Hyperpolarization of 15N-pyridinium and 15N-aniline derivatives by using parahydrogen: new opportunities to store nuclear spin polarization in aqueous media

We introduce 15N quaternary pyridinium as moiety that can be NMR-signal-enhanced by several orders of magnitudes and allows for long-term storage of the so gained hyperpolarization in water.


Introduction
Nuclear magnetic resonance (NMR) spectroscopy and magnetic resonance imaging (MRI) are powerful techniques, which have widely been used for studying molecular structures associated to diseases and to visualize illnesses even in vivo. [1][2][3][4] Both techniques are greatly hampered, due to their low sensitivity. This limitation can be overcome by using hyperpolarization methods, which increase signals of molecules by more than four orders of magnitude. [5][6][7][8][9][10][11][12][13] Several hyperpolarization techniques have evolved to gain new insights e.g. in the elds of structural biology, material science, chemical analysis, biochemistry and biomedical science. With a view on the latter, hyperpolarization allows for creating new contrast agents to study and diagnose diseases in vivo. 14 The technique mainly used for producing hyperpolarized contrast agents is dissolution dynamic nuclear polarization (d-DNP). 5 It enables the hyperpolarization of metabolically active compounds that can be followed during in vivo studies. 8,9,11,[14][15][16] Other methods with biomedical relevance are spin exchange optical pumping (SEOP) [17][18][19][20] of noble gases and para-hydrogen induced polarization (PHIP). [21][22][23][24][25][26][27][28][29][30][31][32] PHIP methods transfer nuclear spin order from para-hydrogen (para-H 2 ) enriched hydrogen over to target molecules for their hyperpolarization. Hydrogenative PHIP adds para-H 2 to unsaturated precursors over suitable hydrogenation catalysts, to create large spin-order in the target compounds, which can be converted into observable magnetization aerwards. Due to the design of suitable precursor molecules, this technique can now be utilized to hyperpolarize metabolically active compounds and to analyze their chemical conversion in vivo. 28,31,32 Within the past ten years, a non-hydrogenative para-H 2based hyperpolarization methods has evolved: signal amplication by reversible exchange (SABRE). [32][33][34][35][36][37] For this method, para-H 2 and a substrate of interest coordinate to a temporarily stable transition metal complex. In this complex, the para-H 2 spin order is converted into observable magnetization at the molecule of interest. Dissociation of the complex leads to free hyperpolarized substrates that have not been altered as in the classical PHIP approach. 38 However, this method has not been shown to be applicable for in vivo applications yet since, SABRE experiments typically need to be performed in organic solvents. However, the eld rapidly progressing and work is on the way to make this technique more biologically applicable in the future. 32,39,40 What all techniques have in common is the desire to store hyperpolarization in contrast agents for long periods of time. To this end, hyperpolarization is typically stored on hetero-nuclei such as in 13 C and 15 N, which possess longitudinal relaxation times (T 1 ) ranging from seconds to minutes. The T 1 of 13 Cpyruvate, the metabolite most commonly hyperpolarized, for example is in the range of 40-60 s. 41 For in vivo applications, this results in a time window of 2-3 minutes, during which pyruvate can be monitored. 11,14,41 To increase tracing times, 15 N nuclei are more favorable than 13 C nuclei since T 1 can be one order of magnitude longer and T 1 > 1200 s (20 minutes) in water have been reported in quaternary nitrogen compounds. 30 Due to its longer T 1 values, 15 N-derived chemical probes have been explored: with respect to PHIP N-ethyl trimethyl ammonium (NETMA) and an allyl choline derivative have been polarized in biocompatible solvents. 30,42,43 Dissolution DNP has demonstrated rst in vivo experiments utilizing 15 N polarized choline and several other applications in vitro such as pH-sensing, Ca 2+ monitoring and enzyme activity. [44][45][46] Degrees of 15 N-polarization have long been rather low until the advancements in crosspolarization (CP) d-DNP have overcome this challenge. 13 SABRE has made great progress in polarizing 15 N spins in the past years. 34,35,47 Demonstrations of over 40% polarization in 15 N pyridine and more than 30% for imidazole have been accomplished in methanol. 48,49 Prospective applications may include pH-sensing 50 or probing of hypoxia. 47,49 The later may in particular become feasible via storage of polarization in a 15 Nnitro group of metronidazole which has a T 1 of about 10 minutes in methanol. 51 Currently the main challenge is to discover molecules that are biological relevant, have long T 1 and can be hyperpolarized to a large degree. Here, we are tackling this challenge and introduce classes of compounds that meet these requirements. Our particular focus is thereby on pyridinium, a compound already relevant in drug applications. [52][53][54][55][56] Experimental The synthesis of the labelled compounds was conducted as follows: to yield 1, we prepared 15 N-pyridine-d 5 starting from protonated 15 N-pyridine, oxidation with meta-chloroperoxybenzoic acid (m-CPBA) followed by H-D exchange reaction under microwave condition in D 2 O (Scheme 1A). Further reduction with PCl 3 in CH 2 Cl 2 yielded 15 N-pyridine-d 5 . 57 Finally, quaternization of 5 was accomplished by the treatment with allyl bromide-d 5 (6) in EtOAc to yield 1 as colourless solid. 58 In order to synthesize the aniline derivatives 2 and 3, we rst synthesized 15 N-aniline-d 5 (7) in a two-step procedure from benzene-d 5 . 59 Mono-allylation of 7 with allyl bromide-d 5 (6) in the presence of K 2 CO 3 and further treatment with CD 3 I in presence of DIPEA, yielded 2 (Scheme 1B). Stirring of 2 in neat CD 3 I leads to the quaternary aniline derivative 3. Further experimental details can be found in the ESI. †

Result and discussion
We have synthesized and investigated a library of 15 N-enriched compounds and report on two novelties: rstly, we have discovered an aniline derivative containing a tertiary amine with a long T 1 of about 10 minutes in methanol-d 4 (MeOD). This is of particular interest since it demonstrates that uncharged nitrogen species, in addition to quaternary compounds, have potential to store polarization for long periods and opens up new possibilities to design contrast agents with lipophilic moieties. Secondly, we are introducing a new class of compounds that can be hyperpolarized and possesses a T 1 of about 8 minutes in water: quaternary pyridinium derivatives. Quaternary pyridinium is a core structure found in many molecules which has been used for investigations of neurodegenerative diseases 60 as well as in drug design and drug-delivery approaches. [52][53][54][55][56] We furthermore present the hyperpolarization of the library of compounds via PHIP and a pulsed transfer method to enhance the 15 N signals. Generating contrast agents in aqueous media becomes possible by utilizing rhodium nanocatalysts (NAC@Rh) that promote the hydrogenation reaction with para-H 2 in water. 30 Table 1 presents the investigated compounds and at the top the general scheme of how the investigated compounds are hyperpolarized with para-H 2 . The precursor compounds prior to hydrogenation are a pyridinium derivative (1), a tert-amine derivative of aniline (2) and a quaternary nitrogen derivative of aniline (3). We have perdeuterated all of the precursors to prolong 15 N-T 1 by weakening dipolar couplings, as compared to the protonated counterparts. As an unsaturated moiety to which para-H 2 will be added during the hydrogenation step, we have chosen deuterated allyl groups. The rationale behind this choice is twofold: rst, the added protons from para-H 2 aer the hydrogenation will be one extra bond away as compared to the vinyl derivatives, thus reducing dipolar interactions that potentially shorten T 1 . Second, the scalar coupling network in the hydrogenation products 1a-3a are thought to be an ideal spin system to apply the recently developed ESOTHERIC (efficient spin order transfer to heteronuclei via relayed INEPT chains) spin order transfer sequence to hyperpolarize the 15 N spins. 61,62 This is because the 3 J H,N coupling is larger than 4 J H,N (see ESI †) and the protons are weakly coupled.
Prior to performing hyperpolarization experiments, we determined 15 N-T 1 for the unsaturated precursor molecules 1-3 in D 2 O, MeOD or mixtures thereof to increase the molecule's solubility. The 15 N-T 1 values obtained in different solvents and at various magnetic elds are summarized in Table 1. For the precursor molecules it is noteworthy to mention that 15 N-T 1 of the tert-amine 2 has a 15 N-T 1 of 570 AE 40 s in MeOD (this compound was not soluble in water) at high eld and the quaternary ammonium compound 3 displays a 15 N-T 1 of 420 AE 100 s in D 2 O. For the unsaturated pyridinium derivative 1, we discovered a 15 N-T 1 of 220 AE 30 s at high eld in D 2 O.
Since we found 15 N-T 1 values of several minutes for all precursor compounds, we performed hydrogenation reactions and investigated 15 N-T 1 of the hydrogenation products. This was done by hyperpolarizing the 15 N nuclei and measuring the polarization decay with low ip angle pulses as described in the next paragraph and in the ESI. † Our rst observation was that the anilinium derivative 3a decomposes upon hydrogenation. This may reect that trimethylanilinium is typically used as a methylation agent 63 and not stable enough for hyperpolarization studies with para-H 2 . Moreover, a similar kind of degradation was reported on 15 N-propargylcholine while performing PHIP. 42 In addition to this, Shchepin et. al. reported lack of the successful 15 N hyperpolarization on other choline derivatives using 15 N-enriched PHIP precursors. 64 The 15 N-T 1 of the tert-amine 2a is strongly reduced aer hydrogenation to 150 AE 20 s. Lastly, the pyridinium derivative 1a has a T 1 of 120 AE 10 s at high eld in D 2 O, but reaches 500 AE 30 s (about 8 minutes) when the eld is lowered to 0.1 T (see also Fig. S1 †). With Decomposed -a A general scheme of hyperpolarization followed by polarization transfer to 15  respect to T 1 , the main relaxation source at high eld appears to be chemical shi anisotropy (CSA). This offers possibilities to make the compound applicable for studies in clinical scanners. Given its long 15 N-T 1 at low eld in water and being an important structure in a variety of biomolecules or drugs, the pyridinium derivative is the most promising compound discovered among the investigated compounds here for future applications.
To obtain the hyperpolarized products, compounds 1-3 were hydrogenated with para-H 2 under two experimental conditions: for preparation in MeOD, we used the homogeneous Rh-catalyst [Rh(dppb)(COD)][BF 4 ] (dppb: diphenylphosphino butane, COD: cyclooctadiene). For hyperpolarization in D 2 O, we used an N-acetylcysteine-capped Rh-nano-catalysts (NAC@Rh). 30 The enrichment of H 2 in its para-state was 80%, as determined experimentally. At rst, we have investigated the 1 H polarization and subsequently the 15 N polarization following the ESOTHERIC sequence. 61,62 The results are summarized in Table 2.
As compound 3a did not form during hydrogenation, no hyperpolarization data is reported here for either the homogeneous or heterogeneous catalyst. Compound 2 turned out to be insoluble in D 2 O; therefore, we chose an equimolar mixture of MeOD and D 2 O for dissolving the heterogeneous catalyst for PHIP experiments. We have found 1% polarization of 1 H and 15 N nuclei respectively in the hydrogenated compound 2a, whereas multiple polarized products were observed in MeOD with the homogeneous catalyst. This result demonstrates that heterogeneous catalysts provide new opportunities for polarizing nitrogen containing compounds that may not be accessible with the standard homogeneous catalyst.
With respect to the pyridinium derivative, we observed signicant 1 H polarization of 11% AE 1.3% in 1a using the homogeneous catalyst in MeOD. We succeeded in transferring this polarization to the 15 N-spin with a signal enhancement (3) of 32 000 (P ¼ 7.4% AE 0.6%) compared to thermal polarization at B 0 ¼ 7 T at 320 K in MeOD. For improved biocompatibility, we performed polarization experiments with the heterogeneous catalyst in water and achieved a highest polarization of 3.1% (3 ¼ 15 000-fold compared to the thermal signal at 353 K, Fig. 1) and an average 2.3% polarization. The spectrum of the hyperpolarized compound in water as well as the T 1 -experiment (inset) with small tip angle pulses at 0.1 T is depicted in Fig. 1.

Conclusions
In conclusion, we have introduced and synthesized perdeuterated 15 N-allyl-pyridinium (1) and -aniline derivatives (2 & 3). We succeeded in forming hyperpolarized addition products of 1 and 2 utilizing para-H 2 . Most notably, a 15 N-pyridinium derivative (1a) provided strong 15 N-polarization of P ¼ 7.4% in methanol and P ¼ 2.3% in water compared to thermal polarization. Polarization in water was achieved via rhodium nanocatalysts that although heterogeneous PHIP catalysts are still in an early development stage show here the possibility to signal enhance molecules that are not polarizable with standard homogeneous metal complexes. In water at 0.1 T eld, we discovered a long 15 N-T 1 of about 8 min. We also found that the tert-amine 2 features notably a slow relaxation time of 10 min for 15 N-nuclei in methanol. This is despite the fact that it is not a quaternary nitrogen compound, and thus could be used as a hydrophobic 15 N-labelled tracer. Overall, our presented studies introduce new possibilities for the molecular design of contrast agents and storage capabilities of hyperpolarized spin states. It is noteworthy to mention that out of all compounds studied here, the highest levels of hyperpolarization ( 1 H and 15 N) were found in pyridinium derivatives, a molecular species present in many bio-relevant molecules. Longer relaxation times of 15 N nuclei of these compounds in combination with targeting moieties will potentially in the future ensure long traceability and opportunity to deliver the hyperpolarization in organisms for biomedical imaging applications.

Conflicts of interest
There are no conicts to declare.  funding and Prof. Christian Griesinger for access to his equipment and facilities. The authors furthermore thank Dr Sergey Korchak and Dr Salvatore Mamone for support. We also thank Dr V. Belov and the chemical synthesis facility for performing the microwave reaction.