Extended Data Fig. 8: Characterisation of neurosphere and xenograft metabolic phenotypes.
a) PCA plot using MSI signal intensities of 13C-labelled pyruvate, lactate, succinate, malate, fumarate, glutamate and glutamine. The loadings plot (inset) shows that the oxidative spheres separate to the right of PC1 and the glycolytic spheres to the left. Although the MSI data were segmented into 3 metabolic phenotypes, the relative activities of the glycolytic and oxidative pathways are a continuum. b) Neurosphere RNA sequencing DGSEA analysis showed upregulation of hallmark genes in glycolysis and hypoxia in the more glycolytic spheres, as determined by MSI (labelled here as group A). c) Neurospheres grown in normoxia and hypoxia (0.5% O2). PCA plot of RNA sequencing data showing separation of neurospheres into three groups. Hypoxia did not shift the position of these groups on PC1 and PC2 significantly. d) Contiguous sections from S2, AT8 and AT5 xenografts stained for Ki-67, CC3 and CD31. e) Ki67, CC3 and CD31 signals normalized to tissue area (mm2) were not significantly different between the three tumour models (n = 3 biological replicates per tumour model; data presented as mean ± SD). f, g, h) Correlation of glutamate labelling in the primary cell lines with response to Imatinib (R2 = .55, F(1, 9) = 11.0, p = 0.009) (f), Gefitinib (R2 = .38, F(1, 9) = 5.46, p = 0.0443) (g) and AZD2014 (R2 = .55, F(1, 9) = 10.83, p = 0.0094) (h) treatment. The y-axis shows cell line response to drug normalised to control, where 1.0 indicates unsuppressed growth in the presence of the drug and values less than 1.0 indicate reduced cell growth relative to control.
