PHIP-X is a complex physical-chemical process with mutually influencing parameters. We distinguish four critical steps in the PHIP-X process (A, B, C and D in Fig. 1a):
A.
Hydrogenation of a precursor1 using pH2 to hyperpolarize a transfer agent2.
B.
Polarization of exchanging protons of transfer agents.
C.
Polarization of the target molecule (Target) via proton exchange.
D.
Polarization of the target nucleus via polarization transfer within Target.
In step A, pH2 is added to an unsaturated precursor (e.g., 1 = propargyl alcohol) with the aid of a catalyst (e.g., [Rh] = [Rh(dppb)(COD)]BF4, dppb = 1,4-Bis-(diphenylphosphino)-butane, COD = 1,5-cyclooctadiene) to generate a transfer agent (e.g., 2 = allyl alcohol). In step B, the labile proton of 2 is polarized. This proton is in exchange with a labile proton of Target, such that the polarization subsequently transfers to Target in step C. In step D, the polarization is transferred to the target nucleus. Each of these steps can be optimized in several ways.Changing the conditions for one step, however, may also influence the condition of another step, which complicates the improvement of the methodology. For example, a change of the conditions for the proton exchange (e.g., by changing the solvent system) also affects the hydrogenation. The matter is even more complicated as the sample is usually hydrogenated, transferred, and measured at different fields. Some of the steps will happen at the same time.Previous results26 showed that step A is already quite efficient, since a pH2-enrichment of only 50% induced a polarization of about 13% on the transfer agent (a pH2-enrichment of about 99% should triple that value). For step B and D, in principle, many of the known polarization transfer techniques can be employed, like free evolution at constant or varying magnetic fields, or a dedicate pulse sequence. The matter is exacerbated, however, by the exchanging target nucleus.In this study, we focus on the following steps C and D. We report on 5 experimental results (E1–E5) and 5 simulation results (S1–S5).Polarization transfer within the target moleculeWhen a polarized proton is transiently bound to the Target, it is essential to transfer its polarization to the desired 13C nuclei or, as we will see later, an intermediary.The straightforward approach is to transfer the polarization from the labile proton (1Hl) directly to 13C e.g. by using a pulse sequence at high magnetic fields tailored to the specific coupling (here: 3 Hz, τj = 167 ms, Fig. 1a, SOT2). Thus, we simulated the effect of DEPT and refocussed INEPT (rINEPT) as described further down, and tested DEPT 135° experimentally (note that DEPT 45 and DEPT 90 work as well.Experimentally, we tested six different hydrogenation fields (BPol0 = 15, 30, 45, 60, 75, 90 mT), where 1 (propargyl alcohol) was hydrogenated for 5 s, resulting in 2 (allyl alcohol)). The number of fields BPol0 was increased for the biologically more relevant lactate. Result E1 (BPol0): After hydrogenation, transfer through Earth’s field and DEPT at high field (1 T), the 13C signal of methanol (Target) was found to be significantly enhanced compared to the thermal signal. In case of 3 Hz a maximum signal amplification of ≈38 was reached at BPol0 = 75 mT (Fig. 2b, red).Fig. 2: 13C-methanol hyperpolarized with PHIP-X.a 13C-NMR spectra of methanol hyperpolarized using PHIP-X. DEPT 135o tuned to J(1Hl-13C) = 3 Hz and J(1Hc-13C) = 145 Hz were compared with the thermally polarized sample (black). Applying the 145-Hz-DEPT resulted in higher polarization than using the 3-Hz-DEPT sequence. The thermal spectrum (black) was amplified by a factor of 20 and recorded by averaging 200 scans. b Signal enhancement of 13C-methanol for different fields BPol0 during hydrogenation (for each point the mean was calculated using 3 measurements with standard deviation). In this experiment, only alcoholic hydroxy groups, but no carboxylic groups, were present.Interestingly, we observed that the methyl protons of Target (1Hc) were polarized, too (similar to step b in reverse). Thus, we tested transferring the polarization from these directly bound, non-exchanging methyl-protons 1Hc to the carbon-13, using DEPT-135° adjusted to the 1Hc-13C J-coupling of ≈145 Hz (SOT1, green, (Fig. 2b). Result E2 (SOT1/SOT2): This experiment resulted in 5–30 times higher signal enhancements compared to SOT2, reaching a maximum of about 150-fold (with respect to 1 T) when the sample was hydrogenated at 15 mT or 90 mT.Both the facts that a) the methyl protons were polarized “spontaneously” and b) the difference in enhancements caused by transfer from the exchanging proton or directly bound protons are correlated: In both cases, the polarization “enters” the Target via the exchanging proton, but the “detour” via the methyl-protons leads to higher 13C polarization in the end. It appears likely that the reason for this is a), that the methyl protons “accumulate” the polarization at BPol0 due to less relaxation, b), that the exchange deteriorates the effectivity of DEPT (a proton would need to be associated with target for the entire duration of DEPT for optimal effect), and c, that 1J(1Hc-13C) > 2J(1Hl-13C) accelerates the polarization transfer. 13C-methanol was chosen as a model target system because of its simple structure and known 2JCH-coupling constant of 3 Hz32. In the following section, we will use these findings to polarize lactic acid.Chemically optimized PHIP-X of lactic acidNext, we tested the biologically relevant molecule 13C3-lactic acid (LA).Result E3 (SOT1/SOT2)When DEPT was tuned to (an estimated) coupling between the labile proton and the 2-13C, J(1Hl-13C) = 1–8 Hz, no 13C signal enhancement of LA was observed (Fig. 3a, red). When DEPT was set to the coupling between the firmly bound 1H and 2-13C, J(1Hc,13C) ≈ 139 Hz), however, strong 13C enhancements were found, both on 2-13C and, notably, on 3-13C, too (Fig. 3a, green). This result is supported by the hyperpolarized 1H spectrum of lactate26, which showed hyperpolarized signal of the proton directly attached to the 2-13C. This enhancement is a significant advance compared to the original method26, where no 13C-signal enhancement was detected by using a simple 90° pulse. Note that the resonance in the SOT2-DEPT spectrum (red, Fig. 3a) at 63.5 ppm originated from hyperpolarized 2 with natural abundance of 1-13C.Fig. 3: PHIP-X hyperpolarized 13C3-lactic acid (LA).a Strong 13C hyperpolarization was observed on 2- and 3-13C lactic acid when the polarization was transferred from the methyl-1H to the methyl-13C using SOT1 (green). No 13C signal enhancement was apparent when the transfer was attempted from the labile proton using SOT2. As reference, 1000 scans of the sample after the PHIP-X experiment were acquired in thermal equilibrium and magnified 10-fold for convenience (gray). b Normalized signal enhancements of 2-13C-LA as a function of concentration ratio c1/c(LA), where 1 = propargyl alcohol. The polarization was found to decrease monotonously around a maximum atc1/c(LA) ≈ 3.5. Changing this ratio affects the proton exchange processes between 1 and LA (Fig. 1a–c). Error bars represent the standard deviation respecting the normalization. Experimental details: c(LA) = 39.3 mM,c1 = 17.3–865 mM. After applying 30 bar pH2 for 5 s at BPol0 = 0.05 mT, the solution was shuttled into the 1 T NMR spectrometer where DEPT 135o (139 Hz) was applied 1.5 s after the beginning of the shuttling.Result E4 (exchange)Next, we investigated how the precursor-to-target concentration ratio affects the 2-13C-polarization of lactic acid (Fig. 3b). For all experiments, we kept the concentration of LA constant at c(LA) = 39.3 mM and varied the concentration of the precursor, c1, from 17.3 to 865 mM. At BPol0 = 0.05 mT, we found a monotonous increase of the 13C signal up to c1/c(LA) ≈ 3.5, and a monotonous decrease thereafter.
B
Pol0-dependance of the 13C-polarization of lactic acidResult E5 (B
Pol0)Next, we investigated the BPol0-dependance of the polarization of 2-13C-LA. Using the optimized ratio of c1/c(LA) ≈ 3.5 and the sequence to transfer from the directly bound 1H to 2-13C (DEPT 135 (J = 139 Hz)), we varied Bpol0 from 0.05 to 95 mT. We found that the 2-13C-LA polarization was greatly affected by Bpol0, roughly speaking increasing with BPol0 (Fig. 4). The highest polarization was found at the highest field tested (95 mT, P(13C) = 0.026%).Fig. 4: 13C hyperpolarization of 2-13C lactic acid as function of BPol0.There is a tendency of higher P for higher fields. The highest polarization was found at BPol0 = 95 mT, and the lowest polarization was found at 5 mT. The experiments were carried out using the optimized ratio of c1/c(LA) ≈ 3.5, c(LA) = 40 mM and DEPT 135° set to 139 Hz (3.6 ms evolution period). Note that the error bars (standard deviation) indicate relatively strong fluctuations (each point is a mean of 3 scans). Enhancement is calculated with respect to thermal signal at 1T.Spin dynamics simulationsTo elucidate these matters further, we performed simulations solving the Liouville von-Neumann equation with relaxation and exchange superoperator33,34,35. We considered a system of two labile protons interacting with a target consisting of one carbon and one proton. The system was defined by the Lamor frequencies, J-couplings, relaxation rates (T1), exchange rates (K) and time-dependent magnetic fields (Fig. 5, The J-couplings are J13 = J23 = −3 Hz, J14 = J24 = 5 Hz and J34 = 140 Hz and magnetic shields are c1 = 6 ppm, c2 = 4 ppm, c3 = 120 ppm and c1 = 6 ppm. K1 = K2 = 200 1/s, BPol0 = 90 mT, BPol1 = 50 µT, \(P\left({S}_{Z}^{1}\right)=P\left({S}_{Z}^{2}\right)=50 \%\) at t = 0, T1 = 1 s for both labile protons (spin No. 1 and 2), T1 = 20 s for the 13C nucleus (spin No. 3) and T1 = 4 s for the fixed target proton (spin No. 4), unless otherwise noted). Pulses were assumed as instantaneous rotations.Fig. 5: Scheme of the spin system simulated.The system consisted of labile spins No. 1 and 2, a carbon spin No. 3, and a (firmly bound) proton spin No. 4 (left side). The spin system transitions between state X and Y happened with rate constants K1 = K2. The spins considered in the corresponding chemical system (right) are marked in red. A is a compound that generates labile protons (examples are given on the bottom of the right side). R1, R2, R3 and R4 are any rests.The initial state was chosen such that the labile protons were hyperpolarized (50% at t = 0) while the carbon and proton spin of the target were thermally polarized (for our simulations it not important whether the labile spins were polarized using DNP, SABRE-RELAY or PHIP-X). We simulated a PHIP-X experiment in which we included a magnetic field cycling (MFC) followed by a pulse sequence. The MFC consisted of free evolution at BPol0 for tPol0 = 2400 ms, a linear drop to BPol1 in tcycle = 15 ms, free evolution at BPol1 = 50 uT for tBpol1 = 600 ms, a linear rise to B0 = 1 T in tcycle = 15 ms and finally a free evolution at B0 for 500 ms. The pulse sequence was either DEPT or refocused INEPT (rINEPT). We simulated time steps of 1 ms.Firstly, we simulated the target polarization during the experiment for different exchange rates K1 and K2, while keeping all other parameters fixed (Fig. 6). Here, we simulated up to the time point where the sequence is initiated, such that the results are independent of the sequence.Fig. 6: Evolution of the 1H polarization \((P({S}_{Z}^{4}))\) in the target for different exchange rates K1 = K2 = 6.12–4800 s−1 during PHIP-X.A maximum target polarization was found at K1 = K2 = 200 s−1 with a monotonic decrease for smaller (a) and larger (b) rates. Note that MFC was initiated at 2400 ms and the dashed lines indicate the transitions between the magnetic fields.Result S1It was found that exchange rates of K1 = K2 = 200 1/s generated the strongest target polarization (blue line in Fig. 6a, b), which showed a maximum polarization of the fixed proton (No. 4) after an evolution of 1283 ms. Higher as well as lower exchange rates generated lower target polarizations, whose maxima are shifted slightly towards later times (Fig. 6a, b).The MFC-induced 13C polarization is proportional to the 1H target polarization (SI). Simulation parameters other than exchange rates were used as stated above.Next, we investigated how the T1 relaxation of the labile protons \({T}_{1}^{L}\) affected the polarization of the target (Fig. 7a, b). The relaxation times are chosen to start at \({T}_{1}^{L}=31.25{{\rm{ms}}}\) and double at each step until 4 s are reached.Fig. 7: Evolution of the polarization of the fixed proton of the target \(\left(\right.({{\boldsymbol{P}}}({{{\boldsymbol{S}}}}_{{{\boldsymbol{Z}}}}^{4}))\) and the 13C of the target \(\left(\right.({{\boldsymbol{P}}}({{{\boldsymbol{S}}}}_{{{\boldsymbol{Z}}}}^{3}))\) for different longitudinal relaxation times \({{{\boldsymbol{T}}}}_{1}^{{{\boldsymbol{L}}}}\) of the labile protons.There is a monotonous increase of the 1H (a) and 13C (b) target polarization with increasing \({T}_{1}^{L}\). Note that MFC is applied at 2400 ms and the dashed lines indicate the transitions between the magnetic fields. Here, BPol0 = 90 mT and BPol1 = 50 µT were used.Result S2As expected, the target polarizations were found to increase with \({T}_{1}^{L}\). Still, the fastest relaxation rate of 31.25 ms generated significant 1H target polarization of \(P\left({S}_{Z}^{4}\right){{\rm{\approx }}}0.4 \%\) (Fig. 7a); for longer times, \(P\left({S}_{Z}^{4},{T}_{1}^{L}=0.5{{\rm{s}}}\right){{\rm{\approx }}}4.0 \%\) and \(P\left({S}_{Z}^{4},{T}_{1}^{L}=4{{\rm{s}}}\right){{\rm{\approx }}}11.4 \%\). These findings appear promising for reaching significant polarizations over a wide range of exchange rates. For 13C, no polarization was observed during BPol0, prior to the MFC (Fig. 7b). At BPol1 = 50 µT, oscillating 13C polarization was found. Oscillations vanished at B0 while the latest polarization level of the oscillations was preserved. K1 = K2 = 200 s−1 was used, and all other parameters were the same as for Fig. 6.Result S3A closer look at the polarization at BPol1 showed that the polarization oscillated between 1H and 13C (Fig. 8a, shown for BPol1 = 5 uT). The amplitude of the oscillations (and thus the 13C polarization) was further increased for lower BPol1 (Fig. 8b). In case of BPol1 < 100 nT, the frequency is given by the J coupling of J34 = 140 Hz. The couplings of 5 Hz and −3 Hz contributed to the oscillations as well (Figs. 7b and 8). A high and stable 13C polarization can be achieved if the magnetic field increases very quickly from BPol1 to B0 at a time when the phase of the oscillations is such that most of the polarization is transferred from 1H to 13C. Hence, the obtained 13C polarization yield depends very critically on the length of the time interval in which BPol1 is applied. Note that our current experimental setup did not allowed us to vary BPol1. However, similar effects of excitation of coherences between protons or between protons and 13C were observed experimentally before in other hyperpolarization experiments36,37,38.Fig. 8: 1H and 13C polarization of the fixed proton and carbon in the target molecule during a PHIP-X experiment.a For the first 2400 ms, the system evolved at BPol0 = 90 mT, and 1H polarization was built up (\(P\left({S}_{Z}^{4}\right)\)). When the system was dropped to BPol1 = 5 µT (in 15 ms), the polarization started to oscillate between \(P\left({S}_{Z}^{4}\right)\) and \(P\left({S}_{Z}^{3}\right)\). When the field was increased (in 15 ms) to B0 = 1 T, the oscillations stopped, and the latest polarization of the oscillation was preserved. The largest amplitude and the lowest frequency were observed (b) for BPol1 < 100 nT. Note that the oscillations are modulated also by lower frequencies (5 Hz or 3 Hz) time, so that significant 13C polarization can be obtained by increasing the field to B0 at the right time.Next, we investigated the 1H-13C polarization transfer with DEPT and refocused INEPT (Fig. 9a) tailored to 140 Hz (corresponding to J34). DEPT has three intervals τ = 1/2J, and rINEPT has four with τ = 1/4J, so that DEPT runs 1.5 times longer than rINEPT. Again, we assumed BPol0 = 90 mT, tpol0 = 2400 ms, BPol1 = 50 µT, tPol1 = 600 ms, B0 = 1 T, tB0 = 500 ms, tcycle = 15 ms, K1 = K2 = 200 1/s, for \({T}_{1}^{L}=1\), \({T}_{2}^{L}=1{{\rm{s}}}\), \({T}_{3}^{L}=20{{\rm{s}}}\), \({T}_{4}^{L}=4{{\rm{s}}}\) and \(P\left({S}_{Z}^{1}\right)=P\left({S}_{Z}^{2}\right)=50 \%\) at t = 0.Fig. 9: 13C polarization during rINEPT and DEPT 90°, and 1Hc polarization during the PHIP-X experiment.Both sequences transfer the proton polarization perfectly (>99.9%), although DEPT takes about 3 ms longer (a). Timings were set for J(1H, 13C) = 140 Hz. Dashed lines indicate pulses, delays, field changes, and the polarization at the onset of the SOT; numbers correspond to individual information; rINEPT: 1: 90°x 1H, 2: 180°x 1H and 13C, 3: 90°y 1H and 90°x 13C, 4: 180°x 1H and 13C, 5: start of FID. DEPT: 1’: 90°x 1H, 2’: 180°x 1H and 90°x 13C, 3’: 90°y 1H and 180°x 13C, 4’: start of FID. The distances between the vertical lines (a) correspond to the evolution periods of τ = 1/(2J) and τ = 1/(4J) respectively. The simulations were done in the lab-frame, which is the reason for the oscillations (a). The 1H target-polarization (b) had a maximum of about 6% and was 3.3% at the time where the SOT was applied.Result S4For J = 140 Hz (Fig. 9a), both sequences transferred all available proton polarization to the carbon (P(1H) = 3.3% is completely transferred to P(13C) = 3.3%). DEPT was longer by ≈3.6 ms.Result S5For J = 3 Hz, the picture is more complex (Fig. 10). In the fully coupled molecule (J13 = J23 = −3 Hz, J14 = J24 = 5 Hz and J34 = 140 Hz), DEPT produced a 13C polarization of ca. 0.7%, and rINEPT of ca. 0.25%. Interestingly, the polarization was strongest right after the last pulse (3’), and not after the latest evolution period (4’), as expected. If all coupling constants in the spin system were set to 0, except the ones between the carbon and the labile protons (J12 = J13 = −3 Hz, J14 = J24 = J34 = 0), the results were the opposite. Now, rINEPT produced about 0.81% and DEPT about 0.49% 13C polarization. In case of 3 Hz, DEPT was about 180 ms longer. The strongest polarization was right after the last evolution period (4’ and 5), which corresponded to the beginning of the FID. These findings show that a) the other couplings affect the SOT if the evolution times are long (for J = 3 Hz), and b), that (some of) the polarization was transferred directly from the exchanging proton, despite the ongoing exchange.Fig. 10: rINEPT and DEPT in the case of J = 3 Hz.13C polarization during rINEPT (blue) and DEPT (orange) with J = 3 Hz for the fully coupled spin system (a, as in Fig. 5) and where all couplings, except the ones between the labile protons and carbon, were 0 (b). For the fully coupled system, the dynamics are affected by the other couplings in the molecule. DEPT achieved almost three times higher polarization as rINEPT. This suggests that other, more efficient sequences maybe found that take all couplings into account. For the simplified system (e.g. a tertiary alcohol), the expected behavior is observed, suggesting that the polarization was transferred from the labile proton despite of the exchange.