Three-bond H – H coupling constants depend on the dihedral angles, and consequently they are key to determining the ring conformation. In general, most hexoses are monoconformational and are generally in a ⁴C₁ or ¹C₄ chair conformation. Due to the cyclic nature of the sugars, many redundant interprotonic three-bond coupling constants are available and it is relatively simple to define the conformation from the Cremer–Pople polar coordinates (Cremer and Pople, 1975). Residual dipolar coupling constants provide also data about the relative orientation of two vectors in partially oriented media (Martin-Pastor and Bush, 2001). In carbohydrates, few examples of conformational equilibrium are known; however, one such example is the iduronate residue, which can exist in at least three conformations, ¹C₄, ⁴C₁, and ²S_O, easily distinguishable by their coupling constants. Generally, internal iduronate residues are found in fast equilibrium between ¹C₄ and ²S_O (Figure 1A,B) (Mulloy et al., 1993). This equilibrium is fundamental in the molecular recognition events in which heparin is involved (Canales et al., 2005) and also has been studied by residual dipolar coupling (Jin et al., 2009)

NOE (and other related parameters such as ROE, transverse NOE, and off-resonance ROE)-based distance constraints are often used to study the glycosidic angle behavior and can thus be used for the determination of the 3D shapes of saccharides. The scarcity of experimental restrictions in the analysis of carbohydrates implies a more accurate quantification of the NOE. Fortunately, for small molecules, distances are easily quantifiable from the NOE by means of an analysis of the growing NOE curve at several mixing times (Macura et al., 1986). Then, assuming equal motional behaviors for an undetermined distance and for a known reference, the initial NOE growth rate (cross-relaxation rate) relationship is inversely proportional to the sixth power of the interprotonic distance (Neuhaus and Williamson, 2000).

Applications of this methodology have increased since the arrival of modern selective 1D NOESY methods based on the double pulsed field gradients spin echo technique (Stott et al., 1997). This scheme has advantages over the bidimensional analogs; it is faster and has better resolution. Moreover, as the protons are relaxing by T_1sel instead of T₁, the growing curves have longer linearity, and thus better-fitting and more precise results can be obtained (Hu and Krishnamurthy, 2006; Munoz-Garcia et al., 2013). The technique has also been applied to off-resonance ROESY, so as to obtain reference-independent distances. Using this approach, it has been possible to obtain the interglycosidic distances for a strongly anisotropic heparin hexasaccharide in which the interprotonic correlation time depends upon the orientation of the vector relative to the axis of the molecule (Munoz-Garcia et al., 2013).

Most conformational analyses of sugars have been based on NOE data. For the determination of the structures of ligands bound to receptor proteins, transfer NOE (TR-NOE)-based experiments are an excellent tool (Ni, 1994). In suitable conditions, a fast equilibrium in the NMR timescale is established between the free ligand (a small to medium-sized fast-tumbling molecule with a short correlation time) and the ligand bound to the high-molecular-weight receptor (a large and slow-tumbling molecule with a long correlation time) and therefore behaving as the receptor, with a large and fast-growing negative NOE. This averaged situation is reflected in the observation of a large-molecule negative NOE in the signals of the ligand (see Figures 1C,D). In general, modifications of the magnetic field strength or changes in the sample conditions such as temperature or viscosity can be used for fine adjustment if the free ligand is not clearly in the fast-tumbling regime.

In such a case, it is possible to obtain the bound ligand NOE by the analysis of the ligand signals (Ni, 1994; Neuhaus and Williamson, 2000). As a result, the qualitative analysis considering the changes between the free ligand conformation and that of the bound ligand is direct. On the other hand, the quantitative interpretation of the spectrum suffers from the potential effects of spin diffusion. In this situation, complete relaxation matrix calculations (CORCEMA or Mardigras (Borgias and James, 1990; Jayalakshmi and Krishna, 2002) should be used.

In principle, any experiment that measures interprotonic relaxation rates can be used for the analysis of the structure of the bound ligand. Thus, ROESY has been used to improve the experiment, taking advantage of its lower sensitivity to spin diffusion (Poveda et al., 1997). As the ROESY experiment suffers from some spurious effects, transverse-ROESY (Hwang and Shaka, 1992), is better suited as it is less prone to spin-lock artifacts. In addition, the quiet-NOESY experiment has been used. In this experiment, in the middle of the mixing time a biselective signal inversion is applied and a selective band inversion pulse is included, which avoids diffusion through the use of inverted spins (Zwahlen et al., 1994).

Special mention should be made of the monodimensional analogs (Stott et al., 1997). By selecting a single signal, not only is a better resolution achieved due to the change from 2D to 1D but a new dependence on T_1sel is created and the NOE exhibits a longer linearity, a feature that is particularly useful when precise distances are required (Hu and Krishnamurthy, 2006).

The STD-NMR (Mayer and Meyer, 1999) experiment consists of the difference between two experiments. It is undertaken with low-power irradiation during the relaxation delay on a sample in equilibrium that comprises a large excess of ligand(s) (from 10:1 up to 1,000:1) relative to the receptor (a large molecule), which is present at low concentration (nM to μM). In one experiment, recorded as the reference, dummy irradiation far from the signals is performed, while in the other experiment some signals from the receptor are selectively irradiated. If binding occurs, magnetization from the receptor is transferred to the ligand through close contacts with the receptor in the complex, and the effect will emerge in the difference experiment (Mayer and Meyer, 1999; Meyer and Peters, 2003).

Since its formulation, several applications have been explored for STD-NMR. The first application proposed for the technique was ligand screening; this consists of the deconvolution of a library of potential binding molecules by using several compounds at once in each experiment (Henrichsen et al., 1999). The molecules that bind better will show STD signals. Epitope mapping was the next application proposed. The experiment can be used to determine the relative importance of diverse regions of the ligand in the interaction with the receptor (Mayer and Meyer, 2001). The technique can also be used for affinity constant evaluation. The first application in which calculation of a binding constant was described used the Cheng and Prusoff equation, since the STD values are biased by the relaxation properties of the ligand proton being considered (Meyer and Peters, 2003). Latterly, the initial growth rate of the STD affinity factor has been used, in order to avoid this dependence upon the relaxation of the protons, see Figure 1E (Angulo et al., 2010). One of the potential complications in the STD is peak overlap that can make difficult the interpretation of the results. One of the solutions relies on adding another transfer step, then the peaks will be spread in two dimensions: STD-TOCSY (Mayer and Meyer, 1999) and STD-HMQC (Vogtherr and Peters, 2000). Similarly, spectral editing can also be performed using other nuclei ¹⁵N or ¹⁹F (Kövér et al., 2007; Diercks et al., 2009).

Using methods based on the relaxation matrix, it is possible to calculate the theoretical STD of a given complex. Thus, a quantitative analysis of the ligand within the complex can be carried out using a 3D structure. Based on a 3D model of the complex, using some NMR parameters and considering the saturation time and the irradiation frequency among others, CORCEMA-ST can estimate the different values of saturation transferred from the protein to a particular proton of the ligand (Jayalakshmi and Krishna, 2004, 2005). An example is given in Figure 2. This methodology has also been extended to the case of multiple binding modes within the same binding site (Angulo et al., 2008) and to the iterative refinement of a complex structure (Jayalakshmi and Krishna, 2005).

T_1sel and T₂ also depend on correlation times and in principle they can be used to study association processes for fast equilibrium between ligand and receptor, observing the ligand signals at low concentration of the receptor. These values can be used for the analysis of binding constants by performing a titration (García-Jiménez et al., 2017). A series of T_1sel /T₂ values are obtained at different ratios of ligand to receptor, and by fitting these to the variation in the ligand/receptor ratio, K_D or IC₅₀ can be extracted. Recently the application of ¹H relaxation dispersion measurements has been described for weak binding systems using both T₂ and T_1ρ (Moschen et al., 2015; Trigo-Mouriño et al., 2017).

The WaterLOGSY experiment (Dalvit et al., 2000) takes into account explicitly the water molecules surrounding the macromolecular complex. It is easy to implement and consists essentially of a selective excitation of the water molecules in a sample prepared in H₂O, followed by a NOESY mixing period and a water suppression scheme for detection (Dalvit et al., 2001).

This experiment traces the mechanisms that follow the magnetization from excited water to ligand protons. It can be a direct transfer of magnetization from bulk water to ligand, through exchangeable protons from the receptor or via water molecules buried into the receptor interface. These molecules have tumbling times closer to the receptor than to the bulk water or ligand. Due to the slow correlation times of the bound ligand or the buried water molecules the peaks through the two later mechanisms will have the opposite sign as the direct transfer of magnetization of the free compounds (Stockman and Dalvit, 2002).

The WaterLOGSY competition experiment is used for ligand library deconvolution, in which several relay pathways are used constructively, transferring bulk water magnetization to the ligand in a selective manner (Dalvit et al., 2002). In this experiment, the resonances of non-binding compounds appear with opposite sign, corresponding to NOE in the narrowing limit, and they tend to be weaker than those of the interacting ligands.

The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.