Extended Data Fig. 2: Multi-tissue metabolomics in the C26 mouse model.
(a-o) See also Fig. 1a for experimental setup. Ctrl: healthy controls, no tumour; Non-cax: non-cachectic tumour mice; Pre-cax: pre-cachectic tumour mice; Cax: cachectic tumour mice. n = 4 animals per group. (a-b) Initial (a) and final (b) body, lean and fat mass. (c) Loss of body, lean and fat mass expressed as percentage of initial mass. (d) Final tissue weights (GC muscle, eWAT, iWAT). Data are mean ± s.e.m. Statistical analysis (a-d): unpaired one-way ANOVA with Dunnett’s post-hoc tests or unpaired Kruskal-Wallis with Dunn’s post-hoc tests vs. Ctrl. *p < 0.05, **p < 0.01, ****p < 0.0001 compared to Ctrl animals. (e) Upset plot showing all metabolites included in the analysis after filtering (see Methods). (f-m) 3D Principal Component Analysis (PCA) score plots of samples based on metabolite log-transformed, imputed and scaled data. Ellipses represent 95% confidence intervals. (n-o) Volcano plots showing the positive and negative alteration in the metabolome in the time course of cachexia development in tumour (n) and the different cachexia target tissues (o). From top to bottom: plasma, liver, eWAT, iWAT, heart, GC and soleus muscles. Data are presented as log2 fold change (Tumour groups/Ctrl). Statistical analysis of filtered data: one way ANOVA following post-hoc correction based on Tukey’s honesty significant difference procedure. Significant metabolites with a p value < 0.05 are highlighted in colored fields on each plot. See also Supplemental Data Table 1 for log-transformed imputed data, log2 fold change and p values for each tissue.
