Microtesla SABRE enables 10% nitrogen-15 nuclear spin polarization.

Parahydrogen is demonstrated to efficiently transfer its nuclear spin hyperpolarization to nitrogen-15 in pyridine and nicotinamide (vitamin B(3) amide) by conducting "signal amplification by reversible exchange" (SABRE) at microtesla fields within a magnetic shield. Following transfer of the sample from the magnetic shield chamber to a conventional NMR spectrometer, the (15)N NMR signals for these molecules are enhanced by ∼30,000- and ∼20,000-fold at 9.4 T, corresponding to ∼10% and ∼7% nuclear spin polarization, respectively. This method, dubbed "SABRE in shield enables alignment transfer to heteronuclei" or "SABRE-SHEATH", promises to be a simple, cost-effective way to hyperpolarize heteronuclei. It may be particularly useful for in vivo applications because of longer hyperpolarization lifetimes, lack of background signal, and facile chemical-shift discrimination of different species.

The traditional limitation of magnetic resonance (MR) is low sensitivity because of low thermal equilibrium polarization (P ∼10–5–10–6). This limitation can be overcome by hyperpolarized magnetic resonance (HP-MR), which can boost P by several orders of magnitude. In addition, HP-MR is sensitive to molecular changes (reported by the chemical shift) and is free of ionizing radiation (unlike the established molecular imaging methods PET or SPECT). Thus, HP-MR enables real-time tracking of metabolic pathways and distribution processes.[1−4] Moreover, HP-MR works for drugs and metabolites in their natural form, free of cumbersome radioactive labels that alter biological function. These unique features of HP-MR create new avenues for interrogating disease states, their aggressiveness, degree of progression, and response to treatment.[5−7]

To date, most biomedical HP-MR applications involve dissolution dynamic nuclear polarization (d-DNP).[1−7] However, d-DNP is associated with high costs and lengthy hyperpolarization times: P ∼70% polarization in 20 min under optimal conditions.[9] An alternative, cost-effective, and continuous source of hyperpolarization is “Signal Amplification By Reversible Exchange” (SABRE), which uses parahydrogen (para-H2) as the source of spin order.[10] SABRE employs an organometallic catalyst that transiently binds both para-H2 and target molecules, permitting the transfer of spin order from para-H2 to the target by matching the resonance frequency difference between para-H2 and protons on the target to the magnitude of through-bond J-couplings. (Typically, the optimal B0 is ∼5–8 mT).[11] The main limitation of SABRE is the requirement for transient binding on a catalyst which currently restricts the potential classes of molecules amenable to SABRE. Nevertheless, a number of pyridine derivatives (e.g., nicotinamide,[10] pyrazinamide, and isoniazid)[12] and DNA building blocks (e.g., adenine and adenosine)[13] have been successfully hyperpolarized. Another significant limitation of SABRE is that it has been primarily used to hyperpolarize protons, which depolarize quickly (typically in seconds), making metabolic tracking on biologically relevant time scales challenging; direct detection of 1H signals also competes with background signals from water. Heteronuclear spins like 15N have lower sensitivity but are attractive because they have long polarization lifetimes, which in special cases can exceed 10 min.[14] SABRE-derived proton hyperpolarization has recently been transferred to 13C, but, thus far, the associated efficiency remains low (P ∼0.03%)—a significant enhancement, but one that leaves room for improvement.[15] Such recent improvements include “high-field SABRE” (via incoherent polarization transfer), which removes the need for sample transfer from the polarization region to the magnet for detection[16] (relying on incoherent polarization transfer in its first demonstration). This was improved again by enabling coherent transfer of para-H2 polarization to targets at high field using rf irradiation and shown to work for 1H as well as for 15N.[17−19]

Here we demonstrate direct production of P ∼10% on 15N by SABRE, exceeding all previous demonstrations by orders of magnitude. We use a field cycling strategy down to extremely low magnetic fields (∼μT) known to transfer para-H2-derived hyperpolarization to heteronuclei in hydrogenative para-H2 induced polarization.[11,15,18,20−31] We call the presented hyperpolarization strategy SABRE in SHield Enables Alignment Transfer to Heteronuclei (SABRE-SHEATH), which we demonstrate using 15N-Py and 15N-nicotinamide, but immediately extends to other 15N enriched heterocyclic compounds. Py,[32,33] nicotinamide,[34] and others[32,35] can be enriched with 15N using simple chemistries that allow either direct heteroatom replacement[32,35] or through ring opening and closure[33] using 15NH4Cl[32−35] as a cheap spin label source. As a result, the labor-intensive synthesis of complex 15N-labeled biomolecules can be largely avoided for this class of heterocycles. Correspondingly, our theoretical analysis easily extends to other biologically relevant molecules already shown to be amenable to 1H SABRE hyperpolarization.

The presented experiments shine in their simplicity: para-H2 is bubbled through a methanol-d4 solution containing activated catalyst ([IrH2(15N-Py)3(IMes)]+) and 15N-Py inside a μ-metal magnetic shield where the hyperpolarization is created; see Figure 1A. Subsequently, the sample is transferred into an NMR magnet for detection after a 90° pulse. The in-phase nature of the resulting signal is of particular importance for imaging applications, where the associated broader lines would suffer signal cancellations if the signals were antiphase.[37]

Figure 1

Setup for the SABRE-SHEATH experiment. (A) shows the sample in an NMR tube held inside the μ-metal shield, where para-H2 is bubbled through the solution. After hyperpolarization the sample is transferred into NMR spectrometer for detection. (B) para-H2 and 15N-pyridine (15N-Py) in exchange with [IrH2(15N-Py)3(IMes)]+. On the catalyst para-H2 transfers its hyperpolarization to 15N-Py through the J-coupling network. The equatorial 15N-Py (blue) is in fast exchange and responsible for the hyperpolarization of the free 15N-Py (green).

The simplest model for SABRE hyperpolarization uses the quantum mechanics of AA′BB′ systems developed in 1957 by Pople et al.[38] Such an AA′BB′ spin system is formed in the catalytic intermediate [IrH2(15N-Py)3(IMes)]+, as illustrated in Figure 1A. The AA′BB′ system consists of the para-H2 protons (AA′) and the 15N-nuclei of the two Py (BB′) in the equatorial plane. The spectral features of importance are the Larmor frequencies νA of 1H, νB of 15N, and the spin couplings JAA, JBB, JAB = JA′B′, and JAB′ = JA′B. At specific magnetic fields different sets of energy levels of the AA′BB′ system are brought very close to one another (they experience anticrossing),[20,21] thereby permitting efficient polarization transfer from the A-spins to the B-spins.

Building on previous work discussing the importance of low fields for efficient polarization transfer from para-H2 to heteronuclei,[11,15,18,20−28] the Supporting Information (SI) provides a theoretical description of the polarization-transfer process. In the SI it is shown that the hyperpolarization process can be intuitively understood as simple rotations on a Blochsphere. This theoretical analysis shows that generally, Δν = νA – νB needs to match one of the following J-terms:When Δν = +J1T or J2T hyperpolarization aligned with the weak external magnetic field is created; reversing the direction of the magnetic field matches to the other state and reverses the direction of the induced magnetization. In eq 1 only Δν depends on the magnetic field, given aswhere B0 is the magnetic field in Tesla, γ is the gyromagnetic ratio in MHz/T, and σ is the chemical shielding (in relative units conventionally related to the chemical shift δ (in units of ppm) as δ = 106 × (σref – σsample), where σref and σsample are the chemical shielding values of the reference sample (i.e., TMS for 1H NMR and liquid 15NH3 for 15N) and the sample of interest. Combining eqs 1 and 2 provides the magnetic field for polarization transfer:

In the original (homonuclear) case for SABRE (γA = γB), the matching condition is only dependent on the chemical shielding difference Δσ = 10–6 × Δδ) between para-H2-derived hydrides, σA, and Py protons σB. In this case, the optimal B0 is ∼6.6 mT (i.e., B0-transfer = 9 Hz/0.000032/42.6 MHz/T, assuming Δδ ∼32 ppm and JT ∼9 Hz);[11] this magnetic field corresponds to the conditions at which 1H-SABRE is most efficient.[10,37,39,40] For the heteronuclear case—of primary focus here—eq 2 gives a significantly different result: For 15N/1H, an optimal transfer field of B0-transfer ∼0.2–0.4 μT is predicted by eq 3 (assuming Δγ ∼47 MHz/T), depending on the exact value of JT; note Δσ is negligible given the large Δγ. However, this condition does not have to be precisely met, as it is limited by the residence times of para-H2 and the substrate molecule (e.g., Py) on the Ir complex, which are typically on the time scale of ∼0.1 s.[41] This new target magnetic field is significantly lower than the Earth’s magnetic field (∼50 μT); thus, 15N-SABRE experiments were conducted in a layer of μ-metal to shield the Earth’s magnetic field.

SABRE-SHEATH polarization is demonstrated by placing the NMR tube in a 305 mm-long magnetic shield (Lake Shore Cryotronics, P/N 4065) during ∼30 s of bubbling. Note that no attempt was made to tune the small residual field to a precise value. After this ∼30 s polarization period, para-H2 delivery is stopped, and the NMR tube is quickly transferred (∼4 s) to a 9.4 T NMR spectrometer to detect the SABRE-SHEATH polarization through conventional 1D-NMR. Both 15N and 1H pulse-acquire NMR experiments were conducted (see SI for further experimental details). Figure 2 highlights the main results: At concentrations of 4 mM 15N-Py (and 0.24 mM catalyst), a 30,000-fold polarization enhancement over the thermal level is achieved (Figure 2A). The thermal 15N polarization at 9.4 T and room temperature is ∼3.3 × 10–6, thus this enhancement corresponds to a P15N ∼10%.

Figure 2

Single-shot 15N NMR SABRE spectra. (A) SABRE-SHEATH experiment with Ir catalyst and 15N-Py concentrations of 0.24 and 4 mM respectively result in 30,000-fold signal enhancement corresponding to P ∼10% with 15N T1 ∼ 42 s at 9.4 T. (B) Shows that increase of Ir catalyst and 15N-Py concentrations results in further increased NMR signals, but relative enhancement and P levels decrease. An in-phase triplet split by 11 Hz is observed in both A and B. SABRE-SHEATH experiments (A, B) also reveal both bound 15N-Py ligand site of the activated Ir catalyst. (C) NMR spectrum of neat 12.5 M 15N-Py used for calibration. (D) 15N-SABRE at conditions optimized for 1H hyperpolarization (6 ± 4 mT) results in antiphase triplets and lower enhancements.

The results obtained with 63 mM 15N-Py (6.3 mM catalyst, Figure 2B) illustrate that with increasing concentrations, the absolute signal can be further increased, but the corresponding enhancement level (and the respective P) is diminished to 3000-fold (P ∼ 1%). The reduced enhancements observed with increasing concentrations are limited by three factors: (i) the finite amount of dissolved para-H2 limits the available hyperpolarization that can be passed to 15N-Py molecules; (ii) the Ir catalyst and 15N-Py concentrations modulate 15N-Py residence time, which affects the efficiency of SABRE polarization transfer;[40] and (iii) the NHC Ir catalyst is a source of T1 relaxation.[40] Optimization of these three and other (e.g., temperature and para-H2 pressure) parameters may in principle lead to higher, even on the order of unity, 15N hyperpolarization, which would be particularly important in the context of biomedical translation of this work.

The spectral pattern observed for free 15N-Py molecules in the SABRE-SHEATH experiments (Figures 2A and B) is a clean, emissive in-phase triplet, as expected, because of the dominant two-bond J-coupling between 15N and the ortho-Py-protons (2JNH is ≈11 Hz). In the traditional low-field SABRE experiment (at ∼6 mT)[10] optimized for protons, the 15N-triplet is typically characterized not only by lower enhancements but also by purely antiphase signatures, which can make such signals less useful in imaging applications because of partial signal cancellation from spectrally broadened lines.

In the SABRE-SHEATH experiments (Figure 2A,B), not only is the free 15N-Py signal detected, but signals from the ligand-bound species are observed as well. The equatorial 15N-Py molecules in the catalytic intermediate (blue) are those that exchange quickly and produce the observed hyperpolarization of the free 15N-Py (green). The axial 15N-Py (red) is also clearly visible in the SABRE-SHEATH experiments. This axial 15N-Py is no longer detectable in the experiments optimized for 1H SABRE (Figure 2D), where antiphase lines result.[37] The NMR lines of the catalyst-bound species are intrinsically broadened. Thus, the in-phase lines are still easily observable in SABRE-SHEATH experiments (Figure 2A,B), however the experiments conducted at ∼6 ± 4 mT (Figure 2D) exhibit antiphase lines with reduced enhancements.

The presented theory predicts the observation of in-phase signals resulting from standard I-magnetization in the SABRE-SHEATH experiments. We speculate that the antiphase signals in the 6 ± 4 mT experiments arise because the initial singlet spin-order of para-H2 is transferred not only into I-magnetization on protons but also into zero-quantum terms (e.g., ZQ = I1I2 + I1I2) on pairs of Py-protons, which are finally transferred (by the 2JNH-coupling ∝IS) into antiphase terms between protons and 15N (e.g., I1I2S), thereby resulting in antiphase signals. Additional indication for the involvement of antiphase and zero-quantum terms are presented in the the SI.

We also point out that polarization transfer strategies, e.g., 1H → 15N via INEPT, are unlikely to be as effective as the direct 15N-hyperpolarization in the SABRE-SHEATH because, as we demonstrate in the SI, standard proton SABRE is much less effective for 15N-Py than on the natural abundant 14N-Py.

This method can also be applied to other biomolecular contrast agents with 15N spin labels, such as 15N-nicotinamide (vitamin B3 amide);[12,36] Figure 3A demonstrates an enhancement of ε ∼20,000 (%P ∼7%) at 48 mM; a significantly higher concentration than that demonstrated for the greatest 15N-Py enhancement and potentially suitable for biomedical applications. Future demonstrations will likely show SABRE-SHEATH to work with heterogeneous catalysts[42] and/or in aqueous media[15,43] bringing this technique closer to its ultimate applications.

Figure 3

(A) 15N SABRE-SHEATH of a 48 mM sample of 15N-nicotinamide (15N enrichment of ∼66%) in methanol-d4 using 2 mM of activated IMes catalyst. An enhancement ε of ∼20,000-fold (corresponding to %P15N ∼ 7%) was achieved in comparison to (B) a spectrum of thermally polarized sample of 12.5 M 15N-Py (note the vertical axis is scaled by 50 fold). The spectrum of HP 15N-nicotinamide shows the expected three resonances of free and bound species. See SI for details.

In conclusion, a simple and cost-effective hyperpolarization scheme to produce high P levels on heteronuclei has greater T1 (P15N ∼10% ) is developed by conducting SABRE experiments with a 15N-labeled substrate in a magnetic shield. Despite the requirement for isotopic labeling, the heterocyclic compounds currently amenable to SABRE are well suited for inexpensive and straightforward enrichment strategies. The resulting hyperpolarization on heteronuclei, which are likely required for tracing hyperpolarized biomarkers in vivo. This technique can easily be extended to other biomolecules bearing 15N in important Py derivatives, but it is to be expected that many other classes of molecules will be enabled given the rapid progress in the field, and SABRE-SHEATH will likely remain an advantageous method to polarize heteronuclei including 15N and 13C. Future applications span from biomolecular imaging, to the study of protein dynamics.