Mitochondrial mutations and metabolic adaptation in pancreatic cancer

Abstract

Background: Pancreatic cancer has a five-year survival rate of ~8%, with characteristic molecular heterogeneity and restricted treatment options. Targeting metabolism has emerged as a potentially effective therapeutic strategy for cancers such as pancreatic cancer, which are driven by genetic alterations that are not tractable drug targets. Although somatic mitochondrial genome (mtDNA) mutations have been observed in various tumors types, understanding of metabolic genotype-phenotype relationships is limited.

Methods: We deployed an integrated approach combining genomics, metabolomics, and phenotypic analysis on a unique cohort of patient-derived pancreatic cancer cell lines (PDCLs). Genome analysis was performed via targeted sequencing of the mitochondrial genome (mtDNA) and nuclear genes encoding mitochondrial components and metabolic genes. Phenotypic characterization of PDCLs included measurement of cellular oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) using a Seahorse XF extracellular flux analyser, targeted metabolomics and pathway profiling, and radiolabelled glutamine tracing.

Results: We identified 24 somatic mutations in the mtDNA of 12 patient-derived pancreatic cancer cell lines (PDCLs). A further 18 mutations were identified in a targeted study of ~1000 nuclear genes important for mitochondrial function and metabolism. Comparison with reference datasets indicated a strong selection bias for non-synonymous mutants with predicted functional effects. Phenotypic analysis showed metabolic changes consistent with mitochondrial dysfunction, including reduced oxygen consumption and increased glycolysis. Metabolomics and radiolabeled substrate tracing indicated the initiation of reductive glutamine metabolism and lipid synthesis in tumours.

Conclusions: The heterogeneous genomic landscape of pancreatic tumours may converge on a common metabolic phenotype, with individual tumours adapting to increased anabolic demands via different genetic mechanisms. Targeting resulting metabolic phenotypes may be a productive therapeutic strategy.

Keywords: Genome; Glutamine; Lipid; Metabolomics; Mitochondria; Pancreas.

Fig. 1

Mitochondrial genome (mtDNA) sequence analysis in pancreatic cancer PDCLs. a Schematic showing the approach used to map genotype:phenotype relationships in pancreatic cancer. b Distribution of somatic mtDNA mutations (red lines, n = 24) in 12 pancreatic PDCLs, showing strong bias towards variants in ETC complex I coding and control regions. ETC subunit coding regions are denoted by subunit (colour coded by ETC complex). Position of tRNAs are noted, with ticks marking 500 bp intervals. c Strong selection for non-synonymous mutants in mtDNA and 1056 nuclear genes important for mitochondrial function and metabolism in pancreatic PDCLs compared with a reference survey of mitochondrial variants in infantile mitochondrial disease (Calvo et al. [39]). Statistical comparisons performed using Chi-squared analysis

Fig. 2

Metabolic profile of TKCC and normal HPDE pancreatic cell lines. Basal oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) during mitochondrial stress test using Seahorse analyser (data presented as mean +/− s.d., n = 5)

Fig. 3

Metabolomics analysis of pancreatic cancer PDCLs. a Partial least squares discrimination analysis (PLS-DA) of metabolomics profiles from all pancreatic cell lines (normal and PDCL) used in this study. b Relative intracellular metabolite levels for 13 pancreatic PDCLs and normal HPDE pancreatic cell line grown in two different media (K-SFM and M199/F12). Colour key (top right) is superimposed by a histogram showing counts of all identified metabolites (Z-score). Rows list the identified metabolite names, and columns list the pancreatic cell lines arranged according to hierarchical clustering analysis. c Heirarchical clustering analysis of changes in intracellular metabolite levels for 13 pancreatic PDCLs relative to HPDE normal cells (in K-SFM media). Fold change of metabolites with significantly different levels by ANOVA (p < 0.05) are shown. d Succinate abundance in PDCLs and HPDE (normalised to protein content)

Fig. 4

Metabolic pathway flux analysis: a. Heirarchical clustering of significantly altered metabolic pathways (ANOVA, p < 0.05) in 13 PDCLs relative to normal HPDE cells (in K-SFM media), identified by PAPi analysis. Various culture media are represented by colours along the top of columns (grey = M199/F12, green = RPMI, purple = HPAC modified, red = IMDM). b Conversion of glutamine to lipid by ¹⁴C glutamine tracing in PDCLs cultured under normoxic and hypoxic conditions. c Metabolomics data are consistent with the activation of reductive carboxylation in pancreatic tumours, driving conversion of glutamine to lipid in the presence of low oxygen and ETC inhibition

Fig. 5

Proposed convergence model of mtDNA mutations driving metabolic adaptation in pancreatic cancer. We propose a model in which the underlying nuclear genomic landscape of pancreatic cancer cells induces a metabolic challenge. As an adaptation to increased biosynthetic requirements, tumours attenuate oxidative phosphorylation through positive selection for diverse somatic mitochondrial mutations, which converge on common metabolic phenotypes