Although NMR methods have largely been employed primarily to identify compounds from pure samples as in classical natural product research, they have recently played a larger role in the analysis of mixtures of unfractionated natural extracts [
123]. The bottleneck in such complicated analysis is the congregation of the
1H NMR spectra, which could be untangled using two main strategies. The first strategy aims to simplify the abundant
1H NMR spectra through the use of diffusion and/or relaxation filters or through creating a projection of the
1H decoupled spectrum like in 2D
1H J-resolved NMR experiments. Relaxation filters are pulse sequences that selectively attenuate, or even remove, signals of components with shorter T1 or spin–spin relaxation times, T2 [
124]. Hence, they could be utilized to simplify the
1H NMR spectra through selective elimination of signals belonging to high molecular weight or rigid structure (quickly relaxing) compounds. They have also been implemented to attenuate water signals using the water attenuation T2 relaxation Carr-Purcell-Meiboom-Gill (WATR-CPMG) pulse sequence [
125]. On the other hand, diffusion filter based experiments such as diffusion ordered spectroscopy (DOSY) commonly simplify
1H NMR spectra based on the difference of the translational diffusion coefficients between molecules of different molecular size, achieving what has be described as “NMR chromatography” [
126]. The DOSY experiments can be performed in 1D mode, where two or more
1H NMR experiments are acquired for compounds possessing different diffusion coefficients or in 2D mode (2D DOSY), where the proton spectrum is depicted in one dimension and the diffusion coefficient is spread in an orthogonal dimension [
127]. Interestingly, the extension of 2D DOSY to 3D DOSY is gaining much popularity in metabolomics analysis via incorporating another proton dimension as in 3D COSY-DOSY and 3D TOCSY-DOSY, or another carbon dimension as in 3D DOSY-HSQC [
128]. Such DOSY methods have been recently applied in quality control and metabolic fingerprinting of commercial natural extracts [
128,
129]. Furthermore, they have also been used to identify novel metabolites from natural resources, for example, novel bromopyrrole alkaloids, and agesamides A and B were identified in the crude bromopyrrole fractions from Okinawan marine sponges using the 2D DOSY method [
130]. 2D
1H J-resolved NMR experiments represent another approach that simplify
1H NMR spectra where the overlapping resonance of the
1H dimension is resolved through representing the coupling constant of each signal (J value) in a second dimension [
131]. 2D
1H J-resolved NMR has been widely used in metabolomics classifications studies [
24,
132], however, prior knowledge of the chemical composition of the studied samples is required, which make this experiment of limited utility in the analysis of novel compound discovery [
108]. On the other hand, a newer approach called pure-shift
1H NMR spectroscopy has emerged in which better removal of homonuclear J couplings has been achieved [
133]. In this experiment, J coupling multiplicities are collapsed into single lines via implementing specific pulse sequences that result in a broadband homo-decoupled spectrum. This method has been applied recently to investigate the metabolome of
Physalis peruviana fruit aqueous extracts [
134]. The simplified data allowed for the identification of glutamic acid, a metabolite not observed in previous studies of the same extract due to the heavy overlap of its NMR signals.
The second strategy that can reliably handle
1H NMR overlapping peaks is the spreading of the crowded
1H NMR peaks in a second frequency dimension via magnetization transfer [
135]. This transfer could be between nuclei of the same type as in H–H correlated spectroscopy (COSY) and H–H total correlated spectroscopy (TOCSY), nuclei of different types such as heteronuclear single quantum coherence spectroscopy (HSQC) and heteronuclear multiple bond correlation (HMBC) experiments, or through space as in nuclear overhauser effect spectroscopy (NOESY) and rotating frame overhauser enhancement spectroscopy (ROESY) experiments [
135]. Full details of these common 2D NMR methods that have been used extensively in metabolomics analysis of natural products, and are thus beyond the scope of this review, but can be found elsewhere [
12,
108,
135,
136]. It is noteworthy to mention that spreading the crowded signals in indirect proton dimensions such as in COSY and TOCSY results in a relatively short acquisition time due to the high natural abundance of protons, however, some overlap might still exist as a region of only 10 ppm is employed. In contrast, using carbon as a second dimension such as in HSQC and HMBC will require a much longer acquisition time, but allow the resolution of the overlapping signals and the indirect dimension to be increased to more than 200 ppm units. Nevertheless, it should be noted that acquiring data using HSQC or HMBC can aid in the detection of less abundant compounds compared to the direct acquisition from the carbon channel as it is typical in
13C-NMR. It is worth mentioning that different pulse sequences have been developed for quantitative HSQC experiments. For example, a new phase modulated pulse sequence called the quantitative, offset-compensated, CPMG-adjusted HSQC (Q-OCCAHSQC) experiment has been proposed to enable the use of HSQC spectra for the quantitative determination of many primary metabolites [
137]. However, a combination of different 2D NMR experiments is favored for better analytical performance and enhanced identification strategies (e.g., COSY, (1H,13C)-HMQC, (1H,13C)-HMBC, and NOESY) were used to analyze the unfractionated defensive secretion of the walking stick
Parectatosoma mocquerysi identifying the novel monoterpene parectadial [
138]. COSY, HMQC, and HMBC have also aided in the identification of six terpenoids/sterol secondary metabolites in the crude extract of
Sarcophyton marine coral [
139]. Similarly, a combination of
1H,
1H COSY, HSQC, and HMBC have identified and quantified structurally related α- and β-bitter acids and their degradation products in hop resin [
140] using run times comparable to those used in HPLC. Interestingly, in the same study, HMBC spectra coupled with multivariate techniques were utilized as a novel approach for the classification of 13 commercial hop cultivars. Application of 2D-NMR metabolomics for drug discovery is considerably easier in cases where the secondary metabolites are naturally abundant as typical in the case of hop resin, which contains much lower levels of primary metabolites [
54]. In an interesting application of 2D NMR methods, differential analyses of NMR spectra (DANS) has been oriented for natural products drug discovery and comparative metabolomics. DANS uses a simple algorithm that graphically compares the 2D NMR spectra of different biological states to spot the peaks that are discriminatory among this set of spectra. For example, 2D double quantum filtered (DQF)-COSY was employed for DANS analysis of seven different culturing protocols of the filamentous fungus,
Tolypocladium cylindrosporum, leading to the detection and identification of two novel indole alkaloids, TC-705A and TC-705B, in the unfractionated extracts [
141]. Furthermore, DQF-COSY experiments have been applied for the identification of signaling molecules in the model organism
Caenorhabditis elegans [
142] and polyene antibiotic “bacillaene” from
Bacillus subtilis [
143]. On the other hand, multidimensional NMR experiments have already been used to increase the resolution of the newly developed benchtop NMR instruments [
144]. Benchtop NMR provides inexpensive and direct NMR access within any metabolomics laboratory, however, they suffer from decreased chemical shift dispersion and peaks overlapping as they are typically operated in the range of 40–80 MHz proton resonance frequency [
145]. Benchtop NMR maintain the same J-coupling frequency as the high field instruments since it is independent of the Larmor frequency, making it suitable for some metabolomics and quality control applications, particularly in 2D modes [
146].