INTRODUCTION: Structural derangement in the mitochondrial lipid cardiolipin (CL) has been associated with mitochondrial insult which drives Warburg’s tumorigenesis hypothesis. Greatest CL acyl chain diversity has been observed in mammalian brain, making it an exemplary tissue type for understanding CL alterations in cancer. We have previously demonstrated improved CL sensitivity using desorption electrospray ionization mass spectrometry (DESI-MS) imaging coupled to high field asymmetric waveform ion mobility mass spectrometry (FAIMS), and now use this approach to investigate CL in gliomas.
OBJECTIVES: We apply DESI-FAIMS-MS imaging to investigate the spatial distribution of altered CL profiles within the heterogeneous tissue environment of human glioma and normal cortex.
METHODS: An ultraFAIMS device (Owlstone Ltd, UK) was fit to a Q Exactive Orbitrap mass spectrometer (Thermo Scientific, CA) and compensation and dispersion field voltages were optimized for CL transmission. Human tissue sections including 45 normal brain cortices and 73 gliomas (WHO grades II-IV) were imaged in negative ion mode at a spatial resolution of 200x200 µm using 1:1 ACN:DMF, with and without the FAIMS device operating at optimal static voltages. Imaged tissue sections were subject to pathological evaluation. Mitochondria were quantified from normal and glioma tissues, and IHC is being performed to correlate mitochondrial distribution with MS imaging results. ESI-CID/CID and CID/UVPD was applied to evaluate structural differences among CL observed in glioma and normal tissues.
RESULTS: Under optimal static FAIMS conditions, the average percent improvement in signal-to-noise ratio for CL 74:9 (m/z 736.485), a major CL species in normal brain tissue, was 532% when compared to DESI-MS analysis. FAIMS allowed detection of 44 different CL species from normal cortex, an average of 14 additional CL species than were detected using DESI-MS alone. Mass spectra from DESI-FAIMS-MS analysis of high-grade glioma tissues presented lower relative abundance of longer-chain CL species compared to those of normal tissues. CL profiles of glioma and normal cortex separated in principal component space, with longer chain CL species contributing to the differences among normal and glioma tissues. Among major CL species detected in both tissue types, no difference in fatty acid (FA) composition was observed. The relative ratio of FA18:1(Δ9) and FA18:1(Δ11) was determined for m/z 727.485 (CL(72:4)2-, homoacylated species) by measuring the abundance ratio of the diagnostic ions by ESI-CID/UVPD. No significant difference was observed among isomeric ratios of glioma and normal cortex.
CONCLUSION: Altered CL profiles were observed in glioma tissues compared with normal cortex, where longer-chain CL species were detected at lower relative abundance in high grade glioma tissues. This observation was facilitated by the improvement in CL sensitivity provided by the optimized DESI-FAIMS-MS platform.