Toward Optimizing and Understanding Reversible Hyperpolarization of Lactate Esters Relayed from para-Hydrogen.

The SABRE-Relay hyperpolarization method is used to enhance the 1H and 13C NMR signals of lactate esters, which find use in a wide range of medical, pharmaceutical, and food science applications. This is achieved by the indirect relay of magnetization from para-hydrogen, a spin isomer of dihydrogen, to OH-containing lactate esters via a SABRE-hyperpolarized NH intermediary. This delivers 1H and 13C NMR signal enhancements as high as 245- and 985-fold, respectively, which makes the lactate esters far more detectable using NMR. DFT-calculated J-couplings and spin dynamics simulations indicate that, while polarization can be transferred from the lactate OH to other 1H nuclei via the J-coupling network, incoherent mechanisms are needed to polarize the 13C nuclei at the 6.5 mT transfer field used. The resulting sensitivity boost is predicted to be of great benefit for the NMR detection and quantification of low concentrations (<mM) of lactate esters and could provide a useful precursor for the production of hyperpolarized lactate, a key metabolite.

Lactate and lactate esters have many uses in medicine, biochemistry, and food science.[1−3] Detection of these molecules is vital for many applications from food quality control to early diagnosis and treatment of diseases such as cancer.[4−8] Nuclear magnetic resonance (NMR) is often the method-of-choice for molecular analysis as it allows direct detection of a molecule without sample destruction. However, NMR suffers from a high detection limit, which is usually compensated by signal averaging or using highly concentrated molecules (millimolar to molar) to generate sufficient signals. Many approaches have been developed to improve the sensitivity of NMR: perhaps the most successful is the advent of hyperpolarization.[9] Such techniques create molecules where the normal Boltzmann spin state population distribution is tipped to create a transient spin state that can be detected more easily.

In this work, we focus on hyperpolarization involving para-hydrogen (pH2), a spin isomer of dihydrogen, as it is a versatile platform for the hyperpolarization of many biologically relevant disease markers.[5,10−13] Specifically, the signal amplification by reversible exchange (SABRE) method is employed here to enhance the NMR signals of lactate esters. SABRE utilizes the oxidative addition of pH2 to form a hyperpolarized iridium dihydride complex (Figure a). Magnetization can then be transferred to other ligands within the complex in a process that is either spontaneous at low magnetic fields (mT[14] and μT[15]) or radiofrequency-driven at high field (T).[16] Spontaneous transfer at high fields (T) is also possible,[16−19] but ∼6.5 mT fields are typically employed for most efficient polarization transfer to the 1H sites.[14,20] Under such conditions, polarization transfer occurs through the temporary J-coupled network formed within the transient organometallic complex[21] and level anticrossing descriptions are used to rationalize these effects.[22,23] Subsequently, dissociation of ligands from the polarization transfer catalyst gives rise to hyperpolarized molecules free in solution. Therefore, SABRE catalytically creates hyperpolarized target molecules continually and allows them to regenerate easily after they relax back to the non-hyperpolarized form. These considerations provide advantages when compared to other hyperpolarization techniques that are one-shot in nature. Furthermore, SABRE does not require target molecules to contain pH2 acceptors or specialist equipment as its pH2 feedstock is cheap and easy to produce.[24]

Figure 1

Summary of the SABRE and SABRE-Relay hyperpolarization methods and the substrates used in this work. (a) SABRE hyperpolarizes molecules without chemical alteration if the target molecule (here, NH2R) and pH2 are in reversible exchange at a polarization transfer catalyst. (b) The variant, SABRE-Relay exploits relayed proton exchange from a hyperpolarized source to enhance NMR signals in a target that is unable to coordinate to a SABRE catalyst. (c) The hyperpolarization carriers and hyperpolarization targets used in this work.

Currently, SABRE hyperpolarization is typically applied to substrates containing N-donor atoms, such as N-heterocycles,[14,20,25−27] amines,[28] and nitriles[29,30] although S-donor and O-donor targets, such as S-heterocycles[31] and ketoacids,[32,33] have also been hyperpolarized. Recently, relayed proton exchange effects (such as SABRE-Relay)[34] have expanded the substrate scope even further. In SABRE-Relay, a hyperpolarization carrier is used to transfer enhanced magnetization to a target molecule by proton exchange (Figure b). Amines or ammonia are typically used as hyperpolarization carriers as they have the N-donor motif required to form active [Ir(H)2(NHC)(NH2R)3]Cl catalysts necessary for the hyperpolarization of an amine or ammonia.[28,34,35] Exchange between carrier NH and OH protons in molecules, such as alcohols,[36] sugars,[37] silanols,[38] and many others[34,39] allows them to become hyperpolarized even though they do not have the typical iridium-ligating motifs required for direct SABRE. Most examples of molecules hyperpolarized using SABRE-Relay are simple OH-containing structures, with sugars representing perhaps the most complex targets hyperpolarized to date. So far, SABRE-Relay has achieved 2.6% 1H,[36] 2.3% 29Si,[38] 1.1% 13C,[36,37] 0.2% 19F,[36] and 0.04% 31P[36] polarization levels. In this work, we aim to expand SABRE-Relay to hyperpolarize molecules of greater biological relevance such as the OH-containing lactate esters methyl lactate (1), ethyl lactate (2), and butyl lactate (3) (Figure c). Key factors that determine the efficiency of SABRE-Relay and how polarization spreads from the exchanging OH group to 1H and 13C sites within the target substrates are explored. We predict that this sensitivity boost will be of great use for the rapid NMR detection of lactate esters and could have implications for mixture analysis and disease diagnosis in the future.

This work begins with the formation of [Ir(H)2(IMes)(NH3)3]Cl and [Ir(H)2(IMes)(NH2Bn-d7)3]Cl catalysts (where IMes is 1,3-bis(2,4,6-trimethyl-phenyl)imidazole-2-ylidene) by reaction of [IrCl(COD)(IMes)] (5 mM) (where COD is cis,cis-1,5-cyclooctadiene) with NH3 (8–12 equiv) or BnNH2-d7 (5 equiv), respectively, alongside H2 (3 bar) in dichloromethane-d2 (0.6 mL) at room temperature overnight. NH3 and BnNH2-d7 were selected as hyperpolarization carriers as they have been reported to yield the highest SABRE-Relay hyperpolarization levels in simple alcohols based on studies on a range of 24 different NH2R carriers.[36] The formation of these polarization transfer catalysts is confirmed by the observation of characteristic 1H NMR resonances.[28,34] At this point, 1, 2, or 3 (5 equiv) were added to the resulting solutions, under an N2 atmosphere, before pH2 (3 bar) was introduced. The NMR tube was then shaken manually for 10 s in the stray field of a 9.4 T spectrometer where the field was ∼6.5 mT. The metal: carrier and carrier: substrate ratios were fixed at 1:5–10 and 1:1, respectively, as these conditions have been reported to give the optimum relayed hyperpolarization levels in related substrates.[36,37] After shaking, the samples are rapidly inserted into the 9.4 T spectrometer for the collection of a single-scan hyperpolarized 1H or 13C NMR spectrum.

When such SABRE-Relay experiments are performed, enhanced 1H NMR signals for 1, 2, or 3 can be observed (Figure a and b). These enhanced signals are broadened relative to their thermally polarized counterparts and this is likely related to sample movement and nonoptimal shimming using this manual sample shaking and insertion method. The use of automated systems could reflect a route to reduce line broadness.[40,41] The highest 1H NMR signal enhancements were seen for the CH group or lactate CH3 group, with polarization transfer across the ester group into the methyl, ethyl, or butyl groups in 1, 2, or 3, respectively, being less significant (Table ). Prior relaxation of the associated nonequilibrium population differences will play an important role in determining these measured NMR signal enhancements and therefore the relaxation times of the 1H resonances of 1–3 in these mixtures were determined (Table ). It might be expected that the aliphatic sites with the highest NMR signal enhancements are associated with long T1 values. This is true for 1 in which the CH site with the longest T1 time (29.3 s) has the highest 1H NMR signal enhancement (55-fold). However, these trends are not always consistent as the lactate CH3 group within 1 contains the shortest T1 within the molecule (5.9 s) and yet exhibits a modest 30-fold 1H NMR signal enhancement. In other words, the low 1H NMR signal enhancements of sites in the ester chains of 1–3 are not necessarily just associated with rapid proton relaxation (Table ). When these experiments were repeated using the BnNH2-d7 carrier, the trends in 1H NMR signal enhancement proved similar, with the polarization gain remaining predominantly localized on the lactate CH and CH3 groups. For the 1H NMR signal enhancements of these substrates, BnNH2-d7 proves to be more efficient than NH3 as it gives rise to higher enhancements for every site of 1 and 2 when compared to those with NH3, whereas the 1H NMR signal enhancements of 3 are comparable for both carriers. Previous investigations have suggested that when NH3 and BnNH2-d7 are compared as SABRE targets, greater NH polarization results for BnNH2-d7.[28,36] We, therefore, link the amount of substrate OH polarization to the finite polarization of the carrier NH available for relay into the lactate esters after exchange. It generally appears that the average 1H NMR signal enhancements per site decrease as the length of the ester chain increases in accordance with the fact that the finite OH polarization is shared among more protons. Greater distances between the exchanging OH group and remote aliphatic sites in the ester side arm are also likely to play a role in reducing the efficiency of polarization transfer to distant sites.[12,42,43]

Figure 2

Hyperpolarized NMR spectra. Selected regions of single-scan (a, b) 1H, (c) 13C, and (d) 1H decoupled 13C INEPT (short-range) NMR spectra recorded at 9.4 T and 298 K after samples containing (a, c) NH3 (8–12 equiv) or (b, d) BnNH2-d7 (5 equiv) with [IrCl(COD)(IMes)] (5 mM) and the indicated lactate ester in 0.6 mL of dichloromethane-d2 are shaken with 3-bar pH2 for 10 s at 6.5 mT. In panels a and b, the corresponding single-scan thermally polarized spectrum for 1 is shown in black for comparison. More detailed NMR spectra are provided in the Supporting Information, section S2. Thermally polarized 13C and INEPT NMR spectra are provided in the Supporting Information, section S3, for comparison.

Table 1

Summary of SABRE-Relay NMR Signal Enhancementsa

					13C NMR signal enhancement (fold)
		1H NMR signal enhancement (fold)		1H T1 (s)	13C NMR		1H and 13C INEPT short-range		1H and 13C INEPT long-range
substrate	site	carrier NH3	carrier BnNH2-d7	carrier NH3	carrier NH3	carrier BnNH2-d7	carrier NH3	carrier BnNH2-d7	carrier NH3	carrier BnNH2-d7
1	a	30 ± 5	110 ± 10	5.9	625 ± 5	310 ± 40	175 ± 10	55 ± 5	30 ± 5	15 ± 5
	b	55 ± 5	115 ± 10	29.3	595 ± 30	335 ± 25	400 ± 55	150 ± 5	390 ± 50	70 ± 5
	c	N/A	N/A	N/A	85 ± 10	35 ± 5	0	305 ± 20	70 ± 15	30 ± 5
	d	15 ± 5	40 ± 5	10.9	0	0	0	0	0	0
	OH	80 ± 5	310 ± 10	5.8	N/A	N/A	N/A	N/A	N/A	N/A
	average	35 ± 5	85 ± 5	N/A	375 ± 10	195 ± 20	320 ± 40	115 ± 5	145 ± 20	35 ± 5
	NH (free)	70 ± 5	N/A	6.2	N/A	N/A	N/A	N/A	N/A	N/A
2	a	15 ± 5	135 ± 5	5.0	640 ± 5	270 ± 5	55 ± 5	55 ± 5	15 ± 5	20 ± 5
	b	15 ± 5b	90 ± 5b	12.0b	420 ± 45	345 ± 20	115 ± 15	185 ± 10	130 ± 40	120 ± 10
	c	N/A	N/A	N/A	85 ± 10	50 ± 5	0	335 ± 50	100 ± 30	120 ± 5
	d	15 ± 5b	90 ± 5b	12.0b	0	0	0	0	0	0
	e	2 ± 1	20 ± 5	8.1	0	0	0	0	0	0
	OH	40 ± 10	165 ± 5	4.8	N/A	N/A	N/A	N/A	N/A	N/A
	average	15 ± 5	80 ± 5	N/A	390 ± 20	140 ± 5	125 ± 20	140 ± 5	85 ± 20	70 ± 5
	NH (free)	30 ± 5	N/A	4.4	N/A	N/A	N/A	N/A	N/A	N/A
3	a	20 ± 5b	20 ± 5b	4.0£	120 ± 25	95 ± 5	30 ± 5	60 ± 5	0	20 ± 5
	b	40 ± 5+	70 ± 5+	12.8+	165 ± 35	255 ± 15	110 ± 10	5 ± 1	45 ± 15	80 ± 15
	c	N/A	N/A	N/A	35 ± 5	40 ± 5	0	3 ± 1	75 ± 5	35 ± 5
	d	10 ± 5+	10 ± 5+	5.6+	0	0	0	0	0	0
	e	25 ± 5	25 ± 5	5.1	0	0	0	0	0	0
	f	20 ± 5b	20 ± 5b	5.4£	0	0	0	0	0	0
	g	7 ± 1	5 ± 1	5.7	0	0	0	0	0	0
	OH	50 ± 5	N/A	4.7	N/A	N/A	N/A	N/A	N/A	N/A
	average	20 ± 5	20 ± 5	N/A	110 ± 20	130 ± 5	95 ± 10	15 ± 5	40 ± 5	15 ± 5
	NH (free)	75 ± 10	N/A	5.8	N/A	N/A	N/A	N/A	N/A	N/A

1H and 13C NMR signal enhancements (per site) for 1–3 when each (5 equiv) is shaken with 3-bar pH2 for 10 s at 6.5 mT in an activated solution of [IrCl(COD)(IMes)] (5 mM) with NH3 (8–12 equiv) or BnNH2-d7 (5 equiv) in dichloromethane-d2 (0.6 mL). The sites are labelled according to Figure c. 1H T1 values were measured at 9.4 T using inversion recovery. Note that the average polarization refers to the total hyperpolarized integral intensity of the substrate divided by its thermally polarized counterpart and reflects an average substrate polarization per 1H or 13C site. NH (free) refers to the polarization (per site) for the carrier NH.

Values are averaged across the two sites due to signal overlap; +,£Signals overlap.

When similar single-scan hyperpolarized 13C NMR measurements are recorded on the samples of 1–3 with NH3 or BnNH2-d7, carriers, enhanced 13C NMR signals can also be detected despite the fact that the lactate esters are in a non-13C-labeled form (Figure c and d, Table ). Significant 13C NMR signal enhancements are seen for the lactate CH and CH3 (>95-fold) groups with smaller values achieved for the ester carbonyl (<85-fold). The site with the highest 13C NMR signal gain is now dependent on the substrate and carrier. For 1, the lactate CH3 site was enhanced to greater extent than the CH site when NH3 was used as a carrier (625- versus 595-fold, respectively), although when BnNH2-d7 was employed the CH polarization was greater (310- versus 335-fold, respectively). However, for 3, the CH site received the greatest share of polarization (165- versus 255-fold with NH3 and BnNH2-d7 carriers, respectively, compared to CH3 gains of 195- and 150-fold, respectively). For 2, NH3 again delivered the highest 13C polarization on the lactate CH3 site when compared to the CH site (640- versus 420-fold) but BnNH2-d7 enhances the CH signal more than the CH3 (345- versus 270-fold). This behavior suggests that the OH/NH exchange rate, which will differ for the two carriers, must influence the degree of polarization transfer from OH into the 13C nuclei. Visible OH polarization is clearly higher using BnNH2-d7 when compared to NH3, which is reflected in generally better 1H performance but worse 13C gains. Exchange of the OH group proton will destroy the J-coupling network that is associated with propagation of polarization. Therefore, the residence time of the enhanced OH proton is concluded to be extremely important.[44] This fact highlights the tension between initial carrier polarization, efficiency of the relayed NH/OH exchange, and subsequent propagation of OH polarization to other 1H and 13C sites within the substrate. At the 65 G polarization transfer fields in these experiments, 13C polarization was localized exclusively within the lactate unit and did not spread across the carbonyl group to the ester chains of the substrates. It is worth noting that transfer to distant 13C sites may be possible using field cycling regimes at low magnetic fields (i.e., μT): such effects have been demonstrated for related molecules.[5,12,42] For utilization of these enhanced 13C MR signals, it is important that hyperpolarization resides most predominantly on sites with the longest relaxation time as this allows the hyperpolarized state to be detected over much longer time windows which is useful for reaction monitoring[45,46] or biomedical imaging applications.[47] Estimates of hyperpolarized 13C magnetization lifetimes of 1–3 were measured at 9.4 T by leaving the hyperpolarized sample (with carrier and catalyst) in the magnet for a varying time interval before recording a single-scan 13C NMR spectrum. This yielded T1 values of ∼<5, ∼20, and ∼50 s for the CH3, CH, and CO sites, respectively (see Figure S17). These compare well with values reported in the literature for lactate and similar molecules and confirm that the carbonyl carbon, which is more isolated from the other NMR-active spins within the molecule, has the longest T1.[5,6,48] Unfortunately, in our SABRE-Relay experiments on 1–3 discussed so far the majority of the attained 13C polarization resides on the more rapidly relaxing CH3 and CH groups and only modest (<100-fold) signal enhancements can be achieved for the slowly relaxing CO site.

To enhance the degree of magnetization transfer to the carbonyl site, 1H → 13C INEPT (insensitive nuclei enhanced by polarization transfer) pulse sequences were used to transfer the SABRE-Relay derived 1H polarization to 13C sites in the molecule. These sequences contain a variable time delay that is related to the size of the J-coupling connecting the 1H–13C spin pair between which magnetization is transferred. NMR pulse sequences were used with two sets of arbitrary time delays corresponding to polarization transfer between 1H–13C spin pairs with 125 Hz (short-range) and 10 Hz (long-range) coupling (see Supporting Information, section S1.2). The effect of these 1H → 13C INEPT sequences, following the hyperpolarization process of 1–3 using SABRE-Relay, was examined by calculating the 13C NMR signal enhancements by reference to thermally polarized 1H → 13C INEPT sequences recorded using the same delay times and acquisition parameters (Table ). 13C NMR signal enhancements for the lactate CH and CH3 sites within 1–3 were higher using direct 13C NMR detection as compared to the 1H → 13C INEPT detection, regardless of the carrier or INEPT sequence used (Table ). Optimized INEPT sequences can theoretically boost 13C polarization by a factor of 4, which reflects the four times larger gyromagnetic ratio of 1H nuclei. The INEPT 13C NMR signal enhancements recorded here do not achieve this level of 13C sensitivity boost (Table ). We expect factors, such as OH exchange and relaxation, during the mixing time of these INEPT experiments to decrease their efficiency, effects that become more evident for “short-range” INEPT sequences containing longer time delays. In our experiments, it is challenging to attribute a combined 13C NMR signal enhancement factor resulting from combined SABRE-Relay and INEPT methods due to differing acquisition parameters. Nevertheless, we suggest that true hyperpolarized 13C INEPT NMR signal enhancements compared to thermally polarized direct 13C NMR detection are ∼1–2 times larger than the signal enhancements presented in Table , which quotes values relative to thermally polarized 1H → 13C INEPT control measurements. There is, however, a significant uplift in the CO polarization level of the longer-lived carbonyl sites. For instance, 305- and 335-fold 13C NMR signal enhancements could be achieved for 1 and 2 respectively using the BnNH2-d7 carrier and short-range 1H → 13C INEPT sequences, which is higher than the 35-fold and 50-fold achieved for the same samples with standard 13C NMR detection. Interestingly, this increase in carbonyl site 13C polarization using the short-range INEPT sequence suggests that polarization transfer from OH to CO sites is now efficient even though the J coupling is mismatched. There is, however, a dramatic difference between differing carriers, which suggests that the OH/NH exchange has a major impact. For example, in samples involving NH3 as a carrier, 13C NMR signal enhancements for the carbonyl site of 2 and 3 detected using INEPT sequences optimized for long-range 1H–13C couplings (10 Hz) were higher when compared to direct 13C NMR detection. However, for samples containing BnNH2-d7, use of short-range (125 Hz) 1H → 13C INEPT sequences yielded much better 13C NMR carbonyl signal enhancement in 1–3 when compared to direct 13C NMR detection. These results highlight that there is no one set of standard conditions that deliver optimal SABRE-Relay NMR signal enhancements in all three substrates and suggest that further development and the use of novel sequences[49−52] with optimized delay times may constitute a valuable route to increase these 13C signal gains further.

The initial SABRE effect, in which spin order is transferred from pH2-derived hydride ligands to the NH site of a bound carrier molecule, is driven by formation of a transient J-coupled network.[21,22,53] Upon NH/OH exchange, polarization propagation to other 1H and 13C sites within the lactate ester could occur through J-coupling in a similar fashion. To gain a greater understanding of these effects, density functional theory (DFT) calculations were performed to predict the J-couplings within 1-3 that are involved in these polarization transfer processes (Supporting Information, Tables S5–7). Overall, the short-range computed coupling constants are in fairly good agreement with values obtained from experiment (Supporting Information, Tables S5–7). Furthermore, these calculations determine long-range J-coupling values that we could not discern from 1H or 13C NMR spectra and could play an important role in polarization propagation. The results show that the J coupling between the OH proton and the carbonyl carbon is just 5–6 Hz. Small J-couplings will also play a role if polarization transfer from OH is indirectly propagated to the carbonyl carbon according to a OH → Hb → Cc pathway, with the latter step no longer being inhibited by OH exchange. Computationally, the OH → Hb → Cc transfer pathway involves small J couplings of <5 Hz. A route involving a larger 56–57 Hz coupling Cb → Cc could operate, but this will be limited by the natural 13C abundance and a large JCC splitting is not observed for the carbonyl signal in hyperpolarized spectra. Crucially, these J-couplings prove to be similar for 1–3, which suggests that the differences in substrate 1H and 13C polarization levels that are observed are not due to significant differences in the spin coupling topology within these CH frameworks.

Spin dynamics simulations were performed to predict how coherent polarization transfer might occur via the J-coupling network from the OH spin to the other spins within the lactate ester framework. Different spin systems were simulated, taking into account the fact that natural abundance 13C was involved in the experiments. Initially, a 5-spin model consisting of OH, Hb, and three Ha protons was used to study polarization transfer within the proton-only spin system. These simulations allow evolution of an initial 100% OH nuclear spin polarization during a 10-s shaking period at a given polarization transfer field (0 T, 6.5 mT, or 9.4 T) to the other 1H spins. These simulations reflect an ideal experiment free from the effects of OH/NH exchange and T1 relaxation. Simulations for this system show that spin polarization transfer occurs between all the included 1H spins at the lower polarization transfer fields (0 T and 6.5 mT), but essentially no spin polarization transfer occurs at 9.4 T (see Supporting Information, Figures S19–S21). Furthermore, 6-spin systems involving one of the 13Ca, 13Cb, or 13Cc centers were simulated to investigate polarization transfer to 13C, as in most 13C-containing molecules there is only one naturally abundant 13C nucleus (see Figure and Supporting Information, Figures S19–S21). Sites in the ester side chain were not included as 13C NMR experiments do not exhibit enhanced magnetization for these sites at a 6.5 mT transfer field. For these 6-spin systems, simulations show that at polarization transfer fields of 0 and 9.4 T they behave similarly to the 5-spin 1H systems, with spin polarization transfer to all spins at 0 T and not to any noticeable extent to any spin at 9.4 T. At a polarization transfer field of 6.5 mT, the simulations show that the presence of a 13C (i.e., in about 1% of the molecules at natural abundance), reduces the spin polarization transfer between the protons by perturbing the spin–spin coupling network, without any noticeable spin polarization being transferred to the 13C. Under such idealized conditions, the simulations reveal that coherent polarization transfer from OH to Ha and Hb can occur at 6.5 mT, but transfer to the 13C subsystem does not. This suggests that other, incoherent routes, such as cross relaxation, have to dominate the OH → 13C polarization propagation at the 6.5 mT polarization transfer fields necessary to achieve optimal 1H polarization via SABRE-Relay. This finding supports previous theoretical studies.[44]

Figure 3

Simulating polarization propagation. Simulated spin polarization as a function of time for the different sites of a 6-spin system (3Ha, Hb, OH, and Cc) at (a) 0 T, (b) 6.5 mT, and (c) 9.4 T.

The results presented so far highlight that there is no one set of standard conditions that achieve optimal SABRE-Relay NMR signal enhancements for all substrates. Optimization of magnetization on a particular site, or particular nuclei (1H or 13C in this work), is likely to require a different set of conditions, even for optimization of NMR signal enhancements on different sites, or for different nuclei, in the same molecule using the same carrier. To highlight this point, SABRE-Relay hyperpolarization of 2 was tested with different amounts of the carrier NH3. These results show that the 1H NMR signal enhancements for the lactate CH3, CH and OH sites could be dramatically increased from 15-, 15-, and 40-fold using 9 equiv NH3 to 215-, 160-, and 245-fold, when the amount of NH3 was lowered to 6.5 equiv (Figure a). Similar increases of 13C NMR signal enhancements from 640-, 420, and 85-fold for the CH3, CH, and CO sites, respectively, to 865-, 985- and 110-fold, respectively (Figure b), were observed. Interestingly, this same improvement in 13C NMR signal enhancement at lower NH3 concentration was not observed in the data collected by 1H → 13C INEPT transfer (Figure c and d). This suggests that a lower NH3 amount corresponds to more optimal, likely slower, NH/OH exchange and benefits the direct 1H and 13C NMR detection.

Figure 4

Toward optimizing SABRE-Relay. (a) 1H and (b) 13C NMR signal enhancements of 2 achieved using SABRE-Relay as a function of carrier NH3 concentration. 13C NMR signal enhancements achieved using (c) short-range and (d) long-range 1H→ 13C INEPT sequences are also shown. A full table of NMR signal enhancements is given in the Supporting Information, Table S9.

The substrates 1–3 used in this work are liquids at room temperature and pressure. Hyperpolarization of protic molecules that are not soluble in nonprotic SABRE-Relay compatible solvents can be challenging because nonspecific proton exchange is undesirable. For example, attempts to hyperpolarize sodium lactate by SABRE-Relay fail due to the presence of water; 10 μL is required to solubilize it in CD2Cl2 or CDCl3 solvents (see Supporting Information, section S10). Furthermore, doping SABRE-Relay samples of 1–3 with water leads to a drop in the delivered NMR signal enhancements (see Supporting Information, section S9). Similar effects have been reported for other substrates.[36,37] Hyperpolarization of 1-3 could therefore provide a route to hyperpolarize sodium lactate. To this end, SABRE-Relay could be used with rapid hydrolysis of the ester side chain to form lactate in an approach that is similar to PHIP-SAH which has already achieved significant lactate 13C polarization.[5] In preliminary experiments we confirmed that hydrolysis and phase separation of 2 could indeed yield aqueous solutions of lactate (see Supporting Information, section S10). However, these steps took ∼90 s in our nonoptimized setup and, therefore, have not yet yielded enhanced NMR signals for lactate. In the future, optimization of 13C NMR signal gains on the carbonyl site, followed by rapid hydrolysis, will likely open a route to produce hyperpolarized lactate. This approach could be extended to a wider range of molecules that contain the OH-groups necessary for SABRE-Relay but are not soluble in compatible solvents. They could be functionalized as esters with a side arm that alters solubility. This would allow them to be hyperpolarized using SABRE-Relay before the target molecule is released in a rapid hydrolysis step.

In conclusion, hyperpolarization of lactate esters using SABRE-Relay has been presented. 1H and 13C polarization is predominantly localized within the lactate CH and CH3 groups with lower signal enhancements for the carbonyl carbon and little polarization transfer to the ester side arm. Optimization and rationalization of SABRE-Relay efficiency can be extremely challenging with many factors such as carrier NH polarization, OH/NH exchange efficiency, and polarization propagation from OH to other sites with the substrate all being important steps that can be influenced by the carrier, substrate, and their ratios. 1H and 13C NMR enhancements as high as 245-fold (0.8% polarization) and 985-fold (0.8% polarization), respectively, for lactate esters have been achieved using the SABRE-Relay approach. We have also demonstrated that 1H → 13C INEPT sequences can in some cases improve polarization of the longer-lived carbonyl site (up to 335-fold 13C NMR enhancement using hyperpolarized INEPT). These sensitivity improvements have allowed the detection of 25 mM concentrations of lactate esters in just a single-scan 13C NMR spectrum, which are not discerned using thermally polarized NMR. Examination of a wider range of INEPT sequences,[6,49,54,55] further optimization of sequence mixing times, or even magnetic field cycling approaches,[5,42] are likely to increase these NMR signal enhancements. Relayed substrate/carrier exchange influences polarization propagation within the substrate, despite theoretical calculations showing similar spin topologies within the substrates used here. Spin dynamics simulations indicate that polarization transfer from the OH proton to the other 1H sites of the substrate readily takes place via the spin–spin coupling network at the polarization transfer field of 6.5 mT. In contrast, polarization of the 13C spin subsystem is deduced to occur via incoherent mechanisms, at this polarization transfer field. It is likely that both J-coupling and cross-relaxation effects can play a role in polarization transfer to heteronuclei depending on the spin topology of the substrate and the polarization transfer field. In the future, further studies to understand and optimize relayed polarization transfer would be highly beneficial because of the wider range of molecules whose NMR signals could be enhanced by the technique.