Spin Relays Enable Efficient Long-Range Heteronuclear Signal Amplification By Reversible Exchange

Abstract

A systematic experimental study is reported on the polarization transfer to distant spins, which do not directly bind to the polarization transfer complexes employed in Signal Amplification By Reversible Exchange (SABRE) experiments. Both, long-range transfer to protons and long-range transfer to heteronuclei i.e. ¹³C and ¹⁵N are examined. Selective destruction of hyperpolarization on ¹H, ¹³C, and ¹⁵N sites is employed, followed by their re-hyperpolarization from neighboring spins within the molecules of interest (pyridine for ¹H studies and metronidazole-¹⁵N₂-¹³C₂ for ¹³C and ¹⁵N studies). We conclude that long-range sites can be efficiently hyperpolarized when a network of spin-½ nuclei enables relayed polarization transfer (i.e. via short-range interactions between sites). In case of proton SABRE in the milli-Tesla regime, a relay network consisting of protons only is sufficient. However, in case ¹³C and ¹⁵N are targeted (i.e. via SABRE in SHield Enables Alignment Transfer to Heteronuclei or SABRE-SHEATH experiment), the presence of a heteronuclear network (e.g. consisting of ¹⁵N) enables a relay mechanism that is significantly more efficient than the direct transfer of spin order from para-H₂-derived hydrides.

Graphical Abstract

INTRODUCTION

Hyperpolarization techniques transiently increase nuclear spin polarization (P) by several orders of magnitude, resulting in corresponding gains in NMR signals.¹ These techniques² enable new applications including in vivo molecular imaging,^3–6 which relies on preparation, administration, and MRI of exogenous hyperpolarized (HP) contrast agents. Several technologies have been developed to produce HP states of low-gamma spin 1/2 nuclei (e.g. ¹³C and ¹⁵N), which retain HP state significantly longer than protons in biomolecular motifs.

Signal Amplification by Reversible Exchange (SABRE)⁷ is the hyperpolarization method, which employs parahydrogen (para-H₂) as the source of spin order.⁸ SABRE relies on spontaneous polarization transfer from para-H₂-derived hydrides to a substrate on polarization transfer catalysts. The SABRE polarization transfer is primarily accomplished via spin-spin interactions (i.e. J-couplings). While other interactions may contribute, they are generally orders of magnitude less efficient.^9–11 When polarization transfer is performed in milli-Tesla (mT) magnetic fields, the polarization from para-H₂-derived hydrides is spontaneously transferred to proton sites in substrates, enabling ¹H polarization values (%P_H) of up to 50%.¹² In this case, proton-proton spin-spin couplings (J_H-H) enable SABRE polarization transfer.^{7, 11, 13} Another variant of this technique dubbed SABRE in SHield Enables Alignment Transfer to Heteronuclei (SABRE-SHEATH) is an efficient process for hyperpolarizing ¹⁵N sites, in one example providing more than 20% polarization in 1 minute.^{14 15}N-hyperpolarization is of particular interest, because it enables long hyperpolarization lifetimes and exponential decay constants. In some cases, ¹⁵N exponential decay constant can exceed 20 minutes.^15–16 SABRE-SHEATH is accomplished at micro-Tesla (μT) magnetic fields, enabling spontaneous polarization transfer from para-H₂-derived hydrides to heteronuclei, as illustrated in Scheme 1a. In case of ¹⁵N SABRE-SHEATH protonnitrogen-15 spin-spin couplings (J_H-15N) enable the spontaneous polarization transfer.^17–18 While polarization transfer between nuclear spins in the same molecule in solution is an extremely well documented area, the case is less clear for SABRE and SABRE-SHEATH, which occur within a complex with finite lifetime, and which involves several changes of the external magnetic field.

SCHEME 1.

a) The overall schematic of SABRE hyperpolarization of ¹⁵N. Spin order is transferred from para-H₂ and mediated by scalar spin-spin couplings within a reversibly-formed Ir-IMes hexacoordinate complex.^{23, 26} Direct SABRE of short-range ¹⁵N sites is accomplished via 2-bond couplings between ¹⁵N and hydride protons.^17–18 b) the molecular framework (note axial ligangd of polarization transfer from hydride protons via short- and long-range spin-spin (J) couplings.

Over the past two years, this approach has been bolstered by a series of advancements from proof-of-principle demonstrations in organic solvents^17–18 (including “neat” substrates¹⁹) all the way to demonstrations of SABRE-SHEATH enhancement under heterogeneous catalytic conditions²⁰ and catalysis in aqueous media,²¹ which in principle will enable the preparation of pure aqueous ¹⁵N HP compounds for potential in vivo use. Moreover, the scope of amendable bio-structures has been expanded from N-heterocycles to Schiff bases,²² diazirines,¹⁵ and nitriles.¹⁶

However, despite these successes, to date SABRE-SHEATH has been primarily employed for hyperpolarization of ¹⁵N sites that bind directly to metal centers via two-bond spin-spin couplings (i.e. short-range couplings to substrate ¹⁵N sites directly binding to catalysts like Ir-IMes [IMes = 1,3-bis(2,4,6-trimethylphenyl)imidazole-2-ylidene], Scheme 1a).²³ Efforts to hyperpolarize long-range ¹⁵N sites (i.e. more than 2 chemical bonds away from para-H₂-hydrides) showed that polarization efficiency is reduced by more than an order of magnitude¹⁴—presenting a clear obstacle for expanding SABRE hyperpolarization technology to broader classes of substrates. Moreover, recent studies of direct ¹³C^{24 and 19}F²⁵ hyperpolarization via SABRE indicated that the problem can be mitigated if the nitrogen sites binding the iridium metal centers are labeled with ¹⁵N nuclei as opposed to naturally occurring quadrupolar ¹⁴N spins.²⁴ However, the obvious question regarding the mechanism of polarization transfer (and more importantly its limitations) was unanswered. Specifically: “Is SABRE of distant heteronuclei accomplished directly via long-range J-couplings, or does it work indirectly via relay through a network of spins (Scheme 1b)? Here, we present experimental evidence that SABRE of long-range ¹⁵N and ¹³C sites is efficient in the presence of a relay network connected by close-range spin-spin couplings. While the data presented does not rule out a direct mechanism via weak long-range J-couplings, the indirect mechanism is significantly more efficient than the direct polarization transfer (i.e., the enhancements observed cannot be fully explained by long-range transfer; although long-range transfer is possible, it leads to weaker enhancements). We hope that the presented work will stimulate future theoretical studies and pave the way to efficient SABRE-SHEATH hyperpolarization of remote heteronuclear sites thereby expanding the reach of this technique.

METHODS

All NMR spectra were recorded using a 9.4 T Bruker Avance III high-resolution NMR spectrometer equipped with a broad-band dual-channel NMR probe. All SABRE experiments were performed using ~50% para-H₂. All sample manipulations related to the sample transfer between magnetic fields (i.e. the Magnetic Field Cycling (MFC) procedures) were performed manually (Figures 1a,b and 2a,b). The NMR spectra from metronidazole-¹⁵N₂-¹³C₂ (Millipore-Sigma P/N 32744-10MG) samples were recorded in medium-wall 5 mm NMR tubes, whereas the spectra from signal reference samples were recorded using standard 5 mm NMR tubes.

Figure 1.

The schematic of SABRE polarization (a) and re-polarization (b) experiments for ¹H hyperpolarization studies.

Figure 2.

The schematic of SABRE-SHEATH polarization (a) and re-polarization (b) experiments for ¹⁵N and ¹³C hyperpolarization studies.

Preparation of solutions

Previously synthesized IrIMes catalyst was used for the described studies.¹⁰ Metronidazole-¹⁵N₂-¹³C₂ (~20 mM final concentration) and pre-activated iridium catalyst ([IrCl(COD)(IMes)], ~1.0 mM final concentration) and methanol-d₄ were prepared from stock solutions of Metronidazole-¹⁵N₂-¹³C₂ (40 mM) and ([IrCl(COD)(IMes)], 2.0 mM). Both solutions were flushed with Argon and vortexed (at least three times) in an Eppendorf safe-lock tube and Duran bottle (10 mL GL25) correspondently. Please note: while Metronidazole-¹⁵N₂-¹³C₂ solution was used over a period of several days (stored at ~6 C°), ([IrCl(COD)(IMes)] was prepared fresh each day and stored for <10 hours in a refrigerator (~6 C°). A portion (~0.3 mL) of each stock solutions was transferred into an Argon-filled medium-walled 5 mm NMR sample tube: 9 in.-long, 3.43-mm inner diameter (ID) (Wilmad Glass, P/N 503-PS-9) equipped with a Teflon tube extension: 0.25 in. outer diameter (OD), 3/16 in. ID. The Teflon extension was approximately 3-in. long. The tube was connected to the previously described setup.²⁷ The best results were obtained by SABRE sample activation by bubbling para-H₂ for ~3 minutes (at 90 sccm & 65 psi) and leaving sample for ~1 hour under para-H₂ at 65 psi.

SABRE hyperpolarization

For SABRE experiments, the para-H₂ flow rate (50 sccm, ~1 minute of bubbling) was controlled using a mass flow controller (Sierra Instruments, Monterey, CA, P/N C100L-DD-OV1-SV1-PV2-V1-S0-C0). The schematic of SABRE experimental setup is shown in Figure 1a, and the corresponding sequence of the events is shown in Figure 1b. Micro-Tesla magnetic field was created by attenuating the Earth’s magnetic field using a three-layered mu-metal shield (Magnetic Shield Corp., Bensenville, IL, P/N ZG-206,). The tuning of SABRE-SHEATH magnetic field was achieved by custom-built solenoid coil (>90% homogeneity over >15 cm length) and a power supply (GW INSTEK, GPRS series) with variable-resistor bank connected in series with the magnet coil. When para-H₂ bubbling was stopped, the sample was transferred from the shield to the Earth’s magnetic field (ca. 50 μT), followed by rapid sample transfer into a 9.4 T magnet of the NMR spectrometer for data acquisition or further manipulations as described in the main text. Typical sample transfer time (from cessation of para-H₂ gas to ¹⁵N/¹³C/¹H signal detection) was 4–7 seconds. The schematic of SABRE-SHEATH experimental setup is shown in Figure 2a, and the corresponding sequence of the events is shown in Figure 2b.

The reader should note use of magnetic field cycling (MFC) to the Earth’s magnetic field, ca. 50 μT, (SABRE-SHEATH, Figure 2b) for ¹⁵N re-polarization, whereas ¹H re-polarization (SABRE, Figure 1b) employs MFC to the fringe field, ca. 6 mT, due to the difference in the matching conditions for efficient ¹⁵N→¹⁵N or ¹H→¹H re-polarization respectively: see the main text for more detailed consideration.

Computation of NMR signal and nuclear spin polarization enhancements

The signal enhancements were computed as the following. ¹H enhancements were computed by diving the HP signal magnitude by the magnitude of thermally polarized signals. In case of ¹⁵N and ¹³C, the thermally polarized signals were generally very low, and therefore, external signal reference was employed. The enhancements were computed as follows:

$$\epsilon =\frac{S_{HP}}{S_{\mathit{REF}}}\times \frac{C_{\mathit{REF}}}{C_{HP}}\times \frac{A_{\mathit{REF}}}{A_{HP}}\times \frac{N_{\mathit{REF}}}{N_{HP}}$$

where S_HP and S_REF are NMR signals for HP state and thermally polarized signal reference samples respectively, C_REF and C_HP are the effective isotope concentrations of thermally polarized signal reference and HP samples respectively, A_REF and A_HP are the solution cross-sections in the NMR tube of thermally polarized signal reference and HP samples respectively, and N_REF and H_HP are the numbers of symmetrical sites per molecule for the thermally polarized signal reference and HP samples respectively (A_REF/A_HP was ~1.85 as described earlier¹⁸). Percentage polarization was computed by multiplying signal enhancement ε by the equilibrium nuclear spin polarization of a given spin at 9.4T and 298K: 3.2×10⁻³% (¹H), 8.1×10⁻⁴% (¹³C), 3.3×10⁻⁴% (¹⁵N).

RESULTS AND DISCUSSION

General Consideration

Briefly, the reader is reminded that in case of traditional SABRE (i.e. hyperpolarization of proton sites), the matching condition for J-coupling (¹H-¹H) mediated polarization transfer occurs at a few mT magnetic field. In case of SABRE-SHEATH (i.e. hyperpolarization of heteronuclear sites, e.g. ¹⁵N and ¹³C), the matching condition for J-coupling (e.g., ¹H-¹⁵N) mediated polarization transfer occurs at μT magnetic field. Detailed theory is provided in earlier works of Duckett et al.^{7, 11, 13} and Theis at al.^17–18

SABRE hyperpolarization of pyridine proton sites

The first reports of SABRE demonstrated that proton sites at least six-bonds away can be efficiently hyperpolarized.^{7, 13} The theoretical basis for SABRE¹¹ indicated that polarization transfer from para-H₂-hydrides to substrate protons is accomplished via a networks of proton-proton couplings, although the recent study by Eshuis and co-workers determined the values of these four-, five-, and six-bond spin-spin couplings, and postulated that these long-range couplings may also enable the canonical SABRE effect via the direct transfer of spin order.²⁸

Here, we have employed the approach previously described for studying ¹H SABRE,²⁷ and performed selective spin destruction of HP resonances of ortho-pyridine protons (denoted as Ha in Figure 3a) after the sample was hyperpolarized at magnetic field of ~6 mT and para-H₂ bubbling was stopped (Figure 3b). After destroying the Ha polarization with a frequency-selective pulse (Figure 3c), the sample was moved back into the ~6 mT field, and finally, the sample was returned to the 9.4 T NMR spectrometer for detection (this procedure effectively represents magnetic field cycling or MFC – the details of the experimental setup and the involved steps are shown in Figure 1a,b of the Methods section). The resulting spectrum shown in Figure 3d indeed shows that the Ha HP state is successfully recreated using the meta- and para- (Hb and Hc) protons as hyperpolarization reservoirs. Corresponding datasets showing HP state destruction and re-creation for the Hb and Hc protons may be found in the Supporting Information (Figure S1).

Figure 3.

¹H NMR spectra of SABRE-hyperpolarized ~100 mM pyridine solutions in methanol-d₄. All NMR spectra were recorded using Bruker Avance III 9.4 T NMR spectrometer. a) Thermally polarized spectrum provided for signal referencing. b) NMR spectrum of SABRE-hyperpolarized solution (at ~6 mT) after cessation of para-H₂ bubbling (Figure 1a). c) NMR spectrum of SABRE-hyperpolarized (at ~6 mT) solution after frequency-selective RF irradiation leading to selective destruction of Ha hyperpolarization; d) the corresponding NMR spectrum after the sample prepared in c) was re-hyperpolarized at ~6 mT without para-H₂ bubbling (Figure 1b). We note that the overall signal intensity of the spectrum shown in d) decreased compared to that in spectrum c) due to relaxation processes leading to polarization decay during the additional ~6 s required for sample shuttling. We also note that the total shuttling time (after cessation of para-H₂ bubbling) for spectrum d is > 10 s, i.e. more than 3*T₁ of hydrides, and significantly shorter than T₁ of aromatic protons (19±1 s),¹⁰ and therefore, it was concluded that the residual polarization of hydrides cannot serve as a source of re-polarization in spectrum in spectrum d.

All-in-all, the experimental results shown in Figure 3a,b,c,d and Figure S1 are clearly consistent with the canonical (i.e. ¹H) SABRE effect in the mT regime may indeed rely on the network of proton-proton couplings, in agreement with pioneering studies by Duckett and co-workers.^{7, 11, 13}

SABRE hyperpolarization of metronidazole-¹⁵N₂-¹³C₂ in micro-Tesla (μT) magnetic fields (SABRE-SHEATH)

Hyperpolarization in μT magnetic fields was performed using a μ-metal magnetic shield that attenuates the Earth’s magnetic field (ca. 50 μT) down to the sub-μT regime (Figure 2a).^{17–18, 29} The structure of the activated complex and the schematic of SABRE-SHEATH process are shown in Scheme 2. Efficient ¹⁵N^{17–18 and 13}C²⁴ SABRE-SHEATH hyperpolarization has been demonstrated previously, and no significant effort was made to optimize polarization parameters (e.g., temperature and μT field) for the present study.

SCHEME 2.

A schematic of the SABRE-SHEATH hyperpolarization process of metronidazole-¹⁵N₂-¹³C₂ using transfer of spin order from para-H₂ on an Ir-IMes hexacoordinate complex.^{23, 26} SABRE-SHEATH is accomplished via spin-spin couplings between para-H₂-derived hydride protons and nuclear spins of the equatorial exchangeable ligands. The axial ligands are not exchangeable.

The following polarization levels were achieved for ~20 mM metronidazole-¹⁵N₂-¹³C₂ solutions in methanol-d₄ using 50% para-H₂:³⁰ ε_15N1 ~ 4,000 (%P_15N1~1.3%), ε_15N3 ~ 4,700 (%P_15N1~1.5%), Figure 4a,b, ε_13C2 ~ 310 (%P_13C2~0.3%), ε_13CH3 ~ 230 (%P_13CH3~0.2%), Figure 4c,d, ε_H ~ 14 (%P_H~0.o4%), Figure 4e,f,g. If near 100% para-H₂ were employed, the polarization values would be effectively tripled. %P_15N values are within the expected ranges for the employed concentration regime.^17–18 %P_13C values are several times lower than %P_15N, because initial micro-Tesla field optimization was performed for ¹⁵N spins, and the optimization of polarization efficiency was outside the scope of the presented mechanistic study. %P_H values are significantly lower than %P_15N and %P_13C values, because μT fields are not optimal for proton SABRE.

Figure 4.

NMR spectra of metronidazole-¹⁵N₂-¹³C₂ hyperpolarized using SABRE-SHEATH setup (Figure 2a). a) HP ¹⁵N NMR spectrum, b) ¹⁵N spectrum from a thermally polarized reference sample, c) HP ¹³C NMR spectrum, d) ¹³C spectrum from a thermally polarized reference sample, e) HP ¹H NMR spectrum (polarization at ~6 mT), f) ¹H spectrum from a thermally-polarized sample, g) HP ¹H NMR spectrum (polarization at <1 μT), All spectra were recorded using Bruker Avance III 9.4 T NMR spectrometer.

SABRE-SHEATH hyperpolarization of long-range ¹⁵N sites

¹⁵N hyperpolarization of metronidazole¹⁴ at natural abundance of ¹⁵N (~0.3%) and ¹³C (~1.1%) isotopes was recently shown. In that work, the efficiency of SABRE-SHEATH hyperpolarization (gauged as %P_15N) of ¹⁵N₁ and -¹⁵NO₂ sites was significantly lower (by more than an order of magnitude) than that of the ¹⁵N₃ site. Low natural ¹³C/¹⁵N abundance results in simplification of the spin system (participating in SABRE-SHEATH), effectively reducing it to a three-spin system (Scheme 3), because the statistical probability of the simultaneous presence of two spins (e.g. ¹⁵N and ¹⁵N or ¹⁵N and ¹³C) is two orders of magnitude lower than the statistical probability of the structures shown in Scheme 3a–c (Note, SABRE relevant spin-spin couplings for direct polarization transfer are shown in Scheme 3e,f).

SCHEME 3.

The relevant spin-spin coupling schemes for three ¹⁵N sites at natural abundance of ¹⁵N and ¹³C (a–c) and the network for labeled metronidazole-¹⁵N₂-¹³C₂ (d). Note that only two-bond heteronuclear couplings are shown in d. (e–f) SABRE relevant spin-spin couplings for direct polarization transfer.

The ¹⁵N SABRE-SHEATH polarization of met-ronidazole-¹⁵N₂-¹³C₂ shows that polarization efficiency of ¹⁵N₁ site is 85% of that of the ¹⁵N₃ site (Figure 4a), which is markedly different from the previously reported corresponding value of 2%.¹⁴ This more than 40-fold improvement of hyperpolarization at ¹⁵N₁ can be rationalized by the presence of ¹⁵N₃ and ¹³C sites in this labeled molecule, which provide a J-coupling network enabling relayed SABRE-SHEATH polarization transfer (instead of direct H→¹⁵N₁ transfer). In the relayed case, two-bond couplings (²J_H-15N3 and ²J’_H-15N3) enable hyperpolarization of the ¹⁵N₃ site, and ¹⁵N hyperpolarization is then propagated to other ¹³C and ¹⁵N sites via the heteronuclear coupling network.

Additional experimental evidence for the relayed nature of such polarization transfer is provided in Figures 5 and 6. For these experiments the HP sample was inserted in the 9.4 T NMR spectrometer and the ¹⁵N₃ hyperpolarization was destroyed using frequency-selective RF pulses (note that the ¹⁵N₁ polarization was preserved in this case, Figure 5a). The relevant setup and sample manipulation steps are shown in Figure 2b. Following this selective polarization-destruction procedure, the sample was transferred into the Earth’s magnetic field (ca. 50 μT) to enable ‘re-mixing’ of heteronuclear polarization (no bubbling of fresh para-H₂ is performed). Lastly, the sample was returned into the NMR spectrometer for read-out. After this magnetic field cycling (MFC), the polarization of ¹⁵N₃ is recovered to nearly the same level as the one for the ¹⁵N₁ site, as shown in Figure 5b. We note that the T₁ of ¹⁵N sites is typically on the order of 1 minute or greater, and a significant fraction of hyperpolarization can be retained after 10–30 seconds of manipulation time required in such experiments. We also note that para-H₂ bubbling employed in the initial SABRE-SHEATH procedure was stopped before the sample left the magnetic shield for the first time. We also note that even when the SABRE hyperpolarization experiment was performed at the (higher) Earth’s magnetic field (control experiment), the achieved polarization on ¹⁵N₁ and ¹⁵N₃ sites (Figure 5c) was lower than the one achieved in Figure 5b. When combined, this evidence supports the conclusion that in the MFC procedure, the ¹⁵N₃ site was re-hyperpolarized using the hyperpolarization pool of the ¹⁵N₁ site. While some polarization transfer from ¹³C is also potentially possible, it is less likely because HP ¹³C sites depolarize more quickly and the initial polarization levels of ¹³C sites were significantly lower than those of ¹⁵N sites (Figure 4a). Moreover, in a separate experiment, we inverted (using a 180° RF pulse) ¹⁵N magnetization of the HP ¹⁵N₁ site (Figure 5d) prior to the second MFC procedure. As a result of the spin inversion of the ¹⁵N₁ site, the resulting polarization on the ¹⁵N₃ site followed this inversion (Figure 5e) after MFC. Furthermore, additional experiments (Figure S2b) shows that MFC to the fringe field of ~6 mT was not sufficient to achieve this effect, and re-hyperpolarization was less efficient.

Figure 5.

¹⁵N NMR spectra of HP metronidazole-¹⁵N₂-¹³C₂: a) after SABRE-SHEATH hyperpolarization at μT field, cessation of para-H₂ bubbling, and HP sample transfer to the 9.4 T NMR spectrometer followed by frequency-selective polarization destruction of ¹⁵N₃ site. b) the spectrum obtained via the procedure described in (a) followed by magnetic field cycling (MFC) to the Earth’s magnetic field and then back to the 9.4 T NMR spectrometer; c) after SABRE hyperpolarization at the Earth’s magnetic field, cessation of para-H₂ bubbling, and HP sample transfer to the 9.4 T NMR spectrometer; d) the spectrum obtained by the procedure described in (a) followed by 180° phase inversion of the ¹⁵N polarization; and e) the spectrum attained by the procedure described in (d), followed by magnetic field cycling (MFC) to the Earth’s magnetic field and then returned to the 9.4 T NMR spectrometer. See Figure 2b for details. All NMR spectra are acquired using 90° excitation RF pulse after the respective manipulations.

Figure 6.

¹⁵N NMR spectra of HP metronidazole-¹⁵N₂-¹³C₂: a) after SABRE-SHEATH hyperpolarization at μT magnetic field, cessation of para-H₂ bubbling, and HP sample transfer into 9.4 T NMR spectrometer followed by frequency-selective polarization destruction of ¹⁵N₁ site; b) the sample produced by the procedure described in (a) followed by the magnetic field cycling (MFC) to the Earth’s magnetic field (ca. 50 μT) and then back in the 9.4 T NMR spectrometer; c) the sample produced by the procedure described in (a) followed by 180° phase inversion of ¹⁵N polarization; d) the sample produced by the procedure described in (c) followed by the magnetic field cycling (MFC) to the Earth’s magnetic field (ca. 50 μT) and then back in the 9.4 T NMR spectrometer. All NMR spectra are acquired using 90° excitation RF pulse after the respective manipulations.

The re-hyperpolarization mechanism requires mixing of spin states from ¹⁵N₃ and ¹⁵N₁ sites. At the Earth’s field (ca. 50 μT) the chemical shift difference of the two sites (ca. 100 ppm away) is only ~0.02 Hz, which is significantly less than their J-coupling (which we estimate to be on the order of a Hertz or less). At 6 mT the frequency difference of the two sites is ~2.6 Hz, which is apparently still too large for efficient level anti-crossing. Accordingly, MFC to the Earth’s field (ca. 50 μT) is required.^31–33 We also note that performing the MFC procedure from the 9.4 T to the magnetic shield (and therefore passing the Earth’s field condition twice) also leads to the ¹⁵N re-polarization (data not shown), but the effect is significantly reduced, which is likely due to additional polarization leaks (e.g. to ¹³C and ¹H sites).

Figure 6 provides additional experimental evidence for polarization transfer from ¹⁵N₃ site to ¹⁵N₁ site. Instead of applying the frequency-selective irradiation on ¹⁵N₃, it was applied to ¹⁵N₁ (Figure 6a). Exactly analogous results are obtained. The MFC to Earth’s field (ca. 50 μT) enables re-hyperpolarization of ¹⁵N₁ using the polarization of HP ¹⁵N₃ (Figure 6b). The corresponding spin inversion experiments (Figure 6c,d) prove that ¹⁵N₁ site was indeed re-hyperpolarized from magnetization of HP ¹⁵N₃ site.

While the possibility of polarization transfer between spin ½ nuclei is predictable, e.g. between two ¹⁵N spins as shown here), demonstrating this phenomenon in the context of SABRE/SABRE-SHEATH repolarization is critical for understanding and proving the mechanism of relayed polarization transfer in weak magnetic fields. Taken together, the above evidence supports the model of relayed ¹⁵N SABRE-SHEATH polarization of long range sites, explaining the efficient polarization of a distant ¹⁵N₁ site that is four bonds away from the para-H₂-derived metal hydrides (Scheme 3d).

SABRE-SHEATH hyperpolarization of long-range ¹³C sites

Figure 4c demonstrates the efficient SABRE-SHEATH hyperpolarization of ¹³C sites three and four bonds away from para-H₂-derived hydrides. Importantly, the efficiency of ¹³C hyperpolarization of the-¹³CH₃ group was ~75% of that of ¹³C₂ site; we note however, that –¹³CH₃ sites may have lost disproportionately more polarization during sample transfer from μT field to 9.4 T of NMR spectrometer, because the ¹³C T₁ in methyl groups is typically significantly shorter than T₁ values of ¹³C sites without directly attached protons (e.g. the ¹³C₂ site). Therefore, the actual efficiency (at the end of the SABRE-SHEATH procedure prior to the HP sample transfer) might have been better than 75%. We have recently demonstrated that the presence of a ¹⁵N nucleus in pyridine and other similar compounds is essential for efficient polarization (i.e. large %P_13C values) of ¹³C sites that are three (ortho-position) and four (meta-position, Figure 3) chemical bonds away from hydride protons (Scheme 1).²⁴ That previous proof-of-principle work did not address the reasons for low heteronuclear polarization efficiency in the case of ¹⁴N spins that would otherwise be present in such compounds. One part of the explanation is enhanced scalar relaxation of the second kind^34–37 suffered by the target spins within the micro-Tesla regime, induced by quadrupolar ¹⁴N sites within the scalar coupling network. A second explanation is the absence of a close spin ½ J-coupling network. All-in-all, the data presented in Figure 7 supports the importance of J-coupling networks for hyperpolarization of ¹³C spins too (in addition to the ¹⁵N sites discussed as the primary topic of this work).

Figure 7.

¹³C NMR spectra of HP metronidazole-¹⁵N₂-¹³C₂. a) Spectrum obtained After SABRE-SHEATH hyperpolarization at μT magnetic field, followed by cessation of para-H₂ bubbling, and sample transfer to the 9.4 T NMR spectrometer, followed by ¹H and ¹³C polarization destruction (via applying a series of 90° RF pulses to ¹³C spins and ¹H decoupling to ¹H spins); b) The spectrum obtained by the procedure described in (a) but followed by magnetic field cycling (MFC) to the μT regime (i.e. within a magnetic shield) and then after the sample was returned to the 9.4 T NMR spectrometer. c) The spectrum obtained by the procedure described in (a) but followed by magnetic field cycling (MFC) to the Earth’s magnetic field prior to return to the 9.4 T NMR spectrometer. All NMR spectra shown were acquired using a 90° excitation RF pulse after the respective manipulations described in the figure caption above.

In Figure 7 we provide additional evidence supporting the need for polarization relays in such strongly coupled networks. First, a metronidazole-¹⁵N₂-¹³C₂ sample was hyperpolarized via SABRE-SHEATH and para-H₂ bubbling was stopped. Next, the sample was rapidly transferred into the 9.4 T NMR spectrometer, and hyperpolarization on the ¹³C and ¹H sites was immediately destroyed (by applying a series of 90° RF pulses to ¹³C spins and ¹H decoupling to ¹H spins) and a proton-decoupled ¹³C spectrum was acquired (Figure 7a). In a separate experiment (instead of recording ¹³C spectrum), the sample was additionally transferred back into the μT field created by the magnetic shield (employed for SABRE-SHEATH polarization in the first step—here employed to re-enable heteronuclear polarization transfer), and ¹³C hyperpolarization was indeed re-created (Figure 7b). Note that since all ¹³C and ¹H sites were depolarized (by broadband irradiation), we conclude that the only remaining source of polarization for such ¹³C re-hyperpolarization is the hyperpolarization pool of ¹⁵N sites. Moreover, the efforts to re-polarize this sample by MFC to the Earth’s magnetic field (Figure 7c) were unsuccessful in comparison to re-hyperpolarization at μT field (Figure 7b)—further indicating that the source of polarization must be from non-carbon spins.

Broader relevance

Metronidazole is an important potential contrast agent (because high %P_15N can be achieved,¹⁴ and it is possible to administer high (~2 g³⁸) dose of this potent antibiotic) that can be potentially employed for hypoxia imaging in a manner similar to fluoromisonidazole (FMISO) and other radiolabeled nitroimidazole derivatives^39–44 used in position emission tomography (PET) imaging. Therefore, this work will certainly be useful for future development and optimization of SABRE-SHEATH hyperpolarization of this and other ¹⁵N and ¹³C HP imaging probes.

More generally, the systematic studies presented in this work provide clear experimental evidence that heteronuclear J-coupling spin-1/2 networks serve as the underlying fundamental basis for efficient relayed polarization transfer from para-H₂-derived metal hydrides (Scheme 2) to substrate heteronuclei that do not directly bind to the Iridium complex. Here, we show that efficient hyperpolarization of long-range ¹³C and ¹⁵N sites is achieved in SABRE when a network of heteronuclear J-couplings is present. The presence of this network translates to efficient hyperpolarization of many heteronuclei, including ¹³C,^{24 15}N,^{17 31}P,^{45, 19}F,²⁵ etc.⁴⁶. Most importantly, the nitrogen site directly interacting with the iridium hexacoordinate complex should be labeled to create a relay of spin ½ nuclei for efficient polarization transmission deeper within the intramolecular space. Therefore, we believe this relayed mechanism opens new opportunities for efficient SABRE hyperpolarization of new biomolecular targets, and informs the rational design of nuclear spin coupling networks in labeled agents for a wide variety of potential biomedical applications. Of note, ¹³C labeling did not significantly affect ¹⁵N T₁ relaxation: T₁(¹⁵N₃) is ca. 36 s with (data not shown) and without ¹³C spin labeling¹⁴, which is relevant for production of HP contrast agents with long-lived HP states in the context of biomedical applications. As another example, uniformly (or backbone-) ¹³C/¹⁵N labeled peptides and proteins can potentially act as efficient networks of SABRE hyperpolarization, and it may be possible to hyperpolarize isotopically labeled peptides and proteins using this approach. We also envision that isotopically labeled DNA, RNA, and other structures can be hyperpolarized via this SABRE-SHEATH approach. The general concept of relayed polarization transfer through a close J-coupling network will also potentially translate to LIGHT-SABRE and RF-SABRE sequences that create hyperpolarization directly in the magnet and avoid the need for sample transfer.^47–49

In the context of SABRE hyperpolarization of proton sites, this work provides experimental evidence in support of relayed polarization transfer. As a result, it would be potentially possible to hyperpolarize long-range proton sites too. While proton-hyperpolarized compounds are rarely used for contrast agents for HP MRI, potential SABRE applications most likely would include analysis of complex mixtures at low concentrations^50–53 with potential detection of nitrogen- and sulfur-containing heterocycles⁵⁴ in oil or refined petroleum products.

CONCLUSION

Efficient SABRE hyperpolarization of long-range ¹³C and ¹⁵N sites was demonstrated in metronidazole-¹⁵N₂-¹³C₂ in the μT field regime. We have shown that long range ¹³C and ¹⁵N sites (i.e. three and four chemical bonds away from para-H₂-derived hydrides) can be hyperpolarized much more efficiently via a mechanism of relayed spin-polarization transfer than via weak long-range J-couplings. Specifically, the short-range ¹⁵N site, directly bound to Iridium, is hyperpolarized first and hyperpolarization is then transferred / relayed to other intramolecular sites via a network of short-range J-couplings involving further spins. The presented evidence opens new opportunities for SABRE-based hyperpolarization of long range spin-½ nuclei in a wide range of applications ranging from biomedical contrast agents, to analysis of complex mixtures, to structural biology.