A Hyperpolarizable 1H Magnetic Resonance Probe for Signal Detection 15 Minutes after Spin Polarization Storage

Abstract Nuclear magnetic resonance (NMR) and magnetic resonance imaging (MRI) are two extremely important techniques with applications ranging from molecular structure determination to human imaging. However, in many cases the applicability of NMR and MRI are limited by inherently poor sensitivity and insufficient nuclear spin lifetime. Here we demonstrate a cost‐efficient and fast technique that tackles both issues simultaneously. We use the signal amplification by reversible exchange (SABRE) technique to hyperpolarize the target 1H nuclei and store this polarization in long‐lived singlet (LLS) form after suitable radiofrequency (rf) pulses. Compared to the normal scenario, we achieve three orders of signal enhancement and one order of lifetime extension, leading to 1H NMR signal detection 15 minutes after the creation of the detected states. The creation of such hyperpolarized long‐lived polarization reflects an important step forward in the pipeline to see such agents used as clinical probes of disease.

Nuclear spin hyperpolarization has evolved as one of most important developments in NMR and MRI in recent years as it starts finding applications in human metabolomics, [1][2][3][4] where their detection holds great potential to create tools for the diagnose of diseases.A mong the various hyperpolarization techniques, [5] dynamic nuclear polarization (DNP) [6] and para-hydrogen-induced hyperpolarization (PHIP) [7] are two of the most popular techniques.I n2 009, an important variant to the PHIP technique [8,9] termed SABRE [10] was described that no longer required am olecular change to use para-hydrogen (p-H 2 )d erived hyperpolarization. Instead, in SABRE am etal catalyst reversibly binds p-H 2 and the hyperpolarization target. Thed ormant magnetism of p-H 2 transfers into the target through the scalar-coupling framework of these catalysts as illustrated in Scheme 1. Since its inception, this method has stimulated many developments which include the hyperpolarization of al arge class of molecules comprising of 1 H, 13 C, 15 N, and 31 Pn uclei. [11][12][13][14] When compared to dissolution DNP,S ABRE provides alow cost alternative that takes just seconds to hyperpolarize the agent in acontinuous process that, while being inherently simple in concept, can be augmented by rf excitation. [15] In order to advance the future integration of SABRE with molecular imaging,i ti sh ighly desirable to create hyperpolarized targets,t he magnetism of which survives transfer into ad iagnostically relevant region of the body.T his requirement is based on observations with DNP and PHIP, techniques that have been used to successfully prepare and detect 13 C-based magnetization in vivo [3,4] and also show potential for 15 N-based agents. [16] These reported low-gamma nuclei-based in vivo studies employ relatively slowly relaxing Zeeman-derived magnetization in order to overcome the rate of signal loss,b ut these approaches inherently measure aw eaker response than would be provided by 1 Hd etection, whilst requiring al arger gradient strength for equivalent spatial resolution. [17] Instead, the detection of hyperpolarized 1 Hn uclei is feasible on all existing clinical MRI systems as they routinely probe aH 2 Or esponse.H ence,w hile hyperpolarized high-gamma 1 Hnuclei detection in vivo is therefore thought to be challenging because of faster relaxation it reflects the optimal direction for clinical MRI to follow.
Formany years,the long-lived singlet state associated with p-H 2 was used to simply access hyperpolarization. [7] However, in 2004 Levitt and co-workers showed that it was possible to create analogous non-magnetic singlet states more generally between pairs of spin-1/2 nuclei that are magnetically inequivalent and have lifetimes that are much longer than T 1 . [18] Consequently,the spin-lattice relaxation time constant T 1 is no longer the time-limiting barrier for nuclear spin memory and such long-lived singlet states (LLS) reflect an Scheme 1. Schematic depiction of the SABRE hyperpolarization technique.
important and rapidly developing area of NMR spectroscopy. [19][20][21][22] Related long-lived states have been prepared under chemically modifying PHIP. [23,24] More recently,T heis et al. demonstrated that long-lived 15 Nm agnetization can be created and integrated into the chemically benign SABRE approach. [25] Aparallel approach of using SABRE to prepare hyperpolarized LLS in weakly coupled 1 Hspin pairs have also been reported but the magnetization lasted under 90 s. [26,27] Thec hoice of spin system is critical in developing av ery long lifetime [28] and providing access to hyperpolarization by SABRE. Here,weuse the pyridazine derivatives of Figure 1.
We selected this class of agent because the pyridazine motif is found in an array of pharmacologically active agents and their future in vivo imaging may yield clinically diagnostic information. [29,30] We also needed to identify atarget that possesses ab inding site for SABRE, and an optimally coupled pair of 1 Hn uclei that resonate at similar frequencies but are magnetically inequivalent.
We started out by considering pyridazine (I)and the need to break the symmetry between H-4 and H-5 in order to generate singlet states by rf pulses.This was achieved in II by replacing one of its two a-proton sites with am ethyl group. We then replaced its remaining a-proton with a 2 Hlabel in III to remove the proton coupling that could reduce the lifetime of the state.Putting 2 Hlabels into both of these positions (IV) makes it possible to further isolate them before preparing the dialkylated forms V and VI where we create more sterically shielded binding sites whilst maintaining the symmetrybreaking process (see Section S3 in the Supporting Information). We expected that this strategy would allow us to explore how to optimally influence relaxation and hence improve lifetime.
Surprisingly,t he chemical shifts of the target spins in III and IV proved to be highly solvent-dependent, while those of I, II, V,and VI were not. Figure 2shows an array of 1 HNMR spectra of target IV in aseries of CD 3 OD-CDCl 3 mixtures to illustrate this point. In 100 %C DCl 3 ,t he chemical shift difference between H-4 and H-5 (Dd, w 0 Dd/2p in a400 MHz spectrometer) is 13.6 Hz. Effectively,a st he J-coupling between them is 8.5 Hz, af irst-order spin system at high field. Remarkably, Dd reduced to only 1Hzwhen in CD 3 OD and as trongly coupled spin pair (Dd ! J)r esults.A s ac onsequence,i ti ss ubject to much smaller chemical shift anisotropy (CSA) mediated relaxation at high field, leading to ap otentially longer LLS lifetime (T LLS ). Furthermore,t he progressive change in Dd between these two extremes with solvent composition means that these systems reflect ar elatively unique opportunity to test the effect of Dd on relaxation without having to complete ah igh-cost study at an array of observation fields.A sp redicted the value of T LLS increases dramatically as Dd falls,reaching 136 sinCD 3 OD when Dd is just 1Hz, but 12.4 si nC DCl 3 where the Dd is 13.6 Hz (Section S6). The T 1 lifetimes were measured by traditional inversion recovery approach, whilst T LLS lifetimes were determined by Levittsprotocol [31] (Section S5).
We tested the applicability of substrates I-VI to hyperpolarization by SABRE method (Section S1). Figure 3a illustrates the result of this process for IV in CD 3 OD solution after 20 sofexposure to p-H 2 as determined at 400 MHz. As expected, substrates I and II polarize well using initial 4 J HH couplings within the catalyst leading to 6.5 %n et 1 Hp olar-   ization rather than the more usual Zeeman level of 0.003 %at this field. Despite the use of unusual 5 J HH coupling for SABRE transfer in III-IV,similar levels of hyperpolarization are seen ( Table 1). Thepresence of asingle methyl substituent does not therefore prevent successful SABRE catalysis (Section S7). However,t he hindered dialkylated pyridazines V and VI do exhibit reduced levels of SABRE enhancement, relative to I (Section S2). Theo ptimum level of hyperpolarization results from transfer in a6 5G field in all cases in agreement with theoretical and simulated calculations (Section S4).
TheM2S-S2M pulse sequence [31] was found most suitable to transfer this polarization into hyperpolarized-singlet states and its subsequent detection (Section S5). State storage was then explored in three ways:a)keeping the sample inside the magnet without further change,b)keeping the sample inside the magnet whilst applying as pin-lock, and c) removing the sample from the magnet to an 10 mT field ( Figure S4). Key results are summarized in Table 1(also Table S4).
Theassociated parameters required for the M2S and S2M conversions were obtained via a J-synchronization experiment in each case (Section S5). We observe a4 5-50 % increase in T LLS lifetime with spin-locking over option one for III-IV.S torage in low-field outside the magnet provides more than 200 %i ncrease in lifetime.D ifferent behavior is observed for V, where its high-field T LLS is just 23 s, but its low-field value is 255 s. Related SABRE-LLS spectra are shown in Figure 3c-e.I ng eneral, we achieve magnetization to singlet conversion of about 66 %i na greement with theoretical estimates. [28] Figure 4s hows the decay of the SABRE-LLS states as af unction of low-field storage time (T S )f or substrates II-VI.E xponential fitting of the experimental points provides the T LLS values to ah igh level of accuracy.T he value for V with the catalyst present is 255 AE 22.8 s, which is an order of magnitiude increment on its corresponding T 1 value.Inafinal refinement, we note that the hyperpolarized results use solutions that contain the SABRE catalyst which influences the T LLS lifetime.I nt he case of V, T LLS extends out to 262 sw hen the catalyst is not present, while for IV it becomes 188.5 s ( Table S3).
In summary,w eh ave demonstrated that SABRE-hyperpolarized 1 Hm agnetization can be stored in relaxation protected singlet states that have lifetimes of several minutes and are an order of magnitude larger than the corresponding T 1 lifetimes.W ea chieve these results in biologically relevant pyridazines that possess an early equivalent 1 Hp air in conjunction with a 2 H-labeling strategy.T he unexpected solvent dependence seen for the chemical shifts between the 1 Hspin pair of III and IV allowed the establishment of aclear link between the corresponding Dd and T LLS ,w hich demonstrates the benefit of as tronger coupling regime.T his approach also results in an in-phase signal which would be desirable for future MRI detection. Our storage strategies allow the successful detection of magnetization 15 minutes after its creation. Thel ow-field storage scheme has the potential to allow the hyperpolarized sample to be transported into the final measurement location whilst keeping any wasteful signal loss to am inimum. These findings therefore illustrate some of the steps needed for successful in vivo measurement with 1 Hdetection. We are currently seeking to develop tracers with higher signal gains and longer lifetimes, and plan to extend this rational-design study into biocompatible media shortly. Table 1: Signal enhancement and lifetimes of substrates (I-VI)dissolved in CD 3 OD. measured in high (9.4 T) and low field ( % 10 mT). The Jcoupling between the 1 Hpair is 8.5 AE 0.1 Hz in all cases.