High-performance chemical isotope labeling (CIL) liquid chromatography-mass spectrometry (LC-MS) is an enabling technology based on rational design of labeling reagents to target a class of metabolites sharing the same functional group (e.g., all the amine-containing metabolites or the amine submetabolome) to provide concomitant improvements in metabolite separation, detection, and quantification. However, identification of labeled metabolites remains to be an analytical challenge. In this work, we describe a library of labeled standards and a search method for metabolite identification in CIL LC-MS. The current library consists of 273 unique metabolites, mainly amines and phenols that are individually labeled by dansylation (Dns). Some of them produced more than one Dns-derivative (isomers or multiple labeled products), resulting in a total of 315 dansyl compounds in the library. These metabolites cover 42 metabolic pathways, allowing the possibility of probing their changes in metabolomics studies. Each labeled metabolite contains three searchable parameters: molecular ion mass, MS/MS spectrum, and retention time (RT). To overcome RT variations caused by experimental conditions used, we have developed a calibration method to normalize RTs of labeled metabolites using a mixture of RT calibrants. A search program, DnsID, has been developed in www.MyCompoundID.org for automated identification of dansyl labeled metabolites in a sample based on matching one or more of the three parameters with those of the library standards. Using human urine as an example, we illustrate the workflow and analytical performance of this method for metabolite identification. This freely accessible resource is expandable by adding more amine and phenol standards in the future. In addition, the same strategy should be applicable for developing other labeled standards libraries to cover different classes of metabolites for comprehensive metabolomics using CIL LC-MS.

We report an analytical tool to facilitate metabolite identification based on an MS/MS spectral match of an unknown to a library of predicted MS/MS spectra of possible human metabolites. To construct the spectral library, the known endogenous human metabolites in the Human Metabolome Database (HMDB) (8,021 metabolites) and their predicted metabolic products via one metabolic reaction in the Evidence-based Metabolome Library (EML) (375,809 predicted metabolites) were subjected to in silico fragmentation to produce the predicted MS/MS spectra. This spectral library is hosted at the public MCID Web site ( www.MyCompoundID.org ), and a spectral search program, MCID MS/MS, has been developed to allow a user to search one or a batch of experimental MS/MS spectra against the library spectra for possible match(s). Using MS/MS spectra generated from standard metabolites and a human urine sample, we demonstrate that this tool is very useful for putative metabolite identification. It allows a user to narrow down many possible structures initially found by using an accurate mass search of an unknown metabolite to only one or a few candidates, thereby saving time and effort in selecting or synthesizing metabolite standard(s) for eventual positive metabolite identification.

No abstract