Aqueous NMR Signal Enhancement by Reversible Exchange in a Single Step Using Water-Soluble Catalysts.

Two synthetic strategies are investigated for the preparation of water-soluble iridium-based catalysts for NMR signal amplification by reversible exchange (SABRE). In one approach, PEGylation of a variant N-heterocyclic carbene provided a novel catalyst with excellent water solubility. However, while SABRE-active in ethanol solutions, the catalyst lost activity in >50% water. In a second approach, synthesis of a novel di-iridium complex precursor where the cyclooctadiene (COD) rings have been replaced by CODDA (1,2-dihydroxy-3,7-cyclooctadiene) leads to the creation of a catalyst [IrCl(CODDA)IMes] that can be dissolved and activated in water-enabling aqueous SABRE in a single step, without need for either an organic cosolvent or solvent removal followed by aqueous reconstitution. The potential utility of the CODDA catalyst for aqueous SABRE is demonstrated with the ∼(-)32-fold enhancement of 1H signals of pyridine in water with only 1 atm of parahydrogen.

Introduction

Because of their inherent advantages (including high spatiotemporal resolution, lack of ionizing radiation, and the ability to spectrally distinguish multiple signal sources), magnetic resonance imaging (MRI)-based molecular imaging[1,2] techniques promise to revolutionize clinical imaging—from the screening and diagnosis of disease, to the assessment of treatment response. However, the inherently low detection sensitivity of conventional magnetic resonance techniques makes it challenging to detect and track low-concentration species in vivo, such as gas species in lung spaces or metabolic biomarkers in blood or other tissues. Hyperpolarization[3] techniques like dissolution dynamic nuclear polarization (d-DNP),[4,5] spin-exchange optical pumping (SEOP),[6,7] and parahydrogen induced polarization (PHIP)[8,9] offer the possibility of overcoming the problem of low agent concentration by increasing the nuclear spin polarization—and hence MR signal—by several orders of magnitude.

Signal amplification by reversible exchange (SABRE)[10] is a relatively new hyperpolarization technique pioneered by Duckett, Green, and co-workers in 2009.[11,12] In SABRE, an organometallic catalyst is used to colocate a molecular substrate to be hyperpolarized and parahydrogen (pH2)—a source of pure nuclear spin order. Like traditional PHIP,[8,9,13−19] SABRE is of interest because it is cost-effective, potentially continuous, scalable, and rapid (achieving polarization enhancement in seconds).[10−12,20−40] However, unlike traditional PHIP, SABRE does not require permanent alteration of the substrate to hyperpolarize it.[11] Since its inception, considerable effort has been put forth to broaden the applicability of SABRE by investigating alternative catalyst structures,[21,28,41−45] improving the nuclear spin polarization achieved for protons[34,46] and various heteronuclei[30,32,47−50] (including through the application of variable applied DC and AC fields), demonstrating high-resolution imaging[25,50] (including at low magnetic field[51]), widening the range of amenable substrate types,[36] achieving enhancement in the limits of both low-[29,52] and high-concentration[49] agents (including in complex mixtures[20]), and demonstrating SABRE with (and separation/reuse of) heterogeneous microscale/nanoscale catalysts.[53,54]

Other efforts have concerned the extension of SABRE to aqueous environments. Because of the poor aqueous solubility of the “standard” SABRE catalyst ([IrCl(COD) (IMes)],[46,55,56] where “COD” = cyclooctadiene and “IMes” = 1,3-bis(2,4,6-trimethylphenyl)imidazol-2-ylidene), recent promising efforts have relied on organic cosolvents to achieve SABRE in aqueous/organic mixtures[34,45,50,57] However, in other previous work we recently found that the chemical changes that accompany this catalyst’s activation also endow it with water solubility;[57] following activation, the organic solvent may be completely removed and the activated catalyst can be subsequently reconstituted in deuterated water to achieve SABRE enhancement.

Here we report our efforts to develop novel homogeneous catalysts that may lead to improved SABRE in aqueous environments, without the need for separate catalyst activation, organic solvent removal, or subsequent aqueous reconstitution.[58,59] Two different strategies were utilized to alter the structure—and hence aqueous solubility—of the original standard catalyst by targeting either the N-heterocyclic carbene moiety or the COD group, respectively (Figure ). For the former, PEGylation[60] of a variant of the aromatic carbene moiety provided much greater aqueous solubility for the catalyst (“7”); however, while that catalyst is SABRE-active in ethanol solutions, it lost activity in >50% water. For the latter, synthesis of a di-iridium complex precursor where the COD rings have been replaced by CODDA (1,2-dihydroxy-3,7-cyclooctadiene) permits creation of a catalyst [IrCl(CODDA)IMes] (“13”) that can be dissolved and activated in water, enabling aqueous SABRE in a single step without need for any organic cosolvent. The potential utility of the CODDA catalyst for aqueous SABRE is demonstrated with the ∼(−)32-fold enhancement of 1H signals of pyridine in water with only 1 atm of pH2. Taken together, these results aid the evaluation of different synthetic approaches for aqueous SABRE that, when improved and combined with other approaches, should help enable a wide range of biological, biomedical, and in vivo spectroscopic and imaging experiments.

Figure 1

Relevant structures for studying SABRE in aqueous environments in the present work, pre (7, 13, and 16) and post (14 and 15) activation in the presence of H2 gas and pyridine (py) substrate. 16 is the “traditional” Ir/IMes SABRE catalyst in its preactivated form, whereas 7 and 13 are the water-soluble PEGylated and cyclooctadiene-diol (“CODDA”) variants, respectively (the numbering of the above structures is explained in the Supporting Information Figures S1 and S2, which, respectively, summarize the synthesis of structures 7 and 13).

Results and Discussion

Exploring SABRE with the PEGylated Catalyst

The PEGylated catalyst 7 was examined to determine its efficacy for SABRE in organic and aqueous environments. SABRE experiments were performed by bubbling pH2 thoroughly into the NMR tube while located outside of the magnet (“low-field”), followed by immediate transfer of the sample into the 9.4 T NMR magnet for “high-field” detection of enhanced 1H NMR spectra. The catalyst was activated via pH2 bubbling in the presence of excess substrate prior to use in SABRE experiments, and the low mixing field was somewhat variable (∼11 ± 5 mT) and was not systematically optimized. Enhancements were recorded for the test substrate pyridine (py); results for all of the experiments described in this work are summarized in Table .

Table 1

Polarization Enhancement (ε) Values for Three Aromatic Proton Sites of Pyridine Observed with Different Catalysts in Aqueous and Nonaqueous Environmentsa

catalyst	solvent	ε (Ho)	ε (Hp)	ε (Hm)
7	100% d6-ethanol	–42	–57	–11
7	100% d6-ethanol	–45	–61	–11
7	13% D2O/87% d6-ethanol	–37	–27	–12
7	13% D2O/87% d6-ethanol	–38	–31	–14
7	43% D2O/57% d6-ethanol	–9.5	–5.7	–1.3
7	43% D2O/57% d6-ethanol	–7.3	–4.9	–0.4
7	63% D2O/37% d6-ethanol	∼0	∼0	∼0
13	100% D2O	–25	–19	–11
13	100% D2O	–32	–25	–16
16	100% D2O	∼0	∼0	∼0

Reported ε values are calculated from spectral integrals and are approximate, with estimated uncertainties of ∼10%. Results from the top two acquisitions for each condition are reported.

In an early set of experiments (not shown), bubbling pH2 at atmospheric pressure gave up to ∼16-fold enhancements for the 1H NMR signals of py in 100% d6-ethanol. The addition of D2O to d4-methanol solutions had lower enhancements than d6-ethanol, with ∼20% D2O/∼80% d4-methanol yielding only ∼6-fold 1H signal enhancements. Higher volume fractions (e.g., 50/50) of D2O in d4-methanol resulted in no observable SABRE enhancements under these conditions.

The lower SABRE enhancements in solutions with increasing water fractions were originally rationalized by the ∼15-fold lower solubility of H2 gas in water compared to that in alcohol-based solvents.[61] To mitigate the H2 solubility limitation of aqueous solutions, the apparatus was altered to allow pH2 pressures of up to ∼60 psi positive pressure (∼5.1 atm total H2 pressure). Bubbling pH2 at 60 psi into a sample containing 100% d6-ethanol, ∼3.5 mM of the catalyst 7, and 35 mM py gave rise to ∼40–60-fold enhancement of the 1H NMR signal from the substrate (e.g., Figure b) compared to the signal acquired at thermal equilibrium (Figure a; the conventional SABRE catalyst 16 is also effective in 100% d6-ethanol[57]). Little dependence on temperature was observed, with similar enhancements attained when the temperature was raised from 301 to 321 K.

Figure 2

SABRE studies with the PEGylated Ir catalyst (7/14) in 100% deuterated ethanol. (a) Thermally polarized reference 1H NMR scan from the solution following activation with pH2 in the presence of excess substrate (py); the spectrum is vertically scaled 5-fold compared to panel b, which shows the successful observation of SABRE enhancement after 1 min of bubbling with ∼5.1 atm of pH2 at ∼11 mT, then transfer to 9.39 T for high-field acquisition; enhancements up to ∼40–60-fold were observed with 3.5 mM catalyst and the given conditions. Panels c and d show spectra from a separate experiment, where no high-field SABRE effect was observed, i.e., where pH2 bubbling/SABRE mixing was performed entirely at 9.39 T (d), compared to a corresponding thermally polarized spectrum (c). (Vertical scale for panels c and d is different from that of panels a and b.) Insets show amplified hydride regions from spectra in panels b and d, respectively.

Next, no SABRE enhancement was observed when pH2 was bubbled in at high field (9.39 T; Figure , parts c and d), unlike the case with the “standard” NHC-Ir catalyst, 16.[24,57] Also unlike the case with 16, no strong, purely absorptive signal at ∼(−)22.8 ppm is observed from magnetically equivalent hyperpolarized hydride spins on the activated catalyst structure. Instead, the hydride region exhibits two relatively weak dispersive doublets at ca. −22.2 and ∼−23.1 ppm. These dispersive signals are reminiscent of the enhanced hydride resonances from organometallic catalysts explored previously with PHIP (e.g., RhH2(PPh3)3Cl[13]) and, thus, are tentatively assigned to the two hydride sites on the activated catalyst (14) rendered effectively inequivalent by the broken symmetry of the PEGylated N-heterocyclic carbene. A pair of additional, much weaker dispersive signals (at ca. −22.6 and −25.9 ppm) likely arise from inequivalent hydride sites on a similar structure to 14 originating from a different chemical pathway. The absence of a high-field SABRE effect is likely a combination of inefficient conversion of spin order from pH2 at high field and the lack of strong z-magnetization of the hydride spins, and is consistent with the current picture for the high-field SABRE mechanism—cross-relaxation akin to the spin-polarization induced nuclear Overhauser effect.[24,57,62,63]

As shown in Figure , parts a and b, modest aqueous fractions (∼13% v/v) had only a minor negative effect on SABRE enhancement (maximum |ε| ∼ 40). Here, the concentration of D2O is already orders of magnitude higher than the concentrations of the catalyst and substrate. Bringing the water fraction to nearly 1:1 dropped the SABRE enhancement by ∼5-fold (Figure , parts c and d); this observation is in reasonable agreement with the ∼15-fold lower solubility of H2 in water versus alcohol-based solvents.[61] However, higher mole fractions of water (e.g., Figure d, inset) have not yielded observable enhancements to date. While this second set of experiments represents a marked improvement over the first in terms of both larger enhancements and larger aqueous fractions for the solvent, the origin of the absence of SABRE at higher aqueous fractions remains unclear. One hint may lie in the changes to the hydride region of the spectrum. For example, while the primary dispersive resonances at ca. −22.2 and ∼−23.1 ppm remain in the spectrum from the ∼13% v/v solution (Figure b, inset), overall the hydride signal is attenuated, there appears to be a new absorptive resonance at ∼(−)22.5 ppm, and the other weak resonances appear to have bifurcated and shifted several parts per million downfield. With ∼43% D2O, only a weak dispersive resonance at ca. −22.3 ppm remains, and with higher aqueous fractions, almost no hydride signal can be detected (not shown).

Figure 3

SABRE studies with the PEGylated Ir catalyst (7/14) in various deuterated ethanol/water mixtures. (a) Thermally polarized reference 1H NMR scan from a D2O/d6-ethanol (∼13%/87% v/v) solution following activation; the spectrum is vertically scaled 5-fold compared to panel b, which shows the successful observation of SABRE enhancement of substrate (py) 1H resonances after 1 min of bubbling with 5.1 atm of pH2 at ∼11 mT, then transfer to 9.39 T for high-field acquisition; enhancements up to ∼40-fold were observed with 3.5 mM catalyst and the given conditions. Changing the D2O/d6-ethanol fraction to ∼43%/57% v/v (c and d) and 63%/37% v/v (inset) significantly impacted the magnitude of the SABRE enhancement; the SABRE spectrum in panel d showed less than an ∼10-fold enhancement compared to the corresponding thermal spectrum (c), and no observed SABRE enhancement was observed in the 63%/37% solution.

The observations of reduced (or no) SABRE enhancements in large aqueous fractions are qualitatively similar to those very recently reported by Fekete et al.,[45] who investigated the use of two different synthetic approaches for generating water-soluble iridium-based SABRE catalysts (respectively featuring sulfonated phosphine groups and IMes NHC variants difunctionalized with triazole groups). For those catalysts, significant 1H NMR enhancements could be observed in organic solvents, but little or no SABRE activity was observed when the aqueous fraction was too great. In that work, the absence of SABRE activity was attributed to the much lower solubility of H2 in water compared to the organic solvents. The observations reported here could be largely explained by the reduced pH2 concentration; however, other effects may be contributing given the complete lack of SABRE activity with high water fractions, as well as the changes in the hydride spectra. As an aside, the solvent environment during activation (i.e., organic vs aqueous) did not affect the results. Thus, the reduced pH2 concentration, possibly combined with structural changes of the catalyst that interfere with the formation of effective hydride species, binding of the substrate, and/or subsequent transfer of spin order from pH2 to substrate spins, likely leads to the loss of SABRE activity with high aqueous fractions—issues that will be the subject of future study.

Exploring SABRE with the CODDA Catalyst

As mentioned above, the standard SABRE catalyst (16) is effectively insoluble in water for the present purposes; however, changes accompanying catalyst activation provide a water-soluble structure (e.g., 15).[57] Thus, in light of the challenges presented by the PEGylated catalyst, an alternative design approach was devised to provide a catalyst structure with improved water solubility (e.g., [IrCl(CODDA)IMes], 13, Figure ) that, once activated, should yield the same SABRE-active structure as 15—with the goal of enabling aqueous SABRE in a single step without need for any organic cosolvent.

Although not as water-soluble as 7 (at least ∼10 mg/mL), according to atomic absorption spectroscopy (AAS) the solubility of the CODDA catalyst (13) in water is ∼0.2 mg/mL; thus, a saturated solution of 13 (with ∼0.3 mM dissolved concentration) was prepared in deuterated water with excess py substrate (∼10 mM). Bubbling with pH2 allowed the activation of the catalyst in an aqueous environment to be monitored in situ via hyperpolarization-enhanced 1H NMR (Figure ). More specifically, spectra from the hydride region acquired during activation of 13 are shown in Figure a, and these results are compared with selected spectra obtained from the standard catalyst (16) in deuterated water (Figure b) and methanol solvents (Figure c), respectively. At first (30 s), the signals from the hydride region for 13 are dramatically different from what is observed during activation of 16. Reflecting the different intermediate structures present, alternating absorptive/emissive (or dispersive) signals downfield of the activated catalyst’s characteristic shift (−22.8 ppm) are virtually absent, and instead the early spectra are dominated by a number of purely absorptive peaks that are mostly further upfield (i.e., with a more negative chemical shift), including a strong peak at −26.2 ppm from a key intermediate structure. Nevertheless, following 180 s of pH2 bubbling, the expected singlet peak at ca. −22.8 ppm is observed, in excellent agreement with the hydride shift of the activated structure 15 obtained from the standard catalyst in methanol (Figure c). However, corresponding efforts to activate 16 directly in D2O were unsuccessful, yielding a cloudy suspension and no discernible enhanced NMR signals from the hydride region (Figure b). In any case, the above results are consistent with successful activation of the novel catalyst 13 in water in just a few minutes to achieve the desired activated structure 15.

Figure 4

Comparison of the hydride regions of 1H NMR spectra acquired during activation of the water-soluble CODDA/Ir SABRE catalyst 13 (a) and the conventional Ir/IMes SABRE catalyst 16 (b and c). (a) Upfield 1H NMR region showing changes to the hydride resonances during activation of 13 in D2O observed at 30, 60, and 180 s after pH2 bubbling began (1 atm). A corresponding attempt to observe activation of the (poorly water-soluble) conventional SABRE catalyst in D2O (16) exhibited no hydride signal (b). Selected spectra obtained separately during activation of 16 in deuterated methanol are shown in panel c, respectively, 60 and 420 s following the onset of pH2 bubbling. As expected, activation of both 13 and 16 trend toward the same final hydride signal (i.e., a strong singlet at ∼22.8 ppm). Spectra in panels a–c possess different vertical scales.

Following successful activation of the CODDA catalyst in deuterated water, the potential of this catalyst for performing SABRE enhancement of 1H NMR in aqueous environments was evaluated using the standard test substrate pyridine (Figure ). With only 1 atm of pH2 bubbling (∼90% pH2 fraction) and catalyst and substrate concentrations of ∼0.3 and ∼10 mM, respectively, an initial enhancement of ca. ε = −25 was achieved for the ortho 1H py position after 30 s of bubbling at ∼10 mT fringe field and subsequent transfer to 9.4 T (Figure b), compared to the signal from a corresponding thermal spectrum (Figure a). The inset of Figure b shows the corresponding hydride regions obtained from the CODDA catalyst during the SABRE experiments, indicating that the CODDA catalyst is essentially activated by the time the SABRE spectra were recorded (total pH2 bubbling time of 210 and 240 s, respectively). Repeating the experiment permitted enhancements as large as ca. −32, −25, and −16 for ortho, para, and meta 1H Py positions to be observed, Figure b; Table . However, the sample from Figure b containing an aqueous suspension of the traditional SABRE catalyst (16) yields no SABRE enhancement, Figure c.

Figure 5

SABRE studies with the water-soluble CODDA/Ir SABRE catalyst (13/15) in 100% D2O. (a) Thermally polarized reference 1H NMR scan from the solution following activation with pH2 in the presence of excess substrate (py); the spectrum is vertically scaled 5-fold compared to panel b, which shows two spectra exhibiting successful observation of SABRE enhancement after bubbling with 1 atm of pH2 at ∼10 mT [total bubbling times of 210 s (30 s immediately prior to acquisition, green dashed curve) and 240 s (30 s immediately prior to acquisition, red solid curve) for the spectra, respectively], then transfer to 9.39 T for high-field acquisition (note that the green curve is shown horizontally offset by a fraction of a ppm to show the enhancement compared to the red curve). Corresponding peak enhancements were ∼25-fold and ∼32-fold for py in water with only 1 atm of pH2 bubbling in the two spectra, using a catalyst concentration of 0.3 mM. The inset shows the corresponding hydride region. A separate experiment where SABRE was attempted using the standard Ir/IMes catalyst in deuterated water exhibited no SABRE enhancement (c). (Vertical scale for panel c is different from that of panels a and b.)

The experiments described above were performed in deuterated water to facilitate spectral interpretation and quantification; however, this practice poses no impediment to broader application of the approach (including for ultimate in vivo experiments) because SABRE hyperpolarization generally works as well (or better) in protonated solution environments, particularly for heteronuclei.[49,64,65] We also note that these results are similar to what has been achieved using the conventional catalyst following dissolution and activation in organic solvents, drying, and reconstitution in D2O (ε ∼ 30), using a weaker substrate (nicotinamide) but higher pH2 pressure (∼5 atm) and greater (∼1:10) catalyst/substrate ratio.[57] In any case, these results indicate the successful preparation, activation, and demonstration of a catalyst capable of easily performing SABRE enhancement in aqueous environments in a single step. This approach obviates the need for either the extra steps associated with reconstitution or the exposure of sensitive biological samples to organic solvents, and thus may also help facilitate biomedical (and ultimately in vivo) applications.

Conclusion

In summary, two novel approaches were investigated for creating water-soluble catalysts to increase the nuclear spin polarization of substrates via SABRE. PEGylation of an asymmetric aromatic carbene ligand provided a highly water-soluble structure that yielded ∼40–60-fold 1H NMR enhancements in alcohol-based solvents and in lean water/alcohol mixtures, but lost SABRE activity in more highly aqueous solvent mixtures. In the second strategy, diol functionalization of the COD ring provided a catalyst structure with lower water solubility, but sufficient to dissolve and activate in water to enable aqueous SABRE in a single step—without need for either an organic cosolvent or solvent removal followed by aqueous reconstitution—here demonstrated for the first time. The >30-fold 1H enhancement under our conditions (with only 1 atm pH2—a mere technical limitation of the bubbler apparatus used for those experiments) is in reasonable agreement with our recent observation of nearly 2000-fold enhancements of 1H signals for the same substrate using the standard SABRE catalyst in deuterated methanol with elevated pH2 pressures,[41] given the expected ∼75-fold difference in pH2 concentration; correspondingly, much larger enhancements should be expected upon implementing experimental approaches to greatly increase the pH2 concentration, including higher-pressure reaction vessels. Moreover, the results presented here likely point the way to achieving higher aqueous catalyst concentrations, which should be possible by employing some combination of the above synthetic approaches (e.g., by functionalizing the COD with moieties that endow greater aqueous solubility). Such improvements, combined with other approaches, should help enable biological and spectroscopic applications that will be pursued in due course.