Hyperpolarized water as universal sensitivity booster in biomolecular NMR

Abstract

Nuclear magnetic resonance (NMR) spectroscopy is the only method to access the structural dynamics of biomolecules at high (atomistic) resolution in their native solution state. However, this method's low sensitivity has two important consequences: 1) typically experiments have to be performed at high concentrations that increase sensitivity but are not physiological; 2) signals have to be accumulated over long periods, complicating the determination of interaction kinetics on the order of seconds, and impeding studies of unstable systems. Both limitations are of equal, fundamental relevance: non-native conditions are of limited pharmacological relevance, and the function of proteins, enzymes, and nucleic acids often relies on their interaction kinetics. To overcome these limitations, we have developed applications that involve ‘hyperpolarized water’ to boost signal intensities in NMR of proteins and nucleic acids. The technique includes four stages: (i) preparation of the biomolecule in partially deuterated buffers; (ii) preparation of “hyperpolarized” water featuring enhanced ¹H NMR signals via cryogenic dynamic nuclear polarization; (iii) sudden melting of the cryogenic pellet and dissolution of the protein or nucleic acid in the hyperpolarized water (enabling spontaneous exchanges of protons between water and target); (iv) recording signal-amplified NMR spectra targeting either labile ¹H or neighbouring ¹⁵N/¹³C nuclei in the biomolecule. Water in the ensuing experiments is used as a universal ‘hyperpolarization’ agent, rendering the approach versatile and applicable to any biomolecule possessing labile hydrogens. Thus, questions can be addressed, ranging from protein and RNA folding problems to resolving structure-function relationships of intrinsically disordered proteins to investigating membrane interactions.

Introduction

Nuclear magnetic resonance (NMR) is one of the main methods to characterize structural environments and chemical dynamics in liquids and solids. It is essential in organic chemistry research and in the pharmaceutical industry to elucidate the structures and properties of small molecules; biophysicists use it to understand the structures of proteins and nucleic acids under physiological conditions; biologists, radiologists, clinicians, and psychologists couple NMR with spatial localization techniques to understand and diagnose health features. Despite these widespread applications, NMR is far from realizing its full potential, as its signals stem from only ~0.01% of the sample under investigation. The rest of the sample remains essentially unused due to the very low magnetic energies involved.¹ Nuclear spins at conventional temperatures reach polarization levels of less than 10⁻⁴, causing a majority of them to remain “indifferent” to any NMR experiment, even when placed in very strong magnetic fields.

There are many ongoing efforts towards improving the sensitivity of NMR measurements –including using stronger magnets, better NMR probes, advanced data acquisition schemes and artificial intelligence processing. At present, however, the approaches that provide the best improvement in sensitivity are those based on nuclear hyperpolarization.

Nuclear hyperpolarization can be achieved by a variety of physics-based experiments, including transferring the perfect spin alignment involved in the para-hydrogen isomer of the H₂ molecule to other spins via chemical reactions,^2-7 or the optical pumping of certain photo-active systems.^8-17 However, of all nuclear hyperpolarization methods, arguably none is as general and powerful as dynamic nuclear polarization, DNP. In the liquid state, DNP can increase nuclear polarizations by exploiting the higher spin alignment of unpaired electrons (radicals) that have been comixed with a targeted solution, up to a maximum that corresponds to ca. 50% of the ratio between the electron and nuclear Larmor frequencies –i.e., ~10²−10³. The polarization transfer is driven by the application of microwaves resonant at the electron Larmor frequency. In unison with suitable relaxation processes, such irradiation provides pathways to the so-called Overhauser enhancement.^18-23 The underlying mechanism proceeds efficiently up to fields on the order of ~1 T (¹H Larmor frequencies of ~40 MHz); past this limit, the efficiency of the process and limited microwave penetration impede Overhauser DNP unless very small samples and/or very strong microwave powers are used. Higher polarizations at higher field strengths are obtained in solid-state implementations of DNP carried out under cryogenic conditions, where electrons are nearly 100% polarized.^24-33 Transferring this electron polarization to neighboring nuclei (Figure 1a) and subsequently dissolving the sample translates into a boost in the NMR signal intensity that enables the detection of target molecules at up to 10⁻⁴ times lower concentrations than in conventional room-temperature NMR measurements; or in other words, to achieve single scan detection of a sample that would otherwise require up to 10⁸ signal averages –NMR signal to noise ratio (SNR) scaling with the square root of the number of averaged signals.

Figure 1.

Instrumental components underlying the dDNP methodology. a) An electron spin with a high polarization interacts with nearby nuclear spins (red). Under microwave irradiation, this interaction results in an overall increase in nuclear polarization. The inset shows the spin polarization of the nuclei, which are distributed among their two Zeeman energy levels before and after the DNP process. b) Prototypical dDNP equipment (University of Vienna). The two-magnet setup includes the DNP system that enables microwave irradiation of samples doped with paramagnetic impurities at low temperatures and in high magnetic fields. A pneumatic dissolution system allows one to dissolve and transfer the samples after the DNP build-up to a solution-state NMR spectrometer operating at ambient temperatures (typically between 4 and 50 °C.).

The last decade has witnessed a dramatic increase in the numbers of such cryogenic DNP techniques for chemical and biophysical applications.^34-39 Hyperpolarisation by DNP is achievable at low temperatures but retained by only a few systems and for short lengths of time when porting them to room temperature; therefore, further advances were required for adapting this approach to solution-state NMR and biomedical applications. ^{20,24-33,40-73}

Solution-oriented DNP experiments are at first carried out in the same way as their solid-state counterparts: by comixing a dilute unpaired electron spin –arising, for instance, from the stable nitroxide or trityl radicals– with the sample containing the nuclear spins of interest, and freezing them together in a suitable glassing solvent or medium that ensures their homogeneous dispersion. In such static solid, an electron-to-nuclear polarization transfer can then be carried out in relatively high fields (B_0,DNP ≈ 3-7 T) and low temperatures (T_DNP ≈ 1-4 K), by irradiating the electrons close to their Larmor resonance with moderately-powered (tens of milli-Watt) microwave fields. Before being irradiated under such conditions, the electron spins will be almost entirely polarized due to their high magnetogyric ratio. Microwave irradiation will reduce the electrons’ spin polarization; in the process of returning to equilibrium, the latter will also disturb the spin populations of the nuclei that surround them. The overall ensemble will thus adopt a metastable state that features high nuclear spin order, reaching the polarization levels that are characteristic of electrons. Mechanisms that result in such high nuclear spin alignment usually involve two-spin (“solid effect”), three-spin (“cross effect”), or multi-spin effects (“thermal mixing”).^{25,26,30,74-78} For any of these mechanisms to proceed efficiently, however, the whole operation needs to take place in a cryogenic solid –a condition that is compatible with solid-state NMR observations, but not with the biomolecular solution-phase NMR applications that constitute the focus of this protocol.

The key step that enables using DNP-enhanced nuclear polarization in liquids under ambient conditions, is the sudden melting that occurs upon transport of the sample from its cryogenic hyperpolarization environment to a conventional solution-state NMR setup operating close to physiological temperatures. In such “dissolution DNP” (dDNP) experiments,^31-33,56,57 the signal-enhanced, hyperpolarized, frozen pellet is rapidly dissolved, flushed away from the DNP magnet by a stream of hot solvent, and injected into another magnet for subsequent NMR (or MRI) observation. Figure 1b illustrates this two-magnet, ex-situ principle underlying dDNP-boosted NMR, for one of the ‘hyperpolarizers’ operating at the University of Vienna. The sudden dissolution and sample porting involved in dDNP provide a liquid state, ambient temperature hyperpolarized sample, whose spin alignment will last for a duration governed by the longitudinal relaxation time T₁ of the hyperpolarized nuclear spins. Clearly, the transfer of a sample that has been hyperpolarized under cryogenic conditions for several minutes or even hours into a conventional room-temperature magnet for NMR observation, will be a one-time, single-shot event unsuited for multi-scan signal averaging. However, since the nuclear polarization is increased to ~10⁴ times its thermal equilibrium value, the high sensitivity upon NMR acquisition renders signal averaging largely unnecessary. Based on this principle, dDNP^31-33 has found numerous applications in preclinical^42,44-47,52 and clinical MRI for cancer diagnosis,^53,54 as well as analytical chemical, physical and biological applications.^60,69,79-88 Due to these successes, dDNP systems have recently also been commercialized⁸⁹ and made available to a broad community. Hence, accessibility to DNP facilities becomes more and more common, allowing many researchers to complement existing assays with this rapidly developing technique.

Despite these achievements, dDNP still suffers from certain limitations that prevent it from reaching its full potential in chemical and biochemical applications. One of these limitations relates to transferring the molten “super sample”, from the DNP to the NMR magnets. This transfer needs to proceed within timescales shorter than the nuclear relaxation time T₁, as otherwise, the metastable nuclear state will return to equilibrium canceling the gains provided by DNP. This transfer condition is not onerously taxing for hyperpolarized low-γ spin-1/2 nuclei like ¹³C or ¹⁵N in small metabolites; ¹H and ¹⁹F probes, too, can be efficiently hyperpolarized and transferred in small molecules. However, independently of the targeted nucleus, spin polarization in large biomolecules will always be severely compromised by the short, sub-second T₁ relaxation times of its spins as they transverse regions of low magnetic field –preventing applications of dDNP NMR to proteins, sugars, and nucleic acids. Specific sites in flexible polypeptides might preserve some of their hyperpolarization as they traverse the low-field regions between the DNP and NMR magnets^84,90; however, as predicted by established relaxation models,^91,92 the T₁ relaxation times of more rigid structures become so short to prevent dDNP from providing significant enhancements.

This shortcoming substantially reduces the scope of dDNP, in particular for biomolecular applications. Indeed, NMR is a key technique in structural and molecular biology as the only method that provides access to protein and nucleic acid structural dynamics at atomic-level resolution, in a native-like, solution-state environment. Therefore, a way of combining dDNP with biomolecular NMR, would be highly desirable: protein and nucleic acid NMR signals could then be boosted by orders-of-magnitude, thereby enabling i) real-time NMR studies of molecular processes on sub-second timescales without a need for signal averaging; and ii) sensitive NMR experiments at nanomolar, physiological concentration regimes. Both aspects are of fundamental relevance: the key to the function of proteins and nucleic acids often resides in their interaction kinetics, while biomolecular studies under high concentrations, non-physiological conditions are of limited relevance in the elucidation of these properties.

To fulfill these goals, recent years have witnessed numerous concurrent efforts to devise routes enabling the applicability of dDNP to biomacromolecules.^22,70,93-98 This Protocol describes one such route, which provides a relatively general solution to this problem by exploiting dynamics in labile biomolecular sites –for instance, amide, amine, hydroxyl, or imino sites in nucleic acids and protein sidechains or backbones. We refer to this as the hyperpolarized water or “HyperW” approach.^99-109 It should be noted that this approach aims to enable studies at physiological protein / nucleic acid conditions –but so far in an in-vitro setting. Spectra from hyperpolarized small metabolites are recorded on a regular basis on various platforms: in cell that continuously feed hyperpolarized substrates¹¹⁰ into an NMR tube,¹¹¹ or by injecting hyperpolarized substrates into animals¹¹² and even humans^113,114 for detection by MRI (magnetic resonance imaging). However, applications of HyperW-enhanced biomolecular NMR to study the multitude of intermolecular interactions in the very crowded cellular milieu or in-vivo remains to be implemented.

Development of the method

HyperW exploits water protons that can be efficiently hyperpolarized, to polarizations >10%, by the aforementioned cryogenic DNP procedure. Moreover, this water hyperpolarization, when suitably handled, can be sustained almost in its entirety upon melting and traversing the low intervening fields between the DNP and NMR magnets.^60,100,115 The spontaneous exchanges that such hyperpolarized water protons will then undergo with the labile, solvent-accessible sites of proteins, peptides, or nucleic acids, can enable sensitivity-enhanced NMR of proteins and nucleic acids as long as the measurements are taken within approximately a minute. Spraying the hyperpolarized water into a sample tube containing an awaiting biomolecule can significantly enhance the signal amplitudes of labile protons, as these spontaneously exchange with the numerous water protons. Hence, the possibility arises to record amide/amino/imino-based NMR spectra of proteins and nucleic acids with superior sensitivity while rendering the water hyperpolarization reservoir independent of the target molecule.

Over the past years, procedures have also been implemented^116-118 to deliver aliquots of hyperpolarized water with very high precision into NMR tubes preloaded with biomolecular samples to be analyzed. This led to excellent repeatability and permitted an accurate, bias-free shimming (B₀ homogenization) of the post-injection samples –leading to high-resolution (ppb) line widths and maximum NMR sensitivity as well as resolution. Further, following its sudden injection, the hyperpolarized water will take a relatively long period –up to ~50 seconds¹¹⁹– to relax back to thermal equilibrium. Throughout this period, spontaneous chemical exchanges can keep on replenishing the biomolecule’s nuclear polarizations. This enables the acquisition of multi-scan NMR experiments while enjoying the benefits of an enhanced nuclear hyperpolarization –including multidimensional correlations, which in many cases rely on observing or exciting labile sites. A suite of NMR experiments has been developed over the past years that combine dDNP-boosted signal enhancements with 1D, 2D, and 3D spectroscopy capitalizing on the intrinsically high, atomistic resolution of NMR. The versatility of the HyperW approach is reflected in a wide array of covered target systems: from classically folded, globular proteins to intrinsically disordered proteins, from membranes to RNAs. In addition, by discerning heterogeneities in these sensitivity enhancements, HyperW also provides a window to discriminate the exposure of various sites in the biomolecules to their surrounding solvents.

As a result of the above described features, we believe that HyperW could become a method of choice for specific investigations of RNA, DNA (e.g., primers, fragments, and aptamers), and polypeptides (e.g., globular as well as intrinsically disordered proteins and domains, membrane interacting peptides), particularly as the signal enhancement reaches the “magic” number of 1 000-fold. Being equivalent to the averaging of 10⁶ scans, this is at the edge of what is practically possible with non-hyperpolarized samples. The present Protocol thus describes the implementation of this powerful emergent methodology.

Overview of the procedure

The implementation of the HyperW method for a biomolecular sample occurs in four phases (Figure 2).

Figure 2.

Flow chart of a dDNP-NMR experiment employing water hyperpolarization.

Phase 1 – Preparation of the reagents

In phase 1, the targeted biomolecule – the protein or the nucleic acid – is prepared and placed in the NMR magnet for observation. The specific protocol for preparing the target molecule may involve recombinant expression of a protein in bacterial host cells that have been genetically modified by introducing a plasmid. Alternatively, chemical synthesis of a peptide / nucleic acid is possible. This sample will usually be produced with suitable isotope labeling, often including ¹⁵N and/or ¹³C enrichment (see also Table 1). In the final step, these labels will allow the execution of multidimensional NMR experiments starting from (and sometimes ending in) the labile proton, followed by hyperpolarization transfer to the macromolecule’s backbone ¹⁵N/¹³C-nuclei by heteronuclear spin-spin couplings.¹²⁰ The target may also be (partially) deuterated, on the aliphatic and/or exchangeable amide positions. Deuteration of aliphatic positions reduces intramolecular polarization transfer through spin diffusion, whereas initial deuteration of exchangeable protons increases the signal contrast from the exchange-based hyperpolarization transfer. After production and purification of the protein, a sample is prepared under physiological solution conditions and at a sufficiently high concentration to observe signals after dilution by a few-fold with hyperpolarized water. This sample may contain additional components needed for the desired experiment, such as ligands that interact with the macromolecule to study ligand binding by polarization transfer from water. Denaturing agents such as urea may be added if the goal is to characterize a chemically unfolded protein.

Table 1.

Pulse sequences used for detection in hyperpolarized water experiments. The correlated nuclei, dimensionality, and suggested uses are indicated. Note that the list only refers to experiments used here for the detection of hyperpolarized biomacromolecules. Other pulse sequences for measuring and quantifying NOE effects and relaxation times, as often used in dDNP applications, are not listed.

Pulse sequence	Correlate nuclei	Dimensionality	Use
zg2d	1H	Pseudo 2D	Real-time monitoring
Selective COSY	1H-1H	2D	short peptides
BEST-HMQC using 90° pulses	1H-15N	2D	folded or short proteins and RNAs
BEST-1HN-CON using 90° pulses	13C-15N	2D	Physiological pH, temperatures
BEST-HNCO using 90° pulses	1H-13C-15N	3D	Crowded spectra; long IDPs and RNAs

In a second, independent step, an aliquot containing the water to be hyperpolarized is prepared and placed in the hyperpolarizer. This aliquot contains an organic component, dimethyl sulfoxide or glycerol, to help the water form a homogenous glass upon freezing. A radical such as TEMPOL is also added, to provide unpaired electron spins for the DNP process. Another sample, usually a buffer solution, is prepared for performing the post-polarization melting and dissolution of the hyperpolarized water aliquot. This buffer is prepared in deuterium oxide, which, when diluting the water, reduces the proton density and thus also relaxation losses.

Phase 2 – DNP hyperpolarization

DNP hyperpolarization is performed as part of phase 2 of the protocol, as summarized in Figure 3. This process begins with cooling the variable temperature bore of the DNP instrument to a temperature near 1.2 K. The aliquot containing water is subsequently loaded and vitrified. DNP is initiated by irradiating the aliquot with microwaves at a frequency close to the central electron paramagnetic resonance (EPR) transition of the used radical. The spin polarization of water protons builds up with a time constant typically on the order of tens of minutes. The polarization is then left to build up to the desired level, preferably within ~90% of the possible maximum.

Figure 3.

DNP of water. The workflow begins with the preparation of the aqueous sample that includes glycerol as cryo-protectant and TEMPOL as paramagnetic species. The sample is then inserted into the DNP system, where it is cooled down to temperatures close to 1.2 K. Once cold, slightly off-resonant microwave irradiation leads to a build-up of hyperpolarization. This process typically takes ca. 2 h to reach a steady state.

At the same time, the NMR spectrometer, dissolution, and injection systems are prepared. The NMR probe is pre-tuned, and the spectrometer is pre-shimmed using a sample with the same solvent composition and volume as the final HyperW sample. Subsequently, the tuning sample is removed, and the sample containing the biomacromolecule to be hyperpolarized loaded into the NMR spectrometer, awaiting in-situ dilution with the hyperpolarized water. Alternatively, the macromolecule may be pre-loaded in a sample injector, for injection into the NMR spectrometer concomitantly with the hyperpolarized water. Then, the dissolution solvent is loaded in the dissolution system of the DNP polarizer to dissolve and flush the hyperpolarized water aliquot from the polarizer to the NMR magnet. Finally, a sample injector may need to be enabled to trigger automatically after the dissolution of the DNP polarized sample.

Phase 3 – Transfer and injection

In phase 3, the dissolution solvent, such as pure D₂O or a buffer in D₂O, is heated to ca. 180 °C within the dissolution system of the DNP polarizer. The solvent reaches a pressure of several bar. It is then suddenly released, melting the hyperpolarized water pellet and pushing it, aided by high-pressure Helium gas, through a tubing towards the NMR spectrometer (Figure 4). The sample injection proceeds through a flow path containing one or more switched valves and may involve metering the liquid volume using an injection loop.^117,118 The dissolved sample can optionally be treated using a filter device or a solvent extraction procedure, to remove the radical.^100,119 Upon reaching the NMR sample tube waiting inside the NMR magnet, the hyperpolarized water mixes with the sample containing the macromolecule. Polarization transfer between the water and the labile protons in the target molecule will occur spontaneously through proton exchanges; further polarization may be transferred to non-labile protons in the biomolecule via inter- or intramolecular cross-relaxation (NOE).

Figure 4.

The dissolution-DNP experiment. 1: The water sample is hyperpolarized up to 2 h at temperatures close to 1.2 K. 2: The hyperpolarized water is dissolved with a burst of superheated, deuterated solvent and pneumatically transferred to the NMR spectrometer for detection. 3: The hyperpolarized water is mixed in-situ within in the NMR tube with the target solution that contains the protein or nucleic acid of interest. Amide proton exchange and NOE effects lead to a transfer of polarization from the medium to the target molecule. 4: An NMR experiment with boosted sensitivity is recorded within the lifetime of the water hyperpolarization. The resulting multidimensional spectrum features substantially improved signal amplitudes.

Phase 4 – Data acquisition

At this time, in phase 4, NMR data acquisition is triggered. The NMR experiment to be performed can be chosen from a suite of pulse sequences that have been developed for use with hyperpolarized water. These pulse sequences may include features such as a pre-scan for determining the amount of water polarization, followed by one or a series of scans that aim to observe the transferred hyperpolarization at the target molecule. These scans may include selective pulses for observation of target spins without perturbing water hyperpolarization. The experiment may also include fast multi-dimensional NMR spectroscopy methods, which can be used to resolve amide and/or aliphatic moieties of the macromolecule. The experiments can additionally aim at the measurement of relaxation rates through the observation of transient signal evolution.

Planning the experiment

Careful method development and optimization prior to the main experiment is essential, because many target molecules might not (or only partially) be recoverable after dilution with hyperpolarized water and mixing with the cryo-protectants and polarization agents.

It may be useful to do some of this development and validation with a similar molecule that is cheaper or more abundant.

For the chosen NMR pulse sequences, we advise that all parameters are evaluated and optimised. These include choosing the most relevant sequence for the experiment, optimising the pulse bandwidth, checking the spectral resolution, etc. Additional details can be found in the Equipment Setup section. The method development and optimisation are typically performed using protein and nucleic acid concentrations that are suitable for conventional NMR. This approach is preferable because the low concentrations used in HyperW experiments might necessitate signal averaging periods that are unfeasibly long for pulse sequence calibration (see the Equipment Setup section for details).

Another part of the setup that needs special attention is the mechanism for transfer and injection of the hyperpolarised solution. Because the signal acquisition needs to start almost immediately on injection, aspects of the NMR preparation need to be done in advance. Detailed advice on setting up the equipment and timing of these checks are also in the Equipment Setup.

Applications of the method

Applications of HyperW include the characterization of protein or nucleic acid structure and dynamics, molecular interactions, and ligand binding. Typically, amides (proteins) and imino resonances (nucleic acids) are targeted. The characterization of hydroxide sites in these molecules as well as in carbohydrates can also be envisioned, albeit spectral proximity between the resonances of these sites and that of water complicates the NMR experiment. Depending on the desired information, either a single strongly signal-enhanced ¹H NMR spectrum can be recorded from the biomolecules (e.g., for experiments at low concentrations), or a series of spectra can be recorded; the latter can allow one to monitor the above-described kinetic processes in real-time, with the benefits of enhanced sensitivity. In addition, the method can be combined with rapid two-dimensional NMR spectroscopy (for specific example applications, see the Anticipated Results section). Dynamic processes that occur on faster time scales can be determined via the intermediary of measuring line shapes or relaxation rates, accessing the microsecond to millisecond or picosecond to nanosecond regimes, respectively.

Specific benefits of the method, which enable these applications, are two-fold. Firstly, the polarization transfer from hyperpolarized water enhances the sensitivity of macromolecular NMR spectra acquired in either a single or a small number of successive scans. These spectra can be obtained within seconds, i.e., in a shorter time than with signal averaging methods in conventional NMR. Secondly, the polarization transfer is selective towards solvent-exposed positions, depending on molecular structures and interactions. In addition, it is modulated by molecular dynamics, and the corresponding macromolecular properties can thus be elucidated.^101,105

Refolding a protein or interactions of unfolded protein domains with binding partners, too, can be investigated with a sub-second to seconds time resolution.^102,107,121 Hyperpolarized water can further be used to investigate protein-ligand interactions.¹²² Information about kinetic rates, site-specific interactions, and possibly existing intermediate structures can be obtained. Through mathematical modeling of polarization transfer, which occurs via proton exchange and cross-relaxation, even faster molecular processes can be accessed.^102,103 A more detailed illustration of possible applications can, again, be found in the ‘Anticipated Results’ section.

Comparison with other methods

Conventional biomolecular NMR includes techniques to determine solvent accessibility and protein hydration through NOE and exchange spectroscopy experiments.^123,124 Acquisition of 2D NMR spectra under proton-deuterium exchange conditions permits the determination of protection factors, informing about solvent exposure and hydrogen bonding.^125,126 Experiments employing water magnetization for the determination of protein-ligand interactions have been further developed.¹²⁷ Fast processes can be investigated by NMR spectroscopy employing stopped-flow¹²⁸ and pressure-jump¹²⁹ devices. These methods are conceptually related to NMR spectroscopy enhanced by hyperpolarized water. In contrast to these techniques, hyperpolarized water provides a deviation from the equilibrium distribution of nuclear spin states by about three orders of magnitude. As a result, the measurement of time-dependent buildup curves is facilitated, and NMR signals are detectable within a shorter timeframe. Indeed, hyperpolarized water enables the characterization of non-equilibrium processes in the sub-second regime.^107,121 In contrast, most conventional NMR experiments access equilibrium states of macromolecules or slow kinetic processes on the order of minutes. The use of hyperpolarized water shortens the time scale of these experiments while retaining the ability to resolve independent sites of a macromolecule.

Other hyperpolarization methods provide information on protein structure, dynamics, and surface accessibility. For example, photo-chemically induced dynamic nuclear polarization (photo-CIDNP)¹³⁰ relies on a radical pair mechanism involving a photosensitizer and a histidine, tryptophan, or tyrosine amino acid side-chain. With photo-CIDNP, surface accessible side-chains in proteins can be hyperpolarized, and this method has further been used for the characterization of protein folding. This technique provides substantially simplified NMR spectra, showing hyperpolarized signals primarily from the mentioned amino acid residues.

Alternatively, Overhauser effect DNP (ODNP) can be used to hyperpolarize nuclear spins directly in solution. ODNP relies on cross-relaxation between an unpaired electron spin and a nuclear spin; although molecules can be directly polarized in solution by ODNP, the experiment most easily works at low magnetic fields, or at moderate fields with very small samples so that microwave heating of the liquid solvent is reduced. Samples can also be polarized by ODNP at low fields and then rapidly shuttled to a higher field¹³¹, where high-resolution spectra with enhanced signals can be measured. Alternatively, ODNP polarization efficiency can be directly measured to obtain information on molecular properties as the polarization efficiency depends on proximity between a radical spin probe and target spin, as well as on molecular dynamics. In combination with site-specific spin labeling of a protein or nucleic acid^132-135, ODNP has been used to probe hydration¹³³, non-native protein conformations¹³⁴, and protein aggregation using low-field NMR spectroscopy¹³⁵. These measurements are highly sensitive to dynamics in the hydration layer of the protein. However, with the site-specific spin labeling, they do not simultaneously resolve multiple sites of the molecule. ODNP can also substantially enhance signals of the low-γ nucleus ¹³C at higher magnetic field strengths, such as 3T.¹³⁶ Based on their mechanisms and resulting hyperpolarization patterns, the above hyperpolarization methods are complementary to the dDNP method described in this article. Indeed, dDNP provides very high signal enhancements of water as a solvent, which can be transferred to a target biomolecule to measure fast multi-dimensional spectra at high magnetic fields. Hence the spectroscopic capabilities of this method appear most closely related to high-resolution biomolecular NMR.

Other methods, including stopped-flow or temperature-jump optical spectroscopy¹³⁷, time-resolved X-ray methods¹³⁸, and related approaches, are also widely used for the characterization of transient macromolecular processes.^139,140 While these techniques offer a high time resolution extending into the milliseconds regime, they do not provide the same level of structural information as can be determined from NMR-based correlation measurements.

Expertise needed to implement the protocol

The HyperW experiment requires access to a dDNP facility, in addition to the availability of an NMR spectrometer. The availability of dDNP capabilities is less widespread than other sensitivity enhancement techniques such as cryoprobes or solid-state DNP-enhanced NMR, even if it provides a much higher enhancement of the signal-to-noise ratio per scan. Nevertheless, with the recent introduction of the SpinAligner DNP polarizer (Polarize Co., Denmark), and the pending availability of a new DNP polarizer by Bruker, the number of available instruments is expected to increase substantially.

The dDNP experiment can be performed by a skilled PhD student or postdoctoral researcher with only minimal guidance and additional help. The steps required to load the dDNP instrument and dissolve the hyperpolarized sample should be reviewed prior to performing the experiments. Written instructions are provided with commercial instruments. Additional training in the use of an add-on sample injector may be required. This can be combined with the training of a user for operating a conventional NMR spectrometer. It should be noted that the newer models of dDNP systems are relatively easy to use. In order to set up the NMR experiments, practical familiarity with liquid-state NMR spectroscopy and basic knowledge of biomolecular NMR techniques are needed. Users should familiarize themselves with the safe handling of cryogenically cooled materials and the heated and pressurized solvent in the dissolution system of the instrument.

Sample preparation requires basic laboratory skills, including accurate weighing and pipetting techniques. If protein or nucleic acid samples are prepared in the laboratory, expertise in biochemistry and molecular biology is further required. This expertise should include the ability to clone inserts, plate bacteria and select colonies, follow or develop protocols for culturing cells and recombinant protein expression, as well as protein purification through chromatography and other suitable methods.

Limitations

At the time of writing, research groups should be prepared to invest ca. 300 to 600 kEUR to connect a new dDNP system to an existing NMR spectrometer and exploit the Protocol presented herein. If a cryogen consumption-free system or a helium recovery apparatus is used, the running costs for cryogenic liquids are negligible; otherwise, cryogenic consumption of several kEUR per year should also be factored in.

Experimental limitations of the method are primarily related to the achievable signal enhancement, to the time scale of real-time NMR problems that can be investigated, and to irreversible changes in the sample composition.

Achievable signal enhancement is less than expected

HyperW’s main aim is to enhance the sensitivity of the biomolecular NMR signals. Theoretically, this enhancement can be very large: protons can be hyperpolarized by factors of ~5 000x over their room-temperature thermal counterparts under cryogenic DNP conditions. This factor is further magnified by the much higher concentration of the hyperpolarized water compared to the biomolecule in the sample following the dissolution and in-situ mixing. Thus, sensitivity enhancements of over 10 000-fold should, in principle, be possible.

In practice, however, multiple sources of loss exist that decrease this maximum enhancement by factors of ~100-fold. These include:

• the substantial dilution that the hyperpolarized water undergoes upon being flushed out of its cryogenic environment by a buffer or D₂O,

• the additional dilution that it experiences by virtue of the solution that is already waiting in the NMR spectrometer with the sample to be hyperpolarized,

• the polarization losses that occur as the sample transfers from one magnet to the other, and

• the enhanced relaxation that water undergoes inside the NMR magnet by virtue of the co-dissolved biomolecule and/or the polarizing radical.

Many of these limitations can be ameliorated by technical means; implementing these are topics of active research in the field.

The effect of the water peak on the spectrum

In addition, there is a limit to the kind of systems and processes that can be probed using HyperW methods. The protons to be targeted need to exchange in a facile manner with the water. Faster exchanges mean a quicker repolarization –but this holds up to a limit, given by the chemical shift separation between the labile sites and the water peak: exchange rates that exceed this separation (in sec⁻¹) will blur the resonances being targeted, melding them into the overwhelming ¹H₂O resonance and preventing their observation and/or the exploitation of their polarization.

Likewise, ¹H resonances like those of sugar hydroxyl groups that fall too close to the water line will usually be invisible in the presence of the blinding hyperpolarized solvent resonance; in this sense, operation at the highest possible field facilitates the execution of the residue-specific excitations underlying HyperW NMR.

Limited time due to loss of spin polarization

In terms of the rates of the processes that can be studied, events can be resolved if occurring within a time range between tens of milliseconds and tens of seconds. At the higher end of this interval, signals are reduced because of the loss of spin polarization due to spin-lattice (T₁) relaxation. This signal reduction occurs because the dDNP experiment produces the observed non-equilibrium spin polarization ex-situ before delivery of the sample to the NMR instrument. This feature of the experiment, on the one hand, enables obtaining NMR spectra more rapidly than by conventional NMR, thus extending the lower limit of the accessible time.

On the other hand, unlike conventional NMR, spin polarization is not renewed between scans. Enhanced signals can therefore be observed for a time of approximately two to five times the T₁ of water. While protons in water normally exhibit a relaxation time of 2-3 s, this time can be extended to >10 s by dissolution in a deuterated buffer, using radical extraction, or, if desired and feasible, by raising the temperature.^100,119

At the lower end of the accessible time interval, the sampling time for acquiring the NMR signal is not long enough to result in high spectral resolutions after Fourier transform. For example, an acquisition time of 100 ms results in a minimal line width of approximately 10 Hz. Further, a mixing dead time on the order of 100 ms is necessary for the dDNP experiment.¹¹⁸ However, similar to conventional NMR experiments, dynamic processes on the picosecond to millisecond time scale can be probed indirectly by measuring relaxation parameters and analyzing line shapes.⁸⁸ Though, as in conventional NMR, this indirect measurement does not lead to the observation of the corresponding process in real-time but instead considers the system to be under equilibrium with respect to the time scale of data acquisition.

Limited capacity for phase cycling

Another aspect concerns phase cycling. Strong signal enhancements by DNP largely obviate the need for signal averaging, which enables the detection of the above-mentioned fast processes. On the other hand, polarization by dDNP is not renewable short of polarizing a fresh sample. This property impacts the ability to perform multiple scans that would typically be used in conventional biomolecular NMR experiments for phase cycling. With dDNP hyperpolarization transferring from water, two-step phase cycles are still practical. In general, however, coherence selection using pulsed-field gradients is preferred.

The hyperpolarized water is diluted

The dissolution process causes a dilution of the hyperpolarized water sample. The dilution offers the opportunity to reduce the proton density and prolong relaxation times through mixing with a deuterated solvent. Upon injection into the NMR spectrometer and mixing with the target protein, the water is further diluted. At the same time the concentration of the target molecule is reduced. This second dilution is typically 2- to 3-fold, depending on the experimental design and can impose a limit to the upper concentration targeted in these studies, as aggregation or precipitation problems might constrain the pre-injection concentration limits. Another issue related to the sudden dilution of the target solution is a possible denaturation of the macromolecule. While denaturation due to heat or pressure shock has not been observed so far, some proteins and nucleic acids adopt different conformations at different concentrations.¹⁴¹

A possible solution to reduce the dilution factor is the use of a flow cell for mixing the hyperpolarized water with the target.^110,142

The polarizing materials for DNP are also added to the target

The DNP process requires the presence of a free radical such as 4-hydroxy-2,2,6,6-tetramethylpiperidin-1-oxyl (TEMPOL) and a glassing agent such as glycerol or dimethyl sulfoxide (DMSO). These compounds often remain in the sample and have the potential of interacting with target molecules. Fortunately, the radicals used for DNP are not reactive, and the radical concentration is reduced to the micromolar range due to dilution in the dissolution process.

A final important limitation that needs to be considered is the “single-shot” nature of dissolution DNP experiments. Dilution, mixing with polarization agents or cryo-protectants, and unfeasible re-purification are all factors that might prevent sample recycling. In favorable cases, the protein or nucleic acid can be recovered by means of dialysis and/or HPLC purification, but the efficiencies are variable and in many cases the analyte cannot be recovered. Hence, concerning the experimental design, particular attention should be paid to correct spectrometer and DNP setups to avoid dDNP experiments at sub-optimal signal enhancements.

Heterogeneous Signal Enhancements

Kaderavek et al.¹⁰⁹ have shown, using ubiquitin as a model system, that signal amplitudes in a HyperW experiment (at a low degree of solvent protonation of 2%) can exceed those of a conventional NMR experiment in a 90% protonated buffer by factors >10. This was the case for residues that exchange hyperpolarization with solvent efficiently on the time scale of the pulse sequence’s recycling delay. However, for many residues that do not efficiently interact with the solvent, the signals in HyperW are often either weaker in comparison to their conventionally detected counterparts, or even entirely undetectable. This circumstance leads to a residue-selective signal enhancement, which can be beneficial as the resulting spectral sparseness reduces signal overlap and improves resolution, but which can also be a limitation since regions that are, for example, buried in the hydrophobic cores of a protein cannot be accessed using HyperW. Hence, another parallel to the classic ‘protection factor’ (vide supra) often used in biochemical studies can be drawn: if the protection factor of a residue is too high, it remains invisible to HyperW. The selective polarization transfer to exchangeable positions therefore confines the strongest signal enhancements to the surface of folded proteins.

While all residues are accessible in denatured or intrinsically disordered proteins, the selective observation of surface-exposed residues could be a limitation or an advantage depending on the question asked. On the one hand, reducing the complexity of the NMR spectra increases the ability to resolve specific surface sites. On the other hand, the hydrophobic cores of proteins are not directly accessible through water hyperpolarization, despite their importance as reactive centers. Similar considerations hold for the imino resonances of nucleic acids, which will exchange more readily with the hyperpolarized solvent if they are located in bulge regions but will be more protected if involved in stable base-paired structures.

Materials

Reagents

• D₂O (D ≥ 98%, Cortecnet, cat. no. CD5251P1000 or Eurisotop, ref. DLM-4-99-PK) (CAS: 7789-20-0)

• Glycerol (C3H8O3, Sigma-Aldrich, cat. no. G5516)

• (Optional) Glycerol-d8 (D ≥ 98%, Sigma-Aldrich, cat. no. 447498)

  • Optional substitution: dimethylsulfoxide-d6 (CAS: 2206-27-1)
• 4-hydroxy-2,2,6,6-tetramethylpiperidin-1-oxyl (TEMPOL). TEMPOL is harmful when swallowed and causes severe eye irritation (97%, Sigma-Aldrich, cat. no. 176141) (CAS: 2226-96-2) [CAUTION]

  • Optional substitution: 4-amino-2,2,6,6-tetramethylpiperidine-1-oxyl (4-amino-TEMPO). 4-amino-TEMPO is harmful when swallowed and causes severe eye irritation (97%, Sigma-Aldrich, cat. no. 163945) (CAS: 14691-88-4) [CAUTION]
• disodium hydrogen phosphate (Na₂HPO₄). (Sigma-Aldrich, cat. no. 1.06580) (CAS: 7558-79-4) or dipotassium hydrogen phosphate.. (CAS: 7758-11-4) [CAUTION causes serious eye damage]

  • Optional substitution: dipotassium deuterium phosphate. (CAS: 22387-03-7) [CAUTION causes serious eye damage]
• sodium dihydrogen phosphate (NaH₂PO₄, Sigma-Aldrich, cat. no. 1.06370) (CAS: 7558-80-7) or potassium dihydrogen phosphate (CAS: 7778-77-0)

  • Optional substitution: potassium dideuterium phosphate (CAS: 13761-79-0)

• sodium chloride (NaCl, Sigma-Aldrich, cat. no. 1.06404) (CAS: 7647-14-5)

• (Optional) Urea-d₄ (CAS: 1433-11-0) as denaturant

  • (Optional) 4,4-dimethyl-4-silapentane-1-sulfonate sodium salt (DSS) for chemical shift referencing. Causes skin irritation, causes serious eye irritation, may cause respiratory irritation (CAS: 2039-96-5) [CAUTION]

• (Optional) Helmanex III (Sigma-Aldrich, cat. no. Z805939). Causes serious eye damage, may cause respiratory irritation. [CAUTION]

• Target protein or target RNA

  • (Optional) ¹⁵N-enriched target protein
  • (Optional) ¹⁵N- and ¹³C-enriched target protein
  • (Optional) ¹⁵N-enriched target RNA. RNA is very sensitive to contamination with RNases and degradation [CAUTION]

Equipment

• (Optional) Sterile working environment. This is only required when working with RNAs or unstable protein samples.

• Centrifuge microtubes (1.5 ml)

• Vortex shaker (e.g., IKA Vortex 3)

• Micropipettes (10 μL, 100 μL and 500 μL)

• Pipette tips

• 5 mL pipette

• Benchtop Centrifuge for 15 and 50 mL tubes (e.g., Thermo Scientific^™ Multifuge X1R)

• Centrifugal filter tubes (15 and 50 mL)

• Lyophilizer (e.g., Christ Alpha 2-4 LDplus)

• 5 mm outer diameter NMR tube (e.g., Wilmad WG-BTNMR-8-25)

• (Optional) high pressure NMR-tube (e.g., Wilmad WILM507-PV-7)

• DNP sample cup or sample holder. This item is implemented differently for the various available dissolution-DNP systems.

• Dissolution-DNP system for ¹H-hyperpolarization. E.g., an Oxford HyperSense system. Other commercial systems (e.g., the Polarize SpinAligner or the GE SpinLab) or custom-made setups based on flow-through cryostats are also suitable.

  • (Optional alternative) A closed-cycle cryostat/magnet combination (e.g., Cryogenic LTD) with a custom dissolution-DNP insert.⁸⁹
  • (Optional alternative) A flow-through cryostat/magnet combination based on, e.g., a Bruker wide-bore magnet with a dissolution-DNP insert.¹⁴³

• Dissolution and transfer system including a 3.2 mm outer diameter (OD), 1.6 mm inner diameter (ID) of 1.6 mm OD, 0.8 mm ID PTFE capillary to guide the liquid from the DNP system to the NMR spectrometer. Home-built systems were used here that transfer the liquid from a HyperSense system^117,118,144 or home-built DNP systems to an NMR spectrometer for detection¹⁴⁵

• (Optional) A magnetic tunnel surrounding the PTFE capillary between DNP system and NMR spectrometer.¹¹⁵ This could also be achieved by using a solenoid coil to cover the PTFE capillary.¹⁴⁶

• (Optional) A BNC cable to connect the NMR spectrometer with the dissolution system for automated detection.

• (Optional) A high-pressure transfer system to accelerate the liquid transfer.¹¹⁸ This system is a user set-up device to inject hyperpolarized samples into a standard NMR tube or into an NMR flow cell using a high-pressure gas or high-pressure liquid pumps.¹¹⁶

• An adaptor to connect the PTFE capillary to the sample tube used with the NMR spectrometer.

• (Optional) A high-pressure NMR tube that enables use of counter pressure upon injection.

• (Optional) Digital thermometer with thermocouple. The thermometer is used for measuring the sample temperature when setting up experiments.

Reagent setup

Nitroxide stock solutions

Prepare aqueous TEMPOL stock solution at 175 mM concentration. This solution can be stored at −20° C for months.

(Optional) Use 4-amino-TEMPO instead of TEMPOL to reduce foaming upon mixing of the hyperpolarized water with the target solution at later stages of the workflow. (See Troubleshooting for step 1,2.).

D₂O stocks

Always use fresh batches of D₂O. Longer air exposure will lead to H/D exchange and reduce the experimental performance. Store D₂O under N₂ or Ar gas. <CRITICAL>

Glycerol stocks

Always use fresh batches of glycerol. Longer air exposure will lead to dilution due to its hygroscopic nature. Store under N₂ or Ar gas.

Sample buffers

For a performant sample buffer, dissolve 2.2 g disodium hydrogen phosphate, 0.2 g sodium dihydrogen phosphate, and 8.5 g sodium chloride in 1 L D₂O to obtain a partially deuterated phosphate-buffered solution (dPBS) at a pH of 7.4.

(Optional) Different target molecules might require different buffers that resemble their respective native environments. Generally, all common buffer recipes are feasible. However, D₂O should be used instead of H₂O for preparation.

DNP samples

Dissolve 10% v/v of the 175 mM nitroxide stock solution in H₂O. Add 15% v/v glycerol to the obtained radical solution for a nitroxide concentration of 15 mM in a 0.85:0.15 v/v mixture of H₂O and glycerol.

(Optional) Prepare a solution of H₂O/DMSO-d6 (v/v 1:1) mixture containing 15 mM TEMPOL.

Protein samples

The target protein needs to be purchased or expressed and purified according to the envisaged NMR experiment. These are the methods that we have used in our lab:

• For ¹H-¹⁵N correlation spectroscopy (e.g., HMQC) uniformly ¹⁵N-enriched proteins were produced following well-established procedures.^106,109,147

• For ¹³C-¹⁵N correlation spectroscopy (e.g., ¹H^N-CON) and three-dimensional ¹H-¹³C-¹⁵N correlation spectroscopy (e.g., HNCO), uniformly ¹⁵N- and ¹³C-enriched proteins were produced.¹⁰⁵

• For ¹H-¹H correlation spectroscopy (e.g., COSY), samples were synthesized or purchased at natural isotope abundance.¹⁰⁴

Protein target solutions

The protein sample needs to be dissolved in deuterated PBS. To achieve this, exchange the purification buffer using, e.g., a centrifugal filter with an appropriate molecular weight cut-off. Here, the final protein concentrations were typically 500 μM. Generally, there is no upper concentration limit related to the HyperW technique. However, concentrations as low as 10 μM are feasible for acquiring 1D spectra.

(Optional) If the protein can be reconstituted after drying, it can be lyophilized following expression and purification and later dissolved in deuterated PBS at the desired concentration.

(Optional) Add DSS for chemical shift referencing.

RNA samples

The target RNA needs to be purchased or synthesized according to the envisaged NMR experiment. Here for ¹H-¹⁵N correlation spectroscopy (e.g., HMQC focused on the imino region), uniformly ¹⁵N-enriched RNAs were produced following well-established procedures.¹⁰⁷

RNA target solutions

The RNA sample (optionally isotope-labeled) needs to be dissolved in deuterated PBS, after lyophilization. Here, the final RNA concentration of the target solution was typically 600 μM. Again no upper concentration limits apply due to HyperW approach. However, as for protein NMR, lower concentrations of 10 μM are feasible for acquiring 1D spectra.

Reference samples

If the target molecule degrades quickly or is available in limited amounts, it is helpful to use a reference sample to perform the initial method optimisation.

For protein work, a 1 mM solution of isotope-enriched (¹⁵N for HMQC, ¹³C and ¹⁵N for CON or HNCO detection) ubiquitin in 10% v/v deuterated PBS at pH 6.5 can be used. An important consideration is the calibration of the proton pulse selectivity. Since water suppression techniques cannot be used, the water signal should not be excited to avoid radiation damping and signal distortions during the NMR detection phase of a HyperW experiment. Figure S4 in the supplementary material shows the influence of well-calibrated vs. miscalibrated BEST excitation on the spectra.

Note that changes in protein concentrations between calibration and HyperW experiments do not affect pulse lengths and widths as much as variations in buffer salts and sample volumes. The pulse parameters can therefore be optimized using more concentrated samples than in the main experiment. Alternatively, if the signal intensities would be high enough, calibration can be done with a sample recovered from an earlier HyperW experiment, providing the same conditions for setup and experiment. Using a recovered sample has the advantage of most closely matching the experimental conditions.

Equipment setup

A liquid-state NMR system for detection.

9.4 T - 18.8 T narrow- or wide-bore NMR spectrometers were used for the work described in this protocol. The use of unshielded magnets might be beneficial, as fringe fields help preserve the hyperpolarization while transiting between magnets. Other commercial or custom-made setups are also suitable if they have the following features:

• 5 mm NMR probe with channels for ¹H and ¹³C or ¹⁵N probe compatible with the detection spectrometer. A broadband or triple resonance probe can be used. A liquid state cryoprobe is optional.

• Pulsed field gradient(s). A gradient along the z-axis is sufficient for coherence selection, while triple axis gradients are preferred.

• Temperature control. The temperature of the injected sample should be equal to the temperature setting of the probe.

• NMR pulse sequences implemented on the spectrometer.

While not essential, it is useful to have second receiver channel for simultaneous acquisition of water signals and target signals from a different nucleus. Using a cryogenically cooled probe improves signal-to-noise ratios upon NMR detection.

Pulse sequences for NMR-detection

Well-designed NMR pulse sequences are key to detecting signal-amplified NMR spectra of the proteins or RNAs under investigation upon dissolution in hyperpolarized water. The pulse sequences must be prepared in view of the unconventionally strong water magnetization in such experiments.

We suggest evaluating all parameters relevant for NMR detection for each target molecule with NMR samples in thermal equilibrium before running the actual dissolution-DNP experiments. To this end, the protein or RNA is ideally provided under conditions that enable conventional NMR detection (μM-mM concentration range, a pH of 5-6, 10% v/v deuterated buffers). If this is not possible, e.g., due to fast sample degradation or low available quantities, a reference sample can be used to set up the detection sequence.

In the Equipment Setup section, we describe all critical pulse sequence parameters that need to be considered when setting up the detection NMR spectrometer for use with hyperpolarized water as a sensitivity booster.

We want to stress the particular importance of a careful experimental setup since many target molecules might not (or only partially) be recoverable after dilution with hyperpolarized water and mixing with the cryo-protectants and polarization agents. This is particularly important when only limited sample quantities are available (see the ‘Limitations of the Method’ section for more details.)

Choice of pulse sequence

For all experiments, we recommend detection pulse sequences that employ gradient-selective coherence editing¹⁴⁸ as these led to the best line shapes in our hands. We suggest using a selective COSY experiment or the BEST (band-selective excitation, short transient)-versions¹⁴⁹ of a 2D HMQC,¹⁰¹ 2D ¹H^N-CON,¹⁰⁵ or 3D HNCO¹⁰⁵ pulse sequence, in which all proton pulses are selective towards the amide resonances of the target protein or amine/imino resonances of a DNA or RNA –and in which all excitations are tuned to their maximal, 90° values. The frequency selection of all these schemes guarantees that the water resonance is neither excited nor suppressed upon NMR acquisition.

This is important as pulsing on the waterline would i) lead to accelerated loss of the water hyperpolarization and ii) can lead to radiation damping effects that may prohibitively broaden the water line obscuring the protein and DNA/RNA resonances upon detection.

The suggested sequences, their ensuing spectra, and their usage are listed in Table 1. We adapted COSY, HMQC, and HNCO experiments from the Bruker TopSpin 3.7 or 4.0 library to obtain these pulse sequences. In particular, we rendered all proton pulses selective towards the resonances >6.5 ppm using band-selective shaped pulses.¹⁴⁹ Several examples can be found in Figures S1-S3 in the Supplementary Material. The basis pulse sequences in the Bruker library can be found under the names “cosyph”, “sfhmqcf3gpph”, and “b_hncogp3d”, respectively. For the ¹H^N-CON the customized pulse sequence can be found in the work by Gil et al.¹⁵⁰ It is originally based on the ‘c_hnco_ia3d’ library entry. In addition, Table 1 lists a ‘zg2d’ sequence, which refers to a time series (pseudo 2D) of ¹⁵N-edited ¹H-1D spectra. In practice, the first increment of a BEST-HMQC is successively detected to obtain proton signals of ¹H^N-amide protons of ¹⁵N-enriched proteins or nucleic acids.

Each sequence provides a particular set of advantages.

1. The ‘zg2d’ pulse sequence is well suited for real-time monitoring applications.

2. The COSY experiment is useful for short peptides and does not require isotope enrichment.

3. The HMQC experiment is best suited to record hyperpolarized 2D spectra of well-folded proteins or RNAs, as well as of short unstructured or intrinsically disordered proteins (IDPs). However, for very crowded spectra, the HMQC cross-peaks are prone to overlap.

4. If there are cross-peaks in the HMQC, the 3D HNCO experiment can help to improve spectral resolution. Nevertheless, due to the higher-dimensionality unconventional detection schemes, such as non-uniform sampling (NUS),^105,151 need to be employed to complete the acquisition before decay of the water hyperpolarization to naught (for details, see section ‘Spectral resolution’), adding another layer of complexity to the experimental workflow.

5. When detecting data under physiological or near-physiological conditions (e.g., pH_NMR = 7.4, T_NMR = 37° C), the ¹H^N-CON experiment is well-suited. The direct ¹³C-detection neutralizes line-broadening effects due to accelerated amide proton exchange, which would impair spectral resolution in ¹H-detected experiments. However, as for the HNCO, costly ¹³C-enrichment of the target molecule is necessary. <CRITICAL> It is important to avoid pulsing on the water line. More advice can be found in the Troubleshooting table – see Step 12,35.

Pulse bandwidth

For ¹³C and ¹⁵N pulses, the conventionally employed pulse calibrations should be used. For ¹H NMR, it is necessary to consider the parameters more carefully, and adjust them as appropriate.

All 90° and 180° pulses should be calibrated following the standard procedures for the employed NMR spectrometer using a sample in thermal equilibrium. All ¹H excitation and inversion pulses should be selective not to excite the water resonance during NMR detection, as this would accelerate the loss of hyperpolarization and cause radiation damping. More advice can be found in the Troubleshooting table – see Step 12,35.

For 90° proton excitations on proteins and polypeptides, PC9 pulses¹⁵² were used with a bandwidth of ~6 ppm, centered at 9.5 ppm.

For 180° proton inversion, REBURP¹⁵³ pulses were used with a bandwidth of ~6 ppm, centered at 9.5 ppm.

In comparison with protein analysis, HyperW experiments on nucleic acids may be more forgiving concerning mis-adjustment of pulses because of a larger frequency difference to the water signal, particularly if targeting the imino proton region resonating in the 11-15 ppm range.

Conversely, targeting hydroxyl groups in sugars resonating in the 5-6 ppm range may demand carefully calibrated, relatively long pulses; in our experience, pulses with Q5 shapes have been best for this application. Note that the water resonance can become unexpectedly broad¹¹⁹ and shift over 1 ppm downfield because of the generation of a large self-acting dipolar field upon hyperpolarization, such that bandwidths that reach chemical shifts below 6.5 ppm can already significantly excite the hyperpolarized water resonance.

Selective pulses can be optimized for flip angles that are smaller than 90 degrees. Such pulses may be used in NMR experiments using exchange transferred polarization.¹⁰² On the other hand, if exchange rates between labile protons and hyperpolarized water protons are fast and the latter become replenished during the duration of each scan, maximal sensitivity is obtained with the use of 90° pulses. It should be noted that the exchange rate strongly depends on the sample pH and temperature.

Spectral resolution

Hyperpolarized water is an out-of-equilibrium substance that unavoidably decays towards thermal equilibrium. Consequently, in hyperpolarized correlation experiments, the number of t₁-increments, i.e., the number of points in the indirectly detected dimension, recorded in a dissolution-DNP experiment depends on the hyperpolarization lifetime and the delay between successive signal acquisitions (see Figure 5). The time constants for the decay of the water magnetization T_1,H2O = 1/R_1,H2O are typically on the order of 10 s < T_1,H2O < 30 s,¹¹⁹ such that the NMR acquisition needs to be completed within a theoretical limit of 50 to 150 s, corresponding to 5·T_1,H2O. It should yet be noted that the proton hyperpolarization lifetime critically depends on the degree of solvent deuteration. Hence, depending on the planned experiment, different degrees of deuteration may be required.

Figure 5.

NMR acquisition in hyperpolarized water experiments. a) The excitation profiles (green shape) of the ¹H pulses need to be selective for amide resonance, typically found between 11.5 and 6.5 ppm. In contrast, the water resonance line (black line) centered at 4.7 ppm must not be excited. b) The water hyperpolarization decays exponentially towards thermal equilibrium after mixing with the target solution (red trace). The hyperpolarization is transferred to the target protein during the recovery delay τ_R and is then read out by an appropriate NMR pulse sequence (indicated by the green boxes). The resulting protein or RNA signals also decay towards thermal equilibrium, together with the water polarization (sketched by the blue NMR lines). NMR signal acquisition needs to be completed before the water hyperpolarization has decayed to naught, i.e., within ca. one minute.

• If very long relaxation times are needed for experiments requiring longer acquisition times, deuteration of more than 99% can lead to T₁ times longer than 30 s¹¹⁹.

• However, if stronger signal enhancements but shorter relaxation times are targeted, lower deuteration, e.g., 96% have been found efficient¹⁰³.

To detect a single HyperW spectrum, choose the latter option. For longer time series, choose the former.

The T₁ decay of the hyperpolarization reservoir will also appear as an additional broadening (similar to T₂ decay) in the indirect-dimension line shapes. Generally, it is desirable to fit as many scans as possible into the detection time window determined by the lifetime of the hyperpolarized water. However, due to the requirement of hyperpolarization replenishment between sequential detections, the delay between two detections should not be chosen too short (see section ‘Recovery delay’).¹⁰²

In practice, for a ¹H-¹⁵N HMQC and ¹H-¹H COSY, a typical compromise will consist of 64 complex t₁-increments detected with two scans incorporating 0°/180°phase cycling of the heteronucleus and the receiver, within 20 s –i.e., some 80 ms between scans.

For ¹H^N-CON spectra, 32 complex t₁-increments were typically recorded within 70 s. Also in this case, two transients were recorded per increment to enable an in-phase/anti-phase (IPAP) acquisition scheme, amounting to an effective homonuclear ¹³C decoupling. Note that IPAP decoupling is also a suitable option if pulsed heteronuclear decoupling needs to be avoided, e.g., for very short recycling delays where decoupling might cause heating of the NMR probe. However, as IPAP (or TROSY)-based sequences require at least two-scan phase cycles, the obtainable time resolution might not necessarily be better than that of a pulse sequence with a longer recovery delay but fewer scans per increment.

For three-dimensional spectra (e.g., HNCO), the required number of transients cannot be recorded within the detection time window when using conventional sampling schemes. For example, 32x32 = 1024 increments require at least 300 s of acquisition. Hence, other strategies need to be employed to achieve hyperpolarized 3D spectra with feasible resolution in all three dimensions. Here we used non-uniform sampling (NUS) at ~10% NUS density. In our hands, the HMS-IST implementation¹⁵⁴ for Bruker TopSpin 3.7 worked reliably and user-friendly. Here, 24 (t₁, t₂) pairs were recorded within 70 s using Poisson-gap sampling leading to 32x32 complex increments after reconstruction.¹⁰⁵ The 24 complex incremented pairs could be sampled with sufficient signal-to-noise before the hyperpolarization decayed. A 10% NUS allowed for the reconstruction of a spectrum with a total of 32x32 data points for the two indirect dimensions, yielding a useful compromise between hyperpolarization lifetime and spectral resolution. It should be noted, however, that some ‘ghost peaks’ could be observed after reconstruction in these experiments.¹⁰⁵ When planning 3D HyperW experiments, one must be careful not to misinterpret such artifacts.

Recovery delay

In conventional NMR experiments, the recovery delay τ_R between signal detection and the start of the next iteration of the pulse sequence ensures that the measured magnetization can relax back towards its equilibrium state before being used for the subsequent detection. In HyperW NMR, the magnetization is replenished by chemical and magnetic exchange between the target molecule and the surrounding buffer during the recovery delay. As a result, the delay between successive NMR acquisitions must be adjusted differently than for thermal equilibrium spectroscopy.

The recovery delay should be long enough to allow the magnetization to flow from the hyperpolarized water to the target molecule. At the same time, τ_R needs to be as short as possible to detect a maximum of t₁-increments during the lifetime of the hyperpolarized water. Hence, a compromise between intensity and resolution is necessary.

For labile groups exchanging with the solvent at ~10 Hz rates –including imino groups in RNA, amides in intrinsically disordered proteins or polypeptides, as well as OH-groups – τ_R ≤100 ms are suitable. In fact, for some 2D HMQC acquisitions, we have observed that setting τ_R to a minimal 5 ms delay (i.e., collecting scans nearly uninterruptedly) provided optimal sensitivity; this was the case for systems where the majority of labile sites exchanged over the course of the t₁ and t₂ evolutions. However, the dependence between exchange efficiency on parameters such as the exposure of the proton exchange sites, intramolecular nuclear Overhauser effects, and temperature and pH conditions, is neither predictable nor expected to be uniform for all sites in a complex biomolecule. The optimum for the recovery delay will thus depend on the target molecule, the information being sought, and the experimental conditions. To determine this delay empirically, we test multiple delay times between 5 ms < τ_R < 500 ms for all target molecules.

Too short recovery delays combined with heteronuclear decoupling during acquisition can lead to heating of the probe or cause arcing. For Bruker Prodigy cryoprobes, e.g., the lower limit for τ_R is ca. 80 ms. When not decoupling, the recycling times can be reduced to ca. 10 ms.

Quantification of polarization transfer

For some applications, such as calculating molecular structures based on transferred hyperpolarization, it is desirable to obtain quantitative results for polarization transfer and cross-relaxation rates. This generally requires knowledge of initial hyperpolarization levels, the concentrations for the different components in the sample, and the spin-lattice relaxation rates of some spins.

The initial hyperpolarization levels can be determined by executing a reference NMR scan before the desired NMR experiment.¹⁰² The reference NMR scan can include a small-flip angle excitation pulse with a flip angle of < 1°, followed by a signal acquisition period. Pulse programming software on commercial NMR spectrometers typically allows for the inclusion of conditional statements, with which this scan can be readily executed. The majority NMR experiments acquire signals of ¹H spins, in which case the determination of the initial hyperpolarization level of water is straightforward by this method. If the NMR experiment requires the acquisition of signals of a nucleus different from ¹H, an NMR spectrometer incorporating two separate receiver channels is required.¹⁵⁵ In both cases, the initial hyperpolarization level of water spins can then be determined by comparing the signal intensity of this scan to that of a reference scan executed after the decay of hyperpolarization.

The final sample concentrations may be estimated from the known concentrations of stock solutions. However, the DNP experiment with hyperpolarized water introduces a dilution during the experiment. The dilution factor and final sample concentrations can be determined experimentally by NMR, from the comparison of signals obtained after the hyperpolarization has decayed, to signals from standard samples. Other methods may be employed as well, such as the determination of analyte concentrations using high-performance liquid chromatography (HPLC).¹⁵⁶

In most cases, modeling of polarization transfer depends on several parameters such as proton exchange and spin relaxation rates. The accuracy of models can be improved if some of the relaxation rates, such as the spin-lattice relaxation rate of water, are known independently under the prevailing experimental conditions.¹⁰² These relaxation rates can be determined directly from hyperpolarized samples using a series of NMR signal acquisitions with small-flip angle excitations.¹⁵⁷

Push times

The timing of the transfer and injection of the hyperpolarized solution into the NMR spectrometer is a critical parameter. It should be well controlled to achieve homogenous mixing of the hyperpolarized water with the target protein solution while yielding the desired final volume. In the following, we describe the parameters that need to be considered when setting up the transfer equipment (Figure 6).

Figure 6.

The dissolution-DNP setup. (Left) The water samples are hyperpolarized in a DNP platform operating at temperatures close to 1.2 K. After hyperpolarization, the water is dissolved and with a pressure of 6 bar pneumatically transferred through a PTFE transfer capillary to the NMR spectrometer for detection. To bridge longer transfer distances d_T, the capillary can be embedded in a magnetic tunnel that provides a static non-zero magnetic field (red arrow) perpendicular to the B₀ field of the NMR magnet. (Right) Inside the NMR spectrometer, the PTFE capillary is connected to an NMR tube through a connector (e.g., by HPLC hand-tight fittings) that avoids liquid leakage in the probe. The connector sits on a standard Bruker 3 cm spinner and can be screwed tightly to guide the liquid into the NMR tube. In addition, it provides an outlet for overpressure upon injection.

The push time t_T denotes the time needed to thrust the hyperpolarized water upon dissolution through a transfer capillary from the DNP system to the detection NMR spectrometer. The time depends on the distance the hyperpolarized solution needs to travel before reaching the NMR spectrometer and the pressure used to push the solution through the transfer capillary pneumatically (chase gas pressure P_T). Here, transfer distances d_T between 2 and 7 m had to be bridged, depending on the layout of the NMR laboratory. PTFE capillaries with 3.2 mm outer diameter and 1.6 mm inner diameter yielded optimal transfer efficiencies. At a typical chase gas pressure of P_T = 6 Bar, these distances resulted in push times of 1 < t_T < 3 s to reach the spectrometer. Figure 6 visualizes the relationship between the push time, pressure, and transfer distance.

<CRITICAL> The push time and chase gas pressure need to be carefully determined for every laboratory layout. In practice, “dummy dissolution” experiments are typically carried out to identify the optimal parameter set. To this end, water/glycerol mixtures are dissolved within the DNP system, transferred to the NMR spectrometer, and the arriving volume is measured as a function of the chase gas pressure and the push time.

Transfer field

Depending on the laboratory layout, the hyperpolarized solution needs to travel through regions with low magnetic fields or even zero-field crossings. Such passages can lead to loss of hyperpolarization as a consequence of accelerated longitudinal relaxation.^158,159 This is not the case when placing the hyperpolarizer within the fringe field of an unshielded magnet, but it can become problematic when using NMR magnets with efficient shielding technologies, where several zero-crossings can be found close to the opening of the magnets’ bore. These can shorten the water relaxation time into the ms timescale, prohibitively diminishing the nuclear polarization reaching the targeted biomolecule. In such situations, the transfer capillary can be surrounded by a “magnetic tunnel” that provides a constant or pulsed magnetic field during the transfer to achieve optimal experimental performance. Two solutions, depending on the distance to be covered, have been employed here. i) For longer distances d_T > 2 m, a 4-element Halbach array was surrounding the transfer capillary.¹¹⁵ The array provides a constant magnetic field strength of 0.9 T perpendicular to the B₀ field of the detection magnet. This design consists of several Halbach bar-magnet arrays, surrounding the transfer capillary, and held together by 3D-printed spacers. The array provides a relatively homogeneous magnetic field within the PTFE capillary. Such a tunnel becomes necessary when the magnetic field between the DNP apparatus and the detection spectrometer drop below about 30 mT.¹⁶⁰ ii) To cover distances d_T < 2 m, the transfer capillary can be surrounded by a copper solenoid that provides a pulsed field of >37 mT during sample transfer.¹⁴⁶ The solenoid option, in contrast to a static Halbach array, retains the capillary’s flexibility and, hence, provides the additional advantage that it can be guided through the zero-field crossing atop Bruker UltraShield Plus magnets directly into the magnets’ bores. Both options can also be combined to connect the outlet of the magnetic tunnel atop the NMR spectrometer with a solenoid. Such longer distances can be covered, and simultaneously magnetic turbulences at the entrance to the magnet bore are mitigated. In either case, the tunnel should be installed so that its entry and exit come as close as possible to the DNP and NMR magnets and reach their bores. Note that the tunnel is not too heavy to be directly attached to the installations on top of an NMR spectrometer using suitable fittings.

Injection and mixing

Once arriving at the NMR spectrometer, the hyperpolarized water needs to be injected into the NMR tube and mixed in-situ with the target solution. To such an end, various implementations of injection devices have been reported so far.

In our lab the setup that worked best was one in which the hyperpolarized solution is directly injected into the NMR tube by guiding the transfer capillary into the latter.

• The injection has been typically carried out 0.6 MPa chase gas pressure.

• The target protein solution waiting in the NMR tube is then mixed with the hyperpolarized solution through turbulence induced by the injection. Completion of the mixing process is achieved with this setup in less than 400 ms.

• Subsequently, the chase gas flow can be interrupted, and NMR detection initiated. Convection, i.e., physical movement of the sample, typically settles in less than 2 s after completion of the mixing event such that gradient encoding is possible.

• For a standard 5 mm OD NMR tube, a volume of 300 μL of the hyperpolarized solution was typically injected and mixed with 200 μL of the target solution.

It should be noted that the device connecting the transfer capillary to the NMR tube needs to be leak-tight, such that no liquid spills into the NMR spectrometer or its probe. At the same time, excessive overpressure within the tube needs to be avoided. In the setup shown in Figure 6, customized PEEK connectors between the PTFE capillary and standard Bruker BioSpin spinners were used.

(Optional) Alternatively, the hyperpolarized solution can be injected at higher pressures, i.e., at P_T = 17 Bar, if a back pressure of 10 Bar is applied inside the NMR tube.¹¹⁸ Under these conditions, completion of the mixing event within 100 ms or less can be achieved. This option requires:

• a prototype injection system (as described in reference ¹¹⁸) that enables timed pressurization of the NMR tube upon injection and mixing with the target solution.

• In addition, medium-walled NMR tubes are recommended, which are designed to withstand elevated pressures.

The setup of such an automated injection system depends on the layout of the laboratory and the used DNP and NMR system. For example, switching from 5 mm to 10 mm outer diameter NMR tubes significantly impacts the required volumes, the amounts of sample, as well as the transfer times. Therefore, giving general guidelines is hardly possible. However, it can be stated that the calibration of the push time depends critically on the distance between the two magnets and needs to be calibrated with particular care.

<CRITICAL> The mixing pressures and times need to be calibrated for each NMR system and laboratory layout, since the magnet bore lengths and heights vary.

Pulse lengths, tuning, matching, and shimming

The NMR spectrometer used for detection needs to be prepared for the dissolution DNP experiment according to the device’s specifications. In the following, we detail the procedure for the widespread Bruker BioSpin spectrometers.

To prepare the NMR spectrometer for the hyperpolarized water experiment, insert a standard reference sample (e.g., for a 5 mm OD NMR tube 500 μL of 9:1 mixture of H₂O and D₂O) into the spectrometer. Tune, match and shim the system following the recommended standard procedures. Then, for the dissolution-DNP experiment, remove the reference sample and insert the NMR tube with the target solution attached to the transfer capillary via the dedicated connector.

Finally, determine the pulse lengths. For salty samples or high-Q probes, pulse lengths can significantly deviate from typically used ones. We recommend using the final sample (e.g., from a preceding experiment) solution to determine the pulse lengths.

NMR tubes and field-lock

For the dissolution-DNP experiment, the NMR tube needs to be provided with the target solution placed at its bottom. If the target solution needs to be based on a deuterated buffer, the deuterated solvent can be used to lock the magnetic field for Bruker BioSpin NMR spectrometers. Insert the sample tube connected to the transfer capillary-spinner connector to ready the system for the dDNP experiment.

In some experimental configurations, the lock system can be turned off after pre-locking with the reference sample. The typical drift of a superconducting magnet is low enough to allow for accurate experiments at least for several hours without requiring drift compensation.

Make sure the system does not attempt autoshimming on the provisional sample; fast shimming on the sample after the hyperpolarized solvent has been injected is also possible, yet at the expense of precious seconds over which the hyperpolarization will decay (not recommended).

<CRITICAL> Lock the field before starting the dissolution DNP experiment, i.e., while the target solution is waiting. This ensures that the selective proton pulses excite the correct frequencies as soon as the detection period starts. Besides, no time is lost while the NMR system scans for the lock signal.

(Optional) Treating the NMR tubes

The NMR tubes can be treated with strongly oxidizing rinsing agents such as Helmanex III. This homogenizes the internal quartz surface and reduces the risk of air-inclusions and foaming upon injection of the hyperpolarized water.¹⁶¹

Procedure

Phase 1 - DNP sample preparation – Timing 1 h

1. Prepare a 17.5 mM aqueous solution of TEMPOL using conventional micropipettes, pipette tips, and tools for biochemistry. <CRITICAL STEP> If you see foaming in later stages of the experiment (see Troubleshooting for Step 1,2), use 4-amino-TEMPO instead of TEMPOL. [Troubleshooting']

2. Mix the TEMPOL solution with 15 % v/v glycerol. <CRITICAL STEP> You could choose to use deuterated glycerol-d₈ to reduce the number of protons; this would make the polarization step more efficient. Normal glycerol is fine for most applications; refer to the Troubleshooting advice for Step 1,2. [Troubleshooting]

3. Homogenize the solution using a vortex shaker, e.g., in a 1.5 mL centrifuge tubes. <CRITICAL STEP> Make sure that the glycerol does not stick to the bottom of the sample tube used for mixing.

[Pause Point] The aqueous DNP sample can be stored in liquid nitrogen Dewars for years.

Phase 1 - NMR sample preparation – Timing 1 h

• 4.
  Obtain the target protein or nucleic acid. Check that you have the correct isotope enrichment for your chosen experiment.

  • Natural abundance for 2D COSY detection
  • ¹⁵N-labeling time-resolved ¹⁵N-edited ¹H detection
  • ¹⁵N-labeling for 2D heteronuclear correlation spectroscopy NMR
  • ¹³C and ¹⁵N-labelling for 2D ¹H^N-CON spectroscopy
  • ¹³C and ¹⁵N-labelling for 3D HNCO spectroscopy

[Pause Point] The sample can be lyophilized for storage and resuspended when needed.

• 5.
  Dissolve the target protein or RNA in a deuterated buffer at μM to mM concentrations using conventional micropipettes, pipette tips, and tools for biochemistry. Here, typically 500 μM concentrations were used. Additional options to consider are:

  • Set the pH of the sample to 7.4 for near-physiological conditions.
  • Use physiological salt concentrations to approach near physiological conditions.
  • Use a benchtop centrifuge and centrifugal filter units to concentrate samples and exchange their buffers.

<CRITICAL STEP> Do not expose the final target solution to air for longer periods of time to avoid H-D exchange.

<CRITICAL STEP> Ensure a sterile working environment, when preparing RNA samples.

[Pause Point] Most biomolecular samples can be stored at −80° C for years, if they do not precipitate upon freezing.

Phase 2 - Prepare the detection NMR spectrometer – Timing 2 h

• 6.Insert a 5 mm NMR tube in the spectrometer containing a reference sample (500 μL) including 10% v/v D₂O.
• 7.Field-lock, tune, match, and shim the spectrometer.
• 8.Set up the desired NMR pulse sequence as described in the ‘Equipment setup’ section.
• 9.Choose a suitable delay between successive NMR detections.

<CRITICAL STEP> A delay of 300-500 ms is often appropriate to achieve maximum sensitivity. However, the timing heavily depends on the target molecule as described in the ‘Equipment setup’ section.

• 10.Make sure the employed ¹H pulses are sufficiently selective that the water resonance is not excited upon detection. Run the chosen pulse sequence with the reference sample to evaluate whether water resonances are present.

  <CRITICAL STEP> If the water resonance is detected, reduce the ¹H-pulse bandwidth or shift the carrier frequency downfield; refer to the Troubleshooting advice for Step 10,39. [Troubleshooting]

• 11.Make sure that the acquisition sequence is completed within 5 times 1/R₁ of the water signal after dissolution. For the final H₂O concentration of 2 to 4%, a 5 times 1/R₁ value of ca. 1 min can be anticipated.

  <CRITICAL STEP> If detection times become too long, only noise will be recorded towards the end of the detection period reducing averaged signal intensities; refer to the Troubleshooting advice for Step 11. [Troubleshooting]

• 12.Remove the reference NMR tube form the spectrometer.
• 13.
  Fill a 5 mm NMR tube with 150 μL of the target solution containing the protein or RNA to be detected. Consider the following options:

  • If the target molecule is not stable for longer waiting periods under ambient conditions, insert the sample after the hyperpolarization build-up.
  • Treat the NMR tube with Helmanex III beforehand to avoid gas inclusions in later stages of the experiment; refer to the Troubleshooting advice for Step 13. [Troubleshooting]
• 14.Connect tube to the injection device that connects the transfer capillary to the NMR tube and the dDNP system.
• 15.
  Insert sample tube together with the injection system into the NMR spectrometer. Consider whether any of these options are appropriate for your experiment:

  • To detect at physiological temperatures, heat the spectrometer to 37° C.
  • To further accelerate proton exchange, heat the spectrometer to temperatures >37° C.
  • To maximize temperature stability, match spectrometer temperature to temperature of injected sample.
• 16.Lock the spectrometer’s magnetic field. When using Bruker BioSpin spectrometers, use the deuterated PBS as lock solvent.

  <CRITICAL STEP> If the spectrometer is not locked properly, the spectra might be distorted.

• 17.(Optional) If an automated trigger initiates the detection sequence, make the necessary connections (e.g., by BNC) between the dissolution system and the NMR spectrometer.

Phase 2 - Prepare the dissolution and transfer system – Timing 1 h

• 18.Set the chase gas pressure and push time needed to transfer the sample from the DNP system upon dissolution to the NMR spectrometer through the transfer PTFE capillary.

  <CRITICAL STEP> Refer to the Equipment Setup for more information.

• 19.
  Determine the injection pressure and time needed to achieve complete mixing of the hyperpolarized solution with the target solution at the desired final volume of 50 to 500 μL. The volume depends on the following:

  • For a 5 mm OD NMR tube, typically a target solution volume of 150 μL was used, such that 350 μL of the hyperpolarized solution need to be injected.
  • Shigemi tubes can be used to reduce the required volume
  • For flow-cell systems volumes down to 50 μL ca be used.
• 20.Clean the transfer capillary of residual water. This can be done by blowing a compressed inert gas through the capillary for >30 s.
• 21.Connect the transfer capillary (typically 3.2 mm outer diameter, 1.6 mm inner diameter PTFE tubing) to the outlet of the dissolution system at the site of the DNP system.
• 22.
  Connect the transfer capillary to the injection system at the site of the NMR spectrometer used for detection. The following options might be useful depending on the type of experiment. Details on how to achieve each point are described in the Equipment setup section.

  • Heat the transfer lines to 50° C. This should be considered if relaxation losses are too large, and the biomolecular sample supports an elevated temperature.
  • To avoid passing the hyperpolarized liquid through low-field regions in later stages of the experiment guide the capillary through a “magnetic tunnel” (e.g., B_Tunnel = 0.9 T, perpendicular to B₀).
  • To avoid passing the sample through zero-field crossings that often appear close to the bores of the DNP and NMR magnets, use a copper solenoid wrapped around the transfer capillary to provide a small, constant field during sample transfer. (Typically 5-80 mT can be obtained using a 0.5 mm diameter copper wire and 3-10 A current.)

  <CAUTION> If you are using a solenoid, only switch it on during sample transfer to avoid excessive heating.

• 23.(Optional) If a filterable radical is used for DNP, place a fresh filter in the dissolution system to remove the radical upon dissolution.

[Pause Point] At this stage everything is prepared for the dDNP experiment. If the target protein is stable, the next stage can be delayed until the start of the hyperpolarization experiment.

Phase 2 - Prepare the DNP system – Timing 30 minutes

• 24.
  Place 250 μL of the DNP sample in a sample holder appropriate for the used DNP system. Three possible options are:

  • An Oxford HyperSense system.
  • A Cryogenic Ltd. cryogen-free magnet/cryostat system with a custom DNP insert.
  • A Bruker BioSpin dDNP prototype system.
• 25.Insert the DNP sample into the polarizer. <CRITICAL STEP> make sure the sample is inserted fast enough to ensure efficient vitrification. You might need to increase the cooling power of the cryostat during sample injection; refer to the Troubleshooting advice for Step 25 [Troubleshooting]
• 26.Cool-down the polarizer to T_DNP < 1.5 K. This is typically achieved by evaporating liquid helium with a strong vacuum pump.

If a flow-through cryostat is used, fill the variable temperature insert with a sufficient volume of liquid helium to ensure that the system retains its low temperature for >2 h.

Phase 2 - Hyperpolarize the DNP sample – Timing 2 h

• 27.Initiate microwave irradiation of the sample with >15 mW power.

  <CRITICAL STEP> Set the microwave frequency to irradiate the positive lope of the TEMPOL EPR spectrum to avoid radiation damping effects in later stages of the experiment. (E.g., ν_microwave = 94.03 GHz for a field of B_0,DNP = 3.36 T; ν_microwave = 188.05 GHz for a field of B_0,DNP = 6.7 T.)

• 28.(Optional) Modulate the microwave with a frequency of 2 kHz over a range of 50 MHz for a field of B_0,DNP = 3.34 T and 100 MHz for a field of B_0,DNP = 6.7 T.
• 29.(Optional) Monitor the ¹H signal build up using small angle (e.g., 0.1° every 5 s) detection pulses.
• 30.Polarize the DNP sample for >2 h.
• 31.(Optional) If a flow-through cryostat is used, wait until the liquid helium reservoir is nearly empty to ensure efficient dissolution in later stages of the experiment.

Phase 3 - Dissolve and transfer the DNP sample – Timing 5 min.

• 32.Load the dissolution system with 3-5 mL D₂O in dependence of the used system. Use deuterated PBS instead of neat D₂O if dilution of the sample buffer is to be avoided upon injection.
• 33.Heat the dissolution liquid to 1-1.5 MPa in dependence of the used dissolution system.
• 34.
  Connect the dissolution system to the hyperpolarized DNP sample within the DNP polarizer. This step heavily depends on the used dissolution-DNP system and needs to be implemented as foreseen for the system in use.

  • For the Oxford HyperSense and SpinAligner system use the provided dissolution system.
  • For a Cryogenic Ltd. cryogen-free magnet/cryostat system with a custom DNP insert use a system as described in reference.¹⁶²
  • For a Bruker BioSpin system with a flow-through cryostat operate as described in reference.¹⁴³
• 35.Dissolve the sample by squirting the heated dissolution liquid on the sample. Then, push the dissolved sample with a chase gas pressure of 0.6 MPa out of the DNP system.
• 36.
  Transfer the sample to the NMR spectrometer and inject it into the NMR tube waiting in the spectrometer. This is typically done by pushing the hyperpolarized water pneumatically through the transfer capillary with a chase gas pressure of 0.6 MPa. There are a few options to consider:

  • Use a magnetic tunnel and/or solenoid system.
  • Pre-heat the transfer capillaries.
  • Use a high-pressure transfer system applying a chase gas pressure of 1.7 MPa as described in the ‘Equipment setup’ section.

  <CRITICAL STEP> The transfer should not take more than 3 s to avoid loss of the water hyperpolarization.

• 37.Upon injection of the hyperpolarized water into the NMR tube, the injected and waiting solutions need to be mixed. This is typically achieved by the force of injection as determined by the chase gas pressure. Inject the hyperpolarized solution to result in the exact sample volume (e.g., 500 μL) that was used for pre-shimming and pre-tuning.
  <CRITICAL STEP> It is important that the shims and tuning/matching conditions are kept constant to avoid distortions of the recorded spectrum.

  • The injection capillary must be tightly connected to the NMR tube to avoid spilling of liquids into the spectrometer or the probe.
  • Avoid excessive overpressure in conventional NMR tubes, by providing appropriate pressure outlets. Alternatively, rely on medium-size wall tubes.

Phase 4 - NMR Detection – Timing 2 min.

• 38.Initiate NMR detection upon completion of the mixing process. It is possible to use an automated trigger system that initiates NMR detection upon completion of the mixing process.

  <CRITICAL STEP> The timing of this step is important. The stabilization of the solutions can take up to 5 s, depending on the employed injection system. If the delays between decoupling pulses are too short this can cause probe heating. [Trouble shooting]

• 39.
  Record signals using the previously set up pulse sequences as described in the ‘Equipment setup’ section. Options include:

  • A selective ¹H-¹H 2D COSY spectrum. Be aware that selective pulses in 2D COSY can lead to artefacts along the main diagonal.
  • An amide selective ¹H-¹⁵N 2D HMQC spectrum (e.g. BEST HMQC).
  • A selective ¹³C-¹⁵N 2D ¹H^N-CON spectrum.

    • An amide selective ¹H-¹³C-¹⁵N 3D HNCO spectrum (e.g. BEST HNCO) using non-uniform sampling techniques. Reconstruct the HNCO spectrum.
  • A series of amide selective ¹⁵N-edited ¹H spectra (e.g., the first transient of a BEST-HMQC experiment).
• 40.Accumulate data until the hyperpolarization has decayed to thermal equilibrium.

  <CRITICAL STEP> An excessively long detection period will reduce the experiment’s performance. [Troubleshooting]

  [Pause Point] Once detection is completed, the remaining steps are not time critical.

Phase 4 - Collect reference spectra – Timing 24 h

• 41.
  After completion of the experiment and complete decay of the hyperpolarization, collect a reference spectrum in thermal equilibrium to determine which signals have been enhanced. The following are example options:

  • Use exactly the same pulse sequence as used for hyperpolarized detection, only via consecutive signal averaging to achieve an acceptable signal-to-noise (SNR) ratio.
  • Prolong the recovery delay between successive detections to assure that the reference is collected in thermal equilibrium to determine enhancement factors ε.
  • Collect a reference spectrum under conditions optimized for prototypical NMR of biomolecules (e.g., pH 6.5, T_NMR = 25° C, 10% D₂O v/v) to compare your result to those anticipated for conventional methods.
  • Compare signals of a reference compound, such as DMSO or DSS, instead of the protein, if the signal of the protein is too weak.

Timing

• Equipment Setup: 1d

• Reagent Setup: 3-4 h

• Prepare Detection spectrometer: 1 h

• Transfer system setup: 1h

• DNP setup: 30 min

• Hyperpolarization build-up: 2 h

• Dissolution, Transfer and Detection: 2 min.

• Collect reference data: 1 d

It should be noted that these times represent averages. In a ‘worst case scenario’, e.g., with very low signal intensities in thermal equilibrium, the equipment setup and the reference data collection can take up to 2-3 days. The setup depends critically on the employed magnetic field B₀ and NMR probe (e.g., room-temperature vs. cryogenically cooled), and it is hence complicated to anticipate exact timings.

Troubleshooting

Troubleshooting advice can be found in Table 2.

Table 2.

Troubleshooting

STEP	PROBLEM	POSSIBLE REASON	SOLUTION
1,2	Too slow polarization build-up	Degradation of radical stock	Check the radicals using, e.g., CW EPR. Use a fresh batch, redo the stock solution.
25	Biexponential, inefficient polarization build-up	Crystallization of DNP sample upon cool-down	Reduce cool-down time to achieve efficient vitrification.
28	Too slow polarization build-up	Microwave excitation inefficient	Recalibrate the microwave frequency and power
36	Loss of hyperpolarization during transfer	Low-field regions or zero-crossing along the transfer path	Use a magnetic tunnel and/or solenoid to cover the transfer capillary
34,36	Incomplete mixing	Too low injection pressure	Increase injection pressure or reduce back pressure
1,2	Foaming upon mixing	High protein concentrations	Switch from TEMPOL to 4-amino-TEMPO or consider applying a higher pressure to the sample upon injection. Additionally, co-solutes can be added to the sample if not interfering the experiment.
13	Gas inclusions upon injection	Microscratches inside the NMR tube	Switch tube, treat tube with Helmanex III
10,39	Distorted or broad resonance lines	Radiation Damping	Reduce excitation and inversion pulse bandwidth
11,39	Spectra appear very sparse	Insufficient polarization transfer from water to target	Use a longer recovery delay
9,39	Weak signal enhancements	Acquisition beyond hyperpolarization lifetime	Reduce number of increments or shorten the recovery delay
14	Weak signal amplitudes	Too slow polarization exchange	Increase temperature and/or pH
9,39	Too low resolution	Insufficient number of increments in the indirect dimension	Re-calibrate pulse sequence and reduce recovery time. Alternatively, use NUS to improve resolution
38	Distorted spectra in indirect dimensions	Bad gradient performance due to sample convection	Prolong delay between mixing and start of the detection period
10,39	Fast loss of water polarization upon NMR-detection	Depletion of water magnetization due to partial excitation	Reduce proton excitation bandwidth or shift the carrier frequency downfield
11,39	Sample heating upon detection	high power deposit in the radiofrequency coils	Prolong the recovery delay or reduce the decoupling power or length

Anticipated results

The central anticipated result of following this protocol is that the signal amplitudes in the multidimensional NMR spectra of biomolecules are substantially enhanced. Since the hyperpolarization source is simply the aqueous buffer surrounding the target molecule, the presented method is versatile, applicable to a vast array of biomolecules and has been observed for several different target molecules and pulse sequences.^{99-109,119,147,163}

It can be used to answer many different research questions, such as mapping solvent-exposed surfaces, protein and RNA folding, or membrane interactions. In this section we describe results for:

• Mapping of solvent interaction surfaces of intrinsically disordered proteins

  1. mapping the Heparin-binding site of the extracellular cytokine Osteopontin (OPN) under near-physiological conditions
  2. assessing the solvent shielding of α-Synuclein (α-Syn) an IDP involved in fibril formation and Parkinson’s diseases
• Monitoring of intermediates in protein and RNA folding upon binding

  1. Folding and unfolding of Ribonuclease Sa (RNase Sa), a protein from E.coli
  2. ligand binding to the guanine-sensing riboswitch aptamer domain (RNA) of the xpt-pbuX operon in Bacillus subtilis (GSRapt)
• Assessment of solvent-exposure in heterogeneous systems

  1. Membrane interactions, e.g. with Ala-Gly dipeptide
  2. Conformationally exchanging systems

Mapping of solvent interaction surfaces of intrinsically disordered proteins

So-called Intrinsically disordered proteins (IDPs) and peptides have unstable secondary and tertiary structural elements. NMR experiments are very helpful towards understanding their function due to the high flexibility and fast dynamics. Characteristically, these substrates have greater solvent exposure than other proteins and lack of solvent-shielded domains. Consequently, exchange with hyperpolarized water is typically possible along the complete primary sequence leading to substantial NMR signal enhancements throughout the entire protein. This feature can be used to (i) map ligand binding sites and (ii) to assess solvent-shielded domains.

i). Example of ligand binding site mapping

The extracellular cytokine Osteopontin (OPN) is a ~300 amino acid-long IDP associated with cell signaling and metastasis in cancer cases.^164,165 The protein features a compacted core region, such that dissolution in hyperpolarized water entails substantial signal enhancements of the entire protein, except for the compacted core. To fulfill its signaling functions, OPN binds to heparin and hyaluronic acid. The binding of these ligands leads to an unfolding of the IDP’s core region and increased exposure to the surrounding buffer (Figure 7a).^166-168 As a result, the binding site is shielded from the solvent upon complex formation leading to reduced signal enhancements due to suppression of proton exchange with the surrounding water, while the core experiences increased enhancements.

Figure 7.

Mapping of solvent exposure in IDPs through hyperpolarized water. a) OPN features a compacted core region that buries most of its glycine residues reducing their hyperpolarization exchange efficiency (left). Upon heparin-binding, the compacted core melts and the glycine residues become visible when OPN is dissolved in hyperpolarized water (right). b) ¹H-¹⁵N HMQC of OPN’s glycine signals at pH 7.5 and 37° C. In conventional NMR, the signals are weak and broadened, causing significant signal overlap (grey). When using hyperpolarized water to boost signal intensities, the glycine resonances, upon Heparin-binding, experience a signal boost such that the individual signals can be detected (red). c) ¹H-¹⁵N HMQC of α-Syn, conventionally detected (blue) and measured in a hyperpolarized buffer. The average signal enhancement amounts to ca. 60-fold. Residues in the CTD remain unaffected by the hyperpolarized water (for a superposition see the supplementary material). d) Mapping of signal enhancement onto α-Syn’s primary sequence. The CTD residues experience much lower signal boosts as a consequence of slow proton exchange. Figure reproduced with permission from refs. 166-168, ACS, Wiley.

This effect enabled us to map the Heparin-binding site under near-physiological conditions, i.e., in physiological saline at a pH of 7.4 and T_NMR = 37° C, where conventional NMR provided only weak and heavily broadened NMR signals impeding data interpretation.¹⁰¹ Figure 7b shows a ¹H-¹⁵N HMQC of OPN’s glycine signals. Under near-physiological conditions, the signals are weak and broadened, causing significant signal overlap (grey). When using hyperpolarized water to boost signal intensities, the glycine resonances are not enhanced. These residues (except for one) are buried in the compacted core. However, ussspon Heparin-binding, some of the glycine signals are enhanced such that the individual signals can be discerned (red).

ii). Example of solvent shielding assessment

α-Synuclein (α-Syn) is an IDP involved in fibril formation and Parkinson’s diseases. Its C-terminal domain (CTD) displays reduced solvent interaction due to electrostatic shielding effects that result from an accumulation of charged amino acids. Consequently, when dissolved in hyperpolarized water α-Syn’s CTD receives hyperpolarization less efficiently than its N-terminal region. Such effects of reduced solvent exchange (even despite the absence of stable structural elements) can hence readily be mapped using the proposed protocol.¹⁰³

Figure 7c shows a hyperpolarized (red) spectrum of α-Syn compared to its conventional counterpart in thermal equilibrium (blue). Averaged over all resonance, the former provides a 60-fold sensitivity enhancement compared to the latter. However, residues located in the CTD (Figure 7d) are not signal-enhanced and remain below the detection threshold in a hyperpolarized water experiment. This type of subtle effect is often overlooked in conventional NMR experiments but can be probed with superior signal intensity using the HyperW approach.

Monitoring of intermediates in protein and RNA folding upon binding

NMR spectra acquired with hyperpolarized water can be sampled at much higher temporal rates, because the sensitivity is much greater. This makes it possible to monitor rapid folding and unfolding events in real time. This has been exemplified using protein as well RNA substrates:

i). Protein folding and unfolding

The Ribonuclease Sa (RNase Sa) from E.coli is a small enzyme (96 amino acids) comprising a two-stranded β-sheet, a four-stranded antiparallel β-sheet, and a turn of 3/10 helix followed by an α-helix. The refolding of RNase Sa upon removing denaturation agents such as urea proceeds fast on the seconds time scale (a folding rate of 1.1 s⁻¹ was reported). Using hyperpolarized water, it was possible to record a 2D correlation spectrum of an RNase Sa during refolding after dilution of the denaturation agent.¹⁴⁷

Figure 8a displays a conventional NMR spectrum of RNase Sa (grey) superimposed on a hyperpolarized spectrum (blue) recorded during the refolding process. In this experiment, hyperpolarized water was injected into a solution of urea-denatured RNase Sa. The urea concentration was reduced upon dilution of the sample with the arriving hyperpolarized solution leading to a refolding of the target protein. An HMQC spectrum was recorded immediately, and thanks to the signal boost, was completed within 5 s. This spectrum hence could provide a snapshot of the forms found during refolding. Figure 8b shows the signal enhancements resulting from HyperW; this complementary information shows that residues in loop regions are preferentially amplified during the refolding event.

Figure 8.

Folding processes assessed by hyperpolarized water. a) ¹H-¹⁵N HMQC spectrum of folded RNase Sa (grey) recorded by conventional thermal equilibrium NMR and corresponding hyperpolarized spectrum (blue) recorded during the refolding process subsequent to removal of urea as denaturation agent. The signal-boost enabled to record the blue spectrum within a few seconds enabling the acquisition of a “snapshot spectrum” during the re-folding event. b) Signal intensities obtained through hyperpolarized water mapped onto the crystal structure of RNase Sa. c) Slices taken from fast 2D NMR spectra measured of RNAse Sa undergoing refolding. The folding rate is modulated by choice of final denaturant concentration. The top traces are from a signal of the folded protein, whereas the bottom traces are from a signal of unfolded protein. d) Real-time NMR of GSR^apt structural adaptions upon binding to its hypoxanthine ligand. The colored strips guide the eye to signals that experience temporal evolution on the 0-10 s timescale subsequent to the binding event. e) Structure of GSR^apt and indication of the signal assignments in panel c (top). The time dependence of normalized signal intensities for residue G29 after binding of GSR^apt to hypoxanthine. The structural changes causing varying signal amplitudes could be monitored with a 2 Hz temporal sampling rate. Figure reproduced with permission from refs. 107,108, Elsevier, National Academy of Sciences.

The folding rate of an initially unfolded protein can be modulated by changing the concentration of a denaturant present in the sample. Folding proceeds more slowly when a larger amount of denaturant is present. Figure 8c shows slices obtained from fast 2D NMR spectra that were measured of RNAse Sa at different final denaturant concentrations.¹⁰⁸ In the different spectra, the extent of completion of folding, therefore, differed at the time of the NMR measurement. This is evidenced by a decreasing intensity of the signal stemming from the folded protein as the denaturant concentration increases and vice-versa.

From this result, we expect that it will be possible to capture snapshots of intermediates formed during other refolding processes that take place at the time scale of a few seconds.

Unlike conventional NMR experiments that may also be carried out at variable denaturant concentrations, the experiment with hyperpolarization is capable of measuring signals under non-equilibrium conditions, i.e., while the folding process is occurring.

ii). RNA-ligand binding

The hyperpolarized water technique can also be used to follow the structural adaptions that nucleic acids undergo upon binding to their target ligands in real-time. In the example shown in Figure 8, the NMR signals of the guanine-sensing riboswitch aptamer domain of the xpt-pbuX operon in Bacillus subtilis (GSR^apt) were amplified over 300-fold enabling real-time monitoring of the structural adaptions the aptamer undertakes when binding to its hypoxanthine target.¹⁰⁷ This high-affinity interaction triggers RNA refolding that proceeds on the timescale of seconds.

Figure 8d shows how the signal amplitudes of GSR^apt vary during the first 10 s after mixing with the ligand. In this experiment, the ligand and the hyperpolarized water were injected into the RNA solution at the same time. This simultaneously led to a substantial boost in the signal, and initiated the binding event resulting in changes in the relative signal amplitudes. Figure 8e shows the structure of GSR^apt and the signal assignment for the lines highlighted in Figure 8d. The inset exemplarily shows how residue G29 could be monitored with a sampling rate of 2 Hz. The normalized signal intensities allow one to follow the completion of the refolding event within the first 10 s after mixing with the hypoxanthine ligand.

Assessment of solvent-exposure in heterogeneous systems

Hyperpolarization transfer efficiency depends on the position of the residue within the tertiary and secondary structure of the target protein.

In folded proteins, residues buried in the hydrophobic core are shielded from the solvent by the external parts of the protein. For example, site-specific signal enhancements have been observed for the well-folded ubiquitin, dominated by strongly primary sequence-dependent proton exchange rates and nuclear Overhauser effects.¹⁰⁹ Hence, in contrast to IDPs where more homogenous signal enhancements can be anticipated, folded proteins display heterogeneous enhancement patterns due to differential surface exposure. Proteins or peptides can also be shielded from the solvent when they interact with other materials present in the matrix, e.g. membrane vesicles or salts. When looking at proteins embedded in membranes, signal enhancements will only occur for residues in solvent-exposed parts of the target molecule. A highly charged protein region may also be shielded by its interaction with polyvalent cations or anions, if these are present.

i). Membrane interactions

Using an Ala-Gly dipeptide as a model system, it was shown that interaction with membrane surfaces (liposomes in this case) leads to almost complete suppression of hyperpolarization exchange to the membrane-inserted Ala-residue.¹⁰⁴ Figure 9a shows a ¹H-¹H COSY spectrum of free (red) and membrane-bound (blue) Ala-Gly recorded in hyperpolarized water. In the former case, all residues are homogenously enhanced (here ca. 120- to 190-fold), yet in the latter, only the Gly residue remains visible with an enhancement of ~20. The insertion and shielding of the Ala residue are visualized in Figure 9b via a snapshot from a molecular dynamics simulation of the encounter. Note how only the Gly-residue remains solvent exposed.

Figure 9.

Mapping of solvent-exposed surfaces. a) ¹H-¹H COSY of an Ala-Gly dipeptide dissolved in hyperpolarized water in its free (red) and liposome-bound (blue) states. Note how the blue spectrum appears sparser as the insertion of the Ala residue into the liposome effectively shields it from the hyperpolarized water. b) Simulated binding mode of the Ala-Gly dipeptide on a membrane surface. Only the Gly-residue remains solvent-exposed and hyperpolarizable through exchange with the surrounding buffer. c) ¹H-¹⁵N HMQC spectrum of hyperpolarized (red) and thermally polarized (blue) R17 showing folded and unfolded forms under slow interconversion. The proposed protocol enhances the disordered residues appearing in the central 8–9 ppm/118–128 ppm ¹H/¹⁵N amide region more strongly than the well-resolved peaks arising from the folded form and appearing in the periphery of this “box”. Figure reproduced with permission from refs. 104,106, ACS.

In a related experiment, interactions of proteins with lipid structures such as micelles can be characterized by observing polarization transfer from hyperpolarized lipids.⁹⁰ This polarization transfer occurs predominantly through cross-relaxation. In ref. ⁹⁰, it was demonstrated that an encoding scheme using selective pulses can be used to distinguish polarization transfer from head or tail groups of the lipid to the unfolded membrane protein OmpX. Thus, this experiment accesses the mode of insertion of the protein into the micelle.

ii). Conformationally exchanging systems

In another application, R17, a 13.3 kDa system that displays folded and unfolded forms under slow interconversion (on the NMR timescale) was investigated using hyperpolarized water.¹⁰⁶ The conformational exchange dynamics between states folded (F) and unfolded states (U) display an exchange rate k_ex = F→U +U→F, with an upper limit of 0.01 s⁻¹ at 37 °C. This provided an exciting platform for assessing the “exchange filter” effect in hyperpolarized water experiments. As individual resonances should be observable for each of these forms, one expects that the enhancement will highlight the unfolded, exposed residues over their folded, protected counterparts. Figure 9c demonstrates that this is indeed the case by comparing hyperpolarized and thermal data recorded at 37° C and 2% H₂O (v/v) on this 13.3 kDa polypeptide, where the [U]/[F] equilibrium constant is ~1. Even a cursory investigation of the spectra shows that the protocol enhances the disordered residues appearing in the central 8–9 ppm/118–128 ppm ¹H/¹⁵N amide region more strongly than the well-resolved peaks arising from the folded form and appearing in the periphery of this “box”.

It should be noted that signal enhancements are not necessarily always stronger for residues included in intrinsically disordered regions than their folded counterparts. Instead, facilitated hyperpolarization exchange rates can be observed in folded proteins due to changing local acidity or charge patterns. In this respect, hyperpolarized water provides a unique tool to observe such effects via differential signal enhancements (see for example Kaderavek et al.)¹⁰⁹.