Achieving High Levels of NMR‐Hyperpolarization in Aqueous Media With Minimal Catalyst Contamination Using SABRE

Abstract Signal amplification by reversible exchange (SABRE) is shown to allow access to strongly enhanced 1H NMR signals in a range of substrates in aqueous media. To achieve this outcome, phase‐transfer catalysis is exploited, which leads to less than 1.5×10−6 mol dm−3 of the iridium catalyst in the aqueous phase. These observations reflect a compelling route to produce a saline‐based hyperpolarized bolus in just a few seconds for subsequent in vivo MRI monitoring. The new process has been called catalyst separated hyperpolarization through signal amplification by reversible exchange or CASH‐SABRE. We illustrate this method for the substrates pyrazine, 5‐methylpyrimidine, 4,6‐d 2‐methyl nicotinate, 4,6‐d 2‐nicotinamide and pyridazine achieving 1H signal gains of approximately 790‐, 340‐, 3000‐, 260‐ and 380‐fold per proton at 9.4 T at the time point at which phase separation is complete.


Materials
All of the experimental procedures associated with this work were carried out under nitrogen using standard Schlenk techniques. The solvents used were dried using an Innovative Technology anhydrous solvent system, or distilled from an appropriate drying age nt under nitrogen. The catalyst precursor [Ir(IMes)(COD)Cl] (1) employed in this work was synthesized by established procedures according to literature methods. 1 Deuterated chloroform (CDCl3), deuterated water (D2O), deuterated ethanol (EtOD), pyrazine (2), 5methylpyrimidine, and pyridazine were purchased from Sigma Aldrich and used as supplied. 4,6-d2-methyl nicotinate and 4,6-d2nicotinamide were synthesized as described by Rayner et al. 1 The level of chloroform contamination in the aqueous phase was assessed by signal comparison to a known internal DMSO reference.

SABRE analysis
All samples have been prepared in standard 5 mm NMR tubes equipped with Young taps. In a typical experiment set, arrays of NMR measurements were collected using either 4 equivalents of substrate (1-fold excess) to 5 mM of iridium or 20 equivalents of substrate (17-fold excess) to 5 mM iridium dissolved in mixtures of CDCl3 and D2O. The exact composition of each sample analyzed is specified detailed in Table S1. After adding p-H2 at 3 bar pressure, 1 H NMR spectra were recorded using π/2 excitation pulses after shaking the sample in a magnetic field between 0 and 65 G. Enhancement factors were calculated by using the ratio of the integral areas of individual resonances in the hyperpolarized NMR spectrum and the corresponding spectrum collected under normal H2 and Boltzmann equilibrium conditions respectively.
Similar experiments have been performed in order to assess the possibility to polarize heteronuclei such as 13 C and 15 N. 13 C hyperpolarization experiments have been performed using polarization transfer fields between 0 and 65 G, while 15 N data has been acquired at 0 G field, by inserting the tube and shaking the sample in a μ-metal shield designed to shield the sample form the environmental magnetic field.
The 13 C and 15 N NMR signal enhancements were calcualted relative to an external standard using the following formula: Spol × Npol Sref unpol × Nref unpol E = Signal enhancement Spol = Integral of polarized samples signal Sref unpol = Intergral of unpolarized reference ( 15 N, 13 C labelled urea of known concentration) E = Npol = Effective concentration of 13 C or 15 N in polarized sample Nref unpol = Effective concentration of 13 C or 15 N in unpolarized reference

1D MRI experiments
All 1D MRI experiments have been performed using a 400 MHz Bruker Avance spectrometer equipped with a z-gradient of maximum strength of 0.536 T/m. As a function of the observed nucleus, the experiments can be separated in two categories: • 1D projections of the 2 H signal amplitude on the z direction (parallel to the magnet bore and the NMR sample tube). These experiments have been performed with the purpose of observing the spatial distribution of the two solvents.
• 1D projections of the 1 H signal amplitude on the z direction (parallel to the magnet bore and the NMR sample tube). In a typical experiment, 128 projections have been acquired immediately after the shaking process. These experiments have been performed with the purpose of observing the spatial distribution of the substrate in thermal equilibrium conditions, after shaking in t he presence of o-H2 and well as after shaking in the presence of p-H2.

2D MRI experiments
All 2D MRI experiments have been performed using a 400 MHz Bruker Avance spectrometer equipped with a micro imaging gradient set with maximum amplitude of 1T/m and a double resonance birdcage coil with a diameter of 30 mm. All samples have been prepared in 10 mm diameter standard NMR tubes. The hyperpolarization step has achieved by shaking the sample under p-H2 for 10 seconds at 65G in the stray field of the magnet. Images have been acquired using the rapid acquisition schemes based on spin echoes (RARE) and gradient echoes respectively (Steady State Free Precession-SSFP). The 1 H MRI acquisition parameters were, as follows: • For SSFP images: field of view 60 x 60 mm 2 , slice thickness 2.5 mm, matrix size 64 x 64 (zero-filled to 128 x 128) leading to a nominal 2D resolution of 0.94 x 0.94 mm 2 (digital resolution 0.47 x 0.47 mm 2 ). TE/TR 2/4 ms. Repetition time between two consecutive image acquisition: 600 ms. Excitation pulse angle: 30°.
A sine bell squared filter has been applied prior to the Fourier transform to minimize the contribution of noise in the images. Data post processing has been done using home developed routines in Prospa (Magritek) and MATLAB (MathWorks).

Single Voxel Spectroscopy (SVS) experiments
SVS experiments have been performed using the setup described in section 1.4. 1 H SABRE hyperpolarized spectra of pyrazine have been acquired by selecting two separate voxels (matrix size 5x5x5 mm 3 ) located parallel to the tube's vertical axis, as depicted in Figure S1. In order to minimize artefacts caused by diffusion and turbulence on the time scale of the experiment, an outer volume suppression scheme (OVS) has been used prior to data acquisition. Figure S1. Schematic drawing of the NMR tube containing the sample and the position of the two voxels used for spectra acquisition.

Optimization experiments performed on pyrazine dissolved in CDCl3/D2O mixtures
In order to determine the optimal composition of the biphasic solvent that is to be used for SABRE catalysis, 1 H hyperpolarization experiments have been performed on a series of samples in which the ratio of the inorganic to organic phase has been varied. For each sample between five and ten hyperpolarization experiments were performed by shaking the sample in the stray field of the magnet at ~ 30 G for 10 seconds and immediately acquiring a 1 H spectrum after the hyperpolarization transfer step had been completed. The enhancement factor presented is the average value, taken from the integral areas of the hyperpolarized resonances divided by the corresponding area in a spectrum that was acquired under Boltzmann equilibrium conditions. The composition of each sample, together with the corresponding enhancement and associated errors, are presented in Table S2.   Table S1).

SABRE hyperpolarization of pyrazine in CDCl3/D2O mixtures in the presence of other salts
Sample 3a: 1-fold excess pyrazine + NaOH A sample was prepared using 1-fold excess of ligand relative to the active iridium catalyst (5 mM) in a mixture of 0.3 ml CDCl3 and 0.3 D2O. NaOH (0.0085 mmol) was then added as the separation promoting agent. When examining the result obtained after acquiring a 90° spectrum under Boltzmann equilibrium conditions, it can be seen that NaOH addition leads to a clear difference in the chemical shift of the pyrazine resonance being detected in the two different solvents. The pyrazine 1 H NMR signal that was typically observed as a singlet in the previous experiments now appears now as a sharp, narrow peak (corresponding to the ligand dissolved in the water phase) and a low-intensity, broad resonance, shifted downfield from the former, corresponding to the substrate that is present in the organic phase ( Figure S8). When examining the analogous spectra acquired after SABRE hyperpolarization, the chemical shift difference cannot be resolved if the measurement is performed immediately after the polarization transfer step, due to the extreme line broadening artefacts that are introduced by the fact that these data are acquired while the mixture separates and stabilizes (motion artefacts reduced). However, if a) b) c) the mixture is allowed to separate prior to data acquisition, the chemical shift difference between the resonances in the distinct organic and inorganic phases can be clearly detected ( Figure S9).

Sample 3b: 1-fold excess pyrazine + Na2CO3
A sample was prepared using 1-fold excess of ligand relative to 5 mM of Ir dissolved in a mixture of 0.3 ml CDCl3 and 0.3 D2O. Na2CO3 was added as a separation agent. When examining the result obtained after acquiring a spectrum under Boltzmann equilibrium conditions, it can be seen that, similarly to the data presented above, that chemical shift separation of the pyrazine resonance as a function of solvent is seen. This chemical shift difference is preserved in the 1 H hyperpolarized spectrum that is acquired after the mixture is allowed to separate. However, as Na2CO3 is a much milder base (pKb 3.67) when compared to NaOH (pKb 0.2). The two substrate resonances appear much narrower and closer together ( Figure S10) with Na2CO3.

Sample 3c: 1-fold excess pyrazine + NaHCO3
A sample was prepared using 1-fold excess of ligand relative to 5 mM of Ir dissolved in a mixture of 0.3 ml CDCl3 and 0.3 D2O. NaHCO3 was then added as the separation agent. When examining the result obtained after acquiring a spectrum under Boltzmann equilibrium conditions, a chemical shift separation similar to the examples presented above can be detected. The analogous spectrum acquired under hyperpolarized conditions, after the mixture was allowed to separate, exhibits several resonances instead of the two observed previously ( Figure S13).

Sample: 1-fold excess pyrazine + CH3COO -NH4 + (3e)
A sample was prepared using 1-fold excess of ligand relative to 5 mM of Ir dissolved in a mixture of 0.3 ml CDCl3 and 0.3 D2O. NH4CO3CH3 was added as a separation agent. The 1 H spectrum acquired under thermal equilibrium conditions exhibits two sharp resonances accompanied by a very broad peak located downfield ( Figure S16).  Sample 3f: 1-fold excess pyrazine + Cl -NH4 + A sample was prepared using 1-fold excess of ligand relative to 5 mM of Ir dissolved in a mixture of 0.3 ml CDCl3 and 0.3 D2O. NH4Cl was added as the phase separation agent. The 1 H spectrum acquired under thermal equilibrium conditions exhibits two resonances ( Figure S18) for the two phases. Phase separation after shaking proceeds as detailed in Figures S18b and S18c.

Sample 3g: 17-fold excess pyrazine + NH4Cl + NaCl
A sample was prepared using 17-fold excess of ligand relative to 5 mM of Ir dissolved in a mixture of 0.3 ml CDCl3 and 0.3 D2O. NH4Cl and NaCl were added to promote phase separation. The 1 H spectrum acquired under thermal equilibrium conditions exhibits two sharp resonances accompanied by a very broad peak located downfield ( Figure S19). While a poor polarization level is observed, the phase separation time is also slow.

SABRE hyperpolarization of pyrazine in CDCl3/D2O mixtures
Figures S20 and S21 reflect sample 1g (without salt) and present a series of time-resolved 1D projections which detail how phase separation proceeds. Figure S20 details a projection of the 2 H distribution (left) and reflects the dominant D2O solvent. Figure S21 (left) shows the weak 1 H response of pyrazine, acquired under thermal polarization conditions. Figure S21 (right) shows the results of shaking this sample under p-H2. Hyperpolarization in the aqueous phase results even though the CDCl3 doping is minimal.

SABRE hyperpolarization of other substrates in CDCl3/D2O mixtures
In order to demonstrate that the approach of CASH SABRE is widely extendable to a whole range of substrates, we have performed hyperpolarization experiments on a set of samples prepared using a 1:1 organic: aqueous ratio (0.3 ml : 0.3 ml) and a series of ligands which present significant interest in terms of biomedical applications. A 5 mM catalyst loading was employed in conjunction with 20 mM of the agent and 0.16w/v NaCl. The agents analyzed are presented in Scheme S1. Figures S53-S56 present 1 H NMR spectra that detail the SABRE hyperpolarization of the 1 H response of these samples and confirm the wider applicability of CASH SABRE. The very high polarization level resulting for d2-methylnicotinate is particularly notable. Figures S57-S68 detail a series of time-resolved 1D projections on these samples to illustrate how phase separation proceeds in a similar way to that presented earlier.

2D MRI of SABRE hyperpolarized pyrazine in CDCl3/D2O mixtures in the presence of NaCl.
Prior to 2D data acquisition, 1D projections of the signal derived from the hyperpolarized protons of pyrazine have been acquired on samples prepared with 1-fold excess of ligand, a 17-fold excess of ligand and a 17-fold excess of ligand in the presence of 50 μl of EtOD. These results are presented in Figures S71 -S75. These data show that in the absence of EtOD the pyrazine is uniformly distributed in both CDCl3 and D2O, with average ratios of ~8:3 for the sample containing 1 fold excess and 5:1.5 for the sample containing 17-fold excess. EtOD addition therefore promotes the transition of the ligand dissolved in CDCl3 towards the interface with D2O ( Figure S72, right).    In order to assess the evolution of the hyperpolarized signal as a function of time in both phases, gradient echo images acquired using a low flip angle pulse for excitation have been acquired each 0.8 s after the hyperpolarization step. The results, presented in Figure S76, show that, in a first instance, a more intense signal is obtained in the CDCl3 phase. After longer acquisition times, the signal intensity in the D2O phase becomes comparable with that of the signal in the CDCl3 phase, demonstrating the continuous diffusion of hyperpolarized pyrazine in water. It is worth noting that, in the sample containing EtOD, the highest signal intensity is obtained at the interface, as previously shown by the 1D projections. This reflects the possibility of using SABRE in the study of interface processes in multi-phase systems.

2D MRI of SABRE hyperpolarized pyrazine in CDCl3/D2O mixtures in the presence of other salts.
Prior to 2D data acquisition, 1D projections of the signal derived from the hyperpolarized protons of pyrazine have been acquired on samples 3f and 3g, prepared with 1-fold excess of ligand and 3.3 mg NH4 + CH3COO -, and 17-fold excess of ligand and 3.3 mg NH4 + CH3COO -+ 1mg NaCl respectively. The results are presented in Figure S78. 2D images of the samples have been acquired using a RARE protocol with centric k-space sampling in thermal equilibrium conditions and after 10 s of shaking in the stray field of the magnet in the presence of p-H2. The results are depicted in Figure S79 for the sample prepared with a 1-fold excess of pyrazine and Figure S80 for the sample prepared with 17-fold excess of pyrazine.

2D MRI of SABRE hyperpolarized pyridazine in CDCl3/D2O mixtures in the presence of NaCl.
In order to demonstrate that the CASH-SABRE approach, exemplified so far mainly for pyrazine, can be extended to other substrates, we have also tested the possibility of MRI detection of hyperpolarized pyridazine, which, as previously shown, tends to successfully concentrate in the aqueous phase, thus making this substrate a good candidate for in vivo MRI applications.
2D images of a sample containing 7-fold excess pyridazine have been acquired using the RARE protocol described previously. The results, presented in Figure S81, show that, after phase separation, excellent signal enhancement can be detected in the aqueous phase (top of the tube, see sagittal image) and almost no hyperpolarized pyridazine response is present in the organic phase. Figure S81. 2D MRI images the 1 H signal of pyradizine (7-fold excess + 20 mg NaCl). From left to right: sagittal image of the whole tube, axial projection through the part of the tube containing CDCl3, axial projection through the part of the tube containing D2O. Top: thermal equilibrium conditions, bottom: SABRE hyperpolarized. We note that the images presented in Figure S81, Figure S64 and Figure S68 show that both 4,6-d2-nicotinamide and pyradizine present signal in the aqueous phase where they are highly soluble.

Quality control by UV-Vis and transport between the phases
a) b) Figure S82. (a) UV-visible spectrum of the chloroform layer after dilution (100 times). In this layer the catalyst is characterized by two absorption bands at λ1 = 374 nm (ε = 6800 l.mol -1 .cm -1 ) and λ2 = 500 nm (ε = 2200 l.mol -1 .cm -1 ) with pyrazine providing the λ1 261 nm signal. (b) UV-visible spectrum of the D2O layer after dilution of 10 times. In this layer, the pyrazine proved to be the main product.
A UV spectrum was recorded to monitor the amount of catalyst present in the aqueous phase 10 seconds after mixing. As detailed in Figure S83 no signal is seen which places an upper limit of 1.5 x 10 -6 mol dm -3 on the iridium concentration based on [Ir(H)2(IMes)(pz)3]Cl. Further working using ICPMS is on-going to quantify this level more accurately. The sample used mimicked that of 2c. Figure S83. UV-visible response of the water phase that results 10 seconds after phase separation has started.
Relative rates of pz transfer from the aqueous phase into chloroform were assessed, with, and without NaCl, by UV monitoring as detailed in Figure S85. This involved layering a sample of H2O containing pyrazine over an equivalent volume of CHCl3 that contained no pyrazine. It appears that while the relative partioning of pz between these phases was unaffected by added NaCl (within the error of a control measurement) but the presence of NaCl was found to increase the rate of pz transfer between the phases. Figure S85. UV-visible response of pz in the chloroform phase as a function of time after the shake.