Use of paramagnetic systems to speed-up NMR data acquisition and for structural and dynamic studies

Abstract

NMR spectroscopy is a powerful experimental technique to study biological systems at the atomic resolution. However, its intrinsic low sensitivity results in long acquisition times that in extreme cases lasts for days (or even weeks) often exceeding the lifetime of the sample under investigation. Different paramagnetic agents have been used in an effort to decrease the spin-lattice (T₁) relaxation times of the studied nuclei, which are the main cause for long acquisition times necessary for signal averaging to enhance the signal-to-noise ratio of NMR spectra. Consequently, most of the experimental time is “wasted” in waiting for the magnetization to recover between successive scans. In this review, we discuss how to set up an optimal paramagnetic relaxation enhancement (PRE) system to effectively reduce the T₁ relaxation times avoiding significant broadening of NMR signals. Additionally, we describe how PRE-agents can be used to provide structural and dynamic information and can even be used to follow the intermediates of chemical reactions and to speed-up data acquisition. We also describe the unique challenges and benefits associated with the application of PRE to solid-state NMR spectroscopy, explaining how the use of PREs is more complex for membrane mimetic systems as PREs can also be exploited to change the alignment of oriented membrane systems. Functionalization of membrane mimetics, such as bicelles, can provide a controlled region of paramagnetic effect that has the potential, together with the desired alignment, to provide crucial biologically relevant structural information. And finally, we discuss how paramagnetic metals can be utilized to further increase the dynamic nuclear polarization (DNP) effects and how to preserve the enhancements when dissolution DNP is implemented.

Graphical Abstract

1. Introduction

Nuclear magnetic resonance (NMR) spectroscopy is a well-established technique that allows the study of a variety of systems ranging from small molecules and macromolecules in solution to crystals, powders, and a variety of supramolecular aggregates, including fibrils and bone, in the solid state[1–15]. Even though NMR can provide a plethora of structural and dynamic information, it suffers from being a relatively insensitive technique due to the small energy difference of the eigenstates of nuclear spins, which translates to low population differences and consequently poor signal-to-noise ratios[12,16,17]. Achieving high sensitivity in NMR is additionally hampered by the fact that many of the most common nuclei (such as ¹⁵N and ¹³C) studied are low natural-abundance isotopes. As a result, NMR requires samples with high concentrations (in the micro– to milli-molar range) and volumes of 200– to 300 microliters or several milligrams of solid materials. When conducting NMR experiments, especially on biomolecules, it is common to rely on isotopic enrichment, which is both time-consuming and expensive[18–22]. Advancements in probe design and the introduction of cryogenic probes, which enable a significant reduction in noise levels, have also contributed to increased sensitivity of NMR spectra [23–25]. Another way to increase NMR sensitivity, which will be the focus of this review, is to affect the relaxation properties of the studied nuclei using paramagnetic probes[26–32].

Relaxation is a process in which the Boltzmann equilibrium of spin states is regained after the perturbation of nuclear spins by radio-frequency (RF) pulses. Even though the relaxation of multiple-spin systems can be a very complex process, often complicated by chemical exchange and local motions, it can be described phenomenologically by longitudinal (or spin-lattice or T₁) and transverse (or spin-spin or T₂) relaxations as defined by Bloch equations[33]. The longitudinal relaxation, characterized by the time constant T₁, describes the energy transfer from the nuclear spin system to the surrounding environment (or lattice). For this process to occur, it is critical that a stochastically fluctuating magnetic field with an appropriate correlation time is present within the system. There are many possible sources for the random fluctuating magnetic fields such as molecular tumbling, internal molecular dynamics and the effects of free electrons. If intramolecular and tumbling motions are comparable and no free electrons are present, the T₁ times of different nuclei within a molecule are closely related to the tumbling rate (also known as the correlation time τ_c) of the molecule while in the presence of an external magnetic field. The nuclei with the shortest T₁ times are in molecules that tumble with a rate that is approximately equal to the Larmor frequency. This means that faster tumbling molecules will have lower T₁ values in higher magnetic fields since the Larmor frequency increases with the magnetic field and can more efficiently match the tumbling rate. Importantly, molecules that tumble both faster or slower than the Larmor frequency will relax much slowly, due to the mismatch between the tumbling rate and the oscillating fields, and therefore exhibit high T₁ values. Consequently, the T₁ times of crystals and rigid solids can be quite long (from seconds to even minutes)[26]. When describing the relaxation processes of a paramagnetic system, it is important to consider the electronic relaxation times, which are mostly dependent on the nature of the chosen paramagnetic ion, the proton effective correlation times (if focused on ¹H relaxation), and the molecular rotational correlation time[34]. It was shown that for fast rotating molecules, or part of molecules, the electronic relaxation times dominate, and it is possible to significantly shorten the T₁ without significantly altering T₂. On the other hand, if the electronic relaxation times are longer than rotational correlation times of molecules, their T₂ times are much shorter than T₁ and consequently significant line-broadening effects are observed in the NMR spectrum[34].

To improve the sensitivity, NMR spectroscopy relies on signal averaging. However, prior to each scan, the spin states must be brought to thermal equilibrium during the recycle delay time for an optimal signal-to-noise ratio. Consequently, systems with long T₁ times are problematic since they require long recycle delays and will significantly prolong the measurement time of an NMR experiment (Figure 1A). In addition to T₁ time, it is important to also consider the potential loss of transverse magnetization, characterized by the T₂ time constant, which leads to the decay of NMR signal. The T₂ times are crucially important for the quality of NMR spectra since the linewidth observed in an NMR spectrum is inversely proportional to T₂ of the corresponding nucleus. The T₂ times are highly dependent on the nuclei dipole-dipole interaction and large macromolecules with slower correlation times have shorter T₂ times. In solid state samples the random motion is usually so small that it minimally effects the T₂ times of nuclei. Instead NMR signals are broadened by effects, often classified under T₂′, dominated by coherent residuals arising from the incomplete averaging of dipole-dipole interactions by magic-angle spinning and/or decoupling by RF pulses, distributions of chemical shifts arising from heterogeneity, external magnetic field inhomogeneity, or magnetic susceptibility effects[35,36]. Many of these effects can be refocused, albeit not all refocusing is easy to predict, with spin-echo experiments[35,37]. Additionally, decoupling RF pulse sequences were developed specifically to increase coherence lifetimes[38,39].

Figure 1. (A) Schematic representation on how the total signal acquisition time is affected by the recycle delay. Shorter T₁ reduces the recycle delay, shortening the experiment. (B) The dependence of relaxation times (T₁ and T₂) on the tumbling rate of molecules.

This scheme shows that the molecular size plays an active role on both tumbling and relaxation. An optimal range (in light-green) is indicated where T₂ is not significantly affected by the decreased T₁.

A plot of T₁ and T₂ dependence in relation to the tumbling rate of molecules exhibit three very distinct regimes as shown in Figure 1B: (i) both T₁ and T₂ are very long (“small” molecules regime), (ii) T₁ is long and T₂ is short (“large” molecules regime), and (iii) T₁ is close or equal T₂ (shown in light green). The latter regime of motion is most desirable for NMR spectroscopy since it enables short recycle delays with reasonably narrow resonance lines. In this review, we will describe how researchers have created ways to reduce the T₁ times of nuclei with minimal reductions of their T₂ times. Our main focus is to describe how to choose and position paramagnetic tags in the systems under investigation and optimize NMR experiments.

2. Systems for T₁ reduction with minimal change on T₂

Different paramagnetic agents have been utilized to reduce the T₁ relaxation times in the efforts to shorten the recycle delay times. Paramagnetic probes can be mainly divided into three classes: the nitroxide stable radicals, linear metal chelators, and macrocyclic metal chelators able to bind paramagnetic metal ions at high affinity (Figure 2)[34,40–45]. Organic radicals, such as nitroxide-based radicals (Figure 2A), are commonly used for PRE investigation of different molecular systems. In addition to generating PRE, these radicals are also used as spin labels for Electronic Paramagnetic Resonance (EPR) spectroscopy[46,47] and Dynamic Nuclear Polarization (DNP) studies[48,49]. In addition to free radicals, different metal ions, mostly from the first row of the d-block and the f-block of the periodic table, can be used as paramagnetic probes to explore a variety of paramagnetic effects[50–53]. The paramagnetic properties of metal ions are dictated by their oxidation state, coordination sphere, and electron spin states; therefore, it is possible to use different ligands to fine-tune their properties (Figure 2B and 2C)[54]. Importantly, we can observe clear differences in the paramagnetic behavior between the d-block and f-block elements of the Periodic Table. The d-block elements exhibit several coordination numbers and oxidation states. Consequently, their electronic and magnetic properties are heavily affected by the type (and number) of coordinating ligands since they interact via their valence electrons where the unpaired electrons are located. The f-block elements form primarily trivalent cations and exhibit high coordination numbers (from 8 to 9). In contrast to d-block, the unpaired electrons of the f-block elements are located in the f-orbitals, which are largely non-bond forming, so the effect of the ligand field on their magnetic properties is much less pronounced. For this reason, it is possible to consider the f-block cations as point charges because any eventual bond with the ligands affects their unpaired electrons only minimally[55,56].

Figure 2. Chemical structures of stable free-radicals and metal chelating agents commonly involved in PRE studies.

(A) Nitroxide spin labels 1-oxyl-2,2,5,5-tetramethyl-3-pyrroline-3-methylmethanethiosulfonate (MTSL) and 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO). (B) Linear metal chelators ethylenediaminetetraacetic acid (EDTA), and diethylenetriaminepentaacetic acid (DTPA). (C) Macrocyclic metal chelating agents 1,4,7-triazacyclononane-1,4,7-triacetic acid (NOTA), and 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA).

Favorable electronic relaxation times, proton effective correlation times and the rotational correlation times of the system can be fine-tuned by varying the ligands coordinated to the paramagnetic metal. In the case of a common chelator such as EDTA (ethylenediaminetetraacetic acid), it has been reported that the Ni(EDTA)²⁻ complex causes significant line-broadening to the extent that no signals were observable in the ¹H-¹⁵N HSQC spectra of the studied protein[34]. If, instead, DO2A (1,7-dicarboxymethyl-1,4,7,10-tetraazacyclododecane) is used as a chelating molecule, which forms 1:1 chelates with most transition metals and lanthanide ions, and a strong complex with Ni²⁺ cations, no significant line-broadening effects and chemical shift changes were observed for the protein resonances. Ni(DO2A) can be used to significantly decrease proton T₁ relaxation times of macromolecules with negligible line-broadening effects[34]. In contrast, metal ions with slow electronic relaxation times, such as Mn²⁺and Gd³⁺ (as well as some organic free radicals, such as TEMPO), decrease the T₂ times more significantly, and thereby induce line-broadening[57,58]. Chelators such as EDTA[34], DTPA[59], TAHA[60], DO2A[34], DOTA[61,62], and derivatives[63,64] (Figure 2) are widely used to develop tagging strategies to generate different paramagnetic effects using the desired metal[65]. Additionally, free metals in solution may cause protein structure modification or degradation and other sample instabilities, e.g., inducing precipitation. In such a case, it is preferable to use paramagnetic metals in their chelated form.

3. PRE molecules can be used to obtain structural information

In addition to expedite the acquisition of NMR spectra, PRE molecules/tags placed at specific locations in the studied system can provide important structural information (Figure 3)[66,67]. When designing the experiment, it is critical to consider that the relaxation effect decreases when moving away from the paramagnetic center with an r⁻⁶ dependence, where r is the distance between the paramagnetic center and nucleus whose signal is being observed. The region closes to the paramagnetic center where NMR signals are too broad to be detected is called the “blind zone”. The most useful region is further away from the PRE center, where the NMR signals are visible but still affected by the unpaired electrons that can be used to obtain structural information[68,69]. The sizes of both “blind” and “useful” regions are heavily dependent on the nature of the paramagnetic center. Even by taking the sharp drop-off of the relaxation effects into account, metal ions in paramagnetic molecules can affect nuclei up to a distance of 35 Å due to the large magnetic moments of the unpaired electrons.

Figure 3. Schematic representation of the tagging strategies for PRE studies.

(A) Direct or intramolecular labeling, (B) Indirect or intermolecular labeling, (C) Solvent PREs arising from random interactions between a macromolecule and paramagnetic co-solute molecules. This illustration is inspired by the models reported in the literature [68].

In biomolecules, paramagnetic probes can be introduced into the sample via simple chemical modifications such as cysteine-cysteine and amide bonds to pre-existing amino acid residues [70–72] or by using a metal binding synthetic amino acids or peptides engineered into a protein (Figure 3A), replacing the metal ion of a metalloprotein, or by adventitious binding to binding sites[73]. However, these methods are often non-trivial and time-consuming. Examples of paramagnetic tags are S-(1-oxyl-2,2,5,5-tetramethyl-2,5-dihydro-¹H-pyrrol-3-yl)methylmethanesulfono-thioate (MTSL) or Ethylenediaminetetraacetic acid chelated with Mn²⁺ (EDTA-Mn²⁺) (Figure 2) [74]. If asymmetric isotope labeling is used, as in a system of two interacting proteins, NMR experiments can be designed to distinguish between intra- and inter-molecular PRE effects which is very useful in the identification of structural changes occurring in both binding partners upon formation of transient and permanent interactions[74,75]. A paramagnetic tag can be used to detect intermolecular interactions. For instance, a modified phospholipid with a paramagnetic tag in the headgroup will affect only those nuclei that are spatially close to the tag, thus providing information about the orientation of the molecule such as a peptide, protein or a ligand) relative to the lipid head group (Figure 3b)[76]. Additionally, a judicious combination of PRE-tagging and isotope labeling can divulge information about interacting domains of different biomolecules.[45,74,76,77]. The simplest way to utilize the PREs effects for structural studies is to dissolve paramagnetic ions into the solution (solvent PREs) that contains macromolecules of interest (Figure 3C)[77]. Importantly, the PRE effect is concentration dependent since multiple PRE centers will increase the relaxation of the exposed nuclei. Paramagnetic ions dissolved in solution with suitable relaxation properties will randomly interact with the macromolecules or macromolecular complexes and will effectively suppress the signals of the nuclei that are close enough to the paramagnetic ion[26,74]. On the other hand, signals from chemical groups (amino acid residues or nucleotides) that are shielded from the solvent, and consequently from the paramagnetic ions in solution, can be observed. For example, Mn²⁺ ion is commonly used to identify the residues from the transmembrane domains of a membrane protein embedded in a lipid bilayer[78,79]. Solvent PRE is also commonly used to identify the solvent exposed residues of a protein[80,81]. Additionally, the paramagnetic effect can be used to obtain local structural information by either modifying the molecule under investigation, for instance proteins, with a conjugated paramagnetic center, or by using their innate paramagnetic center (like in the Cytochrome and heme-containing proteins)[26,82–85].

4. Solid-state NMR and PRE

Solid-state NMR spectra get substantially more complicated with the appearance of chemical shift anisotropy, dipolar and quadrupolar couplings, which are otherwise averaged out under isotropic motion of molecules in solution. In addition, very slow or no molecular tumbling in solids, crystals or other viscous and aligned samples that are studied using solid-state NMR spectroscopy usually give rise to long T₁ and as discussed above broad signals due to T₂′ effects.[86,87]. For these systems, this problem is exacerbated as the T₁ times of the dilute spin 1/2 nuclei are usually long (tens of seconds for ¹³C in powdered organics and minutes for ²⁹Si in framework silicates[88,89]). To deal with long T₁ and short T₂ times of dilute spins, or insensitive nuclei, cross-polarization (CP) methods [88,90–92] are commonly used to take advantage of the relatively short T₁ of the sensitive nuclei (usually ¹H) in addition to enriching the sensitivity of less sensitive nuclei by the highly-abundant and high-γ nuclei like protons. This means that compared to single pulse (i.e., direct detection) experiments conducted on insensitive nuclei the recycle delays can be shorter when cross-polarization (for example, protons to ¹⁵N) is used[93–95]. Therefore, the use of PRE-molecules makes it possible to decrease the T₁ times of protons and speeding-up the acquisition of cross-polarization based NMR spectra of the corresponding nuclei in solid-state.

Many solid-state NMR studies on crystalline samples have utilized paramagnetic dopants to shorten the T₁ of protons[94]. This approach was successfully utilized to either decrease the amount of sample required or shorten the experimental measurement time significantly for studies on amyloid peptides and polymorphic pharmaceutical compounds[96–98]. Using paramagnetic relaxation enhancement, it has also been possible to provide atomic-level insight into the structure and dynamics of the organic matrix (primarily type I collagen) and the mineral surface (primarily poorly crystalline calcium-rich carbonated hydroxyapatite) in bone tissues[99,100]. The intrinsic T₁ relaxation times of ¹H resonances of different amino acid residues were shortened with the use of copper(II) ions coordinated in [Cu(II)(NH₄)₂EDTA][99,100] (Figure 4). Importantly, the linewidth was not significantly affected, suggesting that the spin-spin relaxation was not altered in these samples. Shortening T₁ allowed for the acceleration of data acquisition from cross-polarization magic-angle-spinning (CP-MAS) NMR experiments, which enabled the use of a natural-abundance ¹³C bone sample (Figure 4). Furthermore, it was possible to obtain structural information due to the quenching of specific ¹³C resonances by Cu²⁺ ions in the absence of mineral. These results showed that three main amino acid residues (glycine, proline, and alanine) from the protein backbone are located close to the bone mineral surface[100]. Simply determining T₁ times for nuclei of different isotopes inside solids can also lead to useful structural and kinetic information. In the case of a X-Cu(II)–HY zeolite (Y = the starting ratio of n_si/n_Al=2.8), where X represents the number of Cu²⁺ ions anchored per unit cell, it was possible to identify the preferred binding sites for the Cu²⁺ ions inside the unit cell[101]. Also, employing in situ PRE MAS NMR technique it was possible to determine the reaction pathway of catalytic conversion of acetone to hydrocarbons, enabled by the much shorter ¹H and ¹³C T₁ times of the zeolite bound acetone molecule[101]. Note that in most solid-state NMR experiments radio-frequency induced heating of the sample is a problem and may denature the expensive isotope-labeled membrane proteins and in extreme cases can also damage the probe. Consequently, special care must be given to reduce the sample heating by preparing deuterated samples and using very fast magic angle spinning in combination with specialized low-E NMR probes and RF pulse sequences that utilize low power pulses for magnetization transfer and decupling in order to fully take advantage of the reduced recycle delay.[9,17,96,102–111].

Figure 4.

T₁ ¹H relaxation times observed from collagen, powdered cortical bone, and demineralized bone in the absence and in the presence of Cu–EDTA (30 mM). The T₁ values were determined from ¹H-spin-inversion recovery experiments, and the reported errors were estimated from the best-Fitting of experimental data. All measurements were performed on a 600 MHz Bruker AVANCE solid-state NMR spectrometer. A, alanine; L, leucine; P, proline; E, glutamic acid; O, hydroxyproline; G, glycine; CO, carbonyl. The signals from (Pα, Oα) and (Oα, Eα) overlap in the ¹³C NMR spectrum. Adapted with permission from [100]. Copyright 2010 American Chemical Society.

5. Biological membranes: bicelles and nanodiscs

Biological membranes delimit every cell and all its compartments, playing a pivotal role in basic biological processes. The driving principles of bio-membrane formation lie in the amphipathic properties of phospholipids in an aqueous environment but, despite this “simplicity”, membranes are made of a plethora of different lipids, polysaccharides, and proteins involved in intricate interactions and equilibria. Membrane component diversity is crucial to maintain stability, function, and integrity of membrane proteins[112,113]. Different aspects of membrane lipids can be studied by both solution and solid-state NMR[10,22,114–120]. Solution state NMR works best when structural studies are being performed on lipids dissolved in organic solvents or also in favorable cases when dispersed in solution. To obtain more biologically relevant structural information about the nature of the membrane phase (lamellar, hexagonal, isotropic, etc.), its order/dynamics (fluid or gel, or liquid ordered with cholesterol), and the molecular structure of embedded lipids it is better to utilize solid-state NMR methodology. When studying transmembrane and receptor proteins, studies have shown the importance to develop NMR techniques that would allow for the characterization of both the solvent exposed, more dynamic, and lipid bilayer embedded rigid protein domains at the same time[62]. To achieve this, it is crucial to develop excellent membrane-mimicking systems. The most commonly used membrane-model systems are liposomes (SUVs, LUVs, GUVs, MLVs), bicelles, and nanodiscs[121]. The very large differences in size and tumbling rates of the systems are also reflected in T₁ and T₂times of lipids and any molecules embedded in those lipids. Liposomes are the simplest bilayer-mimicking system, consisting of lipids that self-assemble into vesicles. Size-tunability is possible by mechanically breaking them down (extrusion or sonication)[122]. Bicelles consist of disc-shaped phospholipid bilayers surrounded by a rim containing short chained detergent molecules[10,123–130]. The ratio of lipid to detergent, also known as q-ratio, allows size tunability. Bicelles with large q-ratio and multilamellar vesicles (MLVs) are more suited to solid-state NMR due to being large systems and having slow tumbling rates[10]. Recently, non-covalent disc-shaped nanoparticles known as nanodiscs, have been introduced[131,132]. Several studies have demonstrated how this system represents a suitable membrane-mimicking model and have a resemblance to bicelles. These nanodiscs are made of a planar phospholipid bilayer patch surrounded by an amphiphilic belt made up of proteins (MSP[131] or its derived peptides [133]) or synthetic polymers (SMA[132], DIBMA[134], PMA[135], PAA[136], or designed polymers [135,137–142]). These nanodiscs are free from undesirable detergents and also devoid of membrane curvature. Besides their native lipid membrane-like character, the most important property is that the size of peptide-based and polymerbased nanodiscs is tunable by simply varying the lipid:peptide/polymer ratio. Therefore, these nanodiscs satisfy the requirements of solution NMR experiments whereas macro-nanodiscs (>20 nm diameter) enable the use of solid-state NMR techniques, making them an excellent system to study the structure and dynamics of membrane proteins in a near-native lipid membrane environment [139,143].

As mentioned above, paramagnetic metals have been used to shorten the T₁ relaxation of protons in solution and in solid-state NMR experiments to study membrane proteins. Since membrane proteins are embedded in a lipid bilayer, and paramagnetic metals exhibit less PRE effect for residues in the transmembrane region, a higher amount of salt is required to effectively shorten the T₁ values for the transmembrane residues. However, high amount of salt can have undesired effects on the proteinlipid interactions. To avoid this problem, metal-chelated lipids were developed that can be reconstituted in the lipid bilayer sample ( Figure 5C and D)[45]. Using a copper-chelated lipid, T₁ times of ¹H resonances of membrane lipids (Figure 5A and B), membrane embedded peptides and proteins were effectively shortened[45,76]. Since the metal-chelated lipid is immobilized within the lipid bilayer, the PRE effect has been shown to be dramatic due to the ¹H-¹H dipolar couplings enabled spin diffusion process (Figure 6), and therefore the amount of paramagnetic metal ions required to achieve T₁ reduction was significantly reduced as compared to previous studies that used the paramagnetic metal ions in the bulk solution. An additional benefit is the absence of free metal ions in the samples, which otherwise can induce undesired side effects as mentioned above.

Figure 5.

2D ¹H/¹H chemical shift correlation spectra of bicelles without (A) and with (B) 2.56 mM copper-chelated lipid obtained under 5 kHz MAS with total data collection times of 11 and 1.77 h, respectively. A 6.2-fold reduction in data collection time with a similar S/N ratio was made possible by the use of the copper-chelated lipid, as can be seen from the ID spectral slices taken at (top) 1.34 and (bottom) 3.25 ppm with (red) and without (black) the copper-chelating lipid. An RFDR7 sequence with a 100 ms mixing time and a 100 ms low-power pulse for water saturation at 35 °C was used; 512 t1 experiments with 32 scans were used, with recycle delays of 0.2 s (with copper-chelating lipid) and 2 s (without). (C) The structure of DMPE-DTPA (1,2-ditetradecanoyl-sn-glycero-3-phosphoethanolamine-N-diethylenetriaminepentaacetic acid). (D) Molecular structure of DTPA (diethylenetriaminepentaacetic acid) chelated with a copper ion. DTPA is one of the common metal ion chelators. Adapted with permission from [76], Copyright 2010 American Chemical Society.

Figure 6.

(A) Representation of lipid bilayers containing a paramagnetic copper-chelated lipid and subtilosin A, a 35-residue cyclic antimicrobial peptide that has been shown to interact with lipid bilayers with the membrane orientation depicted in (A). (B) ¹⁵N spectra of aligned 7:3 DMPC/DHPC bicelles containing 12–14% uniformly ¹⁵N labeled (only 70–82 nmol) subtilosin A (red) with and (black) without the 2.56 mM copper-chelated lipid. The spectra were obtained on a 400 MHz Varian NMR spectrometer using a ramped-amplitude cross polarization (ramp-CP) sequence with a contact time of 0.8 ms under static conditions at 37 °C. While the total data collection time was 8 h for both spectra, the recycle delay was different for samples without (2 s) and with (1 s) the copper-chelated lipid. The transfer of the paramagnetic effect in T₁ reduction for nuclei in the membrane via proton spin diffusion is also indicated in (A). (C) Primary structure of the antimicrobial peptide Subtilosin A. Adapted with permission from [76]. Copyright 2010 American Chemical Society.

Using the specially designed chelating phospholipid (i.e. DMPE-DTPA[45,76]), it is possible to enhance the T₁ relaxation without affecting the lipid bilayer orientation (Figure 7). If isotopically tumbling membrane systems that contain metal-chelated lipids are used, as can be done with isotropic bicelles or nanodiscs, the data collection times can be dramatically reduced in solution NMR experiments such as 2D SOFAST-HMQC. As shown in Figure 7, it is remarkable that this approach enabled the acquisition of a 2D ¹H/¹⁵N SOFAST-HMQC spectrum of an antimicrobial peptide without the need for ¹⁵N isotopic enrichment (Figure 7A, C and D)[45]. The used sample preparation approach can be optimized for most membrane proteins to speed-up the acquisition of multidimensional solution NMR experiments for structural and dynamic studies of membrane proteins, and dramatically reduce the amount of membrane protein required to acquire NMR spectra. Consequently, this approach is highly beneficial to overcome many intrinsic limitations in membrane protein structural studies. Some of the immediate applications include: (i) quickly acquiring NMR spectra of unstable membrane proteins or protein-protein complexes; (ii) performing NMR experiments on scarcely available mammalian membrane proteins; (iii) real-time monitoring of protein-lipid interactions induced folding, refolding, misfolding, and oligomerization of amyloid proteins; (iv) investigating the action of membrane active peptides such as antimicrobial peptides and amyloid peptides.

Figure 7.

Four-fold increase in the sensitivity of 2D SOFAST-HMQC experiments. (A) 2D SOFAST-¹H/¹⁵N-HMQC spectrum of a 9.3 mM (unlabeled) MSI-78 (also known as pexiganan) incorporated in DMPC/DHPC isotropic bicelles (q = [DMPC]/[DHPC] = 0.25, DMPC: 1,2-dimyristoyl-sn-glycero-3-phosphocholine, DHPC: l,2-dihexanoyl-sn-glycero-3-phosphocholine) containing a 2.96 mM copper-chelated DMPE lipid. (B) 3D structure of MSI-78 embedded in bicelles along with its amino acid sequence. (C) Signal-to-noise ratio obtained from 2D SOFAST-HMQC spectra of a 9.3 mM unlabeled MSI-78 in q = 0.5 isotropic bicelles without copper-chelated lipid (black) and with a 2.96 mM copper-chelated lipid (red). (D) 1D ¹H chemical shift projection spectrum obtained from 2D SOFAST- ¹H/¹⁵N-HMQC spectra that were obtained with no copper (black) and a 2.96 mM Cu²⁺-DMPE-DTPA (red). All spectra presented in this study were obtained from a 900 MHz Bruker NMR spectrometer at 35 °C using a cryoprobe. Each 2D SOFAST-HMQC spectrum was obtained from 64 t1 experiments, 256 scans, and a 100 ms recycle delay; the total data collection time (including the acquisition time and delays in INEPT) was ~54 min. Adapted with permission from[45]. Copyright 2011 American Chemical Society.

Also useful is the application of solid-state NMR which is one of the most powerful spectroscopic techniques that enables structural studies of membrane proteins in lipids under liquid crystalline state[144–146]. Many successful relaxation enhancement solid-state NMR experiments carried out on membrane proteins utilizing different paramagnetic metal ions (i.e., Co³⁺, Ni²⁺, and Gd³⁺) have been reported[62,147]. Critically, when discussing the NMR studies of membrane mimetics, it is important to consider alignment, since aligned samples can be used to reintroduce anisotropic interactions such as dipolar couplings and chemical shift anisotropy (CSA) into NMR spectra, an invaluable tool to determine the relative orientation of domains in multidomain proteins and solve the high-resolution structure and topology of the protein[10,148]. Determining how proteins orient in a lipid bilayer is crucial for understanding their biological function (Figure 8B). Macroscopically aligned lipid bilayer samples can be prepared by either using mechanical alignment of lipids between glass plates or aluminum discs or by magnetically aligning samples[148–150]. Focusing on magnetic-alignment, it is possible to align lipid bilayers in the form of bicelles or macro-nanodiscs as demonstrated in the literature[139,146]. Indeed, if the diameter of these nanoparticles is larger than 20 nm (called macro-nanodiscs), and their lipid concentration is high enough, they are likely to align in the magnetic field of the spectrometer as successfully demonstrated for many different types of polymer-based nanodiscs[139]. Many different NMR techniques have been developed to monitor the alignment of different membrane systems. For instance, a fast-tumbling lipid bilayer shows a narrow isotropic spectral line. On the other hand, if the bilayer is aligned in an external magnetic field, then the ³¹P chemical shift is anisotropic. Utilizing alignment, two-dimensional separated-local-field (SLF) experiments that correlate ¹⁵N chemical shift and ¹H-¹⁵N dipolar coupling have been used in the structural studies of membrane proteins and other functional peptides (Figure 8)[151]. The prototype of such experiments is the polarization inversion by spin exchange at the magic angle (PISEMA) experiment (and its variants) that allows the display of characteristic patterns based on a molecule’s orientation with respect to the lipid bilayer (Figure 8C and D)[151–156]. The so-called helical wheels can be used to infer a helical peptide’s tilt within the bilayer, but its analysis requires detailed knowledge of the chemical shift anisotropy tensor within the geometry of an amide bond. Several methods can be used to enhance sensitivity of these 2D SLF experiments. The copper-chelated lipid developed in the Ramamoorthy lab has also been successfully utilized for fast acquisition of solid-state NMR spectra in the investigation of structure and membrane orientation of SLN membrane protein (Figure 8A). Due to the hydrocarbon chains of the dominant DMPC lipids, which have a negative magnetic susceptibility anisotropy, the bicelles are normally aligned with the bilayer normal perpendicular to the magnetic field direction[157]. Adding paramagnetic lanthanide ions (Yb³⁺, Tm³⁺) to the studied membrane system causes them to bind to the lipid phosphate head groups which results in a change in their magnetic susceptibility from negative to positive, and consequently alters the tilt angle of bicelles. Such bicelles are characterized as “flipped” and are orientated with the lipid bilayer normal parallel to the main magnetic field[157,158]. It is important to keep this in mind when deciding what kind of paramagnetic system to use for reduction of T₁ times of the studied membrane system. Paramagnetic species have been introduced by either using metal-chelated lipids when forming membrane systems or adding 5-DOXYL stearic acid radical to the studied systems[159]. In addition to increasing the speed of data acquisition the membrane-embedded radicals can also provide structural information since they will reduce the signal intensities of nuclei that are close to the paramagnetic centers near the lipid heads. Using a special chelating phospholipid (i.e. DMPE-DTPA[45,76]) it is possible to affect the relaxation without affecting the bilayer orientation.

Figure 8.

Site-specific ¹H T₁ relaxation times of the backbone resonances of [U-¹⁵N]-SLN oriented in lipid bicelles with 5% Cu²⁺-chelated lipids measured by the 2D inversion recovery SAMPI4 experiment. B) Mapping of the site-specific T₁ relaxation times on SLN structure (blue), where unobservable/overlapped residues in the SAMPI4 spectrum are shown in white. C) 2D SE-SAMPI4 spectra of [U-¹⁵N]-SLN without paramagnetic and d) with 5% Cu²⁺-chelated lipids. The averagel5N linewidths for isolated peaks are 82.9Hz and 86.7 Hz without and with 5% Cu²⁺-chelated lipids, respectively. Adapted with permission from [151]. Copyright 2018 Elsevier.

Similarly, macro-nanodiscs can be successfully aligned in the magnetic field[136,137,139,143,160]. So far, to our knowledge, there is no report in literature about PRE studies using aligned macro-nanodiscs. At this point, several approaches can be followed. One of the straightforward approach involves the use of paramagnetically labeled phospholipids such as DMPE-DTPE, already reported for bicelles [45,76]; while another approach involves the modification of the polymer belt with paramagnetic probes. In both cases, the different position of the probes could lead to different results given the fact that the paramagnetic centers lie on different positions of the nanoparticle (lipid-bilayer surface vs polymeric rim of the nanoparticle).

In addition to free or chelated metal ions, paramagnetic species can be introduced by either using metal-chelated lipids when forming membrane systems or adding 5-DOXYL stearic acid radical to the studied systems[159]. Using paramagnetic ion chelated lipids has an advantage of ion centers being located close to the proteins, which increases the PRE effects. The amount of the metal-chelated lipids needed is also very low, only 5–10%, to ensure a significant PRE effect. Lower metal ion concentrations that are used with chelated lipids, compared to metal ions bound to free chelators, are also advantageous, since they reduce RF heating of the sample and can consequently allow for shorter recycle delay times. In addition to increasing the speed of NMR data acquisition the membrane-embedded radicals can also provide structural information since they will reduce the magnetization of nuclei that are close to the paramagnetic centers near the lipid head group.

6. Dynamic Nuclear Polarization (DNP) to improve ssNMR

Paramagnetic dopants can also be effectively utilized in DNP, a process utilized to transfer polarization from the unpaired electron of a stable radical to nuclear spins, enhancing the NMR sensitivity dramatically at cryogenic temperatures [48,161–174]. With the use of stable radicals, it has been shown to be possible to increase the polarization of the studied species by as much as 10,000 times and more at a very low temperature (^~l–2 K)[49]. This is especially useful for DNP-enhanced solid-state NMR applications [175–177]. Recently, DNP-enhanced MAS NMR has moved well beyond the proof-of-principle phase, particularly in the field of structural biology and to study materials. Various types of polarizing agents, typically stable organic free radicals, used in DNP have different Electron Paramagnetic Resonance (EPR) spectra which exhibit different rates of DNP enhancements[178]. For instance, trityl radicals have narrow EPR spectra, and are ideal for direct polarization of ¹³C nuclei and the polarization process can be additionally enhanced in presence of PRE compounds[179]. A recent study has shown that PRE additives are capable of increasing the DNP enhancement. Additionally, it was demonstrated that the reduction in the electron T₁ time of the polarizing agent by a PRE compound led to an increase in DNP enhancement. Importantly, a beneficial DNP enhancement is not always connected to shorter electron T₁ times but is highly dependent on the studied system. Solid-state ¹³C DNP signals were enhanced by 100–250% with the use of trace amounts of paramagnetic additives, with lanthanide complexes with Gd³⁺, Ho³⁺, Dy³⁺ and Tb³⁺ being the most studied[175,178,180–183]. In contrast, Cu²⁺- and Co²⁺-NOTA complexes, had virtually no impact on DNP enhancement of ¹³C signals[176]. It is also possible to directly use paramagnetic metals for sensitivity enhancements of the studied species[161]. This is useful when exogenous radicals, e.g., nitroxides, limits the sensitivity enhancement to the surface and sub-surface layers of the particles, such is the case in the study of inorganic particles and non-porous materials. Additionally, the high reactivity of radicals can also be an issue when studying solid samples. Recently, it was shown that Mn²⁺ and Gd³⁺ dopants are able to polarize ⁷Li nuclei in the bulk of micron sized LTO (Li₄Ti₅0i₂ spinel) particles with the use of DNP[161]. Compared to DNP experiments conducted with stable radicals the enhancements are low, with enhancement factors ranging from 3 to 14, but still quite significant since they can translate even to two orders of magnitude reduction in experimental time[161]. Chelated paramagnetic metal ions were also used in DNP experiments on biomolecules and molecular crystals[181,184]. Interestingly, applications of paramagnetic metals are also possible in dissolution DNP, where metals such as lanthanides, can be used to increase the polarization in solid-state, but can, due to their relaxation properties, significantly decrease liquid-state polarization[167]. This relaxation by paramagnetic metals can be minimized by chelation of the metal during the dissolution, as was shown in the case of chelation of Gd³⁺ ions by diethylenetriaminepentaacetic acid (DTPA) during the dissolution process. Gd³⁺ ions could not reduce the T₁ of the ¹³C of the pyruvate molecule in the liquid state due to the chelation with DTPA that increases Gd³⁺-¹³C distance due both to electrostatic repulsion and steric hinderance.

7. Conclusions and future perspectives

Doping NMR samples with different paramagnetic relaxation enhancement (PRE) agents is an effective way to selectively or uniformly reduce the T₁ relaxation times of molecules, chemical groups or species in a variety of samples including nanomaterials and biological compounds. As demonstrated successfully for many different systems, this approach has been well utilized in speeding-up the acquisition of NMR spectra recorded both with solution and solid samples. In fact, it has become routine to use this approach to acquire solid-state NMR of crystalline or polymorphic pharmaceutical compounds/drugs, amyloid fibers, and membrane proteins. Bones, nanocomposites, and a variety of different nanomaterials benefit from the use of paramagnetism in solid-state NMR. Aspects such as spin-lattice relaxation shortening, or the chance of selectively detecting certain molecular constituents, boost the utility of solid-state NMR spectroscopy for studying non-soluble/non-crystallizable systems.

It is important to pay special attention to the choice of the paramagnetic center (i.e., radicals vs metal ions with the optimal chelating agent) to ensure that the PRE probes decrease the T₁ times without affecting significantly the T₂ times of the molecules under investigation.

Since studies on functionalizing biomolecules with PRE agents are advancing, further developments to obtain structural and dynamic information and even real time monitoring of the kinetics of chemical reactions are now feasible. Paramagnetic metals have also been shown to be important in dynamic nuclear polarization (DNP) coupled to magic angle spinning (MAS) NMR spectroscopy, since they can offer additional sensitivity enhancement and sometimes even be a source of DNP polarization. In fact, paramagnetic DNP MAS NMR spectroscopy can enable structural studies on samples that were unreachable before due to low-abundance of NMR active nuclei.

The use of paramagnetism in solid-state NMR is challenging when applied to different membrane mimetics and used in magnetically aligned samples, since paramagnetic agents can also be used to magnetically-align and also alter the alignment direction of membrane-mimicking systems and other biomolecules in the presence of an external magnetic field. Different functionalization of membrane mimetic systems, such as bicelles and nanodiscs, can provide a way to control the position of paramagnetic centers without the introduction of “free” or direct labels to the system of interest (reconstituted membrane proteins, for example). This has the potential, together with the desired alignment, to provide crucial biologically relevant structural information using NMR spectroscopy. Such systems would also enable the applications of a combination of NMR and EPR experiments.

Highlights.

• An optimal PRE system to effectively reduce T₁ is discussed.

• PRE agents to speed-up T₁ and to provide structural information are reported.

• Challenges and benefits of using PRE in solid-state NMR studies are included.

• Use of a metal-chelated lipid for fast NMR signal acquisition is discussed.