Single-Cell Lipidomics by LC-MS Interlaboratory Study Reveals the Impact of X-ray Irradiation on a Pancreatic Cancer Cell Line and Its Bystanders

Abstract

Live single-cell lipidomics by liquid chromatography mass spectrometry (LC-MS) is a nascent and rapidly growing field which can shed new light on infectious diseases, cancer, immunology, and drug delivery. There are now a growing number of laboratories that can isolate single cells and laboratories that can perform lipidomics analysis at correspondingly low sample volumes, but there is a lack of validation data. We have carried out the first interlaboratory LC-MS lipidomics experiment for single cells, aimed at filling this gap. We present a novel workflow to enable interlaboratory studies, comprising live-cell imaging and single-cell isolation, followed by freeze-drying, international shipping, reconstitution, and untargeted lipidomics analysis. We applied this methodology to reveal radiation-induced bystander effects in pancreatic cancer cells. X-ray irradiated cells and their bystanders sampled live 48 h postirradiation demonstrated reduced lipid abundance compared to controls, with distinct changes in molar ratios of several polyunsaturated lipids. This demonstrates for the first time that radiation can cause considerable cellular lipid remodelling, not only at the site of delivery. A striking similarity in lipid changes was observed between the two participating laboratories despite differences in sample preparation and analysis methods. Our results are further corroborated by live-cell imaging analysis of lipid droplets. This work serves as an important validation and demonstration of the nascent and rapidly growing field of live single-cell lipidomics.

1

Workflow for cell irradiation, sampling, and analysis. PANC1 cells were irradiated through a slit in lead shielding and incubated for 48 h. The Yokogawa SS2000 was used to sample directly irradiated cells, bystanders, and cells from a control dish. The cells were divided into two batches: the first batch was stored at −80 °C and then analyzed by analytical flow LC-MS using an Orbitrap Q Exactive Plus instrument, and the second batch was freeze-dried, shipped at room temperature to San Jose, and analyzed using a nanoflow LC-MS method on an Orbitrap Exploris 240 instrument.

2

Single-cell lipidomics of shipped PANC-1 cells. (A) % RSD of the internal standards eluted alongside single cells in three batches with and without shipment (batch 1, n = 9; batch 2, n = 9; batch 3, n = 18, unshipped, n = 14). (B) mol % distribution of lipid classes of putatively identified lipids in 12 single PANC-1 cells. (C) Number of lipids detected vs their detection consistency across 12 PANC-1 cells. (D) Distribution of unique lipid identifications made in positive/negative ionization modes in 12 PANC-1 cells.

3

Lipid profile of shipped single PANC-1 cells. (A) PLS-DA plots of lipidomics output from control (n = 12) and directly irradiated (n = 12) cells. (B)­Total molar lipid content of control (n = 12) and directly irradiated (n = 12) cells. (C.) PC mol % of selected species containing arachidonic acid (20:4, ω-6) and docosahexaenoic acid (22:6, ω-3) in unirradiated controls (n = 12) and directly irradiated (n = 12) cells. *p < 0.05 as determined by two-tailed Mann–Whitney U test, **p < 0.01.

4

Lipid content of unshipped (validation) and shipped (discovery) single PANC-1 cells. (A) Total molar content of unshipped validation (control, n = 6; direct irradiation, n = 5) and shipped discovery cells (control, n = 12; direct irradiation, n = 12). (B) Overlap of significantly altered lipids between discovery and validation data sets. (C) Trends in PC(36:4) between unshipped validation (control, n = 6; direct irradiation, n = 5) and shipped discovery cell control (n = 12; direct irradiation, n = 12). *p < 0.05 as determined by two-tailed Mann–Whitney U test, **p < 0.01.

5

Lipid droplet profile of unshipped single PANC-1 cells. (A–C.) PLS-DA plots of lipidomics output from control (n = 12) directly irradiated (n = 12) and bystander (n = 12) cells. (A) Combined fluorescence (blue = Hoechst, pink = BODIPY 493/503) and brightfield channel image with overlay of cells isolated for lipidomics analysis (yellow). Image taken from population of bystander cells. Scale bar: 50 μm. (B) Receiver operating characteristic of the machine learning model used to characterize cells by treatment conditions (AUC = 0.94). (C) Lipid droplet area per cell between control (n = 12), directly irradiated (n = 12), and bystander cells (n = 12) directly sampled for lipidomics. (D) Trends in arachidonic acid-containing PC lipids between control (n = 12), directly irradiated (n = 12), and bystander cells (n = 12). *p < 0.05 as determined by two-tailed Mann–Whitney U test, **p < 0.01.