Pyxis Unveiled: Advancing Single-Cell MALDI MSI Analysis for Deeper Molecular Insights

72nd ASMS Conference on Mass Spectrometry. June 2024

Ismael Zamora1; Rachelle Balez2; Jayden C. McKinnon2; Reuben S.E. Young2; Liam Robinson2; Lezanne Ooi2; Giuseppe Arturi3; Giulia Sorbi3; Shane Ellis2Sara Tortorella3
1Mass Analytica, Sant Cugat del Vallès, Spain; 2Molecular Horizons, University of Wollongong, Wollongong, Australia; 3Mass Analytica, Bettona, Italy

Abstract

Introduction

Single-cell metabolomics and lipidomics using Matrix-Assisted Laser Desorption/Ionization Mass Spectrometry Imaging (MALDI MSI) is emerging as an exciting tool to study metabolic alterations across heterogeneous cell populations. Single-cell analysis using MALDI MSI does however come with technical challenges, including sample preparation workflows compatible with fluorescent microscopy and downstream MS analysis, as well as dedicated data processing workflow to coregister microscopy with MSI and extract profiles from defined cells for later analysis. Here, we introduce an efficient single-cell MSI data analysis pipeline utilising the cutting-edge Pyxis software platform. The performance and versatility of the pipeline are demonstrated by single-cell lipidomics and metabolomics analysis of human-derived astrocytes, unravelling intricate molecular insights at the single-cell level.

Methods

Human astrocytes were generated using induced pluripotent stem cell-derived neurons. Live astrocyte cultures were stained with CellBrite Green and Hoechst, fixed with PFA and washed with cold ammonium acetate prior to microscopy. Following microscopy cells were coated in 2,5-DHA matrix via sublimation for lipid imaging or NEDC matrix using a HTC TM-Sprayer for metabolite imaging for MALDI and MALDI-2 analysis, respectively, using an Orbitrap Elite coupled to a Spectroglyph MALDI/ESI ion source (Spectroglyph LLC, Kennewick, WA, USA). Coregistration with microscopy, segmentation at single-cell level, extraction of spectra generated from single cells, and dedicated statistics to process single-cell spectra were performed using Pyxis (Mass Analytica, Spain). Features of interest were annotated using HMDB and LIPID MAPS databases integrated in Pyxis.

Preliminary data

Rich lipid and metabolite data were recorded from single cells using both MALDI and MALDI-2 with pixel sizes as low as 10 microns. Given the relatively large size of the astrocytes multiple pixels were recorded across single cells, shedding light into sub-cellular metabolite distributions. For example the [M-H]- of adenine was detected predominantly in the soma of the cells.
Ad hoc algorithms were developed and integrated to cope with the data analysis challenges. Rigid and non-rigid strategies to coregister microscopy, fluorescence and MS images were evaluated. An hybrid approach was employed, where rigid registration was applied initially to achieve a coarse alignment, followed by non-rigid registration to refine the alignment at a finer spatial scale. Utilising both approaches in a complementary fashion enhanced the accuracy of spatial integration of multi-modal imaging data. Bisecting k-means and spatial denoising facilitated single-cell border identification. An algorithm was designed to automatically isolate cells, apply user-defined exclusion criteria, calculate single-cell profiles, which then underwent multivariate statistical analysis.
Demonstrated applications of this analytical and data analysis workflow include: (i) single-cell analysis of human-derived astrocyte populations  and (ii) investigating the single-cell metabolic and lipidomic responses to inflammatory cytokine stimulation and how these are correlated with cytockeletal remodelling upon inflammatory activation.
By implementing all the necessary data analysis steps into a single, user-friendly software platform, we enable the comprehensive exploration of single-cell MSI data. This integrated approach not only streamlines the intricate process of biochemical data interpretation but also ensures that users can fully exploit the richness of information within their single-cell datasets with ease.

 

You must be logged in to access this content. Not yet registered? Create a new account

 

 

Tags -