Aqueous, Heterogeneous Parahydrogen-Induced 15N Polarization.

The successful transfer of parahydrogen-induced polarization to 15N spins using heterogeneous catalysts in aqueous solutions was demonstrated. Hydrogenation of a synthesized unsaturated 15N-labeled precursor (neurine) with parahydrogen (p-H2) over Rh/TiO2 heterogeneous catalysts yielded a hyperpolarized structural analog of choline. As a result, 15N polarization enhancements of over two orders of magnitude were achieved for the 15N-ethyl trimethyl ammonium ion product in deuterated water at elevated temperatures. Enhanced 15N NMR spectra were successfully acquired at 9.4 T and 0.05 T. Importantly, long hyperpolarization lifetimes were observed at 9.4 T, with a 15N T1 of ~6 min for the product molecules, and the T1 of the deuterated form exceeded 8 min. Taken together, these results show that this approach for generating hyperpolarized species with extended lifetimes in aqueous, biologically compatible solutions is promising for various biomedical applications.

Introduction

Hyperpolarization—the creation of highly nonequilibrium nuclear spin polarization—has been investigated for years as a way to dramatically improve the detection sensitivity of NMR and MRI.[1−8] Although many hyperpolarization methods have been developed, dissolution dynamic nuclear polarization (d-DNP)[9,10] has become increasingly dominant for biomedical applications because of advanced technology enabling the preparation of hyperpolarized (HP) nuclear spins within a wide range of chemical and biological systems, including metabolic MRI contrast agents now under investigation in clinical trials.[11−13] However, the high costs and infrastructure associated with d-DNP technology, combined with relatively slow production rates, present a challenge for many potential applications.

Approaches exploiting para-hydrogen-induced polarization (PHIP)[14−17] could be attractive alternatives because of their dramatically lower costs and instrumentation demands, much greater hyperpolarization rates (minutes to seconds, even allowing continuous agent production), and potential for scalability. In “traditional” PHIP,[15,18] the pure spin order from parahydrogen (p-H2) gas is transferred to a molecular substrate via the pairwise hydrogenation of asymmetric unsaturated bonds, a process that is typically facilitated with a catalyst. Key recent PHIP developments for biomedical applications include the demonstration of PHIP in aqueous media[19−25] and PHIP using heterogeneous catalysts (HET-PHIP).[26−30] The latter approach enables facile separation of the catalyst from the target molecule and, hence, the potential preparation of “pure” HP agents and catalyst reuse. However, also of great importance is the transfer of spin order from the nascent protons to substrate heteronuclei (e.g., 13C), providing greater hyperpolarization lifetimes compared to 1H spins. 13C hyperpolarization via PHIP has been achieved via both RF-driven polarization transfer[31−36] and polarization transfer in a magnetic shield (i.e., field cycling),[37,38] an approach that very recently has been extended to HET-PHIP conditions to produce aqueous solutions of highly polarized 13C-containing molecules free from the catalyst.[27,39]

Translation of this approach to 15N spins could have many advantages; indeed, Aime and co-workers have recently demonstrated 15N hyperpolarization of propargylcholine-15N via homogeneous PHIP and field cycling in a mixture of acetone and methanol or in water.[40] In addition to greatly increasing agent diversity, agents with HP 15N spins can be spectrally sensitive to the local biochemical environment.[41−43] Importantly, 15N T1 values are often considerably longer than corresponding 13C values,[43−45] thereby enabling longer hyperpolarization storage (either for direct readout or for transfer to 1H for more sensitive detection);[46−52] such T1 values are expected to be even longer at lower magnetic fields.[53,54]

Herein, we report 15N NMR hyperpolarization of a structural analogue of choline via heterogeneous, aqueous-phase hydrogenation of 15N-trimethyl(vinyl)-ammonium (i.e., neurine-15N) bromide over solid Rh/TiO2 catalysts. The PHIP-derived 15N nuclear spin polarization achieved in these experiments is the first reported to date involving heterogeneous catalysis and yielded 15N enhancements of ∼2 × 102-fold and a long relaxation time of ∼350 s at 9.4 T; deuterating the substrate yielded weaker enhancements but a longer relaxation time (15N T1 ≈ 500 s). Finally, significant signal enhancement is shown for the first time to enable detection at low (0.05 T)[55] magnetic field of the 15N resonance for molecules polarized using PHIP. For most of the heterogeneous hydrogenation reactions in this work, hydrogen gas was used with 50% p-H2 enrichment (the normal room-temperature ratio of para- to ortho-hydrogen is 25/75) prepared with a home-built generator (more details concerning experiments, chemical synthesis, and characterization are provided in the Supporting Information (SI)).

Results and Discussion

In one experiment, freshly produced p-H2 was bubbled at 90 psi into a medium-wall NMR tube (using a previously developed setup)[27,39] containing the target substrate (neurine-15N bromide) and the heterogeneous catalyst Rh/TiO2 in water (D2O) at 90 °C, causing the unsaturated substrate to be hydrogenated via pairwise addition. The reaction was performed under ALTADENA[56] conditions (i.e., wherein the hydrogenation reaction was performed outside of the magnet at low field), and results are shown in Figure (see also Figures S5 and S8). In order to effect the transfer of spin order from nascent 1H substrate spins to 15N, the hydrogenation reaction was performed by using a magnetic shield, similar to the recently reported procedure—also know as the magnetic field cycling (MFC) approach,[27,57−59] but in our case, the hydrogenation reaction was carried out directly in the magnetic shield. The level of polarization achieved is strongly dependent upon the speed of the sample transfer from the low (micro-Tesla) field to the Earth’s field;[40] therefore, to avoid related issues, the hydrogenation reaction was performed directly in the magnetic shield, and only after the termination of p-H2 bubbling was the sample quickly transferred to the high-field NMR for analysis. A strong 15N NMR signal was observed for the HP product (Figure B); however, no 15N signal was observed prior to p-H2 bubbling (Figure A).

Figure 1

(A) 15N NMR spectrum of a 0.25 M 15N-neurine substrate in the presence of 1.0% Rh/TiO2 before reaction with p-H2 in D2O, recorded with eight scans and a 30 s repetition delay; no signal of reactant is observed at 58 ppm under these conditions. 15N NMR spectrum of the hyperpolarized product using the same acquisition parameters as those used for spectrum acquisition shown in (A) but taken with 1 scan after transfer of spin order to 15N, achieved with 30 s 50%-enriched p-H2 bubbling inside of the magnetic shield. (C) Same as (B) but with hydrogenation occurring over 23.2% Rh/TiO2. HP 15N spectra are shown with an absorptive phase (i.e., sharing the same phase as a thermally polarized 15N sample); note that the field cycling was not optimized for polarization transfer.

This 15N NMR enhancement was achieved using Rh/TiO2 catalyst with 1.0% Rh loading. Importantly, utilization of Rh/TiO2 solid catalyst for heterogeneous PHIP[60,61] can, in principle, allow one to alter the conversion rate by varying the Rh fraction[62] of the catalytic material without decreasing the achieved polarization level of the products.[27] To investigate this possibility for the present reaction, a second Rh/TiO2 catalyst with 23.2% Rh loading was also used (Figure C). Although an enhanced 15N signal of the product is observed with the 23.2% Rh/TiO2 catalyst, the signal is ∼3.7-fold weaker than that observed with the 1.0% Rh catalyst. The explanation comes from the corresponding 1H HET-PHIP spectra (Figure S8), which indicate that, while the 23.2% Rh catalyst does indeed yield much higher reaction rates (in fact, providing essentially complete conversion of the substrate in 30 s), a smaller 1H polarization enhancement is achieved, giving rise to the weaker 15N enhancement in Figure (possibly reflecting either reduced pairwise H2 addition or different 1H relaxation of species adsorbed onto catalyst particles). Thus, the 1.0% Rh catalyst was used for the subsequent experiments in this work. In any case, these observations are the first reported to date for hyperpolarization of 15N-containing molecules via heterogeneous catalysis with 15N polarization derived from the spin order from p-H2.

The effects of substrate deuteration and increased p-H2 fraction on 15N signal enhancement were also separately investigated. Deuteration has previously been shown to increase heteronuclear (e.g., 13C) T1 in the context of PHIP[63,64] and DNP.[65] Here, following successful observation of 1H HET-PHIP with the fully deuterated substrate (neurine-15N-d12 bromide; see the SI for synthesis and Figure S9 for spectra), the approach described above was used to demonstrate 15N enhancement of the product in the aqueous phase following heterogeneous hydrogenation (Figure A). However, the intensity of this 15N line was slightly lower than that of the fully protonated substrate studied under the same conditions (Figure B). This reduced 15N enhancement with deuterated substrates is analogous to that observed with 15N SABRE-SHEATH (Signal Amplification By Reversible Exchange in SHield Enables Alignment to Heteronuclei)[66] and likely reflects either enhanced 15N relaxation in the μT regime (i.e. within the magnetic shield) or a more direct loss of spin order into the 2H spin degrees of freedom under those conditions.

Figure 2

(A) Single-shot 15N NMR spectrum of the fully deuterated substrate (0.125 M) solution in D2O obtained after 30 s of p-H2 bubbling (50% p-H2 fraction) and polarization transfer to 15N using the magnetic shield. (B) Same as (A) but with the protonated substrate (0.125 M). (C) Same as (B) but with an 80% p-H2 fraction.

On the other hand, use of a greater p-H2 fraction with the protonated substrate did yield an expected increase in 15N signal enhancement (Figure C). For this experiment, the p-H2 fraction was increased to 80% using a cryocooler-based p-H2 generator operating at a temperature lower than the 77 K of lN2 (see the SI for details). Increasing the ratio of para- to ortho-H2 from ∼50/50 (Figure B) to ∼80/20 (Figure C) yielded an improvement in the product’s 15N signal enhancement by nearly 3-fold, approaching the full 3-fold increase that would be theoretically expected if 100% p-H2 had been used. Taking the results from Figures and 2 together, the greatest signal enhancement was achieved using 80% p-H2 on the protonated substrate in the presence of 1.0% Rh/TiO2.

In PHIP, quantification of the sensitivity gain provided by the polarization level requires not only comparison with a signal from a thermally polarized sample but also an estimation of the efficiency of the hydrogenation reaction (and hence, the concentration of the product) at the time of detection. Here, the HP 15N signal was compared to the thermally polarized signal obtained from a 3.2 M 15NH4Cl aqueous solution. Note that the spectrum in Figure C was obtained after the first 30 s of p-H2 bubbling; comparison with thermal 15N (and 1H) spectra obtained with different bubbling times allowed the conversion level of the reagent to the product to be estimated at ∼10% (see Figure S10C). Therefore, the concentration of the product in Figure C is approximately 0.0125 M, yielding a corresponding 15N signal enhancement of ε ≈ 2 × 102 (attempts to detect polarization of 15N nuclei at natural abundance were unsuccessful).

While the 15N signal increased steadily with p-H2 bubbling time for the first 30 s of reaction, after that point, the signal tended to level off until the hydrogenation reaction yield reached 100% (not shown). This behavior could be potentially explained by changes in relaxation of species adsorbed onto the catalyst particles. Importantly, the 15N hyperpolarization was found to be long-lived at 9.4 T for both the protonated and deuterated substrates, with T1 decay constants of 348 ± 10 and 494 ± 13 s, respectively (Figure ), values that are roughly 1–2 orders of magnitude longer that the corresponding lifetimes of 1H hyperpolarization and a factor of 2 larger than that recently reported for 15N derivatives of choline.[40]

Figure 3

HP 15N T1 relaxation curves measured at 9.4 T for the protonated (A) and deuterated product (B); curves are exponential fits, giving the 15N T1 values (not corrected for ∼10° tipping-angle pulses) and error margins. Note the >2-fold time axis scale difference.

Finally, because the magnetization of HP species is not endowed by the NMR/MRI magnet, in principle, strong magnets are not required for detection. To investigate the feasibility of performing low-field NMR/MRI with 15N spins hyperpolarized by HET-PHIP, hydrogenation reaction products for both protonated and deuterated substrates were detected at 0.05 T, Figure . Indeed, 15N NMR resonances were successfully detected via low-field 15N NMR spectroscopy at a 210 kHz resonance frequency, with slightly greater signal observed for the protonated versus the deuterated product, which is qualitatively consistent with the high-field results. Higher polarization enhancements enabled by future experimental refinements should boost the signal to noise ratio to allow measurement of the 15N T1 at low field, where even longer hyperpolarization lifetimes would be expected,[54] thereby improving storage of the HP state.

Figure 4

Single-shot low-field (0.05 T) HP 15N NMR spectra of (A) the fully protonated product and (B) the fully deuterated product obtained via heterogeneous hydrogenation of neurine-15N bromide over a 1.0% Rh/TiO2 catalyst with 80% p-H2. Peaks of interest are at an offset of ∼(−)0.025 kHz.

Conclusions

In summary, heterogeneous Rh/TiO2 catalysts and MFC were used to achieve 15N hyperpolarization via HET-PHIP for the first time; previous transfers of spin order from p-H2 to 15N spins had only been achieved under homogeneous catalytic conditions via PHIP/MFC[40] or SABRE-SHEATH.[67,68]15N nuclear spin polarization enhancements of ∼2 × 102 fold (at 9.4 T) were observed in aqueous solutions following hydrogenation of neurine-15N bromide with p-H2, which yielded a structural analogue of the biological molecule choline (the HP form of which has been shown promising in vivo[45] because of the widespread function of choline in cellular metabolism and its significantly upregulated metabolism in cancer).[69,70] Larger enhancements were observed with 1.0 versus 23.2% Rh/TiO2 as well as with higher p-H2 fractions; however, deuteration of the substrate yielded lower enhancements but a longer hyperpolarization lifetime. Indeed, very long 15N T1 values were observed at 9.4 T for both the protonated and deuterated substrates, ∼6 and >8 min, respectively. The 15N hyperpolarization via HET-PHIP also enabled observation of 15N signals at low (0.05 T) field, where even longer hyperpolarization lifetimes are expected, further demonstrating the potential for wider applicability of the approach. Moreover, these results also expand the range of molecules (including biomolecules) amenable to HET-PHIP hyperpolarization. While the molecule hyperpolarized here lacks the −OH moiety found in choline, PHIP precursors for choline hyperpolarization have been previously described.[71] It should also be noted that the heterogeneous catalysts used (Rh/TiO2) are very stable and do not undergo any modifications during reaction (as was confirmed by XPS analysis; see the SI). Therefore, the absence of leaching of the active component of the catalyst material into solution, combined with the ability to use such supported metal catalysts for aqueous-phase heterogeneous hydrogenation (given their potential for facile separation), should allow not only catalyst recycling and reuse but also the preparation of pure HP substances free from the presence of the catalyst. Although the reported polarization values need to be increased further, taken together, these results open a door to the rapid and inexpensive creation of pure agents with long hyperpolarization lifetimes for various biomedical applications, including in vivo molecular MR imaging.