Looking into Lipids

Lipidomics is one of the youngest branches of “omics” research. Maria Fedorova from Leipzig University, in Leipzig, Germany, discusses the latest trends and challenges in lipidomics research and highlights how innovative bioinformatics solutions are addressing data handling issues in this evolving field.

Q. What is the definition of lipidomics?

A: Lipidomics is the large-scale study of diversified molecular species of lipids, with the aim of addressing the identification and cellular and tissue distribution of lipids as well as their related signalling and metabolic pathways in a variety of organisms.

As with any “omics” study, lipidomics aims to describe the whole variety of lipid species and to provide knowledge on their diversity, distribution, and concentration, which can then be used for further systems biology and systems medicine data integration.

Lipidomics is probably the youngest addition to the family of classical “omics” studies, which includes genomics, proteomics, transcriptomics, and metabolomics. A lot of effort is currently directed at providing the inventory of natural lipidomes. It might sound surprising, but only a limited number of organisms or tissue-specific lipidomes have been characterized so far. In fact, for human tissues, the blood lipidome is probably the only in-depth characterization that has been verified by different laboratories.

Another active area in lipidomics research is the identification of lipid markers that can be associated with human health and pathologies. The dynamic nature of lipids and their deep involvement in a variety of functional activities makes them very attractive biomolecules for diagnostic, prognostic, and therapeutic applications.

Q. What are the main aims of your research group?

A: Our group focuses on the development and optimization of analytical and bioinformatics solutions for high-throughput lipidomics with the aim of studying human metabolic disorders. We would like to have the tools to perform a deep lipidomics profiling of human tissues to create reference lipidomes and integrate the data on existing lipid species in genome-scale metabolic models to describe the whole set of biochemical reactions driven by corresponding enzymes (genes).

This type of deep lipidomics profiling requires a combination of several analytical and computation strategies to ensure the high quality of the data and cannot be called high-throughput. However, having the whole set of lipid species in the tissue (or at least the majority of it) integrated using systems biology and systems medicine tools would allow us to design high-throughput and robust analytical solutions suitable for translation to the clinic.

Another focus of our group is the characterization of oxidized lipids derived using reactions catalyzed by dedicated enzymes, for example, cyclooxygenase and lipoxygenase, or by free-radical-driven oxidation in conditions generally classified as oxidative stress or redox imbalance (1–3).

These modified lipids represent the fraction of what we call epilipidomes, a subset of lipidomes derived by modifying the native lipids. Indeed, similar to epigenomes and proteoforms, which have been shown to play significant regulatory roles on other “omics” levels, modified lipids perform fine-tuning of metabolic and signalling functions. Relatively well studied at the level of free fatty acids (eicosanoids and prostaglandins), oxidative modifications of phospholipids and triglycerides (TGs) are not currently well understood.

The oxidation of fatty acyl chains changes the physicochemical properties of lipids, which causes them to function differently. However, from an analytical perspective, this also requires the optimization of new separation methods and mass spectrometry (MS)-based protocols. We are working on methods that combine liquid chromatography (LC) (mostly reversed phase) and tandem mass spectrometry (MS/MS) to specifically detect, identify, and quantify oxidized lipids in human lipidomes connected to pathologies associated with chronic inflammation where redox dysregulation plays a significant role (4–8).

Q. What are the main challenges facing separation scientists involved in lipidomics and how are they being overcome? Are there any recent developments in lipidomics from an analytical perspective that you think are particularly innovative?

A: Lipids are difficult to analyze because (i) they have very different physicochemical properties and require different extraction and separation methods), (ii) they are usually present at very different concentrations in biological samples so require methods with a wide dynamic range, and (iii) we actually still do not know the whole variety of lipid species and thus their physicochemical properties in natural lipidomes. This makes it difficult-if not impossible-to have an “all-in-one” analytical solution capable of detecting and identifying all the different lipid species in a given biological system.

The combination of separation techniques with modern MS instruments capable of high-resolution, mass accuracy, sensitivity, and speed has significantly improved the dynamic range in lipidomics analysis. This offers us the possibility of identifying hundreds of lipid species from natural lipidomes on the fatty acyl level in one LC–MS/MS analysis. However, the development of analytical tools to define fatty acyl chain positions, for example, sn-1 versus sn-2 in phospholipids, remains challenging for high-throughput applications.

A lot of progress was recently achieved for methods to define double bond positions in esterified fatty acyl chains. Several methods based on different gas-phase fragmentation mechanisms, such as ozone-induced dissociation (OzID) and ultraviolet photodissociation (UVPD), as well as chemical derivatization strategies, such as the Paterno-Büchi reaction or epoxidation, offered the possibility to identify isomeric lipid species (9–14).

The separation of some structural isomeric lipids, such as bis(monoacylglycero)phosphates (BMP) and phosphatidylglycerols (PG), remains challenging as well. Methods based on chemical derivatization (methylation), ion mobility spectrometry (IMS), and separation using zwitterionic hydrophilic interaction liquid chromatography (HILIC) stationary phases using optimized concentrations of ammonium acetate were recently demonstrated (15–17).

Q. There are three major chromatography techniques used in lipidomics: normal phase, reversed phase, and HILIC. Are there distinct application areas where these individual categories of chromatography are being used in lipidomics?

A: As the majority of the lipidomics studies rely on the on-line coupling of LC to electrospray ionization (ESI)-MS, normal-phase chromatography is a less popular method because of the low compatibility of the mobile phase components with ESI (18–20).

Both reversed phase and HILIC are used very widely in lipidomics. Reversed phase remains the most popular choice for lipidomic profiling because of the ability to separate multiple lipid species within the same lipid class based on the length and number of double bonds in fatty acyl chains (21,22). HILIC provides lipid class-based separation and can be the optimal choice for lipid quantification using class-specific internal standards that would closely coelute with multiple lipid species within corresponding lipid classes (21). Moreover, the complementary nature of reversed phase and HILIC separation mechanisms makes the combination of both techniques a very attractive choice for deep lipidomics profiling, especially when dealing with the lipidomes characterized by a wide range of polarities and lipid concentrations.

Q. Can you comment on method selection and practical considerations for the choice of stationary phases and mobile phases for normal phase, reversed phase, and HILIC in lipidomics?

A: In my opinion, the choice of the stationary phase chemistry should be defined by the lipidome that needs to be analyzed. As a result of the differences in the polarities, as well as the range of concentrations for lipids from different classes, there would be no universal choice for LC stationary phases.

Recently we performed a comparison of five reversed-phase columns with different stationary phase surface chemistry (C18 versus C30), types of stationary phase particles (fully porous particles [FPP] versus. solid-core particles [SCP]), and particle size (1.9 µm versus. 2.6 µm versus. 3.0 µm) using the same mobile phases and tandem mass spectrometry method to resolve the human blood plasma lipidome (23). We demonstrated that not all C18 columns are efficient for lipid chromatography and selection should not be based entirely on particle size. Thus, pore size, as well as surface area, can play a significant role for stationary phases with the same surface chemistry. Columns with fully porous sub-2-μm particles and solid‑core 2.6-μm particles usually perform well.

The choice of the surface chemistry depends on the polarity of the studied lipidome. Thus, a C18 stationary phase is recommended for the analysis of lipidomes of intermediate polarity, for example, a plasma lipidome with a relative high content of both phospholipids and triacylglycerol and cholesteryl esters, while C30 columns would be more suitable for samples with a high content of long chain hydrophobic lipids, for example, adipose tissue.

Gradient elution using water–acetonitrile–isopropanol is probably the most popular mobile phase used for reversed-phase chromatography in lipidomics. Methanol can also be included into the eluent system. In HILIC, unbound Si-based columns are the most popular for lipid separation. Among polar-bonded phases are polyvinylalcohol- and dihydroxypropyl-modified silica stationary phases. Recently, interesting applications of zwitterionic HILIC columns have been demonstrated (16). The majority of mobile phases consist of acetonitrile and aqueous buffers (ammonium formate and acetate), supplemented with minor amounts of isopropyl alcohol (IPA), methyl tert-butyl ether (MTBE), methanol, or other polar, water-miscible solvents.

In general, the selection of mobile phase-including suitable additives and their concentrations-is crucial in lipidomics. For example, terminal phosphate groups in some lipid classes interact with stainless steel material in the flow path of the high performance liquid chromatography (HPLC) systems leading to peak tailing. This effect can be eliminated by adding phosphoric acid in the samples or by substituting all HPLC tubing to PEEK material. Another significant challenge in optimizing the mobile phase for optimal separation of complex lipidomes is the different dissociation states of phospholipids at different pHs. The coexistence of a single lipid in a charged, ionized state together with its neutral form would result in peak broadening and tailing. Thus, one should tune the pH of mobile phase to ensure uniform distribution of dissociation state for different lipid classes.

Q. Supercritical fluid chromatography (SFC) is also used to a lesser extent. When is SFC useful?

A: The application of SFC for lipidomics analysis has shown a high potential over the last decade. SFC combines the advantages of both gas chromatography (GC) and HPLC (low back pressure, solubility of analytes, and good kinetic performance) resulting in high efficiency and short separation times. Ultrahigh‑performance SFC separation using sub-2-μm unmodified and functionalized silica stationary phases coupled on-line to ESI-MS was used for the analysis of complex lipidomes as well as the separation of lipids within different lipid classes (24,25). For example, the separation of 30 lipid classes within 6 min was recently demonstrated by the group of Michal Holčapek (26).

Q. Are comprehensive chromatography techniques commonly used in lipidomics?

A: Several very interesting examples of comprehensive chromatography for lipidomics studies have been published. Off-line and even on-line coupling of two orthogonal separation techniques illustrated deep lipidomics coverage (27–32). However, routine application of comprehensive chromatography techniques is not common. Off-line combinations of two chromatographic techniques, such as HILIC and reversed phase (which is the most popular orthogonal system), are relatively easy to perform, but do not provide high-throughput. Automated on-line coupling would provide the most robust solution. However, instrumentation, such as an ultrahigh-pressure liquid chromatography (UHPLC) system with two sets of pumps, would require additional investment, and the analytical workflow would still need to be optimized to ensure full capacity for both separation modes. Instead deep lipidomics profiling usually relies on the combination of different lipid extraction methods, as well as fractionation using solid-phase extraction (SPE) and liquid–liquid extraction (LLE) protocols.

Q. Big data is always a concern in any “omics” field. You recently published two papers related to data handling using open source software: LipidHunter and LPPtiger. What solutions do these offer separation scientists?

A: As I mentioned previously, lipidomics is the youngest addition to the classical “omics” family and computational solutions to support high-throughput analytical workflows are much less developed compared to transcriptomics, proteomics, and metabolomics. One of the main bottlenecks remains reliable high‑throughput identification of lipids from LC–MS/MS datasets. With this in mind, we developed LipidHunter as an open source software to identify lipids from LC–MS/MS datasets obtained using data-dependent acquisition (33). When we started in lipidomics, we went through hundreds of tandem mass spectra manually to learn how we can confidently identify lipids. When we got tired of doing this manually, we created this software to do it for us. This software repeats all the steps of lipid identification one would do during manual identification, but much faster and keeps the whole identification process very transparent and traceable. The first version of this software only dealt with phospholipid identifications, but when we started to work with adipose tissue lipidome we extended it to glycerolipids and the second version is now freely available (https://github.com/SysMedOs/lipidhunter).

LPPtiger (LPP stands for lipid peroxidation products) is a software tool for the analysis of oxidized phospholipids (https://bitbucket.org/SysMedOs/lpptiger). Oxidized phospholipids represent a very interesting fraction of epilipidome. However, the identification from LC–MS/MS cannot be directly translated from native phospholipids. With this software, we implemented several new algorithms including in silico prediction of oxidized epilipidome from native lipidome provided to the software. To perform data-driven prediction of oxidized lipids rather than simple enumeration of oxygen atoms to unsaturated fatty acyl chains, we conducted the meta-study on the available literature describing mechanisms of oxidation for polyunsaturated fatty acids and used the information from more than 170 publications to integrate these data in metabolic networks representing oxidation pathways for ten polyunsaturated fatty acids (PUFAs). LPPtiger relies on these metabolic networks to perform data-driven in silico oxidation (34).

Q. Are the any other “big data” solutions being adopted in lipidomics?

A: There are several other very good open source computational solutions supporting lipid identification from LC–MS/MS datasets including LipidBlast, MSDIAL, and Lipostar just to name a few (35–37).

Further directions supporting big data integration in lipidomics should provide tools for pathway mapping and network integration of lipidomics data. The LipidMaps consortium brings different aspects of lipidomics studies together and provides an integrative platform for lipid analysis (38).

Q. Are there any developments in sample preparation in lipidomics that are worth commenting on?

A: Once again, the choice of extraction method depends on the biological matrix and lipid species variety and concentrations present in the sample. Folch-, Bligh and Dyer-, and MTBE‑based methods are very popular, as well as the butanol–methanol extraction (BUME protocol) (39).

Q. Adipose tissue lipidomes are currently a main focus of research. Why are these molecules important and what analytical strategies are used to analyze these analytes?

A: The role of adipose tissue in human physiology was reconsidered after the discovery of adipokines and their role in the regulation of human metabolism and immune responses (40–42). Adipose tissue metabolism was correlated with insulin sensitivity status as well as chronic inflammation accompanying numerous human pathologies, changing our view on the role of adipose as an inactive lipid storage organ to the active regulator of whole-body metabolism. Together with liver and all lipoproteins in blood, adipose tissue is among the most crucial organs in lipid trafficking, distribution, regulation, and metabolism.

However, it still remains largely unknown which exact lipid species are present in adipose tissue. We know it contains massive amounts of triglycerides, but are they all the same? Which fatty acyl chains are esterified and de-esterified? What is the dynamic of this process and how is it regulated?

Currently we have identified over 1000 individual TG lipids in human white adipose tissue (data unpublished). Why do we have such a large variety of individual species of TGs? What is the difference in lipidomes of white adipose tissue from different depots, for example visceral versus subcutaneous, and different insulin sensitivity states? We try to answer these questions by combining dedicated analytical workflows with systems biology tools to provide an integrative fatty acid-centric view on adipose tissue metabolism.

For example, to uncover the diversity of adipose tissue lipidome, we combined several extraction and fractionation (SPE and LLE) methods followed by nuclear magnetic resonance (NMR), thin-layer chromatography (TLC), HILIC–MS/MS, and reversed-phase LC–MS/MS on C18 and C30 columns (data unpublished). Furthermore, to ensure high-quality, reliable identification of lipids from multiple measurement platforms, we used a combination of three different lipid identification software tools, which allowed us to compose the reference lipidome of human white adipose tissue, including over 1600 lipid molecular species. All these data are currently used for integration into genome-scale metabolic models specific for adipose tissue. The availability of a high-quality in-depth characterized lipidome described by means of a genome-scale metabolic (GEM) model will provide us with the possibility to understand disease‑associated metabolic changes in lipidomes profiled using more targeted techniques applied for a large number of human adipose tissue samples.

Q. What other projects are you working on at the moment?

A: We are also looking into the lipotoxicity effects connected with ectopic lipid accumulation in cardiac cells. Using a cell culture model of mild nitroxidative stress, we demonstrated the formation of lipid droplets in cardiac cells accompanied by the accumulation of oxidized lipids (43). In collaboration with the group of Dolores Perez-Sala in CSIC Madrid (Spain) and the group of Professor Spengler at the University of Giessen (Germany), we combined confocal fluorescent microscopy, LC–MS/MS-based lipidomics, and single-cell matrix-assisted laser desorption–ionization (MALDI) imaging to understand the distribution of lipids and their oxidized forms upon lipid droplets formation and associated the dynamic of these droplets with autophagy‑lysosomal degradation pathway.

Acknowledgements

The project is supported by the German Ministry of Education and Research (BMBF) within the framework of the e:Med research and funding concept for SysMedOS project.

References

1. Z. Ni and M. Fedorova, in Encyclopedia of Lipidomics, M.R. Wenck, Ed. (Springer Science+Busines Media, 2018).

1. C.M. Spickett, M. Fedorova, R. Hoffmann, and H.J. Forman, in Health and Disease, (Taylor and Francis, 2015), pp. 1–22.

1. M. Fedorova and R. Hoffmann, in Lipid Oxidation in Health and Disease, (Taylor and Francis, 2015), pp. 151–172.

1. A. Annibal, T. Riemer, O. Jovanovic, E.E. Pohl, J. Schiller, R. Hoffmann, and M. Fedorova, Free Radical Biology and Medicine95, 293–307 (2016).

1. I. Milic, E. Griesser, V. Vemula, N. Ieda, H. Nakagawa, N. Miyata, J.M. Galano, C. Oger, T. Durand, and M. Fedorova, Analytical and Bioanalytical Chemistry407(19), 5587–602 (2015).

1. D. Weber, Z. Ni, D. Vetter, R. Hoffmann, and M. Fedorova, European Journal of Lipid Science118(2), 325–331 (2015).

1. Z. Ni, I. Milic, and M. Fedorova, Analytical and Bioanalytical Chemistry407(17), 5161–5163 (2015).

1. I. Milic and M. Fedorova, in Methods in Molecular Biology. Advanced Protocols in Oxidative Stress III (Humana Press, vol. 1208, 2015), pp. 3–20.

1. B.L.J. Poad, X. Zheng, T.W. Mitchell, R.D. Smith, E.S. Baker, and S.J. Blanksby, Anal. Chem.90(2), 1292–1300 (2018).

1. D.L. Marshall, H.T. Pham, M. Bhujel, J.S. Chin, J.Y. Yew, K. Mori, T.W. Mitchell, and S.J. Blanksby, Anal. Chem.88(5), 2685–2692 (2016).

1. P.E. Williams, D.R. Klein, S.M. Greer, and J.S. Brodbelt, J. Am. Chem. Soc.139(44), 15681–15690 (2017).

1. A. Korf, V. Jeck, R. Schmid, P.O. Helmer, and H. Hayen, Anal. Chem. (2019). DOI: 10.1021/acs.analchem.8b05493.

1. A. Bednařík, S. Bölsker, J. Soltwisch, and K. Dreisewerd, Angew. Chem. Int. Ed. Engl.57(37), 12092–12096 (2018).

1. Y. Feng, B. Chen, Q. Yu, and L. Li, Anal. Chem.91(3), 1791–1795 (2019).

1. M. Wang, J.P. Palavicini, A. Cseresznye, and X. Han, Anal. Chem.89(16), 8490–8495 (2017).

1. C. Vosse, C. Wienken, C. Cadenas, and H. Hayen, J. Chromatogr. A1565, 105–113 (2018).

1. K.L. Leaptrot, J.C. May, J.N. Dodds, and J.A. McLean, Nat. Commun.10(1), 985 (2019).

1. E. Sokol et al., Anal. Biochem.443, 88–96 (2013).

1. S. Abreu, A. Solgadi, and P. Chaminade, J. Chromatogr. A1514, 54–71 (2017).

1. N. Prache, S. Abreu, P. Sassiat, D. Thiébaut, and P. Chaminade, J. Chromatogr. A1464, 55–63 (2016).

1. M. Lange, Z. Ni, A. Criscuolo, and M. Fedorova, Chromatographia82(1), 77–100 (2019).

1. T. Cajka and O. Fiehn, Trends Analyt. Chem.61, 192–206 (2014).

1. A. Criscuolo et al., Chemistry and Physics of Lipids. SI Contemporary Lipidomics. in press (2019). DOI: 10.1016/j.chemphyslip.2019.03.006

1. Y. Yang, Y. Liang, J. Yang, F. Ye, T. Zhou, and L. Gongke, J. Pharm. Anal.9(1), 1–8 (2019).

1. L. Laboureur, M. Ollero, and D. Touboul, Int. J. Mol. Sci.16(6), 13868–84 (2015).

1. M. Lísa and M. Holčapek, Methods Mol. Biol.1730, 73–82 (2018).

1. S. Wang et al., J. Chromatogr. A1321, 65–72 (2013).

1. M. Narváez-Rivas, N. Vu, G.-Y. Chen, and Q. Zhang, Anal. Chim. Acta954, 140–150 (2017).

1. L. Mondello et al., J. Chromatogr. A1086, 91–98 (2005).

1. P. Dugo, T. Kumm, M.L. Crupi, A. Cotroneo, and L. Mondello, J. Chromatogr. A1112, 269–275 (2006).

1. M. Holčapek, M. Ovčačíková, M. Lísa, E. Cífková, and T. Hájek, Anal. Bioanal. Chem.407(17), 5033–43 (2015).

1. Y.S. Ling et al., Biomed. Chromatogr. 28, 1284–1293 (2014).

1. Z. Ni, G. Angelidou, M. Lange, R. Hoffmann, and M. Fedorova, Anal. Chem.89(17), 8800–8807 (2017).

1. Z. Ni, G. Angelidou, R. Hoffmann, and M. Fedorova, Scientific Reports7(1), 15138 (2017).

1. T. Kind, K.-H. Liu, D.Y. Lee, B. DeFelice, J.K. Meissen, and O. Fiehn, Nat. Methods10, 755–758 (2013).

1. L. Goracci, S. Tortorella, P. Tiberi, R.M. Pellegrino, A. Di Veroli, A. Valeri, and G. Cruciani, Anal. Chem.89, 6257–6264 (2017).

1. H. Tsugawa, T. Cajka, T. Kind, Y. Ma, B. Higgins, K. Ikeda, M. Kanazawa, J. VanderGheynst, O. Fiehn, and M. Arita, Nat. Methods12, 523–526 (2015).

1. E. Fahy et al., J. Lipid Res.50, S9–S14 (2009).

1. S. Tumanov and J. Kamphorst, J. Curr. Opin. Biotechnol.43, 127–133 (2017).

1. A.L. Ghaben and P.E. Scherer, Nat. Rev. Mol. Cell. Biol.20(4), 242–258 (2019).

1. S.M. Reilly and A.R. Saltiel, Nat. Rev. Endocrinol.13(11), 633–643 (2017).

1. M. Kumari, J. Heeren, and L. Scheja, Semin. Immunopathol. 40(2), 189–202 (2018).

1. G. Coliva, S. Duarte, D. Perez-Sala, and M. Fedorova, Redox Biol. 101123 (2019).

Maria Fedorova studied biochemistry at Saint Petersburg State University, in Saint Petersburg, Russia, and obtained her Ph.D. at the Faculty of Chemistry and Mineralogy, at Leipzig University, in Leipzig, Germany. She is a group leader at the Institute of Bioanalytical Chemistry, Faculty of Chemistry and Mineralogy, at the University of Leipzig, Germany. Her research is focused on the development and optimization of chromatography and mass spectrometry methods for the analysis of lipids and their modified forms. Her group works on implementation of high‑throughput LC–MS methods in discovery lipidomics targeting in-depth identification and quantification of human lipidome in a variety of tissues. By combining lipidomics data with investigating related proteins and protein post-translational modifications via a systems medicine approach, she aims for a deeper understanding of pathophysiology of obesity, insulin resistance, type II diabetes, and cardiovascular disorders.