For our primary objective, we build on two central datasets that query different aspects of the lipid mobilization potential of all LTPs. The first dataset resulted from purification of tagged LTPs expressed in human cell lines, and allows lipid mobilization identification in a cellular context. The second dataset resulted from exposure of purified LTPs to a mammalian organ lipid mix, which allows lipid mobilization identification with a broader palette of potential ligands. I contributed to the last stages of the associated screens, and to quality control, analysis refinement, and optimization for integration of both screens with each other and previously published data. Lipid identification happened in collaboration with the Saez-Rodriguez lab, where our manual analysis and derivation of identification rules helped automated pipeline development.
For more than half of the known LTPs, we managed to detect the mobilized lipids. This confirmed many of the known and predicted ligands, and uncovered previously unknown ones. The total observed lipidome showed the expected trends, but upon detailed comparison with expected distributions from the employed cell line we observed that the longest glycerophospholipids are not transported by our successfully researched LTPs, and we could identify several strong chain length and unsaturation specificities for diacylglycerol, phosphatidylserine, phosphatidylinositol, phosphatidic acid, and ceramide. Moreover, many LTPs bind more than one lipid class, which was previously only known behavior for a few of them, and most lipid classes are bound by more than one LTP.
Based on the observation that many LTPs bind more than one lipid class, we aimed to discover general principles linked to the co-transport of lipids. Firstly, we discovered that co-transport of lipids by LTPs explains a part of the co-regulation observed for lipids in the cell, even between classes of lipids that are metabolically not closely related. We researched this by comparison of our integrated datasets with a previously published lipid co-regulation network, generated by systematic mRNA perturbation of sphingolipid metabolic proteins and with clusters relevant for signaling and disease, such as the lysosomal storage diseases Krabbe, Gaucher, Farber and Chediak-Higashi. Secondly, we also discovered that co-transported lipids are more frequently observed together in tissues. To determine the co-localization of lipids in tissues, we employed the METASPACE database developed in the Alexandrov lab, which gathers spatial metabolomics data of mass spectrometry-based imaging of tissues from hundreds of research groups across the world. In collaboration, we determined the optimal co-localization metric, applied it to all tissue-sections, integrated the data with our data, and made the comparison. Moreover, our methodology also allowed to validate their METASPACE database.
Knowing this link between co-transport of lipids by LTPs and both co-regulation of lipids in the cell and co-localization in tissues, we evaluated how LTPs connect metabolic pathways by two approaches that depend on a self-developed innovative metabolic map and database. These contain the manually curated information from literature on the intracellular localization of metabolic reactions based on the subcellular localization of the involved enzymes, as well as the known molecular specificities of the enzymes. In the first approach, we subsetted all known metabolic pathways per organelle and determined how LTPs connect these and in what direction the lipid transport is likely to happen. In the second approach, we further narrowed down interlinked parts of metabolism and determined exceptional specificities for the LTPs by integrating lipid chain length and unsaturation. The first approach uncovered functional collaboration of several LTPs, including a previously unknown switching mechanism between lipid droplet biogenesis and membrane biosynthesis, currently under validation. The second approach uncovered exceptional LTP specificity for some lipids, enabling the inference of their likely location of action and function. Moreover, some of these insights also form the basis for a new ceramide release mechanism for one of the ceramide-transporting LTPs, observed with molecular dynamics by our collaborators of the Reuter lab.
The obtained insights and generated methodologies also led to several other collaborations set up in the context of the LTP-analysis in this fellowship. I contributed to the bioinformatical lipidomics data analysis for research on the isoform-dependent lipidation of Alzheimer-associated apolipoprotein E, within the Gavin lab. With the Risco lab, we identified an LTP that switches ligand upon infection with Bunyavirus, and with the Maeda lab, we helped validate a novel technology for subcellular lipidomics, which will next be applied to cancer research and will also be relevant for this LTP research.
