MS and NMR - the Perfect Couple?

From the search for new drugs to a better understanding of biological systems, the use of analytical tools for qualitative and quantitative characterization of samples is essential. Until very recently, groups that focus their efforts in such procedures used mass spectrometry (MS) or nuclear magnetic resonance (NMR) as separate tools. There has been much debate as to whether MS or NMR is most successful for sample characterization; undoubtedly, both have advantages and limitations.

On one hand, MS has high sensitivity and selectivity but more troublesome quantitation. On the other hand, NMR shows quantitative power and unprecedented reproducibility but poor sensitivity and selectivity. When it comes to structural information, both techniques enable the collection of additional orthogonal “dimensions” to help to gather distinctive information for compound identification. In MS, the selection of a certain ion for controlled fragmentation and detection of the product ions enable us to identify pieces of the structural puzzle for a compound. While in NMR, coupling-constant-based 2D experiments enable us to connect characteristic resonances and even link predefined pieces to find the correct structural formula for a compound. The limitation common to both techniques is how comprehensive the available databases are; MS boasts a richer database compared with NMR, but still has far to go. Considerable effort to increase experimental database size is being made by several groups, and others are developing approaches in database-independent compound identification.

In my field of natural products chemistry, dereplication of natural extracts is used as a screening procedure to collect qualitative data for sample selection in high-throughput drug discovery, most commonly using LCMS. Briefly, the raw extract will be analyzed using a data-dependent analysis LC-MS – a C18 LC column and an electrospray ionization source is a common setup. Once the data have been collected for a whole sample set, a general processing procedure and an extensive database comparison are applied. Bioactive compounds containing sample are then selected for in-depth studies.

In metabolomic studies, the analytical data of a carefully designed study are submitted to multivariate statistical analysis in search of biomarkers that characterize a certain distinction between groups. Once those biomarker features are assigned, the study is largely limited by whether we can establish their identity. Again, there is a preference among researchers to choose a single technique to collect their analytical data. A straightforward, untargeted metabolomics study using only MS data may miss quantitative information, potentially leading to an inconsistent hypothesis. But if only NMR data is used, that same metabolomics study may fail to detect low-concentration biomarkers, even when a targeted sample preparation is used. According to the Metabolomics Standards Initiative (MSI), putative identification using MS or NMR data is level 2 evidence; whereas coanalysis using authentic standards would be level 1.

An obvious solution is to use both MS and NMR, targeted and untargeted, as the source of analytical data for samples (and maybe even fractions) and interpret them all together as a single multiresponse vector in a multivariate scheme, allowing us to access very high sensitivity and direct quantitative information in the same study. The certainty of an MS-based putative identification is greatly increased once characteristic NMR features are confirmed, and vice versa.

However, every metabolomics study relies strongly upon databases for identification of their biomarkers and, in many cases, this is limited by inconsistent cataloging. I would like to make a plea for the community to routinely catalog identified compounds with all possible spectroscopic data for future access, especially for natural products chemistry, where we can detect dozens of compounds during dereplication procedures.

Large-scale cataloging is essential and attention in this area will drive progress in a broad range of areas within life science.