Preneoplastic cells switch to Warburg metabolism from their inception exposing multiple vulnerabilities for targeted elimination

Abstract

Otto Warburg described tumour cells as displaying enhanced aerobic glycolysis whilst maintaining defective oxidative phosphorylation (OXPHOS) for energy production almost 100 years ago [1, 2]. Since then, the 'Warburg effect' has been widely accepted as a key feature of rapidly proliferating cancer cells [3-5]. What is not clear is how early "Warburg metabolism" initiates in cancer and whether changes in energy metabolism might influence tumour progression ab initio. We set out to investigate energy metabolism in the HRAS^G12V driven preneoplastic cell (PNC) at inception, in a zebrafish skin PNC model. We find that, within 24 h of HRAS^G12V induction, PNCs upregulate glycolysis and blocking glycolysis reduces PNC proliferation, whilst increasing available glucose enhances PNC proliferation and reduces apoptosis. Impaired OXPHOS accompanies enhanced glycolysis in PNCs, and a mild complex I inhibitor, metformin, selectively suppresses expansion of PNCs. Enhanced mitochondrial fragmentation might be underlining impaired OXPHOS and blocking mitochondrial fragmentation triggers PNC apoptosis. Our data indicate that altered energy metabolism is one of the earliest events upon oncogene activation in somatic cells, which allows a targeted and effective PNC elimination.

Fig. 1. Glycolysis is important in boosting PNC proliferation and excess glucose promotes PNC survival and expansion.

A Schematic showing the inducible human HRAS^G12V mediated preneoplastic cell in zebrafish larval skin tissue. Seahorse® Metabolic Flux analyses were carried out at 24 h post induction (hpi). A a, explanation of ECAR trace b, representative confocal images of EdU staining (proliferation) and anti-cleaved-Caspase3 staining (apoptosis) that were carried out in this study. B Seahorse Analyser® ECAR readout over time, showing no difference in baseline ECAR (before cycle 6), glycolytic flux (after adding 2-DG) readout at cycle 7 showing enhanced glycolytic flux in HRAS PNCs. Note: 2-DG leads to a transient ECAR change which recovers from cycle 8, this is thought to be due to the whole organism response to 2-DG. FCCP was added (black line) to assess respiratory function using the complimentary OCR readout (data not shown, as similar OCR data were presented in Fig. 2). C Quantification (cycle 7 ECAR) showing PNC have enhanced glycolytic flux (mean +/− SD, From 4 experiments n ≥ 24, p = 0.0042). Hexokinase inhibitor lonidamine (2 nM) treatment leads to reduced EdU positive cells in PNCs (unpaired t test, mean +/− SD, 2 experiments, animal n ≥ 19, p = 0.0001). D Pseudobulk differential expression analysis of single-cell RNA sequencing data, showing significantly up- and down- regulated genes related to glycolysis in HRAS PNCs vs. Control CAAX keratinocytes at 24 hpi. Heatmap depicts log fold-change (EdgeR, n = 2, FDR < 0.05). E Hexokinase inhibitor lonidamine (2 nM) treatment did not change PNC apoptosis (unpaired t test, mean +/− SD, 3 experiments, animal n ≥ 22, p = 0.2042). F Glucose injected larvae show increased PNC proliferation (unpaired t test, mean +/− SD, 2 experiments, animal n ≥ 18, p = 0.0002). G Glucose injected larvae show decreased PNC apoptosis (unpaired t test, mean +/− SD, 2 experiments, animal n ≥ 14, p = 0.0119).

Fig. 2. OXPHOS is impaired in PNCs and complex I inhibitor metformin suppresses PNC proliferation and induces PNC apoptosis in vivo.

A Seahorse Analyser® Oxygen Consumption Rate (OCR) measurement comparing control larvae with PNC bearing larvae, graph showing OCR readout over time. Cycle 4 showing similar baseline respiration. Uncoupler FCCP was added after cycle 4 (black line) which assesses the reserve OCR. Cycle 7 showing significantly reduced maximum OCR in PNC bearing larvae, indicating reduced reserved respiration capacity. After cycle 8, treatment with the complex III inhibitor antimycin and the complex I inhibitor rotenone (black line) allowed the non-respiratory contribution to OCR to be determined, and there was no difference detected. (p < Mean +/− SD, 4 experiments, animal n ≥ 20 embryos, Two-way ANOVA followed by Sidak’s multiple comparisons test). B Quantification showing maximum respiratory capacity is reduced in HRAS expressing PNCs (cycle 6 OCR, unpaired t test, mean +/− SD, n ≥ 20, p = 0.0326). C Quantification showing maximum respiratory capacity is reduced in HRAS expressing PNCs (cycle 7 OCR, mean +/− SD unpaired t test, n ≥ 20, *p = 0.0193) D Gene-set enrichment analysis of single-cell RNA sequencing data shows that oxidative phosphorylation is enriched in HRAS expressing PNCs vs. CAAX expressing control keratinocytes (NES = 1.4401, p = 0.0409, FDR = 0.1308). E Quantification showing metformin treatment induces superficial skin PNC apoptosis within 4hpt (Mean +/− SEM, Mann-Whitney test, 2 experiments, animal n ≥ 12, p = 0.0001). F Quantification showing metformin treatment induces basal skin PNC apoptosis (Mean +/− SD, unpaired t test, 2 experiments, animal n ≥ 10, p = 0.0185). G quantification showing reduced proliferation of basal PNCs upon metformin treatment (Mean +/− SD, unpaired t test, 3 experiments, animal n ≥ 28, p = 0.0072). H Quantification showing metformin treatment leads to reduced PNC burden at 48 hpi (PNC fluorescent volume in defined area per animal; Mean +/− SD, unpaired t test, n ≥ 11, p = 0.0102). I quantification showing reduced proliferation of basal PNCs upon IACS treatment (%EdU incorporation in PNCs; Mean +/− SD, unpaired t test, n ≥ 9, p < 0.0001). J Quantification showing reduced PNC burden at 48 hpi upon IACS treatment (PNC number per animal; Mean +/− SD, unpaired t test, 2 experiments n ≥ 6, p = 0.0016).

Fig. 3. Mitochondria in PNCs are fragmented and have reduced membrane potential.

A Confocal images of mito-trackers CMXRos and Deep-Red stained zebrafish larval skin cells. a, b, c, indicate “Zoom in” area of mito-tracker deep-red image to show details of mitochondrial fragmentation phenotype in HRAS expressing skin PNCs. Scale bar = 10 µm. B Electron microscope image of a mitochondrion in normal skin cells of zebrafish larva. C Electron microscope image of mitochondria in HRAS expressing skin PNC of zebrafish larva. D Quantification showing increased number of mitochondrial fragments in PNCs (Mann-Whitney test, p = 0.0087, 3 experiments, animal n ≥ 8) E Quantification showing decreased mitochondrial fragment size in PNCs (Mann Whitney test, median, p < 0.0001) F Quantification showing decreased mitochondrial membrane potential in PNCs (unpaired t tests, 24 hpi; 2 experiments, animal n ≥ 6, p = 0.004; 8 hpi n ≥ 3, p = 0.0099).

Fig. 4. mdivi suppression of Drp1/Dnml1 blocks mitochondrial fission and induces PNC apoptosis.

A Violin plot, depicting normalized expression of dnml1 in single-cell RNA sequencing dataset, shows that dnml1 is up-regulated in PNCs vs. CAAX control keratinocytes. B Pseudobulk differential expression analysis of single-cell RNA sequencing data, showing genes related to mitochondrial fission and fusion in HRAS PNCs vs. control keratinocytes at 24 hpi. Heatmap depicts log fold-change (EdgeR, n = 2, FDR < 0.05). C Quantification showing increased mitochondrial fragmentation can be reversed by mdivi treatment (Kruskal-Wallis test with Dunn’s multiple comparisons, p = 0.001, 3 experiments, animal n ≥ 8) D Quantification showing reduced size of mitochondrial fragments can be reversed by mdivi treatment (Kruskal-Wallis test with Dunn’s multiple comparisons, p < 0.0001) E Quantification showing mdivi does not alter PNC proliferation (unpaired t test, p = 0.1889, n ≥ 22). F quantification showing mdivi induced PNC apoptosis (mean +/− SD, Mann-Whitney test, p = 0.001, 2 experiments, animal n = 11).