Quantifying the Effects of Quadrupolar Sink via ¹⁵N Relaxation Dynamics in Metronidazoles Hyperpolarized via SABRE-SHEATH

Abstract

¹⁵N spin-lattice relaxation dynamics in metronidazole-¹⁵N₃ and metronidazole-¹⁵N₂ isotopologues are studied for rational design of ¹⁵N-enriched biomolecules for Signal Amplification by Reversible Exchange in microtesla fields. ¹⁵N relaxation dynamics mapping reveals the deleterious effects of interactions with polarization transfer catalyst and quadrupolar ¹⁴N nucleus within the spin-relayed ¹⁵N-¹⁵N network.

Graphical Abstract

Presence of ¹⁴N nucleus in the scalar coupling network causes deleterious effects in SABRE hyperpolarization in microtesla fields resulting in 3-fold decrease of ¹⁵N T₁ and polarization values for all ¹⁵N sites in ¹⁵N₂-isotopologue versus ¹⁵N₃-isotopologue.

The nuclear spin polarization P at thermal equilibrium is governed by a Boltzmann distribution of nuclear spins among Zeeman energy levels. P increases linearly with magnetic field strength. For a conventional high-field NMR spectrometer (e.g., 9.4 T) or clinical Magnetic Resonance Imaging (MRI) scanner (e.g., 3 T) at room temperature, P is typically on the order of 10⁻⁵ to 10⁻⁶, resulting in a relatively low sensitivity of NMR-based applications. For NMR applications where this sensitivity is too low to be useful, various hyperpolarization strategies may be employed to increase P by as much as 4–5 orders of magnitude,^{1, 2} with corresponding signal gains.

One such technique is Signal Amplification by Reversible Exchange (SABRE), pioneered by Duckett et al. in 2009,³ which utilizes simultaneous reversible chemical exchange of parahydrogen (p-H₂) and to-be-hyperpolarized substrate molecules at a metal center. In SABRE, the transfer of nuclear spin polarization from parahydrogen-derived hydrides to a spin-polarizable substrate occurs spontaneously via the network of spin-spin couplings established in a transient polarization transfer catalyst (PTC) complex (Figure 1a).^3–5 While several approaches have been developed for polarization transfer in SABRE,^6–11 a variant of SABRE, SABRE-SHEATH (SABRE in SHield Enables Alignment Transfer to Heteronuclei),^{12, 13} facilitates the generation of highly polarized spin states of heteronuclei^14–17 including nitrogen-15 with high polarization (P_15N >30%¹⁸) persisting for tens of minutes.¹⁹ Nitrogen is found in a wide range of biomolecules including nucleic acids, amino acids, proteins, and drugs. Since SABRE-SHEATH is performed at very low magnetic fields (< 1 μT) and near room temperature, the production of such HP ¹⁵N spin-labeled biomolecules is comparatively simple, fast and inexpensive.¹⁰

Figure 1.

a) Molecular exchange between p-H₂ and substrate, e.g., metronidazole (MNZ) employed here, in SABRE hyperpolarization. b) Structure of pyridine-¹⁵N employed as a signal reference. c-d) Corresponding structures and polarization transfer spin-relays (red overlay) between p-H₂ and ¹⁵N nuclei in corresponding MNZ ¹⁵N-isotopologues. e) Signal reference ¹⁵N NMR spectrum of a thermally polarized neat pyridine-¹⁵N acquired with 8 scans and 10-minute recovery time. f-g) Corresponding ¹⁵N NMR spectra of HP metronidazole-¹⁵N₂ and metronidazole-¹⁵N₃.

Nitroimidazoles can be readily reduced in anaerobic environments. This property has been widely employed in a number of antibiotic drugs,²⁰ Positron Emission Tomography (PET) tracers for hypoxia sensing,²¹ and also in a number of emerging cancer therapeutics: e.g., evofosfamide (a.k.a. TH-302)²² and the radiosensitizer nimorazole.²³

Metronidazole is an FDA-approved antibiotic,²⁴ belonging to the nitroimidazole class of compounds. We envision that ¹⁵N-hyperpolarized metronidazole can be potentially employed for hypoxia sensing in a manner similar to that of nitroimidazole-based Positron Emission Tomography (PET) tracers. One such tracer, ¹⁸F-fluoromisonidazole (FMISO),²¹ undergoes reduction in hypoxic environment (including most notably hypoxic tumors) and the metabolic products of this reduction process become trapped in hypoxic cells, providing contrast in FMISO PET images.²⁵ The enormous potential for using HP MRI to sense metabolic transformations in vivo has been well demonstrated^{10, 26}; correspondingly, HP MRI of metronidazole may obviate the limitations of FMISO PET imaging, including the use of ionizing radiation, the requirement for long clearance time from surrounding tissues, and the inability to spectrally distinguish parent compounds from downstream products.

We have demonstrated efficient SABRE-SHEATH hyperpolarization of the ¹⁵N₃ site in natural abundance metronidazole with %P_15N exceeding 30%.¹⁸ This nitrogen site directly interacts with the PTC, and therefore gains its polarization directly from p-H₂-derived hydrides. Subsequently, commercially available metronidazole-¹⁵N₂-¹³C₂ (Sigma-Aldrich, #32744) was employed for SABRE-SHEATH, but unfortunately yielded significantly lower %P_15N (roughly by an order of magnitude), although all ¹⁵N- and ¹³C-labeled sites have been successfully hyperpolarized.^{27, 28} Most recently we have synthesized metronidazole-¹⁵N₃ and demonstrated ¹⁵N→¹⁵N spin-relayed SABRE-SHEATH hyperpolarization via two-bond ¹⁵N-¹⁵N spin-spin couplings, Figure 1d. This metronidazole-¹⁵N₃ isotopologue exhibited a remarkable %P_15N of ~16% on all three ¹⁵N sites including ¹⁵NO₂, which has a polarization relaxation decay constant T₁ approaching 10 minutes. However, in order to facilitate more effective production of HP biomolecules for bioimaging applications, it is clearly necessary to gain improved understanding of the underlying spin-relaxation phenomena to inform the rational design of these promising imaging agents.

Here, we report a quantitative study of spin relaxation dynamics of metronidazole-¹⁵N₂ and metronidazole-¹⁵N₃ isotopologues, Figure 1c and Figure 1d respectively, using previously described experimental setup (Figure S2).^{27, 29} Catalyst activation was performed for approximately 2 h for each sample studied to ensure reproducibility, see Figure S1 in the Supporting Information (SI). Other experimental parameters (temperature, p-H₂ pressure, flow rate and in-shield magnetic field) were optimized for each isotopologue, Figure S1. Using an overpressure of 94 psig, p-H₂ was bubbled through a solution of MNZ ¹⁵N-isotopologue substrate and Ir-IMes catalyst in 0.6 mL methanol-d₄ for one minute to facilitate polarization transfer in microtesla fields. Following polarization build-up, the sample solution was rapidly transferred (2–4 s to minimize the relaxation losses) to a 1.4 T NMR Pro (Nanalysis, Canada) bench-top NMR spectrometer for ¹⁵N NMR detection (Figure 1h). Each relaxation/build-up curve was obtained by varying the time duration that the sample spent in a given magnetic field.

The key results related to ¹⁵N T₁ relaxation at the optimal magnetic field during the SABRE-SHEATH polarization transfer process (ca. 0.4 μT, Figure S1) for metronidazole-¹⁵N₃ and metronidazole-¹⁵N₂ are shown in Figures 2e and 2f, respectively. As expected for microtesla magnetic fields, all ¹⁵N sites within a given molecule share approximately the same relaxation rate (e.g., 13.8–15.4 s for MNZ-¹⁵N₃, Figure 2e). However, the ¹⁵NO₂ group replacement by ¹⁴NO₂ leads to dramatic, 3-fold shortening of the ¹⁵N T₁ (4.3–4.8 s corresponding T₁ values for MNZ-¹⁵N₂, Figure 2f). This striking effect can be explained by the enhanced scalar relaxation of the second kind^{30, 31} induced by the quadrupolar ¹⁴NO₂ site within the N-N spin-spin coupling network. These results are further supported by the overall similar ¹⁵N T₁ trend at the Earth’s magnetic field (ca. 10 μT in the basement of our lab at Detroit, MI, Table S1). Moreover, we find that each ¹⁵N polarization build-up constant (T_b, Figure 2c and Figure 2d) at 0.4 μT is closely correlated with the corresponding T₁ value. In practice, this means that the increased relaxation rate caused by the presence of the quadrupolar ¹⁴N spin in MNZ-¹⁵N₂ (despite the peripheral position of the NO₂ group) allows for achieving the steady-state %P_15N faster on ¹⁵N₃ and ¹⁵N₁ sites—but at significantly lower levels. The correlation plot of %P_15N versus T₁ at 0.4 μT indeed exhibits a linear trend with R²>0.99 (Figure 2inset). When the magnetic field is sufficiently high (e.g., 1.4 T, Figure 2g and Figure 2h), the increased frequency dispersion of the nuclear spins puts them in a weakly coupled regime with ¹⁵N sites having significantly longer T₁ decay constants—on the order of many minutes.

Figure 2.

a-b) Structures of two metronidazole ¹⁵N-isotopologues. c-d) Corresponding ¹⁵N polarization build-up curves at 0.4 μT. e-f) Corresponding ¹⁵N T₁ decay curves at 0.4 μT. g-h) Corresponding ¹⁵N T₁ decay curves at 1.4 T. The presented data was recorded using a 2 mM IrIMes catalyst concentration and a corresponding 20 mM MNZ isotopologue concentration. All experiments are performed in CD₃OD.

On another note, the realization that scalar-coupled ¹⁴N spins are highly deleterious in the context of SABRE-SHEATH suggests that if these quadrupolar effects would have been avoided, near-unity P_15N would have been potentially achievable in the previous studies.¹⁸

It should be pointed out that substrate exchange of Ir-IMes catalyst may act as the potential source of additional undesirable ¹⁵N relaxation (e.g., due to compounding effects of quadrupolar Ir nucleus and the chemical exchange process).¹³ Consequently, a series of control experiments were performed, where the catalyst concentration was systematically varied from 0.5 mM to 1 mM to 2 mM at a fixed concentration of metronidazole isotopologue (Figures S3a–d). The [catalyst] increase from 0.5 mM to 2 mM results in a stepwise decrease in ¹⁵N T₁ and T_b by approximately 2-fold at 0.4 μT (Figure S3b and Figure S3c). However, because the interplay of T₁ relaxation and catalyst concentration is complex in the SABRE process,^{32, 33} these decreases in ¹⁵N T₁ and T_b at 0.4 μT are offset by the overall increased catalyst-to-substrate ratio (i.e., better substrate access to p-H₂ spin bath), resulting in somewhat greater %P_15N in metronidazole-¹⁵N₃ and similar %P_15N in metronidazole-¹⁵N₂ at higher [catalyst], Figure S3d. Of note, the Ir-IMes catalyst decreases ¹⁵N T₁ even at high magnetic fields (1.4 T, weakly coupled regime) for ¹⁵NO₂ (Figure S3a). Moreover, this observation clearly indicates a second reason (beyond agent purification) that SABRE catalyst removal^{18, 34–36} is warranted as soon polarization build-up is completed to minimize ¹⁵N polarization losses prior to biomedical utilization of HP metronidazole-¹⁵N₃ as a contrast agent.

The high levels of ¹⁵N polarization obtained in ¹⁵N-labeled metronidazole isotopologues enable direct ¹⁵N MRI. Figure 3 demonstrates the 2D ¹⁵N projection images of 5 mm NMR tubes filled with HP solutions of ¹⁵N-labeled metronidazole isotopologues with the highest spatial resolution reported to date to the best of our knowledge.

Figure 3.

¹⁵N MRI of 5 mm NMR tubes filled with hyperpolarized solutions of 0.1 M metronidazole-¹⁵N₃ (MNZ-¹⁵N₃) and metronidazole-¹⁵N₂ (MNZ-¹⁵N₂) respectively using 5 mM of Ir(COD)(IMes)Cl in methanol-d₄ obtained by TrueFISP pulse sequence. Imaging parameters employed: TR = 62.5 ms, TE = 3.6 ms, scan time = 2.0 seconds, flip angle = 15˚, matrix size = 32×32 (zero-filled to 512×512). a) axial projection 2D image of metronidazole-¹⁵N₃ using 1 average (maximum SNR(SNR_MAX) is ~500, b) axial projection 2D image of metronidazole-¹⁵N₂ using 1 average, SNR_MAX is ~410, c) coronal projection 2D image of metronidazole-¹⁵N₃ using 8 averages, SNR_MAX is ~450, d) coronal projection 2D image of metronidazole-¹⁵N₂ using 8 averages, SNR_MAX is ~340.

Conclusions

In summary, the spin-relayed SABRE-SHEATH hyperpolarization approach allows efficient polarization of scalar coupled ¹⁵N-¹⁵N spin networks. These networks may be created via two-bond ¹⁵N-¹⁵N J-couplings at most. The presence of ¹⁴N spins in such networks must be avoided to prevent deleterious polarization losses due to quadrupolar relaxation effects (manifested on ¹⁵N as scalar relaxation of the second kind¹⁶), in order to maximize the resulting %P_15N. Although the catalyst decreases the ¹⁵N spin-relaxation time constant, T₁, of metronidazole isotopologues in the microtesla regime in a concentration-dependent manner, the overall impact on the achievable ¹⁵N polarization level is relatively minor. On the other hand, the presence of a ¹⁴N nucleus in the scalar coupling network results in an approximately 3-fold decrease of microtesla ¹⁵N T₁ values for all ¹⁵N sites in the ¹⁵N₂-isotopologue versus the ¹⁵N₃-isotopologue over a wide range of catalyst concentrations. This ¹⁵N T₁ reduction results in a corresponding 3-fold decrease of ¹⁵N polarization levels. These findings have substantial translational relevance for the rational design of hyperpolarized MRI contrast agents comprising ¹⁵N- and ¹³C-labeled biomolecules—both in general, and in the specific case of SABRE-hyperpolarized metronidazole, an antibiotic that can be potentially employed for non-invasive hypoxia sensing antibiotics,²⁰ emerging cancer therapeutics such as evofosfamide²² and nimorazole radiosensitizers,²³ etc. Feasibility of high resolution MRI of HP metronidazole is shown, which bodes well for potential biomedical applications.