Accepted Manuscript Title: Lipidomics Informatics for Life-Science Authors: D. Schwudke, A. Shevchenko, N. Hoffmann, R. Ahrends PII: DOI: Reference:
S0168-1656(17)31596-1 http://dx.doi.org/doi:10.1016/j.jbiotec.2017.08.010 BIOTEC 7990
To appear in:
Journal of Biotechnology
Received date: Revised date: Accepted date:
17-2-2017 7-8-2017 9-8-2017
Please cite this article as: Schwudke, D., Shevchenko, A., Hoffmann, N., Ahrends, R., Lipidomics Informatics for Life-Science.Journal of Biotechnology http://dx.doi.org/10.1016/j.jbiotec.2017.08.010 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Lipidomics Informatics for Life-Science Schwudke D1, Shevchenko A2, Hoffmann N3, Ahrends R3 1 Research Center Borstel, Leibniz Center for Medicine and Biosciences, Borstel, Germany 2 Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany 3 Leibniz-Institut für Analytische Wissenschaften - ISAS - e.V., Dortmund, Germany
Highlights
Powerful and reliable tools for performing several computational steps within an analytics workflow for lipidomics experiments and lipidomic data interpretation. All services were designed to be as intuitively and user-friendly as possible and yet offer a broad functionality to close the gap between the developers and the end users.
Abstract Lipidomics encompasses analytical approaches that aim to identify and quantify the complete set of lipids, defined as lipidome in a given cell, tissue or organism as well as their interactions with other molecules. The majority of lipidomics workflows is based on mass spectrometry and has been proven as a powerful tool in system biology in concert with other Omics disciplines. Unfortunately, bioinformatics infrastructures for this relatively young discipline are limited only to some specialists. Search engines, quantification algorithms, visualization tools and databases developed by the ‘Lipidomics Informatics for Life-Science’ (LIFS) partners will be restructured and standardized to provide broad access to these specialized bioinformatics pipelines. There are many medical challenges related to lipid metabolic alterations that will be fostered by capacity building suggested by LIFS. LIFS as member of the ‘German Network for Bioinformatics’ (de.NBI) node for ‘Bioinformatics for Proteomics’ (BioInfra.Prot) and will provide access to the described software as well as to tutorials and consulting services via a unified web-portal.
1 Introduction The aim of lipidomics studies is to establish the identity, quantity and time dependent distribution of lipophilic and amphiphilic metabolites in biological systems (Klose et al., 2013; Wenk, 2005). 1
Lipids are involved in key biological mechanisms, and in recent years demands on analytical and informatics workflows have risen to study the influence of lipid metabolic regulation on the health status of organisms (Klose et al., 2012; Sampaio et al., 2011; Shevchenko and Simons, 2010). Lipidomics in concert with genomics, transcriptomics, and proteomics provides new avenues to study diseases within metabolic syndrome complex (Han, 2016), degenerative diseases (Wang and Han, 2016) and cancerogenesis (Beloribi-Djefaflia et al., 2016) to name just the most prominent fields. Unfortunately, search engines, quantification algorithms, visualization, validation and tools for lipidome comparisons exist but are neither streamlined, user friendly nor interconnected. Thus integration of a ‘Lipidomics Informatics for Life-Science’ unit (LIFS) into the ‘German Network for Bioinformatics Infrastructure’ (de.NBI) connected to the ‘Bioinformatics for Proteomics’ hub (BioInfra.Prot) will foster a system biology approaches for studying lipid metabolism. LIFS includes implementation, establishment and provision of bioinformatics services for lipidomics research within one webportal: i) We will provide our existing lipidomics software tools (LipidXplorer (Herzog et al., 2012; Herzog et al., 2011; Herzog et al., 2013), Skyline for Lipidomics (Peng and Ahrends, 2016), LUX Score (Marella et al., 2015), LipidHome (Foster et al., 2013)). ii) We will extend and integrate our tools in user-friendly web interfaces to offer them to a broader public. iii) We will offer bioinformatics consulting services regarding large scale data handling and managing and iv) we will organize workshops for practitioners and bioinformaticians on lipidomics tools and data analysis and participate in the de.NBI-wide education activities.
2 Lipidomics Software Tools
2
2.1 Skyline for lipidomics: a comprehensive platform for targeted assays Lipidomes comprise an extensive spectrum of chemical structures, which only mass spectrometry (MS)-based techniques provide the means to establish the identity and quantities of most lipids including sphingolipids (Bou Khalil et al., 2010; Lam and Shui, 2013; van Meer, 2005). For untargeted liquid chromatography (LC)-based lipidomics, software solutions such as LipidSearch (Taguchi et al., 2007) and LipidBlast (Kind et al., 2013) are available, and for shotgun lipidomics, direct infusion experiments, the software suite LipidXplorer can be utilized (Herzog et al., 2012). LipidSearch and LipidBlast depend on spectral libraries, whereas LipidXplorer uses de novo spectra interpretation for lipid identification. Besides this, other identification software packages, including LipidQA (Song et al., 2007), LIMSA (Haimi et al., 2006), FAAT (Leavell and Leary, 2006), lipid (Hubner et al., 2009), LipidInspector (Schwudke et al., 2006b), ALEX (Husen et al., 2013) and Greasy (Kochen et al., 2016) can be applied for lipid identification. Unfortunately, there are currently no open source tools for targeted lipidomics. In response to the urgent need for an analysis software that is capable of handling data from targeted high-throughput lipidomics experiments, we developed a workflow for straightforward method design and analysis of selected and parallel reaction monitoring data. We used the Skyline platform, primarily designed for proteomics applications (MacLean et al., 2010), as a powerful basis to design a specific pipeline for lipid research (Peng and Ahrends, 2016). This extension offers the unique capability to assemble targeted mass spectrometry methods for complex lipids by making use of lipid fragmentation building blocks (Figure 1A). With simple yet tailored
modifications,
targeted
methods
to
analyze
main
lipid
classes
such
as
glycerophospholipids, sphingolipids, glycerolipids, cholesteryl-esters, and cholesterol can be quickly introduced into Skyline for easy application by end users without distinct bioinformatics skills. During the de.NBI funding period we will adapt the interim version, which is still working 3
with amino acid based pseudo sequence tags to a stand-alone version (LipidCreator) which can be used as a plugin in Skyline (Figure 1B). We will implement organism based pre-calculation for lipid species making it convenient for the user to obtain the transitions and target masses of interest. For a better reviewing process and to further accelerate the understanding of lipid fragmentation we will propose a lipid nomenclature connecting the MS1 with the MS2 fragment ion level, by introducing a standardized MS2 nomenclature. This will further help define molecular lipid species-specific fragments that provide information about the chemical composition of the fatty acyl chain of individual lipid molecules, such as long chain bases (LCBs) or fatty acids linked by ester, ether or vinyl-ether bond to their individual lipid backbone. We will seek feedback from and input by the scientific community, particularly from the LIPID MAPS consortium (Fahy et al., 2007; Sud et al., 2007) and HUPO PSI (Kaiser, 2002) initiative on the proposed nomenclature. In summary, by the end of the funding period we will present a user-friendly “Plug and Play” workflow for lipidomics in response to the demand of a dedicated high throughput targeted software platform. This native cross-vendor tool for targeted lipidomics will allow to (i) create transitions and assays, (ii) optimize collision energies, (iii) visually review the obtained results, and (iv) quantify the lipids of interest. We envision that the lipid building block based Skyline application breaks ground not only for optimized method design, analysis, and data evaluation in targeted lipidomics, but also provide a gateway to spectral libraries and the sharing of experimental data. This will help the research community to build up a comprehensive and vendor independent exchange platform to improve reproducibility and validation processes for lipidomics data.
4
2.2 LipidXplorer: a generic software tool for shotgun lipidomics By definition, shotgun lipidomics relies on a direct infusion of total lipid extracts into a tandem mass spectrometer, however lipids can be identified in many ways. Historically, shotgun lipidomics was mainly associated with triple quadrupole mass spectrometers (reviewed in (Han et al., 2012)). Lipids were identified by a combination of precursor and neutral loss scans and, because of their low mass resolution, never relied on accurate masses of fragment precursors. In contrast, tandem mass spectrometers of the Orbitrap family and modern q-TOF deliver high mass resolution spectra along with low-ppm accuracy. Therefore masses of detected molecules can be associated with their elemental compositions and lipids consistently identified in the series of compositionally related samples by mapping their intact masses (Schwudke et al., 2007; Schwudke et al., 2011). The prerequisites for assigning correct elemental compositions of lipid precursors and fragments are still under debate (Bielow et al., 2017). Experimentally proven improvements for lipid identification were observed when the mass difference for certain elemental compositions can be resolved. The differentiation of 1-between alkyl-2-acyl glycerophospholipid and 1,2-diacyl glycerophospholipid can be achieved when the mass difference of 36.4 mDa is resolved. Furthermore, it is preferable to resolve the overlap of the second isotopic peak and lipid with exactly one less double bond which results in a mass difference of 9.0 mDa (Herzog et al., 2011; Schwudke et al., 2007). Specifically, for shotgun lipidomics, where the separation power is solely depending on the chosen MS instrument platform false assignments of major abundant glycerophospholipids can be minimized. From these estimates, one can state that the specificity of shotgun lipidomics increases for resolution above 100,000 (FWHM at m/z 750) for MS1 and 30,000 (FWHM at m/z 200) for MS2. High acquisition rate and sensitivity of any modern hybrid tandem mass spectrometers enable fragmentation of all candidate peaks independent of their intensities and produce a comprehensive dataset that comprise masses of all detectable precursors and all fragments 5
generated by their collisional fragmentation (Schwudke et al., 2006a; Schwudke et al., 2011). Therefore, shotgun lipidomics software should support any data interpretation routine independent of instrumentation platform, and methods of collisional fragmentation. It should also be able to target any lipid class and species. To this end, we developed a novel concept for the interpretation of large collections of shotgun spectra that does not rely on fixed identification rules of pre-compiled databases of masses and fragments (Herzog et al., 2011) (Herzog et al., 2012; Herzog et al., 2013). Instead, our LipidXplorer software (Figure 2) uses a declarative Molecular Fragmentation Query Language that can describe any user-defined lipid identification routine in simple and intuitive terms. In this way, the same software supports high resolution mass mapping in top-down (Schwudke et al., 2007) and bottom-up (Schuhmann et al., 2011) screens; precursor and neutral loss scanning on low resolution triple quadrupole mass spectrometers (Herzog et al., 2012) as well as combined interpretation routines such as multiple precursor-, neutral loss and boolean scans enabled by data-dependent acquisition (Herzog et al., 2013). LipidXplorer can also identify novel lipids that bear some structural similarity to known lipids (Papan et al., 2014). However, the significant limitation of shotgun datasets is their inherent complexity. In a typical workflow after the extraction with organic solvents and MS analysis a dataset acquired in a single experiment might consist of several hundreds of MS and MS/MS spectra and comprise > 100 000 of unique peaks, however only a few hundred of those might be eventually attributed to lipids (Herzog et al., 2011). Although the software design supports many features valued by experienced users, it lacks the transparency and ease of use important for laboratories entering the field and ready-to-use automation tools that further enable and support routine lipidomics measurements.
6
We therefore will extend LipidXplorer capabilities with data processing modules, which will greatly simplify the interpretation of shotgun datasets while maintaining the key features of LipidXplorer. We will also develop a web-version of LipidXplorer that will offer a generic support for routine shotgun experiments, which will be of interest to a broad community of users willing to apply lipidomics in their focused biological projects. Together with other members of the LIFS consortium, we will integrate LipidXplorer with LipidHome, Skyline and LUX tools.
2.3 LipidHome: a curation model for experimental data LipidHome aims to close a gap between theoretically calculated lipid molecules and metadata by implementing a resource comparable in function to UniProt, (Foster et al., 2013). The database was initially populated with lipids that were generated in-silico based on community-accepted rules concerning chemical and structural properties. In addition, a web interface was created to present the information and provide computational access. Tailored specifically to handle high throughput mass spectrometry based approaches, lipids are organized in the database in a hierarchical manner. The level of structural resolution provided by the applied lipidomics method is further implemented by using the short hand lipid annotation that was recently introduced by Liebisch et al. (Liebisch et al., 2013). To reflect the level of confidence of reported lipids in the database we will validate lipid entries at three levels: level C will include all lipids that were already described in literature, level B will contain lipids that were measured and identified by high resolution mass spectrometer at the MS and/or MS2 level and the A level will contain lipids quantified via an internal standard to provide absolute quantities. Further information will include the source organism and tissue where a particular lipid has been observed. Cross-references to other lipid related resources and scientific articles that cite specific lipids will be reported using a specifically improved ontology for lipids. The web application will be directly linked to the LIFS-platform and allow a matching of MS1 and MS2 7
derived lipidomics data. Data submission will be based on the Metabolights database, hosted by the European Bioinformatics Institute (EBI), which provides unique and stable IDs for each submitted dataset as well as data validation and annotation steps based on a standardized submission format (Haug et al., 2013). The alpha version of LipidHome is currently accessible at https://www.ebi.ac.uk/metabolights/lipidhome/.
2.4 Lipid Homology – LUX Score The concept of lipid homology was recently established to provide a rigorous metric comparable to genetic analyses based on sequences (Marella et al., 2015). From this point of view, a lipidome is the set of lipids that reflect not only biosynthetic capacity of an organism but also functional aspects for a certain cell or cell compartment. Recent studies on lipidomes of biological model organisms (Bozek et al., 2015; Carvalho et al., 2012; Ejsing et al., 2009; Klose et al., 2012; Watschinger et al., 2015) and clinical relevant human samples (Chua et al., 2013; Graessler et al., 2009; Sales et al., 2016; Slatter et al., 2016) emphasize that lipidomes give insight into health status and provide molecular fingerprints similar to expression profiles. From this perspective, we developed a homology metric, the ‘Lipidome jUXtaposition (LUX) score’ that can quantify systematic differences in the composition of a lipidome (Figure 3). In a first step, the structural similarity between all lipids is determined. For that, all identified lipids are converted into SMILES (Weininger, 1988) taking into account all possible isomeric structures that are not distinguishable by the chosen mass spectrometric approach (Liebisch et al., 2013). To date, only a limited number of approaches were developed to determine the connectivity of fatty acyl chains and double bond positions of phosphoglycerolipids. Ozone-induced dissociation produces diagnostic fragments for the position of unsaturation within the precursor ion helping to characterize the natural structural complexity of lipids (Thomas et al., 2008). Furthermore, fragmentation techniques based on radical interaction can potentially enrich lipidomics datasets by definite information on the double bond position (Pham et al., 2012). 8
We showed that Template-based SMILES, in which chemical nomenclature is the basis for unique representation of lipids, can correctly represent the structural diversity of isomeric lipids. Specifically, position of double bonds and hydroxyl groups and elongation of aliphatic chains are accurately documented. On basis of Template-based SMILES, a chemical space model of a lipidome can be computed based on pair wise Levenshtein distances (Levenshtein, 1966). With this approach, a well-defined metric space is computed and an intuitive visualization of the lipidome composition is achieved. Several lipidomes can be projected into the same chemical space model, in which the degree of overlap between lipidomes can be determined. The Hausdorff distance (Hausdorff, 2005) was used to measures the difference of two lipidomes in the chemical space model, which basically embodies the LUX score. We tested the LUX score on four yeast strains with known genetic alteration in fatty acid synthesis (Ejsing et al., 2009) and showed that the underlying genetic relationship and growth temperature are better represented than correlation based clustering based on lipid concentrations. Next, we applied this metric for lipdiomics screen (Schwudke et al., 2007; Schwudke et al., 2011) data of larval tissue lipidomes of Drosophila (Carvalho et al., 2012). This showed that the LUX score is sufficient to determine the impact of nutritional changes in an unbiased manner, despite the limited information on the structural diversity of each lipidome. A software package was developed that covers all necessary steps to determine lipidome homology using the LUX score (http://lifs.isas.de/index.php/lux-score). However, improvements for the general usability, import of result files and integration into the LipidHome database are required to make the lipidome homology concept beneficial for the research community. We will integrate the determination of Template-SMILES into the LipidHome database as a look-up service and use the LUX Score software as a backend for homology calculation between lipidomes. Furthermore, a seamless integration into the reporting structure and lipid naming convention of LipidXplorer and Skyline will be a further focus of software implementation. In view 9
of the growing demand to establish functional association of lipids, the LUX score will foster comparative lipidome studies that can bridge species borders, which is specifically important for translational research and application of disease models.
3 Summary The LIFS consortia will consolidate a number of software for lipidomics that were separately developed. By integration of the shotgun lipidomics software LipidXplorer, the targeted analysis tool Skyline for Lipdomics/LipidCreator, the database LipidHome and the LUX score approach for systematic lipidome comparison, we aim to provide a platform independent Open Source software suite that will foster lipid research (Figure 4). It is of high importance to establish a new quality on how results are communicated in lipidomics through this resource and to improve the transparency of computational methods used for lipidomics data analysis. For that, the tool integration will be one of the crucial tasks for the LIFS consortia to broaden the user and developer base. As a member of BioInfra.Prot and together with a community driven course program, tutorials and computational services, we envisage that LIFS will be a driving force for future lipidomics research.
Declaration of interest This work was supported by the Ministerium für Innovation, Wissenschaft und Forschung des Landes Nordrhein-Westfalen, the Senatsverwaltung für Wirtschaft, Technologie und Forschung des Landes Berlin, and the Bundesministerium für Bildung und Forschung, SUPR-G e:Med (Code 01ZX1401C) and de.NBI program (code 031L0108A, 031L0108B, 031L0108C). The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.
10
References
Beloribi-Djefaflia, S., Vasseur, S., Guillaumond, F., (2016) Lipid metabolic reprogramming in cancer cells. Oncogenesis 5, e189. Bielow, C., Mastrobuoni, G., Orioli, M., Kempa, S., (2017) On Mass Ambiguities in High-Resolution Shotgun Lipidomics. Anal Chem 89, 2986-2994. Bou Khalil, M., Hou, W., Zhou, H., Elisma, F., Swayne, L.A., Blanchard, A.P., Yao, Z., Bennett, S.A., Figeys, D., (2010) Lipidomics era: accomplishments and challenges. Mass Spectrom Rev 29, 877-929. Bozek, K., Wei, Y., Yan, Z., Liu, X., Xiong, J., Sugimoto, M., Tomita, M., Paabo, S., Sherwood, C.C., Hof, P.R., Ely, J.J., Li, Y., Steinhauser, D., Willmitzer, L., Giavalisco, P., Khaitovich, P., (2015) Organization and evolution of brain lipidome revealed by large-scale analysis of human, chimpanzee, macaque, and mouse tissues. Neuron 85, 695-702. Carvalho, M., Sampaio, J.L., Palm, W., Brankatschk, M., Eaton, S., Shevchenko, A., (2012) Effects of diet and development on the Drosophila lipidome. Mol Syst Biol 8, 600. Chua, E.C., Shui, G., Lee, I.T., Lau, P., Tan, L.C., Yeo, S.C., Lam, B.D., Bulchand, S., Summers, S.A., Puvanendran, K., Rozen, S.G., Wenk, M.R., Gooley, J.J., (2013) Extensive diversity in circadian regulation of plasma lipids and evidence for different circadian metabolic phenotypes in humans. Proc Natl Acad Sci U S A 110, 14468-14473. Ejsing, C.S., Sampaio, J.L., Surendranath, V., Duchoslav, E., Ekroos, K., Klemm, R.W., Simons, K., Shevchenko, A., (2009) Global analysis of the yeast lipidome by quantitative shotgun mass spectrometry. Proc Natl Acad Sci U S A 106, 2136-2141. Fahy, E., Sud, M., Cotter, D., Subramaniam, S., (2007) LIPID MAPS online tools for lipid research. Nucleic Acids Res 35, W606-612. Foster, J.M., Moreno, P., Fabregat, A., Hermjakob, H., Steinbeck, C., Apweiler, R., Wakelam, M.J., Vizcaino, J.A., (2013) LipidHome: a database of theoretical lipids optimized for high throughput mass spectrometry lipidomics. PLoS One 8, e61951. Graessler, J., Schwudke, D., Schwarz, P.E., Herzog, R., Shevchenko, A., Bornstein, S.R., (2009) Top-down lipidomics reveals ether lipid deficiency in blood plasma of hypertensive patients. PLoS One 4, e6261. Haimi, P., Uphoff, A., Hermansson, M., Somerharju, P., (2006) Software Tools for Analysis of Mass Spectrometric Lipidome Data. Anal. Chem. 78, 8324-8331. Han, X., (2016) Lipidomics for studying metabolism. Nat Rev Endocrinol 12, 668-679.
11
Han, X., Yang, K., Gross, R.W., (2012) Multi-dimensional mass spectrometry-based shotgun lipidomics and novel strategies for lipidomic analyses. Mass Spectrom Rev 31, 134-178. Haug, K., Salek, R.M., Conesa, P., Hastings, J., de Matos, P., Rijnbeek, M., Mahendraker, T., Williams, M., Neumann, S., Rocca-Serra, P., Maguire, E., González-Beltrán, A., Sansone, S.-A., Griffin, J.L., Steinbeck, C., (2013) MetaboLights—an open-access general-purpose repository for metabolomics studies and associated meta-data. Nucleic Acids Research 41, D781-D786. Hausdorff, F., (2005) Set Theory, 10 ed. AMS Chelsea Pub., American Mathematical Society. Herzog, R., Schuhmann, K., Schwudke, D., Sampaio, J.L., Bornstein, S.R., Schroeder, M., Shevchenko, A., (2012) LipidXplorer: a software for consensual cross-platform lipidomics. PLoS One 7, e29851. Herzog, R., Schwudke, D., Schuhmann, K., Sampaio, J.L., Bornstein, S.R., Schroeder, M., Shevchenko, A., (2011) A novel informatics concept for high-throughput shotgun lipidomics based on the molecular fragmentation query language. Genome Biol 12, R8. Herzog, R., Schwudke, D., Shevchenko, A., (2013) LipidXplorer: Software for Quantitative Shotgun Lipidomics Compatible with Multiple Mass Spectrometry Platforms. Curr Protoc Bioinformatics 43, 14 12 11-30. Hubner, G., Crone, C., Lindner, B., (2009) lipID--a software tool for automated assignment of lipids in mass spectra. J Mass Spectrom 44, 1676-1683. Husen, P., Tarasov, K., Katafiasz, M., Sokol, E., Vogt, J., Baumgart, J., Nitsch, R., Ekroos, K., Ejsing, C.S., (2013) Analysis of lipid experiments (ALEX): a software framework for analysis of high-resolution shotgun lipidomics data. PLoS One 8, e79736. Kaiser, J., (2002) Proteomics. Public-private group maps out initiatives. Science 296, 827. Kind, T., Liu, K.H., Lee do, Y., DeFelice, B., Meissen, J.K., Fiehn, O., (2013) LipidBlast in silico tandem mass spectrometry database for lipid identification. Nat Methods 10, 755-758. Klose, C., Surma, M.A., Gerl, M.J., Meyenhofer, F., Shevchenko, A., Simons, K., (2012) Flexibility of a eukaryotic lipidome--insights from yeast lipidomics. PLoS One 7, e35063. Klose, C., Surma, M.A., Simons, K., (2013) Organellar lipidomics--background and perspectives. Curr Opin Cell Biol 25, 406-413. Kochen, M.A., Chambers, M.C., Holman, J.D., Nesvizhskii, A.I., Weintraub, S.T., Belisle, J.T., Islam, M.N., Griss, J., Tabb, D.L., (2016) Greazy: Open-Source Software for Automated Phospholipid Tandem Mass Spectrometry Identification. Anal Chem 88, 5733-5741. Lam, S.M., Shui, G., (2013) Lipidomics as a principal tool for advancing biomedical research. J Genet Genomics 40, 375-390. Leavell, M.D., Leary, J.A., (2006) Fatty Acid Analysis Tool (FAAT): An FT-ICR MS Lipid Analysis Algorithm. Anal. Chem. 78, 5497–5503. 12
Levenshtein, V.I., (1966) Binary codes capable of correcting deletions, insertions and reversals. Soviet Physics Doklady 10, 707--710. Liebisch, G., Vizcaino, J.A., Kofeler, H., Trotzmuller, M., Griffiths, W.J., Schmitz, G., Spener, F., Wakelam, M.J., (2013) Shorthand notation for lipid structures derived from mass spectrometry. J Lipid Res 54, 1523-1530. MacLean, B., Tomazela, D.M., Shulman, N., Chambers, M., Finney, G.L., Frewen, B., Kern, R., Tabb, D.L., Liebler, D.C., MacCoss, M.J., (2010) Skyline: an open source document editor for creating and analyzing targeted proteomics experiments. Bioinformatics 26, 966-968. Marella, C., Torda, A.E., Schwudke, D., (2015) The LUX Score: A Metric for Lipidome Homology. PLoS Comput Biol 11, e1004511. Papan, C., Penkov, S., Herzog, R., Thiele, C., Kurzchalia, T., Shevchenko, A., (2014) Systematic screening for novel lipids by shotgun lipidomics. Anal Chem 86, 2703-2710. Peng, B., Ahrends, R., (2016) Adaptation of Skyline for Targeted Lipidomics. J Proteome Res 15, 291-301. Pham, H.T., Ly, T., Trevitt, A.J., Mitchell, T.W., Blanksby, S.J., (2012) Differentiation of complex lipid isomers by radical-directed dissociation mass spectrometry. Anal Chem 84, 7525-7532. Sales, S., Graessler, J., Ciucci, S., Al-Atrib, R., Vihervaara, T., Schuhmann, K., Kauhanen, D., Sysi-Aho, M., Bornstein, S.R., Bickle, M., Cannistraci, C.V., Ekroos, K., Shevchenko, A., (2016) Gender, Contraceptives and Individual Metabolic Predisposition Shape a Healthy Plasma Lipidome. Sci Rep 6, 27710. Sampaio, J.L., Gerl, M.J., Klose, C., Ejsing, C.S., Beug, H., Simons, K., Shevchenko, A., (2011) Membrane lipidome of an epithelial cell line. Proc Natl Acad Sci U S A 108, 1903-1907. Schuhmann, K., Herzog, R., Schwudke, D., Metelmann-Strupat, W., Bornstein, S.R., Shevchenko, A., (2011) Bottom-up shotgun lipidomics by higher energy collisional dissociation on LTQ Orbitrap mass spectrometers. Anal Chem 83, 5480-5487. Schwudke, D., Hannich, J.T., Surendranath, V., Grimard, V., Moehring, T., Burton, L., Kurzchalia, T., Shevchenko, A., (2007) Top-down lipidomic screens by multivariate analysis of high-resolution survey mass spectra. Anal Chem 79, 4083-4093. Schwudke, D., Oegema, J., Burton, L., Entchev, E., Hannich, J.T., Ejsing, C.S., Kurzchalia, T., Shevchenko, A., (2006a) Lipid profiling by multiple precursor and neutral loss scanning driven by the data-dependent acquisition. Anal Chem 78, 585-595. Schwudke, D., Oegema, J., Burton, L., Entchev, E., Hannich, J.T., Ejsing, C.S., Kurzchalia, T., Shevchenko, A., (2006b) Lipid Profiling by Multiple Precursor and Neutral Loss Scanning Driven by the DataDependent Acquisition. Anal. Chem. 78, 585–595. Schwudke, D., Schuhmann, K., Herzog, R., Bornstein, S.R., Shevchenko, A., (2011) Shotgun lipidomics on high resolution mass spectrometers. Cold Spring Harb Perspect Biol 3, a004614. 13
Shevchenko, A., Simons, K., (2010) Lipidomics: coming to grips with lipid diversity. Nat Rev Mol Cell Biol 11, 593-598. Slatter, D.A., Aldrovandi, M., O'Connor, A., Allen, S.M., Brasher, C.J., Murphy, R.C., Mecklemann, S., Ravi, S., Darley-Usmar, V., O'Donnell, V.B., (2016) Mapping the Human Platelet Lipidome Reveals Cytosolic Phospholipase A2 as a Regulator of Mitochondrial Bioenergetics during Activation. Cell Metab 23, 930944. Song, H., Hsu, F.F., Ladenson, J., Turk, J., (2007) Algorithm for processing raw mass spectrometric data to identify and quantitate complex lipid molecular species in mixtures by data-dependent scanning and fragment ion database searching. J Am Soc Mass Spectrom 18, 1848-1858. Sud, M., Fahy, E., Cotter, D., Brown, A., Dennis, E.A., Glass, C.K., Merrill, A.H., Jr., Murphy, R.C., Raetz, C.R., Russell, D.W., Subramaniam, S., (2007) LMSD: LIPID MAPS structure database. Nucleic Acids Res 35, D527-532. Taguchi, R., Nishijima, M., Shimizu, T., (2007) Basic Analytical Systems for Lipidomics by Mass Spectrometry in Japan. Methods in Enzymology 432, 185–211. Thomas, M.C., Mitchell, T.W., Harman, D.G., Deeley, J.M., Nealon, J.R., Blanksby, S.J., (2008) Ozoneinduced dissociation: elucidation of double bond position within mass-selected lipid ions. Anal Chem 80, 303-311. van Meer, G., (2005) Cellular lipidomics. The EMBO Journal 24, 3159-3165. Wang, M., Han, X., (2016) Advanced Shotgun Lipidomics for Characterization of Altered Lipid Patterns in Neurodegenerative Diseases and Brain Injury. Methods Mol Biol 1303, 405-422. Watschinger, K., Keller, M.A., McNeill, E., Alam, M.T., Lai, S., Sailer, S., Rauch, V., Patel, J., Hermetter, A., Golderer, G., Geley, S., Werner-Felmayer, G., Plumb, R.S., Astarita, G., Ralser, M., Channon, K.M., Werner, E.R., (2015) Tetrahydrobiopterin and alkylglycerol monooxygenase substantially alter the murine macrophage lipidome. Proc Natl Acad Sci U S A 112, 2431-2436. Weininger, D., (1988) SMILES, a chemical language and information system. J. Chem. Inf. Comput. Sci., 31-36. Wenk, M.R., (2005) The emerging field of lipidomics. Nat Rev Drug Discov 4, 594-610.
14
Figure legends Figure 1 Skyline for lipidomics: a comprehensive platform for targeted assays. A) Main lipid classes can be assembled from lipid building blocks such as fatty acid chain, long chain base and head group to create the precursor mass their fragment ions. B) Utilization of LipidCreator in the Skyline environment. Lipid information is transferred to Skyline by LipidCreator to compute transition lists and in-silico spectral libraries. Afterwards, Skyline can be used for visualization, optimization, data evaluation and quantification
of
targeted
lipidomics
results.
Abbreviations:
PG
(Glycerophosphoglycerol), Cer (Ceramide), DG (Diradylglycerol), TG (Triradylglycerol), CL (Cardiolipin).
Figure 2 LipidXplorer, a generic software tool for shotgun lipidomics. LipidXplorer organizes MS and MS/MS spectra (acquired from all samples of lipidomics experiments, including biological and technical repeats) into a flat-file database termed MasterScan. Scans are averaged and individual spectra aligned considering instrument specific peak attributes, such as mass resolution, mass accuracy, and peak occupancy, among others. The MasterScan is then interrogated by user-defined lipid class –specific and/or lipid species -specific queries written in the Molecular Fragmentation Query Language (MFQL). Identified and annotated lipid species, along with intensities of user-defined fragment or precursor ions, are reported. Figure 3 Lipid homology determination using LUX Score. Result files of LIFS applications LipidXplorer and Skyline/LipidCreator are directly used as input. Other 15
result files can be imported containing lipid names and quantities. B) Template SMILES are received via LipidHome. When Gene Ontology information are provided lipids are scored according to proven repositories for organism, tissues and cell types. C) For all lipids of all lipidomes Template SMILES are generated and used as model to compute chemical space model. D) From reference lipidome maps the pairwise LUX Score are calculated. Figure 4 LIFS web portal implementation. Service requests and input data are processed at three levels. 1) Lipid Identification and design of lipidomics experiments using LipidXplorer and Skyline/LipidCreator. 2) Evaluation of Lipid IDs in context of Gene Ontology and data repositories using LipidHome. 3) Lipidomes are compared using LUX Score for functional associations.
16
Figure 1)
Figure 2)
17
Figure 3)
Figure 4
18
S0168-1656(17)31596-1 http://dx.doi.org/doi:10.1016/j.jbiotec.2017.08.010 BIOTEC 7990
To appear in:
Journal of Biotechnology
Received date: Revised date: Accepted date:
17-2-2017 7-8-2017 9-8-2017
Please cite this article as: Schwudke, D., Shevchenko, A., Hoffmann, N., Ahrends, R., Lipidomics Informatics for Life-Science.Journal of Biotechnology http://dx.doi.org/10.1016/j.jbiotec.2017.08.010 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Lipidomics Informatics for Life-Science Schwudke D1, Shevchenko A2, Hoffmann N3, Ahrends R3 1 Research Center Borstel, Leibniz Center for Medicine and Biosciences, Borstel, Germany 2 Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany 3 Leibniz-Institut für Analytische Wissenschaften - ISAS - e.V., Dortmund, Germany
Highlights
Powerful and reliable tools for performing several computational steps within an analytics workflow for lipidomics experiments and lipidomic data interpretation. All services were designed to be as intuitively and user-friendly as possible and yet offer a broad functionality to close the gap between the developers and the end users.
Abstract Lipidomics encompasses analytical approaches that aim to identify and quantify the complete set of lipids, defined as lipidome in a given cell, tissue or organism as well as their interactions with other molecules. The majority of lipidomics workflows is based on mass spectrometry and has been proven as a powerful tool in system biology in concert with other Omics disciplines. Unfortunately, bioinformatics infrastructures for this relatively young discipline are limited only to some specialists. Search engines, quantification algorithms, visualization tools and databases developed by the ‘Lipidomics Informatics for Life-Science’ (LIFS) partners will be restructured and standardized to provide broad access to these specialized bioinformatics pipelines. There are many medical challenges related to lipid metabolic alterations that will be fostered by capacity building suggested by LIFS. LIFS as member of the ‘German Network for Bioinformatics’ (de.NBI) node for ‘Bioinformatics for Proteomics’ (BioInfra.Prot) and will provide access to the described software as well as to tutorials and consulting services via a unified web-portal.
1 Introduction The aim of lipidomics studies is to establish the identity, quantity and time dependent distribution of lipophilic and amphiphilic metabolites in biological systems (Klose et al., 2013; Wenk, 2005). 1
Lipids are involved in key biological mechanisms, and in recent years demands on analytical and informatics workflows have risen to study the influence of lipid metabolic regulation on the health status of organisms (Klose et al., 2012; Sampaio et al., 2011; Shevchenko and Simons, 2010). Lipidomics in concert with genomics, transcriptomics, and proteomics provides new avenues to study diseases within metabolic syndrome complex (Han, 2016), degenerative diseases (Wang and Han, 2016) and cancerogenesis (Beloribi-Djefaflia et al., 2016) to name just the most prominent fields. Unfortunately, search engines, quantification algorithms, visualization, validation and tools for lipidome comparisons exist but are neither streamlined, user friendly nor interconnected. Thus integration of a ‘Lipidomics Informatics for Life-Science’ unit (LIFS) into the ‘German Network for Bioinformatics Infrastructure’ (de.NBI) connected to the ‘Bioinformatics for Proteomics’ hub (BioInfra.Prot) will foster a system biology approaches for studying lipid metabolism. LIFS includes implementation, establishment and provision of bioinformatics services for lipidomics research within one webportal: i) We will provide our existing lipidomics software tools (LipidXplorer (Herzog et al., 2012; Herzog et al., 2011; Herzog et al., 2013), Skyline for Lipidomics (Peng and Ahrends, 2016), LUX Score (Marella et al., 2015), LipidHome (Foster et al., 2013)). ii) We will extend and integrate our tools in user-friendly web interfaces to offer them to a broader public. iii) We will offer bioinformatics consulting services regarding large scale data handling and managing and iv) we will organize workshops for practitioners and bioinformaticians on lipidomics tools and data analysis and participate in the de.NBI-wide education activities.
2 Lipidomics Software Tools
2
2.1 Skyline for lipidomics: a comprehensive platform for targeted assays Lipidomes comprise an extensive spectrum of chemical structures, which only mass spectrometry (MS)-based techniques provide the means to establish the identity and quantities of most lipids including sphingolipids (Bou Khalil et al., 2010; Lam and Shui, 2013; van Meer, 2005). For untargeted liquid chromatography (LC)-based lipidomics, software solutions such as LipidSearch (Taguchi et al., 2007) and LipidBlast (Kind et al., 2013) are available, and for shotgun lipidomics, direct infusion experiments, the software suite LipidXplorer can be utilized (Herzog et al., 2012). LipidSearch and LipidBlast depend on spectral libraries, whereas LipidXplorer uses de novo spectra interpretation for lipid identification. Besides this, other identification software packages, including LipidQA (Song et al., 2007), LIMSA (Haimi et al., 2006), FAAT (Leavell and Leary, 2006), lipid (Hubner et al., 2009), LipidInspector (Schwudke et al., 2006b), ALEX (Husen et al., 2013) and Greasy (Kochen et al., 2016) can be applied for lipid identification. Unfortunately, there are currently no open source tools for targeted lipidomics. In response to the urgent need for an analysis software that is capable of handling data from targeted high-throughput lipidomics experiments, we developed a workflow for straightforward method design and analysis of selected and parallel reaction monitoring data. We used the Skyline platform, primarily designed for proteomics applications (MacLean et al., 2010), as a powerful basis to design a specific pipeline for lipid research (Peng and Ahrends, 2016). This extension offers the unique capability to assemble targeted mass spectrometry methods for complex lipids by making use of lipid fragmentation building blocks (Figure 1A). With simple yet tailored
modifications,
targeted
methods
to
analyze
main
lipid
classes
such
as
glycerophospholipids, sphingolipids, glycerolipids, cholesteryl-esters, and cholesterol can be quickly introduced into Skyline for easy application by end users without distinct bioinformatics skills. During the de.NBI funding period we will adapt the interim version, which is still working 3
with amino acid based pseudo sequence tags to a stand-alone version (LipidCreator) which can be used as a plugin in Skyline (Figure 1B). We will implement organism based pre-calculation for lipid species making it convenient for the user to obtain the transitions and target masses of interest. For a better reviewing process and to further accelerate the understanding of lipid fragmentation we will propose a lipid nomenclature connecting the MS1 with the MS2 fragment ion level, by introducing a standardized MS2 nomenclature. This will further help define molecular lipid species-specific fragments that provide information about the chemical composition of the fatty acyl chain of individual lipid molecules, such as long chain bases (LCBs) or fatty acids linked by ester, ether or vinyl-ether bond to their individual lipid backbone. We will seek feedback from and input by the scientific community, particularly from the LIPID MAPS consortium (Fahy et al., 2007; Sud et al., 2007) and HUPO PSI (Kaiser, 2002) initiative on the proposed nomenclature. In summary, by the end of the funding period we will present a user-friendly “Plug and Play” workflow for lipidomics in response to the demand of a dedicated high throughput targeted software platform. This native cross-vendor tool for targeted lipidomics will allow to (i) create transitions and assays, (ii) optimize collision energies, (iii) visually review the obtained results, and (iv) quantify the lipids of interest. We envision that the lipid building block based Skyline application breaks ground not only for optimized method design, analysis, and data evaluation in targeted lipidomics, but also provide a gateway to spectral libraries and the sharing of experimental data. This will help the research community to build up a comprehensive and vendor independent exchange platform to improve reproducibility and validation processes for lipidomics data.
4
2.2 LipidXplorer: a generic software tool for shotgun lipidomics By definition, shotgun lipidomics relies on a direct infusion of total lipid extracts into a tandem mass spectrometer, however lipids can be identified in many ways. Historically, shotgun lipidomics was mainly associated with triple quadrupole mass spectrometers (reviewed in (Han et al., 2012)). Lipids were identified by a combination of precursor and neutral loss scans and, because of their low mass resolution, never relied on accurate masses of fragment precursors. In contrast, tandem mass spectrometers of the Orbitrap family and modern q-TOF deliver high mass resolution spectra along with low-ppm accuracy. Therefore masses of detected molecules can be associated with their elemental compositions and lipids consistently identified in the series of compositionally related samples by mapping their intact masses (Schwudke et al., 2007; Schwudke et al., 2011). The prerequisites for assigning correct elemental compositions of lipid precursors and fragments are still under debate (Bielow et al., 2017). Experimentally proven improvements for lipid identification were observed when the mass difference for certain elemental compositions can be resolved. The differentiation of 1-between alkyl-2-acyl glycerophospholipid and 1,2-diacyl glycerophospholipid can be achieved when the mass difference of 36.4 mDa is resolved. Furthermore, it is preferable to resolve the overlap of the second isotopic peak and lipid with exactly one less double bond which results in a mass difference of 9.0 mDa (Herzog et al., 2011; Schwudke et al., 2007). Specifically, for shotgun lipidomics, where the separation power is solely depending on the chosen MS instrument platform false assignments of major abundant glycerophospholipids can be minimized. From these estimates, one can state that the specificity of shotgun lipidomics increases for resolution above 100,000 (FWHM at m/z 750) for MS1 and 30,000 (FWHM at m/z 200) for MS2. High acquisition rate and sensitivity of any modern hybrid tandem mass spectrometers enable fragmentation of all candidate peaks independent of their intensities and produce a comprehensive dataset that comprise masses of all detectable precursors and all fragments 5
generated by their collisional fragmentation (Schwudke et al., 2006a; Schwudke et al., 2011). Therefore, shotgun lipidomics software should support any data interpretation routine independent of instrumentation platform, and methods of collisional fragmentation. It should also be able to target any lipid class and species. To this end, we developed a novel concept for the interpretation of large collections of shotgun spectra that does not rely on fixed identification rules of pre-compiled databases of masses and fragments (Herzog et al., 2011) (Herzog et al., 2012; Herzog et al., 2013). Instead, our LipidXplorer software (Figure 2) uses a declarative Molecular Fragmentation Query Language that can describe any user-defined lipid identification routine in simple and intuitive terms. In this way, the same software supports high resolution mass mapping in top-down (Schwudke et al., 2007) and bottom-up (Schuhmann et al., 2011) screens; precursor and neutral loss scanning on low resolution triple quadrupole mass spectrometers (Herzog et al., 2012) as well as combined interpretation routines such as multiple precursor-, neutral loss and boolean scans enabled by data-dependent acquisition (Herzog et al., 2013). LipidXplorer can also identify novel lipids that bear some structural similarity to known lipids (Papan et al., 2014). However, the significant limitation of shotgun datasets is their inherent complexity. In a typical workflow after the extraction with organic solvents and MS analysis a dataset acquired in a single experiment might consist of several hundreds of MS and MS/MS spectra and comprise > 100 000 of unique peaks, however only a few hundred of those might be eventually attributed to lipids (Herzog et al., 2011). Although the software design supports many features valued by experienced users, it lacks the transparency and ease of use important for laboratories entering the field and ready-to-use automation tools that further enable and support routine lipidomics measurements.
6
We therefore will extend LipidXplorer capabilities with data processing modules, which will greatly simplify the interpretation of shotgun datasets while maintaining the key features of LipidXplorer. We will also develop a web-version of LipidXplorer that will offer a generic support for routine shotgun experiments, which will be of interest to a broad community of users willing to apply lipidomics in their focused biological projects. Together with other members of the LIFS consortium, we will integrate LipidXplorer with LipidHome, Skyline and LUX tools.
2.3 LipidHome: a curation model for experimental data LipidHome aims to close a gap between theoretically calculated lipid molecules and metadata by implementing a resource comparable in function to UniProt, (Foster et al., 2013). The database was initially populated with lipids that were generated in-silico based on community-accepted rules concerning chemical and structural properties. In addition, a web interface was created to present the information and provide computational access. Tailored specifically to handle high throughput mass spectrometry based approaches, lipids are organized in the database in a hierarchical manner. The level of structural resolution provided by the applied lipidomics method is further implemented by using the short hand lipid annotation that was recently introduced by Liebisch et al. (Liebisch et al., 2013). To reflect the level of confidence of reported lipids in the database we will validate lipid entries at three levels: level C will include all lipids that were already described in literature, level B will contain lipids that were measured and identified by high resolution mass spectrometer at the MS and/or MS2 level and the A level will contain lipids quantified via an internal standard to provide absolute quantities. Further information will include the source organism and tissue where a particular lipid has been observed. Cross-references to other lipid related resources and scientific articles that cite specific lipids will be reported using a specifically improved ontology for lipids. The web application will be directly linked to the LIFS-platform and allow a matching of MS1 and MS2 7
derived lipidomics data. Data submission will be based on the Metabolights database, hosted by the European Bioinformatics Institute (EBI), which provides unique and stable IDs for each submitted dataset as well as data validation and annotation steps based on a standardized submission format (Haug et al., 2013). The alpha version of LipidHome is currently accessible at https://www.ebi.ac.uk/metabolights/lipidhome/.
2.4 Lipid Homology – LUX Score The concept of lipid homology was recently established to provide a rigorous metric comparable to genetic analyses based on sequences (Marella et al., 2015). From this point of view, a lipidome is the set of lipids that reflect not only biosynthetic capacity of an organism but also functional aspects for a certain cell or cell compartment. Recent studies on lipidomes of biological model organisms (Bozek et al., 2015; Carvalho et al., 2012; Ejsing et al., 2009; Klose et al., 2012; Watschinger et al., 2015) and clinical relevant human samples (Chua et al., 2013; Graessler et al., 2009; Sales et al., 2016; Slatter et al., 2016) emphasize that lipidomes give insight into health status and provide molecular fingerprints similar to expression profiles. From this perspective, we developed a homology metric, the ‘Lipidome jUXtaposition (LUX) score’ that can quantify systematic differences in the composition of a lipidome (Figure 3). In a first step, the structural similarity between all lipids is determined. For that, all identified lipids are converted into SMILES (Weininger, 1988) taking into account all possible isomeric structures that are not distinguishable by the chosen mass spectrometric approach (Liebisch et al., 2013). To date, only a limited number of approaches were developed to determine the connectivity of fatty acyl chains and double bond positions of phosphoglycerolipids. Ozone-induced dissociation produces diagnostic fragments for the position of unsaturation within the precursor ion helping to characterize the natural structural complexity of lipids (Thomas et al., 2008). Furthermore, fragmentation techniques based on radical interaction can potentially enrich lipidomics datasets by definite information on the double bond position (Pham et al., 2012). 8
We showed that Template-based SMILES, in which chemical nomenclature is the basis for unique representation of lipids, can correctly represent the structural diversity of isomeric lipids. Specifically, position of double bonds and hydroxyl groups and elongation of aliphatic chains are accurately documented. On basis of Template-based SMILES, a chemical space model of a lipidome can be computed based on pair wise Levenshtein distances (Levenshtein, 1966). With this approach, a well-defined metric space is computed and an intuitive visualization of the lipidome composition is achieved. Several lipidomes can be projected into the same chemical space model, in which the degree of overlap between lipidomes can be determined. The Hausdorff distance (Hausdorff, 2005) was used to measures the difference of two lipidomes in the chemical space model, which basically embodies the LUX score. We tested the LUX score on four yeast strains with known genetic alteration in fatty acid synthesis (Ejsing et al., 2009) and showed that the underlying genetic relationship and growth temperature are better represented than correlation based clustering based on lipid concentrations. Next, we applied this metric for lipdiomics screen (Schwudke et al., 2007; Schwudke et al., 2011) data of larval tissue lipidomes of Drosophila (Carvalho et al., 2012). This showed that the LUX score is sufficient to determine the impact of nutritional changes in an unbiased manner, despite the limited information on the structural diversity of each lipidome. A software package was developed that covers all necessary steps to determine lipidome homology using the LUX score (http://lifs.isas.de/index.php/lux-score). However, improvements for the general usability, import of result files and integration into the LipidHome database are required to make the lipidome homology concept beneficial for the research community. We will integrate the determination of Template-SMILES into the LipidHome database as a look-up service and use the LUX Score software as a backend for homology calculation between lipidomes. Furthermore, a seamless integration into the reporting structure and lipid naming convention of LipidXplorer and Skyline will be a further focus of software implementation. In view 9
of the growing demand to establish functional association of lipids, the LUX score will foster comparative lipidome studies that can bridge species borders, which is specifically important for translational research and application of disease models.
3 Summary The LIFS consortia will consolidate a number of software for lipidomics that were separately developed. By integration of the shotgun lipidomics software LipidXplorer, the targeted analysis tool Skyline for Lipdomics/LipidCreator, the database LipidHome and the LUX score approach for systematic lipidome comparison, we aim to provide a platform independent Open Source software suite that will foster lipid research (Figure 4). It is of high importance to establish a new quality on how results are communicated in lipidomics through this resource and to improve the transparency of computational methods used for lipidomics data analysis. For that, the tool integration will be one of the crucial tasks for the LIFS consortia to broaden the user and developer base. As a member of BioInfra.Prot and together with a community driven course program, tutorials and computational services, we envisage that LIFS will be a driving force for future lipidomics research.
Declaration of interest This work was supported by the Ministerium für Innovation, Wissenschaft und Forschung des Landes Nordrhein-Westfalen, the Senatsverwaltung für Wirtschaft, Technologie und Forschung des Landes Berlin, and the Bundesministerium für Bildung und Forschung, SUPR-G e:Med (Code 01ZX1401C) and de.NBI program (code 031L0108A, 031L0108B, 031L0108C). The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.
10
References
Beloribi-Djefaflia, S., Vasseur, S., Guillaumond, F., (2016) Lipid metabolic reprogramming in cancer cells. Oncogenesis 5, e189. Bielow, C., Mastrobuoni, G., Orioli, M., Kempa, S., (2017) On Mass Ambiguities in High-Resolution Shotgun Lipidomics. Anal Chem 89, 2986-2994. Bou Khalil, M., Hou, W., Zhou, H., Elisma, F., Swayne, L.A., Blanchard, A.P., Yao, Z., Bennett, S.A., Figeys, D., (2010) Lipidomics era: accomplishments and challenges. Mass Spectrom Rev 29, 877-929. Bozek, K., Wei, Y., Yan, Z., Liu, X., Xiong, J., Sugimoto, M., Tomita, M., Paabo, S., Sherwood, C.C., Hof, P.R., Ely, J.J., Li, Y., Steinhauser, D., Willmitzer, L., Giavalisco, P., Khaitovich, P., (2015) Organization and evolution of brain lipidome revealed by large-scale analysis of human, chimpanzee, macaque, and mouse tissues. Neuron 85, 695-702. Carvalho, M., Sampaio, J.L., Palm, W., Brankatschk, M., Eaton, S., Shevchenko, A., (2012) Effects of diet and development on the Drosophila lipidome. Mol Syst Biol 8, 600. Chua, E.C., Shui, G., Lee, I.T., Lau, P., Tan, L.C., Yeo, S.C., Lam, B.D., Bulchand, S., Summers, S.A., Puvanendran, K., Rozen, S.G., Wenk, M.R., Gooley, J.J., (2013) Extensive diversity in circadian regulation of plasma lipids and evidence for different circadian metabolic phenotypes in humans. Proc Natl Acad Sci U S A 110, 14468-14473. Ejsing, C.S., Sampaio, J.L., Surendranath, V., Duchoslav, E., Ekroos, K., Klemm, R.W., Simons, K., Shevchenko, A., (2009) Global analysis of the yeast lipidome by quantitative shotgun mass spectrometry. Proc Natl Acad Sci U S A 106, 2136-2141. Fahy, E., Sud, M., Cotter, D., Subramaniam, S., (2007) LIPID MAPS online tools for lipid research. Nucleic Acids Res 35, W606-612. Foster, J.M., Moreno, P., Fabregat, A., Hermjakob, H., Steinbeck, C., Apweiler, R., Wakelam, M.J., Vizcaino, J.A., (2013) LipidHome: a database of theoretical lipids optimized for high throughput mass spectrometry lipidomics. PLoS One 8, e61951. Graessler, J., Schwudke, D., Schwarz, P.E., Herzog, R., Shevchenko, A., Bornstein, S.R., (2009) Top-down lipidomics reveals ether lipid deficiency in blood plasma of hypertensive patients. PLoS One 4, e6261. Haimi, P., Uphoff, A., Hermansson, M., Somerharju, P., (2006) Software Tools for Analysis of Mass Spectrometric Lipidome Data. Anal. Chem. 78, 8324-8331. Han, X., (2016) Lipidomics for studying metabolism. Nat Rev Endocrinol 12, 668-679.
11
Han, X., Yang, K., Gross, R.W., (2012) Multi-dimensional mass spectrometry-based shotgun lipidomics and novel strategies for lipidomic analyses. Mass Spectrom Rev 31, 134-178. Haug, K., Salek, R.M., Conesa, P., Hastings, J., de Matos, P., Rijnbeek, M., Mahendraker, T., Williams, M., Neumann, S., Rocca-Serra, P., Maguire, E., González-Beltrán, A., Sansone, S.-A., Griffin, J.L., Steinbeck, C., (2013) MetaboLights—an open-access general-purpose repository for metabolomics studies and associated meta-data. Nucleic Acids Research 41, D781-D786. Hausdorff, F., (2005) Set Theory, 10 ed. AMS Chelsea Pub., American Mathematical Society. Herzog, R., Schuhmann, K., Schwudke, D., Sampaio, J.L., Bornstein, S.R., Schroeder, M., Shevchenko, A., (2012) LipidXplorer: a software for consensual cross-platform lipidomics. PLoS One 7, e29851. Herzog, R., Schwudke, D., Schuhmann, K., Sampaio, J.L., Bornstein, S.R., Schroeder, M., Shevchenko, A., (2011) A novel informatics concept for high-throughput shotgun lipidomics based on the molecular fragmentation query language. Genome Biol 12, R8. Herzog, R., Schwudke, D., Shevchenko, A., (2013) LipidXplorer: Software for Quantitative Shotgun Lipidomics Compatible with Multiple Mass Spectrometry Platforms. Curr Protoc Bioinformatics 43, 14 12 11-30. Hubner, G., Crone, C., Lindner, B., (2009) lipID--a software tool for automated assignment of lipids in mass spectra. J Mass Spectrom 44, 1676-1683. Husen, P., Tarasov, K., Katafiasz, M., Sokol, E., Vogt, J., Baumgart, J., Nitsch, R., Ekroos, K., Ejsing, C.S., (2013) Analysis of lipid experiments (ALEX): a software framework for analysis of high-resolution shotgun lipidomics data. PLoS One 8, e79736. Kaiser, J., (2002) Proteomics. Public-private group maps out initiatives. Science 296, 827. Kind, T., Liu, K.H., Lee do, Y., DeFelice, B., Meissen, J.K., Fiehn, O., (2013) LipidBlast in silico tandem mass spectrometry database for lipid identification. Nat Methods 10, 755-758. Klose, C., Surma, M.A., Gerl, M.J., Meyenhofer, F., Shevchenko, A., Simons, K., (2012) Flexibility of a eukaryotic lipidome--insights from yeast lipidomics. PLoS One 7, e35063. Klose, C., Surma, M.A., Simons, K., (2013) Organellar lipidomics--background and perspectives. Curr Opin Cell Biol 25, 406-413. Kochen, M.A., Chambers, M.C., Holman, J.D., Nesvizhskii, A.I., Weintraub, S.T., Belisle, J.T., Islam, M.N., Griss, J., Tabb, D.L., (2016) Greazy: Open-Source Software for Automated Phospholipid Tandem Mass Spectrometry Identification. Anal Chem 88, 5733-5741. Lam, S.M., Shui, G., (2013) Lipidomics as a principal tool for advancing biomedical research. J Genet Genomics 40, 375-390. Leavell, M.D., Leary, J.A., (2006) Fatty Acid Analysis Tool (FAAT): An FT-ICR MS Lipid Analysis Algorithm. Anal. Chem. 78, 5497–5503. 12
Levenshtein, V.I., (1966) Binary codes capable of correcting deletions, insertions and reversals. Soviet Physics Doklady 10, 707--710. Liebisch, G., Vizcaino, J.A., Kofeler, H., Trotzmuller, M., Griffiths, W.J., Schmitz, G., Spener, F., Wakelam, M.J., (2013) Shorthand notation for lipid structures derived from mass spectrometry. J Lipid Res 54, 1523-1530. MacLean, B., Tomazela, D.M., Shulman, N., Chambers, M., Finney, G.L., Frewen, B., Kern, R., Tabb, D.L., Liebler, D.C., MacCoss, M.J., (2010) Skyline: an open source document editor for creating and analyzing targeted proteomics experiments. Bioinformatics 26, 966-968. Marella, C., Torda, A.E., Schwudke, D., (2015) The LUX Score: A Metric for Lipidome Homology. PLoS Comput Biol 11, e1004511. Papan, C., Penkov, S., Herzog, R., Thiele, C., Kurzchalia, T., Shevchenko, A., (2014) Systematic screening for novel lipids by shotgun lipidomics. Anal Chem 86, 2703-2710. Peng, B., Ahrends, R., (2016) Adaptation of Skyline for Targeted Lipidomics. J Proteome Res 15, 291-301. Pham, H.T., Ly, T., Trevitt, A.J., Mitchell, T.W., Blanksby, S.J., (2012) Differentiation of complex lipid isomers by radical-directed dissociation mass spectrometry. Anal Chem 84, 7525-7532. Sales, S., Graessler, J., Ciucci, S., Al-Atrib, R., Vihervaara, T., Schuhmann, K., Kauhanen, D., Sysi-Aho, M., Bornstein, S.R., Bickle, M., Cannistraci, C.V., Ekroos, K., Shevchenko, A., (2016) Gender, Contraceptives and Individual Metabolic Predisposition Shape a Healthy Plasma Lipidome. Sci Rep 6, 27710. Sampaio, J.L., Gerl, M.J., Klose, C., Ejsing, C.S., Beug, H., Simons, K., Shevchenko, A., (2011) Membrane lipidome of an epithelial cell line. Proc Natl Acad Sci U S A 108, 1903-1907. Schuhmann, K., Herzog, R., Schwudke, D., Metelmann-Strupat, W., Bornstein, S.R., Shevchenko, A., (2011) Bottom-up shotgun lipidomics by higher energy collisional dissociation on LTQ Orbitrap mass spectrometers. Anal Chem 83, 5480-5487. Schwudke, D., Hannich, J.T., Surendranath, V., Grimard, V., Moehring, T., Burton, L., Kurzchalia, T., Shevchenko, A., (2007) Top-down lipidomic screens by multivariate analysis of high-resolution survey mass spectra. Anal Chem 79, 4083-4093. Schwudke, D., Oegema, J., Burton, L., Entchev, E., Hannich, J.T., Ejsing, C.S., Kurzchalia, T., Shevchenko, A., (2006a) Lipid profiling by multiple precursor and neutral loss scanning driven by the data-dependent acquisition. Anal Chem 78, 585-595. Schwudke, D., Oegema, J., Burton, L., Entchev, E., Hannich, J.T., Ejsing, C.S., Kurzchalia, T., Shevchenko, A., (2006b) Lipid Profiling by Multiple Precursor and Neutral Loss Scanning Driven by the DataDependent Acquisition. Anal. Chem. 78, 585–595. Schwudke, D., Schuhmann, K., Herzog, R., Bornstein, S.R., Shevchenko, A., (2011) Shotgun lipidomics on high resolution mass spectrometers. Cold Spring Harb Perspect Biol 3, a004614. 13
Shevchenko, A., Simons, K., (2010) Lipidomics: coming to grips with lipid diversity. Nat Rev Mol Cell Biol 11, 593-598. Slatter, D.A., Aldrovandi, M., O'Connor, A., Allen, S.M., Brasher, C.J., Murphy, R.C., Mecklemann, S., Ravi, S., Darley-Usmar, V., O'Donnell, V.B., (2016) Mapping the Human Platelet Lipidome Reveals Cytosolic Phospholipase A2 as a Regulator of Mitochondrial Bioenergetics during Activation. Cell Metab 23, 930944. Song, H., Hsu, F.F., Ladenson, J., Turk, J., (2007) Algorithm for processing raw mass spectrometric data to identify and quantitate complex lipid molecular species in mixtures by data-dependent scanning and fragment ion database searching. J Am Soc Mass Spectrom 18, 1848-1858. Sud, M., Fahy, E., Cotter, D., Brown, A., Dennis, E.A., Glass, C.K., Merrill, A.H., Jr., Murphy, R.C., Raetz, C.R., Russell, D.W., Subramaniam, S., (2007) LMSD: LIPID MAPS structure database. Nucleic Acids Res 35, D527-532. Taguchi, R., Nishijima, M., Shimizu, T., (2007) Basic Analytical Systems for Lipidomics by Mass Spectrometry in Japan. Methods in Enzymology 432, 185–211. Thomas, M.C., Mitchell, T.W., Harman, D.G., Deeley, J.M., Nealon, J.R., Blanksby, S.J., (2008) Ozoneinduced dissociation: elucidation of double bond position within mass-selected lipid ions. Anal Chem 80, 303-311. van Meer, G., (2005) Cellular lipidomics. The EMBO Journal 24, 3159-3165. Wang, M., Han, X., (2016) Advanced Shotgun Lipidomics for Characterization of Altered Lipid Patterns in Neurodegenerative Diseases and Brain Injury. Methods Mol Biol 1303, 405-422. Watschinger, K., Keller, M.A., McNeill, E., Alam, M.T., Lai, S., Sailer, S., Rauch, V., Patel, J., Hermetter, A., Golderer, G., Geley, S., Werner-Felmayer, G., Plumb, R.S., Astarita, G., Ralser, M., Channon, K.M., Werner, E.R., (2015) Tetrahydrobiopterin and alkylglycerol monooxygenase substantially alter the murine macrophage lipidome. Proc Natl Acad Sci U S A 112, 2431-2436. Weininger, D., (1988) SMILES, a chemical language and information system. J. Chem. Inf. Comput. Sci., 31-36. Wenk, M.R., (2005) The emerging field of lipidomics. Nat Rev Drug Discov 4, 594-610.
14
Figure legends Figure 1 Skyline for lipidomics: a comprehensive platform for targeted assays. A) Main lipid classes can be assembled from lipid building blocks such as fatty acid chain, long chain base and head group to create the precursor mass their fragment ions. B) Utilization of LipidCreator in the Skyline environment. Lipid information is transferred to Skyline by LipidCreator to compute transition lists and in-silico spectral libraries. Afterwards, Skyline can be used for visualization, optimization, data evaluation and quantification
of
targeted
lipidomics
results.
Abbreviations:
PG
(Glycerophosphoglycerol), Cer (Ceramide), DG (Diradylglycerol), TG (Triradylglycerol), CL (Cardiolipin).
Figure 2 LipidXplorer, a generic software tool for shotgun lipidomics. LipidXplorer organizes MS and MS/MS spectra (acquired from all samples of lipidomics experiments, including biological and technical repeats) into a flat-file database termed MasterScan. Scans are averaged and individual spectra aligned considering instrument specific peak attributes, such as mass resolution, mass accuracy, and peak occupancy, among others. The MasterScan is then interrogated by user-defined lipid class –specific and/or lipid species -specific queries written in the Molecular Fragmentation Query Language (MFQL). Identified and annotated lipid species, along with intensities of user-defined fragment or precursor ions, are reported. Figure 3 Lipid homology determination using LUX Score. Result files of LIFS applications LipidXplorer and Skyline/LipidCreator are directly used as input. Other 15
result files can be imported containing lipid names and quantities. B) Template SMILES are received via LipidHome. When Gene Ontology information are provided lipids are scored according to proven repositories for organism, tissues and cell types. C) For all lipids of all lipidomes Template SMILES are generated and used as model to compute chemical space model. D) From reference lipidome maps the pairwise LUX Score are calculated. Figure 4 LIFS web portal implementation. Service requests and input data are processed at three levels. 1) Lipid Identification and design of lipidomics experiments using LipidXplorer and Skyline/LipidCreator. 2) Evaluation of Lipid IDs in context of Gene Ontology and data repositories using LipidHome. 3) Lipidomes are compared using LUX Score for functional associations.
16
Figure 1)
Figure 2)
17
Figure 3)
Figure 4
18
