Fig. 5: Galactose-induced metabolic stress leads to a global cellular response without clonal selection in POLG^D274A cells.

a Schematic overview of the experiment. CTRL and POLG^D274A KI36 cells were cultured in 10 mM glucose for 3 days or 10 mM galactose for 1 or 3 days, followed by mtscATAC-seq with cell hashing for joint profiling. b UMAP embeddings of KI36 single cells colored by treatment condition. Glu, Glucose; Gal, Galactose. n = cell numbers. c Violin plots of single-cell mtDNA depth normalized to the number of unique ATAC read counts per condition in POLG^D274A KI36 cells. Holm–Bonferroni adjusted p-values from the two-sided Wilcoxon test are shown, with significance at a type I error rate of 0.05. Boxes show the median (center line) and the interquartile range (25th–75th percentiles; box). Whiskers extend to the 5th and 95th percentiles. Outliers are displayed as single points. d Scatter plots showing per-variant pseudobulk VAFs in KI36 cells treated with galactose for 3 days (y-axis) vs. glucose for 3 days (x-axis), per indicated (functional) category. e Cumulative distribution plots of single-cell VAFs for the most shifted variants highlighted in (e), stratified by treatment condition. f UMAP projections of KI36 cells colored by the VAF of subclone-specific somatic mtDNA variants as identified in Fig. 4e. g Volcano plots of differentially accessible genes (DAGs) in subclones C1 (left) and C3 (right) comparing galactose vs. glucose treatment (3 days). Statistically significant upregulated (FDR ≤ 0.05 with log₂FC  >= 0.6) DAGs in galactose treatment are highlighted in red, and downregulated (FDR ≤ 0.05 with log₂FC >= 0.6) DAGs are highlighted in blue. The Wilcoxon pairwise test method was utilized to identify statistically significant DAGs. h Pseudobulk chromatin accessibility track plot at the GDF15 locus. The blue dashed boxes highlight select differentially accessible peaks.