Mass Spectrometry-based Single-Cell Lipidomics: Advancements, Challenges, and the Path Forward

Abstract

In the past decade, lipidomics, now recognized as standalone subdiscipline of metabolomics, has gained considerable attention. Due to its sensitivity and unparalleled versatility, mass spectrometry (MS) has emerged as the tool of choice for lipid identification and detection. Traditional MS-based lipidomics are performed on bulk cell samples. While informative, these bulk-scale cellular lipidome measurements mask cellular heterogeneity across seemingly homogeneous populations of cells. Unfortunately, single cell lipidomics methodology and analyses are considerably behind genomics, transcriptomics, and proteomics. Therefore, the cell-to-cell heterogeneity and related function remains largely unexplored for lipidomics. Herein, we review recent advances in MS-based single cell lipidomics. We also explore the root causes for the slow development of single-cell lipidomics techniques. We aim to provide insights on the pivotal knowledge gaps that have been neglected, prohibiting the propulsion of the single-cell lipidomics field forward, while also providing our perspective towards future methodologies that can pave a path forward.

Keywords: Matrix-Assisted Laser Desorption/Ionization; ambient ionization; bioanalysis; cellular; cellular heterogeneity; data analysis; electrospray ionization; lipidomics; machine learning; mass spectrometry; mass spectrometry imaging; metabolomics; omics technologies; single-cell analysis; subcellular analysis; tandem mass spectrometry.

Figure 1.

Number of publications per year for single cell lipidomics, metabolomics, proteomics, transcriptomics, and genomics. Single cell proteomics, transcriptomics, and genomics have demonstrated considerable advances in the past decades. Comparatively, the single cell lipidomics and metabolomics fields lag considerably behind, as highlighted by the annual trends in publications across all single cell “omics” communities. Data accessed December 29, 2022, from Web of Science using the key words “single cell” and the corresponding “omics” discipline.

Figure 2.

Overview of SCMS lipidomics workflow. (1) The first step in any SCMS lipidomics workflow involves sample collection from tissue, biofluid, etc. (2) Next, metabolic quenching is performed to preserve the lipidome/metabolome composition at the time of sampling. (3) Single cell isolation, (4) cell lysis procedures, and (5) lipid extraction for small volumes are performed prior to analysis. (6) To aid in lipid identification, chemical conjugation strategies are commonly utilized. (7) Following sample isolation and preparation, lipids can be analyzed via MS, and (8) resulting data can be processed using a variety of biostatistical, machine learning methods and tools.

Figure 3.

Schematic overview of single cell isolation strategies frequently used for SCMS lipidomics. (left) Isolation of single cells from heterogenous population are done by (A) micropipetting apparatus and strategies, use of (B) microfluidic devices to isolate each cell in a droplet, or by using (C) fluorescence assisted cell sorting (FACS) techniques to isolate single cells in well plates for downstream analysis.

Figure 4.. Representation of hierarchical classification for lipid structural identification.

The number of potential lipid structures were accessed from the Lipid Maps database on December 13, 2022. (Left) Structural hierarchy of a complex lipid, represented by a phosphatidylcholine molecule: PC 16:0/18:1(9Z). (Right) Example for MS-based annotation to identify PC lipid species.

Figure 5.. Overview figure for the lipidomic MS data analytical platform.

The top panel shows the MS-based cell lipidomic analysis workflow for bulk and single cells. 1) This beings with sample preparations for MS analysis which includes cell lysis and chemical conjugation. 2–3) After MS analysis, data is converted from raw to a user friendly format. 4–5) Two methods, database referencing, or rule-based methods are used for lipid identification. The center panel shows the bulk cell lipidomic bioinformatics workflow. Steps 1–4 follow the first panel workflow. 5a) Bulk lipid identification will result in a lipid signature for each cell type. 6a) A machine learning model will profile the lipid signatures of cell types. 7a) UMAP will be used to capture the high dimensions of lipid signatures and visualized and clustered according to their type. The bottom panel shows the single-cell lipidomic bioinformatics workflow. Steps 1–4 follow the first panel workflow. 5b) Single-cell lipid identification will result in a lipid signature for each cell state. 6b) A machine learning model will profile the lipid signature for each cell state. 7b) UMAP will be used to capture the high dimensions of lipid signatures and visualize and cluster them according to their state.