Nuclear magnetic resonance (NMR) is a spectroscopic technique to study molecules with NMR active nuclei. Not all nuclei can give NMR signals. Only those nuclei possessing non-zero nuclear spin (I¹0) have a nuclear magnetic moment which produces magnetic interactions with an external magnetic field. Sitting in an external magnetic field, the NMR active nucleus would have energy level splitting. A radio frequency pulse of the right frequency would induce transitions of nuclei from the lower energy levels to the higher energy levels. After the radio frequency pulse is switched off, the nucleus at the higher energy states would relax back to the lower states, giving out signals (figure 1). This is called the nuclear Zeeman Effect. Since the energy level splitting is different for different atomic nuclei, for a fixed external magnetic field, the frequency of the signal becomes a characteristic for a specific atom (Larmor frequency). Atoms such as ¹H, ²H, ¹³C, ¹⁵N, ³¹P, etc. are the common nuclei studied using NMR in biomedical research.

[enlarge]

Figure 1. The basics of NMR. The nuclear energy level splits in an external magnetic field Bo. The splitting DE is determined by the field strength Bo and nuclear gyromagnetic ratio γ. The radio frequency pulse induces a transition of nuclei from the low energy level to the high energy level. The nuclei at the high energy level (excited state) would relax back to the low energy level and give out NMR signals.

NMR is a powerful technique applied in a variety of field, especially in biomedical research. NMR can identify molecules, to probe molecular dynamics and study the molecular interactions. In biomedical research, it has been applied to the structure and function determination of important biological molecules, such as drug targets; study of drugs and the interactions with their targets; identification of abnormal metabolites as the biomarkers for specific diseases; diagnoses of disorders by imaging particular tissues or organs. For example, L Pellegrini et al measured the levels of L-Dopa and Dopa through NMR analysis in choriod plexus organoid fluid [4]. Ting CP et al ascertained the structures of reactants and amino acid-derived products of a novel biosynthetic pathways through ¹H, ¹³C, and ³¹P NMR analyses [5]. Kim TH et al probed the interactions within pFMRP-CAPRIN1 biological condensates with solution-state NMR spectroscopy [6].

If there were only nuclear Zeeman interaction in NMR, NMR would not be as powerful as it is. There are five major NMR interactions besides Zeeman interaction [7], including chemical shift, scalar coupling (J), dipolar interaction, paramagnetic interaction (if a paramagnetic nucleus is involved) and quadrupolar interaction (if quadrupolar nucleus with I>1/2 is involved). These interactions are susceptible to the molecular motion and the chemical environment (such as chemical bonding and atoms spacial arrangement, etc.), affecting the position, intensity, and splitting of NMR peaks (figure 2). The size and the motion of the molecules also affect NMR relaxation properties, which are fully exploited in NMR research too. For example, water relaxation behavior in tissues has been utilized in magnetic resonance imaging (MRI) to take images of different parts of the body [8].

[enlarge]

Figure 2. (a) A simplified illustration of one NMR peak at nuclear Larmar frequency. However, the NMR peak position is affected by other factors such as its electronic environment and bonding. (b) The same nucleus from two groups (e.g., H from -CH2- or -CH3 groups) would have different NMR peak positions. (c) The NMR peaks would have further splittings and shifts caused by other NMR interactions such as scalar (J) coupling, etc.

From the macromolecular level, the main building blocks of our body are composed of proteins, nucleic acids, lipids and carbohydrates [8]. Proteins have functions such as transporting molecules or ions across the cell membranes, transducing signals, or carrying out biochemical reactions, etc. Proteins could also be the structural basis of our body, like muscles. Nucleic acids (DNA, RNA) store our genetic information and can function in a complicated scheme to deliver this information from cells to cells and from one generation to the next. Lipids form the structural basis of the organelles, cells, organs and other body compartments membrane surface. Carbohydrates take part in a lot of metabolic processes and are our main source of energy. Carbohydrates, usually conjugated with lipid membranes or proteins, also play important roles in cell recognition and other specific functions. These macromolecules also have the same importance in bacteria or viruses, which are the common pathogen for us. The structure and dynamics of these macromolecules and their complex determine their functions. This information is fundamental to most of our medical problems. NMR provides specific advantages in determining the structure and dynamics of macromolecules. ¹H, ¹²C, ¹⁴N, ¹⁶O and ³¹P nuclei are the main nuclei species composing of these macromolecules. In this group, ¹²C and ¹⁶O are not NMR active nuclei and ¹⁴N is the quadrupolar nucleus, not commonly studied. Usually, isotopic labeling is applied to replace ¹²C and ¹⁴N nuclei in these molecules to NMR active ¹³C and ¹⁵N to facility NMR study.

The importance of using NMR to solve biological macromolecular structures has been recognized by awarding Nobel Prize in chemistry to Kurt Wüthrich in 2002. Wüthrich and co-workers developed NMR techniques to allow them to determine the first solution-state NMR protein structure in 1984. And after that, NMR techniques have progressed dramatically towards solving structures for bigger and more complex biological molecules [9] and high-throughput structure determination. Although there are varieties of organic molecules able to be studied by NMR, in this article, the discussion is mainly focused on the structural study of proteins.

Proteins account for more than 75% of the currently known drug targets. Some proteins are also developed as drugs to supplement to the body. The knowledge of the structure of a particular protein can help us understand how it achieves its specific function and how to fix the problems when something goes wrong. As of May 16, 2020, there are more than 163949 protein structures deposited into the protein data bank (PDB) [10]. Most of the structures are solved by x-ray crystallography. Less than 10% of the structures are solved by NMR (12983 structures), especially solution NMR. However, solution NMR still offers some particular advantages over crystallography. For example, Kitchen P et al determined the conformational change in the AQP4 carboxyl terminus caused by calmodulin binding [11]. Rice HC et al determined the binding of the APP 9-mer to the GABA_BR1a sushi 1 domain with NMR [12]. Solution NMR does not require the proteins to form crystals, which may be problematic for some proteins. By studying proteins in solution, the proteins are provided with an environment strictly mimicking their physiological state and are allowed to have their natural dynamics as well.

Looking into the molecular details, proteins are biopolymers formed by 20 different amino acids organized in different orders and into various lengths (figure3) [13]. Because of the sequence differences, the proteins would organize and form a variety of 3-dimensional structures. The structural difference would affect the atomic electronic environment, causing the atomic chemical shifts dispersion. Statistically, people have summarized the relationship between the protein secondary structures (α-helix, β-sheet, random coil secondary structures) and ¹H, ¹³C, ¹⁵N chemical shifts for the 20 different amino acids [14].

[enlarge]

Figure 3. Schematic representation of protein sequence formed by connecting different amino acids through peptide bonds. The real protein conformation is not a linear chain shown here, but a 3-dimensional structure.

It is not a trivial task to assign each chemical shift (peak) to a specific atom at specific amino acid (residue) for proteins because the one-dimensional spectra are often crowded (figure 4). One way to partially overcome this problem is to use a higher field NMR instrument. For protein NMR, it is good to obtain an NMR spectrum at a higher field because higher field offers higher resolution and sensitivity. With the development of multi-dimensional NMR techniques, the correlations between neighbor atoms or residues can be detected [15]. These correlations greatly facilitate the sequential assignment of proteins, which is the first step in solving a protein structure using NMR. Usually, these correlations are obtained through J coupling interaction between neighboring atoms or through some relaxation mechanisms.

[enlarge]

Figure 4. A ¹³C NMR spectrum of uniformly ¹³C isotopically labeled protein amelogenin at an 11.75 T magnet using Varian (Agilent) NMR instrument.

Although chemical shifts alone have been used to predict the protein secondary structures, a lot more structural constraints are needed to build the full structure of a protein. And those constraints are obtained utilizing the NMR interactions, such as scalar coupling or dipolar coupling interactions or relaxation mechanism. Karplus relation predicts the dependence of three-bond scalar coupling values (3J) to the molecular dihedral angles [16]. RDC (residue dipolar coupling) is a technique to obtain the dipolar coupling value of ¹⁵N-¹H or ¹³C-¹H of proteins in a weak alignment medium, which could help us to figure out the orientation of these bonds with respect to the molecular alignment axis [17]. ¹H-NOE (nuclear Overhauser effect) is used to measure the distance between two nuclei up to 5 Å. The cross-relaxation causes it through intra-molecular dipole-dipole interactions between the two nuclei [18].

The final protein structures are often obtained by running a simulated annealing process, such as Xplor-NIH [19], with all the chemical shifts assignments and structural constraints. After several geometry optimizations and structural refinements, an ensemble of molecules with similar structures would be generated. And the average of these structures would be used to present the final protein 3-dimensional structure.

One example given here is the study on phospholamban, a protein in the sarcoplasmic reticulum (SR) membrane of cardiomyocytes that could regulate the intracellular calcium level, affecting the contraction and relaxation of heart muscle cells. Calcium ATPase in the SR membrane is responsible for pumping calcium from the cytoplasm into the SR. The monomeric form of phospholamban can bind to calcium ATPase and inhibit its function. Phosphorylation of phospholamban relieves the inhibition. Mutations in phospholamban could cause dilated cardiomyopathy and heart failure [20]. Solution NMR has been used to study the structure of phospholamban and its variants (figure 5) [21, 22]. NMR distance and dihedral angle constraints were used as inputs in Xplor, and the final structural ensemble was generated. The structures obtained have been used to predict the binding mode between phospholamban and calcium ATPase and to explain why specific mutations are critical in phospholamban’s function [23].

[enlarge]

Figure 5. A selection of structures of phospholamban incorporated into the phospholipid membranes. The figure was produced using MOLMOL. The two solution NMR structures were generated according to the structural coordinates deposit in the protein data bank. The solid-state NNR structure was generated from the information obtained in the literature [1].

Also, a combination of NMR spectroscopy, small-angle X-ray scattering and mutagenesis has been used to generate AlphaFold models of Rec114-Mei4 and Mer2 complexes [24]. The study characterized the activation of the assembly of the meiotic DSB machinery by Rec114-Mei4 and Mer2 complexes. Furthermore, Mer2 condensation was shown to be the main regulator of the assembly, which was also supported via Rec114-Mei4. Another study applied magnetic resonance, mass spectrometry and small-angle scattering data to verify the solution structure of the polypyrimidine-tract binding protein 1 (PTBP1/hnRNP I) binding to the internal ribosome entry site (IRES) of the encephalomyocarditis virus (EMCV) [25]. The atomic-level analysis of RNP structures showed that this binding induces RNA and protein compaction with pronounced conformational flexibility. It was also revealed that PTBP1 induces the generation of several IRES RNA conformations.

Solid-state NMR is different from solution NMR in that the sample used in solid-state NMR is in the solid or semi-solid form. More than half of our body mass is water. Therefore, a lot of biochemical reactions take place in solution phase. However, there are still certain situations when the studies have to be carried out during the solid-state phase. There are solid tissues like bones, teeth, the structure of which could also be understood with the help of NMR. There are molecules embedded in the biological membranes. Membranes are in gel-like semi-solid form. Most protein drug targets are membrane proteins [26]. They usually act as receptors or transporters to pass the information or molecules in and out of the membranes. Not only in human being, but there are also interesting solid-state protein structures in bacteria and viruses too. For example, the virus capsid (the virus shell) formed by capsid proteins has a unique structure with high symmetry. HIV capsid protein has been a hot topic in solid-state NMR field [27]. The information gathered from the structural study will help us towards a cure for AIDS. Solid-state NMR is necessary to study those proteins while keeping them in the native environment.

The solid or semi-solid form of samples has very different molecular dynamics from their solution state. The molecule in the solid phase has restricted motions, and becomes anisotropic, causing a significant change in the properties of the NMR interactions [28]. The chemical shift becomes chemical shift anisotropy. The chemical shift is orientation dependent and is slightly different for each molecule with a different orientation, causing the broadening of the peaks. The dipolar or quadrupolar interaction increases in the magnitude because of less motion average. The broad lineshape makes the NMR spectrum more difficult to obtain and interpret. There are two ways to circumvent this line broadening problem in solid-state NMR. One way is to align the molecules so that all the molecules in the sample have a similar orientation. This technique has been applied to studying membrane proteins [29]. Phospholipids bilayer membranes can be aligned in the magnetic field mechanically or magnetically; the proteins embedded in the membrane are thereby aligned as well. The protein molecules in the aligned sample have similar chemical shifts and narrower linewidths. The other way is to fast spin the sample at an angle of 54.74◦ with the external magnetic field. This angle is called the magic angle. At this angle, these orientation-dependent NMR interactions are averaged to minimum mathematically [30] (figure 6). Besides these, strong ¹H decoupling has been commonly used to reduce the linewidth caused by large dipolar interactions between ¹H and other diluted spins [31]. Cross-polarization between ¹H and other diluted spins is an efficient way to enhance the signals of the diluted spins [32].

[enlarge]

Figure 6. (a) The molecules in the sample have a random orientation. (b) The molecules in the sample are all aligned in one direction. (c) The sample rotor (the sample holder) has fast spinning motion around an axis which has an orientation of 54.74◦ (θ) with the external magnetic field (Bo).

The general strategy to solve a protein structure using solid-state NMR is similar to solution NMR. The assignment of chemical shifts to the corresponding protein residues is needed first. The distance and orientation constraints have to be obtained as many as possible. The NMR interactions are orientation and distance dependent, giving the opportunities to acquire the structural constraints. The final structure is also generated with the help of simulation programs.

Phospholamban is also a perfect candidate for the solid-state NMR study because it is a membrane protein and most of its interaction partners are membrane proteins too. Interestingly, the structure of phospholamban revealed by solid-state NMR in the phospholipid environment (figure 5) is similar but still somehow different from that obtained using solution NMR techniques purely, showing the complexity of a real structural biology problem [21-23, 33]. Solid-state NMR was also sued to study the structural properties of the toxic and non-toxic types of α-synuclein oligomers [34].

Alzheimer’s disease (AD) is the most common neurodegenerative disease and the 6th leading causing of death in the US. It has been linked to the unusual protein deposit in the brain (brain plaques). The main component of the brain plaques is amyloid fibrils formed by β-amyloid peptide (Aβ). To understand the amyloid fibril structure is fundamental to understanding the cause of this disease. Solid-state NMR has been very successful in solving this structural problem (figure 7) [35, 36]. Fast magic angle spinning techniques have been applied to the fibril samples. A lot of dipolar recoupling methods have also been developed to gain distance constraints of the protein fibrils [37, 38].

[enlarge]

Figure 7. (Left) Transmission electron microscopy image of an Abeta fibril derived from human brain with Alzheimer’s disease. (Right) Structure model of the fibril shown on the left generated mainly using solid-state NMR constraints. The figure was produced using MOLMOL.

The structure of aquaporin Z (AqpZ) by ssNMR in Escherichia coli inner membranes was revealed using solid-state NMR [39]. The study applied 1017 distance restraints obtained from two-dimensional 13C-13C spectra to verify the 1.7-A ssNMR structure of AqpZ in E. coli inner membranes. To increase the signal sensitivity of AqpZ, outer membrane components were removed. Furthermore, antibiotics were added to diminish the interference of background proteins. Solid-state NMR spectroscopy was used to analyze the specific tau filament isoforms and demonstrated that amyloid fibrils consisting of acetylated three-repeat tau varied in composition, suggesting that site-specific acetylation affects tau structure [40]. The study indicated that acetylation regulates the aggregation of three-repeat tau in neurodegenerative diseases.

With the development of technology, a huge number of low molecular weight compounds have been synthesized by the chemists and a large number of biological targets have been characterized by the biologist. These compounds and their derivatives are the potential drugs. High-throughput screening is needed to pick up the right pairs, the compound (as a ligand for its target) and its interacting target, often proteins or nucleic acids. Hopefully, this compound could also intervene with the action of the target in vivo, leading to drug discovery. The understanding of the structure of the target and the compound, the target-ligand complex and the interaction mechanism could help in improving the drug design. NMR is one of the tools that could be used here for the discovery.

NMR, especially solution NMR has been widely applied in chemical synthesis as an analytic tool to identify compounds, for example, chemical syntheis of QC-01–175, a tau protein degrader [41], and biocatalytic manufacture of islatravir [42], among others [43]. Many organizations have built and maintained vast and comprehensive NMR spectra databases (e.g., www.acdlabs.com). NMR could also be a structure elucidation tool for those biological targets or target-ligand complexes, as discussed in the last section. Biological magnetic resonance databank is one of the places where a lot of structures of biological macromolecules are deposited [44]. Furthermore in pharmaceutical research, NMR has also been developed to fast screen the potential drug compounds, lead generation and optimization, namely detecting the ligand binding to their biological macromolecular targets, locating the active sites, measuring the affinity or the kinetics of the binding and optimizing the structure of the ligands based on the results [45, 46].

Using Alzheimer’s disease as an example again, the fibril-forming peptide Aβ is produced when the Abeta precursor protein is sequentially cleaved by two proteases, one of which is called β-secretase (BACE) [47]. BACE is a transmembrane aspartyl protease. The inhibition of BACE is a very promising approach for the treatment of Alzheimer’s disease [48]. Fragment-based NMR screening was applied on a custom-built fragment library of approximately 10000 compounds to identify small molecules that could bind and inhibit the function of BACE [2]. First ¹H-¹⁵N HSQC 2 dimensional NMR experiment of BACE was carried out with or without adding the small molecules (figure 8). A comparison of the chemical shift for each site gave the information whether the binding occurred and where the binding site was. A chemical shift perturbation was expected for the protein active sites once the binding happened. Several lead compounds were identified. The dissociation constants Kd were also measured by integration of corresponding residue NMR peaks as a function of the lead compound concentrations [49]. With the help of other structural analysis (such as X-ray crystallography) along with NMR, a close study of the structure of the protein active sites, as well as the interaction with the small molecule, further indicated the direction of lead compound optimization [2].

[enlarge]

Figure 8. (Left) Selected regions of 2D ¹⁵N-HSQC spectra measured with a sample of 100 µM uniformly ¹⁵N-labeled BACE-1 alone (black), in the presence of 27.5(cyan), 55(magenta), 110(blue), 220(green) and 440(red) µM of a hit compound shown on the right. The chemical shifts change at two aspartates sites (Asp32, Asp228), indicating the hit compound binding location [2]. The figure was reproduced with the copyright permission from ACS.

The NMR drug screening could also be done by observation of the ligand resonance changes. The binding of a small molecule to a macromolecule would make the small molecule behave like a macromolecule, causing a significant change in its dynamics. The chemical shift would be perturbed, the linewidth would be broader, and the molecular NMR relaxation properties would be altered too [50].

In addition, photochemically induced dynamic nuclear polarization (photo-CIDNP), the only NMR method directly applicable in aqueous solution, was used to identify weak binders using low micromolar concentrations of target molecules [51]. The interaction was analyzed by single-scan NMR experiments with a duration of 2 to 5 seconds. Also, samples were evaluated using an automated flow-through platform. Moreover, the obtained data included a 212 compounds photo-CIDNP fragment library.

Metabolites are the intermediates or final products of different biochemical reactions happening in a biological system. Metabolomics is a subset discipline of systems biology where the metabolites from a specific biological system are assessed, identified and quantified to gain information about the functional state of that biological system. Theoretically, it can be used as an early diagnostic tool for diseases. The malfunction of the biological system will cause a disorder in the metabolites. NMR is the best tool to analyze this kind of complex samples, second to Mass Spectrometry [52]. Most of the time, the NMR sample is some biofluids like urine, blood or other fluid from the body. With the introduction of the stronger magnetic fields and better probe designs (cryo-cooled NMR probes or high-resolution magic-angle-spinning probe, etc.), both the sensitivity and resolution of NMR have increased enormously. NMR is, therefore, able to identify the biomarkers or pick up the fine “fingerprints” in the micromolar concentration range in the full spectrum of the biofluids, correlating them with certain diseases.

However, metabolites from a particular biological system are usually complicated and multivariate. Therefore, using NMR as a diagnostic tool for diseases is still mostly at the research stage. Skipping the detail procedure for the sampling and NMR sample preparation, the NMR spectrum is usually taken in a straight forward way as mentioned above. A one-dimensional ¹H-NMR spectrum of the sample would be the first step. A two-dimensional spectrum using correlation spectroscopy (COSY) or other NMR pulse techniques could be used to help identify the components. The spectra of the biofluid samples from multi-groups and the control have to be compared. It is important to know that the analysis and comparison of the spectra are not done in a straight way. Instead, it is done using systematically statistical analyses because of the complexity of the spectra [53].

As example, NMR has been used to identify potential diagnostic biomarkers in Alzheimer’s disease (AD) [54] or the metabolome profiling related to microbiome in irritable bowel syndrome [55]. ¹H-NMR spectroscopy and multivariate statistical methods were used to compare the metabolomics profiles of different brain regions in the wild-type mice and transgenic mice having AD-like symptoms [54]. One significant metabolic change was a decrease in N-acetyl-L-aspartate (NAA), developed even before the appearance of the clinical symptoms in transgenic mice [56].

High-resolution magic angle spinning nuclear magnetic resonance (HRMAS NMR) spectroscopy was used to evaluate cellular metabolism in Clostridioides difficile [57]. The study found recruitment of oxidative and supporting pathways combined with integration of high-flux amino acid and glycolytic metabolism. High-resolution analysis of living cells in small reaction volumes was suggested to be a significant advantage of HRMAS NMR compared to alternative MNR techniques.

Lauterbur and Mansfield first illustrated the principle of acquiring two dimensional (2D) and three dimensional (3D) images using NMR in 1973. After that, MRI has been developed and become perhaps mostly known NMR techniques by the public since they are widely used in clinical diagnosis and preclinical research. Imagines or spectra could be obtained for part of the body of a living subject noninvasively. Lauterbur and Mansfield were awarded the Nobel Prize in Medicine in 2003 for their significant contribution to medical science.

The NMR frequency of a specific nucleus is related to the strength of the external magnetic field. For the research mentioned in the above sections, the sample is usually required to be kept in a homogenous magnetic field to make the NMR peak linewidth as narrow as possible. In MRI, however, the field gradients are purposefully introduced across the sample, so that the NMR frequency of the nucleus at different positions in the sample is slightly different and the difference information is used to map the location of the nucleus in the sample (figure 9).

[enlarge]

Figure 9. The simplified MRI imaging principle. A magnetic field gradient is applied along the x-axis (Gx). Therefore, the magnetic field strength in the z-direction is different for different positions at the x-direction. If an NMR spectrum (e. g. ¹H NMR) is taken for the triangle piece put in the magnetic field, the NMR frequency is proportional to the strength of the field and would be correlated to the position of the triangle piece along the x-axis. The NMR peak intensity is related to the ¹H spin density at the specific position.

For MRI, a 2D or 3D image is taken for a particular section of the body. By applying magnetic field gradients across all three directions (x, y, z in a Cartesian coordinate), a proper section of the body can be selected. Usually, the ¹H NMR signal contributing mainly from water and fat abundantly in the body is obtained for the imaging. The contrast of the images depends on a combination of several factors such as the nuclear spin density (¹H spin density for ¹H NMR), nuclear spin-lattice relaxation (T1) and nuclear spin-spin relaxation (T2), etc. (figure 10). The images could help us identify tumors or blood clot or other abnormality in the body.

[enlarge]

Figure 10. Examples of MRI of axial images of brain (A) FLAIR image (FLAIR is a pulse sequence technique used in MRI to suppress the signals generated by fluids in the body); (B) T1-weighted postcontrast image [3]. The figure was reproduced with the copyright permission from Wiley.

For MRS, a one-dimensional spectrum of a particular location of the body is obtained (in the order of 1 mm3) by using the magnetic field gradients in x, y, z three directions [58]. ¹H nucleus as well as other nuclei such as ³¹P, 19F, ¹³C has been studied. Since a spectrum is taken, the chemical shift information could be obtained to distinguish different species. Therefore MRS has been applied in metabolites study in vivo for the diagnostic purpose. Although there are hundreds and thousands of metabolites in the body, MRS can observe only a few with very high concentration. To get a better resolution and a higher sensitivity, a higher magnetic field is preferred than MRI. In brain MRS, the commonly observed metabolites are N-acetyl aspartate (NAA), creatine (Cre), and choline (Cho), etc. [59]. NAA has been referred to as a neuronal marker; creatine and choline are also groups of molecules that are involved in energy and membrane metabolism pathways. Because the spectrum is obtained for various locations, the 2D or 3D spectroscopic image using a specific chemical shift corresponding to a metabolite could also be generated telling us the distribution of the metabolite in the tissue [60].

As an example, MRI and MRS have also been used in Alzheimer’s disease to diagnose and monitor the treatment of this disease [60, 61]. Shrinkage of the brain volume and tissue loss in specific brain regions, such as the hippocampus and temporal lobes, are useful to predict the disease at the early stage. The decrease of the brain atrophy rate could be used to monitor the treatment effect of newly developed anti-AD drugs. The relative concentration changes of some MRS detectable metabolites such as NAA, creatine, and choline have been studied and proved to be useful in the diagnosis of AD [60].

A variation of MRI is the functional MRI, which studies the metabolic process. It is discussed in more detail here.

Significant progress has recently been achieved in the development of magic-angle spinning (MAS) NMR. This method allows the identification and characterization of protein complexes, such as transmembrane proteins [62] and metalloproteins [63] and provides both structural and dynamical characteristics of the target molecules. MAS NMR identified structural details at the atomic level for T3SS [64], DsbB [65] and FimA [66] proteins.

Regarding the protein dynamics, MAS NMR may be applied to evaluate the activity of multiple regions within a protein complex synchronously. In particular, this method helped to analyze enzymatic catalysis [67] and interactions within protein agglomerates [68]. In addition to protein complexes, MAS NMR was used for analysis of nucleic acids [69] and nucleic acid-protein interactivity [70].

Concerning studies of protein complexes, MAS NMR is also useful for analysis of molecular interactions within cellular membrane, cytoskeleton components, and activity of viral proteins. For instance, this technique was suggested to be used for structural analysis of complexes formed by microtubules and associated small molecules [71, 72].

Also, MAS NMR is an effective method to investigate structural and functional aspects of viral protein complexes, especially when it is combined with crystallography or in silico simulations. In particular, it was used to study functions and conformational modifications of Cyclophilin A, a human protein which forms a complex with HIV capsid and affects HIV-1 infectivity [73]. Also, MAS NMR was recommended for structural and functional analysis of bacteriophage proteins [74, 75].

Solid-state MAS NMR was used to investigate ssDNA-bound conformation in F-specific filamentous phages [76]. The study described the structure of the gVp monomer bound to the ssDNA of fd phage in the nucleoprotein complex. The generated model showed significant conformational changes concerning the free form. These changes affect the binding mechanism and potentially activate cooperative binding in the gVp-ssDNA complex.

NMR is a fast progressing field having a tremendous impact on the biomedical research. Some of the applications are only at the beginning of their developing stage. This article only discussed some basic and widely applied approaches utilized in the research. At the same time, all the different researches also promote the hardware of the instrument to improve too. In 1995 , the first bruker 800MHz NMR magnet was made, which was very big in size and costly in maintenance and usage. Right now it has become the common instrument for protein NMR research to solve the structures of big protein molecules. NMR needs scientists and engineers from all fields to work together to promote the development of this technique.