Recent advances in the application of parahydrogen in catalysis and biochemistry.

Nuclear Magnetic Resonance (NMR) spectroscopy and Magnetic Resonance Imaging (MRI) are analytical and diagnostic tools that are essential for a very broad field of applications, ranging from chemical analytics, to non-destructive testing of materials and the investigation of molecular dynamics, to in vivo medical diagnostics and drug research. One of the major challenges in their application to many problems is the inherent low sensitivity of magnetic resonance, which results from the small energy-differences of the nuclear spin-states. At thermal equilibrium at room temperature the normalized population difference of the spin-states, called the Boltzmann polarization, is only on the order of 10-5. Parahydrogen induced polarization (PHIP) is an efficient and cost-effective hyperpolarization method, which has widespread applications in Chemistry, Physics, Biochemistry, Biophysics, and Medical Imaging. PHIP creates its signal-enhancements by means of a reversible (SABRE) or irreversible (classic PHIP) chemical reaction between the parahydrogen, a catalyst, and a substrate. Here, we first give a short overview about parahydrogen-based hyperpolarization techniques and then review the current literature on method developments and applications of various flavors of the PHIP experiment. This journal is © The Royal Society of Chemistry.

Introduction

For many years there have been strong efforts to create so-called hyperpolarized spin systems with polarization far larger than the Boltzmann polarization. Dynamic nuclear polarization[1-4] schemes solve this problem by employing stable radicals or optically created transient triplet states or noble-gases (see e.g., ref. 5–11). Parahydrogen based hyperpolarization schemes employ a chemical reaction of a substrate with parahydrogen[12] for the creation of the hyperpolarization, the so-called parahydrogen induced polarization (PHIP) experiments, which were initially predicted by Bowers and Weitekamp.[13] The first experimental realizations of the experiment were presented in the seminal papers of Weitekamp et al.[14,15] and, independently, Eisenberg et al.[16] About two decades after the initial discovery of PHIP, Duckett and coworkers[17] proposed a reversible variant of the PHIP reaction, called SABRE (Signal Amplification By Reversible Exchange). In the PHIP techniques, the strong spin order of parahydrogen (p-H2, the H2 molecule in its nuclear singlet spin state) is used to generate NMR signal enhancements. It is relatively easy and cheap to produce p-H2 (enrichment of H2 with >90% of p-H2 has become feasible and even standard), and the starting “spin order” of the protons is very high: nearly all H2 molecules are in the same spin state. In order to render the spin order of the parahydrogen visible in an NMR spectrum, a symmetry-breaking step is needed, which is provided by a chemical reaction. In such a step either p-H2 is chemically attached to a molecule with an unsaturated C–C bond or the spin order of p-H2 is transferred to a substrate in a transient organometallic complex. The former scheme corresponds to the conventional PHIP with catalytic hydrogenation, whereas the second scheme is the SABRE experiment.[17] Both variants of PHIP are widely used to enhance NMR signals in general to investigate chemical processes or to enhance the sensitivity in analytical applications.[18,19]

While there is some application overlap between DNP and PHIP in the fields of solution NMR and medical applications, the two techniques are highly complementary. The main advantages of the DNP schemes are that they are versatile, independent of a catalyst, cause no chemical modification of the substrate, and are applicable both in solution and solid-state NMR spectroscopy as well as MRI (see Handbook of high-field DNP[20] and references therein). Their disadvantages, however, are that they depend on a sophisticated and expensive apparatus, need a doping with stable radicals or chromophores and often have to be performed at cryogenic temperatures to achieve efficient levels of hyperpolarization.

The main advantages of PHIP schemes compared to DNP are their independency of any sophisticated experimental hardware and the opportunity to be implemented on any modern NMR spectrometer for few thousands of Euros. An overview of current PHIP instrumentations is given in a recent review by Schmidt et al.[21] Other advantages are the use of inexpensive chemical compounds and the fast formation of the polarization within seconds. PHIP schemes can be employed at ambient temperature and are thus easily amenable for kinetic studies. In the simplest way, PHIP/SABRE experiments are done by blowing the p-H2 gas through the sample containing the substrate and the catalyst to initiate chemical reactions. The main weaknesses of parahydrogen-based hyperpolarization are its dependences on the presence of a hydrogenation catalyst and, in the case of irreversible hydrogenation, the need of a suitable unsaturated precursor which is chemically modified by the reaction.

PHIP experiments were successfully employed for the detailed investigation of mechanistic problems in organic and inorganic synthesis.[22-29] They have an immense application potential in the field of in situ gas phase NMR. This technique provides a direct monitoring of heterogeneous catalytic reactions such as e.g., molecular dynamics, hydrogenation of unsaturated bonds in olefins, H/D-isotope exchange reactions, Fischer–Tropsch and Haber–Bosch type reactions.[30-35] They are sensitive monitors for chemical processes in microreactors.[36-38] They can be employed for sensitivity enhancement in time-domain NMR analyzers.[39,40] Owing to the strong dependence of the PHIP enhancement and signal pattern on the value of the polarization field, PHIP measurements are powerful monitors for local fields in field-cycling experiments.[41-45] There are several excellent reviews which explain in detail the experimental techniques and their theoretical background.[46-50] Currently, there is only a limited number of suitable substrates or catalysts amenable for SABRE, most of them contain nitrogen heterocycles like pyridyl, pyridazyl, pyrimidyl or similar groups.[51] These groups are of very high pharmacological interest, due to their ubiquitous presence in biochemically relevant molecules. Typical application examples of SABRE are the hyperpolarization of organic polymers,[52] amino acids and peptides,[45] and keto–enol tautomers.[53] Hövener et al.[54] recently reported the so-called PHIP-X (Parahydrogen-Induced Polarization Relayed via Proton Exchange) experiment. In this experiment the hyperpolarization is transferred from the hydrogenated substrate to a target molecule via chemical proton exchange.

An important issue for all medical in vivo applications of PHIP on humans is the efficient separation of the necessary catalysts from the hyperpolarized substrates. While this separation can be easily achieved for gaseous substrates, such as hyperpolarized propene,[55,56] for fluid substrates there is until now no really satisfying answer. Currently there are three possible solutions to this problem, namely (i) the immobilization of an homogeneous catalyst via tethering to e.g. a silica surface,[57-62] (ii) employing supported MNPs and (iii) physical separation via solubility properties,[63] which is discussed in more detail in the application chapter. However, with all these methods great care has to be taken that no metal centers or other potentially harmful fragments leach into the solution.[64,65]

While the main application potential of PHIP is clearly in Chemistry and the Biosciences, there are also a number of “more exotic” applications of PHIP in Physics. To name two recent prominent examples: Based on the concepts of MASER and LASER, Appelt et al.[66,67] developed the concept of a p-H2 pumped RASER (Radiowave Amplification by Stimulated Emission of Radiation), which delivers sub-millihertz resolution in nuclear magnetic resonance. Budker and coworkers have proposed parahydrogen based ultralow-field NMR techniques as part of the cosmic axion spin precession experiment (CASPEr),[68] an NMR-based dark-matter search to probe the axion-fermion “wind” coupling.[69]

Both PHIP and SABRE have a high application potential in the field of hyperpolarized MRI,[63,70-78] where hyperpolarized substances, such as hydroxyethyl propionate, fumarate or succinate, were employed as contrast agents (see e.g. the reviews by Hövener et al.,[79] Reineri et al.[80] and Aime et al.[81]).

Principles of parahydrogen induced polarization

Parahydrogen and orthohydrogen

Molecular hydrogen is composed of the two spin isomers: parahydrogen (p-H2) and orthohydrogen (o-H2). p-H2 is characterized by the product of a symmetric rotational and the antisymmetric nuclear singlet spin wave function. In contrast to this, orthohydrogen is characterized by an antisymmetric rotational and one of the three symmetric nuclear triplet spin wave functions. In thermal equilibrium at room temperature all four states have practically equal probability, i.e., a relative population of 25% parahydrogen and 75% orthohydrogen. At liquid nitrogen temperature (77 K) the equilibrium is shifted to roughly 50% parahydrogen and 50% orthohydrogen. In thermal equilibrium below 20 K only the ground state of the energetically lower parahydrogen is populated (see Fig. 1). In other words, in thermal equilibrium all molecules are in the nuclear singlet state and the hydrogen gas is fully hyperpolarized.

Fig. 1

Equilibrium fractions and ratio of parahydrogen and orthohydrogen as a function of the temperature.

The spin-conversion from orthohydrogen to parahydrogen is a kinetically hindered process, which is very slow in the absence of suitable catalysts enhancing the spin-conversion. This can be e.g., relaxation by a paramagnetic species[12] or via magnetic dipolar interactions due to adsorption to nuclear spins.[82] In the absence of such a catalyst, pure parahydrogen gas is stable in liquid solutions, can be stored for several weeks[83] and used subsequently for hydrogenation reactions.

A sketch of the basic PHIP theory

There are a number of variations of the PHIP technique. They are all based on the original irreversible variants, which were pioneered by the groups of Weitekamp,[13,14] and Eisenberg.[16] In these classic PHIP experiments parahydrogen is added to an unsaturated bond of a molecule mediated by the catalyst. There are two major variants of the experiment, which differ in their reaction conditions. If the hydrogen addition is performed inside the NMR magnet, the acronym PASADENA[84] is used. If the experiment is performed outside the magnet (ideally at 0 T field, in practice close to the earth field), the acronym ALTADENA[15] is employed.

Independent on the actual reaction conditions, the basic irreversible PHIP experiment can be divided into three steps (see Fig. 2A). In the first step, the catalyst, the unsaturated substrate (an alkyne in the example) and the parahydrogen react with the rate k1 to form a transient complex. In the second step, the singlet spin order of the parahydrogen evolves in this transient complex and is – at least partially – converted into visible Zeeman magnetization. In the third step, the transient complex decays to the product (an alkene in the example) and the catalyst with the rate k2.

Fig. 2

Basic reaction pathways of the catalyzed hydrogenation in a PHIP (A, irreversible) and SABRE (B, reversible) reaction. (i) The unsaturated precursor, respectively the SABRE substrate, such as pyridine Py as example, reacts with parahydrogen (marked red) and the catalyst M to form the intermediate (ii), which can be, e.g., a dihydride or dihydrogen complex. In the case of PHIP, this complex decays into the hyperpolarized product (iii) and the catalyst (see ref. 85 for details). In the case of SABRE, the hyperpolarization of the parahydrogen is (at least partially) transferred to the substrate.

There are three similar steps in the case of the reversible PHIP, i.e., SABRE, experiment (see Fig. 2B). In this case, the substrate such as e.g., pyridine, forms a transient complex with the catalyst in the first step. In the second step, the singlet spin order of the parahydrogen is transformed in this complex to the common spin orders of the substrate and the two “nascent” hydride proton spins. In the third step, the complex decays back to catalyst, substrate, and hydrogen.

Polarization and hyperpolarization

The observed intensity of a spectroscopic transition between two energy levels is proportional to their population difference. In the case of the two levels (α and β) of an NMR spin-1/2 system, a relative measure of this population difference is

In thermal equilibrium in the presence of the Zeeman-interaction with a field of B0 the equilibrium polarization is

Even for protons at the highest currently available fields of 28.2 Tesla, corresponding to 1.2 GHz of proton Larmor frequency, the equilibrium polarization at 300 K is only about Peq(1H) ≈ 9.6 × 10−5. The situation is even worse for X-nuclei such as e.g., 13C, where the lower gyromagnetic value reduces this to Peq(13C) ≈ 2.4 × 10−5. In order to achieve higher polarization while being able to measure at ambient temperature, it is necessary to bring the spin system in a hyperpolarized state, which is very far away from thermal equilibrium. The typical measures for this hyperpolarized state are the polarization value Phyp itself, or the enhancement factor

Basic 1H–PHIP theory of a two-spin system

A detailed discussion of the various variants of the PHIP experiment is beyond the scope of the present review. In the following only the basic theory for a two-proton system is sketched. More details are found e.g., in the reviews by Bowers and Weitekamp[86] or Canet et al.[87]

The initial state of a PHIP experiment is created by a chemical reaction of the parahydrogen with the substrate molecule Rto form the product P. From the point of view of the spins, the chemical reaction is a sudden change of the spin Hamiltonian from parahydrogen Hamiltonian (Jpara ≈ 3.65 × 1012 Hz)to the spin Hamiltonian of the reaction product (νS,νI: nuclear Zeeman frequencies; J ≈ 1–20 Hz scalar coupling constant)

Starting point for the NMR description is the spin density matrix of the parahydrogen, which corresponds to the singlet state. It can be written asHere {Î,Ŝ}, k = x, y, z, are the spin operators of the two protons from the parahydrogen molecule, and Ê is a unity operator. As long as the two spins are magnetically equivalent, i.e., have the same Larmor frequency (νS = νI = ν0), this density operator commutes with the spin Hamiltoniani.e.,and p does not change. In the singlet state the two protons don't give an NMR signal, since

The situation changes, when the two spins become inequivalent, i.e., when they have different Larmor frequencies. In the case of a two-spin system, this inequivalence stems from chemical shift differences. For larger systems they can also be the result of scalar couplings to other spins. In this case the spin Hamiltonian is

Its commutator with the singlet spin density matrix of the parahydrogen is no longer zero,

Thus, the density matrix becomes time dependent as soon as the symmetry of the spins is broken by the reaction. Before the actual NMR experiment, the reactions run for typically a few to a few tens of seconds. Because of this finite reaction time all the oscillations are damped out and the initial condition of the experiment at the end of the reaction time is (Δν: = νS − νI):[49]

In particular, for an AX spin system with Δν ≫ J one obtains simply as initial condition

In other words, only the spin-states αβ and βα are populated, and the population differences to the levels αα respectively ββ, which are relevant for the intensities of the NMR signals, are on the order of unity instead of on the order of 10−5, i.e., four to five orders of magnitudes larger.

Singlet–triplet conversion of hydrogen in PHIP

Konstantin Ivanov proposed, in analogy to the well-known spin-correlated radical pair model from Spin Chemistry[88] explaining the origin of the chemically induced dynamic nuclear polarization (CIDNP),[89-92] a pictorial description of the spin-dynamics of the spin-pair (see in Fig. 3). As outlined above, at high magnetic field this proton spin pair undergoes a time dependent spin interconversion with the central triplet T0 state. Singlet and central triplet states are not eigen states of two spin-1/2 system at high magnetic fields, therefore fast oscillations take place between them. These states are invisible by a conventional NMR; however, recently it was proven that singlet–triplet conversion indeed takes place in parahydrogen reversibly bound to a complex[88] and gives rise to formation of orthohydrogen. This process is remarkably efficient and strongly reduces the achievable NMR signal enhancement of a substrate (in the analyzed case the 15N nuclei of pyridine). Clear evidence that singlet–triplet conversion in the bound H2 plays an important role in SABRE experiments was taken from the nutation angle dependence of the 15N signal enhancement shown in Fig. 3. Due to the magnetic non-equivalency of two “nascent” hydride spins in their chemical shifts and spin–spin interactions with nuclear spins of substrate in the complex, this singlet–triplet conversion is sensitive to the spin orientation of the substrate spin as well. As result, the singlet–triplet conversion becomes faster in the catalytic iridium complex with 15N enriched substrates. At high field, the other important consequence of this interconversion is that it favors only one of the three triplet states, here the central triplet state T0, producing polarized o-H2 that does not obey a Boltzmann distribution. The latter gives rise to the possibility of the ultrasensitive indirect NMR detection of catalytic hydrogen complexes, as suggested in ref. 93 (see Chapter 3).

Fig. 3

Left: Diagram explaining singlet–triplet conversion for Δν ≠ 0. The red and blue arrows stand for the spin vectors of the two protons, the S state is the state with anti-parallel spins. While the total spin in the T0 state is nonzero, its z-projection is zero. The spins precess about the B0 field at different frequencies: faster precession of one of the spins (for simplicity, we assume that only I2 precesses) gives rise to coherent S–T0 mixing, i.e., S goes to a superposition of S and T0, then to T0, then again to a superposition and so on. Right: The dependence of 15N signal enhancement of the deuterated 15N-enriched pyridine on the flip angle φ of non-selective proton pulse. For details see ref. 88.

1H–PHIP of larger spin systems

The situation is more complicated in larger spin systems, where analytical solutions are generally not available. However, as long as longitudinal relaxation processes can be neglected, it is relatively easy to estimate the initial conditions numerically by a truncation[94] approach, where the transformed density matrix is averaged over the reaction time Twhich removes all oscillating terms from the density matrix. In this case, the population of the initial singlet state will be distributed in general over more levels, but the population differences will still be orders of magnitude larger than the thermal equilibrium polarization.

Proton detected PHIP

The most common way to detect PHIP hyperpolarization is by 1H-NMR. The actual line-shape of the spectra depends on the spin system, the employed pulse sequence, the strength of the external magnetic field during the hydrogenation and during the detection, the switching time between the fields and kinetic processes in the reaction intermediate. There are several excellent reviews on the 1H-detected variant of PHIP.[26,46,86,95]

The conversion of singlet spin order into nuclear spin order during the hydrogenation reaction leads to typical polarization patterns in the NMR spectra of the hydrogenation products, which depend on the spin system and its parameters, such as chemical shifts, scalar couplings, the strength of the external magnetic field during the reaction and the type of field change (e.g., adiabatic, or sudden change). Ivanov et al.[96] obtained analytical solutions for several types of coupled three spin systems and calculated the effect of the external magnetic field strength on the PHIP signal for them. This dependence is demonstrated in the simulations of Korchak et al.[42] They have calculated the effect of the external magnetic field strength on the PHIP signal for a three spin system (see Fig. 4). It is evident that the signal pattern changes drastically with the strength of the polarization field or the type of field change (adiabatic versus sudden change). Kiryutin et al.[97] analyzed the effects of nuclear spin level anti-crossings on the PHIP pattern. They found that rapid passage through the level anti-crossing enables highly efficient polarization transfer between specific spin orders, and that in general, the experimental protocol for the variation of the magnetic field is a means for manipulating PHIP patterns and transferring polarization to target spins of choice.

Fig. 4

PHIP spectra of a three-spin system calculation for different reaction fields at a detection field of 9.4 T. (a) Adiabatic (slow), (b) sudden field variation. Parameters of the calculations: φ = π/4; δ = {5.8 ppm, 6.05 ppm, 6.25 ppm}; J1,2 = 10.5 Hz; J1,3 = 1.8 Hz; J23 = 17.2 Hz. Figure reproduced from Korchak et al.[42] (reproduced from ref. 42 with permission from the Royal Society of Chemistry, copyright 2009).

PHIP detected by X-nuclei

Owing to the comparable short relaxation time of proton polarization, which is typically on the time scale of few seconds, it is advantageous to transfer the hyperpolarization to an X-nucleus, such as 13C, where the life-time of the hyperpolarization is typically an order of magnitude longer, which strongly enhances the application window of the hyperpolarization. Initial applications employed INEPT type experiments to exploit this idea and developed the PH-INEPT sequence as a very efficient tool to hyperpolarize the X-nuclei.[46,98,99] An overview about these techniques is given by Kuhn and Bargon.[100] Bernarding et al.[75,101,102] described the hyperpolarization of 19F containing substrates, which has a high application potential as contrast agent in MRI, as there is practically no 19F background in living tissue. Glöggler et al.[103,104] reported a pulsed transfer experiment, which can nearly completely transfer the hyperpolarization from the protons to the 13C. Theis et al.[105] discussed SABRE hyperpolarization for sensitivity enhancement of rare X-nuclei. Ivanov et al.[106] proposed the INEPT-SABRE sequence for polarization transfer in SABRE experiments and developed a number of versatile tools based on utilizing level-anti-crossing (LAC) in the rotating frame for transferring spin hyperpolarization. The LAC conditions can be fulfilled by applying resonant RF-excitation at the NMR frequency of the protons and the heteronuclei without necessity to utilize magnetic field cycling. Highly efficient conversion of para-hydrogen-induced polarization into net polarization of spin-1/2 insensitive nuclei by adiabatic passage through the level anti-crossings (LACs) in the doubly rotating frame of reference was demonstrated for 13C in ref. 107 and for 15N in ref. 108 The dependence of the polarization transfer efficiency on the RF-field parameters and on the time profile of switching off the RF-field has been studied in detail; experimental results are in excellent agreement with the theory developed.[109]

The Side Arm Hydrogenation (PHIP-SAH) by Reineri et al.[110] opened up a new range of hyperpolarized molecular probes. In this approach the molecule of interest is functionalized with an unsaturated moiety (the “side arm”[111]) which is chemically removed after hydrogenation with p-H2 and further transfer of the hyperpolarization to the 13C target signal. Its power was demonstrated recently by the Reineri group employing hyperpolarized [1-13C]pyruvate,[112] to detect in vivo maps of pyruvate and its metabolic product lactate on a 1 T MRI scanner. An overview on this technique is found in a number of recent reviews.[79]

Reversible PHIP (SABRE)

In the reversible PHIP (SABRE) scheme,[17] a transient complex of parahydrogen, the substrate and a suitable organometallic complex, which induces the polarization transfer from the parahydrogen to the substrate, is formed. The latter is usually an N-heterocyclic carbene (NHC) complex.[113] In this complex the hyperpolarization is transferred from the parahydrogen to the substrate through scalar couplings in the complex.[114] Later the complex dissociates again into substrate, dihydrogen and catalyst, causing a hyperpolarization of the substrate reservoir. A major advantage of the SABRE scheme is that no substrate is consumed, and many hyperpolarization cycles can be performed, thus allowing a substantial amount of the substrate to be hyperpolarized by “pumping” polarization into the substrate reservoir. The final level of hyperpolarization depends on the production and the relaxation rates of the SABRE process. This hyperpolarization is then used for the NMR detection. Afterwards, as no substrate is consumed, the same molecules can be hyperpolarized again by bubbling new parahydrogen through the solution.

A simple phenomenological model of the polarization transfer processes was given by Barskiy et al.[115] Quantitative theoretical description of the SABRE process with rigorous consideration of spin dynamics, level anti-crossings (LACs) and chemical exchange was proposed by Knecht et al.[116,117] A detailed discussion of these models is given in the review by Ivanov et al.,[118] which includes an extensive list of SABRE active substrates. In the following the salient elements of this description are summarized.

The crucial step in the SABRE process is the polarization transfer from the bound parahydrogen nuclei to the nuclei of the substrate. This transfer can occur either as a direct process via the scalar coupling connecting the two nuclei or as a relayed process, where the polarization is transferred in steps: first to nearby and stronger coupled nuclei and then to the nuclei of interest. A theoretical approach to SABRE-relay, which can treat both spin dynamics and chemical kinetics as well as the interplay between them, and allows optimization of SABRE-relay experiments with the ultimate goal of achieving maximal NMR signal enhancements for substrates of interest, was developed by Ivanov et al.[119]

Polarization transfers between protons at high magnetic field are energy conserving with respect to the dominating Zeeman interaction and occur fast under the influence of the scalar interactions. In contrast, transfers from protons to heteronuclei such as e.g., 15N, are inefficient, as the Zeeman energy is not conserved.

If the hyperpolarization is to be transferred to an X-nucleus, it is necessary to remove this Zeeman barrier and bring the two spin-reservoirs into contact. The easiest solution for establishing the contact is to perform the SABRE experiment at very low magnetic fields, where the Zeeman energies are small and efficient polarization transfer is feasible. The resulting experiment is called SABRE-SHEATH (SABRE in Shield Enables Alignment Transfer to Heteronuclei). It achieves the polarization transfer in the case of N-heterocyclic substrates, such as pyridine-15N, at fields of typically 0.2–0.6 μT.[120] At these fields there are LACs, which cause a strong mixing of the spin-states of the protons and the 15N-nuclei, and enable the transfer, as explained by Ivanov et al.[121]

At high magnetic fields alternative transfer mechanisms to X-nuclei are needed. This can be achieved e.g., by the creation of anti-phase polarization (see review by Ivanov et al.[118] and references therein) or via RF irradiation on the complex. The latter schemes are summarized under the acronym RF-SABRE. Typical examples of it are the LIGHT-SABRE (Low-Irradiation Generation of High Tesla-SABRE)[122] and the SLIC-SABRE (Spin Lock Induced Crossing SABRE)[123-125] experiments.

The essential optimization of the RF-SABRE performance for low γ nuclei as proposed by Ivanov et al.[106] is based on the utilization of the shaped RF pulses with constant adiabaticity. Based on the concept of LAC, they implemented adiabatic RF- field variation to optimize the sweeping profile in order to (a) minimize the time and (b) keep a high degree of adiabaticity. Qualitatively, this can be achieved by slowly passing through the LAC regions in rotating frame and introducing fast passage in-between LACs (see ref. 106 for details). The recent development of this approach deals with the highly efficient transfer of proton hyperpolarization to low γ nuclei by adiabatically sweeping the external magnetic field.[126,127]

Fig. 5, adapted from Chekmenev et al.,[120] displays several of these SABRE effects, employing 15N labeled pyridine as substrate. The upper panel compares the thermal (A) and the SABRE hyperpolarized (B) signals of the free and the complex bound pyridine protons. The lower panel displays the 15N-signal of the pyridine, obtained by SABRE-SHEATH (C and D).

Fig. 5

1H and 15N NMR spectroscopy of SABRE catalyst activation and 15N SABRE-SHEATH build-up. (A) 1H thermal NMR spectrum of 2 mM activated Ir-IMes catalyst solution with 48 mM 15N-pyridine. (B) 1H spectrum of hyperpolarized 15N-pyridine via conventional low-field (6 ± 4 mT) SABRE. The resonances labeled with dashed lines correspond to catalyst-associated pyridine.[59] (C and D) 15N NMR spectra of 15N-pyridine hyperpolarized by SABRE-SHEATH at micro-Tesla fields. (C) NMR spectrum of 15N-pyridine (εfree ∼ 300) sample corresponding to completely activated catalyst solution (as validated by 1H NMR using conventional lowfield SABRE through achieving efficient enhancement of pyridine proton polarization, and also validated through in situ detection of the disappearance of SABRE hyperpolarized Ir-hydride intermediate species). (D) 15N NMR spectrum of 15N-pyridine sample corresponding to maximum SABRE-SHEATH signal intensity (εfree ≈ 3600) achieved with ∼20 min of para-H2 bubbling. This figure has been adapted from ref. 120 with permission from the American Chemical Society (ACS), copyright 2015.

PHIP and SABRE at zero to ultra-low-field

An alternative approach for the hyperpolarization of X-nuclei is the zero-to-ultra-low-field nuclear magnetic resonance (ZULF-NMR) technique, where weak magnetic fields in the range of few nanotesla (nT) are employed.[128-134] At these ultra-low-fields the Zeeman-Interaction is on the order of few tenths or hundreds of millihertz (mHz), which is much smaller than the size of typical scalar interactions between nuclear spins. This creates a strongly coupled spin system, whose eigenstates are determined by the scalar interaction Hamiltonians and no longer by the Zeeman Hamiltonian. As example we consider a spin system consisting of two spin-species I and S, such as e.g., 1H and 13C. The relevant Hamiltonian of the spin system at ultra-low field B = BUL is the sum of the Zeeman terms and the scalar interactionsHere ωk = −γk × BUL, k = I, S is the angular frequency of the Zeeman interaction of the spins and JII, JSS, JIS are the homonuclear and heteronuclear scalar-coupling constants. The sum-indices i,j run over all proton respectively carbon spins in the spin system. At ultra-low fields, the difference in Larmor frequency {ωI − ωS}/2π between I and S spins is much smaller than the heteronuclear coupling constants, JIS. Thus, there is no longer a difference between homo- and heteronuclear spin and the usual distinction between protons and X-nuclei is lifted. As a consequence, polarization can be transported between the I and S nuclei by means of the IS-scalar interactions and later detected at high-field via field-cycling.

Kiryutin et al.[135] recently presented an advanced study of SABRE generated 15N polarization of metronidazole via spin mixing at ultralow magnetic fields at natural 15N isotope abundance. The most efficient polarization transfer from protons to low γ of nuclei S occurs at the level-crossing condition {ωI − ωS}/2π = JIS. The temperature variation revealed crucial difference between SABRE polarization generated for three N-atom positions in metronidazole: only one of them, namely N3, can be directly bound to the SABRE catalyst, while the other two 15N spins get polarization indirectly via coupling to protons of the polarized substrate. As an experimental example of this technique, Fig. 6 displays a magnetic field dependence of metronidazole and natural abundance of 15N. The signal enhancements are reaching 46 000 (15% polarization) for the most efficiently polarized N3-atom. In contrast to the other nitrogen positions, its optimum polarization field strongly depends on the sample temperature, which affects the kinetic parameters (describing SABRE complex formation and dissociation, lifetime of the active polarization transfer complex, τd), but not the spin Hamiltonian (Fig. 6).

Fig. 6

(A) 15N SABRE in DMSO-d6, showing the difference in field dependence of nitrogen atoms in one molecule (central part) and representative NMR spectra (lower part). 15N NMR spectrum of polarized metronidazole at 0.165 μT with proton decoupling is shown in the upper part. Lines show the calculation result, which takes into account four (solid orange) and three (green dash) interacting spins in active complex. (B) Energy levels of a three-spin system – two protons HH′ coupled to a 15N nucleus – at ultralow fields. 15N polarization generated by spin mixing in the and state manifolds (red and blue dashed lines, respectively) and the total 15N polarization (solid black line). (C) 15N SABRE field dependence for N3 at ultralow fields at variable temperature: simulation (upper plot) and experimental data (lower plot). For details see Kiryutin et al.[135]

This observation is explained by the fact that the SABRE field dependence is conditioned by LACs at zero field, which become lifetime broadened. Hence, the LAC-based analysis of SABRE remains valid, although in some cases – like the one studied here – not only the LAC position but also the LAC width becomes important. The width of LAC is conditioned by chemical dynamics of reversible binding that was rigorously described by Knecht et al.[116,119] Such a lifetime broadening of the LAC provides interpretation of the data presented in Fig. 6: upon temperature variation the τd value is strongly affected. Consequently, ΔBLAC changes and so does Bmax. This assumption is in agreement with the calculation of the SABRE field dependence performed for variable complex lifetime τd: when τd becomes short at elevated temperature, the Bmax value increases (see Fig. 6C). The effect of the lifetime becomes important when the product of τd and the characteristic spin mixing frequency νmix is smaller than 1; otherwise Bmax is determined by the Jeff parameter.

The key parameter of PHIP experiments at ultralow fields with detection at strong magnetic field is the rate of the magnetic field switching as compared with the coupling parameter among nuclei. In general, it can be slow (adiabatic), sudden (non-adiabatic)[136] or mixed. For the maximal polarization of PHIP and SABRE substrate, it is necessary to switch the magnetic field adiabatically, but as quickly as possible in order to reduce the relaxation effect. Ivanov et al.[126,127] proposed the optimized methods for adiabatic field switching for ultralow fields: the so-called field cycling (FC) and field sweeping (FS) methods called FS-f-Z and FS-t-Z, meaning field switching from zero and through zero (Fig. 7).

Fig. 7

Eigenvalue plots for the 3-spin system of [1-13C]fumarate with state labels shown. (a) Relevant energy levels for the field sweep from zero field (FS-f-Z) experiment are highlighted with white dashes, and the one relevant LAC is labelled. (b) Constant-adiabaticity profile. (c) Relevant energy levels for the field sweep through zero field (FS-t-Z) experiment are highlighted, and the three relevant LACs are labelled. (d) Constant-adiabaticity profile. Colors distinguish between state manifolds with different angular momentum projection (grey: m = ±3/2; blue: m = +1/2; red: m = −1/2). For details see Rodin et al.[126] (reproduced from ref. 126 with permission from the Royal Society of Chemistry, copyright 2021).

Previously, such methods used either uncontrolled switching of the magnetic field, for example, taking the sample out of the magnetic shield, or linear switching of the magnetic field, which is not the optimum form for ensuring fast and efficient polarization transfer. The point is that on the one hand the field switch should be as fast as possible but on the other hand the field has to change slowly in the LAC regions. Ivanov et al.[126] proposed a switching profile that meets this requirement and provides faster and more efficient polarization transfer (see Fig. 7). They have demonstrated this by the example of polarization transfer from the proton singlet order to the 13C magnetization in PHIP-polarized [1-13C]fumarate obtained from the precursor by hydrogenation with parahydrogen. To provide this shape, we proposed a profile of field variation with a constant adiabaticity parameter ξ, which was previously used to improve methods for generating long-lived singlet order[137]Here (i,j) denote the eigenfunctions of the complete Hamiltonian Ĥ, ω – the energies difference of these levels. The field switching profiles for the FS-f-Z and FS-t-Z methods, obtained using the final formula for the specific molecule fumarate, are shown in Fig. 8.

Fig. 8

Optimization of polarization transfer by adiabatic profiles of magnetic field sweep. (a) Experimental protocols for uniform linear (black) and constant-adiabaticity (blue) field sweeps; b) Comparison of hyperpolarized and thermal equilibrium 13C NMR spectrum of 500 mM [1-13C]fumarate in D2O. The asymmetry in the triplet structure in the hyperpolarized spectrum is due to the nonthermal proton spin polarization. (c and d) Results from the field sweep experiments, contrasting constant-adiabaticity and linear magnetic field profiles for field sweep from zero field (FS-f-Z) and field sweep through zero field (FS-t-Z) experiments. Simulations are shown by the lines fitted to the data. For details see Rodin et al.[126] (reproduced from ref. 126 with permission from the Royal Society of Chemistry, copyright 2021).

As can be seen from the experimental results (Fig. 8), a profile with a constant adiabaticity parameter makes it possible to achieve a maximum of polarization transfer in a shorter time, which makes it possible to reduce relaxation losses and increase the overall quality of polarization transfer. Also, these methods are not very sensitive to the uniformity of the longitudinal and transverse fields (compared to coherent methods), which also improves their potential attractiveness as a new and general method for transferring a singlet hyperpolarized order to a heteroatom near LAC region at ultralow magnetic fields. These results make it possible to use profiles with constant adiabaticity for all PHIP experiments with field cycling, where after the addition of parahydrogen a strong coupling between protons and heteronuclei remains and the protons are in a state close to the equivalence.

SABRE polarization in a wide field range

In the previous sections we have seen examples for SABRE at very low fields and at high fields. In a series of papers[138-142] the Ivanov group has analyzed the full field dependence of SABRE. They found that there are four different field regimes (see Fig. 9).

Fig. 9

(A) 13C SABRE field dependence for ortho carbon of 15N-pyridine, for details see ref. 142. (B) 1H SABRE field dependence for hydride protons on ImesIr-catalyst (unpublished results provided by A. Kiryutin).

Zone 1 is the regime of high field of NMR spectrometers: the SABRE polarization on hydride protons and on free hydrogen at high field increases proportionally to the magnetic field.[138] In this regime the polarization transfer can be split into 3 steps: (i) generation of the II spin order of Ir-HH; (ii) spin conversion of II to I+I, leading to generation of net polarization of Ir-HH; (iii) transfer of net polarization Ir-HH to the nearest nuclei (e.g., protons of substrate) by dipolar cross-relaxation.

Zone 2 is a narrow field range from ca. 30 mT to 0.2 T. In this regime very strong multiplet polarization on 13C nuclei with strong direct spin–spin couplings on the order of hundred Hz is observed. At these fields no net polarization of the carbons is created. Instead, multiplet proton–carbon polarization is formed via LACs, which determine the SABRE field dependence. In particular, the position of the sharp feature in the magnetic field dependence scales by direct JCH spin–spin constant.

Zone 3 with magnetic fields from 5 to 30 mT are typical conditions for the direct proton SABRE polarization transfer from hydrides to protons of bound substrate. E.g., the seminal discovery of proton SABRE by Duckett et al.[17] was done in the field of 20 mT. Later the position was theoretically explained[139] and experimentally proved by thorough measurements in a wide field range.[142] In a simple three spin model AA′B, where A and A′ are the chemically equivalent protons attached to the iridium catalyst and B is target proton in the substrate, the optimal conditions for polarization transfer is given aswhere γH – gyromagnetic ratio of protons, δA, δB – chemical shifts of spins A and B, JAA′ – spin–spin coupling constant between A and A′ spins originating from parahydrogen.[139]

Zone 4 is the regime of ultralow field matching conditions (discussed in the previous section) with strong coupling conditions between protons and heteronuclei, also known as the SABRE–SHEATH[143] regime. The polarization transfer can occur between the hydride protons originating from parahydrogen and any heteronuclei of interest in the substrate that is reversibly bound to the iridium complex.[140,141]

Applications of parahydrogen in NMR

Parahydrogen in catalysis

Most of the initial applications of PHIP were related to the investigation of organometallic catalysts and homogeneous catalysis (see e.g. reviews by Duckett[144,145] and Münnemann et al.[146]). In particular, PHIP is one of the most powerful techniques for the in situ NMR investigation of hydrogenation reactions and their catalytic mechanisms.[145,147,148] In recent years there is a growing number of studies, where metal-nanoparticles or other heterogeneous catalysts were employed for the generation of PHIP hyperpolarized gases like propene.[38,56,149-172] A major driving force for these developments is the search for the cost-efficient production of hyperpolarized gases for contrast enhancement in medical applications of NMR such as MRI imaging of the lung, as PHIP does not depend on expensive additional hardware and is thus more amenable to implementation in medical diagnostics.[173-175]

In parallel to the applications in MR signal enhancement, PHIP is also an excellent tool for the investigation of the basic hydrogenation mechanisms, and, in the case of homogeneous catalysis also for the investigation of catalytic intermediates.[176] An essential condition for the formation of PHIP is a pair-wise hydrogenation of an unsaturated bond, where both hydrogen atoms stem from the same hydrogen molecule. Owing to this condition, most heterogeneous metal catalysts cannot create PHIP, as the hydrogen molecule dissociates on their surfaces to atomic hydrogen, which migrates on the surface, and hydrogen atoms from different molecules react with the substrate. In order to achieve heterogeneous HET-PHIP by such catalysts, it is necessary to reduce the migration radius by partial poisoning of the surface.[149,153] After such a reduction the spin-coherence between the two hydrogen atoms can be conserved, e.g., by dipolar interactions,[177] and result in a PHIP. An overview about these variants of PHIP was recently published by Pokochueva et al.[178]

Ultrasensitive indirect NMR detection of catalytic hydrogen complexes via parahydrogen

Kiryutin et al.[93] revealed and explained spin exchange processes between molecular hydrogen with two equivalent nuclear spins and transient hydride complexes with two inequivalent hydrogens with different chemical shifts, in which singlet–triplet evolution takes place (see Fig. 10). Spectroscopically, this process is observed as nearly complete overlapping of a negative and a positive line with only slightly different frequencies, leading to a negative component on the low-field part of the absorptive peak of dissolved free hydrogen in solution; hence, it was called Partially Negative Line (PNL) of H2 in PHIP experiments. The PNL effect was initially found during the investigation of the PHIP enhancement of small oligopeptides. In these experiments, it was found that the 1H NMR signal of dissolved p-H2 enriched molecular hydrogen exhibits, in the presence of an organometallic hydrogenation catalyst, this unusual PNL-shape. A detailed experimental and theoretical study proved that the PNL results from a strongly enhanced two-spin order, interrelated with selective population of the T0 level of orthohydrogen (o-H2). This two-spin order is made visible by a slow asymmetric exchange process between free hydrogen and a transient catalyst–hydrogen complex.

Fig. 10

(a) Magnetic conversion of p-H2 (spin state S) to o-H2 in spin state T0. (b) The PANEL (PArtially NEgative Line) experiment for the indirect detection of the hydrogen catalyst complex. (c) 1H NMR PHIP spectrum recorded during the hydrogenation with p-H2 of Fmoc-O-allyl-tyrosine (left) to Fmoc-O-propyl-tyrosine (right), showing a strong PNL signal at 4.53 ppm (for details see Kiryutin et al.[93]).

By Only Parahydrogen Spectroscopy (OPSY)[179] it was feasible to selectively detect the two-spin order and suppress the signal from thermal o-H2. It was found that the intensity of the PNL can be strongly affected by cw-like low-power radio frequency (RF) irradiation, resulting in a technique called PANEL (Fig. 10). When the RF-frequency is on resonance with the chemical shift values of one of the hydrogens bound to the elusive catalyst or of the free hydrogen, a strong intensity reduction of the PNL is observed (Fig. 11). Numerical simulations of the experiments performed at 500 and 700 MHz proton frequency showed that the indirect detection has an at least three orders of magnitude higher sensitivity than the normal NMR experiment.

Fig. 11

(A) Thermal 700 MHz 1H NMR spectrum accumulated after bubbling the catalyst solution by n-H2 (512 scans, 30 minutes) and magnification (×1000) of the dihydrogen and dihydride region. Note that there is no signal of any complex between catalyst and hydrogen. (C and D) Experimental and simulated PANEL spectra, showing the signal of the hidden complex (lines at −16.5 ppm and −13.5 ppm). (B) Calculated NMR spectrum of the hidden complex, magnified by factor 107 (for details see Kiryutin et al.[93]).

A theoretical model, including reversible binding and S–T0 evolution, was developed, which reproduces the NMR line-shape, the nutation angle dependence, and the dependence on the frequency of the irradiation field of the PNL, and permits the determination of the proton chemical shift values and the sign of the scalar coupling in the transient NMR-invisible complex, where singlet–triplet conversion takes place. A detailed theoretical analysis of the effect is given in the paper by Kiryutin et al.[93]

Understanding and modelling catalytic hydrogenation reactions for PHIP and SABRE

In the continuation of this work, the possibility to utilize non-hyperpolarized Chemical Exchange Saturation Transfer (CEST) NMR in a similar fashion to study short-lived hydride intermediates in the catalytic cycle of an organometallic complex was explored. As example, the SABRE catalyst [Ir(IMes)(H)2(Py)3]+Cl− was chosen (complex 1 in Fig. 12). In the course of the SABRE formation, transient complexes of the catalyst, the p-H2 and the substrate (e.g., pyridine) are formed. Understanding the spin dynamics in these complexes is necessary for enhancing the performance of the nuclear spin hyperpolarization technique SABRE. These complexes are low concentrated and undergo reversible ligand exchange with the main complex. They are typically not observable by other NMR techniques.

Fig. 12

Upper panel: Molecular diagram of the exchange between the main organometallic complex (SABRE catalyst) [Ir(IMes)(H)2(Py)3]+Cl− (1) and its intermediates 2 and 3 in solution (methanol-d4). Lower panel: (A) Experimental 1H NMR CEST scheme. Continuous wave (CW) irradiation with an amplitude (v1 = 50 Hz) is applied for 3 s at a particular offset frequency vRF and the integral of the free H2 signal is recorded; then the offset frequency is incremented, and the experiment is repeated again yielding the CEST spectrum. (B) Standard 1H NMR (500 MHz) spectrum of 1 in methanol-d4. The inset shows the CEST effect on the free H2 resonance when saturation is applied to the hydrides of 1 (vRF = 22.72 ppm). (C) 1H NMR CEST spectrum of 1 in methanol at 280 K showing the presence of dihydride intermediates 2 and 3. Black: experimental data; red, blue and green lines are Lorentzian fits (for details see Knecht et al.[180]).

Knecht et al.[180] could show that by eliminating [Ir(IMes)(H)2(Py)3]Cl (complex 2 in Fig. 12) and manipulating the spin system by RF-irradiation, the nuclear spin singlet life-time of the hydride protons can be increased by more than an order of magnitude, from 2.2 ± 0.1 s to 27.2 ± 1.2 s, and the intermediate complexes [Ir(IMes)(H)2(Py)3] and [Ir(CD3OD)(IMes) (H)2(Py)2] can be detected, assigned and characterized in solution, in situ and at room temperature (see Fig. 12 and Knecht et al.[180] for details). Due to its simplicity and ability to unravel unobservable chemical species, the utilized CEST NMR approach has a large application potential for studying short-lived hydride intermediates in catalytic reactions. To model such phenomena, Knecht, Ivanov et al.[119] developed techniques for simulating the spin dynamics in the presence of chemical exchange. These techniques allowed them to describe the formation of hyperpolarized o-H2 and to model quantitatively the PANEL experiment. Employing these techniques, it was possible to describe the polarization formation in SABRE-relay, an extension of the traditional SABRE for polarizing a wider range of substrates by exploiting additional exchange processes (see Knecht et al.[119] for details). These techniques were later extended to treat highly complex processes with p-H2, involving several exchange pathways.[181]

In situ gas-phase NMR and para/ortho conversion

In the previous two cases we have observed a magnetic para/ortho-conversion of the parahydrogen in the complex. The p/o conversion can take place via various magnetic mechanisms by binding of a single H2 to a catalytic center or via binding of two hydrogen molecules and reorganization of their bonds. In particular, the magnetic conversion mechanism must be distinguished from the chemical conversion by chemical exchange between two hydrogen molecules under the influence of a catalyst. As the chemical process is the analogue of the isotopic scrambling reaction[182]H

In situ gas-phase NMR provides efficient means to probe for chemical conversion mechanisms. It has been shown, that systems catalyzing isotopic scrambling also catalyze the hydrogenation of alkynes and alkenes.[183] Characteristic for these investigations was that the heterogeneous catalyst was in a diamagnetic – or at least – mostly diamagnetic state and could be employed directly inside the NMR magnet by placing it on the bottom or the inner surface of the NMR tube.

Fig. 13 shows two typical examples of this work. In both cases, transition metal nanoparticles (MNPs) stabilized by an organic coating were employed. The first example, taken from Pery et al.,[184] displays results obtained for Ru nanoparticles stabilized by hexadecylamine as a protecting ligand. These nanoparticles were investigated by 1H-gas-phase NMR spectroscopy. The MNPs contain adsorbed mobile hydrogen atoms. In order to control the surface hydrogen content, the MNPs were evacuated and exposed to a known hydrogen (H2) atmosphere. In the next step the H2 atmosphere is replaced by a D2 atmosphere, and the conversion of HD on the surface and its desorption are monitored by the appearance of the HD-gas peak at 4.5 ppm (note: HD and H2 both resonate at 4.5 ppm, but can be distinguished by their line widths, which differ by a factor of ca. six[186]). The right panels of Fig. 13 display the results of similar experiments on Ru, Pt and Ru/Pt nanoparticles stabilized with 1,4-bis(diphenylphosphino)butane (dppb) at deuterium pressures of 1 bar and 2 bar.

Fig. 13

(A) Upper panel: Gas phase 11.4 Tesla 1H NMR spectra showing the production of HD from the reaction of Ru/hexadecylamine nanoparticles containing adsorbed H with gaseous D2 in a glass sealed NMR tube (for details see Pery et al.[184]). (B and C) Typical 1H gas-phase NMR spectra after exposing solid MNPs at room temperature to 1 bar or 2 bar D2 gas for 16 h with Ru/dppb, RuPt/dppb and Pt/dppb nanoparticles at two different pressures of D2 gas. The chemical shift of the HD signal is set to 4.5 ppm (for details see Rothermel et al.[185]).

Potential of PHIP hyperpolarization in heterogeneous catalysis and solid-state NMR

In all cases described above PHIP was always employed for hyperpolarization of fluid (e.g., liquid, or gaseous) samples. Owing to the versatility and cost-efficacy of PHIP based hyperpolarization, an obvious question is, whether PHIP can be employed for the investigation of solid surfaces, similar to the SENS DNP experiment.[187] This would, in contrast to DNP, permit, e.g., the investigation of surface processes on technical catalysts under real world conditions, e.g., at both ambient and elevated temperatures and pressures. This question was theoretically studied by Buntkowsky and Ivanov,[177] who analyzed the spin-dynamics of a heterogeneous catalyzed hydrogenation reaction on the surface of a metal catalyst as a means for creating hyperpolarized surface sites.[177] They assumed that the two hydrogen atoms, stemming from the parahydrogen molecule, are bound as hydrides to different metal centers on a solid surface (Fig. 14). In this case, the J-coupling between the protons is small, as compared with dipolar interactions, which have dominant effect on the spin dynamics.

Fig. 14

PHIP on a solid metal surface. Left panel: the magnetic dipolar interaction d couples the two protons from the parahydrogen molecule chemisorbed on the metal surface. Right panel: enhancement as a function of the normalized time χ = (dt) (for details see ref. 177).

Analytical expressions for the signal enhancement in solid state PHIP NMR spectroscopy employing off-resonant solid state echo detection are developed and tested numerically. Both the analytical calculations and the numerical simulations show clearly that an efficient enhancement of the proton NMR signal in solid state NMR studies of chemisorbed hydrogen on surfaces is possible employing non-resonant solid echo NMR spectroscopy. Assuming a typical reaction efficacy of ca. 5%, enhancement-factors of ca. 30–40 are expected for two-pulse respectively single pulse experiment, which correspond to three orders of magnitude of sensitivity enhancement. In an extension of work, it was shown that the technique is also applicable in the PASADENA case, where the hydrogenation is performed inside the magnet.[40] This result has important consequences for the practical application of the method, since it allows the design of an in situ flow setup, where the parahydrogen is adsorbed and desorbed from catalyst surfaces inside the NMR magnet.

Application of parahydrogen induced polarization in biochemistry

One of the major application areas of hyperpolarization in general and PHIP in particular is in the field of biochemistry, biophysics, and medical imaging. Characteristic for these applications is the employment of a hyperpolarized marker, which is detected in the NMR spectrometer, respectively the MRI scanner. Such a marker can be a simple noble gas atom like 3He or 129Xe,[188,189] or a small organic molecule, such as hyperpolarized fumarate or a large peptide, like the sunflower-trypsin inhibitor SFTI-1.[190] A very detailed review about the current state of the art of PHIP hyperpolarized metabolites for biological studies with strong emphasis on medical applicability was recently published by Reineri.[80] For this reason the current review focusses on applications related to fumarate, amino acids, peptides and proteins.

PHIP on peptides and amino acids

Owing to the robustness and versatility of solid-phase peptide synthesis (SPPS)[191] on the one hand and the pharmacological importance of peptide based drugs, employing hyperpolarized peptides as markers in magnetic resonance tomography or in biophysical studies on the other hand, PHIP of peptides and amino acids has an immense technical potential.[192] There are several feasible strategies to exploit this potential. Reversible PHIP techniques, like SABRE, can be applied without modifications of native amino acids.[193]

As it is relatively easy to incorporate PHIP markers into peptides, there is a very active research in the field of hyperpolarized amino acids, peptides and small proteins.[50] This interest is driven by the pharmacological importance of peptide based drugs, or their application potential as hyperpolarized peptide markers in biophysical studies or magnetic resonance tomography.

One possible strategy is the chemical modification of amino-acid residues. As an example, Bommerich et al.[194,195] employed the acidification of amino acids for e.g., the hyperpolarization of allylglycine or vigabatrin in aqueous solution.

The more versatile and robust pathway to obtain PHIP enhanced NMR of peptides is the insertion of fluorenylmethyloxycarbonyl (fmoc) protected non-natural amino acids with unsaturated PHIP active groups into a peptide via SPPS. The unsaturated group can be a dehydrogenized backbone bond,[192,193,196] or an unsaturated side chain, such as an allyl- or alkyne-group.[190,197] The latter approach has the advantage that it imposes smaller restrictions on the conformational freedom of the peptide strand and thus has less influence on the secondary structure of the peptide and its biological function. As the allyl-groups are the hydrogenation product of the propargyl-groups, it is more convenient to employ the latter, as this permits two consecutive hydrogenation reaction steps with the p-H2: (i) from the triple to the double bond and (ii) from the double bond to the alkyl.

Fmoc-protected propargyl-glycine (Pra) is the smallest non-natural amino acid with an unsaturated side chain. In a systematic study by Koerner et at.,[197] the PHIP activities of a series of Pra-containing tripeptides and two decapeptides were evaluated. They found that the PHIP activity depends strongly on the amino-acid sequence. While peptides like Ala–Pra–Ala, Phe–Pra–Phe or Met–Pra–Met yielded strong signal enhancements, Cys-containing combinations of amino acid residues, such as Cys–Pra–Cys, did not show any measurable PHIP activity. Moreover, even in the PHIP active peptides the obtained enhancement was about an order of magnitude smaller than the enhancement obtained for neat Pra. In order to achieve higher and more stable enhancements, Sauer et al.[190] systematically searched for PHIP-markers suitable for the hyperpolarization of larger peptides, in aqueous solution, i.e., under biocompatible conditions. Employing various tripeptides as test-systems, they could show that an isolated spatial location of the triple bond within the respective label and its accessibility for the hydrogenation catalyst are essential factors for the amount of signal enhancement. As a result of their research, they could show that the labeling of the peptides with O-propargyl-l-tyrosine stably yields very strong 1H-NMR signal enhancements.

Fig. 15 shows a typical example application. The label is inserted into a bioactive derivative of the sunflower trypsin inhibitor-1 (SFTI-1), a naturally occurring disulfide-bridged cyclic tetradecapeptide,[198-200] employed as potent inhibitor of serine protease trypsin[200,201] and in tumor diagnostics and treatment.[202,203] Initial ALTADENA type experiments[190] yielded signal enhancement of a factor of 70 for the hyperpolarized protons H2 and H3. Under PASADENA conditions, employing an automated setup for PHIP experiments at higher pressures, enhancement factors of about 1200 (Fig. 15A) were achieved.[204] Due to this very high signal enhancement, it was feasible to measure a full 2D-NMR spectrum of the hyperpolarized protons in a single shot,[205] by combining ULTRAFAST[206-212] (UF) 2D-NMR with “out-of-phase echo”[40,213] detection. The example shows the hyperpolarized UF-PHIP-TOCSY spectra obtained after hydrogenation with 90% enriched p-H2 and 0.5 mM and 0.3 mM solutions (Fig. 15B and C) of hyperpolarized labeled SFTI-1 in a single scan. Despite the low concentration of the hyperpolarized samples, the 2D-NMR spectra including cross-peaks are clearly visible and well resolved. Moreover, as only the hyperpolarized signals are visible, the PHIP enhanced UF-2D-NMR approach provides a very efficient suppression of all thermally polarized background signals. The estimated resulting advantage in measurement time, compared to a standard 2D-TOCSY, is a factor of approximately 180.000.[205] This number clearly demonstrates the potential of PHIP enhanced UF-2D-NMR e.g., to monitor kinetics of low-concentrated peptidic analytes in reactions. The comparison of the thermal and hyperpolarized UF-2D-NMR spectra also shows the intrinsic spectra-editing effect of PHIP. Since only the protons stemming from the parahydrogen and neighboring protons with strong scalar couplings are visible, the hyperpolarized spectrum is much simpler than the thermal one.

Fig. 15

PHIP enhanced 1H-NMR spectra of a bioactive derivative of the sunflower trypsin inhibitor-1. (A) 1D-NMR (0.96 mM); (B) single scan UF-PHIP-TOCSY (0.5 mM) and 2.5 mM catalyst; (C) single scan UF-PHIP-TOCSY (0.3 mM) and 1.5 mM catalyst. 90% enriched parahydrogen was injected at 3 bars for a period of 10 s. Catalyst = (Rh(dppb)(COD)BF4), solvent = methanol-d4. Signal marked with an asterisk (*) is an impurity (for details see Sauer et al.[190] and Kiryutin et al.[205]).

Fleckenstein et al.[214] recently reported a PHIP labelled bioactive derivate of eptifibatide (integrilin), an antiplatelet aggregation inhibitor, which derives from the disintegrin protein barbourin in the venom of certain rattlesnakes. The PHIP-label was synthesized and inserted into the disulfide bridge of eptifibatide via reduction of the peptide and insertion by a double Michael addition under physiological conditions (see Fig. 16). This procedure is universally applicable for disulfide-containing biomolecules and preserves their tertiary structure with a minimum of change.

Fig. 16

PHIP labelling of eptifibatide (integrilin, 1) by insertion of the ethynyl-containing label 2 into the disulfide bridge forms 3. Hydrogenation of the side-chain alkin-group with p-H2 yields the hyperpolarized alkyl- (4) and allyl- (4′) product. Note: for clarity only the final alkyl-product is shown. For details see Fleckenstein et al.[214]

Reversible PHIP techniques like SABRE can be applied without modifications of native amino acids as was shown by Glöggler et al.[215] They showed, that non-protected proteinogenic amino acids can be hyperpolarized by SABRE via association of the amino- or hydroxy-group to the hydrogenation catalyst, albeit resulting in relatively low polarization levels. Later, Ratajczyk et al.[216,217] demonstrated that non-natural amino acid residues containing pyridyl-groups are more efficient labels for SABRE hyperpolarization and can be employed in physiologically relevant substrates, such as the PyFALGEA oligopeptide ligand, which is selective towards the epidermal growth factor receptor (EGFR).

PHIP of metabolites: fumarate as a case example

Fumarate has an immense potential for in vivo MRI applications. The hydrogenation of fumarate to malate by the enzyme fumarase is one of the key steps of the Krebs-cycle and thus of interest for monitoring the efficacies of cancer therapies. Employing 13C-DNP hyperpolarization, Gallagher et al.[218] could show that malate production from fumarate is increased in treated lymphoma cells and tumors by tumor cell necrosis. Thus, fumarate is an efficient probe to monitor the efficiency of cancer therapies. This seminal discovery started an intense research on the application of hyperpolarized fumarate by MRI.[219-225] Eills and coworkers realized that also PHIP is able to efficiently hyperpolarize fumarate.[226,227] Strong advantages of the PHIP based approach are that the production of the hyperpolarized substrates is faster and scalable with respect to the produced amounts of hyperpolarized molecules, and that the necessary equipment is at least one order of magnitude less expensive than a DNP apparatus and can be easily operated in a clinic environment.

Salient points for the practical application are that: (a) most clinical scanners can monitor only protons; (b) protons are far more sensitive because of their high gyromagnetic ratio and natural abundance; (c) protons permit higher spatial resolution, as they interact stronger with magnetic field gradients than e.g., 13C. Thus, protons are clearly the nuclei of choice for clinical applications. However, there are also a number of challenges in the application of protons, as compared to 13C. The narrow chemical shift dispersion and the large proton background make it difficult to resolve the weak signals of a protonated substrate, and the comparatively short proton T1 strongly narrows the application time-window of the hyperpolarization. A possible solution to this problem is the application of singlet-state NMR.[137,228-234] When p-H2 is added to an unsaturated precursor molecule, the protons remain in a nonmagnetic singlet state, as long as they remain chemically and magnetically equivalent. This state is neither directly observable in NMR or MRI, nor addressable by RF pulses. Additionally, the proton singlet state is immune to certain relaxation mechanisms,[36,37] and can have a much longer lifetime. Moreover, singlet-state NMR offers a second advantage, which solves the background problem. Employing the “out of phase echo” (OPE)[213] pulse-sequence with a nonselective 45° or selective 90° detection pulse (OPE-45 or OPE-s90, respectively) and phase-cycles, the background signal of the Zeeman-polarized protons can be suppressed, while only the hyperpolarized molecules are detected. The hyperpolarization can be stored in the singlet state until the molecule undergoes a chemical reaction that renders the protons chemically or magnetically inequivalent. This breaks the proton singlet state, and observable hyperpolarized NMR signals are released.

In the case of fumarate, this effect is created in two steps. In the first step, the hyperpolarized singlet state of fumarate is generated from dicarboxylic acid by hydrogenation with p-H2. In the second step (see Fig. 17, upper panel), the fumarate is converted by fumarase to malate, which breaks the magnetic equivalence of the two hyperpolarized protons and renders them NMR active. In a proof of principle experiment (Fig. 17, lower panel), Eills et al.[63] could show that this strategy works indeed by detecting hyperpolarized malate signals of a phantom.

Fig. 17

Upper panel: Sketch of the conversion of hyperpolarized fumarate into malate-D2, by the fumarase, followed by OPE-45 or OPE-s90 to convert the antiphase proton spin order into in-phase magnetization and suppress water background signals. Lower panel: (a) Imaging phantom. (b) Pulse sequence to acquire hyperpolarized 1H images. (c and d) Comparison between the hyperpolarized and thermal 1H images. For details see Eills et al.[63]

Moreover, fumarate also offers an elegant opportunity to solve the problem of separating the hyperpolarized product from contaminants, such as e.g., residues of the catalyst. The hyperpolarized fumarate can be purified by acid precipitation as a pure solid, and later redissolved at a chosen concentration in a clean aqueous solvent. It was found that it is possible to form hyperpolarized fumarate at several hundred millimolar concentrations, at 13C polarization levels of 30–45% (see Knecht et al.[77]). Rodin et al.[127] have developed efficient and robust strategies to convert the proton spin order of fumarate into polarization of 13C nuclei. The new pulse sequences exploit adiabatic RF pulses and result in 96% of the maximal theoretically allowed transfer efficiencies. In high-field experiments with p-H2, this allowed to achieve enhancement factors of the order of 10 000.

Equally important is the optimization of the reaction conditions and the reaction protocol for PHIP of fumarate in water as a physiological medium (Fig. 18). Compared to typical organic solvents, water has a higher viscosity and surface tension and a strong tendency to form bubbles and foams. This problem could be solved by using a glass-capillary which is fixed by spacers inside the NMR tube. In the next step, the dependence of the fumarate production on temperature, pH, and sodium sulfite concentration was determined. Employing these optimized conditions, proton enhancement factors of 3500 at 500 MHz, corresponding to polarizations of P(1H) = 28%, were obtained (for details see Wienands et al.[78]).

Fig. 18

(a) Photo of the tube with optimized spacers and capillary inside. Influence of (b) sodium sulfite concentration and (c) pH on the formation of fumarate; the inset plots show the maximal fumarate concentration reached and the time it took to reach the maximum. For details see Wienands et al.[78]

Outlook and perspectives

With the present review we wanted to demonstrate that parahydrogen-based hyperpolarization techniques have become efficient tools for the sensitivity enhancements of NMR and MRI. With their typical signal enhancement factors of 1000 and above they drastically reduce the time or amount of substance necessary for obtaining high-resolved NMR spectra. In the application area, these signal-enhancements enable the ultra-sensitive detection of catalytic reaction intermediates or the ultra-fast detection of low-concentrated peptides in solution, which would have been completely impossible without them. Moreover, as demonstrated with the fumarate example, it is nowadays feasible to hyperpolarize and clean a biological metabolite, amenable in the future as a marker for the success of cancer therapies.

There are two major factors for this success, namely, on the one-hand the technological advances in the field of reversible and irreversible PHIP and the understanding of the underlying spin-dynamics, and on the other hand the development of efficient transfer complexes, catalysts, and unsaturated substrate precursors. In the field of medical NMR, one can expect the first successful applications of PHIP hyperpolarized substrates, e.g., for the in vivo analysis of pharmacokinetics, in the not-too-distant future.

However, also in the field of catalytic studies, parahydrogen has still an immense application potential. PHIP combined with H/D-exchange, solid-state NMR or gas-phase NMR techniques is extremely powerful for the monitoring of heterogeneously catalyzed surface reactions on metal-nanoparticles,[185] and in particular for para/ortho conversion reactions and isotope scrambling.[82,183,235,236]

A further interesting application in the field of PHIP generated hyperpolarization is benchtop NMR.[237,238] Nowadays, there are several commercial vendors, who offer two-channel high-resolution benchtop solution state NMR spectrometers with magnetic fields on the order of 0.9–1.9 Tesla, with very good resolutions well below 1 Hz for protons at prices well below a standard super-conducting high-field NMR spectrometer. As these benchtop spectrometers work cryogen free and are relatively small, they can be easily employed e.g., in an organic preparative lab, under harsh environments or at places, where no cryogens are available. For a detailed overview of their application potential see the recent special issue on compact NMR.[238] The main drawback of these machines is their reduced sensitivity, owing to the low Boltzmann polarization at these low fields. However, combined with parahydrogen based hyperpolarization, these benchtop spectrometers provide highly sensitive solutions.[239-241]

While the PHIP hyperpolarization is conventionally detected by RF-coils, this is not the only possible detection mechanism. In particular at very low fields or for very small sample volumes, where the sensitivity of RF-detection decreases drastically, techniques which directly measure magnetic field strength, such as atomic magnetometers[129,242] or superconducting quantum interference devices (SQUIDs)[243,244] or nitrogen vacancy (NV)[245] based techniques, which have the potential to detect concentration of spins, comparable to a single cell, have strong advantages.

In summary, recent developments of PHIP have paved the way to new applications of magnetic resonance in chemistry, biophysics, and medicine on “real world systems”. With further improvements in technology and the physical understanding of the underlying transfer processes, the levels of hyperpolarization will further increase and permit new experiments and diagnostic techniques which were considered impossible a decade ago.

Adiabatic longitudinal transport after dissociation engenders net alignment

Cosmic axion spin precession experiment

Chemical exchange saturation transfer

Chemically induced dynamic nuclear polarization

Continuous wave

Dimethyl sulfoxide

Dynamic nuclear polarization

1,4-Bis(diphenylphosphino)butane

Field cycling

Fluorenylmethyloxycarbonyl

Field sweep

Hexadecylamine

PHIP catalyzed by a heterogeneous catalyst

Insensitive nuclei enhancement by polarization transfer

Combination of INEPT and SABRE

Level-anti-crossing

Low-irradiation generation of high Tesla-SABRE

Metal nanoparticles

Magnetic resonance imaging

N-Heterocyclic carbene

Nuclear magnetic resonance

Nitrogen vacancy

Orthohydrogen

Only parahydrogen spectroscopy

Partially negative line experiment

Parahydrogen and synthesis allow dramatically enhanced nuclear alignment

Parahydrogen

Parahydrogen induced polarization

PHIP side arm hydrogenation

PHIP relayed via proton exchange

Partially negative line

Pyridine

Radiowave amplification by stimulated emission of radiation

Radio frequency

SABRE with transfer at high field via RF irradiation

Signal amplification by reversible exchange

SABRE in shield enables alignment transfer to heteronuclei

Surface enhanced NMR

Spin-exchange optical pumping

sunflower-trypsin inhibitor 1

Spin lock induced crossing SABRE

Stimulated nuclear polarization

Solid-phase peptide synthesis

Superconducting quantum interference devices

Total correlation spectroscopy

Ultrafast

Ultrafast 2D-NMR

Conflicts of interest

There are no conflicts to declare.