Homogeneous Catalysts for Hydrogenative PHIP Used in Biomedical Applications

Abstract

At present, two competing hyperpolarization (HP) techniques, dissolution dynamic nuclear polarization (DNP) and parahydrogen (para-H₂) induced polarization (PHIP), can generate sufficiently high liquid state ¹³C signal enhancement for in vivo studies. PHIP utilizes the singlet spin state of para-H₂ to create non-equilibrium spin populations. In hydrogenative PHIP, para-H₂ is irreversibly added to unsaturated precursors, typically in the presence of a homogeneous catalyst. The hydrogenation catalyst plays a crucial role in converting the singlet spin order of para-H₂ into detectable nuclear polarization. Currently, rhodium(I) bisphosphine complexes are the most widely employed catalysts for PHIP, capable of catalyzing the addition of para-H₂ to unsaturated precursors in organic solvents or aqueous media, depending on the ligand. Chiral catalysts enable the stereoselective production of hyperpolarized substrates. Ruthenium(II) piano stool complexes are capable of trans addition and are used to generate hyperpolarized fumarate. However, these catalysts systems are not optimal, and the greatest source of nuclear spin polarization loss is attributed to the mixing of singlet and triplet states of the protons derived from the para-H₂ during the hydrogenation process. Hence, future efforts should focus on enhancing the efficiency and kinetics of these catalysts.

Graphical Abstract

Hydrogenative parahydrogen-induced polarization (PHIP) can generate high nuclear spin polarizations via the addition of para-H₂ to unsaturated precursors. The hydrogenation catalyst plays an important role in converting the singlet spin order of para-H₂ into detectable nuclear polarization. In this review, the transition metal-based homogeneous catalysts used in PHIP are discussed with a focus on in vivo imaging applications.

1. Introduction

Parahydrogen-induced polarization (PHIP) is an NMR signal amplification technique that involves the conversion of the singlet spin state of parahydrogen (para-H₂) to ¹H polarization, which can then be transferred to coupled heteronuclear spins.^[1] Along with dissolution dynamic nuclear polarization (DNP), it is one of the most promising methods to generate highly hyperpolarized ¹³C-labeled molecules for biomedical imaging applications.^[2] Although PHIP is not as general as DNP and produces somewhat lower ¹³C NMR signal enhancements, it offers several technical and logistic advantages making it an attractive alternative to DNP. Hyperpolarized samples can be produced very rapidly (in seconds or minutes, rather than hours). Expensive free radical polarizing agents are not required. Unlike DNP, PHIP is capable of large-scale production of hyperpolarized samples in a continuous manner.^[3] PHIP polarizers are not as complex as DNP polarizers and are designed to be portable so that they can be conveniently moved to the vicinity of MR scanners. PHIP has two main branches.^{[2c, 4]} Hydrogenative PHIP involves the irreversible addition of para-H₂ to unsaturated precursors in the presence of a homogeneous or heterogeneous catalyst.^[5] SABRE (Signal Amplification by Reversible Exchange) is a non-hydrogenative PHIP method in which polarization transfer from para-H₂ to the substrate takes place on an iridium catalyst in reversible exchange processes and chemical modification of the substrate does not occur.^[6] In this review, we limit ourselves to the discussion of homogeneous catalysts used in hydrogenative PHIP with a focus on biomedical applications^[4c]. The important and rapidly evolving fields of heterogeneous^{[5, 7]} and metal free catalyst^{[5, 8]} development for hydrogenative PHIP will not be covered here. Catalysts for SABRE^{[5, 9]} will also not be discussed in detail, as we do not have expertise in these areas.

2. The Scope of Hydrogenative PHIP

Since the singlet spin order has zero magnetic moment, it is NMR invisible. To generate NMR detectable magnetization, the symmetry of the two proton spins must be broken. Hydrogenative PHIP is based on the irreversible catalytic addition of para-H₂ to an unsaturated precursor in which the two ¹H spins derived from the para-H₂ molecule become magnetically inequivalent.^[10] The spin order of para-H₂ can be transformed into hyperpolarization of protons or coupled heteronuclei provided that the correlation between the ¹H spins is maintained during the reaction and in the final product, which can be achieved by the concerted pairwise addition of the two protons to the same substrate molecule. There are two main variants. PASADENA (Parahydrogen and Synthesis Allow Dramatically Enhanced Nuclear Alignment) involves the addition of para-H₂ in high field, often inside a superconducting NMR magnet, where the two ¹H spins are weakly coupled.^{[10b, 10c, 11]} In an ALTADENA (Adiabatic Longitudinal Transport After Dissociation Engenders Nuclear Alignment) experiment the addition of para-H₂ is performed in low field (often in the Earth’s magnetic field), so that the two ¹H spins remain strongly coupled during the process.^[12] Figure 1 shows the energy diagram, spin populations and the appearance of the spectra under thermally polarized, PASADENA and ALTADENA conditions.^[13]

Figure 1.

The energy levels, spin populations and transitions for a thermally polarized, PASADENA and ALTADENA spectrum of an AB spin system.^[13–14]

PHIP can generate strongly enhanced ¹H signals, however, for biomedical imaging applications ¹³C-hyperpolarized substrates are more desirable due to the longer T₁ value and larger chemical shift range of ¹³C compared to ¹H. The ¹H spin order originating from the para-H₂ can be converted to ¹³C polarization with varying degree of efficiency by applying spin order transfer (SOT) rf pulses,^[15] magnetic field cycling (MFC),^[16] adiabatic passage through level anticrossing,^[17] or adiabatic radio frequency sweeps at microtesla fields.^[18] Polarization transfer from ¹H to coupled ¹³C also occurs spontaneously albeit with less efficiency, without the use of pulse sequences or MFC via multiplet polarization transfer.^[19] Quantum mechanical treatment of the spin physics involved in PASADENA, ALTADENA, and the various methods of SOT has been reviewed elsewhere.^{[1, 4c, 13, 19a, 20]} Figure 2 shows representative ¹H PASADENA, ALTADENA and ¹³C PHIP-MFC spectra obtained with the same unsaturated precursor (vinyl acetate).^[21]

Figure 2.

Representative PHIP spectra obtained with vinyl acetate at 9.4 T. A) Addition of para-H₂ to vinyl acetate. B) PASADENA experiment at 9.4 T. C) ALTADENA experiment in Earth's magnetic field with subsequent fast transfer to 9.4 T for signal acquisition. D) Polarization transfer to the carboxylate carbon by MFC.^[21] Reproduced from Ref. 21 with permission from the Royal Society of Chemistry.

It must be emphasized that hydrogenative PHIP is a chemistry-based technique and the substrate must have an unsaturated bond for the addition of para-H₂. If we assume that the hydrogen atoms of the para-H₂ will be part of the desired hyperpolarized product, then it is clear that not all biologically interesting compounds would be amenable for direct hydrogenative PHIP. Consequently, there are only a few biologically relevant substrates that have been polarized using direct addition of para-H₂ to an unsaturated precursor. These include fumarate from acetylenedicarboxylate, phospholactate from phosphoenol pyruvate and amino acids from dehydroamino acids.^{[20, 22]} The ingenious side-arm hydrogenation (SAH) approach developed by Reineri and co-workers in 2015 overcame this limitation and significantly widened the range of substrates that can be polarized by PHIP.^[23] The PHIP-SAH method involves the addition of para-H₂ to an unsaturated functional group such as vinyl, allyl or propargyl (as well as various substituted and/or isotopically labeled derivatives) attached to the substrate through a cleavable bond, typically an ester. The PHIP-SAH is usually combined with the two-phase hydrolysis/extraction method to obtain catalyst free hyperpolarized solutions in water. The addition of para-H₂ is performed in an organic solvent and after SOT, the ester function is rapidly cleaved with aqueous NaOH to produce the hyperpolarized substrate in an aqueous solution ready for in vivo use while the catalyst and organic impurities remain in the organic layer. The PHIP-SAH allowed the generation of hyperpolarized [1-¹³C]-labeled pyruvate from unsaturated ester derivatives of pyruvic acid such as propargyl and vinyl pyruvate.^[20] Initially only modest ¹³C polarization levels were reported (2 to 6%).^{[16a, 23]} However, subsequent developmental efforts dedicated to improving the PHIP-SAH process for pyruvate led to the current state-of-the-art ¹³C polarization levels of approximately 15 to 18% at the time of injection.^[24] Figure 3 shows the highly optimized process that has been developed for use in NVision’s commercial PHIP polarizer.^[18b] Given the proven potential of hyperpolarized ¹³C-labeled pyruvate for metabolic imaging, this accomplishment represents an important milestone for the PHIP technology, rendering it a strong competitor to dissolution DNP.^[25] It is worth noting that PHIP-SAH is not the only para-H₂ based method that can polarize pyruvate. SABRE has also reached this level of sophistication and pyruvate can now be polarized to a similar extent using SABRE as demonstrated by the elegant work of Chekmenev, Theis and co-workers.^{[9e, 26]}

Figure 3.

Optimized process for the PHIP-SAH polarization of [1-¹³C]pyruvate to produce injectable hyperpolarized solutions. This process uses [Rh(COD)(DPPB)]BF₄ (Figure 11) as homogeneous catalyst for the hydrogenation.^[18b]

PHIP-polarized pyruvate and fumarate have been tested in cells and in vivo to assess lactate poll size and cellular necrosis, respectively^{.[18b, 20, 24a, 24c, 24d, 24g, 24l, 26b, 27]} Although these experiments did not reveal any new metabolic transformations that had not been previously observed with DNP-polarized substrates, the in vivo studies demonstrated that the ¹³C polarizations achieved via PHIP were sufficiently high to observe the expected metabolites. For example, PHIP-SAH polarized [1-¹³C]pyruvate was successfully used to visualize the Warburg effect (increased lactate level in tumors) via the exchange into preexisting lactate pool. In the experiment shown in Figure 4, the PHIP-SAH of the deuterated vinyl pyruvate was performed in acetone-d₆ followed by hydrolysis with sodium carbonate. Most of the acetone was evaporated under vacuum followed by filtration of the precipitated catalyst. Quite remarkably, the entire operation took only 67 seconds, affording about 12% ¹³C polarization at the time of injection.^[24l]

Figure 4.

In vivo imaging with PIHP-SAH polarized [1-¹³C, 3-²H₃]pyruvate in a tumor bearing mouse.^[24l] The substrate was polarized by PHIP of the deuterated vinyl ester followed by SOT of the ¹H spin order to a ¹³C using the MINERVA pulse sequence at 7.05 T. A) ¹³C MRS of a melanoma tumor-bearing mouse was performed 15 s after polarization transfer and hydrolysis of the product. The spectra were recorded with 18° pulses every 4 s. ¹³C MRSI (magnetic resonance spectroscopic imaging) data were acquired using an EPSI (echo planar spectroscopic imaging) sequence. A) ¹³C MRS of the tumor region. Pyruvate (170.9 ppm), the alanine (176.5 ppm), (pyruvate-hydrate (179.2 ppm) and lactate (183.1 ppm) are clearly visible. B) Anatomical ¹H image with the tumor region is surrounded in red. C) ¹³C pyruvate image and D) lactate image in the tumor region. Reproduced from Ref. 24l, Copyright (2023) with permission from Wiley-VCH

This recent example realistically represents the current state of PHIP technology. Although major metabolites, in this case, lactate and alanine, are easily detectable, observing in vivo [¹³C]bicarbonate production from HP-[1-¹³C]pyruvate via pyruvate dehydrogenase - a crucial metabolic process in oxidative tissues - is considerably more challenging with current ¹³C polarizations achievable via PHIP.^[28] To date, no human studies have been reported with PHIP-polarized substrates.

Next, we briefly discuss two interesting applications of PHIP, which have the potential to make an impact on HP-MR imaging. RASER (radio amplification by stimulated emission is an analogous phenomenon to LASER and MASER in the radiofrequency range.^[29] It involves the spontaneous, stimulated emission of coherent radiofrequency waves from nuclear spin systems. RASER is based on the nonlinear interaction of the spins system with the resonant circuit (receiver coil) and requires nuclear spin population inversion, which can produce large emissive signals and a resonant circuit (coil) in which the rate of radiation dumping is higher than the transverse (T*₂) relaxation rate. Radiation damping occurs when the precessing transverse magnetization induces an electromagnetic field in the receiver coil, which rotates the magnetization vector towards the z-axis (B₀ magnetic field).^[30] This effect depends on the quality factor (Q) and the filling factor of the coil as well as the nuclear polarization in the sample. In conventional NMR experiments, the effect of radiation damping is negligible. However, generating a large emissive signal is necessary for stimulating the spontaneous emission of coherent radiofrequency bursts, which can be achieved through hyperpolarization techniques such as spin exchange optical pumping, DNP and PHIP.^[31] The advantage of PHIP is its ability to generate nuclear polarization in a continuous or semi-continuous manner by supplying reactants to the system, thereby sustaining the RASER effect indefinitely (Figure 5).^[32]

Figure 5.

An example of a ¹H PHIP RASER experiment at 14.1 T. A) RASER was induced by bubbling para-H₂ into a solution of methyl propiolate and a Rh-catalyst in acetone-d₆. B) Spontaneous emission of NMR signal started after about a minute induction period during which the hyperpolarized product accumulated and lasted until the para-H₂ supply was turned off.^[32] Reproduced from Ref. 32, Copyright (2020), with permission from Wiley-VCH.

RASER has been observed with a variety of nuclei including ³He, ¹²⁷Xe, ¹H, ¹³C and ¹⁵N and ¹⁷O.^{[31c, 33]} From biomedical imaging perspective, RASER MRI is feasible, although it faces some technical challenges. Notably, the detection coil (resonant circuit) must have a quality factor (Q) that is at least two or three orders of magnitude higher than that of conventional transmit/receive coils. However, remarkably, in RASER MRI the signal appears spontaneously out of nuclear spin noise without the need of B₁ rf excitations. A 2D image can be generated via projection reconstruction.^[34]

Nuclear singlet spin states are composed of coupled ½ spins.^[35] The singlet state can have long lifetime especially when the spins are chemically and magnetically equivalent, as the two ¹H spins in para-H₂ or nearly equivalent, as the ¹³C spins in [1,2-¹³C₂]pyruvate in Earth’s magnetic field.^[36] Because of the singlet symmetry, the nuclear singlet states are immune to the conventional dipole-dipole relaxation mechanism and often have much longer lifetimes than the nuclear longitudinal relaxation time.^[37] This characteristic allows them to store nuclear polarization for extended periods.^[38] However, it is important to note that the singlet state has zero magnetic moment and is therefore NMR silent. The symmetry of the singlet state can be broken through a chemical transformation into a product in which the two protons are no longer in an equivalent environment. Near equivalency can also be disrupted by increasing the magnetic field. When the singlet state is composed of magnetically inequivalent spins (e. g. in dimethyl fumarate), the singlet state can be converted to observable hyperpolarized magnetization via a singlet to triplet transition, which is facilitated by level anticrossing that occurs when the magnetic field is lowered.^[39] The difference in the coupling constants of magnetically inequivalent ¹H spins can also be used to convert singlet spin order to nuclear magnetization using singlet to magnetization (S2M) pulse sequences or adiabatic ramping of an rf-field.^[40] Singlet states can be populated by manipulating longitudinal magnetization with pulse sequences or by adiabatic transport to low fields.^{[36, 38, 40a, 40b, 41]} However, since the two proton spins of para-H2 are already in a singlet state, addition of para-H₂ to an unsaturated precursor that produces a product where the two hydrogens derived from para-H₂ remain in chemically and preferably, magnetically equivalent or near equivalent environment is a very convenient approach to generate molecules with highly populated nuclear singlet states.^{[39, 42]} As shown in Figure 6, addition of para-H₂ to bis(methyl-d₃) acetylenedicarboxylate in the presence of a Rh-catalyst produces bis(methyl-d₃) maleate, in which the two protons are chemically and magnetically equivalent (the scalar couplings to the deuteriums are negligible). The symmetry can be broken by a chemical reaction with thiols in a Michael addition reaction to produce enhanced ¹H signals.^[43]

Figure 6.

Addition of para-H₂ to bis(methyl-d₃) acetylenedicarboxylate produced bis(methyl-d₃) maleate, in which the two proton spins derived from para-H₂ maintain the nuclear singlet state. When the chemical equivalence of the two protons were broken in the reaction with thiols, strong PHIP signals (red) were observed with Only Para-H₂ Spectroscopy (OPSY)^[44] under PASADENA conditions (11.7 T).^[43] Reproduced from Ref. 43, Copyright (2014), with permission from Wiley-VCH.

Note that the catalyst used in this process yields the syn (cis) addition product, maleate, which is not a naturally occurring metabolite in mammals.

In another example, acetylenedicaboxylate was hydrogenated with a ruthenium piano stool catalyst, which preferentially produces the anti (trans) product, in this case, fumarate. Fumarate is an important TCA cycle intermediate and the reversible hydration of fumarate to malate is catalyzed by the enzyme fumarase. When the fumarate carrying the ¹H-spin order was exposed to fumarase, the chemical equivalency of the protons was broken by the enzymatic conversion to malate (Figure 7).^[45]

Figure 7.

Addition of para-H₂ to acetylenedicarboxylate catalyzed by RuCp*(CH₃CN)₃]PF₆ produced fumarate, in which the two proton spins derived from para-H₂ maintained the nuclear singlet state. When the chemical equivalence of the two protons were broken in the fumarase catalyzed conversion to malate, strong PHIP signals (orange) were observed under PASADENA conditions (11.7 T).^[45] Reproduced from Ref. 45, Copyright (2021), with permission from Wiley-VCH.

3. Homogeneous Catalysts for Hydrogenative PHIP

The most straightforward method to achieve the pairwise addition of para-H₂ to unsaturated bonds is to use a homogeneous transition metal catalyst such as a rhodium, ruthenium or iridium complex. Heterogeneous catalysts offer poor pairwise selectivity,^{[7a, 7c]} although significant progress has been achieved recently (over 10% pairwise addition) with nanoparticle-based catalysts.^[46]

3.1. Rhodium and Iridium Complexes with Monodentate Ligands as Homogeneous Catalysts for Hydrogenation

Chloridotris(triphenylphosphine)rhodium(I), [RhCl(PPh₃)₃], (Wilkinson’s catalyst), was the first homogeneous catalyst that could hydrogenate unsaturated compounds with reasonably high rate at room temperature and 1 bar H₂ pressure (Figure 8).^[47] [RhCl(PPh₃)₃] is a d⁸ complex with a distorted square planar structure. In the presence of H₂ it is rapidly converted to cis-[H₂RhCl(PPh₃)₃] via oxidative addition^[48] of a H₂ molecule. One of the phosphine ligands in the resulting complex is labile due to the kinetic trans effect^[49] of the hydride and it is likely in exchange with the pentacoordinate [H₂RhCl(PPh₃)₂] and the solvated [H₂RhCl(PPh₃)₂(Solvent)] complexes.

Figure 8.

Historically important catalysts used in PHIP experiments.

These species can rapidly form an 18-electron octahedral dihydride alkene complex. The π-coordination of the alkene cis to the hydride facilitates the intramolecular insertion of the alkene into the rhodium–hydride bond, which is the rate-determining step of the catalytic cycle. The rapid reductive elimination of alkane from the [RhH(alkyl)Cl(PPh₃)₂] intermediate regenerates [RhCl(PPh₃)₂(Solvent)] and completes the cycle. This mechanism, shown in Figure 9, is known as the hydride route as it initially involves the oxidative addition of a H₂ molecule followed by the coordination of the unsaturated substrate.^[50]

Figure 9.

Catalytic hydrogenation with monodentate phosphine complexes of Rh(I) proceeds via the hydride route. The first step is the dihydride formation via oxidative addition. S = Solvent.^[50]

Since the two hydrogens are chemically inequivalent in the dihydride intermediate ([H₂RhCl(PPh₃)₃]) formed via the oxidative addition of para-H₂, strongly enhanced antiphase signals could be observed in the hydride region when para-H₂ was bubbled through a solution of Wilkinson’s complex in high field (PASADENA conditions).^[10c] This catalyst also works for hydrogenative PHIP. In fact, it was used in the groundbreaking papers by Bowers and Weitekamp, in which the PHIP effect was first reported.^{[10c, 12a]} However, Wilkinson’s catalyst yielded ¹H polarization levels that were significantly lower than those obtained with rhodium bisphosphine catalysts largely because this complex catalyzed the para-H₂ to ortho-H₂ conversion concomitantly with the addition of para-H₂.^[51] Improved versions of Wilkinson’s catalyst in which dienes, most commonly norbornadiene (NBD), or 1,5-cyclooctadiene (COD) are substituted for the chloride (Shrock-Osborn catalysts, Figure 8) are cationic and exhibit higher reactivity as there is no need for phosphine dissociation.^{[50, 52]} Early hydrogenation experiments with iridium complexes in coordinating solvents were discouraging. Crabtree et al. later realized that in non-coordinating solvents certain cationic Ir(I) complexes such as [Ir(COD)PCy₃Py]PF₆, (Cy = cyclohexyl and Py = pyridine) (Crabtree’s catalyst) (Figure 8), showed very high activity in hydrogenating sterically hindered alkenes.^[53] PHIP effects have been observed with Ir-complexes of monodentate phosphines such as Vaska’s complex (Figure 8).^{[10c, 54]} However, from PHIP perspective, the significance of the iridium based catalysts was their ability to facilitate polarization transfer from proton spins originating from para-H₂ to coordinated substrates without chemical modification (hydrogenation) of the substrate. Further research culminated in the development of iridium N-heterocyclic carbene–phosphine complexes, which proved to be extremely efficient SABRE catalysts due to their optimal ligand exchange kinetics.^{[9b, 9c, 9k]} Interestingly, these iridium-carbene complexes are capable of catalyzing both SABRE and hydrogenative PHIP with a substrate molecule that contains both N-donor atoms suitable for reversible exchange and unsaturated bonds available for the addition of para-H₂.^[55]

3.2. Rhodium Bisphosphine Complexes

The breakthrough for homogeneous catalytic hydrogenation came from the studies of cationic rhodium complexes formed with chiral chelating bisphosphine ligands.^[56] Bisphosphine complexes behave very differently from cationic monodentate phosphine complexes.^[51a] Oxidative addition of H₂ to the monodentate derivatives involves the concerted addition of a H₂ molecule via a sigma-complex resulting in a cis arrangement of the two hydride ligands. Thus, [L₂Rh(PPh₃)₂]⁺ type complexes preferably form a dihydride in which neither H is trans to a phosphine. However, in Rh-complexes with chelating bisphosphine ligands, the ligand enforces the cis binding of the two P atoms and one of the hydrides will necessarily be trans to one of the phosphorus atoms in the dihydrides. Due to the strong trans effect of H, this Rh-P bond undergoes lengthening (weakening), and the octahedral geometry of the complex is distorted, significantly enhancing the reactivity of the dihydride complex.^[57] Consequently, catalytic hydrogenation with rhodium bisphosphine complexes follows a distinctly different mechanism from that shown in Figure 9. This mechanism, known as the unsaturate route, involves the initial coordination of the unsaturated substrate followed by the oxidative addition of hydrogen (Figure 10).^[58] Due to the high reactivity of the dihydride, the migratory insertion step occurs very rapidly, and therefore, the transient dihydride alkene complexes can only be detected in rare cases by PHIP-NMR.^{[57, 59]} The hydrogenation may follow slightly different mechanism depending on the substrate and the catalyst. It is likely that for some catalysts, the dihydride intermediate does not even form but the hydrogen binds side-on forming a short lived η²-H₂ complex^[60] in which the transfer of one hydrogen to the substrate occurs very rapidly to give the monohydride, which then undergoes reductive elimination.^[61]

Figure 10.

Hydrogenation with cationic rhodium(I) bisphosphine complexes generally proceeds via the unsaturate route, in which the first step after the activation of the catalyst is the coordination of the unsaturated substrate rather than the oxidative addition of a H₂ molecule. The example shown here is the hydrogenation of a dehydro amino acid derivative. Note that the substrate has functional groups (amide) that can also coordinate to the catalyst.^[58]

The formation of solvent-complexes, rather than hydrides, upon hydrogenation of the diene precatalysts, the tendency to form dimers and trimers and the unusually strong binding of normally weakly coordinating ligands such as alkenes and arenes are also in agreement with the unsaturate mechanism.^[58b] The hydrogenation of unsaturated bonds in the presence of homogeneous (as well as heterogeneous) transition metal catalysts generally proceeds with frontier orbital control (HOMO-LUMO interactions) and the stereochemistry is determined by the suprafacial addition of H₂.^[62] This affords almost exclusively the syn product and in the case of alkyne hydrogenation, it results in the formation of Z-alkenes.

3.3. Rhodium Bisphosphine Complexes in PHIP

An extraordinarily large number of bisphosphine complexes have been reported in the literature for homogeneous hydrogenation,^[63] and although rhodium bisphosphine catalysts have been used in PHIP experiments since the early years, only very few have been tested in PHIP (Figure 11). Most PHIP experiments performed in organic solvents use the DPPB complex of Rh(I), likely because this complex has high activity, especially towards alkynes, and therefore, it is well suited for PHIP-SAH, which usually employs a propargyl ester sidearm.

Figure 11.

Examples of cationic bisphosphine Rh(I) catalysts used in PHIP. Ligand abbreviations: DPPE = bis(diphenylphosphino)ethane; DPPP = bis(diphenylphosphino)propane; DPPB = bis(diphenylphosphino)butane; BINAP = 2,2'-bis(diphenylphosphino)-1,1'-binaphthyl), DCPB = 1,4-bis(dicyclohexylphosphino)butane; DCPF = 1,1’-bis(cyclohexylphosphino)ferrocene; NBD = norbornadiene, COD = 1,5-cyclooctadiene.

However, [Rh(COD)DPPB]BF₄ reversibly binds H₂, which promotes para–ortho conversion.^{[51b, 64]} This undesirable process can be eliminated by switching to a different bisphosphine ligand. For example, [Rh(COD)BINAP]BF₄ does not catalyze the para–ortho conversion.^[64] In a recent study, Sando et al. tested a series of rhodium bisphosphine complexes as potential catalysts for PHIP using vinyl acetate and normal hydrogen.^[65] The length and rigidity of the linker between the two phosphorus atoms as well as the electron donating properties of the substituents attached to the P atoms were varied and the conversion of vinyl acetate, the activation rate and the turnover frequency (TOF) for each catalyst were determined. The results revealed that the electron donating properties of the phosphine substituents had the greatest influence and bisphosphine ligands with the strongly electron donating cyclohexyl substituents showed the highest conversion and activity. Overall, the 1,4-bis(dicyclohexylphosphino)butane (DCPB) and the 1,1’-bis(dicyclohexylphosphino)ferrocene (DCPF) complexes were the best performers. They exhibited about twice as high conversion, 4 times higher activation rate and about twice as high TOF than the commonly used catalyst [Rh(NBD)(DPPB)]BF₄. This is in agreement with earlier observations made with non-chelating phosphines that Rh-complexes of phosphines bearing electron donating substituents were the most efficient catalysts. ^[50]

3.4. Singlet–Triplet Mixing

The conversion of the NMR invisible proton singlet state (S) originating from the para-H₂ molecule to detectable triplet (T) state in the product (S-T mixing) is a prerequisite for the PHIP phenomenon.^[10b] However, S-T mixing can also occur in intermediates of the catalytic hydrogenation cycle in which the two hydrogen atoms are chemically and/or magnetically inequivalent positions (Figure 12).^[66]

Figure 12.

DFT calculations on the course of the hydrogenation of acetylenedicarboxylic acid with [Rh(COD)DPPB]BF₄ in acetone suggest the formation of a key monohydride intermediate in which one hydrogen is already transferred to the substrate. This hydrogen is in β-agostic interaction with the rhodium (S = acetone). The two protons derived from para-H₂ form an AA’XX’Y spin system with the ³¹P and ¹⁰³Rh nuclei (AA’ are the agostic and hydride proton, respectively, X and X’ are the two ³¹P nuclei and Y is the spin-1/2 ¹⁰³Rh). Singlet–triplet mixing gives rise to rapidly decaying ¹H signal.^[61a]

The S-T mixing leads to the overpopulation of the central T₀ state and although the singlet state is immune to dipolar relaxation effects, the hyperpolarized T₀ state will decay rapidly via T₁ relaxation. As rhodium-dihydride species have rather short ¹H T₁ values,^[67] S-T mixing in catalytic intermediates can cause significant polarization loss and therefore, it is highly undesirable.^{[2c, 68]} The T₁ relaxation rates and lifetimes of the catalyst-bound hydrogenation intermediates will determine how much polarization is lost during the catalytic cycle. As shown by Bargon and coworkers, efficient mixing of the S and T₀ states occurs when the two coordinated hydrogen atoms in the intermediate are in different chemical environment and the lifetime of the dihydride intermediate is long compared to the time scale defined by the chemical shift difference (in Hz) of the two ¹H spins.^[69] The degree of the S-T mixing occurring on the catalyst will influence the polarization pattern of the product and can be used to obtain information about the transient intermediates of the hydrogenation. Full treatment of the evolution of the para-H₂ density operator during the hydrogenation process has been developed by Bargon and coworkers.^[70] The S-T mixing in catalytic intermediates can be minimized by selecting a catalyst that allows the coordination of the two hydrogens in an equivalent (or near equivalent) chemical and magnetic environment. However, this may be difficult, if not impossible, to achieve due to the stereochemistry of the octahedral complexes. Therefore, it is a more reasonable approach to optimize the kinetics to reduce the residence time of the protons spins in a chemically/magnetically inequivalent environment.^{[14, 68c, 69–71]} Slow hydrogenation kinetics also leads to incomplete conversion (hydrogenation) of the unsaturated precursor.^[72] Thus, choosing catalysts that operate via the unsaturate mechanism is essential.^{[61a, 69]}

S-T mixing and polarization loss due to relaxation of the T₀ state can also be inhibited by applying spin-locking (a resonant continuous wave B₁ field or composite pulse that rotates the spins around the spin-lock B₁ pulse in the rotating frame thereby bringing the S and T states into resonance).^[73]

Several diamagnetic transition metal complexes have been reported to catalyze the ortho-H₂ to para-H₂ conversion.^{[51b, 74]} Reversible exchange of para-H₂ with a transition metal catalyst can give rise to a strongly enhanced, distorted ortho-H₂ signal displaying an antiphase character known as partially negative line shape (PNL).^{[68a, 75]} This phenomenon has been attributed to the conversion of singlet spin order of para-H₂ to polarized ortho-H₂ with a selectively overpopulated T₀ state in a chemical exchange process occurring on the catalyst. This process can take place when the catalyst is exposed to para-H₂ in the absence of a substrate or during PHIP when a second para-H₂ molecule replaces the substrate on the catalyst. It has been observed with Vaska’s complex^[76], [Rh(COD)DPPB]BF₄^[68a] a trinuclear Ru-cluster^[77] as well as iridium catalysts used for SABRE.^[75] Interestingly, polarized ortho-H₂ with an overpopulated T₀ state is not expected to have an NMR signal because the T₀ → T₊ and T₀ → T₋ transitions cancel out each other. However, the exchange of the T₀-polarized ortho-H₂ with the transient catalyst-dihydride complex during the signal acquisition gives rise to two lines having slightly different chemical shift and opposing phase. The partial negative line shape vanishes after the thermal equilibrium between para-H₂ and ortho-H₂ is established. When a selective narrow-band B₁ field with is applied at different frequencies, analogously to chemical exchange saturation transfer, the magnitude of the PNL effect decreases at frequencies corresponding to the hydride resonances of the transient complex [L_nMH₂]. This potentially allows the indirect but rather sensitive detection of intermediary hydride complexes.^[68a] Of note, paramagnetic transition metal complexes that can catalyze the para/ortho conversion have been reported to produce a negative ortho-H₂ signal (not PNL) in strong B₀ field, which has been attributed to polarization transfer from the unpaired electrons to coupled nuclear spins.^[78]

3.5. PHIP with Water-soluble Catalysts

The hyperpolarized agent must eventually be in aqueous solution for biological applications. As we have seen, this can be achieved via the biphasic approach, in which the PHIP is performed in an organic solvent and the hyperpolarized product is extracted into water. This is usually combined with the hydrolysis of a hyperpolarized intermediate, generally an ester, to produce the injectable agent. The advantage of the two-phase method is that most of the water insoluble catalyst remains in the organic layer and hydrogenation reactions are usually more rapid in organic solvents than in water. Another option is to perform the addition of para-H₂ in aqueous solutions using water-soluble hydrogenation catalysts. However, although there are several water-soluble hydrogenation catalysts, the turnover rate of these is generally lower than those achieved with catalysts developed for organic phase hydrogenations.^[79] In addition, since the catalyst is not removed via phase separation, an efficient method is needed to remove the relatively large amount of water-soluble catalyst before in vivo use. Despite these potential drawbacks, several attempts have been made to perform PHIP in aqueous media with water-soluble catalysts (Figure 14). The development of water-soluble phosphine-based transition metal catalysts is a very active research area.^[80] These catalysts are most commonly employed in aqueous–organic biphasic media for conventional organic synthesis where the catalyst is in the aqueous layer and the substrate is in the organic phase. This setup allows the easy separation of the product and the recovery of the catalyst. Most water-soluble phosphines are sulfonated derivatives of previously developed ligands although other hydrophilic groups have also been used to improve the solubility in water. Phosphine ligands are frequently sulfonated on the aromatic rings but alkylsulfonic acid derivatives are also known. Aqueous PHIP experiments almost exclusively have been performed with the Rh(I) complex of the commercially available water-soluble bisphosphine ligand, 1,4-bis[(phenyl-3-propanesulfonate) phosphine]butane disodium salt (Figure 14). The rhodium catalyst can easily be prepared by mixing bis(norbornadiene) rhodium(I) tetrafluoroborate with the ligand in methanol or water-acetone.^[81] Representative examples include the hydrogenation of phosphoenol pyruvate to phospholactate, vinyl pyruvate to ethyl pyruvate, fumarate to succinate, hydroxyethyl acrylate to hydroxyethyl propionate, dehydro amino acids to amino acids and unsaturated amino acid esters to amino acid esters to amino acid esters.^{[68b, 79, 81a, 81b, 82]} Incomplete hydrogenation of hydroxyethyl acrylate (80 mM) with this catalyst (4 mM) was reported with 8 s reaction time at 15 psi of para-H₂.^[72b] A structurally very closely related catalyst, the Rh(I) complex of 1,4-bis[(phenyl-3-propanesulfonate) phosphine]propane disodium salt (Figure 14), was reported to yield up to 50% ¹³C polarization with 2-hydroxyethyl acrylate in 4 s hydrogenation time.^[72a] Rhodium complexes of bis(diphenylphosphino)butane ligands sulfonated on the aromatic rings (Figure 14) have been used in kinetic studies with para-H₂ to investigate the mechanism of hydrogenation of dimethyl maleate in water.^[83] A mechanism for the hydrogenation following the unsaturate route was proposed. The Rh(I) complex of DPPBTS was previously reported to be an efficient catalyst for the hydrogenation of acetamidoacrylic acid (N-acetyldehydro alanine) in water.^[84]

Figure 14.

Rhodium(I) bisphosphine catalysts developed for PHIP in aqueous media. DPPBTS = tetrasulfonated bis(diphenylphosphino)butane; DAPBTS = tetrasulfonated bis(dianisylphosphino)butane; Cp* = pentamethylcyclopentadienyl; THP = tris(hydroxymethyl phosphine.

Tris(hydroxymethyl)phosphine (Figure 14) is a non-sulfonated water-soluble phosphine with a unique coordination chemistry.^[85] The precatalyst [Rh(I)(NBD)(THP)₂]BF₄ was prepared by adding a solution of [Rh(I)(NBD)₂]BF₄ to the phosphine solution in water.^[86] PHIP experiments with hydroxyethyl [1-¹³C, 2,3-²H₃]acrylate afforded 12% ¹³C polarization, which is lower than the polarization achieved with the sulfonated bisphosphine catalyst, reflecting the slower kinetics of the monodentate tris(hydroxymethyl)phosphine derivative. This is not surprising as in general, complexes of chelating bisphosphine ligands are more active than those of monodentate ligands.

Several water-soluble Rh(I)- and Ir(I)-complexes with N-heterocyclic carbene (NHC) and/or sulfonated tertiary phosphine ligands were prepared and were tested in the PHIP of propargyl alcohol and allyl alcohol in aqueous media.^[87] However, these complexes displayed slower hydrogenation kinetics than rhodium bisphosphine complexes as only about 40–50% conversion of the unsaturated alcohols was achieved in 30 s hydrogenation time.

The addition of para-H₂ to acetylenedicarboxylate to produce fumarate is performed in water but it represents a special case as it requires the trans addition of para-H₂ and is discussed in the following subchapter.

3.6. Trans Addition of Para-H₂ and the Hyperpolarization of Fumarate.

Unlike the above-mentioned rhodium catalysts, ruthenium piano stool complexes such as [Cp*Ru(L)₃]⁺ (Cp* = pentamethylcyclopentadienyl) (Figure 14) have the unusual property of being able to reduce internal alkynes to the E-alkenes with excellent stereoselectivity. The pentamethylcyclopentadiene ring forms the top of the piano stool structure while the legs consist of three weakly coordinating monodentate ligands, such as MeCN or a diene and a monodentate ligand such as chloride. Bargon and co-workers reported that the cationic sorbic acid derivative catalyzed the rapid trans hydrogenation of several internal alkynes such as 2-pentyne and 1-phenyl-1-propyne under 1 bar of H₂ atmosphere in deuterated methanol. ¹H PHIP NMR studies revealed that the hydrogenation did not involve Z to E (cis to trans) isomerization but proceeded via direct trans addition of H₂ in a pairwise manner.^[88] The reaction failed with terminal alkynes due to the formation of unreactive ruthenium-vinylidene complexes. Initially, the unexpected formation of E-alkenes was explained by a reaction mechanism involving a binuclear ruthenium complex. Later ¹H PHIP NMR studies and DFT calculations by Furstner et al. proved that the actual mechanism proceeds via Ru-carbene intermediates as outlined in Figure 15.^{[62, 89]}

Figure 15.

The catalytic cycle for the trans addition of H₂ in the presence of Ru-piano stool complexes proceeds via Ru-carbene intermediates. The formation of fully hydrogenated alkane occurs through the geminal hydrogenation pathway. X denotes a weakly coordinating ligand.^{[62, 89]}

The significance of the piano stool ruthenium catalysts for PHIP is that they allow the generation of HP-¹³C-labeled fumarate via the trans addition of para-H₂ to acetylenedicarboxylic acid.^{[27, 90]}

HP-¹³C labeled fumarate is a useful probe with a potential for human applications. Fumarate is a TCA cycle intermediate, and ¹³C-hyperpolarized fumarate labeled at C1, or C1 and C4 produces HP-¹³C-malate via fumarase. However, fumarate is a dianion and its uptake by intact cells is slow on the timescale provided by the lifetime of HP-¹³C. In necrotic tissues, fumarate has easier access to fumarase and the appearance of HP-malate signal is therefore an indicator of cellular necrosis associated with acute kidney injury, tumor necrosis, and myocardial infarction.^[91] The unsaturated precursor to fumarate is acetylenedicarboxylate and the PHIP is performed with the commercially available [Cp*Ru(NCMe)₃]PF₆ catalyst. This complex is soluble in water although its solubility is not as high as that of the sulfonated phosphine derivatives. [Cp*Ru(NCMe)₃]PF₆ alone produced a hydrogenation product in which the fumarate to maleate ratio was 3 to 1. This was a problem due to the toxicity of maleate. It was realized however, that addition of Na₂SO₃ improved this ratio to 500 to 1.^[90c] In addition, it also significantly speeded up the reaction (Figure 16).^[90c] The mechanism by which Na₂SO₃ alters the activity and selectivity of the catalyst is not known, however, Ru(II) is known to form complexes with sulfite by coordinating to the S-atom and the ligand has a strong trans-effect.^[92] It should be noted that small amounts (about 10%) of succinate are also produced via the gem-hydrogenation pathway. The formation of succinate is not a concern because unlike maleate, it is not toxic.

Figure 16.

The influence of the sodium sulfite concentration on the formation of fumarate (top) and succinate (bottom) at pH 7.8 and 88 °C as measured by ¹H NMR after 5 s bubbling of H₂. The concentrations of acetylenedicarboxylate and catalyst were set at 250 and 7 mM, respectively.^[90c] Reproduced from Ref. 91c with permission from the Royal Society of Chemistry.

The reaction conditions (pH, temperature, substrate catalyst and Na₂SO₃ concentrations) were optimized in experiments performed with normal hydrogen. The kinetic curves showing the fumarate and succinate concentrations vs time were recorded at different sodium sulfite concentrations (Figure 16).^[90c] From these data, it was concluded that the optimal substrate, catalyst and Na₂SO₃ concentrations were 250 mM, 7 mM and 250 mM, respectively, at the optimal temperature of 88 °C and pH of 7.8.

It was shown previously that gem-hydrogenation proceeds via a Ru-carbene intermediate and hydrogenation of fumarate to succinate does not occur.^{[62, 90a, 93]} A very interesting side-product was also identified when the addition of para-H₂ was performed in the absence of Na₂SO₃ in D₂O. In addition to fumarate, maleate and succinate, another compound was identified that was formed in small amount by the [3+2] cycloaddition of the ruthenium-carbene intermediate with a second acetylenedicarboxylate in the geminal hydrogenation pathway (Figure 17). The two hydrogens of the para-H₂ molecule end up in the same CH₂ group, however, since the symmetry of the singlet spin order is broken by the presence of the chiral center of the five-membered ring, these protons become hyperpolarized (0.9% polarization) and give rise to PHIP-enhanced signals.^[73a]

Figure 17.

Addition of para-H₂ to acetylenedicarboxylate in D₂O using [Cp*Ru(NCMe)₃]PF₆ catalyst in the abcence of sodium sulfite. The main product is fumarate. Major side-products include maleate and succinate. A small amount of [3+2] cycloaddition product sodium 4-(carboxylatomethyl)-5-oxo-2,5-dihydrofuran-2,3-dicarboxylate-2-d is also formed.^[73a]

3.7. Other Metal Complexes

PHIP has been observed with various complexes of other metals including Pd, Pt, Ir, Fe and Co.^{[10a, 94]} In addition to the mononuclear Rh(I)-bisphosphine complexes, di and trinuclear Rh(I) complexes such as [Rh₂H₂(CO)₂(DPPM)₂] [DPPM = bis(diphenylphosphino)methane] and [Rh₃Cl₂H₂(CO)₂(DPPE)₂]⁺ also gave rise to PHIP effect with phenylacetylene.^{[10a, 95]} Iron and cobalt complexes of the monoanionic bis(carbene) ligand bis(mesityl-benzimidazol-2-ylidene)phenyl (Figure 18) generated modest ¹H and ¹³C signal enhancements of ethyl propionate after the addition of para-H₂ to ethyl acrylate. Although only modest polarization levels were achieved, further tuning of the ligand structure may lead to significant improvements. These metals are more easily available than the members of the platinum group.^[96]

Figure 18.

Iron and cobalt complexes used in PHIP experiments.^{[96a, 96b]}

3.8. PHIP Enhanced NMR Spectroscopy as a Tool to Study Reaction Kinetics and Mechanism.

PHIP-enhanced NMR, owing to its improved sensitivity, enables the real-time detection of unstable intermediates and previously undetectable minor reaction products. This approach can be used to measure both the formation and degradation rates of these intermediates. Furthermore, the PHIP patterns offer a method to discern the structural features of these species. Utilizing para-H₂ in hydrogenation reactions and examining the reaction mixture with ¹H PHIP NMR has proven instrumental in elucidating the mechanism of homogenous catalytic hydrogenations via the detection of intermediates that were previously inaccessible for observation. There are excellent reviews on this topic.^{[4b, 59, 97]}

Here we will only mention one particularly interesting application, namely, the use of PHIP NMR to study the enantioselectivity of asymmetric hydrogenation reactions. The basic idea is that from NMR perspective, HP-¹H and thermally polarized ¹H spins are not identical and during the hydrogenation of certain prochiral substrates in the presence of a chiral catalyst temporary diastereomers are formed in which the steric arrangement of the hyperpolarized and thermally polarized ¹H spins is different (Figure 20). The PHIP NMR patterns of these hyperpolarized temporary diastereomers are different.^[98] This also implies that the PHIP patterns obtained with enantiomeric chiral catalysts should be distinguishable. This method has been used to establish the absolute configuration of the products by comparing the experimentally obtained ¹H PHIP spectra with the simulated PHIP patterns of the possible temporary hyperpolarized diastereoisomers.^[99].

Figure 20.

Chiral catalysts used in stereoselective ¹³C PHIP. DIOP = isopropylidene-2,3-dihydroxy-1,4-bis(diphenylphosphino)butane; DIPAMP = 1,2-bis[(2-methoxyphenyl)(phenylphosphino)]ethane.

3.9. Stereoselective ¹³C Polarization

PHIP of unsaturated prochiral precursors with chiral catalysts (Figure 20) can potentially afford ¹³C-hyperpolarized chiral metabolic substrates in their biologically relevant enantiomeric forms.

Jerschow and co-workers used the Rh(I) complex of (+)-(S,S)-DIOP for the PHIP of N-acetyl dehydro-amino acids to produce hyperpolarized N-acetyl amino acids (Figure 21). The reaction was performed in deuterated methanol under PASADENA conditions followed by polarization transfer to the carboxylate carbon by an INEPT type pulse sequence yielding 1.3% ¹³C polarization. However, only one enantiomer of the catalyst was tested and the enantioselectivity of the para-H₂ addition was not determined.^[100]

Figure 21.

PHIP of deuterated dehydro phenylalanine in the presence of [Rh(COD)(S,S)-DIOP]BF₄.^[100]

Using lactate as a model compound, we have recently shown that by combining chiral catalysts used for conventional homogenous asymmetric hydrogenation with para-H₂, it is possible to enantioselectively hyperpolarize both enantiomers of chiral substrates starting from a common unsaturated prochiral precursor. PHIP experiments were performed in the Earth’s magnetic field (ALTADENA conditions) with both enantiomers of the chiral catalyst Rh(I)-DIPAMP followed by magnetic field cycling (MFC) to transfer polarization to ¹³C. The MFC afforded excellent ¹³C polarization (up to 22%). Chiral HPLC analysis of the hydrogenation products revealed that the stereoselectivity was good (90% L and 10% D, i. e. 80% enantiomeric excess or ee). We coined the term stereoPHIP for this process (Figure 22).^[101]

Figure 22.

Stereoselective ¹³C PHIP of lactate esters. A) Catalytic addition of para-H₂ to O-acetyl ethyl enolpyruvate yields hyperpolarized racemic, L, or D lactate ester depending on the catalyst used. B) A series of ¹³C-NMR spectra at 1 T acquired using 5 ° flip angle and 10 s delay between successive scans of the hyperpolarized L-enantiomer. C) Chiral HPLC traces of authentic samples (black) and hydrogenation products obtained with achiral catalyst (green), (S,S)-enantiomer (blue) and (R,R)-enantiomer (red) of the chiral catalyst Rh-DIPAMP.^[101] Reproduced from Ref. 101, Copyright (2023), with permission from Wiley-VCH.

The chiral catalyst Rh(I)-DIPAMP has been developed based on the studies performed with Rh(I)-DIOP and provides superior stereoselectivity.^[102] Asymmetric hydrogenations with rhodium bisphosphine catalysts proceed by the unsaturate mechanism but to understand the origin of enantioselectivity, the formation of diastereomeric adducts between the chiral rhodium complex and the prochiral alkene must be considered. Assuming that the addition of H₂ is irreversible and occurs to the Rh-coordinated face of the alkene, there are two possible scenarios. The first one, which might seem to be more plausible at first sight is that predominant enantiomer of the final product is derived from the major diastereomeric adduct. The second one is that the enantioselectivity is due to the much higher reactivity of the minor adduct. There is overwhelming kinetic, spectroscopic and computational evidence that the stereochemistry is kinetically controlled, i.e. the minor isomer is sufficiently reactive to determine the chirality of the final product (“anti-lock-and-key” mechanism). The enantioselection likely occurs in the migratory insertion step.^[103]

3.10. Catalyst Toxicity and Removal

Rhodium and ruthenium complexes, like all heavy metal compounds, are toxic. Unfortunately, there are limited toxicological data available for these elements. The LD₅₀ of rhodium trichloride in male Sprague-Dawley rats was 198 mg/kg when administered by a single iv injection^[104] Although no toxicity was observed when male and female rats were treated orally with 5000 mg/kg dose of [ClRh(PPh₃)₃] (Wilkinson’s complex)^[105] preliminary results demonstrated that PHIP hyperpolarized solutions containing the rhodium catalyst had no toxic effects when injected into rats,^[106] this should not be taken as an indication that rhodium complexes could safely be administered by i.v. injection at several g/kg doses. Ruthenium compounds and complexes appear to be somewhat more toxic. Ruthenium trichloride was found to have an LD₅₀ of 108 mg/kg in mice. The LD₅₀ of bis(2,2'-dipyridyl)ruthenium(II) dichloride was reported to be 63 mg/kg in mice and this complex caused convulsions and death within minutes.^[107] However, another paper reported the LD₅₀ of ruthenium(II) bisphosphine complexes to be >2,000 mg/kg (via oral administration) suggesting that the ligand may have a strong influence on the toxicity.^[108] The concentration limit for platinum group metals and their compounds (including rhodium and ruthenium complexes) in drug products is set to 1 μg/g product by the United States Pharmacopeia (USP) when administered parenterally (including intravenous injection).^[109]

Removal of the catalyst is necessary for in vivo applications, in particular, for human studies. When the PHIP is performed using the two-phase method, the bulk of the catalyst and organic impurities remain in the organic layer and the hyperpolarized aqueous solution contains only trace amounts of rhodium.^{[23–24, 110]} The process developed for pyruvate polarization by NVision for the POLARIS commercial polarizer is a modification of the original two-phase method.^[18b] It uses deuterated acetone for the PHIP. As acetone is miscible with water, after the hydrolysis of the hyperpolarized ester with aqueous NaOH, a water immiscible solvent (methyl tert-butyl ether) is added to remove the solvent and catalyst. After the phase separation, the organic solvent concentrations are reduced by passing nitrogen gas through the solution at reduced pressure. This method affords rhodium concentrations less than 8 μg mL⁻¹, which are below the no observed-adverse-effect level (NOAEL) given for rats in the International Council for Harmonisation of Technical Requirements of Pharmaceuticals for Human Use (ICH) guideline.^[18b] The water insoluble catalyst can also be removed by filtration after evaporating the acetone under vacuum.^[24l] Traces of organic solvents from the aqueous solution can also be removed using an adsorbent resin such as TENAX.^[110]

In an interesting variant of biphasic PHIP, an ionic liquid, 1-butyl-3-methylimidazolium tetrafluoroborate, was used to dissolve the catalyst. Neither the substrate, ethyl acrylate, nor the product, ethyl propionate is miscible with the ionic liquid thereby offering a convenient method of separation.^[111]

When the PHIP is performed in water with a water-soluble catalyst, other methods must be developed for catalyst removal. Rhodium belongs to the platinum-group elements (platinum, palladium, rhodium, ruthenium, iridium, and osmium), which all have very similar chemical properties. They form typical soft cations, which have high affinity to sulfur donor atoms. Therefore, the easiest and most efficient way to remove platinum group metal ions from solution is by treatment with thiols or other ligands with sulfur or nitrogen donor atoms immobilized on a solid support.^[112] It was demonstrated that mercaptopropyl silica (QuadraSil MP) rapidly and efficiently removed the iridium catalyst used in SABRE-PHIP experiments in less than 10 s.^[113] [Cp*Ru(NCMe)₃]PF₆ catalyst was successfully removed from PHIP-polarized fumarate by passing the hyperpolarized solution through a column packed with a thiourea-based metal ion scavenger resin (QuadraPure^®TU).^[90d] Other methods involve the deactivation of the catalyst with a strongly binding ligand and remove it using an ion exchanger (DEAE Sephadex A-25).^[114]

A rather interesting method of PHIP-product separation reported by Chekmenev and coworkers involved the removal of the volatile hydrogenated substrate from the aqueous layer in a stream of para-H₂.^[115]

The purification of PHIP polarized fumarate, very ingeniously, takes advantage of the low solubility of fumarate in water at low pH.^[27b] After polarization transfer from ¹H to ¹³C, the hyperpolarized solution was added to concentrated hydrochloric acid and a sonotrode (ultrasonic horn) was used to promote the precipitation of fumaric acid within less than a second. The solid was washed with 1 M HCl and acetone. The hyperpolarized solid was redissolved in warm bicarbonate solution in D₂O to produce a hyperpolarized fumarate solution with a pH between 7 and 8.5. The precipitation and purification were carried out in relatively high magnetic field (500 mT) because low field relaxation effects would have significantly shortened the ¹³C T₁ value below 100 mT. During the precipitation and purification process, the ¹³C polarization decreased from 37% to 13% and the ruthenium content of the resulting solution was 7.9 mg/L. It may seem surprising that the ¹³C polarization is not lost completely in the solid state. However, the dependence of the T₁ relaxation time on the rotational correlation time (τ_R) of the molecule follows a V-shaped curve with a minimum at around 1 nanosecond, corresponding to the τ_R of medium sized molecules and the T₁ is longest when the rotational correlation time is either very short i.e. small molecules in non-viscous solvents or very long, such as in the solid state.^[116]

3.11. Practical Considerations

So far, the most commonly employed bisphosphine rhodium(I) catalyst for hydrogenative PHIP experiments is the commercially available [Rh(diene)(DPPB)]BF₄. This complex extremely rapidly hydrogenates alkynes including propargyl derivatives and therefore, it is the catalyst of choice for PHIP-SAH. For example, the hydrogenation of propargyl pyruvate was complete in 3 s shaking after the frozen sample (0.2 M propargyl pyruvate and 4 mol% catalyst in CHCl₃-EtOH in a pressure NMR tube) was immersed in an 80 °C water bath for 8 s^{[24a, 24b, 110]} In the commercial polarizer, the hydrogenation is performed for 8 s in a heated reactor (60 °C) under a hydrogen pressure of 10 bar (0,22 mM pyruvate ester and 1 mol% catalyst in acetone-d₆) and a hydrogen flow rate of 2 standard L/min.^[18b] The catalyst is somewhat slower with substituted propargyl or alkene derivatives although for most substrates this will not affect the polarization appreciably.^{[24m, 82a] [24i, 117]} As expected, the rate of hydrogenation increases with increasing catalyst concentration, temperature and H₂-pressure, although at high temperatures (above 90 °C) the thermal decomposition of the catalyst may occur.^{[72a, 83] [82e]} The molar ratio of the catalyst varies depending on the substrate reactivity but 1 to 20 mol % is commonly used. Interestingly, PHIP has been performed even with stoichiometric amounts of catalyst, i.e. the unsaturated substrate was added to the catalyst in stoichiometric ratio to form a complex, which was then treated with para-H₂.^[115]

The solvent plays an active role in the catalytic cycle.^[118] The precatalyst is generally a cyclooctadiene or norbornadiene complex of the bisphosphine ligand. As seen in Figure 10, the precatalyst is activated by the reduction of the diene during the induction period, upon which solvent molecules will coordinate to the metal. These will then exchange with the substrate. After the transfer of H-atoms to the substrate, the product may leave the complex via reductive elimination. However, most substrates have functional groups that can coordinate to the metal (e. g. carboxyl), and exchange with a solvent molecule capable of coordinating to the catalyst helps complete the catalytic cycle. Acetone and methanol are both coordinating solvents, and they are widely applied for hydrogenation reactions alone or mixed with a non-coordinating solvent such as chloroform.^[110] Solvent coordination significantly reduces the Gibbs free energy of the transition states in the catalytic cycle^[119] and in the absence of a coordinating solvent, the reaction rate will be very slow.^[110] The solubility of hydrogen in the solvent is also an important factor to consider as low solubility reduces the rate of hydrogenation.^[120]

The simplest way of performing PHIP experiments is to use pressure NMR tubes charged with the solvent, catalyst and the substrate.^{[24f, 101, 110]} The reaction mixture is deoxygenated via freeze-pump-thaw cycles and then pressurized with para-H₂. The hydrogenation is initiated by heating the frozen reaction mixture in hot water and vigorous shaking, generally in the Earth’s magnetic field (ALTADENA conditions). The medium wall 5 mm pressure NMR tubes can accommodate up to about 500 μL reaction mixture with a maximum operating pressure of 150 psi. Heavy wall tubes can withstand higher pressures but hold less volume. This method can be used to perform PHIP on a small scale but it also allows the production of hyperpolarized substrates in sufficient quantities for small animal imaging. Unfortunately, parameters such as the hydrogenation time, para-H₂ pressure, temperature and magnetic field are rather poorly controlled.^[121]

Other experimental setups involve spraying the reaction mixture in a reactor charged with para-H₂^{[81b, 122]} or bubbling para-H₂ through the reaction mixture.^{[82e, 121]} In the SAMBADENA (Synthesis Amid the Magnet Bore, A Dramatically Enhanced Nuclear Alignment) method, the PHIP is performed inside the bore of an MRI scanner.^{[72b, 82e]} While these approaches require custom-built dedicated equipment and lack the simplicity of the pressure NMR tube method, they allow a much better control of the entire PHIP process.

PHIP experiments under PASADENA conditions are usually performed by bubbling para-H₂ gas through a capillary into the reaction mixture in an NMR tube inside the spectroscopy magnet. However, before spectrum acquisition, the flow of para-H₂ must be turned off to allow the solution to settle in the NMR-tube. This disrupts the reaction and lowers the polarization due to the waiting time. It was shown that the use of a gas permeable microporous hollow-fiber membranes (such as the polypropylene Celgard X50 for aqueous media) to deliver the para-H₂ to the reaction mixture can eliminate these problems by providing a continuous supply of para-H₂ without foaming and bubbling. In addition, the large gas–liquid interface provides an efficient gas delivery thereby affording high hydrogenation rates.^[123]

Devices capable of executing the entire PHIP process (hydrogenation, SOT and catalyst removal) to produce injectable hyperpolarized solutions for in vivo applications are referred to as polarizers.^[124] Fully automated PHIP polarizers have been developed with the ultimate goal of clinical use.^{[18b, 122, 125]} These are designed to offer consistent and reproducible polarization along with quality assurance.

PHIP can also be advantageously integrated with microfluidic devices. Microfluidics involves the manipulation of nL to fL amounts of liquids using channels with dimensions of tens to hundreds of micrometers. Potential applications of the technology include, chemical synthesis, biological analysis, optics and information technology and microelectronics.^[126] Microfluidic devices are created using photolithography, originally developed for manufacturing integrated circuits. A lab-on-a-chip is a sophisticated microfluidic device that allows the performance of several laboratory processes such as chemical synthesis, analysis and high-throughput screening on a microscale using a single integrated circuit. All microfluidic systems must be integrated with an analytical method to gain information about the processes occurring on the chip. These analytical methods must have sufficient specificity and sensitivity to reliably detect the molecular species of interest. Most commonly used techniques include electrochemical detection, fluorescence or Raman spectroscopy.^[127] Although NMR offers excellent selectivity, the comparatively low sensitivity of conventional NMR limits its application in microfluidics. However, the integration of microfluidics with HP-NMR technology can overcome this problem.^[127b] PHIP-NMR is especially well suited for this purpose because para-H₂ can be brought into contact with the solution containing the unsaturated substrate and the catalyst through a gas permeable membrane with high enough rate to continuously produce significant amounts of hyperpolarized metabolites in a device that can be placed in the bore of a superconducting spectroscopy magnet. This technology makes it feasible to perform metabolic studies in a highly controlled environment with picomolar sensitivity using a microfluidic device that contains cells or tissue samples (organ-on-a-chip).^[128] Figure 23. shows the design and build of a PHIP@chip microfluidic device and a single-scan steady-state PASADENA spectrum of PHIP polarized allyl acetate obtained during the continuous operation of the PHIP@chip.^[128a] An improved version of this device has recently been used to produce PHIP hyperpolarized [1-¹³C]fumarate with up to 8.5 % ¹³C polarization.^[129]

Figure 23.

The PHIP@chip device. A) Outline of the chip (dimensions are in mm). B) Computer aided design (CAD) rendering of the device. The fluid and the pressurized hydrogen gas are separated by a PDMS layer, which is a barrier to the liquid but permeable to the para-H₂ gas. C) the PHIP reaction and single-scan steady-state spectrum obtained at the optimum flow rate with para-H₂ at 11.7 T.^[128a] Reproduced from Ref. 128a, with permission from The Journal of the American Chemical Society.

Careful cleaning of the NMR tubes and reactors is crucial when working with para-H₂.^{[24a, 24f]} The decomposition of the catalyst deposits a black layer, most likely metallic rhodium, on the glass surface, which can catalyze the conversion of para-H₂ to ortho-H₂ likely via chemical exchange. Cleaning the NMR tubes with hot aqua regia effectively removes this deposit, thereby minimizing experiment failures and enabling significantly higher nuclear polarizations.^{[24a, 90c, 101]}

4. Summary and Outlook

Although PHIP offers several advantages over its main competitor, dissolution DNP, it significantly lags behind in terms of polarization levels. Modern preclinical and clinical polarizers (SpinAligner and SPINlab) can produce over 60% ¹³C polarization albeit using much more expensive instrumentation. Although PHIP can theoretically produce 100% ¹³C polarization for some spins systems, in practice, the reported values rarely exceed 30%. Consider the real-life example of fumarate. In [1-¹³C]fumarate, the two ¹H and the ¹³C spins form an AA′X spin system after the addition of para-H₂ to acetylenedicarboxylate, and theoretically, the ¹H singlet state derived from para-H₂ can be converted into 100% ¹³C polarization.^{[24f, 130]} Experimentally however, only 37 % was realized primarily due to suboptimal catalyst performance.^[27b] Thus, there is room for improvement. PHIP imposes specific demands on hydrogenation catalysts. Given that the main cause of polarization loss is the singlet-triplet mixing occurring on the catalyst during hydrogenation, rapid kinetics are essential to prevent substantial polarization decay. Catalyst performance is determined by a complex interplay of pre-catalyst activation, reaction mechanisms, substrate reactivity, reaction conditions (temperature, pressure), solvent effects and catalyst deactivation.^[131] These parameters (together with the SOT process) must be optimized for each substrate to achieve maximum polarization via PHIP. In the light of the numerous transition metal-based hydrogenation catalysts reported in the literature, it is surprising that most PHIP experiments have been performed with Rh-DPPB derivatives, except for the generation of HP-fumarate from acetylenedicarboxylate. Thus, developing more efficient catalysts will play an important role in the future of PHIP.

Figure 19.

Temporary diastereomers formed by the PHIP of certain prochiral substrates in the presence of chiral catalysts can be distinguished by the ¹H PHIP spectra.^[99]

Biographies

Biographical Sketch. Mai T. Huynh received her Ph.D. in Chemistry from the University of Texas at Dallas. Currently, she is a Postdoctoral Researcher in Zoltan Kovacs’s lab at the UT Southwestern Medical Center. Her research focuses on the synthesis of deuterated, ¹³C and ¹⁵N-labeled molecular probes for dynamic nuclear polarization (DNP) to image oxidative metabolism and to assess various biomarkers of tissue microenvironment (pH, redox, enzyme activity). Recently she developed a parahydrogen-induced polarization (PHIP) method for the stereoselective ¹³C-hypeprolarization of prochiral precursors.

Biographical Sketch. Zoltan Kovacs received his Ph.D. from University of Debrecen, Hungary. As a Postdoctoral Fellow at the University of Texas at Dallas, his research focused on the development of lanthanide-based MRI contrast agents. In 2006, he joined the faculty of the Advanced Imaging Research Center at UT Southwestern Medical Center, Dallas, TX, where his current primary research interests include the synthesis and application of hyperpolarized ¹³C, ¹⁵N, ⁸⁹Y, ^107,109Ag and ⁷⁷Se based magnetic resonance probes, and metal free diamagnetic chemical exchange saturation transfer (diaCEST) agents.