High-Throughput Identification of Lipid Molecular Species in LC-MS/MS Data

Problem Description: LC-MS/MS data from complex lipid samples carries the potential to elucidate many structural features of lipids. It provides information about the fatty acids and in many cases about their regio- isomeric position. However, the MS/MS spectra of lipids can vary tremendously, because the fragmentation process depends on parameters like the used mass spectrometer, fragmentation collision energy, charge state, and adduct ions. Due to this diversity, a generally applicable bioinformatics tool for the automated analysis of lipidomics LC-MS/MS experiments is still missing. Project Aims: This project`s global aim is to develop a versatile and generally applicable method for high throughput determination of lipid structural fatty acid composition from LC-MS/MS data, easily adaptable to different mass spectrometers and experimental setups. The general applicability will be facilitated by a newly developed language for the description of MS/MS fragmentation spectra. Based on this language, a novel algorithm will identify the lipid and its deducible compositional features. The performance of the method will be verified in controlled and biological experiments. Furthermore, we want to supply a graphical user interface for the definition of rules describing the spectra, and supply pre defined rule sets for the most common mass spectrometers. Current State of Research: There exist a few tools for the analysis of lipidomics MS/MS data. However, they either do not exploit the advantages of liquid chromatography or are applicable to specific lipid classes and/or an experimental setup only. Furthermore, end users cannot customize the software for their specific fragmentation spectra. Potential Scientific Impact: LC MS/MS data analysis delivers in a single experiment the most sensitive information about the lipidome of a complex biological sample. A successful implementation of a generally applicable tool for the automated analysis of such data will accelerate lipid analysis enormously. This will give a more global picture on the lipidome and its changes which is ideally suited for top down approaches and investigation of biological questions. Methodology, Human Resources and Time Plan: The three year project will be coordinated by G. Thallinger, head of Bioinformatics at the IGB (TU Graz). A senior postdoctoral researcher (J. Hartler) will develop methods and software. MS experiments and the program usability tests will be conducted by M. Trötzmüller, a trained mass spectrometrist, whose work will be coordinated by H. Köfeler, head of the lipidomics MS core facility at the ZMF (MU Graz). Biological experiments will be performed by G. Hämmerle (IMB, KFU Graz). In order to test the method on different MS devices, we collaborate with M. Wakelem (Babraham Institute, Cambridge, UK) and G. Rechberger (IMB, KFU Graz).

No lipids, no life. In all organisms, lipids form cell walls, store energy and release it when necessary, and play an important role in cell signalling. It has been proved that changes in the composition of lipids play a causal role in illnesses such as cancer, fatty liver and multiple sclerosis. According to rough estimates, there are about 300,000 different lipid species. For the detection of lipids indicative for diseases, healthy and sick organisms are typically compared quantitatively. This comparison requires reliable and detailed information about the structure and composition of lipids from tissue samples and to this end a tool the Lipid Data Analyzer - was developed in the context of the project which was published in the renowned journal Nature Methods. Lipids often just called fats are complex substances which in addition to various other components predominantly consist of fatty acids. In lipid research, however, there are still many things unknown. Also, the detection of structural properties of lipid molecules in high- throughput profiling is still in its infancy. In the presented high-throughput method, a large number of samples are measured using mass spectrometry. These data (i.e., spectra) provide information for identification of the type and class of lipid or the type and position of the fatty acyl chains. However, the measured spectra can differ between one and the same lipid species, because lipids show different fragments in the spectra depending on the setup of the mass spectrometer and the ionization. Due to this spectral diversity, up to now there has been no universally applicable bioinformatics software for the automated detection of lipid structures. Fast and reliable details on the lipid composition of cell samples are a prerequisite for comparisons with reference samples from healthy cells which are required for the detection of biomarkers characteristic for diseases. Lipid Data Analyzer will facilitate work tremendously in biomedical research and definitely accelerate lipid research. The method which was developed at TU Graz in collaboration with colleagues from Med Uni Graz and Uni Graz, interprets lipid spectra using intuitive rule sets and can be as such flexibly accommodated to various fragmentation characteristics. This makes it possible for the first time to identify lipids at a very detailed structural level more precisely and reliably than with previous solutions. In the presented study, Lipid Data Analyzer detected more than 100 novel lipid species, which were previously unreported. The tool can be flexibly adapted and not just for new classes of lipids. It may be used, for instance, to characterise polysaccharides and glycolipids, i.e. lipids with attached sugars. Lipid Data Analyzer is provided as open source to the scientific community.