization of ¹³C-1 in sodium pyruvate as a function of temperature (f) and magnetic transfer field (g). All experiments are performed at 1.4 T using a SpinSolve NMR spectrometer in CD₃OD.

Figure 3b shows the spectrum of HP pyruvate with P_13C-1 of 25.5% and P_13C-2 of 5.3%, indeed revealing a remarkable difference in the polarization values of the two binding sites, Figure 3a. The systematic temperature sweeping of the SABRE-SHEATH polarization conditions (Figure 3f) shows nearly identical position of the maximum for P_13C-1 of and P_13C-2 at 6–10 °C. Moreover, the microtesla magnetic field sweep (Figure 3f) revealed that optimum B_T for ¹³C-2 is slightly smaller (~0.20 μT) versus that for ¹³C-1 (~0.42 μT), indicating that the spin-spin interactions between p-H₂-derived hydrides and ¹³C-2 are ~1.5-times weaker than those for ¹³C-1 site. Furthermore, the relaxation analysis of pyruvate ¹³C-1 and ¹³C-2 sites at 0.42 μT reveals no substantial differences between the two sites, indicating that the minute difference in the rates of polarization build-up and decay for the two molecular sites during the SABRE-SHEATH process cannot explain the remarkable 4–5-fold difference in the P_13C-1 and P_13C-2. It should also be noted that the studies at natural ¹³C abundance level allow probing all ¹³C sites independently, due to low probability (1.1%) of having ¹³C in both positions at the same time. Taking together the key presented observations, we hypothesize that ¹³C-2 in pyruvate coordinates the iridium complex (Figure 3a) weaker due to side-chain dynamics induced by the –CH₃ group—and as a result, the residence lifetime of the bound ¹³C-2 site on the catalyst maybe reduced, in turn resulting in reduced P_13C and lower value of optimal B_T. In the case of α-KG, the side chain is substantially longer than the –CH₃ group of pyruvate, leading to more deleterious effects—more specifically, the complete lack of observable ¹³C-2 polarization in both α-KG isotopologues studied. An additional contribution may arise from coherent leakage from ¹³С−2 via spin-spin coupling with the nearby protons in protonated systems: at the magnetic fields used, these protons can strongly couple to the ¹³C-2 site and hence exchange of magnetization could occur.^73–74 Future site-specific hyperpolarization-EXSY experiments⁷⁵ for determination of dissociation rates in SABRE complexes and computational studies are certainly warranted to support or rule out the presented side-chain and coherent leakage hypotheses to explain the presented observations.

The maximum reported P_13C values for α-KG and pyruvate in their natural abundance form were 17.3% and 25.5% respectively at the time of the detection, and the actual achieved polarization values (inside the apparatus) are likely a few percent higher (the reader is reminded that the produced P_13C decays during the sample shuttling from the polarizer to the NMR detector). These values were achieved through rigorous optimization of matching magnetic field to establish spontaneous polarization transfer and temperature that modulates the rates of simultaneous p-H₂ and to-be-hyperpolarized substrate chemical exchange, Figures 2 and 3. While P_13C values can be potentially further improved, we anticipate that the gains will likely come from the use of higher p-H₂ pressure and flow rates, which have been shown previously to improve access to fresh p-H₂, i.e., the source of hyperpolarization.^{58, 76} The reported studies have been performed at 8 atm, i.e., at the maximum pressure rating for 5 mm NMR tubes employed in our apparatus. Therefore, to achieve these additional potential gains, the design of a specialized hyperpolarizer operating at substantially higher p-H₂ pressure and flow rate will be required. Such HP purpose-built setups have been reported for hydrogenative PHIP studies,^{26, 77–79} and we hope that this report will stimulate the development of corresponding instrumentation for SABRE polarization.

The optimum values of B_T and T_T (and maximum total P_13C) for ¹³C-1 in α-KG isotopologues are nearly identical to those of [1-¹³C]pyruvate reported recently,⁴⁷ indicating that our SABRE-SHEATH polarization transfer approach may be universally tailored to this structural motif of α-ketocarboxylate in other biologically relevant biomolecules, such as [1-¹³C]α-ketoisocaproate.⁸⁰

The main biomedical limitation of the presented study is production of HP substrate in an alcohol-based solution in the presence of an iridium-based catalyst, which may be considered toxic.⁸¹ While the iridium level employed here would likely not impact future feasibility studies in rodents, it would certainly need to be reduced by 2–3 orders of magnitude for clinical use. Several approaches have been recently demonstrated to capture iridium catalyst from SABRE-hyperpolarized solutions.^81–84 Moreover, heterogeneous SABRE approaches have been demonstrated to produce HP compounds free from catalyst.^85–86 In the context of methanol content of HP samples, it certainly needs to be substantially diluted (or preferably removed completely) for future in vivo studies. One way to achieve bio-compatible SABRE-based formulation of HP contrast agent is the use of water-soluble SABRE catalyst approaches, which have also been demonstrated to produce HP compounds in aqueous media to mitigate the use of toxic alcohols.^{39, 41, 43–44} However, the achieved P_13C levels using aqueous media have generally been at least an order of magnitude lower than those in methanol (and thus not suitable for in vivo applications, where a high degree of P_13C is needed), which has dampened enthusiasm for this approach. The development of bio-compatible HP contrast agents’ SABRE-based formulations is an active area of research investigations in our partnering and other laboratories, and we anticipate that this report will further stimulate work in that direction.

CONCLUSION

In summary, ¹³C SABRE-SHEATH enables efficient hyperpolarization (P_13C of up to 17.3%) of ¹³C-1 in two α-KG isotopologues: the natural abundance variant and [1-¹³C,5-¹²C,D₄]α-KG. [1-¹³C,5-¹²C,D₄]α-KG is designed as a next-generation molecular probe to sense real-time metabolism of α-KG in vivo using NMR hyperpolarization. This isotopologue is (1) ¹³C labeled in position 1 for spectroscopic sensing, (2) per-deuterated to maximize the lifetime of the ¹³C-1 HP state in aqueous media at 3 T; and (3) ¹²C labeled in position 5 to minimize background signal from this resonance. Complete deuteration of γ- and δ- positions has only a minor impact on ¹³C-1 relaxation in the microtesla fields under our conditions, indicating that spin-spin interactions between ¹³C-1 and remote protons/deuterons sites and corresponding contributions to ¹³C-1 relaxation are effectively negligible under these conditions. In contrast, no HP signals from ¹³C-2 spins were detected from α-KG, and substantially reduced ¹³C-2 polarization was detected in the shorter pyruvate molecule—we hypothesize that the binding of the ¹³C-2 site is generally weaker than that of ¹³C-1 due to random motion of the side chain (-CH₃ in pyruvate and –CH₂-CH₂-COO- in α-KG). Of critical importance in the context of biomedical utility of HP α-KG as a molecular probe, SABRE-SHEATH hyperpolarization process leads to highly specific polarization of ¹³C-1 site with no detectable polarization on any other sites and most notably ¹³C-5. This polarization specificity provides a clear benefit over d-DNP, which is non-site specific polarization technique, and which yields natural-abundance HP ¹³C-5 α-KG, which overlaps with HP [1-¹³C]2HG metabolic signal produced via metabolism of HP ¹³C-1 α-KG in IDH1 mutants.⁵⁴ Future systematic experimental and computational binding studies are certainly warranted to obtain a better understanding of relaxation dynamics in these important biomolecules, with the goal to further improve accessible ¹³C polarization for biomedical applications. We have also presented the feasibility of studying HP substrates at natural ¹³C abundance (1.1%) using a benchtop NMR spectrometer—indeed, meaningful information was readily obtained at ¹³C concentrations below 100 μM ¹³C concentration. The utility of natural abundance studies in the context of benchtop NMR spectroscopy allows studying a wide range of substrates without ¹³C isotopic enrichment for ¹³C SABRE-SHEATH hyperpolarization in a chemistry laboratory, without the need for sample transportation to a high-field NMR facility. As a result, we anticipate that the scope of amendable substrates (and laboratories) to SABRE-SHEATH can be rapidly expanded.