5. Final Remarks
As we have attempted to emphasize and demonstrate, NMR has a powerful and unique role in drug design. NMR provides detailed structural information about a molecule along with kinetic information over extended time periods, i.e., not just a snapshot [24,25]. Moreover, NMR is quantitative and highly reproducible, allowing applications in diverse fields such as relaxometry, combinatorial chemistry, fluxomics, and targeted analysis [91]. NMR can be combined with other analytical techniques, such as mass spectrometry “in tandem” analysis of the molecules of interest [436,437,438]. 1H 1D-NMR is used particularly in the analysis of metabolites, while the strength and intensity of the recorded signals is directly proportional to the concentration of the sample [44,439,440]. 1H 1D-NMR can also be used to follow “real-time” analysis of different molecules [441]. In the 1D 1H-NMR experiment, there are no polarization transfer techniques required (the 1H atom is already highly sensitive) and covers a spread of interesting nuclei in the molecule(s) being studied [91].
Assuming that the sample can be stored stably for extended periods of time, the non-destructive nature of NMR permits the re-use on the sample for different experiments [50,442]. Aiding in the reproducibility of NMR [91], sample-recycling offers a significant advantage of NMR [443,444].
While always important in drug design, sensitivity and resolution of NMR are two major factors that need special consideration. Since both factors improve with increasing magnetic field strength, we have seen a spike in the demand for ultra-high-field NMR spectrometers. Recently, 28.2 T (i.e., 1.2 GHz for 1H) magnets have become commercially available, and with recent advances in magnet technology, such as liquid helium recycling and magnetic field shielding [91], NMR has begun to offer far better resolution and higher sensitivity while reducing the substantial costs of maintaining the instruments compared to past decades. Other steps have been taken to enhance the sensitivity of NMR including: the development of cryoprobes increasing the signal to noise ratio 3 to 4 times, and micro-coil probes that not only increased sensitivity but also reduced the amount of sample required for the measurements [445].
From another perspective, one can further optimize the process of obtaining spectra by using different methods of measurements. One of these, called SOFAST, helps to reduce the delay between scans resulting in lowering acquisition time for 2D experiments such as HMQC utilizing 1H,15N or 1H,13C [117,446,447]. The basic principle of this method is to use selective 1H pulses that will excite only a small portion of the available nuclei pool, while the unperturbed spins provide a magnetization “heat sink” thus improving the spin-lattice relaxation (T1) rate via dipolar interactions [446]. These methods can be highly efficient when studying drug binding and molecular interactions [117].
Another method, called ultrafast 2D NMR, enables obtaining a 2D spectra within a single scan but with the associated cost of reduced sensitivity [117,448]. The principle of this method is to divide the sample into n number of fractions, and apply the appropriate incremental aspect to each fraction, while recording them all simultaneously within the one scan [448]. This method has been applied to many metabolomic studies [177,447,449] as well as in the analysis of natural products [450]. The difficulty is that the effective concentration of the sample is lowered by the fractionation level. The more slices the sample is split into, the greater the reduction of combined signal that is obtained.
Lastly, a method termed non-uniform sampling (NUS) may provide an advantage by reducing the total time for measurement while maintaining the same resolution of spectra [117]. NUS effectively skips over parts of the total dataset, collecting only around 20 to 30% of the total. Usually sections containing higher concentrations of signal (over noise) are emphasized in the selection scheme, known as non-linear data acquisition. These methods reconstruct the complete data subset by applying various algorithms such as multidimensional decomposition (MDD), which essentially separates the sets of multidimensional data into one dimensional problems that are much easier to solve given the common process of signal overlapping in multidimensional NMR spectra [117]. Other algorithms such as compressed sensing (CS), Maximum entropy method (Max Ent) and Iterative soft threshold (IST) each have their own advantages but all focus on decreasing the time needed for collection of spectra [117,451]. The cost is in signal to noise, as no gain is ever absolutely free.
The inherent advantages of NMR do not eliminate the disadvantages (Table 3), which are namely being limited for some nuclei, and inherent insensitivity for many types of experiments. NMR can provide high quality resolution and sensitivity for some experiments [91,452] but the application can be challenging. When the experiments become multidimensional, there is often a tradeoff between the improved higher resolution and/or the resulting sensitivity and/or the amount of time an experiment takes, i.e., high-quality spectra for multidimensional experiments take much longer than their simple 1D counterparts [117] (Scheme 2). One must always choose between resolution, sensitivity, and the amount time; as you can only ever have two out of the three.
Unfortunately, in the real world, when working with NMR spectroscopy, we are mainly forced to choose between high or low precision data (Scheme 2) with fixed available instrument times. Scheme 2 can be useful, when choosing NMR technique and/or method/approaches in drug studies. It shows the general representation of different NMR techniques and method/approaches in cost-time matrix, bearing in mind that the exact position in the matrix can be influenced by the environmental conditions, pulse sequence and sample preparation (e.g., concentration). Regarding, NMR methods/approaches, the exact position depends on the choice of proper NMR technique. For the purposes of Scheme 2, the 2D techniques were chosen as described in the practical examples. Nevertheless, with relatively short times and at low cost, we can acquire numerous data sets and therefore the low precision can be partially compensated for by statistical analysis.
Fortunately, significant efforts have been undertaken to reduce the amount of time it takes to record multidimensional spectra, especially for 2D NMR, and still obtain high quality spectra (Table 3). Novel pulse sequences have been developed to decouple nuclei in Pure-shift NMR [117]. Dynamic Nuclear Polarization (DNP) can induce hyper-polarization in atoms (13C, 15N) with an inherently low sensitivity [453,454]. Parahydrogen-Induced Polarization (PHIP) and Signal Amplification by Reversible Exchange (SABRE) are other polarization techniques used to increase the sensitivity of inherently insensitive nuclei [455].